{ "best_metric": null, "best_model_checkpoint": null, "epoch": 49.95201096892139, "global_step": 104150, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 9.999999793839668e-05, "loss": 2.4929, "step": 10 }, { "epoch": 0.01, "learning_rate": 9.999999175358685e-05, "loss": 1.0031, "step": 20 }, { "epoch": 0.01, "learning_rate": 9.999998144557106e-05, "loss": 0.7927, "step": 30 }, { "epoch": 0.02, "learning_rate": 9.999996701435012e-05, "loss": 0.6878, "step": 40 }, { "epoch": 0.02, "learning_rate": 9.999994845992525e-05, "loss": 0.6542, "step": 50 }, { "epoch": 0.03, "learning_rate": 9.999992578229798e-05, "loss": 0.5415, "step": 60 }, { "epoch": 0.03, "learning_rate": 9.999989898147015e-05, "loss": 0.5615, "step": 70 }, { "epoch": 0.04, "learning_rate": 9.999986805744399e-05, "loss": 0.4998, "step": 80 }, { "epoch": 0.04, "learning_rate": 9.999983301022207e-05, "loss": 0.525, "step": 90 }, { "epoch": 0.05, "learning_rate": 9.999979383980725e-05, "loss": 0.4853, "step": 100 }, { "epoch": 0.05, "learning_rate": 9.999975054620276e-05, "loss": 0.4533, "step": 110 }, { "epoch": 0.05, "learning_rate": 9.999970312941219e-05, "loss": 0.4882, "step": 120 }, { "epoch": 0.06, "learning_rate": 9.999965158943943e-05, "loss": 0.4492, "step": 130 }, { "epoch": 0.06, "learning_rate": 9.999959592628876e-05, "loss": 0.4057, "step": 140 }, { "epoch": 0.07, "learning_rate": 9.999953613996474e-05, "loss": 0.3605, "step": 150 }, { "epoch": 0.07, "learning_rate": 9.999947223047231e-05, "loss": 0.4862, "step": 160 }, { "epoch": 0.08, "learning_rate": 9.999940419781675e-05, "loss": 0.4623, "step": 170 }, { "epoch": 0.08, "learning_rate": 9.999933204200366e-05, "loss": 0.4671, "step": 180 }, { "epoch": 0.09, "learning_rate": 9.999925576303899e-05, "loss": 0.461, "step": 190 }, { "epoch": 0.09, "learning_rate": 9.999917536092901e-05, "loss": 0.4713, "step": 200 }, { "epoch": 0.1, "learning_rate": 9.999909083568042e-05, "loss": 0.4449, "step": 210 }, { "epoch": 0.1, "learning_rate": 9.999900218730012e-05, "loss": 0.3765, "step": 220 }, { "epoch": 0.11, "learning_rate": 9.999890941579544e-05, "loss": 0.4155, "step": 230 }, { "epoch": 0.11, "learning_rate": 9.999881252117403e-05, "loss": 0.3229, "step": 240 }, { "epoch": 0.11, "learning_rate": 9.999871150344389e-05, "loss": 0.3307, "step": 250 }, { "epoch": 0.12, "learning_rate": 9.999860636261333e-05, "loss": 0.3792, "step": 260 }, { "epoch": 0.12, "learning_rate": 9.999849709869104e-05, "loss": 0.3672, "step": 270 }, { "epoch": 0.13, "learning_rate": 9.999838371168604e-05, "loss": 0.393, "step": 280 }, { "epoch": 0.13, "learning_rate": 9.999826620160766e-05, "loss": 0.4057, "step": 290 }, { "epoch": 0.14, "learning_rate": 9.999814456846558e-05, "loss": 0.3285, "step": 300 }, { "epoch": 0.14, "learning_rate": 9.999801881226988e-05, "loss": 0.3485, "step": 310 }, { "epoch": 0.15, "learning_rate": 9.999788893303087e-05, "loss": 0.3556, "step": 320 }, { "epoch": 0.15, "learning_rate": 9.999775493075929e-05, "loss": 0.4098, "step": 330 }, { "epoch": 0.16, "learning_rate": 9.99976168054662e-05, "loss": 0.358, "step": 340 }, { "epoch": 0.16, "learning_rate": 9.999747455716297e-05, "loss": 0.3701, "step": 350 }, { "epoch": 0.16, "learning_rate": 9.999732818586134e-05, "loss": 0.3267, "step": 360 }, { "epoch": 0.17, "learning_rate": 9.999717769157337e-05, "loss": 0.3216, "step": 370 }, { "epoch": 0.17, "learning_rate": 9.999702307431148e-05, "loss": 0.3046, "step": 380 }, { "epoch": 0.18, "learning_rate": 9.999686433408843e-05, "loss": 0.3143, "step": 390 }, { "epoch": 0.18, "learning_rate": 9.999670147091729e-05, "loss": 0.326, "step": 400 }, { "epoch": 0.19, "learning_rate": 9.99965344848115e-05, "loss": 0.4279, "step": 410 }, { "epoch": 0.19, "learning_rate": 9.999636337578483e-05, "loss": 0.3363, "step": 420 }, { "epoch": 0.2, "learning_rate": 9.999618814385139e-05, "loss": 0.3001, "step": 430 }, { "epoch": 0.2, "learning_rate": 9.999600878902564e-05, "loss": 0.3439, "step": 440 }, { "epoch": 0.21, "learning_rate": 9.999582531132234e-05, "loss": 0.295, "step": 450 }, { "epoch": 0.21, "learning_rate": 9.999563771075666e-05, "loss": 0.365, "step": 460 }, { "epoch": 0.21, "learning_rate": 9.999544598734405e-05, "loss": 0.3086, "step": 470 }, { "epoch": 0.22, "learning_rate": 9.999525014110033e-05, "loss": 0.3364, "step": 480 }, { "epoch": 0.22, "learning_rate": 9.999505017204163e-05, "loss": 0.3188, "step": 490 }, { "epoch": 0.23, "learning_rate": 9.999484608018446e-05, "loss": 0.298, "step": 500 }, { "epoch": 0.23, "learning_rate": 9.999463786554564e-05, "loss": 0.3117, "step": 510 }, { "epoch": 0.24, "learning_rate": 9.999442552814234e-05, "loss": 0.2682, "step": 520 }, { "epoch": 0.24, "learning_rate": 9.999420906799209e-05, "loss": 0.2696, "step": 530 }, { "epoch": 0.25, "learning_rate": 9.99939884851127e-05, "loss": 0.2747, "step": 540 }, { "epoch": 0.25, "learning_rate": 9.999376377952239e-05, "loss": 0.3613, "step": 550 }, { "epoch": 0.26, "learning_rate": 9.999353495123968e-05, "loss": 0.2923, "step": 560 }, { "epoch": 0.26, "learning_rate": 9.999330200028347e-05, "loss": 0.319, "step": 570 }, { "epoch": 0.27, "learning_rate": 9.999306492667291e-05, "loss": 0.2808, "step": 580 }, { "epoch": 0.27, "learning_rate": 9.99928237304276e-05, "loss": 0.2792, "step": 590 }, { "epoch": 0.27, "learning_rate": 9.999257841156742e-05, "loss": 0.263, "step": 600 }, { "epoch": 0.28, "learning_rate": 9.999232897011259e-05, "loss": 0.2652, "step": 610 }, { "epoch": 0.28, "learning_rate": 9.999207540608368e-05, "loss": 0.341, "step": 620 }, { "epoch": 0.29, "learning_rate": 9.999181771950161e-05, "loss": 0.3548, "step": 630 }, { "epoch": 0.29, "learning_rate": 9.999155591038764e-05, "loss": 0.3418, "step": 640 }, { "epoch": 0.3, "learning_rate": 9.999128997876333e-05, "loss": 0.2733, "step": 650 }, { "epoch": 0.3, "learning_rate": 9.99910199246506e-05, "loss": 0.2502, "step": 660 }, { "epoch": 0.31, "learning_rate": 9.999074574807179e-05, "loss": 0.3016, "step": 670 }, { "epoch": 0.31, "learning_rate": 9.999046744904944e-05, "loss": 0.253, "step": 680 }, { "epoch": 0.32, "learning_rate": 9.999018502760651e-05, "loss": 0.2747, "step": 690 }, { "epoch": 0.32, "learning_rate": 9.99898984837663e-05, "loss": 0.2844, "step": 700 }, { "epoch": 0.32, "learning_rate": 9.998960781755247e-05, "loss": 0.2812, "step": 710 }, { "epoch": 0.33, "learning_rate": 9.998931302898895e-05, "loss": 0.3333, "step": 720 }, { "epoch": 0.33, "learning_rate": 9.998901411810006e-05, "loss": 0.2649, "step": 730 }, { "epoch": 0.34, "learning_rate": 9.998871108491044e-05, "loss": 0.2907, "step": 740 }, { "epoch": 0.34, "learning_rate": 9.99884039294451e-05, "loss": 0.268, "step": 750 }, { "epoch": 0.35, "learning_rate": 9.998809265172937e-05, "loss": 0.3123, "step": 760 }, { "epoch": 0.35, "learning_rate": 9.99877772517889e-05, "loss": 0.3011, "step": 770 }, { "epoch": 0.36, "learning_rate": 9.99874577296497e-05, "loss": 0.2511, "step": 780 }, { "epoch": 0.36, "learning_rate": 9.998713408533813e-05, "loss": 0.2875, "step": 790 }, { "epoch": 0.37, "learning_rate": 9.998680631888089e-05, "loss": 0.2456, "step": 800 }, { "epoch": 0.37, "learning_rate": 9.998647443030498e-05, "loss": 0.2709, "step": 810 }, { "epoch": 0.37, "learning_rate": 9.99861384196378e-05, "loss": 0.2833, "step": 820 }, { "epoch": 0.38, "learning_rate": 9.998579828690702e-05, "loss": 0.305, "step": 830 }, { "epoch": 0.38, "learning_rate": 9.998545403214073e-05, "loss": 0.2839, "step": 840 }, { "epoch": 0.39, "learning_rate": 9.99851056553673e-05, "loss": 0.2527, "step": 850 }, { "epoch": 0.39, "learning_rate": 9.998475315661544e-05, "loss": 0.284, "step": 860 }, { "epoch": 0.4, "learning_rate": 9.998439653591425e-05, "loss": 0.2837, "step": 870 }, { "epoch": 0.4, "learning_rate": 9.998403579329313e-05, "loss": 0.2429, "step": 880 }, { "epoch": 0.41, "learning_rate": 9.998367092878182e-05, "loss": 0.2575, "step": 890 }, { "epoch": 0.41, "learning_rate": 9.998330194241041e-05, "loss": 0.2345, "step": 900 }, { "epoch": 0.42, "learning_rate": 9.998292883420933e-05, "loss": 0.2645, "step": 910 }, { "epoch": 0.42, "learning_rate": 9.998255160420935e-05, "loss": 0.2894, "step": 920 }, { "epoch": 0.43, "learning_rate": 9.998217025244159e-05, "loss": 0.3108, "step": 930 }, { "epoch": 0.43, "learning_rate": 9.998178477893746e-05, "loss": 0.2815, "step": 940 }, { "epoch": 0.43, "learning_rate": 9.998139518372879e-05, "loss": 0.2468, "step": 950 }, { "epoch": 0.44, "learning_rate": 9.998100146684768e-05, "loss": 0.3, "step": 960 }, { "epoch": 0.44, "learning_rate": 9.99806036283266e-05, "loss": 0.2489, "step": 970 }, { "epoch": 0.45, "learning_rate": 9.998020166819838e-05, "loss": 0.3076, "step": 980 }, { "epoch": 0.45, "learning_rate": 9.997979558649615e-05, "loss": 0.2972, "step": 990 }, { "epoch": 0.46, "learning_rate": 9.99793853832534e-05, "loss": 0.24, "step": 1000 }, { "epoch": 0.46, "learning_rate": 9.997897105850395e-05, "loss": 0.2688, "step": 1010 }, { "epoch": 0.47, "learning_rate": 9.997855261228197e-05, "loss": 0.2663, "step": 1020 }, { "epoch": 0.47, "learning_rate": 9.997813004462197e-05, "loss": 0.2545, "step": 1030 }, { "epoch": 0.48, "learning_rate": 9.997770335555881e-05, "loss": 0.2625, "step": 1040 }, { "epoch": 0.48, "learning_rate": 9.997727254512766e-05, "loss": 0.2678, "step": 1050 }, { "epoch": 0.48, "learning_rate": 9.997683761336402e-05, "loss": 0.267, "step": 1060 }, { "epoch": 0.49, "learning_rate": 9.997639856030383e-05, "loss": 0.2511, "step": 1070 }, { "epoch": 0.49, "learning_rate": 9.997595538598323e-05, "loss": 0.2468, "step": 1080 }, { "epoch": 0.5, "learning_rate": 9.997550809043879e-05, "loss": 0.2556, "step": 1090 }, { "epoch": 0.5, "learning_rate": 9.997505667370738e-05, "loss": 0.2961, "step": 1100 }, { "epoch": 0.51, "learning_rate": 9.997460113582626e-05, "loss": 0.2985, "step": 1110 }, { "epoch": 0.51, "learning_rate": 9.997414147683298e-05, "loss": 0.276, "step": 1120 }, { "epoch": 0.52, "learning_rate": 9.99736776967654e-05, "loss": 0.271, "step": 1130 }, { "epoch": 0.52, "learning_rate": 9.997320979566185e-05, "loss": 0.2613, "step": 1140 }, { "epoch": 0.53, "learning_rate": 9.997273777356084e-05, "loss": 0.2647, "step": 1150 }, { "epoch": 0.53, "learning_rate": 9.997226163050135e-05, "loss": 0.2364, "step": 1160 }, { "epoch": 0.53, "learning_rate": 9.997178136652261e-05, "loss": 0.232, "step": 1170 }, { "epoch": 0.54, "learning_rate": 9.997129698166423e-05, "loss": 0.3073, "step": 1180 }, { "epoch": 0.54, "learning_rate": 9.997080847596615e-05, "loss": 0.3532, "step": 1190 }, { "epoch": 0.55, "learning_rate": 9.997031584946868e-05, "loss": 0.3047, "step": 1200 }, { "epoch": 0.55, "learning_rate": 9.996981910221244e-05, "loss": 0.3058, "step": 1210 }, { "epoch": 0.56, "learning_rate": 9.996931823423836e-05, "loss": 0.2431, "step": 1220 }, { "epoch": 0.56, "learning_rate": 9.996881324558775e-05, "loss": 0.2243, "step": 1230 }, { "epoch": 0.57, "learning_rate": 9.99683041363023e-05, "loss": 0.2137, "step": 1240 }, { "epoch": 0.57, "learning_rate": 9.996779090642396e-05, "loss": 0.2868, "step": 1250 }, { "epoch": 0.58, "learning_rate": 9.996727355599504e-05, "loss": 0.2503, "step": 1260 }, { "epoch": 0.58, "learning_rate": 9.996675208505821e-05, "loss": 0.2567, "step": 1270 }, { "epoch": 0.59, "learning_rate": 9.99662264936565e-05, "loss": 0.2623, "step": 1280 }, { "epoch": 0.59, "learning_rate": 9.996569678183322e-05, "loss": 0.2723, "step": 1290 }, { "epoch": 0.59, "learning_rate": 9.996516294963207e-05, "loss": 0.246, "step": 1300 }, { "epoch": 0.6, "learning_rate": 9.996462499709707e-05, "loss": 0.2371, "step": 1310 }, { "epoch": 0.6, "learning_rate": 9.996408292427257e-05, "loss": 0.2618, "step": 1320 }, { "epoch": 0.61, "learning_rate": 9.996353673120328e-05, "loss": 0.2821, "step": 1330 }, { "epoch": 0.61, "learning_rate": 9.996298641793425e-05, "loss": 0.2404, "step": 1340 }, { "epoch": 0.62, "learning_rate": 9.996243198451084e-05, "loss": 0.241, "step": 1350 }, { "epoch": 0.62, "learning_rate": 9.996187343097879e-05, "loss": 0.2728, "step": 1360 }, { "epoch": 0.63, "learning_rate": 9.996131075738415e-05, "loss": 0.2517, "step": 1370 }, { "epoch": 0.63, "learning_rate": 9.996074396377333e-05, "loss": 0.2646, "step": 1380 }, { "epoch": 0.64, "learning_rate": 9.996017305019307e-05, "loss": 0.2605, "step": 1390 }, { "epoch": 0.64, "learning_rate": 9.995959801669043e-05, "loss": 0.2543, "step": 1400 }, { "epoch": 0.64, "learning_rate": 9.995901886331286e-05, "loss": 0.2525, "step": 1410 }, { "epoch": 0.65, "learning_rate": 9.995843559010809e-05, "loss": 0.2465, "step": 1420 }, { "epoch": 0.65, "learning_rate": 9.995784819712425e-05, "loss": 0.2588, "step": 1430 }, { "epoch": 0.66, "learning_rate": 9.995725668440974e-05, "loss": 0.2319, "step": 1440 }, { "epoch": 0.66, "learning_rate": 9.995666105201338e-05, "loss": 0.2383, "step": 1450 }, { "epoch": 0.67, "learning_rate": 9.995606129998424e-05, "loss": 0.2481, "step": 1460 }, { "epoch": 0.67, "learning_rate": 9.995545742837185e-05, "loss": 0.2631, "step": 1470 }, { "epoch": 0.68, "learning_rate": 9.995484943722593e-05, "loss": 0.2463, "step": 1480 }, { "epoch": 0.68, "learning_rate": 9.995423732659666e-05, "loss": 0.2894, "step": 1490 }, { "epoch": 0.69, "learning_rate": 9.995362109653451e-05, "loss": 0.2391, "step": 1500 }, { "epoch": 0.69, "learning_rate": 9.995300074709028e-05, "loss": 0.2717, "step": 1510 }, { "epoch": 0.69, "learning_rate": 9.995237627831517e-05, "loss": 0.2448, "step": 1520 }, { "epoch": 0.7, "learning_rate": 9.995174769026064e-05, "loss": 0.2504, "step": 1530 }, { "epoch": 0.7, "learning_rate": 9.995111498297853e-05, "loss": 0.2241, "step": 1540 }, { "epoch": 0.71, "learning_rate": 9.995047815652101e-05, "loss": 0.2323, "step": 1550 }, { "epoch": 0.71, "learning_rate": 9.994983721094062e-05, "loss": 0.2682, "step": 1560 }, { "epoch": 0.72, "learning_rate": 9.99491921462902e-05, "loss": 0.2812, "step": 1570 }, { "epoch": 0.72, "learning_rate": 9.994854296262295e-05, "loss": 0.2563, "step": 1580 }, { "epoch": 0.73, "learning_rate": 9.994788965999237e-05, "loss": 0.2142, "step": 1590 }, { "epoch": 0.73, "learning_rate": 9.99472322384524e-05, "loss": 0.2317, "step": 1600 }, { "epoch": 0.74, "learning_rate": 9.994657069805719e-05, "loss": 0.2254, "step": 1610 }, { "epoch": 0.74, "learning_rate": 9.994590503886134e-05, "loss": 0.2152, "step": 1620 }, { "epoch": 0.74, "learning_rate": 9.994523526091972e-05, "loss": 0.2473, "step": 1630 }, { "epoch": 0.75, "learning_rate": 9.994456136428756e-05, "loss": 0.265, "step": 1640 }, { "epoch": 0.75, "learning_rate": 9.994388334902045e-05, "loss": 0.2941, "step": 1650 }, { "epoch": 0.76, "learning_rate": 9.994320121517427e-05, "loss": 0.2851, "step": 1660 }, { "epoch": 0.76, "learning_rate": 9.994251496280531e-05, "loss": 0.2662, "step": 1670 }, { "epoch": 0.77, "learning_rate": 9.994182459197013e-05, "loss": 0.2998, "step": 1680 }, { "epoch": 0.77, "learning_rate": 9.994113010272568e-05, "loss": 0.2606, "step": 1690 }, { "epoch": 0.78, "learning_rate": 9.994043149512924e-05, "loss": 0.2471, "step": 1700 }, { "epoch": 0.78, "learning_rate": 9.993972876923839e-05, "loss": 0.2555, "step": 1710 }, { "epoch": 0.79, "learning_rate": 9.993902192511109e-05, "loss": 0.2468, "step": 1720 }, { "epoch": 0.79, "learning_rate": 9.993831096280564e-05, "loss": 0.2195, "step": 1730 }, { "epoch": 0.8, "learning_rate": 9.993759588238066e-05, "loss": 0.2146, "step": 1740 }, { "epoch": 0.8, "learning_rate": 9.993687668389513e-05, "loss": 0.2181, "step": 1750 }, { "epoch": 0.8, "learning_rate": 9.993615336740833e-05, "loss": 0.2789, "step": 1760 }, { "epoch": 0.81, "learning_rate": 9.993542593297994e-05, "loss": 0.2274, "step": 1770 }, { "epoch": 0.81, "learning_rate": 9.993469438066993e-05, "loss": 0.2182, "step": 1780 }, { "epoch": 0.82, "learning_rate": 9.993395871053863e-05, "loss": 0.2587, "step": 1790 }, { "epoch": 0.82, "learning_rate": 9.993321892264671e-05, "loss": 0.2175, "step": 1800 }, { "epoch": 0.83, "learning_rate": 9.993247501705518e-05, "loss": 0.2101, "step": 1810 }, { "epoch": 0.83, "learning_rate": 9.993172699382539e-05, "loss": 0.2165, "step": 1820 }, { "epoch": 0.84, "learning_rate": 9.993097485301899e-05, "loss": 0.2768, "step": 1830 }, { "epoch": 0.84, "learning_rate": 9.993021859469802e-05, "loss": 0.2241, "step": 1840 }, { "epoch": 0.85, "learning_rate": 9.992945821892488e-05, "loss": 0.2265, "step": 1850 }, { "epoch": 0.85, "learning_rate": 9.992869372576223e-05, "loss": 0.2196, "step": 1860 }, { "epoch": 0.85, "learning_rate": 9.992792511527313e-05, "loss": 0.2159, "step": 1870 }, { "epoch": 0.86, "learning_rate": 9.992715238752097e-05, "loss": 0.2166, "step": 1880 }, { "epoch": 0.86, "learning_rate": 9.992637554256946e-05, "loss": 0.2495, "step": 1890 }, { "epoch": 0.87, "learning_rate": 9.992559458048267e-05, "loss": 0.2391, "step": 1900 }, { "epoch": 0.87, "learning_rate": 9.9924809501325e-05, "loss": 0.2636, "step": 1910 }, { "epoch": 0.88, "learning_rate": 9.992402030516117e-05, "loss": 0.224, "step": 1920 }, { "epoch": 0.88, "learning_rate": 9.99232269920563e-05, "loss": 0.2646, "step": 1930 }, { "epoch": 0.89, "learning_rate": 9.992242956207578e-05, "loss": 0.2651, "step": 1940 }, { "epoch": 0.89, "learning_rate": 9.992162801528539e-05, "loss": 0.2503, "step": 1950 }, { "epoch": 0.9, "learning_rate": 9.99208223517512e-05, "loss": 0.2242, "step": 1960 }, { "epoch": 0.9, "learning_rate": 9.992001257153968e-05, "loss": 0.2446, "step": 1970 }, { "epoch": 0.9, "learning_rate": 9.991919867471759e-05, "loss": 0.2392, "step": 1980 }, { "epoch": 0.91, "learning_rate": 9.991838066135204e-05, "loss": 0.2312, "step": 1990 }, { "epoch": 0.91, "learning_rate": 9.991755853151051e-05, "loss": 0.2424, "step": 2000 }, { "epoch": 0.92, "learning_rate": 9.991673228526079e-05, "loss": 0.2144, "step": 2010 }, { "epoch": 0.92, "learning_rate": 9.9915901922671e-05, "loss": 0.2344, "step": 2020 }, { "epoch": 0.93, "learning_rate": 9.991506744380962e-05, "loss": 0.2234, "step": 2030 }, { "epoch": 0.93, "learning_rate": 9.991422884874547e-05, "loss": 0.2633, "step": 2040 }, { "epoch": 0.94, "learning_rate": 9.991338613754771e-05, "loss": 0.2156, "step": 2050 }, { "epoch": 0.94, "learning_rate": 9.991253931028582e-05, "loss": 0.2644, "step": 2060 }, { "epoch": 0.95, "learning_rate": 9.991168836702964e-05, "loss": 0.3146, "step": 2070 }, { "epoch": 0.95, "learning_rate": 9.991083330784934e-05, "loss": 0.2395, "step": 2080 }, { "epoch": 1.0, "learning_rate": 9.990997413281544e-05, "loss": 0.332, "step": 2090 }, { "epoch": 1.01, "learning_rate": 9.990911084199878e-05, "loss": 0.2398, "step": 2100 }, { "epoch": 1.01, "learning_rate": 9.990824343547056e-05, "loss": 0.244, "step": 2110 }, { "epoch": 1.02, "learning_rate": 9.990737191330231e-05, "loss": 0.2374, "step": 2120 }, { "epoch": 1.02, "learning_rate": 9.990649627556589e-05, "loss": 0.203, "step": 2130 }, { "epoch": 1.03, "learning_rate": 9.990561652233351e-05, "loss": 0.2499, "step": 2140 }, { "epoch": 1.03, "learning_rate": 9.990473265367773e-05, "loss": 0.2377, "step": 2150 }, { "epoch": 1.04, "learning_rate": 9.990384466967141e-05, "loss": 0.2706, "step": 2160 }, { "epoch": 1.04, "learning_rate": 9.990295257038781e-05, "loss": 0.2566, "step": 2170 }, { "epoch": 1.04, "learning_rate": 9.990205635590049e-05, "loss": 0.2415, "step": 2180 }, { "epoch": 1.05, "learning_rate": 9.990115602628333e-05, "loss": 0.2242, "step": 2190 }, { "epoch": 1.05, "learning_rate": 9.99002515816106e-05, "loss": 0.242, "step": 2200 }, { "epoch": 1.06, "learning_rate": 9.989934302195687e-05, "loss": 0.2293, "step": 2210 }, { "epoch": 1.06, "learning_rate": 9.989843034739708e-05, "loss": 0.2294, "step": 2220 }, { "epoch": 1.07, "learning_rate": 9.989751355800647e-05, "loss": 0.2325, "step": 2230 }, { "epoch": 1.07, "learning_rate": 9.989659265386066e-05, "loss": 0.2352, "step": 2240 }, { "epoch": 1.08, "learning_rate": 9.98956676350356e-05, "loss": 0.2306, "step": 2250 }, { "epoch": 1.08, "learning_rate": 9.989473850160753e-05, "loss": 0.2157, "step": 2260 }, { "epoch": 1.09, "learning_rate": 9.989380525365313e-05, "loss": 0.2315, "step": 2270 }, { "epoch": 1.09, "learning_rate": 9.98928678912493e-05, "loss": 0.2024, "step": 2280 }, { "epoch": 1.09, "learning_rate": 9.989192641447338e-05, "loss": 0.2504, "step": 2290 }, { "epoch": 1.1, "learning_rate": 9.989098082340299e-05, "loss": 0.2356, "step": 2300 }, { "epoch": 1.1, "learning_rate": 9.989003111811612e-05, "loss": 0.2461, "step": 2310 }, { "epoch": 1.11, "learning_rate": 9.988907729869107e-05, "loss": 0.2505, "step": 2320 }, { "epoch": 1.11, "learning_rate": 9.98881193652065e-05, "loss": 0.2534, "step": 2330 }, { "epoch": 1.12, "learning_rate": 9.988715731774142e-05, "loss": 0.2466, "step": 2340 }, { "epoch": 1.12, "learning_rate": 9.988619115637514e-05, "loss": 0.2219, "step": 2350 }, { "epoch": 1.13, "learning_rate": 9.988522088118737e-05, "loss": 0.2135, "step": 2360 }, { "epoch": 1.13, "learning_rate": 9.988424649225808e-05, "loss": 0.1912, "step": 2370 }, { "epoch": 1.14, "learning_rate": 9.988326798966765e-05, "loss": 0.2519, "step": 2380 }, { "epoch": 1.14, "learning_rate": 9.988228537349676e-05, "loss": 0.2148, "step": 2390 }, { "epoch": 1.14, "learning_rate": 9.988129864382645e-05, "loss": 0.2458, "step": 2400 }, { "epoch": 1.15, "learning_rate": 9.988030780073808e-05, "loss": 0.2353, "step": 2410 }, { "epoch": 1.15, "learning_rate": 9.987931284431336e-05, "loss": 0.2457, "step": 2420 }, { "epoch": 1.16, "learning_rate": 9.987831377463435e-05, "loss": 0.221, "step": 2430 }, { "epoch": 1.16, "learning_rate": 9.987731059178342e-05, "loss": 0.2248, "step": 2440 }, { "epoch": 1.17, "learning_rate": 9.98763032958433e-05, "loss": 0.2064, "step": 2450 }, { "epoch": 1.17, "learning_rate": 9.987529188689707e-05, "loss": 0.2526, "step": 2460 }, { "epoch": 1.18, "learning_rate": 9.987427636502812e-05, "loss": 0.2236, "step": 2470 }, { "epoch": 1.18, "learning_rate": 9.987325673032021e-05, "loss": 0.2425, "step": 2480 }, { "epoch": 1.19, "learning_rate": 9.987223298285742e-05, "loss": 0.1952, "step": 2490 }, { "epoch": 1.19, "learning_rate": 9.987120512272414e-05, "loss": 0.2801, "step": 2500 }, { "epoch": 1.2, "learning_rate": 9.987017315000518e-05, "loss": 0.2219, "step": 2510 }, { "epoch": 1.2, "learning_rate": 9.98691370647856e-05, "loss": 0.2442, "step": 2520 }, { "epoch": 1.2, "learning_rate": 9.986809686715088e-05, "loss": 0.1976, "step": 2530 }, { "epoch": 1.21, "learning_rate": 9.986705255718676e-05, "loss": 0.2145, "step": 2540 }, { "epoch": 1.21, "learning_rate": 9.986600413497938e-05, "loss": 0.2266, "step": 2550 }, { "epoch": 1.22, "learning_rate": 9.986495160061519e-05, "loss": 0.2359, "step": 2560 }, { "epoch": 1.22, "learning_rate": 9.9863894954181e-05, "loss": 0.2471, "step": 2570 }, { "epoch": 1.23, "learning_rate": 9.986283419576393e-05, "loss": 0.245, "step": 2580 }, { "epoch": 1.23, "learning_rate": 9.986176932545145e-05, "loss": 0.2728, "step": 2590 }, { "epoch": 1.24, "learning_rate": 9.986070034333139e-05, "loss": 0.2405, "step": 2600 }, { "epoch": 1.24, "learning_rate": 9.98596272494919e-05, "loss": 0.2204, "step": 2610 }, { "epoch": 1.25, "learning_rate": 9.985855004402144e-05, "loss": 0.2311, "step": 2620 }, { "epoch": 1.25, "learning_rate": 9.98574687270089e-05, "loss": 0.2496, "step": 2630 }, { "epoch": 1.25, "learning_rate": 9.98563832985434e-05, "loss": 0.2299, "step": 2640 }, { "epoch": 1.26, "learning_rate": 9.985529375871447e-05, "loss": 0.2347, "step": 2650 }, { "epoch": 1.26, "learning_rate": 9.985420010761195e-05, "loss": 0.2152, "step": 2660 }, { "epoch": 1.27, "learning_rate": 9.985310234532603e-05, "loss": 0.235, "step": 2670 }, { "epoch": 1.27, "learning_rate": 9.985200047194725e-05, "loss": 0.2585, "step": 2680 }, { "epoch": 1.28, "learning_rate": 9.985089448756646e-05, "loss": 0.2401, "step": 2690 }, { "epoch": 1.28, "learning_rate": 9.984978439227486e-05, "loss": 0.2324, "step": 2700 }, { "epoch": 1.29, "learning_rate": 9.984867018616401e-05, "loss": 0.2191, "step": 2710 }, { "epoch": 1.29, "learning_rate": 9.984755186932577e-05, "loss": 0.2153, "step": 2720 }, { "epoch": 1.3, "learning_rate": 9.984642944185238e-05, "loss": 0.2378, "step": 2730 }, { "epoch": 1.3, "learning_rate": 9.984530290383638e-05, "loss": 0.2228, "step": 2740 }, { "epoch": 1.3, "learning_rate": 9.98441722553707e-05, "loss": 0.225, "step": 2750 }, { "epoch": 1.31, "learning_rate": 9.984303749654854e-05, "loss": 0.2563, "step": 2760 }, { "epoch": 1.31, "learning_rate": 9.98418986274635e-05, "loss": 0.211, "step": 2770 }, { "epoch": 1.32, "learning_rate": 9.984075564820951e-05, "loss": 0.2169, "step": 2780 }, { "epoch": 1.32, "learning_rate": 9.98396085588808e-05, "loss": 0.2477, "step": 2790 }, { "epoch": 1.33, "learning_rate": 9.983845735957196e-05, "loss": 0.2521, "step": 2800 }, { "epoch": 1.33, "learning_rate": 9.983730205037794e-05, "loss": 0.2348, "step": 2810 }, { "epoch": 1.34, "learning_rate": 9.9836142631394e-05, "loss": 0.225, "step": 2820 }, { "epoch": 1.34, "learning_rate": 9.983497910271575e-05, "loss": 0.2448, "step": 2830 }, { "epoch": 1.35, "learning_rate": 9.983381146443916e-05, "loss": 0.2186, "step": 2840 }, { "epoch": 1.35, "learning_rate": 9.98326397166605e-05, "loss": 0.2348, "step": 2850 }, { "epoch": 1.36, "learning_rate": 9.98314638594764e-05, "loss": 0.2333, "step": 2860 }, { "epoch": 1.36, "learning_rate": 9.983028389298383e-05, "loss": 0.2449, "step": 2870 }, { "epoch": 1.36, "learning_rate": 9.982909981728007e-05, "loss": 0.223, "step": 2880 }, { "epoch": 1.37, "learning_rate": 9.982791163246282e-05, "loss": 0.2203, "step": 2890 }, { "epoch": 1.37, "learning_rate": 9.982671933862999e-05, "loss": 0.2253, "step": 2900 }, { "epoch": 1.38, "learning_rate": 9.982552293587997e-05, "loss": 0.2265, "step": 2910 }, { "epoch": 1.38, "learning_rate": 9.982432242431137e-05, "loss": 0.231, "step": 2920 }, { "epoch": 1.39, "learning_rate": 9.982311780402322e-05, "loss": 0.2362, "step": 2930 }, { "epoch": 1.39, "learning_rate": 9.982190907511483e-05, "loss": 0.227, "step": 2940 }, { "epoch": 1.4, "learning_rate": 9.98206962376859e-05, "loss": 0.2477, "step": 2950 }, { "epoch": 1.4, "learning_rate": 9.981947929183646e-05, "loss": 0.2403, "step": 2960 }, { "epoch": 1.41, "learning_rate": 9.981825823766682e-05, "loss": 0.2165, "step": 2970 }, { "epoch": 1.41, "learning_rate": 9.98170330752777e-05, "loss": 0.2535, "step": 2980 }, { "epoch": 1.41, "learning_rate": 9.981580380477013e-05, "loss": 0.2516, "step": 2990 }, { "epoch": 1.42, "learning_rate": 9.98145704262455e-05, "loss": 0.2004, "step": 3000 }, { "epoch": 1.42, "learning_rate": 9.981333293980547e-05, "loss": 0.2386, "step": 3010 }, { "epoch": 1.43, "learning_rate": 9.981209134555213e-05, "loss": 0.2128, "step": 3020 }, { "epoch": 1.43, "learning_rate": 9.981084564358785e-05, "loss": 0.2367, "step": 3030 }, { "epoch": 1.44, "learning_rate": 9.980959583401535e-05, "loss": 0.2377, "step": 3040 }, { "epoch": 1.44, "learning_rate": 9.980834191693772e-05, "loss": 0.23, "step": 3050 }, { "epoch": 1.45, "learning_rate": 9.980708389245834e-05, "loss": 0.2205, "step": 3060 }, { "epoch": 1.45, "learning_rate": 9.980582176068096e-05, "loss": 0.2482, "step": 3070 }, { "epoch": 1.46, "learning_rate": 9.980455552170967e-05, "loss": 0.2204, "step": 3080 }, { "epoch": 1.46, "learning_rate": 9.980328517564886e-05, "loss": 0.2401, "step": 3090 }, { "epoch": 1.46, "learning_rate": 9.980201072260332e-05, "loss": 0.2302, "step": 3100 }, { "epoch": 1.47, "learning_rate": 9.980073216267813e-05, "loss": 0.2207, "step": 3110 }, { "epoch": 1.47, "learning_rate": 9.979944949597873e-05, "loss": 0.2364, "step": 3120 }, { "epoch": 1.48, "learning_rate": 9.97981627226109e-05, "loss": 0.2215, "step": 3130 }, { "epoch": 1.48, "learning_rate": 9.979687184268074e-05, "loss": 0.2272, "step": 3140 }, { "epoch": 1.49, "learning_rate": 9.979557685629471e-05, "loss": 0.2427, "step": 3150 }, { "epoch": 1.49, "learning_rate": 9.979427776355959e-05, "loss": 0.2196, "step": 3160 }, { "epoch": 1.5, "learning_rate": 9.979297456458251e-05, "loss": 0.1964, "step": 3170 }, { "epoch": 1.5, "learning_rate": 9.979166725947095e-05, "loss": 0.234, "step": 3180 }, { "epoch": 1.51, "learning_rate": 9.979035584833271e-05, "loss": 0.2214, "step": 3190 }, { "epoch": 1.51, "learning_rate": 9.978904033127593e-05, "loss": 0.2724, "step": 3200 }, { "epoch": 1.52, "learning_rate": 9.978772070840909e-05, "loss": 0.2236, "step": 3210 }, { "epoch": 1.52, "learning_rate": 9.978639697984102e-05, "loss": 0.2374, "step": 3220 }, { "epoch": 1.52, "learning_rate": 9.978506914568089e-05, "loss": 0.2356, "step": 3230 }, { "epoch": 1.53, "learning_rate": 9.978373720603818e-05, "loss": 0.1804, "step": 3240 }, { "epoch": 1.53, "learning_rate": 9.978240116102274e-05, "loss": 0.2171, "step": 3250 }, { "epoch": 1.54, "learning_rate": 9.978106101074473e-05, "loss": 0.219, "step": 3260 }, { "epoch": 1.54, "learning_rate": 9.977971675531467e-05, "loss": 0.2228, "step": 3270 }, { "epoch": 1.55, "learning_rate": 9.977836839484342e-05, "loss": 0.2449, "step": 3280 }, { "epoch": 1.55, "learning_rate": 9.977701592944218e-05, "loss": 0.2237, "step": 3290 }, { "epoch": 1.56, "learning_rate": 9.977565935922244e-05, "loss": 0.2318, "step": 3300 }, { "epoch": 1.56, "learning_rate": 9.977429868429612e-05, "loss": 0.2143, "step": 3310 }, { "epoch": 1.57, "learning_rate": 9.97729339047754e-05, "loss": 0.2313, "step": 3320 }, { "epoch": 1.57, "learning_rate": 9.977156502077282e-05, "loss": 0.2177, "step": 3330 }, { "epoch": 1.57, "learning_rate": 9.977019203240128e-05, "loss": 0.2289, "step": 3340 }, { "epoch": 1.58, "learning_rate": 9.976881493977399e-05, "loss": 0.2386, "step": 3350 }, { "epoch": 1.58, "learning_rate": 9.976743374300451e-05, "loss": 0.2194, "step": 3360 }, { "epoch": 1.59, "learning_rate": 9.976604844220675e-05, "loss": 0.1935, "step": 3370 }, { "epoch": 1.59, "learning_rate": 9.976465903749495e-05, "loss": 0.218, "step": 3380 }, { "epoch": 1.6, "learning_rate": 9.976326552898366e-05, "loss": 0.2208, "step": 3390 }, { "epoch": 1.6, "learning_rate": 9.976186791678782e-05, "loss": 0.2231, "step": 3400 }, { "epoch": 1.61, "learning_rate": 9.976046620102269e-05, "loss": 0.232, "step": 3410 }, { "epoch": 1.61, "learning_rate": 9.975906038180382e-05, "loss": 0.2272, "step": 3420 }, { "epoch": 1.62, "learning_rate": 9.975765045924719e-05, "loss": 0.1815, "step": 3430 }, { "epoch": 1.62, "learning_rate": 9.975623643346903e-05, "loss": 0.2043, "step": 3440 }, { "epoch": 1.62, "learning_rate": 9.975481830458598e-05, "loss": 0.1861, "step": 3450 }, { "epoch": 1.63, "learning_rate": 9.975339607271494e-05, "loss": 0.2283, "step": 3460 }, { "epoch": 1.63, "learning_rate": 9.975196973797323e-05, "loss": 0.2127, "step": 3470 }, { "epoch": 1.64, "learning_rate": 9.975053930047846e-05, "loss": 0.2147, "step": 3480 }, { "epoch": 1.64, "learning_rate": 9.97491047603486e-05, "loss": 0.2187, "step": 3490 }, { "epoch": 1.65, "learning_rate": 9.974766611770192e-05, "loss": 0.2198, "step": 3500 }, { "epoch": 1.65, "learning_rate": 9.974622337265709e-05, "loss": 0.2034, "step": 3510 }, { "epoch": 1.66, "learning_rate": 9.974477652533305e-05, "loss": 0.2323, "step": 3520 }, { "epoch": 1.66, "learning_rate": 9.974332557584916e-05, "loss": 0.2115, "step": 3530 }, { "epoch": 1.67, "learning_rate": 9.974187052432502e-05, "loss": 0.2574, "step": 3540 }, { "epoch": 1.67, "learning_rate": 9.974041137088066e-05, "loss": 0.2195, "step": 3550 }, { "epoch": 1.68, "learning_rate": 9.973894811563637e-05, "loss": 0.2228, "step": 3560 }, { "epoch": 1.68, "learning_rate": 9.973748075871285e-05, "loss": 0.1965, "step": 3570 }, { "epoch": 1.68, "learning_rate": 9.97360093002311e-05, "loss": 0.2062, "step": 3580 }, { "epoch": 1.69, "learning_rate": 9.973453374031244e-05, "loss": 0.1964, "step": 3590 }, { "epoch": 1.69, "learning_rate": 9.973305407907855e-05, "loss": 0.2514, "step": 3600 }, { "epoch": 1.7, "learning_rate": 9.97315703166515e-05, "loss": 0.1983, "step": 3610 }, { "epoch": 1.7, "learning_rate": 9.973008245315358e-05, "loss": 0.2034, "step": 3620 }, { "epoch": 1.71, "learning_rate": 9.972859048870754e-05, "loss": 0.2089, "step": 3630 }, { "epoch": 1.71, "learning_rate": 9.972709442343637e-05, "loss": 0.2204, "step": 3640 }, { "epoch": 1.72, "learning_rate": 9.972559425746347e-05, "loss": 0.3065, "step": 3650 }, { "epoch": 1.72, "learning_rate": 9.972408999091254e-05, "loss": 0.1818, "step": 3660 }, { "epoch": 1.73, "learning_rate": 9.972258162390764e-05, "loss": 0.2105, "step": 3670 }, { "epoch": 1.73, "learning_rate": 9.972106915657313e-05, "loss": 0.2146, "step": 3680 }, { "epoch": 1.73, "learning_rate": 9.971955258903376e-05, "loss": 0.1776, "step": 3690 }, { "epoch": 1.74, "learning_rate": 9.971803192141458e-05, "loss": 0.2166, "step": 3700 }, { "epoch": 1.74, "learning_rate": 9.971650715384098e-05, "loss": 0.2043, "step": 3710 }, { "epoch": 1.75, "learning_rate": 9.971497828643873e-05, "loss": 0.1909, "step": 3720 }, { "epoch": 1.75, "learning_rate": 9.971344531933388e-05, "loss": 0.2476, "step": 3730 }, { "epoch": 1.76, "learning_rate": 9.971190825265285e-05, "loss": 0.2039, "step": 3740 }, { "epoch": 1.76, "learning_rate": 9.97103670865224e-05, "loss": 0.2127, "step": 3750 }, { "epoch": 1.77, "learning_rate": 9.970882182106963e-05, "loss": 0.2287, "step": 3760 }, { "epoch": 1.77, "learning_rate": 9.970727245642192e-05, "loss": 0.2229, "step": 3770 }, { "epoch": 1.78, "learning_rate": 9.970571899270711e-05, "loss": 0.2371, "step": 3780 }, { "epoch": 1.78, "learning_rate": 9.970416143005325e-05, "loss": 0.2545, "step": 3790 }, { "epoch": 1.78, "learning_rate": 9.97025997685888e-05, "loss": 0.1963, "step": 3800 }, { "epoch": 1.79, "learning_rate": 9.970103400844255e-05, "loss": 0.2585, "step": 3810 }, { "epoch": 1.79, "learning_rate": 9.96994641497436e-05, "loss": 0.2514, "step": 3820 }, { "epoch": 1.8, "learning_rate": 9.969789019262143e-05, "loss": 0.2301, "step": 3830 }, { "epoch": 1.8, "learning_rate": 9.96963121372058e-05, "loss": 0.2062, "step": 3840 }, { "epoch": 1.81, "learning_rate": 9.96947299836269e-05, "loss": 0.223, "step": 3850 }, { "epoch": 1.81, "learning_rate": 9.969314373201515e-05, "loss": 0.2218, "step": 3860 }, { "epoch": 1.82, "learning_rate": 9.969155338250137e-05, "loss": 0.2013, "step": 3870 }, { "epoch": 1.82, "learning_rate": 9.968995893521672e-05, "loss": 0.255, "step": 3880 }, { "epoch": 1.83, "learning_rate": 9.968836039029267e-05, "loss": 0.2192, "step": 3890 }, { "epoch": 1.83, "learning_rate": 9.968675774786105e-05, "loss": 0.1972, "step": 3900 }, { "epoch": 1.84, "learning_rate": 9.968515100805402e-05, "loss": 0.2291, "step": 3910 }, { "epoch": 1.84, "learning_rate": 9.968354017100409e-05, "loss": 0.2161, "step": 3920 }, { "epoch": 1.84, "learning_rate": 9.968192523684407e-05, "loss": 0.2281, "step": 3930 }, { "epoch": 1.85, "learning_rate": 9.968030620570716e-05, "loss": 0.2409, "step": 3940 }, { "epoch": 1.85, "learning_rate": 9.967868307772687e-05, "loss": 0.2138, "step": 3950 }, { "epoch": 1.86, "learning_rate": 9.967705585303704e-05, "loss": 0.2066, "step": 3960 }, { "epoch": 1.86, "learning_rate": 9.967542453177183e-05, "loss": 0.2242, "step": 3970 }, { "epoch": 1.87, "learning_rate": 9.967378911406582e-05, "loss": 0.2029, "step": 3980 }, { "epoch": 1.87, "learning_rate": 9.967214960005384e-05, "loss": 0.258, "step": 3990 }, { "epoch": 1.88, "learning_rate": 9.967050598987111e-05, "loss": 0.2367, "step": 4000 }, { "epoch": 1.88, "learning_rate": 9.966885828365314e-05, "loss": 0.2442, "step": 4010 }, { "epoch": 1.89, "learning_rate": 9.966720648153584e-05, "loss": 0.2224, "step": 4020 }, { "epoch": 1.89, "learning_rate": 9.966555058365538e-05, "loss": 0.2447, "step": 4030 }, { "epoch": 1.89, "learning_rate": 9.966389059014838e-05, "loss": 0.2138, "step": 4040 }, { "epoch": 1.9, "learning_rate": 9.966222650115167e-05, "loss": 0.2218, "step": 4050 }, { "epoch": 1.9, "learning_rate": 9.96605583168025e-05, "loss": 0.2032, "step": 4060 }, { "epoch": 1.91, "learning_rate": 9.965888603723843e-05, "loss": 0.2202, "step": 4070 }, { "epoch": 1.91, "learning_rate": 9.965720966259737e-05, "loss": 0.2178, "step": 4080 }, { "epoch": 1.92, "learning_rate": 9.965552919301755e-05, "loss": 0.234, "step": 4090 }, { "epoch": 1.92, "learning_rate": 9.965384462863757e-05, "loss": 0.2202, "step": 4100 }, { "epoch": 1.93, "learning_rate": 9.965215596959633e-05, "loss": 0.2387, "step": 4110 }, { "epoch": 1.93, "learning_rate": 9.965046321603309e-05, "loss": 0.2214, "step": 4120 }, { "epoch": 1.94, "learning_rate": 9.964876636808742e-05, "loss": 0.2318, "step": 4130 }, { "epoch": 1.94, "learning_rate": 9.964706542589928e-05, "loss": 0.1761, "step": 4140 }, { "epoch": 1.94, "learning_rate": 9.964536038960891e-05, "loss": 0.2036, "step": 4150 }, { "epoch": 1.95, "learning_rate": 9.964365125935696e-05, "loss": 0.2154, "step": 4160 }, { "epoch": 2.0, "learning_rate": 9.96419380352843e-05, "loss": 0.2375, "step": 4170 }, { "epoch": 2.01, "learning_rate": 9.964022071753228e-05, "loss": 0.2061, "step": 4180 }, { "epoch": 2.01, "learning_rate": 9.963849930624247e-05, "loss": 0.2223, "step": 4190 }, { "epoch": 2.02, "learning_rate": 9.963677380155683e-05, "loss": 0.2365, "step": 4200 }, { "epoch": 2.02, "learning_rate": 9.963504420361767e-05, "loss": 0.206, "step": 4210 }, { "epoch": 2.02, "learning_rate": 9.963331051256763e-05, "loss": 0.1902, "step": 4220 }, { "epoch": 2.03, "learning_rate": 9.963157272854965e-05, "loss": 0.2061, "step": 4230 }, { "epoch": 2.03, "learning_rate": 9.962983085170705e-05, "loss": 0.1957, "step": 4240 }, { "epoch": 2.04, "learning_rate": 9.962808488218346e-05, "loss": 0.2072, "step": 4250 }, { "epoch": 2.04, "learning_rate": 9.962633482012286e-05, "loss": 0.1937, "step": 4260 }, { "epoch": 2.05, "learning_rate": 9.96245806656696e-05, "loss": 0.1871, "step": 4270 }, { "epoch": 2.05, "learning_rate": 9.962282241896828e-05, "loss": 0.2297, "step": 4280 }, { "epoch": 2.06, "learning_rate": 9.962106008016395e-05, "loss": 0.2081, "step": 4290 }, { "epoch": 2.06, "learning_rate": 9.96192936494019e-05, "loss": 0.1732, "step": 4300 }, { "epoch": 2.07, "learning_rate": 9.961752312682779e-05, "loss": 0.2134, "step": 4310 }, { "epoch": 2.07, "learning_rate": 9.961574851258767e-05, "loss": 0.2537, "step": 4320 }, { "epoch": 2.07, "learning_rate": 9.961396980682784e-05, "loss": 0.2014, "step": 4330 }, { "epoch": 2.08, "learning_rate": 9.9612187009695e-05, "loss": 0.1944, "step": 4340 }, { "epoch": 2.08, "learning_rate": 9.961040012133617e-05, "loss": 0.1923, "step": 4350 }, { "epoch": 2.09, "learning_rate": 9.960860914189869e-05, "loss": 0.2109, "step": 4360 }, { "epoch": 2.09, "learning_rate": 9.960681407153025e-05, "loss": 0.2245, "step": 4370 }, { "epoch": 2.1, "learning_rate": 9.960501491037889e-05, "loss": 0.2049, "step": 4380 }, { "epoch": 2.1, "learning_rate": 9.960321165859297e-05, "loss": 0.2066, "step": 4390 }, { "epoch": 2.11, "learning_rate": 9.960140431632121e-05, "loss": 0.1925, "step": 4400 }, { "epoch": 2.11, "learning_rate": 9.959959288371262e-05, "loss": 0.2483, "step": 4410 }, { "epoch": 2.12, "learning_rate": 9.95977773609166e-05, "loss": 0.2148, "step": 4420 }, { "epoch": 2.12, "learning_rate": 9.959595774808287e-05, "loss": 0.1774, "step": 4430 }, { "epoch": 2.13, "learning_rate": 9.959413404536146e-05, "loss": 0.2238, "step": 4440 }, { "epoch": 2.13, "learning_rate": 9.959230625290279e-05, "loss": 0.2239, "step": 4450 }, { "epoch": 2.13, "learning_rate": 9.959047437085755e-05, "loss": 0.2014, "step": 4460 }, { "epoch": 2.14, "learning_rate": 9.958863839937684e-05, "loss": 0.2041, "step": 4470 }, { "epoch": 2.14, "learning_rate": 9.958679833861205e-05, "loss": 0.2443, "step": 4480 }, { "epoch": 2.15, "learning_rate": 9.958495418871491e-05, "loss": 0.2008, "step": 4490 }, { "epoch": 2.15, "learning_rate": 9.958310594983751e-05, "loss": 0.218, "step": 4500 }, { "epoch": 2.16, "learning_rate": 9.958125362213224e-05, "loss": 0.1982, "step": 4510 }, { "epoch": 2.16, "learning_rate": 9.95793972057519e-05, "loss": 0.2058, "step": 4520 }, { "epoch": 2.17, "learning_rate": 9.95775367008495e-05, "loss": 0.2073, "step": 4530 }, { "epoch": 2.17, "learning_rate": 9.957567210757854e-05, "loss": 0.211, "step": 4540 }, { "epoch": 2.18, "learning_rate": 9.957380342609275e-05, "loss": 0.2115, "step": 4550 }, { "epoch": 2.18, "learning_rate": 9.957193065654623e-05, "loss": 0.1967, "step": 4560 }, { "epoch": 2.18, "learning_rate": 9.957005379909339e-05, "loss": 0.2222, "step": 4570 }, { "epoch": 2.19, "learning_rate": 9.956817285388906e-05, "loss": 0.2124, "step": 4580 }, { "epoch": 2.19, "learning_rate": 9.95662878210883e-05, "loss": 0.2151, "step": 4590 }, { "epoch": 2.2, "learning_rate": 9.95643987008466e-05, "loss": 0.2051, "step": 4600 }, { "epoch": 2.2, "learning_rate": 9.956250549331971e-05, "loss": 0.2002, "step": 4610 }, { "epoch": 2.21, "learning_rate": 9.956060819866376e-05, "loss": 0.2204, "step": 4620 }, { "epoch": 2.21, "learning_rate": 9.955870681703522e-05, "loss": 0.1963, "step": 4630 }, { "epoch": 2.22, "learning_rate": 9.955680134859088e-05, "loss": 0.2101, "step": 4640 }, { "epoch": 2.22, "learning_rate": 9.955489179348788e-05, "loss": 0.212, "step": 4650 }, { "epoch": 2.23, "learning_rate": 9.955297815188367e-05, "loss": 0.2311, "step": 4660 }, { "epoch": 2.23, "learning_rate": 9.955106042393607e-05, "loss": 0.2407, "step": 4670 }, { "epoch": 2.23, "learning_rate": 9.954913860980323e-05, "loss": 0.1916, "step": 4680 }, { "epoch": 2.24, "learning_rate": 9.954721270964363e-05, "loss": 0.2157, "step": 4690 }, { "epoch": 2.24, "learning_rate": 9.954528272361607e-05, "loss": 0.2049, "step": 4700 }, { "epoch": 2.25, "learning_rate": 9.954334865187971e-05, "loss": 0.2041, "step": 4710 }, { "epoch": 2.25, "learning_rate": 9.954141049459406e-05, "loss": 0.2246, "step": 4720 }, { "epoch": 2.26, "learning_rate": 9.953946825191893e-05, "loss": 0.2091, "step": 4730 }, { "epoch": 2.26, "learning_rate": 9.953752192401448e-05, "loss": 0.2108, "step": 4740 }, { "epoch": 2.27, "learning_rate": 9.953557151104124e-05, "loss": 0.2104, "step": 4750 }, { "epoch": 2.27, "learning_rate": 9.953361701316002e-05, "loss": 0.2362, "step": 4760 }, { "epoch": 2.28, "learning_rate": 9.953165843053201e-05, "loss": 0.2066, "step": 4770 }, { "epoch": 2.28, "learning_rate": 9.952969576331873e-05, "loss": 0.1937, "step": 4780 }, { "epoch": 2.29, "learning_rate": 9.952772901168202e-05, "loss": 0.2361, "step": 4790 }, { "epoch": 2.29, "learning_rate": 9.952575817578407e-05, "loss": 0.2141, "step": 4800 }, { "epoch": 2.29, "learning_rate": 9.952378325578738e-05, "loss": 0.2142, "step": 4810 }, { "epoch": 2.3, "learning_rate": 9.952180425185485e-05, "loss": 0.2282, "step": 4820 }, { "epoch": 2.3, "learning_rate": 9.951982116414965e-05, "loss": 0.2093, "step": 4830 }, { "epoch": 2.31, "learning_rate": 9.951783399283533e-05, "loss": 0.2028, "step": 4840 }, { "epoch": 2.31, "learning_rate": 9.951584273807573e-05, "loss": 0.2125, "step": 4850 }, { "epoch": 2.32, "learning_rate": 9.951384740003509e-05, "loss": 0.2355, "step": 4860 }, { "epoch": 2.32, "learning_rate": 9.951184797887796e-05, "loss": 0.1903, "step": 4870 }, { "epoch": 2.33, "learning_rate": 9.950984447476919e-05, "loss": 0.24, "step": 4880 }, { "epoch": 2.33, "learning_rate": 9.950783688787399e-05, "loss": 0.2385, "step": 4890 }, { "epoch": 2.34, "learning_rate": 9.950582521835797e-05, "loss": 0.231, "step": 4900 }, { "epoch": 2.34, "learning_rate": 9.950380946638696e-05, "loss": 0.2105, "step": 4910 }, { "epoch": 2.34, "learning_rate": 9.950178963212722e-05, "loss": 0.2276, "step": 4920 }, { "epoch": 2.35, "learning_rate": 9.94997657157453e-05, "loss": 0.2071, "step": 4930 }, { "epoch": 2.35, "learning_rate": 9.949773771740812e-05, "loss": 0.2306, "step": 4940 }, { "epoch": 2.36, "learning_rate": 9.94957056372829e-05, "loss": 0.2142, "step": 4950 }, { "epoch": 2.36, "learning_rate": 9.94936694755372e-05, "loss": 0.2191, "step": 4960 }, { "epoch": 2.37, "learning_rate": 9.949162923233896e-05, "loss": 0.1868, "step": 4970 }, { "epoch": 2.37, "learning_rate": 9.94895849078564e-05, "loss": 0.2347, "step": 4980 }, { "epoch": 2.38, "learning_rate": 9.948753650225814e-05, "loss": 0.2372, "step": 4990 }, { "epoch": 2.38, "learning_rate": 9.948548401571308e-05, "loss": 0.215, "step": 5000 }, { "epoch": 2.39, "learning_rate": 9.948342744839045e-05, "loss": 0.2589, "step": 5010 }, { "epoch": 2.39, "learning_rate": 9.948136680045987e-05, "loss": 0.2248, "step": 5020 }, { "epoch": 2.39, "learning_rate": 9.947930207209129e-05, "loss": 0.1913, "step": 5030 }, { "epoch": 2.4, "learning_rate": 9.947723326345492e-05, "loss": 0.213, "step": 5040 }, { "epoch": 2.4, "learning_rate": 9.947516037472142e-05, "loss": 0.1743, "step": 5050 }, { "epoch": 2.41, "learning_rate": 9.947308340606171e-05, "loss": 0.1833, "step": 5060 }, { "epoch": 2.41, "learning_rate": 9.947100235764705e-05, "loss": 0.2236, "step": 5070 }, { "epoch": 2.42, "learning_rate": 9.946891722964906e-05, "loss": 0.2029, "step": 5080 }, { "epoch": 2.42, "learning_rate": 9.946682802223968e-05, "loss": 0.2283, "step": 5090 }, { "epoch": 2.43, "learning_rate": 9.946473473559121e-05, "loss": 0.2212, "step": 5100 }, { "epoch": 2.43, "learning_rate": 9.946263736987628e-05, "loss": 0.2371, "step": 5110 }, { "epoch": 2.44, "learning_rate": 9.946053592526783e-05, "loss": 0.2091, "step": 5120 }, { "epoch": 2.44, "learning_rate": 9.945843040193914e-05, "loss": 0.204, "step": 5130 }, { "epoch": 2.45, "learning_rate": 9.945632080006387e-05, "loss": 0.2043, "step": 5140 }, { "epoch": 2.45, "learning_rate": 9.945420711981597e-05, "loss": 0.2144, "step": 5150 }, { "epoch": 2.45, "learning_rate": 9.945208936136975e-05, "loss": 0.2043, "step": 5160 }, { "epoch": 2.46, "learning_rate": 9.944996752489984e-05, "loss": 0.2552, "step": 5170 }, { "epoch": 2.46, "learning_rate": 9.944784161058123e-05, "loss": 0.2316, "step": 5180 }, { "epoch": 2.47, "learning_rate": 9.944571161858922e-05, "loss": 0.1913, "step": 5190 }, { "epoch": 2.47, "learning_rate": 9.944357754909945e-05, "loss": 0.199, "step": 5200 }, { "epoch": 2.48, "learning_rate": 9.944143940228792e-05, "loss": 0.197, "step": 5210 }, { "epoch": 2.48, "learning_rate": 9.943929717833094e-05, "loss": 0.2323, "step": 5220 }, { "epoch": 2.49, "learning_rate": 9.943715087740519e-05, "loss": 0.2242, "step": 5230 }, { "epoch": 2.49, "learning_rate": 9.943500049968764e-05, "loss": 0.2196, "step": 5240 }, { "epoch": 2.5, "learning_rate": 9.94328460453556e-05, "loss": 0.2015, "step": 5250 }, { "epoch": 2.5, "learning_rate": 9.943068751458678e-05, "loss": 0.2068, "step": 5260 }, { "epoch": 2.5, "learning_rate": 9.942852490755916e-05, "loss": 0.1984, "step": 5270 }, { "epoch": 2.51, "learning_rate": 9.942635822445107e-05, "loss": 0.1901, "step": 5280 }, { "epoch": 2.51, "learning_rate": 9.94241874654412e-05, "loss": 0.1998, "step": 5290 }, { "epoch": 2.52, "learning_rate": 9.942201263070854e-05, "loss": 0.195, "step": 5300 }, { "epoch": 2.52, "learning_rate": 9.941983372043246e-05, "loss": 0.2029, "step": 5310 }, { "epoch": 2.53, "learning_rate": 9.941765073479262e-05, "loss": 0.1713, "step": 5320 }, { "epoch": 2.53, "learning_rate": 9.941546367396905e-05, "loss": 0.1728, "step": 5330 }, { "epoch": 2.54, "learning_rate": 9.94132725381421e-05, "loss": 0.2145, "step": 5340 }, { "epoch": 2.54, "learning_rate": 9.941107732749246e-05, "loss": 0.2044, "step": 5350 }, { "epoch": 2.55, "learning_rate": 9.940887804220116e-05, "loss": 0.2158, "step": 5360 }, { "epoch": 2.55, "learning_rate": 9.940667468244955e-05, "loss": 0.199, "step": 5370 }, { "epoch": 2.55, "learning_rate": 9.940446724841934e-05, "loss": 0.2245, "step": 5380 }, { "epoch": 2.56, "learning_rate": 9.940225574029258e-05, "loss": 0.21, "step": 5390 }, { "epoch": 2.56, "learning_rate": 9.940004015825159e-05, "loss": 0.2102, "step": 5400 }, { "epoch": 2.57, "learning_rate": 9.939782050247915e-05, "loss": 0.1874, "step": 5410 }, { "epoch": 2.57, "learning_rate": 9.939559677315823e-05, "loss": 0.1885, "step": 5420 }, { "epoch": 2.58, "learning_rate": 9.939336897047225e-05, "loss": 0.2114, "step": 5430 }, { "epoch": 2.58, "learning_rate": 9.93911370946049e-05, "loss": 0.2171, "step": 5440 }, { "epoch": 2.59, "learning_rate": 9.938890114574024e-05, "loss": 0.2271, "step": 5450 }, { "epoch": 2.59, "learning_rate": 9.938666112406267e-05, "loss": 0.1944, "step": 5460 }, { "epoch": 2.6, "learning_rate": 9.938441702975689e-05, "loss": 0.1895, "step": 5470 }, { "epoch": 2.6, "learning_rate": 9.938216886300797e-05, "loss": 0.2112, "step": 5480 }, { "epoch": 2.61, "learning_rate": 9.93799166240013e-05, "loss": 0.2064, "step": 5490 }, { "epoch": 2.61, "learning_rate": 9.937766031292259e-05, "loss": 0.2003, "step": 5500 }, { "epoch": 2.61, "learning_rate": 9.937539992995794e-05, "loss": 0.2237, "step": 5510 }, { "epoch": 2.62, "learning_rate": 9.937313547529374e-05, "loss": 0.2054, "step": 5520 }, { "epoch": 2.62, "learning_rate": 9.937086694911671e-05, "loss": 0.1597, "step": 5530 }, { "epoch": 2.63, "learning_rate": 9.936859435161393e-05, "loss": 0.2307, "step": 5540 }, { "epoch": 2.63, "learning_rate": 9.936631768297281e-05, "loss": 0.2191, "step": 5550 }, { "epoch": 2.64, "learning_rate": 9.936403694338109e-05, "loss": 0.219, "step": 5560 }, { "epoch": 2.64, "learning_rate": 9.936175213302685e-05, "loss": 0.2068, "step": 5570 }, { "epoch": 2.65, "learning_rate": 9.935946325209851e-05, "loss": 0.1972, "step": 5580 }, { "epoch": 2.65, "learning_rate": 9.935717030078482e-05, "loss": 0.2154, "step": 5590 }, { "epoch": 2.66, "learning_rate": 9.935487327927486e-05, "loss": 0.1994, "step": 5600 }, { "epoch": 2.66, "learning_rate": 9.935257218775806e-05, "loss": 0.2102, "step": 5610 }, { "epoch": 2.66, "learning_rate": 9.935026702642417e-05, "loss": 0.1947, "step": 5620 }, { "epoch": 2.67, "learning_rate": 9.934795779546328e-05, "loss": 0.1856, "step": 5630 }, { "epoch": 2.67, "learning_rate": 9.934564449506584e-05, "loss": 0.2104, "step": 5640 }, { "epoch": 2.68, "learning_rate": 9.934332712542258e-05, "loss": 0.231, "step": 5650 }, { "epoch": 2.68, "learning_rate": 9.934100568672465e-05, "loss": 0.2195, "step": 5660 }, { "epoch": 2.69, "learning_rate": 9.933868017916342e-05, "loss": 0.203, "step": 5670 }, { "epoch": 2.69, "learning_rate": 9.933635060293072e-05, "loss": 0.1912, "step": 5680 }, { "epoch": 2.7, "learning_rate": 9.93340169582186e-05, "loss": 0.2114, "step": 5690 }, { "epoch": 2.7, "learning_rate": 9.933167924521956e-05, "loss": 0.2092, "step": 5700 }, { "epoch": 2.71, "learning_rate": 9.932933746412634e-05, "loss": 0.1812, "step": 5710 }, { "epoch": 2.71, "learning_rate": 9.932699161513208e-05, "loss": 0.175, "step": 5720 }, { "epoch": 2.71, "learning_rate": 9.93246416984302e-05, "loss": 0.1858, "step": 5730 }, { "epoch": 2.72, "learning_rate": 9.93222877142145e-05, "loss": 0.2168, "step": 5740 }, { "epoch": 2.72, "learning_rate": 9.931992966267909e-05, "loss": 0.1813, "step": 5750 }, { "epoch": 2.73, "learning_rate": 9.931756754401844e-05, "loss": 0.1956, "step": 5760 }, { "epoch": 2.73, "learning_rate": 9.931520135842733e-05, "loss": 0.2204, "step": 5770 }, { "epoch": 2.74, "learning_rate": 9.931283110610088e-05, "loss": 0.2017, "step": 5780 }, { "epoch": 2.74, "learning_rate": 9.931045678723457e-05, "loss": 0.183, "step": 5790 }, { "epoch": 2.75, "learning_rate": 9.930807840202416e-05, "loss": 0.1964, "step": 5800 }, { "epoch": 2.75, "learning_rate": 9.930569595066582e-05, "loss": 0.2064, "step": 5810 }, { "epoch": 2.76, "learning_rate": 9.9303309433356e-05, "loss": 0.1737, "step": 5820 }, { "epoch": 2.76, "learning_rate": 9.93009188502915e-05, "loss": 0.1786, "step": 5830 }, { "epoch": 2.77, "learning_rate": 9.929852420166945e-05, "loss": 0.2397, "step": 5840 }, { "epoch": 2.77, "learning_rate": 9.929612548768735e-05, "loss": 0.2325, "step": 5850 }, { "epoch": 2.77, "learning_rate": 9.929372270854298e-05, "loss": 0.1913, "step": 5860 }, { "epoch": 2.78, "learning_rate": 9.92913158644345e-05, "loss": 0.2118, "step": 5870 }, { "epoch": 2.78, "learning_rate": 9.928890495556037e-05, "loss": 0.2176, "step": 5880 }, { "epoch": 2.79, "learning_rate": 9.928648998211945e-05, "loss": 0.225, "step": 5890 }, { "epoch": 2.79, "learning_rate": 9.928407094431083e-05, "loss": 0.2123, "step": 5900 }, { "epoch": 2.8, "learning_rate": 9.928164784233402e-05, "loss": 0.1845, "step": 5910 }, { "epoch": 2.8, "learning_rate": 9.927922067638883e-05, "loss": 0.1896, "step": 5920 }, { "epoch": 2.81, "learning_rate": 9.927678944667545e-05, "loss": 0.2011, "step": 5930 }, { "epoch": 2.81, "learning_rate": 9.927435415339431e-05, "loss": 0.2094, "step": 5940 }, { "epoch": 2.82, "learning_rate": 9.927191479674628e-05, "loss": 0.1872, "step": 5950 }, { "epoch": 2.82, "learning_rate": 9.92694713769325e-05, "loss": 0.1802, "step": 5960 }, { "epoch": 2.82, "learning_rate": 9.926702389415447e-05, "loss": 0.1964, "step": 5970 }, { "epoch": 2.83, "learning_rate": 9.926457234861401e-05, "loss": 0.1885, "step": 5980 }, { "epoch": 2.83, "learning_rate": 9.92621167405133e-05, "loss": 0.1896, "step": 5990 }, { "epoch": 2.84, "learning_rate": 9.925965707005484e-05, "loss": 0.201, "step": 6000 }, { "epoch": 2.84, "learning_rate": 9.925719333744146e-05, "loss": 0.2044, "step": 6010 }, { "epoch": 2.85, "learning_rate": 9.925472554287633e-05, "loss": 0.218, "step": 6020 }, { "epoch": 2.85, "learning_rate": 9.925225368656294e-05, "loss": 0.2077, "step": 6030 }, { "epoch": 2.86, "learning_rate": 9.924977776870514e-05, "loss": 0.246, "step": 6040 }, { "epoch": 2.86, "learning_rate": 9.924729778950712e-05, "loss": 0.2392, "step": 6050 }, { "epoch": 2.87, "learning_rate": 9.924481374917337e-05, "loss": 0.2475, "step": 6060 }, { "epoch": 2.87, "learning_rate": 9.924232564790871e-05, "loss": 0.2034, "step": 6070 }, { "epoch": 2.87, "learning_rate": 9.923983348591839e-05, "loss": 0.1795, "step": 6080 }, { "epoch": 2.88, "learning_rate": 9.923733726340786e-05, "loss": 0.2314, "step": 6090 }, { "epoch": 2.88, "learning_rate": 9.9234836980583e-05, "loss": 0.184, "step": 6100 }, { "epoch": 2.89, "learning_rate": 9.923233263764997e-05, "loss": 0.2041, "step": 6110 }, { "epoch": 2.89, "learning_rate": 9.922982423481532e-05, "loss": 0.1704, "step": 6120 }, { "epoch": 2.9, "learning_rate": 9.922731177228588e-05, "loss": 0.2164, "step": 6130 }, { "epoch": 2.9, "learning_rate": 9.922479525026885e-05, "loss": 0.1713, "step": 6140 }, { "epoch": 2.91, "learning_rate": 9.922227466897174e-05, "loss": 0.1948, "step": 6150 }, { "epoch": 2.91, "learning_rate": 9.92197500286024e-05, "loss": 0.2009, "step": 6160 }, { "epoch": 2.92, "learning_rate": 9.921722132936905e-05, "loss": 0.1715, "step": 6170 }, { "epoch": 2.92, "learning_rate": 9.921468857148022e-05, "loss": 0.2055, "step": 6180 }, { "epoch": 2.93, "learning_rate": 9.921215175514474e-05, "loss": 0.2031, "step": 6190 }, { "epoch": 2.93, "learning_rate": 9.920961088057184e-05, "loss": 0.1882, "step": 6200 }, { "epoch": 2.93, "learning_rate": 9.920706594797102e-05, "loss": 0.1826, "step": 6210 }, { "epoch": 2.94, "learning_rate": 9.920451695755215e-05, "loss": 0.1978, "step": 6220 }, { "epoch": 2.94, "learning_rate": 9.920196390952545e-05, "loss": 0.197, "step": 6230 }, { "epoch": 2.95, "learning_rate": 9.919940680410144e-05, "loss": 0.2348, "step": 6240 }, { "epoch": 3.0, "learning_rate": 9.919684564149099e-05, "loss": 0.2213, "step": 6250 }, { "epoch": 3.01, "learning_rate": 9.91942804219053e-05, "loss": 0.2367, "step": 6260 }, { "epoch": 3.01, "learning_rate": 9.919171114555594e-05, "loss": 0.2178, "step": 6270 }, { "epoch": 3.01, "learning_rate": 9.918913781265475e-05, "loss": 0.1949, "step": 6280 }, { "epoch": 3.02, "learning_rate": 9.918656042341394e-05, "loss": 0.2142, "step": 6290 }, { "epoch": 3.02, "learning_rate": 9.918397897804607e-05, "loss": 0.1862, "step": 6300 }, { "epoch": 3.03, "learning_rate": 9.9181393476764e-05, "loss": 0.1988, "step": 6310 }, { "epoch": 3.03, "learning_rate": 9.917880391978094e-05, "loss": 0.1848, "step": 6320 }, { "epoch": 3.04, "learning_rate": 9.917621030731045e-05, "loss": 0.2015, "step": 6330 }, { "epoch": 3.04, "learning_rate": 9.91736126395664e-05, "loss": 0.2335, "step": 6340 }, { "epoch": 3.05, "learning_rate": 9.917101091676302e-05, "loss": 0.1915, "step": 6350 }, { "epoch": 3.05, "learning_rate": 9.916840513911483e-05, "loss": 0.1961, "step": 6360 }, { "epoch": 3.06, "learning_rate": 9.916579530683673e-05, "loss": 0.2076, "step": 6370 }, { "epoch": 3.06, "learning_rate": 9.916318142014393e-05, "loss": 0.2097, "step": 6380 }, { "epoch": 3.06, "learning_rate": 9.9160563479252e-05, "loss": 0.1721, "step": 6390 }, { "epoch": 3.07, "learning_rate": 9.915794148437683e-05, "loss": 0.2017, "step": 6400 }, { "epoch": 3.07, "learning_rate": 9.91553154357346e-05, "loss": 0.1888, "step": 6410 }, { "epoch": 3.08, "learning_rate": 9.91526853335419e-05, "loss": 0.1875, "step": 6420 }, { "epoch": 3.08, "learning_rate": 9.915005117801561e-05, "loss": 0.1756, "step": 6430 }, { "epoch": 3.09, "learning_rate": 9.914741296937295e-05, "loss": 0.22, "step": 6440 }, { "epoch": 3.09, "learning_rate": 9.914477070783148e-05, "loss": 0.1758, "step": 6450 }, { "epoch": 3.1, "learning_rate": 9.914212439360908e-05, "loss": 0.1829, "step": 6460 }, { "epoch": 3.1, "learning_rate": 9.913947402692401e-05, "loss": 0.2037, "step": 6470 }, { "epoch": 3.11, "learning_rate": 9.913681960799479e-05, "loss": 0.2057, "step": 6480 }, { "epoch": 3.11, "learning_rate": 9.913416113704035e-05, "loss": 0.1837, "step": 6490 }, { "epoch": 3.11, "learning_rate": 9.913149861427988e-05, "loss": 0.1799, "step": 6500 }, { "epoch": 3.12, "learning_rate": 9.912883203993297e-05, "loss": 0.2281, "step": 6510 }, { "epoch": 3.12, "learning_rate": 9.912616141421953e-05, "loss": 0.2102, "step": 6520 }, { "epoch": 3.13, "learning_rate": 9.912348673735975e-05, "loss": 0.217, "step": 6530 }, { "epoch": 3.13, "learning_rate": 9.912080800957422e-05, "loss": 0.1846, "step": 6540 }, { "epoch": 3.14, "learning_rate": 9.911812523108384e-05, "loss": 0.1868, "step": 6550 }, { "epoch": 3.14, "learning_rate": 9.911543840210984e-05, "loss": 0.2084, "step": 6560 }, { "epoch": 3.15, "learning_rate": 9.911274752287378e-05, "loss": 0.2007, "step": 6570 }, { "epoch": 3.15, "learning_rate": 9.911005259359756e-05, "loss": 0.2039, "step": 6580 }, { "epoch": 3.16, "learning_rate": 9.910735361450343e-05, "loss": 0.1926, "step": 6590 }, { "epoch": 3.16, "learning_rate": 9.910465058581395e-05, "loss": 0.21, "step": 6600 }, { "epoch": 3.16, "learning_rate": 9.910194350775201e-05, "loss": 0.2084, "step": 6610 }, { "epoch": 3.17, "learning_rate": 9.909923238054087e-05, "loss": 0.1907, "step": 6620 }, { "epoch": 3.17, "learning_rate": 9.909651720440409e-05, "loss": 0.1759, "step": 6630 }, { "epoch": 3.18, "learning_rate": 9.909379797956557e-05, "loss": 0.1892, "step": 6640 }, { "epoch": 3.18, "learning_rate": 9.909107470624956e-05, "loss": 0.2009, "step": 6650 }, { "epoch": 3.19, "learning_rate": 9.908834738468063e-05, "loss": 0.1691, "step": 6660 }, { "epoch": 3.19, "learning_rate": 9.908561601508369e-05, "loss": 0.1995, "step": 6670 }, { "epoch": 3.2, "learning_rate": 9.908288059768396e-05, "loss": 0.1824, "step": 6680 }, { "epoch": 3.2, "learning_rate": 9.908014113270702e-05, "loss": 0.1871, "step": 6690 }, { "epoch": 3.21, "learning_rate": 9.907739762037879e-05, "loss": 0.2005, "step": 6700 }, { "epoch": 3.21, "learning_rate": 9.90746500609255e-05, "loss": 0.1931, "step": 6710 }, { "epoch": 3.22, "learning_rate": 9.907189845457375e-05, "loss": 0.2458, "step": 6720 }, { "epoch": 3.22, "learning_rate": 9.906914280155039e-05, "loss": 0.2, "step": 6730 }, { "epoch": 3.22, "learning_rate": 9.906638310208273e-05, "loss": 0.1877, "step": 6740 }, { "epoch": 3.23, "learning_rate": 9.906361935639832e-05, "loss": 0.2303, "step": 6750 }, { "epoch": 3.23, "learning_rate": 9.906085156472507e-05, "loss": 0.2029, "step": 6760 }, { "epoch": 3.24, "learning_rate": 9.90580797272912e-05, "loss": 0.2042, "step": 6770 }, { "epoch": 3.24, "learning_rate": 9.905530384432533e-05, "loss": 0.192, "step": 6780 }, { "epoch": 3.25, "learning_rate": 9.905252391605633e-05, "loss": 0.2108, "step": 6790 }, { "epoch": 3.25, "learning_rate": 9.904973994271349e-05, "loss": 0.1847, "step": 6800 }, { "epoch": 3.26, "learning_rate": 9.904695192452633e-05, "loss": 0.1859, "step": 6810 }, { "epoch": 3.26, "learning_rate": 9.904415986172482e-05, "loss": 0.1743, "step": 6820 }, { "epoch": 3.27, "learning_rate": 9.904136375453916e-05, "loss": 0.1885, "step": 6830 }, { "epoch": 3.27, "learning_rate": 9.903856360319995e-05, "loss": 0.1991, "step": 6840 }, { "epoch": 3.27, "learning_rate": 9.90357594079381e-05, "loss": 0.1716, "step": 6850 }, { "epoch": 3.28, "learning_rate": 9.903295116898485e-05, "loss": 0.1882, "step": 6860 }, { "epoch": 3.28, "learning_rate": 9.90301388865718e-05, "loss": 0.2027, "step": 6870 }, { "epoch": 3.29, "learning_rate": 9.902732256093083e-05, "loss": 0.181, "step": 6880 }, { "epoch": 3.29, "learning_rate": 9.902450219229421e-05, "loss": 0.1931, "step": 6890 }, { "epoch": 3.3, "learning_rate": 9.90216777808945e-05, "loss": 0.2151, "step": 6900 }, { "epoch": 3.3, "learning_rate": 9.901884932696464e-05, "loss": 0.2104, "step": 6910 }, { "epoch": 3.31, "learning_rate": 9.901601683073784e-05, "loss": 0.1855, "step": 6920 }, { "epoch": 3.31, "learning_rate": 9.901318029244771e-05, "loss": 0.1903, "step": 6930 }, { "epoch": 3.32, "learning_rate": 9.901033971232816e-05, "loss": 0.2044, "step": 6940 }, { "epoch": 3.32, "learning_rate": 9.900749509061341e-05, "loss": 0.2086, "step": 6950 }, { "epoch": 3.32, "learning_rate": 9.900464642753806e-05, "loss": 0.1746, "step": 6960 }, { "epoch": 3.33, "learning_rate": 9.900179372333702e-05, "loss": 0.2157, "step": 6970 }, { "epoch": 3.33, "learning_rate": 9.899893697824554e-05, "loss": 0.1998, "step": 6980 }, { "epoch": 3.34, "learning_rate": 9.89960761924992e-05, "loss": 0.1979, "step": 6990 }, { "epoch": 3.34, "learning_rate": 9.899321136633389e-05, "loss": 0.1752, "step": 7000 }, { "epoch": 3.35, "learning_rate": 9.899034249998589e-05, "loss": 0.1943, "step": 7010 }, { "epoch": 3.35, "learning_rate": 9.898746959369174e-05, "loss": 0.1821, "step": 7020 }, { "epoch": 3.36, "learning_rate": 9.898459264768838e-05, "loss": 0.1902, "step": 7030 }, { "epoch": 3.36, "learning_rate": 9.898171166221307e-05, "loss": 0.1694, "step": 7040 }, { "epoch": 3.37, "learning_rate": 9.897882663750333e-05, "loss": 0.1952, "step": 7050 }, { "epoch": 3.37, "learning_rate": 9.897593757379712e-05, "loss": 0.2336, "step": 7060 }, { "epoch": 3.38, "learning_rate": 9.897304447133267e-05, "loss": 0.1914, "step": 7070 }, { "epoch": 3.38, "learning_rate": 9.897014733034855e-05, "loss": 0.1819, "step": 7080 }, { "epoch": 3.38, "learning_rate": 9.896724615108368e-05, "loss": 0.1624, "step": 7090 }, { "epoch": 3.39, "learning_rate": 9.896434093377731e-05, "loss": 0.2028, "step": 7100 }, { "epoch": 3.39, "learning_rate": 9.8961431678669e-05, "loss": 0.197, "step": 7110 }, { "epoch": 3.4, "learning_rate": 9.895851838599866e-05, "loss": 0.1723, "step": 7120 }, { "epoch": 3.4, "learning_rate": 9.895560105600654e-05, "loss": 0.2246, "step": 7130 }, { "epoch": 3.41, "learning_rate": 9.895267968893323e-05, "loss": 0.1822, "step": 7140 }, { "epoch": 3.41, "learning_rate": 9.894975428501961e-05, "loss": 0.1888, "step": 7150 }, { "epoch": 3.42, "learning_rate": 9.894682484450693e-05, "loss": 0.1986, "step": 7160 }, { "epoch": 3.42, "learning_rate": 9.894389136763677e-05, "loss": 0.1734, "step": 7170 }, { "epoch": 3.43, "learning_rate": 9.894095385465102e-05, "loss": 0.1853, "step": 7180 }, { "epoch": 3.43, "learning_rate": 9.893801230579194e-05, "loss": 0.2029, "step": 7190 }, { "epoch": 3.43, "learning_rate": 9.893506672130211e-05, "loss": 0.1891, "step": 7200 }, { "epoch": 3.44, "learning_rate": 9.893211710142439e-05, "loss": 0.1771, "step": 7210 }, { "epoch": 3.44, "learning_rate": 9.892916344640206e-05, "loss": 0.1977, "step": 7220 }, { "epoch": 3.45, "learning_rate": 9.892620575647869e-05, "loss": 0.2109, "step": 7230 }, { "epoch": 3.45, "learning_rate": 9.892324403189817e-05, "loss": 0.2187, "step": 7240 }, { "epoch": 3.46, "learning_rate": 9.892027827290471e-05, "loss": 0.1946, "step": 7250 }, { "epoch": 3.46, "learning_rate": 9.891730847974293e-05, "loss": 0.1965, "step": 7260 }, { "epoch": 3.47, "learning_rate": 9.89143346526577e-05, "loss": 0.1725, "step": 7270 }, { "epoch": 3.47, "learning_rate": 9.891135679189428e-05, "loss": 0.1874, "step": 7280 }, { "epoch": 3.48, "learning_rate": 9.890837489769818e-05, "loss": 0.2011, "step": 7290 }, { "epoch": 3.48, "learning_rate": 9.890538897031536e-05, "loss": 0.1841, "step": 7300 }, { "epoch": 3.48, "learning_rate": 9.890239900999204e-05, "loss": 0.175, "step": 7310 }, { "epoch": 3.49, "learning_rate": 9.889940501697475e-05, "loss": 0.1722, "step": 7320 }, { "epoch": 3.49, "learning_rate": 9.889640699151042e-05, "loss": 0.1901, "step": 7330 }, { "epoch": 3.5, "learning_rate": 9.889340493384628e-05, "loss": 0.1965, "step": 7340 }, { "epoch": 3.5, "learning_rate": 9.889039884422988e-05, "loss": 0.2163, "step": 7350 }, { "epoch": 3.51, "learning_rate": 9.88873887229091e-05, "loss": 0.1929, "step": 7360 }, { "epoch": 3.51, "learning_rate": 9.88843745701322e-05, "loss": 0.242, "step": 7370 }, { "epoch": 3.52, "learning_rate": 9.888135638614773e-05, "loss": 0.1987, "step": 7380 }, { "epoch": 3.52, "learning_rate": 9.887833417120455e-05, "loss": 0.2067, "step": 7390 }, { "epoch": 3.53, "learning_rate": 9.887530792555192e-05, "loss": 0.1919, "step": 7400 }, { "epoch": 3.53, "learning_rate": 9.887227764943941e-05, "loss": 0.1717, "step": 7410 }, { "epoch": 3.54, "learning_rate": 9.886924334311687e-05, "loss": 0.2081, "step": 7420 }, { "epoch": 3.54, "learning_rate": 9.886620500683453e-05, "loss": 0.1858, "step": 7430 }, { "epoch": 3.54, "learning_rate": 9.886316264084296e-05, "loss": 0.1944, "step": 7440 }, { "epoch": 3.55, "learning_rate": 9.886011624539303e-05, "loss": 0.1902, "step": 7450 }, { "epoch": 3.55, "learning_rate": 9.885706582073597e-05, "loss": 0.1895, "step": 7460 }, { "epoch": 3.56, "learning_rate": 9.885401136712332e-05, "loss": 0.1766, "step": 7470 }, { "epoch": 3.56, "learning_rate": 9.885095288480698e-05, "loss": 0.1832, "step": 7480 }, { "epoch": 3.57, "learning_rate": 9.884789037403914e-05, "loss": 0.1895, "step": 7490 }, { "epoch": 3.57, "learning_rate": 9.884482383507238e-05, "loss": 0.2084, "step": 7500 }, { "epoch": 3.58, "learning_rate": 9.884175326815954e-05, "loss": 0.2074, "step": 7510 }, { "epoch": 3.58, "learning_rate": 9.883867867355387e-05, "loss": 0.1816, "step": 7520 }, { "epoch": 3.59, "learning_rate": 9.883560005150889e-05, "loss": 0.1639, "step": 7530 }, { "epoch": 3.59, "learning_rate": 9.883251740227848e-05, "loss": 0.1725, "step": 7540 }, { "epoch": 3.59, "learning_rate": 9.882943072611684e-05, "loss": 0.1972, "step": 7550 }, { "epoch": 3.6, "learning_rate": 9.882634002327853e-05, "loss": 0.17, "step": 7560 }, { "epoch": 3.6, "learning_rate": 9.882324529401841e-05, "loss": 0.2091, "step": 7570 }, { "epoch": 3.61, "learning_rate": 9.882014653859169e-05, "loss": 0.1816, "step": 7580 }, { "epoch": 3.61, "learning_rate": 9.881704375725391e-05, "loss": 0.2077, "step": 7590 }, { "epoch": 3.62, "learning_rate": 9.881393695026091e-05, "loss": 0.1892, "step": 7600 }, { "epoch": 3.62, "learning_rate": 9.881082611786893e-05, "loss": 0.1796, "step": 7610 }, { "epoch": 3.63, "learning_rate": 9.880771126033447e-05, "loss": 0.192, "step": 7620 }, { "epoch": 3.63, "learning_rate": 9.880459237791442e-05, "loss": 0.1932, "step": 7630 }, { "epoch": 3.64, "learning_rate": 9.880146947086594e-05, "loss": 0.2103, "step": 7640 }, { "epoch": 3.64, "learning_rate": 9.879834253944662e-05, "loss": 0.1904, "step": 7650 }, { "epoch": 3.64, "learning_rate": 9.879521158391425e-05, "loss": 0.205, "step": 7660 }, { "epoch": 3.65, "learning_rate": 9.879207660452706e-05, "loss": 0.2012, "step": 7670 }, { "epoch": 3.65, "learning_rate": 9.878893760154357e-05, "loss": 0.2133, "step": 7680 }, { "epoch": 3.66, "learning_rate": 9.878579457522262e-05, "loss": 0.1851, "step": 7690 }, { "epoch": 3.66, "learning_rate": 9.878264752582341e-05, "loss": 0.2443, "step": 7700 }, { "epoch": 3.67, "learning_rate": 9.877949645360546e-05, "loss": 0.1705, "step": 7710 }, { "epoch": 3.67, "learning_rate": 9.877634135882862e-05, "loss": 0.2277, "step": 7720 }, { "epoch": 3.68, "learning_rate": 9.877318224175306e-05, "loss": 0.1627, "step": 7730 }, { "epoch": 3.68, "learning_rate": 9.87700191026393e-05, "loss": 0.1853, "step": 7740 }, { "epoch": 3.69, "learning_rate": 9.87668519417482e-05, "loss": 0.1764, "step": 7750 }, { "epoch": 3.69, "learning_rate": 9.876368075934092e-05, "loss": 0.183, "step": 7760 }, { "epoch": 3.7, "learning_rate": 9.876050555567897e-05, "loss": 0.1793, "step": 7770 }, { "epoch": 3.7, "learning_rate": 9.87573263310242e-05, "loss": 0.192, "step": 7780 }, { "epoch": 3.7, "learning_rate": 9.875414308563878e-05, "loss": 0.1772, "step": 7790 }, { "epoch": 3.71, "learning_rate": 9.87509558197852e-05, "loss": 0.1826, "step": 7800 }, { "epoch": 3.71, "learning_rate": 9.874776453372632e-05, "loss": 0.1674, "step": 7810 }, { "epoch": 3.72, "learning_rate": 9.874456922772529e-05, "loss": 0.1765, "step": 7820 }, { "epoch": 3.72, "learning_rate": 9.87413699020456e-05, "loss": 0.1791, "step": 7830 }, { "epoch": 3.73, "learning_rate": 9.873816655695108e-05, "loss": 0.19, "step": 7840 }, { "epoch": 3.73, "learning_rate": 9.873495919270593e-05, "loss": 0.2031, "step": 7850 }, { "epoch": 3.74, "learning_rate": 9.873174780957459e-05, "loss": 0.1814, "step": 7860 }, { "epoch": 3.74, "learning_rate": 9.872853240782191e-05, "loss": 0.2162, "step": 7870 }, { "epoch": 3.75, "learning_rate": 9.872531298771303e-05, "loss": 0.2137, "step": 7880 }, { "epoch": 3.75, "learning_rate": 9.872208954951347e-05, "loss": 0.1878, "step": 7890 }, { "epoch": 3.75, "learning_rate": 9.871886209348903e-05, "loss": 0.2265, "step": 7900 }, { "epoch": 3.76, "learning_rate": 9.871563061990586e-05, "loss": 0.2125, "step": 7910 }, { "epoch": 3.76, "learning_rate": 9.87123951290304e-05, "loss": 0.1871, "step": 7920 }, { "epoch": 3.77, "learning_rate": 9.870915562112954e-05, "loss": 0.1992, "step": 7930 }, { "epoch": 3.77, "learning_rate": 9.870591209647037e-05, "loss": 0.1766, "step": 7940 }, { "epoch": 3.78, "learning_rate": 9.870266455532036e-05, "loss": 0.1989, "step": 7950 }, { "epoch": 3.78, "learning_rate": 9.869941299794736e-05, "loss": 0.202, "step": 7960 }, { "epoch": 3.79, "learning_rate": 9.869615742461946e-05, "loss": 0.1769, "step": 7970 }, { "epoch": 3.79, "learning_rate": 9.869289783560516e-05, "loss": 0.1844, "step": 7980 }, { "epoch": 3.8, "learning_rate": 9.868963423117324e-05, "loss": 0.1894, "step": 7990 }, { "epoch": 3.8, "learning_rate": 9.868636661159283e-05, "loss": 0.2156, "step": 8000 }, { "epoch": 3.8, "learning_rate": 9.868309497713342e-05, "loss": 0.2057, "step": 8010 }, { "epoch": 3.81, "learning_rate": 9.867981932806477e-05, "loss": 0.1889, "step": 8020 }, { "epoch": 3.81, "learning_rate": 9.8676539664657e-05, "loss": 0.2036, "step": 8030 }, { "epoch": 3.82, "learning_rate": 9.867325598718059e-05, "loss": 0.2115, "step": 8040 }, { "epoch": 3.82, "learning_rate": 9.86699682959063e-05, "loss": 0.2155, "step": 8050 }, { "epoch": 3.83, "learning_rate": 9.866667659110529e-05, "loss": 0.2025, "step": 8060 }, { "epoch": 3.83, "learning_rate": 9.866338087304896e-05, "loss": 0.2357, "step": 8070 }, { "epoch": 3.84, "learning_rate": 9.866008114200911e-05, "loss": 0.2174, "step": 8080 }, { "epoch": 3.84, "learning_rate": 9.865677739825785e-05, "loss": 0.1779, "step": 8090 }, { "epoch": 3.85, "learning_rate": 9.86534696420676e-05, "loss": 0.2051, "step": 8100 }, { "epoch": 3.85, "learning_rate": 9.865015787371117e-05, "loss": 0.1772, "step": 8110 }, { "epoch": 3.86, "learning_rate": 9.864684209346164e-05, "loss": 0.1896, "step": 8120 }, { "epoch": 3.86, "learning_rate": 9.864352230159244e-05, "loss": 0.1818, "step": 8130 }, { "epoch": 3.86, "learning_rate": 9.864019849837734e-05, "loss": 0.198, "step": 8140 }, { "epoch": 3.87, "learning_rate": 9.863687068409043e-05, "loss": 0.2007, "step": 8150 }, { "epoch": 3.87, "learning_rate": 9.863353885900612e-05, "loss": 0.1941, "step": 8160 }, { "epoch": 3.88, "learning_rate": 9.863020302339921e-05, "loss": 0.1878, "step": 8170 }, { "epoch": 3.88, "learning_rate": 9.862686317754474e-05, "loss": 0.1736, "step": 8180 }, { "epoch": 3.89, "learning_rate": 9.862351932171816e-05, "loss": 0.1974, "step": 8190 }, { "epoch": 3.89, "learning_rate": 9.86201714561952e-05, "loss": 0.1808, "step": 8200 }, { "epoch": 3.9, "learning_rate": 9.861681958125194e-05, "loss": 0.2081, "step": 8210 }, { "epoch": 3.9, "learning_rate": 9.86134636971648e-05, "loss": 0.203, "step": 8220 }, { "epoch": 3.91, "learning_rate": 9.861010380421051e-05, "loss": 0.204, "step": 8230 }, { "epoch": 3.91, "learning_rate": 9.860673990266615e-05, "loss": 0.223, "step": 8240 }, { "epoch": 3.91, "learning_rate": 9.860337199280912e-05, "loss": 0.2352, "step": 8250 }, { "epoch": 3.92, "learning_rate": 9.860000007491712e-05, "loss": 0.168, "step": 8260 }, { "epoch": 3.92, "learning_rate": 9.859662414926827e-05, "loss": 0.2185, "step": 8270 }, { "epoch": 3.93, "learning_rate": 9.859324421614092e-05, "loss": 0.1993, "step": 8280 }, { "epoch": 3.93, "learning_rate": 9.85898602758138e-05, "loss": 0.2182, "step": 8290 }, { "epoch": 3.94, "learning_rate": 9.8586472328566e-05, "loss": 0.1747, "step": 8300 }, { "epoch": 3.94, "learning_rate": 9.858308037467685e-05, "loss": 0.1683, "step": 8310 }, { "epoch": 3.95, "learning_rate": 9.857968441442607e-05, "loss": 0.1848, "step": 8320 }, { "epoch": 3.95, "learning_rate": 9.857628444809375e-05, "loss": 0.2324, "step": 8330 }, { "epoch": 4.0, "learning_rate": 9.857288047596023e-05, "loss": 0.2228, "step": 8340 }, { "epoch": 4.01, "learning_rate": 9.856947249830624e-05, "loss": 0.1956, "step": 8350 }, { "epoch": 4.01, "learning_rate": 9.856606051541277e-05, "loss": 0.1865, "step": 8360 }, { "epoch": 4.02, "learning_rate": 9.856264452756123e-05, "loss": 0.1836, "step": 8370 }, { "epoch": 4.02, "learning_rate": 9.855922453503331e-05, "loss": 0.2178, "step": 8380 }, { "epoch": 4.03, "learning_rate": 9.855580053811101e-05, "loss": 0.1859, "step": 8390 }, { "epoch": 4.03, "learning_rate": 9.855237253707674e-05, "loss": 0.1895, "step": 8400 }, { "epoch": 4.04, "learning_rate": 9.854894053221313e-05, "loss": 0.1813, "step": 8410 }, { "epoch": 4.04, "learning_rate": 9.854550452380323e-05, "loss": 0.1611, "step": 8420 }, { "epoch": 4.04, "learning_rate": 9.854206451213037e-05, "loss": 0.1852, "step": 8430 }, { "epoch": 4.05, "learning_rate": 9.853862049747824e-05, "loss": 0.2148, "step": 8440 }, { "epoch": 4.05, "learning_rate": 9.853517248013085e-05, "loss": 0.1736, "step": 8450 }, { "epoch": 4.06, "learning_rate": 9.853172046037251e-05, "loss": 0.3012, "step": 8460 }, { "epoch": 4.06, "learning_rate": 9.852826443848793e-05, "loss": 0.1918, "step": 8470 }, { "epoch": 4.07, "learning_rate": 9.852480441476209e-05, "loss": 0.1707, "step": 8480 }, { "epoch": 4.07, "learning_rate": 9.85213403894803e-05, "loss": 0.2108, "step": 8490 }, { "epoch": 4.08, "learning_rate": 9.851787236292824e-05, "loss": 0.1512, "step": 8500 }, { "epoch": 4.08, "learning_rate": 9.851440033539187e-05, "loss": 0.1875, "step": 8510 }, { "epoch": 4.09, "learning_rate": 9.851092430715756e-05, "loss": 0.1687, "step": 8520 }, { "epoch": 4.09, "learning_rate": 9.850744427851191e-05, "loss": 0.2013, "step": 8530 }, { "epoch": 4.1, "learning_rate": 9.85039602497419e-05, "loss": 0.1928, "step": 8540 }, { "epoch": 4.1, "learning_rate": 9.850047222113488e-05, "loss": 0.1939, "step": 8550 }, { "epoch": 4.1, "learning_rate": 9.849698019297845e-05, "loss": 0.187, "step": 8560 }, { "epoch": 4.11, "learning_rate": 9.849348416556058e-05, "loss": 0.2114, "step": 8570 }, { "epoch": 4.11, "learning_rate": 9.848998413916957e-05, "loss": 0.2008, "step": 8580 }, { "epoch": 4.12, "learning_rate": 9.848648011409405e-05, "loss": 0.2302, "step": 8590 }, { "epoch": 4.12, "learning_rate": 9.848297209062299e-05, "loss": 0.1935, "step": 8600 }, { "epoch": 4.13, "learning_rate": 9.847946006904564e-05, "loss": 0.2072, "step": 8610 }, { "epoch": 4.13, "learning_rate": 9.847594404965164e-05, "loss": 0.2266, "step": 8620 }, { "epoch": 4.14, "learning_rate": 9.847242403273094e-05, "loss": 0.2011, "step": 8630 }, { "epoch": 4.14, "learning_rate": 9.846890001857381e-05, "loss": 0.1955, "step": 8640 }, { "epoch": 4.15, "learning_rate": 9.846537200747083e-05, "loss": 0.1729, "step": 8650 }, { "epoch": 4.15, "learning_rate": 9.846183999971298e-05, "loss": 0.2029, "step": 8660 }, { "epoch": 4.15, "learning_rate": 9.845830399559149e-05, "loss": 0.1993, "step": 8670 }, { "epoch": 4.16, "learning_rate": 9.845476399539797e-05, "loss": 0.1974, "step": 8680 }, { "epoch": 4.16, "learning_rate": 9.845121999942434e-05, "loss": 0.1725, "step": 8690 }, { "epoch": 4.17, "learning_rate": 9.844767200796284e-05, "loss": 0.1844, "step": 8700 }, { "epoch": 4.17, "learning_rate": 9.844412002130608e-05, "loss": 0.1741, "step": 8710 }, { "epoch": 4.18, "learning_rate": 9.844056403974693e-05, "loss": 0.1764, "step": 8720 }, { "epoch": 4.18, "learning_rate": 9.843700406357866e-05, "loss": 0.2003, "step": 8730 }, { "epoch": 4.19, "learning_rate": 9.843344009309484e-05, "loss": 0.1923, "step": 8740 }, { "epoch": 4.19, "learning_rate": 9.842987212858936e-05, "loss": 0.2084, "step": 8750 }, { "epoch": 4.2, "learning_rate": 9.842630017035646e-05, "loss": 0.1777, "step": 8760 }, { "epoch": 4.2, "learning_rate": 9.84227242186907e-05, "loss": 0.1906, "step": 8770 }, { "epoch": 4.2, "learning_rate": 9.841914427388694e-05, "loss": 0.17, "step": 8780 }, { "epoch": 4.21, "learning_rate": 9.841556033624043e-05, "loss": 0.1818, "step": 8790 }, { "epoch": 4.21, "learning_rate": 9.841197240604669e-05, "loss": 0.1913, "step": 8800 }, { "epoch": 4.22, "learning_rate": 9.840838048360163e-05, "loss": 0.2264, "step": 8810 }, { "epoch": 4.22, "learning_rate": 9.840478456920141e-05, "loss": 0.1711, "step": 8820 }, { "epoch": 4.23, "learning_rate": 9.840118466314261e-05, "loss": 0.1974, "step": 8830 }, { "epoch": 4.23, "learning_rate": 9.839758076572205e-05, "loss": 0.1829, "step": 8840 }, { "epoch": 4.24, "learning_rate": 9.839397287723695e-05, "loss": 0.1793, "step": 8850 }, { "epoch": 4.24, "learning_rate": 9.839036099798483e-05, "loss": 0.1925, "step": 8860 }, { "epoch": 4.25, "learning_rate": 9.838674512826354e-05, "loss": 0.1931, "step": 8870 }, { "epoch": 4.25, "learning_rate": 9.838312526837123e-05, "loss": 0.1799, "step": 8880 }, { "epoch": 4.26, "learning_rate": 9.837950141860645e-05, "loss": 0.1946, "step": 8890 }, { "epoch": 4.26, "learning_rate": 9.837587357926803e-05, "loss": 0.2064, "step": 8900 }, { "epoch": 4.26, "learning_rate": 9.837224175065511e-05, "loss": 0.1828, "step": 8910 }, { "epoch": 4.27, "learning_rate": 9.836860593306721e-05, "loss": 0.1927, "step": 8920 }, { "epoch": 4.27, "learning_rate": 9.836496612680415e-05, "loss": 0.1783, "step": 8930 }, { "epoch": 4.28, "learning_rate": 9.836132233216606e-05, "loss": 0.2175, "step": 8940 }, { "epoch": 4.28, "learning_rate": 9.835767454945347e-05, "loss": 0.2341, "step": 8950 }, { "epoch": 4.29, "learning_rate": 9.835402277896715e-05, "loss": 0.176, "step": 8960 }, { "epoch": 4.29, "learning_rate": 9.835036702100826e-05, "loss": 0.1877, "step": 8970 }, { "epoch": 4.3, "learning_rate": 9.834670727587826e-05, "loss": 0.1818, "step": 8980 }, { "epoch": 4.3, "learning_rate": 9.834304354387896e-05, "loss": 0.19, "step": 8990 }, { "epoch": 4.31, "learning_rate": 9.833937582531245e-05, "loss": 0.1811, "step": 9000 }, { "epoch": 4.31, "learning_rate": 9.833570412048124e-05, "loss": 0.1687, "step": 9010 }, { "epoch": 4.31, "learning_rate": 9.833202842968807e-05, "loss": 0.1964, "step": 9020 }, { "epoch": 4.32, "learning_rate": 9.832834875323609e-05, "loss": 0.1721, "step": 9030 }, { "epoch": 4.32, "learning_rate": 9.83246650914287e-05, "loss": 0.192, "step": 9040 }, { "epoch": 4.33, "learning_rate": 9.832097744456971e-05, "loss": 0.2159, "step": 9050 }, { "epoch": 4.33, "learning_rate": 9.831728581296318e-05, "loss": 0.198, "step": 9060 }, { "epoch": 4.34, "learning_rate": 9.831359019691358e-05, "loss": 0.1599, "step": 9070 }, { "epoch": 4.34, "learning_rate": 9.830989059672562e-05, "loss": 0.182, "step": 9080 }, { "epoch": 4.35, "learning_rate": 9.830618701270441e-05, "loss": 0.1827, "step": 9090 }, { "epoch": 4.35, "learning_rate": 9.830247944515537e-05, "loss": 0.1797, "step": 9100 }, { "epoch": 4.36, "learning_rate": 9.829876789438421e-05, "loss": 0.2065, "step": 9110 }, { "epoch": 4.36, "learning_rate": 9.829505236069704e-05, "loss": 0.2044, "step": 9120 }, { "epoch": 4.36, "learning_rate": 9.829133284440025e-05, "loss": 0.2046, "step": 9130 }, { "epoch": 4.37, "learning_rate": 9.828760934580052e-05, "loss": 0.1858, "step": 9140 }, { "epoch": 4.37, "learning_rate": 9.828388186520496e-05, "loss": 0.2039, "step": 9150 }, { "epoch": 4.38, "learning_rate": 9.828015040292094e-05, "loss": 0.1857, "step": 9160 }, { "epoch": 4.38, "learning_rate": 9.827641495925616e-05, "loss": 0.1926, "step": 9170 }, { "epoch": 4.39, "learning_rate": 9.827267553451866e-05, "loss": 0.1967, "step": 9180 }, { "epoch": 4.39, "learning_rate": 9.826893212901682e-05, "loss": 0.1816, "step": 9190 }, { "epoch": 4.4, "learning_rate": 9.826518474305932e-05, "loss": 0.2181, "step": 9200 }, { "epoch": 4.4, "learning_rate": 9.826143337695521e-05, "loss": 0.1862, "step": 9210 }, { "epoch": 4.41, "learning_rate": 9.825767803101384e-05, "loss": 0.1775, "step": 9220 }, { "epoch": 4.41, "learning_rate": 9.825391870554486e-05, "loss": 0.2128, "step": 9230 }, { "epoch": 4.41, "learning_rate": 9.825015540085831e-05, "loss": 0.1744, "step": 9240 }, { "epoch": 4.42, "learning_rate": 9.82463881172645e-05, "loss": 0.1828, "step": 9250 }, { "epoch": 4.42, "learning_rate": 9.824261685507414e-05, "loss": 0.1891, "step": 9260 }, { "epoch": 4.43, "learning_rate": 9.823884161459818e-05, "loss": 0.1807, "step": 9270 }, { "epoch": 4.43, "learning_rate": 9.823506239614797e-05, "loss": 0.2068, "step": 9280 }, { "epoch": 4.44, "learning_rate": 9.823127920003514e-05, "loss": 0.1883, "step": 9290 }, { "epoch": 4.44, "learning_rate": 9.822749202657169e-05, "loss": 0.184, "step": 9300 }, { "epoch": 4.45, "learning_rate": 9.82237008760699e-05, "loss": 0.1804, "step": 9310 }, { "epoch": 4.45, "learning_rate": 9.821990574884241e-05, "loss": 0.1873, "step": 9320 }, { "epoch": 4.46, "learning_rate": 9.821610664520221e-05, "loss": 0.1972, "step": 9330 }, { "epoch": 4.46, "learning_rate": 9.821230356546256e-05, "loss": 0.1911, "step": 9340 }, { "epoch": 4.47, "learning_rate": 9.820849650993709e-05, "loss": 0.2082, "step": 9350 }, { "epoch": 4.47, "learning_rate": 9.820468547893973e-05, "loss": 0.1935, "step": 9360 }, { "epoch": 4.47, "learning_rate": 9.820087047278477e-05, "loss": 0.1693, "step": 9370 }, { "epoch": 4.48, "learning_rate": 9.819705149178682e-05, "loss": 0.1819, "step": 9380 }, { "epoch": 4.48, "learning_rate": 9.819322853626078e-05, "loss": 0.1742, "step": 9390 }, { "epoch": 4.49, "learning_rate": 9.818940160652193e-05, "loss": 0.1693, "step": 9400 }, { "epoch": 4.49, "learning_rate": 9.818557070288584e-05, "loss": 0.1837, "step": 9410 }, { "epoch": 4.5, "learning_rate": 9.818173582566844e-05, "loss": 0.228, "step": 9420 }, { "epoch": 4.5, "learning_rate": 9.817789697518595e-05, "loss": 0.1831, "step": 9430 }, { "epoch": 4.51, "learning_rate": 9.817405415175494e-05, "loss": 0.1537, "step": 9440 }, { "epoch": 4.51, "learning_rate": 9.817020735569231e-05, "loss": 0.1954, "step": 9450 }, { "epoch": 4.52, "learning_rate": 9.81663565873153e-05, "loss": 0.1842, "step": 9460 }, { "epoch": 4.52, "learning_rate": 9.816250184694144e-05, "loss": 0.1979, "step": 9470 }, { "epoch": 4.52, "learning_rate": 9.815864313488861e-05, "loss": 0.1672, "step": 9480 }, { "epoch": 4.53, "learning_rate": 9.815478045147501e-05, "loss": 0.1892, "step": 9490 }, { "epoch": 4.53, "learning_rate": 9.81509137970192e-05, "loss": 0.189, "step": 9500 }, { "epoch": 4.54, "learning_rate": 9.814704317184e-05, "loss": 0.1876, "step": 9510 }, { "epoch": 4.54, "learning_rate": 9.814316857625663e-05, "loss": 0.1722, "step": 9520 }, { "epoch": 4.55, "learning_rate": 9.813929001058859e-05, "loss": 0.1645, "step": 9530 }, { "epoch": 4.55, "learning_rate": 9.813540747515572e-05, "loss": 0.1959, "step": 9540 }, { "epoch": 4.56, "learning_rate": 9.81315209702782e-05, "loss": 0.1733, "step": 9550 }, { "epoch": 4.56, "learning_rate": 9.812763049627653e-05, "loss": 0.206, "step": 9560 }, { "epoch": 4.57, "learning_rate": 9.812373605347152e-05, "loss": 0.1708, "step": 9570 }, { "epoch": 4.57, "learning_rate": 9.811983764218433e-05, "loss": 0.1899, "step": 9580 }, { "epoch": 4.57, "learning_rate": 9.811593526273643e-05, "loss": 0.2071, "step": 9590 }, { "epoch": 4.58, "learning_rate": 9.811202891544965e-05, "loss": 0.1727, "step": 9600 }, { "epoch": 4.58, "learning_rate": 9.81081186006461e-05, "loss": 0.1693, "step": 9610 }, { "epoch": 4.59, "learning_rate": 9.810420431864826e-05, "loss": 0.1769, "step": 9620 }, { "epoch": 4.59, "learning_rate": 9.810028606977891e-05, "loss": 0.2006, "step": 9630 }, { "epoch": 4.6, "learning_rate": 9.809636385436114e-05, "loss": 0.2, "step": 9640 }, { "epoch": 4.6, "learning_rate": 9.809243767271843e-05, "loss": 0.1778, "step": 9650 }, { "epoch": 4.61, "learning_rate": 9.808850752517453e-05, "loss": 0.1849, "step": 9660 }, { "epoch": 4.61, "learning_rate": 9.808457341205353e-05, "loss": 0.1915, "step": 9670 }, { "epoch": 4.62, "learning_rate": 9.808063533367988e-05, "loss": 0.1722, "step": 9680 }, { "epoch": 4.62, "learning_rate": 9.80766932903783e-05, "loss": 0.1929, "step": 9690 }, { "epoch": 4.63, "learning_rate": 9.807274728247389e-05, "loss": 0.1627, "step": 9700 }, { "epoch": 4.63, "learning_rate": 9.806879731029203e-05, "loss": 0.182, "step": 9710 }, { "epoch": 4.63, "learning_rate": 9.806484337415849e-05, "loss": 0.1696, "step": 9720 }, { "epoch": 4.64, "learning_rate": 9.80608854743993e-05, "loss": 0.1491, "step": 9730 }, { "epoch": 4.64, "learning_rate": 9.805692361134082e-05, "loss": 0.1772, "step": 9740 }, { "epoch": 4.65, "learning_rate": 9.80529577853098e-05, "loss": 0.2076, "step": 9750 }, { "epoch": 4.65, "learning_rate": 9.804898799663328e-05, "loss": 0.208, "step": 9760 }, { "epoch": 4.66, "learning_rate": 9.80450142456386e-05, "loss": 0.203, "step": 9770 }, { "epoch": 4.66, "learning_rate": 9.804103653265348e-05, "loss": 0.182, "step": 9780 }, { "epoch": 4.67, "learning_rate": 9.803705485800591e-05, "loss": 0.2056, "step": 9790 }, { "epoch": 4.67, "learning_rate": 9.803306922202426e-05, "loss": 0.1794, "step": 9800 }, { "epoch": 4.68, "learning_rate": 9.80290796250372e-05, "loss": 0.2019, "step": 9810 }, { "epoch": 4.68, "learning_rate": 9.802508606737371e-05, "loss": 0.1881, "step": 9820 }, { "epoch": 4.68, "learning_rate": 9.802108854936312e-05, "loss": 0.1785, "step": 9830 }, { "epoch": 4.69, "learning_rate": 9.80170870713351e-05, "loss": 0.188, "step": 9840 }, { "epoch": 4.69, "learning_rate": 9.80130816336196e-05, "loss": 0.1643, "step": 9850 }, { "epoch": 4.7, "learning_rate": 9.800907223654693e-05, "loss": 0.1699, "step": 9860 }, { "epoch": 4.7, "learning_rate": 9.800505888044777e-05, "loss": 0.1823, "step": 9870 }, { "epoch": 4.71, "learning_rate": 9.8001041565653e-05, "loss": 0.1993, "step": 9880 }, { "epoch": 4.71, "learning_rate": 9.799702029249397e-05, "loss": 0.1983, "step": 9890 }, { "epoch": 4.72, "learning_rate": 9.799299506130226e-05, "loss": 0.1992, "step": 9900 }, { "epoch": 4.72, "learning_rate": 9.798896587240981e-05, "loss": 0.1981, "step": 9910 }, { "epoch": 4.73, "learning_rate": 9.798493272614888e-05, "loss": 0.1846, "step": 9920 }, { "epoch": 4.73, "learning_rate": 9.798089562285208e-05, "loss": 0.2004, "step": 9930 }, { "epoch": 4.73, "learning_rate": 9.79768545628523e-05, "loss": 0.1785, "step": 9940 }, { "epoch": 4.74, "learning_rate": 9.797280954648279e-05, "loss": 0.1609, "step": 9950 }, { "epoch": 4.74, "learning_rate": 9.796876057407713e-05, "loss": 0.1895, "step": 9960 }, { "epoch": 4.75, "learning_rate": 9.79647076459692e-05, "loss": 0.1711, "step": 9970 }, { "epoch": 4.75, "learning_rate": 9.796065076249325e-05, "loss": 0.237, "step": 9980 }, { "epoch": 4.76, "learning_rate": 9.79565899239838e-05, "loss": 0.2187, "step": 9990 }, { "epoch": 4.76, "learning_rate": 9.795252513077572e-05, "loss": 0.163, "step": 10000 }, { "epoch": 4.77, "learning_rate": 9.794845638320422e-05, "loss": 0.191, "step": 10010 }, { "epoch": 4.77, "learning_rate": 9.794438368160483e-05, "loss": 0.1994, "step": 10020 }, { "epoch": 4.78, "learning_rate": 9.79403070263134e-05, "loss": 0.1927, "step": 10030 }, { "epoch": 4.78, "learning_rate": 9.793622641766612e-05, "loss": 0.1694, "step": 10040 }, { "epoch": 4.79, "learning_rate": 9.793214185599945e-05, "loss": 0.1837, "step": 10050 }, { "epoch": 4.79, "learning_rate": 9.792805334165024e-05, "loss": 0.156, "step": 10060 }, { "epoch": 4.79, "learning_rate": 9.79239608749557e-05, "loss": 0.1837, "step": 10070 }, { "epoch": 4.8, "learning_rate": 9.791986445625323e-05, "loss": 0.1628, "step": 10080 }, { "epoch": 4.8, "learning_rate": 9.791576408588071e-05, "loss": 0.1788, "step": 10090 }, { "epoch": 4.81, "learning_rate": 9.79116597641762e-05, "loss": 0.1656, "step": 10100 }, { "epoch": 4.81, "learning_rate": 9.790755149147821e-05, "loss": 0.1685, "step": 10110 }, { "epoch": 4.82, "learning_rate": 9.790343926812551e-05, "loss": 0.1601, "step": 10120 }, { "epoch": 4.82, "learning_rate": 9.789932309445721e-05, "loss": 0.1986, "step": 10130 }, { "epoch": 4.83, "learning_rate": 9.789520297081276e-05, "loss": 0.1818, "step": 10140 }, { "epoch": 4.83, "learning_rate": 9.78910788975319e-05, "loss": 0.1789, "step": 10150 }, { "epoch": 4.84, "learning_rate": 9.788695087495475e-05, "loss": 0.187, "step": 10160 }, { "epoch": 4.84, "learning_rate": 9.78828189034217e-05, "loss": 0.2005, "step": 10170 }, { "epoch": 4.84, "learning_rate": 9.787868298327348e-05, "loss": 0.187, "step": 10180 }, { "epoch": 4.85, "learning_rate": 9.787454311485118e-05, "loss": 0.1635, "step": 10190 }, { "epoch": 4.85, "learning_rate": 9.787039929849617e-05, "loss": 0.1961, "step": 10200 }, { "epoch": 4.86, "learning_rate": 9.786625153455017e-05, "loss": 0.172, "step": 10210 }, { "epoch": 4.86, "learning_rate": 9.786209982335525e-05, "loss": 0.1591, "step": 10220 }, { "epoch": 4.87, "learning_rate": 9.785794416525373e-05, "loss": 0.179, "step": 10230 }, { "epoch": 4.87, "learning_rate": 9.785378456058836e-05, "loss": 0.1964, "step": 10240 }, { "epoch": 4.88, "learning_rate": 9.78496210097021e-05, "loss": 0.1846, "step": 10250 }, { "epoch": 4.88, "learning_rate": 9.784545351293833e-05, "loss": 0.1832, "step": 10260 }, { "epoch": 4.89, "learning_rate": 9.784128207064069e-05, "loss": 0.1765, "step": 10270 }, { "epoch": 4.89, "learning_rate": 9.78371066831532e-05, "loss": 0.182, "step": 10280 }, { "epoch": 4.89, "learning_rate": 9.783292735082018e-05, "loss": 0.1964, "step": 10290 }, { "epoch": 4.9, "learning_rate": 9.782874407398626e-05, "loss": 0.1838, "step": 10300 }, { "epoch": 4.9, "learning_rate": 9.782455685299643e-05, "loss": 0.1898, "step": 10310 }, { "epoch": 4.91, "learning_rate": 9.782036568819596e-05, "loss": 0.1855, "step": 10320 }, { "epoch": 4.91, "learning_rate": 9.781617057993048e-05, "loss": 0.2112, "step": 10330 }, { "epoch": 4.92, "learning_rate": 9.781197152854596e-05, "loss": 0.1877, "step": 10340 }, { "epoch": 4.92, "learning_rate": 9.780776853438863e-05, "loss": 0.1975, "step": 10350 }, { "epoch": 4.93, "learning_rate": 9.780356159780511e-05, "loss": 0.1599, "step": 10360 }, { "epoch": 4.93, "learning_rate": 9.779935071914233e-05, "loss": 0.1644, "step": 10370 }, { "epoch": 4.94, "learning_rate": 9.779513589874752e-05, "loss": 0.1919, "step": 10380 }, { "epoch": 4.94, "learning_rate": 9.779091713696827e-05, "loss": 0.1863, "step": 10390 }, { "epoch": 4.95, "learning_rate": 9.778669443415244e-05, "loss": 0.1632, "step": 10400 }, { "epoch": 4.95, "learning_rate": 9.778246779064828e-05, "loss": 0.1712, "step": 10410 }, { "epoch": 5.0, "learning_rate": 9.777823720680434e-05, "loss": 0.1735, "step": 10420 }, { "epoch": 5.01, "learning_rate": 9.777400268296948e-05, "loss": 0.1773, "step": 10430 }, { "epoch": 5.01, "learning_rate": 9.776976421949288e-05, "loss": 0.1878, "step": 10440 }, { "epoch": 5.02, "learning_rate": 9.77655218167241e-05, "loss": 0.168, "step": 10450 }, { "epoch": 5.02, "learning_rate": 9.776127547501297e-05, "loss": 0.1705, "step": 10460 }, { "epoch": 5.03, "learning_rate": 9.775702519470963e-05, "loss": 0.1727, "step": 10470 }, { "epoch": 5.03, "learning_rate": 9.775277097616465e-05, "loss": 0.1809, "step": 10480 }, { "epoch": 5.03, "learning_rate": 9.774851281972877e-05, "loss": 0.1737, "step": 10490 }, { "epoch": 5.04, "learning_rate": 9.774425072575317e-05, "loss": 0.1725, "step": 10500 }, { "epoch": 5.04, "learning_rate": 9.773998469458932e-05, "loss": 0.1782, "step": 10510 }, { "epoch": 5.05, "learning_rate": 9.773571472658901e-05, "loss": 0.1565, "step": 10520 }, { "epoch": 5.05, "learning_rate": 9.773144082210437e-05, "loss": 0.1706, "step": 10530 }, { "epoch": 5.06, "learning_rate": 9.772716298148782e-05, "loss": 0.1648, "step": 10540 }, { "epoch": 5.06, "learning_rate": 9.772288120509217e-05, "loss": 0.2018, "step": 10550 }, { "epoch": 5.07, "learning_rate": 9.771859549327047e-05, "loss": 0.1813, "step": 10560 }, { "epoch": 5.07, "learning_rate": 9.771430584637616e-05, "loss": 0.1766, "step": 10570 }, { "epoch": 5.08, "learning_rate": 9.771001226476297e-05, "loss": 0.1805, "step": 10580 }, { "epoch": 5.08, "learning_rate": 9.770571474878498e-05, "loss": 0.1687, "step": 10590 }, { "epoch": 5.08, "learning_rate": 9.770141329879657e-05, "loss": 0.1916, "step": 10600 }, { "epoch": 5.09, "learning_rate": 9.769710791515246e-05, "loss": 0.1864, "step": 10610 }, { "epoch": 5.09, "learning_rate": 9.76927985982077e-05, "loss": 0.1771, "step": 10620 }, { "epoch": 5.1, "learning_rate": 9.768848534831762e-05, "loss": 0.1768, "step": 10630 }, { "epoch": 5.1, "learning_rate": 9.768416816583795e-05, "loss": 0.1699, "step": 10640 }, { "epoch": 5.11, "learning_rate": 9.767984705112469e-05, "loss": 0.2014, "step": 10650 }, { "epoch": 5.11, "learning_rate": 9.767552200453416e-05, "loss": 0.1819, "step": 10660 }, { "epoch": 5.12, "learning_rate": 9.767119302642302e-05, "loss": 0.2009, "step": 10670 }, { "epoch": 5.12, "learning_rate": 9.766686011714829e-05, "loss": 0.1802, "step": 10680 }, { "epoch": 5.13, "learning_rate": 9.766252327706727e-05, "loss": 0.1695, "step": 10690 }, { "epoch": 5.13, "learning_rate": 9.765818250653756e-05, "loss": 0.1801, "step": 10700 }, { "epoch": 5.13, "learning_rate": 9.765383780591713e-05, "loss": 0.1724, "step": 10710 }, { "epoch": 5.14, "learning_rate": 9.764948917556429e-05, "loss": 0.1699, "step": 10720 }, { "epoch": 5.14, "learning_rate": 9.764513661583761e-05, "loss": 0.1686, "step": 10730 }, { "epoch": 5.15, "learning_rate": 9.764078012709605e-05, "loss": 0.1797, "step": 10740 }, { "epoch": 5.15, "learning_rate": 9.763641970969885e-05, "loss": 0.1615, "step": 10750 }, { "epoch": 5.16, "learning_rate": 9.763205536400558e-05, "loss": 0.169, "step": 10760 }, { "epoch": 5.16, "learning_rate": 9.762768709037616e-05, "loss": 0.1997, "step": 10770 }, { "epoch": 5.17, "learning_rate": 9.76233148891708e-05, "loss": 0.1958, "step": 10780 }, { "epoch": 5.17, "learning_rate": 9.761893876075008e-05, "loss": 0.1952, "step": 10790 }, { "epoch": 5.18, "learning_rate": 9.761455870547482e-05, "loss": 0.1797, "step": 10800 }, { "epoch": 5.18, "learning_rate": 9.761017472370627e-05, "loss": 0.1785, "step": 10810 }, { "epoch": 5.19, "learning_rate": 9.760578681580592e-05, "loss": 0.1731, "step": 10820 }, { "epoch": 5.19, "learning_rate": 9.760139498213562e-05, "loss": 0.1852, "step": 10830 }, { "epoch": 5.19, "learning_rate": 9.759699922305756e-05, "loss": 0.1934, "step": 10840 }, { "epoch": 5.2, "learning_rate": 9.75925995389342e-05, "loss": 0.2066, "step": 10850 }, { "epoch": 5.2, "learning_rate": 9.758819593012838e-05, "loss": 0.2013, "step": 10860 }, { "epoch": 5.21, "learning_rate": 9.758378839700324e-05, "loss": 0.1842, "step": 10870 }, { "epoch": 5.21, "learning_rate": 9.757937693992222e-05, "loss": 0.1898, "step": 10880 }, { "epoch": 5.22, "learning_rate": 9.757496155924912e-05, "loss": 0.1494, "step": 10890 }, { "epoch": 5.22, "learning_rate": 9.757054225534808e-05, "loss": 0.1719, "step": 10900 }, { "epoch": 5.23, "learning_rate": 9.756611902858348e-05, "loss": 0.1569, "step": 10910 }, { "epoch": 5.23, "learning_rate": 9.756169187932011e-05, "loss": 0.1845, "step": 10920 }, { "epoch": 5.24, "learning_rate": 9.755726080792304e-05, "loss": 0.1683, "step": 10930 }, { "epoch": 5.24, "learning_rate": 9.755282581475769e-05, "loss": 0.1877, "step": 10940 }, { "epoch": 5.24, "learning_rate": 9.754838690018976e-05, "loss": 0.174, "step": 10950 }, { "epoch": 5.25, "learning_rate": 9.754394406458533e-05, "loss": 0.1766, "step": 10960 }, { "epoch": 5.25, "learning_rate": 9.753949730831077e-05, "loss": 0.1911, "step": 10970 }, { "epoch": 5.26, "learning_rate": 9.753504663173276e-05, "loss": 0.2098, "step": 10980 }, { "epoch": 5.26, "learning_rate": 9.753059203521834e-05, "loss": 0.1789, "step": 10990 }, { "epoch": 5.27, "learning_rate": 9.752613351913485e-05, "loss": 0.1855, "step": 11000 }, { "epoch": 5.27, "learning_rate": 9.752167108384995e-05, "loss": 0.1786, "step": 11010 }, { "epoch": 5.28, "learning_rate": 9.751720472973164e-05, "loss": 0.1714, "step": 11020 }, { "epoch": 5.28, "learning_rate": 9.751273445714823e-05, "loss": 0.1729, "step": 11030 }, { "epoch": 5.29, "learning_rate": 9.750826026646836e-05, "loss": 0.1664, "step": 11040 }, { "epoch": 5.29, "learning_rate": 9.750378215806097e-05, "loss": 0.1739, "step": 11050 }, { "epoch": 5.29, "learning_rate": 9.749930013229536e-05, "loss": 0.1635, "step": 11060 }, { "epoch": 5.3, "learning_rate": 9.749481418954114e-05, "loss": 0.156, "step": 11070 }, { "epoch": 5.3, "learning_rate": 9.749032433016825e-05, "loss": 0.1713, "step": 11080 }, { "epoch": 5.31, "learning_rate": 9.748583055454692e-05, "loss": 0.1797, "step": 11090 }, { "epoch": 5.31, "learning_rate": 9.748133286304774e-05, "loss": 0.169, "step": 11100 }, { "epoch": 5.32, "learning_rate": 9.74768312560416e-05, "loss": 0.171, "step": 11110 }, { "epoch": 5.32, "learning_rate": 9.747232573389971e-05, "loss": 0.1795, "step": 11120 }, { "epoch": 5.33, "learning_rate": 9.746781629699364e-05, "loss": 0.1644, "step": 11130 }, { "epoch": 5.33, "learning_rate": 9.746330294569523e-05, "loss": 0.17, "step": 11140 }, { "epoch": 5.34, "learning_rate": 9.74587856803767e-05, "loss": 0.1884, "step": 11150 }, { "epoch": 5.34, "learning_rate": 9.745426450141053e-05, "loss": 0.1821, "step": 11160 }, { "epoch": 5.35, "learning_rate": 9.744973940916959e-05, "loss": 0.1845, "step": 11170 }, { "epoch": 5.35, "learning_rate": 9.7445210404027e-05, "loss": 0.166, "step": 11180 }, { "epoch": 5.35, "learning_rate": 9.744067748635627e-05, "loss": 0.1605, "step": 11190 }, { "epoch": 5.36, "learning_rate": 9.743614065653119e-05, "loss": 0.177, "step": 11200 }, { "epoch": 5.36, "learning_rate": 9.743159991492589e-05, "loss": 0.1646, "step": 11210 }, { "epoch": 5.37, "learning_rate": 9.74270552619148e-05, "loss": 0.1888, "step": 11220 }, { "epoch": 5.37, "learning_rate": 9.742250669787272e-05, "loss": 0.1974, "step": 11230 }, { "epoch": 5.38, "learning_rate": 9.741795422317474e-05, "loss": 0.1586, "step": 11240 }, { "epoch": 5.38, "learning_rate": 9.741339783819625e-05, "loss": 0.1744, "step": 11250 }, { "epoch": 5.39, "learning_rate": 9.740883754331302e-05, "loss": 0.1949, "step": 11260 }, { "epoch": 5.39, "learning_rate": 9.740427333890109e-05, "loss": 0.166, "step": 11270 }, { "epoch": 5.4, "learning_rate": 9.739970522533685e-05, "loss": 0.2174, "step": 11280 }, { "epoch": 5.4, "learning_rate": 9.7395133202997e-05, "loss": 0.1847, "step": 11290 }, { "epoch": 5.4, "learning_rate": 9.739055727225856e-05, "loss": 0.181, "step": 11300 }, { "epoch": 5.41, "learning_rate": 9.738597743349892e-05, "loss": 0.1804, "step": 11310 }, { "epoch": 5.41, "learning_rate": 9.73813936870957e-05, "loss": 0.1717, "step": 11320 }, { "epoch": 5.42, "learning_rate": 9.737680603342693e-05, "loss": 0.1554, "step": 11330 }, { "epoch": 5.42, "learning_rate": 9.737221447287093e-05, "loss": 0.2052, "step": 11340 }, { "epoch": 5.43, "learning_rate": 9.73676190058063e-05, "loss": 0.1765, "step": 11350 }, { "epoch": 5.43, "learning_rate": 9.736301963261203e-05, "loss": 0.1932, "step": 11360 }, { "epoch": 5.44, "learning_rate": 9.735841635366742e-05, "loss": 0.2065, "step": 11370 }, { "epoch": 5.44, "learning_rate": 9.735380916935203e-05, "loss": 0.2017, "step": 11380 }, { "epoch": 5.45, "learning_rate": 9.734919808004583e-05, "loss": 0.1914, "step": 11390 }, { "epoch": 5.45, "learning_rate": 9.734458308612906e-05, "loss": 0.1791, "step": 11400 }, { "epoch": 5.45, "learning_rate": 9.733996418798226e-05, "loss": 0.1734, "step": 11410 }, { "epoch": 5.46, "learning_rate": 9.733534138598636e-05, "loss": 0.1916, "step": 11420 }, { "epoch": 5.46, "learning_rate": 9.733071468052256e-05, "loss": 0.1927, "step": 11430 }, { "epoch": 5.47, "learning_rate": 9.73260840719724e-05, "loss": 0.1886, "step": 11440 }, { "epoch": 5.47, "learning_rate": 9.732144956071774e-05, "loss": 0.1992, "step": 11450 }, { "epoch": 5.48, "learning_rate": 9.731681114714076e-05, "loss": 0.1807, "step": 11460 }, { "epoch": 5.48, "learning_rate": 9.731216883162397e-05, "loss": 0.1857, "step": 11470 }, { "epoch": 5.49, "learning_rate": 9.730752261455017e-05, "loss": 0.1749, "step": 11480 }, { "epoch": 5.49, "learning_rate": 9.730287249630253e-05, "loss": 0.16, "step": 11490 }, { "epoch": 5.5, "learning_rate": 9.729821847726452e-05, "loss": 0.1522, "step": 11500 }, { "epoch": 5.5, "learning_rate": 9.729356055781991e-05, "loss": 0.1847, "step": 11510 }, { "epoch": 5.51, "learning_rate": 9.728889873835283e-05, "loss": 0.1761, "step": 11520 }, { "epoch": 5.51, "learning_rate": 9.728423301924771e-05, "loss": 0.1808, "step": 11530 }, { "epoch": 5.51, "learning_rate": 9.727956340088928e-05, "loss": 0.1873, "step": 11540 }, { "epoch": 5.52, "learning_rate": 9.727488988366264e-05, "loss": 0.165, "step": 11550 }, { "epoch": 5.52, "learning_rate": 9.72702124679532e-05, "loss": 0.1905, "step": 11560 }, { "epoch": 5.53, "learning_rate": 9.726553115414665e-05, "loss": 0.1926, "step": 11570 }, { "epoch": 5.53, "learning_rate": 9.726084594262906e-05, "loss": 0.1824, "step": 11580 }, { "epoch": 5.54, "learning_rate": 9.725615683378675e-05, "loss": 0.2172, "step": 11590 }, { "epoch": 5.54, "learning_rate": 9.725146382800643e-05, "loss": 0.1709, "step": 11600 }, { "epoch": 5.55, "learning_rate": 9.724676692567512e-05, "loss": 0.1815, "step": 11610 }, { "epoch": 5.55, "learning_rate": 9.724206612718012e-05, "loss": 0.178, "step": 11620 }, { "epoch": 5.56, "learning_rate": 9.723736143290909e-05, "loss": 0.1946, "step": 11630 }, { "epoch": 5.56, "learning_rate": 9.723265284324998e-05, "loss": 0.1863, "step": 11640 }, { "epoch": 5.56, "learning_rate": 9.722794035859111e-05, "loss": 0.1795, "step": 11650 }, { "epoch": 5.57, "learning_rate": 9.722322397932107e-05, "loss": 0.1814, "step": 11660 }, { "epoch": 5.57, "learning_rate": 9.721850370582879e-05, "loss": 0.1627, "step": 11670 }, { "epoch": 5.58, "learning_rate": 9.721377953850355e-05, "loss": 0.164, "step": 11680 }, { "epoch": 5.58, "learning_rate": 9.720905147773489e-05, "loss": 0.1651, "step": 11690 }, { "epoch": 5.59, "learning_rate": 9.720431952391271e-05, "loss": 0.1795, "step": 11700 }, { "epoch": 5.59, "learning_rate": 9.719958367742725e-05, "loss": 0.1838, "step": 11710 }, { "epoch": 5.6, "learning_rate": 9.719484393866903e-05, "loss": 0.1694, "step": 11720 }, { "epoch": 5.6, "learning_rate": 9.719010030802891e-05, "loss": 0.1536, "step": 11730 }, { "epoch": 5.61, "learning_rate": 9.718535278589808e-05, "loss": 0.1672, "step": 11740 }, { "epoch": 5.61, "learning_rate": 9.718060137266804e-05, "loss": 0.1785, "step": 11750 }, { "epoch": 5.61, "learning_rate": 9.71758460687306e-05, "loss": 0.1953, "step": 11760 }, { "epoch": 5.62, "learning_rate": 9.717108687447789e-05, "loss": 0.1638, "step": 11770 }, { "epoch": 5.62, "learning_rate": 9.716632379030239e-05, "loss": 0.1615, "step": 11780 }, { "epoch": 5.63, "learning_rate": 9.71615568165969e-05, "loss": 0.1699, "step": 11790 }, { "epoch": 5.63, "learning_rate": 9.715678595375449e-05, "loss": 0.183, "step": 11800 }, { "epoch": 5.64, "learning_rate": 9.715201120216861e-05, "loss": 0.18, "step": 11810 }, { "epoch": 5.64, "learning_rate": 9.714723256223299e-05, "loss": 0.1791, "step": 11820 }, { "epoch": 5.65, "learning_rate": 9.714245003434171e-05, "loss": 0.1551, "step": 11830 }, { "epoch": 5.65, "learning_rate": 9.713766361888916e-05, "loss": 0.1535, "step": 11840 }, { "epoch": 5.66, "learning_rate": 9.713287331627001e-05, "loss": 0.1886, "step": 11850 }, { "epoch": 5.66, "learning_rate": 9.712807912687933e-05, "loss": 0.1784, "step": 11860 }, { "epoch": 5.66, "learning_rate": 9.712328105111248e-05, "loss": 0.1617, "step": 11870 }, { "epoch": 5.67, "learning_rate": 9.711847908936506e-05, "loss": 0.1681, "step": 11880 }, { "epoch": 5.67, "learning_rate": 9.711367324203313e-05, "loss": 0.1732, "step": 11890 }, { "epoch": 5.68, "learning_rate": 9.710886350951298e-05, "loss": 0.1507, "step": 11900 }, { "epoch": 5.68, "learning_rate": 9.710404989220123e-05, "loss": 0.1864, "step": 11910 }, { "epoch": 5.69, "learning_rate": 9.709923239049483e-05, "loss": 0.1549, "step": 11920 }, { "epoch": 5.69, "learning_rate": 9.709441100479104e-05, "loss": 0.1697, "step": 11930 }, { "epoch": 5.7, "learning_rate": 9.70895857354875e-05, "loss": 0.164, "step": 11940 }, { "epoch": 5.7, "learning_rate": 9.708475658298206e-05, "loss": 0.1717, "step": 11950 }, { "epoch": 5.71, "learning_rate": 9.707992354767299e-05, "loss": 0.1532, "step": 11960 }, { "epoch": 5.71, "learning_rate": 9.707508662995885e-05, "loss": 0.1638, "step": 11970 }, { "epoch": 5.72, "learning_rate": 9.707024583023847e-05, "loss": 0.1606, "step": 11980 }, { "epoch": 5.72, "learning_rate": 9.706540114891108e-05, "loss": 0.1783, "step": 11990 }, { "epoch": 5.72, "learning_rate": 9.706055258637619e-05, "loss": 0.1729, "step": 12000 }, { "epoch": 5.73, "learning_rate": 9.70557001430336e-05, "loss": 0.1713, "step": 12010 }, { "epoch": 5.73, "learning_rate": 9.70508438192835e-05, "loss": 0.184, "step": 12020 }, { "epoch": 5.74, "learning_rate": 9.704598361552635e-05, "loss": 0.1727, "step": 12030 }, { "epoch": 5.74, "learning_rate": 9.704111953216293e-05, "loss": 0.195, "step": 12040 }, { "epoch": 5.75, "learning_rate": 9.703625156959438e-05, "loss": 0.1589, "step": 12050 }, { "epoch": 5.75, "learning_rate": 9.703137972822209e-05, "loss": 0.1552, "step": 12060 }, { "epoch": 5.76, "learning_rate": 9.702650400844785e-05, "loss": 0.1701, "step": 12070 }, { "epoch": 5.76, "learning_rate": 9.702162441067372e-05, "loss": 0.1963, "step": 12080 }, { "epoch": 5.77, "learning_rate": 9.701674093530208e-05, "loss": 0.1596, "step": 12090 }, { "epoch": 5.77, "learning_rate": 9.701185358273567e-05, "loss": 0.1712, "step": 12100 }, { "epoch": 5.77, "learning_rate": 9.70069623533775e-05, "loss": 0.1802, "step": 12110 }, { "epoch": 5.78, "learning_rate": 9.700206724763093e-05, "loss": 0.1688, "step": 12120 }, { "epoch": 5.78, "learning_rate": 9.699716826589962e-05, "loss": 0.163, "step": 12130 }, { "epoch": 5.79, "learning_rate": 9.699226540858755e-05, "loss": 0.1854, "step": 12140 }, { "epoch": 5.79, "learning_rate": 9.698735867609906e-05, "loss": 0.1977, "step": 12150 }, { "epoch": 5.8, "learning_rate": 9.698244806883876e-05, "loss": 0.1685, "step": 12160 }, { "epoch": 5.8, "learning_rate": 9.697753358721161e-05, "loss": 0.1914, "step": 12170 }, { "epoch": 5.81, "learning_rate": 9.697261523162287e-05, "loss": 0.166, "step": 12180 }, { "epoch": 5.81, "learning_rate": 9.696769300247812e-05, "loss": 0.1894, "step": 12190 }, { "epoch": 5.82, "learning_rate": 9.696276690018329e-05, "loss": 0.1541, "step": 12200 }, { "epoch": 5.82, "learning_rate": 9.69578369251446e-05, "loss": 0.1645, "step": 12210 }, { "epoch": 5.82, "learning_rate": 9.695290307776859e-05, "loss": 0.1799, "step": 12220 }, { "epoch": 5.83, "learning_rate": 9.69479653584621e-05, "loss": 0.1791, "step": 12230 }, { "epoch": 5.83, "learning_rate": 9.694302376763236e-05, "loss": 0.1949, "step": 12240 }, { "epoch": 5.84, "learning_rate": 9.693807830568686e-05, "loss": 0.1877, "step": 12250 }, { "epoch": 5.84, "learning_rate": 9.693312897303341e-05, "loss": 0.1897, "step": 12260 }, { "epoch": 5.85, "learning_rate": 9.692817577008016e-05, "loss": 0.1857, "step": 12270 }, { "epoch": 5.85, "learning_rate": 9.692321869723558e-05, "loss": 0.1958, "step": 12280 }, { "epoch": 5.86, "learning_rate": 9.691825775490844e-05, "loss": 0.1849, "step": 12290 }, { "epoch": 5.86, "learning_rate": 9.691329294350784e-05, "loss": 0.1983, "step": 12300 }, { "epoch": 5.87, "learning_rate": 9.69083242634432e-05, "loss": 0.176, "step": 12310 }, { "epoch": 5.87, "learning_rate": 9.690335171512426e-05, "loss": 0.181, "step": 12320 }, { "epoch": 5.88, "learning_rate": 9.689837529896109e-05, "loss": 0.1713, "step": 12330 }, { "epoch": 5.88, "learning_rate": 9.689339501536403e-05, "loss": 0.1757, "step": 12340 }, { "epoch": 5.88, "learning_rate": 9.688841086474381e-05, "loss": 0.2039, "step": 12350 }, { "epoch": 5.89, "learning_rate": 9.688342284751141e-05, "loss": 0.1957, "step": 12360 }, { "epoch": 5.89, "learning_rate": 9.68784309640782e-05, "loss": 0.2159, "step": 12370 }, { "epoch": 5.9, "learning_rate": 9.687343521485582e-05, "loss": 0.1994, "step": 12380 }, { "epoch": 5.9, "learning_rate": 9.686843560025623e-05, "loss": 0.186, "step": 12390 }, { "epoch": 5.91, "learning_rate": 9.686343212069174e-05, "loss": 0.1885, "step": 12400 }, { "epoch": 5.91, "learning_rate": 9.68584247765749e-05, "loss": 0.1808, "step": 12410 }, { "epoch": 5.92, "learning_rate": 9.685341356831871e-05, "loss": 0.1727, "step": 12420 }, { "epoch": 5.92, "learning_rate": 9.684839849633638e-05, "loss": 0.1974, "step": 12430 }, { "epoch": 5.93, "learning_rate": 9.684337956104146e-05, "loss": 0.1649, "step": 12440 }, { "epoch": 5.93, "learning_rate": 9.683835676284784e-05, "loss": 0.1529, "step": 12450 }, { "epoch": 5.93, "learning_rate": 9.683333010216976e-05, "loss": 0.1527, "step": 12460 }, { "epoch": 5.94, "learning_rate": 9.68282995794217e-05, "loss": 0.1688, "step": 12470 }, { "epoch": 5.94, "learning_rate": 9.68232651950185e-05, "loss": 0.1879, "step": 12480 }, { "epoch": 5.95, "learning_rate": 9.681822694937532e-05, "loss": 0.1941, "step": 12490 }, { "epoch": 6.0, "learning_rate": 9.681318484290762e-05, "loss": 0.2031, "step": 12500 }, { "epoch": 6.01, "learning_rate": 9.680813887603124e-05, "loss": 0.1924, "step": 12510 }, { "epoch": 6.01, "learning_rate": 9.680308904916224e-05, "loss": 0.1559, "step": 12520 }, { "epoch": 6.01, "learning_rate": 9.679803536271707e-05, "loss": 0.1571, "step": 12530 }, { "epoch": 6.02, "learning_rate": 9.67929778171125e-05, "loss": 0.181, "step": 12540 }, { "epoch": 6.02, "learning_rate": 9.678791641276554e-05, "loss": 0.189, "step": 12550 }, { "epoch": 6.03, "learning_rate": 9.678285115009364e-05, "loss": 0.184, "step": 12560 }, { "epoch": 6.03, "learning_rate": 9.677778202951445e-05, "loss": 0.1493, "step": 12570 }, { "epoch": 6.04, "learning_rate": 9.677270905144602e-05, "loss": 0.1627, "step": 12580 }, { "epoch": 6.04, "learning_rate": 9.676763221630668e-05, "loss": 0.1463, "step": 12590 }, { "epoch": 6.05, "learning_rate": 9.676255152451508e-05, "loss": 0.1845, "step": 12600 }, { "epoch": 6.05, "learning_rate": 9.675746697649021e-05, "loss": 0.1611, "step": 12610 }, { "epoch": 6.06, "learning_rate": 9.675237857265136e-05, "loss": 0.1732, "step": 12620 }, { "epoch": 6.06, "learning_rate": 9.674728631341811e-05, "loss": 0.1864, "step": 12630 }, { "epoch": 6.06, "learning_rate": 9.674219019921044e-05, "loss": 0.1662, "step": 12640 }, { "epoch": 6.07, "learning_rate": 9.673709023044857e-05, "loss": 0.1882, "step": 12650 }, { "epoch": 6.07, "learning_rate": 9.673198640755306e-05, "loss": 0.1565, "step": 12660 }, { "epoch": 6.08, "learning_rate": 9.672687873094479e-05, "loss": 0.1762, "step": 12670 }, { "epoch": 6.08, "learning_rate": 9.672176720104497e-05, "loss": 0.1665, "step": 12680 }, { "epoch": 6.09, "learning_rate": 9.671665181827514e-05, "loss": 0.1828, "step": 12690 }, { "epoch": 6.09, "learning_rate": 9.671153258305709e-05, "loss": 0.1636, "step": 12700 }, { "epoch": 6.1, "learning_rate": 9.6706409495813e-05, "loss": 0.1756, "step": 12710 }, { "epoch": 6.1, "learning_rate": 9.670128255696533e-05, "loss": 0.1578, "step": 12720 }, { "epoch": 6.11, "learning_rate": 9.669615176693686e-05, "loss": 0.1783, "step": 12730 }, { "epoch": 6.11, "learning_rate": 9.669101712615074e-05, "loss": 0.1821, "step": 12740 }, { "epoch": 6.12, "learning_rate": 9.668587863503035e-05, "loss": 0.1621, "step": 12750 }, { "epoch": 6.12, "learning_rate": 9.668073629399945e-05, "loss": 0.1745, "step": 12760 }, { "epoch": 6.12, "learning_rate": 9.667559010348209e-05, "loss": 0.1694, "step": 12770 }, { "epoch": 6.13, "learning_rate": 9.667044006390264e-05, "loss": 0.1387, "step": 12780 }, { "epoch": 6.13, "learning_rate": 9.666528617568582e-05, "loss": 0.1611, "step": 12790 }, { "epoch": 6.14, "learning_rate": 9.666012843925663e-05, "loss": 0.1684, "step": 12800 }, { "epoch": 6.14, "learning_rate": 9.665496685504037e-05, "loss": 0.157, "step": 12810 }, { "epoch": 6.15, "learning_rate": 9.664980142346271e-05, "loss": 0.2141, "step": 12820 }, { "epoch": 6.15, "learning_rate": 9.664463214494963e-05, "loss": 0.1914, "step": 12830 }, { "epoch": 6.16, "learning_rate": 9.66394590199274e-05, "loss": 0.1615, "step": 12840 }, { "epoch": 6.16, "learning_rate": 9.663428204882258e-05, "loss": 0.1873, "step": 12850 }, { "epoch": 6.17, "learning_rate": 9.662910123206212e-05, "loss": 0.1816, "step": 12860 }, { "epoch": 6.17, "learning_rate": 9.662391657007324e-05, "loss": 0.174, "step": 12870 }, { "epoch": 6.17, "learning_rate": 9.661872806328352e-05, "loss": 0.1746, "step": 12880 }, { "epoch": 6.18, "learning_rate": 9.661353571212077e-05, "loss": 0.1525, "step": 12890 }, { "epoch": 6.18, "learning_rate": 9.660833951701321e-05, "loss": 0.1725, "step": 12900 }, { "epoch": 6.19, "learning_rate": 9.660313947838934e-05, "loss": 0.1646, "step": 12910 }, { "epoch": 6.19, "learning_rate": 9.659793559667795e-05, "loss": 0.1695, "step": 12920 }, { "epoch": 6.2, "learning_rate": 9.659272787230822e-05, "loss": 0.1915, "step": 12930 }, { "epoch": 6.2, "learning_rate": 9.658751630570955e-05, "loss": 0.1637, "step": 12940 }, { "epoch": 6.21, "learning_rate": 9.658230089731174e-05, "loss": 0.1706, "step": 12950 }, { "epoch": 6.21, "learning_rate": 9.657708164754487e-05, "loss": 0.1884, "step": 12960 }, { "epoch": 6.22, "learning_rate": 9.657185855683932e-05, "loss": 0.1896, "step": 12970 }, { "epoch": 6.22, "learning_rate": 9.656663162562582e-05, "loss": 0.1723, "step": 12980 }, { "epoch": 6.22, "learning_rate": 9.65614008543354e-05, "loss": 0.1686, "step": 12990 }, { "epoch": 6.23, "learning_rate": 9.655616624339945e-05, "loss": 0.1837, "step": 13000 }, { "epoch": 6.23, "learning_rate": 9.655092779324958e-05, "loss": 0.1606, "step": 13010 }, { "epoch": 6.24, "learning_rate": 9.65456855043178e-05, "loss": 0.1801, "step": 13020 }, { "epoch": 6.24, "learning_rate": 9.654043937703642e-05, "loss": 0.1512, "step": 13030 }, { "epoch": 6.25, "learning_rate": 9.653518941183803e-05, "loss": 0.1936, "step": 13040 }, { "epoch": 6.25, "learning_rate": 9.65299356091556e-05, "loss": 0.1727, "step": 13050 }, { "epoch": 6.26, "learning_rate": 9.652467796942236e-05, "loss": 0.1978, "step": 13060 }, { "epoch": 6.26, "learning_rate": 9.651941649307189e-05, "loss": 0.1601, "step": 13070 }, { "epoch": 6.27, "learning_rate": 9.651415118053804e-05, "loss": 0.1485, "step": 13080 }, { "epoch": 6.27, "learning_rate": 9.650888203225503e-05, "loss": 0.1593, "step": 13090 }, { "epoch": 6.28, "learning_rate": 9.650360904865738e-05, "loss": 0.1544, "step": 13100 }, { "epoch": 6.28, "learning_rate": 9.649833223017993e-05, "loss": 0.1768, "step": 13110 }, { "epoch": 6.28, "learning_rate": 9.64930515772578e-05, "loss": 0.1535, "step": 13120 }, { "epoch": 6.29, "learning_rate": 9.64877670903265e-05, "loss": 0.167, "step": 13130 }, { "epoch": 6.29, "learning_rate": 9.648247876982175e-05, "loss": 0.1659, "step": 13140 }, { "epoch": 6.3, "learning_rate": 9.64771866161797e-05, "loss": 0.1705, "step": 13150 }, { "epoch": 6.3, "learning_rate": 9.647189062983673e-05, "loss": 0.1794, "step": 13160 }, { "epoch": 6.31, "learning_rate": 9.646659081122959e-05, "loss": 0.1695, "step": 13170 }, { "epoch": 6.31, "learning_rate": 9.64612871607953e-05, "loss": 0.1964, "step": 13180 }, { "epoch": 6.32, "learning_rate": 9.645597967897124e-05, "loss": 0.1925, "step": 13190 }, { "epoch": 6.32, "learning_rate": 9.64506683661951e-05, "loss": 0.1858, "step": 13200 }, { "epoch": 6.33, "learning_rate": 9.644535322290484e-05, "loss": 0.1725, "step": 13210 }, { "epoch": 6.33, "learning_rate": 9.64400342495388e-05, "loss": 0.144, "step": 13220 }, { "epoch": 6.33, "learning_rate": 9.643471144653557e-05, "loss": 0.1576, "step": 13230 }, { "epoch": 6.34, "learning_rate": 9.642938481433412e-05, "loss": 0.1759, "step": 13240 }, { "epoch": 6.34, "learning_rate": 9.64240543533737e-05, "loss": 0.1546, "step": 13250 }, { "epoch": 6.35, "learning_rate": 9.641872006409387e-05, "loss": 0.1542, "step": 13260 }, { "epoch": 6.35, "learning_rate": 9.641338194693454e-05, "loss": 0.1776, "step": 13270 }, { "epoch": 6.36, "learning_rate": 9.640804000233589e-05, "loss": 0.1571, "step": 13280 }, { "epoch": 6.36, "learning_rate": 9.640269423073844e-05, "loss": 0.1845, "step": 13290 }, { "epoch": 6.37, "learning_rate": 9.639734463258303e-05, "loss": 0.1939, "step": 13300 }, { "epoch": 6.37, "learning_rate": 9.639199120831083e-05, "loss": 0.1752, "step": 13310 }, { "epoch": 6.38, "learning_rate": 9.638663395836327e-05, "loss": 0.1803, "step": 13320 }, { "epoch": 6.38, "learning_rate": 9.638127288318215e-05, "loss": 0.1746, "step": 13330 }, { "epoch": 6.38, "learning_rate": 9.637590798320957e-05, "loss": 0.1787, "step": 13340 }, { "epoch": 6.39, "learning_rate": 9.637053925888794e-05, "loss": 0.1736, "step": 13350 }, { "epoch": 6.39, "learning_rate": 9.636516671065998e-05, "loss": 0.1968, "step": 13360 }, { "epoch": 6.4, "learning_rate": 9.635979033896874e-05, "loss": 0.1459, "step": 13370 }, { "epoch": 6.4, "learning_rate": 9.635441014425757e-05, "loss": 0.1639, "step": 13380 }, { "epoch": 6.41, "learning_rate": 9.634902612697014e-05, "loss": 0.1623, "step": 13390 }, { "epoch": 6.41, "learning_rate": 9.634363828755044e-05, "loss": 0.1869, "step": 13400 }, { "epoch": 6.42, "learning_rate": 9.633824662644279e-05, "loss": 0.1791, "step": 13410 }, { "epoch": 6.42, "learning_rate": 9.633285114409181e-05, "loss": 0.1681, "step": 13420 }, { "epoch": 6.43, "learning_rate": 9.63274518409424e-05, "loss": 0.1597, "step": 13430 }, { "epoch": 6.43, "learning_rate": 9.632204871743984e-05, "loss": 0.1831, "step": 13440 }, { "epoch": 6.44, "learning_rate": 9.631664177402968e-05, "loss": 0.1727, "step": 13450 }, { "epoch": 6.44, "learning_rate": 9.631123101115781e-05, "loss": 0.163, "step": 13460 }, { "epoch": 6.44, "learning_rate": 9.630581642927042e-05, "loss": 0.1791, "step": 13470 }, { "epoch": 6.45, "learning_rate": 9.6300398028814e-05, "loss": 0.1623, "step": 13480 }, { "epoch": 6.45, "learning_rate": 9.62949758102354e-05, "loss": 0.1719, "step": 13490 }, { "epoch": 6.46, "learning_rate": 9.628954977398175e-05, "loss": 0.1626, "step": 13500 }, { "epoch": 6.46, "learning_rate": 9.628411992050049e-05, "loss": 0.1853, "step": 13510 }, { "epoch": 6.47, "learning_rate": 9.627868625023942e-05, "loss": 0.1764, "step": 13520 }, { "epoch": 6.47, "learning_rate": 9.627324876364658e-05, "loss": 0.1689, "step": 13530 }, { "epoch": 6.48, "learning_rate": 9.62678074611704e-05, "loss": 0.1579, "step": 13540 }, { "epoch": 6.48, "learning_rate": 9.626236234325958e-05, "loss": 0.1764, "step": 13550 }, { "epoch": 6.49, "learning_rate": 9.625691341036314e-05, "loss": 0.1765, "step": 13560 }, { "epoch": 6.49, "learning_rate": 9.625146066293045e-05, "loss": 0.18, "step": 13570 }, { "epoch": 6.49, "learning_rate": 9.624600410141113e-05, "loss": 0.1819, "step": 13580 }, { "epoch": 6.5, "learning_rate": 9.624054372625518e-05, "loss": 0.188, "step": 13590 }, { "epoch": 6.5, "learning_rate": 9.623507953791287e-05, "loss": 0.1488, "step": 13600 }, { "epoch": 6.51, "learning_rate": 9.62296115368348e-05, "loss": 0.1588, "step": 13610 }, { "epoch": 6.51, "learning_rate": 9.622413972347188e-05, "loss": 0.1977, "step": 13620 }, { "epoch": 6.52, "learning_rate": 9.621866409827536e-05, "loss": 0.1786, "step": 13630 }, { "epoch": 6.52, "learning_rate": 9.621318466169675e-05, "loss": 0.1621, "step": 13640 }, { "epoch": 6.53, "learning_rate": 9.620770141418793e-05, "loss": 0.1613, "step": 13650 }, { "epoch": 6.53, "learning_rate": 9.620221435620107e-05, "loss": 0.1636, "step": 13660 }, { "epoch": 6.54, "learning_rate": 9.619672348818864e-05, "loss": 0.1769, "step": 13670 }, { "epoch": 6.54, "learning_rate": 9.619122881060347e-05, "loss": 0.1846, "step": 13680 }, { "epoch": 6.54, "learning_rate": 9.618573032389863e-05, "loss": 0.1731, "step": 13690 }, { "epoch": 6.55, "learning_rate": 9.61802280285276e-05, "loss": 0.146, "step": 13700 }, { "epoch": 6.55, "learning_rate": 9.617472192494408e-05, "loss": 0.1679, "step": 13710 }, { "epoch": 6.56, "learning_rate": 9.616921201360214e-05, "loss": 0.1552, "step": 13720 }, { "epoch": 6.56, "learning_rate": 9.616369829495615e-05, "loss": 0.1852, "step": 13730 }, { "epoch": 6.57, "learning_rate": 9.61581807694608e-05, "loss": 0.2094, "step": 13740 }, { "epoch": 6.57, "learning_rate": 9.615265943757106e-05, "loss": 0.1791, "step": 13750 }, { "epoch": 6.58, "learning_rate": 9.614713429974229e-05, "loss": 0.1822, "step": 13760 }, { "epoch": 6.58, "learning_rate": 9.614160535643008e-05, "loss": 0.1718, "step": 13770 }, { "epoch": 6.59, "learning_rate": 9.613607260809037e-05, "loss": 0.1849, "step": 13780 }, { "epoch": 6.59, "learning_rate": 9.613053605517943e-05, "loss": 0.1458, "step": 13790 }, { "epoch": 6.6, "learning_rate": 9.612499569815383e-05, "loss": 0.1791, "step": 13800 }, { "epoch": 6.6, "learning_rate": 9.611945153747044e-05, "loss": 0.1409, "step": 13810 }, { "epoch": 6.6, "learning_rate": 9.611390357358644e-05, "loss": 0.1627, "step": 13820 }, { "epoch": 6.61, "learning_rate": 9.610835180695936e-05, "loss": 0.1782, "step": 13830 }, { "epoch": 6.61, "learning_rate": 9.610279623804701e-05, "loss": 0.1714, "step": 13840 }, { "epoch": 6.62, "learning_rate": 9.609723686730753e-05, "loss": 0.1712, "step": 13850 }, { "epoch": 6.62, "learning_rate": 9.609167369519938e-05, "loss": 0.1813, "step": 13860 }, { "epoch": 6.63, "learning_rate": 9.608610672218132e-05, "loss": 0.1659, "step": 13870 }, { "epoch": 6.63, "learning_rate": 9.608053594871239e-05, "loss": 0.1639, "step": 13880 }, { "epoch": 6.64, "learning_rate": 9.607496137525202e-05, "loss": 0.1768, "step": 13890 }, { "epoch": 6.64, "learning_rate": 9.606938300225989e-05, "loss": 0.1516, "step": 13900 }, { "epoch": 6.65, "learning_rate": 9.606380083019603e-05, "loss": 0.1655, "step": 13910 }, { "epoch": 6.65, "learning_rate": 9.605821485952075e-05, "loss": 0.1859, "step": 13920 }, { "epoch": 6.65, "learning_rate": 9.605262509069471e-05, "loss": 0.1712, "step": 13930 }, { "epoch": 6.66, "learning_rate": 9.604703152417887e-05, "loss": 0.2167, "step": 13940 }, { "epoch": 6.66, "learning_rate": 9.604143416043449e-05, "loss": 0.1777, "step": 13950 }, { "epoch": 6.67, "learning_rate": 9.603583299992313e-05, "loss": 0.1604, "step": 13960 }, { "epoch": 6.67, "learning_rate": 9.603022804310672e-05, "loss": 0.1616, "step": 13970 }, { "epoch": 6.68, "learning_rate": 9.602461929044744e-05, "loss": 0.1684, "step": 13980 }, { "epoch": 6.68, "learning_rate": 9.601900674240785e-05, "loss": 0.1754, "step": 13990 }, { "epoch": 6.69, "learning_rate": 9.601339039945074e-05, "loss": 0.182, "step": 14000 }, { "epoch": 6.69, "learning_rate": 9.600777026203927e-05, "loss": 0.146, "step": 14010 }, { "epoch": 6.7, "learning_rate": 9.60021463306369e-05, "loss": 0.1586, "step": 14020 }, { "epoch": 6.7, "learning_rate": 9.599651860570743e-05, "loss": 0.1568, "step": 14030 }, { "epoch": 6.7, "learning_rate": 9.599088708771491e-05, "loss": 0.1756, "step": 14040 }, { "epoch": 6.71, "learning_rate": 9.598525177712375e-05, "loss": 0.1522, "step": 14050 }, { "epoch": 6.71, "learning_rate": 9.597961267439867e-05, "loss": 0.1653, "step": 14060 }, { "epoch": 6.72, "learning_rate": 9.597396978000468e-05, "loss": 0.184, "step": 14070 }, { "epoch": 6.72, "learning_rate": 9.596832309440712e-05, "loss": 0.1679, "step": 14080 }, { "epoch": 6.73, "learning_rate": 9.596267261807163e-05, "loss": 0.1529, "step": 14090 }, { "epoch": 6.73, "learning_rate": 9.59570183514642e-05, "loss": 0.1717, "step": 14100 }, { "epoch": 6.74, "learning_rate": 9.595136029505108e-05, "loss": 0.1491, "step": 14110 }, { "epoch": 6.74, "learning_rate": 9.594569844929888e-05, "loss": 0.1539, "step": 14120 }, { "epoch": 6.75, "learning_rate": 9.594003281467445e-05, "loss": 0.1429, "step": 14130 }, { "epoch": 6.75, "learning_rate": 9.593436339164506e-05, "loss": 0.182, "step": 14140 }, { "epoch": 6.76, "learning_rate": 9.59286901806782e-05, "loss": 0.1555, "step": 14150 }, { "epoch": 6.76, "learning_rate": 9.592301318224171e-05, "loss": 0.203, "step": 14160 }, { "epoch": 6.76, "learning_rate": 9.591733239680376e-05, "loss": 0.1758, "step": 14170 }, { "epoch": 6.77, "learning_rate": 9.591164782483277e-05, "loss": 0.16, "step": 14180 }, { "epoch": 6.77, "learning_rate": 9.590595946679755e-05, "loss": 0.1704, "step": 14190 }, { "epoch": 6.78, "learning_rate": 9.59002673231672e-05, "loss": 0.1438, "step": 14200 }, { "epoch": 6.78, "learning_rate": 9.589457139441105e-05, "loss": 0.1525, "step": 14210 }, { "epoch": 6.79, "learning_rate": 9.588887168099887e-05, "loss": 0.1677, "step": 14220 }, { "epoch": 6.79, "learning_rate": 9.588316818340067e-05, "loss": 0.1871, "step": 14230 }, { "epoch": 6.8, "learning_rate": 9.587746090208676e-05, "loss": 0.1645, "step": 14240 }, { "epoch": 6.8, "learning_rate": 9.587174983752781e-05, "loss": 0.1567, "step": 14250 }, { "epoch": 6.81, "learning_rate": 9.586603499019476e-05, "loss": 0.1843, "step": 14260 }, { "epoch": 6.81, "learning_rate": 9.58603163605589e-05, "loss": 0.1556, "step": 14270 }, { "epoch": 6.81, "learning_rate": 9.585459394909181e-05, "loss": 0.1535, "step": 14280 }, { "epoch": 6.82, "learning_rate": 9.584886775626537e-05, "loss": 0.1645, "step": 14290 }, { "epoch": 6.82, "learning_rate": 9.584313778255179e-05, "loss": 0.1641, "step": 14300 }, { "epoch": 6.83, "learning_rate": 9.583740402842359e-05, "loss": 0.1709, "step": 14310 }, { "epoch": 6.83, "learning_rate": 9.58316664943536e-05, "loss": 0.1892, "step": 14320 }, { "epoch": 6.84, "learning_rate": 9.582592518081497e-05, "loss": 0.1699, "step": 14330 }, { "epoch": 6.84, "learning_rate": 9.582018008828113e-05, "loss": 0.1609, "step": 14340 }, { "epoch": 6.85, "learning_rate": 9.581443121722585e-05, "loss": 0.1649, "step": 14350 }, { "epoch": 6.85, "learning_rate": 9.580867856812322e-05, "loss": 0.188, "step": 14360 }, { "epoch": 6.86, "learning_rate": 9.580292214144762e-05, "loss": 0.1865, "step": 14370 }, { "epoch": 6.86, "learning_rate": 9.579716193767374e-05, "loss": 0.1647, "step": 14380 }, { "epoch": 6.86, "learning_rate": 9.57913979572766e-05, "loss": 0.1697, "step": 14390 }, { "epoch": 6.87, "learning_rate": 9.578563020073153e-05, "loss": 0.1644, "step": 14400 }, { "epoch": 6.87, "learning_rate": 9.577985866851413e-05, "loss": 0.1575, "step": 14410 }, { "epoch": 6.88, "learning_rate": 9.577408336110038e-05, "loss": 0.1789, "step": 14420 }, { "epoch": 6.88, "learning_rate": 9.576830427896652e-05, "loss": 0.181, "step": 14430 }, { "epoch": 6.89, "learning_rate": 9.576252142258912e-05, "loss": 0.1856, "step": 14440 }, { "epoch": 6.89, "learning_rate": 9.575673479244505e-05, "loss": 0.1833, "step": 14450 }, { "epoch": 6.9, "learning_rate": 9.575094438901151e-05, "loss": 0.1539, "step": 14460 }, { "epoch": 6.9, "learning_rate": 9.5745150212766e-05, "loss": 0.1462, "step": 14470 }, { "epoch": 6.91, "learning_rate": 9.573935226418634e-05, "loss": 0.1392, "step": 14480 }, { "epoch": 6.91, "learning_rate": 9.573355054375063e-05, "loss": 0.1804, "step": 14490 }, { "epoch": 6.91, "learning_rate": 9.572774505193731e-05, "loss": 0.1661, "step": 14500 }, { "epoch": 6.92, "learning_rate": 9.572193578922515e-05, "loss": 0.1682, "step": 14510 }, { "epoch": 6.92, "learning_rate": 9.571612275609318e-05, "loss": 0.1686, "step": 14520 }, { "epoch": 6.93, "learning_rate": 9.571030595302076e-05, "loss": 0.1757, "step": 14530 }, { "epoch": 6.93, "learning_rate": 9.57044853804876e-05, "loss": 0.168, "step": 14540 }, { "epoch": 6.94, "learning_rate": 9.569866103897365e-05, "loss": 0.1837, "step": 14550 }, { "epoch": 6.94, "learning_rate": 9.569283292895923e-05, "loss": 0.1576, "step": 14560 }, { "epoch": 6.95, "learning_rate": 9.568700105092497e-05, "loss": 0.1697, "step": 14570 }, { "epoch": 6.95, "learning_rate": 9.568116540535175e-05, "loss": 0.1644, "step": 14580 }, { "epoch": 7.0, "learning_rate": 9.567532599272082e-05, "loss": 0.1899, "step": 14590 }, { "epoch": 7.01, "learning_rate": 9.566948281351373e-05, "loss": 0.1709, "step": 14600 }, { "epoch": 7.01, "learning_rate": 9.566363586821233e-05, "loss": 0.1688, "step": 14610 }, { "epoch": 7.02, "learning_rate": 9.565778515729878e-05, "loss": 0.1642, "step": 14620 }, { "epoch": 7.02, "learning_rate": 9.565193068125554e-05, "loss": 0.1359, "step": 14630 }, { "epoch": 7.03, "learning_rate": 9.564607244056543e-05, "loss": 0.1591, "step": 14640 }, { "epoch": 7.03, "learning_rate": 9.56402104357115e-05, "loss": 0.1427, "step": 14650 }, { "epoch": 7.04, "learning_rate": 9.563434466717718e-05, "loss": 0.1592, "step": 14660 }, { "epoch": 7.04, "learning_rate": 9.562847513544618e-05, "loss": 0.1609, "step": 14670 }, { "epoch": 7.05, "learning_rate": 9.562260184100253e-05, "loss": 0.1636, "step": 14680 }, { "epoch": 7.05, "learning_rate": 9.561672478433058e-05, "loss": 0.1663, "step": 14690 }, { "epoch": 7.05, "learning_rate": 9.561084396591494e-05, "loss": 0.1509, "step": 14700 }, { "epoch": 7.06, "learning_rate": 9.56049593862406e-05, "loss": 0.158, "step": 14710 }, { "epoch": 7.06, "learning_rate": 9.559907104579282e-05, "loss": 0.1552, "step": 14720 }, { "epoch": 7.07, "learning_rate": 9.559317894505715e-05, "loss": 0.1406, "step": 14730 }, { "epoch": 7.07, "learning_rate": 9.558728308451952e-05, "loss": 0.1561, "step": 14740 }, { "epoch": 7.08, "learning_rate": 9.558138346466609e-05, "loss": 0.1443, "step": 14750 }, { "epoch": 7.08, "learning_rate": 9.557548008598338e-05, "loss": 0.1568, "step": 14760 }, { "epoch": 7.09, "learning_rate": 9.556957294895822e-05, "loss": 0.1554, "step": 14770 }, { "epoch": 7.09, "learning_rate": 9.556366205407772e-05, "loss": 0.1529, "step": 14780 }, { "epoch": 7.1, "learning_rate": 9.555774740182932e-05, "loss": 0.1615, "step": 14790 }, { "epoch": 7.1, "learning_rate": 9.555182899270079e-05, "loss": 0.1831, "step": 14800 }, { "epoch": 7.1, "learning_rate": 9.554590682718014e-05, "loss": 0.1656, "step": 14810 }, { "epoch": 7.11, "learning_rate": 9.553998090575577e-05, "loss": 0.1571, "step": 14820 }, { "epoch": 7.11, "learning_rate": 9.553405122891637e-05, "loss": 0.1854, "step": 14830 }, { "epoch": 7.12, "learning_rate": 9.552811779715087e-05, "loss": 0.1668, "step": 14840 }, { "epoch": 7.12, "learning_rate": 9.552218061094863e-05, "loss": 0.157, "step": 14850 }, { "epoch": 7.13, "learning_rate": 9.55162396707992e-05, "loss": 0.1667, "step": 14860 }, { "epoch": 7.13, "learning_rate": 9.551029497719253e-05, "loss": 0.1687, "step": 14870 }, { "epoch": 7.14, "learning_rate": 9.550434653061883e-05, "loss": 0.1724, "step": 14880 }, { "epoch": 7.14, "learning_rate": 9.549839433156865e-05, "loss": 0.1589, "step": 14890 }, { "epoch": 7.15, "learning_rate": 9.549243838053279e-05, "loss": 0.1472, "step": 14900 }, { "epoch": 7.15, "learning_rate": 9.548647867800244e-05, "loss": 0.1751, "step": 14910 }, { "epoch": 7.15, "learning_rate": 9.548051522446906e-05, "loss": 0.1416, "step": 14920 }, { "epoch": 7.16, "learning_rate": 9.547454802042441e-05, "loss": 0.1653, "step": 14930 }, { "epoch": 7.16, "learning_rate": 9.546857706636059e-05, "loss": 0.1574, "step": 14940 }, { "epoch": 7.17, "learning_rate": 9.546260236276996e-05, "loss": 0.1729, "step": 14950 }, { "epoch": 7.17, "learning_rate": 9.545662391014524e-05, "loss": 0.1725, "step": 14960 }, { "epoch": 7.18, "learning_rate": 9.545064170897941e-05, "loss": 0.1743, "step": 14970 }, { "epoch": 7.18, "learning_rate": 9.544465575976583e-05, "loss": 0.1645, "step": 14980 }, { "epoch": 7.19, "learning_rate": 9.543866606299811e-05, "loss": 0.1771, "step": 14990 }, { "epoch": 7.19, "learning_rate": 9.543267261917016e-05, "loss": 0.1808, "step": 15000 }, { "epoch": 7.2, "learning_rate": 9.542667542877625e-05, "loss": 0.1706, "step": 15010 }, { "epoch": 7.2, "learning_rate": 9.542067449231093e-05, "loss": 0.1419, "step": 15020 }, { "epoch": 7.21, "learning_rate": 9.541466981026905e-05, "loss": 0.1663, "step": 15030 }, { "epoch": 7.21, "learning_rate": 9.540866138314579e-05, "loss": 0.1547, "step": 15040 }, { "epoch": 7.21, "learning_rate": 9.540264921143664e-05, "loss": 0.1558, "step": 15050 }, { "epoch": 7.22, "learning_rate": 9.539663329563736e-05, "loss": 0.1475, "step": 15060 }, { "epoch": 7.22, "learning_rate": 9.539061363624408e-05, "loss": 0.1511, "step": 15070 }, { "epoch": 7.23, "learning_rate": 9.53845902337532e-05, "loss": 0.1971, "step": 15080 }, { "epoch": 7.23, "learning_rate": 9.53785630886614e-05, "loss": 0.1474, "step": 15090 }, { "epoch": 7.24, "learning_rate": 9.537253220146574e-05, "loss": 0.1615, "step": 15100 }, { "epoch": 7.24, "learning_rate": 9.536649757266354e-05, "loss": 0.1465, "step": 15110 }, { "epoch": 7.25, "learning_rate": 9.536045920275242e-05, "loss": 0.2131, "step": 15120 }, { "epoch": 7.25, "learning_rate": 9.535441709223039e-05, "loss": 0.175, "step": 15130 }, { "epoch": 7.26, "learning_rate": 9.534837124159563e-05, "loss": 0.1554, "step": 15140 }, { "epoch": 7.26, "learning_rate": 9.534232165134678e-05, "loss": 0.1552, "step": 15150 }, { "epoch": 7.26, "learning_rate": 9.533626832198264e-05, "loss": 0.1723, "step": 15160 }, { "epoch": 7.27, "learning_rate": 9.533021125400246e-05, "loss": 0.1549, "step": 15170 }, { "epoch": 7.27, "learning_rate": 9.532415044790568e-05, "loss": 0.1492, "step": 15180 }, { "epoch": 7.28, "learning_rate": 9.531808590419214e-05, "loss": 0.1592, "step": 15190 }, { "epoch": 7.28, "learning_rate": 9.53120176233619e-05, "loss": 0.1509, "step": 15200 }, { "epoch": 7.29, "learning_rate": 9.530594560591544e-05, "loss": 0.1555, "step": 15210 }, { "epoch": 7.29, "learning_rate": 9.529986985235342e-05, "loss": 0.1448, "step": 15220 }, { "epoch": 7.3, "learning_rate": 9.52937903631769e-05, "loss": 0.1705, "step": 15230 }, { "epoch": 7.3, "learning_rate": 9.528770713888721e-05, "loss": 0.1612, "step": 15240 }, { "epoch": 7.31, "learning_rate": 9.528162017998603e-05, "loss": 0.1576, "step": 15250 }, { "epoch": 7.31, "learning_rate": 9.527552948697527e-05, "loss": 0.1746, "step": 15260 }, { "epoch": 7.31, "learning_rate": 9.526943506035723e-05, "loss": 0.1565, "step": 15270 }, { "epoch": 7.32, "learning_rate": 9.526333690063446e-05, "loss": 0.1758, "step": 15280 }, { "epoch": 7.32, "learning_rate": 9.525723500830985e-05, "loss": 0.1762, "step": 15290 }, { "epoch": 7.33, "learning_rate": 9.525112938388658e-05, "loss": 0.1551, "step": 15300 }, { "epoch": 7.33, "learning_rate": 9.524502002786814e-05, "loss": 0.1518, "step": 15310 }, { "epoch": 7.34, "learning_rate": 9.523890694075836e-05, "loss": 0.1704, "step": 15320 }, { "epoch": 7.34, "learning_rate": 9.523279012306132e-05, "loss": 0.1925, "step": 15330 }, { "epoch": 7.35, "learning_rate": 9.522666957528146e-05, "loss": 0.1918, "step": 15340 }, { "epoch": 7.35, "learning_rate": 9.522054529792347e-05, "loss": 0.1457, "step": 15350 }, { "epoch": 7.36, "learning_rate": 9.521441729149244e-05, "loss": 0.1565, "step": 15360 }, { "epoch": 7.36, "learning_rate": 9.520828555649365e-05, "loss": 0.1761, "step": 15370 }, { "epoch": 7.37, "learning_rate": 9.52021500934328e-05, "loss": 0.1732, "step": 15380 }, { "epoch": 7.37, "learning_rate": 9.519601090281583e-05, "loss": 0.1802, "step": 15390 }, { "epoch": 7.37, "learning_rate": 9.518986798514898e-05, "loss": 0.1489, "step": 15400 }, { "epoch": 7.38, "learning_rate": 9.518372134093884e-05, "loss": 0.1663, "step": 15410 }, { "epoch": 7.38, "learning_rate": 9.517757097069229e-05, "loss": 0.1468, "step": 15420 }, { "epoch": 7.39, "learning_rate": 9.51714168749165e-05, "loss": 0.1717, "step": 15430 }, { "epoch": 7.39, "learning_rate": 9.516525905411898e-05, "loss": 0.1749, "step": 15440 }, { "epoch": 7.4, "learning_rate": 9.515909750880752e-05, "loss": 0.1601, "step": 15450 }, { "epoch": 7.4, "learning_rate": 9.515293223949022e-05, "loss": 0.1521, "step": 15460 }, { "epoch": 7.41, "learning_rate": 9.514676324667551e-05, "loss": 0.1387, "step": 15470 }, { "epoch": 7.41, "learning_rate": 9.51405905308721e-05, "loss": 0.1529, "step": 15480 }, { "epoch": 7.42, "learning_rate": 9.513441409258901e-05, "loss": 0.1489, "step": 15490 }, { "epoch": 7.42, "learning_rate": 9.512823393233558e-05, "loss": 0.1804, "step": 15500 }, { "epoch": 7.42, "learning_rate": 9.512205005062149e-05, "loss": 0.1629, "step": 15510 }, { "epoch": 7.43, "learning_rate": 9.511586244795662e-05, "loss": 0.1688, "step": 15520 }, { "epoch": 7.43, "learning_rate": 9.510967112485128e-05, "loss": 0.1463, "step": 15530 }, { "epoch": 7.44, "learning_rate": 9.510347608181599e-05, "loss": 0.152, "step": 15540 }, { "epoch": 7.44, "learning_rate": 9.509727731936165e-05, "loss": 0.1622, "step": 15550 }, { "epoch": 7.45, "learning_rate": 9.509107483799943e-05, "loss": 0.1869, "step": 15560 }, { "epoch": 7.45, "learning_rate": 9.50848686382408e-05, "loss": 0.1576, "step": 15570 }, { "epoch": 7.46, "learning_rate": 9.507865872059757e-05, "loss": 0.1891, "step": 15580 }, { "epoch": 7.46, "learning_rate": 9.507244508558182e-05, "loss": 0.1427, "step": 15590 }, { "epoch": 7.47, "learning_rate": 9.506622773370594e-05, "loss": 0.1636, "step": 15600 }, { "epoch": 7.47, "learning_rate": 9.506000666548267e-05, "loss": 0.1373, "step": 15610 }, { "epoch": 7.47, "learning_rate": 9.5053781881425e-05, "loss": 0.1723, "step": 15620 }, { "epoch": 7.48, "learning_rate": 9.504755338204625e-05, "loss": 0.1864, "step": 15630 }, { "epoch": 7.48, "learning_rate": 9.504132116786006e-05, "loss": 0.1472, "step": 15640 }, { "epoch": 7.49, "learning_rate": 9.503508523938039e-05, "loss": 0.1696, "step": 15650 }, { "epoch": 7.49, "learning_rate": 9.502884559712142e-05, "loss": 0.1535, "step": 15660 }, { "epoch": 7.5, "learning_rate": 9.502260224159774e-05, "loss": 0.1657, "step": 15670 }, { "epoch": 7.5, "learning_rate": 9.50163551733242e-05, "loss": 0.1851, "step": 15680 }, { "epoch": 7.51, "learning_rate": 9.501010439281592e-05, "loss": 0.1407, "step": 15690 }, { "epoch": 7.51, "learning_rate": 9.500384990058841e-05, "loss": 0.1717, "step": 15700 }, { "epoch": 7.52, "learning_rate": 9.499759169715746e-05, "loss": 0.1494, "step": 15710 }, { "epoch": 7.52, "learning_rate": 9.499132978303907e-05, "loss": 0.1572, "step": 15720 }, { "epoch": 7.53, "learning_rate": 9.498506415874968e-05, "loss": 0.1764, "step": 15730 }, { "epoch": 7.53, "learning_rate": 9.497879482480597e-05, "loss": 0.1748, "step": 15740 }, { "epoch": 7.53, "learning_rate": 9.497252178172493e-05, "loss": 0.152, "step": 15750 }, { "epoch": 7.54, "learning_rate": 9.496624503002387e-05, "loss": 0.1527, "step": 15760 }, { "epoch": 7.54, "learning_rate": 9.495996457022038e-05, "loss": 0.1654, "step": 15770 }, { "epoch": 7.55, "learning_rate": 9.495368040283241e-05, "loss": 0.1603, "step": 15780 }, { "epoch": 7.55, "learning_rate": 9.494739252837813e-05, "loss": 0.1706, "step": 15790 }, { "epoch": 7.56, "learning_rate": 9.494110094737607e-05, "loss": 0.1669, "step": 15800 }, { "epoch": 7.56, "learning_rate": 9.493480566034509e-05, "loss": 0.1623, "step": 15810 }, { "epoch": 7.57, "learning_rate": 9.49285066678043e-05, "loss": 0.1495, "step": 15820 }, { "epoch": 7.57, "learning_rate": 9.492220397027318e-05, "loss": 0.1532, "step": 15830 }, { "epoch": 7.58, "learning_rate": 9.491589756827141e-05, "loss": 0.1527, "step": 15840 }, { "epoch": 7.58, "learning_rate": 9.49095874623191e-05, "loss": 0.1405, "step": 15850 }, { "epoch": 7.58, "learning_rate": 9.49032736529366e-05, "loss": 0.1503, "step": 15860 }, { "epoch": 7.59, "learning_rate": 9.489695614064454e-05, "loss": 0.144, "step": 15870 }, { "epoch": 7.59, "learning_rate": 9.489063492596392e-05, "loss": 0.1558, "step": 15880 }, { "epoch": 7.6, "learning_rate": 9.4884310009416e-05, "loss": 0.1308, "step": 15890 }, { "epoch": 7.6, "learning_rate": 9.487798139152235e-05, "loss": 0.1721, "step": 15900 }, { "epoch": 7.61, "learning_rate": 9.487164907280488e-05, "loss": 0.1308, "step": 15910 }, { "epoch": 7.61, "learning_rate": 9.486531305378574e-05, "loss": 0.1322, "step": 15920 }, { "epoch": 7.62, "learning_rate": 9.485897333498747e-05, "loss": 0.17, "step": 15930 }, { "epoch": 7.62, "learning_rate": 9.485262991693284e-05, "loss": 0.1699, "step": 15940 }, { "epoch": 7.63, "learning_rate": 9.484628280014496e-05, "loss": 0.1512, "step": 15950 }, { "epoch": 7.63, "learning_rate": 9.483993198514725e-05, "loss": 0.1635, "step": 15960 }, { "epoch": 7.63, "learning_rate": 9.48335774724634e-05, "loss": 0.1722, "step": 15970 }, { "epoch": 7.64, "learning_rate": 9.482721926261747e-05, "loss": 0.1505, "step": 15980 }, { "epoch": 7.64, "learning_rate": 9.482085735613372e-05, "loss": 0.1582, "step": 15990 }, { "epoch": 7.65, "learning_rate": 9.481449175353685e-05, "loss": 0.1558, "step": 16000 }, { "epoch": 7.65, "learning_rate": 9.480812245535176e-05, "loss": 0.1441, "step": 16010 }, { "epoch": 7.66, "learning_rate": 9.480174946210367e-05, "loss": 0.1698, "step": 16020 }, { "epoch": 7.66, "learning_rate": 9.479537277431816e-05, "loss": 0.1681, "step": 16030 }, { "epoch": 7.67, "learning_rate": 9.478899239252107e-05, "loss": 0.1488, "step": 16040 }, { "epoch": 7.67, "learning_rate": 9.478260831723853e-05, "loss": 0.1835, "step": 16050 }, { "epoch": 7.68, "learning_rate": 9.477622054899702e-05, "loss": 0.1657, "step": 16060 }, { "epoch": 7.68, "learning_rate": 9.47698290883233e-05, "loss": 0.167, "step": 16070 }, { "epoch": 7.69, "learning_rate": 9.476343393574442e-05, "loss": 0.1554, "step": 16080 }, { "epoch": 7.69, "learning_rate": 9.475703509178777e-05, "loss": 0.164, "step": 16090 }, { "epoch": 7.69, "learning_rate": 9.4750632556981e-05, "loss": 0.1691, "step": 16100 }, { "epoch": 7.7, "learning_rate": 9.474422633185212e-05, "loss": 0.1546, "step": 16110 }, { "epoch": 7.7, "learning_rate": 9.473781641692939e-05, "loss": 0.1547, "step": 16120 }, { "epoch": 7.71, "learning_rate": 9.473140281274143e-05, "loss": 0.1408, "step": 16130 }, { "epoch": 7.71, "learning_rate": 9.472498551981708e-05, "loss": 0.1579, "step": 16140 }, { "epoch": 7.72, "learning_rate": 9.471856453868558e-05, "loss": 0.1551, "step": 16150 }, { "epoch": 7.72, "learning_rate": 9.471213986987642e-05, "loss": 0.1783, "step": 16160 }, { "epoch": 7.73, "learning_rate": 9.47057115139194e-05, "loss": 0.1715, "step": 16170 }, { "epoch": 7.73, "learning_rate": 9.469927947134462e-05, "loss": 0.1601, "step": 16180 }, { "epoch": 7.74, "learning_rate": 9.469284374268251e-05, "loss": 0.1598, "step": 16190 }, { "epoch": 7.74, "learning_rate": 9.468640432846378e-05, "loss": 0.1694, "step": 16200 }, { "epoch": 7.74, "learning_rate": 9.467996122921946e-05, "loss": 0.1564, "step": 16210 }, { "epoch": 7.75, "learning_rate": 9.467351444548086e-05, "loss": 0.157, "step": 16220 }, { "epoch": 7.75, "learning_rate": 9.466706397777961e-05, "loss": 0.1536, "step": 16230 }, { "epoch": 7.76, "learning_rate": 9.466060982664765e-05, "loss": 0.1789, "step": 16240 }, { "epoch": 7.76, "learning_rate": 9.465415199261721e-05, "loss": 0.1699, "step": 16250 }, { "epoch": 7.77, "learning_rate": 9.464769047622084e-05, "loss": 0.1695, "step": 16260 }, { "epoch": 7.77, "learning_rate": 9.464122527799135e-05, "loss": 0.1634, "step": 16270 }, { "epoch": 7.78, "learning_rate": 9.463475639846193e-05, "loss": 0.1569, "step": 16280 }, { "epoch": 7.78, "learning_rate": 9.462828383816603e-05, "loss": 0.1673, "step": 16290 }, { "epoch": 7.79, "learning_rate": 9.462180759763737e-05, "loss": 0.1713, "step": 16300 }, { "epoch": 7.79, "learning_rate": 9.461532767741001e-05, "loss": 0.1671, "step": 16310 }, { "epoch": 7.79, "learning_rate": 9.460884407801835e-05, "loss": 0.1574, "step": 16320 }, { "epoch": 7.8, "learning_rate": 9.4602356799997e-05, "loss": 0.1405, "step": 16330 }, { "epoch": 7.8, "learning_rate": 9.459586584388098e-05, "loss": 0.1616, "step": 16340 }, { "epoch": 7.81, "learning_rate": 9.458937121020554e-05, "loss": 0.148, "step": 16350 }, { "epoch": 7.81, "learning_rate": 9.458287289950624e-05, "loss": 0.1677, "step": 16360 }, { "epoch": 7.82, "learning_rate": 9.457637091231899e-05, "loss": 0.1744, "step": 16370 }, { "epoch": 7.82, "learning_rate": 9.456986524917993e-05, "loss": 0.1859, "step": 16380 }, { "epoch": 7.83, "learning_rate": 9.456335591062558e-05, "loss": 0.1621, "step": 16390 }, { "epoch": 7.83, "learning_rate": 9.45568428971927e-05, "loss": 0.1452, "step": 16400 }, { "epoch": 7.84, "learning_rate": 9.45503262094184e-05, "loss": 0.1899, "step": 16410 }, { "epoch": 7.84, "learning_rate": 9.454380584784006e-05, "loss": 0.162, "step": 16420 }, { "epoch": 7.85, "learning_rate": 9.453728181299538e-05, "loss": 0.1555, "step": 16430 }, { "epoch": 7.85, "learning_rate": 9.453075410542236e-05, "loss": 0.1705, "step": 16440 }, { "epoch": 7.85, "learning_rate": 9.45242227256593e-05, "loss": 0.1592, "step": 16450 }, { "epoch": 7.86, "learning_rate": 9.451768767424482e-05, "loss": 0.1682, "step": 16460 }, { "epoch": 7.86, "learning_rate": 9.45111489517178e-05, "loss": 0.1737, "step": 16470 }, { "epoch": 7.87, "learning_rate": 9.450460655861748e-05, "loss": 0.1693, "step": 16480 }, { "epoch": 7.87, "learning_rate": 9.449806049548333e-05, "loss": 0.1468, "step": 16490 }, { "epoch": 7.88, "learning_rate": 9.449151076285521e-05, "loss": 0.1728, "step": 16500 }, { "epoch": 7.88, "learning_rate": 9.448495736127322e-05, "loss": 0.1706, "step": 16510 }, { "epoch": 7.89, "learning_rate": 9.447840029127777e-05, "loss": 0.2014, "step": 16520 }, { "epoch": 7.89, "learning_rate": 9.447183955340962e-05, "loss": 0.1704, "step": 16530 }, { "epoch": 7.9, "learning_rate": 9.446527514820973e-05, "loss": 0.1814, "step": 16540 }, { "epoch": 7.9, "learning_rate": 9.445870707621948e-05, "loss": 0.1749, "step": 16550 }, { "epoch": 7.9, "learning_rate": 9.44521353379805e-05, "loss": 0.1662, "step": 16560 }, { "epoch": 7.91, "learning_rate": 9.44455599340347e-05, "loss": 0.1694, "step": 16570 }, { "epoch": 7.91, "learning_rate": 9.443898086492434e-05, "loss": 0.1801, "step": 16580 }, { "epoch": 7.92, "learning_rate": 9.443239813119192e-05, "loss": 0.1616, "step": 16590 }, { "epoch": 7.92, "learning_rate": 9.442581173338032e-05, "loss": 0.1649, "step": 16600 }, { "epoch": 7.93, "learning_rate": 9.441922167203265e-05, "loss": 0.1609, "step": 16610 }, { "epoch": 7.93, "learning_rate": 9.441262794769238e-05, "loss": 0.1746, "step": 16620 }, { "epoch": 7.94, "learning_rate": 9.440603056090323e-05, "loss": 0.1456, "step": 16630 }, { "epoch": 7.94, "learning_rate": 9.439942951220928e-05, "loss": 0.1523, "step": 16640 }, { "epoch": 7.95, "learning_rate": 9.439282480215484e-05, "loss": 0.1617, "step": 16650 }, { "epoch": 7.95, "learning_rate": 9.438621643128458e-05, "loss": 0.1585, "step": 16660 }, { "epoch": 8.0, "learning_rate": 9.437960440014346e-05, "loss": 0.1465, "step": 16670 }, { "epoch": 8.01, "learning_rate": 9.437298870927675e-05, "loss": 0.1754, "step": 16680 }, { "epoch": 8.01, "learning_rate": 9.436636935922997e-05, "loss": 0.1482, "step": 16690 }, { "epoch": 8.02, "learning_rate": 9.4359746350549e-05, "loss": 0.1517, "step": 16700 }, { "epoch": 8.02, "learning_rate": 9.435311968378e-05, "loss": 0.145, "step": 16710 }, { "epoch": 8.03, "learning_rate": 9.434648935946942e-05, "loss": 0.15, "step": 16720 }, { "epoch": 8.03, "learning_rate": 9.433985537816404e-05, "loss": 0.1329, "step": 16730 }, { "epoch": 8.03, "learning_rate": 9.433321774041093e-05, "loss": 0.1329, "step": 16740 }, { "epoch": 8.04, "learning_rate": 9.432657644675744e-05, "loss": 0.1494, "step": 16750 }, { "epoch": 8.04, "learning_rate": 9.431993149775125e-05, "loss": 0.1455, "step": 16760 }, { "epoch": 8.05, "learning_rate": 9.43132828939403e-05, "loss": 0.1518, "step": 16770 }, { "epoch": 8.05, "learning_rate": 9.430663063587291e-05, "loss": 0.1544, "step": 16780 }, { "epoch": 8.06, "learning_rate": 9.429997472409763e-05, "loss": 0.1682, "step": 16790 }, { "epoch": 8.06, "learning_rate": 9.429331515916333e-05, "loss": 0.1373, "step": 16800 }, { "epoch": 8.07, "learning_rate": 9.428665194161919e-05, "loss": 0.1475, "step": 16810 }, { "epoch": 8.07, "learning_rate": 9.427998507201468e-05, "loss": 0.1489, "step": 16820 }, { "epoch": 8.08, "learning_rate": 9.427331455089958e-05, "loss": 0.1609, "step": 16830 }, { "epoch": 8.08, "learning_rate": 9.426664037882398e-05, "loss": 0.1508, "step": 16840 }, { "epoch": 8.09, "learning_rate": 9.425996255633824e-05, "loss": 0.1458, "step": 16850 }, { "epoch": 8.09, "learning_rate": 9.425328108399306e-05, "loss": 0.167, "step": 16860 }, { "epoch": 8.09, "learning_rate": 9.424659596233942e-05, "loss": 0.1532, "step": 16870 }, { "epoch": 8.1, "learning_rate": 9.423990719192858e-05, "loss": 0.1486, "step": 16880 }, { "epoch": 8.1, "learning_rate": 9.423321477331215e-05, "loss": 0.14, "step": 16890 }, { "epoch": 8.11, "learning_rate": 9.4226518707042e-05, "loss": 0.1396, "step": 16900 }, { "epoch": 8.11, "learning_rate": 9.421981899367034e-05, "loss": 0.1475, "step": 16910 }, { "epoch": 8.12, "learning_rate": 9.421311563374962e-05, "loss": 0.1614, "step": 16920 }, { "epoch": 8.12, "learning_rate": 9.420640862783263e-05, "loss": 0.1689, "step": 16930 }, { "epoch": 8.13, "learning_rate": 9.419969797647249e-05, "loss": 0.1475, "step": 16940 }, { "epoch": 8.13, "learning_rate": 9.419298368022256e-05, "loss": 0.148, "step": 16950 }, { "epoch": 8.14, "learning_rate": 9.418626573963654e-05, "loss": 0.1529, "step": 16960 }, { "epoch": 8.14, "learning_rate": 9.417954415526842e-05, "loss": 0.1383, "step": 16970 }, { "epoch": 8.14, "learning_rate": 9.417281892767247e-05, "loss": 0.1624, "step": 16980 }, { "epoch": 8.15, "learning_rate": 9.416609005740331e-05, "loss": 0.1433, "step": 16990 }, { "epoch": 8.15, "learning_rate": 9.415935754501582e-05, "loss": 0.1583, "step": 17000 }, { "epoch": 8.16, "learning_rate": 9.415262139106517e-05, "loss": 0.1413, "step": 17010 }, { "epoch": 8.16, "learning_rate": 9.414588159610688e-05, "loss": 0.1723, "step": 17020 }, { "epoch": 8.17, "learning_rate": 9.413913816069672e-05, "loss": 0.1387, "step": 17030 }, { "epoch": 8.17, "learning_rate": 9.41323910853908e-05, "loss": 0.1614, "step": 17040 }, { "epoch": 8.18, "learning_rate": 9.41256403707455e-05, "loss": 0.1329, "step": 17050 }, { "epoch": 8.18, "learning_rate": 9.41188860173175e-05, "loss": 0.1348, "step": 17060 }, { "epoch": 8.19, "learning_rate": 9.411212802566382e-05, "loss": 0.1724, "step": 17070 }, { "epoch": 8.19, "learning_rate": 9.410536639634175e-05, "loss": 0.1299, "step": 17080 }, { "epoch": 8.19, "learning_rate": 9.409860112990886e-05, "loss": 0.1504, "step": 17090 }, { "epoch": 8.2, "learning_rate": 9.409183222692305e-05, "loss": 0.1407, "step": 17100 }, { "epoch": 8.2, "learning_rate": 9.408505968794253e-05, "loss": 0.1425, "step": 17110 }, { "epoch": 8.21, "learning_rate": 9.407828351352574e-05, "loss": 0.1547, "step": 17120 }, { "epoch": 8.21, "learning_rate": 9.407150370423155e-05, "loss": 0.1627, "step": 17130 }, { "epoch": 8.22, "learning_rate": 9.406472026061898e-05, "loss": 0.1501, "step": 17140 }, { "epoch": 8.22, "learning_rate": 9.405793318324744e-05, "loss": 0.1662, "step": 17150 }, { "epoch": 8.23, "learning_rate": 9.405114247267665e-05, "loss": 0.1619, "step": 17160 }, { "epoch": 8.23, "learning_rate": 9.404434812946658e-05, "loss": 0.1565, "step": 17170 }, { "epoch": 8.24, "learning_rate": 9.403755015417751e-05, "loss": 0.1455, "step": 17180 }, { "epoch": 8.24, "learning_rate": 9.403074854737004e-05, "loss": 0.146, "step": 17190 }, { "epoch": 8.24, "learning_rate": 9.402394330960506e-05, "loss": 0.1453, "step": 17200 }, { "epoch": 8.25, "learning_rate": 9.401713444144374e-05, "loss": 0.141, "step": 17210 }, { "epoch": 8.25, "learning_rate": 9.40103219434476e-05, "loss": 0.158, "step": 17220 }, { "epoch": 8.26, "learning_rate": 9.40035058161784e-05, "loss": 0.1747, "step": 17230 }, { "epoch": 8.26, "learning_rate": 9.399668606019823e-05, "loss": 0.1538, "step": 17240 }, { "epoch": 8.27, "learning_rate": 9.398986267606948e-05, "loss": 0.1572, "step": 17250 }, { "epoch": 8.27, "learning_rate": 9.398303566435485e-05, "loss": 0.1622, "step": 17260 }, { "epoch": 8.28, "learning_rate": 9.39762050256173e-05, "loss": 0.1457, "step": 17270 }, { "epoch": 8.28, "learning_rate": 9.396937076042012e-05, "loss": 0.1586, "step": 17280 }, { "epoch": 8.29, "learning_rate": 9.396253286932689e-05, "loss": 0.1423, "step": 17290 }, { "epoch": 8.29, "learning_rate": 9.39556913529015e-05, "loss": 0.1427, "step": 17300 }, { "epoch": 8.3, "learning_rate": 9.394884621170812e-05, "loss": 0.1365, "step": 17310 }, { "epoch": 8.3, "learning_rate": 9.394199744631123e-05, "loss": 0.128, "step": 17320 }, { "epoch": 8.3, "learning_rate": 9.39351450572756e-05, "loss": 0.1447, "step": 17330 }, { "epoch": 8.31, "learning_rate": 9.392828904516633e-05, "loss": 0.1559, "step": 17340 }, { "epoch": 8.31, "learning_rate": 9.392142941054878e-05, "loss": 0.1334, "step": 17350 }, { "epoch": 8.32, "learning_rate": 9.391456615398861e-05, "loss": 0.1611, "step": 17360 }, { "epoch": 8.32, "learning_rate": 9.390769927605181e-05, "loss": 0.172, "step": 17370 }, { "epoch": 8.33, "learning_rate": 9.390082877730466e-05, "loss": 0.1507, "step": 17380 }, { "epoch": 8.33, "learning_rate": 9.389395465831371e-05, "loss": 0.1645, "step": 17390 }, { "epoch": 8.34, "learning_rate": 9.388707691964584e-05, "loss": 0.1656, "step": 17400 }, { "epoch": 8.34, "learning_rate": 9.388019556186822e-05, "loss": 0.1674, "step": 17410 }, { "epoch": 8.35, "learning_rate": 9.387331058554829e-05, "loss": 0.1452, "step": 17420 }, { "epoch": 8.35, "learning_rate": 9.386642199125384e-05, "loss": 0.1699, "step": 17430 }, { "epoch": 8.35, "learning_rate": 9.385952977955292e-05, "loss": 0.1692, "step": 17440 }, { "epoch": 8.36, "learning_rate": 9.38526339510139e-05, "loss": 0.1556, "step": 17450 }, { "epoch": 8.36, "learning_rate": 9.384573450620543e-05, "loss": 0.1492, "step": 17460 }, { "epoch": 8.37, "learning_rate": 9.383883144569648e-05, "loss": 0.1454, "step": 17470 }, { "epoch": 8.37, "learning_rate": 9.383192477005627e-05, "loss": 0.1556, "step": 17480 }, { "epoch": 8.38, "learning_rate": 9.382501447985439e-05, "loss": 0.1512, "step": 17490 }, { "epoch": 8.38, "learning_rate": 9.381810057566068e-05, "loss": 0.1603, "step": 17500 }, { "epoch": 8.39, "learning_rate": 9.381118305804528e-05, "loss": 0.142, "step": 17510 }, { "epoch": 8.39, "learning_rate": 9.380426192757865e-05, "loss": 0.1736, "step": 17520 }, { "epoch": 8.4, "learning_rate": 9.379733718483154e-05, "loss": 0.1545, "step": 17530 }, { "epoch": 8.4, "learning_rate": 9.379040883037497e-05, "loss": 0.1447, "step": 17540 }, { "epoch": 8.4, "learning_rate": 9.378347686478028e-05, "loss": 0.1463, "step": 17550 }, { "epoch": 8.41, "learning_rate": 9.377654128861914e-05, "loss": 0.1542, "step": 17560 }, { "epoch": 8.41, "learning_rate": 9.376960210246346e-05, "loss": 0.1667, "step": 17570 }, { "epoch": 8.42, "learning_rate": 9.376265930688548e-05, "loss": 0.1373, "step": 17580 }, { "epoch": 8.42, "learning_rate": 9.375571290245774e-05, "loss": 0.1743, "step": 17590 }, { "epoch": 8.43, "learning_rate": 9.374876288975308e-05, "loss": 0.1664, "step": 17600 }, { "epoch": 8.43, "learning_rate": 9.374180926934458e-05, "loss": 0.1719, "step": 17610 }, { "epoch": 8.44, "learning_rate": 9.37348520418057e-05, "loss": 0.1651, "step": 17620 }, { "epoch": 8.44, "learning_rate": 9.372789120771016e-05, "loss": 0.1608, "step": 17630 }, { "epoch": 8.45, "learning_rate": 9.372092676763197e-05, "loss": 0.1877, "step": 17640 }, { "epoch": 8.45, "learning_rate": 9.371395872214546e-05, "loss": 0.1474, "step": 17650 }, { "epoch": 8.46, "learning_rate": 9.370698707182524e-05, "loss": 0.1383, "step": 17660 }, { "epoch": 8.46, "learning_rate": 9.37000118172462e-05, "loss": 0.1742, "step": 17670 }, { "epoch": 8.46, "learning_rate": 9.369303295898356e-05, "loss": 0.1724, "step": 17680 }, { "epoch": 8.47, "learning_rate": 9.368605049761284e-05, "loss": 0.1448, "step": 17690 }, { "epoch": 8.47, "learning_rate": 9.367906443370984e-05, "loss": 0.1619, "step": 17700 }, { "epoch": 8.48, "learning_rate": 9.367207476785065e-05, "loss": 0.1689, "step": 17710 }, { "epoch": 8.48, "learning_rate": 9.366508150061166e-05, "loss": 0.1699, "step": 17720 }, { "epoch": 8.49, "learning_rate": 9.365808463256958e-05, "loss": 0.1493, "step": 17730 }, { "epoch": 8.49, "learning_rate": 9.365108416430137e-05, "loss": 0.1502, "step": 17740 }, { "epoch": 8.5, "learning_rate": 9.364408009638437e-05, "loss": 0.1396, "step": 17750 }, { "epoch": 8.5, "learning_rate": 9.363707242939612e-05, "loss": 0.1262, "step": 17760 }, { "epoch": 8.51, "learning_rate": 9.363006116391453e-05, "loss": 0.1649, "step": 17770 }, { "epoch": 8.51, "learning_rate": 9.362304630051775e-05, "loss": 0.1461, "step": 17780 }, { "epoch": 8.51, "learning_rate": 9.361602783978428e-05, "loss": 0.1367, "step": 17790 }, { "epoch": 8.52, "learning_rate": 9.360900578229288e-05, "loss": 0.1581, "step": 17800 }, { "epoch": 8.52, "learning_rate": 9.36019801286226e-05, "loss": 0.1462, "step": 17810 }, { "epoch": 8.53, "learning_rate": 9.359495087935284e-05, "loss": 0.1587, "step": 17820 }, { "epoch": 8.53, "learning_rate": 9.358791803506324e-05, "loss": 0.1499, "step": 17830 }, { "epoch": 8.54, "learning_rate": 9.358088159633376e-05, "loss": 0.1624, "step": 17840 }, { "epoch": 8.54, "learning_rate": 9.357384156374464e-05, "loss": 0.1569, "step": 17850 }, { "epoch": 8.55, "learning_rate": 9.356679793787646e-05, "loss": 0.1465, "step": 17860 }, { "epoch": 8.55, "learning_rate": 9.355975071931006e-05, "loss": 0.149, "step": 17870 }, { "epoch": 8.56, "learning_rate": 9.355269990862657e-05, "loss": 0.1612, "step": 17880 }, { "epoch": 8.56, "learning_rate": 9.354564550640741e-05, "loss": 0.1537, "step": 17890 }, { "epoch": 8.56, "learning_rate": 9.353858751323436e-05, "loss": 0.1484, "step": 17900 }, { "epoch": 8.57, "learning_rate": 9.353152592968943e-05, "loss": 0.1729, "step": 17910 }, { "epoch": 8.57, "learning_rate": 9.352446075635495e-05, "loss": 0.135, "step": 17920 }, { "epoch": 8.58, "learning_rate": 9.351739199381353e-05, "loss": 0.126, "step": 17930 }, { "epoch": 8.58, "learning_rate": 9.351031964264809e-05, "loss": 0.1559, "step": 17940 }, { "epoch": 8.59, "learning_rate": 9.350324370344188e-05, "loss": 0.1537, "step": 17950 }, { "epoch": 8.59, "learning_rate": 9.349616417677837e-05, "loss": 0.1429, "step": 17960 }, { "epoch": 8.6, "learning_rate": 9.348908106324139e-05, "loss": 0.1438, "step": 17970 }, { "epoch": 8.6, "learning_rate": 9.348199436341502e-05, "loss": 0.1521, "step": 17980 }, { "epoch": 8.61, "learning_rate": 9.34749040778837e-05, "loss": 0.145, "step": 17990 }, { "epoch": 8.61, "learning_rate": 9.346781020723209e-05, "loss": 0.1484, "step": 18000 }, { "epoch": 8.62, "learning_rate": 9.346071275204519e-05, "loss": 0.1542, "step": 18010 }, { "epoch": 8.62, "learning_rate": 9.345361171290827e-05, "loss": 0.1561, "step": 18020 }, { "epoch": 8.62, "learning_rate": 9.344650709040694e-05, "loss": 0.1654, "step": 18030 }, { "epoch": 8.63, "learning_rate": 9.343939888512705e-05, "loss": 0.1487, "step": 18040 }, { "epoch": 8.63, "learning_rate": 9.34322870976548e-05, "loss": 0.1516, "step": 18050 }, { "epoch": 8.64, "learning_rate": 9.342517172857664e-05, "loss": 0.1673, "step": 18060 }, { "epoch": 8.64, "learning_rate": 9.341805277847932e-05, "loss": 0.1639, "step": 18070 }, { "epoch": 8.65, "learning_rate": 9.341093024794991e-05, "loss": 0.1513, "step": 18080 }, { "epoch": 8.65, "learning_rate": 9.340380413757578e-05, "loss": 0.1369, "step": 18090 }, { "epoch": 8.66, "learning_rate": 9.339667444794455e-05, "loss": 0.1629, "step": 18100 }, { "epoch": 8.66, "learning_rate": 9.338954117964419e-05, "loss": 0.1877, "step": 18110 }, { "epoch": 8.67, "learning_rate": 9.338240433326292e-05, "loss": 0.1837, "step": 18120 }, { "epoch": 8.67, "learning_rate": 9.337526390938929e-05, "loss": 0.1287, "step": 18130 }, { "epoch": 8.67, "learning_rate": 9.336811990861211e-05, "loss": 0.1474, "step": 18140 }, { "epoch": 8.68, "learning_rate": 9.33609723315205e-05, "loss": 0.1459, "step": 18150 }, { "epoch": 8.68, "learning_rate": 9.335382117870392e-05, "loss": 0.1567, "step": 18160 }, { "epoch": 8.69, "learning_rate": 9.334666645075203e-05, "loss": 0.1372, "step": 18170 }, { "epoch": 8.69, "learning_rate": 9.333950814825487e-05, "loss": 0.1382, "step": 18180 }, { "epoch": 8.7, "learning_rate": 9.333234627180274e-05, "loss": 0.1484, "step": 18190 }, { "epoch": 8.7, "learning_rate": 9.332518082198624e-05, "loss": 0.1509, "step": 18200 }, { "epoch": 8.71, "learning_rate": 9.331801179939625e-05, "loss": 0.1816, "step": 18210 }, { "epoch": 8.71, "learning_rate": 9.331083920462398e-05, "loss": 0.1452, "step": 18220 }, { "epoch": 8.72, "learning_rate": 9.330366303826088e-05, "loss": 0.1643, "step": 18230 }, { "epoch": 8.72, "learning_rate": 9.329648330089874e-05, "loss": 0.168, "step": 18240 }, { "epoch": 8.72, "learning_rate": 9.328929999312964e-05, "loss": 0.1304, "step": 18250 }, { "epoch": 8.73, "learning_rate": 9.328211311554595e-05, "loss": 0.1667, "step": 18260 }, { "epoch": 8.73, "learning_rate": 9.32749226687403e-05, "loss": 0.1675, "step": 18270 }, { "epoch": 8.74, "learning_rate": 9.326772865330566e-05, "loss": 0.1463, "step": 18280 }, { "epoch": 8.74, "learning_rate": 9.32605310698353e-05, "loss": 0.1461, "step": 18290 }, { "epoch": 8.75, "learning_rate": 9.325332991892273e-05, "loss": 0.1579, "step": 18300 }, { "epoch": 8.75, "learning_rate": 9.324612520116181e-05, "loss": 0.1476, "step": 18310 }, { "epoch": 8.76, "learning_rate": 9.323891691714665e-05, "loss": 0.1622, "step": 18320 }, { "epoch": 8.76, "learning_rate": 9.32317050674717e-05, "loss": 0.1614, "step": 18330 }, { "epoch": 8.77, "learning_rate": 9.322448965273165e-05, "loss": 0.1576, "step": 18340 }, { "epoch": 8.77, "learning_rate": 9.321727067352153e-05, "loss": 0.197, "step": 18350 }, { "epoch": 8.78, "learning_rate": 9.321004813043666e-05, "loss": 0.1647, "step": 18360 }, { "epoch": 8.78, "learning_rate": 9.32028220240726e-05, "loss": 0.161, "step": 18370 }, { "epoch": 8.78, "learning_rate": 9.319559235502528e-05, "loss": 0.154, "step": 18380 }, { "epoch": 8.79, "learning_rate": 9.318835912389088e-05, "loss": 0.1269, "step": 18390 }, { "epoch": 8.79, "learning_rate": 9.318112233126589e-05, "loss": 0.1525, "step": 18400 }, { "epoch": 8.8, "learning_rate": 9.317388197774706e-05, "loss": 0.1384, "step": 18410 }, { "epoch": 8.8, "learning_rate": 9.316663806393148e-05, "loss": 0.1424, "step": 18420 }, { "epoch": 8.81, "learning_rate": 9.315939059041649e-05, "loss": 0.1627, "step": 18430 }, { "epoch": 8.81, "learning_rate": 9.31521395577998e-05, "loss": 0.1399, "step": 18440 }, { "epoch": 8.82, "learning_rate": 9.31448849666793e-05, "loss": 0.1613, "step": 18450 }, { "epoch": 8.82, "learning_rate": 9.313762681765328e-05, "loss": 0.1585, "step": 18460 }, { "epoch": 8.83, "learning_rate": 9.313036511132024e-05, "loss": 0.1674, "step": 18470 }, { "epoch": 8.83, "learning_rate": 9.312309984827903e-05, "loss": 0.1588, "step": 18480 }, { "epoch": 8.83, "learning_rate": 9.311583102912878e-05, "loss": 0.1428, "step": 18490 }, { "epoch": 8.84, "learning_rate": 9.310855865446888e-05, "loss": 0.1346, "step": 18500 }, { "epoch": 8.84, "learning_rate": 9.310128272489906e-05, "loss": 0.1524, "step": 18510 }, { "epoch": 8.85, "learning_rate": 9.309400324101932e-05, "loss": 0.1406, "step": 18520 }, { "epoch": 8.85, "learning_rate": 9.308672020342997e-05, "loss": 0.1392, "step": 18530 }, { "epoch": 8.86, "learning_rate": 9.307943361273157e-05, "loss": 0.1862, "step": 18540 }, { "epoch": 8.86, "learning_rate": 9.307214346952502e-05, "loss": 0.1567, "step": 18550 }, { "epoch": 8.87, "learning_rate": 9.30648497744115e-05, "loss": 0.1727, "step": 18560 }, { "epoch": 8.87, "learning_rate": 9.305755252799246e-05, "loss": 0.1602, "step": 18570 }, { "epoch": 8.88, "learning_rate": 9.30502517308697e-05, "loss": 0.1924, "step": 18580 }, { "epoch": 8.88, "learning_rate": 9.304294738364521e-05, "loss": 0.147, "step": 18590 }, { "epoch": 8.88, "learning_rate": 9.30356394869214e-05, "loss": 0.1625, "step": 18600 }, { "epoch": 8.89, "learning_rate": 9.302832804130087e-05, "loss": 0.1548, "step": 18610 }, { "epoch": 8.89, "learning_rate": 9.302101304738658e-05, "loss": 0.157, "step": 18620 }, { "epoch": 8.9, "learning_rate": 9.301369450578173e-05, "loss": 0.2302, "step": 18630 }, { "epoch": 8.9, "learning_rate": 9.300637241708983e-05, "loss": 0.1471, "step": 18640 }, { "epoch": 8.91, "learning_rate": 9.299904678191472e-05, "loss": 0.1737, "step": 18650 }, { "epoch": 8.91, "learning_rate": 9.299171760086049e-05, "loss": 0.164, "step": 18660 }, { "epoch": 8.92, "learning_rate": 9.298438487453154e-05, "loss": 0.1533, "step": 18670 }, { "epoch": 8.92, "learning_rate": 9.297704860353254e-05, "loss": 0.1771, "step": 18680 }, { "epoch": 8.93, "learning_rate": 9.296970878846847e-05, "loss": 0.1402, "step": 18690 }, { "epoch": 8.93, "learning_rate": 9.296236542994463e-05, "loss": 0.1472, "step": 18700 }, { "epoch": 8.94, "learning_rate": 9.295501852856655e-05, "loss": 0.1889, "step": 18710 }, { "epoch": 8.94, "learning_rate": 9.294766808494012e-05, "loss": 0.1501, "step": 18720 }, { "epoch": 8.94, "learning_rate": 9.294031409967143e-05, "loss": 0.1876, "step": 18730 }, { "epoch": 8.95, "learning_rate": 9.293295657336698e-05, "loss": 0.1459, "step": 18740 }, { "epoch": 9.0, "learning_rate": 9.292559550663349e-05, "loss": 0.1419, "step": 18750 }, { "epoch": 9.01, "learning_rate": 9.291823090007797e-05, "loss": 0.1749, "step": 18760 }, { "epoch": 9.01, "learning_rate": 9.291086275430774e-05, "loss": 0.1403, "step": 18770 }, { "epoch": 9.02, "learning_rate": 9.29034910699304e-05, "loss": 0.1426, "step": 18780 }, { "epoch": 9.02, "learning_rate": 9.289611584755385e-05, "loss": 0.1404, "step": 18790 }, { "epoch": 9.02, "learning_rate": 9.288873708778631e-05, "loss": 0.1249, "step": 18800 }, { "epoch": 9.03, "learning_rate": 9.288135479123623e-05, "loss": 0.16, "step": 18810 }, { "epoch": 9.03, "learning_rate": 9.287396895851239e-05, "loss": 0.1379, "step": 18820 }, { "epoch": 9.04, "learning_rate": 9.286657959022387e-05, "loss": 0.1419, "step": 18830 }, { "epoch": 9.04, "learning_rate": 9.285918668698002e-05, "loss": 0.1311, "step": 18840 }, { "epoch": 9.05, "learning_rate": 9.285179024939048e-05, "loss": 0.121, "step": 18850 }, { "epoch": 9.05, "learning_rate": 9.284439027806522e-05, "loss": 0.1383, "step": 18860 }, { "epoch": 9.06, "learning_rate": 9.283698677361444e-05, "loss": 0.1538, "step": 18870 }, { "epoch": 9.06, "learning_rate": 9.282957973664868e-05, "loss": 0.1391, "step": 18880 }, { "epoch": 9.07, "learning_rate": 9.282216916777874e-05, "loss": 0.1494, "step": 18890 }, { "epoch": 9.07, "learning_rate": 9.281475506761577e-05, "loss": 0.1588, "step": 18900 }, { "epoch": 9.07, "learning_rate": 9.28073374367711e-05, "loss": 0.1316, "step": 18910 }, { "epoch": 9.08, "learning_rate": 9.279991627585648e-05, "loss": 0.1547, "step": 18920 }, { "epoch": 9.08, "learning_rate": 9.279249158548384e-05, "loss": 0.1456, "step": 18930 }, { "epoch": 9.09, "learning_rate": 9.27850633662655e-05, "loss": 0.1662, "step": 18940 }, { "epoch": 9.09, "learning_rate": 9.277763161881397e-05, "loss": 0.1424, "step": 18950 }, { "epoch": 9.1, "learning_rate": 9.277019634374214e-05, "loss": 0.1562, "step": 18960 }, { "epoch": 9.1, "learning_rate": 9.276275754166312e-05, "loss": 0.1545, "step": 18970 }, { "epoch": 9.11, "learning_rate": 9.275531521319039e-05, "loss": 0.1278, "step": 18980 }, { "epoch": 9.11, "learning_rate": 9.274786935893765e-05, "loss": 0.1536, "step": 18990 }, { "epoch": 9.12, "learning_rate": 9.27404199795189e-05, "loss": 0.1203, "step": 19000 }, { "epoch": 9.12, "learning_rate": 9.273296707554849e-05, "loss": 0.1436, "step": 19010 }, { "epoch": 9.12, "learning_rate": 9.272551064764097e-05, "loss": 0.1367, "step": 19020 }, { "epoch": 9.13, "learning_rate": 9.271805069641125e-05, "loss": 0.142, "step": 19030 }, { "epoch": 9.13, "learning_rate": 9.271058722247452e-05, "loss": 0.1538, "step": 19040 }, { "epoch": 9.14, "learning_rate": 9.270312022644623e-05, "loss": 0.1363, "step": 19050 }, { "epoch": 9.14, "learning_rate": 9.269564970894213e-05, "loss": 0.1477, "step": 19060 }, { "epoch": 9.15, "learning_rate": 9.268817567057831e-05, "loss": 0.1412, "step": 19070 }, { "epoch": 9.15, "learning_rate": 9.268069811197106e-05, "loss": 0.1294, "step": 19080 }, { "epoch": 9.16, "learning_rate": 9.267321703373705e-05, "loss": 0.1553, "step": 19090 }, { "epoch": 9.16, "learning_rate": 9.266573243649319e-05, "loss": 0.1319, "step": 19100 }, { "epoch": 9.17, "learning_rate": 9.265824432085668e-05, "loss": 0.1284, "step": 19110 }, { "epoch": 9.17, "learning_rate": 9.265075268744503e-05, "loss": 0.1607, "step": 19120 }, { "epoch": 9.18, "learning_rate": 9.264325753687602e-05, "loss": 0.1265, "step": 19130 }, { "epoch": 9.18, "learning_rate": 9.263575886976775e-05, "loss": 0.1254, "step": 19140 }, { "epoch": 9.18, "learning_rate": 9.262825668673858e-05, "loss": 0.1644, "step": 19150 }, { "epoch": 9.19, "learning_rate": 9.262075098840717e-05, "loss": 0.1299, "step": 19160 }, { "epoch": 9.19, "learning_rate": 9.261324177539247e-05, "loss": 0.1597, "step": 19170 }, { "epoch": 9.2, "learning_rate": 9.260572904831373e-05, "loss": 0.1336, "step": 19180 }, { "epoch": 9.2, "learning_rate": 9.259821280779047e-05, "loss": 0.1413, "step": 19190 }, { "epoch": 9.21, "learning_rate": 9.259069305444252e-05, "loss": 0.1475, "step": 19200 }, { "epoch": 9.21, "learning_rate": 9.258316978888998e-05, "loss": 0.1509, "step": 19210 }, { "epoch": 9.22, "learning_rate": 9.257564301175326e-05, "loss": 0.1329, "step": 19220 }, { "epoch": 9.22, "learning_rate": 9.256811272365303e-05, "loss": 0.1481, "step": 19230 }, { "epoch": 9.23, "learning_rate": 9.256057892521028e-05, "loss": 0.1639, "step": 19240 }, { "epoch": 9.23, "learning_rate": 9.25530416170463e-05, "loss": 0.151, "step": 19250 }, { "epoch": 9.23, "learning_rate": 9.254550079978261e-05, "loss": 0.1572, "step": 19260 }, { "epoch": 9.24, "learning_rate": 9.253795647404108e-05, "loss": 0.1463, "step": 19270 }, { "epoch": 9.24, "learning_rate": 9.253040864044385e-05, "loss": 0.162, "step": 19280 }, { "epoch": 9.25, "learning_rate": 9.252285729961335e-05, "loss": 0.1501, "step": 19290 }, { "epoch": 9.25, "learning_rate": 9.251530245217225e-05, "loss": 0.1361, "step": 19300 }, { "epoch": 9.26, "learning_rate": 9.250774409874359e-05, "loss": 0.1358, "step": 19310 }, { "epoch": 9.26, "learning_rate": 9.250018223995067e-05, "loss": 0.1688, "step": 19320 }, { "epoch": 9.27, "learning_rate": 9.249261687641706e-05, "loss": 0.1564, "step": 19330 }, { "epoch": 9.27, "learning_rate": 9.248504800876662e-05, "loss": 0.1479, "step": 19340 }, { "epoch": 9.28, "learning_rate": 9.247747563762353e-05, "loss": 0.1276, "step": 19350 }, { "epoch": 9.28, "learning_rate": 9.246989976361222e-05, "loss": 0.1614, "step": 19360 }, { "epoch": 9.28, "learning_rate": 9.246232038735744e-05, "loss": 0.1487, "step": 19370 }, { "epoch": 9.29, "learning_rate": 9.245473750948423e-05, "loss": 0.1339, "step": 19380 }, { "epoch": 9.29, "learning_rate": 9.24471511306179e-05, "loss": 0.1493, "step": 19390 }, { "epoch": 9.3, "learning_rate": 9.243956125138402e-05, "loss": 0.1482, "step": 19400 }, { "epoch": 9.3, "learning_rate": 9.24319678724085e-05, "loss": 0.1339, "step": 19410 }, { "epoch": 9.31, "learning_rate": 9.242437099431755e-05, "loss": 0.139, "step": 19420 }, { "epoch": 9.31, "learning_rate": 9.241677061773762e-05, "loss": 0.1207, "step": 19430 }, { "epoch": 9.32, "learning_rate": 9.240916674329547e-05, "loss": 0.1426, "step": 19440 }, { "epoch": 9.32, "learning_rate": 9.240155937161813e-05, "loss": 0.161, "step": 19450 }, { "epoch": 9.33, "learning_rate": 9.239394850333296e-05, "loss": 0.1666, "step": 19460 }, { "epoch": 9.33, "learning_rate": 9.238633413906756e-05, "loss": 0.1394, "step": 19470 }, { "epoch": 9.34, "learning_rate": 9.237871627944987e-05, "loss": 0.1507, "step": 19480 }, { "epoch": 9.34, "learning_rate": 9.237109492510807e-05, "loss": 0.1742, "step": 19490 }, { "epoch": 9.34, "learning_rate": 9.236347007667066e-05, "loss": 0.135, "step": 19500 }, { "epoch": 9.35, "learning_rate": 9.23558417347664e-05, "loss": 0.1412, "step": 19510 }, { "epoch": 9.35, "learning_rate": 9.234820990002437e-05, "loss": 0.1406, "step": 19520 }, { "epoch": 9.36, "learning_rate": 9.234057457307392e-05, "loss": 0.1408, "step": 19530 }, { "epoch": 9.36, "learning_rate": 9.233293575454467e-05, "loss": 0.1302, "step": 19540 }, { "epoch": 9.37, "learning_rate": 9.23252934450666e-05, "loss": 0.1392, "step": 19550 }, { "epoch": 9.37, "learning_rate": 9.231764764526987e-05, "loss": 0.1397, "step": 19560 }, { "epoch": 9.38, "learning_rate": 9.2309998355785e-05, "loss": 0.1307, "step": 19570 }, { "epoch": 9.38, "learning_rate": 9.23023455772428e-05, "loss": 0.146, "step": 19580 }, { "epoch": 9.39, "learning_rate": 9.229468931027435e-05, "loss": 0.1436, "step": 19590 }, { "epoch": 9.39, "learning_rate": 9.228702955551101e-05, "loss": 0.1443, "step": 19600 }, { "epoch": 9.39, "learning_rate": 9.227936631358441e-05, "loss": 0.149, "step": 19610 }, { "epoch": 9.4, "learning_rate": 9.227169958512652e-05, "loss": 0.1625, "step": 19620 }, { "epoch": 9.4, "learning_rate": 9.226402937076958e-05, "loss": 0.1451, "step": 19630 }, { "epoch": 9.41, "learning_rate": 9.225635567114608e-05, "loss": 0.152, "step": 19640 }, { "epoch": 9.41, "learning_rate": 9.224867848688883e-05, "loss": 0.1522, "step": 19650 }, { "epoch": 9.42, "learning_rate": 9.224099781863094e-05, "loss": 0.1578, "step": 19660 }, { "epoch": 9.42, "learning_rate": 9.223331366700578e-05, "loss": 0.1444, "step": 19670 }, { "epoch": 9.43, "learning_rate": 9.222562603264702e-05, "loss": 0.138, "step": 19680 }, { "epoch": 9.43, "learning_rate": 9.22179349161886e-05, "loss": 0.17, "step": 19690 }, { "epoch": 9.44, "learning_rate": 9.221024031826476e-05, "loss": 0.1332, "step": 19700 }, { "epoch": 9.44, "learning_rate": 9.220254223951006e-05, "loss": 0.1605, "step": 19710 }, { "epoch": 9.44, "learning_rate": 9.21948406805593e-05, "loss": 0.1426, "step": 19720 }, { "epoch": 9.45, "learning_rate": 9.218713564204755e-05, "loss": 0.1463, "step": 19730 }, { "epoch": 9.45, "learning_rate": 9.217942712461026e-05, "loss": 0.1471, "step": 19740 }, { "epoch": 9.46, "learning_rate": 9.217171512888304e-05, "loss": 0.1267, "step": 19750 }, { "epoch": 9.46, "learning_rate": 9.216399965550191e-05, "loss": 0.1512, "step": 19760 }, { "epoch": 9.47, "learning_rate": 9.215628070510307e-05, "loss": 0.1481, "step": 19770 }, { "epoch": 9.47, "learning_rate": 9.214855827832312e-05, "loss": 0.1477, "step": 19780 }, { "epoch": 9.48, "learning_rate": 9.214083237579882e-05, "loss": 0.1417, "step": 19790 }, { "epoch": 9.48, "learning_rate": 9.213310299816731e-05, "loss": 0.1633, "step": 19800 }, { "epoch": 9.49, "learning_rate": 9.212537014606597e-05, "loss": 0.164, "step": 19810 }, { "epoch": 9.49, "learning_rate": 9.211763382013253e-05, "loss": 0.1446, "step": 19820 }, { "epoch": 9.49, "learning_rate": 9.210989402100489e-05, "loss": 0.1739, "step": 19830 }, { "epoch": 9.5, "learning_rate": 9.210215074932134e-05, "loss": 0.1436, "step": 19840 }, { "epoch": 9.5, "learning_rate": 9.209440400572044e-05, "loss": 0.1443, "step": 19850 }, { "epoch": 9.51, "learning_rate": 9.2086653790841e-05, "loss": 0.1356, "step": 19860 }, { "epoch": 9.51, "learning_rate": 9.207890010532211e-05, "loss": 0.1583, "step": 19870 }, { "epoch": 9.52, "learning_rate": 9.207114294980323e-05, "loss": 0.1418, "step": 19880 }, { "epoch": 9.52, "learning_rate": 9.2063382324924e-05, "loss": 0.1685, "step": 19890 }, { "epoch": 9.53, "learning_rate": 9.205561823132441e-05, "loss": 0.1277, "step": 19900 }, { "epoch": 9.53, "learning_rate": 9.204785066964472e-05, "loss": 0.1431, "step": 19910 }, { "epoch": 9.54, "learning_rate": 9.204007964052545e-05, "loss": 0.1408, "step": 19920 }, { "epoch": 9.54, "learning_rate": 9.203230514460747e-05, "loss": 0.1584, "step": 19930 }, { "epoch": 9.55, "learning_rate": 9.202452718253187e-05, "loss": 0.1324, "step": 19940 }, { "epoch": 9.55, "learning_rate": 9.201674575494008e-05, "loss": 0.1611, "step": 19950 }, { "epoch": 9.55, "learning_rate": 9.200896086247375e-05, "loss": 0.1295, "step": 19960 }, { "epoch": 9.56, "learning_rate": 9.200117250577489e-05, "loss": 0.141, "step": 19970 }, { "epoch": 9.56, "learning_rate": 9.199338068548574e-05, "loss": 0.1559, "step": 19980 }, { "epoch": 9.57, "learning_rate": 9.198558540224885e-05, "loss": 0.1363, "step": 19990 }, { "epoch": 9.57, "learning_rate": 9.197778665670706e-05, "loss": 0.127, "step": 20000 }, { "epoch": 9.58, "learning_rate": 9.196998444950347e-05, "loss": 0.1407, "step": 20010 }, { "epoch": 9.58, "learning_rate": 9.19621787812815e-05, "loss": 0.1411, "step": 20020 }, { "epoch": 9.59, "learning_rate": 9.195436965268483e-05, "loss": 0.1418, "step": 20030 }, { "epoch": 9.59, "learning_rate": 9.194655706435741e-05, "loss": 0.1444, "step": 20040 }, { "epoch": 9.6, "learning_rate": 9.193874101694355e-05, "loss": 0.1241, "step": 20050 }, { "epoch": 9.6, "learning_rate": 9.193092151108776e-05, "loss": 0.1422, "step": 20060 }, { "epoch": 9.6, "learning_rate": 9.192309854743485e-05, "loss": 0.1383, "step": 20070 }, { "epoch": 9.61, "learning_rate": 9.191527212662998e-05, "loss": 0.1484, "step": 20080 }, { "epoch": 9.61, "learning_rate": 9.19074422493185e-05, "loss": 0.153, "step": 20090 }, { "epoch": 9.62, "learning_rate": 9.189960891614615e-05, "loss": 0.1507, "step": 20100 }, { "epoch": 9.62, "learning_rate": 9.189177212775886e-05, "loss": 0.1311, "step": 20110 }, { "epoch": 9.63, "learning_rate": 9.188393188480289e-05, "loss": 0.1734, "step": 20120 }, { "epoch": 9.63, "learning_rate": 9.187608818792476e-05, "loss": 0.1495, "step": 20130 }, { "epoch": 9.64, "learning_rate": 9.186824103777135e-05, "loss": 0.1291, "step": 20140 }, { "epoch": 9.64, "learning_rate": 9.186039043498972e-05, "loss": 0.1303, "step": 20150 }, { "epoch": 9.65, "learning_rate": 9.185253638022729e-05, "loss": 0.1426, "step": 20160 }, { "epoch": 9.65, "learning_rate": 9.184467887413171e-05, "loss": 0.1262, "step": 20170 }, { "epoch": 9.65, "learning_rate": 9.183681791735095e-05, "loss": 0.153, "step": 20180 }, { "epoch": 9.66, "learning_rate": 9.182895351053327e-05, "loss": 0.1328, "step": 20190 }, { "epoch": 9.66, "learning_rate": 9.18210856543272e-05, "loss": 0.1453, "step": 20200 }, { "epoch": 9.67, "learning_rate": 9.181321434938154e-05, "loss": 0.1413, "step": 20210 }, { "epoch": 9.67, "learning_rate": 9.180533959634544e-05, "loss": 0.1444, "step": 20220 }, { "epoch": 9.68, "learning_rate": 9.17974613958682e-05, "loss": 0.1393, "step": 20230 }, { "epoch": 9.68, "learning_rate": 9.178957974859956e-05, "loss": 0.1445, "step": 20240 }, { "epoch": 9.69, "learning_rate": 9.178169465518944e-05, "loss": 0.1355, "step": 20250 }, { "epoch": 9.69, "learning_rate": 9.177380611628809e-05, "loss": 0.1562, "step": 20260 }, { "epoch": 9.7, "learning_rate": 9.176591413254603e-05, "loss": 0.1513, "step": 20270 }, { "epoch": 9.7, "learning_rate": 9.175801870461407e-05, "loss": 0.1288, "step": 20280 }, { "epoch": 9.71, "learning_rate": 9.175011983314328e-05, "loss": 0.1312, "step": 20290 }, { "epoch": 9.71, "learning_rate": 9.174221751878505e-05, "loss": 0.1376, "step": 20300 }, { "epoch": 9.71, "learning_rate": 9.173431176219104e-05, "loss": 0.1566, "step": 20310 }, { "epoch": 9.72, "learning_rate": 9.172640256401318e-05, "loss": 0.1351, "step": 20320 }, { "epoch": 9.72, "learning_rate": 9.171848992490371e-05, "loss": 0.142, "step": 20330 }, { "epoch": 9.73, "learning_rate": 9.171057384551511e-05, "loss": 0.1612, "step": 20340 }, { "epoch": 9.73, "learning_rate": 9.17026543265002e-05, "loss": 0.1363, "step": 20350 }, { "epoch": 9.74, "learning_rate": 9.169473136851205e-05, "loss": 0.1407, "step": 20360 }, { "epoch": 9.74, "learning_rate": 9.168680497220404e-05, "loss": 0.1396, "step": 20370 }, { "epoch": 9.75, "learning_rate": 9.167887513822976e-05, "loss": 0.1443, "step": 20380 }, { "epoch": 9.75, "learning_rate": 9.167094186724318e-05, "loss": 0.1326, "step": 20390 }, { "epoch": 9.76, "learning_rate": 9.16630051598985e-05, "loss": 0.1531, "step": 20400 }, { "epoch": 9.76, "learning_rate": 9.165506501685022e-05, "loss": 0.1422, "step": 20410 }, { "epoch": 9.76, "learning_rate": 9.16471214387531e-05, "loss": 0.1663, "step": 20420 }, { "epoch": 9.77, "learning_rate": 9.163917442626219e-05, "loss": 0.1385, "step": 20430 }, { "epoch": 9.77, "learning_rate": 9.163122398003289e-05, "loss": 0.1367, "step": 20440 }, { "epoch": 9.78, "learning_rate": 9.162327010072079e-05, "loss": 0.1643, "step": 20450 }, { "epoch": 9.78, "learning_rate": 9.161531278898176e-05, "loss": 0.1445, "step": 20460 }, { "epoch": 9.79, "learning_rate": 9.160735204547206e-05, "loss": 0.1405, "step": 20470 }, { "epoch": 9.79, "learning_rate": 9.159938787084814e-05, "loss": 0.1561, "step": 20480 }, { "epoch": 9.8, "learning_rate": 9.159142026576675e-05, "loss": 0.1535, "step": 20490 }, { "epoch": 9.8, "learning_rate": 9.158344923088496e-05, "loss": 0.1598, "step": 20500 }, { "epoch": 9.81, "learning_rate": 9.157547476686005e-05, "loss": 0.1451, "step": 20510 }, { "epoch": 9.81, "learning_rate": 9.156749687434965e-05, "loss": 0.163, "step": 20520 }, { "epoch": 9.81, "learning_rate": 9.155951555401165e-05, "loss": 0.1474, "step": 20530 }, { "epoch": 9.82, "learning_rate": 9.155153080650424e-05, "loss": 0.1524, "step": 20540 }, { "epoch": 9.82, "learning_rate": 9.154354263248585e-05, "loss": 0.1445, "step": 20550 }, { "epoch": 9.83, "learning_rate": 9.153555103261522e-05, "loss": 0.1398, "step": 20560 }, { "epoch": 9.83, "learning_rate": 9.152755600755138e-05, "loss": 0.1849, "step": 20570 }, { "epoch": 9.84, "learning_rate": 9.151955755795363e-05, "loss": 0.1447, "step": 20580 }, { "epoch": 9.84, "learning_rate": 9.151155568448157e-05, "loss": 0.1448, "step": 20590 }, { "epoch": 9.85, "learning_rate": 9.150355038779504e-05, "loss": 0.151, "step": 20600 }, { "epoch": 9.85, "learning_rate": 9.149554166855418e-05, "loss": 0.1384, "step": 20610 }, { "epoch": 9.86, "learning_rate": 9.148752952741948e-05, "loss": 0.1382, "step": 20620 }, { "epoch": 9.86, "learning_rate": 9.14795139650516e-05, "loss": 0.1782, "step": 20630 }, { "epoch": 9.87, "learning_rate": 9.147149498211157e-05, "loss": 0.134, "step": 20640 }, { "epoch": 9.87, "learning_rate": 9.146347257926062e-05, "loss": 0.1594, "step": 20650 }, { "epoch": 9.87, "learning_rate": 9.145544675716038e-05, "loss": 0.156, "step": 20660 }, { "epoch": 9.88, "learning_rate": 9.144741751647263e-05, "loss": 0.1597, "step": 20670 }, { "epoch": 9.88, "learning_rate": 9.143938485785955e-05, "loss": 0.1489, "step": 20680 }, { "epoch": 9.89, "learning_rate": 9.143134878198349e-05, "loss": 0.1525, "step": 20690 }, { "epoch": 9.89, "learning_rate": 9.142330928950718e-05, "loss": 0.1326, "step": 20700 }, { "epoch": 9.9, "learning_rate": 9.141526638109357e-05, "loss": 0.1519, "step": 20710 }, { "epoch": 9.9, "learning_rate": 9.140722005740592e-05, "loss": 0.14, "step": 20720 }, { "epoch": 9.91, "learning_rate": 9.139917031910775e-05, "loss": 0.1295, "step": 20730 }, { "epoch": 9.91, "learning_rate": 9.139111716686289e-05, "loss": 0.1545, "step": 20740 }, { "epoch": 9.92, "learning_rate": 9.138306060133544e-05, "loss": 0.134, "step": 20750 }, { "epoch": 9.92, "learning_rate": 9.137500062318976e-05, "loss": 0.1459, "step": 20760 }, { "epoch": 9.92, "learning_rate": 9.13669372330905e-05, "loss": 0.1431, "step": 20770 }, { "epoch": 9.93, "learning_rate": 9.135887043170264e-05, "loss": 0.1437, "step": 20780 }, { "epoch": 9.93, "learning_rate": 9.135080021969137e-05, "loss": 0.1599, "step": 20790 }, { "epoch": 9.94, "learning_rate": 9.134272659772221e-05, "loss": 0.1813, "step": 20800 }, { "epoch": 9.94, "learning_rate": 9.133464956646091e-05, "loss": 0.1627, "step": 20810 }, { "epoch": 9.95, "learning_rate": 9.13265691265736e-05, "loss": 0.1469, "step": 20820 }, { "epoch": 9.95, "learning_rate": 9.131848527872657e-05, "loss": 0.1695, "step": 20830 }, { "epoch": 10.0, "learning_rate": 9.131039802358647e-05, "loss": 0.1804, "step": 20840 }, { "epoch": 10.01, "learning_rate": 9.13023073618202e-05, "loss": 0.1375, "step": 20850 }, { "epoch": 10.01, "learning_rate": 9.129421329409495e-05, "loss": 0.1429, "step": 20860 }, { "epoch": 10.02, "learning_rate": 9.12861158210782e-05, "loss": 0.14, "step": 20870 }, { "epoch": 10.02, "learning_rate": 9.127801494343771e-05, "loss": 0.1611, "step": 20880 }, { "epoch": 10.03, "learning_rate": 9.126991066184148e-05, "loss": 0.1477, "step": 20890 }, { "epoch": 10.03, "learning_rate": 9.126180297695783e-05, "loss": 0.1472, "step": 20900 }, { "epoch": 10.04, "learning_rate": 9.125369188945539e-05, "loss": 0.15, "step": 20910 }, { "epoch": 10.04, "learning_rate": 9.124557740000298e-05, "loss": 0.1247, "step": 20920 }, { "epoch": 10.05, "learning_rate": 9.12374595092698e-05, "loss": 0.1443, "step": 20930 }, { "epoch": 10.05, "learning_rate": 9.122933821792525e-05, "loss": 0.1319, "step": 20940 }, { "epoch": 10.05, "learning_rate": 9.122121352663908e-05, "loss": 0.1563, "step": 20950 }, { "epoch": 10.06, "learning_rate": 9.121308543608125e-05, "loss": 0.1407, "step": 20960 }, { "epoch": 10.06, "learning_rate": 9.120495394692206e-05, "loss": 0.1447, "step": 20970 }, { "epoch": 10.07, "learning_rate": 9.119681905983206e-05, "loss": 0.1443, "step": 20980 }, { "epoch": 10.07, "learning_rate": 9.118868077548208e-05, "loss": 0.1322, "step": 20990 }, { "epoch": 10.08, "learning_rate": 9.118053909454325e-05, "loss": 0.1292, "step": 21000 }, { "epoch": 10.08, "learning_rate": 9.117239401768695e-05, "loss": 0.131, "step": 21010 }, { "epoch": 10.09, "learning_rate": 9.116424554558487e-05, "loss": 0.131, "step": 21020 }, { "epoch": 10.09, "learning_rate": 9.115609367890896e-05, "loss": 0.1367, "step": 21030 }, { "epoch": 10.1, "learning_rate": 9.114793841833146e-05, "loss": 0.1517, "step": 21040 }, { "epoch": 10.1, "learning_rate": 9.113977976452488e-05, "loss": 0.1438, "step": 21050 }, { "epoch": 10.11, "learning_rate": 9.113161771816203e-05, "loss": 0.1346, "step": 21060 }, { "epoch": 10.11, "learning_rate": 9.112345227991598e-05, "loss": 0.1357, "step": 21070 }, { "epoch": 10.11, "learning_rate": 9.111528345046008e-05, "loss": 0.1301, "step": 21080 }, { "epoch": 10.12, "learning_rate": 9.110711123046796e-05, "loss": 0.1488, "step": 21090 }, { "epoch": 10.12, "learning_rate": 9.109893562061354e-05, "loss": 0.1704, "step": 21100 }, { "epoch": 10.13, "learning_rate": 9.109075662157103e-05, "loss": 0.1369, "step": 21110 }, { "epoch": 10.13, "learning_rate": 9.108257423401489e-05, "loss": 0.1354, "step": 21120 }, { "epoch": 10.14, "learning_rate": 9.107438845861987e-05, "loss": 0.1202, "step": 21130 }, { "epoch": 10.14, "learning_rate": 9.106619929606102e-05, "loss": 0.1491, "step": 21140 }, { "epoch": 10.15, "learning_rate": 9.105800674701364e-05, "loss": 0.1605, "step": 21150 }, { "epoch": 10.15, "learning_rate": 9.104981081215331e-05, "loss": 0.1618, "step": 21160 }, { "epoch": 10.16, "learning_rate": 9.104161149215593e-05, "loss": 0.1198, "step": 21170 }, { "epoch": 10.16, "learning_rate": 9.103340878769762e-05, "loss": 0.137, "step": 21180 }, { "epoch": 10.16, "learning_rate": 9.102520269945483e-05, "loss": 0.1239, "step": 21190 }, { "epoch": 10.17, "learning_rate": 9.101699322810424e-05, "loss": 0.1429, "step": 21200 }, { "epoch": 10.17, "learning_rate": 9.100878037432288e-05, "loss": 0.145, "step": 21210 }, { "epoch": 10.18, "learning_rate": 9.100056413878797e-05, "loss": 0.1354, "step": 21220 }, { "epoch": 10.18, "learning_rate": 9.09923445221771e-05, "loss": 0.1301, "step": 21230 }, { "epoch": 10.19, "learning_rate": 9.098412152516807e-05, "loss": 0.1324, "step": 21240 }, { "epoch": 10.19, "learning_rate": 9.097589514843896e-05, "loss": 0.1267, "step": 21250 }, { "epoch": 10.2, "learning_rate": 9.09676653926682e-05, "loss": 0.1385, "step": 21260 }, { "epoch": 10.2, "learning_rate": 9.095943225853441e-05, "loss": 0.1445, "step": 21270 }, { "epoch": 10.21, "learning_rate": 9.095119574671653e-05, "loss": 0.1256, "step": 21280 }, { "epoch": 10.21, "learning_rate": 9.094295585789381e-05, "loss": 0.1454, "step": 21290 }, { "epoch": 10.21, "learning_rate": 9.093471259274572e-05, "loss": 0.12, "step": 21300 }, { "epoch": 10.22, "learning_rate": 9.092646595195205e-05, "loss": 0.1362, "step": 21310 }, { "epoch": 10.22, "learning_rate": 9.091821593619281e-05, "loss": 0.1396, "step": 21320 }, { "epoch": 10.23, "learning_rate": 9.090996254614838e-05, "loss": 0.1299, "step": 21330 }, { "epoch": 10.23, "learning_rate": 9.090170578249934e-05, "loss": 0.1323, "step": 21340 }, { "epoch": 10.24, "learning_rate": 9.089344564592659e-05, "loss": 0.1484, "step": 21350 }, { "epoch": 10.24, "learning_rate": 9.08851821371113e-05, "loss": 0.1271, "step": 21360 }, { "epoch": 10.25, "learning_rate": 9.087691525673489e-05, "loss": 0.1378, "step": 21370 }, { "epoch": 10.25, "learning_rate": 9.08686450054791e-05, "loss": 0.1725, "step": 21380 }, { "epoch": 10.26, "learning_rate": 9.086037138402592e-05, "loss": 0.1438, "step": 21390 }, { "epoch": 10.26, "learning_rate": 9.085209439305764e-05, "loss": 0.1334, "step": 21400 }, { "epoch": 10.27, "learning_rate": 9.084381403325681e-05, "loss": 0.1486, "step": 21410 }, { "epoch": 10.27, "learning_rate": 9.083553030530625e-05, "loss": 0.138, "step": 21420 }, { "epoch": 10.27, "learning_rate": 9.082724320988908e-05, "loss": 0.141, "step": 21430 }, { "epoch": 10.28, "learning_rate": 9.081895274768868e-05, "loss": 0.1307, "step": 21440 }, { "epoch": 10.28, "learning_rate": 9.081065891938873e-05, "loss": 0.1626, "step": 21450 }, { "epoch": 10.29, "learning_rate": 9.080236172567318e-05, "loss": 0.1412, "step": 21460 }, { "epoch": 10.29, "learning_rate": 9.079406116722622e-05, "loss": 0.1543, "step": 21470 }, { "epoch": 10.3, "learning_rate": 9.078575724473238e-05, "loss": 0.1581, "step": 21480 }, { "epoch": 10.3, "learning_rate": 9.077744995887641e-05, "loss": 0.1296, "step": 21490 }, { "epoch": 10.31, "learning_rate": 9.076913931034337e-05, "loss": 0.136, "step": 21500 }, { "epoch": 10.31, "learning_rate": 9.07608252998186e-05, "loss": 0.1578, "step": 21510 }, { "epoch": 10.32, "learning_rate": 9.07525079279877e-05, "loss": 0.1163, "step": 21520 }, { "epoch": 10.32, "learning_rate": 9.074418719553656e-05, "loss": 0.1353, "step": 21530 }, { "epoch": 10.32, "learning_rate": 9.073586310315136e-05, "loss": 0.1242, "step": 21540 }, { "epoch": 10.33, "learning_rate": 9.07275356515185e-05, "loss": 0.1397, "step": 21550 }, { "epoch": 10.33, "learning_rate": 9.071920484132471e-05, "loss": 0.1335, "step": 21560 }, { "epoch": 10.34, "learning_rate": 9.071087067325699e-05, "loss": 0.1535, "step": 21570 }, { "epoch": 10.34, "learning_rate": 9.070253314800261e-05, "loss": 0.1297, "step": 21580 }, { "epoch": 10.35, "learning_rate": 9.069419226624913e-05, "loss": 0.1392, "step": 21590 }, { "epoch": 10.35, "learning_rate": 9.068584802868433e-05, "loss": 0.1341, "step": 21600 }, { "epoch": 10.36, "learning_rate": 9.067750043599637e-05, "loss": 0.1293, "step": 21610 }, { "epoch": 10.36, "learning_rate": 9.066914948887357e-05, "loss": 0.1414, "step": 21620 }, { "epoch": 10.37, "learning_rate": 9.066079518800462e-05, "loss": 0.1287, "step": 21630 }, { "epoch": 10.37, "learning_rate": 9.065243753407844e-05, "loss": 0.1511, "step": 21640 }, { "epoch": 10.37, "learning_rate": 9.064407652778423e-05, "loss": 0.1267, "step": 21650 }, { "epoch": 10.38, "learning_rate": 9.063571216981149e-05, "loss": 0.1337, "step": 21660 }, { "epoch": 10.38, "learning_rate": 9.062734446084995e-05, "loss": 0.145, "step": 21670 }, { "epoch": 10.39, "learning_rate": 9.061897340158967e-05, "loss": 0.1397, "step": 21680 }, { "epoch": 10.39, "learning_rate": 9.061059899272096e-05, "loss": 0.1156, "step": 21690 }, { "epoch": 10.4, "learning_rate": 9.060222123493441e-05, "loss": 0.1304, "step": 21700 }, { "epoch": 10.4, "learning_rate": 9.059384012892088e-05, "loss": 0.1457, "step": 21710 }, { "epoch": 10.41, "learning_rate": 9.058545567537151e-05, "loss": 0.1267, "step": 21720 }, { "epoch": 10.41, "learning_rate": 9.05770678749777e-05, "loss": 0.1358, "step": 21730 }, { "epoch": 10.42, "learning_rate": 9.056867672843118e-05, "loss": 0.1348, "step": 21740 }, { "epoch": 10.42, "learning_rate": 9.056028223642389e-05, "loss": 0.1329, "step": 21750 }, { "epoch": 10.43, "learning_rate": 9.055188439964807e-05, "loss": 0.1357, "step": 21760 }, { "epoch": 10.43, "learning_rate": 9.054348321879628e-05, "loss": 0.1336, "step": 21770 }, { "epoch": 10.43, "learning_rate": 9.053507869456127e-05, "loss": 0.1166, "step": 21780 }, { "epoch": 10.44, "learning_rate": 9.052667082763612e-05, "loss": 0.1369, "step": 21790 }, { "epoch": 10.44, "learning_rate": 9.051825961871423e-05, "loss": 0.1513, "step": 21800 }, { "epoch": 10.45, "learning_rate": 9.050984506848915e-05, "loss": 0.1779, "step": 21810 }, { "epoch": 10.45, "learning_rate": 9.050142717765481e-05, "loss": 0.1441, "step": 21820 }, { "epoch": 10.46, "learning_rate": 9.049300594690539e-05, "loss": 0.1401, "step": 21830 }, { "epoch": 10.46, "learning_rate": 9.048458137693535e-05, "loss": 0.1457, "step": 21840 }, { "epoch": 10.47, "learning_rate": 9.047615346843938e-05, "loss": 0.139, "step": 21850 }, { "epoch": 10.47, "learning_rate": 9.046772222211249e-05, "loss": 0.1275, "step": 21860 }, { "epoch": 10.48, "learning_rate": 9.045928763864997e-05, "loss": 0.1541, "step": 21870 }, { "epoch": 10.48, "learning_rate": 9.045084971874738e-05, "loss": 0.1262, "step": 21880 }, { "epoch": 10.48, "learning_rate": 9.044240846310053e-05, "loss": 0.1295, "step": 21890 }, { "epoch": 10.49, "learning_rate": 9.043396387240551e-05, "loss": 0.1315, "step": 21900 }, { "epoch": 10.49, "learning_rate": 9.042551594735869e-05, "loss": 0.1473, "step": 21910 }, { "epoch": 10.5, "learning_rate": 9.041706468865676e-05, "loss": 0.1421, "step": 21920 }, { "epoch": 10.5, "learning_rate": 9.040861009699661e-05, "loss": 0.1447, "step": 21930 }, { "epoch": 10.51, "learning_rate": 9.040015217307547e-05, "loss": 0.128, "step": 21940 }, { "epoch": 10.51, "learning_rate": 9.03916909175908e-05, "loss": 0.1376, "step": 21950 }, { "epoch": 10.52, "learning_rate": 9.038322633124035e-05, "loss": 0.1225, "step": 21960 }, { "epoch": 10.52, "learning_rate": 9.037475841472212e-05, "loss": 0.147, "step": 21970 }, { "epoch": 10.53, "learning_rate": 9.036628716873445e-05, "loss": 0.1434, "step": 21980 }, { "epoch": 10.53, "learning_rate": 9.035781259397592e-05, "loss": 0.1315, "step": 21990 }, { "epoch": 10.53, "learning_rate": 9.034933469114532e-05, "loss": 0.1605, "step": 22000 }, { "epoch": 10.54, "learning_rate": 9.034085346094183e-05, "loss": 0.1433, "step": 22010 }, { "epoch": 10.54, "learning_rate": 9.033236890406482e-05, "loss": 0.1353, "step": 22020 }, { "epoch": 10.55, "learning_rate": 9.032388102121396e-05, "loss": 0.1271, "step": 22030 }, { "epoch": 10.55, "learning_rate": 9.031538981308923e-05, "loss": 0.1154, "step": 22040 }, { "epoch": 10.56, "learning_rate": 9.03068952803908e-05, "loss": 0.1545, "step": 22050 }, { "epoch": 10.56, "learning_rate": 9.029839742381919e-05, "loss": 0.1275, "step": 22060 }, { "epoch": 10.57, "learning_rate": 9.028989624407517e-05, "loss": 0.1354, "step": 22070 }, { "epoch": 10.57, "learning_rate": 9.028139174185979e-05, "loss": 0.137, "step": 22080 }, { "epoch": 10.58, "learning_rate": 9.027288391787434e-05, "loss": 0.148, "step": 22090 }, { "epoch": 10.58, "learning_rate": 9.026437277282044e-05, "loss": 0.143, "step": 22100 }, { "epoch": 10.59, "learning_rate": 9.025585830739993e-05, "loss": 0.1371, "step": 22110 }, { "epoch": 10.59, "learning_rate": 9.024734052231495e-05, "loss": 0.1472, "step": 22120 }, { "epoch": 10.59, "learning_rate": 9.023881941826795e-05, "loss": 0.1334, "step": 22130 }, { "epoch": 10.6, "learning_rate": 9.023029499596156e-05, "loss": 0.1433, "step": 22140 }, { "epoch": 10.6, "learning_rate": 9.022176725609877e-05, "loss": 0.1281, "step": 22150 }, { "epoch": 10.61, "learning_rate": 9.02132361993828e-05, "loss": 0.1297, "step": 22160 }, { "epoch": 10.61, "learning_rate": 9.020470182651718e-05, "loss": 0.1435, "step": 22170 }, { "epoch": 10.62, "learning_rate": 9.019616413820566e-05, "loss": 0.1318, "step": 22180 }, { "epoch": 10.62, "learning_rate": 9.018762313515232e-05, "loss": 0.1407, "step": 22190 }, { "epoch": 10.63, "learning_rate": 9.017907881806146e-05, "loss": 0.1433, "step": 22200 }, { "epoch": 10.63, "learning_rate": 9.01705311876377e-05, "loss": 0.1362, "step": 22210 }, { "epoch": 10.64, "learning_rate": 9.016198024458588e-05, "loss": 0.1472, "step": 22220 }, { "epoch": 10.64, "learning_rate": 9.015342598961121e-05, "loss": 0.1404, "step": 22230 }, { "epoch": 10.64, "learning_rate": 9.014486842341906e-05, "loss": 0.1443, "step": 22240 }, { "epoch": 10.65, "learning_rate": 9.013630754671511e-05, "loss": 0.1232, "step": 22250 }, { "epoch": 10.65, "learning_rate": 9.012774336020537e-05, "loss": 0.1377, "step": 22260 }, { "epoch": 10.66, "learning_rate": 9.011917586459603e-05, "loss": 0.1447, "step": 22270 }, { "epoch": 10.66, "learning_rate": 9.011060506059366e-05, "loss": 0.1545, "step": 22280 }, { "epoch": 10.67, "learning_rate": 9.010203094890499e-05, "loss": 0.1304, "step": 22290 }, { "epoch": 10.67, "learning_rate": 9.009345353023711e-05, "loss": 0.1263, "step": 22300 }, { "epoch": 10.68, "learning_rate": 9.008487280529734e-05, "loss": 0.1425, "step": 22310 }, { "epoch": 10.68, "learning_rate": 9.007628877479326e-05, "loss": 0.1462, "step": 22320 }, { "epoch": 10.69, "learning_rate": 9.006770143943279e-05, "loss": 0.1431, "step": 22330 }, { "epoch": 10.69, "learning_rate": 9.005911079992402e-05, "loss": 0.1344, "step": 22340 }, { "epoch": 10.69, "learning_rate": 9.005051685697543e-05, "loss": 0.1158, "step": 22350 }, { "epoch": 10.7, "learning_rate": 9.004191961129567e-05, "loss": 0.1129, "step": 22360 }, { "epoch": 10.7, "learning_rate": 9.003331906359371e-05, "loss": 0.143, "step": 22370 }, { "epoch": 10.71, "learning_rate": 9.002471521457881e-05, "loss": 0.1299, "step": 22380 }, { "epoch": 10.71, "learning_rate": 9.001610806496045e-05, "loss": 0.134, "step": 22390 }, { "epoch": 10.72, "learning_rate": 9.000749761544842e-05, "loss": 0.1515, "step": 22400 }, { "epoch": 10.72, "learning_rate": 8.999888386675279e-05, "loss": 0.1318, "step": 22410 }, { "epoch": 10.73, "learning_rate": 8.999026681958387e-05, "loss": 0.1452, "step": 22420 }, { "epoch": 10.73, "learning_rate": 8.998164647465226e-05, "loss": 0.1363, "step": 22430 }, { "epoch": 10.74, "learning_rate": 8.997302283266882e-05, "loss": 0.1309, "step": 22440 }, { "epoch": 10.74, "learning_rate": 8.996439589434471e-05, "loss": 0.1447, "step": 22450 }, { "epoch": 10.74, "learning_rate": 8.995576566039133e-05, "loss": 0.1531, "step": 22460 }, { "epoch": 10.75, "learning_rate": 8.994713213152038e-05, "loss": 0.1457, "step": 22470 }, { "epoch": 10.75, "learning_rate": 8.993849530844379e-05, "loss": 0.1538, "step": 22480 }, { "epoch": 10.76, "learning_rate": 8.99298551918738e-05, "loss": 0.1339, "step": 22490 }, { "epoch": 10.76, "learning_rate": 8.992121178252292e-05, "loss": 0.1464, "step": 22500 }, { "epoch": 10.77, "learning_rate": 8.991256508110391e-05, "loss": 0.1351, "step": 22510 }, { "epoch": 10.77, "learning_rate": 8.990391508832983e-05, "loss": 0.1392, "step": 22520 }, { "epoch": 10.78, "learning_rate": 8.989526180491396e-05, "loss": 0.1359, "step": 22530 }, { "epoch": 10.78, "learning_rate": 8.988660523156993e-05, "loss": 0.1562, "step": 22540 }, { "epoch": 10.79, "learning_rate": 8.987794536901156e-05, "loss": 0.1502, "step": 22550 }, { "epoch": 10.79, "learning_rate": 8.986928221795299e-05, "loss": 0.1539, "step": 22560 }, { "epoch": 10.8, "learning_rate": 8.986061577910861e-05, "loss": 0.145, "step": 22570 }, { "epoch": 10.8, "learning_rate": 8.985194605319312e-05, "loss": 0.137, "step": 22580 }, { "epoch": 10.8, "learning_rate": 8.984327304092143e-05, "loss": 0.147, "step": 22590 }, { "epoch": 10.81, "learning_rate": 8.983459674300876e-05, "loss": 0.1401, "step": 22600 }, { "epoch": 10.81, "learning_rate": 8.982591716017062e-05, "loss": 0.1243, "step": 22610 }, { "epoch": 10.82, "learning_rate": 8.981723429312271e-05, "loss": 0.1378, "step": 22620 }, { "epoch": 10.82, "learning_rate": 8.980854814258111e-05, "loss": 0.1437, "step": 22630 }, { "epoch": 10.83, "learning_rate": 8.979985870926209e-05, "loss": 0.1202, "step": 22640 }, { "epoch": 10.83, "learning_rate": 8.979116599388222e-05, "loss": 0.136, "step": 22650 }, { "epoch": 10.84, "learning_rate": 8.978246999715834e-05, "loss": 0.1478, "step": 22660 }, { "epoch": 10.84, "learning_rate": 8.977377071980755e-05, "loss": 0.1464, "step": 22670 }, { "epoch": 10.85, "learning_rate": 8.976506816254721e-05, "loss": 0.1265, "step": 22680 }, { "epoch": 10.85, "learning_rate": 8.975636232609503e-05, "loss": 0.149, "step": 22690 }, { "epoch": 10.85, "learning_rate": 8.974765321116886e-05, "loss": 0.1443, "step": 22700 }, { "epoch": 10.86, "learning_rate": 8.973894081848695e-05, "loss": 0.1403, "step": 22710 }, { "epoch": 10.86, "learning_rate": 8.973022514876772e-05, "loss": 0.1436, "step": 22720 }, { "epoch": 10.87, "learning_rate": 8.972150620272991e-05, "loss": 0.1317, "step": 22730 }, { "epoch": 10.87, "learning_rate": 8.971278398109251e-05, "loss": 0.1368, "step": 22740 }, { "epoch": 10.88, "learning_rate": 8.970405848457482e-05, "loss": 0.1224, "step": 22750 }, { "epoch": 10.88, "learning_rate": 8.969532971389636e-05, "loss": 0.1288, "step": 22760 }, { "epoch": 10.89, "learning_rate": 8.968659766977693e-05, "loss": 0.1428, "step": 22770 }, { "epoch": 10.89, "learning_rate": 8.967786235293662e-05, "loss": 0.1534, "step": 22780 }, { "epoch": 10.9, "learning_rate": 8.96691237640958e-05, "loss": 0.1385, "step": 22790 }, { "epoch": 10.9, "learning_rate": 8.966038190397508e-05, "loss": 0.1295, "step": 22800 }, { "epoch": 10.9, "learning_rate": 8.96516367732953e-05, "loss": 0.142, "step": 22810 }, { "epoch": 10.91, "learning_rate": 8.96428883727777e-05, "loss": 0.1359, "step": 22820 }, { "epoch": 10.91, "learning_rate": 8.963413670314368e-05, "loss": 0.1349, "step": 22830 }, { "epoch": 10.92, "learning_rate": 8.962538176511492e-05, "loss": 0.1344, "step": 22840 }, { "epoch": 10.92, "learning_rate": 8.961662355941339e-05, "loss": 0.1416, "step": 22850 }, { "epoch": 10.93, "learning_rate": 8.960786208676135e-05, "loss": 0.1395, "step": 22860 }, { "epoch": 10.93, "learning_rate": 8.959909734788128e-05, "loss": 0.1357, "step": 22870 }, { "epoch": 10.94, "learning_rate": 8.959032934349598e-05, "loss": 0.148, "step": 22880 }, { "epoch": 10.94, "learning_rate": 8.958155807432848e-05, "loss": 0.1657, "step": 22890 }, { "epoch": 10.95, "learning_rate": 8.957278354110212e-05, "loss": 0.1706, "step": 22900 }, { "epoch": 10.95, "learning_rate": 8.956400574454046e-05, "loss": 0.1737, "step": 22910 }, { "epoch": 11.0, "learning_rate": 8.955522468536736e-05, "loss": 0.1482, "step": 22920 }, { "epoch": 11.01, "learning_rate": 8.954644036430694e-05, "loss": 0.1271, "step": 22930 }, { "epoch": 11.01, "learning_rate": 8.953765278208359e-05, "loss": 0.1632, "step": 22940 }, { "epoch": 11.02, "learning_rate": 8.952886193942198e-05, "loss": 0.138, "step": 22950 }, { "epoch": 11.02, "learning_rate": 8.952006783704703e-05, "loss": 0.1409, "step": 22960 }, { "epoch": 11.03, "learning_rate": 8.951127047568395e-05, "loss": 0.1221, "step": 22970 }, { "epoch": 11.03, "learning_rate": 8.950246985605818e-05, "loss": 0.1283, "step": 22980 }, { "epoch": 11.04, "learning_rate": 8.949366597889548e-05, "loss": 0.1413, "step": 22990 }, { "epoch": 11.04, "learning_rate": 8.948485884492186e-05, "loss": 0.1449, "step": 23000 }, { "epoch": 11.04, "learning_rate": 8.947604845486357e-05, "loss": 0.1449, "step": 23010 }, { "epoch": 11.05, "learning_rate": 8.946723480944717e-05, "loss": 0.1517, "step": 23020 }, { "epoch": 11.05, "learning_rate": 8.945841790939947e-05, "loss": 0.1266, "step": 23030 }, { "epoch": 11.06, "learning_rate": 8.94495977554475e-05, "loss": 0.1388, "step": 23040 }, { "epoch": 11.06, "learning_rate": 8.944077434831867e-05, "loss": 0.1313, "step": 23050 }, { "epoch": 11.07, "learning_rate": 8.943194768874058e-05, "loss": 0.1084, "step": 23060 }, { "epoch": 11.07, "learning_rate": 8.942311777744111e-05, "loss": 0.1262, "step": 23070 }, { "epoch": 11.08, "learning_rate": 8.941428461514837e-05, "loss": 0.1306, "step": 23080 }, { "epoch": 11.08, "learning_rate": 8.940544820259082e-05, "loss": 0.116, "step": 23090 }, { "epoch": 11.09, "learning_rate": 8.939660854049716e-05, "loss": 0.124, "step": 23100 }, { "epoch": 11.09, "learning_rate": 8.938776562959632e-05, "loss": 0.1151, "step": 23110 }, { "epoch": 11.09, "learning_rate": 8.937891947061753e-05, "loss": 0.1267, "step": 23120 }, { "epoch": 11.1, "learning_rate": 8.937007006429027e-05, "loss": 0.1359, "step": 23130 }, { "epoch": 11.1, "learning_rate": 8.936121741134431e-05, "loss": 0.1278, "step": 23140 }, { "epoch": 11.11, "learning_rate": 8.935236151250967e-05, "loss": 0.1614, "step": 23150 }, { "epoch": 11.11, "learning_rate": 8.934350236851666e-05, "loss": 0.1479, "step": 23160 }, { "epoch": 11.12, "learning_rate": 8.933463998009582e-05, "loss": 0.1447, "step": 23170 }, { "epoch": 11.12, "learning_rate": 8.9325774347978e-05, "loss": 0.1239, "step": 23180 }, { "epoch": 11.13, "learning_rate": 8.931690547289429e-05, "loss": 0.1394, "step": 23190 }, { "epoch": 11.13, "learning_rate": 8.930803335557602e-05, "loss": 0.1269, "step": 23200 }, { "epoch": 11.14, "learning_rate": 8.929915799675488e-05, "loss": 0.1241, "step": 23210 }, { "epoch": 11.14, "learning_rate": 8.929027939716272e-05, "loss": 0.1597, "step": 23220 }, { "epoch": 11.14, "learning_rate": 8.928139755753174e-05, "loss": 0.1304, "step": 23230 }, { "epoch": 11.15, "learning_rate": 8.927251247859436e-05, "loss": 0.1287, "step": 23240 }, { "epoch": 11.15, "learning_rate": 8.926362416108327e-05, "loss": 0.1109, "step": 23250 }, { "epoch": 11.16, "learning_rate": 8.925473260573145e-05, "loss": 0.1311, "step": 23260 }, { "epoch": 11.16, "learning_rate": 8.924583781327213e-05, "loss": 0.1345, "step": 23270 }, { "epoch": 11.17, "learning_rate": 8.923693978443884e-05, "loss": 0.1442, "step": 23280 }, { "epoch": 11.17, "learning_rate": 8.922803851996529e-05, "loss": 0.1276, "step": 23290 }, { "epoch": 11.18, "learning_rate": 8.921913402058555e-05, "loss": 0.1327, "step": 23300 }, { "epoch": 11.18, "learning_rate": 8.921022628703392e-05, "loss": 0.123, "step": 23310 }, { "epoch": 11.19, "learning_rate": 8.920131532004497e-05, "loss": 0.1255, "step": 23320 }, { "epoch": 11.19, "learning_rate": 8.919240112035352e-05, "loss": 0.1518, "step": 23330 }, { "epoch": 11.2, "learning_rate": 8.918348368869469e-05, "loss": 0.1456, "step": 23340 }, { "epoch": 11.2, "learning_rate": 8.917456302580385e-05, "loss": 0.128, "step": 23350 }, { "epoch": 11.2, "learning_rate": 8.91656391324166e-05, "loss": 0.1322, "step": 23360 }, { "epoch": 11.21, "learning_rate": 8.91567120092689e-05, "loss": 0.1301, "step": 23370 }, { "epoch": 11.21, "learning_rate": 8.914778165709687e-05, "loss": 0.1229, "step": 23380 }, { "epoch": 11.22, "learning_rate": 8.913884807663697e-05, "loss": 0.1421, "step": 23390 }, { "epoch": 11.22, "learning_rate": 8.912991126862586e-05, "loss": 0.139, "step": 23400 }, { "epoch": 11.23, "learning_rate": 8.912097123380057e-05, "loss": 0.1466, "step": 23410 }, { "epoch": 11.23, "learning_rate": 8.911202797289829e-05, "loss": 0.1301, "step": 23420 }, { "epoch": 11.24, "learning_rate": 8.91030814866565e-05, "loss": 0.137, "step": 23430 }, { "epoch": 11.24, "learning_rate": 8.909413177581301e-05, "loss": 0.1263, "step": 23440 }, { "epoch": 11.25, "learning_rate": 8.908517884110582e-05, "loss": 0.1287, "step": 23450 }, { "epoch": 11.25, "learning_rate": 8.907622268327324e-05, "loss": 0.131, "step": 23460 }, { "epoch": 11.25, "learning_rate": 8.906726330305383e-05, "loss": 0.1218, "step": 23470 }, { "epoch": 11.26, "learning_rate": 8.905830070118642e-05, "loss": 0.1513, "step": 23480 }, { "epoch": 11.26, "learning_rate": 8.904933487841007e-05, "loss": 0.1168, "step": 23490 }, { "epoch": 11.27, "learning_rate": 8.90403658354642e-05, "loss": 0.1365, "step": 23500 }, { "epoch": 11.27, "learning_rate": 8.903139357308838e-05, "loss": 0.1453, "step": 23510 }, { "epoch": 11.28, "learning_rate": 8.902241809202254e-05, "loss": 0.125, "step": 23520 }, { "epoch": 11.28, "learning_rate": 8.90134393930068e-05, "loss": 0.1301, "step": 23530 }, { "epoch": 11.29, "learning_rate": 8.900445747678161e-05, "loss": 0.1207, "step": 23540 }, { "epoch": 11.29, "learning_rate": 8.899547234408763e-05, "loss": 0.1195, "step": 23550 }, { "epoch": 11.3, "learning_rate": 8.898648399566583e-05, "loss": 0.12, "step": 23560 }, { "epoch": 11.3, "learning_rate": 8.897749243225742e-05, "loss": 0.1366, "step": 23570 }, { "epoch": 11.3, "learning_rate": 8.896849765460389e-05, "loss": 0.1369, "step": 23580 }, { "epoch": 11.31, "learning_rate": 8.895949966344697e-05, "loss": 0.1404, "step": 23590 }, { "epoch": 11.31, "learning_rate": 8.895049845952868e-05, "loss": 0.1413, "step": 23600 }, { "epoch": 11.32, "learning_rate": 8.894149404359129e-05, "loss": 0.1283, "step": 23610 }, { "epoch": 11.32, "learning_rate": 8.893248641637736e-05, "loss": 0.1445, "step": 23620 }, { "epoch": 11.33, "learning_rate": 8.892347557862968e-05, "loss": 0.117, "step": 23630 }, { "epoch": 11.33, "learning_rate": 8.891446153109132e-05, "loss": 0.125, "step": 23640 }, { "epoch": 11.34, "learning_rate": 8.890544427450563e-05, "loss": 0.1284, "step": 23650 }, { "epoch": 11.34, "learning_rate": 8.88964238096162e-05, "loss": 0.1206, "step": 23660 }, { "epoch": 11.35, "learning_rate": 8.888740013716688e-05, "loss": 0.129, "step": 23670 }, { "epoch": 11.35, "learning_rate": 8.887837325790183e-05, "loss": 0.1322, "step": 23680 }, { "epoch": 11.36, "learning_rate": 8.886934317256542e-05, "loss": 0.1396, "step": 23690 }, { "epoch": 11.36, "learning_rate": 8.886030988190232e-05, "loss": 0.1322, "step": 23700 }, { "epoch": 11.36, "learning_rate": 8.885127338665746e-05, "loss": 0.1348, "step": 23710 }, { "epoch": 11.37, "learning_rate": 8.8842233687576e-05, "loss": 0.122, "step": 23720 }, { "epoch": 11.37, "learning_rate": 8.883319078540342e-05, "loss": 0.1256, "step": 23730 }, { "epoch": 11.38, "learning_rate": 8.882414468088541e-05, "loss": 0.1241, "step": 23740 }, { "epoch": 11.38, "learning_rate": 8.881509537476797e-05, "loss": 0.1404, "step": 23750 }, { "epoch": 11.39, "learning_rate": 8.880604286779733e-05, "loss": 0.1291, "step": 23760 }, { "epoch": 11.39, "learning_rate": 8.879698716072e-05, "loss": 0.1211, "step": 23770 }, { "epoch": 11.4, "learning_rate": 8.878792825428277e-05, "loss": 0.1389, "step": 23780 }, { "epoch": 11.4, "learning_rate": 8.877886614923264e-05, "loss": 0.1181, "step": 23790 }, { "epoch": 11.41, "learning_rate": 8.876980084631693e-05, "loss": 0.1155, "step": 23800 }, { "epoch": 11.41, "learning_rate": 8.87607323462832e-05, "loss": 0.1458, "step": 23810 }, { "epoch": 11.41, "learning_rate": 8.875166064987928e-05, "loss": 0.1255, "step": 23820 }, { "epoch": 11.42, "learning_rate": 8.874258575785323e-05, "loss": 0.1258, "step": 23830 }, { "epoch": 11.42, "learning_rate": 8.873350767095346e-05, "loss": 0.1282, "step": 23840 }, { "epoch": 11.43, "learning_rate": 8.872442638992853e-05, "loss": 0.1654, "step": 23850 }, { "epoch": 11.43, "learning_rate": 8.871534191552735e-05, "loss": 0.1402, "step": 23860 }, { "epoch": 11.44, "learning_rate": 8.870625424849906e-05, "loss": 0.1341, "step": 23870 }, { "epoch": 11.44, "learning_rate": 8.869716338959307e-05, "loss": 0.129, "step": 23880 }, { "epoch": 11.45, "learning_rate": 8.868806933955903e-05, "loss": 0.1219, "step": 23890 }, { "epoch": 11.45, "learning_rate": 8.86789720991469e-05, "loss": 0.1419, "step": 23900 }, { "epoch": 11.46, "learning_rate": 8.866987166910685e-05, "loss": 0.1396, "step": 23910 }, { "epoch": 11.46, "learning_rate": 8.866076805018937e-05, "loss": 0.1137, "step": 23920 }, { "epoch": 11.46, "learning_rate": 8.865166124314515e-05, "loss": 0.148, "step": 23930 }, { "epoch": 11.47, "learning_rate": 8.864255124872518e-05, "loss": 0.1324, "step": 23940 }, { "epoch": 11.47, "learning_rate": 8.863343806768071e-05, "loss": 0.1354, "step": 23950 }, { "epoch": 11.48, "learning_rate": 8.862432170076329e-05, "loss": 0.1246, "step": 23960 }, { "epoch": 11.48, "learning_rate": 8.861520214872464e-05, "loss": 0.1194, "step": 23970 }, { "epoch": 11.49, "learning_rate": 8.860607941231682e-05, "loss": 0.1212, "step": 23980 }, { "epoch": 11.49, "learning_rate": 8.859695349229212e-05, "loss": 0.1335, "step": 23990 }, { "epoch": 11.5, "learning_rate": 8.858782438940312e-05, "loss": 0.1299, "step": 24000 }, { "epoch": 11.5, "learning_rate": 8.857869210440262e-05, "loss": 0.1224, "step": 24010 }, { "epoch": 11.51, "learning_rate": 8.856955663804371e-05, "loss": 0.1307, "step": 24020 }, { "epoch": 11.51, "learning_rate": 8.856041799107974e-05, "loss": 0.1502, "step": 24030 }, { "epoch": 11.52, "learning_rate": 8.855127616426434e-05, "loss": 0.1316, "step": 24040 }, { "epoch": 11.52, "learning_rate": 8.854213115835134e-05, "loss": 0.132, "step": 24050 }, { "epoch": 11.52, "learning_rate": 8.853298297409494e-05, "loss": 0.1294, "step": 24060 }, { "epoch": 11.53, "learning_rate": 8.852383161224946e-05, "loss": 0.1511, "step": 24070 }, { "epoch": 11.53, "learning_rate": 8.851467707356963e-05, "loss": 0.1446, "step": 24080 }, { "epoch": 11.54, "learning_rate": 8.850551935881033e-05, "loss": 0.1245, "step": 24090 }, { "epoch": 11.54, "learning_rate": 8.849635846872675e-05, "loss": 0.1285, "step": 24100 }, { "epoch": 11.55, "learning_rate": 8.848719440407434e-05, "loss": 0.1231, "step": 24110 }, { "epoch": 11.55, "learning_rate": 8.84780271656088e-05, "loss": 0.1253, "step": 24120 }, { "epoch": 11.56, "learning_rate": 8.84688567540861e-05, "loss": 0.1577, "step": 24130 }, { "epoch": 11.56, "learning_rate": 8.845968317026247e-05, "loss": 0.1244, "step": 24140 }, { "epoch": 11.57, "learning_rate": 8.84505064148944e-05, "loss": 0.1268, "step": 24150 }, { "epoch": 11.57, "learning_rate": 8.844132648873866e-05, "loss": 0.1214, "step": 24160 }, { "epoch": 11.57, "learning_rate": 8.843214339255226e-05, "loss": 0.131, "step": 24170 }, { "epoch": 11.58, "learning_rate": 8.842295712709245e-05, "loss": 0.1332, "step": 24180 }, { "epoch": 11.58, "learning_rate": 8.84137676931168e-05, "loss": 0.1328, "step": 24190 }, { "epoch": 11.59, "learning_rate": 8.840457509138307e-05, "loss": 0.1326, "step": 24200 }, { "epoch": 11.59, "learning_rate": 8.839537932264936e-05, "loss": 0.1337, "step": 24210 }, { "epoch": 11.6, "learning_rate": 8.838618038767397e-05, "loss": 0.1219, "step": 24220 }, { "epoch": 11.6, "learning_rate": 8.83769782872155e-05, "loss": 0.1535, "step": 24230 }, { "epoch": 11.61, "learning_rate": 8.836777302203277e-05, "loss": 0.131, "step": 24240 }, { "epoch": 11.61, "learning_rate": 8.835856459288489e-05, "loss": 0.1308, "step": 24250 }, { "epoch": 11.62, "learning_rate": 8.834935300053125e-05, "loss": 0.141, "step": 24260 }, { "epoch": 11.62, "learning_rate": 8.834013824573143e-05, "loss": 0.127, "step": 24270 }, { "epoch": 11.62, "learning_rate": 8.833092032924535e-05, "loss": 0.127, "step": 24280 }, { "epoch": 11.63, "learning_rate": 8.832169925183315e-05, "loss": 0.126, "step": 24290 }, { "epoch": 11.63, "learning_rate": 8.831247501425524e-05, "loss": 0.1327, "step": 24300 }, { "epoch": 11.64, "learning_rate": 8.830324761727229e-05, "loss": 0.1502, "step": 24310 }, { "epoch": 11.64, "learning_rate": 8.829401706164523e-05, "loss": 0.1614, "step": 24320 }, { "epoch": 11.65, "learning_rate": 8.828478334813523e-05, "loss": 0.137, "step": 24330 }, { "epoch": 11.65, "learning_rate": 8.827554647750378e-05, "loss": 0.1335, "step": 24340 }, { "epoch": 11.66, "learning_rate": 8.826630645051255e-05, "loss": 0.1571, "step": 24350 }, { "epoch": 11.66, "learning_rate": 8.825706326792354e-05, "loss": 0.1229, "step": 24360 }, { "epoch": 11.67, "learning_rate": 8.824781693049895e-05, "loss": 0.1376, "step": 24370 }, { "epoch": 11.67, "learning_rate": 8.82385674390013e-05, "loss": 0.1251, "step": 24380 }, { "epoch": 11.68, "learning_rate": 8.822931479419334e-05, "loss": 0.1355, "step": 24390 }, { "epoch": 11.68, "learning_rate": 8.822005899683805e-05, "loss": 0.1439, "step": 24400 }, { "epoch": 11.68, "learning_rate": 8.821080004769874e-05, "loss": 0.1202, "step": 24410 }, { "epoch": 11.69, "learning_rate": 8.820153794753891e-05, "loss": 0.1552, "step": 24420 }, { "epoch": 11.69, "learning_rate": 8.81922726971224e-05, "loss": 0.1397, "step": 24430 }, { "epoch": 11.7, "learning_rate": 8.81830042972132e-05, "loss": 0.1217, "step": 24440 }, { "epoch": 11.7, "learning_rate": 8.817373274857564e-05, "loss": 0.1338, "step": 24450 }, { "epoch": 11.71, "learning_rate": 8.816445805197431e-05, "loss": 0.1244, "step": 24460 }, { "epoch": 11.71, "learning_rate": 8.815518020817402e-05, "loss": 0.1436, "step": 24470 }, { "epoch": 11.72, "learning_rate": 8.814589921793987e-05, "loss": 0.123, "step": 24480 }, { "epoch": 11.72, "learning_rate": 8.813661508203721e-05, "loss": 0.1371, "step": 24490 }, { "epoch": 11.73, "learning_rate": 8.812732780123165e-05, "loss": 0.124, "step": 24500 }, { "epoch": 11.73, "learning_rate": 8.811803737628905e-05, "loss": 0.128, "step": 24510 }, { "epoch": 11.73, "learning_rate": 8.810874380797553e-05, "loss": 0.1312, "step": 24520 }, { "epoch": 11.74, "learning_rate": 8.809944709705748e-05, "loss": 0.1446, "step": 24530 }, { "epoch": 11.74, "learning_rate": 8.809014724430156e-05, "loss": 0.145, "step": 24540 }, { "epoch": 11.75, "learning_rate": 8.808084425047465e-05, "loss": 0.1323, "step": 24550 }, { "epoch": 11.75, "learning_rate": 8.807153811634394e-05, "loss": 0.139, "step": 24560 }, { "epoch": 11.76, "learning_rate": 8.806222884267683e-05, "loss": 0.1288, "step": 24570 }, { "epoch": 11.76, "learning_rate": 8.805291643024102e-05, "loss": 0.1277, "step": 24580 }, { "epoch": 11.77, "learning_rate": 8.804360087980443e-05, "loss": 0.1292, "step": 24590 }, { "epoch": 11.77, "learning_rate": 8.803428219213528e-05, "loss": 0.1536, "step": 24600 }, { "epoch": 11.78, "learning_rate": 8.802496036800201e-05, "loss": 0.1376, "step": 24610 }, { "epoch": 11.78, "learning_rate": 8.801563540817335e-05, "loss": 0.1131, "step": 24620 }, { "epoch": 11.78, "learning_rate": 8.800630731341825e-05, "loss": 0.1292, "step": 24630 }, { "epoch": 11.79, "learning_rate": 8.799697608450599e-05, "loss": 0.1368, "step": 24640 }, { "epoch": 11.79, "learning_rate": 8.798764172220601e-05, "loss": 0.1341, "step": 24650 }, { "epoch": 11.8, "learning_rate": 8.797830422728809e-05, "loss": 0.152, "step": 24660 }, { "epoch": 11.8, "learning_rate": 8.796896360052222e-05, "loss": 0.1179, "step": 24670 }, { "epoch": 11.81, "learning_rate": 8.795961984267869e-05, "loss": 0.1248, "step": 24680 }, { "epoch": 11.81, "learning_rate": 8.795027295452802e-05, "loss": 0.1228, "step": 24690 }, { "epoch": 11.82, "learning_rate": 8.794092293684098e-05, "loss": 0.1503, "step": 24700 }, { "epoch": 11.82, "learning_rate": 8.793156979038861e-05, "loss": 0.1282, "step": 24710 }, { "epoch": 11.83, "learning_rate": 8.792221351594222e-05, "loss": 0.1402, "step": 24720 }, { "epoch": 11.83, "learning_rate": 8.791285411427337e-05, "loss": 0.128, "step": 24730 }, { "epoch": 11.84, "learning_rate": 8.790349158615385e-05, "loss": 0.1531, "step": 24740 }, { "epoch": 11.84, "learning_rate": 8.789412593235578e-05, "loss": 0.1367, "step": 24750 }, { "epoch": 11.84, "learning_rate": 8.788475715365144e-05, "loss": 0.1254, "step": 24760 }, { "epoch": 11.85, "learning_rate": 8.787538525081344e-05, "loss": 0.128, "step": 24770 }, { "epoch": 11.85, "learning_rate": 8.786601022461463e-05, "loss": 0.1356, "step": 24780 }, { "epoch": 11.86, "learning_rate": 8.78566320758281e-05, "loss": 0.1112, "step": 24790 }, { "epoch": 11.86, "learning_rate": 8.784725080522723e-05, "loss": 0.1359, "step": 24800 }, { "epoch": 11.87, "learning_rate": 8.783786641358562e-05, "loss": 0.13, "step": 24810 }, { "epoch": 11.87, "learning_rate": 8.782847890167716e-05, "loss": 0.1406, "step": 24820 }, { "epoch": 11.88, "learning_rate": 8.781908827027596e-05, "loss": 0.1301, "step": 24830 }, { "epoch": 11.88, "learning_rate": 8.780969452015645e-05, "loss": 0.1287, "step": 24840 }, { "epoch": 11.89, "learning_rate": 8.780029765209325e-05, "loss": 0.1344, "step": 24850 }, { "epoch": 11.89, "learning_rate": 8.779089766686127e-05, "loss": 0.1314, "step": 24860 }, { "epoch": 11.89, "learning_rate": 8.778149456523565e-05, "loss": 0.1402, "step": 24870 }, { "epoch": 11.9, "learning_rate": 8.777208834799187e-05, "loss": 0.1368, "step": 24880 }, { "epoch": 11.9, "learning_rate": 8.776267901590552e-05, "loss": 0.13, "step": 24890 }, { "epoch": 11.91, "learning_rate": 8.77532665697526e-05, "loss": 0.1349, "step": 24900 }, { "epoch": 11.91, "learning_rate": 8.774385101030928e-05, "loss": 0.1386, "step": 24910 }, { "epoch": 11.92, "learning_rate": 8.773443233835199e-05, "loss": 0.1475, "step": 24920 }, { "epoch": 11.92, "learning_rate": 8.772501055465745e-05, "loss": 0.1213, "step": 24930 }, { "epoch": 11.93, "learning_rate": 8.771558566000261e-05, "loss": 0.1244, "step": 24940 }, { "epoch": 11.93, "learning_rate": 8.770615765516471e-05, "loss": 0.1437, "step": 24950 }, { "epoch": 11.94, "learning_rate": 8.769672654092118e-05, "loss": 0.1284, "step": 24960 }, { "epoch": 11.94, "learning_rate": 8.76872923180498e-05, "loss": 0.139, "step": 24970 }, { "epoch": 11.94, "learning_rate": 8.767785498732849e-05, "loss": 0.1302, "step": 24980 }, { "epoch": 11.95, "learning_rate": 8.766841454953554e-05, "loss": 0.1408, "step": 24990 }, { "epoch": 12.0, "learning_rate": 8.765897100544945e-05, "loss": 0.1364, "step": 25000 }, { "epoch": 12.01, "learning_rate": 8.764952435584894e-05, "loss": 0.1251, "step": 25010 }, { "epoch": 12.01, "learning_rate": 8.764007460151304e-05, "loss": 0.1307, "step": 25020 }, { "epoch": 12.02, "learning_rate": 8.763062174322103e-05, "loss": 0.1405, "step": 25030 }, { "epoch": 12.02, "learning_rate": 8.76211657817524e-05, "loss": 0.1331, "step": 25040 }, { "epoch": 12.02, "learning_rate": 8.761170671788694e-05, "loss": 0.1369, "step": 25050 }, { "epoch": 12.03, "learning_rate": 8.76022445524047e-05, "loss": 0.1258, "step": 25060 }, { "epoch": 12.03, "learning_rate": 8.759277928608596e-05, "loss": 0.1125, "step": 25070 }, { "epoch": 12.04, "learning_rate": 8.758331091971123e-05, "loss": 0.1255, "step": 25080 }, { "epoch": 12.04, "learning_rate": 8.757383945406138e-05, "loss": 0.1298, "step": 25090 }, { "epoch": 12.05, "learning_rate": 8.756436488991741e-05, "loss": 0.1232, "step": 25100 }, { "epoch": 12.05, "learning_rate": 8.755488722806066e-05, "loss": 0.1301, "step": 25110 }, { "epoch": 12.06, "learning_rate": 8.754540646927267e-05, "loss": 0.1222, "step": 25120 }, { "epoch": 12.06, "learning_rate": 8.75359226143353e-05, "loss": 0.1192, "step": 25130 }, { "epoch": 12.07, "learning_rate": 8.75264356640306e-05, "loss": 0.1065, "step": 25140 }, { "epoch": 12.07, "learning_rate": 8.75169456191409e-05, "loss": 0.1225, "step": 25150 }, { "epoch": 12.07, "learning_rate": 8.750745248044881e-05, "loss": 0.1207, "step": 25160 }, { "epoch": 12.08, "learning_rate": 8.749795624873716e-05, "loss": 0.1297, "step": 25170 }, { "epoch": 12.08, "learning_rate": 8.748845692478905e-05, "loss": 0.1125, "step": 25180 }, { "epoch": 12.09, "learning_rate": 8.747895450938782e-05, "loss": 0.1036, "step": 25190 }, { "epoch": 12.09, "learning_rate": 8.746944900331711e-05, "loss": 0.1352, "step": 25200 }, { "epoch": 12.1, "learning_rate": 8.745994040736077e-05, "loss": 0.1247, "step": 25210 }, { "epoch": 12.1, "learning_rate": 8.74504287223029e-05, "loss": 0.1134, "step": 25220 }, { "epoch": 12.11, "learning_rate": 8.74409139489279e-05, "loss": 0.1192, "step": 25230 }, { "epoch": 12.11, "learning_rate": 8.743139608802038e-05, "loss": 0.1206, "step": 25240 }, { "epoch": 12.12, "learning_rate": 8.742187514036522e-05, "loss": 0.1326, "step": 25250 }, { "epoch": 12.12, "learning_rate": 8.741235110674757e-05, "loss": 0.1309, "step": 25260 }, { "epoch": 12.13, "learning_rate": 8.740282398795281e-05, "loss": 0.1285, "step": 25270 }, { "epoch": 12.13, "learning_rate": 8.73932937847666e-05, "loss": 0.1349, "step": 25280 }, { "epoch": 12.13, "learning_rate": 8.738376049797482e-05, "loss": 0.1257, "step": 25290 }, { "epoch": 12.14, "learning_rate": 8.737422412836364e-05, "loss": 0.1109, "step": 25300 }, { "epoch": 12.14, "learning_rate": 8.736468467671945e-05, "loss": 0.1329, "step": 25310 }, { "epoch": 12.15, "learning_rate": 8.735514214382894e-05, "loss": 0.1126, "step": 25320 }, { "epoch": 12.15, "learning_rate": 8.734559653047901e-05, "loss": 0.1288, "step": 25330 }, { "epoch": 12.16, "learning_rate": 8.733604783745683e-05, "loss": 0.1289, "step": 25340 }, { "epoch": 12.16, "learning_rate": 8.732649606554984e-05, "loss": 0.137, "step": 25350 }, { "epoch": 12.17, "learning_rate": 8.731694121554569e-05, "loss": 0.124, "step": 25360 }, { "epoch": 12.17, "learning_rate": 8.730738328823234e-05, "loss": 0.1232, "step": 25370 }, { "epoch": 12.18, "learning_rate": 8.729782228439796e-05, "loss": 0.1361, "step": 25380 }, { "epoch": 12.18, "learning_rate": 8.728825820483099e-05, "loss": 0.1233, "step": 25390 }, { "epoch": 12.18, "learning_rate": 8.727869105032014e-05, "loss": 0.1233, "step": 25400 }, { "epoch": 12.19, "learning_rate": 8.726912082165432e-05, "loss": 0.1145, "step": 25410 }, { "epoch": 12.19, "learning_rate": 8.725954751962279e-05, "loss": 0.1081, "step": 25420 }, { "epoch": 12.2, "learning_rate": 8.724997114501495e-05, "loss": 0.1372, "step": 25430 }, { "epoch": 12.2, "learning_rate": 8.724039169862053e-05, "loss": 0.1235, "step": 25440 }, { "epoch": 12.21, "learning_rate": 8.72308091812295e-05, "loss": 0.1213, "step": 25450 }, { "epoch": 12.21, "learning_rate": 8.722122359363204e-05, "loss": 0.1263, "step": 25460 }, { "epoch": 12.22, "learning_rate": 8.721163493661865e-05, "loss": 0.1314, "step": 25470 }, { "epoch": 12.22, "learning_rate": 8.720204321098004e-05, "loss": 0.1206, "step": 25480 }, { "epoch": 12.23, "learning_rate": 8.719244841750719e-05, "loss": 0.1326, "step": 25490 }, { "epoch": 12.23, "learning_rate": 8.718285055699131e-05, "loss": 0.1408, "step": 25500 }, { "epoch": 12.23, "learning_rate": 8.71732496302239e-05, "loss": 0.1311, "step": 25510 }, { "epoch": 12.24, "learning_rate": 8.716364563799667e-05, "loss": 0.1319, "step": 25520 }, { "epoch": 12.24, "learning_rate": 8.715403858110161e-05, "loss": 0.1247, "step": 25530 }, { "epoch": 12.25, "learning_rate": 8.714442846033097e-05, "loss": 0.1335, "step": 25540 }, { "epoch": 12.25, "learning_rate": 8.713481527647724e-05, "loss": 0.1171, "step": 25550 }, { "epoch": 12.26, "learning_rate": 8.712519903033317e-05, "loss": 0.1326, "step": 25560 }, { "epoch": 12.26, "learning_rate": 8.711557972269171e-05, "loss": 0.1167, "step": 25570 }, { "epoch": 12.27, "learning_rate": 8.710595735434615e-05, "loss": 0.1133, "step": 25580 }, { "epoch": 12.27, "learning_rate": 8.709633192608999e-05, "loss": 0.1115, "step": 25590 }, { "epoch": 12.28, "learning_rate": 8.708670343871697e-05, "loss": 0.1223, "step": 25600 }, { "epoch": 12.28, "learning_rate": 8.707707189302108e-05, "loss": 0.1352, "step": 25610 }, { "epoch": 12.29, "learning_rate": 8.70674372897966e-05, "loss": 0.1185, "step": 25620 }, { "epoch": 12.29, "learning_rate": 8.705779962983804e-05, "loss": 0.1106, "step": 25630 }, { "epoch": 12.29, "learning_rate": 8.704815891394016e-05, "loss": 0.1171, "step": 25640 }, { "epoch": 12.3, "learning_rate": 8.703851514289797e-05, "loss": 0.1355, "step": 25650 }, { "epoch": 12.3, "learning_rate": 8.702886831750673e-05, "loss": 0.1165, "step": 25660 }, { "epoch": 12.31, "learning_rate": 8.701921843856196e-05, "loss": 0.1096, "step": 25670 }, { "epoch": 12.31, "learning_rate": 8.700956550685942e-05, "loss": 0.1195, "step": 25680 }, { "epoch": 12.32, "learning_rate": 8.699990952319515e-05, "loss": 0.1376, "step": 25690 }, { "epoch": 12.32, "learning_rate": 8.699025048836541e-05, "loss": 0.1207, "step": 25700 }, { "epoch": 12.33, "learning_rate": 8.698058840316672e-05, "loss": 0.1148, "step": 25710 }, { "epoch": 12.33, "learning_rate": 8.697092326839586e-05, "loss": 0.134, "step": 25720 }, { "epoch": 12.34, "learning_rate": 8.696125508484989e-05, "loss": 0.1259, "step": 25730 }, { "epoch": 12.34, "learning_rate": 8.695158385332603e-05, "loss": 0.1122, "step": 25740 }, { "epoch": 12.34, "learning_rate": 8.694190957462183e-05, "loss": 0.1298, "step": 25750 }, { "epoch": 12.35, "learning_rate": 8.693223224953509e-05, "loss": 0.1356, "step": 25760 }, { "epoch": 12.35, "learning_rate": 8.692255187886384e-05, "loss": 0.131, "step": 25770 }, { "epoch": 12.36, "learning_rate": 8.691286846340634e-05, "loss": 0.1284, "step": 25780 }, { "epoch": 12.36, "learning_rate": 8.690318200396113e-05, "loss": 0.121, "step": 25790 }, { "epoch": 12.37, "learning_rate": 8.689349250132702e-05, "loss": 0.1138, "step": 25800 }, { "epoch": 12.37, "learning_rate": 8.688379995630302e-05, "loss": 0.1241, "step": 25810 }, { "epoch": 12.38, "learning_rate": 8.687410436968842e-05, "loss": 0.1258, "step": 25820 }, { "epoch": 12.38, "learning_rate": 8.686440574228279e-05, "loss": 0.1247, "step": 25830 }, { "epoch": 12.39, "learning_rate": 8.685470407488587e-05, "loss": 0.1307, "step": 25840 }, { "epoch": 12.39, "learning_rate": 8.684499936829774e-05, "loss": 0.1498, "step": 25850 }, { "epoch": 12.39, "learning_rate": 8.683529162331866e-05, "loss": 0.117, "step": 25860 }, { "epoch": 12.4, "learning_rate": 8.682558084074919e-05, "loss": 0.1116, "step": 25870 }, { "epoch": 12.4, "learning_rate": 8.681586702139012e-05, "loss": 0.1306, "step": 25880 }, { "epoch": 12.41, "learning_rate": 8.680615016604249e-05, "loss": 0.1151, "step": 25890 }, { "epoch": 12.41, "learning_rate": 8.67964302755076e-05, "loss": 0.1167, "step": 25900 }, { "epoch": 12.42, "learning_rate": 8.678670735058696e-05, "loss": 0.1219, "step": 25910 }, { "epoch": 12.42, "learning_rate": 8.677698139208238e-05, "loss": 0.1141, "step": 25920 }, { "epoch": 12.43, "learning_rate": 8.676725240079593e-05, "loss": 0.1116, "step": 25930 }, { "epoch": 12.43, "learning_rate": 8.675752037752987e-05, "loss": 0.1611, "step": 25940 }, { "epoch": 12.44, "learning_rate": 8.674778532308677e-05, "loss": 0.1452, "step": 25950 }, { "epoch": 12.44, "learning_rate": 8.673804723826938e-05, "loss": 0.1353, "step": 25960 }, { "epoch": 12.45, "learning_rate": 8.672830612388076e-05, "loss": 0.1214, "step": 25970 }, { "epoch": 12.45, "learning_rate": 8.671856198072425e-05, "loss": 0.1369, "step": 25980 }, { "epoch": 12.45, "learning_rate": 8.670881480960333e-05, "loss": 0.1321, "step": 25990 }, { "epoch": 12.46, "learning_rate": 8.669906461132182e-05, "loss": 0.1235, "step": 26000 }, { "epoch": 12.46, "learning_rate": 8.668931138668374e-05, "loss": 0.1273, "step": 26010 }, { "epoch": 12.47, "learning_rate": 8.667955513649343e-05, "loss": 0.142, "step": 26020 }, { "epoch": 12.47, "learning_rate": 8.666979586155538e-05, "loss": 0.1262, "step": 26030 }, { "epoch": 12.48, "learning_rate": 8.666003356267441e-05, "loss": 0.1124, "step": 26040 }, { "epoch": 12.48, "learning_rate": 8.665026824065554e-05, "loss": 0.1234, "step": 26050 }, { "epoch": 12.49, "learning_rate": 8.664049989630408e-05, "loss": 0.1263, "step": 26060 }, { "epoch": 12.49, "learning_rate": 8.663072853042554e-05, "loss": 0.1265, "step": 26070 }, { "epoch": 12.5, "learning_rate": 8.662095414382571e-05, "loss": 0.1194, "step": 26080 }, { "epoch": 12.5, "learning_rate": 8.661117673731067e-05, "loss": 0.1226, "step": 26090 }, { "epoch": 12.5, "learning_rate": 8.660139631168667e-05, "loss": 0.1473, "step": 26100 }, { "epoch": 12.51, "learning_rate": 8.659161286776023e-05, "loss": 0.1497, "step": 26110 }, { "epoch": 12.51, "learning_rate": 8.658182640633817e-05, "loss": 0.1175, "step": 26120 }, { "epoch": 12.52, "learning_rate": 8.65720369282275e-05, "loss": 0.1198, "step": 26130 }, { "epoch": 12.52, "learning_rate": 8.65622444342355e-05, "loss": 0.1244, "step": 26140 }, { "epoch": 12.53, "learning_rate": 8.655244892516971e-05, "loss": 0.1391, "step": 26150 }, { "epoch": 12.53, "learning_rate": 8.654265040183788e-05, "loss": 0.1223, "step": 26160 }, { "epoch": 12.54, "learning_rate": 8.653284886504808e-05, "loss": 0.141, "step": 26170 }, { "epoch": 12.54, "learning_rate": 8.652304431560858e-05, "loss": 0.1124, "step": 26180 }, { "epoch": 12.55, "learning_rate": 8.651323675432787e-05, "loss": 0.1144, "step": 26190 }, { "epoch": 12.55, "learning_rate": 8.650342618201475e-05, "loss": 0.131, "step": 26200 }, { "epoch": 12.55, "learning_rate": 8.649361259947821e-05, "loss": 0.1327, "step": 26210 }, { "epoch": 12.56, "learning_rate": 8.648379600752756e-05, "loss": 0.1423, "step": 26220 }, { "epoch": 12.56, "learning_rate": 8.64739764069723e-05, "loss": 0.1159, "step": 26230 }, { "epoch": 12.57, "learning_rate": 8.646415379862217e-05, "loss": 0.1189, "step": 26240 }, { "epoch": 12.57, "learning_rate": 8.645432818328721e-05, "loss": 0.1403, "step": 26250 }, { "epoch": 12.58, "learning_rate": 8.644449956177769e-05, "loss": 0.1373, "step": 26260 }, { "epoch": 12.58, "learning_rate": 8.643466793490409e-05, "loss": 0.1334, "step": 26270 }, { "epoch": 12.59, "learning_rate": 8.642483330347718e-05, "loss": 0.1062, "step": 26280 }, { "epoch": 12.59, "learning_rate": 8.641499566830798e-05, "loss": 0.1444, "step": 26290 }, { "epoch": 12.6, "learning_rate": 8.640515503020769e-05, "loss": 0.1274, "step": 26300 }, { "epoch": 12.6, "learning_rate": 8.63953113899879e-05, "loss": 0.1465, "step": 26310 }, { "epoch": 12.61, "learning_rate": 8.638546474846025e-05, "loss": 0.1263, "step": 26320 }, { "epoch": 12.61, "learning_rate": 8.637561510643681e-05, "loss": 0.1294, "step": 26330 }, { "epoch": 12.61, "learning_rate": 8.63657624647298e-05, "loss": 0.1169, "step": 26340 }, { "epoch": 12.62, "learning_rate": 8.635590682415171e-05, "loss": 0.1445, "step": 26350 }, { "epoch": 12.62, "learning_rate": 8.634604818551529e-05, "loss": 0.1267, "step": 26360 }, { "epoch": 12.63, "learning_rate": 8.63361865496335e-05, "loss": 0.1373, "step": 26370 }, { "epoch": 12.63, "learning_rate": 8.632632191731957e-05, "loss": 0.1173, "step": 26380 }, { "epoch": 12.64, "learning_rate": 8.631645428938701e-05, "loss": 0.1217, "step": 26390 }, { "epoch": 12.64, "learning_rate": 8.630658366664951e-05, "loss": 0.1167, "step": 26400 }, { "epoch": 12.65, "learning_rate": 8.629671004992106e-05, "loss": 0.114, "step": 26410 }, { "epoch": 12.65, "learning_rate": 8.628683344001589e-05, "loss": 0.1283, "step": 26420 }, { "epoch": 12.66, "learning_rate": 8.627695383774845e-05, "loss": 0.125, "step": 26430 }, { "epoch": 12.66, "learning_rate": 8.626707124393345e-05, "loss": 0.1369, "step": 26440 }, { "epoch": 12.66, "learning_rate": 8.625718565938585e-05, "loss": 0.1303, "step": 26450 }, { "epoch": 12.67, "learning_rate": 8.624729708492088e-05, "loss": 0.1279, "step": 26460 }, { "epoch": 12.67, "learning_rate": 8.623740552135396e-05, "loss": 0.1145, "step": 26470 }, { "epoch": 12.68, "learning_rate": 8.622751096950081e-05, "loss": 0.1301, "step": 26480 }, { "epoch": 12.68, "learning_rate": 8.621761343017737e-05, "loss": 0.116, "step": 26490 }, { "epoch": 12.69, "learning_rate": 8.620771290419982e-05, "loss": 0.1246, "step": 26500 }, { "epoch": 12.69, "learning_rate": 8.619780939238462e-05, "loss": 0.1196, "step": 26510 }, { "epoch": 12.7, "learning_rate": 8.618790289554845e-05, "loss": 0.1227, "step": 26520 }, { "epoch": 12.7, "learning_rate": 8.617799341450823e-05, "loss": 0.1283, "step": 26530 }, { "epoch": 12.71, "learning_rate": 8.616808095008114e-05, "loss": 0.1163, "step": 26540 }, { "epoch": 12.71, "learning_rate": 8.61581655030846e-05, "loss": 0.1124, "step": 26550 }, { "epoch": 12.71, "learning_rate": 8.61482470743363e-05, "loss": 0.1323, "step": 26560 }, { "epoch": 12.72, "learning_rate": 8.613832566465412e-05, "loss": 0.123, "step": 26570 }, { "epoch": 12.72, "learning_rate": 8.612840127485624e-05, "loss": 0.1364, "step": 26580 }, { "epoch": 12.73, "learning_rate": 8.611847390576107e-05, "loss": 0.1287, "step": 26590 }, { "epoch": 12.73, "learning_rate": 8.610854355818727e-05, "loss": 0.1278, "step": 26600 }, { "epoch": 12.74, "learning_rate": 8.609861023295372e-05, "loss": 0.1321, "step": 26610 }, { "epoch": 12.74, "learning_rate": 8.608867393087955e-05, "loss": 0.1156, "step": 26620 }, { "epoch": 12.75, "learning_rate": 8.607873465278418e-05, "loss": 0.1172, "step": 26630 }, { "epoch": 12.75, "learning_rate": 8.606879239948722e-05, "loss": 0.1163, "step": 26640 }, { "epoch": 12.76, "learning_rate": 8.605884717180856e-05, "loss": 0.1407, "step": 26650 }, { "epoch": 12.76, "learning_rate": 8.604889897056831e-05, "loss": 0.1103, "step": 26660 }, { "epoch": 12.77, "learning_rate": 8.603894779658688e-05, "loss": 0.1359, "step": 26670 }, { "epoch": 12.77, "learning_rate": 8.602899365068484e-05, "loss": 0.1273, "step": 26680 }, { "epoch": 12.77, "learning_rate": 8.601903653368308e-05, "loss": 0.1249, "step": 26690 }, { "epoch": 12.78, "learning_rate": 8.60090764464027e-05, "loss": 0.1129, "step": 26700 }, { "epoch": 12.78, "learning_rate": 8.599911338966502e-05, "loss": 0.1262, "step": 26710 }, { "epoch": 12.79, "learning_rate": 8.598914736429166e-05, "loss": 0.1418, "step": 26720 }, { "epoch": 12.79, "learning_rate": 8.597917837110448e-05, "loss": 0.1415, "step": 26730 }, { "epoch": 12.8, "learning_rate": 8.596920641092553e-05, "loss": 0.1352, "step": 26740 }, { "epoch": 12.8, "learning_rate": 8.595923148457715e-05, "loss": 0.1182, "step": 26750 }, { "epoch": 12.81, "learning_rate": 8.594925359288191e-05, "loss": 0.1317, "step": 26760 }, { "epoch": 12.81, "learning_rate": 8.593927273666265e-05, "loss": 0.1303, "step": 26770 }, { "epoch": 12.82, "learning_rate": 8.592928891674239e-05, "loss": 0.1448, "step": 26780 }, { "epoch": 12.82, "learning_rate": 8.591930213394449e-05, "loss": 0.1305, "step": 26790 }, { "epoch": 12.82, "learning_rate": 8.590931238909245e-05, "loss": 0.1205, "step": 26800 }, { "epoch": 12.83, "learning_rate": 8.589931968301011e-05, "loss": 0.1099, "step": 26810 }, { "epoch": 12.83, "learning_rate": 8.588932401652148e-05, "loss": 0.1314, "step": 26820 }, { "epoch": 12.84, "learning_rate": 8.587932539045085e-05, "loss": 0.1253, "step": 26830 }, { "epoch": 12.84, "learning_rate": 8.586932380562276e-05, "loss": 0.1278, "step": 26840 }, { "epoch": 12.85, "learning_rate": 8.585931926286198e-05, "loss": 0.1326, "step": 26850 }, { "epoch": 12.85, "learning_rate": 8.584931176299348e-05, "loss": 0.1411, "step": 26860 }, { "epoch": 12.86, "learning_rate": 8.58393013068426e-05, "loss": 0.1194, "step": 26870 }, { "epoch": 12.86, "learning_rate": 8.582928789523479e-05, "loss": 0.1197, "step": 26880 }, { "epoch": 12.87, "learning_rate": 8.581927152899582e-05, "loss": 0.1312, "step": 26890 }, { "epoch": 12.87, "learning_rate": 8.580925220895166e-05, "loss": 0.1269, "step": 26900 }, { "epoch": 12.87, "learning_rate": 8.579922993592854e-05, "loss": 0.1533, "step": 26910 }, { "epoch": 12.88, "learning_rate": 8.578920471075298e-05, "loss": 0.1322, "step": 26920 }, { "epoch": 12.88, "learning_rate": 8.577917653425167e-05, "loss": 0.1292, "step": 26930 }, { "epoch": 12.89, "learning_rate": 8.576914540725159e-05, "loss": 0.1294, "step": 26940 }, { "epoch": 12.89, "learning_rate": 8.575911133057992e-05, "loss": 0.1253, "step": 26950 }, { "epoch": 12.9, "learning_rate": 8.574907430506413e-05, "loss": 0.1553, "step": 26960 }, { "epoch": 12.9, "learning_rate": 8.573903433153192e-05, "loss": 0.1284, "step": 26970 }, { "epoch": 12.91, "learning_rate": 8.572899141081124e-05, "loss": 0.1258, "step": 26980 }, { "epoch": 12.91, "learning_rate": 8.571894554373021e-05, "loss": 0.1192, "step": 26990 }, { "epoch": 12.92, "learning_rate": 8.570889673111733e-05, "loss": 0.1288, "step": 27000 }, { "epoch": 12.92, "learning_rate": 8.569884497380123e-05, "loss": 0.1438, "step": 27010 }, { "epoch": 12.93, "learning_rate": 8.56887902726108e-05, "loss": 0.1158, "step": 27020 }, { "epoch": 12.93, "learning_rate": 8.567873262837525e-05, "loss": 0.1363, "step": 27030 }, { "epoch": 12.93, "learning_rate": 8.566867204192392e-05, "loss": 0.1588, "step": 27040 }, { "epoch": 12.94, "learning_rate": 8.565860851408646e-05, "loss": 0.1633, "step": 27050 }, { "epoch": 12.94, "learning_rate": 8.564854204569277e-05, "loss": 0.1365, "step": 27060 }, { "epoch": 12.95, "learning_rate": 8.563847263757297e-05, "loss": 0.1164, "step": 27070 }, { "epoch": 13.0, "learning_rate": 8.562840029055739e-05, "loss": 0.1464, "step": 27080 }, { "epoch": 13.01, "learning_rate": 8.561832500547668e-05, "loss": 0.1292, "step": 27090 }, { "epoch": 13.01, "learning_rate": 8.560824678316167e-05, "loss": 0.1245, "step": 27100 }, { "epoch": 13.01, "learning_rate": 8.559816562444344e-05, "loss": 0.1119, "step": 27110 }, { "epoch": 13.02, "learning_rate": 8.558808153015335e-05, "loss": 0.1111, "step": 27120 }, { "epoch": 13.02, "learning_rate": 8.557799450112296e-05, "loss": 0.1144, "step": 27130 }, { "epoch": 13.03, "learning_rate": 8.55679045381841e-05, "loss": 0.1293, "step": 27140 }, { "epoch": 13.03, "learning_rate": 8.555781164216883e-05, "loss": 0.1241, "step": 27150 }, { "epoch": 13.04, "learning_rate": 8.554771581390942e-05, "loss": 0.1136, "step": 27160 }, { "epoch": 13.04, "learning_rate": 8.553761705423845e-05, "loss": 0.1285, "step": 27170 }, { "epoch": 13.05, "learning_rate": 8.552751536398869e-05, "loss": 0.1127, "step": 27180 }, { "epoch": 13.05, "learning_rate": 8.551741074399318e-05, "loss": 0.1539, "step": 27190 }, { "epoch": 13.06, "learning_rate": 8.550730319508516e-05, "loss": 0.1187, "step": 27200 }, { "epoch": 13.06, "learning_rate": 8.549719271809818e-05, "loss": 0.1175, "step": 27210 }, { "epoch": 13.06, "learning_rate": 8.548707931386597e-05, "loss": 0.1286, "step": 27220 }, { "epoch": 13.07, "learning_rate": 8.547696298322251e-05, "loss": 0.1106, "step": 27230 }, { "epoch": 13.07, "learning_rate": 8.546684372700205e-05, "loss": 0.1248, "step": 27240 }, { "epoch": 13.08, "learning_rate": 8.545672154603908e-05, "loss": 0.1072, "step": 27250 }, { "epoch": 13.08, "learning_rate": 8.544659644116829e-05, "loss": 0.1091, "step": 27260 }, { "epoch": 13.09, "learning_rate": 8.543646841322464e-05, "loss": 0.1231, "step": 27270 }, { "epoch": 13.09, "learning_rate": 8.542633746304335e-05, "loss": 0.1441, "step": 27280 }, { "epoch": 13.1, "learning_rate": 8.541620359145985e-05, "loss": 0.1199, "step": 27290 }, { "epoch": 13.1, "learning_rate": 8.540606679930982e-05, "loss": 0.1153, "step": 27300 }, { "epoch": 13.11, "learning_rate": 8.539592708742918e-05, "loss": 0.1297, "step": 27310 }, { "epoch": 13.11, "learning_rate": 8.538578445665409e-05, "loss": 0.1226, "step": 27320 }, { "epoch": 13.11, "learning_rate": 8.537563890782096e-05, "loss": 0.1057, "step": 27330 }, { "epoch": 13.12, "learning_rate": 8.536549044176643e-05, "loss": 0.1251, "step": 27340 }, { "epoch": 13.12, "learning_rate": 8.535533905932738e-05, "loss": 0.1382, "step": 27350 }, { "epoch": 13.13, "learning_rate": 8.534518476134094e-05, "loss": 0.1218, "step": 27360 }, { "epoch": 13.13, "learning_rate": 8.533502754864449e-05, "loss": 0.1199, "step": 27370 }, { "epoch": 13.14, "learning_rate": 8.532486742207561e-05, "loss": 0.1348, "step": 27380 }, { "epoch": 13.14, "learning_rate": 8.531470438247215e-05, "loss": 0.1109, "step": 27390 }, { "epoch": 13.15, "learning_rate": 8.530453843067222e-05, "loss": 0.1169, "step": 27400 }, { "epoch": 13.15, "learning_rate": 8.529436956751412e-05, "loss": 0.1255, "step": 27410 }, { "epoch": 13.16, "learning_rate": 8.528419779383642e-05, "loss": 0.1317, "step": 27420 }, { "epoch": 13.16, "learning_rate": 8.527402311047794e-05, "loss": 0.1212, "step": 27430 }, { "epoch": 13.16, "learning_rate": 8.526384551827773e-05, "loss": 0.1122, "step": 27440 }, { "epoch": 13.17, "learning_rate": 8.525366501807506e-05, "loss": 0.1264, "step": 27450 }, { "epoch": 13.17, "learning_rate": 8.524348161070945e-05, "loss": 0.1042, "step": 27460 }, { "epoch": 13.18, "learning_rate": 8.52332952970207e-05, "loss": 0.1073, "step": 27470 }, { "epoch": 13.18, "learning_rate": 8.522310607784878e-05, "loss": 0.1235, "step": 27480 }, { "epoch": 13.19, "learning_rate": 8.521291395403395e-05, "loss": 0.1155, "step": 27490 }, { "epoch": 13.19, "learning_rate": 8.520271892641668e-05, "loss": 0.1191, "step": 27500 }, { "epoch": 13.2, "learning_rate": 8.519252099583772e-05, "loss": 0.1106, "step": 27510 }, { "epoch": 13.2, "learning_rate": 8.518232016313802e-05, "loss": 0.133, "step": 27520 }, { "epoch": 13.21, "learning_rate": 8.517211642915878e-05, "loss": 0.1053, "step": 27530 }, { "epoch": 13.21, "learning_rate": 8.516190979474146e-05, "loss": 0.1386, "step": 27540 }, { "epoch": 13.22, "learning_rate": 8.51517002607277e-05, "loss": 0.1091, "step": 27550 }, { "epoch": 13.22, "learning_rate": 8.514148782795948e-05, "loss": 0.1379, "step": 27560 }, { "epoch": 13.22, "learning_rate": 8.513127249727889e-05, "loss": 0.1185, "step": 27570 }, { "epoch": 13.23, "learning_rate": 8.512105426952839e-05, "loss": 0.1262, "step": 27580 }, { "epoch": 13.23, "learning_rate": 8.511083314555059e-05, "loss": 0.1273, "step": 27590 }, { "epoch": 13.24, "learning_rate": 8.510060912618835e-05, "loss": 0.1181, "step": 27600 }, { "epoch": 13.24, "learning_rate": 8.509038221228483e-05, "loss": 0.1383, "step": 27610 }, { "epoch": 13.25, "learning_rate": 8.508015240468335e-05, "loss": 0.1108, "step": 27620 }, { "epoch": 13.25, "learning_rate": 8.50699197042275e-05, "loss": 0.1195, "step": 27630 }, { "epoch": 13.26, "learning_rate": 8.505968411176112e-05, "loss": 0.1081, "step": 27640 }, { "epoch": 13.26, "learning_rate": 8.504944562812829e-05, "loss": 0.1249, "step": 27650 }, { "epoch": 13.27, "learning_rate": 8.503920425417329e-05, "loss": 0.1269, "step": 27660 }, { "epoch": 13.27, "learning_rate": 8.50289599907407e-05, "loss": 0.1245, "step": 27670 }, { "epoch": 13.27, "learning_rate": 8.501871283867527e-05, "loss": 0.1137, "step": 27680 }, { "epoch": 13.28, "learning_rate": 8.500846279882205e-05, "loss": 0.1159, "step": 27690 }, { "epoch": 13.28, "learning_rate": 8.49982098720263e-05, "loss": 0.1107, "step": 27700 }, { "epoch": 13.29, "learning_rate": 8.498795405913349e-05, "loss": 0.1088, "step": 27710 }, { "epoch": 13.29, "learning_rate": 8.497769536098938e-05, "loss": 0.1301, "step": 27720 }, { "epoch": 13.3, "learning_rate": 8.496743377843995e-05, "loss": 0.1187, "step": 27730 }, { "epoch": 13.3, "learning_rate": 8.49571693123314e-05, "loss": 0.1152, "step": 27740 }, { "epoch": 13.31, "learning_rate": 8.494690196351018e-05, "loss": 0.1189, "step": 27750 }, { "epoch": 13.31, "learning_rate": 8.493663173282298e-05, "loss": 0.1201, "step": 27760 }, { "epoch": 13.32, "learning_rate": 8.492635862111674e-05, "loss": 0.1083, "step": 27770 }, { "epoch": 13.32, "learning_rate": 8.491608262923859e-05, "loss": 0.1266, "step": 27780 }, { "epoch": 13.32, "learning_rate": 8.490580375803596e-05, "loss": 0.1239, "step": 27790 }, { "epoch": 13.33, "learning_rate": 8.489552200835648e-05, "loss": 0.1359, "step": 27800 }, { "epoch": 13.33, "learning_rate": 8.488523738104803e-05, "loss": 0.1207, "step": 27810 }, { "epoch": 13.34, "learning_rate": 8.48749498769587e-05, "loss": 0.1288, "step": 27820 }, { "epoch": 13.34, "learning_rate": 8.486465949693687e-05, "loss": 0.1279, "step": 27830 }, { "epoch": 13.35, "learning_rate": 8.485436624183112e-05, "loss": 0.1144, "step": 27840 }, { "epoch": 13.35, "learning_rate": 8.484407011249027e-05, "loss": 0.1196, "step": 27850 }, { "epoch": 13.36, "learning_rate": 8.483377110976336e-05, "loss": 0.1186, "step": 27860 }, { "epoch": 13.36, "learning_rate": 8.482346923449974e-05, "loss": 0.1266, "step": 27870 }, { "epoch": 13.37, "learning_rate": 8.48131644875489e-05, "loss": 0.1396, "step": 27880 }, { "epoch": 13.37, "learning_rate": 8.480285686976062e-05, "loss": 0.1187, "step": 27890 }, { "epoch": 13.38, "learning_rate": 8.479254638198493e-05, "loss": 0.1163, "step": 27900 }, { "epoch": 13.38, "learning_rate": 8.478223302507203e-05, "loss": 0.1177, "step": 27910 }, { "epoch": 13.38, "learning_rate": 8.477191679987246e-05, "loss": 0.1192, "step": 27920 }, { "epoch": 13.39, "learning_rate": 8.47615977072369e-05, "loss": 0.1246, "step": 27930 }, { "epoch": 13.39, "learning_rate": 8.475127574801633e-05, "loss": 0.1437, "step": 27940 }, { "epoch": 13.4, "learning_rate": 8.474095092306191e-05, "loss": 0.1198, "step": 27950 }, { "epoch": 13.4, "learning_rate": 8.473062323322509e-05, "loss": 0.1264, "step": 27960 }, { "epoch": 13.41, "learning_rate": 8.472029267935753e-05, "loss": 0.1342, "step": 27970 }, { "epoch": 13.41, "learning_rate": 8.470995926231114e-05, "loss": 0.1271, "step": 27980 }, { "epoch": 13.42, "learning_rate": 8.469962298293803e-05, "loss": 0.1232, "step": 27990 }, { "epoch": 13.42, "learning_rate": 8.46892838420906e-05, "loss": 0.1231, "step": 28000 }, { "epoch": 13.43, "learning_rate": 8.467894184062144e-05, "loss": 0.128, "step": 28010 }, { "epoch": 13.43, "learning_rate": 8.466859697938339e-05, "loss": 0.1359, "step": 28020 }, { "epoch": 13.43, "learning_rate": 8.465824925922956e-05, "loss": 0.1432, "step": 28030 }, { "epoch": 13.44, "learning_rate": 8.464789868101324e-05, "loss": 0.1191, "step": 28040 }, { "epoch": 13.44, "learning_rate": 8.463754524558797e-05, "loss": 0.1054, "step": 28050 }, { "epoch": 13.45, "learning_rate": 8.462718895380759e-05, "loss": 0.1144, "step": 28060 }, { "epoch": 13.45, "learning_rate": 8.461682980652607e-05, "loss": 0.1179, "step": 28070 }, { "epoch": 13.46, "learning_rate": 8.46064678045977e-05, "loss": 0.137, "step": 28080 }, { "epoch": 13.46, "learning_rate": 8.459610294887692e-05, "loss": 0.1147, "step": 28090 }, { "epoch": 13.47, "learning_rate": 8.458573524021854e-05, "loss": 0.1308, "step": 28100 }, { "epoch": 13.47, "learning_rate": 8.457536467947745e-05, "loss": 0.1175, "step": 28110 }, { "epoch": 13.48, "learning_rate": 8.45649912675089e-05, "loss": 0.1211, "step": 28120 }, { "epoch": 13.48, "learning_rate": 8.45546150051683e-05, "loss": 0.1123, "step": 28130 }, { "epoch": 13.48, "learning_rate": 8.454423589331132e-05, "loss": 0.114, "step": 28140 }, { "epoch": 13.49, "learning_rate": 8.453385393279387e-05, "loss": 0.1256, "step": 28150 }, { "epoch": 13.49, "learning_rate": 8.45234691244721e-05, "loss": 0.1301, "step": 28160 }, { "epoch": 13.5, "learning_rate": 8.451308146920236e-05, "loss": 0.1182, "step": 28170 }, { "epoch": 13.5, "learning_rate": 8.450269096784129e-05, "loss": 0.1335, "step": 28180 }, { "epoch": 13.51, "learning_rate": 8.449229762124571e-05, "loss": 0.1267, "step": 28190 }, { "epoch": 13.51, "learning_rate": 8.448190143027269e-05, "loss": 0.1268, "step": 28200 }, { "epoch": 13.52, "learning_rate": 8.447150239577957e-05, "loss": 0.1239, "step": 28210 }, { "epoch": 13.52, "learning_rate": 8.446110051862388e-05, "loss": 0.1142, "step": 28220 }, { "epoch": 13.53, "learning_rate": 8.44506957996634e-05, "loss": 0.1109, "step": 28230 }, { "epoch": 13.53, "learning_rate": 8.444028823975618e-05, "loss": 0.1216, "step": 28240 }, { "epoch": 13.54, "learning_rate": 8.442987783976042e-05, "loss": 0.1165, "step": 28250 }, { "epoch": 13.54, "learning_rate": 8.441946460053463e-05, "loss": 0.1382, "step": 28260 }, { "epoch": 13.54, "learning_rate": 8.440904852293752e-05, "loss": 0.1283, "step": 28270 }, { "epoch": 13.55, "learning_rate": 8.439862960782807e-05, "loss": 0.1333, "step": 28280 }, { "epoch": 13.55, "learning_rate": 8.438820785606543e-05, "loss": 0.1206, "step": 28290 }, { "epoch": 13.56, "learning_rate": 8.437778326850903e-05, "loss": 0.1264, "step": 28300 }, { "epoch": 13.56, "learning_rate": 8.436735584601853e-05, "loss": 0.1469, "step": 28310 }, { "epoch": 13.57, "learning_rate": 8.435692558945385e-05, "loss": 0.1235, "step": 28320 }, { "epoch": 13.57, "learning_rate": 8.434649249967506e-05, "loss": 0.1319, "step": 28330 }, { "epoch": 13.58, "learning_rate": 8.433605657754254e-05, "loss": 0.1236, "step": 28340 }, { "epoch": 13.58, "learning_rate": 8.432561782391687e-05, "loss": 0.141, "step": 28350 }, { "epoch": 13.59, "learning_rate": 8.431517623965888e-05, "loss": 0.1188, "step": 28360 }, { "epoch": 13.59, "learning_rate": 8.430473182562964e-05, "loss": 0.142, "step": 28370 }, { "epoch": 13.59, "learning_rate": 8.429428458269041e-05, "loss": 0.1314, "step": 28380 }, { "epoch": 13.6, "learning_rate": 8.428383451170273e-05, "loss": 0.1324, "step": 28390 }, { "epoch": 13.6, "learning_rate": 8.427338161352836e-05, "loss": 0.1217, "step": 28400 }, { "epoch": 13.61, "learning_rate": 8.426292588902928e-05, "loss": 0.1156, "step": 28410 }, { "epoch": 13.61, "learning_rate": 8.42524673390677e-05, "loss": 0.1223, "step": 28420 }, { "epoch": 13.62, "learning_rate": 8.424200596450611e-05, "loss": 0.1327, "step": 28430 }, { "epoch": 13.62, "learning_rate": 8.423154176620717e-05, "loss": 0.1242, "step": 28440 }, { "epoch": 13.63, "learning_rate": 8.422107474503382e-05, "loss": 0.1266, "step": 28450 }, { "epoch": 13.63, "learning_rate": 8.42106049018492e-05, "loss": 0.1527, "step": 28460 }, { "epoch": 13.64, "learning_rate": 8.42001322375167e-05, "loss": 0.1336, "step": 28470 }, { "epoch": 13.64, "learning_rate": 8.418965675289994e-05, "loss": 0.1225, "step": 28480 }, { "epoch": 13.64, "learning_rate": 8.417917844886278e-05, "loss": 0.1272, "step": 28490 }, { "epoch": 13.65, "learning_rate": 8.416869732626928e-05, "loss": 0.1217, "step": 28500 }, { "epoch": 13.65, "learning_rate": 8.415821338598377e-05, "loss": 0.119, "step": 28510 }, { "epoch": 13.66, "learning_rate": 8.414772662887081e-05, "loss": 0.1184, "step": 28520 }, { "epoch": 13.66, "learning_rate": 8.413723705579517e-05, "loss": 0.1226, "step": 28530 }, { "epoch": 13.67, "learning_rate": 8.412674466762186e-05, "loss": 0.1175, "step": 28540 }, { "epoch": 13.67, "learning_rate": 8.411624946521615e-05, "loss": 0.1192, "step": 28550 }, { "epoch": 13.68, "learning_rate": 8.410575144944348e-05, "loss": 0.1076, "step": 28560 }, { "epoch": 13.68, "learning_rate": 8.409525062116958e-05, "loss": 0.1296, "step": 28570 }, { "epoch": 13.69, "learning_rate": 8.40847469812604e-05, "loss": 0.1366, "step": 28580 }, { "epoch": 13.69, "learning_rate": 8.40742405305821e-05, "loss": 0.1204, "step": 28590 }, { "epoch": 13.7, "learning_rate": 8.40637312700011e-05, "loss": 0.1073, "step": 28600 }, { "epoch": 13.7, "learning_rate": 8.405321920038402e-05, "loss": 0.1381, "step": 28610 }, { "epoch": 13.7, "learning_rate": 8.404270432259773e-05, "loss": 0.1091, "step": 28620 }, { "epoch": 13.71, "learning_rate": 8.403218663750933e-05, "loss": 0.1142, "step": 28630 }, { "epoch": 13.71, "learning_rate": 8.402166614598619e-05, "loss": 0.1132, "step": 28640 }, { "epoch": 13.72, "learning_rate": 8.40111428488958e-05, "loss": 0.1137, "step": 28650 }, { "epoch": 13.72, "learning_rate": 8.400061674710603e-05, "loss": 0.1203, "step": 28660 }, { "epoch": 13.73, "learning_rate": 8.399008784148486e-05, "loss": 0.1031, "step": 28670 }, { "epoch": 13.73, "learning_rate": 8.397955613290054e-05, "loss": 0.1109, "step": 28680 }, { "epoch": 13.74, "learning_rate": 8.39690216222216e-05, "loss": 0.1237, "step": 28690 }, { "epoch": 13.74, "learning_rate": 8.395848431031672e-05, "loss": 0.1245, "step": 28700 }, { "epoch": 13.75, "learning_rate": 8.394794419805488e-05, "loss": 0.1234, "step": 28710 }, { "epoch": 13.75, "learning_rate": 8.393740128630526e-05, "loss": 0.1322, "step": 28720 }, { "epoch": 13.75, "learning_rate": 8.392685557593723e-05, "loss": 0.1348, "step": 28730 }, { "epoch": 13.76, "learning_rate": 8.391630706782047e-05, "loss": 0.1249, "step": 28740 }, { "epoch": 13.76, "learning_rate": 8.390575576282484e-05, "loss": 0.1352, "step": 28750 }, { "epoch": 13.77, "learning_rate": 8.389520166182047e-05, "loss": 0.1259, "step": 28760 }, { "epoch": 13.77, "learning_rate": 8.388464476567766e-05, "loss": 0.1425, "step": 28770 }, { "epoch": 13.78, "learning_rate": 8.3874085075267e-05, "loss": 0.1182, "step": 28780 }, { "epoch": 13.78, "learning_rate": 8.386352259145926e-05, "loss": 0.1243, "step": 28790 }, { "epoch": 13.79, "learning_rate": 8.38529573151255e-05, "loss": 0.1219, "step": 28800 }, { "epoch": 13.79, "learning_rate": 8.384238924713695e-05, "loss": 0.1349, "step": 28810 }, { "epoch": 13.8, "learning_rate": 8.383181838836511e-05, "loss": 0.1153, "step": 28820 }, { "epoch": 13.8, "learning_rate": 8.38212447396817e-05, "loss": 0.132, "step": 28830 }, { "epoch": 13.8, "learning_rate": 8.381066830195863e-05, "loss": 0.1343, "step": 28840 }, { "epoch": 13.81, "learning_rate": 8.380008907606813e-05, "loss": 0.1279, "step": 28850 }, { "epoch": 13.81, "learning_rate": 8.378950706288257e-05, "loss": 0.0946, "step": 28860 }, { "epoch": 13.82, "learning_rate": 8.37789222632746e-05, "loss": 0.1314, "step": 28870 }, { "epoch": 13.82, "learning_rate": 8.376833467811709e-05, "loss": 0.1088, "step": 28880 }, { "epoch": 13.83, "learning_rate": 8.375774430828314e-05, "loss": 0.1234, "step": 28890 }, { "epoch": 13.83, "learning_rate": 8.374715115464605e-05, "loss": 0.1175, "step": 28900 }, { "epoch": 13.84, "learning_rate": 8.37365552180794e-05, "loss": 0.1096, "step": 28910 }, { "epoch": 13.84, "learning_rate": 8.372595649945695e-05, "loss": 0.1134, "step": 28920 }, { "epoch": 13.85, "learning_rate": 8.371535499965276e-05, "loss": 0.1209, "step": 28930 }, { "epoch": 13.85, "learning_rate": 8.370475071954103e-05, "loss": 0.1181, "step": 28940 }, { "epoch": 13.86, "learning_rate": 8.369414365999624e-05, "loss": 0.114, "step": 28950 }, { "epoch": 13.86, "learning_rate": 8.368353382189311e-05, "loss": 0.1254, "step": 28960 }, { "epoch": 13.86, "learning_rate": 8.367292120610655e-05, "loss": 0.1228, "step": 28970 }, { "epoch": 13.87, "learning_rate": 8.366230581351173e-05, "loss": 0.1024, "step": 28980 }, { "epoch": 13.87, "learning_rate": 8.365168764498403e-05, "loss": 0.1188, "step": 28990 }, { "epoch": 13.88, "learning_rate": 8.36410667013991e-05, "loss": 0.122, "step": 29000 }, { "epoch": 13.88, "learning_rate": 8.363044298363275e-05, "loss": 0.114, "step": 29010 }, { "epoch": 13.89, "learning_rate": 8.361981649256108e-05, "loss": 0.1141, "step": 29020 }, { "epoch": 13.89, "learning_rate": 8.360918722906039e-05, "loss": 0.1256, "step": 29030 }, { "epoch": 13.9, "learning_rate": 8.359855519400719e-05, "loss": 0.1346, "step": 29040 }, { "epoch": 13.9, "learning_rate": 8.358792038827826e-05, "loss": 0.1215, "step": 29050 }, { "epoch": 13.91, "learning_rate": 8.357728281275058e-05, "loss": 0.1201, "step": 29060 }, { "epoch": 13.91, "learning_rate": 8.356664246830141e-05, "loss": 0.1108, "step": 29070 }, { "epoch": 13.91, "learning_rate": 8.355599935580813e-05, "loss": 0.1073, "step": 29080 }, { "epoch": 13.92, "learning_rate": 8.354535347614848e-05, "loss": 0.1133, "step": 29090 }, { "epoch": 13.92, "learning_rate": 8.353470483020031e-05, "loss": 0.1269, "step": 29100 }, { "epoch": 13.93, "learning_rate": 8.352405341884179e-05, "loss": 0.1269, "step": 29110 }, { "epoch": 13.93, "learning_rate": 8.351339924295125e-05, "loss": 0.1193, "step": 29120 }, { "epoch": 13.94, "learning_rate": 8.350274230340729e-05, "loss": 0.0971, "step": 29130 }, { "epoch": 13.94, "learning_rate": 8.349208260108874e-05, "loss": 0.1197, "step": 29140 }, { "epoch": 13.95, "learning_rate": 8.348142013687461e-05, "loss": 0.1183, "step": 29150 }, { "epoch": 13.95, "learning_rate": 8.34707549116442e-05, "loss": 0.1148, "step": 29160 }, { "epoch": 14.0, "learning_rate": 8.3460086926277e-05, "loss": 0.1167, "step": 29170 }, { "epoch": 14.01, "learning_rate": 8.344941618165275e-05, "loss": 0.1063, "step": 29180 }, { "epoch": 14.01, "learning_rate": 8.343874267865136e-05, "loss": 0.1243, "step": 29190 }, { "epoch": 14.02, "learning_rate": 8.342806641815304e-05, "loss": 0.1139, "step": 29200 }, { "epoch": 14.02, "learning_rate": 8.341738740103821e-05, "loss": 0.1289, "step": 29210 }, { "epoch": 14.03, "learning_rate": 8.34067056281875e-05, "loss": 0.1226, "step": 29220 }, { "epoch": 14.03, "learning_rate": 8.339602110048175e-05, "loss": 0.1291, "step": 29230 }, { "epoch": 14.04, "learning_rate": 8.338533381880208e-05, "loss": 0.1174, "step": 29240 }, { "epoch": 14.04, "learning_rate": 8.33746437840298e-05, "loss": 0.1062, "step": 29250 }, { "epoch": 14.04, "learning_rate": 8.336395099704643e-05, "loss": 0.1259, "step": 29260 }, { "epoch": 14.05, "learning_rate": 8.335325545873378e-05, "loss": 0.1115, "step": 29270 }, { "epoch": 14.05, "learning_rate": 8.334255716997382e-05, "loss": 0.1043, "step": 29280 }, { "epoch": 14.06, "learning_rate": 8.333185613164877e-05, "loss": 0.1379, "step": 29290 }, { "epoch": 14.06, "learning_rate": 8.332115234464113e-05, "loss": 0.1212, "step": 29300 }, { "epoch": 14.07, "learning_rate": 8.331044580983352e-05, "loss": 0.1192, "step": 29310 }, { "epoch": 14.07, "learning_rate": 8.329973652810888e-05, "loss": 0.1122, "step": 29320 }, { "epoch": 14.08, "learning_rate": 8.32890245003503e-05, "loss": 0.1061, "step": 29330 }, { "epoch": 14.08, "learning_rate": 8.32783097274412e-05, "loss": 0.1177, "step": 29340 }, { "epoch": 14.09, "learning_rate": 8.326759221026512e-05, "loss": 0.1112, "step": 29350 }, { "epoch": 14.09, "learning_rate": 8.325687194970591e-05, "loss": 0.1125, "step": 29360 }, { "epoch": 14.1, "learning_rate": 8.324614894664756e-05, "loss": 0.1247, "step": 29370 }, { "epoch": 14.1, "learning_rate": 8.323542320197434e-05, "loss": 0.1168, "step": 29380 }, { "epoch": 14.1, "learning_rate": 8.32246947165708e-05, "loss": 0.1225, "step": 29390 }, { "epoch": 14.11, "learning_rate": 8.321396349132156e-05, "loss": 0.1158, "step": 29400 }, { "epoch": 14.11, "learning_rate": 8.320322952711164e-05, "loss": 0.1215, "step": 29410 }, { "epoch": 14.12, "learning_rate": 8.319249282482617e-05, "loss": 0.1361, "step": 29420 }, { "epoch": 14.12, "learning_rate": 8.318175338535055e-05, "loss": 0.1209, "step": 29430 }, { "epoch": 14.13, "learning_rate": 8.317101120957039e-05, "loss": 0.1261, "step": 29440 }, { "epoch": 14.13, "learning_rate": 8.316026629837156e-05, "loss": 0.1139, "step": 29450 }, { "epoch": 14.14, "learning_rate": 8.31495186526401e-05, "loss": 0.1066, "step": 29460 }, { "epoch": 14.14, "learning_rate": 8.313876827326233e-05, "loss": 0.1267, "step": 29470 }, { "epoch": 14.15, "learning_rate": 8.312801516112474e-05, "loss": 0.1106, "step": 29480 }, { "epoch": 14.15, "learning_rate": 8.31172593171141e-05, "loss": 0.1104, "step": 29490 }, { "epoch": 14.15, "learning_rate": 8.31065007421174e-05, "loss": 0.1013, "step": 29500 }, { "epoch": 14.16, "learning_rate": 8.309573943702177e-05, "loss": 0.1035, "step": 29510 }, { "epoch": 14.16, "learning_rate": 8.308497540271471e-05, "loss": 0.1159, "step": 29520 }, { "epoch": 14.17, "learning_rate": 8.307420864008382e-05, "loss": 0.0961, "step": 29530 }, { "epoch": 14.17, "learning_rate": 8.306343915001697e-05, "loss": 0.1191, "step": 29540 }, { "epoch": 14.18, "learning_rate": 8.305266693340231e-05, "loss": 0.1178, "step": 29550 }, { "epoch": 14.18, "learning_rate": 8.304189199112808e-05, "loss": 0.1118, "step": 29560 }, { "epoch": 14.19, "learning_rate": 8.303111432408289e-05, "loss": 0.1193, "step": 29570 }, { "epoch": 14.19, "learning_rate": 8.302033393315548e-05, "loss": 0.1199, "step": 29580 }, { "epoch": 14.2, "learning_rate": 8.300955081923483e-05, "loss": 0.1067, "step": 29590 }, { "epoch": 14.2, "learning_rate": 8.299876498321022e-05, "loss": 0.1169, "step": 29600 }, { "epoch": 14.2, "learning_rate": 8.298797642597105e-05, "loss": 0.1212, "step": 29610 }, { "epoch": 14.21, "learning_rate": 8.297718514840698e-05, "loss": 0.1192, "step": 29620 }, { "epoch": 14.21, "learning_rate": 8.296639115140794e-05, "loss": 0.1221, "step": 29630 }, { "epoch": 14.22, "learning_rate": 8.295559443586402e-05, "loss": 0.1144, "step": 29640 }, { "epoch": 14.22, "learning_rate": 8.294479500266557e-05, "loss": 0.1191, "step": 29650 }, { "epoch": 14.23, "learning_rate": 8.293399285270317e-05, "loss": 0.1124, "step": 29660 }, { "epoch": 14.23, "learning_rate": 8.292318798686757e-05, "loss": 0.1236, "step": 29670 }, { "epoch": 14.24, "learning_rate": 8.291238040604984e-05, "loss": 0.1202, "step": 29680 }, { "epoch": 14.24, "learning_rate": 8.290157011114117e-05, "loss": 0.1252, "step": 29690 }, { "epoch": 14.25, "learning_rate": 8.289075710303305e-05, "loss": 0.131, "step": 29700 }, { "epoch": 14.25, "learning_rate": 8.287994138261714e-05, "loss": 0.1163, "step": 29710 }, { "epoch": 14.26, "learning_rate": 8.286912295078539e-05, "loss": 0.1076, "step": 29720 }, { "epoch": 14.26, "learning_rate": 8.285830180842989e-05, "loss": 0.1159, "step": 29730 }, { "epoch": 14.26, "learning_rate": 8.284747795644301e-05, "loss": 0.1144, "step": 29740 }, { "epoch": 14.27, "learning_rate": 8.283665139571732e-05, "loss": 0.1249, "step": 29750 }, { "epoch": 14.27, "learning_rate": 8.282582212714565e-05, "loss": 0.1145, "step": 29760 }, { "epoch": 14.28, "learning_rate": 8.281499015162102e-05, "loss": 0.1075, "step": 29770 }, { "epoch": 14.28, "learning_rate": 8.280415547003665e-05, "loss": 0.1048, "step": 29780 }, { "epoch": 14.29, "learning_rate": 8.279331808328605e-05, "loss": 0.1193, "step": 29790 }, { "epoch": 14.29, "learning_rate": 8.278247799226291e-05, "loss": 0.1325, "step": 29800 }, { "epoch": 14.3, "learning_rate": 8.27716351978611e-05, "loss": 0.1035, "step": 29810 }, { "epoch": 14.3, "learning_rate": 8.276078970097483e-05, "loss": 0.1097, "step": 29820 }, { "epoch": 14.31, "learning_rate": 8.274994150249843e-05, "loss": 0.1102, "step": 29830 }, { "epoch": 14.31, "learning_rate": 8.273909060332649e-05, "loss": 0.1152, "step": 29840 }, { "epoch": 14.31, "learning_rate": 8.272823700435382e-05, "loss": 0.1457, "step": 29850 }, { "epoch": 14.32, "learning_rate": 8.271738070647546e-05, "loss": 0.1152, "step": 29860 }, { "epoch": 14.32, "learning_rate": 8.270652171058664e-05, "loss": 0.1141, "step": 29870 }, { "epoch": 14.33, "learning_rate": 8.269566001758287e-05, "loss": 0.1295, "step": 29880 }, { "epoch": 14.33, "learning_rate": 8.268479562835984e-05, "loss": 0.1245, "step": 29890 }, { "epoch": 14.34, "learning_rate": 8.267392854381347e-05, "loss": 0.1215, "step": 29900 }, { "epoch": 14.34, "learning_rate": 8.266305876483988e-05, "loss": 0.1144, "step": 29910 }, { "epoch": 14.35, "learning_rate": 8.265218629233549e-05, "loss": 0.1086, "step": 29920 }, { "epoch": 14.35, "learning_rate": 8.264131112719685e-05, "loss": 0.12, "step": 29930 }, { "epoch": 14.36, "learning_rate": 8.263043327032078e-05, "loss": 0.1014, "step": 29940 }, { "epoch": 14.36, "learning_rate": 8.261955272260432e-05, "loss": 0.1283, "step": 29950 }, { "epoch": 14.36, "learning_rate": 8.260866948494472e-05, "loss": 0.1221, "step": 29960 }, { "epoch": 14.37, "learning_rate": 8.259778355823943e-05, "loss": 0.1142, "step": 29970 }, { "epoch": 14.37, "learning_rate": 8.258689494338622e-05, "loss": 0.0982, "step": 29980 }, { "epoch": 14.38, "learning_rate": 8.257600364128293e-05, "loss": 0.1153, "step": 29990 }, { "epoch": 14.38, "learning_rate": 8.256510965282776e-05, "loss": 0.1181, "step": 30000 }, { "epoch": 14.39, "learning_rate": 8.255421297891902e-05, "loss": 0.1172, "step": 30010 }, { "epoch": 14.39, "learning_rate": 8.254331362045534e-05, "loss": 0.1125, "step": 30020 }, { "epoch": 14.4, "learning_rate": 8.253241157833552e-05, "loss": 0.1148, "step": 30030 }, { "epoch": 14.4, "learning_rate": 8.252150685345857e-05, "loss": 0.1186, "step": 30040 }, { "epoch": 14.41, "learning_rate": 8.251059944672373e-05, "loss": 0.1092, "step": 30050 }, { "epoch": 14.41, "learning_rate": 8.249968935903051e-05, "loss": 0.1132, "step": 30060 }, { "epoch": 14.41, "learning_rate": 8.248877659127858e-05, "loss": 0.1281, "step": 30070 }, { "epoch": 14.42, "learning_rate": 8.247786114436783e-05, "loss": 0.1073, "step": 30080 }, { "epoch": 14.42, "learning_rate": 8.246694301919842e-05, "loss": 0.1041, "step": 30090 }, { "epoch": 14.43, "learning_rate": 8.24560222166707e-05, "loss": 0.121, "step": 30100 }, { "epoch": 14.43, "learning_rate": 8.244509873768524e-05, "loss": 0.1097, "step": 30110 }, { "epoch": 14.44, "learning_rate": 8.243417258314282e-05, "loss": 0.1023, "step": 30120 }, { "epoch": 14.44, "learning_rate": 8.242324375394449e-05, "loss": 0.122, "step": 30130 }, { "epoch": 14.45, "learning_rate": 8.241231225099146e-05, "loss": 0.1067, "step": 30140 }, { "epoch": 14.45, "learning_rate": 8.24013780751852e-05, "loss": 0.1124, "step": 30150 }, { "epoch": 14.46, "learning_rate": 8.239044122742739e-05, "loss": 0.118, "step": 30160 }, { "epoch": 14.46, "learning_rate": 8.23795017086199e-05, "loss": 0.1182, "step": 30170 }, { "epoch": 14.47, "learning_rate": 8.236855951966487e-05, "loss": 0.1196, "step": 30180 }, { "epoch": 14.47, "learning_rate": 8.235761466146465e-05, "loss": 0.1153, "step": 30190 }, { "epoch": 14.47, "learning_rate": 8.234666713492178e-05, "loss": 0.1184, "step": 30200 }, { "epoch": 14.48, "learning_rate": 8.233571694093905e-05, "loss": 0.1147, "step": 30210 }, { "epoch": 14.48, "learning_rate": 8.232476408041943e-05, "loss": 0.1095, "step": 30220 }, { "epoch": 14.49, "learning_rate": 8.231380855426617e-05, "loss": 0.1133, "step": 30230 }, { "epoch": 14.49, "learning_rate": 8.23028503633827e-05, "loss": 0.1121, "step": 30240 }, { "epoch": 14.5, "learning_rate": 8.229188950867268e-05, "loss": 0.1145, "step": 30250 }, { "epoch": 14.5, "learning_rate": 8.228092599103995e-05, "loss": 0.1302, "step": 30260 }, { "epoch": 14.51, "learning_rate": 8.226995981138867e-05, "loss": 0.1063, "step": 30270 }, { "epoch": 14.51, "learning_rate": 8.225899097062313e-05, "loss": 0.1071, "step": 30280 }, { "epoch": 14.52, "learning_rate": 8.224801946964785e-05, "loss": 0.1143, "step": 30290 }, { "epoch": 14.52, "learning_rate": 8.22370453093676e-05, "loss": 0.1264, "step": 30300 }, { "epoch": 14.52, "learning_rate": 8.222606849068735e-05, "loss": 0.1103, "step": 30310 }, { "epoch": 14.53, "learning_rate": 8.221508901451228e-05, "loss": 0.1157, "step": 30320 }, { "epoch": 14.53, "learning_rate": 8.220410688174784e-05, "loss": 0.1294, "step": 30330 }, { "epoch": 14.54, "learning_rate": 8.219312209329964e-05, "loss": 0.1104, "step": 30340 }, { "epoch": 14.54, "learning_rate": 8.218213465007351e-05, "loss": 0.1258, "step": 30350 }, { "epoch": 14.55, "learning_rate": 8.217114455297557e-05, "loss": 0.1195, "step": 30360 }, { "epoch": 14.55, "learning_rate": 8.216015180291205e-05, "loss": 0.1243, "step": 30370 }, { "epoch": 14.56, "learning_rate": 8.214915640078951e-05, "loss": 0.1313, "step": 30380 }, { "epoch": 14.56, "learning_rate": 8.213815834751464e-05, "loss": 0.1191, "step": 30390 }, { "epoch": 14.57, "learning_rate": 8.212715764399441e-05, "loss": 0.1087, "step": 30400 }, { "epoch": 14.57, "learning_rate": 8.211615429113598e-05, "loss": 0.1273, "step": 30410 }, { "epoch": 14.57, "learning_rate": 8.210514828984668e-05, "loss": 0.1119, "step": 30420 }, { "epoch": 14.58, "learning_rate": 8.209413964103421e-05, "loss": 0.1241, "step": 30430 }, { "epoch": 14.58, "learning_rate": 8.20831283456063e-05, "loss": 0.1166, "step": 30440 }, { "epoch": 14.59, "learning_rate": 8.207211440447104e-05, "loss": 0.1105, "step": 30450 }, { "epoch": 14.59, "learning_rate": 8.206109781853664e-05, "loss": 0.1228, "step": 30460 }, { "epoch": 14.6, "learning_rate": 8.205007858871162e-05, "loss": 0.1226, "step": 30470 }, { "epoch": 14.6, "learning_rate": 8.203905671590464e-05, "loss": 0.1163, "step": 30480 }, { "epoch": 14.61, "learning_rate": 8.202803220102462e-05, "loss": 0.1163, "step": 30490 }, { "epoch": 14.61, "learning_rate": 8.20170050449807e-05, "loss": 0.1085, "step": 30500 }, { "epoch": 14.62, "learning_rate": 8.20059752486822e-05, "loss": 0.1281, "step": 30510 }, { "epoch": 14.62, "learning_rate": 8.19949428130387e-05, "loss": 0.1231, "step": 30520 }, { "epoch": 14.63, "learning_rate": 8.198390773895996e-05, "loss": 0.1145, "step": 30530 }, { "epoch": 14.63, "learning_rate": 8.197287002735601e-05, "loss": 0.1093, "step": 30540 }, { "epoch": 14.63, "learning_rate": 8.196182967913702e-05, "loss": 0.116, "step": 30550 }, { "epoch": 14.64, "learning_rate": 8.195078669521348e-05, "loss": 0.1224, "step": 30560 }, { "epoch": 14.64, "learning_rate": 8.193974107649599e-05, "loss": 0.1134, "step": 30570 }, { "epoch": 14.65, "learning_rate": 8.192869282389544e-05, "loss": 0.1081, "step": 30580 }, { "epoch": 14.65, "learning_rate": 8.191764193832291e-05, "loss": 0.1248, "step": 30590 }, { "epoch": 14.66, "learning_rate": 8.190658842068972e-05, "loss": 0.1031, "step": 30600 }, { "epoch": 14.66, "learning_rate": 8.189553227190736e-05, "loss": 0.1115, "step": 30610 }, { "epoch": 14.67, "learning_rate": 8.188447349288758e-05, "loss": 0.1303, "step": 30620 }, { "epoch": 14.67, "learning_rate": 8.187341208454233e-05, "loss": 0.1029, "step": 30630 }, { "epoch": 14.68, "learning_rate": 8.186234804778377e-05, "loss": 0.1018, "step": 30640 }, { "epoch": 14.68, "learning_rate": 8.185128138352431e-05, "loss": 0.1137, "step": 30650 }, { "epoch": 14.68, "learning_rate": 8.184021209267654e-05, "loss": 0.1095, "step": 30660 }, { "epoch": 14.69, "learning_rate": 8.182914017615327e-05, "loss": 0.1206, "step": 30670 }, { "epoch": 14.69, "learning_rate": 8.181806563486756e-05, "loss": 0.1625, "step": 30680 }, { "epoch": 14.7, "learning_rate": 8.180698846973263e-05, "loss": 0.1183, "step": 30690 }, { "epoch": 14.7, "learning_rate": 8.179590868166196e-05, "loss": 0.1166, "step": 30700 }, { "epoch": 14.71, "learning_rate": 8.178482627156926e-05, "loss": 0.1236, "step": 30710 }, { "epoch": 14.71, "learning_rate": 8.17737412403684e-05, "loss": 0.1191, "step": 30720 }, { "epoch": 14.72, "learning_rate": 8.176265358897353e-05, "loss": 0.1077, "step": 30730 }, { "epoch": 14.72, "learning_rate": 8.175156331829895e-05, "loss": 0.1085, "step": 30740 }, { "epoch": 14.73, "learning_rate": 8.174047042925921e-05, "loss": 0.1111, "step": 30750 }, { "epoch": 14.73, "learning_rate": 8.17293749227691e-05, "loss": 0.1174, "step": 30760 }, { "epoch": 14.73, "learning_rate": 8.17182767997436e-05, "loss": 0.1125, "step": 30770 }, { "epoch": 14.74, "learning_rate": 8.170717606109788e-05, "loss": 0.1184, "step": 30780 }, { "epoch": 14.74, "learning_rate": 8.169607270774737e-05, "loss": 0.1265, "step": 30790 }, { "epoch": 14.75, "learning_rate": 8.16849667406077e-05, "loss": 0.1189, "step": 30800 }, { "epoch": 14.75, "learning_rate": 8.167385816059471e-05, "loss": 0.1199, "step": 30810 }, { "epoch": 14.76, "learning_rate": 8.166274696862448e-05, "loss": 0.1127, "step": 30820 }, { "epoch": 14.76, "learning_rate": 8.165163316561326e-05, "loss": 0.132, "step": 30830 }, { "epoch": 14.77, "learning_rate": 8.164051675247753e-05, "loss": 0.13, "step": 30840 }, { "epoch": 14.77, "learning_rate": 8.162939773013402e-05, "loss": 0.1134, "step": 30850 }, { "epoch": 14.78, "learning_rate": 8.161827609949966e-05, "loss": 0.1247, "step": 30860 }, { "epoch": 14.78, "learning_rate": 8.160715186149154e-05, "loss": 0.1101, "step": 30870 }, { "epoch": 14.79, "learning_rate": 8.159602501702706e-05, "loss": 0.1215, "step": 30880 }, { "epoch": 14.79, "learning_rate": 8.158489556702375e-05, "loss": 0.119, "step": 30890 }, { "epoch": 14.79, "learning_rate": 8.157376351239941e-05, "loss": 0.141, "step": 30900 }, { "epoch": 14.8, "learning_rate": 8.156262885407202e-05, "loss": 0.1336, "step": 30910 }, { "epoch": 14.8, "learning_rate": 8.15514915929598e-05, "loss": 0.1158, "step": 30920 }, { "epoch": 14.81, "learning_rate": 8.15403517299812e-05, "loss": 0.1137, "step": 30930 }, { "epoch": 14.81, "learning_rate": 8.15292092660548e-05, "loss": 0.1169, "step": 30940 }, { "epoch": 14.82, "learning_rate": 8.15180642020995e-05, "loss": 0.1266, "step": 30950 }, { "epoch": 14.82, "learning_rate": 8.150691653903437e-05, "loss": 0.1239, "step": 30960 }, { "epoch": 14.83, "learning_rate": 8.149576627777866e-05, "loss": 0.1133, "step": 30970 }, { "epoch": 14.83, "learning_rate": 8.148461341925188e-05, "loss": 0.1123, "step": 30980 }, { "epoch": 14.84, "learning_rate": 8.147345796437378e-05, "loss": 0.1288, "step": 30990 }, { "epoch": 14.84, "learning_rate": 8.146229991406422e-05, "loss": 0.1209, "step": 31000 }, { "epoch": 14.84, "learning_rate": 8.145113926924337e-05, "loss": 0.1306, "step": 31010 }, { "epoch": 14.85, "learning_rate": 8.143997603083159e-05, "loss": 0.1082, "step": 31020 }, { "epoch": 14.85, "learning_rate": 8.142881019974943e-05, "loss": 0.1073, "step": 31030 }, { "epoch": 14.86, "learning_rate": 8.141764177691768e-05, "loss": 0.1415, "step": 31040 }, { "epoch": 14.86, "learning_rate": 8.140647076325734e-05, "loss": 0.1245, "step": 31050 }, { "epoch": 14.87, "learning_rate": 8.139529715968961e-05, "loss": 0.1157, "step": 31060 }, { "epoch": 14.87, "learning_rate": 8.13841209671359e-05, "loss": 0.116, "step": 31070 }, { "epoch": 14.88, "learning_rate": 8.137294218651787e-05, "loss": 0.1056, "step": 31080 }, { "epoch": 14.88, "learning_rate": 8.136176081875736e-05, "loss": 0.1244, "step": 31090 }, { "epoch": 14.89, "learning_rate": 8.135057686477643e-05, "loss": 0.1089, "step": 31100 }, { "epoch": 14.89, "learning_rate": 8.133939032549736e-05, "loss": 0.1155, "step": 31110 }, { "epoch": 14.89, "learning_rate": 8.13282012018426e-05, "loss": 0.1194, "step": 31120 }, { "epoch": 14.9, "learning_rate": 8.131700949473491e-05, "loss": 0.1184, "step": 31130 }, { "epoch": 14.9, "learning_rate": 8.130581520509717e-05, "loss": 0.1314, "step": 31140 }, { "epoch": 14.91, "learning_rate": 8.129461833385252e-05, "loss": 0.1432, "step": 31150 }, { "epoch": 14.91, "learning_rate": 8.128341888192428e-05, "loss": 0.1214, "step": 31160 }, { "epoch": 14.92, "learning_rate": 8.127221685023603e-05, "loss": 0.1043, "step": 31170 }, { "epoch": 14.92, "learning_rate": 8.12610122397115e-05, "loss": 0.1203, "step": 31180 }, { "epoch": 14.93, "learning_rate": 8.124980505127473e-05, "loss": 0.1167, "step": 31190 }, { "epoch": 14.93, "learning_rate": 8.123859528584985e-05, "loss": 0.1272, "step": 31200 }, { "epoch": 14.94, "learning_rate": 8.12273829443613e-05, "loss": 0.1248, "step": 31210 }, { "epoch": 14.94, "learning_rate": 8.121616802773367e-05, "loss": 0.1035, "step": 31220 }, { "epoch": 14.95, "learning_rate": 8.120495053689181e-05, "loss": 0.1358, "step": 31230 }, { "epoch": 14.95, "learning_rate": 8.119373047276075e-05, "loss": 0.1122, "step": 31240 }, { "epoch": 15.0, "learning_rate": 8.118250783626575e-05, "loss": 0.1218, "step": 31250 }, { "epoch": 15.01, "learning_rate": 8.117128262833226e-05, "loss": 0.1135, "step": 31260 }, { "epoch": 15.01, "learning_rate": 8.116005484988597e-05, "loss": 0.1207, "step": 31270 }, { "epoch": 15.02, "learning_rate": 8.114882450185277e-05, "loss": 0.1057, "step": 31280 }, { "epoch": 15.02, "learning_rate": 8.113759158515877e-05, "loss": 0.108, "step": 31290 }, { "epoch": 15.03, "learning_rate": 8.112635610073025e-05, "loss": 0.1092, "step": 31300 }, { "epoch": 15.03, "learning_rate": 8.111511804949375e-05, "loss": 0.1087, "step": 31310 }, { "epoch": 15.03, "learning_rate": 8.110387743237602e-05, "loss": 0.1247, "step": 31320 }, { "epoch": 15.04, "learning_rate": 8.1092634250304e-05, "loss": 0.1219, "step": 31330 }, { "epoch": 15.04, "learning_rate": 8.108138850420483e-05, "loss": 0.0952, "step": 31340 }, { "epoch": 15.05, "learning_rate": 8.107014019500592e-05, "loss": 0.1062, "step": 31350 }, { "epoch": 15.05, "learning_rate": 8.105888932363483e-05, "loss": 0.1138, "step": 31360 }, { "epoch": 15.06, "learning_rate": 8.104763589101935e-05, "loss": 0.1176, "step": 31370 }, { "epoch": 15.06, "learning_rate": 8.103637989808748e-05, "loss": 0.1137, "step": 31380 }, { "epoch": 15.07, "learning_rate": 8.102512134576745e-05, "loss": 0.1109, "step": 31390 }, { "epoch": 15.07, "learning_rate": 8.101386023498769e-05, "loss": 0.1147, "step": 31400 }, { "epoch": 15.08, "learning_rate": 8.100259656667682e-05, "loss": 0.1163, "step": 31410 }, { "epoch": 15.08, "learning_rate": 8.099133034176368e-05, "loss": 0.1122, "step": 31420 }, { "epoch": 15.08, "learning_rate": 8.098006156117737e-05, "loss": 0.0969, "step": 31430 }, { "epoch": 15.09, "learning_rate": 8.096879022584712e-05, "loss": 0.1027, "step": 31440 }, { "epoch": 15.09, "learning_rate": 8.095751633670243e-05, "loss": 0.1031, "step": 31450 }, { "epoch": 15.1, "learning_rate": 8.094623989467301e-05, "loss": 0.101, "step": 31460 }, { "epoch": 15.1, "learning_rate": 8.093496090068872e-05, "loss": 0.1103, "step": 31470 }, { "epoch": 15.11, "learning_rate": 8.092367935567969e-05, "loss": 0.1121, "step": 31480 }, { "epoch": 15.11, "learning_rate": 8.091239526057626e-05, "loss": 0.1092, "step": 31490 }, { "epoch": 15.12, "learning_rate": 8.090110861630894e-05, "loss": 0.1233, "step": 31500 }, { "epoch": 15.12, "learning_rate": 8.088981942380849e-05, "loss": 0.1013, "step": 31510 }, { "epoch": 15.13, "learning_rate": 8.087852768400586e-05, "loss": 0.1141, "step": 31520 }, { "epoch": 15.13, "learning_rate": 8.08672333978322e-05, "loss": 0.1079, "step": 31530 }, { "epoch": 15.13, "learning_rate": 8.085593656621889e-05, "loss": 0.1022, "step": 31540 }, { "epoch": 15.14, "learning_rate": 8.084463719009753e-05, "loss": 0.1014, "step": 31550 }, { "epoch": 15.14, "learning_rate": 8.083333527039989e-05, "loss": 0.1262, "step": 31560 }, { "epoch": 15.15, "learning_rate": 8.082203080805799e-05, "loss": 0.1215, "step": 31570 }, { "epoch": 15.15, "learning_rate": 8.081072380400403e-05, "loss": 0.122, "step": 31580 }, { "epoch": 15.16, "learning_rate": 8.079941425917043e-05, "loss": 0.1173, "step": 31590 }, { "epoch": 15.16, "learning_rate": 8.078810217448986e-05, "loss": 0.1225, "step": 31600 }, { "epoch": 15.17, "learning_rate": 8.07767875508951e-05, "loss": 0.1157, "step": 31610 }, { "epoch": 15.17, "learning_rate": 8.076547038931924e-05, "loss": 0.1145, "step": 31620 }, { "epoch": 15.18, "learning_rate": 8.075415069069554e-05, "loss": 0.11, "step": 31630 }, { "epoch": 15.18, "learning_rate": 8.074282845595745e-05, "loss": 0.1053, "step": 31640 }, { "epoch": 15.19, "learning_rate": 8.073150368603866e-05, "loss": 0.0982, "step": 31650 }, { "epoch": 15.19, "learning_rate": 8.072017638187306e-05, "loss": 0.1058, "step": 31660 }, { "epoch": 15.19, "learning_rate": 8.070884654439475e-05, "loss": 0.1076, "step": 31670 }, { "epoch": 15.2, "learning_rate": 8.069751417453802e-05, "loss": 0.1003, "step": 31680 }, { "epoch": 15.2, "learning_rate": 8.068617927323739e-05, "loss": 0.0947, "step": 31690 }, { "epoch": 15.21, "learning_rate": 8.067484184142759e-05, "loss": 0.1165, "step": 31700 }, { "epoch": 15.21, "learning_rate": 8.066350188004354e-05, "loss": 0.1163, "step": 31710 }, { "epoch": 15.22, "learning_rate": 8.06521593900204e-05, "loss": 0.1125, "step": 31720 }, { "epoch": 15.22, "learning_rate": 8.064081437229349e-05, "loss": 0.1263, "step": 31730 }, { "epoch": 15.23, "learning_rate": 8.062946682779839e-05, "loss": 0.1198, "step": 31740 }, { "epoch": 15.23, "learning_rate": 8.061811675747084e-05, "loss": 0.1065, "step": 31750 }, { "epoch": 15.24, "learning_rate": 8.060676416224686e-05, "loss": 0.1148, "step": 31760 }, { "epoch": 15.24, "learning_rate": 8.059540904306257e-05, "loss": 0.1224, "step": 31770 }, { "epoch": 15.24, "learning_rate": 8.058405140085442e-05, "loss": 0.1178, "step": 31780 }, { "epoch": 15.25, "learning_rate": 8.057269123655897e-05, "loss": 0.1122, "step": 31790 }, { "epoch": 15.25, "learning_rate": 8.056132855111305e-05, "loss": 0.114, "step": 31800 }, { "epoch": 15.26, "learning_rate": 8.054996334545366e-05, "loss": 0.1185, "step": 31810 }, { "epoch": 15.26, "learning_rate": 8.0538595620518e-05, "loss": 0.1159, "step": 31820 }, { "epoch": 15.27, "learning_rate": 8.052722537724354e-05, "loss": 0.1135, "step": 31830 }, { "epoch": 15.27, "learning_rate": 8.05158526165679e-05, "loss": 0.1085, "step": 31840 }, { "epoch": 15.28, "learning_rate": 8.050447733942892e-05, "loss": 0.1125, "step": 31850 }, { "epoch": 15.28, "learning_rate": 8.049309954676465e-05, "loss": 0.1138, "step": 31860 }, { "epoch": 15.29, "learning_rate": 8.048171923951337e-05, "loss": 0.1209, "step": 31870 }, { "epoch": 15.29, "learning_rate": 8.047033641861353e-05, "loss": 0.1162, "step": 31880 }, { "epoch": 15.29, "learning_rate": 8.045895108500382e-05, "loss": 0.1184, "step": 31890 }, { "epoch": 15.3, "learning_rate": 8.04475632396231e-05, "loss": 0.1164, "step": 31900 }, { "epoch": 15.3, "learning_rate": 8.043617288341047e-05, "loss": 0.0926, "step": 31910 }, { "epoch": 15.31, "learning_rate": 8.042478001730522e-05, "loss": 0.117, "step": 31920 }, { "epoch": 15.31, "learning_rate": 8.041338464224686e-05, "loss": 0.1079, "step": 31930 }, { "epoch": 15.32, "learning_rate": 8.04019867591751e-05, "loss": 0.1176, "step": 31940 }, { "epoch": 15.32, "learning_rate": 8.039058636902985e-05, "loss": 0.1005, "step": 31950 }, { "epoch": 15.33, "learning_rate": 8.037918347275124e-05, "loss": 0.1053, "step": 31960 }, { "epoch": 15.33, "learning_rate": 8.03677780712796e-05, "loss": 0.113, "step": 31970 }, { "epoch": 15.34, "learning_rate": 8.035637016555546e-05, "loss": 0.1047, "step": 31980 }, { "epoch": 15.34, "learning_rate": 8.034495975651957e-05, "loss": 0.1039, "step": 31990 }, { "epoch": 15.35, "learning_rate": 8.033354684511287e-05, "loss": 0.1094, "step": 32000 }, { "epoch": 15.35, "learning_rate": 8.032213143227654e-05, "loss": 0.1106, "step": 32010 }, { "epoch": 15.35, "learning_rate": 8.031071351895191e-05, "loss": 0.1137, "step": 32020 }, { "epoch": 15.36, "learning_rate": 8.029929310608057e-05, "loss": 0.1088, "step": 32030 }, { "epoch": 15.36, "learning_rate": 8.028787019460427e-05, "loss": 0.1239, "step": 32040 }, { "epoch": 15.37, "learning_rate": 8.027644478546503e-05, "loss": 0.1095, "step": 32050 }, { "epoch": 15.37, "learning_rate": 8.026501687960499e-05, "loss": 0.1284, "step": 32060 }, { "epoch": 15.38, "learning_rate": 8.025358647796659e-05, "loss": 0.1213, "step": 32070 }, { "epoch": 15.38, "learning_rate": 8.024215358149238e-05, "loss": 0.0956, "step": 32080 }, { "epoch": 15.39, "learning_rate": 8.023071819112522e-05, "loss": 0.1308, "step": 32090 }, { "epoch": 15.39, "learning_rate": 8.021928030780806e-05, "loss": 0.124, "step": 32100 }, { "epoch": 15.4, "learning_rate": 8.020783993248415e-05, "loss": 0.1003, "step": 32110 }, { "epoch": 15.4, "learning_rate": 8.019639706609692e-05, "loss": 0.1131, "step": 32120 }, { "epoch": 15.4, "learning_rate": 8.018495170958996e-05, "loss": 0.1093, "step": 32130 }, { "epoch": 15.41, "learning_rate": 8.017350386390713e-05, "loss": 0.1092, "step": 32140 }, { "epoch": 15.41, "learning_rate": 8.016205352999246e-05, "loss": 0.1006, "step": 32150 }, { "epoch": 15.42, "learning_rate": 8.015060070879019e-05, "loss": 0.1171, "step": 32160 }, { "epoch": 15.42, "learning_rate": 8.013914540124477e-05, "loss": 0.1064, "step": 32170 }, { "epoch": 15.43, "learning_rate": 8.012768760830084e-05, "loss": 0.1034, "step": 32180 }, { "epoch": 15.43, "learning_rate": 8.011622733090326e-05, "loss": 0.1084, "step": 32190 }, { "epoch": 15.44, "learning_rate": 8.010476456999712e-05, "loss": 0.1161, "step": 32200 }, { "epoch": 15.44, "learning_rate": 8.009329932652766e-05, "loss": 0.1096, "step": 32210 }, { "epoch": 15.45, "learning_rate": 8.008183160144033e-05, "loss": 0.1115, "step": 32220 }, { "epoch": 15.45, "learning_rate": 8.007036139568085e-05, "loss": 0.1183, "step": 32230 }, { "epoch": 15.45, "learning_rate": 8.005888871019507e-05, "loss": 0.1082, "step": 32240 }, { "epoch": 15.46, "learning_rate": 8.004741354592911e-05, "loss": 0.1277, "step": 32250 }, { "epoch": 15.46, "learning_rate": 8.003593590382922e-05, "loss": 0.1274, "step": 32260 }, { "epoch": 15.47, "learning_rate": 8.002445578484192e-05, "loss": 0.1075, "step": 32270 }, { "epoch": 15.47, "learning_rate": 8.00129731899139e-05, "loss": 0.1097, "step": 32280 }, { "epoch": 15.48, "learning_rate": 8.000148811999204e-05, "loss": 0.1204, "step": 32290 }, { "epoch": 15.48, "learning_rate": 7.99900005760235e-05, "loss": 0.1052, "step": 32300 }, { "epoch": 15.49, "learning_rate": 7.997851055895553e-05, "loss": 0.1186, "step": 32310 }, { "epoch": 15.49, "learning_rate": 7.996701806973568e-05, "loss": 0.1141, "step": 32320 }, { "epoch": 15.5, "learning_rate": 7.995552310931168e-05, "loss": 0.1136, "step": 32330 }, { "epoch": 15.5, "learning_rate": 7.99440256786314e-05, "loss": 0.1251, "step": 32340 }, { "epoch": 15.51, "learning_rate": 7.9932525778643e-05, "loss": 0.1128, "step": 32350 }, { "epoch": 15.51, "learning_rate": 7.992102341029483e-05, "loss": 0.1252, "step": 32360 }, { "epoch": 15.51, "learning_rate": 7.990951857453539e-05, "loss": 0.1068, "step": 32370 }, { "epoch": 15.52, "learning_rate": 7.989801127231342e-05, "loss": 0.104, "step": 32380 }, { "epoch": 15.52, "learning_rate": 7.988650150457787e-05, "loss": 0.1215, "step": 32390 }, { "epoch": 15.53, "learning_rate": 7.987498927227788e-05, "loss": 0.1111, "step": 32400 }, { "epoch": 15.53, "learning_rate": 7.986347457636279e-05, "loss": 0.128, "step": 32410 }, { "epoch": 15.54, "learning_rate": 7.985195741778216e-05, "loss": 0.1223, "step": 32420 }, { "epoch": 15.54, "learning_rate": 7.984043779748573e-05, "loss": 0.111, "step": 32430 }, { "epoch": 15.55, "learning_rate": 7.982891571642347e-05, "loss": 0.1067, "step": 32440 }, { "epoch": 15.55, "learning_rate": 7.981739117554551e-05, "loss": 0.1152, "step": 32450 }, { "epoch": 15.56, "learning_rate": 7.980586417580227e-05, "loss": 0.1199, "step": 32460 }, { "epoch": 15.56, "learning_rate": 7.979433471814426e-05, "loss": 0.1096, "step": 32470 }, { "epoch": 15.56, "learning_rate": 7.978280280352226e-05, "loss": 0.1081, "step": 32480 }, { "epoch": 15.57, "learning_rate": 7.977126843288724e-05, "loss": 0.1121, "step": 32490 }, { "epoch": 15.57, "learning_rate": 7.975973160719036e-05, "loss": 0.0948, "step": 32500 }, { "epoch": 15.58, "learning_rate": 7.974819232738305e-05, "loss": 0.1046, "step": 32510 }, { "epoch": 15.58, "learning_rate": 7.973665059441681e-05, "loss": 0.1414, "step": 32520 }, { "epoch": 15.59, "learning_rate": 7.972510640924346e-05, "loss": 0.1156, "step": 32530 }, { "epoch": 15.59, "learning_rate": 7.971355977281499e-05, "loss": 0.1147, "step": 32540 }, { "epoch": 15.6, "learning_rate": 7.970201068608355e-05, "loss": 0.1201, "step": 32550 }, { "epoch": 15.6, "learning_rate": 7.969045915000155e-05, "loss": 0.111, "step": 32560 }, { "epoch": 15.61, "learning_rate": 7.967890516552158e-05, "loss": 0.1173, "step": 32570 }, { "epoch": 15.61, "learning_rate": 7.96673487335964e-05, "loss": 0.1127, "step": 32580 }, { "epoch": 15.61, "learning_rate": 7.965578985517905e-05, "loss": 0.117, "step": 32590 }, { "epoch": 15.62, "learning_rate": 7.964422853122268e-05, "loss": 0.109, "step": 32600 }, { "epoch": 15.62, "learning_rate": 7.963266476268069e-05, "loss": 0.1115, "step": 32610 }, { "epoch": 15.63, "learning_rate": 7.96210985505067e-05, "loss": 0.1044, "step": 32620 }, { "epoch": 15.63, "learning_rate": 7.96095298956545e-05, "loss": 0.1084, "step": 32630 }, { "epoch": 15.64, "learning_rate": 7.959795879907807e-05, "loss": 0.1186, "step": 32640 }, { "epoch": 15.64, "learning_rate": 7.958638526173164e-05, "loss": 0.1178, "step": 32650 }, { "epoch": 15.65, "learning_rate": 7.957480928456957e-05, "loss": 0.1059, "step": 32660 }, { "epoch": 15.65, "learning_rate": 7.956323086854649e-05, "loss": 0.1233, "step": 32670 }, { "epoch": 15.66, "learning_rate": 7.955165001461722e-05, "loss": 0.1158, "step": 32680 }, { "epoch": 15.66, "learning_rate": 7.954006672373674e-05, "loss": 0.1169, "step": 32690 }, { "epoch": 15.66, "learning_rate": 7.952848099686025e-05, "loss": 0.1178, "step": 32700 }, { "epoch": 15.67, "learning_rate": 7.951689283494319e-05, "loss": 0.1222, "step": 32710 }, { "epoch": 15.67, "learning_rate": 7.950530223894113e-05, "loss": 0.118, "step": 32720 }, { "epoch": 15.68, "learning_rate": 7.949370920980989e-05, "loss": 0.1218, "step": 32730 }, { "epoch": 15.68, "learning_rate": 7.948211374850551e-05, "loss": 0.1167, "step": 32740 }, { "epoch": 15.69, "learning_rate": 7.947051585598415e-05, "loss": 0.1168, "step": 32750 }, { "epoch": 15.69, "learning_rate": 7.945891553320227e-05, "loss": 0.1072, "step": 32760 }, { "epoch": 15.7, "learning_rate": 7.944731278111644e-05, "loss": 0.1106, "step": 32770 }, { "epoch": 15.7, "learning_rate": 7.943570760068348e-05, "loss": 0.12, "step": 32780 }, { "epoch": 15.71, "learning_rate": 7.942409999286042e-05, "loss": 0.0973, "step": 32790 }, { "epoch": 15.71, "learning_rate": 7.941248995860445e-05, "loss": 0.1132, "step": 32800 }, { "epoch": 15.72, "learning_rate": 7.9400877498873e-05, "loss": 0.1089, "step": 32810 }, { "epoch": 15.72, "learning_rate": 7.938926261462366e-05, "loss": 0.1007, "step": 32820 }, { "epoch": 15.72, "learning_rate": 7.937764530681426e-05, "loss": 0.1066, "step": 32830 }, { "epoch": 15.73, "learning_rate": 7.93660255764028e-05, "loss": 0.1068, "step": 32840 }, { "epoch": 15.73, "learning_rate": 7.93544034243475e-05, "loss": 0.1081, "step": 32850 }, { "epoch": 15.74, "learning_rate": 7.934277885160676e-05, "loss": 0.1251, "step": 32860 }, { "epoch": 15.74, "learning_rate": 7.933115185913921e-05, "loss": 0.1265, "step": 32870 }, { "epoch": 15.75, "learning_rate": 7.931952244790361e-05, "loss": 0.1193, "step": 32880 }, { "epoch": 15.75, "learning_rate": 7.930789061885905e-05, "loss": 0.107, "step": 32890 }, { "epoch": 15.76, "learning_rate": 7.929625637296467e-05, "loss": 0.1166, "step": 32900 }, { "epoch": 15.76, "learning_rate": 7.928461971117991e-05, "loss": 0.1227, "step": 32910 }, { "epoch": 15.77, "learning_rate": 7.927298063446436e-05, "loss": 0.1124, "step": 32920 }, { "epoch": 15.77, "learning_rate": 7.926133914377785e-05, "loss": 0.1268, "step": 32930 }, { "epoch": 15.77, "learning_rate": 7.924969524008035e-05, "loss": 0.121, "step": 32940 }, { "epoch": 15.78, "learning_rate": 7.92380489243321e-05, "loss": 0.0956, "step": 32950 }, { "epoch": 15.78, "learning_rate": 7.922640019749349e-05, "loss": 0.1037, "step": 32960 }, { "epoch": 15.79, "learning_rate": 7.921474906052512e-05, "loss": 0.1162, "step": 32970 }, { "epoch": 15.79, "learning_rate": 7.920309551438779e-05, "loss": 0.122, "step": 32980 }, { "epoch": 15.8, "learning_rate": 7.919143956004249e-05, "loss": 0.1355, "step": 32990 }, { "epoch": 15.8, "learning_rate": 7.917978119845045e-05, "loss": 0.1186, "step": 33000 }, { "epoch": 15.81, "learning_rate": 7.916812043057302e-05, "loss": 0.1029, "step": 33010 }, { "epoch": 15.81, "learning_rate": 7.915645725737184e-05, "loss": 0.1174, "step": 33020 }, { "epoch": 15.82, "learning_rate": 7.914479167980869e-05, "loss": 0.1097, "step": 33030 }, { "epoch": 15.82, "learning_rate": 7.913312369884555e-05, "loss": 0.1098, "step": 33040 }, { "epoch": 15.82, "learning_rate": 7.912145331544462e-05, "loss": 0.1323, "step": 33050 }, { "epoch": 15.83, "learning_rate": 7.910978053056828e-05, "loss": 0.1282, "step": 33060 }, { "epoch": 15.83, "learning_rate": 7.909810534517912e-05, "loss": 0.1255, "step": 33070 }, { "epoch": 15.84, "learning_rate": 7.908642776023992e-05, "loss": 0.1074, "step": 33080 }, { "epoch": 15.84, "learning_rate": 7.907474777671366e-05, "loss": 0.1055, "step": 33090 }, { "epoch": 15.85, "learning_rate": 7.906306539556354e-05, "loss": 0.1258, "step": 33100 }, { "epoch": 15.85, "learning_rate": 7.905138061775292e-05, "loss": 0.1252, "step": 33110 }, { "epoch": 15.86, "learning_rate": 7.90396934442454e-05, "loss": 0.1152, "step": 33120 }, { "epoch": 15.86, "learning_rate": 7.90280038760047e-05, "loss": 0.1151, "step": 33130 }, { "epoch": 15.87, "learning_rate": 7.901631191399484e-05, "loss": 0.1207, "step": 33140 }, { "epoch": 15.87, "learning_rate": 7.900461755917996e-05, "loss": 0.1086, "step": 33150 }, { "epoch": 15.88, "learning_rate": 7.899292081252443e-05, "loss": 0.1057, "step": 33160 }, { "epoch": 15.88, "learning_rate": 7.898122167499282e-05, "loss": 0.1141, "step": 33170 }, { "epoch": 15.88, "learning_rate": 7.89695201475499e-05, "loss": 0.1211, "step": 33180 }, { "epoch": 15.89, "learning_rate": 7.895781623116059e-05, "loss": 0.1171, "step": 33190 }, { "epoch": 15.89, "learning_rate": 7.894610992679008e-05, "loss": 0.1147, "step": 33200 }, { "epoch": 15.9, "learning_rate": 7.893440123540371e-05, "loss": 0.1134, "step": 33210 }, { "epoch": 15.9, "learning_rate": 7.8922690157967e-05, "loss": 0.1125, "step": 33220 }, { "epoch": 15.91, "learning_rate": 7.891097669544574e-05, "loss": 0.1188, "step": 33230 }, { "epoch": 15.91, "learning_rate": 7.889926084880582e-05, "loss": 0.1273, "step": 33240 }, { "epoch": 15.92, "learning_rate": 7.888754261901343e-05, "loss": 0.1168, "step": 33250 }, { "epoch": 15.92, "learning_rate": 7.887582200703486e-05, "loss": 0.113, "step": 33260 }, { "epoch": 15.93, "learning_rate": 7.886409901383666e-05, "loss": 0.1135, "step": 33270 }, { "epoch": 15.93, "learning_rate": 7.885237364038556e-05, "loss": 0.1317, "step": 33280 }, { "epoch": 15.93, "learning_rate": 7.884064588764848e-05, "loss": 0.1118, "step": 33290 }, { "epoch": 15.94, "learning_rate": 7.882891575659253e-05, "loss": 0.1104, "step": 33300 }, { "epoch": 15.94, "learning_rate": 7.881718324818503e-05, "loss": 0.1143, "step": 33310 }, { "epoch": 15.95, "learning_rate": 7.880544836339349e-05, "loss": 0.1102, "step": 33320 }, { "epoch": 16.0, "learning_rate": 7.879371110318562e-05, "loss": 0.1517, "step": 33330 }, { "epoch": 16.01, "learning_rate": 7.878197146852932e-05, "loss": 0.116, "step": 33340 }, { "epoch": 16.01, "learning_rate": 7.87702294603927e-05, "loss": 0.1152, "step": 33350 }, { "epoch": 16.01, "learning_rate": 7.875848507974404e-05, "loss": 0.1204, "step": 33360 }, { "epoch": 16.02, "learning_rate": 7.874673832755183e-05, "loss": 0.0955, "step": 33370 }, { "epoch": 16.02, "learning_rate": 7.873498920478477e-05, "loss": 0.1048, "step": 33380 }, { "epoch": 16.03, "learning_rate": 7.872323771241172e-05, "loss": 0.1067, "step": 33390 }, { "epoch": 16.03, "learning_rate": 7.871148385140179e-05, "loss": 0.1132, "step": 33400 }, { "epoch": 16.04, "learning_rate": 7.869972762272421e-05, "loss": 0.1285, "step": 33410 }, { "epoch": 16.04, "learning_rate": 7.868796902734849e-05, "loss": 0.1084, "step": 33420 }, { "epoch": 16.05, "learning_rate": 7.867620806624425e-05, "loss": 0.1059, "step": 33430 }, { "epoch": 16.05, "learning_rate": 7.866444474038137e-05, "loss": 0.112, "step": 33440 }, { "epoch": 16.06, "learning_rate": 7.865267905072992e-05, "loss": 0.1041, "step": 33450 }, { "epoch": 16.06, "learning_rate": 7.864091099826011e-05, "loss": 0.1196, "step": 33460 }, { "epoch": 16.06, "learning_rate": 7.862914058394239e-05, "loss": 0.1134, "step": 33470 }, { "epoch": 16.07, "learning_rate": 7.861736780874744e-05, "loss": 0.1113, "step": 33480 }, { "epoch": 16.07, "learning_rate": 7.860559267364604e-05, "loss": 0.1096, "step": 33490 }, { "epoch": 16.08, "learning_rate": 7.859381517960924e-05, "loss": 0.1195, "step": 33500 }, { "epoch": 16.08, "learning_rate": 7.858203532760824e-05, "loss": 0.1181, "step": 33510 }, { "epoch": 16.09, "learning_rate": 7.857025311861449e-05, "loss": 0.1042, "step": 33520 }, { "epoch": 16.09, "learning_rate": 7.855846855359957e-05, "loss": 0.1068, "step": 33530 }, { "epoch": 16.1, "learning_rate": 7.85466816335353e-05, "loss": 0.0931, "step": 33540 }, { "epoch": 16.1, "learning_rate": 7.853489235939367e-05, "loss": 0.0959, "step": 33550 }, { "epoch": 16.11, "learning_rate": 7.852310073214689e-05, "loss": 0.094, "step": 33560 }, { "epoch": 16.11, "learning_rate": 7.851130675276732e-05, "loss": 0.0961, "step": 33570 }, { "epoch": 16.12, "learning_rate": 7.849951042222754e-05, "loss": 0.1266, "step": 33580 }, { "epoch": 16.12, "learning_rate": 7.848771174150035e-05, "loss": 0.11, "step": 33590 }, { "epoch": 16.12, "learning_rate": 7.847591071155871e-05, "loss": 0.1054, "step": 33600 }, { "epoch": 16.13, "learning_rate": 7.846410733337578e-05, "loss": 0.0988, "step": 33610 }, { "epoch": 16.13, "learning_rate": 7.845230160792489e-05, "loss": 0.1178, "step": 33620 }, { "epoch": 16.14, "learning_rate": 7.844049353617962e-05, "loss": 0.1197, "step": 33630 }, { "epoch": 16.14, "learning_rate": 7.842868311911369e-05, "loss": 0.1066, "step": 33640 }, { "epoch": 16.15, "learning_rate": 7.841687035770106e-05, "loss": 0.1078, "step": 33650 }, { "epoch": 16.15, "learning_rate": 7.840505525291586e-05, "loss": 0.105, "step": 33660 }, { "epoch": 16.16, "learning_rate": 7.839323780573237e-05, "loss": 0.1178, "step": 33670 }, { "epoch": 16.16, "learning_rate": 7.838141801712518e-05, "loss": 0.1037, "step": 33680 }, { "epoch": 16.17, "learning_rate": 7.836959588806892e-05, "loss": 0.1032, "step": 33690 }, { "epoch": 16.17, "learning_rate": 7.835777141953853e-05, "loss": 0.1112, "step": 33700 }, { "epoch": 16.17, "learning_rate": 7.834594461250909e-05, "loss": 0.0972, "step": 33710 }, { "epoch": 16.18, "learning_rate": 7.833411546795593e-05, "loss": 0.1031, "step": 33720 }, { "epoch": 16.18, "learning_rate": 7.832228398685446e-05, "loss": 0.0988, "step": 33730 }, { "epoch": 16.19, "learning_rate": 7.831045017018042e-05, "loss": 0.1004, "step": 33740 }, { "epoch": 16.19, "learning_rate": 7.829861401890964e-05, "loss": 0.1231, "step": 33750 }, { "epoch": 16.2, "learning_rate": 7.828677553401818e-05, "loss": 0.1095, "step": 33760 }, { "epoch": 16.2, "learning_rate": 7.827493471648228e-05, "loss": 0.1075, "step": 33770 }, { "epoch": 16.21, "learning_rate": 7.82630915672784e-05, "loss": 0.1032, "step": 33780 }, { "epoch": 16.21, "learning_rate": 7.825124608738318e-05, "loss": 0.112, "step": 33790 }, { "epoch": 16.22, "learning_rate": 7.823939827777344e-05, "loss": 0.1013, "step": 33800 }, { "epoch": 16.22, "learning_rate": 7.82275481394262e-05, "loss": 0.1163, "step": 33810 }, { "epoch": 16.22, "learning_rate": 7.821569567331868e-05, "loss": 0.0946, "step": 33820 }, { "epoch": 16.23, "learning_rate": 7.820384088042827e-05, "loss": 0.1177, "step": 33830 }, { "epoch": 16.23, "learning_rate": 7.819198376173255e-05, "loss": 0.104, "step": 33840 }, { "epoch": 16.24, "learning_rate": 7.818012431820935e-05, "loss": 0.1118, "step": 33850 }, { "epoch": 16.24, "learning_rate": 7.816826255083661e-05, "loss": 0.109, "step": 33860 }, { "epoch": 16.25, "learning_rate": 7.815639846059251e-05, "loss": 0.107, "step": 33870 }, { "epoch": 16.25, "learning_rate": 7.814453204845543e-05, "loss": 0.1205, "step": 33880 }, { "epoch": 16.26, "learning_rate": 7.81326633154039e-05, "loss": 0.1014, "step": 33890 }, { "epoch": 16.26, "learning_rate": 7.812079226241669e-05, "loss": 0.109, "step": 33900 }, { "epoch": 16.27, "learning_rate": 7.810891889047271e-05, "loss": 0.1049, "step": 33910 }, { "epoch": 16.27, "learning_rate": 7.80970432005511e-05, "loss": 0.1276, "step": 33920 }, { "epoch": 16.28, "learning_rate": 7.808516519363116e-05, "loss": 0.1085, "step": 33930 }, { "epoch": 16.28, "learning_rate": 7.807328487069243e-05, "loss": 0.1068, "step": 33940 }, { "epoch": 16.28, "learning_rate": 7.806140223271461e-05, "loss": 0.1065, "step": 33950 }, { "epoch": 16.29, "learning_rate": 7.804951728067756e-05, "loss": 0.102, "step": 33960 }, { "epoch": 16.29, "learning_rate": 7.803763001556139e-05, "loss": 0.1105, "step": 33970 }, { "epoch": 16.3, "learning_rate": 7.802574043834635e-05, "loss": 0.1038, "step": 33980 }, { "epoch": 16.3, "learning_rate": 7.801384855001293e-05, "loss": 0.0951, "step": 33990 }, { "epoch": 16.31, "learning_rate": 7.800195435154178e-05, "loss": 0.1256, "step": 34000 }, { "epoch": 16.31, "learning_rate": 7.799005784391372e-05, "loss": 0.1066, "step": 34010 }, { "epoch": 16.32, "learning_rate": 7.79781590281098e-05, "loss": 0.117, "step": 34020 }, { "epoch": 16.32, "learning_rate": 7.796625790511125e-05, "loss": 0.1274, "step": 34030 }, { "epoch": 16.33, "learning_rate": 7.79543544758995e-05, "loss": 0.1232, "step": 34040 }, { "epoch": 16.33, "learning_rate": 7.794244874145611e-05, "loss": 0.1155, "step": 34050 }, { "epoch": 16.33, "learning_rate": 7.793054070276294e-05, "loss": 0.1309, "step": 34060 }, { "epoch": 16.34, "learning_rate": 7.791863036080191e-05, "loss": 0.125, "step": 34070 }, { "epoch": 16.34, "learning_rate": 7.790671771655525e-05, "loss": 0.1101, "step": 34080 }, { "epoch": 16.35, "learning_rate": 7.789480277100527e-05, "loss": 0.117, "step": 34090 }, { "epoch": 16.35, "learning_rate": 7.788288552513458e-05, "loss": 0.1112, "step": 34100 }, { "epoch": 16.36, "learning_rate": 7.787096597992592e-05, "loss": 0.1079, "step": 34110 }, { "epoch": 16.36, "learning_rate": 7.785904413636218e-05, "loss": 0.1208, "step": 34120 }, { "epoch": 16.37, "learning_rate": 7.784711999542653e-05, "loss": 0.1208, "step": 34130 }, { "epoch": 16.37, "learning_rate": 7.783519355810225e-05, "loss": 0.1016, "step": 34140 }, { "epoch": 16.38, "learning_rate": 7.782326482537287e-05, "loss": 0.1112, "step": 34150 }, { "epoch": 16.38, "learning_rate": 7.781133379822207e-05, "loss": 0.1111, "step": 34160 }, { "epoch": 16.38, "learning_rate": 7.779940047763375e-05, "loss": 0.1068, "step": 34170 }, { "epoch": 16.39, "learning_rate": 7.778746486459196e-05, "loss": 0.1041, "step": 34180 }, { "epoch": 16.39, "learning_rate": 7.777552696008097e-05, "loss": 0.1178, "step": 34190 }, { "epoch": 16.4, "learning_rate": 7.776358676508522e-05, "loss": 0.1192, "step": 34200 }, { "epoch": 16.4, "learning_rate": 7.775164428058934e-05, "loss": 0.0945, "step": 34210 }, { "epoch": 16.41, "learning_rate": 7.773969950757817e-05, "loss": 0.0932, "step": 34220 }, { "epoch": 16.41, "learning_rate": 7.772775244703674e-05, "loss": 0.1082, "step": 34230 }, { "epoch": 16.42, "learning_rate": 7.771580309995024e-05, "loss": 0.106, "step": 34240 }, { "epoch": 16.42, "learning_rate": 7.770385146730405e-05, "loss": 0.1103, "step": 34250 }, { "epoch": 16.43, "learning_rate": 7.769189755008377e-05, "loss": 0.1077, "step": 34260 }, { "epoch": 16.43, "learning_rate": 7.767994134927515e-05, "loss": 0.1063, "step": 34270 }, { "epoch": 16.44, "learning_rate": 7.766798286586416e-05, "loss": 0.1167, "step": 34280 }, { "epoch": 16.44, "learning_rate": 7.765602210083695e-05, "loss": 0.1111, "step": 34290 }, { "epoch": 16.44, "learning_rate": 7.764405905517986e-05, "loss": 0.1033, "step": 34300 }, { "epoch": 16.45, "learning_rate": 7.76320937298794e-05, "loss": 0.1149, "step": 34310 }, { "epoch": 16.45, "learning_rate": 7.762012612592227e-05, "loss": 0.1061, "step": 34320 }, { "epoch": 16.46, "learning_rate": 7.760815624429538e-05, "loss": 0.102, "step": 34330 }, { "epoch": 16.46, "learning_rate": 7.759618408598583e-05, "loss": 0.1157, "step": 34340 }, { "epoch": 16.47, "learning_rate": 7.758420965198087e-05, "loss": 0.1005, "step": 34350 }, { "epoch": 16.47, "learning_rate": 7.757223294326797e-05, "loss": 0.1115, "step": 34360 }, { "epoch": 16.48, "learning_rate": 7.756025396083477e-05, "loss": 0.1009, "step": 34370 }, { "epoch": 16.48, "learning_rate": 7.754827270566913e-05, "loss": 0.0949, "step": 34380 }, { "epoch": 16.49, "learning_rate": 7.753628917875906e-05, "loss": 0.1055, "step": 34390 }, { "epoch": 16.49, "learning_rate": 7.752430338109277e-05, "loss": 0.1017, "step": 34400 }, { "epoch": 16.49, "learning_rate": 7.751231531365866e-05, "loss": 0.1241, "step": 34410 }, { "epoch": 16.5, "learning_rate": 7.750032497744532e-05, "loss": 0.1099, "step": 34420 }, { "epoch": 16.5, "learning_rate": 7.748833237344151e-05, "loss": 0.1084, "step": 34430 }, { "epoch": 16.51, "learning_rate": 7.747633750263619e-05, "loss": 0.1088, "step": 34440 }, { "epoch": 16.51, "learning_rate": 7.746434036601853e-05, "loss": 0.1045, "step": 34450 }, { "epoch": 16.52, "learning_rate": 7.745234096457782e-05, "loss": 0.1009, "step": 34460 }, { "epoch": 16.52, "learning_rate": 7.744033929930365e-05, "loss": 0.1091, "step": 34470 }, { "epoch": 16.53, "learning_rate": 7.742833537118566e-05, "loss": 0.1082, "step": 34480 }, { "epoch": 16.53, "learning_rate": 7.741632918121376e-05, "loss": 0.1136, "step": 34490 }, { "epoch": 16.54, "learning_rate": 7.740432073037804e-05, "loss": 0.1158, "step": 34500 }, { "epoch": 16.54, "learning_rate": 7.739231001966878e-05, "loss": 0.1161, "step": 34510 }, { "epoch": 16.54, "learning_rate": 7.738029705007641e-05, "loss": 0.1244, "step": 34520 }, { "epoch": 16.55, "learning_rate": 7.736828182259159e-05, "loss": 0.1102, "step": 34530 }, { "epoch": 16.55, "learning_rate": 7.73562643382051e-05, "loss": 0.1053, "step": 34540 }, { "epoch": 16.56, "learning_rate": 7.7344244597908e-05, "loss": 0.1159, "step": 34550 }, { "epoch": 16.56, "learning_rate": 7.73322226026915e-05, "loss": 0.1052, "step": 34560 }, { "epoch": 16.57, "learning_rate": 7.732019835354691e-05, "loss": 0.1047, "step": 34570 }, { "epoch": 16.57, "learning_rate": 7.730817185146588e-05, "loss": 0.111, "step": 34580 }, { "epoch": 16.58, "learning_rate": 7.72961430974401e-05, "loss": 0.0991, "step": 34590 }, { "epoch": 16.58, "learning_rate": 7.728411209246156e-05, "loss": 0.1096, "step": 34600 }, { "epoch": 16.59, "learning_rate": 7.727207883752235e-05, "loss": 0.1069, "step": 34610 }, { "epoch": 16.59, "learning_rate": 7.72600433336148e-05, "loss": 0.1061, "step": 34620 }, { "epoch": 16.6, "learning_rate": 7.724800558173142e-05, "loss": 0.1076, "step": 34630 }, { "epoch": 16.6, "learning_rate": 7.723596558286486e-05, "loss": 0.1145, "step": 34640 }, { "epoch": 16.6, "learning_rate": 7.722392333800802e-05, "loss": 0.1075, "step": 34650 }, { "epoch": 16.61, "learning_rate": 7.721187884815392e-05, "loss": 0.1095, "step": 34660 }, { "epoch": 16.61, "learning_rate": 7.719983211429585e-05, "loss": 0.1255, "step": 34670 }, { "epoch": 16.62, "learning_rate": 7.718778313742719e-05, "loss": 0.1075, "step": 34680 }, { "epoch": 16.62, "learning_rate": 7.717573191854154e-05, "loss": 0.0928, "step": 34690 }, { "epoch": 16.63, "learning_rate": 7.716367845863273e-05, "loss": 0.1199, "step": 34700 }, { "epoch": 16.63, "learning_rate": 7.71516227586947e-05, "loss": 0.1185, "step": 34710 }, { "epoch": 16.64, "learning_rate": 7.713956481972167e-05, "loss": 0.1196, "step": 34720 }, { "epoch": 16.64, "learning_rate": 7.712750464270792e-05, "loss": 0.1136, "step": 34730 }, { "epoch": 16.65, "learning_rate": 7.711544222864804e-05, "loss": 0.1216, "step": 34740 }, { "epoch": 16.65, "learning_rate": 7.71033775785367e-05, "loss": 0.1183, "step": 34750 }, { "epoch": 16.65, "learning_rate": 7.709131069336883e-05, "loss": 0.118, "step": 34760 }, { "epoch": 16.66, "learning_rate": 7.707924157413949e-05, "loss": 0.12, "step": 34770 }, { "epoch": 16.66, "learning_rate": 7.706717022184398e-05, "loss": 0.1105, "step": 34780 }, { "epoch": 16.67, "learning_rate": 7.705509663747774e-05, "loss": 0.1022, "step": 34790 }, { "epoch": 16.67, "learning_rate": 7.704302082203641e-05, "loss": 0.1098, "step": 34800 }, { "epoch": 16.68, "learning_rate": 7.70309427765158e-05, "loss": 0.1203, "step": 34810 }, { "epoch": 16.68, "learning_rate": 7.701886250191193e-05, "loss": 0.1106, "step": 34820 }, { "epoch": 16.69, "learning_rate": 7.700677999922097e-05, "loss": 0.1175, "step": 34830 }, { "epoch": 16.69, "learning_rate": 7.699469526943933e-05, "loss": 0.1353, "step": 34840 }, { "epoch": 16.7, "learning_rate": 7.698260831356352e-05, "loss": 0.1171, "step": 34850 }, { "epoch": 16.7, "learning_rate": 7.697051913259033e-05, "loss": 0.1054, "step": 34860 }, { "epoch": 16.7, "learning_rate": 7.695842772751663e-05, "loss": 0.1056, "step": 34870 }, { "epoch": 16.71, "learning_rate": 7.694633409933958e-05, "loss": 0.1142, "step": 34880 }, { "epoch": 16.71, "learning_rate": 7.693423824905641e-05, "loss": 0.0967, "step": 34890 }, { "epoch": 16.72, "learning_rate": 7.692214017766466e-05, "loss": 0.1151, "step": 34900 }, { "epoch": 16.72, "learning_rate": 7.691003988616194e-05, "loss": 0.11, "step": 34910 }, { "epoch": 16.73, "learning_rate": 7.689793737554614e-05, "loss": 0.1099, "step": 34920 }, { "epoch": 16.73, "learning_rate": 7.68858326468152e-05, "loss": 0.1012, "step": 34930 }, { "epoch": 16.74, "learning_rate": 7.68737257009674e-05, "loss": 0.1101, "step": 34940 }, { "epoch": 16.74, "learning_rate": 7.686161653900111e-05, "loss": 0.1076, "step": 34950 }, { "epoch": 16.75, "learning_rate": 7.68495051619149e-05, "loss": 0.1064, "step": 34960 }, { "epoch": 16.75, "learning_rate": 7.68373915707075e-05, "loss": 0.1168, "step": 34970 }, { "epoch": 16.76, "learning_rate": 7.682527576637787e-05, "loss": 0.1178, "step": 34980 }, { "epoch": 16.76, "learning_rate": 7.681315774992514e-05, "loss": 0.1086, "step": 34990 }, { "epoch": 16.76, "learning_rate": 7.680103752234858e-05, "loss": 0.1123, "step": 35000 }, { "epoch": 16.77, "learning_rate": 7.678891508464769e-05, "loss": 0.1109, "step": 35010 }, { "epoch": 16.77, "learning_rate": 7.677679043782215e-05, "loss": 0.108, "step": 35020 }, { "epoch": 16.78, "learning_rate": 7.67646635828718e-05, "loss": 0.1102, "step": 35030 }, { "epoch": 16.78, "learning_rate": 7.675253452079664e-05, "loss": 0.1154, "step": 35040 }, { "epoch": 16.79, "learning_rate": 7.674040325259695e-05, "loss": 0.1005, "step": 35050 }, { "epoch": 16.79, "learning_rate": 7.672826977927306e-05, "loss": 0.1067, "step": 35060 }, { "epoch": 16.8, "learning_rate": 7.671613410182558e-05, "loss": 0.1239, "step": 35070 }, { "epoch": 16.8, "learning_rate": 7.670399622125524e-05, "loss": 0.1174, "step": 35080 }, { "epoch": 16.81, "learning_rate": 7.669185613856303e-05, "loss": 0.1128, "step": 35090 }, { "epoch": 16.81, "learning_rate": 7.667971385475001e-05, "loss": 0.1007, "step": 35100 }, { "epoch": 16.81, "learning_rate": 7.666756937081753e-05, "loss": 0.1001, "step": 35110 }, { "epoch": 16.82, "learning_rate": 7.665542268776705e-05, "loss": 0.1076, "step": 35120 }, { "epoch": 16.82, "learning_rate": 7.664327380660024e-05, "loss": 0.1017, "step": 35130 }, { "epoch": 16.83, "learning_rate": 7.663112272831895e-05, "loss": 0.1111, "step": 35140 }, { "epoch": 16.83, "learning_rate": 7.661896945392522e-05, "loss": 0.1034, "step": 35150 }, { "epoch": 16.84, "learning_rate": 7.660681398442124e-05, "loss": 0.1078, "step": 35160 }, { "epoch": 16.84, "learning_rate": 7.65946563208094e-05, "loss": 0.1068, "step": 35170 }, { "epoch": 16.85, "learning_rate": 7.658249646409228e-05, "loss": 0.1078, "step": 35180 }, { "epoch": 16.85, "learning_rate": 7.657033441527263e-05, "loss": 0.1276, "step": 35190 }, { "epoch": 16.86, "learning_rate": 7.655817017535339e-05, "loss": 0.1091, "step": 35200 }, { "epoch": 16.86, "learning_rate": 7.654600374533767e-05, "loss": 0.1079, "step": 35210 }, { "epoch": 16.86, "learning_rate": 7.653383512622877e-05, "loss": 0.1094, "step": 35220 }, { "epoch": 16.87, "learning_rate": 7.652166431903013e-05, "loss": 0.1052, "step": 35230 }, { "epoch": 16.87, "learning_rate": 7.650949132474545e-05, "loss": 0.1116, "step": 35240 }, { "epoch": 16.88, "learning_rate": 7.649731614437854e-05, "loss": 0.1043, "step": 35250 }, { "epoch": 16.88, "learning_rate": 7.648513877893342e-05, "loss": 0.1135, "step": 35260 }, { "epoch": 16.89, "learning_rate": 7.647295922941428e-05, "loss": 0.1144, "step": 35270 }, { "epoch": 16.89, "learning_rate": 7.646077749682552e-05, "loss": 0.1118, "step": 35280 }, { "epoch": 16.9, "learning_rate": 7.644859358217167e-05, "loss": 0.1069, "step": 35290 }, { "epoch": 16.9, "learning_rate": 7.643640748645748e-05, "loss": 0.1088, "step": 35300 }, { "epoch": 16.91, "learning_rate": 7.642421921068785e-05, "loss": 0.0951, "step": 35310 }, { "epoch": 16.91, "learning_rate": 7.64120287558679e-05, "loss": 0.1166, "step": 35320 }, { "epoch": 16.91, "learning_rate": 7.63998361230029e-05, "loss": 0.1106, "step": 35330 }, { "epoch": 16.92, "learning_rate": 7.638764131309827e-05, "loss": 0.1124, "step": 35340 }, { "epoch": 16.92, "learning_rate": 7.637544432715969e-05, "loss": 0.1013, "step": 35350 }, { "epoch": 16.93, "learning_rate": 7.636324516619296e-05, "loss": 0.1064, "step": 35360 }, { "epoch": 16.93, "learning_rate": 7.635104383120406e-05, "loss": 0.1075, "step": 35370 }, { "epoch": 16.94, "learning_rate": 7.633884032319917e-05, "loss": 0.1242, "step": 35380 }, { "epoch": 16.94, "learning_rate": 7.632663464318463e-05, "loss": 0.106, "step": 35390 }, { "epoch": 16.95, "learning_rate": 7.631442679216703e-05, "loss": 0.1148, "step": 35400 }, { "epoch": 16.95, "learning_rate": 7.630221677115299e-05, "loss": 0.103, "step": 35410 }, { "epoch": 17.0, "learning_rate": 7.629000458114945e-05, "loss": 0.1022, "step": 35420 }, { "epoch": 17.01, "learning_rate": 7.627779022316348e-05, "loss": 0.1061, "step": 35430 }, { "epoch": 17.01, "learning_rate": 7.626557369820231e-05, "loss": 0.1083, "step": 35440 }, { "epoch": 17.02, "learning_rate": 7.625335500727338e-05, "loss": 0.1089, "step": 35450 }, { "epoch": 17.02, "learning_rate": 7.624113415138427e-05, "loss": 0.1043, "step": 35460 }, { "epoch": 17.03, "learning_rate": 7.622891113154279e-05, "loss": 0.1299, "step": 35470 }, { "epoch": 17.03, "learning_rate": 7.621668594875689e-05, "loss": 0.1084, "step": 35480 }, { "epoch": 17.04, "learning_rate": 7.620445860403472e-05, "loss": 0.1045, "step": 35490 }, { "epoch": 17.04, "learning_rate": 7.619222909838456e-05, "loss": 0.106, "step": 35500 }, { "epoch": 17.05, "learning_rate": 7.617999743281495e-05, "loss": 0.0982, "step": 35510 }, { "epoch": 17.05, "learning_rate": 7.616776360833451e-05, "loss": 0.1082, "step": 35520 }, { "epoch": 17.05, "learning_rate": 7.615552762595214e-05, "loss": 0.1044, "step": 35530 }, { "epoch": 17.06, "learning_rate": 7.614328948667688e-05, "loss": 0.1039, "step": 35540 }, { "epoch": 17.06, "learning_rate": 7.61310491915179e-05, "loss": 0.1021, "step": 35550 }, { "epoch": 17.07, "learning_rate": 7.611880674148459e-05, "loss": 0.1113, "step": 35560 }, { "epoch": 17.07, "learning_rate": 7.610656213758654e-05, "loss": 0.1176, "step": 35570 }, { "epoch": 17.08, "learning_rate": 7.609431538083348e-05, "loss": 0.1048, "step": 35580 }, { "epoch": 17.08, "learning_rate": 7.608206647223528e-05, "loss": 0.1072, "step": 35590 }, { "epoch": 17.09, "learning_rate": 7.606981541280212e-05, "loss": 0.1131, "step": 35600 }, { "epoch": 17.09, "learning_rate": 7.605756220354418e-05, "loss": 0.1106, "step": 35610 }, { "epoch": 17.1, "learning_rate": 7.6045306845472e-05, "loss": 0.1092, "step": 35620 }, { "epoch": 17.1, "learning_rate": 7.603304933959614e-05, "loss": 0.1116, "step": 35630 }, { "epoch": 17.1, "learning_rate": 7.602078968692746e-05, "loss": 0.117, "step": 35640 }, { "epoch": 17.11, "learning_rate": 7.600852788847688e-05, "loss": 0.1159, "step": 35650 }, { "epoch": 17.11, "learning_rate": 7.599626394525562e-05, "loss": 0.1018, "step": 35660 }, { "epoch": 17.12, "learning_rate": 7.5983997858275e-05, "loss": 0.1281, "step": 35670 }, { "epoch": 17.12, "learning_rate": 7.597172962854649e-05, "loss": 0.0963, "step": 35680 }, { "epoch": 17.13, "learning_rate": 7.59594592570818e-05, "loss": 0.0996, "step": 35690 }, { "epoch": 17.13, "learning_rate": 7.594718674489281e-05, "loss": 0.0928, "step": 35700 }, { "epoch": 17.14, "learning_rate": 7.593491209299157e-05, "loss": 0.1056, "step": 35710 }, { "epoch": 17.14, "learning_rate": 7.592263530239028e-05, "loss": 0.1114, "step": 35720 }, { "epoch": 17.15, "learning_rate": 7.591035637410133e-05, "loss": 0.1173, "step": 35730 }, { "epoch": 17.15, "learning_rate": 7.589807530913731e-05, "loss": 0.1112, "step": 35740 }, { "epoch": 17.15, "learning_rate": 7.588579210851097e-05, "loss": 0.0991, "step": 35750 }, { "epoch": 17.16, "learning_rate": 7.587350677323519e-05, "loss": 0.1139, "step": 35760 }, { "epoch": 17.16, "learning_rate": 7.586121930432313e-05, "loss": 0.1037, "step": 35770 }, { "epoch": 17.17, "learning_rate": 7.584892970278803e-05, "loss": 0.1021, "step": 35780 }, { "epoch": 17.17, "learning_rate": 7.583663796964334e-05, "loss": 0.0999, "step": 35790 }, { "epoch": 17.18, "learning_rate": 7.582434410590269e-05, "loss": 0.0989, "step": 35800 }, { "epoch": 17.18, "learning_rate": 7.581204811257989e-05, "loss": 0.1172, "step": 35810 }, { "epoch": 17.19, "learning_rate": 7.579974999068893e-05, "loss": 0.1178, "step": 35820 }, { "epoch": 17.19, "learning_rate": 7.578744974124393e-05, "loss": 0.1109, "step": 35830 }, { "epoch": 17.2, "learning_rate": 7.577514736525926e-05, "loss": 0.1035, "step": 35840 }, { "epoch": 17.2, "learning_rate": 7.57628428637494e-05, "loss": 0.1077, "step": 35850 }, { "epoch": 17.21, "learning_rate": 7.575053623772902e-05, "loss": 0.1313, "step": 35860 }, { "epoch": 17.21, "learning_rate": 7.5738227488213e-05, "loss": 0.1044, "step": 35870 }, { "epoch": 17.21, "learning_rate": 7.572591661621636e-05, "loss": 0.1116, "step": 35880 }, { "epoch": 17.22, "learning_rate": 7.57136036227543e-05, "loss": 0.1104, "step": 35890 }, { "epoch": 17.22, "learning_rate": 7.570128850884221e-05, "loss": 0.1066, "step": 35900 }, { "epoch": 17.23, "learning_rate": 7.568897127549563e-05, "loss": 0.1071, "step": 35910 }, { "epoch": 17.23, "learning_rate": 7.567665192373033e-05, "loss": 0.1069, "step": 35920 }, { "epoch": 17.24, "learning_rate": 7.566433045456216e-05, "loss": 0.0989, "step": 35930 }, { "epoch": 17.24, "learning_rate": 7.565200686900723e-05, "loss": 0.1071, "step": 35940 }, { "epoch": 17.25, "learning_rate": 7.563968116808178e-05, "loss": 0.102, "step": 35950 }, { "epoch": 17.25, "learning_rate": 7.562735335280227e-05, "loss": 0.1115, "step": 35960 }, { "epoch": 17.26, "learning_rate": 7.561502342418525e-05, "loss": 0.0981, "step": 35970 }, { "epoch": 17.26, "learning_rate": 7.560269138324755e-05, "loss": 0.0979, "step": 35980 }, { "epoch": 17.26, "learning_rate": 7.559035723100607e-05, "loss": 0.1111, "step": 35990 }, { "epoch": 17.27, "learning_rate": 7.557802096847799e-05, "loss": 0.1013, "step": 36000 }, { "epoch": 17.27, "learning_rate": 7.556568259668058e-05, "loss": 0.111, "step": 36010 }, { "epoch": 17.28, "learning_rate": 7.55533421166313e-05, "loss": 0.1174, "step": 36020 }, { "epoch": 17.28, "learning_rate": 7.554099952934782e-05, "loss": 0.1071, "step": 36030 }, { "epoch": 17.29, "learning_rate": 7.552865483584794e-05, "loss": 0.1063, "step": 36040 }, { "epoch": 17.29, "learning_rate": 7.55163080371497e-05, "loss": 0.1034, "step": 36050 }, { "epoch": 17.3, "learning_rate": 7.55039591342712e-05, "loss": 0.0955, "step": 36060 }, { "epoch": 17.3, "learning_rate": 7.549160812823084e-05, "loss": 0.1075, "step": 36070 }, { "epoch": 17.31, "learning_rate": 7.547925502004709e-05, "loss": 0.1066, "step": 36080 }, { "epoch": 17.31, "learning_rate": 7.54668998107387e-05, "loss": 0.0997, "step": 36090 }, { "epoch": 17.31, "learning_rate": 7.545454250132445e-05, "loss": 0.1086, "step": 36100 }, { "epoch": 17.32, "learning_rate": 7.544218309282343e-05, "loss": 0.1002, "step": 36110 }, { "epoch": 17.32, "learning_rate": 7.542982158625484e-05, "loss": 0.112, "step": 36120 }, { "epoch": 17.33, "learning_rate": 7.541745798263805e-05, "loss": 0.104, "step": 36130 }, { "epoch": 17.33, "learning_rate": 7.540509228299261e-05, "loss": 0.1059, "step": 36140 }, { "epoch": 17.34, "learning_rate": 7.539272448833824e-05, "loss": 0.1038, "step": 36150 }, { "epoch": 17.34, "learning_rate": 7.53803545996949e-05, "loss": 0.1074, "step": 36160 }, { "epoch": 17.35, "learning_rate": 7.53679826180826e-05, "loss": 0.106, "step": 36170 }, { "epoch": 17.35, "learning_rate": 7.535560854452159e-05, "loss": 0.1064, "step": 36180 }, { "epoch": 17.36, "learning_rate": 7.534323238003232e-05, "loss": 0.1157, "step": 36190 }, { "epoch": 17.36, "learning_rate": 7.533085412563534e-05, "loss": 0.118, "step": 36200 }, { "epoch": 17.37, "learning_rate": 7.531847378235144e-05, "loss": 0.1033, "step": 36210 }, { "epoch": 17.37, "learning_rate": 7.530609135120155e-05, "loss": 0.1067, "step": 36220 }, { "epoch": 17.37, "learning_rate": 7.529370683320675e-05, "loss": 0.0942, "step": 36230 }, { "epoch": 17.38, "learning_rate": 7.528132022938836e-05, "loss": 0.1053, "step": 36240 }, { "epoch": 17.38, "learning_rate": 7.526893154076782e-05, "loss": 0.1023, "step": 36250 }, { "epoch": 17.39, "learning_rate": 7.525654076836672e-05, "loss": 0.1026, "step": 36260 }, { "epoch": 17.39, "learning_rate": 7.52441479132069e-05, "loss": 0.1075, "step": 36270 }, { "epoch": 17.4, "learning_rate": 7.523175297631029e-05, "loss": 0.088, "step": 36280 }, { "epoch": 17.4, "learning_rate": 7.521935595869904e-05, "loss": 0.1015, "step": 36290 }, { "epoch": 17.41, "learning_rate": 7.520695686139547e-05, "loss": 0.1018, "step": 36300 }, { "epoch": 17.41, "learning_rate": 7.519455568542205e-05, "loss": 0.0945, "step": 36310 }, { "epoch": 17.42, "learning_rate": 7.518215243180142e-05, "loss": 0.1212, "step": 36320 }, { "epoch": 17.42, "learning_rate": 7.516974710155643e-05, "loss": 0.1137, "step": 36330 }, { "epoch": 17.42, "learning_rate": 7.515733969571006e-05, "loss": 0.0984, "step": 36340 }, { "epoch": 17.43, "learning_rate": 7.514493021528548e-05, "loss": 0.1167, "step": 36350 }, { "epoch": 17.43, "learning_rate": 7.513251866130601e-05, "loss": 0.0969, "step": 36360 }, { "epoch": 17.44, "learning_rate": 7.512010503479519e-05, "loss": 0.12, "step": 36370 }, { "epoch": 17.44, "learning_rate": 7.510768933677668e-05, "loss": 0.1112, "step": 36380 }, { "epoch": 17.45, "learning_rate": 7.509527156827434e-05, "loss": 0.1075, "step": 36390 }, { "epoch": 17.45, "learning_rate": 7.508285173031216e-05, "loss": 0.1196, "step": 36400 }, { "epoch": 17.46, "learning_rate": 7.507042982391437e-05, "loss": 0.1122, "step": 36410 }, { "epoch": 17.46, "learning_rate": 7.505800585010531e-05, "loss": 0.0938, "step": 36420 }, { "epoch": 17.47, "learning_rate": 7.504557980990953e-05, "loss": 0.1073, "step": 36430 }, { "epoch": 17.47, "learning_rate": 7.50331517043517e-05, "loss": 0.1105, "step": 36440 }, { "epoch": 17.47, "learning_rate": 7.502072153445672e-05, "loss": 0.1152, "step": 36450 }, { "epoch": 17.48, "learning_rate": 7.500828930124963e-05, "loss": 0.1134, "step": 36460 }, { "epoch": 17.48, "learning_rate": 7.499585500575564e-05, "loss": 0.0974, "step": 36470 }, { "epoch": 17.49, "learning_rate": 7.498341864900013e-05, "loss": 0.1028, "step": 36480 }, { "epoch": 17.49, "learning_rate": 7.497098023200866e-05, "loss": 0.103, "step": 36490 }, { "epoch": 17.5, "learning_rate": 7.495853975580695e-05, "loss": 0.1039, "step": 36500 }, { "epoch": 17.5, "learning_rate": 7.494609722142089e-05, "loss": 0.1099, "step": 36510 }, { "epoch": 17.51, "learning_rate": 7.493365262987656e-05, "loss": 0.1115, "step": 36520 }, { "epoch": 17.51, "learning_rate": 7.492120598220017e-05, "loss": 0.1101, "step": 36530 }, { "epoch": 17.52, "learning_rate": 7.490875727941812e-05, "loss": 0.1002, "step": 36540 }, { "epoch": 17.52, "learning_rate": 7.489630652255699e-05, "loss": 0.1072, "step": 36550 }, { "epoch": 17.53, "learning_rate": 7.488385371264353e-05, "loss": 0.093, "step": 36560 }, { "epoch": 17.53, "learning_rate": 7.487139885070464e-05, "loss": 0.1099, "step": 36570 }, { "epoch": 17.53, "learning_rate": 7.485894193776742e-05, "loss": 0.1144, "step": 36580 }, { "epoch": 17.54, "learning_rate": 7.484648297485906e-05, "loss": 0.1133, "step": 36590 }, { "epoch": 17.54, "learning_rate": 7.483402196300705e-05, "loss": 0.0929, "step": 36600 }, { "epoch": 17.55, "learning_rate": 7.482155890323894e-05, "loss": 0.1108, "step": 36610 }, { "epoch": 17.55, "learning_rate": 7.48090937965825e-05, "loss": 0.1, "step": 36620 }, { "epoch": 17.56, "learning_rate": 7.479662664406561e-05, "loss": 0.1039, "step": 36630 }, { "epoch": 17.56, "learning_rate": 7.478415744671644e-05, "loss": 0.1098, "step": 36640 }, { "epoch": 17.57, "learning_rate": 7.477168620556319e-05, "loss": 0.1073, "step": 36650 }, { "epoch": 17.57, "learning_rate": 7.47592129216343e-05, "loss": 0.1267, "step": 36660 }, { "epoch": 17.58, "learning_rate": 7.474673759595838e-05, "loss": 0.1123, "step": 36670 }, { "epoch": 17.58, "learning_rate": 7.47342602295642e-05, "loss": 0.1071, "step": 36680 }, { "epoch": 17.58, "learning_rate": 7.47217808234807e-05, "loss": 0.1136, "step": 36690 }, { "epoch": 17.59, "learning_rate": 7.470929937873695e-05, "loss": 0.1058, "step": 36700 }, { "epoch": 17.59, "learning_rate": 7.469681589636226e-05, "loss": 0.107, "step": 36710 }, { "epoch": 17.6, "learning_rate": 7.468433037738606e-05, "loss": 0.1053, "step": 36720 }, { "epoch": 17.6, "learning_rate": 7.467184282283795e-05, "loss": 0.1236, "step": 36730 }, { "epoch": 17.61, "learning_rate": 7.465935323374769e-05, "loss": 0.1181, "step": 36740 }, { "epoch": 17.61, "learning_rate": 7.464686161114524e-05, "loss": 0.1101, "step": 36750 }, { "epoch": 17.62, "learning_rate": 7.463436795606073e-05, "loss": 0.109, "step": 36760 }, { "epoch": 17.62, "learning_rate": 7.462187226952442e-05, "loss": 0.1238, "step": 36770 }, { "epoch": 17.63, "learning_rate": 7.460937455256673e-05, "loss": 0.1054, "step": 36780 }, { "epoch": 17.63, "learning_rate": 7.459687480621831e-05, "loss": 0.1151, "step": 36790 }, { "epoch": 17.63, "learning_rate": 7.458437303150994e-05, "loss": 0.104, "step": 36800 }, { "epoch": 17.64, "learning_rate": 7.457186922947254e-05, "loss": 0.1047, "step": 36810 }, { "epoch": 17.64, "learning_rate": 7.455936340113726e-05, "loss": 0.1013, "step": 36820 }, { "epoch": 17.65, "learning_rate": 7.454685554753535e-05, "loss": 0.1055, "step": 36830 }, { "epoch": 17.65, "learning_rate": 7.45343456696983e-05, "loss": 0.1031, "step": 36840 }, { "epoch": 17.66, "learning_rate": 7.452183376865767e-05, "loss": 0.1032, "step": 36850 }, { "epoch": 17.66, "learning_rate": 7.45093198454453e-05, "loss": 0.1062, "step": 36860 }, { "epoch": 17.67, "learning_rate": 7.449680390109309e-05, "loss": 0.1064, "step": 36870 }, { "epoch": 17.67, "learning_rate": 7.448428593663319e-05, "loss": 0.1035, "step": 36880 }, { "epoch": 17.68, "learning_rate": 7.447176595309789e-05, "loss": 0.1117, "step": 36890 }, { "epoch": 17.68, "learning_rate": 7.44592439515196e-05, "loss": 0.1042, "step": 36900 }, { "epoch": 17.69, "learning_rate": 7.444671993293096e-05, "loss": 0.107, "step": 36910 }, { "epoch": 17.69, "learning_rate": 7.443419389836476e-05, "loss": 0.1075, "step": 36920 }, { "epoch": 17.69, "learning_rate": 7.442166584885393e-05, "loss": 0.1145, "step": 36930 }, { "epoch": 17.7, "learning_rate": 7.440913578543159e-05, "loss": 0.1051, "step": 36940 }, { "epoch": 17.7, "learning_rate": 7.439660370913104e-05, "loss": 0.0945, "step": 36950 }, { "epoch": 17.71, "learning_rate": 7.43840696209857e-05, "loss": 0.1058, "step": 36960 }, { "epoch": 17.71, "learning_rate": 7.43715335220292e-05, "loss": 0.102, "step": 36970 }, { "epoch": 17.72, "learning_rate": 7.435899541329532e-05, "loss": 0.1192, "step": 36980 }, { "epoch": 17.72, "learning_rate": 7.4346455295818e-05, "loss": 0.1147, "step": 36990 }, { "epoch": 17.73, "learning_rate": 7.433391317063134e-05, "loss": 0.1014, "step": 37000 }, { "epoch": 17.73, "learning_rate": 7.43213690387696e-05, "loss": 0.106, "step": 37010 }, { "epoch": 17.74, "learning_rate": 7.430882290126728e-05, "loss": 0.11, "step": 37020 }, { "epoch": 17.74, "learning_rate": 7.429627475915893e-05, "loss": 0.1054, "step": 37030 }, { "epoch": 17.74, "learning_rate": 7.428372461347935e-05, "loss": 0.1094, "step": 37040 }, { "epoch": 17.75, "learning_rate": 7.427117246526347e-05, "loss": 0.1055, "step": 37050 }, { "epoch": 17.75, "learning_rate": 7.42586183155464e-05, "loss": 0.1081, "step": 37060 }, { "epoch": 17.76, "learning_rate": 7.424606216536337e-05, "loss": 0.1161, "step": 37070 }, { "epoch": 17.76, "learning_rate": 7.423350401574985e-05, "loss": 0.1027, "step": 37080 }, { "epoch": 17.77, "learning_rate": 7.422094386774143e-05, "loss": 0.1075, "step": 37090 }, { "epoch": 17.77, "learning_rate": 7.420838172237388e-05, "loss": 0.1107, "step": 37100 }, { "epoch": 17.78, "learning_rate": 7.41958175806831e-05, "loss": 0.1082, "step": 37110 }, { "epoch": 17.78, "learning_rate": 7.418325144370519e-05, "loss": 0.097, "step": 37120 }, { "epoch": 17.79, "learning_rate": 7.417068331247643e-05, "loss": 0.1038, "step": 37130 }, { "epoch": 17.79, "learning_rate": 7.415811318803322e-05, "loss": 0.1051, "step": 37140 }, { "epoch": 17.79, "learning_rate": 7.414554107141214e-05, "loss": 0.1011, "step": 37150 }, { "epoch": 17.8, "learning_rate": 7.413296696364993e-05, "loss": 0.1056, "step": 37160 }, { "epoch": 17.8, "learning_rate": 7.412039086578353e-05, "loss": 0.0975, "step": 37170 }, { "epoch": 17.81, "learning_rate": 7.410781277885002e-05, "loss": 0.1021, "step": 37180 }, { "epoch": 17.81, "learning_rate": 7.409523270388661e-05, "loss": 0.1141, "step": 37190 }, { "epoch": 17.82, "learning_rate": 7.408265064193071e-05, "loss": 0.0968, "step": 37200 }, { "epoch": 17.82, "learning_rate": 7.407006659401992e-05, "loss": 0.1075, "step": 37210 }, { "epoch": 17.83, "learning_rate": 7.405748056119195e-05, "loss": 0.1073, "step": 37220 }, { "epoch": 17.83, "learning_rate": 7.404489254448468e-05, "loss": 0.1035, "step": 37230 }, { "epoch": 17.84, "learning_rate": 7.40323025449362e-05, "loss": 0.1001, "step": 37240 }, { "epoch": 17.84, "learning_rate": 7.401971056358473e-05, "loss": 0.0975, "step": 37250 }, { "epoch": 17.85, "learning_rate": 7.400711660146865e-05, "loss": 0.1018, "step": 37260 }, { "epoch": 17.85, "learning_rate": 7.39945206596265e-05, "loss": 0.1066, "step": 37270 }, { "epoch": 17.85, "learning_rate": 7.398192273909699e-05, "loss": 0.1113, "step": 37280 }, { "epoch": 17.86, "learning_rate": 7.396932284091902e-05, "loss": 0.1038, "step": 37290 }, { "epoch": 17.86, "learning_rate": 7.395672096613163e-05, "loss": 0.1011, "step": 37300 }, { "epoch": 17.87, "learning_rate": 7.394411711577398e-05, "loss": 0.122, "step": 37310 }, { "epoch": 17.87, "learning_rate": 7.393151129088549e-05, "loss": 0.0975, "step": 37320 }, { "epoch": 17.88, "learning_rate": 7.391890349250567e-05, "loss": 0.1147, "step": 37330 }, { "epoch": 17.88, "learning_rate": 7.390629372167419e-05, "loss": 0.1036, "step": 37340 }, { "epoch": 17.89, "learning_rate": 7.389368197943091e-05, "loss": 0.1059, "step": 37350 }, { "epoch": 17.89, "learning_rate": 7.388106826681589e-05, "loss": 0.1052, "step": 37360 }, { "epoch": 17.9, "learning_rate": 7.386845258486926e-05, "loss": 0.1075, "step": 37370 }, { "epoch": 17.9, "learning_rate": 7.385583493463138e-05, "loss": 0.1055, "step": 37380 }, { "epoch": 17.9, "learning_rate": 7.384321531714273e-05, "loss": 0.1069, "step": 37390 }, { "epoch": 17.91, "learning_rate": 7.383059373344402e-05, "loss": 0.1162, "step": 37400 }, { "epoch": 17.91, "learning_rate": 7.381797018457606e-05, "loss": 0.1015, "step": 37410 }, { "epoch": 17.92, "learning_rate": 7.38053446715798e-05, "loss": 0.1077, "step": 37420 }, { "epoch": 17.92, "learning_rate": 7.379271719549645e-05, "loss": 0.0972, "step": 37430 }, { "epoch": 17.93, "learning_rate": 7.378008775736729e-05, "loss": 0.1293, "step": 37440 }, { "epoch": 17.93, "learning_rate": 7.376745635823382e-05, "loss": 0.0977, "step": 37450 }, { "epoch": 17.94, "learning_rate": 7.375482299913764e-05, "loss": 0.1164, "step": 37460 }, { "epoch": 17.94, "learning_rate": 7.37421876811206e-05, "loss": 0.103, "step": 37470 }, { "epoch": 17.95, "learning_rate": 7.372955040522461e-05, "loss": 0.1167, "step": 37480 }, { "epoch": 17.95, "learning_rate": 7.371691117249184e-05, "loss": 0.1024, "step": 37490 }, { "epoch": 18.0, "learning_rate": 7.370426998396452e-05, "loss": 0.1227, "step": 37500 }, { "epoch": 18.01, "learning_rate": 7.369162684068512e-05, "loss": 0.1147, "step": 37510 }, { "epoch": 18.01, "learning_rate": 7.367898174369627e-05, "loss": 0.1063, "step": 37520 }, { "epoch": 18.02, "learning_rate": 7.366633469404071e-05, "loss": 0.0921, "step": 37530 }, { "epoch": 18.02, "learning_rate": 7.365368569276137e-05, "loss": 0.1039, "step": 37540 }, { "epoch": 18.03, "learning_rate": 7.364103474090133e-05, "loss": 0.1001, "step": 37550 }, { "epoch": 18.03, "learning_rate": 7.362838183950387e-05, "loss": 0.1157, "step": 37560 }, { "epoch": 18.03, "learning_rate": 7.361572698961238e-05, "loss": 0.108, "step": 37570 }, { "epoch": 18.04, "learning_rate": 7.360307019227042e-05, "loss": 0.1104, "step": 37580 }, { "epoch": 18.04, "learning_rate": 7.359041144852174e-05, "loss": 0.0987, "step": 37590 }, { "epoch": 18.05, "learning_rate": 7.357775075941025e-05, "loss": 0.1026, "step": 37600 }, { "epoch": 18.05, "learning_rate": 7.356508812597995e-05, "loss": 0.1036, "step": 37610 }, { "epoch": 18.06, "learning_rate": 7.35524235492751e-05, "loss": 0.1058, "step": 37620 }, { "epoch": 18.06, "learning_rate": 7.353975703034006e-05, "loss": 0.0995, "step": 37630 }, { "epoch": 18.07, "learning_rate": 7.352708857021934e-05, "loss": 0.0976, "step": 37640 }, { "epoch": 18.07, "learning_rate": 7.351441816995767e-05, "loss": 0.0967, "step": 37650 }, { "epoch": 18.08, "learning_rate": 7.350174583059989e-05, "loss": 0.1089, "step": 37660 }, { "epoch": 18.08, "learning_rate": 7.348907155319101e-05, "loss": 0.0971, "step": 37670 }, { "epoch": 18.09, "learning_rate": 7.34763953387762e-05, "loss": 0.094, "step": 37680 }, { "epoch": 18.09, "learning_rate": 7.346371718840079e-05, "loss": 0.094, "step": 37690 }, { "epoch": 18.09, "learning_rate": 7.345103710311031e-05, "loss": 0.0982, "step": 37700 }, { "epoch": 18.1, "learning_rate": 7.343835508395036e-05, "loss": 0.0951, "step": 37710 }, { "epoch": 18.1, "learning_rate": 7.342567113196677e-05, "loss": 0.0859, "step": 37720 }, { "epoch": 18.11, "learning_rate": 7.341298524820552e-05, "loss": 0.0966, "step": 37730 }, { "epoch": 18.11, "learning_rate": 7.340029743371275e-05, "loss": 0.1148, "step": 37740 }, { "epoch": 18.12, "learning_rate": 7.338760768953473e-05, "loss": 0.0923, "step": 37750 }, { "epoch": 18.12, "learning_rate": 7.337491601671789e-05, "loss": 0.1052, "step": 37760 }, { "epoch": 18.13, "learning_rate": 7.336222241630887e-05, "loss": 0.0992, "step": 37770 }, { "epoch": 18.13, "learning_rate": 7.334952688935444e-05, "loss": 0.0948, "step": 37780 }, { "epoch": 18.14, "learning_rate": 7.333682943690152e-05, "loss": 0.0934, "step": 37790 }, { "epoch": 18.14, "learning_rate": 7.332413005999717e-05, "loss": 0.0936, "step": 37800 }, { "epoch": 18.14, "learning_rate": 7.331142875968867e-05, "loss": 0.1072, "step": 37810 }, { "epoch": 18.15, "learning_rate": 7.329872553702339e-05, "loss": 0.1067, "step": 37820 }, { "epoch": 18.15, "learning_rate": 7.328602039304893e-05, "loss": 0.1084, "step": 37830 }, { "epoch": 18.16, "learning_rate": 7.327331332881296e-05, "loss": 0.0957, "step": 37840 }, { "epoch": 18.16, "learning_rate": 7.326060434536337e-05, "loss": 0.0984, "step": 37850 }, { "epoch": 18.17, "learning_rate": 7.324789344374822e-05, "loss": 0.1016, "step": 37860 }, { "epoch": 18.17, "learning_rate": 7.323518062501569e-05, "loss": 0.1215, "step": 37870 }, { "epoch": 18.18, "learning_rate": 7.322246589021411e-05, "loss": 0.1001, "step": 37880 }, { "epoch": 18.18, "learning_rate": 7.320974924039203e-05, "loss": 0.0975, "step": 37890 }, { "epoch": 18.19, "learning_rate": 7.319703067659809e-05, "loss": 0.1035, "step": 37900 }, { "epoch": 18.19, "learning_rate": 7.318431019988112e-05, "loss": 0.0956, "step": 37910 }, { "epoch": 18.19, "learning_rate": 7.31715878112901e-05, "loss": 0.1081, "step": 37920 }, { "epoch": 18.2, "learning_rate": 7.31588635118742e-05, "loss": 0.1018, "step": 37930 }, { "epoch": 18.2, "learning_rate": 7.314613730268267e-05, "loss": 0.0994, "step": 37940 }, { "epoch": 18.21, "learning_rate": 7.3133409184765e-05, "loss": 0.1092, "step": 37950 }, { "epoch": 18.21, "learning_rate": 7.312067915917077e-05, "loss": 0.1123, "step": 37960 }, { "epoch": 18.22, "learning_rate": 7.310794722694982e-05, "loss": 0.1172, "step": 37970 }, { "epoch": 18.22, "learning_rate": 7.3095213389152e-05, "loss": 0.1056, "step": 37980 }, { "epoch": 18.23, "learning_rate": 7.308247764682743e-05, "loss": 0.1108, "step": 37990 }, { "epoch": 18.23, "learning_rate": 7.306974000102636e-05, "loss": 0.1045, "step": 38000 }, { "epoch": 18.24, "learning_rate": 7.305700045279918e-05, "loss": 0.0978, "step": 38010 }, { "epoch": 18.24, "learning_rate": 7.304425900319643e-05, "loss": 0.0926, "step": 38020 }, { "epoch": 18.24, "learning_rate": 7.303151565326882e-05, "loss": 0.0996, "step": 38030 }, { "epoch": 18.25, "learning_rate": 7.301877040406727e-05, "loss": 0.1186, "step": 38040 }, { "epoch": 18.25, "learning_rate": 7.300602325664276e-05, "loss": 0.0987, "step": 38050 }, { "epoch": 18.26, "learning_rate": 7.299327421204647e-05, "loss": 0.0989, "step": 38060 }, { "epoch": 18.26, "learning_rate": 7.298052327132976e-05, "loss": 0.0982, "step": 38070 }, { "epoch": 18.27, "learning_rate": 7.296777043554412e-05, "loss": 0.0982, "step": 38080 }, { "epoch": 18.27, "learning_rate": 7.29550157057412e-05, "loss": 0.1068, "step": 38090 }, { "epoch": 18.28, "learning_rate": 7.29422590829728e-05, "loss": 0.106, "step": 38100 }, { "epoch": 18.28, "learning_rate": 7.292950056829091e-05, "loss": 0.1056, "step": 38110 }, { "epoch": 18.29, "learning_rate": 7.291674016274761e-05, "loss": 0.104, "step": 38120 }, { "epoch": 18.29, "learning_rate": 7.290397786739522e-05, "loss": 0.1048, "step": 38130 }, { "epoch": 18.3, "learning_rate": 7.28912136832861e-05, "loss": 0.0964, "step": 38140 }, { "epoch": 18.3, "learning_rate": 7.287844761147293e-05, "loss": 0.0911, "step": 38150 }, { "epoch": 18.3, "learning_rate": 7.28656796530084e-05, "loss": 0.0984, "step": 38160 }, { "epoch": 18.31, "learning_rate": 7.285290980894541e-05, "loss": 0.1005, "step": 38170 }, { "epoch": 18.31, "learning_rate": 7.284013808033703e-05, "loss": 0.1024, "step": 38180 }, { "epoch": 18.32, "learning_rate": 7.282736446823646e-05, "loss": 0.1098, "step": 38190 }, { "epoch": 18.32, "learning_rate": 7.281458897369707e-05, "loss": 0.1048, "step": 38200 }, { "epoch": 18.33, "learning_rate": 7.280181159777236e-05, "loss": 0.097, "step": 38210 }, { "epoch": 18.33, "learning_rate": 7.278903234151602e-05, "loss": 0.1039, "step": 38220 }, { "epoch": 18.34, "learning_rate": 7.277625120598189e-05, "loss": 0.107, "step": 38230 }, { "epoch": 18.34, "learning_rate": 7.276346819222394e-05, "loss": 0.0994, "step": 38240 }, { "epoch": 18.35, "learning_rate": 7.275068330129633e-05, "loss": 0.1095, "step": 38250 }, { "epoch": 18.35, "learning_rate": 7.273789653425334e-05, "loss": 0.1022, "step": 38260 }, { "epoch": 18.35, "learning_rate": 7.27251078921494e-05, "loss": 0.1029, "step": 38270 }, { "epoch": 18.36, "learning_rate": 7.271231737603916e-05, "loss": 0.1063, "step": 38280 }, { "epoch": 18.36, "learning_rate": 7.269952498697734e-05, "loss": 0.1062, "step": 38290 }, { "epoch": 18.37, "learning_rate": 7.268673072601888e-05, "loss": 0.1007, "step": 38300 }, { "epoch": 18.37, "learning_rate": 7.267393459421883e-05, "loss": 0.1149, "step": 38310 }, { "epoch": 18.38, "learning_rate": 7.266113659263242e-05, "loss": 0.0968, "step": 38320 }, { "epoch": 18.38, "learning_rate": 7.264833672231502e-05, "loss": 0.1045, "step": 38330 }, { "epoch": 18.39, "learning_rate": 7.263553498432215e-05, "loss": 0.093, "step": 38340 }, { "epoch": 18.39, "learning_rate": 7.262273137970953e-05, "loss": 0.1097, "step": 38350 }, { "epoch": 18.4, "learning_rate": 7.260992590953296e-05, "loss": 0.1065, "step": 38360 }, { "epoch": 18.4, "learning_rate": 7.259711857484845e-05, "loss": 0.1014, "step": 38370 }, { "epoch": 18.4, "learning_rate": 7.258430937671215e-05, "loss": 0.1035, "step": 38380 }, { "epoch": 18.41, "learning_rate": 7.257149831618036e-05, "loss": 0.109, "step": 38390 }, { "epoch": 18.41, "learning_rate": 7.255868539430951e-05, "loss": 0.1024, "step": 38400 }, { "epoch": 18.42, "learning_rate": 7.254587061215622e-05, "loss": 0.1013, "step": 38410 }, { "epoch": 18.42, "learning_rate": 7.253305397077727e-05, "loss": 0.099, "step": 38420 }, { "epoch": 18.43, "learning_rate": 7.252023547122954e-05, "loss": 0.1127, "step": 38430 }, { "epoch": 18.43, "learning_rate": 7.250741511457013e-05, "loss": 0.1348, "step": 38440 }, { "epoch": 18.44, "learning_rate": 7.249459290185623e-05, "loss": 0.1021, "step": 38450 }, { "epoch": 18.44, "learning_rate": 7.248176883414523e-05, "loss": 0.0937, "step": 38460 }, { "epoch": 18.45, "learning_rate": 7.246894291249466e-05, "loss": 0.1264, "step": 38470 }, { "epoch": 18.45, "learning_rate": 7.245611513796217e-05, "loss": 0.1117, "step": 38480 }, { "epoch": 18.46, "learning_rate": 7.244328551160562e-05, "loss": 0.1073, "step": 38490 }, { "epoch": 18.46, "learning_rate": 7.243045403448298e-05, "loss": 0.1019, "step": 38500 }, { "epoch": 18.46, "learning_rate": 7.241762070765239e-05, "loss": 0.0999, "step": 38510 }, { "epoch": 18.47, "learning_rate": 7.240478553217213e-05, "loss": 0.1065, "step": 38520 }, { "epoch": 18.47, "learning_rate": 7.239194850910065e-05, "loss": 0.102, "step": 38530 }, { "epoch": 18.48, "learning_rate": 7.237910963949658e-05, "loss": 0.1071, "step": 38540 }, { "epoch": 18.48, "learning_rate": 7.23662689244186e-05, "loss": 0.1027, "step": 38550 }, { "epoch": 18.49, "learning_rate": 7.235342636492564e-05, "loss": 0.0914, "step": 38560 }, { "epoch": 18.49, "learning_rate": 7.234058196207676e-05, "loss": 0.1002, "step": 38570 }, { "epoch": 18.5, "learning_rate": 7.232773571693117e-05, "loss": 0.1079, "step": 38580 }, { "epoch": 18.5, "learning_rate": 7.231488763054818e-05, "loss": 0.0982, "step": 38590 }, { "epoch": 18.51, "learning_rate": 7.230203770398734e-05, "loss": 0.1015, "step": 38600 }, { "epoch": 18.51, "learning_rate": 7.228918593830827e-05, "loss": 0.1028, "step": 38610 }, { "epoch": 18.51, "learning_rate": 7.227633233457083e-05, "loss": 0.0999, "step": 38620 }, { "epoch": 18.52, "learning_rate": 7.226347689383494e-05, "loss": 0.0878, "step": 38630 }, { "epoch": 18.52, "learning_rate": 7.225061961716072e-05, "loss": 0.101, "step": 38640 }, { "epoch": 18.53, "learning_rate": 7.223776050560845e-05, "loss": 0.0947, "step": 38650 }, { "epoch": 18.53, "learning_rate": 7.222489956023852e-05, "loss": 0.0958, "step": 38660 }, { "epoch": 18.54, "learning_rate": 7.221203678211153e-05, "loss": 0.1052, "step": 38670 }, { "epoch": 18.54, "learning_rate": 7.219917217228816e-05, "loss": 0.1104, "step": 38680 }, { "epoch": 18.55, "learning_rate": 7.218630573182933e-05, "loss": 0.1197, "step": 38690 }, { "epoch": 18.55, "learning_rate": 7.217343746179601e-05, "loss": 0.098, "step": 38700 }, { "epoch": 18.56, "learning_rate": 7.216056736324939e-05, "loss": 0.0986, "step": 38710 }, { "epoch": 18.56, "learning_rate": 7.21476954372508e-05, "loss": 0.0962, "step": 38720 }, { "epoch": 18.56, "learning_rate": 7.213482168486167e-05, "loss": 0.114, "step": 38730 }, { "epoch": 18.57, "learning_rate": 7.212194610714369e-05, "loss": 0.1181, "step": 38740 }, { "epoch": 18.57, "learning_rate": 7.210906870515858e-05, "loss": 0.0997, "step": 38750 }, { "epoch": 18.58, "learning_rate": 7.20961894799683e-05, "loss": 0.1084, "step": 38760 }, { "epoch": 18.58, "learning_rate": 7.208330843263491e-05, "loss": 0.0982, "step": 38770 }, { "epoch": 18.59, "learning_rate": 7.207042556422062e-05, "loss": 0.1121, "step": 38780 }, { "epoch": 18.59, "learning_rate": 7.205754087578781e-05, "loss": 0.1109, "step": 38790 }, { "epoch": 18.6, "learning_rate": 7.204465436839903e-05, "loss": 0.0985, "step": 38800 }, { "epoch": 18.6, "learning_rate": 7.203176604311692e-05, "loss": 0.0873, "step": 38810 }, { "epoch": 18.61, "learning_rate": 7.201887590100433e-05, "loss": 0.1002, "step": 38820 }, { "epoch": 18.61, "learning_rate": 7.200598394312422e-05, "loss": 0.0993, "step": 38830 }, { "epoch": 18.62, "learning_rate": 7.199309017053971e-05, "loss": 0.1045, "step": 38840 }, { "epoch": 18.62, "learning_rate": 7.19801945843141e-05, "loss": 0.1015, "step": 38850 }, { "epoch": 18.62, "learning_rate": 7.196729718551078e-05, "loss": 0.098, "step": 38860 }, { "epoch": 18.63, "learning_rate": 7.195439797519335e-05, "loss": 0.0995, "step": 38870 }, { "epoch": 18.63, "learning_rate": 7.194149695442551e-05, "loss": 0.1071, "step": 38880 }, { "epoch": 18.64, "learning_rate": 7.192859412427115e-05, "loss": 0.0966, "step": 38890 }, { "epoch": 18.64, "learning_rate": 7.191568948579428e-05, "loss": 0.1121, "step": 38900 }, { "epoch": 18.65, "learning_rate": 7.190278304005908e-05, "loss": 0.1172, "step": 38910 }, { "epoch": 18.65, "learning_rate": 7.188987478812986e-05, "loss": 0.1007, "step": 38920 }, { "epoch": 18.66, "learning_rate": 7.187696473107109e-05, "loss": 0.0904, "step": 38930 }, { "epoch": 18.66, "learning_rate": 7.186405286994737e-05, "loss": 0.0946, "step": 38940 }, { "epoch": 18.67, "learning_rate": 7.185113920582349e-05, "loss": 0.0991, "step": 38950 }, { "epoch": 18.67, "learning_rate": 7.183822373976437e-05, "loss": 0.1012, "step": 38960 }, { "epoch": 18.67, "learning_rate": 7.182530647283505e-05, "loss": 0.1026, "step": 38970 }, { "epoch": 18.68, "learning_rate": 7.181238740610074e-05, "loss": 0.0881, "step": 38980 }, { "epoch": 18.68, "learning_rate": 7.179946654062683e-05, "loss": 0.0966, "step": 38990 }, { "epoch": 18.69, "learning_rate": 7.178654387747879e-05, "loss": 0.0918, "step": 39000 }, { "epoch": 18.69, "learning_rate": 7.177361941772229e-05, "loss": 0.0998, "step": 39010 }, { "epoch": 18.7, "learning_rate": 7.176069316242315e-05, "loss": 0.1017, "step": 39020 }, { "epoch": 18.7, "learning_rate": 7.174776511264731e-05, "loss": 0.1017, "step": 39030 }, { "epoch": 18.71, "learning_rate": 7.173483526946087e-05, "loss": 0.0951, "step": 39040 }, { "epoch": 18.71, "learning_rate": 7.172190363393007e-05, "loss": 0.1032, "step": 39050 }, { "epoch": 18.72, "learning_rate": 7.170897020712133e-05, "loss": 0.1098, "step": 39060 }, { "epoch": 18.72, "learning_rate": 7.169603499010116e-05, "loss": 0.1116, "step": 39070 }, { "epoch": 18.72, "learning_rate": 7.168309798393627e-05, "loss": 0.1001, "step": 39080 }, { "epoch": 18.73, "learning_rate": 7.167015918969351e-05, "loss": 0.0978, "step": 39090 }, { "epoch": 18.73, "learning_rate": 7.165721860843987e-05, "loss": 0.0963, "step": 39100 }, { "epoch": 18.74, "learning_rate": 7.164427624124245e-05, "loss": 0.1016, "step": 39110 }, { "epoch": 18.74, "learning_rate": 7.163133208916856e-05, "loss": 0.092, "step": 39120 }, { "epoch": 18.75, "learning_rate": 7.161838615328561e-05, "loss": 0.093, "step": 39130 }, { "epoch": 18.75, "learning_rate": 7.160543843466119e-05, "loss": 0.127, "step": 39140 }, { "epoch": 18.76, "learning_rate": 7.159248893436304e-05, "loss": 0.1039, "step": 39150 }, { "epoch": 18.76, "learning_rate": 7.157953765345896e-05, "loss": 0.1239, "step": 39160 }, { "epoch": 18.77, "learning_rate": 7.156658459301705e-05, "loss": 0.1116, "step": 39170 }, { "epoch": 18.77, "learning_rate": 7.155362975410541e-05, "loss": 0.1107, "step": 39180 }, { "epoch": 18.78, "learning_rate": 7.15406731377924e-05, "loss": 0.0929, "step": 39190 }, { "epoch": 18.78, "learning_rate": 7.152771474514642e-05, "loss": 0.1121, "step": 39200 }, { "epoch": 18.78, "learning_rate": 7.151475457723614e-05, "loss": 0.1347, "step": 39210 }, { "epoch": 18.79, "learning_rate": 7.150179263513024e-05, "loss": 0.1097, "step": 39220 }, { "epoch": 18.79, "learning_rate": 7.148882891989766e-05, "loss": 0.1092, "step": 39230 }, { "epoch": 18.8, "learning_rate": 7.147586343260744e-05, "loss": 0.1148, "step": 39240 }, { "epoch": 18.8, "learning_rate": 7.146289617432874e-05, "loss": 0.0995, "step": 39250 }, { "epoch": 18.81, "learning_rate": 7.144992714613092e-05, "loss": 0.1181, "step": 39260 }, { "epoch": 18.81, "learning_rate": 7.143695634908345e-05, "loss": 0.1195, "step": 39270 }, { "epoch": 18.82, "learning_rate": 7.142398378425596e-05, "loss": 0.0993, "step": 39280 }, { "epoch": 18.82, "learning_rate": 7.14110094527182e-05, "loss": 0.0958, "step": 39290 }, { "epoch": 18.83, "learning_rate": 7.139803335554013e-05, "loss": 0.1069, "step": 39300 }, { "epoch": 18.83, "learning_rate": 7.138505549379175e-05, "loss": 0.1068, "step": 39310 }, { "epoch": 18.83, "learning_rate": 7.137207586854335e-05, "loss": 0.1051, "step": 39320 }, { "epoch": 18.84, "learning_rate": 7.135909448086522e-05, "loss": 0.1111, "step": 39330 }, { "epoch": 18.84, "learning_rate": 7.134611133182788e-05, "loss": 0.1012, "step": 39340 }, { "epoch": 18.85, "learning_rate": 7.133312642250196e-05, "loss": 0.1057, "step": 39350 }, { "epoch": 18.85, "learning_rate": 7.132013975395826e-05, "loss": 0.1144, "step": 39360 }, { "epoch": 18.86, "learning_rate": 7.130715132726773e-05, "loss": 0.1051, "step": 39370 }, { "epoch": 18.86, "learning_rate": 7.129416114350143e-05, "loss": 0.1166, "step": 39380 }, { "epoch": 18.87, "learning_rate": 7.128116920373059e-05, "loss": 0.1072, "step": 39390 }, { "epoch": 18.87, "learning_rate": 7.126817550902655e-05, "loss": 0.104, "step": 39400 }, { "epoch": 18.88, "learning_rate": 7.12551800604609e-05, "loss": 0.1155, "step": 39410 }, { "epoch": 18.88, "learning_rate": 7.124218285910522e-05, "loss": 0.1239, "step": 39420 }, { "epoch": 18.88, "learning_rate": 7.122918390603135e-05, "loss": 0.1025, "step": 39430 }, { "epoch": 18.89, "learning_rate": 7.121618320231122e-05, "loss": 0.0997, "step": 39440 }, { "epoch": 18.89, "learning_rate": 7.120318074901695e-05, "loss": 0.1156, "step": 39450 }, { "epoch": 18.9, "learning_rate": 7.119017654722075e-05, "loss": 0.096, "step": 39460 }, { "epoch": 18.9, "learning_rate": 7.1177170597995e-05, "loss": 0.0969, "step": 39470 }, { "epoch": 18.91, "learning_rate": 7.116416290241226e-05, "loss": 0.1056, "step": 39480 }, { "epoch": 18.91, "learning_rate": 7.115115346154514e-05, "loss": 0.1099, "step": 39490 }, { "epoch": 18.92, "learning_rate": 7.11381422764665e-05, "loss": 0.1073, "step": 39500 }, { "epoch": 18.92, "learning_rate": 7.112512934824927e-05, "loss": 0.1047, "step": 39510 }, { "epoch": 18.93, "learning_rate": 7.111211467796656e-05, "loss": 0.1048, "step": 39520 }, { "epoch": 18.93, "learning_rate": 7.109909826669164e-05, "loss": 0.115, "step": 39530 }, { "epoch": 18.94, "learning_rate": 7.108608011549783e-05, "loss": 0.1067, "step": 39540 }, { "epoch": 18.94, "learning_rate": 7.107306022545872e-05, "loss": 0.1149, "step": 39550 }, { "epoch": 18.94, "learning_rate": 7.106003859764796e-05, "loss": 0.1061, "step": 39560 }, { "epoch": 18.95, "learning_rate": 7.104701523313939e-05, "loss": 0.1111, "step": 39570 }, { "epoch": 19.0, "learning_rate": 7.103399013300692e-05, "loss": 0.1334, "step": 39580 }, { "epoch": 19.01, "learning_rate": 7.10209632983247e-05, "loss": 0.0972, "step": 39590 }, { "epoch": 19.01, "learning_rate": 7.100793473016699e-05, "loss": 0.1077, "step": 39600 }, { "epoch": 19.02, "learning_rate": 7.099490442960812e-05, "loss": 0.0967, "step": 39610 }, { "epoch": 19.02, "learning_rate": 7.098187239772266e-05, "loss": 0.1154, "step": 39620 }, { "epoch": 19.02, "learning_rate": 7.096883863558528e-05, "loss": 0.0991, "step": 39630 }, { "epoch": 19.03, "learning_rate": 7.095580314427079e-05, "loss": 0.1121, "step": 39640 }, { "epoch": 19.03, "learning_rate": 7.094276592485417e-05, "loss": 0.095, "step": 39650 }, { "epoch": 19.04, "learning_rate": 7.09297269784105e-05, "loss": 0.1017, "step": 39660 }, { "epoch": 19.04, "learning_rate": 7.091668630601504e-05, "loss": 0.1021, "step": 39670 }, { "epoch": 19.05, "learning_rate": 7.09036439087432e-05, "loss": 0.0969, "step": 39680 }, { "epoch": 19.05, "learning_rate": 7.089059978767045e-05, "loss": 0.1049, "step": 39690 }, { "epoch": 19.06, "learning_rate": 7.087755394387251e-05, "loss": 0.1092, "step": 39700 }, { "epoch": 19.06, "learning_rate": 7.086450637842518e-05, "loss": 0.1107, "step": 39710 }, { "epoch": 19.07, "learning_rate": 7.085145709240442e-05, "loss": 0.0888, "step": 39720 }, { "epoch": 19.07, "learning_rate": 7.083840608688633e-05, "loss": 0.0963, "step": 39730 }, { "epoch": 19.07, "learning_rate": 7.082535336294713e-05, "loss": 0.1019, "step": 39740 }, { "epoch": 19.08, "learning_rate": 7.08122989216632e-05, "loss": 0.104, "step": 39750 }, { "epoch": 19.08, "learning_rate": 7.079924276411111e-05, "loss": 0.0919, "step": 39760 }, { "epoch": 19.09, "learning_rate": 7.078618489136748e-05, "loss": 0.111, "step": 39770 }, { "epoch": 19.09, "learning_rate": 7.077312530450914e-05, "loss": 0.101, "step": 39780 }, { "epoch": 19.1, "learning_rate": 7.076006400461303e-05, "loss": 0.1121, "step": 39790 }, { "epoch": 19.1, "learning_rate": 7.074700099275623e-05, "loss": 0.0959, "step": 39800 }, { "epoch": 19.11, "learning_rate": 7.073393627001596e-05, "loss": 0.1027, "step": 39810 }, { "epoch": 19.11, "learning_rate": 7.072086983746964e-05, "loss": 0.0989, "step": 39820 }, { "epoch": 19.12, "learning_rate": 7.070780169619473e-05, "loss": 0.0906, "step": 39830 }, { "epoch": 19.12, "learning_rate": 7.06947318472689e-05, "loss": 0.1066, "step": 39840 }, { "epoch": 19.12, "learning_rate": 7.068166029176996e-05, "loss": 0.1123, "step": 39850 }, { "epoch": 19.13, "learning_rate": 7.066858703077582e-05, "loss": 0.1162, "step": 39860 }, { "epoch": 19.13, "learning_rate": 7.065551206536456e-05, "loss": 0.0985, "step": 39870 }, { "epoch": 19.14, "learning_rate": 7.064243539661441e-05, "loss": 0.095, "step": 39880 }, { "epoch": 19.14, "learning_rate": 7.062935702560372e-05, "loss": 0.1045, "step": 39890 }, { "epoch": 19.15, "learning_rate": 7.061627695341099e-05, "loss": 0.0942, "step": 39900 }, { "epoch": 19.15, "learning_rate": 7.060319518111483e-05, "loss": 0.0963, "step": 39910 }, { "epoch": 19.16, "learning_rate": 7.059011170979404e-05, "loss": 0.1024, "step": 39920 }, { "epoch": 19.16, "learning_rate": 7.057702654052757e-05, "loss": 0.1009, "step": 39930 }, { "epoch": 19.17, "learning_rate": 7.056393967439442e-05, "loss": 0.0996, "step": 39940 }, { "epoch": 19.17, "learning_rate": 7.055085111247379e-05, "loss": 0.0989, "step": 39950 }, { "epoch": 19.18, "learning_rate": 7.053776085584508e-05, "loss": 0.1026, "step": 39960 }, { "epoch": 19.18, "learning_rate": 7.052466890558769e-05, "loss": 0.0959, "step": 39970 }, { "epoch": 19.18, "learning_rate": 7.051157526278127e-05, "loss": 0.0937, "step": 39980 }, { "epoch": 19.19, "learning_rate": 7.049847992850558e-05, "loss": 0.1168, "step": 39990 }, { "epoch": 19.19, "learning_rate": 7.048538290384051e-05, "loss": 0.0945, "step": 40000 }, { "epoch": 19.2, "learning_rate": 7.04722841898661e-05, "loss": 0.1009, "step": 40010 }, { "epoch": 19.2, "learning_rate": 7.045918378766252e-05, "loss": 0.1023, "step": 40020 }, { "epoch": 19.21, "learning_rate": 7.044608169831007e-05, "loss": 0.0919, "step": 40030 }, { "epoch": 19.21, "learning_rate": 7.043297792288923e-05, "loss": 0.1076, "step": 40040 }, { "epoch": 19.22, "learning_rate": 7.041987246248058e-05, "loss": 0.0985, "step": 40050 }, { "epoch": 19.22, "learning_rate": 7.040676531816484e-05, "loss": 0.1066, "step": 40060 }, { "epoch": 19.23, "learning_rate": 7.039365649102287e-05, "loss": 0.1023, "step": 40070 }, { "epoch": 19.23, "learning_rate": 7.038054598213571e-05, "loss": 0.1064, "step": 40080 }, { "epoch": 19.23, "learning_rate": 7.036743379258451e-05, "loss": 0.1075, "step": 40090 }, { "epoch": 19.24, "learning_rate": 7.035431992345051e-05, "loss": 0.0938, "step": 40100 }, { "epoch": 19.24, "learning_rate": 7.034120437581516e-05, "loss": 0.0992, "step": 40110 }, { "epoch": 19.25, "learning_rate": 7.032808715076003e-05, "loss": 0.1032, "step": 40120 }, { "epoch": 19.25, "learning_rate": 7.031496824936682e-05, "loss": 0.0954, "step": 40130 }, { "epoch": 19.26, "learning_rate": 7.030184767271735e-05, "loss": 0.0922, "step": 40140 }, { "epoch": 19.26, "learning_rate": 7.028872542189362e-05, "loss": 0.1007, "step": 40150 }, { "epoch": 19.27, "learning_rate": 7.027560149797774e-05, "loss": 0.0982, "step": 40160 }, { "epoch": 19.27, "learning_rate": 7.026247590205194e-05, "loss": 0.1019, "step": 40170 }, { "epoch": 19.28, "learning_rate": 7.024934863519863e-05, "loss": 0.1027, "step": 40180 }, { "epoch": 19.28, "learning_rate": 7.023621969850034e-05, "loss": 0.1023, "step": 40190 }, { "epoch": 19.28, "learning_rate": 7.022308909303974e-05, "loss": 0.0969, "step": 40200 }, { "epoch": 19.29, "learning_rate": 7.020995681989962e-05, "loss": 0.0916, "step": 40210 }, { "epoch": 19.29, "learning_rate": 7.019682288016293e-05, "loss": 0.0936, "step": 40220 }, { "epoch": 19.3, "learning_rate": 7.018368727491275e-05, "loss": 0.099, "step": 40230 }, { "epoch": 19.3, "learning_rate": 7.017055000523229e-05, "loss": 0.0971, "step": 40240 }, { "epoch": 19.31, "learning_rate": 7.015741107220491e-05, "loss": 0.0997, "step": 40250 }, { "epoch": 19.31, "learning_rate": 7.01442704769141e-05, "loss": 0.0911, "step": 40260 }, { "epoch": 19.32, "learning_rate": 7.013112822044348e-05, "loss": 0.1018, "step": 40270 }, { "epoch": 19.32, "learning_rate": 7.011798430387683e-05, "loss": 0.103, "step": 40280 }, { "epoch": 19.33, "learning_rate": 7.010483872829803e-05, "loss": 0.1, "step": 40290 }, { "epoch": 19.33, "learning_rate": 7.009169149479115e-05, "loss": 0.1084, "step": 40300 }, { "epoch": 19.34, "learning_rate": 7.007854260444033e-05, "loss": 0.0937, "step": 40310 }, { "epoch": 19.34, "learning_rate": 7.006539205832991e-05, "loss": 0.104, "step": 40320 }, { "epoch": 19.34, "learning_rate": 7.00522398575443e-05, "loss": 0.094, "step": 40330 }, { "epoch": 19.35, "learning_rate": 7.003908600316814e-05, "loss": 0.1017, "step": 40340 }, { "epoch": 19.35, "learning_rate": 7.002593049628611e-05, "loss": 0.1042, "step": 40350 }, { "epoch": 19.36, "learning_rate": 7.001277333798308e-05, "loss": 0.0944, "step": 40360 }, { "epoch": 19.36, "learning_rate": 6.999961452934404e-05, "loss": 0.1002, "step": 40370 }, { "epoch": 19.37, "learning_rate": 6.998645407145415e-05, "loss": 0.1096, "step": 40380 }, { "epoch": 19.37, "learning_rate": 6.997329196539862e-05, "loss": 0.1136, "step": 40390 }, { "epoch": 19.38, "learning_rate": 6.996012821226288e-05, "loss": 0.1015, "step": 40400 }, { "epoch": 19.38, "learning_rate": 6.994696281313248e-05, "loss": 0.118, "step": 40410 }, { "epoch": 19.39, "learning_rate": 6.993379576909309e-05, "loss": 0.114, "step": 40420 }, { "epoch": 19.39, "learning_rate": 6.99206270812305e-05, "loss": 0.1058, "step": 40430 }, { "epoch": 19.39, "learning_rate": 6.990745675063066e-05, "loss": 0.0924, "step": 40440 }, { "epoch": 19.4, "learning_rate": 6.989428477837965e-05, "loss": 0.1039, "step": 40450 }, { "epoch": 19.4, "learning_rate": 6.988111116556371e-05, "loss": 0.0992, "step": 40460 }, { "epoch": 19.41, "learning_rate": 6.986793591326915e-05, "loss": 0.0984, "step": 40470 }, { "epoch": 19.41, "learning_rate": 6.98547590225825e-05, "loss": 0.1025, "step": 40480 }, { "epoch": 19.42, "learning_rate": 6.984158049459033e-05, "loss": 0.1034, "step": 40490 }, { "epoch": 19.42, "learning_rate": 6.982840033037944e-05, "loss": 0.1076, "step": 40500 }, { "epoch": 19.43, "learning_rate": 6.98152185310367e-05, "loss": 0.1044, "step": 40510 }, { "epoch": 19.43, "learning_rate": 6.980203509764913e-05, "loss": 0.1036, "step": 40520 }, { "epoch": 19.44, "learning_rate": 6.978885003130391e-05, "loss": 0.0929, "step": 40530 }, { "epoch": 19.44, "learning_rate": 6.977566333308834e-05, "loss": 0.1051, "step": 40540 }, { "epoch": 19.44, "learning_rate": 6.976247500408981e-05, "loss": 0.0978, "step": 40550 }, { "epoch": 19.45, "learning_rate": 6.974928504539591e-05, "loss": 0.0965, "step": 40560 }, { "epoch": 19.45, "learning_rate": 6.973609345809436e-05, "loss": 0.1043, "step": 40570 }, { "epoch": 19.46, "learning_rate": 6.972290024327295e-05, "loss": 0.1066, "step": 40580 }, { "epoch": 19.46, "learning_rate": 6.970970540201967e-05, "loss": 0.101, "step": 40590 }, { "epoch": 19.47, "learning_rate": 6.96965089354226e-05, "loss": 0.0971, "step": 40600 }, { "epoch": 19.47, "learning_rate": 6.968331084457003e-05, "loss": 0.0972, "step": 40610 }, { "epoch": 19.48, "learning_rate": 6.967011113055028e-05, "loss": 0.1039, "step": 40620 }, { "epoch": 19.48, "learning_rate": 6.965690979445185e-05, "loss": 0.0947, "step": 40630 }, { "epoch": 19.49, "learning_rate": 6.96437068373634e-05, "loss": 0.098, "step": 40640 }, { "epoch": 19.49, "learning_rate": 6.963050226037371e-05, "loss": 0.1039, "step": 40650 }, { "epoch": 19.49, "learning_rate": 6.961729606457166e-05, "loss": 0.1144, "step": 40660 }, { "epoch": 19.5, "learning_rate": 6.960408825104627e-05, "loss": 0.1087, "step": 40670 }, { "epoch": 19.5, "learning_rate": 6.959087882088677e-05, "loss": 0.1001, "step": 40680 }, { "epoch": 19.51, "learning_rate": 6.95776677751824e-05, "loss": 0.0858, "step": 40690 }, { "epoch": 19.51, "learning_rate": 6.956445511502264e-05, "loss": 0.108, "step": 40700 }, { "epoch": 19.52, "learning_rate": 6.955124084149704e-05, "loss": 0.107, "step": 40710 }, { "epoch": 19.52, "learning_rate": 6.953802495569533e-05, "loss": 0.0903, "step": 40720 }, { "epoch": 19.53, "learning_rate": 6.952480745870731e-05, "loss": 0.1053, "step": 40730 }, { "epoch": 19.53, "learning_rate": 6.951158835162297e-05, "loss": 0.1089, "step": 40740 }, { "epoch": 19.54, "learning_rate": 6.94983676355324e-05, "loss": 0.1172, "step": 40750 }, { "epoch": 19.54, "learning_rate": 6.948514531152586e-05, "loss": 0.106, "step": 40760 }, { "epoch": 19.55, "learning_rate": 6.947192138069369e-05, "loss": 0.1009, "step": 40770 }, { "epoch": 19.55, "learning_rate": 6.94586958441264e-05, "loss": 0.0999, "step": 40780 }, { "epoch": 19.55, "learning_rate": 6.944546870291464e-05, "loss": 0.1018, "step": 40790 }, { "epoch": 19.56, "learning_rate": 6.943223995814913e-05, "loss": 0.1036, "step": 40800 }, { "epoch": 19.56, "learning_rate": 6.941900961092083e-05, "loss": 0.0924, "step": 40810 }, { "epoch": 19.57, "learning_rate": 6.940577766232072e-05, "loss": 0.1009, "step": 40820 }, { "epoch": 19.57, "learning_rate": 6.939254411343998e-05, "loss": 0.1148, "step": 40830 }, { "epoch": 19.58, "learning_rate": 6.937930896536989e-05, "loss": 0.0988, "step": 40840 }, { "epoch": 19.58, "learning_rate": 6.936607221920188e-05, "loss": 0.1009, "step": 40850 }, { "epoch": 19.59, "learning_rate": 6.935283387602752e-05, "loss": 0.0938, "step": 40860 }, { "epoch": 19.59, "learning_rate": 6.933959393693848e-05, "loss": 0.1003, "step": 40870 }, { "epoch": 19.6, "learning_rate": 6.932635240302661e-05, "loss": 0.1043, "step": 40880 }, { "epoch": 19.6, "learning_rate": 6.931310927538383e-05, "loss": 0.0992, "step": 40890 }, { "epoch": 19.6, "learning_rate": 6.929986455510222e-05, "loss": 0.0992, "step": 40900 }, { "epoch": 19.61, "learning_rate": 6.928661824327403e-05, "loss": 0.1072, "step": 40910 }, { "epoch": 19.61, "learning_rate": 6.927337034099157e-05, "loss": 0.1106, "step": 40920 }, { "epoch": 19.62, "learning_rate": 6.926012084934735e-05, "loss": 0.0972, "step": 40930 }, { "epoch": 19.62, "learning_rate": 6.924686976943393e-05, "loss": 0.1066, "step": 40940 }, { "epoch": 19.63, "learning_rate": 6.92336171023441e-05, "loss": 0.0903, "step": 40950 }, { "epoch": 19.63, "learning_rate": 6.922036284917069e-05, "loss": 0.1149, "step": 40960 }, { "epoch": 19.64, "learning_rate": 6.920710701100672e-05, "loss": 0.0986, "step": 40970 }, { "epoch": 19.64, "learning_rate": 6.919384958894532e-05, "loss": 0.0974, "step": 40980 }, { "epoch": 19.65, "learning_rate": 6.918059058407978e-05, "loss": 0.1053, "step": 40990 }, { "epoch": 19.65, "learning_rate": 6.916732999750344e-05, "loss": 0.1069, "step": 41000 }, { "epoch": 19.65, "learning_rate": 6.915406783030983e-05, "loss": 0.0978, "step": 41010 }, { "epoch": 19.66, "learning_rate": 6.914080408359264e-05, "loss": 0.0981, "step": 41020 }, { "epoch": 19.66, "learning_rate": 6.912753875844562e-05, "loss": 0.0961, "step": 41030 }, { "epoch": 19.67, "learning_rate": 6.911427185596271e-05, "loss": 0.0885, "step": 41040 }, { "epoch": 19.67, "learning_rate": 6.910100337723793e-05, "loss": 0.0966, "step": 41050 }, { "epoch": 19.68, "learning_rate": 6.908773332336546e-05, "loss": 0.1044, "step": 41060 }, { "epoch": 19.68, "learning_rate": 6.907446169543963e-05, "loss": 0.0927, "step": 41070 }, { "epoch": 19.69, "learning_rate": 6.906118849455481e-05, "loss": 0.0935, "step": 41080 }, { "epoch": 19.69, "learning_rate": 6.904791372180563e-05, "loss": 0.0978, "step": 41090 }, { "epoch": 19.7, "learning_rate": 6.903463737828675e-05, "loss": 0.1246, "step": 41100 }, { "epoch": 19.7, "learning_rate": 6.9021359465093e-05, "loss": 0.0889, "step": 41110 }, { "epoch": 19.71, "learning_rate": 6.900807998331933e-05, "loss": 0.1, "step": 41120 }, { "epoch": 19.71, "learning_rate": 6.89947989340608e-05, "loss": 0.0934, "step": 41130 }, { "epoch": 19.71, "learning_rate": 6.898151631841266e-05, "loss": 0.0914, "step": 41140 }, { "epoch": 19.72, "learning_rate": 6.896823213747024e-05, "loss": 0.0948, "step": 41150 }, { "epoch": 19.72, "learning_rate": 6.895494639232898e-05, "loss": 0.1082, "step": 41160 }, { "epoch": 19.73, "learning_rate": 6.89416590840845e-05, "loss": 0.0897, "step": 41170 }, { "epoch": 19.73, "learning_rate": 6.892837021383253e-05, "loss": 0.0922, "step": 41180 }, { "epoch": 19.74, "learning_rate": 6.891507978266892e-05, "loss": 0.1121, "step": 41190 }, { "epoch": 19.74, "learning_rate": 6.890178779168963e-05, "loss": 0.1013, "step": 41200 }, { "epoch": 19.75, "learning_rate": 6.88884942419908e-05, "loss": 0.0995, "step": 41210 }, { "epoch": 19.75, "learning_rate": 6.887519913466868e-05, "loss": 0.0924, "step": 41220 }, { "epoch": 19.76, "learning_rate": 6.886190247081961e-05, "loss": 0.0982, "step": 41230 }, { "epoch": 19.76, "learning_rate": 6.88486042515401e-05, "loss": 0.1069, "step": 41240 }, { "epoch": 19.76, "learning_rate": 6.883530447792679e-05, "loss": 0.1058, "step": 41250 }, { "epoch": 19.77, "learning_rate": 6.882200315107642e-05, "loss": 0.1031, "step": 41260 }, { "epoch": 19.77, "learning_rate": 6.880870027208587e-05, "loss": 0.0995, "step": 41270 }, { "epoch": 19.78, "learning_rate": 6.879539584205215e-05, "loss": 0.1098, "step": 41280 }, { "epoch": 19.78, "learning_rate": 6.878208986207242e-05, "loss": 0.1023, "step": 41290 }, { "epoch": 19.79, "learning_rate": 6.876878233324391e-05, "loss": 0.0966, "step": 41300 }, { "epoch": 19.79, "learning_rate": 6.875547325666404e-05, "loss": 0.097, "step": 41310 }, { "epoch": 19.8, "learning_rate": 6.874216263343032e-05, "loss": 0.1092, "step": 41320 }, { "epoch": 19.8, "learning_rate": 6.872885046464044e-05, "loss": 0.1051, "step": 41330 }, { "epoch": 19.81, "learning_rate": 6.871553675139211e-05, "loss": 0.0905, "step": 41340 }, { "epoch": 19.81, "learning_rate": 6.870222149478326e-05, "loss": 0.1032, "step": 41350 }, { "epoch": 19.81, "learning_rate": 6.868890469591194e-05, "loss": 0.1037, "step": 41360 }, { "epoch": 19.82, "learning_rate": 6.867558635587628e-05, "loss": 0.1015, "step": 41370 }, { "epoch": 19.82, "learning_rate": 6.86622664757746e-05, "loss": 0.1126, "step": 41380 }, { "epoch": 19.83, "learning_rate": 6.864894505670527e-05, "loss": 0.09, "step": 41390 }, { "epoch": 19.83, "learning_rate": 6.863562209976685e-05, "loss": 0.1031, "step": 41400 }, { "epoch": 19.84, "learning_rate": 6.862229760605802e-05, "loss": 0.1024, "step": 41410 }, { "epoch": 19.84, "learning_rate": 6.860897157667755e-05, "loss": 0.1054, "step": 41420 }, { "epoch": 19.85, "learning_rate": 6.859564401272438e-05, "loss": 0.0946, "step": 41430 }, { "epoch": 19.85, "learning_rate": 6.858231491529754e-05, "loss": 0.1097, "step": 41440 }, { "epoch": 19.86, "learning_rate": 6.85689842854962e-05, "loss": 0.0882, "step": 41450 }, { "epoch": 19.86, "learning_rate": 6.855565212441967e-05, "loss": 0.0902, "step": 41460 }, { "epoch": 19.87, "learning_rate": 6.854231843316738e-05, "loss": 0.1167, "step": 41470 }, { "epoch": 19.87, "learning_rate": 6.852898321283886e-05, "loss": 0.0961, "step": 41480 }, { "epoch": 19.87, "learning_rate": 6.851564646453382e-05, "loss": 0.1042, "step": 41490 }, { "epoch": 19.88, "learning_rate": 6.850230818935201e-05, "loss": 0.1027, "step": 41500 }, { "epoch": 19.88, "learning_rate": 6.848896838839341e-05, "loss": 0.1208, "step": 41510 }, { "epoch": 19.89, "learning_rate": 6.847562706275808e-05, "loss": 0.0968, "step": 41520 }, { "epoch": 19.89, "learning_rate": 6.846228421354614e-05, "loss": 0.1298, "step": 41530 }, { "epoch": 19.9, "learning_rate": 6.844893984185796e-05, "loss": 0.1016, "step": 41540 }, { "epoch": 19.9, "learning_rate": 6.843559394879395e-05, "loss": 0.0927, "step": 41550 }, { "epoch": 19.91, "learning_rate": 6.842224653545465e-05, "loss": 0.0882, "step": 41560 }, { "epoch": 19.91, "learning_rate": 6.840889760294076e-05, "loss": 0.1029, "step": 41570 }, { "epoch": 19.92, "learning_rate": 6.839554715235309e-05, "loss": 0.0957, "step": 41580 }, { "epoch": 19.92, "learning_rate": 6.838219518479256e-05, "loss": 0.0877, "step": 41590 }, { "epoch": 19.92, "learning_rate": 6.836884170136027e-05, "loss": 0.1029, "step": 41600 }, { "epoch": 19.93, "learning_rate": 6.835548670315733e-05, "loss": 0.1112, "step": 41610 }, { "epoch": 19.93, "learning_rate": 6.834213019128511e-05, "loss": 0.1007, "step": 41620 }, { "epoch": 19.94, "learning_rate": 6.832877216684503e-05, "loss": 0.1679, "step": 41630 }, { "epoch": 19.94, "learning_rate": 6.831541263093863e-05, "loss": 0.0997, "step": 41640 }, { "epoch": 19.95, "learning_rate": 6.830205158466759e-05, "loss": 0.1148, "step": 41650 }, { "epoch": 19.95, "learning_rate": 6.828868902913375e-05, "loss": 0.1185, "step": 41660 }, { "epoch": 20.0, "learning_rate": 6.827532496543903e-05, "loss": 0.1097, "step": 41670 }, { "epoch": 20.01, "learning_rate": 6.826195939468544e-05, "loss": 0.1083, "step": 41680 }, { "epoch": 20.01, "learning_rate": 6.824859231797523e-05, "loss": 0.0991, "step": 41690 }, { "epoch": 20.02, "learning_rate": 6.823522373641066e-05, "loss": 0.0926, "step": 41700 }, { "epoch": 20.02, "learning_rate": 6.822185365109419e-05, "loss": 0.1118, "step": 41710 }, { "epoch": 20.03, "learning_rate": 6.820848206312833e-05, "loss": 0.0908, "step": 41720 }, { "epoch": 20.03, "learning_rate": 6.819510897361577e-05, "loss": 0.0995, "step": 41730 }, { "epoch": 20.04, "learning_rate": 6.818173438365934e-05, "loss": 0.1055, "step": 41740 }, { "epoch": 20.04, "learning_rate": 6.816835829436194e-05, "loss": 0.0905, "step": 41750 }, { "epoch": 20.05, "learning_rate": 6.815498070682663e-05, "loss": 0.0973, "step": 41760 }, { "epoch": 20.05, "learning_rate": 6.814160162215658e-05, "loss": 0.089, "step": 41770 }, { "epoch": 20.05, "learning_rate": 6.812822104145505e-05, "loss": 0.1171, "step": 41780 }, { "epoch": 20.06, "learning_rate": 6.81148389658255e-05, "loss": 0.1013, "step": 41790 }, { "epoch": 20.06, "learning_rate": 6.810145539637145e-05, "loss": 0.0963, "step": 41800 }, { "epoch": 20.07, "learning_rate": 6.808807033419657e-05, "loss": 0.0914, "step": 41810 }, { "epoch": 20.07, "learning_rate": 6.807468378040465e-05, "loss": 0.1079, "step": 41820 }, { "epoch": 20.08, "learning_rate": 6.80612957360996e-05, "loss": 0.094, "step": 41830 }, { "epoch": 20.08, "learning_rate": 6.804790620238546e-05, "loss": 0.1018, "step": 41840 }, { "epoch": 20.09, "learning_rate": 6.803451518036636e-05, "loss": 0.0987, "step": 41850 }, { "epoch": 20.09, "learning_rate": 6.802112267114663e-05, "loss": 0.0983, "step": 41860 }, { "epoch": 20.1, "learning_rate": 6.800772867583061e-05, "loss": 0.1052, "step": 41870 }, { "epoch": 20.1, "learning_rate": 6.799433319552286e-05, "loss": 0.0959, "step": 41880 }, { "epoch": 20.11, "learning_rate": 6.798093623132803e-05, "loss": 0.1077, "step": 41890 }, { "epoch": 20.11, "learning_rate": 6.796753778435087e-05, "loss": 0.0948, "step": 41900 }, { "epoch": 20.11, "learning_rate": 6.79541378556963e-05, "loss": 0.1147, "step": 41910 }, { "epoch": 20.12, "learning_rate": 6.79407364464693e-05, "loss": 0.1026, "step": 41920 }, { "epoch": 20.12, "learning_rate": 6.7927333557775e-05, "loss": 0.0987, "step": 41930 }, { "epoch": 20.13, "learning_rate": 6.791392919071871e-05, "loss": 0.1026, "step": 41940 }, { "epoch": 20.13, "learning_rate": 6.790052334640576e-05, "loss": 0.097, "step": 41950 }, { "epoch": 20.14, "learning_rate": 6.788711602594167e-05, "loss": 0.0937, "step": 41960 }, { "epoch": 20.14, "learning_rate": 6.787370723043206e-05, "loss": 0.0932, "step": 41970 }, { "epoch": 20.15, "learning_rate": 6.786029696098268e-05, "loss": 0.107, "step": 41980 }, { "epoch": 20.15, "learning_rate": 6.784688521869938e-05, "loss": 0.116, "step": 41990 }, { "epoch": 20.16, "learning_rate": 6.783347200468817e-05, "loss": 0.0956, "step": 42000 }, { "epoch": 20.16, "learning_rate": 6.782005732005517e-05, "loss": 0.093, "step": 42010 }, { "epoch": 20.16, "learning_rate": 6.780664116590656e-05, "loss": 0.1047, "step": 42020 }, { "epoch": 20.17, "learning_rate": 6.779322354334871e-05, "loss": 0.0923, "step": 42030 }, { "epoch": 20.17, "learning_rate": 6.777980445348815e-05, "loss": 0.102, "step": 42040 }, { "epoch": 20.18, "learning_rate": 6.77663838974314e-05, "loss": 0.0973, "step": 42050 }, { "epoch": 20.18, "learning_rate": 6.77529618762852e-05, "loss": 0.0853, "step": 42060 }, { "epoch": 20.19, "learning_rate": 6.773953839115637e-05, "loss": 0.1014, "step": 42070 }, { "epoch": 20.19, "learning_rate": 6.772611344315192e-05, "loss": 0.1005, "step": 42080 }, { "epoch": 20.2, "learning_rate": 6.771268703337887e-05, "loss": 0.1085, "step": 42090 }, { "epoch": 20.2, "learning_rate": 6.769925916294444e-05, "loss": 0.094, "step": 42100 }, { "epoch": 20.21, "learning_rate": 6.768582983295593e-05, "loss": 0.107, "step": 42110 }, { "epoch": 20.21, "learning_rate": 6.767239904452081e-05, "loss": 0.1065, "step": 42120 }, { "epoch": 20.21, "learning_rate": 6.765896679874663e-05, "loss": 0.0959, "step": 42130 }, { "epoch": 20.22, "learning_rate": 6.764553309674105e-05, "loss": 0.0994, "step": 42140 }, { "epoch": 20.22, "learning_rate": 6.763209793961187e-05, "loss": 0.1027, "step": 42150 }, { "epoch": 20.23, "learning_rate": 6.761866132846702e-05, "loss": 0.0976, "step": 42160 }, { "epoch": 20.23, "learning_rate": 6.760522326441454e-05, "loss": 0.1072, "step": 42170 }, { "epoch": 20.24, "learning_rate": 6.759178374856257e-05, "loss": 0.1013, "step": 42180 }, { "epoch": 20.24, "learning_rate": 6.757834278201943e-05, "loss": 0.09, "step": 42190 }, { "epoch": 20.25, "learning_rate": 6.756490036589346e-05, "loss": 0.1038, "step": 42200 }, { "epoch": 20.25, "learning_rate": 6.755145650129322e-05, "loss": 0.1001, "step": 42210 }, { "epoch": 20.26, "learning_rate": 6.753801118932733e-05, "loss": 0.0941, "step": 42220 }, { "epoch": 20.26, "learning_rate": 6.752456443110456e-05, "loss": 0.1009, "step": 42230 }, { "epoch": 20.27, "learning_rate": 6.751111622773376e-05, "loss": 0.0978, "step": 42240 }, { "epoch": 20.27, "learning_rate": 6.749766658032393e-05, "loss": 0.0873, "step": 42250 }, { "epoch": 20.27, "learning_rate": 6.748421548998422e-05, "loss": 0.0988, "step": 42260 }, { "epoch": 20.28, "learning_rate": 6.747076295782381e-05, "loss": 0.0923, "step": 42270 }, { "epoch": 20.28, "learning_rate": 6.74573089849521e-05, "loss": 0.1002, "step": 42280 }, { "epoch": 20.29, "learning_rate": 6.744385357247851e-05, "loss": 0.0949, "step": 42290 }, { "epoch": 20.29, "learning_rate": 6.743039672151267e-05, "loss": 0.1021, "step": 42300 }, { "epoch": 20.3, "learning_rate": 6.741693843316427e-05, "loss": 0.0972, "step": 42310 }, { "epoch": 20.3, "learning_rate": 6.740347870854315e-05, "loss": 0.0923, "step": 42320 }, { "epoch": 20.31, "learning_rate": 6.739001754875922e-05, "loss": 0.0992, "step": 42330 }, { "epoch": 20.31, "learning_rate": 6.737655495492259e-05, "loss": 0.095, "step": 42340 }, { "epoch": 20.32, "learning_rate": 6.736309092814342e-05, "loss": 0.1, "step": 42350 }, { "epoch": 20.32, "learning_rate": 6.734962546953198e-05, "loss": 0.0978, "step": 42360 }, { "epoch": 20.32, "learning_rate": 6.733615858019873e-05, "loss": 0.1035, "step": 42370 }, { "epoch": 20.33, "learning_rate": 6.732269026125419e-05, "loss": 0.1034, "step": 42380 }, { "epoch": 20.33, "learning_rate": 6.730922051380901e-05, "loss": 0.1034, "step": 42390 }, { "epoch": 20.34, "learning_rate": 6.729574933897397e-05, "loss": 0.1003, "step": 42400 }, { "epoch": 20.34, "learning_rate": 6.728227673785993e-05, "loss": 0.1091, "step": 42410 }, { "epoch": 20.35, "learning_rate": 6.726880271157795e-05, "loss": 0.1041, "step": 42420 }, { "epoch": 20.35, "learning_rate": 6.725532726123911e-05, "loss": 0.1111, "step": 42430 }, { "epoch": 20.36, "learning_rate": 6.724185038795466e-05, "loss": 0.1048, "step": 42440 }, { "epoch": 20.36, "learning_rate": 6.722837209283595e-05, "loss": 0.0902, "step": 42450 }, { "epoch": 20.37, "learning_rate": 6.72148923769945e-05, "loss": 0.106, "step": 42460 }, { "epoch": 20.37, "learning_rate": 6.720141124154187e-05, "loss": 0.1038, "step": 42470 }, { "epoch": 20.37, "learning_rate": 6.718792868758974e-05, "loss": 0.1125, "step": 42480 }, { "epoch": 20.38, "learning_rate": 6.717444471625e-05, "loss": 0.0914, "step": 42490 }, { "epoch": 20.38, "learning_rate": 6.716095932863454e-05, "loss": 0.0988, "step": 42500 }, { "epoch": 20.39, "learning_rate": 6.714747252585547e-05, "loss": 0.0958, "step": 42510 }, { "epoch": 20.39, "learning_rate": 6.713398430902491e-05, "loss": 0.1018, "step": 42520 }, { "epoch": 20.4, "learning_rate": 6.712049467925521e-05, "loss": 0.0971, "step": 42530 }, { "epoch": 20.4, "learning_rate": 6.710700363765874e-05, "loss": 0.1049, "step": 42540 }, { "epoch": 20.41, "learning_rate": 6.709351118534806e-05, "loss": 0.0971, "step": 42550 }, { "epoch": 20.41, "learning_rate": 6.708001732343578e-05, "loss": 0.1039, "step": 42560 }, { "epoch": 20.42, "learning_rate": 6.706652205303468e-05, "loss": 0.1092, "step": 42570 }, { "epoch": 20.42, "learning_rate": 6.705302537525765e-05, "loss": 0.1115, "step": 42580 }, { "epoch": 20.43, "learning_rate": 6.703952729121765e-05, "loss": 0.0873, "step": 42590 }, { "epoch": 20.43, "learning_rate": 6.702602780202779e-05, "loss": 0.0971, "step": 42600 }, { "epoch": 20.43, "learning_rate": 6.701252690880131e-05, "loss": 0.106, "step": 42610 }, { "epoch": 20.44, "learning_rate": 6.699902461265157e-05, "loss": 0.0945, "step": 42620 }, { "epoch": 20.44, "learning_rate": 6.698552091469198e-05, "loss": 0.0964, "step": 42630 }, { "epoch": 20.45, "learning_rate": 6.697201581603613e-05, "loss": 0.1004, "step": 42640 }, { "epoch": 20.45, "learning_rate": 6.695850931779771e-05, "loss": 0.0923, "step": 42650 }, { "epoch": 20.46, "learning_rate": 6.694500142109052e-05, "loss": 0.0962, "step": 42660 }, { "epoch": 20.46, "learning_rate": 6.693149212702847e-05, "loss": 0.0869, "step": 42670 }, { "epoch": 20.47, "learning_rate": 6.69179814367256e-05, "loss": 0.0941, "step": 42680 }, { "epoch": 20.47, "learning_rate": 6.690446935129606e-05, "loss": 0.1123, "step": 42690 }, { "epoch": 20.48, "learning_rate": 6.68909558718541e-05, "loss": 0.0918, "step": 42700 }, { "epoch": 20.48, "learning_rate": 6.68774409995141e-05, "loss": 0.1021, "step": 42710 }, { "epoch": 20.48, "learning_rate": 6.686392473539057e-05, "loss": 0.0942, "step": 42720 }, { "epoch": 20.49, "learning_rate": 6.685040708059811e-05, "loss": 0.0996, "step": 42730 }, { "epoch": 20.49, "learning_rate": 6.683688803625144e-05, "loss": 0.0944, "step": 42740 }, { "epoch": 20.5, "learning_rate": 6.682336760346536e-05, "loss": 0.0958, "step": 42750 }, { "epoch": 20.5, "learning_rate": 6.680984578335487e-05, "loss": 0.1087, "step": 42760 }, { "epoch": 20.51, "learning_rate": 6.679632257703502e-05, "loss": 0.0905, "step": 42770 }, { "epoch": 20.51, "learning_rate": 6.678279798562098e-05, "loss": 0.0951, "step": 42780 }, { "epoch": 20.52, "learning_rate": 6.676927201022807e-05, "loss": 0.0931, "step": 42790 }, { "epoch": 20.52, "learning_rate": 6.675574465197167e-05, "loss": 0.0939, "step": 42800 }, { "epoch": 20.53, "learning_rate": 6.674221591196729e-05, "loss": 0.091, "step": 42810 }, { "epoch": 20.53, "learning_rate": 6.672868579133061e-05, "loss": 0.0991, "step": 42820 }, { "epoch": 20.53, "learning_rate": 6.671515429117736e-05, "loss": 0.104, "step": 42830 }, { "epoch": 20.54, "learning_rate": 6.67016214126234e-05, "loss": 0.1006, "step": 42840 }, { "epoch": 20.54, "learning_rate": 6.66880871567847e-05, "loss": 0.096, "step": 42850 }, { "epoch": 20.55, "learning_rate": 6.667455152477734e-05, "loss": 0.094, "step": 42860 }, { "epoch": 20.55, "learning_rate": 6.666101451771757e-05, "loss": 0.0931, "step": 42870 }, { "epoch": 20.56, "learning_rate": 6.664747613672167e-05, "loss": 0.0929, "step": 42880 }, { "epoch": 20.56, "learning_rate": 6.663393638290608e-05, "loss": 0.0928, "step": 42890 }, { "epoch": 20.57, "learning_rate": 6.662039525738732e-05, "loss": 0.0962, "step": 42900 }, { "epoch": 20.57, "learning_rate": 6.660685276128211e-05, "loss": 0.0997, "step": 42910 }, { "epoch": 20.58, "learning_rate": 6.659330889570717e-05, "loss": 0.1045, "step": 42920 }, { "epoch": 20.58, "learning_rate": 6.657976366177939e-05, "loss": 0.1077, "step": 42930 }, { "epoch": 20.59, "learning_rate": 6.656621706061576e-05, "loss": 0.1013, "step": 42940 }, { "epoch": 20.59, "learning_rate": 6.655266909333341e-05, "loss": 0.0899, "step": 42950 }, { "epoch": 20.59, "learning_rate": 6.653911976104956e-05, "loss": 0.104, "step": 42960 }, { "epoch": 20.6, "learning_rate": 6.652556906488154e-05, "loss": 0.1029, "step": 42970 }, { "epoch": 20.6, "learning_rate": 6.651201700594676e-05, "loss": 0.098, "step": 42980 }, { "epoch": 20.61, "learning_rate": 6.649846358536282e-05, "loss": 0.0909, "step": 42990 }, { "epoch": 20.61, "learning_rate": 6.64849088042474e-05, "loss": 0.095, "step": 43000 }, { "epoch": 20.62, "learning_rate": 6.647135266371825e-05, "loss": 0.1067, "step": 43010 }, { "epoch": 20.62, "learning_rate": 6.645779516489329e-05, "loss": 0.0986, "step": 43020 }, { "epoch": 20.63, "learning_rate": 6.644423630889052e-05, "loss": 0.0902, "step": 43030 }, { "epoch": 20.63, "learning_rate": 6.643067609682803e-05, "loss": 0.0895, "step": 43040 }, { "epoch": 20.64, "learning_rate": 6.64171145298241e-05, "loss": 0.0964, "step": 43050 }, { "epoch": 20.64, "learning_rate": 6.640355160899705e-05, "loss": 0.0983, "step": 43060 }, { "epoch": 20.64, "learning_rate": 6.638998733546533e-05, "loss": 0.0968, "step": 43070 }, { "epoch": 20.65, "learning_rate": 6.63764217103475e-05, "loss": 0.1157, "step": 43080 }, { "epoch": 20.65, "learning_rate": 6.636285473476227e-05, "loss": 0.1053, "step": 43090 }, { "epoch": 20.66, "learning_rate": 6.63492864098284e-05, "loss": 0.1028, "step": 43100 }, { "epoch": 20.66, "learning_rate": 6.633571673666479e-05, "loss": 0.0986, "step": 43110 }, { "epoch": 20.67, "learning_rate": 6.632214571639046e-05, "loss": 0.0975, "step": 43120 }, { "epoch": 20.67, "learning_rate": 6.630857335012454e-05, "loss": 0.0837, "step": 43130 }, { "epoch": 20.68, "learning_rate": 6.629499963898625e-05, "loss": 0.0877, "step": 43140 }, { "epoch": 20.68, "learning_rate": 6.628142458409494e-05, "loss": 0.0942, "step": 43150 }, { "epoch": 20.69, "learning_rate": 6.626784818657006e-05, "loss": 0.0984, "step": 43160 }, { "epoch": 20.69, "learning_rate": 6.62542704475312e-05, "loss": 0.0971, "step": 43170 }, { "epoch": 20.69, "learning_rate": 6.6240691368098e-05, "loss": 0.107, "step": 43180 }, { "epoch": 20.7, "learning_rate": 6.622711094939027e-05, "loss": 0.1005, "step": 43190 }, { "epoch": 20.7, "learning_rate": 6.621352919252788e-05, "loss": 0.0924, "step": 43200 }, { "epoch": 20.71, "learning_rate": 6.619994609863087e-05, "loss": 0.0878, "step": 43210 }, { "epoch": 20.71, "learning_rate": 6.618636166881936e-05, "loss": 0.1109, "step": 43220 }, { "epoch": 20.72, "learning_rate": 6.617277590421356e-05, "loss": 0.1014, "step": 43230 }, { "epoch": 20.72, "learning_rate": 6.615918880593379e-05, "loss": 0.0931, "step": 43240 }, { "epoch": 20.73, "learning_rate": 6.614560037510056e-05, "loss": 0.1032, "step": 43250 }, { "epoch": 20.73, "learning_rate": 6.613201061283437e-05, "loss": 0.0978, "step": 43260 }, { "epoch": 20.74, "learning_rate": 6.61184195202559e-05, "loss": 0.0903, "step": 43270 }, { "epoch": 20.74, "learning_rate": 6.610482709848595e-05, "loss": 0.1001, "step": 43280 }, { "epoch": 20.74, "learning_rate": 6.60912333486454e-05, "loss": 0.1098, "step": 43290 }, { "epoch": 20.75, "learning_rate": 6.607763827185522e-05, "loss": 0.105, "step": 43300 }, { "epoch": 20.75, "learning_rate": 6.606404186923654e-05, "loss": 0.0925, "step": 43310 }, { "epoch": 20.76, "learning_rate": 6.605044414191057e-05, "loss": 0.0957, "step": 43320 }, { "epoch": 20.76, "learning_rate": 6.603684509099865e-05, "loss": 0.1, "step": 43330 }, { "epoch": 20.77, "learning_rate": 6.602324471762219e-05, "loss": 0.0988, "step": 43340 }, { "epoch": 20.77, "learning_rate": 6.600964302290274e-05, "loss": 0.096, "step": 43350 }, { "epoch": 20.78, "learning_rate": 6.599604000796196e-05, "loss": 0.1033, "step": 43360 }, { "epoch": 20.78, "learning_rate": 6.598243567392161e-05, "loss": 0.1061, "step": 43370 }, { "epoch": 20.79, "learning_rate": 6.596883002190355e-05, "loss": 0.0996, "step": 43380 }, { "epoch": 20.79, "learning_rate": 6.595522305302976e-05, "loss": 0.1033, "step": 43390 }, { "epoch": 20.8, "learning_rate": 6.594161476842234e-05, "loss": 0.1011, "step": 43400 }, { "epoch": 20.8, "learning_rate": 6.592800516920348e-05, "loss": 0.0998, "step": 43410 }, { "epoch": 20.8, "learning_rate": 6.591439425649546e-05, "loss": 0.1027, "step": 43420 }, { "epoch": 20.81, "learning_rate": 6.590078203142072e-05, "loss": 0.1043, "step": 43430 }, { "epoch": 20.81, "learning_rate": 6.58871684951018e-05, "loss": 0.0938, "step": 43440 }, { "epoch": 20.82, "learning_rate": 6.587355364866126e-05, "loss": 0.1119, "step": 43450 }, { "epoch": 20.82, "learning_rate": 6.58599374932219e-05, "loss": 0.0986, "step": 43460 }, { "epoch": 20.83, "learning_rate": 6.584632002990653e-05, "loss": 0.0921, "step": 43470 }, { "epoch": 20.83, "learning_rate": 6.583270125983812e-05, "loss": 0.098, "step": 43480 }, { "epoch": 20.84, "learning_rate": 6.581908118413973e-05, "loss": 0.1094, "step": 43490 }, { "epoch": 20.84, "learning_rate": 6.580545980393451e-05, "loss": 0.0977, "step": 43500 }, { "epoch": 20.85, "learning_rate": 6.579183712034576e-05, "loss": 0.0933, "step": 43510 }, { "epoch": 20.85, "learning_rate": 6.577821313449685e-05, "loss": 0.1005, "step": 43520 }, { "epoch": 20.85, "learning_rate": 6.576458784751127e-05, "loss": 0.1041, "step": 43530 }, { "epoch": 20.86, "learning_rate": 6.57509612605126e-05, "loss": 0.0996, "step": 43540 }, { "epoch": 20.86, "learning_rate": 6.573733337462458e-05, "loss": 0.1021, "step": 43550 }, { "epoch": 20.87, "learning_rate": 6.572370419097099e-05, "loss": 0.1051, "step": 43560 }, { "epoch": 20.87, "learning_rate": 6.57100737106758e-05, "loss": 0.1014, "step": 43570 }, { "epoch": 20.88, "learning_rate": 6.569644193486293e-05, "loss": 0.0926, "step": 43580 }, { "epoch": 20.88, "learning_rate": 6.568280886465664e-05, "loss": 0.0989, "step": 43590 }, { "epoch": 20.89, "learning_rate": 6.566917450118108e-05, "loss": 0.1013, "step": 43600 }, { "epoch": 20.89, "learning_rate": 6.565553884556065e-05, "loss": 0.1337, "step": 43610 }, { "epoch": 20.9, "learning_rate": 6.564190189891978e-05, "loss": 0.0961, "step": 43620 }, { "epoch": 20.9, "learning_rate": 6.562826366238301e-05, "loss": 0.1014, "step": 43630 }, { "epoch": 20.9, "learning_rate": 6.561462413707504e-05, "loss": 0.1009, "step": 43640 }, { "epoch": 20.91, "learning_rate": 6.560098332412061e-05, "loss": 0.0837, "step": 43650 }, { "epoch": 20.91, "learning_rate": 6.558734122464462e-05, "loss": 0.1085, "step": 43660 }, { "epoch": 20.92, "learning_rate": 6.557369783977204e-05, "loss": 0.0904, "step": 43670 }, { "epoch": 20.92, "learning_rate": 6.556005317062798e-05, "loss": 0.0992, "step": 43680 }, { "epoch": 20.93, "learning_rate": 6.554640721833761e-05, "loss": 0.0942, "step": 43690 }, { "epoch": 20.93, "learning_rate": 6.553275998402625e-05, "loss": 0.0948, "step": 43700 }, { "epoch": 20.94, "learning_rate": 6.55191114688193e-05, "loss": 0.1204, "step": 43710 }, { "epoch": 20.94, "learning_rate": 6.550546167384228e-05, "loss": 0.1066, "step": 43720 }, { "epoch": 20.95, "learning_rate": 6.549181060022079e-05, "loss": 0.1001, "step": 43730 }, { "epoch": 20.95, "learning_rate": 6.547815824908057e-05, "loss": 0.1059, "step": 43740 }, { "epoch": 21.0, "learning_rate": 6.546450462154748e-05, "loss": 0.1056, "step": 43750 }, { "epoch": 21.01, "learning_rate": 6.545084971874738e-05, "loss": 0.1046, "step": 43760 }, { "epoch": 21.01, "learning_rate": 6.543719354180635e-05, "loss": 0.0961, "step": 43770 }, { "epoch": 21.02, "learning_rate": 6.542353609185057e-05, "loss": 0.1067, "step": 43780 }, { "epoch": 21.02, "learning_rate": 6.540987737000622e-05, "loss": 0.0985, "step": 43790 }, { "epoch": 21.03, "learning_rate": 6.53962173773997e-05, "loss": 0.0903, "step": 43800 }, { "epoch": 21.03, "learning_rate": 6.538255611515747e-05, "loss": 0.0968, "step": 43810 }, { "epoch": 21.04, "learning_rate": 6.536889358440608e-05, "loss": 0.0922, "step": 43820 }, { "epoch": 21.04, "learning_rate": 6.53552297862722e-05, "loss": 0.0904, "step": 43830 }, { "epoch": 21.04, "learning_rate": 6.53415647218826e-05, "loss": 0.091, "step": 43840 }, { "epoch": 21.05, "learning_rate": 6.532789839236416e-05, "loss": 0.1006, "step": 43850 }, { "epoch": 21.05, "learning_rate": 6.531423079884388e-05, "loss": 0.0928, "step": 43860 }, { "epoch": 21.06, "learning_rate": 6.530056194244883e-05, "loss": 0.0994, "step": 43870 }, { "epoch": 21.06, "learning_rate": 6.528689182430619e-05, "loss": 0.0976, "step": 43880 }, { "epoch": 21.07, "learning_rate": 6.527322044554327e-05, "loss": 0.0997, "step": 43890 }, { "epoch": 21.07, "learning_rate": 6.525954780728747e-05, "loss": 0.0955, "step": 43900 }, { "epoch": 21.08, "learning_rate": 6.524587391066629e-05, "loss": 0.0906, "step": 43910 }, { "epoch": 21.08, "learning_rate": 6.52321987568073e-05, "loss": 0.0947, "step": 43920 }, { "epoch": 21.09, "learning_rate": 6.52185223468383e-05, "loss": 0.0988, "step": 43930 }, { "epoch": 21.09, "learning_rate": 6.5204844681887e-05, "loss": 0.0884, "step": 43940 }, { "epoch": 21.09, "learning_rate": 6.519116576308138e-05, "loss": 0.0955, "step": 43950 }, { "epoch": 21.1, "learning_rate": 6.517748559154945e-05, "loss": 0.093, "step": 43960 }, { "epoch": 21.1, "learning_rate": 6.51638041684193e-05, "loss": 0.0981, "step": 43970 }, { "epoch": 21.11, "learning_rate": 6.515012149481922e-05, "loss": 0.0961, "step": 43980 }, { "epoch": 21.11, "learning_rate": 6.513643757187748e-05, "loss": 0.0962, "step": 43990 }, { "epoch": 21.12, "learning_rate": 6.512275240072253e-05, "loss": 0.1011, "step": 44000 }, { "epoch": 21.12, "learning_rate": 6.510906598248292e-05, "loss": 0.0951, "step": 44010 }, { "epoch": 21.13, "learning_rate": 6.509537831828729e-05, "loss": 0.0863, "step": 44020 }, { "epoch": 21.13, "learning_rate": 6.508168940926436e-05, "loss": 0.0894, "step": 44030 }, { "epoch": 21.14, "learning_rate": 6.5067999256543e-05, "loss": 0.0926, "step": 44040 }, { "epoch": 21.14, "learning_rate": 6.505430786125212e-05, "loss": 0.0977, "step": 44050 }, { "epoch": 21.14, "learning_rate": 6.50406152245208e-05, "loss": 0.0946, "step": 44060 }, { "epoch": 21.15, "learning_rate": 6.502692134747818e-05, "loss": 0.0932, "step": 44070 }, { "epoch": 21.15, "learning_rate": 6.501322623125352e-05, "loss": 0.0939, "step": 44080 }, { "epoch": 21.16, "learning_rate": 6.499952987697618e-05, "loss": 0.0937, "step": 44090 }, { "epoch": 21.16, "learning_rate": 6.498583228577559e-05, "loss": 0.0846, "step": 44100 }, { "epoch": 21.17, "learning_rate": 6.497213345878135e-05, "loss": 0.1007, "step": 44110 }, { "epoch": 21.17, "learning_rate": 6.495843339712309e-05, "loss": 0.0822, "step": 44120 }, { "epoch": 21.18, "learning_rate": 6.49447321019306e-05, "loss": 0.0843, "step": 44130 }, { "epoch": 21.18, "learning_rate": 6.493102957433371e-05, "loss": 0.0908, "step": 44140 }, { "epoch": 21.19, "learning_rate": 6.491732581546242e-05, "loss": 0.0904, "step": 44150 }, { "epoch": 21.19, "learning_rate": 6.49036208264468e-05, "loss": 0.1035, "step": 44160 }, { "epoch": 21.2, "learning_rate": 6.488991460841698e-05, "loss": 0.0964, "step": 44170 }, { "epoch": 21.2, "learning_rate": 6.487620716250328e-05, "loss": 0.0891, "step": 44180 }, { "epoch": 21.2, "learning_rate": 6.486249848983603e-05, "loss": 0.0963, "step": 44190 }, { "epoch": 21.21, "learning_rate": 6.484878859154576e-05, "loss": 0.0965, "step": 44200 }, { "epoch": 21.21, "learning_rate": 6.483507746876297e-05, "loss": 0.0993, "step": 44210 }, { "epoch": 21.22, "learning_rate": 6.48213651226184e-05, "loss": 0.085, "step": 44220 }, { "epoch": 21.22, "learning_rate": 6.48076515542428e-05, "loss": 0.1094, "step": 44230 }, { "epoch": 21.23, "learning_rate": 6.479393676476704e-05, "loss": 0.0945, "step": 44240 }, { "epoch": 21.23, "learning_rate": 6.478022075532212e-05, "loss": 0.0876, "step": 44250 }, { "epoch": 21.24, "learning_rate": 6.476650352703908e-05, "loss": 0.0912, "step": 44260 }, { "epoch": 21.24, "learning_rate": 6.475278508104915e-05, "loss": 0.0924, "step": 44270 }, { "epoch": 21.25, "learning_rate": 6.473906541848358e-05, "loss": 0.0931, "step": 44280 }, { "epoch": 21.25, "learning_rate": 6.472534454047376e-05, "loss": 0.096, "step": 44290 }, { "epoch": 21.25, "learning_rate": 6.471162244815115e-05, "loss": 0.1032, "step": 44300 }, { "epoch": 21.26, "learning_rate": 6.469789914264736e-05, "loss": 0.0969, "step": 44310 }, { "epoch": 21.26, "learning_rate": 6.468417462509408e-05, "loss": 0.0976, "step": 44320 }, { "epoch": 21.27, "learning_rate": 6.467044889662304e-05, "loss": 0.1026, "step": 44330 }, { "epoch": 21.27, "learning_rate": 6.465672195836615e-05, "loss": 0.1074, "step": 44340 }, { "epoch": 21.28, "learning_rate": 6.46429938114554e-05, "loss": 0.1012, "step": 44350 }, { "epoch": 21.28, "learning_rate": 6.462926445702284e-05, "loss": 0.0958, "step": 44360 }, { "epoch": 21.29, "learning_rate": 6.461553389620068e-05, "loss": 0.1015, "step": 44370 }, { "epoch": 21.29, "learning_rate": 6.460180213012117e-05, "loss": 0.097, "step": 44380 }, { "epoch": 21.3, "learning_rate": 6.458806915991672e-05, "loss": 0.0989, "step": 44390 }, { "epoch": 21.3, "learning_rate": 6.457433498671978e-05, "loss": 0.0917, "step": 44400 }, { "epoch": 21.3, "learning_rate": 6.456059961166295e-05, "loss": 0.0863, "step": 44410 }, { "epoch": 21.31, "learning_rate": 6.454686303587889e-05, "loss": 0.1044, "step": 44420 }, { "epoch": 21.31, "learning_rate": 6.453312526050039e-05, "loss": 0.1004, "step": 44430 }, { "epoch": 21.32, "learning_rate": 6.45193862866603e-05, "loss": 0.0988, "step": 44440 }, { "epoch": 21.32, "learning_rate": 6.45056461154916e-05, "loss": 0.1063, "step": 44450 }, { "epoch": 21.33, "learning_rate": 6.449190474812736e-05, "loss": 0.0978, "step": 44460 }, { "epoch": 21.33, "learning_rate": 6.447816218570078e-05, "loss": 0.0975, "step": 44470 }, { "epoch": 21.34, "learning_rate": 6.44644184293451e-05, "loss": 0.0953, "step": 44480 }, { "epoch": 21.34, "learning_rate": 6.445067348019367e-05, "loss": 0.0983, "step": 44490 }, { "epoch": 21.35, "learning_rate": 6.443692733938e-05, "loss": 0.0904, "step": 44500 }, { "epoch": 21.35, "learning_rate": 6.442318000803763e-05, "loss": 0.0946, "step": 44510 }, { "epoch": 21.36, "learning_rate": 6.440943148730021e-05, "loss": 0.0987, "step": 44520 }, { "epoch": 21.36, "learning_rate": 6.439568177830152e-05, "loss": 0.0923, "step": 44530 }, { "epoch": 21.36, "learning_rate": 6.438193088217541e-05, "loss": 0.0918, "step": 44540 }, { "epoch": 21.37, "learning_rate": 6.436817880005583e-05, "loss": 0.1026, "step": 44550 }, { "epoch": 21.37, "learning_rate": 6.435442553307685e-05, "loss": 0.11, "step": 44560 }, { "epoch": 21.38, "learning_rate": 6.434067108237261e-05, "loss": 0.1056, "step": 44570 }, { "epoch": 21.38, "learning_rate": 6.432691544907735e-05, "loss": 0.0987, "step": 44580 }, { "epoch": 21.39, "learning_rate": 6.431315863432544e-05, "loss": 0.0964, "step": 44590 }, { "epoch": 21.39, "learning_rate": 6.429940063925128e-05, "loss": 0.0886, "step": 44600 }, { "epoch": 21.4, "learning_rate": 6.428564146498948e-05, "loss": 0.1182, "step": 44610 }, { "epoch": 21.4, "learning_rate": 6.427188111267461e-05, "loss": 0.099, "step": 44620 }, { "epoch": 21.41, "learning_rate": 6.425811958344145e-05, "loss": 0.0956, "step": 44630 }, { "epoch": 21.41, "learning_rate": 6.42443568784248e-05, "loss": 0.0959, "step": 44640 }, { "epoch": 21.41, "learning_rate": 6.423059299875963e-05, "loss": 0.1038, "step": 44650 }, { "epoch": 21.42, "learning_rate": 6.421682794558093e-05, "loss": 0.102, "step": 44660 }, { "epoch": 21.42, "learning_rate": 6.420306172002385e-05, "loss": 0.0889, "step": 44670 }, { "epoch": 21.43, "learning_rate": 6.418929432322358e-05, "loss": 0.0881, "step": 44680 }, { "epoch": 21.43, "learning_rate": 6.417552575631545e-05, "loss": 0.0911, "step": 44690 }, { "epoch": 21.44, "learning_rate": 6.416175602043489e-05, "loss": 0.0933, "step": 44700 }, { "epoch": 21.44, "learning_rate": 6.414798511671739e-05, "loss": 0.0999, "step": 44710 }, { "epoch": 21.45, "learning_rate": 6.413421304629856e-05, "loss": 0.0994, "step": 44720 }, { "epoch": 21.45, "learning_rate": 6.41204398103141e-05, "loss": 0.0999, "step": 44730 }, { "epoch": 21.46, "learning_rate": 6.410666540989983e-05, "loss": 0.0961, "step": 44740 }, { "epoch": 21.46, "learning_rate": 6.40928898461916e-05, "loss": 0.0936, "step": 44750 }, { "epoch": 21.46, "learning_rate": 6.407911312032542e-05, "loss": 0.0871, "step": 44760 }, { "epoch": 21.47, "learning_rate": 6.406533523343741e-05, "loss": 0.093, "step": 44770 }, { "epoch": 21.47, "learning_rate": 6.40515561866637e-05, "loss": 0.0925, "step": 44780 }, { "epoch": 21.48, "learning_rate": 6.403777598114057e-05, "loss": 0.101, "step": 44790 }, { "epoch": 21.48, "learning_rate": 6.402399461800443e-05, "loss": 0.0949, "step": 44800 }, { "epoch": 21.49, "learning_rate": 6.401021209839175e-05, "loss": 0.0883, "step": 44810 }, { "epoch": 21.49, "learning_rate": 6.399642842343906e-05, "loss": 0.0963, "step": 44820 }, { "epoch": 21.5, "learning_rate": 6.398264359428302e-05, "loss": 0.084, "step": 44830 }, { "epoch": 21.5, "learning_rate": 6.396885761206042e-05, "loss": 0.0985, "step": 44840 }, { "epoch": 21.51, "learning_rate": 6.395507047790806e-05, "loss": 0.0993, "step": 44850 }, { "epoch": 21.51, "learning_rate": 6.394128219296292e-05, "loss": 0.081, "step": 44860 }, { "epoch": 21.52, "learning_rate": 6.392749275836203e-05, "loss": 0.099, "step": 44870 }, { "epoch": 21.52, "learning_rate": 6.391370217524252e-05, "loss": 0.0831, "step": 44880 }, { "epoch": 21.52, "learning_rate": 6.389991044474162e-05, "loss": 0.0874, "step": 44890 }, { "epoch": 21.53, "learning_rate": 6.388611756799666e-05, "loss": 0.0932, "step": 44900 }, { "epoch": 21.53, "learning_rate": 6.387232354614504e-05, "loss": 0.0973, "step": 44910 }, { "epoch": 21.54, "learning_rate": 6.385852838032429e-05, "loss": 0.0971, "step": 44920 }, { "epoch": 21.54, "learning_rate": 6.384473207167201e-05, "loss": 0.1054, "step": 44930 }, { "epoch": 21.55, "learning_rate": 6.383093462132588e-05, "loss": 0.097, "step": 44940 }, { "epoch": 21.55, "learning_rate": 6.381713603042374e-05, "loss": 0.1161, "step": 44950 }, { "epoch": 21.56, "learning_rate": 6.380333630010345e-05, "loss": 0.1002, "step": 44960 }, { "epoch": 21.56, "learning_rate": 6.3789535431503e-05, "loss": 0.0957, "step": 44970 }, { "epoch": 21.57, "learning_rate": 6.377573342576044e-05, "loss": 0.1024, "step": 44980 }, { "epoch": 21.57, "learning_rate": 6.376193028401398e-05, "loss": 0.0956, "step": 44990 }, { "epoch": 21.57, "learning_rate": 6.374812600740187e-05, "loss": 0.0916, "step": 45000 }, { "epoch": 21.58, "learning_rate": 6.373432059706247e-05, "loss": 0.0962, "step": 45010 }, { "epoch": 21.58, "learning_rate": 6.372051405413421e-05, "loss": 0.0983, "step": 45020 }, { "epoch": 21.59, "learning_rate": 6.370670637975566e-05, "loss": 0.0971, "step": 45030 }, { "epoch": 21.59, "learning_rate": 6.369289757506547e-05, "loss": 0.1055, "step": 45040 }, { "epoch": 21.6, "learning_rate": 6.367908764120232e-05, "loss": 0.0841, "step": 45050 }, { "epoch": 21.6, "learning_rate": 6.366527657930508e-05, "loss": 0.0926, "step": 45060 }, { "epoch": 21.61, "learning_rate": 6.365146439051263e-05, "loss": 0.0893, "step": 45070 }, { "epoch": 21.61, "learning_rate": 6.363765107596404e-05, "loss": 0.0856, "step": 45080 }, { "epoch": 21.62, "learning_rate": 6.362383663679836e-05, "loss": 0.0941, "step": 45090 }, { "epoch": 21.62, "learning_rate": 6.361002107415477e-05, "loss": 0.0937, "step": 45100 }, { "epoch": 21.62, "learning_rate": 6.359620438917263e-05, "loss": 0.0874, "step": 45110 }, { "epoch": 21.63, "learning_rate": 6.358238658299127e-05, "loss": 0.0924, "step": 45120 }, { "epoch": 21.63, "learning_rate": 6.356856765675016e-05, "loss": 0.0949, "step": 45130 }, { "epoch": 21.64, "learning_rate": 6.355474761158888e-05, "loss": 0.1181, "step": 45140 }, { "epoch": 21.64, "learning_rate": 6.35409264486471e-05, "loss": 0.0953, "step": 45150 }, { "epoch": 21.65, "learning_rate": 6.352710416906456e-05, "loss": 0.1004, "step": 45160 }, { "epoch": 21.65, "learning_rate": 6.351328077398109e-05, "loss": 0.0968, "step": 45170 }, { "epoch": 21.66, "learning_rate": 6.349945626453663e-05, "loss": 0.101, "step": 45180 }, { "epoch": 21.66, "learning_rate": 6.348563064187122e-05, "loss": 0.0962, "step": 45190 }, { "epoch": 21.67, "learning_rate": 6.347180390712497e-05, "loss": 0.0897, "step": 45200 }, { "epoch": 21.67, "learning_rate": 6.345797606143808e-05, "loss": 0.0918, "step": 45210 }, { "epoch": 21.68, "learning_rate": 6.344414710595087e-05, "loss": 0.0961, "step": 45220 }, { "epoch": 21.68, "learning_rate": 6.34303170418037e-05, "loss": 0.1037, "step": 45230 }, { "epoch": 21.68, "learning_rate": 6.341648587013708e-05, "loss": 0.1044, "step": 45240 }, { "epoch": 21.69, "learning_rate": 6.340265359209161e-05, "loss": 0.0953, "step": 45250 }, { "epoch": 21.69, "learning_rate": 6.338882020880792e-05, "loss": 0.111, "step": 45260 }, { "epoch": 21.7, "learning_rate": 6.337498572142677e-05, "loss": 0.0907, "step": 45270 }, { "epoch": 21.7, "learning_rate": 6.3361150131089e-05, "loss": 0.0857, "step": 45280 }, { "epoch": 21.71, "learning_rate": 6.334731343893559e-05, "loss": 0.0989, "step": 45290 }, { "epoch": 21.71, "learning_rate": 6.333347564610753e-05, "loss": 0.1034, "step": 45300 }, { "epoch": 21.72, "learning_rate": 6.331963675374597e-05, "loss": 0.0982, "step": 45310 }, { "epoch": 21.72, "learning_rate": 6.33057967629921e-05, "loss": 0.1048, "step": 45320 }, { "epoch": 21.73, "learning_rate": 6.329195567498724e-05, "loss": 0.1002, "step": 45330 }, { "epoch": 21.73, "learning_rate": 6.327811349087279e-05, "loss": 0.1031, "step": 45340 }, { "epoch": 21.73, "learning_rate": 6.32642702117902e-05, "loss": 0.0983, "step": 45350 }, { "epoch": 21.74, "learning_rate": 6.325042583888106e-05, "loss": 0.1015, "step": 45360 }, { "epoch": 21.74, "learning_rate": 6.323658037328706e-05, "loss": 0.0913, "step": 45370 }, { "epoch": 21.75, "learning_rate": 6.322273381614992e-05, "loss": 0.0894, "step": 45380 }, { "epoch": 21.75, "learning_rate": 6.32088861686115e-05, "loss": 0.1087, "step": 45390 }, { "epoch": 21.76, "learning_rate": 6.319503743181372e-05, "loss": 0.1032, "step": 45400 }, { "epoch": 21.76, "learning_rate": 6.318118760689863e-05, "loss": 0.0974, "step": 45410 }, { "epoch": 21.77, "learning_rate": 6.316733669500833e-05, "loss": 0.0957, "step": 45420 }, { "epoch": 21.77, "learning_rate": 6.3153484697285e-05, "loss": 0.0902, "step": 45430 }, { "epoch": 21.78, "learning_rate": 6.313963161487098e-05, "loss": 0.0893, "step": 45440 }, { "epoch": 21.78, "learning_rate": 6.312577744890864e-05, "loss": 0.1027, "step": 45450 }, { "epoch": 21.78, "learning_rate": 6.31119222005404e-05, "loss": 0.092, "step": 45460 }, { "epoch": 21.79, "learning_rate": 6.309806587090889e-05, "loss": 0.0899, "step": 45470 }, { "epoch": 21.79, "learning_rate": 6.308420846115672e-05, "loss": 0.0959, "step": 45480 }, { "epoch": 21.8, "learning_rate": 6.307034997242665e-05, "loss": 0.0944, "step": 45490 }, { "epoch": 21.8, "learning_rate": 6.30564904058615e-05, "loss": 0.1007, "step": 45500 }, { "epoch": 21.81, "learning_rate": 6.304262976260418e-05, "loss": 0.0978, "step": 45510 }, { "epoch": 21.81, "learning_rate": 6.302876804379769e-05, "loss": 0.0973, "step": 45520 }, { "epoch": 21.82, "learning_rate": 6.301490525058515e-05, "loss": 0.0917, "step": 45530 }, { "epoch": 21.82, "learning_rate": 6.300104138410974e-05, "loss": 0.0988, "step": 45540 }, { "epoch": 21.83, "learning_rate": 6.298717644551469e-05, "loss": 0.1003, "step": 45550 }, { "epoch": 21.83, "learning_rate": 6.297331043594343e-05, "loss": 0.0844, "step": 45560 }, { "epoch": 21.84, "learning_rate": 6.295944335653934e-05, "loss": 0.093, "step": 45570 }, { "epoch": 21.84, "learning_rate": 6.2945575208446e-05, "loss": 0.0937, "step": 45580 }, { "epoch": 21.84, "learning_rate": 6.2931705992807e-05, "loss": 0.1009, "step": 45590 }, { "epoch": 21.85, "learning_rate": 6.291783571076611e-05, "loss": 0.0929, "step": 45600 }, { "epoch": 21.85, "learning_rate": 6.290396436346708e-05, "loss": 0.1057, "step": 45610 }, { "epoch": 21.86, "learning_rate": 6.28900919520538e-05, "loss": 0.0974, "step": 45620 }, { "epoch": 21.86, "learning_rate": 6.28762184776703e-05, "loss": 0.0929, "step": 45630 }, { "epoch": 21.87, "learning_rate": 6.286234394146055e-05, "loss": 0.1052, "step": 45640 }, { "epoch": 21.87, "learning_rate": 6.28484683445688e-05, "loss": 0.097, "step": 45650 }, { "epoch": 21.88, "learning_rate": 6.283459168813921e-05, "loss": 0.0813, "step": 45660 }, { "epoch": 21.88, "learning_rate": 6.282071397331616e-05, "loss": 0.0993, "step": 45670 }, { "epoch": 21.89, "learning_rate": 6.280683520124406e-05, "loss": 0.0992, "step": 45680 }, { "epoch": 21.89, "learning_rate": 6.279295537306738e-05, "loss": 0.1249, "step": 45690 }, { "epoch": 21.89, "learning_rate": 6.277907448993072e-05, "loss": 0.091, "step": 45700 }, { "epoch": 21.9, "learning_rate": 6.276519255297876e-05, "loss": 0.0981, "step": 45710 }, { "epoch": 21.9, "learning_rate": 6.275130956335628e-05, "loss": 0.0959, "step": 45720 }, { "epoch": 21.91, "learning_rate": 6.273742552220807e-05, "loss": 0.0992, "step": 45730 }, { "epoch": 21.91, "learning_rate": 6.272354043067915e-05, "loss": 0.1001, "step": 45740 }, { "epoch": 21.92, "learning_rate": 6.270965428991447e-05, "loss": 0.1003, "step": 45750 }, { "epoch": 21.92, "learning_rate": 6.269576710105917e-05, "loss": 0.0969, "step": 45760 }, { "epoch": 21.93, "learning_rate": 6.268187886525842e-05, "loss": 0.1056, "step": 45770 }, { "epoch": 21.93, "learning_rate": 6.266798958365756e-05, "loss": 0.0991, "step": 45780 }, { "epoch": 21.94, "learning_rate": 6.265409925740191e-05, "loss": 0.1065, "step": 45790 }, { "epoch": 21.94, "learning_rate": 6.264020788763691e-05, "loss": 0.0988, "step": 45800 }, { "epoch": 21.94, "learning_rate": 6.262631547550815e-05, "loss": 0.0972, "step": 45810 }, { "epoch": 21.95, "learning_rate": 6.261242202216121e-05, "loss": 0.1129, "step": 45820 }, { "epoch": 22.0, "learning_rate": 6.259852752874184e-05, "loss": 0.1007, "step": 45830 }, { "epoch": 22.01, "learning_rate": 6.25846319963958e-05, "loss": 0.0896, "step": 45840 }, { "epoch": 22.01, "learning_rate": 6.257073542626899e-05, "loss": 0.1023, "step": 45850 }, { "epoch": 22.02, "learning_rate": 6.255683781950739e-05, "loss": 0.0869, "step": 45860 }, { "epoch": 22.02, "learning_rate": 6.254293917725704e-05, "loss": 0.1061, "step": 45870 }, { "epoch": 22.02, "learning_rate": 6.252903950066409e-05, "loss": 0.0855, "step": 45880 }, { "epoch": 22.03, "learning_rate": 6.251513879087474e-05, "loss": 0.0984, "step": 45890 }, { "epoch": 22.03, "learning_rate": 6.250123704903534e-05, "loss": 0.089, "step": 45900 }, { "epoch": 22.04, "learning_rate": 6.248733427629225e-05, "loss": 0.0962, "step": 45910 }, { "epoch": 22.04, "learning_rate": 6.247343047379194e-05, "loss": 0.0876, "step": 45920 }, { "epoch": 22.05, "learning_rate": 6.245952564268103e-05, "loss": 0.0933, "step": 45930 }, { "epoch": 22.05, "learning_rate": 6.244561978410613e-05, "loss": 0.0971, "step": 45940 }, { "epoch": 22.06, "learning_rate": 6.243171289921399e-05, "loss": 0.0957, "step": 45950 }, { "epoch": 22.06, "learning_rate": 6.24178049891514e-05, "loss": 0.0887, "step": 45960 }, { "epoch": 22.07, "learning_rate": 6.24038960550653e-05, "loss": 0.0862, "step": 45970 }, { "epoch": 22.07, "learning_rate": 6.238998609810266e-05, "loss": 0.0987, "step": 45980 }, { "epoch": 22.07, "learning_rate": 6.237607511941056e-05, "loss": 0.0955, "step": 45990 }, { "epoch": 22.08, "learning_rate": 6.236216312013614e-05, "loss": 0.099, "step": 46000 }, { "epoch": 22.08, "learning_rate": 6.234825010142668e-05, "loss": 0.0918, "step": 46010 }, { "epoch": 22.09, "learning_rate": 6.233433606442945e-05, "loss": 0.0869, "step": 46020 }, { "epoch": 22.09, "learning_rate": 6.23204210102919e-05, "loss": 0.0956, "step": 46030 }, { "epoch": 22.1, "learning_rate": 6.23065049401615e-05, "loss": 0.0971, "step": 46040 }, { "epoch": 22.1, "learning_rate": 6.229258785518585e-05, "loss": 0.1017, "step": 46050 }, { "epoch": 22.11, "learning_rate": 6.227866975651261e-05, "loss": 0.0923, "step": 46060 }, { "epoch": 22.11, "learning_rate": 6.226475064528948e-05, "loss": 0.103, "step": 46070 }, { "epoch": 22.12, "learning_rate": 6.225083052266435e-05, "loss": 0.0965, "step": 46080 }, { "epoch": 22.12, "learning_rate": 6.223690938978508e-05, "loss": 0.0951, "step": 46090 }, { "epoch": 22.13, "learning_rate": 6.222298724779971e-05, "loss": 0.0927, "step": 46100 }, { "epoch": 22.13, "learning_rate": 6.220906409785625e-05, "loss": 0.1066, "step": 46110 }, { "epoch": 22.13, "learning_rate": 6.219513994110293e-05, "loss": 0.092, "step": 46120 }, { "epoch": 22.14, "learning_rate": 6.218121477868795e-05, "loss": 0.1064, "step": 46130 }, { "epoch": 22.14, "learning_rate": 6.216728861175967e-05, "loss": 0.0943, "step": 46140 }, { "epoch": 22.15, "learning_rate": 6.215336144146646e-05, "loss": 0.0874, "step": 46150 }, { "epoch": 22.15, "learning_rate": 6.213943326895685e-05, "loss": 0.1068, "step": 46160 }, { "epoch": 22.16, "learning_rate": 6.212550409537939e-05, "loss": 0.0916, "step": 46170 }, { "epoch": 22.16, "learning_rate": 6.211157392188274e-05, "loss": 0.0971, "step": 46180 }, { "epoch": 22.17, "learning_rate": 6.209764274961566e-05, "loss": 0.0974, "step": 46190 }, { "epoch": 22.17, "learning_rate": 6.208371057972694e-05, "loss": 0.0958, "step": 46200 }, { "epoch": 22.18, "learning_rate": 6.206977741336553e-05, "loss": 0.0939, "step": 46210 }, { "epoch": 22.18, "learning_rate": 6.205584325168036e-05, "loss": 0.0944, "step": 46220 }, { "epoch": 22.18, "learning_rate": 6.204190809582053e-05, "loss": 0.1, "step": 46230 }, { "epoch": 22.19, "learning_rate": 6.202797194693519e-05, "loss": 0.0878, "step": 46240 }, { "epoch": 22.19, "learning_rate": 6.201403480617357e-05, "loss": 0.0887, "step": 46250 }, { "epoch": 22.2, "learning_rate": 6.200009667468498e-05, "loss": 0.0807, "step": 46260 }, { "epoch": 22.2, "learning_rate": 6.198615755361883e-05, "loss": 0.096, "step": 46270 }, { "epoch": 22.21, "learning_rate": 6.197221744412458e-05, "loss": 0.0962, "step": 46280 }, { "epoch": 22.21, "learning_rate": 6.19582763473518e-05, "loss": 0.0907, "step": 46290 }, { "epoch": 22.22, "learning_rate": 6.19443342644501e-05, "loss": 0.1013, "step": 46300 }, { "epoch": 22.22, "learning_rate": 6.193039119656926e-05, "loss": 0.0913, "step": 46310 }, { "epoch": 22.23, "learning_rate": 6.191644714485904e-05, "loss": 0.1091, "step": 46320 }, { "epoch": 22.23, "learning_rate": 6.190250211046934e-05, "loss": 0.0923, "step": 46330 }, { "epoch": 22.23, "learning_rate": 6.188855609455012e-05, "loss": 0.0937, "step": 46340 }, { "epoch": 22.24, "learning_rate": 6.187460909825142e-05, "loss": 0.0976, "step": 46350 }, { "epoch": 22.24, "learning_rate": 6.186066112272337e-05, "loss": 0.0999, "step": 46360 }, { "epoch": 22.25, "learning_rate": 6.184671216911619e-05, "loss": 0.0822, "step": 46370 }, { "epoch": 22.25, "learning_rate": 6.183276223858016e-05, "loss": 0.1066, "step": 46380 }, { "epoch": 22.26, "learning_rate": 6.181881133226565e-05, "loss": 0.0957, "step": 46390 }, { "epoch": 22.26, "learning_rate": 6.18048594513231e-05, "loss": 0.0888, "step": 46400 }, { "epoch": 22.27, "learning_rate": 6.179090659690305e-05, "loss": 0.0917, "step": 46410 }, { "epoch": 22.27, "learning_rate": 6.177695277015612e-05, "loss": 0.1, "step": 46420 }, { "epoch": 22.28, "learning_rate": 6.176299797223296e-05, "loss": 0.1033, "step": 46430 }, { "epoch": 22.28, "learning_rate": 6.17490422042844e-05, "loss": 0.0847, "step": 46440 }, { "epoch": 22.29, "learning_rate": 6.173508546746122e-05, "loss": 0.1042, "step": 46450 }, { "epoch": 22.29, "learning_rate": 6.172112776291444e-05, "loss": 0.1014, "step": 46460 }, { "epoch": 22.29, "learning_rate": 6.170716909179498e-05, "loss": 0.0948, "step": 46470 }, { "epoch": 22.3, "learning_rate": 6.169320945525398e-05, "loss": 0.0901, "step": 46480 }, { "epoch": 22.3, "learning_rate": 6.167924885444259e-05, "loss": 0.0929, "step": 46490 }, { "epoch": 22.31, "learning_rate": 6.166528729051209e-05, "loss": 0.0919, "step": 46500 }, { "epoch": 22.31, "learning_rate": 6.165132476461377e-05, "loss": 0.093, "step": 46510 }, { "epoch": 22.32, "learning_rate": 6.163736127789906e-05, "loss": 0.1125, "step": 46520 }, { "epoch": 22.32, "learning_rate": 6.162339683151942e-05, "loss": 0.094, "step": 46530 }, { "epoch": 22.33, "learning_rate": 6.160943142662644e-05, "loss": 0.0727, "step": 46540 }, { "epoch": 22.33, "learning_rate": 6.159546506437177e-05, "loss": 0.0872, "step": 46550 }, { "epoch": 22.34, "learning_rate": 6.158149774590712e-05, "loss": 0.1055, "step": 46560 }, { "epoch": 22.34, "learning_rate": 6.15675294723843e-05, "loss": 0.0866, "step": 46570 }, { "epoch": 22.34, "learning_rate": 6.15535602449552e-05, "loss": 0.0895, "step": 46580 }, { "epoch": 22.35, "learning_rate": 6.153959006477175e-05, "loss": 0.0881, "step": 46590 }, { "epoch": 22.35, "learning_rate": 6.1525618932986e-05, "loss": 0.1011, "step": 46600 }, { "epoch": 22.36, "learning_rate": 6.151164685075009e-05, "loss": 0.096, "step": 46610 }, { "epoch": 22.36, "learning_rate": 6.14976738192162e-05, "loss": 0.0932, "step": 46620 }, { "epoch": 22.37, "learning_rate": 6.148369983953662e-05, "loss": 0.0919, "step": 46630 }, { "epoch": 22.37, "learning_rate": 6.146972491286365e-05, "loss": 0.0828, "step": 46640 }, { "epoch": 22.38, "learning_rate": 6.145574904034979e-05, "loss": 0.0981, "step": 46650 }, { "epoch": 22.38, "learning_rate": 6.144177222314751e-05, "loss": 0.0935, "step": 46660 }, { "epoch": 22.39, "learning_rate": 6.142779446240939e-05, "loss": 0.0905, "step": 46670 }, { "epoch": 22.39, "learning_rate": 6.141381575928811e-05, "loss": 0.0884, "step": 46680 }, { "epoch": 22.39, "learning_rate": 6.13998361149364e-05, "loss": 0.096, "step": 46690 }, { "epoch": 22.4, "learning_rate": 6.13858555305071e-05, "loss": 0.0932, "step": 46700 }, { "epoch": 22.4, "learning_rate": 6.137187400715307e-05, "loss": 0.0902, "step": 46710 }, { "epoch": 22.41, "learning_rate": 6.135789154602733e-05, "loss": 0.0985, "step": 46720 }, { "epoch": 22.41, "learning_rate": 6.13439081482829e-05, "loss": 0.1041, "step": 46730 }, { "epoch": 22.42, "learning_rate": 6.132992381507292e-05, "loss": 0.0924, "step": 46740 }, { "epoch": 22.42, "learning_rate": 6.13159385475506e-05, "loss": 0.1005, "step": 46750 }, { "epoch": 22.43, "learning_rate": 6.130195234686921e-05, "loss": 0.0893, "step": 46760 }, { "epoch": 22.43, "learning_rate": 6.128796521418211e-05, "loss": 0.0895, "step": 46770 }, { "epoch": 22.44, "learning_rate": 6.127397715064274e-05, "loss": 0.0992, "step": 46780 }, { "epoch": 22.44, "learning_rate": 6.125998815740461e-05, "loss": 0.1107, "step": 46790 }, { "epoch": 22.45, "learning_rate": 6.124599823562135e-05, "loss": 0.0898, "step": 46800 }, { "epoch": 22.45, "learning_rate": 6.123200738644657e-05, "loss": 0.1089, "step": 46810 }, { "epoch": 22.45, "learning_rate": 6.121801561103404e-05, "loss": 0.0924, "step": 46820 }, { "epoch": 22.46, "learning_rate": 6.120402291053757e-05, "loss": 0.1032, "step": 46830 }, { "epoch": 22.46, "learning_rate": 6.119002928611107e-05, "loss": 0.0956, "step": 46840 }, { "epoch": 22.47, "learning_rate": 6.117603473890851e-05, "loss": 0.112, "step": 46850 }, { "epoch": 22.47, "learning_rate": 6.116203927008392e-05, "loss": 0.0943, "step": 46860 }, { "epoch": 22.48, "learning_rate": 6.114804288079145e-05, "loss": 0.1032, "step": 46870 }, { "epoch": 22.48, "learning_rate": 6.113404557218527e-05, "loss": 0.1137, "step": 46880 }, { "epoch": 22.49, "learning_rate": 6.11200473454197e-05, "loss": 0.0903, "step": 46890 }, { "epoch": 22.49, "learning_rate": 6.110604820164903e-05, "loss": 0.0873, "step": 46900 }, { "epoch": 22.5, "learning_rate": 6.109204814202774e-05, "loss": 0.1071, "step": 46910 }, { "epoch": 22.5, "learning_rate": 6.10780471677103e-05, "loss": 0.0983, "step": 46920 }, { "epoch": 22.5, "learning_rate": 6.106404527985133e-05, "loss": 0.1022, "step": 46930 }, { "epoch": 22.51, "learning_rate": 6.105004247960544e-05, "loss": 0.089, "step": 46940 }, { "epoch": 22.51, "learning_rate": 6.103603876812737e-05, "loss": 0.0872, "step": 46950 }, { "epoch": 22.52, "learning_rate": 6.102203414657195e-05, "loss": 0.0997, "step": 46960 }, { "epoch": 22.52, "learning_rate": 6.1008028616094015e-05, "loss": 0.0855, "step": 46970 }, { "epoch": 22.53, "learning_rate": 6.099402217784854e-05, "loss": 0.0931, "step": 46980 }, { "epoch": 22.53, "learning_rate": 6.098001483299057e-05, "loss": 0.106, "step": 46990 }, { "epoch": 22.54, "learning_rate": 6.096600658267519e-05, "loss": 0.0955, "step": 47000 }, { "epoch": 22.54, "learning_rate": 6.095199742805758e-05, "loss": 0.0851, "step": 47010 }, { "epoch": 22.55, "learning_rate": 6.0937987370293005e-05, "loss": 0.0967, "step": 47020 }, { "epoch": 22.55, "learning_rate": 6.0923976410536774e-05, "loss": 0.0986, "step": 47030 }, { "epoch": 22.55, "learning_rate": 6.09099645499443e-05, "loss": 0.0961, "step": 47040 }, { "epoch": 22.56, "learning_rate": 6.089595178967104e-05, "loss": 0.1072, "step": 47050 }, { "epoch": 22.56, "learning_rate": 6.088193813087258e-05, "loss": 0.102, "step": 47060 }, { "epoch": 22.57, "learning_rate": 6.086792357470452e-05, "loss": 0.1067, "step": 47070 }, { "epoch": 22.57, "learning_rate": 6.085390812232257e-05, "loss": 0.0919, "step": 47080 }, { "epoch": 22.58, "learning_rate": 6.083989177488247e-05, "loss": 0.09, "step": 47090 }, { "epoch": 22.58, "learning_rate": 6.0825874533540116e-05, "loss": 0.0862, "step": 47100 }, { "epoch": 22.59, "learning_rate": 6.081185639945139e-05, "loss": 0.0948, "step": 47110 }, { "epoch": 22.59, "learning_rate": 6.079783737377231e-05, "loss": 0.0893, "step": 47120 }, { "epoch": 22.6, "learning_rate": 6.078381745765891e-05, "loss": 0.09, "step": 47130 }, { "epoch": 22.6, "learning_rate": 6.0769796652267375e-05, "loss": 0.0936, "step": 47140 }, { "epoch": 22.61, "learning_rate": 6.0755774958753877e-05, "loss": 0.0986, "step": 47150 }, { "epoch": 22.61, "learning_rate": 6.074175237827473e-05, "loss": 0.0987, "step": 47160 }, { "epoch": 22.61, "learning_rate": 6.072772891198626e-05, "loss": 0.09, "step": 47170 }, { "epoch": 22.62, "learning_rate": 6.071370456104495e-05, "loss": 0.089, "step": 47180 }, { "epoch": 22.62, "learning_rate": 6.0699679326607285e-05, "loss": 0.0939, "step": 47190 }, { "epoch": 22.63, "learning_rate": 6.068565320982982e-05, "loss": 0.1021, "step": 47200 }, { "epoch": 22.63, "learning_rate": 6.067162621186923e-05, "loss": 0.1036, "step": 47210 }, { "epoch": 22.64, "learning_rate": 6.065759833388223e-05, "loss": 0.1029, "step": 47220 }, { "epoch": 22.64, "learning_rate": 6.0643569577025635e-05, "loss": 0.1012, "step": 47230 }, { "epoch": 22.65, "learning_rate": 6.062953994245629e-05, "loss": 0.1003, "step": 47240 }, { "epoch": 22.65, "learning_rate": 6.061550943133114e-05, "loss": 0.0935, "step": 47250 }, { "epoch": 22.66, "learning_rate": 6.0601478044807215e-05, "loss": 0.0902, "step": 47260 }, { "epoch": 22.66, "learning_rate": 6.0587445784041606e-05, "loss": 0.0983, "step": 47270 }, { "epoch": 22.66, "learning_rate": 6.057341265019144e-05, "loss": 0.1023, "step": 47280 }, { "epoch": 22.67, "learning_rate": 6.055937864441397e-05, "loss": 0.0884, "step": 47290 }, { "epoch": 22.67, "learning_rate": 6.054534376786649e-05, "loss": 0.0947, "step": 47300 }, { "epoch": 22.68, "learning_rate": 6.053130802170638e-05, "loss": 0.0958, "step": 47310 }, { "epoch": 22.68, "learning_rate": 6.051727140709108e-05, "loss": 0.1009, "step": 47320 }, { "epoch": 22.69, "learning_rate": 6.0503233925178104e-05, "loss": 0.0886, "step": 47330 }, { "epoch": 22.69, "learning_rate": 6.048919557712507e-05, "loss": 0.0967, "step": 47340 }, { "epoch": 22.7, "learning_rate": 6.047515636408959e-05, "loss": 0.0944, "step": 47350 }, { "epoch": 22.7, "learning_rate": 6.046111628722942e-05, "loss": 0.0989, "step": 47360 }, { "epoch": 22.71, "learning_rate": 6.044707534770239e-05, "loss": 0.0953, "step": 47370 }, { "epoch": 22.71, "learning_rate": 6.0433033546666315e-05, "loss": 0.0902, "step": 47380 }, { "epoch": 22.71, "learning_rate": 6.041899088527919e-05, "loss": 0.0888, "step": 47390 }, { "epoch": 22.72, "learning_rate": 6.0404947364699004e-05, "loss": 0.0995, "step": 47400 }, { "epoch": 22.72, "learning_rate": 6.0390902986083865e-05, "loss": 0.0893, "step": 47410 }, { "epoch": 22.73, "learning_rate": 6.037685775059191e-05, "loss": 0.0888, "step": 47420 }, { "epoch": 22.73, "learning_rate": 6.036281165938137e-05, "loss": 0.0871, "step": 47430 }, { "epoch": 22.74, "learning_rate": 6.034876471361055e-05, "loss": 0.0888, "step": 47440 }, { "epoch": 22.74, "learning_rate": 6.033471691443783e-05, "loss": 0.0979, "step": 47450 }, { "epoch": 22.75, "learning_rate": 6.032066826302163e-05, "loss": 0.0934, "step": 47460 }, { "epoch": 22.75, "learning_rate": 6.0306618760520463e-05, "loss": 0.0909, "step": 47470 }, { "epoch": 22.76, "learning_rate": 6.029256840809292e-05, "loss": 0.1025, "step": 47480 }, { "epoch": 22.76, "learning_rate": 6.027851720689763e-05, "loss": 0.0935, "step": 47490 }, { "epoch": 22.77, "learning_rate": 6.026446515809335e-05, "loss": 0.1038, "step": 47500 }, { "epoch": 22.77, "learning_rate": 6.025041226283883e-05, "loss": 0.0931, "step": 47510 }, { "epoch": 22.77, "learning_rate": 6.0236358522292967e-05, "loss": 0.0954, "step": 47520 }, { "epoch": 22.78, "learning_rate": 6.022230393761465e-05, "loss": 0.0879, "step": 47530 }, { "epoch": 22.78, "learning_rate": 6.020824850996292e-05, "loss": 0.1087, "step": 47540 }, { "epoch": 22.79, "learning_rate": 6.0194192240496816e-05, "loss": 0.0878, "step": 47550 }, { "epoch": 22.79, "learning_rate": 6.018013513037548e-05, "loss": 0.1101, "step": 47560 }, { "epoch": 22.8, "learning_rate": 6.016607718075813e-05, "loss": 0.0916, "step": 47570 }, { "epoch": 22.8, "learning_rate": 6.015201839280403e-05, "loss": 0.1066, "step": 47580 }, { "epoch": 22.81, "learning_rate": 6.0137958767672544e-05, "loss": 0.1119, "step": 47590 }, { "epoch": 22.81, "learning_rate": 6.0123898306523067e-05, "loss": 0.0951, "step": 47600 }, { "epoch": 22.82, "learning_rate": 6.01098370105151e-05, "loss": 0.0953, "step": 47610 }, { "epoch": 22.82, "learning_rate": 6.009577488080817e-05, "loss": 0.0978, "step": 47620 }, { "epoch": 22.82, "learning_rate": 6.008171191856193e-05, "loss": 0.0698, "step": 47630 }, { "epoch": 22.83, "learning_rate": 6.006764812493606e-05, "loss": 0.0872, "step": 47640 }, { "epoch": 22.83, "learning_rate": 6.00535835010903e-05, "loss": 0.0897, "step": 47650 }, { "epoch": 22.84, "learning_rate": 6.0039518048184495e-05, "loss": 0.0986, "step": 47660 }, { "epoch": 22.84, "learning_rate": 6.002545176737854e-05, "loss": 0.0931, "step": 47670 }, { "epoch": 22.85, "learning_rate": 6.001138465983241e-05, "loss": 0.0966, "step": 47680 }, { "epoch": 22.85, "learning_rate": 5.999731672670609e-05, "loss": 0.0954, "step": 47690 }, { "epoch": 22.86, "learning_rate": 5.998324796915973e-05, "loss": 0.0895, "step": 47700 }, { "epoch": 22.86, "learning_rate": 5.9969178388353474e-05, "loss": 0.0865, "step": 47710 }, { "epoch": 22.87, "learning_rate": 5.995510798544757e-05, "loss": 0.0986, "step": 47720 }, { "epoch": 22.87, "learning_rate": 5.994103676160231e-05, "loss": 0.0858, "step": 47730 }, { "epoch": 22.87, "learning_rate": 5.9926964717978064e-05, "loss": 0.0889, "step": 47740 }, { "epoch": 22.88, "learning_rate": 5.9912891855735285e-05, "loss": 0.0981, "step": 47750 }, { "epoch": 22.88, "learning_rate": 5.989881817603448e-05, "loss": 0.091, "step": 47760 }, { "epoch": 22.89, "learning_rate": 5.988474368003619e-05, "loss": 0.0896, "step": 47770 }, { "epoch": 22.89, "learning_rate": 5.987066836890109e-05, "loss": 0.0937, "step": 47780 }, { "epoch": 22.9, "learning_rate": 5.98565922437899e-05, "loss": 0.1055, "step": 47790 }, { "epoch": 22.9, "learning_rate": 5.984251530586337e-05, "loss": 0.0887, "step": 47800 }, { "epoch": 22.91, "learning_rate": 5.9828437556282314e-05, "loss": 0.0868, "step": 47810 }, { "epoch": 22.91, "learning_rate": 5.981435899620771e-05, "loss": 0.087, "step": 47820 }, { "epoch": 22.92, "learning_rate": 5.9800279626800485e-05, "loss": 0.0943, "step": 47830 }, { "epoch": 22.92, "learning_rate": 5.978619944922169e-05, "loss": 0.0854, "step": 47840 }, { "epoch": 22.93, "learning_rate": 5.9772118464632435e-05, "loss": 0.0926, "step": 47850 }, { "epoch": 22.93, "learning_rate": 5.975803667419392e-05, "loss": 0.0913, "step": 47860 }, { "epoch": 22.93, "learning_rate": 5.974395407906736e-05, "loss": 0.0847, "step": 47870 }, { "epoch": 22.94, "learning_rate": 5.972987068041408e-05, "loss": 0.1088, "step": 47880 }, { "epoch": 22.94, "learning_rate": 5.971578647939543e-05, "loss": 0.0922, "step": 47890 }, { "epoch": 22.95, "learning_rate": 5.970170147717288e-05, "loss": 0.0875, "step": 47900 }, { "epoch": 23.0, "learning_rate": 5.968761567490794e-05, "loss": 0.1198, "step": 47910 }, { "epoch": 23.01, "learning_rate": 5.9673529073762146e-05, "loss": 0.0979, "step": 47920 }, { "epoch": 23.01, "learning_rate": 5.9659441674897176e-05, "loss": 0.0886, "step": 47930 }, { "epoch": 23.01, "learning_rate": 5.964535347947471e-05, "loss": 0.0973, "step": 47940 }, { "epoch": 23.02, "learning_rate": 5.963126448865655e-05, "loss": 0.098, "step": 47950 }, { "epoch": 23.02, "learning_rate": 5.96171747036045e-05, "loss": 0.0914, "step": 47960 }, { "epoch": 23.03, "learning_rate": 5.9603084125480465e-05, "loss": 0.1043, "step": 47970 }, { "epoch": 23.03, "learning_rate": 5.958899275544644e-05, "loss": 0.1055, "step": 47980 }, { "epoch": 23.04, "learning_rate": 5.957490059466443e-05, "loss": 0.0938, "step": 47990 }, { "epoch": 23.04, "learning_rate": 5.956080764429653e-05, "loss": 0.103, "step": 48000 }, { "epoch": 23.05, "learning_rate": 5.9546713905504925e-05, "loss": 0.0999, "step": 48010 }, { "epoch": 23.05, "learning_rate": 5.9532619379451846e-05, "loss": 0.0923, "step": 48020 }, { "epoch": 23.06, "learning_rate": 5.9518524067299554e-05, "loss": 0.0918, "step": 48030 }, { "epoch": 23.06, "learning_rate": 5.950442797021043e-05, "loss": 0.0888, "step": 48040 }, { "epoch": 23.06, "learning_rate": 5.94903310893469e-05, "loss": 0.0865, "step": 48050 }, { "epoch": 23.07, "learning_rate": 5.9476233425871444e-05, "loss": 0.0996, "step": 48060 }, { "epoch": 23.07, "learning_rate": 5.9462134980946594e-05, "loss": 0.0905, "step": 48070 }, { "epoch": 23.08, "learning_rate": 5.9448035755735e-05, "loss": 0.1002, "step": 48080 }, { "epoch": 23.08, "learning_rate": 5.9433935751399325e-05, "loss": 0.0966, "step": 48090 }, { "epoch": 23.09, "learning_rate": 5.941983496910232e-05, "loss": 0.0818, "step": 48100 }, { "epoch": 23.09, "learning_rate": 5.940573341000677e-05, "loss": 0.0983, "step": 48110 }, { "epoch": 23.1, "learning_rate": 5.9391631075275575e-05, "loss": 0.0919, "step": 48120 }, { "epoch": 23.1, "learning_rate": 5.937752796607168e-05, "loss": 0.0919, "step": 48130 }, { "epoch": 23.11, "learning_rate": 5.936342408355806e-05, "loss": 0.0909, "step": 48140 }, { "epoch": 23.11, "learning_rate": 5.934931942889778e-05, "loss": 0.1008, "step": 48150 }, { "epoch": 23.11, "learning_rate": 5.9335214003253995e-05, "loss": 0.0978, "step": 48160 }, { "epoch": 23.12, "learning_rate": 5.9321107807789866e-05, "loss": 0.0891, "step": 48170 }, { "epoch": 23.12, "learning_rate": 5.930700084366867e-05, "loss": 0.092, "step": 48180 }, { "epoch": 23.13, "learning_rate": 5.9292893112053694e-05, "loss": 0.0886, "step": 48190 }, { "epoch": 23.13, "learning_rate": 5.9278784614108375e-05, "loss": 0.091, "step": 48200 }, { "epoch": 23.14, "learning_rate": 5.92646753509961e-05, "loss": 0.0974, "step": 48210 }, { "epoch": 23.14, "learning_rate": 5.9250565323880424e-05, "loss": 0.0836, "step": 48220 }, { "epoch": 23.15, "learning_rate": 5.923645453392487e-05, "loss": 0.08, "step": 48230 }, { "epoch": 23.15, "learning_rate": 5.922234298229312e-05, "loss": 0.0849, "step": 48240 }, { "epoch": 23.16, "learning_rate": 5.920823067014885e-05, "loss": 0.0972, "step": 48250 }, { "epoch": 23.16, "learning_rate": 5.9194117598655805e-05, "loss": 0.095, "step": 48260 }, { "epoch": 23.16, "learning_rate": 5.918000376897783e-05, "loss": 0.0973, "step": 48270 }, { "epoch": 23.17, "learning_rate": 5.91658891822788e-05, "loss": 0.0963, "step": 48280 }, { "epoch": 23.17, "learning_rate": 5.915177383972268e-05, "loss": 0.0872, "step": 48290 }, { "epoch": 23.18, "learning_rate": 5.913765774247343e-05, "loss": 0.0997, "step": 48300 }, { "epoch": 23.18, "learning_rate": 5.9123540891695185e-05, "loss": 0.0909, "step": 48310 }, { "epoch": 23.19, "learning_rate": 5.910942328855202e-05, "loss": 0.1009, "step": 48320 }, { "epoch": 23.19, "learning_rate": 5.909530493420818e-05, "loss": 0.0965, "step": 48330 }, { "epoch": 23.2, "learning_rate": 5.9081185829827904e-05, "loss": 0.0952, "step": 48340 }, { "epoch": 23.2, "learning_rate": 5.9067065976575496e-05, "loss": 0.0948, "step": 48350 }, { "epoch": 23.21, "learning_rate": 5.905294537561537e-05, "loss": 0.0952, "step": 48360 }, { "epoch": 23.21, "learning_rate": 5.903882402811194e-05, "loss": 0.0979, "step": 48370 }, { "epoch": 23.22, "learning_rate": 5.902470193522972e-05, "loss": 0.0908, "step": 48380 }, { "epoch": 23.22, "learning_rate": 5.9010579098133276e-05, "loss": 0.0932, "step": 48390 }, { "epoch": 23.22, "learning_rate": 5.899645551798725e-05, "loss": 0.0953, "step": 48400 }, { "epoch": 23.23, "learning_rate": 5.89823311959563e-05, "loss": 0.0848, "step": 48410 }, { "epoch": 23.23, "learning_rate": 5.896820613320519e-05, "loss": 0.0938, "step": 48420 }, { "epoch": 23.24, "learning_rate": 5.8954080330898764e-05, "loss": 0.0883, "step": 48430 }, { "epoch": 23.24, "learning_rate": 5.893995379020184e-05, "loss": 0.0805, "step": 48440 }, { "epoch": 23.25, "learning_rate": 5.892582651227938e-05, "loss": 0.0882, "step": 48450 }, { "epoch": 23.25, "learning_rate": 5.891169849829637e-05, "loss": 0.0869, "step": 48460 }, { "epoch": 23.26, "learning_rate": 5.8897569749417894e-05, "loss": 0.085, "step": 48470 }, { "epoch": 23.26, "learning_rate": 5.888344026680901e-05, "loss": 0.0877, "step": 48480 }, { "epoch": 23.27, "learning_rate": 5.886931005163493e-05, "loss": 0.0979, "step": 48490 }, { "epoch": 23.27, "learning_rate": 5.88551791050609e-05, "loss": 0.1028, "step": 48500 }, { "epoch": 23.27, "learning_rate": 5.8841047428252184e-05, "loss": 0.089, "step": 48510 }, { "epoch": 23.28, "learning_rate": 5.882691502237417e-05, "loss": 0.0889, "step": 48520 }, { "epoch": 23.28, "learning_rate": 5.881278188859224e-05, "loss": 0.0907, "step": 48530 }, { "epoch": 23.29, "learning_rate": 5.879864802807191e-05, "loss": 0.0809, "step": 48540 }, { "epoch": 23.29, "learning_rate": 5.878451344197868e-05, "loss": 0.0935, "step": 48550 }, { "epoch": 23.3, "learning_rate": 5.877037813147819e-05, "loss": 0.0946, "step": 48560 }, { "epoch": 23.3, "learning_rate": 5.8756242097736035e-05, "loss": 0.0891, "step": 48570 }, { "epoch": 23.31, "learning_rate": 5.874210534191798e-05, "loss": 0.098, "step": 48580 }, { "epoch": 23.31, "learning_rate": 5.8727967865189804e-05, "loss": 0.0993, "step": 48590 }, { "epoch": 23.32, "learning_rate": 5.871382966871729e-05, "loss": 0.0952, "step": 48600 }, { "epoch": 23.32, "learning_rate": 5.869969075366639e-05, "loss": 0.102, "step": 48610 }, { "epoch": 23.32, "learning_rate": 5.868555112120302e-05, "loss": 0.0917, "step": 48620 }, { "epoch": 23.33, "learning_rate": 5.867141077249323e-05, "loss": 0.0968, "step": 48630 }, { "epoch": 23.33, "learning_rate": 5.865726970870302e-05, "loss": 0.0887, "step": 48640 }, { "epoch": 23.34, "learning_rate": 5.8643127930998607e-05, "loss": 0.0904, "step": 48650 }, { "epoch": 23.34, "learning_rate": 5.862898544054612e-05, "loss": 0.0923, "step": 48660 }, { "epoch": 23.35, "learning_rate": 5.8614842238511855e-05, "loss": 0.0833, "step": 48670 }, { "epoch": 23.35, "learning_rate": 5.860069832606208e-05, "loss": 0.0912, "step": 48680 }, { "epoch": 23.36, "learning_rate": 5.858655370436315e-05, "loss": 0.0901, "step": 48690 }, { "epoch": 23.36, "learning_rate": 5.857240837458155e-05, "loss": 0.0929, "step": 48700 }, { "epoch": 23.37, "learning_rate": 5.8558262337883686e-05, "loss": 0.0896, "step": 48710 }, { "epoch": 23.37, "learning_rate": 5.854411559543616e-05, "loss": 0.0869, "step": 48720 }, { "epoch": 23.38, "learning_rate": 5.8529968148405535e-05, "loss": 0.1071, "step": 48730 }, { "epoch": 23.38, "learning_rate": 5.851581999795849e-05, "loss": 0.0929, "step": 48740 }, { "epoch": 23.38, "learning_rate": 5.8501671145261715e-05, "loss": 0.0977, "step": 48750 }, { "epoch": 23.39, "learning_rate": 5.8487521591482005e-05, "loss": 0.0922, "step": 48760 }, { "epoch": 23.39, "learning_rate": 5.84733713377862e-05, "loss": 0.1065, "step": 48770 }, { "epoch": 23.4, "learning_rate": 5.8459220385341165e-05, "loss": 0.099, "step": 48780 }, { "epoch": 23.4, "learning_rate": 5.8445068735313834e-05, "loss": 0.0994, "step": 48790 }, { "epoch": 23.41, "learning_rate": 5.843091638887125e-05, "loss": 0.0896, "step": 48800 }, { "epoch": 23.41, "learning_rate": 5.8416763347180446e-05, "loss": 0.0977, "step": 48810 }, { "epoch": 23.42, "learning_rate": 5.8402609611408554e-05, "loss": 0.0944, "step": 48820 }, { "epoch": 23.42, "learning_rate": 5.8388455182722735e-05, "loss": 0.0944, "step": 48830 }, { "epoch": 23.43, "learning_rate": 5.8374300062290235e-05, "loss": 0.1005, "step": 48840 }, { "epoch": 23.43, "learning_rate": 5.8360144251278345e-05, "loss": 0.0924, "step": 48850 }, { "epoch": 23.43, "learning_rate": 5.83459877508544e-05, "loss": 0.0949, "step": 48860 }, { "epoch": 23.44, "learning_rate": 5.833183056218581e-05, "loss": 0.0976, "step": 48870 }, { "epoch": 23.44, "learning_rate": 5.8317672686440036e-05, "loss": 0.087, "step": 48880 }, { "epoch": 23.45, "learning_rate": 5.830351412478459e-05, "loss": 0.0836, "step": 48890 }, { "epoch": 23.45, "learning_rate": 5.828935487838705e-05, "loss": 0.0873, "step": 48900 }, { "epoch": 23.46, "learning_rate": 5.827519494841504e-05, "loss": 0.1089, "step": 48910 }, { "epoch": 23.46, "learning_rate": 5.826103433603627e-05, "loss": 0.0859, "step": 48920 }, { "epoch": 23.47, "learning_rate": 5.824687304241845e-05, "loss": 0.094, "step": 48930 }, { "epoch": 23.47, "learning_rate": 5.8232711068729384e-05, "loss": 0.0912, "step": 48940 }, { "epoch": 23.48, "learning_rate": 5.8218548416136956e-05, "loss": 0.0835, "step": 48950 }, { "epoch": 23.48, "learning_rate": 5.8204385085809046e-05, "loss": 0.1002, "step": 48960 }, { "epoch": 23.48, "learning_rate": 5.819022107891363e-05, "loss": 0.0861, "step": 48970 }, { "epoch": 23.49, "learning_rate": 5.8176056396618725e-05, "loss": 0.0894, "step": 48980 }, { "epoch": 23.49, "learning_rate": 5.816189104009243e-05, "loss": 0.0834, "step": 48990 }, { "epoch": 23.5, "learning_rate": 5.8147725010502864e-05, "loss": 0.0966, "step": 49000 }, { "epoch": 23.5, "learning_rate": 5.813355830901822e-05, "loss": 0.0842, "step": 49010 }, { "epoch": 23.51, "learning_rate": 5.8119390936806714e-05, "loss": 0.0917, "step": 49020 }, { "epoch": 23.51, "learning_rate": 5.8105222895036704e-05, "loss": 0.0967, "step": 49030 }, { "epoch": 23.52, "learning_rate": 5.809105418487651e-05, "loss": 0.0963, "step": 49040 }, { "epoch": 23.52, "learning_rate": 5.8076884807494536e-05, "loss": 0.0903, "step": 49050 }, { "epoch": 23.53, "learning_rate": 5.8062714764059254e-05, "loss": 0.0949, "step": 49060 }, { "epoch": 23.53, "learning_rate": 5.80485440557392e-05, "loss": 0.1035, "step": 49070 }, { "epoch": 23.54, "learning_rate": 5.803437268370293e-05, "loss": 0.0959, "step": 49080 }, { "epoch": 23.54, "learning_rate": 5.8020200649119075e-05, "loss": 0.0932, "step": 49090 }, { "epoch": 23.54, "learning_rate": 5.800602795315633e-05, "loss": 0.1014, "step": 49100 }, { "epoch": 23.55, "learning_rate": 5.7991854596983445e-05, "loss": 0.091, "step": 49110 }, { "epoch": 23.55, "learning_rate": 5.797768058176917e-05, "loss": 0.1043, "step": 49120 }, { "epoch": 23.56, "learning_rate": 5.796350590868239e-05, "loss": 0.0936, "step": 49130 }, { "epoch": 23.56, "learning_rate": 5.7949330578892e-05, "loss": 0.0955, "step": 49140 }, { "epoch": 23.57, "learning_rate": 5.7935154593566965e-05, "loss": 0.0964, "step": 49150 }, { "epoch": 23.57, "learning_rate": 5.7920977953876274e-05, "loss": 0.0928, "step": 49160 }, { "epoch": 23.58, "learning_rate": 5.7906800660989004e-05, "loss": 0.1013, "step": 49170 }, { "epoch": 23.58, "learning_rate": 5.789262271607426e-05, "loss": 0.0981, "step": 49180 }, { "epoch": 23.59, "learning_rate": 5.787844412030124e-05, "loss": 0.0883, "step": 49190 }, { "epoch": 23.59, "learning_rate": 5.7864264874839144e-05, "loss": 0.0905, "step": 49200 }, { "epoch": 23.59, "learning_rate": 5.785008498085727e-05, "loss": 0.0861, "step": 49210 }, { "epoch": 23.6, "learning_rate": 5.7835904439524946e-05, "loss": 0.0828, "step": 49220 }, { "epoch": 23.6, "learning_rate": 5.782172325201155e-05, "loss": 0.0926, "step": 49230 }, { "epoch": 23.61, "learning_rate": 5.780754141948652e-05, "loss": 0.0988, "step": 49240 }, { "epoch": 23.61, "learning_rate": 5.7793358943119355e-05, "loss": 0.0857, "step": 49250 }, { "epoch": 23.62, "learning_rate": 5.777917582407961e-05, "loss": 0.0841, "step": 49260 }, { "epoch": 23.62, "learning_rate": 5.7764992063536875e-05, "loss": 0.0829, "step": 49270 }, { "epoch": 23.63, "learning_rate": 5.775080766266078e-05, "loss": 0.1116, "step": 49280 }, { "epoch": 23.63, "learning_rate": 5.7736622622621074e-05, "loss": 0.0977, "step": 49290 }, { "epoch": 23.64, "learning_rate": 5.772243694458748e-05, "loss": 0.0994, "step": 49300 }, { "epoch": 23.64, "learning_rate": 5.770825062972981e-05, "loss": 0.0843, "step": 49310 }, { "epoch": 23.64, "learning_rate": 5.7694063679217925e-05, "loss": 0.0819, "step": 49320 }, { "epoch": 23.65, "learning_rate": 5.7679876094221774e-05, "loss": 0.0827, "step": 49330 }, { "epoch": 23.65, "learning_rate": 5.766568787591128e-05, "loss": 0.0974, "step": 49340 }, { "epoch": 23.66, "learning_rate": 5.765149902545649e-05, "loss": 0.1064, "step": 49350 }, { "epoch": 23.66, "learning_rate": 5.763730954402744e-05, "loss": 0.0948, "step": 49360 }, { "epoch": 23.67, "learning_rate": 5.7623119432794316e-05, "loss": 0.0977, "step": 49370 }, { "epoch": 23.67, "learning_rate": 5.760892869292724e-05, "loss": 0.0874, "step": 49380 }, { "epoch": 23.68, "learning_rate": 5.759473732559645e-05, "loss": 0.0967, "step": 49390 }, { "epoch": 23.68, "learning_rate": 5.758054533197223e-05, "loss": 0.0895, "step": 49400 }, { "epoch": 23.69, "learning_rate": 5.75663527132249e-05, "loss": 0.0939, "step": 49410 }, { "epoch": 23.69, "learning_rate": 5.755215947052488e-05, "loss": 0.101, "step": 49420 }, { "epoch": 23.7, "learning_rate": 5.7537965605042554e-05, "loss": 0.0959, "step": 49430 }, { "epoch": 23.7, "learning_rate": 5.752377111794843e-05, "loss": 0.0906, "step": 49440 }, { "epoch": 23.7, "learning_rate": 5.750957601041305e-05, "loss": 0.088, "step": 49450 }, { "epoch": 23.71, "learning_rate": 5.749538028360699e-05, "loss": 0.0902, "step": 49460 }, { "epoch": 23.71, "learning_rate": 5.748118393870089e-05, "loss": 0.0889, "step": 49470 }, { "epoch": 23.72, "learning_rate": 5.746698697686543e-05, "loss": 0.1053, "step": 49480 }, { "epoch": 23.72, "learning_rate": 5.745278939927139e-05, "loss": 0.0887, "step": 49490 }, { "epoch": 23.73, "learning_rate": 5.743859120708951e-05, "loss": 0.1032, "step": 49500 }, { "epoch": 23.73, "learning_rate": 5.742439240149066e-05, "loss": 0.0971, "step": 49510 }, { "epoch": 23.74, "learning_rate": 5.7410192983645714e-05, "loss": 0.0871, "step": 49520 }, { "epoch": 23.74, "learning_rate": 5.739599295472564e-05, "loss": 0.1047, "step": 49530 }, { "epoch": 23.75, "learning_rate": 5.738179231590142e-05, "loss": 0.1069, "step": 49540 }, { "epoch": 23.75, "learning_rate": 5.7367591068344085e-05, "loss": 0.0984, "step": 49550 }, { "epoch": 23.75, "learning_rate": 5.735338921322475e-05, "loss": 0.1068, "step": 49560 }, { "epoch": 23.76, "learning_rate": 5.7339186751714544e-05, "loss": 0.0922, "step": 49570 }, { "epoch": 23.76, "learning_rate": 5.732498368498466e-05, "loss": 0.1072, "step": 49580 }, { "epoch": 23.77, "learning_rate": 5.731078001420635e-05, "loss": 0.0949, "step": 49590 }, { "epoch": 23.77, "learning_rate": 5.72965757405509e-05, "loss": 0.0881, "step": 49600 }, { "epoch": 23.78, "learning_rate": 5.728237086518966e-05, "loss": 0.1047, "step": 49610 }, { "epoch": 23.78, "learning_rate": 5.7268165389294004e-05, "loss": 0.096, "step": 49620 }, { "epoch": 23.79, "learning_rate": 5.72539593140354e-05, "loss": 0.0961, "step": 49630 }, { "epoch": 23.79, "learning_rate": 5.7239752640585345e-05, "loss": 0.0888, "step": 49640 }, { "epoch": 23.8, "learning_rate": 5.7225545370115355e-05, "loss": 0.1027, "step": 49650 }, { "epoch": 23.8, "learning_rate": 5.7211337503797e-05, "loss": 0.0947, "step": 49660 }, { "epoch": 23.8, "learning_rate": 5.719712904280198e-05, "loss": 0.0919, "step": 49670 }, { "epoch": 23.81, "learning_rate": 5.718291998830194e-05, "loss": 0.0931, "step": 49680 }, { "epoch": 23.81, "learning_rate": 5.716871034146865e-05, "loss": 0.0856, "step": 49690 }, { "epoch": 23.82, "learning_rate": 5.715450010347384e-05, "loss": 0.0964, "step": 49700 }, { "epoch": 23.82, "learning_rate": 5.71402892754894e-05, "loss": 0.0976, "step": 49710 }, { "epoch": 23.83, "learning_rate": 5.71260778586872e-05, "loss": 0.0882, "step": 49720 }, { "epoch": 23.83, "learning_rate": 5.711186585423915e-05, "loss": 0.1069, "step": 49730 }, { "epoch": 23.84, "learning_rate": 5.7097653263317266e-05, "loss": 0.0846, "step": 49740 }, { "epoch": 23.84, "learning_rate": 5.708344008709354e-05, "loss": 0.0956, "step": 49750 }, { "epoch": 23.85, "learning_rate": 5.7069226326740076e-05, "loss": 0.079, "step": 49760 }, { "epoch": 23.85, "learning_rate": 5.705501198342899e-05, "loss": 0.0853, "step": 49770 }, { "epoch": 23.86, "learning_rate": 5.704079705833245e-05, "loss": 0.0786, "step": 49780 }, { "epoch": 23.86, "learning_rate": 5.702658155262268e-05, "loss": 0.0928, "step": 49790 }, { "epoch": 23.86, "learning_rate": 5.7012365467471975e-05, "loss": 0.0902, "step": 49800 }, { "epoch": 23.87, "learning_rate": 5.699814880405261e-05, "loss": 0.0864, "step": 49810 }, { "epoch": 23.87, "learning_rate": 5.6983931563536976e-05, "loss": 0.088, "step": 49820 }, { "epoch": 23.88, "learning_rate": 5.6969713747097486e-05, "loss": 0.0933, "step": 49830 }, { "epoch": 23.88, "learning_rate": 5.695549535590659e-05, "loss": 0.0935, "step": 49840 }, { "epoch": 23.89, "learning_rate": 5.694127639113679e-05, "loss": 0.0875, "step": 49850 }, { "epoch": 23.89, "learning_rate": 5.6927056853960646e-05, "loss": 0.0971, "step": 49860 }, { "epoch": 23.9, "learning_rate": 5.691283674555078e-05, "loss": 0.0945, "step": 49870 }, { "epoch": 23.9, "learning_rate": 5.6898616067079813e-05, "loss": 0.1051, "step": 49880 }, { "epoch": 23.91, "learning_rate": 5.688439481972044e-05, "loss": 0.0852, "step": 49890 }, { "epoch": 23.91, "learning_rate": 5.687017300464544e-05, "loss": 0.0936, "step": 49900 }, { "epoch": 23.91, "learning_rate": 5.6855950623027565e-05, "loss": 0.0953, "step": 49910 }, { "epoch": 23.92, "learning_rate": 5.684172767603967e-05, "loss": 0.0938, "step": 49920 }, { "epoch": 23.92, "learning_rate": 5.6827504164854616e-05, "loss": 0.0928, "step": 49930 }, { "epoch": 23.93, "learning_rate": 5.681328009064537e-05, "loss": 0.0895, "step": 49940 }, { "epoch": 23.93, "learning_rate": 5.679905545458488e-05, "loss": 0.097, "step": 49950 }, { "epoch": 23.94, "learning_rate": 5.678483025784617e-05, "loss": 0.1005, "step": 49960 }, { "epoch": 23.94, "learning_rate": 5.677060450160232e-05, "loss": 0.0982, "step": 49970 }, { "epoch": 23.95, "learning_rate": 5.675637818702644e-05, "loss": 0.0986, "step": 49980 }, { "epoch": 23.95, "learning_rate": 5.674215131529169e-05, "loss": 0.1126, "step": 49990 }, { "epoch": 24.0, "learning_rate": 5.6727923887571265e-05, "loss": 0.0921, "step": 50000 }, { "epoch": 24.01, "learning_rate": 5.671369590503843e-05, "loss": 0.0866, "step": 50010 }, { "epoch": 24.01, "learning_rate": 5.669946736886649e-05, "loss": 0.0895, "step": 50020 }, { "epoch": 24.02, "learning_rate": 5.668523828022878e-05, "loss": 0.0826, "step": 50030 }, { "epoch": 24.02, "learning_rate": 5.667100864029868e-05, "loss": 0.0916, "step": 50040 }, { "epoch": 24.03, "learning_rate": 5.665677845024965e-05, "loss": 0.0829, "step": 50050 }, { "epoch": 24.03, "learning_rate": 5.6642547711255145e-05, "loss": 0.0968, "step": 50060 }, { "epoch": 24.04, "learning_rate": 5.66283164244887e-05, "loss": 0.0886, "step": 50070 }, { "epoch": 24.04, "learning_rate": 5.66140845911239e-05, "loss": 0.1084, "step": 50080 }, { "epoch": 24.04, "learning_rate": 5.659985221233432e-05, "loss": 0.0829, "step": 50090 }, { "epoch": 24.05, "learning_rate": 5.6585619289293676e-05, "loss": 0.0863, "step": 50100 }, { "epoch": 24.05, "learning_rate": 5.657138582317564e-05, "loss": 0.082, "step": 50110 }, { "epoch": 24.06, "learning_rate": 5.6557151815153955e-05, "loss": 0.0856, "step": 50120 }, { "epoch": 24.06, "learning_rate": 5.654291726640243e-05, "loss": 0.0865, "step": 50130 }, { "epoch": 24.07, "learning_rate": 5.6528682178094915e-05, "loss": 0.082, "step": 50140 }, { "epoch": 24.07, "learning_rate": 5.651444655140528e-05, "loss": 0.0876, "step": 50150 }, { "epoch": 24.08, "learning_rate": 5.650021038750745e-05, "loss": 0.093, "step": 50160 }, { "epoch": 24.08, "learning_rate": 5.6485973687575424e-05, "loss": 0.0942, "step": 50170 }, { "epoch": 24.09, "learning_rate": 5.647173645278318e-05, "loss": 0.0852, "step": 50180 }, { "epoch": 24.09, "learning_rate": 5.64574986843048e-05, "loss": 0.0956, "step": 50190 }, { "epoch": 24.1, "learning_rate": 5.644326038331439e-05, "loss": 0.0925, "step": 50200 }, { "epoch": 24.1, "learning_rate": 5.6429021550986105e-05, "loss": 0.0868, "step": 50210 }, { "epoch": 24.1, "learning_rate": 5.641478218849412e-05, "loss": 0.0887, "step": 50220 }, { "epoch": 24.11, "learning_rate": 5.640054229701269e-05, "loss": 0.0775, "step": 50230 }, { "epoch": 24.11, "learning_rate": 5.638630187771609e-05, "loss": 0.0917, "step": 50240 }, { "epoch": 24.12, "learning_rate": 5.6372060931778633e-05, "loss": 0.085, "step": 50250 }, { "epoch": 24.12, "learning_rate": 5.6357819460374695e-05, "loss": 0.0921, "step": 50260 }, { "epoch": 24.13, "learning_rate": 5.634357746467869e-05, "loss": 0.0834, "step": 50270 }, { "epoch": 24.13, "learning_rate": 5.632933494586507e-05, "loss": 0.09, "step": 50280 }, { "epoch": 24.14, "learning_rate": 5.631509190510833e-05, "loss": 0.0899, "step": 50290 }, { "epoch": 24.14, "learning_rate": 5.6300848343583004e-05, "loss": 0.0895, "step": 50300 }, { "epoch": 24.15, "learning_rate": 5.6286604262463684e-05, "loss": 0.0965, "step": 50310 }, { "epoch": 24.15, "learning_rate": 5.627235966292499e-05, "loss": 0.0971, "step": 50320 }, { "epoch": 24.15, "learning_rate": 5.62581145461416e-05, "loss": 0.0901, "step": 50330 }, { "epoch": 24.16, "learning_rate": 5.6243868913288215e-05, "loss": 0.092, "step": 50340 }, { "epoch": 24.16, "learning_rate": 5.622962276553959e-05, "loss": 0.0813, "step": 50350 }, { "epoch": 24.17, "learning_rate": 5.6215376104070525e-05, "loss": 0.0813, "step": 50360 }, { "epoch": 24.17, "learning_rate": 5.620112893005587e-05, "loss": 0.0814, "step": 50370 }, { "epoch": 24.18, "learning_rate": 5.6186881244670465e-05, "loss": 0.0876, "step": 50380 }, { "epoch": 24.18, "learning_rate": 5.6172633049089275e-05, "loss": 0.0921, "step": 50390 }, { "epoch": 24.19, "learning_rate": 5.615838434448726e-05, "loss": 0.085, "step": 50400 }, { "epoch": 24.19, "learning_rate": 5.614413513203941e-05, "loss": 0.0941, "step": 50410 }, { "epoch": 24.2, "learning_rate": 5.612988541292078e-05, "loss": 0.092, "step": 50420 }, { "epoch": 24.2, "learning_rate": 5.611563518830646e-05, "loss": 0.0961, "step": 50430 }, { "epoch": 24.2, "learning_rate": 5.6101384459371596e-05, "loss": 0.0879, "step": 50440 }, { "epoch": 24.21, "learning_rate": 5.608713322729134e-05, "loss": 0.0911, "step": 50450 }, { "epoch": 24.21, "learning_rate": 5.6072881493240924e-05, "loss": 0.0886, "step": 50460 }, { "epoch": 24.22, "learning_rate": 5.60586292583956e-05, "loss": 0.0892, "step": 50470 }, { "epoch": 24.22, "learning_rate": 5.604437652393067e-05, "loss": 0.0913, "step": 50480 }, { "epoch": 24.23, "learning_rate": 5.6030123291021466e-05, "loss": 0.0932, "step": 50490 }, { "epoch": 24.23, "learning_rate": 5.601586956084336e-05, "loss": 0.0934, "step": 50500 }, { "epoch": 24.24, "learning_rate": 5.60016153345718e-05, "loss": 0.0989, "step": 50510 }, { "epoch": 24.24, "learning_rate": 5.598736061338223e-05, "loss": 0.107, "step": 50520 }, { "epoch": 24.25, "learning_rate": 5.5973105398450155e-05, "loss": 0.0886, "step": 50530 }, { "epoch": 24.25, "learning_rate": 5.595884969095111e-05, "loss": 0.0943, "step": 50540 }, { "epoch": 24.26, "learning_rate": 5.594459349206072e-05, "loss": 0.0927, "step": 50550 }, { "epoch": 24.26, "learning_rate": 5.593033680295455e-05, "loss": 0.0934, "step": 50560 }, { "epoch": 24.26, "learning_rate": 5.5916079624808305e-05, "loss": 0.0885, "step": 50570 }, { "epoch": 24.27, "learning_rate": 5.590182195879768e-05, "loss": 0.09, "step": 50580 }, { "epoch": 24.27, "learning_rate": 5.588756380609844e-05, "loss": 0.1069, "step": 50590 }, { "epoch": 24.28, "learning_rate": 5.5873305167886334e-05, "loss": 0.0863, "step": 50600 }, { "epoch": 24.28, "learning_rate": 5.5859046045337216e-05, "loss": 0.0906, "step": 50610 }, { "epoch": 24.29, "learning_rate": 5.584478643962695e-05, "loss": 0.0948, "step": 50620 }, { "epoch": 24.29, "learning_rate": 5.583052635193142e-05, "loss": 0.0816, "step": 50630 }, { "epoch": 24.3, "learning_rate": 5.58162657834266e-05, "loss": 0.0901, "step": 50640 }, { "epoch": 24.3, "learning_rate": 5.580200473528846e-05, "loss": 0.102, "step": 50650 }, { "epoch": 24.31, "learning_rate": 5.578774320869305e-05, "loss": 0.0867, "step": 50660 }, { "epoch": 24.31, "learning_rate": 5.577348120481638e-05, "loss": 0.092, "step": 50670 }, { "epoch": 24.31, "learning_rate": 5.5759218724834596e-05, "loss": 0.0986, "step": 50680 }, { "epoch": 24.32, "learning_rate": 5.5744955769923844e-05, "loss": 0.0914, "step": 50690 }, { "epoch": 24.32, "learning_rate": 5.5730692341260294e-05, "loss": 0.091, "step": 50700 }, { "epoch": 24.33, "learning_rate": 5.571642844002016e-05, "loss": 0.0817, "step": 50710 }, { "epoch": 24.33, "learning_rate": 5.5702164067379694e-05, "loss": 0.0825, "step": 50720 }, { "epoch": 24.34, "learning_rate": 5.568789922451524e-05, "loss": 0.0886, "step": 50730 }, { "epoch": 24.34, "learning_rate": 5.567363391260309e-05, "loss": 0.0924, "step": 50740 }, { "epoch": 24.35, "learning_rate": 5.565936813281964e-05, "loss": 0.0931, "step": 50750 }, { "epoch": 24.35, "learning_rate": 5.56451018863413e-05, "loss": 0.0829, "step": 50760 }, { "epoch": 24.36, "learning_rate": 5.563083517434453e-05, "loss": 0.0958, "step": 50770 }, { "epoch": 24.36, "learning_rate": 5.561656799800583e-05, "loss": 0.099, "step": 50780 }, { "epoch": 24.36, "learning_rate": 5.56023003585017e-05, "loss": 0.1006, "step": 50790 }, { "epoch": 24.37, "learning_rate": 5.558803225700873e-05, "loss": 0.0896, "step": 50800 }, { "epoch": 24.37, "learning_rate": 5.5573763694703516e-05, "loss": 0.0836, "step": 50810 }, { "epoch": 24.38, "learning_rate": 5.5559494672762734e-05, "loss": 0.0822, "step": 50820 }, { "epoch": 24.38, "learning_rate": 5.5545225192363025e-05, "loss": 0.0826, "step": 50830 }, { "epoch": 24.39, "learning_rate": 5.553095525468113e-05, "loss": 0.0859, "step": 50840 }, { "epoch": 24.39, "learning_rate": 5.55166848608938e-05, "loss": 0.1009, "step": 50850 }, { "epoch": 24.4, "learning_rate": 5.550241401217785e-05, "loss": 0.0911, "step": 50860 }, { "epoch": 24.4, "learning_rate": 5.548814270971009e-05, "loss": 0.0841, "step": 50870 }, { "epoch": 24.41, "learning_rate": 5.5473870954667384e-05, "loss": 0.0793, "step": 50880 }, { "epoch": 24.41, "learning_rate": 5.5459598748226684e-05, "loss": 0.0772, "step": 50890 }, { "epoch": 24.41, "learning_rate": 5.544532609156488e-05, "loss": 0.0896, "step": 50900 }, { "epoch": 24.42, "learning_rate": 5.543105298585899e-05, "loss": 0.0848, "step": 50910 }, { "epoch": 24.42, "learning_rate": 5.541677943228601e-05, "loss": 0.0864, "step": 50920 }, { "epoch": 24.43, "learning_rate": 5.540250543202303e-05, "loss": 0.0909, "step": 50930 }, { "epoch": 24.43, "learning_rate": 5.5388230986247117e-05, "loss": 0.0865, "step": 50940 }, { "epoch": 24.44, "learning_rate": 5.5373956096135395e-05, "loss": 0.0854, "step": 50950 }, { "epoch": 24.44, "learning_rate": 5.535968076286506e-05, "loss": 0.0822, "step": 50960 }, { "epoch": 24.45, "learning_rate": 5.534540498761328e-05, "loss": 0.0831, "step": 50970 }, { "epoch": 24.45, "learning_rate": 5.533112877155732e-05, "loss": 0.0895, "step": 50980 }, { "epoch": 24.46, "learning_rate": 5.5316852115874427e-05, "loss": 0.0881, "step": 50990 }, { "epoch": 24.46, "learning_rate": 5.5302575021741967e-05, "loss": 0.1147, "step": 51000 }, { "epoch": 24.47, "learning_rate": 5.5288297490337225e-05, "loss": 0.0824, "step": 51010 }, { "epoch": 24.47, "learning_rate": 5.527401952283763e-05, "loss": 0.1006, "step": 51020 }, { "epoch": 24.47, "learning_rate": 5.525974112042058e-05, "loss": 0.085, "step": 51030 }, { "epoch": 24.48, "learning_rate": 5.5245462284263547e-05, "loss": 0.0852, "step": 51040 }, { "epoch": 24.48, "learning_rate": 5.523118301554401e-05, "loss": 0.0953, "step": 51050 }, { "epoch": 24.49, "learning_rate": 5.5216903315439486e-05, "loss": 0.0909, "step": 51060 }, { "epoch": 24.49, "learning_rate": 5.5202623185127564e-05, "loss": 0.0901, "step": 51070 }, { "epoch": 24.5, "learning_rate": 5.518834262578583e-05, "loss": 0.0935, "step": 51080 }, { "epoch": 24.5, "learning_rate": 5.5174061638591926e-05, "loss": 0.093, "step": 51090 }, { "epoch": 24.51, "learning_rate": 5.515978022472349e-05, "loss": 0.0953, "step": 51100 }, { "epoch": 24.51, "learning_rate": 5.514549838535827e-05, "loss": 0.0914, "step": 51110 }, { "epoch": 24.52, "learning_rate": 5.513121612167399e-05, "loss": 0.0982, "step": 51120 }, { "epoch": 24.52, "learning_rate": 5.511693343484842e-05, "loss": 0.0885, "step": 51130 }, { "epoch": 24.52, "learning_rate": 5.510265032605935e-05, "loss": 0.1037, "step": 51140 }, { "epoch": 24.53, "learning_rate": 5.508836679648465e-05, "loss": 0.0984, "step": 51150 }, { "epoch": 24.53, "learning_rate": 5.507408284730221e-05, "loss": 0.09, "step": 51160 }, { "epoch": 24.54, "learning_rate": 5.505979847968991e-05, "loss": 0.093, "step": 51170 }, { "epoch": 24.54, "learning_rate": 5.504551369482571e-05, "loss": 0.0951, "step": 51180 }, { "epoch": 24.55, "learning_rate": 5.50312284938876e-05, "loss": 0.0864, "step": 51190 }, { "epoch": 24.55, "learning_rate": 5.501694287805361e-05, "loss": 0.0878, "step": 51200 }, { "epoch": 24.56, "learning_rate": 5.500265684850176e-05, "loss": 0.0904, "step": 51210 }, { "epoch": 24.56, "learning_rate": 5.498837040641015e-05, "loss": 0.0946, "step": 51220 }, { "epoch": 24.57, "learning_rate": 5.4974083552956904e-05, "loss": 0.0965, "step": 51230 }, { "epoch": 24.57, "learning_rate": 5.495979628932017e-05, "loss": 0.1038, "step": 51240 }, { "epoch": 24.57, "learning_rate": 5.494550861667812e-05, "loss": 0.0945, "step": 51250 }, { "epoch": 24.58, "learning_rate": 5.493122053620899e-05, "loss": 0.0917, "step": 51260 }, { "epoch": 24.58, "learning_rate": 5.491693204909105e-05, "loss": 0.0933, "step": 51270 }, { "epoch": 24.59, "learning_rate": 5.4902643156502554e-05, "loss": 0.0934, "step": 51280 }, { "epoch": 24.59, "learning_rate": 5.488835385962184e-05, "loss": 0.0848, "step": 51290 }, { "epoch": 24.6, "learning_rate": 5.487406415962728e-05, "loss": 0.0886, "step": 51300 }, { "epoch": 24.6, "learning_rate": 5.485977405769723e-05, "loss": 0.0915, "step": 51310 }, { "epoch": 24.61, "learning_rate": 5.484548355501012e-05, "loss": 0.0943, "step": 51320 }, { "epoch": 24.61, "learning_rate": 5.483119265274442e-05, "loss": 0.082, "step": 51330 }, { "epoch": 24.62, "learning_rate": 5.481690135207861e-05, "loss": 0.0889, "step": 51340 }, { "epoch": 24.62, "learning_rate": 5.480260965419119e-05, "loss": 0.0859, "step": 51350 }, { "epoch": 24.63, "learning_rate": 5.478831756026074e-05, "loss": 0.0905, "step": 51360 }, { "epoch": 24.63, "learning_rate": 5.477402507146582e-05, "loss": 0.086, "step": 51370 }, { "epoch": 24.63, "learning_rate": 5.475973218898508e-05, "loss": 0.0898, "step": 51380 }, { "epoch": 24.64, "learning_rate": 5.4745438913997146e-05, "loss": 0.0996, "step": 51390 }, { "epoch": 24.64, "learning_rate": 5.473114524768068e-05, "loss": 0.1026, "step": 51400 }, { "epoch": 24.65, "learning_rate": 5.4716851191214446e-05, "loss": 0.0988, "step": 51410 }, { "epoch": 24.65, "learning_rate": 5.470255674577716e-05, "loss": 0.0897, "step": 51420 }, { "epoch": 24.66, "learning_rate": 5.468826191254762e-05, "loss": 0.0877, "step": 51430 }, { "epoch": 24.66, "learning_rate": 5.46739666927046e-05, "loss": 0.0832, "step": 51440 }, { "epoch": 24.67, "learning_rate": 5.465967108742699e-05, "loss": 0.0914, "step": 51450 }, { "epoch": 24.67, "learning_rate": 5.4645375097893626e-05, "loss": 0.0937, "step": 51460 }, { "epoch": 24.68, "learning_rate": 5.463107872528346e-05, "loss": 0.0853, "step": 51470 }, { "epoch": 24.68, "learning_rate": 5.461678197077538e-05, "loss": 0.099, "step": 51480 }, { "epoch": 24.68, "learning_rate": 5.460248483554836e-05, "loss": 0.0928, "step": 51490 }, { "epoch": 24.69, "learning_rate": 5.4588187320781445e-05, "loss": 0.1062, "step": 51500 }, { "epoch": 24.69, "learning_rate": 5.4573889427653626e-05, "loss": 0.0897, "step": 51510 }, { "epoch": 24.7, "learning_rate": 5.455959115734398e-05, "loss": 0.2018, "step": 51520 }, { "epoch": 24.7, "learning_rate": 5.454529251103161e-05, "loss": 0.0891, "step": 51530 }, { "epoch": 24.71, "learning_rate": 5.453099348989563e-05, "loss": 0.0982, "step": 51540 }, { "epoch": 24.71, "learning_rate": 5.451669409511518e-05, "loss": 0.0926, "step": 51550 }, { "epoch": 24.72, "learning_rate": 5.45023943278695e-05, "loss": 0.0924, "step": 51560 }, { "epoch": 24.72, "learning_rate": 5.448809418933777e-05, "loss": 0.0929, "step": 51570 }, { "epoch": 24.73, "learning_rate": 5.4473793680699225e-05, "loss": 0.096, "step": 51580 }, { "epoch": 24.73, "learning_rate": 5.445949280313316e-05, "loss": 0.1009, "step": 51590 }, { "epoch": 24.73, "learning_rate": 5.4445191557818894e-05, "loss": 0.0884, "step": 51600 }, { "epoch": 24.74, "learning_rate": 5.443088994593577e-05, "loss": 0.0983, "step": 51610 }, { "epoch": 24.74, "learning_rate": 5.441658796866313e-05, "loss": 0.0906, "step": 51620 }, { "epoch": 24.75, "learning_rate": 5.4402285627180394e-05, "loss": 0.0888, "step": 51630 }, { "epoch": 24.75, "learning_rate": 5.438798292266699e-05, "loss": 0.0864, "step": 51640 }, { "epoch": 24.76, "learning_rate": 5.4373679856302395e-05, "loss": 0.095, "step": 51650 }, { "epoch": 24.76, "learning_rate": 5.435937642926606e-05, "loss": 0.099, "step": 51660 }, { "epoch": 24.77, "learning_rate": 5.434507264273754e-05, "loss": 0.0866, "step": 51670 }, { "epoch": 24.77, "learning_rate": 5.4330768497896365e-05, "loss": 0.1039, "step": 51680 }, { "epoch": 24.78, "learning_rate": 5.431646399592213e-05, "loss": 0.1042, "step": 51690 }, { "epoch": 24.78, "learning_rate": 5.430215913799441e-05, "loss": 0.0963, "step": 51700 }, { "epoch": 24.79, "learning_rate": 5.4287853925292887e-05, "loss": 0.0952, "step": 51710 }, { "epoch": 24.79, "learning_rate": 5.427354835899721e-05, "loss": 0.0925, "step": 51720 }, { "epoch": 24.79, "learning_rate": 5.425924244028706e-05, "loss": 0.0834, "step": 51730 }, { "epoch": 24.8, "learning_rate": 5.4244936170342166e-05, "loss": 0.0869, "step": 51740 }, { "epoch": 24.8, "learning_rate": 5.4230629550342305e-05, "loss": 0.0829, "step": 51750 }, { "epoch": 24.81, "learning_rate": 5.421632258146724e-05, "loss": 0.0913, "step": 51760 }, { "epoch": 24.81, "learning_rate": 5.420201526489679e-05, "loss": 0.0876, "step": 51770 }, { "epoch": 24.82, "learning_rate": 5.418770760181078e-05, "loss": 0.0872, "step": 51780 }, { "epoch": 24.82, "learning_rate": 5.417339959338911e-05, "loss": 0.0933, "step": 51790 }, { "epoch": 24.83, "learning_rate": 5.4159091240811646e-05, "loss": 0.0877, "step": 51800 }, { "epoch": 24.83, "learning_rate": 5.414478254525833e-05, "loss": 0.0815, "step": 51810 }, { "epoch": 24.84, "learning_rate": 5.413047350790913e-05, "loss": 0.084, "step": 51820 }, { "epoch": 24.84, "learning_rate": 5.4116164129944006e-05, "loss": 0.0977, "step": 51830 }, { "epoch": 24.84, "learning_rate": 5.410185441254297e-05, "loss": 0.0885, "step": 51840 }, { "epoch": 24.85, "learning_rate": 5.408754435688606e-05, "loss": 0.0861, "step": 51850 }, { "epoch": 24.85, "learning_rate": 5.407323396415336e-05, "loss": 0.0935, "step": 51860 }, { "epoch": 24.86, "learning_rate": 5.405892323552494e-05, "loss": 0.0873, "step": 51870 }, { "epoch": 24.86, "learning_rate": 5.404461217218094e-05, "loss": 0.093, "step": 51880 }, { "epoch": 24.87, "learning_rate": 5.403030077530149e-05, "loss": 0.0834, "step": 51890 }, { "epoch": 24.87, "learning_rate": 5.4015989046066804e-05, "loss": 0.0829, "step": 51900 }, { "epoch": 24.88, "learning_rate": 5.4001676985657054e-05, "loss": 0.0816, "step": 51910 }, { "epoch": 24.88, "learning_rate": 5.398736459525248e-05, "loss": 0.0931, "step": 51920 }, { "epoch": 24.89, "learning_rate": 5.397305187603333e-05, "loss": 0.0911, "step": 51930 }, { "epoch": 24.89, "learning_rate": 5.39587388291799e-05, "loss": 0.1013, "step": 51940 }, { "epoch": 24.89, "learning_rate": 5.394442545587253e-05, "loss": 0.0892, "step": 51950 }, { "epoch": 24.9, "learning_rate": 5.393011175729151e-05, "loss": 0.0863, "step": 51960 }, { "epoch": 24.9, "learning_rate": 5.3915797734617236e-05, "loss": 0.0868, "step": 51970 }, { "epoch": 24.91, "learning_rate": 5.390148338903008e-05, "loss": 0.0836, "step": 51980 }, { "epoch": 24.91, "learning_rate": 5.388716872171051e-05, "loss": 0.1055, "step": 51990 }, { "epoch": 24.92, "learning_rate": 5.387285373383892e-05, "loss": 0.0759, "step": 52000 }, { "epoch": 24.92, "learning_rate": 5.385853842659581e-05, "loss": 0.0904, "step": 52010 }, { "epoch": 24.93, "learning_rate": 5.3844222801161684e-05, "loss": 0.0915, "step": 52020 }, { "epoch": 24.93, "learning_rate": 5.3829906858717036e-05, "loss": 0.09, "step": 52030 }, { "epoch": 24.94, "learning_rate": 5.3815590600442445e-05, "loss": 0.0936, "step": 52040 }, { "epoch": 24.94, "learning_rate": 5.380127402751848e-05, "loss": 0.0856, "step": 52050 }, { "epoch": 24.95, "learning_rate": 5.378695714112576e-05, "loss": 0.0902, "step": 52060 }, { "epoch": 24.95, "learning_rate": 5.3772639942444894e-05, "loss": 0.1077, "step": 52070 }, { "epoch": 25.0, "learning_rate": 5.375832243265654e-05, "loss": 0.1215, "step": 52080 }, { "epoch": 25.01, "learning_rate": 5.37440046129414e-05, "loss": 0.0914, "step": 52090 }, { "epoch": 25.01, "learning_rate": 5.372968648448015e-05, "loss": 0.0933, "step": 52100 }, { "epoch": 25.02, "learning_rate": 5.371536804845354e-05, "loss": 0.0914, "step": 52110 }, { "epoch": 25.02, "learning_rate": 5.3701049306042315e-05, "loss": 0.0947, "step": 52120 }, { "epoch": 25.03, "learning_rate": 5.368673025842729e-05, "loss": 0.0968, "step": 52130 }, { "epoch": 25.03, "learning_rate": 5.3672410906789237e-05, "loss": 0.0944, "step": 52140 }, { "epoch": 25.03, "learning_rate": 5.365809125230899e-05, "loss": 0.0955, "step": 52150 }, { "epoch": 25.04, "learning_rate": 5.364377129616743e-05, "loss": 0.0944, "step": 52160 }, { "epoch": 25.04, "learning_rate": 5.362945103954544e-05, "loss": 0.0902, "step": 52170 }, { "epoch": 25.05, "learning_rate": 5.361513048362392e-05, "loss": 0.0981, "step": 52180 }, { "epoch": 25.05, "learning_rate": 5.360080962958377e-05, "loss": 0.0867, "step": 52190 }, { "epoch": 25.06, "learning_rate": 5.358648847860599e-05, "loss": 0.0963, "step": 52200 }, { "epoch": 25.06, "learning_rate": 5.3572167031871544e-05, "loss": 0.0857, "step": 52210 }, { "epoch": 25.07, "learning_rate": 5.3557845290561446e-05, "loss": 0.1004, "step": 52220 }, { "epoch": 25.07, "learning_rate": 5.3543523255856706e-05, "loss": 0.0923, "step": 52230 }, { "epoch": 25.08, "learning_rate": 5.3529200928938406e-05, "loss": 0.0854, "step": 52240 }, { "epoch": 25.08, "learning_rate": 5.35148783109876e-05, "loss": 0.0878, "step": 52250 }, { "epoch": 25.08, "learning_rate": 5.350055540318542e-05, "loss": 0.0821, "step": 52260 }, { "epoch": 25.09, "learning_rate": 5.348623220671296e-05, "loss": 0.0998, "step": 52270 }, { "epoch": 25.09, "learning_rate": 5.3471908722751375e-05, "loss": 0.0951, "step": 52280 }, { "epoch": 25.1, "learning_rate": 5.3457584952481865e-05, "loss": 0.0997, "step": 52290 }, { "epoch": 25.1, "learning_rate": 5.344326089708559e-05, "loss": 0.0851, "step": 52300 }, { "epoch": 25.11, "learning_rate": 5.3428936557743804e-05, "loss": 0.0952, "step": 52310 }, { "epoch": 25.11, "learning_rate": 5.341461193563773e-05, "loss": 0.0824, "step": 52320 }, { "epoch": 25.12, "learning_rate": 5.3400287031948635e-05, "loss": 0.0845, "step": 52330 }, { "epoch": 25.12, "learning_rate": 5.338596184785782e-05, "loss": 0.0857, "step": 52340 }, { "epoch": 25.13, "learning_rate": 5.337163638454661e-05, "loss": 0.0936, "step": 52350 }, { "epoch": 25.13, "learning_rate": 5.3357310643196324e-05, "loss": 0.0959, "step": 52360 }, { "epoch": 25.13, "learning_rate": 5.3342984624988326e-05, "loss": 0.0895, "step": 52370 }, { "epoch": 25.14, "learning_rate": 5.332865833110399e-05, "loss": 0.0805, "step": 52380 }, { "epoch": 25.14, "learning_rate": 5.3314331762724726e-05, "loss": 0.0919, "step": 52390 }, { "epoch": 25.15, "learning_rate": 5.3300004921031974e-05, "loss": 0.0941, "step": 52400 }, { "epoch": 25.15, "learning_rate": 5.328567780720718e-05, "loss": 0.0905, "step": 52410 }, { "epoch": 25.16, "learning_rate": 5.327135042243181e-05, "loss": 0.1109, "step": 52420 }, { "epoch": 25.16, "learning_rate": 5.325702276788736e-05, "loss": 0.0969, "step": 52430 }, { "epoch": 25.17, "learning_rate": 5.324269484475536e-05, "loss": 0.0829, "step": 52440 }, { "epoch": 25.17, "learning_rate": 5.322836665421733e-05, "loss": 0.0877, "step": 52450 }, { "epoch": 25.18, "learning_rate": 5.321403819745484e-05, "loss": 0.0928, "step": 52460 }, { "epoch": 25.18, "learning_rate": 5.31997094756495e-05, "loss": 0.0961, "step": 52470 }, { "epoch": 25.19, "learning_rate": 5.3185380489982864e-05, "loss": 0.0903, "step": 52480 }, { "epoch": 25.19, "learning_rate": 5.3171051241636595e-05, "loss": 0.0867, "step": 52490 }, { "epoch": 25.19, "learning_rate": 5.3156721731792336e-05, "loss": 0.0764, "step": 52500 }, { "epoch": 25.2, "learning_rate": 5.3142391961631765e-05, "loss": 0.0818, "step": 52510 }, { "epoch": 25.2, "learning_rate": 5.312806193233656e-05, "loss": 0.082, "step": 52520 }, { "epoch": 25.21, "learning_rate": 5.311373164508842e-05, "loss": 0.0846, "step": 52530 }, { "epoch": 25.21, "learning_rate": 5.309940110106912e-05, "loss": 0.0869, "step": 52540 }, { "epoch": 25.22, "learning_rate": 5.3085070301460384e-05, "loss": 0.0797, "step": 52550 }, { "epoch": 25.22, "learning_rate": 5.307073924744401e-05, "loss": 0.081, "step": 52560 }, { "epoch": 25.23, "learning_rate": 5.305640794020177e-05, "loss": 0.0818, "step": 52570 }, { "epoch": 25.23, "learning_rate": 5.304207638091552e-05, "loss": 0.0956, "step": 52580 }, { "epoch": 25.24, "learning_rate": 5.302774457076706e-05, "loss": 0.0775, "step": 52590 }, { "epoch": 25.24, "learning_rate": 5.301341251093828e-05, "loss": 0.0895, "step": 52600 }, { "epoch": 25.24, "learning_rate": 5.299908020261104e-05, "loss": 0.1019, "step": 52610 }, { "epoch": 25.25, "learning_rate": 5.2984747646967244e-05, "loss": 0.0889, "step": 52620 }, { "epoch": 25.25, "learning_rate": 5.297041484518884e-05, "loss": 0.0761, "step": 52630 }, { "epoch": 25.26, "learning_rate": 5.295608179845774e-05, "loss": 0.0897, "step": 52640 }, { "epoch": 25.26, "learning_rate": 5.2941748507955914e-05, "loss": 0.0892, "step": 52650 }, { "epoch": 25.27, "learning_rate": 5.292741497486535e-05, "loss": 0.09, "step": 52660 }, { "epoch": 25.27, "learning_rate": 5.291308120036805e-05, "loss": 0.0866, "step": 52670 }, { "epoch": 25.28, "learning_rate": 5.289874718564604e-05, "loss": 0.0875, "step": 52680 }, { "epoch": 25.28, "learning_rate": 5.2884412931881346e-05, "loss": 0.1015, "step": 52690 }, { "epoch": 25.29, "learning_rate": 5.287007844025604e-05, "loss": 0.085, "step": 52700 }, { "epoch": 25.29, "learning_rate": 5.285574371195221e-05, "loss": 0.0899, "step": 52710 }, { "epoch": 25.29, "learning_rate": 5.284140874815194e-05, "loss": 0.0924, "step": 52720 }, { "epoch": 25.3, "learning_rate": 5.2827073550037365e-05, "loss": 0.0874, "step": 52730 }, { "epoch": 25.3, "learning_rate": 5.2812738118790625e-05, "loss": 0.0861, "step": 52740 }, { "epoch": 25.31, "learning_rate": 5.279840245559387e-05, "loss": 0.0948, "step": 52750 }, { "epoch": 25.31, "learning_rate": 5.278406656162928e-05, "loss": 0.0861, "step": 52760 }, { "epoch": 25.32, "learning_rate": 5.2769730438079056e-05, "loss": 0.0975, "step": 52770 }, { "epoch": 25.32, "learning_rate": 5.275539408612542e-05, "loss": 0.0833, "step": 52780 }, { "epoch": 25.33, "learning_rate": 5.2741057506950586e-05, "loss": 0.0976, "step": 52790 }, { "epoch": 25.33, "learning_rate": 5.272672070173682e-05, "loss": 0.0991, "step": 52800 }, { "epoch": 25.34, "learning_rate": 5.271238367166641e-05, "loss": 0.0915, "step": 52810 }, { "epoch": 25.34, "learning_rate": 5.269804641792162e-05, "loss": 0.0838, "step": 52820 }, { "epoch": 25.35, "learning_rate": 5.268370894168477e-05, "loss": 0.0918, "step": 52830 }, { "epoch": 25.35, "learning_rate": 5.266937124413819e-05, "loss": 0.0857, "step": 52840 }, { "epoch": 25.35, "learning_rate": 5.2655033326464245e-05, "loss": 0.0889, "step": 52850 }, { "epoch": 25.36, "learning_rate": 5.264069518984525e-05, "loss": 0.0837, "step": 52860 }, { "epoch": 25.36, "learning_rate": 5.262635683546362e-05, "loss": 0.0895, "step": 52870 }, { "epoch": 25.37, "learning_rate": 5.261201826450176e-05, "loss": 0.0864, "step": 52880 }, { "epoch": 25.37, "learning_rate": 5.259767947814207e-05, "loss": 0.0924, "step": 52890 }, { "epoch": 25.38, "learning_rate": 5.2583340477567e-05, "loss": 0.0936, "step": 52900 }, { "epoch": 25.38, "learning_rate": 5.256900126395898e-05, "loss": 0.0901, "step": 52910 }, { "epoch": 25.39, "learning_rate": 5.255466183850052e-05, "loss": 0.1042, "step": 52920 }, { "epoch": 25.39, "learning_rate": 5.254032220237407e-05, "loss": 0.088, "step": 52930 }, { "epoch": 25.4, "learning_rate": 5.252598235676217e-05, "loss": 0.083, "step": 52940 }, { "epoch": 25.4, "learning_rate": 5.2511642302847295e-05, "loss": 0.0879, "step": 52950 }, { "epoch": 25.4, "learning_rate": 5.249730204181205e-05, "loss": 0.0913, "step": 52960 }, { "epoch": 25.41, "learning_rate": 5.248296157483894e-05, "loss": 0.0801, "step": 52970 }, { "epoch": 25.41, "learning_rate": 5.246862090311054e-05, "loss": 0.0863, "step": 52980 }, { "epoch": 25.42, "learning_rate": 5.2454280027809476e-05, "loss": 0.0796, "step": 52990 }, { "epoch": 25.42, "learning_rate": 5.243993895011833e-05, "loss": 0.0905, "step": 53000 }, { "epoch": 25.43, "learning_rate": 5.242559767121974e-05, "loss": 0.0833, "step": 53010 }, { "epoch": 25.43, "learning_rate": 5.241125619229633e-05, "loss": 0.0773, "step": 53020 }, { "epoch": 25.44, "learning_rate": 5.239691451453078e-05, "loss": 0.0837, "step": 53030 }, { "epoch": 25.44, "learning_rate": 5.238257263910575e-05, "loss": 0.0902, "step": 53040 }, { "epoch": 25.45, "learning_rate": 5.2368230567203924e-05, "loss": 0.0824, "step": 53050 }, { "epoch": 25.45, "learning_rate": 5.2353888300008025e-05, "loss": 0.0831, "step": 53060 }, { "epoch": 25.45, "learning_rate": 5.233954583870075e-05, "loss": 0.0905, "step": 53070 }, { "epoch": 25.46, "learning_rate": 5.232520318446489e-05, "loss": 0.0824, "step": 53080 }, { "epoch": 25.46, "learning_rate": 5.231086033848314e-05, "loss": 0.0856, "step": 53090 }, { "epoch": 25.47, "learning_rate": 5.2296517301938294e-05, "loss": 0.0819, "step": 53100 }, { "epoch": 25.47, "learning_rate": 5.2282174076013144e-05, "loss": 0.0761, "step": 53110 }, { "epoch": 25.48, "learning_rate": 5.2267830661890495e-05, "loss": 0.0869, "step": 53120 }, { "epoch": 25.48, "learning_rate": 5.225348706075315e-05, "loss": 0.0872, "step": 53130 }, { "epoch": 25.49, "learning_rate": 5.223914327378394e-05, "loss": 0.0858, "step": 53140 }, { "epoch": 25.49, "learning_rate": 5.222479930216574e-05, "loss": 0.0842, "step": 53150 }, { "epoch": 25.5, "learning_rate": 5.221045514708137e-05, "loss": 0.0833, "step": 53160 }, { "epoch": 25.5, "learning_rate": 5.219611080971375e-05, "loss": 0.0927, "step": 53170 }, { "epoch": 25.51, "learning_rate": 5.218176629124575e-05, "loss": 0.0877, "step": 53180 }, { "epoch": 25.51, "learning_rate": 5.216742159286029e-05, "loss": 0.0708, "step": 53190 }, { "epoch": 25.51, "learning_rate": 5.215307671574027e-05, "loss": 0.0952, "step": 53200 }, { "epoch": 25.52, "learning_rate": 5.2138731661068653e-05, "loss": 0.0804, "step": 53210 }, { "epoch": 25.52, "learning_rate": 5.2124386430028395e-05, "loss": 0.0876, "step": 53220 }, { "epoch": 25.53, "learning_rate": 5.211004102380244e-05, "loss": 0.082, "step": 53230 }, { "epoch": 25.53, "learning_rate": 5.209569544357379e-05, "loss": 0.0921, "step": 53240 }, { "epoch": 25.54, "learning_rate": 5.2081349690525416e-05, "loss": 0.0891, "step": 53250 }, { "epoch": 25.54, "learning_rate": 5.206700376584036e-05, "loss": 0.0952, "step": 53260 }, { "epoch": 25.55, "learning_rate": 5.205265767070162e-05, "loss": 0.0874, "step": 53270 }, { "epoch": 25.55, "learning_rate": 5.203831140629225e-05, "loss": 0.0966, "step": 53280 }, { "epoch": 25.56, "learning_rate": 5.202396497379528e-05, "loss": 0.0893, "step": 53290 }, { "epoch": 25.56, "learning_rate": 5.200961837439382e-05, "loss": 0.0825, "step": 53300 }, { "epoch": 25.56, "learning_rate": 5.199527160927092e-05, "loss": 0.0868, "step": 53310 }, { "epoch": 25.57, "learning_rate": 5.198092467960966e-05, "loss": 0.0934, "step": 53320 }, { "epoch": 25.57, "learning_rate": 5.196657758659317e-05, "loss": 0.0889, "step": 53330 }, { "epoch": 25.58, "learning_rate": 5.1952230331404564e-05, "loss": 0.089, "step": 53340 }, { "epoch": 25.58, "learning_rate": 5.1937882915226974e-05, "loss": 0.0859, "step": 53350 }, { "epoch": 25.59, "learning_rate": 5.1923535339243554e-05, "loss": 0.0861, "step": 53360 }, { "epoch": 25.59, "learning_rate": 5.190918760463745e-05, "loss": 0.0828, "step": 53370 }, { "epoch": 25.6, "learning_rate": 5.189483971259185e-05, "loss": 0.0815, "step": 53380 }, { "epoch": 25.6, "learning_rate": 5.188049166428994e-05, "loss": 0.087, "step": 53390 }, { "epoch": 25.61, "learning_rate": 5.186614346091491e-05, "loss": 0.0913, "step": 53400 }, { "epoch": 25.61, "learning_rate": 5.185179510364998e-05, "loss": 0.0916, "step": 53410 }, { "epoch": 25.61, "learning_rate": 5.183744659367837e-05, "loss": 0.1032, "step": 53420 }, { "epoch": 25.62, "learning_rate": 5.182309793218332e-05, "loss": 0.0839, "step": 53430 }, { "epoch": 25.62, "learning_rate": 5.180874912034808e-05, "loss": 0.0927, "step": 53440 }, { "epoch": 25.63, "learning_rate": 5.1794400159355907e-05, "loss": 0.0881, "step": 53450 }, { "epoch": 25.63, "learning_rate": 5.178005105039009e-05, "loss": 0.0957, "step": 53460 }, { "epoch": 25.64, "learning_rate": 5.176570179463389e-05, "loss": 0.094, "step": 53470 }, { "epoch": 25.64, "learning_rate": 5.175135239327064e-05, "loss": 0.0937, "step": 53480 }, { "epoch": 25.65, "learning_rate": 5.173700284748363e-05, "loss": 0.0938, "step": 53490 }, { "epoch": 25.65, "learning_rate": 5.172265315845618e-05, "loss": 0.0983, "step": 53500 }, { "epoch": 25.66, "learning_rate": 5.170830332737162e-05, "loss": 0.1006, "step": 53510 }, { "epoch": 25.66, "learning_rate": 5.169395335541332e-05, "loss": 0.0863, "step": 53520 }, { "epoch": 25.66, "learning_rate": 5.167960324376463e-05, "loss": 0.09, "step": 53530 }, { "epoch": 25.67, "learning_rate": 5.1665252993608906e-05, "loss": 0.098, "step": 53540 }, { "epoch": 25.67, "learning_rate": 5.1650902606129536e-05, "loss": 0.0875, "step": 53550 }, { "epoch": 25.68, "learning_rate": 5.1636552082509915e-05, "loss": 0.0932, "step": 53560 }, { "epoch": 25.68, "learning_rate": 5.162220142393345e-05, "loss": 0.0894, "step": 53570 }, { "epoch": 25.69, "learning_rate": 5.160785063158355e-05, "loss": 0.0865, "step": 53580 }, { "epoch": 25.69, "learning_rate": 5.159349970664363e-05, "loss": 0.0796, "step": 53590 }, { "epoch": 25.7, "learning_rate": 5.157914865029715e-05, "loss": 0.0788, "step": 53600 }, { "epoch": 25.7, "learning_rate": 5.156479746372754e-05, "loss": 0.0959, "step": 53610 }, { "epoch": 25.71, "learning_rate": 5.155044614811827e-05, "loss": 0.081, "step": 53620 }, { "epoch": 25.71, "learning_rate": 5.1536094704652794e-05, "loss": 0.0892, "step": 53630 }, { "epoch": 25.72, "learning_rate": 5.15217431345146e-05, "loss": 0.0876, "step": 53640 }, { "epoch": 25.72, "learning_rate": 5.1507391438887184e-05, "loss": 0.0955, "step": 53650 }, { "epoch": 25.72, "learning_rate": 5.149303961895404e-05, "loss": 0.091, "step": 53660 }, { "epoch": 25.73, "learning_rate": 5.147868767589867e-05, "loss": 0.0971, "step": 53670 }, { "epoch": 25.73, "learning_rate": 5.14643356109046e-05, "loss": 0.0948, "step": 53680 }, { "epoch": 25.74, "learning_rate": 5.144998342515538e-05, "loss": 0.0827, "step": 53690 }, { "epoch": 25.74, "learning_rate": 5.1435631119834526e-05, "loss": 0.0846, "step": 53700 }, { "epoch": 25.75, "learning_rate": 5.1421278696125586e-05, "loss": 0.1058, "step": 53710 }, { "epoch": 25.75, "learning_rate": 5.140692615521213e-05, "loss": 0.097, "step": 53720 }, { "epoch": 25.76, "learning_rate": 5.139257349827775e-05, "loss": 0.0971, "step": 53730 }, { "epoch": 25.76, "learning_rate": 5.137822072650599e-05, "loss": 0.0843, "step": 53740 }, { "epoch": 25.77, "learning_rate": 5.136386784108045e-05, "loss": 0.0895, "step": 53750 }, { "epoch": 25.77, "learning_rate": 5.134951484318473e-05, "loss": 0.0863, "step": 53760 }, { "epoch": 25.77, "learning_rate": 5.133516173400245e-05, "loss": 0.0876, "step": 53770 }, { "epoch": 25.78, "learning_rate": 5.13208085147172e-05, "loss": 0.0936, "step": 53780 }, { "epoch": 25.78, "learning_rate": 5.1306455186512614e-05, "loss": 0.0872, "step": 53790 }, { "epoch": 25.79, "learning_rate": 5.129210175057236e-05, "loss": 0.0917, "step": 53800 }, { "epoch": 25.79, "learning_rate": 5.127774820808004e-05, "loss": 0.0929, "step": 53810 }, { "epoch": 25.8, "learning_rate": 5.1263394560219316e-05, "loss": 0.0958, "step": 53820 }, { "epoch": 25.8, "learning_rate": 5.124904080817387e-05, "loss": 0.0898, "step": 53830 }, { "epoch": 25.81, "learning_rate": 5.123468695312734e-05, "loss": 0.0916, "step": 53840 }, { "epoch": 25.81, "learning_rate": 5.122033299626343e-05, "loss": 0.0982, "step": 53850 }, { "epoch": 25.82, "learning_rate": 5.1205978938765806e-05, "loss": 0.0889, "step": 53860 }, { "epoch": 25.82, "learning_rate": 5.119162478181819e-05, "loss": 0.0893, "step": 53870 }, { "epoch": 25.82, "learning_rate": 5.117727052660427e-05, "loss": 0.0874, "step": 53880 }, { "epoch": 25.83, "learning_rate": 5.116291617430774e-05, "loss": 0.0905, "step": 53890 }, { "epoch": 25.83, "learning_rate": 5.114856172611234e-05, "loss": 0.0911, "step": 53900 }, { "epoch": 25.84, "learning_rate": 5.113420718320181e-05, "loss": 0.0927, "step": 53910 }, { "epoch": 25.84, "learning_rate": 5.111985254675986e-05, "loss": 0.0857, "step": 53920 }, { "epoch": 25.85, "learning_rate": 5.110549781797023e-05, "loss": 0.0855, "step": 53930 }, { "epoch": 25.85, "learning_rate": 5.1091142998016697e-05, "loss": 0.0918, "step": 53940 }, { "epoch": 25.86, "learning_rate": 5.107678808808298e-05, "loss": 0.0863, "step": 53950 }, { "epoch": 25.86, "learning_rate": 5.1062433089352894e-05, "loss": 0.1009, "step": 53960 }, { "epoch": 25.87, "learning_rate": 5.104807800301015e-05, "loss": 0.0976, "step": 53970 }, { "epoch": 25.87, "learning_rate": 5.103372283023859e-05, "loss": 0.0969, "step": 53980 }, { "epoch": 25.88, "learning_rate": 5.1019367572221955e-05, "loss": 0.0892, "step": 53990 }, { "epoch": 25.88, "learning_rate": 5.100501223014408e-05, "loss": 0.1088, "step": 54000 }, { "epoch": 25.88, "learning_rate": 5.099065680518872e-05, "loss": 0.0782, "step": 54010 }, { "epoch": 25.89, "learning_rate": 5.097630129853971e-05, "loss": 0.0957, "step": 54020 }, { "epoch": 25.89, "learning_rate": 5.096194571138088e-05, "loss": 0.0859, "step": 54030 }, { "epoch": 25.9, "learning_rate": 5.0947590044896e-05, "loss": 0.0936, "step": 54040 }, { "epoch": 25.9, "learning_rate": 5.093323430026894e-05, "loss": 0.0889, "step": 54050 }, { "epoch": 25.91, "learning_rate": 5.091887847868352e-05, "loss": 0.0807, "step": 54060 }, { "epoch": 25.91, "learning_rate": 5.09045225813236e-05, "loss": 0.0808, "step": 54070 }, { "epoch": 25.92, "learning_rate": 5.089016660937299e-05, "loss": 0.0841, "step": 54080 }, { "epoch": 25.92, "learning_rate": 5.087581056401557e-05, "loss": 0.0867, "step": 54090 }, { "epoch": 25.93, "learning_rate": 5.0861454446435195e-05, "loss": 0.0857, "step": 54100 }, { "epoch": 25.93, "learning_rate": 5.084709825781572e-05, "loss": 0.0898, "step": 54110 }, { "epoch": 25.93, "learning_rate": 5.0832741999341015e-05, "loss": 0.0995, "step": 54120 }, { "epoch": 25.94, "learning_rate": 5.0818385672194966e-05, "loss": 0.0953, "step": 54130 }, { "epoch": 25.94, "learning_rate": 5.080402927756146e-05, "loss": 0.086, "step": 54140 }, { "epoch": 25.95, "learning_rate": 5.0789672816624366e-05, "loss": 0.0998, "step": 54150 }, { "epoch": 26.0, "learning_rate": 5.07753162905676e-05, "loss": 0.0974, "step": 54160 }, { "epoch": 26.01, "learning_rate": 5.076095970057504e-05, "loss": 0.0886, "step": 54170 }, { "epoch": 26.01, "learning_rate": 5.0746603047830596e-05, "loss": 0.0875, "step": 54180 }, { "epoch": 26.01, "learning_rate": 5.073224633351819e-05, "loss": 0.0927, "step": 54190 }, { "epoch": 26.02, "learning_rate": 5.071788955882171e-05, "loss": 0.0937, "step": 54200 }, { "epoch": 26.02, "learning_rate": 5.0703532724925105e-05, "loss": 0.0797, "step": 54210 }, { "epoch": 26.03, "learning_rate": 5.068917583301227e-05, "loss": 0.0971, "step": 54220 }, { "epoch": 26.03, "learning_rate": 5.067481888426715e-05, "loss": 0.0837, "step": 54230 }, { "epoch": 26.04, "learning_rate": 5.0660461879873676e-05, "loss": 0.0856, "step": 54240 }, { "epoch": 26.04, "learning_rate": 5.0646104821015795e-05, "loss": 0.0871, "step": 54250 }, { "epoch": 26.05, "learning_rate": 5.063174770887743e-05, "loss": 0.0928, "step": 54260 }, { "epoch": 26.05, "learning_rate": 5.061739054464254e-05, "loss": 0.0892, "step": 54270 }, { "epoch": 26.06, "learning_rate": 5.0603033329495085e-05, "loss": 0.0852, "step": 54280 }, { "epoch": 26.06, "learning_rate": 5.0588676064619e-05, "loss": 0.0837, "step": 54290 }, { "epoch": 26.06, "learning_rate": 5.0574318751198254e-05, "loss": 0.0907, "step": 54300 }, { "epoch": 26.07, "learning_rate": 5.05599613904168e-05, "loss": 0.0805, "step": 54310 }, { "epoch": 26.07, "learning_rate": 5.0545603983458636e-05, "loss": 0.0816, "step": 54320 }, { "epoch": 26.08, "learning_rate": 5.053124653150769e-05, "loss": 0.0806, "step": 54330 }, { "epoch": 26.08, "learning_rate": 5.051688903574798e-05, "loss": 0.0822, "step": 54340 }, { "epoch": 26.09, "learning_rate": 5.050253149736344e-05, "loss": 0.0807, "step": 54350 }, { "epoch": 26.09, "learning_rate": 5.0488173917538094e-05, "loss": 0.0993, "step": 54360 }, { "epoch": 26.1, "learning_rate": 5.047381629745591e-05, "loss": 0.0907, "step": 54370 }, { "epoch": 26.1, "learning_rate": 5.045945863830087e-05, "loss": 0.0854, "step": 54380 }, { "epoch": 26.11, "learning_rate": 5.0445100941256976e-05, "loss": 0.0854, "step": 54390 }, { "epoch": 26.11, "learning_rate": 5.04307432075082e-05, "loss": 0.0837, "step": 54400 }, { "epoch": 26.12, "learning_rate": 5.041638543823859e-05, "loss": 0.0894, "step": 54410 }, { "epoch": 26.12, "learning_rate": 5.0402027634632073e-05, "loss": 0.0845, "step": 54420 }, { "epoch": 26.12, "learning_rate": 5.038766979787274e-05, "loss": 0.0801, "step": 54430 }, { "epoch": 26.13, "learning_rate": 5.037331192914454e-05, "loss": 0.1129, "step": 54440 }, { "epoch": 26.13, "learning_rate": 5.035895402963149e-05, "loss": 0.0856, "step": 54450 }, { "epoch": 26.14, "learning_rate": 5.034459610051759e-05, "loss": 0.095, "step": 54460 }, { "epoch": 26.14, "learning_rate": 5.03302381429869e-05, "loss": 0.0853, "step": 54470 }, { "epoch": 26.15, "learning_rate": 5.031588015822339e-05, "loss": 0.091, "step": 54480 }, { "epoch": 26.15, "learning_rate": 5.030152214741111e-05, "loss": 0.1032, "step": 54490 }, { "epoch": 26.16, "learning_rate": 5.028716411173405e-05, "loss": 0.0896, "step": 54500 }, { "epoch": 26.16, "learning_rate": 5.027280605237626e-05, "loss": 0.0891, "step": 54510 }, { "epoch": 26.17, "learning_rate": 5.025844797052176e-05, "loss": 0.0794, "step": 54520 }, { "epoch": 26.17, "learning_rate": 5.0244089867354574e-05, "loss": 0.0899, "step": 54530 }, { "epoch": 26.17, "learning_rate": 5.0229731744058714e-05, "loss": 0.087, "step": 54540 }, { "epoch": 26.18, "learning_rate": 5.021537360181824e-05, "loss": 0.0815, "step": 54550 }, { "epoch": 26.18, "learning_rate": 5.020101544181716e-05, "loss": 0.0919, "step": 54560 }, { "epoch": 26.19, "learning_rate": 5.018665726523951e-05, "loss": 0.0855, "step": 54570 }, { "epoch": 26.19, "learning_rate": 5.017229907326934e-05, "loss": 0.0815, "step": 54580 }, { "epoch": 26.2, "learning_rate": 5.015794086709068e-05, "loss": 0.0825, "step": 54590 }, { "epoch": 26.2, "learning_rate": 5.0143582647887556e-05, "loss": 0.0839, "step": 54600 }, { "epoch": 26.21, "learning_rate": 5.012922441684401e-05, "loss": 0.0907, "step": 54610 }, { "epoch": 26.21, "learning_rate": 5.01148661751441e-05, "loss": 0.0886, "step": 54620 }, { "epoch": 26.22, "learning_rate": 5.010050792397184e-05, "loss": 0.0861, "step": 54630 }, { "epoch": 26.22, "learning_rate": 5.008614966451128e-05, "loss": 0.0832, "step": 54640 }, { "epoch": 26.22, "learning_rate": 5.007179139794645e-05, "loss": 0.082, "step": 54650 }, { "epoch": 26.23, "learning_rate": 5.0057433125461426e-05, "loss": 0.0867, "step": 54660 }, { "epoch": 26.23, "learning_rate": 5.0043074848240224e-05, "loss": 0.085, "step": 54670 }, { "epoch": 26.24, "learning_rate": 5.002871656746689e-05, "loss": 0.0932, "step": 54680 }, { "epoch": 26.24, "learning_rate": 5.001435828432546e-05, "loss": 0.091, "step": 54690 }, { "epoch": 26.25, "learning_rate": 5e-05, "loss": 0.0892, "step": 54700 }, { "epoch": 26.25, "learning_rate": 4.998564171567454e-05, "loss": 0.097, "step": 54710 }, { "epoch": 26.26, "learning_rate": 4.997128343253312e-05, "loss": 0.0818, "step": 54720 }, { "epoch": 26.26, "learning_rate": 4.995692515175979e-05, "loss": 0.0845, "step": 54730 }, { "epoch": 26.27, "learning_rate": 4.994256687453858e-05, "loss": 0.0805, "step": 54740 }, { "epoch": 26.27, "learning_rate": 4.9928208602053546e-05, "loss": 0.0865, "step": 54750 }, { "epoch": 26.28, "learning_rate": 4.991385033548874e-05, "loss": 0.0761, "step": 54760 }, { "epoch": 26.28, "learning_rate": 4.989949207602818e-05, "loss": 0.0995, "step": 54770 }, { "epoch": 26.28, "learning_rate": 4.9885133824855915e-05, "loss": 0.0859, "step": 54780 }, { "epoch": 26.29, "learning_rate": 4.9870775583155996e-05, "loss": 0.085, "step": 54790 }, { "epoch": 26.29, "learning_rate": 4.9856417352112456e-05, "loss": 0.0959, "step": 54800 }, { "epoch": 26.3, "learning_rate": 4.984205913290933e-05, "loss": 0.093, "step": 54810 }, { "epoch": 26.3, "learning_rate": 4.982770092673066e-05, "loss": 0.0954, "step": 54820 }, { "epoch": 26.31, "learning_rate": 4.98133427347605e-05, "loss": 0.0852, "step": 54830 }, { "epoch": 26.31, "learning_rate": 4.979898455818285e-05, "loss": 0.0818, "step": 54840 }, { "epoch": 26.32, "learning_rate": 4.9784626398181775e-05, "loss": 0.0821, "step": 54850 }, { "epoch": 26.32, "learning_rate": 4.977026825594129e-05, "loss": 0.0894, "step": 54860 }, { "epoch": 26.33, "learning_rate": 4.9755910132645444e-05, "loss": 0.083, "step": 54870 }, { "epoch": 26.33, "learning_rate": 4.974155202947825e-05, "loss": 0.0875, "step": 54880 }, { "epoch": 26.33, "learning_rate": 4.972719394762374e-05, "loss": 0.0848, "step": 54890 }, { "epoch": 26.34, "learning_rate": 4.971283588826597e-05, "loss": 0.0844, "step": 54900 }, { "epoch": 26.34, "learning_rate": 4.96984778525889e-05, "loss": 0.0904, "step": 54910 }, { "epoch": 26.35, "learning_rate": 4.968411984177662e-05, "loss": 0.0779, "step": 54920 }, { "epoch": 26.35, "learning_rate": 4.9669761857013106e-05, "loss": 0.0908, "step": 54930 }, { "epoch": 26.36, "learning_rate": 4.965540389948242e-05, "loss": 0.0835, "step": 54940 }, { "epoch": 26.36, "learning_rate": 4.9641045970368525e-05, "loss": 0.083, "step": 54950 }, { "epoch": 26.37, "learning_rate": 4.9626688070855476e-05, "loss": 0.0831, "step": 54960 }, { "epoch": 26.37, "learning_rate": 4.961233020212729e-05, "loss": 0.0935, "step": 54970 }, { "epoch": 26.38, "learning_rate": 4.959797236536793e-05, "loss": 0.0934, "step": 54980 }, { "epoch": 26.38, "learning_rate": 4.958361456176143e-05, "loss": 0.0825, "step": 54990 }, { "epoch": 26.38, "learning_rate": 4.95692567924918e-05, "loss": 0.0842, "step": 55000 }, { "epoch": 26.39, "learning_rate": 4.9554899058743056e-05, "loss": 0.094, "step": 55010 }, { "epoch": 26.39, "learning_rate": 4.954054136169914e-05, "loss": 0.0912, "step": 55020 }, { "epoch": 26.4, "learning_rate": 4.9526183702544096e-05, "loss": 0.0879, "step": 55030 }, { "epoch": 26.4, "learning_rate": 4.9511826082461904e-05, "loss": 0.0858, "step": 55040 }, { "epoch": 26.41, "learning_rate": 4.949746850263657e-05, "loss": 0.089, "step": 55050 }, { "epoch": 26.41, "learning_rate": 4.9483110964252036e-05, "loss": 0.0811, "step": 55060 }, { "epoch": 26.42, "learning_rate": 4.9468753468492305e-05, "loss": 0.088, "step": 55070 }, { "epoch": 26.42, "learning_rate": 4.9454396016541396e-05, "loss": 0.0873, "step": 55080 }, { "epoch": 26.43, "learning_rate": 4.944003860958321e-05, "loss": 0.0942, "step": 55090 }, { "epoch": 26.43, "learning_rate": 4.942568124880175e-05, "loss": 0.0928, "step": 55100 }, { "epoch": 26.44, "learning_rate": 4.9411323935381006e-05, "loss": 0.0855, "step": 55110 }, { "epoch": 26.44, "learning_rate": 4.9396966670504934e-05, "loss": 0.0784, "step": 55120 }, { "epoch": 26.44, "learning_rate": 4.938260945535747e-05, "loss": 0.0884, "step": 55130 }, { "epoch": 26.45, "learning_rate": 4.936825229112256e-05, "loss": 0.0772, "step": 55140 }, { "epoch": 26.45, "learning_rate": 4.935389517898422e-05, "loss": 0.0898, "step": 55150 }, { "epoch": 26.46, "learning_rate": 4.9339538120126336e-05, "loss": 0.0961, "step": 55160 }, { "epoch": 26.46, "learning_rate": 4.932518111573286e-05, "loss": 0.0924, "step": 55170 }, { "epoch": 26.47, "learning_rate": 4.931082416698772e-05, "loss": 0.0804, "step": 55180 }, { "epoch": 26.47, "learning_rate": 4.929646727507491e-05, "loss": 0.0921, "step": 55190 }, { "epoch": 26.48, "learning_rate": 4.92821104411783e-05, "loss": 0.0864, "step": 55200 }, { "epoch": 26.48, "learning_rate": 4.9267753666481824e-05, "loss": 0.08, "step": 55210 }, { "epoch": 26.49, "learning_rate": 4.9253396952169395e-05, "loss": 0.0906, "step": 55220 }, { "epoch": 26.49, "learning_rate": 4.923904029942498e-05, "loss": 0.0906, "step": 55230 }, { "epoch": 26.49, "learning_rate": 4.9224683709432414e-05, "loss": 0.0859, "step": 55240 }, { "epoch": 26.5, "learning_rate": 4.921032718337564e-05, "loss": 0.0813, "step": 55250 }, { "epoch": 26.5, "learning_rate": 4.919597072243856e-05, "loss": 0.0906, "step": 55260 }, { "epoch": 26.51, "learning_rate": 4.918161432780504e-05, "loss": 0.0892, "step": 55270 }, { "epoch": 26.51, "learning_rate": 4.9167258000658996e-05, "loss": 0.0918, "step": 55280 }, { "epoch": 26.52, "learning_rate": 4.915290174218429e-05, "loss": 0.0838, "step": 55290 }, { "epoch": 26.52, "learning_rate": 4.913854555356483e-05, "loss": 0.0837, "step": 55300 }, { "epoch": 26.53, "learning_rate": 4.912418943598445e-05, "loss": 0.0891, "step": 55310 }, { "epoch": 26.53, "learning_rate": 4.910983339062702e-05, "loss": 0.096, "step": 55320 }, { "epoch": 26.54, "learning_rate": 4.909547741867642e-05, "loss": 0.0856, "step": 55330 }, { "epoch": 26.54, "learning_rate": 4.908112152131649e-05, "loss": 0.0928, "step": 55340 }, { "epoch": 26.54, "learning_rate": 4.906676569973107e-05, "loss": 0.0896, "step": 55350 }, { "epoch": 26.55, "learning_rate": 4.905240995510401e-05, "loss": 0.0877, "step": 55360 }, { "epoch": 26.55, "learning_rate": 4.903805428861915e-05, "loss": 0.0913, "step": 55370 }, { "epoch": 26.56, "learning_rate": 4.9023698701460304e-05, "loss": 0.082, "step": 55380 }, { "epoch": 26.56, "learning_rate": 4.900934319481129e-05, "loss": 0.0881, "step": 55390 }, { "epoch": 26.57, "learning_rate": 4.8994987769855946e-05, "loss": 0.084, "step": 55400 }, { "epoch": 26.57, "learning_rate": 4.8980632427778056e-05, "loss": 0.0992, "step": 55410 }, { "epoch": 26.58, "learning_rate": 4.8966277169761425e-05, "loss": 0.0867, "step": 55420 }, { "epoch": 26.58, "learning_rate": 4.895192199698985e-05, "loss": 0.0884, "step": 55430 }, { "epoch": 26.59, "learning_rate": 4.893756691064714e-05, "loss": 0.0946, "step": 55440 }, { "epoch": 26.59, "learning_rate": 4.8923211911917036e-05, "loss": 0.0941, "step": 55450 }, { "epoch": 26.6, "learning_rate": 4.890885700198332e-05, "loss": 0.0854, "step": 55460 }, { "epoch": 26.6, "learning_rate": 4.8894502182029775e-05, "loss": 0.0824, "step": 55470 }, { "epoch": 26.6, "learning_rate": 4.888014745324016e-05, "loss": 0.0881, "step": 55480 }, { "epoch": 26.61, "learning_rate": 4.88657928167982e-05, "loss": 0.0937, "step": 55490 }, { "epoch": 26.61, "learning_rate": 4.885143827388766e-05, "loss": 0.0864, "step": 55500 }, { "epoch": 26.62, "learning_rate": 4.883708382569227e-05, "loss": 0.0916, "step": 55510 }, { "epoch": 26.62, "learning_rate": 4.8822729473395745e-05, "loss": 0.0907, "step": 55520 }, { "epoch": 26.63, "learning_rate": 4.880837521818182e-05, "loss": 0.0858, "step": 55530 }, { "epoch": 26.63, "learning_rate": 4.879402106123419e-05, "loss": 0.086, "step": 55540 }, { "epoch": 26.64, "learning_rate": 4.877966700373659e-05, "loss": 0.0914, "step": 55550 }, { "epoch": 26.64, "learning_rate": 4.876531304687267e-05, "loss": 0.085, "step": 55560 }, { "epoch": 26.65, "learning_rate": 4.875095919182614e-05, "loss": 0.0805, "step": 55570 }, { "epoch": 26.65, "learning_rate": 4.8736605439780696e-05, "loss": 0.0898, "step": 55580 }, { "epoch": 26.65, "learning_rate": 4.8722251791919975e-05, "loss": 0.0997, "step": 55590 }, { "epoch": 26.66, "learning_rate": 4.870789824942765e-05, "loss": 0.0918, "step": 55600 }, { "epoch": 26.66, "learning_rate": 4.869354481348738e-05, "loss": 0.0883, "step": 55610 }, { "epoch": 26.67, "learning_rate": 4.8679191485282816e-05, "loss": 0.0953, "step": 55620 }, { "epoch": 26.67, "learning_rate": 4.866483826599757e-05, "loss": 0.089, "step": 55630 }, { "epoch": 26.68, "learning_rate": 4.865048515681527e-05, "loss": 0.0811, "step": 55640 }, { "epoch": 26.68, "learning_rate": 4.863613215891956e-05, "loss": 0.0882, "step": 55650 }, { "epoch": 26.69, "learning_rate": 4.862177927349402e-05, "loss": 0.0929, "step": 55660 }, { "epoch": 26.69, "learning_rate": 4.8607426501722256e-05, "loss": 0.0774, "step": 55670 }, { "epoch": 26.7, "learning_rate": 4.859307384478786e-05, "loss": 0.0887, "step": 55680 }, { "epoch": 26.7, "learning_rate": 4.857872130387444e-05, "loss": 0.087, "step": 55690 }, { "epoch": 26.7, "learning_rate": 4.856436888016549e-05, "loss": 0.0918, "step": 55700 }, { "epoch": 26.71, "learning_rate": 4.855001657484463e-05, "loss": 0.0768, "step": 55710 }, { "epoch": 26.71, "learning_rate": 4.853566438909539e-05, "loss": 0.0905, "step": 55720 }, { "epoch": 26.72, "learning_rate": 4.852131232410135e-05, "loss": 0.0839, "step": 55730 }, { "epoch": 26.72, "learning_rate": 4.850696038104597e-05, "loss": 0.0945, "step": 55740 }, { "epoch": 26.73, "learning_rate": 4.849260856111282e-05, "loss": 0.0774, "step": 55750 }, { "epoch": 26.73, "learning_rate": 4.847825686548542e-05, "loss": 0.0888, "step": 55760 }, { "epoch": 26.74, "learning_rate": 4.8463905295347225e-05, "loss": 0.0854, "step": 55770 }, { "epoch": 26.74, "learning_rate": 4.844955385188173e-05, "loss": 0.0855, "step": 55780 }, { "epoch": 26.75, "learning_rate": 4.8435202536272455e-05, "loss": 0.0909, "step": 55790 }, { "epoch": 26.75, "learning_rate": 4.842085134970287e-05, "loss": 0.0984, "step": 55800 }, { "epoch": 26.76, "learning_rate": 4.840650029335638e-05, "loss": 0.0866, "step": 55810 }, { "epoch": 26.76, "learning_rate": 4.839214936841645e-05, "loss": 0.0801, "step": 55820 }, { "epoch": 26.76, "learning_rate": 4.837779857606655e-05, "loss": 0.0887, "step": 55830 }, { "epoch": 26.77, "learning_rate": 4.8363447917490104e-05, "loss": 0.0851, "step": 55840 }, { "epoch": 26.77, "learning_rate": 4.834909739387048e-05, "loss": 0.0866, "step": 55850 }, { "epoch": 26.78, "learning_rate": 4.833474700639109e-05, "loss": 0.0875, "step": 55860 }, { "epoch": 26.78, "learning_rate": 4.8320396756235394e-05, "loss": 0.0973, "step": 55870 }, { "epoch": 26.79, "learning_rate": 4.8306046644586686e-05, "loss": 0.0845, "step": 55880 }, { "epoch": 26.79, "learning_rate": 4.829169667262838e-05, "loss": 0.0881, "step": 55890 }, { "epoch": 26.8, "learning_rate": 4.827734684154383e-05, "loss": 0.0912, "step": 55900 }, { "epoch": 26.8, "learning_rate": 4.8262997152516394e-05, "loss": 0.0896, "step": 55910 }, { "epoch": 26.81, "learning_rate": 4.8248647606729374e-05, "loss": 0.0817, "step": 55920 }, { "epoch": 26.81, "learning_rate": 4.8234298205366114e-05, "loss": 0.0746, "step": 55930 }, { "epoch": 26.81, "learning_rate": 4.821994894960993e-05, "loss": 0.095, "step": 55940 }, { "epoch": 26.82, "learning_rate": 4.8205599840644105e-05, "loss": 0.0801, "step": 55950 }, { "epoch": 26.82, "learning_rate": 4.819125087965193e-05, "loss": 0.0877, "step": 55960 }, { "epoch": 26.83, "learning_rate": 4.8176902067816685e-05, "loss": 0.0864, "step": 55970 }, { "epoch": 26.83, "learning_rate": 4.816255340632165e-05, "loss": 0.0894, "step": 55980 }, { "epoch": 26.84, "learning_rate": 4.8148204896350036e-05, "loss": 0.0861, "step": 55990 }, { "epoch": 26.84, "learning_rate": 4.8133856539085096e-05, "loss": 0.0969, "step": 56000 }, { "epoch": 26.85, "learning_rate": 4.8119508335710064e-05, "loss": 0.0855, "step": 56010 }, { "epoch": 26.85, "learning_rate": 4.8105160287408164e-05, "loss": 0.0965, "step": 56020 }, { "epoch": 26.86, "learning_rate": 4.809081239536256e-05, "loss": 0.09, "step": 56030 }, { "epoch": 26.86, "learning_rate": 4.807646466075646e-05, "loss": 0.0899, "step": 56040 }, { "epoch": 26.86, "learning_rate": 4.8062117084773044e-05, "loss": 0.0801, "step": 56050 }, { "epoch": 26.87, "learning_rate": 4.8047769668595455e-05, "loss": 0.0806, "step": 56060 }, { "epoch": 26.87, "learning_rate": 4.8033422413406844e-05, "loss": 0.0861, "step": 56070 }, { "epoch": 26.88, "learning_rate": 4.801907532039035e-05, "loss": 0.0752, "step": 56080 }, { "epoch": 26.88, "learning_rate": 4.8004728390729103e-05, "loss": 0.0944, "step": 56090 }, { "epoch": 26.89, "learning_rate": 4.799038162560619e-05, "loss": 0.0946, "step": 56100 }, { "epoch": 26.89, "learning_rate": 4.797603502620471e-05, "loss": 0.0805, "step": 56110 }, { "epoch": 26.9, "learning_rate": 4.796168859370777e-05, "loss": 0.097, "step": 56120 }, { "epoch": 26.9, "learning_rate": 4.794734232929839e-05, "loss": 0.0816, "step": 56130 }, { "epoch": 26.91, "learning_rate": 4.793299623415965e-05, "loss": 0.0782, "step": 56140 }, { "epoch": 26.91, "learning_rate": 4.791865030947458e-05, "loss": 0.0924, "step": 56150 }, { "epoch": 26.91, "learning_rate": 4.790430455642623e-05, "loss": 0.0927, "step": 56160 }, { "epoch": 26.92, "learning_rate": 4.7889958976197566e-05, "loss": 0.0884, "step": 56170 }, { "epoch": 26.92, "learning_rate": 4.787561356997161e-05, "loss": 0.0882, "step": 56180 }, { "epoch": 26.93, "learning_rate": 4.786126833893135e-05, "loss": 0.083, "step": 56190 }, { "epoch": 26.93, "learning_rate": 4.784692328425974e-05, "loss": 0.0845, "step": 56200 }, { "epoch": 26.94, "learning_rate": 4.783257840713973e-05, "loss": 0.0826, "step": 56210 }, { "epoch": 26.94, "learning_rate": 4.7818233708754255e-05, "loss": 0.076, "step": 56220 }, { "epoch": 26.95, "learning_rate": 4.7803889190286265e-05, "loss": 0.0923, "step": 56230 }, { "epoch": 26.95, "learning_rate": 4.778954485291864e-05, "loss": 0.1028, "step": 56240 }, { "epoch": 27.0, "learning_rate": 4.7775200697834277e-05, "loss": 0.0913, "step": 56250 }, { "epoch": 27.01, "learning_rate": 4.7760856726216056e-05, "loss": 0.0891, "step": 56260 }, { "epoch": 27.01, "learning_rate": 4.774651293924686e-05, "loss": 0.09, "step": 56270 }, { "epoch": 27.02, "learning_rate": 4.773216933810952e-05, "loss": 0.0996, "step": 56280 }, { "epoch": 27.02, "learning_rate": 4.771782592398686e-05, "loss": 0.0733, "step": 56290 }, { "epoch": 27.03, "learning_rate": 4.770348269806172e-05, "loss": 0.0782, "step": 56300 }, { "epoch": 27.03, "learning_rate": 4.768913966151687e-05, "loss": 0.0888, "step": 56310 }, { "epoch": 27.04, "learning_rate": 4.767479681553513e-05, "loss": 0.0771, "step": 56320 }, { "epoch": 27.04, "learning_rate": 4.766045416129924e-05, "loss": 0.0867, "step": 56330 }, { "epoch": 27.05, "learning_rate": 4.764611169999199e-05, "loss": 0.0811, "step": 56340 }, { "epoch": 27.05, "learning_rate": 4.763176943279609e-05, "loss": 0.0926, "step": 56350 }, { "epoch": 27.05, "learning_rate": 4.761742736089426e-05, "loss": 0.0956, "step": 56360 }, { "epoch": 27.06, "learning_rate": 4.760308548546925e-05, "loss": 0.0816, "step": 56370 }, { "epoch": 27.06, "learning_rate": 4.7588743807703674e-05, "loss": 0.0905, "step": 56380 }, { "epoch": 27.07, "learning_rate": 4.7574402328780265e-05, "loss": 0.0876, "step": 56390 }, { "epoch": 27.07, "learning_rate": 4.7560061049881674e-05, "loss": 0.0893, "step": 56400 }, { "epoch": 27.08, "learning_rate": 4.754571997219054e-05, "loss": 0.0804, "step": 56410 }, { "epoch": 27.08, "learning_rate": 4.753137909688946e-05, "loss": 0.0843, "step": 56420 }, { "epoch": 27.09, "learning_rate": 4.7517038425161076e-05, "loss": 0.0827, "step": 56430 }, { "epoch": 27.09, "learning_rate": 4.7502697958187956e-05, "loss": 0.0788, "step": 56440 }, { "epoch": 27.1, "learning_rate": 4.748835769715271e-05, "loss": 0.0891, "step": 56450 }, { "epoch": 27.1, "learning_rate": 4.7474017643237836e-05, "loss": 0.0894, "step": 56460 }, { "epoch": 27.1, "learning_rate": 4.745967779762592e-05, "loss": 0.0884, "step": 56470 }, { "epoch": 27.11, "learning_rate": 4.7445338161499505e-05, "loss": 0.0772, "step": 56480 }, { "epoch": 27.11, "learning_rate": 4.7430998736041024e-05, "loss": 0.0799, "step": 56490 }, { "epoch": 27.12, "learning_rate": 4.7416659522433006e-05, "loss": 0.0821, "step": 56500 }, { "epoch": 27.12, "learning_rate": 4.740232052185793e-05, "loss": 0.083, "step": 56510 }, { "epoch": 27.13, "learning_rate": 4.738798173549826e-05, "loss": 0.0923, "step": 56520 }, { "epoch": 27.13, "learning_rate": 4.737364316453639e-05, "loss": 0.0877, "step": 56530 }, { "epoch": 27.14, "learning_rate": 4.7359304810154756e-05, "loss": 0.0795, "step": 56540 }, { "epoch": 27.14, "learning_rate": 4.734496667353579e-05, "loss": 0.0839, "step": 56550 }, { "epoch": 27.15, "learning_rate": 4.733062875586182e-05, "loss": 0.0908, "step": 56560 }, { "epoch": 27.15, "learning_rate": 4.731629105831524e-05, "loss": 0.0872, "step": 56570 }, { "epoch": 27.15, "learning_rate": 4.730195358207838e-05, "loss": 0.0919, "step": 56580 }, { "epoch": 27.16, "learning_rate": 4.728761632833361e-05, "loss": 0.0818, "step": 56590 }, { "epoch": 27.16, "learning_rate": 4.727327929826319e-05, "loss": 0.0898, "step": 56600 }, { "epoch": 27.17, "learning_rate": 4.7258942493049426e-05, "loss": 0.0931, "step": 56610 }, { "epoch": 27.17, "learning_rate": 4.7244605913874586e-05, "loss": 0.0789, "step": 56620 }, { "epoch": 27.18, "learning_rate": 4.723026956192096e-05, "loss": 0.0972, "step": 56630 }, { "epoch": 27.18, "learning_rate": 4.721593343837073e-05, "loss": 0.0924, "step": 56640 }, { "epoch": 27.19, "learning_rate": 4.720159754440614e-05, "loss": 0.0836, "step": 56650 }, { "epoch": 27.19, "learning_rate": 4.718726188120939e-05, "loss": 0.0841, "step": 56660 }, { "epoch": 27.2, "learning_rate": 4.7172926449962646e-05, "loss": 0.0781, "step": 56670 }, { "epoch": 27.2, "learning_rate": 4.715859125184807e-05, "loss": 0.0868, "step": 56680 }, { "epoch": 27.21, "learning_rate": 4.71442562880478e-05, "loss": 0.0897, "step": 56690 }, { "epoch": 27.21, "learning_rate": 4.712992155974397e-05, "loss": 0.0884, "step": 56700 }, { "epoch": 27.21, "learning_rate": 4.7115587068118665e-05, "loss": 0.0816, "step": 56710 }, { "epoch": 27.22, "learning_rate": 4.7101252814353973e-05, "loss": 0.0835, "step": 56720 }, { "epoch": 27.22, "learning_rate": 4.708691879963196e-05, "loss": 0.0888, "step": 56730 }, { "epoch": 27.23, "learning_rate": 4.707258502513466e-05, "loss": 0.0888, "step": 56740 }, { "epoch": 27.23, "learning_rate": 4.705825149204409e-05, "loss": 0.0784, "step": 56750 }, { "epoch": 27.24, "learning_rate": 4.704391820154227e-05, "loss": 0.0759, "step": 56760 }, { "epoch": 27.24, "learning_rate": 4.702958515481118e-05, "loss": 0.0926, "step": 56770 }, { "epoch": 27.25, "learning_rate": 4.701525235303276e-05, "loss": 0.0885, "step": 56780 }, { "epoch": 27.25, "learning_rate": 4.700091979738897e-05, "loss": 0.0874, "step": 56790 }, { "epoch": 27.26, "learning_rate": 4.6986587489061746e-05, "loss": 0.0863, "step": 56800 }, { "epoch": 27.26, "learning_rate": 4.697225542923295e-05, "loss": 0.081, "step": 56810 }, { "epoch": 27.26, "learning_rate": 4.69579236190845e-05, "loss": 0.0886, "step": 56820 }, { "epoch": 27.27, "learning_rate": 4.694359205979823e-05, "loss": 0.0933, "step": 56830 }, { "epoch": 27.27, "learning_rate": 4.692926075255601e-05, "loss": 0.0823, "step": 56840 }, { "epoch": 27.28, "learning_rate": 4.691492969853963e-05, "loss": 0.0849, "step": 56850 }, { "epoch": 27.28, "learning_rate": 4.690059889893089e-05, "loss": 0.0868, "step": 56860 }, { "epoch": 27.29, "learning_rate": 4.688626835491158e-05, "loss": 0.0794, "step": 56870 }, { "epoch": 27.29, "learning_rate": 4.687193806766346e-05, "loss": 0.0807, "step": 56880 }, { "epoch": 27.3, "learning_rate": 4.685760803836825e-05, "loss": 0.09, "step": 56890 }, { "epoch": 27.3, "learning_rate": 4.684327826820767e-05, "loss": 0.0728, "step": 56900 }, { "epoch": 27.31, "learning_rate": 4.682894875836342e-05, "loss": 0.0741, "step": 56910 }, { "epoch": 27.31, "learning_rate": 4.681461951001714e-05, "loss": 0.0993, "step": 56920 }, { "epoch": 27.31, "learning_rate": 4.6800290524350514e-05, "loss": 0.09, "step": 56930 }, { "epoch": 27.32, "learning_rate": 4.678596180254515e-05, "loss": 0.0804, "step": 56940 }, { "epoch": 27.32, "learning_rate": 4.6771633345782676e-05, "loss": 0.0914, "step": 56950 }, { "epoch": 27.33, "learning_rate": 4.675730515524465e-05, "loss": 0.0853, "step": 56960 }, { "epoch": 27.33, "learning_rate": 4.674297723211264e-05, "loss": 0.0901, "step": 56970 }, { "epoch": 27.34, "learning_rate": 4.67286495775682e-05, "loss": 0.0924, "step": 56980 }, { "epoch": 27.34, "learning_rate": 4.671432219279283e-05, "loss": 0.0793, "step": 56990 }, { "epoch": 27.35, "learning_rate": 4.669999507896803e-05, "loss": 0.0844, "step": 57000 }, { "epoch": 27.35, "learning_rate": 4.668566823727528e-05, "loss": 0.087, "step": 57010 }, { "epoch": 27.36, "learning_rate": 4.6671341668896023e-05, "loss": 0.0888, "step": 57020 }, { "epoch": 27.36, "learning_rate": 4.665701537501169e-05, "loss": 0.0931, "step": 57030 }, { "epoch": 27.37, "learning_rate": 4.664268935680369e-05, "loss": 0.0839, "step": 57040 }, { "epoch": 27.37, "learning_rate": 4.662836361545339e-05, "loss": 0.0783, "step": 57050 }, { "epoch": 27.37, "learning_rate": 4.661403815214218e-05, "loss": 0.1076, "step": 57060 }, { "epoch": 27.38, "learning_rate": 4.659971296805137e-05, "loss": 0.0875, "step": 57070 }, { "epoch": 27.38, "learning_rate": 4.6585388064362276e-05, "loss": 0.0838, "step": 57080 }, { "epoch": 27.39, "learning_rate": 4.657106344225622e-05, "loss": 0.0841, "step": 57090 }, { "epoch": 27.39, "learning_rate": 4.6556739102914416e-05, "loss": 0.0821, "step": 57100 }, { "epoch": 27.4, "learning_rate": 4.6542415047518146e-05, "loss": 0.0836, "step": 57110 }, { "epoch": 27.4, "learning_rate": 4.652809127724862e-05, "loss": 0.0871, "step": 57120 }, { "epoch": 27.41, "learning_rate": 4.6513767793287065e-05, "loss": 0.0797, "step": 57130 }, { "epoch": 27.41, "learning_rate": 4.649944459681459e-05, "loss": 0.0791, "step": 57140 }, { "epoch": 27.42, "learning_rate": 4.6485121689012396e-05, "loss": 0.0847, "step": 57150 }, { "epoch": 27.42, "learning_rate": 4.647079907106161e-05, "loss": 0.0866, "step": 57160 }, { "epoch": 27.42, "learning_rate": 4.6456476744143305e-05, "loss": 0.0745, "step": 57170 }, { "epoch": 27.43, "learning_rate": 4.6442154709438565e-05, "loss": 0.0916, "step": 57180 }, { "epoch": 27.43, "learning_rate": 4.6427832968128454e-05, "loss": 0.0871, "step": 57190 }, { "epoch": 27.44, "learning_rate": 4.6413511521394026e-05, "loss": 0.0955, "step": 57200 }, { "epoch": 27.44, "learning_rate": 4.639919037041624e-05, "loss": 0.102, "step": 57210 }, { "epoch": 27.45, "learning_rate": 4.63848695163761e-05, "loss": 0.0954, "step": 57220 }, { "epoch": 27.45, "learning_rate": 4.637054896045456e-05, "loss": 0.0758, "step": 57230 }, { "epoch": 27.46, "learning_rate": 4.635622870383258e-05, "loss": 0.0773, "step": 57240 }, { "epoch": 27.46, "learning_rate": 4.634190874769102e-05, "loss": 0.0882, "step": 57250 }, { "epoch": 27.47, "learning_rate": 4.632758909321077e-05, "loss": 0.0906, "step": 57260 }, { "epoch": 27.47, "learning_rate": 4.631326974157273e-05, "loss": 0.0805, "step": 57270 }, { "epoch": 27.47, "learning_rate": 4.629895069395769e-05, "loss": 0.0902, "step": 57280 }, { "epoch": 27.48, "learning_rate": 4.628463195154648e-05, "loss": 0.0899, "step": 57290 }, { "epoch": 27.48, "learning_rate": 4.627031351551985e-05, "loss": 0.0829, "step": 57300 }, { "epoch": 27.49, "learning_rate": 4.625599538705863e-05, "loss": 0.0842, "step": 57310 }, { "epoch": 27.49, "learning_rate": 4.624167756734347e-05, "loss": 0.0808, "step": 57320 }, { "epoch": 27.5, "learning_rate": 4.622736005755511e-05, "loss": 0.0886, "step": 57330 }, { "epoch": 27.5, "learning_rate": 4.621304285887426e-05, "loss": 0.0824, "step": 57340 }, { "epoch": 27.51, "learning_rate": 4.6198725972481526e-05, "loss": 0.0822, "step": 57350 }, { "epoch": 27.51, "learning_rate": 4.6184409399557566e-05, "loss": 0.0897, "step": 57360 }, { "epoch": 27.52, "learning_rate": 4.617009314128297e-05, "loss": 0.0875, "step": 57370 }, { "epoch": 27.52, "learning_rate": 4.615577719883835e-05, "loss": 0.0979, "step": 57380 }, { "epoch": 27.53, "learning_rate": 4.6141461573404204e-05, "loss": 0.0786, "step": 57390 }, { "epoch": 27.53, "learning_rate": 4.612714626616109e-05, "loss": 0.0862, "step": 57400 }, { "epoch": 27.53, "learning_rate": 4.611283127828951e-05, "loss": 0.0935, "step": 57410 }, { "epoch": 27.54, "learning_rate": 4.609851661096992e-05, "loss": 0.0859, "step": 57420 }, { "epoch": 27.54, "learning_rate": 4.608420226538278e-05, "loss": 0.0817, "step": 57430 }, { "epoch": 27.55, "learning_rate": 4.60698882427085e-05, "loss": 0.0898, "step": 57440 }, { "epoch": 27.55, "learning_rate": 4.60555745441275e-05, "loss": 0.0982, "step": 57450 }, { "epoch": 27.56, "learning_rate": 4.604126117082011e-05, "loss": 0.0835, "step": 57460 }, { "epoch": 27.56, "learning_rate": 4.6026948123966676e-05, "loss": 0.0873, "step": 57470 }, { "epoch": 27.57, "learning_rate": 4.6012635404747534e-05, "loss": 0.0863, "step": 57480 }, { "epoch": 27.57, "learning_rate": 4.5998323014342964e-05, "loss": 0.0839, "step": 57490 }, { "epoch": 27.58, "learning_rate": 4.598401095393321e-05, "loss": 0.0822, "step": 57500 }, { "epoch": 27.58, "learning_rate": 4.596969922469851e-05, "loss": 0.0912, "step": 57510 }, { "epoch": 27.58, "learning_rate": 4.595538782781908e-05, "loss": 0.0919, "step": 57520 }, { "epoch": 27.59, "learning_rate": 4.594107676447507e-05, "loss": 0.0892, "step": 57530 }, { "epoch": 27.59, "learning_rate": 4.592676603584666e-05, "loss": 0.0817, "step": 57540 }, { "epoch": 27.6, "learning_rate": 4.591245564311394e-05, "loss": 0.0899, "step": 57550 }, { "epoch": 27.6, "learning_rate": 4.5898145587457054e-05, "loss": 0.093, "step": 57560 }, { "epoch": 27.61, "learning_rate": 4.588383587005601e-05, "loss": 0.0902, "step": 57570 }, { "epoch": 27.61, "learning_rate": 4.586952649209088e-05, "loss": 0.0866, "step": 57580 }, { "epoch": 27.62, "learning_rate": 4.585521745474168e-05, "loss": 0.0936, "step": 57590 }, { "epoch": 27.62, "learning_rate": 4.584090875918836e-05, "loss": 0.0799, "step": 57600 }, { "epoch": 27.63, "learning_rate": 4.58266004066109e-05, "loss": 0.0877, "step": 57610 }, { "epoch": 27.63, "learning_rate": 4.5812292398189226e-05, "loss": 0.0859, "step": 57620 }, { "epoch": 27.63, "learning_rate": 4.5797984735103236e-05, "loss": 0.0878, "step": 57630 }, { "epoch": 27.64, "learning_rate": 4.578367741853278e-05, "loss": 0.0942, "step": 57640 }, { "epoch": 27.64, "learning_rate": 4.576937044965771e-05, "loss": 0.089, "step": 57650 }, { "epoch": 27.65, "learning_rate": 4.575506382965784e-05, "loss": 0.0813, "step": 57660 }, { "epoch": 27.65, "learning_rate": 4.574075755971296e-05, "loss": 0.0839, "step": 57670 }, { "epoch": 27.66, "learning_rate": 4.572645164100281e-05, "loss": 0.0902, "step": 57680 }, { "epoch": 27.66, "learning_rate": 4.571214607470712e-05, "loss": 0.0873, "step": 57690 }, { "epoch": 27.67, "learning_rate": 4.56978408620056e-05, "loss": 0.0814, "step": 57700 }, { "epoch": 27.67, "learning_rate": 4.568353600407789e-05, "loss": 0.085, "step": 57710 }, { "epoch": 27.68, "learning_rate": 4.566923150210364e-05, "loss": 0.0811, "step": 57720 }, { "epoch": 27.68, "learning_rate": 4.565492735726246e-05, "loss": 0.0913, "step": 57730 }, { "epoch": 27.69, "learning_rate": 4.5640623570733945e-05, "loss": 0.0803, "step": 57740 }, { "epoch": 27.69, "learning_rate": 4.562632014369762e-05, "loss": 0.0826, "step": 57750 }, { "epoch": 27.69, "learning_rate": 4.5612017077333e-05, "loss": 0.0916, "step": 57760 }, { "epoch": 27.7, "learning_rate": 4.5597714372819624e-05, "loss": 0.0896, "step": 57770 }, { "epoch": 27.7, "learning_rate": 4.558341203133688e-05, "loss": 0.0793, "step": 57780 }, { "epoch": 27.71, "learning_rate": 4.556911005406424e-05, "loss": 0.0789, "step": 57790 }, { "epoch": 27.71, "learning_rate": 4.555480844218111e-05, "loss": 0.0871, "step": 57800 }, { "epoch": 27.72, "learning_rate": 4.5540507196866864e-05, "loss": 0.091, "step": 57810 }, { "epoch": 27.72, "learning_rate": 4.5526206319300793e-05, "loss": 0.0811, "step": 57820 }, { "epoch": 27.73, "learning_rate": 4.551190581066225e-05, "loss": 0.0923, "step": 57830 }, { "epoch": 27.73, "learning_rate": 4.54976056721305e-05, "loss": 0.0805, "step": 57840 }, { "epoch": 27.74, "learning_rate": 4.548330590488482e-05, "loss": 0.0852, "step": 57850 }, { "epoch": 27.74, "learning_rate": 4.546900651010438e-05, "loss": 0.1, "step": 57860 }, { "epoch": 27.74, "learning_rate": 4.545470748896839e-05, "loss": 0.0908, "step": 57870 }, { "epoch": 27.75, "learning_rate": 4.544040884265604e-05, "loss": 0.0809, "step": 57880 }, { "epoch": 27.75, "learning_rate": 4.542611057234639e-05, "loss": 0.0853, "step": 57890 }, { "epoch": 27.76, "learning_rate": 4.541181267921856e-05, "loss": 0.0964, "step": 57900 }, { "epoch": 27.76, "learning_rate": 4.539751516445163e-05, "loss": 0.0825, "step": 57910 }, { "epoch": 27.77, "learning_rate": 4.538321802922465e-05, "loss": 0.0896, "step": 57920 }, { "epoch": 27.77, "learning_rate": 4.536892127471656e-05, "loss": 0.0873, "step": 57930 }, { "epoch": 27.78, "learning_rate": 4.5354624902106365e-05, "loss": 0.0884, "step": 57940 }, { "epoch": 27.78, "learning_rate": 4.534032891257303e-05, "loss": 0.0826, "step": 57950 }, { "epoch": 27.79, "learning_rate": 4.532603330729541e-05, "loss": 0.0795, "step": 57960 }, { "epoch": 27.79, "learning_rate": 4.531173808745239e-05, "loss": 0.0864, "step": 57970 }, { "epoch": 27.79, "learning_rate": 4.529744325422284e-05, "loss": 0.0895, "step": 57980 }, { "epoch": 27.8, "learning_rate": 4.528314880878557e-05, "loss": 0.0829, "step": 57990 }, { "epoch": 27.8, "learning_rate": 4.5268854752319326e-05, "loss": 0.0792, "step": 58000 }, { "epoch": 27.81, "learning_rate": 4.5254561086002866e-05, "loss": 0.0809, "step": 58010 }, { "epoch": 27.81, "learning_rate": 4.524026781101492e-05, "loss": 0.0957, "step": 58020 }, { "epoch": 27.82, "learning_rate": 4.5225974928534186e-05, "loss": 0.0841, "step": 58030 }, { "epoch": 27.82, "learning_rate": 4.521168243973927e-05, "loss": 0.0815, "step": 58040 }, { "epoch": 27.83, "learning_rate": 4.5197390345808805e-05, "loss": 0.0874, "step": 58050 }, { "epoch": 27.83, "learning_rate": 4.518309864792141e-05, "loss": 0.0883, "step": 58060 }, { "epoch": 27.84, "learning_rate": 4.516880734725559e-05, "loss": 0.0873, "step": 58070 }, { "epoch": 27.84, "learning_rate": 4.5154516444989884e-05, "loss": 0.0857, "step": 58080 }, { "epoch": 27.85, "learning_rate": 4.514022594230277e-05, "loss": 0.0806, "step": 58090 }, { "epoch": 27.85, "learning_rate": 4.512593584037274e-05, "loss": 0.0912, "step": 58100 }, { "epoch": 27.85, "learning_rate": 4.511164614037816e-05, "loss": 0.0893, "step": 58110 }, { "epoch": 27.86, "learning_rate": 4.509735684349745e-05, "loss": 0.0841, "step": 58120 }, { "epoch": 27.86, "learning_rate": 4.508306795090897e-05, "loss": 0.0764, "step": 58130 }, { "epoch": 27.87, "learning_rate": 4.5068779463791014e-05, "loss": 0.0859, "step": 58140 }, { "epoch": 27.87, "learning_rate": 4.5054491383321896e-05, "loss": 0.0847, "step": 58150 }, { "epoch": 27.88, "learning_rate": 4.504020371067985e-05, "loss": 0.081, "step": 58160 }, { "epoch": 27.88, "learning_rate": 4.5025916447043114e-05, "loss": 0.0934, "step": 58170 }, { "epoch": 27.89, "learning_rate": 4.501162959358987e-05, "loss": 0.0886, "step": 58180 }, { "epoch": 27.89, "learning_rate": 4.499734315149825e-05, "loss": 0.0907, "step": 58190 }, { "epoch": 27.9, "learning_rate": 4.4983057121946414e-05, "loss": 0.0875, "step": 58200 }, { "epoch": 27.9, "learning_rate": 4.496877150611241e-05, "loss": 0.0832, "step": 58210 }, { "epoch": 27.9, "learning_rate": 4.4954486305174295e-05, "loss": 0.0906, "step": 58220 }, { "epoch": 27.91, "learning_rate": 4.49402015203101e-05, "loss": 0.0862, "step": 58230 }, { "epoch": 27.91, "learning_rate": 4.492591715269782e-05, "loss": 0.081, "step": 58240 }, { "epoch": 27.92, "learning_rate": 4.4911633203515355e-05, "loss": 0.0818, "step": 58250 }, { "epoch": 27.92, "learning_rate": 4.4897349673940655e-05, "loss": 0.0877, "step": 58260 }, { "epoch": 27.93, "learning_rate": 4.48830665651516e-05, "loss": 0.0797, "step": 58270 }, { "epoch": 27.93, "learning_rate": 4.4868783878326024e-05, "loss": 0.0754, "step": 58280 }, { "epoch": 27.94, "learning_rate": 4.485450161464174e-05, "loss": 0.0847, "step": 58290 }, { "epoch": 27.94, "learning_rate": 4.484021977527651e-05, "loss": 0.0785, "step": 58300 }, { "epoch": 27.95, "learning_rate": 4.482593836140809e-05, "loss": 0.0792, "step": 58310 }, { "epoch": 27.95, "learning_rate": 4.4811657374214174e-05, "loss": 0.0843, "step": 58320 }, { "epoch": 28.0, "learning_rate": 4.479737681487244e-05, "loss": 0.1048, "step": 58330 }, { "epoch": 28.01, "learning_rate": 4.478309668456052e-05, "loss": 0.09, "step": 58340 }, { "epoch": 28.01, "learning_rate": 4.476881698445601e-05, "loss": 0.0844, "step": 58350 }, { "epoch": 28.02, "learning_rate": 4.4754537715736465e-05, "loss": 0.0824, "step": 58360 }, { "epoch": 28.02, "learning_rate": 4.4740258879579424e-05, "loss": 0.0822, "step": 58370 }, { "epoch": 28.03, "learning_rate": 4.472598047716238e-05, "loss": 0.0896, "step": 58380 }, { "epoch": 28.03, "learning_rate": 4.471170250966278e-05, "loss": 0.0873, "step": 58390 }, { "epoch": 28.03, "learning_rate": 4.4697424978258045e-05, "loss": 0.0881, "step": 58400 }, { "epoch": 28.04, "learning_rate": 4.4683147884125565e-05, "loss": 0.0947, "step": 58410 }, { "epoch": 28.04, "learning_rate": 4.46688712284427e-05, "loss": 0.0841, "step": 58420 }, { "epoch": 28.05, "learning_rate": 4.465459501238673e-05, "loss": 0.0823, "step": 58430 }, { "epoch": 28.05, "learning_rate": 4.4640319237134956e-05, "loss": 0.0887, "step": 58440 }, { "epoch": 28.06, "learning_rate": 4.46260439038646e-05, "loss": 0.0883, "step": 58450 }, { "epoch": 28.06, "learning_rate": 4.4611769013752895e-05, "loss": 0.0803, "step": 58460 }, { "epoch": 28.07, "learning_rate": 4.459749456797698e-05, "loss": 0.0854, "step": 58470 }, { "epoch": 28.07, "learning_rate": 4.4583220567713985e-05, "loss": 0.0788, "step": 58480 }, { "epoch": 28.08, "learning_rate": 4.456894701414103e-05, "loss": 0.0826, "step": 58490 }, { "epoch": 28.08, "learning_rate": 4.4554673908435126e-05, "loss": 0.0953, "step": 58500 }, { "epoch": 28.09, "learning_rate": 4.4540401251773334e-05, "loss": 0.0852, "step": 58510 }, { "epoch": 28.09, "learning_rate": 4.452612904533261e-05, "loss": 0.084, "step": 58520 }, { "epoch": 28.09, "learning_rate": 4.4511857290289926e-05, "loss": 0.0744, "step": 58530 }, { "epoch": 28.1, "learning_rate": 4.449758598782216e-05, "loss": 0.0829, "step": 58540 }, { "epoch": 28.1, "learning_rate": 4.44833151391062e-05, "loss": 0.0745, "step": 58550 }, { "epoch": 28.11, "learning_rate": 4.446904474531889e-05, "loss": 0.0853, "step": 58560 }, { "epoch": 28.11, "learning_rate": 4.4454774807636986e-05, "loss": 0.0959, "step": 58570 }, { "epoch": 28.12, "learning_rate": 4.444050532723728e-05, "loss": 0.081, "step": 58580 }, { "epoch": 28.12, "learning_rate": 4.4426236305296475e-05, "loss": 0.0866, "step": 58590 }, { "epoch": 28.13, "learning_rate": 4.441196774299129e-05, "loss": 0.0848, "step": 58600 }, { "epoch": 28.13, "learning_rate": 4.439769964149831e-05, "loss": 0.0906, "step": 58610 }, { "epoch": 28.14, "learning_rate": 4.438343200199419e-05, "loss": 0.082, "step": 58620 }, { "epoch": 28.14, "learning_rate": 4.4369164825655466e-05, "loss": 0.0765, "step": 58630 }, { "epoch": 28.14, "learning_rate": 4.4354898113658715e-05, "loss": 0.0773, "step": 58640 }, { "epoch": 28.15, "learning_rate": 4.434063186718036e-05, "loss": 0.0802, "step": 58650 }, { "epoch": 28.15, "learning_rate": 4.432636608739691e-05, "loss": 0.0906, "step": 58660 }, { "epoch": 28.16, "learning_rate": 4.4312100775484785e-05, "loss": 0.0863, "step": 58670 }, { "epoch": 28.16, "learning_rate": 4.429783593262031e-05, "loss": 0.085, "step": 58680 }, { "epoch": 28.17, "learning_rate": 4.428357155997985e-05, "loss": 0.0795, "step": 58690 }, { "epoch": 28.17, "learning_rate": 4.426930765873971e-05, "loss": 0.081, "step": 58700 }, { "epoch": 28.18, "learning_rate": 4.4255044230076174e-05, "loss": 0.0814, "step": 58710 }, { "epoch": 28.18, "learning_rate": 4.424078127516541e-05, "loss": 0.0854, "step": 58720 }, { "epoch": 28.19, "learning_rate": 4.422651879518362e-05, "loss": 0.0797, "step": 58730 }, { "epoch": 28.19, "learning_rate": 4.4212256791306985e-05, "loss": 0.0849, "step": 58740 }, { "epoch": 28.19, "learning_rate": 4.419799526471155e-05, "loss": 0.0779, "step": 58750 }, { "epoch": 28.2, "learning_rate": 4.418373421657341e-05, "loss": 0.0811, "step": 58760 }, { "epoch": 28.2, "learning_rate": 4.416947364806857e-05, "loss": 0.0794, "step": 58770 }, { "epoch": 28.21, "learning_rate": 4.4155213560373074e-05, "loss": 0.0802, "step": 58780 }, { "epoch": 28.21, "learning_rate": 4.41409539546628e-05, "loss": 0.0942, "step": 58790 }, { "epoch": 28.22, "learning_rate": 4.412669483211368e-05, "loss": 0.0856, "step": 58800 }, { "epoch": 28.22, "learning_rate": 4.411243619390158e-05, "loss": 0.0828, "step": 58810 }, { "epoch": 28.23, "learning_rate": 4.409817804120233e-05, "loss": 0.0784, "step": 58820 }, { "epoch": 28.23, "learning_rate": 4.408392037519171e-05, "loss": 0.0825, "step": 58830 }, { "epoch": 28.24, "learning_rate": 4.406966319704546e-05, "loss": 0.0942, "step": 58840 }, { "epoch": 28.24, "learning_rate": 4.4055406507939314e-05, "loss": 0.0922, "step": 58850 }, { "epoch": 28.24, "learning_rate": 4.404115030904889e-05, "loss": 0.0886, "step": 58860 }, { "epoch": 28.25, "learning_rate": 4.4026894601549864e-05, "loss": 0.0864, "step": 58870 }, { "epoch": 28.25, "learning_rate": 4.401263938661778e-05, "loss": 0.074, "step": 58880 }, { "epoch": 28.26, "learning_rate": 4.3998384665428214e-05, "loss": 0.0844, "step": 58890 }, { "epoch": 28.26, "learning_rate": 4.398413043915665e-05, "loss": 0.0807, "step": 58900 }, { "epoch": 28.27, "learning_rate": 4.396987670897855e-05, "loss": 0.0866, "step": 58910 }, { "epoch": 28.27, "learning_rate": 4.395562347606935e-05, "loss": 0.0768, "step": 58920 }, { "epoch": 28.28, "learning_rate": 4.3941370741604414e-05, "loss": 0.0884, "step": 58930 }, { "epoch": 28.28, "learning_rate": 4.392711850675909e-05, "loss": 0.0849, "step": 58940 }, { "epoch": 28.29, "learning_rate": 4.3912866772708664e-05, "loss": 0.0948, "step": 58950 }, { "epoch": 28.29, "learning_rate": 4.389861554062842e-05, "loss": 0.0842, "step": 58960 }, { "epoch": 28.3, "learning_rate": 4.388436481169355e-05, "loss": 0.0821, "step": 58970 }, { "epoch": 28.3, "learning_rate": 4.387011458707923e-05, "loss": 0.0832, "step": 58980 }, { "epoch": 28.3, "learning_rate": 4.3855864867960614e-05, "loss": 0.0818, "step": 58990 }, { "epoch": 28.31, "learning_rate": 4.384161565551276e-05, "loss": 0.0894, "step": 59000 }, { "epoch": 28.31, "learning_rate": 4.382736695091073e-05, "loss": 0.086, "step": 59010 }, { "epoch": 28.32, "learning_rate": 4.381311875532954e-05, "loss": 0.0868, "step": 59020 }, { "epoch": 28.32, "learning_rate": 4.379887106994416e-05, "loss": 0.0957, "step": 59030 }, { "epoch": 28.33, "learning_rate": 4.3784623895929487e-05, "loss": 0.0863, "step": 59040 }, { "epoch": 28.33, "learning_rate": 4.377037723446042e-05, "loss": 0.0828, "step": 59050 }, { "epoch": 28.34, "learning_rate": 4.375613108671179e-05, "loss": 0.0828, "step": 59060 }, { "epoch": 28.34, "learning_rate": 4.374188545385841e-05, "loss": 0.0933, "step": 59070 }, { "epoch": 28.35, "learning_rate": 4.3727640337075014e-05, "loss": 0.0829, "step": 59080 }, { "epoch": 28.35, "learning_rate": 4.371339573753632e-05, "loss": 0.0804, "step": 59090 }, { "epoch": 28.35, "learning_rate": 4.369915165641701e-05, "loss": 0.0877, "step": 59100 }, { "epoch": 28.36, "learning_rate": 4.3684908094891686e-05, "loss": 0.0941, "step": 59110 }, { "epoch": 28.36, "learning_rate": 4.367066505413493e-05, "loss": 0.0851, "step": 59120 }, { "epoch": 28.37, "learning_rate": 4.365642253532131e-05, "loss": 0.0928, "step": 59130 }, { "epoch": 28.37, "learning_rate": 4.364218053962531e-05, "loss": 0.0791, "step": 59140 }, { "epoch": 28.38, "learning_rate": 4.362793906822137e-05, "loss": 0.0747, "step": 59150 }, { "epoch": 28.38, "learning_rate": 4.361369812228392e-05, "loss": 0.0856, "step": 59160 }, { "epoch": 28.39, "learning_rate": 4.359945770298732e-05, "loss": 0.0872, "step": 59170 }, { "epoch": 28.39, "learning_rate": 4.358521781150588e-05, "loss": 0.0903, "step": 59180 }, { "epoch": 28.4, "learning_rate": 4.357097844901391e-05, "loss": 0.0804, "step": 59190 }, { "epoch": 28.4, "learning_rate": 4.355673961668561e-05, "loss": 0.0927, "step": 59200 }, { "epoch": 28.4, "learning_rate": 4.354250131569521e-05, "loss": 0.0868, "step": 59210 }, { "epoch": 28.41, "learning_rate": 4.352826354721683e-05, "loss": 0.0809, "step": 59220 }, { "epoch": 28.41, "learning_rate": 4.3514026312424594e-05, "loss": 0.0859, "step": 59230 }, { "epoch": 28.42, "learning_rate": 4.349978961249255e-05, "loss": 0.0846, "step": 59240 }, { "epoch": 28.42, "learning_rate": 4.3485553448594735e-05, "loss": 0.0875, "step": 59250 }, { "epoch": 28.43, "learning_rate": 4.3471317821905096e-05, "loss": 0.0845, "step": 59260 }, { "epoch": 28.43, "learning_rate": 4.345708273359757e-05, "loss": 0.0813, "step": 59270 }, { "epoch": 28.44, "learning_rate": 4.344284818484607e-05, "loss": 0.0793, "step": 59280 }, { "epoch": 28.44, "learning_rate": 4.342861417682438e-05, "loss": 0.0923, "step": 59290 }, { "epoch": 28.45, "learning_rate": 4.341438071070633e-05, "loss": 0.0866, "step": 59300 }, { "epoch": 28.45, "learning_rate": 4.3400147787665675e-05, "loss": 0.0849, "step": 59310 }, { "epoch": 28.46, "learning_rate": 4.338591540887613e-05, "loss": 0.0838, "step": 59320 }, { "epoch": 28.46, "learning_rate": 4.33716835755113e-05, "loss": 0.0936, "step": 59330 }, { "epoch": 28.46, "learning_rate": 4.335745228874486e-05, "loss": 0.0809, "step": 59340 }, { "epoch": 28.47, "learning_rate": 4.3343221549750376e-05, "loss": 0.0766, "step": 59350 }, { "epoch": 28.47, "learning_rate": 4.332899135970133e-05, "loss": 0.0823, "step": 59360 }, { "epoch": 28.48, "learning_rate": 4.331476171977122e-05, "loss": 0.0891, "step": 59370 }, { "epoch": 28.48, "learning_rate": 4.33005326311335e-05, "loss": 0.0855, "step": 59380 }, { "epoch": 28.49, "learning_rate": 4.3286304094961586e-05, "loss": 0.086, "step": 59390 }, { "epoch": 28.49, "learning_rate": 4.327207611242875e-05, "loss": 0.0786, "step": 59400 }, { "epoch": 28.5, "learning_rate": 4.325784868470832e-05, "loss": 0.082, "step": 59410 }, { "epoch": 28.5, "learning_rate": 4.324362181297358e-05, "loss": 0.0796, "step": 59420 }, { "epoch": 28.51, "learning_rate": 4.322939549839769e-05, "loss": 0.0796, "step": 59430 }, { "epoch": 28.51, "learning_rate": 4.321516974215384e-05, "loss": 0.0858, "step": 59440 }, { "epoch": 28.51, "learning_rate": 4.320094454541512e-05, "loss": 0.0844, "step": 59450 }, { "epoch": 28.52, "learning_rate": 4.3186719909354646e-05, "loss": 0.082, "step": 59460 }, { "epoch": 28.52, "learning_rate": 4.317249583514539e-05, "loss": 0.0808, "step": 59470 }, { "epoch": 28.53, "learning_rate": 4.315827232396035e-05, "loss": 0.0853, "step": 59480 }, { "epoch": 28.53, "learning_rate": 4.314404937697243e-05, "loss": 0.0783, "step": 59490 }, { "epoch": 28.54, "learning_rate": 4.3129826995354575e-05, "loss": 0.0799, "step": 59500 }, { "epoch": 28.54, "learning_rate": 4.311560518027956e-05, "loss": 0.083, "step": 59510 }, { "epoch": 28.55, "learning_rate": 4.3101383932920205e-05, "loss": 0.0766, "step": 59520 }, { "epoch": 28.55, "learning_rate": 4.308716325444924e-05, "loss": 0.0847, "step": 59530 }, { "epoch": 28.56, "learning_rate": 4.3072943146039366e-05, "loss": 0.0851, "step": 59540 }, { "epoch": 28.56, "learning_rate": 4.3058723608863225e-05, "loss": 0.0896, "step": 59550 }, { "epoch": 28.56, "learning_rate": 4.304450464409342e-05, "loss": 0.0873, "step": 59560 }, { "epoch": 28.57, "learning_rate": 4.303028625290253e-05, "loss": 0.0743, "step": 59570 }, { "epoch": 28.57, "learning_rate": 4.3016068436463036e-05, "loss": 0.0784, "step": 59580 }, { "epoch": 28.58, "learning_rate": 4.3001851195947396e-05, "loss": 0.0766, "step": 59590 }, { "epoch": 28.58, "learning_rate": 4.298763453252805e-05, "loss": 0.0815, "step": 59600 }, { "epoch": 28.59, "learning_rate": 4.2973418447377324e-05, "loss": 0.0872, "step": 59610 }, { "epoch": 28.59, "learning_rate": 4.295920294166756e-05, "loss": 0.0948, "step": 59620 }, { "epoch": 28.6, "learning_rate": 4.294498801657102e-05, "loss": 0.0886, "step": 59630 }, { "epoch": 28.6, "learning_rate": 4.2930773673259935e-05, "loss": 0.0872, "step": 59640 }, { "epoch": 28.61, "learning_rate": 4.291655991290647e-05, "loss": 0.0838, "step": 59650 }, { "epoch": 28.61, "learning_rate": 4.290234673668275e-05, "loss": 0.0894, "step": 59660 }, { "epoch": 28.62, "learning_rate": 4.288813414576085e-05, "loss": 0.0852, "step": 59670 }, { "epoch": 28.62, "learning_rate": 4.2873922141312814e-05, "loss": 0.0809, "step": 59680 }, { "epoch": 28.62, "learning_rate": 4.285971072451061e-05, "loss": 0.0874, "step": 59690 }, { "epoch": 28.63, "learning_rate": 4.2845499896526165e-05, "loss": 0.0773, "step": 59700 }, { "epoch": 28.63, "learning_rate": 4.283128965853138e-05, "loss": 0.0824, "step": 59710 }, { "epoch": 28.64, "learning_rate": 4.2817080011698066e-05, "loss": 0.0918, "step": 59720 }, { "epoch": 28.64, "learning_rate": 4.2802870957198026e-05, "loss": 0.0873, "step": 59730 }, { "epoch": 28.65, "learning_rate": 4.2788662496202995e-05, "loss": 0.0877, "step": 59740 }, { "epoch": 28.65, "learning_rate": 4.2774454629884677e-05, "loss": 0.0839, "step": 59750 }, { "epoch": 28.66, "learning_rate": 4.276024735941467e-05, "loss": 0.0887, "step": 59760 }, { "epoch": 28.66, "learning_rate": 4.27460406859646e-05, "loss": 0.0862, "step": 59770 }, { "epoch": 28.67, "learning_rate": 4.2731834610706e-05, "loss": 0.09, "step": 59780 }, { "epoch": 28.67, "learning_rate": 4.2717629134810354e-05, "loss": 0.0871, "step": 59790 }, { "epoch": 28.67, "learning_rate": 4.2703424259449106e-05, "loss": 0.092, "step": 59800 }, { "epoch": 28.68, "learning_rate": 4.268921998579366e-05, "loss": 0.09, "step": 59810 }, { "epoch": 28.68, "learning_rate": 4.267501631501535e-05, "loss": 0.0832, "step": 59820 }, { "epoch": 28.69, "learning_rate": 4.266081324828547e-05, "loss": 0.0929, "step": 59830 }, { "epoch": 28.69, "learning_rate": 4.2646610786775254e-05, "loss": 0.0839, "step": 59840 }, { "epoch": 28.7, "learning_rate": 4.263240893165591e-05, "loss": 0.0861, "step": 59850 }, { "epoch": 28.7, "learning_rate": 4.2618207684098595e-05, "loss": 0.0786, "step": 59860 }, { "epoch": 28.71, "learning_rate": 4.2604007045274365e-05, "loss": 0.0788, "step": 59870 }, { "epoch": 28.71, "learning_rate": 4.2589807016354284e-05, "loss": 0.0826, "step": 59880 }, { "epoch": 28.72, "learning_rate": 4.257560759850936e-05, "loss": 0.0839, "step": 59890 }, { "epoch": 28.72, "learning_rate": 4.256140879291051e-05, "loss": 0.0748, "step": 59900 }, { "epoch": 28.72, "learning_rate": 4.2547210600728626e-05, "loss": 0.0895, "step": 59910 }, { "epoch": 28.73, "learning_rate": 4.253301302313457e-05, "loss": 0.0938, "step": 59920 }, { "epoch": 28.73, "learning_rate": 4.2518816061299125e-05, "loss": 0.0857, "step": 59930 }, { "epoch": 28.74, "learning_rate": 4.250461971639302e-05, "loss": 0.0772, "step": 59940 }, { "epoch": 28.74, "learning_rate": 4.249042398958696e-05, "loss": 0.0857, "step": 59950 }, { "epoch": 28.75, "learning_rate": 4.2476228882051595e-05, "loss": 0.0804, "step": 59960 }, { "epoch": 28.75, "learning_rate": 4.2462034394957464e-05, "loss": 0.0864, "step": 59970 }, { "epoch": 28.76, "learning_rate": 4.244784052947514e-05, "loss": 0.0846, "step": 59980 }, { "epoch": 28.76, "learning_rate": 4.243364728677509e-05, "loss": 0.0865, "step": 59990 }, { "epoch": 28.77, "learning_rate": 4.241945466802779e-05, "loss": 0.0799, "step": 60000 }, { "epoch": 28.77, "learning_rate": 4.240526267440356e-05, "loss": 0.0756, "step": 60010 }, { "epoch": 28.78, "learning_rate": 4.239107130707277e-05, "loss": 0.0743, "step": 60020 }, { "epoch": 28.78, "learning_rate": 4.237688056720569e-05, "loss": 0.0802, "step": 60030 }, { "epoch": 28.78, "learning_rate": 4.2362690455972567e-05, "loss": 0.0777, "step": 60040 }, { "epoch": 28.79, "learning_rate": 4.234850097454352e-05, "loss": 0.0765, "step": 60050 }, { "epoch": 28.79, "learning_rate": 4.233431212408872e-05, "loss": 0.1004, "step": 60060 }, { "epoch": 28.8, "learning_rate": 4.232012390577825e-05, "loss": 0.0923, "step": 60070 }, { "epoch": 28.8, "learning_rate": 4.2305936320782086e-05, "loss": 0.0868, "step": 60080 }, { "epoch": 28.81, "learning_rate": 4.2291749370270195e-05, "loss": 0.0831, "step": 60090 }, { "epoch": 28.81, "learning_rate": 4.2277563055412525e-05, "loss": 0.0801, "step": 60100 }, { "epoch": 28.82, "learning_rate": 4.226337737737895e-05, "loss": 0.1, "step": 60110 }, { "epoch": 28.82, "learning_rate": 4.2249192337339226e-05, "loss": 0.0768, "step": 60120 }, { "epoch": 28.83, "learning_rate": 4.223500793646313e-05, "loss": 0.0888, "step": 60130 }, { "epoch": 28.83, "learning_rate": 4.222082417592041e-05, "loss": 0.075, "step": 60140 }, { "epoch": 28.83, "learning_rate": 4.220664105688066e-05, "loss": 0.0932, "step": 60150 }, { "epoch": 28.84, "learning_rate": 4.2192458580513486e-05, "loss": 0.0871, "step": 60160 }, { "epoch": 28.84, "learning_rate": 4.2178276747988446e-05, "loss": 0.0815, "step": 60170 }, { "epoch": 28.85, "learning_rate": 4.216409556047507e-05, "loss": 0.0938, "step": 60180 }, { "epoch": 28.85, "learning_rate": 4.214991501914274e-05, "loss": 0.0904, "step": 60190 }, { "epoch": 28.86, "learning_rate": 4.213573512516086e-05, "loss": 0.0859, "step": 60200 }, { "epoch": 28.86, "learning_rate": 4.2121555879698777e-05, "loss": 0.0952, "step": 60210 }, { "epoch": 28.87, "learning_rate": 4.2107377283925755e-05, "loss": 0.0781, "step": 60220 }, { "epoch": 28.87, "learning_rate": 4.2093199339011015e-05, "loss": 0.0766, "step": 60230 }, { "epoch": 28.88, "learning_rate": 4.207902204612373e-05, "loss": 0.0862, "step": 60240 }, { "epoch": 28.88, "learning_rate": 4.206484540643306e-05, "loss": 0.0958, "step": 60250 }, { "epoch": 28.88, "learning_rate": 4.2050669421108016e-05, "loss": 0.0821, "step": 60260 }, { "epoch": 28.89, "learning_rate": 4.203649409131762e-05, "loss": 0.0918, "step": 60270 }, { "epoch": 28.89, "learning_rate": 4.202231941823083e-05, "loss": 0.0936, "step": 60280 }, { "epoch": 28.9, "learning_rate": 4.200814540301659e-05, "loss": 0.0807, "step": 60290 }, { "epoch": 28.9, "learning_rate": 4.199397204684368e-05, "loss": 0.0845, "step": 60300 }, { "epoch": 28.91, "learning_rate": 4.197979935088093e-05, "loss": 0.088, "step": 60310 }, { "epoch": 28.91, "learning_rate": 4.196562731629709e-05, "loss": 0.0837, "step": 60320 }, { "epoch": 28.92, "learning_rate": 4.195145594426082e-05, "loss": 0.0832, "step": 60330 }, { "epoch": 28.92, "learning_rate": 4.193728523594075e-05, "loss": 0.0871, "step": 60340 }, { "epoch": 28.93, "learning_rate": 4.1923115192505476e-05, "loss": 0.088, "step": 60350 }, { "epoch": 28.93, "learning_rate": 4.190894581512351e-05, "loss": 0.082, "step": 60360 }, { "epoch": 28.94, "learning_rate": 4.18947771049633e-05, "loss": 0.0832, "step": 60370 }, { "epoch": 28.94, "learning_rate": 4.188060906319328e-05, "loss": 0.0866, "step": 60380 }, { "epoch": 28.94, "learning_rate": 4.18664416909818e-05, "loss": 0.0885, "step": 60390 }, { "epoch": 28.95, "learning_rate": 4.185227498949715e-05, "loss": 0.108, "step": 60400 }, { "epoch": 29.0, "learning_rate": 4.183810895990757e-05, "loss": 0.1029, "step": 60410 }, { "epoch": 29.01, "learning_rate": 4.182394360338127e-05, "loss": 0.086, "step": 60420 }, { "epoch": 29.01, "learning_rate": 4.180977892108638e-05, "loss": 0.0879, "step": 60430 }, { "epoch": 29.02, "learning_rate": 4.1795614914190966e-05, "loss": 0.0869, "step": 60440 }, { "epoch": 29.02, "learning_rate": 4.1781451583863055e-05, "loss": 0.0814, "step": 60450 }, { "epoch": 29.02, "learning_rate": 4.1767288931270614e-05, "loss": 0.0939, "step": 60460 }, { "epoch": 29.03, "learning_rate": 4.1753126957581565e-05, "loss": 0.0809, "step": 60470 }, { "epoch": 29.03, "learning_rate": 4.173896566396375e-05, "loss": 0.086, "step": 60480 }, { "epoch": 29.04, "learning_rate": 4.172480505158496e-05, "loss": 0.076, "step": 60490 }, { "epoch": 29.04, "learning_rate": 4.171064512161296e-05, "loss": 0.0853, "step": 60500 }, { "epoch": 29.05, "learning_rate": 4.1696485875215416e-05, "loss": 0.0798, "step": 60510 }, { "epoch": 29.05, "learning_rate": 4.1682327313559975e-05, "loss": 0.0804, "step": 60520 }, { "epoch": 29.06, "learning_rate": 4.16681694378142e-05, "loss": 0.0834, "step": 60530 }, { "epoch": 29.06, "learning_rate": 4.165401224914561e-05, "loss": 0.0808, "step": 60540 }, { "epoch": 29.07, "learning_rate": 4.1639855748721666e-05, "loss": 0.0777, "step": 60550 }, { "epoch": 29.07, "learning_rate": 4.162569993770977e-05, "loss": 0.082, "step": 60560 }, { "epoch": 29.07, "learning_rate": 4.1611544817277276e-05, "loss": 0.0836, "step": 60570 }, { "epoch": 29.08, "learning_rate": 4.159739038859146e-05, "loss": 0.0917, "step": 60580 }, { "epoch": 29.08, "learning_rate": 4.158323665281956e-05, "loss": 0.0863, "step": 60590 }, { "epoch": 29.09, "learning_rate": 4.156908361112876e-05, "loss": 0.0841, "step": 60600 }, { "epoch": 29.09, "learning_rate": 4.155493126468617e-05, "loss": 0.095, "step": 60610 }, { "epoch": 29.1, "learning_rate": 4.154077961465885e-05, "loss": 0.0834, "step": 60620 }, { "epoch": 29.1, "learning_rate": 4.152662866221381e-05, "loss": 0.0773, "step": 60630 }, { "epoch": 29.11, "learning_rate": 4.151247840851799e-05, "loss": 0.0897, "step": 60640 }, { "epoch": 29.11, "learning_rate": 4.149832885473829e-05, "loss": 0.0757, "step": 60650 }, { "epoch": 29.12, "learning_rate": 4.148418000204152e-05, "loss": 0.092, "step": 60660 }, { "epoch": 29.12, "learning_rate": 4.147003185159447e-05, "loss": 0.0811, "step": 60670 }, { "epoch": 29.12, "learning_rate": 4.1455884404563874e-05, "loss": 0.0883, "step": 60680 }, { "epoch": 29.13, "learning_rate": 4.144173766211632e-05, "loss": 0.0884, "step": 60690 }, { "epoch": 29.13, "learning_rate": 4.142759162541847e-05, "loss": 0.0744, "step": 60700 }, { "epoch": 29.14, "learning_rate": 4.141344629563685e-05, "loss": 0.0811, "step": 60710 }, { "epoch": 29.14, "learning_rate": 4.1399301673937954e-05, "loss": 0.0865, "step": 60720 }, { "epoch": 29.15, "learning_rate": 4.1385157761488156e-05, "loss": 0.0819, "step": 60730 }, { "epoch": 29.15, "learning_rate": 4.137101455945387e-05, "loss": 0.084, "step": 60740 }, { "epoch": 29.16, "learning_rate": 4.135687206900141e-05, "loss": 0.0826, "step": 60750 }, { "epoch": 29.16, "learning_rate": 4.1342730291296984e-05, "loss": 0.0727, "step": 60760 }, { "epoch": 29.17, "learning_rate": 4.132858922750679e-05, "loss": 0.0863, "step": 60770 }, { "epoch": 29.17, "learning_rate": 4.131444887879697e-05, "loss": 0.0813, "step": 60780 }, { "epoch": 29.18, "learning_rate": 4.130030924633363e-05, "loss": 0.0806, "step": 60790 }, { "epoch": 29.18, "learning_rate": 4.128617033128271e-05, "loss": 0.0817, "step": 60800 }, { "epoch": 29.18, "learning_rate": 4.127203213481021e-05, "loss": 0.0868, "step": 60810 }, { "epoch": 29.19, "learning_rate": 4.125789465808203e-05, "loss": 0.0922, "step": 60820 }, { "epoch": 29.19, "learning_rate": 4.124375790226398e-05, "loss": 0.0858, "step": 60830 }, { "epoch": 29.2, "learning_rate": 4.122962186852182e-05, "loss": 0.0813, "step": 60840 }, { "epoch": 29.2, "learning_rate": 4.1215486558021315e-05, "loss": 0.0836, "step": 60850 }, { "epoch": 29.21, "learning_rate": 4.120135197192811e-05, "loss": 0.0762, "step": 60860 }, { "epoch": 29.21, "learning_rate": 4.118721811140777e-05, "loss": 0.0925, "step": 60870 }, { "epoch": 29.22, "learning_rate": 4.1173084977625834e-05, "loss": 0.0911, "step": 60880 }, { "epoch": 29.22, "learning_rate": 4.1158952571747814e-05, "loss": 0.0785, "step": 60890 }, { "epoch": 29.23, "learning_rate": 4.1144820894939126e-05, "loss": 0.087, "step": 60900 }, { "epoch": 29.23, "learning_rate": 4.1130689948365077e-05, "loss": 0.0796, "step": 60910 }, { "epoch": 29.23, "learning_rate": 4.111655973319099e-05, "loss": 0.0748, "step": 60920 }, { "epoch": 29.24, "learning_rate": 4.110243025058214e-05, "loss": 0.0785, "step": 60930 }, { "epoch": 29.24, "learning_rate": 4.108830150170363e-05, "loss": 0.0767, "step": 60940 }, { "epoch": 29.25, "learning_rate": 4.107417348772063e-05, "loss": 0.0879, "step": 60950 }, { "epoch": 29.25, "learning_rate": 4.106004620979816e-05, "loss": 0.0943, "step": 60960 }, { "epoch": 29.26, "learning_rate": 4.104591966910126e-05, "loss": 0.0845, "step": 60970 }, { "epoch": 29.26, "learning_rate": 4.103179386679481e-05, "loss": 0.0795, "step": 60980 }, { "epoch": 29.27, "learning_rate": 4.101766880404371e-05, "loss": 0.0817, "step": 60990 }, { "epoch": 29.27, "learning_rate": 4.100354448201278e-05, "loss": 0.0843, "step": 61000 }, { "epoch": 29.28, "learning_rate": 4.0989420901866735e-05, "loss": 0.084, "step": 61010 }, { "epoch": 29.28, "learning_rate": 4.0975298064770294e-05, "loss": 0.0786, "step": 61020 }, { "epoch": 29.28, "learning_rate": 4.096117597188807e-05, "loss": 0.0845, "step": 61030 }, { "epoch": 29.29, "learning_rate": 4.094705462438465e-05, "loss": 0.0826, "step": 61040 }, { "epoch": 29.29, "learning_rate": 4.093293402342451e-05, "loss": 0.0735, "step": 61050 }, { "epoch": 29.3, "learning_rate": 4.09188141701721e-05, "loss": 0.0791, "step": 61060 }, { "epoch": 29.3, "learning_rate": 4.090469506579182e-05, "loss": 0.071, "step": 61070 }, { "epoch": 29.31, "learning_rate": 4.089057671144798e-05, "loss": 0.0777, "step": 61080 }, { "epoch": 29.31, "learning_rate": 4.087645910830483e-05, "loss": 0.086, "step": 61090 }, { "epoch": 29.32, "learning_rate": 4.086234225752657e-05, "loss": 0.086, "step": 61100 }, { "epoch": 29.32, "learning_rate": 4.084822616027735e-05, "loss": 0.0804, "step": 61110 }, { "epoch": 29.33, "learning_rate": 4.083411081772121e-05, "loss": 0.0863, "step": 61120 }, { "epoch": 29.33, "learning_rate": 4.081999623102218e-05, "loss": 0.0756, "step": 61130 }, { "epoch": 29.34, "learning_rate": 4.08058824013442e-05, "loss": 0.0828, "step": 61140 }, { "epoch": 29.34, "learning_rate": 4.079176932985117e-05, "loss": 0.0786, "step": 61150 }, { "epoch": 29.34, "learning_rate": 4.0777657017706895e-05, "loss": 0.0871, "step": 61160 }, { "epoch": 29.35, "learning_rate": 4.076354546607513e-05, "loss": 0.0852, "step": 61170 }, { "epoch": 29.35, "learning_rate": 4.07494346761196e-05, "loss": 0.0814, "step": 61180 }, { "epoch": 29.36, "learning_rate": 4.073532464900391e-05, "loss": 0.0868, "step": 61190 }, { "epoch": 29.36, "learning_rate": 4.072121538589164e-05, "loss": 0.0807, "step": 61200 }, { "epoch": 29.37, "learning_rate": 4.0707106887946304e-05, "loss": 0.0833, "step": 61210 }, { "epoch": 29.37, "learning_rate": 4.069299915633135e-05, "loss": 0.08, "step": 61220 }, { "epoch": 29.38, "learning_rate": 4.067889219221015e-05, "loss": 0.0806, "step": 61230 }, { "epoch": 29.38, "learning_rate": 4.066478599674602e-05, "loss": 0.0837, "step": 61240 }, { "epoch": 29.39, "learning_rate": 4.065068057110222e-05, "loss": 0.0785, "step": 61250 }, { "epoch": 29.39, "learning_rate": 4.063657591644195e-05, "loss": 0.0812, "step": 61260 }, { "epoch": 29.39, "learning_rate": 4.062247203392833e-05, "loss": 0.083, "step": 61270 }, { "epoch": 29.4, "learning_rate": 4.0608368924724423e-05, "loss": 0.0847, "step": 61280 }, { "epoch": 29.4, "learning_rate": 4.059426658999324e-05, "loss": 0.0871, "step": 61290 }, { "epoch": 29.41, "learning_rate": 4.05801650308977e-05, "loss": 0.0752, "step": 61300 }, { "epoch": 29.41, "learning_rate": 4.056606424860068e-05, "loss": 0.0858, "step": 61310 }, { "epoch": 29.42, "learning_rate": 4.0551964244265005e-05, "loss": 0.0881, "step": 61320 }, { "epoch": 29.42, "learning_rate": 4.053786501905341e-05, "loss": 0.0788, "step": 61330 }, { "epoch": 29.43, "learning_rate": 4.0523766574128574e-05, "loss": 0.081, "step": 61340 }, { "epoch": 29.43, "learning_rate": 4.0509668910653106e-05, "loss": 0.0833, "step": 61350 }, { "epoch": 29.44, "learning_rate": 4.049557202978959e-05, "loss": 0.0969, "step": 61360 }, { "epoch": 29.44, "learning_rate": 4.048147593270045e-05, "loss": 0.0791, "step": 61370 }, { "epoch": 29.44, "learning_rate": 4.0467380620548165e-05, "loss": 0.0939, "step": 61380 }, { "epoch": 29.45, "learning_rate": 4.045328609449507e-05, "loss": 0.0894, "step": 61390 }, { "epoch": 29.45, "learning_rate": 4.043919235570348e-05, "loss": 0.0889, "step": 61400 }, { "epoch": 29.46, "learning_rate": 4.0425099405335587e-05, "loss": 0.0768, "step": 61410 }, { "epoch": 29.46, "learning_rate": 4.041100724455357e-05, "loss": 0.087, "step": 61420 }, { "epoch": 29.47, "learning_rate": 4.039691587451956e-05, "loss": 0.0954, "step": 61430 }, { "epoch": 29.47, "learning_rate": 4.0382825296395515e-05, "loss": 0.0775, "step": 61440 }, { "epoch": 29.48, "learning_rate": 4.0368735511343464e-05, "loss": 0.0852, "step": 61450 }, { "epoch": 29.48, "learning_rate": 4.0354646520525294e-05, "loss": 0.0937, "step": 61460 }, { "epoch": 29.49, "learning_rate": 4.034055832510285e-05, "loss": 0.0842, "step": 61470 }, { "epoch": 29.49, "learning_rate": 4.0326470926237865e-05, "loss": 0.087, "step": 61480 }, { "epoch": 29.49, "learning_rate": 4.031238432509208e-05, "loss": 0.0813, "step": 61490 }, { "epoch": 29.5, "learning_rate": 4.0298298522827124e-05, "loss": 0.0926, "step": 61500 }, { "epoch": 29.5, "learning_rate": 4.028421352060459e-05, "loss": 0.0761, "step": 61510 }, { "epoch": 29.51, "learning_rate": 4.0270129319585936e-05, "loss": 0.0771, "step": 61520 }, { "epoch": 29.51, "learning_rate": 4.025604592093265e-05, "loss": 0.0834, "step": 61530 }, { "epoch": 29.52, "learning_rate": 4.0241963325806104e-05, "loss": 0.0809, "step": 61540 }, { "epoch": 29.52, "learning_rate": 4.022788153536757e-05, "loss": 0.0822, "step": 61550 }, { "epoch": 29.53, "learning_rate": 4.021380055077831e-05, "loss": 0.0827, "step": 61560 }, { "epoch": 29.53, "learning_rate": 4.019972037319952e-05, "loss": 0.0874, "step": 61570 }, { "epoch": 29.54, "learning_rate": 4.018564100379231e-05, "loss": 0.0832, "step": 61580 }, { "epoch": 29.54, "learning_rate": 4.017156244371769e-05, "loss": 0.0851, "step": 61590 }, { "epoch": 29.55, "learning_rate": 4.015748469413664e-05, "loss": 0.0827, "step": 61600 }, { "epoch": 29.55, "learning_rate": 4.014340775621012e-05, "loss": 0.0875, "step": 61610 }, { "epoch": 29.55, "learning_rate": 4.012933163109891e-05, "loss": 0.0903, "step": 61620 }, { "epoch": 29.56, "learning_rate": 4.0115256319963814e-05, "loss": 0.0837, "step": 61630 }, { "epoch": 29.56, "learning_rate": 4.0101181823965526e-05, "loss": 0.0923, "step": 61640 }, { "epoch": 29.57, "learning_rate": 4.008710814426473e-05, "loss": 0.0855, "step": 61650 }, { "epoch": 29.57, "learning_rate": 4.007303528202195e-05, "loss": 0.0861, "step": 61660 }, { "epoch": 29.58, "learning_rate": 4.0058963238397703e-05, "loss": 0.0827, "step": 61670 }, { "epoch": 29.58, "learning_rate": 4.004489201455243e-05, "loss": 0.0861, "step": 61680 }, { "epoch": 29.59, "learning_rate": 4.0030821611646544e-05, "loss": 0.0872, "step": 61690 }, { "epoch": 29.59, "learning_rate": 4.001675203084029e-05, "loss": 0.0827, "step": 61700 }, { "epoch": 29.6, "learning_rate": 4.0002683273293916e-05, "loss": 0.0834, "step": 61710 }, { "epoch": 29.6, "learning_rate": 3.9988615340167625e-05, "loss": 0.0854, "step": 61720 }, { "epoch": 29.6, "learning_rate": 3.997454823262147e-05, "loss": 0.0806, "step": 61730 }, { "epoch": 29.61, "learning_rate": 3.9960481951815516e-05, "loss": 0.0801, "step": 61740 }, { "epoch": 29.61, "learning_rate": 3.994641649890971e-05, "loss": 0.0811, "step": 61750 }, { "epoch": 29.62, "learning_rate": 3.993235187506396e-05, "loss": 0.0873, "step": 61760 }, { "epoch": 29.62, "learning_rate": 3.9918288081438085e-05, "loss": 0.0776, "step": 61770 }, { "epoch": 29.63, "learning_rate": 3.990422511919183e-05, "loss": 0.0793, "step": 61780 }, { "epoch": 29.63, "learning_rate": 3.9890162989484925e-05, "loss": 0.082, "step": 61790 }, { "epoch": 29.64, "learning_rate": 3.9876101693476945e-05, "loss": 0.0766, "step": 61800 }, { "epoch": 29.64, "learning_rate": 3.9862041232327474e-05, "loss": 0.0896, "step": 61810 }, { "epoch": 29.65, "learning_rate": 3.984798160719598e-05, "loss": 0.0816, "step": 61820 }, { "epoch": 29.65, "learning_rate": 3.983392281924189e-05, "loss": 0.0806, "step": 61830 }, { "epoch": 29.65, "learning_rate": 3.981986486962453e-05, "loss": 0.0796, "step": 61840 }, { "epoch": 29.66, "learning_rate": 3.98058077595032e-05, "loss": 0.0757, "step": 61850 }, { "epoch": 29.66, "learning_rate": 3.979175149003709e-05, "loss": 0.0857, "step": 61860 }, { "epoch": 29.67, "learning_rate": 3.977769606238536e-05, "loss": 0.0762, "step": 61870 }, { "epoch": 29.67, "learning_rate": 3.9763641477707045e-05, "loss": 0.072, "step": 61880 }, { "epoch": 29.68, "learning_rate": 3.974958773716117e-05, "loss": 0.0793, "step": 61890 }, { "epoch": 29.68, "learning_rate": 3.973553484190666e-05, "loss": 0.0861, "step": 61900 }, { "epoch": 29.69, "learning_rate": 3.972148279310237e-05, "loss": 0.0806, "step": 61910 }, { "epoch": 29.69, "learning_rate": 3.970743159190709e-05, "loss": 0.086, "step": 61920 }, { "epoch": 29.7, "learning_rate": 3.969338123947954e-05, "loss": 0.0849, "step": 61930 }, { "epoch": 29.7, "learning_rate": 3.9679331736978386e-05, "loss": 0.0777, "step": 61940 }, { "epoch": 29.71, "learning_rate": 3.966528308556218e-05, "loss": 0.0838, "step": 61950 }, { "epoch": 29.71, "learning_rate": 3.9651235286389456e-05, "loss": 0.0765, "step": 61960 }, { "epoch": 29.71, "learning_rate": 3.963718834061865e-05, "loss": 0.0757, "step": 61970 }, { "epoch": 29.72, "learning_rate": 3.9623142249408104e-05, "loss": 0.0813, "step": 61980 }, { "epoch": 29.72, "learning_rate": 3.960909701391615e-05, "loss": 0.0776, "step": 61990 }, { "epoch": 29.73, "learning_rate": 3.9595052635301e-05, "loss": 0.0812, "step": 62000 }, { "epoch": 29.73, "learning_rate": 3.9581009114720825e-05, "loss": 0.0848, "step": 62010 }, { "epoch": 29.74, "learning_rate": 3.956696645333369e-05, "loss": 0.0827, "step": 62020 }, { "epoch": 29.74, "learning_rate": 3.9552924652297624e-05, "loss": 0.0821, "step": 62030 }, { "epoch": 29.75, "learning_rate": 3.9538883712770584e-05, "loss": 0.0841, "step": 62040 }, { "epoch": 29.75, "learning_rate": 3.952484363591042e-05, "loss": 0.0827, "step": 62050 }, { "epoch": 29.76, "learning_rate": 3.951080442287495e-05, "loss": 0.0767, "step": 62060 }, { "epoch": 29.76, "learning_rate": 3.949676607482189e-05, "loss": 0.0859, "step": 62070 }, { "epoch": 29.76, "learning_rate": 3.948272859290893e-05, "loss": 0.0908, "step": 62080 }, { "epoch": 29.77, "learning_rate": 3.9468691978293624e-05, "loss": 0.0798, "step": 62090 }, { "epoch": 29.77, "learning_rate": 3.945465623213351e-05, "loss": 0.0751, "step": 62100 }, { "epoch": 29.78, "learning_rate": 3.944062135558603e-05, "loss": 0.0858, "step": 62110 }, { "epoch": 29.78, "learning_rate": 3.942658734980857e-05, "loss": 0.0916, "step": 62120 }, { "epoch": 29.79, "learning_rate": 3.9412554215958406e-05, "loss": 0.1003, "step": 62130 }, { "epoch": 29.79, "learning_rate": 3.939852195519278e-05, "loss": 0.0769, "step": 62140 }, { "epoch": 29.8, "learning_rate": 3.938449056866887e-05, "loss": 0.0813, "step": 62150 }, { "epoch": 29.8, "learning_rate": 3.937046005754372e-05, "loss": 0.0787, "step": 62160 }, { "epoch": 29.81, "learning_rate": 3.9356430422974376e-05, "loss": 0.0883, "step": 62170 }, { "epoch": 29.81, "learning_rate": 3.934240166611777e-05, "loss": 0.0839, "step": 62180 }, { "epoch": 29.81, "learning_rate": 3.93283737881308e-05, "loss": 0.0803, "step": 62190 }, { "epoch": 29.82, "learning_rate": 3.931434679017019e-05, "loss": 0.0808, "step": 62200 }, { "epoch": 29.82, "learning_rate": 3.930032067339273e-05, "loss": 0.0828, "step": 62210 }, { "epoch": 29.83, "learning_rate": 3.928629543895507e-05, "loss": 0.089, "step": 62220 }, { "epoch": 29.83, "learning_rate": 3.9272271088013744e-05, "loss": 0.0855, "step": 62230 }, { "epoch": 29.84, "learning_rate": 3.925824762172528e-05, "loss": 0.0779, "step": 62240 }, { "epoch": 29.84, "learning_rate": 3.924422504124612e-05, "loss": 0.0885, "step": 62250 }, { "epoch": 29.85, "learning_rate": 3.923020334773265e-05, "loss": 0.0854, "step": 62260 }, { "epoch": 29.85, "learning_rate": 3.92161825423411e-05, "loss": 0.0889, "step": 62270 }, { "epoch": 29.86, "learning_rate": 3.920216262622769e-05, "loss": 0.0759, "step": 62280 }, { "epoch": 29.86, "learning_rate": 3.9188143600548605e-05, "loss": 0.0844, "step": 62290 }, { "epoch": 29.87, "learning_rate": 3.91741254664599e-05, "loss": 0.094, "step": 62300 }, { "epoch": 29.87, "learning_rate": 3.916010822511753e-05, "loss": 0.0811, "step": 62310 }, { "epoch": 29.87, "learning_rate": 3.914609187767744e-05, "loss": 0.0822, "step": 62320 }, { "epoch": 29.88, "learning_rate": 3.913207642529549e-05, "loss": 0.0808, "step": 62330 }, { "epoch": 29.88, "learning_rate": 3.911806186912743e-05, "loss": 0.079, "step": 62340 }, { "epoch": 29.89, "learning_rate": 3.910404821032896e-05, "loss": 0.082, "step": 62350 }, { "epoch": 29.89, "learning_rate": 3.90900354500557e-05, "loss": 0.0866, "step": 62360 }, { "epoch": 29.9, "learning_rate": 3.9076023589463245e-05, "loss": 0.0862, "step": 62370 }, { "epoch": 29.9, "learning_rate": 3.906201262970701e-05, "loss": 0.083, "step": 62380 }, { "epoch": 29.91, "learning_rate": 3.9048002571942425e-05, "loss": 0.0849, "step": 62390 }, { "epoch": 29.91, "learning_rate": 3.9033993417324826e-05, "loss": 0.0874, "step": 62400 }, { "epoch": 29.92, "learning_rate": 3.901998516700944e-05, "loss": 0.0933, "step": 62410 }, { "epoch": 29.92, "learning_rate": 3.9005977822151466e-05, "loss": 0.0888, "step": 62420 }, { "epoch": 29.92, "learning_rate": 3.8991971383906e-05, "loss": 0.0779, "step": 62430 }, { "epoch": 29.93, "learning_rate": 3.8977965853428076e-05, "loss": 0.0823, "step": 62440 }, { "epoch": 29.93, "learning_rate": 3.8963961231872635e-05, "loss": 0.0821, "step": 62450 }, { "epoch": 29.94, "learning_rate": 3.894995752039457e-05, "loss": 0.1164, "step": 62460 }, { "epoch": 29.94, "learning_rate": 3.893595472014868e-05, "loss": 0.0899, "step": 62470 }, { "epoch": 29.95, "learning_rate": 3.89219528322897e-05, "loss": 0.0867, "step": 62480 }, { "epoch": 29.95, "learning_rate": 3.8907951857972276e-05, "loss": 0.0936, "step": 62490 }, { "epoch": 30.0, "learning_rate": 3.889395179835098e-05, "loss": 0.0815, "step": 62500 }, { "epoch": 30.01, "learning_rate": 3.887995265458033e-05, "loss": 0.0772, "step": 62510 }, { "epoch": 30.01, "learning_rate": 3.886595442781474e-05, "loss": 0.0791, "step": 62520 }, { "epoch": 30.02, "learning_rate": 3.885195711920857e-05, "loss": 0.0751, "step": 62530 }, { "epoch": 30.02, "learning_rate": 3.883796072991609e-05, "loss": 0.0841, "step": 62540 }, { "epoch": 30.03, "learning_rate": 3.882396526109151e-05, "loss": 0.0829, "step": 62550 }, { "epoch": 30.03, "learning_rate": 3.880997071388894e-05, "loss": 0.0775, "step": 62560 }, { "epoch": 30.04, "learning_rate": 3.879597708946243e-05, "loss": 0.07, "step": 62570 }, { "epoch": 30.04, "learning_rate": 3.878198438896598e-05, "loss": 0.0758, "step": 62580 }, { "epoch": 30.05, "learning_rate": 3.876799261355344e-05, "loss": 0.0824, "step": 62590 }, { "epoch": 30.05, "learning_rate": 3.8754001764378664e-05, "loss": 0.0849, "step": 62600 }, { "epoch": 30.05, "learning_rate": 3.874001184259538e-05, "loss": 0.0732, "step": 62610 }, { "epoch": 30.06, "learning_rate": 3.872602284935727e-05, "loss": 0.0915, "step": 62620 }, { "epoch": 30.06, "learning_rate": 3.871203478581791e-05, "loss": 0.0885, "step": 62630 }, { "epoch": 30.07, "learning_rate": 3.869804765313081e-05, "loss": 0.0747, "step": 62640 }, { "epoch": 30.07, "learning_rate": 3.868406145244941e-05, "loss": 0.0869, "step": 62650 }, { "epoch": 30.08, "learning_rate": 3.867007618492709e-05, "loss": 0.0881, "step": 62660 }, { "epoch": 30.08, "learning_rate": 3.86560918517171e-05, "loss": 0.0784, "step": 62670 }, { "epoch": 30.09, "learning_rate": 3.864210845397267e-05, "loss": 0.0825, "step": 62680 }, { "epoch": 30.09, "learning_rate": 3.862812599284693e-05, "loss": 0.0728, "step": 62690 }, { "epoch": 30.1, "learning_rate": 3.8614144469492914e-05, "loss": 0.0853, "step": 62700 }, { "epoch": 30.1, "learning_rate": 3.86001638850636e-05, "loss": 0.0864, "step": 62710 }, { "epoch": 30.11, "learning_rate": 3.85861842407119e-05, "loss": 0.0959, "step": 62720 }, { "epoch": 30.11, "learning_rate": 3.857220553759062e-05, "loss": 0.0871, "step": 62730 }, { "epoch": 30.11, "learning_rate": 3.855822777685251e-05, "loss": 0.0872, "step": 62740 }, { "epoch": 30.12, "learning_rate": 3.854425095965021e-05, "loss": 0.0801, "step": 62750 }, { "epoch": 30.12, "learning_rate": 3.853027508713635e-05, "loss": 0.0743, "step": 62760 }, { "epoch": 30.13, "learning_rate": 3.8516300160463395e-05, "loss": 0.0815, "step": 62770 }, { "epoch": 30.13, "learning_rate": 3.85023261807838e-05, "loss": 0.075, "step": 62780 }, { "epoch": 30.14, "learning_rate": 3.848835314924991e-05, "loss": 0.0833, "step": 62790 }, { "epoch": 30.14, "learning_rate": 3.8474381067014e-05, "loss": 0.0861, "step": 62800 }, { "epoch": 30.15, "learning_rate": 3.846040993522827e-05, "loss": 0.0792, "step": 62810 }, { "epoch": 30.15, "learning_rate": 3.8446439755044815e-05, "loss": 0.0835, "step": 62820 }, { "epoch": 30.16, "learning_rate": 3.8432470527615725e-05, "loss": 0.0743, "step": 62830 }, { "epoch": 30.16, "learning_rate": 3.8418502254092886e-05, "loss": 0.0881, "step": 62840 }, { "epoch": 30.16, "learning_rate": 3.840453493562823e-05, "loss": 0.0812, "step": 62850 }, { "epoch": 30.17, "learning_rate": 3.839056857337356e-05, "loss": 0.0881, "step": 62860 }, { "epoch": 30.17, "learning_rate": 3.837660316848061e-05, "loss": 0.0888, "step": 62870 }, { "epoch": 30.18, "learning_rate": 3.836263872210096e-05, "loss": 0.0807, "step": 62880 }, { "epoch": 30.18, "learning_rate": 3.834867523538625e-05, "loss": 0.0819, "step": 62890 }, { "epoch": 30.19, "learning_rate": 3.833471270948792e-05, "loss": 0.0859, "step": 62900 }, { "epoch": 30.19, "learning_rate": 3.832075114555742e-05, "loss": 0.0784, "step": 62910 }, { "epoch": 30.2, "learning_rate": 3.830679054474602e-05, "loss": 0.0777, "step": 62920 }, { "epoch": 30.2, "learning_rate": 3.8292830908205025e-05, "loss": 0.0845, "step": 62930 }, { "epoch": 30.21, "learning_rate": 3.827887223708559e-05, "loss": 0.0787, "step": 62940 }, { "epoch": 30.21, "learning_rate": 3.826491453253878e-05, "loss": 0.0882, "step": 62950 }, { "epoch": 30.21, "learning_rate": 3.825095779571561e-05, "loss": 0.0779, "step": 62960 }, { "epoch": 30.22, "learning_rate": 3.823700202776703e-05, "loss": 0.08, "step": 62970 }, { "epoch": 30.22, "learning_rate": 3.822304722984391e-05, "loss": 0.0855, "step": 62980 }, { "epoch": 30.23, "learning_rate": 3.820909340309696e-05, "loss": 0.0742, "step": 62990 }, { "epoch": 30.23, "learning_rate": 3.819514054867691e-05, "loss": 0.0885, "step": 63000 }, { "epoch": 30.24, "learning_rate": 3.8181188667734375e-05, "loss": 0.0741, "step": 63010 }, { "epoch": 30.24, "learning_rate": 3.816723776141986e-05, "loss": 0.0802, "step": 63020 }, { "epoch": 30.25, "learning_rate": 3.815328783088382e-05, "loss": 0.0881, "step": 63030 }, { "epoch": 30.25, "learning_rate": 3.8139338877276624e-05, "loss": 0.0773, "step": 63040 }, { "epoch": 30.26, "learning_rate": 3.8125390901748604e-05, "loss": 0.0783, "step": 63050 }, { "epoch": 30.26, "learning_rate": 3.81114439054499e-05, "loss": 0.0796, "step": 63060 }, { "epoch": 30.27, "learning_rate": 3.8097497889530675e-05, "loss": 0.0819, "step": 63070 }, { "epoch": 30.27, "learning_rate": 3.808355285514096e-05, "loss": 0.0906, "step": 63080 }, { "epoch": 30.27, "learning_rate": 3.8069608803430755e-05, "loss": 0.0832, "step": 63090 }, { "epoch": 30.28, "learning_rate": 3.80556657355499e-05, "loss": 0.0802, "step": 63100 }, { "epoch": 30.28, "learning_rate": 3.804172365264822e-05, "loss": 0.0854, "step": 63110 }, { "epoch": 30.29, "learning_rate": 3.802778255587544e-05, "loss": 0.0861, "step": 63120 }, { "epoch": 30.29, "learning_rate": 3.8013842446381186e-05, "loss": 0.0829, "step": 63130 }, { "epoch": 30.3, "learning_rate": 3.799990332531503e-05, "loss": 0.0776, "step": 63140 }, { "epoch": 30.3, "learning_rate": 3.7985965193826423e-05, "loss": 0.0804, "step": 63150 }, { "epoch": 30.31, "learning_rate": 3.7972028053064823e-05, "loss": 0.0771, "step": 63160 }, { "epoch": 30.31, "learning_rate": 3.795809190417948e-05, "loss": 0.0802, "step": 63170 }, { "epoch": 30.32, "learning_rate": 3.794415674831965e-05, "loss": 0.0862, "step": 63180 }, { "epoch": 30.32, "learning_rate": 3.793022258663449e-05, "loss": 0.0843, "step": 63190 }, { "epoch": 30.32, "learning_rate": 3.791628942027307e-05, "loss": 0.0944, "step": 63200 }, { "epoch": 30.33, "learning_rate": 3.790235725038435e-05, "loss": 0.0785, "step": 63210 }, { "epoch": 30.33, "learning_rate": 3.7888426078117254e-05, "loss": 0.0794, "step": 63220 }, { "epoch": 30.34, "learning_rate": 3.787449590462062e-05, "loss": 0.096, "step": 63230 }, { "epoch": 30.34, "learning_rate": 3.7860566731043164e-05, "loss": 0.0686, "step": 63240 }, { "epoch": 30.35, "learning_rate": 3.784663855853354e-05, "loss": 0.0879, "step": 63250 }, { "epoch": 30.35, "learning_rate": 3.783271138824034e-05, "loss": 0.0817, "step": 63260 }, { "epoch": 30.36, "learning_rate": 3.781878522131206e-05, "loss": 0.0847, "step": 63270 }, { "epoch": 30.36, "learning_rate": 3.7804860058897086e-05, "loss": 0.0692, "step": 63280 }, { "epoch": 30.37, "learning_rate": 3.779093590214376e-05, "loss": 0.0828, "step": 63290 }, { "epoch": 30.37, "learning_rate": 3.7777012752200325e-05, "loss": 0.076, "step": 63300 }, { "epoch": 30.37, "learning_rate": 3.776309061021493e-05, "loss": 0.0724, "step": 63310 }, { "epoch": 30.38, "learning_rate": 3.774916947733566e-05, "loss": 0.0802, "step": 63320 }, { "epoch": 30.38, "learning_rate": 3.773524935471051e-05, "loss": 0.0814, "step": 63330 }, { "epoch": 30.39, "learning_rate": 3.772133024348741e-05, "loss": 0.0849, "step": 63340 }, { "epoch": 30.39, "learning_rate": 3.770741214481415e-05, "loss": 0.0888, "step": 63350 }, { "epoch": 30.4, "learning_rate": 3.7693495059838494e-05, "loss": 0.0796, "step": 63360 }, { "epoch": 30.4, "learning_rate": 3.7679578989708116e-05, "loss": 0.0824, "step": 63370 }, { "epoch": 30.41, "learning_rate": 3.766566393557056e-05, "loss": 0.0794, "step": 63380 }, { "epoch": 30.41, "learning_rate": 3.765174989857334e-05, "loss": 0.0851, "step": 63390 }, { "epoch": 30.42, "learning_rate": 3.7637836879863855e-05, "loss": 0.0735, "step": 63400 }, { "epoch": 30.42, "learning_rate": 3.7623924880589455e-05, "loss": 0.0911, "step": 63410 }, { "epoch": 30.43, "learning_rate": 3.761001390189735e-05, "loss": 0.0822, "step": 63420 }, { "epoch": 30.43, "learning_rate": 3.759610394493471e-05, "loss": 0.0829, "step": 63430 }, { "epoch": 30.43, "learning_rate": 3.758219501084861e-05, "loss": 0.0722, "step": 63440 }, { "epoch": 30.44, "learning_rate": 3.756828710078603e-05, "loss": 0.0896, "step": 63450 }, { "epoch": 30.44, "learning_rate": 3.755438021589388e-05, "loss": 0.0805, "step": 63460 }, { "epoch": 30.45, "learning_rate": 3.754047435731897e-05, "loss": 0.0785, "step": 63470 }, { "epoch": 30.45, "learning_rate": 3.752656952620806e-05, "loss": 0.0784, "step": 63480 }, { "epoch": 30.46, "learning_rate": 3.751266572370777e-05, "loss": 0.0882, "step": 63490 }, { "epoch": 30.46, "learning_rate": 3.749876295096468e-05, "loss": 0.0717, "step": 63500 }, { "epoch": 30.47, "learning_rate": 3.748486120912526e-05, "loss": 0.0843, "step": 63510 }, { "epoch": 30.47, "learning_rate": 3.747096049933593e-05, "loss": 0.0762, "step": 63520 }, { "epoch": 30.48, "learning_rate": 3.7457060822742964e-05, "loss": 0.0754, "step": 63530 }, { "epoch": 30.48, "learning_rate": 3.7443162180492615e-05, "loss": 0.068, "step": 63540 }, { "epoch": 30.48, "learning_rate": 3.7429264573731024e-05, "loss": 0.0783, "step": 63550 }, { "epoch": 30.49, "learning_rate": 3.741536800360421e-05, "loss": 0.0706, "step": 63560 }, { "epoch": 30.49, "learning_rate": 3.740147247125817e-05, "loss": 0.0729, "step": 63570 }, { "epoch": 30.5, "learning_rate": 3.7387577977838786e-05, "loss": 0.08, "step": 63580 }, { "epoch": 30.5, "learning_rate": 3.7373684524491875e-05, "loss": 0.0769, "step": 63590 }, { "epoch": 30.51, "learning_rate": 3.735979211236309e-05, "loss": 0.0908, "step": 63600 }, { "epoch": 30.51, "learning_rate": 3.7345900742598106e-05, "loss": 0.078, "step": 63610 }, { "epoch": 30.52, "learning_rate": 3.7332010416342465e-05, "loss": 0.0781, "step": 63620 }, { "epoch": 30.52, "learning_rate": 3.7318121134741585e-05, "loss": 0.0758, "step": 63630 }, { "epoch": 30.53, "learning_rate": 3.7304232898940836e-05, "loss": 0.0797, "step": 63640 }, { "epoch": 30.53, "learning_rate": 3.729034571008554e-05, "loss": 0.0902, "step": 63650 }, { "epoch": 30.53, "learning_rate": 3.7276459569320886e-05, "loss": 0.081, "step": 63660 }, { "epoch": 30.54, "learning_rate": 3.726257447779194e-05, "loss": 0.0844, "step": 63670 }, { "epoch": 30.54, "learning_rate": 3.7248690436643743e-05, "loss": 0.0771, "step": 63680 }, { "epoch": 30.55, "learning_rate": 3.7234807447021235e-05, "loss": 0.0796, "step": 63690 }, { "epoch": 30.55, "learning_rate": 3.72209255100693e-05, "loss": 0.089, "step": 63700 }, { "epoch": 30.56, "learning_rate": 3.720704462693264e-05, "loss": 0.0763, "step": 63710 }, { "epoch": 30.56, "learning_rate": 3.719316479875595e-05, "loss": 0.0818, "step": 63720 }, { "epoch": 30.57, "learning_rate": 3.717928602668385e-05, "loss": 0.0931, "step": 63730 }, { "epoch": 30.57, "learning_rate": 3.71654083118608e-05, "loss": 0.0864, "step": 63740 }, { "epoch": 30.58, "learning_rate": 3.7151531655431204e-05, "loss": 0.0922, "step": 63750 }, { "epoch": 30.58, "learning_rate": 3.713765605853944e-05, "loss": 0.0761, "step": 63760 }, { "epoch": 30.59, "learning_rate": 3.712378152232974e-05, "loss": 0.0886, "step": 63770 }, { "epoch": 30.59, "learning_rate": 3.71099080479462e-05, "loss": 0.0846, "step": 63780 }, { "epoch": 30.59, "learning_rate": 3.709603563653292e-05, "loss": 0.0833, "step": 63790 }, { "epoch": 30.6, "learning_rate": 3.708216428923391e-05, "loss": 0.0795, "step": 63800 }, { "epoch": 30.6, "learning_rate": 3.706829400719299e-05, "loss": 0.0802, "step": 63810 }, { "epoch": 30.61, "learning_rate": 3.705442479155401e-05, "loss": 0.0836, "step": 63820 }, { "epoch": 30.61, "learning_rate": 3.704055664346065e-05, "loss": 0.0775, "step": 63830 }, { "epoch": 30.62, "learning_rate": 3.702668956405659e-05, "loss": 0.0849, "step": 63840 }, { "epoch": 30.62, "learning_rate": 3.701282355448531e-05, "loss": 0.0781, "step": 63850 }, { "epoch": 30.63, "learning_rate": 3.699895861589028e-05, "loss": 0.0847, "step": 63860 }, { "epoch": 30.63, "learning_rate": 3.698509474941484e-05, "loss": 0.0757, "step": 63870 }, { "epoch": 30.64, "learning_rate": 3.6971231956202315e-05, "loss": 0.0865, "step": 63880 }, { "epoch": 30.64, "learning_rate": 3.695737023739584e-05, "loss": 0.0889, "step": 63890 }, { "epoch": 30.64, "learning_rate": 3.6943509594138516e-05, "loss": 0.087, "step": 63900 }, { "epoch": 30.65, "learning_rate": 3.6929650027573365e-05, "loss": 0.0839, "step": 63910 }, { "epoch": 30.65, "learning_rate": 3.691579153884329e-05, "loss": 0.0773, "step": 63920 }, { "epoch": 30.66, "learning_rate": 3.690193412909112e-05, "loss": 0.0907, "step": 63930 }, { "epoch": 30.66, "learning_rate": 3.68880777994596e-05, "loss": 0.0861, "step": 63940 }, { "epoch": 30.67, "learning_rate": 3.6874222551091386e-05, "loss": 0.086, "step": 63950 }, { "epoch": 30.67, "learning_rate": 3.686036838512903e-05, "loss": 0.086, "step": 63960 }, { "epoch": 30.68, "learning_rate": 3.6846515302714994e-05, "loss": 0.0888, "step": 63970 }, { "epoch": 30.68, "learning_rate": 3.683266330499169e-05, "loss": 0.0909, "step": 63980 }, { "epoch": 30.69, "learning_rate": 3.6818812393101384e-05, "loss": 0.0759, "step": 63990 }, { "epoch": 30.69, "learning_rate": 3.6804962568186286e-05, "loss": 0.0858, "step": 64000 }, { "epoch": 30.69, "learning_rate": 3.6791113831388516e-05, "loss": 0.0832, "step": 64010 }, { "epoch": 30.7, "learning_rate": 3.67772661838501e-05, "loss": 0.0837, "step": 64020 }, { "epoch": 30.7, "learning_rate": 3.6763419626712955e-05, "loss": 0.0817, "step": 64030 }, { "epoch": 30.71, "learning_rate": 3.674957416111894e-05, "loss": 0.0826, "step": 64040 }, { "epoch": 30.71, "learning_rate": 3.673572978820982e-05, "loss": 0.0856, "step": 64050 }, { "epoch": 30.72, "learning_rate": 3.672188650912723e-05, "loss": 0.0833, "step": 64060 }, { "epoch": 30.72, "learning_rate": 3.670804432501277e-05, "loss": 0.0727, "step": 64070 }, { "epoch": 30.73, "learning_rate": 3.66942032370079e-05, "loss": 0.0875, "step": 64080 }, { "epoch": 30.73, "learning_rate": 3.668036324625405e-05, "loss": 0.0807, "step": 64090 }, { "epoch": 30.74, "learning_rate": 3.666652435389248e-05, "loss": 0.0886, "step": 64100 }, { "epoch": 30.74, "learning_rate": 3.665268656106442e-05, "loss": 0.0875, "step": 64110 }, { "epoch": 30.74, "learning_rate": 3.6638849868910995e-05, "loss": 0.0833, "step": 64120 }, { "epoch": 30.75, "learning_rate": 3.662501427857325e-05, "loss": 0.0766, "step": 64130 }, { "epoch": 30.75, "learning_rate": 3.66111797911921e-05, "loss": 0.0769, "step": 64140 }, { "epoch": 30.76, "learning_rate": 3.6597346407908396e-05, "loss": 0.0717, "step": 64150 }, { "epoch": 30.76, "learning_rate": 3.6583514129862915e-05, "loss": 0.0933, "step": 64160 }, { "epoch": 30.77, "learning_rate": 3.6569682958196305e-05, "loss": 0.083, "step": 64170 }, { "epoch": 30.77, "learning_rate": 3.6555852894049146e-05, "loss": 0.0721, "step": 64180 }, { "epoch": 30.78, "learning_rate": 3.6542023938561925e-05, "loss": 0.0847, "step": 64190 }, { "epoch": 30.78, "learning_rate": 3.6528196092875044e-05, "loss": 0.0857, "step": 64200 }, { "epoch": 30.79, "learning_rate": 3.651436935812879e-05, "loss": 0.0825, "step": 64210 }, { "epoch": 30.79, "learning_rate": 3.650054373546337e-05, "loss": 0.0798, "step": 64220 }, { "epoch": 30.8, "learning_rate": 3.648671922601894e-05, "loss": 0.0801, "step": 64230 }, { "epoch": 30.8, "learning_rate": 3.647289583093545e-05, "loss": 0.0763, "step": 64240 }, { "epoch": 30.8, "learning_rate": 3.6459073551352906e-05, "loss": 0.0773, "step": 64250 }, { "epoch": 30.81, "learning_rate": 3.6445252388411115e-05, "loss": 0.0858, "step": 64260 }, { "epoch": 30.81, "learning_rate": 3.643143234324986e-05, "loss": 0.0768, "step": 64270 }, { "epoch": 30.82, "learning_rate": 3.6417613417008744e-05, "loss": 0.0817, "step": 64280 }, { "epoch": 30.82, "learning_rate": 3.640379561082737e-05, "loss": 0.0774, "step": 64290 }, { "epoch": 30.83, "learning_rate": 3.638997892584522e-05, "loss": 0.0802, "step": 64300 }, { "epoch": 30.83, "learning_rate": 3.637616336320167e-05, "loss": 0.0925, "step": 64310 }, { "epoch": 30.84, "learning_rate": 3.6362348924035974e-05, "loss": 0.0786, "step": 64320 }, { "epoch": 30.84, "learning_rate": 3.6348535609487357e-05, "loss": 0.0817, "step": 64330 }, { "epoch": 30.85, "learning_rate": 3.6334723420694945e-05, "loss": 0.0742, "step": 64340 }, { "epoch": 30.85, "learning_rate": 3.632091235879769e-05, "loss": 0.091, "step": 64350 }, { "epoch": 30.85, "learning_rate": 3.6307102424934545e-05, "loss": 0.0742, "step": 64360 }, { "epoch": 30.86, "learning_rate": 3.629329362024433e-05, "loss": 0.0756, "step": 64370 }, { "epoch": 30.86, "learning_rate": 3.62794859458658e-05, "loss": 0.0839, "step": 64380 }, { "epoch": 30.87, "learning_rate": 3.6265679402937534e-05, "loss": 0.0774, "step": 64390 }, { "epoch": 30.87, "learning_rate": 3.625187399259813e-05, "loss": 0.0783, "step": 64400 }, { "epoch": 30.88, "learning_rate": 3.623806971598604e-05, "loss": 0.083, "step": 64410 }, { "epoch": 30.88, "learning_rate": 3.6224266574239575e-05, "loss": 0.0793, "step": 64420 }, { "epoch": 30.89, "learning_rate": 3.6210464568497015e-05, "loss": 0.0921, "step": 64430 }, { "epoch": 30.89, "learning_rate": 3.619666369989655e-05, "loss": 0.0803, "step": 64440 }, { "epoch": 30.9, "learning_rate": 3.6182863969576274e-05, "loss": 0.0958, "step": 64450 }, { "epoch": 30.9, "learning_rate": 3.6169065378674126e-05, "loss": 0.0737, "step": 64460 }, { "epoch": 30.9, "learning_rate": 3.6155267928328004e-05, "loss": 0.0821, "step": 64470 }, { "epoch": 30.91, "learning_rate": 3.614147161967572e-05, "loss": 0.0787, "step": 64480 }, { "epoch": 30.91, "learning_rate": 3.612767645385498e-05, "loss": 0.0799, "step": 64490 }, { "epoch": 30.92, "learning_rate": 3.611388243200337e-05, "loss": 0.0744, "step": 64500 }, { "epoch": 30.92, "learning_rate": 3.610008955525838e-05, "loss": 0.0792, "step": 64510 }, { "epoch": 30.93, "learning_rate": 3.60862978247575e-05, "loss": 0.0915, "step": 64520 }, { "epoch": 30.93, "learning_rate": 3.6072507241637984e-05, "loss": 0.0824, "step": 64530 }, { "epoch": 30.94, "learning_rate": 3.605871780703709e-05, "loss": 0.0858, "step": 64540 }, { "epoch": 30.94, "learning_rate": 3.604492952209194e-05, "loss": 0.0875, "step": 64550 }, { "epoch": 30.95, "learning_rate": 3.603114238793961e-05, "loss": 0.0902, "step": 64560 }, { "epoch": 30.95, "learning_rate": 3.601735640571698e-05, "loss": 0.0889, "step": 64570 }, { "epoch": 31.0, "learning_rate": 3.600357157656096e-05, "loss": 0.0912, "step": 64580 }, { "epoch": 31.01, "learning_rate": 3.5989787901608266e-05, "loss": 0.0827, "step": 64590 }, { "epoch": 31.01, "learning_rate": 3.597600538199557e-05, "loss": 0.0838, "step": 64600 }, { "epoch": 31.02, "learning_rate": 3.596222401885943e-05, "loss": 0.0823, "step": 64610 }, { "epoch": 31.02, "learning_rate": 3.594844381333632e-05, "loss": 0.0813, "step": 64620 }, { "epoch": 31.03, "learning_rate": 3.593466476656262e-05, "loss": 0.0822, "step": 64630 }, { "epoch": 31.03, "learning_rate": 3.592088687967459e-05, "loss": 0.0746, "step": 64640 }, { "epoch": 31.04, "learning_rate": 3.590711015380842e-05, "loss": 0.0778, "step": 64650 }, { "epoch": 31.04, "learning_rate": 3.589333459010019e-05, "loss": 0.078, "step": 64660 }, { "epoch": 31.04, "learning_rate": 3.587956018968591e-05, "loss": 0.0756, "step": 64670 }, { "epoch": 31.05, "learning_rate": 3.586578695370145e-05, "loss": 0.0774, "step": 64680 }, { "epoch": 31.05, "learning_rate": 3.585201488328262e-05, "loss": 0.0799, "step": 64690 }, { "epoch": 31.06, "learning_rate": 3.5838243979565126e-05, "loss": 0.0792, "step": 64700 }, { "epoch": 31.06, "learning_rate": 3.582447424368456e-05, "loss": 0.0889, "step": 64710 }, { "epoch": 31.07, "learning_rate": 3.581070567677644e-05, "loss": 0.0829, "step": 64720 }, { "epoch": 31.07, "learning_rate": 3.579693827997617e-05, "loss": 0.0772, "step": 64730 }, { "epoch": 31.08, "learning_rate": 3.578317205441908e-05, "loss": 0.0772, "step": 64740 }, { "epoch": 31.08, "learning_rate": 3.5769407001240384e-05, "loss": 0.0772, "step": 64750 }, { "epoch": 31.09, "learning_rate": 3.57556431215752e-05, "loss": 0.0815, "step": 64760 }, { "epoch": 31.09, "learning_rate": 3.574188041655857e-05, "loss": 0.0757, "step": 64770 }, { "epoch": 31.09, "learning_rate": 3.57281188873254e-05, "loss": 0.0725, "step": 64780 }, { "epoch": 31.1, "learning_rate": 3.5714358535010536e-05, "loss": 0.0748, "step": 64790 }, { "epoch": 31.1, "learning_rate": 3.5700599360748716e-05, "loss": 0.0727, "step": 64800 }, { "epoch": 31.11, "learning_rate": 3.568684136567458e-05, "loss": 0.0744, "step": 64810 }, { "epoch": 31.11, "learning_rate": 3.567308455092266e-05, "loss": 0.0779, "step": 64820 }, { "epoch": 31.12, "learning_rate": 3.56593289176274e-05, "loss": 0.088, "step": 64830 }, { "epoch": 31.12, "learning_rate": 3.5645574466923163e-05, "loss": 0.0951, "step": 64840 }, { "epoch": 31.13, "learning_rate": 3.563182119994417e-05, "loss": 0.0816, "step": 64850 }, { "epoch": 31.13, "learning_rate": 3.561806911782459e-05, "loss": 0.0763, "step": 64860 }, { "epoch": 31.14, "learning_rate": 3.5604318221698483e-05, "loss": 0.0926, "step": 64870 }, { "epoch": 31.14, "learning_rate": 3.55905685126998e-05, "loss": 0.0869, "step": 64880 }, { "epoch": 31.14, "learning_rate": 3.5576819991962385e-05, "loss": 0.0772, "step": 64890 }, { "epoch": 31.15, "learning_rate": 3.556307266062e-05, "loss": 0.0768, "step": 64900 }, { "epoch": 31.15, "learning_rate": 3.5549326519806325e-05, "loss": 0.0814, "step": 64910 }, { "epoch": 31.16, "learning_rate": 3.553558157065492e-05, "loss": 0.0825, "step": 64920 }, { "epoch": 31.16, "learning_rate": 3.552183781429923e-05, "loss": 0.0755, "step": 64930 }, { "epoch": 31.17, "learning_rate": 3.550809525187263e-05, "loss": 0.0871, "step": 64940 }, { "epoch": 31.17, "learning_rate": 3.549435388450841e-05, "loss": 0.0821, "step": 64950 }, { "epoch": 31.18, "learning_rate": 3.5480613713339716e-05, "loss": 0.0806, "step": 64960 }, { "epoch": 31.18, "learning_rate": 3.546687473949962e-05, "loss": 0.0855, "step": 64970 }, { "epoch": 31.19, "learning_rate": 3.545313696412111e-05, "loss": 0.0846, "step": 64980 }, { "epoch": 31.19, "learning_rate": 3.543940038833705e-05, "loss": 0.0773, "step": 64990 }, { "epoch": 31.2, "learning_rate": 3.5425665013280216e-05, "loss": 0.0779, "step": 65000 }, { "epoch": 31.2, "learning_rate": 3.541193084008329e-05, "loss": 0.08, "step": 65010 }, { "epoch": 31.2, "learning_rate": 3.539819786987884e-05, "loss": 0.0843, "step": 65020 }, { "epoch": 31.21, "learning_rate": 3.5384466103799334e-05, "loss": 0.083, "step": 65030 }, { "epoch": 31.21, "learning_rate": 3.537073554297716e-05, "loss": 0.0777, "step": 65040 }, { "epoch": 31.22, "learning_rate": 3.535700618854461e-05, "loss": 0.0805, "step": 65050 }, { "epoch": 31.22, "learning_rate": 3.534327804163387e-05, "loss": 0.0742, "step": 65060 }, { "epoch": 31.23, "learning_rate": 3.532955110337697e-05, "loss": 0.0726, "step": 65070 }, { "epoch": 31.23, "learning_rate": 3.531582537490593e-05, "loss": 0.0699, "step": 65080 }, { "epoch": 31.24, "learning_rate": 3.530210085735263e-05, "loss": 0.0863, "step": 65090 }, { "epoch": 31.24, "learning_rate": 3.528837755184885e-05, "loss": 0.0767, "step": 65100 }, { "epoch": 31.25, "learning_rate": 3.527465545952624e-05, "loss": 0.0807, "step": 65110 }, { "epoch": 31.25, "learning_rate": 3.526093458151642e-05, "loss": 0.0778, "step": 65120 }, { "epoch": 31.25, "learning_rate": 3.5247214918950864e-05, "loss": 0.0767, "step": 65130 }, { "epoch": 31.26, "learning_rate": 3.5233496472960934e-05, "loss": 0.0819, "step": 65140 }, { "epoch": 31.26, "learning_rate": 3.5219779244677895e-05, "loss": 0.0773, "step": 65150 }, { "epoch": 31.27, "learning_rate": 3.520606323523297e-05, "loss": 0.0846, "step": 65160 }, { "epoch": 31.27, "learning_rate": 3.519234844575723e-05, "loss": 0.0889, "step": 65170 }, { "epoch": 31.28, "learning_rate": 3.5178634877381625e-05, "loss": 0.0854, "step": 65180 }, { "epoch": 31.28, "learning_rate": 3.516492253123703e-05, "loss": 0.0819, "step": 65190 }, { "epoch": 31.29, "learning_rate": 3.5151211408454276e-05, "loss": 0.0823, "step": 65200 }, { "epoch": 31.29, "learning_rate": 3.5137501510163976e-05, "loss": 0.0836, "step": 65210 }, { "epoch": 31.3, "learning_rate": 3.512379283749674e-05, "loss": 0.0777, "step": 65220 }, { "epoch": 31.3, "learning_rate": 3.511008539158301e-05, "loss": 0.0801, "step": 65230 }, { "epoch": 31.3, "learning_rate": 3.509637917355323e-05, "loss": 0.0801, "step": 65240 }, { "epoch": 31.31, "learning_rate": 3.508267418453759e-05, "loss": 0.0888, "step": 65250 }, { "epoch": 31.31, "learning_rate": 3.5068970425666295e-05, "loss": 0.0811, "step": 65260 }, { "epoch": 31.32, "learning_rate": 3.5055267898069406e-05, "loss": 0.082, "step": 65270 }, { "epoch": 31.32, "learning_rate": 3.504156660287693e-05, "loss": 0.0815, "step": 65280 }, { "epoch": 31.33, "learning_rate": 3.5027866541218665e-05, "loss": 0.0794, "step": 65290 }, { "epoch": 31.33, "learning_rate": 3.501416771422441e-05, "loss": 0.079, "step": 65300 }, { "epoch": 31.34, "learning_rate": 3.500047012302384e-05, "loss": 0.0827, "step": 65310 }, { "epoch": 31.34, "learning_rate": 3.49867737687465e-05, "loss": 0.0941, "step": 65320 }, { "epoch": 31.35, "learning_rate": 3.497307865252183e-05, "loss": 0.0765, "step": 65330 }, { "epoch": 31.35, "learning_rate": 3.495938477547921e-05, "loss": 0.0781, "step": 65340 }, { "epoch": 31.36, "learning_rate": 3.49456921387479e-05, "loss": 0.086, "step": 65350 }, { "epoch": 31.36, "learning_rate": 3.493200074345703e-05, "loss": 0.074, "step": 65360 }, { "epoch": 31.36, "learning_rate": 3.4918310590735646e-05, "loss": 0.08, "step": 65370 }, { "epoch": 31.37, "learning_rate": 3.490462168171273e-05, "loss": 0.0792, "step": 65380 }, { "epoch": 31.37, "learning_rate": 3.489093401751709e-05, "loss": 0.0726, "step": 65390 }, { "epoch": 31.38, "learning_rate": 3.487724759927747e-05, "loss": 0.0778, "step": 65400 }, { "epoch": 31.38, "learning_rate": 3.486356242812253e-05, "loss": 0.0939, "step": 65410 }, { "epoch": 31.39, "learning_rate": 3.4849878505180804e-05, "loss": 0.0776, "step": 65420 }, { "epoch": 31.39, "learning_rate": 3.48361958315807e-05, "loss": 0.0913, "step": 65430 }, { "epoch": 31.4, "learning_rate": 3.482251440845057e-05, "loss": 0.0794, "step": 65440 }, { "epoch": 31.4, "learning_rate": 3.480883423691863e-05, "loss": 0.0853, "step": 65450 }, { "epoch": 31.41, "learning_rate": 3.479515531811301e-05, "loss": 0.0832, "step": 65460 }, { "epoch": 31.41, "learning_rate": 3.478147765316172e-05, "loss": 0.0705, "step": 65470 }, { "epoch": 31.41, "learning_rate": 3.4767801243192685e-05, "loss": 0.0767, "step": 65480 }, { "epoch": 31.42, "learning_rate": 3.475412608933373e-05, "loss": 0.0698, "step": 65490 }, { "epoch": 31.42, "learning_rate": 3.4740452192712544e-05, "loss": 0.0928, "step": 65500 }, { "epoch": 31.43, "learning_rate": 3.472677955445673e-05, "loss": 0.0847, "step": 65510 }, { "epoch": 31.43, "learning_rate": 3.471310817569382e-05, "loss": 0.081, "step": 65520 }, { "epoch": 31.44, "learning_rate": 3.469943805755119e-05, "loss": 0.0825, "step": 65530 }, { "epoch": 31.44, "learning_rate": 3.468576920115613e-05, "loss": 0.0784, "step": 65540 }, { "epoch": 31.45, "learning_rate": 3.467210160763584e-05, "loss": 0.083, "step": 65550 }, { "epoch": 31.45, "learning_rate": 3.465843527811741e-05, "loss": 0.0717, "step": 65560 }, { "epoch": 31.46, "learning_rate": 3.4644770213727814e-05, "loss": 0.0849, "step": 65570 }, { "epoch": 31.46, "learning_rate": 3.463110641559393e-05, "loss": 0.0729, "step": 65580 }, { "epoch": 31.46, "learning_rate": 3.4617443884842536e-05, "loss": 0.0892, "step": 65590 }, { "epoch": 31.47, "learning_rate": 3.4603782622600305e-05, "loss": 0.0769, "step": 65600 }, { "epoch": 31.47, "learning_rate": 3.4590122629993785e-05, "loss": 0.0775, "step": 65610 }, { "epoch": 31.48, "learning_rate": 3.4576463908149446e-05, "loss": 0.0857, "step": 65620 }, { "epoch": 31.48, "learning_rate": 3.4562806458193654e-05, "loss": 0.0713, "step": 65630 }, { "epoch": 31.49, "learning_rate": 3.4549150281252636e-05, "loss": 0.0764, "step": 65640 }, { "epoch": 31.49, "learning_rate": 3.453549537845254e-05, "loss": 0.0795, "step": 65650 }, { "epoch": 31.5, "learning_rate": 3.4521841750919425e-05, "loss": 0.0787, "step": 65660 }, { "epoch": 31.5, "learning_rate": 3.4508189399779215e-05, "loss": 0.0825, "step": 65670 }, { "epoch": 31.51, "learning_rate": 3.449453832615773e-05, "loss": 0.0795, "step": 65680 }, { "epoch": 31.51, "learning_rate": 3.448088853118071e-05, "loss": 0.077, "step": 65690 }, { "epoch": 31.52, "learning_rate": 3.446724001597375e-05, "loss": 0.0797, "step": 65700 }, { "epoch": 31.52, "learning_rate": 3.44535927816624e-05, "loss": 0.0739, "step": 65710 }, { "epoch": 31.52, "learning_rate": 3.443994682937203e-05, "loss": 0.0897, "step": 65720 }, { "epoch": 31.53, "learning_rate": 3.442630216022796e-05, "loss": 0.084, "step": 65730 }, { "epoch": 31.53, "learning_rate": 3.441265877535541e-05, "loss": 0.0903, "step": 65740 }, { "epoch": 31.54, "learning_rate": 3.43990166758794e-05, "loss": 0.0781, "step": 65750 }, { "epoch": 31.54, "learning_rate": 3.4385375862924975e-05, "loss": 0.086, "step": 65760 }, { "epoch": 31.55, "learning_rate": 3.4371736337616985e-05, "loss": 0.0823, "step": 65770 }, { "epoch": 31.55, "learning_rate": 3.435809810108025e-05, "loss": 0.0822, "step": 65780 }, { "epoch": 31.56, "learning_rate": 3.434446115443935e-05, "loss": 0.08, "step": 65790 }, { "epoch": 31.56, "learning_rate": 3.433082549881891e-05, "loss": 0.0783, "step": 65800 }, { "epoch": 31.57, "learning_rate": 3.431719113534338e-05, "loss": 0.0747, "step": 65810 }, { "epoch": 31.57, "learning_rate": 3.430355806513707e-05, "loss": 0.0792, "step": 65820 }, { "epoch": 31.57, "learning_rate": 3.4289926289324225e-05, "loss": 0.0888, "step": 65830 }, { "epoch": 31.58, "learning_rate": 3.4276295809029e-05, "loss": 0.0817, "step": 65840 }, { "epoch": 31.58, "learning_rate": 3.4262666625375436e-05, "loss": 0.0662, "step": 65850 }, { "epoch": 31.59, "learning_rate": 3.424903873948742e-05, "loss": 0.0759, "step": 65860 }, { "epoch": 31.59, "learning_rate": 3.4235412152488747e-05, "loss": 0.0879, "step": 65870 }, { "epoch": 31.6, "learning_rate": 3.422178686550316e-05, "loss": 0.0847, "step": 65880 }, { "epoch": 31.6, "learning_rate": 3.420816287965426e-05, "loss": 0.0768, "step": 65890 }, { "epoch": 31.61, "learning_rate": 3.41945401960655e-05, "loss": 0.0799, "step": 65900 }, { "epoch": 31.61, "learning_rate": 3.418091881586028e-05, "loss": 0.0762, "step": 65910 }, { "epoch": 31.62, "learning_rate": 3.4167298740161894e-05, "loss": 0.0788, "step": 65920 }, { "epoch": 31.62, "learning_rate": 3.4153679970093486e-05, "loss": 0.0766, "step": 65930 }, { "epoch": 31.62, "learning_rate": 3.4140062506778116e-05, "loss": 0.0678, "step": 65940 }, { "epoch": 31.63, "learning_rate": 3.4126446351338734e-05, "loss": 0.0817, "step": 65950 }, { "epoch": 31.63, "learning_rate": 3.4112831504898235e-05, "loss": 0.0812, "step": 65960 }, { "epoch": 31.64, "learning_rate": 3.409921796857929e-05, "loss": 0.0776, "step": 65970 }, { "epoch": 31.64, "learning_rate": 3.4085605743504547e-05, "loss": 0.0849, "step": 65980 }, { "epoch": 31.65, "learning_rate": 3.407199483079655e-05, "loss": 0.0761, "step": 65990 }, { "epoch": 31.65, "learning_rate": 3.4058385231577674e-05, "loss": 0.0941, "step": 66000 }, { "epoch": 31.66, "learning_rate": 3.404477694697025e-05, "loss": 0.0789, "step": 66010 }, { "epoch": 31.66, "learning_rate": 3.403116997809646e-05, "loss": 0.0992, "step": 66020 }, { "epoch": 31.67, "learning_rate": 3.401756432607841e-05, "loss": 0.0923, "step": 66030 }, { "epoch": 31.67, "learning_rate": 3.400395999203805e-05, "loss": 0.0799, "step": 66040 }, { "epoch": 31.68, "learning_rate": 3.3990356977097274e-05, "loss": 0.0899, "step": 66050 }, { "epoch": 31.68, "learning_rate": 3.397675528237783e-05, "loss": 0.0829, "step": 66060 }, { "epoch": 31.68, "learning_rate": 3.396315490900137e-05, "loss": 0.0804, "step": 66070 }, { "epoch": 31.69, "learning_rate": 3.394955585808944e-05, "loss": 0.0807, "step": 66080 }, { "epoch": 31.69, "learning_rate": 3.393595813076347e-05, "loss": 0.0919, "step": 66090 }, { "epoch": 31.7, "learning_rate": 3.3922361728144804e-05, "loss": 0.0959, "step": 66100 }, { "epoch": 31.7, "learning_rate": 3.3908766651354625e-05, "loss": 0.0805, "step": 66110 }, { "epoch": 31.71, "learning_rate": 3.389517290151406e-05, "loss": 0.0754, "step": 66120 }, { "epoch": 31.71, "learning_rate": 3.38815804797441e-05, "loss": 0.0818, "step": 66130 }, { "epoch": 31.72, "learning_rate": 3.3867989387165646e-05, "loss": 0.0777, "step": 66140 }, { "epoch": 31.72, "learning_rate": 3.385439962489946e-05, "loss": 0.0793, "step": 66150 }, { "epoch": 31.73, "learning_rate": 3.3840811194066205e-05, "loss": 0.0765, "step": 66160 }, { "epoch": 31.73, "learning_rate": 3.3827224095786456e-05, "loss": 0.0743, "step": 66170 }, { "epoch": 31.73, "learning_rate": 3.3813638331180655e-05, "loss": 0.0858, "step": 66180 }, { "epoch": 31.74, "learning_rate": 3.3800053901369135e-05, "loss": 0.0743, "step": 66190 }, { "epoch": 31.74, "learning_rate": 3.378647080747213e-05, "loss": 0.0789, "step": 66200 }, { "epoch": 31.75, "learning_rate": 3.377288905060976e-05, "loss": 0.0845, "step": 66210 }, { "epoch": 31.75, "learning_rate": 3.375930863190202e-05, "loss": 0.0781, "step": 66220 }, { "epoch": 31.76, "learning_rate": 3.374572955246882e-05, "loss": 0.0727, "step": 66230 }, { "epoch": 31.76, "learning_rate": 3.3732151813429945e-05, "loss": 0.0845, "step": 66240 }, { "epoch": 31.77, "learning_rate": 3.371857541590507e-05, "loss": 0.0769, "step": 66250 }, { "epoch": 31.77, "learning_rate": 3.370500036101375e-05, "loss": 0.0847, "step": 66260 }, { "epoch": 31.78, "learning_rate": 3.369142664987546e-05, "loss": 0.0757, "step": 66270 }, { "epoch": 31.78, "learning_rate": 3.367785428360955e-05, "loss": 0.0812, "step": 66280 }, { "epoch": 31.78, "learning_rate": 3.3664283263335214e-05, "loss": 0.0721, "step": 66290 }, { "epoch": 31.79, "learning_rate": 3.365071359017161e-05, "loss": 0.0764, "step": 66300 }, { "epoch": 31.79, "learning_rate": 3.3637145265237735e-05, "loss": 0.0814, "step": 66310 }, { "epoch": 31.8, "learning_rate": 3.36235782896525e-05, "loss": 0.0859, "step": 66320 }, { "epoch": 31.8, "learning_rate": 3.361001266453468e-05, "loss": 0.0707, "step": 66330 }, { "epoch": 31.81, "learning_rate": 3.359644839100295e-05, "loss": 0.082, "step": 66340 }, { "epoch": 31.81, "learning_rate": 3.3582885470175906e-05, "loss": 0.0839, "step": 66350 }, { "epoch": 31.82, "learning_rate": 3.356932390317197e-05, "loss": 0.086, "step": 66360 }, { "epoch": 31.82, "learning_rate": 3.3555763691109496e-05, "loss": 0.0823, "step": 66370 }, { "epoch": 31.83, "learning_rate": 3.354220483510672e-05, "loss": 0.0853, "step": 66380 }, { "epoch": 31.83, "learning_rate": 3.3528647336281755e-05, "loss": 0.0806, "step": 66390 }, { "epoch": 31.84, "learning_rate": 3.35150911957526e-05, "loss": 0.0808, "step": 66400 }, { "epoch": 31.84, "learning_rate": 3.350153641463717e-05, "loss": 0.0861, "step": 66410 }, { "epoch": 31.84, "learning_rate": 3.3487982994053255e-05, "loss": 0.0784, "step": 66420 }, { "epoch": 31.85, "learning_rate": 3.3474430935118487e-05, "loss": 0.0861, "step": 66430 }, { "epoch": 31.85, "learning_rate": 3.346088023895044e-05, "loss": 0.0726, "step": 66440 }, { "epoch": 31.86, "learning_rate": 3.3447330906666584e-05, "loss": 0.0748, "step": 66450 }, { "epoch": 31.86, "learning_rate": 3.343378293938425e-05, "loss": 0.0828, "step": 66460 }, { "epoch": 31.87, "learning_rate": 3.342023633822062e-05, "loss": 0.0853, "step": 66470 }, { "epoch": 31.87, "learning_rate": 3.3406691104292844e-05, "loss": 0.0811, "step": 66480 }, { "epoch": 31.88, "learning_rate": 3.339314723871789e-05, "loss": 0.0776, "step": 66490 }, { "epoch": 31.88, "learning_rate": 3.337960474261268e-05, "loss": 0.0767, "step": 66500 }, { "epoch": 31.89, "learning_rate": 3.336606361709393e-05, "loss": 0.0801, "step": 66510 }, { "epoch": 31.89, "learning_rate": 3.335252386327834e-05, "loss": 0.0709, "step": 66520 }, { "epoch": 31.89, "learning_rate": 3.3338985482282453e-05, "loss": 0.0781, "step": 66530 }, { "epoch": 31.9, "learning_rate": 3.332544847522267e-05, "loss": 0.0897, "step": 66540 }, { "epoch": 31.9, "learning_rate": 3.331191284321531e-05, "loss": 0.0798, "step": 66550 }, { "epoch": 31.91, "learning_rate": 3.329837858737661e-05, "loss": 0.0767, "step": 66560 }, { "epoch": 31.91, "learning_rate": 3.3284845708822656e-05, "loss": 0.0829, "step": 66570 }, { "epoch": 31.92, "learning_rate": 3.32713142086694e-05, "loss": 0.0822, "step": 66580 }, { "epoch": 31.92, "learning_rate": 3.32577840880327e-05, "loss": 0.0818, "step": 66590 }, { "epoch": 31.93, "learning_rate": 3.3244255348028355e-05, "loss": 0.0743, "step": 66600 }, { "epoch": 31.93, "learning_rate": 3.3230727989771945e-05, "loss": 0.083, "step": 66610 }, { "epoch": 31.94, "learning_rate": 3.3217202014379024e-05, "loss": 0.0729, "step": 66620 }, { "epoch": 31.94, "learning_rate": 3.320367742296498e-05, "loss": 0.0912, "step": 66630 }, { "epoch": 31.94, "learning_rate": 3.319015421664514e-05, "loss": 0.0753, "step": 66640 }, { "epoch": 31.95, "learning_rate": 3.317663239653466e-05, "loss": 0.0807, "step": 66650 }, { "epoch": 32.0, "learning_rate": 3.3163111963748576e-05, "loss": 0.0791, "step": 66660 }, { "epoch": 32.01, "learning_rate": 3.3149592919401894e-05, "loss": 0.0797, "step": 66670 }, { "epoch": 32.01, "learning_rate": 3.313607526460944e-05, "loss": 0.0869, "step": 66680 }, { "epoch": 32.02, "learning_rate": 3.31225590004859e-05, "loss": 0.0812, "step": 66690 }, { "epoch": 32.02, "learning_rate": 3.31090441281459e-05, "loss": 0.0821, "step": 66700 }, { "epoch": 32.02, "learning_rate": 3.309553064870397e-05, "loss": 0.075, "step": 66710 }, { "epoch": 32.03, "learning_rate": 3.308201856327441e-05, "loss": 0.0911, "step": 66720 }, { "epoch": 32.03, "learning_rate": 3.3068507872971545e-05, "loss": 0.0905, "step": 66730 }, { "epoch": 32.04, "learning_rate": 3.305499857890948e-05, "loss": 0.0838, "step": 66740 }, { "epoch": 32.04, "learning_rate": 3.3041490682202306e-05, "loss": 0.0756, "step": 66750 }, { "epoch": 32.05, "learning_rate": 3.3027984183963887e-05, "loss": 0.0784, "step": 66760 }, { "epoch": 32.05, "learning_rate": 3.301447908530804e-05, "loss": 0.0859, "step": 66770 }, { "epoch": 32.06, "learning_rate": 3.3000975387348454e-05, "loss": 0.0735, "step": 66780 }, { "epoch": 32.06, "learning_rate": 3.29874730911987e-05, "loss": 0.0832, "step": 66790 }, { "epoch": 32.07, "learning_rate": 3.2973972197972216e-05, "loss": 0.0755, "step": 66800 }, { "epoch": 32.07, "learning_rate": 3.2960472708782365e-05, "loss": 0.0862, "step": 66810 }, { "epoch": 32.07, "learning_rate": 3.2946974624742374e-05, "loss": 0.0804, "step": 66820 }, { "epoch": 32.08, "learning_rate": 3.293347794696532e-05, "loss": 0.0846, "step": 66830 }, { "epoch": 32.08, "learning_rate": 3.2919982676564227e-05, "loss": 0.0789, "step": 66840 }, { "epoch": 32.09, "learning_rate": 3.2906488814651956e-05, "loss": 0.0772, "step": 66850 }, { "epoch": 32.09, "learning_rate": 3.289299636234127e-05, "loss": 0.0856, "step": 66860 }, { "epoch": 32.1, "learning_rate": 3.287950532074481e-05, "loss": 0.0786, "step": 66870 }, { "epoch": 32.1, "learning_rate": 3.28660156909751e-05, "loss": 0.0821, "step": 66880 }, { "epoch": 32.11, "learning_rate": 3.285252747414456e-05, "loss": 0.0804, "step": 66890 }, { "epoch": 32.11, "learning_rate": 3.2839040671365474e-05, "loss": 0.0802, "step": 66900 }, { "epoch": 32.12, "learning_rate": 3.282555528375002e-05, "loss": 0.0818, "step": 66910 }, { "epoch": 32.12, "learning_rate": 3.2812071312410265e-05, "loss": 0.0755, "step": 66920 }, { "epoch": 32.13, "learning_rate": 3.279858875845816e-05, "loss": 0.087, "step": 66930 }, { "epoch": 32.13, "learning_rate": 3.278510762300551e-05, "loss": 0.0748, "step": 66940 }, { "epoch": 32.13, "learning_rate": 3.2771627907164043e-05, "loss": 0.0751, "step": 66950 }, { "epoch": 32.14, "learning_rate": 3.2758149612045355e-05, "loss": 0.0854, "step": 66960 }, { "epoch": 32.14, "learning_rate": 3.274467273876091e-05, "loss": 0.0792, "step": 66970 }, { "epoch": 32.15, "learning_rate": 3.2731197288422065e-05, "loss": 0.0813, "step": 66980 }, { "epoch": 32.15, "learning_rate": 3.2717723262140065e-05, "loss": 0.0781, "step": 66990 }, { "epoch": 32.16, "learning_rate": 3.2704250661026045e-05, "loss": 0.0891, "step": 67000 }, { "epoch": 32.16, "learning_rate": 3.2690779486191e-05, "loss": 0.0849, "step": 67010 }, { "epoch": 32.17, "learning_rate": 3.267730973874582e-05, "loss": 0.0739, "step": 67020 }, { "epoch": 32.17, "learning_rate": 3.266384141980129e-05, "loss": 0.0832, "step": 67030 }, { "epoch": 32.18, "learning_rate": 3.265037453046803e-05, "loss": 0.0809, "step": 67040 }, { "epoch": 32.18, "learning_rate": 3.26369090718566e-05, "loss": 0.0862, "step": 67050 }, { "epoch": 32.18, "learning_rate": 3.262344504507742e-05, "loss": 0.0873, "step": 67060 }, { "epoch": 32.19, "learning_rate": 3.260998245124078e-05, "loss": 0.0885, "step": 67070 }, { "epoch": 32.19, "learning_rate": 3.2596521291456864e-05, "loss": 0.0799, "step": 67080 }, { "epoch": 32.2, "learning_rate": 3.258306156683573e-05, "loss": 0.0807, "step": 67090 }, { "epoch": 32.2, "learning_rate": 3.256960327848733e-05, "loss": 0.0839, "step": 67100 }, { "epoch": 32.21, "learning_rate": 3.2556146427521493e-05, "loss": 0.0711, "step": 67110 }, { "epoch": 32.21, "learning_rate": 3.254269101504792e-05, "loss": 0.0889, "step": 67120 }, { "epoch": 32.22, "learning_rate": 3.2529237042176184e-05, "loss": 0.0687, "step": 67130 }, { "epoch": 32.22, "learning_rate": 3.251578451001581e-05, "loss": 0.0777, "step": 67140 }, { "epoch": 32.23, "learning_rate": 3.2502333419676066e-05, "loss": 0.0813, "step": 67150 }, { "epoch": 32.23, "learning_rate": 3.248888377226626e-05, "loss": 0.0791, "step": 67160 }, { "epoch": 32.23, "learning_rate": 3.247543556889545e-05, "loss": 0.0726, "step": 67170 }, { "epoch": 32.24, "learning_rate": 3.246198881067269e-05, "loss": 0.0765, "step": 67180 }, { "epoch": 32.24, "learning_rate": 3.2448543498706786e-05, "loss": 0.0808, "step": 67190 }, { "epoch": 32.25, "learning_rate": 3.2435099634106545e-05, "loss": 0.0772, "step": 67200 }, { "epoch": 32.25, "learning_rate": 3.2421657217980606e-05, "loss": 0.0756, "step": 67210 }, { "epoch": 32.26, "learning_rate": 3.240821625143743e-05, "loss": 0.0812, "step": 67220 }, { "epoch": 32.26, "learning_rate": 3.2394776735585466e-05, "loss": 0.0835, "step": 67230 }, { "epoch": 32.27, "learning_rate": 3.2381338671532984e-05, "loss": 0.0834, "step": 67240 }, { "epoch": 32.27, "learning_rate": 3.2367902060388155e-05, "loss": 0.0891, "step": 67250 }, { "epoch": 32.28, "learning_rate": 3.235446690325897e-05, "loss": 0.0806, "step": 67260 }, { "epoch": 32.28, "learning_rate": 3.234103320125338e-05, "loss": 0.0747, "step": 67270 }, { "epoch": 32.29, "learning_rate": 3.2327600955479184e-05, "loss": 0.0856, "step": 67280 }, { "epoch": 32.29, "learning_rate": 3.231417016704408e-05, "loss": 0.072, "step": 67290 }, { "epoch": 32.29, "learning_rate": 3.2300740837055567e-05, "loss": 0.0859, "step": 67300 }, { "epoch": 32.3, "learning_rate": 3.2287312966621136e-05, "loss": 0.0717, "step": 67310 }, { "epoch": 32.3, "learning_rate": 3.2273886556848104e-05, "loss": 0.0707, "step": 67320 }, { "epoch": 32.31, "learning_rate": 3.226046160884364e-05, "loss": 0.0756, "step": 67330 }, { "epoch": 32.31, "learning_rate": 3.2247038123714815e-05, "loss": 0.0901, "step": 67340 }, { "epoch": 32.32, "learning_rate": 3.223361610256861e-05, "loss": 0.0828, "step": 67350 }, { "epoch": 32.32, "learning_rate": 3.222019554651188e-05, "loss": 0.0792, "step": 67360 }, { "epoch": 32.33, "learning_rate": 3.220677645665129e-05, "loss": 0.0763, "step": 67370 }, { "epoch": 32.33, "learning_rate": 3.219335883409344e-05, "loss": 0.0732, "step": 67380 }, { "epoch": 32.34, "learning_rate": 3.2179942679944865e-05, "loss": 0.079, "step": 67390 }, { "epoch": 32.34, "learning_rate": 3.2166527995311836e-05, "loss": 0.0769, "step": 67400 }, { "epoch": 32.34, "learning_rate": 3.215311478130062e-05, "loss": 0.0754, "step": 67410 }, { "epoch": 32.35, "learning_rate": 3.213970303901732e-05, "loss": 0.0779, "step": 67420 }, { "epoch": 32.35, "learning_rate": 3.212629276956796e-05, "loss": 0.0728, "step": 67430 }, { "epoch": 32.36, "learning_rate": 3.211288397405835e-05, "loss": 0.079, "step": 67440 }, { "epoch": 32.36, "learning_rate": 3.209947665359426e-05, "loss": 0.0691, "step": 67450 }, { "epoch": 32.37, "learning_rate": 3.2086070809281314e-05, "loss": 0.0782, "step": 67460 }, { "epoch": 32.37, "learning_rate": 3.207266644222501e-05, "loss": 0.0819, "step": 67470 }, { "epoch": 32.38, "learning_rate": 3.205926355353073e-05, "loss": 0.0743, "step": 67480 }, { "epoch": 32.38, "learning_rate": 3.204586214430372e-05, "loss": 0.0865, "step": 67490 }, { "epoch": 32.39, "learning_rate": 3.203246221564914e-05, "loss": 0.0762, "step": 67500 }, { "epoch": 32.39, "learning_rate": 3.201906376867198e-05, "loss": 0.0976, "step": 67510 }, { "epoch": 32.39, "learning_rate": 3.200566680447714e-05, "loss": 0.0799, "step": 67520 }, { "epoch": 32.4, "learning_rate": 3.19922713241694e-05, "loss": 0.0759, "step": 67530 }, { "epoch": 32.4, "learning_rate": 3.1978877328853396e-05, "loss": 0.0798, "step": 67540 }, { "epoch": 32.41, "learning_rate": 3.196548481963364e-05, "loss": 0.0792, "step": 67550 }, { "epoch": 32.41, "learning_rate": 3.1952093797614554e-05, "loss": 0.0766, "step": 67560 }, { "epoch": 32.42, "learning_rate": 3.193870426390041e-05, "loss": 0.0865, "step": 67570 }, { "epoch": 32.42, "learning_rate": 3.192531621959536e-05, "loss": 0.0728, "step": 67580 }, { "epoch": 32.43, "learning_rate": 3.191192966580344e-05, "loss": 0.0746, "step": 67590 }, { "epoch": 32.43, "learning_rate": 3.1898544603628563e-05, "loss": 0.078, "step": 67600 }, { "epoch": 32.44, "learning_rate": 3.1885161034174524e-05, "loss": 0.0835, "step": 67610 }, { "epoch": 32.44, "learning_rate": 3.1871778958544965e-05, "loss": 0.0773, "step": 67620 }, { "epoch": 32.45, "learning_rate": 3.185839837784345e-05, "loss": 0.085, "step": 67630 }, { "epoch": 32.45, "learning_rate": 3.1845019293173386e-05, "loss": 0.069, "step": 67640 }, { "epoch": 32.45, "learning_rate": 3.183164170563806e-05, "loss": 0.0792, "step": 67650 }, { "epoch": 32.46, "learning_rate": 3.181826561634066e-05, "loss": 0.0804, "step": 67660 }, { "epoch": 32.46, "learning_rate": 3.180489102638423e-05, "loss": 0.0815, "step": 67670 }, { "epoch": 32.47, "learning_rate": 3.179151793687169e-05, "loss": 0.0761, "step": 67680 }, { "epoch": 32.47, "learning_rate": 3.177814634890583e-05, "loss": 0.0771, "step": 67690 }, { "epoch": 32.48, "learning_rate": 3.176477626358935e-05, "loss": 0.0767, "step": 67700 }, { "epoch": 32.48, "learning_rate": 3.175140768202477e-05, "loss": 0.0842, "step": 67710 }, { "epoch": 32.49, "learning_rate": 3.173804060531456e-05, "loss": 0.0742, "step": 67720 }, { "epoch": 32.49, "learning_rate": 3.1724675034560994e-05, "loss": 0.0866, "step": 67730 }, { "epoch": 32.5, "learning_rate": 3.1711310970866254e-05, "loss": 0.0834, "step": 67740 }, { "epoch": 32.5, "learning_rate": 3.169794841533241e-05, "loss": 0.0755, "step": 67750 }, { "epoch": 32.5, "learning_rate": 3.1684587369061385e-05, "loss": 0.0788, "step": 67760 }, { "epoch": 32.51, "learning_rate": 3.167122783315498e-05, "loss": 0.0773, "step": 67770 }, { "epoch": 32.51, "learning_rate": 3.1657869808714894e-05, "loss": 0.0729, "step": 67780 }, { "epoch": 32.52, "learning_rate": 3.164451329684267e-05, "loss": 0.0742, "step": 67790 }, { "epoch": 32.52, "learning_rate": 3.163115829863975e-05, "loss": 0.0828, "step": 67800 }, { "epoch": 32.53, "learning_rate": 3.161780481520743e-05, "loss": 0.1054, "step": 67810 }, { "epoch": 32.53, "learning_rate": 3.160445284764692e-05, "loss": 0.0766, "step": 67820 }, { "epoch": 32.54, "learning_rate": 3.1591102397059246e-05, "loss": 0.0763, "step": 67830 }, { "epoch": 32.54, "learning_rate": 3.157775346454536e-05, "loss": 0.0898, "step": 67840 }, { "epoch": 32.55, "learning_rate": 3.1564406051206064e-05, "loss": 0.0721, "step": 67850 }, { "epoch": 32.55, "learning_rate": 3.1551060158142055e-05, "loss": 0.0761, "step": 67860 }, { "epoch": 32.55, "learning_rate": 3.153771578645386e-05, "loss": 0.0841, "step": 67870 }, { "epoch": 32.56, "learning_rate": 3.152437293724194e-05, "loss": 0.0782, "step": 67880 }, { "epoch": 32.56, "learning_rate": 3.151103161160659e-05, "loss": 0.0791, "step": 67890 }, { "epoch": 32.57, "learning_rate": 3.1497691810647993e-05, "loss": 0.0761, "step": 67900 }, { "epoch": 32.57, "learning_rate": 3.148435353546621e-05, "loss": 0.0771, "step": 67910 }, { "epoch": 32.58, "learning_rate": 3.147101678716114e-05, "loss": 0.0859, "step": 67920 }, { "epoch": 32.58, "learning_rate": 3.1457681566832656e-05, "loss": 0.0731, "step": 67930 }, { "epoch": 32.59, "learning_rate": 3.144434787558034e-05, "loss": 0.079, "step": 67940 }, { "epoch": 32.59, "learning_rate": 3.1431015714503806e-05, "loss": 0.0764, "step": 67950 }, { "epoch": 32.6, "learning_rate": 3.141768508470247e-05, "loss": 0.0811, "step": 67960 }, { "epoch": 32.6, "learning_rate": 3.140435598727565e-05, "loss": 0.077, "step": 67970 }, { "epoch": 32.61, "learning_rate": 3.139102842332246e-05, "loss": 0.0798, "step": 67980 }, { "epoch": 32.61, "learning_rate": 3.137770239394199e-05, "loss": 0.0828, "step": 67990 }, { "epoch": 32.61, "learning_rate": 3.136437790023317e-05, "loss": 0.0743, "step": 68000 }, { "epoch": 32.62, "learning_rate": 3.135105494329476e-05, "loss": 0.0731, "step": 68010 }, { "epoch": 32.62, "learning_rate": 3.133773352422542e-05, "loss": 0.0741, "step": 68020 }, { "epoch": 32.63, "learning_rate": 3.132441364412372e-05, "loss": 0.0832, "step": 68030 }, { "epoch": 32.63, "learning_rate": 3.1311095304088086e-05, "loss": 0.0896, "step": 68040 }, { "epoch": 32.64, "learning_rate": 3.1297778505216756e-05, "loss": 0.0741, "step": 68050 }, { "epoch": 32.64, "learning_rate": 3.12844632486079e-05, "loss": 0.0804, "step": 68060 }, { "epoch": 32.65, "learning_rate": 3.127114953535959e-05, "loss": 0.0866, "step": 68070 }, { "epoch": 32.65, "learning_rate": 3.1257837366569675e-05, "loss": 0.0903, "step": 68080 }, { "epoch": 32.66, "learning_rate": 3.124452674333597e-05, "loss": 0.0824, "step": 68090 }, { "epoch": 32.66, "learning_rate": 3.1231217666756086e-05, "loss": 0.0814, "step": 68100 }, { "epoch": 32.66, "learning_rate": 3.1217910137927605e-05, "loss": 0.0831, "step": 68110 }, { "epoch": 32.67, "learning_rate": 3.120460415794786e-05, "loss": 0.0804, "step": 68120 }, { "epoch": 32.67, "learning_rate": 3.1191299727914146e-05, "loss": 0.0723, "step": 68130 }, { "epoch": 32.68, "learning_rate": 3.117799684892359e-05, "loss": 0.0876, "step": 68140 }, { "epoch": 32.68, "learning_rate": 3.1164695522073226e-05, "loss": 0.0825, "step": 68150 }, { "epoch": 32.69, "learning_rate": 3.115139574845991e-05, "loss": 0.0835, "step": 68160 }, { "epoch": 32.69, "learning_rate": 3.11380975291804e-05, "loss": 0.0863, "step": 68170 }, { "epoch": 32.7, "learning_rate": 3.112480086533134e-05, "loss": 0.0807, "step": 68180 }, { "epoch": 32.7, "learning_rate": 3.111150575800921e-05, "loss": 0.0747, "step": 68190 }, { "epoch": 32.71, "learning_rate": 3.1098212208310385e-05, "loss": 0.0792, "step": 68200 }, { "epoch": 32.71, "learning_rate": 3.10849202173311e-05, "loss": 0.0822, "step": 68210 }, { "epoch": 32.71, "learning_rate": 3.1071629786167486e-05, "loss": 0.0797, "step": 68220 }, { "epoch": 32.72, "learning_rate": 3.1058340915915506e-05, "loss": 0.0795, "step": 68230 }, { "epoch": 32.72, "learning_rate": 3.104505360767103e-05, "loss": 0.0802, "step": 68240 }, { "epoch": 32.73, "learning_rate": 3.103176786252978e-05, "loss": 0.0745, "step": 68250 }, { "epoch": 32.73, "learning_rate": 3.101848368158734e-05, "loss": 0.0739, "step": 68260 }, { "epoch": 32.74, "learning_rate": 3.10052010659392e-05, "loss": 0.0755, "step": 68270 }, { "epoch": 32.74, "learning_rate": 3.0991920016680686e-05, "loss": 0.0739, "step": 68280 }, { "epoch": 32.75, "learning_rate": 3.0978640534907016e-05, "loss": 0.0697, "step": 68290 }, { "epoch": 32.75, "learning_rate": 3.096536262171326e-05, "loss": 0.0867, "step": 68300 }, { "epoch": 32.76, "learning_rate": 3.095208627819438e-05, "loss": 0.0749, "step": 68310 }, { "epoch": 32.76, "learning_rate": 3.0938811505445186e-05, "loss": 0.0812, "step": 68320 }, { "epoch": 32.77, "learning_rate": 3.09255383045604e-05, "loss": 0.0835, "step": 68330 }, { "epoch": 32.77, "learning_rate": 3.091226667663455e-05, "loss": 0.0745, "step": 68340 }, { "epoch": 32.77, "learning_rate": 3.0898996622762075e-05, "loss": 0.0796, "step": 68350 }, { "epoch": 32.78, "learning_rate": 3.088572814403731e-05, "loss": 0.0823, "step": 68360 }, { "epoch": 32.78, "learning_rate": 3.087246124155438e-05, "loss": 0.0745, "step": 68370 }, { "epoch": 32.79, "learning_rate": 3.085919591640737e-05, "loss": 0.0804, "step": 68380 }, { "epoch": 32.79, "learning_rate": 3.084593216969017e-05, "loss": 0.0719, "step": 68390 }, { "epoch": 32.8, "learning_rate": 3.0832670002496584e-05, "loss": 0.0795, "step": 68400 }, { "epoch": 32.8, "learning_rate": 3.0819409415920245e-05, "loss": 0.0832, "step": 68410 }, { "epoch": 32.81, "learning_rate": 3.080615041105467e-05, "loss": 0.0862, "step": 68420 }, { "epoch": 32.81, "learning_rate": 3.079289298899328e-05, "loss": 0.0809, "step": 68430 }, { "epoch": 32.82, "learning_rate": 3.077963715082932e-05, "loss": 0.0767, "step": 68440 }, { "epoch": 32.82, "learning_rate": 3.076638289765591e-05, "loss": 0.0759, "step": 68450 }, { "epoch": 32.82, "learning_rate": 3.075313023056607e-05, "loss": 0.0712, "step": 68460 }, { "epoch": 32.83, "learning_rate": 3.073987915065267e-05, "loss": 0.0818, "step": 68470 }, { "epoch": 32.83, "learning_rate": 3.0726629659008436e-05, "loss": 0.0738, "step": 68480 }, { "epoch": 32.84, "learning_rate": 3.071338175672598e-05, "loss": 0.0886, "step": 68490 }, { "epoch": 32.84, "learning_rate": 3.070013544489778e-05, "loss": 0.0798, "step": 68500 }, { "epoch": 32.85, "learning_rate": 3.068689072461619e-05, "loss": 0.0858, "step": 68510 }, { "epoch": 32.85, "learning_rate": 3.06736475969734e-05, "loss": 0.0719, "step": 68520 }, { "epoch": 32.86, "learning_rate": 3.066040606306151e-05, "loss": 0.0868, "step": 68530 }, { "epoch": 32.86, "learning_rate": 3.0647166123972495e-05, "loss": 0.0772, "step": 68540 }, { "epoch": 32.87, "learning_rate": 3.063392778079812e-05, "loss": 0.0868, "step": 68550 }, { "epoch": 32.87, "learning_rate": 3.062069103463012e-05, "loss": 0.0729, "step": 68560 }, { "epoch": 32.87, "learning_rate": 3.060745588656003e-05, "loss": 0.0831, "step": 68570 }, { "epoch": 32.88, "learning_rate": 3.059422233767929e-05, "loss": 0.0857, "step": 68580 }, { "epoch": 32.88, "learning_rate": 3.058099038907918e-05, "loss": 0.0737, "step": 68590 }, { "epoch": 32.89, "learning_rate": 3.0567760041850854e-05, "loss": 0.0745, "step": 68600 }, { "epoch": 32.89, "learning_rate": 3.055453129708539e-05, "loss": 0.0823, "step": 68610 }, { "epoch": 32.9, "learning_rate": 3.05413041558736e-05, "loss": 0.0763, "step": 68620 }, { "epoch": 32.9, "learning_rate": 3.052807861930631e-05, "loss": 0.0824, "step": 68630 }, { "epoch": 32.91, "learning_rate": 3.051485468847415e-05, "loss": 0.0796, "step": 68640 }, { "epoch": 32.91, "learning_rate": 3.050163236446762e-05, "loss": 0.0811, "step": 68650 }, { "epoch": 32.92, "learning_rate": 3.0488411648377047e-05, "loss": 0.0794, "step": 68660 }, { "epoch": 32.92, "learning_rate": 3.0475192541292702e-05, "loss": 0.0833, "step": 68670 }, { "epoch": 32.93, "learning_rate": 3.046197504430468e-05, "loss": 0.0901, "step": 68680 }, { "epoch": 32.93, "learning_rate": 3.0448759158502975e-05, "loss": 0.0768, "step": 68690 }, { "epoch": 32.93, "learning_rate": 3.0435544884977368e-05, "loss": 0.0748, "step": 68700 }, { "epoch": 32.94, "learning_rate": 3.0422332224817607e-05, "loss": 0.0827, "step": 68710 }, { "epoch": 32.94, "learning_rate": 3.0409121179113266e-05, "loss": 0.0795, "step": 68720 }, { "epoch": 32.95, "learning_rate": 3.0395911748953743e-05, "loss": 0.0852, "step": 68730 }, { "epoch": 33.0, "learning_rate": 3.0382703935428358e-05, "loss": 0.0813, "step": 68740 }, { "epoch": 33.01, "learning_rate": 3.0369497739626296e-05, "loss": 0.078, "step": 68750 }, { "epoch": 33.01, "learning_rate": 3.0356293162636608e-05, "loss": 0.0749, "step": 68760 }, { "epoch": 33.01, "learning_rate": 3.0343090205548157e-05, "loss": 0.0804, "step": 68770 }, { "epoch": 33.02, "learning_rate": 3.0329888869449728e-05, "loss": 0.0804, "step": 68780 }, { "epoch": 33.02, "learning_rate": 3.031668915542999e-05, "loss": 0.0773, "step": 68790 }, { "epoch": 33.03, "learning_rate": 3.03034910645774e-05, "loss": 0.0814, "step": 68800 }, { "epoch": 33.03, "learning_rate": 3.0290294597980352e-05, "loss": 0.0762, "step": 68810 }, { "epoch": 33.04, "learning_rate": 3.0277099756727057e-05, "loss": 0.0829, "step": 68820 }, { "epoch": 33.04, "learning_rate": 3.026390654190567e-05, "loss": 0.079, "step": 68830 }, { "epoch": 33.05, "learning_rate": 3.02507149546041e-05, "loss": 0.0808, "step": 68840 }, { "epoch": 33.05, "learning_rate": 3.0237524995910205e-05, "loss": 0.0779, "step": 68850 }, { "epoch": 33.06, "learning_rate": 3.022433666691169e-05, "loss": 0.0749, "step": 68860 }, { "epoch": 33.06, "learning_rate": 3.02111499686961e-05, "loss": 0.0841, "step": 68870 }, { "epoch": 33.06, "learning_rate": 3.0197964902350873e-05, "loss": 0.0792, "step": 68880 }, { "epoch": 33.07, "learning_rate": 3.018478146896331e-05, "loss": 0.0812, "step": 68890 }, { "epoch": 33.07, "learning_rate": 3.0171599669620577e-05, "loss": 0.0795, "step": 68900 }, { "epoch": 33.08, "learning_rate": 3.0158419505409687e-05, "loss": 0.0809, "step": 68910 }, { "epoch": 33.08, "learning_rate": 3.0145240977417522e-05, "loss": 0.083, "step": 68920 }, { "epoch": 33.09, "learning_rate": 3.0132064086730855e-05, "loss": 0.0754, "step": 68930 }, { "epoch": 33.09, "learning_rate": 3.0118888834436308e-05, "loss": 0.0802, "step": 68940 }, { "epoch": 33.1, "learning_rate": 3.0105715221620356e-05, "loss": 0.0874, "step": 68950 }, { "epoch": 33.1, "learning_rate": 3.0092543249369354e-05, "loss": 0.0838, "step": 68960 }, { "epoch": 33.11, "learning_rate": 3.007937291876952e-05, "loss": 0.0759, "step": 68970 }, { "epoch": 33.11, "learning_rate": 3.006620423090693e-05, "loss": 0.0794, "step": 68980 }, { "epoch": 33.11, "learning_rate": 3.0053037186867526e-05, "loss": 0.0763, "step": 68990 }, { "epoch": 33.12, "learning_rate": 3.0039871787737118e-05, "loss": 0.0726, "step": 69000 }, { "epoch": 33.12, "learning_rate": 3.0026708034601398e-05, "loss": 0.0848, "step": 69010 }, { "epoch": 33.13, "learning_rate": 3.001354592854587e-05, "loss": 0.0738, "step": 69020 }, { "epoch": 33.13, "learning_rate": 3.0000385470655955e-05, "loss": 0.0735, "step": 69030 }, { "epoch": 33.14, "learning_rate": 2.998722666201693e-05, "loss": 0.0708, "step": 69040 }, { "epoch": 33.14, "learning_rate": 2.99740695037139e-05, "loss": 0.0758, "step": 69050 }, { "epoch": 33.15, "learning_rate": 2.9960913996831875e-05, "loss": 0.0775, "step": 69060 }, { "epoch": 33.15, "learning_rate": 2.9947760142455694e-05, "loss": 0.0819, "step": 69070 }, { "epoch": 33.16, "learning_rate": 2.993460794167011e-05, "loss": 0.0804, "step": 69080 }, { "epoch": 33.16, "learning_rate": 2.992145739555968e-05, "loss": 0.0853, "step": 69090 }, { "epoch": 33.16, "learning_rate": 2.9908308505208866e-05, "loss": 0.0761, "step": 69100 }, { "epoch": 33.17, "learning_rate": 2.9895161271701967e-05, "loss": 0.075, "step": 69110 }, { "epoch": 33.17, "learning_rate": 2.9882015696123183e-05, "loss": 0.0749, "step": 69120 }, { "epoch": 33.18, "learning_rate": 2.986887177955653e-05, "loss": 0.0826, "step": 69130 }, { "epoch": 33.18, "learning_rate": 2.9855729523085907e-05, "loss": 0.0784, "step": 69140 }, { "epoch": 33.19, "learning_rate": 2.98425889277951e-05, "loss": 0.0755, "step": 69150 }, { "epoch": 33.19, "learning_rate": 2.982944999476772e-05, "loss": 0.0805, "step": 69160 }, { "epoch": 33.2, "learning_rate": 2.9816312725087258e-05, "loss": 0.0707, "step": 69170 }, { "epoch": 33.2, "learning_rate": 2.9803177119837073e-05, "loss": 0.0775, "step": 69180 }, { "epoch": 33.21, "learning_rate": 2.9790043180100393e-05, "loss": 0.0748, "step": 69190 }, { "epoch": 33.21, "learning_rate": 2.977691090696027e-05, "loss": 0.0733, "step": 69200 }, { "epoch": 33.22, "learning_rate": 2.976378030149966e-05, "loss": 0.0837, "step": 69210 }, { "epoch": 33.22, "learning_rate": 2.9750651364801385e-05, "loss": 0.0724, "step": 69220 }, { "epoch": 33.22, "learning_rate": 2.973752409794807e-05, "loss": 0.0846, "step": 69230 }, { "epoch": 33.23, "learning_rate": 2.9724398502022276e-05, "loss": 0.0772, "step": 69240 }, { "epoch": 33.23, "learning_rate": 2.9711274578106383e-05, "loss": 0.0779, "step": 69250 }, { "epoch": 33.24, "learning_rate": 2.9698152327282657e-05, "loss": 0.0777, "step": 69260 }, { "epoch": 33.24, "learning_rate": 2.968503175063319e-05, "loss": 0.0795, "step": 69270 }, { "epoch": 33.25, "learning_rate": 2.967191284923997e-05, "loss": 0.0851, "step": 69280 }, { "epoch": 33.25, "learning_rate": 2.9658795624184845e-05, "loss": 0.0833, "step": 69290 }, { "epoch": 33.26, "learning_rate": 2.9645680076549503e-05, "loss": 0.0733, "step": 69300 }, { "epoch": 33.26, "learning_rate": 2.9632566207415514e-05, "loss": 0.0792, "step": 69310 }, { "epoch": 33.27, "learning_rate": 2.961945401786429e-05, "loss": 0.0789, "step": 69320 }, { "epoch": 33.27, "learning_rate": 2.960634350897714e-05, "loss": 0.0765, "step": 69330 }, { "epoch": 33.27, "learning_rate": 2.959323468183518e-05, "loss": 0.0798, "step": 69340 }, { "epoch": 33.28, "learning_rate": 2.958012753751943e-05, "loss": 0.0841, "step": 69350 }, { "epoch": 33.28, "learning_rate": 2.9567022077110773e-05, "loss": 0.0749, "step": 69360 }, { "epoch": 33.29, "learning_rate": 2.955391830168994e-05, "loss": 0.0769, "step": 69370 }, { "epoch": 33.29, "learning_rate": 2.954081621233749e-05, "loss": 0.0762, "step": 69380 }, { "epoch": 33.3, "learning_rate": 2.9527715810133905e-05, "loss": 0.075, "step": 69390 }, { "epoch": 33.3, "learning_rate": 2.9514617096159507e-05, "loss": 0.086, "step": 69400 }, { "epoch": 33.31, "learning_rate": 2.9501520071494436e-05, "loss": 0.0892, "step": 69410 }, { "epoch": 33.31, "learning_rate": 2.9488424737218733e-05, "loss": 0.0751, "step": 69420 }, { "epoch": 33.32, "learning_rate": 2.947533109441232e-05, "loss": 0.0792, "step": 69430 }, { "epoch": 33.32, "learning_rate": 2.9462239144154956e-05, "loss": 0.0841, "step": 69440 }, { "epoch": 33.32, "learning_rate": 2.944914888752621e-05, "loss": 0.0769, "step": 69450 }, { "epoch": 33.33, "learning_rate": 2.9436060325605585e-05, "loss": 0.0727, "step": 69460 }, { "epoch": 33.33, "learning_rate": 2.942297345947246e-05, "loss": 0.0731, "step": 69470 }, { "epoch": 33.34, "learning_rate": 2.9409888290205954e-05, "loss": 0.0757, "step": 69480 }, { "epoch": 33.34, "learning_rate": 2.9396804818885176e-05, "loss": 0.0735, "step": 69490 }, { "epoch": 33.35, "learning_rate": 2.938372304658902e-05, "loss": 0.0839, "step": 69500 }, { "epoch": 33.35, "learning_rate": 2.9370642974396295e-05, "loss": 0.0805, "step": 69510 }, { "epoch": 33.36, "learning_rate": 2.93575646033856e-05, "loss": 0.0792, "step": 69520 }, { "epoch": 33.36, "learning_rate": 2.9344487934635444e-05, "loss": 0.0758, "step": 69530 }, { "epoch": 33.37, "learning_rate": 2.933141296922418e-05, "loss": 0.078, "step": 69540 }, { "epoch": 33.37, "learning_rate": 2.9318339708230058e-05, "loss": 0.0684, "step": 69550 }, { "epoch": 33.38, "learning_rate": 2.9305268152731107e-05, "loss": 0.0799, "step": 69560 }, { "epoch": 33.38, "learning_rate": 2.9292198303805275e-05, "loss": 0.0715, "step": 69570 }, { "epoch": 33.38, "learning_rate": 2.9279130162530377e-05, "loss": 0.0794, "step": 69580 }, { "epoch": 33.39, "learning_rate": 2.926606372998404e-05, "loss": 0.0725, "step": 69590 }, { "epoch": 33.39, "learning_rate": 2.9252999007243785e-05, "loss": 0.0713, "step": 69600 }, { "epoch": 33.4, "learning_rate": 2.923993599538697e-05, "loss": 0.09, "step": 69610 }, { "epoch": 33.4, "learning_rate": 2.9226874695490867e-05, "loss": 0.0752, "step": 69620 }, { "epoch": 33.41, "learning_rate": 2.9213815108632527e-05, "loss": 0.0748, "step": 69630 }, { "epoch": 33.41, "learning_rate": 2.9200757235888897e-05, "loss": 0.0908, "step": 69640 }, { "epoch": 33.42, "learning_rate": 2.9187701078336803e-05, "loss": 0.0744, "step": 69650 }, { "epoch": 33.42, "learning_rate": 2.9174646637052893e-05, "loss": 0.0882, "step": 69660 }, { "epoch": 33.43, "learning_rate": 2.9161593913113694e-05, "loss": 0.0796, "step": 69670 }, { "epoch": 33.43, "learning_rate": 2.9148542907595588e-05, "loss": 0.0665, "step": 69680 }, { "epoch": 33.43, "learning_rate": 2.9135493621574844e-05, "loss": 0.0833, "step": 69690 }, { "epoch": 33.44, "learning_rate": 2.912244605612749e-05, "loss": 0.0738, "step": 69700 }, { "epoch": 33.44, "learning_rate": 2.9109400212329546e-05, "loss": 0.0801, "step": 69710 }, { "epoch": 33.45, "learning_rate": 2.9096356091256803e-05, "loss": 0.0729, "step": 69720 }, { "epoch": 33.45, "learning_rate": 2.9083313693984958e-05, "loss": 0.0765, "step": 69730 }, { "epoch": 33.46, "learning_rate": 2.90702730215895e-05, "loss": 0.0826, "step": 69740 }, { "epoch": 33.46, "learning_rate": 2.9057234075145832e-05, "loss": 0.0855, "step": 69750 }, { "epoch": 33.47, "learning_rate": 2.9044196855729232e-05, "loss": 0.0901, "step": 69760 }, { "epoch": 33.47, "learning_rate": 2.903116136441475e-05, "loss": 0.0774, "step": 69770 }, { "epoch": 33.48, "learning_rate": 2.9018127602277346e-05, "loss": 0.0781, "step": 69780 }, { "epoch": 33.48, "learning_rate": 2.9005095570391882e-05, "loss": 0.0756, "step": 69790 }, { "epoch": 33.48, "learning_rate": 2.8992065269833036e-05, "loss": 0.0729, "step": 69800 }, { "epoch": 33.49, "learning_rate": 2.89790367016753e-05, "loss": 0.0831, "step": 69810 }, { "epoch": 33.49, "learning_rate": 2.8966009866993084e-05, "loss": 0.0849, "step": 69820 }, { "epoch": 33.5, "learning_rate": 2.895298476686064e-05, "loss": 0.0817, "step": 69830 }, { "epoch": 33.5, "learning_rate": 2.8939961402352055e-05, "loss": 0.0767, "step": 69840 }, { "epoch": 33.51, "learning_rate": 2.892693977454127e-05, "loss": 0.0734, "step": 69850 }, { "epoch": 33.51, "learning_rate": 2.891391988450216e-05, "loss": 0.0796, "step": 69860 }, { "epoch": 33.52, "learning_rate": 2.8900901733308382e-05, "loss": 0.0762, "step": 69870 }, { "epoch": 33.52, "learning_rate": 2.8887885322033437e-05, "loss": 0.074, "step": 69880 }, { "epoch": 33.53, "learning_rate": 2.887487065175073e-05, "loss": 0.0752, "step": 69890 }, { "epoch": 33.53, "learning_rate": 2.8861857723533502e-05, "loss": 0.0755, "step": 69900 }, { "epoch": 33.54, "learning_rate": 2.8848846538454877e-05, "loss": 0.0733, "step": 69910 }, { "epoch": 33.54, "learning_rate": 2.883583709758777e-05, "loss": 0.0812, "step": 69920 }, { "epoch": 33.54, "learning_rate": 2.882282940200499e-05, "loss": 0.0794, "step": 69930 }, { "epoch": 33.55, "learning_rate": 2.8809823452779262e-05, "loss": 0.074, "step": 69940 }, { "epoch": 33.55, "learning_rate": 2.8796819250983066e-05, "loss": 0.0704, "step": 69950 }, { "epoch": 33.56, "learning_rate": 2.878381679768878e-05, "loss": 0.0718, "step": 69960 }, { "epoch": 33.56, "learning_rate": 2.8770816093968656e-05, "loss": 0.0687, "step": 69970 }, { "epoch": 33.57, "learning_rate": 2.8757817140894804e-05, "loss": 0.0874, "step": 69980 }, { "epoch": 33.57, "learning_rate": 2.8744819939539125e-05, "loss": 0.0863, "step": 69990 }, { "epoch": 33.58, "learning_rate": 2.8731824490973446e-05, "loss": 0.0825, "step": 70000 }, { "epoch": 33.58, "learning_rate": 2.8718830796269435e-05, "loss": 0.0769, "step": 70010 }, { "epoch": 33.59, "learning_rate": 2.8705838856498584e-05, "loss": 0.0757, "step": 70020 }, { "epoch": 33.59, "learning_rate": 2.869284867273228e-05, "loss": 0.0831, "step": 70030 }, { "epoch": 33.59, "learning_rate": 2.867986024604174e-05, "loss": 0.0683, "step": 70040 }, { "epoch": 33.6, "learning_rate": 2.8666873577498066e-05, "loss": 0.0798, "step": 70050 }, { "epoch": 33.6, "learning_rate": 2.8653888668172146e-05, "loss": 0.0782, "step": 70060 }, { "epoch": 33.61, "learning_rate": 2.8640905519134797e-05, "loss": 0.0784, "step": 70070 }, { "epoch": 33.61, "learning_rate": 2.8627924131456664e-05, "loss": 0.0796, "step": 70080 }, { "epoch": 33.62, "learning_rate": 2.861494450620824e-05, "loss": 0.0813, "step": 70090 }, { "epoch": 33.62, "learning_rate": 2.8601966644459883e-05, "loss": 0.0853, "step": 70100 }, { "epoch": 33.63, "learning_rate": 2.8588990547281797e-05, "loss": 0.082, "step": 70110 }, { "epoch": 33.63, "learning_rate": 2.8576016215744068e-05, "loss": 0.0728, "step": 70120 }, { "epoch": 33.64, "learning_rate": 2.8563043650916565e-05, "loss": 0.0881, "step": 70130 }, { "epoch": 33.64, "learning_rate": 2.8550072853869092e-05, "loss": 0.0775, "step": 70140 }, { "epoch": 33.64, "learning_rate": 2.8537103825671267e-05, "loss": 0.0893, "step": 70150 }, { "epoch": 33.65, "learning_rate": 2.8524136567392577e-05, "loss": 0.0888, "step": 70160 }, { "epoch": 33.65, "learning_rate": 2.851117108010234e-05, "loss": 0.0757, "step": 70170 }, { "epoch": 33.66, "learning_rate": 2.8498207364869762e-05, "loss": 0.0893, "step": 70180 }, { "epoch": 33.66, "learning_rate": 2.848524542276389e-05, "loss": 0.0847, "step": 70190 }, { "epoch": 33.67, "learning_rate": 2.8472285254853593e-05, "loss": 0.0794, "step": 70200 }, { "epoch": 33.67, "learning_rate": 2.8459326862207625e-05, "loss": 0.0792, "step": 70210 }, { "epoch": 33.68, "learning_rate": 2.8446370245894593e-05, "loss": 0.0793, "step": 70220 }, { "epoch": 33.68, "learning_rate": 2.8433415406982965e-05, "loss": 0.0949, "step": 70230 }, { "epoch": 33.69, "learning_rate": 2.842046234654104e-05, "loss": 0.08, "step": 70240 }, { "epoch": 33.69, "learning_rate": 2.8407511065636978e-05, "loss": 0.0741, "step": 70250 }, { "epoch": 33.7, "learning_rate": 2.8394561565338823e-05, "loss": 0.0771, "step": 70260 }, { "epoch": 33.7, "learning_rate": 2.83816138467144e-05, "loss": 0.0824, "step": 70270 }, { "epoch": 33.7, "learning_rate": 2.8368667910831453e-05, "loss": 0.0773, "step": 70280 }, { "epoch": 33.71, "learning_rate": 2.835572375875756e-05, "loss": 0.0755, "step": 70290 }, { "epoch": 33.71, "learning_rate": 2.834278139156014e-05, "loss": 0.0809, "step": 70300 }, { "epoch": 33.72, "learning_rate": 2.8329840810306483e-05, "loss": 0.0818, "step": 70310 }, { "epoch": 33.72, "learning_rate": 2.8316902016063718e-05, "loss": 0.0745, "step": 70320 }, { "epoch": 33.73, "learning_rate": 2.8303965009898837e-05, "loss": 0.0766, "step": 70330 }, { "epoch": 33.73, "learning_rate": 2.82910297928787e-05, "loss": 0.0768, "step": 70340 }, { "epoch": 33.74, "learning_rate": 2.8278096366069946e-05, "loss": 0.079, "step": 70350 }, { "epoch": 33.74, "learning_rate": 2.826516473053915e-05, "loss": 0.0828, "step": 70360 }, { "epoch": 33.75, "learning_rate": 2.8252234887352725e-05, "loss": 0.0781, "step": 70370 }, { "epoch": 33.75, "learning_rate": 2.8239306837576857e-05, "loss": 0.0768, "step": 70380 }, { "epoch": 33.75, "learning_rate": 2.8226380582277712e-05, "loss": 0.0805, "step": 70390 }, { "epoch": 33.76, "learning_rate": 2.8213456122521215e-05, "loss": 0.0803, "step": 70400 }, { "epoch": 33.76, "learning_rate": 2.8200533459373197e-05, "loss": 0.0803, "step": 70410 }, { "epoch": 33.77, "learning_rate": 2.8187612593899272e-05, "loss": 0.0705, "step": 70420 }, { "epoch": 33.77, "learning_rate": 2.8174693527164965e-05, "loss": 0.0736, "step": 70430 }, { "epoch": 33.78, "learning_rate": 2.816177626023566e-05, "loss": 0.0735, "step": 70440 }, { "epoch": 33.78, "learning_rate": 2.8148860794176507e-05, "loss": 0.0843, "step": 70450 }, { "epoch": 33.79, "learning_rate": 2.813594713005263e-05, "loss": 0.0908, "step": 70460 }, { "epoch": 33.79, "learning_rate": 2.8123035268928923e-05, "loss": 0.0904, "step": 70470 }, { "epoch": 33.8, "learning_rate": 2.8110125211870163e-05, "loss": 0.0888, "step": 70480 }, { "epoch": 33.8, "learning_rate": 2.809721695994093e-05, "loss": 0.0757, "step": 70490 }, { "epoch": 33.8, "learning_rate": 2.8084310514205725e-05, "loss": 0.0859, "step": 70500 }, { "epoch": 33.81, "learning_rate": 2.8071405875728852e-05, "loss": 0.0813, "step": 70510 }, { "epoch": 33.81, "learning_rate": 2.8058503045574514e-05, "loss": 0.0837, "step": 70520 }, { "epoch": 33.82, "learning_rate": 2.804560202480665e-05, "loss": 0.0867, "step": 70530 }, { "epoch": 33.82, "learning_rate": 2.803270281448922e-05, "loss": 0.0795, "step": 70540 }, { "epoch": 33.83, "learning_rate": 2.8019805415685918e-05, "loss": 0.0792, "step": 70550 }, { "epoch": 33.83, "learning_rate": 2.80069098294603e-05, "loss": 0.0813, "step": 70560 }, { "epoch": 33.84, "learning_rate": 2.7994016056875793e-05, "loss": 0.0783, "step": 70570 }, { "epoch": 33.84, "learning_rate": 2.7981124098995683e-05, "loss": 0.0798, "step": 70580 }, { "epoch": 33.85, "learning_rate": 2.7968233956883106e-05, "loss": 0.0667, "step": 70590 }, { "epoch": 33.85, "learning_rate": 2.7955345631600993e-05, "loss": 0.0859, "step": 70600 }, { "epoch": 33.86, "learning_rate": 2.7942459124212184e-05, "loss": 0.0815, "step": 70610 }, { "epoch": 33.86, "learning_rate": 2.7929574435779393e-05, "loss": 0.0771, "step": 70620 }, { "epoch": 33.86, "learning_rate": 2.7916691567365106e-05, "loss": 0.08, "step": 70630 }, { "epoch": 33.87, "learning_rate": 2.7903810520031703e-05, "loss": 0.0785, "step": 70640 }, { "epoch": 33.87, "learning_rate": 2.7890931294841417e-05, "loss": 0.0775, "step": 70650 }, { "epoch": 33.88, "learning_rate": 2.7878053892856333e-05, "loss": 0.0739, "step": 70660 }, { "epoch": 33.88, "learning_rate": 2.786517831513834e-05, "loss": 0.0749, "step": 70670 }, { "epoch": 33.89, "learning_rate": 2.785230456274923e-05, "loss": 0.0837, "step": 70680 }, { "epoch": 33.89, "learning_rate": 2.7839432636750605e-05, "loss": 0.0763, "step": 70690 }, { "epoch": 33.9, "learning_rate": 2.7826562538204004e-05, "loss": 0.0878, "step": 70700 }, { "epoch": 33.9, "learning_rate": 2.7813694268170687e-05, "loss": 0.0763, "step": 70710 }, { "epoch": 33.91, "learning_rate": 2.7800827827711828e-05, "loss": 0.0783, "step": 70720 }, { "epoch": 33.91, "learning_rate": 2.778796321788849e-05, "loss": 0.0778, "step": 70730 }, { "epoch": 33.91, "learning_rate": 2.7775100439761493e-05, "loss": 0.0796, "step": 70740 }, { "epoch": 33.92, "learning_rate": 2.7762239494391573e-05, "loss": 0.0776, "step": 70750 }, { "epoch": 33.92, "learning_rate": 2.7749380382839296e-05, "loss": 0.0881, "step": 70760 }, { "epoch": 33.93, "learning_rate": 2.773652310616508e-05, "loss": 0.0728, "step": 70770 }, { "epoch": 33.93, "learning_rate": 2.772366766542919e-05, "loss": 0.0823, "step": 70780 }, { "epoch": 33.94, "learning_rate": 2.7710814061691724e-05, "loss": 0.0854, "step": 70790 }, { "epoch": 33.94, "learning_rate": 2.7697962296012693e-05, "loss": 0.0884, "step": 70800 }, { "epoch": 33.95, "learning_rate": 2.7685112369451838e-05, "loss": 0.0822, "step": 70810 }, { "epoch": 33.95, "learning_rate": 2.767226428306886e-05, "loss": 0.1181, "step": 70820 }, { "epoch": 34.0, "learning_rate": 2.7659418037923245e-05, "loss": 0.0771, "step": 70830 }, { "epoch": 34.01, "learning_rate": 2.7646573635074363e-05, "loss": 0.0811, "step": 70840 }, { "epoch": 34.01, "learning_rate": 2.7633731075581405e-05, "loss": 0.0773, "step": 70850 }, { "epoch": 34.02, "learning_rate": 2.762089036050343e-05, "loss": 0.0803, "step": 70860 }, { "epoch": 34.02, "learning_rate": 2.7608051490899358e-05, "loss": 0.0882, "step": 70870 }, { "epoch": 34.03, "learning_rate": 2.7595214467827885e-05, "loss": 0.0735, "step": 70880 }, { "epoch": 34.03, "learning_rate": 2.7582379292347633e-05, "loss": 0.0776, "step": 70890 }, { "epoch": 34.04, "learning_rate": 2.756954596551704e-05, "loss": 0.0698, "step": 70900 }, { "epoch": 34.04, "learning_rate": 2.7556714488394398e-05, "loss": 0.0796, "step": 70910 }, { "epoch": 34.04, "learning_rate": 2.754388486203784e-05, "loss": 0.0733, "step": 70920 }, { "epoch": 34.05, "learning_rate": 2.753105708750535e-05, "loss": 0.0834, "step": 70930 }, { "epoch": 34.05, "learning_rate": 2.751823116585477e-05, "loss": 0.0909, "step": 70940 }, { "epoch": 34.06, "learning_rate": 2.7505407098143775e-05, "loss": 0.0843, "step": 70950 }, { "epoch": 34.06, "learning_rate": 2.7492584885429877e-05, "loss": 0.071, "step": 70960 }, { "epoch": 34.07, "learning_rate": 2.7479764528770457e-05, "loss": 0.0781, "step": 70970 }, { "epoch": 34.07, "learning_rate": 2.746694602922273e-05, "loss": 0.0773, "step": 70980 }, { "epoch": 34.08, "learning_rate": 2.7454129387843774e-05, "loss": 0.0693, "step": 70990 }, { "epoch": 34.08, "learning_rate": 2.7441314605690488e-05, "loss": 0.0749, "step": 71000 }, { "epoch": 34.09, "learning_rate": 2.7428501683819642e-05, "loss": 0.0778, "step": 71010 }, { "epoch": 34.09, "learning_rate": 2.741569062328786e-05, "loss": 0.0792, "step": 71020 }, { "epoch": 34.1, "learning_rate": 2.7402881425151562e-05, "loss": 0.0742, "step": 71030 }, { "epoch": 34.1, "learning_rate": 2.739007409046705e-05, "loss": 0.0774, "step": 71040 }, { "epoch": 34.1, "learning_rate": 2.7377268620290486e-05, "loss": 0.0772, "step": 71050 }, { "epoch": 34.11, "learning_rate": 2.7364465015677854e-05, "loss": 0.0837, "step": 71060 }, { "epoch": 34.11, "learning_rate": 2.7351663277684993e-05, "loss": 0.0774, "step": 71070 }, { "epoch": 34.12, "learning_rate": 2.7338863407367586e-05, "loss": 0.0762, "step": 71080 }, { "epoch": 34.12, "learning_rate": 2.7326065405781186e-05, "loss": 0.072, "step": 71090 }, { "epoch": 34.13, "learning_rate": 2.7313269273981135e-05, "loss": 0.0818, "step": 71100 }, { "epoch": 34.13, "learning_rate": 2.7300475013022663e-05, "loss": 0.079, "step": 71110 }, { "epoch": 34.14, "learning_rate": 2.7287682623960848e-05, "loss": 0.0816, "step": 71120 }, { "epoch": 34.14, "learning_rate": 2.7274892107850598e-05, "loss": 0.0748, "step": 71130 }, { "epoch": 34.15, "learning_rate": 2.726210346574667e-05, "loss": 0.0817, "step": 71140 }, { "epoch": 34.15, "learning_rate": 2.724931669870367e-05, "loss": 0.0838, "step": 71150 }, { "epoch": 34.15, "learning_rate": 2.7236531807776068e-05, "loss": 0.0796, "step": 71160 }, { "epoch": 34.16, "learning_rate": 2.7223748794018124e-05, "loss": 0.076, "step": 71170 }, { "epoch": 34.16, "learning_rate": 2.7210967658483986e-05, "loss": 0.0751, "step": 71180 }, { "epoch": 34.17, "learning_rate": 2.7198188402227653e-05, "loss": 0.0798, "step": 71190 }, { "epoch": 34.17, "learning_rate": 2.7185411026302964e-05, "loss": 0.0769, "step": 71200 }, { "epoch": 34.18, "learning_rate": 2.7172635531763548e-05, "loss": 0.0786, "step": 71210 }, { "epoch": 34.18, "learning_rate": 2.7159861919662967e-05, "loss": 0.0784, "step": 71220 }, { "epoch": 34.19, "learning_rate": 2.7147090191054602e-05, "loss": 0.0784, "step": 71230 }, { "epoch": 34.19, "learning_rate": 2.713432034699161e-05, "loss": 0.0742, "step": 71240 }, { "epoch": 34.2, "learning_rate": 2.712155238852707e-05, "loss": 0.0808, "step": 71250 }, { "epoch": 34.2, "learning_rate": 2.7108786316713885e-05, "loss": 0.0874, "step": 71260 }, { "epoch": 34.2, "learning_rate": 2.7096022132604813e-05, "loss": 0.0759, "step": 71270 }, { "epoch": 34.21, "learning_rate": 2.7083259837252407e-05, "loss": 0.0814, "step": 71280 }, { "epoch": 34.21, "learning_rate": 2.7070499431709097e-05, "loss": 0.0797, "step": 71290 }, { "epoch": 34.22, "learning_rate": 2.7057740917027186e-05, "loss": 0.0762, "step": 71300 }, { "epoch": 34.22, "learning_rate": 2.704498429425881e-05, "loss": 0.0811, "step": 71310 }, { "epoch": 34.23, "learning_rate": 2.7032229564455886e-05, "loss": 0.0753, "step": 71320 }, { "epoch": 34.23, "learning_rate": 2.7019476728670246e-05, "loss": 0.0818, "step": 71330 }, { "epoch": 34.24, "learning_rate": 2.7006725787953546e-05, "loss": 0.0695, "step": 71340 }, { "epoch": 34.24, "learning_rate": 2.6993976743357264e-05, "loss": 0.0807, "step": 71350 }, { "epoch": 34.25, "learning_rate": 2.698122959593275e-05, "loss": 0.0825, "step": 71360 }, { "epoch": 34.25, "learning_rate": 2.696848434673116e-05, "loss": 0.0885, "step": 71370 }, { "epoch": 34.26, "learning_rate": 2.695574099680359e-05, "loss": 0.0716, "step": 71380 }, { "epoch": 34.26, "learning_rate": 2.6942999547200838e-05, "loss": 0.078, "step": 71390 }, { "epoch": 34.26, "learning_rate": 2.6930259998973646e-05, "loss": 0.0916, "step": 71400 }, { "epoch": 34.27, "learning_rate": 2.6917522353172587e-05, "loss": 0.0847, "step": 71410 }, { "epoch": 34.27, "learning_rate": 2.6904786610848014e-05, "loss": 0.0813, "step": 71420 }, { "epoch": 34.28, "learning_rate": 2.68920527730502e-05, "loss": 0.0749, "step": 71430 }, { "epoch": 34.28, "learning_rate": 2.6879320840829224e-05, "loss": 0.078, "step": 71440 }, { "epoch": 34.29, "learning_rate": 2.6866590815235016e-05, "loss": 0.0821, "step": 71450 }, { "epoch": 34.29, "learning_rate": 2.685386269731734e-05, "loss": 0.0756, "step": 71460 }, { "epoch": 34.3, "learning_rate": 2.6841136488125817e-05, "loss": 0.0798, "step": 71470 }, { "epoch": 34.3, "learning_rate": 2.6828412188709916e-05, "loss": 0.076, "step": 71480 }, { "epoch": 34.31, "learning_rate": 2.6815689800118897e-05, "loss": 0.0697, "step": 71490 }, { "epoch": 34.31, "learning_rate": 2.680296932340193e-05, "loss": 0.0765, "step": 71500 }, { "epoch": 34.31, "learning_rate": 2.6790250759607986e-05, "loss": 0.0785, "step": 71510 }, { "epoch": 34.32, "learning_rate": 2.6777534109785894e-05, "loss": 0.0801, "step": 71520 }, { "epoch": 34.32, "learning_rate": 2.676481937498433e-05, "loss": 0.081, "step": 71530 }, { "epoch": 34.33, "learning_rate": 2.675210655625179e-05, "loss": 0.0747, "step": 71540 }, { "epoch": 34.33, "learning_rate": 2.6739395654636633e-05, "loss": 0.0702, "step": 71550 }, { "epoch": 34.34, "learning_rate": 2.6726686671187072e-05, "loss": 0.0689, "step": 71560 }, { "epoch": 34.34, "learning_rate": 2.6713979606951102e-05, "loss": 0.0857, "step": 71570 }, { "epoch": 34.35, "learning_rate": 2.6701274462976615e-05, "loss": 0.0765, "step": 71580 }, { "epoch": 34.35, "learning_rate": 2.668857124031134e-05, "loss": 0.0798, "step": 71590 }, { "epoch": 34.36, "learning_rate": 2.6675869940002828e-05, "loss": 0.073, "step": 71600 }, { "epoch": 34.36, "learning_rate": 2.6663170563098485e-05, "loss": 0.0738, "step": 71610 }, { "epoch": 34.36, "learning_rate": 2.6650473110645557e-05, "loss": 0.0706, "step": 71620 }, { "epoch": 34.37, "learning_rate": 2.6637777583691137e-05, "loss": 0.068, "step": 71630 }, { "epoch": 34.37, "learning_rate": 2.6625083983282124e-05, "loss": 0.0776, "step": 71640 }, { "epoch": 34.38, "learning_rate": 2.6612392310465294e-05, "loss": 0.0782, "step": 71650 }, { "epoch": 34.38, "learning_rate": 2.6599702566287266e-05, "loss": 0.0741, "step": 71660 }, { "epoch": 34.39, "learning_rate": 2.658701475179448e-05, "loss": 0.0726, "step": 71670 }, { "epoch": 34.39, "learning_rate": 2.6574328868033226e-05, "loss": 0.0762, "step": 71680 }, { "epoch": 34.4, "learning_rate": 2.6561644916049643e-05, "loss": 0.0746, "step": 71690 }, { "epoch": 34.4, "learning_rate": 2.654896289688972e-05, "loss": 0.0824, "step": 71700 }, { "epoch": 34.41, "learning_rate": 2.653628281159921e-05, "loss": 0.0691, "step": 71710 }, { "epoch": 34.41, "learning_rate": 2.652360466122381e-05, "loss": 0.0694, "step": 71720 }, { "epoch": 34.41, "learning_rate": 2.6510928446808996e-05, "loss": 0.0832, "step": 71730 }, { "epoch": 34.42, "learning_rate": 2.6498254169400112e-05, "loss": 0.079, "step": 71740 }, { "epoch": 34.42, "learning_rate": 2.6485581830042328e-05, "loss": 0.0724, "step": 71750 }, { "epoch": 34.43, "learning_rate": 2.647291142978065e-05, "loss": 0.0733, "step": 71760 }, { "epoch": 34.43, "learning_rate": 2.6460242969659966e-05, "loss": 0.0852, "step": 71770 }, { "epoch": 34.44, "learning_rate": 2.6447576450724915e-05, "loss": 0.0793, "step": 71780 }, { "epoch": 34.44, "learning_rate": 2.6434911874020057e-05, "loss": 0.0882, "step": 71790 }, { "epoch": 34.45, "learning_rate": 2.6422249240589768e-05, "loss": 0.0757, "step": 71800 }, { "epoch": 34.45, "learning_rate": 2.640958855147826e-05, "loss": 0.081, "step": 71810 }, { "epoch": 34.46, "learning_rate": 2.6396929807729582e-05, "loss": 0.0766, "step": 71820 }, { "epoch": 34.46, "learning_rate": 2.638427301038763e-05, "loss": 0.0725, "step": 71830 }, { "epoch": 34.47, "learning_rate": 2.6371618160496148e-05, "loss": 0.0691, "step": 71840 }, { "epoch": 34.47, "learning_rate": 2.635896525909868e-05, "loss": 0.079, "step": 71850 }, { "epoch": 34.47, "learning_rate": 2.634631430723865e-05, "loss": 0.0803, "step": 71860 }, { "epoch": 34.48, "learning_rate": 2.6333665305959304e-05, "loss": 0.0827, "step": 71870 }, { "epoch": 34.48, "learning_rate": 2.6321018256303754e-05, "loss": 0.0753, "step": 71880 }, { "epoch": 34.49, "learning_rate": 2.6308373159314876e-05, "loss": 0.0795, "step": 71890 }, { "epoch": 34.49, "learning_rate": 2.629573001603548e-05, "loss": 0.0737, "step": 71900 }, { "epoch": 34.5, "learning_rate": 2.6283088827508173e-05, "loss": 0.086, "step": 71910 }, { "epoch": 34.5, "learning_rate": 2.6270449594775405e-05, "loss": 0.081, "step": 71920 }, { "epoch": 34.51, "learning_rate": 2.6257812318879415e-05, "loss": 0.0838, "step": 71930 }, { "epoch": 34.51, "learning_rate": 2.6245177000862352e-05, "loss": 0.0701, "step": 71940 }, { "epoch": 34.52, "learning_rate": 2.6232543641766206e-05, "loss": 0.0756, "step": 71950 }, { "epoch": 34.52, "learning_rate": 2.6219912242632726e-05, "loss": 0.0785, "step": 71960 }, { "epoch": 34.52, "learning_rate": 2.6207282804503548e-05, "loss": 0.0768, "step": 71970 }, { "epoch": 34.53, "learning_rate": 2.6194655328420192e-05, "loss": 0.0724, "step": 71980 }, { "epoch": 34.53, "learning_rate": 2.618202981542397e-05, "loss": 0.0914, "step": 71990 }, { "epoch": 34.54, "learning_rate": 2.6169406266555983e-05, "loss": 0.0783, "step": 72000 }, { "epoch": 34.54, "learning_rate": 2.6156784682857267e-05, "loss": 0.074, "step": 72010 }, { "epoch": 34.55, "learning_rate": 2.6144165065368653e-05, "loss": 0.0817, "step": 72020 }, { "epoch": 34.55, "learning_rate": 2.6131547415130763e-05, "loss": 0.0767, "step": 72030 }, { "epoch": 34.56, "learning_rate": 2.6118931733184126e-05, "loss": 0.0731, "step": 72040 }, { "epoch": 34.56, "learning_rate": 2.6106318020569064e-05, "loss": 0.0814, "step": 72050 }, { "epoch": 34.57, "learning_rate": 2.6093706278325825e-05, "loss": 0.0774, "step": 72060 }, { "epoch": 34.57, "learning_rate": 2.6081096507494345e-05, "loss": 0.0749, "step": 72070 }, { "epoch": 34.57, "learning_rate": 2.6068488709114513e-05, "loss": 0.0787, "step": 72080 }, { "epoch": 34.58, "learning_rate": 2.6055882884226035e-05, "loss": 0.0702, "step": 72090 }, { "epoch": 34.58, "learning_rate": 2.6043279033868407e-05, "loss": 0.072, "step": 72100 }, { "epoch": 34.59, "learning_rate": 2.6030677159080995e-05, "loss": 0.075, "step": 72110 }, { "epoch": 34.59, "learning_rate": 2.601807726090303e-05, "loss": 0.0786, "step": 72120 }, { "epoch": 34.6, "learning_rate": 2.600547934037352e-05, "loss": 0.0739, "step": 72130 }, { "epoch": 34.6, "learning_rate": 2.599288339853137e-05, "loss": 0.0762, "step": 72140 }, { "epoch": 34.61, "learning_rate": 2.5980289436415277e-05, "loss": 0.0721, "step": 72150 }, { "epoch": 34.61, "learning_rate": 2.59676974550638e-05, "loss": 0.0756, "step": 72160 }, { "epoch": 34.62, "learning_rate": 2.5955107455515336e-05, "loss": 0.0802, "step": 72170 }, { "epoch": 34.62, "learning_rate": 2.5942519438808078e-05, "loss": 0.0773, "step": 72180 }, { "epoch": 34.63, "learning_rate": 2.5929933405980094e-05, "loss": 0.0803, "step": 72190 }, { "epoch": 34.63, "learning_rate": 2.591734935806929e-05, "loss": 0.0808, "step": 72200 }, { "epoch": 34.63, "learning_rate": 2.5904767296113404e-05, "loss": 0.0816, "step": 72210 }, { "epoch": 34.64, "learning_rate": 2.5892187221149993e-05, "loss": 0.0721, "step": 72220 }, { "epoch": 34.64, "learning_rate": 2.5879609134216463e-05, "loss": 0.0766, "step": 72230 }, { "epoch": 34.65, "learning_rate": 2.5867033036350085e-05, "loss": 0.0753, "step": 72240 }, { "epoch": 34.65, "learning_rate": 2.5854458928587888e-05, "loss": 0.0777, "step": 72250 }, { "epoch": 34.66, "learning_rate": 2.58418868119668e-05, "loss": 0.0729, "step": 72260 }, { "epoch": 34.66, "learning_rate": 2.5829316687523586e-05, "loss": 0.0741, "step": 72270 }, { "epoch": 34.67, "learning_rate": 2.581674855629481e-05, "loss": 0.0721, "step": 72280 }, { "epoch": 34.67, "learning_rate": 2.5804182419316908e-05, "loss": 0.0787, "step": 72290 }, { "epoch": 34.68, "learning_rate": 2.579161827762613e-05, "loss": 0.0899, "step": 72300 }, { "epoch": 34.68, "learning_rate": 2.5779056132258582e-05, "loss": 0.0745, "step": 72310 }, { "epoch": 34.68, "learning_rate": 2.576649598425016e-05, "loss": 0.0727, "step": 72320 }, { "epoch": 34.69, "learning_rate": 2.5753937834636643e-05, "loss": 0.0764, "step": 72330 }, { "epoch": 34.69, "learning_rate": 2.5741381684453625e-05, "loss": 0.0754, "step": 72340 }, { "epoch": 34.7, "learning_rate": 2.572882753473654e-05, "loss": 0.0747, "step": 72350 }, { "epoch": 34.7, "learning_rate": 2.571627538652065e-05, "loss": 0.0714, "step": 72360 }, { "epoch": 34.71, "learning_rate": 2.570372524084107e-05, "loss": 0.0845, "step": 72370 }, { "epoch": 34.71, "learning_rate": 2.5691177098732738e-05, "loss": 0.0816, "step": 72380 }, { "epoch": 34.72, "learning_rate": 2.5678630961230397e-05, "loss": 0.0802, "step": 72390 }, { "epoch": 34.72, "learning_rate": 2.5666086829368678e-05, "loss": 0.0745, "step": 72400 }, { "epoch": 34.73, "learning_rate": 2.5653544704182014e-05, "loss": 0.0725, "step": 72410 }, { "epoch": 34.73, "learning_rate": 2.5641004586704688e-05, "loss": 0.074, "step": 72420 }, { "epoch": 34.73, "learning_rate": 2.5628466477970793e-05, "loss": 0.0796, "step": 72430 }, { "epoch": 34.74, "learning_rate": 2.5615930379014296e-05, "loss": 0.0835, "step": 72440 }, { "epoch": 34.74, "learning_rate": 2.5603396290868975e-05, "loss": 0.0723, "step": 72450 }, { "epoch": 34.75, "learning_rate": 2.5590864214568415e-05, "loss": 0.0808, "step": 72460 }, { "epoch": 34.75, "learning_rate": 2.5578334151146082e-05, "loss": 0.0722, "step": 72470 }, { "epoch": 34.76, "learning_rate": 2.5565806101635258e-05, "loss": 0.0829, "step": 72480 }, { "epoch": 34.76, "learning_rate": 2.555328006706905e-05, "loss": 0.0813, "step": 72490 }, { "epoch": 34.77, "learning_rate": 2.554075604848041e-05, "loss": 0.0792, "step": 72500 }, { "epoch": 34.77, "learning_rate": 2.5528234046902122e-05, "loss": 0.0759, "step": 72510 }, { "epoch": 34.78, "learning_rate": 2.55157140633668e-05, "loss": 0.0761, "step": 72520 }, { "epoch": 34.78, "learning_rate": 2.550319609890692e-05, "loss": 0.0702, "step": 72530 }, { "epoch": 34.79, "learning_rate": 2.5490680154554712e-05, "loss": 0.0811, "step": 72540 }, { "epoch": 34.79, "learning_rate": 2.5478166231342333e-05, "loss": 0.0795, "step": 72550 }, { "epoch": 34.79, "learning_rate": 2.546565433030173e-05, "loss": 0.0832, "step": 72560 }, { "epoch": 34.8, "learning_rate": 2.545314445246464e-05, "loss": 0.0732, "step": 72570 }, { "epoch": 34.8, "learning_rate": 2.544063659886274e-05, "loss": 0.0751, "step": 72580 }, { "epoch": 34.81, "learning_rate": 2.5428130770527448e-05, "loss": 0.0861, "step": 72590 }, { "epoch": 34.81, "learning_rate": 2.5415626968490074e-05, "loss": 0.0686, "step": 72600 }, { "epoch": 34.82, "learning_rate": 2.5403125193781696e-05, "loss": 0.0775, "step": 72610 }, { "epoch": 34.82, "learning_rate": 2.539062544743328e-05, "loss": 0.0753, "step": 72620 }, { "epoch": 34.83, "learning_rate": 2.537812773047562e-05, "loss": 0.086, "step": 72630 }, { "epoch": 34.83, "learning_rate": 2.5365632043939293e-05, "loss": 0.0784, "step": 72640 }, { "epoch": 34.84, "learning_rate": 2.5353138388854754e-05, "loss": 0.0858, "step": 72650 }, { "epoch": 34.84, "learning_rate": 2.5340646766252308e-05, "loss": 0.0751, "step": 72660 }, { "epoch": 34.84, "learning_rate": 2.5328157177162076e-05, "loss": 0.085, "step": 72670 }, { "epoch": 34.85, "learning_rate": 2.5315669622613957e-05, "loss": 0.0668, "step": 72680 }, { "epoch": 34.85, "learning_rate": 2.5303184103637745e-05, "loss": 0.0795, "step": 72690 }, { "epoch": 34.86, "learning_rate": 2.5290700621263046e-05, "loss": 0.0773, "step": 72700 }, { "epoch": 34.86, "learning_rate": 2.5278219176519325e-05, "loss": 0.074, "step": 72710 }, { "epoch": 34.87, "learning_rate": 2.5265739770435813e-05, "loss": 0.0736, "step": 72720 }, { "epoch": 34.87, "learning_rate": 2.525326240404161e-05, "loss": 0.0783, "step": 72730 }, { "epoch": 34.88, "learning_rate": 2.5240787078365715e-05, "loss": 0.0784, "step": 72740 }, { "epoch": 34.88, "learning_rate": 2.5228313794436832e-05, "loss": 0.075, "step": 72750 }, { "epoch": 34.89, "learning_rate": 2.5215842553283576e-05, "loss": 0.0744, "step": 72760 }, { "epoch": 34.89, "learning_rate": 2.5203373355934377e-05, "loss": 0.0787, "step": 72770 }, { "epoch": 34.89, "learning_rate": 2.519090620341753e-05, "loss": 0.0779, "step": 72780 }, { "epoch": 34.9, "learning_rate": 2.5178441096761075e-05, "loss": 0.0731, "step": 72790 }, { "epoch": 34.9, "learning_rate": 2.5165978036992943e-05, "loss": 0.0818, "step": 72800 }, { "epoch": 34.91, "learning_rate": 2.515351702514094e-05, "loss": 0.0764, "step": 72810 }, { "epoch": 34.91, "learning_rate": 2.5141058062232602e-05, "loss": 0.0697, "step": 72820 }, { "epoch": 34.92, "learning_rate": 2.5128601149295362e-05, "loss": 0.0749, "step": 72830 }, { "epoch": 34.92, "learning_rate": 2.5116146287356474e-05, "loss": 0.0724, "step": 72840 }, { "epoch": 34.93, "learning_rate": 2.510369347744303e-05, "loss": 0.0787, "step": 72850 }, { "epoch": 34.93, "learning_rate": 2.5091242720581902e-05, "loss": 0.07, "step": 72860 }, { "epoch": 34.94, "learning_rate": 2.5078794017799857e-05, "loss": 0.0851, "step": 72870 }, { "epoch": 34.94, "learning_rate": 2.5066347370123456e-05, "loss": 0.077, "step": 72880 }, { "epoch": 34.95, "learning_rate": 2.5053902778579115e-05, "loss": 0.0824, "step": 72890 }, { "epoch": 34.95, "learning_rate": 2.5041460244193047e-05, "loss": 0.0797, "step": 72900 }, { "epoch": 35.0, "learning_rate": 2.5029019767991335e-05, "loss": 0.0933, "step": 72910 }, { "epoch": 35.01, "learning_rate": 2.5016581350999884e-05, "loss": 0.0739, "step": 72920 }, { "epoch": 35.01, "learning_rate": 2.500414499424437e-05, "loss": 0.0825, "step": 72930 }, { "epoch": 35.02, "learning_rate": 2.499171069875038e-05, "loss": 0.0741, "step": 72940 }, { "epoch": 35.02, "learning_rate": 2.4979278465543287e-05, "loss": 0.072, "step": 72950 }, { "epoch": 35.03, "learning_rate": 2.4966848295648303e-05, "loss": 0.0776, "step": 72960 }, { "epoch": 35.03, "learning_rate": 2.4954420190090484e-05, "loss": 0.0759, "step": 72970 }, { "epoch": 35.03, "learning_rate": 2.4941994149894687e-05, "loss": 0.0806, "step": 72980 }, { "epoch": 35.04, "learning_rate": 2.4929570176085648e-05, "loss": 0.074, "step": 72990 }, { "epoch": 35.04, "learning_rate": 2.491714826968785e-05, "loss": 0.0741, "step": 73000 }, { "epoch": 35.05, "learning_rate": 2.4904728431725682e-05, "loss": 0.0785, "step": 73010 }, { "epoch": 35.05, "learning_rate": 2.4892310663223328e-05, "loss": 0.0822, "step": 73020 }, { "epoch": 35.06, "learning_rate": 2.4879894965204813e-05, "loss": 0.0721, "step": 73030 }, { "epoch": 35.06, "learning_rate": 2.4867481338693988e-05, "loss": 0.0718, "step": 73040 }, { "epoch": 35.07, "learning_rate": 2.4855069784714528e-05, "loss": 0.0718, "step": 73050 }, { "epoch": 35.07, "learning_rate": 2.484266030428996e-05, "loss": 0.0699, "step": 73060 }, { "epoch": 35.08, "learning_rate": 2.483025289844359e-05, "loss": 0.0677, "step": 73070 }, { "epoch": 35.08, "learning_rate": 2.4817847568198595e-05, "loss": 0.074, "step": 73080 }, { "epoch": 35.08, "learning_rate": 2.480544431457797e-05, "loss": 0.0759, "step": 73090 }, { "epoch": 35.09, "learning_rate": 2.4793043138604544e-05, "loss": 0.0739, "step": 73100 }, { "epoch": 35.09, "learning_rate": 2.4780644041300966e-05, "loss": 0.0796, "step": 73110 }, { "epoch": 35.1, "learning_rate": 2.4768247023689716e-05, "loss": 0.0668, "step": 73120 }, { "epoch": 35.1, "learning_rate": 2.47558520867931e-05, "loss": 0.0673, "step": 73130 }, { "epoch": 35.11, "learning_rate": 2.4743459231633288e-05, "loss": 0.0837, "step": 73140 }, { "epoch": 35.11, "learning_rate": 2.4731068459232197e-05, "loss": 0.0788, "step": 73150 }, { "epoch": 35.12, "learning_rate": 2.4718679770611643e-05, "loss": 0.0876, "step": 73160 }, { "epoch": 35.12, "learning_rate": 2.4706293166793244e-05, "loss": 0.0786, "step": 73170 }, { "epoch": 35.13, "learning_rate": 2.4693908648798458e-05, "loss": 0.078, "step": 73180 }, { "epoch": 35.13, "learning_rate": 2.468152621764856e-05, "loss": 0.0756, "step": 73190 }, { "epoch": 35.13, "learning_rate": 2.4669145874364658e-05, "loss": 0.0759, "step": 73200 }, { "epoch": 35.14, "learning_rate": 2.46567676199677e-05, "loss": 0.0835, "step": 73210 }, { "epoch": 35.14, "learning_rate": 2.4644391455478418e-05, "loss": 0.0768, "step": 73220 }, { "epoch": 35.15, "learning_rate": 2.4632017381917412e-05, "loss": 0.0778, "step": 73230 }, { "epoch": 35.15, "learning_rate": 2.4619645400305125e-05, "loss": 0.0821, "step": 73240 }, { "epoch": 35.16, "learning_rate": 2.4607275511661743e-05, "loss": 0.0804, "step": 73250 }, { "epoch": 35.16, "learning_rate": 2.4594907717007393e-05, "loss": 0.0784, "step": 73260 }, { "epoch": 35.17, "learning_rate": 2.458254201736196e-05, "loss": 0.0775, "step": 73270 }, { "epoch": 35.17, "learning_rate": 2.4570178413745187e-05, "loss": 0.073, "step": 73280 }, { "epoch": 35.18, "learning_rate": 2.455781690717659e-05, "loss": 0.0782, "step": 73290 }, { "epoch": 35.18, "learning_rate": 2.4545457498675567e-05, "loss": 0.0715, "step": 73300 }, { "epoch": 35.19, "learning_rate": 2.4533100189261327e-05, "loss": 0.0734, "step": 73310 }, { "epoch": 35.19, "learning_rate": 2.4520744979952926e-05, "loss": 0.076, "step": 73320 }, { "epoch": 35.19, "learning_rate": 2.4508391871769165e-05, "loss": 0.081, "step": 73330 }, { "epoch": 35.2, "learning_rate": 2.449604086572879e-05, "loss": 0.0737, "step": 73340 }, { "epoch": 35.2, "learning_rate": 2.4483691962850326e-05, "loss": 0.0805, "step": 73350 }, { "epoch": 35.21, "learning_rate": 2.4471345164152058e-05, "loss": 0.0822, "step": 73360 }, { "epoch": 35.21, "learning_rate": 2.4459000470652188e-05, "loss": 0.0765, "step": 73370 }, { "epoch": 35.22, "learning_rate": 2.4446657883368702e-05, "loss": 0.0797, "step": 73380 }, { "epoch": 35.22, "learning_rate": 2.4434317403319445e-05, "loss": 0.0733, "step": 73390 }, { "epoch": 35.23, "learning_rate": 2.4421979031522006e-05, "loss": 0.0745, "step": 73400 }, { "epoch": 35.23, "learning_rate": 2.4409642768993917e-05, "loss": 0.075, "step": 73410 }, { "epoch": 35.24, "learning_rate": 2.4397308616752475e-05, "loss": 0.077, "step": 73420 }, { "epoch": 35.24, "learning_rate": 2.438497657581476e-05, "loss": 0.0793, "step": 73430 }, { "epoch": 35.24, "learning_rate": 2.437264664719775e-05, "loss": 0.0705, "step": 73440 }, { "epoch": 35.25, "learning_rate": 2.4360318831918228e-05, "loss": 0.0781, "step": 73450 }, { "epoch": 35.25, "learning_rate": 2.4347993130992804e-05, "loss": 0.0751, "step": 73460 }, { "epoch": 35.26, "learning_rate": 2.4335669545437867e-05, "loss": 0.0892, "step": 73470 }, { "epoch": 35.26, "learning_rate": 2.4323348076269682e-05, "loss": 0.0775, "step": 73480 }, { "epoch": 35.27, "learning_rate": 2.4311028724504375e-05, "loss": 0.0735, "step": 73490 }, { "epoch": 35.27, "learning_rate": 2.4298711491157806e-05, "loss": 0.0879, "step": 73500 }, { "epoch": 35.28, "learning_rate": 2.4286396377245714e-05, "loss": 0.0899, "step": 73510 }, { "epoch": 35.28, "learning_rate": 2.427408338378365e-05, "loss": 0.0784, "step": 73520 }, { "epoch": 35.29, "learning_rate": 2.4261772511787023e-05, "loss": 0.0857, "step": 73530 }, { "epoch": 35.29, "learning_rate": 2.4249463762271e-05, "loss": 0.0837, "step": 73540 }, { "epoch": 35.29, "learning_rate": 2.4237157136250628e-05, "loss": 0.0728, "step": 73550 }, { "epoch": 35.3, "learning_rate": 2.4224852634740742e-05, "loss": 0.0814, "step": 73560 }, { "epoch": 35.3, "learning_rate": 2.421255025875608e-05, "loss": 0.0686, "step": 73570 }, { "epoch": 35.31, "learning_rate": 2.4200250009311083e-05, "loss": 0.0831, "step": 73580 }, { "epoch": 35.31, "learning_rate": 2.4187951887420114e-05, "loss": 0.076, "step": 73590 }, { "epoch": 35.32, "learning_rate": 2.4175655894097333e-05, "loss": 0.0787, "step": 73600 }, { "epoch": 35.32, "learning_rate": 2.416336203035669e-05, "loss": 0.078, "step": 73610 }, { "epoch": 35.33, "learning_rate": 2.4151070297211997e-05, "loss": 0.0721, "step": 73620 }, { "epoch": 35.33, "learning_rate": 2.413878069567689e-05, "loss": 0.0728, "step": 73630 }, { "epoch": 35.34, "learning_rate": 2.4126493226764812e-05, "loss": 0.0754, "step": 73640 }, { "epoch": 35.34, "learning_rate": 2.4114207891489047e-05, "loss": 0.0724, "step": 73650 }, { "epoch": 35.35, "learning_rate": 2.4101924690862694e-05, "loss": 0.087, "step": 73660 }, { "epoch": 35.35, "learning_rate": 2.4089643625898684e-05, "loss": 0.0762, "step": 73670 }, { "epoch": 35.35, "learning_rate": 2.4077364697609738e-05, "loss": 0.0842, "step": 73680 }, { "epoch": 35.36, "learning_rate": 2.4065087907008442e-05, "loss": 0.0766, "step": 73690 }, { "epoch": 35.36, "learning_rate": 2.4052813255107198e-05, "loss": 0.0751, "step": 73700 }, { "epoch": 35.37, "learning_rate": 2.404054074291821e-05, "loss": 0.0738, "step": 73710 }, { "epoch": 35.37, "learning_rate": 2.4028270371453532e-05, "loss": 0.0885, "step": 73720 }, { "epoch": 35.38, "learning_rate": 2.4016002141725025e-05, "loss": 0.0796, "step": 73730 }, { "epoch": 35.38, "learning_rate": 2.400373605474438e-05, "loss": 0.0778, "step": 73740 }, { "epoch": 35.39, "learning_rate": 2.399147211152313e-05, "loss": 0.085, "step": 73750 }, { "epoch": 35.39, "learning_rate": 2.397921031307256e-05, "loss": 0.0763, "step": 73760 }, { "epoch": 35.4, "learning_rate": 2.3966950660403863e-05, "loss": 0.0801, "step": 73770 }, { "epoch": 35.4, "learning_rate": 2.3954693154528012e-05, "loss": 0.0867, "step": 73780 }, { "epoch": 35.4, "learning_rate": 2.3942437796455818e-05, "loss": 0.0686, "step": 73790 }, { "epoch": 35.41, "learning_rate": 2.3930184587197903e-05, "loss": 0.0794, "step": 73800 }, { "epoch": 35.41, "learning_rate": 2.391793352776472e-05, "loss": 0.0749, "step": 73810 }, { "epoch": 35.42, "learning_rate": 2.3905684619166556e-05, "loss": 0.0717, "step": 73820 }, { "epoch": 35.42, "learning_rate": 2.3893437862413475e-05, "loss": 0.0775, "step": 73830 }, { "epoch": 35.43, "learning_rate": 2.388119325851541e-05, "loss": 0.0761, "step": 73840 }, { "epoch": 35.43, "learning_rate": 2.3868950808482104e-05, "loss": 0.0757, "step": 73850 }, { "epoch": 35.44, "learning_rate": 2.385671051332312e-05, "loss": 0.0726, "step": 73860 }, { "epoch": 35.44, "learning_rate": 2.3844472374047844e-05, "loss": 0.0865, "step": 73870 }, { "epoch": 35.45, "learning_rate": 2.383223639166548e-05, "loss": 0.0805, "step": 73880 }, { "epoch": 35.45, "learning_rate": 2.3820002567185084e-05, "loss": 0.0791, "step": 73890 }, { "epoch": 35.45, "learning_rate": 2.3807770901615463e-05, "loss": 0.0738, "step": 73900 }, { "epoch": 35.46, "learning_rate": 2.3795541395965306e-05, "loss": 0.0839, "step": 73910 }, { "epoch": 35.46, "learning_rate": 2.3783314051243116e-05, "loss": 0.0767, "step": 73920 }, { "epoch": 35.47, "learning_rate": 2.377108886845721e-05, "loss": 0.0788, "step": 73930 }, { "epoch": 35.47, "learning_rate": 2.3758865848615725e-05, "loss": 0.0809, "step": 73940 }, { "epoch": 35.48, "learning_rate": 2.3746644992726623e-05, "loss": 0.0777, "step": 73950 }, { "epoch": 35.48, "learning_rate": 2.3734426301797702e-05, "loss": 0.082, "step": 73960 }, { "epoch": 35.49, "learning_rate": 2.3722209776836534e-05, "loss": 0.0815, "step": 73970 }, { "epoch": 35.49, "learning_rate": 2.3709995418850557e-05, "loss": 0.074, "step": 73980 }, { "epoch": 35.5, "learning_rate": 2.369778322884702e-05, "loss": 0.0853, "step": 73990 }, { "epoch": 35.5, "learning_rate": 2.368557320783299e-05, "loss": 0.0793, "step": 74000 }, { "epoch": 35.51, "learning_rate": 2.3673365356815357e-05, "loss": 0.077, "step": 74010 }, { "epoch": 35.51, "learning_rate": 2.366115967680083e-05, "loss": 0.0754, "step": 74020 }, { "epoch": 35.51, "learning_rate": 2.364895616879596e-05, "loss": 0.078, "step": 74030 }, { "epoch": 35.52, "learning_rate": 2.363675483380706e-05, "loss": 0.0802, "step": 74040 }, { "epoch": 35.52, "learning_rate": 2.362455567284032e-05, "loss": 0.0789, "step": 74050 }, { "epoch": 35.53, "learning_rate": 2.3612358686901737e-05, "loss": 0.0707, "step": 74060 }, { "epoch": 35.53, "learning_rate": 2.3600163876997134e-05, "loss": 0.0758, "step": 74070 }, { "epoch": 35.54, "learning_rate": 2.35879712441321e-05, "loss": 0.0706, "step": 74080 }, { "epoch": 35.54, "learning_rate": 2.3575780789312145e-05, "loss": 0.0665, "step": 74090 }, { "epoch": 35.55, "learning_rate": 2.3563592513542544e-05, "loss": 0.0854, "step": 74100 }, { "epoch": 35.55, "learning_rate": 2.355140641782835e-05, "loss": 0.0787, "step": 74110 }, { "epoch": 35.56, "learning_rate": 2.3539222503174495e-05, "loss": 0.0762, "step": 74120 }, { "epoch": 35.56, "learning_rate": 2.3527040770585723e-05, "loss": 0.0787, "step": 74130 }, { "epoch": 35.56, "learning_rate": 2.3514861221066608e-05, "loss": 0.065, "step": 74140 }, { "epoch": 35.57, "learning_rate": 2.3502683855621487e-05, "loss": 0.0645, "step": 74150 }, { "epoch": 35.57, "learning_rate": 2.3490508675254554e-05, "loss": 0.0848, "step": 74160 }, { "epoch": 35.58, "learning_rate": 2.3478335680969865e-05, "loss": 0.0799, "step": 74170 }, { "epoch": 35.58, "learning_rate": 2.346616487377125e-05, "loss": 0.078, "step": 74180 }, { "epoch": 35.59, "learning_rate": 2.3453996254662334e-05, "loss": 0.0744, "step": 74190 }, { "epoch": 35.59, "learning_rate": 2.3441829824646604e-05, "loss": 0.0779, "step": 74200 }, { "epoch": 35.6, "learning_rate": 2.3429665584727384e-05, "loss": 0.0801, "step": 74210 }, { "epoch": 35.6, "learning_rate": 2.3417503535907736e-05, "loss": 0.0737, "step": 74220 }, { "epoch": 35.61, "learning_rate": 2.3405343679190616e-05, "loss": 0.0821, "step": 74230 }, { "epoch": 35.61, "learning_rate": 2.3393186015578762e-05, "loss": 0.0837, "step": 74240 }, { "epoch": 35.61, "learning_rate": 2.3381030546074796e-05, "loss": 0.0677, "step": 74250 }, { "epoch": 35.62, "learning_rate": 2.3368877271681057e-05, "loss": 0.0784, "step": 74260 }, { "epoch": 35.62, "learning_rate": 2.3356726193399764e-05, "loss": 0.0736, "step": 74270 }, { "epoch": 35.63, "learning_rate": 2.3344577312232974e-05, "loss": 0.0718, "step": 74280 }, { "epoch": 35.63, "learning_rate": 2.3332430629182493e-05, "loss": 0.0774, "step": 74290 }, { "epoch": 35.64, "learning_rate": 2.3320286145250002e-05, "loss": 0.0774, "step": 74300 }, { "epoch": 35.64, "learning_rate": 2.3308143861436992e-05, "loss": 0.0737, "step": 74310 }, { "epoch": 35.65, "learning_rate": 2.329600377874476e-05, "loss": 0.0749, "step": 74320 }, { "epoch": 35.65, "learning_rate": 2.3283865898174435e-05, "loss": 0.0771, "step": 74330 }, { "epoch": 35.66, "learning_rate": 2.3271730220726945e-05, "loss": 0.0702, "step": 74340 }, { "epoch": 35.66, "learning_rate": 2.3259596747403057e-05, "loss": 0.0785, "step": 74350 }, { "epoch": 35.66, "learning_rate": 2.3247465479203362e-05, "loss": 0.0744, "step": 74360 }, { "epoch": 35.67, "learning_rate": 2.3235336417128222e-05, "loss": 0.0746, "step": 74370 }, { "epoch": 35.67, "learning_rate": 2.322320956217786e-05, "loss": 0.0841, "step": 74380 }, { "epoch": 35.68, "learning_rate": 2.3211084915352316e-05, "loss": 0.0687, "step": 74390 }, { "epoch": 35.68, "learning_rate": 2.3198962477651432e-05, "loss": 0.0752, "step": 74400 }, { "epoch": 35.69, "learning_rate": 2.3186842250074875e-05, "loss": 0.0783, "step": 74410 }, { "epoch": 35.69, "learning_rate": 2.3174724233622135e-05, "loss": 0.0694, "step": 74420 }, { "epoch": 35.7, "learning_rate": 2.3162608429292526e-05, "loss": 0.0742, "step": 74430 }, { "epoch": 35.7, "learning_rate": 2.315049483808513e-05, "loss": 0.0793, "step": 74440 }, { "epoch": 35.71, "learning_rate": 2.3138383460998903e-05, "loss": 0.0793, "step": 74450 }, { "epoch": 35.71, "learning_rate": 2.3126274299032603e-05, "loss": 0.086, "step": 74460 }, { "epoch": 35.72, "learning_rate": 2.3114167353184797e-05, "loss": 0.0818, "step": 74470 }, { "epoch": 35.72, "learning_rate": 2.310206262445388e-05, "loss": 0.0807, "step": 74480 }, { "epoch": 35.72, "learning_rate": 2.3089960113838048e-05, "loss": 0.0785, "step": 74490 }, { "epoch": 35.73, "learning_rate": 2.3077859822335353e-05, "loss": 0.0726, "step": 74500 }, { "epoch": 35.73, "learning_rate": 2.3065761750943594e-05, "loss": 0.0717, "step": 74510 }, { "epoch": 35.74, "learning_rate": 2.3053665900660443e-05, "loss": 0.0805, "step": 74520 }, { "epoch": 35.74, "learning_rate": 2.3041572272483374e-05, "loss": 0.0819, "step": 74530 }, { "epoch": 35.75, "learning_rate": 2.3029480867409686e-05, "loss": 0.0797, "step": 74540 }, { "epoch": 35.75, "learning_rate": 2.301739168643648e-05, "loss": 0.0844, "step": 74550 }, { "epoch": 35.76, "learning_rate": 2.300530473056067e-05, "loss": 0.0869, "step": 74560 }, { "epoch": 35.76, "learning_rate": 2.2993220000779037e-05, "loss": 0.0826, "step": 74570 }, { "epoch": 35.77, "learning_rate": 2.2981137498088086e-05, "loss": 0.0793, "step": 74580 }, { "epoch": 35.77, "learning_rate": 2.2969057223484208e-05, "loss": 0.0736, "step": 74590 }, { "epoch": 35.77, "learning_rate": 2.29569791779636e-05, "loss": 0.0755, "step": 74600 }, { "epoch": 35.78, "learning_rate": 2.2944903362522264e-05, "loss": 0.0762, "step": 74610 }, { "epoch": 35.78, "learning_rate": 2.2932829778156017e-05, "loss": 0.0732, "step": 74620 }, { "epoch": 35.79, "learning_rate": 2.2920758425860506e-05, "loss": 0.0821, "step": 74630 }, { "epoch": 35.79, "learning_rate": 2.2908689306631192e-05, "loss": 0.0654, "step": 74640 }, { "epoch": 35.8, "learning_rate": 2.2896622421463314e-05, "loss": 0.0763, "step": 74650 }, { "epoch": 35.8, "learning_rate": 2.2884557771351977e-05, "loss": 0.0811, "step": 74660 }, { "epoch": 35.81, "learning_rate": 2.2872495357292085e-05, "loss": 0.0787, "step": 74670 }, { "epoch": 35.81, "learning_rate": 2.2860435180278343e-05, "loss": 0.0778, "step": 74680 }, { "epoch": 35.82, "learning_rate": 2.284837724130529e-05, "loss": 0.0678, "step": 74690 }, { "epoch": 35.82, "learning_rate": 2.2836321541367272e-05, "loss": 0.0788, "step": 74700 }, { "epoch": 35.82, "learning_rate": 2.2824268081458454e-05, "loss": 0.0741, "step": 74710 }, { "epoch": 35.83, "learning_rate": 2.2812216862572833e-05, "loss": 0.0776, "step": 74720 }, { "epoch": 35.83, "learning_rate": 2.2800167885704164e-05, "loss": 0.0864, "step": 74730 }, { "epoch": 35.84, "learning_rate": 2.2788121151846076e-05, "loss": 0.0853, "step": 74740 }, { "epoch": 35.84, "learning_rate": 2.2776076661992e-05, "loss": 0.0788, "step": 74750 }, { "epoch": 35.85, "learning_rate": 2.2764034417135138e-05, "loss": 0.0844, "step": 74760 }, { "epoch": 35.85, "learning_rate": 2.2751994418268578e-05, "loss": 0.0783, "step": 74770 }, { "epoch": 35.86, "learning_rate": 2.2739956666385186e-05, "loss": 0.0769, "step": 74780 }, { "epoch": 35.86, "learning_rate": 2.272792116247766e-05, "loss": 0.0722, "step": 74790 }, { "epoch": 35.87, "learning_rate": 2.2715887907538454e-05, "loss": 0.0778, "step": 74800 }, { "epoch": 35.87, "learning_rate": 2.27038569025599e-05, "loss": 0.076, "step": 74810 }, { "epoch": 35.88, "learning_rate": 2.269182814853415e-05, "loss": 0.0817, "step": 74820 }, { "epoch": 35.88, "learning_rate": 2.2679801646453098e-05, "loss": 0.0765, "step": 74830 }, { "epoch": 35.88, "learning_rate": 2.266777739730851e-05, "loss": 0.0872, "step": 74840 }, { "epoch": 35.89, "learning_rate": 2.2655755402091976e-05, "loss": 0.0725, "step": 74850 }, { "epoch": 35.89, "learning_rate": 2.2643735661794896e-05, "loss": 0.0701, "step": 74860 }, { "epoch": 35.9, "learning_rate": 2.2631718177408423e-05, "loss": 0.0744, "step": 74870 }, { "epoch": 35.9, "learning_rate": 2.2619702949923593e-05, "loss": 0.0737, "step": 74880 }, { "epoch": 35.91, "learning_rate": 2.2607689980331237e-05, "loss": 0.0782, "step": 74890 }, { "epoch": 35.91, "learning_rate": 2.2595679269621968e-05, "loss": 0.0707, "step": 74900 }, { "epoch": 35.92, "learning_rate": 2.2583670818786255e-05, "loss": 0.0818, "step": 74910 }, { "epoch": 35.92, "learning_rate": 2.2571664628814344e-05, "loss": 0.0662, "step": 74920 }, { "epoch": 35.93, "learning_rate": 2.255966070069637e-05, "loss": 0.0708, "step": 74930 }, { "epoch": 35.93, "learning_rate": 2.2547659035422175e-05, "loss": 0.0711, "step": 74940 }, { "epoch": 35.93, "learning_rate": 2.253565963398148e-05, "loss": 0.0723, "step": 74950 }, { "epoch": 35.94, "learning_rate": 2.2523662497363813e-05, "loss": 0.0822, "step": 74960 }, { "epoch": 35.94, "learning_rate": 2.2511667626558517e-05, "loss": 0.0753, "step": 74970 }, { "epoch": 35.95, "learning_rate": 2.2499675022554708e-05, "loss": 0.0803, "step": 74980 }, { "epoch": 36.0, "learning_rate": 2.2487684686341352e-05, "loss": 0.0902, "step": 74990 }, { "epoch": 36.01, "learning_rate": 2.2475696618907237e-05, "loss": 0.071, "step": 75000 }, { "epoch": 36.01, "learning_rate": 2.2463710821240947e-05, "loss": 0.0754, "step": 75010 }, { "epoch": 36.01, "learning_rate": 2.2451727294330872e-05, "loss": 0.0704, "step": 75020 }, { "epoch": 36.02, "learning_rate": 2.2439746039165222e-05, "loss": 0.079, "step": 75030 }, { "epoch": 36.02, "learning_rate": 2.2427767056732053e-05, "loss": 0.0774, "step": 75040 }, { "epoch": 36.03, "learning_rate": 2.241579034801915e-05, "loss": 0.0708, "step": 75050 }, { "epoch": 36.03, "learning_rate": 2.2403815914014187e-05, "loss": 0.0728, "step": 75060 }, { "epoch": 36.04, "learning_rate": 2.2391843755704624e-05, "loss": 0.0739, "step": 75070 }, { "epoch": 36.04, "learning_rate": 2.237987387407774e-05, "loss": 0.0797, "step": 75080 }, { "epoch": 36.05, "learning_rate": 2.236790627012061e-05, "loss": 0.0775, "step": 75090 }, { "epoch": 36.05, "learning_rate": 2.2355940944820143e-05, "loss": 0.0662, "step": 75100 }, { "epoch": 36.06, "learning_rate": 2.234397789916306e-05, "loss": 0.0756, "step": 75110 }, { "epoch": 36.06, "learning_rate": 2.2332017134135845e-05, "loss": 0.0768, "step": 75120 }, { "epoch": 36.06, "learning_rate": 2.232005865072486e-05, "loss": 0.073, "step": 75130 }, { "epoch": 36.07, "learning_rate": 2.2308102449916247e-05, "loss": 0.0807, "step": 75140 }, { "epoch": 36.07, "learning_rate": 2.229614853269596e-05, "loss": 0.0643, "step": 75150 }, { "epoch": 36.08, "learning_rate": 2.228419690004977e-05, "loss": 0.0778, "step": 75160 }, { "epoch": 36.08, "learning_rate": 2.227224755296326e-05, "loss": 0.0761, "step": 75170 }, { "epoch": 36.09, "learning_rate": 2.2260300492421836e-05, "loss": 0.0783, "step": 75180 }, { "epoch": 36.09, "learning_rate": 2.2248355719410675e-05, "loss": 0.0839, "step": 75190 }, { "epoch": 36.1, "learning_rate": 2.2236413234914805e-05, "loss": 0.0805, "step": 75200 }, { "epoch": 36.1, "learning_rate": 2.2224473039919047e-05, "loss": 0.0738, "step": 75210 }, { "epoch": 36.11, "learning_rate": 2.2212535135408046e-05, "loss": 0.0809, "step": 75220 }, { "epoch": 36.11, "learning_rate": 2.220059952236625e-05, "loss": 0.0817, "step": 75230 }, { "epoch": 36.12, "learning_rate": 2.218866620177792e-05, "loss": 0.0863, "step": 75240 }, { "epoch": 36.12, "learning_rate": 2.2176735174627138e-05, "loss": 0.0846, "step": 75250 }, { "epoch": 36.12, "learning_rate": 2.216480644189776e-05, "loss": 0.0784, "step": 75260 }, { "epoch": 36.13, "learning_rate": 2.215288000457349e-05, "loss": 0.0728, "step": 75270 }, { "epoch": 36.13, "learning_rate": 2.214095586363783e-05, "loss": 0.0779, "step": 75280 }, { "epoch": 36.14, "learning_rate": 2.2129034020074102e-05, "loss": 0.0726, "step": 75290 }, { "epoch": 36.14, "learning_rate": 2.211711447486542e-05, "loss": 0.079, "step": 75300 }, { "epoch": 36.15, "learning_rate": 2.210519722899473e-05, "loss": 0.0803, "step": 75310 }, { "epoch": 36.15, "learning_rate": 2.2093282283444768e-05, "loss": 0.0779, "step": 75320 }, { "epoch": 36.16, "learning_rate": 2.2081369639198106e-05, "loss": 0.0805, "step": 75330 }, { "epoch": 36.16, "learning_rate": 2.2069459297237082e-05, "loss": 0.0745, "step": 75340 }, { "epoch": 36.17, "learning_rate": 2.205755125854389e-05, "loss": 0.0726, "step": 75350 }, { "epoch": 36.17, "learning_rate": 2.2045645524100516e-05, "loss": 0.0928, "step": 75360 }, { "epoch": 36.17, "learning_rate": 2.2033742094888748e-05, "loss": 0.0818, "step": 75370 }, { "epoch": 36.18, "learning_rate": 2.20218409718902e-05, "loss": 0.0815, "step": 75380 }, { "epoch": 36.18, "learning_rate": 2.2009942156086284e-05, "loss": 0.0732, "step": 75390 }, { "epoch": 36.19, "learning_rate": 2.1998045648458245e-05, "loss": 0.0775, "step": 75400 }, { "epoch": 36.19, "learning_rate": 2.198615144998708e-05, "loss": 0.0803, "step": 75410 }, { "epoch": 36.2, "learning_rate": 2.1974259561653654e-05, "loss": 0.0678, "step": 75420 }, { "epoch": 36.2, "learning_rate": 2.1962369984438642e-05, "loss": 0.0735, "step": 75430 }, { "epoch": 36.21, "learning_rate": 2.1950482719322446e-05, "loss": 0.0694, "step": 75440 }, { "epoch": 36.21, "learning_rate": 2.1938597767285397e-05, "loss": 0.0688, "step": 75450 }, { "epoch": 36.22, "learning_rate": 2.192671512930756e-05, "loss": 0.0764, "step": 75460 }, { "epoch": 36.22, "learning_rate": 2.191483480636885e-05, "loss": 0.0687, "step": 75470 }, { "epoch": 36.22, "learning_rate": 2.190295679944892e-05, "loss": 0.0707, "step": 75480 }, { "epoch": 36.23, "learning_rate": 2.1891081109527305e-05, "loss": 0.0809, "step": 75490 }, { "epoch": 36.23, "learning_rate": 2.187920773758334e-05, "loss": 0.0838, "step": 75500 }, { "epoch": 36.24, "learning_rate": 2.186733668459612e-05, "loss": 0.0689, "step": 75510 }, { "epoch": 36.24, "learning_rate": 2.1855467951544572e-05, "loss": 0.0813, "step": 75520 }, { "epoch": 36.25, "learning_rate": 2.1843601539407482e-05, "loss": 0.0703, "step": 75530 }, { "epoch": 36.25, "learning_rate": 2.1831737449163402e-05, "loss": 0.0609, "step": 75540 }, { "epoch": 36.26, "learning_rate": 2.1819875681790662e-05, "loss": 0.0761, "step": 75550 }, { "epoch": 36.26, "learning_rate": 2.1808016238267446e-05, "loss": 0.0721, "step": 75560 }, { "epoch": 36.27, "learning_rate": 2.1796159119571742e-05, "loss": 0.0865, "step": 75570 }, { "epoch": 36.27, "learning_rate": 2.1784304326681338e-05, "loss": 0.0749, "step": 75580 }, { "epoch": 36.28, "learning_rate": 2.1772451860573807e-05, "loss": 0.0714, "step": 75590 }, { "epoch": 36.28, "learning_rate": 2.176060172222654e-05, "loss": 0.0764, "step": 75600 }, { "epoch": 36.28, "learning_rate": 2.174875391261682e-05, "loss": 0.079, "step": 75610 }, { "epoch": 36.29, "learning_rate": 2.1736908432721598e-05, "loss": 0.0742, "step": 75620 }, { "epoch": 36.29, "learning_rate": 2.1725065283517727e-05, "loss": 0.0747, "step": 75630 }, { "epoch": 36.3, "learning_rate": 2.1713224465981837e-05, "loss": 0.0811, "step": 75640 }, { "epoch": 36.3, "learning_rate": 2.1701385981090393e-05, "loss": 0.0807, "step": 75650 }, { "epoch": 36.31, "learning_rate": 2.1689549829819605e-05, "loss": 0.0703, "step": 75660 }, { "epoch": 36.31, "learning_rate": 2.1677716013145532e-05, "loss": 0.0818, "step": 75670 }, { "epoch": 36.32, "learning_rate": 2.1665884532044096e-05, "loss": 0.0797, "step": 75680 }, { "epoch": 36.32, "learning_rate": 2.1654055387490912e-05, "loss": 0.0781, "step": 75690 }, { "epoch": 36.33, "learning_rate": 2.164222858046148e-05, "loss": 0.0751, "step": 75700 }, { "epoch": 36.33, "learning_rate": 2.163040411193109e-05, "loss": 0.0739, "step": 75710 }, { "epoch": 36.33, "learning_rate": 2.161858198287485e-05, "loss": 0.0799, "step": 75720 }, { "epoch": 36.34, "learning_rate": 2.1606762194267637e-05, "loss": 0.0735, "step": 75730 }, { "epoch": 36.34, "learning_rate": 2.159494474708416e-05, "loss": 0.0778, "step": 75740 }, { "epoch": 36.35, "learning_rate": 2.1583129642298927e-05, "loss": 0.0802, "step": 75750 }, { "epoch": 36.35, "learning_rate": 2.157131688088631e-05, "loss": 0.0714, "step": 75760 }, { "epoch": 36.36, "learning_rate": 2.1559506463820394e-05, "loss": 0.0736, "step": 75770 }, { "epoch": 36.36, "learning_rate": 2.154769839207512e-05, "loss": 0.0783, "step": 75780 }, { "epoch": 36.37, "learning_rate": 2.1535892666624253e-05, "loss": 0.0726, "step": 75790 }, { "epoch": 36.37, "learning_rate": 2.152408928844131e-05, "loss": 0.0803, "step": 75800 }, { "epoch": 36.38, "learning_rate": 2.1512288258499657e-05, "loss": 0.0808, "step": 75810 }, { "epoch": 36.38, "learning_rate": 2.1500489577772464e-05, "loss": 0.0766, "step": 75820 }, { "epoch": 36.38, "learning_rate": 2.1488693247232696e-05, "loss": 0.0706, "step": 75830 }, { "epoch": 36.39, "learning_rate": 2.1476899267853124e-05, "loss": 0.0786, "step": 75840 }, { "epoch": 36.39, "learning_rate": 2.146510764060633e-05, "loss": 0.0825, "step": 75850 }, { "epoch": 36.4, "learning_rate": 2.1453318366464714e-05, "loss": 0.075, "step": 75860 }, { "epoch": 36.4, "learning_rate": 2.1441531446400444e-05, "loss": 0.0751, "step": 75870 }, { "epoch": 36.41, "learning_rate": 2.1429746881385525e-05, "loss": 0.0792, "step": 75880 }, { "epoch": 36.41, "learning_rate": 2.1417964672391767e-05, "loss": 0.075, "step": 75890 }, { "epoch": 36.42, "learning_rate": 2.1406184820390774e-05, "loss": 0.0789, "step": 75900 }, { "epoch": 36.42, "learning_rate": 2.139440732635397e-05, "loss": 0.0709, "step": 75910 }, { "epoch": 36.43, "learning_rate": 2.1382632191252567e-05, "loss": 0.0691, "step": 75920 }, { "epoch": 36.43, "learning_rate": 2.1370859416057597e-05, "loss": 0.072, "step": 75930 }, { "epoch": 36.44, "learning_rate": 2.1359089001739907e-05, "loss": 0.0826, "step": 75940 }, { "epoch": 36.44, "learning_rate": 2.13473209492701e-05, "loss": 0.0857, "step": 75950 }, { "epoch": 36.44, "learning_rate": 2.1335555259618634e-05, "loss": 0.0656, "step": 75960 }, { "epoch": 36.45, "learning_rate": 2.132379193375576e-05, "loss": 0.071, "step": 75970 }, { "epoch": 36.45, "learning_rate": 2.1312030972651525e-05, "loss": 0.0765, "step": 75980 }, { "epoch": 36.46, "learning_rate": 2.1300272377275786e-05, "loss": 0.0777, "step": 75990 }, { "epoch": 36.46, "learning_rate": 2.1288516148598214e-05, "loss": 0.0821, "step": 76000 }, { "epoch": 36.47, "learning_rate": 2.1276762287588288e-05, "loss": 0.0717, "step": 76010 }, { "epoch": 36.47, "learning_rate": 2.1265010795215247e-05, "loss": 0.0724, "step": 76020 }, { "epoch": 36.48, "learning_rate": 2.1253261672448176e-05, "loss": 0.0701, "step": 76030 }, { "epoch": 36.48, "learning_rate": 2.1241514920255972e-05, "loss": 0.0625, "step": 76040 }, { "epoch": 36.49, "learning_rate": 2.122977053960731e-05, "loss": 0.0706, "step": 76050 }, { "epoch": 36.49, "learning_rate": 2.121802853147068e-05, "loss": 0.0673, "step": 76060 }, { "epoch": 36.49, "learning_rate": 2.120628889681438e-05, "loss": 0.0786, "step": 76070 }, { "epoch": 36.5, "learning_rate": 2.1194551636606525e-05, "loss": 0.0689, "step": 76080 }, { "epoch": 36.5, "learning_rate": 2.118281675181499e-05, "loss": 0.0779, "step": 76090 }, { "epoch": 36.51, "learning_rate": 2.1171084243407485e-05, "loss": 0.0691, "step": 76100 }, { "epoch": 36.51, "learning_rate": 2.1159354112351548e-05, "loss": 0.0809, "step": 76110 }, { "epoch": 36.52, "learning_rate": 2.1147626359614443e-05, "loss": 0.069, "step": 76120 }, { "epoch": 36.52, "learning_rate": 2.113590098616334e-05, "loss": 0.0778, "step": 76130 }, { "epoch": 36.53, "learning_rate": 2.1124177992965137e-05, "loss": 0.0691, "step": 76140 }, { "epoch": 36.53, "learning_rate": 2.1112457380986587e-05, "loss": 0.0742, "step": 76150 }, { "epoch": 36.54, "learning_rate": 2.1100739151194188e-05, "loss": 0.0787, "step": 76160 }, { "epoch": 36.54, "learning_rate": 2.1089023304554278e-05, "loss": 0.0778, "step": 76170 }, { "epoch": 36.54, "learning_rate": 2.1077309842033003e-05, "loss": 0.0717, "step": 76180 }, { "epoch": 36.55, "learning_rate": 2.106559876459632e-05, "loss": 0.0747, "step": 76190 }, { "epoch": 36.55, "learning_rate": 2.105389007320992e-05, "loss": 0.0764, "step": 76200 }, { "epoch": 36.56, "learning_rate": 2.1042183768839402e-05, "loss": 0.0821, "step": 76210 }, { "epoch": 36.56, "learning_rate": 2.103047985245012e-05, "loss": 0.073, "step": 76220 }, { "epoch": 36.57, "learning_rate": 2.101877832500718e-05, "loss": 0.0852, "step": 76230 }, { "epoch": 36.57, "learning_rate": 2.100707918747557e-05, "loss": 0.0745, "step": 76240 }, { "epoch": 36.58, "learning_rate": 2.0995382440820045e-05, "loss": 0.0655, "step": 76250 }, { "epoch": 36.58, "learning_rate": 2.0983688086005182e-05, "loss": 0.0777, "step": 76260 }, { "epoch": 36.59, "learning_rate": 2.0971996123995297e-05, "loss": 0.0758, "step": 76270 }, { "epoch": 36.59, "learning_rate": 2.0960306555754606e-05, "loss": 0.0774, "step": 76280 }, { "epoch": 36.6, "learning_rate": 2.094861938224708e-05, "loss": 0.0841, "step": 76290 }, { "epoch": 36.6, "learning_rate": 2.0936934604436466e-05, "loss": 0.0708, "step": 76300 }, { "epoch": 36.6, "learning_rate": 2.092525222328634e-05, "loss": 0.0786, "step": 76310 }, { "epoch": 36.61, "learning_rate": 2.0913572239760088e-05, "loss": 0.0784, "step": 76320 }, { "epoch": 36.61, "learning_rate": 2.0901894654820908e-05, "loss": 0.0696, "step": 76330 }, { "epoch": 36.62, "learning_rate": 2.089021946943175e-05, "loss": 0.0839, "step": 76340 }, { "epoch": 36.62, "learning_rate": 2.0878546684555385e-05, "loss": 0.0708, "step": 76350 }, { "epoch": 36.63, "learning_rate": 2.086687630115445e-05, "loss": 0.0782, "step": 76360 }, { "epoch": 36.63, "learning_rate": 2.085520832019132e-05, "loss": 0.0731, "step": 76370 }, { "epoch": 36.64, "learning_rate": 2.0843542742628164e-05, "loss": 0.0774, "step": 76380 }, { "epoch": 36.64, "learning_rate": 2.0831879569426978e-05, "loss": 0.0754, "step": 76390 }, { "epoch": 36.65, "learning_rate": 2.082021880154958e-05, "loss": 0.0705, "step": 76400 }, { "epoch": 36.65, "learning_rate": 2.0808560439957525e-05, "loss": 0.0713, "step": 76410 }, { "epoch": 36.65, "learning_rate": 2.0796904485612234e-05, "loss": 0.0682, "step": 76420 }, { "epoch": 36.66, "learning_rate": 2.0785250939474883e-05, "loss": 0.0872, "step": 76430 }, { "epoch": 36.66, "learning_rate": 2.077359980250652e-05, "loss": 0.0848, "step": 76440 }, { "epoch": 36.67, "learning_rate": 2.0761951075667908e-05, "loss": 0.077, "step": 76450 }, { "epoch": 36.67, "learning_rate": 2.0750304759919652e-05, "loss": 0.0736, "step": 76460 }, { "epoch": 36.68, "learning_rate": 2.073866085622218e-05, "loss": 0.0781, "step": 76470 }, { "epoch": 36.68, "learning_rate": 2.0727019365535654e-05, "loss": 0.0744, "step": 76480 }, { "epoch": 36.69, "learning_rate": 2.071538028882011e-05, "loss": 0.0809, "step": 76490 }, { "epoch": 36.69, "learning_rate": 2.0703743627035344e-05, "loss": 0.0801, "step": 76500 }, { "epoch": 36.7, "learning_rate": 2.0692109381140963e-05, "loss": 0.0749, "step": 76510 }, { "epoch": 36.7, "learning_rate": 2.0680477552096383e-05, "loss": 0.0727, "step": 76520 }, { "epoch": 36.7, "learning_rate": 2.0668848140860802e-05, "loss": 0.0732, "step": 76530 }, { "epoch": 36.71, "learning_rate": 2.065722114839324e-05, "loss": 0.08, "step": 76540 }, { "epoch": 36.71, "learning_rate": 2.0645596575652522e-05, "loss": 0.0683, "step": 76550 }, { "epoch": 36.72, "learning_rate": 2.0633974423597214e-05, "loss": 0.0774, "step": 76560 }, { "epoch": 36.72, "learning_rate": 2.0622354693185752e-05, "loss": 0.0742, "step": 76570 }, { "epoch": 36.73, "learning_rate": 2.061073738537635e-05, "loss": 0.0754, "step": 76580 }, { "epoch": 36.73, "learning_rate": 2.059912250112701e-05, "loss": 0.0742, "step": 76590 }, { "epoch": 36.74, "learning_rate": 2.0587510041395554e-05, "loss": 0.0768, "step": 76600 }, { "epoch": 36.74, "learning_rate": 2.057590000713958e-05, "loss": 0.0846, "step": 76610 }, { "epoch": 36.75, "learning_rate": 2.0564292399316533e-05, "loss": 0.0754, "step": 76620 }, { "epoch": 36.75, "learning_rate": 2.0552687218883575e-05, "loss": 0.0814, "step": 76630 }, { "epoch": 36.76, "learning_rate": 2.0541084466797745e-05, "loss": 0.0734, "step": 76640 }, { "epoch": 36.76, "learning_rate": 2.052948414401585e-05, "loss": 0.08, "step": 76650 }, { "epoch": 36.76, "learning_rate": 2.05178862514945e-05, "loss": 0.0707, "step": 76660 }, { "epoch": 36.77, "learning_rate": 2.05062907901901e-05, "loss": 0.0884, "step": 76670 }, { "epoch": 36.77, "learning_rate": 2.0494697761058873e-05, "loss": 0.0743, "step": 76680 }, { "epoch": 36.78, "learning_rate": 2.0483107165056838e-05, "loss": 0.0791, "step": 76690 }, { "epoch": 36.78, "learning_rate": 2.0471519003139762e-05, "loss": 0.0763, "step": 76700 }, { "epoch": 36.79, "learning_rate": 2.0459933276263276e-05, "loss": 0.0865, "step": 76710 }, { "epoch": 36.79, "learning_rate": 2.0448349985382792e-05, "loss": 0.0784, "step": 76720 }, { "epoch": 36.8, "learning_rate": 2.043676913145351e-05, "loss": 0.0884, "step": 76730 }, { "epoch": 36.8, "learning_rate": 2.0425190715430433e-05, "loss": 0.0807, "step": 76740 }, { "epoch": 36.81, "learning_rate": 2.0413614738268373e-05, "loss": 0.079, "step": 76750 }, { "epoch": 36.81, "learning_rate": 2.0402041200921945e-05, "loss": 0.0715, "step": 76760 }, { "epoch": 36.81, "learning_rate": 2.0390470104345516e-05, "loss": 0.0807, "step": 76770 }, { "epoch": 36.82, "learning_rate": 2.0378901449493306e-05, "loss": 0.0778, "step": 76780 }, { "epoch": 36.82, "learning_rate": 2.0367335237319306e-05, "loss": 0.0778, "step": 76790 }, { "epoch": 36.83, "learning_rate": 2.0355771468777323e-05, "loss": 0.0817, "step": 76800 }, { "epoch": 36.83, "learning_rate": 2.0344210144820958e-05, "loss": 0.0684, "step": 76810 }, { "epoch": 36.84, "learning_rate": 2.033265126640359e-05, "loss": 0.0844, "step": 76820 }, { "epoch": 36.84, "learning_rate": 2.032109483447844e-05, "loss": 0.0767, "step": 76830 }, { "epoch": 36.85, "learning_rate": 2.030954084999846e-05, "loss": 0.0692, "step": 76840 }, { "epoch": 36.85, "learning_rate": 2.029798931391646e-05, "loss": 0.0766, "step": 76850 }, { "epoch": 36.86, "learning_rate": 2.0286440227185026e-05, "loss": 0.0647, "step": 76860 }, { "epoch": 36.86, "learning_rate": 2.0274893590756545e-05, "loss": 0.0746, "step": 76870 }, { "epoch": 36.86, "learning_rate": 2.0263349405583194e-05, "loss": 0.0718, "step": 76880 }, { "epoch": 36.87, "learning_rate": 2.025180767261696e-05, "loss": 0.0671, "step": 76890 }, { "epoch": 36.87, "learning_rate": 2.024026839280964e-05, "loss": 0.0746, "step": 76900 }, { "epoch": 36.88, "learning_rate": 2.0228731567112773e-05, "loss": 0.0757, "step": 76910 }, { "epoch": 36.88, "learning_rate": 2.0217197196477754e-05, "loss": 0.0771, "step": 76920 }, { "epoch": 36.89, "learning_rate": 2.0205665281855752e-05, "loss": 0.0794, "step": 76930 }, { "epoch": 36.89, "learning_rate": 2.019413582419775e-05, "loss": 0.0878, "step": 76940 }, { "epoch": 36.9, "learning_rate": 2.0182608824454475e-05, "loss": 0.0848, "step": 76950 }, { "epoch": 36.9, "learning_rate": 2.017108428357653e-05, "loss": 0.0687, "step": 76960 }, { "epoch": 36.91, "learning_rate": 2.0159562202514263e-05, "loss": 0.0877, "step": 76970 }, { "epoch": 36.91, "learning_rate": 2.0148042582217857e-05, "loss": 0.08, "step": 76980 }, { "epoch": 36.91, "learning_rate": 2.013652542363722e-05, "loss": 0.0836, "step": 76990 }, { "epoch": 36.92, "learning_rate": 2.012501072772213e-05, "loss": 0.0719, "step": 77000 }, { "epoch": 36.92, "learning_rate": 2.0113498495422155e-05, "loss": 0.0816, "step": 77010 }, { "epoch": 36.93, "learning_rate": 2.01019887276866e-05, "loss": 0.0727, "step": 77020 }, { "epoch": 36.93, "learning_rate": 2.0090481425464614e-05, "loss": 0.0816, "step": 77030 }, { "epoch": 36.94, "learning_rate": 2.0078976589705174e-05, "loss": 0.081, "step": 77040 }, { "epoch": 36.94, "learning_rate": 2.0067474221357003e-05, "loss": 0.074, "step": 77050 }, { "epoch": 36.95, "learning_rate": 2.0055974321368615e-05, "loss": 0.0783, "step": 77060 }, { "epoch": 36.95, "learning_rate": 2.004447689068834e-05, "loss": 0.0944, "step": 77070 }, { "epoch": 37.0, "learning_rate": 2.003298193026434e-05, "loss": 0.0781, "step": 77080 }, { "epoch": 37.01, "learning_rate": 2.002148944104449e-05, "loss": 0.0805, "step": 77090 }, { "epoch": 37.01, "learning_rate": 2.0009999423976528e-05, "loss": 0.0786, "step": 77100 }, { "epoch": 37.02, "learning_rate": 1.999851188000795e-05, "loss": 0.0789, "step": 77110 }, { "epoch": 37.02, "learning_rate": 1.9987026810086117e-05, "loss": 0.0778, "step": 77120 }, { "epoch": 37.03, "learning_rate": 1.9975544215158087e-05, "loss": 0.0719, "step": 77130 }, { "epoch": 37.03, "learning_rate": 1.996406409617078e-05, "loss": 0.0771, "step": 77140 }, { "epoch": 37.04, "learning_rate": 1.9952586454070895e-05, "loss": 0.0729, "step": 77150 }, { "epoch": 37.04, "learning_rate": 1.9941111289804938e-05, "loss": 0.0666, "step": 77160 }, { "epoch": 37.05, "learning_rate": 1.9929638604319167e-05, "loss": 0.0763, "step": 77170 }, { "epoch": 37.05, "learning_rate": 1.991816839855968e-05, "loss": 0.0809, "step": 77180 }, { "epoch": 37.05, "learning_rate": 1.9906700673472366e-05, "loss": 0.0777, "step": 77190 }, { "epoch": 37.06, "learning_rate": 1.9895235430002894e-05, "loss": 0.0807, "step": 77200 }, { "epoch": 37.06, "learning_rate": 1.988377266909674e-05, "loss": 0.0827, "step": 77210 }, { "epoch": 37.07, "learning_rate": 1.9872312391699167e-05, "loss": 0.0776, "step": 77220 }, { "epoch": 37.07, "learning_rate": 1.9860854598755253e-05, "loss": 0.0755, "step": 77230 }, { "epoch": 37.08, "learning_rate": 1.9849399291209826e-05, "loss": 0.0723, "step": 77240 }, { "epoch": 37.08, "learning_rate": 1.9837946470007552e-05, "loss": 0.0655, "step": 77250 }, { "epoch": 37.09, "learning_rate": 1.9826496136092875e-05, "loss": 0.0752, "step": 77260 }, { "epoch": 37.09, "learning_rate": 1.9815048290410044e-05, "loss": 0.0793, "step": 77270 }, { "epoch": 37.1, "learning_rate": 1.9803602933903088e-05, "loss": 0.0807, "step": 77280 }, { "epoch": 37.1, "learning_rate": 1.979216006751584e-05, "loss": 0.0678, "step": 77290 }, { "epoch": 37.1, "learning_rate": 1.978071969219195e-05, "loss": 0.0825, "step": 77300 }, { "epoch": 37.11, "learning_rate": 1.9769281808874803e-05, "loss": 0.0863, "step": 77310 }, { "epoch": 37.11, "learning_rate": 1.9757846418507624e-05, "loss": 0.0802, "step": 77320 }, { "epoch": 37.12, "learning_rate": 1.9746413522033426e-05, "loss": 0.085, "step": 77330 }, { "epoch": 37.12, "learning_rate": 1.9734983120395014e-05, "loss": 0.0669, "step": 77340 }, { "epoch": 37.13, "learning_rate": 1.9723555214534983e-05, "loss": 0.0791, "step": 77350 }, { "epoch": 37.13, "learning_rate": 1.9712129805395728e-05, "loss": 0.083, "step": 77360 }, { "epoch": 37.14, "learning_rate": 1.9700706893919456e-05, "loss": 0.0748, "step": 77370 }, { "epoch": 37.14, "learning_rate": 1.9689286481048106e-05, "loss": 0.0772, "step": 77380 }, { "epoch": 37.15, "learning_rate": 1.967786856772348e-05, "loss": 0.0755, "step": 77390 }, { "epoch": 37.15, "learning_rate": 1.9666453154887134e-05, "loss": 0.0755, "step": 77400 }, { "epoch": 37.15, "learning_rate": 1.9655040243480437e-05, "loss": 0.073, "step": 77410 }, { "epoch": 37.16, "learning_rate": 1.9643629834444542e-05, "loss": 0.0758, "step": 77420 }, { "epoch": 37.16, "learning_rate": 1.96322219287204e-05, "loss": 0.0714, "step": 77430 }, { "epoch": 37.17, "learning_rate": 1.9620816527248774e-05, "loss": 0.0815, "step": 77440 }, { "epoch": 37.17, "learning_rate": 1.9609413630970165e-05, "loss": 0.0706, "step": 77450 }, { "epoch": 37.18, "learning_rate": 1.9598013240824914e-05, "loss": 0.0729, "step": 77460 }, { "epoch": 37.18, "learning_rate": 1.958661535775315e-05, "loss": 0.0864, "step": 77470 }, { "epoch": 37.19, "learning_rate": 1.9575219982694788e-05, "loss": 0.0684, "step": 77480 }, { "epoch": 37.19, "learning_rate": 1.956382711658954e-05, "loss": 0.0763, "step": 77490 }, { "epoch": 37.2, "learning_rate": 1.9552436760376907e-05, "loss": 0.0726, "step": 77500 }, { "epoch": 37.2, "learning_rate": 1.95410489149962e-05, "loss": 0.0723, "step": 77510 }, { "epoch": 37.21, "learning_rate": 1.9529663581386477e-05, "loss": 0.0746, "step": 77520 }, { "epoch": 37.21, "learning_rate": 1.9518280760486635e-05, "loss": 0.0824, "step": 77530 }, { "epoch": 37.21, "learning_rate": 1.950690045323535e-05, "loss": 0.0865, "step": 77540 }, { "epoch": 37.22, "learning_rate": 1.9495522660571085e-05, "loss": 0.0728, "step": 77550 }, { "epoch": 37.22, "learning_rate": 1.9484147383432104e-05, "loss": 0.0734, "step": 77560 }, { "epoch": 37.23, "learning_rate": 1.947277462275646e-05, "loss": 0.0726, "step": 77570 }, { "epoch": 37.23, "learning_rate": 1.9461404379481996e-05, "loss": 0.075, "step": 77580 }, { "epoch": 37.24, "learning_rate": 1.9450036654546367e-05, "loss": 0.0776, "step": 77590 }, { "epoch": 37.24, "learning_rate": 1.9438671448886965e-05, "loss": 0.0743, "step": 77600 }, { "epoch": 37.25, "learning_rate": 1.9427308763441032e-05, "loss": 0.0738, "step": 77610 }, { "epoch": 37.25, "learning_rate": 1.94159485991456e-05, "loss": 0.0713, "step": 77620 }, { "epoch": 37.26, "learning_rate": 1.9404590956937417e-05, "loss": 0.0742, "step": 77630 }, { "epoch": 37.26, "learning_rate": 1.9393235837753147e-05, "loss": 0.0778, "step": 77640 }, { "epoch": 37.26, "learning_rate": 1.9381883242529146e-05, "loss": 0.0718, "step": 77650 }, { "epoch": 37.27, "learning_rate": 1.9370533172201626e-05, "loss": 0.0797, "step": 77660 }, { "epoch": 37.27, "learning_rate": 1.9359185627706523e-05, "loss": 0.0703, "step": 77670 }, { "epoch": 37.28, "learning_rate": 1.9347840609979617e-05, "loss": 0.0782, "step": 77680 }, { "epoch": 37.28, "learning_rate": 1.9336498119956482e-05, "loss": 0.0776, "step": 77690 }, { "epoch": 37.29, "learning_rate": 1.9325158158572433e-05, "loss": 0.0735, "step": 77700 }, { "epoch": 37.29, "learning_rate": 1.931382072676261e-05, "loss": 0.0739, "step": 77710 }, { "epoch": 37.3, "learning_rate": 1.930248582546198e-05, "loss": 0.0687, "step": 77720 }, { "epoch": 37.3, "learning_rate": 1.9291153455605264e-05, "loss": 0.0738, "step": 77730 }, { "epoch": 37.31, "learning_rate": 1.927982361812694e-05, "loss": 0.0836, "step": 77740 }, { "epoch": 37.31, "learning_rate": 1.9268496313961343e-05, "loss": 0.0739, "step": 77750 }, { "epoch": 37.31, "learning_rate": 1.925717154404255e-05, "loss": 0.0858, "step": 77760 }, { "epoch": 37.32, "learning_rate": 1.924584930930448e-05, "loss": 0.073, "step": 77770 }, { "epoch": 37.32, "learning_rate": 1.9234529610680773e-05, "loss": 0.0756, "step": 77780 }, { "epoch": 37.33, "learning_rate": 1.9223212449104894e-05, "loss": 0.0723, "step": 77790 }, { "epoch": 37.33, "learning_rate": 1.9211897825510162e-05, "loss": 0.0718, "step": 77800 }, { "epoch": 37.34, "learning_rate": 1.9200585740829567e-05, "loss": 0.0753, "step": 77810 }, { "epoch": 37.34, "learning_rate": 1.9189276195995974e-05, "loss": 0.0747, "step": 77820 }, { "epoch": 37.35, "learning_rate": 1.9177969191942013e-05, "loss": 0.0671, "step": 77830 }, { "epoch": 37.35, "learning_rate": 1.9166664729600125e-05, "loss": 0.0693, "step": 77840 }, { "epoch": 37.36, "learning_rate": 1.9155362809902488e-05, "loss": 0.0752, "step": 77850 }, { "epoch": 37.36, "learning_rate": 1.9144063433781118e-05, "loss": 0.0779, "step": 77860 }, { "epoch": 37.37, "learning_rate": 1.9132766602167812e-05, "loss": 0.0799, "step": 77870 }, { "epoch": 37.37, "learning_rate": 1.9121472315994155e-05, "loss": 0.0696, "step": 77880 }, { "epoch": 37.37, "learning_rate": 1.911018057619151e-05, "loss": 0.0697, "step": 77890 }, { "epoch": 37.38, "learning_rate": 1.9098891383691058e-05, "loss": 0.0851, "step": 77900 }, { "epoch": 37.38, "learning_rate": 1.9087604739423758e-05, "loss": 0.0715, "step": 77910 }, { "epoch": 37.39, "learning_rate": 1.907632064432032e-05, "loss": 0.0811, "step": 77920 }, { "epoch": 37.39, "learning_rate": 1.9065039099311294e-05, "loss": 0.0816, "step": 77930 }, { "epoch": 37.4, "learning_rate": 1.9053760105327006e-05, "loss": 0.0814, "step": 77940 }, { "epoch": 37.4, "learning_rate": 1.9042483663297572e-05, "loss": 0.073, "step": 77950 }, { "epoch": 37.41, "learning_rate": 1.903120977415288e-05, "loss": 0.0851, "step": 77960 }, { "epoch": 37.41, "learning_rate": 1.9019938438822638e-05, "loss": 0.072, "step": 77970 }, { "epoch": 37.42, "learning_rate": 1.9008669658236332e-05, "loss": 0.0692, "step": 77980 }, { "epoch": 37.42, "learning_rate": 1.8997403433323208e-05, "loss": 0.071, "step": 77990 }, { "epoch": 37.42, "learning_rate": 1.898613976501233e-05, "loss": 0.0719, "step": 78000 }, { "epoch": 37.43, "learning_rate": 1.8974878654232558e-05, "loss": 0.0717, "step": 78010 }, { "epoch": 37.43, "learning_rate": 1.896362010191252e-05, "loss": 0.0723, "step": 78020 }, { "epoch": 37.44, "learning_rate": 1.8952364108980658e-05, "loss": 0.0768, "step": 78030 }, { "epoch": 37.44, "learning_rate": 1.894111067636517e-05, "loss": 0.0778, "step": 78040 }, { "epoch": 37.45, "learning_rate": 1.8929859804994086e-05, "loss": 0.084, "step": 78050 }, { "epoch": 37.45, "learning_rate": 1.891861149579517e-05, "loss": 0.0729, "step": 78060 }, { "epoch": 37.46, "learning_rate": 1.8907365749696016e-05, "loss": 0.0817, "step": 78070 }, { "epoch": 37.46, "learning_rate": 1.8896122567623985e-05, "loss": 0.0758, "step": 78080 }, { "epoch": 37.47, "learning_rate": 1.8884881950506256e-05, "loss": 0.0718, "step": 78090 }, { "epoch": 37.47, "learning_rate": 1.887364389926976e-05, "loss": 0.0717, "step": 78100 }, { "epoch": 37.47, "learning_rate": 1.8862408414841243e-05, "loss": 0.0737, "step": 78110 }, { "epoch": 37.48, "learning_rate": 1.8851175498147244e-05, "loss": 0.0826, "step": 78120 }, { "epoch": 37.48, "learning_rate": 1.883994515011404e-05, "loss": 0.0687, "step": 78130 }, { "epoch": 37.49, "learning_rate": 1.8828717371667752e-05, "loss": 0.0742, "step": 78140 }, { "epoch": 37.49, "learning_rate": 1.8817492163734263e-05, "loss": 0.0715, "step": 78150 }, { "epoch": 37.5, "learning_rate": 1.8806269527239258e-05, "loss": 0.0752, "step": 78160 }, { "epoch": 37.5, "learning_rate": 1.8795049463108196e-05, "loss": 0.0734, "step": 78170 }, { "epoch": 37.51, "learning_rate": 1.878383197226633e-05, "loss": 0.0788, "step": 78180 }, { "epoch": 37.51, "learning_rate": 1.87726170556387e-05, "loss": 0.0727, "step": 78190 }, { "epoch": 37.52, "learning_rate": 1.876140471415016e-05, "loss": 0.0803, "step": 78200 }, { "epoch": 37.52, "learning_rate": 1.8750194948725286e-05, "loss": 0.0731, "step": 78210 }, { "epoch": 37.53, "learning_rate": 1.8738987760288495e-05, "loss": 0.0707, "step": 78220 }, { "epoch": 37.53, "learning_rate": 1.8727783149763984e-05, "loss": 0.0692, "step": 78230 }, { "epoch": 37.53, "learning_rate": 1.8716581118075726e-05, "loss": 0.0738, "step": 78240 }, { "epoch": 37.54, "learning_rate": 1.8705381666147493e-05, "loss": 0.08, "step": 78250 }, { "epoch": 37.54, "learning_rate": 1.8694184794902837e-05, "loss": 0.0732, "step": 78260 }, { "epoch": 37.55, "learning_rate": 1.8682990505265107e-05, "loss": 0.0838, "step": 78270 }, { "epoch": 37.55, "learning_rate": 1.867179879815741e-05, "loss": 0.0811, "step": 78280 }, { "epoch": 37.56, "learning_rate": 1.8660609674502666e-05, "loss": 0.0798, "step": 78290 }, { "epoch": 37.56, "learning_rate": 1.8649423135223593e-05, "loss": 0.071, "step": 78300 }, { "epoch": 37.57, "learning_rate": 1.863823918124264e-05, "loss": 0.0727, "step": 78310 }, { "epoch": 37.57, "learning_rate": 1.8627057813482124e-05, "loss": 0.0746, "step": 78320 }, { "epoch": 37.58, "learning_rate": 1.8615879032864087e-05, "loss": 0.0854, "step": 78330 }, { "epoch": 37.58, "learning_rate": 1.8604702840310406e-05, "loss": 0.0808, "step": 78340 }, { "epoch": 37.58, "learning_rate": 1.859352923674267e-05, "loss": 0.0726, "step": 78350 }, { "epoch": 37.59, "learning_rate": 1.8582358223082324e-05, "loss": 0.0722, "step": 78360 }, { "epoch": 37.59, "learning_rate": 1.8571189800250578e-05, "loss": 0.0715, "step": 78370 }, { "epoch": 37.6, "learning_rate": 1.8560023969168432e-05, "loss": 0.0757, "step": 78380 }, { "epoch": 37.6, "learning_rate": 1.854886073075663e-05, "loss": 0.0688, "step": 78390 }, { "epoch": 37.61, "learning_rate": 1.8537700085935782e-05, "loss": 0.0715, "step": 78400 }, { "epoch": 37.61, "learning_rate": 1.8526542035626244e-05, "loss": 0.0762, "step": 78410 }, { "epoch": 37.62, "learning_rate": 1.851538658074811e-05, "loss": 0.0701, "step": 78420 }, { "epoch": 37.62, "learning_rate": 1.8504233722221343e-05, "loss": 0.0699, "step": 78430 }, { "epoch": 37.63, "learning_rate": 1.8493083460965638e-05, "loss": 0.0668, "step": 78440 }, { "epoch": 37.63, "learning_rate": 1.8481935797900505e-05, "loss": 0.0784, "step": 78450 }, { "epoch": 37.63, "learning_rate": 1.8470790733945208e-05, "loss": 0.0742, "step": 78460 }, { "epoch": 37.64, "learning_rate": 1.845964827001881e-05, "loss": 0.0853, "step": 78470 }, { "epoch": 37.64, "learning_rate": 1.8448508407040205e-05, "loss": 0.0771, "step": 78480 }, { "epoch": 37.65, "learning_rate": 1.843737114592799e-05, "loss": 0.0718, "step": 78490 }, { "epoch": 37.65, "learning_rate": 1.8426236487600606e-05, "loss": 0.0833, "step": 78500 }, { "epoch": 37.66, "learning_rate": 1.841510443297626e-05, "loss": 0.0795, "step": 78510 }, { "epoch": 37.66, "learning_rate": 1.840397498297297e-05, "loss": 0.0692, "step": 78520 }, { "epoch": 37.67, "learning_rate": 1.8392848138508477e-05, "loss": 0.0762, "step": 78530 }, { "epoch": 37.67, "learning_rate": 1.838172390050037e-05, "loss": 0.0717, "step": 78540 }, { "epoch": 37.68, "learning_rate": 1.8370602269865973e-05, "loss": 0.0717, "step": 78550 }, { "epoch": 37.68, "learning_rate": 1.835948324752248e-05, "loss": 0.0789, "step": 78560 }, { "epoch": 37.69, "learning_rate": 1.8348366834386755e-05, "loss": 0.0718, "step": 78570 }, { "epoch": 37.69, "learning_rate": 1.833725303137553e-05, "loss": 0.0736, "step": 78580 }, { "epoch": 37.69, "learning_rate": 1.8326141839405296e-05, "loss": 0.075, "step": 78590 }, { "epoch": 37.7, "learning_rate": 1.831503325939231e-05, "loss": 0.0809, "step": 78600 }, { "epoch": 37.7, "learning_rate": 1.830392729225264e-05, "loss": 0.0748, "step": 78610 }, { "epoch": 37.71, "learning_rate": 1.8292823938902122e-05, "loss": 0.0749, "step": 78620 }, { "epoch": 37.71, "learning_rate": 1.8281723200256418e-05, "loss": 0.0768, "step": 78630 }, { "epoch": 37.72, "learning_rate": 1.8270625077230908e-05, "loss": 0.0686, "step": 78640 }, { "epoch": 37.72, "learning_rate": 1.825952957074079e-05, "loss": 0.0779, "step": 78650 }, { "epoch": 37.73, "learning_rate": 1.8248436681701082e-05, "loss": 0.0811, "step": 78660 }, { "epoch": 37.73, "learning_rate": 1.8237346411026497e-05, "loss": 0.0696, "step": 78670 }, { "epoch": 37.74, "learning_rate": 1.8226258759631602e-05, "loss": 0.0734, "step": 78680 }, { "epoch": 37.74, "learning_rate": 1.8215173728430744e-05, "loss": 0.0797, "step": 78690 }, { "epoch": 37.74, "learning_rate": 1.820409131833804e-05, "loss": 0.0722, "step": 78700 }, { "epoch": 37.75, "learning_rate": 1.819301153026738e-05, "loss": 0.0768, "step": 78710 }, { "epoch": 37.75, "learning_rate": 1.8181934365132453e-05, "loss": 0.0802, "step": 78720 }, { "epoch": 37.76, "learning_rate": 1.8170859823846742e-05, "loss": 0.075, "step": 78730 }, { "epoch": 37.76, "learning_rate": 1.815978790732347e-05, "loss": 0.0727, "step": 78740 }, { "epoch": 37.77, "learning_rate": 1.8148718616475696e-05, "loss": 0.0788, "step": 78750 }, { "epoch": 37.77, "learning_rate": 1.8137651952216235e-05, "loss": 0.0739, "step": 78760 }, { "epoch": 37.78, "learning_rate": 1.8126587915457676e-05, "loss": 0.071, "step": 78770 }, { "epoch": 37.78, "learning_rate": 1.811552650711243e-05, "loss": 0.0785, "step": 78780 }, { "epoch": 37.79, "learning_rate": 1.8104467728092644e-05, "loss": 0.0683, "step": 78790 }, { "epoch": 37.79, "learning_rate": 1.8093411579310283e-05, "loss": 0.0775, "step": 78800 }, { "epoch": 37.79, "learning_rate": 1.8082358061677097e-05, "loss": 0.0759, "step": 78810 }, { "epoch": 37.8, "learning_rate": 1.8071307176104567e-05, "loss": 0.0725, "step": 78820 }, { "epoch": 37.8, "learning_rate": 1.806025892350402e-05, "loss": 0.0862, "step": 78830 }, { "epoch": 37.81, "learning_rate": 1.8049213304786533e-05, "loss": 0.0645, "step": 78840 }, { "epoch": 37.81, "learning_rate": 1.803817032086298e-05, "loss": 0.0827, "step": 78850 }, { "epoch": 37.82, "learning_rate": 1.8027129972644e-05, "loss": 0.0691, "step": 78860 }, { "epoch": 37.82, "learning_rate": 1.801609226104004e-05, "loss": 0.0753, "step": 78870 }, { "epoch": 37.83, "learning_rate": 1.800505718696132e-05, "loss": 0.0696, "step": 78880 }, { "epoch": 37.83, "learning_rate": 1.7994024751317817e-05, "loss": 0.0718, "step": 78890 }, { "epoch": 37.84, "learning_rate": 1.798299495501931e-05, "loss": 0.0714, "step": 78900 }, { "epoch": 37.84, "learning_rate": 1.797196779897538e-05, "loss": 0.0789, "step": 78910 }, { "epoch": 37.85, "learning_rate": 1.796094328409536e-05, "loss": 0.0785, "step": 78920 }, { "epoch": 37.85, "learning_rate": 1.7949921411288384e-05, "loss": 0.0673, "step": 78930 }, { "epoch": 37.85, "learning_rate": 1.793890218146336e-05, "loss": 0.0682, "step": 78940 }, { "epoch": 37.86, "learning_rate": 1.7927885595528988e-05, "loss": 0.0748, "step": 78950 }, { "epoch": 37.86, "learning_rate": 1.7916871654393713e-05, "loss": 0.0758, "step": 78960 }, { "epoch": 37.87, "learning_rate": 1.790586035896581e-05, "loss": 0.0706, "step": 78970 }, { "epoch": 37.87, "learning_rate": 1.7894851710153314e-05, "loss": 0.0722, "step": 78980 }, { "epoch": 37.88, "learning_rate": 1.7883845708864045e-05, "loss": 0.0911, "step": 78990 }, { "epoch": 37.88, "learning_rate": 1.7872842356005597e-05, "loss": 0.0794, "step": 79000 }, { "epoch": 37.89, "learning_rate": 1.786184165248536e-05, "loss": 0.0774, "step": 79010 }, { "epoch": 37.89, "learning_rate": 1.7850843599210514e-05, "loss": 0.0723, "step": 79020 }, { "epoch": 37.9, "learning_rate": 1.783984819708796e-05, "loss": 0.0704, "step": 79030 }, { "epoch": 37.9, "learning_rate": 1.7828855447024455e-05, "loss": 0.068, "step": 79040 }, { "epoch": 37.9, "learning_rate": 1.7817865349926494e-05, "loss": 0.0668, "step": 79050 }, { "epoch": 37.91, "learning_rate": 1.7806877906700392e-05, "loss": 0.0746, "step": 79060 }, { "epoch": 37.91, "learning_rate": 1.7795893118252167e-05, "loss": 0.0706, "step": 79070 }, { "epoch": 37.92, "learning_rate": 1.7784910985487713e-05, "loss": 0.0748, "step": 79080 }, { "epoch": 37.92, "learning_rate": 1.7773931509312675e-05, "loss": 0.075, "step": 79090 }, { "epoch": 37.93, "learning_rate": 1.7762954690632417e-05, "loss": 0.0738, "step": 79100 }, { "epoch": 37.93, "learning_rate": 1.7751980530352164e-05, "loss": 0.0843, "step": 79110 }, { "epoch": 37.94, "learning_rate": 1.774100902937688e-05, "loss": 0.079, "step": 79120 }, { "epoch": 37.94, "learning_rate": 1.773004018861134e-05, "loss": 0.0787, "step": 79130 }, { "epoch": 37.95, "learning_rate": 1.771907400896005e-05, "loss": 0.0771, "step": 79140 }, { "epoch": 37.95, "learning_rate": 1.7708110491327327e-05, "loss": 0.0708, "step": 79150 }, { "epoch": 38.0, "learning_rate": 1.7697149636617293e-05, "loss": 0.0751, "step": 79160 }, { "epoch": 38.01, "learning_rate": 1.7686191445733842e-05, "loss": 0.0659, "step": 79170 }, { "epoch": 38.01, "learning_rate": 1.7675235919580573e-05, "loss": 0.0711, "step": 79180 }, { "epoch": 38.02, "learning_rate": 1.7664283059060964e-05, "loss": 0.0721, "step": 79190 }, { "epoch": 38.02, "learning_rate": 1.7653332865078242e-05, "loss": 0.0741, "step": 79200 }, { "epoch": 38.03, "learning_rate": 1.7642385338535367e-05, "loss": 0.0807, "step": 79210 }, { "epoch": 38.03, "learning_rate": 1.763144048033512e-05, "loss": 0.0816, "step": 79220 }, { "epoch": 38.03, "learning_rate": 1.7620498291380096e-05, "loss": 0.0715, "step": 79230 }, { "epoch": 38.04, "learning_rate": 1.7609558772572632e-05, "loss": 0.0751, "step": 79240 }, { "epoch": 38.04, "learning_rate": 1.7598621924814806e-05, "loss": 0.0815, "step": 79250 }, { "epoch": 38.05, "learning_rate": 1.758768774900854e-05, "loss": 0.068, "step": 79260 }, { "epoch": 38.05, "learning_rate": 1.757675624605553e-05, "loss": 0.0699, "step": 79270 }, { "epoch": 38.06, "learning_rate": 1.7565827416857193e-05, "loss": 0.0807, "step": 79280 }, { "epoch": 38.06, "learning_rate": 1.7554901262314784e-05, "loss": 0.0855, "step": 79290 }, { "epoch": 38.07, "learning_rate": 1.75439777833293e-05, "loss": 0.0804, "step": 79300 }, { "epoch": 38.07, "learning_rate": 1.7533056980801594e-05, "loss": 0.0816, "step": 79310 }, { "epoch": 38.08, "learning_rate": 1.7522138855632182e-05, "loss": 0.0832, "step": 79320 }, { "epoch": 38.08, "learning_rate": 1.751122340872144e-05, "loss": 0.0737, "step": 79330 }, { "epoch": 38.09, "learning_rate": 1.7500310640969496e-05, "loss": 0.0773, "step": 79340 }, { "epoch": 38.09, "learning_rate": 1.7489400553276282e-05, "loss": 0.0755, "step": 79350 }, { "epoch": 38.09, "learning_rate": 1.7478493146541453e-05, "loss": 0.0831, "step": 79360 }, { "epoch": 38.1, "learning_rate": 1.74675884216645e-05, "loss": 0.072, "step": 79370 }, { "epoch": 38.1, "learning_rate": 1.745668637954467e-05, "loss": 0.0736, "step": 79380 }, { "epoch": 38.11, "learning_rate": 1.744578702108099e-05, "loss": 0.0721, "step": 79390 }, { "epoch": 38.11, "learning_rate": 1.743489034717226e-05, "loss": 0.0741, "step": 79400 }, { "epoch": 38.12, "learning_rate": 1.742399635871707e-05, "loss": 0.0766, "step": 79410 }, { "epoch": 38.12, "learning_rate": 1.741310505661381e-05, "loss": 0.078, "step": 79420 }, { "epoch": 38.13, "learning_rate": 1.7402216441760572e-05, "loss": 0.0789, "step": 79430 }, { "epoch": 38.13, "learning_rate": 1.73913305150553e-05, "loss": 0.0843, "step": 79440 }, { "epoch": 38.14, "learning_rate": 1.738044727739569e-05, "loss": 0.0825, "step": 79450 }, { "epoch": 38.14, "learning_rate": 1.7369566729679227e-05, "loss": 0.074, "step": 79460 }, { "epoch": 38.14, "learning_rate": 1.7358688872803153e-05, "loss": 0.0767, "step": 79470 }, { "epoch": 38.15, "learning_rate": 1.7347813707664513e-05, "loss": 0.0745, "step": 79480 }, { "epoch": 38.15, "learning_rate": 1.7336941235160126e-05, "loss": 0.068, "step": 79490 }, { "epoch": 38.16, "learning_rate": 1.7326071456186555e-05, "loss": 0.085, "step": 79500 }, { "epoch": 38.16, "learning_rate": 1.7315204371640175e-05, "loss": 0.0781, "step": 79510 }, { "epoch": 38.17, "learning_rate": 1.730433998241714e-05, "loss": 0.0761, "step": 79520 }, { "epoch": 38.17, "learning_rate": 1.7293478289413366e-05, "loss": 0.0689, "step": 79530 }, { "epoch": 38.18, "learning_rate": 1.7282619293524555e-05, "loss": 0.0749, "step": 79540 }, { "epoch": 38.18, "learning_rate": 1.7271762995646185e-05, "loss": 0.0776, "step": 79550 }, { "epoch": 38.19, "learning_rate": 1.726090939667353e-05, "loss": 0.0778, "step": 79560 }, { "epoch": 38.19, "learning_rate": 1.7250058497501587e-05, "loss": 0.0807, "step": 79570 }, { "epoch": 38.19, "learning_rate": 1.7239210299025178e-05, "loss": 0.0703, "step": 79580 }, { "epoch": 38.2, "learning_rate": 1.7228364802138897e-05, "loss": 0.082, "step": 79590 }, { "epoch": 38.2, "learning_rate": 1.7217522007737106e-05, "loss": 0.0816, "step": 79600 }, { "epoch": 38.21, "learning_rate": 1.720668191671395e-05, "loss": 0.0662, "step": 79610 }, { "epoch": 38.21, "learning_rate": 1.7195844529963344e-05, "loss": 0.0685, "step": 79620 }, { "epoch": 38.22, "learning_rate": 1.7185009848379e-05, "loss": 0.0749, "step": 79630 }, { "epoch": 38.22, "learning_rate": 1.7174177872854356e-05, "loss": 0.074, "step": 79640 }, { "epoch": 38.23, "learning_rate": 1.7163348604282686e-05, "loss": 0.0709, "step": 79650 }, { "epoch": 38.23, "learning_rate": 1.7152522043557008e-05, "loss": 0.0696, "step": 79660 }, { "epoch": 38.24, "learning_rate": 1.7141698191570128e-05, "loss": 0.0758, "step": 79670 }, { "epoch": 38.24, "learning_rate": 1.7130877049214626e-05, "loss": 0.0731, "step": 79680 }, { "epoch": 38.24, "learning_rate": 1.7120058617382855e-05, "loss": 0.0689, "step": 79690 }, { "epoch": 38.25, "learning_rate": 1.710924289696697e-05, "loss": 0.0762, "step": 79700 }, { "epoch": 38.25, "learning_rate": 1.7098429888858843e-05, "loss": 0.0729, "step": 79710 }, { "epoch": 38.26, "learning_rate": 1.7087619593950173e-05, "loss": 0.0756, "step": 79720 }, { "epoch": 38.26, "learning_rate": 1.7076812013132428e-05, "loss": 0.0841, "step": 79730 }, { "epoch": 38.27, "learning_rate": 1.7066007147296842e-05, "loss": 0.0698, "step": 79740 }, { "epoch": 38.27, "learning_rate": 1.7055204997334427e-05, "loss": 0.0742, "step": 79750 }, { "epoch": 38.28, "learning_rate": 1.704440556413598e-05, "loss": 0.0798, "step": 79760 }, { "epoch": 38.28, "learning_rate": 1.7033608848592058e-05, "loss": 0.0749, "step": 79770 }, { "epoch": 38.29, "learning_rate": 1.7022814851593028e-05, "loss": 0.067, "step": 79780 }, { "epoch": 38.29, "learning_rate": 1.701202357402897e-05, "loss": 0.0749, "step": 79790 }, { "epoch": 38.3, "learning_rate": 1.700123501678979e-05, "loss": 0.0735, "step": 79800 }, { "epoch": 38.3, "learning_rate": 1.6990449180765183e-05, "loss": 0.0725, "step": 79810 }, { "epoch": 38.3, "learning_rate": 1.6979666066844537e-05, "loss": 0.066, "step": 79820 }, { "epoch": 38.31, "learning_rate": 1.696888567591712e-05, "loss": 0.0694, "step": 79830 }, { "epoch": 38.31, "learning_rate": 1.695810800887192e-05, "loss": 0.0726, "step": 79840 }, { "epoch": 38.32, "learning_rate": 1.694733306659772e-05, "loss": 0.0795, "step": 79850 }, { "epoch": 38.32, "learning_rate": 1.693656084998303e-05, "loss": 0.075, "step": 79860 }, { "epoch": 38.33, "learning_rate": 1.6925791359916187e-05, "loss": 0.083, "step": 79870 }, { "epoch": 38.33, "learning_rate": 1.6915024597285312e-05, "loss": 0.0792, "step": 79880 }, { "epoch": 38.34, "learning_rate": 1.6904260562978237e-05, "loss": 0.0753, "step": 79890 }, { "epoch": 38.34, "learning_rate": 1.689349925788261e-05, "loss": 0.0786, "step": 79900 }, { "epoch": 38.35, "learning_rate": 1.688274068288589e-05, "loss": 0.072, "step": 79910 }, { "epoch": 38.35, "learning_rate": 1.687198483887527e-05, "loss": 0.0662, "step": 79920 }, { "epoch": 38.35, "learning_rate": 1.6861231726737686e-05, "loss": 0.073, "step": 79930 }, { "epoch": 38.36, "learning_rate": 1.685048134735991e-05, "loss": 0.0692, "step": 79940 }, { "epoch": 38.36, "learning_rate": 1.6839733701628447e-05, "loss": 0.0781, "step": 79950 }, { "epoch": 38.37, "learning_rate": 1.6828988790429623e-05, "loss": 0.0763, "step": 79960 }, { "epoch": 38.37, "learning_rate": 1.681824661464947e-05, "loss": 0.0628, "step": 79970 }, { "epoch": 38.38, "learning_rate": 1.6807507175173827e-05, "loss": 0.0735, "step": 79980 }, { "epoch": 38.38, "learning_rate": 1.6796770472888374e-05, "loss": 0.066, "step": 79990 }, { "epoch": 38.39, "learning_rate": 1.6786036508678437e-05, "loss": 0.0736, "step": 80000 }, { "epoch": 38.39, "learning_rate": 1.677530528342922e-05, "loss": 0.0761, "step": 80010 }, { "epoch": 38.4, "learning_rate": 1.676457679802565e-05, "loss": 0.0684, "step": 80020 }, { "epoch": 38.4, "learning_rate": 1.675385105335246e-05, "loss": 0.0775, "step": 80030 }, { "epoch": 38.4, "learning_rate": 1.674312805029411e-05, "loss": 0.0705, "step": 80040 }, { "epoch": 38.41, "learning_rate": 1.6732407789734884e-05, "loss": 0.0746, "step": 80050 }, { "epoch": 38.41, "learning_rate": 1.6721690272558803e-05, "loss": 0.0788, "step": 80060 }, { "epoch": 38.42, "learning_rate": 1.67109754996497e-05, "loss": 0.0758, "step": 80070 }, { "epoch": 38.42, "learning_rate": 1.670026347189114e-05, "loss": 0.0685, "step": 80080 }, { "epoch": 38.43, "learning_rate": 1.668955419016649e-05, "loss": 0.0834, "step": 80090 }, { "epoch": 38.43, "learning_rate": 1.6678847655358897e-05, "loss": 0.0724, "step": 80100 }, { "epoch": 38.44, "learning_rate": 1.6668143868351237e-05, "loss": 0.075, "step": 80110 }, { "epoch": 38.44, "learning_rate": 1.66574428300262e-05, "loss": 0.0691, "step": 80120 }, { "epoch": 38.45, "learning_rate": 1.6646744541266235e-05, "loss": 0.0672, "step": 80130 }, { "epoch": 38.45, "learning_rate": 1.6636049002953575e-05, "loss": 0.0762, "step": 80140 }, { "epoch": 38.46, "learning_rate": 1.662535621597021e-05, "loss": 0.0766, "step": 80150 }, { "epoch": 38.46, "learning_rate": 1.6614666181197924e-05, "loss": 0.064, "step": 80160 }, { "epoch": 38.46, "learning_rate": 1.6603978899518262e-05, "loss": 0.0714, "step": 80170 }, { "epoch": 38.47, "learning_rate": 1.6593294371812518e-05, "loss": 0.0733, "step": 80180 }, { "epoch": 38.47, "learning_rate": 1.65826125989618e-05, "loss": 0.0795, "step": 80190 }, { "epoch": 38.48, "learning_rate": 1.6571933581846965e-05, "loss": 0.0853, "step": 80200 }, { "epoch": 38.48, "learning_rate": 1.656125732134865e-05, "loss": 0.0671, "step": 80210 }, { "epoch": 38.49, "learning_rate": 1.655058381834727e-05, "loss": 0.0894, "step": 80220 }, { "epoch": 38.49, "learning_rate": 1.6539913073723003e-05, "loss": 0.0713, "step": 80230 }, { "epoch": 38.5, "learning_rate": 1.652924508835581e-05, "loss": 0.0806, "step": 80240 }, { "epoch": 38.5, "learning_rate": 1.6518579863125398e-05, "loss": 0.0776, "step": 80250 }, { "epoch": 38.51, "learning_rate": 1.6507917398911276e-05, "loss": 0.0803, "step": 80260 }, { "epoch": 38.51, "learning_rate": 1.6497257696592718e-05, "loss": 0.0815, "step": 80270 }, { "epoch": 38.51, "learning_rate": 1.648660075704876e-05, "loss": 0.0706, "step": 80280 }, { "epoch": 38.52, "learning_rate": 1.6475946581158218e-05, "loss": 0.071, "step": 80290 }, { "epoch": 38.52, "learning_rate": 1.6465295169799687e-05, "loss": 0.0797, "step": 80300 }, { "epoch": 38.53, "learning_rate": 1.645464652385154e-05, "loss": 0.0689, "step": 80310 }, { "epoch": 38.53, "learning_rate": 1.644400064419187e-05, "loss": 0.0672, "step": 80320 }, { "epoch": 38.54, "learning_rate": 1.6433357531698607e-05, "loss": 0.07, "step": 80330 }, { "epoch": 38.54, "learning_rate": 1.6422717187249414e-05, "loss": 0.0746, "step": 80340 }, { "epoch": 38.55, "learning_rate": 1.6412079611721747e-05, "loss": 0.0767, "step": 80350 }, { "epoch": 38.55, "learning_rate": 1.640144480599282e-05, "loss": 0.0684, "step": 80360 }, { "epoch": 38.56, "learning_rate": 1.6390812770939622e-05, "loss": 0.0718, "step": 80370 }, { "epoch": 38.56, "learning_rate": 1.638018350743892e-05, "loss": 0.0775, "step": 80380 }, { "epoch": 38.56, "learning_rate": 1.636955701636726e-05, "loss": 0.0805, "step": 80390 }, { "epoch": 38.57, "learning_rate": 1.635893329860091e-05, "loss": 0.0814, "step": 80400 }, { "epoch": 38.57, "learning_rate": 1.634831235501597e-05, "loss": 0.0665, "step": 80410 }, { "epoch": 38.58, "learning_rate": 1.6337694186488277e-05, "loss": 0.0666, "step": 80420 }, { "epoch": 38.58, "learning_rate": 1.6327078793893463e-05, "loss": 0.0739, "step": 80430 }, { "epoch": 38.59, "learning_rate": 1.6316466178106904e-05, "loss": 0.0739, "step": 80440 }, { "epoch": 38.59, "learning_rate": 1.6305856340003767e-05, "loss": 0.0685, "step": 80450 }, { "epoch": 38.6, "learning_rate": 1.6295249280458996e-05, "loss": 0.0751, "step": 80460 }, { "epoch": 38.6, "learning_rate": 1.628464500034726e-05, "loss": 0.0683, "step": 80470 }, { "epoch": 38.61, "learning_rate": 1.6274043500543053e-05, "loss": 0.0825, "step": 80480 }, { "epoch": 38.61, "learning_rate": 1.6263444781920628e-05, "loss": 0.0781, "step": 80490 }, { "epoch": 38.62, "learning_rate": 1.6252848845353958e-05, "loss": 0.0691, "step": 80500 }, { "epoch": 38.62, "learning_rate": 1.624225569171687e-05, "loss": 0.0803, "step": 80510 }, { "epoch": 38.62, "learning_rate": 1.6231665321882905e-05, "loss": 0.0762, "step": 80520 }, { "epoch": 38.63, "learning_rate": 1.622107773672541e-05, "loss": 0.0887, "step": 80530 }, { "epoch": 38.63, "learning_rate": 1.6210492937117437e-05, "loss": 0.0834, "step": 80540 }, { "epoch": 38.64, "learning_rate": 1.6199910923931878e-05, "loss": 0.0809, "step": 80550 }, { "epoch": 38.64, "learning_rate": 1.6189331698041372e-05, "loss": 0.069, "step": 80560 }, { "epoch": 38.65, "learning_rate": 1.617875526031833e-05, "loss": 0.0811, "step": 80570 }, { "epoch": 38.65, "learning_rate": 1.6168181611634892e-05, "loss": 0.0786, "step": 80580 }, { "epoch": 38.66, "learning_rate": 1.6157610752863046e-05, "loss": 0.0699, "step": 80590 }, { "epoch": 38.66, "learning_rate": 1.6147042684874508e-05, "loss": 0.0728, "step": 80600 }, { "epoch": 38.67, "learning_rate": 1.613647740854074e-05, "loss": 0.0666, "step": 80610 }, { "epoch": 38.67, "learning_rate": 1.612591492473301e-05, "loss": 0.0761, "step": 80620 }, { "epoch": 38.67, "learning_rate": 1.6115355234322343e-05, "loss": 0.0694, "step": 80630 }, { "epoch": 38.68, "learning_rate": 1.610479833817955e-05, "loss": 0.0612, "step": 80640 }, { "epoch": 38.68, "learning_rate": 1.609424423717517e-05, "loss": 0.0787, "step": 80650 }, { "epoch": 38.69, "learning_rate": 1.608369293217953e-05, "loss": 0.0672, "step": 80660 }, { "epoch": 38.69, "learning_rate": 1.6073144424062787e-05, "loss": 0.0821, "step": 80670 }, { "epoch": 38.7, "learning_rate": 1.6062598713694764e-05, "loss": 0.0844, "step": 80680 }, { "epoch": 38.7, "learning_rate": 1.6052055801945125e-05, "loss": 0.0677, "step": 80690 }, { "epoch": 38.71, "learning_rate": 1.604151568968328e-05, "loss": 0.0807, "step": 80700 }, { "epoch": 38.71, "learning_rate": 1.6030978377778415e-05, "loss": 0.0761, "step": 80710 }, { "epoch": 38.72, "learning_rate": 1.6020443867099467e-05, "loss": 0.0728, "step": 80720 }, { "epoch": 38.72, "learning_rate": 1.6009912158515162e-05, "loss": 0.0818, "step": 80730 }, { "epoch": 38.72, "learning_rate": 1.5999383252893983e-05, "loss": 0.0764, "step": 80740 }, { "epoch": 38.73, "learning_rate": 1.5988857151104197e-05, "loss": 0.0818, "step": 80750 }, { "epoch": 38.73, "learning_rate": 1.5978333854013823e-05, "loss": 0.0797, "step": 80760 }, { "epoch": 38.74, "learning_rate": 1.596781336249066e-05, "loss": 0.0666, "step": 80770 }, { "epoch": 38.74, "learning_rate": 1.5957295677402284e-05, "loss": 0.0733, "step": 80780 }, { "epoch": 38.75, "learning_rate": 1.5946780799615995e-05, "loss": 0.0734, "step": 80790 }, { "epoch": 38.75, "learning_rate": 1.5936268729998914e-05, "loss": 0.0721, "step": 80800 }, { "epoch": 38.76, "learning_rate": 1.5925759469417904e-05, "loss": 0.0745, "step": 80810 }, { "epoch": 38.76, "learning_rate": 1.5915253018739606e-05, "loss": 0.0677, "step": 80820 }, { "epoch": 38.77, "learning_rate": 1.5904749378830418e-05, "loss": 0.0738, "step": 80830 }, { "epoch": 38.77, "learning_rate": 1.5894248550556524e-05, "loss": 0.0759, "step": 80840 }, { "epoch": 38.78, "learning_rate": 1.5883750534783875e-05, "loss": 0.0828, "step": 80850 }, { "epoch": 38.78, "learning_rate": 1.587325533237815e-05, "loss": 0.0674, "step": 80860 }, { "epoch": 38.78, "learning_rate": 1.5862762944204844e-05, "loss": 0.0692, "step": 80870 }, { "epoch": 38.79, "learning_rate": 1.58522733711292e-05, "loss": 0.0757, "step": 80880 }, { "epoch": 38.79, "learning_rate": 1.584178661401624e-05, "loss": 0.0771, "step": 80890 }, { "epoch": 38.8, "learning_rate": 1.583130267373073e-05, "loss": 0.0763, "step": 80900 }, { "epoch": 38.8, "learning_rate": 1.5820821551137234e-05, "loss": 0.0678, "step": 80910 }, { "epoch": 38.81, "learning_rate": 1.5810343247100078e-05, "loss": 0.0743, "step": 80920 }, { "epoch": 38.81, "learning_rate": 1.5799867762483312e-05, "loss": 0.0642, "step": 80930 }, { "epoch": 38.82, "learning_rate": 1.578939509815081e-05, "loss": 0.0764, "step": 80940 }, { "epoch": 38.82, "learning_rate": 1.577892525496618e-05, "loss": 0.0754, "step": 80950 }, { "epoch": 38.83, "learning_rate": 1.5768458233792828e-05, "loss": 0.067, "step": 80960 }, { "epoch": 38.83, "learning_rate": 1.575799403549389e-05, "loss": 0.0774, "step": 80970 }, { "epoch": 38.83, "learning_rate": 1.574753266093229e-05, "loss": 0.0741, "step": 80980 }, { "epoch": 38.84, "learning_rate": 1.5737074110970723e-05, "loss": 0.0759, "step": 80990 }, { "epoch": 38.84, "learning_rate": 1.572661838647166e-05, "loss": 0.0761, "step": 81000 }, { "epoch": 38.85, "learning_rate": 1.571616548829728e-05, "loss": 0.0753, "step": 81010 }, { "epoch": 38.85, "learning_rate": 1.57057154173096e-05, "loss": 0.0801, "step": 81020 }, { "epoch": 38.86, "learning_rate": 1.5695268174370376e-05, "loss": 0.0778, "step": 81030 }, { "epoch": 38.86, "learning_rate": 1.5684823760341123e-05, "loss": 0.0753, "step": 81040 }, { "epoch": 38.87, "learning_rate": 1.5674382176083135e-05, "loss": 0.0721, "step": 81050 }, { "epoch": 38.87, "learning_rate": 1.5663943422457467e-05, "loss": 0.076, "step": 81060 }, { "epoch": 38.88, "learning_rate": 1.565350750032496e-05, "loss": 0.0794, "step": 81070 }, { "epoch": 38.88, "learning_rate": 1.5643074410546165e-05, "loss": 0.0767, "step": 81080 }, { "epoch": 38.88, "learning_rate": 1.5632644153981463e-05, "loss": 0.0749, "step": 81090 }, { "epoch": 38.89, "learning_rate": 1.5622216731490973e-05, "loss": 0.0793, "step": 81100 }, { "epoch": 38.89, "learning_rate": 1.5611792143934577e-05, "loss": 0.0707, "step": 81110 }, { "epoch": 38.9, "learning_rate": 1.560137039217194e-05, "loss": 0.0777, "step": 81120 }, { "epoch": 38.9, "learning_rate": 1.559095147706247e-05, "loss": 0.0772, "step": 81130 }, { "epoch": 38.91, "learning_rate": 1.5580535399465383e-05, "loss": 0.0776, "step": 81140 }, { "epoch": 38.91, "learning_rate": 1.557012216023959e-05, "loss": 0.0739, "step": 81150 }, { "epoch": 38.92, "learning_rate": 1.5559711760243833e-05, "loss": 0.0757, "step": 81160 }, { "epoch": 38.92, "learning_rate": 1.5549304200336595e-05, "loss": 0.0717, "step": 81170 }, { "epoch": 38.93, "learning_rate": 1.5538899481376117e-05, "loss": 0.0734, "step": 81180 }, { "epoch": 38.93, "learning_rate": 1.5528497604220427e-05, "loss": 0.0786, "step": 81190 }, { "epoch": 38.94, "learning_rate": 1.55180985697273e-05, "loss": 0.0822, "step": 81200 }, { "epoch": 38.94, "learning_rate": 1.550770237875431e-05, "loss": 0.0785, "step": 81210 }, { "epoch": 38.94, "learning_rate": 1.5497309032158723e-05, "loss": 0.078, "step": 81220 }, { "epoch": 38.95, "learning_rate": 1.548691853079764e-05, "loss": 0.0725, "step": 81230 }, { "epoch": 39.0, "learning_rate": 1.5476530875527907e-05, "loss": 0.086, "step": 81240 }, { "epoch": 39.01, "learning_rate": 1.5466146067206145e-05, "loss": 0.082, "step": 81250 }, { "epoch": 39.01, "learning_rate": 1.545576410668868e-05, "loss": 0.0744, "step": 81260 }, { "epoch": 39.02, "learning_rate": 1.54453849948317e-05, "loss": 0.0651, "step": 81270 }, { "epoch": 39.02, "learning_rate": 1.543500873249112e-05, "loss": 0.0685, "step": 81280 }, { "epoch": 39.02, "learning_rate": 1.5424635320522563e-05, "loss": 0.0688, "step": 81290 }, { "epoch": 39.03, "learning_rate": 1.5414264759781483e-05, "loss": 0.0844, "step": 81300 }, { "epoch": 39.03, "learning_rate": 1.5403897051123085e-05, "loss": 0.0714, "step": 81310 }, { "epoch": 39.04, "learning_rate": 1.5393532195402337e-05, "loss": 0.0706, "step": 81320 }, { "epoch": 39.04, "learning_rate": 1.538317019347395e-05, "loss": 0.073, "step": 81330 }, { "epoch": 39.05, "learning_rate": 1.537281104619241e-05, "loss": 0.0746, "step": 81340 }, { "epoch": 39.05, "learning_rate": 1.536245475441201e-05, "loss": 0.0681, "step": 81350 }, { "epoch": 39.06, "learning_rate": 1.535210131898677e-05, "loss": 0.0752, "step": 81360 }, { "epoch": 39.06, "learning_rate": 1.5341750740770444e-05, "loss": 0.0768, "step": 81370 }, { "epoch": 39.07, "learning_rate": 1.5331403020616608e-05, "loss": 0.0691, "step": 81380 }, { "epoch": 39.07, "learning_rate": 1.5321058159378583e-05, "loss": 0.0805, "step": 81390 }, { "epoch": 39.07, "learning_rate": 1.531071615790942e-05, "loss": 0.0771, "step": 81400 }, { "epoch": 39.08, "learning_rate": 1.5300377017061985e-05, "loss": 0.0781, "step": 81410 }, { "epoch": 39.08, "learning_rate": 1.529004073768886e-05, "loss": 0.0717, "step": 81420 }, { "epoch": 39.09, "learning_rate": 1.5279707320642477e-05, "loss": 0.0803, "step": 81430 }, { "epoch": 39.09, "learning_rate": 1.5269376766774916e-05, "loss": 0.0712, "step": 81440 }, { "epoch": 39.1, "learning_rate": 1.5259049076938098e-05, "loss": 0.0744, "step": 81450 }, { "epoch": 39.1, "learning_rate": 1.5248724251983699e-05, "loss": 0.0674, "step": 81460 }, { "epoch": 39.11, "learning_rate": 1.5238402292763115e-05, "loss": 0.0718, "step": 81470 }, { "epoch": 39.11, "learning_rate": 1.5228083200127557e-05, "loss": 0.0812, "step": 81480 }, { "epoch": 39.12, "learning_rate": 1.521776697492796e-05, "loss": 0.085, "step": 81490 }, { "epoch": 39.12, "learning_rate": 1.5207453618015094e-05, "loss": 0.0762, "step": 81500 }, { "epoch": 39.12, "learning_rate": 1.5197143130239388e-05, "loss": 0.0717, "step": 81510 }, { "epoch": 39.13, "learning_rate": 1.5186835512451114e-05, "loss": 0.0762, "step": 81520 }, { "epoch": 39.13, "learning_rate": 1.5176530765500285e-05, "loss": 0.0662, "step": 81530 }, { "epoch": 39.14, "learning_rate": 1.5166228890236644e-05, "loss": 0.0691, "step": 81540 }, { "epoch": 39.14, "learning_rate": 1.5155929887509746e-05, "loss": 0.0744, "step": 81550 }, { "epoch": 39.15, "learning_rate": 1.5145633758168888e-05, "loss": 0.0721, "step": 81560 }, { "epoch": 39.15, "learning_rate": 1.5135340503063133e-05, "loss": 0.0758, "step": 81570 }, { "epoch": 39.16, "learning_rate": 1.51250501230413e-05, "loss": 0.0688, "step": 81580 }, { "epoch": 39.16, "learning_rate": 1.5114762618951978e-05, "loss": 0.0795, "step": 81590 }, { "epoch": 39.17, "learning_rate": 1.5104477991643517e-05, "loss": 0.0811, "step": 81600 }, { "epoch": 39.17, "learning_rate": 1.5094196241964054e-05, "loss": 0.0815, "step": 81610 }, { "epoch": 39.18, "learning_rate": 1.508391737076142e-05, "loss": 0.0725, "step": 81620 }, { "epoch": 39.18, "learning_rate": 1.507364137888328e-05, "loss": 0.0753, "step": 81630 }, { "epoch": 39.18, "learning_rate": 1.5063368267177025e-05, "loss": 0.0784, "step": 81640 }, { "epoch": 39.19, "learning_rate": 1.5053098036489826e-05, "loss": 0.0763, "step": 81650 }, { "epoch": 39.19, "learning_rate": 1.5042830687668602e-05, "loss": 0.0725, "step": 81660 }, { "epoch": 39.2, "learning_rate": 1.5032566221560052e-05, "loss": 0.072, "step": 81670 }, { "epoch": 39.2, "learning_rate": 1.5022304639010627e-05, "loss": 0.0779, "step": 81680 }, { "epoch": 39.21, "learning_rate": 1.501204594086652e-05, "loss": 0.065, "step": 81690 }, { "epoch": 39.21, "learning_rate": 1.5001790127973719e-05, "loss": 0.0779, "step": 81700 }, { "epoch": 39.22, "learning_rate": 1.4991537201177952e-05, "loss": 0.0717, "step": 81710 }, { "epoch": 39.22, "learning_rate": 1.498128716132473e-05, "loss": 0.0735, "step": 81720 }, { "epoch": 39.23, "learning_rate": 1.4971040009259307e-05, "loss": 0.0736, "step": 81730 }, { "epoch": 39.23, "learning_rate": 1.4960795745826712e-05, "loss": 0.0802, "step": 81740 }, { "epoch": 39.23, "learning_rate": 1.4950554371871733e-05, "loss": 0.0761, "step": 81750 }, { "epoch": 39.24, "learning_rate": 1.4940315888238892e-05, "loss": 0.0676, "step": 81760 }, { "epoch": 39.24, "learning_rate": 1.4930080295772514e-05, "loss": 0.0887, "step": 81770 }, { "epoch": 39.25, "learning_rate": 1.4919847595316666e-05, "loss": 0.0725, "step": 81780 }, { "epoch": 39.25, "learning_rate": 1.490961778771518e-05, "loss": 0.0714, "step": 81790 }, { "epoch": 39.26, "learning_rate": 1.4899390873811642e-05, "loss": 0.0642, "step": 81800 }, { "epoch": 39.26, "learning_rate": 1.4889166854449415e-05, "loss": 0.0816, "step": 81810 }, { "epoch": 39.27, "learning_rate": 1.4878945730471628e-05, "loss": 0.0752, "step": 81820 }, { "epoch": 39.27, "learning_rate": 1.4868727502721114e-05, "loss": 0.0744, "step": 81830 }, { "epoch": 39.28, "learning_rate": 1.485851217204054e-05, "loss": 0.0782, "step": 81840 }, { "epoch": 39.28, "learning_rate": 1.4848299739272303e-05, "loss": 0.0747, "step": 81850 }, { "epoch": 39.28, "learning_rate": 1.4838090205258554e-05, "loss": 0.0787, "step": 81860 }, { "epoch": 39.29, "learning_rate": 1.4827883570841222e-05, "loss": 0.0731, "step": 81870 }, { "epoch": 39.29, "learning_rate": 1.481767983686198e-05, "loss": 0.069, "step": 81880 }, { "epoch": 39.3, "learning_rate": 1.4807479004162294e-05, "loss": 0.0763, "step": 81890 }, { "epoch": 39.3, "learning_rate": 1.479728107358333e-05, "loss": 0.0737, "step": 81900 }, { "epoch": 39.31, "learning_rate": 1.478708604596607e-05, "loss": 0.0745, "step": 81910 }, { "epoch": 39.31, "learning_rate": 1.4776893922151236e-05, "loss": 0.0738, "step": 81920 }, { "epoch": 39.32, "learning_rate": 1.4766704702979328e-05, "loss": 0.0687, "step": 81930 }, { "epoch": 39.32, "learning_rate": 1.4756518389290546e-05, "loss": 0.0743, "step": 81940 }, { "epoch": 39.33, "learning_rate": 1.4746334981924943e-05, "loss": 0.0637, "step": 81950 }, { "epoch": 39.33, "learning_rate": 1.4736154481722264e-05, "loss": 0.0759, "step": 81960 }, { "epoch": 39.34, "learning_rate": 1.4725976889522058e-05, "loss": 0.0784, "step": 81970 }, { "epoch": 39.34, "learning_rate": 1.4715802206163582e-05, "loss": 0.0734, "step": 81980 }, { "epoch": 39.34, "learning_rate": 1.4705630432485884e-05, "loss": 0.0706, "step": 81990 }, { "epoch": 39.35, "learning_rate": 1.46954615693278e-05, "loss": 0.0697, "step": 82000 }, { "epoch": 39.35, "learning_rate": 1.468529561752786e-05, "loss": 0.0669, "step": 82010 }, { "epoch": 39.36, "learning_rate": 1.4675132577924394e-05, "loss": 0.0758, "step": 82020 }, { "epoch": 39.36, "learning_rate": 1.4664972451355514e-05, "loss": 0.0696, "step": 82030 }, { "epoch": 39.37, "learning_rate": 1.4654815238659065e-05, "loss": 0.0723, "step": 82040 }, { "epoch": 39.37, "learning_rate": 1.4644660940672627e-05, "loss": 0.071, "step": 82050 }, { "epoch": 39.38, "learning_rate": 1.4634509558233577e-05, "loss": 0.0674, "step": 82060 }, { "epoch": 39.38, "learning_rate": 1.4624361092179062e-05, "loss": 0.074, "step": 82070 }, { "epoch": 39.39, "learning_rate": 1.4614215543345927e-05, "loss": 0.0696, "step": 82080 }, { "epoch": 39.39, "learning_rate": 1.460407291257082e-05, "loss": 0.0836, "step": 82090 }, { "epoch": 39.39, "learning_rate": 1.4593933200690179e-05, "loss": 0.0764, "step": 82100 }, { "epoch": 39.4, "learning_rate": 1.4583796408540157e-05, "loss": 0.0648, "step": 82110 }, { "epoch": 39.4, "learning_rate": 1.457366253695665e-05, "loss": 0.076, "step": 82120 }, { "epoch": 39.41, "learning_rate": 1.4563531586775359e-05, "loss": 0.0721, "step": 82130 }, { "epoch": 39.41, "learning_rate": 1.4553403558831735e-05, "loss": 0.0731, "step": 82140 }, { "epoch": 39.42, "learning_rate": 1.454327845396094e-05, "loss": 0.0739, "step": 82150 }, { "epoch": 39.42, "learning_rate": 1.4533156272997961e-05, "loss": 0.0656, "step": 82160 }, { "epoch": 39.43, "learning_rate": 1.4523037016777486e-05, "loss": 0.0642, "step": 82170 }, { "epoch": 39.43, "learning_rate": 1.4512920686134046e-05, "loss": 0.0716, "step": 82180 }, { "epoch": 39.44, "learning_rate": 1.4502807281901832e-05, "loss": 0.07, "step": 82190 }, { "epoch": 39.44, "learning_rate": 1.449269680491484e-05, "loss": 0.0655, "step": 82200 }, { "epoch": 39.44, "learning_rate": 1.4482589256006829e-05, "loss": 0.0628, "step": 82210 }, { "epoch": 39.45, "learning_rate": 1.4472484636011325e-05, "loss": 0.0791, "step": 82220 }, { "epoch": 39.45, "learning_rate": 1.4462382945761566e-05, "loss": 0.0747, "step": 82230 }, { "epoch": 39.46, "learning_rate": 1.4452284186090592e-05, "loss": 0.0807, "step": 82240 }, { "epoch": 39.46, "learning_rate": 1.4442188357831187e-05, "loss": 0.0693, "step": 82250 }, { "epoch": 39.47, "learning_rate": 1.4432095461815903e-05, "loss": 0.0743, "step": 82260 }, { "epoch": 39.47, "learning_rate": 1.442200549887704e-05, "loss": 0.0676, "step": 82270 }, { "epoch": 39.48, "learning_rate": 1.4411918469846648e-05, "loss": 0.0694, "step": 82280 }, { "epoch": 39.48, "learning_rate": 1.4401834375556572e-05, "loss": 0.0841, "step": 82290 }, { "epoch": 39.49, "learning_rate": 1.4391753216838349e-05, "loss": 0.0731, "step": 82300 }, { "epoch": 39.49, "learning_rate": 1.438167499452333e-05, "loss": 0.0768, "step": 82310 }, { "epoch": 39.49, "learning_rate": 1.4371599709442618e-05, "loss": 0.0727, "step": 82320 }, { "epoch": 39.5, "learning_rate": 1.4361527362427046e-05, "loss": 0.0734, "step": 82330 }, { "epoch": 39.5, "learning_rate": 1.435145795430723e-05, "loss": 0.0719, "step": 82340 }, { "epoch": 39.51, "learning_rate": 1.4341391485913535e-05, "loss": 0.0722, "step": 82350 }, { "epoch": 39.51, "learning_rate": 1.4331327958076097e-05, "loss": 0.0661, "step": 82360 }, { "epoch": 39.52, "learning_rate": 1.432126737162477e-05, "loss": 0.0688, "step": 82370 }, { "epoch": 39.52, "learning_rate": 1.4311209727389197e-05, "loss": 0.0724, "step": 82380 }, { "epoch": 39.53, "learning_rate": 1.4301155026198781e-05, "loss": 0.0734, "step": 82390 }, { "epoch": 39.53, "learning_rate": 1.4291103268882677e-05, "loss": 0.0702, "step": 82400 }, { "epoch": 39.54, "learning_rate": 1.4281054456269788e-05, "loss": 0.0784, "step": 82410 }, { "epoch": 39.54, "learning_rate": 1.4271008589188777e-05, "loss": 0.0824, "step": 82420 }, { "epoch": 39.55, "learning_rate": 1.4260965668468096e-05, "loss": 0.0799, "step": 82430 }, { "epoch": 39.55, "learning_rate": 1.4250925694935879e-05, "loss": 0.0685, "step": 82440 }, { "epoch": 39.55, "learning_rate": 1.4240888669420094e-05, "loss": 0.0691, "step": 82450 }, { "epoch": 39.56, "learning_rate": 1.4230854592748427e-05, "loss": 0.0699, "step": 82460 }, { "epoch": 39.56, "learning_rate": 1.4220823465748334e-05, "loss": 0.0719, "step": 82470 }, { "epoch": 39.57, "learning_rate": 1.4210795289247019e-05, "loss": 0.0798, "step": 82480 }, { "epoch": 39.57, "learning_rate": 1.420077006407145e-05, "loss": 0.08, "step": 82490 }, { "epoch": 39.58, "learning_rate": 1.4190747791048359e-05, "loss": 0.0721, "step": 82500 }, { "epoch": 39.58, "learning_rate": 1.4180728471004201e-05, "loss": 0.0768, "step": 82510 }, { "epoch": 39.59, "learning_rate": 1.4170712104765215e-05, "loss": 0.0724, "step": 82520 }, { "epoch": 39.59, "learning_rate": 1.4160698693157404e-05, "loss": 0.0716, "step": 82530 }, { "epoch": 39.6, "learning_rate": 1.4150688237006509e-05, "loss": 0.0713, "step": 82540 }, { "epoch": 39.6, "learning_rate": 1.414068073713804e-05, "loss": 0.0698, "step": 82550 }, { "epoch": 39.6, "learning_rate": 1.4130676194377246e-05, "loss": 0.074, "step": 82560 }, { "epoch": 39.61, "learning_rate": 1.412067460954915e-05, "loss": 0.0838, "step": 82570 }, { "epoch": 39.61, "learning_rate": 1.4110675983478532e-05, "loss": 0.0775, "step": 82580 }, { "epoch": 39.62, "learning_rate": 1.41006803169899e-05, "loss": 0.0721, "step": 82590 }, { "epoch": 39.62, "learning_rate": 1.409068761090755e-05, "loss": 0.0713, "step": 82600 }, { "epoch": 39.63, "learning_rate": 1.4080697866055536e-05, "loss": 0.0701, "step": 82610 }, { "epoch": 39.63, "learning_rate": 1.4070711083257603e-05, "loss": 0.0828, "step": 82620 }, { "epoch": 39.64, "learning_rate": 1.4060727263337359e-05, "loss": 0.0683, "step": 82630 }, { "epoch": 39.64, "learning_rate": 1.4050746407118082e-05, "loss": 0.0745, "step": 82640 }, { "epoch": 39.65, "learning_rate": 1.4040768515422864e-05, "loss": 0.0687, "step": 82650 }, { "epoch": 39.65, "learning_rate": 1.4030793589074481e-05, "loss": 0.0689, "step": 82660 }, { "epoch": 39.65, "learning_rate": 1.4020821628895531e-05, "loss": 0.0772, "step": 82670 }, { "epoch": 39.66, "learning_rate": 1.401085263570835e-05, "loss": 0.0768, "step": 82680 }, { "epoch": 39.66, "learning_rate": 1.4000886610334979e-05, "loss": 0.0739, "step": 82690 }, { "epoch": 39.67, "learning_rate": 1.3990923553597312e-05, "loss": 0.071, "step": 82700 }, { "epoch": 39.67, "learning_rate": 1.3980963466316916e-05, "loss": 0.0727, "step": 82710 }, { "epoch": 39.68, "learning_rate": 1.3971006349315163e-05, "loss": 0.0779, "step": 82720 }, { "epoch": 39.68, "learning_rate": 1.3961052203413132e-05, "loss": 0.0778, "step": 82730 }, { "epoch": 39.69, "learning_rate": 1.395110102943169e-05, "loss": 0.0868, "step": 82740 }, { "epoch": 39.69, "learning_rate": 1.3941152828191467e-05, "loss": 0.0713, "step": 82750 }, { "epoch": 39.7, "learning_rate": 1.3931207600512807e-05, "loss": 0.0725, "step": 82760 }, { "epoch": 39.7, "learning_rate": 1.3921265347215833e-05, "loss": 0.0762, "step": 82770 }, { "epoch": 39.71, "learning_rate": 1.3911326069120456e-05, "loss": 0.0742, "step": 82780 }, { "epoch": 39.71, "learning_rate": 1.3901389767046302e-05, "loss": 0.0754, "step": 82790 }, { "epoch": 39.71, "learning_rate": 1.3891456441812745e-05, "loss": 0.0767, "step": 82800 }, { "epoch": 39.72, "learning_rate": 1.3881526094238928e-05, "loss": 0.0706, "step": 82810 }, { "epoch": 39.72, "learning_rate": 1.3871598725143758e-05, "loss": 0.0734, "step": 82820 }, { "epoch": 39.73, "learning_rate": 1.3861674335345897e-05, "loss": 0.0782, "step": 82830 }, { "epoch": 39.73, "learning_rate": 1.385175292566372e-05, "loss": 0.0706, "step": 82840 }, { "epoch": 39.74, "learning_rate": 1.384183449691539e-05, "loss": 0.0707, "step": 82850 }, { "epoch": 39.74, "learning_rate": 1.3831919049918873e-05, "loss": 0.0791, "step": 82860 }, { "epoch": 39.75, "learning_rate": 1.3822006585491781e-05, "loss": 0.0841, "step": 82870 }, { "epoch": 39.75, "learning_rate": 1.3812097104451554e-05, "loss": 0.0679, "step": 82880 }, { "epoch": 39.76, "learning_rate": 1.3802190607615379e-05, "loss": 0.0706, "step": 82890 }, { "epoch": 39.76, "learning_rate": 1.3792287095800193e-05, "loss": 0.0812, "step": 82900 }, { "epoch": 39.76, "learning_rate": 1.3782386569822647e-05, "loss": 0.0756, "step": 82910 }, { "epoch": 39.77, "learning_rate": 1.3772489030499207e-05, "loss": 0.0677, "step": 82920 }, { "epoch": 39.77, "learning_rate": 1.3762594478646051e-05, "loss": 0.0798, "step": 82930 }, { "epoch": 39.78, "learning_rate": 1.3752702915079136e-05, "loss": 0.0799, "step": 82940 }, { "epoch": 39.78, "learning_rate": 1.3742814340614152e-05, "loss": 0.0758, "step": 82950 }, { "epoch": 39.79, "learning_rate": 1.373292875606656e-05, "loss": 0.0744, "step": 82960 }, { "epoch": 39.79, "learning_rate": 1.3723046162251573e-05, "loss": 0.0727, "step": 82970 }, { "epoch": 39.8, "learning_rate": 1.3713166559984125e-05, "loss": 0.0814, "step": 82980 }, { "epoch": 39.8, "learning_rate": 1.3703289950078945e-05, "loss": 0.0663, "step": 82990 }, { "epoch": 39.81, "learning_rate": 1.3693416333350501e-05, "loss": 0.072, "step": 83000 }, { "epoch": 39.81, "learning_rate": 1.3683545710613006e-05, "loss": 0.0846, "step": 83010 }, { "epoch": 39.81, "learning_rate": 1.3673678082680435e-05, "loss": 0.081, "step": 83020 }, { "epoch": 39.82, "learning_rate": 1.3663813450366514e-05, "loss": 0.065, "step": 83030 }, { "epoch": 39.82, "learning_rate": 1.3653951814484733e-05, "loss": 0.0654, "step": 83040 }, { "epoch": 39.83, "learning_rate": 1.36440931758483e-05, "loss": 0.0762, "step": 83050 }, { "epoch": 39.83, "learning_rate": 1.3634237535270206e-05, "loss": 0.0838, "step": 83060 }, { "epoch": 39.84, "learning_rate": 1.3624384893563197e-05, "loss": 0.0796, "step": 83070 }, { "epoch": 39.84, "learning_rate": 1.3614535251539756e-05, "loss": 0.0691, "step": 83080 }, { "epoch": 39.85, "learning_rate": 1.3604688610012123e-05, "loss": 0.0744, "step": 83090 }, { "epoch": 39.85, "learning_rate": 1.3594844969792303e-05, "loss": 0.0771, "step": 83100 }, { "epoch": 39.86, "learning_rate": 1.3585004331692048e-05, "loss": 0.0635, "step": 83110 }, { "epoch": 39.86, "learning_rate": 1.3575166696522835e-05, "loss": 0.0728, "step": 83120 }, { "epoch": 39.87, "learning_rate": 1.356533206509592e-05, "loss": 0.0723, "step": 83130 }, { "epoch": 39.87, "learning_rate": 1.3555500438222323e-05, "loss": 0.0689, "step": 83140 }, { "epoch": 39.87, "learning_rate": 1.354567181671279e-05, "loss": 0.081, "step": 83150 }, { "epoch": 39.88, "learning_rate": 1.3535846201377834e-05, "loss": 0.0776, "step": 83160 }, { "epoch": 39.88, "learning_rate": 1.3526023593027709e-05, "loss": 0.08, "step": 83170 }, { "epoch": 39.89, "learning_rate": 1.3516203992472437e-05, "loss": 0.0713, "step": 83180 }, { "epoch": 39.89, "learning_rate": 1.3506387400521792e-05, "loss": 0.0875, "step": 83190 }, { "epoch": 39.9, "learning_rate": 1.3496573817985264e-05, "loss": 0.08, "step": 83200 }, { "epoch": 39.9, "learning_rate": 1.3486763245672135e-05, "loss": 0.0756, "step": 83210 }, { "epoch": 39.91, "learning_rate": 1.3476955684391429e-05, "loss": 0.0791, "step": 83220 }, { "epoch": 39.91, "learning_rate": 1.3467151134951911e-05, "loss": 0.0702, "step": 83230 }, { "epoch": 39.92, "learning_rate": 1.3457349598162106e-05, "loss": 0.0743, "step": 83240 }, { "epoch": 39.92, "learning_rate": 1.3447551074830295e-05, "loss": 0.0731, "step": 83250 }, { "epoch": 39.92, "learning_rate": 1.3437755565764514e-05, "loss": 0.0735, "step": 83260 }, { "epoch": 39.93, "learning_rate": 1.3427963071772515e-05, "loss": 0.0747, "step": 83270 }, { "epoch": 39.93, "learning_rate": 1.3418173593661843e-05, "loss": 0.0736, "step": 83280 }, { "epoch": 39.94, "learning_rate": 1.3408387132239774e-05, "loss": 0.0831, "step": 83290 }, { "epoch": 39.94, "learning_rate": 1.3398603688313344e-05, "loss": 0.0686, "step": 83300 }, { "epoch": 39.95, "learning_rate": 1.3388823262689331e-05, "loss": 0.0722, "step": 83310 }, { "epoch": 39.95, "learning_rate": 1.3379045856174277e-05, "loss": 0.0971, "step": 83320 }, { "epoch": 40.0, "learning_rate": 1.336927146957448e-05, "loss": 0.0715, "step": 83330 }, { "epoch": 40.01, "learning_rate": 1.3359500103695943e-05, "loss": 0.0705, "step": 83340 }, { "epoch": 40.01, "learning_rate": 1.334973175934447e-05, "loss": 0.0642, "step": 83350 }, { "epoch": 40.02, "learning_rate": 1.3339966437325602e-05, "loss": 0.0845, "step": 83360 }, { "epoch": 40.02, "learning_rate": 1.3330204138444624e-05, "loss": 0.0656, "step": 83370 }, { "epoch": 40.03, "learning_rate": 1.3320444863506577e-05, "loss": 0.0764, "step": 83380 }, { "epoch": 40.03, "learning_rate": 1.3310688613316247e-05, "loss": 0.0773, "step": 83390 }, { "epoch": 40.04, "learning_rate": 1.3300935388678199e-05, "loss": 0.0674, "step": 83400 }, { "epoch": 40.04, "learning_rate": 1.3291185190396682e-05, "loss": 0.0745, "step": 83410 }, { "epoch": 40.05, "learning_rate": 1.3281438019275766e-05, "loss": 0.0689, "step": 83420 }, { "epoch": 40.05, "learning_rate": 1.3271693876119235e-05, "loss": 0.0735, "step": 83430 }, { "epoch": 40.05, "learning_rate": 1.3261952761730644e-05, "loss": 0.0725, "step": 83440 }, { "epoch": 40.06, "learning_rate": 1.3252214676913249e-05, "loss": 0.0738, "step": 83450 }, { "epoch": 40.06, "learning_rate": 1.3242479622470127e-05, "loss": 0.0849, "step": 83460 }, { "epoch": 40.07, "learning_rate": 1.3232747599204082e-05, "loss": 0.0684, "step": 83470 }, { "epoch": 40.07, "learning_rate": 1.322301860791762e-05, "loss": 0.0735, "step": 83480 }, { "epoch": 40.08, "learning_rate": 1.321329264941305e-05, "loss": 0.069, "step": 83490 }, { "epoch": 40.08, "learning_rate": 1.3203569724492415e-05, "loss": 0.0726, "step": 83500 }, { "epoch": 40.09, "learning_rate": 1.3193849833957528e-05, "loss": 0.0723, "step": 83510 }, { "epoch": 40.09, "learning_rate": 1.3184132978609892e-05, "loss": 0.0741, "step": 83520 }, { "epoch": 40.1, "learning_rate": 1.31744191592508e-05, "loss": 0.0686, "step": 83530 }, { "epoch": 40.1, "learning_rate": 1.3164708376681345e-05, "loss": 0.0801, "step": 83540 }, { "epoch": 40.11, "learning_rate": 1.3155000631702275e-05, "loss": 0.0741, "step": 83550 }, { "epoch": 40.11, "learning_rate": 1.3145295925114132e-05, "loss": 0.0691, "step": 83560 }, { "epoch": 40.11, "learning_rate": 1.3135594257717222e-05, "loss": 0.0717, "step": 83570 }, { "epoch": 40.12, "learning_rate": 1.312589563031159e-05, "loss": 0.0805, "step": 83580 }, { "epoch": 40.12, "learning_rate": 1.3116200043696996e-05, "loss": 0.0782, "step": 83590 }, { "epoch": 40.13, "learning_rate": 1.3106507498672999e-05, "loss": 0.0771, "step": 83600 }, { "epoch": 40.13, "learning_rate": 1.3096817996038863e-05, "loss": 0.0784, "step": 83610 }, { "epoch": 40.14, "learning_rate": 1.3087131536593678e-05, "loss": 0.0744, "step": 83620 }, { "epoch": 40.14, "learning_rate": 1.3077448121136177e-05, "loss": 0.0743, "step": 83630 }, { "epoch": 40.15, "learning_rate": 1.3067767750464915e-05, "loss": 0.0699, "step": 83640 }, { "epoch": 40.15, "learning_rate": 1.3058090425378183e-05, "loss": 0.0824, "step": 83650 }, { "epoch": 40.16, "learning_rate": 1.3048416146673991e-05, "loss": 0.0774, "step": 83660 }, { "epoch": 40.16, "learning_rate": 1.3038744915150136e-05, "loss": 0.0736, "step": 83670 }, { "epoch": 40.16, "learning_rate": 1.302907673160414e-05, "loss": 0.0747, "step": 83680 }, { "epoch": 40.17, "learning_rate": 1.3019411596833287e-05, "loss": 0.0754, "step": 83690 }, { "epoch": 40.17, "learning_rate": 1.30097495116346e-05, "loss": 0.0755, "step": 83700 }, { "epoch": 40.18, "learning_rate": 1.3000090476804855e-05, "loss": 0.084, "step": 83710 }, { "epoch": 40.18, "learning_rate": 1.2990434493140596e-05, "loss": 0.0754, "step": 83720 }, { "epoch": 40.19, "learning_rate": 1.298078156143806e-05, "loss": 0.079, "step": 83730 }, { "epoch": 40.19, "learning_rate": 1.2971131682493287e-05, "loss": 0.0774, "step": 83740 }, { "epoch": 40.2, "learning_rate": 1.296148485710204e-05, "loss": 0.0672, "step": 83750 }, { "epoch": 40.2, "learning_rate": 1.295184108605984e-05, "loss": 0.0687, "step": 83760 }, { "epoch": 40.21, "learning_rate": 1.2942200370161956e-05, "loss": 0.0664, "step": 83770 }, { "epoch": 40.21, "learning_rate": 1.2932562710203394e-05, "loss": 0.0758, "step": 83780 }, { "epoch": 40.21, "learning_rate": 1.292292810697892e-05, "loss": 0.0655, "step": 83790 }, { "epoch": 40.22, "learning_rate": 1.2913296561283057e-05, "loss": 0.0734, "step": 83800 }, { "epoch": 40.22, "learning_rate": 1.290366807391003e-05, "loss": 0.0775, "step": 83810 }, { "epoch": 40.23, "learning_rate": 1.289404264565386e-05, "loss": 0.0675, "step": 83820 }, { "epoch": 40.23, "learning_rate": 1.2884420277308302e-05, "loss": 0.0722, "step": 83830 }, { "epoch": 40.24, "learning_rate": 1.287480096966685e-05, "loss": 0.0701, "step": 83840 }, { "epoch": 40.24, "learning_rate": 1.2865184723522761e-05, "loss": 0.0666, "step": 83850 }, { "epoch": 40.25, "learning_rate": 1.2855571539669021e-05, "loss": 0.0739, "step": 83860 }, { "epoch": 40.25, "learning_rate": 1.2845961418898394e-05, "loss": 0.0715, "step": 83870 }, { "epoch": 40.26, "learning_rate": 1.2836354362003344e-05, "loss": 0.0675, "step": 83880 }, { "epoch": 40.26, "learning_rate": 1.2826750369776113e-05, "loss": 0.0638, "step": 83890 }, { "epoch": 40.27, "learning_rate": 1.2817149443008691e-05, "loss": 0.0695, "step": 83900 }, { "epoch": 40.27, "learning_rate": 1.2807551582492816e-05, "loss": 0.0732, "step": 83910 }, { "epoch": 40.27, "learning_rate": 1.2797956789019955e-05, "loss": 0.0696, "step": 83920 }, { "epoch": 40.28, "learning_rate": 1.2788365063381346e-05, "loss": 0.0795, "step": 83930 }, { "epoch": 40.28, "learning_rate": 1.2778776406367971e-05, "loss": 0.0725, "step": 83940 }, { "epoch": 40.29, "learning_rate": 1.2769190818770521e-05, "loss": 0.0851, "step": 83950 }, { "epoch": 40.29, "learning_rate": 1.2759608301379477e-05, "loss": 0.0726, "step": 83960 }, { "epoch": 40.3, "learning_rate": 1.2750028854985058e-05, "loss": 0.0735, "step": 83970 }, { "epoch": 40.3, "learning_rate": 1.274045248037722e-05, "loss": 0.0691, "step": 83980 }, { "epoch": 40.31, "learning_rate": 1.2730879178345673e-05, "loss": 0.0723, "step": 83990 }, { "epoch": 40.31, "learning_rate": 1.2721308949679866e-05, "loss": 0.0664, "step": 84000 }, { "epoch": 40.32, "learning_rate": 1.2711741795169024e-05, "loss": 0.0682, "step": 84010 }, { "epoch": 40.32, "learning_rate": 1.2702177715602054e-05, "loss": 0.0665, "step": 84020 }, { "epoch": 40.32, "learning_rate": 1.2692616711767674e-05, "loss": 0.0648, "step": 84030 }, { "epoch": 40.33, "learning_rate": 1.2683058784454316e-05, "loss": 0.0774, "step": 84040 }, { "epoch": 40.33, "learning_rate": 1.2673503934450171e-05, "loss": 0.0732, "step": 84050 }, { "epoch": 40.34, "learning_rate": 1.266395216254317e-05, "loss": 0.0736, "step": 84060 }, { "epoch": 40.34, "learning_rate": 1.2654403469520986e-05, "loss": 0.069, "step": 84070 }, { "epoch": 40.35, "learning_rate": 1.2644857856171066e-05, "loss": 0.0804, "step": 84080 }, { "epoch": 40.35, "learning_rate": 1.2635315323280555e-05, "loss": 0.0733, "step": 84090 }, { "epoch": 40.36, "learning_rate": 1.2625775871636375e-05, "loss": 0.0743, "step": 84100 }, { "epoch": 40.36, "learning_rate": 1.261623950202519e-05, "loss": 0.0776, "step": 84110 }, { "epoch": 40.37, "learning_rate": 1.2606706215233427e-05, "loss": 0.0769, "step": 84120 }, { "epoch": 40.37, "learning_rate": 1.259717601204719e-05, "loss": 0.0698, "step": 84130 }, { "epoch": 40.37, "learning_rate": 1.2587648893252434e-05, "loss": 0.0754, "step": 84140 }, { "epoch": 40.38, "learning_rate": 1.2578124859634793e-05, "loss": 0.0719, "step": 84150 }, { "epoch": 40.38, "learning_rate": 1.2568603911979637e-05, "loss": 0.0719, "step": 84160 }, { "epoch": 40.39, "learning_rate": 1.2559086051072116e-05, "loss": 0.0645, "step": 84170 }, { "epoch": 40.39, "learning_rate": 1.2549571277697109e-05, "loss": 0.0834, "step": 84180 }, { "epoch": 40.4, "learning_rate": 1.2540059592639258e-05, "loss": 0.0698, "step": 84190 }, { "epoch": 40.4, "learning_rate": 1.2530550996682905e-05, "loss": 0.0792, "step": 84200 }, { "epoch": 40.41, "learning_rate": 1.2521045490612177e-05, "loss": 0.0686, "step": 84210 }, { "epoch": 40.41, "learning_rate": 1.2511543075210957e-05, "loss": 0.0733, "step": 84220 }, { "epoch": 40.42, "learning_rate": 1.250204375126286e-05, "loss": 0.0686, "step": 84230 }, { "epoch": 40.42, "learning_rate": 1.2492547519551207e-05, "loss": 0.068, "step": 84240 }, { "epoch": 40.43, "learning_rate": 1.248305438085911e-05, "loss": 0.0732, "step": 84250 }, { "epoch": 40.43, "learning_rate": 1.2473564335969429e-05, "loss": 0.0712, "step": 84260 }, { "epoch": 40.43, "learning_rate": 1.2464077385664725e-05, "loss": 0.069, "step": 84270 }, { "epoch": 40.44, "learning_rate": 1.2454593530727338e-05, "loss": 0.0785, "step": 84280 }, { "epoch": 40.44, "learning_rate": 1.2445112771939343e-05, "loss": 0.0736, "step": 84290 }, { "epoch": 40.45, "learning_rate": 1.24356351100826e-05, "loss": 0.0736, "step": 84300 }, { "epoch": 40.45, "learning_rate": 1.2426160545938625e-05, "loss": 0.0719, "step": 84310 }, { "epoch": 40.46, "learning_rate": 1.2416689080288763e-05, "loss": 0.0687, "step": 84320 }, { "epoch": 40.46, "learning_rate": 1.2407220713914064e-05, "loss": 0.0816, "step": 84330 }, { "epoch": 40.47, "learning_rate": 1.2397755447595317e-05, "loss": 0.0712, "step": 84340 }, { "epoch": 40.47, "learning_rate": 1.2388293282113067e-05, "loss": 0.0693, "step": 84350 }, { "epoch": 40.48, "learning_rate": 1.2378834218247614e-05, "loss": 0.0772, "step": 84360 }, { "epoch": 40.48, "learning_rate": 1.2369378256778985e-05, "loss": 0.0695, "step": 84370 }, { "epoch": 40.48, "learning_rate": 1.235992539848696e-05, "loss": 0.0819, "step": 84380 }, { "epoch": 40.49, "learning_rate": 1.2350475644151067e-05, "loss": 0.0769, "step": 84390 }, { "epoch": 40.49, "learning_rate": 1.2341028994550558e-05, "loss": 0.0792, "step": 84400 }, { "epoch": 40.5, "learning_rate": 1.2331585450464472e-05, "loss": 0.07, "step": 84410 }, { "epoch": 40.5, "learning_rate": 1.2322145012671521e-05, "loss": 0.0751, "step": 84420 }, { "epoch": 40.51, "learning_rate": 1.2312707681950225e-05, "loss": 0.0778, "step": 84430 }, { "epoch": 40.51, "learning_rate": 1.2303273459078824e-05, "loss": 0.0718, "step": 84440 }, { "epoch": 40.52, "learning_rate": 1.22938423448353e-05, "loss": 0.0747, "step": 84450 }, { "epoch": 40.52, "learning_rate": 1.228441433999738e-05, "loss": 0.065, "step": 84460 }, { "epoch": 40.53, "learning_rate": 1.2274989445342544e-05, "loss": 0.0717, "step": 84470 }, { "epoch": 40.53, "learning_rate": 1.226556766164802e-05, "loss": 0.0889, "step": 84480 }, { "epoch": 40.53, "learning_rate": 1.2256148989690735e-05, "loss": 0.0777, "step": 84490 }, { "epoch": 40.54, "learning_rate": 1.2246733430247409e-05, "loss": 0.0653, "step": 84500 }, { "epoch": 40.54, "learning_rate": 1.2237320984094481e-05, "loss": 0.082, "step": 84510 }, { "epoch": 40.55, "learning_rate": 1.222791165200815e-05, "loss": 0.0711, "step": 84520 }, { "epoch": 40.55, "learning_rate": 1.2218505434764343e-05, "loss": 0.0709, "step": 84530 }, { "epoch": 40.56, "learning_rate": 1.220910233313874e-05, "loss": 0.0692, "step": 84540 }, { "epoch": 40.56, "learning_rate": 1.2199702347906767e-05, "loss": 0.0822, "step": 84550 }, { "epoch": 40.57, "learning_rate": 1.2190305479843561e-05, "loss": 0.0802, "step": 84560 }, { "epoch": 40.57, "learning_rate": 1.2180911729724043e-05, "loss": 0.0624, "step": 84570 }, { "epoch": 40.58, "learning_rate": 1.2171521098322852e-05, "loss": 0.0801, "step": 84580 }, { "epoch": 40.58, "learning_rate": 1.2162133586414388e-05, "loss": 0.0684, "step": 84590 }, { "epoch": 40.59, "learning_rate": 1.2152749194772784e-05, "loss": 0.0789, "step": 84600 }, { "epoch": 40.59, "learning_rate": 1.2143367924171905e-05, "loss": 0.0764, "step": 84610 }, { "epoch": 40.59, "learning_rate": 1.2133989775385396e-05, "loss": 0.077, "step": 84620 }, { "epoch": 40.6, "learning_rate": 1.212461474918658e-05, "loss": 0.072, "step": 84630 }, { "epoch": 40.6, "learning_rate": 1.2115242846348579e-05, "loss": 0.0719, "step": 84640 }, { "epoch": 40.61, "learning_rate": 1.210587406764424e-05, "loss": 0.0724, "step": 84650 }, { "epoch": 40.61, "learning_rate": 1.2096508413846153e-05, "loss": 0.0662, "step": 84660 }, { "epoch": 40.62, "learning_rate": 1.2087145885726637e-05, "loss": 0.0668, "step": 84670 }, { "epoch": 40.62, "learning_rate": 1.2077786484057779e-05, "loss": 0.0712, "step": 84680 }, { "epoch": 40.63, "learning_rate": 1.2068430209611403e-05, "loss": 0.0836, "step": 84690 }, { "epoch": 40.63, "learning_rate": 1.2059077063159035e-05, "loss": 0.0695, "step": 84700 }, { "epoch": 40.64, "learning_rate": 1.2049727045471986e-05, "loss": 0.0757, "step": 84710 }, { "epoch": 40.64, "learning_rate": 1.2040380157321307e-05, "loss": 0.0716, "step": 84720 }, { "epoch": 40.64, "learning_rate": 1.2031036399477774e-05, "loss": 0.0809, "step": 84730 }, { "epoch": 40.65, "learning_rate": 1.202169577271191e-05, "loss": 0.0762, "step": 84740 }, { "epoch": 40.65, "learning_rate": 1.201235827779399e-05, "loss": 0.0766, "step": 84750 }, { "epoch": 40.66, "learning_rate": 1.2003023915494027e-05, "loss": 0.071, "step": 84760 }, { "epoch": 40.66, "learning_rate": 1.1993692686581747e-05, "loss": 0.0738, "step": 84770 }, { "epoch": 40.67, "learning_rate": 1.1984364591826658e-05, "loss": 0.0791, "step": 84780 }, { "epoch": 40.67, "learning_rate": 1.1975039631997992e-05, "loss": 0.0653, "step": 84790 }, { "epoch": 40.68, "learning_rate": 1.196571780786474e-05, "loss": 0.0775, "step": 84800 }, { "epoch": 40.68, "learning_rate": 1.1956399120195566e-05, "loss": 0.077, "step": 84810 }, { "epoch": 40.69, "learning_rate": 1.1947083569758982e-05, "loss": 0.0634, "step": 84820 }, { "epoch": 40.69, "learning_rate": 1.1937771157323163e-05, "loss": 0.0721, "step": 84830 }, { "epoch": 40.69, "learning_rate": 1.1928461883656073e-05, "loss": 0.0756, "step": 84840 }, { "epoch": 40.7, "learning_rate": 1.1919155749525357e-05, "loss": 0.0695, "step": 84850 }, { "epoch": 40.7, "learning_rate": 1.1909852755698453e-05, "loss": 0.0636, "step": 84860 }, { "epoch": 40.71, "learning_rate": 1.190055290294254e-05, "loss": 0.0713, "step": 84870 }, { "epoch": 40.71, "learning_rate": 1.189125619202449e-05, "loss": 0.0726, "step": 84880 }, { "epoch": 40.72, "learning_rate": 1.1881962623710957e-05, "loss": 0.068, "step": 84890 }, { "epoch": 40.72, "learning_rate": 1.1872672198768347e-05, "loss": 0.0648, "step": 84900 }, { "epoch": 40.73, "learning_rate": 1.1863384917962794e-05, "loss": 0.0724, "step": 84910 }, { "epoch": 40.73, "learning_rate": 1.1854100782060129e-05, "loss": 0.0755, "step": 84920 }, { "epoch": 40.74, "learning_rate": 1.1844819791825978e-05, "loss": 0.06, "step": 84930 }, { "epoch": 40.74, "learning_rate": 1.1835541948025708e-05, "loss": 0.0675, "step": 84940 }, { "epoch": 40.74, "learning_rate": 1.1826267251424372e-05, "loss": 0.0739, "step": 84950 }, { "epoch": 40.75, "learning_rate": 1.181699570278681e-05, "loss": 0.075, "step": 84960 }, { "epoch": 40.75, "learning_rate": 1.1807727302877609e-05, "loss": 0.0669, "step": 84970 }, { "epoch": 40.76, "learning_rate": 1.179846205246109e-05, "loss": 0.0717, "step": 84980 }, { "epoch": 40.76, "learning_rate": 1.1789199952301266e-05, "loss": 0.0842, "step": 84990 }, { "epoch": 40.77, "learning_rate": 1.1779941003161955e-05, "loss": 0.0727, "step": 85000 }, { "epoch": 40.77, "learning_rate": 1.177068520580667e-05, "loss": 0.0697, "step": 85010 }, { "epoch": 40.78, "learning_rate": 1.1761432560998715e-05, "loss": 0.0729, "step": 85020 }, { "epoch": 40.78, "learning_rate": 1.175218306950106e-05, "loss": 0.0789, "step": 85030 }, { "epoch": 40.79, "learning_rate": 1.1742936732076466e-05, "loss": 0.0696, "step": 85040 }, { "epoch": 40.79, "learning_rate": 1.1733693549487457e-05, "loss": 0.0749, "step": 85050 }, { "epoch": 40.8, "learning_rate": 1.1724453522496232e-05, "loss": 0.0712, "step": 85060 }, { "epoch": 40.8, "learning_rate": 1.1715216651864763e-05, "loss": 0.0816, "step": 85070 }, { "epoch": 40.8, "learning_rate": 1.170598293835477e-05, "loss": 0.0709, "step": 85080 }, { "epoch": 40.81, "learning_rate": 1.1696752382727722e-05, "loss": 0.0682, "step": 85090 }, { "epoch": 40.81, "learning_rate": 1.1687524985744764e-05, "loss": 0.0698, "step": 85100 }, { "epoch": 40.82, "learning_rate": 1.1678300748166853e-05, "loss": 0.0826, "step": 85110 }, { "epoch": 40.82, "learning_rate": 1.1669079670754656e-05, "loss": 0.074, "step": 85120 }, { "epoch": 40.83, "learning_rate": 1.1659861754268575e-05, "loss": 0.0721, "step": 85130 }, { "epoch": 40.83, "learning_rate": 1.1650646999468766e-05, "loss": 0.0828, "step": 85140 }, { "epoch": 40.84, "learning_rate": 1.164143540711511e-05, "loss": 0.0758, "step": 85150 }, { "epoch": 40.84, "learning_rate": 1.1632226977967247e-05, "loss": 0.0743, "step": 85160 }, { "epoch": 40.85, "learning_rate": 1.1623021712784515e-05, "loss": 0.0707, "step": 85170 }, { "epoch": 40.85, "learning_rate": 1.1613819612326032e-05, "loss": 0.0703, "step": 85180 }, { "epoch": 40.85, "learning_rate": 1.1604620677350647e-05, "loss": 0.0723, "step": 85190 }, { "epoch": 40.86, "learning_rate": 1.1595424908616931e-05, "loss": 0.0717, "step": 85200 }, { "epoch": 40.86, "learning_rate": 1.1586232306883215e-05, "loss": 0.0741, "step": 85210 }, { "epoch": 40.87, "learning_rate": 1.1577042872907557e-05, "loss": 0.0725, "step": 85220 }, { "epoch": 40.87, "learning_rate": 1.1567856607447763e-05, "loss": 0.0723, "step": 85230 }, { "epoch": 40.88, "learning_rate": 1.1558673511261348e-05, "loss": 0.0799, "step": 85240 }, { "epoch": 40.88, "learning_rate": 1.1549493585105602e-05, "loss": 0.066, "step": 85250 }, { "epoch": 40.89, "learning_rate": 1.1540316829737541e-05, "loss": 0.0693, "step": 85260 }, { "epoch": 40.89, "learning_rate": 1.1531143245913912e-05, "loss": 0.0832, "step": 85270 }, { "epoch": 40.9, "learning_rate": 1.1521972834391215e-05, "loss": 0.0832, "step": 85280 }, { "epoch": 40.9, "learning_rate": 1.1512805595925669e-05, "loss": 0.0775, "step": 85290 }, { "epoch": 40.9, "learning_rate": 1.1503641531273267e-05, "loss": 0.0722, "step": 85300 }, { "epoch": 40.91, "learning_rate": 1.1494480641189687e-05, "loss": 0.0694, "step": 85310 }, { "epoch": 40.91, "learning_rate": 1.148532292643038e-05, "loss": 0.0727, "step": 85320 }, { "epoch": 40.92, "learning_rate": 1.1476168387750536e-05, "loss": 0.0729, "step": 85330 }, { "epoch": 40.92, "learning_rate": 1.1467017025905075e-05, "loss": 0.0722, "step": 85340 }, { "epoch": 40.93, "learning_rate": 1.1457868841648655e-05, "loss": 0.0754, "step": 85350 }, { "epoch": 40.93, "learning_rate": 1.144872383573567e-05, "loss": 0.0901, "step": 85360 }, { "epoch": 40.94, "learning_rate": 1.1439582008920257e-05, "loss": 0.0784, "step": 85370 }, { "epoch": 40.94, "learning_rate": 1.1430443361956312e-05, "loss": 0.072, "step": 85380 }, { "epoch": 40.95, "learning_rate": 1.14213078955974e-05, "loss": 0.0782, "step": 85390 }, { "epoch": 40.95, "learning_rate": 1.1412175610596897e-05, "loss": 0.0761, "step": 85400 }, { "epoch": 41.0, "learning_rate": 1.1403046507707883e-05, "loss": 0.0767, "step": 85410 }, { "epoch": 41.01, "learning_rate": 1.1393920587683188e-05, "loss": 0.0672, "step": 85420 }, { "epoch": 41.01, "learning_rate": 1.1384797851275364e-05, "loss": 0.0677, "step": 85430 }, { "epoch": 41.02, "learning_rate": 1.1375678299236715e-05, "loss": 0.076, "step": 85440 }, { "epoch": 41.02, "learning_rate": 1.1366561932319286e-05, "loss": 0.0683, "step": 85450 }, { "epoch": 41.03, "learning_rate": 1.135744875127483e-05, "loss": 0.0729, "step": 85460 }, { "epoch": 41.03, "learning_rate": 1.1348338756854865e-05, "loss": 0.066, "step": 85470 }, { "epoch": 41.04, "learning_rate": 1.1339231949810657e-05, "loss": 0.0785, "step": 85480 }, { "epoch": 41.04, "learning_rate": 1.1330128330893153e-05, "loss": 0.0644, "step": 85490 }, { "epoch": 41.04, "learning_rate": 1.1321027900853103e-05, "loss": 0.0685, "step": 85500 }, { "epoch": 41.05, "learning_rate": 1.1311930660440967e-05, "loss": 0.0734, "step": 85510 }, { "epoch": 41.05, "learning_rate": 1.1302836610406941e-05, "loss": 0.0703, "step": 85520 }, { "epoch": 41.06, "learning_rate": 1.1293745751500944e-05, "loss": 0.0767, "step": 85530 }, { "epoch": 41.06, "learning_rate": 1.1284658084472655e-05, "loss": 0.0713, "step": 85540 }, { "epoch": 41.07, "learning_rate": 1.1275573610071488e-05, "loss": 0.0765, "step": 85550 }, { "epoch": 41.07, "learning_rate": 1.1266492329046551e-05, "loss": 0.0723, "step": 85560 }, { "epoch": 41.08, "learning_rate": 1.1257414242146763e-05, "loss": 0.0719, "step": 85570 }, { "epoch": 41.08, "learning_rate": 1.1248339350120729e-05, "loss": 0.0718, "step": 85580 }, { "epoch": 41.09, "learning_rate": 1.1239267653716812e-05, "loss": 0.0655, "step": 85590 }, { "epoch": 41.09, "learning_rate": 1.123019915368308e-05, "loss": 0.0822, "step": 85600 }, { "epoch": 41.09, "learning_rate": 1.1221133850767368e-05, "loss": 0.072, "step": 85610 }, { "epoch": 41.1, "learning_rate": 1.121207174571724e-05, "loss": 0.0713, "step": 85620 }, { "epoch": 41.1, "learning_rate": 1.1203012839280008e-05, "loss": 0.0723, "step": 85630 }, { "epoch": 41.11, "learning_rate": 1.1193957132202665e-05, "loss": 0.0675, "step": 85640 }, { "epoch": 41.11, "learning_rate": 1.1184904625232028e-05, "loss": 0.0762, "step": 85650 }, { "epoch": 41.12, "learning_rate": 1.1175855319114597e-05, "loss": 0.0686, "step": 85660 }, { "epoch": 41.12, "learning_rate": 1.1166809214596591e-05, "loss": 0.0741, "step": 85670 }, { "epoch": 41.13, "learning_rate": 1.1157766312424007e-05, "loss": 0.0707, "step": 85680 }, { "epoch": 41.13, "learning_rate": 1.1148726613342547e-05, "loss": 0.0774, "step": 85690 }, { "epoch": 41.14, "learning_rate": 1.113969011809769e-05, "loss": 0.0818, "step": 85700 }, { "epoch": 41.14, "learning_rate": 1.1130656827434593e-05, "loss": 0.0736, "step": 85710 }, { "epoch": 41.14, "learning_rate": 1.1121626742098174e-05, "loss": 0.0721, "step": 85720 }, { "epoch": 41.15, "learning_rate": 1.1112599862833129e-05, "loss": 0.0736, "step": 85730 }, { "epoch": 41.15, "learning_rate": 1.1103576190383819e-05, "loss": 0.0721, "step": 85740 }, { "epoch": 41.16, "learning_rate": 1.1094555725494382e-05, "loss": 0.0706, "step": 85750 }, { "epoch": 41.16, "learning_rate": 1.1085538468908684e-05, "loss": 0.0702, "step": 85760 }, { "epoch": 41.17, "learning_rate": 1.1076524421370338e-05, "loss": 0.0838, "step": 85770 }, { "epoch": 41.17, "learning_rate": 1.1067513583622657e-05, "loss": 0.0746, "step": 85780 }, { "epoch": 41.18, "learning_rate": 1.105850595640872e-05, "loss": 0.0762, "step": 85790 }, { "epoch": 41.18, "learning_rate": 1.1049501540471325e-05, "loss": 0.0767, "step": 85800 }, { "epoch": 41.19, "learning_rate": 1.1040500336553045e-05, "loss": 0.0713, "step": 85810 }, { "epoch": 41.19, "learning_rate": 1.1031502345396122e-05, "loss": 0.0698, "step": 85820 }, { "epoch": 41.2, "learning_rate": 1.1022507567742585e-05, "loss": 0.0728, "step": 85830 }, { "epoch": 41.2, "learning_rate": 1.1013516004334184e-05, "loss": 0.0691, "step": 85840 }, { "epoch": 41.2, "learning_rate": 1.1004527655912384e-05, "loss": 0.0704, "step": 85850 }, { "epoch": 41.21, "learning_rate": 1.0995542523218411e-05, "loss": 0.0771, "step": 85860 }, { "epoch": 41.21, "learning_rate": 1.0986560606993207e-05, "loss": 0.0668, "step": 85870 }, { "epoch": 41.22, "learning_rate": 1.0977581907977474e-05, "loss": 0.0765, "step": 85880 }, { "epoch": 41.22, "learning_rate": 1.096860642691162e-05, "loss": 0.0694, "step": 85890 }, { "epoch": 41.23, "learning_rate": 1.0959634164535808e-05, "loss": 0.071, "step": 85900 }, { "epoch": 41.23, "learning_rate": 1.0950665121589937e-05, "loss": 0.0582, "step": 85910 }, { "epoch": 41.24, "learning_rate": 1.0941699298813606e-05, "loss": 0.0611, "step": 85920 }, { "epoch": 41.24, "learning_rate": 1.0932736696946183e-05, "loss": 0.0742, "step": 85930 }, { "epoch": 41.25, "learning_rate": 1.0923777316726768e-05, "loss": 0.0771, "step": 85940 }, { "epoch": 41.25, "learning_rate": 1.0914821158894189e-05, "loss": 0.0656, "step": 85950 }, { "epoch": 41.25, "learning_rate": 1.0905868224187e-05, "loss": 0.072, "step": 85960 }, { "epoch": 41.26, "learning_rate": 1.0896918513343501e-05, "loss": 0.0734, "step": 85970 }, { "epoch": 41.26, "learning_rate": 1.0887972027101729e-05, "loss": 0.0706, "step": 85980 }, { "epoch": 41.27, "learning_rate": 1.087902876619945e-05, "loss": 0.0673, "step": 85990 }, { "epoch": 41.27, "learning_rate": 1.0870088731374139e-05, "loss": 0.0657, "step": 86000 }, { "epoch": 41.28, "learning_rate": 1.086115192336305e-05, "loss": 0.0721, "step": 86010 }, { "epoch": 41.28, "learning_rate": 1.0852218342903137e-05, "loss": 0.0743, "step": 86020 }, { "epoch": 41.29, "learning_rate": 1.0843287990731105e-05, "loss": 0.0726, "step": 86030 }, { "epoch": 41.29, "learning_rate": 1.0834360867583388e-05, "loss": 0.0724, "step": 86040 }, { "epoch": 41.3, "learning_rate": 1.0825436974196157e-05, "loss": 0.0812, "step": 86050 }, { "epoch": 41.3, "learning_rate": 1.0816516311305324e-05, "loss": 0.0755, "step": 86060 }, { "epoch": 41.3, "learning_rate": 1.0807598879646491e-05, "loss": 0.0667, "step": 86070 }, { "epoch": 41.31, "learning_rate": 1.0798684679955046e-05, "loss": 0.0663, "step": 86080 }, { "epoch": 41.31, "learning_rate": 1.0789773712966094e-05, "loss": 0.0651, "step": 86090 }, { "epoch": 41.32, "learning_rate": 1.0780865979414462e-05, "loss": 0.0758, "step": 86100 }, { "epoch": 41.32, "learning_rate": 1.077196148003472e-05, "loss": 0.0702, "step": 86110 }, { "epoch": 41.33, "learning_rate": 1.0763060215561177e-05, "loss": 0.0764, "step": 86120 }, { "epoch": 41.33, "learning_rate": 1.0754162186727873e-05, "loss": 0.0698, "step": 86130 }, { "epoch": 41.34, "learning_rate": 1.0745267394268555e-05, "loss": 0.082, "step": 86140 }, { "epoch": 41.34, "learning_rate": 1.0736375838916734e-05, "loss": 0.0706, "step": 86150 }, { "epoch": 41.35, "learning_rate": 1.0727487521405648e-05, "loss": 0.073, "step": 86160 }, { "epoch": 41.35, "learning_rate": 1.0718602442468261e-05, "loss": 0.073, "step": 86170 }, { "epoch": 41.36, "learning_rate": 1.0709720602837275e-05, "loss": 0.0752, "step": 86180 }, { "epoch": 41.36, "learning_rate": 1.0700842003245121e-05, "loss": 0.0747, "step": 86190 }, { "epoch": 41.36, "learning_rate": 1.0691966644423985e-05, "loss": 0.0747, "step": 86200 }, { "epoch": 41.37, "learning_rate": 1.068309452710573e-05, "loss": 0.0694, "step": 86210 }, { "epoch": 41.37, "learning_rate": 1.0674225652022007e-05, "loss": 0.0738, "step": 86220 }, { "epoch": 41.38, "learning_rate": 1.066536001990418e-05, "loss": 0.0715, "step": 86230 }, { "epoch": 41.38, "learning_rate": 1.0656497631483342e-05, "loss": 0.0909, "step": 86240 }, { "epoch": 41.39, "learning_rate": 1.0647638487490324e-05, "loss": 0.0705, "step": 86250 }, { "epoch": 41.39, "learning_rate": 1.0638782588655689e-05, "loss": 0.0797, "step": 86260 }, { "epoch": 41.4, "learning_rate": 1.0629929935709742e-05, "loss": 0.0729, "step": 86270 }, { "epoch": 41.4, "learning_rate": 1.0621080529382483e-05, "loss": 0.0689, "step": 86280 }, { "epoch": 41.41, "learning_rate": 1.061223437040369e-05, "loss": 0.0807, "step": 86290 }, { "epoch": 41.41, "learning_rate": 1.0603391459502842e-05, "loss": 0.068, "step": 86300 }, { "epoch": 41.41, "learning_rate": 1.0594551797409186e-05, "loss": 0.0831, "step": 86310 }, { "epoch": 41.42, "learning_rate": 1.058571538485163e-05, "loss": 0.0763, "step": 86320 }, { "epoch": 41.42, "learning_rate": 1.057688222255891e-05, "loss": 0.0746, "step": 86330 }, { "epoch": 41.43, "learning_rate": 1.0568052311259436e-05, "loss": 0.0755, "step": 86340 }, { "epoch": 41.43, "learning_rate": 1.0559225651681331e-05, "loss": 0.0757, "step": 86350 }, { "epoch": 41.44, "learning_rate": 1.0550402244552503e-05, "loss": 0.0702, "step": 86360 }, { "epoch": 41.44, "learning_rate": 1.0541582090600555e-05, "loss": 0.0805, "step": 86370 }, { "epoch": 41.45, "learning_rate": 1.0532765190552851e-05, "loss": 0.0792, "step": 86380 }, { "epoch": 41.45, "learning_rate": 1.0523951545136446e-05, "loss": 0.0726, "step": 86390 }, { "epoch": 41.46, "learning_rate": 1.051514115507814e-05, "loss": 0.0672, "step": 86400 }, { "epoch": 41.46, "learning_rate": 1.0506334021104513e-05, "loss": 0.0769, "step": 86410 }, { "epoch": 41.46, "learning_rate": 1.0497530143941826e-05, "loss": 0.0731, "step": 86420 }, { "epoch": 41.47, "learning_rate": 1.0488729524316065e-05, "loss": 0.0735, "step": 86430 }, { "epoch": 41.47, "learning_rate": 1.0479932162952976e-05, "loss": 0.0838, "step": 86440 }, { "epoch": 41.48, "learning_rate": 1.0471138060578039e-05, "loss": 0.0792, "step": 86450 }, { "epoch": 41.48, "learning_rate": 1.0462347217916424e-05, "loss": 0.0819, "step": 86460 }, { "epoch": 41.49, "learning_rate": 1.0453559635693077e-05, "loss": 0.0817, "step": 86470 }, { "epoch": 41.49, "learning_rate": 1.0444775314632638e-05, "loss": 0.0821, "step": 86480 }, { "epoch": 41.5, "learning_rate": 1.0435994255459546e-05, "loss": 0.0697, "step": 86490 }, { "epoch": 41.5, "learning_rate": 1.0427216458897881e-05, "loss": 0.0764, "step": 86500 }, { "epoch": 41.51, "learning_rate": 1.0418441925671512e-05, "loss": 0.0704, "step": 86510 }, { "epoch": 41.51, "learning_rate": 1.0409670656504029e-05, "loss": 0.0664, "step": 86520 }, { "epoch": 41.52, "learning_rate": 1.0400902652118727e-05, "loss": 0.0702, "step": 86530 }, { "epoch": 41.52, "learning_rate": 1.0392137913238665e-05, "loss": 0.0835, "step": 86540 }, { "epoch": 41.52, "learning_rate": 1.0383376440586617e-05, "loss": 0.0723, "step": 86550 }, { "epoch": 41.53, "learning_rate": 1.0374618234885092e-05, "loss": 0.069, "step": 86560 }, { "epoch": 41.53, "learning_rate": 1.0365863296856333e-05, "loss": 0.0709, "step": 86570 }, { "epoch": 41.54, "learning_rate": 1.0357111627222293e-05, "loss": 0.0596, "step": 86580 }, { "epoch": 41.54, "learning_rate": 1.0348363226704683e-05, "loss": 0.0724, "step": 86590 }, { "epoch": 41.55, "learning_rate": 1.0339618096024945e-05, "loss": 0.0693, "step": 86600 }, { "epoch": 41.55, "learning_rate": 1.0330876235904213e-05, "loss": 0.0728, "step": 86610 }, { "epoch": 41.56, "learning_rate": 1.0322137647063385e-05, "loss": 0.0712, "step": 86620 }, { "epoch": 41.56, "learning_rate": 1.0313402330223081e-05, "loss": 0.0856, "step": 86630 }, { "epoch": 41.57, "learning_rate": 1.0304670286103651e-05, "loss": 0.0723, "step": 86640 }, { "epoch": 41.57, "learning_rate": 1.0295941515425184e-05, "loss": 0.0741, "step": 86650 }, { "epoch": 41.57, "learning_rate": 1.0287216018907486e-05, "loss": 0.0793, "step": 86660 }, { "epoch": 41.58, "learning_rate": 1.0278493797270106e-05, "loss": 0.0728, "step": 86670 }, { "epoch": 41.58, "learning_rate": 1.0269774851232294e-05, "loss": 0.0863, "step": 86680 }, { "epoch": 41.59, "learning_rate": 1.026105918151306e-05, "loss": 0.075, "step": 86690 }, { "epoch": 41.59, "learning_rate": 1.0252346788831136e-05, "loss": 0.0705, "step": 86700 }, { "epoch": 41.6, "learning_rate": 1.0243637673904977e-05, "loss": 0.0758, "step": 86710 }, { "epoch": 41.6, "learning_rate": 1.0234931837452782e-05, "loss": 0.0698, "step": 86720 }, { "epoch": 41.61, "learning_rate": 1.0226229280192463e-05, "loss": 0.0704, "step": 86730 }, { "epoch": 41.61, "learning_rate": 1.0217530002841679e-05, "loss": 0.0714, "step": 86740 }, { "epoch": 41.62, "learning_rate": 1.0208834006117796e-05, "loss": 0.079, "step": 86750 }, { "epoch": 41.62, "learning_rate": 1.020014129073792e-05, "loss": 0.0756, "step": 86760 }, { "epoch": 41.62, "learning_rate": 1.01914518574189e-05, "loss": 0.074, "step": 86770 }, { "epoch": 41.63, "learning_rate": 1.018276570687729e-05, "loss": 0.0743, "step": 86780 }, { "epoch": 41.63, "learning_rate": 1.0174082839829397e-05, "loss": 0.0721, "step": 86790 }, { "epoch": 41.64, "learning_rate": 1.016540325699124e-05, "loss": 0.0743, "step": 86800 }, { "epoch": 41.64, "learning_rate": 1.0156726959078589e-05, "loss": 0.0678, "step": 86810 }, { "epoch": 41.65, "learning_rate": 1.0148053946806896e-05, "loss": 0.084, "step": 86820 }, { "epoch": 41.65, "learning_rate": 1.0139384220891396e-05, "loss": 0.0778, "step": 86830 }, { "epoch": 41.66, "learning_rate": 1.0130717782047027e-05, "loss": 0.0774, "step": 86840 }, { "epoch": 41.66, "learning_rate": 1.0122054630988453e-05, "loss": 0.0615, "step": 86850 }, { "epoch": 41.67, "learning_rate": 1.011339476843008e-05, "loss": 0.0688, "step": 86860 }, { "epoch": 41.67, "learning_rate": 1.0104738195086034e-05, "loss": 0.071, "step": 86870 }, { "epoch": 41.68, "learning_rate": 1.009608491167019e-05, "loss": 0.0747, "step": 86880 }, { "epoch": 41.68, "learning_rate": 1.0087434918896094e-05, "loss": 0.0714, "step": 86890 }, { "epoch": 41.68, "learning_rate": 1.0078788217477086e-05, "loss": 0.072, "step": 86900 }, { "epoch": 41.69, "learning_rate": 1.0070144808126202e-05, "loss": 0.0739, "step": 86910 }, { "epoch": 41.69, "learning_rate": 1.0061504691556217e-05, "loss": 0.0788, "step": 86920 }, { "epoch": 41.7, "learning_rate": 1.0052867868479631e-05, "loss": 0.0706, "step": 86930 }, { "epoch": 41.7, "learning_rate": 1.0044234339608666e-05, "loss": 0.0725, "step": 86940 }, { "epoch": 41.71, "learning_rate": 1.00356041056553e-05, "loss": 0.0703, "step": 86950 }, { "epoch": 41.71, "learning_rate": 1.0026977167331187e-05, "loss": 0.077, "step": 86960 }, { "epoch": 41.72, "learning_rate": 1.001835352534775e-05, "loss": 0.0729, "step": 86970 }, { "epoch": 41.72, "learning_rate": 1.0009733180416136e-05, "loss": 0.08, "step": 86980 }, { "epoch": 41.73, "learning_rate": 1.000111613324723e-05, "loss": 0.0676, "step": 86990 }, { "epoch": 41.73, "learning_rate": 9.992502384551577e-06, "loss": 0.0745, "step": 87000 }, { "epoch": 41.73, "learning_rate": 9.983891935039552e-06, "loss": 0.0666, "step": 87010 }, { "epoch": 41.74, "learning_rate": 9.975284785421196e-06, "loss": 0.0715, "step": 87020 }, { "epoch": 41.74, "learning_rate": 9.966680936406298e-06, "loss": 0.0688, "step": 87030 }, { "epoch": 41.75, "learning_rate": 9.958080388704343e-06, "loss": 0.0821, "step": 87040 }, { "epoch": 41.75, "learning_rate": 9.949483143024585e-06, "loss": 0.0738, "step": 87050 }, { "epoch": 41.76, "learning_rate": 9.940889200075992e-06, "loss": 0.0727, "step": 87060 }, { "epoch": 41.76, "learning_rate": 9.932298560567237e-06, "loss": 0.0684, "step": 87070 }, { "epoch": 41.77, "learning_rate": 9.92371122520674e-06, "loss": 0.0712, "step": 87080 }, { "epoch": 41.77, "learning_rate": 9.915127194702668e-06, "loss": 0.0709, "step": 87090 }, { "epoch": 41.78, "learning_rate": 9.9065464697629e-06, "loss": 0.0718, "step": 87100 }, { "epoch": 41.78, "learning_rate": 9.897969051095012e-06, "loss": 0.0684, "step": 87110 }, { "epoch": 41.78, "learning_rate": 9.88939493940635e-06, "loss": 0.0768, "step": 87120 }, { "epoch": 41.79, "learning_rate": 9.880824135403976e-06, "loss": 0.0703, "step": 87130 }, { "epoch": 41.79, "learning_rate": 9.872256639794653e-06, "loss": 0.0688, "step": 87140 }, { "epoch": 41.8, "learning_rate": 9.863692453284901e-06, "loss": 0.0687, "step": 87150 }, { "epoch": 41.8, "learning_rate": 9.855131576580951e-06, "loss": 0.066, "step": 87160 }, { "epoch": 41.81, "learning_rate": 9.846574010388804e-06, "loss": 0.0668, "step": 87170 }, { "epoch": 41.81, "learning_rate": 9.838019755414113e-06, "loss": 0.0729, "step": 87180 }, { "epoch": 41.82, "learning_rate": 9.829468812362314e-06, "loss": 0.0705, "step": 87190 }, { "epoch": 41.82, "learning_rate": 9.820921181938547e-06, "loss": 0.0824, "step": 87200 }, { "epoch": 41.83, "learning_rate": 9.8123768648477e-06, "loss": 0.0733, "step": 87210 }, { "epoch": 41.83, "learning_rate": 9.803835861794348e-06, "loss": 0.071, "step": 87220 }, { "epoch": 41.84, "learning_rate": 9.795298173482831e-06, "loss": 0.0644, "step": 87230 }, { "epoch": 41.84, "learning_rate": 9.7867638006172e-06, "loss": 0.0713, "step": 87240 }, { "epoch": 41.84, "learning_rate": 9.778232743901233e-06, "loss": 0.0766, "step": 87250 }, { "epoch": 41.85, "learning_rate": 9.769705004038442e-06, "loss": 0.0686, "step": 87260 }, { "epoch": 41.85, "learning_rate": 9.76118058173206e-06, "loss": 0.0619, "step": 87270 }, { "epoch": 41.86, "learning_rate": 9.75265947768505e-06, "loss": 0.0705, "step": 87280 }, { "epoch": 41.86, "learning_rate": 9.744141692600078e-06, "loss": 0.0767, "step": 87290 }, { "epoch": 41.87, "learning_rate": 9.735627227179572e-06, "loss": 0.0765, "step": 87300 }, { "epoch": 41.87, "learning_rate": 9.727116082125664e-06, "loss": 0.074, "step": 87310 }, { "epoch": 41.88, "learning_rate": 9.718608258140217e-06, "loss": 0.0731, "step": 87320 }, { "epoch": 41.88, "learning_rate": 9.710103755924827e-06, "loss": 0.0593, "step": 87330 }, { "epoch": 41.89, "learning_rate": 9.701602576180807e-06, "loss": 0.0691, "step": 87340 }, { "epoch": 41.89, "learning_rate": 9.693104719609213e-06, "loss": 0.0748, "step": 87350 }, { "epoch": 41.89, "learning_rate": 9.684610186910792e-06, "loss": 0.0761, "step": 87360 }, { "epoch": 41.9, "learning_rate": 9.676118978786042e-06, "loss": 0.0714, "step": 87370 }, { "epoch": 41.9, "learning_rate": 9.667631095935187e-06, "loss": 0.0707, "step": 87380 }, { "epoch": 41.91, "learning_rate": 9.659146539058178e-06, "loss": 0.0752, "step": 87390 }, { "epoch": 41.91, "learning_rate": 9.650665308854679e-06, "loss": 0.0701, "step": 87400 }, { "epoch": 41.92, "learning_rate": 9.642187406024095e-06, "loss": 0.0673, "step": 87410 }, { "epoch": 41.92, "learning_rate": 9.633712831265551e-06, "loss": 0.0699, "step": 87420 }, { "epoch": 41.93, "learning_rate": 9.625241585277884e-06, "loss": 0.0668, "step": 87430 }, { "epoch": 41.93, "learning_rate": 9.616773668759671e-06, "loss": 0.0704, "step": 87440 }, { "epoch": 41.94, "learning_rate": 9.608309082409211e-06, "loss": 0.0872, "step": 87450 }, { "epoch": 41.94, "learning_rate": 9.59984782692453e-06, "loss": 0.0898, "step": 87460 }, { "epoch": 41.94, "learning_rate": 9.591389903003384e-06, "loss": 0.0751, "step": 87470 }, { "epoch": 41.95, "learning_rate": 9.582935311343245e-06, "loss": 0.0754, "step": 87480 }, { "epoch": 42.0, "learning_rate": 9.57448405264132e-06, "loss": 0.0759, "step": 87490 }, { "epoch": 42.01, "learning_rate": 9.566036127594514e-06, "loss": 0.0698, "step": 87500 }, { "epoch": 42.01, "learning_rate": 9.557591536899496e-06, "loss": 0.0671, "step": 87510 }, { "epoch": 42.02, "learning_rate": 9.549150281252633e-06, "loss": 0.0735, "step": 87520 }, { "epoch": 42.02, "learning_rate": 9.54071236135003e-06, "loss": 0.0788, "step": 87530 }, { "epoch": 42.02, "learning_rate": 9.53227777788751e-06, "loss": 0.0684, "step": 87540 }, { "epoch": 42.03, "learning_rate": 9.523846531560631e-06, "loss": 0.0709, "step": 87550 }, { "epoch": 42.03, "learning_rate": 9.515418623064676e-06, "loss": 0.0742, "step": 87560 }, { "epoch": 42.04, "learning_rate": 9.506994053094615e-06, "loss": 0.0749, "step": 87570 }, { "epoch": 42.04, "learning_rate": 9.498572822345197e-06, "loss": 0.0657, "step": 87580 }, { "epoch": 42.05, "learning_rate": 9.490154931510859e-06, "loss": 0.0691, "step": 87590 }, { "epoch": 42.05, "learning_rate": 9.481740381285781e-06, "loss": 0.0668, "step": 87600 }, { "epoch": 42.06, "learning_rate": 9.473329172363865e-06, "loss": 0.0704, "step": 87610 }, { "epoch": 42.06, "learning_rate": 9.46492130543873e-06, "loss": 0.0777, "step": 87620 }, { "epoch": 42.07, "learning_rate": 9.456516781203722e-06, "loss": 0.0714, "step": 87630 }, { "epoch": 42.07, "learning_rate": 9.448115600351931e-06, "loss": 0.0788, "step": 87640 }, { "epoch": 42.07, "learning_rate": 9.439717763576123e-06, "loss": 0.0752, "step": 87650 }, { "epoch": 42.08, "learning_rate": 9.431323271568831e-06, "loss": 0.0718, "step": 87660 }, { "epoch": 42.08, "learning_rate": 9.422932125022315e-06, "loss": 0.0762, "step": 87670 }, { "epoch": 42.09, "learning_rate": 9.414544324628504e-06, "loss": 0.0689, "step": 87680 }, { "epoch": 42.09, "learning_rate": 9.40615987107913e-06, "loss": 0.0687, "step": 87690 }, { "epoch": 42.1, "learning_rate": 9.39777876506559e-06, "loss": 0.0855, "step": 87700 }, { "epoch": 42.1, "learning_rate": 9.389401007279047e-06, "loss": 0.0739, "step": 87710 }, { "epoch": 42.11, "learning_rate": 9.381026598410336e-06, "loss": 0.0777, "step": 87720 }, { "epoch": 42.11, "learning_rate": 9.37265553915006e-06, "loss": 0.0681, "step": 87730 }, { "epoch": 42.12, "learning_rate": 9.364287830188539e-06, "loss": 0.0777, "step": 87740 }, { "epoch": 42.12, "learning_rate": 9.355923472215789e-06, "loss": 0.0641, "step": 87750 }, { "epoch": 42.13, "learning_rate": 9.347562465921567e-06, "loss": 0.072, "step": 87760 }, { "epoch": 42.13, "learning_rate": 9.339204811995383e-06, "loss": 0.0725, "step": 87770 }, { "epoch": 42.13, "learning_rate": 9.330850511126444e-06, "loss": 0.0734, "step": 87780 }, { "epoch": 42.14, "learning_rate": 9.322499564003645e-06, "loss": 0.0721, "step": 87790 }, { "epoch": 42.14, "learning_rate": 9.314151971315666e-06, "loss": 0.0781, "step": 87800 }, { "epoch": 42.15, "learning_rate": 9.305807733750882e-06, "loss": 0.0716, "step": 87810 }, { "epoch": 42.15, "learning_rate": 9.2974668519974e-06, "loss": 0.0743, "step": 87820 }, { "epoch": 42.16, "learning_rate": 9.289129326743019e-06, "loss": 0.0728, "step": 87830 }, { "epoch": 42.16, "learning_rate": 9.280795158675287e-06, "loss": 0.0691, "step": 87840 }, { "epoch": 42.17, "learning_rate": 9.272464348481512e-06, "loss": 0.0738, "step": 87850 }, { "epoch": 42.17, "learning_rate": 9.264136896848652e-06, "loss": 0.0738, "step": 87860 }, { "epoch": 42.18, "learning_rate": 9.255812804463431e-06, "loss": 0.0699, "step": 87870 }, { "epoch": 42.18, "learning_rate": 9.247492072012292e-06, "loss": 0.0693, "step": 87880 }, { "epoch": 42.18, "learning_rate": 9.239174700181407e-06, "loss": 0.0765, "step": 87890 }, { "epoch": 42.19, "learning_rate": 9.23086068965664e-06, "loss": 0.074, "step": 87900 }, { "epoch": 42.19, "learning_rate": 9.222550041123596e-06, "loss": 0.0677, "step": 87910 }, { "epoch": 42.2, "learning_rate": 9.214242755267638e-06, "loss": 0.0788, "step": 87920 }, { "epoch": 42.2, "learning_rate": 9.205938832773786e-06, "loss": 0.069, "step": 87930 }, { "epoch": 42.21, "learning_rate": 9.197638274326831e-06, "loss": 0.0742, "step": 87940 }, { "epoch": 42.21, "learning_rate": 9.189341080611264e-06, "loss": 0.0703, "step": 87950 }, { "epoch": 42.22, "learning_rate": 9.181047252311326e-06, "loss": 0.074, "step": 87960 }, { "epoch": 42.22, "learning_rate": 9.172756790110931e-06, "loss": 0.0664, "step": 87970 }, { "epoch": 42.23, "learning_rate": 9.164469694693762e-06, "loss": 0.0745, "step": 87980 }, { "epoch": 42.23, "learning_rate": 9.156185966743192e-06, "loss": 0.0719, "step": 87990 }, { "epoch": 42.23, "learning_rate": 9.147905606942364e-06, "loss": 0.0734, "step": 88000 }, { "epoch": 42.24, "learning_rate": 9.139628615974077e-06, "loss": 0.0825, "step": 88010 }, { "epoch": 42.24, "learning_rate": 9.131354994520902e-06, "loss": 0.0766, "step": 88020 }, { "epoch": 42.25, "learning_rate": 9.12308474326512e-06, "loss": 0.0768, "step": 88030 }, { "epoch": 42.25, "learning_rate": 9.114817862888714e-06, "loss": 0.0648, "step": 88040 }, { "epoch": 42.26, "learning_rate": 9.106554354073415e-06, "loss": 0.0709, "step": 88050 }, { "epoch": 42.26, "learning_rate": 9.098294217500664e-06, "loss": 0.0726, "step": 88060 }, { "epoch": 42.27, "learning_rate": 9.09003745385163e-06, "loss": 0.0624, "step": 88070 }, { "epoch": 42.27, "learning_rate": 9.081784063807191e-06, "loss": 0.0748, "step": 88080 }, { "epoch": 42.28, "learning_rate": 9.073534048047966e-06, "loss": 0.0709, "step": 88090 }, { "epoch": 42.28, "learning_rate": 9.065287407254292e-06, "loss": 0.0754, "step": 88100 }, { "epoch": 42.29, "learning_rate": 9.057044142106198e-06, "loss": 0.0679, "step": 88110 }, { "epoch": 42.29, "learning_rate": 9.048804253283471e-06, "loss": 0.0755, "step": 88120 }, { "epoch": 42.29, "learning_rate": 9.040567741465606e-06, "loss": 0.0695, "step": 88130 }, { "epoch": 42.3, "learning_rate": 9.032334607331816e-06, "loss": 0.0762, "step": 88140 }, { "epoch": 42.3, "learning_rate": 9.024104851561045e-06, "loss": 0.0626, "step": 88150 }, { "epoch": 42.31, "learning_rate": 9.015878474831946e-06, "loss": 0.0674, "step": 88160 }, { "epoch": 42.31, "learning_rate": 9.007655477822919e-06, "loss": 0.0783, "step": 88170 }, { "epoch": 42.32, "learning_rate": 8.999435861212036e-06, "loss": 0.0732, "step": 88180 }, { "epoch": 42.32, "learning_rate": 8.99121962567714e-06, "loss": 0.0687, "step": 88190 }, { "epoch": 42.33, "learning_rate": 8.983006771895763e-06, "loss": 0.0655, "step": 88200 }, { "epoch": 42.33, "learning_rate": 8.97479730054519e-06, "loss": 0.0746, "step": 88210 }, { "epoch": 42.34, "learning_rate": 8.966591212302388e-06, "loss": 0.0734, "step": 88220 }, { "epoch": 42.34, "learning_rate": 8.958388507844085e-06, "loss": 0.0721, "step": 88230 }, { "epoch": 42.34, "learning_rate": 8.950189187846691e-06, "loss": 0.0676, "step": 88240 }, { "epoch": 42.35, "learning_rate": 8.94199325298638e-06, "loss": 0.0753, "step": 88250 }, { "epoch": 42.35, "learning_rate": 8.933800703938993e-06, "loss": 0.0769, "step": 88260 }, { "epoch": 42.36, "learning_rate": 8.925611541380136e-06, "loss": 0.0709, "step": 88270 }, { "epoch": 42.36, "learning_rate": 8.917425765985115e-06, "loss": 0.0713, "step": 88280 }, { "epoch": 42.37, "learning_rate": 8.909243378428971e-06, "loss": 0.0671, "step": 88290 }, { "epoch": 42.37, "learning_rate": 8.901064379386459e-06, "loss": 0.0776, "step": 88300 }, { "epoch": 42.38, "learning_rate": 8.892888769532048e-06, "loss": 0.0759, "step": 88310 }, { "epoch": 42.38, "learning_rate": 8.88471654953994e-06, "loss": 0.0729, "step": 88320 }, { "epoch": 42.39, "learning_rate": 8.87654772008404e-06, "loss": 0.0713, "step": 88330 }, { "epoch": 42.39, "learning_rate": 8.868382281837977e-06, "loss": 0.0726, "step": 88340 }, { "epoch": 42.39, "learning_rate": 8.860220235475136e-06, "loss": 0.0697, "step": 88350 }, { "epoch": 42.4, "learning_rate": 8.852061581668548e-06, "loss": 0.0701, "step": 88360 }, { "epoch": 42.4, "learning_rate": 8.843906321091044e-06, "loss": 0.0722, "step": 88370 }, { "epoch": 42.41, "learning_rate": 8.83575445441513e-06, "loss": 0.0733, "step": 88380 }, { "epoch": 42.41, "learning_rate": 8.827605982313059e-06, "loss": 0.078, "step": 88390 }, { "epoch": 42.42, "learning_rate": 8.819460905456762e-06, "loss": 0.0753, "step": 88400 }, { "epoch": 42.42, "learning_rate": 8.811319224517928e-06, "loss": 0.0714, "step": 88410 }, { "epoch": 42.43, "learning_rate": 8.80318094016795e-06, "loss": 0.0743, "step": 88420 }, { "epoch": 42.43, "learning_rate": 8.79504605307796e-06, "loss": 0.0794, "step": 88430 }, { "epoch": 42.44, "learning_rate": 8.786914563918753e-06, "loss": 0.0773, "step": 88440 }, { "epoch": 42.44, "learning_rate": 8.778786473360928e-06, "loss": 0.074, "step": 88450 }, { "epoch": 42.45, "learning_rate": 8.770661782074757e-06, "loss": 0.0708, "step": 88460 }, { "epoch": 42.45, "learning_rate": 8.762540490730214e-06, "loss": 0.0749, "step": 88470 }, { "epoch": 42.45, "learning_rate": 8.754422599997025e-06, "loss": 0.0718, "step": 88480 }, { "epoch": 42.46, "learning_rate": 8.746308110544627e-06, "loss": 0.075, "step": 88490 }, { "epoch": 42.46, "learning_rate": 8.738197023042183e-06, "loss": 0.0671, "step": 88500 }, { "epoch": 42.47, "learning_rate": 8.730089338158532e-06, "loss": 0.069, "step": 88510 }, { "epoch": 42.47, "learning_rate": 8.721985056562298e-06, "loss": 0.0665, "step": 88520 }, { "epoch": 42.48, "learning_rate": 8.713884178921804e-06, "loss": 0.0819, "step": 88530 }, { "epoch": 42.48, "learning_rate": 8.705786705905056e-06, "loss": 0.0749, "step": 88540 }, { "epoch": 42.49, "learning_rate": 8.697692638179811e-06, "loss": 0.0746, "step": 88550 }, { "epoch": 42.49, "learning_rate": 8.689601976413542e-06, "loss": 0.0746, "step": 88560 }, { "epoch": 42.5, "learning_rate": 8.681514721273449e-06, "loss": 0.0763, "step": 88570 }, { "epoch": 42.5, "learning_rate": 8.673430873426419e-06, "loss": 0.0758, "step": 88580 }, { "epoch": 42.5, "learning_rate": 8.665350433539077e-06, "loss": 0.0797, "step": 88590 }, { "epoch": 42.51, "learning_rate": 8.657273402277799e-06, "loss": 0.073, "step": 88600 }, { "epoch": 42.51, "learning_rate": 8.649199780308642e-06, "loss": 0.0738, "step": 88610 }, { "epoch": 42.52, "learning_rate": 8.641129568297374e-06, "loss": 0.07, "step": 88620 }, { "epoch": 42.52, "learning_rate": 8.633062766909505e-06, "loss": 0.0727, "step": 88630 }, { "epoch": 42.53, "learning_rate": 8.624999376810267e-06, "loss": 0.0704, "step": 88640 }, { "epoch": 42.53, "learning_rate": 8.616939398664587e-06, "loss": 0.0687, "step": 88650 }, { "epoch": 42.54, "learning_rate": 8.608882833137122e-06, "loss": 0.0694, "step": 88660 }, { "epoch": 42.54, "learning_rate": 8.600829680892248e-06, "loss": 0.0714, "step": 88670 }, { "epoch": 42.55, "learning_rate": 8.592779942594097e-06, "loss": 0.0696, "step": 88680 }, { "epoch": 42.55, "learning_rate": 8.584733618906437e-06, "loss": 0.0695, "step": 88690 }, { "epoch": 42.55, "learning_rate": 8.57669071049283e-06, "loss": 0.0685, "step": 88700 }, { "epoch": 42.56, "learning_rate": 8.568651218016522e-06, "loss": 0.0682, "step": 88710 }, { "epoch": 42.56, "learning_rate": 8.560615142140472e-06, "loss": 0.0719, "step": 88720 }, { "epoch": 42.57, "learning_rate": 8.552582483527371e-06, "loss": 0.0683, "step": 88730 }, { "epoch": 42.57, "learning_rate": 8.544553242839637e-06, "loss": 0.0727, "step": 88740 }, { "epoch": 42.58, "learning_rate": 8.536527420739377e-06, "loss": 0.0772, "step": 88750 }, { "epoch": 42.58, "learning_rate": 8.52850501788845e-06, "loss": 0.074, "step": 88760 }, { "epoch": 42.59, "learning_rate": 8.520486034948406e-06, "loss": 0.072, "step": 88770 }, { "epoch": 42.59, "learning_rate": 8.512470472580542e-06, "loss": 0.0753, "step": 88780 }, { "epoch": 42.6, "learning_rate": 8.504458331445825e-06, "loss": 0.0698, "step": 88790 }, { "epoch": 42.6, "learning_rate": 8.496449612204982e-06, "loss": 0.0668, "step": 88800 }, { "epoch": 42.61, "learning_rate": 8.488444315518452e-06, "loss": 0.0699, "step": 88810 }, { "epoch": 42.61, "learning_rate": 8.48044244204637e-06, "loss": 0.0661, "step": 88820 }, { "epoch": 42.61, "learning_rate": 8.472443992448619e-06, "loss": 0.0652, "step": 88830 }, { "epoch": 42.62, "learning_rate": 8.46444896738478e-06, "loss": 0.0682, "step": 88840 }, { "epoch": 42.62, "learning_rate": 8.456457367514154e-06, "loss": 0.0776, "step": 88850 }, { "epoch": 42.63, "learning_rate": 8.448469193495774e-06, "loss": 0.0895, "step": 88860 }, { "epoch": 42.63, "learning_rate": 8.440484445988355e-06, "loss": 0.0713, "step": 88870 }, { "epoch": 42.64, "learning_rate": 8.432503125650359e-06, "loss": 0.0741, "step": 88880 }, { "epoch": 42.64, "learning_rate": 8.424525233139963e-06, "loss": 0.062, "step": 88890 }, { "epoch": 42.65, "learning_rate": 8.416550769115056e-06, "loss": 0.0693, "step": 88900 }, { "epoch": 42.65, "learning_rate": 8.408579734233246e-06, "loss": 0.0736, "step": 88910 }, { "epoch": 42.66, "learning_rate": 8.400612129151859e-06, "loss": 0.0684, "step": 88920 }, { "epoch": 42.66, "learning_rate": 8.392647954527944e-06, "loss": 0.0712, "step": 88930 }, { "epoch": 42.66, "learning_rate": 8.384687211018244e-06, "loss": 0.0692, "step": 88940 }, { "epoch": 42.67, "learning_rate": 8.376729899279234e-06, "loss": 0.0701, "step": 88950 }, { "epoch": 42.67, "learning_rate": 8.368776019967117e-06, "loss": 0.0668, "step": 88960 }, { "epoch": 42.68, "learning_rate": 8.360825573737802e-06, "loss": 0.0774, "step": 88970 }, { "epoch": 42.68, "learning_rate": 8.352878561246913e-06, "loss": 0.0844, "step": 88980 }, { "epoch": 42.69, "learning_rate": 8.344934983149793e-06, "loss": 0.0766, "step": 88990 }, { "epoch": 42.69, "learning_rate": 8.336994840101514e-06, "loss": 0.0755, "step": 89000 }, { "epoch": 42.7, "learning_rate": 8.329058132756834e-06, "loss": 0.0753, "step": 89010 }, { "epoch": 42.7, "learning_rate": 8.321124861770251e-06, "loss": 0.0691, "step": 89020 }, { "epoch": 42.71, "learning_rate": 8.313195027795977e-06, "loss": 0.0793, "step": 89030 }, { "epoch": 42.71, "learning_rate": 8.30526863148795e-06, "loss": 0.0734, "step": 89040 }, { "epoch": 42.71, "learning_rate": 8.2973456734998e-06, "loss": 0.0716, "step": 89050 }, { "epoch": 42.72, "learning_rate": 8.289426154484891e-06, "loss": 0.0728, "step": 89060 }, { "epoch": 42.72, "learning_rate": 8.28151007509631e-06, "loss": 0.077, "step": 89070 }, { "epoch": 42.73, "learning_rate": 8.27359743598683e-06, "loss": 0.0789, "step": 89080 }, { "epoch": 42.73, "learning_rate": 8.265688237808972e-06, "loss": 0.0682, "step": 89090 }, { "epoch": 42.74, "learning_rate": 8.257782481214953e-06, "loss": 0.0633, "step": 89100 }, { "epoch": 42.74, "learning_rate": 8.249880166856728e-06, "loss": 0.0788, "step": 89110 }, { "epoch": 42.75, "learning_rate": 8.241981295385937e-06, "loss": 0.0774, "step": 89120 }, { "epoch": 42.75, "learning_rate": 8.234085867453967e-06, "loss": 0.0787, "step": 89130 }, { "epoch": 42.76, "learning_rate": 8.226193883711919e-06, "loss": 0.0677, "step": 89140 }, { "epoch": 42.76, "learning_rate": 8.218305344810568e-06, "loss": 0.0817, "step": 89150 }, { "epoch": 42.77, "learning_rate": 8.210420251400448e-06, "loss": 0.0679, "step": 89160 }, { "epoch": 42.77, "learning_rate": 8.202538604131798e-06, "loss": 0.0677, "step": 89170 }, { "epoch": 42.77, "learning_rate": 8.194660403654591e-06, "loss": 0.0701, "step": 89180 }, { "epoch": 42.78, "learning_rate": 8.18678565061845e-06, "loss": 0.0756, "step": 89190 }, { "epoch": 42.78, "learning_rate": 8.1789143456728e-06, "loss": 0.0709, "step": 89200 }, { "epoch": 42.79, "learning_rate": 8.171046489466728e-06, "loss": 0.068, "step": 89210 }, { "epoch": 42.79, "learning_rate": 8.163182082649063e-06, "loss": 0.0706, "step": 89220 }, { "epoch": 42.8, "learning_rate": 8.155321125868309e-06, "loss": 0.061, "step": 89230 }, { "epoch": 42.8, "learning_rate": 8.147463619772727e-06, "loss": 0.0708, "step": 89240 }, { "epoch": 42.81, "learning_rate": 8.139609565010297e-06, "loss": 0.072, "step": 89250 }, { "epoch": 42.81, "learning_rate": 8.131758962228663e-06, "loss": 0.0743, "step": 89260 }, { "epoch": 42.82, "learning_rate": 8.123911812075224e-06, "loss": 0.0821, "step": 89270 }, { "epoch": 42.82, "learning_rate": 8.116068115197117e-06, "loss": 0.0699, "step": 89280 }, { "epoch": 42.82, "learning_rate": 8.108227872241153e-06, "loss": 0.087, "step": 89290 }, { "epoch": 42.83, "learning_rate": 8.10039108385386e-06, "loss": 0.0676, "step": 89300 }, { "epoch": 42.83, "learning_rate": 8.092557750681496e-06, "loss": 0.0724, "step": 89310 }, { "epoch": 42.84, "learning_rate": 8.084727873370045e-06, "loss": 0.0725, "step": 89320 }, { "epoch": 42.84, "learning_rate": 8.07690145256516e-06, "loss": 0.068, "step": 89330 }, { "epoch": 42.85, "learning_rate": 8.06907848891227e-06, "loss": 0.0743, "step": 89340 }, { "epoch": 42.85, "learning_rate": 8.061258983056452e-06, "loss": 0.0731, "step": 89350 }, { "epoch": 42.86, "learning_rate": 8.053442935642592e-06, "loss": 0.0721, "step": 89360 }, { "epoch": 42.86, "learning_rate": 8.045630347315186e-06, "loss": 0.0681, "step": 89370 }, { "epoch": 42.87, "learning_rate": 8.03782121871851e-06, "loss": 0.0665, "step": 89380 }, { "epoch": 42.87, "learning_rate": 8.030015550496545e-06, "loss": 0.0738, "step": 89390 }, { "epoch": 42.87, "learning_rate": 8.022213343292956e-06, "loss": 0.0629, "step": 89400 }, { "epoch": 42.88, "learning_rate": 8.014414597751158e-06, "loss": 0.0743, "step": 89410 }, { "epoch": 42.88, "learning_rate": 8.00661931451427e-06, "loss": 0.072, "step": 89420 }, { "epoch": 42.89, "learning_rate": 7.99882749422512e-06, "loss": 0.0774, "step": 89430 }, { "epoch": 42.89, "learning_rate": 7.991039137526252e-06, "loss": 0.0804, "step": 89440 }, { "epoch": 42.9, "learning_rate": 7.98325424505993e-06, "loss": 0.0733, "step": 89450 }, { "epoch": 42.9, "learning_rate": 7.975472817468126e-06, "loss": 0.0598, "step": 89460 }, { "epoch": 42.91, "learning_rate": 7.967694855392543e-06, "loss": 0.0689, "step": 89470 }, { "epoch": 42.91, "learning_rate": 7.959920359474555e-06, "loss": 0.0894, "step": 89480 }, { "epoch": 42.92, "learning_rate": 7.952149330355296e-06, "loss": 0.0722, "step": 89490 }, { "epoch": 42.92, "learning_rate": 7.944381768675602e-06, "loss": 0.078, "step": 89500 }, { "epoch": 42.93, "learning_rate": 7.936617675076002e-06, "loss": 0.0791, "step": 89510 }, { "epoch": 42.93, "learning_rate": 7.928857050196775e-06, "loss": 0.0782, "step": 89520 }, { "epoch": 42.93, "learning_rate": 7.92109989467788e-06, "loss": 0.0688, "step": 89530 }, { "epoch": 42.94, "learning_rate": 7.913346209159016e-06, "loss": 0.0716, "step": 89540 }, { "epoch": 42.94, "learning_rate": 7.905595994279568e-06, "loss": 0.071, "step": 89550 }, { "epoch": 42.95, "learning_rate": 7.897849250678662e-06, "loss": 0.0863, "step": 89560 }, { "epoch": 43.0, "learning_rate": 7.890105978995122e-06, "loss": 0.0648, "step": 89570 }, { "epoch": 43.01, "learning_rate": 7.88236617986749e-06, "loss": 0.07, "step": 89580 }, { "epoch": 43.01, "learning_rate": 7.87462985393403e-06, "loss": 0.0792, "step": 89590 }, { "epoch": 43.01, "learning_rate": 7.866897001832696e-06, "loss": 0.0765, "step": 89600 }, { "epoch": 43.02, "learning_rate": 7.859167624201197e-06, "loss": 0.0701, "step": 89610 }, { "epoch": 43.02, "learning_rate": 7.851441721676905e-06, "loss": 0.0696, "step": 89620 }, { "epoch": 43.03, "learning_rate": 7.843719294896929e-06, "loss": 0.074, "step": 89630 }, { "epoch": 43.03, "learning_rate": 7.836000344498107e-06, "loss": 0.0727, "step": 89640 }, { "epoch": 43.04, "learning_rate": 7.828284871116964e-06, "loss": 0.0677, "step": 89650 }, { "epoch": 43.04, "learning_rate": 7.820572875389758e-06, "loss": 0.0693, "step": 89660 }, { "epoch": 43.05, "learning_rate": 7.812864357952448e-06, "loss": 0.0644, "step": 89670 }, { "epoch": 43.05, "learning_rate": 7.805159319440725e-06, "loss": 0.0628, "step": 89680 }, { "epoch": 43.06, "learning_rate": 7.797457760489951e-06, "loss": 0.0689, "step": 89690 }, { "epoch": 43.06, "learning_rate": 7.789759681735242e-06, "loss": 0.0669, "step": 89700 }, { "epoch": 43.06, "learning_rate": 7.782065083811414e-06, "loss": 0.0726, "step": 89710 }, { "epoch": 43.07, "learning_rate": 7.774373967352994e-06, "loss": 0.0654, "step": 89720 }, { "epoch": 43.07, "learning_rate": 7.766686332994227e-06, "loss": 0.0698, "step": 89730 }, { "epoch": 43.08, "learning_rate": 7.759002181369063e-06, "loss": 0.0696, "step": 89740 }, { "epoch": 43.08, "learning_rate": 7.751321513111176e-06, "loss": 0.0726, "step": 89750 }, { "epoch": 43.09, "learning_rate": 7.743644328853933e-06, "loss": 0.079, "step": 89760 }, { "epoch": 43.09, "learning_rate": 7.735970629230432e-06, "loss": 0.0711, "step": 89770 }, { "epoch": 43.1, "learning_rate": 7.72830041487348e-06, "loss": 0.072, "step": 89780 }, { "epoch": 43.1, "learning_rate": 7.720633686415596e-06, "loss": 0.0638, "step": 89790 }, { "epoch": 43.11, "learning_rate": 7.712970444489004e-06, "loss": 0.0759, "step": 89800 }, { "epoch": 43.11, "learning_rate": 7.705310689725647e-06, "loss": 0.0704, "step": 89810 }, { "epoch": 43.11, "learning_rate": 7.697654422757188e-06, "loss": 0.0837, "step": 89820 }, { "epoch": 43.12, "learning_rate": 7.690001644215e-06, "loss": 0.0854, "step": 89830 }, { "epoch": 43.12, "learning_rate": 7.682352354730144e-06, "loss": 0.0779, "step": 89840 }, { "epoch": 43.13, "learning_rate": 7.674706554933414e-06, "loss": 0.0723, "step": 89850 }, { "epoch": 43.13, "learning_rate": 7.667064245455335e-06, "loss": 0.074, "step": 89860 }, { "epoch": 43.14, "learning_rate": 7.659425426926086e-06, "loss": 0.0756, "step": 89870 }, { "epoch": 43.14, "learning_rate": 7.651790099975637e-06, "loss": 0.076, "step": 89880 }, { "epoch": 43.15, "learning_rate": 7.644158265233603e-06, "loss": 0.0785, "step": 89890 }, { "epoch": 43.15, "learning_rate": 7.636529923329355e-06, "loss": 0.0798, "step": 89900 }, { "epoch": 43.16, "learning_rate": 7.6289050748919396e-06, "loss": 0.0697, "step": 89910 }, { "epoch": 43.16, "learning_rate": 7.621283720550137e-06, "loss": 0.0693, "step": 89920 }, { "epoch": 43.16, "learning_rate": 7.613665860932451e-06, "loss": 0.0732, "step": 89930 }, { "epoch": 43.17, "learning_rate": 7.606051496667061e-06, "loss": 0.0777, "step": 89940 }, { "epoch": 43.17, "learning_rate": 7.59844062838187e-06, "loss": 0.0751, "step": 89950 }, { "epoch": 43.18, "learning_rate": 7.590833256704533e-06, "loss": 0.0816, "step": 89960 }, { "epoch": 43.18, "learning_rate": 7.583229382262386e-06, "loss": 0.0789, "step": 89970 }, { "epoch": 43.19, "learning_rate": 7.575629005682449e-06, "loss": 0.0745, "step": 89980 }, { "epoch": 43.19, "learning_rate": 7.56803212759149e-06, "loss": 0.0693, "step": 89990 }, { "epoch": 43.2, "learning_rate": 7.5604387486159836e-06, "loss": 0.0841, "step": 90000 }, { "epoch": 43.2, "learning_rate": 7.552848869382123e-06, "loss": 0.0733, "step": 90010 }, { "epoch": 43.21, "learning_rate": 7.545262490515775e-06, "loss": 0.0758, "step": 90020 }, { "epoch": 43.21, "learning_rate": 7.537679612642546e-06, "loss": 0.0696, "step": 90030 }, { "epoch": 43.22, "learning_rate": 7.530100236387782e-06, "loss": 0.0729, "step": 90040 }, { "epoch": 43.22, "learning_rate": 7.522524362376482e-06, "loss": 0.0751, "step": 90050 }, { "epoch": 43.22, "learning_rate": 7.514951991233387e-06, "loss": 0.0742, "step": 90060 }, { "epoch": 43.23, "learning_rate": 7.50738312358295e-06, "loss": 0.0719, "step": 90070 }, { "epoch": 43.23, "learning_rate": 7.499817760049344e-06, "loss": 0.0644, "step": 90080 }, { "epoch": 43.24, "learning_rate": 7.4922559012564175e-06, "loss": 0.0684, "step": 90090 }, { "epoch": 43.24, "learning_rate": 7.484697547827762e-06, "loss": 0.0854, "step": 90100 }, { "epoch": 43.25, "learning_rate": 7.477142700386674e-06, "loss": 0.0727, "step": 90110 }, { "epoch": 43.25, "learning_rate": 7.469591359556149e-06, "loss": 0.0744, "step": 90120 }, { "epoch": 43.26, "learning_rate": 7.462043525958912e-06, "loss": 0.0755, "step": 90130 }, { "epoch": 43.26, "learning_rate": 7.454499200217385e-06, "loss": 0.0809, "step": 90140 }, { "epoch": 43.27, "learning_rate": 7.446958382953712e-06, "loss": 0.0725, "step": 90150 }, { "epoch": 43.27, "learning_rate": 7.4394210747897205e-06, "loss": 0.0658, "step": 90160 }, { "epoch": 43.27, "learning_rate": 7.431887276346977e-06, "loss": 0.0701, "step": 90170 }, { "epoch": 43.28, "learning_rate": 7.424356988246756e-06, "loss": 0.0736, "step": 90180 }, { "epoch": 43.28, "learning_rate": 7.4168302111100265e-06, "loss": 0.0727, "step": 90190 }, { "epoch": 43.29, "learning_rate": 7.409306945557487e-06, "loss": 0.0669, "step": 90200 }, { "epoch": 43.29, "learning_rate": 7.401787192209531e-06, "loss": 0.0683, "step": 90210 }, { "epoch": 43.3, "learning_rate": 7.394270951686283e-06, "loss": 0.0723, "step": 90220 }, { "epoch": 43.3, "learning_rate": 7.386758224607543e-06, "loss": 0.0769, "step": 90230 }, { "epoch": 43.31, "learning_rate": 7.379249011592842e-06, "loss": 0.0738, "step": 90240 }, { "epoch": 43.31, "learning_rate": 7.371743313261431e-06, "loss": 0.0768, "step": 90250 }, { "epoch": 43.32, "learning_rate": 7.364241130232258e-06, "loss": 0.0814, "step": 90260 }, { "epoch": 43.32, "learning_rate": 7.356742463123983e-06, "loss": 0.0773, "step": 90270 }, { "epoch": 43.32, "learning_rate": 7.34924731255498e-06, "loss": 0.0611, "step": 90280 }, { "epoch": 43.33, "learning_rate": 7.341755679143336e-06, "loss": 0.0753, "step": 90290 }, { "epoch": 43.33, "learning_rate": 7.334267563506825e-06, "loss": 0.0706, "step": 90300 }, { "epoch": 43.34, "learning_rate": 7.326782966262957e-06, "loss": 0.0757, "step": 90310 }, { "epoch": 43.34, "learning_rate": 7.319301888028945e-06, "loss": 0.0729, "step": 90320 }, { "epoch": 43.35, "learning_rate": 7.311824329421707e-06, "loss": 0.0761, "step": 90330 }, { "epoch": 43.35, "learning_rate": 7.304350291057871e-06, "loss": 0.0727, "step": 90340 }, { "epoch": 43.36, "learning_rate": 7.296879773553783e-06, "loss": 0.0699, "step": 90350 }, { "epoch": 43.36, "learning_rate": 7.289412777525495e-06, "loss": 0.0723, "step": 90360 }, { "epoch": 43.37, "learning_rate": 7.2819493035887585e-06, "loss": 0.067, "step": 90370 }, { "epoch": 43.37, "learning_rate": 7.274489352359038e-06, "loss": 0.0643, "step": 90380 }, { "epoch": 43.38, "learning_rate": 7.267032924451522e-06, "loss": 0.0718, "step": 90390 }, { "epoch": 43.38, "learning_rate": 7.259580020481093e-06, "loss": 0.0665, "step": 90400 }, { "epoch": 43.38, "learning_rate": 7.252130641062354e-06, "loss": 0.0769, "step": 90410 }, { "epoch": 43.39, "learning_rate": 7.244684786809608e-06, "loss": 0.0691, "step": 90420 }, { "epoch": 43.39, "learning_rate": 7.237242458336868e-06, "loss": 0.0689, "step": 90430 }, { "epoch": 43.4, "learning_rate": 7.229803656257878e-06, "loss": 0.0705, "step": 90440 }, { "epoch": 43.4, "learning_rate": 7.22236838118604e-06, "loss": 0.0698, "step": 90450 }, { "epoch": 43.41, "learning_rate": 7.21493663373452e-06, "loss": 0.0755, "step": 90460 }, { "epoch": 43.41, "learning_rate": 7.207508414516162e-06, "loss": 0.0715, "step": 90470 }, { "epoch": 43.42, "learning_rate": 7.20008372414353e-06, "loss": 0.0624, "step": 90480 }, { "epoch": 43.42, "learning_rate": 7.192662563228897e-06, "loss": 0.0719, "step": 90490 }, { "epoch": 43.43, "learning_rate": 7.185244932384239e-06, "loss": 0.071, "step": 90500 }, { "epoch": 43.43, "learning_rate": 7.177830832221255e-06, "loss": 0.0686, "step": 90510 }, { "epoch": 43.43, "learning_rate": 7.170420263351324e-06, "loss": 0.0708, "step": 90520 }, { "epoch": 43.44, "learning_rate": 7.163013226385562e-06, "loss": 0.0692, "step": 90530 }, { "epoch": 43.44, "learning_rate": 7.155609721934797e-06, "loss": 0.0747, "step": 90540 }, { "epoch": 43.45, "learning_rate": 7.1482097506095135e-06, "loss": 0.0618, "step": 90550 }, { "epoch": 43.45, "learning_rate": 7.140813313019984e-06, "loss": 0.072, "step": 90560 }, { "epoch": 43.46, "learning_rate": 7.133420409776131e-06, "loss": 0.0743, "step": 90570 }, { "epoch": 43.46, "learning_rate": 7.1260310414876185e-06, "loss": 0.0822, "step": 90580 }, { "epoch": 43.47, "learning_rate": 7.118645208763785e-06, "loss": 0.0702, "step": 90590 }, { "epoch": 43.47, "learning_rate": 7.1112629122137066e-06, "loss": 0.0801, "step": 90600 }, { "epoch": 43.48, "learning_rate": 7.10388415244615e-06, "loss": 0.0735, "step": 90610 }, { "epoch": 43.48, "learning_rate": 7.096508930069623e-06, "loss": 0.0647, "step": 90620 }, { "epoch": 43.48, "learning_rate": 7.089137245692273e-06, "loss": 0.0762, "step": 90630 }, { "epoch": 43.49, "learning_rate": 7.081769099922036e-06, "loss": 0.0767, "step": 90640 }, { "epoch": 43.49, "learning_rate": 7.074404493366521e-06, "loss": 0.0669, "step": 90650 }, { "epoch": 43.5, "learning_rate": 7.067043426633019e-06, "loss": 0.066, "step": 90660 }, { "epoch": 43.5, "learning_rate": 7.059685900328572e-06, "loss": 0.0799, "step": 90670 }, { "epoch": 43.51, "learning_rate": 7.052331915059901e-06, "loss": 0.075, "step": 90680 }, { "epoch": 43.51, "learning_rate": 7.0449814714334646e-06, "loss": 0.0745, "step": 90690 }, { "epoch": 43.52, "learning_rate": 7.03763457005539e-06, "loss": 0.0711, "step": 90700 }, { "epoch": 43.52, "learning_rate": 7.030291211531526e-06, "loss": 0.0791, "step": 90710 }, { "epoch": 43.53, "learning_rate": 7.022951396467481e-06, "loss": 0.0782, "step": 90720 }, { "epoch": 43.53, "learning_rate": 7.015615125468478e-06, "loss": 0.0821, "step": 90730 }, { "epoch": 43.54, "learning_rate": 7.008282399139515e-06, "loss": 0.0765, "step": 90740 }, { "epoch": 43.54, "learning_rate": 7.0009532180852826e-06, "loss": 0.0751, "step": 90750 }, { "epoch": 43.54, "learning_rate": 6.993627582910184e-06, "loss": 0.0747, "step": 90760 }, { "epoch": 43.55, "learning_rate": 6.986305494218298e-06, "loss": 0.0669, "step": 90770 }, { "epoch": 43.55, "learning_rate": 6.978986952613431e-06, "loss": 0.0848, "step": 90780 }, { "epoch": 43.56, "learning_rate": 6.971671958699138e-06, "loss": 0.0663, "step": 90790 }, { "epoch": 43.56, "learning_rate": 6.9643605130786106e-06, "loss": 0.0697, "step": 90800 }, { "epoch": 43.57, "learning_rate": 6.957052616354792e-06, "loss": 0.0728, "step": 90810 }, { "epoch": 43.57, "learning_rate": 6.949748269130313e-06, "loss": 0.0788, "step": 90820 }, { "epoch": 43.58, "learning_rate": 6.942447472007546e-06, "loss": 0.0685, "step": 90830 }, { "epoch": 43.58, "learning_rate": 6.93515022558851e-06, "loss": 0.0681, "step": 90840 }, { "epoch": 43.59, "learning_rate": 6.9278565304749854e-06, "loss": 0.0718, "step": 90850 }, { "epoch": 43.59, "learning_rate": 6.920566387268429e-06, "loss": 0.0732, "step": 90860 }, { "epoch": 43.59, "learning_rate": 6.913279796570043e-06, "loss": 0.0755, "step": 90870 }, { "epoch": 43.6, "learning_rate": 6.905996758980682e-06, "loss": 0.0754, "step": 90880 }, { "epoch": 43.6, "learning_rate": 6.898717275100941e-06, "loss": 0.0688, "step": 90890 }, { "epoch": 43.61, "learning_rate": 6.8914413455311356e-06, "loss": 0.0766, "step": 90900 }, { "epoch": 43.61, "learning_rate": 6.884168970871241e-06, "loss": 0.0726, "step": 90910 }, { "epoch": 43.62, "learning_rate": 6.876900151720977e-06, "loss": 0.0636, "step": 90920 }, { "epoch": 43.62, "learning_rate": 6.869634888679771e-06, "loss": 0.0711, "step": 90930 }, { "epoch": 43.63, "learning_rate": 6.862373182346732e-06, "loss": 0.071, "step": 90940 }, { "epoch": 43.63, "learning_rate": 6.855115033320697e-06, "loss": 0.0721, "step": 90950 }, { "epoch": 43.64, "learning_rate": 6.847860442200204e-06, "loss": 0.0722, "step": 90960 }, { "epoch": 43.64, "learning_rate": 6.840609409583509e-06, "loss": 0.0667, "step": 90970 }, { "epoch": 43.64, "learning_rate": 6.833361936068538e-06, "loss": 0.0647, "step": 90980 }, { "epoch": 43.65, "learning_rate": 6.826118022252959e-06, "loss": 0.0776, "step": 90990 }, { "epoch": 43.65, "learning_rate": 6.81887766873413e-06, "loss": 0.0712, "step": 91000 }, { "epoch": 43.66, "learning_rate": 6.811640876109126e-06, "loss": 0.0749, "step": 91010 }, { "epoch": 43.66, "learning_rate": 6.8044076449747255e-06, "loss": 0.0623, "step": 91020 }, { "epoch": 43.67, "learning_rate": 6.797177975927405e-06, "loss": 0.0802, "step": 91030 }, { "epoch": 43.67, "learning_rate": 6.789951869563355e-06, "loss": 0.0707, "step": 91040 }, { "epoch": 43.68, "learning_rate": 6.782729326478482e-06, "loss": 0.0823, "step": 91050 }, { "epoch": 43.68, "learning_rate": 6.7755103472683605e-06, "loss": 0.0655, "step": 91060 }, { "epoch": 43.69, "learning_rate": 6.768294932528313e-06, "loss": 0.0756, "step": 91070 }, { "epoch": 43.69, "learning_rate": 6.761083082853353e-06, "loss": 0.0689, "step": 91080 }, { "epoch": 43.7, "learning_rate": 6.753874798838194e-06, "loss": 0.0772, "step": 91090 }, { "epoch": 43.7, "learning_rate": 6.746670081077267e-06, "loss": 0.0683, "step": 91100 }, { "epoch": 43.7, "learning_rate": 6.7394689301647e-06, "loss": 0.0703, "step": 91110 }, { "epoch": 43.71, "learning_rate": 6.732271346694341e-06, "loss": 0.071, "step": 91120 }, { "epoch": 43.71, "learning_rate": 6.725077331259711e-06, "loss": 0.0742, "step": 91130 }, { "epoch": 43.72, "learning_rate": 6.717886884454066e-06, "loss": 0.0834, "step": 91140 }, { "epoch": 43.72, "learning_rate": 6.710700006870363e-06, "loss": 0.0758, "step": 91150 }, { "epoch": 43.73, "learning_rate": 6.7035166991012645e-06, "loss": 0.0821, "step": 91160 }, { "epoch": 43.73, "learning_rate": 6.696336961739125e-06, "loss": 0.07, "step": 91170 }, { "epoch": 43.74, "learning_rate": 6.689160795376031e-06, "loss": 0.0721, "step": 91180 }, { "epoch": 43.74, "learning_rate": 6.681988200603756e-06, "loss": 0.0752, "step": 91190 }, { "epoch": 43.75, "learning_rate": 6.674819178013769e-06, "loss": 0.0696, "step": 91200 }, { "epoch": 43.75, "learning_rate": 6.667653728197259e-06, "loss": 0.0677, "step": 91210 }, { "epoch": 43.75, "learning_rate": 6.660491851745132e-06, "loss": 0.0719, "step": 91220 }, { "epoch": 43.76, "learning_rate": 6.653333549247976e-06, "loss": 0.0781, "step": 91230 }, { "epoch": 43.76, "learning_rate": 6.64617882129609e-06, "loss": 0.0713, "step": 91240 }, { "epoch": 43.77, "learning_rate": 6.639027668479497e-06, "loss": 0.0769, "step": 91250 }, { "epoch": 43.77, "learning_rate": 6.63188009138791e-06, "loss": 0.0706, "step": 91260 }, { "epoch": 43.78, "learning_rate": 6.624736090610728e-06, "loss": 0.0733, "step": 91270 }, { "epoch": 43.78, "learning_rate": 6.6175956667370865e-06, "loss": 0.0751, "step": 91280 }, { "epoch": 43.79, "learning_rate": 6.610458820355814e-06, "loss": 0.071, "step": 91290 }, { "epoch": 43.79, "learning_rate": 6.6033255520554615e-06, "loss": 0.072, "step": 91300 }, { "epoch": 43.8, "learning_rate": 6.5961958624242255e-06, "loss": 0.0769, "step": 91310 }, { "epoch": 43.8, "learning_rate": 6.5890697520500864e-06, "loss": 0.0754, "step": 91320 }, { "epoch": 43.8, "learning_rate": 6.581947221520695e-06, "loss": 0.0701, "step": 91330 }, { "epoch": 43.81, "learning_rate": 6.5748282714233774e-06, "loss": 0.0758, "step": 91340 }, { "epoch": 43.81, "learning_rate": 6.567712902345208e-06, "loss": 0.071, "step": 91350 }, { "epoch": 43.82, "learning_rate": 6.5606011148729495e-06, "loss": 0.073, "step": 91360 }, { "epoch": 43.82, "learning_rate": 6.5534929095930785e-06, "loss": 0.0675, "step": 91370 }, { "epoch": 43.83, "learning_rate": 6.546388287091731e-06, "loss": 0.0695, "step": 91380 }, { "epoch": 43.83, "learning_rate": 6.539287247954817e-06, "loss": 0.0671, "step": 91390 }, { "epoch": 43.84, "learning_rate": 6.532189792767924e-06, "loss": 0.0655, "step": 91400 }, { "epoch": 43.84, "learning_rate": 6.5250959221163085e-06, "loss": 0.0691, "step": 91410 }, { "epoch": 43.85, "learning_rate": 6.518005636584973e-06, "loss": 0.066, "step": 91420 }, { "epoch": 43.85, "learning_rate": 6.5109189367586144e-06, "loss": 0.0703, "step": 91430 }, { "epoch": 43.86, "learning_rate": 6.5038358232216435e-06, "loss": 0.0674, "step": 91440 }, { "epoch": 43.86, "learning_rate": 6.496756296558137e-06, "loss": 0.0698, "step": 91450 }, { "epoch": 43.86, "learning_rate": 6.489680357351901e-06, "loss": 0.0671, "step": 91460 }, { "epoch": 43.87, "learning_rate": 6.482608006186469e-06, "loss": 0.077, "step": 91470 }, { "epoch": 43.87, "learning_rate": 6.475539243645062e-06, "loss": 0.0716, "step": 91480 }, { "epoch": 43.88, "learning_rate": 6.468474070310576e-06, "loss": 0.0779, "step": 91490 }, { "epoch": 43.88, "learning_rate": 6.461412486765639e-06, "loss": 0.0698, "step": 91500 }, { "epoch": 43.89, "learning_rate": 6.454354493592596e-06, "loss": 0.0712, "step": 91510 }, { "epoch": 43.89, "learning_rate": 6.447300091373448e-06, "loss": 0.0722, "step": 91520 }, { "epoch": 43.9, "learning_rate": 6.440249280689953e-06, "loss": 0.0718, "step": 91530 }, { "epoch": 43.9, "learning_rate": 6.4332020621235336e-06, "loss": 0.0735, "step": 91540 }, { "epoch": 43.91, "learning_rate": 6.426158436255358e-06, "loss": 0.0632, "step": 91550 }, { "epoch": 43.91, "learning_rate": 6.4191184036662524e-06, "loss": 0.0835, "step": 91560 }, { "epoch": 43.91, "learning_rate": 6.412081964936773e-06, "loss": 0.0711, "step": 91570 }, { "epoch": 43.92, "learning_rate": 6.405049120647177e-06, "loss": 0.0786, "step": 91580 }, { "epoch": 43.92, "learning_rate": 6.398019871377414e-06, "loss": 0.0671, "step": 91590 }, { "epoch": 43.93, "learning_rate": 6.390994217707141e-06, "loss": 0.07, "step": 91600 }, { "epoch": 43.93, "learning_rate": 6.383972160215734e-06, "loss": 0.077, "step": 91610 }, { "epoch": 43.94, "learning_rate": 6.376953699482257e-06, "loss": 0.0708, "step": 91620 }, { "epoch": 43.94, "learning_rate": 6.369938836085481e-06, "loss": 0.0719, "step": 91630 }, { "epoch": 43.95, "learning_rate": 6.362927570603882e-06, "loss": 0.0735, "step": 91640 }, { "epoch": 43.95, "learning_rate": 6.3559199036156366e-06, "loss": 0.1584, "step": 91650 }, { "epoch": 44.0, "learning_rate": 6.348915835698632e-06, "loss": 0.0707, "step": 91660 }, { "epoch": 44.01, "learning_rate": 6.34191536743044e-06, "loss": 0.0628, "step": 91670 }, { "epoch": 44.01, "learning_rate": 6.334918499388354e-06, "loss": 0.0654, "step": 91680 }, { "epoch": 44.02, "learning_rate": 6.3279252321493675e-06, "loss": 0.0715, "step": 91690 }, { "epoch": 44.02, "learning_rate": 6.32093556629017e-06, "loss": 0.0711, "step": 91700 }, { "epoch": 44.03, "learning_rate": 6.313949502387162e-06, "loss": 0.0768, "step": 91710 }, { "epoch": 44.03, "learning_rate": 6.306967041016437e-06, "loss": 0.0759, "step": 91720 }, { "epoch": 44.04, "learning_rate": 6.299988182753819e-06, "loss": 0.0718, "step": 91730 }, { "epoch": 44.04, "learning_rate": 6.293012928174779e-06, "loss": 0.069, "step": 91740 }, { "epoch": 44.04, "learning_rate": 6.286041277854549e-06, "loss": 0.071, "step": 91750 }, { "epoch": 44.05, "learning_rate": 6.2790732323680325e-06, "loss": 0.0653, "step": 91760 }, { "epoch": 44.05, "learning_rate": 6.272108792289844e-06, "loss": 0.0686, "step": 91770 }, { "epoch": 44.06, "learning_rate": 6.265147958194301e-06, "loss": 0.0648, "step": 91780 }, { "epoch": 44.06, "learning_rate": 6.258190730655422e-06, "loss": 0.0687, "step": 91790 }, { "epoch": 44.07, "learning_rate": 6.251237110246944e-06, "loss": 0.0785, "step": 91800 }, { "epoch": 44.07, "learning_rate": 6.244287097542262e-06, "loss": 0.0723, "step": 91810 }, { "epoch": 44.08, "learning_rate": 6.2373406931145185e-06, "loss": 0.0787, "step": 91820 }, { "epoch": 44.08, "learning_rate": 6.230397897536544e-06, "loss": 0.065, "step": 91830 }, { "epoch": 44.09, "learning_rate": 6.223458711380864e-06, "loss": 0.0709, "step": 91840 }, { "epoch": 44.09, "learning_rate": 6.216523135219715e-06, "loss": 0.0763, "step": 91850 }, { "epoch": 44.1, "learning_rate": 6.2095911696250394e-06, "loss": 0.0699, "step": 91860 }, { "epoch": 44.1, "learning_rate": 6.202662815168481e-06, "loss": 0.084, "step": 91870 }, { "epoch": 44.1, "learning_rate": 6.195738072421359e-06, "loss": 0.0668, "step": 91880 }, { "epoch": 44.11, "learning_rate": 6.1888169419547225e-06, "loss": 0.0754, "step": 91890 }, { "epoch": 44.11, "learning_rate": 6.181899424339327e-06, "loss": 0.0652, "step": 91900 }, { "epoch": 44.12, "learning_rate": 6.174985520145615e-06, "loss": 0.072, "step": 91910 }, { "epoch": 44.12, "learning_rate": 6.168075229943732e-06, "loss": 0.0693, "step": 91920 }, { "epoch": 44.13, "learning_rate": 6.161168554303537e-06, "loss": 0.0757, "step": 91930 }, { "epoch": 44.13, "learning_rate": 6.154265493794581e-06, "loss": 0.0692, "step": 91940 }, { "epoch": 44.14, "learning_rate": 6.1473660489861084e-06, "loss": 0.0681, "step": 91950 }, { "epoch": 44.14, "learning_rate": 6.140470220447087e-06, "loss": 0.0827, "step": 91960 }, { "epoch": 44.15, "learning_rate": 6.133578008746166e-06, "loss": 0.0652, "step": 91970 }, { "epoch": 44.15, "learning_rate": 6.126689414451719e-06, "loss": 0.0708, "step": 91980 }, { "epoch": 44.15, "learning_rate": 6.119804438131793e-06, "loss": 0.0742, "step": 91990 }, { "epoch": 44.16, "learning_rate": 6.112923080354166e-06, "loss": 0.0691, "step": 92000 }, { "epoch": 44.16, "learning_rate": 6.1060453416862895e-06, "loss": 0.0797, "step": 92010 }, { "epoch": 44.17, "learning_rate": 6.099171222695349e-06, "loss": 0.0754, "step": 92020 }, { "epoch": 44.17, "learning_rate": 6.092300723948191e-06, "loss": 0.0667, "step": 92030 }, { "epoch": 44.18, "learning_rate": 6.0854338460113955e-06, "loss": 0.076, "step": 92040 }, { "epoch": 44.18, "learning_rate": 6.078570589451244e-06, "loss": 0.0722, "step": 92050 }, { "epoch": 44.19, "learning_rate": 6.071710954833676e-06, "loss": 0.0729, "step": 92060 }, { "epoch": 44.19, "learning_rate": 6.064854942724402e-06, "loss": 0.0745, "step": 92070 }, { "epoch": 44.2, "learning_rate": 6.05800255368878e-06, "loss": 0.0747, "step": 92080 }, { "epoch": 44.2, "learning_rate": 6.051153788291897e-06, "loss": 0.0689, "step": 92090 }, { "epoch": 44.2, "learning_rate": 6.044308647098512e-06, "loss": 0.0673, "step": 92100 }, { "epoch": 44.21, "learning_rate": 6.037467130673119e-06, "loss": 0.0703, "step": 92110 }, { "epoch": 44.21, "learning_rate": 6.030629239579899e-06, "loss": 0.0777, "step": 92120 }, { "epoch": 44.22, "learning_rate": 6.023794974382713e-06, "loss": 0.067, "step": 92130 }, { "epoch": 44.22, "learning_rate": 6.016964335645153e-06, "loss": 0.0677, "step": 92140 }, { "epoch": 44.23, "learning_rate": 6.010137323930509e-06, "loss": 0.0684, "step": 92150 }, { "epoch": 44.23, "learning_rate": 6.003313939801775e-06, "loss": 0.0674, "step": 92160 }, { "epoch": 44.24, "learning_rate": 5.996494183821611e-06, "loss": 0.0719, "step": 92170 }, { "epoch": 44.24, "learning_rate": 5.989678056552412e-06, "loss": 0.0631, "step": 92180 }, { "epoch": 44.25, "learning_rate": 5.982865558556272e-06, "loss": 0.0724, "step": 92190 }, { "epoch": 44.25, "learning_rate": 5.97605669039496e-06, "loss": 0.0728, "step": 92200 }, { "epoch": 44.26, "learning_rate": 5.969251452629976e-06, "loss": 0.0644, "step": 92210 }, { "epoch": 44.26, "learning_rate": 5.962449845822493e-06, "loss": 0.064, "step": 92220 }, { "epoch": 44.26, "learning_rate": 5.95565187053343e-06, "loss": 0.0637, "step": 92230 }, { "epoch": 44.27, "learning_rate": 5.948857527323354e-06, "loss": 0.0763, "step": 92240 }, { "epoch": 44.27, "learning_rate": 5.942066816752556e-06, "loss": 0.0659, "step": 92250 }, { "epoch": 44.28, "learning_rate": 5.935279739381028e-06, "loss": 0.0688, "step": 92260 }, { "epoch": 44.28, "learning_rate": 5.9284962957684705e-06, "loss": 0.0709, "step": 92270 }, { "epoch": 44.29, "learning_rate": 5.92171648647426e-06, "loss": 0.0746, "step": 92280 }, { "epoch": 44.29, "learning_rate": 5.914940312057488e-06, "loss": 0.0762, "step": 92290 }, { "epoch": 44.3, "learning_rate": 5.90816777307695e-06, "loss": 0.0743, "step": 92300 }, { "epoch": 44.3, "learning_rate": 5.901398870091146e-06, "loss": 0.065, "step": 92310 }, { "epoch": 44.31, "learning_rate": 5.894633603658251e-06, "loss": 0.0762, "step": 92320 }, { "epoch": 44.31, "learning_rate": 5.887871974336168e-06, "loss": 0.0682, "step": 92330 }, { "epoch": 44.31, "learning_rate": 5.8811139826825015e-06, "loss": 0.0747, "step": 92340 }, { "epoch": 44.32, "learning_rate": 5.874359629254511e-06, "loss": 0.0757, "step": 92350 }, { "epoch": 44.32, "learning_rate": 5.867608914609213e-06, "loss": 0.0684, "step": 92360 }, { "epoch": 44.33, "learning_rate": 5.860861839303289e-06, "loss": 0.0821, "step": 92370 }, { "epoch": 44.33, "learning_rate": 5.854118403893133e-06, "loss": 0.077, "step": 92380 }, { "epoch": 44.34, "learning_rate": 5.847378608934839e-06, "loss": 0.0799, "step": 92390 }, { "epoch": 44.34, "learning_rate": 5.840642454984197e-06, "loss": 0.0718, "step": 92400 }, { "epoch": 44.35, "learning_rate": 5.833909942596705e-06, "loss": 0.0784, "step": 92410 }, { "epoch": 44.35, "learning_rate": 5.827181072327542e-06, "loss": 0.0694, "step": 92420 }, { "epoch": 44.36, "learning_rate": 5.820455844731598e-06, "loss": 0.076, "step": 92430 }, { "epoch": 44.36, "learning_rate": 5.813734260363474e-06, "loss": 0.0701, "step": 92440 }, { "epoch": 44.36, "learning_rate": 5.8070163197774476e-06, "loss": 0.0776, "step": 92450 }, { "epoch": 44.37, "learning_rate": 5.800302023527521e-06, "loss": 0.075, "step": 92460 }, { "epoch": 44.37, "learning_rate": 5.793591372167367e-06, "loss": 0.0692, "step": 92470 }, { "epoch": 44.38, "learning_rate": 5.786884366250401e-06, "loss": 0.0699, "step": 92480 }, { "epoch": 44.38, "learning_rate": 5.780181006329677e-06, "loss": 0.0664, "step": 92490 }, { "epoch": 44.39, "learning_rate": 5.7734812929579995e-06, "loss": 0.0774, "step": 92500 }, { "epoch": 44.39, "learning_rate": 5.766785226687854e-06, "loss": 0.0724, "step": 92510 }, { "epoch": 44.4, "learning_rate": 5.760092808071421e-06, "loss": 0.0744, "step": 92520 }, { "epoch": 44.4, "learning_rate": 5.7534040376605866e-06, "loss": 0.0715, "step": 92530 }, { "epoch": 44.41, "learning_rate": 5.746718916006938e-06, "loss": 0.0682, "step": 92540 }, { "epoch": 44.41, "learning_rate": 5.740037443661766e-06, "loss": 0.0648, "step": 92550 }, { "epoch": 44.41, "learning_rate": 5.733359621176032e-06, "loss": 0.0654, "step": 92560 }, { "epoch": 44.42, "learning_rate": 5.726685449100427e-06, "loss": 0.0674, "step": 92570 }, { "epoch": 44.42, "learning_rate": 5.720014927985329e-06, "loss": 0.0739, "step": 92580 }, { "epoch": 44.43, "learning_rate": 5.713348058380819e-06, "loss": 0.0721, "step": 92590 }, { "epoch": 44.43, "learning_rate": 5.706684840836674e-06, "loss": 0.0673, "step": 92600 }, { "epoch": 44.44, "learning_rate": 5.700025275902371e-06, "loss": 0.0738, "step": 92610 }, { "epoch": 44.44, "learning_rate": 5.693369364127088e-06, "loss": 0.0678, "step": 92620 }, { "epoch": 44.45, "learning_rate": 5.686717106059702e-06, "loss": 0.0668, "step": 92630 }, { "epoch": 44.45, "learning_rate": 5.68006850224877e-06, "loss": 0.073, "step": 92640 }, { "epoch": 44.46, "learning_rate": 5.673423553242568e-06, "loss": 0.0735, "step": 92650 }, { "epoch": 44.46, "learning_rate": 5.6667822595890765e-06, "loss": 0.0732, "step": 92660 }, { "epoch": 44.47, "learning_rate": 5.660144621835961e-06, "loss": 0.072, "step": 92670 }, { "epoch": 44.47, "learning_rate": 5.6535106405305795e-06, "loss": 0.067, "step": 92680 }, { "epoch": 44.47, "learning_rate": 5.646880316220004e-06, "loss": 0.0719, "step": 92690 }, { "epoch": 44.48, "learning_rate": 5.640253649451016e-06, "loss": 0.07, "step": 92700 }, { "epoch": 44.48, "learning_rate": 5.6336306407700435e-06, "loss": 0.0747, "step": 92710 }, { "epoch": 44.49, "learning_rate": 5.627011290723267e-06, "loss": 0.0651, "step": 92720 }, { "epoch": 44.49, "learning_rate": 5.620395599856548e-06, "loss": 0.0718, "step": 92730 }, { "epoch": 44.5, "learning_rate": 5.613783568715419e-06, "loss": 0.0649, "step": 92740 }, { "epoch": 44.5, "learning_rate": 5.607175197845166e-06, "loss": 0.0707, "step": 92750 }, { "epoch": 44.51, "learning_rate": 5.600570487790735e-06, "loss": 0.0759, "step": 92760 }, { "epoch": 44.51, "learning_rate": 5.59396943909678e-06, "loss": 0.0754, "step": 92770 }, { "epoch": 44.52, "learning_rate": 5.587372052307632e-06, "loss": 0.0667, "step": 92780 }, { "epoch": 44.52, "learning_rate": 5.580778327967351e-06, "loss": 0.0753, "step": 92790 }, { "epoch": 44.52, "learning_rate": 5.574188266619696e-06, "loss": 0.0729, "step": 92800 }, { "epoch": 44.53, "learning_rate": 5.567601868808087e-06, "loss": 0.0709, "step": 92810 }, { "epoch": 44.53, "learning_rate": 5.561019135075668e-06, "loss": 0.0757, "step": 92820 }, { "epoch": 44.54, "learning_rate": 5.554440065965294e-06, "loss": 0.0754, "step": 92830 }, { "epoch": 44.54, "learning_rate": 5.547864662019503e-06, "loss": 0.0752, "step": 92840 }, { "epoch": 44.55, "learning_rate": 5.541292923780517e-06, "loss": 0.0645, "step": 92850 }, { "epoch": 44.55, "learning_rate": 5.534724851790274e-06, "loss": 0.0706, "step": 92860 }, { "epoch": 44.56, "learning_rate": 5.528160446590397e-06, "loss": 0.0701, "step": 92870 }, { "epoch": 44.56, "learning_rate": 5.521599708722236e-06, "loss": 0.067, "step": 92880 }, { "epoch": 44.57, "learning_rate": 5.515042638726792e-06, "loss": 0.0732, "step": 92890 }, { "epoch": 44.57, "learning_rate": 5.508489237144787e-06, "loss": 0.0664, "step": 92900 }, { "epoch": 44.57, "learning_rate": 5.501939504516673e-06, "loss": 0.0738, "step": 92910 }, { "epoch": 44.58, "learning_rate": 5.495393441382535e-06, "loss": 0.0722, "step": 92920 }, { "epoch": 44.58, "learning_rate": 5.488851048282201e-06, "loss": 0.0678, "step": 92930 }, { "epoch": 44.59, "learning_rate": 5.482312325755185e-06, "loss": 0.0766, "step": 92940 }, { "epoch": 44.59, "learning_rate": 5.475777274340704e-06, "loss": 0.0723, "step": 92950 }, { "epoch": 44.6, "learning_rate": 5.469245894577646e-06, "loss": 0.0737, "step": 92960 }, { "epoch": 44.6, "learning_rate": 5.4627181870046284e-06, "loss": 0.0694, "step": 92970 }, { "epoch": 44.61, "learning_rate": 5.456194152159949e-06, "loss": 0.0704, "step": 92980 }, { "epoch": 44.61, "learning_rate": 5.449673790581611e-06, "loss": 0.0727, "step": 92990 }, { "epoch": 44.62, "learning_rate": 5.443157102807306e-06, "loss": 0.0777, "step": 93000 }, { "epoch": 44.62, "learning_rate": 5.436644089374432e-06, "loss": 0.0749, "step": 93010 }, { "epoch": 44.63, "learning_rate": 5.430134750820082e-06, "loss": 0.0796, "step": 93020 }, { "epoch": 44.63, "learning_rate": 5.423629087681026e-06, "loss": 0.0725, "step": 93030 }, { "epoch": 44.63, "learning_rate": 5.417127100493763e-06, "loss": 0.0826, "step": 93040 }, { "epoch": 44.64, "learning_rate": 5.41062878979447e-06, "loss": 0.0754, "step": 93050 }, { "epoch": 44.64, "learning_rate": 5.404134156119023e-06, "loss": 0.0686, "step": 93060 }, { "epoch": 44.65, "learning_rate": 5.397643200002994e-06, "loss": 0.0698, "step": 93070 }, { "epoch": 44.65, "learning_rate": 5.391155921981661e-06, "loss": 0.0728, "step": 93080 }, { "epoch": 44.66, "learning_rate": 5.384672322590001e-06, "loss": 0.0634, "step": 93090 }, { "epoch": 44.66, "learning_rate": 5.378192402362653e-06, "loss": 0.0734, "step": 93100 }, { "epoch": 44.67, "learning_rate": 5.37171616183399e-06, "loss": 0.0665, "step": 93110 }, { "epoch": 44.67, "learning_rate": 5.365243601538067e-06, "loss": 0.0733, "step": 93120 }, { "epoch": 44.68, "learning_rate": 5.358774722008647e-06, "loss": 0.0699, "step": 93130 }, { "epoch": 44.68, "learning_rate": 5.35230952377917e-06, "loss": 0.0659, "step": 93140 }, { "epoch": 44.68, "learning_rate": 5.345848007382792e-06, "loss": 0.0688, "step": 93150 }, { "epoch": 44.69, "learning_rate": 5.33939017335236e-06, "loss": 0.0716, "step": 93160 }, { "epoch": 44.69, "learning_rate": 5.332936022220397e-06, "loss": 0.0657, "step": 93170 }, { "epoch": 44.7, "learning_rate": 5.32648555451915e-06, "loss": 0.0707, "step": 93180 }, { "epoch": 44.7, "learning_rate": 5.320038770780544e-06, "loss": 0.0648, "step": 93190 }, { "epoch": 44.71, "learning_rate": 5.3135956715362205e-06, "loss": 0.0725, "step": 93200 }, { "epoch": 44.71, "learning_rate": 5.3071562573174915e-06, "loss": 0.0672, "step": 93210 }, { "epoch": 44.72, "learning_rate": 5.300720528655379e-06, "loss": 0.072, "step": 93220 }, { "epoch": 44.72, "learning_rate": 5.2942884860806064e-06, "loss": 0.0679, "step": 93230 }, { "epoch": 44.73, "learning_rate": 5.2878601301235955e-06, "loss": 0.0738, "step": 93240 }, { "epoch": 44.73, "learning_rate": 5.281435461314427e-06, "loss": 0.07, "step": 93250 }, { "epoch": 44.73, "learning_rate": 5.275014480182927e-06, "loss": 0.0699, "step": 93260 }, { "epoch": 44.74, "learning_rate": 5.268597187258589e-06, "loss": 0.0756, "step": 93270 }, { "epoch": 44.74, "learning_rate": 5.262183583070607e-06, "loss": 0.0725, "step": 93280 }, { "epoch": 44.75, "learning_rate": 5.255773668147884e-06, "loss": 0.0644, "step": 93290 }, { "epoch": 44.75, "learning_rate": 5.249367443018999e-06, "loss": 0.0704, "step": 93300 }, { "epoch": 44.76, "learning_rate": 5.242964908212244e-06, "loss": 0.0742, "step": 93310 }, { "epoch": 44.76, "learning_rate": 5.236566064255588e-06, "loss": 0.078, "step": 93320 }, { "epoch": 44.77, "learning_rate": 5.230170911676713e-06, "loss": 0.0716, "step": 93330 }, { "epoch": 44.77, "learning_rate": 5.223779451002986e-06, "loss": 0.069, "step": 93340 }, { "epoch": 44.78, "learning_rate": 5.2173916827614686e-06, "loss": 0.0693, "step": 93350 }, { "epoch": 44.78, "learning_rate": 5.211007607478935e-06, "loss": 0.0715, "step": 93360 }, { "epoch": 44.79, "learning_rate": 5.204627225681835e-06, "loss": 0.07, "step": 93370 }, { "epoch": 44.79, "learning_rate": 5.198250537896332e-06, "loss": 0.0746, "step": 93380 }, { "epoch": 44.79, "learning_rate": 5.191877544648255e-06, "loss": 0.0685, "step": 93390 }, { "epoch": 44.8, "learning_rate": 5.185508246463161e-06, "loss": 0.0699, "step": 93400 }, { "epoch": 44.8, "learning_rate": 5.179142643866286e-06, "loss": 0.0688, "step": 93410 }, { "epoch": 44.81, "learning_rate": 5.172780737382549e-06, "loss": 0.0736, "step": 93420 }, { "epoch": 44.81, "learning_rate": 5.166422527536602e-06, "loss": 0.065, "step": 93430 }, { "epoch": 44.82, "learning_rate": 5.160068014852759e-06, "loss": 0.0675, "step": 93440 }, { "epoch": 44.82, "learning_rate": 5.1537171998550515e-06, "loss": 0.0759, "step": 93450 }, { "epoch": 44.83, "learning_rate": 5.1473700830671715e-06, "loss": 0.0771, "step": 93460 }, { "epoch": 44.83, "learning_rate": 5.141026665012539e-06, "loss": 0.0763, "step": 93470 }, { "epoch": 44.84, "learning_rate": 5.134686946214257e-06, "loss": 0.074, "step": 93480 }, { "epoch": 44.84, "learning_rate": 5.1283509271951414e-06, "loss": 0.0636, "step": 93490 }, { "epoch": 44.84, "learning_rate": 5.122018608477652e-06, "loss": 0.0757, "step": 93500 }, { "epoch": 44.85, "learning_rate": 5.115689990584005e-06, "loss": 0.069, "step": 93510 }, { "epoch": 44.85, "learning_rate": 5.109365074036088e-06, "loss": 0.0699, "step": 93520 }, { "epoch": 44.86, "learning_rate": 5.103043859355461e-06, "loss": 0.0707, "step": 93530 }, { "epoch": 44.86, "learning_rate": 5.096726347063408e-06, "loss": 0.0799, "step": 93540 }, { "epoch": 44.87, "learning_rate": 5.090412537680894e-06, "loss": 0.0658, "step": 93550 }, { "epoch": 44.87, "learning_rate": 5.084102431728588e-06, "loss": 0.0668, "step": 93560 }, { "epoch": 44.88, "learning_rate": 5.077796029726839e-06, "loss": 0.0683, "step": 93570 }, { "epoch": 44.88, "learning_rate": 5.071493332195687e-06, "loss": 0.0705, "step": 93580 }, { "epoch": 44.89, "learning_rate": 5.065194339654916e-06, "loss": 0.0713, "step": 93590 }, { "epoch": 44.89, "learning_rate": 5.058899052623933e-06, "loss": 0.0782, "step": 93600 }, { "epoch": 44.89, "learning_rate": 5.052607471621889e-06, "loss": 0.0649, "step": 93610 }, { "epoch": 44.9, "learning_rate": 5.046319597167609e-06, "loss": 0.0731, "step": 93620 }, { "epoch": 44.9, "learning_rate": 5.0400354297796215e-06, "loss": 0.0765, "step": 93630 }, { "epoch": 44.91, "learning_rate": 5.03375496997614e-06, "loss": 0.0805, "step": 93640 }, { "epoch": 44.91, "learning_rate": 5.027478218275072e-06, "loss": 0.0772, "step": 93650 }, { "epoch": 44.92, "learning_rate": 5.0212051751940274e-06, "loss": 0.0669, "step": 93660 }, { "epoch": 44.92, "learning_rate": 5.014935841250324e-06, "loss": 0.0698, "step": 93670 }, { "epoch": 44.93, "learning_rate": 5.008670216960936e-06, "loss": 0.0723, "step": 93680 }, { "epoch": 44.93, "learning_rate": 5.002408302842565e-06, "loss": 0.0814, "step": 93690 }, { "epoch": 44.94, "learning_rate": 4.996150099411595e-06, "loss": 0.0754, "step": 93700 }, { "epoch": 44.94, "learning_rate": 4.989895607184087e-06, "loss": 0.0731, "step": 93710 }, { "epoch": 44.95, "learning_rate": 4.983644826675826e-06, "loss": 0.0827, "step": 93720 }, { "epoch": 44.95, "learning_rate": 4.977397758402263e-06, "loss": 0.0722, "step": 93730 }, { "epoch": 45.0, "learning_rate": 4.971154402878592e-06, "loss": 0.0716, "step": 93740 }, { "epoch": 45.01, "learning_rate": 4.9649147606196265e-06, "loss": 0.0685, "step": 93750 }, { "epoch": 45.01, "learning_rate": 4.958678832139935e-06, "loss": 0.0801, "step": 93760 }, { "epoch": 45.02, "learning_rate": 4.95244661795376e-06, "loss": 0.0729, "step": 93770 }, { "epoch": 45.02, "learning_rate": 4.9462181185750185e-06, "loss": 0.0617, "step": 93780 }, { "epoch": 45.03, "learning_rate": 4.939993334517346e-06, "loss": 0.0702, "step": 93790 }, { "epoch": 45.03, "learning_rate": 4.933772266294068e-06, "loss": 0.068, "step": 93800 }, { "epoch": 45.03, "learning_rate": 4.927554914418192e-06, "loss": 0.0673, "step": 93810 }, { "epoch": 45.04, "learning_rate": 4.921341279402436e-06, "loss": 0.0791, "step": 93820 }, { "epoch": 45.04, "learning_rate": 4.915131361759201e-06, "loss": 0.0743, "step": 93830 }, { "epoch": 45.05, "learning_rate": 4.908925162000572e-06, "loss": 0.0725, "step": 93840 }, { "epoch": 45.05, "learning_rate": 4.902722680638355e-06, "loss": 0.0645, "step": 93850 }, { "epoch": 45.06, "learning_rate": 4.896523918184015e-06, "loss": 0.0756, "step": 93860 }, { "epoch": 45.06, "learning_rate": 4.890328875148737e-06, "loss": 0.073, "step": 93870 }, { "epoch": 45.07, "learning_rate": 4.884137552043389e-06, "loss": 0.0705, "step": 93880 }, { "epoch": 45.07, "learning_rate": 4.877949949378524e-06, "loss": 0.0896, "step": 93890 }, { "epoch": 45.08, "learning_rate": 4.871766067664413e-06, "loss": 0.0792, "step": 93900 }, { "epoch": 45.08, "learning_rate": 4.865585907410991e-06, "loss": 0.072, "step": 93910 }, { "epoch": 45.08, "learning_rate": 4.859409469127912e-06, "loss": 0.0778, "step": 93920 }, { "epoch": 45.09, "learning_rate": 4.853236753324503e-06, "loss": 0.0694, "step": 93930 }, { "epoch": 45.09, "learning_rate": 4.847067760509788e-06, "loss": 0.0716, "step": 93940 }, { "epoch": 45.1, "learning_rate": 4.8409024911924894e-06, "loss": 0.0718, "step": 93950 }, { "epoch": 45.1, "learning_rate": 4.834740945881028e-06, "loss": 0.0796, "step": 93960 }, { "epoch": 45.11, "learning_rate": 4.8285831250835025e-06, "loss": 0.0733, "step": 93970 }, { "epoch": 45.11, "learning_rate": 4.822429029307713e-06, "loss": 0.0784, "step": 93980 }, { "epoch": 45.12, "learning_rate": 4.816278659061168e-06, "loss": 0.0698, "step": 93990 }, { "epoch": 45.12, "learning_rate": 4.810132014851027e-06, "loss": 0.0663, "step": 94000 }, { "epoch": 45.13, "learning_rate": 4.80398909718418e-06, "loss": 0.0736, "step": 94010 }, { "epoch": 45.13, "learning_rate": 4.797849906567197e-06, "loss": 0.066, "step": 94020 }, { "epoch": 45.13, "learning_rate": 4.7917144435063396e-06, "loss": 0.0674, "step": 94030 }, { "epoch": 45.14, "learning_rate": 4.785582708507563e-06, "loss": 0.0788, "step": 94040 }, { "epoch": 45.14, "learning_rate": 4.779454702076519e-06, "loss": 0.0797, "step": 94050 }, { "epoch": 45.15, "learning_rate": 4.773330424718558e-06, "loss": 0.0741, "step": 94060 }, { "epoch": 45.15, "learning_rate": 4.767209876938688e-06, "loss": 0.0737, "step": 94070 }, { "epoch": 45.16, "learning_rate": 4.76109305924165e-06, "loss": 0.0748, "step": 94080 }, { "epoch": 45.16, "learning_rate": 4.754979972131857e-06, "loss": 0.0733, "step": 94090 }, { "epoch": 45.17, "learning_rate": 4.748870616113427e-06, "loss": 0.069, "step": 94100 }, { "epoch": 45.17, "learning_rate": 4.742764991690158e-06, "loss": 0.0733, "step": 94110 }, { "epoch": 45.18, "learning_rate": 4.7366630993655404e-06, "loss": 0.08, "step": 94120 }, { "epoch": 45.18, "learning_rate": 4.730564939642779e-06, "loss": 0.0652, "step": 94130 }, { "epoch": 45.19, "learning_rate": 4.724470513024731e-06, "loss": 0.0676, "step": 94140 }, { "epoch": 45.19, "learning_rate": 4.718379820013979e-06, "loss": 0.0742, "step": 94150 }, { "epoch": 45.19, "learning_rate": 4.712292861112788e-06, "loss": 0.0731, "step": 94160 }, { "epoch": 45.2, "learning_rate": 4.706209636823117e-06, "loss": 0.0762, "step": 94170 }, { "epoch": 45.2, "learning_rate": 4.700130147646586e-06, "loss": 0.0686, "step": 94180 }, { "epoch": 45.21, "learning_rate": 4.694054394084574e-06, "loss": 0.0725, "step": 94190 }, { "epoch": 45.21, "learning_rate": 4.687982376638101e-06, "loss": 0.0736, "step": 94200 }, { "epoch": 45.22, "learning_rate": 4.681914095807876e-06, "loss": 0.0661, "step": 94210 }, { "epoch": 45.22, "learning_rate": 4.675849552094319e-06, "loss": 0.0738, "step": 94220 }, { "epoch": 45.23, "learning_rate": 4.66978874599755e-06, "loss": 0.067, "step": 94230 }, { "epoch": 45.23, "learning_rate": 4.663731678017363e-06, "loss": 0.0716, "step": 94240 }, { "epoch": 45.24, "learning_rate": 4.657678348653244e-06, "loss": 0.0892, "step": 94250 }, { "epoch": 45.24, "learning_rate": 4.651628758404364e-06, "loss": 0.0695, "step": 94260 }, { "epoch": 45.24, "learning_rate": 4.645582907769619e-06, "loss": 0.0693, "step": 94270 }, { "epoch": 45.25, "learning_rate": 4.639540797247577e-06, "loss": 0.0647, "step": 94280 }, { "epoch": 45.25, "learning_rate": 4.633502427336478e-06, "loss": 0.0779, "step": 94290 }, { "epoch": 45.26, "learning_rate": 4.627467798534274e-06, "loss": 0.0702, "step": 94300 }, { "epoch": 45.26, "learning_rate": 4.621436911338623e-06, "loss": 0.0769, "step": 94310 }, { "epoch": 45.27, "learning_rate": 4.615409766246831e-06, "loss": 0.0683, "step": 94320 }, { "epoch": 45.27, "learning_rate": 4.6093863637559244e-06, "loss": 0.0781, "step": 94330 }, { "epoch": 45.28, "learning_rate": 4.603366704362639e-06, "loss": 0.0648, "step": 94340 }, { "epoch": 45.28, "learning_rate": 4.597350788563376e-06, "loss": 0.0606, "step": 94350 }, { "epoch": 45.29, "learning_rate": 4.591338616854213e-06, "loss": 0.073, "step": 94360 }, { "epoch": 45.29, "learning_rate": 4.585330189730957e-06, "loss": 0.0747, "step": 94370 }, { "epoch": 45.29, "learning_rate": 4.5793255076890865e-06, "loss": 0.0672, "step": 94380 }, { "epoch": 45.3, "learning_rate": 4.5733245712237634e-06, "loss": 0.0689, "step": 94390 }, { "epoch": 45.3, "learning_rate": 4.56732738082985e-06, "loss": 0.0724, "step": 94400 }, { "epoch": 45.31, "learning_rate": 4.5613339370018995e-06, "loss": 0.074, "step": 94410 }, { "epoch": 45.31, "learning_rate": 4.555344240234171e-06, "loss": 0.0746, "step": 94420 }, { "epoch": 45.32, "learning_rate": 4.549358291020583e-06, "loss": 0.0719, "step": 94430 }, { "epoch": 45.32, "learning_rate": 4.543376089854768e-06, "loss": 0.0712, "step": 94440 }, { "epoch": 45.33, "learning_rate": 4.5373976372300454e-06, "loss": 0.0764, "step": 94450 }, { "epoch": 45.33, "learning_rate": 4.531422933639423e-06, "loss": 0.0693, "step": 94460 }, { "epoch": 45.34, "learning_rate": 4.52545197957559e-06, "loss": 0.0757, "step": 94470 }, { "epoch": 45.34, "learning_rate": 4.5194847755309415e-06, "loss": 0.0729, "step": 94480 }, { "epoch": 45.35, "learning_rate": 4.513521321997561e-06, "loss": 0.0693, "step": 94490 }, { "epoch": 45.35, "learning_rate": 4.507561619467215e-06, "loss": 0.0732, "step": 94500 }, { "epoch": 45.35, "learning_rate": 4.5016056684313744e-06, "loss": 0.0634, "step": 94510 }, { "epoch": 45.36, "learning_rate": 4.495653469381178e-06, "loss": 0.073, "step": 94520 }, { "epoch": 45.36, "learning_rate": 4.4897050228074835e-06, "loss": 0.0634, "step": 94530 }, { "epoch": 45.37, "learning_rate": 4.483760329200809e-06, "loss": 0.0734, "step": 94540 }, { "epoch": 45.37, "learning_rate": 4.477819389051391e-06, "loss": 0.0775, "step": 94550 }, { "epoch": 45.38, "learning_rate": 4.471882202849131e-06, "loss": 0.0636, "step": 94560 }, { "epoch": 45.38, "learning_rate": 4.465948771083644e-06, "loss": 0.0665, "step": 94570 }, { "epoch": 45.39, "learning_rate": 4.460019094244228e-06, "loss": 0.0778, "step": 94580 }, { "epoch": 45.39, "learning_rate": 4.454093172819856e-06, "loss": 0.0675, "step": 94590 }, { "epoch": 45.4, "learning_rate": 4.448171007299229e-06, "loss": 0.0665, "step": 94600 }, { "epoch": 45.4, "learning_rate": 4.442252598170682e-06, "loss": 0.0723, "step": 94610 }, { "epoch": 45.4, "learning_rate": 4.436337945922287e-06, "loss": 0.0691, "step": 94620 }, { "epoch": 45.41, "learning_rate": 4.430427051041786e-06, "loss": 0.0761, "step": 94630 }, { "epoch": 45.41, "learning_rate": 4.424519914016623e-06, "loss": 0.0634, "step": 94640 }, { "epoch": 45.42, "learning_rate": 4.418616535333919e-06, "loss": 0.0729, "step": 94650 }, { "epoch": 45.42, "learning_rate": 4.412716915480492e-06, "loss": 0.0693, "step": 94660 }, { "epoch": 45.43, "learning_rate": 4.406821054942861e-06, "loss": 0.0734, "step": 94670 }, { "epoch": 45.43, "learning_rate": 4.400928954207201e-06, "loss": 0.0759, "step": 94680 }, { "epoch": 45.44, "learning_rate": 4.39504061375941e-06, "loss": 0.0633, "step": 94690 }, { "epoch": 45.44, "learning_rate": 4.389156034085062e-06, "loss": 0.075, "step": 94700 }, { "epoch": 45.45, "learning_rate": 4.38327521566943e-06, "loss": 0.0705, "step": 94710 }, { "epoch": 45.45, "learning_rate": 4.37739815899747e-06, "loss": 0.0804, "step": 94720 }, { "epoch": 45.45, "learning_rate": 4.371524864553822e-06, "loss": 0.069, "step": 94730 }, { "epoch": 45.46, "learning_rate": 4.365655332822838e-06, "loss": 0.0727, "step": 94740 }, { "epoch": 45.46, "learning_rate": 4.359789564288519e-06, "loss": 0.0752, "step": 94750 }, { "epoch": 45.47, "learning_rate": 4.353927559434589e-06, "loss": 0.0672, "step": 94760 }, { "epoch": 45.47, "learning_rate": 4.348069318744463e-06, "loss": 0.0739, "step": 94770 }, { "epoch": 45.48, "learning_rate": 4.3422148427012314e-06, "loss": 0.0747, "step": 94780 }, { "epoch": 45.48, "learning_rate": 4.336364131787674e-06, "loss": 0.0651, "step": 94790 }, { "epoch": 45.49, "learning_rate": 4.330517186486266e-06, "loss": 0.0685, "step": 94800 }, { "epoch": 45.49, "learning_rate": 4.3246740072791835e-06, "loss": 0.075, "step": 94810 }, { "epoch": 45.5, "learning_rate": 4.318834594648258e-06, "loss": 0.0694, "step": 94820 }, { "epoch": 45.5, "learning_rate": 4.312998949075042e-06, "loss": 0.0712, "step": 94830 }, { "epoch": 45.51, "learning_rate": 4.307167071040768e-06, "loss": 0.0695, "step": 94840 }, { "epoch": 45.51, "learning_rate": 4.301338961026363e-06, "loss": 0.075, "step": 94850 }, { "epoch": 45.51, "learning_rate": 4.295514619512414e-06, "loss": 0.0707, "step": 94860 }, { "epoch": 45.52, "learning_rate": 4.2896940469792425e-06, "loss": 0.07, "step": 94870 }, { "epoch": 45.52, "learning_rate": 4.283877243906831e-06, "loss": 0.0716, "step": 94880 }, { "epoch": 45.53, "learning_rate": 4.2780642107748625e-06, "loss": 0.0783, "step": 94890 }, { "epoch": 45.53, "learning_rate": 4.2722549480626925e-06, "loss": 0.0707, "step": 94900 }, { "epoch": 45.54, "learning_rate": 4.266449456249378e-06, "loss": 0.0725, "step": 94910 }, { "epoch": 45.54, "learning_rate": 4.260647735813678e-06, "loss": 0.0716, "step": 94920 }, { "epoch": 45.55, "learning_rate": 4.254849787234e-06, "loss": 0.074, "step": 94930 }, { "epoch": 45.55, "learning_rate": 4.249055610988489e-06, "loss": 0.0768, "step": 94940 }, { "epoch": 45.56, "learning_rate": 4.243265207554947e-06, "loss": 0.068, "step": 94950 }, { "epoch": 45.56, "learning_rate": 4.237478577410892e-06, "loss": 0.0674, "step": 94960 }, { "epoch": 45.56, "learning_rate": 4.2316957210334905e-06, "loss": 0.0695, "step": 94970 }, { "epoch": 45.57, "learning_rate": 4.2259166388996295e-06, "loss": 0.0694, "step": 94980 }, { "epoch": 45.57, "learning_rate": 4.220141331485888e-06, "loss": 0.0706, "step": 94990 }, { "epoch": 45.58, "learning_rate": 4.214369799268497e-06, "loss": 0.0696, "step": 95000 }, { "epoch": 45.58, "learning_rate": 4.208602042723409e-06, "loss": 0.0637, "step": 95010 }, { "epoch": 45.59, "learning_rate": 4.202838062326264e-06, "loss": 0.0682, "step": 95020 }, { "epoch": 45.59, "learning_rate": 4.197077858552395e-06, "loss": 0.0738, "step": 95030 }, { "epoch": 45.6, "learning_rate": 4.191321431876788e-06, "loss": 0.0721, "step": 95040 }, { "epoch": 45.6, "learning_rate": 4.1855687827741575e-06, "loss": 0.0712, "step": 95050 }, { "epoch": 45.61, "learning_rate": 4.1798199117188795e-06, "loss": 0.0664, "step": 95060 }, { "epoch": 45.61, "learning_rate": 4.174074819185048e-06, "loss": 0.0671, "step": 95070 }, { "epoch": 45.61, "learning_rate": 4.168333505646404e-06, "loss": 0.066, "step": 95080 }, { "epoch": 45.62, "learning_rate": 4.162595971576405e-06, "loss": 0.0734, "step": 95090 }, { "epoch": 45.62, "learning_rate": 4.1568622174482155e-06, "loss": 0.0752, "step": 95100 }, { "epoch": 45.63, "learning_rate": 4.151132243734634e-06, "loss": 0.073, "step": 95110 }, { "epoch": 45.63, "learning_rate": 4.145406050908196e-06, "loss": 0.0666, "step": 95120 }, { "epoch": 45.64, "learning_rate": 4.1396836394411e-06, "loss": 0.0707, "step": 95130 }, { "epoch": 45.64, "learning_rate": 4.1339650098052464e-06, "loss": 0.0647, "step": 95140 }, { "epoch": 45.65, "learning_rate": 4.128250162472208e-06, "loss": 0.074, "step": 95150 }, { "epoch": 45.65, "learning_rate": 4.122539097913253e-06, "loss": 0.0732, "step": 95160 }, { "epoch": 45.66, "learning_rate": 4.116831816599348e-06, "loss": 0.0692, "step": 95170 }, { "epoch": 45.66, "learning_rate": 4.1111283190011395e-06, "loss": 0.0748, "step": 95180 }, { "epoch": 45.66, "learning_rate": 4.105428605588956e-06, "loss": 0.0699, "step": 95190 }, { "epoch": 45.67, "learning_rate": 4.099732676832818e-06, "loss": 0.0683, "step": 95200 }, { "epoch": 45.67, "learning_rate": 4.094040533202453e-06, "loss": 0.0745, "step": 95210 }, { "epoch": 45.68, "learning_rate": 4.088352175167232e-06, "loss": 0.0748, "step": 95220 }, { "epoch": 45.68, "learning_rate": 4.082667603196255e-06, "loss": 0.0714, "step": 95230 }, { "epoch": 45.69, "learning_rate": 4.0769868177582936e-06, "loss": 0.0696, "step": 95240 }, { "epoch": 45.69, "learning_rate": 4.071309819321806e-06, "loss": 0.0637, "step": 95250 }, { "epoch": 45.7, "learning_rate": 4.065636608354944e-06, "loss": 0.065, "step": 95260 }, { "epoch": 45.7, "learning_rate": 4.059967185325547e-06, "loss": 0.0745, "step": 95270 }, { "epoch": 45.71, "learning_rate": 4.054301550701139e-06, "loss": 0.0733, "step": 95280 }, { "epoch": 45.71, "learning_rate": 4.048639704948926e-06, "loss": 0.0708, "step": 95290 }, { "epoch": 45.72, "learning_rate": 4.042981648535804e-06, "loss": 0.0661, "step": 95300 }, { "epoch": 45.72, "learning_rate": 4.037327381928369e-06, "loss": 0.075, "step": 95310 }, { "epoch": 45.72, "learning_rate": 4.03167690559289e-06, "loss": 0.0801, "step": 95320 }, { "epoch": 45.73, "learning_rate": 4.02603021999533e-06, "loss": 0.0715, "step": 95330 }, { "epoch": 45.73, "learning_rate": 4.020387325601338e-06, "loss": 0.0738, "step": 95340 }, { "epoch": 45.74, "learning_rate": 4.014748222876258e-06, "loss": 0.0816, "step": 95350 }, { "epoch": 45.74, "learning_rate": 4.009112912285096e-06, "loss": 0.0817, "step": 95360 }, { "epoch": 45.75, "learning_rate": 4.0034813942925744e-06, "loss": 0.0681, "step": 95370 }, { "epoch": 45.75, "learning_rate": 3.997853669363094e-06, "loss": 0.0754, "step": 95380 }, { "epoch": 45.76, "learning_rate": 3.992229737960735e-06, "loss": 0.0689, "step": 95390 }, { "epoch": 45.76, "learning_rate": 3.986609600549268e-06, "loss": 0.0725, "step": 95400 }, { "epoch": 45.77, "learning_rate": 3.980993257592158e-06, "loss": 0.0629, "step": 95410 }, { "epoch": 45.77, "learning_rate": 3.975380709552562e-06, "loss": 0.0744, "step": 95420 }, { "epoch": 45.77, "learning_rate": 3.969771956893287e-06, "loss": 0.0646, "step": 95430 }, { "epoch": 45.78, "learning_rate": 3.964167000076874e-06, "loss": 0.0688, "step": 95440 }, { "epoch": 45.78, "learning_rate": 3.958565839565526e-06, "loss": 0.0679, "step": 95450 }, { "epoch": 45.79, "learning_rate": 3.952968475821133e-06, "loss": 0.0678, "step": 95460 }, { "epoch": 45.79, "learning_rate": 3.94737490930529e-06, "loss": 0.0692, "step": 95470 }, { "epoch": 45.8, "learning_rate": 3.941785140479249e-06, "loss": 0.072, "step": 95480 }, { "epoch": 45.8, "learning_rate": 3.936199169803978e-06, "loss": 0.0695, "step": 95490 }, { "epoch": 45.81, "learning_rate": 3.930616997740122e-06, "loss": 0.0707, "step": 95500 }, { "epoch": 45.81, "learning_rate": 3.925038624747995e-06, "loss": 0.0667, "step": 95510 }, { "epoch": 45.82, "learning_rate": 3.919464051287619e-06, "loss": 0.0696, "step": 95520 }, { "epoch": 45.82, "learning_rate": 3.913893277818698e-06, "loss": 0.0743, "step": 95530 }, { "epoch": 45.82, "learning_rate": 3.908326304800619e-06, "loss": 0.0697, "step": 95540 }, { "epoch": 45.83, "learning_rate": 3.902763132692461e-06, "loss": 0.0765, "step": 95550 }, { "epoch": 45.83, "learning_rate": 3.897203761952983e-06, "loss": 0.0725, "step": 95560 }, { "epoch": 45.84, "learning_rate": 3.891648193040648e-06, "loss": 0.0756, "step": 95570 }, { "epoch": 45.84, "learning_rate": 3.886096426413566e-06, "loss": 0.07, "step": 95580 }, { "epoch": 45.85, "learning_rate": 3.880548462529576e-06, "loss": 0.0739, "step": 95590 }, { "epoch": 45.85, "learning_rate": 3.8750043018461855e-06, "loss": 0.064, "step": 95600 }, { "epoch": 45.86, "learning_rate": 3.869463944820567e-06, "loss": 0.0725, "step": 95610 }, { "epoch": 45.86, "learning_rate": 3.863927391909627e-06, "loss": 0.0725, "step": 95620 }, { "epoch": 45.87, "learning_rate": 3.858394643569929e-06, "loss": 0.0789, "step": 95630 }, { "epoch": 45.87, "learning_rate": 3.852865700257724e-06, "loss": 0.0761, "step": 95640 }, { "epoch": 45.88, "learning_rate": 3.847340562428942e-06, "loss": 0.0711, "step": 95650 }, { "epoch": 45.88, "learning_rate": 3.841819230539218e-06, "loss": 0.0727, "step": 95660 }, { "epoch": 45.88, "learning_rate": 3.83630170504386e-06, "loss": 0.0643, "step": 95670 }, { "epoch": 45.89, "learning_rate": 3.830787986397877e-06, "loss": 0.0717, "step": 95680 }, { "epoch": 45.89, "learning_rate": 3.825278075055927e-06, "loss": 0.0719, "step": 95690 }, { "epoch": 45.9, "learning_rate": 3.8197719714724075e-06, "loss": 0.0704, "step": 95700 }, { "epoch": 45.9, "learning_rate": 3.8142696761013676e-06, "loss": 0.0781, "step": 95710 }, { "epoch": 45.91, "learning_rate": 3.808771189396537e-06, "loss": 0.0698, "step": 95720 }, { "epoch": 45.91, "learning_rate": 3.8032765118113543e-06, "loss": 0.0713, "step": 95730 }, { "epoch": 45.92, "learning_rate": 3.7977856437989346e-06, "loss": 0.0733, "step": 95740 }, { "epoch": 45.92, "learning_rate": 3.7922985858120763e-06, "loss": 0.0714, "step": 95750 }, { "epoch": 45.93, "learning_rate": 3.7868153383032567e-06, "loss": 0.0808, "step": 95760 }, { "epoch": 45.93, "learning_rate": 3.781335901724642e-06, "loss": 0.078, "step": 95770 }, { "epoch": 45.93, "learning_rate": 3.7758602765281204e-06, "loss": 0.0684, "step": 95780 }, { "epoch": 45.94, "learning_rate": 3.770388463165209e-06, "loss": 0.0778, "step": 95790 }, { "epoch": 45.94, "learning_rate": 3.7649204620871348e-06, "loss": 0.0741, "step": 95800 }, { "epoch": 45.95, "learning_rate": 3.7594562737448213e-06, "loss": 0.0708, "step": 95810 }, { "epoch": 46.0, "learning_rate": 3.7539958985888744e-06, "loss": 0.0793, "step": 95820 }, { "epoch": 46.01, "learning_rate": 3.7485393370695622e-06, "loss": 0.0646, "step": 95830 }, { "epoch": 46.01, "learning_rate": 3.7430865896368626e-06, "loss": 0.0683, "step": 95840 }, { "epoch": 46.01, "learning_rate": 3.7376376567404227e-06, "loss": 0.0683, "step": 95850 }, { "epoch": 46.02, "learning_rate": 3.7321925388296098e-06, "loss": 0.0742, "step": 95860 }, { "epoch": 46.02, "learning_rate": 3.7267512363534264e-06, "loss": 0.0739, "step": 95870 }, { "epoch": 46.03, "learning_rate": 3.7213137497605967e-06, "loss": 0.0704, "step": 95880 }, { "epoch": 46.03, "learning_rate": 3.7158800794995175e-06, "loss": 0.0747, "step": 95890 }, { "epoch": 46.04, "learning_rate": 3.7104502260182637e-06, "loss": 0.0742, "step": 95900 }, { "epoch": 46.04, "learning_rate": 3.7050241897646108e-06, "loss": 0.0727, "step": 95910 }, { "epoch": 46.05, "learning_rate": 3.6996019711860065e-06, "loss": 0.08, "step": 95920 }, { "epoch": 46.05, "learning_rate": 3.6941835707295923e-06, "loss": 0.0663, "step": 95930 }, { "epoch": 46.06, "learning_rate": 3.688768988842195e-06, "loss": 0.0798, "step": 95940 }, { "epoch": 46.06, "learning_rate": 3.683358225970318e-06, "loss": 0.066, "step": 95950 }, { "epoch": 46.06, "learning_rate": 3.6779512825601715e-06, "loss": 0.0719, "step": 95960 }, { "epoch": 46.07, "learning_rate": 3.6725481590576096e-06, "loss": 0.0761, "step": 95970 }, { "epoch": 46.07, "learning_rate": 3.6671488559082034e-06, "loss": 0.072, "step": 95980 }, { "epoch": 46.08, "learning_rate": 3.6617533735572083e-06, "loss": 0.0647, "step": 95990 }, { "epoch": 46.08, "learning_rate": 3.6563617124495575e-06, "loss": 0.073, "step": 96000 }, { "epoch": 46.09, "learning_rate": 3.650973873029867e-06, "loss": 0.0756, "step": 96010 }, { "epoch": 46.09, "learning_rate": 3.645589855742443e-06, "loss": 0.073, "step": 96020 }, { "epoch": 46.1, "learning_rate": 3.6402096610312743e-06, "loss": 0.0701, "step": 96030 }, { "epoch": 46.1, "learning_rate": 3.634833289340034e-06, "loss": 0.0666, "step": 96040 }, { "epoch": 46.11, "learning_rate": 3.629460741112073e-06, "loss": 0.0785, "step": 96050 }, { "epoch": 46.11, "learning_rate": 3.6240920167904312e-06, "loss": 0.0718, "step": 96060 }, { "epoch": 46.12, "learning_rate": 3.6187271168178496e-06, "loss": 0.0666, "step": 96070 }, { "epoch": 46.12, "learning_rate": 3.6133660416367345e-06, "loss": 0.078, "step": 96080 }, { "epoch": 46.12, "learning_rate": 3.6080087916891767e-06, "loss": 0.0773, "step": 96090 }, { "epoch": 46.13, "learning_rate": 3.6026553674169683e-06, "loss": 0.072, "step": 96100 }, { "epoch": 46.13, "learning_rate": 3.5973057692615717e-06, "loss": 0.0706, "step": 96110 }, { "epoch": 46.14, "learning_rate": 3.591959997664124e-06, "loss": 0.0743, "step": 96120 }, { "epoch": 46.14, "learning_rate": 3.5866180530654712e-06, "loss": 0.0673, "step": 96130 }, { "epoch": 46.15, "learning_rate": 3.581279935906129e-06, "loss": 0.0693, "step": 96140 }, { "epoch": 46.15, "learning_rate": 3.5759456466262997e-06, "loss": 0.068, "step": 96150 }, { "epoch": 46.16, "learning_rate": 3.5706151856658764e-06, "loss": 0.0715, "step": 96160 }, { "epoch": 46.16, "learning_rate": 3.565288553464424e-06, "loss": 0.0671, "step": 96170 }, { "epoch": 46.17, "learning_rate": 3.559965750461214e-06, "loss": 0.0693, "step": 96180 }, { "epoch": 46.17, "learning_rate": 3.5546467770951607e-06, "loss": 0.0856, "step": 96190 }, { "epoch": 46.17, "learning_rate": 3.5493316338049086e-06, "loss": 0.0698, "step": 96200 }, { "epoch": 46.18, "learning_rate": 3.544020321028757e-06, "loss": 0.0742, "step": 96210 }, { "epoch": 46.18, "learning_rate": 3.538712839204705e-06, "loss": 0.0678, "step": 96220 }, { "epoch": 46.19, "learning_rate": 3.5334091887704203e-06, "loss": 0.067, "step": 96230 }, { "epoch": 46.19, "learning_rate": 3.528109370163274e-06, "loss": 0.0692, "step": 96240 }, { "epoch": 46.2, "learning_rate": 3.522813383820317e-06, "loss": 0.0771, "step": 96250 }, { "epoch": 46.2, "learning_rate": 3.5175212301782555e-06, "loss": 0.0658, "step": 96260 }, { "epoch": 46.21, "learning_rate": 3.5122329096735185e-06, "loss": 0.0765, "step": 96270 }, { "epoch": 46.21, "learning_rate": 3.506948422742201e-06, "loss": 0.0641, "step": 96280 }, { "epoch": 46.22, "learning_rate": 3.501667769820077e-06, "loss": 0.0768, "step": 96290 }, { "epoch": 46.22, "learning_rate": 3.4963909513426206e-06, "loss": 0.0685, "step": 96300 }, { "epoch": 46.22, "learning_rate": 3.4911179677449722e-06, "loss": 0.0671, "step": 96310 }, { "epoch": 46.23, "learning_rate": 3.4858488194619734e-06, "loss": 0.0642, "step": 96320 }, { "epoch": 46.23, "learning_rate": 3.480583506928131e-06, "loss": 0.062, "step": 96330 }, { "epoch": 46.24, "learning_rate": 3.4753220305776435e-06, "loss": 0.0634, "step": 96340 }, { "epoch": 46.24, "learning_rate": 3.470064390844402e-06, "loss": 0.0752, "step": 96350 }, { "epoch": 46.25, "learning_rate": 3.4648105881619706e-06, "loss": 0.0733, "step": 96360 }, { "epoch": 46.25, "learning_rate": 3.4595606229635868e-06, "loss": 0.0691, "step": 96370 }, { "epoch": 46.26, "learning_rate": 3.454314495682198e-06, "loss": 0.0774, "step": 96380 }, { "epoch": 46.26, "learning_rate": 3.4490722067504313e-06, "loss": 0.0666, "step": 96390 }, { "epoch": 46.27, "learning_rate": 3.4438337566005675e-06, "loss": 0.0715, "step": 96400 }, { "epoch": 46.27, "learning_rate": 3.438599145664595e-06, "loss": 0.0756, "step": 96410 }, { "epoch": 46.28, "learning_rate": 3.433368374374185e-06, "loss": 0.0716, "step": 96420 }, { "epoch": 46.28, "learning_rate": 3.4281414431606986e-06, "loss": 0.0739, "step": 96430 }, { "epoch": 46.28, "learning_rate": 3.422918352455151e-06, "loss": 0.0668, "step": 96440 }, { "epoch": 46.29, "learning_rate": 3.4176991026882653e-06, "loss": 0.0678, "step": 96450 }, { "epoch": 46.29, "learning_rate": 3.412483694290447e-06, "loss": 0.0715, "step": 96460 }, { "epoch": 46.3, "learning_rate": 3.4072721276917906e-06, "loss": 0.0823, "step": 96470 }, { "epoch": 46.3, "learning_rate": 3.4020644033220417e-06, "loss": 0.066, "step": 96480 }, { "epoch": 46.31, "learning_rate": 3.3968605216106674e-06, "loss": 0.0737, "step": 96490 }, { "epoch": 46.31, "learning_rate": 3.391660482986797e-06, "loss": 0.0636, "step": 96500 }, { "epoch": 46.32, "learning_rate": 3.3864642878792374e-06, "loss": 0.0788, "step": 96510 }, { "epoch": 46.32, "learning_rate": 3.381271936716496e-06, "loss": 0.0749, "step": 96520 }, { "epoch": 46.33, "learning_rate": 3.376083429926752e-06, "loss": 0.0671, "step": 96530 }, { "epoch": 46.33, "learning_rate": 3.3708987679378858e-06, "loss": 0.0676, "step": 96540 }, { "epoch": 46.33, "learning_rate": 3.365717951177433e-06, "loss": 0.0748, "step": 96550 }, { "epoch": 46.34, "learning_rate": 3.3605409800726185e-06, "loss": 0.0644, "step": 96560 }, { "epoch": 46.34, "learning_rate": 3.3553678550503786e-06, "loss": 0.0769, "step": 96570 }, { "epoch": 46.35, "learning_rate": 3.350198576537289e-06, "loss": 0.0705, "step": 96580 }, { "epoch": 46.35, "learning_rate": 3.345033144959642e-06, "loss": 0.0696, "step": 96590 }, { "epoch": 46.36, "learning_rate": 3.3398715607433796e-06, "loss": 0.0719, "step": 96600 }, { "epoch": 46.36, "learning_rate": 3.3347138243141897e-06, "loss": 0.0728, "step": 96610 }, { "epoch": 46.37, "learning_rate": 3.3295599360973594e-06, "loss": 0.0737, "step": 96620 }, { "epoch": 46.37, "learning_rate": 3.3244098965179215e-06, "loss": 0.0678, "step": 96630 }, { "epoch": 46.38, "learning_rate": 3.319263706000558e-06, "loss": 0.0793, "step": 96640 }, { "epoch": 46.38, "learning_rate": 3.3141213649696636e-06, "loss": 0.0728, "step": 96650 }, { "epoch": 46.38, "learning_rate": 3.30898287384927e-06, "loss": 0.0696, "step": 96660 }, { "epoch": 46.39, "learning_rate": 3.303848233063139e-06, "loss": 0.0712, "step": 96670 }, { "epoch": 46.39, "learning_rate": 3.2987174430346823e-06, "loss": 0.0596, "step": 96680 }, { "epoch": 46.4, "learning_rate": 3.2935905041870164e-06, "loss": 0.0772, "step": 96690 }, { "epoch": 46.4, "learning_rate": 3.2884674169429195e-06, "loss": 0.0677, "step": 96700 }, { "epoch": 46.41, "learning_rate": 3.283348181724871e-06, "loss": 0.0763, "step": 96710 }, { "epoch": 46.41, "learning_rate": 3.2782327989550275e-06, "loss": 0.0772, "step": 96720 }, { "epoch": 46.42, "learning_rate": 3.2731212690552126e-06, "loss": 0.0759, "step": 96730 }, { "epoch": 46.42, "learning_rate": 3.26801359244695e-06, "loss": 0.0655, "step": 96740 }, { "epoch": 46.43, "learning_rate": 3.2629097695514365e-06, "loss": 0.0648, "step": 96750 }, { "epoch": 46.43, "learning_rate": 3.2578098007895575e-06, "loss": 0.0841, "step": 96760 }, { "epoch": 46.44, "learning_rate": 3.252713686581882e-06, "loss": 0.0771, "step": 96770 }, { "epoch": 46.44, "learning_rate": 3.247621427348646e-06, "loss": 0.0622, "step": 96780 }, { "epoch": 46.44, "learning_rate": 3.2425330235097973e-06, "loss": 0.0661, "step": 96790 }, { "epoch": 46.45, "learning_rate": 3.237448475484922e-06, "loss": 0.0726, "step": 96800 }, { "epoch": 46.45, "learning_rate": 3.23236778369333e-06, "loss": 0.0694, "step": 96810 }, { "epoch": 46.46, "learning_rate": 3.2272909485539847e-06, "loss": 0.0728, "step": 96820 }, { "epoch": 46.46, "learning_rate": 3.2222179704855527e-06, "loss": 0.0816, "step": 96830 }, { "epoch": 46.47, "learning_rate": 3.2171488499063707e-06, "loss": 0.0632, "step": 96840 }, { "epoch": 46.47, "learning_rate": 3.2120835872344545e-06, "loss": 0.0718, "step": 96850 }, { "epoch": 46.48, "learning_rate": 3.20702218288752e-06, "loss": 0.0686, "step": 96860 }, { "epoch": 46.48, "learning_rate": 3.2019646372829335e-06, "loss": 0.0771, "step": 96870 }, { "epoch": 46.49, "learning_rate": 3.196910950837767e-06, "loss": 0.0713, "step": 96880 }, { "epoch": 46.49, "learning_rate": 3.1918611239687757e-06, "loss": 0.0716, "step": 96890 }, { "epoch": 46.49, "learning_rate": 3.1868151570923765e-06, "loss": 0.0744, "step": 96900 }, { "epoch": 46.5, "learning_rate": 3.1817730506246923e-06, "loss": 0.0641, "step": 96910 }, { "epoch": 46.5, "learning_rate": 3.1767348049815126e-06, "loss": 0.0776, "step": 96920 }, { "epoch": 46.51, "learning_rate": 3.1717004205783163e-06, "loss": 0.0661, "step": 96930 }, { "epoch": 46.51, "learning_rate": 3.166669897830249e-06, "loss": 0.0669, "step": 96940 }, { "epoch": 46.52, "learning_rate": 3.161643237152151e-06, "loss": 0.075, "step": 96950 }, { "epoch": 46.52, "learning_rate": 3.1566204389585463e-06, "loss": 0.0723, "step": 96960 }, { "epoch": 46.53, "learning_rate": 3.151601503663637e-06, "loss": 0.0776, "step": 96970 }, { "epoch": 46.53, "learning_rate": 3.1465864316812975e-06, "loss": 0.07, "step": 96980 }, { "epoch": 46.54, "learning_rate": 3.141575223425097e-06, "loss": 0.067, "step": 96990 }, { "epoch": 46.54, "learning_rate": 3.136567879308283e-06, "loss": 0.0714, "step": 97000 }, { "epoch": 46.54, "learning_rate": 3.131564399743775e-06, "loss": 0.0806, "step": 97010 }, { "epoch": 46.55, "learning_rate": 3.1265647851441814e-06, "loss": 0.0676, "step": 97020 }, { "epoch": 46.55, "learning_rate": 3.1215690359217954e-06, "loss": 0.0782, "step": 97030 }, { "epoch": 46.56, "learning_rate": 3.1165771524885924e-06, "loss": 0.0681, "step": 97040 }, { "epoch": 46.56, "learning_rate": 3.1115891352561987e-06, "loss": 0.0726, "step": 97050 }, { "epoch": 46.57, "learning_rate": 3.1066049846359747e-06, "loss": 0.0678, "step": 97060 }, { "epoch": 46.57, "learning_rate": 3.1016247010389243e-06, "loss": 0.07, "step": 97070 }, { "epoch": 46.58, "learning_rate": 3.0966482848757473e-06, "loss": 0.0692, "step": 97080 }, { "epoch": 46.58, "learning_rate": 3.0916757365568094e-06, "loss": 0.0673, "step": 97090 }, { "epoch": 46.59, "learning_rate": 3.086707056492166e-06, "loss": 0.0601, "step": 97100 }, { "epoch": 46.59, "learning_rate": 3.0817422450915735e-06, "loss": 0.0605, "step": 97110 }, { "epoch": 46.6, "learning_rate": 3.076781302764431e-06, "loss": 0.0638, "step": 97120 }, { "epoch": 46.6, "learning_rate": 3.0718242299198342e-06, "loss": 0.0809, "step": 97130 }, { "epoch": 46.6, "learning_rate": 3.066871026966589e-06, "loss": 0.0666, "step": 97140 }, { "epoch": 46.61, "learning_rate": 3.061921694313147e-06, "loss": 0.0708, "step": 97150 }, { "epoch": 46.61, "learning_rate": 3.0569762323676422e-06, "loss": 0.0821, "step": 97160 }, { "epoch": 46.62, "learning_rate": 3.0520346415378985e-06, "loss": 0.0691, "step": 97170 }, { "epoch": 46.62, "learning_rate": 3.047096922231435e-06, "loss": 0.074, "step": 97180 }, { "epoch": 46.63, "learning_rate": 3.0421630748554143e-06, "loss": 0.071, "step": 97190 }, { "epoch": 46.63, "learning_rate": 3.037233099816705e-06, "loss": 0.0737, "step": 97200 }, { "epoch": 46.64, "learning_rate": 3.0323069975218723e-06, "loss": 0.0661, "step": 97210 }, { "epoch": 46.64, "learning_rate": 3.0273847683771395e-06, "loss": 0.0752, "step": 97220 }, { "epoch": 46.65, "learning_rate": 3.0224664127883996e-06, "loss": 0.0744, "step": 97230 }, { "epoch": 46.65, "learning_rate": 3.0175519311612445e-06, "loss": 0.0704, "step": 97240 }, { "epoch": 46.65, "learning_rate": 3.0126413239009443e-06, "loss": 0.0708, "step": 97250 }, { "epoch": 46.66, "learning_rate": 3.0077345914124587e-06, "loss": 0.0713, "step": 97260 }, { "epoch": 46.66, "learning_rate": 3.002831734100403e-06, "loss": 0.0734, "step": 97270 }, { "epoch": 46.67, "learning_rate": 2.997932752369076e-06, "loss": 0.0758, "step": 97280 }, { "epoch": 46.67, "learning_rate": 2.993037646622504e-06, "loss": 0.0829, "step": 97290 }, { "epoch": 46.68, "learning_rate": 2.9881464172643314e-06, "loss": 0.0767, "step": 97300 }, { "epoch": 46.68, "learning_rate": 2.9832590646979133e-06, "loss": 0.0609, "step": 97310 }, { "epoch": 46.69, "learning_rate": 2.978375589326282e-06, "loss": 0.0724, "step": 97320 }, { "epoch": 46.69, "learning_rate": 2.9734959915521553e-06, "loss": 0.0641, "step": 97330 }, { "epoch": 46.7, "learning_rate": 2.9686202717779165e-06, "loss": 0.0708, "step": 97340 }, { "epoch": 46.7, "learning_rate": 2.9637484304056384e-06, "loss": 0.0739, "step": 97350 }, { "epoch": 46.7, "learning_rate": 2.958880467837072e-06, "loss": 0.0779, "step": 97360 }, { "epoch": 46.71, "learning_rate": 2.9540163844736567e-06, "loss": 0.0665, "step": 97370 }, { "epoch": 46.71, "learning_rate": 2.9491561807165004e-06, "loss": 0.0643, "step": 97380 }, { "epoch": 46.72, "learning_rate": 2.9442998569663983e-06, "loss": 0.0749, "step": 97390 }, { "epoch": 46.72, "learning_rate": 2.9394474136238246e-06, "loss": 0.0754, "step": 97400 }, { "epoch": 46.73, "learning_rate": 2.9345988510889266e-06, "loss": 0.0688, "step": 97410 }, { "epoch": 46.73, "learning_rate": 2.929754169761534e-06, "loss": 0.0693, "step": 97420 }, { "epoch": 46.74, "learning_rate": 2.9249133700411655e-06, "loss": 0.0756, "step": 97430 }, { "epoch": 46.74, "learning_rate": 2.9200764523270084e-06, "loss": 0.0666, "step": 97440 }, { "epoch": 46.75, "learning_rate": 2.9152434170179434e-06, "loss": 0.0766, "step": 97450 }, { "epoch": 46.75, "learning_rate": 2.910414264512512e-06, "loss": 0.0799, "step": 97460 }, { "epoch": 46.76, "learning_rate": 2.905588995208963e-06, "loss": 0.0763, "step": 97470 }, { "epoch": 46.76, "learning_rate": 2.900767609505184e-06, "loss": 0.0711, "step": 97480 }, { "epoch": 46.76, "learning_rate": 2.8959501077987837e-06, "loss": 0.0717, "step": 97490 }, { "epoch": 46.77, "learning_rate": 2.8911364904870284e-06, "loss": 0.0795, "step": 97500 }, { "epoch": 46.77, "learning_rate": 2.8863267579668675e-06, "loss": 0.0713, "step": 97510 }, { "epoch": 46.78, "learning_rate": 2.8815209106349326e-06, "loss": 0.072, "step": 97520 }, { "epoch": 46.78, "learning_rate": 2.87671894888753e-06, "loss": 0.0724, "step": 97530 }, { "epoch": 46.79, "learning_rate": 2.8719208731206647e-06, "loss": 0.0711, "step": 97540 }, { "epoch": 46.79, "learning_rate": 2.867126683729987e-06, "loss": 0.0759, "step": 97550 }, { "epoch": 46.8, "learning_rate": 2.8623363811108527e-06, "loss": 0.0669, "step": 97560 }, { "epoch": 46.8, "learning_rate": 2.8575499656582903e-06, "loss": 0.0752, "step": 97570 }, { "epoch": 46.81, "learning_rate": 2.852767437767012e-06, "loss": 0.0716, "step": 97580 }, { "epoch": 46.81, "learning_rate": 2.8479887978313912e-06, "loss": 0.0775, "step": 97590 }, { "epoch": 46.81, "learning_rate": 2.843214046245507e-06, "loss": 0.0708, "step": 97600 }, { "epoch": 46.82, "learning_rate": 2.8384431834031113e-06, "loss": 0.0746, "step": 97610 }, { "epoch": 46.82, "learning_rate": 2.833676209697611e-06, "loss": 0.0785, "step": 97620 }, { "epoch": 46.83, "learning_rate": 2.8289131255221147e-06, "loss": 0.0727, "step": 97630 }, { "epoch": 46.83, "learning_rate": 2.824153931269413e-06, "loss": 0.0713, "step": 97640 }, { "epoch": 46.84, "learning_rate": 2.819398627331965e-06, "loss": 0.0729, "step": 97650 }, { "epoch": 46.84, "learning_rate": 2.8146472141019176e-06, "loss": 0.0635, "step": 97660 }, { "epoch": 46.85, "learning_rate": 2.8098996919710795e-06, "loss": 0.0688, "step": 97670 }, { "epoch": 46.85, "learning_rate": 2.805156061330966e-06, "loss": 0.0701, "step": 97680 }, { "epoch": 46.86, "learning_rate": 2.8004163225727574e-06, "loss": 0.0761, "step": 97690 }, { "epoch": 46.86, "learning_rate": 2.7956804760872923e-06, "loss": 0.0718, "step": 97700 }, { "epoch": 46.86, "learning_rate": 2.7909485222651246e-06, "loss": 0.0636, "step": 97710 }, { "epoch": 46.87, "learning_rate": 2.7862204614964694e-06, "loss": 0.075, "step": 97720 }, { "epoch": 46.87, "learning_rate": 2.7814962941712098e-06, "loss": 0.0649, "step": 97730 }, { "epoch": 46.88, "learning_rate": 2.776776020678934e-06, "loss": 0.0755, "step": 97740 }, { "epoch": 46.88, "learning_rate": 2.7720596414088917e-06, "loss": 0.0782, "step": 97750 }, { "epoch": 46.89, "learning_rate": 2.767347156750022e-06, "loss": 0.0691, "step": 97760 }, { "epoch": 46.89, "learning_rate": 2.7626385670909195e-06, "loss": 0.0749, "step": 97770 }, { "epoch": 46.9, "learning_rate": 2.7579338728198846e-06, "loss": 0.0693, "step": 97780 }, { "epoch": 46.9, "learning_rate": 2.75323307432489e-06, "loss": 0.0731, "step": 97790 }, { "epoch": 46.91, "learning_rate": 2.7485361719935655e-06, "loss": 0.076, "step": 97800 }, { "epoch": 46.91, "learning_rate": 2.7438431662132503e-06, "loss": 0.0715, "step": 97810 }, { "epoch": 46.91, "learning_rate": 2.739154057370952e-06, "loss": 0.0735, "step": 97820 }, { "epoch": 46.92, "learning_rate": 2.7344688458533495e-06, "loss": 0.074, "step": 97830 }, { "epoch": 46.92, "learning_rate": 2.7297875320468013e-06, "loss": 0.0703, "step": 97840 }, { "epoch": 46.93, "learning_rate": 2.725110116337354e-06, "loss": 0.0709, "step": 97850 }, { "epoch": 46.93, "learning_rate": 2.7204365991107205e-06, "loss": 0.0722, "step": 97860 }, { "epoch": 46.94, "learning_rate": 2.71576698075231e-06, "loss": 0.0704, "step": 97870 }, { "epoch": 46.94, "learning_rate": 2.71110126164717e-06, "loss": 0.0644, "step": 97880 }, { "epoch": 46.95, "learning_rate": 2.7064394421800866e-06, "loss": 0.0783, "step": 97890 }, { "epoch": 46.95, "learning_rate": 2.7017815227354857e-06, "loss": 0.0888, "step": 97900 }, { "epoch": 47.0, "learning_rate": 2.697127503697472e-06, "loss": 0.0789, "step": 97910 }, { "epoch": 47.01, "learning_rate": 2.692477385449832e-06, "loss": 0.0729, "step": 97920 }, { "epoch": 47.01, "learning_rate": 2.6878311683760427e-06, "loss": 0.0732, "step": 97930 }, { "epoch": 47.02, "learning_rate": 2.6831888528592474e-06, "loss": 0.0659, "step": 97940 }, { "epoch": 47.02, "learning_rate": 2.678550439282268e-06, "loss": 0.0659, "step": 97950 }, { "epoch": 47.03, "learning_rate": 2.673915928027598e-06, "loss": 0.0691, "step": 97960 }, { "epoch": 47.03, "learning_rate": 2.6692853194774435e-06, "loss": 0.0707, "step": 97970 }, { "epoch": 47.04, "learning_rate": 2.6646586140136433e-06, "loss": 0.0673, "step": 97980 }, { "epoch": 47.04, "learning_rate": 2.6600358120177414e-06, "loss": 0.0755, "step": 97990 }, { "epoch": 47.05, "learning_rate": 2.655416913870956e-06, "loss": 0.076, "step": 98000 }, { "epoch": 47.05, "learning_rate": 2.6508019199541767e-06, "loss": 0.0652, "step": 98010 }, { "epoch": 47.05, "learning_rate": 2.646190830647971e-06, "loss": 0.0747, "step": 98020 }, { "epoch": 47.06, "learning_rate": 2.641583646332596e-06, "loss": 0.0698, "step": 98030 }, { "epoch": 47.06, "learning_rate": 2.6369803673879643e-06, "loss": 0.0675, "step": 98040 }, { "epoch": 47.07, "learning_rate": 2.632380994193706e-06, "loss": 0.0725, "step": 98050 }, { "epoch": 47.07, "learning_rate": 2.6277855271290897e-06, "loss": 0.0719, "step": 98060 }, { "epoch": 47.08, "learning_rate": 2.6231939665730733e-06, "loss": 0.0732, "step": 98070 }, { "epoch": 47.08, "learning_rate": 2.618606312904309e-06, "loss": 0.071, "step": 98080 }, { "epoch": 47.09, "learning_rate": 2.614022566501095e-06, "loss": 0.072, "step": 98090 }, { "epoch": 47.09, "learning_rate": 2.60944272774144e-06, "loss": 0.0782, "step": 98100 }, { "epoch": 47.1, "learning_rate": 2.6048667970030137e-06, "loss": 0.0661, "step": 98110 }, { "epoch": 47.1, "learning_rate": 2.6002947746631643e-06, "loss": 0.0672, "step": 98120 }, { "epoch": 47.1, "learning_rate": 2.5957266610989182e-06, "loss": 0.0696, "step": 98130 }, { "epoch": 47.11, "learning_rate": 2.5911624566869853e-06, "loss": 0.0684, "step": 98140 }, { "epoch": 47.11, "learning_rate": 2.586602161803753e-06, "loss": 0.0735, "step": 98150 }, { "epoch": 47.12, "learning_rate": 2.58204577682527e-06, "loss": 0.072, "step": 98160 }, { "epoch": 47.12, "learning_rate": 2.577493302127282e-06, "loss": 0.063, "step": 98170 }, { "epoch": 47.13, "learning_rate": 2.572944738085198e-06, "loss": 0.084, "step": 98180 }, { "epoch": 47.13, "learning_rate": 2.5684000850741243e-06, "loss": 0.0769, "step": 98190 }, { "epoch": 47.14, "learning_rate": 2.563859343468822e-06, "loss": 0.0687, "step": 98200 }, { "epoch": 47.14, "learning_rate": 2.559322513643736e-06, "loss": 0.0676, "step": 98210 }, { "epoch": 47.15, "learning_rate": 2.554789595973012e-06, "loss": 0.0693, "step": 98220 }, { "epoch": 47.15, "learning_rate": 2.5502605908304277e-06, "loss": 0.0758, "step": 98230 }, { "epoch": 47.15, "learning_rate": 2.5457354985894742e-06, "loss": 0.0647, "step": 98240 }, { "epoch": 47.16, "learning_rate": 2.541214319623314e-06, "loss": 0.0731, "step": 98250 }, { "epoch": 47.16, "learning_rate": 2.536697054304771e-06, "loss": 0.0688, "step": 98260 }, { "epoch": 47.17, "learning_rate": 2.5321837030063688e-06, "loss": 0.076, "step": 98270 }, { "epoch": 47.17, "learning_rate": 2.5276742661002938e-06, "loss": 0.0756, "step": 98280 }, { "epoch": 47.18, "learning_rate": 2.523168743958415e-06, "loss": 0.0725, "step": 98290 }, { "epoch": 47.18, "learning_rate": 2.518667136952274e-06, "loss": 0.0677, "step": 98300 }, { "epoch": 47.19, "learning_rate": 2.514169445453085e-06, "loss": 0.0813, "step": 98310 }, { "epoch": 47.19, "learning_rate": 2.509675669831757e-06, "loss": 0.0756, "step": 98320 }, { "epoch": 47.2, "learning_rate": 2.5051858104588544e-06, "loss": 0.0693, "step": 98330 }, { "epoch": 47.2, "learning_rate": 2.500699867704642e-06, "loss": 0.0676, "step": 98340 }, { "epoch": 47.21, "learning_rate": 2.4962178419390357e-06, "loss": 0.0744, "step": 98350 }, { "epoch": 47.21, "learning_rate": 2.491739733531656e-06, "loss": 0.0708, "step": 98360 }, { "epoch": 47.21, "learning_rate": 2.4872655428517857e-06, "loss": 0.0688, "step": 98370 }, { "epoch": 47.22, "learning_rate": 2.4827952702683677e-06, "loss": 0.0645, "step": 98380 }, { "epoch": 47.22, "learning_rate": 2.4783289161500522e-06, "loss": 0.0752, "step": 98390 }, { "epoch": 47.23, "learning_rate": 2.4738664808651502e-06, "loss": 0.0722, "step": 98400 }, { "epoch": 47.23, "learning_rate": 2.469407964781656e-06, "loss": 0.0815, "step": 98410 }, { "epoch": 47.24, "learning_rate": 2.4649533682672367e-06, "loss": 0.0728, "step": 98420 }, { "epoch": 47.24, "learning_rate": 2.4605026916892317e-06, "loss": 0.0695, "step": 98430 }, { "epoch": 47.25, "learning_rate": 2.4560559354146696e-06, "loss": 0.069, "step": 98440 }, { "epoch": 47.25, "learning_rate": 2.4516130998102405e-06, "loss": 0.0724, "step": 98450 }, { "epoch": 47.26, "learning_rate": 2.4471741852423237e-06, "loss": 0.0699, "step": 98460 }, { "epoch": 47.26, "learning_rate": 2.442739192076965e-06, "loss": 0.0768, "step": 98470 }, { "epoch": 47.26, "learning_rate": 2.4383081206798996e-06, "loss": 0.0663, "step": 98480 }, { "epoch": 47.27, "learning_rate": 2.43388097141653e-06, "loss": 0.0694, "step": 98490 }, { "epoch": 47.27, "learning_rate": 2.429457744651936e-06, "loss": 0.0659, "step": 98500 }, { "epoch": 47.28, "learning_rate": 2.4250384407508818e-06, "loss": 0.0681, "step": 98510 }, { "epoch": 47.28, "learning_rate": 2.420623060077787e-06, "loss": 0.0697, "step": 98520 }, { "epoch": 47.29, "learning_rate": 2.416211602996771e-06, "loss": 0.0706, "step": 98530 }, { "epoch": 47.29, "learning_rate": 2.4118040698716204e-06, "loss": 0.0757, "step": 98540 }, { "epoch": 47.3, "learning_rate": 2.407400461065801e-06, "loss": 0.0747, "step": 98550 }, { "epoch": 47.3, "learning_rate": 2.4030007769424436e-06, "loss": 0.0778, "step": 98560 }, { "epoch": 47.31, "learning_rate": 2.3986050178643693e-06, "loss": 0.0655, "step": 98570 }, { "epoch": 47.31, "learning_rate": 2.394213184194083e-06, "loss": 0.0649, "step": 98580 }, { "epoch": 47.31, "learning_rate": 2.3898252762937334e-06, "loss": 0.0683, "step": 98590 }, { "epoch": 47.32, "learning_rate": 2.385441294525176e-06, "loss": 0.0684, "step": 98600 }, { "epoch": 47.32, "learning_rate": 2.381061239249932e-06, "loss": 0.0711, "step": 98610 }, { "epoch": 47.33, "learning_rate": 2.376685110829202e-06, "loss": 0.0726, "step": 98620 }, { "epoch": 47.33, "learning_rate": 2.3723129096238472e-06, "loss": 0.0691, "step": 98630 }, { "epoch": 47.34, "learning_rate": 2.367944635994418e-06, "loss": 0.0752, "step": 98640 }, { "epoch": 47.34, "learning_rate": 2.363580290301154e-06, "loss": 0.0793, "step": 98650 }, { "epoch": 47.35, "learning_rate": 2.3592198729039615e-06, "loss": 0.0755, "step": 98660 }, { "epoch": 47.35, "learning_rate": 2.354863384162398e-06, "loss": 0.0775, "step": 98670 }, { "epoch": 47.36, "learning_rate": 2.3505108244357253e-06, "loss": 0.0671, "step": 98680 }, { "epoch": 47.36, "learning_rate": 2.3461621940828847e-06, "loss": 0.069, "step": 98690 }, { "epoch": 47.37, "learning_rate": 2.3418174934624614e-06, "loss": 0.0715, "step": 98700 }, { "epoch": 47.37, "learning_rate": 2.337476722932752e-06, "loss": 0.0749, "step": 98710 }, { "epoch": 47.37, "learning_rate": 2.3331398828517037e-06, "loss": 0.0695, "step": 98720 }, { "epoch": 47.38, "learning_rate": 2.3288069735769745e-06, "loss": 0.0697, "step": 98730 }, { "epoch": 47.38, "learning_rate": 2.324477995465846e-06, "loss": 0.0729, "step": 98740 }, { "epoch": 47.39, "learning_rate": 2.3201529488753203e-06, "loss": 0.0725, "step": 98750 }, { "epoch": 47.39, "learning_rate": 2.3158318341620577e-06, "loss": 0.0604, "step": 98760 }, { "epoch": 47.4, "learning_rate": 2.3115146516823783e-06, "loss": 0.0714, "step": 98770 }, { "epoch": 47.4, "learning_rate": 2.3072014017923138e-06, "loss": 0.068, "step": 98780 }, { "epoch": 47.41, "learning_rate": 2.3028920848475468e-06, "loss": 0.0695, "step": 98790 }, { "epoch": 47.41, "learning_rate": 2.298586701203437e-06, "loss": 0.0638, "step": 98800 }, { "epoch": 47.42, "learning_rate": 2.294285251215028e-06, "loss": 0.0743, "step": 98810 }, { "epoch": 47.42, "learning_rate": 2.2899877352370313e-06, "loss": 0.07, "step": 98820 }, { "epoch": 47.42, "learning_rate": 2.285694153623852e-06, "loss": 0.0675, "step": 98830 }, { "epoch": 47.43, "learning_rate": 2.2814045067295397e-06, "loss": 0.0678, "step": 98840 }, { "epoch": 47.43, "learning_rate": 2.2771187949078452e-06, "loss": 0.0738, "step": 98850 }, { "epoch": 47.44, "learning_rate": 2.2728370185121804e-06, "loss": 0.0707, "step": 98860 }, { "epoch": 47.44, "learning_rate": 2.2685591778956406e-06, "loss": 0.0696, "step": 98870 }, { "epoch": 47.45, "learning_rate": 2.264285273410999e-06, "loss": 0.0727, "step": 98880 }, { "epoch": 47.45, "learning_rate": 2.2600153054106897e-06, "loss": 0.067, "step": 98890 }, { "epoch": 47.46, "learning_rate": 2.2557492742468378e-06, "loss": 0.0721, "step": 98900 }, { "epoch": 47.46, "learning_rate": 2.25148718027125e-06, "loss": 0.063, "step": 98910 }, { "epoch": 47.47, "learning_rate": 2.2472290238353733e-06, "loss": 0.0737, "step": 98920 }, { "epoch": 47.47, "learning_rate": 2.2429748052903654e-06, "loss": 0.067, "step": 98930 }, { "epoch": 47.47, "learning_rate": 2.23872452498704e-06, "loss": 0.0641, "step": 98940 }, { "epoch": 47.48, "learning_rate": 2.2344781832759008e-06, "loss": 0.068, "step": 98950 }, { "epoch": 47.48, "learning_rate": 2.2302357805071174e-06, "loss": 0.0788, "step": 98960 }, { "epoch": 47.49, "learning_rate": 2.2259973170305316e-06, "loss": 0.0688, "step": 98970 }, { "epoch": 47.49, "learning_rate": 2.2217627931956753e-06, "loss": 0.0821, "step": 98980 }, { "epoch": 47.5, "learning_rate": 2.2175322093517302e-06, "loss": 0.0685, "step": 98990 }, { "epoch": 47.5, "learning_rate": 2.213305565847573e-06, "loss": 0.0743, "step": 99000 }, { "epoch": 47.51, "learning_rate": 2.209082863031747e-06, "loss": 0.0635, "step": 99010 }, { "epoch": 47.51, "learning_rate": 2.2048641012524852e-06, "loss": 0.0695, "step": 99020 }, { "epoch": 47.52, "learning_rate": 2.2006492808576753e-06, "loss": 0.0673, "step": 99030 }, { "epoch": 47.52, "learning_rate": 2.1964384021948837e-06, "loss": 0.0736, "step": 99040 }, { "epoch": 47.53, "learning_rate": 2.1922314656113774e-06, "loss": 0.0718, "step": 99050 }, { "epoch": 47.53, "learning_rate": 2.1880284714540568e-06, "loss": 0.0731, "step": 99060 }, { "epoch": 47.53, "learning_rate": 2.183829420069522e-06, "loss": 0.0652, "step": 99070 }, { "epoch": 47.54, "learning_rate": 2.1796343118040464e-06, "loss": 0.0616, "step": 99080 }, { "epoch": 47.54, "learning_rate": 2.1754431470035807e-06, "loss": 0.0734, "step": 99090 }, { "epoch": 47.55, "learning_rate": 2.1712559260137434e-06, "loss": 0.0674, "step": 99100 }, { "epoch": 47.55, "learning_rate": 2.1670726491798244e-06, "loss": 0.0717, "step": 99110 }, { "epoch": 47.56, "learning_rate": 2.1628933168468034e-06, "loss": 0.0647, "step": 99120 }, { "epoch": 47.56, "learning_rate": 2.1587179293593164e-06, "loss": 0.0673, "step": 99130 }, { "epoch": 47.57, "learning_rate": 2.1545464870616873e-06, "loss": 0.0644, "step": 99140 }, { "epoch": 47.57, "learning_rate": 2.1503789902979134e-06, "loss": 0.071, "step": 99150 }, { "epoch": 47.58, "learning_rate": 2.146215439411653e-06, "loss": 0.0741, "step": 99160 }, { "epoch": 47.58, "learning_rate": 2.142055834746265e-06, "loss": 0.0658, "step": 99170 }, { "epoch": 47.58, "learning_rate": 2.1379001766447527e-06, "loss": 0.0704, "step": 99180 }, { "epoch": 47.59, "learning_rate": 2.1337484654498253e-06, "loss": 0.068, "step": 99190 }, { "epoch": 47.59, "learning_rate": 2.1296007015038366e-06, "loss": 0.0681, "step": 99200 }, { "epoch": 47.6, "learning_rate": 2.12545688514883e-06, "loss": 0.0681, "step": 99210 }, { "epoch": 47.6, "learning_rate": 2.1213170167265207e-06, "loss": 0.0687, "step": 99220 }, { "epoch": 47.61, "learning_rate": 2.1171810965783133e-06, "loss": 0.0686, "step": 99230 }, { "epoch": 47.61, "learning_rate": 2.113049125045252e-06, "loss": 0.0684, "step": 99240 }, { "epoch": 47.62, "learning_rate": 2.1089211024680864e-06, "loss": 0.0701, "step": 99250 }, { "epoch": 47.62, "learning_rate": 2.1047970291872386e-06, "loss": 0.0754, "step": 99260 }, { "epoch": 47.63, "learning_rate": 2.100676905542787e-06, "loss": 0.0779, "step": 99270 }, { "epoch": 47.63, "learning_rate": 2.0965607318744916e-06, "loss": 0.0746, "step": 99280 }, { "epoch": 47.63, "learning_rate": 2.0924485085217937e-06, "loss": 0.0767, "step": 99290 }, { "epoch": 47.64, "learning_rate": 2.0883402358238157e-06, "loss": 0.0737, "step": 99300 }, { "epoch": 47.64, "learning_rate": 2.0842359141193147e-06, "loss": 0.0755, "step": 99310 }, { "epoch": 47.65, "learning_rate": 2.080135543746764e-06, "loss": 0.071, "step": 99320 }, { "epoch": 47.65, "learning_rate": 2.07603912504431e-06, "loss": 0.0758, "step": 99330 }, { "epoch": 47.66, "learning_rate": 2.0719466583497493e-06, "loss": 0.0739, "step": 99340 }, { "epoch": 47.66, "learning_rate": 2.067858144000562e-06, "loss": 0.0696, "step": 99350 }, { "epoch": 47.67, "learning_rate": 2.0637735823339e-06, "loss": 0.0725, "step": 99360 }, { "epoch": 47.67, "learning_rate": 2.059692973686611e-06, "loss": 0.0722, "step": 99370 }, { "epoch": 47.68, "learning_rate": 2.055616318395176e-06, "loss": 0.0745, "step": 99380 }, { "epoch": 47.68, "learning_rate": 2.051543616795787e-06, "loss": 0.0639, "step": 99390 }, { "epoch": 47.69, "learning_rate": 2.0474748692242864e-06, "loss": 0.0786, "step": 99400 }, { "epoch": 47.69, "learning_rate": 2.0434100760162168e-06, "loss": 0.0726, "step": 99410 }, { "epoch": 47.69, "learning_rate": 2.0393492375067602e-06, "loss": 0.0665, "step": 99420 }, { "epoch": 47.7, "learning_rate": 2.0352923540307933e-06, "loss": 0.0795, "step": 99430 }, { "epoch": 47.7, "learning_rate": 2.031239425922876e-06, "loss": 0.0693, "step": 99440 }, { "epoch": 47.71, "learning_rate": 2.0271904535172193e-06, "loss": 0.0746, "step": 99450 }, { "epoch": 47.71, "learning_rate": 2.023145437147711e-06, "loss": 0.0716, "step": 99460 }, { "epoch": 47.72, "learning_rate": 2.0191043771479347e-06, "loss": 0.0815, "step": 99470 }, { "epoch": 47.72, "learning_rate": 2.0150672738511234e-06, "loss": 0.0768, "step": 99480 }, { "epoch": 47.73, "learning_rate": 2.0110341275902e-06, "loss": 0.0714, "step": 99490 }, { "epoch": 47.73, "learning_rate": 2.0070049386977484e-06, "loss": 0.0711, "step": 99500 }, { "epoch": 47.74, "learning_rate": 2.002979707506031e-06, "loss": 0.0698, "step": 99510 }, { "epoch": 47.74, "learning_rate": 1.998958434346998e-06, "loss": 0.0687, "step": 99520 }, { "epoch": 47.74, "learning_rate": 1.994941119552246e-06, "loss": 0.0724, "step": 99530 }, { "epoch": 47.75, "learning_rate": 1.9909277634530655e-06, "loss": 0.0698, "step": 99540 }, { "epoch": 47.75, "learning_rate": 1.986918366380408e-06, "loss": 0.0711, "step": 99550 }, { "epoch": 47.76, "learning_rate": 1.982912928664915e-06, "loss": 0.071, "step": 99560 }, { "epoch": 47.76, "learning_rate": 1.9789114506368833e-06, "loss": 0.0812, "step": 99570 }, { "epoch": 47.77, "learning_rate": 1.9749139326262988e-06, "loss": 0.074, "step": 99580 }, { "epoch": 47.77, "learning_rate": 1.9709203749628093e-06, "loss": 0.0713, "step": 99590 }, { "epoch": 47.78, "learning_rate": 1.966930777975734e-06, "loss": 0.0715, "step": 99600 }, { "epoch": 47.78, "learning_rate": 1.9629451419940826e-06, "loss": 0.0661, "step": 99610 }, { "epoch": 47.79, "learning_rate": 1.9589634673465194e-06, "loss": 0.0746, "step": 99620 }, { "epoch": 47.79, "learning_rate": 1.9549857543613935e-06, "loss": 0.0734, "step": 99630 }, { "epoch": 47.79, "learning_rate": 1.95101200336672e-06, "loss": 0.0694, "step": 99640 }, { "epoch": 47.8, "learning_rate": 1.9470422146901923e-06, "loss": 0.0678, "step": 99650 }, { "epoch": 47.8, "learning_rate": 1.9430763886591873e-06, "loss": 0.0719, "step": 99660 }, { "epoch": 47.81, "learning_rate": 1.9391145256007215e-06, "loss": 0.0753, "step": 99670 }, { "epoch": 47.81, "learning_rate": 1.935156625841522e-06, "loss": 0.0665, "step": 99680 }, { "epoch": 47.82, "learning_rate": 1.9312026897079672e-06, "loss": 0.0767, "step": 99690 }, { "epoch": 47.82, "learning_rate": 1.927252717526118e-06, "loss": 0.0756, "step": 99700 }, { "epoch": 47.83, "learning_rate": 1.9233067096217027e-06, "loss": 0.0639, "step": 99710 }, { "epoch": 47.83, "learning_rate": 1.9193646663201227e-06, "loss": 0.0719, "step": 99720 }, { "epoch": 47.84, "learning_rate": 1.915426587946473e-06, "loss": 0.0692, "step": 99730 }, { "epoch": 47.84, "learning_rate": 1.9114924748254826e-06, "loss": 0.0717, "step": 99740 }, { "epoch": 47.85, "learning_rate": 1.9075623272815813e-06, "loss": 0.0666, "step": 99750 }, { "epoch": 47.85, "learning_rate": 1.903636145638865e-06, "loss": 0.0725, "step": 99760 }, { "epoch": 47.85, "learning_rate": 1.8997139302211088e-06, "loss": 0.0613, "step": 99770 }, { "epoch": 47.86, "learning_rate": 1.8957956813517474e-06, "loss": 0.0769, "step": 99780 }, { "epoch": 47.86, "learning_rate": 1.8918813993539008e-06, "loss": 0.0708, "step": 99790 }, { "epoch": 47.87, "learning_rate": 1.8879710845503606e-06, "loss": 0.0691, "step": 99800 }, { "epoch": 47.87, "learning_rate": 1.8840647372635744e-06, "loss": 0.0591, "step": 99810 }, { "epoch": 47.88, "learning_rate": 1.8801623578156791e-06, "loss": 0.0828, "step": 99820 }, { "epoch": 47.88, "learning_rate": 1.8762639465284948e-06, "loss": 0.0666, "step": 99830 }, { "epoch": 47.89, "learning_rate": 1.8723695037234812e-06, "loss": 0.0651, "step": 99840 }, { "epoch": 47.89, "learning_rate": 1.8684790297218036e-06, "loss": 0.0739, "step": 99850 }, { "epoch": 47.9, "learning_rate": 1.8645925248442887e-06, "loss": 0.0677, "step": 99860 }, { "epoch": 47.9, "learning_rate": 1.8607099894114189e-06, "loss": 0.0738, "step": 99870 }, { "epoch": 47.9, "learning_rate": 1.8568314237433825e-06, "loss": 0.0658, "step": 99880 }, { "epoch": 47.91, "learning_rate": 1.8529568281600075e-06, "loss": 0.067, "step": 99890 }, { "epoch": 47.91, "learning_rate": 1.8490862029808154e-06, "loss": 0.0817, "step": 99900 }, { "epoch": 47.92, "learning_rate": 1.8452195485249956e-06, "loss": 0.0633, "step": 99910 }, { "epoch": 47.92, "learning_rate": 1.8413568651113933e-06, "loss": 0.0704, "step": 99920 }, { "epoch": 47.93, "learning_rate": 1.8374981530585644e-06, "loss": 0.062, "step": 99930 }, { "epoch": 47.93, "learning_rate": 1.8336434126846991e-06, "loss": 0.0659, "step": 99940 }, { "epoch": 47.94, "learning_rate": 1.8297926443076873e-06, "loss": 0.0653, "step": 99950 }, { "epoch": 47.94, "learning_rate": 1.8259458482450642e-06, "loss": 0.0691, "step": 99960 }, { "epoch": 47.95, "learning_rate": 1.8221030248140648e-06, "loss": 0.0771, "step": 99970 }, { "epoch": 47.95, "learning_rate": 1.8182641743315798e-06, "loss": 0.0664, "step": 99980 }, { "epoch": 48.0, "learning_rate": 1.814429297114173e-06, "loss": 0.0709, "step": 99990 }, { "epoch": 48.01, "learning_rate": 1.81059839347808e-06, "loss": 0.0649, "step": 100000 }, { "epoch": 48.01, "learning_rate": 1.8067714637392263e-06, "loss": 0.0678, "step": 100010 }, { "epoch": 48.02, "learning_rate": 1.8029485082131926e-06, "loss": 0.0746, "step": 100020 }, { "epoch": 48.02, "learning_rate": 1.799129527215232e-06, "loss": 0.0632, "step": 100030 }, { "epoch": 48.03, "learning_rate": 1.7953145210602762e-06, "loss": 0.0684, "step": 100040 }, { "epoch": 48.03, "learning_rate": 1.7915034900629234e-06, "loss": 0.0806, "step": 100050 }, { "epoch": 48.03, "learning_rate": 1.7876964345374503e-06, "loss": 0.0736, "step": 100060 }, { "epoch": 48.04, "learning_rate": 1.7838933547977998e-06, "loss": 0.0728, "step": 100070 }, { "epoch": 48.04, "learning_rate": 1.7800942511575823e-06, "loss": 0.067, "step": 100080 }, { "epoch": 48.05, "learning_rate": 1.7762991239301086e-06, "loss": 0.0605, "step": 100090 }, { "epoch": 48.05, "learning_rate": 1.7725079734283223e-06, "loss": 0.0795, "step": 100100 }, { "epoch": 48.06, "learning_rate": 1.7687207999648625e-06, "loss": 0.0678, "step": 100110 }, { "epoch": 48.06, "learning_rate": 1.7649376038520348e-06, "loss": 0.0734, "step": 100120 }, { "epoch": 48.07, "learning_rate": 1.7611583854018232e-06, "loss": 0.0722, "step": 100130 }, { "epoch": 48.07, "learning_rate": 1.757383144925867e-06, "loss": 0.0677, "step": 100140 }, { "epoch": 48.08, "learning_rate": 1.7536118827354897e-06, "loss": 0.069, "step": 100150 }, { "epoch": 48.08, "learning_rate": 1.7498445991416979e-06, "loss": 0.0732, "step": 100160 }, { "epoch": 48.09, "learning_rate": 1.746081294455143e-06, "loss": 0.0748, "step": 100170 }, { "epoch": 48.09, "learning_rate": 1.7423219689861713e-06, "loss": 0.0809, "step": 100180 }, { "epoch": 48.09, "learning_rate": 1.738566623044785e-06, "loss": 0.0759, "step": 100190 }, { "epoch": 48.1, "learning_rate": 1.734815256940675e-06, "loss": 0.0689, "step": 100200 }, { "epoch": 48.1, "learning_rate": 1.7310678709831884e-06, "loss": 0.0755, "step": 100210 }, { "epoch": 48.11, "learning_rate": 1.7273244654813448e-06, "loss": 0.0705, "step": 100220 }, { "epoch": 48.11, "learning_rate": 1.7235850407438525e-06, "loss": 0.0675, "step": 100230 }, { "epoch": 48.12, "learning_rate": 1.7198495970790706e-06, "loss": 0.0629, "step": 100240 }, { "epoch": 48.12, "learning_rate": 1.7161181347950417e-06, "loss": 0.0816, "step": 100250 }, { "epoch": 48.13, "learning_rate": 1.7123906541994806e-06, "loss": 0.0772, "step": 100260 }, { "epoch": 48.13, "learning_rate": 1.7086671555997747e-06, "loss": 0.0687, "step": 100270 }, { "epoch": 48.14, "learning_rate": 1.7049476393029673e-06, "loss": 0.0575, "step": 100280 }, { "epoch": 48.14, "learning_rate": 1.7012321056157908e-06, "loss": 0.0643, "step": 100290 }, { "epoch": 48.14, "learning_rate": 1.6975205548446448e-06, "loss": 0.0686, "step": 100300 }, { "epoch": 48.15, "learning_rate": 1.6938129872955954e-06, "loss": 0.0751, "step": 100310 }, { "epoch": 48.15, "learning_rate": 1.6901094032743868e-06, "loss": 0.0747, "step": 100320 }, { "epoch": 48.16, "learning_rate": 1.686409803086436e-06, "loss": 0.0707, "step": 100330 }, { "epoch": 48.16, "learning_rate": 1.6827141870368268e-06, "loss": 0.0702, "step": 100340 }, { "epoch": 48.17, "learning_rate": 1.679022555430304e-06, "loss": 0.068, "step": 100350 }, { "epoch": 48.17, "learning_rate": 1.6753349085713022e-06, "loss": 0.0642, "step": 100360 }, { "epoch": 48.18, "learning_rate": 1.6716512467639168e-06, "loss": 0.0757, "step": 100370 }, { "epoch": 48.18, "learning_rate": 1.667971570311927e-06, "loss": 0.0831, "step": 100380 }, { "epoch": 48.19, "learning_rate": 1.6642958795187624e-06, "loss": 0.0776, "step": 100390 }, { "epoch": 48.19, "learning_rate": 1.6606241746875473e-06, "loss": 0.0737, "step": 100400 }, { "epoch": 48.19, "learning_rate": 1.6569564561210615e-06, "loss": 0.0692, "step": 100410 }, { "epoch": 48.2, "learning_rate": 1.653292724121752e-06, "loss": 0.0719, "step": 100420 }, { "epoch": 48.2, "learning_rate": 1.6496329789917498e-06, "loss": 0.0725, "step": 100430 }, { "epoch": 48.21, "learning_rate": 1.6459772210328573e-06, "loss": 0.0746, "step": 100440 }, { "epoch": 48.21, "learning_rate": 1.642325450546539e-06, "loss": 0.069, "step": 100450 }, { "epoch": 48.22, "learning_rate": 1.6386776678339377e-06, "loss": 0.0679, "step": 100460 }, { "epoch": 48.22, "learning_rate": 1.6350338731958626e-06, "loss": 0.0721, "step": 100470 }, { "epoch": 48.23, "learning_rate": 1.6313940669327955e-06, "loss": 0.0869, "step": 100480 }, { "epoch": 48.23, "learning_rate": 1.6277582493449018e-06, "loss": 0.0644, "step": 100490 }, { "epoch": 48.24, "learning_rate": 1.6241264207319806e-06, "loss": 0.066, "step": 100500 }, { "epoch": 48.24, "learning_rate": 1.620498581393548e-06, "loss": 0.0754, "step": 100510 }, { "epoch": 48.24, "learning_rate": 1.6168747316287647e-06, "loss": 0.0685, "step": 100520 }, { "epoch": 48.25, "learning_rate": 1.6132548717364693e-06, "loss": 0.0651, "step": 100530 }, { "epoch": 48.25, "learning_rate": 1.609639002015173e-06, "loss": 0.077, "step": 100540 }, { "epoch": 48.26, "learning_rate": 1.6060271227630487e-06, "loss": 0.0692, "step": 100550 }, { "epoch": 48.26, "learning_rate": 1.602419234277952e-06, "loss": 0.0696, "step": 100560 }, { "epoch": 48.27, "learning_rate": 1.598815336857401e-06, "loss": 0.0703, "step": 100570 }, { "epoch": 48.27, "learning_rate": 1.5952154307985857e-06, "loss": 0.0728, "step": 100580 }, { "epoch": 48.28, "learning_rate": 1.591619516398385e-06, "loss": 0.0766, "step": 100590 }, { "epoch": 48.28, "learning_rate": 1.5880275939533063e-06, "loss": 0.0682, "step": 100600 }, { "epoch": 48.29, "learning_rate": 1.5844396637595738e-06, "loss": 0.0728, "step": 100610 }, { "epoch": 48.29, "learning_rate": 1.5808557261130563e-06, "loss": 0.0695, "step": 100620 }, { "epoch": 48.3, "learning_rate": 1.577275781309312e-06, "loss": 0.0747, "step": 100630 }, { "epoch": 48.3, "learning_rate": 1.5736998296435434e-06, "loss": 0.0649, "step": 100640 }, { "epoch": 48.3, "learning_rate": 1.5701278714106371e-06, "loss": 0.0737, "step": 100650 }, { "epoch": 48.31, "learning_rate": 1.566559906905163e-06, "loss": 0.0809, "step": 100660 }, { "epoch": 48.31, "learning_rate": 1.5629959364213464e-06, "loss": 0.0716, "step": 100670 }, { "epoch": 48.32, "learning_rate": 1.559435960253075e-06, "loss": 0.0717, "step": 100680 }, { "epoch": 48.32, "learning_rate": 1.555879978693936e-06, "loss": 0.0713, "step": 100690 }, { "epoch": 48.33, "learning_rate": 1.552327992037167e-06, "loss": 0.0664, "step": 100700 }, { "epoch": 48.33, "learning_rate": 1.5487800005756726e-06, "loss": 0.074, "step": 100710 }, { "epoch": 48.34, "learning_rate": 1.5452360046020352e-06, "loss": 0.0704, "step": 100720 }, { "epoch": 48.34, "learning_rate": 1.5416960044085105e-06, "loss": 0.0764, "step": 100730 }, { "epoch": 48.35, "learning_rate": 1.5381600002870312e-06, "loss": 0.0665, "step": 100740 }, { "epoch": 48.35, "learning_rate": 1.5346279925291696e-06, "loss": 0.0697, "step": 100750 }, { "epoch": 48.35, "learning_rate": 1.531099981426204e-06, "loss": 0.0658, "step": 100760 }, { "epoch": 48.36, "learning_rate": 1.5275759672690737e-06, "loss": 0.0735, "step": 100770 }, { "epoch": 48.36, "learning_rate": 1.5240559503483686e-06, "loss": 0.074, "step": 100780 }, { "epoch": 48.37, "learning_rate": 1.5205399309543678e-06, "loss": 0.0805, "step": 100790 }, { "epoch": 48.37, "learning_rate": 1.5170279093770279e-06, "loss": 0.0738, "step": 100800 }, { "epoch": 48.38, "learning_rate": 1.5135198859059562e-06, "loss": 0.0647, "step": 100810 }, { "epoch": 48.38, "learning_rate": 1.5100158608304382e-06, "loss": 0.0714, "step": 100820 }, { "epoch": 48.39, "learning_rate": 1.5065158344394203e-06, "loss": 0.0773, "step": 100830 }, { "epoch": 48.39, "learning_rate": 1.503019807021555e-06, "loss": 0.0693, "step": 100840 }, { "epoch": 48.4, "learning_rate": 1.499527778865123e-06, "loss": 0.0618, "step": 100850 }, { "epoch": 48.4, "learning_rate": 1.4960397502580881e-06, "loss": 0.0751, "step": 100860 }, { "epoch": 48.4, "learning_rate": 1.4925557214880925e-06, "loss": 0.0698, "step": 100870 }, { "epoch": 48.41, "learning_rate": 1.489075692842451e-06, "loss": 0.0718, "step": 100880 }, { "epoch": 48.41, "learning_rate": 1.4855996646081227e-06, "loss": 0.0701, "step": 100890 }, { "epoch": 48.42, "learning_rate": 1.4821276370717729e-06, "loss": 0.0736, "step": 100900 }, { "epoch": 48.42, "learning_rate": 1.478659610519706e-06, "loss": 0.0691, "step": 100910 }, { "epoch": 48.43, "learning_rate": 1.4751955852379206e-06, "loss": 0.0635, "step": 100920 }, { "epoch": 48.43, "learning_rate": 1.471735561512072e-06, "loss": 0.067, "step": 100930 }, { "epoch": 48.44, "learning_rate": 1.4682795396274873e-06, "loss": 0.072, "step": 100940 }, { "epoch": 48.44, "learning_rate": 1.4648275198691663e-06, "loss": 0.0722, "step": 100950 }, { "epoch": 48.45, "learning_rate": 1.4613795025217647e-06, "loss": 0.0756, "step": 100960 }, { "epoch": 48.45, "learning_rate": 1.4579354878696328e-06, "loss": 0.0634, "step": 100970 }, { "epoch": 48.46, "learning_rate": 1.4544954761967767e-06, "loss": 0.078, "step": 100980 }, { "epoch": 48.46, "learning_rate": 1.451059467786875e-06, "loss": 0.0718, "step": 100990 }, { "epoch": 48.46, "learning_rate": 1.4476274629232678e-06, "loss": 0.0749, "step": 101000 }, { "epoch": 48.47, "learning_rate": 1.4441994618889788e-06, "loss": 0.0668, "step": 101010 }, { "epoch": 48.47, "learning_rate": 1.4407754649666926e-06, "loss": 0.0754, "step": 101020 }, { "epoch": 48.48, "learning_rate": 1.4373554724387671e-06, "loss": 0.0728, "step": 101030 }, { "epoch": 48.48, "learning_rate": 1.4339394845872323e-06, "loss": 0.0712, "step": 101040 }, { "epoch": 48.49, "learning_rate": 1.430527501693779e-06, "loss": 0.0769, "step": 101050 }, { "epoch": 48.49, "learning_rate": 1.4271195240397717e-06, "loss": 0.0704, "step": 101060 }, { "epoch": 48.5, "learning_rate": 1.4237155519062518e-06, "loss": 0.0779, "step": 101070 }, { "epoch": 48.5, "learning_rate": 1.420315585573928e-06, "loss": 0.0762, "step": 101080 }, { "epoch": 48.51, "learning_rate": 1.4169196253231654e-06, "loss": 0.0658, "step": 101090 }, { "epoch": 48.51, "learning_rate": 1.4135276714340229e-06, "loss": 0.071, "step": 101100 }, { "epoch": 48.51, "learning_rate": 1.4101397241861992e-06, "loss": 0.0735, "step": 101110 }, { "epoch": 48.52, "learning_rate": 1.4067557838590873e-06, "loss": 0.0755, "step": 101120 }, { "epoch": 48.52, "learning_rate": 1.4033758507317363e-06, "loss": 0.0791, "step": 101130 }, { "epoch": 48.53, "learning_rate": 1.3999999250828788e-06, "loss": 0.0842, "step": 101140 }, { "epoch": 48.53, "learning_rate": 1.3966280071908977e-06, "loss": 0.0681, "step": 101150 }, { "epoch": 48.54, "learning_rate": 1.3932600973338594e-06, "loss": 0.0682, "step": 101160 }, { "epoch": 48.54, "learning_rate": 1.389896195789503e-06, "loss": 0.0693, "step": 101170 }, { "epoch": 48.55, "learning_rate": 1.386536302835212e-06, "loss": 0.0681, "step": 101180 }, { "epoch": 48.55, "learning_rate": 1.3831804187480701e-06, "loss": 0.0631, "step": 101190 }, { "epoch": 48.56, "learning_rate": 1.379828543804812e-06, "loss": 0.0741, "step": 101200 }, { "epoch": 48.56, "learning_rate": 1.3764806782818497e-06, "loss": 0.0682, "step": 101210 }, { "epoch": 48.56, "learning_rate": 1.3731368224552622e-06, "loss": 0.085, "step": 101220 }, { "epoch": 48.57, "learning_rate": 1.3697969766008011e-06, "loss": 0.076, "step": 101230 }, { "epoch": 48.57, "learning_rate": 1.36646114099388e-06, "loss": 0.068, "step": 101240 }, { "epoch": 48.58, "learning_rate": 1.3631293159095836e-06, "loss": 0.0687, "step": 101250 }, { "epoch": 48.58, "learning_rate": 1.3598015016226706e-06, "loss": 0.0672, "step": 101260 }, { "epoch": 48.59, "learning_rate": 1.3564776984075656e-06, "loss": 0.0822, "step": 101270 }, { "epoch": 48.59, "learning_rate": 1.3531579065383604e-06, "loss": 0.0682, "step": 101280 }, { "epoch": 48.6, "learning_rate": 1.349842126288825e-06, "loss": 0.0758, "step": 101290 }, { "epoch": 48.6, "learning_rate": 1.3465303579323906e-06, "loss": 0.0707, "step": 101300 }, { "epoch": 48.61, "learning_rate": 1.3432226017421557e-06, "loss": 0.0715, "step": 101310 }, { "epoch": 48.61, "learning_rate": 1.3399188579908962e-06, "loss": 0.0771, "step": 101320 }, { "epoch": 48.62, "learning_rate": 1.3366191269510441e-06, "loss": 0.0716, "step": 101330 }, { "epoch": 48.62, "learning_rate": 1.3333234088947211e-06, "loss": 0.0728, "step": 101340 }, { "epoch": 48.62, "learning_rate": 1.3300317040936926e-06, "loss": 0.0662, "step": 101350 }, { "epoch": 48.63, "learning_rate": 1.3267440128194198e-06, "loss": 0.0777, "step": 101360 }, { "epoch": 48.63, "learning_rate": 1.3234603353430076e-06, "loss": 0.0664, "step": 101370 }, { "epoch": 48.64, "learning_rate": 1.3201806719352505e-06, "loss": 0.0683, "step": 101380 }, { "epoch": 48.64, "learning_rate": 1.316905022866599e-06, "loss": 0.0753, "step": 101390 }, { "epoch": 48.65, "learning_rate": 1.3136333884071706e-06, "loss": 0.0778, "step": 101400 }, { "epoch": 48.65, "learning_rate": 1.3103657688267657e-06, "loss": 0.0789, "step": 101410 }, { "epoch": 48.66, "learning_rate": 1.3071021643948522e-06, "loss": 0.0835, "step": 101420 }, { "epoch": 48.66, "learning_rate": 1.3038425753805428e-06, "loss": 0.0688, "step": 101430 }, { "epoch": 48.67, "learning_rate": 1.3005870020526445e-06, "loss": 0.0715, "step": 101440 }, { "epoch": 48.67, "learning_rate": 1.297335444679637e-06, "loss": 0.0677, "step": 101450 }, { "epoch": 48.67, "learning_rate": 1.2940879035296394e-06, "loss": 0.0665, "step": 101460 }, { "epoch": 48.68, "learning_rate": 1.2908443788704649e-06, "loss": 0.0745, "step": 101470 }, { "epoch": 48.68, "learning_rate": 1.2876048709695942e-06, "loss": 0.0705, "step": 101480 }, { "epoch": 48.69, "learning_rate": 1.2843693800941636e-06, "loss": 0.0712, "step": 101490 }, { "epoch": 48.69, "learning_rate": 1.281137906510982e-06, "loss": 0.0744, "step": 101500 }, { "epoch": 48.7, "learning_rate": 1.2779104504865246e-06, "loss": 0.0723, "step": 101510 }, { "epoch": 48.7, "learning_rate": 1.2746870122869624e-06, "loss": 0.0672, "step": 101520 }, { "epoch": 48.71, "learning_rate": 1.2714675921780995e-06, "loss": 0.0781, "step": 101530 }, { "epoch": 48.71, "learning_rate": 1.268252190425423e-06, "loss": 0.0653, "step": 101540 }, { "epoch": 48.72, "learning_rate": 1.2650408072940878e-06, "loss": 0.072, "step": 101550 }, { "epoch": 48.72, "learning_rate": 1.2618334430489265e-06, "loss": 0.0648, "step": 101560 }, { "epoch": 48.72, "learning_rate": 1.2586300979544163e-06, "loss": 0.0699, "step": 101570 }, { "epoch": 48.73, "learning_rate": 1.2554307722747293e-06, "loss": 0.0758, "step": 101580 }, { "epoch": 48.73, "learning_rate": 1.252235466273688e-06, "loss": 0.0685, "step": 101590 }, { "epoch": 48.74, "learning_rate": 1.2490441802148033e-06, "loss": 0.0688, "step": 101600 }, { "epoch": 48.74, "learning_rate": 1.2458569143612265e-06, "loss": 0.0697, "step": 101610 }, { "epoch": 48.75, "learning_rate": 1.2426736689758022e-06, "loss": 0.0713, "step": 101620 }, { "epoch": 48.75, "learning_rate": 1.2394944443210322e-06, "loss": 0.0707, "step": 101630 }, { "epoch": 48.76, "learning_rate": 1.2363192406590895e-06, "loss": 0.0634, "step": 101640 }, { "epoch": 48.76, "learning_rate": 1.2331480582518095e-06, "loss": 0.0697, "step": 101650 }, { "epoch": 48.77, "learning_rate": 1.2299808973606996e-06, "loss": 0.0682, "step": 101660 }, { "epoch": 48.77, "learning_rate": 1.2268177582469454e-06, "loss": 0.0697, "step": 101670 }, { "epoch": 48.78, "learning_rate": 1.2236586411713936e-06, "loss": 0.0635, "step": 101680 }, { "epoch": 48.78, "learning_rate": 1.2205035463945468e-06, "loss": 0.0694, "step": 101690 }, { "epoch": 48.78, "learning_rate": 1.2173524741765917e-06, "loss": 0.0747, "step": 101700 }, { "epoch": 48.79, "learning_rate": 1.2142054247773926e-06, "loss": 0.0689, "step": 101710 }, { "epoch": 48.79, "learning_rate": 1.2110623984564473e-06, "loss": 0.0708, "step": 101720 }, { "epoch": 48.8, "learning_rate": 1.2079233954729484e-06, "loss": 0.0716, "step": 101730 }, { "epoch": 48.8, "learning_rate": 1.204788416085756e-06, "loss": 0.0752, "step": 101740 }, { "epoch": 48.81, "learning_rate": 1.201657460553396e-06, "loss": 0.0678, "step": 101750 }, { "epoch": 48.81, "learning_rate": 1.198530529134051e-06, "loss": 0.0699, "step": 101760 }, { "epoch": 48.82, "learning_rate": 1.1954076220855869e-06, "loss": 0.0632, "step": 101770 }, { "epoch": 48.82, "learning_rate": 1.1922887396655313e-06, "loss": 0.0711, "step": 101780 }, { "epoch": 48.83, "learning_rate": 1.1891738821310782e-06, "loss": 0.0646, "step": 101790 }, { "epoch": 48.83, "learning_rate": 1.186063049739089e-06, "loss": 0.0721, "step": 101800 }, { "epoch": 48.83, "learning_rate": 1.1829562427461026e-06, "loss": 0.0792, "step": 101810 }, { "epoch": 48.84, "learning_rate": 1.179853461408309e-06, "loss": 0.0724, "step": 101820 }, { "epoch": 48.84, "learning_rate": 1.176754705981592e-06, "loss": 0.0721, "step": 101830 }, { "epoch": 48.85, "learning_rate": 1.1736599767214696e-06, "loss": 0.0769, "step": 101840 }, { "epoch": 48.85, "learning_rate": 1.1705692738831654e-06, "loss": 0.0652, "step": 101850 }, { "epoch": 48.86, "learning_rate": 1.1674825977215365e-06, "loss": 0.0696, "step": 101860 }, { "epoch": 48.86, "learning_rate": 1.164399948491124e-06, "loss": 0.0682, "step": 101870 }, { "epoch": 48.87, "learning_rate": 1.1613213264461408e-06, "loss": 0.0636, "step": 101880 }, { "epoch": 48.87, "learning_rate": 1.158246731840462e-06, "loss": 0.0676, "step": 101890 }, { "epoch": 48.88, "learning_rate": 1.1551761649276293e-06, "loss": 0.0744, "step": 101900 }, { "epoch": 48.88, "learning_rate": 1.1521096259608566e-06, "loss": 0.0706, "step": 101910 }, { "epoch": 48.88, "learning_rate": 1.1490471151930304e-06, "loss": 0.0703, "step": 101920 }, { "epoch": 48.89, "learning_rate": 1.1459886328766822e-06, "loss": 0.0695, "step": 101930 }, { "epoch": 48.89, "learning_rate": 1.142934179264038e-06, "loss": 0.073, "step": 101940 }, { "epoch": 48.9, "learning_rate": 1.139883754606974e-06, "loss": 0.06, "step": 101950 }, { "epoch": 48.9, "learning_rate": 1.1368373591570447e-06, "loss": 0.0709, "step": 101960 }, { "epoch": 48.91, "learning_rate": 1.1337949931654713e-06, "loss": 0.0693, "step": 101970 }, { "epoch": 48.91, "learning_rate": 1.1307566568831362e-06, "loss": 0.0638, "step": 101980 }, { "epoch": 48.92, "learning_rate": 1.1277223505606005e-06, "loss": 0.0645, "step": 101990 }, { "epoch": 48.92, "learning_rate": 1.1246920744480693e-06, "loss": 0.0661, "step": 102000 }, { "epoch": 48.93, "learning_rate": 1.1216658287954486e-06, "loss": 0.0707, "step": 102010 }, { "epoch": 48.93, "learning_rate": 1.1186436138522827e-06, "loss": 0.067, "step": 102020 }, { "epoch": 48.94, "learning_rate": 1.1156254298678005e-06, "loss": 0.0793, "step": 102030 }, { "epoch": 48.94, "learning_rate": 1.1126112770908969e-06, "loss": 0.0673, "step": 102040 }, { "epoch": 48.94, "learning_rate": 1.1096011557701291e-06, "loss": 0.0715, "step": 102050 }, { "epoch": 48.95, "learning_rate": 1.1065950661537206e-06, "loss": 0.0734, "step": 102060 }, { "epoch": 49.0, "learning_rate": 1.1035930084895784e-06, "loss": 0.0753, "step": 102070 }, { "epoch": 49.01, "learning_rate": 1.1005949830252494e-06, "loss": 0.0714, "step": 102080 }, { "epoch": 49.01, "learning_rate": 1.0976009900079743e-06, "loss": 0.0656, "step": 102090 }, { "epoch": 49.02, "learning_rate": 1.0946110296846446e-06, "loss": 0.0811, "step": 102100 }, { "epoch": 49.02, "learning_rate": 1.0916251023018187e-06, "loss": 0.0725, "step": 102110 }, { "epoch": 49.02, "learning_rate": 1.0886432081057386e-06, "loss": 0.0766, "step": 102120 }, { "epoch": 49.03, "learning_rate": 1.0856653473423017e-06, "loss": 0.0718, "step": 102130 }, { "epoch": 49.03, "learning_rate": 1.0826915202570787e-06, "loss": 0.0713, "step": 102140 }, { "epoch": 49.04, "learning_rate": 1.0797217270952897e-06, "loss": 0.0742, "step": 102150 }, { "epoch": 49.04, "learning_rate": 1.0767559681018446e-06, "loss": 0.0704, "step": 102160 }, { "epoch": 49.05, "learning_rate": 1.07379424352132e-06, "loss": 0.0643, "step": 102170 }, { "epoch": 49.05, "learning_rate": 1.0708365535979425e-06, "loss": 0.0713, "step": 102180 }, { "epoch": 49.06, "learning_rate": 1.067882898575606e-06, "loss": 0.0674, "step": 102190 }, { "epoch": 49.06, "learning_rate": 1.064933278697905e-06, "loss": 0.0796, "step": 102200 }, { "epoch": 49.07, "learning_rate": 1.061987694208061e-06, "loss": 0.0627, "step": 102210 }, { "epoch": 49.07, "learning_rate": 1.0590461453489853e-06, "loss": 0.0707, "step": 102220 }, { "epoch": 49.07, "learning_rate": 1.0561086323632396e-06, "loss": 0.0664, "step": 102230 }, { "epoch": 49.08, "learning_rate": 1.0531751554930857e-06, "loss": 0.0798, "step": 102240 }, { "epoch": 49.08, "learning_rate": 1.0502457149804025e-06, "loss": 0.0718, "step": 102250 }, { "epoch": 49.09, "learning_rate": 1.0473203110667794e-06, "loss": 0.0704, "step": 102260 }, { "epoch": 49.09, "learning_rate": 1.0443989439934521e-06, "loss": 0.0663, "step": 102270 }, { "epoch": 49.1, "learning_rate": 1.041481614001344e-06, "loss": 0.0703, "step": 102280 }, { "epoch": 49.1, "learning_rate": 1.0385683213310072e-06, "loss": 0.0687, "step": 102290 }, { "epoch": 49.11, "learning_rate": 1.0356590662226994e-06, "loss": 0.0814, "step": 102300 }, { "epoch": 49.11, "learning_rate": 1.0327538489163236e-06, "loss": 0.0718, "step": 102310 }, { "epoch": 49.12, "learning_rate": 1.0298526696514598e-06, "loss": 0.0859, "step": 102320 }, { "epoch": 49.12, "learning_rate": 1.026955528667345e-06, "loss": 0.0692, "step": 102330 }, { "epoch": 49.12, "learning_rate": 1.0240624262028931e-06, "loss": 0.0695, "step": 102340 }, { "epoch": 49.13, "learning_rate": 1.0211733624966802e-06, "loss": 0.0684, "step": 102350 }, { "epoch": 49.13, "learning_rate": 1.018288337786949e-06, "loss": 0.0687, "step": 102360 }, { "epoch": 49.14, "learning_rate": 1.0154073523116203e-06, "loss": 0.07, "step": 102370 }, { "epoch": 49.14, "learning_rate": 1.0125304063082597e-06, "loss": 0.0638, "step": 102380 }, { "epoch": 49.15, "learning_rate": 1.0096575000141218e-06, "loss": 0.0702, "step": 102390 }, { "epoch": 49.15, "learning_rate": 1.0067886336661114e-06, "loss": 0.0722, "step": 102400 }, { "epoch": 49.16, "learning_rate": 1.0039238075008117e-06, "loss": 0.0723, "step": 102410 }, { "epoch": 49.16, "learning_rate": 1.0010630217544614e-06, "loss": 0.0672, "step": 102420 }, { "epoch": 49.17, "learning_rate": 9.982062766629775e-07, "loss": 0.0746, "step": 102430 }, { "epoch": 49.17, "learning_rate": 9.953535724619379e-07, "loss": 0.0712, "step": 102440 }, { "epoch": 49.18, "learning_rate": 9.925049093865934e-07, "loss": 0.0709, "step": 102450 }, { "epoch": 49.18, "learning_rate": 9.896602876718564e-07, "loss": 0.0717, "step": 102460 }, { "epoch": 49.18, "learning_rate": 9.868197075522945e-07, "loss": 0.0754, "step": 102470 }, { "epoch": 49.19, "learning_rate": 9.839831692621593e-07, "loss": 0.0668, "step": 102480 }, { "epoch": 49.19, "learning_rate": 9.811506730353693e-07, "loss": 0.0766, "step": 102490 }, { "epoch": 49.2, "learning_rate": 9.783222191054986e-07, "loss": 0.0697, "step": 102500 }, { "epoch": 49.2, "learning_rate": 9.754978077057942e-07, "loss": 0.0662, "step": 102510 }, { "epoch": 49.21, "learning_rate": 9.726774390691696e-07, "loss": 0.0723, "step": 102520 }, { "epoch": 49.21, "learning_rate": 9.69861113428211e-07, "loss": 0.0734, "step": 102530 }, { "epoch": 49.22, "learning_rate": 9.670488310151494e-07, "loss": 0.0651, "step": 102540 }, { "epoch": 49.22, "learning_rate": 9.642405920619046e-07, "loss": 0.0677, "step": 102550 }, { "epoch": 49.23, "learning_rate": 9.614363968000528e-07, "loss": 0.0748, "step": 102560 }, { "epoch": 49.23, "learning_rate": 9.586362454608476e-07, "loss": 0.0792, "step": 102570 }, { "epoch": 49.23, "learning_rate": 9.558401382751935e-07, "loss": 0.067, "step": 102580 }, { "epoch": 49.24, "learning_rate": 9.53048075473667e-07, "loss": 0.0693, "step": 102590 }, { "epoch": 49.24, "learning_rate": 9.502600572865284e-07, "loss": 0.0718, "step": 102600 }, { "epoch": 49.25, "learning_rate": 9.474760839436713e-07, "loss": 0.0696, "step": 102610 }, { "epoch": 49.25, "learning_rate": 9.446961556746792e-07, "loss": 0.0715, "step": 102620 }, { "epoch": 49.26, "learning_rate": 9.419202727088017e-07, "loss": 0.0809, "step": 102630 }, { "epoch": 49.26, "learning_rate": 9.391484352749446e-07, "loss": 0.0684, "step": 102640 }, { "epoch": 49.27, "learning_rate": 9.363806436016809e-07, "loss": 0.0673, "step": 102650 }, { "epoch": 49.27, "learning_rate": 9.336168979172666e-07, "loss": 0.0697, "step": 102660 }, { "epoch": 49.28, "learning_rate": 9.308571984496029e-07, "loss": 0.0666, "step": 102670 }, { "epoch": 49.28, "learning_rate": 9.281015454262687e-07, "loss": 0.071, "step": 102680 }, { "epoch": 49.28, "learning_rate": 9.253499390745046e-07, "loss": 0.0694, "step": 102690 }, { "epoch": 49.29, "learning_rate": 9.22602379621218e-07, "loss": 0.0667, "step": 102700 }, { "epoch": 49.29, "learning_rate": 9.198588672929886e-07, "loss": 0.0773, "step": 102710 }, { "epoch": 49.3, "learning_rate": 9.171194023160523e-07, "loss": 0.073, "step": 102720 }, { "epoch": 49.3, "learning_rate": 9.143839849163227e-07, "loss": 0.0706, "step": 102730 }, { "epoch": 49.31, "learning_rate": 9.116526153193694e-07, "loss": 0.074, "step": 102740 }, { "epoch": 49.31, "learning_rate": 9.0892529375044e-07, "loss": 0.0749, "step": 102750 }, { "epoch": 49.32, "learning_rate": 9.062020204344268e-07, "loss": 0.077, "step": 102760 }, { "epoch": 49.32, "learning_rate": 9.034827955959114e-07, "loss": 0.0662, "step": 102770 }, { "epoch": 49.33, "learning_rate": 9.007676194591363e-07, "loss": 0.0681, "step": 102780 }, { "epoch": 49.33, "learning_rate": 8.980564922479895e-07, "loss": 0.0711, "step": 102790 }, { "epoch": 49.34, "learning_rate": 8.953494141860585e-07, "loss": 0.0711, "step": 102800 }, { "epoch": 49.34, "learning_rate": 8.926463854965761e-07, "loss": 0.0675, "step": 102810 }, { "epoch": 49.34, "learning_rate": 8.899474064024471e-07, "loss": 0.0723, "step": 102820 }, { "epoch": 49.35, "learning_rate": 8.872524771262325e-07, "loss": 0.0726, "step": 102830 }, { "epoch": 49.35, "learning_rate": 8.845615978901712e-07, "loss": 0.0741, "step": 102840 }, { "epoch": 49.36, "learning_rate": 8.818747689161689e-07, "loss": 0.0664, "step": 102850 }, { "epoch": 49.36, "learning_rate": 8.791919904257873e-07, "loss": 0.0653, "step": 102860 }, { "epoch": 49.37, "learning_rate": 8.76513262640255e-07, "loss": 0.071, "step": 102870 }, { "epoch": 49.37, "learning_rate": 8.738385857804787e-07, "loss": 0.066, "step": 102880 }, { "epoch": 49.38, "learning_rate": 8.711679600670264e-07, "loss": 0.0647, "step": 102890 }, { "epoch": 49.38, "learning_rate": 8.685013857201219e-07, "loss": 0.0733, "step": 102900 }, { "epoch": 49.39, "learning_rate": 8.658388629596614e-07, "loss": 0.0656, "step": 102910 }, { "epoch": 49.39, "learning_rate": 8.631803920052084e-07, "loss": 0.0689, "step": 102920 }, { "epoch": 49.39, "learning_rate": 8.605259730759985e-07, "loss": 0.0757, "step": 102930 }, { "epoch": 49.4, "learning_rate": 8.578756063909177e-07, "loss": 0.0776, "step": 102940 }, { "epoch": 49.4, "learning_rate": 8.552292921685246e-07, "loss": 0.0741, "step": 102950 }, { "epoch": 49.41, "learning_rate": 8.525870306270611e-07, "loss": 0.0637, "step": 102960 }, { "epoch": 49.41, "learning_rate": 8.499488219843976e-07, "loss": 0.0683, "step": 102970 }, { "epoch": 49.42, "learning_rate": 8.473146664581044e-07, "loss": 0.0674, "step": 102980 }, { "epoch": 49.42, "learning_rate": 8.446845642654022e-07, "loss": 0.0655, "step": 102990 }, { "epoch": 49.43, "learning_rate": 8.420585156231842e-07, "loss": 0.0744, "step": 103000 }, { "epoch": 49.43, "learning_rate": 8.394365207479993e-07, "loss": 0.0709, "step": 103010 }, { "epoch": 49.44, "learning_rate": 8.368185798560635e-07, "loss": 0.0754, "step": 103020 }, { "epoch": 49.44, "learning_rate": 8.342046931632764e-07, "loss": 0.0706, "step": 103030 }, { "epoch": 49.44, "learning_rate": 8.315948608851764e-07, "loss": 0.0675, "step": 103040 }, { "epoch": 49.45, "learning_rate": 8.289890832369918e-07, "loss": 0.0589, "step": 103050 }, { "epoch": 49.45, "learning_rate": 8.263873604336003e-07, "loss": 0.0685, "step": 103060 }, { "epoch": 49.46, "learning_rate": 8.237896926895527e-07, "loss": 0.0694, "step": 103070 }, { "epoch": 49.46, "learning_rate": 8.21196080219061e-07, "loss": 0.0657, "step": 103080 }, { "epoch": 49.47, "learning_rate": 8.186065232360096e-07, "loss": 0.0713, "step": 103090 }, { "epoch": 49.47, "learning_rate": 8.160210219539333e-07, "loss": 0.0676, "step": 103100 }, { "epoch": 49.48, "learning_rate": 8.134395765860614e-07, "loss": 0.064, "step": 103110 }, { "epoch": 49.48, "learning_rate": 8.108621873452571e-07, "loss": 0.072, "step": 103120 }, { "epoch": 49.49, "learning_rate": 8.082888544440614e-07, "loss": 0.068, "step": 103130 }, { "epoch": 49.49, "learning_rate": 8.057195780946936e-07, "loss": 0.0696, "step": 103140 }, { "epoch": 49.49, "learning_rate": 8.031543585090173e-07, "loss": 0.0708, "step": 103150 }, { "epoch": 49.5, "learning_rate": 8.005931958985691e-07, "loss": 0.0726, "step": 103160 }, { "epoch": 49.5, "learning_rate": 7.980360904745632e-07, "loss": 0.0636, "step": 103170 }, { "epoch": 49.51, "learning_rate": 7.954830424478588e-07, "loss": 0.0743, "step": 103180 }, { "epoch": 49.51, "learning_rate": 7.929340520289929e-07, "loss": 0.0687, "step": 103190 }, { "epoch": 49.52, "learning_rate": 7.903891194281754e-07, "loss": 0.0655, "step": 103200 }, { "epoch": 49.52, "learning_rate": 7.87848244855266e-07, "loss": 0.078, "step": 103210 }, { "epoch": 49.53, "learning_rate": 7.85311428519786e-07, "loss": 0.0657, "step": 103220 }, { "epoch": 49.53, "learning_rate": 7.827786706309459e-07, "loss": 0.0703, "step": 103230 }, { "epoch": 49.54, "learning_rate": 7.802499713975952e-07, "loss": 0.0724, "step": 103240 }, { "epoch": 49.54, "learning_rate": 7.777253310282729e-07, "loss": 0.0749, "step": 103250 }, { "epoch": 49.55, "learning_rate": 7.752047497311621e-07, "loss": 0.0675, "step": 103260 }, { "epoch": 49.55, "learning_rate": 7.726882277141301e-07, "loss": 0.0727, "step": 103270 }, { "epoch": 49.55, "learning_rate": 7.701757651846886e-07, "loss": 0.0692, "step": 103280 }, { "epoch": 49.56, "learning_rate": 7.67667362350033e-07, "loss": 0.0767, "step": 103290 }, { "epoch": 49.56, "learning_rate": 7.651630194170145e-07, "loss": 0.0705, "step": 103300 }, { "epoch": 49.57, "learning_rate": 7.626627365921458e-07, "loss": 0.0694, "step": 103310 }, { "epoch": 49.57, "learning_rate": 7.601665140816228e-07, "loss": 0.0921, "step": 103320 }, { "epoch": 49.58, "learning_rate": 7.57674352091281e-07, "loss": 0.0718, "step": 103330 }, { "epoch": 49.58, "learning_rate": 7.55186250826645e-07, "loss": 0.062, "step": 103340 }, { "epoch": 49.59, "learning_rate": 7.527022104928894e-07, "loss": 0.0766, "step": 103350 }, { "epoch": 49.59, "learning_rate": 7.502222312948615e-07, "loss": 0.0705, "step": 103360 }, { "epoch": 49.6, "learning_rate": 7.477463134370644e-07, "loss": 0.0655, "step": 103370 }, { "epoch": 49.6, "learning_rate": 7.452744571236792e-07, "loss": 0.0777, "step": 103380 }, { "epoch": 49.6, "learning_rate": 7.428066625585373e-07, "loss": 0.0628, "step": 103390 }, { "epoch": 49.61, "learning_rate": 7.403429299451537e-07, "loss": 0.0693, "step": 103400 }, { "epoch": 49.61, "learning_rate": 7.37883259486688e-07, "loss": 0.0674, "step": 103410 }, { "epoch": 49.62, "learning_rate": 7.354276513859837e-07, "loss": 0.067, "step": 103420 }, { "epoch": 49.62, "learning_rate": 7.329761058455342e-07, "loss": 0.0755, "step": 103430 }, { "epoch": 49.63, "learning_rate": 7.305286230675057e-07, "loss": 0.0693, "step": 103440 }, { "epoch": 49.63, "learning_rate": 7.280852032537255e-07, "loss": 0.0657, "step": 103450 }, { "epoch": 49.64, "learning_rate": 7.256458466056992e-07, "loss": 0.0675, "step": 103460 }, { "epoch": 49.64, "learning_rate": 7.232105533245658e-07, "loss": 0.0743, "step": 103470 }, { "epoch": 49.65, "learning_rate": 7.207793236111649e-07, "loss": 0.0729, "step": 103480 }, { "epoch": 49.65, "learning_rate": 7.183521576659858e-07, "loss": 0.0732, "step": 103490 }, { "epoch": 49.65, "learning_rate": 7.159290556891796e-07, "loss": 0.0754, "step": 103500 }, { "epoch": 49.66, "learning_rate": 7.135100178805643e-07, "loss": 0.0719, "step": 103510 }, { "epoch": 49.66, "learning_rate": 7.110950444396192e-07, "loss": 0.0768, "step": 103520 }, { "epoch": 49.67, "learning_rate": 7.086841355655016e-07, "loss": 0.072, "step": 103530 }, { "epoch": 49.67, "learning_rate": 7.062772914570248e-07, "loss": 0.076, "step": 103540 }, { "epoch": 49.68, "learning_rate": 7.038745123126578e-07, "loss": 0.067, "step": 103550 }, { "epoch": 49.68, "learning_rate": 7.014757983305476e-07, "loss": 0.0671, "step": 103560 }, { "epoch": 49.69, "learning_rate": 6.990811497085136e-07, "loss": 0.0764, "step": 103570 }, { "epoch": 49.69, "learning_rate": 6.966905666440148e-07, "loss": 0.0742, "step": 103580 }, { "epoch": 49.7, "learning_rate": 6.943040493341879e-07, "loss": 0.067, "step": 103590 }, { "epoch": 49.7, "learning_rate": 6.919215979758476e-07, "loss": 0.074, "step": 103600 }, { "epoch": 49.71, "learning_rate": 6.895432127654533e-07, "loss": 0.0741, "step": 103610 }, { "epoch": 49.71, "learning_rate": 6.871688938991261e-07, "loss": 0.0719, "step": 103620 }, { "epoch": 49.71, "learning_rate": 6.847986415726759e-07, "loss": 0.0666, "step": 103630 }, { "epoch": 49.72, "learning_rate": 6.824324559815632e-07, "loss": 0.0711, "step": 103640 }, { "epoch": 49.72, "learning_rate": 6.800703373209094e-07, "loss": 0.0732, "step": 103650 }, { "epoch": 49.73, "learning_rate": 6.777122857855034e-07, "loss": 0.0772, "step": 103660 }, { "epoch": 49.73, "learning_rate": 6.753583015698062e-07, "loss": 0.0693, "step": 103670 }, { "epoch": 49.74, "learning_rate": 6.730083848679291e-07, "loss": 0.0632, "step": 103680 }, { "epoch": 49.74, "learning_rate": 6.706625358736563e-07, "loss": 0.0712, "step": 103690 }, { "epoch": 49.75, "learning_rate": 6.683207547804382e-07, "loss": 0.066, "step": 103700 }, { "epoch": 49.75, "learning_rate": 6.659830417813928e-07, "loss": 0.0767, "step": 103710 }, { "epoch": 49.76, "learning_rate": 6.636493970692992e-07, "loss": 0.0727, "step": 103720 }, { "epoch": 49.76, "learning_rate": 6.613198208365867e-07, "loss": 0.0716, "step": 103730 }, { "epoch": 49.76, "learning_rate": 6.589943132753684e-07, "loss": 0.0712, "step": 103740 }, { "epoch": 49.77, "learning_rate": 6.566728745774186e-07, "loss": 0.0662, "step": 103750 }, { "epoch": 49.77, "learning_rate": 6.543555049341676e-07, "loss": 0.0695, "step": 103760 }, { "epoch": 49.78, "learning_rate": 6.52042204536718e-07, "loss": 0.0737, "step": 103770 }, { "epoch": 49.78, "learning_rate": 6.497329735758339e-07, "loss": 0.0657, "step": 103780 }, { "epoch": 49.79, "learning_rate": 6.474278122419464e-07, "loss": 0.062, "step": 103790 }, { "epoch": 49.79, "learning_rate": 6.451267207251422e-07, "loss": 0.0748, "step": 103800 }, { "epoch": 49.8, "learning_rate": 6.428296992151805e-07, "loss": 0.0664, "step": 103810 }, { "epoch": 49.8, "learning_rate": 6.405367479014934e-07, "loss": 0.0695, "step": 103820 }, { "epoch": 49.81, "learning_rate": 6.382478669731518e-07, "loss": 0.0784, "step": 103830 }, { "epoch": 49.81, "learning_rate": 6.359630566189156e-07, "loss": 0.0693, "step": 103840 }, { "epoch": 49.81, "learning_rate": 6.336823170272011e-07, "loss": 0.0676, "step": 103850 }, { "epoch": 49.82, "learning_rate": 6.314056483860798e-07, "loss": 0.0649, "step": 103860 }, { "epoch": 49.82, "learning_rate": 6.29133050883296e-07, "loss": 0.074, "step": 103870 }, { "epoch": 49.83, "learning_rate": 6.268645247062666e-07, "loss": 0.0742, "step": 103880 }, { "epoch": 49.83, "learning_rate": 6.246000700420584e-07, "loss": 0.0657, "step": 103890 }, { "epoch": 49.84, "learning_rate": 6.223396870774056e-07, "loss": 0.0686, "step": 103900 }, { "epoch": 49.84, "learning_rate": 6.200833759987091e-07, "loss": 0.0655, "step": 103910 }, { "epoch": 49.85, "learning_rate": 6.178311369920364e-07, "loss": 0.0691, "step": 103920 }, { "epoch": 49.85, "learning_rate": 6.15582970243117e-07, "loss": 0.0764, "step": 103930 }, { "epoch": 49.86, "learning_rate": 6.13338875937336e-07, "loss": 0.0774, "step": 103940 }, { "epoch": 49.86, "learning_rate": 6.110988542597617e-07, "loss": 0.0736, "step": 103950 }, { "epoch": 49.87, "learning_rate": 6.088629053951079e-07, "loss": 0.099, "step": 103960 }, { "epoch": 49.87, "learning_rate": 6.066310295277655e-07, "loss": 0.0801, "step": 103970 }, { "epoch": 49.87, "learning_rate": 6.044032268417821e-07, "loss": 0.0685, "step": 103980 }, { "epoch": 49.88, "learning_rate": 6.021794975208661e-07, "loss": 0.0753, "step": 103990 }, { "epoch": 49.88, "learning_rate": 5.999598417484042e-07, "loss": 0.0677, "step": 104000 }, { "epoch": 49.89, "learning_rate": 5.977442597074334e-07, "loss": 0.0742, "step": 104010 }, { "epoch": 49.89, "learning_rate": 5.955327515806575e-07, "loss": 0.0721, "step": 104020 }, { "epoch": 49.9, "learning_rate": 5.933253175504528e-07, "loss": 0.0772, "step": 104030 }, { "epoch": 49.9, "learning_rate": 5.911219577988514e-07, "loss": 0.0671, "step": 104040 }, { "epoch": 49.91, "learning_rate": 5.889226725075525e-07, "loss": 0.0689, "step": 104050 }, { "epoch": 49.91, "learning_rate": 5.867274618579111e-07, "loss": 0.0753, "step": 104060 }, { "epoch": 49.92, "learning_rate": 5.845363260309599e-07, "loss": 0.0702, "step": 104070 }, { "epoch": 49.92, "learning_rate": 5.823492652073881e-07, "loss": 0.0745, "step": 104080 }, { "epoch": 49.92, "learning_rate": 5.801662795675511e-07, "loss": 0.0793, "step": 104090 }, { "epoch": 49.93, "learning_rate": 5.779873692914606e-07, "loss": 0.0686, "step": 104100 }, { "epoch": 49.93, "learning_rate": 5.758125345588117e-07, "loss": 0.0769, "step": 104110 }, { "epoch": 49.94, "learning_rate": 5.736417755489332e-07, "loss": 0.0759, "step": 104120 }, { "epoch": 49.94, "learning_rate": 5.714750924408485e-07, "loss": 0.0666, "step": 104130 }, { "epoch": 49.95, "learning_rate": 5.693124854132259e-07, "loss": 0.0742, "step": 104140 }, { "epoch": 49.95, "learning_rate": 5.671539546444004e-07, "loss": 0.0774, "step": 104150 } ], "max_steps": 109400, "num_train_epochs": 50, "total_flos": 0.0, "trial_name": null, "trial_params": null }