{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.29246358341037, "global_step": 24000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.05, "learning_rate": 1.978544061302682e-05, "loss": 9.6964, "step": 20 }, { "epoch": 0.09, "learning_rate": 1.9478927203065136e-05, "loss": 7.2368, "step": 40 }, { "epoch": 0.14, "learning_rate": 1.917241379310345e-05, "loss": 6.2605, "step": 60 }, { "epoch": 0.18, "learning_rate": 1.8865900383141766e-05, "loss": 4.7284, "step": 80 }, { "epoch": 0.23, "learning_rate": 1.855938697318008e-05, "loss": 4.0496, "step": 100 }, { "epoch": 0.28, "learning_rate": 1.8252873563218392e-05, "loss": 3.7911, "step": 120 }, { "epoch": 0.32, "learning_rate": 1.7946360153256706e-05, "loss": 3.6108, "step": 140 }, { "epoch": 0.37, "learning_rate": 1.7639846743295022e-05, "loss": 3.5038, "step": 160 }, { "epoch": 0.41, "learning_rate": 1.7333333333333336e-05, "loss": 3.409, "step": 180 }, { "epoch": 0.46, "learning_rate": 1.702681992337165e-05, "loss": 3.3203, "step": 200 }, { "epoch": 0.51, "learning_rate": 1.6720306513409962e-05, "loss": 3.2725, "step": 220 }, { "epoch": 0.55, "learning_rate": 1.6413793103448276e-05, "loss": 3.2151, "step": 240 }, { "epoch": 0.6, "learning_rate": 1.610727969348659e-05, "loss": 3.1602, "step": 260 }, { "epoch": 0.64, "learning_rate": 1.5800766283524906e-05, "loss": 3.1164, "step": 280 }, { "epoch": 0.69, "learning_rate": 1.549425287356322e-05, "loss": 3.0772, "step": 300 }, { "epoch": 0.74, "learning_rate": 1.5187739463601534e-05, "loss": 3.0443, "step": 320 }, { "epoch": 0.78, "learning_rate": 1.4881226053639847e-05, "loss": 3.022, "step": 340 }, { "epoch": 0.83, "learning_rate": 1.4574712643678162e-05, "loss": 3.0014, "step": 360 }, { "epoch": 0.87, "learning_rate": 1.4268199233716476e-05, "loss": 2.9582, "step": 380 }, { "epoch": 0.92, "learning_rate": 1.396168582375479e-05, "loss": 2.9272, "step": 400 }, { "epoch": 0.97, "learning_rate": 1.3655172413793106e-05, "loss": 2.8951, "step": 420 }, { "epoch": 1.01, "learning_rate": 1.3348659003831419e-05, "loss": 2.9101, "step": 440 }, { "epoch": 1.06, "learning_rate": 1.3042145593869732e-05, "loss": 2.8377, "step": 460 }, { "epoch": 1.1, "learning_rate": 1.2735632183908047e-05, "loss": 2.8211, "step": 480 }, { "epoch": 1.15, "learning_rate": 1.242911877394636e-05, "loss": 2.7957, "step": 500 }, { "epoch": 0.03, "learning_rate": 1.981905326081103e-05, "loss": 2.8029, "step": 510 }, { "epoch": 0.03, "learning_rate": 1.9815463047731884e-05, "loss": 2.8451, "step": 520 }, { "epoch": 0.03, "learning_rate": 1.981187283465274e-05, "loss": 2.9085, "step": 530 }, { "epoch": 0.03, "learning_rate": 1.9808282621573592e-05, "loss": 2.9984, "step": 540 }, { "epoch": 0.03, "learning_rate": 1.9804692408494444e-05, "loss": 2.8696, "step": 550 }, { "epoch": 0.03, "learning_rate": 1.98011021954153e-05, "loss": 2.8919, "step": 560 }, { "epoch": 0.03, "learning_rate": 1.9797511982336152e-05, "loss": 2.8115, "step": 570 }, { "epoch": 0.03, "learning_rate": 1.9793921769257008e-05, "loss": 2.7233, "step": 580 }, { "epoch": 0.03, "learning_rate": 1.979033155617786e-05, "loss": 2.9758, "step": 590 }, { "epoch": 0.03, "learning_rate": 1.9786741343098715e-05, "loss": 2.8503, "step": 600 }, { "epoch": 0.03, "learning_rate": 1.9783151130019567e-05, "loss": 2.7586, "step": 610 }, { "epoch": 0.03, "learning_rate": 1.977956091694042e-05, "loss": 2.898, "step": 620 }, { "epoch": 0.03, "learning_rate": 1.9775970703861275e-05, "loss": 2.7126, "step": 630 }, { "epoch": 0.03, "learning_rate": 1.977238049078213e-05, "loss": 2.8214, "step": 640 }, { "epoch": 0.04, "learning_rate": 1.9768790277702983e-05, "loss": 2.8384, "step": 650 }, { "epoch": 0.04, "learning_rate": 1.976520006462384e-05, "loss": 2.8592, "step": 660 }, { "epoch": 0.04, "learning_rate": 1.976160985154469e-05, "loss": 2.7752, "step": 670 }, { "epoch": 0.04, "learning_rate": 1.9758019638465543e-05, "loss": 2.9652, "step": 680 }, { "epoch": 0.04, "learning_rate": 1.9754429425386398e-05, "loss": 2.8639, "step": 690 }, { "epoch": 0.04, "learning_rate": 1.975083921230725e-05, "loss": 3.5679, "step": 700 }, { "epoch": 0.04, "learning_rate": 1.9747248999228106e-05, "loss": 3.842, "step": 710 }, { "epoch": 0.04, "learning_rate": 1.974365878614896e-05, "loss": 2.9028, "step": 720 }, { "epoch": 0.04, "learning_rate": 1.9740068573069814e-05, "loss": 2.9596, "step": 730 }, { "epoch": 0.04, "learning_rate": 1.9736478359990666e-05, "loss": 3.5545, "step": 740 }, { "epoch": 0.04, "learning_rate": 1.9733247168219437e-05, "loss": 9.8081, "step": 750 }, { "epoch": 0.04, "learning_rate": 1.972965695514029e-05, "loss": 4.6766, "step": 760 }, { "epoch": 0.04, "learning_rate": 1.9726784784676972e-05, "loss": 6.1851, "step": 770 }, { "epoch": 0.04, "learning_rate": 1.9723194571597824e-05, "loss": 15.1653, "step": 780 }, { "epoch": 0.04, "learning_rate": 1.971960435851868e-05, "loss": 16.7837, "step": 790 }, { "epoch": 0.04, "learning_rate": 1.9716014145439532e-05, "loss": 16.0732, "step": 800 }, { "epoch": 0.04, "learning_rate": 1.9712423932360387e-05, "loss": 15.0973, "step": 810 }, { "epoch": 0.04, "learning_rate": 1.9708833719281243e-05, "loss": 14.4623, "step": 820 }, { "epoch": 0.04, "learning_rate": 1.9705243506202095e-05, "loss": 13.6565, "step": 830 }, { "epoch": 0.05, "learning_rate": 1.9701653293122947e-05, "loss": 13.3604, "step": 840 }, { "epoch": 0.05, "learning_rate": 1.9698063080043803e-05, "loss": 12.3429, "step": 850 }, { "epoch": 0.05, "learning_rate": 1.9694472866964655e-05, "loss": 11.7149, "step": 860 }, { "epoch": 0.05, "learning_rate": 1.969088265388551e-05, "loss": 8.7238, "step": 870 }, { "epoch": 0.05, "learning_rate": 1.9687292440806363e-05, "loss": 7.7324, "step": 880 }, { "epoch": 0.05, "learning_rate": 1.9683702227727218e-05, "loss": 9.1205, "step": 890 }, { "epoch": 0.05, "learning_rate": 1.968011201464807e-05, "loss": 5.5195, "step": 900 }, { "epoch": 0.05, "learning_rate": 1.9676521801568923e-05, "loss": 4.6745, "step": 910 }, { "epoch": 0.05, "learning_rate": 1.9672931588489778e-05, "loss": 3.2255, "step": 920 }, { "epoch": 0.05, "learning_rate": 1.9669341375410634e-05, "loss": 3.071, "step": 930 }, { "epoch": 0.05, "learning_rate": 1.9665751162331486e-05, "loss": 7.5739, "step": 940 }, { "epoch": 0.05, "learning_rate": 1.966216094925234e-05, "loss": 10.6607, "step": 950 }, { "epoch": 0.05, "learning_rate": 1.9658570736173194e-05, "loss": 9.1666, "step": 960 }, { "epoch": 0.05, "learning_rate": 1.9654980523094046e-05, "loss": 3.3352, "step": 970 }, { "epoch": 0.05, "learning_rate": 1.96513903100149e-05, "loss": 3.0204, "step": 980 }, { "epoch": 0.05, "learning_rate": 1.9647800096935753e-05, "loss": 3.0261, "step": 990 }, { "epoch": 0.05, "learning_rate": 1.964420988385661e-05, "loss": 2.9761, "step": 1000 }, { "epoch": 0.05, "learning_rate": 1.9640619670777464e-05, "loss": 2.9513, "step": 1010 }, { "epoch": 0.05, "learning_rate": 1.9637029457698317e-05, "loss": 2.9555, "step": 1020 }, { "epoch": 0.06, "learning_rate": 1.963343924461917e-05, "loss": 2.8127, "step": 1030 }, { "epoch": 0.06, "learning_rate": 1.962984903154002e-05, "loss": 2.8066, "step": 1040 }, { "epoch": 0.06, "learning_rate": 1.9626258818460876e-05, "loss": 2.8596, "step": 1050 }, { "epoch": 0.06, "learning_rate": 1.9622668605381732e-05, "loss": 2.8612, "step": 1060 }, { "epoch": 0.06, "learning_rate": 1.9619078392302584e-05, "loss": 2.8639, "step": 1070 }, { "epoch": 0.06, "learning_rate": 1.961548817922344e-05, "loss": 2.8358, "step": 1080 }, { "epoch": 0.06, "learning_rate": 1.9611897966144292e-05, "loss": 2.951, "step": 1090 }, { "epoch": 0.06, "learning_rate": 1.9608307753065144e-05, "loss": 2.961, "step": 1100 }, { "epoch": 0.06, "learning_rate": 1.9604717539986e-05, "loss": 2.7049, "step": 1110 }, { "epoch": 0.06, "learning_rate": 1.9601127326906852e-05, "loss": 2.9308, "step": 1120 }, { "epoch": 0.06, "learning_rate": 1.9597537113827707e-05, "loss": 2.9189, "step": 1130 }, { "epoch": 0.06, "learning_rate": 1.9593946900748563e-05, "loss": 2.8701, "step": 1140 }, { "epoch": 0.06, "learning_rate": 1.9590356687669415e-05, "loss": 3.0191, "step": 1150 }, { "epoch": 0.06, "learning_rate": 1.9586766474590267e-05, "loss": 2.851, "step": 1160 }, { "epoch": 0.06, "learning_rate": 1.9583176261511123e-05, "loss": 2.9326, "step": 1170 }, { "epoch": 0.06, "learning_rate": 1.9579586048431975e-05, "loss": 2.916, "step": 1180 }, { "epoch": 0.06, "learning_rate": 1.957599583535283e-05, "loss": 2.8135, "step": 1190 }, { "epoch": 0.06, "learning_rate": 1.9572405622273683e-05, "loss": 2.8072, "step": 1200 }, { "epoch": 0.07, "learning_rate": 1.9568815409194538e-05, "loss": 2.7695, "step": 1210 }, { "epoch": 0.07, "learning_rate": 1.956522519611539e-05, "loss": 2.7315, "step": 1220 }, { "epoch": 0.07, "learning_rate": 1.9561634983036242e-05, "loss": 2.7673, "step": 1230 }, { "epoch": 0.07, "learning_rate": 1.9558044769957098e-05, "loss": 2.8883, "step": 1240 }, { "epoch": 0.07, "learning_rate": 1.9554454556877954e-05, "loss": 2.6834, "step": 1250 }, { "epoch": 0.07, "learning_rate": 1.9550864343798806e-05, "loss": 2.8404, "step": 1260 }, { "epoch": 0.07, "learning_rate": 1.954727413071966e-05, "loss": 2.7066, "step": 1270 }, { "epoch": 0.07, "learning_rate": 1.9543683917640513e-05, "loss": 2.8951, "step": 1280 }, { "epoch": 0.07, "learning_rate": 1.9540093704561366e-05, "loss": 2.7959, "step": 1290 }, { "epoch": 0.07, "learning_rate": 1.953650349148222e-05, "loss": 2.7034, "step": 1300 }, { "epoch": 0.07, "learning_rate": 1.9532913278403073e-05, "loss": 2.838, "step": 1310 }, { "epoch": 0.07, "learning_rate": 1.952932306532393e-05, "loss": 2.8125, "step": 1320 }, { "epoch": 0.07, "learning_rate": 1.9525732852244784e-05, "loss": 2.7806, "step": 1330 }, { "epoch": 0.07, "learning_rate": 1.9522142639165636e-05, "loss": 2.8089, "step": 1340 }, { "epoch": 0.07, "learning_rate": 1.951855242608649e-05, "loss": 2.8305, "step": 1350 }, { "epoch": 0.07, "learning_rate": 1.9514962213007344e-05, "loss": 2.7721, "step": 1360 }, { "epoch": 0.07, "learning_rate": 1.9511371999928196e-05, "loss": 2.8546, "step": 1370 }, { "epoch": 0.07, "learning_rate": 1.9507781786849052e-05, "loss": 2.6916, "step": 1380 }, { "epoch": 0.07, "learning_rate": 1.9504191573769904e-05, "loss": 2.8091, "step": 1390 }, { "epoch": 0.08, "learning_rate": 1.950060136069076e-05, "loss": 2.7681, "step": 1400 }, { "epoch": 0.08, "learning_rate": 1.9497011147611612e-05, "loss": 2.8239, "step": 1410 }, { "epoch": 0.08, "learning_rate": 1.9493420934532464e-05, "loss": 2.8446, "step": 1420 }, { "epoch": 0.08, "learning_rate": 1.948983072145332e-05, "loss": 2.8963, "step": 1430 }, { "epoch": 0.08, "learning_rate": 1.9486240508374175e-05, "loss": 2.8163, "step": 1440 }, { "epoch": 0.08, "learning_rate": 1.9482650295295027e-05, "loss": 2.7866, "step": 1450 }, { "epoch": 0.08, "learning_rate": 1.9479060082215883e-05, "loss": 2.78, "step": 1460 }, { "epoch": 0.08, "learning_rate": 1.9475469869136735e-05, "loss": 2.7873, "step": 1470 }, { "epoch": 0.08, "learning_rate": 1.9471879656057587e-05, "loss": 2.7717, "step": 1480 }, { "epoch": 0.08, "learning_rate": 1.9468289442978443e-05, "loss": 3.0517, "step": 1490 }, { "epoch": 0.08, "learning_rate": 1.9464699229899295e-05, "loss": 2.8272, "step": 1500 }, { "epoch": 0.08, "learning_rate": 1.946110901682015e-05, "loss": 2.8358, "step": 1510 }, { "epoch": 0.08, "learning_rate": 1.9457518803741006e-05, "loss": 2.6929, "step": 1520 }, { "epoch": 0.08, "learning_rate": 1.9453928590661858e-05, "loss": 2.7598, "step": 1530 }, { "epoch": 0.08, "learning_rate": 1.945033837758271e-05, "loss": 2.7529, "step": 1540 }, { "epoch": 0.08, "learning_rate": 1.9446748164503562e-05, "loss": 2.6802, "step": 1550 }, { "epoch": 0.08, "learning_rate": 1.9443157951424418e-05, "loss": 2.8318, "step": 1560 }, { "epoch": 0.08, "learning_rate": 1.9439567738345273e-05, "loss": 2.8196, "step": 1570 }, { "epoch": 0.09, "learning_rate": 1.9435977525266125e-05, "loss": 2.7937, "step": 1580 }, { "epoch": 0.09, "learning_rate": 1.943238731218698e-05, "loss": 2.7443, "step": 1590 }, { "epoch": 0.09, "learning_rate": 1.9428797099107833e-05, "loss": 2.7506, "step": 1600 }, { "epoch": 0.09, "learning_rate": 1.9425206886028685e-05, "loss": 2.8196, "step": 1610 }, { "epoch": 0.09, "learning_rate": 1.942161667294954e-05, "loss": 2.8199, "step": 1620 }, { "epoch": 0.09, "learning_rate": 1.9418026459870393e-05, "loss": 2.7978, "step": 1630 }, { "epoch": 0.09, "learning_rate": 1.941443624679125e-05, "loss": 2.835, "step": 1640 }, { "epoch": 0.09, "learning_rate": 1.9410846033712104e-05, "loss": 2.6939, "step": 1650 }, { "epoch": 0.09, "learning_rate": 1.9407255820632956e-05, "loss": 2.6551, "step": 1660 }, { "epoch": 0.09, "learning_rate": 1.940366560755381e-05, "loss": 2.9222, "step": 1670 }, { "epoch": 0.09, "learning_rate": 1.9400075394474664e-05, "loss": 2.5975, "step": 1680 }, { "epoch": 0.09, "learning_rate": 1.9396485181395516e-05, "loss": 2.7544, "step": 1690 }, { "epoch": 0.09, "learning_rate": 1.9392894968316372e-05, "loss": 2.6074, "step": 1700 }, { "epoch": 0.09, "learning_rate": 1.9389304755237227e-05, "loss": 2.6687, "step": 1710 }, { "epoch": 0.09, "learning_rate": 1.938571454215808e-05, "loss": 2.7668, "step": 1720 }, { "epoch": 0.09, "learning_rate": 1.938212432907893e-05, "loss": 2.8017, "step": 1730 }, { "epoch": 0.09, "learning_rate": 1.9378534115999784e-05, "loss": 2.8552, "step": 1740 }, { "epoch": 0.09, "learning_rate": 1.937494390292064e-05, "loss": 2.6845, "step": 1750 }, { "epoch": 0.09, "learning_rate": 1.9371353689841495e-05, "loss": 2.7429, "step": 1760 }, { "epoch": 0.1, "learning_rate": 1.9367763476762347e-05, "loss": 2.8029, "step": 1770 }, { "epoch": 0.1, "learning_rate": 1.9364173263683203e-05, "loss": 2.7683, "step": 1780 }, { "epoch": 0.1, "learning_rate": 1.9360583050604055e-05, "loss": 2.8638, "step": 1790 }, { "epoch": 0.1, "learning_rate": 1.9356992837524907e-05, "loss": 2.75, "step": 1800 }, { "epoch": 0.1, "learning_rate": 1.9353402624445762e-05, "loss": 2.6314, "step": 1810 }, { "epoch": 0.1, "learning_rate": 1.9349812411366615e-05, "loss": 2.766, "step": 1820 }, { "epoch": 0.1, "learning_rate": 1.934622219828747e-05, "loss": 2.6745, "step": 1830 }, { "epoch": 0.1, "learning_rate": 1.9342631985208326e-05, "loss": 2.7482, "step": 1840 }, { "epoch": 0.1, "learning_rate": 1.9339041772129178e-05, "loss": 2.723, "step": 1850 }, { "epoch": 0.1, "learning_rate": 1.933545155905003e-05, "loss": 2.7507, "step": 1860 }, { "epoch": 0.1, "learning_rate": 1.9331861345970885e-05, "loss": 2.7127, "step": 1870 }, { "epoch": 0.1, "learning_rate": 1.9328271132891738e-05, "loss": 2.6458, "step": 1880 }, { "epoch": 0.1, "learning_rate": 1.9324680919812593e-05, "loss": 2.7481, "step": 1890 }, { "epoch": 0.1, "learning_rate": 1.9321090706733445e-05, "loss": 2.6548, "step": 1900 }, { "epoch": 0.1, "learning_rate": 1.93175004936543e-05, "loss": 2.6856, "step": 1910 }, { "epoch": 0.1, "learning_rate": 1.9313910280575153e-05, "loss": 2.7039, "step": 1920 }, { "epoch": 0.1, "learning_rate": 1.9310320067496005e-05, "loss": 2.7392, "step": 1930 }, { "epoch": 0.1, "learning_rate": 1.930672985441686e-05, "loss": 2.6267, "step": 1940 }, { "epoch": 0.11, "learning_rate": 1.9303139641337716e-05, "loss": 2.7609, "step": 1950 }, { "epoch": 0.11, "learning_rate": 1.929954942825857e-05, "loss": 2.7663, "step": 1960 }, { "epoch": 0.11, "learning_rate": 1.9295959215179424e-05, "loss": 2.7452, "step": 1970 }, { "epoch": 0.11, "learning_rate": 1.9292369002100276e-05, "loss": 2.791, "step": 1980 }, { "epoch": 0.11, "learning_rate": 1.9288778789021128e-05, "loss": 2.8045, "step": 1990 }, { "epoch": 0.11, "learning_rate": 1.9285188575941984e-05, "loss": 2.6488, "step": 2000 }, { "epoch": 0.11, "learning_rate": 1.9281598362862836e-05, "loss": 2.6689, "step": 2010 }, { "epoch": 0.11, "learning_rate": 1.927800814978369e-05, "loss": 2.5595, "step": 2020 }, { "epoch": 0.11, "learning_rate": 1.9274417936704547e-05, "loss": 2.6463, "step": 2030 }, { "epoch": 0.11, "learning_rate": 1.92708277236254e-05, "loss": 2.7077, "step": 2040 }, { "epoch": 0.11, "learning_rate": 1.926723751054625e-05, "loss": 2.726, "step": 2050 }, { "epoch": 0.11, "learning_rate": 1.9263647297467104e-05, "loss": 2.6916, "step": 2060 }, { "epoch": 0.11, "learning_rate": 1.926005708438796e-05, "loss": 2.7012, "step": 2070 }, { "epoch": 0.11, "learning_rate": 1.9256466871308815e-05, "loss": 2.601, "step": 2080 }, { "epoch": 0.11, "learning_rate": 1.9252876658229667e-05, "loss": 2.751, "step": 2090 }, { "epoch": 0.11, "learning_rate": 1.9249286445150522e-05, "loss": 2.5344, "step": 2100 }, { "epoch": 0.11, "learning_rate": 1.9245696232071375e-05, "loss": 2.589, "step": 2110 }, { "epoch": 0.11, "learning_rate": 1.9242106018992227e-05, "loss": 2.6888, "step": 2120 }, { "epoch": 0.11, "learning_rate": 1.9238515805913082e-05, "loss": 2.6296, "step": 2130 }, { "epoch": 0.12, "learning_rate": 1.9234925592833938e-05, "loss": 2.5918, "step": 2140 }, { "epoch": 0.12, "learning_rate": 1.923133537975479e-05, "loss": 2.5673, "step": 2150 }, { "epoch": 0.12, "learning_rate": 1.9227745166675645e-05, "loss": 2.5967, "step": 2160 }, { "epoch": 0.12, "learning_rate": 1.9224154953596498e-05, "loss": 2.6612, "step": 2170 }, { "epoch": 0.12, "learning_rate": 1.922056474051735e-05, "loss": 2.7359, "step": 2180 }, { "epoch": 0.12, "learning_rate": 1.9216974527438205e-05, "loss": 2.593, "step": 2190 }, { "epoch": 0.12, "learning_rate": 1.9213384314359057e-05, "loss": 2.63, "step": 2200 }, { "epoch": 0.12, "learning_rate": 1.9209794101279913e-05, "loss": 2.6424, "step": 2210 }, { "epoch": 0.12, "learning_rate": 1.920620388820077e-05, "loss": 2.7708, "step": 2220 }, { "epoch": 0.12, "learning_rate": 1.920261367512162e-05, "loss": 2.6115, "step": 2230 }, { "epoch": 0.12, "learning_rate": 1.9199023462042473e-05, "loss": 2.7522, "step": 2240 }, { "epoch": 0.12, "learning_rate": 1.9195433248963325e-05, "loss": 2.7271, "step": 2250 }, { "epoch": 0.12, "learning_rate": 1.919184303588418e-05, "loss": 2.5119, "step": 2260 }, { "epoch": 0.12, "learning_rate": 1.9188252822805036e-05, "loss": 2.7152, "step": 2270 }, { "epoch": 0.12, "learning_rate": 1.9184662609725888e-05, "loss": 2.6712, "step": 2280 }, { "epoch": 0.12, "learning_rate": 1.9181072396646744e-05, "loss": 2.6579, "step": 2290 }, { "epoch": 0.12, "learning_rate": 1.9177482183567596e-05, "loss": 2.556, "step": 2300 }, { "epoch": 0.12, "learning_rate": 1.9173891970488448e-05, "loss": 2.755, "step": 2310 }, { "epoch": 0.12, "learning_rate": 1.9170301757409304e-05, "loss": 2.5799, "step": 2320 }, { "epoch": 0.13, "learning_rate": 1.9166711544330156e-05, "loss": 2.8061, "step": 2330 }, { "epoch": 0.13, "learning_rate": 1.916312133125101e-05, "loss": 2.5838, "step": 2340 }, { "epoch": 0.13, "learning_rate": 1.9159531118171867e-05, "loss": 2.5489, "step": 2350 }, { "epoch": 0.13, "learning_rate": 1.915594090509272e-05, "loss": 2.6664, "step": 2360 }, { "epoch": 0.13, "learning_rate": 1.915235069201357e-05, "loss": 2.638, "step": 2370 }, { "epoch": 0.13, "learning_rate": 1.9148760478934427e-05, "loss": 2.6145, "step": 2380 }, { "epoch": 0.13, "learning_rate": 1.914517026585528e-05, "loss": 2.6475, "step": 2390 }, { "epoch": 0.13, "learning_rate": 1.9141580052776134e-05, "loss": 2.731, "step": 2400 }, { "epoch": 0.13, "learning_rate": 1.9137989839696987e-05, "loss": 2.624, "step": 2410 }, { "epoch": 0.13, "learning_rate": 1.9134399626617842e-05, "loss": 2.6253, "step": 2420 }, { "epoch": 0.13, "learning_rate": 1.9130809413538694e-05, "loss": 2.6922, "step": 2430 }, { "epoch": 0.13, "learning_rate": 1.9127219200459546e-05, "loss": 2.7505, "step": 2440 }, { "epoch": 0.13, "learning_rate": 1.9123628987380402e-05, "loss": 2.7062, "step": 2450 }, { "epoch": 0.13, "learning_rate": 1.9120038774301258e-05, "loss": 2.6434, "step": 2460 }, { "epoch": 0.13, "learning_rate": 1.911644856122211e-05, "loss": 2.7078, "step": 2470 }, { "epoch": 0.13, "learning_rate": 1.9112858348142965e-05, "loss": 2.66, "step": 2480 }, { "epoch": 0.13, "learning_rate": 1.9109268135063817e-05, "loss": 2.7149, "step": 2490 }, { "epoch": 0.13, "learning_rate": 1.910567792198467e-05, "loss": 2.6047, "step": 2500 }, { "epoch": 0.14, "learning_rate": 1.9102087708905525e-05, "loss": 2.6814, "step": 2510 }, { "epoch": 0.14, "learning_rate": 1.9098497495826377e-05, "loss": 2.5845, "step": 2520 }, { "epoch": 0.14, "learning_rate": 1.9094907282747233e-05, "loss": 2.5178, "step": 2530 }, { "epoch": 0.14, "learning_rate": 1.909131706966809e-05, "loss": 2.5956, "step": 2540 }, { "epoch": 0.14, "learning_rate": 1.908772685658894e-05, "loss": 2.5979, "step": 2550 }, { "epoch": 0.14, "learning_rate": 1.9084136643509793e-05, "loss": 2.7305, "step": 2560 }, { "epoch": 0.14, "learning_rate": 1.9080546430430648e-05, "loss": 2.7569, "step": 2570 }, { "epoch": 0.14, "learning_rate": 1.90769562173515e-05, "loss": 2.5463, "step": 2580 }, { "epoch": 0.14, "learning_rate": 1.9073366004272356e-05, "loss": 2.6905, "step": 2590 }, { "epoch": 0.14, "learning_rate": 1.9069775791193208e-05, "loss": 2.6339, "step": 2600 }, { "epoch": 0.14, "learning_rate": 1.9066185578114064e-05, "loss": 2.6937, "step": 2610 }, { "epoch": 0.14, "learning_rate": 1.9062595365034916e-05, "loss": 2.6612, "step": 2620 }, { "epoch": 0.14, "learning_rate": 1.9059005151955768e-05, "loss": 2.6495, "step": 2630 }, { "epoch": 0.14, "learning_rate": 1.9055414938876624e-05, "loss": 2.6662, "step": 2640 }, { "epoch": 0.14, "learning_rate": 1.905182472579748e-05, "loss": 2.6109, "step": 2650 }, { "epoch": 0.14, "learning_rate": 1.904823451271833e-05, "loss": 2.6967, "step": 2660 }, { "epoch": 0.14, "learning_rate": 1.9044644299639187e-05, "loss": 2.6573, "step": 2670 }, { "epoch": 0.14, "learning_rate": 1.904105408656004e-05, "loss": 2.6437, "step": 2680 }, { "epoch": 0.14, "learning_rate": 1.903746387348089e-05, "loss": 2.6396, "step": 2690 }, { "epoch": 0.15, "learning_rate": 1.9033873660401747e-05, "loss": 2.5431, "step": 2700 }, { "epoch": 0.15, "learning_rate": 1.90302834473226e-05, "loss": 2.6333, "step": 2710 }, { "epoch": 0.15, "learning_rate": 1.9026693234243454e-05, "loss": 2.562, "step": 2720 }, { "epoch": 0.15, "learning_rate": 1.902310302116431e-05, "loss": 2.6076, "step": 2730 }, { "epoch": 0.15, "learning_rate": 1.9019512808085162e-05, "loss": 2.7656, "step": 2740 }, { "epoch": 0.15, "learning_rate": 1.9015922595006014e-05, "loss": 2.6859, "step": 2750 }, { "epoch": 0.15, "learning_rate": 1.9012332381926866e-05, "loss": 2.5723, "step": 2760 }, { "epoch": 0.15, "learning_rate": 1.9008742168847722e-05, "loss": 2.5823, "step": 2770 }, { "epoch": 0.15, "learning_rate": 1.9005151955768577e-05, "loss": 2.6527, "step": 2780 }, { "epoch": 0.15, "learning_rate": 1.900156174268943e-05, "loss": 2.6323, "step": 2790 }, { "epoch": 0.15, "learning_rate": 1.8997971529610285e-05, "loss": 2.5642, "step": 2800 }, { "epoch": 0.15, "learning_rate": 1.8994381316531137e-05, "loss": 2.706, "step": 2810 }, { "epoch": 0.15, "learning_rate": 1.899079110345199e-05, "loss": 2.4924, "step": 2820 }, { "epoch": 0.15, "learning_rate": 1.8987200890372845e-05, "loss": 2.6242, "step": 2830 }, { "epoch": 0.15, "learning_rate": 1.8983610677293697e-05, "loss": 2.503, "step": 2840 }, { "epoch": 0.15, "learning_rate": 1.8980020464214553e-05, "loss": 2.6463, "step": 2850 }, { "epoch": 0.15, "learning_rate": 1.8976430251135408e-05, "loss": 2.6109, "step": 2860 }, { "epoch": 0.15, "learning_rate": 1.897284003805626e-05, "loss": 2.6042, "step": 2870 }, { "epoch": 0.16, "learning_rate": 1.8969249824977113e-05, "loss": 2.6578, "step": 2880 }, { "epoch": 0.16, "learning_rate": 1.8965659611897968e-05, "loss": 2.6372, "step": 2890 }, { "epoch": 0.16, "learning_rate": 1.896206939881882e-05, "loss": 2.5956, "step": 2900 }, { "epoch": 0.16, "learning_rate": 1.8958479185739676e-05, "loss": 2.5386, "step": 2910 }, { "epoch": 0.16, "learning_rate": 1.8954888972660528e-05, "loss": 2.6261, "step": 2920 }, { "epoch": 0.16, "learning_rate": 1.8951298759581383e-05, "loss": 2.5414, "step": 2930 }, { "epoch": 0.16, "learning_rate": 1.8947708546502236e-05, "loss": 2.6658, "step": 2940 }, { "epoch": 0.16, "learning_rate": 1.8944118333423088e-05, "loss": 2.5249, "step": 2950 }, { "epoch": 0.16, "learning_rate": 1.8940528120343943e-05, "loss": 2.5465, "step": 2960 }, { "epoch": 0.16, "learning_rate": 1.89369379072648e-05, "loss": 2.5472, "step": 2970 }, { "epoch": 0.16, "learning_rate": 1.893334769418565e-05, "loss": 2.659, "step": 2980 }, { "epoch": 0.16, "learning_rate": 1.8929757481106507e-05, "loss": 2.6079, "step": 2990 }, { "epoch": 0.16, "learning_rate": 1.892616726802736e-05, "loss": 2.508, "step": 3000 }, { "epoch": 0.16, "learning_rate": 1.892257705494821e-05, "loss": 2.496, "step": 3010 }, { "epoch": 0.16, "learning_rate": 1.8918986841869066e-05, "loss": 2.6098, "step": 3020 }, { "epoch": 0.16, "learning_rate": 1.891539662878992e-05, "loss": 2.5876, "step": 3030 }, { "epoch": 0.16, "learning_rate": 1.8911806415710774e-05, "loss": 2.6408, "step": 3040 }, { "epoch": 0.16, "learning_rate": 1.890821620263163e-05, "loss": 2.5723, "step": 3050 }, { "epoch": 0.16, "learning_rate": 1.8904625989552482e-05, "loss": 2.5794, "step": 3060 }, { "epoch": 0.17, "learning_rate": 1.8901035776473334e-05, "loss": 2.6547, "step": 3070 }, { "epoch": 0.17, "learning_rate": 1.889744556339419e-05, "loss": 2.729, "step": 3080 }, { "epoch": 0.17, "learning_rate": 1.8893855350315042e-05, "loss": 2.6229, "step": 3090 }, { "epoch": 0.17, "learning_rate": 1.8890265137235897e-05, "loss": 2.5611, "step": 3100 }, { "epoch": 0.17, "learning_rate": 1.888667492415675e-05, "loss": 2.6038, "step": 3110 }, { "epoch": 0.17, "learning_rate": 1.8883084711077605e-05, "loss": 2.5261, "step": 3120 }, { "epoch": 0.17, "learning_rate": 1.8879494497998457e-05, "loss": 2.5488, "step": 3130 }, { "epoch": 0.17, "learning_rate": 1.887590428491931e-05, "loss": 2.4828, "step": 3140 }, { "epoch": 0.17, "learning_rate": 1.8872314071840165e-05, "loss": 2.6341, "step": 3150 }, { "epoch": 0.17, "learning_rate": 1.886872385876102e-05, "loss": 2.5842, "step": 3160 }, { "epoch": 0.17, "learning_rate": 1.8865133645681873e-05, "loss": 2.4895, "step": 3170 }, { "epoch": 0.17, "learning_rate": 1.8861543432602728e-05, "loss": 2.6954, "step": 3180 }, { "epoch": 0.17, "learning_rate": 1.885795321952358e-05, "loss": 2.3721, "step": 3190 }, { "epoch": 0.17, "learning_rate": 1.8854363006444432e-05, "loss": 2.5129, "step": 3200 }, { "epoch": 0.17, "learning_rate": 1.8850772793365288e-05, "loss": 2.5214, "step": 3210 }, { "epoch": 0.17, "learning_rate": 1.884718258028614e-05, "loss": 2.4711, "step": 3220 }, { "epoch": 0.17, "learning_rate": 1.8843592367206996e-05, "loss": 2.6295, "step": 3230 }, { "epoch": 0.17, "learning_rate": 1.884000215412785e-05, "loss": 2.7043, "step": 3240 }, { "epoch": 0.18, "learning_rate": 1.8836411941048703e-05, "loss": 2.5009, "step": 3250 }, { "epoch": 0.18, "learning_rate": 1.8832821727969555e-05, "loss": 2.5467, "step": 3260 }, { "epoch": 0.18, "learning_rate": 1.8829231514890408e-05, "loss": 2.5931, "step": 3270 }, { "epoch": 0.18, "learning_rate": 1.8825641301811263e-05, "loss": 2.4969, "step": 3280 }, { "epoch": 0.18, "learning_rate": 1.882205108873212e-05, "loss": 2.4967, "step": 3290 }, { "epoch": 0.18, "learning_rate": 1.881846087565297e-05, "loss": 2.6285, "step": 3300 }, { "epoch": 0.18, "learning_rate": 1.8814870662573826e-05, "loss": 2.5414, "step": 3310 }, { "epoch": 0.18, "learning_rate": 1.881128044949468e-05, "loss": 2.6132, "step": 3320 }, { "epoch": 0.18, "learning_rate": 1.880769023641553e-05, "loss": 2.5526, "step": 3330 }, { "epoch": 0.18, "learning_rate": 1.8804100023336386e-05, "loss": 2.424, "step": 3340 }, { "epoch": 0.18, "learning_rate": 1.880050981025724e-05, "loss": 2.4913, "step": 3350 }, { "epoch": 0.18, "learning_rate": 1.8796919597178094e-05, "loss": 2.5996, "step": 3360 }, { "epoch": 0.18, "learning_rate": 1.879332938409895e-05, "loss": 2.5916, "step": 3370 }, { "epoch": 0.18, "learning_rate": 1.87897391710198e-05, "loss": 2.6065, "step": 3380 }, { "epoch": 0.18, "learning_rate": 1.8786148957940654e-05, "loss": 2.5286, "step": 3390 }, { "epoch": 0.18, "learning_rate": 1.878255874486151e-05, "loss": 2.3606, "step": 3400 }, { "epoch": 0.18, "learning_rate": 1.877896853178236e-05, "loss": 2.5906, "step": 3410 }, { "epoch": 0.18, "learning_rate": 1.8775378318703217e-05, "loss": 2.5598, "step": 3420 }, { "epoch": 0.18, "learning_rate": 1.8771788105624073e-05, "loss": 2.567, "step": 3430 }, { "epoch": 0.19, "learning_rate": 1.8768197892544925e-05, "loss": 2.4522, "step": 3440 }, { "epoch": 0.19, "learning_rate": 1.8764607679465777e-05, "loss": 2.5184, "step": 3450 }, { "epoch": 0.19, "learning_rate": 1.876101746638663e-05, "loss": 2.5049, "step": 3460 }, { "epoch": 0.19, "learning_rate": 1.8757427253307485e-05, "loss": 2.5587, "step": 3470 }, { "epoch": 0.19, "learning_rate": 1.875383704022834e-05, "loss": 2.5336, "step": 3480 }, { "epoch": 0.19, "learning_rate": 1.8750246827149192e-05, "loss": 2.5556, "step": 3490 }, { "epoch": 0.19, "learning_rate": 1.8746656614070048e-05, "loss": 2.5977, "step": 3500 }, { "epoch": 0.19, "learning_rate": 1.87430664009909e-05, "loss": 2.635, "step": 3510 }, { "epoch": 0.19, "learning_rate": 1.8739476187911752e-05, "loss": 2.4791, "step": 3520 }, { "epoch": 0.19, "learning_rate": 1.8735885974832608e-05, "loss": 2.4704, "step": 3530 }, { "epoch": 0.19, "learning_rate": 1.873229576175346e-05, "loss": 2.5109, "step": 3540 }, { "epoch": 0.19, "learning_rate": 1.8728705548674315e-05, "loss": 2.5131, "step": 3550 }, { "epoch": 0.19, "learning_rate": 1.872511533559517e-05, "loss": 2.6331, "step": 3560 }, { "epoch": 0.19, "learning_rate": 1.8721525122516023e-05, "loss": 2.4951, "step": 3570 }, { "epoch": 0.19, "learning_rate": 1.8717934909436875e-05, "loss": 2.3377, "step": 3580 }, { "epoch": 0.19, "learning_rate": 1.871434469635773e-05, "loss": 2.5551, "step": 3590 }, { "epoch": 0.19, "learning_rate": 1.8710754483278583e-05, "loss": 2.5181, "step": 3600 }, { "epoch": 0.19, "learning_rate": 1.870716427019944e-05, "loss": 2.3867, "step": 3610 }, { "epoch": 0.19, "learning_rate": 1.870357405712029e-05, "loss": 2.4953, "step": 3620 }, { "epoch": 0.2, "learning_rate": 1.8699983844041146e-05, "loss": 2.4266, "step": 3630 }, { "epoch": 0.2, "learning_rate": 1.8696393630962e-05, "loss": 2.5761, "step": 3640 }, { "epoch": 0.2, "learning_rate": 1.869280341788285e-05, "loss": 2.5007, "step": 3650 }, { "epoch": 0.2, "learning_rate": 1.8689213204803706e-05, "loss": 2.641, "step": 3660 }, { "epoch": 0.2, "learning_rate": 1.868562299172456e-05, "loss": 2.509, "step": 3670 }, { "epoch": 0.2, "learning_rate": 1.8682032778645414e-05, "loss": 2.4467, "step": 3680 }, { "epoch": 0.2, "learning_rate": 1.867844256556627e-05, "loss": 2.5679, "step": 3690 }, { "epoch": 0.2, "learning_rate": 1.867485235248712e-05, "loss": 2.4515, "step": 3700 }, { "epoch": 0.2, "learning_rate": 1.8671262139407974e-05, "loss": 2.5014, "step": 3710 }, { "epoch": 0.2, "learning_rate": 1.866767192632883e-05, "loss": 2.5634, "step": 3720 }, { "epoch": 0.2, "learning_rate": 1.866408171324968e-05, "loss": 2.5036, "step": 3730 }, { "epoch": 0.2, "learning_rate": 1.8660491500170537e-05, "loss": 2.4928, "step": 3740 }, { "epoch": 0.2, "learning_rate": 1.8656901287091392e-05, "loss": 2.4937, "step": 3750 }, { "epoch": 0.2, "learning_rate": 1.8653311074012245e-05, "loss": 2.5571, "step": 3760 }, { "epoch": 0.2, "learning_rate": 1.8649720860933097e-05, "loss": 2.5574, "step": 3770 }, { "epoch": 0.2, "learning_rate": 1.864613064785395e-05, "loss": 2.4821, "step": 3780 }, { "epoch": 0.2, "learning_rate": 1.8642540434774804e-05, "loss": 2.4531, "step": 3790 }, { "epoch": 0.2, "learning_rate": 1.863895022169566e-05, "loss": 2.5505, "step": 3800 }, { "epoch": 0.21, "learning_rate": 1.8635360008616512e-05, "loss": 2.4979, "step": 3810 }, { "epoch": 0.21, "learning_rate": 1.8631769795537368e-05, "loss": 2.5249, "step": 3820 }, { "epoch": 0.21, "learning_rate": 1.862817958245822e-05, "loss": 2.4761, "step": 3830 }, { "epoch": 0.21, "learning_rate": 1.8624589369379072e-05, "loss": 2.4575, "step": 3840 }, { "epoch": 0.21, "learning_rate": 1.8620999156299928e-05, "loss": 2.5435, "step": 3850 }, { "epoch": 0.21, "learning_rate": 1.8617408943220783e-05, "loss": 2.5849, "step": 3860 }, { "epoch": 0.21, "learning_rate": 1.8613818730141635e-05, "loss": 2.5028, "step": 3870 }, { "epoch": 0.21, "learning_rate": 1.861022851706249e-05, "loss": 2.4835, "step": 3880 }, { "epoch": 0.21, "learning_rate": 1.8606638303983343e-05, "loss": 2.6491, "step": 3890 }, { "epoch": 0.21, "learning_rate": 1.8603048090904195e-05, "loss": 2.4101, "step": 3900 }, { "epoch": 0.21, "learning_rate": 1.859945787782505e-05, "loss": 2.5407, "step": 3910 }, { "epoch": 0.21, "learning_rate": 1.859622668605382e-05, "loss": 6.4295, "step": 3920 }, { "epoch": 0.21, "learning_rate": 1.8592636472974674e-05, "loss": 11.3457, "step": 3930 }, { "epoch": 0.21, "learning_rate": 1.8589046259895526e-05, "loss": 15.0764, "step": 3940 }, { "epoch": 0.21, "learning_rate": 1.8585456046816378e-05, "loss": 14.1912, "step": 3950 }, { "epoch": 0.21, "learning_rate": 1.8581865833737234e-05, "loss": 12.5133, "step": 3960 }, { "epoch": 0.21, "learning_rate": 1.857827562065809e-05, "loss": 11.572, "step": 3970 }, { "epoch": 0.21, "learning_rate": 1.857468540757894e-05, "loss": 11.2024, "step": 3980 }, { "epoch": 0.21, "learning_rate": 1.8571095194499794e-05, "loss": 10.5553, "step": 3990 }, { "epoch": 0.22, "learning_rate": 1.856750498142065e-05, "loss": 9.9466, "step": 4000 }, { "epoch": 0.22, "learning_rate": 1.85639147683415e-05, "loss": 5.7554, "step": 4010 }, { "epoch": 0.22, "learning_rate": 1.8560324555262357e-05, "loss": 2.8853, "step": 4020 }, { "epoch": 0.22, "learning_rate": 1.855673434218321e-05, "loss": 2.7084, "step": 4030 }, { "epoch": 0.22, "learning_rate": 1.8553144129104065e-05, "loss": 3.5929, "step": 4040 }, { "epoch": 0.22, "learning_rate": 1.8549553916024917e-05, "loss": 5.2618, "step": 4050 }, { "epoch": 0.22, "learning_rate": 1.8545963702945772e-05, "loss": 4.1107, "step": 4060 }, { "epoch": 0.22, "learning_rate": 1.8542373489866624e-05, "loss": 3.6029, "step": 4070 }, { "epoch": 0.22, "learning_rate": 1.853878327678748e-05, "loss": 3.5567, "step": 4080 }, { "epoch": 0.22, "learning_rate": 1.8535193063708332e-05, "loss": 3.7352, "step": 4090 }, { "epoch": 0.22, "learning_rate": 1.8531602850629188e-05, "loss": 3.6084, "step": 4100 }, { "epoch": 0.22, "learning_rate": 1.852801263755004e-05, "loss": 3.0987, "step": 4110 }, { "epoch": 0.22, "learning_rate": 1.8524422424470895e-05, "loss": 3.2507, "step": 4120 }, { "epoch": 0.22, "learning_rate": 1.8520832211391747e-05, "loss": 3.3224, "step": 4130 }, { "epoch": 0.22, "learning_rate": 1.85172419983126e-05, "loss": 3.1285, "step": 4140 }, { "epoch": 0.22, "learning_rate": 1.8513651785233455e-05, "loss": 2.9518, "step": 4150 }, { "epoch": 0.22, "learning_rate": 1.851006157215431e-05, "loss": 3.2708, "step": 4160 }, { "epoch": 0.22, "learning_rate": 1.8506471359075163e-05, "loss": 3.1881, "step": 4170 }, { "epoch": 0.23, "learning_rate": 1.8502881145996015e-05, "loss": 2.9374, "step": 4180 }, { "epoch": 0.23, "learning_rate": 1.849929093291687e-05, "loss": 2.8588, "step": 4190 }, { "epoch": 0.23, "learning_rate": 1.8495700719837723e-05, "loss": 2.8595, "step": 4200 }, { "epoch": 0.23, "learning_rate": 1.8492110506758578e-05, "loss": 2.8636, "step": 4210 }, { "epoch": 0.23, "learning_rate": 1.848852029367943e-05, "loss": 2.839, "step": 4220 }, { "epoch": 0.23, "learning_rate": 1.8484930080600286e-05, "loss": 2.8617, "step": 4230 }, { "epoch": 0.23, "learning_rate": 1.8481339867521138e-05, "loss": 2.8889, "step": 4240 }, { "epoch": 0.23, "learning_rate": 1.8477749654441994e-05, "loss": 2.8111, "step": 4250 }, { "epoch": 0.23, "learning_rate": 1.8474159441362846e-05, "loss": 2.805, "step": 4260 }, { "epoch": 0.23, "learning_rate": 1.84705692282837e-05, "loss": 2.7259, "step": 4270 }, { "epoch": 0.23, "learning_rate": 1.8466979015204554e-05, "loss": 2.8082, "step": 4280 }, { "epoch": 0.23, "learning_rate": 1.846338880212541e-05, "loss": 2.822, "step": 4290 }, { "epoch": 0.23, "learning_rate": 1.845979858904626e-05, "loss": 2.7717, "step": 4300 }, { "epoch": 0.23, "learning_rate": 1.8456208375967117e-05, "loss": 2.8857, "step": 4310 }, { "epoch": 0.23, "learning_rate": 1.845261816288797e-05, "loss": 2.8493, "step": 4320 }, { "epoch": 0.23, "learning_rate": 1.844902794980882e-05, "loss": 2.9915, "step": 4330 }, { "epoch": 0.23, "learning_rate": 1.8445437736729677e-05, "loss": 3.125, "step": 4340 }, { "epoch": 0.23, "learning_rate": 1.8441847523650532e-05, "loss": 2.8721, "step": 4350 }, { "epoch": 0.23, "learning_rate": 1.8438257310571384e-05, "loss": 2.8248, "step": 4360 }, { "epoch": 0.24, "learning_rate": 1.8434667097492237e-05, "loss": 2.6402, "step": 4370 }, { "epoch": 0.24, "learning_rate": 1.8431076884413092e-05, "loss": 2.7797, "step": 4380 }, { "epoch": 0.24, "learning_rate": 1.8427486671333944e-05, "loss": 2.913, "step": 4390 }, { "epoch": 0.24, "learning_rate": 1.84238964582548e-05, "loss": 2.7321, "step": 4400 }, { "epoch": 0.24, "learning_rate": 1.8420306245175652e-05, "loss": 2.7276, "step": 4410 }, { "epoch": 0.24, "learning_rate": 1.8416716032096507e-05, "loss": 2.8455, "step": 4420 }, { "epoch": 0.24, "learning_rate": 1.841312581901736e-05, "loss": 2.5952, "step": 4430 }, { "epoch": 0.24, "learning_rate": 1.8409535605938215e-05, "loss": 2.6574, "step": 4440 }, { "epoch": 0.24, "learning_rate": 1.8405945392859067e-05, "loss": 2.7382, "step": 4450 }, { "epoch": 0.24, "learning_rate": 1.840235517977992e-05, "loss": 2.7432, "step": 4460 }, { "epoch": 0.24, "learning_rate": 1.8398764966700775e-05, "loss": 2.8787, "step": 4470 }, { "epoch": 0.24, "learning_rate": 1.839517475362163e-05, "loss": 2.7913, "step": 4480 }, { "epoch": 0.24, "learning_rate": 1.8391584540542483e-05, "loss": 2.6114, "step": 4490 }, { "epoch": 0.24, "learning_rate": 1.8387994327463335e-05, "loss": 2.7506, "step": 4500 }, { "epoch": 0.24, "learning_rate": 1.838440411438419e-05, "loss": 2.6786, "step": 4510 }, { "epoch": 0.24, "learning_rate": 1.8380813901305043e-05, "loss": 2.7556, "step": 4520 }, { "epoch": 0.24, "learning_rate": 1.8377223688225898e-05, "loss": 2.5142, "step": 4530 }, { "epoch": 0.24, "learning_rate": 1.837363347514675e-05, "loss": 2.5796, "step": 4540 }, { "epoch": 0.25, "learning_rate": 1.8370043262067606e-05, "loss": 2.5861, "step": 4550 }, { "epoch": 0.25, "learning_rate": 1.8366453048988458e-05, "loss": 2.4686, "step": 4560 }, { "epoch": 0.25, "learning_rate": 1.8362862835909314e-05, "loss": 2.5343, "step": 4570 }, { "epoch": 0.25, "learning_rate": 1.8359272622830166e-05, "loss": 2.4751, "step": 4580 }, { "epoch": 0.25, "learning_rate": 1.835568240975102e-05, "loss": 2.5097, "step": 4590 }, { "epoch": 0.25, "learning_rate": 1.8352092196671873e-05, "loss": 2.4695, "step": 4600 }, { "epoch": 0.25, "learning_rate": 1.834850198359273e-05, "loss": 2.3584, "step": 4610 }, { "epoch": 0.25, "learning_rate": 1.834491177051358e-05, "loss": 2.5559, "step": 4620 }, { "epoch": 0.25, "learning_rate": 1.8341321557434437e-05, "loss": 2.501, "step": 4630 }, { "epoch": 0.25, "learning_rate": 1.833773134435529e-05, "loss": 2.3888, "step": 4640 }, { "epoch": 0.25, "learning_rate": 1.833414113127614e-05, "loss": 2.4763, "step": 4650 }, { "epoch": 0.25, "learning_rate": 1.8330550918196996e-05, "loss": 2.4321, "step": 4660 }, { "epoch": 0.25, "learning_rate": 1.8326960705117852e-05, "loss": 2.5815, "step": 4670 }, { "epoch": 0.25, "learning_rate": 1.8323370492038704e-05, "loss": 2.5292, "step": 4680 }, { "epoch": 0.25, "learning_rate": 1.8319780278959556e-05, "loss": 2.4254, "step": 4690 }, { "epoch": 0.25, "learning_rate": 1.8316190065880412e-05, "loss": 2.4508, "step": 4700 }, { "epoch": 0.25, "learning_rate": 1.8312599852801264e-05, "loss": 2.5653, "step": 4710 }, { "epoch": 0.25, "learning_rate": 1.830900963972212e-05, "loss": 2.8117, "step": 4720 }, { "epoch": 0.25, "learning_rate": 1.8305419426642972e-05, "loss": 3.3514, "step": 4730 }, { "epoch": 0.26, "learning_rate": 1.8301829213563827e-05, "loss": 3.1663, "step": 4740 }, { "epoch": 0.26, "learning_rate": 1.829823900048468e-05, "loss": 2.6344, "step": 4750 }, { "epoch": 0.26, "learning_rate": 1.8294648787405535e-05, "loss": 2.5842, "step": 4760 }, { "epoch": 0.26, "learning_rate": 1.8291058574326387e-05, "loss": 2.5508, "step": 4770 }, { "epoch": 0.26, "learning_rate": 1.8287468361247243e-05, "loss": 2.5438, "step": 4780 }, { "epoch": 0.26, "learning_rate": 1.8283878148168095e-05, "loss": 2.5024, "step": 4790 }, { "epoch": 0.26, "learning_rate": 1.828028793508895e-05, "loss": 2.524, "step": 4800 }, { "epoch": 0.26, "learning_rate": 1.8276697722009803e-05, "loss": 2.5516, "step": 4810 }, { "epoch": 0.26, "learning_rate": 1.8273107508930658e-05, "loss": 2.5282, "step": 4820 }, { "epoch": 0.26, "learning_rate": 1.826951729585151e-05, "loss": 2.5368, "step": 4830 }, { "epoch": 0.26, "learning_rate": 1.8265927082772362e-05, "loss": 2.6432, "step": 4840 }, { "epoch": 0.26, "learning_rate": 1.8262336869693218e-05, "loss": 2.4558, "step": 4850 }, { "epoch": 0.26, "learning_rate": 1.8258746656614074e-05, "loss": 2.5538, "step": 4860 }, { "epoch": 0.26, "learning_rate": 1.8255156443534926e-05, "loss": 2.5044, "step": 4870 }, { "epoch": 0.26, "learning_rate": 1.8251566230455778e-05, "loss": 2.4873, "step": 4880 }, { "epoch": 0.26, "learning_rate": 1.8247976017376633e-05, "loss": 2.5056, "step": 4890 }, { "epoch": 0.26, "learning_rate": 1.8244385804297486e-05, "loss": 2.506, "step": 4900 }, { "epoch": 0.26, "learning_rate": 1.824079559121834e-05, "loss": 2.4669, "step": 4910 }, { "epoch": 0.26, "learning_rate": 1.8237205378139193e-05, "loss": 2.416, "step": 4920 }, { "epoch": 0.27, "learning_rate": 1.823361516506005e-05, "loss": 2.4406, "step": 4930 }, { "epoch": 0.27, "learning_rate": 1.82300249519809e-05, "loss": 2.5012, "step": 4940 }, { "epoch": 0.27, "learning_rate": 1.8226434738901756e-05, "loss": 2.5044, "step": 4950 }, { "epoch": 0.27, "learning_rate": 1.822284452582261e-05, "loss": 2.6877, "step": 4960 }, { "epoch": 0.27, "learning_rate": 1.821925431274346e-05, "loss": 2.4853, "step": 4970 }, { "epoch": 0.27, "learning_rate": 1.8215664099664316e-05, "loss": 2.4769, "step": 4980 }, { "epoch": 0.27, "learning_rate": 1.8212073886585172e-05, "loss": 2.6282, "step": 4990 }, { "epoch": 0.27, "learning_rate": 1.8208483673506024e-05, "loss": 2.612, "step": 5000 }, { "epoch": 0.27, "learning_rate": 1.8204893460426876e-05, "loss": 2.7836, "step": 5010 }, { "epoch": 0.27, "learning_rate": 1.8201303247347732e-05, "loss": 2.8112, "step": 5020 }, { "epoch": 0.27, "learning_rate": 1.8197713034268584e-05, "loss": 2.3967, "step": 5030 }, { "epoch": 0.27, "learning_rate": 1.819412282118944e-05, "loss": 2.5992, "step": 5040 }, { "epoch": 0.27, "learning_rate": 1.819053260811029e-05, "loss": 2.5028, "step": 5050 }, { "epoch": 0.27, "learning_rate": 1.8186942395031147e-05, "loss": 2.3993, "step": 5060 }, { "epoch": 0.27, "learning_rate": 1.8183352181952e-05, "loss": 2.4996, "step": 5070 }, { "epoch": 0.27, "learning_rate": 1.8179761968872855e-05, "loss": 2.5207, "step": 5080 }, { "epoch": 0.27, "learning_rate": 1.8176171755793707e-05, "loss": 2.5934, "step": 5090 }, { "epoch": 0.27, "learning_rate": 1.8172581542714563e-05, "loss": 2.4164, "step": 5100 }, { "epoch": 0.28, "learning_rate": 1.8168991329635415e-05, "loss": 2.4869, "step": 5110 }, { "epoch": 0.28, "learning_rate": 1.816540111655627e-05, "loss": 2.562, "step": 5120 }, { "epoch": 0.28, "learning_rate": 1.8161810903477122e-05, "loss": 2.4455, "step": 5130 }, { "epoch": 0.28, "learning_rate": 1.8158220690397978e-05, "loss": 2.5116, "step": 5140 }, { "epoch": 0.28, "learning_rate": 1.815463047731883e-05, "loss": 2.5017, "step": 5150 }, { "epoch": 0.28, "learning_rate": 1.8151040264239682e-05, "loss": 2.411, "step": 5160 }, { "epoch": 0.28, "learning_rate": 1.8147450051160538e-05, "loss": 2.4743, "step": 5170 }, { "epoch": 0.28, "learning_rate": 1.8143859838081393e-05, "loss": 2.2396, "step": 5180 }, { "epoch": 0.28, "learning_rate": 1.8140269625002246e-05, "loss": 2.532, "step": 5190 }, { "epoch": 0.28, "learning_rate": 1.8136679411923098e-05, "loss": 2.4305, "step": 5200 }, { "epoch": 0.28, "learning_rate": 1.8133089198843953e-05, "loss": 2.4409, "step": 5210 }, { "epoch": 0.28, "learning_rate": 1.8129498985764805e-05, "loss": 2.5456, "step": 5220 }, { "epoch": 0.28, "learning_rate": 1.812590877268566e-05, "loss": 2.4283, "step": 5230 }, { "epoch": 0.28, "learning_rate": 1.8122318559606513e-05, "loss": 2.3817, "step": 5240 }, { "epoch": 0.28, "learning_rate": 1.811872834652737e-05, "loss": 2.4492, "step": 5250 }, { "epoch": 0.28, "learning_rate": 1.811513813344822e-05, "loss": 2.4183, "step": 5260 }, { "epoch": 0.28, "learning_rate": 1.8111547920369076e-05, "loss": 2.3994, "step": 5270 }, { "epoch": 0.28, "learning_rate": 1.810795770728993e-05, "loss": 2.5341, "step": 5280 }, { "epoch": 0.28, "learning_rate": 1.8104367494210784e-05, "loss": 2.5021, "step": 5290 }, { "epoch": 0.29, "learning_rate": 1.8100777281131636e-05, "loss": 2.3663, "step": 5300 }, { "epoch": 0.29, "learning_rate": 1.8097187068052492e-05, "loss": 2.3879, "step": 5310 }, { "epoch": 0.29, "learning_rate": 1.8093596854973344e-05, "loss": 2.4557, "step": 5320 }, { "epoch": 0.29, "learning_rate": 1.80900066418942e-05, "loss": 2.5436, "step": 5330 }, { "epoch": 0.29, "learning_rate": 1.808641642881505e-05, "loss": 2.4606, "step": 5340 }, { "epoch": 0.29, "learning_rate": 1.8082826215735904e-05, "loss": 2.3418, "step": 5350 }, { "epoch": 0.29, "learning_rate": 1.807923600265676e-05, "loss": 2.4659, "step": 5360 }, { "epoch": 0.29, "learning_rate": 1.8075645789577615e-05, "loss": 2.3854, "step": 5370 }, { "epoch": 0.29, "learning_rate": 1.8072055576498467e-05, "loss": 2.3809, "step": 5380 }, { "epoch": 0.29, "learning_rate": 1.806846536341932e-05, "loss": 2.4259, "step": 5390 }, { "epoch": 0.29, "learning_rate": 1.8064875150340175e-05, "loss": 2.3579, "step": 5400 }, { "epoch": 0.29, "learning_rate": 1.8061284937261027e-05, "loss": 2.4097, "step": 5410 }, { "epoch": 0.29, "learning_rate": 1.8057694724181882e-05, "loss": 2.5435, "step": 5420 }, { "epoch": 0.29, "learning_rate": 1.8054104511102735e-05, "loss": 2.4496, "step": 5430 }, { "epoch": 0.29, "learning_rate": 1.805051429802359e-05, "loss": 2.4193, "step": 5440 }, { "epoch": 0.29, "learning_rate": 1.8046924084944442e-05, "loss": 2.4353, "step": 5450 }, { "epoch": 0.29, "learning_rate": 1.8043333871865298e-05, "loss": 2.3613, "step": 5460 }, { "epoch": 0.29, "learning_rate": 1.803974365878615e-05, "loss": 2.4537, "step": 5470 }, { "epoch": 0.3, "learning_rate": 1.8036153445707002e-05, "loss": 2.3768, "step": 5480 }, { "epoch": 0.3, "learning_rate": 1.8032563232627858e-05, "loss": 2.3984, "step": 5490 }, { "epoch": 0.3, "learning_rate": 1.8028973019548713e-05, "loss": 2.5764, "step": 5500 }, { "epoch": 0.3, "learning_rate": 1.8025382806469565e-05, "loss": 2.5871, "step": 5510 }, { "epoch": 0.3, "learning_rate": 1.802179259339042e-05, "loss": 2.4675, "step": 5520 }, { "epoch": 0.3, "learning_rate": 1.8018202380311273e-05, "loss": 2.4377, "step": 5530 }, { "epoch": 0.3, "learning_rate": 1.8014612167232125e-05, "loss": 2.3109, "step": 5540 }, { "epoch": 0.3, "learning_rate": 1.801102195415298e-05, "loss": 2.2933, "step": 5550 }, { "epoch": 0.3, "learning_rate": 1.8007431741073836e-05, "loss": 2.3383, "step": 5560 }, { "epoch": 0.3, "learning_rate": 1.800384152799469e-05, "loss": 2.4377, "step": 5570 }, { "epoch": 0.3, "learning_rate": 1.800025131491554e-05, "loss": 2.4252, "step": 5580 }, { "epoch": 0.3, "learning_rate": 1.7996661101836396e-05, "loss": 2.5476, "step": 5590 }, { "epoch": 0.3, "learning_rate": 1.7993070888757248e-05, "loss": 2.251, "step": 5600 }, { "epoch": 0.3, "learning_rate": 1.7989480675678104e-05, "loss": 2.2933, "step": 5610 }, { "epoch": 0.3, "learning_rate": 1.7985890462598956e-05, "loss": 2.5557, "step": 5620 }, { "epoch": 0.3, "learning_rate": 1.798230024951981e-05, "loss": 2.5928, "step": 5630 }, { "epoch": 0.3, "learning_rate": 1.7978710036440664e-05, "loss": 2.4235, "step": 5640 }, { "epoch": 0.3, "learning_rate": 1.797511982336152e-05, "loss": 2.3717, "step": 5650 }, { "epoch": 0.3, "learning_rate": 1.797152961028237e-05, "loss": 2.4914, "step": 5660 }, { "epoch": 0.31, "learning_rate": 1.7967939397203224e-05, "loss": 2.779, "step": 5670 }, { "epoch": 0.31, "learning_rate": 1.796434918412408e-05, "loss": 2.8167, "step": 5680 }, { "epoch": 0.31, "learning_rate": 1.7960758971044935e-05, "loss": 2.5284, "step": 5690 }, { "epoch": 0.31, "learning_rate": 1.7957168757965787e-05, "loss": 2.3563, "step": 5700 }, { "epoch": 0.31, "learning_rate": 1.795357854488664e-05, "loss": 2.4062, "step": 5710 }, { "epoch": 0.31, "learning_rate": 1.7949988331807495e-05, "loss": 2.2945, "step": 5720 }, { "epoch": 0.31, "learning_rate": 1.7946398118728347e-05, "loss": 2.4604, "step": 5730 }, { "epoch": 0.31, "learning_rate": 1.7942807905649202e-05, "loss": 2.4568, "step": 5740 }, { "epoch": 0.31, "learning_rate": 1.7939217692570054e-05, "loss": 2.4982, "step": 5750 }, { "epoch": 0.31, "learning_rate": 1.793562747949091e-05, "loss": 2.525, "step": 5760 }, { "epoch": 0.31, "learning_rate": 1.7932037266411762e-05, "loss": 2.2969, "step": 5770 }, { "epoch": 0.31, "learning_rate": 1.7928447053332618e-05, "loss": 2.3002, "step": 5780 }, { "epoch": 0.31, "learning_rate": 1.792485684025347e-05, "loss": 2.3354, "step": 5790 }, { "epoch": 0.31, "learning_rate": 1.7921266627174325e-05, "loss": 2.46, "step": 5800 }, { "epoch": 0.31, "learning_rate": 1.7917676414095177e-05, "loss": 2.2436, "step": 5810 }, { "epoch": 0.31, "learning_rate": 1.7914086201016033e-05, "loss": 2.3628, "step": 5820 }, { "epoch": 0.31, "learning_rate": 1.7910495987936885e-05, "loss": 2.4019, "step": 5830 }, { "epoch": 0.31, "learning_rate": 1.790690577485774e-05, "loss": 2.3565, "step": 5840 }, { "epoch": 0.32, "learning_rate": 1.7903315561778593e-05, "loss": 2.3806, "step": 5850 }, { "epoch": 0.32, "learning_rate": 1.7899725348699445e-05, "loss": 2.4735, "step": 5860 }, { "epoch": 0.32, "learning_rate": 1.78961351356203e-05, "loss": 2.2858, "step": 5870 }, { "epoch": 0.32, "learning_rate": 1.7892544922541156e-05, "loss": 2.4405, "step": 5880 }, { "epoch": 0.32, "learning_rate": 1.7888954709462008e-05, "loss": 2.51, "step": 5890 }, { "epoch": 0.32, "learning_rate": 1.788536449638286e-05, "loss": 2.3536, "step": 5900 }, { "epoch": 0.32, "learning_rate": 1.7881774283303716e-05, "loss": 2.3228, "step": 5910 }, { "epoch": 0.32, "learning_rate": 1.7878184070224568e-05, "loss": 2.3953, "step": 5920 }, { "epoch": 0.32, "learning_rate": 1.7874593857145424e-05, "loss": 2.3964, "step": 5930 }, { "epoch": 0.32, "learning_rate": 1.7871003644066276e-05, "loss": 2.2246, "step": 5940 }, { "epoch": 0.32, "learning_rate": 1.786741343098713e-05, "loss": 2.4914, "step": 5950 }, { "epoch": 0.32, "learning_rate": 1.7863823217907984e-05, "loss": 2.4168, "step": 5960 }, { "epoch": 0.32, "learning_rate": 1.786023300482884e-05, "loss": 2.386, "step": 5970 }, { "epoch": 0.32, "learning_rate": 1.785664279174969e-05, "loss": 2.4283, "step": 5980 }, { "epoch": 0.32, "learning_rate": 1.7853052578670547e-05, "loss": 2.4177, "step": 5990 }, { "epoch": 0.32, "learning_rate": 1.78494623655914e-05, "loss": 2.5387, "step": 6000 }, { "epoch": 0.32, "learning_rate": 1.7845872152512254e-05, "loss": 2.4892, "step": 6010 }, { "epoch": 0.32, "learning_rate": 1.7842281939433107e-05, "loss": 2.3955, "step": 6020 }, { "epoch": 0.32, "learning_rate": 1.7838691726353962e-05, "loss": 2.2883, "step": 6030 }, { "epoch": 0.33, "learning_rate": 1.7835101513274814e-05, "loss": 2.3865, "step": 6040 }, { "epoch": 0.33, "learning_rate": 1.7831511300195667e-05, "loss": 2.4296, "step": 6050 }, { "epoch": 0.33, "learning_rate": 1.7827921087116522e-05, "loss": 2.1582, "step": 6060 }, { "epoch": 0.33, "learning_rate": 1.7824330874037378e-05, "loss": 2.3292, "step": 6070 }, { "epoch": 0.33, "learning_rate": 1.782074066095823e-05, "loss": 2.2857, "step": 6080 }, { "epoch": 0.33, "learning_rate": 1.7817150447879082e-05, "loss": 2.3803, "step": 6090 }, { "epoch": 0.33, "learning_rate": 1.7813560234799937e-05, "loss": 2.2857, "step": 6100 }, { "epoch": 0.33, "learning_rate": 1.780997002172079e-05, "loss": 2.4408, "step": 6110 }, { "epoch": 0.33, "learning_rate": 1.7806379808641645e-05, "loss": 2.3569, "step": 6120 }, { "epoch": 0.33, "learning_rate": 1.7802789595562497e-05, "loss": 2.3872, "step": 6130 }, { "epoch": 0.33, "learning_rate": 1.7799199382483353e-05, "loss": 2.476, "step": 6140 }, { "epoch": 0.33, "learning_rate": 1.7795609169404205e-05, "loss": 2.3956, "step": 6150 }, { "epoch": 0.33, "learning_rate": 1.779201895632506e-05, "loss": 2.3636, "step": 6160 }, { "epoch": 0.33, "learning_rate": 1.7788428743245913e-05, "loss": 2.4033, "step": 6170 }, { "epoch": 0.33, "learning_rate": 1.7784838530166765e-05, "loss": 2.4184, "step": 6180 }, { "epoch": 0.33, "learning_rate": 1.778124831708762e-05, "loss": 2.3507, "step": 6190 }, { "epoch": 0.33, "learning_rate": 1.7777658104008476e-05, "loss": 2.2903, "step": 6200 }, { "epoch": 0.33, "learning_rate": 1.7774067890929328e-05, "loss": 2.3092, "step": 6210 }, { "epoch": 0.33, "learning_rate": 1.777047767785018e-05, "loss": 2.4227, "step": 6220 }, { "epoch": 0.34, "learning_rate": 1.7766887464771036e-05, "loss": 2.2057, "step": 6230 }, { "epoch": 0.34, "learning_rate": 1.7763297251691888e-05, "loss": 2.3168, "step": 6240 }, { "epoch": 0.34, "learning_rate": 1.7759707038612744e-05, "loss": 2.2859, "step": 6250 }, { "epoch": 0.34, "learning_rate": 1.7756116825533596e-05, "loss": 2.3328, "step": 6260 }, { "epoch": 0.34, "learning_rate": 1.775252661245445e-05, "loss": 2.3735, "step": 6270 }, { "epoch": 0.34, "learning_rate": 1.7748936399375303e-05, "loss": 2.3695, "step": 6280 }, { "epoch": 0.34, "learning_rate": 1.774534618629616e-05, "loss": 2.2232, "step": 6290 }, { "epoch": 0.34, "learning_rate": 1.774175597321701e-05, "loss": 2.2689, "step": 6300 }, { "epoch": 0.34, "learning_rate": 1.7738165760137867e-05, "loss": 2.3546, "step": 6310 }, { "epoch": 0.34, "learning_rate": 1.773457554705872e-05, "loss": 2.2545, "step": 6320 }, { "epoch": 0.34, "learning_rate": 1.7730985333979574e-05, "loss": 2.4258, "step": 6330 }, { "epoch": 0.34, "learning_rate": 1.7727395120900426e-05, "loss": 2.3306, "step": 6340 }, { "epoch": 0.34, "learning_rate": 1.7723804907821282e-05, "loss": 2.4172, "step": 6350 }, { "epoch": 0.34, "learning_rate": 1.7720214694742134e-05, "loss": 2.3494, "step": 6360 }, { "epoch": 0.34, "learning_rate": 1.7716624481662986e-05, "loss": 2.511, "step": 6370 }, { "epoch": 0.34, "learning_rate": 1.7713034268583842e-05, "loss": 2.2993, "step": 6380 }, { "epoch": 0.34, "learning_rate": 1.7709444055504697e-05, "loss": 2.4266, "step": 6390 }, { "epoch": 0.34, "learning_rate": 1.770585384242555e-05, "loss": 2.3397, "step": 6400 }, { "epoch": 0.35, "learning_rate": 1.7702263629346402e-05, "loss": 2.2552, "step": 6410 }, { "epoch": 0.35, "learning_rate": 1.7698673416267257e-05, "loss": 2.287, "step": 6420 }, { "epoch": 0.35, "learning_rate": 1.769508320318811e-05, "loss": 2.2259, "step": 6430 }, { "epoch": 0.35, "learning_rate": 1.7691492990108965e-05, "loss": 2.3773, "step": 6440 }, { "epoch": 0.35, "learning_rate": 1.7687902777029817e-05, "loss": 2.273, "step": 6450 }, { "epoch": 0.35, "learning_rate": 1.7684312563950673e-05, "loss": 2.2841, "step": 6460 }, { "epoch": 0.35, "learning_rate": 1.7680722350871525e-05, "loss": 2.3063, "step": 6470 }, { "epoch": 0.35, "learning_rate": 1.767713213779238e-05, "loss": 2.2447, "step": 6480 }, { "epoch": 0.35, "learning_rate": 1.7673541924713233e-05, "loss": 2.3697, "step": 6490 }, { "epoch": 0.35, "learning_rate": 1.7669951711634088e-05, "loss": 2.19, "step": 6500 }, { "epoch": 0.35, "learning_rate": 1.766636149855494e-05, "loss": 2.2496, "step": 6510 }, { "epoch": 0.35, "learning_rate": 1.7662771285475796e-05, "loss": 2.3629, "step": 6520 }, { "epoch": 0.35, "learning_rate": 1.7659181072396648e-05, "loss": 2.3822, "step": 6530 }, { "epoch": 0.35, "learning_rate": 1.7655590859317503e-05, "loss": 2.3546, "step": 6540 }, { "epoch": 0.35, "learning_rate": 1.7652000646238356e-05, "loss": 2.3238, "step": 6550 }, { "epoch": 0.35, "learning_rate": 1.7648410433159208e-05, "loss": 2.4195, "step": 6560 }, { "epoch": 0.35, "learning_rate": 1.7644820220080063e-05, "loss": 2.3532, "step": 6570 }, { "epoch": 0.35, "learning_rate": 1.764123000700092e-05, "loss": 2.407, "step": 6580 }, { "epoch": 0.35, "learning_rate": 1.763763979392177e-05, "loss": 2.2077, "step": 6590 }, { "epoch": 0.36, "learning_rate": 1.7634049580842623e-05, "loss": 2.3895, "step": 6600 }, { "epoch": 0.36, "learning_rate": 1.763045936776348e-05, "loss": 2.2603, "step": 6610 }, { "epoch": 0.36, "learning_rate": 1.762686915468433e-05, "loss": 2.2514, "step": 6620 }, { "epoch": 0.36, "learning_rate": 1.7623278941605186e-05, "loss": 2.4831, "step": 6630 }, { "epoch": 0.36, "learning_rate": 1.761968872852604e-05, "loss": 2.3825, "step": 6640 }, { "epoch": 0.36, "learning_rate": 1.7616098515446894e-05, "loss": 2.2755, "step": 6650 }, { "epoch": 0.36, "learning_rate": 1.7612508302367746e-05, "loss": 2.2863, "step": 6660 }, { "epoch": 0.36, "learning_rate": 1.7608918089288602e-05, "loss": 2.3306, "step": 6670 }, { "epoch": 0.36, "learning_rate": 1.7605327876209454e-05, "loss": 2.5017, "step": 6680 }, { "epoch": 0.36, "learning_rate": 1.7601737663130306e-05, "loss": 2.4517, "step": 6690 }, { "epoch": 0.36, "learning_rate": 1.7598147450051162e-05, "loss": 2.3501, "step": 6700 }, { "epoch": 0.36, "learning_rate": 1.7594557236972017e-05, "loss": 2.753, "step": 6710 }, { "epoch": 0.36, "learning_rate": 1.759096702389287e-05, "loss": 5.5076, "step": 6720 }, { "epoch": 0.36, "learning_rate": 1.7587376810813725e-05, "loss": 3.9846, "step": 6730 }, { "epoch": 0.36, "learning_rate": 1.7583786597734577e-05, "loss": 2.4476, "step": 6740 }, { "epoch": 0.36, "learning_rate": 1.758019638465543e-05, "loss": 2.3937, "step": 6750 }, { "epoch": 0.36, "learning_rate": 1.7576606171576285e-05, "loss": 2.3642, "step": 6760 }, { "epoch": 0.36, "learning_rate": 1.7573015958497137e-05, "loss": 2.5941, "step": 6770 }, { "epoch": 0.37, "learning_rate": 1.7569425745417993e-05, "loss": 2.4031, "step": 6780 }, { "epoch": 0.37, "learning_rate": 1.7565835532338845e-05, "loss": 2.3498, "step": 6790 }, { "epoch": 0.37, "learning_rate": 1.75622453192597e-05, "loss": 2.4207, "step": 6800 }, { "epoch": 0.37, "learning_rate": 1.7558655106180552e-05, "loss": 2.3942, "step": 6810 }, { "epoch": 0.37, "learning_rate": 1.7555064893101408e-05, "loss": 2.5219, "step": 6820 }, { "epoch": 0.37, "learning_rate": 1.755147468002226e-05, "loss": 2.5068, "step": 6830 }, { "epoch": 0.37, "learning_rate": 1.7547884466943116e-05, "loss": 2.6013, "step": 6840 }, { "epoch": 0.37, "learning_rate": 1.7544294253863968e-05, "loss": 2.4904, "step": 6850 }, { "epoch": 0.37, "learning_rate": 1.7540704040784823e-05, "loss": 2.575, "step": 6860 }, { "epoch": 0.37, "learning_rate": 1.7537113827705675e-05, "loss": 2.3668, "step": 6870 }, { "epoch": 0.37, "learning_rate": 1.7533523614626528e-05, "loss": 2.5214, "step": 6880 }, { "epoch": 0.37, "learning_rate": 1.7529933401547383e-05, "loss": 2.3257, "step": 6890 }, { "epoch": 0.37, "learning_rate": 1.752634318846824e-05, "loss": 2.2955, "step": 6900 }, { "epoch": 0.37, "learning_rate": 1.752275297538909e-05, "loss": 2.1785, "step": 6910 }, { "epoch": 0.37, "learning_rate": 1.7519162762309943e-05, "loss": 2.3821, "step": 6920 }, { "epoch": 0.37, "learning_rate": 1.75155725492308e-05, "loss": 2.4626, "step": 6930 }, { "epoch": 0.37, "learning_rate": 1.751198233615165e-05, "loss": 2.2706, "step": 6940 }, { "epoch": 0.37, "learning_rate": 1.7508392123072506e-05, "loss": 2.2919, "step": 6950 }, { "epoch": 0.37, "learning_rate": 1.750480190999336e-05, "loss": 2.2875, "step": 6960 }, { "epoch": 0.38, "learning_rate": 1.7501211696914214e-05, "loss": 2.4828, "step": 6970 }, { "epoch": 0.38, "learning_rate": 1.7497621483835066e-05, "loss": 2.3753, "step": 6980 }, { "epoch": 0.38, "learning_rate": 1.7494031270755922e-05, "loss": 2.2839, "step": 6990 }, { "epoch": 0.38, "learning_rate": 1.7490441057676774e-05, "loss": 2.2531, "step": 7000 }, { "epoch": 0.38, "learning_rate": 1.748685084459763e-05, "loss": 2.3371, "step": 7010 }, { "epoch": 0.38, "learning_rate": 1.748326063151848e-05, "loss": 2.3223, "step": 7020 }, { "epoch": 0.38, "learning_rate": 1.7479670418439337e-05, "loss": 2.3399, "step": 7030 }, { "epoch": 0.38, "learning_rate": 1.747608020536019e-05, "loss": 2.3653, "step": 7040 }, { "epoch": 0.38, "learning_rate": 1.7472489992281045e-05, "loss": 2.3339, "step": 7050 }, { "epoch": 0.38, "learning_rate": 1.7468899779201897e-05, "loss": 2.2744, "step": 7060 }, { "epoch": 0.38, "learning_rate": 1.746530956612275e-05, "loss": 2.2977, "step": 7070 }, { "epoch": 0.38, "learning_rate": 1.7461719353043605e-05, "loss": 2.2017, "step": 7080 }, { "epoch": 0.38, "learning_rate": 1.745812913996446e-05, "loss": 2.3658, "step": 7090 }, { "epoch": 0.38, "learning_rate": 1.7454538926885312e-05, "loss": 2.3233, "step": 7100 }, { "epoch": 0.38, "learning_rate": 1.7450948713806165e-05, "loss": 2.3938, "step": 7110 }, { "epoch": 0.38, "learning_rate": 1.744735850072702e-05, "loss": 2.3039, "step": 7120 }, { "epoch": 0.38, "learning_rate": 1.7443768287647872e-05, "loss": 2.4032, "step": 7130 }, { "epoch": 0.38, "learning_rate": 1.7440178074568728e-05, "loss": 2.4282, "step": 7140 }, { "epoch": 0.39, "learning_rate": 1.743658786148958e-05, "loss": 2.3364, "step": 7150 }, { "epoch": 0.39, "learning_rate": 1.7432997648410435e-05, "loss": 2.3028, "step": 7160 }, { "epoch": 0.39, "learning_rate": 1.7429407435331288e-05, "loss": 2.3196, "step": 7170 }, { "epoch": 0.39, "learning_rate": 1.7425817222252143e-05, "loss": 2.4483, "step": 7180 }, { "epoch": 0.39, "learning_rate": 1.7422227009172995e-05, "loss": 2.3709, "step": 7190 }, { "epoch": 0.39, "learning_rate": 1.7418636796093847e-05, "loss": 2.2443, "step": 7200 }, { "epoch": 0.39, "learning_rate": 1.7415046583014703e-05, "loss": 2.501, "step": 7210 }, { "epoch": 0.39, "learning_rate": 1.741145636993556e-05, "loss": 2.3832, "step": 7220 }, { "epoch": 0.39, "learning_rate": 1.740786615685641e-05, "loss": 2.3266, "step": 7230 }, { "epoch": 0.39, "learning_rate": 1.7404275943777266e-05, "loss": 2.2613, "step": 7240 }, { "epoch": 0.39, "learning_rate": 1.740068573069812e-05, "loss": 2.362, "step": 7250 }, { "epoch": 0.39, "learning_rate": 1.739709551761897e-05, "loss": 2.2679, "step": 7260 }, { "epoch": 0.39, "learning_rate": 1.7393505304539826e-05, "loss": 2.2904, "step": 7270 }, { "epoch": 0.39, "learning_rate": 1.738991509146068e-05, "loss": 2.108, "step": 7280 }, { "epoch": 0.39, "learning_rate": 1.7386324878381534e-05, "loss": 2.2689, "step": 7290 }, { "epoch": 0.39, "learning_rate": 1.7382734665302386e-05, "loss": 2.317, "step": 7300 }, { "epoch": 0.39, "learning_rate": 1.737914445222324e-05, "loss": 2.3217, "step": 7310 }, { "epoch": 0.39, "learning_rate": 1.7375554239144094e-05, "loss": 2.2032, "step": 7320 }, { "epoch": 0.39, "learning_rate": 1.737196402606495e-05, "loss": 2.3136, "step": 7330 }, { "epoch": 0.4, "learning_rate": 1.73683738129858e-05, "loss": 2.381, "step": 7340 }, { "epoch": 0.4, "learning_rate": 1.7364783599906657e-05, "loss": 2.1501, "step": 7350 }, { "epoch": 0.4, "learning_rate": 1.736119338682751e-05, "loss": 2.215, "step": 7360 }, { "epoch": 0.4, "learning_rate": 1.7357603173748365e-05, "loss": 2.2151, "step": 7370 }, { "epoch": 0.4, "learning_rate": 1.7354012960669217e-05, "loss": 2.2561, "step": 7380 }, { "epoch": 0.4, "learning_rate": 1.735042274759007e-05, "loss": 2.2621, "step": 7390 }, { "epoch": 0.4, "learning_rate": 1.7346832534510924e-05, "loss": 2.2839, "step": 7400 }, { "epoch": 0.4, "learning_rate": 1.734324232143178e-05, "loss": 2.2633, "step": 7410 }, { "epoch": 0.4, "learning_rate": 1.7339652108352632e-05, "loss": 2.2914, "step": 7420 }, { "epoch": 0.4, "learning_rate": 1.7336061895273484e-05, "loss": 2.2161, "step": 7430 }, { "epoch": 0.4, "learning_rate": 1.733247168219434e-05, "loss": 2.1786, "step": 7440 }, { "epoch": 0.4, "learning_rate": 1.7328881469115192e-05, "loss": 2.2575, "step": 7450 }, { "epoch": 0.4, "learning_rate": 1.7325291256036048e-05, "loss": 2.2606, "step": 7460 }, { "epoch": 0.4, "learning_rate": 1.73217010429569e-05, "loss": 2.2184, "step": 7470 }, { "epoch": 0.4, "learning_rate": 1.7318110829877755e-05, "loss": 2.2805, "step": 7480 }, { "epoch": 0.4, "learning_rate": 1.7314520616798607e-05, "loss": 2.3843, "step": 7490 }, { "epoch": 0.4, "learning_rate": 1.7310930403719463e-05, "loss": 2.3597, "step": 7500 }, { "epoch": 0.4, "learning_rate": 1.7307340190640315e-05, "loss": 2.2498, "step": 7510 }, { "epoch": 0.4, "learning_rate": 1.730374997756117e-05, "loss": 2.3065, "step": 7520 }, { "epoch": 0.41, "learning_rate": 1.7300159764482023e-05, "loss": 2.2381, "step": 7530 }, { "epoch": 0.41, "learning_rate": 1.729656955140288e-05, "loss": 2.1535, "step": 7540 }, { "epoch": 0.41, "learning_rate": 1.729297933832373e-05, "loss": 2.3371, "step": 7550 }, { "epoch": 0.41, "learning_rate": 1.7289389125244586e-05, "loss": 2.143, "step": 7560 }, { "epoch": 0.41, "learning_rate": 1.7285798912165438e-05, "loss": 2.4662, "step": 7570 }, { "epoch": 0.41, "learning_rate": 1.728220869908629e-05, "loss": 2.2632, "step": 7580 }, { "epoch": 0.41, "learning_rate": 1.7278618486007146e-05, "loss": 2.2501, "step": 7590 }, { "epoch": 0.41, "learning_rate": 1.7275028272928e-05, "loss": 2.2385, "step": 7600 }, { "epoch": 0.41, "learning_rate": 1.7271438059848854e-05, "loss": 2.1904, "step": 7610 }, { "epoch": 0.41, "learning_rate": 1.7267847846769706e-05, "loss": 2.2508, "step": 7620 }, { "epoch": 0.41, "learning_rate": 1.726425763369056e-05, "loss": 2.2618, "step": 7630 }, { "epoch": 0.41, "learning_rate": 1.7260667420611414e-05, "loss": 2.3615, "step": 7640 }, { "epoch": 0.41, "learning_rate": 1.725707720753227e-05, "loss": 2.273, "step": 7650 }, { "epoch": 0.41, "learning_rate": 1.725348699445312e-05, "loss": 2.2675, "step": 7660 }, { "epoch": 0.41, "learning_rate": 1.7249896781373977e-05, "loss": 2.1635, "step": 7670 }, { "epoch": 0.41, "learning_rate": 1.724630656829483e-05, "loss": 2.2832, "step": 7680 }, { "epoch": 0.41, "learning_rate": 1.7242716355215684e-05, "loss": 2.2099, "step": 7690 }, { "epoch": 0.41, "learning_rate": 1.7239126142136537e-05, "loss": 2.3396, "step": 7700 }, { "epoch": 0.42, "learning_rate": 1.7235535929057392e-05, "loss": 2.2692, "step": 7710 }, { "epoch": 0.42, "learning_rate": 1.7231945715978244e-05, "loss": 2.3635, "step": 7720 }, { "epoch": 0.42, "learning_rate": 1.72283555028991e-05, "loss": 2.2206, "step": 7730 }, { "epoch": 0.42, "learning_rate": 1.7224765289819952e-05, "loss": 2.2761, "step": 7740 }, { "epoch": 0.42, "learning_rate": 1.7221175076740808e-05, "loss": 2.3127, "step": 7750 }, { "epoch": 0.42, "learning_rate": 1.721758486366166e-05, "loss": 2.2085, "step": 7760 }, { "epoch": 0.42, "learning_rate": 1.7213994650582512e-05, "loss": 2.2609, "step": 7770 }, { "epoch": 0.42, "learning_rate": 1.7210404437503367e-05, "loss": 2.1903, "step": 7780 }, { "epoch": 0.42, "learning_rate": 1.7206814224424223e-05, "loss": 2.2663, "step": 7790 }, { "epoch": 0.42, "learning_rate": 1.7203224011345075e-05, "loss": 2.2721, "step": 7800 }, { "epoch": 0.42, "learning_rate": 1.7199633798265927e-05, "loss": 2.3648, "step": 7810 }, { "epoch": 0.42, "learning_rate": 1.7196043585186783e-05, "loss": 2.1694, "step": 7820 }, { "epoch": 0.42, "learning_rate": 1.7192453372107635e-05, "loss": 2.2729, "step": 7830 }, { "epoch": 0.42, "learning_rate": 1.718886315902849e-05, "loss": 2.2186, "step": 7840 }, { "epoch": 0.42, "learning_rate": 1.7185272945949343e-05, "loss": 2.4567, "step": 7850 }, { "epoch": 0.42, "learning_rate": 1.7181682732870198e-05, "loss": 2.4213, "step": 7860 }, { "epoch": 0.42, "learning_rate": 1.717809251979105e-05, "loss": 2.2289, "step": 7870 }, { "epoch": 0.42, "learning_rate": 1.7174502306711906e-05, "loss": 2.2743, "step": 7880 }, { "epoch": 0.42, "learning_rate": 1.7170912093632758e-05, "loss": 2.4067, "step": 7890 }, { "epoch": 0.43, "learning_rate": 1.716732188055361e-05, "loss": 2.2565, "step": 7900 }, { "epoch": 0.43, "learning_rate": 1.7163731667474466e-05, "loss": 2.1664, "step": 7910 }, { "epoch": 0.43, "learning_rate": 1.716014145439532e-05, "loss": 2.3281, "step": 7920 }, { "epoch": 0.43, "learning_rate": 1.7156551241316174e-05, "loss": 2.2045, "step": 7930 }, { "epoch": 0.43, "learning_rate": 1.7152961028237026e-05, "loss": 2.2015, "step": 7940 }, { "epoch": 0.43, "learning_rate": 1.714937081515788e-05, "loss": 2.2742, "step": 7950 }, { "epoch": 0.43, "learning_rate": 1.7145780602078733e-05, "loss": 2.3435, "step": 7960 }, { "epoch": 0.43, "learning_rate": 1.714219038899959e-05, "loss": 2.2946, "step": 7970 }, { "epoch": 0.43, "learning_rate": 1.713860017592044e-05, "loss": 2.3054, "step": 7980 }, { "epoch": 0.43, "learning_rate": 1.7135009962841297e-05, "loss": 2.2602, "step": 7990 }, { "epoch": 0.43, "learning_rate": 1.713141974976215e-05, "loss": 2.1613, "step": 8000 }, { "epoch": 0.43, "learning_rate": 1.7127829536683004e-05, "loss": 2.1763, "step": 8010 }, { "epoch": 0.43, "learning_rate": 1.7124239323603856e-05, "loss": 2.195, "step": 8020 }, { "epoch": 0.43, "learning_rate": 1.7120649110524712e-05, "loss": 2.1862, "step": 8030 }, { "epoch": 0.43, "learning_rate": 1.7117058897445564e-05, "loss": 2.1873, "step": 8040 }, { "epoch": 0.43, "learning_rate": 1.711346868436642e-05, "loss": 2.3154, "step": 8050 }, { "epoch": 0.43, "learning_rate": 1.7109878471287272e-05, "loss": 2.3125, "step": 8060 }, { "epoch": 0.43, "learning_rate": 1.7106288258208127e-05, "loss": 2.3045, "step": 8070 }, { "epoch": 0.44, "learning_rate": 1.710269804512898e-05, "loss": 2.2409, "step": 8080 }, { "epoch": 0.44, "learning_rate": 1.7099107832049832e-05, "loss": 2.1399, "step": 8090 }, { "epoch": 0.44, "learning_rate": 1.7095517618970687e-05, "loss": 2.2779, "step": 8100 }, { "epoch": 0.44, "learning_rate": 1.7091927405891543e-05, "loss": 2.2896, "step": 8110 }, { "epoch": 0.44, "learning_rate": 1.7088337192812395e-05, "loss": 2.1175, "step": 8120 }, { "epoch": 0.44, "learning_rate": 1.7084746979733247e-05, "loss": 2.3036, "step": 8130 }, { "epoch": 0.44, "learning_rate": 1.7081156766654103e-05, "loss": 2.1518, "step": 8140 }, { "epoch": 0.44, "learning_rate": 1.7077566553574955e-05, "loss": 2.265, "step": 8150 }, { "epoch": 0.44, "learning_rate": 1.707397634049581e-05, "loss": 2.2341, "step": 8160 }, { "epoch": 0.44, "learning_rate": 1.7070386127416663e-05, "loss": 2.1126, "step": 8170 }, { "epoch": 0.44, "learning_rate": 1.7066795914337518e-05, "loss": 2.2285, "step": 8180 }, { "epoch": 0.44, "learning_rate": 1.706320570125837e-05, "loss": 2.298, "step": 8190 }, { "epoch": 0.44, "learning_rate": 1.7059615488179226e-05, "loss": 2.343, "step": 8200 }, { "epoch": 0.44, "learning_rate": 1.7056025275100078e-05, "loss": 2.2423, "step": 8210 }, { "epoch": 0.44, "learning_rate": 1.7052435062020933e-05, "loss": 2.2843, "step": 8220 }, { "epoch": 0.44, "learning_rate": 1.7048844848941786e-05, "loss": 2.2622, "step": 8230 }, { "epoch": 0.44, "learning_rate": 1.704525463586264e-05, "loss": 2.2375, "step": 8240 }, { "epoch": 0.44, "learning_rate": 1.7041664422783493e-05, "loss": 2.2831, "step": 8250 }, { "epoch": 0.44, "learning_rate": 1.703807420970435e-05, "loss": 2.1778, "step": 8260 }, { "epoch": 0.45, "learning_rate": 1.70344839966252e-05, "loss": 2.1791, "step": 8270 }, { "epoch": 0.45, "learning_rate": 1.7030893783546053e-05, "loss": 2.306, "step": 8280 }, { "epoch": 0.45, "learning_rate": 1.702730357046691e-05, "loss": 2.3793, "step": 8290 }, { "epoch": 0.45, "learning_rate": 1.7023713357387764e-05, "loss": 2.2369, "step": 8300 }, { "epoch": 0.45, "learning_rate": 1.7020123144308616e-05, "loss": 2.2126, "step": 8310 }, { "epoch": 0.45, "learning_rate": 1.701653293122947e-05, "loss": 2.226, "step": 8320 }, { "epoch": 0.45, "learning_rate": 1.7012942718150324e-05, "loss": 2.2276, "step": 8330 }, { "epoch": 0.45, "learning_rate": 1.7009352505071176e-05, "loss": 2.2843, "step": 8340 }, { "epoch": 0.45, "learning_rate": 1.7005762291992032e-05, "loss": 2.36, "step": 8350 }, { "epoch": 0.45, "learning_rate": 1.7002172078912884e-05, "loss": 2.2753, "step": 8360 }, { "epoch": 0.45, "learning_rate": 1.699858186583374e-05, "loss": 2.3235, "step": 8370 }, { "epoch": 0.45, "learning_rate": 1.6994991652754592e-05, "loss": 2.2682, "step": 8380 }, { "epoch": 0.45, "learning_rate": 1.6991401439675447e-05, "loss": 2.2015, "step": 8390 }, { "epoch": 0.45, "learning_rate": 1.69878112265963e-05, "loss": 2.1672, "step": 8400 }, { "epoch": 0.45, "learning_rate": 1.698422101351715e-05, "loss": 2.2932, "step": 8410 }, { "epoch": 0.45, "learning_rate": 1.6980630800438007e-05, "loss": 2.1829, "step": 8420 }, { "epoch": 0.45, "learning_rate": 1.6977040587358863e-05, "loss": 2.1634, "step": 8430 }, { "epoch": 0.45, "learning_rate": 1.6973450374279715e-05, "loss": 2.0578, "step": 8440 }, { "epoch": 0.46, "learning_rate": 1.696986016120057e-05, "loss": 2.2226, "step": 8450 }, { "epoch": 0.46, "learning_rate": 1.6966269948121423e-05, "loss": 2.3056, "step": 8460 }, { "epoch": 0.46, "learning_rate": 1.6962679735042275e-05, "loss": 2.2172, "step": 8470 }, { "epoch": 0.46, "learning_rate": 1.695908952196313e-05, "loss": 2.2584, "step": 8480 }, { "epoch": 0.46, "learning_rate": 1.6955499308883982e-05, "loss": 2.0855, "step": 8490 }, { "epoch": 0.46, "learning_rate": 1.6951909095804838e-05, "loss": 2.1562, "step": 8500 }, { "epoch": 0.46, "learning_rate": 1.694831888272569e-05, "loss": 2.1618, "step": 8510 }, { "epoch": 0.46, "learning_rate": 1.6944728669646546e-05, "loss": 2.2511, "step": 8520 }, { "epoch": 0.46, "learning_rate": 1.6941138456567398e-05, "loss": 2.1883, "step": 8530 }, { "epoch": 0.46, "learning_rate": 1.6937548243488253e-05, "loss": 2.1983, "step": 8540 }, { "epoch": 0.46, "learning_rate": 1.6933958030409105e-05, "loss": 2.3444, "step": 8550 }, { "epoch": 0.46, "learning_rate": 1.693036781732996e-05, "loss": 2.2075, "step": 8560 }, { "epoch": 0.46, "learning_rate": 1.6926777604250813e-05, "loss": 2.1159, "step": 8570 }, { "epoch": 0.46, "learning_rate": 1.692318739117167e-05, "loss": 2.2593, "step": 8580 }, { "epoch": 0.46, "learning_rate": 1.691959717809252e-05, "loss": 2.235, "step": 8590 }, { "epoch": 0.46, "learning_rate": 1.6916006965013373e-05, "loss": 2.2385, "step": 8600 }, { "epoch": 0.46, "learning_rate": 1.691241675193423e-05, "loss": 2.262, "step": 8610 }, { "epoch": 0.46, "learning_rate": 1.6908826538855084e-05, "loss": 2.3144, "step": 8620 }, { "epoch": 0.46, "learning_rate": 1.6905236325775936e-05, "loss": 2.1893, "step": 8630 }, { "epoch": 0.47, "learning_rate": 1.690164611269679e-05, "loss": 2.144, "step": 8640 }, { "epoch": 0.47, "learning_rate": 1.6898055899617644e-05, "loss": 2.196, "step": 8650 }, { "epoch": 0.47, "learning_rate": 1.6894465686538496e-05, "loss": 2.2927, "step": 8660 }, { "epoch": 0.47, "learning_rate": 1.689087547345935e-05, "loss": 2.2725, "step": 8670 }, { "epoch": 0.47, "learning_rate": 1.6887285260380204e-05, "loss": 2.2725, "step": 8680 }, { "epoch": 0.47, "learning_rate": 1.688369504730106e-05, "loss": 2.2988, "step": 8690 }, { "epoch": 0.47, "learning_rate": 1.688010483422191e-05, "loss": 2.1989, "step": 8700 }, { "epoch": 0.47, "learning_rate": 1.6876514621142767e-05, "loss": 2.1761, "step": 8710 }, { "epoch": 0.47, "learning_rate": 1.687292440806362e-05, "loss": 2.1727, "step": 8720 }, { "epoch": 0.47, "learning_rate": 1.6869334194984475e-05, "loss": 2.3115, "step": 8730 }, { "epoch": 0.47, "learning_rate": 1.6865743981905327e-05, "loss": 2.2573, "step": 8740 }, { "epoch": 0.47, "learning_rate": 1.6862153768826182e-05, "loss": 2.0392, "step": 8750 }, { "epoch": 0.47, "learning_rate": 1.6858563555747035e-05, "loss": 2.2875, "step": 8760 }, { "epoch": 0.47, "learning_rate": 1.685497334266789e-05, "loss": 2.2708, "step": 8770 }, { "epoch": 0.47, "learning_rate": 1.6851383129588742e-05, "loss": 2.1666, "step": 8780 }, { "epoch": 0.47, "learning_rate": 1.6847792916509595e-05, "loss": 2.2829, "step": 8790 }, { "epoch": 0.47, "learning_rate": 1.684420270343045e-05, "loss": 2.0808, "step": 8800 }, { "epoch": 0.47, "learning_rate": 1.6840612490351306e-05, "loss": 2.2349, "step": 8810 }, { "epoch": 0.47, "learning_rate": 1.6837022277272158e-05, "loss": 2.2609, "step": 8820 }, { "epoch": 0.48, "learning_rate": 1.683343206419301e-05, "loss": 2.0985, "step": 8830 }, { "epoch": 0.48, "learning_rate": 1.6829841851113865e-05, "loss": 2.3048, "step": 8840 }, { "epoch": 0.48, "learning_rate": 1.6826251638034718e-05, "loss": 2.255, "step": 8850 }, { "epoch": 0.48, "learning_rate": 1.6822661424955573e-05, "loss": 2.3968, "step": 8860 }, { "epoch": 0.48, "learning_rate": 1.6819071211876425e-05, "loss": 2.2403, "step": 8870 }, { "epoch": 0.48, "learning_rate": 1.681548099879728e-05, "loss": 2.1985, "step": 8880 }, { "epoch": 0.48, "learning_rate": 1.6811890785718133e-05, "loss": 2.0891, "step": 8890 }, { "epoch": 0.48, "learning_rate": 1.680830057263899e-05, "loss": 2.1553, "step": 8900 }, { "epoch": 0.48, "learning_rate": 1.680471035955984e-05, "loss": 2.2571, "step": 8910 }, { "epoch": 0.48, "learning_rate": 1.6801120146480696e-05, "loss": 2.3044, "step": 8920 }, { "epoch": 0.48, "learning_rate": 1.679752993340155e-05, "loss": 2.2925, "step": 8930 }, { "epoch": 0.48, "learning_rate": 1.6793939720322404e-05, "loss": 2.3496, "step": 8940 }, { "epoch": 0.48, "learning_rate": 1.6790349507243256e-05, "loss": 2.1817, "step": 8950 }, { "epoch": 0.48, "learning_rate": 1.678675929416411e-05, "loss": 2.3405, "step": 8960 }, { "epoch": 0.48, "learning_rate": 1.6783169081084964e-05, "loss": 2.3, "step": 8970 }, { "epoch": 0.48, "learning_rate": 1.6779578868005816e-05, "loss": 2.238, "step": 8980 }, { "epoch": 0.48, "learning_rate": 1.677598865492667e-05, "loss": 2.3269, "step": 8990 }, { "epoch": 0.48, "learning_rate": 1.6772398441847527e-05, "loss": 2.154, "step": 9000 }, { "epoch": 0.49, "learning_rate": 1.676880822876838e-05, "loss": 2.1695, "step": 9010 }, { "epoch": 0.49, "learning_rate": 1.676521801568923e-05, "loss": 2.3699, "step": 9020 }, { "epoch": 0.49, "learning_rate": 1.6761627802610087e-05, "loss": 2.2296, "step": 9030 }, { "epoch": 0.49, "learning_rate": 1.675803758953094e-05, "loss": 2.2475, "step": 9040 }, { "epoch": 0.49, "learning_rate": 1.6754447376451795e-05, "loss": 2.3329, "step": 9050 }, { "epoch": 0.49, "learning_rate": 1.6750857163372647e-05, "loss": 2.1067, "step": 9060 }, { "epoch": 0.49, "learning_rate": 1.6747266950293502e-05, "loss": 2.0579, "step": 9070 }, { "epoch": 0.49, "learning_rate": 1.6743676737214354e-05, "loss": 2.3367, "step": 9080 }, { "epoch": 0.49, "learning_rate": 1.674008652413521e-05, "loss": 2.2078, "step": 9090 }, { "epoch": 0.49, "learning_rate": 1.6736496311056062e-05, "loss": 2.1925, "step": 9100 }, { "epoch": 0.49, "learning_rate": 1.6732906097976914e-05, "loss": 2.3188, "step": 9110 }, { "epoch": 0.49, "learning_rate": 1.672931588489777e-05, "loss": 2.1495, "step": 9120 }, { "epoch": 0.49, "learning_rate": 1.6725725671818625e-05, "loss": 2.3221, "step": 9130 }, { "epoch": 0.49, "learning_rate": 1.6722135458739478e-05, "loss": 2.3298, "step": 9140 }, { "epoch": 0.49, "learning_rate": 1.671854524566033e-05, "loss": 2.2786, "step": 9150 }, { "epoch": 0.49, "learning_rate": 1.6714955032581185e-05, "loss": 2.1649, "step": 9160 }, { "epoch": 0.49, "learning_rate": 1.6711364819502037e-05, "loss": 2.1034, "step": 9170 }, { "epoch": 0.49, "learning_rate": 1.6707774606422893e-05, "loss": 2.2087, "step": 9180 }, { "epoch": 0.49, "learning_rate": 1.6704184393343745e-05, "loss": 2.2867, "step": 9190 }, { "epoch": 0.5, "learning_rate": 1.67005941802646e-05, "loss": 2.2975, "step": 9200 }, { "epoch": 0.5, "learning_rate": 1.6697003967185453e-05, "loss": 2.2516, "step": 9210 }, { "epoch": 0.5, "learning_rate": 1.669341375410631e-05, "loss": 2.1547, "step": 9220 }, { "epoch": 0.5, "learning_rate": 1.668982354102716e-05, "loss": 2.2671, "step": 9230 }, { "epoch": 0.5, "learning_rate": 1.6686233327948016e-05, "loss": 2.2782, "step": 9240 }, { "epoch": 0.5, "learning_rate": 1.6682643114868868e-05, "loss": 2.1954, "step": 9250 }, { "epoch": 0.5, "learning_rate": 1.6679052901789724e-05, "loss": 2.266, "step": 9260 }, { "epoch": 0.5, "learning_rate": 1.6675462688710576e-05, "loss": 2.1991, "step": 9270 }, { "epoch": 0.5, "learning_rate": 1.667187247563143e-05, "loss": 2.1759, "step": 9280 }, { "epoch": 0.5, "learning_rate": 1.6668282262552284e-05, "loss": 2.2107, "step": 9290 }, { "epoch": 0.5, "learning_rate": 1.6664692049473136e-05, "loss": 2.3168, "step": 9300 }, { "epoch": 0.5, "learning_rate": 1.666110183639399e-05, "loss": 2.306, "step": 9310 }, { "epoch": 0.5, "learning_rate": 1.6657511623314847e-05, "loss": 2.2466, "step": 9320 }, { "epoch": 0.5, "learning_rate": 1.66539214102357e-05, "loss": 2.192, "step": 9330 }, { "epoch": 0.5, "learning_rate": 1.665033119715655e-05, "loss": 2.2591, "step": 9340 }, { "epoch": 0.5, "learning_rate": 1.6646740984077407e-05, "loss": 2.231, "step": 9350 }, { "epoch": 0.5, "learning_rate": 1.664315077099826e-05, "loss": 2.1856, "step": 9360 }, { "epoch": 0.5, "learning_rate": 1.6639560557919114e-05, "loss": 2.2281, "step": 9370 }, { "epoch": 0.51, "learning_rate": 1.6635970344839967e-05, "loss": 2.2623, "step": 9380 }, { "epoch": 0.51, "learning_rate": 1.6632380131760822e-05, "loss": 2.2053, "step": 9390 }, { "epoch": 0.51, "learning_rate": 1.6628789918681674e-05, "loss": 2.2343, "step": 9400 }, { "epoch": 0.51, "learning_rate": 1.662519970560253e-05, "loss": 2.1216, "step": 9410 }, { "epoch": 0.51, "learning_rate": 1.6621609492523382e-05, "loss": 2.1708, "step": 9420 }, { "epoch": 0.51, "learning_rate": 1.6618019279444238e-05, "loss": 2.3139, "step": 9430 }, { "epoch": 0.51, "learning_rate": 1.661442906636509e-05, "loss": 2.0632, "step": 9440 }, { "epoch": 0.51, "learning_rate": 1.6610838853285945e-05, "loss": 2.1803, "step": 9450 }, { "epoch": 0.51, "learning_rate": 1.6607248640206797e-05, "loss": 2.0717, "step": 9460 }, { "epoch": 0.51, "learning_rate": 1.6603658427127653e-05, "loss": 2.2347, "step": 9470 }, { "epoch": 0.51, "learning_rate": 1.6600068214048505e-05, "loss": 2.2618, "step": 9480 }, { "epoch": 0.51, "learning_rate": 1.6596478000969357e-05, "loss": 2.3069, "step": 9490 }, { "epoch": 0.51, "learning_rate": 1.6592887787890213e-05, "loss": 2.2402, "step": 9500 }, { "epoch": 0.51, "learning_rate": 1.658929757481107e-05, "loss": 2.2092, "step": 9510 }, { "epoch": 0.51, "learning_rate": 1.658570736173192e-05, "loss": 2.2205, "step": 9520 }, { "epoch": 0.51, "learning_rate": 1.6582117148652773e-05, "loss": 2.1354, "step": 9530 }, { "epoch": 0.51, "learning_rate": 1.6578526935573628e-05, "loss": 2.2772, "step": 9540 }, { "epoch": 0.51, "learning_rate": 1.657493672249448e-05, "loss": 2.1454, "step": 9550 }, { "epoch": 0.51, "learning_rate": 1.6571346509415336e-05, "loss": 2.1655, "step": 9560 }, { "epoch": 0.52, "learning_rate": 1.6567756296336188e-05, "loss": 2.1073, "step": 9570 }, { "epoch": 0.52, "learning_rate": 1.6564166083257044e-05, "loss": 2.2159, "step": 9580 }, { "epoch": 0.52, "learning_rate": 1.6560575870177896e-05, "loss": 2.1162, "step": 9590 }, { "epoch": 0.52, "learning_rate": 1.655698565709875e-05, "loss": 2.2512, "step": 9600 }, { "epoch": 0.52, "learning_rate": 1.6553395444019603e-05, "loss": 2.1312, "step": 9610 }, { "epoch": 0.52, "learning_rate": 1.6549805230940456e-05, "loss": 2.0549, "step": 9620 }, { "epoch": 0.52, "learning_rate": 1.654621501786131e-05, "loss": 2.2253, "step": 9630 }, { "epoch": 0.52, "learning_rate": 1.6542624804782167e-05, "loss": 2.248, "step": 9640 }, { "epoch": 0.52, "learning_rate": 1.653903459170302e-05, "loss": 2.2771, "step": 9650 }, { "epoch": 0.52, "learning_rate": 1.6535444378623874e-05, "loss": 2.1893, "step": 9660 }, { "epoch": 0.52, "learning_rate": 1.6531854165544727e-05, "loss": 2.2424, "step": 9670 }, { "epoch": 0.52, "learning_rate": 1.652826395246558e-05, "loss": 2.0632, "step": 9680 }, { "epoch": 0.52, "learning_rate": 1.6524673739386434e-05, "loss": 2.2094, "step": 9690 }, { "epoch": 0.52, "learning_rate": 1.6521083526307286e-05, "loss": 2.1398, "step": 9700 }, { "epoch": 0.52, "learning_rate": 1.6517493313228142e-05, "loss": 2.1862, "step": 9710 }, { "epoch": 0.52, "learning_rate": 1.6513903100148994e-05, "loss": 2.2011, "step": 9720 }, { "epoch": 0.52, "learning_rate": 1.651031288706985e-05, "loss": 2.2074, "step": 9730 }, { "epoch": 0.52, "learning_rate": 1.6506722673990702e-05, "loss": 2.2983, "step": 9740 }, { "epoch": 0.53, "learning_rate": 1.6503132460911557e-05, "loss": 2.1001, "step": 9750 }, { "epoch": 0.53, "learning_rate": 1.649954224783241e-05, "loss": 2.2327, "step": 9760 }, { "epoch": 0.53, "learning_rate": 1.6495952034753265e-05, "loss": 2.2643, "step": 9770 }, { "epoch": 0.53, "learning_rate": 1.6492361821674117e-05, "loss": 2.2373, "step": 9780 }, { "epoch": 0.53, "learning_rate": 1.6488771608594973e-05, "loss": 2.2869, "step": 9790 }, { "epoch": 0.53, "learning_rate": 1.6485181395515825e-05, "loss": 2.162, "step": 9800 }, { "epoch": 0.53, "learning_rate": 1.6481591182436677e-05, "loss": 2.1644, "step": 9810 }, { "epoch": 0.53, "learning_rate": 1.6478000969357533e-05, "loss": 2.1258, "step": 9820 }, { "epoch": 0.53, "learning_rate": 1.6474410756278388e-05, "loss": 2.2174, "step": 9830 }, { "epoch": 0.53, "learning_rate": 1.647082054319924e-05, "loss": 2.2735, "step": 9840 }, { "epoch": 0.53, "learning_rate": 1.6467230330120093e-05, "loss": 2.1586, "step": 9850 }, { "epoch": 0.53, "learning_rate": 1.6463640117040948e-05, "loss": 2.1769, "step": 9860 }, { "epoch": 0.53, "learning_rate": 1.64600499039618e-05, "loss": 2.134, "step": 9870 }, { "epoch": 0.53, "learning_rate": 1.6456459690882656e-05, "loss": 2.1595, "step": 9880 }, { "epoch": 0.53, "learning_rate": 1.6452869477803508e-05, "loss": 2.207, "step": 9890 }, { "epoch": 0.53, "learning_rate": 1.6449279264724363e-05, "loss": 2.1611, "step": 9900 }, { "epoch": 0.53, "learning_rate": 1.6445689051645216e-05, "loss": 2.1584, "step": 9910 }, { "epoch": 0.53, "learning_rate": 1.644209883856607e-05, "loss": 2.1888, "step": 9920 }, { "epoch": 0.53, "learning_rate": 1.6438508625486923e-05, "loss": 2.2346, "step": 9930 }, { "epoch": 0.54, "learning_rate": 1.643491841240778e-05, "loss": 2.2044, "step": 9940 }, { "epoch": 0.54, "learning_rate": 1.643132819932863e-05, "loss": 2.2151, "step": 9950 }, { "epoch": 0.54, "learning_rate": 1.6427737986249487e-05, "loss": 2.1221, "step": 9960 }, { "epoch": 0.54, "learning_rate": 1.642414777317034e-05, "loss": 2.1223, "step": 9970 }, { "epoch": 0.54, "learning_rate": 1.6420557560091194e-05, "loss": 2.1505, "step": 9980 }, { "epoch": 0.54, "learning_rate": 1.6416967347012046e-05, "loss": 2.1762, "step": 9990 }, { "epoch": 0.54, "learning_rate": 1.64133771339329e-05, "loss": 2.1924, "step": 10000 }, { "epoch": 0.54, "learning_rate": 1.6409786920853754e-05, "loss": 2.0863, "step": 10010 }, { "epoch": 0.54, "learning_rate": 1.640619670777461e-05, "loss": 2.1743, "step": 10020 }, { "epoch": 0.54, "learning_rate": 1.6402606494695462e-05, "loss": 2.1435, "step": 10030 }, { "epoch": 0.54, "learning_rate": 1.6399016281616314e-05, "loss": 2.1934, "step": 10040 }, { "epoch": 0.54, "learning_rate": 1.639542606853717e-05, "loss": 2.2459, "step": 10050 }, { "epoch": 0.54, "learning_rate": 1.639183585545802e-05, "loss": 2.0843, "step": 10060 }, { "epoch": 0.54, "learning_rate": 1.6388245642378877e-05, "loss": 2.1792, "step": 10070 }, { "epoch": 0.54, "learning_rate": 1.638465542929973e-05, "loss": 2.0811, "step": 10080 }, { "epoch": 0.54, "learning_rate": 1.6381065216220585e-05, "loss": 2.1952, "step": 10090 }, { "epoch": 0.54, "learning_rate": 1.6377475003141437e-05, "loss": 2.1716, "step": 10100 }, { "epoch": 0.54, "learning_rate": 1.6373884790062293e-05, "loss": 2.203, "step": 10110 }, { "epoch": 0.54, "learning_rate": 1.6370294576983145e-05, "loss": 2.1518, "step": 10120 }, { "epoch": 0.55, "learning_rate": 1.6366704363903997e-05, "loss": 2.2372, "step": 10130 }, { "epoch": 0.55, "learning_rate": 1.6363114150824852e-05, "loss": 2.1836, "step": 10140 }, { "epoch": 0.55, "learning_rate": 1.6359523937745708e-05, "loss": 2.1725, "step": 10150 }, { "epoch": 0.55, "learning_rate": 1.635593372466656e-05, "loss": 2.1132, "step": 10160 }, { "epoch": 0.55, "learning_rate": 1.6352343511587416e-05, "loss": 2.1585, "step": 10170 }, { "epoch": 0.55, "learning_rate": 1.6348753298508268e-05, "loss": 2.2595, "step": 10180 }, { "epoch": 0.55, "learning_rate": 1.634516308542912e-05, "loss": 2.1541, "step": 10190 }, { "epoch": 0.55, "learning_rate": 1.6341572872349976e-05, "loss": 2.2525, "step": 10200 }, { "epoch": 0.55, "learning_rate": 1.633798265927083e-05, "loss": 2.156, "step": 10210 }, { "epoch": 0.55, "learning_rate": 1.6334392446191683e-05, "loss": 2.168, "step": 10220 }, { "epoch": 0.55, "learning_rate": 1.6330802233112535e-05, "loss": 2.1093, "step": 10230 }, { "epoch": 0.55, "learning_rate": 1.632721202003339e-05, "loss": 2.1389, "step": 10240 }, { "epoch": 0.55, "learning_rate": 1.6323621806954243e-05, "loss": 2.217, "step": 10250 }, { "epoch": 0.55, "learning_rate": 1.63200315938751e-05, "loss": 2.152, "step": 10260 }, { "epoch": 0.55, "learning_rate": 1.631644138079595e-05, "loss": 2.1542, "step": 10270 }, { "epoch": 0.55, "learning_rate": 1.6312851167716806e-05, "loss": 2.1434, "step": 10280 }, { "epoch": 0.55, "learning_rate": 1.630926095463766e-05, "loss": 2.1697, "step": 10290 }, { "epoch": 0.55, "learning_rate": 1.6305670741558514e-05, "loss": 2.1033, "step": 10300 }, { "epoch": 0.56, "learning_rate": 1.6302080528479366e-05, "loss": 2.1527, "step": 10310 }, { "epoch": 0.56, "learning_rate": 1.629849031540022e-05, "loss": 2.3524, "step": 10320 }, { "epoch": 0.56, "learning_rate": 1.6294900102321074e-05, "loss": 2.0593, "step": 10330 }, { "epoch": 0.56, "learning_rate": 1.629130988924193e-05, "loss": 2.1977, "step": 10340 }, { "epoch": 0.56, "learning_rate": 1.628771967616278e-05, "loss": 2.2231, "step": 10350 }, { "epoch": 0.56, "learning_rate": 1.6284129463083634e-05, "loss": 2.0381, "step": 10360 }, { "epoch": 0.56, "learning_rate": 1.628053925000449e-05, "loss": 2.1322, "step": 10370 }, { "epoch": 0.56, "learning_rate": 1.627694903692534e-05, "loss": 2.0896, "step": 10380 }, { "epoch": 0.56, "learning_rate": 1.6273358823846197e-05, "loss": 2.1107, "step": 10390 }, { "epoch": 0.56, "learning_rate": 1.626976861076705e-05, "loss": 2.0914, "step": 10400 }, { "epoch": 0.56, "learning_rate": 1.6266178397687905e-05, "loss": 2.0941, "step": 10410 }, { "epoch": 0.56, "learning_rate": 1.6262588184608757e-05, "loss": 2.073, "step": 10420 }, { "epoch": 0.56, "learning_rate": 1.6258997971529612e-05, "loss": 2.2026, "step": 10430 }, { "epoch": 0.56, "learning_rate": 1.6255407758450465e-05, "loss": 2.1048, "step": 10440 }, { "epoch": 0.56, "learning_rate": 1.625181754537132e-05, "loss": 2.2007, "step": 10450 }, { "epoch": 0.56, "learning_rate": 1.6248227332292172e-05, "loss": 2.3039, "step": 10460 }, { "epoch": 0.56, "learning_rate": 1.6244637119213028e-05, "loss": 2.2859, "step": 10470 }, { "epoch": 0.56, "learning_rate": 1.624104690613388e-05, "loss": 2.1524, "step": 10480 }, { "epoch": 0.56, "learning_rate": 1.6237456693054736e-05, "loss": 2.1549, "step": 10490 }, { "epoch": 0.57, "learning_rate": 1.6233866479975588e-05, "loss": 1.9481, "step": 10500 }, { "epoch": 0.57, "learning_rate": 1.623027626689644e-05, "loss": 2.1315, "step": 10510 }, { "epoch": 0.57, "learning_rate": 1.6226686053817295e-05, "loss": 2.1259, "step": 10520 }, { "epoch": 0.57, "learning_rate": 1.622309584073815e-05, "loss": 2.1968, "step": 10530 }, { "epoch": 0.57, "learning_rate": 1.6219505627659003e-05, "loss": 2.1678, "step": 10540 }, { "epoch": 0.57, "learning_rate": 1.6215915414579855e-05, "loss": 2.1433, "step": 10550 }, { "epoch": 0.57, "learning_rate": 1.621232520150071e-05, "loss": 2.2864, "step": 10560 }, { "epoch": 0.57, "learning_rate": 1.6208734988421563e-05, "loss": 2.1106, "step": 10570 }, { "epoch": 0.57, "learning_rate": 1.620514477534242e-05, "loss": 1.9924, "step": 10580 }, { "epoch": 0.57, "learning_rate": 1.620155456226327e-05, "loss": 2.1775, "step": 10590 }, { "epoch": 0.57, "learning_rate": 1.6197964349184126e-05, "loss": 2.1438, "step": 10600 }, { "epoch": 0.57, "learning_rate": 1.619437413610498e-05, "loss": 2.128, "step": 10610 }, { "epoch": 0.57, "learning_rate": 1.6190783923025834e-05, "loss": 2.1338, "step": 10620 }, { "epoch": 0.57, "learning_rate": 1.6187193709946686e-05, "loss": 2.1487, "step": 10630 }, { "epoch": 0.57, "learning_rate": 1.618360349686754e-05, "loss": 2.1434, "step": 10640 }, { "epoch": 0.57, "learning_rate": 1.6180013283788394e-05, "loss": 2.14, "step": 10650 }, { "epoch": 0.57, "learning_rate": 1.617642307070925e-05, "loss": 2.1477, "step": 10660 }, { "epoch": 0.57, "learning_rate": 1.61728328576301e-05, "loss": 2.0712, "step": 10670 }, { "epoch": 0.58, "learning_rate": 1.6169242644550957e-05, "loss": 2.2026, "step": 10680 }, { "epoch": 0.58, "learning_rate": 1.616565243147181e-05, "loss": 2.2072, "step": 10690 }, { "epoch": 0.58, "learning_rate": 1.616206221839266e-05, "loss": 2.1297, "step": 10700 }, { "epoch": 0.58, "learning_rate": 1.6158472005313517e-05, "loss": 2.1002, "step": 10710 }, { "epoch": 0.58, "learning_rate": 1.6154881792234372e-05, "loss": 2.183, "step": 10720 }, { "epoch": 0.58, "learning_rate": 1.6151291579155225e-05, "loss": 2.2539, "step": 10730 }, { "epoch": 0.58, "learning_rate": 1.6147701366076077e-05, "loss": 2.1698, "step": 10740 }, { "epoch": 0.58, "learning_rate": 1.6144111152996932e-05, "loss": 2.1343, "step": 10750 }, { "epoch": 0.58, "learning_rate": 1.6140520939917784e-05, "loss": 2.0151, "step": 10760 }, { "epoch": 0.58, "learning_rate": 1.613693072683864e-05, "loss": 2.1985, "step": 10770 }, { "epoch": 0.58, "learning_rate": 1.6133340513759492e-05, "loss": 2.0488, "step": 10780 }, { "epoch": 0.58, "learning_rate": 1.6129750300680348e-05, "loss": 2.1105, "step": 10790 }, { "epoch": 0.58, "learning_rate": 1.61261600876012e-05, "loss": 2.1872, "step": 10800 }, { "epoch": 0.58, "learning_rate": 1.6122569874522055e-05, "loss": 2.1352, "step": 10810 }, { "epoch": 0.58, "learning_rate": 1.6118979661442908e-05, "loss": 2.0997, "step": 10820 }, { "epoch": 0.58, "learning_rate": 1.611538944836376e-05, "loss": 2.2875, "step": 10830 }, { "epoch": 0.58, "learning_rate": 1.6111799235284615e-05, "loss": 2.206, "step": 10840 }, { "epoch": 0.58, "learning_rate": 1.610820902220547e-05, "loss": 2.2479, "step": 10850 }, { "epoch": 0.58, "learning_rate": 1.6104618809126323e-05, "loss": 2.0381, "step": 10860 }, { "epoch": 0.59, "learning_rate": 1.610102859604718e-05, "loss": 2.2111, "step": 10870 }, { "epoch": 0.59, "learning_rate": 1.609743838296803e-05, "loss": 2.2088, "step": 10880 }, { "epoch": 0.59, "learning_rate": 1.6093848169888883e-05, "loss": 2.2253, "step": 10890 }, { "epoch": 0.59, "learning_rate": 1.609025795680974e-05, "loss": 2.0664, "step": 10900 }, { "epoch": 0.59, "learning_rate": 1.608666774373059e-05, "loss": 2.0922, "step": 10910 }, { "epoch": 0.59, "learning_rate": 1.6083077530651446e-05, "loss": 2.0279, "step": 10920 }, { "epoch": 0.59, "learning_rate": 1.6079487317572298e-05, "loss": 2.164, "step": 10930 }, { "epoch": 0.59, "learning_rate": 1.6075897104493154e-05, "loss": 2.0937, "step": 10940 }, { "epoch": 0.59, "learning_rate": 1.6072306891414006e-05, "loss": 2.1731, "step": 10950 }, { "epoch": 0.59, "learning_rate": 1.606871667833486e-05, "loss": 2.1772, "step": 10960 }, { "epoch": 0.59, "learning_rate": 1.6065126465255714e-05, "loss": 2.1858, "step": 10970 }, { "epoch": 0.59, "learning_rate": 1.606153625217657e-05, "loss": 2.1996, "step": 10980 }, { "epoch": 0.59, "learning_rate": 1.605794603909742e-05, "loss": 2.0568, "step": 10990 }, { "epoch": 0.59, "learning_rate": 1.6054355826018277e-05, "loss": 2.2148, "step": 11000 }, { "epoch": 0.59, "learning_rate": 1.605076561293913e-05, "loss": 2.0837, "step": 11010 }, { "epoch": 0.59, "learning_rate": 1.604717539985998e-05, "loss": 2.2407, "step": 11020 }, { "epoch": 0.59, "learning_rate": 1.6043585186780837e-05, "loss": 2.1737, "step": 11030 }, { "epoch": 0.59, "learning_rate": 1.6039994973701692e-05, "loss": 2.1715, "step": 11040 }, { "epoch": 0.6, "learning_rate": 1.6036404760622544e-05, "loss": 2.1524, "step": 11050 }, { "epoch": 0.6, "learning_rate": 1.6032814547543397e-05, "loss": 2.1151, "step": 11060 }, { "epoch": 0.6, "learning_rate": 1.6029224334464252e-05, "loss": 1.9862, "step": 11070 }, { "epoch": 0.6, "learning_rate": 1.6025634121385104e-05, "loss": 2.174, "step": 11080 }, { "epoch": 0.6, "learning_rate": 1.602204390830596e-05, "loss": 2.1074, "step": 11090 }, { "epoch": 0.6, "learning_rate": 1.6018453695226812e-05, "loss": 1.9895, "step": 11100 }, { "epoch": 0.6, "learning_rate": 1.6014863482147668e-05, "loss": 2.1324, "step": 11110 }, { "epoch": 0.6, "learning_rate": 1.601127326906852e-05, "loss": 2.0449, "step": 11120 }, { "epoch": 0.6, "learning_rate": 1.6007683055989375e-05, "loss": 2.1068, "step": 11130 }, { "epoch": 0.6, "learning_rate": 1.6004092842910227e-05, "loss": 2.2069, "step": 11140 }, { "epoch": 0.6, "learning_rate": 1.6000502629831083e-05, "loss": 2.1625, "step": 11150 }, { "epoch": 0.6, "learning_rate": 1.5996912416751935e-05, "loss": 2.2651, "step": 11160 }, { "epoch": 0.6, "learning_rate": 1.599332220367279e-05, "loss": 2.2055, "step": 11170 }, { "epoch": 0.6, "learning_rate": 1.5989731990593643e-05, "loss": 2.0406, "step": 11180 }, { "epoch": 0.6, "learning_rate": 1.59861417775145e-05, "loss": 2.1785, "step": 11190 }, { "epoch": 0.6, "learning_rate": 1.598255156443535e-05, "loss": 2.211, "step": 11200 }, { "epoch": 0.6, "learning_rate": 1.5978961351356203e-05, "loss": 2.171, "step": 11210 }, { "epoch": 0.6, "learning_rate": 1.5975371138277058e-05, "loss": 2.0174, "step": 11220 }, { "epoch": 0.6, "learning_rate": 1.5971780925197914e-05, "loss": 2.1268, "step": 11230 }, { "epoch": 0.61, "learning_rate": 1.5968190712118766e-05, "loss": 2.1038, "step": 11240 }, { "epoch": 0.61, "learning_rate": 1.5964600499039618e-05, "loss": 2.1649, "step": 11250 }, { "epoch": 0.61, "learning_rate": 1.5961010285960474e-05, "loss": 2.0736, "step": 11260 }, { "epoch": 0.61, "learning_rate": 1.5957420072881326e-05, "loss": 2.0837, "step": 11270 }, { "epoch": 0.61, "learning_rate": 1.595382985980218e-05, "loss": 2.0133, "step": 11280 }, { "epoch": 0.61, "learning_rate": 1.5950239646723033e-05, "loss": 2.2018, "step": 11290 }, { "epoch": 0.61, "learning_rate": 1.594664943364389e-05, "loss": 2.1764, "step": 11300 }, { "epoch": 0.61, "learning_rate": 1.594305922056474e-05, "loss": 2.0341, "step": 11310 }, { "epoch": 0.61, "learning_rate": 1.5939469007485597e-05, "loss": 2.0903, "step": 11320 }, { "epoch": 0.61, "learning_rate": 1.593587879440645e-05, "loss": 1.9631, "step": 11330 }, { "epoch": 0.61, "learning_rate": 1.59322885813273e-05, "loss": 2.195, "step": 11340 }, { "epoch": 0.61, "learning_rate": 1.5928698368248157e-05, "loss": 2.2418, "step": 11350 }, { "epoch": 0.61, "learning_rate": 1.5925108155169012e-05, "loss": 2.1835, "step": 11360 }, { "epoch": 0.61, "learning_rate": 1.5921517942089864e-05, "loss": 2.1581, "step": 11370 }, { "epoch": 0.61, "learning_rate": 1.591792772901072e-05, "loss": 2.0861, "step": 11380 }, { "epoch": 0.61, "learning_rate": 1.5914337515931572e-05, "loss": 2.1837, "step": 11390 }, { "epoch": 0.61, "learning_rate": 1.5910747302852424e-05, "loss": 2.1113, "step": 11400 }, { "epoch": 0.61, "learning_rate": 1.590715708977328e-05, "loss": 2.0892, "step": 11410 }, { "epoch": 0.61, "learning_rate": 1.5903566876694132e-05, "loss": 2.1853, "step": 11420 }, { "epoch": 0.62, "learning_rate": 1.5899976663614987e-05, "loss": 2.1656, "step": 11430 }, { "epoch": 0.62, "learning_rate": 1.589638645053584e-05, "loss": 2.1389, "step": 11440 }, { "epoch": 0.62, "learning_rate": 1.5892796237456695e-05, "loss": 2.271, "step": 11450 }, { "epoch": 0.62, "learning_rate": 1.5889206024377547e-05, "loss": 2.1676, "step": 11460 }, { "epoch": 0.62, "learning_rate": 1.5885615811298403e-05, "loss": 2.1593, "step": 11470 }, { "epoch": 0.62, "learning_rate": 1.5882025598219255e-05, "loss": 2.1308, "step": 11480 }, { "epoch": 0.62, "learning_rate": 1.587843538514011e-05, "loss": 2.12, "step": 11490 }, { "epoch": 0.62, "learning_rate": 1.5874845172060963e-05, "loss": 2.0811, "step": 11500 }, { "epoch": 0.62, "learning_rate": 1.5871254958981818e-05, "loss": 2.1657, "step": 11510 }, { "epoch": 0.62, "learning_rate": 1.586766474590267e-05, "loss": 2.0265, "step": 11520 }, { "epoch": 0.62, "learning_rate": 1.5864074532823523e-05, "loss": 2.1235, "step": 11530 }, { "epoch": 0.62, "learning_rate": 1.5860484319744378e-05, "loss": 2.1472, "step": 11540 }, { "epoch": 0.62, "learning_rate": 1.5856894106665234e-05, "loss": 2.0441, "step": 11550 }, { "epoch": 0.62, "learning_rate": 1.5853303893586086e-05, "loss": 2.1707, "step": 11560 }, { "epoch": 0.62, "learning_rate": 1.5849713680506938e-05, "loss": 2.1224, "step": 11570 }, { "epoch": 0.62, "learning_rate": 1.5846123467427793e-05, "loss": 1.9841, "step": 11580 }, { "epoch": 0.62, "learning_rate": 1.5842533254348646e-05, "loss": 2.1761, "step": 11590 }, { "epoch": 0.62, "learning_rate": 1.58389430412695e-05, "loss": 2.1094, "step": 11600 }, { "epoch": 0.63, "learning_rate": 1.5835352828190353e-05, "loss": 2.0055, "step": 11610 }, { "epoch": 0.63, "learning_rate": 1.583176261511121e-05, "loss": 2.0425, "step": 11620 }, { "epoch": 0.63, "learning_rate": 1.582817240203206e-05, "loss": 2.1102, "step": 11630 }, { "epoch": 0.63, "learning_rate": 1.5824582188952917e-05, "loss": 2.1263, "step": 11640 }, { "epoch": 0.63, "learning_rate": 1.582099197587377e-05, "loss": 2.1232, "step": 11650 }, { "epoch": 0.63, "learning_rate": 1.5817401762794624e-05, "loss": 2.0697, "step": 11660 }, { "epoch": 0.63, "learning_rate": 1.5813811549715476e-05, "loss": 2.0262, "step": 11670 }, { "epoch": 0.63, "learning_rate": 1.5810221336636332e-05, "loss": 2.0511, "step": 11680 }, { "epoch": 0.63, "learning_rate": 1.5806631123557184e-05, "loss": 2.0816, "step": 11690 }, { "epoch": 0.63, "learning_rate": 1.580304091047804e-05, "loss": 2.0309, "step": 11700 }, { "epoch": 0.63, "learning_rate": 1.5799450697398892e-05, "loss": 2.208, "step": 11710 }, { "epoch": 0.63, "learning_rate": 1.5795860484319744e-05, "loss": 2.1646, "step": 11720 }, { "epoch": 0.63, "learning_rate": 1.57922702712406e-05, "loss": 2.1067, "step": 11730 }, { "epoch": 0.63, "learning_rate": 1.5788680058161455e-05, "loss": 2.1174, "step": 11740 }, { "epoch": 0.63, "learning_rate": 1.5785089845082307e-05, "loss": 2.1511, "step": 11750 }, { "epoch": 0.63, "learning_rate": 1.578149963200316e-05, "loss": 2.111, "step": 11760 }, { "epoch": 0.63, "learning_rate": 1.5777909418924015e-05, "loss": 2.1089, "step": 11770 }, { "epoch": 0.63, "learning_rate": 1.5774319205844867e-05, "loss": 2.0037, "step": 11780 }, { "epoch": 0.63, "learning_rate": 1.5770728992765723e-05, "loss": 2.1375, "step": 11790 }, { "epoch": 0.64, "learning_rate": 1.5767138779686575e-05, "loss": 2.069, "step": 11800 }, { "epoch": 0.64, "learning_rate": 1.576354856660743e-05, "loss": 2.1082, "step": 11810 }, { "epoch": 0.64, "learning_rate": 1.5759958353528282e-05, "loss": 2.2734, "step": 11820 }, { "epoch": 0.64, "learning_rate": 1.5756368140449138e-05, "loss": 2.1316, "step": 11830 }, { "epoch": 0.64, "learning_rate": 1.575277792736999e-05, "loss": 2.0294, "step": 11840 }, { "epoch": 0.64, "learning_rate": 1.5749187714290842e-05, "loss": 2.1099, "step": 11850 }, { "epoch": 0.64, "learning_rate": 1.5745597501211698e-05, "loss": 2.0493, "step": 11860 }, { "epoch": 0.64, "learning_rate": 1.5742007288132553e-05, "loss": 2.1372, "step": 11870 }, { "epoch": 0.64, "learning_rate": 1.5738417075053406e-05, "loss": 2.0806, "step": 11880 }, { "epoch": 0.64, "learning_rate": 1.573482686197426e-05, "loss": 2.0454, "step": 11890 }, { "epoch": 0.64, "learning_rate": 1.5731236648895113e-05, "loss": 2.0622, "step": 11900 }, { "epoch": 0.64, "learning_rate": 1.5727646435815965e-05, "loss": 2.1437, "step": 11910 }, { "epoch": 0.64, "learning_rate": 1.572405622273682e-05, "loss": 2.0992, "step": 11920 }, { "epoch": 0.64, "learning_rate": 1.5720466009657677e-05, "loss": 2.1536, "step": 11930 }, { "epoch": 0.64, "learning_rate": 1.571687579657853e-05, "loss": 2.0224, "step": 11940 }, { "epoch": 0.64, "learning_rate": 1.571328558349938e-05, "loss": 2.0274, "step": 11950 }, { "epoch": 0.64, "learning_rate": 1.5709695370420236e-05, "loss": 2.1237, "step": 11960 }, { "epoch": 0.64, "learning_rate": 1.570610515734109e-05, "loss": 2.1309, "step": 11970 }, { "epoch": 0.65, "learning_rate": 1.5702514944261944e-05, "loss": 2.036, "step": 11980 }, { "epoch": 0.65, "learning_rate": 1.5698924731182796e-05, "loss": 2.0992, "step": 11990 }, { "epoch": 0.65, "learning_rate": 1.5695334518103652e-05, "loss": 2.1507, "step": 12000 }, { "epoch": 0.65, "learning_rate": 1.5691744305024504e-05, "loss": 2.1355, "step": 12010 }, { "epoch": 0.65, "learning_rate": 1.568815409194536e-05, "loss": 2.1131, "step": 12020 }, { "epoch": 0.65, "learning_rate": 1.568456387886621e-05, "loss": 1.9487, "step": 12030 }, { "epoch": 0.65, "learning_rate": 1.5680973665787064e-05, "loss": 2.1121, "step": 12040 }, { "epoch": 0.65, "learning_rate": 1.567738345270792e-05, "loss": 2.0161, "step": 12050 }, { "epoch": 0.65, "learning_rate": 1.5673793239628775e-05, "loss": 2.0167, "step": 12060 }, { "epoch": 0.65, "learning_rate": 1.5670203026549627e-05, "loss": 2.0283, "step": 12070 }, { "epoch": 0.65, "learning_rate": 1.566661281347048e-05, "loss": 2.0438, "step": 12080 }, { "epoch": 0.65, "learning_rate": 1.5663022600391335e-05, "loss": 2.0318, "step": 12090 }, { "epoch": 0.65, "learning_rate": 1.5659432387312187e-05, "loss": 2.0962, "step": 12100 }, { "epoch": 0.65, "learning_rate": 1.5655842174233042e-05, "loss": 1.9641, "step": 12110 }, { "epoch": 0.65, "learning_rate": 1.5652251961153895e-05, "loss": 1.9245, "step": 12120 }, { "epoch": 0.65, "learning_rate": 1.564866174807475e-05, "loss": 2.1494, "step": 12130 }, { "epoch": 0.65, "learning_rate": 1.5645071534995602e-05, "loss": 2.1068, "step": 12140 }, { "epoch": 0.65, "learning_rate": 1.5641481321916458e-05, "loss": 2.1053, "step": 12150 }, { "epoch": 0.65, "learning_rate": 1.563789110883731e-05, "loss": 2.0999, "step": 12160 }, { "epoch": 0.66, "learning_rate": 1.5634300895758166e-05, "loss": 2.0399, "step": 12170 }, { "epoch": 0.66, "learning_rate": 1.5630710682679018e-05, "loss": 1.9359, "step": 12180 }, { "epoch": 0.66, "learning_rate": 1.5627120469599873e-05, "loss": 2.1474, "step": 12190 }, { "epoch": 0.66, "learning_rate": 1.5623530256520725e-05, "loss": 2.1611, "step": 12200 }, { "epoch": 0.66, "learning_rate": 1.561994004344158e-05, "loss": 2.0974, "step": 12210 }, { "epoch": 0.66, "learning_rate": 1.5616349830362433e-05, "loss": 2.0282, "step": 12220 }, { "epoch": 0.66, "learning_rate": 1.5612759617283285e-05, "loss": 2.1181, "step": 12230 }, { "epoch": 0.66, "learning_rate": 1.560916940420414e-05, "loss": 2.0439, "step": 12240 }, { "epoch": 0.66, "learning_rate": 1.5605579191124996e-05, "loss": 2.0225, "step": 12250 }, { "epoch": 0.66, "learning_rate": 1.560198897804585e-05, "loss": 2.0476, "step": 12260 }, { "epoch": 0.66, "learning_rate": 1.55983987649667e-05, "loss": 2.1115, "step": 12270 }, { "epoch": 0.66, "learning_rate": 1.5594808551887556e-05, "loss": 2.1125, "step": 12280 }, { "epoch": 0.66, "learning_rate": 1.559121833880841e-05, "loss": 2.0522, "step": 12290 }, { "epoch": 0.66, "learning_rate": 1.5587628125729264e-05, "loss": 2.1925, "step": 12300 }, { "epoch": 0.66, "learning_rate": 1.5584037912650116e-05, "loss": 2.1551, "step": 12310 }, { "epoch": 0.66, "learning_rate": 1.558044769957097e-05, "loss": 2.0679, "step": 12320 }, { "epoch": 0.66, "learning_rate": 1.5576857486491824e-05, "loss": 2.0695, "step": 12330 }, { "epoch": 0.66, "learning_rate": 1.557326727341268e-05, "loss": 2.1098, "step": 12340 }, { "epoch": 0.67, "learning_rate": 1.556967706033353e-05, "loss": 2.05, "step": 12350 }, { "epoch": 0.67, "learning_rate": 1.5566086847254387e-05, "loss": 2.1581, "step": 12360 }, { "epoch": 0.67, "learning_rate": 1.556249663417524e-05, "loss": 2.0917, "step": 12370 }, { "epoch": 0.67, "learning_rate": 1.5558906421096095e-05, "loss": 1.9694, "step": 12380 }, { "epoch": 0.67, "learning_rate": 1.5555316208016947e-05, "loss": 2.0252, "step": 12390 }, { "epoch": 0.67, "learning_rate": 1.5551725994937802e-05, "loss": 2.115, "step": 12400 }, { "epoch": 0.67, "learning_rate": 1.5548135781858655e-05, "loss": 2.2182, "step": 12410 }, { "epoch": 0.67, "learning_rate": 1.5544545568779507e-05, "loss": 1.9931, "step": 12420 }, { "epoch": 0.67, "learning_rate": 1.5540955355700362e-05, "loss": 2.0317, "step": 12430 }, { "epoch": 0.67, "learning_rate": 1.5537365142621218e-05, "loss": 2.0119, "step": 12440 }, { "epoch": 0.67, "learning_rate": 1.553377492954207e-05, "loss": 2.1493, "step": 12450 }, { "epoch": 0.67, "learning_rate": 1.5530184716462922e-05, "loss": 2.0019, "step": 12460 }, { "epoch": 0.67, "learning_rate": 1.5526594503383778e-05, "loss": 2.003, "step": 12470 }, { "epoch": 0.67, "learning_rate": 1.552300429030463e-05, "loss": 2.1463, "step": 12480 }, { "epoch": 0.67, "learning_rate": 1.5519414077225485e-05, "loss": 2.2, "step": 12490 }, { "epoch": 0.67, "learning_rate": 1.5515823864146338e-05, "loss": 2.0664, "step": 12500 }, { "epoch": 0.67, "learning_rate": 1.5512233651067193e-05, "loss": 1.9624, "step": 12510 }, { "epoch": 0.67, "learning_rate": 1.5508643437988045e-05, "loss": 2.1482, "step": 12520 }, { "epoch": 0.67, "learning_rate": 1.55050532249089e-05, "loss": 2.0371, "step": 12530 }, { "epoch": 0.68, "learning_rate": 1.5501463011829753e-05, "loss": 2.0824, "step": 12540 }, { "epoch": 0.68, "learning_rate": 1.5497872798750605e-05, "loss": 2.0686, "step": 12550 }, { "epoch": 0.68, "learning_rate": 1.549428258567146e-05, "loss": 2.0953, "step": 12560 }, { "epoch": 0.68, "learning_rate": 1.5490692372592316e-05, "loss": 2.1463, "step": 12570 }, { "epoch": 0.68, "learning_rate": 1.548710215951317e-05, "loss": 1.8882, "step": 12580 }, { "epoch": 0.68, "learning_rate": 1.5483511946434024e-05, "loss": 2.2288, "step": 12590 }, { "epoch": 0.68, "learning_rate": 1.5479921733354876e-05, "loss": 2.2213, "step": 12600 }, { "epoch": 0.68, "learning_rate": 1.5476331520275728e-05, "loss": 2.0162, "step": 12610 }, { "epoch": 0.68, "learning_rate": 1.54731003285045e-05, "loss": 1.9999, "step": 12620 }, { "epoch": 0.68, "learning_rate": 1.546951011542535e-05, "loss": 2.1581, "step": 12630 }, { "epoch": 0.68, "learning_rate": 1.5465919902346204e-05, "loss": 2.1017, "step": 12640 }, { "epoch": 0.68, "learning_rate": 1.546232968926706e-05, "loss": 2.1257, "step": 12650 }, { "epoch": 0.68, "learning_rate": 1.5458739476187915e-05, "loss": 2.0674, "step": 12660 }, { "epoch": 0.68, "learning_rate": 1.5455149263108767e-05, "loss": 2.0509, "step": 12670 }, { "epoch": 0.68, "learning_rate": 1.5451559050029622e-05, "loss": 1.9216, "step": 12680 }, { "epoch": 0.68, "learning_rate": 1.5447968836950474e-05, "loss": 2.1356, "step": 12690 }, { "epoch": 0.68, "learning_rate": 1.5444378623871327e-05, "loss": 2.024, "step": 12700 }, { "epoch": 0.68, "learning_rate": 1.5440788410792182e-05, "loss": 2.2308, "step": 12710 }, { "epoch": 0.68, "learning_rate": 1.5437198197713034e-05, "loss": 1.9717, "step": 12720 }, { "epoch": 0.69, "learning_rate": 1.543360798463389e-05, "loss": 2.0596, "step": 12730 }, { "epoch": 0.69, "learning_rate": 1.5430017771554745e-05, "loss": 2.0887, "step": 12740 }, { "epoch": 0.69, "learning_rate": 1.5426427558475598e-05, "loss": 2.1571, "step": 12750 }, { "epoch": 0.69, "learning_rate": 1.542283734539645e-05, "loss": 1.9607, "step": 12760 }, { "epoch": 0.69, "learning_rate": 1.5419247132317302e-05, "loss": 1.9414, "step": 12770 }, { "epoch": 0.69, "learning_rate": 1.5415656919238157e-05, "loss": 2.1463, "step": 12780 }, { "epoch": 0.69, "learning_rate": 1.5412066706159013e-05, "loss": 1.9298, "step": 12790 }, { "epoch": 0.69, "learning_rate": 1.5408476493079865e-05, "loss": 2.0287, "step": 12800 }, { "epoch": 0.69, "learning_rate": 1.540488628000072e-05, "loss": 2.0871, "step": 12810 }, { "epoch": 0.69, "learning_rate": 1.5401296066921573e-05, "loss": 2.0881, "step": 12820 }, { "epoch": 0.69, "learning_rate": 1.5397705853842425e-05, "loss": 2.1333, "step": 12830 }, { "epoch": 0.69, "learning_rate": 1.539411564076328e-05, "loss": 2.0392, "step": 12840 }, { "epoch": 0.69, "learning_rate": 1.5390525427684136e-05, "loss": 1.9655, "step": 12850 }, { "epoch": 0.69, "learning_rate": 1.5386935214604988e-05, "loss": 2.1195, "step": 12860 }, { "epoch": 0.69, "learning_rate": 1.5383345001525844e-05, "loss": 1.9787, "step": 12870 }, { "epoch": 0.69, "learning_rate": 1.5379754788446696e-05, "loss": 1.9632, "step": 12880 }, { "epoch": 0.69, "learning_rate": 1.5376164575367548e-05, "loss": 2.1126, "step": 12890 }, { "epoch": 0.69, "learning_rate": 1.5372574362288404e-05, "loss": 2.0399, "step": 12900 }, { "epoch": 0.7, "learning_rate": 1.5368984149209256e-05, "loss": 2.0546, "step": 12910 }, { "epoch": 0.7, "learning_rate": 1.536539393613011e-05, "loss": 2.1329, "step": 12920 }, { "epoch": 0.7, "learning_rate": 1.5361803723050967e-05, "loss": 2.0799, "step": 12930 }, { "epoch": 0.7, "learning_rate": 1.535821350997182e-05, "loss": 1.9909, "step": 12940 }, { "epoch": 0.7, "learning_rate": 1.535462329689267e-05, "loss": 2.0849, "step": 12950 }, { "epoch": 0.7, "learning_rate": 1.5351033083813523e-05, "loss": 2.1127, "step": 12960 }, { "epoch": 0.7, "learning_rate": 1.534744287073438e-05, "loss": 2.0281, "step": 12970 }, { "epoch": 0.7, "learning_rate": 1.5343852657655234e-05, "loss": 2.0845, "step": 12980 }, { "epoch": 0.7, "learning_rate": 1.5340262444576087e-05, "loss": 2.0396, "step": 12990 }, { "epoch": 0.7, "learning_rate": 1.5336672231496942e-05, "loss": 1.9809, "step": 13000 }, { "epoch": 0.7, "learning_rate": 1.5333082018417794e-05, "loss": 2.0528, "step": 13010 }, { "epoch": 0.7, "learning_rate": 1.5329491805338646e-05, "loss": 2.1109, "step": 13020 }, { "epoch": 0.7, "learning_rate": 1.5325901592259502e-05, "loss": 2.0509, "step": 13030 }, { "epoch": 0.7, "learning_rate": 1.5322311379180354e-05, "loss": 2.2027, "step": 13040 }, { "epoch": 0.7, "learning_rate": 1.531872116610121e-05, "loss": 1.9651, "step": 13050 }, { "epoch": 0.7, "learning_rate": 1.5315130953022065e-05, "loss": 2.1208, "step": 13060 }, { "epoch": 0.7, "learning_rate": 1.5311540739942917e-05, "loss": 1.9164, "step": 13070 }, { "epoch": 0.7, "learning_rate": 1.530795052686377e-05, "loss": 1.9972, "step": 13080 }, { "epoch": 0.7, "learning_rate": 1.5304360313784625e-05, "loss": 2.0601, "step": 13090 }, { "epoch": 0.71, "learning_rate": 1.5300770100705477e-05, "loss": 2.0548, "step": 13100 }, { "epoch": 0.71, "learning_rate": 1.5297179887626333e-05, "loss": 2.1394, "step": 13110 }, { "epoch": 0.71, "learning_rate": 1.5293589674547185e-05, "loss": 1.9684, "step": 13120 }, { "epoch": 0.71, "learning_rate": 1.528999946146804e-05, "loss": 2.1786, "step": 13130 }, { "epoch": 0.71, "learning_rate": 1.5286409248388893e-05, "loss": 2.0281, "step": 13140 }, { "epoch": 0.71, "learning_rate": 1.5282819035309745e-05, "loss": 2.0376, "step": 13150 }, { "epoch": 0.71, "learning_rate": 1.52792288222306e-05, "loss": 2.0923, "step": 13160 }, { "epoch": 0.71, "learning_rate": 1.5275638609151456e-05, "loss": 1.9774, "step": 13170 }, { "epoch": 0.71, "learning_rate": 1.5272048396072308e-05, "loss": 2.0755, "step": 13180 }, { "epoch": 0.71, "learning_rate": 1.5268458182993164e-05, "loss": 2.0806, "step": 13190 }, { "epoch": 0.71, "learning_rate": 1.5264867969914016e-05, "loss": 2.0583, "step": 13200 }, { "epoch": 0.71, "learning_rate": 1.5261277756834868e-05, "loss": 2.1806, "step": 13210 }, { "epoch": 0.71, "learning_rate": 1.5257687543755724e-05, "loss": 2.1308, "step": 13220 }, { "epoch": 0.71, "learning_rate": 1.5254097330676576e-05, "loss": 2.0235, "step": 13230 }, { "epoch": 0.71, "learning_rate": 1.5250507117597431e-05, "loss": 2.0108, "step": 13240 }, { "epoch": 0.71, "learning_rate": 1.5246916904518285e-05, "loss": 2.1065, "step": 13250 }, { "epoch": 0.71, "learning_rate": 1.5243326691439137e-05, "loss": 1.982, "step": 13260 }, { "epoch": 0.71, "learning_rate": 1.5239736478359993e-05, "loss": 2.0293, "step": 13270 }, { "epoch": 0.72, "learning_rate": 1.5236146265280847e-05, "loss": 2.088, "step": 13280 }, { "epoch": 0.72, "learning_rate": 1.5232556052201699e-05, "loss": 1.9779, "step": 13290 }, { "epoch": 0.72, "learning_rate": 1.5228965839122554e-05, "loss": 2.0171, "step": 13300 }, { "epoch": 0.72, "learning_rate": 1.5225375626043406e-05, "loss": 2.0627, "step": 13310 }, { "epoch": 0.72, "learning_rate": 1.522178541296426e-05, "loss": 2.1406, "step": 13320 }, { "epoch": 0.72, "learning_rate": 1.5218195199885116e-05, "loss": 2.0995, "step": 13330 }, { "epoch": 0.72, "learning_rate": 1.5214604986805968e-05, "loss": 2.1292, "step": 13340 }, { "epoch": 0.72, "learning_rate": 1.5211014773726822e-05, "loss": 1.9974, "step": 13350 }, { "epoch": 0.72, "learning_rate": 1.5207424560647677e-05, "loss": 2.0344, "step": 13360 }, { "epoch": 0.72, "learning_rate": 1.520383434756853e-05, "loss": 2.1315, "step": 13370 }, { "epoch": 0.72, "learning_rate": 1.5200244134489383e-05, "loss": 2.0566, "step": 13380 }, { "epoch": 0.72, "learning_rate": 1.5196653921410236e-05, "loss": 2.0072, "step": 13390 }, { "epoch": 0.72, "learning_rate": 1.5193063708331091e-05, "loss": 2.0746, "step": 13400 }, { "epoch": 0.72, "learning_rate": 1.5189473495251945e-05, "loss": 1.8783, "step": 13410 }, { "epoch": 0.72, "learning_rate": 1.5185883282172797e-05, "loss": 2.0765, "step": 13420 }, { "epoch": 0.72, "learning_rate": 1.5182293069093653e-05, "loss": 2.0715, "step": 13430 }, { "epoch": 0.72, "learning_rate": 1.5178702856014507e-05, "loss": 2.0494, "step": 13440 }, { "epoch": 0.72, "learning_rate": 1.5175112642935359e-05, "loss": 2.0454, "step": 13450 }, { "epoch": 0.72, "learning_rate": 1.5171522429856214e-05, "loss": 2.0833, "step": 13460 }, { "epoch": 0.73, "learning_rate": 1.5167932216777066e-05, "loss": 2.1092, "step": 13470 }, { "epoch": 0.73, "learning_rate": 1.516434200369792e-05, "loss": 2.1402, "step": 13480 }, { "epoch": 0.73, "learning_rate": 1.5160751790618776e-05, "loss": 2.0447, "step": 13490 }, { "epoch": 0.73, "learning_rate": 1.5157161577539628e-05, "loss": 1.8994, "step": 13500 }, { "epoch": 0.73, "learning_rate": 1.5153571364460482e-05, "loss": 2.0731, "step": 13510 }, { "epoch": 0.73, "learning_rate": 1.5149981151381337e-05, "loss": 2.1427, "step": 13520 }, { "epoch": 0.73, "learning_rate": 1.514639093830219e-05, "loss": 2.0815, "step": 13530 }, { "epoch": 0.73, "learning_rate": 1.5142800725223043e-05, "loss": 2.0199, "step": 13540 }, { "epoch": 0.73, "learning_rate": 1.5139210512143895e-05, "loss": 1.9998, "step": 13550 }, { "epoch": 0.73, "learning_rate": 1.5135620299064751e-05, "loss": 2.0558, "step": 13560 }, { "epoch": 0.73, "learning_rate": 1.5132030085985605e-05, "loss": 2.0485, "step": 13570 }, { "epoch": 0.73, "learning_rate": 1.5128439872906457e-05, "loss": 2.0875, "step": 13580 }, { "epoch": 0.73, "learning_rate": 1.5124849659827313e-05, "loss": 2.0702, "step": 13590 }, { "epoch": 0.73, "learning_rate": 1.5121259446748166e-05, "loss": 2.0934, "step": 13600 }, { "epoch": 0.73, "learning_rate": 1.5117669233669019e-05, "loss": 2.095, "step": 13610 }, { "epoch": 0.73, "learning_rate": 1.5114079020589874e-05, "loss": 2.0092, "step": 13620 }, { "epoch": 0.73, "learning_rate": 1.5110488807510726e-05, "loss": 2.0886, "step": 13630 }, { "epoch": 0.73, "learning_rate": 1.510689859443158e-05, "loss": 2.2902, "step": 13640 }, { "epoch": 0.74, "learning_rate": 1.5103308381352436e-05, "loss": 2.0329, "step": 13650 }, { "epoch": 0.74, "learning_rate": 1.5099718168273288e-05, "loss": 2.0693, "step": 13660 }, { "epoch": 0.74, "learning_rate": 1.5096127955194142e-05, "loss": 2.0849, "step": 13670 }, { "epoch": 0.74, "learning_rate": 1.5092537742114997e-05, "loss": 2.0251, "step": 13680 }, { "epoch": 0.74, "learning_rate": 1.508894752903585e-05, "loss": 2.1715, "step": 13690 }, { "epoch": 0.74, "learning_rate": 1.5085357315956703e-05, "loss": 2.0221, "step": 13700 }, { "epoch": 0.74, "learning_rate": 1.5081767102877559e-05, "loss": 2.1422, "step": 13710 }, { "epoch": 0.74, "learning_rate": 1.5078176889798411e-05, "loss": 2.042, "step": 13720 }, { "epoch": 0.74, "learning_rate": 1.5074586676719265e-05, "loss": 2.06, "step": 13730 }, { "epoch": 0.74, "learning_rate": 1.5070996463640117e-05, "loss": 2.0753, "step": 13740 }, { "epoch": 0.74, "learning_rate": 1.5067406250560973e-05, "loss": 1.9367, "step": 13750 }, { "epoch": 0.74, "learning_rate": 1.5063816037481826e-05, "loss": 2.1012, "step": 13760 }, { "epoch": 0.74, "learning_rate": 1.5060225824402679e-05, "loss": 2.052, "step": 13770 }, { "epoch": 0.74, "learning_rate": 1.5056635611323534e-05, "loss": 2.0435, "step": 13780 }, { "epoch": 0.74, "learning_rate": 1.5053045398244388e-05, "loss": 2.0605, "step": 13790 }, { "epoch": 0.74, "learning_rate": 1.504945518516524e-05, "loss": 1.9417, "step": 13800 }, { "epoch": 0.74, "learning_rate": 1.5045864972086096e-05, "loss": 2.0748, "step": 13810 }, { "epoch": 0.74, "learning_rate": 1.5042274759006948e-05, "loss": 2.127, "step": 13820 }, { "epoch": 0.74, "learning_rate": 1.5038684545927802e-05, "loss": 2.0128, "step": 13830 }, { "epoch": 0.75, "learning_rate": 1.5035094332848657e-05, "loss": 1.896, "step": 13840 }, { "epoch": 0.75, "learning_rate": 1.503150411976951e-05, "loss": 2.0379, "step": 13850 }, { "epoch": 0.75, "learning_rate": 1.5027913906690363e-05, "loss": 2.1346, "step": 13860 }, { "epoch": 0.75, "learning_rate": 1.5024323693611219e-05, "loss": 2.1245, "step": 13870 }, { "epoch": 0.75, "learning_rate": 1.5020733480532071e-05, "loss": 2.1242, "step": 13880 }, { "epoch": 0.75, "learning_rate": 1.5017143267452925e-05, "loss": 2.0489, "step": 13890 }, { "epoch": 0.75, "learning_rate": 1.5013553054373777e-05, "loss": 2.1228, "step": 13900 }, { "epoch": 0.75, "learning_rate": 1.5009962841294632e-05, "loss": 1.9814, "step": 13910 }, { "epoch": 0.75, "learning_rate": 1.5006372628215486e-05, "loss": 2.0464, "step": 13920 }, { "epoch": 0.75, "learning_rate": 1.5002782415136338e-05, "loss": 2.0132, "step": 13930 }, { "epoch": 0.75, "learning_rate": 1.4999192202057194e-05, "loss": 2.1623, "step": 13940 }, { "epoch": 0.75, "learning_rate": 1.4995601988978048e-05, "loss": 2.0603, "step": 13950 }, { "epoch": 0.75, "learning_rate": 1.49920117758989e-05, "loss": 1.9995, "step": 13960 }, { "epoch": 0.75, "learning_rate": 1.4988421562819756e-05, "loss": 2.0911, "step": 13970 }, { "epoch": 0.75, "learning_rate": 1.4984831349740608e-05, "loss": 2.1758, "step": 13980 }, { "epoch": 0.75, "learning_rate": 1.4981241136661462e-05, "loss": 2.0097, "step": 13990 }, { "epoch": 0.75, "learning_rate": 1.4977650923582317e-05, "loss": 2.0739, "step": 14000 }, { "epoch": 0.75, "learning_rate": 1.497406071050317e-05, "loss": 1.9435, "step": 14010 }, { "epoch": 0.75, "learning_rate": 1.4970470497424023e-05, "loss": 1.8872, "step": 14020 }, { "epoch": 0.76, "learning_rate": 1.4966880284344879e-05, "loss": 2.1015, "step": 14030 }, { "epoch": 0.76, "learning_rate": 1.496329007126573e-05, "loss": 2.0666, "step": 14040 }, { "epoch": 0.76, "learning_rate": 1.4959699858186585e-05, "loss": 2.0813, "step": 14050 }, { "epoch": 0.76, "learning_rate": 1.495610964510744e-05, "loss": 1.9736, "step": 14060 }, { "epoch": 0.76, "learning_rate": 1.4952519432028292e-05, "loss": 2.1561, "step": 14070 }, { "epoch": 0.76, "learning_rate": 1.4948929218949146e-05, "loss": 2.0205, "step": 14080 }, { "epoch": 0.76, "learning_rate": 1.4945339005869998e-05, "loss": 2.0563, "step": 14090 }, { "epoch": 0.76, "learning_rate": 1.4941748792790854e-05, "loss": 2.0275, "step": 14100 }, { "epoch": 0.76, "learning_rate": 1.4938158579711708e-05, "loss": 1.931, "step": 14110 }, { "epoch": 0.76, "learning_rate": 1.493456836663256e-05, "loss": 2.2797, "step": 14120 }, { "epoch": 0.76, "learning_rate": 1.4930978153553415e-05, "loss": 2.0612, "step": 14130 }, { "epoch": 0.76, "learning_rate": 1.492738794047427e-05, "loss": 1.9289, "step": 14140 }, { "epoch": 0.76, "learning_rate": 1.4923797727395121e-05, "loss": 2.0389, "step": 14150 }, { "epoch": 0.76, "learning_rate": 1.4920207514315977e-05, "loss": 2.1303, "step": 14160 }, { "epoch": 0.76, "learning_rate": 1.491661730123683e-05, "loss": 2.0961, "step": 14170 }, { "epoch": 0.76, "learning_rate": 1.4913027088157683e-05, "loss": 1.9737, "step": 14180 }, { "epoch": 0.76, "learning_rate": 1.4909436875078539e-05, "loss": 2.1415, "step": 14190 }, { "epoch": 0.76, "learning_rate": 1.490584666199939e-05, "loss": 1.9394, "step": 14200 }, { "epoch": 0.77, "learning_rate": 1.4902256448920245e-05, "loss": 2.0567, "step": 14210 }, { "epoch": 0.77, "learning_rate": 1.48986662358411e-05, "loss": 1.9274, "step": 14220 }, { "epoch": 0.77, "learning_rate": 1.4895076022761952e-05, "loss": 1.8914, "step": 14230 }, { "epoch": 0.77, "learning_rate": 1.4891485809682806e-05, "loss": 2.092, "step": 14240 }, { "epoch": 0.77, "learning_rate": 1.4887895596603658e-05, "loss": 1.9887, "step": 14250 }, { "epoch": 0.77, "learning_rate": 1.4884305383524514e-05, "loss": 2.0835, "step": 14260 }, { "epoch": 0.77, "learning_rate": 1.4880715170445368e-05, "loss": 2.123, "step": 14270 }, { "epoch": 0.77, "learning_rate": 1.487712495736622e-05, "loss": 2.0234, "step": 14280 }, { "epoch": 0.77, "learning_rate": 1.4873534744287075e-05, "loss": 2.1607, "step": 14290 }, { "epoch": 0.77, "learning_rate": 1.486994453120793e-05, "loss": 2.028, "step": 14300 }, { "epoch": 0.77, "learning_rate": 1.4866354318128781e-05, "loss": 1.939, "step": 14310 }, { "epoch": 0.77, "learning_rate": 1.4862764105049637e-05, "loss": 2.238, "step": 14320 }, { "epoch": 0.77, "learning_rate": 1.4859173891970489e-05, "loss": 2.1677, "step": 14330 }, { "epoch": 0.77, "learning_rate": 1.4855583678891343e-05, "loss": 1.9644, "step": 14340 }, { "epoch": 0.77, "learning_rate": 1.4851993465812198e-05, "loss": 2.0189, "step": 14350 }, { "epoch": 0.77, "learning_rate": 1.484840325273305e-05, "loss": 2.0919, "step": 14360 }, { "epoch": 0.77, "learning_rate": 1.4844813039653904e-05, "loss": 1.9994, "step": 14370 }, { "epoch": 0.77, "learning_rate": 1.484122282657476e-05, "loss": 1.9655, "step": 14380 }, { "epoch": 0.77, "learning_rate": 1.4837632613495612e-05, "loss": 1.9701, "step": 14390 }, { "epoch": 0.78, "learning_rate": 1.4834042400416466e-05, "loss": 2.0238, "step": 14400 }, { "epoch": 0.78, "learning_rate": 1.4830452187337318e-05, "loss": 1.9132, "step": 14410 }, { "epoch": 0.78, "learning_rate": 1.4826861974258174e-05, "loss": 2.0609, "step": 14420 }, { "epoch": 0.78, "learning_rate": 1.4823271761179028e-05, "loss": 2.0056, "step": 14430 }, { "epoch": 0.78, "learning_rate": 1.481968154809988e-05, "loss": 1.9965, "step": 14440 }, { "epoch": 0.78, "learning_rate": 1.4816091335020735e-05, "loss": 2.0528, "step": 14450 }, { "epoch": 0.78, "learning_rate": 1.4812501121941589e-05, "loss": 2.0794, "step": 14460 }, { "epoch": 0.78, "learning_rate": 1.4808910908862441e-05, "loss": 1.9269, "step": 14470 }, { "epoch": 0.78, "learning_rate": 1.4805320695783297e-05, "loss": 2.0307, "step": 14480 }, { "epoch": 0.78, "learning_rate": 1.480173048270415e-05, "loss": 2.013, "step": 14490 }, { "epoch": 0.78, "learning_rate": 1.4798140269625003e-05, "loss": 1.9453, "step": 14500 }, { "epoch": 0.78, "learning_rate": 1.4794550056545858e-05, "loss": 2.0528, "step": 14510 }, { "epoch": 0.78, "learning_rate": 1.479095984346671e-05, "loss": 2.0623, "step": 14520 }, { "epoch": 0.78, "learning_rate": 1.4787369630387564e-05, "loss": 1.9508, "step": 14530 }, { "epoch": 0.78, "learning_rate": 1.478377941730842e-05, "loss": 2.1089, "step": 14540 }, { "epoch": 0.78, "learning_rate": 1.4780189204229272e-05, "loss": 2.1795, "step": 14550 }, { "epoch": 0.78, "learning_rate": 1.4776598991150126e-05, "loss": 1.9582, "step": 14560 }, { "epoch": 0.78, "learning_rate": 1.4773008778070981e-05, "loss": 2.0255, "step": 14570 }, { "epoch": 0.79, "learning_rate": 1.4769418564991834e-05, "loss": 1.9702, "step": 14580 }, { "epoch": 0.79, "learning_rate": 1.4765828351912688e-05, "loss": 2.0792, "step": 14590 }, { "epoch": 0.79, "learning_rate": 1.476223813883354e-05, "loss": 2.1122, "step": 14600 }, { "epoch": 0.79, "learning_rate": 1.4758647925754395e-05, "loss": 2.0807, "step": 14610 }, { "epoch": 0.79, "learning_rate": 1.4755057712675249e-05, "loss": 1.9668, "step": 14620 }, { "epoch": 0.79, "learning_rate": 1.4751467499596101e-05, "loss": 1.9507, "step": 14630 }, { "epoch": 0.79, "learning_rate": 1.4747877286516957e-05, "loss": 1.9836, "step": 14640 }, { "epoch": 0.79, "learning_rate": 1.474428707343781e-05, "loss": 2.1809, "step": 14650 }, { "epoch": 0.79, "learning_rate": 1.4740696860358663e-05, "loss": 2.1095, "step": 14660 }, { "epoch": 0.79, "learning_rate": 1.4737106647279518e-05, "loss": 2.1119, "step": 14670 }, { "epoch": 0.79, "learning_rate": 1.473351643420037e-05, "loss": 1.9775, "step": 14680 }, { "epoch": 0.79, "learning_rate": 1.4729926221121224e-05, "loss": 1.9746, "step": 14690 }, { "epoch": 0.79, "learning_rate": 1.472633600804208e-05, "loss": 2.1225, "step": 14700 }, { "epoch": 0.79, "learning_rate": 1.4722745794962932e-05, "loss": 2.0044, "step": 14710 }, { "epoch": 0.79, "learning_rate": 1.4719155581883786e-05, "loss": 2.0326, "step": 14720 }, { "epoch": 0.79, "learning_rate": 1.4715565368804641e-05, "loss": 1.991, "step": 14730 }, { "epoch": 0.79, "learning_rate": 1.4711975155725494e-05, "loss": 1.9945, "step": 14740 }, { "epoch": 0.79, "learning_rate": 1.4708384942646347e-05, "loss": 2.0258, "step": 14750 }, { "epoch": 0.79, "learning_rate": 1.47047947295672e-05, "loss": 2.0528, "step": 14760 }, { "epoch": 0.8, "learning_rate": 1.4701204516488055e-05, "loss": 2.0156, "step": 14770 }, { "epoch": 0.8, "learning_rate": 1.4697614303408909e-05, "loss": 2.1252, "step": 14780 }, { "epoch": 0.8, "learning_rate": 1.4694024090329761e-05, "loss": 2.0191, "step": 14790 }, { "epoch": 0.8, "learning_rate": 1.4690433877250617e-05, "loss": 2.0197, "step": 14800 }, { "epoch": 0.8, "learning_rate": 1.468684366417147e-05, "loss": 2.1201, "step": 14810 }, { "epoch": 0.8, "learning_rate": 1.4683253451092323e-05, "loss": 1.9993, "step": 14820 }, { "epoch": 0.8, "learning_rate": 1.4679663238013178e-05, "loss": 2.0877, "step": 14830 }, { "epoch": 0.8, "learning_rate": 1.467607302493403e-05, "loss": 2.018, "step": 14840 }, { "epoch": 0.8, "learning_rate": 1.4672482811854884e-05, "loss": 1.9555, "step": 14850 }, { "epoch": 0.8, "learning_rate": 1.466889259877574e-05, "loss": 1.9872, "step": 14860 }, { "epoch": 0.8, "learning_rate": 1.4665302385696592e-05, "loss": 2.1174, "step": 14870 }, { "epoch": 0.8, "learning_rate": 1.4661712172617446e-05, "loss": 1.9459, "step": 14880 }, { "epoch": 0.8, "learning_rate": 1.4658121959538301e-05, "loss": 2.1038, "step": 14890 }, { "epoch": 0.8, "learning_rate": 1.4654531746459153e-05, "loss": 1.9867, "step": 14900 }, { "epoch": 0.8, "learning_rate": 1.4650941533380007e-05, "loss": 1.997, "step": 14910 }, { "epoch": 0.8, "learning_rate": 1.4647351320300863e-05, "loss": 2.0763, "step": 14920 }, { "epoch": 0.8, "learning_rate": 1.4643761107221715e-05, "loss": 1.968, "step": 14930 }, { "epoch": 0.8, "learning_rate": 1.4640170894142569e-05, "loss": 1.98, "step": 14940 }, { "epoch": 0.81, "learning_rate": 1.4636580681063421e-05, "loss": 2.0142, "step": 14950 }, { "epoch": 0.81, "learning_rate": 1.4632990467984277e-05, "loss": 1.9931, "step": 14960 }, { "epoch": 0.81, "learning_rate": 1.462940025490513e-05, "loss": 1.9904, "step": 14970 }, { "epoch": 0.81, "learning_rate": 1.4625810041825983e-05, "loss": 1.9204, "step": 14980 }, { "epoch": 0.81, "learning_rate": 1.4622219828746838e-05, "loss": 2.0475, "step": 14990 }, { "epoch": 0.81, "learning_rate": 1.4618629615667692e-05, "loss": 1.8839, "step": 15000 }, { "epoch": 0.81, "learning_rate": 1.4615398423896458e-05, "loss": 2.0259, "step": 15010 }, { "epoch": 0.81, "learning_rate": 1.4611808210817313e-05, "loss": 2.0052, "step": 15020 }, { "epoch": 0.81, "learning_rate": 1.4608217997738167e-05, "loss": 2.0268, "step": 15030 }, { "epoch": 0.81, "learning_rate": 1.460462778465902e-05, "loss": 2.1336, "step": 15040 }, { "epoch": 0.81, "learning_rate": 1.4601037571579875e-05, "loss": 1.9692, "step": 15050 }, { "epoch": 0.81, "learning_rate": 1.4597447358500729e-05, "loss": 1.9419, "step": 15060 }, { "epoch": 0.81, "learning_rate": 1.4593857145421581e-05, "loss": 2.0268, "step": 15070 }, { "epoch": 0.81, "learning_rate": 1.4590266932342437e-05, "loss": 1.938, "step": 15080 }, { "epoch": 0.81, "learning_rate": 1.4586676719263289e-05, "loss": 1.9639, "step": 15090 }, { "epoch": 0.81, "learning_rate": 1.4583086506184143e-05, "loss": 2.0612, "step": 15100 }, { "epoch": 0.81, "learning_rate": 1.4579496293104998e-05, "loss": 1.9251, "step": 15110 }, { "epoch": 0.81, "learning_rate": 1.457590608002585e-05, "loss": 1.8879, "step": 15120 }, { "epoch": 0.81, "learning_rate": 1.4572315866946704e-05, "loss": 2.0217, "step": 15130 }, { "epoch": 0.82, "learning_rate": 1.456872565386756e-05, "loss": 2.0998, "step": 15140 }, { "epoch": 0.82, "learning_rate": 1.4565135440788412e-05, "loss": 2.068, "step": 15150 }, { "epoch": 0.82, "learning_rate": 1.4561545227709266e-05, "loss": 2.0515, "step": 15160 }, { "epoch": 0.82, "learning_rate": 1.4557955014630118e-05, "loss": 2.1081, "step": 15170 }, { "epoch": 0.82, "learning_rate": 1.4554364801550973e-05, "loss": 2.0565, "step": 15180 }, { "epoch": 0.82, "learning_rate": 1.4550774588471827e-05, "loss": 2.0125, "step": 15190 }, { "epoch": 0.82, "learning_rate": 1.454718437539268e-05, "loss": 1.9999, "step": 15200 }, { "epoch": 0.82, "learning_rate": 1.4543594162313535e-05, "loss": 2.0545, "step": 15210 }, { "epoch": 0.82, "learning_rate": 1.4540003949234389e-05, "loss": 2.0233, "step": 15220 }, { "epoch": 0.82, "learning_rate": 1.4536413736155241e-05, "loss": 1.9643, "step": 15230 }, { "epoch": 0.82, "learning_rate": 1.4532823523076096e-05, "loss": 1.9595, "step": 15240 }, { "epoch": 0.82, "learning_rate": 1.4529233309996949e-05, "loss": 2.1415, "step": 15250 }, { "epoch": 0.82, "learning_rate": 1.4525643096917802e-05, "loss": 1.9609, "step": 15260 }, { "epoch": 0.82, "learning_rate": 1.4522052883838658e-05, "loss": 2.0199, "step": 15270 }, { "epoch": 0.82, "learning_rate": 1.451846267075951e-05, "loss": 2.1178, "step": 15280 }, { "epoch": 0.82, "learning_rate": 1.4514872457680364e-05, "loss": 2.0005, "step": 15290 }, { "epoch": 0.82, "learning_rate": 1.451128224460122e-05, "loss": 2.0603, "step": 15300 }, { "epoch": 0.82, "learning_rate": 1.4507692031522072e-05, "loss": 1.9754, "step": 15310 }, { "epoch": 0.82, "learning_rate": 1.4504101818442926e-05, "loss": 2.1138, "step": 15320 }, { "epoch": 0.83, "learning_rate": 1.4500511605363778e-05, "loss": 2.1264, "step": 15330 }, { "epoch": 0.83, "learning_rate": 1.4496921392284633e-05, "loss": 2.074, "step": 15340 }, { "epoch": 0.83, "learning_rate": 1.4493331179205487e-05, "loss": 1.9235, "step": 15350 }, { "epoch": 0.83, "learning_rate": 1.448974096612634e-05, "loss": 1.9561, "step": 15360 }, { "epoch": 0.83, "learning_rate": 1.4486150753047195e-05, "loss": 1.9477, "step": 15370 }, { "epoch": 0.83, "learning_rate": 1.4482560539968049e-05, "loss": 1.9683, "step": 15380 }, { "epoch": 0.83, "learning_rate": 1.4478970326888901e-05, "loss": 2.0304, "step": 15390 }, { "epoch": 0.83, "learning_rate": 1.4475380113809756e-05, "loss": 2.1361, "step": 15400 }, { "epoch": 0.83, "learning_rate": 1.447178990073061e-05, "loss": 1.9325, "step": 15410 }, { "epoch": 0.83, "learning_rate": 1.4468199687651462e-05, "loss": 2.1411, "step": 15420 }, { "epoch": 0.83, "learning_rate": 1.4464609474572318e-05, "loss": 2.0253, "step": 15430 }, { "epoch": 0.83, "learning_rate": 1.446101926149317e-05, "loss": 2.0638, "step": 15440 }, { "epoch": 0.83, "learning_rate": 1.4457429048414024e-05, "loss": 2.1194, "step": 15450 }, { "epoch": 0.83, "learning_rate": 1.445383883533488e-05, "loss": 1.96, "step": 15460 }, { "epoch": 0.83, "learning_rate": 1.4450248622255732e-05, "loss": 2.0183, "step": 15470 }, { "epoch": 0.83, "learning_rate": 1.4446658409176586e-05, "loss": 2.045, "step": 15480 }, { "epoch": 0.83, "learning_rate": 1.4443068196097441e-05, "loss": 2.1903, "step": 15490 }, { "epoch": 0.83, "learning_rate": 1.4439477983018293e-05, "loss": 1.7952, "step": 15500 }, { "epoch": 0.84, "learning_rate": 1.4435887769939147e-05, "loss": 1.9941, "step": 15510 }, { "epoch": 0.84, "learning_rate": 1.443229755686e-05, "loss": 2.0757, "step": 15520 }, { "epoch": 0.84, "learning_rate": 1.4428707343780855e-05, "loss": 1.9646, "step": 15530 }, { "epoch": 0.84, "learning_rate": 1.4425117130701709e-05, "loss": 1.9857, "step": 15540 }, { "epoch": 0.84, "learning_rate": 1.442152691762256e-05, "loss": 2.0288, "step": 15550 }, { "epoch": 0.84, "learning_rate": 1.4417936704543416e-05, "loss": 1.9306, "step": 15560 }, { "epoch": 0.84, "learning_rate": 1.441434649146427e-05, "loss": 1.887, "step": 15570 }, { "epoch": 0.84, "learning_rate": 1.4410756278385122e-05, "loss": 2.0224, "step": 15580 }, { "epoch": 0.84, "learning_rate": 1.4407166065305978e-05, "loss": 1.9507, "step": 15590 }, { "epoch": 0.84, "learning_rate": 1.440357585222683e-05, "loss": 1.9471, "step": 15600 }, { "epoch": 0.84, "learning_rate": 1.4399985639147684e-05, "loss": 1.8485, "step": 15610 }, { "epoch": 0.84, "learning_rate": 1.439639542606854e-05, "loss": 1.9715, "step": 15620 }, { "epoch": 0.84, "learning_rate": 1.4392805212989392e-05, "loss": 1.9989, "step": 15630 }, { "epoch": 0.84, "learning_rate": 1.4389214999910245e-05, "loss": 2.0961, "step": 15640 }, { "epoch": 0.84, "learning_rate": 1.4385624786831101e-05, "loss": 2.066, "step": 15650 }, { "epoch": 0.84, "learning_rate": 1.4382034573751953e-05, "loss": 2.1179, "step": 15660 }, { "epoch": 0.84, "learning_rate": 1.4378444360672807e-05, "loss": 1.9075, "step": 15670 }, { "epoch": 0.84, "learning_rate": 1.4374854147593659e-05, "loss": 1.9378, "step": 15680 }, { "epoch": 0.84, "learning_rate": 1.4371263934514515e-05, "loss": 1.9555, "step": 15690 }, { "epoch": 0.85, "learning_rate": 1.4367673721435369e-05, "loss": 1.923, "step": 15700 }, { "epoch": 0.85, "learning_rate": 1.436408350835622e-05, "loss": 1.9627, "step": 15710 }, { "epoch": 0.85, "learning_rate": 1.4360493295277076e-05, "loss": 2.0411, "step": 15720 }, { "epoch": 0.85, "learning_rate": 1.435690308219793e-05, "loss": 1.8912, "step": 15730 }, { "epoch": 0.85, "learning_rate": 1.4353312869118782e-05, "loss": 2.0325, "step": 15740 }, { "epoch": 0.85, "learning_rate": 1.4349722656039638e-05, "loss": 1.9367, "step": 15750 }, { "epoch": 0.85, "learning_rate": 1.434613244296049e-05, "loss": 1.9747, "step": 15760 }, { "epoch": 0.85, "learning_rate": 1.4342542229881344e-05, "loss": 2.0033, "step": 15770 }, { "epoch": 0.85, "learning_rate": 1.43389520168022e-05, "loss": 2.024, "step": 15780 }, { "epoch": 0.85, "learning_rate": 1.4335361803723052e-05, "loss": 2.026, "step": 15790 }, { "epoch": 0.85, "learning_rate": 1.4331771590643905e-05, "loss": 1.9525, "step": 15800 }, { "epoch": 0.85, "learning_rate": 1.4328181377564761e-05, "loss": 1.8893, "step": 15810 }, { "epoch": 0.85, "learning_rate": 1.4324591164485613e-05, "loss": 1.9991, "step": 15820 }, { "epoch": 0.85, "learning_rate": 1.4321000951406467e-05, "loss": 2.0067, "step": 15830 }, { "epoch": 0.85, "learning_rate": 1.4317410738327322e-05, "loss": 2.0351, "step": 15840 }, { "epoch": 0.85, "learning_rate": 1.4313820525248175e-05, "loss": 1.9715, "step": 15850 }, { "epoch": 0.85, "learning_rate": 1.4310230312169028e-05, "loss": 1.9715, "step": 15860 }, { "epoch": 0.85, "learning_rate": 1.430664009908988e-05, "loss": 1.9043, "step": 15870 }, { "epoch": 0.86, "learning_rate": 1.4303049886010736e-05, "loss": 1.8549, "step": 15880 }, { "epoch": 0.86, "learning_rate": 1.429945967293159e-05, "loss": 1.9681, "step": 15890 }, { "epoch": 0.86, "learning_rate": 1.4295869459852442e-05, "loss": 1.8661, "step": 15900 }, { "epoch": 0.86, "learning_rate": 1.4292279246773298e-05, "loss": 2.0524, "step": 15910 }, { "epoch": 0.86, "learning_rate": 1.4288689033694152e-05, "loss": 1.9066, "step": 15920 }, { "epoch": 0.86, "learning_rate": 1.4285098820615004e-05, "loss": 2.0746, "step": 15930 }, { "epoch": 0.86, "learning_rate": 1.428150860753586e-05, "loss": 2.0448, "step": 15940 }, { "epoch": 0.86, "learning_rate": 1.4277918394456711e-05, "loss": 1.9522, "step": 15950 }, { "epoch": 0.86, "learning_rate": 1.4274328181377565e-05, "loss": 2.0247, "step": 15960 }, { "epoch": 0.86, "learning_rate": 1.427073796829842e-05, "loss": 2.0537, "step": 15970 }, { "epoch": 0.86, "learning_rate": 1.4267147755219273e-05, "loss": 1.968, "step": 15980 }, { "epoch": 0.86, "learning_rate": 1.4263557542140127e-05, "loss": 1.9622, "step": 15990 }, { "epoch": 0.86, "learning_rate": 1.4259967329060982e-05, "loss": 1.8356, "step": 16000 }, { "epoch": 0.86, "learning_rate": 1.4256377115981835e-05, "loss": 1.9053, "step": 16010 }, { "epoch": 0.86, "learning_rate": 1.4252786902902688e-05, "loss": 1.9957, "step": 16020 }, { "epoch": 0.86, "learning_rate": 1.424919668982354e-05, "loss": 1.8931, "step": 16030 }, { "epoch": 0.86, "learning_rate": 1.4245606476744396e-05, "loss": 1.9105, "step": 16040 }, { "epoch": 0.86, "learning_rate": 1.424201626366525e-05, "loss": 1.876, "step": 16050 }, { "epoch": 0.86, "learning_rate": 1.4238426050586102e-05, "loss": 1.8605, "step": 16060 }, { "epoch": 0.87, "learning_rate": 1.4234835837506958e-05, "loss": 1.9075, "step": 16070 }, { "epoch": 0.87, "learning_rate": 1.4231245624427811e-05, "loss": 1.9312, "step": 16080 }, { "epoch": 0.87, "learning_rate": 1.4227655411348664e-05, "loss": 2.0342, "step": 16090 }, { "epoch": 0.87, "learning_rate": 1.422406519826952e-05, "loss": 2.0015, "step": 16100 }, { "epoch": 0.87, "learning_rate": 1.4220474985190371e-05, "loss": 1.9941, "step": 16110 }, { "epoch": 0.87, "learning_rate": 1.4216884772111225e-05, "loss": 2.0914, "step": 16120 }, { "epoch": 0.87, "learning_rate": 1.421329455903208e-05, "loss": 1.8741, "step": 16130 }, { "epoch": 0.87, "learning_rate": 1.4209704345952933e-05, "loss": 2.0152, "step": 16140 }, { "epoch": 0.87, "learning_rate": 1.4206114132873787e-05, "loss": 1.9645, "step": 16150 }, { "epoch": 0.87, "learning_rate": 1.4202523919794642e-05, "loss": 1.8885, "step": 16160 }, { "epoch": 0.87, "learning_rate": 1.4198933706715494e-05, "loss": 1.9489, "step": 16170 }, { "epoch": 0.87, "learning_rate": 1.4195343493636348e-05, "loss": 2.0266, "step": 16180 }, { "epoch": 0.87, "learning_rate": 1.41917532805572e-05, "loss": 2.0092, "step": 16190 }, { "epoch": 0.87, "learning_rate": 1.4188163067478056e-05, "loss": 1.971, "step": 16200 }, { "epoch": 0.87, "learning_rate": 1.418457285439891e-05, "loss": 1.9781, "step": 16210 }, { "epoch": 0.87, "learning_rate": 1.4180982641319762e-05, "loss": 1.8303, "step": 16220 }, { "epoch": 0.87, "learning_rate": 1.4177392428240618e-05, "loss": 2.0398, "step": 16230 }, { "epoch": 0.87, "learning_rate": 1.4173802215161471e-05, "loss": 1.9476, "step": 16240 }, { "epoch": 0.88, "learning_rate": 1.4170212002082324e-05, "loss": 1.9903, "step": 16250 }, { "epoch": 0.88, "learning_rate": 1.4166621789003179e-05, "loss": 1.9999, "step": 16260 }, { "epoch": 0.88, "learning_rate": 1.4163031575924033e-05, "loss": 2.0079, "step": 16270 }, { "epoch": 0.88, "learning_rate": 1.4159441362844885e-05, "loss": 1.9331, "step": 16280 }, { "epoch": 0.88, "learning_rate": 1.415585114976574e-05, "loss": 1.9041, "step": 16290 }, { "epoch": 0.88, "learning_rate": 1.4152260936686593e-05, "loss": 1.9733, "step": 16300 }, { "epoch": 0.88, "learning_rate": 1.4148670723607447e-05, "loss": 1.9513, "step": 16310 }, { "epoch": 0.88, "learning_rate": 1.4145080510528302e-05, "loss": 1.8792, "step": 16320 }, { "epoch": 0.88, "learning_rate": 1.4141490297449154e-05, "loss": 1.978, "step": 16330 }, { "epoch": 0.88, "learning_rate": 1.4137900084370008e-05, "loss": 1.9788, "step": 16340 }, { "epoch": 0.88, "learning_rate": 1.4134309871290864e-05, "loss": 2.028, "step": 16350 }, { "epoch": 0.88, "learning_rate": 1.4130719658211716e-05, "loss": 1.9102, "step": 16360 }, { "epoch": 0.88, "learning_rate": 1.412712944513257e-05, "loss": 1.9286, "step": 16370 }, { "epoch": 0.88, "learning_rate": 1.4123539232053422e-05, "loss": 2.0198, "step": 16380 }, { "epoch": 0.88, "learning_rate": 1.4119949018974277e-05, "loss": 1.9118, "step": 16390 }, { "epoch": 0.88, "learning_rate": 1.4116358805895131e-05, "loss": 2.0512, "step": 16400 }, { "epoch": 0.88, "learning_rate": 1.4112768592815983e-05, "loss": 1.9825, "step": 16410 }, { "epoch": 0.88, "learning_rate": 1.4109178379736839e-05, "loss": 2.0601, "step": 16420 }, { "epoch": 0.88, "learning_rate": 1.4105588166657693e-05, "loss": 2.0098, "step": 16430 }, { "epoch": 0.89, "learning_rate": 1.4101997953578545e-05, "loss": 2.0663, "step": 16440 }, { "epoch": 0.89, "learning_rate": 1.40984077404994e-05, "loss": 1.94, "step": 16450 }, { "epoch": 0.89, "learning_rate": 1.4094817527420253e-05, "loss": 1.9743, "step": 16460 }, { "epoch": 0.89, "learning_rate": 1.4091227314341107e-05, "loss": 2.0206, "step": 16470 }, { "epoch": 0.89, "learning_rate": 1.4087637101261962e-05, "loss": 1.9322, "step": 16480 }, { "epoch": 0.89, "learning_rate": 1.4084046888182814e-05, "loss": 1.9907, "step": 16490 }, { "epoch": 0.89, "learning_rate": 1.4080456675103668e-05, "loss": 1.9723, "step": 16500 }, { "epoch": 0.89, "learning_rate": 1.4076866462024524e-05, "loss": 1.9327, "step": 16510 }, { "epoch": 0.89, "learning_rate": 1.4073276248945376e-05, "loss": 1.9044, "step": 16520 }, { "epoch": 0.89, "learning_rate": 1.406968603586623e-05, "loss": 1.8921, "step": 16530 }, { "epoch": 0.89, "learning_rate": 1.4066095822787082e-05, "loss": 2.0041, "step": 16540 }, { "epoch": 0.89, "learning_rate": 1.4062505609707937e-05, "loss": 1.88, "step": 16550 }, { "epoch": 0.89, "learning_rate": 1.4058915396628791e-05, "loss": 2.041, "step": 16560 }, { "epoch": 0.89, "learning_rate": 1.4055325183549643e-05, "loss": 1.9944, "step": 16570 }, { "epoch": 0.89, "learning_rate": 1.4051734970470499e-05, "loss": 1.9913, "step": 16580 }, { "epoch": 0.89, "learning_rate": 1.4048144757391353e-05, "loss": 1.9204, "step": 16590 }, { "epoch": 0.89, "learning_rate": 1.4044554544312205e-05, "loss": 2.0074, "step": 16600 }, { "epoch": 0.89, "learning_rate": 1.404096433123306e-05, "loss": 1.8801, "step": 16610 }, { "epoch": 0.89, "learning_rate": 1.4037374118153913e-05, "loss": 1.9455, "step": 16620 }, { "epoch": 0.9, "learning_rate": 1.4033783905074766e-05, "loss": 1.9434, "step": 16630 }, { "epoch": 0.9, "learning_rate": 1.4030193691995622e-05, "loss": 2.0602, "step": 16640 }, { "epoch": 0.9, "learning_rate": 1.4026603478916474e-05, "loss": 1.7402, "step": 16650 }, { "epoch": 0.9, "learning_rate": 1.4023013265837328e-05, "loss": 1.995, "step": 16660 }, { "epoch": 0.9, "learning_rate": 1.4019423052758184e-05, "loss": 1.7645, "step": 16670 }, { "epoch": 0.9, "learning_rate": 1.4015832839679036e-05, "loss": 2.0947, "step": 16680 }, { "epoch": 0.9, "learning_rate": 1.401224262659989e-05, "loss": 1.8972, "step": 16690 }, { "epoch": 0.9, "learning_rate": 1.4008652413520745e-05, "loss": 1.9739, "step": 16700 }, { "epoch": 0.9, "learning_rate": 1.4005062200441597e-05, "loss": 1.9881, "step": 16710 }, { "epoch": 0.9, "learning_rate": 1.4001471987362451e-05, "loss": 1.8128, "step": 16720 }, { "epoch": 0.9, "learning_rate": 1.3997881774283303e-05, "loss": 1.8953, "step": 16730 }, { "epoch": 0.9, "learning_rate": 1.3994291561204159e-05, "loss": 2.0479, "step": 16740 }, { "epoch": 0.9, "learning_rate": 1.3990701348125013e-05, "loss": 2.0189, "step": 16750 }, { "epoch": 0.9, "learning_rate": 1.3987111135045865e-05, "loss": 1.9738, "step": 16760 }, { "epoch": 0.9, "learning_rate": 1.398352092196672e-05, "loss": 2.0089, "step": 16770 }, { "epoch": 0.9, "learning_rate": 1.3979930708887574e-05, "loss": 1.9707, "step": 16780 }, { "epoch": 0.9, "learning_rate": 1.3976340495808426e-05, "loss": 1.9936, "step": 16790 }, { "epoch": 0.9, "learning_rate": 1.3972750282729282e-05, "loss": 2.0583, "step": 16800 }, { "epoch": 0.91, "learning_rate": 1.3969160069650134e-05, "loss": 1.9089, "step": 16810 }, { "epoch": 0.91, "learning_rate": 1.3965569856570988e-05, "loss": 2.0011, "step": 16820 }, { "epoch": 0.91, "learning_rate": 1.3961979643491844e-05, "loss": 1.9584, "step": 16830 }, { "epoch": 0.91, "learning_rate": 1.3958389430412696e-05, "loss": 1.8561, "step": 16840 }, { "epoch": 0.91, "learning_rate": 1.395479921733355e-05, "loss": 2.0188, "step": 16850 }, { "epoch": 0.91, "learning_rate": 1.3951209004254405e-05, "loss": 1.944, "step": 16860 }, { "epoch": 0.91, "learning_rate": 1.3947618791175257e-05, "loss": 1.8993, "step": 16870 }, { "epoch": 0.91, "learning_rate": 1.3944028578096111e-05, "loss": 1.7442, "step": 16880 }, { "epoch": 0.91, "learning_rate": 1.3940438365016963e-05, "loss": 1.9484, "step": 16890 }, { "epoch": 0.91, "learning_rate": 1.3936848151937819e-05, "loss": 1.8741, "step": 16900 }, { "epoch": 0.91, "learning_rate": 1.3933257938858673e-05, "loss": 1.9448, "step": 16910 }, { "epoch": 0.91, "learning_rate": 1.3929667725779525e-05, "loss": 1.8326, "step": 16920 }, { "epoch": 0.91, "learning_rate": 1.392607751270038e-05, "loss": 1.9772, "step": 16930 }, { "epoch": 0.91, "learning_rate": 1.3922487299621234e-05, "loss": 1.9518, "step": 16940 }, { "epoch": 0.91, "learning_rate": 1.3918897086542086e-05, "loss": 1.8712, "step": 16950 }, { "epoch": 0.91, "learning_rate": 1.3915306873462942e-05, "loss": 1.9424, "step": 16960 }, { "epoch": 0.91, "learning_rate": 1.3911716660383794e-05, "loss": 2.0596, "step": 16970 }, { "epoch": 0.91, "learning_rate": 1.3908126447304648e-05, "loss": 2.0653, "step": 16980 }, { "epoch": 0.91, "learning_rate": 1.3904536234225503e-05, "loss": 2.0152, "step": 16990 }, { "epoch": 0.92, "learning_rate": 1.3900946021146356e-05, "loss": 1.8808, "step": 17000 }, { "epoch": 0.92, "learning_rate": 1.389735580806721e-05, "loss": 2.0699, "step": 17010 }, { "epoch": 0.92, "learning_rate": 1.3893765594988065e-05, "loss": 1.9926, "step": 17020 }, { "epoch": 0.92, "learning_rate": 1.3890175381908917e-05, "loss": 1.8643, "step": 17030 }, { "epoch": 0.92, "learning_rate": 1.3886585168829771e-05, "loss": 1.9813, "step": 17040 }, { "epoch": 0.92, "learning_rate": 1.3882994955750623e-05, "loss": 1.8834, "step": 17050 }, { "epoch": 0.92, "learning_rate": 1.3879404742671479e-05, "loss": 2.0034, "step": 17060 }, { "epoch": 0.92, "learning_rate": 1.3875814529592333e-05, "loss": 1.9453, "step": 17070 }, { "epoch": 0.92, "learning_rate": 1.3872224316513185e-05, "loss": 1.949, "step": 17080 }, { "epoch": 0.92, "learning_rate": 1.386863410343404e-05, "loss": 2.0749, "step": 17090 }, { "epoch": 0.92, "learning_rate": 1.3865043890354894e-05, "loss": 2.0284, "step": 17100 }, { "epoch": 0.92, "learning_rate": 1.3861453677275746e-05, "loss": 1.8916, "step": 17110 }, { "epoch": 0.92, "learning_rate": 1.3857863464196602e-05, "loss": 1.9627, "step": 17120 }, { "epoch": 0.92, "learning_rate": 1.3854273251117456e-05, "loss": 2.0131, "step": 17130 }, { "epoch": 0.92, "learning_rate": 1.3850683038038308e-05, "loss": 2.022, "step": 17140 }, { "epoch": 0.92, "learning_rate": 1.3847092824959163e-05, "loss": 1.9384, "step": 17150 }, { "epoch": 0.92, "learning_rate": 1.3843502611880016e-05, "loss": 1.8964, "step": 17160 }, { "epoch": 0.92, "learning_rate": 1.383991239880087e-05, "loss": 1.8729, "step": 17170 }, { "epoch": 0.93, "learning_rate": 1.3836322185721725e-05, "loss": 1.8538, "step": 17180 }, { "epoch": 0.93, "learning_rate": 1.3832731972642577e-05, "loss": 1.9941, "step": 17190 }, { "epoch": 0.93, "learning_rate": 1.3829141759563431e-05, "loss": 1.9676, "step": 17200 }, { "epoch": 0.93, "learning_rate": 1.3825551546484286e-05, "loss": 1.9209, "step": 17210 }, { "epoch": 0.93, "learning_rate": 1.3821961333405139e-05, "loss": 1.8978, "step": 17220 }, { "epoch": 0.93, "learning_rate": 1.3818371120325992e-05, "loss": 1.7926, "step": 17230 }, { "epoch": 0.93, "learning_rate": 1.3814780907246845e-05, "loss": 1.934, "step": 17240 }, { "epoch": 0.93, "learning_rate": 1.38111906941677e-05, "loss": 1.9259, "step": 17250 }, { "epoch": 0.93, "learning_rate": 1.3807600481088554e-05, "loss": 1.84, "step": 17260 }, { "epoch": 0.93, "learning_rate": 1.3804010268009406e-05, "loss": 2.0435, "step": 17270 }, { "epoch": 0.93, "learning_rate": 1.3800420054930262e-05, "loss": 2.0319, "step": 17280 }, { "epoch": 0.93, "learning_rate": 1.3796829841851116e-05, "loss": 1.8894, "step": 17290 }, { "epoch": 0.93, "learning_rate": 1.3793239628771968e-05, "loss": 2.0701, "step": 17300 }, { "epoch": 0.93, "learning_rate": 1.3789649415692823e-05, "loss": 1.907, "step": 17310 }, { "epoch": 0.93, "learning_rate": 1.3786059202613675e-05, "loss": 2.0582, "step": 17320 }, { "epoch": 0.93, "learning_rate": 1.378246898953453e-05, "loss": 1.9768, "step": 17330 }, { "epoch": 0.93, "learning_rate": 1.3778878776455385e-05, "loss": 2.0354, "step": 17340 }, { "epoch": 0.93, "learning_rate": 1.3775288563376237e-05, "loss": 1.9354, "step": 17350 }, { "epoch": 0.93, "learning_rate": 1.377169835029709e-05, "loss": 1.8514, "step": 17360 }, { "epoch": 0.94, "learning_rate": 1.3768108137217946e-05, "loss": 1.8775, "step": 17370 }, { "epoch": 0.94, "learning_rate": 1.3764517924138799e-05, "loss": 1.8588, "step": 17380 }, { "epoch": 0.94, "learning_rate": 1.3760927711059652e-05, "loss": 1.9315, "step": 17390 }, { "epoch": 0.94, "learning_rate": 1.3757337497980505e-05, "loss": 1.9144, "step": 17400 }, { "epoch": 0.94, "learning_rate": 1.375374728490136e-05, "loss": 1.8276, "step": 17410 }, { "epoch": 0.94, "learning_rate": 1.3750157071822214e-05, "loss": 1.9901, "step": 17420 }, { "epoch": 0.94, "learning_rate": 1.3746566858743066e-05, "loss": 1.9458, "step": 17430 }, { "epoch": 0.94, "learning_rate": 1.3742976645663922e-05, "loss": 1.9794, "step": 17440 }, { "epoch": 0.94, "learning_rate": 1.3739386432584775e-05, "loss": 1.9105, "step": 17450 }, { "epoch": 0.94, "learning_rate": 1.3735796219505628e-05, "loss": 1.9361, "step": 17460 }, { "epoch": 0.94, "learning_rate": 1.3732206006426483e-05, "loss": 2.0216, "step": 17470 }, { "epoch": 0.94, "learning_rate": 1.3728615793347335e-05, "loss": 2.0762, "step": 17480 }, { "epoch": 0.94, "learning_rate": 1.372502558026819e-05, "loss": 1.8754, "step": 17490 }, { "epoch": 0.94, "learning_rate": 1.3721435367189045e-05, "loss": 1.9706, "step": 17500 }, { "epoch": 0.94, "learning_rate": 1.3717845154109897e-05, "loss": 1.9341, "step": 17510 }, { "epoch": 0.94, "learning_rate": 1.371425494103075e-05, "loss": 1.8783, "step": 17520 }, { "epoch": 0.94, "learning_rate": 1.3710664727951606e-05, "loss": 1.8955, "step": 17530 }, { "epoch": 0.94, "learning_rate": 1.3707074514872458e-05, "loss": 1.9838, "step": 17540 }, { "epoch": 0.95, "learning_rate": 1.3703484301793312e-05, "loss": 2.0297, "step": 17550 }, { "epoch": 0.95, "learning_rate": 1.3699894088714168e-05, "loss": 1.9573, "step": 17560 }, { "epoch": 0.95, "learning_rate": 1.369630387563502e-05, "loss": 1.8823, "step": 17570 }, { "epoch": 0.95, "learning_rate": 1.3692713662555874e-05, "loss": 1.8877, "step": 17580 }, { "epoch": 0.95, "learning_rate": 1.3689123449476726e-05, "loss": 1.8792, "step": 17590 }, { "epoch": 0.95, "learning_rate": 1.3685533236397582e-05, "loss": 2.0049, "step": 17600 }, { "epoch": 0.95, "learning_rate": 1.3681943023318435e-05, "loss": 1.9629, "step": 17610 }, { "epoch": 0.95, "learning_rate": 1.3678352810239288e-05, "loss": 2.0267, "step": 17620 }, { "epoch": 0.95, "learning_rate": 1.3674762597160143e-05, "loss": 1.9491, "step": 17630 }, { "epoch": 0.95, "learning_rate": 1.3671172384080997e-05, "loss": 1.8555, "step": 17640 }, { "epoch": 0.95, "learning_rate": 1.3667582171001849e-05, "loss": 1.9679, "step": 17650 }, { "epoch": 0.95, "learning_rate": 1.3663991957922705e-05, "loss": 1.9188, "step": 17660 }, { "epoch": 0.95, "learning_rate": 1.3660401744843557e-05, "loss": 2.1323, "step": 17670 }, { "epoch": 0.95, "learning_rate": 1.365681153176441e-05, "loss": 2.0195, "step": 17680 }, { "epoch": 0.95, "learning_rate": 1.3653221318685266e-05, "loss": 1.8739, "step": 17690 }, { "epoch": 0.95, "learning_rate": 1.3649631105606118e-05, "loss": 2.0275, "step": 17700 }, { "epoch": 0.95, "learning_rate": 1.3646040892526972e-05, "loss": 2.0296, "step": 17710 }, { "epoch": 0.95, "learning_rate": 1.3642450679447828e-05, "loss": 2.0098, "step": 17720 }, { "epoch": 0.95, "learning_rate": 1.363886046636868e-05, "loss": 1.9721, "step": 17730 }, { "epoch": 0.96, "learning_rate": 1.3635270253289534e-05, "loss": 1.7587, "step": 17740 }, { "epoch": 0.96, "learning_rate": 1.3631680040210386e-05, "loss": 1.8962, "step": 17750 }, { "epoch": 0.96, "learning_rate": 1.3628089827131241e-05, "loss": 1.939, "step": 17760 }, { "epoch": 0.96, "learning_rate": 1.3624499614052095e-05, "loss": 1.8013, "step": 17770 }, { "epoch": 0.96, "learning_rate": 1.3620909400972947e-05, "loss": 1.9124, "step": 17780 }, { "epoch": 0.96, "learning_rate": 1.3617319187893803e-05, "loss": 1.9174, "step": 17790 }, { "epoch": 0.96, "learning_rate": 1.3613728974814657e-05, "loss": 2.0872, "step": 17800 }, { "epoch": 0.96, "learning_rate": 1.3610138761735509e-05, "loss": 1.8096, "step": 17810 }, { "epoch": 0.96, "learning_rate": 1.3606548548656365e-05, "loss": 1.8067, "step": 17820 }, { "epoch": 0.96, "learning_rate": 1.3602958335577217e-05, "loss": 1.9801, "step": 17830 }, { "epoch": 0.96, "learning_rate": 1.359936812249807e-05, "loss": 1.926, "step": 17840 }, { "epoch": 0.96, "learning_rate": 1.3595777909418926e-05, "loss": 1.9291, "step": 17850 }, { "epoch": 0.96, "learning_rate": 1.3592187696339778e-05, "loss": 2.0261, "step": 17860 }, { "epoch": 0.96, "learning_rate": 1.3588597483260632e-05, "loss": 1.8809, "step": 17870 }, { "epoch": 0.96, "learning_rate": 1.3585007270181488e-05, "loss": 1.8724, "step": 17880 }, { "epoch": 0.96, "learning_rate": 1.358141705710234e-05, "loss": 1.8935, "step": 17890 }, { "epoch": 0.96, "learning_rate": 1.3577826844023194e-05, "loss": 2.0121, "step": 17900 }, { "epoch": 0.96, "learning_rate": 1.3574236630944046e-05, "loss": 1.8836, "step": 17910 }, { "epoch": 0.97, "learning_rate": 1.3570646417864901e-05, "loss": 1.9438, "step": 17920 }, { "epoch": 0.97, "learning_rate": 1.3567056204785755e-05, "loss": 1.8996, "step": 17930 }, { "epoch": 0.97, "learning_rate": 1.3563465991706607e-05, "loss": 2.0233, "step": 17940 }, { "epoch": 0.97, "learning_rate": 1.3559875778627463e-05, "loss": 1.9133, "step": 17950 }, { "epoch": 0.97, "learning_rate": 1.3556285565548317e-05, "loss": 2.1215, "step": 17960 }, { "epoch": 0.97, "learning_rate": 1.3552695352469169e-05, "loss": 1.869, "step": 17970 }, { "epoch": 0.97, "learning_rate": 1.3549105139390024e-05, "loss": 1.9853, "step": 17980 }, { "epoch": 0.97, "learning_rate": 1.3545514926310878e-05, "loss": 1.9775, "step": 17990 }, { "epoch": 0.97, "learning_rate": 1.354192471323173e-05, "loss": 1.9134, "step": 18000 }, { "epoch": 0.97, "learning_rate": 1.3538334500152586e-05, "loss": 1.9826, "step": 18010 }, { "epoch": 0.97, "learning_rate": 1.3534744287073438e-05, "loss": 1.9269, "step": 18020 }, { "epoch": 0.97, "learning_rate": 1.3531154073994292e-05, "loss": 1.8932, "step": 18030 }, { "epoch": 0.97, "learning_rate": 1.3527563860915148e-05, "loss": 2.1225, "step": 18040 }, { "epoch": 0.97, "learning_rate": 1.3523973647836e-05, "loss": 1.9616, "step": 18050 }, { "epoch": 0.97, "learning_rate": 1.3520383434756854e-05, "loss": 1.9021, "step": 18060 }, { "epoch": 0.97, "learning_rate": 1.3516793221677709e-05, "loss": 1.9402, "step": 18070 }, { "epoch": 0.97, "learning_rate": 1.3513203008598561e-05, "loss": 1.9571, "step": 18080 }, { "epoch": 0.97, "learning_rate": 1.3509612795519415e-05, "loss": 1.9891, "step": 18090 }, { "epoch": 0.97, "learning_rate": 1.3506022582440267e-05, "loss": 1.8689, "step": 18100 }, { "epoch": 0.98, "learning_rate": 1.3502432369361123e-05, "loss": 1.9705, "step": 18110 }, { "epoch": 0.98, "learning_rate": 1.3498842156281977e-05, "loss": 1.9255, "step": 18120 }, { "epoch": 0.98, "learning_rate": 1.3495251943202829e-05, "loss": 1.8557, "step": 18130 }, { "epoch": 0.98, "learning_rate": 1.3491661730123684e-05, "loss": 1.9083, "step": 18140 }, { "epoch": 0.98, "learning_rate": 1.3488071517044538e-05, "loss": 1.9571, "step": 18150 }, { "epoch": 0.98, "learning_rate": 1.348448130396539e-05, "loss": 1.9429, "step": 18160 }, { "epoch": 0.98, "learning_rate": 1.3480891090886246e-05, "loss": 1.9167, "step": 18170 }, { "epoch": 0.98, "learning_rate": 1.3477300877807098e-05, "loss": 1.9541, "step": 18180 }, { "epoch": 0.98, "learning_rate": 1.3473710664727952e-05, "loss": 1.8566, "step": 18190 }, { "epoch": 0.98, "learning_rate": 1.3470120451648808e-05, "loss": 1.855, "step": 18200 }, { "epoch": 0.98, "learning_rate": 1.346653023856966e-05, "loss": 1.8863, "step": 18210 }, { "epoch": 0.98, "learning_rate": 1.3462940025490514e-05, "loss": 2.0019, "step": 18220 }, { "epoch": 0.98, "learning_rate": 1.3459349812411369e-05, "loss": 1.9807, "step": 18230 }, { "epoch": 0.98, "learning_rate": 1.3455759599332221e-05, "loss": 1.9471, "step": 18240 }, { "epoch": 0.98, "learning_rate": 1.3452169386253075e-05, "loss": 1.9109, "step": 18250 }, { "epoch": 0.98, "learning_rate": 1.3448579173173927e-05, "loss": 1.9753, "step": 18260 }, { "epoch": 0.98, "learning_rate": 1.3444988960094783e-05, "loss": 1.9699, "step": 18270 }, { "epoch": 0.98, "learning_rate": 1.3441398747015637e-05, "loss": 2.0114, "step": 18280 }, { "epoch": 0.98, "learning_rate": 1.3437808533936489e-05, "loss": 1.9242, "step": 18290 }, { "epoch": 0.99, "learning_rate": 1.3434218320857344e-05, "loss": 1.8851, "step": 18300 }, { "epoch": 0.99, "learning_rate": 1.3430628107778198e-05, "loss": 1.8244, "step": 18310 }, { "epoch": 0.99, "learning_rate": 1.342703789469905e-05, "loss": 1.8548, "step": 18320 }, { "epoch": 0.99, "learning_rate": 1.3423447681619906e-05, "loss": 1.8949, "step": 18330 }, { "epoch": 0.99, "learning_rate": 1.3419857468540758e-05, "loss": 1.9934, "step": 18340 }, { "epoch": 0.99, "learning_rate": 1.3416267255461612e-05, "loss": 1.9683, "step": 18350 }, { "epoch": 0.99, "learning_rate": 1.3412677042382467e-05, "loss": 1.9204, "step": 18360 }, { "epoch": 0.99, "learning_rate": 1.340908682930332e-05, "loss": 1.9867, "step": 18370 }, { "epoch": 0.99, "learning_rate": 1.3405496616224173e-05, "loss": 1.8859, "step": 18380 }, { "epoch": 0.99, "learning_rate": 1.3401906403145029e-05, "loss": 1.7884, "step": 18390 }, { "epoch": 0.99, "learning_rate": 1.3398316190065881e-05, "loss": 1.981, "step": 18400 }, { "epoch": 0.99, "learning_rate": 1.3394725976986735e-05, "loss": 2.0169, "step": 18410 }, { "epoch": 0.99, "learning_rate": 1.339113576390759e-05, "loss": 2.0133, "step": 18420 }, { "epoch": 0.99, "learning_rate": 1.3387545550828443e-05, "loss": 1.8447, "step": 18430 }, { "epoch": 0.99, "learning_rate": 1.3383955337749297e-05, "loss": 1.9768, "step": 18440 }, { "epoch": 0.99, "learning_rate": 1.3380365124670149e-05, "loss": 1.9563, "step": 18450 }, { "epoch": 0.99, "learning_rate": 1.3376774911591004e-05, "loss": 1.9422, "step": 18460 }, { "epoch": 0.99, "learning_rate": 1.3373184698511858e-05, "loss": 2.0337, "step": 18470 }, { "epoch": 1.0, "learning_rate": 1.336959448543271e-05, "loss": 1.957, "step": 18480 }, { "epoch": 1.0, "learning_rate": 1.3366004272353566e-05, "loss": 1.9534, "step": 18490 }, { "epoch": 1.0, "learning_rate": 1.336241405927442e-05, "loss": 1.9711, "step": 18500 }, { "epoch": 1.0, "learning_rate": 1.3358823846195272e-05, "loss": 2.0202, "step": 18510 }, { "epoch": 1.0, "learning_rate": 1.3355233633116127e-05, "loss": 2.0269, "step": 18520 }, { "epoch": 1.0, "learning_rate": 1.335164342003698e-05, "loss": 1.8893, "step": 18530 }, { "epoch": 1.0, "learning_rate": 1.3348053206957833e-05, "loss": 1.7941, "step": 18540 }, { "epoch": 1.0, "learning_rate": 1.3344462993878689e-05, "loss": 1.9511, "step": 18550 }, { "epoch": 1.0, "learning_rate": 1.3340872780799541e-05, "loss": 1.928, "step": 18560 }, { "epoch": 1.0, "learning_rate": 1.3337282567720395e-05, "loss": 1.9972, "step": 18570 }, { "epoch": 1.0, "learning_rate": 1.333369235464125e-05, "loss": 1.9407, "step": 18580 }, { "epoch": 1.0, "learning_rate": 1.3330102141562103e-05, "loss": 1.696, "step": 18590 }, { "epoch": 1.0, "learning_rate": 1.3326511928482956e-05, "loss": 1.7199, "step": 18600 }, { "epoch": 1.0, "learning_rate": 1.3322921715403809e-05, "loss": 1.806, "step": 18610 }, { "epoch": 1.0, "learning_rate": 1.3319331502324664e-05, "loss": 1.8597, "step": 18620 }, { "epoch": 1.0, "learning_rate": 1.3315741289245518e-05, "loss": 1.7483, "step": 18630 }, { "epoch": 1.0, "learning_rate": 1.331215107616637e-05, "loss": 1.797, "step": 18640 }, { "epoch": 1.0, "learning_rate": 1.3308560863087226e-05, "loss": 1.7938, "step": 18650 }, { "epoch": 1.0, "learning_rate": 1.330497065000808e-05, "loss": 1.8008, "step": 18660 }, { "epoch": 1.01, "learning_rate": 1.3301380436928932e-05, "loss": 1.7371, "step": 18670 }, { "epoch": 1.01, "learning_rate": 1.3297790223849787e-05, "loss": 1.861, "step": 18680 }, { "epoch": 1.01, "learning_rate": 1.329420001077064e-05, "loss": 1.7679, "step": 18690 }, { "epoch": 1.01, "learning_rate": 1.3290609797691493e-05, "loss": 1.7499, "step": 18700 }, { "epoch": 1.01, "learning_rate": 1.3287019584612349e-05, "loss": 1.768, "step": 18710 }, { "epoch": 1.01, "learning_rate": 1.3283429371533201e-05, "loss": 1.8267, "step": 18720 }, { "epoch": 1.01, "learning_rate": 1.3279839158454055e-05, "loss": 1.8017, "step": 18730 }, { "epoch": 1.01, "learning_rate": 1.327624894537491e-05, "loss": 1.7822, "step": 18740 }, { "epoch": 1.01, "learning_rate": 1.3272658732295763e-05, "loss": 1.7097, "step": 18750 }, { "epoch": 1.01, "learning_rate": 1.3269068519216616e-05, "loss": 1.6784, "step": 18760 }, { "epoch": 1.01, "learning_rate": 1.3265478306137472e-05, "loss": 1.6903, "step": 18770 }, { "epoch": 1.01, "learning_rate": 1.3261888093058324e-05, "loss": 1.7884, "step": 18780 }, { "epoch": 1.01, "learning_rate": 1.3258297879979178e-05, "loss": 1.6404, "step": 18790 }, { "epoch": 1.01, "learning_rate": 1.325470766690003e-05, "loss": 1.7782, "step": 18800 }, { "epoch": 1.01, "learning_rate": 1.3251117453820886e-05, "loss": 1.8474, "step": 18810 }, { "epoch": 1.01, "learning_rate": 1.324752724074174e-05, "loss": 1.8297, "step": 18820 }, { "epoch": 1.01, "learning_rate": 1.3243937027662592e-05, "loss": 1.8117, "step": 18830 }, { "epoch": 1.01, "learning_rate": 1.3240346814583447e-05, "loss": 1.7813, "step": 18840 }, { "epoch": 1.02, "learning_rate": 1.3236756601504301e-05, "loss": 1.8464, "step": 18850 }, { "epoch": 1.02, "learning_rate": 1.3233166388425153e-05, "loss": 1.8961, "step": 18860 }, { "epoch": 1.02, "learning_rate": 1.3229576175346009e-05, "loss": 1.7762, "step": 18870 }, { "epoch": 1.02, "learning_rate": 1.3225985962266861e-05, "loss": 1.7805, "step": 18880 }, { "epoch": 1.02, "learning_rate": 1.3222395749187715e-05, "loss": 1.9418, "step": 18890 }, { "epoch": 1.02, "learning_rate": 1.321880553610857e-05, "loss": 1.7455, "step": 18900 }, { "epoch": 1.02, "learning_rate": 1.3215215323029422e-05, "loss": 1.7082, "step": 18910 }, { "epoch": 1.02, "learning_rate": 1.3211625109950276e-05, "loss": 1.7467, "step": 18920 }, { "epoch": 1.02, "learning_rate": 1.3208034896871132e-05, "loss": 1.9134, "step": 18930 }, { "epoch": 1.02, "learning_rate": 1.3204444683791984e-05, "loss": 1.7263, "step": 18940 }, { "epoch": 1.02, "learning_rate": 1.3200854470712838e-05, "loss": 1.8755, "step": 18950 }, { "epoch": 1.02, "learning_rate": 1.319726425763369e-05, "loss": 1.7216, "step": 18960 }, { "epoch": 1.02, "learning_rate": 1.3193674044554546e-05, "loss": 1.7302, "step": 18970 }, { "epoch": 1.02, "learning_rate": 1.31900838314754e-05, "loss": 1.7747, "step": 18980 }, { "epoch": 1.02, "learning_rate": 1.3186493618396252e-05, "loss": 1.6233, "step": 18990 }, { "epoch": 1.02, "learning_rate": 1.3182903405317107e-05, "loss": 1.8281, "step": 19000 }, { "epoch": 1.02, "learning_rate": 1.3179313192237961e-05, "loss": 1.8092, "step": 19010 }, { "epoch": 1.02, "learning_rate": 1.3175722979158813e-05, "loss": 1.8647, "step": 19020 }, { "epoch": 1.02, "learning_rate": 1.3172132766079669e-05, "loss": 1.7834, "step": 19030 }, { "epoch": 1.03, "learning_rate": 1.316854255300052e-05, "loss": 1.9089, "step": 19040 }, { "epoch": 1.03, "learning_rate": 1.3164952339921375e-05, "loss": 1.7163, "step": 19050 }, { "epoch": 1.03, "learning_rate": 1.316136212684223e-05, "loss": 1.7653, "step": 19060 }, { "epoch": 1.03, "learning_rate": 1.3157771913763082e-05, "loss": 1.8077, "step": 19070 }, { "epoch": 1.03, "learning_rate": 1.3154181700683936e-05, "loss": 1.7835, "step": 19080 }, { "epoch": 1.03, "learning_rate": 1.3150591487604792e-05, "loss": 1.8502, "step": 19090 }, { "epoch": 1.03, "learning_rate": 1.3147001274525644e-05, "loss": 1.8244, "step": 19100 }, { "epoch": 1.03, "learning_rate": 1.3143411061446498e-05, "loss": 1.6629, "step": 19110 }, { "epoch": 1.03, "learning_rate": 1.313982084836735e-05, "loss": 1.8535, "step": 19120 }, { "epoch": 1.03, "learning_rate": 1.3136230635288205e-05, "loss": 1.841, "step": 19130 }, { "epoch": 1.03, "learning_rate": 1.313264042220906e-05, "loss": 1.8037, "step": 19140 }, { "epoch": 1.03, "learning_rate": 1.3129050209129911e-05, "loss": 1.8704, "step": 19150 }, { "epoch": 1.03, "learning_rate": 1.3125459996050767e-05, "loss": 1.8562, "step": 19160 }, { "epoch": 1.03, "learning_rate": 1.3121869782971621e-05, "loss": 1.6579, "step": 19170 }, { "epoch": 1.03, "learning_rate": 1.3118279569892473e-05, "loss": 1.7192, "step": 19180 }, { "epoch": 1.03, "learning_rate": 1.3114689356813329e-05, "loss": 1.8458, "step": 19190 }, { "epoch": 1.03, "learning_rate": 1.3111099143734182e-05, "loss": 1.722, "step": 19200 }, { "epoch": 1.03, "learning_rate": 1.3107508930655035e-05, "loss": 1.8337, "step": 19210 }, { "epoch": 1.04, "learning_rate": 1.310391871757589e-05, "loss": 1.8254, "step": 19220 }, { "epoch": 1.04, "learning_rate": 1.3100328504496742e-05, "loss": 1.6823, "step": 19230 }, { "epoch": 1.04, "learning_rate": 1.3096738291417596e-05, "loss": 1.7848, "step": 19240 }, { "epoch": 1.04, "learning_rate": 1.3093148078338452e-05, "loss": 1.7432, "step": 19250 }, { "epoch": 1.04, "learning_rate": 1.308991688656722e-05, "loss": 1.7444, "step": 19260 }, { "epoch": 1.04, "learning_rate": 1.3086326673488071e-05, "loss": 1.8404, "step": 19270 }, { "epoch": 1.04, "learning_rate": 1.3082736460408927e-05, "loss": 1.7824, "step": 19280 }, { "epoch": 1.04, "learning_rate": 1.3079146247329779e-05, "loss": 1.7939, "step": 19290 }, { "epoch": 1.04, "learning_rate": 1.3075556034250633e-05, "loss": 1.8649, "step": 19300 }, { "epoch": 1.04, "learning_rate": 1.3071965821171489e-05, "loss": 1.7568, "step": 19310 }, { "epoch": 1.04, "learning_rate": 1.306837560809234e-05, "loss": 1.8244, "step": 19320 }, { "epoch": 1.04, "learning_rate": 1.3064785395013195e-05, "loss": 1.6341, "step": 19330 }, { "epoch": 1.04, "learning_rate": 1.306119518193405e-05, "loss": 1.6557, "step": 19340 }, { "epoch": 1.04, "learning_rate": 1.3057604968854902e-05, "loss": 1.6874, "step": 19350 }, { "epoch": 1.04, "learning_rate": 1.3054014755775756e-05, "loss": 1.7809, "step": 19360 }, { "epoch": 1.04, "learning_rate": 1.305042454269661e-05, "loss": 1.7649, "step": 19370 }, { "epoch": 1.04, "learning_rate": 1.3046834329617464e-05, "loss": 1.7408, "step": 19380 }, { "epoch": 1.04, "learning_rate": 1.3043244116538318e-05, "loss": 1.8852, "step": 19390 }, { "epoch": 1.04, "learning_rate": 1.3039653903459172e-05, "loss": 1.7279, "step": 19400 }, { "epoch": 1.05, "learning_rate": 1.3036063690380025e-05, "loss": 1.7686, "step": 19410 }, { "epoch": 1.05, "learning_rate": 1.303247347730088e-05, "loss": 1.8555, "step": 19420 }, { "epoch": 1.05, "learning_rate": 1.3028883264221733e-05, "loss": 1.8016, "step": 19430 }, { "epoch": 1.05, "learning_rate": 1.3025293051142587e-05, "loss": 1.7659, "step": 19440 }, { "epoch": 1.05, "learning_rate": 1.3021702838063439e-05, "loss": 1.8517, "step": 19450 }, { "epoch": 1.05, "learning_rate": 1.3018112624984293e-05, "loss": 1.8932, "step": 19460 }, { "epoch": 1.05, "learning_rate": 1.3014522411905148e-05, "loss": 1.8636, "step": 19470 }, { "epoch": 1.05, "learning_rate": 1.3010932198826e-05, "loss": 1.9053, "step": 19480 }, { "epoch": 1.05, "learning_rate": 1.3007341985746854e-05, "loss": 1.7368, "step": 19490 }, { "epoch": 1.05, "learning_rate": 1.300375177266771e-05, "loss": 1.7335, "step": 19500 }, { "epoch": 1.05, "learning_rate": 1.3000161559588562e-05, "loss": 1.6094, "step": 19510 }, { "epoch": 1.05, "learning_rate": 1.2996571346509416e-05, "loss": 1.6847, "step": 19520 }, { "epoch": 1.05, "learning_rate": 1.299298113343027e-05, "loss": 1.8355, "step": 19530 }, { "epoch": 1.05, "learning_rate": 1.2989390920351124e-05, "loss": 1.86, "step": 19540 }, { "epoch": 1.05, "learning_rate": 1.2985800707271978e-05, "loss": 1.676, "step": 19550 }, { "epoch": 1.05, "learning_rate": 1.2982210494192831e-05, "loss": 1.6875, "step": 19560 }, { "epoch": 1.05, "learning_rate": 1.2978620281113685e-05, "loss": 1.7743, "step": 19570 }, { "epoch": 1.05, "learning_rate": 1.2975030068034539e-05, "loss": 1.7975, "step": 19580 }, { "epoch": 1.05, "learning_rate": 1.2971439854955393e-05, "loss": 1.8773, "step": 19590 }, { "epoch": 1.06, "learning_rate": 1.2967849641876247e-05, "loss": 1.7479, "step": 19600 }, { "epoch": 1.06, "learning_rate": 1.2964259428797099e-05, "loss": 1.6837, "step": 19610 }, { "epoch": 1.06, "learning_rate": 1.2960669215717953e-05, "loss": 1.7116, "step": 19620 }, { "epoch": 1.06, "learning_rate": 1.2957079002638808e-05, "loss": 1.5853, "step": 19630 }, { "epoch": 1.06, "learning_rate": 1.295348878955966e-05, "loss": 1.718, "step": 19640 }, { "epoch": 1.06, "learning_rate": 1.2949898576480514e-05, "loss": 1.906, "step": 19650 }, { "epoch": 1.06, "learning_rate": 1.294630836340137e-05, "loss": 1.7537, "step": 19660 }, { "epoch": 1.06, "learning_rate": 1.2942718150322222e-05, "loss": 1.7788, "step": 19670 }, { "epoch": 1.06, "learning_rate": 1.2939127937243076e-05, "loss": 1.7932, "step": 19680 }, { "epoch": 1.06, "learning_rate": 1.2935537724163931e-05, "loss": 1.95, "step": 19690 }, { "epoch": 1.06, "learning_rate": 1.2931947511084784e-05, "loss": 1.7249, "step": 19700 }, { "epoch": 1.06, "learning_rate": 1.2928357298005637e-05, "loss": 1.6228, "step": 19710 }, { "epoch": 1.06, "learning_rate": 1.2924767084926491e-05, "loss": 1.7451, "step": 19720 }, { "epoch": 1.06, "learning_rate": 1.2921176871847345e-05, "loss": 1.712, "step": 19730 }, { "epoch": 1.06, "learning_rate": 1.2917586658768199e-05, "loss": 1.7371, "step": 19740 }, { "epoch": 1.06, "learning_rate": 1.2913996445689053e-05, "loss": 1.7924, "step": 19750 }, { "epoch": 1.06, "learning_rate": 1.2910406232609907e-05, "loss": 1.7688, "step": 19760 }, { "epoch": 1.06, "learning_rate": 1.290681601953076e-05, "loss": 1.7632, "step": 19770 }, { "epoch": 1.07, "learning_rate": 1.2903225806451613e-05, "loss": 1.6841, "step": 19780 }, { "epoch": 1.07, "learning_rate": 1.2899635593372468e-05, "loss": 1.7368, "step": 19790 }, { "epoch": 1.07, "learning_rate": 1.289604538029332e-05, "loss": 1.8147, "step": 19800 }, { "epoch": 1.07, "learning_rate": 1.2892455167214174e-05, "loss": 1.65, "step": 19810 }, { "epoch": 1.07, "learning_rate": 1.288886495413503e-05, "loss": 1.7752, "step": 19820 }, { "epoch": 1.07, "learning_rate": 1.2885274741055882e-05, "loss": 1.7301, "step": 19830 }, { "epoch": 1.07, "learning_rate": 1.2881684527976736e-05, "loss": 1.7366, "step": 19840 }, { "epoch": 1.07, "learning_rate": 1.2878094314897591e-05, "loss": 1.6592, "step": 19850 }, { "epoch": 1.07, "learning_rate": 1.2874504101818444e-05, "loss": 1.7674, "step": 19860 }, { "epoch": 1.07, "learning_rate": 1.2870913888739297e-05, "loss": 1.7379, "step": 19870 }, { "epoch": 1.07, "learning_rate": 1.2867323675660151e-05, "loss": 1.8247, "step": 19880 }, { "epoch": 1.07, "learning_rate": 1.2863733462581005e-05, "loss": 1.8423, "step": 19890 }, { "epoch": 1.07, "learning_rate": 1.2860143249501859e-05, "loss": 1.8379, "step": 19900 }, { "epoch": 1.07, "learning_rate": 1.2856553036422713e-05, "loss": 1.7165, "step": 19910 }, { "epoch": 1.07, "learning_rate": 1.2852962823343567e-05, "loss": 1.7198, "step": 19920 }, { "epoch": 1.07, "learning_rate": 1.284937261026442e-05, "loss": 1.6824, "step": 19930 }, { "epoch": 1.07, "learning_rate": 1.2845782397185274e-05, "loss": 1.7188, "step": 19940 }, { "epoch": 1.07, "learning_rate": 1.2842192184106128e-05, "loss": 1.866, "step": 19950 }, { "epoch": 1.07, "learning_rate": 1.283860197102698e-05, "loss": 1.6882, "step": 19960 }, { "epoch": 1.08, "learning_rate": 1.2835011757947834e-05, "loss": 1.7889, "step": 19970 }, { "epoch": 1.08, "learning_rate": 1.283142154486869e-05, "loss": 1.9276, "step": 19980 }, { "epoch": 1.08, "learning_rate": 1.2827831331789542e-05, "loss": 1.651, "step": 19990 }, { "epoch": 1.08, "learning_rate": 1.2824241118710396e-05, "loss": 1.8214, "step": 20000 }, { "epoch": 1.08, "learning_rate": 1.2820650905631251e-05, "loss": 1.8692, "step": 20010 }, { "epoch": 1.08, "learning_rate": 1.2817060692552103e-05, "loss": 1.8377, "step": 20020 }, { "epoch": 1.08, "learning_rate": 1.2813470479472957e-05, "loss": 1.8392, "step": 20030 }, { "epoch": 1.08, "learning_rate": 1.2809880266393811e-05, "loss": 1.7332, "step": 20040 }, { "epoch": 1.08, "learning_rate": 1.2806290053314665e-05, "loss": 1.9028, "step": 20050 }, { "epoch": 1.08, "learning_rate": 1.2802699840235519e-05, "loss": 1.8507, "step": 20060 }, { "epoch": 1.08, "learning_rate": 1.2799109627156373e-05, "loss": 1.8539, "step": 20070 }, { "epoch": 1.08, "learning_rate": 1.2795519414077227e-05, "loss": 1.7467, "step": 20080 }, { "epoch": 1.08, "learning_rate": 1.279192920099808e-05, "loss": 1.6939, "step": 20090 }, { "epoch": 1.08, "learning_rate": 1.2788338987918934e-05, "loss": 1.7934, "step": 20100 }, { "epoch": 1.08, "learning_rate": 1.2784748774839788e-05, "loss": 1.7787, "step": 20110 }, { "epoch": 1.08, "learning_rate": 1.2781158561760642e-05, "loss": 1.7349, "step": 20120 }, { "epoch": 1.08, "learning_rate": 1.2777568348681494e-05, "loss": 1.81, "step": 20130 }, { "epoch": 1.08, "learning_rate": 1.277397813560235e-05, "loss": 1.6787, "step": 20140 }, { "epoch": 1.09, "learning_rate": 1.2770387922523202e-05, "loss": 1.8114, "step": 20150 }, { "epoch": 1.09, "learning_rate": 1.2766797709444056e-05, "loss": 1.8685, "step": 20160 }, { "epoch": 1.09, "learning_rate": 1.2763207496364911e-05, "loss": 1.6548, "step": 20170 }, { "epoch": 1.09, "learning_rate": 1.2759617283285763e-05, "loss": 1.7486, "step": 20180 }, { "epoch": 1.09, "learning_rate": 1.2756027070206617e-05, "loss": 1.7679, "step": 20190 }, { "epoch": 1.09, "learning_rate": 1.2752436857127473e-05, "loss": 1.7955, "step": 20200 }, { "epoch": 1.09, "learning_rate": 1.2748846644048325e-05, "loss": 1.6551, "step": 20210 }, { "epoch": 1.09, "learning_rate": 1.2745256430969179e-05, "loss": 1.762, "step": 20220 }, { "epoch": 1.09, "learning_rate": 1.2741666217890033e-05, "loss": 1.8542, "step": 20230 }, { "epoch": 1.09, "learning_rate": 1.2738076004810887e-05, "loss": 1.7594, "step": 20240 }, { "epoch": 1.09, "learning_rate": 1.273448579173174e-05, "loss": 1.8916, "step": 20250 }, { "epoch": 1.09, "learning_rate": 1.2730895578652594e-05, "loss": 1.6549, "step": 20260 }, { "epoch": 1.09, "learning_rate": 1.2727305365573448e-05, "loss": 1.9288, "step": 20270 }, { "epoch": 1.09, "learning_rate": 1.2723715152494302e-05, "loss": 1.8467, "step": 20280 }, { "epoch": 1.09, "learning_rate": 1.2720124939415156e-05, "loss": 1.8236, "step": 20290 }, { "epoch": 1.09, "learning_rate": 1.271653472633601e-05, "loss": 1.8475, "step": 20300 }, { "epoch": 1.09, "learning_rate": 1.2712944513256862e-05, "loss": 1.6939, "step": 20310 }, { "epoch": 1.09, "learning_rate": 1.2709354300177716e-05, "loss": 1.7849, "step": 20320 }, { "epoch": 1.09, "learning_rate": 1.2705764087098571e-05, "loss": 1.8494, "step": 20330 }, { "epoch": 1.1, "learning_rate": 1.2702173874019423e-05, "loss": 1.8983, "step": 20340 }, { "epoch": 1.1, "learning_rate": 1.2698583660940277e-05, "loss": 1.7474, "step": 20350 }, { "epoch": 1.1, "learning_rate": 1.2694993447861133e-05, "loss": 1.7819, "step": 20360 }, { "epoch": 1.1, "learning_rate": 1.2691403234781985e-05, "loss": 1.9049, "step": 20370 }, { "epoch": 1.1, "learning_rate": 1.2687813021702839e-05, "loss": 1.7426, "step": 20380 }, { "epoch": 1.1, "learning_rate": 1.2684222808623693e-05, "loss": 1.6417, "step": 20390 }, { "epoch": 1.1, "learning_rate": 1.2680632595544546e-05, "loss": 1.8293, "step": 20400 }, { "epoch": 1.1, "learning_rate": 1.26770423824654e-05, "loss": 1.707, "step": 20410 }, { "epoch": 1.1, "learning_rate": 1.2673452169386254e-05, "loss": 1.9008, "step": 20420 }, { "epoch": 1.1, "learning_rate": 1.2669861956307108e-05, "loss": 1.6466, "step": 20430 }, { "epoch": 1.1, "learning_rate": 1.2666271743227962e-05, "loss": 1.7816, "step": 20440 }, { "epoch": 1.1, "learning_rate": 1.2662681530148816e-05, "loss": 1.7785, "step": 20450 }, { "epoch": 1.1, "learning_rate": 1.265909131706967e-05, "loss": 1.7442, "step": 20460 }, { "epoch": 1.1, "learning_rate": 1.2655501103990522e-05, "loss": 1.6724, "step": 20470 }, { "epoch": 1.1, "learning_rate": 1.2651910890911376e-05, "loss": 1.7555, "step": 20480 }, { "epoch": 1.1, "learning_rate": 1.2648320677832231e-05, "loss": 1.7557, "step": 20490 }, { "epoch": 1.1, "learning_rate": 1.2644730464753083e-05, "loss": 1.6681, "step": 20500 }, { "epoch": 1.1, "learning_rate": 1.2641140251673937e-05, "loss": 1.8153, "step": 20510 }, { "epoch": 1.11, "learning_rate": 1.2637550038594793e-05, "loss": 1.9002, "step": 20520 }, { "epoch": 1.11, "learning_rate": 1.2633959825515645e-05, "loss": 1.7669, "step": 20530 }, { "epoch": 1.11, "learning_rate": 1.2630369612436499e-05, "loss": 1.7453, "step": 20540 }, { "epoch": 1.11, "learning_rate": 1.2626779399357354e-05, "loss": 1.7848, "step": 20550 }, { "epoch": 1.11, "learning_rate": 1.2623189186278206e-05, "loss": 1.6359, "step": 20560 }, { "epoch": 1.11, "learning_rate": 1.261959897319906e-05, "loss": 1.7285, "step": 20570 }, { "epoch": 1.11, "learning_rate": 1.2616008760119914e-05, "loss": 1.6288, "step": 20580 }, { "epoch": 1.11, "learning_rate": 1.2612418547040768e-05, "loss": 1.7294, "step": 20590 }, { "epoch": 1.11, "learning_rate": 1.2608828333961622e-05, "loss": 1.6621, "step": 20600 }, { "epoch": 1.11, "learning_rate": 1.2605238120882476e-05, "loss": 1.7675, "step": 20610 }, { "epoch": 1.11, "learning_rate": 1.260164790780333e-05, "loss": 1.7777, "step": 20620 }, { "epoch": 1.11, "learning_rate": 1.2598057694724183e-05, "loss": 1.8369, "step": 20630 }, { "epoch": 1.11, "learning_rate": 1.2594467481645035e-05, "loss": 1.86, "step": 20640 }, { "epoch": 1.11, "learning_rate": 1.2590877268565891e-05, "loss": 1.6539, "step": 20650 }, { "epoch": 1.11, "learning_rate": 1.2587287055486743e-05, "loss": 1.8151, "step": 20660 }, { "epoch": 1.11, "learning_rate": 1.2583696842407597e-05, "loss": 1.8652, "step": 20670 }, { "epoch": 1.11, "learning_rate": 1.2580106629328453e-05, "loss": 1.8267, "step": 20680 }, { "epoch": 1.11, "learning_rate": 1.2576516416249305e-05, "loss": 1.8106, "step": 20690 }, { "epoch": 1.11, "learning_rate": 1.2572926203170159e-05, "loss": 1.852, "step": 20700 }, { "epoch": 1.12, "learning_rate": 1.2569335990091014e-05, "loss": 1.7338, "step": 20710 }, { "epoch": 1.12, "learning_rate": 1.2565745777011866e-05, "loss": 1.7288, "step": 20720 }, { "epoch": 1.12, "learning_rate": 1.256215556393272e-05, "loss": 1.7089, "step": 20730 }, { "epoch": 1.12, "learning_rate": 1.2558565350853574e-05, "loss": 1.6135, "step": 20740 }, { "epoch": 1.12, "learning_rate": 1.2554975137774428e-05, "loss": 1.7718, "step": 20750 }, { "epoch": 1.12, "learning_rate": 1.2551384924695282e-05, "loss": 1.7169, "step": 20760 }, { "epoch": 1.12, "learning_rate": 1.2547794711616136e-05, "loss": 1.7485, "step": 20770 }, { "epoch": 1.12, "learning_rate": 1.254420449853699e-05, "loss": 1.7821, "step": 20780 }, { "epoch": 1.12, "learning_rate": 1.2540614285457843e-05, "loss": 1.7196, "step": 20790 }, { "epoch": 1.12, "learning_rate": 1.2537024072378697e-05, "loss": 1.8118, "step": 20800 }, { "epoch": 1.12, "learning_rate": 1.2533433859299551e-05, "loss": 1.7692, "step": 20810 }, { "epoch": 1.12, "learning_rate": 1.2529843646220403e-05, "loss": 1.7816, "step": 20820 }, { "epoch": 1.12, "learning_rate": 1.2526253433141257e-05, "loss": 1.7582, "step": 20830 }, { "epoch": 1.12, "learning_rate": 1.2522663220062112e-05, "loss": 1.8181, "step": 20840 }, { "epoch": 1.12, "learning_rate": 1.2519073006982965e-05, "loss": 1.7159, "step": 20850 }, { "epoch": 1.12, "learning_rate": 1.2515482793903818e-05, "loss": 1.7464, "step": 20860 }, { "epoch": 1.12, "learning_rate": 1.2511892580824674e-05, "loss": 1.6706, "step": 20870 }, { "epoch": 1.12, "learning_rate": 1.2508302367745526e-05, "loss": 1.8153, "step": 20880 }, { "epoch": 1.12, "learning_rate": 1.250471215466638e-05, "loss": 1.8627, "step": 20890 }, { "epoch": 1.13, "learning_rate": 1.2501121941587234e-05, "loss": 1.6918, "step": 20900 }, { "epoch": 1.13, "learning_rate": 1.2497531728508088e-05, "loss": 1.8377, "step": 20910 }, { "epoch": 1.13, "learning_rate": 1.2493941515428942e-05, "loss": 1.7511, "step": 20920 }, { "epoch": 1.13, "learning_rate": 1.2490351302349795e-05, "loss": 1.8068, "step": 20930 }, { "epoch": 1.13, "learning_rate": 1.248676108927065e-05, "loss": 1.6007, "step": 20940 }, { "epoch": 1.13, "learning_rate": 1.2483170876191503e-05, "loss": 1.6785, "step": 20950 }, { "epoch": 1.13, "learning_rate": 1.2479580663112357e-05, "loss": 1.7841, "step": 20960 }, { "epoch": 1.13, "learning_rate": 1.247599045003321e-05, "loss": 1.6256, "step": 20970 }, { "epoch": 1.13, "learning_rate": 1.2472400236954065e-05, "loss": 1.7497, "step": 20980 }, { "epoch": 1.13, "learning_rate": 1.2468810023874917e-05, "loss": 1.773, "step": 20990 }, { "epoch": 1.13, "learning_rate": 1.2465219810795772e-05, "loss": 1.8381, "step": 21000 }, { "epoch": 1.13, "learning_rate": 1.2461629597716625e-05, "loss": 1.7507, "step": 21010 }, { "epoch": 1.13, "learning_rate": 1.2458039384637478e-05, "loss": 1.761, "step": 21020 }, { "epoch": 1.13, "learning_rate": 1.2454449171558334e-05, "loss": 1.7833, "step": 21030 }, { "epoch": 1.13, "learning_rate": 1.2450858958479186e-05, "loss": 1.7085, "step": 21040 }, { "epoch": 1.13, "learning_rate": 1.244726874540004e-05, "loss": 1.7134, "step": 21050 }, { "epoch": 1.13, "learning_rate": 1.2443678532320895e-05, "loss": 1.815, "step": 21060 }, { "epoch": 1.13, "learning_rate": 1.2440088319241748e-05, "loss": 1.7771, "step": 21070 }, { "epoch": 1.14, "learning_rate": 1.2436498106162601e-05, "loss": 1.7808, "step": 21080 }, { "epoch": 1.14, "learning_rate": 1.2432907893083455e-05, "loss": 1.774, "step": 21090 }, { "epoch": 1.14, "learning_rate": 1.242931768000431e-05, "loss": 1.8024, "step": 21100 }, { "epoch": 1.14, "learning_rate": 1.2425727466925163e-05, "loss": 1.7885, "step": 21110 }, { "epoch": 1.14, "learning_rate": 1.2422137253846017e-05, "loss": 1.7049, "step": 21120 }, { "epoch": 1.14, "learning_rate": 1.241854704076687e-05, "loss": 1.6753, "step": 21130 }, { "epoch": 1.14, "learning_rate": 1.2414956827687725e-05, "loss": 1.8175, "step": 21140 }, { "epoch": 1.14, "learning_rate": 1.2411366614608578e-05, "loss": 1.765, "step": 21150 }, { "epoch": 1.14, "learning_rate": 1.2407776401529432e-05, "loss": 1.7993, "step": 21160 }, { "epoch": 1.14, "learning_rate": 1.2404186188450284e-05, "loss": 1.6822, "step": 21170 }, { "epoch": 1.14, "learning_rate": 1.2400595975371138e-05, "loss": 1.6994, "step": 21180 }, { "epoch": 1.14, "learning_rate": 1.2397005762291994e-05, "loss": 1.723, "step": 21190 }, { "epoch": 1.14, "learning_rate": 1.2393415549212846e-05, "loss": 1.8503, "step": 21200 }, { "epoch": 1.14, "learning_rate": 1.23898253361337e-05, "loss": 1.7853, "step": 21210 }, { "epoch": 1.14, "learning_rate": 1.2386235123054555e-05, "loss": 1.764, "step": 21220 }, { "epoch": 1.14, "learning_rate": 1.2382644909975408e-05, "loss": 1.8558, "step": 21230 }, { "epoch": 1.14, "learning_rate": 1.2379054696896261e-05, "loss": 1.8028, "step": 21240 }, { "epoch": 1.14, "learning_rate": 1.2375464483817115e-05, "loss": 1.8349, "step": 21250 }, { "epoch": 1.14, "learning_rate": 1.2371874270737969e-05, "loss": 1.8957, "step": 21260 }, { "epoch": 1.15, "learning_rate": 1.2368284057658823e-05, "loss": 1.7902, "step": 21270 }, { "epoch": 1.15, "learning_rate": 1.2364693844579677e-05, "loss": 1.7482, "step": 21280 }, { "epoch": 1.15, "learning_rate": 1.236110363150053e-05, "loss": 1.7468, "step": 21290 }, { "epoch": 1.15, "learning_rate": 1.2357513418421385e-05, "loss": 1.7689, "step": 21300 }, { "epoch": 1.15, "learning_rate": 1.2353923205342238e-05, "loss": 1.5882, "step": 21310 }, { "epoch": 1.15, "learning_rate": 1.2350332992263092e-05, "loss": 1.6365, "step": 21320 }, { "epoch": 1.15, "learning_rate": 1.2346742779183944e-05, "loss": 1.5739, "step": 21330 }, { "epoch": 1.15, "learning_rate": 1.2343152566104798e-05, "loss": 1.7283, "step": 21340 }, { "epoch": 1.15, "learning_rate": 1.2339562353025654e-05, "loss": 1.8061, "step": 21350 }, { "epoch": 1.15, "learning_rate": 1.2335972139946506e-05, "loss": 1.8564, "step": 21360 }, { "epoch": 1.15, "learning_rate": 1.233238192686736e-05, "loss": 1.7724, "step": 21370 }, { "epoch": 1.15, "learning_rate": 1.2328791713788215e-05, "loss": 1.7041, "step": 21380 }, { "epoch": 1.15, "learning_rate": 1.2325201500709067e-05, "loss": 1.8317, "step": 21390 }, { "epoch": 1.15, "learning_rate": 1.2321611287629921e-05, "loss": 1.6964, "step": 21400 }, { "epoch": 1.15, "learning_rate": 1.2318021074550777e-05, "loss": 1.7921, "step": 21410 }, { "epoch": 1.15, "learning_rate": 1.2314430861471629e-05, "loss": 1.7006, "step": 21420 }, { "epoch": 1.15, "learning_rate": 1.2310840648392483e-05, "loss": 1.8333, "step": 21430 }, { "epoch": 1.15, "learning_rate": 1.2307250435313337e-05, "loss": 1.7508, "step": 21440 }, { "epoch": 1.16, "learning_rate": 1.230366022223419e-05, "loss": 1.78, "step": 21450 }, { "epoch": 1.16, "learning_rate": 1.2300070009155044e-05, "loss": 1.7815, "step": 21460 }, { "epoch": 1.16, "learning_rate": 1.2296479796075898e-05, "loss": 1.7311, "step": 21470 }, { "epoch": 1.16, "learning_rate": 1.2292889582996752e-05, "loss": 1.8281, "step": 21480 }, { "epoch": 1.16, "learning_rate": 1.2289299369917606e-05, "loss": 1.8276, "step": 21490 }, { "epoch": 1.16, "learning_rate": 1.228570915683846e-05, "loss": 1.6702, "step": 21500 }, { "epoch": 1.16, "learning_rate": 1.2282118943759314e-05, "loss": 1.6485, "step": 21510 }, { "epoch": 1.16, "learning_rate": 1.2278528730680166e-05, "loss": 1.7481, "step": 21520 }, { "epoch": 1.16, "learning_rate": 1.227493851760102e-05, "loss": 1.6782, "step": 21530 }, { "epoch": 1.16, "learning_rate": 1.2271348304521875e-05, "loss": 1.8494, "step": 21540 }, { "epoch": 1.16, "learning_rate": 1.2267758091442727e-05, "loss": 1.7176, "step": 21550 }, { "epoch": 1.16, "learning_rate": 1.2264167878363581e-05, "loss": 1.7551, "step": 21560 }, { "epoch": 1.16, "learning_rate": 1.2260577665284437e-05, "loss": 1.6861, "step": 21570 }, { "epoch": 1.16, "learning_rate": 1.2256987452205289e-05, "loss": 1.8086, "step": 21580 }, { "epoch": 1.16, "learning_rate": 1.2253397239126143e-05, "loss": 1.7569, "step": 21590 }, { "epoch": 1.16, "learning_rate": 1.2249807026046997e-05, "loss": 1.7652, "step": 21600 }, { "epoch": 1.16, "learning_rate": 1.224621681296785e-05, "loss": 1.8044, "step": 21610 }, { "epoch": 1.16, "learning_rate": 1.2242626599888704e-05, "loss": 1.7165, "step": 21620 }, { "epoch": 1.16, "learning_rate": 1.2239036386809558e-05, "loss": 1.834, "step": 21630 }, { "epoch": 1.17, "learning_rate": 1.2235805195038326e-05, "loss": 1.9121, "step": 21640 }, { "epoch": 1.17, "learning_rate": 1.2232214981959181e-05, "loss": 2.1681, "step": 21650 }, { "epoch": 1.17, "learning_rate": 1.2228624768880034e-05, "loss": 1.9449, "step": 21660 }, { "epoch": 1.17, "learning_rate": 1.2225034555800887e-05, "loss": 1.7481, "step": 21670 }, { "epoch": 1.17, "learning_rate": 1.2221444342721743e-05, "loss": 1.782, "step": 21680 }, { "epoch": 1.17, "learning_rate": 1.2217854129642595e-05, "loss": 1.753, "step": 21690 }, { "epoch": 1.17, "learning_rate": 1.2214263916563449e-05, "loss": 1.7713, "step": 21700 }, { "epoch": 1.17, "learning_rate": 1.2210673703484304e-05, "loss": 1.6513, "step": 21710 }, { "epoch": 1.17, "learning_rate": 1.2207083490405157e-05, "loss": 1.6801, "step": 21720 }, { "epoch": 1.17, "learning_rate": 1.220349327732601e-05, "loss": 1.7796, "step": 21730 }, { "epoch": 1.17, "learning_rate": 1.2199903064246863e-05, "loss": 1.784, "step": 21740 }, { "epoch": 1.17, "learning_rate": 1.2196312851167718e-05, "loss": 1.6147, "step": 21750 }, { "epoch": 1.17, "learning_rate": 1.2192722638088572e-05, "loss": 1.6939, "step": 21760 }, { "epoch": 1.17, "learning_rate": 1.2189132425009424e-05, "loss": 1.6936, "step": 21770 }, { "epoch": 1.17, "learning_rate": 1.218554221193028e-05, "loss": 1.829, "step": 21780 }, { "epoch": 1.17, "learning_rate": 1.2181951998851134e-05, "loss": 1.8211, "step": 21790 }, { "epoch": 1.17, "learning_rate": 1.2178361785771986e-05, "loss": 1.8724, "step": 21800 }, { "epoch": 1.17, "learning_rate": 1.2174771572692841e-05, "loss": 1.8389, "step": 21810 }, { "epoch": 1.18, "learning_rate": 1.2171181359613693e-05, "loss": 1.7052, "step": 21820 }, { "epoch": 1.18, "learning_rate": 1.2167591146534547e-05, "loss": 1.8421, "step": 21830 }, { "epoch": 1.18, "learning_rate": 1.2164000933455403e-05, "loss": 1.6461, "step": 21840 }, { "epoch": 1.18, "learning_rate": 1.2160410720376255e-05, "loss": 1.6211, "step": 21850 }, { "epoch": 1.18, "learning_rate": 1.2156820507297109e-05, "loss": 1.761, "step": 21860 }, { "epoch": 1.18, "learning_rate": 1.2153230294217964e-05, "loss": 1.7755, "step": 21870 }, { "epoch": 1.18, "learning_rate": 1.2149640081138817e-05, "loss": 1.7774, "step": 21880 }, { "epoch": 1.18, "learning_rate": 1.214604986805967e-05, "loss": 1.7768, "step": 21890 }, { "epoch": 1.18, "learning_rate": 1.2142459654980526e-05, "loss": 1.6478, "step": 21900 }, { "epoch": 1.18, "learning_rate": 1.2138869441901378e-05, "loss": 1.7341, "step": 21910 }, { "epoch": 1.18, "learning_rate": 1.2135279228822232e-05, "loss": 1.7083, "step": 21920 }, { "epoch": 1.18, "learning_rate": 1.2131689015743084e-05, "loss": 1.7937, "step": 21930 }, { "epoch": 1.18, "learning_rate": 1.212809880266394e-05, "loss": 1.7279, "step": 21940 }, { "epoch": 1.18, "learning_rate": 1.2124508589584794e-05, "loss": 1.7483, "step": 21950 }, { "epoch": 1.18, "learning_rate": 1.2120918376505646e-05, "loss": 1.7798, "step": 21960 }, { "epoch": 1.18, "learning_rate": 1.2117328163426501e-05, "loss": 1.8101, "step": 21970 }, { "epoch": 1.18, "learning_rate": 1.2113737950347355e-05, "loss": 1.5854, "step": 21980 }, { "epoch": 1.18, "learning_rate": 1.2110147737268207e-05, "loss": 1.6443, "step": 21990 }, { "epoch": 1.18, "learning_rate": 1.2106557524189063e-05, "loss": 1.701, "step": 22000 }, { "epoch": 1.19, "learning_rate": 1.2102967311109915e-05, "loss": 1.7943, "step": 22010 }, { "epoch": 1.19, "learning_rate": 1.2099377098030769e-05, "loss": 1.7075, "step": 22020 }, { "epoch": 1.19, "learning_rate": 1.2095786884951624e-05, "loss": 1.8885, "step": 22030 }, { "epoch": 1.19, "learning_rate": 1.2092196671872476e-05, "loss": 1.6302, "step": 22040 }, { "epoch": 1.19, "learning_rate": 1.208860645879333e-05, "loss": 1.7562, "step": 22050 }, { "epoch": 1.19, "learning_rate": 1.2085016245714186e-05, "loss": 1.8803, "step": 22060 }, { "epoch": 1.19, "learning_rate": 1.2081426032635038e-05, "loss": 1.8528, "step": 22070 }, { "epoch": 1.19, "learning_rate": 1.2077835819555892e-05, "loss": 1.813, "step": 22080 }, { "epoch": 1.19, "learning_rate": 1.2074245606476744e-05, "loss": 1.6714, "step": 22090 }, { "epoch": 1.19, "learning_rate": 1.20706553933976e-05, "loss": 1.8006, "step": 22100 }, { "epoch": 1.19, "learning_rate": 1.2067065180318453e-05, "loss": 1.626, "step": 22110 }, { "epoch": 1.19, "learning_rate": 1.2063474967239306e-05, "loss": 1.7842, "step": 22120 }, { "epoch": 1.19, "learning_rate": 1.2059884754160161e-05, "loss": 1.5872, "step": 22130 }, { "epoch": 1.19, "learning_rate": 1.2056294541081015e-05, "loss": 1.7081, "step": 22140 }, { "epoch": 1.19, "learning_rate": 1.2052704328001867e-05, "loss": 1.7212, "step": 22150 }, { "epoch": 1.19, "learning_rate": 1.2049114114922723e-05, "loss": 1.7808, "step": 22160 }, { "epoch": 1.19, "learning_rate": 1.2045523901843575e-05, "loss": 1.7566, "step": 22170 }, { "epoch": 1.19, "learning_rate": 1.2041933688764429e-05, "loss": 1.7821, "step": 22180 }, { "epoch": 1.19, "learning_rate": 1.2038343475685284e-05, "loss": 1.831, "step": 22190 }, { "epoch": 1.2, "learning_rate": 1.2034753262606136e-05, "loss": 1.8323, "step": 22200 }, { "epoch": 1.2, "learning_rate": 1.203116304952699e-05, "loss": 1.8933, "step": 22210 }, { "epoch": 1.2, "learning_rate": 1.2027572836447846e-05, "loss": 1.7978, "step": 22220 }, { "epoch": 1.2, "learning_rate": 1.2023982623368698e-05, "loss": 1.8204, "step": 22230 }, { "epoch": 1.2, "learning_rate": 1.2020392410289552e-05, "loss": 1.5931, "step": 22240 }, { "epoch": 1.2, "learning_rate": 1.2016802197210407e-05, "loss": 1.715, "step": 22250 }, { "epoch": 1.2, "learning_rate": 1.201321198413126e-05, "loss": 1.7191, "step": 22260 }, { "epoch": 1.2, "learning_rate": 1.2009621771052113e-05, "loss": 1.6852, "step": 22270 }, { "epoch": 1.2, "learning_rate": 1.2006031557972965e-05, "loss": 1.7392, "step": 22280 }, { "epoch": 1.2, "learning_rate": 1.2002441344893821e-05, "loss": 1.7142, "step": 22290 }, { "epoch": 1.2, "learning_rate": 1.1998851131814675e-05, "loss": 1.8304, "step": 22300 }, { "epoch": 1.2, "learning_rate": 1.1995260918735527e-05, "loss": 1.7863, "step": 22310 }, { "epoch": 1.2, "learning_rate": 1.1991670705656383e-05, "loss": 1.7987, "step": 22320 }, { "epoch": 1.2, "learning_rate": 1.1988080492577236e-05, "loss": 1.8618, "step": 22330 }, { "epoch": 1.2, "learning_rate": 1.1984490279498089e-05, "loss": 1.6974, "step": 22340 }, { "epoch": 1.2, "learning_rate": 1.1980900066418944e-05, "loss": 1.8575, "step": 22350 }, { "epoch": 1.2, "learning_rate": 1.1977309853339796e-05, "loss": 1.6146, "step": 22360 }, { "epoch": 1.2, "learning_rate": 1.197371964026065e-05, "loss": 1.6902, "step": 22370 }, { "epoch": 1.21, "learning_rate": 1.1970129427181506e-05, "loss": 1.6845, "step": 22380 }, { "epoch": 1.21, "learning_rate": 1.1966539214102358e-05, "loss": 1.6762, "step": 22390 }, { "epoch": 1.21, "learning_rate": 1.1962949001023212e-05, "loss": 1.8003, "step": 22400 }, { "epoch": 1.21, "learning_rate": 1.1959358787944067e-05, "loss": 1.6465, "step": 22410 }, { "epoch": 1.21, "learning_rate": 1.195576857486492e-05, "loss": 1.8532, "step": 22420 }, { "epoch": 1.21, "learning_rate": 1.1952178361785773e-05, "loss": 1.6153, "step": 22430 }, { "epoch": 1.21, "learning_rate": 1.1948588148706625e-05, "loss": 1.7068, "step": 22440 }, { "epoch": 1.21, "learning_rate": 1.1944997935627481e-05, "loss": 1.776, "step": 22450 }, { "epoch": 1.21, "learning_rate": 1.1941407722548335e-05, "loss": 1.7797, "step": 22460 }, { "epoch": 1.21, "learning_rate": 1.1937817509469187e-05, "loss": 1.691, "step": 22470 }, { "epoch": 1.21, "learning_rate": 1.1934227296390043e-05, "loss": 1.8263, "step": 22480 }, { "epoch": 1.21, "learning_rate": 1.1930637083310896e-05, "loss": 1.6666, "step": 22490 }, { "epoch": 1.21, "learning_rate": 1.1927046870231749e-05, "loss": 1.8092, "step": 22500 }, { "epoch": 1.21, "learning_rate": 1.1923456657152604e-05, "loss": 1.6762, "step": 22510 }, { "epoch": 1.21, "learning_rate": 1.1919866444073456e-05, "loss": 1.7895, "step": 22520 }, { "epoch": 1.21, "learning_rate": 1.191627623099431e-05, "loss": 1.7721, "step": 22530 }, { "epoch": 1.21, "learning_rate": 1.1912686017915166e-05, "loss": 1.7304, "step": 22540 }, { "epoch": 1.21, "learning_rate": 1.1909095804836018e-05, "loss": 1.6836, "step": 22550 }, { "epoch": 1.21, "learning_rate": 1.1905505591756872e-05, "loss": 1.7381, "step": 22560 }, { "epoch": 1.22, "learning_rate": 1.1901915378677727e-05, "loss": 1.6696, "step": 22570 }, { "epoch": 1.22, "learning_rate": 1.189832516559858e-05, "loss": 1.8457, "step": 22580 }, { "epoch": 1.22, "learning_rate": 1.1894734952519433e-05, "loss": 1.7814, "step": 22590 }, { "epoch": 1.22, "learning_rate": 1.1891144739440285e-05, "loss": 1.6799, "step": 22600 }, { "epoch": 1.22, "learning_rate": 1.1887554526361141e-05, "loss": 1.802, "step": 22610 }, { "epoch": 1.22, "learning_rate": 1.1883964313281995e-05, "loss": 1.7901, "step": 22620 }, { "epoch": 1.22, "learning_rate": 1.1880374100202847e-05, "loss": 1.7148, "step": 22630 }, { "epoch": 1.22, "learning_rate": 1.1876783887123702e-05, "loss": 1.7515, "step": 22640 }, { "epoch": 1.22, "learning_rate": 1.1873193674044556e-05, "loss": 1.7677, "step": 22650 }, { "epoch": 1.22, "learning_rate": 1.1869603460965408e-05, "loss": 1.6447, "step": 22660 }, { "epoch": 1.22, "learning_rate": 1.1866013247886264e-05, "loss": 1.7662, "step": 22670 }, { "epoch": 1.22, "learning_rate": 1.1862423034807118e-05, "loss": 1.6552, "step": 22680 }, { "epoch": 1.22, "learning_rate": 1.185883282172797e-05, "loss": 1.7894, "step": 22690 }, { "epoch": 1.22, "learning_rate": 1.1855242608648826e-05, "loss": 1.8249, "step": 22700 }, { "epoch": 1.22, "learning_rate": 1.1851652395569678e-05, "loss": 1.7412, "step": 22710 }, { "epoch": 1.22, "learning_rate": 1.1848062182490532e-05, "loss": 1.74, "step": 22720 }, { "epoch": 1.22, "learning_rate": 1.1844471969411387e-05, "loss": 1.6933, "step": 22730 }, { "epoch": 1.22, "learning_rate": 1.184088175633224e-05, "loss": 1.7274, "step": 22740 }, { "epoch": 1.23, "learning_rate": 1.1837291543253093e-05, "loss": 1.7166, "step": 22750 }, { "epoch": 1.23, "learning_rate": 1.1833701330173949e-05, "loss": 1.6276, "step": 22760 }, { "epoch": 1.23, "learning_rate": 1.18301111170948e-05, "loss": 1.7742, "step": 22770 }, { "epoch": 1.23, "learning_rate": 1.1826520904015655e-05, "loss": 1.6555, "step": 22780 }, { "epoch": 1.23, "learning_rate": 1.1822930690936507e-05, "loss": 1.6934, "step": 22790 }, { "epoch": 1.23, "learning_rate": 1.1819340477857362e-05, "loss": 1.6607, "step": 22800 }, { "epoch": 1.23, "learning_rate": 1.1815750264778216e-05, "loss": 1.7239, "step": 22810 }, { "epoch": 1.23, "learning_rate": 1.1812160051699068e-05, "loss": 1.7697, "step": 22820 }, { "epoch": 1.23, "learning_rate": 1.1808569838619924e-05, "loss": 1.7235, "step": 22830 }, { "epoch": 1.23, "learning_rate": 1.1804979625540778e-05, "loss": 1.7153, "step": 22840 }, { "epoch": 1.23, "learning_rate": 1.180138941246163e-05, "loss": 1.7798, "step": 22850 }, { "epoch": 1.23, "learning_rate": 1.1797799199382485e-05, "loss": 1.7678, "step": 22860 }, { "epoch": 1.23, "learning_rate": 1.1794208986303338e-05, "loss": 1.7366, "step": 22870 }, { "epoch": 1.23, "learning_rate": 1.1790618773224191e-05, "loss": 1.7102, "step": 22880 }, { "epoch": 1.23, "learning_rate": 1.1787028560145047e-05, "loss": 1.7875, "step": 22890 }, { "epoch": 1.23, "learning_rate": 1.17834383470659e-05, "loss": 1.8191, "step": 22900 }, { "epoch": 1.23, "learning_rate": 1.1779848133986753e-05, "loss": 1.7766, "step": 22910 }, { "epoch": 1.23, "learning_rate": 1.1776257920907609e-05, "loss": 1.6979, "step": 22920 }, { "epoch": 1.23, "learning_rate": 1.177266770782846e-05, "loss": 1.6685, "step": 22930 }, { "epoch": 1.24, "learning_rate": 1.1769077494749315e-05, "loss": 1.7044, "step": 22940 }, { "epoch": 1.24, "learning_rate": 1.1765487281670167e-05, "loss": 1.7609, "step": 22950 }, { "epoch": 1.24, "learning_rate": 1.1761897068591022e-05, "loss": 1.7921, "step": 22960 }, { "epoch": 1.24, "learning_rate": 1.1758306855511876e-05, "loss": 1.7488, "step": 22970 }, { "epoch": 1.24, "learning_rate": 1.1754716642432728e-05, "loss": 1.7598, "step": 22980 }, { "epoch": 1.24, "learning_rate": 1.1751126429353584e-05, "loss": 1.6986, "step": 22990 }, { "epoch": 1.24, "learning_rate": 1.1747536216274438e-05, "loss": 1.7319, "step": 23000 }, { "epoch": 1.24, "learning_rate": 1.174394600319529e-05, "loss": 1.7286, "step": 23010 }, { "epoch": 1.24, "learning_rate": 1.1740355790116145e-05, "loss": 1.7466, "step": 23020 }, { "epoch": 1.24, "learning_rate": 1.1736765577036998e-05, "loss": 1.7435, "step": 23030 }, { "epoch": 1.24, "learning_rate": 1.1733175363957851e-05, "loss": 1.7518, "step": 23040 }, { "epoch": 1.24, "learning_rate": 1.1729585150878707e-05, "loss": 1.6624, "step": 23050 }, { "epoch": 1.24, "learning_rate": 1.1725994937799559e-05, "loss": 1.656, "step": 23060 }, { "epoch": 1.24, "learning_rate": 1.1722404724720413e-05, "loss": 1.7089, "step": 23070 }, { "epoch": 1.24, "learning_rate": 1.1718814511641268e-05, "loss": 1.6022, "step": 23080 }, { "epoch": 1.24, "learning_rate": 1.171522429856212e-05, "loss": 1.6208, "step": 23090 }, { "epoch": 1.24, "learning_rate": 1.1711634085482974e-05, "loss": 1.9315, "step": 23100 }, { "epoch": 1.24, "learning_rate": 1.170804387240383e-05, "loss": 1.7092, "step": 23110 }, { "epoch": 1.25, "learning_rate": 1.1704453659324682e-05, "loss": 1.7155, "step": 23120 }, { "epoch": 1.25, "learning_rate": 1.1700863446245536e-05, "loss": 1.5854, "step": 23130 }, { "epoch": 1.25, "learning_rate": 1.1697273233166388e-05, "loss": 1.6777, "step": 23140 }, { "epoch": 1.25, "learning_rate": 1.1693683020087244e-05, "loss": 1.7284, "step": 23150 }, { "epoch": 1.25, "learning_rate": 1.1690092807008098e-05, "loss": 1.7222, "step": 23160 }, { "epoch": 1.25, "learning_rate": 1.168650259392895e-05, "loss": 1.7567, "step": 23170 }, { "epoch": 1.25, "learning_rate": 1.1682912380849805e-05, "loss": 1.7747, "step": 23180 }, { "epoch": 1.25, "learning_rate": 1.1679322167770659e-05, "loss": 1.6788, "step": 23190 }, { "epoch": 1.25, "learning_rate": 1.1675731954691511e-05, "loss": 1.7406, "step": 23200 }, { "epoch": 1.25, "learning_rate": 1.1672141741612367e-05, "loss": 1.6847, "step": 23210 }, { "epoch": 1.25, "learning_rate": 1.1668551528533219e-05, "loss": 1.9008, "step": 23220 }, { "epoch": 1.25, "learning_rate": 1.1664961315454073e-05, "loss": 1.7076, "step": 23230 }, { "epoch": 1.25, "learning_rate": 1.1661371102374928e-05, "loss": 1.6606, "step": 23240 }, { "epoch": 1.25, "learning_rate": 1.165778088929578e-05, "loss": 1.7151, "step": 23250 }, { "epoch": 1.25, "learning_rate": 1.1654190676216634e-05, "loss": 1.8361, "step": 23260 }, { "epoch": 1.25, "learning_rate": 1.165060046313749e-05, "loss": 1.8569, "step": 23270 }, { "epoch": 1.25, "learning_rate": 1.1647010250058342e-05, "loss": 1.7445, "step": 23280 }, { "epoch": 1.25, "learning_rate": 1.1643420036979196e-05, "loss": 1.6659, "step": 23290 }, { "epoch": 1.25, "learning_rate": 1.1639829823900048e-05, "loss": 1.6378, "step": 23300 }, { "epoch": 1.26, "learning_rate": 1.1636239610820904e-05, "loss": 1.8148, "step": 23310 }, { "epoch": 1.26, "learning_rate": 1.1632649397741758e-05, "loss": 1.7935, "step": 23320 }, { "epoch": 1.26, "learning_rate": 1.162905918466261e-05, "loss": 1.8219, "step": 23330 }, { "epoch": 1.26, "learning_rate": 1.1625468971583465e-05, "loss": 1.8538, "step": 23340 }, { "epoch": 1.26, "learning_rate": 1.1621878758504319e-05, "loss": 1.7643, "step": 23350 }, { "epoch": 1.26, "learning_rate": 1.1618288545425171e-05, "loss": 1.6763, "step": 23360 }, { "epoch": 1.26, "learning_rate": 1.1614698332346027e-05, "loss": 1.8272, "step": 23370 }, { "epoch": 1.26, "learning_rate": 1.1611108119266879e-05, "loss": 1.7382, "step": 23380 }, { "epoch": 1.26, "learning_rate": 1.1607517906187733e-05, "loss": 1.699, "step": 23390 }, { "epoch": 1.26, "learning_rate": 1.1603927693108588e-05, "loss": 1.6477, "step": 23400 }, { "epoch": 1.26, "learning_rate": 1.160033748002944e-05, "loss": 1.7422, "step": 23410 }, { "epoch": 1.26, "learning_rate": 1.1596747266950294e-05, "loss": 1.6654, "step": 23420 }, { "epoch": 1.26, "learning_rate": 1.159315705387115e-05, "loss": 1.7119, "step": 23430 }, { "epoch": 1.26, "learning_rate": 1.1589566840792002e-05, "loss": 1.8142, "step": 23440 }, { "epoch": 1.26, "learning_rate": 1.1585976627712856e-05, "loss": 1.8485, "step": 23450 }, { "epoch": 1.26, "learning_rate": 1.1582386414633708e-05, "loss": 1.6374, "step": 23460 }, { "epoch": 1.26, "learning_rate": 1.1578796201554564e-05, "loss": 1.7546, "step": 23470 }, { "epoch": 1.26, "learning_rate": 1.1575205988475417e-05, "loss": 1.6024, "step": 23480 }, { "epoch": 1.26, "learning_rate": 1.157161577539627e-05, "loss": 1.7212, "step": 23490 }, { "epoch": 1.27, "learning_rate": 1.1568025562317125e-05, "loss": 1.6594, "step": 23500 }, { "epoch": 1.27, "learning_rate": 1.1564435349237979e-05, "loss": 1.6101, "step": 23510 }, { "epoch": 1.27, "learning_rate": 1.1560845136158831e-05, "loss": 1.8018, "step": 23520 }, { "epoch": 1.27, "learning_rate": 1.1557254923079687e-05, "loss": 1.7027, "step": 23530 }, { "epoch": 1.27, "learning_rate": 1.155366471000054e-05, "loss": 1.6612, "step": 23540 }, { "epoch": 1.27, "learning_rate": 1.1550074496921393e-05, "loss": 1.7788, "step": 23550 }, { "epoch": 1.27, "learning_rate": 1.1546484283842248e-05, "loss": 1.5938, "step": 23560 }, { "epoch": 1.27, "learning_rate": 1.15428940707631e-05, "loss": 1.7915, "step": 23570 }, { "epoch": 1.27, "learning_rate": 1.1539303857683954e-05, "loss": 1.6829, "step": 23580 }, { "epoch": 1.27, "learning_rate": 1.153571364460481e-05, "loss": 1.6506, "step": 23590 }, { "epoch": 1.27, "learning_rate": 1.1532123431525662e-05, "loss": 1.671, "step": 23600 }, { "epoch": 1.27, "learning_rate": 1.1528533218446516e-05, "loss": 1.7698, "step": 23610 }, { "epoch": 1.27, "learning_rate": 1.1524943005367371e-05, "loss": 1.5942, "step": 23620 }, { "epoch": 1.27, "learning_rate": 1.1521352792288223e-05, "loss": 1.6354, "step": 23630 }, { "epoch": 1.27, "learning_rate": 1.1517762579209077e-05, "loss": 1.7906, "step": 23640 }, { "epoch": 1.27, "learning_rate": 1.151417236612993e-05, "loss": 1.6929, "step": 23650 }, { "epoch": 1.27, "learning_rate": 1.1510582153050785e-05, "loss": 1.8026, "step": 23660 }, { "epoch": 1.27, "learning_rate": 1.1506991939971639e-05, "loss": 1.6577, "step": 23670 }, { "epoch": 1.28, "learning_rate": 1.1503401726892491e-05, "loss": 1.7668, "step": 23680 }, { "epoch": 1.28, "learning_rate": 1.1499811513813347e-05, "loss": 1.659, "step": 23690 }, { "epoch": 1.28, "learning_rate": 1.14962213007342e-05, "loss": 1.7382, "step": 23700 }, { "epoch": 1.28, "learning_rate": 1.1492631087655053e-05, "loss": 1.7625, "step": 23710 }, { "epoch": 1.28, "learning_rate": 1.1489040874575908e-05, "loss": 1.7755, "step": 23720 }, { "epoch": 1.28, "learning_rate": 1.148545066149676e-05, "loss": 1.7658, "step": 23730 }, { "epoch": 1.28, "learning_rate": 1.1481860448417614e-05, "loss": 1.7286, "step": 23740 }, { "epoch": 1.28, "learning_rate": 1.147827023533847e-05, "loss": 1.8849, "step": 23750 }, { "epoch": 1.28, "learning_rate": 1.1474680022259322e-05, "loss": 1.8382, "step": 23760 }, { "epoch": 1.28, "learning_rate": 1.1471089809180176e-05, "loss": 1.8739, "step": 23770 }, { "epoch": 1.28, "learning_rate": 1.1467499596101031e-05, "loss": 1.7771, "step": 23780 }, { "epoch": 1.28, "learning_rate": 1.1463909383021883e-05, "loss": 1.8168, "step": 23790 }, { "epoch": 1.28, "learning_rate": 1.1460319169942737e-05, "loss": 1.7607, "step": 23800 }, { "epoch": 1.28, "learning_rate": 1.145672895686359e-05, "loss": 1.8435, "step": 23810 }, { "epoch": 1.28, "learning_rate": 1.1453138743784445e-05, "loss": 1.7058, "step": 23820 }, { "epoch": 1.28, "learning_rate": 1.1449548530705299e-05, "loss": 1.6961, "step": 23830 }, { "epoch": 1.28, "learning_rate": 1.1445958317626151e-05, "loss": 1.6084, "step": 23840 }, { "epoch": 1.28, "learning_rate": 1.1442368104547007e-05, "loss": 1.784, "step": 23850 }, { "epoch": 1.28, "learning_rate": 1.143877789146786e-05, "loss": 1.6311, "step": 23860 }, { "epoch": 1.29, "learning_rate": 1.1435187678388713e-05, "loss": 1.6207, "step": 23870 }, { "epoch": 1.29, "learning_rate": 1.1431597465309568e-05, "loss": 1.7628, "step": 23880 }, { "epoch": 1.29, "learning_rate": 1.142800725223042e-05, "loss": 1.6632, "step": 23890 }, { "epoch": 1.29, "learning_rate": 1.1424417039151274e-05, "loss": 1.6529, "step": 23900 }, { "epoch": 1.29, "learning_rate": 1.142082682607213e-05, "loss": 1.7288, "step": 23910 }, { "epoch": 1.29, "learning_rate": 1.1417236612992982e-05, "loss": 1.7961, "step": 23920 }, { "epoch": 1.29, "learning_rate": 1.1413646399913836e-05, "loss": 1.7683, "step": 23930 }, { "epoch": 1.29, "learning_rate": 1.1410056186834691e-05, "loss": 1.6715, "step": 23940 }, { "epoch": 1.29, "learning_rate": 1.1406465973755543e-05, "loss": 1.809, "step": 23950 }, { "epoch": 1.29, "learning_rate": 1.1402875760676397e-05, "loss": 1.7715, "step": 23960 }, { "epoch": 1.29, "learning_rate": 1.1399285547597253e-05, "loss": 1.7344, "step": 23970 }, { "epoch": 1.29, "learning_rate": 1.1395695334518105e-05, "loss": 1.7077, "step": 23980 }, { "epoch": 1.29, "learning_rate": 1.1392105121438959e-05, "loss": 1.6736, "step": 23990 }, { "epoch": 1.29, "learning_rate": 1.1388514908359811e-05, "loss": 1.7176, "step": 24000 } ], "max_steps": 55707, "num_train_epochs": 3, "total_flos": 2.91547218375082e+17, "trial_name": null, "trial_params": null }