{ "best_metric": null, "best_model_checkpoint": null, "epoch": 4.0, "eval_steps": 61000, "global_step": 244460, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00016362595107584062, "grad_norm": 6.462503910064697, "learning_rate": 4.0906487768960156e-09, "loss": 0.4892, "step": 10 }, { "epoch": 0.00032725190215168124, "grad_norm": 6.010731220245361, "learning_rate": 8.181297553792031e-09, "loss": 0.4798, "step": 20 }, { "epoch": 0.0004908778532275219, "grad_norm": 7.63542366027832, "learning_rate": 1.2271946330688047e-08, "loss": 0.477, "step": 30 }, { "epoch": 0.0006545038043033625, "grad_norm": 4.759828567504883, "learning_rate": 1.6362595107584062e-08, "loss": 0.4863, "step": 40 }, { "epoch": 0.0008181297553792031, "grad_norm": 10.127399444580078, "learning_rate": 2.045324388448008e-08, "loss": 0.5023, "step": 50 }, { "epoch": 0.0009817557064550437, "grad_norm": 7.6865620613098145, "learning_rate": 2.4543892661376094e-08, "loss": 0.4574, "step": 60 }, { "epoch": 0.0011453816575308843, "grad_norm": 6.511442184448242, "learning_rate": 2.8634541438272112e-08, "loss": 0.4838, "step": 70 }, { "epoch": 0.001309007608606725, "grad_norm": 5.831765651702881, "learning_rate": 3.2725190215168125e-08, "loss": 0.472, "step": 80 }, { "epoch": 0.0014726335596825656, "grad_norm": 6.472559452056885, "learning_rate": 3.6815838992064144e-08, "loss": 0.4623, "step": 90 }, { "epoch": 0.0016362595107584062, "grad_norm": 8.431915283203125, "learning_rate": 4.090648776896016e-08, "loss": 0.4781, "step": 100 }, { "epoch": 0.0017998854618342468, "grad_norm": 7.033874988555908, "learning_rate": 4.4997136545856175e-08, "loss": 0.4614, "step": 110 }, { "epoch": 0.0019635114129100874, "grad_norm": 6.592417240142822, "learning_rate": 4.908778532275219e-08, "loss": 0.4885, "step": 120 }, { "epoch": 0.002127137363985928, "grad_norm": 7.651928424835205, "learning_rate": 5.317843409964821e-08, "loss": 0.455, "step": 130 }, { "epoch": 0.0022907633150617687, "grad_norm": 6.515140533447266, "learning_rate": 5.7269082876544225e-08, "loss": 0.4526, "step": 140 }, { "epoch": 0.0024543892661376093, "grad_norm": 7.782172203063965, "learning_rate": 6.135973165344024e-08, "loss": 0.428, "step": 150 }, { "epoch": 0.00261801521721345, "grad_norm": 7.287140846252441, "learning_rate": 6.545038043033625e-08, "loss": 0.441, "step": 160 }, { "epoch": 0.0027816411682892906, "grad_norm": 5.192974090576172, "learning_rate": 6.954102920723227e-08, "loss": 0.4102, "step": 170 }, { "epoch": 0.002945267119365131, "grad_norm": 8.087858200073242, "learning_rate": 7.363167798412829e-08, "loss": 0.4278, "step": 180 }, { "epoch": 0.003108893070440972, "grad_norm": 6.700540542602539, "learning_rate": 7.77223267610243e-08, "loss": 0.4276, "step": 190 }, { "epoch": 0.0032725190215168124, "grad_norm": 6.550505638122559, "learning_rate": 8.181297553792032e-08, "loss": 0.3764, "step": 200 }, { "epoch": 0.003436144972592653, "grad_norm": 5.587803840637207, "learning_rate": 8.590362431481634e-08, "loss": 0.3503, "step": 210 }, { "epoch": 0.0035997709236684937, "grad_norm": 4.2618536949157715, "learning_rate": 8.999427309171235e-08, "loss": 0.3397, "step": 220 }, { "epoch": 0.0037633968747443343, "grad_norm": 4.231817245483398, "learning_rate": 9.408492186860837e-08, "loss": 0.327, "step": 230 }, { "epoch": 0.003927022825820175, "grad_norm": 5.085544109344482, "learning_rate": 9.817557064550437e-08, "loss": 0.3113, "step": 240 }, { "epoch": 0.004090648776896016, "grad_norm": 3.684473991394043, "learning_rate": 1.022662194224004e-07, "loss": 0.2757, "step": 250 }, { "epoch": 0.004254274727971856, "grad_norm": 3.9479005336761475, "learning_rate": 1.0635686819929643e-07, "loss": 0.2883, "step": 260 }, { "epoch": 0.004417900679047697, "grad_norm": 2.5990707874298096, "learning_rate": 1.1044751697619242e-07, "loss": 0.2506, "step": 270 }, { "epoch": 0.004581526630123537, "grad_norm": 2.5024545192718506, "learning_rate": 1.1453816575308845e-07, "loss": 0.2131, "step": 280 }, { "epoch": 0.004745152581199378, "grad_norm": 2.5615949630737305, "learning_rate": 1.1862881452998446e-07, "loss": 0.1941, "step": 290 }, { "epoch": 0.004908778532275219, "grad_norm": 2.0188534259796143, "learning_rate": 1.2271946330688047e-07, "loss": 0.1851, "step": 300 }, { "epoch": 0.00507240448335106, "grad_norm": 2.1463961601257324, "learning_rate": 1.268101120837765e-07, "loss": 0.181, "step": 310 }, { "epoch": 0.0052360304344269, "grad_norm": 2.032957077026367, "learning_rate": 1.309007608606725e-07, "loss": 0.1573, "step": 320 }, { "epoch": 0.005399656385502741, "grad_norm": 1.6918240785598755, "learning_rate": 1.3499140963756852e-07, "loss": 0.1551, "step": 330 }, { "epoch": 0.005563282336578581, "grad_norm": 1.6664615869522095, "learning_rate": 1.3908205841446455e-07, "loss": 0.1305, "step": 340 }, { "epoch": 0.005726908287654422, "grad_norm": 2.0269060134887695, "learning_rate": 1.4317270719136055e-07, "loss": 0.1523, "step": 350 }, { "epoch": 0.005890534238730262, "grad_norm": 1.656173586845398, "learning_rate": 1.4726335596825657e-07, "loss": 0.1388, "step": 360 }, { "epoch": 0.006054160189806103, "grad_norm": 1.7295620441436768, "learning_rate": 1.513540047451526e-07, "loss": 0.1357, "step": 370 }, { "epoch": 0.006217786140881944, "grad_norm": 2.333462953567505, "learning_rate": 1.554446535220486e-07, "loss": 0.1285, "step": 380 }, { "epoch": 0.006381412091957785, "grad_norm": 1.943618655204773, "learning_rate": 1.5953530229894462e-07, "loss": 0.1001, "step": 390 }, { "epoch": 0.006545038043033625, "grad_norm": 1.6467885971069336, "learning_rate": 1.6362595107584065e-07, "loss": 0.1306, "step": 400 }, { "epoch": 0.006708663994109466, "grad_norm": 1.7440624237060547, "learning_rate": 1.6771659985273668e-07, "loss": 0.1139, "step": 410 }, { "epoch": 0.006872289945185306, "grad_norm": 1.743868112564087, "learning_rate": 1.7180724862963267e-07, "loss": 0.1128, "step": 420 }, { "epoch": 0.007035915896261147, "grad_norm": 1.8379240036010742, "learning_rate": 1.7589789740652867e-07, "loss": 0.0942, "step": 430 }, { "epoch": 0.007199541847336987, "grad_norm": 1.4758986234664917, "learning_rate": 1.799885461834247e-07, "loss": 0.1107, "step": 440 }, { "epoch": 0.007363167798412828, "grad_norm": 1.4652776718139648, "learning_rate": 1.8407919496032072e-07, "loss": 0.1063, "step": 450 }, { "epoch": 0.0075267937494886685, "grad_norm": 1.6025725603103638, "learning_rate": 1.8816984373721675e-07, "loss": 0.0973, "step": 460 }, { "epoch": 0.00769041970056451, "grad_norm": 1.2838658094406128, "learning_rate": 1.9226049251411275e-07, "loss": 0.0851, "step": 470 }, { "epoch": 0.00785404565164035, "grad_norm": 1.4988064765930176, "learning_rate": 1.9635114129100875e-07, "loss": 0.1003, "step": 480 }, { "epoch": 0.00801767160271619, "grad_norm": 1.7101094722747803, "learning_rate": 2.0044179006790477e-07, "loss": 0.0842, "step": 490 }, { "epoch": 0.008181297553792032, "grad_norm": 2.590806484222412, "learning_rate": 2.045324388448008e-07, "loss": 0.084, "step": 500 }, { "epoch": 0.008344923504867872, "grad_norm": 1.3689161539077759, "learning_rate": 2.0862308762169682e-07, "loss": 0.0781, "step": 510 }, { "epoch": 0.008508549455943712, "grad_norm": 1.7421581745147705, "learning_rate": 2.1271373639859285e-07, "loss": 0.0978, "step": 520 }, { "epoch": 0.008672175407019552, "grad_norm": 1.7616751194000244, "learning_rate": 2.1680438517548885e-07, "loss": 0.0872, "step": 530 }, { "epoch": 0.008835801358095394, "grad_norm": 1.2861080169677734, "learning_rate": 2.2089503395238485e-07, "loss": 0.0758, "step": 540 }, { "epoch": 0.008999427309171235, "grad_norm": 1.2443004846572876, "learning_rate": 2.2498568272928087e-07, "loss": 0.0681, "step": 550 }, { "epoch": 0.009163053260247075, "grad_norm": 1.6169759035110474, "learning_rate": 2.290763315061769e-07, "loss": 0.0763, "step": 560 }, { "epoch": 0.009326679211322917, "grad_norm": 1.7666412591934204, "learning_rate": 2.3316698028307292e-07, "loss": 0.0678, "step": 570 }, { "epoch": 0.009490305162398757, "grad_norm": 1.6621071100234985, "learning_rate": 2.3725762905996892e-07, "loss": 0.0827, "step": 580 }, { "epoch": 0.009653931113474597, "grad_norm": 1.4472569227218628, "learning_rate": 2.413482778368649e-07, "loss": 0.0727, "step": 590 }, { "epoch": 0.009817557064550437, "grad_norm": 1.4462445974349976, "learning_rate": 2.4543892661376095e-07, "loss": 0.0671, "step": 600 }, { "epoch": 0.00998118301562628, "grad_norm": 1.1776124238967896, "learning_rate": 2.49529575390657e-07, "loss": 0.0662, "step": 610 }, { "epoch": 0.01014480896670212, "grad_norm": 1.5504887104034424, "learning_rate": 2.53620224167553e-07, "loss": 0.0656, "step": 620 }, { "epoch": 0.01030843491777796, "grad_norm": 1.54075026512146, "learning_rate": 2.57710872944449e-07, "loss": 0.0615, "step": 630 }, { "epoch": 0.0104720608688538, "grad_norm": 1.2261064052581787, "learning_rate": 2.61801521721345e-07, "loss": 0.0769, "step": 640 }, { "epoch": 0.010635686819929642, "grad_norm": 1.3045021295547485, "learning_rate": 2.65892170498241e-07, "loss": 0.0532, "step": 650 }, { "epoch": 0.010799312771005482, "grad_norm": 1.0263113975524902, "learning_rate": 2.6998281927513705e-07, "loss": 0.0586, "step": 660 }, { "epoch": 0.010962938722081322, "grad_norm": 1.992857575416565, "learning_rate": 2.740734680520331e-07, "loss": 0.0656, "step": 670 }, { "epoch": 0.011126564673157162, "grad_norm": 1.5376744270324707, "learning_rate": 2.781641168289291e-07, "loss": 0.069, "step": 680 }, { "epoch": 0.011290190624233004, "grad_norm": 1.2060158252716064, "learning_rate": 2.8225476560582507e-07, "loss": 0.0592, "step": 690 }, { "epoch": 0.011453816575308844, "grad_norm": 1.4092153310775757, "learning_rate": 2.863454143827211e-07, "loss": 0.0605, "step": 700 }, { "epoch": 0.011617442526384684, "grad_norm": 1.3497917652130127, "learning_rate": 2.904360631596171e-07, "loss": 0.0577, "step": 710 }, { "epoch": 0.011781068477460525, "grad_norm": 1.2837356328964233, "learning_rate": 2.9452671193651315e-07, "loss": 0.0617, "step": 720 }, { "epoch": 0.011944694428536367, "grad_norm": 1.1854506731033325, "learning_rate": 2.986173607134092e-07, "loss": 0.0533, "step": 730 }, { "epoch": 0.012108320379612207, "grad_norm": 1.338873267173767, "learning_rate": 3.027080094903052e-07, "loss": 0.0621, "step": 740 }, { "epoch": 0.012271946330688047, "grad_norm": 1.3774230480194092, "learning_rate": 3.0679865826720117e-07, "loss": 0.063, "step": 750 }, { "epoch": 0.012435572281763887, "grad_norm": 1.3706445693969727, "learning_rate": 3.108893070440972e-07, "loss": 0.0579, "step": 760 }, { "epoch": 0.012599198232839729, "grad_norm": 1.367600679397583, "learning_rate": 3.149799558209932e-07, "loss": 0.0701, "step": 770 }, { "epoch": 0.01276282418391557, "grad_norm": 1.4418561458587646, "learning_rate": 3.1907060459788925e-07, "loss": 0.0571, "step": 780 }, { "epoch": 0.01292645013499141, "grad_norm": 1.0723310708999634, "learning_rate": 3.231612533747853e-07, "loss": 0.0629, "step": 790 }, { "epoch": 0.01309007608606725, "grad_norm": 1.7749966382980347, "learning_rate": 3.272519021516813e-07, "loss": 0.0581, "step": 800 }, { "epoch": 0.013253702037143092, "grad_norm": 1.0387691259384155, "learning_rate": 3.313425509285773e-07, "loss": 0.0445, "step": 810 }, { "epoch": 0.013417327988218932, "grad_norm": 1.178432822227478, "learning_rate": 3.3543319970547335e-07, "loss": 0.0617, "step": 820 }, { "epoch": 0.013580953939294772, "grad_norm": 1.630421757698059, "learning_rate": 3.395238484823693e-07, "loss": 0.0555, "step": 830 }, { "epoch": 0.013744579890370612, "grad_norm": 1.1873332262039185, "learning_rate": 3.4361449725926535e-07, "loss": 0.0522, "step": 840 }, { "epoch": 0.013908205841446454, "grad_norm": 1.3297563791275024, "learning_rate": 3.477051460361613e-07, "loss": 0.0489, "step": 850 }, { "epoch": 0.014071831792522294, "grad_norm": 1.3731971979141235, "learning_rate": 3.5179579481305735e-07, "loss": 0.0519, "step": 860 }, { "epoch": 0.014235457743598134, "grad_norm": 1.222265362739563, "learning_rate": 3.5588644358995337e-07, "loss": 0.0458, "step": 870 }, { "epoch": 0.014399083694673975, "grad_norm": 1.5020134449005127, "learning_rate": 3.599770923668494e-07, "loss": 0.0577, "step": 880 }, { "epoch": 0.014562709645749817, "grad_norm": 1.040066123008728, "learning_rate": 3.640677411437454e-07, "loss": 0.0401, "step": 890 }, { "epoch": 0.014726335596825657, "grad_norm": 1.1723414659500122, "learning_rate": 3.6815838992064145e-07, "loss": 0.0474, "step": 900 }, { "epoch": 0.014889961547901497, "grad_norm": 1.069694995880127, "learning_rate": 3.722490386975375e-07, "loss": 0.0411, "step": 910 }, { "epoch": 0.015053587498977337, "grad_norm": 1.3606572151184082, "learning_rate": 3.763396874744335e-07, "loss": 0.0455, "step": 920 }, { "epoch": 0.015217213450053179, "grad_norm": 1.4972037076950073, "learning_rate": 3.804303362513295e-07, "loss": 0.0435, "step": 930 }, { "epoch": 0.01538083940112902, "grad_norm": 1.39334237575531, "learning_rate": 3.845209850282255e-07, "loss": 0.0443, "step": 940 }, { "epoch": 0.01554446535220486, "grad_norm": 1.2184045314788818, "learning_rate": 3.886116338051215e-07, "loss": 0.0457, "step": 950 }, { "epoch": 0.0157080913032807, "grad_norm": 1.113869547843933, "learning_rate": 3.927022825820175e-07, "loss": 0.0468, "step": 960 }, { "epoch": 0.01587171725435654, "grad_norm": 1.450608253479004, "learning_rate": 3.967929313589135e-07, "loss": 0.0319, "step": 970 }, { "epoch": 0.01603534320543238, "grad_norm": 1.1869876384735107, "learning_rate": 4.0088358013580955e-07, "loss": 0.0363, "step": 980 }, { "epoch": 0.016198969156508224, "grad_norm": 1.1743232011795044, "learning_rate": 4.0497422891270557e-07, "loss": 0.0566, "step": 990 }, { "epoch": 0.016362595107584064, "grad_norm": 1.1873711347579956, "learning_rate": 4.090648776896016e-07, "loss": 0.0308, "step": 1000 }, { "epoch": 0.016526221058659904, "grad_norm": 1.512641191482544, "learning_rate": 4.131555264664976e-07, "loss": 0.0415, "step": 1010 }, { "epoch": 0.016689847009735744, "grad_norm": 1.2807750701904297, "learning_rate": 4.1724617524339365e-07, "loss": 0.0427, "step": 1020 }, { "epoch": 0.016853472960811584, "grad_norm": 1.2817078828811646, "learning_rate": 4.213368240202897e-07, "loss": 0.044, "step": 1030 }, { "epoch": 0.017017098911887425, "grad_norm": 1.2551891803741455, "learning_rate": 4.254274727971857e-07, "loss": 0.0457, "step": 1040 }, { "epoch": 0.017180724862963265, "grad_norm": 1.2710257768630981, "learning_rate": 4.2951812157408167e-07, "loss": 0.0479, "step": 1050 }, { "epoch": 0.017344350814039105, "grad_norm": 1.4306761026382446, "learning_rate": 4.336087703509777e-07, "loss": 0.037, "step": 1060 }, { "epoch": 0.01750797676511495, "grad_norm": 1.2414277791976929, "learning_rate": 4.376994191278737e-07, "loss": 0.0437, "step": 1070 }, { "epoch": 0.01767160271619079, "grad_norm": 0.9635937213897705, "learning_rate": 4.417900679047697e-07, "loss": 0.0391, "step": 1080 }, { "epoch": 0.01783522866726663, "grad_norm": 1.3641034364700317, "learning_rate": 4.458807166816657e-07, "loss": 0.0367, "step": 1090 }, { "epoch": 0.01799885461834247, "grad_norm": 1.0497424602508545, "learning_rate": 4.4997136545856175e-07, "loss": 0.0489, "step": 1100 }, { "epoch": 0.01816248056941831, "grad_norm": 1.198820948600769, "learning_rate": 4.5406201423545777e-07, "loss": 0.0343, "step": 1110 }, { "epoch": 0.01832610652049415, "grad_norm": 0.7283077239990234, "learning_rate": 4.581526630123538e-07, "loss": 0.0311, "step": 1120 }, { "epoch": 0.01848973247156999, "grad_norm": 1.6635894775390625, "learning_rate": 4.622433117892498e-07, "loss": 0.0426, "step": 1130 }, { "epoch": 0.018653358422645833, "grad_norm": 0.9116859436035156, "learning_rate": 4.6633396056614585e-07, "loss": 0.031, "step": 1140 }, { "epoch": 0.018816984373721674, "grad_norm": 1.0714006423950195, "learning_rate": 4.704246093430419e-07, "loss": 0.0407, "step": 1150 }, { "epoch": 0.018980610324797514, "grad_norm": 0.9308425188064575, "learning_rate": 4.7451525811993785e-07, "loss": 0.0301, "step": 1160 }, { "epoch": 0.019144236275873354, "grad_norm": 1.1003210544586182, "learning_rate": 4.786059068968338e-07, "loss": 0.029, "step": 1170 }, { "epoch": 0.019307862226949194, "grad_norm": 1.092533826828003, "learning_rate": 4.826965556737298e-07, "loss": 0.0315, "step": 1180 }, { "epoch": 0.019471488178025034, "grad_norm": 1.0536627769470215, "learning_rate": 4.867872044506259e-07, "loss": 0.0264, "step": 1190 }, { "epoch": 0.019635114129100874, "grad_norm": 1.1207045316696167, "learning_rate": 4.908778532275219e-07, "loss": 0.0435, "step": 1200 }, { "epoch": 0.019798740080176715, "grad_norm": 1.0102251768112183, "learning_rate": 4.949685020044179e-07, "loss": 0.0351, "step": 1210 }, { "epoch": 0.01996236603125256, "grad_norm": 1.0621376037597656, "learning_rate": 4.99059150781314e-07, "loss": 0.0306, "step": 1220 }, { "epoch": 0.0201259919823284, "grad_norm": 1.04286789894104, "learning_rate": 5.0314979955821e-07, "loss": 0.0344, "step": 1230 }, { "epoch": 0.02028961793340424, "grad_norm": 1.3077759742736816, "learning_rate": 5.07240448335106e-07, "loss": 0.0337, "step": 1240 }, { "epoch": 0.02045324388448008, "grad_norm": 1.0296199321746826, "learning_rate": 5.11331097112002e-07, "loss": 0.0343, "step": 1250 }, { "epoch": 0.02061686983555592, "grad_norm": 0.9476549625396729, "learning_rate": 5.15421745888898e-07, "loss": 0.0356, "step": 1260 }, { "epoch": 0.02078049578663176, "grad_norm": 1.3184281587600708, "learning_rate": 5.19512394665794e-07, "loss": 0.0326, "step": 1270 }, { "epoch": 0.0209441217377076, "grad_norm": 0.9071869850158691, "learning_rate": 5.2360304344269e-07, "loss": 0.0306, "step": 1280 }, { "epoch": 0.02110774768878344, "grad_norm": 1.1159783601760864, "learning_rate": 5.27693692219586e-07, "loss": 0.0372, "step": 1290 }, { "epoch": 0.021271373639859283, "grad_norm": 1.6537309885025024, "learning_rate": 5.31784340996482e-07, "loss": 0.0346, "step": 1300 }, { "epoch": 0.021434999590935123, "grad_norm": 0.9327256679534912, "learning_rate": 5.358749897733781e-07, "loss": 0.0322, "step": 1310 }, { "epoch": 0.021598625542010964, "grad_norm": 1.5405960083007812, "learning_rate": 5.399656385502741e-07, "loss": 0.0346, "step": 1320 }, { "epoch": 0.021762251493086804, "grad_norm": 1.0464937686920166, "learning_rate": 5.440562873271701e-07, "loss": 0.0258, "step": 1330 }, { "epoch": 0.021925877444162644, "grad_norm": 0.9697508215904236, "learning_rate": 5.481469361040661e-07, "loss": 0.0254, "step": 1340 }, { "epoch": 0.022089503395238484, "grad_norm": 1.1420433521270752, "learning_rate": 5.522375848809622e-07, "loss": 0.0342, "step": 1350 }, { "epoch": 0.022253129346314324, "grad_norm": 1.1603723764419556, "learning_rate": 5.563282336578582e-07, "loss": 0.0328, "step": 1360 }, { "epoch": 0.022416755297390165, "grad_norm": 1.5856233835220337, "learning_rate": 5.604188824347542e-07, "loss": 0.0403, "step": 1370 }, { "epoch": 0.022580381248466008, "grad_norm": 1.0629204511642456, "learning_rate": 5.645095312116501e-07, "loss": 0.036, "step": 1380 }, { "epoch": 0.02274400719954185, "grad_norm": 1.0553789138793945, "learning_rate": 5.686001799885462e-07, "loss": 0.0301, "step": 1390 }, { "epoch": 0.02290763315061769, "grad_norm": 1.2036844491958618, "learning_rate": 5.726908287654422e-07, "loss": 0.0327, "step": 1400 }, { "epoch": 0.02307125910169353, "grad_norm": 1.0642132759094238, "learning_rate": 5.767814775423382e-07, "loss": 0.0341, "step": 1410 }, { "epoch": 0.02323488505276937, "grad_norm": 0.8616418838500977, "learning_rate": 5.808721263192342e-07, "loss": 0.0338, "step": 1420 }, { "epoch": 0.02339851100384521, "grad_norm": 1.2722835540771484, "learning_rate": 5.849627750961303e-07, "loss": 0.0319, "step": 1430 }, { "epoch": 0.02356213695492105, "grad_norm": 0.9430190324783325, "learning_rate": 5.890534238730263e-07, "loss": 0.0373, "step": 1440 }, { "epoch": 0.02372576290599689, "grad_norm": 1.1763486862182617, "learning_rate": 5.931440726499223e-07, "loss": 0.0287, "step": 1450 }, { "epoch": 0.023889388857072733, "grad_norm": 0.8227166533470154, "learning_rate": 5.972347214268183e-07, "loss": 0.0298, "step": 1460 }, { "epoch": 0.024053014808148573, "grad_norm": 1.1035717725753784, "learning_rate": 6.013253702037144e-07, "loss": 0.0222, "step": 1470 }, { "epoch": 0.024216640759224414, "grad_norm": 1.4184423685073853, "learning_rate": 6.054160189806104e-07, "loss": 0.0316, "step": 1480 }, { "epoch": 0.024380266710300254, "grad_norm": 0.9370906949043274, "learning_rate": 6.095066677575063e-07, "loss": 0.0255, "step": 1490 }, { "epoch": 0.024543892661376094, "grad_norm": 1.0215215682983398, "learning_rate": 6.135973165344023e-07, "loss": 0.0271, "step": 1500 }, { "epoch": 0.024707518612451934, "grad_norm": 1.2105132341384888, "learning_rate": 6.176879653112984e-07, "loss": 0.0265, "step": 1510 }, { "epoch": 0.024871144563527774, "grad_norm": 1.1781824827194214, "learning_rate": 6.217786140881944e-07, "loss": 0.0301, "step": 1520 }, { "epoch": 0.025034770514603615, "grad_norm": 0.9946390986442566, "learning_rate": 6.258692628650904e-07, "loss": 0.036, "step": 1530 }, { "epoch": 0.025198396465679458, "grad_norm": 0.7864744663238525, "learning_rate": 6.299599116419864e-07, "loss": 0.0386, "step": 1540 }, { "epoch": 0.0253620224167553, "grad_norm": 1.0190725326538086, "learning_rate": 6.340505604188825e-07, "loss": 0.0294, "step": 1550 }, { "epoch": 0.02552564836783114, "grad_norm": 0.9382844567298889, "learning_rate": 6.381412091957785e-07, "loss": 0.0264, "step": 1560 }, { "epoch": 0.02568927431890698, "grad_norm": 0.8204848170280457, "learning_rate": 6.422318579726745e-07, "loss": 0.0257, "step": 1570 }, { "epoch": 0.02585290026998282, "grad_norm": 0.9049162268638611, "learning_rate": 6.463225067495705e-07, "loss": 0.0249, "step": 1580 }, { "epoch": 0.02601652622105866, "grad_norm": 0.8795179724693298, "learning_rate": 6.504131555264666e-07, "loss": 0.0302, "step": 1590 }, { "epoch": 0.0261801521721345, "grad_norm": 0.7912503480911255, "learning_rate": 6.545038043033626e-07, "loss": 0.031, "step": 1600 }, { "epoch": 0.02634377812321034, "grad_norm": 1.1222286224365234, "learning_rate": 6.585944530802586e-07, "loss": 0.0275, "step": 1610 }, { "epoch": 0.026507404074286183, "grad_norm": 0.953707218170166, "learning_rate": 6.626851018571547e-07, "loss": 0.025, "step": 1620 }, { "epoch": 0.026671030025362023, "grad_norm": 0.8810185790061951, "learning_rate": 6.667757506340507e-07, "loss": 0.0189, "step": 1630 }, { "epoch": 0.026834655976437864, "grad_norm": 1.1936429738998413, "learning_rate": 6.708663994109467e-07, "loss": 0.0254, "step": 1640 }, { "epoch": 0.026998281927513704, "grad_norm": 1.115318775177002, "learning_rate": 6.749570481878426e-07, "loss": 0.0296, "step": 1650 }, { "epoch": 0.027161907878589544, "grad_norm": 1.1250269412994385, "learning_rate": 6.790476969647386e-07, "loss": 0.0289, "step": 1660 }, { "epoch": 0.027325533829665384, "grad_norm": 1.0015441179275513, "learning_rate": 6.831383457416347e-07, "loss": 0.026, "step": 1670 }, { "epoch": 0.027489159780741224, "grad_norm": 1.0860515832901, "learning_rate": 6.872289945185307e-07, "loss": 0.0299, "step": 1680 }, { "epoch": 0.027652785731817068, "grad_norm": 0.9313044548034668, "learning_rate": 6.913196432954266e-07, "loss": 0.0273, "step": 1690 }, { "epoch": 0.027816411682892908, "grad_norm": 1.0411732196807861, "learning_rate": 6.954102920723226e-07, "loss": 0.0269, "step": 1700 }, { "epoch": 0.02798003763396875, "grad_norm": 0.9204750061035156, "learning_rate": 6.995009408492187e-07, "loss": 0.0285, "step": 1710 }, { "epoch": 0.02814366358504459, "grad_norm": 0.8619980216026306, "learning_rate": 7.035915896261147e-07, "loss": 0.0335, "step": 1720 }, { "epoch": 0.02830728953612043, "grad_norm": 1.1429946422576904, "learning_rate": 7.076822384030107e-07, "loss": 0.0219, "step": 1730 }, { "epoch": 0.02847091548719627, "grad_norm": 0.5820915699005127, "learning_rate": 7.117728871799067e-07, "loss": 0.021, "step": 1740 }, { "epoch": 0.02863454143827211, "grad_norm": 1.1792618036270142, "learning_rate": 7.158635359568028e-07, "loss": 0.0227, "step": 1750 }, { "epoch": 0.02879816738934795, "grad_norm": 0.8747755289077759, "learning_rate": 7.199541847336988e-07, "loss": 0.0269, "step": 1760 }, { "epoch": 0.028961793340423793, "grad_norm": 1.06398606300354, "learning_rate": 7.240448335105948e-07, "loss": 0.032, "step": 1770 }, { "epoch": 0.029125419291499633, "grad_norm": 1.0716880559921265, "learning_rate": 7.281354822874908e-07, "loss": 0.0284, "step": 1780 }, { "epoch": 0.029289045242575473, "grad_norm": 0.965604305267334, "learning_rate": 7.322261310643869e-07, "loss": 0.0238, "step": 1790 }, { "epoch": 0.029452671193651313, "grad_norm": 0.965664267539978, "learning_rate": 7.363167798412829e-07, "loss": 0.0255, "step": 1800 }, { "epoch": 0.029616297144727154, "grad_norm": 1.0312159061431885, "learning_rate": 7.404074286181789e-07, "loss": 0.0156, "step": 1810 }, { "epoch": 0.029779923095802994, "grad_norm": 1.1121737957000732, "learning_rate": 7.44498077395075e-07, "loss": 0.0174, "step": 1820 }, { "epoch": 0.029943549046878834, "grad_norm": 1.2629731893539429, "learning_rate": 7.48588726171971e-07, "loss": 0.0304, "step": 1830 }, { "epoch": 0.030107174997954674, "grad_norm": 0.9692763090133667, "learning_rate": 7.52679374948867e-07, "loss": 0.0202, "step": 1840 }, { "epoch": 0.030270800949030518, "grad_norm": 0.7943910360336304, "learning_rate": 7.56770023725763e-07, "loss": 0.0258, "step": 1850 }, { "epoch": 0.030434426900106358, "grad_norm": 0.8956258893013, "learning_rate": 7.60860672502659e-07, "loss": 0.0232, "step": 1860 }, { "epoch": 0.030598052851182198, "grad_norm": 0.786642849445343, "learning_rate": 7.64951321279555e-07, "loss": 0.032, "step": 1870 }, { "epoch": 0.03076167880225804, "grad_norm": 0.9094133973121643, "learning_rate": 7.69041970056451e-07, "loss": 0.0298, "step": 1880 }, { "epoch": 0.03092530475333388, "grad_norm": 0.8209534287452698, "learning_rate": 7.73132618833347e-07, "loss": 0.0229, "step": 1890 }, { "epoch": 0.03108893070440972, "grad_norm": 0.9857476949691772, "learning_rate": 7.77223267610243e-07, "loss": 0.0224, "step": 1900 }, { "epoch": 0.03125255665548556, "grad_norm": 1.1389381885528564, "learning_rate": 7.813139163871391e-07, "loss": 0.0245, "step": 1910 }, { "epoch": 0.0314161826065614, "grad_norm": 1.18366277217865, "learning_rate": 7.85404565164035e-07, "loss": 0.0219, "step": 1920 }, { "epoch": 0.03157980855763724, "grad_norm": 1.040653109550476, "learning_rate": 7.89495213940931e-07, "loss": 0.0217, "step": 1930 }, { "epoch": 0.03174343450871308, "grad_norm": 0.8729476928710938, "learning_rate": 7.93585862717827e-07, "loss": 0.0226, "step": 1940 }, { "epoch": 0.03190706045978892, "grad_norm": 1.0330930948257446, "learning_rate": 7.976765114947231e-07, "loss": 0.0179, "step": 1950 }, { "epoch": 0.03207068641086476, "grad_norm": 0.615511953830719, "learning_rate": 8.017671602716191e-07, "loss": 0.0266, "step": 1960 }, { "epoch": 0.032234312361940604, "grad_norm": 0.7662327885627747, "learning_rate": 8.058578090485151e-07, "loss": 0.0192, "step": 1970 }, { "epoch": 0.03239793831301645, "grad_norm": 0.6774893999099731, "learning_rate": 8.099484578254111e-07, "loss": 0.0263, "step": 1980 }, { "epoch": 0.032561564264092284, "grad_norm": 0.8957571387290955, "learning_rate": 8.140391066023072e-07, "loss": 0.0237, "step": 1990 }, { "epoch": 0.03272519021516813, "grad_norm": 0.8837447762489319, "learning_rate": 8.181297553792032e-07, "loss": 0.0237, "step": 2000 }, { "epoch": 0.032888816166243964, "grad_norm": 1.1863794326782227, "learning_rate": 8.222204041560992e-07, "loss": 0.0275, "step": 2010 }, { "epoch": 0.03305244211731981, "grad_norm": 0.7342771291732788, "learning_rate": 8.263110529329952e-07, "loss": 0.0265, "step": 2020 }, { "epoch": 0.033216068068395645, "grad_norm": 0.7886518836021423, "learning_rate": 8.304017017098913e-07, "loss": 0.0185, "step": 2030 }, { "epoch": 0.03337969401947149, "grad_norm": 0.9511569142341614, "learning_rate": 8.344923504867873e-07, "loss": 0.0208, "step": 2040 }, { "epoch": 0.03354331997054733, "grad_norm": 0.8712888956069946, "learning_rate": 8.385829992636833e-07, "loss": 0.0262, "step": 2050 }, { "epoch": 0.03370694592162317, "grad_norm": 0.6433917284011841, "learning_rate": 8.426736480405793e-07, "loss": 0.0181, "step": 2060 }, { "epoch": 0.03387057187269901, "grad_norm": 0.7254625558853149, "learning_rate": 8.467642968174754e-07, "loss": 0.0198, "step": 2070 }, { "epoch": 0.03403419782377485, "grad_norm": 0.9536096453666687, "learning_rate": 8.508549455943714e-07, "loss": 0.0199, "step": 2080 }, { "epoch": 0.03419782377485069, "grad_norm": 1.5916779041290283, "learning_rate": 8.549455943712673e-07, "loss": 0.0188, "step": 2090 }, { "epoch": 0.03436144972592653, "grad_norm": 0.8822453022003174, "learning_rate": 8.590362431481633e-07, "loss": 0.0189, "step": 2100 }, { "epoch": 0.03452507567700237, "grad_norm": 0.7866227030754089, "learning_rate": 8.631268919250594e-07, "loss": 0.0347, "step": 2110 }, { "epoch": 0.03468870162807821, "grad_norm": 0.8466185331344604, "learning_rate": 8.672175407019554e-07, "loss": 0.0195, "step": 2120 }, { "epoch": 0.034852327579154053, "grad_norm": 0.821190595626831, "learning_rate": 8.713081894788514e-07, "loss": 0.0252, "step": 2130 }, { "epoch": 0.0350159535302299, "grad_norm": 0.5729047060012817, "learning_rate": 8.753988382557474e-07, "loss": 0.0163, "step": 2140 }, { "epoch": 0.035179579481305734, "grad_norm": 0.6806867718696594, "learning_rate": 8.794894870326434e-07, "loss": 0.0274, "step": 2150 }, { "epoch": 0.03534320543238158, "grad_norm": 0.6294782757759094, "learning_rate": 8.835801358095394e-07, "loss": 0.0207, "step": 2160 }, { "epoch": 0.035506831383457414, "grad_norm": 0.7756246328353882, "learning_rate": 8.876707845864354e-07, "loss": 0.0199, "step": 2170 }, { "epoch": 0.03567045733453326, "grad_norm": 0.8734956383705139, "learning_rate": 8.917614333633314e-07, "loss": 0.0234, "step": 2180 }, { "epoch": 0.035834083285609095, "grad_norm": 1.2174321413040161, "learning_rate": 8.958520821402275e-07, "loss": 0.021, "step": 2190 }, { "epoch": 0.03599770923668494, "grad_norm": 0.9613479971885681, "learning_rate": 8.999427309171235e-07, "loss": 0.0193, "step": 2200 }, { "epoch": 0.03616133518776078, "grad_norm": 1.1292650699615479, "learning_rate": 9.040333796940195e-07, "loss": 0.0249, "step": 2210 }, { "epoch": 0.03632496113883662, "grad_norm": 1.7488288879394531, "learning_rate": 9.081240284709155e-07, "loss": 0.0205, "step": 2220 }, { "epoch": 0.03648858708991246, "grad_norm": 1.0636547803878784, "learning_rate": 9.122146772478116e-07, "loss": 0.0285, "step": 2230 }, { "epoch": 0.0366522130409883, "grad_norm": 0.9664870500564575, "learning_rate": 9.163053260247076e-07, "loss": 0.0171, "step": 2240 }, { "epoch": 0.03681583899206414, "grad_norm": 0.7720797657966614, "learning_rate": 9.203959748016036e-07, "loss": 0.0217, "step": 2250 }, { "epoch": 0.03697946494313998, "grad_norm": 0.7935715913772583, "learning_rate": 9.244866235784996e-07, "loss": 0.0178, "step": 2260 }, { "epoch": 0.03714309089421582, "grad_norm": 0.9548618793487549, "learning_rate": 9.285772723553957e-07, "loss": 0.0187, "step": 2270 }, { "epoch": 0.03730671684529167, "grad_norm": 0.6137745380401611, "learning_rate": 9.326679211322917e-07, "loss": 0.0184, "step": 2280 }, { "epoch": 0.0374703427963675, "grad_norm": 0.7197610139846802, "learning_rate": 9.367585699091877e-07, "loss": 0.0171, "step": 2290 }, { "epoch": 0.03763396874744335, "grad_norm": 1.2089803218841553, "learning_rate": 9.408492186860837e-07, "loss": 0.0241, "step": 2300 }, { "epoch": 0.037797594698519184, "grad_norm": 0.6038383841514587, "learning_rate": 9.449398674629797e-07, "loss": 0.0205, "step": 2310 }, { "epoch": 0.03796122064959503, "grad_norm": 0.7549110651016235, "learning_rate": 9.490305162398757e-07, "loss": 0.0186, "step": 2320 }, { "epoch": 0.038124846600670864, "grad_norm": 1.0074551105499268, "learning_rate": 9.531211650167717e-07, "loss": 0.0243, "step": 2330 }, { "epoch": 0.03828847255174671, "grad_norm": 1.0781223773956299, "learning_rate": 9.572118137936676e-07, "loss": 0.019, "step": 2340 }, { "epoch": 0.038452098502822545, "grad_norm": 0.7320874333381653, "learning_rate": 9.613024625705638e-07, "loss": 0.0196, "step": 2350 }, { "epoch": 0.03861572445389839, "grad_norm": 0.7214975357055664, "learning_rate": 9.653931113474597e-07, "loss": 0.0181, "step": 2360 }, { "epoch": 0.03877935040497423, "grad_norm": 0.5130652785301208, "learning_rate": 9.694837601243558e-07, "loss": 0.017, "step": 2370 }, { "epoch": 0.03894297635605007, "grad_norm": 0.7154262065887451, "learning_rate": 9.735744089012517e-07, "loss": 0.0176, "step": 2380 }, { "epoch": 0.03910660230712591, "grad_norm": 1.1331579685211182, "learning_rate": 9.776650576781479e-07, "loss": 0.0165, "step": 2390 }, { "epoch": 0.03927022825820175, "grad_norm": 0.7700015306472778, "learning_rate": 9.817557064550438e-07, "loss": 0.0142, "step": 2400 }, { "epoch": 0.03943385420927759, "grad_norm": 0.9046688675880432, "learning_rate": 9.8584635523194e-07, "loss": 0.0153, "step": 2410 }, { "epoch": 0.03959748016035343, "grad_norm": 0.6018112897872925, "learning_rate": 9.899370040088358e-07, "loss": 0.0157, "step": 2420 }, { "epoch": 0.03976110611142927, "grad_norm": 0.9699082970619202, "learning_rate": 9.94027652785732e-07, "loss": 0.0196, "step": 2430 }, { "epoch": 0.03992473206250512, "grad_norm": 0.8703541159629822, "learning_rate": 9.98118301562628e-07, "loss": 0.0213, "step": 2440 }, { "epoch": 0.04008835801358095, "grad_norm": 0.6621701717376709, "learning_rate": 1.002208950339524e-06, "loss": 0.0218, "step": 2450 }, { "epoch": 0.0402519839646568, "grad_norm": 0.8218420147895813, "learning_rate": 1.00629959911642e-06, "loss": 0.0208, "step": 2460 }, { "epoch": 0.040415609915732634, "grad_norm": 0.9209209680557251, "learning_rate": 1.010390247893316e-06, "loss": 0.0155, "step": 2470 }, { "epoch": 0.04057923586680848, "grad_norm": 0.6691835522651672, "learning_rate": 1.014480896670212e-06, "loss": 0.0154, "step": 2480 }, { "epoch": 0.040742861817884314, "grad_norm": 0.9384796619415283, "learning_rate": 1.0185715454471081e-06, "loss": 0.0206, "step": 2490 }, { "epoch": 0.04090648776896016, "grad_norm": 0.8038955926895142, "learning_rate": 1.022662194224004e-06, "loss": 0.0155, "step": 2500 }, { "epoch": 0.041070113720035994, "grad_norm": 0.7362964749336243, "learning_rate": 1.0267528430009002e-06, "loss": 0.0206, "step": 2510 }, { "epoch": 0.04123373967111184, "grad_norm": 0.9192692637443542, "learning_rate": 1.030843491777796e-06, "loss": 0.0182, "step": 2520 }, { "epoch": 0.04139736562218768, "grad_norm": 0.7515204548835754, "learning_rate": 1.034934140554692e-06, "loss": 0.018, "step": 2530 }, { "epoch": 0.04156099157326352, "grad_norm": 1.057660460472107, "learning_rate": 1.039024789331588e-06, "loss": 0.0192, "step": 2540 }, { "epoch": 0.04172461752433936, "grad_norm": 0.7640735507011414, "learning_rate": 1.043115438108484e-06, "loss": 0.0179, "step": 2550 }, { "epoch": 0.0418882434754152, "grad_norm": 0.8635754585266113, "learning_rate": 1.04720608688538e-06, "loss": 0.0182, "step": 2560 }, { "epoch": 0.04205186942649104, "grad_norm": 0.8443044424057007, "learning_rate": 1.0512967356622761e-06, "loss": 0.0124, "step": 2570 }, { "epoch": 0.04221549537756688, "grad_norm": 0.9158975481987, "learning_rate": 1.055387384439172e-06, "loss": 0.0124, "step": 2580 }, { "epoch": 0.04237912132864272, "grad_norm": 0.5584027171134949, "learning_rate": 1.0594780332160682e-06, "loss": 0.0178, "step": 2590 }, { "epoch": 0.04254274727971857, "grad_norm": 0.27069151401519775, "learning_rate": 1.063568681992964e-06, "loss": 0.0133, "step": 2600 }, { "epoch": 0.0427063732307944, "grad_norm": 0.5137176513671875, "learning_rate": 1.0676593307698602e-06, "loss": 0.016, "step": 2610 }, { "epoch": 0.04286999918187025, "grad_norm": 1.3019955158233643, "learning_rate": 1.0717499795467561e-06, "loss": 0.0134, "step": 2620 }, { "epoch": 0.043033625132946084, "grad_norm": 0.7433570623397827, "learning_rate": 1.0758406283236523e-06, "loss": 0.0148, "step": 2630 }, { "epoch": 0.04319725108402193, "grad_norm": 1.3966814279556274, "learning_rate": 1.0799312771005482e-06, "loss": 0.0195, "step": 2640 }, { "epoch": 0.043360877035097764, "grad_norm": 0.7880895733833313, "learning_rate": 1.0840219258774443e-06, "loss": 0.0152, "step": 2650 }, { "epoch": 0.04352450298617361, "grad_norm": 1.008904218673706, "learning_rate": 1.0881125746543402e-06, "loss": 0.0197, "step": 2660 }, { "epoch": 0.043688128937249444, "grad_norm": 1.024035096168518, "learning_rate": 1.0922032234312364e-06, "loss": 0.0137, "step": 2670 }, { "epoch": 0.04385175488832529, "grad_norm": 0.8459672927856445, "learning_rate": 1.0962938722081323e-06, "loss": 0.0141, "step": 2680 }, { "epoch": 0.04401538083940113, "grad_norm": 0.6067708134651184, "learning_rate": 1.1003845209850284e-06, "loss": 0.0169, "step": 2690 }, { "epoch": 0.04417900679047697, "grad_norm": 0.7578173875808716, "learning_rate": 1.1044751697619243e-06, "loss": 0.0173, "step": 2700 }, { "epoch": 0.04434263274155281, "grad_norm": 0.8268752098083496, "learning_rate": 1.1085658185388205e-06, "loss": 0.0179, "step": 2710 }, { "epoch": 0.04450625869262865, "grad_norm": 0.6216982007026672, "learning_rate": 1.1126564673157164e-06, "loss": 0.0154, "step": 2720 }, { "epoch": 0.04466988464370449, "grad_norm": 0.5965304970741272, "learning_rate": 1.1167471160926125e-06, "loss": 0.0163, "step": 2730 }, { "epoch": 0.04483351059478033, "grad_norm": 0.661740779876709, "learning_rate": 1.1208377648695084e-06, "loss": 0.0143, "step": 2740 }, { "epoch": 0.04499713654585617, "grad_norm": 1.0739951133728027, "learning_rate": 1.1249284136464044e-06, "loss": 0.0204, "step": 2750 }, { "epoch": 0.045160762496932016, "grad_norm": 0.5748319029808044, "learning_rate": 1.1290190624233003e-06, "loss": 0.0137, "step": 2760 }, { "epoch": 0.04532438844800785, "grad_norm": 0.725404679775238, "learning_rate": 1.1331097112001964e-06, "loss": 0.0142, "step": 2770 }, { "epoch": 0.0454880143990837, "grad_norm": 1.0682239532470703, "learning_rate": 1.1372003599770923e-06, "loss": 0.0237, "step": 2780 }, { "epoch": 0.045651640350159534, "grad_norm": 1.1173665523529053, "learning_rate": 1.1412910087539885e-06, "loss": 0.0159, "step": 2790 }, { "epoch": 0.04581526630123538, "grad_norm": 0.5693539977073669, "learning_rate": 1.1453816575308844e-06, "loss": 0.0166, "step": 2800 }, { "epoch": 0.045978892252311214, "grad_norm": 0.5482837557792664, "learning_rate": 1.1494723063077805e-06, "loss": 0.0185, "step": 2810 }, { "epoch": 0.04614251820338706, "grad_norm": 0.3773319721221924, "learning_rate": 1.1535629550846764e-06, "loss": 0.0129, "step": 2820 }, { "epoch": 0.0463061441544629, "grad_norm": 1.0107671022415161, "learning_rate": 1.1576536038615726e-06, "loss": 0.0154, "step": 2830 }, { "epoch": 0.04646977010553874, "grad_norm": 0.758152425289154, "learning_rate": 1.1617442526384685e-06, "loss": 0.0162, "step": 2840 }, { "epoch": 0.04663339605661458, "grad_norm": 0.6462787985801697, "learning_rate": 1.1658349014153646e-06, "loss": 0.0186, "step": 2850 }, { "epoch": 0.04679702200769042, "grad_norm": 0.5293020009994507, "learning_rate": 1.1699255501922605e-06, "loss": 0.0158, "step": 2860 }, { "epoch": 0.04696064795876626, "grad_norm": 0.5496254563331604, "learning_rate": 1.1740161989691567e-06, "loss": 0.0159, "step": 2870 }, { "epoch": 0.0471242739098421, "grad_norm": 1.1118664741516113, "learning_rate": 1.1781068477460526e-06, "loss": 0.0211, "step": 2880 }, { "epoch": 0.04728789986091794, "grad_norm": 0.8163902759552002, "learning_rate": 1.1821974965229487e-06, "loss": 0.0148, "step": 2890 }, { "epoch": 0.04745152581199378, "grad_norm": 0.38576290011405945, "learning_rate": 1.1862881452998446e-06, "loss": 0.0176, "step": 2900 }, { "epoch": 0.04761515176306962, "grad_norm": 1.2882270812988281, "learning_rate": 1.1903787940767408e-06, "loss": 0.0181, "step": 2910 }, { "epoch": 0.047778777714145466, "grad_norm": 0.9855552315711975, "learning_rate": 1.1944694428536367e-06, "loss": 0.0195, "step": 2920 }, { "epoch": 0.0479424036652213, "grad_norm": 0.48401790857315063, "learning_rate": 1.1985600916305328e-06, "loss": 0.0138, "step": 2930 }, { "epoch": 0.04810602961629715, "grad_norm": 1.1864231824874878, "learning_rate": 1.2026507404074287e-06, "loss": 0.0189, "step": 2940 }, { "epoch": 0.048269655567372984, "grad_norm": 0.49471181631088257, "learning_rate": 1.2067413891843249e-06, "loss": 0.017, "step": 2950 }, { "epoch": 0.04843328151844883, "grad_norm": 0.6354305148124695, "learning_rate": 1.2108320379612208e-06, "loss": 0.0147, "step": 2960 }, { "epoch": 0.048596907469524664, "grad_norm": 0.6612855195999146, "learning_rate": 1.2149226867381167e-06, "loss": 0.0147, "step": 2970 }, { "epoch": 0.04876053342060051, "grad_norm": 0.7974123358726501, "learning_rate": 1.2190133355150126e-06, "loss": 0.0162, "step": 2980 }, { "epoch": 0.04892415937167635, "grad_norm": 0.7535789608955383, "learning_rate": 1.2231039842919088e-06, "loss": 0.0152, "step": 2990 }, { "epoch": 0.04908778532275219, "grad_norm": 0.7143415212631226, "learning_rate": 1.2271946330688047e-06, "loss": 0.0151, "step": 3000 }, { "epoch": 0.04925141127382803, "grad_norm": 0.7587301135063171, "learning_rate": 1.2312852818457008e-06, "loss": 0.0173, "step": 3010 }, { "epoch": 0.04941503722490387, "grad_norm": 0.7240614891052246, "learning_rate": 1.2353759306225967e-06, "loss": 0.0139, "step": 3020 }, { "epoch": 0.04957866317597971, "grad_norm": 1.402193307876587, "learning_rate": 1.2394665793994929e-06, "loss": 0.0115, "step": 3030 }, { "epoch": 0.04974228912705555, "grad_norm": 0.5034580230712891, "learning_rate": 1.2435572281763888e-06, "loss": 0.0187, "step": 3040 }, { "epoch": 0.04990591507813139, "grad_norm": 0.8597303032875061, "learning_rate": 1.247647876953285e-06, "loss": 0.0158, "step": 3050 }, { "epoch": 0.05006954102920723, "grad_norm": 0.7847544550895691, "learning_rate": 1.2517385257301808e-06, "loss": 0.0191, "step": 3060 }, { "epoch": 0.05023316698028307, "grad_norm": 0.8109354376792908, "learning_rate": 1.255829174507077e-06, "loss": 0.0123, "step": 3070 }, { "epoch": 0.050396792931358916, "grad_norm": 0.6916020512580872, "learning_rate": 1.2599198232839729e-06, "loss": 0.0189, "step": 3080 }, { "epoch": 0.05056041888243475, "grad_norm": 0.8129985928535461, "learning_rate": 1.264010472060869e-06, "loss": 0.0158, "step": 3090 }, { "epoch": 0.0507240448335106, "grad_norm": 0.7774389982223511, "learning_rate": 1.268101120837765e-06, "loss": 0.0147, "step": 3100 }, { "epoch": 0.05088767078458643, "grad_norm": 1.3621916770935059, "learning_rate": 1.272191769614661e-06, "loss": 0.0141, "step": 3110 }, { "epoch": 0.05105129673566228, "grad_norm": 1.349725604057312, "learning_rate": 1.276282418391557e-06, "loss": 0.0178, "step": 3120 }, { "epoch": 0.051214922686738114, "grad_norm": 0.6559675931930542, "learning_rate": 1.2803730671684531e-06, "loss": 0.0144, "step": 3130 }, { "epoch": 0.05137854863781396, "grad_norm": 0.4959447383880615, "learning_rate": 1.284463715945349e-06, "loss": 0.0127, "step": 3140 }, { "epoch": 0.0515421745888898, "grad_norm": 0.712787926197052, "learning_rate": 1.2885543647222452e-06, "loss": 0.0149, "step": 3150 }, { "epoch": 0.05170580053996564, "grad_norm": 0.5305976867675781, "learning_rate": 1.292645013499141e-06, "loss": 0.0162, "step": 3160 }, { "epoch": 0.05186942649104148, "grad_norm": 0.6919953227043152, "learning_rate": 1.2967356622760372e-06, "loss": 0.0112, "step": 3170 }, { "epoch": 0.05203305244211732, "grad_norm": 0.7500140070915222, "learning_rate": 1.3008263110529331e-06, "loss": 0.0181, "step": 3180 }, { "epoch": 0.05219667839319316, "grad_norm": 0.5839250683784485, "learning_rate": 1.3049169598298293e-06, "loss": 0.0147, "step": 3190 }, { "epoch": 0.052360304344269, "grad_norm": 0.7700366973876953, "learning_rate": 1.3090076086067252e-06, "loss": 0.0179, "step": 3200 }, { "epoch": 0.05252393029534484, "grad_norm": 0.7092366218566895, "learning_rate": 1.3130982573836213e-06, "loss": 0.0157, "step": 3210 }, { "epoch": 0.05268755624642068, "grad_norm": 0.6009559035301208, "learning_rate": 1.3171889061605172e-06, "loss": 0.017, "step": 3220 }, { "epoch": 0.05285118219749652, "grad_norm": 0.8292373418807983, "learning_rate": 1.3212795549374134e-06, "loss": 0.0145, "step": 3230 }, { "epoch": 0.053014808148572366, "grad_norm": 0.7454757690429688, "learning_rate": 1.3253702037143093e-06, "loss": 0.0164, "step": 3240 }, { "epoch": 0.0531784340996482, "grad_norm": 0.8205867409706116, "learning_rate": 1.3294608524912052e-06, "loss": 0.0134, "step": 3250 }, { "epoch": 0.05334206005072405, "grad_norm": 0.7466033697128296, "learning_rate": 1.3335515012681014e-06, "loss": 0.0153, "step": 3260 }, { "epoch": 0.05350568600179988, "grad_norm": 0.8189770579338074, "learning_rate": 1.3376421500449973e-06, "loss": 0.0161, "step": 3270 }, { "epoch": 0.05366931195287573, "grad_norm": 0.26645174622535706, "learning_rate": 1.3417327988218934e-06, "loss": 0.0141, "step": 3280 }, { "epoch": 0.053832937903951564, "grad_norm": 0.5707406997680664, "learning_rate": 1.3458234475987893e-06, "loss": 0.0187, "step": 3290 }, { "epoch": 0.05399656385502741, "grad_norm": 0.4595284163951874, "learning_rate": 1.3499140963756852e-06, "loss": 0.0145, "step": 3300 }, { "epoch": 0.05416018980610325, "grad_norm": 0.4943508505821228, "learning_rate": 1.3540047451525812e-06, "loss": 0.0123, "step": 3310 }, { "epoch": 0.05432381575717909, "grad_norm": 0.6578164100646973, "learning_rate": 1.3580953939294773e-06, "loss": 0.0115, "step": 3320 }, { "epoch": 0.05448744170825493, "grad_norm": 0.3397503197193146, "learning_rate": 1.3621860427063732e-06, "loss": 0.0128, "step": 3330 }, { "epoch": 0.05465106765933077, "grad_norm": 0.5769123435020447, "learning_rate": 1.3662766914832693e-06, "loss": 0.0113, "step": 3340 }, { "epoch": 0.05481469361040661, "grad_norm": 0.6306193470954895, "learning_rate": 1.3703673402601653e-06, "loss": 0.0124, "step": 3350 }, { "epoch": 0.05497831956148245, "grad_norm": 0.612381100654602, "learning_rate": 1.3744579890370614e-06, "loss": 0.0115, "step": 3360 }, { "epoch": 0.05514194551255829, "grad_norm": 1.0355913639068604, "learning_rate": 1.3785486378139573e-06, "loss": 0.0124, "step": 3370 }, { "epoch": 0.055305571463634136, "grad_norm": 0.8097784519195557, "learning_rate": 1.3826392865908532e-06, "loss": 0.0236, "step": 3380 }, { "epoch": 0.05546919741470997, "grad_norm": 0.39903342723846436, "learning_rate": 1.3867299353677494e-06, "loss": 0.0188, "step": 3390 }, { "epoch": 0.055632823365785816, "grad_norm": 0.6708441972732544, "learning_rate": 1.3908205841446453e-06, "loss": 0.011, "step": 3400 }, { "epoch": 0.05579644931686165, "grad_norm": 0.8678045868873596, "learning_rate": 1.3949112329215414e-06, "loss": 0.0153, "step": 3410 }, { "epoch": 0.0559600752679375, "grad_norm": 0.49642881751060486, "learning_rate": 1.3990018816984373e-06, "loss": 0.0159, "step": 3420 }, { "epoch": 0.05612370121901333, "grad_norm": 0.3772575557231903, "learning_rate": 1.4030925304753335e-06, "loss": 0.0127, "step": 3430 }, { "epoch": 0.05628732717008918, "grad_norm": 0.8412145972251892, "learning_rate": 1.4071831792522294e-06, "loss": 0.015, "step": 3440 }, { "epoch": 0.056450953121165014, "grad_norm": 0.8376142382621765, "learning_rate": 1.4112738280291255e-06, "loss": 0.0164, "step": 3450 }, { "epoch": 0.05661457907224086, "grad_norm": 0.34573403000831604, "learning_rate": 1.4153644768060214e-06, "loss": 0.0127, "step": 3460 }, { "epoch": 0.0567782050233167, "grad_norm": 0.5954613089561462, "learning_rate": 1.4194551255829176e-06, "loss": 0.0162, "step": 3470 }, { "epoch": 0.05694183097439254, "grad_norm": 0.9560826420783997, "learning_rate": 1.4235457743598135e-06, "loss": 0.0128, "step": 3480 }, { "epoch": 0.05710545692546838, "grad_norm": 0.507256805896759, "learning_rate": 1.4276364231367096e-06, "loss": 0.0121, "step": 3490 }, { "epoch": 0.05726908287654422, "grad_norm": 0.7701252102851868, "learning_rate": 1.4317270719136055e-06, "loss": 0.0109, "step": 3500 }, { "epoch": 0.05743270882762006, "grad_norm": 0.4697326421737671, "learning_rate": 1.4358177206905017e-06, "loss": 0.0149, "step": 3510 }, { "epoch": 0.0575963347786959, "grad_norm": 0.5739027857780457, "learning_rate": 1.4399083694673976e-06, "loss": 0.0139, "step": 3520 }, { "epoch": 0.05775996072977174, "grad_norm": 0.9096561670303345, "learning_rate": 1.4439990182442937e-06, "loss": 0.0154, "step": 3530 }, { "epoch": 0.057923586680847586, "grad_norm": 0.40334752202033997, "learning_rate": 1.4480896670211896e-06, "loss": 0.0172, "step": 3540 }, { "epoch": 0.05808721263192342, "grad_norm": 0.6278005838394165, "learning_rate": 1.4521803157980858e-06, "loss": 0.0145, "step": 3550 }, { "epoch": 0.058250838582999266, "grad_norm": 0.8693646192550659, "learning_rate": 1.4562709645749817e-06, "loss": 0.0126, "step": 3560 }, { "epoch": 0.0584144645340751, "grad_norm": 0.8312264680862427, "learning_rate": 1.4603616133518778e-06, "loss": 0.0149, "step": 3570 }, { "epoch": 0.058578090485150947, "grad_norm": 0.8749430775642395, "learning_rate": 1.4644522621287737e-06, "loss": 0.0149, "step": 3580 }, { "epoch": 0.05874171643622678, "grad_norm": 0.9583225250244141, "learning_rate": 1.4685429109056699e-06, "loss": 0.0132, "step": 3590 }, { "epoch": 0.05890534238730263, "grad_norm": 0.7419372797012329, "learning_rate": 1.4726335596825658e-06, "loss": 0.0134, "step": 3600 }, { "epoch": 0.059068968338378464, "grad_norm": 1.7683305740356445, "learning_rate": 1.476724208459462e-06, "loss": 0.015, "step": 3610 }, { "epoch": 0.05923259428945431, "grad_norm": 0.46562284231185913, "learning_rate": 1.4808148572363578e-06, "loss": 0.0132, "step": 3620 }, { "epoch": 0.05939622024053015, "grad_norm": 0.862454891204834, "learning_rate": 1.484905506013254e-06, "loss": 0.0174, "step": 3630 }, { "epoch": 0.05955984619160599, "grad_norm": 0.43820881843566895, "learning_rate": 1.48899615479015e-06, "loss": 0.0157, "step": 3640 }, { "epoch": 0.05972347214268183, "grad_norm": 0.8450514078140259, "learning_rate": 1.493086803567046e-06, "loss": 0.0129, "step": 3650 }, { "epoch": 0.05988709809375767, "grad_norm": 0.8324694633483887, "learning_rate": 1.497177452343942e-06, "loss": 0.0178, "step": 3660 }, { "epoch": 0.06005072404483351, "grad_norm": 0.6049464344978333, "learning_rate": 1.501268101120838e-06, "loss": 0.015, "step": 3670 }, { "epoch": 0.06021434999590935, "grad_norm": 0.5698825716972351, "learning_rate": 1.505358749897734e-06, "loss": 0.0116, "step": 3680 }, { "epoch": 0.06037797594698519, "grad_norm": 0.8258331418037415, "learning_rate": 1.5094493986746301e-06, "loss": 0.0111, "step": 3690 }, { "epoch": 0.060541601898061036, "grad_norm": 0.6101037859916687, "learning_rate": 1.513540047451526e-06, "loss": 0.0175, "step": 3700 }, { "epoch": 0.06070522784913687, "grad_norm": 0.6410166621208191, "learning_rate": 1.517630696228422e-06, "loss": 0.0094, "step": 3710 }, { "epoch": 0.060868853800212716, "grad_norm": 0.5575524568557739, "learning_rate": 1.521721345005318e-06, "loss": 0.017, "step": 3720 }, { "epoch": 0.06103247975128855, "grad_norm": 0.6818864941596985, "learning_rate": 1.525811993782214e-06, "loss": 0.015, "step": 3730 }, { "epoch": 0.061196105702364396, "grad_norm": 0.7768665552139282, "learning_rate": 1.52990264255911e-06, "loss": 0.0137, "step": 3740 }, { "epoch": 0.06135973165344023, "grad_norm": 0.9026640057563782, "learning_rate": 1.5339932913360059e-06, "loss": 0.0161, "step": 3750 }, { "epoch": 0.06152335760451608, "grad_norm": 0.6262550354003906, "learning_rate": 1.538083940112902e-06, "loss": 0.0155, "step": 3760 }, { "epoch": 0.061686983555591914, "grad_norm": 0.6269939541816711, "learning_rate": 1.542174588889798e-06, "loss": 0.0156, "step": 3770 }, { "epoch": 0.06185060950666776, "grad_norm": 0.6901090741157532, "learning_rate": 1.546265237666694e-06, "loss": 0.0133, "step": 3780 }, { "epoch": 0.0620142354577436, "grad_norm": 0.549107551574707, "learning_rate": 1.55035588644359e-06, "loss": 0.0132, "step": 3790 }, { "epoch": 0.06217786140881944, "grad_norm": 0.5338832139968872, "learning_rate": 1.554446535220486e-06, "loss": 0.013, "step": 3800 }, { "epoch": 0.06234148735989528, "grad_norm": 0.9736700057983398, "learning_rate": 1.558537183997382e-06, "loss": 0.0118, "step": 3810 }, { "epoch": 0.06250511331097112, "grad_norm": 0.5628230571746826, "learning_rate": 1.5626278327742781e-06, "loss": 0.0114, "step": 3820 }, { "epoch": 0.06266873926204695, "grad_norm": 0.5485469102859497, "learning_rate": 1.566718481551174e-06, "loss": 0.0203, "step": 3830 }, { "epoch": 0.0628323652131228, "grad_norm": 0.6979078650474548, "learning_rate": 1.57080913032807e-06, "loss": 0.013, "step": 3840 }, { "epoch": 0.06299599116419864, "grad_norm": 0.3915221393108368, "learning_rate": 1.5748997791049661e-06, "loss": 0.0102, "step": 3850 }, { "epoch": 0.06315961711527449, "grad_norm": 0.6288328170776367, "learning_rate": 1.578990427881862e-06, "loss": 0.017, "step": 3860 }, { "epoch": 0.06332324306635033, "grad_norm": 0.37258294224739075, "learning_rate": 1.5830810766587582e-06, "loss": 0.012, "step": 3870 }, { "epoch": 0.06348686901742616, "grad_norm": 0.9199112057685852, "learning_rate": 1.587171725435654e-06, "loss": 0.0097, "step": 3880 }, { "epoch": 0.063650494968502, "grad_norm": 0.5273042321205139, "learning_rate": 1.5912623742125502e-06, "loss": 0.0149, "step": 3890 }, { "epoch": 0.06381412091957785, "grad_norm": 0.7047915458679199, "learning_rate": 1.5953530229894461e-06, "loss": 0.0086, "step": 3900 }, { "epoch": 0.06397774687065369, "grad_norm": 0.5367410182952881, "learning_rate": 1.5994436717663423e-06, "loss": 0.0114, "step": 3910 }, { "epoch": 0.06414137282172952, "grad_norm": 0.784299373626709, "learning_rate": 1.6035343205432382e-06, "loss": 0.0125, "step": 3920 }, { "epoch": 0.06430499877280536, "grad_norm": 0.7520073652267456, "learning_rate": 1.6076249693201343e-06, "loss": 0.0147, "step": 3930 }, { "epoch": 0.06446862472388121, "grad_norm": 0.6631868481636047, "learning_rate": 1.6117156180970302e-06, "loss": 0.0086, "step": 3940 }, { "epoch": 0.06463225067495705, "grad_norm": 0.5261515378952026, "learning_rate": 1.6158062668739264e-06, "loss": 0.012, "step": 3950 }, { "epoch": 0.0647958766260329, "grad_norm": 0.34229394793510437, "learning_rate": 1.6198969156508223e-06, "loss": 0.0116, "step": 3960 }, { "epoch": 0.06495950257710872, "grad_norm": 0.8410035967826843, "learning_rate": 1.6239875644277184e-06, "loss": 0.0136, "step": 3970 }, { "epoch": 0.06512312852818457, "grad_norm": 0.5108805894851685, "learning_rate": 1.6280782132046143e-06, "loss": 0.0145, "step": 3980 }, { "epoch": 0.06528675447926041, "grad_norm": 0.3938635587692261, "learning_rate": 1.6321688619815105e-06, "loss": 0.0139, "step": 3990 }, { "epoch": 0.06545038043033626, "grad_norm": 0.5281301140785217, "learning_rate": 1.6362595107584064e-06, "loss": 0.016, "step": 4000 }, { "epoch": 0.0656140063814121, "grad_norm": 0.7819935083389282, "learning_rate": 1.6403501595353025e-06, "loss": 0.0123, "step": 4010 }, { "epoch": 0.06577763233248793, "grad_norm": 0.6288535594940186, "learning_rate": 1.6444408083121984e-06, "loss": 0.0116, "step": 4020 }, { "epoch": 0.06594125828356377, "grad_norm": 0.550206184387207, "learning_rate": 1.6485314570890946e-06, "loss": 0.0164, "step": 4030 }, { "epoch": 0.06610488423463962, "grad_norm": 0.5400760769844055, "learning_rate": 1.6526221058659905e-06, "loss": 0.0105, "step": 4040 }, { "epoch": 0.06626851018571546, "grad_norm": 0.4235306978225708, "learning_rate": 1.6567127546428866e-06, "loss": 0.0152, "step": 4050 }, { "epoch": 0.06643213613679129, "grad_norm": 0.34356123208999634, "learning_rate": 1.6608034034197825e-06, "loss": 0.0109, "step": 4060 }, { "epoch": 0.06659576208786713, "grad_norm": 0.34012627601623535, "learning_rate": 1.6648940521966787e-06, "loss": 0.0102, "step": 4070 }, { "epoch": 0.06675938803894298, "grad_norm": 0.5730460286140442, "learning_rate": 1.6689847009735746e-06, "loss": 0.0108, "step": 4080 }, { "epoch": 0.06692301399001882, "grad_norm": 0.8617515563964844, "learning_rate": 1.6730753497504707e-06, "loss": 0.0139, "step": 4090 }, { "epoch": 0.06708663994109466, "grad_norm": 0.6126902103424072, "learning_rate": 1.6771659985273666e-06, "loss": 0.0093, "step": 4100 }, { "epoch": 0.0672502658921705, "grad_norm": 0.8658061623573303, "learning_rate": 1.6812566473042628e-06, "loss": 0.0127, "step": 4110 }, { "epoch": 0.06741389184324634, "grad_norm": 0.6324352025985718, "learning_rate": 1.6853472960811587e-06, "loss": 0.0179, "step": 4120 }, { "epoch": 0.06757751779432218, "grad_norm": 1.390407681465149, "learning_rate": 1.6894379448580548e-06, "loss": 0.0116, "step": 4130 }, { "epoch": 0.06774114374539802, "grad_norm": 0.6377262473106384, "learning_rate": 1.6935285936349507e-06, "loss": 0.011, "step": 4140 }, { "epoch": 0.06790476969647385, "grad_norm": 0.6767842769622803, "learning_rate": 1.6976192424118469e-06, "loss": 0.013, "step": 4150 }, { "epoch": 0.0680683956475497, "grad_norm": 0.5550398826599121, "learning_rate": 1.7017098911887428e-06, "loss": 0.0102, "step": 4160 }, { "epoch": 0.06823202159862554, "grad_norm": 0.36330094933509827, "learning_rate": 1.7058005399656387e-06, "loss": 0.0068, "step": 4170 }, { "epoch": 0.06839564754970139, "grad_norm": 0.34109383821487427, "learning_rate": 1.7098911887425346e-06, "loss": 0.0123, "step": 4180 }, { "epoch": 0.06855927350077723, "grad_norm": 0.44411882758140564, "learning_rate": 1.7139818375194306e-06, "loss": 0.0084, "step": 4190 }, { "epoch": 0.06872289945185306, "grad_norm": 0.6136140823364258, "learning_rate": 1.7180724862963267e-06, "loss": 0.0154, "step": 4200 }, { "epoch": 0.0688865254029289, "grad_norm": 0.5123714208602905, "learning_rate": 1.7221631350732226e-06, "loss": 0.0125, "step": 4210 }, { "epoch": 0.06905015135400475, "grad_norm": 0.7241730093955994, "learning_rate": 1.7262537838501187e-06, "loss": 0.0102, "step": 4220 }, { "epoch": 0.06921377730508059, "grad_norm": 0.8738928437232971, "learning_rate": 1.7303444326270147e-06, "loss": 0.01, "step": 4230 }, { "epoch": 0.06937740325615642, "grad_norm": 0.6852147579193115, "learning_rate": 1.7344350814039108e-06, "loss": 0.0124, "step": 4240 }, { "epoch": 0.06954102920723226, "grad_norm": 0.5979969501495361, "learning_rate": 1.7385257301808067e-06, "loss": 0.027, "step": 4250 }, { "epoch": 0.06970465515830811, "grad_norm": 0.3118220567703247, "learning_rate": 1.7426163789577028e-06, "loss": 0.0151, "step": 4260 }, { "epoch": 0.06986828110938395, "grad_norm": 0.5463794469833374, "learning_rate": 1.7467070277345988e-06, "loss": 0.0131, "step": 4270 }, { "epoch": 0.0700319070604598, "grad_norm": 0.5436028838157654, "learning_rate": 1.750797676511495e-06, "loss": 0.0093, "step": 4280 }, { "epoch": 0.07019553301153562, "grad_norm": 0.8300839066505432, "learning_rate": 1.7548883252883908e-06, "loss": 0.0138, "step": 4290 }, { "epoch": 0.07035915896261147, "grad_norm": 0.4434642791748047, "learning_rate": 1.7589789740652867e-06, "loss": 0.0099, "step": 4300 }, { "epoch": 0.07052278491368731, "grad_norm": 0.6233537197113037, "learning_rate": 1.7630696228421829e-06, "loss": 0.0089, "step": 4310 }, { "epoch": 0.07068641086476316, "grad_norm": 0.7118611335754395, "learning_rate": 1.7671602716190788e-06, "loss": 0.0095, "step": 4320 }, { "epoch": 0.070850036815839, "grad_norm": 0.7702072858810425, "learning_rate": 1.771250920395975e-06, "loss": 0.0172, "step": 4330 }, { "epoch": 0.07101366276691483, "grad_norm": 0.5972219705581665, "learning_rate": 1.7753415691728708e-06, "loss": 0.0124, "step": 4340 }, { "epoch": 0.07117728871799067, "grad_norm": 0.5806770324707031, "learning_rate": 1.779432217949767e-06, "loss": 0.0096, "step": 4350 }, { "epoch": 0.07134091466906652, "grad_norm": 0.4935813844203949, "learning_rate": 1.7835228667266629e-06, "loss": 0.011, "step": 4360 }, { "epoch": 0.07150454062014236, "grad_norm": 0.5056700706481934, "learning_rate": 1.787613515503559e-06, "loss": 0.017, "step": 4370 }, { "epoch": 0.07166816657121819, "grad_norm": 0.08893541991710663, "learning_rate": 1.791704164280455e-06, "loss": 0.0187, "step": 4380 }, { "epoch": 0.07183179252229403, "grad_norm": 0.431772381067276, "learning_rate": 1.795794813057351e-06, "loss": 0.0132, "step": 4390 }, { "epoch": 0.07199541847336988, "grad_norm": 0.6524070501327515, "learning_rate": 1.799885461834247e-06, "loss": 0.0123, "step": 4400 }, { "epoch": 0.07215904442444572, "grad_norm": 0.4881684184074402, "learning_rate": 1.8039761106111431e-06, "loss": 0.0077, "step": 4410 }, { "epoch": 0.07232267037552156, "grad_norm": 0.40165388584136963, "learning_rate": 1.808066759388039e-06, "loss": 0.0117, "step": 4420 }, { "epoch": 0.0724862963265974, "grad_norm": 0.508890688419342, "learning_rate": 1.8121574081649352e-06, "loss": 0.0141, "step": 4430 }, { "epoch": 0.07264992227767324, "grad_norm": 0.7069491147994995, "learning_rate": 1.816248056941831e-06, "loss": 0.012, "step": 4440 }, { "epoch": 0.07281354822874908, "grad_norm": 0.6757910251617432, "learning_rate": 1.8203387057187272e-06, "loss": 0.0155, "step": 4450 }, { "epoch": 0.07297717417982492, "grad_norm": 0.36266452074050903, "learning_rate": 1.8244293544956231e-06, "loss": 0.0113, "step": 4460 }, { "epoch": 0.07314080013090075, "grad_norm": 0.34122607111930847, "learning_rate": 1.8285200032725193e-06, "loss": 0.0103, "step": 4470 }, { "epoch": 0.0733044260819766, "grad_norm": 0.3820841908454895, "learning_rate": 1.8326106520494152e-06, "loss": 0.0098, "step": 4480 }, { "epoch": 0.07346805203305244, "grad_norm": 0.4844144880771637, "learning_rate": 1.8367013008263113e-06, "loss": 0.0089, "step": 4490 }, { "epoch": 0.07363167798412829, "grad_norm": 0.8034864664077759, "learning_rate": 1.8407919496032072e-06, "loss": 0.0107, "step": 4500 }, { "epoch": 0.07379530393520413, "grad_norm": 0.5484949350357056, "learning_rate": 1.8448825983801034e-06, "loss": 0.016, "step": 4510 }, { "epoch": 0.07395892988627996, "grad_norm": 0.3402491509914398, "learning_rate": 1.8489732471569993e-06, "loss": 0.0135, "step": 4520 }, { "epoch": 0.0741225558373558, "grad_norm": 0.6615390777587891, "learning_rate": 1.8530638959338954e-06, "loss": 0.0161, "step": 4530 }, { "epoch": 0.07428618178843165, "grad_norm": 0.43279364705085754, "learning_rate": 1.8571545447107913e-06, "loss": 0.0117, "step": 4540 }, { "epoch": 0.07444980773950749, "grad_norm": 0.40396544337272644, "learning_rate": 1.8612451934876875e-06, "loss": 0.0113, "step": 4550 }, { "epoch": 0.07461343369058333, "grad_norm": 0.436628520488739, "learning_rate": 1.8653358422645834e-06, "loss": 0.0136, "step": 4560 }, { "epoch": 0.07477705964165916, "grad_norm": 0.37469497323036194, "learning_rate": 1.8694264910414795e-06, "loss": 0.0097, "step": 4570 }, { "epoch": 0.074940685592735, "grad_norm": 0.47834062576293945, "learning_rate": 1.8735171398183754e-06, "loss": 0.0095, "step": 4580 }, { "epoch": 0.07510431154381085, "grad_norm": 0.5727904438972473, "learning_rate": 1.8776077885952716e-06, "loss": 0.0086, "step": 4590 }, { "epoch": 0.0752679374948867, "grad_norm": 0.5979715585708618, "learning_rate": 1.8816984373721675e-06, "loss": 0.0114, "step": 4600 }, { "epoch": 0.07543156344596252, "grad_norm": 0.6282607316970825, "learning_rate": 1.8857890861490636e-06, "loss": 0.0108, "step": 4610 }, { "epoch": 0.07559518939703837, "grad_norm": 0.7292720079421997, "learning_rate": 1.8898797349259593e-06, "loss": 0.0101, "step": 4620 }, { "epoch": 0.07575881534811421, "grad_norm": 0.5749276876449585, "learning_rate": 1.8939703837028553e-06, "loss": 0.0137, "step": 4630 }, { "epoch": 0.07592244129919005, "grad_norm": 0.6188210248947144, "learning_rate": 1.8980610324797514e-06, "loss": 0.0148, "step": 4640 }, { "epoch": 0.0760860672502659, "grad_norm": 0.5189754962921143, "learning_rate": 1.9021516812566473e-06, "loss": 0.0125, "step": 4650 }, { "epoch": 0.07624969320134173, "grad_norm": 0.6029978394508362, "learning_rate": 1.9062423300335434e-06, "loss": 0.0121, "step": 4660 }, { "epoch": 0.07641331915241757, "grad_norm": 0.38304024934768677, "learning_rate": 1.9103329788104396e-06, "loss": 0.0132, "step": 4670 }, { "epoch": 0.07657694510349342, "grad_norm": 0.5108842849731445, "learning_rate": 1.9144236275873353e-06, "loss": 0.0114, "step": 4680 }, { "epoch": 0.07674057105456926, "grad_norm": 0.6131144165992737, "learning_rate": 1.9185142763642314e-06, "loss": 0.0126, "step": 4690 }, { "epoch": 0.07690419700564509, "grad_norm": 0.9674975275993347, "learning_rate": 1.9226049251411275e-06, "loss": 0.0121, "step": 4700 }, { "epoch": 0.07706782295672093, "grad_norm": 0.6585379242897034, "learning_rate": 1.9266955739180237e-06, "loss": 0.0118, "step": 4710 }, { "epoch": 0.07723144890779678, "grad_norm": 0.4793098270893097, "learning_rate": 1.9307862226949194e-06, "loss": 0.0097, "step": 4720 }, { "epoch": 0.07739507485887262, "grad_norm": 0.5050153136253357, "learning_rate": 1.9348768714718155e-06, "loss": 0.0123, "step": 4730 }, { "epoch": 0.07755870080994846, "grad_norm": 0.46781304478645325, "learning_rate": 1.9389675202487116e-06, "loss": 0.0117, "step": 4740 }, { "epoch": 0.0777223267610243, "grad_norm": 0.8317105174064636, "learning_rate": 1.9430581690256078e-06, "loss": 0.0167, "step": 4750 }, { "epoch": 0.07788595271210014, "grad_norm": 0.6884296536445618, "learning_rate": 1.9471488178025035e-06, "loss": 0.0081, "step": 4760 }, { "epoch": 0.07804957866317598, "grad_norm": 0.46151047945022583, "learning_rate": 1.9512394665793996e-06, "loss": 0.0103, "step": 4770 }, { "epoch": 0.07821320461425182, "grad_norm": 0.44711849093437195, "learning_rate": 1.9553301153562957e-06, "loss": 0.0084, "step": 4780 }, { "epoch": 0.07837683056532765, "grad_norm": 0.5170176029205322, "learning_rate": 1.9594207641331915e-06, "loss": 0.0097, "step": 4790 }, { "epoch": 0.0785404565164035, "grad_norm": 0.5548386573791504, "learning_rate": 1.9635114129100876e-06, "loss": 0.01, "step": 4800 }, { "epoch": 0.07870408246747934, "grad_norm": 0.8234551548957825, "learning_rate": 1.9676020616869837e-06, "loss": 0.0103, "step": 4810 }, { "epoch": 0.07886770841855519, "grad_norm": 0.7186896800994873, "learning_rate": 1.97169271046388e-06, "loss": 0.0108, "step": 4820 }, { "epoch": 0.07903133436963103, "grad_norm": 0.9302345514297485, "learning_rate": 1.9757833592407756e-06, "loss": 0.0109, "step": 4830 }, { "epoch": 0.07919496032070686, "grad_norm": 0.6536460518836975, "learning_rate": 1.9798740080176717e-06, "loss": 0.0102, "step": 4840 }, { "epoch": 0.0793585862717827, "grad_norm": 0.4088923931121826, "learning_rate": 1.983964656794568e-06, "loss": 0.0095, "step": 4850 }, { "epoch": 0.07952221222285855, "grad_norm": 0.5232491493225098, "learning_rate": 1.988055305571464e-06, "loss": 0.0126, "step": 4860 }, { "epoch": 0.07968583817393439, "grad_norm": 0.32782286405563354, "learning_rate": 1.9921459543483597e-06, "loss": 0.0101, "step": 4870 }, { "epoch": 0.07984946412501023, "grad_norm": 0.35001909732818604, "learning_rate": 1.996236603125256e-06, "loss": 0.0097, "step": 4880 }, { "epoch": 0.08001309007608606, "grad_norm": 0.40097835659980774, "learning_rate": 2.000327251902152e-06, "loss": 0.0072, "step": 4890 }, { "epoch": 0.0801767160271619, "grad_norm": 0.47307050228118896, "learning_rate": 2.004417900679048e-06, "loss": 0.0112, "step": 4900 }, { "epoch": 0.08034034197823775, "grad_norm": 0.5360323786735535, "learning_rate": 2.0085085494559438e-06, "loss": 0.0086, "step": 4910 }, { "epoch": 0.0805039679293136, "grad_norm": 0.5396776795387268, "learning_rate": 2.01259919823284e-06, "loss": 0.0078, "step": 4920 }, { "epoch": 0.08066759388038942, "grad_norm": 0.5715587139129639, "learning_rate": 2.016689847009736e-06, "loss": 0.0116, "step": 4930 }, { "epoch": 0.08083121983146527, "grad_norm": 0.7345837354660034, "learning_rate": 2.020780495786632e-06, "loss": 0.0107, "step": 4940 }, { "epoch": 0.08099484578254111, "grad_norm": 0.5504596829414368, "learning_rate": 2.024871144563528e-06, "loss": 0.0096, "step": 4950 }, { "epoch": 0.08115847173361695, "grad_norm": 0.6429846286773682, "learning_rate": 2.028961793340424e-06, "loss": 0.013, "step": 4960 }, { "epoch": 0.0813220976846928, "grad_norm": 0.34800758957862854, "learning_rate": 2.03305244211732e-06, "loss": 0.0095, "step": 4970 }, { "epoch": 0.08148572363576863, "grad_norm": 0.8073736429214478, "learning_rate": 2.0371430908942163e-06, "loss": 0.0107, "step": 4980 }, { "epoch": 0.08164934958684447, "grad_norm": 0.5357869267463684, "learning_rate": 2.041233739671112e-06, "loss": 0.0096, "step": 4990 }, { "epoch": 0.08181297553792032, "grad_norm": 0.4288693368434906, "learning_rate": 2.045324388448008e-06, "loss": 0.0116, "step": 5000 }, { "epoch": 0.08197660148899616, "grad_norm": 0.45683252811431885, "learning_rate": 2.0494150372249042e-06, "loss": 0.0112, "step": 5010 }, { "epoch": 0.08214022744007199, "grad_norm": 0.5369279980659485, "learning_rate": 2.0535056860018004e-06, "loss": 0.0076, "step": 5020 }, { "epoch": 0.08230385339114783, "grad_norm": 0.8051870465278625, "learning_rate": 2.057596334778696e-06, "loss": 0.0105, "step": 5030 }, { "epoch": 0.08246747934222368, "grad_norm": 0.3576186001300812, "learning_rate": 2.061686983555592e-06, "loss": 0.0114, "step": 5040 }, { "epoch": 0.08263110529329952, "grad_norm": 0.5585474967956543, "learning_rate": 2.0657776323324883e-06, "loss": 0.0116, "step": 5050 }, { "epoch": 0.08279473124437536, "grad_norm": 0.5253618955612183, "learning_rate": 2.069868281109384e-06, "loss": 0.0101, "step": 5060 }, { "epoch": 0.0829583571954512, "grad_norm": 0.3295694887638092, "learning_rate": 2.07395892988628e-06, "loss": 0.0106, "step": 5070 }, { "epoch": 0.08312198314652704, "grad_norm": 0.5517259836196899, "learning_rate": 2.078049578663176e-06, "loss": 0.0092, "step": 5080 }, { "epoch": 0.08328560909760288, "grad_norm": 0.4592241942882538, "learning_rate": 2.082140227440072e-06, "loss": 0.0105, "step": 5090 }, { "epoch": 0.08344923504867872, "grad_norm": 0.25137263536453247, "learning_rate": 2.086230876216968e-06, "loss": 0.0083, "step": 5100 }, { "epoch": 0.08361286099975457, "grad_norm": 0.5853001475334167, "learning_rate": 2.0903215249938643e-06, "loss": 0.0096, "step": 5110 }, { "epoch": 0.0837764869508304, "grad_norm": 0.6381774544715881, "learning_rate": 2.09441217377076e-06, "loss": 0.0149, "step": 5120 }, { "epoch": 0.08394011290190624, "grad_norm": 0.837694525718689, "learning_rate": 2.098502822547656e-06, "loss": 0.0087, "step": 5130 }, { "epoch": 0.08410373885298209, "grad_norm": 0.456706702709198, "learning_rate": 2.1025934713245522e-06, "loss": 0.0132, "step": 5140 }, { "epoch": 0.08426736480405793, "grad_norm": 0.33764880895614624, "learning_rate": 2.1066841201014484e-06, "loss": 0.0079, "step": 5150 }, { "epoch": 0.08443099075513376, "grad_norm": 0.479375422000885, "learning_rate": 2.110774768878344e-06, "loss": 0.0149, "step": 5160 }, { "epoch": 0.0845946167062096, "grad_norm": 0.9286848306655884, "learning_rate": 2.1148654176552402e-06, "loss": 0.0142, "step": 5170 }, { "epoch": 0.08475824265728545, "grad_norm": 0.5799455046653748, "learning_rate": 2.1189560664321363e-06, "loss": 0.0093, "step": 5180 }, { "epoch": 0.08492186860836129, "grad_norm": 0.35933926701545715, "learning_rate": 2.1230467152090325e-06, "loss": 0.009, "step": 5190 }, { "epoch": 0.08508549455943713, "grad_norm": 0.8341226577758789, "learning_rate": 2.127137363985928e-06, "loss": 0.0145, "step": 5200 }, { "epoch": 0.08524912051051296, "grad_norm": 0.2809264063835144, "learning_rate": 2.1312280127628243e-06, "loss": 0.0114, "step": 5210 }, { "epoch": 0.0854127464615888, "grad_norm": 0.4433784782886505, "learning_rate": 2.1353186615397204e-06, "loss": 0.0138, "step": 5220 }, { "epoch": 0.08557637241266465, "grad_norm": 0.45161765813827515, "learning_rate": 2.139409310316616e-06, "loss": 0.0091, "step": 5230 }, { "epoch": 0.0857399983637405, "grad_norm": 0.6701156497001648, "learning_rate": 2.1434999590935123e-06, "loss": 0.0101, "step": 5240 }, { "epoch": 0.08590362431481632, "grad_norm": 0.2068326622247696, "learning_rate": 2.1475906078704084e-06, "loss": 0.0105, "step": 5250 }, { "epoch": 0.08606725026589217, "grad_norm": 0.33273452520370483, "learning_rate": 2.1516812566473045e-06, "loss": 0.01, "step": 5260 }, { "epoch": 0.08623087621696801, "grad_norm": 0.5258915424346924, "learning_rate": 2.1557719054242003e-06, "loss": 0.0068, "step": 5270 }, { "epoch": 0.08639450216804385, "grad_norm": 0.42602652311325073, "learning_rate": 2.1598625542010964e-06, "loss": 0.0086, "step": 5280 }, { "epoch": 0.0865581281191197, "grad_norm": 0.5108977556228638, "learning_rate": 2.1639532029779925e-06, "loss": 0.0096, "step": 5290 }, { "epoch": 0.08672175407019553, "grad_norm": 0.4123256802558899, "learning_rate": 2.1680438517548887e-06, "loss": 0.0106, "step": 5300 }, { "epoch": 0.08688538002127137, "grad_norm": 0.6202502250671387, "learning_rate": 2.1721345005317844e-06, "loss": 0.0075, "step": 5310 }, { "epoch": 0.08704900597234722, "grad_norm": 0.40042591094970703, "learning_rate": 2.1762251493086805e-06, "loss": 0.0123, "step": 5320 }, { "epoch": 0.08721263192342306, "grad_norm": 0.45790234208106995, "learning_rate": 2.1803157980855766e-06, "loss": 0.012, "step": 5330 }, { "epoch": 0.08737625787449889, "grad_norm": 0.5915714502334595, "learning_rate": 2.1844064468624728e-06, "loss": 0.0095, "step": 5340 }, { "epoch": 0.08753988382557473, "grad_norm": 0.5131071209907532, "learning_rate": 2.1884970956393685e-06, "loss": 0.0084, "step": 5350 }, { "epoch": 0.08770350977665058, "grad_norm": 0.9336416721343994, "learning_rate": 2.1925877444162646e-06, "loss": 0.0074, "step": 5360 }, { "epoch": 0.08786713572772642, "grad_norm": 0.29183948040008545, "learning_rate": 2.1966783931931607e-06, "loss": 0.0089, "step": 5370 }, { "epoch": 0.08803076167880226, "grad_norm": 0.4270169734954834, "learning_rate": 2.200769041970057e-06, "loss": 0.0142, "step": 5380 }, { "epoch": 0.0881943876298781, "grad_norm": 0.5773008465766907, "learning_rate": 2.2048596907469526e-06, "loss": 0.0113, "step": 5390 }, { "epoch": 0.08835801358095394, "grad_norm": 0.4308386445045471, "learning_rate": 2.2089503395238487e-06, "loss": 0.009, "step": 5400 }, { "epoch": 0.08852163953202978, "grad_norm": 0.37513744831085205, "learning_rate": 2.213040988300745e-06, "loss": 0.0083, "step": 5410 }, { "epoch": 0.08868526548310562, "grad_norm": 0.5529701113700867, "learning_rate": 2.217131637077641e-06, "loss": 0.0067, "step": 5420 }, { "epoch": 0.08884889143418147, "grad_norm": 0.3665136992931366, "learning_rate": 2.2212222858545367e-06, "loss": 0.0077, "step": 5430 }, { "epoch": 0.0890125173852573, "grad_norm": 0.4690069556236267, "learning_rate": 2.225312934631433e-06, "loss": 0.0087, "step": 5440 }, { "epoch": 0.08917614333633314, "grad_norm": 0.20975327491760254, "learning_rate": 2.229403583408329e-06, "loss": 0.0064, "step": 5450 }, { "epoch": 0.08933976928740898, "grad_norm": 0.4744136333465576, "learning_rate": 2.233494232185225e-06, "loss": 0.0106, "step": 5460 }, { "epoch": 0.08950339523848483, "grad_norm": 0.5571690797805786, "learning_rate": 2.2375848809621208e-06, "loss": 0.0064, "step": 5470 }, { "epoch": 0.08966702118956066, "grad_norm": 0.4103950262069702, "learning_rate": 2.241675529739017e-06, "loss": 0.012, "step": 5480 }, { "epoch": 0.0898306471406365, "grad_norm": 0.3762763738632202, "learning_rate": 2.245766178515913e-06, "loss": 0.0084, "step": 5490 }, { "epoch": 0.08999427309171235, "grad_norm": 0.5339162945747375, "learning_rate": 2.2498568272928087e-06, "loss": 0.0093, "step": 5500 }, { "epoch": 0.09015789904278819, "grad_norm": 0.32454484701156616, "learning_rate": 2.253947476069705e-06, "loss": 0.009, "step": 5510 }, { "epoch": 0.09032152499386403, "grad_norm": 0.4830230176448822, "learning_rate": 2.2580381248466006e-06, "loss": 0.0081, "step": 5520 }, { "epoch": 0.09048515094493986, "grad_norm": 0.3017253577709198, "learning_rate": 2.2621287736234967e-06, "loss": 0.0103, "step": 5530 }, { "epoch": 0.0906487768960157, "grad_norm": 0.4473549723625183, "learning_rate": 2.266219422400393e-06, "loss": 0.0092, "step": 5540 }, { "epoch": 0.09081240284709155, "grad_norm": 0.46569833159446716, "learning_rate": 2.270310071177289e-06, "loss": 0.0111, "step": 5550 }, { "epoch": 0.0909760287981674, "grad_norm": 0.5446906089782715, "learning_rate": 2.2744007199541847e-06, "loss": 0.0083, "step": 5560 }, { "epoch": 0.09113965474924322, "grad_norm": 0.4112224280834198, "learning_rate": 2.278491368731081e-06, "loss": 0.0117, "step": 5570 }, { "epoch": 0.09130328070031907, "grad_norm": 0.502993106842041, "learning_rate": 2.282582017507977e-06, "loss": 0.0103, "step": 5580 }, { "epoch": 0.09146690665139491, "grad_norm": 0.26549339294433594, "learning_rate": 2.286672666284873e-06, "loss": 0.0078, "step": 5590 }, { "epoch": 0.09163053260247075, "grad_norm": 0.7301869988441467, "learning_rate": 2.2907633150617688e-06, "loss": 0.0102, "step": 5600 }, { "epoch": 0.0917941585535466, "grad_norm": 0.5565994381904602, "learning_rate": 2.294853963838665e-06, "loss": 0.008, "step": 5610 }, { "epoch": 0.09195778450462243, "grad_norm": 0.577850341796875, "learning_rate": 2.298944612615561e-06, "loss": 0.0137, "step": 5620 }, { "epoch": 0.09212141045569827, "grad_norm": 0.5726605653762817, "learning_rate": 2.303035261392457e-06, "loss": 0.0102, "step": 5630 }, { "epoch": 0.09228503640677412, "grad_norm": 0.42456308007240295, "learning_rate": 2.307125910169353e-06, "loss": 0.0133, "step": 5640 }, { "epoch": 0.09244866235784996, "grad_norm": 0.4474571943283081, "learning_rate": 2.311216558946249e-06, "loss": 0.0087, "step": 5650 }, { "epoch": 0.0926122883089258, "grad_norm": 0.3489013612270355, "learning_rate": 2.315307207723145e-06, "loss": 0.0076, "step": 5660 }, { "epoch": 0.09277591426000163, "grad_norm": 0.4597572684288025, "learning_rate": 2.3193978565000413e-06, "loss": 0.0101, "step": 5670 }, { "epoch": 0.09293954021107748, "grad_norm": 0.39873847365379333, "learning_rate": 2.323488505276937e-06, "loss": 0.0104, "step": 5680 }, { "epoch": 0.09310316616215332, "grad_norm": 0.4917791187763214, "learning_rate": 2.327579154053833e-06, "loss": 0.0082, "step": 5690 }, { "epoch": 0.09326679211322916, "grad_norm": 0.296065092086792, "learning_rate": 2.3316698028307292e-06, "loss": 0.0106, "step": 5700 }, { "epoch": 0.093430418064305, "grad_norm": 0.3405877351760864, "learning_rate": 2.335760451607625e-06, "loss": 0.0173, "step": 5710 }, { "epoch": 0.09359404401538084, "grad_norm": 0.33221426606178284, "learning_rate": 2.339851100384521e-06, "loss": 0.0086, "step": 5720 }, { "epoch": 0.09375766996645668, "grad_norm": 0.3432377576828003, "learning_rate": 2.3439417491614172e-06, "loss": 0.0088, "step": 5730 }, { "epoch": 0.09392129591753252, "grad_norm": 1.0555815696716309, "learning_rate": 2.3480323979383133e-06, "loss": 0.01, "step": 5740 }, { "epoch": 0.09408492186860837, "grad_norm": 0.23515449464321136, "learning_rate": 2.352123046715209e-06, "loss": 0.0079, "step": 5750 }, { "epoch": 0.0942485478196842, "grad_norm": 0.5707820653915405, "learning_rate": 2.356213695492105e-06, "loss": 0.0119, "step": 5760 }, { "epoch": 0.09441217377076004, "grad_norm": 0.5417932271957397, "learning_rate": 2.3603043442690013e-06, "loss": 0.01, "step": 5770 }, { "epoch": 0.09457579972183588, "grad_norm": 0.44638481736183167, "learning_rate": 2.3643949930458975e-06, "loss": 0.0092, "step": 5780 }, { "epoch": 0.09473942567291173, "grad_norm": 0.3540875017642975, "learning_rate": 2.368485641822793e-06, "loss": 0.0103, "step": 5790 }, { "epoch": 0.09490305162398756, "grad_norm": 0.4761086702346802, "learning_rate": 2.3725762905996893e-06, "loss": 0.0105, "step": 5800 }, { "epoch": 0.0950666775750634, "grad_norm": 0.40597155690193176, "learning_rate": 2.3766669393765854e-06, "loss": 0.0085, "step": 5810 }, { "epoch": 0.09523030352613925, "grad_norm": 0.6491827368736267, "learning_rate": 2.3807575881534816e-06, "loss": 0.0092, "step": 5820 }, { "epoch": 0.09539392947721509, "grad_norm": 0.5546706914901733, "learning_rate": 2.3848482369303773e-06, "loss": 0.0118, "step": 5830 }, { "epoch": 0.09555755542829093, "grad_norm": 0.14426052570343018, "learning_rate": 2.3889388857072734e-06, "loss": 0.0083, "step": 5840 }, { "epoch": 0.09572118137936676, "grad_norm": 0.4116996228694916, "learning_rate": 2.3930295344841695e-06, "loss": 0.0099, "step": 5850 }, { "epoch": 0.0958848073304426, "grad_norm": 0.36243778467178345, "learning_rate": 2.3971201832610657e-06, "loss": 0.0099, "step": 5860 }, { "epoch": 0.09604843328151845, "grad_norm": 0.32407140731811523, "learning_rate": 2.4012108320379614e-06, "loss": 0.011, "step": 5870 }, { "epoch": 0.0962120592325943, "grad_norm": 0.36773157119750977, "learning_rate": 2.4053014808148575e-06, "loss": 0.0057, "step": 5880 }, { "epoch": 0.09637568518367012, "grad_norm": 0.45417511463165283, "learning_rate": 2.4093921295917536e-06, "loss": 0.0073, "step": 5890 }, { "epoch": 0.09653931113474597, "grad_norm": 0.4947884976863861, "learning_rate": 2.4134827783686498e-06, "loss": 0.0111, "step": 5900 }, { "epoch": 0.09670293708582181, "grad_norm": 0.33010333776474, "learning_rate": 2.4175734271455455e-06, "loss": 0.0089, "step": 5910 }, { "epoch": 0.09686656303689765, "grad_norm": 0.4588622748851776, "learning_rate": 2.4216640759224416e-06, "loss": 0.0081, "step": 5920 }, { "epoch": 0.0970301889879735, "grad_norm": 0.3258035182952881, "learning_rate": 2.4257547246993377e-06, "loss": 0.0071, "step": 5930 }, { "epoch": 0.09719381493904933, "grad_norm": 0.335820734500885, "learning_rate": 2.4298453734762334e-06, "loss": 0.0079, "step": 5940 }, { "epoch": 0.09735744089012517, "grad_norm": 0.49288108944892883, "learning_rate": 2.4339360222531296e-06, "loss": 0.0115, "step": 5950 }, { "epoch": 0.09752106684120102, "grad_norm": 0.6643833518028259, "learning_rate": 2.4380266710300253e-06, "loss": 0.0103, "step": 5960 }, { "epoch": 0.09768469279227686, "grad_norm": 0.5905048251152039, "learning_rate": 2.4421173198069214e-06, "loss": 0.0104, "step": 5970 }, { "epoch": 0.0978483187433527, "grad_norm": 0.4968055784702301, "learning_rate": 2.4462079685838175e-06, "loss": 0.0091, "step": 5980 }, { "epoch": 0.09801194469442853, "grad_norm": 0.7845309972763062, "learning_rate": 2.4502986173607137e-06, "loss": 0.0066, "step": 5990 }, { "epoch": 0.09817557064550438, "grad_norm": 0.5478293299674988, "learning_rate": 2.4543892661376094e-06, "loss": 0.0093, "step": 6000 }, { "epoch": 0.09833919659658022, "grad_norm": 0.16949962079524994, "learning_rate": 2.4584799149145055e-06, "loss": 0.0111, "step": 6010 }, { "epoch": 0.09850282254765606, "grad_norm": 0.3224465847015381, "learning_rate": 2.4625705636914016e-06, "loss": 0.0093, "step": 6020 }, { "epoch": 0.09866644849873189, "grad_norm": 0.30649334192276, "learning_rate": 2.4666612124682978e-06, "loss": 0.0066, "step": 6030 }, { "epoch": 0.09883007444980774, "grad_norm": 0.5914590358734131, "learning_rate": 2.4707518612451935e-06, "loss": 0.0092, "step": 6040 }, { "epoch": 0.09899370040088358, "grad_norm": 0.3313790559768677, "learning_rate": 2.4748425100220896e-06, "loss": 0.0079, "step": 6050 }, { "epoch": 0.09915732635195942, "grad_norm": 0.09751972556114197, "learning_rate": 2.4789331587989857e-06, "loss": 0.0067, "step": 6060 }, { "epoch": 0.09932095230303527, "grad_norm": 0.5551877617835999, "learning_rate": 2.483023807575882e-06, "loss": 0.0074, "step": 6070 }, { "epoch": 0.0994845782541111, "grad_norm": 0.4545195400714874, "learning_rate": 2.4871144563527776e-06, "loss": 0.0069, "step": 6080 }, { "epoch": 0.09964820420518694, "grad_norm": 0.46272385120391846, "learning_rate": 2.4912051051296737e-06, "loss": 0.0111, "step": 6090 }, { "epoch": 0.09981183015626278, "grad_norm": 0.43556222319602966, "learning_rate": 2.49529575390657e-06, "loss": 0.0055, "step": 6100 }, { "epoch": 0.09997545610733863, "grad_norm": 0.2892121970653534, "learning_rate": 2.499386402683466e-06, "loss": 0.0077, "step": 6110 }, { "epoch": 0.10013908205841446, "grad_norm": 0.31710416078567505, "learning_rate": 2.5034770514603617e-06, "loss": 0.0059, "step": 6120 }, { "epoch": 0.1003027080094903, "grad_norm": 0.37537866830825806, "learning_rate": 2.507567700237258e-06, "loss": 0.0108, "step": 6130 }, { "epoch": 0.10046633396056615, "grad_norm": 0.5153443217277527, "learning_rate": 2.511658349014154e-06, "loss": 0.0078, "step": 6140 }, { "epoch": 0.10062995991164199, "grad_norm": 0.46366602182388306, "learning_rate": 2.5157489977910497e-06, "loss": 0.01, "step": 6150 }, { "epoch": 0.10079358586271783, "grad_norm": 0.5856673717498779, "learning_rate": 2.5198396465679458e-06, "loss": 0.0104, "step": 6160 }, { "epoch": 0.10095721181379366, "grad_norm": 0.40452584624290466, "learning_rate": 2.523930295344842e-06, "loss": 0.0105, "step": 6170 }, { "epoch": 0.1011208377648695, "grad_norm": 0.5706951022148132, "learning_rate": 2.528020944121738e-06, "loss": 0.0096, "step": 6180 }, { "epoch": 0.10128446371594535, "grad_norm": 0.5840921998023987, "learning_rate": 2.5321115928986338e-06, "loss": 0.0098, "step": 6190 }, { "epoch": 0.1014480896670212, "grad_norm": 0.7124074697494507, "learning_rate": 2.53620224167553e-06, "loss": 0.0092, "step": 6200 }, { "epoch": 0.10161171561809704, "grad_norm": 0.5009989142417908, "learning_rate": 2.540292890452426e-06, "loss": 0.006, "step": 6210 }, { "epoch": 0.10177534156917287, "grad_norm": 0.30747249722480774, "learning_rate": 2.544383539229322e-06, "loss": 0.0098, "step": 6220 }, { "epoch": 0.10193896752024871, "grad_norm": 0.4699268937110901, "learning_rate": 2.548474188006218e-06, "loss": 0.0084, "step": 6230 }, { "epoch": 0.10210259347132455, "grad_norm": 0.6360853314399719, "learning_rate": 2.552564836783114e-06, "loss": 0.0078, "step": 6240 }, { "epoch": 0.1022662194224004, "grad_norm": 0.3140677213668823, "learning_rate": 2.55665548556001e-06, "loss": 0.007, "step": 6250 }, { "epoch": 0.10242984537347623, "grad_norm": 0.3698727786540985, "learning_rate": 2.5607461343369063e-06, "loss": 0.0061, "step": 6260 }, { "epoch": 0.10259347132455207, "grad_norm": 0.3195227384567261, "learning_rate": 2.564836783113802e-06, "loss": 0.0074, "step": 6270 }, { "epoch": 0.10275709727562791, "grad_norm": 0.32058024406433105, "learning_rate": 2.568927431890698e-06, "loss": 0.0097, "step": 6280 }, { "epoch": 0.10292072322670376, "grad_norm": 0.5562793612480164, "learning_rate": 2.5730180806675942e-06, "loss": 0.0092, "step": 6290 }, { "epoch": 0.1030843491777796, "grad_norm": 0.32858577370643616, "learning_rate": 2.5771087294444904e-06, "loss": 0.0074, "step": 6300 }, { "epoch": 0.10324797512885543, "grad_norm": 0.44006767868995667, "learning_rate": 2.581199378221386e-06, "loss": 0.008, "step": 6310 }, { "epoch": 0.10341160107993128, "grad_norm": 0.49075889587402344, "learning_rate": 2.585290026998282e-06, "loss": 0.0135, "step": 6320 }, { "epoch": 0.10357522703100712, "grad_norm": 0.5965510606765747, "learning_rate": 2.5893806757751783e-06, "loss": 0.0107, "step": 6330 }, { "epoch": 0.10373885298208296, "grad_norm": 0.3659735321998596, "learning_rate": 2.5934713245520745e-06, "loss": 0.0094, "step": 6340 }, { "epoch": 0.10390247893315879, "grad_norm": 0.4217313230037689, "learning_rate": 2.59756197332897e-06, "loss": 0.009, "step": 6350 }, { "epoch": 0.10406610488423464, "grad_norm": 0.41710585355758667, "learning_rate": 2.6016526221058663e-06, "loss": 0.0109, "step": 6360 }, { "epoch": 0.10422973083531048, "grad_norm": 0.40991777181625366, "learning_rate": 2.6057432708827624e-06, "loss": 0.0085, "step": 6370 }, { "epoch": 0.10439335678638632, "grad_norm": 0.643492579460144, "learning_rate": 2.6098339196596586e-06, "loss": 0.0064, "step": 6380 }, { "epoch": 0.10455698273746217, "grad_norm": 0.2730458080768585, "learning_rate": 2.6139245684365543e-06, "loss": 0.0083, "step": 6390 }, { "epoch": 0.104720608688538, "grad_norm": 0.5287303924560547, "learning_rate": 2.6180152172134504e-06, "loss": 0.0097, "step": 6400 }, { "epoch": 0.10488423463961384, "grad_norm": 0.7431327104568481, "learning_rate": 2.6221058659903465e-06, "loss": 0.0104, "step": 6410 }, { "epoch": 0.10504786059068968, "grad_norm": 0.7869155406951904, "learning_rate": 2.6261965147672427e-06, "loss": 0.0077, "step": 6420 }, { "epoch": 0.10521148654176553, "grad_norm": 0.5374842882156372, "learning_rate": 2.6302871635441384e-06, "loss": 0.0099, "step": 6430 }, { "epoch": 0.10537511249284136, "grad_norm": 0.3151998519897461, "learning_rate": 2.6343778123210345e-06, "loss": 0.0071, "step": 6440 }, { "epoch": 0.1055387384439172, "grad_norm": 0.18780340254306793, "learning_rate": 2.6384684610979306e-06, "loss": 0.0077, "step": 6450 }, { "epoch": 0.10570236439499305, "grad_norm": 0.24858160316944122, "learning_rate": 2.6425591098748268e-06, "loss": 0.0093, "step": 6460 }, { "epoch": 0.10586599034606889, "grad_norm": 0.45628657937049866, "learning_rate": 2.6466497586517225e-06, "loss": 0.011, "step": 6470 }, { "epoch": 0.10602961629714473, "grad_norm": 0.241099014878273, "learning_rate": 2.6507404074286186e-06, "loss": 0.009, "step": 6480 }, { "epoch": 0.10619324224822056, "grad_norm": 0.8430644869804382, "learning_rate": 2.6548310562055147e-06, "loss": 0.0123, "step": 6490 }, { "epoch": 0.1063568681992964, "grad_norm": 0.35791537165641785, "learning_rate": 2.6589217049824104e-06, "loss": 0.0059, "step": 6500 }, { "epoch": 0.10652049415037225, "grad_norm": 0.4280668795108795, "learning_rate": 2.6630123537593066e-06, "loss": 0.0104, "step": 6510 }, { "epoch": 0.1066841201014481, "grad_norm": 0.45465826988220215, "learning_rate": 2.6671030025362027e-06, "loss": 0.0084, "step": 6520 }, { "epoch": 0.10684774605252394, "grad_norm": 0.3093436062335968, "learning_rate": 2.671193651313099e-06, "loss": 0.008, "step": 6530 }, { "epoch": 0.10701137200359977, "grad_norm": 0.5135814547538757, "learning_rate": 2.6752843000899945e-06, "loss": 0.01, "step": 6540 }, { "epoch": 0.10717499795467561, "grad_norm": 0.45517611503601074, "learning_rate": 2.6793749488668907e-06, "loss": 0.0062, "step": 6550 }, { "epoch": 0.10733862390575145, "grad_norm": 1.0340131521224976, "learning_rate": 2.683465597643787e-06, "loss": 0.0077, "step": 6560 }, { "epoch": 0.1075022498568273, "grad_norm": 0.6461662650108337, "learning_rate": 2.687556246420683e-06, "loss": 0.0108, "step": 6570 }, { "epoch": 0.10766587580790313, "grad_norm": 0.5310466885566711, "learning_rate": 2.6916468951975786e-06, "loss": 0.007, "step": 6580 }, { "epoch": 0.10782950175897897, "grad_norm": 0.4045282006263733, "learning_rate": 2.6957375439744748e-06, "loss": 0.0098, "step": 6590 }, { "epoch": 0.10799312771005481, "grad_norm": 0.44540032744407654, "learning_rate": 2.6998281927513705e-06, "loss": 0.0086, "step": 6600 }, { "epoch": 0.10815675366113066, "grad_norm": 0.5340131521224976, "learning_rate": 2.703918841528266e-06, "loss": 0.0152, "step": 6610 }, { "epoch": 0.1083203796122065, "grad_norm": 0.4870081841945648, "learning_rate": 2.7080094903051623e-06, "loss": 0.0073, "step": 6620 }, { "epoch": 0.10848400556328233, "grad_norm": 0.3604751527309418, "learning_rate": 2.7121001390820585e-06, "loss": 0.0056, "step": 6630 }, { "epoch": 0.10864763151435818, "grad_norm": 0.5349363088607788, "learning_rate": 2.7161907878589546e-06, "loss": 0.012, "step": 6640 }, { "epoch": 0.10881125746543402, "grad_norm": 0.29263758659362793, "learning_rate": 2.7202814366358503e-06, "loss": 0.0049, "step": 6650 }, { "epoch": 0.10897488341650986, "grad_norm": 0.2576255202293396, "learning_rate": 2.7243720854127464e-06, "loss": 0.0073, "step": 6660 }, { "epoch": 0.10913850936758569, "grad_norm": 0.2516396641731262, "learning_rate": 2.7284627341896426e-06, "loss": 0.0085, "step": 6670 }, { "epoch": 0.10930213531866154, "grad_norm": 0.4531979262828827, "learning_rate": 2.7325533829665387e-06, "loss": 0.0077, "step": 6680 }, { "epoch": 0.10946576126973738, "grad_norm": 0.6014463901519775, "learning_rate": 2.7366440317434344e-06, "loss": 0.0075, "step": 6690 }, { "epoch": 0.10962938722081322, "grad_norm": 0.28603222966194153, "learning_rate": 2.7407346805203305e-06, "loss": 0.0121, "step": 6700 }, { "epoch": 0.10979301317188907, "grad_norm": 0.38990697264671326, "learning_rate": 2.7448253292972267e-06, "loss": 0.0076, "step": 6710 }, { "epoch": 0.1099566391229649, "grad_norm": 0.7766194343566895, "learning_rate": 2.748915978074123e-06, "loss": 0.0136, "step": 6720 }, { "epoch": 0.11012026507404074, "grad_norm": 0.3501160740852356, "learning_rate": 2.7530066268510185e-06, "loss": 0.007, "step": 6730 }, { "epoch": 0.11028389102511658, "grad_norm": 0.4071316421031952, "learning_rate": 2.7570972756279146e-06, "loss": 0.0058, "step": 6740 }, { "epoch": 0.11044751697619243, "grad_norm": 0.23193137347698212, "learning_rate": 2.7611879244048108e-06, "loss": 0.0063, "step": 6750 }, { "epoch": 0.11061114292726827, "grad_norm": 0.5020021796226501, "learning_rate": 2.7652785731817065e-06, "loss": 0.0087, "step": 6760 }, { "epoch": 0.1107747688783441, "grad_norm": 0.4488350450992584, "learning_rate": 2.7693692219586026e-06, "loss": 0.007, "step": 6770 }, { "epoch": 0.11093839482941995, "grad_norm": 0.42864692211151123, "learning_rate": 2.7734598707354987e-06, "loss": 0.0072, "step": 6780 }, { "epoch": 0.11110202078049579, "grad_norm": 0.3764258027076721, "learning_rate": 2.777550519512395e-06, "loss": 0.0055, "step": 6790 }, { "epoch": 0.11126564673157163, "grad_norm": 0.3585885465145111, "learning_rate": 2.7816411682892906e-06, "loss": 0.0059, "step": 6800 }, { "epoch": 0.11142927268264746, "grad_norm": 0.356059730052948, "learning_rate": 2.7857318170661867e-06, "loss": 0.0119, "step": 6810 }, { "epoch": 0.1115928986337233, "grad_norm": 0.31881484389305115, "learning_rate": 2.789822465843083e-06, "loss": 0.0114, "step": 6820 }, { "epoch": 0.11175652458479915, "grad_norm": 0.5487472414970398, "learning_rate": 2.793913114619979e-06, "loss": 0.0102, "step": 6830 }, { "epoch": 0.111920150535875, "grad_norm": 0.17071464657783508, "learning_rate": 2.7980037633968747e-06, "loss": 0.0097, "step": 6840 }, { "epoch": 0.11208377648695084, "grad_norm": 0.3571915328502655, "learning_rate": 2.802094412173771e-06, "loss": 0.0088, "step": 6850 }, { "epoch": 0.11224740243802667, "grad_norm": 0.4432998597621918, "learning_rate": 2.806185060950667e-06, "loss": 0.0099, "step": 6860 }, { "epoch": 0.11241102838910251, "grad_norm": 0.4738149344921112, "learning_rate": 2.810275709727563e-06, "loss": 0.0153, "step": 6870 }, { "epoch": 0.11257465434017835, "grad_norm": 0.7653093338012695, "learning_rate": 2.8143663585044588e-06, "loss": 0.0094, "step": 6880 }, { "epoch": 0.1127382802912542, "grad_norm": 0.3688839375972748, "learning_rate": 2.818457007281355e-06, "loss": 0.0045, "step": 6890 }, { "epoch": 0.11290190624233003, "grad_norm": 0.13069376349449158, "learning_rate": 2.822547656058251e-06, "loss": 0.0123, "step": 6900 }, { "epoch": 0.11306553219340587, "grad_norm": 0.33732107281684875, "learning_rate": 2.826638304835147e-06, "loss": 0.0081, "step": 6910 }, { "epoch": 0.11322915814448171, "grad_norm": 0.20888225734233856, "learning_rate": 2.830728953612043e-06, "loss": 0.0067, "step": 6920 }, { "epoch": 0.11339278409555756, "grad_norm": 0.3723088204860687, "learning_rate": 2.834819602388939e-06, "loss": 0.0079, "step": 6930 }, { "epoch": 0.1135564100466334, "grad_norm": 0.4187650680541992, "learning_rate": 2.838910251165835e-06, "loss": 0.0072, "step": 6940 }, { "epoch": 0.11372003599770923, "grad_norm": 0.4980318546295166, "learning_rate": 2.8430008999427313e-06, "loss": 0.007, "step": 6950 }, { "epoch": 0.11388366194878508, "grad_norm": 0.3013603091239929, "learning_rate": 2.847091548719627e-06, "loss": 0.0079, "step": 6960 }, { "epoch": 0.11404728789986092, "grad_norm": 0.3569165766239166, "learning_rate": 2.851182197496523e-06, "loss": 0.0104, "step": 6970 }, { "epoch": 0.11421091385093676, "grad_norm": 0.25310376286506653, "learning_rate": 2.8552728462734192e-06, "loss": 0.0093, "step": 6980 }, { "epoch": 0.11437453980201259, "grad_norm": 0.43188145756721497, "learning_rate": 2.8593634950503154e-06, "loss": 0.0085, "step": 6990 }, { "epoch": 0.11453816575308844, "grad_norm": 0.46280843019485474, "learning_rate": 2.863454143827211e-06, "loss": 0.0071, "step": 7000 }, { "epoch": 0.11470179170416428, "grad_norm": 0.39597222208976746, "learning_rate": 2.8675447926041072e-06, "loss": 0.0067, "step": 7010 }, { "epoch": 0.11486541765524012, "grad_norm": 0.37816449999809265, "learning_rate": 2.8716354413810033e-06, "loss": 0.0075, "step": 7020 }, { "epoch": 0.11502904360631597, "grad_norm": 0.371407151222229, "learning_rate": 2.8757260901578995e-06, "loss": 0.0097, "step": 7030 }, { "epoch": 0.1151926695573918, "grad_norm": 0.6381232142448425, "learning_rate": 2.879816738934795e-06, "loss": 0.0108, "step": 7040 }, { "epoch": 0.11535629550846764, "grad_norm": 1.037317156791687, "learning_rate": 2.8839073877116913e-06, "loss": 0.0086, "step": 7050 }, { "epoch": 0.11551992145954348, "grad_norm": 0.18345707654953003, "learning_rate": 2.8879980364885874e-06, "loss": 0.0075, "step": 7060 }, { "epoch": 0.11568354741061933, "grad_norm": 0.35895949602127075, "learning_rate": 2.892088685265483e-06, "loss": 0.0099, "step": 7070 }, { "epoch": 0.11584717336169517, "grad_norm": 0.552554726600647, "learning_rate": 2.8961793340423793e-06, "loss": 0.0118, "step": 7080 }, { "epoch": 0.116010799312771, "grad_norm": 0.16110466420650482, "learning_rate": 2.9002699828192754e-06, "loss": 0.0098, "step": 7090 }, { "epoch": 0.11617442526384684, "grad_norm": 0.6598643660545349, "learning_rate": 2.9043606315961715e-06, "loss": 0.0085, "step": 7100 }, { "epoch": 0.11633805121492269, "grad_norm": 0.4332290291786194, "learning_rate": 2.9084512803730673e-06, "loss": 0.0085, "step": 7110 }, { "epoch": 0.11650167716599853, "grad_norm": 0.2834670841693878, "learning_rate": 2.9125419291499634e-06, "loss": 0.0083, "step": 7120 }, { "epoch": 0.11666530311707436, "grad_norm": 0.577721893787384, "learning_rate": 2.9166325779268595e-06, "loss": 0.0094, "step": 7130 }, { "epoch": 0.1168289290681502, "grad_norm": 0.3473556935787201, "learning_rate": 2.9207232267037557e-06, "loss": 0.0061, "step": 7140 }, { "epoch": 0.11699255501922605, "grad_norm": 0.316207617521286, "learning_rate": 2.9248138754806514e-06, "loss": 0.0091, "step": 7150 }, { "epoch": 0.11715618097030189, "grad_norm": 0.49986520409584045, "learning_rate": 2.9289045242575475e-06, "loss": 0.0094, "step": 7160 }, { "epoch": 0.11731980692137774, "grad_norm": 0.6766525506973267, "learning_rate": 2.9329951730344436e-06, "loss": 0.007, "step": 7170 }, { "epoch": 0.11748343287245357, "grad_norm": 0.2860639691352844, "learning_rate": 2.9370858218113398e-06, "loss": 0.0078, "step": 7180 }, { "epoch": 0.11764705882352941, "grad_norm": 0.6770176291465759, "learning_rate": 2.9411764705882355e-06, "loss": 0.0097, "step": 7190 }, { "epoch": 0.11781068477460525, "grad_norm": 0.5268541574478149, "learning_rate": 2.9452671193651316e-06, "loss": 0.0057, "step": 7200 }, { "epoch": 0.1179743107256811, "grad_norm": 0.317954421043396, "learning_rate": 2.9493577681420277e-06, "loss": 0.0081, "step": 7210 }, { "epoch": 0.11813793667675693, "grad_norm": 0.7357617616653442, "learning_rate": 2.953448416918924e-06, "loss": 0.0083, "step": 7220 }, { "epoch": 0.11830156262783277, "grad_norm": 0.5980863571166992, "learning_rate": 2.9575390656958196e-06, "loss": 0.0057, "step": 7230 }, { "epoch": 0.11846518857890861, "grad_norm": 0.40455174446105957, "learning_rate": 2.9616297144727157e-06, "loss": 0.0106, "step": 7240 }, { "epoch": 0.11862881452998446, "grad_norm": 0.21107572317123413, "learning_rate": 2.965720363249612e-06, "loss": 0.0043, "step": 7250 }, { "epoch": 0.1187924404810603, "grad_norm": 0.09314513951539993, "learning_rate": 2.969811012026508e-06, "loss": 0.0114, "step": 7260 }, { "epoch": 0.11895606643213613, "grad_norm": 0.21563343703746796, "learning_rate": 2.9739016608034037e-06, "loss": 0.0068, "step": 7270 }, { "epoch": 0.11911969238321198, "grad_norm": 1.262451410293579, "learning_rate": 2.9779923095803e-06, "loss": 0.0073, "step": 7280 }, { "epoch": 0.11928331833428782, "grad_norm": 0.4878429174423218, "learning_rate": 2.982082958357196e-06, "loss": 0.0083, "step": 7290 }, { "epoch": 0.11944694428536366, "grad_norm": 0.21024684607982635, "learning_rate": 2.986173607134092e-06, "loss": 0.0062, "step": 7300 }, { "epoch": 0.11961057023643949, "grad_norm": 0.3616296350955963, "learning_rate": 2.9902642559109878e-06, "loss": 0.0083, "step": 7310 }, { "epoch": 0.11977419618751534, "grad_norm": 0.33539247512817383, "learning_rate": 2.994354904687884e-06, "loss": 0.0075, "step": 7320 }, { "epoch": 0.11993782213859118, "grad_norm": 0.303585410118103, "learning_rate": 2.99844555346478e-06, "loss": 0.007, "step": 7330 }, { "epoch": 0.12010144808966702, "grad_norm": 0.5015720725059509, "learning_rate": 3.002536202241676e-06, "loss": 0.0068, "step": 7340 }, { "epoch": 0.12026507404074287, "grad_norm": 0.4114151895046234, "learning_rate": 3.006626851018572e-06, "loss": 0.0133, "step": 7350 }, { "epoch": 0.1204286999918187, "grad_norm": 0.39374297857284546, "learning_rate": 3.010717499795468e-06, "loss": 0.0079, "step": 7360 }, { "epoch": 0.12059232594289454, "grad_norm": 0.7024542689323425, "learning_rate": 3.014808148572364e-06, "loss": 0.01, "step": 7370 }, { "epoch": 0.12075595189397038, "grad_norm": 0.36882203817367554, "learning_rate": 3.0188987973492603e-06, "loss": 0.0071, "step": 7380 }, { "epoch": 0.12091957784504623, "grad_norm": 0.5441009998321533, "learning_rate": 3.022989446126156e-06, "loss": 0.0135, "step": 7390 }, { "epoch": 0.12108320379612207, "grad_norm": 0.45814159512519836, "learning_rate": 3.027080094903052e-06, "loss": 0.0061, "step": 7400 }, { "epoch": 0.1212468297471979, "grad_norm": 0.29454901814460754, "learning_rate": 3.0311707436799482e-06, "loss": 0.007, "step": 7410 }, { "epoch": 0.12141045569827374, "grad_norm": 0.2151595503091812, "learning_rate": 3.035261392456844e-06, "loss": 0.0107, "step": 7420 }, { "epoch": 0.12157408164934959, "grad_norm": 0.3251050114631653, "learning_rate": 3.03935204123374e-06, "loss": 0.0089, "step": 7430 }, { "epoch": 0.12173770760042543, "grad_norm": 0.4839686453342438, "learning_rate": 3.043442690010636e-06, "loss": 0.006, "step": 7440 }, { "epoch": 0.12190133355150126, "grad_norm": 0.10744336992502213, "learning_rate": 3.0475333387875323e-06, "loss": 0.0087, "step": 7450 }, { "epoch": 0.1220649595025771, "grad_norm": 0.38201701641082764, "learning_rate": 3.051623987564428e-06, "loss": 0.0101, "step": 7460 }, { "epoch": 0.12222858545365295, "grad_norm": 0.5543551445007324, "learning_rate": 3.055714636341324e-06, "loss": 0.0053, "step": 7470 }, { "epoch": 0.12239221140472879, "grad_norm": 0.25225088000297546, "learning_rate": 3.05980528511822e-06, "loss": 0.0057, "step": 7480 }, { "epoch": 0.12255583735580464, "grad_norm": 0.2500893175601959, "learning_rate": 3.0638959338951156e-06, "loss": 0.0076, "step": 7490 }, { "epoch": 0.12271946330688047, "grad_norm": 0.39099547266960144, "learning_rate": 3.0679865826720117e-06, "loss": 0.0085, "step": 7500 }, { "epoch": 0.12288308925795631, "grad_norm": 0.44680076837539673, "learning_rate": 3.072077231448908e-06, "loss": 0.0121, "step": 7510 }, { "epoch": 0.12304671520903215, "grad_norm": 1.2046732902526855, "learning_rate": 3.076167880225804e-06, "loss": 0.0087, "step": 7520 }, { "epoch": 0.123210341160108, "grad_norm": 0.6178827285766602, "learning_rate": 3.0802585290026997e-06, "loss": 0.0057, "step": 7530 }, { "epoch": 0.12337396711118383, "grad_norm": 0.4025334417819977, "learning_rate": 3.084349177779596e-06, "loss": 0.0102, "step": 7540 }, { "epoch": 0.12353759306225967, "grad_norm": 0.3102932572364807, "learning_rate": 3.088439826556492e-06, "loss": 0.0086, "step": 7550 }, { "epoch": 0.12370121901333551, "grad_norm": 0.22910022735595703, "learning_rate": 3.092530475333388e-06, "loss": 0.0079, "step": 7560 }, { "epoch": 0.12386484496441136, "grad_norm": 0.6808933615684509, "learning_rate": 3.096621124110284e-06, "loss": 0.0081, "step": 7570 }, { "epoch": 0.1240284709154872, "grad_norm": 0.37424010038375854, "learning_rate": 3.10071177288718e-06, "loss": 0.0065, "step": 7580 }, { "epoch": 0.12419209686656303, "grad_norm": 0.451021671295166, "learning_rate": 3.104802421664076e-06, "loss": 0.0096, "step": 7590 }, { "epoch": 0.12435572281763888, "grad_norm": 0.49796557426452637, "learning_rate": 3.108893070440972e-06, "loss": 0.008, "step": 7600 }, { "epoch": 0.12451934876871472, "grad_norm": 0.6674835085868835, "learning_rate": 3.112983719217868e-06, "loss": 0.0083, "step": 7610 }, { "epoch": 0.12468297471979056, "grad_norm": 0.4237474203109741, "learning_rate": 3.117074367994764e-06, "loss": 0.0088, "step": 7620 }, { "epoch": 0.1248466006708664, "grad_norm": 0.7019473314285278, "learning_rate": 3.12116501677166e-06, "loss": 0.0069, "step": 7630 }, { "epoch": 0.12501022662194225, "grad_norm": 0.4018499553203583, "learning_rate": 3.1252556655485563e-06, "loss": 0.0059, "step": 7640 }, { "epoch": 0.12517385257301808, "grad_norm": 0.779705822467804, "learning_rate": 3.129346314325452e-06, "loss": 0.0069, "step": 7650 }, { "epoch": 0.1253374785240939, "grad_norm": 0.6239704489707947, "learning_rate": 3.133436963102348e-06, "loss": 0.0073, "step": 7660 }, { "epoch": 0.12550110447516977, "grad_norm": 0.40643253922462463, "learning_rate": 3.1375276118792443e-06, "loss": 0.006, "step": 7670 }, { "epoch": 0.1256647304262456, "grad_norm": 0.4331013858318329, "learning_rate": 3.14161826065614e-06, "loss": 0.0101, "step": 7680 }, { "epoch": 0.12582835637732145, "grad_norm": 0.45190146565437317, "learning_rate": 3.145708909433036e-06, "loss": 0.0124, "step": 7690 }, { "epoch": 0.12599198232839728, "grad_norm": 0.3329585790634155, "learning_rate": 3.1497995582099322e-06, "loss": 0.0075, "step": 7700 }, { "epoch": 0.1261556082794731, "grad_norm": 0.2454882562160492, "learning_rate": 3.1538902069868284e-06, "loss": 0.0059, "step": 7710 }, { "epoch": 0.12631923423054897, "grad_norm": 0.2036505490541458, "learning_rate": 3.157980855763724e-06, "loss": 0.0061, "step": 7720 }, { "epoch": 0.1264828601816248, "grad_norm": 0.1963091641664505, "learning_rate": 3.16207150454062e-06, "loss": 0.007, "step": 7730 }, { "epoch": 0.12664648613270066, "grad_norm": 0.36304619908332825, "learning_rate": 3.1661621533175163e-06, "loss": 0.0068, "step": 7740 }, { "epoch": 0.1268101120837765, "grad_norm": 0.5582684278488159, "learning_rate": 3.1702528020944125e-06, "loss": 0.007, "step": 7750 }, { "epoch": 0.12697373803485232, "grad_norm": 0.31853148341178894, "learning_rate": 3.174343450871308e-06, "loss": 0.0096, "step": 7760 }, { "epoch": 0.12713736398592818, "grad_norm": 0.34923887252807617, "learning_rate": 3.1784340996482043e-06, "loss": 0.0112, "step": 7770 }, { "epoch": 0.127300989937004, "grad_norm": 0.48723575472831726, "learning_rate": 3.1825247484251004e-06, "loss": 0.0057, "step": 7780 }, { "epoch": 0.12746461588807986, "grad_norm": 0.3825056552886963, "learning_rate": 3.1866153972019966e-06, "loss": 0.0095, "step": 7790 }, { "epoch": 0.1276282418391557, "grad_norm": 0.5003986358642578, "learning_rate": 3.1907060459788923e-06, "loss": 0.0092, "step": 7800 }, { "epoch": 0.12779186779023152, "grad_norm": 0.482189416885376, "learning_rate": 3.1947966947557884e-06, "loss": 0.0105, "step": 7810 }, { "epoch": 0.12795549374130738, "grad_norm": 0.38571682572364807, "learning_rate": 3.1988873435326845e-06, "loss": 0.0042, "step": 7820 }, { "epoch": 0.1281191196923832, "grad_norm": 0.5222697257995605, "learning_rate": 3.2029779923095807e-06, "loss": 0.0076, "step": 7830 }, { "epoch": 0.12828274564345904, "grad_norm": 0.7520259618759155, "learning_rate": 3.2070686410864764e-06, "loss": 0.0095, "step": 7840 }, { "epoch": 0.1284463715945349, "grad_norm": 0.21808841824531555, "learning_rate": 3.2111592898633725e-06, "loss": 0.0042, "step": 7850 }, { "epoch": 0.12860999754561073, "grad_norm": 0.3986198604106903, "learning_rate": 3.2152499386402686e-06, "loss": 0.0102, "step": 7860 }, { "epoch": 0.12877362349668658, "grad_norm": 0.3703888952732086, "learning_rate": 3.2193405874171648e-06, "loss": 0.006, "step": 7870 }, { "epoch": 0.12893724944776241, "grad_norm": 0.222367525100708, "learning_rate": 3.2234312361940605e-06, "loss": 0.0084, "step": 7880 }, { "epoch": 0.12910087539883824, "grad_norm": 0.3565288782119751, "learning_rate": 3.2275218849709566e-06, "loss": 0.0073, "step": 7890 }, { "epoch": 0.1292645013499141, "grad_norm": 0.3214541971683502, "learning_rate": 3.2316125337478527e-06, "loss": 0.0084, "step": 7900 }, { "epoch": 0.12942812730098993, "grad_norm": 0.5277888178825378, "learning_rate": 3.235703182524749e-06, "loss": 0.0107, "step": 7910 }, { "epoch": 0.1295917532520658, "grad_norm": 0.24659627676010132, "learning_rate": 3.2397938313016446e-06, "loss": 0.0054, "step": 7920 }, { "epoch": 0.12975537920314162, "grad_norm": 0.45949915051460266, "learning_rate": 3.2438844800785407e-06, "loss": 0.0071, "step": 7930 }, { "epoch": 0.12991900515421745, "grad_norm": 0.21689455211162567, "learning_rate": 3.247975128855437e-06, "loss": 0.0057, "step": 7940 }, { "epoch": 0.1300826311052933, "grad_norm": 0.3042140007019043, "learning_rate": 3.252065777632333e-06, "loss": 0.0085, "step": 7950 }, { "epoch": 0.13024625705636914, "grad_norm": 0.7746505737304688, "learning_rate": 3.2561564264092287e-06, "loss": 0.0062, "step": 7960 }, { "epoch": 0.130409883007445, "grad_norm": 0.28350093960762024, "learning_rate": 3.260247075186125e-06, "loss": 0.0066, "step": 7970 }, { "epoch": 0.13057350895852082, "grad_norm": 0.23668046295642853, "learning_rate": 3.264337723963021e-06, "loss": 0.0063, "step": 7980 }, { "epoch": 0.13073713490959665, "grad_norm": 0.3025921881198883, "learning_rate": 3.268428372739917e-06, "loss": 0.0088, "step": 7990 }, { "epoch": 0.1309007608606725, "grad_norm": 0.5452218651771545, "learning_rate": 3.2725190215168128e-06, "loss": 0.0097, "step": 8000 }, { "epoch": 0.13106438681174834, "grad_norm": 0.674902617931366, "learning_rate": 3.276609670293709e-06, "loss": 0.0082, "step": 8010 }, { "epoch": 0.1312280127628242, "grad_norm": 0.33774250745773315, "learning_rate": 3.280700319070605e-06, "loss": 0.006, "step": 8020 }, { "epoch": 0.13139163871390003, "grad_norm": 0.48532918095588684, "learning_rate": 3.2847909678475008e-06, "loss": 0.0071, "step": 8030 }, { "epoch": 0.13155526466497586, "grad_norm": 0.1250019520521164, "learning_rate": 3.288881616624397e-06, "loss": 0.0071, "step": 8040 }, { "epoch": 0.13171889061605171, "grad_norm": 0.33212506771087646, "learning_rate": 3.292972265401293e-06, "loss": 0.0061, "step": 8050 }, { "epoch": 0.13188251656712754, "grad_norm": 0.2655091881752014, "learning_rate": 3.297062914178189e-06, "loss": 0.0043, "step": 8060 }, { "epoch": 0.13204614251820337, "grad_norm": 0.4257424771785736, "learning_rate": 3.301153562955085e-06, "loss": 0.0068, "step": 8070 }, { "epoch": 0.13220976846927923, "grad_norm": 0.26023197174072266, "learning_rate": 3.305244211731981e-06, "loss": 0.0074, "step": 8080 }, { "epoch": 0.13237339442035506, "grad_norm": 0.31127622723579407, "learning_rate": 3.309334860508877e-06, "loss": 0.0061, "step": 8090 }, { "epoch": 0.13253702037143092, "grad_norm": 0.5677733421325684, "learning_rate": 3.3134255092857733e-06, "loss": 0.0049, "step": 8100 }, { "epoch": 0.13270064632250675, "grad_norm": 0.24019241333007812, "learning_rate": 3.317516158062669e-06, "loss": 0.0058, "step": 8110 }, { "epoch": 0.13286427227358258, "grad_norm": 0.3517206311225891, "learning_rate": 3.321606806839565e-06, "loss": 0.0056, "step": 8120 }, { "epoch": 0.13302789822465844, "grad_norm": 0.4195566773414612, "learning_rate": 3.3256974556164612e-06, "loss": 0.0073, "step": 8130 }, { "epoch": 0.13319152417573427, "grad_norm": 0.5175689458847046, "learning_rate": 3.3297881043933574e-06, "loss": 0.0053, "step": 8140 }, { "epoch": 0.13335515012681012, "grad_norm": 0.497042715549469, "learning_rate": 3.333878753170253e-06, "loss": 0.0079, "step": 8150 }, { "epoch": 0.13351877607788595, "grad_norm": 0.49808087944984436, "learning_rate": 3.337969401947149e-06, "loss": 0.0107, "step": 8160 }, { "epoch": 0.13368240202896178, "grad_norm": 0.37722522020339966, "learning_rate": 3.3420600507240453e-06, "loss": 0.0066, "step": 8170 }, { "epoch": 0.13384602798003764, "grad_norm": 0.5292481780052185, "learning_rate": 3.3461506995009415e-06, "loss": 0.0088, "step": 8180 }, { "epoch": 0.13400965393111347, "grad_norm": 0.19702689349651337, "learning_rate": 3.350241348277837e-06, "loss": 0.0068, "step": 8190 }, { "epoch": 0.13417327988218933, "grad_norm": 0.3749525547027588, "learning_rate": 3.3543319970547333e-06, "loss": 0.0111, "step": 8200 }, { "epoch": 0.13433690583326516, "grad_norm": 0.4897129237651825, "learning_rate": 3.3584226458316294e-06, "loss": 0.0077, "step": 8210 }, { "epoch": 0.134500531784341, "grad_norm": 0.3965592384338379, "learning_rate": 3.3625132946085256e-06, "loss": 0.0068, "step": 8220 }, { "epoch": 0.13466415773541685, "grad_norm": 0.45412561297416687, "learning_rate": 3.3666039433854213e-06, "loss": 0.0077, "step": 8230 }, { "epoch": 0.13482778368649267, "grad_norm": 0.4656006693840027, "learning_rate": 3.3706945921623174e-06, "loss": 0.0073, "step": 8240 }, { "epoch": 0.1349914096375685, "grad_norm": 0.29646602272987366, "learning_rate": 3.3747852409392135e-06, "loss": 0.0057, "step": 8250 }, { "epoch": 0.13515503558864436, "grad_norm": 0.44818490743637085, "learning_rate": 3.3788758897161097e-06, "loss": 0.0066, "step": 8260 }, { "epoch": 0.1353186615397202, "grad_norm": 0.34393343329429626, "learning_rate": 3.3829665384930054e-06, "loss": 0.0054, "step": 8270 }, { "epoch": 0.13548228749079605, "grad_norm": 0.40935346484184265, "learning_rate": 3.3870571872699015e-06, "loss": 0.0058, "step": 8280 }, { "epoch": 0.13564591344187188, "grad_norm": 0.5371100306510925, "learning_rate": 3.3911478360467976e-06, "loss": 0.0081, "step": 8290 }, { "epoch": 0.1358095393929477, "grad_norm": 0.6072351932525635, "learning_rate": 3.3952384848236938e-06, "loss": 0.0106, "step": 8300 }, { "epoch": 0.13597316534402357, "grad_norm": 0.27352070808410645, "learning_rate": 3.3993291336005895e-06, "loss": 0.0111, "step": 8310 }, { "epoch": 0.1361367912950994, "grad_norm": 0.5729878544807434, "learning_rate": 3.4034197823774856e-06, "loss": 0.0086, "step": 8320 }, { "epoch": 0.13630041724617525, "grad_norm": 0.2698054015636444, "learning_rate": 3.4075104311543817e-06, "loss": 0.0066, "step": 8330 }, { "epoch": 0.13646404319725108, "grad_norm": 0.23836910724639893, "learning_rate": 3.4116010799312774e-06, "loss": 0.0075, "step": 8340 }, { "epoch": 0.1366276691483269, "grad_norm": 0.42263901233673096, "learning_rate": 3.4156917287081736e-06, "loss": 0.0063, "step": 8350 }, { "epoch": 0.13679129509940277, "grad_norm": 0.5607989430427551, "learning_rate": 3.4197823774850693e-06, "loss": 0.0049, "step": 8360 }, { "epoch": 0.1369549210504786, "grad_norm": 0.2147042453289032, "learning_rate": 3.423873026261965e-06, "loss": 0.0056, "step": 8370 }, { "epoch": 0.13711854700155446, "grad_norm": 0.5855520963668823, "learning_rate": 3.427963675038861e-06, "loss": 0.0083, "step": 8380 }, { "epoch": 0.1372821729526303, "grad_norm": 0.5900159478187561, "learning_rate": 3.4320543238157573e-06, "loss": 0.0075, "step": 8390 }, { "epoch": 0.13744579890370612, "grad_norm": 0.306379497051239, "learning_rate": 3.4361449725926534e-06, "loss": 0.0083, "step": 8400 }, { "epoch": 0.13760942485478198, "grad_norm": 0.45768922567367554, "learning_rate": 3.440235621369549e-06, "loss": 0.0066, "step": 8410 }, { "epoch": 0.1377730508058578, "grad_norm": 0.2933987081050873, "learning_rate": 3.4443262701464452e-06, "loss": 0.0061, "step": 8420 }, { "epoch": 0.13793667675693366, "grad_norm": 0.4279216527938843, "learning_rate": 3.4484169189233414e-06, "loss": 0.0084, "step": 8430 }, { "epoch": 0.1381003027080095, "grad_norm": 0.2626897394657135, "learning_rate": 3.4525075677002375e-06, "loss": 0.0072, "step": 8440 }, { "epoch": 0.13826392865908532, "grad_norm": 0.42999786138534546, "learning_rate": 3.456598216477133e-06, "loss": 0.0074, "step": 8450 }, { "epoch": 0.13842755461016118, "grad_norm": 0.3273768126964569, "learning_rate": 3.4606888652540293e-06, "loss": 0.0069, "step": 8460 }, { "epoch": 0.138591180561237, "grad_norm": 0.06941913813352585, "learning_rate": 3.4647795140309255e-06, "loss": 0.0089, "step": 8470 }, { "epoch": 0.13875480651231284, "grad_norm": 0.3182835578918457, "learning_rate": 3.4688701628078216e-06, "loss": 0.0067, "step": 8480 }, { "epoch": 0.1389184324633887, "grad_norm": 0.09858960658311844, "learning_rate": 3.4729608115847173e-06, "loss": 0.0099, "step": 8490 }, { "epoch": 0.13908205841446453, "grad_norm": 0.34095507860183716, "learning_rate": 3.4770514603616134e-06, "loss": 0.0061, "step": 8500 }, { "epoch": 0.13924568436554038, "grad_norm": 0.47347864508628845, "learning_rate": 3.4811421091385096e-06, "loss": 0.0076, "step": 8510 }, { "epoch": 0.13940931031661621, "grad_norm": 0.4379952847957611, "learning_rate": 3.4852327579154057e-06, "loss": 0.0054, "step": 8520 }, { "epoch": 0.13957293626769204, "grad_norm": 0.14146015048027039, "learning_rate": 3.4893234066923014e-06, "loss": 0.009, "step": 8530 }, { "epoch": 0.1397365622187679, "grad_norm": 0.21573321521282196, "learning_rate": 3.4934140554691975e-06, "loss": 0.007, "step": 8540 }, { "epoch": 0.13990018816984373, "grad_norm": 0.23727820813655853, "learning_rate": 3.4975047042460937e-06, "loss": 0.006, "step": 8550 }, { "epoch": 0.1400638141209196, "grad_norm": 0.2250244915485382, "learning_rate": 3.50159535302299e-06, "loss": 0.0061, "step": 8560 }, { "epoch": 0.14022744007199542, "grad_norm": 0.15550285577774048, "learning_rate": 3.5056860017998855e-06, "loss": 0.0078, "step": 8570 }, { "epoch": 0.14039106602307125, "grad_norm": 0.7467495203018188, "learning_rate": 3.5097766505767816e-06, "loss": 0.0078, "step": 8580 }, { "epoch": 0.1405546919741471, "grad_norm": 0.48337849974632263, "learning_rate": 3.5138672993536778e-06, "loss": 0.0089, "step": 8590 }, { "epoch": 0.14071831792522294, "grad_norm": 0.2958015203475952, "learning_rate": 3.5179579481305735e-06, "loss": 0.009, "step": 8600 }, { "epoch": 0.1408819438762988, "grad_norm": 0.22649645805358887, "learning_rate": 3.5220485969074696e-06, "loss": 0.005, "step": 8610 }, { "epoch": 0.14104556982737462, "grad_norm": 0.2475706934928894, "learning_rate": 3.5261392456843657e-06, "loss": 0.0061, "step": 8620 }, { "epoch": 0.14120919577845045, "grad_norm": 0.37954050302505493, "learning_rate": 3.530229894461262e-06, "loss": 0.008, "step": 8630 }, { "epoch": 0.1413728217295263, "grad_norm": 1.0334120988845825, "learning_rate": 3.5343205432381576e-06, "loss": 0.01, "step": 8640 }, { "epoch": 0.14153644768060214, "grad_norm": 0.20481878519058228, "learning_rate": 3.5384111920150537e-06, "loss": 0.008, "step": 8650 }, { "epoch": 0.141700073631678, "grad_norm": 0.25706836581230164, "learning_rate": 3.54250184079195e-06, "loss": 0.0066, "step": 8660 }, { "epoch": 0.14186369958275383, "grad_norm": 0.47084149718284607, "learning_rate": 3.546592489568846e-06, "loss": 0.0099, "step": 8670 }, { "epoch": 0.14202732553382966, "grad_norm": 0.5283222198486328, "learning_rate": 3.5506831383457417e-06, "loss": 0.0078, "step": 8680 }, { "epoch": 0.14219095148490551, "grad_norm": 0.24488379061222076, "learning_rate": 3.554773787122638e-06, "loss": 0.0051, "step": 8690 }, { "epoch": 0.14235457743598134, "grad_norm": 0.24622181057929993, "learning_rate": 3.558864435899534e-06, "loss": 0.0079, "step": 8700 }, { "epoch": 0.14251820338705717, "grad_norm": 0.39647823572158813, "learning_rate": 3.56295508467643e-06, "loss": 0.0079, "step": 8710 }, { "epoch": 0.14268182933813303, "grad_norm": 0.7268316745758057, "learning_rate": 3.5670457334533258e-06, "loss": 0.0098, "step": 8720 }, { "epoch": 0.14284545528920886, "grad_norm": 0.10956976562738419, "learning_rate": 3.571136382230222e-06, "loss": 0.013, "step": 8730 }, { "epoch": 0.14300908124028472, "grad_norm": 0.4719236493110657, "learning_rate": 3.575227031007118e-06, "loss": 0.0075, "step": 8740 }, { "epoch": 0.14317270719136055, "grad_norm": 0.429869681596756, "learning_rate": 3.579317679784014e-06, "loss": 0.0092, "step": 8750 }, { "epoch": 0.14333633314243638, "grad_norm": 0.26199933886528015, "learning_rate": 3.58340832856091e-06, "loss": 0.0068, "step": 8760 }, { "epoch": 0.14349995909351224, "grad_norm": 0.7678824067115784, "learning_rate": 3.587498977337806e-06, "loss": 0.01, "step": 8770 }, { "epoch": 0.14366358504458807, "grad_norm": 0.25995609164237976, "learning_rate": 3.591589626114702e-06, "loss": 0.0059, "step": 8780 }, { "epoch": 0.14382721099566392, "grad_norm": 0.5889927744865417, "learning_rate": 3.5956802748915983e-06, "loss": 0.0061, "step": 8790 }, { "epoch": 0.14399083694673975, "grad_norm": 0.44174692034721375, "learning_rate": 3.599770923668494e-06, "loss": 0.0061, "step": 8800 }, { "epoch": 0.14415446289781558, "grad_norm": 0.32798492908477783, "learning_rate": 3.60386157244539e-06, "loss": 0.0047, "step": 8810 }, { "epoch": 0.14431808884889144, "grad_norm": 0.28921255469322205, "learning_rate": 3.6079522212222862e-06, "loss": 0.006, "step": 8820 }, { "epoch": 0.14448171479996727, "grad_norm": 0.4168439209461212, "learning_rate": 3.6120428699991824e-06, "loss": 0.0071, "step": 8830 }, { "epoch": 0.14464534075104313, "grad_norm": 0.07107597589492798, "learning_rate": 3.616133518776078e-06, "loss": 0.0045, "step": 8840 }, { "epoch": 0.14480896670211896, "grad_norm": 0.18367886543273926, "learning_rate": 3.6202241675529742e-06, "loss": 0.0073, "step": 8850 }, { "epoch": 0.1449725926531948, "grad_norm": 0.21223200857639313, "learning_rate": 3.6243148163298703e-06, "loss": 0.009, "step": 8860 }, { "epoch": 0.14513621860427064, "grad_norm": 0.20582692325115204, "learning_rate": 3.6284054651067665e-06, "loss": 0.006, "step": 8870 }, { "epoch": 0.14529984455534647, "grad_norm": 0.13987311720848083, "learning_rate": 3.632496113883662e-06, "loss": 0.0072, "step": 8880 }, { "epoch": 0.14546347050642233, "grad_norm": 0.23706193268299103, "learning_rate": 3.6365867626605583e-06, "loss": 0.0059, "step": 8890 }, { "epoch": 0.14562709645749816, "grad_norm": 0.49697041511535645, "learning_rate": 3.6406774114374544e-06, "loss": 0.0056, "step": 8900 }, { "epoch": 0.145790722408574, "grad_norm": 0.39989998936653137, "learning_rate": 3.6447680602143506e-06, "loss": 0.0055, "step": 8910 }, { "epoch": 0.14595434835964985, "grad_norm": 0.6293586492538452, "learning_rate": 3.6488587089912463e-06, "loss": 0.0083, "step": 8920 }, { "epoch": 0.14611797431072568, "grad_norm": 0.23673556745052338, "learning_rate": 3.6529493577681424e-06, "loss": 0.0078, "step": 8930 }, { "epoch": 0.1462816002618015, "grad_norm": 0.3978474736213684, "learning_rate": 3.6570400065450385e-06, "loss": 0.0056, "step": 8940 }, { "epoch": 0.14644522621287737, "grad_norm": 0.22527693212032318, "learning_rate": 3.6611306553219343e-06, "loss": 0.0055, "step": 8950 }, { "epoch": 0.1466088521639532, "grad_norm": 0.4276365339756012, "learning_rate": 3.6652213040988304e-06, "loss": 0.005, "step": 8960 }, { "epoch": 0.14677247811502905, "grad_norm": 0.04873589053750038, "learning_rate": 3.6693119528757265e-06, "loss": 0.0067, "step": 8970 }, { "epoch": 0.14693610406610488, "grad_norm": 0.7406699061393738, "learning_rate": 3.6734026016526226e-06, "loss": 0.0058, "step": 8980 }, { "epoch": 0.1470997300171807, "grad_norm": 0.468553751707077, "learning_rate": 3.6774932504295184e-06, "loss": 0.0059, "step": 8990 }, { "epoch": 0.14726335596825657, "grad_norm": 0.3678469955921173, "learning_rate": 3.6815838992064145e-06, "loss": 0.0054, "step": 9000 }, { "epoch": 0.1474269819193324, "grad_norm": 0.2844921946525574, "learning_rate": 3.6856745479833106e-06, "loss": 0.0071, "step": 9010 }, { "epoch": 0.14759060787040826, "grad_norm": 0.3633035123348236, "learning_rate": 3.6897651967602068e-06, "loss": 0.0061, "step": 9020 }, { "epoch": 0.1477542338214841, "grad_norm": 0.39038974046707153, "learning_rate": 3.6938558455371025e-06, "loss": 0.0061, "step": 9030 }, { "epoch": 0.14791785977255992, "grad_norm": 0.32494351267814636, "learning_rate": 3.6979464943139986e-06, "loss": 0.0071, "step": 9040 }, { "epoch": 0.14808148572363578, "grad_norm": 0.37955132126808167, "learning_rate": 3.7020371430908947e-06, "loss": 0.0072, "step": 9050 }, { "epoch": 0.1482451116747116, "grad_norm": 0.30948764085769653, "learning_rate": 3.706127791867791e-06, "loss": 0.0078, "step": 9060 }, { "epoch": 0.14840873762578746, "grad_norm": 0.24738672375679016, "learning_rate": 3.7102184406446866e-06, "loss": 0.0083, "step": 9070 }, { "epoch": 0.1485723635768633, "grad_norm": 0.5252488851547241, "learning_rate": 3.7143090894215827e-06, "loss": 0.0084, "step": 9080 }, { "epoch": 0.14873598952793912, "grad_norm": 0.31364017724990845, "learning_rate": 3.718399738198479e-06, "loss": 0.0054, "step": 9090 }, { "epoch": 0.14889961547901498, "grad_norm": 0.2760404348373413, "learning_rate": 3.722490386975375e-06, "loss": 0.0062, "step": 9100 }, { "epoch": 0.1490632414300908, "grad_norm": 0.3392021358013153, "learning_rate": 3.7265810357522707e-06, "loss": 0.0057, "step": 9110 }, { "epoch": 0.14922686738116667, "grad_norm": 0.18432380259037018, "learning_rate": 3.730671684529167e-06, "loss": 0.0102, "step": 9120 }, { "epoch": 0.1493904933322425, "grad_norm": 0.1979115903377533, "learning_rate": 3.734762333306063e-06, "loss": 0.0084, "step": 9130 }, { "epoch": 0.14955411928331833, "grad_norm": 0.14122186601161957, "learning_rate": 3.738852982082959e-06, "loss": 0.0054, "step": 9140 }, { "epoch": 0.14971774523439418, "grad_norm": 0.35552433133125305, "learning_rate": 3.7429436308598548e-06, "loss": 0.0064, "step": 9150 }, { "epoch": 0.14988137118547, "grad_norm": 0.5318500399589539, "learning_rate": 3.747034279636751e-06, "loss": 0.0085, "step": 9160 }, { "epoch": 0.15004499713654584, "grad_norm": 0.1631288230419159, "learning_rate": 3.751124928413647e-06, "loss": 0.0068, "step": 9170 }, { "epoch": 0.1502086230876217, "grad_norm": 0.765034556388855, "learning_rate": 3.755215577190543e-06, "loss": 0.0059, "step": 9180 }, { "epoch": 0.15037224903869753, "grad_norm": 0.6723112463951111, "learning_rate": 3.759306225967439e-06, "loss": 0.007, "step": 9190 }, { "epoch": 0.1505358749897734, "grad_norm": 0.3604050874710083, "learning_rate": 3.763396874744335e-06, "loss": 0.0073, "step": 9200 }, { "epoch": 0.15069950094084922, "grad_norm": 0.42386680841445923, "learning_rate": 3.767487523521231e-06, "loss": 0.0061, "step": 9210 }, { "epoch": 0.15086312689192505, "grad_norm": 0.09864157438278198, "learning_rate": 3.7715781722981273e-06, "loss": 0.0047, "step": 9220 }, { "epoch": 0.1510267528430009, "grad_norm": 0.41507938504219055, "learning_rate": 3.775668821075023e-06, "loss": 0.0082, "step": 9230 }, { "epoch": 0.15119037879407674, "grad_norm": 0.15359948575496674, "learning_rate": 3.7797594698519187e-06, "loss": 0.0057, "step": 9240 }, { "epoch": 0.1513540047451526, "grad_norm": 0.3864249885082245, "learning_rate": 3.7838501186288144e-06, "loss": 0.0042, "step": 9250 }, { "epoch": 0.15151763069622842, "grad_norm": 0.3802049458026886, "learning_rate": 3.7879407674057105e-06, "loss": 0.0056, "step": 9260 }, { "epoch": 0.15168125664730425, "grad_norm": 0.6256120204925537, "learning_rate": 3.7920314161826066e-06, "loss": 0.0051, "step": 9270 }, { "epoch": 0.1518448825983801, "grad_norm": 0.24934342503547668, "learning_rate": 3.7961220649595028e-06, "loss": 0.0046, "step": 9280 }, { "epoch": 0.15200850854945594, "grad_norm": 0.34026098251342773, "learning_rate": 3.8002127137363985e-06, "loss": 0.0056, "step": 9290 }, { "epoch": 0.1521721345005318, "grad_norm": 0.5085468292236328, "learning_rate": 3.8043033625132946e-06, "loss": 0.0056, "step": 9300 }, { "epoch": 0.15233576045160763, "grad_norm": 0.23638816177845, "learning_rate": 3.8083940112901908e-06, "loss": 0.0091, "step": 9310 }, { "epoch": 0.15249938640268346, "grad_norm": 0.3247488737106323, "learning_rate": 3.812484660067087e-06, "loss": 0.0085, "step": 9320 }, { "epoch": 0.15266301235375931, "grad_norm": 0.1689610481262207, "learning_rate": 3.816575308843983e-06, "loss": 0.0053, "step": 9330 }, { "epoch": 0.15282663830483514, "grad_norm": 0.570988655090332, "learning_rate": 3.820665957620879e-06, "loss": 0.0085, "step": 9340 }, { "epoch": 0.15299026425591097, "grad_norm": 0.5647450089454651, "learning_rate": 3.824756606397775e-06, "loss": 0.0121, "step": 9350 }, { "epoch": 0.15315389020698683, "grad_norm": 0.4744463860988617, "learning_rate": 3.8288472551746706e-06, "loss": 0.007, "step": 9360 }, { "epoch": 0.15331751615806266, "grad_norm": 0.5375319123268127, "learning_rate": 3.832937903951567e-06, "loss": 0.0084, "step": 9370 }, { "epoch": 0.15348114210913852, "grad_norm": 0.3785806894302368, "learning_rate": 3.837028552728463e-06, "loss": 0.0068, "step": 9380 }, { "epoch": 0.15364476806021435, "grad_norm": 0.5662693977355957, "learning_rate": 3.8411192015053585e-06, "loss": 0.0054, "step": 9390 }, { "epoch": 0.15380839401129018, "grad_norm": 0.361285001039505, "learning_rate": 3.845209850282255e-06, "loss": 0.0067, "step": 9400 }, { "epoch": 0.15397201996236604, "grad_norm": 0.5910766124725342, "learning_rate": 3.849300499059151e-06, "loss": 0.007, "step": 9410 }, { "epoch": 0.15413564591344187, "grad_norm": 0.48576393723487854, "learning_rate": 3.853391147836047e-06, "loss": 0.0052, "step": 9420 }, { "epoch": 0.15429927186451772, "grad_norm": 0.49207016825675964, "learning_rate": 3.857481796612943e-06, "loss": 0.0074, "step": 9430 }, { "epoch": 0.15446289781559355, "grad_norm": 0.10530605167150497, "learning_rate": 3.861572445389839e-06, "loss": 0.0088, "step": 9440 }, { "epoch": 0.15462652376666938, "grad_norm": 0.4658294916152954, "learning_rate": 3.865663094166735e-06, "loss": 0.0093, "step": 9450 }, { "epoch": 0.15479014971774524, "grad_norm": 0.567507266998291, "learning_rate": 3.869753742943631e-06, "loss": 0.0085, "step": 9460 }, { "epoch": 0.15495377566882107, "grad_norm": 0.35576754808425903, "learning_rate": 3.873844391720527e-06, "loss": 0.008, "step": 9470 }, { "epoch": 0.15511740161989693, "grad_norm": 0.36761316657066345, "learning_rate": 3.877935040497423e-06, "loss": 0.0087, "step": 9480 }, { "epoch": 0.15528102757097276, "grad_norm": 0.2626834511756897, "learning_rate": 3.882025689274319e-06, "loss": 0.0078, "step": 9490 }, { "epoch": 0.1554446535220486, "grad_norm": 0.3308922350406647, "learning_rate": 3.8861163380512156e-06, "loss": 0.0059, "step": 9500 }, { "epoch": 0.15560827947312444, "grad_norm": 0.31624552607536316, "learning_rate": 3.890206986828111e-06, "loss": 0.006, "step": 9510 }, { "epoch": 0.15577190542420027, "grad_norm": 0.19489829242229462, "learning_rate": 3.894297635605007e-06, "loss": 0.004, "step": 9520 }, { "epoch": 0.15593553137527613, "grad_norm": 0.21668119728565216, "learning_rate": 3.8983882843819035e-06, "loss": 0.0085, "step": 9530 }, { "epoch": 0.15609915732635196, "grad_norm": 0.16764439642429352, "learning_rate": 3.902478933158799e-06, "loss": 0.0061, "step": 9540 }, { "epoch": 0.1562627832774278, "grad_norm": 0.2651609778404236, "learning_rate": 3.906569581935695e-06, "loss": 0.0058, "step": 9550 }, { "epoch": 0.15642640922850365, "grad_norm": 0.43974924087524414, "learning_rate": 3.9106602307125915e-06, "loss": 0.0057, "step": 9560 }, { "epoch": 0.15659003517957948, "grad_norm": 1.2607057094573975, "learning_rate": 3.914750879489487e-06, "loss": 0.0087, "step": 9570 }, { "epoch": 0.1567536611306553, "grad_norm": 0.21439142525196075, "learning_rate": 3.918841528266383e-06, "loss": 0.0054, "step": 9580 }, { "epoch": 0.15691728708173117, "grad_norm": 0.9617077112197876, "learning_rate": 3.9229321770432795e-06, "loss": 0.008, "step": 9590 }, { "epoch": 0.157080913032807, "grad_norm": 0.3692317306995392, "learning_rate": 3.927022825820175e-06, "loss": 0.0098, "step": 9600 }, { "epoch": 0.15724453898388285, "grad_norm": 0.32081103324890137, "learning_rate": 3.931113474597072e-06, "loss": 0.0069, "step": 9610 }, { "epoch": 0.15740816493495868, "grad_norm": 0.24258305132389069, "learning_rate": 3.9352041233739674e-06, "loss": 0.0055, "step": 9620 }, { "epoch": 0.1575717908860345, "grad_norm": 0.10042114555835724, "learning_rate": 3.939294772150863e-06, "loss": 0.0043, "step": 9630 }, { "epoch": 0.15773541683711037, "grad_norm": 0.4679971933364868, "learning_rate": 3.94338542092776e-06, "loss": 0.0098, "step": 9640 }, { "epoch": 0.1578990427881862, "grad_norm": 0.3239520490169525, "learning_rate": 3.947476069704655e-06, "loss": 0.0073, "step": 9650 }, { "epoch": 0.15806266873926206, "grad_norm": 0.3165862262248993, "learning_rate": 3.951566718481551e-06, "loss": 0.007, "step": 9660 }, { "epoch": 0.1582262946903379, "grad_norm": 0.27481207251548767, "learning_rate": 3.955657367258448e-06, "loss": 0.0082, "step": 9670 }, { "epoch": 0.15838992064141372, "grad_norm": 0.15646636486053467, "learning_rate": 3.959748016035343e-06, "loss": 0.0076, "step": 9680 }, { "epoch": 0.15855354659248957, "grad_norm": 0.3996772766113281, "learning_rate": 3.96383866481224e-06, "loss": 0.0093, "step": 9690 }, { "epoch": 0.1587171725435654, "grad_norm": 0.3623597025871277, "learning_rate": 3.967929313589136e-06, "loss": 0.0065, "step": 9700 }, { "epoch": 0.15888079849464126, "grad_norm": 0.40755218267440796, "learning_rate": 3.972019962366031e-06, "loss": 0.0063, "step": 9710 }, { "epoch": 0.1590444244457171, "grad_norm": 0.20658060908317566, "learning_rate": 3.976110611142928e-06, "loss": 0.0053, "step": 9720 }, { "epoch": 0.15920805039679292, "grad_norm": 0.4787262976169586, "learning_rate": 3.980201259919824e-06, "loss": 0.0067, "step": 9730 }, { "epoch": 0.15937167634786878, "grad_norm": 0.2056826949119568, "learning_rate": 3.984291908696719e-06, "loss": 0.0043, "step": 9740 }, { "epoch": 0.1595353022989446, "grad_norm": 0.3455033004283905, "learning_rate": 3.988382557473616e-06, "loss": 0.0053, "step": 9750 }, { "epoch": 0.15969892825002047, "grad_norm": 0.11885394155979156, "learning_rate": 3.992473206250512e-06, "loss": 0.0053, "step": 9760 }, { "epoch": 0.1598625542010963, "grad_norm": 0.11894296109676361, "learning_rate": 3.996563855027408e-06, "loss": 0.0056, "step": 9770 }, { "epoch": 0.16002618015217213, "grad_norm": 0.15304909646511078, "learning_rate": 4.000654503804304e-06, "loss": 0.0061, "step": 9780 }, { "epoch": 0.16018980610324798, "grad_norm": 0.48488661646842957, "learning_rate": 4.0047451525811996e-06, "loss": 0.0077, "step": 9790 }, { "epoch": 0.1603534320543238, "grad_norm": 0.2977035343647003, "learning_rate": 4.008835801358096e-06, "loss": 0.0082, "step": 9800 }, { "epoch": 0.16051705800539964, "grad_norm": 0.2649109959602356, "learning_rate": 4.012926450134992e-06, "loss": 0.0072, "step": 9810 }, { "epoch": 0.1606806839564755, "grad_norm": 0.28319841623306274, "learning_rate": 4.0170170989118875e-06, "loss": 0.0059, "step": 9820 }, { "epoch": 0.16084430990755133, "grad_norm": 0.522306501865387, "learning_rate": 4.021107747688784e-06, "loss": 0.0077, "step": 9830 }, { "epoch": 0.1610079358586272, "grad_norm": 0.3319797217845917, "learning_rate": 4.02519839646568e-06, "loss": 0.0058, "step": 9840 }, { "epoch": 0.16117156180970302, "grad_norm": 0.29219916462898254, "learning_rate": 4.0292890452425755e-06, "loss": 0.0093, "step": 9850 }, { "epoch": 0.16133518776077885, "grad_norm": 0.21502402424812317, "learning_rate": 4.033379694019472e-06, "loss": 0.0081, "step": 9860 }, { "epoch": 0.1614988137118547, "grad_norm": 0.42148274183273315, "learning_rate": 4.037470342796368e-06, "loss": 0.0085, "step": 9870 }, { "epoch": 0.16166243966293053, "grad_norm": 0.21281130611896515, "learning_rate": 4.041560991573264e-06, "loss": 0.0075, "step": 9880 }, { "epoch": 0.1618260656140064, "grad_norm": 0.43234172463417053, "learning_rate": 4.04565164035016e-06, "loss": 0.0033, "step": 9890 }, { "epoch": 0.16198969156508222, "grad_norm": 0.3127169907093048, "learning_rate": 4.049742289127056e-06, "loss": 0.0076, "step": 9900 }, { "epoch": 0.16215331751615805, "grad_norm": 0.3405517339706421, "learning_rate": 4.053832937903952e-06, "loss": 0.0085, "step": 9910 }, { "epoch": 0.1623169434672339, "grad_norm": 0.11197216063737869, "learning_rate": 4.057923586680848e-06, "loss": 0.0059, "step": 9920 }, { "epoch": 0.16248056941830974, "grad_norm": 0.22729948163032532, "learning_rate": 4.062014235457744e-06, "loss": 0.0068, "step": 9930 }, { "epoch": 0.1626441953693856, "grad_norm": 0.18688653409481049, "learning_rate": 4.06610488423464e-06, "loss": 0.008, "step": 9940 }, { "epoch": 0.16280782132046143, "grad_norm": 0.22636111080646515, "learning_rate": 4.070195533011536e-06, "loss": 0.0076, "step": 9950 }, { "epoch": 0.16297144727153726, "grad_norm": 0.11180245876312256, "learning_rate": 4.0742861817884325e-06, "loss": 0.0036, "step": 9960 }, { "epoch": 0.16313507322261311, "grad_norm": 0.24625404179096222, "learning_rate": 4.078376830565328e-06, "loss": 0.0072, "step": 9970 }, { "epoch": 0.16329869917368894, "grad_norm": 0.4976617097854614, "learning_rate": 4.082467479342224e-06, "loss": 0.0061, "step": 9980 }, { "epoch": 0.1634623251247648, "grad_norm": 0.31449639797210693, "learning_rate": 4.0865581281191205e-06, "loss": 0.0073, "step": 9990 }, { "epoch": 0.16362595107584063, "grad_norm": 0.3601400554180145, "learning_rate": 4.090648776896016e-06, "loss": 0.0053, "step": 10000 }, { "epoch": 0.16378957702691646, "grad_norm": 0.32192179560661316, "learning_rate": 4.094739425672912e-06, "loss": 0.0084, "step": 10010 }, { "epoch": 0.16395320297799232, "grad_norm": 0.24399849772453308, "learning_rate": 4.0988300744498085e-06, "loss": 0.0069, "step": 10020 }, { "epoch": 0.16411682892906815, "grad_norm": 0.4611990451812744, "learning_rate": 4.102920723226704e-06, "loss": 0.0093, "step": 10030 }, { "epoch": 0.16428045488014398, "grad_norm": 0.36032283306121826, "learning_rate": 4.107011372003601e-06, "loss": 0.0066, "step": 10040 }, { "epoch": 0.16444408083121984, "grad_norm": 0.3563063144683838, "learning_rate": 4.1111020207804964e-06, "loss": 0.0059, "step": 10050 }, { "epoch": 0.16460770678229567, "grad_norm": 0.1926143765449524, "learning_rate": 4.115192669557392e-06, "loss": 0.0055, "step": 10060 }, { "epoch": 0.16477133273337152, "grad_norm": 0.34634092450141907, "learning_rate": 4.119283318334289e-06, "loss": 0.0072, "step": 10070 }, { "epoch": 0.16493495868444735, "grad_norm": 0.2805430591106415, "learning_rate": 4.123373967111184e-06, "loss": 0.0062, "step": 10080 }, { "epoch": 0.16509858463552318, "grad_norm": 0.4078565537929535, "learning_rate": 4.12746461588808e-06, "loss": 0.0064, "step": 10090 }, { "epoch": 0.16526221058659904, "grad_norm": 0.12962278723716736, "learning_rate": 4.131555264664977e-06, "loss": 0.006, "step": 10100 }, { "epoch": 0.16542583653767487, "grad_norm": 0.555002748966217, "learning_rate": 4.135645913441872e-06, "loss": 0.0089, "step": 10110 }, { "epoch": 0.16558946248875073, "grad_norm": 0.29776695370674133, "learning_rate": 4.139736562218768e-06, "loss": 0.0059, "step": 10120 }, { "epoch": 0.16575308843982656, "grad_norm": 0.3102855086326599, "learning_rate": 4.143827210995664e-06, "loss": 0.0068, "step": 10130 }, { "epoch": 0.1659167143909024, "grad_norm": 0.19820088148117065, "learning_rate": 4.14791785977256e-06, "loss": 0.0063, "step": 10140 }, { "epoch": 0.16608034034197824, "grad_norm": 0.24998965859413147, "learning_rate": 4.152008508549456e-06, "loss": 0.0078, "step": 10150 }, { "epoch": 0.16624396629305407, "grad_norm": 0.8987391591072083, "learning_rate": 4.156099157326352e-06, "loss": 0.0067, "step": 10160 }, { "epoch": 0.16640759224412993, "grad_norm": 0.4655391573905945, "learning_rate": 4.160189806103248e-06, "loss": 0.0063, "step": 10170 }, { "epoch": 0.16657121819520576, "grad_norm": 0.3683518171310425, "learning_rate": 4.164280454880144e-06, "loss": 0.0082, "step": 10180 }, { "epoch": 0.1667348441462816, "grad_norm": 0.1902153640985489, "learning_rate": 4.16837110365704e-06, "loss": 0.0039, "step": 10190 }, { "epoch": 0.16689847009735745, "grad_norm": 0.18356649577617645, "learning_rate": 4.172461752433936e-06, "loss": 0.0055, "step": 10200 }, { "epoch": 0.16706209604843328, "grad_norm": 0.2644200623035431, "learning_rate": 4.176552401210832e-06, "loss": 0.0066, "step": 10210 }, { "epoch": 0.16722572199950914, "grad_norm": 0.26600381731987, "learning_rate": 4.1806430499877285e-06, "loss": 0.005, "step": 10220 }, { "epoch": 0.16738934795058497, "grad_norm": 0.4145239591598511, "learning_rate": 4.184733698764624e-06, "loss": 0.0055, "step": 10230 }, { "epoch": 0.1675529739016608, "grad_norm": 0.607559323310852, "learning_rate": 4.18882434754152e-06, "loss": 0.0083, "step": 10240 }, { "epoch": 0.16771659985273665, "grad_norm": 0.21513056755065918, "learning_rate": 4.1929149963184165e-06, "loss": 0.0058, "step": 10250 }, { "epoch": 0.16788022580381248, "grad_norm": 0.318538099527359, "learning_rate": 4.197005645095312e-06, "loss": 0.0063, "step": 10260 }, { "epoch": 0.1680438517548883, "grad_norm": 0.2217639833688736, "learning_rate": 4.201096293872208e-06, "loss": 0.0094, "step": 10270 }, { "epoch": 0.16820747770596417, "grad_norm": 0.31767144799232483, "learning_rate": 4.2051869426491045e-06, "loss": 0.0089, "step": 10280 }, { "epoch": 0.16837110365704, "grad_norm": 0.3173513412475586, "learning_rate": 4.209277591426e-06, "loss": 0.0061, "step": 10290 }, { "epoch": 0.16853472960811586, "grad_norm": 0.1854225993156433, "learning_rate": 4.213368240202897e-06, "loss": 0.0083, "step": 10300 }, { "epoch": 0.1686983555591917, "grad_norm": 0.3832815885543823, "learning_rate": 4.2174588889797925e-06, "loss": 0.0069, "step": 10310 }, { "epoch": 0.16886198151026752, "grad_norm": 0.11185244470834732, "learning_rate": 4.221549537756688e-06, "loss": 0.0076, "step": 10320 }, { "epoch": 0.16902560746134337, "grad_norm": 0.16015222668647766, "learning_rate": 4.225640186533585e-06, "loss": 0.0087, "step": 10330 }, { "epoch": 0.1691892334124192, "grad_norm": 0.35454490780830383, "learning_rate": 4.2297308353104804e-06, "loss": 0.0058, "step": 10340 }, { "epoch": 0.16935285936349506, "grad_norm": 0.5603348016738892, "learning_rate": 4.233821484087376e-06, "loss": 0.0081, "step": 10350 }, { "epoch": 0.1695164853145709, "grad_norm": 0.4495898485183716, "learning_rate": 4.237912132864273e-06, "loss": 0.0072, "step": 10360 }, { "epoch": 0.16968011126564672, "grad_norm": 0.23535388708114624, "learning_rate": 4.242002781641168e-06, "loss": 0.0054, "step": 10370 }, { "epoch": 0.16984373721672258, "grad_norm": 0.12101545929908752, "learning_rate": 4.246093430418065e-06, "loss": 0.0052, "step": 10380 }, { "epoch": 0.1700073631677984, "grad_norm": 0.22193896770477295, "learning_rate": 4.250184079194961e-06, "loss": 0.0104, "step": 10390 }, { "epoch": 0.17017098911887427, "grad_norm": 0.25616130232810974, "learning_rate": 4.254274727971856e-06, "loss": 0.0054, "step": 10400 }, { "epoch": 0.1703346150699501, "grad_norm": 0.2591128945350647, "learning_rate": 4.258365376748753e-06, "loss": 0.0058, "step": 10410 }, { "epoch": 0.17049824102102593, "grad_norm": 0.1352667212486267, "learning_rate": 4.262456025525649e-06, "loss": 0.0049, "step": 10420 }, { "epoch": 0.17066186697210178, "grad_norm": 0.47396889328956604, "learning_rate": 4.266546674302544e-06, "loss": 0.0082, "step": 10430 }, { "epoch": 0.1708254929231776, "grad_norm": 0.3998880088329315, "learning_rate": 4.270637323079441e-06, "loss": 0.0049, "step": 10440 }, { "epoch": 0.17098911887425344, "grad_norm": 0.1897314190864563, "learning_rate": 4.274727971856337e-06, "loss": 0.0049, "step": 10450 }, { "epoch": 0.1711527448253293, "grad_norm": 0.1091860681772232, "learning_rate": 4.278818620633232e-06, "loss": 0.0078, "step": 10460 }, { "epoch": 0.17131637077640513, "grad_norm": 0.34856221079826355, "learning_rate": 4.282909269410129e-06, "loss": 0.0071, "step": 10470 }, { "epoch": 0.171479996727481, "grad_norm": 0.2046232372522354, "learning_rate": 4.2869999181870246e-06, "loss": 0.0089, "step": 10480 }, { "epoch": 0.17164362267855682, "grad_norm": 0.2474735826253891, "learning_rate": 4.291090566963921e-06, "loss": 0.0103, "step": 10490 }, { "epoch": 0.17180724862963265, "grad_norm": 0.292987585067749, "learning_rate": 4.295181215740817e-06, "loss": 0.0046, "step": 10500 }, { "epoch": 0.1719708745807085, "grad_norm": 0.23074445128440857, "learning_rate": 4.2992718645177125e-06, "loss": 0.0078, "step": 10510 }, { "epoch": 0.17213450053178433, "grad_norm": 0.5717023611068726, "learning_rate": 4.303362513294609e-06, "loss": 0.008, "step": 10520 }, { "epoch": 0.1722981264828602, "grad_norm": 0.4172489047050476, "learning_rate": 4.307453162071505e-06, "loss": 0.0059, "step": 10530 }, { "epoch": 0.17246175243393602, "grad_norm": 0.18148262798786163, "learning_rate": 4.3115438108484005e-06, "loss": 0.0081, "step": 10540 }, { "epoch": 0.17262537838501185, "grad_norm": 0.3314947187900543, "learning_rate": 4.315634459625297e-06, "loss": 0.0077, "step": 10550 }, { "epoch": 0.1727890043360877, "grad_norm": 0.21466121077537537, "learning_rate": 4.319725108402193e-06, "loss": 0.0077, "step": 10560 }, { "epoch": 0.17295263028716354, "grad_norm": 0.10223405808210373, "learning_rate": 4.323815757179089e-06, "loss": 0.0081, "step": 10570 }, { "epoch": 0.1731162562382394, "grad_norm": 0.1599915325641632, "learning_rate": 4.327906405955985e-06, "loss": 0.0052, "step": 10580 }, { "epoch": 0.17327988218931523, "grad_norm": 0.11431307345628738, "learning_rate": 4.331997054732881e-06, "loss": 0.0066, "step": 10590 }, { "epoch": 0.17344350814039106, "grad_norm": 0.25003182888031006, "learning_rate": 4.336087703509777e-06, "loss": 0.0076, "step": 10600 }, { "epoch": 0.1736071340914669, "grad_norm": 0.2527673542499542, "learning_rate": 4.340178352286673e-06, "loss": 0.0041, "step": 10610 }, { "epoch": 0.17377076004254274, "grad_norm": 0.2978326678276062, "learning_rate": 4.344269001063569e-06, "loss": 0.0064, "step": 10620 }, { "epoch": 0.1739343859936186, "grad_norm": 0.318464994430542, "learning_rate": 4.348359649840465e-06, "loss": 0.0048, "step": 10630 }, { "epoch": 0.17409801194469443, "grad_norm": 0.2701649069786072, "learning_rate": 4.352450298617361e-06, "loss": 0.0059, "step": 10640 }, { "epoch": 0.17426163789577026, "grad_norm": 0.4103914797306061, "learning_rate": 4.3565409473942575e-06, "loss": 0.0047, "step": 10650 }, { "epoch": 0.17442526384684612, "grad_norm": 0.4058134853839874, "learning_rate": 4.360631596171153e-06, "loss": 0.0049, "step": 10660 }, { "epoch": 0.17458888979792195, "grad_norm": 0.24684330821037292, "learning_rate": 4.364722244948049e-06, "loss": 0.005, "step": 10670 }, { "epoch": 0.17475251574899778, "grad_norm": 0.37260669469833374, "learning_rate": 4.3688128937249455e-06, "loss": 0.0091, "step": 10680 }, { "epoch": 0.17491614170007364, "grad_norm": 0.37129220366477966, "learning_rate": 4.372903542501841e-06, "loss": 0.004, "step": 10690 }, { "epoch": 0.17507976765114946, "grad_norm": 0.217575341463089, "learning_rate": 4.376994191278737e-06, "loss": 0.0059, "step": 10700 }, { "epoch": 0.17524339360222532, "grad_norm": 0.29052358865737915, "learning_rate": 4.3810848400556335e-06, "loss": 0.0048, "step": 10710 }, { "epoch": 0.17540701955330115, "grad_norm": 0.38673990964889526, "learning_rate": 4.385175488832529e-06, "loss": 0.0068, "step": 10720 }, { "epoch": 0.17557064550437698, "grad_norm": 0.1569669097661972, "learning_rate": 4.389266137609426e-06, "loss": 0.0062, "step": 10730 }, { "epoch": 0.17573427145545284, "grad_norm": 0.21959932148456573, "learning_rate": 4.3933567863863214e-06, "loss": 0.0057, "step": 10740 }, { "epoch": 0.17589789740652867, "grad_norm": 0.23419082164764404, "learning_rate": 4.397447435163217e-06, "loss": 0.0033, "step": 10750 }, { "epoch": 0.17606152335760453, "grad_norm": 0.5058455467224121, "learning_rate": 4.401538083940114e-06, "loss": 0.0072, "step": 10760 }, { "epoch": 0.17622514930868036, "grad_norm": 0.5645384192466736, "learning_rate": 4.405628732717009e-06, "loss": 0.0058, "step": 10770 }, { "epoch": 0.1763887752597562, "grad_norm": 0.2857092320919037, "learning_rate": 4.409719381493905e-06, "loss": 0.005, "step": 10780 }, { "epoch": 0.17655240121083204, "grad_norm": 0.3288016617298126, "learning_rate": 4.413810030270802e-06, "loss": 0.0049, "step": 10790 }, { "epoch": 0.17671602716190787, "grad_norm": 0.040123678743839264, "learning_rate": 4.417900679047697e-06, "loss": 0.0045, "step": 10800 }, { "epoch": 0.17687965311298373, "grad_norm": 0.2644427716732025, "learning_rate": 4.421991327824593e-06, "loss": 0.0055, "step": 10810 }, { "epoch": 0.17704327906405956, "grad_norm": 0.11512814462184906, "learning_rate": 4.42608197660149e-06, "loss": 0.008, "step": 10820 }, { "epoch": 0.1772069050151354, "grad_norm": 0.1733572781085968, "learning_rate": 4.430172625378385e-06, "loss": 0.0052, "step": 10830 }, { "epoch": 0.17737053096621125, "grad_norm": 0.25138598680496216, "learning_rate": 4.434263274155282e-06, "loss": 0.0037, "step": 10840 }, { "epoch": 0.17753415691728708, "grad_norm": 0.10439487546682358, "learning_rate": 4.438353922932178e-06, "loss": 0.0073, "step": 10850 }, { "epoch": 0.17769778286836294, "grad_norm": 0.3217577040195465, "learning_rate": 4.442444571709073e-06, "loss": 0.006, "step": 10860 }, { "epoch": 0.17786140881943877, "grad_norm": 0.45750030875205994, "learning_rate": 4.44653522048597e-06, "loss": 0.0069, "step": 10870 }, { "epoch": 0.1780250347705146, "grad_norm": 0.4229678809642792, "learning_rate": 4.450625869262866e-06, "loss": 0.0048, "step": 10880 }, { "epoch": 0.17818866072159045, "grad_norm": 0.3212835192680359, "learning_rate": 4.454716518039761e-06, "loss": 0.0052, "step": 10890 }, { "epoch": 0.17835228667266628, "grad_norm": 0.2603631615638733, "learning_rate": 4.458807166816658e-06, "loss": 0.006, "step": 10900 }, { "epoch": 0.1785159126237421, "grad_norm": 0.44148221611976624, "learning_rate": 4.4628978155935536e-06, "loss": 0.0058, "step": 10910 }, { "epoch": 0.17867953857481797, "grad_norm": 0.24088291823863983, "learning_rate": 4.46698846437045e-06, "loss": 0.0053, "step": 10920 }, { "epoch": 0.1788431645258938, "grad_norm": 0.05094173178076744, "learning_rate": 4.471079113147346e-06, "loss": 0.0075, "step": 10930 }, { "epoch": 0.17900679047696966, "grad_norm": 0.2019433230161667, "learning_rate": 4.4751697619242415e-06, "loss": 0.0051, "step": 10940 }, { "epoch": 0.1791704164280455, "grad_norm": 0.16214290261268616, "learning_rate": 4.479260410701138e-06, "loss": 0.0071, "step": 10950 }, { "epoch": 0.17933404237912132, "grad_norm": 0.3339186906814575, "learning_rate": 4.483351059478034e-06, "loss": 0.0065, "step": 10960 }, { "epoch": 0.17949766833019717, "grad_norm": 0.30227893590927124, "learning_rate": 4.4874417082549295e-06, "loss": 0.0068, "step": 10970 }, { "epoch": 0.179661294281273, "grad_norm": 0.2975096106529236, "learning_rate": 4.491532357031826e-06, "loss": 0.0052, "step": 10980 }, { "epoch": 0.17982492023234886, "grad_norm": 0.2426649034023285, "learning_rate": 4.495623005808722e-06, "loss": 0.0067, "step": 10990 }, { "epoch": 0.1799885461834247, "grad_norm": 0.2166430950164795, "learning_rate": 4.4997136545856175e-06, "loss": 0.0076, "step": 11000 }, { "epoch": 0.18015217213450052, "grad_norm": 0.2573049068450928, "learning_rate": 4.503804303362513e-06, "loss": 0.0057, "step": 11010 }, { "epoch": 0.18031579808557638, "grad_norm": 0.274663507938385, "learning_rate": 4.50789495213941e-06, "loss": 0.0078, "step": 11020 }, { "epoch": 0.1804794240366522, "grad_norm": 0.13890759646892548, "learning_rate": 4.5119856009163054e-06, "loss": 0.009, "step": 11030 }, { "epoch": 0.18064304998772807, "grad_norm": 0.4240662157535553, "learning_rate": 4.516076249693201e-06, "loss": 0.0071, "step": 11040 }, { "epoch": 0.1808066759388039, "grad_norm": 0.24942529201507568, "learning_rate": 4.520166898470098e-06, "loss": 0.005, "step": 11050 }, { "epoch": 0.18097030188987973, "grad_norm": 0.4940030574798584, "learning_rate": 4.524257547246993e-06, "loss": 0.0077, "step": 11060 }, { "epoch": 0.18113392784095558, "grad_norm": 0.07209077477455139, "learning_rate": 4.528348196023889e-06, "loss": 0.0077, "step": 11070 }, { "epoch": 0.1812975537920314, "grad_norm": 0.4373304843902588, "learning_rate": 4.532438844800786e-06, "loss": 0.0093, "step": 11080 }, { "epoch": 0.18146117974310727, "grad_norm": 0.3054561913013458, "learning_rate": 4.536529493577681e-06, "loss": 0.0068, "step": 11090 }, { "epoch": 0.1816248056941831, "grad_norm": 0.1635497510433197, "learning_rate": 4.540620142354578e-06, "loss": 0.0039, "step": 11100 }, { "epoch": 0.18178843164525893, "grad_norm": 0.3546565771102905, "learning_rate": 4.544710791131474e-06, "loss": 0.0067, "step": 11110 }, { "epoch": 0.1819520575963348, "grad_norm": 0.46289438009262085, "learning_rate": 4.548801439908369e-06, "loss": 0.0055, "step": 11120 }, { "epoch": 0.18211568354741062, "grad_norm": 0.20432139933109283, "learning_rate": 4.552892088685266e-06, "loss": 0.0038, "step": 11130 }, { "epoch": 0.18227930949848645, "grad_norm": 0.295704185962677, "learning_rate": 4.556982737462162e-06, "loss": 0.0052, "step": 11140 }, { "epoch": 0.1824429354495623, "grad_norm": 0.6937462687492371, "learning_rate": 4.561073386239057e-06, "loss": 0.0049, "step": 11150 }, { "epoch": 0.18260656140063813, "grad_norm": 0.12342402338981628, "learning_rate": 4.565164035015954e-06, "loss": 0.0056, "step": 11160 }, { "epoch": 0.182770187351714, "grad_norm": 0.1431237906217575, "learning_rate": 4.56925468379285e-06, "loss": 0.0055, "step": 11170 }, { "epoch": 0.18293381330278982, "grad_norm": 0.20269453525543213, "learning_rate": 4.573345332569746e-06, "loss": 0.005, "step": 11180 }, { "epoch": 0.18309743925386565, "grad_norm": 0.1598280966281891, "learning_rate": 4.577435981346642e-06, "loss": 0.0074, "step": 11190 }, { "epoch": 0.1832610652049415, "grad_norm": 0.32835787534713745, "learning_rate": 4.5815266301235376e-06, "loss": 0.0035, "step": 11200 }, { "epoch": 0.18342469115601734, "grad_norm": 0.32326799631118774, "learning_rate": 4.585617278900434e-06, "loss": 0.0071, "step": 11210 }, { "epoch": 0.1835883171070932, "grad_norm": 0.40954193472862244, "learning_rate": 4.58970792767733e-06, "loss": 0.0054, "step": 11220 }, { "epoch": 0.18375194305816903, "grad_norm": 0.30036476254463196, "learning_rate": 4.5937985764542255e-06, "loss": 0.0044, "step": 11230 }, { "epoch": 0.18391556900924486, "grad_norm": 0.3911738693714142, "learning_rate": 4.597889225231122e-06, "loss": 0.0064, "step": 11240 }, { "epoch": 0.1840791949603207, "grad_norm": 0.20620127022266388, "learning_rate": 4.601979874008018e-06, "loss": 0.006, "step": 11250 }, { "epoch": 0.18424282091139654, "grad_norm": 0.21057777106761932, "learning_rate": 4.606070522784914e-06, "loss": 0.0055, "step": 11260 }, { "epoch": 0.1844064468624724, "grad_norm": 0.41327470541000366, "learning_rate": 4.61016117156181e-06, "loss": 0.0066, "step": 11270 }, { "epoch": 0.18457007281354823, "grad_norm": 0.14875183999538422, "learning_rate": 4.614251820338706e-06, "loss": 0.0052, "step": 11280 }, { "epoch": 0.18473369876462406, "grad_norm": 0.19140979647636414, "learning_rate": 4.618342469115602e-06, "loss": 0.0061, "step": 11290 }, { "epoch": 0.18489732471569992, "grad_norm": 0.24508711695671082, "learning_rate": 4.622433117892498e-06, "loss": 0.0059, "step": 11300 }, { "epoch": 0.18506095066677575, "grad_norm": 0.1613132804632187, "learning_rate": 4.626523766669394e-06, "loss": 0.0094, "step": 11310 }, { "epoch": 0.1852245766178516, "grad_norm": 0.17226490378379822, "learning_rate": 4.63061441544629e-06, "loss": 0.0043, "step": 11320 }, { "epoch": 0.18538820256892743, "grad_norm": 0.09170756489038467, "learning_rate": 4.634705064223186e-06, "loss": 0.0058, "step": 11330 }, { "epoch": 0.18555182852000326, "grad_norm": 0.11842875927686691, "learning_rate": 4.6387957130000826e-06, "loss": 0.0066, "step": 11340 }, { "epoch": 0.18571545447107912, "grad_norm": 0.253249853849411, "learning_rate": 4.642886361776978e-06, "loss": 0.0046, "step": 11350 }, { "epoch": 0.18587908042215495, "grad_norm": 0.21617116034030914, "learning_rate": 4.646977010553874e-06, "loss": 0.0072, "step": 11360 }, { "epoch": 0.18604270637323078, "grad_norm": 0.19860310852527618, "learning_rate": 4.6510676593307705e-06, "loss": 0.0085, "step": 11370 }, { "epoch": 0.18620633232430664, "grad_norm": 0.7381302714347839, "learning_rate": 4.655158308107666e-06, "loss": 0.0092, "step": 11380 }, { "epoch": 0.18636995827538247, "grad_norm": 0.1908685266971588, "learning_rate": 4.659248956884562e-06, "loss": 0.0087, "step": 11390 }, { "epoch": 0.18653358422645833, "grad_norm": 0.33350157737731934, "learning_rate": 4.6633396056614585e-06, "loss": 0.0069, "step": 11400 }, { "epoch": 0.18669721017753416, "grad_norm": 0.44991880655288696, "learning_rate": 4.667430254438354e-06, "loss": 0.0061, "step": 11410 }, { "epoch": 0.18686083612861, "grad_norm": 0.6459864974021912, "learning_rate": 4.67152090321525e-06, "loss": 0.0078, "step": 11420 }, { "epoch": 0.18702446207968584, "grad_norm": 0.4520348608493805, "learning_rate": 4.6756115519921465e-06, "loss": 0.0051, "step": 11430 }, { "epoch": 0.18718808803076167, "grad_norm": 0.37368884682655334, "learning_rate": 4.679702200769042e-06, "loss": 0.0058, "step": 11440 }, { "epoch": 0.18735171398183753, "grad_norm": 0.6532014608383179, "learning_rate": 4.683792849545939e-06, "loss": 0.0099, "step": 11450 }, { "epoch": 0.18751533993291336, "grad_norm": 0.22395190596580505, "learning_rate": 4.6878834983228344e-06, "loss": 0.0058, "step": 11460 }, { "epoch": 0.1876789658839892, "grad_norm": 0.16034214198589325, "learning_rate": 4.69197414709973e-06, "loss": 0.0063, "step": 11470 }, { "epoch": 0.18784259183506505, "grad_norm": 0.15379777550697327, "learning_rate": 4.696064795876627e-06, "loss": 0.0056, "step": 11480 }, { "epoch": 0.18800621778614088, "grad_norm": 0.3850826919078827, "learning_rate": 4.700155444653522e-06, "loss": 0.0055, "step": 11490 }, { "epoch": 0.18816984373721674, "grad_norm": 0.16058777272701263, "learning_rate": 4.704246093430418e-06, "loss": 0.005, "step": 11500 }, { "epoch": 0.18833346968829257, "grad_norm": 0.43356195092201233, "learning_rate": 4.708336742207315e-06, "loss": 0.0046, "step": 11510 }, { "epoch": 0.1884970956393684, "grad_norm": 0.3228779137134552, "learning_rate": 4.71242739098421e-06, "loss": 0.0089, "step": 11520 }, { "epoch": 0.18866072159044425, "grad_norm": 0.2570642828941345, "learning_rate": 4.716518039761107e-06, "loss": 0.008, "step": 11530 }, { "epoch": 0.18882434754152008, "grad_norm": 0.13985690474510193, "learning_rate": 4.720608688538003e-06, "loss": 0.0054, "step": 11540 }, { "epoch": 0.1889879734925959, "grad_norm": 0.21010899543762207, "learning_rate": 4.724699337314898e-06, "loss": 0.0067, "step": 11550 }, { "epoch": 0.18915159944367177, "grad_norm": 0.18383213877677917, "learning_rate": 4.728789986091795e-06, "loss": 0.0063, "step": 11560 }, { "epoch": 0.1893152253947476, "grad_norm": 0.398485004901886, "learning_rate": 4.732880634868691e-06, "loss": 0.0077, "step": 11570 }, { "epoch": 0.18947885134582346, "grad_norm": 0.19336490333080292, "learning_rate": 4.736971283645586e-06, "loss": 0.0057, "step": 11580 }, { "epoch": 0.1896424772968993, "grad_norm": 0.06239166855812073, "learning_rate": 4.741061932422483e-06, "loss": 0.0043, "step": 11590 }, { "epoch": 0.18980610324797512, "grad_norm": 0.11733987182378769, "learning_rate": 4.745152581199379e-06, "loss": 0.0076, "step": 11600 }, { "epoch": 0.18996972919905097, "grad_norm": 0.35899195075035095, "learning_rate": 4.749243229976275e-06, "loss": 0.0041, "step": 11610 }, { "epoch": 0.1901333551501268, "grad_norm": 0.32030799984931946, "learning_rate": 4.753333878753171e-06, "loss": 0.0076, "step": 11620 }, { "epoch": 0.19029698110120266, "grad_norm": 0.17217855155467987, "learning_rate": 4.7574245275300666e-06, "loss": 0.0046, "step": 11630 }, { "epoch": 0.1904606070522785, "grad_norm": 0.36323297023773193, "learning_rate": 4.761515176306963e-06, "loss": 0.0066, "step": 11640 }, { "epoch": 0.19062423300335432, "grad_norm": 0.35365089774131775, "learning_rate": 4.765605825083859e-06, "loss": 0.0045, "step": 11650 }, { "epoch": 0.19078785895443018, "grad_norm": 0.550478994846344, "learning_rate": 4.7696964738607545e-06, "loss": 0.0051, "step": 11660 }, { "epoch": 0.190951484905506, "grad_norm": 0.3063075542449951, "learning_rate": 4.773787122637651e-06, "loss": 0.0051, "step": 11670 }, { "epoch": 0.19111511085658187, "grad_norm": 0.0457979291677475, "learning_rate": 4.777877771414547e-06, "loss": 0.005, "step": 11680 }, { "epoch": 0.1912787368076577, "grad_norm": 0.3218732476234436, "learning_rate": 4.781968420191443e-06, "loss": 0.0073, "step": 11690 }, { "epoch": 0.19144236275873353, "grad_norm": 0.15582577884197235, "learning_rate": 4.786059068968339e-06, "loss": 0.0039, "step": 11700 }, { "epoch": 0.19160598870980938, "grad_norm": 0.28192299604415894, "learning_rate": 4.790149717745235e-06, "loss": 0.0069, "step": 11710 }, { "epoch": 0.1917696146608852, "grad_norm": 0.3666723966598511, "learning_rate": 4.794240366522131e-06, "loss": 0.0062, "step": 11720 }, { "epoch": 0.19193324061196107, "grad_norm": 0.21619901061058044, "learning_rate": 4.798331015299027e-06, "loss": 0.0053, "step": 11730 }, { "epoch": 0.1920968665630369, "grad_norm": 0.42905810475349426, "learning_rate": 4.802421664075923e-06, "loss": 0.0034, "step": 11740 }, { "epoch": 0.19226049251411273, "grad_norm": 0.04460269212722778, "learning_rate": 4.806512312852819e-06, "loss": 0.0059, "step": 11750 }, { "epoch": 0.1924241184651886, "grad_norm": 0.19645920395851135, "learning_rate": 4.810602961629715e-06, "loss": 0.0045, "step": 11760 }, { "epoch": 0.19258774441626442, "grad_norm": 0.266537606716156, "learning_rate": 4.814693610406611e-06, "loss": 0.0065, "step": 11770 }, { "epoch": 0.19275137036734025, "grad_norm": 0.31572315096855164, "learning_rate": 4.818784259183507e-06, "loss": 0.0045, "step": 11780 }, { "epoch": 0.1929149963184161, "grad_norm": 0.14216381311416626, "learning_rate": 4.822874907960403e-06, "loss": 0.0045, "step": 11790 }, { "epoch": 0.19307862226949193, "grad_norm": 0.9160482883453369, "learning_rate": 4.8269655567372995e-06, "loss": 0.0057, "step": 11800 }, { "epoch": 0.1932422482205678, "grad_norm": 0.1419849395751953, "learning_rate": 4.831056205514195e-06, "loss": 0.0041, "step": 11810 }, { "epoch": 0.19340587417164362, "grad_norm": 0.39523741602897644, "learning_rate": 4.835146854291091e-06, "loss": 0.0074, "step": 11820 }, { "epoch": 0.19356950012271945, "grad_norm": 0.15790234506130219, "learning_rate": 4.8392375030679875e-06, "loss": 0.0054, "step": 11830 }, { "epoch": 0.1937331260737953, "grad_norm": 0.26259472966194153, "learning_rate": 4.843328151844883e-06, "loss": 0.0042, "step": 11840 }, { "epoch": 0.19389675202487114, "grad_norm": 0.373803049325943, "learning_rate": 4.847418800621779e-06, "loss": 0.0052, "step": 11850 }, { "epoch": 0.194060377975947, "grad_norm": 0.19831988215446472, "learning_rate": 4.8515094493986755e-06, "loss": 0.0069, "step": 11860 }, { "epoch": 0.19422400392702283, "grad_norm": 0.2537267506122589, "learning_rate": 4.855600098175571e-06, "loss": 0.0075, "step": 11870 }, { "epoch": 0.19438762987809866, "grad_norm": 0.400096595287323, "learning_rate": 4.859690746952467e-06, "loss": 0.0075, "step": 11880 }, { "epoch": 0.1945512558291745, "grad_norm": 0.3654206693172455, "learning_rate": 4.863781395729363e-06, "loss": 0.0052, "step": 11890 }, { "epoch": 0.19471488178025034, "grad_norm": 0.48343920707702637, "learning_rate": 4.867872044506259e-06, "loss": 0.0076, "step": 11900 }, { "epoch": 0.1948785077313262, "grad_norm": 0.3201134204864502, "learning_rate": 4.871962693283155e-06, "loss": 0.0041, "step": 11910 }, { "epoch": 0.19504213368240203, "grad_norm": 0.13583214581012726, "learning_rate": 4.8760533420600506e-06, "loss": 0.0053, "step": 11920 }, { "epoch": 0.19520575963347786, "grad_norm": 0.1420280486345291, "learning_rate": 4.880143990836947e-06, "loss": 0.0047, "step": 11930 }, { "epoch": 0.19536938558455372, "grad_norm": 0.21825622022151947, "learning_rate": 4.884234639613843e-06, "loss": 0.0052, "step": 11940 }, { "epoch": 0.19553301153562955, "grad_norm": 0.11790311336517334, "learning_rate": 4.8883252883907385e-06, "loss": 0.0062, "step": 11950 }, { "epoch": 0.1956966374867054, "grad_norm": 0.2594107389450073, "learning_rate": 4.892415937167635e-06, "loss": 0.0047, "step": 11960 }, { "epoch": 0.19586026343778123, "grad_norm": 0.30148184299468994, "learning_rate": 4.896506585944531e-06, "loss": 0.0059, "step": 11970 }, { "epoch": 0.19602388938885706, "grad_norm": 0.31892696022987366, "learning_rate": 4.900597234721427e-06, "loss": 0.006, "step": 11980 }, { "epoch": 0.19618751533993292, "grad_norm": 0.3130878210067749, "learning_rate": 4.904687883498323e-06, "loss": 0.0102, "step": 11990 }, { "epoch": 0.19635114129100875, "grad_norm": 0.23477788269519806, "learning_rate": 4.908778532275219e-06, "loss": 0.0064, "step": 12000 }, { "epoch": 0.19651476724208458, "grad_norm": 0.1322139948606491, "learning_rate": 4.912869181052115e-06, "loss": 0.0071, "step": 12010 }, { "epoch": 0.19667839319316044, "grad_norm": 0.23168422281742096, "learning_rate": 4.916959829829011e-06, "loss": 0.0074, "step": 12020 }, { "epoch": 0.19684201914423627, "grad_norm": 0.2383849173784256, "learning_rate": 4.921050478605907e-06, "loss": 0.0067, "step": 12030 }, { "epoch": 0.19700564509531213, "grad_norm": 0.047537025064229965, "learning_rate": 4.925141127382803e-06, "loss": 0.007, "step": 12040 }, { "epoch": 0.19716927104638796, "grad_norm": 0.2416282743215561, "learning_rate": 4.929231776159699e-06, "loss": 0.0066, "step": 12050 }, { "epoch": 0.19733289699746379, "grad_norm": 0.36442631483078003, "learning_rate": 4.9333224249365955e-06, "loss": 0.0101, "step": 12060 }, { "epoch": 0.19749652294853964, "grad_norm": 0.22719527781009674, "learning_rate": 4.937413073713491e-06, "loss": 0.0056, "step": 12070 }, { "epoch": 0.19766014889961547, "grad_norm": 0.2919733226299286, "learning_rate": 4.941503722490387e-06, "loss": 0.0056, "step": 12080 }, { "epoch": 0.19782377485069133, "grad_norm": 0.5045968294143677, "learning_rate": 4.9455943712672835e-06, "loss": 0.0069, "step": 12090 }, { "epoch": 0.19798740080176716, "grad_norm": 0.4553527534008026, "learning_rate": 4.949685020044179e-06, "loss": 0.0077, "step": 12100 }, { "epoch": 0.198151026752843, "grad_norm": 0.25020548701286316, "learning_rate": 4.953775668821075e-06, "loss": 0.0054, "step": 12110 }, { "epoch": 0.19831465270391885, "grad_norm": 0.385056734085083, "learning_rate": 4.9578663175979715e-06, "loss": 0.0052, "step": 12120 }, { "epoch": 0.19847827865499468, "grad_norm": 0.08680994808673859, "learning_rate": 4.961956966374867e-06, "loss": 0.0039, "step": 12130 }, { "epoch": 0.19864190460607054, "grad_norm": 0.3920186161994934, "learning_rate": 4.966047615151764e-06, "loss": 0.0061, "step": 12140 }, { "epoch": 0.19880553055714636, "grad_norm": 0.4898313581943512, "learning_rate": 4.9701382639286595e-06, "loss": 0.0079, "step": 12150 }, { "epoch": 0.1989691565082222, "grad_norm": 0.14415888488292694, "learning_rate": 4.974228912705555e-06, "loss": 0.0044, "step": 12160 }, { "epoch": 0.19913278245929805, "grad_norm": 0.21137873828411102, "learning_rate": 4.978319561482452e-06, "loss": 0.01, "step": 12170 }, { "epoch": 0.19929640841037388, "grad_norm": 0.271688848733902, "learning_rate": 4.9824102102593474e-06, "loss": 0.007, "step": 12180 }, { "epoch": 0.19946003436144974, "grad_norm": 0.18128825724124908, "learning_rate": 4.986500859036243e-06, "loss": 0.0045, "step": 12190 }, { "epoch": 0.19962366031252557, "grad_norm": 0.3777793347835541, "learning_rate": 4.99059150781314e-06, "loss": 0.0059, "step": 12200 }, { "epoch": 0.1997872862636014, "grad_norm": 0.19831782579421997, "learning_rate": 4.994682156590035e-06, "loss": 0.004, "step": 12210 }, { "epoch": 0.19995091221467726, "grad_norm": 0.4344349205493927, "learning_rate": 4.998772805366932e-06, "loss": 0.0069, "step": 12220 }, { "epoch": 0.2001145381657531, "grad_norm": 0.2862706780433655, "learning_rate": 5.0028634541438285e-06, "loss": 0.007, "step": 12230 }, { "epoch": 0.20027816411682892, "grad_norm": 0.23682764172554016, "learning_rate": 5.006954102920723e-06, "loss": 0.0069, "step": 12240 }, { "epoch": 0.20044179006790477, "grad_norm": 0.5163635611534119, "learning_rate": 5.01104475169762e-06, "loss": 0.0069, "step": 12250 }, { "epoch": 0.2006054160189806, "grad_norm": 0.2805919051170349, "learning_rate": 5.015135400474516e-06, "loss": 0.0041, "step": 12260 }, { "epoch": 0.20076904197005646, "grad_norm": 0.1742028146982193, "learning_rate": 5.019226049251412e-06, "loss": 0.0038, "step": 12270 }, { "epoch": 0.2009326679211323, "grad_norm": 0.31259241700172424, "learning_rate": 5.023316698028308e-06, "loss": 0.005, "step": 12280 }, { "epoch": 0.20109629387220812, "grad_norm": 0.3883764445781708, "learning_rate": 5.0274073468052044e-06, "loss": 0.0076, "step": 12290 }, { "epoch": 0.20125991982328398, "grad_norm": 0.19468890130519867, "learning_rate": 5.031497995582099e-06, "loss": 0.0051, "step": 12300 }, { "epoch": 0.2014235457743598, "grad_norm": 0.6576398015022278, "learning_rate": 5.035588644358997e-06, "loss": 0.0064, "step": 12310 }, { "epoch": 0.20158717172543567, "grad_norm": 0.2135811746120453, "learning_rate": 5.0396792931358916e-06, "loss": 0.0056, "step": 12320 }, { "epoch": 0.2017507976765115, "grad_norm": 0.4390604794025421, "learning_rate": 5.043769941912787e-06, "loss": 0.0072, "step": 12330 }, { "epoch": 0.20191442362758732, "grad_norm": 0.6014368534088135, "learning_rate": 5.047860590689684e-06, "loss": 0.0051, "step": 12340 }, { "epoch": 0.20207804957866318, "grad_norm": 0.29724380373954773, "learning_rate": 5.0519512394665795e-06, "loss": 0.0067, "step": 12350 }, { "epoch": 0.202241675529739, "grad_norm": 0.8124383091926575, "learning_rate": 5.056041888243476e-06, "loss": 0.0056, "step": 12360 }, { "epoch": 0.20240530148081487, "grad_norm": 0.26274561882019043, "learning_rate": 5.060132537020371e-06, "loss": 0.0058, "step": 12370 }, { "epoch": 0.2025689274318907, "grad_norm": 0.2812219560146332, "learning_rate": 5.0642231857972675e-06, "loss": 0.0055, "step": 12380 }, { "epoch": 0.20273255338296653, "grad_norm": 0.23372188210487366, "learning_rate": 5.068313834574163e-06, "loss": 0.012, "step": 12390 }, { "epoch": 0.2028961793340424, "grad_norm": 0.1705169528722763, "learning_rate": 5.07240448335106e-06, "loss": 0.0035, "step": 12400 }, { "epoch": 0.20305980528511822, "grad_norm": 0.341804176568985, "learning_rate": 5.0764951321279555e-06, "loss": 0.0057, "step": 12410 }, { "epoch": 0.20322343123619407, "grad_norm": 0.4900599718093872, "learning_rate": 5.080585780904852e-06, "loss": 0.0062, "step": 12420 }, { "epoch": 0.2033870571872699, "grad_norm": 0.47706037759780884, "learning_rate": 5.084676429681748e-06, "loss": 0.0061, "step": 12430 }, { "epoch": 0.20355068313834573, "grad_norm": 0.31758180260658264, "learning_rate": 5.088767078458644e-06, "loss": 0.0066, "step": 12440 }, { "epoch": 0.2037143090894216, "grad_norm": 0.28509029746055603, "learning_rate": 5.092857727235539e-06, "loss": 0.0044, "step": 12450 }, { "epoch": 0.20387793504049742, "grad_norm": 0.18527908623218536, "learning_rate": 5.096948376012436e-06, "loss": 0.0058, "step": 12460 }, { "epoch": 0.20404156099157325, "grad_norm": 0.25118526816368103, "learning_rate": 5.1010390247893314e-06, "loss": 0.0054, "step": 12470 }, { "epoch": 0.2042051869426491, "grad_norm": 0.376507043838501, "learning_rate": 5.105129673566228e-06, "loss": 0.0056, "step": 12480 }, { "epoch": 0.20436881289372494, "grad_norm": 0.29002055525779724, "learning_rate": 5.109220322343124e-06, "loss": 0.0049, "step": 12490 }, { "epoch": 0.2045324388448008, "grad_norm": 0.17401456832885742, "learning_rate": 5.11331097112002e-06, "loss": 0.0075, "step": 12500 }, { "epoch": 0.20469606479587663, "grad_norm": 0.1565149873495102, "learning_rate": 5.117401619896916e-06, "loss": 0.0068, "step": 12510 }, { "epoch": 0.20485969074695246, "grad_norm": 0.3050566613674164, "learning_rate": 5.1214922686738125e-06, "loss": 0.0053, "step": 12520 }, { "epoch": 0.2050233166980283, "grad_norm": 0.233909010887146, "learning_rate": 5.125582917450707e-06, "loss": 0.0072, "step": 12530 }, { "epoch": 0.20518694264910414, "grad_norm": 0.2051415592432022, "learning_rate": 5.129673566227604e-06, "loss": 0.006, "step": 12540 }, { "epoch": 0.20535056860018, "grad_norm": 0.1716548055410385, "learning_rate": 5.1337642150045e-06, "loss": 0.0041, "step": 12550 }, { "epoch": 0.20551419455125583, "grad_norm": 0.3126802444458008, "learning_rate": 5.137854863781396e-06, "loss": 0.0054, "step": 12560 }, { "epoch": 0.20567782050233166, "grad_norm": 0.3838905096054077, "learning_rate": 5.141945512558292e-06, "loss": 0.0047, "step": 12570 }, { "epoch": 0.20584144645340752, "grad_norm": 0.2420954704284668, "learning_rate": 5.1460361613351884e-06, "loss": 0.0079, "step": 12580 }, { "epoch": 0.20600507240448335, "grad_norm": 0.23210854828357697, "learning_rate": 5.150126810112084e-06, "loss": 0.0056, "step": 12590 }, { "epoch": 0.2061686983555592, "grad_norm": 0.2516229450702667, "learning_rate": 5.154217458888981e-06, "loss": 0.0059, "step": 12600 }, { "epoch": 0.20633232430663503, "grad_norm": 0.25838491320610046, "learning_rate": 5.1583081076658756e-06, "loss": 0.0052, "step": 12610 }, { "epoch": 0.20649595025771086, "grad_norm": 0.09571520239114761, "learning_rate": 5.162398756442772e-06, "loss": 0.0048, "step": 12620 }, { "epoch": 0.20665957620878672, "grad_norm": 0.19840897619724274, "learning_rate": 5.166489405219668e-06, "loss": 0.0066, "step": 12630 }, { "epoch": 0.20682320215986255, "grad_norm": 0.35993286967277527, "learning_rate": 5.170580053996564e-06, "loss": 0.0058, "step": 12640 }, { "epoch": 0.20698682811093838, "grad_norm": 0.29612278938293457, "learning_rate": 5.17467070277346e-06, "loss": 0.0084, "step": 12650 }, { "epoch": 0.20715045406201424, "grad_norm": 0.2622396945953369, "learning_rate": 5.178761351550357e-06, "loss": 0.0049, "step": 12660 }, { "epoch": 0.20731408001309007, "grad_norm": 0.19489862024784088, "learning_rate": 5.182852000327252e-06, "loss": 0.0066, "step": 12670 }, { "epoch": 0.20747770596416593, "grad_norm": 0.5107905864715576, "learning_rate": 5.186942649104149e-06, "loss": 0.0085, "step": 12680 }, { "epoch": 0.20764133191524176, "grad_norm": 0.10897552222013474, "learning_rate": 5.191033297881044e-06, "loss": 0.0051, "step": 12690 }, { "epoch": 0.20780495786631759, "grad_norm": 0.29415425658226013, "learning_rate": 5.19512394665794e-06, "loss": 0.0059, "step": 12700 }, { "epoch": 0.20796858381739344, "grad_norm": 0.30819106101989746, "learning_rate": 5.199214595434836e-06, "loss": 0.0055, "step": 12710 }, { "epoch": 0.20813220976846927, "grad_norm": 0.12865321338176727, "learning_rate": 5.203305244211733e-06, "loss": 0.0035, "step": 12720 }, { "epoch": 0.20829583571954513, "grad_norm": 0.21491898596286774, "learning_rate": 5.207395892988628e-06, "loss": 0.0049, "step": 12730 }, { "epoch": 0.20845946167062096, "grad_norm": 0.576177716255188, "learning_rate": 5.211486541765525e-06, "loss": 0.0086, "step": 12740 }, { "epoch": 0.2086230876216968, "grad_norm": 0.25540775060653687, "learning_rate": 5.2155771905424206e-06, "loss": 0.0057, "step": 12750 }, { "epoch": 0.20878671357277265, "grad_norm": 0.17240296304225922, "learning_rate": 5.219667839319317e-06, "loss": 0.0076, "step": 12760 }, { "epoch": 0.20895033952384848, "grad_norm": 0.3096192479133606, "learning_rate": 5.223758488096212e-06, "loss": 0.0083, "step": 12770 }, { "epoch": 0.20911396547492433, "grad_norm": 0.10653487592935562, "learning_rate": 5.2278491368731085e-06, "loss": 0.0047, "step": 12780 }, { "epoch": 0.20927759142600016, "grad_norm": 0.2059009075164795, "learning_rate": 5.231939785650004e-06, "loss": 0.0081, "step": 12790 }, { "epoch": 0.209441217377076, "grad_norm": 0.07613440603017807, "learning_rate": 5.236030434426901e-06, "loss": 0.0083, "step": 12800 }, { "epoch": 0.20960484332815185, "grad_norm": 0.2324797660112381, "learning_rate": 5.2401210832037965e-06, "loss": 0.0049, "step": 12810 }, { "epoch": 0.20976846927922768, "grad_norm": 0.39309024810791016, "learning_rate": 5.244211731980693e-06, "loss": 0.0055, "step": 12820 }, { "epoch": 0.20993209523030354, "grad_norm": 0.10823529958724976, "learning_rate": 5.248302380757589e-06, "loss": 0.0065, "step": 12830 }, { "epoch": 0.21009572118137937, "grad_norm": 0.10268852859735489, "learning_rate": 5.252393029534485e-06, "loss": 0.0061, "step": 12840 }, { "epoch": 0.2102593471324552, "grad_norm": 0.3961237668991089, "learning_rate": 5.25648367831138e-06, "loss": 0.0055, "step": 12850 }, { "epoch": 0.21042297308353106, "grad_norm": 0.09107676893472672, "learning_rate": 5.260574327088277e-06, "loss": 0.0056, "step": 12860 }, { "epoch": 0.21058659903460689, "grad_norm": 0.2935256063938141, "learning_rate": 5.2646649758651724e-06, "loss": 0.004, "step": 12870 }, { "epoch": 0.21075022498568272, "grad_norm": 0.3953011929988861, "learning_rate": 5.268755624642069e-06, "loss": 0.0047, "step": 12880 }, { "epoch": 0.21091385093675857, "grad_norm": 0.19132383167743683, "learning_rate": 5.272846273418965e-06, "loss": 0.0041, "step": 12890 }, { "epoch": 0.2110774768878344, "grad_norm": 0.32521113753318787, "learning_rate": 5.276936922195861e-06, "loss": 0.006, "step": 12900 }, { "epoch": 0.21124110283891026, "grad_norm": 0.22624783217906952, "learning_rate": 5.281027570972756e-06, "loss": 0.0062, "step": 12910 }, { "epoch": 0.2114047287899861, "grad_norm": 0.20094077289104462, "learning_rate": 5.2851182197496535e-06, "loss": 0.0058, "step": 12920 }, { "epoch": 0.21156835474106192, "grad_norm": 0.27441850304603577, "learning_rate": 5.289208868526548e-06, "loss": 0.004, "step": 12930 }, { "epoch": 0.21173198069213778, "grad_norm": 0.3558763861656189, "learning_rate": 5.293299517303445e-06, "loss": 0.0053, "step": 12940 }, { "epoch": 0.2118956066432136, "grad_norm": 0.15553462505340576, "learning_rate": 5.297390166080341e-06, "loss": 0.0036, "step": 12950 }, { "epoch": 0.21205923259428947, "grad_norm": 0.19737417995929718, "learning_rate": 5.301480814857237e-06, "loss": 0.0056, "step": 12960 }, { "epoch": 0.2122228585453653, "grad_norm": 0.23187093436717987, "learning_rate": 5.305571463634133e-06, "loss": 0.0053, "step": 12970 }, { "epoch": 0.21238648449644112, "grad_norm": 0.3074353039264679, "learning_rate": 5.3096621124110295e-06, "loss": 0.0069, "step": 12980 }, { "epoch": 0.21255011044751698, "grad_norm": 0.2877925932407379, "learning_rate": 5.313752761187924e-06, "loss": 0.0071, "step": 12990 }, { "epoch": 0.2127137363985928, "grad_norm": 0.2555560767650604, "learning_rate": 5.317843409964821e-06, "loss": 0.0049, "step": 13000 }, { "epoch": 0.21287736234966867, "grad_norm": 0.3059338629245758, "learning_rate": 5.321934058741717e-06, "loss": 0.0033, "step": 13010 }, { "epoch": 0.2130409883007445, "grad_norm": 0.21236415207386017, "learning_rate": 5.326024707518613e-06, "loss": 0.0044, "step": 13020 }, { "epoch": 0.21320461425182033, "grad_norm": 0.27880361676216125, "learning_rate": 5.330115356295509e-06, "loss": 0.006, "step": 13030 }, { "epoch": 0.2133682402028962, "grad_norm": 0.2796422243118286, "learning_rate": 5.334206005072405e-06, "loss": 0.0068, "step": 13040 }, { "epoch": 0.21353186615397202, "grad_norm": 0.20327647030353546, "learning_rate": 5.338296653849301e-06, "loss": 0.006, "step": 13050 }, { "epoch": 0.21369549210504787, "grad_norm": 0.2529746890068054, "learning_rate": 5.342387302626198e-06, "loss": 0.0047, "step": 13060 }, { "epoch": 0.2138591180561237, "grad_norm": 0.3821907341480255, "learning_rate": 5.3464779514030925e-06, "loss": 0.0051, "step": 13070 }, { "epoch": 0.21402274400719953, "grad_norm": 0.23533156514167786, "learning_rate": 5.350568600179989e-06, "loss": 0.006, "step": 13080 }, { "epoch": 0.2141863699582754, "grad_norm": 0.44652244448661804, "learning_rate": 5.354659248956885e-06, "loss": 0.0044, "step": 13090 }, { "epoch": 0.21434999590935122, "grad_norm": 0.14073534309864044, "learning_rate": 5.358749897733781e-06, "loss": 0.0061, "step": 13100 }, { "epoch": 0.21451362186042705, "grad_norm": 0.13494594395160675, "learning_rate": 5.362840546510677e-06, "loss": 0.0043, "step": 13110 }, { "epoch": 0.2146772478115029, "grad_norm": 0.07882718741893768, "learning_rate": 5.366931195287574e-06, "loss": 0.004, "step": 13120 }, { "epoch": 0.21484087376257874, "grad_norm": 0.2565968632698059, "learning_rate": 5.371021844064469e-06, "loss": 0.0054, "step": 13130 }, { "epoch": 0.2150044997136546, "grad_norm": 0.1400710791349411, "learning_rate": 5.375112492841366e-06, "loss": 0.0062, "step": 13140 }, { "epoch": 0.21516812566473043, "grad_norm": 0.20806723833084106, "learning_rate": 5.379203141618261e-06, "loss": 0.0064, "step": 13150 }, { "epoch": 0.21533175161580626, "grad_norm": 0.470234215259552, "learning_rate": 5.383293790395157e-06, "loss": 0.0069, "step": 13160 }, { "epoch": 0.2154953775668821, "grad_norm": 0.30703094601631165, "learning_rate": 5.387384439172053e-06, "loss": 0.0064, "step": 13170 }, { "epoch": 0.21565900351795794, "grad_norm": 0.2885434925556183, "learning_rate": 5.3914750879489496e-06, "loss": 0.008, "step": 13180 }, { "epoch": 0.2158226294690338, "grad_norm": 0.16507022082805634, "learning_rate": 5.395565736725845e-06, "loss": 0.0082, "step": 13190 }, { "epoch": 0.21598625542010963, "grad_norm": 0.3633711040019989, "learning_rate": 5.399656385502741e-06, "loss": 0.0057, "step": 13200 }, { "epoch": 0.21614988137118546, "grad_norm": 0.36370086669921875, "learning_rate": 5.4037470342796375e-06, "loss": 0.006, "step": 13210 }, { "epoch": 0.21631350732226132, "grad_norm": 0.7826770544052124, "learning_rate": 5.407837683056532e-06, "loss": 0.0054, "step": 13220 }, { "epoch": 0.21647713327333715, "grad_norm": 0.2468007206916809, "learning_rate": 5.411928331833429e-06, "loss": 0.0064, "step": 13230 }, { "epoch": 0.216640759224413, "grad_norm": 0.046535465866327286, "learning_rate": 5.416018980610325e-06, "loss": 0.0052, "step": 13240 }, { "epoch": 0.21680438517548883, "grad_norm": 0.17650368809700012, "learning_rate": 5.420109629387221e-06, "loss": 0.0059, "step": 13250 }, { "epoch": 0.21696801112656466, "grad_norm": 0.2192971110343933, "learning_rate": 5.424200278164117e-06, "loss": 0.007, "step": 13260 }, { "epoch": 0.21713163707764052, "grad_norm": 0.36229148507118225, "learning_rate": 5.4282909269410135e-06, "loss": 0.0056, "step": 13270 }, { "epoch": 0.21729526302871635, "grad_norm": 0.16386255621910095, "learning_rate": 5.432381575717909e-06, "loss": 0.0051, "step": 13280 }, { "epoch": 0.2174588889797922, "grad_norm": 0.16969378292560577, "learning_rate": 5.436472224494806e-06, "loss": 0.0055, "step": 13290 }, { "epoch": 0.21762251493086804, "grad_norm": 0.27967017889022827, "learning_rate": 5.440562873271701e-06, "loss": 0.0054, "step": 13300 }, { "epoch": 0.21778614088194387, "grad_norm": 0.10583193600177765, "learning_rate": 5.444653522048597e-06, "loss": 0.0063, "step": 13310 }, { "epoch": 0.21794976683301973, "grad_norm": 0.24081790447235107, "learning_rate": 5.448744170825493e-06, "loss": 0.0034, "step": 13320 }, { "epoch": 0.21811339278409556, "grad_norm": 0.3276372253894806, "learning_rate": 5.452834819602389e-06, "loss": 0.0064, "step": 13330 }, { "epoch": 0.21827701873517139, "grad_norm": 0.2675396502017975, "learning_rate": 5.456925468379285e-06, "loss": 0.0053, "step": 13340 }, { "epoch": 0.21844064468624724, "grad_norm": 0.17607930302619934, "learning_rate": 5.461016117156182e-06, "loss": 0.0076, "step": 13350 }, { "epoch": 0.21860427063732307, "grad_norm": 0.4253722131252289, "learning_rate": 5.465106765933077e-06, "loss": 0.0094, "step": 13360 }, { "epoch": 0.21876789658839893, "grad_norm": 0.642120897769928, "learning_rate": 5.469197414709974e-06, "loss": 0.0078, "step": 13370 }, { "epoch": 0.21893152253947476, "grad_norm": 0.13732227683067322, "learning_rate": 5.473288063486869e-06, "loss": 0.0045, "step": 13380 }, { "epoch": 0.2190951484905506, "grad_norm": 0.1933867335319519, "learning_rate": 5.477378712263765e-06, "loss": 0.0059, "step": 13390 }, { "epoch": 0.21925877444162645, "grad_norm": 0.13849449157714844, "learning_rate": 5.481469361040661e-06, "loss": 0.006, "step": 13400 }, { "epoch": 0.21942240039270228, "grad_norm": 0.5225475430488586, "learning_rate": 5.485560009817558e-06, "loss": 0.0073, "step": 13410 }, { "epoch": 0.21958602634377813, "grad_norm": 0.1187896803021431, "learning_rate": 5.489650658594453e-06, "loss": 0.0033, "step": 13420 }, { "epoch": 0.21974965229485396, "grad_norm": 0.3173638582229614, "learning_rate": 5.49374130737135e-06, "loss": 0.0072, "step": 13430 }, { "epoch": 0.2199132782459298, "grad_norm": 0.09578055888414383, "learning_rate": 5.497831956148246e-06, "loss": 0.0074, "step": 13440 }, { "epoch": 0.22007690419700565, "grad_norm": 0.17457492649555206, "learning_rate": 5.501922604925142e-06, "loss": 0.0052, "step": 13450 }, { "epoch": 0.22024053014808148, "grad_norm": 0.1466219425201416, "learning_rate": 5.506013253702037e-06, "loss": 0.0035, "step": 13460 }, { "epoch": 0.22040415609915734, "grad_norm": 0.33679795265197754, "learning_rate": 5.5101039024789336e-06, "loss": 0.0061, "step": 13470 }, { "epoch": 0.22056778205023317, "grad_norm": 0.2536139488220215, "learning_rate": 5.514194551255829e-06, "loss": 0.004, "step": 13480 }, { "epoch": 0.220731408001309, "grad_norm": 0.18243299424648285, "learning_rate": 5.518285200032726e-06, "loss": 0.0056, "step": 13490 }, { "epoch": 0.22089503395238486, "grad_norm": 0.19751858711242676, "learning_rate": 5.5223758488096215e-06, "loss": 0.0073, "step": 13500 }, { "epoch": 0.22105865990346069, "grad_norm": 0.2707878053188324, "learning_rate": 5.526466497586518e-06, "loss": 0.0038, "step": 13510 }, { "epoch": 0.22122228585453654, "grad_norm": 0.15621255338191986, "learning_rate": 5.530557146363413e-06, "loss": 0.0066, "step": 13520 }, { "epoch": 0.22138591180561237, "grad_norm": 0.5182904005050659, "learning_rate": 5.53464779514031e-06, "loss": 0.0064, "step": 13530 }, { "epoch": 0.2215495377566882, "grad_norm": 0.4845060110092163, "learning_rate": 5.538738443917205e-06, "loss": 0.0061, "step": 13540 }, { "epoch": 0.22171316370776406, "grad_norm": 0.14424413442611694, "learning_rate": 5.542829092694102e-06, "loss": 0.0044, "step": 13550 }, { "epoch": 0.2218767896588399, "grad_norm": 0.1840703785419464, "learning_rate": 5.5469197414709975e-06, "loss": 0.0051, "step": 13560 }, { "epoch": 0.22204041560991572, "grad_norm": 0.26374658942222595, "learning_rate": 5.551010390247894e-06, "loss": 0.0093, "step": 13570 }, { "epoch": 0.22220404156099158, "grad_norm": 0.13048377633094788, "learning_rate": 5.55510103902479e-06, "loss": 0.005, "step": 13580 }, { "epoch": 0.2223676675120674, "grad_norm": 0.16124430298805237, "learning_rate": 5.559191687801686e-06, "loss": 0.0058, "step": 13590 }, { "epoch": 0.22253129346314326, "grad_norm": 0.2681410312652588, "learning_rate": 5.563282336578581e-06, "loss": 0.0059, "step": 13600 }, { "epoch": 0.2226949194142191, "grad_norm": 0.3200013339519501, "learning_rate": 5.567372985355478e-06, "loss": 0.0063, "step": 13610 }, { "epoch": 0.22285854536529492, "grad_norm": 0.13598883152008057, "learning_rate": 5.571463634132373e-06, "loss": 0.007, "step": 13620 }, { "epoch": 0.22302217131637078, "grad_norm": 0.2553706765174866, "learning_rate": 5.57555428290927e-06, "loss": 0.0069, "step": 13630 }, { "epoch": 0.2231857972674466, "grad_norm": 0.2136799693107605, "learning_rate": 5.579644931686166e-06, "loss": 0.0051, "step": 13640 }, { "epoch": 0.22334942321852247, "grad_norm": 0.2547089755535126, "learning_rate": 5.583735580463062e-06, "loss": 0.0038, "step": 13650 }, { "epoch": 0.2235130491695983, "grad_norm": 0.24491526186466217, "learning_rate": 5.587826229239958e-06, "loss": 0.0052, "step": 13660 }, { "epoch": 0.22367667512067413, "grad_norm": 0.3228518068790436, "learning_rate": 5.5919168780168545e-06, "loss": 0.0055, "step": 13670 }, { "epoch": 0.22384030107175, "grad_norm": 0.31561779975891113, "learning_rate": 5.596007526793749e-06, "loss": 0.0062, "step": 13680 }, { "epoch": 0.22400392702282582, "grad_norm": 0.2978101968765259, "learning_rate": 5.600098175570646e-06, "loss": 0.0043, "step": 13690 }, { "epoch": 0.22416755297390167, "grad_norm": 0.2244068831205368, "learning_rate": 5.604188824347542e-06, "loss": 0.0063, "step": 13700 }, { "epoch": 0.2243311789249775, "grad_norm": 0.18937498331069946, "learning_rate": 5.608279473124438e-06, "loss": 0.0032, "step": 13710 }, { "epoch": 0.22449480487605333, "grad_norm": 0.40831300616264343, "learning_rate": 5.612370121901334e-06, "loss": 0.0049, "step": 13720 }, { "epoch": 0.2246584308271292, "grad_norm": 0.3180380165576935, "learning_rate": 5.6164607706782304e-06, "loss": 0.0079, "step": 13730 }, { "epoch": 0.22482205677820502, "grad_norm": 0.33835768699645996, "learning_rate": 5.620551419455126e-06, "loss": 0.0067, "step": 13740 }, { "epoch": 0.22498568272928085, "grad_norm": 0.3634449541568756, "learning_rate": 5.624642068232023e-06, "loss": 0.005, "step": 13750 }, { "epoch": 0.2251493086803567, "grad_norm": 0.20941343903541565, "learning_rate": 5.6287327170089176e-06, "loss": 0.005, "step": 13760 }, { "epoch": 0.22531293463143254, "grad_norm": 0.2777906060218811, "learning_rate": 5.632823365785814e-06, "loss": 0.0105, "step": 13770 }, { "epoch": 0.2254765605825084, "grad_norm": 0.10400465130805969, "learning_rate": 5.63691401456271e-06, "loss": 0.0051, "step": 13780 }, { "epoch": 0.22564018653358422, "grad_norm": 0.34114325046539307, "learning_rate": 5.641004663339606e-06, "loss": 0.0074, "step": 13790 }, { "epoch": 0.22580381248466005, "grad_norm": 0.15259796380996704, "learning_rate": 5.645095312116502e-06, "loss": 0.0056, "step": 13800 }, { "epoch": 0.2259674384357359, "grad_norm": 0.18578624725341797, "learning_rate": 5.649185960893399e-06, "loss": 0.0048, "step": 13810 }, { "epoch": 0.22613106438681174, "grad_norm": 0.30198147892951965, "learning_rate": 5.653276609670294e-06, "loss": 0.0066, "step": 13820 }, { "epoch": 0.2262946903378876, "grad_norm": 0.1933286190032959, "learning_rate": 5.657367258447191e-06, "loss": 0.0052, "step": 13830 }, { "epoch": 0.22645831628896343, "grad_norm": 0.1645790934562683, "learning_rate": 5.661457907224086e-06, "loss": 0.008, "step": 13840 }, { "epoch": 0.22662194224003926, "grad_norm": 0.40378186106681824, "learning_rate": 5.665548556000982e-06, "loss": 0.006, "step": 13850 }, { "epoch": 0.22678556819111512, "grad_norm": 0.34223198890686035, "learning_rate": 5.669639204777878e-06, "loss": 0.0059, "step": 13860 }, { "epoch": 0.22694919414219095, "grad_norm": 0.2772032916545868, "learning_rate": 5.6737298535547746e-06, "loss": 0.0067, "step": 13870 }, { "epoch": 0.2271128200932668, "grad_norm": 0.31348899006843567, "learning_rate": 5.67782050233167e-06, "loss": 0.0064, "step": 13880 }, { "epoch": 0.22727644604434263, "grad_norm": 0.2948260009288788, "learning_rate": 5.681911151108567e-06, "loss": 0.0056, "step": 13890 }, { "epoch": 0.22744007199541846, "grad_norm": 0.17786602675914764, "learning_rate": 5.6860017998854625e-06, "loss": 0.0063, "step": 13900 }, { "epoch": 0.22760369794649432, "grad_norm": 0.231853187084198, "learning_rate": 5.690092448662359e-06, "loss": 0.0056, "step": 13910 }, { "epoch": 0.22776732389757015, "grad_norm": 0.15944230556488037, "learning_rate": 5.694183097439254e-06, "loss": 0.0049, "step": 13920 }, { "epoch": 0.227930949848646, "grad_norm": 0.11967223882675171, "learning_rate": 5.6982737462161505e-06, "loss": 0.0061, "step": 13930 }, { "epoch": 0.22809457579972184, "grad_norm": 0.13277150690555573, "learning_rate": 5.702364394993046e-06, "loss": 0.0044, "step": 13940 }, { "epoch": 0.22825820175079767, "grad_norm": 0.18200495839118958, "learning_rate": 5.706455043769943e-06, "loss": 0.0037, "step": 13950 }, { "epoch": 0.22842182770187353, "grad_norm": 0.12429895997047424, "learning_rate": 5.7105456925468385e-06, "loss": 0.004, "step": 13960 }, { "epoch": 0.22858545365294936, "grad_norm": 0.356438010931015, "learning_rate": 5.714636341323735e-06, "loss": 0.0059, "step": 13970 }, { "epoch": 0.22874907960402519, "grad_norm": 0.2690216898918152, "learning_rate": 5.718726990100631e-06, "loss": 0.0054, "step": 13980 }, { "epoch": 0.22891270555510104, "grad_norm": 0.3261720538139343, "learning_rate": 5.722817638877527e-06, "loss": 0.0069, "step": 13990 }, { "epoch": 0.22907633150617687, "grad_norm": 0.2134871482849121, "learning_rate": 5.726908287654422e-06, "loss": 0.0062, "step": 14000 }, { "epoch": 0.22923995745725273, "grad_norm": 0.21844306588172913, "learning_rate": 5.730998936431319e-06, "loss": 0.0047, "step": 14010 }, { "epoch": 0.22940358340832856, "grad_norm": 0.3882652223110199, "learning_rate": 5.7350895852082144e-06, "loss": 0.009, "step": 14020 }, { "epoch": 0.2295672093594044, "grad_norm": 0.14818373322486877, "learning_rate": 5.739180233985111e-06, "loss": 0.0036, "step": 14030 }, { "epoch": 0.22973083531048025, "grad_norm": 0.1437871754169464, "learning_rate": 5.743270882762007e-06, "loss": 0.0059, "step": 14040 }, { "epoch": 0.22989446126155608, "grad_norm": 0.314844012260437, "learning_rate": 5.747361531538903e-06, "loss": 0.005, "step": 14050 }, { "epoch": 0.23005808721263193, "grad_norm": 0.2389242798089981, "learning_rate": 5.751452180315799e-06, "loss": 0.0048, "step": 14060 }, { "epoch": 0.23022171316370776, "grad_norm": 0.1123579666018486, "learning_rate": 5.755542829092694e-06, "loss": 0.0044, "step": 14070 }, { "epoch": 0.2303853391147836, "grad_norm": 0.4211624562740326, "learning_rate": 5.75963347786959e-06, "loss": 0.0093, "step": 14080 }, { "epoch": 0.23054896506585945, "grad_norm": 0.3057348132133484, "learning_rate": 5.763724126646486e-06, "loss": 0.005, "step": 14090 }, { "epoch": 0.23071259101693528, "grad_norm": 0.44600942730903625, "learning_rate": 5.767814775423383e-06, "loss": 0.0044, "step": 14100 }, { "epoch": 0.23087621696801114, "grad_norm": 0.2926277220249176, "learning_rate": 5.771905424200278e-06, "loss": 0.0028, "step": 14110 }, { "epoch": 0.23103984291908697, "grad_norm": 0.39210575819015503, "learning_rate": 5.775996072977175e-06, "loss": 0.0052, "step": 14120 }, { "epoch": 0.2312034688701628, "grad_norm": 0.3104311525821686, "learning_rate": 5.78008672175407e-06, "loss": 0.0068, "step": 14130 }, { "epoch": 0.23136709482123866, "grad_norm": 0.10409224778413773, "learning_rate": 5.784177370530966e-06, "loss": 0.0057, "step": 14140 }, { "epoch": 0.23153072077231449, "grad_norm": 0.06968032568693161, "learning_rate": 5.788268019307862e-06, "loss": 0.0073, "step": 14150 }, { "epoch": 0.23169434672339034, "grad_norm": 0.30508825182914734, "learning_rate": 5.7923586680847586e-06, "loss": 0.0047, "step": 14160 }, { "epoch": 0.23185797267446617, "grad_norm": 0.4546557068824768, "learning_rate": 5.796449316861654e-06, "loss": 0.0048, "step": 14170 }, { "epoch": 0.232021598625542, "grad_norm": 0.4909617006778717, "learning_rate": 5.800539965638551e-06, "loss": 0.0071, "step": 14180 }, { "epoch": 0.23218522457661786, "grad_norm": 0.6030717492103577, "learning_rate": 5.8046306144154465e-06, "loss": 0.0067, "step": 14190 }, { "epoch": 0.2323488505276937, "grad_norm": 0.3564246892929077, "learning_rate": 5.808721263192343e-06, "loss": 0.0067, "step": 14200 }, { "epoch": 0.23251247647876952, "grad_norm": 0.2273881733417511, "learning_rate": 5.812811911969238e-06, "loss": 0.0073, "step": 14210 }, { "epoch": 0.23267610242984538, "grad_norm": 0.07840313017368317, "learning_rate": 5.8169025607461345e-06, "loss": 0.0067, "step": 14220 }, { "epoch": 0.2328397283809212, "grad_norm": 0.24078573286533356, "learning_rate": 5.82099320952303e-06, "loss": 0.0081, "step": 14230 }, { "epoch": 0.23300335433199706, "grad_norm": 0.3099464178085327, "learning_rate": 5.825083858299927e-06, "loss": 0.0052, "step": 14240 }, { "epoch": 0.2331669802830729, "grad_norm": 0.29668036103248596, "learning_rate": 5.8291745070768225e-06, "loss": 0.0067, "step": 14250 }, { "epoch": 0.23333060623414872, "grad_norm": 0.36040419340133667, "learning_rate": 5.833265155853719e-06, "loss": 0.0051, "step": 14260 }, { "epoch": 0.23349423218522458, "grad_norm": 0.18252961337566376, "learning_rate": 5.837355804630615e-06, "loss": 0.0064, "step": 14270 }, { "epoch": 0.2336578581363004, "grad_norm": 0.20247343182563782, "learning_rate": 5.841446453407511e-06, "loss": 0.004, "step": 14280 }, { "epoch": 0.23382148408737627, "grad_norm": 0.14536994695663452, "learning_rate": 5.845537102184406e-06, "loss": 0.0052, "step": 14290 }, { "epoch": 0.2339851100384521, "grad_norm": 0.39473190903663635, "learning_rate": 5.849627750961303e-06, "loss": 0.0069, "step": 14300 }, { "epoch": 0.23414873598952793, "grad_norm": 0.5212802886962891, "learning_rate": 5.8537183997381984e-06, "loss": 0.0051, "step": 14310 }, { "epoch": 0.23431236194060379, "grad_norm": 0.4780243933200836, "learning_rate": 5.857809048515095e-06, "loss": 0.009, "step": 14320 }, { "epoch": 0.23447598789167962, "grad_norm": 0.21021021902561188, "learning_rate": 5.861899697291991e-06, "loss": 0.0057, "step": 14330 }, { "epoch": 0.23463961384275547, "grad_norm": 0.5564619898796082, "learning_rate": 5.865990346068887e-06, "loss": 0.0057, "step": 14340 }, { "epoch": 0.2348032397938313, "grad_norm": 0.04012363776564598, "learning_rate": 5.870080994845783e-06, "loss": 0.0076, "step": 14350 }, { "epoch": 0.23496686574490713, "grad_norm": 0.75568687915802, "learning_rate": 5.8741716436226795e-06, "loss": 0.0079, "step": 14360 }, { "epoch": 0.235130491695983, "grad_norm": 0.15325964987277985, "learning_rate": 5.878262292399574e-06, "loss": 0.0058, "step": 14370 }, { "epoch": 0.23529411764705882, "grad_norm": 0.2867588698863983, "learning_rate": 5.882352941176471e-06, "loss": 0.0074, "step": 14380 }, { "epoch": 0.23545774359813468, "grad_norm": 0.10078386962413788, "learning_rate": 5.886443589953367e-06, "loss": 0.0081, "step": 14390 }, { "epoch": 0.2356213695492105, "grad_norm": 0.15865343809127808, "learning_rate": 5.890534238730263e-06, "loss": 0.0077, "step": 14400 }, { "epoch": 0.23578499550028634, "grad_norm": 0.23428471386432648, "learning_rate": 5.894624887507159e-06, "loss": 0.0065, "step": 14410 }, { "epoch": 0.2359486214513622, "grad_norm": 0.27213379740715027, "learning_rate": 5.8987155362840554e-06, "loss": 0.0044, "step": 14420 }, { "epoch": 0.23611224740243802, "grad_norm": 0.33400261402130127, "learning_rate": 5.902806185060951e-06, "loss": 0.0066, "step": 14430 }, { "epoch": 0.23627587335351385, "grad_norm": 0.4187803566455841, "learning_rate": 5.906896833837848e-06, "loss": 0.0056, "step": 14440 }, { "epoch": 0.2364394993045897, "grad_norm": 0.20335114002227783, "learning_rate": 5.9109874826147426e-06, "loss": 0.008, "step": 14450 }, { "epoch": 0.23660312525566554, "grad_norm": 0.18088027834892273, "learning_rate": 5.915078131391639e-06, "loss": 0.0047, "step": 14460 }, { "epoch": 0.2367667512067414, "grad_norm": 0.42793741822242737, "learning_rate": 5.919168780168535e-06, "loss": 0.0054, "step": 14470 }, { "epoch": 0.23693037715781723, "grad_norm": 0.23710022866725922, "learning_rate": 5.923259428945431e-06, "loss": 0.004, "step": 14480 }, { "epoch": 0.23709400310889306, "grad_norm": 0.5717060565948486, "learning_rate": 5.927350077722327e-06, "loss": 0.0049, "step": 14490 }, { "epoch": 0.23725762905996892, "grad_norm": 0.11903871595859528, "learning_rate": 5.931440726499224e-06, "loss": 0.0032, "step": 14500 }, { "epoch": 0.23742125501104475, "grad_norm": 0.03658638522028923, "learning_rate": 5.935531375276119e-06, "loss": 0.0049, "step": 14510 }, { "epoch": 0.2375848809621206, "grad_norm": 0.2194136381149292, "learning_rate": 5.939622024053016e-06, "loss": 0.0043, "step": 14520 }, { "epoch": 0.23774850691319643, "grad_norm": 0.2365874946117401, "learning_rate": 5.943712672829911e-06, "loss": 0.0067, "step": 14530 }, { "epoch": 0.23791213286427226, "grad_norm": 0.3578321635723114, "learning_rate": 5.947803321606807e-06, "loss": 0.0039, "step": 14540 }, { "epoch": 0.23807575881534812, "grad_norm": 0.3870665431022644, "learning_rate": 5.951893970383703e-06, "loss": 0.005, "step": 14550 }, { "epoch": 0.23823938476642395, "grad_norm": 0.14380130171775818, "learning_rate": 5.9559846191606e-06, "loss": 0.0046, "step": 14560 }, { "epoch": 0.2384030107174998, "grad_norm": 0.12963660061359406, "learning_rate": 5.960075267937495e-06, "loss": 0.0046, "step": 14570 }, { "epoch": 0.23856663666857564, "grad_norm": 0.17952145636081696, "learning_rate": 5.964165916714392e-06, "loss": 0.0053, "step": 14580 }, { "epoch": 0.23873026261965147, "grad_norm": 0.0959174633026123, "learning_rate": 5.9682565654912876e-06, "loss": 0.0078, "step": 14590 }, { "epoch": 0.23889388857072733, "grad_norm": 0.1235545426607132, "learning_rate": 5.972347214268184e-06, "loss": 0.0036, "step": 14600 }, { "epoch": 0.23905751452180315, "grad_norm": 0.3402208089828491, "learning_rate": 5.976437863045079e-06, "loss": 0.0068, "step": 14610 }, { "epoch": 0.23922114047287898, "grad_norm": 0.07358551025390625, "learning_rate": 5.9805285118219755e-06, "loss": 0.0047, "step": 14620 }, { "epoch": 0.23938476642395484, "grad_norm": 0.07803614437580109, "learning_rate": 5.984619160598871e-06, "loss": 0.0039, "step": 14630 }, { "epoch": 0.23954839237503067, "grad_norm": 0.31854450702667236, "learning_rate": 5.988709809375768e-06, "loss": 0.005, "step": 14640 }, { "epoch": 0.23971201832610653, "grad_norm": 0.19740121066570282, "learning_rate": 5.9928004581526635e-06, "loss": 0.0047, "step": 14650 }, { "epoch": 0.23987564427718236, "grad_norm": 0.28780025243759155, "learning_rate": 5.99689110692956e-06, "loss": 0.0072, "step": 14660 }, { "epoch": 0.2400392702282582, "grad_norm": 0.30051249265670776, "learning_rate": 6.000981755706456e-06, "loss": 0.0067, "step": 14670 }, { "epoch": 0.24020289617933405, "grad_norm": 0.12498542666435242, "learning_rate": 6.005072404483352e-06, "loss": 0.005, "step": 14680 }, { "epoch": 0.24036652213040988, "grad_norm": 0.09199636429548264, "learning_rate": 6.009163053260247e-06, "loss": 0.0031, "step": 14690 }, { "epoch": 0.24053014808148573, "grad_norm": 0.05660368502140045, "learning_rate": 6.013253702037144e-06, "loss": 0.0038, "step": 14700 }, { "epoch": 0.24069377403256156, "grad_norm": 0.5036044120788574, "learning_rate": 6.0173443508140394e-06, "loss": 0.0078, "step": 14710 }, { "epoch": 0.2408573999836374, "grad_norm": 0.25386109948158264, "learning_rate": 6.021434999590936e-06, "loss": 0.0044, "step": 14720 }, { "epoch": 0.24102102593471325, "grad_norm": 0.1535719335079193, "learning_rate": 6.025525648367832e-06, "loss": 0.005, "step": 14730 }, { "epoch": 0.24118465188578908, "grad_norm": 0.10951091349124908, "learning_rate": 6.029616297144728e-06, "loss": 0.0052, "step": 14740 }, { "epoch": 0.24134827783686494, "grad_norm": 0.3470185697078705, "learning_rate": 6.033706945921623e-06, "loss": 0.0052, "step": 14750 }, { "epoch": 0.24151190378794077, "grad_norm": 0.3239732086658478, "learning_rate": 6.0377975946985205e-06, "loss": 0.0064, "step": 14760 }, { "epoch": 0.2416755297390166, "grad_norm": 0.1985783874988556, "learning_rate": 6.041888243475415e-06, "loss": 0.0061, "step": 14770 }, { "epoch": 0.24183915569009246, "grad_norm": 0.1293640434741974, "learning_rate": 6.045978892252312e-06, "loss": 0.0034, "step": 14780 }, { "epoch": 0.24200278164116829, "grad_norm": 0.14833714067935944, "learning_rate": 6.050069541029208e-06, "loss": 0.0039, "step": 14790 }, { "epoch": 0.24216640759224414, "grad_norm": 0.13785865902900696, "learning_rate": 6.054160189806104e-06, "loss": 0.0073, "step": 14800 }, { "epoch": 0.24233003354331997, "grad_norm": 0.29925528168678284, "learning_rate": 6.058250838583e-06, "loss": 0.0054, "step": 14810 }, { "epoch": 0.2424936594943958, "grad_norm": 0.2730848789215088, "learning_rate": 6.0623414873598965e-06, "loss": 0.0062, "step": 14820 }, { "epoch": 0.24265728544547166, "grad_norm": 0.19348080456256866, "learning_rate": 6.066432136136791e-06, "loss": 0.0089, "step": 14830 }, { "epoch": 0.2428209113965475, "grad_norm": 0.08180736750364304, "learning_rate": 6.070522784913688e-06, "loss": 0.0055, "step": 14840 }, { "epoch": 0.24298453734762332, "grad_norm": 0.1534387320280075, "learning_rate": 6.074613433690584e-06, "loss": 0.0054, "step": 14850 }, { "epoch": 0.24314816329869918, "grad_norm": 0.14646729826927185, "learning_rate": 6.07870408246748e-06, "loss": 0.0067, "step": 14860 }, { "epoch": 0.243311789249775, "grad_norm": 0.14335399866104126, "learning_rate": 6.082794731244376e-06, "loss": 0.0037, "step": 14870 }, { "epoch": 0.24347541520085086, "grad_norm": 0.3513915240764618, "learning_rate": 6.086885380021272e-06, "loss": 0.0044, "step": 14880 }, { "epoch": 0.2436390411519267, "grad_norm": 0.1069168895483017, "learning_rate": 6.090976028798168e-06, "loss": 0.0046, "step": 14890 }, { "epoch": 0.24380266710300252, "grad_norm": 0.13567444682121277, "learning_rate": 6.095066677575065e-06, "loss": 0.0033, "step": 14900 }, { "epoch": 0.24396629305407838, "grad_norm": 0.18619506061077118, "learning_rate": 6.0991573263519595e-06, "loss": 0.0074, "step": 14910 }, { "epoch": 0.2441299190051542, "grad_norm": 0.1902233511209488, "learning_rate": 6.103247975128856e-06, "loss": 0.0041, "step": 14920 }, { "epoch": 0.24429354495623007, "grad_norm": 0.1372702270746231, "learning_rate": 6.107338623905752e-06, "loss": 0.0071, "step": 14930 }, { "epoch": 0.2444571709073059, "grad_norm": 0.09901681542396545, "learning_rate": 6.111429272682648e-06, "loss": 0.0038, "step": 14940 }, { "epoch": 0.24462079685838173, "grad_norm": 0.13739995658397675, "learning_rate": 6.115519921459544e-06, "loss": 0.0038, "step": 14950 }, { "epoch": 0.24478442280945759, "grad_norm": 0.1895793080329895, "learning_rate": 6.11961057023644e-06, "loss": 0.005, "step": 14960 }, { "epoch": 0.24494804876053342, "grad_norm": 0.21043923497200012, "learning_rate": 6.123701219013336e-06, "loss": 0.0064, "step": 14970 }, { "epoch": 0.24511167471160927, "grad_norm": 0.3341471254825592, "learning_rate": 6.127791867790231e-06, "loss": 0.0073, "step": 14980 }, { "epoch": 0.2452753006626851, "grad_norm": 0.09192818403244019, "learning_rate": 6.131882516567128e-06, "loss": 0.0048, "step": 14990 }, { "epoch": 0.24543892661376093, "grad_norm": 0.3511482775211334, "learning_rate": 6.1359731653440234e-06, "loss": 0.0064, "step": 15000 }, { "epoch": 0.2456025525648368, "grad_norm": 0.15601126849651337, "learning_rate": 6.14006381412092e-06, "loss": 0.0028, "step": 15010 }, { "epoch": 0.24576617851591262, "grad_norm": 0.3366992771625519, "learning_rate": 6.144154462897816e-06, "loss": 0.0055, "step": 15020 }, { "epoch": 0.24592980446698848, "grad_norm": 0.09476987272500992, "learning_rate": 6.148245111674712e-06, "loss": 0.0058, "step": 15030 }, { "epoch": 0.2460934304180643, "grad_norm": 0.2929367125034332, "learning_rate": 6.152335760451608e-06, "loss": 0.007, "step": 15040 }, { "epoch": 0.24625705636914014, "grad_norm": 0.12254449725151062, "learning_rate": 6.1564264092285045e-06, "loss": 0.0069, "step": 15050 }, { "epoch": 0.246420682320216, "grad_norm": 0.25186407566070557, "learning_rate": 6.160517058005399e-06, "loss": 0.0049, "step": 15060 }, { "epoch": 0.24658430827129182, "grad_norm": 0.11779588460922241, "learning_rate": 6.164607706782296e-06, "loss": 0.0051, "step": 15070 }, { "epoch": 0.24674793422236765, "grad_norm": 0.6581495404243469, "learning_rate": 6.168698355559192e-06, "loss": 0.0068, "step": 15080 }, { "epoch": 0.2469115601734435, "grad_norm": 0.07045186311006546, "learning_rate": 6.172789004336088e-06, "loss": 0.0034, "step": 15090 }, { "epoch": 0.24707518612451934, "grad_norm": 0.25319018959999084, "learning_rate": 6.176879653112984e-06, "loss": 0.0043, "step": 15100 }, { "epoch": 0.2472388120755952, "grad_norm": 0.11416971683502197, "learning_rate": 6.1809703018898805e-06, "loss": 0.0071, "step": 15110 }, { "epoch": 0.24740243802667103, "grad_norm": 0.25165674090385437, "learning_rate": 6.185060950666776e-06, "loss": 0.0048, "step": 15120 }, { "epoch": 0.24756606397774686, "grad_norm": 0.2683257758617401, "learning_rate": 6.189151599443673e-06, "loss": 0.0044, "step": 15130 }, { "epoch": 0.24772968992882272, "grad_norm": 0.2208983153104782, "learning_rate": 6.193242248220568e-06, "loss": 0.0054, "step": 15140 }, { "epoch": 0.24789331587989855, "grad_norm": 0.02654389850795269, "learning_rate": 6.197332896997464e-06, "loss": 0.0039, "step": 15150 }, { "epoch": 0.2480569418309744, "grad_norm": 0.24881313741207123, "learning_rate": 6.20142354577436e-06, "loss": 0.0046, "step": 15160 }, { "epoch": 0.24822056778205023, "grad_norm": 0.1501893848180771, "learning_rate": 6.205514194551256e-06, "loss": 0.0045, "step": 15170 }, { "epoch": 0.24838419373312606, "grad_norm": 0.21987955272197723, "learning_rate": 6.209604843328152e-06, "loss": 0.0059, "step": 15180 }, { "epoch": 0.24854781968420192, "grad_norm": 0.1687382161617279, "learning_rate": 6.213695492105049e-06, "loss": 0.0057, "step": 15190 }, { "epoch": 0.24871144563527775, "grad_norm": 0.22444510459899902, "learning_rate": 6.217786140881944e-06, "loss": 0.0048, "step": 15200 }, { "epoch": 0.2488750715863536, "grad_norm": 0.04187791422009468, "learning_rate": 6.221876789658841e-06, "loss": 0.0026, "step": 15210 }, { "epoch": 0.24903869753742944, "grad_norm": 0.29989373683929443, "learning_rate": 6.225967438435736e-06, "loss": 0.0042, "step": 15220 }, { "epoch": 0.24920232348850527, "grad_norm": 0.08411089330911636, "learning_rate": 6.230058087212632e-06, "loss": 0.0069, "step": 15230 }, { "epoch": 0.24936594943958112, "grad_norm": 0.0845126286149025, "learning_rate": 6.234148735989528e-06, "loss": 0.0045, "step": 15240 }, { "epoch": 0.24952957539065695, "grad_norm": 0.13904179632663727, "learning_rate": 6.238239384766425e-06, "loss": 0.0037, "step": 15250 }, { "epoch": 0.2496932013417328, "grad_norm": 0.26953741908073425, "learning_rate": 6.24233003354332e-06, "loss": 0.0056, "step": 15260 }, { "epoch": 0.24985682729280864, "grad_norm": 0.21368801593780518, "learning_rate": 6.246420682320217e-06, "loss": 0.0046, "step": 15270 }, { "epoch": 0.2500204532438845, "grad_norm": 0.34250885248184204, "learning_rate": 6.250511331097113e-06, "loss": 0.0052, "step": 15280 }, { "epoch": 0.25018407919496033, "grad_norm": 0.23057936131954193, "learning_rate": 6.254601979874009e-06, "loss": 0.0043, "step": 15290 }, { "epoch": 0.25034770514603616, "grad_norm": 0.39956244826316833, "learning_rate": 6.258692628650904e-06, "loss": 0.0055, "step": 15300 }, { "epoch": 0.250511331097112, "grad_norm": 0.12587584555149078, "learning_rate": 6.2627832774278006e-06, "loss": 0.0043, "step": 15310 }, { "epoch": 0.2506749570481878, "grad_norm": 0.18822215497493744, "learning_rate": 6.266873926204696e-06, "loss": 0.0045, "step": 15320 }, { "epoch": 0.2508385829992637, "grad_norm": 0.1432047188282013, "learning_rate": 6.270964574981593e-06, "loss": 0.0046, "step": 15330 }, { "epoch": 0.25100220895033953, "grad_norm": 0.22165043652057648, "learning_rate": 6.2750552237584885e-06, "loss": 0.0043, "step": 15340 }, { "epoch": 0.25116583490141536, "grad_norm": 0.5780850648880005, "learning_rate": 6.279145872535385e-06, "loss": 0.0057, "step": 15350 }, { "epoch": 0.2513294608524912, "grad_norm": 0.6724796891212463, "learning_rate": 6.28323652131228e-06, "loss": 0.0078, "step": 15360 }, { "epoch": 0.251493086803567, "grad_norm": 0.3123752772808075, "learning_rate": 6.287327170089177e-06, "loss": 0.0043, "step": 15370 }, { "epoch": 0.2516567127546429, "grad_norm": 0.2853545844554901, "learning_rate": 6.291417818866072e-06, "loss": 0.0051, "step": 15380 }, { "epoch": 0.25182033870571874, "grad_norm": 0.3671239912509918, "learning_rate": 6.295508467642969e-06, "loss": 0.0067, "step": 15390 }, { "epoch": 0.25198396465679457, "grad_norm": 0.1831258088350296, "learning_rate": 6.2995991164198645e-06, "loss": 0.0068, "step": 15400 }, { "epoch": 0.2521475906078704, "grad_norm": 0.12181084603071213, "learning_rate": 6.303689765196761e-06, "loss": 0.0043, "step": 15410 }, { "epoch": 0.2523112165589462, "grad_norm": 0.1676037460565567, "learning_rate": 6.307780413973657e-06, "loss": 0.0061, "step": 15420 }, { "epoch": 0.2524748425100221, "grad_norm": 0.5814197659492493, "learning_rate": 6.311871062750553e-06, "loss": 0.0037, "step": 15430 }, { "epoch": 0.25263846846109794, "grad_norm": 0.15786194801330566, "learning_rate": 6.315961711527448e-06, "loss": 0.0058, "step": 15440 }, { "epoch": 0.2528020944121738, "grad_norm": 0.25556448101997375, "learning_rate": 6.320052360304345e-06, "loss": 0.0041, "step": 15450 }, { "epoch": 0.2529657203632496, "grad_norm": 0.3170509338378906, "learning_rate": 6.32414300908124e-06, "loss": 0.0084, "step": 15460 }, { "epoch": 0.25312934631432543, "grad_norm": 0.30223548412323, "learning_rate": 6.328233657858137e-06, "loss": 0.0031, "step": 15470 }, { "epoch": 0.2532929722654013, "grad_norm": 0.3219361901283264, "learning_rate": 6.332324306635033e-06, "loss": 0.0074, "step": 15480 }, { "epoch": 0.25345659821647715, "grad_norm": 0.47689446806907654, "learning_rate": 6.336414955411929e-06, "loss": 0.0098, "step": 15490 }, { "epoch": 0.253620224167553, "grad_norm": 0.26821568608283997, "learning_rate": 6.340505604188825e-06, "loss": 0.0089, "step": 15500 }, { "epoch": 0.2537838501186288, "grad_norm": 0.43441054224967957, "learning_rate": 6.3445962529657215e-06, "loss": 0.0067, "step": 15510 }, { "epoch": 0.25394747606970464, "grad_norm": 0.14736980199813843, "learning_rate": 6.348686901742616e-06, "loss": 0.0045, "step": 15520 }, { "epoch": 0.2541111020207805, "grad_norm": 0.31045329570770264, "learning_rate": 6.352777550519513e-06, "loss": 0.0047, "step": 15530 }, { "epoch": 0.25427472797185635, "grad_norm": 0.14488083124160767, "learning_rate": 6.356868199296409e-06, "loss": 0.0043, "step": 15540 }, { "epoch": 0.2544383539229322, "grad_norm": 0.18014821410179138, "learning_rate": 6.360958848073305e-06, "loss": 0.0094, "step": 15550 }, { "epoch": 0.254601979874008, "grad_norm": 0.34195029735565186, "learning_rate": 6.365049496850201e-06, "loss": 0.0059, "step": 15560 }, { "epoch": 0.25476560582508384, "grad_norm": 0.3033911883831024, "learning_rate": 6.3691401456270974e-06, "loss": 0.0037, "step": 15570 }, { "epoch": 0.2549292317761597, "grad_norm": 0.34467455744743347, "learning_rate": 6.373230794403993e-06, "loss": 0.0047, "step": 15580 }, { "epoch": 0.25509285772723556, "grad_norm": 0.3673767149448395, "learning_rate": 6.37732144318089e-06, "loss": 0.0055, "step": 15590 }, { "epoch": 0.2552564836783114, "grad_norm": 0.38794153928756714, "learning_rate": 6.3814120919577845e-06, "loss": 0.0052, "step": 15600 }, { "epoch": 0.2554201096293872, "grad_norm": 0.39927300810813904, "learning_rate": 6.385502740734681e-06, "loss": 0.0064, "step": 15610 }, { "epoch": 0.25558373558046305, "grad_norm": 0.15061934292316437, "learning_rate": 6.389593389511577e-06, "loss": 0.005, "step": 15620 }, { "epoch": 0.2557473615315389, "grad_norm": 0.2666986882686615, "learning_rate": 6.393684038288473e-06, "loss": 0.0039, "step": 15630 }, { "epoch": 0.25591098748261476, "grad_norm": 0.2232005000114441, "learning_rate": 6.397774687065369e-06, "loss": 0.0036, "step": 15640 }, { "epoch": 0.2560746134336906, "grad_norm": 0.28430745005607605, "learning_rate": 6.401865335842266e-06, "loss": 0.0058, "step": 15650 }, { "epoch": 0.2562382393847664, "grad_norm": 0.20765452086925507, "learning_rate": 6.405955984619161e-06, "loss": 0.0036, "step": 15660 }, { "epoch": 0.25640186533584225, "grad_norm": 0.12396921962499619, "learning_rate": 6.410046633396058e-06, "loss": 0.0061, "step": 15670 }, { "epoch": 0.2565654912869181, "grad_norm": 0.03155027702450752, "learning_rate": 6.414137282172953e-06, "loss": 0.0043, "step": 15680 }, { "epoch": 0.25672911723799396, "grad_norm": 0.08579002320766449, "learning_rate": 6.418227930949849e-06, "loss": 0.0047, "step": 15690 }, { "epoch": 0.2568927431890698, "grad_norm": 0.2501731812953949, "learning_rate": 6.422318579726745e-06, "loss": 0.0049, "step": 15700 }, { "epoch": 0.2570563691401456, "grad_norm": 0.1593310534954071, "learning_rate": 6.4264092285036416e-06, "loss": 0.0056, "step": 15710 }, { "epoch": 0.25721999509122145, "grad_norm": 0.2098381370306015, "learning_rate": 6.430499877280537e-06, "loss": 0.0068, "step": 15720 }, { "epoch": 0.2573836210422973, "grad_norm": 0.1671796441078186, "learning_rate": 6.434590526057434e-06, "loss": 0.0071, "step": 15730 }, { "epoch": 0.25754724699337317, "grad_norm": 0.1863090693950653, "learning_rate": 6.4386811748343295e-06, "loss": 0.0097, "step": 15740 }, { "epoch": 0.257710872944449, "grad_norm": 0.28753289580345154, "learning_rate": 6.442771823611226e-06, "loss": 0.006, "step": 15750 }, { "epoch": 0.25787449889552483, "grad_norm": 0.04778561368584633, "learning_rate": 6.446862472388121e-06, "loss": 0.0041, "step": 15760 }, { "epoch": 0.25803812484660066, "grad_norm": 0.26718056201934814, "learning_rate": 6.4509531211650175e-06, "loss": 0.0053, "step": 15770 }, { "epoch": 0.2582017507976765, "grad_norm": 0.15029019117355347, "learning_rate": 6.455043769941913e-06, "loss": 0.004, "step": 15780 }, { "epoch": 0.2583653767487524, "grad_norm": 0.19623981416225433, "learning_rate": 6.45913441871881e-06, "loss": 0.0031, "step": 15790 }, { "epoch": 0.2585290026998282, "grad_norm": 0.3236330449581146, "learning_rate": 6.4632250674957055e-06, "loss": 0.0044, "step": 15800 }, { "epoch": 0.25869262865090403, "grad_norm": 0.025501620024442673, "learning_rate": 6.467315716272602e-06, "loss": 0.0041, "step": 15810 }, { "epoch": 0.25885625460197986, "grad_norm": 0.1859510838985443, "learning_rate": 6.471406365049498e-06, "loss": 0.004, "step": 15820 }, { "epoch": 0.2590198805530557, "grad_norm": 0.2503608763217926, "learning_rate": 6.475497013826393e-06, "loss": 0.0037, "step": 15830 }, { "epoch": 0.2591835065041316, "grad_norm": 0.23659338057041168, "learning_rate": 6.479587662603289e-06, "loss": 0.0035, "step": 15840 }, { "epoch": 0.2593471324552074, "grad_norm": 0.17410370707511902, "learning_rate": 6.483678311380185e-06, "loss": 0.0051, "step": 15850 }, { "epoch": 0.25951075840628324, "grad_norm": 0.13967087864875793, "learning_rate": 6.4877689601570814e-06, "loss": 0.0037, "step": 15860 }, { "epoch": 0.25967438435735907, "grad_norm": 0.06533373147249222, "learning_rate": 6.491859608933977e-06, "loss": 0.0042, "step": 15870 }, { "epoch": 0.2598380103084349, "grad_norm": 0.22366955876350403, "learning_rate": 6.495950257710874e-06, "loss": 0.0061, "step": 15880 }, { "epoch": 0.2600016362595108, "grad_norm": 0.30437174439430237, "learning_rate": 6.500040906487769e-06, "loss": 0.0041, "step": 15890 }, { "epoch": 0.2601652622105866, "grad_norm": 0.31927600502967834, "learning_rate": 6.504131555264666e-06, "loss": 0.0046, "step": 15900 }, { "epoch": 0.26032888816166244, "grad_norm": 0.2994828224182129, "learning_rate": 6.508222204041561e-06, "loss": 0.0048, "step": 15910 }, { "epoch": 0.26049251411273827, "grad_norm": 0.10696202516555786, "learning_rate": 6.512312852818457e-06, "loss": 0.0082, "step": 15920 }, { "epoch": 0.2606561400638141, "grad_norm": 0.292820006608963, "learning_rate": 6.516403501595353e-06, "loss": 0.0073, "step": 15930 }, { "epoch": 0.26081976601489, "grad_norm": 0.13571535050868988, "learning_rate": 6.52049415037225e-06, "loss": 0.0034, "step": 15940 }, { "epoch": 0.2609833919659658, "grad_norm": 0.13646425306797028, "learning_rate": 6.524584799149145e-06, "loss": 0.0035, "step": 15950 }, { "epoch": 0.26114701791704165, "grad_norm": 0.7637732028961182, "learning_rate": 6.528675447926042e-06, "loss": 0.0117, "step": 15960 }, { "epoch": 0.2613106438681175, "grad_norm": 0.24349670112133026, "learning_rate": 6.532766096702937e-06, "loss": 0.0068, "step": 15970 }, { "epoch": 0.2614742698191933, "grad_norm": 0.19349224865436554, "learning_rate": 6.536856745479834e-06, "loss": 0.0045, "step": 15980 }, { "epoch": 0.2616378957702692, "grad_norm": 0.23084218800067902, "learning_rate": 6.540947394256729e-06, "loss": 0.0064, "step": 15990 }, { "epoch": 0.261801521721345, "grad_norm": 0.08095081895589828, "learning_rate": 6.5450380430336256e-06, "loss": 0.0056, "step": 16000 }, { "epoch": 0.26196514767242085, "grad_norm": 0.15526416897773743, "learning_rate": 6.549128691810521e-06, "loss": 0.006, "step": 16010 }, { "epoch": 0.2621287736234967, "grad_norm": 0.18197907507419586, "learning_rate": 6.553219340587418e-06, "loss": 0.0047, "step": 16020 }, { "epoch": 0.2622923995745725, "grad_norm": 0.12251053750514984, "learning_rate": 6.5573099893643135e-06, "loss": 0.008, "step": 16030 }, { "epoch": 0.2624560255256484, "grad_norm": 0.2623133361339569, "learning_rate": 6.56140063814121e-06, "loss": 0.0054, "step": 16040 }, { "epoch": 0.2626196514767242, "grad_norm": 0.25632718205451965, "learning_rate": 6.565491286918105e-06, "loss": 0.006, "step": 16050 }, { "epoch": 0.26278327742780005, "grad_norm": 0.4878097474575043, "learning_rate": 6.5695819356950015e-06, "loss": 0.0056, "step": 16060 }, { "epoch": 0.2629469033788759, "grad_norm": 0.23200906813144684, "learning_rate": 6.573672584471897e-06, "loss": 0.0039, "step": 16070 }, { "epoch": 0.2631105293299517, "grad_norm": 0.3379689157009125, "learning_rate": 6.577763233248794e-06, "loss": 0.0036, "step": 16080 }, { "epoch": 0.26327415528102754, "grad_norm": 0.613802969455719, "learning_rate": 6.5818538820256895e-06, "loss": 0.006, "step": 16090 }, { "epoch": 0.26343778123210343, "grad_norm": 0.3824259042739868, "learning_rate": 6.585944530802586e-06, "loss": 0.0064, "step": 16100 }, { "epoch": 0.26360140718317926, "grad_norm": 0.2794356644153595, "learning_rate": 6.590035179579482e-06, "loss": 0.0064, "step": 16110 }, { "epoch": 0.2637650331342551, "grad_norm": 0.13561318814754486, "learning_rate": 6.594125828356378e-06, "loss": 0.0053, "step": 16120 }, { "epoch": 0.2639286590853309, "grad_norm": 0.1534125953912735, "learning_rate": 6.598216477133273e-06, "loss": 0.0048, "step": 16130 }, { "epoch": 0.26409228503640675, "grad_norm": 0.4351784586906433, "learning_rate": 6.60230712591017e-06, "loss": 0.0039, "step": 16140 }, { "epoch": 0.26425591098748263, "grad_norm": 0.07826736569404602, "learning_rate": 6.6063977746870654e-06, "loss": 0.0051, "step": 16150 }, { "epoch": 0.26441953693855846, "grad_norm": 0.3567856550216675, "learning_rate": 6.610488423463962e-06, "loss": 0.0042, "step": 16160 }, { "epoch": 0.2645831628896343, "grad_norm": 0.22316698729991913, "learning_rate": 6.614579072240858e-06, "loss": 0.0033, "step": 16170 }, { "epoch": 0.2647467888407101, "grad_norm": 0.32276371121406555, "learning_rate": 6.618669721017754e-06, "loss": 0.0075, "step": 16180 }, { "epoch": 0.26491041479178595, "grad_norm": 0.19045418500900269, "learning_rate": 6.62276036979465e-06, "loss": 0.0111, "step": 16190 }, { "epoch": 0.26507404074286184, "grad_norm": 0.09470469504594803, "learning_rate": 6.6268510185715465e-06, "loss": 0.0026, "step": 16200 }, { "epoch": 0.26523766669393767, "grad_norm": 0.2835683226585388, "learning_rate": 6.630941667348441e-06, "loss": 0.0049, "step": 16210 }, { "epoch": 0.2654012926450135, "grad_norm": 0.2065233588218689, "learning_rate": 6.635032316125338e-06, "loss": 0.0039, "step": 16220 }, { "epoch": 0.2655649185960893, "grad_norm": 0.2749060094356537, "learning_rate": 6.639122964902234e-06, "loss": 0.0067, "step": 16230 }, { "epoch": 0.26572854454716516, "grad_norm": 0.13738185167312622, "learning_rate": 6.64321361367913e-06, "loss": 0.0055, "step": 16240 }, { "epoch": 0.26589217049824104, "grad_norm": 0.1671639382839203, "learning_rate": 6.647304262456026e-06, "loss": 0.0038, "step": 16250 }, { "epoch": 0.2660557964493169, "grad_norm": 0.18006369471549988, "learning_rate": 6.6513949112329224e-06, "loss": 0.0033, "step": 16260 }, { "epoch": 0.2662194224003927, "grad_norm": 0.3271634578704834, "learning_rate": 6.655485560009818e-06, "loss": 0.0062, "step": 16270 }, { "epoch": 0.26638304835146853, "grad_norm": 0.2002531886100769, "learning_rate": 6.659576208786715e-06, "loss": 0.0034, "step": 16280 }, { "epoch": 0.26654667430254436, "grad_norm": 0.2232416570186615, "learning_rate": 6.6636668575636096e-06, "loss": 0.0045, "step": 16290 }, { "epoch": 0.26671030025362025, "grad_norm": 0.04324057325720787, "learning_rate": 6.667757506340506e-06, "loss": 0.0037, "step": 16300 }, { "epoch": 0.2668739262046961, "grad_norm": 0.2106238454580307, "learning_rate": 6.671848155117402e-06, "loss": 0.0041, "step": 16310 }, { "epoch": 0.2670375521557719, "grad_norm": 0.17994728684425354, "learning_rate": 6.675938803894298e-06, "loss": 0.0051, "step": 16320 }, { "epoch": 0.26720117810684774, "grad_norm": 0.10701189190149307, "learning_rate": 6.680029452671194e-06, "loss": 0.006, "step": 16330 }, { "epoch": 0.26736480405792357, "grad_norm": 0.3325197100639343, "learning_rate": 6.684120101448091e-06, "loss": 0.0065, "step": 16340 }, { "epoch": 0.26752843000899945, "grad_norm": 0.09425511956214905, "learning_rate": 6.688210750224986e-06, "loss": 0.0037, "step": 16350 }, { "epoch": 0.2676920559600753, "grad_norm": 0.21500110626220703, "learning_rate": 6.692301399001883e-06, "loss": 0.0067, "step": 16360 }, { "epoch": 0.2678556819111511, "grad_norm": 0.14832596480846405, "learning_rate": 6.696392047778778e-06, "loss": 0.0055, "step": 16370 }, { "epoch": 0.26801930786222694, "grad_norm": 0.1404012143611908, "learning_rate": 6.700482696555674e-06, "loss": 0.006, "step": 16380 }, { "epoch": 0.26818293381330277, "grad_norm": 0.22498805820941925, "learning_rate": 6.70457334533257e-06, "loss": 0.0095, "step": 16390 }, { "epoch": 0.26834655976437866, "grad_norm": 0.19096927344799042, "learning_rate": 6.708663994109467e-06, "loss": 0.0044, "step": 16400 }, { "epoch": 0.2685101857154545, "grad_norm": 0.35353168845176697, "learning_rate": 6.712754642886362e-06, "loss": 0.0038, "step": 16410 }, { "epoch": 0.2686738116665303, "grad_norm": 0.4161711633205414, "learning_rate": 6.716845291663259e-06, "loss": 0.0077, "step": 16420 }, { "epoch": 0.26883743761760615, "grad_norm": 0.16533902287483215, "learning_rate": 6.7209359404401546e-06, "loss": 0.0045, "step": 16430 }, { "epoch": 0.269001063568682, "grad_norm": 0.5903836488723755, "learning_rate": 6.725026589217051e-06, "loss": 0.0055, "step": 16440 }, { "epoch": 0.26916468951975786, "grad_norm": 0.25705355405807495, "learning_rate": 6.729117237993946e-06, "loss": 0.005, "step": 16450 }, { "epoch": 0.2693283154708337, "grad_norm": 0.2878263294696808, "learning_rate": 6.7332078867708425e-06, "loss": 0.0049, "step": 16460 }, { "epoch": 0.2694919414219095, "grad_norm": 0.522221565246582, "learning_rate": 6.737298535547738e-06, "loss": 0.0076, "step": 16470 }, { "epoch": 0.26965556737298535, "grad_norm": 0.32265564799308777, "learning_rate": 6.741389184324635e-06, "loss": 0.0074, "step": 16480 }, { "epoch": 0.2698191933240612, "grad_norm": 0.08602648973464966, "learning_rate": 6.7454798331015305e-06, "loss": 0.0047, "step": 16490 }, { "epoch": 0.269982819275137, "grad_norm": 0.18631140887737274, "learning_rate": 6.749570481878427e-06, "loss": 0.0051, "step": 16500 }, { "epoch": 0.2701464452262129, "grad_norm": 0.2951951324939728, "learning_rate": 6.753661130655323e-06, "loss": 0.0063, "step": 16510 }, { "epoch": 0.2703100711772887, "grad_norm": 0.12326254695653915, "learning_rate": 6.757751779432219e-06, "loss": 0.006, "step": 16520 }, { "epoch": 0.27047369712836455, "grad_norm": 0.3129864037036896, "learning_rate": 6.761842428209114e-06, "loss": 0.0051, "step": 16530 }, { "epoch": 0.2706373230794404, "grad_norm": 0.411050945520401, "learning_rate": 6.765933076986011e-06, "loss": 0.0036, "step": 16540 }, { "epoch": 0.2708009490305162, "grad_norm": 0.3037816882133484, "learning_rate": 6.7700237257629064e-06, "loss": 0.0101, "step": 16550 }, { "epoch": 0.2709645749815921, "grad_norm": 0.21663936972618103, "learning_rate": 6.774114374539803e-06, "loss": 0.0069, "step": 16560 }, { "epoch": 0.27112820093266793, "grad_norm": 0.1212368831038475, "learning_rate": 6.778205023316699e-06, "loss": 0.0045, "step": 16570 }, { "epoch": 0.27129182688374376, "grad_norm": 0.23089103400707245, "learning_rate": 6.782295672093595e-06, "loss": 0.0054, "step": 16580 }, { "epoch": 0.2714554528348196, "grad_norm": 0.08057130128145218, "learning_rate": 6.78638632087049e-06, "loss": 0.0041, "step": 16590 }, { "epoch": 0.2716190787858954, "grad_norm": 0.5196241140365601, "learning_rate": 6.7904769696473875e-06, "loss": 0.0051, "step": 16600 }, { "epoch": 0.2717827047369713, "grad_norm": 0.2149159014225006, "learning_rate": 6.794567618424282e-06, "loss": 0.0064, "step": 16610 }, { "epoch": 0.27194633068804713, "grad_norm": 0.12387235462665558, "learning_rate": 6.798658267201179e-06, "loss": 0.0037, "step": 16620 }, { "epoch": 0.27210995663912296, "grad_norm": 0.0738256424665451, "learning_rate": 6.802748915978075e-06, "loss": 0.0035, "step": 16630 }, { "epoch": 0.2722735825901988, "grad_norm": 0.19157706201076508, "learning_rate": 6.806839564754971e-06, "loss": 0.005, "step": 16640 }, { "epoch": 0.2724372085412746, "grad_norm": 0.2832675576210022, "learning_rate": 6.810930213531867e-06, "loss": 0.0047, "step": 16650 }, { "epoch": 0.2726008344923505, "grad_norm": 0.20407840609550476, "learning_rate": 6.8150208623087635e-06, "loss": 0.0065, "step": 16660 }, { "epoch": 0.27276446044342634, "grad_norm": 0.22679053246974945, "learning_rate": 6.819111511085658e-06, "loss": 0.005, "step": 16670 }, { "epoch": 0.27292808639450217, "grad_norm": 0.15578868985176086, "learning_rate": 6.823202159862555e-06, "loss": 0.0031, "step": 16680 }, { "epoch": 0.273091712345578, "grad_norm": 0.5440335869789124, "learning_rate": 6.827292808639451e-06, "loss": 0.0056, "step": 16690 }, { "epoch": 0.2732553382966538, "grad_norm": 0.1290455013513565, "learning_rate": 6.831383457416347e-06, "loss": 0.0068, "step": 16700 }, { "epoch": 0.2734189642477297, "grad_norm": 0.154061421751976, "learning_rate": 6.835474106193243e-06, "loss": 0.005, "step": 16710 }, { "epoch": 0.27358259019880554, "grad_norm": 0.3540516793727875, "learning_rate": 6.8395647549701386e-06, "loss": 0.0038, "step": 16720 }, { "epoch": 0.27374621614988137, "grad_norm": 0.31063932180404663, "learning_rate": 6.843655403747035e-06, "loss": 0.0084, "step": 16730 }, { "epoch": 0.2739098421009572, "grad_norm": 0.2314666509628296, "learning_rate": 6.84774605252393e-06, "loss": 0.0051, "step": 16740 }, { "epoch": 0.27407346805203303, "grad_norm": 0.43002989888191223, "learning_rate": 6.8518367013008265e-06, "loss": 0.0038, "step": 16750 }, { "epoch": 0.2742370940031089, "grad_norm": 0.4898504912853241, "learning_rate": 6.855927350077722e-06, "loss": 0.0052, "step": 16760 }, { "epoch": 0.27440071995418475, "grad_norm": 0.12422274053096771, "learning_rate": 6.860017998854619e-06, "loss": 0.0064, "step": 16770 }, { "epoch": 0.2745643459052606, "grad_norm": 0.1528148204088211, "learning_rate": 6.8641086476315145e-06, "loss": 0.005, "step": 16780 }, { "epoch": 0.2747279718563364, "grad_norm": 0.3730817139148712, "learning_rate": 6.868199296408411e-06, "loss": 0.0059, "step": 16790 }, { "epoch": 0.27489159780741224, "grad_norm": 0.23787443339824677, "learning_rate": 6.872289945185307e-06, "loss": 0.0047, "step": 16800 }, { "epoch": 0.2750552237584881, "grad_norm": 0.20724879205226898, "learning_rate": 6.876380593962203e-06, "loss": 0.0035, "step": 16810 }, { "epoch": 0.27521884970956395, "grad_norm": 0.07592567056417465, "learning_rate": 6.880471242739098e-06, "loss": 0.0045, "step": 16820 }, { "epoch": 0.2753824756606398, "grad_norm": 0.2696102261543274, "learning_rate": 6.884561891515995e-06, "loss": 0.0041, "step": 16830 }, { "epoch": 0.2755461016117156, "grad_norm": 0.1510450541973114, "learning_rate": 6.8886525402928904e-06, "loss": 0.0073, "step": 16840 }, { "epoch": 0.27570972756279144, "grad_norm": 0.07801207154989243, "learning_rate": 6.892743189069787e-06, "loss": 0.0027, "step": 16850 }, { "epoch": 0.2758733535138673, "grad_norm": 0.22183087468147278, "learning_rate": 6.896833837846683e-06, "loss": 0.0069, "step": 16860 }, { "epoch": 0.27603697946494316, "grad_norm": 0.0752909928560257, "learning_rate": 6.900924486623579e-06, "loss": 0.0051, "step": 16870 }, { "epoch": 0.276200605416019, "grad_norm": 0.24921782314777374, "learning_rate": 6.905015135400475e-06, "loss": 0.0055, "step": 16880 }, { "epoch": 0.2763642313670948, "grad_norm": 0.3096201717853546, "learning_rate": 6.9091057841773715e-06, "loss": 0.0034, "step": 16890 }, { "epoch": 0.27652785731817064, "grad_norm": 0.533718466758728, "learning_rate": 6.913196432954266e-06, "loss": 0.0073, "step": 16900 }, { "epoch": 0.27669148326924653, "grad_norm": 0.2875325381755829, "learning_rate": 6.917287081731163e-06, "loss": 0.0054, "step": 16910 }, { "epoch": 0.27685510922032236, "grad_norm": 0.2637481987476349, "learning_rate": 6.921377730508059e-06, "loss": 0.0064, "step": 16920 }, { "epoch": 0.2770187351713982, "grad_norm": 0.28419989347457886, "learning_rate": 6.925468379284955e-06, "loss": 0.004, "step": 16930 }, { "epoch": 0.277182361122474, "grad_norm": 0.14696654677391052, "learning_rate": 6.929559028061851e-06, "loss": 0.0051, "step": 16940 }, { "epoch": 0.27734598707354985, "grad_norm": 0.2792978882789612, "learning_rate": 6.9336496768387475e-06, "loss": 0.0064, "step": 16950 }, { "epoch": 0.2775096130246257, "grad_norm": 0.20526666939258575, "learning_rate": 6.937740325615643e-06, "loss": 0.0066, "step": 16960 }, { "epoch": 0.27767323897570156, "grad_norm": 0.7601214647293091, "learning_rate": 6.94183097439254e-06, "loss": 0.0049, "step": 16970 }, { "epoch": 0.2778368649267774, "grad_norm": 0.045588064938783646, "learning_rate": 6.945921623169435e-06, "loss": 0.0065, "step": 16980 }, { "epoch": 0.2780004908778532, "grad_norm": 0.23847539722919464, "learning_rate": 6.950012271946331e-06, "loss": 0.0046, "step": 16990 }, { "epoch": 0.27816411682892905, "grad_norm": 0.33378732204437256, "learning_rate": 6.954102920723227e-06, "loss": 0.0038, "step": 17000 }, { "epoch": 0.2783277427800049, "grad_norm": 0.14488115906715393, "learning_rate": 6.958193569500123e-06, "loss": 0.0062, "step": 17010 }, { "epoch": 0.27849136873108077, "grad_norm": 0.15549837052822113, "learning_rate": 6.962284218277019e-06, "loss": 0.0093, "step": 17020 }, { "epoch": 0.2786549946821566, "grad_norm": 0.18969117105007172, "learning_rate": 6.966374867053916e-06, "loss": 0.0028, "step": 17030 }, { "epoch": 0.27881862063323243, "grad_norm": 0.031048908829689026, "learning_rate": 6.970465515830811e-06, "loss": 0.0039, "step": 17040 }, { "epoch": 0.27898224658430826, "grad_norm": 0.17645825445652008, "learning_rate": 6.974556164607708e-06, "loss": 0.004, "step": 17050 }, { "epoch": 0.2791458725353841, "grad_norm": 0.2408183217048645, "learning_rate": 6.978646813384603e-06, "loss": 0.0049, "step": 17060 }, { "epoch": 0.27930949848646, "grad_norm": 0.16109289228916168, "learning_rate": 6.982737462161499e-06, "loss": 0.0039, "step": 17070 }, { "epoch": 0.2794731244375358, "grad_norm": 0.1018131896853447, "learning_rate": 6.986828110938395e-06, "loss": 0.0056, "step": 17080 }, { "epoch": 0.27963675038861163, "grad_norm": 0.14227068424224854, "learning_rate": 6.990918759715292e-06, "loss": 0.0029, "step": 17090 }, { "epoch": 0.27980037633968746, "grad_norm": 0.13403025269508362, "learning_rate": 6.995009408492187e-06, "loss": 0.0059, "step": 17100 }, { "epoch": 0.2799640022907633, "grad_norm": 0.8118295669555664, "learning_rate": 6.999100057269084e-06, "loss": 0.0045, "step": 17110 }, { "epoch": 0.2801276282418392, "grad_norm": 0.1951279491186142, "learning_rate": 7.00319070604598e-06, "loss": 0.0093, "step": 17120 }, { "epoch": 0.280291254192915, "grad_norm": 0.24189841747283936, "learning_rate": 7.007281354822876e-06, "loss": 0.0044, "step": 17130 }, { "epoch": 0.28045488014399084, "grad_norm": 0.22562536597251892, "learning_rate": 7.011372003599771e-06, "loss": 0.0037, "step": 17140 }, { "epoch": 0.28061850609506667, "grad_norm": 0.18976938724517822, "learning_rate": 7.0154626523766676e-06, "loss": 0.0089, "step": 17150 }, { "epoch": 0.2807821320461425, "grad_norm": 0.24779732525348663, "learning_rate": 7.019553301153563e-06, "loss": 0.0055, "step": 17160 }, { "epoch": 0.2809457579972184, "grad_norm": 0.1888822317123413, "learning_rate": 7.02364394993046e-06, "loss": 0.0054, "step": 17170 }, { "epoch": 0.2811093839482942, "grad_norm": 0.6627678871154785, "learning_rate": 7.0277345987073555e-06, "loss": 0.0065, "step": 17180 }, { "epoch": 0.28127300989937004, "grad_norm": 0.27115777134895325, "learning_rate": 7.031825247484252e-06, "loss": 0.0054, "step": 17190 }, { "epoch": 0.28143663585044587, "grad_norm": 0.25892120599746704, "learning_rate": 7.035915896261147e-06, "loss": 0.0069, "step": 17200 }, { "epoch": 0.2816002618015217, "grad_norm": 0.30582934617996216, "learning_rate": 7.040006545038044e-06, "loss": 0.0049, "step": 17210 }, { "epoch": 0.2817638877525976, "grad_norm": 0.26105257868766785, "learning_rate": 7.044097193814939e-06, "loss": 0.0061, "step": 17220 }, { "epoch": 0.2819275137036734, "grad_norm": 0.6026048064231873, "learning_rate": 7.048187842591836e-06, "loss": 0.0046, "step": 17230 }, { "epoch": 0.28209113965474925, "grad_norm": 0.14918570220470428, "learning_rate": 7.0522784913687315e-06, "loss": 0.0049, "step": 17240 }, { "epoch": 0.2822547656058251, "grad_norm": 0.21756070852279663, "learning_rate": 7.056369140145628e-06, "loss": 0.0047, "step": 17250 }, { "epoch": 0.2824183915569009, "grad_norm": 0.3691888749599457, "learning_rate": 7.060459788922524e-06, "loss": 0.0052, "step": 17260 }, { "epoch": 0.2825820175079768, "grad_norm": 0.09177648276090622, "learning_rate": 7.06455043769942e-06, "loss": 0.0057, "step": 17270 }, { "epoch": 0.2827456434590526, "grad_norm": 0.2515797019004822, "learning_rate": 7.068641086476315e-06, "loss": 0.0048, "step": 17280 }, { "epoch": 0.28290926941012845, "grad_norm": 0.10462643206119537, "learning_rate": 7.072731735253212e-06, "loss": 0.0055, "step": 17290 }, { "epoch": 0.2830728953612043, "grad_norm": 0.0832565426826477, "learning_rate": 7.076822384030107e-06, "loss": 0.0045, "step": 17300 }, { "epoch": 0.2832365213122801, "grad_norm": 0.19747138023376465, "learning_rate": 7.080913032807004e-06, "loss": 0.0036, "step": 17310 }, { "epoch": 0.283400147263356, "grad_norm": 0.16253244876861572, "learning_rate": 7.0850036815839e-06, "loss": 0.0044, "step": 17320 }, { "epoch": 0.2835637732144318, "grad_norm": 0.17164021730422974, "learning_rate": 7.089094330360796e-06, "loss": 0.0047, "step": 17330 }, { "epoch": 0.28372739916550765, "grad_norm": 0.22622981667518616, "learning_rate": 7.093184979137692e-06, "loss": 0.0054, "step": 17340 }, { "epoch": 0.2838910251165835, "grad_norm": 0.19495829939842224, "learning_rate": 7.0972756279145885e-06, "loss": 0.0036, "step": 17350 }, { "epoch": 0.2840546510676593, "grad_norm": 0.4779374599456787, "learning_rate": 7.101366276691483e-06, "loss": 0.0046, "step": 17360 }, { "epoch": 0.28421827701873514, "grad_norm": 0.13583843410015106, "learning_rate": 7.10545692546838e-06, "loss": 0.004, "step": 17370 }, { "epoch": 0.28438190296981103, "grad_norm": 0.26904454827308655, "learning_rate": 7.109547574245276e-06, "loss": 0.0042, "step": 17380 }, { "epoch": 0.28454552892088686, "grad_norm": 0.09131120145320892, "learning_rate": 7.113638223022172e-06, "loss": 0.0057, "step": 17390 }, { "epoch": 0.2847091548719627, "grad_norm": 0.09867345541715622, "learning_rate": 7.117728871799068e-06, "loss": 0.005, "step": 17400 }, { "epoch": 0.2848727808230385, "grad_norm": 0.09620248526334763, "learning_rate": 7.1218195205759644e-06, "loss": 0.0041, "step": 17410 }, { "epoch": 0.28503640677411435, "grad_norm": 0.1676165908575058, "learning_rate": 7.12591016935286e-06, "loss": 0.0055, "step": 17420 }, { "epoch": 0.28520003272519023, "grad_norm": 0.23453105986118317, "learning_rate": 7.130000818129757e-06, "loss": 0.0054, "step": 17430 }, { "epoch": 0.28536365867626606, "grad_norm": 0.08915801346302032, "learning_rate": 7.1340914669066515e-06, "loss": 0.0043, "step": 17440 }, { "epoch": 0.2855272846273419, "grad_norm": 0.285888671875, "learning_rate": 7.138182115683548e-06, "loss": 0.0063, "step": 17450 }, { "epoch": 0.2856909105784177, "grad_norm": 0.08646760880947113, "learning_rate": 7.142272764460444e-06, "loss": 0.0066, "step": 17460 }, { "epoch": 0.28585453652949355, "grad_norm": 0.1747852861881256, "learning_rate": 7.14636341323734e-06, "loss": 0.0053, "step": 17470 }, { "epoch": 0.28601816248056944, "grad_norm": 0.20636416971683502, "learning_rate": 7.150454062014236e-06, "loss": 0.0031, "step": 17480 }, { "epoch": 0.28618178843164527, "grad_norm": 0.4373856484889984, "learning_rate": 7.154544710791133e-06, "loss": 0.0046, "step": 17490 }, { "epoch": 0.2863454143827211, "grad_norm": 0.08666010200977325, "learning_rate": 7.158635359568028e-06, "loss": 0.0042, "step": 17500 }, { "epoch": 0.2865090403337969, "grad_norm": 0.16355401277542114, "learning_rate": 7.162726008344925e-06, "loss": 0.0026, "step": 17510 }, { "epoch": 0.28667266628487276, "grad_norm": 0.47033005952835083, "learning_rate": 7.16681665712182e-06, "loss": 0.0083, "step": 17520 }, { "epoch": 0.28683629223594864, "grad_norm": 0.2540893852710724, "learning_rate": 7.170907305898716e-06, "loss": 0.0043, "step": 17530 }, { "epoch": 0.28699991818702447, "grad_norm": 0.2372291088104248, "learning_rate": 7.174997954675612e-06, "loss": 0.004, "step": 17540 }, { "epoch": 0.2871635441381003, "grad_norm": 0.25841793417930603, "learning_rate": 7.1790886034525086e-06, "loss": 0.0055, "step": 17550 }, { "epoch": 0.28732717008917613, "grad_norm": 0.38014817237854004, "learning_rate": 7.183179252229404e-06, "loss": 0.0076, "step": 17560 }, { "epoch": 0.28749079604025196, "grad_norm": 0.3096122443675995, "learning_rate": 7.187269901006301e-06, "loss": 0.0056, "step": 17570 }, { "epoch": 0.28765442199132785, "grad_norm": 0.2213359773159027, "learning_rate": 7.1913605497831965e-06, "loss": 0.0063, "step": 17580 }, { "epoch": 0.2878180479424037, "grad_norm": 0.15548378229141235, "learning_rate": 7.195451198560091e-06, "loss": 0.0044, "step": 17590 }, { "epoch": 0.2879816738934795, "grad_norm": 0.6089572310447693, "learning_rate": 7.199541847336988e-06, "loss": 0.0089, "step": 17600 }, { "epoch": 0.28814529984455534, "grad_norm": 0.11487782746553421, "learning_rate": 7.203632496113884e-06, "loss": 0.0041, "step": 17610 }, { "epoch": 0.28830892579563117, "grad_norm": 0.06798506528139114, "learning_rate": 7.20772314489078e-06, "loss": 0.0031, "step": 17620 }, { "epoch": 0.28847255174670705, "grad_norm": 0.11315786093473434, "learning_rate": 7.211813793667676e-06, "loss": 0.0087, "step": 17630 }, { "epoch": 0.2886361776977829, "grad_norm": 0.24774329364299774, "learning_rate": 7.2159044424445725e-06, "loss": 0.005, "step": 17640 }, { "epoch": 0.2887998036488587, "grad_norm": 0.15388353168964386, "learning_rate": 7.219995091221468e-06, "loss": 0.0037, "step": 17650 }, { "epoch": 0.28896342959993454, "grad_norm": 0.12828485667705536, "learning_rate": 7.224085739998365e-06, "loss": 0.0044, "step": 17660 }, { "epoch": 0.28912705555101037, "grad_norm": 0.14320962131023407, "learning_rate": 7.22817638877526e-06, "loss": 0.0086, "step": 17670 }, { "epoch": 0.28929068150208626, "grad_norm": 0.2661992609500885, "learning_rate": 7.232267037552156e-06, "loss": 0.0042, "step": 17680 }, { "epoch": 0.2894543074531621, "grad_norm": 0.1722348928451538, "learning_rate": 7.236357686329052e-06, "loss": 0.0043, "step": 17690 }, { "epoch": 0.2896179334042379, "grad_norm": 0.030284078791737556, "learning_rate": 7.2404483351059484e-06, "loss": 0.0038, "step": 17700 }, { "epoch": 0.28978155935531374, "grad_norm": 0.24819263815879822, "learning_rate": 7.244538983882844e-06, "loss": 0.0039, "step": 17710 }, { "epoch": 0.2899451853063896, "grad_norm": 0.30302855372428894, "learning_rate": 7.248629632659741e-06, "loss": 0.0045, "step": 17720 }, { "epoch": 0.29010881125746546, "grad_norm": 0.20657287538051605, "learning_rate": 7.252720281436636e-06, "loss": 0.0057, "step": 17730 }, { "epoch": 0.2902724372085413, "grad_norm": 0.2904598116874695, "learning_rate": 7.256810930213533e-06, "loss": 0.0055, "step": 17740 }, { "epoch": 0.2904360631596171, "grad_norm": 0.09727419167757034, "learning_rate": 7.260901578990428e-06, "loss": 0.0054, "step": 17750 }, { "epoch": 0.29059968911069295, "grad_norm": 0.23081184923648834, "learning_rate": 7.264992227767324e-06, "loss": 0.0049, "step": 17760 }, { "epoch": 0.2907633150617688, "grad_norm": 0.12650440633296967, "learning_rate": 7.26908287654422e-06, "loss": 0.0061, "step": 17770 }, { "epoch": 0.29092694101284466, "grad_norm": 0.26246377825737, "learning_rate": 7.273173525321117e-06, "loss": 0.0042, "step": 17780 }, { "epoch": 0.2910905669639205, "grad_norm": 0.07062461972236633, "learning_rate": 7.277264174098012e-06, "loss": 0.0061, "step": 17790 }, { "epoch": 0.2912541929149963, "grad_norm": 0.14631223678588867, "learning_rate": 7.281354822874909e-06, "loss": 0.0036, "step": 17800 }, { "epoch": 0.29141781886607215, "grad_norm": 0.21712523698806763, "learning_rate": 7.285445471651804e-06, "loss": 0.0049, "step": 17810 }, { "epoch": 0.291581444817148, "grad_norm": 0.2919718623161316, "learning_rate": 7.289536120428701e-06, "loss": 0.0088, "step": 17820 }, { "epoch": 0.2917450707682238, "grad_norm": 0.1402592658996582, "learning_rate": 7.293626769205596e-06, "loss": 0.0035, "step": 17830 }, { "epoch": 0.2919086967192997, "grad_norm": 0.07435724139213562, "learning_rate": 7.2977174179824926e-06, "loss": 0.0028, "step": 17840 }, { "epoch": 0.29207232267037553, "grad_norm": 0.25881344079971313, "learning_rate": 7.301808066759388e-06, "loss": 0.0038, "step": 17850 }, { "epoch": 0.29223594862145136, "grad_norm": 0.1921069473028183, "learning_rate": 7.305898715536285e-06, "loss": 0.0072, "step": 17860 }, { "epoch": 0.2923995745725272, "grad_norm": 0.33964893221855164, "learning_rate": 7.3099893643131805e-06, "loss": 0.0046, "step": 17870 }, { "epoch": 0.292563200523603, "grad_norm": 0.2901919186115265, "learning_rate": 7.314080013090077e-06, "loss": 0.0038, "step": 17880 }, { "epoch": 0.2927268264746789, "grad_norm": 0.45158663392066956, "learning_rate": 7.318170661866972e-06, "loss": 0.0056, "step": 17890 }, { "epoch": 0.29289045242575473, "grad_norm": 0.22703143954277039, "learning_rate": 7.3222613106438685e-06, "loss": 0.0048, "step": 17900 }, { "epoch": 0.29305407837683056, "grad_norm": 0.34966686367988586, "learning_rate": 7.326351959420764e-06, "loss": 0.0076, "step": 17910 }, { "epoch": 0.2932177043279064, "grad_norm": 0.22989550232887268, "learning_rate": 7.330442608197661e-06, "loss": 0.0045, "step": 17920 }, { "epoch": 0.2933813302789822, "grad_norm": 0.2388298660516739, "learning_rate": 7.3345332569745565e-06, "loss": 0.004, "step": 17930 }, { "epoch": 0.2935449562300581, "grad_norm": 0.24046891927719116, "learning_rate": 7.338623905751453e-06, "loss": 0.0054, "step": 17940 }, { "epoch": 0.29370858218113394, "grad_norm": 0.13769195973873138, "learning_rate": 7.342714554528349e-06, "loss": 0.0045, "step": 17950 }, { "epoch": 0.29387220813220977, "grad_norm": 0.13098104298114777, "learning_rate": 7.346805203305245e-06, "loss": 0.0028, "step": 17960 }, { "epoch": 0.2940358340832856, "grad_norm": 0.10277196764945984, "learning_rate": 7.35089585208214e-06, "loss": 0.0072, "step": 17970 }, { "epoch": 0.2941994600343614, "grad_norm": 0.1661752611398697, "learning_rate": 7.354986500859037e-06, "loss": 0.0047, "step": 17980 }, { "epoch": 0.2943630859854373, "grad_norm": 0.22722065448760986, "learning_rate": 7.359077149635932e-06, "loss": 0.006, "step": 17990 }, { "epoch": 0.29452671193651314, "grad_norm": 0.09039205312728882, "learning_rate": 7.363167798412829e-06, "loss": 0.0041, "step": 18000 }, { "epoch": 0.29469033788758897, "grad_norm": 0.1161094382405281, "learning_rate": 7.367258447189725e-06, "loss": 0.0037, "step": 18010 }, { "epoch": 0.2948539638386648, "grad_norm": 0.47167229652404785, "learning_rate": 7.371349095966621e-06, "loss": 0.0058, "step": 18020 }, { "epoch": 0.29501758978974063, "grad_norm": 0.18339763581752777, "learning_rate": 7.375439744743517e-06, "loss": 0.009, "step": 18030 }, { "epoch": 0.2951812157408165, "grad_norm": 0.3071402609348297, "learning_rate": 7.3795303935204135e-06, "loss": 0.0036, "step": 18040 }, { "epoch": 0.29534484169189235, "grad_norm": 0.22482460737228394, "learning_rate": 7.383621042297308e-06, "loss": 0.0036, "step": 18050 }, { "epoch": 0.2955084676429682, "grad_norm": 0.7473443746566772, "learning_rate": 7.387711691074205e-06, "loss": 0.0073, "step": 18060 }, { "epoch": 0.295672093594044, "grad_norm": 0.02185545302927494, "learning_rate": 7.391802339851101e-06, "loss": 0.0079, "step": 18070 }, { "epoch": 0.29583571954511984, "grad_norm": 0.11892666667699814, "learning_rate": 7.395892988627997e-06, "loss": 0.0048, "step": 18080 }, { "epoch": 0.2959993454961957, "grad_norm": 0.5051997900009155, "learning_rate": 7.399983637404893e-06, "loss": 0.0059, "step": 18090 }, { "epoch": 0.29616297144727155, "grad_norm": 0.44782519340515137, "learning_rate": 7.4040742861817894e-06, "loss": 0.0065, "step": 18100 }, { "epoch": 0.2963265973983474, "grad_norm": 0.17810332775115967, "learning_rate": 7.408164934958685e-06, "loss": 0.0057, "step": 18110 }, { "epoch": 0.2964902233494232, "grad_norm": 0.2504522204399109, "learning_rate": 7.412255583735582e-06, "loss": 0.0073, "step": 18120 }, { "epoch": 0.29665384930049904, "grad_norm": 0.17219947278499603, "learning_rate": 7.4163462325124766e-06, "loss": 0.0045, "step": 18130 }, { "epoch": 0.2968174752515749, "grad_norm": 0.12781786918640137, "learning_rate": 7.420436881289373e-06, "loss": 0.0065, "step": 18140 }, { "epoch": 0.29698110120265075, "grad_norm": 0.16578535735607147, "learning_rate": 7.424527530066269e-06, "loss": 0.0063, "step": 18150 }, { "epoch": 0.2971447271537266, "grad_norm": 0.15519273281097412, "learning_rate": 7.428618178843165e-06, "loss": 0.0053, "step": 18160 }, { "epoch": 0.2973083531048024, "grad_norm": 0.37240874767303467, "learning_rate": 7.432708827620061e-06, "loss": 0.0051, "step": 18170 }, { "epoch": 0.29747197905587824, "grad_norm": 0.26183152198791504, "learning_rate": 7.436799476396958e-06, "loss": 0.006, "step": 18180 }, { "epoch": 0.29763560500695413, "grad_norm": 0.18613168597221375, "learning_rate": 7.440890125173853e-06, "loss": 0.0067, "step": 18190 }, { "epoch": 0.29779923095802996, "grad_norm": 0.24026289582252502, "learning_rate": 7.44498077395075e-06, "loss": 0.0055, "step": 18200 }, { "epoch": 0.2979628569091058, "grad_norm": 0.2119658887386322, "learning_rate": 7.449071422727645e-06, "loss": 0.0043, "step": 18210 }, { "epoch": 0.2981264828601816, "grad_norm": 0.15999020636081696, "learning_rate": 7.453162071504541e-06, "loss": 0.0037, "step": 18220 }, { "epoch": 0.29829010881125745, "grad_norm": 0.12330754846334457, "learning_rate": 7.457252720281437e-06, "loss": 0.0044, "step": 18230 }, { "epoch": 0.29845373476233333, "grad_norm": 0.3837892413139343, "learning_rate": 7.461343369058334e-06, "loss": 0.0052, "step": 18240 }, { "epoch": 0.29861736071340916, "grad_norm": 0.1017920970916748, "learning_rate": 7.465434017835229e-06, "loss": 0.0034, "step": 18250 }, { "epoch": 0.298780986664485, "grad_norm": 0.2537844479084015, "learning_rate": 7.469524666612126e-06, "loss": 0.0053, "step": 18260 }, { "epoch": 0.2989446126155608, "grad_norm": 0.2478175312280655, "learning_rate": 7.4736153153890216e-06, "loss": 0.0029, "step": 18270 }, { "epoch": 0.29910823856663665, "grad_norm": 0.33007654547691345, "learning_rate": 7.477705964165918e-06, "loss": 0.0042, "step": 18280 }, { "epoch": 0.2992718645177125, "grad_norm": 0.4075591266155243, "learning_rate": 7.481796612942813e-06, "loss": 0.0049, "step": 18290 }, { "epoch": 0.29943549046878837, "grad_norm": 0.2058347463607788, "learning_rate": 7.4858872617197095e-06, "loss": 0.004, "step": 18300 }, { "epoch": 0.2995991164198642, "grad_norm": 0.14779265224933624, "learning_rate": 7.489977910496605e-06, "loss": 0.0059, "step": 18310 }, { "epoch": 0.29976274237094, "grad_norm": 0.11837135255336761, "learning_rate": 7.494068559273502e-06, "loss": 0.0057, "step": 18320 }, { "epoch": 0.29992636832201586, "grad_norm": 0.2260010540485382, "learning_rate": 7.4981592080503975e-06, "loss": 0.0052, "step": 18330 }, { "epoch": 0.3000899942730917, "grad_norm": 0.09538309276103973, "learning_rate": 7.502249856827294e-06, "loss": 0.0053, "step": 18340 }, { "epoch": 0.30025362022416757, "grad_norm": 0.34263426065444946, "learning_rate": 7.50634050560419e-06, "loss": 0.0037, "step": 18350 }, { "epoch": 0.3004172461752434, "grad_norm": 0.1547912210226059, "learning_rate": 7.510431154381086e-06, "loss": 0.0055, "step": 18360 }, { "epoch": 0.30058087212631923, "grad_norm": 0.08638154715299606, "learning_rate": 7.514521803157981e-06, "loss": 0.0048, "step": 18370 }, { "epoch": 0.30074449807739506, "grad_norm": 0.15340624749660492, "learning_rate": 7.518612451934878e-06, "loss": 0.0045, "step": 18380 }, { "epoch": 0.3009081240284709, "grad_norm": 0.2953304052352905, "learning_rate": 7.5227031007117734e-06, "loss": 0.0056, "step": 18390 }, { "epoch": 0.3010717499795468, "grad_norm": 0.30856576561927795, "learning_rate": 7.52679374948867e-06, "loss": 0.0037, "step": 18400 }, { "epoch": 0.3012353759306226, "grad_norm": 0.12718188762664795, "learning_rate": 7.530884398265566e-06, "loss": 0.0042, "step": 18410 }, { "epoch": 0.30139900188169844, "grad_norm": 0.12337753176689148, "learning_rate": 7.534975047042462e-06, "loss": 0.0049, "step": 18420 }, { "epoch": 0.30156262783277427, "grad_norm": 0.5804848670959473, "learning_rate": 7.539065695819357e-06, "loss": 0.0077, "step": 18430 }, { "epoch": 0.3017262537838501, "grad_norm": 0.12283514440059662, "learning_rate": 7.5431563445962545e-06, "loss": 0.0055, "step": 18440 }, { "epoch": 0.301889879734926, "grad_norm": 0.09182077646255493, "learning_rate": 7.547246993373149e-06, "loss": 0.0042, "step": 18450 }, { "epoch": 0.3020535056860018, "grad_norm": 0.22442393004894257, "learning_rate": 7.551337642150046e-06, "loss": 0.0052, "step": 18460 }, { "epoch": 0.30221713163707764, "grad_norm": 0.13014543056488037, "learning_rate": 7.555428290926942e-06, "loss": 0.006, "step": 18470 }, { "epoch": 0.30238075758815347, "grad_norm": 0.15237633883953094, "learning_rate": 7.559518939703837e-06, "loss": 0.0036, "step": 18480 }, { "epoch": 0.3025443835392293, "grad_norm": 0.2913595736026764, "learning_rate": 7.563609588480734e-06, "loss": 0.0058, "step": 18490 }, { "epoch": 0.3027080094903052, "grad_norm": 0.22992025315761566, "learning_rate": 7.567700237257629e-06, "loss": 0.0077, "step": 18500 }, { "epoch": 0.302871635441381, "grad_norm": 0.26717495918273926, "learning_rate": 7.571790886034525e-06, "loss": 0.0041, "step": 18510 }, { "epoch": 0.30303526139245685, "grad_norm": 0.30904456973075867, "learning_rate": 7.575881534811421e-06, "loss": 0.0062, "step": 18520 }, { "epoch": 0.3031988873435327, "grad_norm": 0.3449600040912628, "learning_rate": 7.579972183588318e-06, "loss": 0.0037, "step": 18530 }, { "epoch": 0.3033625132946085, "grad_norm": 0.2026180624961853, "learning_rate": 7.584062832365213e-06, "loss": 0.004, "step": 18540 }, { "epoch": 0.3035261392456844, "grad_norm": 0.12346754968166351, "learning_rate": 7.58815348114211e-06, "loss": 0.0056, "step": 18550 }, { "epoch": 0.3036897651967602, "grad_norm": 0.3728863298892975, "learning_rate": 7.5922441299190056e-06, "loss": 0.0055, "step": 18560 }, { "epoch": 0.30385339114783605, "grad_norm": 0.22810564935207367, "learning_rate": 7.596334778695902e-06, "loss": 0.0072, "step": 18570 }, { "epoch": 0.3040170170989119, "grad_norm": 0.1458268016576767, "learning_rate": 7.600425427472797e-06, "loss": 0.0042, "step": 18580 }, { "epoch": 0.3041806430499877, "grad_norm": 0.1446843296289444, "learning_rate": 7.6045160762496935e-06, "loss": 0.0034, "step": 18590 }, { "epoch": 0.3043442690010636, "grad_norm": 0.1632337123155594, "learning_rate": 7.608606725026589e-06, "loss": 0.0047, "step": 18600 }, { "epoch": 0.3045078949521394, "grad_norm": 0.3342912495136261, "learning_rate": 7.612697373803486e-06, "loss": 0.0047, "step": 18610 }, { "epoch": 0.30467152090321525, "grad_norm": 0.352647602558136, "learning_rate": 7.6167880225803815e-06, "loss": 0.0058, "step": 18620 }, { "epoch": 0.3048351468542911, "grad_norm": 0.12880577147006989, "learning_rate": 7.620878671357278e-06, "loss": 0.0063, "step": 18630 }, { "epoch": 0.3049987728053669, "grad_norm": 0.14008937776088715, "learning_rate": 7.624969320134174e-06, "loss": 0.0069, "step": 18640 }, { "epoch": 0.3051623987564428, "grad_norm": 0.0768856331706047, "learning_rate": 7.62905996891107e-06, "loss": 0.0048, "step": 18650 }, { "epoch": 0.30532602470751863, "grad_norm": 0.043743979185819626, "learning_rate": 7.633150617687965e-06, "loss": 0.0071, "step": 18660 }, { "epoch": 0.30548965065859446, "grad_norm": 0.10971121490001678, "learning_rate": 7.637241266464862e-06, "loss": 0.0037, "step": 18670 }, { "epoch": 0.3056532766096703, "grad_norm": 0.2039702832698822, "learning_rate": 7.641331915241758e-06, "loss": 0.0036, "step": 18680 }, { "epoch": 0.3058169025607461, "grad_norm": 0.2806364595890045, "learning_rate": 7.645422564018655e-06, "loss": 0.0046, "step": 18690 }, { "epoch": 0.30598052851182195, "grad_norm": 0.2097017467021942, "learning_rate": 7.64951321279555e-06, "loss": 0.007, "step": 18700 }, { "epoch": 0.30614415446289783, "grad_norm": 0.1897299438714981, "learning_rate": 7.653603861572446e-06, "loss": 0.0053, "step": 18710 }, { "epoch": 0.30630778041397366, "grad_norm": 0.2692115008831024, "learning_rate": 7.657694510349341e-06, "loss": 0.0037, "step": 18720 }, { "epoch": 0.3064714063650495, "grad_norm": 0.09947997331619263, "learning_rate": 7.661785159126238e-06, "loss": 0.007, "step": 18730 }, { "epoch": 0.3066350323161253, "grad_norm": 0.05276409536600113, "learning_rate": 7.665875807903134e-06, "loss": 0.0042, "step": 18740 }, { "epoch": 0.30679865826720115, "grad_norm": 0.17177321016788483, "learning_rate": 7.66996645668003e-06, "loss": 0.0053, "step": 18750 }, { "epoch": 0.30696228421827704, "grad_norm": 0.06253622472286224, "learning_rate": 7.674057105456926e-06, "loss": 0.0049, "step": 18760 }, { "epoch": 0.30712591016935287, "grad_norm": 0.15851595997810364, "learning_rate": 7.678147754233822e-06, "loss": 0.0035, "step": 18770 }, { "epoch": 0.3072895361204287, "grad_norm": 0.18347696959972382, "learning_rate": 7.682238403010717e-06, "loss": 0.0053, "step": 18780 }, { "epoch": 0.3074531620715045, "grad_norm": 0.2729877829551697, "learning_rate": 7.686329051787614e-06, "loss": 0.0052, "step": 18790 }, { "epoch": 0.30761678802258036, "grad_norm": 0.17296148836612701, "learning_rate": 7.69041970056451e-06, "loss": 0.003, "step": 18800 }, { "epoch": 0.30778041397365624, "grad_norm": 0.2672906517982483, "learning_rate": 7.694510349341407e-06, "loss": 0.007, "step": 18810 }, { "epoch": 0.30794403992473207, "grad_norm": 0.23469918966293335, "learning_rate": 7.698600998118302e-06, "loss": 0.003, "step": 18820 }, { "epoch": 0.3081076658758079, "grad_norm": 0.13074682652950287, "learning_rate": 7.702691646895198e-06, "loss": 0.0044, "step": 18830 }, { "epoch": 0.30827129182688373, "grad_norm": 0.14064204692840576, "learning_rate": 7.706782295672095e-06, "loss": 0.0045, "step": 18840 }, { "epoch": 0.30843491777795956, "grad_norm": 0.17295491695404053, "learning_rate": 7.710872944448991e-06, "loss": 0.0043, "step": 18850 }, { "epoch": 0.30859854372903545, "grad_norm": 0.07957793772220612, "learning_rate": 7.714963593225886e-06, "loss": 0.0037, "step": 18860 }, { "epoch": 0.3087621696801113, "grad_norm": 0.23941302299499512, "learning_rate": 7.719054242002783e-06, "loss": 0.0068, "step": 18870 }, { "epoch": 0.3089257956311871, "grad_norm": 0.002763977274298668, "learning_rate": 7.723144890779678e-06, "loss": 0.0022, "step": 18880 }, { "epoch": 0.30908942158226294, "grad_norm": 0.26649191975593567, "learning_rate": 7.727235539556574e-06, "loss": 0.0071, "step": 18890 }, { "epoch": 0.30925304753333877, "grad_norm": 0.22485825419425964, "learning_rate": 7.73132618833347e-06, "loss": 0.006, "step": 18900 }, { "epoch": 0.30941667348441465, "grad_norm": 0.16092853248119354, "learning_rate": 7.735416837110367e-06, "loss": 0.0051, "step": 18910 }, { "epoch": 0.3095802994354905, "grad_norm": 0.1497368961572647, "learning_rate": 7.739507485887262e-06, "loss": 0.0031, "step": 18920 }, { "epoch": 0.3097439253865663, "grad_norm": 0.09456463903188705, "learning_rate": 7.743598134664159e-06, "loss": 0.0029, "step": 18930 }, { "epoch": 0.30990755133764214, "grad_norm": 0.04766058549284935, "learning_rate": 7.747688783441053e-06, "loss": 0.0048, "step": 18940 }, { "epoch": 0.31007117728871797, "grad_norm": 0.42687052488327026, "learning_rate": 7.75177943221795e-06, "loss": 0.0043, "step": 18950 }, { "epoch": 0.31023480323979385, "grad_norm": 0.5042826533317566, "learning_rate": 7.755870080994847e-06, "loss": 0.0063, "step": 18960 }, { "epoch": 0.3103984291908697, "grad_norm": 0.13414916396141052, "learning_rate": 7.759960729771743e-06, "loss": 0.0046, "step": 18970 }, { "epoch": 0.3105620551419455, "grad_norm": 0.19323979318141937, "learning_rate": 7.764051378548638e-06, "loss": 0.0053, "step": 18980 }, { "epoch": 0.31072568109302134, "grad_norm": 0.26154085993766785, "learning_rate": 7.768142027325535e-06, "loss": 0.0057, "step": 18990 }, { "epoch": 0.3108893070440972, "grad_norm": 0.06799019873142242, "learning_rate": 7.772232676102431e-06, "loss": 0.0047, "step": 19000 }, { "epoch": 0.31105293299517306, "grad_norm": 0.1372680515050888, "learning_rate": 7.776323324879328e-06, "loss": 0.005, "step": 19010 }, { "epoch": 0.3112165589462489, "grad_norm": 0.3767656981945038, "learning_rate": 7.780413973656223e-06, "loss": 0.0055, "step": 19020 }, { "epoch": 0.3113801848973247, "grad_norm": 0.410710871219635, "learning_rate": 7.784504622433119e-06, "loss": 0.0079, "step": 19030 }, { "epoch": 0.31154381084840055, "grad_norm": 0.20782385766506195, "learning_rate": 7.788595271210014e-06, "loss": 0.004, "step": 19040 }, { "epoch": 0.3117074367994764, "grad_norm": 0.18275180459022522, "learning_rate": 7.79268591998691e-06, "loss": 0.0048, "step": 19050 }, { "epoch": 0.31187106275055226, "grad_norm": 0.10391122847795486, "learning_rate": 7.796776568763807e-06, "loss": 0.0039, "step": 19060 }, { "epoch": 0.3120346887016281, "grad_norm": 0.04727224260568619, "learning_rate": 7.800867217540704e-06, "loss": 0.005, "step": 19070 }, { "epoch": 0.3121983146527039, "grad_norm": 0.1769312173128128, "learning_rate": 7.804957866317598e-06, "loss": 0.0074, "step": 19080 }, { "epoch": 0.31236194060377975, "grad_norm": 0.1780601590871811, "learning_rate": 7.809048515094495e-06, "loss": 0.0038, "step": 19090 }, { "epoch": 0.3125255665548556, "grad_norm": 0.08103331923484802, "learning_rate": 7.81313916387139e-06, "loss": 0.0028, "step": 19100 }, { "epoch": 0.31268919250593147, "grad_norm": 0.3151041567325592, "learning_rate": 7.817229812648286e-06, "loss": 0.0039, "step": 19110 }, { "epoch": 0.3128528184570073, "grad_norm": 0.2823973298072815, "learning_rate": 7.821320461425183e-06, "loss": 0.006, "step": 19120 }, { "epoch": 0.3130164444080831, "grad_norm": 0.34604787826538086, "learning_rate": 7.82541111020208e-06, "loss": 0.0042, "step": 19130 }, { "epoch": 0.31318007035915896, "grad_norm": 0.07746456563472748, "learning_rate": 7.829501758978974e-06, "loss": 0.0033, "step": 19140 }, { "epoch": 0.3133436963102348, "grad_norm": 0.28007733821868896, "learning_rate": 7.833592407755871e-06, "loss": 0.0049, "step": 19150 }, { "epoch": 0.3135073222613106, "grad_norm": 0.13396677374839783, "learning_rate": 7.837683056532766e-06, "loss": 0.0047, "step": 19160 }, { "epoch": 0.3136709482123865, "grad_norm": 0.3597058951854706, "learning_rate": 7.841773705309662e-06, "loss": 0.0055, "step": 19170 }, { "epoch": 0.31383457416346233, "grad_norm": 0.13958048820495605, "learning_rate": 7.845864354086559e-06, "loss": 0.0045, "step": 19180 }, { "epoch": 0.31399820011453816, "grad_norm": 0.20837749540805817, "learning_rate": 7.849955002863455e-06, "loss": 0.0056, "step": 19190 }, { "epoch": 0.314161826065614, "grad_norm": 0.277681440114975, "learning_rate": 7.85404565164035e-06, "loss": 0.0055, "step": 19200 }, { "epoch": 0.3143254520166898, "grad_norm": 0.1765432208776474, "learning_rate": 7.858136300417247e-06, "loss": 0.0044, "step": 19210 }, { "epoch": 0.3144890779677657, "grad_norm": 0.23526066541671753, "learning_rate": 7.862226949194143e-06, "loss": 0.0067, "step": 19220 }, { "epoch": 0.31465270391884154, "grad_norm": 0.12809814512729645, "learning_rate": 7.86631759797104e-06, "loss": 0.0038, "step": 19230 }, { "epoch": 0.31481632986991737, "grad_norm": 0.15444503724575043, "learning_rate": 7.870408246747935e-06, "loss": 0.0043, "step": 19240 }, { "epoch": 0.3149799558209932, "grad_norm": 0.3523310422897339, "learning_rate": 7.874498895524831e-06, "loss": 0.0052, "step": 19250 }, { "epoch": 0.315143581772069, "grad_norm": 0.14869391918182373, "learning_rate": 7.878589544301726e-06, "loss": 0.0044, "step": 19260 }, { "epoch": 0.3153072077231449, "grad_norm": 0.29279232025146484, "learning_rate": 7.882680193078623e-06, "loss": 0.0044, "step": 19270 }, { "epoch": 0.31547083367422074, "grad_norm": 0.11961070448160172, "learning_rate": 7.88677084185552e-06, "loss": 0.0036, "step": 19280 }, { "epoch": 0.31563445962529657, "grad_norm": 0.25717243552207947, "learning_rate": 7.890861490632416e-06, "loss": 0.0043, "step": 19290 }, { "epoch": 0.3157980855763724, "grad_norm": 0.1517508327960968, "learning_rate": 7.89495213940931e-06, "loss": 0.0048, "step": 19300 }, { "epoch": 0.31596171152744823, "grad_norm": 0.4031235873699188, "learning_rate": 7.899042788186207e-06, "loss": 0.009, "step": 19310 }, { "epoch": 0.3161253374785241, "grad_norm": 0.2463417947292328, "learning_rate": 7.903133436963102e-06, "loss": 0.0039, "step": 19320 }, { "epoch": 0.31628896342959995, "grad_norm": 0.2547284960746765, "learning_rate": 7.907224085739999e-06, "loss": 0.004, "step": 19330 }, { "epoch": 0.3164525893806758, "grad_norm": 0.39489611983299255, "learning_rate": 7.911314734516895e-06, "loss": 0.0062, "step": 19340 }, { "epoch": 0.3166162153317516, "grad_norm": 0.27700304985046387, "learning_rate": 7.91540538329379e-06, "loss": 0.0053, "step": 19350 }, { "epoch": 0.31677984128282743, "grad_norm": 0.12871511280536652, "learning_rate": 7.919496032070687e-06, "loss": 0.0047, "step": 19360 }, { "epoch": 0.3169434672339033, "grad_norm": 0.25581035017967224, "learning_rate": 7.923586680847583e-06, "loss": 0.0053, "step": 19370 }, { "epoch": 0.31710709318497915, "grad_norm": 0.27006617188453674, "learning_rate": 7.92767732962448e-06, "loss": 0.0067, "step": 19380 }, { "epoch": 0.317270719136055, "grad_norm": 0.29440680146217346, "learning_rate": 7.931767978401375e-06, "loss": 0.0064, "step": 19390 }, { "epoch": 0.3174343450871308, "grad_norm": 0.24534447491168976, "learning_rate": 7.935858627178271e-06, "loss": 0.0046, "step": 19400 }, { "epoch": 0.31759797103820664, "grad_norm": 0.3082183003425598, "learning_rate": 7.939949275955166e-06, "loss": 0.0037, "step": 19410 }, { "epoch": 0.3177615969892825, "grad_norm": 0.4193437397480011, "learning_rate": 7.944039924732063e-06, "loss": 0.0049, "step": 19420 }, { "epoch": 0.31792522294035835, "grad_norm": 0.13350224494934082, "learning_rate": 7.94813057350896e-06, "loss": 0.0072, "step": 19430 }, { "epoch": 0.3180888488914342, "grad_norm": 0.14556393027305603, "learning_rate": 7.952221222285856e-06, "loss": 0.0047, "step": 19440 }, { "epoch": 0.31825247484251, "grad_norm": 0.3152436316013336, "learning_rate": 7.95631187106275e-06, "loss": 0.0057, "step": 19450 }, { "epoch": 0.31841610079358584, "grad_norm": 0.16616572439670563, "learning_rate": 7.960402519839647e-06, "loss": 0.0039, "step": 19460 }, { "epoch": 0.31857972674466173, "grad_norm": 0.13911588490009308, "learning_rate": 7.964493168616542e-06, "loss": 0.005, "step": 19470 }, { "epoch": 0.31874335269573756, "grad_norm": 0.21921111643314362, "learning_rate": 7.968583817393439e-06, "loss": 0.0031, "step": 19480 }, { "epoch": 0.3189069786468134, "grad_norm": 0.2710632383823395, "learning_rate": 7.972674466170335e-06, "loss": 0.0059, "step": 19490 }, { "epoch": 0.3190706045978892, "grad_norm": 0.25952523946762085, "learning_rate": 7.976765114947232e-06, "loss": 0.0057, "step": 19500 }, { "epoch": 0.31923423054896505, "grad_norm": 0.1300297975540161, "learning_rate": 7.980855763724127e-06, "loss": 0.005, "step": 19510 }, { "epoch": 0.31939785650004093, "grad_norm": 0.23904241621494293, "learning_rate": 7.984946412501023e-06, "loss": 0.0033, "step": 19520 }, { "epoch": 0.31956148245111676, "grad_norm": 0.19017618894577026, "learning_rate": 7.98903706127792e-06, "loss": 0.0057, "step": 19530 }, { "epoch": 0.3197251084021926, "grad_norm": 0.3804857134819031, "learning_rate": 7.993127710054816e-06, "loss": 0.0044, "step": 19540 }, { "epoch": 0.3198887343532684, "grad_norm": 0.23853324353694916, "learning_rate": 7.997218358831711e-06, "loss": 0.0039, "step": 19550 }, { "epoch": 0.32005236030434425, "grad_norm": 0.2351067066192627, "learning_rate": 8.001309007608608e-06, "loss": 0.003, "step": 19560 }, { "epoch": 0.3202159862554201, "grad_norm": 0.11607512086629868, "learning_rate": 8.005399656385503e-06, "loss": 0.0041, "step": 19570 }, { "epoch": 0.32037961220649597, "grad_norm": 0.3082251250743866, "learning_rate": 8.009490305162399e-06, "loss": 0.0063, "step": 19580 }, { "epoch": 0.3205432381575718, "grad_norm": 0.19097323715686798, "learning_rate": 8.013580953939296e-06, "loss": 0.0048, "step": 19590 }, { "epoch": 0.3207068641086476, "grad_norm": 0.16701151430606842, "learning_rate": 8.017671602716192e-06, "loss": 0.005, "step": 19600 }, { "epoch": 0.32087049005972346, "grad_norm": 0.15833081305027008, "learning_rate": 8.021762251493087e-06, "loss": 0.0055, "step": 19610 }, { "epoch": 0.3210341160107993, "grad_norm": 0.13259179890155792, "learning_rate": 8.025852900269984e-06, "loss": 0.0048, "step": 19620 }, { "epoch": 0.32119774196187517, "grad_norm": 0.19049565494060516, "learning_rate": 8.029943549046878e-06, "loss": 0.0065, "step": 19630 }, { "epoch": 0.321361367912951, "grad_norm": 0.09582779556512833, "learning_rate": 8.034034197823775e-06, "loss": 0.0044, "step": 19640 }, { "epoch": 0.32152499386402683, "grad_norm": 0.6378761529922485, "learning_rate": 8.038124846600672e-06, "loss": 0.0064, "step": 19650 }, { "epoch": 0.32168861981510266, "grad_norm": 0.2533129155635834, "learning_rate": 8.042215495377568e-06, "loss": 0.0055, "step": 19660 }, { "epoch": 0.3218522457661785, "grad_norm": 0.14883220195770264, "learning_rate": 8.046306144154463e-06, "loss": 0.0054, "step": 19670 }, { "epoch": 0.3220158717172544, "grad_norm": 0.19980032742023468, "learning_rate": 8.05039679293136e-06, "loss": 0.0065, "step": 19680 }, { "epoch": 0.3221794976683302, "grad_norm": 0.3442917764186859, "learning_rate": 8.054487441708254e-06, "loss": 0.0068, "step": 19690 }, { "epoch": 0.32234312361940604, "grad_norm": 0.23766648769378662, "learning_rate": 8.058578090485151e-06, "loss": 0.0044, "step": 19700 }, { "epoch": 0.32250674957048187, "grad_norm": 0.07321497052907944, "learning_rate": 8.062668739262048e-06, "loss": 0.0026, "step": 19710 }, { "epoch": 0.3226703755215577, "grad_norm": 0.09744125604629517, "learning_rate": 8.066759388038944e-06, "loss": 0.0045, "step": 19720 }, { "epoch": 0.3228340014726336, "grad_norm": 0.10075178742408752, "learning_rate": 8.070850036815839e-06, "loss": 0.0041, "step": 19730 }, { "epoch": 0.3229976274237094, "grad_norm": 0.25104907155036926, "learning_rate": 8.074940685592736e-06, "loss": 0.0053, "step": 19740 }, { "epoch": 0.32316125337478524, "grad_norm": 0.1305944174528122, "learning_rate": 8.079031334369632e-06, "loss": 0.0058, "step": 19750 }, { "epoch": 0.32332487932586107, "grad_norm": 0.1521109938621521, "learning_rate": 8.083121983146529e-06, "loss": 0.0035, "step": 19760 }, { "epoch": 0.3234885052769369, "grad_norm": 0.11625616997480392, "learning_rate": 8.087212631923423e-06, "loss": 0.004, "step": 19770 }, { "epoch": 0.3236521312280128, "grad_norm": 0.1586543321609497, "learning_rate": 8.09130328070032e-06, "loss": 0.0026, "step": 19780 }, { "epoch": 0.3238157571790886, "grad_norm": 0.10577210038900375, "learning_rate": 8.095393929477215e-06, "loss": 0.0045, "step": 19790 }, { "epoch": 0.32397938313016444, "grad_norm": 0.29894617199897766, "learning_rate": 8.099484578254111e-06, "loss": 0.0063, "step": 19800 }, { "epoch": 0.3241430090812403, "grad_norm": 0.49183616042137146, "learning_rate": 8.103575227031008e-06, "loss": 0.0053, "step": 19810 }, { "epoch": 0.3243066350323161, "grad_norm": 0.07915569096803665, "learning_rate": 8.107665875807905e-06, "loss": 0.0039, "step": 19820 }, { "epoch": 0.324470260983392, "grad_norm": 0.12183774262666702, "learning_rate": 8.1117565245848e-06, "loss": 0.0051, "step": 19830 }, { "epoch": 0.3246338869344678, "grad_norm": 0.2872908115386963, "learning_rate": 8.115847173361696e-06, "loss": 0.0039, "step": 19840 }, { "epoch": 0.32479751288554365, "grad_norm": 0.3507295548915863, "learning_rate": 8.11993782213859e-06, "loss": 0.0043, "step": 19850 }, { "epoch": 0.3249611388366195, "grad_norm": 0.13108202815055847, "learning_rate": 8.124028470915487e-06, "loss": 0.0059, "step": 19860 }, { "epoch": 0.3251247647876953, "grad_norm": 0.2939903438091278, "learning_rate": 8.128119119692384e-06, "loss": 0.0048, "step": 19870 }, { "epoch": 0.3252883907387712, "grad_norm": 0.062342189252376556, "learning_rate": 8.13220976846928e-06, "loss": 0.0071, "step": 19880 }, { "epoch": 0.325452016689847, "grad_norm": 0.11442258208990097, "learning_rate": 8.136300417246175e-06, "loss": 0.005, "step": 19890 }, { "epoch": 0.32561564264092285, "grad_norm": 0.5130837559700012, "learning_rate": 8.140391066023072e-06, "loss": 0.0064, "step": 19900 }, { "epoch": 0.3257792685919987, "grad_norm": 0.27795618772506714, "learning_rate": 8.144481714799968e-06, "loss": 0.0064, "step": 19910 }, { "epoch": 0.3259428945430745, "grad_norm": 0.14897869527339935, "learning_rate": 8.148572363576865e-06, "loss": 0.0047, "step": 19920 }, { "epoch": 0.3261065204941504, "grad_norm": 0.25517538189888, "learning_rate": 8.15266301235376e-06, "loss": 0.0048, "step": 19930 }, { "epoch": 0.32627014644522623, "grad_norm": 0.24248509109020233, "learning_rate": 8.156753661130656e-06, "loss": 0.0049, "step": 19940 }, { "epoch": 0.32643377239630206, "grad_norm": 0.09647409617900848, "learning_rate": 8.160844309907551e-06, "loss": 0.0056, "step": 19950 }, { "epoch": 0.3265973983473779, "grad_norm": 0.20471543073654175, "learning_rate": 8.164934958684448e-06, "loss": 0.0035, "step": 19960 }, { "epoch": 0.3267610242984537, "grad_norm": 0.17148776352405548, "learning_rate": 8.169025607461344e-06, "loss": 0.0059, "step": 19970 }, { "epoch": 0.3269246502495296, "grad_norm": 0.18397770822048187, "learning_rate": 8.173116256238241e-06, "loss": 0.0039, "step": 19980 }, { "epoch": 0.32708827620060543, "grad_norm": 0.320706844329834, "learning_rate": 8.177206905015136e-06, "loss": 0.0045, "step": 19990 }, { "epoch": 0.32725190215168126, "grad_norm": 0.24201466143131256, "learning_rate": 8.181297553792032e-06, "loss": 0.0082, "step": 20000 }, { "epoch": 0.3274155281027571, "grad_norm": 0.13273926079273224, "learning_rate": 8.185388202568927e-06, "loss": 0.0043, "step": 20010 }, { "epoch": 0.3275791540538329, "grad_norm": 0.10021848976612091, "learning_rate": 8.189478851345824e-06, "loss": 0.0045, "step": 20020 }, { "epoch": 0.32774278000490875, "grad_norm": 0.22692611813545227, "learning_rate": 8.19356950012272e-06, "loss": 0.0031, "step": 20030 }, { "epoch": 0.32790640595598464, "grad_norm": 0.187375009059906, "learning_rate": 8.197660148899617e-06, "loss": 0.0057, "step": 20040 }, { "epoch": 0.32807003190706047, "grad_norm": 0.21428370475769043, "learning_rate": 8.201750797676512e-06, "loss": 0.0055, "step": 20050 }, { "epoch": 0.3282336578581363, "grad_norm": 0.25382089614868164, "learning_rate": 8.205841446453408e-06, "loss": 0.0071, "step": 20060 }, { "epoch": 0.3283972838092121, "grad_norm": 0.12623588740825653, "learning_rate": 8.209932095230305e-06, "loss": 0.0059, "step": 20070 }, { "epoch": 0.32856090976028796, "grad_norm": 0.20093154907226562, "learning_rate": 8.214022744007201e-06, "loss": 0.0031, "step": 20080 }, { "epoch": 0.32872453571136384, "grad_norm": 0.4679361879825592, "learning_rate": 8.218113392784096e-06, "loss": 0.008, "step": 20090 }, { "epoch": 0.32888816166243967, "grad_norm": 0.15006762742996216, "learning_rate": 8.222204041560993e-06, "loss": 0.0038, "step": 20100 }, { "epoch": 0.3290517876135155, "grad_norm": 0.16611838340759277, "learning_rate": 8.226294690337888e-06, "loss": 0.004, "step": 20110 }, { "epoch": 0.32921541356459133, "grad_norm": 0.16666080057621002, "learning_rate": 8.230385339114784e-06, "loss": 0.0068, "step": 20120 }, { "epoch": 0.32937903951566716, "grad_norm": 0.05329088866710663, "learning_rate": 8.23447598789168e-06, "loss": 0.0066, "step": 20130 }, { "epoch": 0.32954266546674305, "grad_norm": 0.3946906626224518, "learning_rate": 8.238566636668577e-06, "loss": 0.0061, "step": 20140 }, { "epoch": 0.3297062914178189, "grad_norm": 0.13600410521030426, "learning_rate": 8.242657285445472e-06, "loss": 0.0045, "step": 20150 }, { "epoch": 0.3298699173688947, "grad_norm": 0.2781405746936798, "learning_rate": 8.246747934222369e-06, "loss": 0.0056, "step": 20160 }, { "epoch": 0.33003354331997053, "grad_norm": 0.2997875213623047, "learning_rate": 8.250838582999264e-06, "loss": 0.0089, "step": 20170 }, { "epoch": 0.33019716927104636, "grad_norm": 0.3011452555656433, "learning_rate": 8.25492923177616e-06, "loss": 0.0046, "step": 20180 }, { "epoch": 0.33036079522212225, "grad_norm": 0.12200075387954712, "learning_rate": 8.259019880553057e-06, "loss": 0.0053, "step": 20190 }, { "epoch": 0.3305244211731981, "grad_norm": 0.2633582353591919, "learning_rate": 8.263110529329953e-06, "loss": 0.0053, "step": 20200 }, { "epoch": 0.3306880471242739, "grad_norm": 0.22029295563697815, "learning_rate": 8.267201178106848e-06, "loss": 0.0064, "step": 20210 }, { "epoch": 0.33085167307534974, "grad_norm": 0.2690644860267639, "learning_rate": 8.271291826883745e-06, "loss": 0.0079, "step": 20220 }, { "epoch": 0.33101529902642557, "grad_norm": 0.1486840695142746, "learning_rate": 8.275382475660641e-06, "loss": 0.0041, "step": 20230 }, { "epoch": 0.33117892497750145, "grad_norm": 0.1798991709947586, "learning_rate": 8.279473124437536e-06, "loss": 0.0031, "step": 20240 }, { "epoch": 0.3313425509285773, "grad_norm": 0.48375213146209717, "learning_rate": 8.283563773214433e-06, "loss": 0.0054, "step": 20250 }, { "epoch": 0.3315061768796531, "grad_norm": 0.09920594096183777, "learning_rate": 8.287654421991328e-06, "loss": 0.0047, "step": 20260 }, { "epoch": 0.33166980283072894, "grad_norm": 0.1942196935415268, "learning_rate": 8.291745070768224e-06, "loss": 0.0031, "step": 20270 }, { "epoch": 0.3318334287818048, "grad_norm": 0.15283474326133728, "learning_rate": 8.29583571954512e-06, "loss": 0.0038, "step": 20280 }, { "epoch": 0.33199705473288066, "grad_norm": 0.19673120975494385, "learning_rate": 8.299926368322017e-06, "loss": 0.0038, "step": 20290 }, { "epoch": 0.3321606806839565, "grad_norm": 0.10388221591711044, "learning_rate": 8.304017017098912e-06, "loss": 0.0062, "step": 20300 }, { "epoch": 0.3323243066350323, "grad_norm": 0.18718284368515015, "learning_rate": 8.308107665875809e-06, "loss": 0.004, "step": 20310 }, { "epoch": 0.33248793258610815, "grad_norm": 0.10253486782312393, "learning_rate": 8.312198314652704e-06, "loss": 0.003, "step": 20320 }, { "epoch": 0.332651558537184, "grad_norm": 0.21905989944934845, "learning_rate": 8.3162889634296e-06, "loss": 0.0051, "step": 20330 }, { "epoch": 0.33281518448825986, "grad_norm": 0.23274537920951843, "learning_rate": 8.320379612206497e-06, "loss": 0.006, "step": 20340 }, { "epoch": 0.3329788104393357, "grad_norm": 0.1298518180847168, "learning_rate": 8.324470260983393e-06, "loss": 0.0032, "step": 20350 }, { "epoch": 0.3331424363904115, "grad_norm": 0.11103275418281555, "learning_rate": 8.328560909760288e-06, "loss": 0.0054, "step": 20360 }, { "epoch": 0.33330606234148735, "grad_norm": 0.10212615877389908, "learning_rate": 8.332651558537185e-06, "loss": 0.0037, "step": 20370 }, { "epoch": 0.3334696882925632, "grad_norm": 0.32354986667633057, "learning_rate": 8.33674220731408e-06, "loss": 0.0041, "step": 20380 }, { "epoch": 0.33363331424363907, "grad_norm": 0.11438208818435669, "learning_rate": 8.340832856090976e-06, "loss": 0.0051, "step": 20390 }, { "epoch": 0.3337969401947149, "grad_norm": 0.27901801466941833, "learning_rate": 8.344923504867873e-06, "loss": 0.0047, "step": 20400 }, { "epoch": 0.3339605661457907, "grad_norm": 0.20371009409427643, "learning_rate": 8.349014153644769e-06, "loss": 0.0059, "step": 20410 }, { "epoch": 0.33412419209686656, "grad_norm": 0.3445451855659485, "learning_rate": 8.353104802421664e-06, "loss": 0.0068, "step": 20420 }, { "epoch": 0.3342878180479424, "grad_norm": 0.4437674283981323, "learning_rate": 8.35719545119856e-06, "loss": 0.0063, "step": 20430 }, { "epoch": 0.33445144399901827, "grad_norm": 0.12597471475601196, "learning_rate": 8.361286099975457e-06, "loss": 0.0047, "step": 20440 }, { "epoch": 0.3346150699500941, "grad_norm": 0.3394995629787445, "learning_rate": 8.365376748752354e-06, "loss": 0.0048, "step": 20450 }, { "epoch": 0.33477869590116993, "grad_norm": 0.04256118834018707, "learning_rate": 8.369467397529249e-06, "loss": 0.0034, "step": 20460 }, { "epoch": 0.33494232185224576, "grad_norm": 0.13757580518722534, "learning_rate": 8.373558046306145e-06, "loss": 0.0036, "step": 20470 }, { "epoch": 0.3351059478033216, "grad_norm": 0.3718339204788208, "learning_rate": 8.37764869508304e-06, "loss": 0.0117, "step": 20480 }, { "epoch": 0.3352695737543974, "grad_norm": 0.09792344272136688, "learning_rate": 8.381739343859936e-06, "loss": 0.0041, "step": 20490 }, { "epoch": 0.3354331997054733, "grad_norm": 0.2893122434616089, "learning_rate": 8.385829992636833e-06, "loss": 0.0055, "step": 20500 }, { "epoch": 0.33559682565654914, "grad_norm": 0.38381460309028625, "learning_rate": 8.38992064141373e-06, "loss": 0.0053, "step": 20510 }, { "epoch": 0.33576045160762497, "grad_norm": 0.09499429166316986, "learning_rate": 8.394011290190624e-06, "loss": 0.0053, "step": 20520 }, { "epoch": 0.3359240775587008, "grad_norm": 0.1613861471414566, "learning_rate": 8.398101938967521e-06, "loss": 0.0084, "step": 20530 }, { "epoch": 0.3360877035097766, "grad_norm": 0.4145393967628479, "learning_rate": 8.402192587744416e-06, "loss": 0.0067, "step": 20540 }, { "epoch": 0.3362513294608525, "grad_norm": 0.2662180960178375, "learning_rate": 8.406283236521312e-06, "loss": 0.0046, "step": 20550 }, { "epoch": 0.33641495541192834, "grad_norm": 0.18077141046524048, "learning_rate": 8.410373885298209e-06, "loss": 0.0046, "step": 20560 }, { "epoch": 0.33657858136300417, "grad_norm": 0.10458344221115112, "learning_rate": 8.414464534075106e-06, "loss": 0.0037, "step": 20570 }, { "epoch": 0.33674220731408, "grad_norm": 0.5180356502532959, "learning_rate": 8.418555182852e-06, "loss": 0.0057, "step": 20580 }, { "epoch": 0.33690583326515583, "grad_norm": 0.14359240233898163, "learning_rate": 8.422645831628897e-06, "loss": 0.0049, "step": 20590 }, { "epoch": 0.3370694592162317, "grad_norm": 0.08442745357751846, "learning_rate": 8.426736480405793e-06, "loss": 0.0044, "step": 20600 }, { "epoch": 0.33723308516730754, "grad_norm": 0.19987064599990845, "learning_rate": 8.43082712918269e-06, "loss": 0.0037, "step": 20610 }, { "epoch": 0.3373967111183834, "grad_norm": 0.05664944648742676, "learning_rate": 8.434917777959585e-06, "loss": 0.0029, "step": 20620 }, { "epoch": 0.3375603370694592, "grad_norm": 0.19470560550689697, "learning_rate": 8.439008426736481e-06, "loss": 0.0036, "step": 20630 }, { "epoch": 0.33772396302053503, "grad_norm": 0.07094257324934006, "learning_rate": 8.443099075513376e-06, "loss": 0.0038, "step": 20640 }, { "epoch": 0.3378875889716109, "grad_norm": 0.16477927565574646, "learning_rate": 8.447189724290273e-06, "loss": 0.0058, "step": 20650 }, { "epoch": 0.33805121492268675, "grad_norm": 0.11610040068626404, "learning_rate": 8.45128037306717e-06, "loss": 0.0034, "step": 20660 }, { "epoch": 0.3382148408737626, "grad_norm": 0.10579053312540054, "learning_rate": 8.455371021844066e-06, "loss": 0.004, "step": 20670 }, { "epoch": 0.3383784668248384, "grad_norm": 0.128859743475914, "learning_rate": 8.459461670620961e-06, "loss": 0.0059, "step": 20680 }, { "epoch": 0.33854209277591424, "grad_norm": 0.1648133099079132, "learning_rate": 8.463552319397857e-06, "loss": 0.0034, "step": 20690 }, { "epoch": 0.3387057187269901, "grad_norm": 0.3360321819782257, "learning_rate": 8.467642968174752e-06, "loss": 0.0062, "step": 20700 }, { "epoch": 0.33886934467806595, "grad_norm": 0.17909955978393555, "learning_rate": 8.471733616951649e-06, "loss": 0.0043, "step": 20710 }, { "epoch": 0.3390329706291418, "grad_norm": 0.05808741971850395, "learning_rate": 8.475824265728545e-06, "loss": 0.0036, "step": 20720 }, { "epoch": 0.3391965965802176, "grad_norm": 0.2061193585395813, "learning_rate": 8.479914914505442e-06, "loss": 0.0061, "step": 20730 }, { "epoch": 0.33936022253129344, "grad_norm": 0.20300054550170898, "learning_rate": 8.484005563282337e-06, "loss": 0.0044, "step": 20740 }, { "epoch": 0.33952384848236933, "grad_norm": 0.19833914935588837, "learning_rate": 8.488096212059233e-06, "loss": 0.0037, "step": 20750 }, { "epoch": 0.33968747443344516, "grad_norm": 0.08807452768087387, "learning_rate": 8.49218686083613e-06, "loss": 0.0039, "step": 20760 }, { "epoch": 0.339851100384521, "grad_norm": 0.08674616366624832, "learning_rate": 8.496277509613026e-06, "loss": 0.0036, "step": 20770 }, { "epoch": 0.3400147263355968, "grad_norm": 0.12078633159399033, "learning_rate": 8.500368158389921e-06, "loss": 0.0041, "step": 20780 }, { "epoch": 0.34017835228667265, "grad_norm": 0.2678808569908142, "learning_rate": 8.504458807166818e-06, "loss": 0.004, "step": 20790 }, { "epoch": 0.34034197823774853, "grad_norm": 0.25788435339927673, "learning_rate": 8.508549455943713e-06, "loss": 0.0065, "step": 20800 }, { "epoch": 0.34050560418882436, "grad_norm": 0.13729840517044067, "learning_rate": 8.51264010472061e-06, "loss": 0.0034, "step": 20810 }, { "epoch": 0.3406692301399002, "grad_norm": 0.2293776422739029, "learning_rate": 8.516730753497506e-06, "loss": 0.0045, "step": 20820 }, { "epoch": 0.340832856090976, "grad_norm": 0.07786042988300323, "learning_rate": 8.520821402274402e-06, "loss": 0.0051, "step": 20830 }, { "epoch": 0.34099648204205185, "grad_norm": 0.16890448331832886, "learning_rate": 8.524912051051297e-06, "loss": 0.0052, "step": 20840 }, { "epoch": 0.34116010799312774, "grad_norm": 0.18066124618053436, "learning_rate": 8.529002699828194e-06, "loss": 0.0069, "step": 20850 }, { "epoch": 0.34132373394420357, "grad_norm": 0.12973280251026154, "learning_rate": 8.533093348605089e-06, "loss": 0.0066, "step": 20860 }, { "epoch": 0.3414873598952794, "grad_norm": 0.07088779658079147, "learning_rate": 8.537183997381985e-06, "loss": 0.0041, "step": 20870 }, { "epoch": 0.3416509858463552, "grad_norm": 0.12644897401332855, "learning_rate": 8.541274646158882e-06, "loss": 0.003, "step": 20880 }, { "epoch": 0.34181461179743106, "grad_norm": 0.22792212665081024, "learning_rate": 8.545365294935778e-06, "loss": 0.0044, "step": 20890 }, { "epoch": 0.3419782377485069, "grad_norm": 0.14202633500099182, "learning_rate": 8.549455943712673e-06, "loss": 0.0102, "step": 20900 }, { "epoch": 0.34214186369958277, "grad_norm": 0.4567568600177765, "learning_rate": 8.55354659248957e-06, "loss": 0.0074, "step": 20910 }, { "epoch": 0.3423054896506586, "grad_norm": 0.31625422835350037, "learning_rate": 8.557637241266465e-06, "loss": 0.0055, "step": 20920 }, { "epoch": 0.34246911560173443, "grad_norm": 0.2108621597290039, "learning_rate": 8.561727890043363e-06, "loss": 0.0058, "step": 20930 }, { "epoch": 0.34263274155281026, "grad_norm": 0.15458016097545624, "learning_rate": 8.565818538820258e-06, "loss": 0.0045, "step": 20940 }, { "epoch": 0.3427963675038861, "grad_norm": 0.06926602125167847, "learning_rate": 8.569909187597154e-06, "loss": 0.0032, "step": 20950 }, { "epoch": 0.342959993454962, "grad_norm": 0.13370223343372345, "learning_rate": 8.573999836374049e-06, "loss": 0.0049, "step": 20960 }, { "epoch": 0.3431236194060378, "grad_norm": 0.0604298897087574, "learning_rate": 8.578090485150946e-06, "loss": 0.0097, "step": 20970 }, { "epoch": 0.34328724535711364, "grad_norm": 0.08349194377660751, "learning_rate": 8.582181133927842e-06, "loss": 0.0039, "step": 20980 }, { "epoch": 0.34345087130818946, "grad_norm": 0.12087934464216232, "learning_rate": 8.586271782704739e-06, "loss": 0.0035, "step": 20990 }, { "epoch": 0.3436144972592653, "grad_norm": 0.19481709599494934, "learning_rate": 8.590362431481634e-06, "loss": 0.0033, "step": 21000 }, { "epoch": 0.3437781232103412, "grad_norm": 0.1556348204612732, "learning_rate": 8.59445308025853e-06, "loss": 0.0052, "step": 21010 }, { "epoch": 0.343941749161417, "grad_norm": 0.07341060787439346, "learning_rate": 8.598543729035425e-06, "loss": 0.0032, "step": 21020 }, { "epoch": 0.34410537511249284, "grad_norm": 0.24655956029891968, "learning_rate": 8.602634377812322e-06, "loss": 0.0036, "step": 21030 }, { "epoch": 0.34426900106356867, "grad_norm": 0.1589750200510025, "learning_rate": 8.606725026589218e-06, "loss": 0.0051, "step": 21040 }, { "epoch": 0.3444326270146445, "grad_norm": 0.20792227983474731, "learning_rate": 8.610815675366115e-06, "loss": 0.0034, "step": 21050 }, { "epoch": 0.3445962529657204, "grad_norm": 0.13694500923156738, "learning_rate": 8.61490632414301e-06, "loss": 0.0049, "step": 21060 }, { "epoch": 0.3447598789167962, "grad_norm": 0.25741446018218994, "learning_rate": 8.618996972919906e-06, "loss": 0.005, "step": 21070 }, { "epoch": 0.34492350486787204, "grad_norm": 0.10180523246526718, "learning_rate": 8.623087621696801e-06, "loss": 0.0038, "step": 21080 }, { "epoch": 0.3450871308189479, "grad_norm": 0.15457716584205627, "learning_rate": 8.627178270473698e-06, "loss": 0.0036, "step": 21090 }, { "epoch": 0.3452507567700237, "grad_norm": 0.07752872258424759, "learning_rate": 8.631268919250594e-06, "loss": 0.0051, "step": 21100 }, { "epoch": 0.3454143827210996, "grad_norm": 0.24382314085960388, "learning_rate": 8.635359568027489e-06, "loss": 0.0052, "step": 21110 }, { "epoch": 0.3455780086721754, "grad_norm": 0.18617527186870575, "learning_rate": 8.639450216804386e-06, "loss": 0.0043, "step": 21120 }, { "epoch": 0.34574163462325125, "grad_norm": 0.28618744015693665, "learning_rate": 8.643540865581282e-06, "loss": 0.0054, "step": 21130 }, { "epoch": 0.3459052605743271, "grad_norm": 0.2752063274383545, "learning_rate": 8.647631514358179e-06, "loss": 0.0039, "step": 21140 }, { "epoch": 0.3460688865254029, "grad_norm": 0.07854831218719482, "learning_rate": 8.651722163135074e-06, "loss": 0.0056, "step": 21150 }, { "epoch": 0.3462325124764788, "grad_norm": 0.5704415440559387, "learning_rate": 8.65581281191197e-06, "loss": 0.0046, "step": 21160 }, { "epoch": 0.3463961384275546, "grad_norm": 0.08530449867248535, "learning_rate": 8.659903460688865e-06, "loss": 0.0062, "step": 21170 }, { "epoch": 0.34655976437863045, "grad_norm": 0.23941287398338318, "learning_rate": 8.663994109465761e-06, "loss": 0.0071, "step": 21180 }, { "epoch": 0.3467233903297063, "grad_norm": 0.18629103899002075, "learning_rate": 8.668084758242658e-06, "loss": 0.0049, "step": 21190 }, { "epoch": 0.3468870162807821, "grad_norm": 0.19518473744392395, "learning_rate": 8.672175407019555e-06, "loss": 0.0042, "step": 21200 }, { "epoch": 0.347050642231858, "grad_norm": 0.3767547905445099, "learning_rate": 8.67626605579645e-06, "loss": 0.0074, "step": 21210 }, { "epoch": 0.3472142681829338, "grad_norm": 0.3724960386753082, "learning_rate": 8.680356704573346e-06, "loss": 0.0051, "step": 21220 }, { "epoch": 0.34737789413400966, "grad_norm": 0.08176866918802261, "learning_rate": 8.684447353350241e-06, "loss": 0.0057, "step": 21230 }, { "epoch": 0.3475415200850855, "grad_norm": 0.08813058584928513, "learning_rate": 8.688538002127137e-06, "loss": 0.0033, "step": 21240 }, { "epoch": 0.3477051460361613, "grad_norm": 2.233001708984375, "learning_rate": 8.692628650904034e-06, "loss": 0.0071, "step": 21250 }, { "epoch": 0.3478687719872372, "grad_norm": 0.6641151905059814, "learning_rate": 8.69671929968093e-06, "loss": 0.0048, "step": 21260 }, { "epoch": 0.34803239793831303, "grad_norm": 0.22105421125888824, "learning_rate": 8.700809948457825e-06, "loss": 0.0055, "step": 21270 }, { "epoch": 0.34819602388938886, "grad_norm": 0.015857327729463577, "learning_rate": 8.704900597234722e-06, "loss": 0.0033, "step": 21280 }, { "epoch": 0.3483596498404647, "grad_norm": 0.2630769610404968, "learning_rate": 8.708991246011619e-06, "loss": 0.0039, "step": 21290 }, { "epoch": 0.3485232757915405, "grad_norm": 0.5350326299667358, "learning_rate": 8.713081894788515e-06, "loss": 0.0038, "step": 21300 }, { "epoch": 0.3486869017426164, "grad_norm": 0.11582928150892258, "learning_rate": 8.71717254356541e-06, "loss": 0.0057, "step": 21310 }, { "epoch": 0.34885052769369224, "grad_norm": 0.11121343821287155, "learning_rate": 8.721263192342306e-06, "loss": 0.0048, "step": 21320 }, { "epoch": 0.34901415364476807, "grad_norm": 0.04576209560036659, "learning_rate": 8.725353841119201e-06, "loss": 0.0035, "step": 21330 }, { "epoch": 0.3491777795958439, "grad_norm": 0.33002373576164246, "learning_rate": 8.729444489896098e-06, "loss": 0.0063, "step": 21340 }, { "epoch": 0.3493414055469197, "grad_norm": 0.1032777652144432, "learning_rate": 8.733535138672994e-06, "loss": 0.0037, "step": 21350 }, { "epoch": 0.34950503149799556, "grad_norm": 0.07112009078264236, "learning_rate": 8.737625787449891e-06, "loss": 0.0064, "step": 21360 }, { "epoch": 0.34966865744907144, "grad_norm": 0.05648186802864075, "learning_rate": 8.741716436226786e-06, "loss": 0.0051, "step": 21370 }, { "epoch": 0.34983228340014727, "grad_norm": 0.0511622317135334, "learning_rate": 8.745807085003682e-06, "loss": 0.0043, "step": 21380 }, { "epoch": 0.3499959093512231, "grad_norm": 0.14490967988967896, "learning_rate": 8.749897733780577e-06, "loss": 0.0104, "step": 21390 }, { "epoch": 0.35015953530229893, "grad_norm": 0.1327885538339615, "learning_rate": 8.753988382557474e-06, "loss": 0.0045, "step": 21400 }, { "epoch": 0.35032316125337476, "grad_norm": 0.26864880323410034, "learning_rate": 8.75807903133437e-06, "loss": 0.0052, "step": 21410 }, { "epoch": 0.35048678720445064, "grad_norm": 0.21027100086212158, "learning_rate": 8.762169680111267e-06, "loss": 0.0043, "step": 21420 }, { "epoch": 0.3506504131555265, "grad_norm": 0.12389078736305237, "learning_rate": 8.766260328888162e-06, "loss": 0.0033, "step": 21430 }, { "epoch": 0.3508140391066023, "grad_norm": 0.28328317403793335, "learning_rate": 8.770350977665058e-06, "loss": 0.0025, "step": 21440 }, { "epoch": 0.35097766505767813, "grad_norm": 0.17781242728233337, "learning_rate": 8.774441626441955e-06, "loss": 0.0068, "step": 21450 }, { "epoch": 0.35114129100875396, "grad_norm": 0.2873762249946594, "learning_rate": 8.778532275218851e-06, "loss": 0.0069, "step": 21460 }, { "epoch": 0.35130491695982985, "grad_norm": 0.49797704815864563, "learning_rate": 8.782622923995746e-06, "loss": 0.006, "step": 21470 }, { "epoch": 0.3514685429109057, "grad_norm": 0.2703474760055542, "learning_rate": 8.786713572772643e-06, "loss": 0.0047, "step": 21480 }, { "epoch": 0.3516321688619815, "grad_norm": 0.19876666367053986, "learning_rate": 8.790804221549538e-06, "loss": 0.0069, "step": 21490 }, { "epoch": 0.35179579481305734, "grad_norm": 0.5667471289634705, "learning_rate": 8.794894870326434e-06, "loss": 0.0062, "step": 21500 }, { "epoch": 0.35195942076413317, "grad_norm": 0.11301826685667038, "learning_rate": 8.798985519103331e-06, "loss": 0.004, "step": 21510 }, { "epoch": 0.35212304671520905, "grad_norm": 0.08300937712192535, "learning_rate": 8.803076167880227e-06, "loss": 0.0034, "step": 21520 }, { "epoch": 0.3522866726662849, "grad_norm": 0.2404436618089676, "learning_rate": 8.807166816657122e-06, "loss": 0.0047, "step": 21530 }, { "epoch": 0.3524502986173607, "grad_norm": 0.13877421617507935, "learning_rate": 8.811257465434019e-06, "loss": 0.0046, "step": 21540 }, { "epoch": 0.35261392456843654, "grad_norm": 0.1315915286540985, "learning_rate": 8.815348114210914e-06, "loss": 0.0032, "step": 21550 }, { "epoch": 0.3527775505195124, "grad_norm": 0.22448566555976868, "learning_rate": 8.81943876298781e-06, "loss": 0.004, "step": 21560 }, { "epoch": 0.35294117647058826, "grad_norm": 0.2019607126712799, "learning_rate": 8.823529411764707e-06, "loss": 0.0037, "step": 21570 }, { "epoch": 0.3531048024216641, "grad_norm": 0.13282088935375214, "learning_rate": 8.827620060541603e-06, "loss": 0.0055, "step": 21580 }, { "epoch": 0.3532684283727399, "grad_norm": 0.180799201130867, "learning_rate": 8.831710709318498e-06, "loss": 0.0043, "step": 21590 }, { "epoch": 0.35343205432381575, "grad_norm": 0.13726092875003815, "learning_rate": 8.835801358095395e-06, "loss": 0.0044, "step": 21600 }, { "epoch": 0.3535956802748916, "grad_norm": 0.1499367505311966, "learning_rate": 8.83989200687229e-06, "loss": 0.0059, "step": 21610 }, { "epoch": 0.35375930622596746, "grad_norm": 0.17077206075191498, "learning_rate": 8.843982655649186e-06, "loss": 0.0053, "step": 21620 }, { "epoch": 0.3539229321770433, "grad_norm": 0.1917620450258255, "learning_rate": 8.848073304426083e-06, "loss": 0.0051, "step": 21630 }, { "epoch": 0.3540865581281191, "grad_norm": 0.27524930238723755, "learning_rate": 8.85216395320298e-06, "loss": 0.0058, "step": 21640 }, { "epoch": 0.35425018407919495, "grad_norm": 0.23236794769763947, "learning_rate": 8.856254601979874e-06, "loss": 0.0036, "step": 21650 }, { "epoch": 0.3544138100302708, "grad_norm": 0.20862087607383728, "learning_rate": 8.86034525075677e-06, "loss": 0.0087, "step": 21660 }, { "epoch": 0.35457743598134667, "grad_norm": 0.18572506308555603, "learning_rate": 8.864435899533667e-06, "loss": 0.0062, "step": 21670 }, { "epoch": 0.3547410619324225, "grad_norm": 0.22928810119628906, "learning_rate": 8.868526548310564e-06, "loss": 0.0033, "step": 21680 }, { "epoch": 0.3549046878834983, "grad_norm": 0.18449503183364868, "learning_rate": 8.872617197087459e-06, "loss": 0.0095, "step": 21690 }, { "epoch": 0.35506831383457416, "grad_norm": 0.06720693409442902, "learning_rate": 8.876707845864355e-06, "loss": 0.0037, "step": 21700 }, { "epoch": 0.35523193978565, "grad_norm": 0.24558401107788086, "learning_rate": 8.88079849464125e-06, "loss": 0.0053, "step": 21710 }, { "epoch": 0.35539556573672587, "grad_norm": 0.13239549100399017, "learning_rate": 8.884889143418147e-06, "loss": 0.0043, "step": 21720 }, { "epoch": 0.3555591916878017, "grad_norm": 0.0794357880949974, "learning_rate": 8.888979792195043e-06, "loss": 0.0043, "step": 21730 }, { "epoch": 0.35572281763887753, "grad_norm": 0.16611962020397186, "learning_rate": 8.89307044097194e-06, "loss": 0.0052, "step": 21740 }, { "epoch": 0.35588644358995336, "grad_norm": 0.17523154616355896, "learning_rate": 8.897161089748835e-06, "loss": 0.0048, "step": 21750 }, { "epoch": 0.3560500695410292, "grad_norm": 0.14425332844257355, "learning_rate": 8.901251738525731e-06, "loss": 0.0043, "step": 21760 }, { "epoch": 0.356213695492105, "grad_norm": 0.15739698708057404, "learning_rate": 8.905342387302626e-06, "loss": 0.0067, "step": 21770 }, { "epoch": 0.3563773214431809, "grad_norm": 0.41962000727653503, "learning_rate": 8.909433036079523e-06, "loss": 0.0064, "step": 21780 }, { "epoch": 0.35654094739425674, "grad_norm": 0.1400892734527588, "learning_rate": 8.913523684856419e-06, "loss": 0.0027, "step": 21790 }, { "epoch": 0.35670457334533257, "grad_norm": 0.31727856397628784, "learning_rate": 8.917614333633316e-06, "loss": 0.0083, "step": 21800 }, { "epoch": 0.3568681992964084, "grad_norm": 0.17142167687416077, "learning_rate": 8.92170498241021e-06, "loss": 0.0038, "step": 21810 }, { "epoch": 0.3570318252474842, "grad_norm": 0.21454480290412903, "learning_rate": 8.925795631187107e-06, "loss": 0.0045, "step": 21820 }, { "epoch": 0.3571954511985601, "grad_norm": 0.17677193880081177, "learning_rate": 8.929886279964004e-06, "loss": 0.0037, "step": 21830 }, { "epoch": 0.35735907714963594, "grad_norm": 0.2132851481437683, "learning_rate": 8.9339769287409e-06, "loss": 0.0039, "step": 21840 }, { "epoch": 0.35752270310071177, "grad_norm": 0.05725911259651184, "learning_rate": 8.938067577517795e-06, "loss": 0.0047, "step": 21850 }, { "epoch": 0.3576863290517876, "grad_norm": 0.15161773562431335, "learning_rate": 8.942158226294692e-06, "loss": 0.0038, "step": 21860 }, { "epoch": 0.35784995500286343, "grad_norm": 0.09197467565536499, "learning_rate": 8.946248875071587e-06, "loss": 0.0047, "step": 21870 }, { "epoch": 0.3580135809539393, "grad_norm": 0.5142105221748352, "learning_rate": 8.950339523848483e-06, "loss": 0.0034, "step": 21880 }, { "epoch": 0.35817720690501514, "grad_norm": 0.12274603545665741, "learning_rate": 8.95443017262538e-06, "loss": 0.0046, "step": 21890 }, { "epoch": 0.358340832856091, "grad_norm": 0.37482950091362, "learning_rate": 8.958520821402276e-06, "loss": 0.0062, "step": 21900 }, { "epoch": 0.3585044588071668, "grad_norm": 0.23216994106769562, "learning_rate": 8.962611470179171e-06, "loss": 0.0055, "step": 21910 }, { "epoch": 0.35866808475824263, "grad_norm": 0.24686093628406525, "learning_rate": 8.966702118956068e-06, "loss": 0.0055, "step": 21920 }, { "epoch": 0.3588317107093185, "grad_norm": 0.17744821310043335, "learning_rate": 8.970792767732962e-06, "loss": 0.0068, "step": 21930 }, { "epoch": 0.35899533666039435, "grad_norm": 0.05231819301843643, "learning_rate": 8.974883416509859e-06, "loss": 0.0033, "step": 21940 }, { "epoch": 0.3591589626114702, "grad_norm": 0.1520070731639862, "learning_rate": 8.978974065286756e-06, "loss": 0.005, "step": 21950 }, { "epoch": 0.359322588562546, "grad_norm": 0.4787937104701996, "learning_rate": 8.983064714063652e-06, "loss": 0.0033, "step": 21960 }, { "epoch": 0.35948621451362184, "grad_norm": 0.04311678186058998, "learning_rate": 8.987155362840547e-06, "loss": 0.0045, "step": 21970 }, { "epoch": 0.3596498404646977, "grad_norm": 0.06038455292582512, "learning_rate": 8.991246011617444e-06, "loss": 0.0063, "step": 21980 }, { "epoch": 0.35981346641577355, "grad_norm": 0.11675078421831131, "learning_rate": 8.99533666039434e-06, "loss": 0.0051, "step": 21990 }, { "epoch": 0.3599770923668494, "grad_norm": 0.027839265763759613, "learning_rate": 8.999427309171235e-06, "loss": 0.0046, "step": 22000 }, { "epoch": 0.3601407183179252, "grad_norm": 0.10224975645542145, "learning_rate": 9.003517957948132e-06, "loss": 0.0034, "step": 22010 }, { "epoch": 0.36030434426900104, "grad_norm": 0.24953122437000275, "learning_rate": 9.007608606725026e-06, "loss": 0.0056, "step": 22020 }, { "epoch": 0.3604679702200769, "grad_norm": 0.188195139169693, "learning_rate": 9.011699255501923e-06, "loss": 0.0056, "step": 22030 }, { "epoch": 0.36063159617115276, "grad_norm": 0.10194826871156693, "learning_rate": 9.01578990427882e-06, "loss": 0.0053, "step": 22040 }, { "epoch": 0.3607952221222286, "grad_norm": 0.03883948549628258, "learning_rate": 9.019880553055716e-06, "loss": 0.0038, "step": 22050 }, { "epoch": 0.3609588480733044, "grad_norm": 0.06035114824771881, "learning_rate": 9.023971201832611e-06, "loss": 0.0044, "step": 22060 }, { "epoch": 0.36112247402438025, "grad_norm": 0.5380794405937195, "learning_rate": 9.028061850609507e-06, "loss": 0.0066, "step": 22070 }, { "epoch": 0.36128609997545613, "grad_norm": 0.5984293818473816, "learning_rate": 9.032152499386402e-06, "loss": 0.0051, "step": 22080 }, { "epoch": 0.36144972592653196, "grad_norm": 0.14698141813278198, "learning_rate": 9.036243148163299e-06, "loss": 0.0046, "step": 22090 }, { "epoch": 0.3616133518776078, "grad_norm": 0.2986476421356201, "learning_rate": 9.040333796940195e-06, "loss": 0.0051, "step": 22100 }, { "epoch": 0.3617769778286836, "grad_norm": 0.09871439635753632, "learning_rate": 9.044424445717092e-06, "loss": 0.005, "step": 22110 }, { "epoch": 0.36194060377975945, "grad_norm": 0.3814399540424347, "learning_rate": 9.048515094493987e-06, "loss": 0.004, "step": 22120 }, { "epoch": 0.36210422973083534, "grad_norm": 0.1017717570066452, "learning_rate": 9.052605743270883e-06, "loss": 0.0061, "step": 22130 }, { "epoch": 0.36226785568191117, "grad_norm": 0.15506720542907715, "learning_rate": 9.056696392047778e-06, "loss": 0.0051, "step": 22140 }, { "epoch": 0.362431481632987, "grad_norm": 0.11817537248134613, "learning_rate": 9.060787040824675e-06, "loss": 0.0051, "step": 22150 }, { "epoch": 0.3625951075840628, "grad_norm": 0.06475069373846054, "learning_rate": 9.064877689601571e-06, "loss": 0.0057, "step": 22160 }, { "epoch": 0.36275873353513866, "grad_norm": 0.24207590520381927, "learning_rate": 9.068968338378468e-06, "loss": 0.0034, "step": 22170 }, { "epoch": 0.36292235948621454, "grad_norm": 0.07822246104478836, "learning_rate": 9.073058987155363e-06, "loss": 0.0062, "step": 22180 }, { "epoch": 0.36308598543729037, "grad_norm": 0.08871735632419586, "learning_rate": 9.07714963593226e-06, "loss": 0.0068, "step": 22190 }, { "epoch": 0.3632496113883662, "grad_norm": 0.19820338487625122, "learning_rate": 9.081240284709156e-06, "loss": 0.005, "step": 22200 }, { "epoch": 0.36341323733944203, "grad_norm": 0.13168364763259888, "learning_rate": 9.085330933486052e-06, "loss": 0.0061, "step": 22210 }, { "epoch": 0.36357686329051786, "grad_norm": 0.1346927434206009, "learning_rate": 9.089421582262947e-06, "loss": 0.0052, "step": 22220 }, { "epoch": 0.3637404892415937, "grad_norm": 0.20508906245231628, "learning_rate": 9.093512231039844e-06, "loss": 0.0071, "step": 22230 }, { "epoch": 0.3639041151926696, "grad_norm": 0.09205359220504761, "learning_rate": 9.097602879816739e-06, "loss": 0.0043, "step": 22240 }, { "epoch": 0.3640677411437454, "grad_norm": 0.1732301115989685, "learning_rate": 9.101693528593635e-06, "loss": 0.0059, "step": 22250 }, { "epoch": 0.36423136709482123, "grad_norm": 0.041214555501937866, "learning_rate": 9.105784177370532e-06, "loss": 0.0045, "step": 22260 }, { "epoch": 0.36439499304589706, "grad_norm": 0.09344473481178284, "learning_rate": 9.109874826147428e-06, "loss": 0.0048, "step": 22270 }, { "epoch": 0.3645586189969729, "grad_norm": 0.13867849111557007, "learning_rate": 9.113965474924323e-06, "loss": 0.0042, "step": 22280 }, { "epoch": 0.3647222449480488, "grad_norm": 0.2681144177913666, "learning_rate": 9.11805612370122e-06, "loss": 0.0039, "step": 22290 }, { "epoch": 0.3648858708991246, "grad_norm": 0.22986134886741638, "learning_rate": 9.122146772478115e-06, "loss": 0.0077, "step": 22300 }, { "epoch": 0.36504949685020044, "grad_norm": 0.2135639637708664, "learning_rate": 9.126237421255011e-06, "loss": 0.0043, "step": 22310 }, { "epoch": 0.36521312280127627, "grad_norm": 0.043512701988220215, "learning_rate": 9.130328070031908e-06, "loss": 0.0032, "step": 22320 }, { "epoch": 0.3653767487523521, "grad_norm": 0.2695474624633789, "learning_rate": 9.134418718808804e-06, "loss": 0.0034, "step": 22330 }, { "epoch": 0.365540374703428, "grad_norm": 0.07508651167154312, "learning_rate": 9.1385093675857e-06, "loss": 0.0035, "step": 22340 }, { "epoch": 0.3657040006545038, "grad_norm": 0.014311243779957294, "learning_rate": 9.142600016362596e-06, "loss": 0.0068, "step": 22350 }, { "epoch": 0.36586762660557964, "grad_norm": 0.1704816073179245, "learning_rate": 9.146690665139492e-06, "loss": 0.003, "step": 22360 }, { "epoch": 0.3660312525566555, "grad_norm": 0.05056550353765488, "learning_rate": 9.150781313916389e-06, "loss": 0.0045, "step": 22370 }, { "epoch": 0.3661948785077313, "grad_norm": 0.0928158313035965, "learning_rate": 9.154871962693284e-06, "loss": 0.0044, "step": 22380 }, { "epoch": 0.3663585044588072, "grad_norm": 0.255362868309021, "learning_rate": 9.15896261147018e-06, "loss": 0.0046, "step": 22390 }, { "epoch": 0.366522130409883, "grad_norm": 0.07483069598674774, "learning_rate": 9.163053260247075e-06, "loss": 0.0029, "step": 22400 }, { "epoch": 0.36668575636095885, "grad_norm": 0.08020536601543427, "learning_rate": 9.167143909023972e-06, "loss": 0.0046, "step": 22410 }, { "epoch": 0.3668493823120347, "grad_norm": 0.046727146953344345, "learning_rate": 9.171234557800868e-06, "loss": 0.0044, "step": 22420 }, { "epoch": 0.3670130082631105, "grad_norm": 0.2200734168291092, "learning_rate": 9.175325206577765e-06, "loss": 0.003, "step": 22430 }, { "epoch": 0.3671766342141864, "grad_norm": 0.04855937510728836, "learning_rate": 9.17941585535466e-06, "loss": 0.0046, "step": 22440 }, { "epoch": 0.3673402601652622, "grad_norm": 0.13849368691444397, "learning_rate": 9.183506504131556e-06, "loss": 0.0029, "step": 22450 }, { "epoch": 0.36750388611633805, "grad_norm": 0.18201111257076263, "learning_rate": 9.187597152908451e-06, "loss": 0.0046, "step": 22460 }, { "epoch": 0.3676675120674139, "grad_norm": 0.2613840401172638, "learning_rate": 9.191687801685348e-06, "loss": 0.0044, "step": 22470 }, { "epoch": 0.3678311380184897, "grad_norm": 0.21508777141571045, "learning_rate": 9.195778450462244e-06, "loss": 0.0057, "step": 22480 }, { "epoch": 0.3679947639695656, "grad_norm": 0.08759069442749023, "learning_rate": 9.19986909923914e-06, "loss": 0.006, "step": 22490 }, { "epoch": 0.3681583899206414, "grad_norm": 0.06278969347476959, "learning_rate": 9.203959748016036e-06, "loss": 0.0035, "step": 22500 }, { "epoch": 0.36832201587171726, "grad_norm": 0.1803193837404251, "learning_rate": 9.208050396792932e-06, "loss": 0.0072, "step": 22510 }, { "epoch": 0.3684856418227931, "grad_norm": 0.0776112899184227, "learning_rate": 9.212141045569829e-06, "loss": 0.0046, "step": 22520 }, { "epoch": 0.3686492677738689, "grad_norm": 0.12017883360385895, "learning_rate": 9.216231694346725e-06, "loss": 0.0039, "step": 22530 }, { "epoch": 0.3688128937249448, "grad_norm": 0.27362507581710815, "learning_rate": 9.22032234312362e-06, "loss": 0.0049, "step": 22540 }, { "epoch": 0.36897651967602063, "grad_norm": 0.1422208994626999, "learning_rate": 9.224412991900517e-06, "loss": 0.0055, "step": 22550 }, { "epoch": 0.36914014562709646, "grad_norm": 0.14262330532073975, "learning_rate": 9.228503640677412e-06, "loss": 0.0039, "step": 22560 }, { "epoch": 0.3693037715781723, "grad_norm": 0.08472730964422226, "learning_rate": 9.232594289454308e-06, "loss": 0.0047, "step": 22570 }, { "epoch": 0.3694673975292481, "grad_norm": 0.14583709836006165, "learning_rate": 9.236684938231205e-06, "loss": 0.0032, "step": 22580 }, { "epoch": 0.369631023480324, "grad_norm": 0.10728612542152405, "learning_rate": 9.240775587008101e-06, "loss": 0.0041, "step": 22590 }, { "epoch": 0.36979464943139984, "grad_norm": 0.1927308887243271, "learning_rate": 9.244866235784996e-06, "loss": 0.0052, "step": 22600 }, { "epoch": 0.36995827538247567, "grad_norm": 0.17712165415287018, "learning_rate": 9.248956884561893e-06, "loss": 0.0048, "step": 22610 }, { "epoch": 0.3701219013335515, "grad_norm": 0.16923968493938446, "learning_rate": 9.253047533338787e-06, "loss": 0.0045, "step": 22620 }, { "epoch": 0.3702855272846273, "grad_norm": 0.21556515991687775, "learning_rate": 9.257138182115684e-06, "loss": 0.0076, "step": 22630 }, { "epoch": 0.3704491532357032, "grad_norm": 0.09332364052534103, "learning_rate": 9.26122883089258e-06, "loss": 0.0041, "step": 22640 }, { "epoch": 0.37061277918677904, "grad_norm": 0.2369449883699417, "learning_rate": 9.265319479669477e-06, "loss": 0.0056, "step": 22650 }, { "epoch": 0.37077640513785487, "grad_norm": 0.18861030042171478, "learning_rate": 9.269410128446372e-06, "loss": 0.0064, "step": 22660 }, { "epoch": 0.3709400310889307, "grad_norm": 0.2040271759033203, "learning_rate": 9.273500777223269e-06, "loss": 0.0087, "step": 22670 }, { "epoch": 0.37110365704000653, "grad_norm": 0.3186517059803009, "learning_rate": 9.277591426000165e-06, "loss": 0.0065, "step": 22680 }, { "epoch": 0.37126728299108236, "grad_norm": 0.06650704890489578, "learning_rate": 9.281682074777062e-06, "loss": 0.0045, "step": 22690 }, { "epoch": 0.37143090894215824, "grad_norm": 0.12271928787231445, "learning_rate": 9.285772723553957e-06, "loss": 0.006, "step": 22700 }, { "epoch": 0.3715945348932341, "grad_norm": 0.09046670794487, "learning_rate": 9.289863372330853e-06, "loss": 0.0036, "step": 22710 }, { "epoch": 0.3717581608443099, "grad_norm": 0.303998738527298, "learning_rate": 9.293954021107748e-06, "loss": 0.0028, "step": 22720 }, { "epoch": 0.37192178679538573, "grad_norm": 0.05583928897976875, "learning_rate": 9.298044669884644e-06, "loss": 0.0037, "step": 22730 }, { "epoch": 0.37208541274646156, "grad_norm": 0.2206975519657135, "learning_rate": 9.302135318661541e-06, "loss": 0.0024, "step": 22740 }, { "epoch": 0.37224903869753745, "grad_norm": 0.38729727268218994, "learning_rate": 9.306225967438438e-06, "loss": 0.0039, "step": 22750 }, { "epoch": 0.3724126646486133, "grad_norm": 0.29606953263282776, "learning_rate": 9.310316616215332e-06, "loss": 0.0042, "step": 22760 }, { "epoch": 0.3725762905996891, "grad_norm": 0.06541324406862259, "learning_rate": 9.314407264992229e-06, "loss": 0.0033, "step": 22770 }, { "epoch": 0.37273991655076494, "grad_norm": 0.1531689316034317, "learning_rate": 9.318497913769124e-06, "loss": 0.0035, "step": 22780 }, { "epoch": 0.37290354250184077, "grad_norm": 0.009701278060674667, "learning_rate": 9.32258856254602e-06, "loss": 0.0046, "step": 22790 }, { "epoch": 0.37306716845291665, "grad_norm": 0.18347367644309998, "learning_rate": 9.326679211322917e-06, "loss": 0.0062, "step": 22800 }, { "epoch": 0.3732307944039925, "grad_norm": 0.09910908341407776, "learning_rate": 9.330769860099814e-06, "loss": 0.0043, "step": 22810 }, { "epoch": 0.3733944203550683, "grad_norm": 0.20214706659317017, "learning_rate": 9.334860508876708e-06, "loss": 0.0048, "step": 22820 }, { "epoch": 0.37355804630614414, "grad_norm": 0.17023234069347382, "learning_rate": 9.338951157653605e-06, "loss": 0.0042, "step": 22830 }, { "epoch": 0.37372167225722, "grad_norm": 0.1548735499382019, "learning_rate": 9.3430418064305e-06, "loss": 0.0032, "step": 22840 }, { "epoch": 0.37388529820829586, "grad_norm": 0.1988033503293991, "learning_rate": 9.347132455207396e-06, "loss": 0.0042, "step": 22850 }, { "epoch": 0.3740489241593717, "grad_norm": 0.14209190011024475, "learning_rate": 9.351223103984293e-06, "loss": 0.0063, "step": 22860 }, { "epoch": 0.3742125501104475, "grad_norm": 0.13470007479190826, "learning_rate": 9.355313752761188e-06, "loss": 0.0047, "step": 22870 }, { "epoch": 0.37437617606152335, "grad_norm": 0.1602434515953064, "learning_rate": 9.359404401538084e-06, "loss": 0.0049, "step": 22880 }, { "epoch": 0.3745398020125992, "grad_norm": 0.22728559374809265, "learning_rate": 9.363495050314981e-06, "loss": 0.0036, "step": 22890 }, { "epoch": 0.37470342796367506, "grad_norm": 0.1100834533572197, "learning_rate": 9.367585699091877e-06, "loss": 0.0033, "step": 22900 }, { "epoch": 0.3748670539147509, "grad_norm": 0.11184380203485489, "learning_rate": 9.371676347868772e-06, "loss": 0.004, "step": 22910 }, { "epoch": 0.3750306798658267, "grad_norm": 0.08237633109092712, "learning_rate": 9.375766996645669e-06, "loss": 0.0032, "step": 22920 }, { "epoch": 0.37519430581690255, "grad_norm": 0.11020543426275253, "learning_rate": 9.379857645422564e-06, "loss": 0.003, "step": 22930 }, { "epoch": 0.3753579317679784, "grad_norm": 0.24248459935188293, "learning_rate": 9.38394829419946e-06, "loss": 0.0052, "step": 22940 }, { "epoch": 0.37552155771905427, "grad_norm": 0.13667307794094086, "learning_rate": 9.388038942976357e-06, "loss": 0.0047, "step": 22950 }, { "epoch": 0.3756851836701301, "grad_norm": 0.2448524534702301, "learning_rate": 9.392129591753253e-06, "loss": 0.0037, "step": 22960 }, { "epoch": 0.3758488096212059, "grad_norm": 0.09055765718221664, "learning_rate": 9.396220240530148e-06, "loss": 0.0035, "step": 22970 }, { "epoch": 0.37601243557228176, "grad_norm": 0.10678678005933762, "learning_rate": 9.400310889307045e-06, "loss": 0.003, "step": 22980 }, { "epoch": 0.3761760615233576, "grad_norm": 0.23421886563301086, "learning_rate": 9.40440153808394e-06, "loss": 0.0055, "step": 22990 }, { "epoch": 0.37633968747443347, "grad_norm": 0.17917338013648987, "learning_rate": 9.408492186860836e-06, "loss": 0.0058, "step": 23000 }, { "epoch": 0.3765033134255093, "grad_norm": 0.1587824821472168, "learning_rate": 9.412582835637733e-06, "loss": 0.0045, "step": 23010 }, { "epoch": 0.37666693937658513, "grad_norm": 0.0629500299692154, "learning_rate": 9.41667348441463e-06, "loss": 0.0045, "step": 23020 }, { "epoch": 0.37683056532766096, "grad_norm": 0.21350431442260742, "learning_rate": 9.420764133191524e-06, "loss": 0.0043, "step": 23030 }, { "epoch": 0.3769941912787368, "grad_norm": 0.07687771320343018, "learning_rate": 9.42485478196842e-06, "loss": 0.0064, "step": 23040 }, { "epoch": 0.3771578172298127, "grad_norm": 0.2771693468093872, "learning_rate": 9.428945430745317e-06, "loss": 0.0055, "step": 23050 }, { "epoch": 0.3773214431808885, "grad_norm": 0.12499131262302399, "learning_rate": 9.433036079522214e-06, "loss": 0.0044, "step": 23060 }, { "epoch": 0.37748506913196433, "grad_norm": 0.18079163134098053, "learning_rate": 9.437126728299109e-06, "loss": 0.0032, "step": 23070 }, { "epoch": 0.37764869508304016, "grad_norm": 0.048156045377254486, "learning_rate": 9.441217377076005e-06, "loss": 0.0042, "step": 23080 }, { "epoch": 0.377812321034116, "grad_norm": 0.1754903644323349, "learning_rate": 9.4453080258529e-06, "loss": 0.0048, "step": 23090 }, { "epoch": 0.3779759469851918, "grad_norm": 0.28425905108451843, "learning_rate": 9.449398674629797e-06, "loss": 0.0034, "step": 23100 }, { "epoch": 0.3781395729362677, "grad_norm": 0.07343924790620804, "learning_rate": 9.453489323406693e-06, "loss": 0.0038, "step": 23110 }, { "epoch": 0.37830319888734354, "grad_norm": 0.06746002286672592, "learning_rate": 9.45757997218359e-06, "loss": 0.0032, "step": 23120 }, { "epoch": 0.37846682483841937, "grad_norm": 0.15183481574058533, "learning_rate": 9.461670620960485e-06, "loss": 0.0057, "step": 23130 }, { "epoch": 0.3786304507894952, "grad_norm": 0.31659454107284546, "learning_rate": 9.465761269737381e-06, "loss": 0.0038, "step": 23140 }, { "epoch": 0.37879407674057103, "grad_norm": 0.3467671275138855, "learning_rate": 9.469851918514276e-06, "loss": 0.0051, "step": 23150 }, { "epoch": 0.3789577026916469, "grad_norm": 0.08848075568675995, "learning_rate": 9.473942567291173e-06, "loss": 0.0041, "step": 23160 }, { "epoch": 0.37912132864272274, "grad_norm": 0.289042592048645, "learning_rate": 9.47803321606807e-06, "loss": 0.0048, "step": 23170 }, { "epoch": 0.3792849545937986, "grad_norm": 0.17555688321590424, "learning_rate": 9.482123864844966e-06, "loss": 0.0046, "step": 23180 }, { "epoch": 0.3794485805448744, "grad_norm": 0.07031220197677612, "learning_rate": 9.48621451362186e-06, "loss": 0.0038, "step": 23190 }, { "epoch": 0.37961220649595023, "grad_norm": 0.33143553137779236, "learning_rate": 9.490305162398757e-06, "loss": 0.0033, "step": 23200 }, { "epoch": 0.3797758324470261, "grad_norm": 0.27846482396125793, "learning_rate": 9.494395811175654e-06, "loss": 0.0066, "step": 23210 }, { "epoch": 0.37993945839810195, "grad_norm": 0.24830101430416107, "learning_rate": 9.49848645995255e-06, "loss": 0.0056, "step": 23220 }, { "epoch": 0.3801030843491778, "grad_norm": 0.22130486369132996, "learning_rate": 9.502577108729445e-06, "loss": 0.0038, "step": 23230 }, { "epoch": 0.3802667103002536, "grad_norm": 0.0590287446975708, "learning_rate": 9.506667757506342e-06, "loss": 0.0042, "step": 23240 }, { "epoch": 0.38043033625132944, "grad_norm": 0.1556752473115921, "learning_rate": 9.510758406283237e-06, "loss": 0.0037, "step": 23250 }, { "epoch": 0.3805939622024053, "grad_norm": 0.21244369447231293, "learning_rate": 9.514849055060133e-06, "loss": 0.0034, "step": 23260 }, { "epoch": 0.38075758815348115, "grad_norm": 0.2829434871673584, "learning_rate": 9.51893970383703e-06, "loss": 0.0054, "step": 23270 }, { "epoch": 0.380921214104557, "grad_norm": 0.1576375514268875, "learning_rate": 9.523030352613926e-06, "loss": 0.0036, "step": 23280 }, { "epoch": 0.3810848400556328, "grad_norm": 0.2944735288619995, "learning_rate": 9.527121001390821e-06, "loss": 0.0058, "step": 23290 }, { "epoch": 0.38124846600670864, "grad_norm": 0.1357761174440384, "learning_rate": 9.531211650167718e-06, "loss": 0.0035, "step": 23300 }, { "epoch": 0.3814120919577845, "grad_norm": 0.16953496634960175, "learning_rate": 9.535302298944612e-06, "loss": 0.0056, "step": 23310 }, { "epoch": 0.38157571790886036, "grad_norm": 0.100553497672081, "learning_rate": 9.539392947721509e-06, "loss": 0.0034, "step": 23320 }, { "epoch": 0.3817393438599362, "grad_norm": 0.11355781555175781, "learning_rate": 9.543483596498406e-06, "loss": 0.0039, "step": 23330 }, { "epoch": 0.381902969811012, "grad_norm": 0.08497854322195053, "learning_rate": 9.547574245275302e-06, "loss": 0.0044, "step": 23340 }, { "epoch": 0.38206659576208785, "grad_norm": 0.08253499120473862, "learning_rate": 9.551664894052197e-06, "loss": 0.0044, "step": 23350 }, { "epoch": 0.38223022171316373, "grad_norm": 0.26761943101882935, "learning_rate": 9.555755542829094e-06, "loss": 0.0045, "step": 23360 }, { "epoch": 0.38239384766423956, "grad_norm": 0.22954662144184113, "learning_rate": 9.559846191605988e-06, "loss": 0.0047, "step": 23370 }, { "epoch": 0.3825574736153154, "grad_norm": 0.6663384437561035, "learning_rate": 9.563936840382887e-06, "loss": 0.0054, "step": 23380 }, { "epoch": 0.3827210995663912, "grad_norm": 0.15105237066745758, "learning_rate": 9.568027489159782e-06, "loss": 0.0046, "step": 23390 }, { "epoch": 0.38288472551746705, "grad_norm": 0.10548959672451019, "learning_rate": 9.572118137936678e-06, "loss": 0.004, "step": 23400 }, { "epoch": 0.38304835146854294, "grad_norm": 0.24891695380210876, "learning_rate": 9.576208786713573e-06, "loss": 0.0045, "step": 23410 }, { "epoch": 0.38321197741961877, "grad_norm": 0.08492522686719894, "learning_rate": 9.58029943549047e-06, "loss": 0.0041, "step": 23420 }, { "epoch": 0.3833756033706946, "grad_norm": 0.15219536423683167, "learning_rate": 9.584390084267366e-06, "loss": 0.0071, "step": 23430 }, { "epoch": 0.3835392293217704, "grad_norm": 0.08458176255226135, "learning_rate": 9.588480733044263e-06, "loss": 0.0032, "step": 23440 }, { "epoch": 0.38370285527284625, "grad_norm": 0.1511332392692566, "learning_rate": 9.592571381821157e-06, "loss": 0.0069, "step": 23450 }, { "epoch": 0.38386648122392214, "grad_norm": 0.21422572433948517, "learning_rate": 9.596662030598054e-06, "loss": 0.0038, "step": 23460 }, { "epoch": 0.38403010717499797, "grad_norm": 0.05362638086080551, "learning_rate": 9.600752679374949e-06, "loss": 0.005, "step": 23470 }, { "epoch": 0.3841937331260738, "grad_norm": 0.22935301065444946, "learning_rate": 9.604843328151845e-06, "loss": 0.0054, "step": 23480 }, { "epoch": 0.38435735907714963, "grad_norm": 0.18938273191452026, "learning_rate": 9.608933976928742e-06, "loss": 0.0031, "step": 23490 }, { "epoch": 0.38452098502822546, "grad_norm": 0.22983835637569427, "learning_rate": 9.613024625705639e-06, "loss": 0.0043, "step": 23500 }, { "epoch": 0.38468461097930134, "grad_norm": 0.1469614952802658, "learning_rate": 9.617115274482533e-06, "loss": 0.0033, "step": 23510 }, { "epoch": 0.3848482369303772, "grad_norm": 0.20555458962917328, "learning_rate": 9.62120592325943e-06, "loss": 0.0093, "step": 23520 }, { "epoch": 0.385011862881453, "grad_norm": 0.16882050037384033, "learning_rate": 9.625296572036325e-06, "loss": 0.0044, "step": 23530 }, { "epoch": 0.38517548883252883, "grad_norm": 0.06599155068397522, "learning_rate": 9.629387220813221e-06, "loss": 0.0031, "step": 23540 }, { "epoch": 0.38533911478360466, "grad_norm": 0.11514567583799362, "learning_rate": 9.633477869590118e-06, "loss": 0.0053, "step": 23550 }, { "epoch": 0.3855027407346805, "grad_norm": 0.17259375751018524, "learning_rate": 9.637568518367015e-06, "loss": 0.0031, "step": 23560 }, { "epoch": 0.3856663666857564, "grad_norm": 0.29099732637405396, "learning_rate": 9.64165916714391e-06, "loss": 0.005, "step": 23570 }, { "epoch": 0.3858299926368322, "grad_norm": 0.4632797837257385, "learning_rate": 9.645749815920806e-06, "loss": 0.0043, "step": 23580 }, { "epoch": 0.38599361858790804, "grad_norm": 0.5670897960662842, "learning_rate": 9.649840464697702e-06, "loss": 0.0034, "step": 23590 }, { "epoch": 0.38615724453898387, "grad_norm": 0.23511441051959991, "learning_rate": 9.653931113474599e-06, "loss": 0.0075, "step": 23600 }, { "epoch": 0.3863208704900597, "grad_norm": 0.16084155440330505, "learning_rate": 9.658021762251494e-06, "loss": 0.0049, "step": 23610 }, { "epoch": 0.3864844964411356, "grad_norm": 0.25248411297798157, "learning_rate": 9.66211241102839e-06, "loss": 0.0033, "step": 23620 }, { "epoch": 0.3866481223922114, "grad_norm": 0.24997346103191376, "learning_rate": 9.666203059805285e-06, "loss": 0.0029, "step": 23630 }, { "epoch": 0.38681174834328724, "grad_norm": 0.056109312921762466, "learning_rate": 9.670293708582182e-06, "loss": 0.0049, "step": 23640 }, { "epoch": 0.3869753742943631, "grad_norm": 0.11116239428520203, "learning_rate": 9.674384357359078e-06, "loss": 0.0036, "step": 23650 }, { "epoch": 0.3871390002454389, "grad_norm": 0.2773296535015106, "learning_rate": 9.678475006135975e-06, "loss": 0.0053, "step": 23660 }, { "epoch": 0.3873026261965148, "grad_norm": 0.06932026892900467, "learning_rate": 9.68256565491287e-06, "loss": 0.004, "step": 23670 }, { "epoch": 0.3874662521475906, "grad_norm": 0.09216681122779846, "learning_rate": 9.686656303689766e-06, "loss": 0.0064, "step": 23680 }, { "epoch": 0.38762987809866645, "grad_norm": 0.18986600637435913, "learning_rate": 9.690746952466661e-06, "loss": 0.0053, "step": 23690 }, { "epoch": 0.3877935040497423, "grad_norm": 0.07712690532207489, "learning_rate": 9.694837601243558e-06, "loss": 0.0035, "step": 23700 }, { "epoch": 0.3879571300008181, "grad_norm": 0.054018620401620865, "learning_rate": 9.698928250020454e-06, "loss": 0.005, "step": 23710 }, { "epoch": 0.388120755951894, "grad_norm": 0.12356428056955338, "learning_rate": 9.703018898797351e-06, "loss": 0.0053, "step": 23720 }, { "epoch": 0.3882843819029698, "grad_norm": 0.17293617129325867, "learning_rate": 9.707109547574246e-06, "loss": 0.0036, "step": 23730 }, { "epoch": 0.38844800785404565, "grad_norm": 0.2207067310810089, "learning_rate": 9.711200196351142e-06, "loss": 0.0033, "step": 23740 }, { "epoch": 0.3886116338051215, "grad_norm": 0.12133341282606125, "learning_rate": 9.715290845128039e-06, "loss": 0.0041, "step": 23750 }, { "epoch": 0.3887752597561973, "grad_norm": 0.421698659658432, "learning_rate": 9.719381493904934e-06, "loss": 0.0037, "step": 23760 }, { "epoch": 0.3889388857072732, "grad_norm": 0.10737305879592896, "learning_rate": 9.72347214268183e-06, "loss": 0.0062, "step": 23770 }, { "epoch": 0.389102511658349, "grad_norm": 0.4675261378288269, "learning_rate": 9.727562791458725e-06, "loss": 0.0046, "step": 23780 }, { "epoch": 0.38926613760942486, "grad_norm": 0.12439744919538498, "learning_rate": 9.731653440235622e-06, "loss": 0.0048, "step": 23790 }, { "epoch": 0.3894297635605007, "grad_norm": 0.21090586483478546, "learning_rate": 9.735744089012518e-06, "loss": 0.0066, "step": 23800 }, { "epoch": 0.3895933895115765, "grad_norm": 0.08181467652320862, "learning_rate": 9.739834737789415e-06, "loss": 0.0046, "step": 23810 }, { "epoch": 0.3897570154626524, "grad_norm": 0.17643025517463684, "learning_rate": 9.74392538656631e-06, "loss": 0.0075, "step": 23820 }, { "epoch": 0.38992064141372823, "grad_norm": 0.056413788348436356, "learning_rate": 9.748016035343206e-06, "loss": 0.0051, "step": 23830 }, { "epoch": 0.39008426736480406, "grad_norm": 0.07529550790786743, "learning_rate": 9.752106684120101e-06, "loss": 0.0036, "step": 23840 }, { "epoch": 0.3902478933158799, "grad_norm": 0.06806400418281555, "learning_rate": 9.756197332896998e-06, "loss": 0.0066, "step": 23850 }, { "epoch": 0.3904115192669557, "grad_norm": 0.14307600259780884, "learning_rate": 9.760287981673894e-06, "loss": 0.0038, "step": 23860 }, { "epoch": 0.3905751452180316, "grad_norm": 0.16854174435138702, "learning_rate": 9.76437863045079e-06, "loss": 0.0045, "step": 23870 }, { "epoch": 0.39073877116910744, "grad_norm": 0.29509541392326355, "learning_rate": 9.768469279227686e-06, "loss": 0.0069, "step": 23880 }, { "epoch": 0.39090239712018326, "grad_norm": 0.3630253076553345, "learning_rate": 9.772559928004582e-06, "loss": 0.0035, "step": 23890 }, { "epoch": 0.3910660230712591, "grad_norm": 0.1080494076013565, "learning_rate": 9.776650576781477e-06, "loss": 0.0042, "step": 23900 }, { "epoch": 0.3912296490223349, "grad_norm": 0.026498494669795036, "learning_rate": 9.780741225558375e-06, "loss": 0.0033, "step": 23910 }, { "epoch": 0.3913932749734108, "grad_norm": 0.17583703994750977, "learning_rate": 9.78483187433527e-06, "loss": 0.0033, "step": 23920 }, { "epoch": 0.39155690092448664, "grad_norm": 0.11554347723722458, "learning_rate": 9.788922523112167e-06, "loss": 0.003, "step": 23930 }, { "epoch": 0.39172052687556247, "grad_norm": 0.18944105505943298, "learning_rate": 9.793013171889062e-06, "loss": 0.0058, "step": 23940 }, { "epoch": 0.3918841528266383, "grad_norm": 0.1040097326040268, "learning_rate": 9.797103820665958e-06, "loss": 0.0057, "step": 23950 }, { "epoch": 0.39204777877771413, "grad_norm": 0.28679007291793823, "learning_rate": 9.801194469442855e-06, "loss": 0.004, "step": 23960 }, { "epoch": 0.39221140472878996, "grad_norm": 0.4044707417488098, "learning_rate": 9.805285118219751e-06, "loss": 0.0057, "step": 23970 }, { "epoch": 0.39237503067986584, "grad_norm": 0.0821269229054451, "learning_rate": 9.809375766996646e-06, "loss": 0.0038, "step": 23980 }, { "epoch": 0.3925386566309417, "grad_norm": 0.025821613147854805, "learning_rate": 9.813466415773543e-06, "loss": 0.0053, "step": 23990 }, { "epoch": 0.3927022825820175, "grad_norm": 0.14595523476600647, "learning_rate": 9.817557064550438e-06, "loss": 0.0041, "step": 24000 }, { "epoch": 0.39286590853309333, "grad_norm": 0.201743483543396, "learning_rate": 9.821647713327334e-06, "loss": 0.0048, "step": 24010 }, { "epoch": 0.39302953448416916, "grad_norm": 0.35324200987815857, "learning_rate": 9.82573836210423e-06, "loss": 0.0035, "step": 24020 }, { "epoch": 0.39319316043524505, "grad_norm": 0.19706648588180542, "learning_rate": 9.829829010881127e-06, "loss": 0.0052, "step": 24030 }, { "epoch": 0.3933567863863209, "grad_norm": 0.12601634860038757, "learning_rate": 9.833919659658022e-06, "loss": 0.004, "step": 24040 }, { "epoch": 0.3935204123373967, "grad_norm": 0.1772211194038391, "learning_rate": 9.838010308434919e-06, "loss": 0.0054, "step": 24050 }, { "epoch": 0.39368403828847254, "grad_norm": 0.14259088039398193, "learning_rate": 9.842100957211813e-06, "loss": 0.0026, "step": 24060 }, { "epoch": 0.39384766423954837, "grad_norm": 0.1450604498386383, "learning_rate": 9.84619160598871e-06, "loss": 0.0032, "step": 24070 }, { "epoch": 0.39401129019062425, "grad_norm": 0.4755989909172058, "learning_rate": 9.850282254765607e-06, "loss": 0.0069, "step": 24080 }, { "epoch": 0.3941749161417001, "grad_norm": 0.09204227477312088, "learning_rate": 9.854372903542503e-06, "loss": 0.005, "step": 24090 }, { "epoch": 0.3943385420927759, "grad_norm": 0.05107157304883003, "learning_rate": 9.858463552319398e-06, "loss": 0.0027, "step": 24100 }, { "epoch": 0.39450216804385174, "grad_norm": 0.2852765619754791, "learning_rate": 9.862554201096295e-06, "loss": 0.0042, "step": 24110 }, { "epoch": 0.39466579399492757, "grad_norm": 0.21650436520576477, "learning_rate": 9.866644849873191e-06, "loss": 0.0092, "step": 24120 }, { "epoch": 0.39482941994600346, "grad_norm": 0.8082996010780334, "learning_rate": 9.870735498650088e-06, "loss": 0.0042, "step": 24130 }, { "epoch": 0.3949930458970793, "grad_norm": 0.6313358545303345, "learning_rate": 9.874826147426983e-06, "loss": 0.0039, "step": 24140 }, { "epoch": 0.3951566718481551, "grad_norm": 0.3295305669307709, "learning_rate": 9.878916796203879e-06, "loss": 0.0051, "step": 24150 }, { "epoch": 0.39532029779923095, "grad_norm": 0.6190198659896851, "learning_rate": 9.883007444980774e-06, "loss": 0.0079, "step": 24160 }, { "epoch": 0.3954839237503068, "grad_norm": 0.3497302532196045, "learning_rate": 9.88709809375767e-06, "loss": 0.0045, "step": 24170 }, { "epoch": 0.39564754970138266, "grad_norm": 0.12648363411426544, "learning_rate": 9.891188742534567e-06, "loss": 0.0036, "step": 24180 }, { "epoch": 0.3958111756524585, "grad_norm": 0.11969157308340073, "learning_rate": 9.895279391311464e-06, "loss": 0.0044, "step": 24190 }, { "epoch": 0.3959748016035343, "grad_norm": 0.16596895456314087, "learning_rate": 9.899370040088358e-06, "loss": 0.0034, "step": 24200 }, { "epoch": 0.39613842755461015, "grad_norm": 0.09344743937253952, "learning_rate": 9.903460688865255e-06, "loss": 0.0057, "step": 24210 }, { "epoch": 0.396302053505686, "grad_norm": 0.15033163130283356, "learning_rate": 9.90755133764215e-06, "loss": 0.0051, "step": 24220 }, { "epoch": 0.39646567945676187, "grad_norm": 0.29229214787483215, "learning_rate": 9.911641986419046e-06, "loss": 0.0039, "step": 24230 }, { "epoch": 0.3966293054078377, "grad_norm": 0.18242864310741425, "learning_rate": 9.915732635195943e-06, "loss": 0.0043, "step": 24240 }, { "epoch": 0.3967929313589135, "grad_norm": 0.15203256905078888, "learning_rate": 9.91982328397284e-06, "loss": 0.0036, "step": 24250 }, { "epoch": 0.39695655730998936, "grad_norm": 0.2579713761806488, "learning_rate": 9.923913932749734e-06, "loss": 0.0058, "step": 24260 }, { "epoch": 0.3971201832610652, "grad_norm": 0.15356788039207458, "learning_rate": 9.928004581526631e-06, "loss": 0.0051, "step": 24270 }, { "epoch": 0.39728380921214107, "grad_norm": 0.11238787323236465, "learning_rate": 9.932095230303527e-06, "loss": 0.0033, "step": 24280 }, { "epoch": 0.3974474351632169, "grad_norm": 0.09681972861289978, "learning_rate": 9.936185879080424e-06, "loss": 0.0027, "step": 24290 }, { "epoch": 0.39761106111429273, "grad_norm": 0.15361787378787994, "learning_rate": 9.940276527857319e-06, "loss": 0.004, "step": 24300 }, { "epoch": 0.39777468706536856, "grad_norm": 0.14386487007141113, "learning_rate": 9.944367176634215e-06, "loss": 0.0076, "step": 24310 }, { "epoch": 0.3979383130164444, "grad_norm": 0.4206905961036682, "learning_rate": 9.94845782541111e-06, "loss": 0.0042, "step": 24320 }, { "epoch": 0.3981019389675203, "grad_norm": 0.052798766642808914, "learning_rate": 9.952548474188007e-06, "loss": 0.0028, "step": 24330 }, { "epoch": 0.3982655649185961, "grad_norm": 0.2567293643951416, "learning_rate": 9.956639122964903e-06, "loss": 0.0044, "step": 24340 }, { "epoch": 0.39842919086967193, "grad_norm": 0.09716498851776123, "learning_rate": 9.9607297717418e-06, "loss": 0.0028, "step": 24350 }, { "epoch": 0.39859281682074776, "grad_norm": 0.3586733937263489, "learning_rate": 9.964820420518695e-06, "loss": 0.0056, "step": 24360 }, { "epoch": 0.3987564427718236, "grad_norm": 0.27824023365974426, "learning_rate": 9.968911069295591e-06, "loss": 0.0042, "step": 24370 }, { "epoch": 0.3989200687228995, "grad_norm": 0.2321470081806183, "learning_rate": 9.973001718072486e-06, "loss": 0.0046, "step": 24380 }, { "epoch": 0.3990836946739753, "grad_norm": 0.18242955207824707, "learning_rate": 9.977092366849383e-06, "loss": 0.0045, "step": 24390 }, { "epoch": 0.39924732062505114, "grad_norm": 0.19941671192646027, "learning_rate": 9.98118301562628e-06, "loss": 0.0037, "step": 24400 }, { "epoch": 0.39941094657612697, "grad_norm": 0.10845080763101578, "learning_rate": 9.985273664403176e-06, "loss": 0.003, "step": 24410 }, { "epoch": 0.3995745725272028, "grad_norm": 0.030282288789749146, "learning_rate": 9.98936431318007e-06, "loss": 0.0051, "step": 24420 }, { "epoch": 0.39973819847827863, "grad_norm": 0.14798863232135773, "learning_rate": 9.993454961956967e-06, "loss": 0.0054, "step": 24430 }, { "epoch": 0.3999018244293545, "grad_norm": 0.13766653835773468, "learning_rate": 9.997545610733864e-06, "loss": 0.0031, "step": 24440 }, { "epoch": 0.40006545038043034, "grad_norm": 0.06036052480340004, "learning_rate": 9.99999999184434e-06, "loss": 0.0049, "step": 24450 }, { "epoch": 0.4002290763315062, "grad_norm": 0.1102435365319252, "learning_rate": 9.999999900093168e-06, "loss": 0.0044, "step": 24460 }, { "epoch": 0.400392702282582, "grad_norm": 0.12202160805463791, "learning_rate": 9.99999970639625e-06, "loss": 0.0033, "step": 24470 }, { "epoch": 0.40055632823365783, "grad_norm": 0.09905244410037994, "learning_rate": 9.99999941075359e-06, "loss": 0.0042, "step": 24480 }, { "epoch": 0.4007199541847337, "grad_norm": 0.06508225947618484, "learning_rate": 9.999999013165194e-06, "loss": 0.0052, "step": 24490 }, { "epoch": 0.40088358013580955, "grad_norm": 0.12556862831115723, "learning_rate": 9.999998513631071e-06, "loss": 0.0036, "step": 24500 }, { "epoch": 0.4010472060868854, "grad_norm": 0.0585654154419899, "learning_rate": 9.999997912151232e-06, "loss": 0.0043, "step": 24510 }, { "epoch": 0.4012108320379612, "grad_norm": 0.07464836537837982, "learning_rate": 9.999997208725685e-06, "loss": 0.0034, "step": 24520 }, { "epoch": 0.40137445798903704, "grad_norm": 0.06693817675113678, "learning_rate": 9.999996403354448e-06, "loss": 0.0038, "step": 24530 }, { "epoch": 0.4015380839401129, "grad_norm": 0.3313537836074829, "learning_rate": 9.999995496037538e-06, "loss": 0.0078, "step": 24540 }, { "epoch": 0.40170170989118875, "grad_norm": 0.08874393254518509, "learning_rate": 9.99999448677497e-06, "loss": 0.0047, "step": 24550 }, { "epoch": 0.4018653358422646, "grad_norm": 0.36500483751296997, "learning_rate": 9.99999337556677e-06, "loss": 0.0052, "step": 24560 }, { "epoch": 0.4020289617933404, "grad_norm": 0.06597773730754852, "learning_rate": 9.999992162412954e-06, "loss": 0.0043, "step": 24570 }, { "epoch": 0.40219258774441624, "grad_norm": 0.14203812181949615, "learning_rate": 9.999990847313552e-06, "loss": 0.0042, "step": 24580 }, { "epoch": 0.4023562136954921, "grad_norm": 0.17366263270378113, "learning_rate": 9.99998943026859e-06, "loss": 0.0031, "step": 24590 }, { "epoch": 0.40251983964656796, "grad_norm": 0.2920961081981659, "learning_rate": 9.999987911278094e-06, "loss": 0.005, "step": 24600 }, { "epoch": 0.4026834655976438, "grad_norm": 0.10078559070825577, "learning_rate": 9.999986290342099e-06, "loss": 0.0041, "step": 24610 }, { "epoch": 0.4028470915487196, "grad_norm": 0.04006541892886162, "learning_rate": 9.999984567460635e-06, "loss": 0.002, "step": 24620 }, { "epoch": 0.40301071749979545, "grad_norm": 0.08892949670553207, "learning_rate": 9.999982742633736e-06, "loss": 0.003, "step": 24630 }, { "epoch": 0.40317434345087133, "grad_norm": 0.12582775950431824, "learning_rate": 9.999980815861443e-06, "loss": 0.0065, "step": 24640 }, { "epoch": 0.40333796940194716, "grad_norm": 0.08963852375745773, "learning_rate": 9.999978787143793e-06, "loss": 0.0049, "step": 24650 }, { "epoch": 0.403501595353023, "grad_norm": 0.10660353302955627, "learning_rate": 9.999976656480828e-06, "loss": 0.0041, "step": 24660 }, { "epoch": 0.4036652213040988, "grad_norm": 0.05753873661160469, "learning_rate": 9.999974423872591e-06, "loss": 0.0057, "step": 24670 }, { "epoch": 0.40382884725517465, "grad_norm": 0.19236977398395538, "learning_rate": 9.999972089319127e-06, "loss": 0.0039, "step": 24680 }, { "epoch": 0.40399247320625054, "grad_norm": 0.10060194879770279, "learning_rate": 9.999969652820487e-06, "loss": 0.0041, "step": 24690 }, { "epoch": 0.40415609915732637, "grad_norm": 0.04839000478386879, "learning_rate": 9.999967114376716e-06, "loss": 0.0054, "step": 24700 }, { "epoch": 0.4043197251084022, "grad_norm": 0.16000086069107056, "learning_rate": 9.99996447398787e-06, "loss": 0.0038, "step": 24710 }, { "epoch": 0.404483351059478, "grad_norm": 0.1464492827653885, "learning_rate": 9.999961731653998e-06, "loss": 0.0043, "step": 24720 }, { "epoch": 0.40464697701055385, "grad_norm": 0.10031981766223907, "learning_rate": 9.999958887375159e-06, "loss": 0.0035, "step": 24730 }, { "epoch": 0.40481060296162974, "grad_norm": 0.11082616448402405, "learning_rate": 9.99995594115141e-06, "loss": 0.0056, "step": 24740 }, { "epoch": 0.40497422891270557, "grad_norm": 0.24518175423145294, "learning_rate": 9.999952892982815e-06, "loss": 0.0053, "step": 24750 }, { "epoch": 0.4051378548637814, "grad_norm": 0.18666844069957733, "learning_rate": 9.999949742869431e-06, "loss": 0.0044, "step": 24760 }, { "epoch": 0.40530148081485723, "grad_norm": 0.15410616993904114, "learning_rate": 9.999946490811323e-06, "loss": 0.0033, "step": 24770 }, { "epoch": 0.40546510676593306, "grad_norm": 0.13189055025577545, "learning_rate": 9.99994313680856e-06, "loss": 0.0045, "step": 24780 }, { "epoch": 0.40562873271700894, "grad_norm": 0.15981779992580414, "learning_rate": 9.999939680861208e-06, "loss": 0.0052, "step": 24790 }, { "epoch": 0.4057923586680848, "grad_norm": 0.12059137970209122, "learning_rate": 9.999936122969339e-06, "loss": 0.0043, "step": 24800 }, { "epoch": 0.4059559846191606, "grad_norm": 0.24657613039016724, "learning_rate": 9.999932463133023e-06, "loss": 0.0033, "step": 24810 }, { "epoch": 0.40611961057023643, "grad_norm": 0.13819099962711334, "learning_rate": 9.999928701352337e-06, "loss": 0.0037, "step": 24820 }, { "epoch": 0.40628323652131226, "grad_norm": 0.21539317071437836, "learning_rate": 9.999924837627357e-06, "loss": 0.006, "step": 24830 }, { "epoch": 0.40644686247238815, "grad_norm": 0.21200107038021088, "learning_rate": 9.999920871958163e-06, "loss": 0.0047, "step": 24840 }, { "epoch": 0.406610488423464, "grad_norm": 0.22419607639312744, "learning_rate": 9.999916804344834e-06, "loss": 0.0033, "step": 24850 }, { "epoch": 0.4067741143745398, "grad_norm": 0.051459506154060364, "learning_rate": 9.999912634787453e-06, "loss": 0.0031, "step": 24860 }, { "epoch": 0.40693774032561564, "grad_norm": 0.17161522805690765, "learning_rate": 9.999908363286106e-06, "loss": 0.004, "step": 24870 }, { "epoch": 0.40710136627669147, "grad_norm": 0.08215340226888657, "learning_rate": 9.99990398984088e-06, "loss": 0.0037, "step": 24880 }, { "epoch": 0.4072649922277673, "grad_norm": 0.17910446226596832, "learning_rate": 9.999899514451864e-06, "loss": 0.0054, "step": 24890 }, { "epoch": 0.4074286181788432, "grad_norm": 0.14583317935466766, "learning_rate": 9.999894937119149e-06, "loss": 0.0033, "step": 24900 }, { "epoch": 0.407592244129919, "grad_norm": 0.515279233455658, "learning_rate": 9.999890257842828e-06, "loss": 0.0029, "step": 24910 }, { "epoch": 0.40775587008099484, "grad_norm": 0.12449789047241211, "learning_rate": 9.999885476622998e-06, "loss": 0.0034, "step": 24920 }, { "epoch": 0.40791949603207067, "grad_norm": 0.1295752376317978, "learning_rate": 9.999880593459754e-06, "loss": 0.0055, "step": 24930 }, { "epoch": 0.4080831219831465, "grad_norm": 0.21455809473991394, "learning_rate": 9.999875608353199e-06, "loss": 0.0042, "step": 24940 }, { "epoch": 0.4082467479342224, "grad_norm": 0.19430391490459442, "learning_rate": 9.99987052130343e-06, "loss": 0.0053, "step": 24950 }, { "epoch": 0.4084103738852982, "grad_norm": 0.17783723771572113, "learning_rate": 9.999865332310556e-06, "loss": 0.0044, "step": 24960 }, { "epoch": 0.40857399983637405, "grad_norm": 0.0693744346499443, "learning_rate": 9.999860041374679e-06, "loss": 0.0064, "step": 24970 }, { "epoch": 0.4087376257874499, "grad_norm": 0.1459283083677292, "learning_rate": 9.999854648495907e-06, "loss": 0.0056, "step": 24980 }, { "epoch": 0.4089012517385257, "grad_norm": 0.10276496410369873, "learning_rate": 9.999849153674351e-06, "loss": 0.0041, "step": 24990 }, { "epoch": 0.4090648776896016, "grad_norm": 0.10199088603258133, "learning_rate": 9.999843556910125e-06, "loss": 0.0036, "step": 25000 }, { "epoch": 0.4092285036406774, "grad_norm": 0.22493737936019897, "learning_rate": 9.999837858203338e-06, "loss": 0.0057, "step": 25010 }, { "epoch": 0.40939212959175325, "grad_norm": 0.07949916273355484, "learning_rate": 9.999832057554111e-06, "loss": 0.0044, "step": 25020 }, { "epoch": 0.4095557555428291, "grad_norm": 0.16311852633953094, "learning_rate": 9.999826154962562e-06, "loss": 0.0039, "step": 25030 }, { "epoch": 0.4097193814939049, "grad_norm": 0.0386163704097271, "learning_rate": 9.999820150428808e-06, "loss": 0.0064, "step": 25040 }, { "epoch": 0.4098830074449808, "grad_norm": 0.13025905191898346, "learning_rate": 9.999814043952974e-06, "loss": 0.0044, "step": 25050 }, { "epoch": 0.4100466333960566, "grad_norm": 0.2470933198928833, "learning_rate": 9.999807835535184e-06, "loss": 0.0057, "step": 25060 }, { "epoch": 0.41021025934713246, "grad_norm": 0.01312474999576807, "learning_rate": 9.999801525175564e-06, "loss": 0.0024, "step": 25070 }, { "epoch": 0.4103738852982083, "grad_norm": 0.0827040895819664, "learning_rate": 9.999795112874242e-06, "loss": 0.004, "step": 25080 }, { "epoch": 0.4105375112492841, "grad_norm": 0.054043009877204895, "learning_rate": 9.999788598631351e-06, "loss": 0.0036, "step": 25090 }, { "epoch": 0.41070113720036, "grad_norm": 0.11571065336465836, "learning_rate": 9.999781982447024e-06, "loss": 0.0043, "step": 25100 }, { "epoch": 0.41086476315143583, "grad_norm": 0.12566331028938293, "learning_rate": 9.999775264321392e-06, "loss": 0.004, "step": 25110 }, { "epoch": 0.41102838910251166, "grad_norm": 0.2587388753890991, "learning_rate": 9.999768444254596e-06, "loss": 0.0042, "step": 25120 }, { "epoch": 0.4111920150535875, "grad_norm": 0.23783445358276367, "learning_rate": 9.999761522246773e-06, "loss": 0.0045, "step": 25130 }, { "epoch": 0.4113556410046633, "grad_norm": 0.05245461314916611, "learning_rate": 9.999754498298064e-06, "loss": 0.0079, "step": 25140 }, { "epoch": 0.4115192669557392, "grad_norm": 0.20317251980304718, "learning_rate": 9.999747372408615e-06, "loss": 0.0041, "step": 25150 }, { "epoch": 0.41168289290681503, "grad_norm": 0.2727523446083069, "learning_rate": 9.999740144578568e-06, "loss": 0.0042, "step": 25160 }, { "epoch": 0.41184651885789086, "grad_norm": 0.2234259694814682, "learning_rate": 9.999732814808071e-06, "loss": 0.0074, "step": 25170 }, { "epoch": 0.4120101448089667, "grad_norm": 0.1002168208360672, "learning_rate": 9.999725383097275e-06, "loss": 0.0053, "step": 25180 }, { "epoch": 0.4121737707600425, "grad_norm": 0.1090393140912056, "learning_rate": 9.99971784944633e-06, "loss": 0.0042, "step": 25190 }, { "epoch": 0.4123373967111184, "grad_norm": 0.059359900653362274, "learning_rate": 9.99971021385539e-06, "loss": 0.007, "step": 25200 }, { "epoch": 0.41250102266219424, "grad_norm": 0.07507307827472687, "learning_rate": 9.999702476324611e-06, "loss": 0.0067, "step": 25210 }, { "epoch": 0.41266464861327007, "grad_norm": 0.2514195442199707, "learning_rate": 9.999694636854151e-06, "loss": 0.0038, "step": 25220 }, { "epoch": 0.4128282745643459, "grad_norm": 0.2539943754673004, "learning_rate": 9.99968669544417e-06, "loss": 0.0068, "step": 25230 }, { "epoch": 0.41299190051542173, "grad_norm": 0.0930694043636322, "learning_rate": 9.999678652094828e-06, "loss": 0.0028, "step": 25240 }, { "epoch": 0.4131555264664976, "grad_norm": 0.10800737887620926, "learning_rate": 9.999670506806292e-06, "loss": 0.0042, "step": 25250 }, { "epoch": 0.41331915241757344, "grad_norm": 0.11834069341421127, "learning_rate": 9.999662259578725e-06, "loss": 0.0047, "step": 25260 }, { "epoch": 0.4134827783686493, "grad_norm": 0.23048332333564758, "learning_rate": 9.999653910412298e-06, "loss": 0.0028, "step": 25270 }, { "epoch": 0.4136464043197251, "grad_norm": 0.11871960014104843, "learning_rate": 9.999645459307176e-06, "loss": 0.0042, "step": 25280 }, { "epoch": 0.41381003027080093, "grad_norm": 0.11515631526708603, "learning_rate": 9.999636906263539e-06, "loss": 0.004, "step": 25290 }, { "epoch": 0.41397365622187676, "grad_norm": 0.16639158129692078, "learning_rate": 9.999628251281556e-06, "loss": 0.0039, "step": 25300 }, { "epoch": 0.41413728217295265, "grad_norm": 0.09989313781261444, "learning_rate": 9.999619494361405e-06, "loss": 0.0037, "step": 25310 }, { "epoch": 0.4143009081240285, "grad_norm": 0.19848747551441193, "learning_rate": 9.999610635503266e-06, "loss": 0.0058, "step": 25320 }, { "epoch": 0.4144645340751043, "grad_norm": 0.059781014919281006, "learning_rate": 9.999601674707317e-06, "loss": 0.0061, "step": 25330 }, { "epoch": 0.41462816002618014, "grad_norm": 0.6605257391929626, "learning_rate": 9.999592611973743e-06, "loss": 0.0059, "step": 25340 }, { "epoch": 0.41479178597725597, "grad_norm": 0.04650753736495972, "learning_rate": 9.999583447302726e-06, "loss": 0.0035, "step": 25350 }, { "epoch": 0.41495541192833185, "grad_norm": 0.041029080748558044, "learning_rate": 9.999574180694456e-06, "loss": 0.0039, "step": 25360 }, { "epoch": 0.4151190378794077, "grad_norm": 0.10975819826126099, "learning_rate": 9.999564812149118e-06, "loss": 0.0051, "step": 25370 }, { "epoch": 0.4152826638304835, "grad_norm": 0.07286175340414047, "learning_rate": 9.999555341666908e-06, "loss": 0.0046, "step": 25380 }, { "epoch": 0.41544628978155934, "grad_norm": 0.11840367317199707, "learning_rate": 9.999545769248016e-06, "loss": 0.0037, "step": 25390 }, { "epoch": 0.41560991573263517, "grad_norm": 0.05487668514251709, "learning_rate": 9.999536094892637e-06, "loss": 0.0033, "step": 25400 }, { "epoch": 0.41577354168371106, "grad_norm": 0.07930178940296173, "learning_rate": 9.99952631860097e-06, "loss": 0.0046, "step": 25410 }, { "epoch": 0.4159371676347869, "grad_norm": 0.24946336448192596, "learning_rate": 9.999516440373212e-06, "loss": 0.0043, "step": 25420 }, { "epoch": 0.4161007935858627, "grad_norm": 0.12467186152935028, "learning_rate": 9.999506460209566e-06, "loss": 0.0028, "step": 25430 }, { "epoch": 0.41626441953693855, "grad_norm": 0.25923803448677063, "learning_rate": 9.999496378110236e-06, "loss": 0.0052, "step": 25440 }, { "epoch": 0.4164280454880144, "grad_norm": 0.28901058435440063, "learning_rate": 9.999486194075424e-06, "loss": 0.0055, "step": 25450 }, { "epoch": 0.41659167143909026, "grad_norm": 0.036517467349767685, "learning_rate": 9.999475908105345e-06, "loss": 0.0046, "step": 25460 }, { "epoch": 0.4167552973901661, "grad_norm": 0.1596677303314209, "learning_rate": 9.999465520200201e-06, "loss": 0.0047, "step": 25470 }, { "epoch": 0.4169189233412419, "grad_norm": 0.12417147308588028, "learning_rate": 9.999455030360207e-06, "loss": 0.0031, "step": 25480 }, { "epoch": 0.41708254929231775, "grad_norm": 0.3233848810195923, "learning_rate": 9.99944443858558e-06, "loss": 0.0049, "step": 25490 }, { "epoch": 0.4172461752433936, "grad_norm": 0.12631428241729736, "learning_rate": 9.999433744876528e-06, "loss": 0.0052, "step": 25500 }, { "epoch": 0.41740980119446947, "grad_norm": 0.18793383240699768, "learning_rate": 9.999422949233278e-06, "loss": 0.0041, "step": 25510 }, { "epoch": 0.4175734271455453, "grad_norm": 0.09556079655885696, "learning_rate": 9.999412051656044e-06, "loss": 0.004, "step": 25520 }, { "epoch": 0.4177370530966211, "grad_norm": 0.15479198098182678, "learning_rate": 9.99940105214505e-06, "loss": 0.0043, "step": 25530 }, { "epoch": 0.41790067904769695, "grad_norm": 0.06880859285593033, "learning_rate": 9.99938995070052e-06, "loss": 0.0046, "step": 25540 }, { "epoch": 0.4180643049987728, "grad_norm": 0.10976146161556244, "learning_rate": 9.999378747322681e-06, "loss": 0.0038, "step": 25550 }, { "epoch": 0.41822793094984867, "grad_norm": 0.2192991077899933, "learning_rate": 9.999367442011763e-06, "loss": 0.0052, "step": 25560 }, { "epoch": 0.4183915569009245, "grad_norm": 0.20336294174194336, "learning_rate": 9.999356034767991e-06, "loss": 0.0031, "step": 25570 }, { "epoch": 0.41855518285200033, "grad_norm": 0.04404453560709953, "learning_rate": 9.999344525591604e-06, "loss": 0.004, "step": 25580 }, { "epoch": 0.41871880880307616, "grad_norm": 0.03359917551279068, "learning_rate": 9.999332914482833e-06, "loss": 0.0054, "step": 25590 }, { "epoch": 0.418882434754152, "grad_norm": 0.22117958962917328, "learning_rate": 9.999321201441916e-06, "loss": 0.0084, "step": 25600 }, { "epoch": 0.4190460607052279, "grad_norm": 0.36323994398117065, "learning_rate": 9.999309386469091e-06, "loss": 0.0037, "step": 25610 }, { "epoch": 0.4192096866563037, "grad_norm": 0.35884690284729004, "learning_rate": 9.999297469564601e-06, "loss": 0.0069, "step": 25620 }, { "epoch": 0.41937331260737953, "grad_norm": 0.40806958079338074, "learning_rate": 9.999285450728686e-06, "loss": 0.0046, "step": 25630 }, { "epoch": 0.41953693855845536, "grad_norm": 0.06866832822561264, "learning_rate": 9.999273329961594e-06, "loss": 0.0042, "step": 25640 }, { "epoch": 0.4197005645095312, "grad_norm": 0.16907472908496857, "learning_rate": 9.999261107263568e-06, "loss": 0.003, "step": 25650 }, { "epoch": 0.4198641904606071, "grad_norm": 0.14983923733234406, "learning_rate": 9.99924878263486e-06, "loss": 0.0041, "step": 25660 }, { "epoch": 0.4200278164116829, "grad_norm": 0.10785099864006042, "learning_rate": 9.999236356075723e-06, "loss": 0.0056, "step": 25670 }, { "epoch": 0.42019144236275874, "grad_norm": 0.07436855882406235, "learning_rate": 9.999223827586406e-06, "loss": 0.0036, "step": 25680 }, { "epoch": 0.42035506831383457, "grad_norm": 0.18278750777244568, "learning_rate": 9.999211197167168e-06, "loss": 0.0035, "step": 25690 }, { "epoch": 0.4205186942649104, "grad_norm": 0.03764623403549194, "learning_rate": 9.999198464818268e-06, "loss": 0.0025, "step": 25700 }, { "epoch": 0.4206823202159863, "grad_norm": 0.10185063630342484, "learning_rate": 9.999185630539959e-06, "loss": 0.0044, "step": 25710 }, { "epoch": 0.4208459461670621, "grad_norm": 0.09264640510082245, "learning_rate": 9.999172694332508e-06, "loss": 0.0031, "step": 25720 }, { "epoch": 0.42100957211813794, "grad_norm": 0.260682076215744, "learning_rate": 9.999159656196177e-06, "loss": 0.0047, "step": 25730 }, { "epoch": 0.42117319806921377, "grad_norm": 0.37670665979385376, "learning_rate": 9.999146516131234e-06, "loss": 0.0028, "step": 25740 }, { "epoch": 0.4213368240202896, "grad_norm": 0.10291631519794464, "learning_rate": 9.999133274137943e-06, "loss": 0.0034, "step": 25750 }, { "epoch": 0.42150044997136543, "grad_norm": 0.3429001569747925, "learning_rate": 9.999119930216576e-06, "loss": 0.0051, "step": 25760 }, { "epoch": 0.4216640759224413, "grad_norm": 0.3090563416481018, "learning_rate": 9.999106484367407e-06, "loss": 0.0045, "step": 25770 }, { "epoch": 0.42182770187351715, "grad_norm": 0.09472007304430008, "learning_rate": 9.999092936590708e-06, "loss": 0.0045, "step": 25780 }, { "epoch": 0.421991327824593, "grad_norm": 0.1163785457611084, "learning_rate": 9.999079286886756e-06, "loss": 0.0046, "step": 25790 }, { "epoch": 0.4221549537756688, "grad_norm": 0.06592093408107758, "learning_rate": 9.999065535255828e-06, "loss": 0.0056, "step": 25800 }, { "epoch": 0.42231857972674464, "grad_norm": 0.2928982079029083, "learning_rate": 9.999051681698204e-06, "loss": 0.0037, "step": 25810 }, { "epoch": 0.4224822056778205, "grad_norm": 0.10722280293703079, "learning_rate": 9.99903772621417e-06, "loss": 0.0034, "step": 25820 }, { "epoch": 0.42264583162889635, "grad_norm": 0.10507380962371826, "learning_rate": 9.999023668804008e-06, "loss": 0.0036, "step": 25830 }, { "epoch": 0.4228094575799722, "grad_norm": 0.05868520960211754, "learning_rate": 9.999009509468003e-06, "loss": 0.004, "step": 25840 }, { "epoch": 0.422973083531048, "grad_norm": 0.1118895635008812, "learning_rate": 9.998995248206447e-06, "loss": 0.0054, "step": 25850 }, { "epoch": 0.42313670948212384, "grad_norm": 0.07243648171424866, "learning_rate": 9.99898088501963e-06, "loss": 0.0037, "step": 25860 }, { "epoch": 0.4233003354331997, "grad_norm": 0.17189763486385345, "learning_rate": 9.998966419907845e-06, "loss": 0.0046, "step": 25870 }, { "epoch": 0.42346396138427556, "grad_norm": 0.20694151520729065, "learning_rate": 9.998951852871384e-06, "loss": 0.0036, "step": 25880 }, { "epoch": 0.4236275873353514, "grad_norm": 0.06915944069623947, "learning_rate": 9.998937183910546e-06, "loss": 0.0029, "step": 25890 }, { "epoch": 0.4237912132864272, "grad_norm": 0.1516818106174469, "learning_rate": 9.998922413025632e-06, "loss": 0.0039, "step": 25900 }, { "epoch": 0.42395483923750305, "grad_norm": 0.15778373181819916, "learning_rate": 9.99890754021694e-06, "loss": 0.0042, "step": 25910 }, { "epoch": 0.42411846518857893, "grad_norm": 0.1296018362045288, "learning_rate": 9.998892565484776e-06, "loss": 0.0042, "step": 25920 }, { "epoch": 0.42428209113965476, "grad_norm": 0.15442849695682526, "learning_rate": 9.998877488829443e-06, "loss": 0.0042, "step": 25930 }, { "epoch": 0.4244457170907306, "grad_norm": 0.025703877210617065, "learning_rate": 9.99886231025125e-06, "loss": 0.0049, "step": 25940 }, { "epoch": 0.4246093430418064, "grad_norm": 0.10746867209672928, "learning_rate": 9.998847029750504e-06, "loss": 0.002, "step": 25950 }, { "epoch": 0.42477296899288225, "grad_norm": 0.15868765115737915, "learning_rate": 9.998831647327521e-06, "loss": 0.0039, "step": 25960 }, { "epoch": 0.42493659494395813, "grad_norm": 0.18104416131973267, "learning_rate": 9.998816162982612e-06, "loss": 0.0033, "step": 25970 }, { "epoch": 0.42510022089503396, "grad_norm": 0.09707491844892502, "learning_rate": 9.998800576716092e-06, "loss": 0.0035, "step": 25980 }, { "epoch": 0.4252638468461098, "grad_norm": 0.25578773021698, "learning_rate": 9.99878488852828e-06, "loss": 0.0028, "step": 25990 }, { "epoch": 0.4254274727971856, "grad_norm": 0.2557174563407898, "learning_rate": 9.998769098419494e-06, "loss": 0.0035, "step": 26000 }, { "epoch": 0.42559109874826145, "grad_norm": 0.12440783530473709, "learning_rate": 9.998753206390058e-06, "loss": 0.0054, "step": 26010 }, { "epoch": 0.42575472469933734, "grad_norm": 0.3831712305545807, "learning_rate": 9.998737212440296e-06, "loss": 0.0048, "step": 26020 }, { "epoch": 0.42591835065041317, "grad_norm": 0.2286456674337387, "learning_rate": 9.998721116570533e-06, "loss": 0.0042, "step": 26030 }, { "epoch": 0.426081976601489, "grad_norm": 0.09556512534618378, "learning_rate": 9.998704918781097e-06, "loss": 0.0029, "step": 26040 }, { "epoch": 0.42624560255256483, "grad_norm": 0.20679213106632233, "learning_rate": 9.99868861907232e-06, "loss": 0.005, "step": 26050 }, { "epoch": 0.42640922850364066, "grad_norm": 0.04784760996699333, "learning_rate": 9.998672217444533e-06, "loss": 0.003, "step": 26060 }, { "epoch": 0.42657285445471654, "grad_norm": 0.09368787705898285, "learning_rate": 9.99865571389807e-06, "loss": 0.0034, "step": 26070 }, { "epoch": 0.4267364804057924, "grad_norm": 0.3422050178050995, "learning_rate": 9.99863910843327e-06, "loss": 0.0051, "step": 26080 }, { "epoch": 0.4269001063568682, "grad_norm": 0.17890796065330505, "learning_rate": 9.99862240105047e-06, "loss": 0.0053, "step": 26090 }, { "epoch": 0.42706373230794403, "grad_norm": 0.21800026297569275, "learning_rate": 9.998605591750009e-06, "loss": 0.0044, "step": 26100 }, { "epoch": 0.42722735825901986, "grad_norm": 0.2139187455177307, "learning_rate": 9.998588680532231e-06, "loss": 0.0035, "step": 26110 }, { "epoch": 0.42739098421009575, "grad_norm": 0.19632987678050995, "learning_rate": 9.998571667397481e-06, "loss": 0.0053, "step": 26120 }, { "epoch": 0.4275546101611716, "grad_norm": 0.2859506905078888, "learning_rate": 9.998554552346108e-06, "loss": 0.0048, "step": 26130 }, { "epoch": 0.4277182361122474, "grad_norm": 0.19365176558494568, "learning_rate": 9.998537335378456e-06, "loss": 0.0054, "step": 26140 }, { "epoch": 0.42788186206332324, "grad_norm": 0.12902939319610596, "learning_rate": 9.99852001649488e-06, "loss": 0.0023, "step": 26150 }, { "epoch": 0.42804548801439907, "grad_norm": 0.26238876581192017, "learning_rate": 9.998502595695732e-06, "loss": 0.0036, "step": 26160 }, { "epoch": 0.4282091139654749, "grad_norm": 0.3157866895198822, "learning_rate": 9.998485072981365e-06, "loss": 0.0058, "step": 26170 }, { "epoch": 0.4283727399165508, "grad_norm": 0.3600223660469055, "learning_rate": 9.998467448352141e-06, "loss": 0.0038, "step": 26180 }, { "epoch": 0.4285363658676266, "grad_norm": 0.11996397376060486, "learning_rate": 9.998449721808416e-06, "loss": 0.0039, "step": 26190 }, { "epoch": 0.42869999181870244, "grad_norm": 0.160811185836792, "learning_rate": 9.998431893350552e-06, "loss": 0.0047, "step": 26200 }, { "epoch": 0.42886361776977827, "grad_norm": 0.3165494203567505, "learning_rate": 9.998413962978913e-06, "loss": 0.0066, "step": 26210 }, { "epoch": 0.4290272437208541, "grad_norm": 0.06887270510196686, "learning_rate": 9.998395930693865e-06, "loss": 0.0074, "step": 26220 }, { "epoch": 0.42919086967193, "grad_norm": 0.08147931843996048, "learning_rate": 9.998377796495773e-06, "loss": 0.0038, "step": 26230 }, { "epoch": 0.4293544956230058, "grad_norm": 0.09643200039863586, "learning_rate": 9.998359560385011e-06, "loss": 0.0032, "step": 26240 }, { "epoch": 0.42951812157408165, "grad_norm": 0.1182512566447258, "learning_rate": 9.998341222361947e-06, "loss": 0.0046, "step": 26250 }, { "epoch": 0.4296817475251575, "grad_norm": 0.24211357533931732, "learning_rate": 9.998322782426957e-06, "loss": 0.0064, "step": 26260 }, { "epoch": 0.4298453734762333, "grad_norm": 0.14535945653915405, "learning_rate": 9.998304240580416e-06, "loss": 0.0027, "step": 26270 }, { "epoch": 0.4300089994273092, "grad_norm": 0.18549613654613495, "learning_rate": 9.998285596822704e-06, "loss": 0.0033, "step": 26280 }, { "epoch": 0.430172625378385, "grad_norm": 0.21193449199199677, "learning_rate": 9.998266851154197e-06, "loss": 0.0045, "step": 26290 }, { "epoch": 0.43033625132946085, "grad_norm": 0.0509624145925045, "learning_rate": 9.998248003575282e-06, "loss": 0.0041, "step": 26300 }, { "epoch": 0.4304998772805367, "grad_norm": 0.300417423248291, "learning_rate": 9.99822905408634e-06, "loss": 0.0056, "step": 26310 }, { "epoch": 0.4306635032316125, "grad_norm": 0.04854543134570122, "learning_rate": 9.998210002687758e-06, "loss": 0.0042, "step": 26320 }, { "epoch": 0.4308271291826884, "grad_norm": 0.21527865529060364, "learning_rate": 9.998190849379924e-06, "loss": 0.005, "step": 26330 }, { "epoch": 0.4309907551337642, "grad_norm": 0.14639009535312653, "learning_rate": 9.998171594163232e-06, "loss": 0.0072, "step": 26340 }, { "epoch": 0.43115438108484005, "grad_norm": 0.17406724393367767, "learning_rate": 9.99815223703807e-06, "loss": 0.0028, "step": 26350 }, { "epoch": 0.4313180070359159, "grad_norm": 0.5569946765899658, "learning_rate": 9.998132778004837e-06, "loss": 0.0047, "step": 26360 }, { "epoch": 0.4314816329869917, "grad_norm": 0.25458425283432007, "learning_rate": 9.998113217063925e-06, "loss": 0.0033, "step": 26370 }, { "epoch": 0.4316452589380676, "grad_norm": 0.06832278519868851, "learning_rate": 9.998093554215735e-06, "loss": 0.0047, "step": 26380 }, { "epoch": 0.43180888488914343, "grad_norm": 0.08607838302850723, "learning_rate": 9.99807378946067e-06, "loss": 0.0032, "step": 26390 }, { "epoch": 0.43197251084021926, "grad_norm": 0.19025413691997528, "learning_rate": 9.998053922799131e-06, "loss": 0.0041, "step": 26400 }, { "epoch": 0.4321361367912951, "grad_norm": 0.06855814903974533, "learning_rate": 9.998033954231523e-06, "loss": 0.0042, "step": 26410 }, { "epoch": 0.4322997627423709, "grad_norm": 0.08008967339992523, "learning_rate": 9.998013883758252e-06, "loss": 0.0072, "step": 26420 }, { "epoch": 0.4324633886934468, "grad_norm": 0.04564782604575157, "learning_rate": 9.99799371137973e-06, "loss": 0.0124, "step": 26430 }, { "epoch": 0.43262701464452263, "grad_norm": 0.10168755054473877, "learning_rate": 9.997973437096366e-06, "loss": 0.0039, "step": 26440 }, { "epoch": 0.43279064059559846, "grad_norm": 0.28457680344581604, "learning_rate": 9.997953060908574e-06, "loss": 0.0039, "step": 26450 }, { "epoch": 0.4329542665466743, "grad_norm": 0.08336926996707916, "learning_rate": 9.997932582816771e-06, "loss": 0.0036, "step": 26460 }, { "epoch": 0.4331178924977501, "grad_norm": 0.32391616702079773, "learning_rate": 9.997912002821373e-06, "loss": 0.0032, "step": 26470 }, { "epoch": 0.433281518448826, "grad_norm": 0.5596672296524048, "learning_rate": 9.9978913209228e-06, "loss": 0.0033, "step": 26480 }, { "epoch": 0.43344514439990184, "grad_norm": 0.4474188983440399, "learning_rate": 9.997870537121472e-06, "loss": 0.0047, "step": 26490 }, { "epoch": 0.43360877035097767, "grad_norm": 1.1210988759994507, "learning_rate": 9.997849651417815e-06, "loss": 0.0117, "step": 26500 }, { "epoch": 0.4337723963020535, "grad_norm": 0.07779864966869354, "learning_rate": 9.997828663812254e-06, "loss": 0.0051, "step": 26510 }, { "epoch": 0.4339360222531293, "grad_norm": 0.17434799671173096, "learning_rate": 9.997807574305218e-06, "loss": 0.0032, "step": 26520 }, { "epoch": 0.4340996482042052, "grad_norm": 0.07303110510110855, "learning_rate": 9.997786382897135e-06, "loss": 0.0035, "step": 26530 }, { "epoch": 0.43426327415528104, "grad_norm": 0.04681973159313202, "learning_rate": 9.997765089588439e-06, "loss": 0.0029, "step": 26540 }, { "epoch": 0.4344269001063569, "grad_norm": 0.11589141935110092, "learning_rate": 9.997743694379562e-06, "loss": 0.0026, "step": 26550 }, { "epoch": 0.4345905260574327, "grad_norm": 0.10688954591751099, "learning_rate": 9.997722197270942e-06, "loss": 0.003, "step": 26560 }, { "epoch": 0.43475415200850853, "grad_norm": 0.15923893451690674, "learning_rate": 9.997700598263019e-06, "loss": 0.0044, "step": 26570 }, { "epoch": 0.4349177779595844, "grad_norm": 0.1487172245979309, "learning_rate": 9.997678897356227e-06, "loss": 0.0031, "step": 26580 }, { "epoch": 0.43508140391066025, "grad_norm": 0.09937702119350433, "learning_rate": 9.997657094551017e-06, "loss": 0.0077, "step": 26590 }, { "epoch": 0.4352450298617361, "grad_norm": 0.24548287689685822, "learning_rate": 9.997635189847827e-06, "loss": 0.0044, "step": 26600 }, { "epoch": 0.4354086558128119, "grad_norm": 0.09195926785469055, "learning_rate": 9.997613183247106e-06, "loss": 0.0035, "step": 26610 }, { "epoch": 0.43557228176388774, "grad_norm": 0.10624075680971146, "learning_rate": 9.997591074749302e-06, "loss": 0.0032, "step": 26620 }, { "epoch": 0.43573590771496357, "grad_norm": 0.37142980098724365, "learning_rate": 9.997568864354866e-06, "loss": 0.0104, "step": 26630 }, { "epoch": 0.43589953366603945, "grad_norm": 0.07850992679595947, "learning_rate": 9.997546552064252e-06, "loss": 0.0048, "step": 26640 }, { "epoch": 0.4360631596171153, "grad_norm": 0.2238306850194931, "learning_rate": 9.997524137877913e-06, "loss": 0.0041, "step": 26650 }, { "epoch": 0.4362267855681911, "grad_norm": 0.06877344846725464, "learning_rate": 9.997501621796309e-06, "loss": 0.0046, "step": 26660 }, { "epoch": 0.43639041151926694, "grad_norm": 0.16973277926445007, "learning_rate": 9.997479003819896e-06, "loss": 0.0058, "step": 26670 }, { "epoch": 0.43655403747034277, "grad_norm": 0.3797495365142822, "learning_rate": 9.997456283949135e-06, "loss": 0.0039, "step": 26680 }, { "epoch": 0.43671766342141866, "grad_norm": 0.09545832127332687, "learning_rate": 9.997433462184494e-06, "loss": 0.0071, "step": 26690 }, { "epoch": 0.4368812893724945, "grad_norm": 0.13027063012123108, "learning_rate": 9.997410538526434e-06, "loss": 0.0036, "step": 26700 }, { "epoch": 0.4370449153235703, "grad_norm": 0.07889262586832047, "learning_rate": 9.997387512975421e-06, "loss": 0.0057, "step": 26710 }, { "epoch": 0.43720854127464615, "grad_norm": 0.15415535867214203, "learning_rate": 9.997364385531928e-06, "loss": 0.0031, "step": 26720 }, { "epoch": 0.437372167225722, "grad_norm": 0.4882834553718567, "learning_rate": 9.997341156196425e-06, "loss": 0.0054, "step": 26730 }, { "epoch": 0.43753579317679786, "grad_norm": 0.21822094917297363, "learning_rate": 9.997317824969385e-06, "loss": 0.0036, "step": 26740 }, { "epoch": 0.4376994191278737, "grad_norm": 0.2246321737766266, "learning_rate": 9.997294391851287e-06, "loss": 0.0048, "step": 26750 }, { "epoch": 0.4378630450789495, "grad_norm": 0.39662232995033264, "learning_rate": 9.997270856842605e-06, "loss": 0.0078, "step": 26760 }, { "epoch": 0.43802667103002535, "grad_norm": 0.18292704224586487, "learning_rate": 9.99724721994382e-06, "loss": 0.0038, "step": 26770 }, { "epoch": 0.4381902969811012, "grad_norm": 0.3387145698070526, "learning_rate": 9.997223481155417e-06, "loss": 0.0047, "step": 26780 }, { "epoch": 0.43835392293217706, "grad_norm": 0.30814608931541443, "learning_rate": 9.997199640477873e-06, "loss": 0.0061, "step": 26790 }, { "epoch": 0.4385175488832529, "grad_norm": 0.15218864381313324, "learning_rate": 9.99717569791168e-06, "loss": 0.0097, "step": 26800 }, { "epoch": 0.4386811748343287, "grad_norm": 0.13944163918495178, "learning_rate": 9.997151653457324e-06, "loss": 0.0031, "step": 26810 }, { "epoch": 0.43884480078540455, "grad_norm": 0.0670444518327713, "learning_rate": 9.997127507115297e-06, "loss": 0.0049, "step": 26820 }, { "epoch": 0.4390084267364804, "grad_norm": 0.3063841164112091, "learning_rate": 9.997103258886089e-06, "loss": 0.005, "step": 26830 }, { "epoch": 0.43917205268755627, "grad_norm": 0.16914385557174683, "learning_rate": 9.997078908770197e-06, "loss": 0.0035, "step": 26840 }, { "epoch": 0.4393356786386321, "grad_norm": 0.32465919852256775, "learning_rate": 9.997054456768114e-06, "loss": 0.0063, "step": 26850 }, { "epoch": 0.43949930458970793, "grad_norm": 0.1115717738866806, "learning_rate": 9.99702990288034e-06, "loss": 0.0039, "step": 26860 }, { "epoch": 0.43966293054078376, "grad_norm": 0.11917252093553543, "learning_rate": 9.997005247107378e-06, "loss": 0.0037, "step": 26870 }, { "epoch": 0.4398265564918596, "grad_norm": 0.22692342102527618, "learning_rate": 9.996980489449728e-06, "loss": 0.0059, "step": 26880 }, { "epoch": 0.4399901824429355, "grad_norm": 0.18923920392990112, "learning_rate": 9.996955629907896e-06, "loss": 0.0035, "step": 26890 }, { "epoch": 0.4401538083940113, "grad_norm": 0.09172652661800385, "learning_rate": 9.996930668482388e-06, "loss": 0.0034, "step": 26900 }, { "epoch": 0.44031743434508713, "grad_norm": 0.052121348679065704, "learning_rate": 9.996905605173712e-06, "loss": 0.0034, "step": 26910 }, { "epoch": 0.44048106029616296, "grad_norm": 0.1711583435535431, "learning_rate": 9.996880439982382e-06, "loss": 0.0039, "step": 26920 }, { "epoch": 0.4406446862472388, "grad_norm": 0.26988691091537476, "learning_rate": 9.99685517290891e-06, "loss": 0.0068, "step": 26930 }, { "epoch": 0.4408083121983147, "grad_norm": 0.2070457935333252, "learning_rate": 9.99682980395381e-06, "loss": 0.0039, "step": 26940 }, { "epoch": 0.4409719381493905, "grad_norm": 0.11718105524778366, "learning_rate": 9.9968043331176e-06, "loss": 0.0037, "step": 26950 }, { "epoch": 0.44113556410046634, "grad_norm": 0.053028304129838943, "learning_rate": 9.9967787604008e-06, "loss": 0.0028, "step": 26960 }, { "epoch": 0.44129919005154217, "grad_norm": 0.2117151916027069, "learning_rate": 9.99675308580393e-06, "loss": 0.0058, "step": 26970 }, { "epoch": 0.441462816002618, "grad_norm": 0.1964861899614334, "learning_rate": 9.996727309327514e-06, "loss": 0.0037, "step": 26980 }, { "epoch": 0.4416264419536939, "grad_norm": 0.16663330793380737, "learning_rate": 9.996701430972078e-06, "loss": 0.0038, "step": 26990 }, { "epoch": 0.4417900679047697, "grad_norm": 0.15710590779781342, "learning_rate": 9.99667545073815e-06, "loss": 0.003, "step": 27000 }, { "epoch": 0.44195369385584554, "grad_norm": 0.05967897176742554, "learning_rate": 9.996649368626258e-06, "loss": 0.0038, "step": 27010 }, { "epoch": 0.44211731980692137, "grad_norm": 0.16013340651988983, "learning_rate": 9.996623184636936e-06, "loss": 0.006, "step": 27020 }, { "epoch": 0.4422809457579972, "grad_norm": 0.26836228370666504, "learning_rate": 9.996596898770715e-06, "loss": 0.0038, "step": 27030 }, { "epoch": 0.4424445717090731, "grad_norm": 0.0562361441552639, "learning_rate": 9.996570511028135e-06, "loss": 0.0046, "step": 27040 }, { "epoch": 0.4426081976601489, "grad_norm": 0.0926424115896225, "learning_rate": 9.996544021409729e-06, "loss": 0.0037, "step": 27050 }, { "epoch": 0.44277182361122475, "grad_norm": 0.25835689902305603, "learning_rate": 9.996517429916041e-06, "loss": 0.0037, "step": 27060 }, { "epoch": 0.4429354495623006, "grad_norm": 0.06540341675281525, "learning_rate": 9.996490736547612e-06, "loss": 0.003, "step": 27070 }, { "epoch": 0.4430990755133764, "grad_norm": 0.11224102973937988, "learning_rate": 9.996463941304987e-06, "loss": 0.0053, "step": 27080 }, { "epoch": 0.44326270146445224, "grad_norm": 0.191273033618927, "learning_rate": 9.996437044188712e-06, "loss": 0.0033, "step": 27090 }, { "epoch": 0.4434263274155281, "grad_norm": 0.11718302965164185, "learning_rate": 9.996410045199334e-06, "loss": 0.0056, "step": 27100 }, { "epoch": 0.44358995336660395, "grad_norm": 0.04910613223910332, "learning_rate": 9.996382944337404e-06, "loss": 0.0034, "step": 27110 }, { "epoch": 0.4437535793176798, "grad_norm": 0.1824340969324112, "learning_rate": 9.996355741603475e-06, "loss": 0.0037, "step": 27120 }, { "epoch": 0.4439172052687556, "grad_norm": 0.08041592687368393, "learning_rate": 9.996328436998102e-06, "loss": 0.0045, "step": 27130 }, { "epoch": 0.44408083121983144, "grad_norm": 0.05174966901540756, "learning_rate": 9.996301030521842e-06, "loss": 0.0037, "step": 27140 }, { "epoch": 0.4442444571709073, "grad_norm": 0.14784163236618042, "learning_rate": 9.996273522175253e-06, "loss": 0.0036, "step": 27150 }, { "epoch": 0.44440808312198316, "grad_norm": 0.1644139438867569, "learning_rate": 9.996245911958896e-06, "loss": 0.0036, "step": 27160 }, { "epoch": 0.444571709073059, "grad_norm": 0.2815445065498352, "learning_rate": 9.996218199873333e-06, "loss": 0.0046, "step": 27170 }, { "epoch": 0.4447353350241348, "grad_norm": 0.11031584441661835, "learning_rate": 9.996190385919131e-06, "loss": 0.0035, "step": 27180 }, { "epoch": 0.44489896097521064, "grad_norm": 0.27610912919044495, "learning_rate": 9.996162470096857e-06, "loss": 0.0049, "step": 27190 }, { "epoch": 0.44506258692628653, "grad_norm": 0.24276669323444366, "learning_rate": 9.996134452407077e-06, "loss": 0.0064, "step": 27200 }, { "epoch": 0.44522621287736236, "grad_norm": 0.29813361167907715, "learning_rate": 9.996106332850366e-06, "loss": 0.0048, "step": 27210 }, { "epoch": 0.4453898388284382, "grad_norm": 0.18097570538520813, "learning_rate": 9.996078111427297e-06, "loss": 0.0049, "step": 27220 }, { "epoch": 0.445553464779514, "grad_norm": 0.13288721442222595, "learning_rate": 9.996049788138443e-06, "loss": 0.0029, "step": 27230 }, { "epoch": 0.44571709073058985, "grad_norm": 0.14311382174491882, "learning_rate": 9.996021362984383e-06, "loss": 0.0034, "step": 27240 }, { "epoch": 0.44588071668166573, "grad_norm": 0.11784293502569199, "learning_rate": 9.995992835965695e-06, "loss": 0.0024, "step": 27250 }, { "epoch": 0.44604434263274156, "grad_norm": 0.5263943076133728, "learning_rate": 9.995964207082964e-06, "loss": 0.0059, "step": 27260 }, { "epoch": 0.4462079685838174, "grad_norm": 0.13926151394844055, "learning_rate": 9.995935476336772e-06, "loss": 0.0048, "step": 27270 }, { "epoch": 0.4463715945348932, "grad_norm": 0.17433682084083557, "learning_rate": 9.995906643727703e-06, "loss": 0.0038, "step": 27280 }, { "epoch": 0.44653522048596905, "grad_norm": 0.11859584599733353, "learning_rate": 9.995877709256348e-06, "loss": 0.0036, "step": 27290 }, { "epoch": 0.44669884643704494, "grad_norm": 0.27433714270591736, "learning_rate": 9.995848672923293e-06, "loss": 0.0029, "step": 27300 }, { "epoch": 0.44686247238812077, "grad_norm": 0.23860201239585876, "learning_rate": 9.995819534729134e-06, "loss": 0.0029, "step": 27310 }, { "epoch": 0.4470260983391966, "grad_norm": 0.08167732506990433, "learning_rate": 9.995790294674464e-06, "loss": 0.0043, "step": 27320 }, { "epoch": 0.44718972429027243, "grad_norm": 0.12910307943820953, "learning_rate": 9.995760952759877e-06, "loss": 0.0022, "step": 27330 }, { "epoch": 0.44735335024134826, "grad_norm": 0.06776918470859528, "learning_rate": 9.995731508985974e-06, "loss": 0.0057, "step": 27340 }, { "epoch": 0.44751697619242414, "grad_norm": 0.08393682539463043, "learning_rate": 9.995701963353354e-06, "loss": 0.0039, "step": 27350 }, { "epoch": 0.4476806021435, "grad_norm": 0.18717533349990845, "learning_rate": 9.99567231586262e-06, "loss": 0.004, "step": 27360 }, { "epoch": 0.4478442280945758, "grad_norm": 0.1391838639974594, "learning_rate": 9.995642566514374e-06, "loss": 0.0039, "step": 27370 }, { "epoch": 0.44800785404565163, "grad_norm": 0.23270200192928314, "learning_rate": 9.995612715309228e-06, "loss": 0.0042, "step": 27380 }, { "epoch": 0.44817147999672746, "grad_norm": 0.08955033868551254, "learning_rate": 9.995582762247786e-06, "loss": 0.0041, "step": 27390 }, { "epoch": 0.44833510594780335, "grad_norm": 0.18732483685016632, "learning_rate": 9.99555270733066e-06, "loss": 0.0036, "step": 27400 }, { "epoch": 0.4484987318988792, "grad_norm": 0.12323611974716187, "learning_rate": 9.995522550558462e-06, "loss": 0.0031, "step": 27410 }, { "epoch": 0.448662357849955, "grad_norm": 0.12637698650360107, "learning_rate": 9.99549229193181e-06, "loss": 0.0036, "step": 27420 }, { "epoch": 0.44882598380103084, "grad_norm": 0.14145420491695404, "learning_rate": 9.995461931451316e-06, "loss": 0.0025, "step": 27430 }, { "epoch": 0.44898960975210667, "grad_norm": 0.09392669796943665, "learning_rate": 9.995431469117604e-06, "loss": 0.0025, "step": 27440 }, { "epoch": 0.44915323570318255, "grad_norm": 0.14458806812763214, "learning_rate": 9.995400904931292e-06, "loss": 0.0046, "step": 27450 }, { "epoch": 0.4493168616542584, "grad_norm": 0.1254132241010666, "learning_rate": 9.995370238893002e-06, "loss": 0.0059, "step": 27460 }, { "epoch": 0.4494804876053342, "grad_norm": 0.07990530878305435, "learning_rate": 9.995339471003364e-06, "loss": 0.0053, "step": 27470 }, { "epoch": 0.44964411355641004, "grad_norm": 0.015056951902806759, "learning_rate": 9.995308601263001e-06, "loss": 0.0045, "step": 27480 }, { "epoch": 0.44980773950748587, "grad_norm": 0.2548316419124603, "learning_rate": 9.995277629672546e-06, "loss": 0.0033, "step": 27490 }, { "epoch": 0.4499713654585617, "grad_norm": 0.20976446568965912, "learning_rate": 9.995246556232626e-06, "loss": 0.0033, "step": 27500 }, { "epoch": 0.4501349914096376, "grad_norm": 0.16374512016773224, "learning_rate": 9.995215380943878e-06, "loss": 0.0043, "step": 27510 }, { "epoch": 0.4502986173607134, "grad_norm": 0.10583820194005966, "learning_rate": 9.995184103806937e-06, "loss": 0.0038, "step": 27520 }, { "epoch": 0.45046224331178925, "grad_norm": 0.1769176721572876, "learning_rate": 9.99515272482244e-06, "loss": 0.0049, "step": 27530 }, { "epoch": 0.4506258692628651, "grad_norm": 0.1833527535200119, "learning_rate": 9.995121243991028e-06, "loss": 0.0046, "step": 27540 }, { "epoch": 0.4507894952139409, "grad_norm": 0.2012864202260971, "learning_rate": 9.99508966131334e-06, "loss": 0.0035, "step": 27550 }, { "epoch": 0.4509531211650168, "grad_norm": 0.10869038105010986, "learning_rate": 9.995057976790024e-06, "loss": 0.0037, "step": 27560 }, { "epoch": 0.4511167471160926, "grad_norm": 0.23595736920833588, "learning_rate": 9.995026190421722e-06, "loss": 0.003, "step": 27570 }, { "epoch": 0.45128037306716845, "grad_norm": 0.2734525501728058, "learning_rate": 9.994994302209087e-06, "loss": 0.0031, "step": 27580 }, { "epoch": 0.4514439990182443, "grad_norm": 0.19757907092571259, "learning_rate": 9.994962312152764e-06, "loss": 0.0107, "step": 27590 }, { "epoch": 0.4516076249693201, "grad_norm": 0.17800143361091614, "learning_rate": 9.994930220253407e-06, "loss": 0.0054, "step": 27600 }, { "epoch": 0.451771250920396, "grad_norm": 0.10832612216472626, "learning_rate": 9.994898026511673e-06, "loss": 0.0035, "step": 27610 }, { "epoch": 0.4519348768714718, "grad_norm": 0.2706403434276581, "learning_rate": 9.994865730928214e-06, "loss": 0.0052, "step": 27620 }, { "epoch": 0.45209850282254765, "grad_norm": 0.0667344480752945, "learning_rate": 9.994833333503693e-06, "loss": 0.004, "step": 27630 }, { "epoch": 0.4522621287736235, "grad_norm": 0.08147048205137253, "learning_rate": 9.994800834238767e-06, "loss": 0.0043, "step": 27640 }, { "epoch": 0.4524257547246993, "grad_norm": 0.19588790833950043, "learning_rate": 9.9947682331341e-06, "loss": 0.0039, "step": 27650 }, { "epoch": 0.4525893806757752, "grad_norm": 0.1040431335568428, "learning_rate": 9.994735530190356e-06, "loss": 0.0045, "step": 27660 }, { "epoch": 0.45275300662685103, "grad_norm": 0.2052241414785385, "learning_rate": 9.994702725408204e-06, "loss": 0.0047, "step": 27670 }, { "epoch": 0.45291663257792686, "grad_norm": 0.08924125134944916, "learning_rate": 9.994669818788311e-06, "loss": 0.0048, "step": 27680 }, { "epoch": 0.4530802585290027, "grad_norm": 0.1974812000989914, "learning_rate": 9.994636810331348e-06, "loss": 0.0035, "step": 27690 }, { "epoch": 0.4532438844800785, "grad_norm": 0.08930552750825882, "learning_rate": 9.994603700037988e-06, "loss": 0.0045, "step": 27700 }, { "epoch": 0.4534075104311544, "grad_norm": 0.1495504230260849, "learning_rate": 9.994570487908906e-06, "loss": 0.0034, "step": 27710 }, { "epoch": 0.45357113638223023, "grad_norm": 0.15513291954994202, "learning_rate": 9.994537173944779e-06, "loss": 0.0026, "step": 27720 }, { "epoch": 0.45373476233330606, "grad_norm": 0.11846410483121872, "learning_rate": 9.994503758146288e-06, "loss": 0.0032, "step": 27730 }, { "epoch": 0.4538983882843819, "grad_norm": 0.09204913675785065, "learning_rate": 9.994470240514111e-06, "loss": 0.0049, "step": 27740 }, { "epoch": 0.4540620142354577, "grad_norm": 0.1841985136270523, "learning_rate": 9.994436621048935e-06, "loss": 0.0037, "step": 27750 }, { "epoch": 0.4542256401865336, "grad_norm": 0.12644842267036438, "learning_rate": 9.994402899751445e-06, "loss": 0.0029, "step": 27760 }, { "epoch": 0.45438926613760944, "grad_norm": 0.16739171743392944, "learning_rate": 9.994369076622325e-06, "loss": 0.0055, "step": 27770 }, { "epoch": 0.45455289208868527, "grad_norm": 0.25355854630470276, "learning_rate": 9.994335151662268e-06, "loss": 0.0036, "step": 27780 }, { "epoch": 0.4547165180397611, "grad_norm": 0.04607543721795082, "learning_rate": 9.994301124871965e-06, "loss": 0.0034, "step": 27790 }, { "epoch": 0.4548801439908369, "grad_norm": 0.3245411515235901, "learning_rate": 9.994266996252111e-06, "loss": 0.0028, "step": 27800 }, { "epoch": 0.4550437699419128, "grad_norm": 0.15082235634326935, "learning_rate": 9.9942327658034e-06, "loss": 0.0047, "step": 27810 }, { "epoch": 0.45520739589298864, "grad_norm": 0.24894031882286072, "learning_rate": 9.994198433526529e-06, "loss": 0.0047, "step": 27820 }, { "epoch": 0.45537102184406447, "grad_norm": 0.023628007620573044, "learning_rate": 9.9941639994222e-06, "loss": 0.0027, "step": 27830 }, { "epoch": 0.4555346477951403, "grad_norm": 0.1640615463256836, "learning_rate": 9.994129463491114e-06, "loss": 0.0035, "step": 27840 }, { "epoch": 0.45569827374621613, "grad_norm": 0.15352194011211395, "learning_rate": 9.994094825733975e-06, "loss": 0.005, "step": 27850 }, { "epoch": 0.455861899697292, "grad_norm": 0.07925118505954742, "learning_rate": 9.994060086151491e-06, "loss": 0.0042, "step": 27860 }, { "epoch": 0.45602552564836785, "grad_norm": 0.11744213849306107, "learning_rate": 9.99402524474437e-06, "loss": 0.0035, "step": 27870 }, { "epoch": 0.4561891515994437, "grad_norm": 0.20660056173801422, "learning_rate": 9.99399030151332e-06, "loss": 0.004, "step": 27880 }, { "epoch": 0.4563527775505195, "grad_norm": 0.10965823382139206, "learning_rate": 9.993955256459055e-06, "loss": 0.0033, "step": 27890 }, { "epoch": 0.45651640350159534, "grad_norm": 0.22864864766597748, "learning_rate": 9.99392010958229e-06, "loss": 0.0033, "step": 27900 }, { "epoch": 0.4566800294526712, "grad_norm": 0.05846943333745003, "learning_rate": 9.99388486088374e-06, "loss": 0.0042, "step": 27910 }, { "epoch": 0.45684365540374705, "grad_norm": 0.045472562313079834, "learning_rate": 9.993849510364127e-06, "loss": 0.0052, "step": 27920 }, { "epoch": 0.4570072813548229, "grad_norm": 0.14438344538211823, "learning_rate": 9.993814058024167e-06, "loss": 0.0043, "step": 27930 }, { "epoch": 0.4571709073058987, "grad_norm": 0.4087490737438202, "learning_rate": 9.993778503864588e-06, "loss": 0.0065, "step": 27940 }, { "epoch": 0.45733453325697454, "grad_norm": 0.08334951102733612, "learning_rate": 9.993742847886111e-06, "loss": 0.0039, "step": 27950 }, { "epoch": 0.45749815920805037, "grad_norm": 0.1459195911884308, "learning_rate": 9.993707090089463e-06, "loss": 0.0027, "step": 27960 }, { "epoch": 0.45766178515912626, "grad_norm": 0.14969177544116974, "learning_rate": 9.993671230475378e-06, "loss": 0.0023, "step": 27970 }, { "epoch": 0.4578254111102021, "grad_norm": 0.29071202874183655, "learning_rate": 9.993635269044582e-06, "loss": 0.0056, "step": 27980 }, { "epoch": 0.4579890370612779, "grad_norm": 0.09931430965662003, "learning_rate": 9.993599205797808e-06, "loss": 0.0025, "step": 27990 }, { "epoch": 0.45815266301235374, "grad_norm": 0.11988617479801178, "learning_rate": 9.993563040735796e-06, "loss": 0.003, "step": 28000 }, { "epoch": 0.4583162889634296, "grad_norm": 0.15163767337799072, "learning_rate": 9.993526773859278e-06, "loss": 0.0033, "step": 28010 }, { "epoch": 0.45847991491450546, "grad_norm": 0.17001581192016602, "learning_rate": 9.993490405168997e-06, "loss": 0.0059, "step": 28020 }, { "epoch": 0.4586435408655813, "grad_norm": 0.011412421241402626, "learning_rate": 9.993453934665693e-06, "loss": 0.0038, "step": 28030 }, { "epoch": 0.4588071668166571, "grad_norm": 0.12588201463222504, "learning_rate": 9.99341736235011e-06, "loss": 0.0054, "step": 28040 }, { "epoch": 0.45897079276773295, "grad_norm": 0.16009269654750824, "learning_rate": 9.993380688222996e-06, "loss": 0.0062, "step": 28050 }, { "epoch": 0.4591344187188088, "grad_norm": 0.05625995993614197, "learning_rate": 9.993343912285093e-06, "loss": 0.0034, "step": 28060 }, { "epoch": 0.45929804466988466, "grad_norm": 0.09073406457901001, "learning_rate": 9.993307034537156e-06, "loss": 0.0047, "step": 28070 }, { "epoch": 0.4594616706209605, "grad_norm": 0.06390167772769928, "learning_rate": 9.993270054979935e-06, "loss": 0.0036, "step": 28080 }, { "epoch": 0.4596252965720363, "grad_norm": 0.27866801619529724, "learning_rate": 9.993232973614185e-06, "loss": 0.0053, "step": 28090 }, { "epoch": 0.45978892252311215, "grad_norm": 0.1632014364004135, "learning_rate": 9.993195790440661e-06, "loss": 0.0041, "step": 28100 }, { "epoch": 0.459952548474188, "grad_norm": 0.16526079177856445, "learning_rate": 9.993158505460121e-06, "loss": 0.004, "step": 28110 }, { "epoch": 0.46011617442526387, "grad_norm": 0.23027385771274567, "learning_rate": 9.993121118673326e-06, "loss": 0.0044, "step": 28120 }, { "epoch": 0.4602798003763397, "grad_norm": 0.2643786072731018, "learning_rate": 9.993083630081038e-06, "loss": 0.0029, "step": 28130 }, { "epoch": 0.46044342632741553, "grad_norm": 0.09273754060268402, "learning_rate": 9.99304603968402e-06, "loss": 0.0033, "step": 28140 }, { "epoch": 0.46060705227849136, "grad_norm": 0.19377370178699493, "learning_rate": 9.99300834748304e-06, "loss": 0.0033, "step": 28150 }, { "epoch": 0.4607706782295672, "grad_norm": 0.18764221668243408, "learning_rate": 9.992970553478867e-06, "loss": 0.0048, "step": 28160 }, { "epoch": 0.4609343041806431, "grad_norm": 0.18193836510181427, "learning_rate": 9.99293265767227e-06, "loss": 0.0064, "step": 28170 }, { "epoch": 0.4610979301317189, "grad_norm": 0.11493314057588577, "learning_rate": 9.992894660064023e-06, "loss": 0.0046, "step": 28180 }, { "epoch": 0.46126155608279473, "grad_norm": 0.07859452068805695, "learning_rate": 9.9928565606549e-06, "loss": 0.0047, "step": 28190 }, { "epoch": 0.46142518203387056, "grad_norm": 0.06853849440813065, "learning_rate": 9.992818359445678e-06, "loss": 0.0029, "step": 28200 }, { "epoch": 0.4615888079849464, "grad_norm": 0.05688709765672684, "learning_rate": 9.992780056437135e-06, "loss": 0.0041, "step": 28210 }, { "epoch": 0.4617524339360223, "grad_norm": 0.0614645816385746, "learning_rate": 9.992741651630055e-06, "loss": 0.0024, "step": 28220 }, { "epoch": 0.4619160598870981, "grad_norm": 0.45396459102630615, "learning_rate": 9.992703145025217e-06, "loss": 0.0049, "step": 28230 }, { "epoch": 0.46207968583817394, "grad_norm": 0.11350274831056595, "learning_rate": 9.992664536623409e-06, "loss": 0.0063, "step": 28240 }, { "epoch": 0.46224331178924977, "grad_norm": 0.3227901756763458, "learning_rate": 9.992625826425417e-06, "loss": 0.0031, "step": 28250 }, { "epoch": 0.4624069377403256, "grad_norm": 0.13260935246944427, "learning_rate": 9.99258701443203e-06, "loss": 0.0034, "step": 28260 }, { "epoch": 0.4625705636914015, "grad_norm": 0.09039606153964996, "learning_rate": 9.99254810064404e-06, "loss": 0.0027, "step": 28270 }, { "epoch": 0.4627341896424773, "grad_norm": 0.09105551242828369, "learning_rate": 9.992509085062241e-06, "loss": 0.0054, "step": 28280 }, { "epoch": 0.46289781559355314, "grad_norm": 0.16750673949718475, "learning_rate": 9.992469967687427e-06, "loss": 0.0032, "step": 28290 }, { "epoch": 0.46306144154462897, "grad_norm": 0.143649160861969, "learning_rate": 9.992430748520396e-06, "loss": 0.0042, "step": 28300 }, { "epoch": 0.4632250674957048, "grad_norm": 0.4388536214828491, "learning_rate": 9.992391427561949e-06, "loss": 0.0085, "step": 28310 }, { "epoch": 0.4633886934467807, "grad_norm": 0.3366541266441345, "learning_rate": 9.992352004812887e-06, "loss": 0.0037, "step": 28320 }, { "epoch": 0.4635523193978565, "grad_norm": 0.16876201331615448, "learning_rate": 9.99231248027401e-06, "loss": 0.0054, "step": 28330 }, { "epoch": 0.46371594534893235, "grad_norm": 0.116632841527462, "learning_rate": 9.992272853946133e-06, "loss": 0.0047, "step": 28340 }, { "epoch": 0.4638795713000082, "grad_norm": 0.11052442342042923, "learning_rate": 9.992233125830054e-06, "loss": 0.003, "step": 28350 }, { "epoch": 0.464043197251084, "grad_norm": 0.2388480007648468, "learning_rate": 9.99219329592659e-06, "loss": 0.0051, "step": 28360 }, { "epoch": 0.46420682320215984, "grad_norm": 0.01891777478158474, "learning_rate": 9.992153364236548e-06, "loss": 0.0036, "step": 28370 }, { "epoch": 0.4643704491532357, "grad_norm": 0.027410203590989113, "learning_rate": 9.992113330760744e-06, "loss": 0.0037, "step": 28380 }, { "epoch": 0.46453407510431155, "grad_norm": 0.13110503554344177, "learning_rate": 9.992073195499997e-06, "loss": 0.003, "step": 28390 }, { "epoch": 0.4646977010553874, "grad_norm": 0.318270742893219, "learning_rate": 9.992032958455122e-06, "loss": 0.0052, "step": 28400 }, { "epoch": 0.4648613270064632, "grad_norm": 0.23975637555122375, "learning_rate": 9.99199261962694e-06, "loss": 0.0049, "step": 28410 }, { "epoch": 0.46502495295753904, "grad_norm": 0.0998329445719719, "learning_rate": 9.991952179016277e-06, "loss": 0.0034, "step": 28420 }, { "epoch": 0.4651885789086149, "grad_norm": 0.10514581203460693, "learning_rate": 9.991911636623953e-06, "loss": 0.0031, "step": 28430 }, { "epoch": 0.46535220485969075, "grad_norm": 0.14132778346538544, "learning_rate": 9.991870992450794e-06, "loss": 0.0033, "step": 28440 }, { "epoch": 0.4655158308107666, "grad_norm": 0.12868468463420868, "learning_rate": 9.991830246497633e-06, "loss": 0.004, "step": 28450 }, { "epoch": 0.4656794567618424, "grad_norm": 0.07125670462846756, "learning_rate": 9.9917893987653e-06, "loss": 0.0029, "step": 28460 }, { "epoch": 0.46584308271291824, "grad_norm": 0.05923925340175629, "learning_rate": 9.991748449254624e-06, "loss": 0.0051, "step": 28470 }, { "epoch": 0.46600670866399413, "grad_norm": 0.06435070931911469, "learning_rate": 9.991707397966443e-06, "loss": 0.0031, "step": 28480 }, { "epoch": 0.46617033461506996, "grad_norm": 0.09230821579694748, "learning_rate": 9.991666244901594e-06, "loss": 0.0033, "step": 28490 }, { "epoch": 0.4663339605661458, "grad_norm": 0.1277211755514145, "learning_rate": 9.991624990060915e-06, "loss": 0.0028, "step": 28500 }, { "epoch": 0.4664975865172216, "grad_norm": 0.13298146426677704, "learning_rate": 9.991583633445248e-06, "loss": 0.0054, "step": 28510 }, { "epoch": 0.46666121246829745, "grad_norm": 0.08152301609516144, "learning_rate": 9.991542175055436e-06, "loss": 0.0041, "step": 28520 }, { "epoch": 0.46682483841937333, "grad_norm": 0.08023188263177872, "learning_rate": 9.991500614892324e-06, "loss": 0.0039, "step": 28530 }, { "epoch": 0.46698846437044916, "grad_norm": 0.12477965652942657, "learning_rate": 9.99145895295676e-06, "loss": 0.0026, "step": 28540 }, { "epoch": 0.467152090321525, "grad_norm": 0.21476998925209045, "learning_rate": 9.991417189249592e-06, "loss": 0.004, "step": 28550 }, { "epoch": 0.4673157162726008, "grad_norm": 0.13494491577148438, "learning_rate": 9.991375323771673e-06, "loss": 0.0047, "step": 28560 }, { "epoch": 0.46747934222367665, "grad_norm": 0.06163588538765907, "learning_rate": 9.991333356523858e-06, "loss": 0.0048, "step": 28570 }, { "epoch": 0.46764296817475254, "grad_norm": 0.23640471696853638, "learning_rate": 9.991291287506998e-06, "loss": 0.0072, "step": 28580 }, { "epoch": 0.46780659412582837, "grad_norm": 0.12952841818332672, "learning_rate": 9.991249116721954e-06, "loss": 0.0031, "step": 28590 }, { "epoch": 0.4679702200769042, "grad_norm": 0.22316324710845947, "learning_rate": 9.991206844169588e-06, "loss": 0.0026, "step": 28600 }, { "epoch": 0.46813384602798, "grad_norm": 0.05955805629491806, "learning_rate": 9.991164469850756e-06, "loss": 0.0032, "step": 28610 }, { "epoch": 0.46829747197905586, "grad_norm": 0.029345855116844177, "learning_rate": 9.991121993766329e-06, "loss": 0.0043, "step": 28620 }, { "epoch": 0.46846109793013174, "grad_norm": 0.054199740290641785, "learning_rate": 9.991079415917166e-06, "loss": 0.003, "step": 28630 }, { "epoch": 0.46862472388120757, "grad_norm": 0.03526965528726578, "learning_rate": 9.99103673630414e-06, "loss": 0.0076, "step": 28640 }, { "epoch": 0.4687883498322834, "grad_norm": 0.200664684176445, "learning_rate": 9.990993954928118e-06, "loss": 0.0049, "step": 28650 }, { "epoch": 0.46895197578335923, "grad_norm": 0.17916527390480042, "learning_rate": 9.990951071789977e-06, "loss": 0.0054, "step": 28660 }, { "epoch": 0.46911560173443506, "grad_norm": 0.07058257609605789, "learning_rate": 9.990908086890586e-06, "loss": 0.0052, "step": 28670 }, { "epoch": 0.46927922768551095, "grad_norm": 0.12913282215595245, "learning_rate": 9.990865000230825e-06, "loss": 0.0027, "step": 28680 }, { "epoch": 0.4694428536365868, "grad_norm": 0.15732742846012115, "learning_rate": 9.99082181181157e-06, "loss": 0.004, "step": 28690 }, { "epoch": 0.4696064795876626, "grad_norm": 0.20564952492713928, "learning_rate": 9.990778521633703e-06, "loss": 0.0078, "step": 28700 }, { "epoch": 0.46977010553873844, "grad_norm": 0.10583022981882095, "learning_rate": 9.990735129698108e-06, "loss": 0.0042, "step": 28710 }, { "epoch": 0.46993373148981427, "grad_norm": 0.11958803236484528, "learning_rate": 9.990691636005667e-06, "loss": 0.0046, "step": 28720 }, { "epoch": 0.47009735744089015, "grad_norm": 0.10276207327842712, "learning_rate": 9.990648040557268e-06, "loss": 0.0043, "step": 28730 }, { "epoch": 0.470260983391966, "grad_norm": 0.1346631646156311, "learning_rate": 9.990604343353799e-06, "loss": 0.0021, "step": 28740 }, { "epoch": 0.4704246093430418, "grad_norm": 0.20833595097064972, "learning_rate": 9.990560544396154e-06, "loss": 0.0059, "step": 28750 }, { "epoch": 0.47058823529411764, "grad_norm": 0.15876638889312744, "learning_rate": 9.990516643685222e-06, "loss": 0.0033, "step": 28760 }, { "epoch": 0.47075186124519347, "grad_norm": 0.03844869136810303, "learning_rate": 9.9904726412219e-06, "loss": 0.0024, "step": 28770 }, { "epoch": 0.47091548719626936, "grad_norm": 0.5361451506614685, "learning_rate": 9.990428537007084e-06, "loss": 0.004, "step": 28780 }, { "epoch": 0.4710791131473452, "grad_norm": 0.2800006866455078, "learning_rate": 9.990384331041677e-06, "loss": 0.0058, "step": 28790 }, { "epoch": 0.471242739098421, "grad_norm": 0.4387984275817871, "learning_rate": 9.990340023326577e-06, "loss": 0.0053, "step": 28800 }, { "epoch": 0.47140636504949684, "grad_norm": 0.2280067652463913, "learning_rate": 9.990295613862686e-06, "loss": 0.0029, "step": 28810 }, { "epoch": 0.4715699910005727, "grad_norm": 0.37110304832458496, "learning_rate": 9.990251102650914e-06, "loss": 0.0042, "step": 28820 }, { "epoch": 0.4717336169516485, "grad_norm": 0.18031258881092072, "learning_rate": 9.990206489692165e-06, "loss": 0.0041, "step": 28830 }, { "epoch": 0.4718972429027244, "grad_norm": 0.0776405930519104, "learning_rate": 9.99016177498735e-06, "loss": 0.0133, "step": 28840 }, { "epoch": 0.4720608688538002, "grad_norm": 0.1698601096868515, "learning_rate": 9.990116958537379e-06, "loss": 0.003, "step": 28850 }, { "epoch": 0.47222449480487605, "grad_norm": 0.6324454545974731, "learning_rate": 9.990072040343169e-06, "loss": 0.004, "step": 28860 }, { "epoch": 0.4723881207559519, "grad_norm": 0.38532182574272156, "learning_rate": 9.990027020405633e-06, "loss": 0.0039, "step": 28870 }, { "epoch": 0.4725517467070277, "grad_norm": 0.023404866456985474, "learning_rate": 9.989981898725692e-06, "loss": 0.0061, "step": 28880 }, { "epoch": 0.4727153726581036, "grad_norm": 0.11049871146678925, "learning_rate": 9.989936675304262e-06, "loss": 0.0034, "step": 28890 }, { "epoch": 0.4728789986091794, "grad_norm": 0.07357291877269745, "learning_rate": 9.989891350142266e-06, "loss": 0.0034, "step": 28900 }, { "epoch": 0.47304262456025525, "grad_norm": 0.26542168855667114, "learning_rate": 9.98984592324063e-06, "loss": 0.0033, "step": 28910 }, { "epoch": 0.4732062505113311, "grad_norm": 0.2597133219242096, "learning_rate": 9.98980039460028e-06, "loss": 0.0084, "step": 28920 }, { "epoch": 0.4733698764624069, "grad_norm": 0.15031291544437408, "learning_rate": 9.989754764222144e-06, "loss": 0.0034, "step": 28930 }, { "epoch": 0.4735335024134828, "grad_norm": 0.0715213343501091, "learning_rate": 9.989709032107151e-06, "loss": 0.0062, "step": 28940 }, { "epoch": 0.47369712836455863, "grad_norm": 0.17136460542678833, "learning_rate": 9.989663198256234e-06, "loss": 0.0067, "step": 28950 }, { "epoch": 0.47386075431563446, "grad_norm": 0.08632353693246841, "learning_rate": 9.989617262670328e-06, "loss": 0.0042, "step": 28960 }, { "epoch": 0.4740243802667103, "grad_norm": 0.0633331909775734, "learning_rate": 9.98957122535037e-06, "loss": 0.0028, "step": 28970 }, { "epoch": 0.4741880062177861, "grad_norm": 0.3133474886417389, "learning_rate": 9.989525086297299e-06, "loss": 0.0035, "step": 28980 }, { "epoch": 0.474351632168862, "grad_norm": 0.1960539072751999, "learning_rate": 9.989478845512053e-06, "loss": 0.0057, "step": 28990 }, { "epoch": 0.47451525811993783, "grad_norm": 0.22735944390296936, "learning_rate": 9.98943250299558e-06, "loss": 0.0049, "step": 29000 }, { "epoch": 0.47467888407101366, "grad_norm": 0.0779023990035057, "learning_rate": 9.989386058748819e-06, "loss": 0.0043, "step": 29010 }, { "epoch": 0.4748425100220895, "grad_norm": 0.18344761431217194, "learning_rate": 9.98933951277272e-06, "loss": 0.0037, "step": 29020 }, { "epoch": 0.4750061359731653, "grad_norm": 0.09890209138393402, "learning_rate": 9.989292865068231e-06, "loss": 0.0042, "step": 29030 }, { "epoch": 0.4751697619242412, "grad_norm": 0.030051127076148987, "learning_rate": 9.989246115636304e-06, "loss": 0.003, "step": 29040 }, { "epoch": 0.47533338787531704, "grad_norm": 0.2599758505821228, "learning_rate": 9.989199264477894e-06, "loss": 0.0045, "step": 29050 }, { "epoch": 0.47549701382639287, "grad_norm": 0.15279127657413483, "learning_rate": 9.989152311593953e-06, "loss": 0.0054, "step": 29060 }, { "epoch": 0.4756606397774687, "grad_norm": 0.14165113866329193, "learning_rate": 9.989105256985438e-06, "loss": 0.0037, "step": 29070 }, { "epoch": 0.4758242657285445, "grad_norm": 0.07169383019208908, "learning_rate": 9.989058100653312e-06, "loss": 0.0035, "step": 29080 }, { "epoch": 0.4759878916796204, "grad_norm": 0.30929306149482727, "learning_rate": 9.989010842598533e-06, "loss": 0.0054, "step": 29090 }, { "epoch": 0.47615151763069624, "grad_norm": 0.1315333992242813, "learning_rate": 9.988963482822066e-06, "loss": 0.0026, "step": 29100 }, { "epoch": 0.47631514358177207, "grad_norm": 0.057822972536087036, "learning_rate": 9.988916021324877e-06, "loss": 0.0038, "step": 29110 }, { "epoch": 0.4764787695328479, "grad_norm": 0.06550322473049164, "learning_rate": 9.988868458107931e-06, "loss": 0.0045, "step": 29120 }, { "epoch": 0.47664239548392373, "grad_norm": 0.1842125654220581, "learning_rate": 9.988820793172204e-06, "loss": 0.0042, "step": 29130 }, { "epoch": 0.4768060214349996, "grad_norm": 0.13346388936042786, "learning_rate": 9.988773026518661e-06, "loss": 0.0051, "step": 29140 }, { "epoch": 0.47696964738607545, "grad_norm": 0.17408321797847748, "learning_rate": 9.98872515814828e-06, "loss": 0.0045, "step": 29150 }, { "epoch": 0.4771332733371513, "grad_norm": 0.06286933273077011, "learning_rate": 9.988677188062036e-06, "loss": 0.0031, "step": 29160 }, { "epoch": 0.4772968992882271, "grad_norm": 0.05195888504385948, "learning_rate": 9.988629116260905e-06, "loss": 0.0027, "step": 29170 }, { "epoch": 0.47746052523930294, "grad_norm": 0.0836397260427475, "learning_rate": 9.988580942745871e-06, "loss": 0.0042, "step": 29180 }, { "epoch": 0.4776241511903788, "grad_norm": 0.12860994040966034, "learning_rate": 9.988532667517914e-06, "loss": 0.0034, "step": 29190 }, { "epoch": 0.47778777714145465, "grad_norm": 0.08134452998638153, "learning_rate": 9.988484290578018e-06, "loss": 0.0068, "step": 29200 }, { "epoch": 0.4779514030925305, "grad_norm": 0.06096518784761429, "learning_rate": 9.98843581192717e-06, "loss": 0.0032, "step": 29210 }, { "epoch": 0.4781150290436063, "grad_norm": 0.2294633537530899, "learning_rate": 9.988387231566358e-06, "loss": 0.0038, "step": 29220 }, { "epoch": 0.47827865499468214, "grad_norm": 0.14104579389095306, "learning_rate": 9.988338549496574e-06, "loss": 0.0028, "step": 29230 }, { "epoch": 0.47844228094575797, "grad_norm": 0.16806736588478088, "learning_rate": 9.988289765718808e-06, "loss": 0.004, "step": 29240 }, { "epoch": 0.47860590689683385, "grad_norm": 0.045745979994535446, "learning_rate": 9.988240880234057e-06, "loss": 0.0039, "step": 29250 }, { "epoch": 0.4787695328479097, "grad_norm": 0.0668676346540451, "learning_rate": 9.988191893043317e-06, "loss": 0.0064, "step": 29260 }, { "epoch": 0.4789331587989855, "grad_norm": 0.12628376483917236, "learning_rate": 9.988142804147584e-06, "loss": 0.0037, "step": 29270 }, { "epoch": 0.47909678475006134, "grad_norm": 0.20727810263633728, "learning_rate": 9.988093613547864e-06, "loss": 0.0056, "step": 29280 }, { "epoch": 0.4792604107011372, "grad_norm": 0.04622520133852959, "learning_rate": 9.988044321245159e-06, "loss": 0.0027, "step": 29290 }, { "epoch": 0.47942403665221306, "grad_norm": 0.033117543905973434, "learning_rate": 9.987994927240469e-06, "loss": 0.003, "step": 29300 }, { "epoch": 0.4795876626032889, "grad_norm": 0.08094853162765503, "learning_rate": 9.987945431534805e-06, "loss": 0.0025, "step": 29310 }, { "epoch": 0.4797512885543647, "grad_norm": 0.13202032446861267, "learning_rate": 9.987895834129177e-06, "loss": 0.0029, "step": 29320 }, { "epoch": 0.47991491450544055, "grad_norm": 0.1861954629421234, "learning_rate": 9.987846135024594e-06, "loss": 0.0052, "step": 29330 }, { "epoch": 0.4800785404565164, "grad_norm": 0.11726203560829163, "learning_rate": 9.987796334222073e-06, "loss": 0.0036, "step": 29340 }, { "epoch": 0.48024216640759226, "grad_norm": 0.04961273446679115, "learning_rate": 9.987746431722625e-06, "loss": 0.0052, "step": 29350 }, { "epoch": 0.4804057923586681, "grad_norm": 0.033906035125255585, "learning_rate": 9.987696427527268e-06, "loss": 0.0037, "step": 29360 }, { "epoch": 0.4805694183097439, "grad_norm": 0.14672502875328064, "learning_rate": 9.987646321637023e-06, "loss": 0.0045, "step": 29370 }, { "epoch": 0.48073304426081975, "grad_norm": 0.1271412968635559, "learning_rate": 9.987596114052912e-06, "loss": 0.004, "step": 29380 }, { "epoch": 0.4808966702118956, "grad_norm": 0.09657371044158936, "learning_rate": 9.987545804775958e-06, "loss": 0.0034, "step": 29390 }, { "epoch": 0.48106029616297147, "grad_norm": 0.05304640531539917, "learning_rate": 9.987495393807188e-06, "loss": 0.0032, "step": 29400 }, { "epoch": 0.4812239221140473, "grad_norm": 0.038368843495845795, "learning_rate": 9.987444881147627e-06, "loss": 0.0023, "step": 29410 }, { "epoch": 0.4813875480651231, "grad_norm": 0.10408424586057663, "learning_rate": 9.987394266798308e-06, "loss": 0.0038, "step": 29420 }, { "epoch": 0.48155117401619896, "grad_norm": 0.06250657141208649, "learning_rate": 9.98734355076026e-06, "loss": 0.0036, "step": 29430 }, { "epoch": 0.4817147999672748, "grad_norm": 0.06074332818388939, "learning_rate": 9.987292733034518e-06, "loss": 0.0032, "step": 29440 }, { "epoch": 0.4818784259183507, "grad_norm": 0.07642721384763718, "learning_rate": 9.98724181362212e-06, "loss": 0.0024, "step": 29450 }, { "epoch": 0.4820420518694265, "grad_norm": 0.026789678260684013, "learning_rate": 9.987190792524101e-06, "loss": 0.0036, "step": 29460 }, { "epoch": 0.48220567782050233, "grad_norm": 0.18355298042297363, "learning_rate": 9.987139669741506e-06, "loss": 0.0049, "step": 29470 }, { "epoch": 0.48236930377157816, "grad_norm": 0.11854942888021469, "learning_rate": 9.987088445275375e-06, "loss": 0.0057, "step": 29480 }, { "epoch": 0.482532929722654, "grad_norm": 0.0933738648891449, "learning_rate": 9.98703711912675e-06, "loss": 0.0028, "step": 29490 }, { "epoch": 0.4826965556737299, "grad_norm": 0.26762449741363525, "learning_rate": 9.986985691296679e-06, "loss": 0.0052, "step": 29500 }, { "epoch": 0.4828601816248057, "grad_norm": 0.3076751232147217, "learning_rate": 9.986934161786212e-06, "loss": 0.005, "step": 29510 }, { "epoch": 0.48302380757588154, "grad_norm": 0.17378439009189606, "learning_rate": 9.986882530596398e-06, "loss": 0.0058, "step": 29520 }, { "epoch": 0.48318743352695737, "grad_norm": 0.10537422448396683, "learning_rate": 9.986830797728292e-06, "loss": 0.0039, "step": 29530 }, { "epoch": 0.4833510594780332, "grad_norm": 0.14493772387504578, "learning_rate": 9.986778963182945e-06, "loss": 0.0034, "step": 29540 }, { "epoch": 0.4835146854291091, "grad_norm": 0.14214463531970978, "learning_rate": 9.986727026961419e-06, "loss": 0.0033, "step": 29550 }, { "epoch": 0.4836783113801849, "grad_norm": 0.1347922831773758, "learning_rate": 9.986674989064768e-06, "loss": 0.0048, "step": 29560 }, { "epoch": 0.48384193733126074, "grad_norm": 0.026316815987229347, "learning_rate": 9.986622849494054e-06, "loss": 0.0038, "step": 29570 }, { "epoch": 0.48400556328233657, "grad_norm": 0.17439676821231842, "learning_rate": 9.986570608250343e-06, "loss": 0.0053, "step": 29580 }, { "epoch": 0.4841691892334124, "grad_norm": 1.7108169794082642, "learning_rate": 9.986518265334696e-06, "loss": 0.0081, "step": 29590 }, { "epoch": 0.4843328151844883, "grad_norm": 0.22583450376987457, "learning_rate": 9.986465820748186e-06, "loss": 0.0067, "step": 29600 }, { "epoch": 0.4844964411355641, "grad_norm": 0.22242532670497894, "learning_rate": 9.986413274491876e-06, "loss": 0.0029, "step": 29610 }, { "epoch": 0.48466006708663995, "grad_norm": 0.06477893143892288, "learning_rate": 9.98636062656684e-06, "loss": 0.0036, "step": 29620 }, { "epoch": 0.4848236930377158, "grad_norm": 0.20788834989070892, "learning_rate": 9.986307876974153e-06, "loss": 0.0046, "step": 29630 }, { "epoch": 0.4849873189887916, "grad_norm": 0.07671323418617249, "learning_rate": 9.986255025714888e-06, "loss": 0.0036, "step": 29640 }, { "epoch": 0.4851509449398675, "grad_norm": 0.1357870101928711, "learning_rate": 9.986202072790125e-06, "loss": 0.0042, "step": 29650 }, { "epoch": 0.4853145708909433, "grad_norm": 0.34036198258399963, "learning_rate": 9.98614901820094e-06, "loss": 0.004, "step": 29660 }, { "epoch": 0.48547819684201915, "grad_norm": 0.11370505392551422, "learning_rate": 9.986095861948419e-06, "loss": 0.0039, "step": 29670 }, { "epoch": 0.485641822793095, "grad_norm": 0.14968065917491913, "learning_rate": 9.986042604033644e-06, "loss": 0.0039, "step": 29680 }, { "epoch": 0.4858054487441708, "grad_norm": 0.11812610179185867, "learning_rate": 9.9859892444577e-06, "loss": 0.0059, "step": 29690 }, { "epoch": 0.48596907469524664, "grad_norm": 0.14844104647636414, "learning_rate": 9.985935783221676e-06, "loss": 0.003, "step": 29700 }, { "epoch": 0.4861327006463225, "grad_norm": 0.14661206305027008, "learning_rate": 9.98588222032666e-06, "loss": 0.004, "step": 29710 }, { "epoch": 0.48629632659739835, "grad_norm": 0.2372039258480072, "learning_rate": 9.985828555773748e-06, "loss": 0.0053, "step": 29720 }, { "epoch": 0.4864599525484742, "grad_norm": 0.196861132979393, "learning_rate": 9.985774789564033e-06, "loss": 0.0035, "step": 29730 }, { "epoch": 0.48662357849955, "grad_norm": 0.1689983457326889, "learning_rate": 9.98572092169861e-06, "loss": 0.0042, "step": 29740 }, { "epoch": 0.48678720445062584, "grad_norm": 0.08349242806434631, "learning_rate": 9.985666952178576e-06, "loss": 0.0041, "step": 29750 }, { "epoch": 0.48695083040170173, "grad_norm": 0.26917514204978943, "learning_rate": 9.985612881005034e-06, "loss": 0.0036, "step": 29760 }, { "epoch": 0.48711445635277756, "grad_norm": 0.0674847811460495, "learning_rate": 9.985558708179083e-06, "loss": 0.0033, "step": 29770 }, { "epoch": 0.4872780823038534, "grad_norm": 0.359036386013031, "learning_rate": 9.985504433701834e-06, "loss": 0.0059, "step": 29780 }, { "epoch": 0.4874417082549292, "grad_norm": 0.11143141239881516, "learning_rate": 9.985450057574387e-06, "loss": 0.0024, "step": 29790 }, { "epoch": 0.48760533420600505, "grad_norm": 0.12133438140153885, "learning_rate": 9.985395579797855e-06, "loss": 0.0064, "step": 29800 }, { "epoch": 0.48776896015708093, "grad_norm": 0.05216781795024872, "learning_rate": 9.985341000373345e-06, "loss": 0.0029, "step": 29810 }, { "epoch": 0.48793258610815676, "grad_norm": 0.07204172015190125, "learning_rate": 9.985286319301972e-06, "loss": 0.003, "step": 29820 }, { "epoch": 0.4880962120592326, "grad_norm": 0.02983291633427143, "learning_rate": 9.985231536584852e-06, "loss": 0.0026, "step": 29830 }, { "epoch": 0.4882598380103084, "grad_norm": 0.09569099545478821, "learning_rate": 9.985176652223101e-06, "loss": 0.0045, "step": 29840 }, { "epoch": 0.48842346396138425, "grad_norm": 0.14981193840503693, "learning_rate": 9.985121666217838e-06, "loss": 0.0036, "step": 29850 }, { "epoch": 0.48858708991246014, "grad_norm": 0.09515146166086197, "learning_rate": 9.985066578570184e-06, "loss": 0.0041, "step": 29860 }, { "epoch": 0.48875071586353597, "grad_norm": 0.269700288772583, "learning_rate": 9.98501138928126e-06, "loss": 0.0047, "step": 29870 }, { "epoch": 0.4889143418146118, "grad_norm": 0.2091132551431656, "learning_rate": 9.984956098352194e-06, "loss": 0.0048, "step": 29880 }, { "epoch": 0.4890779677656876, "grad_norm": 0.15794916450977325, "learning_rate": 9.984900705784115e-06, "loss": 0.0039, "step": 29890 }, { "epoch": 0.48924159371676346, "grad_norm": 0.189207524061203, "learning_rate": 9.984845211578147e-06, "loss": 0.003, "step": 29900 }, { "epoch": 0.48940521966783934, "grad_norm": 0.2440270483493805, "learning_rate": 9.984789615735427e-06, "loss": 0.0038, "step": 29910 }, { "epoch": 0.48956884561891517, "grad_norm": 0.1895362138748169, "learning_rate": 9.984733918257085e-06, "loss": 0.0032, "step": 29920 }, { "epoch": 0.489732471569991, "grad_norm": 0.2710678279399872, "learning_rate": 9.984678119144258e-06, "loss": 0.0029, "step": 29930 }, { "epoch": 0.48989609752106683, "grad_norm": 0.035962022840976715, "learning_rate": 9.984622218398083e-06, "loss": 0.0023, "step": 29940 }, { "epoch": 0.49005972347214266, "grad_norm": 0.13316458463668823, "learning_rate": 9.984566216019701e-06, "loss": 0.0036, "step": 29950 }, { "epoch": 0.49022334942321855, "grad_norm": 0.20803581178188324, "learning_rate": 9.984510112010253e-06, "loss": 0.0024, "step": 29960 }, { "epoch": 0.4903869753742944, "grad_norm": 0.16329339146614075, "learning_rate": 9.984453906370882e-06, "loss": 0.0045, "step": 29970 }, { "epoch": 0.4905506013253702, "grad_norm": 0.17576371133327484, "learning_rate": 9.984397599102735e-06, "loss": 0.0039, "step": 29980 }, { "epoch": 0.49071422727644604, "grad_norm": 0.11398075520992279, "learning_rate": 9.984341190206961e-06, "loss": 0.0039, "step": 29990 }, { "epoch": 0.49087785322752187, "grad_norm": 0.19422316551208496, "learning_rate": 9.98428467968471e-06, "loss": 0.0022, "step": 30000 }, { "epoch": 0.49104147917859775, "grad_norm": 0.2841072678565979, "learning_rate": 9.984228067537132e-06, "loss": 0.0026, "step": 30010 }, { "epoch": 0.4912051051296736, "grad_norm": 0.19866299629211426, "learning_rate": 9.984171353765383e-06, "loss": 0.0041, "step": 30020 }, { "epoch": 0.4913687310807494, "grad_norm": 0.41484084725379944, "learning_rate": 9.984114538370619e-06, "loss": 0.0037, "step": 30030 }, { "epoch": 0.49153235703182524, "grad_norm": 0.13038912415504456, "learning_rate": 9.984057621353997e-06, "loss": 0.0044, "step": 30040 }, { "epoch": 0.49169598298290107, "grad_norm": 0.04600489139556885, "learning_rate": 9.984000602716681e-06, "loss": 0.0023, "step": 30050 }, { "epoch": 0.49185960893397696, "grad_norm": 0.09092388302087784, "learning_rate": 9.98394348245983e-06, "loss": 0.0057, "step": 30060 }, { "epoch": 0.4920232348850528, "grad_norm": 0.12364479154348373, "learning_rate": 9.983886260584612e-06, "loss": 0.0041, "step": 30070 }, { "epoch": 0.4921868608361286, "grad_norm": 0.2692112326622009, "learning_rate": 9.98382893709219e-06, "loss": 0.0037, "step": 30080 }, { "epoch": 0.49235048678720444, "grad_norm": 0.23460355401039124, "learning_rate": 9.983771511983735e-06, "loss": 0.0038, "step": 30090 }, { "epoch": 0.4925141127382803, "grad_norm": 0.06248888745903969, "learning_rate": 9.983713985260418e-06, "loss": 0.0027, "step": 30100 }, { "epoch": 0.49267773868935616, "grad_norm": 0.09914436936378479, "learning_rate": 9.98365635692341e-06, "loss": 0.0035, "step": 30110 }, { "epoch": 0.492841364640432, "grad_norm": 0.210234597325325, "learning_rate": 9.98359862697389e-06, "loss": 0.004, "step": 30120 }, { "epoch": 0.4930049905915078, "grad_norm": 0.10494538396596909, "learning_rate": 9.98354079541303e-06, "loss": 0.0035, "step": 30130 }, { "epoch": 0.49316861654258365, "grad_norm": 0.05746842920780182, "learning_rate": 9.983482862242011e-06, "loss": 0.0035, "step": 30140 }, { "epoch": 0.4933322424936595, "grad_norm": 0.16199208796024323, "learning_rate": 9.983424827462016e-06, "loss": 0.0047, "step": 30150 }, { "epoch": 0.4934958684447353, "grad_norm": 0.20254850387573242, "learning_rate": 9.983366691074228e-06, "loss": 0.0038, "step": 30160 }, { "epoch": 0.4936594943958112, "grad_norm": 1.3658519983291626, "learning_rate": 9.98330845307983e-06, "loss": 0.004, "step": 30170 }, { "epoch": 0.493823120346887, "grad_norm": 0.10895184427499771, "learning_rate": 9.983250113480009e-06, "loss": 0.0033, "step": 30180 }, { "epoch": 0.49398674629796285, "grad_norm": 0.11468274146318436, "learning_rate": 9.98319167227596e-06, "loss": 0.0035, "step": 30190 }, { "epoch": 0.4941503722490387, "grad_norm": 0.11654379963874817, "learning_rate": 9.983133129468869e-06, "loss": 0.0043, "step": 30200 }, { "epoch": 0.4943139982001145, "grad_norm": 0.3892858028411865, "learning_rate": 9.98307448505993e-06, "loss": 0.0029, "step": 30210 }, { "epoch": 0.4944776241511904, "grad_norm": 0.23351286351680756, "learning_rate": 9.983015739050343e-06, "loss": 0.0054, "step": 30220 }, { "epoch": 0.49464125010226623, "grad_norm": 0.10672592371702194, "learning_rate": 9.982956891441303e-06, "loss": 0.0047, "step": 30230 }, { "epoch": 0.49480487605334206, "grad_norm": 0.05478107929229736, "learning_rate": 9.982897942234008e-06, "loss": 0.0037, "step": 30240 }, { "epoch": 0.4949685020044179, "grad_norm": 0.06325937807559967, "learning_rate": 9.982838891429662e-06, "loss": 0.0031, "step": 30250 }, { "epoch": 0.4951321279554937, "grad_norm": 0.07520751655101776, "learning_rate": 9.98277973902947e-06, "loss": 0.0041, "step": 30260 }, { "epoch": 0.4952957539065696, "grad_norm": 0.07135716825723648, "learning_rate": 9.982720485034637e-06, "loss": 0.0032, "step": 30270 }, { "epoch": 0.49545937985764543, "grad_norm": 0.12278308719396591, "learning_rate": 9.982661129446369e-06, "loss": 0.0024, "step": 30280 }, { "epoch": 0.49562300580872126, "grad_norm": 0.42663830518722534, "learning_rate": 9.982601672265879e-06, "loss": 0.0036, "step": 30290 }, { "epoch": 0.4957866317597971, "grad_norm": 0.21377742290496826, "learning_rate": 9.982542113494378e-06, "loss": 0.0021, "step": 30300 }, { "epoch": 0.4959502577108729, "grad_norm": 0.1452416330575943, "learning_rate": 9.982482453133083e-06, "loss": 0.0037, "step": 30310 }, { "epoch": 0.4961138836619488, "grad_norm": 0.07587427645921707, "learning_rate": 9.982422691183206e-06, "loss": 0.0036, "step": 30320 }, { "epoch": 0.49627750961302464, "grad_norm": 0.18142090737819672, "learning_rate": 9.982362827645968e-06, "loss": 0.003, "step": 30330 }, { "epoch": 0.49644113556410047, "grad_norm": 0.2890045940876007, "learning_rate": 9.982302862522591e-06, "loss": 0.0039, "step": 30340 }, { "epoch": 0.4966047615151763, "grad_norm": 0.09478327631950378, "learning_rate": 9.982242795814294e-06, "loss": 0.0046, "step": 30350 }, { "epoch": 0.4967683874662521, "grad_norm": 0.03056584671139717, "learning_rate": 9.982182627522304e-06, "loss": 0.0037, "step": 30360 }, { "epoch": 0.496932013417328, "grad_norm": 0.05623801052570343, "learning_rate": 9.982122357647848e-06, "loss": 0.0031, "step": 30370 }, { "epoch": 0.49709563936840384, "grad_norm": 0.2949344515800476, "learning_rate": 9.982061986192153e-06, "loss": 0.0039, "step": 30380 }, { "epoch": 0.49725926531947967, "grad_norm": 0.14083367586135864, "learning_rate": 9.982001513156452e-06, "loss": 0.0065, "step": 30390 }, { "epoch": 0.4974228912705555, "grad_norm": 0.16157329082489014, "learning_rate": 9.981940938541977e-06, "loss": 0.0042, "step": 30400 }, { "epoch": 0.49758651722163133, "grad_norm": 0.06314056366682053, "learning_rate": 9.981880262349963e-06, "loss": 0.0047, "step": 30410 }, { "epoch": 0.4977501431727072, "grad_norm": 0.18740496039390564, "learning_rate": 9.981819484581649e-06, "loss": 0.0067, "step": 30420 }, { "epoch": 0.49791376912378305, "grad_norm": 0.04666244238615036, "learning_rate": 9.981758605238272e-06, "loss": 0.0037, "step": 30430 }, { "epoch": 0.4980773950748589, "grad_norm": 0.1789308488368988, "learning_rate": 9.981697624321073e-06, "loss": 0.0021, "step": 30440 }, { "epoch": 0.4982410210259347, "grad_norm": 0.06411601603031158, "learning_rate": 9.981636541831297e-06, "loss": 0.0057, "step": 30450 }, { "epoch": 0.49840464697701053, "grad_norm": 0.1273791342973709, "learning_rate": 9.981575357770187e-06, "loss": 0.0031, "step": 30460 }, { "epoch": 0.4985682729280864, "grad_norm": 0.10507764667272568, "learning_rate": 9.981514072138993e-06, "loss": 0.0036, "step": 30470 }, { "epoch": 0.49873189887916225, "grad_norm": 0.08814719319343567, "learning_rate": 9.981452684938966e-06, "loss": 0.0032, "step": 30480 }, { "epoch": 0.4988955248302381, "grad_norm": 0.13190239667892456, "learning_rate": 9.981391196171353e-06, "loss": 0.0026, "step": 30490 }, { "epoch": 0.4990591507813139, "grad_norm": 0.09077088534832001, "learning_rate": 9.981329605837412e-06, "loss": 0.0047, "step": 30500 }, { "epoch": 0.49922277673238974, "grad_norm": 0.11965163052082062, "learning_rate": 9.981267913938393e-06, "loss": 0.0045, "step": 30510 }, { "epoch": 0.4993864026834656, "grad_norm": 0.13430318236351013, "learning_rate": 9.981206120475561e-06, "loss": 0.0041, "step": 30520 }, { "epoch": 0.49955002863454145, "grad_norm": 0.1174863651394844, "learning_rate": 9.98114422545017e-06, "loss": 0.0026, "step": 30530 }, { "epoch": 0.4997136545856173, "grad_norm": 0.1295466423034668, "learning_rate": 9.981082228863487e-06, "loss": 0.0039, "step": 30540 }, { "epoch": 0.4998772805366931, "grad_norm": 0.08947909623384476, "learning_rate": 9.981020130716772e-06, "loss": 0.0049, "step": 30550 }, { "epoch": 0.500040906487769, "grad_norm": 0.08584802597761154, "learning_rate": 9.980957931011294e-06, "loss": 0.0039, "step": 30560 }, { "epoch": 0.5002045324388448, "grad_norm": 0.1037212386727333, "learning_rate": 9.98089562974832e-06, "loss": 0.0048, "step": 30570 }, { "epoch": 0.5003681583899207, "grad_norm": 0.14456059038639069, "learning_rate": 9.980833226929118e-06, "loss": 0.0033, "step": 30580 }, { "epoch": 0.5005317843409964, "grad_norm": 0.15957953035831451, "learning_rate": 9.980770722554965e-06, "loss": 0.0037, "step": 30590 }, { "epoch": 0.5006954102920723, "grad_norm": 0.07862874865531921, "learning_rate": 9.98070811662713e-06, "loss": 0.0029, "step": 30600 }, { "epoch": 0.5008590362431482, "grad_norm": 0.13384631276130676, "learning_rate": 9.980645409146895e-06, "loss": 0.0046, "step": 30610 }, { "epoch": 0.501022662194224, "grad_norm": 0.14607301354408264, "learning_rate": 9.980582600115536e-06, "loss": 0.0044, "step": 30620 }, { "epoch": 0.5011862881452999, "grad_norm": 0.1722712516784668, "learning_rate": 9.980519689534334e-06, "loss": 0.0034, "step": 30630 }, { "epoch": 0.5013499140963756, "grad_norm": 0.05313633754849434, "learning_rate": 9.98045667740457e-06, "loss": 0.0052, "step": 30640 }, { "epoch": 0.5015135400474515, "grad_norm": 0.10245150327682495, "learning_rate": 9.98039356372753e-06, "loss": 0.0039, "step": 30650 }, { "epoch": 0.5016771659985274, "grad_norm": 0.11192922294139862, "learning_rate": 9.980330348504502e-06, "loss": 0.0043, "step": 30660 }, { "epoch": 0.5018407919496032, "grad_norm": 0.02683756873011589, "learning_rate": 9.980267031736774e-06, "loss": 0.0037, "step": 30670 }, { "epoch": 0.5020044179006791, "grad_norm": 0.14306199550628662, "learning_rate": 9.980203613425636e-06, "loss": 0.0022, "step": 30680 }, { "epoch": 0.5021680438517548, "grad_norm": 0.11620167642831802, "learning_rate": 9.98014009357238e-06, "loss": 0.0029, "step": 30690 }, { "epoch": 0.5023316698028307, "grad_norm": 0.0703018382191658, "learning_rate": 9.980076472178307e-06, "loss": 0.0029, "step": 30700 }, { "epoch": 0.5024952957539066, "grad_norm": 0.4483928680419922, "learning_rate": 9.980012749244707e-06, "loss": 0.0068, "step": 30710 }, { "epoch": 0.5026589217049824, "grad_norm": 0.01901552639901638, "learning_rate": 9.979948924772884e-06, "loss": 0.005, "step": 30720 }, { "epoch": 0.5028225476560583, "grad_norm": 0.11628527194261551, "learning_rate": 9.979884998764136e-06, "loss": 0.0033, "step": 30730 }, { "epoch": 0.502986173607134, "grad_norm": 0.20760121941566467, "learning_rate": 9.979820971219768e-06, "loss": 0.0027, "step": 30740 }, { "epoch": 0.5031497995582099, "grad_norm": 0.06874076277017593, "learning_rate": 9.979756842141088e-06, "loss": 0.0039, "step": 30750 }, { "epoch": 0.5033134255092858, "grad_norm": 0.07576856762170792, "learning_rate": 9.9796926115294e-06, "loss": 0.0032, "step": 30760 }, { "epoch": 0.5034770514603616, "grad_norm": 0.1414451152086258, "learning_rate": 9.979628279386015e-06, "loss": 0.0034, "step": 30770 }, { "epoch": 0.5036406774114375, "grad_norm": 0.021483853459358215, "learning_rate": 9.979563845712244e-06, "loss": 0.0033, "step": 30780 }, { "epoch": 0.5038043033625133, "grad_norm": 0.07633334398269653, "learning_rate": 9.979499310509402e-06, "loss": 0.0039, "step": 30790 }, { "epoch": 0.5039679293135891, "grad_norm": 0.14658431708812714, "learning_rate": 9.979434673778803e-06, "loss": 0.0029, "step": 30800 }, { "epoch": 0.504131555264665, "grad_norm": 0.11777408421039581, "learning_rate": 9.979369935521766e-06, "loss": 0.0026, "step": 30810 }, { "epoch": 0.5042951812157408, "grad_norm": 0.04312283918261528, "learning_rate": 9.97930509573961e-06, "loss": 0.002, "step": 30820 }, { "epoch": 0.5044588071668167, "grad_norm": 0.06003951653838158, "learning_rate": 9.979240154433658e-06, "loss": 0.0041, "step": 30830 }, { "epoch": 0.5046224331178925, "grad_norm": 0.28367507457733154, "learning_rate": 9.979175111605235e-06, "loss": 0.0029, "step": 30840 }, { "epoch": 0.5047860590689683, "grad_norm": 0.10103906691074371, "learning_rate": 9.979109967255666e-06, "loss": 0.0033, "step": 30850 }, { "epoch": 0.5049496850200442, "grad_norm": 0.23807662725448608, "learning_rate": 9.97904472138628e-06, "loss": 0.0047, "step": 30860 }, { "epoch": 0.50511331097112, "grad_norm": 0.4172917306423187, "learning_rate": 9.978979373998405e-06, "loss": 0.0041, "step": 30870 }, { "epoch": 0.5052769369221959, "grad_norm": 0.271973192691803, "learning_rate": 9.978913925093375e-06, "loss": 0.0042, "step": 30880 }, { "epoch": 0.5054405628732717, "grad_norm": 0.23000481724739075, "learning_rate": 9.978848374672525e-06, "loss": 0.0033, "step": 30890 }, { "epoch": 0.5056041888243475, "grad_norm": 0.1570647656917572, "learning_rate": 9.978782722737192e-06, "loss": 0.0038, "step": 30900 }, { "epoch": 0.5057678147754234, "grad_norm": 0.17170806229114532, "learning_rate": 9.978716969288713e-06, "loss": 0.0039, "step": 30910 }, { "epoch": 0.5059314407264992, "grad_norm": 0.22527004778385162, "learning_rate": 9.978651114328429e-06, "loss": 0.002, "step": 30920 }, { "epoch": 0.5060950666775751, "grad_norm": 0.2012941837310791, "learning_rate": 9.978585157857683e-06, "loss": 0.0033, "step": 30930 }, { "epoch": 0.5062586926286509, "grad_norm": 0.07829193025827408, "learning_rate": 9.978519099877819e-06, "loss": 0.0022, "step": 30940 }, { "epoch": 0.5064223185797267, "grad_norm": 0.3031608760356903, "learning_rate": 9.978452940390186e-06, "loss": 0.0033, "step": 30950 }, { "epoch": 0.5065859445308026, "grad_norm": 0.17665942013263702, "learning_rate": 9.97838667939613e-06, "loss": 0.0025, "step": 30960 }, { "epoch": 0.5067495704818784, "grad_norm": 0.12963825464248657, "learning_rate": 9.978320316897005e-06, "loss": 0.0033, "step": 30970 }, { "epoch": 0.5069131964329543, "grad_norm": 0.17414751648902893, "learning_rate": 9.978253852894162e-06, "loss": 0.0034, "step": 30980 }, { "epoch": 0.5070768223840301, "grad_norm": 0.13365744054317474, "learning_rate": 9.978187287388958e-06, "loss": 0.0057, "step": 30990 }, { "epoch": 0.507240448335106, "grad_norm": 0.18238407373428345, "learning_rate": 9.978120620382748e-06, "loss": 0.0038, "step": 31000 }, { "epoch": 0.5074040742861818, "grad_norm": 0.03989141434431076, "learning_rate": 9.978053851876892e-06, "loss": 0.0029, "step": 31010 }, { "epoch": 0.5075677002372576, "grad_norm": 0.25647038221359253, "learning_rate": 9.97798698187275e-06, "loss": 0.0051, "step": 31020 }, { "epoch": 0.5077313261883335, "grad_norm": 0.08180467784404755, "learning_rate": 9.977920010371691e-06, "loss": 0.0035, "step": 31030 }, { "epoch": 0.5078949521394093, "grad_norm": 0.10880085080862045, "learning_rate": 9.977852937375074e-06, "loss": 0.0057, "step": 31040 }, { "epoch": 0.5080585780904852, "grad_norm": 0.1811857372522354, "learning_rate": 9.97778576288427e-06, "loss": 0.0024, "step": 31050 }, { "epoch": 0.508222204041561, "grad_norm": 0.1547127515077591, "learning_rate": 9.977718486900647e-06, "loss": 0.0044, "step": 31060 }, { "epoch": 0.5083858299926368, "grad_norm": 0.23319856822490692, "learning_rate": 9.97765110942558e-06, "loss": 0.0052, "step": 31070 }, { "epoch": 0.5085494559437127, "grad_norm": 0.13729974627494812, "learning_rate": 9.977583630460437e-06, "loss": 0.0033, "step": 31080 }, { "epoch": 0.5087130818947885, "grad_norm": 0.0479813776910305, "learning_rate": 9.977516050006597e-06, "loss": 0.0035, "step": 31090 }, { "epoch": 0.5088767078458644, "grad_norm": 0.0657070055603981, "learning_rate": 9.977448368065438e-06, "loss": 0.0024, "step": 31100 }, { "epoch": 0.5090403337969402, "grad_norm": 0.21247324347496033, "learning_rate": 9.977380584638343e-06, "loss": 0.0044, "step": 31110 }, { "epoch": 0.509203959748016, "grad_norm": 0.03586040064692497, "learning_rate": 9.977312699726689e-06, "loss": 0.0061, "step": 31120 }, { "epoch": 0.5093675856990919, "grad_norm": 0.21733413636684418, "learning_rate": 9.977244713331861e-06, "loss": 0.0041, "step": 31130 }, { "epoch": 0.5095312116501677, "grad_norm": 0.12997771799564362, "learning_rate": 9.977176625455248e-06, "loss": 0.0037, "step": 31140 }, { "epoch": 0.5096948376012436, "grad_norm": 0.16956081986427307, "learning_rate": 9.977108436098237e-06, "loss": 0.0023, "step": 31150 }, { "epoch": 0.5098584635523195, "grad_norm": 0.2069341540336609, "learning_rate": 9.977040145262216e-06, "loss": 0.0027, "step": 31160 }, { "epoch": 0.5100220895033952, "grad_norm": 0.0917835608124733, "learning_rate": 9.976971752948582e-06, "loss": 0.0039, "step": 31170 }, { "epoch": 0.5101857154544711, "grad_norm": 0.24272052943706512, "learning_rate": 9.976903259158723e-06, "loss": 0.0026, "step": 31180 }, { "epoch": 0.5103493414055469, "grad_norm": 0.07934727519750595, "learning_rate": 9.976834663894042e-06, "loss": 0.0027, "step": 31190 }, { "epoch": 0.5105129673566228, "grad_norm": 0.0712093785405159, "learning_rate": 9.976765967155933e-06, "loss": 0.004, "step": 31200 }, { "epoch": 0.5106765933076987, "grad_norm": 0.11668846756219864, "learning_rate": 9.976697168945799e-06, "loss": 0.0029, "step": 31210 }, { "epoch": 0.5108402192587744, "grad_norm": 0.07452614605426788, "learning_rate": 9.976628269265044e-06, "loss": 0.0029, "step": 31220 }, { "epoch": 0.5110038452098503, "grad_norm": 0.04538750648498535, "learning_rate": 9.976559268115069e-06, "loss": 0.0027, "step": 31230 }, { "epoch": 0.5111674711609261, "grad_norm": 0.011634523048996925, "learning_rate": 9.976490165497283e-06, "loss": 0.0037, "step": 31240 }, { "epoch": 0.511331097112002, "grad_norm": 0.1263612061738968, "learning_rate": 9.976420961413096e-06, "loss": 0.0024, "step": 31250 }, { "epoch": 0.5114947230630777, "grad_norm": 0.1750522404909134, "learning_rate": 9.976351655863919e-06, "loss": 0.003, "step": 31260 }, { "epoch": 0.5116583490141536, "grad_norm": 0.06611786037683487, "learning_rate": 9.976282248851162e-06, "loss": 0.0041, "step": 31270 }, { "epoch": 0.5118219749652295, "grad_norm": 0.10906612128019333, "learning_rate": 9.976212740376241e-06, "loss": 0.0029, "step": 31280 }, { "epoch": 0.5119856009163053, "grad_norm": 0.07158281654119492, "learning_rate": 9.976143130440577e-06, "loss": 0.0033, "step": 31290 }, { "epoch": 0.5121492268673812, "grad_norm": 0.1089095026254654, "learning_rate": 9.976073419045585e-06, "loss": 0.004, "step": 31300 }, { "epoch": 0.512312852818457, "grad_norm": 0.16829991340637207, "learning_rate": 9.976003606192689e-06, "loss": 0.0036, "step": 31310 }, { "epoch": 0.5124764787695328, "grad_norm": 0.18437013030052185, "learning_rate": 9.97593369188331e-06, "loss": 0.0044, "step": 31320 }, { "epoch": 0.5126401047206087, "grad_norm": 0.14432205259799957, "learning_rate": 9.975863676118877e-06, "loss": 0.0039, "step": 31330 }, { "epoch": 0.5128037306716845, "grad_norm": 0.16965900361537933, "learning_rate": 9.975793558900814e-06, "loss": 0.0039, "step": 31340 }, { "epoch": 0.5129673566227604, "grad_norm": 0.1584232747554779, "learning_rate": 9.975723340230553e-06, "loss": 0.0043, "step": 31350 }, { "epoch": 0.5131309825738362, "grad_norm": 0.2336905598640442, "learning_rate": 9.975653020109524e-06, "loss": 0.0068, "step": 31360 }, { "epoch": 0.513294608524912, "grad_norm": 0.13500230014324188, "learning_rate": 9.975582598539161e-06, "loss": 0.0056, "step": 31370 }, { "epoch": 0.5134582344759879, "grad_norm": 0.0401373952627182, "learning_rate": 9.975512075520901e-06, "loss": 0.0042, "step": 31380 }, { "epoch": 0.5136218604270637, "grad_norm": 0.04150618240237236, "learning_rate": 9.975441451056183e-06, "loss": 0.0023, "step": 31390 }, { "epoch": 0.5137854863781396, "grad_norm": 0.1845923364162445, "learning_rate": 9.975370725146443e-06, "loss": 0.0044, "step": 31400 }, { "epoch": 0.5139491123292154, "grad_norm": 0.18148362636566162, "learning_rate": 9.975299897793124e-06, "loss": 0.0029, "step": 31410 }, { "epoch": 0.5141127382802912, "grad_norm": 0.1584545522928238, "learning_rate": 9.975228968997674e-06, "loss": 0.0061, "step": 31420 }, { "epoch": 0.5142763642313671, "grad_norm": 0.05871844291687012, "learning_rate": 9.975157938761536e-06, "loss": 0.0044, "step": 31430 }, { "epoch": 0.5144399901824429, "grad_norm": 0.15211592614650726, "learning_rate": 9.97508680708616e-06, "loss": 0.0028, "step": 31440 }, { "epoch": 0.5146036161335188, "grad_norm": 0.14249347150325775, "learning_rate": 9.975015573972994e-06, "loss": 0.0043, "step": 31450 }, { "epoch": 0.5147672420845946, "grad_norm": 0.11959750205278397, "learning_rate": 9.974944239423492e-06, "loss": 0.0049, "step": 31460 }, { "epoch": 0.5149308680356705, "grad_norm": 0.17216582596302032, "learning_rate": 9.974872803439108e-06, "loss": 0.0041, "step": 31470 }, { "epoch": 0.5150944939867463, "grad_norm": 0.14213907718658447, "learning_rate": 9.974801266021296e-06, "loss": 0.0054, "step": 31480 }, { "epoch": 0.5152581199378221, "grad_norm": 0.14354561269283295, "learning_rate": 9.97472962717152e-06, "loss": 0.0101, "step": 31490 }, { "epoch": 0.515421745888898, "grad_norm": 0.19279952347278595, "learning_rate": 9.974657886891237e-06, "loss": 0.0056, "step": 31500 }, { "epoch": 0.5155853718399738, "grad_norm": 0.1345345675945282, "learning_rate": 9.97458604518191e-06, "loss": 0.0036, "step": 31510 }, { "epoch": 0.5157489977910497, "grad_norm": 0.23949047923088074, "learning_rate": 9.974514102045006e-06, "loss": 0.0028, "step": 31520 }, { "epoch": 0.5159126237421255, "grad_norm": 0.18810169398784637, "learning_rate": 9.974442057481989e-06, "loss": 0.0033, "step": 31530 }, { "epoch": 0.5160762496932013, "grad_norm": 0.15832187235355377, "learning_rate": 9.974369911494329e-06, "loss": 0.0027, "step": 31540 }, { "epoch": 0.5162398756442772, "grad_norm": 0.15148797631263733, "learning_rate": 9.974297664083496e-06, "loss": 0.0042, "step": 31550 }, { "epoch": 0.516403501595353, "grad_norm": 0.12291761487722397, "learning_rate": 9.974225315250965e-06, "loss": 0.0067, "step": 31560 }, { "epoch": 0.5165671275464289, "grad_norm": 0.15339897572994232, "learning_rate": 9.974152864998212e-06, "loss": 0.0025, "step": 31570 }, { "epoch": 0.5167307534975047, "grad_norm": 0.1379196047782898, "learning_rate": 9.97408031332671e-06, "loss": 0.0069, "step": 31580 }, { "epoch": 0.5168943794485805, "grad_norm": 0.26243141293525696, "learning_rate": 9.97400766023794e-06, "loss": 0.006, "step": 31590 }, { "epoch": 0.5170580053996564, "grad_norm": 0.0924554318189621, "learning_rate": 9.973934905733387e-06, "loss": 0.005, "step": 31600 }, { "epoch": 0.5172216313507322, "grad_norm": 0.1453755497932434, "learning_rate": 9.97386204981453e-06, "loss": 0.0047, "step": 31610 }, { "epoch": 0.5173852573018081, "grad_norm": 0.1966267079114914, "learning_rate": 9.973789092482856e-06, "loss": 0.0037, "step": 31620 }, { "epoch": 0.517548883252884, "grad_norm": 0.05384514853358269, "learning_rate": 9.973716033739853e-06, "loss": 0.0042, "step": 31630 }, { "epoch": 0.5177125092039597, "grad_norm": 0.11059384793043137, "learning_rate": 9.973642873587009e-06, "loss": 0.0048, "step": 31640 }, { "epoch": 0.5178761351550356, "grad_norm": 0.05042577162384987, "learning_rate": 9.973569612025817e-06, "loss": 0.0037, "step": 31650 }, { "epoch": 0.5180397611061114, "grad_norm": 0.15426835417747498, "learning_rate": 9.97349624905777e-06, "loss": 0.0034, "step": 31660 }, { "epoch": 0.5182033870571873, "grad_norm": 0.12459656596183777, "learning_rate": 9.973422784684365e-06, "loss": 0.0037, "step": 31670 }, { "epoch": 0.5183670130082632, "grad_norm": 0.07639476656913757, "learning_rate": 9.9733492189071e-06, "loss": 0.0044, "step": 31680 }, { "epoch": 0.5185306389593389, "grad_norm": 0.21424080431461334, "learning_rate": 9.973275551727473e-06, "loss": 0.0035, "step": 31690 }, { "epoch": 0.5186942649104148, "grad_norm": 0.11782193928956985, "learning_rate": 9.973201783146989e-06, "loss": 0.0037, "step": 31700 }, { "epoch": 0.5188578908614906, "grad_norm": 0.08843778818845749, "learning_rate": 9.973127913167147e-06, "loss": 0.0041, "step": 31710 }, { "epoch": 0.5190215168125665, "grad_norm": 0.17665037512779236, "learning_rate": 9.973053941789458e-06, "loss": 0.0048, "step": 31720 }, { "epoch": 0.5191851427636424, "grad_norm": 0.2969178557395935, "learning_rate": 9.97297986901543e-06, "loss": 0.0033, "step": 31730 }, { "epoch": 0.5193487687147181, "grad_norm": 0.1458924412727356, "learning_rate": 9.972905694846569e-06, "loss": 0.0032, "step": 31740 }, { "epoch": 0.519512394665794, "grad_norm": 0.1267065405845642, "learning_rate": 9.972831419284392e-06, "loss": 0.0041, "step": 31750 }, { "epoch": 0.5196760206168698, "grad_norm": 0.07648555189371109, "learning_rate": 9.97275704233041e-06, "loss": 0.0033, "step": 31760 }, { "epoch": 0.5198396465679457, "grad_norm": 0.07803991436958313, "learning_rate": 9.972682563986142e-06, "loss": 0.0028, "step": 31770 }, { "epoch": 0.5200032725190216, "grad_norm": 0.0907871425151825, "learning_rate": 9.972607984253107e-06, "loss": 0.0044, "step": 31780 }, { "epoch": 0.5201668984700973, "grad_norm": 0.277720183134079, "learning_rate": 9.972533303132822e-06, "loss": 0.0055, "step": 31790 }, { "epoch": 0.5203305244211732, "grad_norm": 0.09113767743110657, "learning_rate": 9.972458520626814e-06, "loss": 0.0033, "step": 31800 }, { "epoch": 0.520494150372249, "grad_norm": 0.0695081427693367, "learning_rate": 9.972383636736604e-06, "loss": 0.0024, "step": 31810 }, { "epoch": 0.5206577763233249, "grad_norm": 0.02042817510664463, "learning_rate": 9.972308651463722e-06, "loss": 0.0047, "step": 31820 }, { "epoch": 0.5208214022744008, "grad_norm": 0.08694383502006531, "learning_rate": 9.972233564809695e-06, "loss": 0.0043, "step": 31830 }, { "epoch": 0.5209850282254765, "grad_norm": 0.11548412591218948, "learning_rate": 9.972158376776053e-06, "loss": 0.0028, "step": 31840 }, { "epoch": 0.5211486541765524, "grad_norm": 0.21227474510669708, "learning_rate": 9.97208308736433e-06, "loss": 0.0032, "step": 31850 }, { "epoch": 0.5213122801276282, "grad_norm": 0.05764858424663544, "learning_rate": 9.972007696576065e-06, "loss": 0.0021, "step": 31860 }, { "epoch": 0.5214759060787041, "grad_norm": 0.16237956285476685, "learning_rate": 9.971932204412788e-06, "loss": 0.0046, "step": 31870 }, { "epoch": 0.52163953202978, "grad_norm": 0.260830283164978, "learning_rate": 9.971856610876043e-06, "loss": 0.0048, "step": 31880 }, { "epoch": 0.5218031579808557, "grad_norm": 0.09277407824993134, "learning_rate": 9.97178091596737e-06, "loss": 0.0037, "step": 31890 }, { "epoch": 0.5219667839319316, "grad_norm": 0.03722226619720459, "learning_rate": 9.971705119688314e-06, "loss": 0.0033, "step": 31900 }, { "epoch": 0.5221304098830074, "grad_norm": 0.09974231570959091, "learning_rate": 9.971629222040418e-06, "loss": 0.004, "step": 31910 }, { "epoch": 0.5222940358340833, "grad_norm": 0.24337823688983917, "learning_rate": 9.97155322302523e-06, "loss": 0.0025, "step": 31920 }, { "epoch": 0.5224576617851592, "grad_norm": 0.07261182367801666, "learning_rate": 9.971477122644299e-06, "loss": 0.0027, "step": 31930 }, { "epoch": 0.522621287736235, "grad_norm": 0.13581582903862, "learning_rate": 9.97140092089918e-06, "loss": 0.0026, "step": 31940 }, { "epoch": 0.5227849136873108, "grad_norm": 0.15152695775032043, "learning_rate": 9.971324617791423e-06, "loss": 0.0035, "step": 31950 }, { "epoch": 0.5229485396383866, "grad_norm": 0.21998485922813416, "learning_rate": 9.971248213322585e-06, "loss": 0.0036, "step": 31960 }, { "epoch": 0.5231121655894625, "grad_norm": 0.05461184307932854, "learning_rate": 9.971171707494223e-06, "loss": 0.0028, "step": 31970 }, { "epoch": 0.5232757915405384, "grad_norm": 0.2140350341796875, "learning_rate": 9.971095100307898e-06, "loss": 0.0036, "step": 31980 }, { "epoch": 0.5234394174916142, "grad_norm": 0.2744661569595337, "learning_rate": 9.971018391765173e-06, "loss": 0.0029, "step": 31990 }, { "epoch": 0.52360304344269, "grad_norm": 0.012217170558869839, "learning_rate": 9.970941581867608e-06, "loss": 0.0017, "step": 32000 }, { "epoch": 0.5237666693937658, "grad_norm": 0.06265369802713394, "learning_rate": 9.970864670616775e-06, "loss": 0.0032, "step": 32010 }, { "epoch": 0.5239302953448417, "grad_norm": 0.11410092562437057, "learning_rate": 9.970787658014235e-06, "loss": 0.0034, "step": 32020 }, { "epoch": 0.5240939212959176, "grad_norm": 0.13382509350776672, "learning_rate": 9.970710544061566e-06, "loss": 0.005, "step": 32030 }, { "epoch": 0.5242575472469934, "grad_norm": 0.17374125123023987, "learning_rate": 9.970633328760335e-06, "loss": 0.0041, "step": 32040 }, { "epoch": 0.5244211731980692, "grad_norm": 0.10937086492776871, "learning_rate": 9.970556012112117e-06, "loss": 0.003, "step": 32050 }, { "epoch": 0.524584799149145, "grad_norm": 0.14367762207984924, "learning_rate": 9.970478594118491e-06, "loss": 0.0034, "step": 32060 }, { "epoch": 0.5247484251002209, "grad_norm": 0.18541327118873596, "learning_rate": 9.970401074781034e-06, "loss": 0.0048, "step": 32070 }, { "epoch": 0.5249120510512968, "grad_norm": 0.07718891650438309, "learning_rate": 9.970323454101324e-06, "loss": 0.003, "step": 32080 }, { "epoch": 0.5250756770023726, "grad_norm": 0.27203840017318726, "learning_rate": 9.970245732080947e-06, "loss": 0.002, "step": 32090 }, { "epoch": 0.5252393029534485, "grad_norm": 0.05518939718604088, "learning_rate": 9.970167908721486e-06, "loss": 0.0031, "step": 32100 }, { "epoch": 0.5254029289045242, "grad_norm": 0.10870165377855301, "learning_rate": 9.970089984024529e-06, "loss": 0.0026, "step": 32110 }, { "epoch": 0.5255665548556001, "grad_norm": 0.10360418260097504, "learning_rate": 9.970011957991664e-06, "loss": 0.0031, "step": 32120 }, { "epoch": 0.5257301808066759, "grad_norm": 0.25197213888168335, "learning_rate": 9.969933830624482e-06, "loss": 0.0061, "step": 32130 }, { "epoch": 0.5258938067577518, "grad_norm": 0.16036970913410187, "learning_rate": 9.969855601924577e-06, "loss": 0.0038, "step": 32140 }, { "epoch": 0.5260574327088277, "grad_norm": 0.03433137759566307, "learning_rate": 9.969777271893542e-06, "loss": 0.0017, "step": 32150 }, { "epoch": 0.5262210586599034, "grad_norm": 0.20450782775878906, "learning_rate": 9.969698840532974e-06, "loss": 0.0032, "step": 32160 }, { "epoch": 0.5263846846109793, "grad_norm": 0.20976878702640533, "learning_rate": 9.969620307844474e-06, "loss": 0.0039, "step": 32170 }, { "epoch": 0.5265483105620551, "grad_norm": 0.06672423332929611, "learning_rate": 9.969541673829643e-06, "loss": 0.0053, "step": 32180 }, { "epoch": 0.526711936513131, "grad_norm": 0.0768200159072876, "learning_rate": 9.969462938490084e-06, "loss": 0.0032, "step": 32190 }, { "epoch": 0.5268755624642069, "grad_norm": 0.147769957780838, "learning_rate": 9.9693841018274e-06, "loss": 0.0052, "step": 32200 }, { "epoch": 0.5270391884152826, "grad_norm": 0.1060827448964119, "learning_rate": 9.969305163843202e-06, "loss": 0.0022, "step": 32210 }, { "epoch": 0.5272028143663585, "grad_norm": 0.17721708118915558, "learning_rate": 9.969226124539097e-06, "loss": 0.0031, "step": 32220 }, { "epoch": 0.5273664403174343, "grad_norm": 0.17007555067539215, "learning_rate": 9.9691469839167e-06, "loss": 0.0037, "step": 32230 }, { "epoch": 0.5275300662685102, "grad_norm": 0.12176928669214249, "learning_rate": 9.96906774197762e-06, "loss": 0.0017, "step": 32240 }, { "epoch": 0.5276936922195861, "grad_norm": 0.31315597891807556, "learning_rate": 9.968988398723476e-06, "loss": 0.0025, "step": 32250 }, { "epoch": 0.5278573181706618, "grad_norm": 0.14350135624408722, "learning_rate": 9.968908954155883e-06, "loss": 0.0035, "step": 32260 }, { "epoch": 0.5280209441217377, "grad_norm": 0.11943665146827698, "learning_rate": 9.968829408276463e-06, "loss": 0.0047, "step": 32270 }, { "epoch": 0.5281845700728135, "grad_norm": 0.1978522390127182, "learning_rate": 9.968749761086837e-06, "loss": 0.0046, "step": 32280 }, { "epoch": 0.5283481960238894, "grad_norm": 0.09488698095083237, "learning_rate": 9.96867001258863e-06, "loss": 0.0032, "step": 32290 }, { "epoch": 0.5285118219749653, "grad_norm": 0.15692515671253204, "learning_rate": 9.968590162783467e-06, "loss": 0.0043, "step": 32300 }, { "epoch": 0.528675447926041, "grad_norm": 0.15200050175189972, "learning_rate": 9.968510211672975e-06, "loss": 0.0045, "step": 32310 }, { "epoch": 0.5288390738771169, "grad_norm": 0.0860099047422409, "learning_rate": 9.968430159258785e-06, "loss": 0.003, "step": 32320 }, { "epoch": 0.5290026998281927, "grad_norm": 0.13482075929641724, "learning_rate": 9.968350005542532e-06, "loss": 0.0042, "step": 32330 }, { "epoch": 0.5291663257792686, "grad_norm": 0.09036947786808014, "learning_rate": 9.968269750525846e-06, "loss": 0.0056, "step": 32340 }, { "epoch": 0.5293299517303445, "grad_norm": 0.35281214118003845, "learning_rate": 9.968189394210366e-06, "loss": 0.004, "step": 32350 }, { "epoch": 0.5294935776814202, "grad_norm": 0.03177090734243393, "learning_rate": 9.968108936597729e-06, "loss": 0.002, "step": 32360 }, { "epoch": 0.5296572036324961, "grad_norm": 0.24317790567874908, "learning_rate": 9.968028377689575e-06, "loss": 0.0031, "step": 32370 }, { "epoch": 0.5298208295835719, "grad_norm": 0.060691870748996735, "learning_rate": 9.967947717487549e-06, "loss": 0.0036, "step": 32380 }, { "epoch": 0.5299844555346478, "grad_norm": 0.19004839658737183, "learning_rate": 9.967866955993296e-06, "loss": 0.0039, "step": 32390 }, { "epoch": 0.5301480814857237, "grad_norm": 0.16221140325069427, "learning_rate": 9.967786093208457e-06, "loss": 0.0032, "step": 32400 }, { "epoch": 0.5303117074367995, "grad_norm": 0.2469356209039688, "learning_rate": 9.967705129134686e-06, "loss": 0.0029, "step": 32410 }, { "epoch": 0.5304753333878753, "grad_norm": 0.09275887906551361, "learning_rate": 9.967624063773634e-06, "loss": 0.004, "step": 32420 }, { "epoch": 0.5306389593389511, "grad_norm": 0.1345721185207367, "learning_rate": 9.967542897126949e-06, "loss": 0.0023, "step": 32430 }, { "epoch": 0.530802585290027, "grad_norm": 0.05243934690952301, "learning_rate": 9.967461629196291e-06, "loss": 0.0042, "step": 32440 }, { "epoch": 0.5309662112411029, "grad_norm": 0.12230829894542694, "learning_rate": 9.967380259983315e-06, "loss": 0.0042, "step": 32450 }, { "epoch": 0.5311298371921787, "grad_norm": 0.06251143664121628, "learning_rate": 9.96729878948968e-06, "loss": 0.0039, "step": 32460 }, { "epoch": 0.5312934631432545, "grad_norm": 0.2818416357040405, "learning_rate": 9.967217217717047e-06, "loss": 0.0047, "step": 32470 }, { "epoch": 0.5314570890943303, "grad_norm": 0.1809210628271103, "learning_rate": 9.96713554466708e-06, "loss": 0.0054, "step": 32480 }, { "epoch": 0.5316207150454062, "grad_norm": 0.08285737782716751, "learning_rate": 9.967053770341442e-06, "loss": 0.0032, "step": 32490 }, { "epoch": 0.5317843409964821, "grad_norm": 0.1592283993959427, "learning_rate": 9.966971894741804e-06, "loss": 0.0047, "step": 32500 }, { "epoch": 0.5319479669475579, "grad_norm": 0.07124653458595276, "learning_rate": 9.966889917869833e-06, "loss": 0.0029, "step": 32510 }, { "epoch": 0.5321115928986337, "grad_norm": 0.14599202573299408, "learning_rate": 9.9668078397272e-06, "loss": 0.0046, "step": 32520 }, { "epoch": 0.5322752188497095, "grad_norm": 0.22009068727493286, "learning_rate": 9.966725660315578e-06, "loss": 0.0046, "step": 32530 }, { "epoch": 0.5324388448007854, "grad_norm": 0.4525461792945862, "learning_rate": 9.966643379636646e-06, "loss": 0.0042, "step": 32540 }, { "epoch": 0.5326024707518613, "grad_norm": 0.03433551266789436, "learning_rate": 9.966560997692078e-06, "loss": 0.0043, "step": 32550 }, { "epoch": 0.5327660967029371, "grad_norm": 0.08348521590232849, "learning_rate": 9.966478514483557e-06, "loss": 0.0037, "step": 32560 }, { "epoch": 0.532929722654013, "grad_norm": 0.06387443840503693, "learning_rate": 9.966395930012762e-06, "loss": 0.0024, "step": 32570 }, { "epoch": 0.5330933486050887, "grad_norm": 0.2717062532901764, "learning_rate": 9.966313244281377e-06, "loss": 0.0031, "step": 32580 }, { "epoch": 0.5332569745561646, "grad_norm": 0.11144217103719711, "learning_rate": 9.966230457291089e-06, "loss": 0.0026, "step": 32590 }, { "epoch": 0.5334206005072405, "grad_norm": 0.08930373936891556, "learning_rate": 9.966147569043584e-06, "loss": 0.0024, "step": 32600 }, { "epoch": 0.5335842264583163, "grad_norm": 0.1380859762430191, "learning_rate": 9.966064579540558e-06, "loss": 0.0032, "step": 32610 }, { "epoch": 0.5337478524093922, "grad_norm": 0.22643807530403137, "learning_rate": 9.965981488783697e-06, "loss": 0.0035, "step": 32620 }, { "epoch": 0.5339114783604679, "grad_norm": 0.09475122392177582, "learning_rate": 9.965898296774695e-06, "loss": 0.0031, "step": 32630 }, { "epoch": 0.5340751043115438, "grad_norm": 0.02509239688515663, "learning_rate": 9.96581500351525e-06, "loss": 0.0022, "step": 32640 }, { "epoch": 0.5342387302626197, "grad_norm": 0.08504031598567963, "learning_rate": 9.965731609007062e-06, "loss": 0.0023, "step": 32650 }, { "epoch": 0.5344023562136955, "grad_norm": 0.14511090517044067, "learning_rate": 9.965648113251828e-06, "loss": 0.0025, "step": 32660 }, { "epoch": 0.5345659821647714, "grad_norm": 0.09983053803443909, "learning_rate": 9.965564516251253e-06, "loss": 0.002, "step": 32670 }, { "epoch": 0.5347296081158471, "grad_norm": 0.12329120188951492, "learning_rate": 9.965480818007042e-06, "loss": 0.0025, "step": 32680 }, { "epoch": 0.534893234066923, "grad_norm": 0.10277808457612991, "learning_rate": 9.965397018520898e-06, "loss": 0.0026, "step": 32690 }, { "epoch": 0.5350568600179989, "grad_norm": 0.14021733403205872, "learning_rate": 9.965313117794532e-06, "loss": 0.0025, "step": 32700 }, { "epoch": 0.5352204859690747, "grad_norm": 0.08235494047403336, "learning_rate": 9.965229115829655e-06, "loss": 0.0022, "step": 32710 }, { "epoch": 0.5353841119201506, "grad_norm": 0.11372055113315582, "learning_rate": 9.96514501262798e-06, "loss": 0.0028, "step": 32720 }, { "epoch": 0.5355477378712263, "grad_norm": 0.13262084126472473, "learning_rate": 9.965060808191221e-06, "loss": 0.0036, "step": 32730 }, { "epoch": 0.5357113638223022, "grad_norm": 0.1994280219078064, "learning_rate": 9.964976502521093e-06, "loss": 0.0033, "step": 32740 }, { "epoch": 0.5358749897733781, "grad_norm": 0.06389732658863068, "learning_rate": 9.964892095619318e-06, "loss": 0.0035, "step": 32750 }, { "epoch": 0.5360386157244539, "grad_norm": 0.0815129280090332, "learning_rate": 9.964807587487614e-06, "loss": 0.0046, "step": 32760 }, { "epoch": 0.5362022416755298, "grad_norm": 0.07941795140504837, "learning_rate": 9.964722978127708e-06, "loss": 0.0032, "step": 32770 }, { "epoch": 0.5363658676266055, "grad_norm": 0.11588133126497269, "learning_rate": 9.964638267541321e-06, "loss": 0.0091, "step": 32780 }, { "epoch": 0.5365294935776814, "grad_norm": 0.12795761227607727, "learning_rate": 9.964553455730183e-06, "loss": 0.0039, "step": 32790 }, { "epoch": 0.5366931195287573, "grad_norm": 0.08228202909231186, "learning_rate": 9.964468542696022e-06, "loss": 0.0032, "step": 32800 }, { "epoch": 0.5368567454798331, "grad_norm": 0.15834614634513855, "learning_rate": 9.964383528440568e-06, "loss": 0.0057, "step": 32810 }, { "epoch": 0.537020371430909, "grad_norm": 0.08455131947994232, "learning_rate": 9.964298412965558e-06, "loss": 0.003, "step": 32820 }, { "epoch": 0.5371839973819847, "grad_norm": 0.19896654784679413, "learning_rate": 9.964213196272725e-06, "loss": 0.0031, "step": 32830 }, { "epoch": 0.5373476233330606, "grad_norm": 0.056586068123579025, "learning_rate": 9.964127878363805e-06, "loss": 0.0032, "step": 32840 }, { "epoch": 0.5375112492841365, "grad_norm": 0.03857789188623428, "learning_rate": 9.964042459240542e-06, "loss": 0.0025, "step": 32850 }, { "epoch": 0.5376748752352123, "grad_norm": 0.045681972056627274, "learning_rate": 9.963956938904674e-06, "loss": 0.0022, "step": 32860 }, { "epoch": 0.5378385011862882, "grad_norm": 0.15979351103305817, "learning_rate": 9.963871317357946e-06, "loss": 0.0044, "step": 32870 }, { "epoch": 0.538002127137364, "grad_norm": 0.18461719155311584, "learning_rate": 9.963785594602103e-06, "loss": 0.0034, "step": 32880 }, { "epoch": 0.5381657530884398, "grad_norm": 0.06537924706935883, "learning_rate": 9.963699770638893e-06, "loss": 0.003, "step": 32890 }, { "epoch": 0.5383293790395157, "grad_norm": 0.08357591927051544, "learning_rate": 9.963613845470066e-06, "loss": 0.0045, "step": 32900 }, { "epoch": 0.5384930049905915, "grad_norm": 0.13711190223693848, "learning_rate": 9.963527819097376e-06, "loss": 0.0019, "step": 32910 }, { "epoch": 0.5386566309416674, "grad_norm": 0.22606053948402405, "learning_rate": 9.963441691522573e-06, "loss": 0.0031, "step": 32920 }, { "epoch": 0.5388202568927432, "grad_norm": 0.04841788113117218, "learning_rate": 9.963355462747417e-06, "loss": 0.0049, "step": 32930 }, { "epoch": 0.538983882843819, "grad_norm": 0.06703105568885803, "learning_rate": 9.963269132773661e-06, "loss": 0.0045, "step": 32940 }, { "epoch": 0.5391475087948949, "grad_norm": 0.14072708785533905, "learning_rate": 9.963182701603071e-06, "loss": 0.0024, "step": 32950 }, { "epoch": 0.5393111347459707, "grad_norm": 0.07079524546861649, "learning_rate": 9.963096169237407e-06, "loss": 0.0024, "step": 32960 }, { "epoch": 0.5394747606970466, "grad_norm": 0.11690954864025116, "learning_rate": 9.963009535678432e-06, "loss": 0.0041, "step": 32970 }, { "epoch": 0.5396383866481224, "grad_norm": 0.17085812985897064, "learning_rate": 9.962922800927915e-06, "loss": 0.0029, "step": 32980 }, { "epoch": 0.5398020125991982, "grad_norm": 0.12784487009048462, "learning_rate": 9.962835964987621e-06, "loss": 0.0027, "step": 32990 }, { "epoch": 0.539965638550274, "grad_norm": 0.23226426541805267, "learning_rate": 9.962749027859325e-06, "loss": 0.0032, "step": 33000 }, { "epoch": 0.5401292645013499, "grad_norm": 0.4497329890727997, "learning_rate": 9.962661989544794e-06, "loss": 0.0047, "step": 33010 }, { "epoch": 0.5402928904524258, "grad_norm": 0.0938999131321907, "learning_rate": 9.962574850045807e-06, "loss": 0.0027, "step": 33020 }, { "epoch": 0.5404565164035016, "grad_norm": 0.08476103097200394, "learning_rate": 9.962487609364138e-06, "loss": 0.0045, "step": 33030 }, { "epoch": 0.5406201423545774, "grad_norm": 0.1042250245809555, "learning_rate": 9.962400267501568e-06, "loss": 0.0035, "step": 33040 }, { "epoch": 0.5407837683056532, "grad_norm": 0.08511903136968613, "learning_rate": 9.962312824459879e-06, "loss": 0.0025, "step": 33050 }, { "epoch": 0.5409473942567291, "grad_norm": 0.10290273278951645, "learning_rate": 9.96222528024085e-06, "loss": 0.0045, "step": 33060 }, { "epoch": 0.541111020207805, "grad_norm": 0.037532366812229156, "learning_rate": 9.962137634846268e-06, "loss": 0.0023, "step": 33070 }, { "epoch": 0.5412746461588808, "grad_norm": 0.05857984721660614, "learning_rate": 9.962049888277918e-06, "loss": 0.0039, "step": 33080 }, { "epoch": 0.5414382721099567, "grad_norm": 0.0975385308265686, "learning_rate": 9.961962040537594e-06, "loss": 0.0028, "step": 33090 }, { "epoch": 0.5416018980610324, "grad_norm": 0.17210792005062103, "learning_rate": 9.961874091627082e-06, "loss": 0.0035, "step": 33100 }, { "epoch": 0.5417655240121083, "grad_norm": 0.11422216892242432, "learning_rate": 9.961786041548177e-06, "loss": 0.0036, "step": 33110 }, { "epoch": 0.5419291499631842, "grad_norm": 0.06392106413841248, "learning_rate": 9.961697890302675e-06, "loss": 0.004, "step": 33120 }, { "epoch": 0.54209277591426, "grad_norm": 0.07996931672096252, "learning_rate": 9.961609637892374e-06, "loss": 0.0038, "step": 33130 }, { "epoch": 0.5422564018653359, "grad_norm": 0.07083261013031006, "learning_rate": 9.96152128431907e-06, "loss": 0.0036, "step": 33140 }, { "epoch": 0.5424200278164116, "grad_norm": 0.25559183955192566, "learning_rate": 9.961432829584567e-06, "loss": 0.003, "step": 33150 }, { "epoch": 0.5425836537674875, "grad_norm": 0.15613792836666107, "learning_rate": 9.96134427369067e-06, "loss": 0.0055, "step": 33160 }, { "epoch": 0.5427472797185634, "grad_norm": 0.2325993925333023, "learning_rate": 9.96125561663918e-06, "loss": 0.0036, "step": 33170 }, { "epoch": 0.5429109056696392, "grad_norm": 0.05904138833284378, "learning_rate": 9.96116685843191e-06, "loss": 0.0032, "step": 33180 }, { "epoch": 0.5430745316207151, "grad_norm": 0.0766424685716629, "learning_rate": 9.961077999070666e-06, "loss": 0.0039, "step": 33190 }, { "epoch": 0.5432381575717908, "grad_norm": 0.0728909894824028, "learning_rate": 9.96098903855726e-06, "loss": 0.003, "step": 33200 }, { "epoch": 0.5434017835228667, "grad_norm": 0.048065684735774994, "learning_rate": 9.960899976893508e-06, "loss": 0.003, "step": 33210 }, { "epoch": 0.5435654094739426, "grad_norm": 0.14613692462444305, "learning_rate": 9.960810814081225e-06, "loss": 0.0035, "step": 33220 }, { "epoch": 0.5437290354250184, "grad_norm": 0.02359084226191044, "learning_rate": 9.960721550122226e-06, "loss": 0.0014, "step": 33230 }, { "epoch": 0.5438926613760943, "grad_norm": 0.15326407551765442, "learning_rate": 9.960632185018335e-06, "loss": 0.0094, "step": 33240 }, { "epoch": 0.54405628732717, "grad_norm": 0.19867601990699768, "learning_rate": 9.960542718771373e-06, "loss": 0.0031, "step": 33250 }, { "epoch": 0.5442199132782459, "grad_norm": 0.05074797943234444, "learning_rate": 9.960453151383164e-06, "loss": 0.0039, "step": 33260 }, { "epoch": 0.5443835392293218, "grad_norm": 0.11424938589334488, "learning_rate": 9.960363482855534e-06, "loss": 0.0028, "step": 33270 }, { "epoch": 0.5445471651803976, "grad_norm": 0.2020004689693451, "learning_rate": 9.96027371319031e-06, "loss": 0.002, "step": 33280 }, { "epoch": 0.5447107911314735, "grad_norm": 0.09233748912811279, "learning_rate": 9.960183842389323e-06, "loss": 0.0036, "step": 33290 }, { "epoch": 0.5448744170825492, "grad_norm": 0.21593010425567627, "learning_rate": 9.960093870454408e-06, "loss": 0.0034, "step": 33300 }, { "epoch": 0.5450380430336251, "grad_norm": 0.02988452836871147, "learning_rate": 9.960003797387397e-06, "loss": 0.0049, "step": 33310 }, { "epoch": 0.545201668984701, "grad_norm": 0.0933699831366539, "learning_rate": 9.959913623190127e-06, "loss": 0.0023, "step": 33320 }, { "epoch": 0.5453652949357768, "grad_norm": 0.3159290552139282, "learning_rate": 9.959823347864434e-06, "loss": 0.0035, "step": 33330 }, { "epoch": 0.5455289208868527, "grad_norm": 0.03786079213023186, "learning_rate": 9.959732971412165e-06, "loss": 0.0059, "step": 33340 }, { "epoch": 0.5456925468379284, "grad_norm": 0.1381710022687912, "learning_rate": 9.959642493835155e-06, "loss": 0.0032, "step": 33350 }, { "epoch": 0.5458561727890043, "grad_norm": 0.05632227659225464, "learning_rate": 9.959551915135255e-06, "loss": 0.0016, "step": 33360 }, { "epoch": 0.5460197987400802, "grad_norm": 0.16760320961475372, "learning_rate": 9.95946123531431e-06, "loss": 0.004, "step": 33370 }, { "epoch": 0.546183424691156, "grad_norm": 0.1086471825838089, "learning_rate": 9.959370454374166e-06, "loss": 0.004, "step": 33380 }, { "epoch": 0.5463470506422319, "grad_norm": 0.09788025915622711, "learning_rate": 9.959279572316678e-06, "loss": 0.0036, "step": 33390 }, { "epoch": 0.5465106765933077, "grad_norm": 0.07638402283191681, "learning_rate": 9.959188589143695e-06, "loss": 0.0042, "step": 33400 }, { "epoch": 0.5466743025443835, "grad_norm": 0.08575642853975296, "learning_rate": 9.959097504857077e-06, "loss": 0.0052, "step": 33410 }, { "epoch": 0.5468379284954594, "grad_norm": 0.06756045669317245, "learning_rate": 9.959006319458676e-06, "loss": 0.0028, "step": 33420 }, { "epoch": 0.5470015544465352, "grad_norm": 0.15578626096248627, "learning_rate": 9.958915032950356e-06, "loss": 0.0029, "step": 33430 }, { "epoch": 0.5471651803976111, "grad_norm": 0.16509877145290375, "learning_rate": 9.958823645333975e-06, "loss": 0.0035, "step": 33440 }, { "epoch": 0.5473288063486869, "grad_norm": 0.3733545243740082, "learning_rate": 9.958732156611397e-06, "loss": 0.0029, "step": 33450 }, { "epoch": 0.5474924322997627, "grad_norm": 0.09957145899534225, "learning_rate": 9.958640566784488e-06, "loss": 0.0021, "step": 33460 }, { "epoch": 0.5476560582508386, "grad_norm": 0.0721987709403038, "learning_rate": 9.958548875855114e-06, "loss": 0.0023, "step": 33470 }, { "epoch": 0.5478196842019144, "grad_norm": 0.07883435487747192, "learning_rate": 9.958457083825147e-06, "loss": 0.0028, "step": 33480 }, { "epoch": 0.5479833101529903, "grad_norm": 0.6021624207496643, "learning_rate": 9.958365190696456e-06, "loss": 0.0036, "step": 33490 }, { "epoch": 0.5481469361040661, "grad_norm": 0.14627084136009216, "learning_rate": 9.958273196470915e-06, "loss": 0.0035, "step": 33500 }, { "epoch": 0.548310562055142, "grad_norm": 0.09402304142713547, "learning_rate": 9.958181101150401e-06, "loss": 0.003, "step": 33510 }, { "epoch": 0.5484741880062178, "grad_norm": 0.09519172459840775, "learning_rate": 9.958088904736793e-06, "loss": 0.0023, "step": 33520 }, { "epoch": 0.5486378139572936, "grad_norm": 0.08432353287935257, "learning_rate": 9.957996607231968e-06, "loss": 0.0062, "step": 33530 }, { "epoch": 0.5488014399083695, "grad_norm": 0.12092642486095428, "learning_rate": 9.957904208637807e-06, "loss": 0.0019, "step": 33540 }, { "epoch": 0.5489650658594453, "grad_norm": 0.13545246422290802, "learning_rate": 9.957811708956197e-06, "loss": 0.0041, "step": 33550 }, { "epoch": 0.5491286918105212, "grad_norm": 0.24060407280921936, "learning_rate": 9.957719108189023e-06, "loss": 0.0027, "step": 33560 }, { "epoch": 0.549292317761597, "grad_norm": 0.05800843983888626, "learning_rate": 9.957626406338172e-06, "loss": 0.0031, "step": 33570 }, { "epoch": 0.5494559437126728, "grad_norm": 0.14253117144107819, "learning_rate": 9.957533603405536e-06, "loss": 0.0045, "step": 33580 }, { "epoch": 0.5496195696637487, "grad_norm": 0.22844666242599487, "learning_rate": 9.957440699393006e-06, "loss": 0.0034, "step": 33590 }, { "epoch": 0.5497831956148245, "grad_norm": 0.14066684246063232, "learning_rate": 9.957347694302475e-06, "loss": 0.0027, "step": 33600 }, { "epoch": 0.5499468215659004, "grad_norm": 0.049563225358724594, "learning_rate": 9.95725458813584e-06, "loss": 0.0034, "step": 33610 }, { "epoch": 0.5501104475169762, "grad_norm": 0.01803707890212536, "learning_rate": 9.957161380895002e-06, "loss": 0.0021, "step": 33620 }, { "epoch": 0.550274073468052, "grad_norm": 0.0664411261677742, "learning_rate": 9.95706807258186e-06, "loss": 0.0042, "step": 33630 }, { "epoch": 0.5504376994191279, "grad_norm": 0.024293283000588417, "learning_rate": 9.956974663198314e-06, "loss": 0.0025, "step": 33640 }, { "epoch": 0.5506013253702037, "grad_norm": 0.09425840526819229, "learning_rate": 9.956881152746269e-06, "loss": 0.0038, "step": 33650 }, { "epoch": 0.5507649513212796, "grad_norm": 0.05422205105423927, "learning_rate": 9.956787541227635e-06, "loss": 0.0031, "step": 33660 }, { "epoch": 0.5509285772723554, "grad_norm": 0.11054264008998871, "learning_rate": 9.956693828644318e-06, "loss": 0.0022, "step": 33670 }, { "epoch": 0.5510922032234312, "grad_norm": 0.12063618004322052, "learning_rate": 9.95660001499823e-06, "loss": 0.003, "step": 33680 }, { "epoch": 0.5512558291745071, "grad_norm": 0.09101159870624542, "learning_rate": 9.956506100291283e-06, "loss": 0.003, "step": 33690 }, { "epoch": 0.5514194551255829, "grad_norm": 0.06721699237823486, "learning_rate": 9.956412084525392e-06, "loss": 0.0038, "step": 33700 }, { "epoch": 0.5515830810766588, "grad_norm": 0.034821514040231705, "learning_rate": 9.956317967702474e-06, "loss": 0.0017, "step": 33710 }, { "epoch": 0.5517467070277347, "grad_norm": 0.15465065836906433, "learning_rate": 9.956223749824447e-06, "loss": 0.0026, "step": 33720 }, { "epoch": 0.5519103329788104, "grad_norm": 0.11142609268426895, "learning_rate": 9.956129430893233e-06, "loss": 0.0028, "step": 33730 }, { "epoch": 0.5520739589298863, "grad_norm": 0.24316005408763885, "learning_rate": 9.956035010910757e-06, "loss": 0.0048, "step": 33740 }, { "epoch": 0.5522375848809621, "grad_norm": 0.007888305932283401, "learning_rate": 9.955940489878939e-06, "loss": 0.0019, "step": 33750 }, { "epoch": 0.552401210832038, "grad_norm": 0.1077987402677536, "learning_rate": 9.95584586779971e-06, "loss": 0.0034, "step": 33760 }, { "epoch": 0.5525648367831139, "grad_norm": 0.05376173183321953, "learning_rate": 9.955751144674999e-06, "loss": 0.005, "step": 33770 }, { "epoch": 0.5527284627341896, "grad_norm": 0.052521031349897385, "learning_rate": 9.955656320506738e-06, "loss": 0.0021, "step": 33780 }, { "epoch": 0.5528920886852655, "grad_norm": 0.046039313077926636, "learning_rate": 9.955561395296858e-06, "loss": 0.0022, "step": 33790 }, { "epoch": 0.5530557146363413, "grad_norm": 0.19020861387252808, "learning_rate": 9.955466369047297e-06, "loss": 0.0038, "step": 33800 }, { "epoch": 0.5532193405874172, "grad_norm": 0.06185857206583023, "learning_rate": 9.95537124175999e-06, "loss": 0.0036, "step": 33810 }, { "epoch": 0.5533829665384931, "grad_norm": 0.1536354422569275, "learning_rate": 9.955276013436877e-06, "loss": 0.0039, "step": 33820 }, { "epoch": 0.5535465924895688, "grad_norm": 0.11529579758644104, "learning_rate": 9.955180684079901e-06, "loss": 0.0023, "step": 33830 }, { "epoch": 0.5537102184406447, "grad_norm": 0.06778623163700104, "learning_rate": 9.955085253691006e-06, "loss": 0.0023, "step": 33840 }, { "epoch": 0.5538738443917205, "grad_norm": 0.28956931829452515, "learning_rate": 9.954989722272137e-06, "loss": 0.005, "step": 33850 }, { "epoch": 0.5540374703427964, "grad_norm": 0.04703707620501518, "learning_rate": 9.95489408982524e-06, "loss": 0.0033, "step": 33860 }, { "epoch": 0.5542010962938722, "grad_norm": 0.0974058210849762, "learning_rate": 9.95479835635227e-06, "loss": 0.0033, "step": 33870 }, { "epoch": 0.554364722244948, "grad_norm": 0.17017142474651337, "learning_rate": 9.954702521855171e-06, "loss": 0.0046, "step": 33880 }, { "epoch": 0.5545283481960239, "grad_norm": 0.043577831238508224, "learning_rate": 9.954606586335904e-06, "loss": 0.0028, "step": 33890 }, { "epoch": 0.5546919741470997, "grad_norm": 0.15279719233512878, "learning_rate": 9.954510549796421e-06, "loss": 0.0023, "step": 33900 }, { "epoch": 0.5548556000981756, "grad_norm": 0.7441000938415527, "learning_rate": 9.954414412238681e-06, "loss": 0.003, "step": 33910 }, { "epoch": 0.5550192260492514, "grad_norm": 0.4778856635093689, "learning_rate": 9.954318173664648e-06, "loss": 0.0026, "step": 33920 }, { "epoch": 0.5551828520003272, "grad_norm": 0.2715802788734436, "learning_rate": 9.954221834076278e-06, "loss": 0.0038, "step": 33930 }, { "epoch": 0.5553464779514031, "grad_norm": 0.2194146066904068, "learning_rate": 9.95412539347554e-06, "loss": 0.0041, "step": 33940 }, { "epoch": 0.5555101039024789, "grad_norm": 0.039383240044116974, "learning_rate": 9.954028851864398e-06, "loss": 0.0051, "step": 33950 }, { "epoch": 0.5556737298535548, "grad_norm": 0.21796464920043945, "learning_rate": 9.95393220924482e-06, "loss": 0.0038, "step": 33960 }, { "epoch": 0.5558373558046306, "grad_norm": 0.014923643320798874, "learning_rate": 9.953835465618777e-06, "loss": 0.0039, "step": 33970 }, { "epoch": 0.5560009817557064, "grad_norm": 0.060972753912210464, "learning_rate": 9.953738620988244e-06, "loss": 0.0027, "step": 33980 }, { "epoch": 0.5561646077067823, "grad_norm": 0.1582566648721695, "learning_rate": 9.953641675355193e-06, "loss": 0.004, "step": 33990 }, { "epoch": 0.5563282336578581, "grad_norm": 0.02304948680102825, "learning_rate": 9.9535446287216e-06, "loss": 0.0033, "step": 34000 }, { "epoch": 0.556491859608934, "grad_norm": 0.04167890548706055, "learning_rate": 9.953447481089446e-06, "loss": 0.0028, "step": 34010 }, { "epoch": 0.5566554855600098, "grad_norm": 0.06242075935006142, "learning_rate": 9.95335023246071e-06, "loss": 0.0028, "step": 34020 }, { "epoch": 0.5568191115110857, "grad_norm": 0.06653702259063721, "learning_rate": 9.953252882837376e-06, "loss": 0.0038, "step": 34030 }, { "epoch": 0.5569827374621615, "grad_norm": 0.07954331487417221, "learning_rate": 9.953155432221428e-06, "loss": 0.0029, "step": 34040 }, { "epoch": 0.5571463634132373, "grad_norm": 0.1751769632101059, "learning_rate": 9.953057880614853e-06, "loss": 0.0038, "step": 34050 }, { "epoch": 0.5573099893643132, "grad_norm": 0.12834465503692627, "learning_rate": 9.95296022801964e-06, "loss": 0.0026, "step": 34060 }, { "epoch": 0.557473615315389, "grad_norm": 0.08995863795280457, "learning_rate": 9.95286247443778e-06, "loss": 0.0026, "step": 34070 }, { "epoch": 0.5576372412664649, "grad_norm": 0.10422423481941223, "learning_rate": 9.95276461987127e-06, "loss": 0.0042, "step": 34080 }, { "epoch": 0.5578008672175407, "grad_norm": 0.18861302733421326, "learning_rate": 9.952666664322097e-06, "loss": 0.0043, "step": 34090 }, { "epoch": 0.5579644931686165, "grad_norm": 0.1867142617702484, "learning_rate": 9.952568607792265e-06, "loss": 0.0041, "step": 34100 }, { "epoch": 0.5581281191196924, "grad_norm": 0.04031140357255936, "learning_rate": 9.95247045028377e-06, "loss": 0.0037, "step": 34110 }, { "epoch": 0.5582917450707682, "grad_norm": 0.04537266120314598, "learning_rate": 9.952372191798615e-06, "loss": 0.0039, "step": 34120 }, { "epoch": 0.5584553710218441, "grad_norm": 0.06030448153614998, "learning_rate": 9.952273832338803e-06, "loss": 0.0021, "step": 34130 }, { "epoch": 0.55861899697292, "grad_norm": 0.24142535030841827, "learning_rate": 9.952175371906339e-06, "loss": 0.0046, "step": 34140 }, { "epoch": 0.5587826229239957, "grad_norm": 0.05431438237428665, "learning_rate": 9.95207681050323e-06, "loss": 0.0024, "step": 34150 }, { "epoch": 0.5589462488750716, "grad_norm": 0.04971565678715706, "learning_rate": 9.951978148131489e-06, "loss": 0.0023, "step": 34160 }, { "epoch": 0.5591098748261474, "grad_norm": 0.04999755322933197, "learning_rate": 9.951879384793124e-06, "loss": 0.0023, "step": 34170 }, { "epoch": 0.5592735007772233, "grad_norm": 0.23727037012577057, "learning_rate": 9.951780520490149e-06, "loss": 0.0039, "step": 34180 }, { "epoch": 0.5594371267282992, "grad_norm": 0.01600191928446293, "learning_rate": 9.95168155522458e-06, "loss": 0.003, "step": 34190 }, { "epoch": 0.5596007526793749, "grad_norm": 0.04376046359539032, "learning_rate": 9.951582488998436e-06, "loss": 0.0038, "step": 34200 }, { "epoch": 0.5597643786304508, "grad_norm": 0.086293525993824, "learning_rate": 9.951483321813737e-06, "loss": 0.0033, "step": 34210 }, { "epoch": 0.5599280045815266, "grad_norm": 0.04714641720056534, "learning_rate": 9.951384053672504e-06, "loss": 0.0027, "step": 34220 }, { "epoch": 0.5600916305326025, "grad_norm": 0.09308035671710968, "learning_rate": 9.951284684576762e-06, "loss": 0.0024, "step": 34230 }, { "epoch": 0.5602552564836784, "grad_norm": 0.2594463527202606, "learning_rate": 9.951185214528534e-06, "loss": 0.004, "step": 34240 }, { "epoch": 0.5604188824347541, "grad_norm": 0.03100936859846115, "learning_rate": 9.95108564352985e-06, "loss": 0.0019, "step": 34250 }, { "epoch": 0.56058250838583, "grad_norm": 0.18479646742343903, "learning_rate": 9.950985971582743e-06, "loss": 0.0027, "step": 34260 }, { "epoch": 0.5607461343369058, "grad_norm": 0.6609426736831665, "learning_rate": 9.95088619868924e-06, "loss": 0.0043, "step": 34270 }, { "epoch": 0.5609097602879817, "grad_norm": 0.16748012602329254, "learning_rate": 9.95078632485138e-06, "loss": 0.0027, "step": 34280 }, { "epoch": 0.5610733862390576, "grad_norm": 0.1738748848438263, "learning_rate": 9.950686350071197e-06, "loss": 0.0047, "step": 34290 }, { "epoch": 0.5612370121901333, "grad_norm": 0.18086744844913483, "learning_rate": 9.950586274350728e-06, "loss": 0.0042, "step": 34300 }, { "epoch": 0.5614006381412092, "grad_norm": 0.13787643611431122, "learning_rate": 9.950486097692019e-06, "loss": 0.0031, "step": 34310 }, { "epoch": 0.561564264092285, "grad_norm": 0.1542452871799469, "learning_rate": 9.950385820097104e-06, "loss": 0.0043, "step": 34320 }, { "epoch": 0.5617278900433609, "grad_norm": 0.2638024687767029, "learning_rate": 9.950285441568036e-06, "loss": 0.0033, "step": 34330 }, { "epoch": 0.5618915159944368, "grad_norm": 0.14743083715438843, "learning_rate": 9.950184962106855e-06, "loss": 0.0048, "step": 34340 }, { "epoch": 0.5620551419455125, "grad_norm": 0.13663847744464874, "learning_rate": 9.950084381715614e-06, "loss": 0.0041, "step": 34350 }, { "epoch": 0.5622187678965884, "grad_norm": 0.24376371502876282, "learning_rate": 9.949983700396363e-06, "loss": 0.0032, "step": 34360 }, { "epoch": 0.5623823938476642, "grad_norm": 0.062250591814517975, "learning_rate": 9.949882918151155e-06, "loss": 0.0016, "step": 34370 }, { "epoch": 0.5625460197987401, "grad_norm": 0.16325074434280396, "learning_rate": 9.949782034982042e-06, "loss": 0.0043, "step": 34380 }, { "epoch": 0.562709645749816, "grad_norm": 0.14869815111160278, "learning_rate": 9.949681050891083e-06, "loss": 0.0058, "step": 34390 }, { "epoch": 0.5628732717008917, "grad_norm": 0.070064976811409, "learning_rate": 9.949579965880338e-06, "loss": 0.002, "step": 34400 }, { "epoch": 0.5630368976519676, "grad_norm": 0.12701670825481415, "learning_rate": 9.949478779951866e-06, "loss": 0.0019, "step": 34410 }, { "epoch": 0.5632005236030434, "grad_norm": 0.10213713347911835, "learning_rate": 9.949377493107732e-06, "loss": 0.0034, "step": 34420 }, { "epoch": 0.5633641495541193, "grad_norm": 0.15470647811889648, "learning_rate": 9.94927610535e-06, "loss": 0.0034, "step": 34430 }, { "epoch": 0.5635277755051952, "grad_norm": 0.26735919713974, "learning_rate": 9.949174616680736e-06, "loss": 0.003, "step": 34440 }, { "epoch": 0.563691401456271, "grad_norm": 0.05675894394516945, "learning_rate": 9.949073027102012e-06, "loss": 0.0026, "step": 34450 }, { "epoch": 0.5638550274073468, "grad_norm": 0.1670236736536026, "learning_rate": 9.948971336615897e-06, "loss": 0.0051, "step": 34460 }, { "epoch": 0.5640186533584226, "grad_norm": 0.10303352028131485, "learning_rate": 9.948869545224469e-06, "loss": 0.0041, "step": 34470 }, { "epoch": 0.5641822793094985, "grad_norm": 0.0850718691945076, "learning_rate": 9.948767652929796e-06, "loss": 0.0031, "step": 34480 }, { "epoch": 0.5643459052605744, "grad_norm": 0.14619317650794983, "learning_rate": 9.94866565973396e-06, "loss": 0.0044, "step": 34490 }, { "epoch": 0.5645095312116502, "grad_norm": 0.1330094188451767, "learning_rate": 9.948563565639041e-06, "loss": 0.004, "step": 34500 }, { "epoch": 0.564673157162726, "grad_norm": 0.09338099509477615, "learning_rate": 9.94846137064712e-06, "loss": 0.0021, "step": 34510 }, { "epoch": 0.5648367831138018, "grad_norm": 0.057484980672597885, "learning_rate": 9.948359074760277e-06, "loss": 0.0037, "step": 34520 }, { "epoch": 0.5650004090648777, "grad_norm": 0.13169962167739868, "learning_rate": 9.948256677980603e-06, "loss": 0.0043, "step": 34530 }, { "epoch": 0.5651640350159536, "grad_norm": 0.0541047602891922, "learning_rate": 9.948154180310184e-06, "loss": 0.0025, "step": 34540 }, { "epoch": 0.5653276609670294, "grad_norm": 0.02761739492416382, "learning_rate": 9.94805158175111e-06, "loss": 0.0025, "step": 34550 }, { "epoch": 0.5654912869181052, "grad_norm": 0.10916019231081009, "learning_rate": 9.94794888230547e-06, "loss": 0.0037, "step": 34560 }, { "epoch": 0.565654912869181, "grad_norm": 0.19711266458034515, "learning_rate": 9.947846081975363e-06, "loss": 0.0042, "step": 34570 }, { "epoch": 0.5658185388202569, "grad_norm": 0.14312979578971863, "learning_rate": 9.947743180762881e-06, "loss": 0.0034, "step": 34580 }, { "epoch": 0.5659821647713328, "grad_norm": 0.1706724911928177, "learning_rate": 9.947640178670123e-06, "loss": 0.0029, "step": 34590 }, { "epoch": 0.5661457907224086, "grad_norm": 0.2194891721010208, "learning_rate": 9.947537075699193e-06, "loss": 0.0037, "step": 34600 }, { "epoch": 0.5663094166734844, "grad_norm": 0.20066037774085999, "learning_rate": 9.947433871852186e-06, "loss": 0.0029, "step": 34610 }, { "epoch": 0.5664730426245602, "grad_norm": 0.060425326228141785, "learning_rate": 9.94733056713121e-06, "loss": 0.0067, "step": 34620 }, { "epoch": 0.5666366685756361, "grad_norm": 0.0407162569463253, "learning_rate": 9.947227161538372e-06, "loss": 0.0018, "step": 34630 }, { "epoch": 0.566800294526712, "grad_norm": 0.2065051645040512, "learning_rate": 9.94712365507578e-06, "loss": 0.0037, "step": 34640 }, { "epoch": 0.5669639204777878, "grad_norm": 0.03449328988790512, "learning_rate": 9.947020047745544e-06, "loss": 0.0026, "step": 34650 }, { "epoch": 0.5671275464288636, "grad_norm": 0.022027960047125816, "learning_rate": 9.946916339549777e-06, "loss": 0.0032, "step": 34660 }, { "epoch": 0.5672911723799394, "grad_norm": 0.15739288926124573, "learning_rate": 9.946812530490593e-06, "loss": 0.0044, "step": 34670 }, { "epoch": 0.5674547983310153, "grad_norm": 0.09893932193517685, "learning_rate": 9.946708620570108e-06, "loss": 0.0036, "step": 34680 }, { "epoch": 0.5676184242820912, "grad_norm": 0.09867346286773682, "learning_rate": 9.946604609790442e-06, "loss": 0.0031, "step": 34690 }, { "epoch": 0.567782050233167, "grad_norm": 0.10683468729257584, "learning_rate": 9.946500498153712e-06, "loss": 0.0035, "step": 34700 }, { "epoch": 0.5679456761842429, "grad_norm": 0.10989606380462646, "learning_rate": 9.946396285662047e-06, "loss": 0.0024, "step": 34710 }, { "epoch": 0.5681093021353186, "grad_norm": 0.10422082990407944, "learning_rate": 9.946291972317567e-06, "loss": 0.0025, "step": 34720 }, { "epoch": 0.5682729280863945, "grad_norm": 0.1076616421341896, "learning_rate": 9.946187558122402e-06, "loss": 0.0025, "step": 34730 }, { "epoch": 0.5684365540374703, "grad_norm": 0.07876507192850113, "learning_rate": 9.946083043078677e-06, "loss": 0.004, "step": 34740 }, { "epoch": 0.5686001799885462, "grad_norm": 0.14891023933887482, "learning_rate": 9.945978427188527e-06, "loss": 0.0024, "step": 34750 }, { "epoch": 0.5687638059396221, "grad_norm": 0.10869897156953812, "learning_rate": 9.945873710454084e-06, "loss": 0.0028, "step": 34760 }, { "epoch": 0.5689274318906978, "grad_norm": 0.058674875646829605, "learning_rate": 9.94576889287748e-06, "loss": 0.0044, "step": 34770 }, { "epoch": 0.5690910578417737, "grad_norm": 0.07079245150089264, "learning_rate": 9.945663974460856e-06, "loss": 0.0045, "step": 34780 }, { "epoch": 0.5692546837928495, "grad_norm": 0.08065475523471832, "learning_rate": 9.94555895520635e-06, "loss": 0.0049, "step": 34790 }, { "epoch": 0.5694183097439254, "grad_norm": 0.09464599937200546, "learning_rate": 9.945453835116101e-06, "loss": 0.004, "step": 34800 }, { "epoch": 0.5695819356950013, "grad_norm": 0.15773901343345642, "learning_rate": 9.945348614192256e-06, "loss": 0.0035, "step": 34810 }, { "epoch": 0.569745561646077, "grad_norm": 0.06196501478552818, "learning_rate": 9.945243292436958e-06, "loss": 0.0035, "step": 34820 }, { "epoch": 0.5699091875971529, "grad_norm": 0.18624082207679749, "learning_rate": 9.945137869852356e-06, "loss": 0.0039, "step": 34830 }, { "epoch": 0.5700728135482287, "grad_norm": 0.12569838762283325, "learning_rate": 9.945032346440597e-06, "loss": 0.0038, "step": 34840 }, { "epoch": 0.5702364394993046, "grad_norm": 0.08257696032524109, "learning_rate": 9.944926722203833e-06, "loss": 0.0025, "step": 34850 }, { "epoch": 0.5704000654503805, "grad_norm": 0.07869430631399155, "learning_rate": 9.944820997144222e-06, "loss": 0.0019, "step": 34860 }, { "epoch": 0.5705636914014562, "grad_norm": 0.2028905153274536, "learning_rate": 9.944715171263915e-06, "loss": 0.0058, "step": 34870 }, { "epoch": 0.5707273173525321, "grad_norm": 0.36887964606285095, "learning_rate": 9.94460924456507e-06, "loss": 0.0038, "step": 34880 }, { "epoch": 0.5708909433036079, "grad_norm": 0.138101264834404, "learning_rate": 9.944503217049847e-06, "loss": 0.0033, "step": 34890 }, { "epoch": 0.5710545692546838, "grad_norm": 0.16581174731254578, "learning_rate": 9.944397088720412e-06, "loss": 0.0053, "step": 34900 }, { "epoch": 0.5712181952057597, "grad_norm": 0.026357300579547882, "learning_rate": 9.944290859578923e-06, "loss": 0.0039, "step": 34910 }, { "epoch": 0.5713818211568354, "grad_norm": 0.0961824506521225, "learning_rate": 9.944184529627549e-06, "loss": 0.0024, "step": 34920 }, { "epoch": 0.5715454471079113, "grad_norm": 0.08350483328104019, "learning_rate": 9.944078098868457e-06, "loss": 0.0036, "step": 34930 }, { "epoch": 0.5717090730589871, "grad_norm": 0.30496495962142944, "learning_rate": 9.943971567303815e-06, "loss": 0.0055, "step": 34940 }, { "epoch": 0.571872699010063, "grad_norm": 0.18487006425857544, "learning_rate": 9.943864934935802e-06, "loss": 0.0029, "step": 34950 }, { "epoch": 0.5720363249611389, "grad_norm": 0.0545002818107605, "learning_rate": 9.943758201766585e-06, "loss": 0.0041, "step": 34960 }, { "epoch": 0.5721999509122146, "grad_norm": 0.36326920986175537, "learning_rate": 9.943651367798343e-06, "loss": 0.0042, "step": 34970 }, { "epoch": 0.5723635768632905, "grad_norm": 0.06574776023626328, "learning_rate": 9.943544433033254e-06, "loss": 0.0035, "step": 34980 }, { "epoch": 0.5725272028143663, "grad_norm": 0.08507231622934341, "learning_rate": 9.943437397473499e-06, "loss": 0.0029, "step": 34990 }, { "epoch": 0.5726908287654422, "grad_norm": 0.17854782938957214, "learning_rate": 9.94333026112126e-06, "loss": 0.0051, "step": 35000 }, { "epoch": 0.5728544547165181, "grad_norm": 0.04359865561127663, "learning_rate": 9.94322302397872e-06, "loss": 0.0028, "step": 35010 }, { "epoch": 0.5730180806675939, "grad_norm": 0.18073321878910065, "learning_rate": 9.943115686048067e-06, "loss": 0.0036, "step": 35020 }, { "epoch": 0.5731817066186697, "grad_norm": 0.003446343122050166, "learning_rate": 9.94300824733149e-06, "loss": 0.0023, "step": 35030 }, { "epoch": 0.5733453325697455, "grad_norm": 0.1148553267121315, "learning_rate": 9.942900707831178e-06, "loss": 0.0022, "step": 35040 }, { "epoch": 0.5735089585208214, "grad_norm": 0.03779436647891998, "learning_rate": 9.942793067549324e-06, "loss": 0.003, "step": 35050 }, { "epoch": 0.5736725844718973, "grad_norm": 0.09162017703056335, "learning_rate": 9.942685326488122e-06, "loss": 0.0027, "step": 35060 }, { "epoch": 0.5738362104229731, "grad_norm": 0.1446458250284195, "learning_rate": 9.942577484649771e-06, "loss": 0.0039, "step": 35070 }, { "epoch": 0.5739998363740489, "grad_norm": 0.05705275014042854, "learning_rate": 9.942469542036468e-06, "loss": 0.0025, "step": 35080 }, { "epoch": 0.5741634623251247, "grad_norm": 0.1711822748184204, "learning_rate": 9.942361498650415e-06, "loss": 0.0041, "step": 35090 }, { "epoch": 0.5743270882762006, "grad_norm": 0.14566859602928162, "learning_rate": 9.942253354493816e-06, "loss": 0.0035, "step": 35100 }, { "epoch": 0.5744907142272765, "grad_norm": 0.07573870569467545, "learning_rate": 9.942145109568873e-06, "loss": 0.0032, "step": 35110 }, { "epoch": 0.5746543401783523, "grad_norm": 0.15446388721466064, "learning_rate": 9.942036763877794e-06, "loss": 0.0033, "step": 35120 }, { "epoch": 0.5748179661294281, "grad_norm": 0.12120422720909119, "learning_rate": 9.941928317422789e-06, "loss": 0.0026, "step": 35130 }, { "epoch": 0.5749815920805039, "grad_norm": 0.04920819401741028, "learning_rate": 9.941819770206067e-06, "loss": 0.003, "step": 35140 }, { "epoch": 0.5751452180315798, "grad_norm": 0.13556945323944092, "learning_rate": 9.941711122229844e-06, "loss": 0.0026, "step": 35150 }, { "epoch": 0.5753088439826557, "grad_norm": 0.04951773211359978, "learning_rate": 9.941602373496334e-06, "loss": 0.0028, "step": 35160 }, { "epoch": 0.5754724699337315, "grad_norm": 0.16583800315856934, "learning_rate": 9.941493524007755e-06, "loss": 0.0037, "step": 35170 }, { "epoch": 0.5756360958848074, "grad_norm": 0.1371152698993683, "learning_rate": 9.941384573766324e-06, "loss": 0.0019, "step": 35180 }, { "epoch": 0.5757997218358831, "grad_norm": 0.2509312927722931, "learning_rate": 9.941275522774265e-06, "loss": 0.0057, "step": 35190 }, { "epoch": 0.575963347786959, "grad_norm": 0.09644955396652222, "learning_rate": 9.9411663710338e-06, "loss": 0.0033, "step": 35200 }, { "epoch": 0.5761269737380349, "grad_norm": 0.14495238661766052, "learning_rate": 9.941057118547155e-06, "loss": 0.005, "step": 35210 }, { "epoch": 0.5762905996891107, "grad_norm": 0.14938023686408997, "learning_rate": 9.940947765316559e-06, "loss": 0.0036, "step": 35220 }, { "epoch": 0.5764542256401866, "grad_norm": 0.06916598975658417, "learning_rate": 9.940838311344238e-06, "loss": 0.0048, "step": 35230 }, { "epoch": 0.5766178515912623, "grad_norm": 0.15974310040473938, "learning_rate": 9.940728756632427e-06, "loss": 0.0027, "step": 35240 }, { "epoch": 0.5767814775423382, "grad_norm": 0.0466461218893528, "learning_rate": 9.94061910118336e-06, "loss": 0.003, "step": 35250 }, { "epoch": 0.5769451034934141, "grad_norm": 0.13920700550079346, "learning_rate": 9.94050934499927e-06, "loss": 0.0023, "step": 35260 }, { "epoch": 0.5771087294444899, "grad_norm": 0.1094503402709961, "learning_rate": 9.940399488082395e-06, "loss": 0.0019, "step": 35270 }, { "epoch": 0.5772723553955658, "grad_norm": 0.054353486746549606, "learning_rate": 9.940289530434978e-06, "loss": 0.0026, "step": 35280 }, { "epoch": 0.5774359813466415, "grad_norm": 0.011951545253396034, "learning_rate": 9.940179472059259e-06, "loss": 0.0026, "step": 35290 }, { "epoch": 0.5775996072977174, "grad_norm": 0.05541665852069855, "learning_rate": 9.940069312957481e-06, "loss": 0.0021, "step": 35300 }, { "epoch": 0.5777632332487933, "grad_norm": 0.09778507798910141, "learning_rate": 9.93995905313189e-06, "loss": 0.0031, "step": 35310 }, { "epoch": 0.5779268591998691, "grad_norm": 0.16715042293071747, "learning_rate": 9.939848692584737e-06, "loss": 0.0039, "step": 35320 }, { "epoch": 0.578090485150945, "grad_norm": 0.14543917775154114, "learning_rate": 9.93973823131827e-06, "loss": 0.0039, "step": 35330 }, { "epoch": 0.5782541111020207, "grad_norm": 0.11386677622795105, "learning_rate": 9.939627669334741e-06, "loss": 0.0047, "step": 35340 }, { "epoch": 0.5784177370530966, "grad_norm": 0.05537869408726692, "learning_rate": 9.939517006636405e-06, "loss": 0.004, "step": 35350 }, { "epoch": 0.5785813630041725, "grad_norm": 0.30286920070648193, "learning_rate": 9.93940624322552e-06, "loss": 0.0029, "step": 35360 }, { "epoch": 0.5787449889552483, "grad_norm": 0.26753318309783936, "learning_rate": 9.939295379104342e-06, "loss": 0.0035, "step": 35370 }, { "epoch": 0.5789086149063242, "grad_norm": 0.35843145847320557, "learning_rate": 9.93918441427513e-06, "loss": 0.0038, "step": 35380 }, { "epoch": 0.5790722408574, "grad_norm": 0.15335777401924133, "learning_rate": 9.93907334874015e-06, "loss": 0.0046, "step": 35390 }, { "epoch": 0.5792358668084758, "grad_norm": 0.14955832064151764, "learning_rate": 9.938962182501663e-06, "loss": 0.0061, "step": 35400 }, { "epoch": 0.5793994927595517, "grad_norm": 0.15107515454292297, "learning_rate": 9.93885091556194e-06, "loss": 0.0051, "step": 35410 }, { "epoch": 0.5795631187106275, "grad_norm": 0.05472130700945854, "learning_rate": 9.938739547923245e-06, "loss": 0.0032, "step": 35420 }, { "epoch": 0.5797267446617034, "grad_norm": 0.19933344423770905, "learning_rate": 9.938628079587853e-06, "loss": 0.0028, "step": 35430 }, { "epoch": 0.5798903706127791, "grad_norm": 0.20176301896572113, "learning_rate": 9.938516510558034e-06, "loss": 0.0045, "step": 35440 }, { "epoch": 0.580053996563855, "grad_norm": 0.22002927958965302, "learning_rate": 9.938404840836062e-06, "loss": 0.0036, "step": 35450 }, { "epoch": 0.5802176225149309, "grad_norm": 0.17419160902500153, "learning_rate": 9.938293070424217e-06, "loss": 0.0033, "step": 35460 }, { "epoch": 0.5803812484660067, "grad_norm": 0.05503339692950249, "learning_rate": 9.938181199324775e-06, "loss": 0.004, "step": 35470 }, { "epoch": 0.5805448744170826, "grad_norm": 0.14341700077056885, "learning_rate": 9.938069227540017e-06, "loss": 0.0034, "step": 35480 }, { "epoch": 0.5807085003681584, "grad_norm": 0.16850212216377258, "learning_rate": 9.937957155072228e-06, "loss": 0.003, "step": 35490 }, { "epoch": 0.5808721263192342, "grad_norm": 0.033911023288965225, "learning_rate": 9.937844981923695e-06, "loss": 0.0021, "step": 35500 }, { "epoch": 0.5810357522703101, "grad_norm": 0.013117808848619461, "learning_rate": 9.937732708096698e-06, "loss": 0.0022, "step": 35510 }, { "epoch": 0.5811993782213859, "grad_norm": 0.06335977464914322, "learning_rate": 9.937620333593534e-06, "loss": 0.0032, "step": 35520 }, { "epoch": 0.5813630041724618, "grad_norm": 0.04076981917023659, "learning_rate": 9.937507858416489e-06, "loss": 0.0041, "step": 35530 }, { "epoch": 0.5815266301235376, "grad_norm": 0.0725402906537056, "learning_rate": 9.937395282567859e-06, "loss": 0.0036, "step": 35540 }, { "epoch": 0.5816902560746134, "grad_norm": 0.01851271279156208, "learning_rate": 9.937282606049935e-06, "loss": 0.003, "step": 35550 }, { "epoch": 0.5818538820256893, "grad_norm": 0.07503094524145126, "learning_rate": 9.937169828865023e-06, "loss": 0.0018, "step": 35560 }, { "epoch": 0.5820175079767651, "grad_norm": 0.05131140351295471, "learning_rate": 9.937056951015413e-06, "loss": 0.0033, "step": 35570 }, { "epoch": 0.582181133927841, "grad_norm": 0.02958175167441368, "learning_rate": 9.936943972503412e-06, "loss": 0.0038, "step": 35580 }, { "epoch": 0.5823447598789168, "grad_norm": 0.10060955584049225, "learning_rate": 9.936830893331322e-06, "loss": 0.0028, "step": 35590 }, { "epoch": 0.5825083858299926, "grad_norm": 0.19774305820465088, "learning_rate": 9.93671771350145e-06, "loss": 0.005, "step": 35600 }, { "epoch": 0.5826720117810685, "grad_norm": 0.05936412885785103, "learning_rate": 9.936604433016099e-06, "loss": 0.0033, "step": 35610 }, { "epoch": 0.5828356377321443, "grad_norm": 0.023220546543598175, "learning_rate": 9.936491051877584e-06, "loss": 0.0035, "step": 35620 }, { "epoch": 0.5829992636832202, "grad_norm": 0.12107620388269424, "learning_rate": 9.936377570088216e-06, "loss": 0.005, "step": 35630 }, { "epoch": 0.583162889634296, "grad_norm": 0.31321418285369873, "learning_rate": 9.936263987650305e-06, "loss": 0.0048, "step": 35640 }, { "epoch": 0.5833265155853719, "grad_norm": 0.40815430879592896, "learning_rate": 9.936150304566173e-06, "loss": 0.0042, "step": 35650 }, { "epoch": 0.5834901415364476, "grad_norm": 0.13484618067741394, "learning_rate": 9.936036520838131e-06, "loss": 0.0041, "step": 35660 }, { "epoch": 0.5836537674875235, "grad_norm": 0.04688865318894386, "learning_rate": 9.935922636468503e-06, "loss": 0.0029, "step": 35670 }, { "epoch": 0.5838173934385994, "grad_norm": 0.07624917477369308, "learning_rate": 9.93580865145961e-06, "loss": 0.0026, "step": 35680 }, { "epoch": 0.5839810193896752, "grad_norm": 0.11773691326379776, "learning_rate": 9.935694565813778e-06, "loss": 0.0031, "step": 35690 }, { "epoch": 0.5841446453407511, "grad_norm": 0.01429735217243433, "learning_rate": 9.935580379533331e-06, "loss": 0.0012, "step": 35700 }, { "epoch": 0.5843082712918268, "grad_norm": 0.08978551626205444, "learning_rate": 9.935466092620598e-06, "loss": 0.004, "step": 35710 }, { "epoch": 0.5844718972429027, "grad_norm": 0.25593701004981995, "learning_rate": 9.935351705077907e-06, "loss": 0.0037, "step": 35720 }, { "epoch": 0.5846355231939786, "grad_norm": 0.07539242506027222, "learning_rate": 9.935237216907595e-06, "loss": 0.0032, "step": 35730 }, { "epoch": 0.5847991491450544, "grad_norm": 0.09403977543115616, "learning_rate": 9.93512262811199e-06, "loss": 0.0033, "step": 35740 }, { "epoch": 0.5849627750961303, "grad_norm": 0.07867195457220078, "learning_rate": 9.935007938693433e-06, "loss": 0.0029, "step": 35750 }, { "epoch": 0.585126401047206, "grad_norm": 0.08491113781929016, "learning_rate": 9.934893148654263e-06, "loss": 0.0083, "step": 35760 }, { "epoch": 0.5852900269982819, "grad_norm": 0.08775237947702408, "learning_rate": 9.934778257996817e-06, "loss": 0.0043, "step": 35770 }, { "epoch": 0.5854536529493578, "grad_norm": 0.06166079267859459, "learning_rate": 9.934663266723438e-06, "loss": 0.0031, "step": 35780 }, { "epoch": 0.5856172789004336, "grad_norm": 0.06222157925367355, "learning_rate": 9.934548174836473e-06, "loss": 0.0026, "step": 35790 }, { "epoch": 0.5857809048515095, "grad_norm": 0.0877544954419136, "learning_rate": 9.934432982338268e-06, "loss": 0.0039, "step": 35800 }, { "epoch": 0.5859445308025852, "grad_norm": 0.09067311137914658, "learning_rate": 9.934317689231172e-06, "loss": 0.0026, "step": 35810 }, { "epoch": 0.5861081567536611, "grad_norm": 0.12574119865894318, "learning_rate": 9.934202295517533e-06, "loss": 0.0032, "step": 35820 }, { "epoch": 0.586271782704737, "grad_norm": 0.10065688192844391, "learning_rate": 9.934086801199707e-06, "loss": 0.0019, "step": 35830 }, { "epoch": 0.5864354086558128, "grad_norm": 0.04277476668357849, "learning_rate": 9.933971206280047e-06, "loss": 0.0023, "step": 35840 }, { "epoch": 0.5865990346068887, "grad_norm": 0.08897034823894501, "learning_rate": 9.933855510760908e-06, "loss": 0.0023, "step": 35850 }, { "epoch": 0.5867626605579644, "grad_norm": 0.03773576021194458, "learning_rate": 9.933739714644653e-06, "loss": 0.0057, "step": 35860 }, { "epoch": 0.5869262865090403, "grad_norm": 0.05865627899765968, "learning_rate": 9.933623817933642e-06, "loss": 0.0018, "step": 35870 }, { "epoch": 0.5870899124601162, "grad_norm": 0.02294173091650009, "learning_rate": 9.933507820630237e-06, "loss": 0.0026, "step": 35880 }, { "epoch": 0.587253538411192, "grad_norm": 0.11480830609798431, "learning_rate": 9.933391722736802e-06, "loss": 0.0045, "step": 35890 }, { "epoch": 0.5874171643622679, "grad_norm": 0.016196081414818764, "learning_rate": 9.933275524255707e-06, "loss": 0.0025, "step": 35900 }, { "epoch": 0.5875807903133436, "grad_norm": 0.14820881187915802, "learning_rate": 9.933159225189318e-06, "loss": 0.0022, "step": 35910 }, { "epoch": 0.5877444162644195, "grad_norm": 0.029537130147218704, "learning_rate": 9.93304282554001e-06, "loss": 0.0031, "step": 35920 }, { "epoch": 0.5879080422154954, "grad_norm": 0.07905814051628113, "learning_rate": 9.932926325310153e-06, "loss": 0.0028, "step": 35930 }, { "epoch": 0.5880716681665712, "grad_norm": 0.17525748908519745, "learning_rate": 9.932809724502124e-06, "loss": 0.0033, "step": 35940 }, { "epoch": 0.5882352941176471, "grad_norm": 0.20093615353107452, "learning_rate": 9.932693023118299e-06, "loss": 0.0029, "step": 35950 }, { "epoch": 0.5883989200687229, "grad_norm": 0.15446005761623383, "learning_rate": 9.93257622116106e-06, "loss": 0.0025, "step": 35960 }, { "epoch": 0.5885625460197987, "grad_norm": 0.04283367842435837, "learning_rate": 9.932459318632783e-06, "loss": 0.0028, "step": 35970 }, { "epoch": 0.5887261719708746, "grad_norm": 0.07666134834289551, "learning_rate": 9.93234231553586e-06, "loss": 0.0036, "step": 35980 }, { "epoch": 0.5888897979219504, "grad_norm": 0.07410679012537003, "learning_rate": 9.932225211872669e-06, "loss": 0.0038, "step": 35990 }, { "epoch": 0.5890534238730263, "grad_norm": 0.06421056389808655, "learning_rate": 9.932108007645602e-06, "loss": 0.0021, "step": 36000 }, { "epoch": 0.5892170498241021, "grad_norm": 0.08304264396429062, "learning_rate": 9.931990702857047e-06, "loss": 0.0028, "step": 36010 }, { "epoch": 0.5893806757751779, "grad_norm": 0.07150162011384964, "learning_rate": 9.931873297509396e-06, "loss": 0.0021, "step": 36020 }, { "epoch": 0.5895443017262538, "grad_norm": 0.342092901468277, "learning_rate": 9.931755791605043e-06, "loss": 0.0039, "step": 36030 }, { "epoch": 0.5897079276773296, "grad_norm": 0.05948949605226517, "learning_rate": 9.931638185146383e-06, "loss": 0.0033, "step": 36040 }, { "epoch": 0.5898715536284055, "grad_norm": 0.1427544206380844, "learning_rate": 9.931520478135817e-06, "loss": 0.0031, "step": 36050 }, { "epoch": 0.5900351795794813, "grad_norm": 0.12840580940246582, "learning_rate": 9.93140267057574e-06, "loss": 0.0036, "step": 36060 }, { "epoch": 0.5901988055305571, "grad_norm": 0.09037903696298599, "learning_rate": 9.931284762468558e-06, "loss": 0.0028, "step": 36070 }, { "epoch": 0.590362431481633, "grad_norm": 0.15646789968013763, "learning_rate": 9.931166753816673e-06, "loss": 0.0035, "step": 36080 }, { "epoch": 0.5905260574327088, "grad_norm": 0.16927802562713623, "learning_rate": 9.931048644622491e-06, "loss": 0.0016, "step": 36090 }, { "epoch": 0.5906896833837847, "grad_norm": 0.020790643990039825, "learning_rate": 9.930930434888422e-06, "loss": 0.0022, "step": 36100 }, { "epoch": 0.5908533093348605, "grad_norm": 0.1678667664527893, "learning_rate": 9.930812124616876e-06, "loss": 0.0032, "step": 36110 }, { "epoch": 0.5910169352859364, "grad_norm": 0.10719109326601028, "learning_rate": 9.930693713810262e-06, "loss": 0.0024, "step": 36120 }, { "epoch": 0.5911805612370122, "grad_norm": 0.05181542783975601, "learning_rate": 9.930575202470998e-06, "loss": 0.0021, "step": 36130 }, { "epoch": 0.591344187188088, "grad_norm": 0.12804168462753296, "learning_rate": 9.930456590601499e-06, "loss": 0.0036, "step": 36140 }, { "epoch": 0.5915078131391639, "grad_norm": 0.050599128007888794, "learning_rate": 9.930337878204182e-06, "loss": 0.002, "step": 36150 }, { "epoch": 0.5916714390902397, "grad_norm": 0.0891394391655922, "learning_rate": 9.93021906528147e-06, "loss": 0.0031, "step": 36160 }, { "epoch": 0.5918350650413156, "grad_norm": 0.21372178196907043, "learning_rate": 9.930100151835784e-06, "loss": 0.0033, "step": 36170 }, { "epoch": 0.5919986909923914, "grad_norm": 0.1381070613861084, "learning_rate": 9.929981137869548e-06, "loss": 0.0048, "step": 36180 }, { "epoch": 0.5921623169434672, "grad_norm": 0.22583429515361786, "learning_rate": 9.92986202338519e-06, "loss": 0.0043, "step": 36190 }, { "epoch": 0.5923259428945431, "grad_norm": 0.033615972846746445, "learning_rate": 9.929742808385139e-06, "loss": 0.0026, "step": 36200 }, { "epoch": 0.5924895688456189, "grad_norm": 0.11740265786647797, "learning_rate": 9.929623492871823e-06, "loss": 0.0041, "step": 36210 }, { "epoch": 0.5926531947966948, "grad_norm": 0.30242079496383667, "learning_rate": 9.929504076847677e-06, "loss": 0.0048, "step": 36220 }, { "epoch": 0.5928168207477706, "grad_norm": 0.05738528072834015, "learning_rate": 9.929384560315135e-06, "loss": 0.0026, "step": 36230 }, { "epoch": 0.5929804466988464, "grad_norm": 0.07402373105287552, "learning_rate": 9.929264943276635e-06, "loss": 0.0059, "step": 36240 }, { "epoch": 0.5931440726499223, "grad_norm": 0.10457896441221237, "learning_rate": 9.929145225734615e-06, "loss": 0.004, "step": 36250 }, { "epoch": 0.5933076986009981, "grad_norm": 0.09303756803274155, "learning_rate": 9.929025407691516e-06, "loss": 0.0025, "step": 36260 }, { "epoch": 0.593471324552074, "grad_norm": 0.37263035774230957, "learning_rate": 9.92890548914978e-06, "loss": 0.0041, "step": 36270 }, { "epoch": 0.5936349505031498, "grad_norm": 0.18307627737522125, "learning_rate": 9.928785470111852e-06, "loss": 0.0024, "step": 36280 }, { "epoch": 0.5937985764542256, "grad_norm": 0.028225671499967575, "learning_rate": 9.928665350580182e-06, "loss": 0.0023, "step": 36290 }, { "epoch": 0.5939622024053015, "grad_norm": 0.11651282012462616, "learning_rate": 9.928545130557216e-06, "loss": 0.0028, "step": 36300 }, { "epoch": 0.5941258283563773, "grad_norm": 0.2828836441040039, "learning_rate": 9.928424810045407e-06, "loss": 0.0047, "step": 36310 }, { "epoch": 0.5942894543074532, "grad_norm": 0.0644621029496193, "learning_rate": 9.928304389047209e-06, "loss": 0.0025, "step": 36320 }, { "epoch": 0.594453080258529, "grad_norm": 0.03878481313586235, "learning_rate": 9.928183867565074e-06, "loss": 0.0045, "step": 36330 }, { "epoch": 0.5946167062096048, "grad_norm": 0.1487322747707367, "learning_rate": 9.928063245601463e-06, "loss": 0.0046, "step": 36340 }, { "epoch": 0.5947803321606807, "grad_norm": 0.0435752235352993, "learning_rate": 9.927942523158831e-06, "loss": 0.0028, "step": 36350 }, { "epoch": 0.5949439581117565, "grad_norm": 0.2003866583108902, "learning_rate": 9.927821700239643e-06, "loss": 0.0029, "step": 36360 }, { "epoch": 0.5951075840628324, "grad_norm": 0.3755134046077728, "learning_rate": 9.927700776846362e-06, "loss": 0.0037, "step": 36370 }, { "epoch": 0.5952712100139083, "grad_norm": 0.0582570880651474, "learning_rate": 9.927579752981454e-06, "loss": 0.0021, "step": 36380 }, { "epoch": 0.595434835964984, "grad_norm": 0.06495503336191177, "learning_rate": 9.927458628647384e-06, "loss": 0.0025, "step": 36390 }, { "epoch": 0.5955984619160599, "grad_norm": 0.09407453238964081, "learning_rate": 9.927337403846622e-06, "loss": 0.0045, "step": 36400 }, { "epoch": 0.5957620878671357, "grad_norm": 0.28359636664390564, "learning_rate": 9.927216078581644e-06, "loss": 0.0039, "step": 36410 }, { "epoch": 0.5959257138182116, "grad_norm": 0.08559110015630722, "learning_rate": 9.927094652854919e-06, "loss": 0.0038, "step": 36420 }, { "epoch": 0.5960893397692875, "grad_norm": 0.18429267406463623, "learning_rate": 9.926973126668923e-06, "loss": 0.0043, "step": 36430 }, { "epoch": 0.5962529657203632, "grad_norm": 0.05833103507757187, "learning_rate": 9.926851500026138e-06, "loss": 0.0047, "step": 36440 }, { "epoch": 0.5964165916714391, "grad_norm": 0.047782596200704575, "learning_rate": 9.92672977292904e-06, "loss": 0.0042, "step": 36450 }, { "epoch": 0.5965802176225149, "grad_norm": 0.07527091354131699, "learning_rate": 9.926607945380111e-06, "loss": 0.0019, "step": 36460 }, { "epoch": 0.5967438435735908, "grad_norm": 0.008818336762487888, "learning_rate": 9.926486017381838e-06, "loss": 0.0018, "step": 36470 }, { "epoch": 0.5969074695246667, "grad_norm": 0.04545479267835617, "learning_rate": 9.926363988936703e-06, "loss": 0.0021, "step": 36480 }, { "epoch": 0.5970710954757424, "grad_norm": 0.15527421236038208, "learning_rate": 9.926241860047197e-06, "loss": 0.0027, "step": 36490 }, { "epoch": 0.5972347214268183, "grad_norm": 0.0678648129105568, "learning_rate": 9.926119630715808e-06, "loss": 0.0017, "step": 36500 }, { "epoch": 0.5973983473778941, "grad_norm": 0.12282824516296387, "learning_rate": 9.925997300945029e-06, "loss": 0.004, "step": 36510 }, { "epoch": 0.59756197332897, "grad_norm": 0.0667882114648819, "learning_rate": 9.925874870737356e-06, "loss": 0.003, "step": 36520 }, { "epoch": 0.5977255992800458, "grad_norm": 0.15810398757457733, "learning_rate": 9.925752340095283e-06, "loss": 0.0025, "step": 36530 }, { "epoch": 0.5978892252311216, "grad_norm": 0.09229476749897003, "learning_rate": 9.92562970902131e-06, "loss": 0.0042, "step": 36540 }, { "epoch": 0.5980528511821975, "grad_norm": 0.2527562081813812, "learning_rate": 9.925506977517934e-06, "loss": 0.0053, "step": 36550 }, { "epoch": 0.5982164771332733, "grad_norm": 0.16690213978290558, "learning_rate": 9.925384145587662e-06, "loss": 0.0036, "step": 36560 }, { "epoch": 0.5983801030843492, "grad_norm": 0.29717326164245605, "learning_rate": 9.925261213232995e-06, "loss": 0.003, "step": 36570 }, { "epoch": 0.598543729035425, "grad_norm": 0.06168719008564949, "learning_rate": 9.92513818045644e-06, "loss": 0.0041, "step": 36580 }, { "epoch": 0.5987073549865009, "grad_norm": 0.10478661954402924, "learning_rate": 9.925015047260508e-06, "loss": 0.0024, "step": 36590 }, { "epoch": 0.5988709809375767, "grad_norm": 0.10058804601430893, "learning_rate": 9.924891813647707e-06, "loss": 0.0044, "step": 36600 }, { "epoch": 0.5990346068886525, "grad_norm": 0.12004987895488739, "learning_rate": 9.92476847962055e-06, "loss": 0.0042, "step": 36610 }, { "epoch": 0.5991982328397284, "grad_norm": 0.07407871633768082, "learning_rate": 9.924645045181552e-06, "loss": 0.0032, "step": 36620 }, { "epoch": 0.5993618587908042, "grad_norm": 0.03402946889400482, "learning_rate": 9.924521510333231e-06, "loss": 0.003, "step": 36630 }, { "epoch": 0.59952548474188, "grad_norm": 0.2208137810230255, "learning_rate": 9.924397875078103e-06, "loss": 0.0041, "step": 36640 }, { "epoch": 0.5996891106929559, "grad_norm": 0.13660487532615662, "learning_rate": 9.92427413941869e-06, "loss": 0.003, "step": 36650 }, { "epoch": 0.5998527366440317, "grad_norm": 0.21275994181632996, "learning_rate": 9.924150303357517e-06, "loss": 0.0035, "step": 36660 }, { "epoch": 0.6000163625951076, "grad_norm": 0.36380282044410706, "learning_rate": 9.924026366897106e-06, "loss": 0.0048, "step": 36670 }, { "epoch": 0.6001799885461834, "grad_norm": 1.4959594011306763, "learning_rate": 9.923902330039986e-06, "loss": 0.0052, "step": 36680 }, { "epoch": 0.6003436144972593, "grad_norm": 0.16231313347816467, "learning_rate": 9.923778192788685e-06, "loss": 0.0031, "step": 36690 }, { "epoch": 0.6005072404483351, "grad_norm": 0.11729563772678375, "learning_rate": 9.923653955145733e-06, "loss": 0.0042, "step": 36700 }, { "epoch": 0.6006708663994109, "grad_norm": 0.0901319831609726, "learning_rate": 9.923529617113665e-06, "loss": 0.0044, "step": 36710 }, { "epoch": 0.6008344923504868, "grad_norm": 0.15310680866241455, "learning_rate": 9.923405178695016e-06, "loss": 0.003, "step": 36720 }, { "epoch": 0.6009981183015626, "grad_norm": 0.02770262025296688, "learning_rate": 9.923280639892321e-06, "loss": 0.0051, "step": 36730 }, { "epoch": 0.6011617442526385, "grad_norm": 0.04368910938501358, "learning_rate": 9.923156000708123e-06, "loss": 0.0027, "step": 36740 }, { "epoch": 0.6013253702037143, "grad_norm": 0.11642678081989288, "learning_rate": 9.92303126114496e-06, "loss": 0.0038, "step": 36750 }, { "epoch": 0.6014889961547901, "grad_norm": 0.15905441343784332, "learning_rate": 9.922906421205376e-06, "loss": 0.0035, "step": 36760 }, { "epoch": 0.601652622105866, "grad_norm": 0.15553593635559082, "learning_rate": 9.922781480891918e-06, "loss": 0.0032, "step": 36770 }, { "epoch": 0.6018162480569418, "grad_norm": 0.08895600587129593, "learning_rate": 9.922656440207133e-06, "loss": 0.0038, "step": 36780 }, { "epoch": 0.6019798740080177, "grad_norm": 0.16574913263320923, "learning_rate": 9.922531299153567e-06, "loss": 0.0029, "step": 36790 }, { "epoch": 0.6021434999590936, "grad_norm": 0.038676243275403976, "learning_rate": 9.922406057733776e-06, "loss": 0.0019, "step": 36800 }, { "epoch": 0.6023071259101693, "grad_norm": 0.16262957453727722, "learning_rate": 9.922280715950313e-06, "loss": 0.0026, "step": 36810 }, { "epoch": 0.6024707518612452, "grad_norm": 0.06429680436849594, "learning_rate": 9.92215527380573e-06, "loss": 0.0026, "step": 36820 }, { "epoch": 0.602634377812321, "grad_norm": 0.21636322140693665, "learning_rate": 9.922029731302587e-06, "loss": 0.0034, "step": 36830 }, { "epoch": 0.6027980037633969, "grad_norm": 0.023974796757102013, "learning_rate": 9.921904088443447e-06, "loss": 0.0063, "step": 36840 }, { "epoch": 0.6029616297144728, "grad_norm": 0.06435230374336243, "learning_rate": 9.921778345230867e-06, "loss": 0.0017, "step": 36850 }, { "epoch": 0.6031252556655485, "grad_norm": 0.15381374955177307, "learning_rate": 9.92165250166741e-06, "loss": 0.0049, "step": 36860 }, { "epoch": 0.6032888816166244, "grad_norm": 0.0123814782127738, "learning_rate": 9.921526557755647e-06, "loss": 0.004, "step": 36870 }, { "epoch": 0.6034525075677002, "grad_norm": 0.07016071677207947, "learning_rate": 9.92140051349814e-06, "loss": 0.0044, "step": 36880 }, { "epoch": 0.6036161335187761, "grad_norm": 0.08831898868083954, "learning_rate": 9.921274368897465e-06, "loss": 0.0038, "step": 36890 }, { "epoch": 0.603779759469852, "grad_norm": 0.1637003868818283, "learning_rate": 9.921148123956191e-06, "loss": 0.004, "step": 36900 }, { "epoch": 0.6039433854209277, "grad_norm": 0.08297514170408249, "learning_rate": 9.921021778676892e-06, "loss": 0.0043, "step": 36910 }, { "epoch": 0.6041070113720036, "grad_norm": 0.1162319928407669, "learning_rate": 9.920895333062142e-06, "loss": 0.0034, "step": 36920 }, { "epoch": 0.6042706373230794, "grad_norm": 0.04002273455262184, "learning_rate": 9.920768787114523e-06, "loss": 0.0033, "step": 36930 }, { "epoch": 0.6044342632741553, "grad_norm": 0.14166095852851868, "learning_rate": 9.920642140836613e-06, "loss": 0.0041, "step": 36940 }, { "epoch": 0.6045978892252312, "grad_norm": 0.05039995163679123, "learning_rate": 9.920515394230994e-06, "loss": 0.0037, "step": 36950 }, { "epoch": 0.6047615151763069, "grad_norm": 0.1033048927783966, "learning_rate": 9.920388547300252e-06, "loss": 0.0024, "step": 36960 }, { "epoch": 0.6049251411273828, "grad_norm": 0.10160665959119797, "learning_rate": 9.920261600046972e-06, "loss": 0.0052, "step": 36970 }, { "epoch": 0.6050887670784586, "grad_norm": 0.07268615067005157, "learning_rate": 9.920134552473741e-06, "loss": 0.0045, "step": 36980 }, { "epoch": 0.6052523930295345, "grad_norm": 0.3050188720226288, "learning_rate": 9.920007404583153e-06, "loss": 0.0026, "step": 36990 }, { "epoch": 0.6054160189806104, "grad_norm": 0.07098031789064407, "learning_rate": 9.919880156377796e-06, "loss": 0.0031, "step": 37000 }, { "epoch": 0.6055796449316861, "grad_norm": 0.09979721903800964, "learning_rate": 9.91975280786027e-06, "loss": 0.0034, "step": 37010 }, { "epoch": 0.605743270882762, "grad_norm": 0.04441436007618904, "learning_rate": 9.919625359033166e-06, "loss": 0.0032, "step": 37020 }, { "epoch": 0.6059068968338378, "grad_norm": 0.09820044040679932, "learning_rate": 9.919497809899086e-06, "loss": 0.0032, "step": 37030 }, { "epoch": 0.6060705227849137, "grad_norm": 0.1345708966255188, "learning_rate": 9.919370160460629e-06, "loss": 0.0026, "step": 37040 }, { "epoch": 0.6062341487359896, "grad_norm": 0.04649454727768898, "learning_rate": 9.919242410720397e-06, "loss": 0.0082, "step": 37050 }, { "epoch": 0.6063977746870653, "grad_norm": 0.3657276928424835, "learning_rate": 9.919114560680997e-06, "loss": 0.0057, "step": 37060 }, { "epoch": 0.6065614006381412, "grad_norm": 0.12013202905654907, "learning_rate": 9.918986610345036e-06, "loss": 0.0034, "step": 37070 }, { "epoch": 0.606725026589217, "grad_norm": 0.5230374932289124, "learning_rate": 9.91885855971512e-06, "loss": 0.0038, "step": 37080 }, { "epoch": 0.6068886525402929, "grad_norm": 0.028463061898946762, "learning_rate": 9.918730408793861e-06, "loss": 0.0018, "step": 37090 }, { "epoch": 0.6070522784913688, "grad_norm": 0.0755695104598999, "learning_rate": 9.918602157583874e-06, "loss": 0.0038, "step": 37100 }, { "epoch": 0.6072159044424446, "grad_norm": 0.08192359656095505, "learning_rate": 9.918473806087769e-06, "loss": 0.0048, "step": 37110 }, { "epoch": 0.6073795303935204, "grad_norm": 0.09471998363733292, "learning_rate": 9.918345354308169e-06, "loss": 0.0032, "step": 37120 }, { "epoch": 0.6075431563445962, "grad_norm": 0.05932093784213066, "learning_rate": 9.91821680224769e-06, "loss": 0.0026, "step": 37130 }, { "epoch": 0.6077067822956721, "grad_norm": 0.2354872077703476, "learning_rate": 9.918088149908951e-06, "loss": 0.0022, "step": 37140 }, { "epoch": 0.607870408246748, "grad_norm": 0.11630737036466599, "learning_rate": 9.917959397294578e-06, "loss": 0.0046, "step": 37150 }, { "epoch": 0.6080340341978238, "grad_norm": 0.09637027978897095, "learning_rate": 9.917830544407197e-06, "loss": 0.0031, "step": 37160 }, { "epoch": 0.6081976601488996, "grad_norm": 0.03533605486154556, "learning_rate": 9.917701591249432e-06, "loss": 0.0023, "step": 37170 }, { "epoch": 0.6083612860999754, "grad_norm": 0.11813351511955261, "learning_rate": 9.917572537823915e-06, "loss": 0.0047, "step": 37180 }, { "epoch": 0.6085249120510513, "grad_norm": 0.05641971901059151, "learning_rate": 9.917443384133275e-06, "loss": 0.0033, "step": 37190 }, { "epoch": 0.6086885380021272, "grad_norm": 0.14030763506889343, "learning_rate": 9.917314130180149e-06, "loss": 0.003, "step": 37200 }, { "epoch": 0.608852163953203, "grad_norm": 0.12399761378765106, "learning_rate": 9.917184775967168e-06, "loss": 0.0029, "step": 37210 }, { "epoch": 0.6090157899042788, "grad_norm": 0.27373266220092773, "learning_rate": 9.917055321496972e-06, "loss": 0.0093, "step": 37220 }, { "epoch": 0.6091794158553546, "grad_norm": 0.2558335065841675, "learning_rate": 9.916925766772198e-06, "loss": 0.003, "step": 37230 }, { "epoch": 0.6093430418064305, "grad_norm": 0.14289845526218414, "learning_rate": 9.916796111795491e-06, "loss": 0.0027, "step": 37240 }, { "epoch": 0.6095066677575064, "grad_norm": 0.12105997651815414, "learning_rate": 9.916666356569493e-06, "loss": 0.0027, "step": 37250 }, { "epoch": 0.6096702937085822, "grad_norm": 0.052745696157217026, "learning_rate": 9.916536501096849e-06, "loss": 0.0025, "step": 37260 }, { "epoch": 0.609833919659658, "grad_norm": 0.28806161880493164, "learning_rate": 9.916406545380207e-06, "loss": 0.0054, "step": 37270 }, { "epoch": 0.6099975456107338, "grad_norm": 0.03294442221522331, "learning_rate": 9.916276489422218e-06, "loss": 0.0025, "step": 37280 }, { "epoch": 0.6101611715618097, "grad_norm": 0.21975286304950714, "learning_rate": 9.916146333225532e-06, "loss": 0.0042, "step": 37290 }, { "epoch": 0.6103247975128856, "grad_norm": 0.2020818591117859, "learning_rate": 9.916016076792802e-06, "loss": 0.0024, "step": 37300 }, { "epoch": 0.6104884234639614, "grad_norm": 0.5058445930480957, "learning_rate": 9.915885720126685e-06, "loss": 0.0029, "step": 37310 }, { "epoch": 0.6106520494150373, "grad_norm": 0.13238525390625, "learning_rate": 9.91575526322984e-06, "loss": 0.0031, "step": 37320 }, { "epoch": 0.610815675366113, "grad_norm": 0.18898938596248627, "learning_rate": 9.915624706104926e-06, "loss": 0.0023, "step": 37330 }, { "epoch": 0.6109793013171889, "grad_norm": 0.4076012969017029, "learning_rate": 9.915494048754605e-06, "loss": 0.0049, "step": 37340 }, { "epoch": 0.6111429272682648, "grad_norm": 0.07381382584571838, "learning_rate": 9.91536329118154e-06, "loss": 0.0034, "step": 37350 }, { "epoch": 0.6113065532193406, "grad_norm": 0.14857052266597748, "learning_rate": 9.915232433388397e-06, "loss": 0.0039, "step": 37360 }, { "epoch": 0.6114701791704165, "grad_norm": 0.026397282257676125, "learning_rate": 9.915101475377847e-06, "loss": 0.0025, "step": 37370 }, { "epoch": 0.6116338051214922, "grad_norm": 0.1670699417591095, "learning_rate": 9.914970417152558e-06, "loss": 0.0039, "step": 37380 }, { "epoch": 0.6117974310725681, "grad_norm": 0.11151667684316635, "learning_rate": 9.914839258715201e-06, "loss": 0.002, "step": 37390 }, { "epoch": 0.6119610570236439, "grad_norm": 0.07161849737167358, "learning_rate": 9.914708000068452e-06, "loss": 0.0015, "step": 37400 }, { "epoch": 0.6121246829747198, "grad_norm": 0.10908344388008118, "learning_rate": 9.914576641214987e-06, "loss": 0.0035, "step": 37410 }, { "epoch": 0.6122883089257957, "grad_norm": 0.06375255435705185, "learning_rate": 9.914445182157484e-06, "loss": 0.0023, "step": 37420 }, { "epoch": 0.6124519348768714, "grad_norm": 0.06880512088537216, "learning_rate": 9.914313622898622e-06, "loss": 0.0037, "step": 37430 }, { "epoch": 0.6126155608279473, "grad_norm": 0.11759839951992035, "learning_rate": 9.914181963441087e-06, "loss": 0.0025, "step": 37440 }, { "epoch": 0.6127791867790231, "grad_norm": 0.11998670548200607, "learning_rate": 9.914050203787559e-06, "loss": 0.0021, "step": 37450 }, { "epoch": 0.612942812730099, "grad_norm": 0.08501125872135162, "learning_rate": 9.913918343940728e-06, "loss": 0.0043, "step": 37460 }, { "epoch": 0.6131064386811749, "grad_norm": 0.21158826351165771, "learning_rate": 9.91378638390328e-06, "loss": 0.0026, "step": 37470 }, { "epoch": 0.6132700646322506, "grad_norm": 0.06731773167848587, "learning_rate": 9.913654323677907e-06, "loss": 0.0026, "step": 37480 }, { "epoch": 0.6134336905833265, "grad_norm": 0.1324690580368042, "learning_rate": 9.913522163267302e-06, "loss": 0.0038, "step": 37490 }, { "epoch": 0.6135973165344023, "grad_norm": 0.050356000661849976, "learning_rate": 9.913389902674158e-06, "loss": 0.0036, "step": 37500 }, { "epoch": 0.6137609424854782, "grad_norm": 0.08342663943767548, "learning_rate": 9.913257541901172e-06, "loss": 0.0045, "step": 37510 }, { "epoch": 0.6139245684365541, "grad_norm": 0.2870792746543884, "learning_rate": 9.913125080951046e-06, "loss": 0.0051, "step": 37520 }, { "epoch": 0.6140881943876298, "grad_norm": 0.04826710373163223, "learning_rate": 9.912992519826474e-06, "loss": 0.0017, "step": 37530 }, { "epoch": 0.6142518203387057, "grad_norm": 0.016608064994215965, "learning_rate": 9.912859858530167e-06, "loss": 0.0049, "step": 37540 }, { "epoch": 0.6144154462897815, "grad_norm": 0.1234755665063858, "learning_rate": 9.912727097064822e-06, "loss": 0.0021, "step": 37550 }, { "epoch": 0.6145790722408574, "grad_norm": 0.009607202373445034, "learning_rate": 9.912594235433152e-06, "loss": 0.0023, "step": 37560 }, { "epoch": 0.6147426981919333, "grad_norm": 0.044833190739154816, "learning_rate": 9.912461273637863e-06, "loss": 0.0035, "step": 37570 }, { "epoch": 0.614906324143009, "grad_norm": 0.12280149012804031, "learning_rate": 9.912328211681667e-06, "loss": 0.0039, "step": 37580 }, { "epoch": 0.6150699500940849, "grad_norm": 0.11728564649820328, "learning_rate": 9.912195049567275e-06, "loss": 0.0049, "step": 37590 }, { "epoch": 0.6152335760451607, "grad_norm": 0.06474844366312027, "learning_rate": 9.912061787297405e-06, "loss": 0.0078, "step": 37600 }, { "epoch": 0.6153972019962366, "grad_norm": 0.032887157052755356, "learning_rate": 9.911928424874773e-06, "loss": 0.0021, "step": 37610 }, { "epoch": 0.6155608279473125, "grad_norm": 0.44525888562202454, "learning_rate": 9.911794962302098e-06, "loss": 0.0034, "step": 37620 }, { "epoch": 0.6157244538983883, "grad_norm": 0.06811727583408356, "learning_rate": 9.9116613995821e-06, "loss": 0.0022, "step": 37630 }, { "epoch": 0.6158880798494641, "grad_norm": 0.1474253088235855, "learning_rate": 9.911527736717503e-06, "loss": 0.0038, "step": 37640 }, { "epoch": 0.6160517058005399, "grad_norm": 0.15572436153888702, "learning_rate": 9.911393973711031e-06, "loss": 0.0039, "step": 37650 }, { "epoch": 0.6162153317516158, "grad_norm": 0.27177467942237854, "learning_rate": 9.911260110565416e-06, "loss": 0.0035, "step": 37660 }, { "epoch": 0.6163789577026917, "grad_norm": 0.14696092903614044, "learning_rate": 9.911126147283383e-06, "loss": 0.0033, "step": 37670 }, { "epoch": 0.6165425836537675, "grad_norm": 0.26636818051338196, "learning_rate": 9.910992083867665e-06, "loss": 0.0047, "step": 37680 }, { "epoch": 0.6167062096048433, "grad_norm": 0.1550528109073639, "learning_rate": 9.910857920320995e-06, "loss": 0.0027, "step": 37690 }, { "epoch": 0.6168698355559191, "grad_norm": 0.10983145982027054, "learning_rate": 9.910723656646108e-06, "loss": 0.003, "step": 37700 }, { "epoch": 0.617033461506995, "grad_norm": 0.08848784118890762, "learning_rate": 9.910589292845742e-06, "loss": 0.0037, "step": 37710 }, { "epoch": 0.6171970874580709, "grad_norm": 0.06501125544309616, "learning_rate": 9.910454828922638e-06, "loss": 0.0033, "step": 37720 }, { "epoch": 0.6173607134091467, "grad_norm": 0.14493057131767273, "learning_rate": 9.910320264879536e-06, "loss": 0.0042, "step": 37730 }, { "epoch": 0.6175243393602226, "grad_norm": 0.043875016272068024, "learning_rate": 9.910185600719179e-06, "loss": 0.0046, "step": 37740 }, { "epoch": 0.6176879653112983, "grad_norm": 0.0887451097369194, "learning_rate": 9.910050836444314e-06, "loss": 0.0041, "step": 37750 }, { "epoch": 0.6178515912623742, "grad_norm": 0.07818350940942764, "learning_rate": 9.909915972057688e-06, "loss": 0.0027, "step": 37760 }, { "epoch": 0.6180152172134501, "grad_norm": 0.28528571128845215, "learning_rate": 9.909781007562052e-06, "loss": 0.0045, "step": 37770 }, { "epoch": 0.6181788431645259, "grad_norm": 0.23261074721813202, "learning_rate": 9.909645942960156e-06, "loss": 0.0042, "step": 37780 }, { "epoch": 0.6183424691156018, "grad_norm": 0.13355617225170135, "learning_rate": 9.909510778254754e-06, "loss": 0.0022, "step": 37790 }, { "epoch": 0.6185060950666775, "grad_norm": 0.006417717318981886, "learning_rate": 9.909375513448603e-06, "loss": 0.0044, "step": 37800 }, { "epoch": 0.6186697210177534, "grad_norm": 0.03255303576588631, "learning_rate": 9.90924014854446e-06, "loss": 0.0041, "step": 37810 }, { "epoch": 0.6188333469688293, "grad_norm": 0.1925007402896881, "learning_rate": 9.909104683545088e-06, "loss": 0.0059, "step": 37820 }, { "epoch": 0.6189969729199051, "grad_norm": 0.2600554823875427, "learning_rate": 9.908969118453244e-06, "loss": 0.0037, "step": 37830 }, { "epoch": 0.619160598870981, "grad_norm": 0.49556082487106323, "learning_rate": 9.908833453271695e-06, "loss": 0.0049, "step": 37840 }, { "epoch": 0.6193242248220567, "grad_norm": 0.19424335658550262, "learning_rate": 9.908697688003208e-06, "loss": 0.0048, "step": 37850 }, { "epoch": 0.6194878507731326, "grad_norm": 0.2291473001241684, "learning_rate": 9.90856182265055e-06, "loss": 0.0047, "step": 37860 }, { "epoch": 0.6196514767242085, "grad_norm": 0.1466854363679886, "learning_rate": 9.90842585721649e-06, "loss": 0.0032, "step": 37870 }, { "epoch": 0.6198151026752843, "grad_norm": 0.14718101918697357, "learning_rate": 9.908289791703801e-06, "loss": 0.0021, "step": 37880 }, { "epoch": 0.6199787286263602, "grad_norm": 0.22129136323928833, "learning_rate": 9.908153626115258e-06, "loss": 0.0047, "step": 37890 }, { "epoch": 0.6201423545774359, "grad_norm": 0.05742582678794861, "learning_rate": 9.908017360453636e-06, "loss": 0.0027, "step": 37900 }, { "epoch": 0.6203059805285118, "grad_norm": 0.0704508125782013, "learning_rate": 9.907880994721714e-06, "loss": 0.0039, "step": 37910 }, { "epoch": 0.6204696064795877, "grad_norm": 0.06838686764240265, "learning_rate": 9.907744528922274e-06, "loss": 0.0027, "step": 37920 }, { "epoch": 0.6206332324306635, "grad_norm": 0.16333092749118805, "learning_rate": 9.907607963058096e-06, "loss": 0.0034, "step": 37930 }, { "epoch": 0.6207968583817394, "grad_norm": 0.1378253698348999, "learning_rate": 9.907471297131967e-06, "loss": 0.0064, "step": 37940 }, { "epoch": 0.6209604843328151, "grad_norm": 0.03919297084212303, "learning_rate": 9.90733453114667e-06, "loss": 0.0028, "step": 37950 }, { "epoch": 0.621124110283891, "grad_norm": 0.1876300573348999, "learning_rate": 9.907197665104997e-06, "loss": 0.004, "step": 37960 }, { "epoch": 0.6212877362349669, "grad_norm": 0.15879355370998383, "learning_rate": 9.907060699009736e-06, "loss": 0.0024, "step": 37970 }, { "epoch": 0.6214513621860427, "grad_norm": 0.14715738594532013, "learning_rate": 9.906923632863682e-06, "loss": 0.0025, "step": 37980 }, { "epoch": 0.6216149881371186, "grad_norm": 0.125774547457695, "learning_rate": 9.906786466669624e-06, "loss": 0.0032, "step": 37990 }, { "epoch": 0.6217786140881943, "grad_norm": 0.09183130413293839, "learning_rate": 9.906649200430367e-06, "loss": 0.0022, "step": 38000 }, { "epoch": 0.6219422400392702, "grad_norm": 0.0563751719892025, "learning_rate": 9.906511834148706e-06, "loss": 0.003, "step": 38010 }, { "epoch": 0.6221058659903461, "grad_norm": 0.05822442099452019, "learning_rate": 9.90637436782744e-06, "loss": 0.0023, "step": 38020 }, { "epoch": 0.6222694919414219, "grad_norm": 0.06481768190860748, "learning_rate": 9.906236801469373e-06, "loss": 0.0031, "step": 38030 }, { "epoch": 0.6224331178924978, "grad_norm": 0.1655895859003067, "learning_rate": 9.906099135077312e-06, "loss": 0.0039, "step": 38040 }, { "epoch": 0.6225967438435736, "grad_norm": 0.034479837864637375, "learning_rate": 9.90596136865406e-06, "loss": 0.0042, "step": 38050 }, { "epoch": 0.6227603697946494, "grad_norm": 0.06327641755342484, "learning_rate": 9.90582350220243e-06, "loss": 0.0019, "step": 38060 }, { "epoch": 0.6229239957457253, "grad_norm": 0.09807563573122025, "learning_rate": 9.905685535725228e-06, "loss": 0.0032, "step": 38070 }, { "epoch": 0.6230876216968011, "grad_norm": 0.0971381738781929, "learning_rate": 9.905547469225274e-06, "loss": 0.0027, "step": 38080 }, { "epoch": 0.623251247647877, "grad_norm": 0.014928369782865047, "learning_rate": 9.905409302705377e-06, "loss": 0.0034, "step": 38090 }, { "epoch": 0.6234148735989528, "grad_norm": 0.11387858539819717, "learning_rate": 9.905271036168357e-06, "loss": 0.0028, "step": 38100 }, { "epoch": 0.6235784995500286, "grad_norm": 0.5354669690132141, "learning_rate": 9.905132669617033e-06, "loss": 0.0055, "step": 38110 }, { "epoch": 0.6237421255011045, "grad_norm": 0.0616229809820652, "learning_rate": 9.904994203054224e-06, "loss": 0.0017, "step": 38120 }, { "epoch": 0.6239057514521803, "grad_norm": 0.1180015280842781, "learning_rate": 9.904855636482755e-06, "loss": 0.0019, "step": 38130 }, { "epoch": 0.6240693774032562, "grad_norm": 0.011107313446700573, "learning_rate": 9.90471696990545e-06, "loss": 0.0039, "step": 38140 }, { "epoch": 0.624233003354332, "grad_norm": 0.04674747213721275, "learning_rate": 9.90457820332514e-06, "loss": 0.0047, "step": 38150 }, { "epoch": 0.6243966293054078, "grad_norm": 0.1691882461309433, "learning_rate": 9.90443933674465e-06, "loss": 0.0051, "step": 38160 }, { "epoch": 0.6245602552564837, "grad_norm": 0.050062134861946106, "learning_rate": 9.904300370166813e-06, "loss": 0.0029, "step": 38170 }, { "epoch": 0.6247238812075595, "grad_norm": 0.16498225927352905, "learning_rate": 9.904161303594461e-06, "loss": 0.0032, "step": 38180 }, { "epoch": 0.6248875071586354, "grad_norm": 0.0054762824438512325, "learning_rate": 9.904022137030433e-06, "loss": 0.0041, "step": 38190 }, { "epoch": 0.6250511331097112, "grad_norm": 0.4126860499382019, "learning_rate": 9.903882870477563e-06, "loss": 0.0026, "step": 38200 }, { "epoch": 0.625214759060787, "grad_norm": 0.19153623282909393, "learning_rate": 9.903743503938693e-06, "loss": 0.0026, "step": 38210 }, { "epoch": 0.6253783850118629, "grad_norm": 0.11115068942308426, "learning_rate": 9.903604037416664e-06, "loss": 0.0026, "step": 38220 }, { "epoch": 0.6255420109629387, "grad_norm": 0.13211697340011597, "learning_rate": 9.903464470914317e-06, "loss": 0.0025, "step": 38230 }, { "epoch": 0.6257056369140146, "grad_norm": 0.038136664777994156, "learning_rate": 9.903324804434503e-06, "loss": 0.0025, "step": 38240 }, { "epoch": 0.6258692628650904, "grad_norm": 0.2067689448595047, "learning_rate": 9.903185037980064e-06, "loss": 0.0026, "step": 38250 }, { "epoch": 0.6260328888161663, "grad_norm": 0.10593418776988983, "learning_rate": 9.903045171553851e-06, "loss": 0.0032, "step": 38260 }, { "epoch": 0.626196514767242, "grad_norm": 0.04602759703993797, "learning_rate": 9.902905205158719e-06, "loss": 0.0024, "step": 38270 }, { "epoch": 0.6263601407183179, "grad_norm": 0.26689964532852173, "learning_rate": 9.90276513879752e-06, "loss": 0.0043, "step": 38280 }, { "epoch": 0.6265237666693938, "grad_norm": 0.16202399134635925, "learning_rate": 9.902624972473107e-06, "loss": 0.0032, "step": 38290 }, { "epoch": 0.6266873926204696, "grad_norm": 0.09605458378791809, "learning_rate": 9.902484706188341e-06, "loss": 0.0028, "step": 38300 }, { "epoch": 0.6268510185715455, "grad_norm": 0.23712900280952454, "learning_rate": 9.902344339946082e-06, "loss": 0.0026, "step": 38310 }, { "epoch": 0.6270146445226212, "grad_norm": 0.07135287672281265, "learning_rate": 9.90220387374919e-06, "loss": 0.005, "step": 38320 }, { "epoch": 0.6271782704736971, "grad_norm": 0.12095287442207336, "learning_rate": 9.902063307600532e-06, "loss": 0.0027, "step": 38330 }, { "epoch": 0.627341896424773, "grad_norm": 0.09234654158353806, "learning_rate": 9.901922641502972e-06, "loss": 0.0027, "step": 38340 }, { "epoch": 0.6275055223758488, "grad_norm": 0.043618861585855484, "learning_rate": 9.901781875459377e-06, "loss": 0.0023, "step": 38350 }, { "epoch": 0.6276691483269247, "grad_norm": 0.007149709854274988, "learning_rate": 9.90164100947262e-06, "loss": 0.0047, "step": 38360 }, { "epoch": 0.6278327742780004, "grad_norm": 0.09364354610443115, "learning_rate": 9.901500043545569e-06, "loss": 0.0024, "step": 38370 }, { "epoch": 0.6279964002290763, "grad_norm": 0.11993347108364105, "learning_rate": 9.901358977681103e-06, "loss": 0.0026, "step": 38380 }, { "epoch": 0.6281600261801522, "grad_norm": 0.19823050498962402, "learning_rate": 9.901217811882096e-06, "loss": 0.0022, "step": 38390 }, { "epoch": 0.628323652131228, "grad_norm": 0.1746380627155304, "learning_rate": 9.901076546151425e-06, "loss": 0.0023, "step": 38400 }, { "epoch": 0.6284872780823039, "grad_norm": 0.0576469823718071, "learning_rate": 9.90093518049197e-06, "loss": 0.0039, "step": 38410 }, { "epoch": 0.6286509040333796, "grad_norm": 0.11299474537372589, "learning_rate": 9.900793714906618e-06, "loss": 0.0049, "step": 38420 }, { "epoch": 0.6288145299844555, "grad_norm": 0.06734330207109451, "learning_rate": 9.900652149398249e-06, "loss": 0.0033, "step": 38430 }, { "epoch": 0.6289781559355314, "grad_norm": 0.004926139954477549, "learning_rate": 9.900510483969749e-06, "loss": 0.0029, "step": 38440 }, { "epoch": 0.6291417818866072, "grad_norm": 0.08655058592557907, "learning_rate": 9.90036871862401e-06, "loss": 0.0042, "step": 38450 }, { "epoch": 0.6293054078376831, "grad_norm": 0.04247899353504181, "learning_rate": 9.900226853363919e-06, "loss": 0.0027, "step": 38460 }, { "epoch": 0.6294690337887588, "grad_norm": 0.0811295285820961, "learning_rate": 9.90008488819237e-06, "loss": 0.0029, "step": 38470 }, { "epoch": 0.6296326597398347, "grad_norm": 0.033104993402957916, "learning_rate": 9.899942823112259e-06, "loss": 0.0027, "step": 38480 }, { "epoch": 0.6297962856909106, "grad_norm": 0.08117479085922241, "learning_rate": 9.89980065812648e-06, "loss": 0.0023, "step": 38490 }, { "epoch": 0.6299599116419864, "grad_norm": 0.050993360579013824, "learning_rate": 9.899658393237934e-06, "loss": 0.0026, "step": 38500 }, { "epoch": 0.6301235375930623, "grad_norm": 0.1152283325791359, "learning_rate": 9.899516028449519e-06, "loss": 0.0019, "step": 38510 }, { "epoch": 0.630287163544138, "grad_norm": 0.018274378031492233, "learning_rate": 9.899373563764138e-06, "loss": 0.0038, "step": 38520 }, { "epoch": 0.6304507894952139, "grad_norm": 0.04154553264379501, "learning_rate": 9.899230999184698e-06, "loss": 0.0023, "step": 38530 }, { "epoch": 0.6306144154462898, "grad_norm": 0.07623428106307983, "learning_rate": 9.899088334714106e-06, "loss": 0.0037, "step": 38540 }, { "epoch": 0.6307780413973656, "grad_norm": 0.14213111996650696, "learning_rate": 9.898945570355267e-06, "loss": 0.0024, "step": 38550 }, { "epoch": 0.6309416673484415, "grad_norm": 0.06029045209288597, "learning_rate": 9.898802706111095e-06, "loss": 0.0039, "step": 38560 }, { "epoch": 0.6311052932995173, "grad_norm": 0.14483420550823212, "learning_rate": 9.898659741984502e-06, "loss": 0.003, "step": 38570 }, { "epoch": 0.6312689192505931, "grad_norm": 0.2079179733991623, "learning_rate": 9.898516677978404e-06, "loss": 0.006, "step": 38580 }, { "epoch": 0.631432545201669, "grad_norm": 0.21550872921943665, "learning_rate": 9.898373514095717e-06, "loss": 0.0043, "step": 38590 }, { "epoch": 0.6315961711527448, "grad_norm": 0.0716831386089325, "learning_rate": 9.89823025033936e-06, "loss": 0.0025, "step": 38600 }, { "epoch": 0.6317597971038207, "grad_norm": 0.08701667189598083, "learning_rate": 9.898086886712252e-06, "loss": 0.0024, "step": 38610 }, { "epoch": 0.6319234230548965, "grad_norm": 0.10656403750181198, "learning_rate": 9.89794342321732e-06, "loss": 0.0022, "step": 38620 }, { "epoch": 0.6320870490059723, "grad_norm": 0.09030712395906448, "learning_rate": 9.897799859857486e-06, "loss": 0.0033, "step": 38630 }, { "epoch": 0.6322506749570482, "grad_norm": 0.04510744661092758, "learning_rate": 9.897656196635678e-06, "loss": 0.0028, "step": 38640 }, { "epoch": 0.632414300908124, "grad_norm": 0.096710205078125, "learning_rate": 9.897512433554828e-06, "loss": 0.0044, "step": 38650 }, { "epoch": 0.6325779268591999, "grad_norm": 0.1258869469165802, "learning_rate": 9.897368570617862e-06, "loss": 0.0025, "step": 38660 }, { "epoch": 0.6327415528102757, "grad_norm": 0.07378030568361282, "learning_rate": 9.897224607827717e-06, "loss": 0.0022, "step": 38670 }, { "epoch": 0.6329051787613516, "grad_norm": 0.13257750868797302, "learning_rate": 9.897080545187328e-06, "loss": 0.0027, "step": 38680 }, { "epoch": 0.6330688047124274, "grad_norm": 0.12059115618467331, "learning_rate": 9.896936382699628e-06, "loss": 0.0028, "step": 38690 }, { "epoch": 0.6332324306635032, "grad_norm": 0.03077046200633049, "learning_rate": 9.896792120367564e-06, "loss": 0.0032, "step": 38700 }, { "epoch": 0.6333960566145791, "grad_norm": 0.01565617509186268, "learning_rate": 9.896647758194071e-06, "loss": 0.0024, "step": 38710 }, { "epoch": 0.6335596825656549, "grad_norm": 0.016274888068437576, "learning_rate": 9.896503296182096e-06, "loss": 0.003, "step": 38720 }, { "epoch": 0.6337233085167308, "grad_norm": 0.03334270417690277, "learning_rate": 9.896358734334583e-06, "loss": 0.0023, "step": 38730 }, { "epoch": 0.6338869344678066, "grad_norm": 0.034932609647512436, "learning_rate": 9.896214072654478e-06, "loss": 0.0034, "step": 38740 }, { "epoch": 0.6340505604188824, "grad_norm": 0.1593017727136612, "learning_rate": 9.896069311144734e-06, "loss": 0.0028, "step": 38750 }, { "epoch": 0.6342141863699583, "grad_norm": 0.024335822090506554, "learning_rate": 9.8959244498083e-06, "loss": 0.0024, "step": 38760 }, { "epoch": 0.6343778123210341, "grad_norm": 0.1574372798204422, "learning_rate": 9.895779488648129e-06, "loss": 0.004, "step": 38770 }, { "epoch": 0.63454143827211, "grad_norm": 0.10908947139978409, "learning_rate": 9.89563442766718e-06, "loss": 0.0032, "step": 38780 }, { "epoch": 0.6347050642231858, "grad_norm": 0.026478221639990807, "learning_rate": 9.895489266868408e-06, "loss": 0.0017, "step": 38790 }, { "epoch": 0.6348686901742616, "grad_norm": 0.052796799689531326, "learning_rate": 9.895344006254773e-06, "loss": 0.0034, "step": 38800 }, { "epoch": 0.6350323161253375, "grad_norm": 0.13456590473651886, "learning_rate": 9.895198645829237e-06, "loss": 0.004, "step": 38810 }, { "epoch": 0.6351959420764133, "grad_norm": 0.10282973200082779, "learning_rate": 9.895053185594762e-06, "loss": 0.0039, "step": 38820 }, { "epoch": 0.6353595680274892, "grad_norm": 0.0715046226978302, "learning_rate": 9.89490762555432e-06, "loss": 0.0032, "step": 38830 }, { "epoch": 0.635523193978565, "grad_norm": 0.1260751634836197, "learning_rate": 9.894761965710871e-06, "loss": 0.0024, "step": 38840 }, { "epoch": 0.6356868199296408, "grad_norm": 0.028826827183365822, "learning_rate": 9.894616206067389e-06, "loss": 0.0038, "step": 38850 }, { "epoch": 0.6358504458807167, "grad_norm": 0.051343221217393875, "learning_rate": 9.894470346626846e-06, "loss": 0.0034, "step": 38860 }, { "epoch": 0.6360140718317925, "grad_norm": 0.06183408200740814, "learning_rate": 9.894324387392216e-06, "loss": 0.0037, "step": 38870 }, { "epoch": 0.6361776977828684, "grad_norm": 0.07487699389457703, "learning_rate": 9.894178328366473e-06, "loss": 0.003, "step": 38880 }, { "epoch": 0.6363413237339443, "grad_norm": 0.23917171359062195, "learning_rate": 9.894032169552596e-06, "loss": 0.0037, "step": 38890 }, { "epoch": 0.63650494968502, "grad_norm": 0.13589614629745483, "learning_rate": 9.893885910953564e-06, "loss": 0.0026, "step": 38900 }, { "epoch": 0.6366685756360959, "grad_norm": 0.2512050271034241, "learning_rate": 9.893739552572363e-06, "loss": 0.0032, "step": 38910 }, { "epoch": 0.6368322015871717, "grad_norm": 0.21717974543571472, "learning_rate": 9.893593094411973e-06, "loss": 0.0029, "step": 38920 }, { "epoch": 0.6369958275382476, "grad_norm": 0.07785177230834961, "learning_rate": 9.893446536475383e-06, "loss": 0.0039, "step": 38930 }, { "epoch": 0.6371594534893235, "grad_norm": 0.11839593946933746, "learning_rate": 9.89329987876558e-06, "loss": 0.0041, "step": 38940 }, { "epoch": 0.6373230794403992, "grad_norm": 0.06187107414007187, "learning_rate": 9.893153121285552e-06, "loss": 0.0032, "step": 38950 }, { "epoch": 0.6374867053914751, "grad_norm": 0.09474199265241623, "learning_rate": 9.893006264038294e-06, "loss": 0.0035, "step": 38960 }, { "epoch": 0.6376503313425509, "grad_norm": 0.14430944621562958, "learning_rate": 9.892859307026801e-06, "loss": 0.0068, "step": 38970 }, { "epoch": 0.6378139572936268, "grad_norm": 0.06610716134309769, "learning_rate": 9.892712250254067e-06, "loss": 0.0034, "step": 38980 }, { "epoch": 0.6379775832447027, "grad_norm": 0.03981900215148926, "learning_rate": 9.892565093723091e-06, "loss": 0.0031, "step": 38990 }, { "epoch": 0.6381412091957784, "grad_norm": 0.11940053105354309, "learning_rate": 9.892417837436874e-06, "loss": 0.0033, "step": 39000 }, { "epoch": 0.6383048351468543, "grad_norm": 0.06617771834135056, "learning_rate": 9.892270481398419e-06, "loss": 0.0058, "step": 39010 }, { "epoch": 0.6384684610979301, "grad_norm": 0.09681250154972076, "learning_rate": 9.892123025610728e-06, "loss": 0.0035, "step": 39020 }, { "epoch": 0.638632087049006, "grad_norm": 0.043846819549798965, "learning_rate": 9.89197547007681e-06, "loss": 0.0016, "step": 39030 }, { "epoch": 0.6387957130000819, "grad_norm": 0.15572737157344818, "learning_rate": 9.891827814799672e-06, "loss": 0.0046, "step": 39040 }, { "epoch": 0.6389593389511576, "grad_norm": 0.12031382322311401, "learning_rate": 9.891680059782326e-06, "loss": 0.0021, "step": 39050 }, { "epoch": 0.6391229649022335, "grad_norm": 0.342399001121521, "learning_rate": 9.891532205027783e-06, "loss": 0.0022, "step": 39060 }, { "epoch": 0.6392865908533093, "grad_norm": 0.06032680720090866, "learning_rate": 9.89138425053906e-06, "loss": 0.0027, "step": 39070 }, { "epoch": 0.6394502168043852, "grad_norm": 0.02009478583931923, "learning_rate": 9.891236196319172e-06, "loss": 0.0024, "step": 39080 }, { "epoch": 0.6396138427554611, "grad_norm": 0.15418677031993866, "learning_rate": 9.891088042371136e-06, "loss": 0.004, "step": 39090 }, { "epoch": 0.6397774687065368, "grad_norm": 0.04442372918128967, "learning_rate": 9.890939788697975e-06, "loss": 0.0039, "step": 39100 }, { "epoch": 0.6399410946576127, "grad_norm": 0.11542436480522156, "learning_rate": 9.89079143530271e-06, "loss": 0.004, "step": 39110 }, { "epoch": 0.6401047206086885, "grad_norm": 0.07558836787939072, "learning_rate": 9.890642982188372e-06, "loss": 0.0017, "step": 39120 }, { "epoch": 0.6402683465597644, "grad_norm": 0.17231884598731995, "learning_rate": 9.890494429357979e-06, "loss": 0.0033, "step": 39130 }, { "epoch": 0.6404319725108402, "grad_norm": 0.0734085962176323, "learning_rate": 9.890345776814565e-06, "loss": 0.0027, "step": 39140 }, { "epoch": 0.640595598461916, "grad_norm": 0.11366374045610428, "learning_rate": 9.89019702456116e-06, "loss": 0.0025, "step": 39150 }, { "epoch": 0.6407592244129919, "grad_norm": 0.07525376230478287, "learning_rate": 9.890048172600795e-06, "loss": 0.0028, "step": 39160 }, { "epoch": 0.6409228503640677, "grad_norm": 0.06043832749128342, "learning_rate": 9.889899220936507e-06, "loss": 0.0032, "step": 39170 }, { "epoch": 0.6410864763151436, "grad_norm": 0.021829726174473763, "learning_rate": 9.889750169571332e-06, "loss": 0.003, "step": 39180 }, { "epoch": 0.6412501022662194, "grad_norm": 0.20577505230903625, "learning_rate": 9.889601018508312e-06, "loss": 0.0029, "step": 39190 }, { "epoch": 0.6414137282172953, "grad_norm": 0.09856659919023514, "learning_rate": 9.889451767750484e-06, "loss": 0.0031, "step": 39200 }, { "epoch": 0.6415773541683711, "grad_norm": 0.06730175018310547, "learning_rate": 9.889302417300894e-06, "loss": 0.0048, "step": 39210 }, { "epoch": 0.6417409801194469, "grad_norm": 0.07655136287212372, "learning_rate": 9.889152967162586e-06, "loss": 0.0052, "step": 39220 }, { "epoch": 0.6419046060705228, "grad_norm": 0.2301681637763977, "learning_rate": 9.889003417338604e-06, "loss": 0.0046, "step": 39230 }, { "epoch": 0.6420682320215986, "grad_norm": 0.17960461974143982, "learning_rate": 9.888853767832003e-06, "loss": 0.0057, "step": 39240 }, { "epoch": 0.6422318579726745, "grad_norm": 0.10316240042448044, "learning_rate": 9.888704018645831e-06, "loss": 0.0044, "step": 39250 }, { "epoch": 0.6423954839237503, "grad_norm": 0.1299852877855301, "learning_rate": 9.888554169783143e-06, "loss": 0.0036, "step": 39260 }, { "epoch": 0.6425591098748261, "grad_norm": 0.17825517058372498, "learning_rate": 9.88840422124699e-06, "loss": 0.0018, "step": 39270 }, { "epoch": 0.642722735825902, "grad_norm": 0.030977703630924225, "learning_rate": 9.888254173040434e-06, "loss": 0.0031, "step": 39280 }, { "epoch": 0.6428863617769778, "grad_norm": 0.05662672594189644, "learning_rate": 9.888104025166533e-06, "loss": 0.0025, "step": 39290 }, { "epoch": 0.6430499877280537, "grad_norm": 0.2172756791114807, "learning_rate": 9.887953777628349e-06, "loss": 0.0033, "step": 39300 }, { "epoch": 0.6432136136791295, "grad_norm": 0.08615817874670029, "learning_rate": 9.887803430428944e-06, "loss": 0.0024, "step": 39310 }, { "epoch": 0.6433772396302053, "grad_norm": 0.2732553780078888, "learning_rate": 9.887652983571383e-06, "loss": 0.0035, "step": 39320 }, { "epoch": 0.6435408655812812, "grad_norm": 0.1442805677652359, "learning_rate": 9.887502437058735e-06, "loss": 0.0045, "step": 39330 }, { "epoch": 0.643704491532357, "grad_norm": 0.04210524633526802, "learning_rate": 9.887351790894069e-06, "loss": 0.0026, "step": 39340 }, { "epoch": 0.6438681174834329, "grad_norm": 0.28563523292541504, "learning_rate": 9.887201045080457e-06, "loss": 0.0036, "step": 39350 }, { "epoch": 0.6440317434345088, "grad_norm": 0.08012040704488754, "learning_rate": 9.887050199620972e-06, "loss": 0.0019, "step": 39360 }, { "epoch": 0.6441953693855845, "grad_norm": 0.11860110610723495, "learning_rate": 9.886899254518688e-06, "loss": 0.0031, "step": 39370 }, { "epoch": 0.6443589953366604, "grad_norm": 0.04950643330812454, "learning_rate": 9.886748209776687e-06, "loss": 0.003, "step": 39380 }, { "epoch": 0.6445226212877362, "grad_norm": 0.15256378054618835, "learning_rate": 9.886597065398043e-06, "loss": 0.0042, "step": 39390 }, { "epoch": 0.6446862472388121, "grad_norm": 0.36426305770874023, "learning_rate": 9.886445821385844e-06, "loss": 0.0018, "step": 39400 }, { "epoch": 0.644849873189888, "grad_norm": 0.2747892439365387, "learning_rate": 9.886294477743168e-06, "loss": 0.0049, "step": 39410 }, { "epoch": 0.6450134991409637, "grad_norm": 0.06749335676431656, "learning_rate": 9.886143034473104e-06, "loss": 0.0055, "step": 39420 }, { "epoch": 0.6451771250920396, "grad_norm": 0.05010202154517174, "learning_rate": 9.885991491578739e-06, "loss": 0.0025, "step": 39430 }, { "epoch": 0.6453407510431154, "grad_norm": 0.09193619340658188, "learning_rate": 9.885839849063163e-06, "loss": 0.0026, "step": 39440 }, { "epoch": 0.6455043769941913, "grad_norm": 0.0737474113702774, "learning_rate": 9.885688106929468e-06, "loss": 0.0028, "step": 39450 }, { "epoch": 0.6456680029452672, "grad_norm": 0.16302211582660675, "learning_rate": 9.885536265180748e-06, "loss": 0.0045, "step": 39460 }, { "epoch": 0.6458316288963429, "grad_norm": 0.12426792085170746, "learning_rate": 9.885384323820098e-06, "loss": 0.0066, "step": 39470 }, { "epoch": 0.6459952548474188, "grad_norm": 0.12219715118408203, "learning_rate": 9.885232282850616e-06, "loss": 0.0041, "step": 39480 }, { "epoch": 0.6461588807984946, "grad_norm": 0.01052816491574049, "learning_rate": 9.885080142275402e-06, "loss": 0.002, "step": 39490 }, { "epoch": 0.6463225067495705, "grad_norm": 0.08820316195487976, "learning_rate": 9.884927902097561e-06, "loss": 0.0033, "step": 39500 }, { "epoch": 0.6464861327006464, "grad_norm": 0.07392338663339615, "learning_rate": 9.884775562320192e-06, "loss": 0.0076, "step": 39510 }, { "epoch": 0.6466497586517221, "grad_norm": 0.058278489857912064, "learning_rate": 9.884623122946405e-06, "loss": 0.0028, "step": 39520 }, { "epoch": 0.646813384602798, "grad_norm": 0.020921604707837105, "learning_rate": 9.884470583979306e-06, "loss": 0.0032, "step": 39530 }, { "epoch": 0.6469770105538738, "grad_norm": 0.0479092113673687, "learning_rate": 9.884317945422007e-06, "loss": 0.003, "step": 39540 }, { "epoch": 0.6471406365049497, "grad_norm": 0.035598404705524445, "learning_rate": 9.884165207277619e-06, "loss": 0.0052, "step": 39550 }, { "epoch": 0.6473042624560256, "grad_norm": 0.21489793062210083, "learning_rate": 9.884012369549255e-06, "loss": 0.0037, "step": 39560 }, { "epoch": 0.6474678884071013, "grad_norm": 0.060852061957120895, "learning_rate": 9.883859432240032e-06, "loss": 0.0042, "step": 39570 }, { "epoch": 0.6476315143581772, "grad_norm": 0.2339639812707901, "learning_rate": 9.883706395353072e-06, "loss": 0.0025, "step": 39580 }, { "epoch": 0.647795140309253, "grad_norm": 0.06198565289378166, "learning_rate": 9.88355325889149e-06, "loss": 0.0038, "step": 39590 }, { "epoch": 0.6479587662603289, "grad_norm": 0.07975541055202484, "learning_rate": 9.88340002285841e-06, "loss": 0.0021, "step": 39600 }, { "epoch": 0.6481223922114048, "grad_norm": 0.08949963003396988, "learning_rate": 9.883246687256956e-06, "loss": 0.0034, "step": 39610 }, { "epoch": 0.6482860181624805, "grad_norm": 0.17046545445919037, "learning_rate": 9.883093252090257e-06, "loss": 0.0045, "step": 39620 }, { "epoch": 0.6484496441135564, "grad_norm": 0.06771198660135269, "learning_rate": 9.882939717361437e-06, "loss": 0.0033, "step": 39630 }, { "epoch": 0.6486132700646322, "grad_norm": 0.0738171637058258, "learning_rate": 9.882786083073632e-06, "loss": 0.0024, "step": 39640 }, { "epoch": 0.6487768960157081, "grad_norm": 0.11500179767608643, "learning_rate": 9.88263234922997e-06, "loss": 0.003, "step": 39650 }, { "epoch": 0.648940521966784, "grad_norm": 0.08991681039333344, "learning_rate": 9.882478515833587e-06, "loss": 0.002, "step": 39660 }, { "epoch": 0.6491041479178598, "grad_norm": 0.0860883966088295, "learning_rate": 9.88232458288762e-06, "loss": 0.0032, "step": 39670 }, { "epoch": 0.6492677738689356, "grad_norm": 0.04854914918541908, "learning_rate": 9.882170550395205e-06, "loss": 0.0034, "step": 39680 }, { "epoch": 0.6494313998200114, "grad_norm": 0.0935431495308876, "learning_rate": 9.882016418359486e-06, "loss": 0.0028, "step": 39690 }, { "epoch": 0.6495950257710873, "grad_norm": 0.07998931407928467, "learning_rate": 9.881862186783605e-06, "loss": 0.002, "step": 39700 }, { "epoch": 0.6497586517221632, "grad_norm": 0.17518122494220734, "learning_rate": 9.881707855670704e-06, "loss": 0.0029, "step": 39710 }, { "epoch": 0.649922277673239, "grad_norm": 0.20413893461227417, "learning_rate": 9.881553425023933e-06, "loss": 0.002, "step": 39720 }, { "epoch": 0.6500859036243148, "grad_norm": 0.0311913900077343, "learning_rate": 9.881398894846439e-06, "loss": 0.0024, "step": 39730 }, { "epoch": 0.6502495295753906, "grad_norm": 0.10733817517757416, "learning_rate": 9.881244265141374e-06, "loss": 0.0028, "step": 39740 }, { "epoch": 0.6504131555264665, "grad_norm": 0.06755837798118591, "learning_rate": 9.881089535911887e-06, "loss": 0.0047, "step": 39750 }, { "epoch": 0.6505767814775424, "grad_norm": 0.013972317799925804, "learning_rate": 9.880934707161138e-06, "loss": 0.0026, "step": 39760 }, { "epoch": 0.6507404074286182, "grad_norm": 0.20944489538669586, "learning_rate": 9.880779778892282e-06, "loss": 0.0025, "step": 39770 }, { "epoch": 0.650904033379694, "grad_norm": 0.0191704910248518, "learning_rate": 9.880624751108476e-06, "loss": 0.0024, "step": 39780 }, { "epoch": 0.6510676593307698, "grad_norm": 0.090189628303051, "learning_rate": 9.880469623812882e-06, "loss": 0.0036, "step": 39790 }, { "epoch": 0.6512312852818457, "grad_norm": 0.0786043331027031, "learning_rate": 9.880314397008663e-06, "loss": 0.0039, "step": 39800 }, { "epoch": 0.6513949112329216, "grad_norm": 0.13325923681259155, "learning_rate": 9.880159070698984e-06, "loss": 0.0029, "step": 39810 }, { "epoch": 0.6515585371839974, "grad_norm": 0.024909093976020813, "learning_rate": 9.880003644887013e-06, "loss": 0.005, "step": 39820 }, { "epoch": 0.6517221631350733, "grad_norm": 0.10260387510061264, "learning_rate": 9.879848119575917e-06, "loss": 0.0041, "step": 39830 }, { "epoch": 0.651885789086149, "grad_norm": 0.07352505624294281, "learning_rate": 9.879692494768868e-06, "loss": 0.0029, "step": 39840 }, { "epoch": 0.6520494150372249, "grad_norm": 0.0945623517036438, "learning_rate": 9.879536770469039e-06, "loss": 0.0024, "step": 39850 }, { "epoch": 0.6522130409883008, "grad_norm": 0.07430195808410645, "learning_rate": 9.879380946679605e-06, "loss": 0.0036, "step": 39860 }, { "epoch": 0.6523766669393766, "grad_norm": 0.20260505378246307, "learning_rate": 9.879225023403744e-06, "loss": 0.0057, "step": 39870 }, { "epoch": 0.6525402928904525, "grad_norm": 0.15074804425239563, "learning_rate": 9.879069000644635e-06, "loss": 0.0036, "step": 39880 }, { "epoch": 0.6527039188415282, "grad_norm": 0.19584380090236664, "learning_rate": 9.878912878405456e-06, "loss": 0.0041, "step": 39890 }, { "epoch": 0.6528675447926041, "grad_norm": 0.16875670850276947, "learning_rate": 9.878756656689395e-06, "loss": 0.0021, "step": 39900 }, { "epoch": 0.65303117074368, "grad_norm": 0.06408625841140747, "learning_rate": 9.878600335499636e-06, "loss": 0.0049, "step": 39910 }, { "epoch": 0.6531947966947558, "grad_norm": 0.13871942460536957, "learning_rate": 9.878443914839362e-06, "loss": 0.004, "step": 39920 }, { "epoch": 0.6533584226458317, "grad_norm": 0.1277013123035431, "learning_rate": 9.878287394711767e-06, "loss": 0.0023, "step": 39930 }, { "epoch": 0.6535220485969074, "grad_norm": 0.2226894199848175, "learning_rate": 9.878130775120041e-06, "loss": 0.0024, "step": 39940 }, { "epoch": 0.6536856745479833, "grad_norm": 0.08430884033441544, "learning_rate": 9.877974056067379e-06, "loss": 0.0041, "step": 39950 }, { "epoch": 0.6538493004990592, "grad_norm": 0.04147782549262047, "learning_rate": 9.877817237556972e-06, "loss": 0.0039, "step": 39960 }, { "epoch": 0.654012926450135, "grad_norm": 0.09764457494020462, "learning_rate": 9.87766031959202e-06, "loss": 0.0034, "step": 39970 }, { "epoch": 0.6541765524012109, "grad_norm": 0.06971295922994614, "learning_rate": 9.877503302175724e-06, "loss": 0.0044, "step": 39980 }, { "epoch": 0.6543401783522866, "grad_norm": 0.06916817277669907, "learning_rate": 9.877346185311283e-06, "loss": 0.0034, "step": 39990 }, { "epoch": 0.6545038043033625, "grad_norm": 0.130192831158638, "learning_rate": 9.8771889690019e-06, "loss": 0.0021, "step": 40000 }, { "epoch": 0.6546674302544384, "grad_norm": 0.09757371991872787, "learning_rate": 9.877031653250784e-06, "loss": 0.0027, "step": 40010 }, { "epoch": 0.6548310562055142, "grad_norm": 0.007362202741205692, "learning_rate": 9.87687423806114e-06, "loss": 0.0048, "step": 40020 }, { "epoch": 0.6549946821565901, "grad_norm": 0.07809104770421982, "learning_rate": 9.876716723436177e-06, "loss": 0.0032, "step": 40030 }, { "epoch": 0.6551583081076658, "grad_norm": 0.1492537260055542, "learning_rate": 9.876559109379108e-06, "loss": 0.0024, "step": 40040 }, { "epoch": 0.6553219340587417, "grad_norm": 0.05646463483572006, "learning_rate": 9.876401395893147e-06, "loss": 0.002, "step": 40050 }, { "epoch": 0.6554855600098175, "grad_norm": 0.14171238243579865, "learning_rate": 9.876243582981507e-06, "loss": 0.0027, "step": 40060 }, { "epoch": 0.6556491859608934, "grad_norm": 0.13456851243972778, "learning_rate": 9.876085670647409e-06, "loss": 0.0035, "step": 40070 }, { "epoch": 0.6558128119119693, "grad_norm": 0.03615734353661537, "learning_rate": 9.87592765889407e-06, "loss": 0.0025, "step": 40080 }, { "epoch": 0.655976437863045, "grad_norm": 0.04019275680184364, "learning_rate": 9.875769547724714e-06, "loss": 0.002, "step": 40090 }, { "epoch": 0.6561400638141209, "grad_norm": 0.10700186342000961, "learning_rate": 9.875611337142561e-06, "loss": 0.002, "step": 40100 }, { "epoch": 0.6563036897651967, "grad_norm": 0.20998942852020264, "learning_rate": 9.875453027150844e-06, "loss": 0.0024, "step": 40110 }, { "epoch": 0.6564673157162726, "grad_norm": 0.04590162634849548, "learning_rate": 9.875294617752782e-06, "loss": 0.0024, "step": 40120 }, { "epoch": 0.6566309416673485, "grad_norm": 0.03895952180027962, "learning_rate": 9.87513610895161e-06, "loss": 0.0053, "step": 40130 }, { "epoch": 0.6567945676184243, "grad_norm": 0.12128430604934692, "learning_rate": 9.87497750075056e-06, "loss": 0.0041, "step": 40140 }, { "epoch": 0.6569581935695001, "grad_norm": 0.31396007537841797, "learning_rate": 9.874818793152865e-06, "loss": 0.0094, "step": 40150 }, { "epoch": 0.6571218195205759, "grad_norm": 0.13139638304710388, "learning_rate": 9.874659986161758e-06, "loss": 0.0026, "step": 40160 }, { "epoch": 0.6572854454716518, "grad_norm": 0.15447881817817688, "learning_rate": 9.874501079780483e-06, "loss": 0.0031, "step": 40170 }, { "epoch": 0.6574490714227277, "grad_norm": 0.04102480411529541, "learning_rate": 9.874342074012275e-06, "loss": 0.0038, "step": 40180 }, { "epoch": 0.6576126973738035, "grad_norm": 0.3970644176006317, "learning_rate": 9.874182968860378e-06, "loss": 0.0027, "step": 40190 }, { "epoch": 0.6577763233248793, "grad_norm": 0.09813760966062546, "learning_rate": 9.874023764328034e-06, "loss": 0.0053, "step": 40200 }, { "epoch": 0.6579399492759551, "grad_norm": 0.17029523849487305, "learning_rate": 9.873864460418493e-06, "loss": 0.0043, "step": 40210 }, { "epoch": 0.658103575227031, "grad_norm": 0.10528775304555893, "learning_rate": 9.873705057134998e-06, "loss": 0.0048, "step": 40220 }, { "epoch": 0.6582672011781069, "grad_norm": 0.11468102782964706, "learning_rate": 9.873545554480806e-06, "loss": 0.0027, "step": 40230 }, { "epoch": 0.6584308271291827, "grad_norm": 0.10369814932346344, "learning_rate": 9.87338595245916e-06, "loss": 0.0038, "step": 40240 }, { "epoch": 0.6585944530802585, "grad_norm": 0.04351894557476044, "learning_rate": 9.873226251073322e-06, "loss": 0.0036, "step": 40250 }, { "epoch": 0.6587580790313343, "grad_norm": 0.11772704869508743, "learning_rate": 9.873066450326546e-06, "loss": 0.0035, "step": 40260 }, { "epoch": 0.6589217049824102, "grad_norm": 0.0429873950779438, "learning_rate": 9.872906550222087e-06, "loss": 0.0018, "step": 40270 }, { "epoch": 0.6590853309334861, "grad_norm": 0.05499604344367981, "learning_rate": 9.87274655076321e-06, "loss": 0.0042, "step": 40280 }, { "epoch": 0.6592489568845619, "grad_norm": 0.1000373363494873, "learning_rate": 9.872586451953175e-06, "loss": 0.0052, "step": 40290 }, { "epoch": 0.6594125828356378, "grad_norm": 0.02823615074157715, "learning_rate": 9.872426253795244e-06, "loss": 0.004, "step": 40300 }, { "epoch": 0.6595762087867135, "grad_norm": 0.05176930129528046, "learning_rate": 9.872265956292687e-06, "loss": 0.0027, "step": 40310 }, { "epoch": 0.6597398347377894, "grad_norm": 0.09854617714881897, "learning_rate": 9.87210555944877e-06, "loss": 0.0036, "step": 40320 }, { "epoch": 0.6599034606888653, "grad_norm": 0.13369372487068176, "learning_rate": 9.871945063266765e-06, "loss": 0.0056, "step": 40330 }, { "epoch": 0.6600670866399411, "grad_norm": 0.07178002595901489, "learning_rate": 9.871784467749944e-06, "loss": 0.0024, "step": 40340 }, { "epoch": 0.660230712591017, "grad_norm": 0.1779090166091919, "learning_rate": 9.87162377290158e-06, "loss": 0.0031, "step": 40350 }, { "epoch": 0.6603943385420927, "grad_norm": 0.24710668623447418, "learning_rate": 9.87146297872495e-06, "loss": 0.0052, "step": 40360 }, { "epoch": 0.6605579644931686, "grad_norm": 0.07842343300580978, "learning_rate": 9.871302085223336e-06, "loss": 0.004, "step": 40370 }, { "epoch": 0.6607215904442445, "grad_norm": 0.03234952688217163, "learning_rate": 9.871141092400012e-06, "loss": 0.0036, "step": 40380 }, { "epoch": 0.6608852163953203, "grad_norm": 0.056190021336078644, "learning_rate": 9.870980000258265e-06, "loss": 0.0033, "step": 40390 }, { "epoch": 0.6610488423463962, "grad_norm": 0.04890386015176773, "learning_rate": 9.87081880880138e-06, "loss": 0.0024, "step": 40400 }, { "epoch": 0.6612124682974719, "grad_norm": 0.07793574035167694, "learning_rate": 9.87065751803264e-06, "loss": 0.0025, "step": 40410 }, { "epoch": 0.6613760942485478, "grad_norm": 0.06824146211147308, "learning_rate": 9.870496127955335e-06, "loss": 0.0027, "step": 40420 }, { "epoch": 0.6615397201996237, "grad_norm": 0.10983072221279144, "learning_rate": 9.870334638572758e-06, "loss": 0.0025, "step": 40430 }, { "epoch": 0.6617033461506995, "grad_norm": 0.051976751536130905, "learning_rate": 9.8701730498882e-06, "loss": 0.0044, "step": 40440 }, { "epoch": 0.6618669721017754, "grad_norm": 0.08021211624145508, "learning_rate": 9.870011361904953e-06, "loss": 0.0025, "step": 40450 }, { "epoch": 0.6620305980528511, "grad_norm": 0.2322685569524765, "learning_rate": 9.869849574626317e-06, "loss": 0.0039, "step": 40460 }, { "epoch": 0.662194224003927, "grad_norm": 0.2386362999677658, "learning_rate": 9.86968768805559e-06, "loss": 0.0032, "step": 40470 }, { "epoch": 0.6623578499550029, "grad_norm": 0.18037950992584229, "learning_rate": 9.869525702196074e-06, "loss": 0.0018, "step": 40480 }, { "epoch": 0.6625214759060787, "grad_norm": 0.10438757389783859, "learning_rate": 9.869363617051069e-06, "loss": 0.0027, "step": 40490 }, { "epoch": 0.6626851018571546, "grad_norm": 0.058175310492515564, "learning_rate": 9.86920143262388e-06, "loss": 0.0025, "step": 40500 }, { "epoch": 0.6628487278082303, "grad_norm": 0.3740994930267334, "learning_rate": 9.869039148917816e-06, "loss": 0.0053, "step": 40510 }, { "epoch": 0.6630123537593062, "grad_norm": 0.2086252123117447, "learning_rate": 9.868876765936186e-06, "loss": 0.0028, "step": 40520 }, { "epoch": 0.6631759797103821, "grad_norm": 0.1848360300064087, "learning_rate": 9.868714283682297e-06, "loss": 0.0033, "step": 40530 }, { "epoch": 0.6633396056614579, "grad_norm": 0.1434709131717682, "learning_rate": 9.868551702159466e-06, "loss": 0.0025, "step": 40540 }, { "epoch": 0.6635032316125338, "grad_norm": 0.016255835071206093, "learning_rate": 9.868389021371007e-06, "loss": 0.0034, "step": 40550 }, { "epoch": 0.6636668575636095, "grad_norm": 0.1157134473323822, "learning_rate": 9.868226241320237e-06, "loss": 0.0033, "step": 40560 }, { "epoch": 0.6638304835146854, "grad_norm": 0.10432381927967072, "learning_rate": 9.868063362010472e-06, "loss": 0.0043, "step": 40570 }, { "epoch": 0.6639941094657613, "grad_norm": 0.027257010340690613, "learning_rate": 9.867900383445035e-06, "loss": 0.0037, "step": 40580 }, { "epoch": 0.6641577354168371, "grad_norm": 0.09254457801580429, "learning_rate": 9.867737305627251e-06, "loss": 0.0027, "step": 40590 }, { "epoch": 0.664321361367913, "grad_norm": 0.22699908912181854, "learning_rate": 9.867574128560442e-06, "loss": 0.0036, "step": 40600 }, { "epoch": 0.6644849873189888, "grad_norm": 0.12526603043079376, "learning_rate": 9.867410852247936e-06, "loss": 0.0043, "step": 40610 }, { "epoch": 0.6646486132700646, "grad_norm": 0.02070995792746544, "learning_rate": 9.867247476693064e-06, "loss": 0.0025, "step": 40620 }, { "epoch": 0.6648122392211405, "grad_norm": 0.011296287178993225, "learning_rate": 9.867084001899155e-06, "loss": 0.0031, "step": 40630 }, { "epoch": 0.6649758651722163, "grad_norm": 0.11782683432102203, "learning_rate": 9.86692042786954e-06, "loss": 0.0026, "step": 40640 }, { "epoch": 0.6651394911232922, "grad_norm": 0.09236036241054535, "learning_rate": 9.866756754607558e-06, "loss": 0.0037, "step": 40650 }, { "epoch": 0.665303117074368, "grad_norm": 0.09906945377588272, "learning_rate": 9.866592982116547e-06, "loss": 0.0035, "step": 40660 }, { "epoch": 0.6654667430254438, "grad_norm": 0.16777123510837555, "learning_rate": 9.866429110399842e-06, "loss": 0.0023, "step": 40670 }, { "epoch": 0.6656303689765197, "grad_norm": 0.07415185868740082, "learning_rate": 9.866265139460787e-06, "loss": 0.0041, "step": 40680 }, { "epoch": 0.6657939949275955, "grad_norm": 0.07174521684646606, "learning_rate": 9.866101069302723e-06, "loss": 0.0032, "step": 40690 }, { "epoch": 0.6659576208786714, "grad_norm": 0.0705026388168335, "learning_rate": 9.865936899928998e-06, "loss": 0.0026, "step": 40700 }, { "epoch": 0.6661212468297472, "grad_norm": 0.12775926291942596, "learning_rate": 9.865772631342956e-06, "loss": 0.0044, "step": 40710 }, { "epoch": 0.666284872780823, "grad_norm": 0.1549355536699295, "learning_rate": 9.86560826354795e-06, "loss": 0.0038, "step": 40720 }, { "epoch": 0.6664484987318989, "grad_norm": 0.04994513839483261, "learning_rate": 9.865443796547329e-06, "loss": 0.0031, "step": 40730 }, { "epoch": 0.6666121246829747, "grad_norm": 0.2746388912200928, "learning_rate": 9.865279230344448e-06, "loss": 0.0033, "step": 40740 }, { "epoch": 0.6667757506340506, "grad_norm": 0.012171761132776737, "learning_rate": 9.86511456494266e-06, "loss": 0.0021, "step": 40750 }, { "epoch": 0.6669393765851264, "grad_norm": 0.06325621157884598, "learning_rate": 9.864949800345325e-06, "loss": 0.002, "step": 40760 }, { "epoch": 0.6671030025362022, "grad_norm": 0.06368610262870789, "learning_rate": 9.864784936555799e-06, "loss": 0.0024, "step": 40770 }, { "epoch": 0.6672666284872781, "grad_norm": 0.25992268323898315, "learning_rate": 9.864619973577448e-06, "loss": 0.0036, "step": 40780 }, { "epoch": 0.6674302544383539, "grad_norm": 0.15245050191879272, "learning_rate": 9.86445491141363e-06, "loss": 0.0024, "step": 40790 }, { "epoch": 0.6675938803894298, "grad_norm": 0.11415904760360718, "learning_rate": 9.864289750067715e-06, "loss": 0.0025, "step": 40800 }, { "epoch": 0.6677575063405056, "grad_norm": 0.18931223452091217, "learning_rate": 9.86412448954307e-06, "loss": 0.0038, "step": 40810 }, { "epoch": 0.6679211322915815, "grad_norm": 0.013915250077843666, "learning_rate": 9.863959129843061e-06, "loss": 0.0011, "step": 40820 }, { "epoch": 0.6680847582426573, "grad_norm": 0.04009123146533966, "learning_rate": 9.863793670971065e-06, "loss": 0.0016, "step": 40830 }, { "epoch": 0.6682483841937331, "grad_norm": 0.11267693340778351, "learning_rate": 9.86362811293045e-06, "loss": 0.0029, "step": 40840 }, { "epoch": 0.668412010144809, "grad_norm": 0.14914385974407196, "learning_rate": 9.863462455724596e-06, "loss": 0.0077, "step": 40850 }, { "epoch": 0.6685756360958848, "grad_norm": 0.0654551088809967, "learning_rate": 9.863296699356876e-06, "loss": 0.0042, "step": 40860 }, { "epoch": 0.6687392620469607, "grad_norm": 0.12592904269695282, "learning_rate": 9.863130843830676e-06, "loss": 0.0043, "step": 40870 }, { "epoch": 0.6689028879980365, "grad_norm": 0.11217638850212097, "learning_rate": 9.862964889149372e-06, "loss": 0.0033, "step": 40880 }, { "epoch": 0.6690665139491123, "grad_norm": 0.04158490151166916, "learning_rate": 9.86279883531635e-06, "loss": 0.0025, "step": 40890 }, { "epoch": 0.6692301399001882, "grad_norm": 0.10852818936109543, "learning_rate": 9.862632682334994e-06, "loss": 0.0032, "step": 40900 }, { "epoch": 0.669393765851264, "grad_norm": 0.05358128994703293, "learning_rate": 9.862466430208694e-06, "loss": 0.0023, "step": 40910 }, { "epoch": 0.6695573918023399, "grad_norm": 0.4772360324859619, "learning_rate": 9.86230007894084e-06, "loss": 0.0027, "step": 40920 }, { "epoch": 0.6697210177534156, "grad_norm": 0.12135154008865356, "learning_rate": 9.862133628534822e-06, "loss": 0.0027, "step": 40930 }, { "epoch": 0.6698846437044915, "grad_norm": 0.06184634938836098, "learning_rate": 9.861967078994035e-06, "loss": 0.0058, "step": 40940 }, { "epoch": 0.6700482696555674, "grad_norm": 0.054414052516222, "learning_rate": 9.861800430321873e-06, "loss": 0.0028, "step": 40950 }, { "epoch": 0.6702118956066432, "grad_norm": 0.049953643232584, "learning_rate": 9.861633682521736e-06, "loss": 0.0087, "step": 40960 }, { "epoch": 0.6703755215577191, "grad_norm": 0.43331629037857056, "learning_rate": 9.861466835597022e-06, "loss": 0.0039, "step": 40970 }, { "epoch": 0.6705391475087948, "grad_norm": 0.03574656695127487, "learning_rate": 9.861299889551135e-06, "loss": 0.0017, "step": 40980 }, { "epoch": 0.6707027734598707, "grad_norm": 0.1379830688238144, "learning_rate": 9.861132844387476e-06, "loss": 0.0028, "step": 40990 }, { "epoch": 0.6708663994109466, "grad_norm": 0.074065662920475, "learning_rate": 9.860965700109453e-06, "loss": 0.0035, "step": 41000 }, { "epoch": 0.6710300253620224, "grad_norm": 0.06319969147443771, "learning_rate": 9.860798456720475e-06, "loss": 0.0018, "step": 41010 }, { "epoch": 0.6711936513130983, "grad_norm": 0.066305972635746, "learning_rate": 9.860631114223948e-06, "loss": 0.0026, "step": 41020 }, { "epoch": 0.671357277264174, "grad_norm": 0.04666053503751755, "learning_rate": 9.860463672623289e-06, "loss": 0.0028, "step": 41030 }, { "epoch": 0.6715209032152499, "grad_norm": 0.11729065328836441, "learning_rate": 9.860296131921909e-06, "loss": 0.0033, "step": 41040 }, { "epoch": 0.6716845291663258, "grad_norm": 0.26807671785354614, "learning_rate": 9.860128492123222e-06, "loss": 0.0043, "step": 41050 }, { "epoch": 0.6718481551174016, "grad_norm": 0.028942910954356194, "learning_rate": 9.859960753230651e-06, "loss": 0.0014, "step": 41060 }, { "epoch": 0.6720117810684775, "grad_norm": 0.018349051475524902, "learning_rate": 9.859792915247612e-06, "loss": 0.0028, "step": 41070 }, { "epoch": 0.6721754070195533, "grad_norm": 0.12847883999347687, "learning_rate": 9.859624978177529e-06, "loss": 0.0034, "step": 41080 }, { "epoch": 0.6723390329706291, "grad_norm": 0.019613200798630714, "learning_rate": 9.859456942023825e-06, "loss": 0.0018, "step": 41090 }, { "epoch": 0.672502658921705, "grad_norm": 0.094123475253582, "learning_rate": 9.859288806789929e-06, "loss": 0.0029, "step": 41100 }, { "epoch": 0.6726662848727808, "grad_norm": 0.3816341757774353, "learning_rate": 9.859120572479265e-06, "loss": 0.0037, "step": 41110 }, { "epoch": 0.6728299108238567, "grad_norm": 0.04508957266807556, "learning_rate": 9.858952239095265e-06, "loss": 0.0027, "step": 41120 }, { "epoch": 0.6729935367749325, "grad_norm": 0.09533313661813736, "learning_rate": 9.858783806641363e-06, "loss": 0.003, "step": 41130 }, { "epoch": 0.6731571627260083, "grad_norm": 0.09755261987447739, "learning_rate": 9.858615275120989e-06, "loss": 0.0022, "step": 41140 }, { "epoch": 0.6733207886770842, "grad_norm": 0.0316133089363575, "learning_rate": 9.858446644537583e-06, "loss": 0.0023, "step": 41150 }, { "epoch": 0.67348441462816, "grad_norm": 0.07658029347658157, "learning_rate": 9.858277914894581e-06, "loss": 0.0032, "step": 41160 }, { "epoch": 0.6736480405792359, "grad_norm": 0.06479310244321823, "learning_rate": 9.858109086195426e-06, "loss": 0.0026, "step": 41170 }, { "epoch": 0.6738116665303117, "grad_norm": 0.12080501019954681, "learning_rate": 9.857940158443558e-06, "loss": 0.003, "step": 41180 }, { "epoch": 0.6739752924813875, "grad_norm": 0.06928355991840363, "learning_rate": 9.85777113164242e-06, "loss": 0.0035, "step": 41190 }, { "epoch": 0.6741389184324634, "grad_norm": 0.22845478355884552, "learning_rate": 9.85760200579546e-06, "loss": 0.0036, "step": 41200 }, { "epoch": 0.6743025443835392, "grad_norm": 0.10625754296779633, "learning_rate": 9.857432780906128e-06, "loss": 0.0018, "step": 41210 }, { "epoch": 0.6744661703346151, "grad_norm": 0.2053966522216797, "learning_rate": 9.857263456977872e-06, "loss": 0.0038, "step": 41220 }, { "epoch": 0.6746297962856909, "grad_norm": 0.2431829422712326, "learning_rate": 9.857094034014146e-06, "loss": 0.0043, "step": 41230 }, { "epoch": 0.6747934222367667, "grad_norm": 0.08167760074138641, "learning_rate": 9.8569245120184e-06, "loss": 0.0041, "step": 41240 }, { "epoch": 0.6749570481878426, "grad_norm": 0.24822281301021576, "learning_rate": 9.856754890994097e-06, "loss": 0.0025, "step": 41250 }, { "epoch": 0.6751206741389184, "grad_norm": 0.27027231454849243, "learning_rate": 9.856585170944693e-06, "loss": 0.0075, "step": 41260 }, { "epoch": 0.6752843000899943, "grad_norm": 0.11435329169034958, "learning_rate": 9.856415351873645e-06, "loss": 0.0037, "step": 41270 }, { "epoch": 0.6754479260410701, "grad_norm": 0.1587599664926529, "learning_rate": 9.856245433784419e-06, "loss": 0.0022, "step": 41280 }, { "epoch": 0.675611551992146, "grad_norm": 0.11113511770963669, "learning_rate": 9.85607541668048e-06, "loss": 0.0051, "step": 41290 }, { "epoch": 0.6757751779432218, "grad_norm": 0.10176198184490204, "learning_rate": 9.855905300565293e-06, "loss": 0.0035, "step": 41300 }, { "epoch": 0.6759388038942976, "grad_norm": 0.09941761940717697, "learning_rate": 9.855735085442326e-06, "loss": 0.0036, "step": 41310 }, { "epoch": 0.6761024298453735, "grad_norm": 0.11637639999389648, "learning_rate": 9.85556477131505e-06, "loss": 0.0026, "step": 41320 }, { "epoch": 0.6762660557964493, "grad_norm": 0.034350473433732986, "learning_rate": 9.855394358186937e-06, "loss": 0.0034, "step": 41330 }, { "epoch": 0.6764296817475252, "grad_norm": 0.09146591275930405, "learning_rate": 9.855223846061466e-06, "loss": 0.0054, "step": 41340 }, { "epoch": 0.676593307698601, "grad_norm": 0.16053631901741028, "learning_rate": 9.855053234942108e-06, "loss": 0.0023, "step": 41350 }, { "epoch": 0.6767569336496768, "grad_norm": 0.090670645236969, "learning_rate": 9.854882524832343e-06, "loss": 0.0021, "step": 41360 }, { "epoch": 0.6769205596007527, "grad_norm": 0.08906329423189163, "learning_rate": 9.854711715735652e-06, "loss": 0.001, "step": 41370 }, { "epoch": 0.6770841855518285, "grad_norm": 0.27419915795326233, "learning_rate": 9.854540807655519e-06, "loss": 0.0038, "step": 41380 }, { "epoch": 0.6772478115029044, "grad_norm": 0.08886449784040451, "learning_rate": 9.854369800595427e-06, "loss": 0.0022, "step": 41390 }, { "epoch": 0.6774114374539802, "grad_norm": 0.01140960119664669, "learning_rate": 9.854198694558862e-06, "loss": 0.0026, "step": 41400 }, { "epoch": 0.677575063405056, "grad_norm": 0.1473616063594818, "learning_rate": 9.854027489549315e-06, "loss": 0.0047, "step": 41410 }, { "epoch": 0.6777386893561319, "grad_norm": 0.1326199769973755, "learning_rate": 9.853856185570276e-06, "loss": 0.003, "step": 41420 }, { "epoch": 0.6779023153072077, "grad_norm": 0.03985225036740303, "learning_rate": 9.853684782625238e-06, "loss": 0.0023, "step": 41430 }, { "epoch": 0.6780659412582836, "grad_norm": 0.038550008088350296, "learning_rate": 9.853513280717695e-06, "loss": 0.0023, "step": 41440 }, { "epoch": 0.6782295672093595, "grad_norm": 0.11070293933153152, "learning_rate": 9.853341679851142e-06, "loss": 0.0016, "step": 41450 }, { "epoch": 0.6783931931604352, "grad_norm": 0.13525570929050446, "learning_rate": 9.853169980029083e-06, "loss": 0.0038, "step": 41460 }, { "epoch": 0.6785568191115111, "grad_norm": 0.08288578689098358, "learning_rate": 9.852998181255013e-06, "loss": 0.0031, "step": 41470 }, { "epoch": 0.6787204450625869, "grad_norm": 0.14620912075042725, "learning_rate": 9.852826283532439e-06, "loss": 0.0028, "step": 41480 }, { "epoch": 0.6788840710136628, "grad_norm": 0.06953957676887512, "learning_rate": 9.852654286864862e-06, "loss": 0.0035, "step": 41490 }, { "epoch": 0.6790476969647387, "grad_norm": 0.1464170664548874, "learning_rate": 9.852482191255794e-06, "loss": 0.0046, "step": 41500 }, { "epoch": 0.6792113229158144, "grad_norm": 0.05585693567991257, "learning_rate": 9.85230999670874e-06, "loss": 0.0021, "step": 41510 }, { "epoch": 0.6793749488668903, "grad_norm": 0.14276130497455597, "learning_rate": 9.852137703227212e-06, "loss": 0.0022, "step": 41520 }, { "epoch": 0.6795385748179661, "grad_norm": 0.16480430960655212, "learning_rate": 9.851965310814722e-06, "loss": 0.0021, "step": 41530 }, { "epoch": 0.679702200769042, "grad_norm": 0.12121809273958206, "learning_rate": 9.851792819474785e-06, "loss": 0.0038, "step": 41540 }, { "epoch": 0.6798658267201179, "grad_norm": 0.18462353944778442, "learning_rate": 9.85162022921092e-06, "loss": 0.0027, "step": 41550 }, { "epoch": 0.6800294526711936, "grad_norm": 0.11869004368782043, "learning_rate": 9.851447540026645e-06, "loss": 0.0043, "step": 41560 }, { "epoch": 0.6801930786222695, "grad_norm": 0.07362282276153564, "learning_rate": 9.85127475192548e-06, "loss": 0.0034, "step": 41570 }, { "epoch": 0.6803567045733453, "grad_norm": 0.05356535688042641, "learning_rate": 9.851101864910949e-06, "loss": 0.0022, "step": 41580 }, { "epoch": 0.6805203305244212, "grad_norm": 0.1480821669101715, "learning_rate": 9.850928878986576e-06, "loss": 0.0025, "step": 41590 }, { "epoch": 0.6806839564754971, "grad_norm": 0.10870128870010376, "learning_rate": 9.850755794155891e-06, "loss": 0.003, "step": 41600 }, { "epoch": 0.6808475824265728, "grad_norm": 0.044383302330970764, "learning_rate": 9.850582610422417e-06, "loss": 0.0035, "step": 41610 }, { "epoch": 0.6810112083776487, "grad_norm": 0.20696525275707245, "learning_rate": 9.850409327789692e-06, "loss": 0.0034, "step": 41620 }, { "epoch": 0.6811748343287245, "grad_norm": 0.167523592710495, "learning_rate": 9.850235946261244e-06, "loss": 0.0026, "step": 41630 }, { "epoch": 0.6813384602798004, "grad_norm": 0.05519760027527809, "learning_rate": 9.850062465840611e-06, "loss": 0.0053, "step": 41640 }, { "epoch": 0.6815020862308763, "grad_norm": 0.047909289598464966, "learning_rate": 9.84988888653133e-06, "loss": 0.0024, "step": 41650 }, { "epoch": 0.681665712181952, "grad_norm": 0.024988509714603424, "learning_rate": 9.849715208336938e-06, "loss": 0.0028, "step": 41660 }, { "epoch": 0.6818293381330279, "grad_norm": 0.07336828112602234, "learning_rate": 9.849541431260979e-06, "loss": 0.0029, "step": 41670 }, { "epoch": 0.6819929640841037, "grad_norm": 0.08521151542663574, "learning_rate": 9.849367555306993e-06, "loss": 0.0031, "step": 41680 }, { "epoch": 0.6821565900351796, "grad_norm": 0.04987331107258797, "learning_rate": 9.849193580478526e-06, "loss": 0.0019, "step": 41690 }, { "epoch": 0.6823202159862555, "grad_norm": 0.05163633078336716, "learning_rate": 9.849019506779127e-06, "loss": 0.0017, "step": 41700 }, { "epoch": 0.6824838419373312, "grad_norm": 0.1420142501592636, "learning_rate": 9.848845334212344e-06, "loss": 0.0025, "step": 41710 }, { "epoch": 0.6826474678884071, "grad_norm": 0.07418406754732132, "learning_rate": 9.84867106278173e-06, "loss": 0.0019, "step": 41720 }, { "epoch": 0.6828110938394829, "grad_norm": 0.013063437305390835, "learning_rate": 9.848496692490835e-06, "loss": 0.0018, "step": 41730 }, { "epoch": 0.6829747197905588, "grad_norm": 0.07301737368106842, "learning_rate": 9.848322223343217e-06, "loss": 0.0024, "step": 41740 }, { "epoch": 0.6831383457416347, "grad_norm": 0.0951089933514595, "learning_rate": 9.84814765534243e-06, "loss": 0.0057, "step": 41750 }, { "epoch": 0.6833019716927105, "grad_norm": 0.04238731041550636, "learning_rate": 9.847972988492038e-06, "loss": 0.0016, "step": 41760 }, { "epoch": 0.6834655976437863, "grad_norm": 0.07355024665594101, "learning_rate": 9.8477982227956e-06, "loss": 0.0029, "step": 41770 }, { "epoch": 0.6836292235948621, "grad_norm": 0.06040283665060997, "learning_rate": 9.847623358256678e-06, "loss": 0.0021, "step": 41780 }, { "epoch": 0.683792849545938, "grad_norm": 0.10334978252649307, "learning_rate": 9.847448394878838e-06, "loss": 0.0044, "step": 41790 }, { "epoch": 0.6839564754970138, "grad_norm": 0.2523374557495117, "learning_rate": 9.847273332665648e-06, "loss": 0.0031, "step": 41800 }, { "epoch": 0.6841201014480897, "grad_norm": 0.010403482243418694, "learning_rate": 9.847098171620678e-06, "loss": 0.0027, "step": 41810 }, { "epoch": 0.6842837273991655, "grad_norm": 0.035346295684576035, "learning_rate": 9.846922911747498e-06, "loss": 0.0042, "step": 41820 }, { "epoch": 0.6844473533502413, "grad_norm": 0.09193457663059235, "learning_rate": 9.846747553049682e-06, "loss": 0.0034, "step": 41830 }, { "epoch": 0.6846109793013172, "grad_norm": 0.07155217975378036, "learning_rate": 9.846572095530807e-06, "loss": 0.0033, "step": 41840 }, { "epoch": 0.684774605252393, "grad_norm": 0.10444560647010803, "learning_rate": 9.846396539194447e-06, "loss": 0.0024, "step": 41850 }, { "epoch": 0.6849382312034689, "grad_norm": 0.024706212803721428, "learning_rate": 9.846220884044183e-06, "loss": 0.0024, "step": 41860 }, { "epoch": 0.6851018571545447, "grad_norm": 0.21703557670116425, "learning_rate": 9.846045130083597e-06, "loss": 0.0017, "step": 41870 }, { "epoch": 0.6852654831056205, "grad_norm": 0.05481688305735588, "learning_rate": 9.845869277316273e-06, "loss": 0.0058, "step": 41880 }, { "epoch": 0.6854291090566964, "grad_norm": 0.23019297420978546, "learning_rate": 9.845693325745795e-06, "loss": 0.0028, "step": 41890 }, { "epoch": 0.6855927350077722, "grad_norm": 0.28986406326293945, "learning_rate": 9.845517275375752e-06, "loss": 0.004, "step": 41900 }, { "epoch": 0.6857563609588481, "grad_norm": 0.23879460990428925, "learning_rate": 9.845341126209733e-06, "loss": 0.0026, "step": 41910 }, { "epoch": 0.685919986909924, "grad_norm": 0.03717726096510887, "learning_rate": 9.845164878251327e-06, "loss": 0.0045, "step": 41920 }, { "epoch": 0.6860836128609997, "grad_norm": 0.17686645686626434, "learning_rate": 9.844988531504132e-06, "loss": 0.0036, "step": 41930 }, { "epoch": 0.6862472388120756, "grad_norm": 0.11270167678594589, "learning_rate": 9.84481208597174e-06, "loss": 0.0047, "step": 41940 }, { "epoch": 0.6864108647631514, "grad_norm": 0.16190236806869507, "learning_rate": 9.844635541657752e-06, "loss": 0.0026, "step": 41950 }, { "epoch": 0.6865744907142273, "grad_norm": 0.05602660030126572, "learning_rate": 9.844458898565763e-06, "loss": 0.0033, "step": 41960 }, { "epoch": 0.6867381166653032, "grad_norm": 0.14890341460704803, "learning_rate": 9.844282156699378e-06, "loss": 0.0029, "step": 41970 }, { "epoch": 0.6869017426163789, "grad_norm": 0.04054934158921242, "learning_rate": 9.8441053160622e-06, "loss": 0.0023, "step": 41980 }, { "epoch": 0.6870653685674548, "grad_norm": 0.24899771809577942, "learning_rate": 9.843928376657836e-06, "loss": 0.0051, "step": 41990 }, { "epoch": 0.6872289945185306, "grad_norm": 0.05639079958200455, "learning_rate": 9.84375133848989e-06, "loss": 0.0029, "step": 42000 }, { "epoch": 0.6873926204696065, "grad_norm": 0.16043952107429504, "learning_rate": 9.843574201561973e-06, "loss": 0.0035, "step": 42010 }, { "epoch": 0.6875562464206824, "grad_norm": 0.12931540608406067, "learning_rate": 9.843396965877698e-06, "loss": 0.0021, "step": 42020 }, { "epoch": 0.6877198723717581, "grad_norm": 0.05526356026530266, "learning_rate": 9.84321963144068e-06, "loss": 0.002, "step": 42030 }, { "epoch": 0.687883498322834, "grad_norm": 0.08453048020601273, "learning_rate": 9.843042198254531e-06, "loss": 0.0048, "step": 42040 }, { "epoch": 0.6880471242739098, "grad_norm": 0.1850995421409607, "learning_rate": 9.84286466632287e-06, "loss": 0.0037, "step": 42050 }, { "epoch": 0.6882107502249857, "grad_norm": 0.0299760103225708, "learning_rate": 9.842687035649319e-06, "loss": 0.0028, "step": 42060 }, { "epoch": 0.6883743761760616, "grad_norm": 0.07078180462121964, "learning_rate": 9.842509306237495e-06, "loss": 0.0034, "step": 42070 }, { "epoch": 0.6885380021271373, "grad_norm": 0.0470513179898262, "learning_rate": 9.842331478091027e-06, "loss": 0.004, "step": 42080 }, { "epoch": 0.6887016280782132, "grad_norm": 0.12903571128845215, "learning_rate": 9.842153551213538e-06, "loss": 0.0039, "step": 42090 }, { "epoch": 0.688865254029289, "grad_norm": 0.04146236181259155, "learning_rate": 9.841975525608655e-06, "loss": 0.0017, "step": 42100 }, { "epoch": 0.6890288799803649, "grad_norm": 9.635364532470703, "learning_rate": 9.841797401280012e-06, "loss": 0.0074, "step": 42110 }, { "epoch": 0.6891925059314408, "grad_norm": 0.12732818722724915, "learning_rate": 9.841619178231234e-06, "loss": 0.0026, "step": 42120 }, { "epoch": 0.6893561318825165, "grad_norm": 0.2447766810655594, "learning_rate": 9.84144085646596e-06, "loss": 0.0047, "step": 42130 }, { "epoch": 0.6895197578335924, "grad_norm": 0.14923837780952454, "learning_rate": 9.841262435987825e-06, "loss": 0.0046, "step": 42140 }, { "epoch": 0.6896833837846682, "grad_norm": 0.08472929149866104, "learning_rate": 9.841083916800468e-06, "loss": 0.0023, "step": 42150 }, { "epoch": 0.6898470097357441, "grad_norm": 0.11935146152973175, "learning_rate": 9.840905298907523e-06, "loss": 0.0052, "step": 42160 }, { "epoch": 0.69001063568682, "grad_norm": 0.10566152632236481, "learning_rate": 9.840726582312639e-06, "loss": 0.0033, "step": 42170 }, { "epoch": 0.6901742616378957, "grad_norm": 0.11551082134246826, "learning_rate": 9.840547767019456e-06, "loss": 0.0018, "step": 42180 }, { "epoch": 0.6903378875889716, "grad_norm": 0.15986007452011108, "learning_rate": 9.84036885303162e-06, "loss": 0.0052, "step": 42190 }, { "epoch": 0.6905015135400474, "grad_norm": 0.1138385757803917, "learning_rate": 9.84018984035278e-06, "loss": 0.0037, "step": 42200 }, { "epoch": 0.6906651394911233, "grad_norm": 0.11291395872831345, "learning_rate": 9.840010728986586e-06, "loss": 0.0027, "step": 42210 }, { "epoch": 0.6908287654421992, "grad_norm": 0.1384618878364563, "learning_rate": 9.83983151893669e-06, "loss": 0.0026, "step": 42220 }, { "epoch": 0.690992391393275, "grad_norm": 0.06459318101406097, "learning_rate": 9.839652210206746e-06, "loss": 0.0037, "step": 42230 }, { "epoch": 0.6911560173443508, "grad_norm": 0.12383589893579483, "learning_rate": 9.839472802800407e-06, "loss": 0.0018, "step": 42240 }, { "epoch": 0.6913196432954266, "grad_norm": 0.04691753163933754, "learning_rate": 9.839293296721335e-06, "loss": 0.0014, "step": 42250 }, { "epoch": 0.6914832692465025, "grad_norm": 0.2157626748085022, "learning_rate": 9.839113691973188e-06, "loss": 0.0038, "step": 42260 }, { "epoch": 0.6916468951975784, "grad_norm": 0.03732483834028244, "learning_rate": 9.838933988559629e-06, "loss": 0.0048, "step": 42270 }, { "epoch": 0.6918105211486542, "grad_norm": 0.0637301653623581, "learning_rate": 9.838754186484321e-06, "loss": 0.0048, "step": 42280 }, { "epoch": 0.69197414709973, "grad_norm": 0.2086525410413742, "learning_rate": 9.83857428575093e-06, "loss": 0.0072, "step": 42290 }, { "epoch": 0.6921377730508058, "grad_norm": 0.11707808822393417, "learning_rate": 9.838394286363124e-06, "loss": 0.0031, "step": 42300 }, { "epoch": 0.6923013990018817, "grad_norm": 0.05959474295377731, "learning_rate": 9.838214188324576e-06, "loss": 0.0027, "step": 42310 }, { "epoch": 0.6924650249529576, "grad_norm": 0.043860726058483124, "learning_rate": 9.83803399163895e-06, "loss": 0.0029, "step": 42320 }, { "epoch": 0.6926286509040334, "grad_norm": 0.18076156079769135, "learning_rate": 9.837853696309932e-06, "loss": 0.0028, "step": 42330 }, { "epoch": 0.6927922768551092, "grad_norm": 0.06909358501434326, "learning_rate": 9.837673302341187e-06, "loss": 0.0028, "step": 42340 }, { "epoch": 0.692955902806185, "grad_norm": 0.017848331481218338, "learning_rate": 9.8374928097364e-06, "loss": 0.0028, "step": 42350 }, { "epoch": 0.6931195287572609, "grad_norm": 0.01810034178197384, "learning_rate": 9.837312218499248e-06, "loss": 0.0035, "step": 42360 }, { "epoch": 0.6932831547083368, "grad_norm": 0.07195191830396652, "learning_rate": 9.837131528633414e-06, "loss": 0.003, "step": 42370 }, { "epoch": 0.6934467806594126, "grad_norm": 0.1767512857913971, "learning_rate": 9.83695074014258e-06, "loss": 0.0023, "step": 42380 }, { "epoch": 0.6936104066104885, "grad_norm": 0.04846373572945595, "learning_rate": 9.836769853030438e-06, "loss": 0.0022, "step": 42390 }, { "epoch": 0.6937740325615642, "grad_norm": 0.32237508893013, "learning_rate": 9.836588867300669e-06, "loss": 0.0032, "step": 42400 }, { "epoch": 0.6939376585126401, "grad_norm": 0.03659368306398392, "learning_rate": 9.836407782956967e-06, "loss": 0.0018, "step": 42410 }, { "epoch": 0.694101284463716, "grad_norm": 0.16458043456077576, "learning_rate": 9.836226600003025e-06, "loss": 0.0023, "step": 42420 }, { "epoch": 0.6942649104147918, "grad_norm": 0.11829166114330292, "learning_rate": 9.836045318442533e-06, "loss": 0.003, "step": 42430 }, { "epoch": 0.6944285363658677, "grad_norm": 0.03232857212424278, "learning_rate": 9.835863938279193e-06, "loss": 0.0021, "step": 42440 }, { "epoch": 0.6945921623169434, "grad_norm": 0.07435762137174606, "learning_rate": 9.835682459516697e-06, "loss": 0.0027, "step": 42450 }, { "epoch": 0.6947557882680193, "grad_norm": 0.04996451735496521, "learning_rate": 9.83550088215875e-06, "loss": 0.0025, "step": 42460 }, { "epoch": 0.6949194142190952, "grad_norm": 0.0981188639998436, "learning_rate": 9.835319206209054e-06, "loss": 0.0044, "step": 42470 }, { "epoch": 0.695083040170171, "grad_norm": 0.11858845502138138, "learning_rate": 9.835137431671308e-06, "loss": 0.0042, "step": 42480 }, { "epoch": 0.6952466661212469, "grad_norm": 0.0390016995370388, "learning_rate": 9.834955558549224e-06, "loss": 0.0044, "step": 42490 }, { "epoch": 0.6954102920723226, "grad_norm": 0.19089801609516144, "learning_rate": 9.834773586846509e-06, "loss": 0.002, "step": 42500 }, { "epoch": 0.6955739180233985, "grad_norm": 0.0345117449760437, "learning_rate": 9.83459151656687e-06, "loss": 0.0028, "step": 42510 }, { "epoch": 0.6957375439744744, "grad_norm": 0.06933096051216125, "learning_rate": 9.834409347714025e-06, "loss": 0.0022, "step": 42520 }, { "epoch": 0.6959011699255502, "grad_norm": 0.09161041676998138, "learning_rate": 9.834227080291681e-06, "loss": 0.0016, "step": 42530 }, { "epoch": 0.6960647958766261, "grad_norm": 0.09689223766326904, "learning_rate": 9.834044714303562e-06, "loss": 0.0025, "step": 42540 }, { "epoch": 0.6962284218277018, "grad_norm": 0.11612505465745926, "learning_rate": 9.83386224975338e-06, "loss": 0.0054, "step": 42550 }, { "epoch": 0.6963920477787777, "grad_norm": 0.05863359570503235, "learning_rate": 9.833679686644858e-06, "loss": 0.0023, "step": 42560 }, { "epoch": 0.6965556737298536, "grad_norm": 0.0496952086687088, "learning_rate": 9.833497024981719e-06, "loss": 0.0026, "step": 42570 }, { "epoch": 0.6967192996809294, "grad_norm": 0.07727567851543427, "learning_rate": 9.833314264767686e-06, "loss": 0.0044, "step": 42580 }, { "epoch": 0.6968829256320053, "grad_norm": 0.056543659418821335, "learning_rate": 9.833131406006484e-06, "loss": 0.0025, "step": 42590 }, { "epoch": 0.697046551583081, "grad_norm": 0.15698623657226562, "learning_rate": 9.832948448701846e-06, "loss": 0.0019, "step": 42600 }, { "epoch": 0.6972101775341569, "grad_norm": 0.04090658202767372, "learning_rate": 9.832765392857497e-06, "loss": 0.0033, "step": 42610 }, { "epoch": 0.6973738034852328, "grad_norm": 0.12148510664701462, "learning_rate": 9.832582238477173e-06, "loss": 0.0017, "step": 42620 }, { "epoch": 0.6975374294363086, "grad_norm": 0.23208025097846985, "learning_rate": 9.832398985564607e-06, "loss": 0.0027, "step": 42630 }, { "epoch": 0.6977010553873845, "grad_norm": 0.04539055377244949, "learning_rate": 9.832215634123537e-06, "loss": 0.0025, "step": 42640 }, { "epoch": 0.6978646813384602, "grad_norm": 0.07503404468297958, "learning_rate": 9.832032184157696e-06, "loss": 0.0046, "step": 42650 }, { "epoch": 0.6980283072895361, "grad_norm": 0.17545457184314728, "learning_rate": 9.831848635670831e-06, "loss": 0.0046, "step": 42660 }, { "epoch": 0.6981919332406119, "grad_norm": 0.1363619714975357, "learning_rate": 9.831664988666683e-06, "loss": 0.0024, "step": 42670 }, { "epoch": 0.6983555591916878, "grad_norm": 0.07460838556289673, "learning_rate": 9.831481243148992e-06, "loss": 0.0025, "step": 42680 }, { "epoch": 0.6985191851427637, "grad_norm": 0.048054736107587814, "learning_rate": 9.831297399121508e-06, "loss": 0.0028, "step": 42690 }, { "epoch": 0.6986828110938395, "grad_norm": 0.06863340735435486, "learning_rate": 9.831113456587981e-06, "loss": 0.0016, "step": 42700 }, { "epoch": 0.6988464370449153, "grad_norm": 0.14542849361896515, "learning_rate": 9.830929415552157e-06, "loss": 0.0035, "step": 42710 }, { "epoch": 0.6990100629959911, "grad_norm": 0.042103566229343414, "learning_rate": 9.830745276017792e-06, "loss": 0.0019, "step": 42720 }, { "epoch": 0.699173688947067, "grad_norm": 0.14125099778175354, "learning_rate": 9.83056103798864e-06, "loss": 0.0034, "step": 42730 }, { "epoch": 0.6993373148981429, "grad_norm": 0.014587906189262867, "learning_rate": 9.830376701468456e-06, "loss": 0.0028, "step": 42740 }, { "epoch": 0.6995009408492187, "grad_norm": 0.0914643183350563, "learning_rate": 9.830192266460998e-06, "loss": 0.0021, "step": 42750 }, { "epoch": 0.6996645668002945, "grad_norm": 0.19832000136375427, "learning_rate": 9.830007732970028e-06, "loss": 0.0023, "step": 42760 }, { "epoch": 0.6998281927513703, "grad_norm": 0.06493548303842545, "learning_rate": 9.829823100999309e-06, "loss": 0.0027, "step": 42770 }, { "epoch": 0.6999918187024462, "grad_norm": 0.12377274036407471, "learning_rate": 9.829638370552605e-06, "loss": 0.0032, "step": 42780 }, { "epoch": 0.7001554446535221, "grad_norm": 0.04548317566514015, "learning_rate": 9.82945354163368e-06, "loss": 0.0016, "step": 42790 }, { "epoch": 0.7003190706045979, "grad_norm": 0.038685888051986694, "learning_rate": 9.829268614246306e-06, "loss": 0.0023, "step": 42800 }, { "epoch": 0.7004826965556737, "grad_norm": 0.21114659309387207, "learning_rate": 9.829083588394253e-06, "loss": 0.0027, "step": 42810 }, { "epoch": 0.7006463225067495, "grad_norm": 0.06751938909292221, "learning_rate": 9.828898464081291e-06, "loss": 0.0035, "step": 42820 }, { "epoch": 0.7008099484578254, "grad_norm": 0.10093922168016434, "learning_rate": 9.828713241311196e-06, "loss": 0.0022, "step": 42830 }, { "epoch": 0.7009735744089013, "grad_norm": 0.16676314175128937, "learning_rate": 9.828527920087746e-06, "loss": 0.0024, "step": 42840 }, { "epoch": 0.7011372003599771, "grad_norm": 0.012430752627551556, "learning_rate": 9.828342500414717e-06, "loss": 0.0031, "step": 42850 }, { "epoch": 0.701300826311053, "grad_norm": 0.1617567241191864, "learning_rate": 9.828156982295893e-06, "loss": 0.0029, "step": 42860 }, { "epoch": 0.7014644522621287, "grad_norm": 0.04069804400205612, "learning_rate": 9.827971365735051e-06, "loss": 0.0034, "step": 42870 }, { "epoch": 0.7016280782132046, "grad_norm": 0.2438546121120453, "learning_rate": 9.827785650735982e-06, "loss": 0.004, "step": 42880 }, { "epoch": 0.7017917041642805, "grad_norm": 0.09727134555578232, "learning_rate": 9.827599837302468e-06, "loss": 0.0025, "step": 42890 }, { "epoch": 0.7019553301153563, "grad_norm": 0.07649528235197067, "learning_rate": 9.827413925438299e-06, "loss": 0.0036, "step": 42900 }, { "epoch": 0.7021189560664322, "grad_norm": 0.12396731227636337, "learning_rate": 9.827227915147266e-06, "loss": 0.0024, "step": 42910 }, { "epoch": 0.7022825820175079, "grad_norm": 0.07873599976301193, "learning_rate": 9.82704180643316e-06, "loss": 0.0029, "step": 42920 }, { "epoch": 0.7024462079685838, "grad_norm": 0.08151212334632874, "learning_rate": 9.826855599299779e-06, "loss": 0.0019, "step": 42930 }, { "epoch": 0.7026098339196597, "grad_norm": 0.08233390003442764, "learning_rate": 9.826669293750915e-06, "loss": 0.0034, "step": 42940 }, { "epoch": 0.7027734598707355, "grad_norm": 0.031621240079402924, "learning_rate": 9.826482889790371e-06, "loss": 0.0033, "step": 42950 }, { "epoch": 0.7029370858218114, "grad_norm": 0.060222528874874115, "learning_rate": 9.826296387421945e-06, "loss": 0.0033, "step": 42960 }, { "epoch": 0.7031007117728871, "grad_norm": 0.14794889092445374, "learning_rate": 9.826109786649441e-06, "loss": 0.0023, "step": 42970 }, { "epoch": 0.703264337723963, "grad_norm": 0.07250554859638214, "learning_rate": 9.825923087476662e-06, "loss": 0.0033, "step": 42980 }, { "epoch": 0.7034279636750389, "grad_norm": 0.02376159094274044, "learning_rate": 9.825736289907414e-06, "loss": 0.0048, "step": 42990 }, { "epoch": 0.7035915896261147, "grad_norm": 0.13682201504707336, "learning_rate": 9.82554939394551e-06, "loss": 0.0015, "step": 43000 }, { "epoch": 0.7037552155771906, "grad_norm": 0.19004955887794495, "learning_rate": 9.825362399594756e-06, "loss": 0.0031, "step": 43010 }, { "epoch": 0.7039188415282663, "grad_norm": 0.06614803522825241, "learning_rate": 9.825175306858968e-06, "loss": 0.0022, "step": 43020 }, { "epoch": 0.7040824674793422, "grad_norm": 0.07687593251466751, "learning_rate": 9.824988115741959e-06, "loss": 0.002, "step": 43030 }, { "epoch": 0.7042460934304181, "grad_norm": 0.08164342492818832, "learning_rate": 9.824800826247544e-06, "loss": 0.0051, "step": 43040 }, { "epoch": 0.7044097193814939, "grad_norm": 0.03556409478187561, "learning_rate": 9.824613438379545e-06, "loss": 0.0028, "step": 43050 }, { "epoch": 0.7045733453325698, "grad_norm": 0.04737414792180061, "learning_rate": 9.824425952141781e-06, "loss": 0.0052, "step": 43060 }, { "epoch": 0.7047369712836455, "grad_norm": 0.060558971017599106, "learning_rate": 9.824238367538074e-06, "loss": 0.003, "step": 43070 }, { "epoch": 0.7049005972347214, "grad_norm": 0.2587260901927948, "learning_rate": 9.824050684572251e-06, "loss": 0.0026, "step": 43080 }, { "epoch": 0.7050642231857973, "grad_norm": 0.04088201746344566, "learning_rate": 9.823862903248136e-06, "loss": 0.0028, "step": 43090 }, { "epoch": 0.7052278491368731, "grad_norm": 0.07648684084415436, "learning_rate": 9.82367502356956e-06, "loss": 0.0032, "step": 43100 }, { "epoch": 0.705391475087949, "grad_norm": 0.35123857855796814, "learning_rate": 9.823487045540352e-06, "loss": 0.0034, "step": 43110 }, { "epoch": 0.7055551010390247, "grad_norm": 0.33388805389404297, "learning_rate": 9.823298969164345e-06, "loss": 0.0032, "step": 43120 }, { "epoch": 0.7057187269901006, "grad_norm": 0.009662275202572346, "learning_rate": 9.823110794445374e-06, "loss": 0.0033, "step": 43130 }, { "epoch": 0.7058823529411765, "grad_norm": 0.08940360695123672, "learning_rate": 9.822922521387277e-06, "loss": 0.0019, "step": 43140 }, { "epoch": 0.7060459788922523, "grad_norm": 0.036306556314229965, "learning_rate": 9.822734149993892e-06, "loss": 0.0029, "step": 43150 }, { "epoch": 0.7062096048433282, "grad_norm": 0.1327640265226364, "learning_rate": 9.822545680269056e-06, "loss": 0.003, "step": 43160 }, { "epoch": 0.706373230794404, "grad_norm": 0.14455074071884155, "learning_rate": 9.822357112216619e-06, "loss": 0.0029, "step": 43170 }, { "epoch": 0.7065368567454798, "grad_norm": 0.008310317061841488, "learning_rate": 9.822168445840418e-06, "loss": 0.0038, "step": 43180 }, { "epoch": 0.7067004826965557, "grad_norm": 0.03896079957485199, "learning_rate": 9.821979681144307e-06, "loss": 0.0028, "step": 43190 }, { "epoch": 0.7068641086476315, "grad_norm": 0.07149850577116013, "learning_rate": 9.821790818132128e-06, "loss": 0.0047, "step": 43200 }, { "epoch": 0.7070277345987074, "grad_norm": 0.15884733200073242, "learning_rate": 9.821601856807738e-06, "loss": 0.0044, "step": 43210 }, { "epoch": 0.7071913605497832, "grad_norm": 0.21633541584014893, "learning_rate": 9.821412797174985e-06, "loss": 0.0047, "step": 43220 }, { "epoch": 0.707354986500859, "grad_norm": 0.10643122345209122, "learning_rate": 9.821223639237727e-06, "loss": 0.0089, "step": 43230 }, { "epoch": 0.7075186124519349, "grad_norm": 0.04317478463053703, "learning_rate": 9.821034382999817e-06, "loss": 0.0028, "step": 43240 }, { "epoch": 0.7076822384030107, "grad_norm": 0.05479319766163826, "learning_rate": 9.820845028465119e-06, "loss": 0.002, "step": 43250 }, { "epoch": 0.7078458643540866, "grad_norm": 0.026026679202914238, "learning_rate": 9.820655575637488e-06, "loss": 0.0034, "step": 43260 }, { "epoch": 0.7080094903051624, "grad_norm": 0.06970680505037308, "learning_rate": 9.820466024520791e-06, "loss": 0.0024, "step": 43270 }, { "epoch": 0.7081731162562382, "grad_norm": 0.056593865156173706, "learning_rate": 9.82027637511889e-06, "loss": 0.002, "step": 43280 }, { "epoch": 0.7083367422073141, "grad_norm": 0.030452068895101547, "learning_rate": 9.820086627435655e-06, "loss": 0.0021, "step": 43290 }, { "epoch": 0.7085003681583899, "grad_norm": 0.2573172152042389, "learning_rate": 9.819896781474953e-06, "loss": 0.0028, "step": 43300 }, { "epoch": 0.7086639941094658, "grad_norm": 0.25506624579429626, "learning_rate": 9.819706837240655e-06, "loss": 0.0042, "step": 43310 }, { "epoch": 0.7088276200605416, "grad_norm": 0.10421574115753174, "learning_rate": 9.819516794736632e-06, "loss": 0.0022, "step": 43320 }, { "epoch": 0.7089912460116174, "grad_norm": 0.1738891452550888, "learning_rate": 9.819326653966761e-06, "loss": 0.0034, "step": 43330 }, { "epoch": 0.7091548719626933, "grad_norm": 0.14306990802288055, "learning_rate": 9.81913641493492e-06, "loss": 0.0058, "step": 43340 }, { "epoch": 0.7093184979137691, "grad_norm": 0.17230625450611115, "learning_rate": 9.818946077644984e-06, "loss": 0.0052, "step": 43350 }, { "epoch": 0.709482123864845, "grad_norm": 0.027592523023486137, "learning_rate": 9.818755642100836e-06, "loss": 0.0022, "step": 43360 }, { "epoch": 0.7096457498159208, "grad_norm": 0.02722185291349888, "learning_rate": 9.81856510830636e-06, "loss": 0.0019, "step": 43370 }, { "epoch": 0.7098093757669967, "grad_norm": 0.04472848400473595, "learning_rate": 9.818374476265438e-06, "loss": 0.0023, "step": 43380 }, { "epoch": 0.7099730017180725, "grad_norm": 0.1214161291718483, "learning_rate": 9.818183745981958e-06, "loss": 0.0027, "step": 43390 }, { "epoch": 0.7101366276691483, "grad_norm": 0.14374306797981262, "learning_rate": 9.81799291745981e-06, "loss": 0.0049, "step": 43400 }, { "epoch": 0.7103002536202242, "grad_norm": 0.19747889041900635, "learning_rate": 9.817801990702885e-06, "loss": 0.0034, "step": 43410 }, { "epoch": 0.7104638795713, "grad_norm": 0.08799836039543152, "learning_rate": 9.817610965715072e-06, "loss": 0.0029, "step": 43420 }, { "epoch": 0.7106275055223759, "grad_norm": 0.0437697134912014, "learning_rate": 9.817419842500271e-06, "loss": 0.0041, "step": 43430 }, { "epoch": 0.7107911314734517, "grad_norm": 0.14782284200191498, "learning_rate": 9.817228621062377e-06, "loss": 0.0021, "step": 43440 }, { "epoch": 0.7109547574245275, "grad_norm": 0.014923573471605778, "learning_rate": 9.817037301405286e-06, "loss": 0.0023, "step": 43450 }, { "epoch": 0.7111183833756034, "grad_norm": 0.20492838323116302, "learning_rate": 9.816845883532902e-06, "loss": 0.0026, "step": 43460 }, { "epoch": 0.7112820093266792, "grad_norm": 0.23000240325927734, "learning_rate": 9.816654367449128e-06, "loss": 0.0028, "step": 43470 }, { "epoch": 0.7114456352777551, "grad_norm": 0.20560063421726227, "learning_rate": 9.816462753157867e-06, "loss": 0.0054, "step": 43480 }, { "epoch": 0.711609261228831, "grad_norm": 0.09783950448036194, "learning_rate": 9.816271040663026e-06, "loss": 0.0031, "step": 43490 }, { "epoch": 0.7117728871799067, "grad_norm": 0.107417032122612, "learning_rate": 9.816079229968515e-06, "loss": 0.0019, "step": 43500 }, { "epoch": 0.7119365131309826, "grad_norm": 0.13816292583942413, "learning_rate": 9.815887321078245e-06, "loss": 0.0023, "step": 43510 }, { "epoch": 0.7121001390820584, "grad_norm": 0.19078758358955383, "learning_rate": 9.81569531399613e-06, "loss": 0.0041, "step": 43520 }, { "epoch": 0.7122637650331343, "grad_norm": 0.05488825961947441, "learning_rate": 9.81550320872608e-06, "loss": 0.0031, "step": 43530 }, { "epoch": 0.71242739098421, "grad_norm": 0.11297845840454102, "learning_rate": 9.815311005272017e-06, "loss": 0.0028, "step": 43540 }, { "epoch": 0.7125910169352859, "grad_norm": 0.1595943570137024, "learning_rate": 9.815118703637858e-06, "loss": 0.0025, "step": 43550 }, { "epoch": 0.7127546428863618, "grad_norm": 0.0949808657169342, "learning_rate": 9.814926303827524e-06, "loss": 0.0019, "step": 43560 }, { "epoch": 0.7129182688374376, "grad_norm": 0.05624634400010109, "learning_rate": 9.814733805844936e-06, "loss": 0.003, "step": 43570 }, { "epoch": 0.7130818947885135, "grad_norm": 0.04622925445437431, "learning_rate": 9.81454120969402e-06, "loss": 0.0022, "step": 43580 }, { "epoch": 0.7132455207395892, "grad_norm": 0.0758991539478302, "learning_rate": 9.814348515378704e-06, "loss": 0.0031, "step": 43590 }, { "epoch": 0.7134091466906651, "grad_norm": 0.053786322474479675, "learning_rate": 9.814155722902919e-06, "loss": 0.0032, "step": 43600 }, { "epoch": 0.713572772641741, "grad_norm": 0.055534135550260544, "learning_rate": 9.81396283227059e-06, "loss": 0.0045, "step": 43610 }, { "epoch": 0.7137363985928168, "grad_norm": 0.10076974332332611, "learning_rate": 9.813769843485654e-06, "loss": 0.0032, "step": 43620 }, { "epoch": 0.7139000245438927, "grad_norm": 0.0619572252035141, "learning_rate": 9.813576756552044e-06, "loss": 0.0047, "step": 43630 }, { "epoch": 0.7140636504949684, "grad_norm": 0.032825637608766556, "learning_rate": 9.813383571473699e-06, "loss": 0.003, "step": 43640 }, { "epoch": 0.7142272764460443, "grad_norm": 0.21396471560001373, "learning_rate": 9.813190288254555e-06, "loss": 0.004, "step": 43650 }, { "epoch": 0.7143909023971202, "grad_norm": 0.13447722792625427, "learning_rate": 9.812996906898556e-06, "loss": 0.0029, "step": 43660 }, { "epoch": 0.714554528348196, "grad_norm": 0.1236262172460556, "learning_rate": 9.812803427409642e-06, "loss": 0.0025, "step": 43670 }, { "epoch": 0.7147181542992719, "grad_norm": 0.02427619881927967, "learning_rate": 9.812609849791762e-06, "loss": 0.0025, "step": 43680 }, { "epoch": 0.7148817802503477, "grad_norm": 0.048693329095840454, "learning_rate": 9.812416174048858e-06, "loss": 0.0029, "step": 43690 }, { "epoch": 0.7150454062014235, "grad_norm": 0.09621987491846085, "learning_rate": 9.81222240018488e-06, "loss": 0.0047, "step": 43700 }, { "epoch": 0.7152090321524994, "grad_norm": 0.20514409244060516, "learning_rate": 9.812028528203781e-06, "loss": 0.0024, "step": 43710 }, { "epoch": 0.7153726581035752, "grad_norm": 0.054572854191064835, "learning_rate": 9.811834558109514e-06, "loss": 0.003, "step": 43720 }, { "epoch": 0.7155362840546511, "grad_norm": 0.014185233041644096, "learning_rate": 9.81164048990603e-06, "loss": 0.0026, "step": 43730 }, { "epoch": 0.7156999100057269, "grad_norm": 0.17476168274879456, "learning_rate": 9.81144632359729e-06, "loss": 0.0042, "step": 43740 }, { "epoch": 0.7158635359568027, "grad_norm": 0.17911222577095032, "learning_rate": 9.811252059187253e-06, "loss": 0.0025, "step": 43750 }, { "epoch": 0.7160271619078786, "grad_norm": 0.21996647119522095, "learning_rate": 9.811057696679878e-06, "loss": 0.003, "step": 43760 }, { "epoch": 0.7161907878589544, "grad_norm": 0.12906548380851746, "learning_rate": 9.810863236079126e-06, "loss": 0.0018, "step": 43770 }, { "epoch": 0.7163544138100303, "grad_norm": 0.18263012170791626, "learning_rate": 9.810668677388967e-06, "loss": 0.0033, "step": 43780 }, { "epoch": 0.7165180397611061, "grad_norm": 0.16824966669082642, "learning_rate": 9.810474020613363e-06, "loss": 0.0021, "step": 43790 }, { "epoch": 0.716681665712182, "grad_norm": 0.10567908734083176, "learning_rate": 9.810279265756285e-06, "loss": 0.004, "step": 43800 }, { "epoch": 0.7168452916632578, "grad_norm": 0.03795129805803299, "learning_rate": 9.810084412821703e-06, "loss": 0.0037, "step": 43810 }, { "epoch": 0.7170089176143336, "grad_norm": 0.024171339347958565, "learning_rate": 9.809889461813592e-06, "loss": 0.0026, "step": 43820 }, { "epoch": 0.7171725435654095, "grad_norm": 0.19626858830451965, "learning_rate": 9.809694412735925e-06, "loss": 0.0036, "step": 43830 }, { "epoch": 0.7173361695164853, "grad_norm": 0.2425263375043869, "learning_rate": 9.809499265592678e-06, "loss": 0.0044, "step": 43840 }, { "epoch": 0.7174997954675612, "grad_norm": 0.04055437818169594, "learning_rate": 9.809304020387834e-06, "loss": 0.0026, "step": 43850 }, { "epoch": 0.717663421418637, "grad_norm": 0.21291111409664154, "learning_rate": 9.809108677125371e-06, "loss": 0.0023, "step": 43860 }, { "epoch": 0.7178270473697128, "grad_norm": 0.13939140737056732, "learning_rate": 9.80891323580927e-06, "loss": 0.0026, "step": 43870 }, { "epoch": 0.7179906733207887, "grad_norm": 0.0670999065041542, "learning_rate": 9.80871769644352e-06, "loss": 0.0031, "step": 43880 }, { "epoch": 0.7181542992718645, "grad_norm": 0.11055275052785873, "learning_rate": 9.808522059032104e-06, "loss": 0.0039, "step": 43890 }, { "epoch": 0.7183179252229404, "grad_norm": 0.043317921459674835, "learning_rate": 9.808326323579014e-06, "loss": 0.0032, "step": 43900 }, { "epoch": 0.7184815511740162, "grad_norm": 0.0022606272250413895, "learning_rate": 9.80813049008824e-06, "loss": 0.0022, "step": 43910 }, { "epoch": 0.718645177125092, "grad_norm": 0.07302048057317734, "learning_rate": 9.807934558563774e-06, "loss": 0.0021, "step": 43920 }, { "epoch": 0.7188088030761679, "grad_norm": 0.12072784453630447, "learning_rate": 9.807738529009611e-06, "loss": 0.0039, "step": 43930 }, { "epoch": 0.7189724290272437, "grad_norm": 0.24046100676059723, "learning_rate": 9.807542401429748e-06, "loss": 0.0014, "step": 43940 }, { "epoch": 0.7191360549783196, "grad_norm": 0.1819590926170349, "learning_rate": 9.807346175828184e-06, "loss": 0.0025, "step": 43950 }, { "epoch": 0.7192996809293954, "grad_norm": 0.038836970925331116, "learning_rate": 9.807149852208922e-06, "loss": 0.0022, "step": 43960 }, { "epoch": 0.7194633068804712, "grad_norm": 0.07732320576906204, "learning_rate": 9.80695343057596e-06, "loss": 0.0028, "step": 43970 }, { "epoch": 0.7196269328315471, "grad_norm": 0.15066413581371307, "learning_rate": 9.806756910933308e-06, "loss": 0.0026, "step": 43980 }, { "epoch": 0.7197905587826229, "grad_norm": 0.04754606634378433, "learning_rate": 9.806560293284967e-06, "loss": 0.0017, "step": 43990 }, { "epoch": 0.7199541847336988, "grad_norm": 0.24383534491062164, "learning_rate": 9.806363577634954e-06, "loss": 0.008, "step": 44000 }, { "epoch": 0.7201178106847747, "grad_norm": 0.12651684880256653, "learning_rate": 9.806166763987272e-06, "loss": 0.0026, "step": 44010 }, { "epoch": 0.7202814366358504, "grad_norm": 0.015381730161607265, "learning_rate": 9.80596985234594e-06, "loss": 0.0022, "step": 44020 }, { "epoch": 0.7204450625869263, "grad_norm": 0.026080502197146416, "learning_rate": 9.805772842714968e-06, "loss": 0.0109, "step": 44030 }, { "epoch": 0.7206086885380021, "grad_norm": 0.06381905823945999, "learning_rate": 9.805575735098376e-06, "loss": 0.0027, "step": 44040 }, { "epoch": 0.720772314489078, "grad_norm": 0.049906786531209946, "learning_rate": 9.80537852950018e-06, "loss": 0.0034, "step": 44050 }, { "epoch": 0.7209359404401539, "grad_norm": 0.035226862877607346, "learning_rate": 9.805181225924406e-06, "loss": 0.0019, "step": 44060 }, { "epoch": 0.7210995663912296, "grad_norm": 0.07376007735729218, "learning_rate": 9.804983824375069e-06, "loss": 0.0035, "step": 44070 }, { "epoch": 0.7212631923423055, "grad_norm": 0.04051455110311508, "learning_rate": 9.804786324856202e-06, "loss": 0.0031, "step": 44080 }, { "epoch": 0.7214268182933813, "grad_norm": 0.08301025629043579, "learning_rate": 9.804588727371827e-06, "loss": 0.0019, "step": 44090 }, { "epoch": 0.7215904442444572, "grad_norm": 0.08333950489759445, "learning_rate": 9.804391031925972e-06, "loss": 0.0035, "step": 44100 }, { "epoch": 0.7217540701955331, "grad_norm": 0.06914415210485458, "learning_rate": 9.804193238522673e-06, "loss": 0.0029, "step": 44110 }, { "epoch": 0.7219176961466088, "grad_norm": 0.07608119398355484, "learning_rate": 9.803995347165958e-06, "loss": 0.0033, "step": 44120 }, { "epoch": 0.7220813220976847, "grad_norm": 0.02542872726917267, "learning_rate": 9.803797357859863e-06, "loss": 0.0023, "step": 44130 }, { "epoch": 0.7222449480487605, "grad_norm": 0.018710918724536896, "learning_rate": 9.803599270608427e-06, "loss": 0.002, "step": 44140 }, { "epoch": 0.7224085739998364, "grad_norm": 0.013810411095619202, "learning_rate": 9.803401085415686e-06, "loss": 0.0025, "step": 44150 }, { "epoch": 0.7225721999509123, "grad_norm": 0.2185066193342209, "learning_rate": 9.803202802285682e-06, "loss": 0.0044, "step": 44160 }, { "epoch": 0.722735825901988, "grad_norm": 0.0407746359705925, "learning_rate": 9.80300442122246e-06, "loss": 0.0028, "step": 44170 }, { "epoch": 0.7228994518530639, "grad_norm": 0.06290512531995773, "learning_rate": 9.802805942230061e-06, "loss": 0.0019, "step": 44180 }, { "epoch": 0.7230630778041397, "grad_norm": 0.028632184490561485, "learning_rate": 9.802607365312533e-06, "loss": 0.0027, "step": 44190 }, { "epoch": 0.7232267037552156, "grad_norm": 0.17602133750915527, "learning_rate": 9.802408690473925e-06, "loss": 0.002, "step": 44200 }, { "epoch": 0.7233903297062915, "grad_norm": 0.06467541307210922, "learning_rate": 9.802209917718288e-06, "loss": 0.0025, "step": 44210 }, { "epoch": 0.7235539556573672, "grad_norm": 0.05923107638955116, "learning_rate": 9.802011047049675e-06, "loss": 0.0023, "step": 44220 }, { "epoch": 0.7237175816084431, "grad_norm": 0.09855934232473373, "learning_rate": 9.801812078472142e-06, "loss": 0.0027, "step": 44230 }, { "epoch": 0.7238812075595189, "grad_norm": 0.10856425762176514, "learning_rate": 9.801613011989745e-06, "loss": 0.0022, "step": 44240 }, { "epoch": 0.7240448335105948, "grad_norm": 0.02521812729537487, "learning_rate": 9.80141384760654e-06, "loss": 0.0031, "step": 44250 }, { "epoch": 0.7242084594616707, "grad_norm": 0.0376301147043705, "learning_rate": 9.801214585326592e-06, "loss": 0.0022, "step": 44260 }, { "epoch": 0.7243720854127464, "grad_norm": 0.14163464307785034, "learning_rate": 9.801015225153962e-06, "loss": 0.0029, "step": 44270 }, { "epoch": 0.7245357113638223, "grad_norm": 0.060477565973997116, "learning_rate": 9.800815767092712e-06, "loss": 0.0024, "step": 44280 }, { "epoch": 0.7246993373148981, "grad_norm": 0.11996287852525711, "learning_rate": 9.800616211146914e-06, "loss": 0.0027, "step": 44290 }, { "epoch": 0.724862963265974, "grad_norm": 0.07877856492996216, "learning_rate": 9.800416557320634e-06, "loss": 0.0022, "step": 44300 }, { "epoch": 0.7250265892170499, "grad_norm": 0.06741829216480255, "learning_rate": 9.800216805617942e-06, "loss": 0.0023, "step": 44310 }, { "epoch": 0.7251902151681257, "grad_norm": 0.08566074073314667, "learning_rate": 9.800016956042913e-06, "loss": 0.0025, "step": 44320 }, { "epoch": 0.7253538411192015, "grad_norm": 0.07404974102973938, "learning_rate": 9.79981700859962e-06, "loss": 0.0032, "step": 44330 }, { "epoch": 0.7255174670702773, "grad_norm": 0.06160774081945419, "learning_rate": 9.79961696329214e-06, "loss": 0.0017, "step": 44340 }, { "epoch": 0.7256810930213532, "grad_norm": 0.09825453907251358, "learning_rate": 9.799416820124552e-06, "loss": 0.0023, "step": 44350 }, { "epoch": 0.7258447189724291, "grad_norm": 0.12130559980869293, "learning_rate": 9.799216579100937e-06, "loss": 0.0021, "step": 44360 }, { "epoch": 0.7260083449235049, "grad_norm": 0.19333811104297638, "learning_rate": 9.799016240225379e-06, "loss": 0.0025, "step": 44370 }, { "epoch": 0.7261719708745807, "grad_norm": 0.010152071714401245, "learning_rate": 9.798815803501958e-06, "loss": 0.0012, "step": 44380 }, { "epoch": 0.7263355968256565, "grad_norm": 0.17399998009204865, "learning_rate": 9.798615268934766e-06, "loss": 0.003, "step": 44390 }, { "epoch": 0.7264992227767324, "grad_norm": 0.576988697052002, "learning_rate": 9.798414636527891e-06, "loss": 0.002, "step": 44400 }, { "epoch": 0.7266628487278082, "grad_norm": 0.1084401085972786, "learning_rate": 9.79821390628542e-06, "loss": 0.0046, "step": 44410 }, { "epoch": 0.7268264746788841, "grad_norm": 0.13188078999519348, "learning_rate": 9.79801307821145e-06, "loss": 0.0022, "step": 44420 }, { "epoch": 0.72699010062996, "grad_norm": 0.21797557175159454, "learning_rate": 9.797812152310074e-06, "loss": 0.005, "step": 44430 }, { "epoch": 0.7271537265810357, "grad_norm": 0.10807088017463684, "learning_rate": 9.797611128585388e-06, "loss": 0.0025, "step": 44440 }, { "epoch": 0.7273173525321116, "grad_norm": 0.035494908690452576, "learning_rate": 9.79741000704149e-06, "loss": 0.0029, "step": 44450 }, { "epoch": 0.7274809784831874, "grad_norm": 0.13293850421905518, "learning_rate": 9.797208787682482e-06, "loss": 0.0025, "step": 44460 }, { "epoch": 0.7276446044342633, "grad_norm": 0.10688387602567673, "learning_rate": 9.797007470512469e-06, "loss": 0.0051, "step": 44470 }, { "epoch": 0.7278082303853391, "grad_norm": 0.106321781873703, "learning_rate": 9.796806055535551e-06, "loss": 0.0018, "step": 44480 }, { "epoch": 0.7279718563364149, "grad_norm": 0.07344456017017365, "learning_rate": 9.796604542755839e-06, "loss": 0.0021, "step": 44490 }, { "epoch": 0.7281354822874908, "grad_norm": 0.05840141698718071, "learning_rate": 9.796402932177438e-06, "loss": 0.0023, "step": 44500 }, { "epoch": 0.7282991082385666, "grad_norm": 0.05820420756936073, "learning_rate": 9.796201223804462e-06, "loss": 0.0032, "step": 44510 }, { "epoch": 0.7284627341896425, "grad_norm": 1.2518428564071655, "learning_rate": 9.795999417641022e-06, "loss": 0.0036, "step": 44520 }, { "epoch": 0.7286263601407184, "grad_norm": 0.03911508247256279, "learning_rate": 9.795797513691232e-06, "loss": 0.0024, "step": 44530 }, { "epoch": 0.7287899860917941, "grad_norm": 0.33814549446105957, "learning_rate": 9.795595511959208e-06, "loss": 0.0056, "step": 44540 }, { "epoch": 0.72895361204287, "grad_norm": 0.355214923620224, "learning_rate": 9.795393412449072e-06, "loss": 0.0035, "step": 44550 }, { "epoch": 0.7291172379939458, "grad_norm": 0.046059418469667435, "learning_rate": 9.795191215164943e-06, "loss": 0.0037, "step": 44560 }, { "epoch": 0.7292808639450217, "grad_norm": 0.10430242121219635, "learning_rate": 9.79498892011094e-06, "loss": 0.0033, "step": 44570 }, { "epoch": 0.7294444898960976, "grad_norm": 0.073418989777565, "learning_rate": 9.794786527291196e-06, "loss": 0.0031, "step": 44580 }, { "epoch": 0.7296081158471733, "grad_norm": 0.44493022561073303, "learning_rate": 9.79458403670983e-06, "loss": 0.006, "step": 44590 }, { "epoch": 0.7297717417982492, "grad_norm": 0.2116163671016693, "learning_rate": 9.794381448370974e-06, "loss": 0.0037, "step": 44600 }, { "epoch": 0.729935367749325, "grad_norm": 0.10657970607280731, "learning_rate": 9.794178762278756e-06, "loss": 0.0028, "step": 44610 }, { "epoch": 0.7300989937004009, "grad_norm": 0.20461271703243256, "learning_rate": 9.793975978437312e-06, "loss": 0.0034, "step": 44620 }, { "epoch": 0.7302626196514768, "grad_norm": 0.041067611426115036, "learning_rate": 9.793773096850773e-06, "loss": 0.002, "step": 44630 }, { "epoch": 0.7304262456025525, "grad_norm": 0.04362873360514641, "learning_rate": 9.793570117523279e-06, "loss": 0.0023, "step": 44640 }, { "epoch": 0.7305898715536284, "grad_norm": 0.06886371970176697, "learning_rate": 9.793367040458966e-06, "loss": 0.0044, "step": 44650 }, { "epoch": 0.7307534975047042, "grad_norm": 0.10333307832479477, "learning_rate": 9.793163865661977e-06, "loss": 0.0024, "step": 44660 }, { "epoch": 0.7309171234557801, "grad_norm": 0.22636538743972778, "learning_rate": 9.79296059313645e-06, "loss": 0.0033, "step": 44670 }, { "epoch": 0.731080749406856, "grad_norm": 0.26287880539894104, "learning_rate": 9.792757222886538e-06, "loss": 0.0024, "step": 44680 }, { "epoch": 0.7312443753579317, "grad_norm": 0.11475876718759537, "learning_rate": 9.79255375491638e-06, "loss": 0.0018, "step": 44690 }, { "epoch": 0.7314080013090076, "grad_norm": 0.24791474640369415, "learning_rate": 9.792350189230125e-06, "loss": 0.005, "step": 44700 }, { "epoch": 0.7315716272600834, "grad_norm": 0.1436145156621933, "learning_rate": 9.792146525831928e-06, "loss": 0.0023, "step": 44710 }, { "epoch": 0.7317352532111593, "grad_norm": 0.0213567353785038, "learning_rate": 9.791942764725938e-06, "loss": 0.002, "step": 44720 }, { "epoch": 0.7318988791622352, "grad_norm": 0.1989593505859375, "learning_rate": 9.79173890591631e-06, "loss": 0.0037, "step": 44730 }, { "epoch": 0.732062505113311, "grad_norm": 0.08524028956890106, "learning_rate": 9.791534949407202e-06, "loss": 0.002, "step": 44740 }, { "epoch": 0.7322261310643868, "grad_norm": 0.07885408401489258, "learning_rate": 9.791330895202768e-06, "loss": 0.0041, "step": 44750 }, { "epoch": 0.7323897570154626, "grad_norm": 0.12538421154022217, "learning_rate": 9.791126743307176e-06, "loss": 0.0026, "step": 44760 }, { "epoch": 0.7325533829665385, "grad_norm": 0.1148659959435463, "learning_rate": 9.790922493724583e-06, "loss": 0.0045, "step": 44770 }, { "epoch": 0.7327170089176144, "grad_norm": 0.08049943298101425, "learning_rate": 9.790718146459154e-06, "loss": 0.0026, "step": 44780 }, { "epoch": 0.7328806348686902, "grad_norm": 0.4611159563064575, "learning_rate": 9.790513701515058e-06, "loss": 0.0028, "step": 44790 }, { "epoch": 0.733044260819766, "grad_norm": 0.16672304272651672, "learning_rate": 9.79030915889646e-06, "loss": 0.0021, "step": 44800 }, { "epoch": 0.7332078867708418, "grad_norm": 0.10810426622629166, "learning_rate": 9.790104518607535e-06, "loss": 0.0015, "step": 44810 }, { "epoch": 0.7333715127219177, "grad_norm": 0.16319534182548523, "learning_rate": 9.78989978065245e-06, "loss": 0.0017, "step": 44820 }, { "epoch": 0.7335351386729936, "grad_norm": 0.19981221854686737, "learning_rate": 9.789694945035383e-06, "loss": 0.0035, "step": 44830 }, { "epoch": 0.7336987646240694, "grad_norm": 0.18697872757911682, "learning_rate": 9.78949001176051e-06, "loss": 0.0031, "step": 44840 }, { "epoch": 0.7338623905751452, "grad_norm": 0.05668473616242409, "learning_rate": 9.789284980832009e-06, "loss": 0.0029, "step": 44850 }, { "epoch": 0.734026016526221, "grad_norm": 0.12394070625305176, "learning_rate": 9.78907985225406e-06, "loss": 0.0054, "step": 44860 }, { "epoch": 0.7341896424772969, "grad_norm": 0.03345782309770584, "learning_rate": 9.788874626030846e-06, "loss": 0.0025, "step": 44870 }, { "epoch": 0.7343532684283728, "grad_norm": 0.3549953103065491, "learning_rate": 9.788669302166551e-06, "loss": 0.0048, "step": 44880 }, { "epoch": 0.7345168943794486, "grad_norm": 0.04275945574045181, "learning_rate": 9.788463880665361e-06, "loss": 0.0037, "step": 44890 }, { "epoch": 0.7346805203305244, "grad_norm": 0.04341256618499756, "learning_rate": 9.788258361531466e-06, "loss": 0.0024, "step": 44900 }, { "epoch": 0.7348441462816002, "grad_norm": 0.13246269524097443, "learning_rate": 9.788052744769054e-06, "loss": 0.0027, "step": 44910 }, { "epoch": 0.7350077722326761, "grad_norm": 0.10529180616140366, "learning_rate": 9.78784703038232e-06, "loss": 0.0035, "step": 44920 }, { "epoch": 0.735171398183752, "grad_norm": 0.046023279428482056, "learning_rate": 9.787641218375456e-06, "loss": 0.0038, "step": 44930 }, { "epoch": 0.7353350241348278, "grad_norm": 0.12781111896038055, "learning_rate": 9.78743530875266e-06, "loss": 0.0029, "step": 44940 }, { "epoch": 0.7354986500859036, "grad_norm": 0.029971975833177567, "learning_rate": 9.787229301518128e-06, "loss": 0.0018, "step": 44950 }, { "epoch": 0.7356622760369794, "grad_norm": 0.011334256269037724, "learning_rate": 9.787023196676064e-06, "loss": 0.0017, "step": 44960 }, { "epoch": 0.7358259019880553, "grad_norm": 0.316170334815979, "learning_rate": 9.786816994230668e-06, "loss": 0.0036, "step": 44970 }, { "epoch": 0.7359895279391312, "grad_norm": 0.10515231639146805, "learning_rate": 9.786610694186143e-06, "loss": 0.0026, "step": 44980 }, { "epoch": 0.736153153890207, "grad_norm": 0.05518409237265587, "learning_rate": 9.786404296546698e-06, "loss": 0.0019, "step": 44990 }, { "epoch": 0.7363167798412829, "grad_norm": 0.08654700964689255, "learning_rate": 9.786197801316539e-06, "loss": 0.0025, "step": 45000 }, { "epoch": 0.7364804057923586, "grad_norm": 0.03393130004405975, "learning_rate": 9.785991208499878e-06, "loss": 0.0044, "step": 45010 }, { "epoch": 0.7366440317434345, "grad_norm": 0.1705559343099594, "learning_rate": 9.785784518100928e-06, "loss": 0.0036, "step": 45020 }, { "epoch": 0.7368076576945104, "grad_norm": 0.10219141095876694, "learning_rate": 9.7855777301239e-06, "loss": 0.0028, "step": 45030 }, { "epoch": 0.7369712836455862, "grad_norm": 0.032892607152462006, "learning_rate": 9.785370844573014e-06, "loss": 0.0019, "step": 45040 }, { "epoch": 0.7371349095966621, "grad_norm": 0.03448035940527916, "learning_rate": 9.785163861452485e-06, "loss": 0.0025, "step": 45050 }, { "epoch": 0.7372985355477378, "grad_norm": 0.11431076377630234, "learning_rate": 9.784956780766534e-06, "loss": 0.0019, "step": 45060 }, { "epoch": 0.7374621614988137, "grad_norm": 0.15446516871452332, "learning_rate": 9.784749602519383e-06, "loss": 0.003, "step": 45070 }, { "epoch": 0.7376257874498896, "grad_norm": 0.10187547653913498, "learning_rate": 9.784542326715259e-06, "loss": 0.0038, "step": 45080 }, { "epoch": 0.7377894134009654, "grad_norm": 0.09889063239097595, "learning_rate": 9.784334953358386e-06, "loss": 0.0037, "step": 45090 }, { "epoch": 0.7379530393520413, "grad_norm": 0.05878574773669243, "learning_rate": 9.78412748245299e-06, "loss": 0.0037, "step": 45100 }, { "epoch": 0.738116665303117, "grad_norm": 0.1768224984407425, "learning_rate": 9.783919914003306e-06, "loss": 0.0022, "step": 45110 }, { "epoch": 0.7382802912541929, "grad_norm": 0.05048155039548874, "learning_rate": 9.783712248013561e-06, "loss": 0.0028, "step": 45120 }, { "epoch": 0.7384439172052688, "grad_norm": 0.1879453808069229, "learning_rate": 9.783504484487994e-06, "loss": 0.0039, "step": 45130 }, { "epoch": 0.7386075431563446, "grad_norm": 0.08732523024082184, "learning_rate": 9.783296623430838e-06, "loss": 0.0019, "step": 45140 }, { "epoch": 0.7387711691074205, "grad_norm": 0.12023834884166718, "learning_rate": 9.783088664846332e-06, "loss": 0.0018, "step": 45150 }, { "epoch": 0.7389347950584962, "grad_norm": 0.0768180638551712, "learning_rate": 9.782880608738714e-06, "loss": 0.0014, "step": 45160 }, { "epoch": 0.7390984210095721, "grad_norm": 0.05600440502166748, "learning_rate": 9.78267245511223e-06, "loss": 0.0029, "step": 45170 }, { "epoch": 0.739262046960648, "grad_norm": 0.1441577672958374, "learning_rate": 9.782464203971122e-06, "loss": 0.0035, "step": 45180 }, { "epoch": 0.7394256729117238, "grad_norm": 0.24062299728393555, "learning_rate": 9.782255855319636e-06, "loss": 0.0022, "step": 45190 }, { "epoch": 0.7395892988627997, "grad_norm": 0.07244294136762619, "learning_rate": 9.78204740916202e-06, "loss": 0.0035, "step": 45200 }, { "epoch": 0.7397529248138754, "grad_norm": 0.07723509520292282, "learning_rate": 9.781838865502525e-06, "loss": 0.0054, "step": 45210 }, { "epoch": 0.7399165507649513, "grad_norm": 0.145652174949646, "learning_rate": 9.781630224345402e-06, "loss": 0.0013, "step": 45220 }, { "epoch": 0.7400801767160272, "grad_norm": 0.07704487442970276, "learning_rate": 9.781421485694906e-06, "loss": 0.0023, "step": 45230 }, { "epoch": 0.740243802667103, "grad_norm": 0.10151775181293488, "learning_rate": 9.781212649555291e-06, "loss": 0.0029, "step": 45240 }, { "epoch": 0.7404074286181789, "grad_norm": 0.08714748173952103, "learning_rate": 9.781003715930817e-06, "loss": 0.0022, "step": 45250 }, { "epoch": 0.7405710545692546, "grad_norm": 0.06606046110391617, "learning_rate": 9.780794684825743e-06, "loss": 0.0028, "step": 45260 }, { "epoch": 0.7407346805203305, "grad_norm": 0.23763804137706757, "learning_rate": 9.780585556244332e-06, "loss": 0.003, "step": 45270 }, { "epoch": 0.7408983064714064, "grad_norm": 0.0766981840133667, "learning_rate": 9.780376330190847e-06, "loss": 0.002, "step": 45280 }, { "epoch": 0.7410619324224822, "grad_norm": 0.24504989385604858, "learning_rate": 9.780167006669554e-06, "loss": 0.0036, "step": 45290 }, { "epoch": 0.7412255583735581, "grad_norm": 0.04044891893863678, "learning_rate": 9.779957585684721e-06, "loss": 0.0017, "step": 45300 }, { "epoch": 0.7413891843246339, "grad_norm": 0.16524098813533783, "learning_rate": 9.77974806724062e-06, "loss": 0.0058, "step": 45310 }, { "epoch": 0.7415528102757097, "grad_norm": 0.184432715177536, "learning_rate": 9.779538451341519e-06, "loss": 0.0041, "step": 45320 }, { "epoch": 0.7417164362267855, "grad_norm": 0.05139269679784775, "learning_rate": 9.779328737991694e-06, "loss": 0.005, "step": 45330 }, { "epoch": 0.7418800621778614, "grad_norm": 0.06401035189628601, "learning_rate": 9.77911892719542e-06, "loss": 0.0024, "step": 45340 }, { "epoch": 0.7420436881289373, "grad_norm": 0.09870250523090363, "learning_rate": 9.778909018956977e-06, "loss": 0.0022, "step": 45350 }, { "epoch": 0.7422073140800131, "grad_norm": 0.020407257601618767, "learning_rate": 9.778699013280642e-06, "loss": 0.0015, "step": 45360 }, { "epoch": 0.7423709400310889, "grad_norm": 0.06867571920156479, "learning_rate": 9.778488910170699e-06, "loss": 0.0041, "step": 45370 }, { "epoch": 0.7425345659821647, "grad_norm": 0.13049520552158356, "learning_rate": 9.77827870963143e-06, "loss": 0.0031, "step": 45380 }, { "epoch": 0.7426981919332406, "grad_norm": 0.19489416480064392, "learning_rate": 9.778068411667124e-06, "loss": 0.0021, "step": 45390 }, { "epoch": 0.7428618178843165, "grad_norm": 0.15758655965328217, "learning_rate": 9.777858016282067e-06, "loss": 0.0038, "step": 45400 }, { "epoch": 0.7430254438353923, "grad_norm": 0.12114369124174118, "learning_rate": 9.777647523480547e-06, "loss": 0.0016, "step": 45410 }, { "epoch": 0.7431890697864681, "grad_norm": 0.09712661057710648, "learning_rate": 9.777436933266857e-06, "loss": 0.0021, "step": 45420 }, { "epoch": 0.7433526957375439, "grad_norm": 0.13094022870063782, "learning_rate": 9.77722624564529e-06, "loss": 0.004, "step": 45430 }, { "epoch": 0.7435163216886198, "grad_norm": 0.22410163283348083, "learning_rate": 9.777015460620144e-06, "loss": 0.0032, "step": 45440 }, { "epoch": 0.7436799476396957, "grad_norm": 0.03716503083705902, "learning_rate": 9.776804578195718e-06, "loss": 0.0013, "step": 45450 }, { "epoch": 0.7438435735907715, "grad_norm": 0.2147616297006607, "learning_rate": 9.776593598376305e-06, "loss": 0.0019, "step": 45460 }, { "epoch": 0.7440071995418474, "grad_norm": 0.1225711926817894, "learning_rate": 9.776382521166214e-06, "loss": 0.0011, "step": 45470 }, { "epoch": 0.7441708254929231, "grad_norm": 0.11914604157209396, "learning_rate": 9.776171346569743e-06, "loss": 0.0028, "step": 45480 }, { "epoch": 0.744334451443999, "grad_norm": 0.08077944070100784, "learning_rate": 9.775960074591202e-06, "loss": 0.0027, "step": 45490 }, { "epoch": 0.7444980773950749, "grad_norm": 0.04823603108525276, "learning_rate": 9.775748705234897e-06, "loss": 0.0031, "step": 45500 }, { "epoch": 0.7446617033461507, "grad_norm": 0.007279330398887396, "learning_rate": 9.775537238505136e-06, "loss": 0.0021, "step": 45510 }, { "epoch": 0.7448253292972266, "grad_norm": 0.08081437647342682, "learning_rate": 9.775325674406233e-06, "loss": 0.0019, "step": 45520 }, { "epoch": 0.7449889552483023, "grad_norm": 0.07863911986351013, "learning_rate": 9.775114012942501e-06, "loss": 0.0017, "step": 45530 }, { "epoch": 0.7451525811993782, "grad_norm": 0.016316983848810196, "learning_rate": 9.774902254118255e-06, "loss": 0.0012, "step": 45540 }, { "epoch": 0.7453162071504541, "grad_norm": 0.020287347957491875, "learning_rate": 9.774690397937814e-06, "loss": 0.0026, "step": 45550 }, { "epoch": 0.7454798331015299, "grad_norm": 0.10449451953172684, "learning_rate": 9.774478444405494e-06, "loss": 0.0023, "step": 45560 }, { "epoch": 0.7456434590526058, "grad_norm": 0.0947481095790863, "learning_rate": 9.774266393525621e-06, "loss": 0.0037, "step": 45570 }, { "epoch": 0.7458070850036815, "grad_norm": 0.08189667016267776, "learning_rate": 9.774054245302516e-06, "loss": 0.0027, "step": 45580 }, { "epoch": 0.7459707109547574, "grad_norm": 0.00909357238560915, "learning_rate": 9.773841999740506e-06, "loss": 0.0051, "step": 45590 }, { "epoch": 0.7461343369058333, "grad_norm": 0.0861947312951088, "learning_rate": 9.773629656843917e-06, "loss": 0.0019, "step": 45600 }, { "epoch": 0.7462979628569091, "grad_norm": 0.046073660254478455, "learning_rate": 9.773417216617079e-06, "loss": 0.0028, "step": 45610 }, { "epoch": 0.746461588807985, "grad_norm": 0.20828597247600555, "learning_rate": 9.773204679064324e-06, "loss": 0.0034, "step": 45620 }, { "epoch": 0.7466252147590607, "grad_norm": 0.12020736187696457, "learning_rate": 9.772992044189984e-06, "loss": 0.0029, "step": 45630 }, { "epoch": 0.7467888407101366, "grad_norm": 0.05431856960058212, "learning_rate": 9.772779311998398e-06, "loss": 0.004, "step": 45640 }, { "epoch": 0.7469524666612125, "grad_norm": 0.243337020277977, "learning_rate": 9.7725664824939e-06, "loss": 0.0048, "step": 45650 }, { "epoch": 0.7471160926122883, "grad_norm": 0.04290987551212311, "learning_rate": 9.772353555680828e-06, "loss": 0.0016, "step": 45660 }, { "epoch": 0.7472797185633642, "grad_norm": 0.10696326196193695, "learning_rate": 9.772140531563527e-06, "loss": 0.0016, "step": 45670 }, { "epoch": 0.74744334451444, "grad_norm": 0.1067265123128891, "learning_rate": 9.77192741014634e-06, "loss": 0.0018, "step": 45680 }, { "epoch": 0.7476069704655158, "grad_norm": 0.09243091940879822, "learning_rate": 9.771714191433612e-06, "loss": 0.0015, "step": 45690 }, { "epoch": 0.7477705964165917, "grad_norm": 0.06521312892436981, "learning_rate": 9.771500875429687e-06, "loss": 0.0013, "step": 45700 }, { "epoch": 0.7479342223676675, "grad_norm": 0.16387376189231873, "learning_rate": 9.77128746213892e-06, "loss": 0.0035, "step": 45710 }, { "epoch": 0.7480978483187434, "grad_norm": 0.05783892050385475, "learning_rate": 9.771073951565657e-06, "loss": 0.0026, "step": 45720 }, { "epoch": 0.7482614742698191, "grad_norm": 0.21741926670074463, "learning_rate": 9.770860343714256e-06, "loss": 0.004, "step": 45730 }, { "epoch": 0.748425100220895, "grad_norm": 0.0997646227478981, "learning_rate": 9.770646638589069e-06, "loss": 0.0033, "step": 45740 }, { "epoch": 0.7485887261719709, "grad_norm": 0.08884342014789581, "learning_rate": 9.770432836194454e-06, "loss": 0.0034, "step": 45750 }, { "epoch": 0.7487523521230467, "grad_norm": 0.12249887734651566, "learning_rate": 9.770218936534773e-06, "loss": 0.0028, "step": 45760 }, { "epoch": 0.7489159780741226, "grad_norm": 0.08099710196256638, "learning_rate": 9.770004939614383e-06, "loss": 0.0029, "step": 45770 }, { "epoch": 0.7490796040251984, "grad_norm": 0.08112912625074387, "learning_rate": 9.769790845437647e-06, "loss": 0.0018, "step": 45780 }, { "epoch": 0.7492432299762742, "grad_norm": 0.07284128665924072, "learning_rate": 9.769576654008936e-06, "loss": 0.0017, "step": 45790 }, { "epoch": 0.7494068559273501, "grad_norm": 0.09556381404399872, "learning_rate": 9.769362365332611e-06, "loss": 0.0042, "step": 45800 }, { "epoch": 0.7495704818784259, "grad_norm": 0.10694733262062073, "learning_rate": 9.769147979413044e-06, "loss": 0.0034, "step": 45810 }, { "epoch": 0.7497341078295018, "grad_norm": 0.051127899438142776, "learning_rate": 9.768933496254607e-06, "loss": 0.0017, "step": 45820 }, { "epoch": 0.7498977337805776, "grad_norm": 0.0850977972149849, "learning_rate": 9.76871891586167e-06, "loss": 0.0012, "step": 45830 }, { "epoch": 0.7500613597316534, "grad_norm": 0.13684287667274475, "learning_rate": 9.76850423823861e-06, "loss": 0.0047, "step": 45840 }, { "epoch": 0.7502249856827293, "grad_norm": 0.06218522787094116, "learning_rate": 9.768289463389807e-06, "loss": 0.0021, "step": 45850 }, { "epoch": 0.7503886116338051, "grad_norm": 0.08200713247060776, "learning_rate": 9.768074591319634e-06, "loss": 0.0022, "step": 45860 }, { "epoch": 0.750552237584881, "grad_norm": 0.04231148958206177, "learning_rate": 9.767859622032476e-06, "loss": 0.0026, "step": 45870 }, { "epoch": 0.7507158635359568, "grad_norm": 0.14967189729213715, "learning_rate": 9.767644555532715e-06, "loss": 0.0043, "step": 45880 }, { "epoch": 0.7508794894870326, "grad_norm": 0.026081779971718788, "learning_rate": 9.767429391824736e-06, "loss": 0.0028, "step": 45890 }, { "epoch": 0.7510431154381085, "grad_norm": 0.06019185110926628, "learning_rate": 9.767214130912928e-06, "loss": 0.0035, "step": 45900 }, { "epoch": 0.7512067413891843, "grad_norm": 0.12409131228923798, "learning_rate": 9.766998772801676e-06, "loss": 0.006, "step": 45910 }, { "epoch": 0.7513703673402602, "grad_norm": 0.2900870144367218, "learning_rate": 9.766783317495373e-06, "loss": 0.0065, "step": 45920 }, { "epoch": 0.751533993291336, "grad_norm": 0.0832093209028244, "learning_rate": 9.766567764998417e-06, "loss": 0.0027, "step": 45930 }, { "epoch": 0.7516976192424119, "grad_norm": 0.03838086128234863, "learning_rate": 9.766352115315191e-06, "loss": 0.004, "step": 45940 }, { "epoch": 0.7518612451934877, "grad_norm": 0.0283611211925745, "learning_rate": 9.766136368450104e-06, "loss": 0.0022, "step": 45950 }, { "epoch": 0.7520248711445635, "grad_norm": 0.04620694741606712, "learning_rate": 9.765920524407548e-06, "loss": 0.0038, "step": 45960 }, { "epoch": 0.7521884970956394, "grad_norm": 0.06304480880498886, "learning_rate": 9.765704583191926e-06, "loss": 0.0024, "step": 45970 }, { "epoch": 0.7523521230467152, "grad_norm": 0.10843624174594879, "learning_rate": 9.765488544807642e-06, "loss": 0.0029, "step": 45980 }, { "epoch": 0.7525157489977911, "grad_norm": 0.09157103300094604, "learning_rate": 9.765272409259098e-06, "loss": 0.0018, "step": 45990 }, { "epoch": 0.7526793749488669, "grad_norm": 0.05723569914698601, "learning_rate": 9.765056176550703e-06, "loss": 0.0019, "step": 46000 }, { "epoch": 0.7528430008999427, "grad_norm": 0.07801765948534012, "learning_rate": 9.764839846686865e-06, "loss": 0.0029, "step": 46010 }, { "epoch": 0.7530066268510186, "grad_norm": 0.048715896904468536, "learning_rate": 9.764623419671995e-06, "loss": 0.0021, "step": 46020 }, { "epoch": 0.7531702528020944, "grad_norm": 0.10928426682949066, "learning_rate": 9.764406895510504e-06, "loss": 0.003, "step": 46030 }, { "epoch": 0.7533338787531703, "grad_norm": 0.030075401067733765, "learning_rate": 9.76419027420681e-06, "loss": 0.0019, "step": 46040 }, { "epoch": 0.7534975047042461, "grad_norm": 0.07996998727321625, "learning_rate": 9.763973555765326e-06, "loss": 0.0035, "step": 46050 }, { "epoch": 0.7536611306553219, "grad_norm": 0.05251079797744751, "learning_rate": 9.763756740190475e-06, "loss": 0.0035, "step": 46060 }, { "epoch": 0.7538247566063978, "grad_norm": 0.06734722852706909, "learning_rate": 9.763539827486677e-06, "loss": 0.0025, "step": 46070 }, { "epoch": 0.7539883825574736, "grad_norm": 0.119187131524086, "learning_rate": 9.76332281765835e-06, "loss": 0.0024, "step": 46080 }, { "epoch": 0.7541520085085495, "grad_norm": 0.1638759970664978, "learning_rate": 9.763105710709923e-06, "loss": 0.0017, "step": 46090 }, { "epoch": 0.7543156344596254, "grad_norm": 0.04908619448542595, "learning_rate": 9.762888506645822e-06, "loss": 0.0022, "step": 46100 }, { "epoch": 0.7544792604107011, "grad_norm": 0.03631407767534256, "learning_rate": 9.762671205470474e-06, "loss": 0.0017, "step": 46110 }, { "epoch": 0.754642886361777, "grad_norm": 0.10562891513109207, "learning_rate": 9.76245380718831e-06, "loss": 0.0037, "step": 46120 }, { "epoch": 0.7548065123128528, "grad_norm": 0.10250034183263779, "learning_rate": 9.762236311803766e-06, "loss": 0.0025, "step": 46130 }, { "epoch": 0.7549701382639287, "grad_norm": 0.09247435629367828, "learning_rate": 9.762018719321272e-06, "loss": 0.0029, "step": 46140 }, { "epoch": 0.7551337642150046, "grad_norm": 0.06103226915001869, "learning_rate": 9.761801029745266e-06, "loss": 0.0026, "step": 46150 }, { "epoch": 0.7552973901660803, "grad_norm": 0.09051872789859772, "learning_rate": 9.761583243080187e-06, "loss": 0.0016, "step": 46160 }, { "epoch": 0.7554610161171562, "grad_norm": 0.08331990242004395, "learning_rate": 9.761365359330475e-06, "loss": 0.0021, "step": 46170 }, { "epoch": 0.755624642068232, "grad_norm": 0.04727861285209656, "learning_rate": 9.761147378500573e-06, "loss": 0.002, "step": 46180 }, { "epoch": 0.7557882680193079, "grad_norm": 0.0715755969285965, "learning_rate": 9.760929300594926e-06, "loss": 0.0013, "step": 46190 }, { "epoch": 0.7559518939703836, "grad_norm": 0.03973992168903351, "learning_rate": 9.760711125617979e-06, "loss": 0.0046, "step": 46200 }, { "epoch": 0.7561155199214595, "grad_norm": 0.07738612592220306, "learning_rate": 9.76049285357418e-06, "loss": 0.0018, "step": 46210 }, { "epoch": 0.7562791458725354, "grad_norm": 0.034473635256290436, "learning_rate": 9.760274484467981e-06, "loss": 0.0017, "step": 46220 }, { "epoch": 0.7564427718236112, "grad_norm": 0.092461496591568, "learning_rate": 9.760056018303834e-06, "loss": 0.0015, "step": 46230 }, { "epoch": 0.7566063977746871, "grad_norm": 0.04530169069766998, "learning_rate": 9.759837455086193e-06, "loss": 0.0032, "step": 46240 }, { "epoch": 0.7567700237257629, "grad_norm": 0.13158944249153137, "learning_rate": 9.759618794819514e-06, "loss": 0.0038, "step": 46250 }, { "epoch": 0.7569336496768387, "grad_norm": 0.19514302909374237, "learning_rate": 9.759400037508257e-06, "loss": 0.0037, "step": 46260 }, { "epoch": 0.7570972756279146, "grad_norm": 0.08501393347978592, "learning_rate": 9.75918118315688e-06, "loss": 0.0051, "step": 46270 }, { "epoch": 0.7572609015789904, "grad_norm": 0.028479844331741333, "learning_rate": 9.758962231769846e-06, "loss": 0.0022, "step": 46280 }, { "epoch": 0.7574245275300663, "grad_norm": 0.06916332244873047, "learning_rate": 9.75874318335162e-06, "loss": 0.0025, "step": 46290 }, { "epoch": 0.7575881534811421, "grad_norm": 0.03860297054052353, "learning_rate": 9.758524037906666e-06, "loss": 0.0033, "step": 46300 }, { "epoch": 0.7577517794322179, "grad_norm": 0.07455075532197952, "learning_rate": 9.758304795439454e-06, "loss": 0.0028, "step": 46310 }, { "epoch": 0.7579154053832938, "grad_norm": 0.03571438416838646, "learning_rate": 9.758085455954457e-06, "loss": 0.0048, "step": 46320 }, { "epoch": 0.7580790313343696, "grad_norm": 0.19071762263774872, "learning_rate": 9.75786601945614e-06, "loss": 0.002, "step": 46330 }, { "epoch": 0.7582426572854455, "grad_norm": 0.2027096450328827, "learning_rate": 9.757646485948986e-06, "loss": 0.0032, "step": 46340 }, { "epoch": 0.7584062832365213, "grad_norm": 0.1549309343099594, "learning_rate": 9.757426855437464e-06, "loss": 0.0023, "step": 46350 }, { "epoch": 0.7585699091875971, "grad_norm": 0.11152104288339615, "learning_rate": 9.757207127926054e-06, "loss": 0.003, "step": 46360 }, { "epoch": 0.758733535138673, "grad_norm": 0.036685261875391006, "learning_rate": 9.756987303419236e-06, "loss": 0.0017, "step": 46370 }, { "epoch": 0.7588971610897488, "grad_norm": 0.11827714741230011, "learning_rate": 9.756767381921495e-06, "loss": 0.0031, "step": 46380 }, { "epoch": 0.7590607870408247, "grad_norm": 0.07287953048944473, "learning_rate": 9.756547363437312e-06, "loss": 0.0015, "step": 46390 }, { "epoch": 0.7592244129919005, "grad_norm": 0.07416464388370514, "learning_rate": 9.756327247971171e-06, "loss": 0.0027, "step": 46400 }, { "epoch": 0.7593880389429764, "grad_norm": 0.11255506426095963, "learning_rate": 9.756107035527564e-06, "loss": 0.0051, "step": 46410 }, { "epoch": 0.7595516648940522, "grad_norm": 0.06275136768817902, "learning_rate": 9.75588672611098e-06, "loss": 0.003, "step": 46420 }, { "epoch": 0.759715290845128, "grad_norm": 0.16329994797706604, "learning_rate": 9.75566631972591e-06, "loss": 0.0041, "step": 46430 }, { "epoch": 0.7598789167962039, "grad_norm": 0.1303407847881317, "learning_rate": 9.75544581637685e-06, "loss": 0.0036, "step": 46440 }, { "epoch": 0.7600425427472797, "grad_norm": 0.06302310526371002, "learning_rate": 9.755225216068292e-06, "loss": 0.0047, "step": 46450 }, { "epoch": 0.7602061686983556, "grad_norm": 0.04518478736281395, "learning_rate": 9.755004518804736e-06, "loss": 0.0026, "step": 46460 }, { "epoch": 0.7603697946494314, "grad_norm": 0.04658970981836319, "learning_rate": 9.754783724590681e-06, "loss": 0.0028, "step": 46470 }, { "epoch": 0.7605334206005072, "grad_norm": 0.19948288798332214, "learning_rate": 9.754562833430632e-06, "loss": 0.0039, "step": 46480 }, { "epoch": 0.7606970465515831, "grad_norm": 0.09420941770076752, "learning_rate": 9.75434184532909e-06, "loss": 0.0048, "step": 46490 }, { "epoch": 0.7608606725026589, "grad_norm": 0.08019019663333893, "learning_rate": 9.75412076029056e-06, "loss": 0.0026, "step": 46500 }, { "epoch": 0.7610242984537348, "grad_norm": 0.025455081835389137, "learning_rate": 9.753899578319552e-06, "loss": 0.0031, "step": 46510 }, { "epoch": 0.7611879244048106, "grad_norm": 0.061003610491752625, "learning_rate": 9.753678299420574e-06, "loss": 0.002, "step": 46520 }, { "epoch": 0.7613515503558864, "grad_norm": 0.26109057664871216, "learning_rate": 9.75345692359814e-06, "loss": 0.0045, "step": 46530 }, { "epoch": 0.7615151763069623, "grad_norm": 0.09738463163375854, "learning_rate": 9.75323545085676e-06, "loss": 0.0034, "step": 46540 }, { "epoch": 0.7616788022580381, "grad_norm": 0.05687416344881058, "learning_rate": 9.75301388120095e-06, "loss": 0.0018, "step": 46550 }, { "epoch": 0.761842428209114, "grad_norm": 0.06727267056703568, "learning_rate": 9.752792214635232e-06, "loss": 0.004, "step": 46560 }, { "epoch": 0.7620060541601898, "grad_norm": 0.11700495332479477, "learning_rate": 9.752570451164122e-06, "loss": 0.0029, "step": 46570 }, { "epoch": 0.7621696801112656, "grad_norm": 0.05372490733861923, "learning_rate": 9.752348590792144e-06, "loss": 0.0022, "step": 46580 }, { "epoch": 0.7623333060623415, "grad_norm": 0.039416879415512085, "learning_rate": 9.75212663352382e-06, "loss": 0.0033, "step": 46590 }, { "epoch": 0.7624969320134173, "grad_norm": 0.03207907825708389, "learning_rate": 9.751904579363673e-06, "loss": 0.0027, "step": 46600 }, { "epoch": 0.7626605579644932, "grad_norm": 0.21323177218437195, "learning_rate": 9.751682428316234e-06, "loss": 0.0057, "step": 46610 }, { "epoch": 0.762824183915569, "grad_norm": 0.08379209041595459, "learning_rate": 9.751460180386032e-06, "loss": 0.0038, "step": 46620 }, { "epoch": 0.7629878098666448, "grad_norm": 0.07013723254203796, "learning_rate": 9.751237835577596e-06, "loss": 0.0023, "step": 46630 }, { "epoch": 0.7631514358177207, "grad_norm": 0.16795386373996735, "learning_rate": 9.751015393895465e-06, "loss": 0.0029, "step": 46640 }, { "epoch": 0.7633150617687965, "grad_norm": 0.32339537143707275, "learning_rate": 9.750792855344166e-06, "loss": 0.0015, "step": 46650 }, { "epoch": 0.7634786877198724, "grad_norm": 0.05880977585911751, "learning_rate": 9.750570219928245e-06, "loss": 0.003, "step": 46660 }, { "epoch": 0.7636423136709483, "grad_norm": 0.14274504780769348, "learning_rate": 9.750347487652237e-06, "loss": 0.0019, "step": 46670 }, { "epoch": 0.763805939622024, "grad_norm": 0.07184871286153793, "learning_rate": 9.750124658520682e-06, "loss": 0.003, "step": 46680 }, { "epoch": 0.7639695655730999, "grad_norm": 0.09070820361375809, "learning_rate": 9.749901732538125e-06, "loss": 0.0021, "step": 46690 }, { "epoch": 0.7641331915241757, "grad_norm": 0.07012701034545898, "learning_rate": 9.749678709709112e-06, "loss": 0.003, "step": 46700 }, { "epoch": 0.7642968174752516, "grad_norm": 0.10238684713840485, "learning_rate": 9.74945559003819e-06, "loss": 0.0027, "step": 46710 }, { "epoch": 0.7644604434263275, "grad_norm": 0.209114208817482, "learning_rate": 9.749232373529907e-06, "loss": 0.003, "step": 46720 }, { "epoch": 0.7646240693774032, "grad_norm": 0.040122851729393005, "learning_rate": 9.749009060188817e-06, "loss": 0.002, "step": 46730 }, { "epoch": 0.7647876953284791, "grad_norm": 0.16671933233737946, "learning_rate": 9.748785650019468e-06, "loss": 0.0032, "step": 46740 }, { "epoch": 0.7649513212795549, "grad_norm": 0.04585683345794678, "learning_rate": 9.74856214302642e-06, "loss": 0.0022, "step": 46750 }, { "epoch": 0.7651149472306308, "grad_norm": 0.1562906801700592, "learning_rate": 9.748338539214229e-06, "loss": 0.0029, "step": 46760 }, { "epoch": 0.7652785731817067, "grad_norm": 0.13379500806331635, "learning_rate": 9.748114838587452e-06, "loss": 0.0038, "step": 46770 }, { "epoch": 0.7654421991327824, "grad_norm": 0.6217193603515625, "learning_rate": 9.747891041150654e-06, "loss": 0.0041, "step": 46780 }, { "epoch": 0.7656058250838583, "grad_norm": 0.1037740558385849, "learning_rate": 9.747667146908393e-06, "loss": 0.003, "step": 46790 }, { "epoch": 0.7657694510349341, "grad_norm": 0.25511634349823, "learning_rate": 9.747443155865238e-06, "loss": 0.0027, "step": 46800 }, { "epoch": 0.76593307698601, "grad_norm": 0.18197022378444672, "learning_rate": 9.747219068025755e-06, "loss": 0.0018, "step": 46810 }, { "epoch": 0.7660967029370859, "grad_norm": 0.06601916253566742, "learning_rate": 9.746994883394512e-06, "loss": 0.0041, "step": 46820 }, { "epoch": 0.7662603288881616, "grad_norm": 0.12055713683366776, "learning_rate": 9.74677060197608e-06, "loss": 0.0025, "step": 46830 }, { "epoch": 0.7664239548392375, "grad_norm": 0.0165539663285017, "learning_rate": 9.746546223775032e-06, "loss": 0.0049, "step": 46840 }, { "epoch": 0.7665875807903133, "grad_norm": 0.0450606569647789, "learning_rate": 9.746321748795944e-06, "loss": 0.0017, "step": 46850 }, { "epoch": 0.7667512067413892, "grad_norm": 0.06225024163722992, "learning_rate": 9.746097177043393e-06, "loss": 0.0027, "step": 46860 }, { "epoch": 0.7669148326924651, "grad_norm": 0.13746477663516998, "learning_rate": 9.745872508521955e-06, "loss": 0.003, "step": 46870 }, { "epoch": 0.7670784586435409, "grad_norm": 0.09262300282716751, "learning_rate": 9.745647743236216e-06, "loss": 0.0025, "step": 46880 }, { "epoch": 0.7672420845946167, "grad_norm": 0.15985800325870514, "learning_rate": 9.745422881190752e-06, "loss": 0.0027, "step": 46890 }, { "epoch": 0.7674057105456925, "grad_norm": 0.18650297820568085, "learning_rate": 9.745197922390153e-06, "loss": 0.0032, "step": 46900 }, { "epoch": 0.7675693364967684, "grad_norm": 0.054992835968732834, "learning_rate": 9.744972866839003e-06, "loss": 0.0025, "step": 46910 }, { "epoch": 0.7677329624478443, "grad_norm": 0.11908558011054993, "learning_rate": 9.744747714541894e-06, "loss": 0.0027, "step": 46920 }, { "epoch": 0.76789658839892, "grad_norm": 0.05694097280502319, "learning_rate": 9.744522465503412e-06, "loss": 0.0032, "step": 46930 }, { "epoch": 0.7680602143499959, "grad_norm": 0.16750960052013397, "learning_rate": 9.744297119728152e-06, "loss": 0.0041, "step": 46940 }, { "epoch": 0.7682238403010717, "grad_norm": 0.054366596043109894, "learning_rate": 9.74407167722071e-06, "loss": 0.0034, "step": 46950 }, { "epoch": 0.7683874662521476, "grad_norm": 0.1750924438238144, "learning_rate": 9.74384613798568e-06, "loss": 0.0045, "step": 46960 }, { "epoch": 0.7685510922032235, "grad_norm": 0.06236717849969864, "learning_rate": 9.743620502027662e-06, "loss": 0.0021, "step": 46970 }, { "epoch": 0.7687147181542993, "grad_norm": 0.049553852528333664, "learning_rate": 9.743394769351258e-06, "loss": 0.002, "step": 46980 }, { "epoch": 0.7688783441053751, "grad_norm": 0.09212444722652435, "learning_rate": 9.743168939961067e-06, "loss": 0.0021, "step": 46990 }, { "epoch": 0.7690419700564509, "grad_norm": 0.029286758974194527, "learning_rate": 9.742943013861695e-06, "loss": 0.0032, "step": 47000 }, { "epoch": 0.7692055960075268, "grad_norm": 0.11678091436624527, "learning_rate": 9.74271699105775e-06, "loss": 0.002, "step": 47010 }, { "epoch": 0.7693692219586027, "grad_norm": 0.013645494356751442, "learning_rate": 9.742490871553837e-06, "loss": 0.0022, "step": 47020 }, { "epoch": 0.7695328479096785, "grad_norm": 0.11028126627206802, "learning_rate": 9.742264655354571e-06, "loss": 0.0039, "step": 47030 }, { "epoch": 0.7696964738607543, "grad_norm": 0.06393654644489288, "learning_rate": 9.74203834246456e-06, "loss": 0.0026, "step": 47040 }, { "epoch": 0.7698600998118301, "grad_norm": 0.06776327639818192, "learning_rate": 9.74181193288842e-06, "loss": 0.0068, "step": 47050 }, { "epoch": 0.770023725762906, "grad_norm": 0.1270781010389328, "learning_rate": 9.74158542663077e-06, "loss": 0.002, "step": 47060 }, { "epoch": 0.7701873517139818, "grad_norm": 0.05257631465792656, "learning_rate": 9.741358823696224e-06, "loss": 0.0036, "step": 47070 }, { "epoch": 0.7703509776650577, "grad_norm": 0.04238073155283928, "learning_rate": 9.741132124089403e-06, "loss": 0.0019, "step": 47080 }, { "epoch": 0.7705146036161336, "grad_norm": 0.06982503086328506, "learning_rate": 9.740905327814931e-06, "loss": 0.0025, "step": 47090 }, { "epoch": 0.7706782295672093, "grad_norm": 0.037962909787893295, "learning_rate": 9.740678434877433e-06, "loss": 0.0032, "step": 47100 }, { "epoch": 0.7708418555182852, "grad_norm": 0.021394450217485428, "learning_rate": 9.740451445281531e-06, "loss": 0.0022, "step": 47110 }, { "epoch": 0.771005481469361, "grad_norm": 0.04623113572597504, "learning_rate": 9.740224359031858e-06, "loss": 0.0024, "step": 47120 }, { "epoch": 0.7711691074204369, "grad_norm": 0.041595038026571274, "learning_rate": 9.73999717613304e-06, "loss": 0.0043, "step": 47130 }, { "epoch": 0.7713327333715128, "grad_norm": 0.12272101640701294, "learning_rate": 9.73976989658971e-06, "loss": 0.0025, "step": 47140 }, { "epoch": 0.7714963593225885, "grad_norm": 0.07439558953046799, "learning_rate": 9.739542520406504e-06, "loss": 0.0032, "step": 47150 }, { "epoch": 0.7716599852736644, "grad_norm": 0.0967576876282692, "learning_rate": 9.739315047588059e-06, "loss": 0.0021, "step": 47160 }, { "epoch": 0.7718236112247402, "grad_norm": 0.0821041390299797, "learning_rate": 9.739087478139008e-06, "loss": 0.0019, "step": 47170 }, { "epoch": 0.7719872371758161, "grad_norm": 0.07549381256103516, "learning_rate": 9.738859812063994e-06, "loss": 0.0019, "step": 47180 }, { "epoch": 0.772150863126892, "grad_norm": 0.19292210042476654, "learning_rate": 9.73863204936766e-06, "loss": 0.003, "step": 47190 }, { "epoch": 0.7723144890779677, "grad_norm": 0.20088626444339752, "learning_rate": 9.738404190054646e-06, "loss": 0.0037, "step": 47200 }, { "epoch": 0.7724781150290436, "grad_norm": 0.14525693655014038, "learning_rate": 9.738176234129604e-06, "loss": 0.002, "step": 47210 }, { "epoch": 0.7726417409801194, "grad_norm": 0.060114648193120956, "learning_rate": 9.737948181597176e-06, "loss": 0.0025, "step": 47220 }, { "epoch": 0.7728053669311953, "grad_norm": 0.347993940114975, "learning_rate": 9.737720032462014e-06, "loss": 0.0034, "step": 47230 }, { "epoch": 0.7729689928822712, "grad_norm": 0.25476476550102234, "learning_rate": 9.73749178672877e-06, "loss": 0.003, "step": 47240 }, { "epoch": 0.7731326188333469, "grad_norm": 0.13154540956020355, "learning_rate": 9.737263444402098e-06, "loss": 0.0031, "step": 47250 }, { "epoch": 0.7732962447844228, "grad_norm": 0.10499826073646545, "learning_rate": 9.737035005486653e-06, "loss": 0.0016, "step": 47260 }, { "epoch": 0.7734598707354986, "grad_norm": 0.31142866611480713, "learning_rate": 9.736806469987094e-06, "loss": 0.0044, "step": 47270 }, { "epoch": 0.7736234966865745, "grad_norm": 0.12050152570009232, "learning_rate": 9.73657783790808e-06, "loss": 0.0024, "step": 47280 }, { "epoch": 0.7737871226376504, "grad_norm": 0.04629393666982651, "learning_rate": 9.736349109254268e-06, "loss": 0.004, "step": 47290 }, { "epoch": 0.7739507485887261, "grad_norm": 0.03307654708623886, "learning_rate": 9.73612028403033e-06, "loss": 0.003, "step": 47300 }, { "epoch": 0.774114374539802, "grad_norm": 0.08769938349723816, "learning_rate": 9.735891362240926e-06, "loss": 0.0016, "step": 47310 }, { "epoch": 0.7742780004908778, "grad_norm": 0.1961381584405899, "learning_rate": 9.735662343890723e-06, "loss": 0.0022, "step": 47320 }, { "epoch": 0.7744416264419537, "grad_norm": 0.10523116588592529, "learning_rate": 9.735433228984397e-06, "loss": 0.0028, "step": 47330 }, { "epoch": 0.7746052523930296, "grad_norm": 0.05042312294244766, "learning_rate": 9.735204017526611e-06, "loss": 0.0032, "step": 47340 }, { "epoch": 0.7747688783441053, "grad_norm": 0.14206130802631378, "learning_rate": 9.734974709522042e-06, "loss": 0.0026, "step": 47350 }, { "epoch": 0.7749325042951812, "grad_norm": 0.06831593066453934, "learning_rate": 9.734745304975368e-06, "loss": 0.0021, "step": 47360 }, { "epoch": 0.775096130246257, "grad_norm": 0.015889719128608704, "learning_rate": 9.734515803891263e-06, "loss": 0.0025, "step": 47370 }, { "epoch": 0.7752597561973329, "grad_norm": 0.01943323016166687, "learning_rate": 9.734286206274408e-06, "loss": 0.0021, "step": 47380 }, { "epoch": 0.7754233821484088, "grad_norm": 0.07826309651136398, "learning_rate": 9.734056512129481e-06, "loss": 0.0017, "step": 47390 }, { "epoch": 0.7755870080994846, "grad_norm": 0.09219307452440262, "learning_rate": 9.733826721461171e-06, "loss": 0.0022, "step": 47400 }, { "epoch": 0.7757506340505604, "grad_norm": 0.08384527266025543, "learning_rate": 9.733596834274158e-06, "loss": 0.0038, "step": 47410 }, { "epoch": 0.7759142600016362, "grad_norm": 0.10139086097478867, "learning_rate": 9.733366850573133e-06, "loss": 0.0022, "step": 47420 }, { "epoch": 0.7760778859527121, "grad_norm": 0.0739239826798439, "learning_rate": 9.733136770362783e-06, "loss": 0.002, "step": 47430 }, { "epoch": 0.776241511903788, "grad_norm": 0.29241400957107544, "learning_rate": 9.732906593647799e-06, "loss": 0.0037, "step": 47440 }, { "epoch": 0.7764051378548638, "grad_norm": 0.15663881599903107, "learning_rate": 9.732676320432874e-06, "loss": 0.0025, "step": 47450 }, { "epoch": 0.7765687638059396, "grad_norm": 0.19817429780960083, "learning_rate": 9.732445950722706e-06, "loss": 0.0046, "step": 47460 }, { "epoch": 0.7767323897570154, "grad_norm": 0.06631200760602951, "learning_rate": 9.732215484521988e-06, "loss": 0.0037, "step": 47470 }, { "epoch": 0.7768960157080913, "grad_norm": 0.12339597195386887, "learning_rate": 9.731984921835422e-06, "loss": 0.002, "step": 47480 }, { "epoch": 0.7770596416591672, "grad_norm": 0.09765121340751648, "learning_rate": 9.731754262667707e-06, "loss": 0.0043, "step": 47490 }, { "epoch": 0.777223267610243, "grad_norm": 0.09059354662895203, "learning_rate": 9.731523507023547e-06, "loss": 0.0019, "step": 47500 }, { "epoch": 0.7773868935613188, "grad_norm": 0.04670633003115654, "learning_rate": 9.731292654907648e-06, "loss": 0.002, "step": 47510 }, { "epoch": 0.7775505195123946, "grad_norm": 0.05789300054311752, "learning_rate": 9.731061706324714e-06, "loss": 0.0021, "step": 47520 }, { "epoch": 0.7777141454634705, "grad_norm": 0.09849626570940018, "learning_rate": 9.730830661279456e-06, "loss": 0.0033, "step": 47530 }, { "epoch": 0.7778777714145464, "grad_norm": 0.13100282847881317, "learning_rate": 9.730599519776584e-06, "loss": 0.0051, "step": 47540 }, { "epoch": 0.7780413973656222, "grad_norm": 0.04953023046255112, "learning_rate": 9.730368281820811e-06, "loss": 0.0022, "step": 47550 }, { "epoch": 0.778205023316698, "grad_norm": 0.07568024843931198, "learning_rate": 9.730136947416853e-06, "loss": 0.0026, "step": 47560 }, { "epoch": 0.7783686492677738, "grad_norm": 0.08602076768875122, "learning_rate": 9.729905516569425e-06, "loss": 0.0026, "step": 47570 }, { "epoch": 0.7785322752188497, "grad_norm": 0.09011707454919815, "learning_rate": 9.729673989283244e-06, "loss": 0.0029, "step": 47580 }, { "epoch": 0.7786959011699256, "grad_norm": 0.10884258151054382, "learning_rate": 9.729442365563035e-06, "loss": 0.0034, "step": 47590 }, { "epoch": 0.7788595271210014, "grad_norm": 0.1586916297674179, "learning_rate": 9.729210645413518e-06, "loss": 0.0018, "step": 47600 }, { "epoch": 0.7790231530720773, "grad_norm": 0.04442888870835304, "learning_rate": 9.728978828839417e-06, "loss": 0.0013, "step": 47610 }, { "epoch": 0.779186779023153, "grad_norm": 0.12428022921085358, "learning_rate": 9.728746915845461e-06, "loss": 0.003, "step": 47620 }, { "epoch": 0.7793504049742289, "grad_norm": 0.07956744730472565, "learning_rate": 9.728514906436378e-06, "loss": 0.0019, "step": 47630 }, { "epoch": 0.7795140309253048, "grad_norm": 0.10614572465419769, "learning_rate": 9.728282800616896e-06, "loss": 0.0029, "step": 47640 }, { "epoch": 0.7796776568763806, "grad_norm": 0.06806833297014236, "learning_rate": 9.728050598391747e-06, "loss": 0.0026, "step": 47650 }, { "epoch": 0.7798412828274565, "grad_norm": 0.09675050526857376, "learning_rate": 9.72781829976567e-06, "loss": 0.0021, "step": 47660 }, { "epoch": 0.7800049087785322, "grad_norm": 0.08166264742612839, "learning_rate": 9.727585904743398e-06, "loss": 0.0027, "step": 47670 }, { "epoch": 0.7801685347296081, "grad_norm": 0.035482924431562424, "learning_rate": 9.72735341332967e-06, "loss": 0.0037, "step": 47680 }, { "epoch": 0.780332160680684, "grad_norm": 0.11753704398870468, "learning_rate": 9.727120825529225e-06, "loss": 0.0029, "step": 47690 }, { "epoch": 0.7804957866317598, "grad_norm": 0.13314154744148254, "learning_rate": 9.72688814134681e-06, "loss": 0.0023, "step": 47700 }, { "epoch": 0.7806594125828357, "grad_norm": 0.13910514116287231, "learning_rate": 9.726655360787163e-06, "loss": 0.0042, "step": 47710 }, { "epoch": 0.7808230385339114, "grad_norm": 0.05691445991396904, "learning_rate": 9.726422483855034e-06, "loss": 0.0034, "step": 47720 }, { "epoch": 0.7809866644849873, "grad_norm": 0.12178466469049454, "learning_rate": 9.72618951055517e-06, "loss": 0.0033, "step": 47730 }, { "epoch": 0.7811502904360632, "grad_norm": 0.07652205973863602, "learning_rate": 9.72595644089232e-06, "loss": 0.0049, "step": 47740 }, { "epoch": 0.781313916387139, "grad_norm": 0.04036548361182213, "learning_rate": 9.725723274871241e-06, "loss": 0.0016, "step": 47750 }, { "epoch": 0.7814775423382149, "grad_norm": 0.05735372379422188, "learning_rate": 9.725490012496682e-06, "loss": 0.0041, "step": 47760 }, { "epoch": 0.7816411682892906, "grad_norm": 0.11945325136184692, "learning_rate": 9.7252566537734e-06, "loss": 0.0038, "step": 47770 }, { "epoch": 0.7818047942403665, "grad_norm": 0.13808202743530273, "learning_rate": 9.725023198706154e-06, "loss": 0.006, "step": 47780 }, { "epoch": 0.7819684201914424, "grad_norm": 0.05489223822951317, "learning_rate": 9.724789647299702e-06, "loss": 0.0019, "step": 47790 }, { "epoch": 0.7821320461425182, "grad_norm": 0.14344577491283417, "learning_rate": 9.724555999558809e-06, "loss": 0.0044, "step": 47800 }, { "epoch": 0.7822956720935941, "grad_norm": 0.12232688814401627, "learning_rate": 9.724322255488237e-06, "loss": 0.0043, "step": 47810 }, { "epoch": 0.7824592980446698, "grad_norm": 0.06129299849271774, "learning_rate": 9.724088415092752e-06, "loss": 0.0037, "step": 47820 }, { "epoch": 0.7826229239957457, "grad_norm": 0.04617205634713173, "learning_rate": 9.723854478377122e-06, "loss": 0.0016, "step": 47830 }, { "epoch": 0.7827865499468216, "grad_norm": 0.04711601510643959, "learning_rate": 9.723620445346116e-06, "loss": 0.0029, "step": 47840 }, { "epoch": 0.7829501758978974, "grad_norm": 0.05082441866397858, "learning_rate": 9.723386316004507e-06, "loss": 0.0027, "step": 47850 }, { "epoch": 0.7831138018489733, "grad_norm": 0.26346391439437866, "learning_rate": 9.723152090357066e-06, "loss": 0.0022, "step": 47860 }, { "epoch": 0.783277427800049, "grad_norm": 0.10622677206993103, "learning_rate": 9.722917768408574e-06, "loss": 0.003, "step": 47870 }, { "epoch": 0.7834410537511249, "grad_norm": 0.09940803050994873, "learning_rate": 9.722683350163804e-06, "loss": 0.003, "step": 47880 }, { "epoch": 0.7836046797022008, "grad_norm": 0.20736342668533325, "learning_rate": 9.722448835627537e-06, "loss": 0.0018, "step": 47890 }, { "epoch": 0.7837683056532766, "grad_norm": 0.10485286265611649, "learning_rate": 9.722214224804555e-06, "loss": 0.0027, "step": 47900 }, { "epoch": 0.7839319316043525, "grad_norm": 0.10018691420555115, "learning_rate": 9.72197951769964e-06, "loss": 0.0028, "step": 47910 }, { "epoch": 0.7840955575554283, "grad_norm": 0.030494332313537598, "learning_rate": 9.72174471431758e-06, "loss": 0.0024, "step": 47920 }, { "epoch": 0.7842591835065041, "grad_norm": 0.033588893711566925, "learning_rate": 9.72150981466316e-06, "loss": 0.0027, "step": 47930 }, { "epoch": 0.7844228094575799, "grad_norm": 0.04430209472775459, "learning_rate": 9.721274818741171e-06, "loss": 0.0059, "step": 47940 }, { "epoch": 0.7845864354086558, "grad_norm": 0.08697672933340073, "learning_rate": 9.721039726556402e-06, "loss": 0.004, "step": 47950 }, { "epoch": 0.7847500613597317, "grad_norm": 0.026969779282808304, "learning_rate": 9.720804538113651e-06, "loss": 0.0022, "step": 47960 }, { "epoch": 0.7849136873108075, "grad_norm": 0.12007205188274384, "learning_rate": 9.720569253417707e-06, "loss": 0.003, "step": 47970 }, { "epoch": 0.7850773132618833, "grad_norm": 0.3360818326473236, "learning_rate": 9.720333872473373e-06, "loss": 0.003, "step": 47980 }, { "epoch": 0.7852409392129591, "grad_norm": 0.09613803029060364, "learning_rate": 9.720098395285445e-06, "loss": 0.0032, "step": 47990 }, { "epoch": 0.785404565164035, "grad_norm": 0.12578800320625305, "learning_rate": 9.719862821858726e-06, "loss": 0.0036, "step": 48000 }, { "epoch": 0.7855681911151109, "grad_norm": 0.31998902559280396, "learning_rate": 9.719627152198017e-06, "loss": 0.0032, "step": 48010 }, { "epoch": 0.7857318170661867, "grad_norm": 0.05854915827512741, "learning_rate": 9.719391386308125e-06, "loss": 0.0022, "step": 48020 }, { "epoch": 0.7858954430172626, "grad_norm": 0.1067088320851326, "learning_rate": 9.719155524193855e-06, "loss": 0.0029, "step": 48030 }, { "epoch": 0.7860590689683383, "grad_norm": 0.036176647990942, "learning_rate": 9.71891956586002e-06, "loss": 0.0032, "step": 48040 }, { "epoch": 0.7862226949194142, "grad_norm": 0.07329155504703522, "learning_rate": 9.718683511311426e-06, "loss": 0.0018, "step": 48050 }, { "epoch": 0.7863863208704901, "grad_norm": 0.09733510762453079, "learning_rate": 9.71844736055289e-06, "loss": 0.0014, "step": 48060 }, { "epoch": 0.7865499468215659, "grad_norm": 0.12931400537490845, "learning_rate": 9.718211113589224e-06, "loss": 0.0031, "step": 48070 }, { "epoch": 0.7867135727726418, "grad_norm": 0.07886756211519241, "learning_rate": 9.717974770425246e-06, "loss": 0.0022, "step": 48080 }, { "epoch": 0.7868771987237175, "grad_norm": 0.07586278021335602, "learning_rate": 9.717738331065774e-06, "loss": 0.0019, "step": 48090 }, { "epoch": 0.7870408246747934, "grad_norm": 0.044749170541763306, "learning_rate": 9.717501795515632e-06, "loss": 0.0023, "step": 48100 }, { "epoch": 0.7872044506258693, "grad_norm": 0.10603630542755127, "learning_rate": 9.71726516377964e-06, "loss": 0.0017, "step": 48110 }, { "epoch": 0.7873680765769451, "grad_norm": 0.04047068580985069, "learning_rate": 9.717028435862623e-06, "loss": 0.0032, "step": 48120 }, { "epoch": 0.787531702528021, "grad_norm": 0.00767413480207324, "learning_rate": 9.716791611769407e-06, "loss": 0.0019, "step": 48130 }, { "epoch": 0.7876953284790967, "grad_norm": 0.27833443880081177, "learning_rate": 9.716554691504822e-06, "loss": 0.0017, "step": 48140 }, { "epoch": 0.7878589544301726, "grad_norm": 0.05714798718690872, "learning_rate": 9.716317675073697e-06, "loss": 0.002, "step": 48150 }, { "epoch": 0.7880225803812485, "grad_norm": 0.12286387383937836, "learning_rate": 9.716080562480867e-06, "loss": 0.0029, "step": 48160 }, { "epoch": 0.7881862063323243, "grad_norm": 0.1094760149717331, "learning_rate": 9.715843353731166e-06, "loss": 0.0026, "step": 48170 }, { "epoch": 0.7883498322834002, "grad_norm": 0.11086010932922363, "learning_rate": 9.715606048829429e-06, "loss": 0.0027, "step": 48180 }, { "epoch": 0.7885134582344759, "grad_norm": 0.02748928591609001, "learning_rate": 9.715368647780494e-06, "loss": 0.0021, "step": 48190 }, { "epoch": 0.7886770841855518, "grad_norm": 0.025180891156196594, "learning_rate": 9.715131150589203e-06, "loss": 0.0042, "step": 48200 }, { "epoch": 0.7888407101366277, "grad_norm": 0.07970261573791504, "learning_rate": 9.7148935572604e-06, "loss": 0.0028, "step": 48210 }, { "epoch": 0.7890043360877035, "grad_norm": 0.09481535851955414, "learning_rate": 9.714655867798926e-06, "loss": 0.0015, "step": 48220 }, { "epoch": 0.7891679620387794, "grad_norm": 0.02162362076342106, "learning_rate": 9.714418082209627e-06, "loss": 0.003, "step": 48230 }, { "epoch": 0.7893315879898551, "grad_norm": 0.05748597905039787, "learning_rate": 9.714180200497353e-06, "loss": 0.0026, "step": 48240 }, { "epoch": 0.789495213940931, "grad_norm": 0.05273944512009621, "learning_rate": 9.713942222666957e-06, "loss": 0.0022, "step": 48250 }, { "epoch": 0.7896588398920069, "grad_norm": 0.010867610573768616, "learning_rate": 9.713704148723286e-06, "loss": 0.0039, "step": 48260 }, { "epoch": 0.7898224658430827, "grad_norm": 0.06923382729291916, "learning_rate": 9.713465978671195e-06, "loss": 0.003, "step": 48270 }, { "epoch": 0.7899860917941586, "grad_norm": 0.09379014372825623, "learning_rate": 9.713227712515543e-06, "loss": 0.0025, "step": 48280 }, { "epoch": 0.7901497177452343, "grad_norm": 0.06543421000242233, "learning_rate": 9.712989350261187e-06, "loss": 0.0022, "step": 48290 }, { "epoch": 0.7903133436963102, "grad_norm": 0.16828572750091553, "learning_rate": 9.712750891912986e-06, "loss": 0.0043, "step": 48300 }, { "epoch": 0.7904769696473861, "grad_norm": 0.03354249894618988, "learning_rate": 9.712512337475802e-06, "loss": 0.0022, "step": 48310 }, { "epoch": 0.7906405955984619, "grad_norm": 0.1578199863433838, "learning_rate": 9.712273686954498e-06, "loss": 0.0036, "step": 48320 }, { "epoch": 0.7908042215495378, "grad_norm": 0.041253529489040375, "learning_rate": 9.712034940353943e-06, "loss": 0.0025, "step": 48330 }, { "epoch": 0.7909678475006136, "grad_norm": 0.04498960077762604, "learning_rate": 9.711796097679e-06, "loss": 0.0036, "step": 48340 }, { "epoch": 0.7911314734516894, "grad_norm": 0.15542519092559814, "learning_rate": 9.711557158934546e-06, "loss": 0.0039, "step": 48350 }, { "epoch": 0.7912950994027653, "grad_norm": 0.05623500421643257, "learning_rate": 9.711318124125445e-06, "loss": 0.0036, "step": 48360 }, { "epoch": 0.7914587253538411, "grad_norm": 0.07412418723106384, "learning_rate": 9.711078993256578e-06, "loss": 0.004, "step": 48370 }, { "epoch": 0.791622351304917, "grad_norm": 0.15220195055007935, "learning_rate": 9.710839766332814e-06, "loss": 0.0034, "step": 48380 }, { "epoch": 0.7917859772559928, "grad_norm": 0.08521812409162521, "learning_rate": 9.710600443359035e-06, "loss": 0.0036, "step": 48390 }, { "epoch": 0.7919496032070686, "grad_norm": 0.048148609697818756, "learning_rate": 9.710361024340118e-06, "loss": 0.0016, "step": 48400 }, { "epoch": 0.7921132291581445, "grad_norm": 0.1092853844165802, "learning_rate": 9.710121509280946e-06, "loss": 0.0024, "step": 48410 }, { "epoch": 0.7922768551092203, "grad_norm": 0.05501531437039375, "learning_rate": 9.709881898186403e-06, "loss": 0.002, "step": 48420 }, { "epoch": 0.7924404810602962, "grad_norm": 0.045999232679605484, "learning_rate": 9.709642191061374e-06, "loss": 0.0028, "step": 48430 }, { "epoch": 0.792604107011372, "grad_norm": 0.21159063279628754, "learning_rate": 9.709402387910745e-06, "loss": 0.0029, "step": 48440 }, { "epoch": 0.7927677329624478, "grad_norm": 0.041300561279058456, "learning_rate": 9.709162488739408e-06, "loss": 0.0016, "step": 48450 }, { "epoch": 0.7929313589135237, "grad_norm": 0.05519705265760422, "learning_rate": 9.70892249355225e-06, "loss": 0.0018, "step": 48460 }, { "epoch": 0.7930949848645995, "grad_norm": 0.07458744198083878, "learning_rate": 9.70868240235417e-06, "loss": 0.0023, "step": 48470 }, { "epoch": 0.7932586108156754, "grad_norm": 0.31479743123054504, "learning_rate": 9.70844221515006e-06, "loss": 0.0019, "step": 48480 }, { "epoch": 0.7934222367667512, "grad_norm": 0.06033928319811821, "learning_rate": 9.708201931944816e-06, "loss": 0.0033, "step": 48490 }, { "epoch": 0.793585862717827, "grad_norm": 0.14628978073596954, "learning_rate": 9.707961552743341e-06, "loss": 0.0035, "step": 48500 }, { "epoch": 0.7937494886689029, "grad_norm": 0.053767867386341095, "learning_rate": 9.707721077550531e-06, "loss": 0.0029, "step": 48510 }, { "epoch": 0.7939131146199787, "grad_norm": 0.18293796479701996, "learning_rate": 9.707480506371295e-06, "loss": 0.0067, "step": 48520 }, { "epoch": 0.7940767405710546, "grad_norm": 0.11483065783977509, "learning_rate": 9.707239839210534e-06, "loss": 0.0016, "step": 48530 }, { "epoch": 0.7942403665221304, "grad_norm": 0.13869036734104156, "learning_rate": 9.706999076073157e-06, "loss": 0.0031, "step": 48540 }, { "epoch": 0.7944039924732063, "grad_norm": 0.1027173101902008, "learning_rate": 9.70675821696407e-06, "loss": 0.0039, "step": 48550 }, { "epoch": 0.7945676184242821, "grad_norm": 0.09573974460363388, "learning_rate": 9.706517261888187e-06, "loss": 0.0028, "step": 48560 }, { "epoch": 0.7947312443753579, "grad_norm": 0.09659865498542786, "learning_rate": 9.706276210850419e-06, "loss": 0.0033, "step": 48570 }, { "epoch": 0.7948948703264338, "grad_norm": 0.031234873458743095, "learning_rate": 9.706035063855682e-06, "loss": 0.0023, "step": 48580 }, { "epoch": 0.7950584962775096, "grad_norm": 0.0507713221013546, "learning_rate": 9.705793820908893e-06, "loss": 0.0019, "step": 48590 }, { "epoch": 0.7952221222285855, "grad_norm": 0.14141270518302917, "learning_rate": 9.705552482014969e-06, "loss": 0.0021, "step": 48600 }, { "epoch": 0.7953857481796613, "grad_norm": 0.023594321683049202, "learning_rate": 9.705311047178832e-06, "loss": 0.0017, "step": 48610 }, { "epoch": 0.7955493741307371, "grad_norm": 0.14653995633125305, "learning_rate": 9.705069516405405e-06, "loss": 0.0022, "step": 48620 }, { "epoch": 0.795713000081813, "grad_norm": 0.23661306500434875, "learning_rate": 9.704827889699612e-06, "loss": 0.0025, "step": 48630 }, { "epoch": 0.7958766260328888, "grad_norm": 0.21054713428020477, "learning_rate": 9.704586167066382e-06, "loss": 0.0022, "step": 48640 }, { "epoch": 0.7960402519839647, "grad_norm": 0.27099117636680603, "learning_rate": 9.704344348510637e-06, "loss": 0.0035, "step": 48650 }, { "epoch": 0.7962038779350405, "grad_norm": 0.06323729455471039, "learning_rate": 9.704102434037314e-06, "loss": 0.0036, "step": 48660 }, { "epoch": 0.7963675038861163, "grad_norm": 0.0648345947265625, "learning_rate": 9.703860423651343e-06, "loss": 0.0026, "step": 48670 }, { "epoch": 0.7965311298371922, "grad_norm": 0.022481955587863922, "learning_rate": 9.703618317357657e-06, "loss": 0.0033, "step": 48680 }, { "epoch": 0.796694755788268, "grad_norm": 0.03263377398252487, "learning_rate": 9.703376115161195e-06, "loss": 0.0027, "step": 48690 }, { "epoch": 0.7968583817393439, "grad_norm": 0.3201417922973633, "learning_rate": 9.703133817066894e-06, "loss": 0.0057, "step": 48700 }, { "epoch": 0.7970220076904198, "grad_norm": 0.35108059644699097, "learning_rate": 9.702891423079695e-06, "loss": 0.0078, "step": 48710 }, { "epoch": 0.7971856336414955, "grad_norm": 0.05510510876774788, "learning_rate": 9.702648933204537e-06, "loss": 0.0036, "step": 48720 }, { "epoch": 0.7973492595925714, "grad_norm": 0.06543893367052078, "learning_rate": 9.70240634744637e-06, "loss": 0.0025, "step": 48730 }, { "epoch": 0.7975128855436472, "grad_norm": 0.05259516090154648, "learning_rate": 9.702163665810135e-06, "loss": 0.0029, "step": 48740 }, { "epoch": 0.7976765114947231, "grad_norm": 0.20444706082344055, "learning_rate": 9.701920888300783e-06, "loss": 0.0027, "step": 48750 }, { "epoch": 0.797840137445799, "grad_norm": 0.15156283974647522, "learning_rate": 9.70167801492326e-06, "loss": 0.0028, "step": 48760 }, { "epoch": 0.7980037633968747, "grad_norm": 0.04095473885536194, "learning_rate": 9.701435045682524e-06, "loss": 0.0022, "step": 48770 }, { "epoch": 0.7981673893479506, "grad_norm": 0.0713505819439888, "learning_rate": 9.701191980583524e-06, "loss": 0.0042, "step": 48780 }, { "epoch": 0.7983310152990264, "grad_norm": 0.33138391375541687, "learning_rate": 9.70094881963122e-06, "loss": 0.0024, "step": 48790 }, { "epoch": 0.7984946412501023, "grad_norm": 0.04352826997637749, "learning_rate": 9.700705562830566e-06, "loss": 0.0021, "step": 48800 }, { "epoch": 0.798658267201178, "grad_norm": 0.09954825043678284, "learning_rate": 9.700462210186524e-06, "loss": 0.0022, "step": 48810 }, { "epoch": 0.7988218931522539, "grad_norm": 0.07862277328968048, "learning_rate": 9.700218761704054e-06, "loss": 0.0015, "step": 48820 }, { "epoch": 0.7989855191033298, "grad_norm": 0.10388215631246567, "learning_rate": 9.699975217388123e-06, "loss": 0.005, "step": 48830 }, { "epoch": 0.7991491450544056, "grad_norm": 0.15805985033512115, "learning_rate": 9.699731577243692e-06, "loss": 0.0046, "step": 48840 }, { "epoch": 0.7993127710054815, "grad_norm": 0.10244891792535782, "learning_rate": 9.699487841275732e-06, "loss": 0.0021, "step": 48850 }, { "epoch": 0.7994763969565573, "grad_norm": 0.20062406361103058, "learning_rate": 9.69924400948921e-06, "loss": 0.0029, "step": 48860 }, { "epoch": 0.7996400229076331, "grad_norm": 0.08950719982385635, "learning_rate": 9.699000081889102e-06, "loss": 0.0025, "step": 48870 }, { "epoch": 0.799803648858709, "grad_norm": 0.05560501292347908, "learning_rate": 9.698756058480378e-06, "loss": 0.0024, "step": 48880 }, { "epoch": 0.7999672748097848, "grad_norm": 0.05409432575106621, "learning_rate": 9.698511939268014e-06, "loss": 0.0029, "step": 48890 }, { "epoch": 0.8001309007608607, "grad_norm": 0.0680970698595047, "learning_rate": 9.698267724256988e-06, "loss": 0.0028, "step": 48900 }, { "epoch": 0.8002945267119365, "grad_norm": 0.12141153216362, "learning_rate": 9.698023413452278e-06, "loss": 0.0025, "step": 48910 }, { "epoch": 0.8004581526630123, "grad_norm": 0.08561242371797562, "learning_rate": 9.697779006858866e-06, "loss": 0.0032, "step": 48920 }, { "epoch": 0.8006217786140882, "grad_norm": 0.06284994632005692, "learning_rate": 9.697534504481735e-06, "loss": 0.0028, "step": 48930 }, { "epoch": 0.800785404565164, "grad_norm": 0.12991926074028015, "learning_rate": 9.697289906325873e-06, "loss": 0.0031, "step": 48940 }, { "epoch": 0.8009490305162399, "grad_norm": 0.03884465619921684, "learning_rate": 9.697045212396263e-06, "loss": 0.0013, "step": 48950 }, { "epoch": 0.8011126564673157, "grad_norm": 0.07366728782653809, "learning_rate": 9.696800422697896e-06, "loss": 0.0027, "step": 48960 }, { "epoch": 0.8012762824183915, "grad_norm": 0.2973185181617737, "learning_rate": 9.696555537235761e-06, "loss": 0.0056, "step": 48970 }, { "epoch": 0.8014399083694674, "grad_norm": 0.04394815117120743, "learning_rate": 9.696310556014856e-06, "loss": 0.0015, "step": 48980 }, { "epoch": 0.8016035343205432, "grad_norm": 0.003998989704996347, "learning_rate": 9.69606547904017e-06, "loss": 0.0019, "step": 48990 }, { "epoch": 0.8017671602716191, "grad_norm": 0.1085531935095787, "learning_rate": 9.695820306316705e-06, "loss": 0.0023, "step": 49000 }, { "epoch": 0.8019307862226949, "grad_norm": 0.019548537209630013, "learning_rate": 9.695575037849456e-06, "loss": 0.0041, "step": 49010 }, { "epoch": 0.8020944121737708, "grad_norm": 0.025338442996144295, "learning_rate": 9.695329673643427e-06, "loss": 0.0023, "step": 49020 }, { "epoch": 0.8022580381248466, "grad_norm": 0.04242786765098572, "learning_rate": 9.695084213703618e-06, "loss": 0.0046, "step": 49030 }, { "epoch": 0.8024216640759224, "grad_norm": 0.23687203228473663, "learning_rate": 9.694838658035034e-06, "loss": 0.0037, "step": 49040 }, { "epoch": 0.8025852900269983, "grad_norm": 0.12351427227258682, "learning_rate": 9.694593006642686e-06, "loss": 0.0036, "step": 49050 }, { "epoch": 0.8027489159780741, "grad_norm": 0.1299629509449005, "learning_rate": 9.694347259531576e-06, "loss": 0.0028, "step": 49060 }, { "epoch": 0.80291254192915, "grad_norm": 0.08962411433458328, "learning_rate": 9.694101416706719e-06, "loss": 0.002, "step": 49070 }, { "epoch": 0.8030761678802258, "grad_norm": 0.20519688725471497, "learning_rate": 9.693855478173127e-06, "loss": 0.0024, "step": 49080 }, { "epoch": 0.8032397938313016, "grad_norm": 0.08468257635831833, "learning_rate": 9.693609443935812e-06, "loss": 0.0017, "step": 49090 }, { "epoch": 0.8034034197823775, "grad_norm": 0.04861167445778847, "learning_rate": 9.693363313999792e-06, "loss": 0.0036, "step": 49100 }, { "epoch": 0.8035670457334533, "grad_norm": 0.22557592391967773, "learning_rate": 9.693117088370087e-06, "loss": 0.0019, "step": 49110 }, { "epoch": 0.8037306716845292, "grad_norm": 0.13949231803417206, "learning_rate": 9.692870767051717e-06, "loss": 0.0028, "step": 49120 }, { "epoch": 0.803894297635605, "grad_norm": 0.039685264229774475, "learning_rate": 9.692624350049701e-06, "loss": 0.0037, "step": 49130 }, { "epoch": 0.8040579235866808, "grad_norm": 0.12810340523719788, "learning_rate": 9.692377837369066e-06, "loss": 0.0018, "step": 49140 }, { "epoch": 0.8042215495377567, "grad_norm": 0.023451054468750954, "learning_rate": 9.692131229014839e-06, "loss": 0.003, "step": 49150 }, { "epoch": 0.8043851754888325, "grad_norm": 0.12018232047557831, "learning_rate": 9.691884524992045e-06, "loss": 0.0027, "step": 49160 }, { "epoch": 0.8045488014399084, "grad_norm": 0.07223574072122574, "learning_rate": 9.691637725305715e-06, "loss": 0.0048, "step": 49170 }, { "epoch": 0.8047124273909843, "grad_norm": 0.0477636456489563, "learning_rate": 9.691390829960886e-06, "loss": 0.0018, "step": 49180 }, { "epoch": 0.80487605334206, "grad_norm": 0.052055809646844864, "learning_rate": 9.691143838962584e-06, "loss": 0.0044, "step": 49190 }, { "epoch": 0.8050396792931359, "grad_norm": 0.06388148665428162, "learning_rate": 9.690896752315851e-06, "loss": 0.0025, "step": 49200 }, { "epoch": 0.8052033052442117, "grad_norm": 0.10655470937490463, "learning_rate": 9.69064957002572e-06, "loss": 0.0019, "step": 49210 }, { "epoch": 0.8053669311952876, "grad_norm": 0.05479537695646286, "learning_rate": 9.690402292097237e-06, "loss": 0.0021, "step": 49220 }, { "epoch": 0.8055305571463635, "grad_norm": 0.14785219728946686, "learning_rate": 9.690154918535437e-06, "loss": 0.0025, "step": 49230 }, { "epoch": 0.8056941830974392, "grad_norm": 0.07313426584005356, "learning_rate": 9.689907449345369e-06, "loss": 0.0023, "step": 49240 }, { "epoch": 0.8058578090485151, "grad_norm": 0.20396806299686432, "learning_rate": 9.689659884532076e-06, "loss": 0.0045, "step": 49250 }, { "epoch": 0.8060214349995909, "grad_norm": 0.02227986976504326, "learning_rate": 9.689412224100607e-06, "loss": 0.0028, "step": 49260 }, { "epoch": 0.8061850609506668, "grad_norm": 0.028481468558311462, "learning_rate": 9.68916446805601e-06, "loss": 0.0012, "step": 49270 }, { "epoch": 0.8063486869017427, "grad_norm": 0.08457043766975403, "learning_rate": 9.688916616403338e-06, "loss": 0.0019, "step": 49280 }, { "epoch": 0.8065123128528184, "grad_norm": 0.1076870784163475, "learning_rate": 9.688668669147644e-06, "loss": 0.0021, "step": 49290 }, { "epoch": 0.8066759388038943, "grad_norm": 0.08043389022350311, "learning_rate": 9.688420626293984e-06, "loss": 0.0025, "step": 49300 }, { "epoch": 0.8068395647549701, "grad_norm": 0.2953455150127411, "learning_rate": 9.688172487847413e-06, "loss": 0.0065, "step": 49310 }, { "epoch": 0.807003190706046, "grad_norm": 0.03834301978349686, "learning_rate": 9.687924253812994e-06, "loss": 0.0026, "step": 49320 }, { "epoch": 0.8071668166571219, "grad_norm": 0.10327063500881195, "learning_rate": 9.687675924195786e-06, "loss": 0.0017, "step": 49330 }, { "epoch": 0.8073304426081976, "grad_norm": 0.05738653242588043, "learning_rate": 9.687427499000852e-06, "loss": 0.0021, "step": 49340 }, { "epoch": 0.8074940685592735, "grad_norm": 0.06337284296751022, "learning_rate": 9.68717897823326e-06, "loss": 0.002, "step": 49350 }, { "epoch": 0.8076576945103493, "grad_norm": 0.20215433835983276, "learning_rate": 9.686930361898073e-06, "loss": 0.0031, "step": 49360 }, { "epoch": 0.8078213204614252, "grad_norm": 0.04510684683918953, "learning_rate": 9.686681650000361e-06, "loss": 0.0022, "step": 49370 }, { "epoch": 0.8079849464125011, "grad_norm": 0.14065484702587128, "learning_rate": 9.686432842545197e-06, "loss": 0.0013, "step": 49380 }, { "epoch": 0.8081485723635768, "grad_norm": 0.009278920479118824, "learning_rate": 9.686183939537655e-06, "loss": 0.0022, "step": 49390 }, { "epoch": 0.8083121983146527, "grad_norm": 0.028190432116389275, "learning_rate": 9.685934940982806e-06, "loss": 0.0025, "step": 49400 }, { "epoch": 0.8084758242657285, "grad_norm": 0.19077745079994202, "learning_rate": 9.685685846885728e-06, "loss": 0.0031, "step": 49410 }, { "epoch": 0.8086394502168044, "grad_norm": 0.10305574536323547, "learning_rate": 9.685436657251501e-06, "loss": 0.0024, "step": 49420 }, { "epoch": 0.8088030761678803, "grad_norm": 0.14640216529369354, "learning_rate": 9.685187372085207e-06, "loss": 0.0029, "step": 49430 }, { "epoch": 0.808966702118956, "grad_norm": 0.03381755203008652, "learning_rate": 9.684937991391924e-06, "loss": 0.0011, "step": 49440 }, { "epoch": 0.8091303280700319, "grad_norm": 0.06910159438848495, "learning_rate": 9.684688515176744e-06, "loss": 0.0027, "step": 49450 }, { "epoch": 0.8092939540211077, "grad_norm": 0.11108586192131042, "learning_rate": 9.684438943444747e-06, "loss": 0.0023, "step": 49460 }, { "epoch": 0.8094575799721836, "grad_norm": 0.0801311582326889, "learning_rate": 9.684189276201023e-06, "loss": 0.0022, "step": 49470 }, { "epoch": 0.8096212059232595, "grad_norm": 0.051979780197143555, "learning_rate": 9.683939513450665e-06, "loss": 0.0016, "step": 49480 }, { "epoch": 0.8097848318743353, "grad_norm": 0.07544364035129547, "learning_rate": 9.683689655198763e-06, "loss": 0.0021, "step": 49490 }, { "epoch": 0.8099484578254111, "grad_norm": 0.09281856566667557, "learning_rate": 9.683439701450413e-06, "loss": 0.0037, "step": 49500 }, { "epoch": 0.8101120837764869, "grad_norm": 0.09419342875480652, "learning_rate": 9.68318965221071e-06, "loss": 0.0027, "step": 49510 }, { "epoch": 0.8102757097275628, "grad_norm": 0.05050797760486603, "learning_rate": 9.682939507484754e-06, "loss": 0.0023, "step": 49520 }, { "epoch": 0.8104393356786387, "grad_norm": 0.01959739439189434, "learning_rate": 9.682689267277643e-06, "loss": 0.0035, "step": 49530 }, { "epoch": 0.8106029616297145, "grad_norm": 0.2361486405134201, "learning_rate": 9.68243893159448e-06, "loss": 0.0037, "step": 49540 }, { "epoch": 0.8107665875807903, "grad_norm": 0.09247185289859772, "learning_rate": 9.682188500440371e-06, "loss": 0.0032, "step": 49550 }, { "epoch": 0.8109302135318661, "grad_norm": 0.06315923482179642, "learning_rate": 9.68193797382042e-06, "loss": 0.0024, "step": 49560 }, { "epoch": 0.811093839482942, "grad_norm": 0.3020116686820984, "learning_rate": 9.681687351739735e-06, "loss": 0.0023, "step": 49570 }, { "epoch": 0.8112574654340179, "grad_norm": 0.2189563810825348, "learning_rate": 9.681436634203426e-06, "loss": 0.0027, "step": 49580 }, { "epoch": 0.8114210913850937, "grad_norm": 0.07801032811403275, "learning_rate": 9.681185821216605e-06, "loss": 0.0037, "step": 49590 }, { "epoch": 0.8115847173361695, "grad_norm": 0.03418737277388573, "learning_rate": 9.680934912784388e-06, "loss": 0.0017, "step": 49600 }, { "epoch": 0.8117483432872453, "grad_norm": 0.07852039486169815, "learning_rate": 9.68068390891189e-06, "loss": 0.0028, "step": 49610 }, { "epoch": 0.8119119692383212, "grad_norm": 0.19629617035388947, "learning_rate": 9.680432809604225e-06, "loss": 0.0032, "step": 49620 }, { "epoch": 0.8120755951893971, "grad_norm": 0.06929346919059753, "learning_rate": 9.680181614866518e-06, "loss": 0.0025, "step": 49630 }, { "epoch": 0.8122392211404729, "grad_norm": 0.021289078518748283, "learning_rate": 9.679930324703888e-06, "loss": 0.0029, "step": 49640 }, { "epoch": 0.8124028470915488, "grad_norm": 0.10128162056207657, "learning_rate": 9.679678939121458e-06, "loss": 0.0024, "step": 49650 }, { "epoch": 0.8125664730426245, "grad_norm": 0.03344869241118431, "learning_rate": 9.679427458124352e-06, "loss": 0.0032, "step": 49660 }, { "epoch": 0.8127300989937004, "grad_norm": 0.10105393826961517, "learning_rate": 9.679175881717703e-06, "loss": 0.0023, "step": 49670 }, { "epoch": 0.8128937249447763, "grad_norm": 0.13163653016090393, "learning_rate": 9.678924209906637e-06, "loss": 0.0025, "step": 49680 }, { "epoch": 0.8130573508958521, "grad_norm": 0.08668633550405502, "learning_rate": 9.678672442696284e-06, "loss": 0.0014, "step": 49690 }, { "epoch": 0.813220976846928, "grad_norm": 0.09215854108333588, "learning_rate": 9.678420580091781e-06, "loss": 0.0021, "step": 49700 }, { "epoch": 0.8133846027980037, "grad_norm": 0.06636171787977219, "learning_rate": 9.67816862209826e-06, "loss": 0.0024, "step": 49710 }, { "epoch": 0.8135482287490796, "grad_norm": 0.07725433260202408, "learning_rate": 9.67791656872086e-06, "loss": 0.0022, "step": 49720 }, { "epoch": 0.8137118547001554, "grad_norm": 0.09806828200817108, "learning_rate": 9.67766441996472e-06, "loss": 0.0039, "step": 49730 }, { "epoch": 0.8138754806512313, "grad_norm": 0.24938426911830902, "learning_rate": 9.67741217583498e-06, "loss": 0.0022, "step": 49740 }, { "epoch": 0.8140391066023072, "grad_norm": 0.04602925106883049, "learning_rate": 9.677159836336784e-06, "loss": 0.0018, "step": 49750 }, { "epoch": 0.8142027325533829, "grad_norm": 0.11057339608669281, "learning_rate": 9.676907401475277e-06, "loss": 0.0034, "step": 49760 }, { "epoch": 0.8143663585044588, "grad_norm": 0.04751643165946007, "learning_rate": 9.676654871255604e-06, "loss": 0.0012, "step": 49770 }, { "epoch": 0.8145299844555346, "grad_norm": 0.1577099859714508, "learning_rate": 9.676402245682916e-06, "loss": 0.0033, "step": 49780 }, { "epoch": 0.8146936104066105, "grad_norm": 0.032889850437641144, "learning_rate": 9.676149524762365e-06, "loss": 0.0035, "step": 49790 }, { "epoch": 0.8148572363576864, "grad_norm": 0.05311069265007973, "learning_rate": 9.675896708499102e-06, "loss": 0.002, "step": 49800 }, { "epoch": 0.8150208623087621, "grad_norm": 0.00864950567483902, "learning_rate": 9.67564379689828e-06, "loss": 0.0023, "step": 49810 }, { "epoch": 0.815184488259838, "grad_norm": 0.05207619071006775, "learning_rate": 9.675390789965058e-06, "loss": 0.0047, "step": 49820 }, { "epoch": 0.8153481142109138, "grad_norm": 0.05980609729886055, "learning_rate": 9.675137687704594e-06, "loss": 0.0029, "step": 49830 }, { "epoch": 0.8155117401619897, "grad_norm": 0.05348074808716774, "learning_rate": 9.67488449012205e-06, "loss": 0.0021, "step": 49840 }, { "epoch": 0.8156753661130656, "grad_norm": 0.05830732733011246, "learning_rate": 9.674631197222586e-06, "loss": 0.0024, "step": 49850 }, { "epoch": 0.8158389920641413, "grad_norm": 0.14693240821361542, "learning_rate": 9.674377809011368e-06, "loss": 0.0057, "step": 49860 }, { "epoch": 0.8160026180152172, "grad_norm": 0.07146044820547104, "learning_rate": 9.67412432549356e-06, "loss": 0.0024, "step": 49870 }, { "epoch": 0.816166243966293, "grad_norm": 0.07387471944093704, "learning_rate": 9.673870746674336e-06, "loss": 0.0029, "step": 49880 }, { "epoch": 0.8163298699173689, "grad_norm": 0.12878385186195374, "learning_rate": 9.67361707255886e-06, "loss": 0.0054, "step": 49890 }, { "epoch": 0.8164934958684448, "grad_norm": 0.06833955645561218, "learning_rate": 9.673363303152306e-06, "loss": 0.0038, "step": 49900 }, { "epoch": 0.8166571218195205, "grad_norm": 0.04959404468536377, "learning_rate": 9.673109438459851e-06, "loss": 0.0039, "step": 49910 }, { "epoch": 0.8168207477705964, "grad_norm": 0.028919225558638573, "learning_rate": 9.672855478486668e-06, "loss": 0.0024, "step": 49920 }, { "epoch": 0.8169843737216722, "grad_norm": 0.0390721894800663, "learning_rate": 9.672601423237934e-06, "loss": 0.0023, "step": 49930 }, { "epoch": 0.8171479996727481, "grad_norm": 0.08679869771003723, "learning_rate": 9.672347272718835e-06, "loss": 0.0025, "step": 49940 }, { "epoch": 0.817311625623824, "grad_norm": 0.07350997626781464, "learning_rate": 9.672093026934544e-06, "loss": 0.0035, "step": 49950 }, { "epoch": 0.8174752515748998, "grad_norm": 0.06121387705206871, "learning_rate": 9.671838685890252e-06, "loss": 0.0027, "step": 49960 }, { "epoch": 0.8176388775259756, "grad_norm": 0.07041393965482712, "learning_rate": 9.671584249591143e-06, "loss": 0.0018, "step": 49970 }, { "epoch": 0.8178025034770514, "grad_norm": 0.12060726433992386, "learning_rate": 9.671329718042404e-06, "loss": 0.002, "step": 49980 }, { "epoch": 0.8179661294281273, "grad_norm": 0.021124226972460747, "learning_rate": 9.671075091249224e-06, "loss": 0.0033, "step": 49990 }, { "epoch": 0.8181297553792032, "grad_norm": 0.05862497538328171, "learning_rate": 9.670820369216795e-06, "loss": 0.0022, "step": 50000 }, { "epoch": 0.818293381330279, "grad_norm": 0.131966695189476, "learning_rate": 9.670565551950312e-06, "loss": 0.0037, "step": 50010 }, { "epoch": 0.8184570072813548, "grad_norm": 0.14693523943424225, "learning_rate": 9.670310639454969e-06, "loss": 0.0031, "step": 50020 }, { "epoch": 0.8186206332324306, "grad_norm": 0.149275004863739, "learning_rate": 9.670055631735965e-06, "loss": 0.0033, "step": 50030 }, { "epoch": 0.8187842591835065, "grad_norm": 0.15670731663703918, "learning_rate": 9.669800528798498e-06, "loss": 0.0035, "step": 50040 }, { "epoch": 0.8189478851345824, "grad_norm": 0.1389196217060089, "learning_rate": 9.66954533064777e-06, "loss": 0.0026, "step": 50050 }, { "epoch": 0.8191115110856582, "grad_norm": 0.012280361726880074, "learning_rate": 9.669290037288984e-06, "loss": 0.0027, "step": 50060 }, { "epoch": 0.819275137036734, "grad_norm": 0.03682423382997513, "learning_rate": 9.669034648727345e-06, "loss": 0.0026, "step": 50070 }, { "epoch": 0.8194387629878098, "grad_norm": 0.10147271305322647, "learning_rate": 9.668779164968061e-06, "loss": 0.002, "step": 50080 }, { "epoch": 0.8196023889388857, "grad_norm": 0.34381839632987976, "learning_rate": 9.668523586016338e-06, "loss": 0.0029, "step": 50090 }, { "epoch": 0.8197660148899616, "grad_norm": 0.13903748989105225, "learning_rate": 9.668267911877392e-06, "loss": 0.0036, "step": 50100 }, { "epoch": 0.8199296408410374, "grad_norm": 0.1477908045053482, "learning_rate": 9.668012142556434e-06, "loss": 0.0032, "step": 50110 }, { "epoch": 0.8200932667921133, "grad_norm": 0.050265174359083176, "learning_rate": 9.667756278058676e-06, "loss": 0.0022, "step": 50120 }, { "epoch": 0.820256892743189, "grad_norm": 0.1716839224100113, "learning_rate": 9.66750031838934e-06, "loss": 0.0042, "step": 50130 }, { "epoch": 0.8204205186942649, "grad_norm": 0.11369640380144119, "learning_rate": 9.667244263553639e-06, "loss": 0.003, "step": 50140 }, { "epoch": 0.8205841446453408, "grad_norm": 0.09122850000858307, "learning_rate": 9.666988113556799e-06, "loss": 0.0028, "step": 50150 }, { "epoch": 0.8207477705964166, "grad_norm": 0.14187799394130707, "learning_rate": 9.66673186840404e-06, "loss": 0.0025, "step": 50160 }, { "epoch": 0.8209113965474925, "grad_norm": 0.17925108969211578, "learning_rate": 9.666475528100587e-06, "loss": 0.0027, "step": 50170 }, { "epoch": 0.8210750224985682, "grad_norm": 0.10018332302570343, "learning_rate": 9.666219092651667e-06, "loss": 0.0034, "step": 50180 }, { "epoch": 0.8212386484496441, "grad_norm": 0.1883748173713684, "learning_rate": 9.665962562062508e-06, "loss": 0.0019, "step": 50190 }, { "epoch": 0.82140227440072, "grad_norm": 0.06900092959403992, "learning_rate": 9.66570593633834e-06, "loss": 0.0039, "step": 50200 }, { "epoch": 0.8215659003517958, "grad_norm": 0.21583062410354614, "learning_rate": 9.665449215484397e-06, "loss": 0.0022, "step": 50210 }, { "epoch": 0.8217295263028717, "grad_norm": 0.09620422124862671, "learning_rate": 9.665192399505912e-06, "loss": 0.0024, "step": 50220 }, { "epoch": 0.8218931522539474, "grad_norm": 0.08340482413768768, "learning_rate": 9.66493548840812e-06, "loss": 0.0025, "step": 50230 }, { "epoch": 0.8220567782050233, "grad_norm": 0.05388401076197624, "learning_rate": 9.664678482196264e-06, "loss": 0.0026, "step": 50240 }, { "epoch": 0.8222204041560992, "grad_norm": 0.07345008850097656, "learning_rate": 9.664421380875577e-06, "loss": 0.0034, "step": 50250 }, { "epoch": 0.822384030107175, "grad_norm": 0.11372125893831253, "learning_rate": 9.664164184451309e-06, "loss": 0.0021, "step": 50260 }, { "epoch": 0.8225476560582509, "grad_norm": 0.10623518377542496, "learning_rate": 9.663906892928698e-06, "loss": 0.0022, "step": 50270 }, { "epoch": 0.8227112820093266, "grad_norm": 0.3014882802963257, "learning_rate": 9.663649506312991e-06, "loss": 0.0024, "step": 50280 }, { "epoch": 0.8228749079604025, "grad_norm": 0.019799960777163506, "learning_rate": 9.663392024609439e-06, "loss": 0.0018, "step": 50290 }, { "epoch": 0.8230385339114784, "grad_norm": 0.09246491640806198, "learning_rate": 9.663134447823287e-06, "loss": 0.0029, "step": 50300 }, { "epoch": 0.8232021598625542, "grad_norm": 0.21695543825626373, "learning_rate": 9.662876775959791e-06, "loss": 0.002, "step": 50310 }, { "epoch": 0.8233657858136301, "grad_norm": 0.05919507518410683, "learning_rate": 9.662619009024203e-06, "loss": 0.004, "step": 50320 }, { "epoch": 0.8235294117647058, "grad_norm": 0.11945618689060211, "learning_rate": 9.66236114702178e-06, "loss": 0.003, "step": 50330 }, { "epoch": 0.8236930377157817, "grad_norm": 0.22266869246959686, "learning_rate": 9.662103189957777e-06, "loss": 0.0023, "step": 50340 }, { "epoch": 0.8238566636668576, "grad_norm": 0.06330099701881409, "learning_rate": 9.661845137837456e-06, "loss": 0.0057, "step": 50350 }, { "epoch": 0.8240202896179334, "grad_norm": 0.14564426243305206, "learning_rate": 9.661586990666077e-06, "loss": 0.0024, "step": 50360 }, { "epoch": 0.8241839155690093, "grad_norm": 0.0715978816151619, "learning_rate": 9.661328748448904e-06, "loss": 0.0017, "step": 50370 }, { "epoch": 0.824347541520085, "grad_norm": 0.050409946590662, "learning_rate": 9.6610704111912e-06, "loss": 0.0013, "step": 50380 }, { "epoch": 0.8245111674711609, "grad_norm": 0.05388601869344711, "learning_rate": 9.660811978898236e-06, "loss": 0.0022, "step": 50390 }, { "epoch": 0.8246747934222368, "grad_norm": 0.08749005943536758, "learning_rate": 9.66055345157528e-06, "loss": 0.0025, "step": 50400 }, { "epoch": 0.8248384193733126, "grad_norm": 0.19783787429332733, "learning_rate": 9.660294829227603e-06, "loss": 0.0032, "step": 50410 }, { "epoch": 0.8250020453243885, "grad_norm": 0.08277029544115067, "learning_rate": 9.660036111860478e-06, "loss": 0.0028, "step": 50420 }, { "epoch": 0.8251656712754643, "grad_norm": 0.07121231406927109, "learning_rate": 9.659777299479179e-06, "loss": 0.0022, "step": 50430 }, { "epoch": 0.8253292972265401, "grad_norm": 0.16005487740039825, "learning_rate": 9.659518392088985e-06, "loss": 0.0024, "step": 50440 }, { "epoch": 0.825492923177616, "grad_norm": 0.0677947849035263, "learning_rate": 9.659259389695172e-06, "loss": 0.0015, "step": 50450 }, { "epoch": 0.8256565491286918, "grad_norm": 0.04735248535871506, "learning_rate": 9.659000292303023e-06, "loss": 0.0019, "step": 50460 }, { "epoch": 0.8258201750797677, "grad_norm": 0.06862716376781464, "learning_rate": 9.65874109991782e-06, "loss": 0.0041, "step": 50470 }, { "epoch": 0.8259838010308435, "grad_norm": 0.11790227890014648, "learning_rate": 9.658481812544851e-06, "loss": 0.0025, "step": 50480 }, { "epoch": 0.8261474269819193, "grad_norm": 0.1961442083120346, "learning_rate": 9.658222430189396e-06, "loss": 0.0032, "step": 50490 }, { "epoch": 0.8263110529329952, "grad_norm": 0.07243924587965012, "learning_rate": 9.65796295285675e-06, "loss": 0.0034, "step": 50500 }, { "epoch": 0.826474678884071, "grad_norm": 0.14968670904636383, "learning_rate": 9.6577033805522e-06, "loss": 0.0035, "step": 50510 }, { "epoch": 0.8266383048351469, "grad_norm": 0.24105805158615112, "learning_rate": 9.65744371328104e-06, "loss": 0.0019, "step": 50520 }, { "epoch": 0.8268019307862227, "grad_norm": 0.0974859967827797, "learning_rate": 9.657183951048562e-06, "loss": 0.0026, "step": 50530 }, { "epoch": 0.8269655567372985, "grad_norm": 0.07340002059936523, "learning_rate": 9.656924093860064e-06, "loss": 0.0027, "step": 50540 }, { "epoch": 0.8271291826883744, "grad_norm": 0.03923872485756874, "learning_rate": 9.656664141720847e-06, "loss": 0.0021, "step": 50550 }, { "epoch": 0.8272928086394502, "grad_norm": 0.07354146242141724, "learning_rate": 9.656404094636207e-06, "loss": 0.0027, "step": 50560 }, { "epoch": 0.8274564345905261, "grad_norm": 0.04071897268295288, "learning_rate": 9.656143952611447e-06, "loss": 0.0016, "step": 50570 }, { "epoch": 0.8276200605416019, "grad_norm": 0.0978303775191307, "learning_rate": 9.655883715651873e-06, "loss": 0.0016, "step": 50580 }, { "epoch": 0.8277836864926778, "grad_norm": 0.09125334769487381, "learning_rate": 9.655623383762787e-06, "loss": 0.002, "step": 50590 }, { "epoch": 0.8279473124437535, "grad_norm": 0.059409063309431076, "learning_rate": 9.655362956949503e-06, "loss": 0.0016, "step": 50600 }, { "epoch": 0.8281109383948294, "grad_norm": 0.05440349876880646, "learning_rate": 9.655102435217325e-06, "loss": 0.0019, "step": 50610 }, { "epoch": 0.8282745643459053, "grad_norm": 0.11952885240316391, "learning_rate": 9.654841818571568e-06, "loss": 0.0028, "step": 50620 }, { "epoch": 0.8284381902969811, "grad_norm": 0.027014872059226036, "learning_rate": 9.654581107017546e-06, "loss": 0.0024, "step": 50630 }, { "epoch": 0.828601816248057, "grad_norm": 0.06745032221078873, "learning_rate": 9.654320300560573e-06, "loss": 0.0029, "step": 50640 }, { "epoch": 0.8287654421991327, "grad_norm": 0.03176699951291084, "learning_rate": 9.654059399205968e-06, "loss": 0.0028, "step": 50650 }, { "epoch": 0.8289290681502086, "grad_norm": 0.11653466522693634, "learning_rate": 9.65379840295905e-06, "loss": 0.0034, "step": 50660 }, { "epoch": 0.8290926941012845, "grad_norm": 0.037585724145174026, "learning_rate": 9.65353731182514e-06, "loss": 0.0025, "step": 50670 }, { "epoch": 0.8292563200523603, "grad_norm": 0.07670770585536957, "learning_rate": 9.653276125809564e-06, "loss": 0.0024, "step": 50680 }, { "epoch": 0.8294199460034362, "grad_norm": 0.18080614507198334, "learning_rate": 9.653014844917642e-06, "loss": 0.0021, "step": 50690 }, { "epoch": 0.8295835719545119, "grad_norm": 0.12606766819953918, "learning_rate": 9.652753469154707e-06, "loss": 0.0033, "step": 50700 }, { "epoch": 0.8297471979055878, "grad_norm": 0.05596032738685608, "learning_rate": 9.652491998526084e-06, "loss": 0.0021, "step": 50710 }, { "epoch": 0.8299108238566637, "grad_norm": 0.07460125535726547, "learning_rate": 9.652230433037106e-06, "loss": 0.0033, "step": 50720 }, { "epoch": 0.8300744498077395, "grad_norm": 0.01866178773343563, "learning_rate": 9.651968772693108e-06, "loss": 0.0014, "step": 50730 }, { "epoch": 0.8302380757588154, "grad_norm": 0.0429743230342865, "learning_rate": 9.651707017499421e-06, "loss": 0.0023, "step": 50740 }, { "epoch": 0.8304017017098911, "grad_norm": 0.23695142567157745, "learning_rate": 9.651445167461385e-06, "loss": 0.0026, "step": 50750 }, { "epoch": 0.830565327660967, "grad_norm": 0.035303689539432526, "learning_rate": 9.651183222584338e-06, "loss": 0.0027, "step": 50760 }, { "epoch": 0.8307289536120429, "grad_norm": 0.4837470054626465, "learning_rate": 9.65092118287362e-06, "loss": 0.0033, "step": 50770 }, { "epoch": 0.8308925795631187, "grad_norm": 0.1959744244813919, "learning_rate": 9.650659048334577e-06, "loss": 0.0022, "step": 50780 }, { "epoch": 0.8310562055141946, "grad_norm": 0.07652280479669571, "learning_rate": 9.650396818972549e-06, "loss": 0.0033, "step": 50790 }, { "epoch": 0.8312198314652703, "grad_norm": 0.13121481239795685, "learning_rate": 9.650134494792884e-06, "loss": 0.0035, "step": 50800 }, { "epoch": 0.8313834574163462, "grad_norm": 0.014614158309996128, "learning_rate": 9.649872075800933e-06, "loss": 0.0052, "step": 50810 }, { "epoch": 0.8315470833674221, "grad_norm": 0.03001992218196392, "learning_rate": 9.649609562002045e-06, "loss": 0.0037, "step": 50820 }, { "epoch": 0.8317107093184979, "grad_norm": 0.07578999549150467, "learning_rate": 9.649346953401573e-06, "loss": 0.0014, "step": 50830 }, { "epoch": 0.8318743352695738, "grad_norm": 0.1088339313864708, "learning_rate": 9.64908425000487e-06, "loss": 0.0036, "step": 50840 }, { "epoch": 0.8320379612206495, "grad_norm": 0.05486440286040306, "learning_rate": 9.648821451817292e-06, "loss": 0.0025, "step": 50850 }, { "epoch": 0.8322015871717254, "grad_norm": 0.16763150691986084, "learning_rate": 9.648558558844198e-06, "loss": 0.0022, "step": 50860 }, { "epoch": 0.8323652131228013, "grad_norm": 0.06533413380384445, "learning_rate": 9.648295571090952e-06, "loss": 0.0041, "step": 50870 }, { "epoch": 0.8325288390738771, "grad_norm": 0.1104370579123497, "learning_rate": 9.64803248856291e-06, "loss": 0.0027, "step": 50880 }, { "epoch": 0.832692465024953, "grad_norm": 0.2400471270084381, "learning_rate": 9.647769311265438e-06, "loss": 0.0025, "step": 50890 }, { "epoch": 0.8328560909760288, "grad_norm": 0.06163942813873291, "learning_rate": 9.647506039203902e-06, "loss": 0.0031, "step": 50900 }, { "epoch": 0.8330197169271046, "grad_norm": 0.05411505326628685, "learning_rate": 9.647242672383672e-06, "loss": 0.0033, "step": 50910 }, { "epoch": 0.8331833428781805, "grad_norm": 0.13377022743225098, "learning_rate": 9.646979210810117e-06, "loss": 0.0018, "step": 50920 }, { "epoch": 0.8333469688292563, "grad_norm": 0.01610630936920643, "learning_rate": 9.646715654488607e-06, "loss": 0.0031, "step": 50930 }, { "epoch": 0.8335105947803322, "grad_norm": 0.02581963688135147, "learning_rate": 9.646452003424515e-06, "loss": 0.003, "step": 50940 }, { "epoch": 0.833674220731408, "grad_norm": 0.2019493728876114, "learning_rate": 9.64618825762322e-06, "loss": 0.003, "step": 50950 }, { "epoch": 0.8338378466824838, "grad_norm": 0.07141926884651184, "learning_rate": 9.645924417090098e-06, "loss": 0.0033, "step": 50960 }, { "epoch": 0.8340014726335597, "grad_norm": 0.06272360682487488, "learning_rate": 9.645660481830528e-06, "loss": 0.0031, "step": 50970 }, { "epoch": 0.8341650985846355, "grad_norm": 0.1110135018825531, "learning_rate": 9.645396451849893e-06, "loss": 0.0022, "step": 50980 }, { "epoch": 0.8343287245357114, "grad_norm": 0.16069768369197845, "learning_rate": 9.645132327153573e-06, "loss": 0.0043, "step": 50990 }, { "epoch": 0.8344923504867872, "grad_norm": 0.06618189811706543, "learning_rate": 9.644868107746957e-06, "loss": 0.0024, "step": 51000 }, { "epoch": 0.834655976437863, "grad_norm": 0.10082022845745087, "learning_rate": 9.64460379363543e-06, "loss": 0.0032, "step": 51010 }, { "epoch": 0.8348196023889389, "grad_norm": 0.20413430035114288, "learning_rate": 9.644339384824381e-06, "loss": 0.0047, "step": 51020 }, { "epoch": 0.8349832283400147, "grad_norm": 0.0794934630393982, "learning_rate": 9.644074881319203e-06, "loss": 0.0035, "step": 51030 }, { "epoch": 0.8351468542910906, "grad_norm": 0.23977714776992798, "learning_rate": 9.643810283125287e-06, "loss": 0.0021, "step": 51040 }, { "epoch": 0.8353104802421664, "grad_norm": 0.01649298518896103, "learning_rate": 9.64354559024803e-06, "loss": 0.0021, "step": 51050 }, { "epoch": 0.8354741061932422, "grad_norm": 0.1320723444223404, "learning_rate": 9.643280802692827e-06, "loss": 0.0025, "step": 51060 }, { "epoch": 0.8356377321443181, "grad_norm": 0.08078517764806747, "learning_rate": 9.643015920465078e-06, "loss": 0.0033, "step": 51070 }, { "epoch": 0.8358013580953939, "grad_norm": 0.10810387134552002, "learning_rate": 9.64275094357018e-06, "loss": 0.0021, "step": 51080 }, { "epoch": 0.8359649840464698, "grad_norm": 0.016725048422813416, "learning_rate": 9.642485872013543e-06, "loss": 0.0027, "step": 51090 }, { "epoch": 0.8361286099975456, "grad_norm": 0.03319329395890236, "learning_rate": 9.642220705800565e-06, "loss": 0.0019, "step": 51100 }, { "epoch": 0.8362922359486215, "grad_norm": 0.1352405697107315, "learning_rate": 9.641955444936655e-06, "loss": 0.0039, "step": 51110 }, { "epoch": 0.8364558618996973, "grad_norm": 0.32675474882125854, "learning_rate": 9.641690089427222e-06, "loss": 0.0021, "step": 51120 }, { "epoch": 0.8366194878507731, "grad_norm": 0.12190156430006027, "learning_rate": 9.641424639277674e-06, "loss": 0.0035, "step": 51130 }, { "epoch": 0.836783113801849, "grad_norm": 0.13167284429073334, "learning_rate": 9.641159094493428e-06, "loss": 0.0037, "step": 51140 }, { "epoch": 0.8369467397529248, "grad_norm": 0.05040110647678375, "learning_rate": 9.640893455079893e-06, "loss": 0.0032, "step": 51150 }, { "epoch": 0.8371103657040007, "grad_norm": 0.02888505347073078, "learning_rate": 9.640627721042488e-06, "loss": 0.0015, "step": 51160 }, { "epoch": 0.8372739916550765, "grad_norm": 0.08166094124317169, "learning_rate": 9.640361892386629e-06, "loss": 0.0026, "step": 51170 }, { "epoch": 0.8374376176061523, "grad_norm": 0.053631123155355453, "learning_rate": 9.64009596911774e-06, "loss": 0.0035, "step": 51180 }, { "epoch": 0.8376012435572282, "grad_norm": 0.035475872457027435, "learning_rate": 9.63982995124124e-06, "loss": 0.0035, "step": 51190 }, { "epoch": 0.837764869508304, "grad_norm": 0.17113080620765686, "learning_rate": 9.639563838762552e-06, "loss": 0.0028, "step": 51200 }, { "epoch": 0.8379284954593799, "grad_norm": 0.19872623682022095, "learning_rate": 9.639297631687104e-06, "loss": 0.0028, "step": 51210 }, { "epoch": 0.8380921214104557, "grad_norm": 0.11533582210540771, "learning_rate": 9.639031330020323e-06, "loss": 0.0021, "step": 51220 }, { "epoch": 0.8382557473615315, "grad_norm": 0.05582445114850998, "learning_rate": 9.638764933767637e-06, "loss": 0.0027, "step": 51230 }, { "epoch": 0.8384193733126074, "grad_norm": 0.105367511510849, "learning_rate": 9.63849844293448e-06, "loss": 0.0039, "step": 51240 }, { "epoch": 0.8385829992636832, "grad_norm": 0.05712570995092392, "learning_rate": 9.638231857526286e-06, "loss": 0.0031, "step": 51250 }, { "epoch": 0.8387466252147591, "grad_norm": 0.11286108195781708, "learning_rate": 9.637965177548488e-06, "loss": 0.0028, "step": 51260 }, { "epoch": 0.838910251165835, "grad_norm": 0.0439751036465168, "learning_rate": 9.637698403006525e-06, "loss": 0.0037, "step": 51270 }, { "epoch": 0.8390738771169107, "grad_norm": 0.15601837635040283, "learning_rate": 9.637431533905834e-06, "loss": 0.0024, "step": 51280 }, { "epoch": 0.8392375030679866, "grad_norm": 0.06329236179590225, "learning_rate": 9.63716457025186e-06, "loss": 0.0029, "step": 51290 }, { "epoch": 0.8394011290190624, "grad_norm": 0.034813422709703445, "learning_rate": 9.636897512050044e-06, "loss": 0.0024, "step": 51300 }, { "epoch": 0.8395647549701383, "grad_norm": 0.21170611679553986, "learning_rate": 9.63663035930583e-06, "loss": 0.0029, "step": 51310 }, { "epoch": 0.8397283809212142, "grad_norm": 0.09532210230827332, "learning_rate": 9.636363112024668e-06, "loss": 0.0025, "step": 51320 }, { "epoch": 0.8398920068722899, "grad_norm": 0.09057831764221191, "learning_rate": 9.636095770212004e-06, "loss": 0.0018, "step": 51330 }, { "epoch": 0.8400556328233658, "grad_norm": 0.1164049357175827, "learning_rate": 9.635828333873291e-06, "loss": 0.0029, "step": 51340 }, { "epoch": 0.8402192587744416, "grad_norm": 0.059679463505744934, "learning_rate": 9.63556080301398e-06, "loss": 0.0037, "step": 51350 }, { "epoch": 0.8403828847255175, "grad_norm": 0.15949715673923492, "learning_rate": 9.635293177639526e-06, "loss": 0.0025, "step": 51360 }, { "epoch": 0.8405465106765934, "grad_norm": 0.09036628901958466, "learning_rate": 9.635025457755389e-06, "loss": 0.0021, "step": 51370 }, { "epoch": 0.8407101366276691, "grad_norm": 0.10026098787784576, "learning_rate": 9.634757643367023e-06, "loss": 0.003, "step": 51380 }, { "epoch": 0.840873762578745, "grad_norm": 0.017771149054169655, "learning_rate": 9.63448973447989e-06, "loss": 0.0016, "step": 51390 }, { "epoch": 0.8410373885298208, "grad_norm": 0.01599142514169216, "learning_rate": 9.634221731099454e-06, "loss": 0.0021, "step": 51400 }, { "epoch": 0.8412010144808967, "grad_norm": 0.029874783009290695, "learning_rate": 9.633953633231177e-06, "loss": 0.0018, "step": 51410 }, { "epoch": 0.8413646404319726, "grad_norm": 0.07459372282028198, "learning_rate": 9.633685440880527e-06, "loss": 0.0027, "step": 51420 }, { "epoch": 0.8415282663830483, "grad_norm": 0.24061495065689087, "learning_rate": 9.63341715405297e-06, "loss": 0.0032, "step": 51430 }, { "epoch": 0.8416918923341242, "grad_norm": 0.05475475639104843, "learning_rate": 9.633148772753981e-06, "loss": 0.0024, "step": 51440 }, { "epoch": 0.8418555182852, "grad_norm": 0.07795213162899017, "learning_rate": 9.632880296989027e-06, "loss": 0.0022, "step": 51450 }, { "epoch": 0.8420191442362759, "grad_norm": 0.10870074480772018, "learning_rate": 9.632611726763584e-06, "loss": 0.0025, "step": 51460 }, { "epoch": 0.8421827701873517, "grad_norm": 0.06703323125839233, "learning_rate": 9.632343062083128e-06, "loss": 0.0038, "step": 51470 }, { "epoch": 0.8423463961384275, "grad_norm": 0.21682631969451904, "learning_rate": 9.632074302953135e-06, "loss": 0.0025, "step": 51480 }, { "epoch": 0.8425100220895034, "grad_norm": 0.16769030690193176, "learning_rate": 9.63180544937909e-06, "loss": 0.002, "step": 51490 }, { "epoch": 0.8426736480405792, "grad_norm": 0.09619239717721939, "learning_rate": 9.63153650136647e-06, "loss": 0.0027, "step": 51500 }, { "epoch": 0.8428372739916551, "grad_norm": 0.06273549795150757, "learning_rate": 9.631267458920757e-06, "loss": 0.0031, "step": 51510 }, { "epoch": 0.8430008999427309, "grad_norm": 0.19191090762615204, "learning_rate": 9.630998322047442e-06, "loss": 0.0021, "step": 51520 }, { "epoch": 0.8431645258938067, "grad_norm": 0.06194857880473137, "learning_rate": 9.63072909075201e-06, "loss": 0.0019, "step": 51530 }, { "epoch": 0.8433281518448826, "grad_norm": 0.060409702360630035, "learning_rate": 9.630459765039948e-06, "loss": 0.0044, "step": 51540 }, { "epoch": 0.8434917777959584, "grad_norm": 0.06496908515691757, "learning_rate": 9.63019034491675e-06, "loss": 0.0022, "step": 51550 }, { "epoch": 0.8436554037470343, "grad_norm": 0.1393449455499649, "learning_rate": 9.629920830387908e-06, "loss": 0.0028, "step": 51560 }, { "epoch": 0.8438190296981101, "grad_norm": 0.0374143160879612, "learning_rate": 9.62965122145892e-06, "loss": 0.0023, "step": 51570 }, { "epoch": 0.843982655649186, "grad_norm": 0.08272173255681992, "learning_rate": 9.62938151813528e-06, "loss": 0.0023, "step": 51580 }, { "epoch": 0.8441462816002618, "grad_norm": 0.10602203756570816, "learning_rate": 9.629111720422488e-06, "loss": 0.0024, "step": 51590 }, { "epoch": 0.8443099075513376, "grad_norm": 0.12805239856243134, "learning_rate": 9.628841828326046e-06, "loss": 0.0037, "step": 51600 }, { "epoch": 0.8444735335024135, "grad_norm": 0.02662556804716587, "learning_rate": 9.628571841851452e-06, "loss": 0.0016, "step": 51610 }, { "epoch": 0.8446371594534893, "grad_norm": 0.08782085031270981, "learning_rate": 9.628301761004219e-06, "loss": 0.0032, "step": 51620 }, { "epoch": 0.8448007854045652, "grad_norm": 0.10015296936035156, "learning_rate": 9.628031585789848e-06, "loss": 0.002, "step": 51630 }, { "epoch": 0.844964411355641, "grad_norm": 0.08973491191864014, "learning_rate": 9.627761316213848e-06, "loss": 0.0027, "step": 51640 }, { "epoch": 0.8451280373067168, "grad_norm": 0.03794463723897934, "learning_rate": 9.62749095228173e-06, "loss": 0.004, "step": 51650 }, { "epoch": 0.8452916632577927, "grad_norm": 0.049685701727867126, "learning_rate": 9.627220493999008e-06, "loss": 0.0031, "step": 51660 }, { "epoch": 0.8454552892088685, "grad_norm": 0.09121789783239365, "learning_rate": 9.626949941371196e-06, "loss": 0.0058, "step": 51670 }, { "epoch": 0.8456189151599444, "grad_norm": 0.088957279920578, "learning_rate": 9.626679294403809e-06, "loss": 0.003, "step": 51680 }, { "epoch": 0.8457825411110202, "grad_norm": 0.02059510350227356, "learning_rate": 9.626408553102367e-06, "loss": 0.0035, "step": 51690 }, { "epoch": 0.845946167062096, "grad_norm": 0.025986632332205772, "learning_rate": 9.626137717472387e-06, "loss": 0.0022, "step": 51700 }, { "epoch": 0.8461097930131719, "grad_norm": 0.03521787375211716, "learning_rate": 9.625866787519397e-06, "loss": 0.0025, "step": 51710 }, { "epoch": 0.8462734189642477, "grad_norm": 0.08303161710500717, "learning_rate": 9.625595763248915e-06, "loss": 0.002, "step": 51720 }, { "epoch": 0.8464370449153236, "grad_norm": 0.06768321245908737, "learning_rate": 9.62532464466647e-06, "loss": 0.0021, "step": 51730 }, { "epoch": 0.8466006708663995, "grad_norm": 0.14760102331638336, "learning_rate": 9.62505343177759e-06, "loss": 0.0038, "step": 51740 }, { "epoch": 0.8467642968174752, "grad_norm": 0.023937223479151726, "learning_rate": 9.624782124587802e-06, "loss": 0.002, "step": 51750 }, { "epoch": 0.8469279227685511, "grad_norm": 0.12368994951248169, "learning_rate": 9.62451072310264e-06, "loss": 0.0036, "step": 51760 }, { "epoch": 0.8470915487196269, "grad_norm": 0.19133426249027252, "learning_rate": 9.624239227327638e-06, "loss": 0.0028, "step": 51770 }, { "epoch": 0.8472551746707028, "grad_norm": 0.02647722326219082, "learning_rate": 9.62396763726833e-06, "loss": 0.0039, "step": 51780 }, { "epoch": 0.8474188006217787, "grad_norm": 0.04900602623820305, "learning_rate": 9.623695952930255e-06, "loss": 0.0017, "step": 51790 }, { "epoch": 0.8475824265728544, "grad_norm": 0.19394294917583466, "learning_rate": 9.623424174318953e-06, "loss": 0.0034, "step": 51800 }, { "epoch": 0.8477460525239303, "grad_norm": 0.048846401274204254, "learning_rate": 9.623152301439963e-06, "loss": 0.0034, "step": 51810 }, { "epoch": 0.8479096784750061, "grad_norm": 0.08402048796415329, "learning_rate": 9.62288033429883e-06, "loss": 0.0031, "step": 51820 }, { "epoch": 0.848073304426082, "grad_norm": 0.24493998289108276, "learning_rate": 9.622608272901097e-06, "loss": 0.0052, "step": 51830 }, { "epoch": 0.8482369303771579, "grad_norm": 0.05313286557793617, "learning_rate": 9.622336117252314e-06, "loss": 0.0019, "step": 51840 }, { "epoch": 0.8484005563282336, "grad_norm": 0.07478287816047668, "learning_rate": 9.622063867358029e-06, "loss": 0.0032, "step": 51850 }, { "epoch": 0.8485641822793095, "grad_norm": 0.07796069234609604, "learning_rate": 9.621791523223792e-06, "loss": 0.0024, "step": 51860 }, { "epoch": 0.8487278082303853, "grad_norm": 0.13828009366989136, "learning_rate": 9.621519084855157e-06, "loss": 0.0029, "step": 51870 }, { "epoch": 0.8488914341814612, "grad_norm": 0.26562175154685974, "learning_rate": 9.62124655225768e-06, "loss": 0.003, "step": 51880 }, { "epoch": 0.8490550601325371, "grad_norm": 0.04413464292883873, "learning_rate": 9.620973925436913e-06, "loss": 0.0021, "step": 51890 }, { "epoch": 0.8492186860836128, "grad_norm": 0.06096407026052475, "learning_rate": 9.620701204398419e-06, "loss": 0.0034, "step": 51900 }, { "epoch": 0.8493823120346887, "grad_norm": 0.006159317679703236, "learning_rate": 9.620428389147756e-06, "loss": 0.0014, "step": 51910 }, { "epoch": 0.8495459379857645, "grad_norm": 0.21517223119735718, "learning_rate": 9.62015547969049e-06, "loss": 0.0037, "step": 51920 }, { "epoch": 0.8497095639368404, "grad_norm": 0.033268868923187256, "learning_rate": 9.619882476032181e-06, "loss": 0.0026, "step": 51930 }, { "epoch": 0.8498731898879163, "grad_norm": 0.005219984333962202, "learning_rate": 9.619609378178398e-06, "loss": 0.0025, "step": 51940 }, { "epoch": 0.850036815838992, "grad_norm": 0.1811348795890808, "learning_rate": 9.619336186134709e-06, "loss": 0.0023, "step": 51950 }, { "epoch": 0.8502004417900679, "grad_norm": 0.09947478026151657, "learning_rate": 9.619062899906684e-06, "loss": 0.0084, "step": 51960 }, { "epoch": 0.8503640677411437, "grad_norm": 0.05105424299836159, "learning_rate": 9.618789519499894e-06, "loss": 0.002, "step": 51970 }, { "epoch": 0.8505276936922196, "grad_norm": 0.10065452754497528, "learning_rate": 9.618516044919914e-06, "loss": 0.0031, "step": 51980 }, { "epoch": 0.8506913196432955, "grad_norm": 0.1850442886352539, "learning_rate": 9.618242476172321e-06, "loss": 0.0029, "step": 51990 }, { "epoch": 0.8508549455943712, "grad_norm": 0.07297413796186447, "learning_rate": 9.61796881326269e-06, "loss": 0.0019, "step": 52000 }, { "epoch": 0.8510185715454471, "grad_norm": 0.12526778876781464, "learning_rate": 9.617695056196603e-06, "loss": 0.0019, "step": 52010 }, { "epoch": 0.8511821974965229, "grad_norm": 0.09726908057928085, "learning_rate": 9.617421204979642e-06, "loss": 0.0038, "step": 52020 }, { "epoch": 0.8513458234475988, "grad_norm": 0.07551705092191696, "learning_rate": 9.617147259617387e-06, "loss": 0.002, "step": 52030 }, { "epoch": 0.8515094493986747, "grad_norm": 0.07560122013092041, "learning_rate": 9.616873220115429e-06, "loss": 0.0027, "step": 52040 }, { "epoch": 0.8516730753497505, "grad_norm": 0.09840092062950134, "learning_rate": 9.61659908647935e-06, "loss": 0.0038, "step": 52050 }, { "epoch": 0.8518367013008263, "grad_norm": 0.04832102358341217, "learning_rate": 9.616324858714743e-06, "loss": 0.0016, "step": 52060 }, { "epoch": 0.8520003272519021, "grad_norm": 0.0734056904911995, "learning_rate": 9.616050536827199e-06, "loss": 0.0015, "step": 52070 }, { "epoch": 0.852163953202978, "grad_norm": 0.048609182238578796, "learning_rate": 9.61577612082231e-06, "loss": 0.0015, "step": 52080 }, { "epoch": 0.8523275791540539, "grad_norm": 0.09903980046510696, "learning_rate": 9.61550161070567e-06, "loss": 0.0019, "step": 52090 }, { "epoch": 0.8524912051051297, "grad_norm": 0.0684807077050209, "learning_rate": 9.61522700648288e-06, "loss": 0.0019, "step": 52100 }, { "epoch": 0.8526548310562055, "grad_norm": 0.08901026844978333, "learning_rate": 9.614952308159535e-06, "loss": 0.002, "step": 52110 }, { "epoch": 0.8528184570072813, "grad_norm": 0.014284414239227772, "learning_rate": 9.614677515741238e-06, "loss": 0.0026, "step": 52120 }, { "epoch": 0.8529820829583572, "grad_norm": 0.05737559497356415, "learning_rate": 9.61440262923359e-06, "loss": 0.0027, "step": 52130 }, { "epoch": 0.8531457089094331, "grad_norm": 0.0647118017077446, "learning_rate": 9.614127648642197e-06, "loss": 0.0018, "step": 52140 }, { "epoch": 0.8533093348605089, "grad_norm": 0.11131154745817184, "learning_rate": 9.613852573972666e-06, "loss": 0.0027, "step": 52150 }, { "epoch": 0.8534729608115847, "grad_norm": 0.09298146516084671, "learning_rate": 9.613577405230605e-06, "loss": 0.0021, "step": 52160 }, { "epoch": 0.8536365867626605, "grad_norm": 0.0993051752448082, "learning_rate": 9.613302142421623e-06, "loss": 0.003, "step": 52170 }, { "epoch": 0.8538002127137364, "grad_norm": 0.1007479652762413, "learning_rate": 9.613026785551336e-06, "loss": 0.0019, "step": 52180 }, { "epoch": 0.8539638386648123, "grad_norm": 0.09583553671836853, "learning_rate": 9.612751334625355e-06, "loss": 0.0013, "step": 52190 }, { "epoch": 0.8541274646158881, "grad_norm": 0.025533199310302734, "learning_rate": 9.612475789649297e-06, "loss": 0.0041, "step": 52200 }, { "epoch": 0.854291090566964, "grad_norm": 0.14080895483493805, "learning_rate": 9.612200150628781e-06, "loss": 0.0021, "step": 52210 }, { "epoch": 0.8544547165180397, "grad_norm": 0.16400188207626343, "learning_rate": 9.611924417569424e-06, "loss": 0.0038, "step": 52220 }, { "epoch": 0.8546183424691156, "grad_norm": 0.05451652780175209, "learning_rate": 9.611648590476855e-06, "loss": 0.0035, "step": 52230 }, { "epoch": 0.8547819684201915, "grad_norm": 0.06889555603265762, "learning_rate": 9.61137266935669e-06, "loss": 0.0014, "step": 52240 }, { "epoch": 0.8549455943712673, "grad_norm": 0.11965793371200562, "learning_rate": 9.61109665421456e-06, "loss": 0.0034, "step": 52250 }, { "epoch": 0.8551092203223432, "grad_norm": 0.036249756813049316, "learning_rate": 9.610820545056089e-06, "loss": 0.002, "step": 52260 }, { "epoch": 0.8552728462734189, "grad_norm": 0.1369425505399704, "learning_rate": 9.61054434188691e-06, "loss": 0.002, "step": 52270 }, { "epoch": 0.8554364722244948, "grad_norm": 0.47064632177352905, "learning_rate": 9.610268044712651e-06, "loss": 0.002, "step": 52280 }, { "epoch": 0.8556000981755707, "grad_norm": 0.02238067053258419, "learning_rate": 9.60999165353895e-06, "loss": 0.0023, "step": 52290 }, { "epoch": 0.8557637241266465, "grad_norm": 0.014495587907731533, "learning_rate": 9.609715168371439e-06, "loss": 0.004, "step": 52300 }, { "epoch": 0.8559273500777224, "grad_norm": 0.09188485890626907, "learning_rate": 9.609438589215755e-06, "loss": 0.0019, "step": 52310 }, { "epoch": 0.8560909760287981, "grad_norm": 0.14291232824325562, "learning_rate": 9.609161916077538e-06, "loss": 0.0017, "step": 52320 }, { "epoch": 0.856254601979874, "grad_norm": 0.06383360177278519, "learning_rate": 9.60888514896243e-06, "loss": 0.002, "step": 52330 }, { "epoch": 0.8564182279309498, "grad_norm": 0.23415899276733398, "learning_rate": 9.608608287876075e-06, "loss": 0.0029, "step": 52340 }, { "epoch": 0.8565818538820257, "grad_norm": 0.09575607627630234, "learning_rate": 9.608331332824116e-06, "loss": 0.0022, "step": 52350 }, { "epoch": 0.8567454798331016, "grad_norm": 0.041918374598026276, "learning_rate": 9.608054283812199e-06, "loss": 0.0025, "step": 52360 }, { "epoch": 0.8569091057841773, "grad_norm": 0.06201367452740669, "learning_rate": 9.607777140845975e-06, "loss": 0.0029, "step": 52370 }, { "epoch": 0.8570727317352532, "grad_norm": 0.055141184478998184, "learning_rate": 9.607499903931094e-06, "loss": 0.0038, "step": 52380 }, { "epoch": 0.857236357686329, "grad_norm": 0.06747132539749146, "learning_rate": 9.607222573073209e-06, "loss": 0.0022, "step": 52390 }, { "epoch": 0.8573999836374049, "grad_norm": 0.11090201139450073, "learning_rate": 9.606945148277974e-06, "loss": 0.0024, "step": 52400 }, { "epoch": 0.8575636095884808, "grad_norm": 0.03525586798787117, "learning_rate": 9.606667629551045e-06, "loss": 0.0024, "step": 52410 }, { "epoch": 0.8577272355395565, "grad_norm": 0.06260069459676743, "learning_rate": 9.606390016898081e-06, "loss": 0.0033, "step": 52420 }, { "epoch": 0.8578908614906324, "grad_norm": 0.07434321939945221, "learning_rate": 9.60611231032474e-06, "loss": 0.0026, "step": 52430 }, { "epoch": 0.8580544874417082, "grad_norm": 0.08914104849100113, "learning_rate": 9.605834509836688e-06, "loss": 0.0033, "step": 52440 }, { "epoch": 0.8582181133927841, "grad_norm": 0.10508634150028229, "learning_rate": 9.60555661543959e-06, "loss": 0.0025, "step": 52450 }, { "epoch": 0.85838173934386, "grad_norm": 0.12989120185375214, "learning_rate": 9.605278627139107e-06, "loss": 0.0022, "step": 52460 }, { "epoch": 0.8585453652949357, "grad_norm": 0.1363268792629242, "learning_rate": 9.60500054494091e-06, "loss": 0.0045, "step": 52470 }, { "epoch": 0.8587089912460116, "grad_norm": 0.04716430604457855, "learning_rate": 9.604722368850668e-06, "loss": 0.0023, "step": 52480 }, { "epoch": 0.8588726171970874, "grad_norm": 0.2414788454771042, "learning_rate": 9.604444098874052e-06, "loss": 0.0034, "step": 52490 }, { "epoch": 0.8590362431481633, "grad_norm": 0.16054204106330872, "learning_rate": 9.60416573501674e-06, "loss": 0.0032, "step": 52500 }, { "epoch": 0.8591998690992392, "grad_norm": 0.05403166636824608, "learning_rate": 9.6038872772844e-06, "loss": 0.0036, "step": 52510 }, { "epoch": 0.859363495050315, "grad_norm": 0.11466041952371597, "learning_rate": 9.603608725682717e-06, "loss": 0.003, "step": 52520 }, { "epoch": 0.8595271210013908, "grad_norm": 0.09208092093467712, "learning_rate": 9.603330080217366e-06, "loss": 0.0038, "step": 52530 }, { "epoch": 0.8596907469524666, "grad_norm": 0.13112366199493408, "learning_rate": 9.603051340894031e-06, "loss": 0.0018, "step": 52540 }, { "epoch": 0.8598543729035425, "grad_norm": 0.10859636962413788, "learning_rate": 9.602772507718393e-06, "loss": 0.0033, "step": 52550 }, { "epoch": 0.8600179988546184, "grad_norm": 0.21014368534088135, "learning_rate": 9.602493580696137e-06, "loss": 0.0019, "step": 52560 }, { "epoch": 0.8601816248056942, "grad_norm": 0.06781033426523209, "learning_rate": 9.602214559832953e-06, "loss": 0.0026, "step": 52570 }, { "epoch": 0.86034525075677, "grad_norm": 0.23799176514148712, "learning_rate": 9.601935445134528e-06, "loss": 0.0029, "step": 52580 }, { "epoch": 0.8605088767078458, "grad_norm": 0.09702961146831512, "learning_rate": 9.601656236606553e-06, "loss": 0.0028, "step": 52590 }, { "epoch": 0.8606725026589217, "grad_norm": 0.10607483237981796, "learning_rate": 9.60137693425472e-06, "loss": 0.0022, "step": 52600 }, { "epoch": 0.8608361286099976, "grad_norm": 0.02545224502682686, "learning_rate": 9.601097538084724e-06, "loss": 0.002, "step": 52610 }, { "epoch": 0.8609997545610734, "grad_norm": 0.08961234241724014, "learning_rate": 9.600818048102265e-06, "loss": 0.0066, "step": 52620 }, { "epoch": 0.8611633805121492, "grad_norm": 0.02086871862411499, "learning_rate": 9.600538464313036e-06, "loss": 0.0034, "step": 52630 }, { "epoch": 0.861327006463225, "grad_norm": 0.17413410544395447, "learning_rate": 9.600258786722743e-06, "loss": 0.0033, "step": 52640 }, { "epoch": 0.8614906324143009, "grad_norm": 0.1064295545220375, "learning_rate": 9.599979015337084e-06, "loss": 0.0039, "step": 52650 }, { "epoch": 0.8616542583653768, "grad_norm": 0.10944345593452454, "learning_rate": 9.599699150161765e-06, "loss": 0.0038, "step": 52660 }, { "epoch": 0.8618178843164526, "grad_norm": 0.11148863285779953, "learning_rate": 9.599419191202492e-06, "loss": 0.0025, "step": 52670 }, { "epoch": 0.8619815102675285, "grad_norm": 0.146977037191391, "learning_rate": 9.599139138464975e-06, "loss": 0.0019, "step": 52680 }, { "epoch": 0.8621451362186042, "grad_norm": 0.03597179427742958, "learning_rate": 9.59885899195492e-06, "loss": 0.0021, "step": 52690 }, { "epoch": 0.8623087621696801, "grad_norm": 0.12595489621162415, "learning_rate": 9.598578751678042e-06, "loss": 0.0021, "step": 52700 }, { "epoch": 0.862472388120756, "grad_norm": 0.09466482698917389, "learning_rate": 9.598298417640056e-06, "loss": 0.0018, "step": 52710 }, { "epoch": 0.8626360140718318, "grad_norm": 0.1120712161064148, "learning_rate": 9.598017989846675e-06, "loss": 0.0029, "step": 52720 }, { "epoch": 0.8627996400229077, "grad_norm": 0.012389987707138062, "learning_rate": 9.597737468303616e-06, "loss": 0.0024, "step": 52730 }, { "epoch": 0.8629632659739834, "grad_norm": 1.258385419845581, "learning_rate": 9.597456853016602e-06, "loss": 0.0025, "step": 52740 }, { "epoch": 0.8631268919250593, "grad_norm": 0.20502237975597382, "learning_rate": 9.597176143991354e-06, "loss": 0.003, "step": 52750 }, { "epoch": 0.8632905178761352, "grad_norm": 0.05570090934634209, "learning_rate": 9.596895341233592e-06, "loss": 0.0031, "step": 52760 }, { "epoch": 0.863454143827211, "grad_norm": 0.12746763229370117, "learning_rate": 9.596614444749044e-06, "loss": 0.0054, "step": 52770 }, { "epoch": 0.8636177697782869, "grad_norm": 0.10286043584346771, "learning_rate": 9.596333454543436e-06, "loss": 0.0027, "step": 52780 }, { "epoch": 0.8637813957293626, "grad_norm": 0.03344634920358658, "learning_rate": 9.5960523706225e-06, "loss": 0.0043, "step": 52790 }, { "epoch": 0.8639450216804385, "grad_norm": 0.2499445080757141, "learning_rate": 9.595771192991962e-06, "loss": 0.0027, "step": 52800 }, { "epoch": 0.8641086476315144, "grad_norm": 0.08369261026382446, "learning_rate": 9.595489921657561e-06, "loss": 0.0038, "step": 52810 }, { "epoch": 0.8642722735825902, "grad_norm": 0.054892003536224365, "learning_rate": 9.595208556625027e-06, "loss": 0.0036, "step": 52820 }, { "epoch": 0.8644358995336661, "grad_norm": 0.09123466163873672, "learning_rate": 9.5949270979001e-06, "loss": 0.0049, "step": 52830 }, { "epoch": 0.8645995254847418, "grad_norm": 0.06404386460781097, "learning_rate": 9.594645545488516e-06, "loss": 0.0026, "step": 52840 }, { "epoch": 0.8647631514358177, "grad_norm": 0.07602690160274506, "learning_rate": 9.594363899396017e-06, "loss": 0.0032, "step": 52850 }, { "epoch": 0.8649267773868936, "grad_norm": 0.05652705952525139, "learning_rate": 9.594082159628346e-06, "loss": 0.0023, "step": 52860 }, { "epoch": 0.8650904033379694, "grad_norm": 0.05874099209904671, "learning_rate": 9.593800326191247e-06, "loss": 0.0029, "step": 52870 }, { "epoch": 0.8652540292890453, "grad_norm": 0.07953371852636337, "learning_rate": 9.593518399090467e-06, "loss": 0.0032, "step": 52880 }, { "epoch": 0.865417655240121, "grad_norm": 0.09407541155815125, "learning_rate": 9.593236378331752e-06, "loss": 0.0018, "step": 52890 }, { "epoch": 0.8655812811911969, "grad_norm": 0.13764885067939758, "learning_rate": 9.592954263920854e-06, "loss": 0.0026, "step": 52900 }, { "epoch": 0.8657449071422728, "grad_norm": 0.17107166349887848, "learning_rate": 9.592672055863526e-06, "loss": 0.0024, "step": 52910 }, { "epoch": 0.8659085330933486, "grad_norm": 0.09284523874521255, "learning_rate": 9.592389754165518e-06, "loss": 0.0026, "step": 52920 }, { "epoch": 0.8660721590444245, "grad_norm": 0.1356283575296402, "learning_rate": 9.592107358832591e-06, "loss": 0.0021, "step": 52930 }, { "epoch": 0.8662357849955002, "grad_norm": 0.031475383788347244, "learning_rate": 9.591824869870498e-06, "loss": 0.0016, "step": 52940 }, { "epoch": 0.8663994109465761, "grad_norm": 0.21384744346141815, "learning_rate": 9.591542287285005e-06, "loss": 0.0041, "step": 52950 }, { "epoch": 0.866563036897652, "grad_norm": 0.11062335968017578, "learning_rate": 9.591259611081868e-06, "loss": 0.0027, "step": 52960 }, { "epoch": 0.8667266628487278, "grad_norm": 0.12457171827554703, "learning_rate": 9.59097684126685e-06, "loss": 0.0029, "step": 52970 }, { "epoch": 0.8668902887998037, "grad_norm": 0.05341683328151703, "learning_rate": 9.59069397784572e-06, "loss": 0.0032, "step": 52980 }, { "epoch": 0.8670539147508795, "grad_norm": 0.11978679150342941, "learning_rate": 9.590411020824245e-06, "loss": 0.0037, "step": 52990 }, { "epoch": 0.8672175407019553, "grad_norm": 0.10608495026826859, "learning_rate": 9.590127970208193e-06, "loss": 0.005, "step": 53000 }, { "epoch": 0.8673811666530312, "grad_norm": 0.11128443479537964, "learning_rate": 9.589844826003335e-06, "loss": 0.0037, "step": 53010 }, { "epoch": 0.867544792604107, "grad_norm": 0.04346242919564247, "learning_rate": 9.589561588215444e-06, "loss": 0.0028, "step": 53020 }, { "epoch": 0.8677084185551829, "grad_norm": 0.010786750353872776, "learning_rate": 9.589278256850296e-06, "loss": 0.0011, "step": 53030 }, { "epoch": 0.8678720445062587, "grad_norm": 0.05752936005592346, "learning_rate": 9.588994831913668e-06, "loss": 0.0033, "step": 53040 }, { "epoch": 0.8680356704573345, "grad_norm": 0.09142913669347763, "learning_rate": 9.588711313411339e-06, "loss": 0.0032, "step": 53050 }, { "epoch": 0.8681992964084104, "grad_norm": 0.10135944932699203, "learning_rate": 9.588427701349086e-06, "loss": 0.002, "step": 53060 }, { "epoch": 0.8683629223594862, "grad_norm": 0.08594942837953568, "learning_rate": 9.588143995732696e-06, "loss": 0.0018, "step": 53070 }, { "epoch": 0.8685265483105621, "grad_norm": 0.032918550074100494, "learning_rate": 9.587860196567954e-06, "loss": 0.0025, "step": 53080 }, { "epoch": 0.8686901742616379, "grad_norm": 0.017636023461818695, "learning_rate": 9.587576303860641e-06, "loss": 0.0032, "step": 53090 }, { "epoch": 0.8688538002127137, "grad_norm": 0.12390098720788956, "learning_rate": 9.58729231761655e-06, "loss": 0.0029, "step": 53100 }, { "epoch": 0.8690174261637896, "grad_norm": 0.10850567370653152, "learning_rate": 9.587008237841471e-06, "loss": 0.003, "step": 53110 }, { "epoch": 0.8691810521148654, "grad_norm": 0.019853414967656136, "learning_rate": 9.586724064541195e-06, "loss": 0.0017, "step": 53120 }, { "epoch": 0.8693446780659413, "grad_norm": 0.04399218410253525, "learning_rate": 9.586439797721515e-06, "loss": 0.0022, "step": 53130 }, { "epoch": 0.8695083040170171, "grad_norm": 0.07243865728378296, "learning_rate": 9.586155437388229e-06, "loss": 0.0018, "step": 53140 }, { "epoch": 0.869671929968093, "grad_norm": 0.11330705136060715, "learning_rate": 9.585870983547134e-06, "loss": 0.0027, "step": 53150 }, { "epoch": 0.8698355559191688, "grad_norm": 0.061940569430589676, "learning_rate": 9.585586436204028e-06, "loss": 0.0053, "step": 53160 }, { "epoch": 0.8699991818702446, "grad_norm": 0.1306333839893341, "learning_rate": 9.585301795364716e-06, "loss": 0.003, "step": 53170 }, { "epoch": 0.8701628078213205, "grad_norm": 0.07640305161476135, "learning_rate": 9.585017061035e-06, "loss": 0.002, "step": 53180 }, { "epoch": 0.8703264337723963, "grad_norm": 0.5528672933578491, "learning_rate": 9.584732233220687e-06, "loss": 0.0031, "step": 53190 }, { "epoch": 0.8704900597234722, "grad_norm": 0.0881960541009903, "learning_rate": 9.584447311927582e-06, "loss": 0.0032, "step": 53200 }, { "epoch": 0.8706536856745479, "grad_norm": 0.1179308146238327, "learning_rate": 9.584162297161494e-06, "loss": 0.0041, "step": 53210 }, { "epoch": 0.8708173116256238, "grad_norm": 0.09850706160068512, "learning_rate": 9.583877188928236e-06, "loss": 0.0014, "step": 53220 }, { "epoch": 0.8709809375766997, "grad_norm": 0.06143893674015999, "learning_rate": 9.583591987233622e-06, "loss": 0.0059, "step": 53230 }, { "epoch": 0.8711445635277755, "grad_norm": 0.127852663397789, "learning_rate": 9.583306692083464e-06, "loss": 0.0034, "step": 53240 }, { "epoch": 0.8713081894788514, "grad_norm": 0.007084489334374666, "learning_rate": 9.58302130348358e-06, "loss": 0.0011, "step": 53250 }, { "epoch": 0.8714718154299271, "grad_norm": 0.5313198566436768, "learning_rate": 9.58273582143979e-06, "loss": 0.0023, "step": 53260 }, { "epoch": 0.871635441381003, "grad_norm": 0.05250276252627373, "learning_rate": 9.582450245957915e-06, "loss": 0.0019, "step": 53270 }, { "epoch": 0.8717990673320789, "grad_norm": 0.037608951330184937, "learning_rate": 9.582164577043776e-06, "loss": 0.0025, "step": 53280 }, { "epoch": 0.8719626932831547, "grad_norm": 0.15086929500102997, "learning_rate": 9.581878814703198e-06, "loss": 0.0029, "step": 53290 }, { "epoch": 0.8721263192342306, "grad_norm": 0.05848635360598564, "learning_rate": 9.581592958942008e-06, "loss": 0.0024, "step": 53300 }, { "epoch": 0.8722899451853063, "grad_norm": 0.14189480245113373, "learning_rate": 9.581307009766034e-06, "loss": 0.0036, "step": 53310 }, { "epoch": 0.8724535711363822, "grad_norm": 0.14678668975830078, "learning_rate": 9.581020967181106e-06, "loss": 0.0027, "step": 53320 }, { "epoch": 0.8726171970874581, "grad_norm": 0.09608741849660873, "learning_rate": 9.58073483119306e-06, "loss": 0.0046, "step": 53330 }, { "epoch": 0.8727808230385339, "grad_norm": 0.12252368032932281, "learning_rate": 9.58044860180772e-06, "loss": 0.0029, "step": 53340 }, { "epoch": 0.8729444489896098, "grad_norm": 0.26075541973114014, "learning_rate": 9.580162279030933e-06, "loss": 0.0031, "step": 53350 }, { "epoch": 0.8731080749406855, "grad_norm": 0.2753729522228241, "learning_rate": 9.579875862868533e-06, "loss": 0.0014, "step": 53360 }, { "epoch": 0.8732717008917614, "grad_norm": 0.09387636929750443, "learning_rate": 9.579589353326357e-06, "loss": 0.0032, "step": 53370 }, { "epoch": 0.8734353268428373, "grad_norm": 0.07718998938798904, "learning_rate": 9.57930275041025e-06, "loss": 0.0028, "step": 53380 }, { "epoch": 0.8735989527939131, "grad_norm": 0.11781702935695648, "learning_rate": 9.579016054126054e-06, "loss": 0.0068, "step": 53390 }, { "epoch": 0.873762578744989, "grad_norm": 0.06752943247556686, "learning_rate": 9.578729264479615e-06, "loss": 0.0021, "step": 53400 }, { "epoch": 0.8739262046960647, "grad_norm": 0.06745772063732147, "learning_rate": 9.57844238147678e-06, "loss": 0.0027, "step": 53410 }, { "epoch": 0.8740898306471406, "grad_norm": 0.05314052477478981, "learning_rate": 9.5781554051234e-06, "loss": 0.0021, "step": 53420 }, { "epoch": 0.8742534565982165, "grad_norm": 0.07847918570041656, "learning_rate": 9.577868335425323e-06, "loss": 0.0017, "step": 53430 }, { "epoch": 0.8744170825492923, "grad_norm": 0.08980578184127808, "learning_rate": 9.577581172388406e-06, "loss": 0.0031, "step": 53440 }, { "epoch": 0.8745807085003682, "grad_norm": 0.1013021171092987, "learning_rate": 9.5772939160185e-06, "loss": 0.0023, "step": 53450 }, { "epoch": 0.874744334451444, "grad_norm": 0.16358695924282074, "learning_rate": 9.577006566321465e-06, "loss": 0.0021, "step": 53460 }, { "epoch": 0.8749079604025198, "grad_norm": 0.0903020054101944, "learning_rate": 9.576719123303158e-06, "loss": 0.004, "step": 53470 }, { "epoch": 0.8750715863535957, "grad_norm": 0.08821499347686768, "learning_rate": 9.576431586969441e-06, "loss": 0.0038, "step": 53480 }, { "epoch": 0.8752352123046715, "grad_norm": 0.04367559775710106, "learning_rate": 9.576143957326177e-06, "loss": 0.0017, "step": 53490 }, { "epoch": 0.8753988382557474, "grad_norm": 0.1289144903421402, "learning_rate": 9.575856234379228e-06, "loss": 0.003, "step": 53500 }, { "epoch": 0.8755624642068232, "grad_norm": 0.22532901167869568, "learning_rate": 9.575568418134463e-06, "loss": 0.0029, "step": 53510 }, { "epoch": 0.875726090157899, "grad_norm": 0.07569871097803116, "learning_rate": 9.575280508597748e-06, "loss": 0.002, "step": 53520 }, { "epoch": 0.8758897161089749, "grad_norm": 0.20920537412166595, "learning_rate": 9.574992505774956e-06, "loss": 0.0023, "step": 53530 }, { "epoch": 0.8760533420600507, "grad_norm": 0.10273563116788864, "learning_rate": 9.574704409671958e-06, "loss": 0.0039, "step": 53540 }, { "epoch": 0.8762169680111266, "grad_norm": 0.1885158270597458, "learning_rate": 9.574416220294626e-06, "loss": 0.0023, "step": 53550 }, { "epoch": 0.8763805939622024, "grad_norm": 0.13255798816680908, "learning_rate": 9.574127937648838e-06, "loss": 0.0023, "step": 53560 }, { "epoch": 0.8765442199132782, "grad_norm": 0.04263634234666824, "learning_rate": 9.573839561740472e-06, "loss": 0.0012, "step": 53570 }, { "epoch": 0.8767078458643541, "grad_norm": 0.03930245712399483, "learning_rate": 9.573551092575407e-06, "loss": 0.0017, "step": 53580 }, { "epoch": 0.8768714718154299, "grad_norm": 0.0897020623087883, "learning_rate": 9.573262530159526e-06, "loss": 0.0014, "step": 53590 }, { "epoch": 0.8770350977665058, "grad_norm": 0.20007702708244324, "learning_rate": 9.572973874498711e-06, "loss": 0.0025, "step": 53600 }, { "epoch": 0.8771987237175816, "grad_norm": 0.03152500092983246, "learning_rate": 9.572685125598848e-06, "loss": 0.0025, "step": 53610 }, { "epoch": 0.8773623496686574, "grad_norm": 0.29023414850234985, "learning_rate": 9.572396283465824e-06, "loss": 0.002, "step": 53620 }, { "epoch": 0.8775259756197333, "grad_norm": 0.04953942447900772, "learning_rate": 9.572107348105529e-06, "loss": 0.0027, "step": 53630 }, { "epoch": 0.8776896015708091, "grad_norm": 0.08470457047224045, "learning_rate": 9.571818319523851e-06, "loss": 0.0019, "step": 53640 }, { "epoch": 0.877853227521885, "grad_norm": 0.13768483698368073, "learning_rate": 9.571529197726688e-06, "loss": 0.0016, "step": 53650 }, { "epoch": 0.8780168534729608, "grad_norm": 0.04462156444787979, "learning_rate": 9.571239982719932e-06, "loss": 0.0018, "step": 53660 }, { "epoch": 0.8781804794240367, "grad_norm": 0.20762935280799866, "learning_rate": 9.57095067450948e-06, "loss": 0.0033, "step": 53670 }, { "epoch": 0.8783441053751125, "grad_norm": 0.08783992379903793, "learning_rate": 9.570661273101233e-06, "loss": 0.0034, "step": 53680 }, { "epoch": 0.8785077313261883, "grad_norm": 0.1511642336845398, "learning_rate": 9.570371778501088e-06, "loss": 0.003, "step": 53690 }, { "epoch": 0.8786713572772642, "grad_norm": 0.07341808825731277, "learning_rate": 9.57008219071495e-06, "loss": 0.0018, "step": 53700 }, { "epoch": 0.87883498322834, "grad_norm": 0.09529939293861389, "learning_rate": 9.569792509748723e-06, "loss": 0.0026, "step": 53710 }, { "epoch": 0.8789986091794159, "grad_norm": 0.013662241399288177, "learning_rate": 9.569502735608314e-06, "loss": 0.0014, "step": 53720 }, { "epoch": 0.8791622351304917, "grad_norm": 0.0594184547662735, "learning_rate": 9.569212868299628e-06, "loss": 0.0016, "step": 53730 }, { "epoch": 0.8793258610815675, "grad_norm": 0.02800312452018261, "learning_rate": 9.568922907828579e-06, "loss": 0.0033, "step": 53740 }, { "epoch": 0.8794894870326434, "grad_norm": 0.1796637773513794, "learning_rate": 9.568632854201078e-06, "loss": 0.0026, "step": 53750 }, { "epoch": 0.8796531129837192, "grad_norm": 0.18796366453170776, "learning_rate": 9.568342707423038e-06, "loss": 0.0037, "step": 53760 }, { "epoch": 0.8798167389347951, "grad_norm": 0.028798678889870644, "learning_rate": 9.568052467500377e-06, "loss": 0.0029, "step": 53770 }, { "epoch": 0.879980364885871, "grad_norm": 0.0769519954919815, "learning_rate": 9.567762134439009e-06, "loss": 0.0025, "step": 53780 }, { "epoch": 0.8801439908369467, "grad_norm": 0.07982081919908524, "learning_rate": 9.567471708244857e-06, "loss": 0.0038, "step": 53790 }, { "epoch": 0.8803076167880226, "grad_norm": 0.0377269983291626, "learning_rate": 9.567181188923841e-06, "loss": 0.0011, "step": 53800 }, { "epoch": 0.8804712427390984, "grad_norm": 0.04987208545207977, "learning_rate": 9.566890576481885e-06, "loss": 0.0027, "step": 53810 }, { "epoch": 0.8806348686901743, "grad_norm": 0.08779017627239227, "learning_rate": 9.566599870924914e-06, "loss": 0.0028, "step": 53820 }, { "epoch": 0.8807984946412502, "grad_norm": 0.028649374842643738, "learning_rate": 9.566309072258854e-06, "loss": 0.0026, "step": 53830 }, { "epoch": 0.8809621205923259, "grad_norm": 0.02745603397488594, "learning_rate": 9.566018180489638e-06, "loss": 0.0034, "step": 53840 }, { "epoch": 0.8811257465434018, "grad_norm": 0.09050844609737396, "learning_rate": 9.565727195623194e-06, "loss": 0.0014, "step": 53850 }, { "epoch": 0.8812893724944776, "grad_norm": 0.09002186357975006, "learning_rate": 9.565436117665455e-06, "loss": 0.0019, "step": 53860 }, { "epoch": 0.8814529984455535, "grad_norm": 0.08195699751377106, "learning_rate": 9.565144946622356e-06, "loss": 0.0017, "step": 53870 }, { "epoch": 0.8816166243966294, "grad_norm": 0.15341678261756897, "learning_rate": 9.564853682499835e-06, "loss": 0.0023, "step": 53880 }, { "epoch": 0.8817802503477051, "grad_norm": 0.11997649818658829, "learning_rate": 9.56456232530383e-06, "loss": 0.0027, "step": 53890 }, { "epoch": 0.881943876298781, "grad_norm": 0.054406486451625824, "learning_rate": 9.564270875040279e-06, "loss": 0.0022, "step": 53900 }, { "epoch": 0.8821075022498568, "grad_norm": 0.05114016309380531, "learning_rate": 9.563979331715127e-06, "loss": 0.0026, "step": 53910 }, { "epoch": 0.8822711282009327, "grad_norm": 0.045420531183481216, "learning_rate": 9.563687695334321e-06, "loss": 0.0017, "step": 53920 }, { "epoch": 0.8824347541520086, "grad_norm": 0.043209608644247055, "learning_rate": 9.563395965903801e-06, "loss": 0.0025, "step": 53930 }, { "epoch": 0.8825983801030843, "grad_norm": 0.12632648646831512, "learning_rate": 9.56310414342952e-06, "loss": 0.0039, "step": 53940 }, { "epoch": 0.8827620060541602, "grad_norm": 0.055554043501615524, "learning_rate": 9.562812227917425e-06, "loss": 0.0017, "step": 53950 }, { "epoch": 0.882925632005236, "grad_norm": 0.1221635639667511, "learning_rate": 9.56252021937347e-06, "loss": 0.0025, "step": 53960 }, { "epoch": 0.8830892579563119, "grad_norm": 0.02618161402642727, "learning_rate": 9.562228117803607e-06, "loss": 0.0024, "step": 53970 }, { "epoch": 0.8832528839073878, "grad_norm": 0.12749531865119934, "learning_rate": 9.561935923213794e-06, "loss": 0.003, "step": 53980 }, { "epoch": 0.8834165098584635, "grad_norm": 0.10604847222566605, "learning_rate": 9.561643635609987e-06, "loss": 0.0018, "step": 53990 }, { "epoch": 0.8835801358095394, "grad_norm": 0.014578521251678467, "learning_rate": 9.561351254998146e-06, "loss": 0.0019, "step": 54000 }, { "epoch": 0.8837437617606152, "grad_norm": 0.04756443202495575, "learning_rate": 9.561058781384232e-06, "loss": 0.0023, "step": 54010 }, { "epoch": 0.8839073877116911, "grad_norm": 0.05209808424115181, "learning_rate": 9.560766214774208e-06, "loss": 0.0015, "step": 54020 }, { "epoch": 0.884071013662767, "grad_norm": 0.04682512953877449, "learning_rate": 9.56047355517404e-06, "loss": 0.0015, "step": 54030 }, { "epoch": 0.8842346396138427, "grad_norm": 0.012812407687306404, "learning_rate": 9.560180802589695e-06, "loss": 0.0035, "step": 54040 }, { "epoch": 0.8843982655649186, "grad_norm": 0.023143291473388672, "learning_rate": 9.55988795702714e-06, "loss": 0.0038, "step": 54050 }, { "epoch": 0.8845618915159944, "grad_norm": 0.09212926030158997, "learning_rate": 9.559595018492351e-06, "loss": 0.002, "step": 54060 }, { "epoch": 0.8847255174670703, "grad_norm": 0.10041774064302444, "learning_rate": 9.559301986991296e-06, "loss": 0.003, "step": 54070 }, { "epoch": 0.8848891434181462, "grad_norm": 0.16437667608261108, "learning_rate": 9.55900886252995e-06, "loss": 0.0023, "step": 54080 }, { "epoch": 0.885052769369222, "grad_norm": 0.13295753300189972, "learning_rate": 9.55871564511429e-06, "loss": 0.0025, "step": 54090 }, { "epoch": 0.8852163953202978, "grad_norm": 0.038814976811409, "learning_rate": 9.558422334750297e-06, "loss": 0.004, "step": 54100 }, { "epoch": 0.8853800212713736, "grad_norm": 0.05001340061426163, "learning_rate": 9.558128931443949e-06, "loss": 0.0016, "step": 54110 }, { "epoch": 0.8855436472224495, "grad_norm": 0.03200601786375046, "learning_rate": 9.557835435201228e-06, "loss": 0.0025, "step": 54120 }, { "epoch": 0.8857072731735253, "grad_norm": 0.1038254052400589, "learning_rate": 9.557541846028117e-06, "loss": 0.0035, "step": 54130 }, { "epoch": 0.8858708991246012, "grad_norm": 0.11069774627685547, "learning_rate": 9.557248163930606e-06, "loss": 0.0017, "step": 54140 }, { "epoch": 0.886034525075677, "grad_norm": 0.21968747675418854, "learning_rate": 9.556954388914682e-06, "loss": 0.0045, "step": 54150 }, { "epoch": 0.8861981510267528, "grad_norm": 0.12075332552194595, "learning_rate": 9.556660520986332e-06, "loss": 0.0017, "step": 54160 }, { "epoch": 0.8863617769778287, "grad_norm": 0.11352811753749847, "learning_rate": 9.556366560151549e-06, "loss": 0.0024, "step": 54170 }, { "epoch": 0.8865254029289045, "grad_norm": 0.04224403202533722, "learning_rate": 9.556072506416327e-06, "loss": 0.0024, "step": 54180 }, { "epoch": 0.8866890288799804, "grad_norm": 0.10203434526920319, "learning_rate": 9.555778359786662e-06, "loss": 0.0033, "step": 54190 }, { "epoch": 0.8868526548310562, "grad_norm": 0.06881732493638992, "learning_rate": 9.55548412026855e-06, "loss": 0.0033, "step": 54200 }, { "epoch": 0.887016280782132, "grad_norm": 0.037233490496873856, "learning_rate": 9.555189787867992e-06, "loss": 0.0028, "step": 54210 }, { "epoch": 0.8871799067332079, "grad_norm": 0.011352908797562122, "learning_rate": 9.554895362590988e-06, "loss": 0.0022, "step": 54220 }, { "epoch": 0.8873435326842837, "grad_norm": 0.25443562865257263, "learning_rate": 9.55460084444354e-06, "loss": 0.0017, "step": 54230 }, { "epoch": 0.8875071586353596, "grad_norm": 0.10848424583673477, "learning_rate": 9.554306233431656e-06, "loss": 0.0019, "step": 54240 }, { "epoch": 0.8876707845864354, "grad_norm": 0.011485688388347626, "learning_rate": 9.554011529561342e-06, "loss": 0.0022, "step": 54250 }, { "epoch": 0.8878344105375112, "grad_norm": 0.110061876475811, "learning_rate": 9.553716732838603e-06, "loss": 0.0034, "step": 54260 }, { "epoch": 0.8879980364885871, "grad_norm": 0.012095319107174873, "learning_rate": 9.553421843269454e-06, "loss": 0.0029, "step": 54270 }, { "epoch": 0.8881616624396629, "grad_norm": 0.06140303611755371, "learning_rate": 9.553126860859908e-06, "loss": 0.0026, "step": 54280 }, { "epoch": 0.8883252883907388, "grad_norm": 0.19773633778095245, "learning_rate": 9.552831785615975e-06, "loss": 0.0037, "step": 54290 }, { "epoch": 0.8884889143418147, "grad_norm": 0.11399158835411072, "learning_rate": 9.552536617543675e-06, "loss": 0.0024, "step": 54300 }, { "epoch": 0.8886525402928904, "grad_norm": 0.043944716453552246, "learning_rate": 9.552241356649025e-06, "loss": 0.0011, "step": 54310 }, { "epoch": 0.8888161662439663, "grad_norm": 0.06667038053274155, "learning_rate": 9.551946002938046e-06, "loss": 0.0033, "step": 54320 }, { "epoch": 0.8889797921950421, "grad_norm": 0.1587071716785431, "learning_rate": 9.551650556416758e-06, "loss": 0.0024, "step": 54330 }, { "epoch": 0.889143418146118, "grad_norm": 0.008233374916017056, "learning_rate": 9.551355017091189e-06, "loss": 0.0019, "step": 54340 }, { "epoch": 0.8893070440971939, "grad_norm": 0.12527155876159668, "learning_rate": 9.551059384967359e-06, "loss": 0.0024, "step": 54350 }, { "epoch": 0.8894706700482696, "grad_norm": 0.05712573230266571, "learning_rate": 9.550763660051299e-06, "loss": 0.0025, "step": 54360 }, { "epoch": 0.8896342959993455, "grad_norm": 0.33300167322158813, "learning_rate": 9.550467842349039e-06, "loss": 0.0036, "step": 54370 }, { "epoch": 0.8897979219504213, "grad_norm": 0.027307454496622086, "learning_rate": 9.55017193186661e-06, "loss": 0.0024, "step": 54380 }, { "epoch": 0.8899615479014972, "grad_norm": 0.21386107802391052, "learning_rate": 9.549875928610045e-06, "loss": 0.0049, "step": 54390 }, { "epoch": 0.8901251738525731, "grad_norm": 0.050544627010822296, "learning_rate": 9.549579832585379e-06, "loss": 0.0017, "step": 54400 }, { "epoch": 0.8902887998036488, "grad_norm": 0.04817807301878929, "learning_rate": 9.549283643798647e-06, "loss": 0.0025, "step": 54410 }, { "epoch": 0.8904524257547247, "grad_norm": 0.02797362022101879, "learning_rate": 9.548987362255896e-06, "loss": 0.0038, "step": 54420 }, { "epoch": 0.8906160517058005, "grad_norm": 0.026934903115034103, "learning_rate": 9.548690987963156e-06, "loss": 0.002, "step": 54430 }, { "epoch": 0.8907796776568764, "grad_norm": 0.0345788337290287, "learning_rate": 9.548394520926479e-06, "loss": 0.0038, "step": 54440 }, { "epoch": 0.8909433036079523, "grad_norm": 0.027998387813568115, "learning_rate": 9.548097961151905e-06, "loss": 0.0013, "step": 54450 }, { "epoch": 0.891106929559028, "grad_norm": 0.1190439760684967, "learning_rate": 9.547801308645482e-06, "loss": 0.0034, "step": 54460 }, { "epoch": 0.8912705555101039, "grad_norm": 0.0030528800562024117, "learning_rate": 9.547504563413257e-06, "loss": 0.0023, "step": 54470 }, { "epoch": 0.8914341814611797, "grad_norm": 0.13413408398628235, "learning_rate": 9.547207725461281e-06, "loss": 0.0044, "step": 54480 }, { "epoch": 0.8915978074122556, "grad_norm": 0.07790248841047287, "learning_rate": 9.546910794795608e-06, "loss": 0.0028, "step": 54490 }, { "epoch": 0.8917614333633315, "grad_norm": 0.10676666349172592, "learning_rate": 9.54661377142229e-06, "loss": 0.0045, "step": 54500 }, { "epoch": 0.8919250593144072, "grad_norm": 0.16265210509300232, "learning_rate": 9.546316655347385e-06, "loss": 0.0022, "step": 54510 }, { "epoch": 0.8920886852654831, "grad_norm": 0.09942714869976044, "learning_rate": 9.546019446576949e-06, "loss": 0.0018, "step": 54520 }, { "epoch": 0.8922523112165589, "grad_norm": 0.23793692886829376, "learning_rate": 9.545722145117043e-06, "loss": 0.0019, "step": 54530 }, { "epoch": 0.8924159371676348, "grad_norm": 0.07803639024496078, "learning_rate": 9.545424750973728e-06, "loss": 0.002, "step": 54540 }, { "epoch": 0.8925795631187107, "grad_norm": 0.20059166848659515, "learning_rate": 9.545127264153069e-06, "loss": 0.0038, "step": 54550 }, { "epoch": 0.8927431890697864, "grad_norm": 0.048696357756853104, "learning_rate": 9.54482968466113e-06, "loss": 0.0034, "step": 54560 }, { "epoch": 0.8929068150208623, "grad_norm": 0.08145347982645035, "learning_rate": 9.544532012503979e-06, "loss": 0.0026, "step": 54570 }, { "epoch": 0.8930704409719381, "grad_norm": 0.0702957957983017, "learning_rate": 9.544234247687688e-06, "loss": 0.0017, "step": 54580 }, { "epoch": 0.893234066923014, "grad_norm": 0.051775671541690826, "learning_rate": 9.543936390218321e-06, "loss": 0.0022, "step": 54590 }, { "epoch": 0.8933976928740899, "grad_norm": 0.0734405368566513, "learning_rate": 9.543638440101958e-06, "loss": 0.0021, "step": 54600 }, { "epoch": 0.8935613188251657, "grad_norm": 0.06779645383358002, "learning_rate": 9.54334039734467e-06, "loss": 0.003, "step": 54610 }, { "epoch": 0.8937249447762415, "grad_norm": 0.039400696754455566, "learning_rate": 9.543042261952537e-06, "loss": 0.0026, "step": 54620 }, { "epoch": 0.8938885707273173, "grad_norm": 0.06378346681594849, "learning_rate": 9.542744033931635e-06, "loss": 0.0024, "step": 54630 }, { "epoch": 0.8940521966783932, "grad_norm": 0.009414619766175747, "learning_rate": 9.542445713288044e-06, "loss": 0.002, "step": 54640 }, { "epoch": 0.8942158226294691, "grad_norm": 0.06544134765863419, "learning_rate": 9.54214730002785e-06, "loss": 0.0045, "step": 54650 }, { "epoch": 0.8943794485805449, "grad_norm": 0.058589305728673935, "learning_rate": 9.541848794157136e-06, "loss": 0.0023, "step": 54660 }, { "epoch": 0.8945430745316207, "grad_norm": 0.4506185054779053, "learning_rate": 9.541550195681987e-06, "loss": 0.0027, "step": 54670 }, { "epoch": 0.8947067004826965, "grad_norm": 0.14765502512454987, "learning_rate": 9.541251504608493e-06, "loss": 0.0015, "step": 54680 }, { "epoch": 0.8948703264337724, "grad_norm": 0.0939253568649292, "learning_rate": 9.540952720942741e-06, "loss": 0.002, "step": 54690 }, { "epoch": 0.8950339523848483, "grad_norm": 0.05415812134742737, "learning_rate": 9.540653844690826e-06, "loss": 0.0028, "step": 54700 }, { "epoch": 0.8951975783359241, "grad_norm": 0.06453116983175278, "learning_rate": 9.540354875858842e-06, "loss": 0.0043, "step": 54710 }, { "epoch": 0.895361204287, "grad_norm": 0.11500495672225952, "learning_rate": 9.540055814452883e-06, "loss": 0.0027, "step": 54720 }, { "epoch": 0.8955248302380757, "grad_norm": 0.033245526254177094, "learning_rate": 9.539756660479044e-06, "loss": 0.003, "step": 54730 }, { "epoch": 0.8956884561891516, "grad_norm": 0.03849424794316292, "learning_rate": 9.53945741394343e-06, "loss": 0.0021, "step": 54740 }, { "epoch": 0.8958520821402275, "grad_norm": 0.059765249490737915, "learning_rate": 9.539158074852142e-06, "loss": 0.002, "step": 54750 }, { "epoch": 0.8960157080913033, "grad_norm": 0.09765031188726425, "learning_rate": 9.538858643211278e-06, "loss": 0.0026, "step": 54760 }, { "epoch": 0.8961793340423791, "grad_norm": 0.04766819253563881, "learning_rate": 9.538559119026948e-06, "loss": 0.0018, "step": 54770 }, { "epoch": 0.8963429599934549, "grad_norm": 0.05045752227306366, "learning_rate": 9.538259502305257e-06, "loss": 0.0029, "step": 54780 }, { "epoch": 0.8965065859445308, "grad_norm": 0.10672380775213242, "learning_rate": 9.537959793052315e-06, "loss": 0.0024, "step": 54790 }, { "epoch": 0.8966702118956067, "grad_norm": 0.04293384402990341, "learning_rate": 9.537659991274232e-06, "loss": 0.0024, "step": 54800 }, { "epoch": 0.8968338378466825, "grad_norm": 0.13742229342460632, "learning_rate": 9.53736009697712e-06, "loss": 0.0024, "step": 54810 }, { "epoch": 0.8969974637977584, "grad_norm": 0.02264195680618286, "learning_rate": 9.537060110167095e-06, "loss": 0.0017, "step": 54820 }, { "epoch": 0.8971610897488341, "grad_norm": 0.031036363914608955, "learning_rate": 9.536760030850274e-06, "loss": 0.0021, "step": 54830 }, { "epoch": 0.89732471569991, "grad_norm": 0.1024598553776741, "learning_rate": 9.536459859032774e-06, "loss": 0.002, "step": 54840 }, { "epoch": 0.8974883416509859, "grad_norm": 0.16526782512664795, "learning_rate": 9.536159594720716e-06, "loss": 0.0041, "step": 54850 }, { "epoch": 0.8976519676020617, "grad_norm": 0.06376354396343231, "learning_rate": 9.535859237920219e-06, "loss": 0.0044, "step": 54860 }, { "epoch": 0.8978155935531376, "grad_norm": 0.03358641639351845, "learning_rate": 9.535558788637412e-06, "loss": 0.002, "step": 54870 }, { "epoch": 0.8979792195042133, "grad_norm": 0.14605191349983215, "learning_rate": 9.535258246878418e-06, "loss": 0.004, "step": 54880 }, { "epoch": 0.8981428454552892, "grad_norm": 0.09803743660449982, "learning_rate": 9.534957612649368e-06, "loss": 0.0017, "step": 54890 }, { "epoch": 0.8983064714063651, "grad_norm": 0.0706992968916893, "learning_rate": 9.534656885956386e-06, "loss": 0.0013, "step": 54900 }, { "epoch": 0.8984700973574409, "grad_norm": 0.025372015312314034, "learning_rate": 9.534356066805608e-06, "loss": 0.0021, "step": 54910 }, { "epoch": 0.8986337233085168, "grad_norm": 0.060549505054950714, "learning_rate": 9.534055155203164e-06, "loss": 0.0019, "step": 54920 }, { "epoch": 0.8987973492595925, "grad_norm": 0.2722710967063904, "learning_rate": 9.533754151155194e-06, "loss": 0.004, "step": 54930 }, { "epoch": 0.8989609752106684, "grad_norm": 0.12339803576469421, "learning_rate": 9.533453054667831e-06, "loss": 0.0023, "step": 54940 }, { "epoch": 0.8991246011617443, "grad_norm": 0.02769821137189865, "learning_rate": 9.533151865747218e-06, "loss": 0.0034, "step": 54950 }, { "epoch": 0.8992882271128201, "grad_norm": 0.13168689608573914, "learning_rate": 9.53285058439949e-06, "loss": 0.0013, "step": 54960 }, { "epoch": 0.899451853063896, "grad_norm": 0.08996118605136871, "learning_rate": 9.532549210630798e-06, "loss": 0.0028, "step": 54970 }, { "epoch": 0.8996154790149717, "grad_norm": 0.42861416935920715, "learning_rate": 9.532247744447281e-06, "loss": 0.0026, "step": 54980 }, { "epoch": 0.8997791049660476, "grad_norm": 0.023833613842725754, "learning_rate": 9.531946185855085e-06, "loss": 0.0032, "step": 54990 }, { "epoch": 0.8999427309171234, "grad_norm": 0.09675250202417374, "learning_rate": 9.531644534860363e-06, "loss": 0.0027, "step": 55000 }, { "epoch": 0.9001063568681993, "grad_norm": 0.05815926939249039, "learning_rate": 9.531342791469263e-06, "loss": 0.0031, "step": 55010 }, { "epoch": 0.9002699828192752, "grad_norm": 0.04331735149025917, "learning_rate": 9.531040955687935e-06, "loss": 0.0051, "step": 55020 }, { "epoch": 0.900433608770351, "grad_norm": 0.10155657678842545, "learning_rate": 9.530739027522537e-06, "loss": 0.0026, "step": 55030 }, { "epoch": 0.9005972347214268, "grad_norm": 0.033184975385665894, "learning_rate": 9.530437006979223e-06, "loss": 0.002, "step": 55040 }, { "epoch": 0.9007608606725026, "grad_norm": 0.08578771352767944, "learning_rate": 9.530134894064152e-06, "loss": 0.0025, "step": 55050 }, { "epoch": 0.9009244866235785, "grad_norm": 0.15480558574199677, "learning_rate": 9.529832688783483e-06, "loss": 0.0034, "step": 55060 }, { "epoch": 0.9010881125746544, "grad_norm": 0.06855463981628418, "learning_rate": 9.529530391143377e-06, "loss": 0.0025, "step": 55070 }, { "epoch": 0.9012517385257302, "grad_norm": 0.022339871153235435, "learning_rate": 9.52922800115e-06, "loss": 0.0054, "step": 55080 }, { "epoch": 0.901415364476806, "grad_norm": 0.04896406829357147, "learning_rate": 9.528925518809516e-06, "loss": 0.0023, "step": 55090 }, { "epoch": 0.9015789904278818, "grad_norm": 0.07121200859546661, "learning_rate": 9.528622944128089e-06, "loss": 0.0026, "step": 55100 }, { "epoch": 0.9017426163789577, "grad_norm": 0.15845845639705658, "learning_rate": 9.528320277111895e-06, "loss": 0.0027, "step": 55110 }, { "epoch": 0.9019062423300336, "grad_norm": 0.27113330364227295, "learning_rate": 9.528017517767103e-06, "loss": 0.0038, "step": 55120 }, { "epoch": 0.9020698682811094, "grad_norm": 0.09511388838291168, "learning_rate": 9.527714666099881e-06, "loss": 0.0027, "step": 55130 }, { "epoch": 0.9022334942321852, "grad_norm": 0.07102517038583755, "learning_rate": 9.52741172211641e-06, "loss": 0.0027, "step": 55140 }, { "epoch": 0.902397120183261, "grad_norm": 0.062374282628297806, "learning_rate": 9.527108685822863e-06, "loss": 0.0021, "step": 55150 }, { "epoch": 0.9025607461343369, "grad_norm": 0.3564642369747162, "learning_rate": 9.526805557225422e-06, "loss": 0.0029, "step": 55160 }, { "epoch": 0.9027243720854128, "grad_norm": 0.05562877655029297, "learning_rate": 9.526502336330264e-06, "loss": 0.0024, "step": 55170 }, { "epoch": 0.9028879980364886, "grad_norm": 0.0298482496291399, "learning_rate": 9.526199023143574e-06, "loss": 0.0025, "step": 55180 }, { "epoch": 0.9030516239875644, "grad_norm": 0.009480059146881104, "learning_rate": 9.525895617671534e-06, "loss": 0.0029, "step": 55190 }, { "epoch": 0.9032152499386402, "grad_norm": 0.06376902759075165, "learning_rate": 9.525592119920333e-06, "loss": 0.0031, "step": 55200 }, { "epoch": 0.9033788758897161, "grad_norm": 0.0394447036087513, "learning_rate": 9.525288529896157e-06, "loss": 0.0018, "step": 55210 }, { "epoch": 0.903542501840792, "grad_norm": 0.048978377133607864, "learning_rate": 9.524984847605197e-06, "loss": 0.0025, "step": 55220 }, { "epoch": 0.9037061277918678, "grad_norm": 0.20087552070617676, "learning_rate": 9.524681073053642e-06, "loss": 0.0039, "step": 55230 }, { "epoch": 0.9038697537429436, "grad_norm": 0.03934411704540253, "learning_rate": 9.524377206247691e-06, "loss": 0.0019, "step": 55240 }, { "epoch": 0.9040333796940194, "grad_norm": 0.049552544951438904, "learning_rate": 9.524073247193535e-06, "loss": 0.0036, "step": 55250 }, { "epoch": 0.9041970056450953, "grad_norm": 0.04648866876959801, "learning_rate": 9.523769195897374e-06, "loss": 0.0033, "step": 55260 }, { "epoch": 0.9043606315961712, "grad_norm": 0.1415618509054184, "learning_rate": 9.523465052365405e-06, "loss": 0.0028, "step": 55270 }, { "epoch": 0.904524257547247, "grad_norm": 0.08807181566953659, "learning_rate": 9.523160816603832e-06, "loss": 0.0017, "step": 55280 }, { "epoch": 0.9046878834983229, "grad_norm": 0.0673745647072792, "learning_rate": 9.522856488618856e-06, "loss": 0.0017, "step": 55290 }, { "epoch": 0.9048515094493986, "grad_norm": 0.19807232916355133, "learning_rate": 9.522552068416684e-06, "loss": 0.0022, "step": 55300 }, { "epoch": 0.9050151354004745, "grad_norm": 0.07490464299917221, "learning_rate": 9.522247556003521e-06, "loss": 0.003, "step": 55310 }, { "epoch": 0.9051787613515504, "grad_norm": 0.02489493228495121, "learning_rate": 9.521942951385576e-06, "loss": 0.0032, "step": 55320 }, { "epoch": 0.9053423873026262, "grad_norm": 0.15011675655841827, "learning_rate": 9.52163825456906e-06, "loss": 0.0028, "step": 55330 }, { "epoch": 0.9055060132537021, "grad_norm": 0.10482406616210938, "learning_rate": 9.521333465560186e-06, "loss": 0.0028, "step": 55340 }, { "epoch": 0.9056696392047778, "grad_norm": 0.07138556987047195, "learning_rate": 9.521028584365166e-06, "loss": 0.0029, "step": 55350 }, { "epoch": 0.9058332651558537, "grad_norm": 0.0436316579580307, "learning_rate": 9.52072361099022e-06, "loss": 0.0015, "step": 55360 }, { "epoch": 0.9059968911069296, "grad_norm": 0.034406282007694244, "learning_rate": 9.520418545441564e-06, "loss": 0.0015, "step": 55370 }, { "epoch": 0.9061605170580054, "grad_norm": 0.06747978925704956, "learning_rate": 9.520113387725418e-06, "loss": 0.0029, "step": 55380 }, { "epoch": 0.9063241430090813, "grad_norm": 0.0884062647819519, "learning_rate": 9.519808137848005e-06, "loss": 0.0024, "step": 55390 }, { "epoch": 0.906487768960157, "grad_norm": 0.13554178178310394, "learning_rate": 9.519502795815548e-06, "loss": 0.0034, "step": 55400 }, { "epoch": 0.9066513949112329, "grad_norm": 0.051258910447359085, "learning_rate": 9.519197361634271e-06, "loss": 0.0016, "step": 55410 }, { "epoch": 0.9068150208623088, "grad_norm": 0.06723268330097198, "learning_rate": 9.518891835310404e-06, "loss": 0.0017, "step": 55420 }, { "epoch": 0.9069786468133846, "grad_norm": 0.033747266978025436, "learning_rate": 9.518586216850176e-06, "loss": 0.0047, "step": 55430 }, { "epoch": 0.9071422727644605, "grad_norm": 0.10488495975732803, "learning_rate": 9.518280506259819e-06, "loss": 0.0032, "step": 55440 }, { "epoch": 0.9073058987155362, "grad_norm": 0.022250862792134285, "learning_rate": 9.517974703545564e-06, "loss": 0.0025, "step": 55450 }, { "epoch": 0.9074695246666121, "grad_norm": 0.049980755895376205, "learning_rate": 9.517668808713647e-06, "loss": 0.0019, "step": 55460 }, { "epoch": 0.907633150617688, "grad_norm": 0.051618095487356186, "learning_rate": 9.517362821770306e-06, "loss": 0.0022, "step": 55470 }, { "epoch": 0.9077967765687638, "grad_norm": 0.0711418017745018, "learning_rate": 9.517056742721779e-06, "loss": 0.0022, "step": 55480 }, { "epoch": 0.9079604025198397, "grad_norm": 0.10291346162557602, "learning_rate": 9.516750571574306e-06, "loss": 0.0015, "step": 55490 }, { "epoch": 0.9081240284709154, "grad_norm": 0.029800264164805412, "learning_rate": 9.51644430833413e-06, "loss": 0.0026, "step": 55500 }, { "epoch": 0.9082876544219913, "grad_norm": 0.09840551763772964, "learning_rate": 9.516137953007496e-06, "loss": 0.0037, "step": 55510 }, { "epoch": 0.9084512803730672, "grad_norm": 0.06032801792025566, "learning_rate": 9.51583150560065e-06, "loss": 0.002, "step": 55520 }, { "epoch": 0.908614906324143, "grad_norm": 0.8066959977149963, "learning_rate": 9.51552496611984e-06, "loss": 0.0031, "step": 55530 }, { "epoch": 0.9087785322752189, "grad_norm": 0.14513057470321655, "learning_rate": 9.515218334571317e-06, "loss": 0.0035, "step": 55540 }, { "epoch": 0.9089421582262946, "grad_norm": 0.13899222016334534, "learning_rate": 9.51491161096133e-06, "loss": 0.0026, "step": 55550 }, { "epoch": 0.9091057841773705, "grad_norm": 0.12392260134220123, "learning_rate": 9.514604795296139e-06, "loss": 0.0031, "step": 55560 }, { "epoch": 0.9092694101284464, "grad_norm": 0.0668749138712883, "learning_rate": 9.514297887581991e-06, "loss": 0.002, "step": 55570 }, { "epoch": 0.9094330360795222, "grad_norm": 0.09502653032541275, "learning_rate": 9.513990887825151e-06, "loss": 0.0023, "step": 55580 }, { "epoch": 0.9095966620305981, "grad_norm": 0.026073558256030083, "learning_rate": 9.513683796031876e-06, "loss": 0.0021, "step": 55590 }, { "epoch": 0.9097602879816739, "grad_norm": 0.09188463538885117, "learning_rate": 9.513376612208427e-06, "loss": 0.0034, "step": 55600 }, { "epoch": 0.9099239139327497, "grad_norm": 0.06103824824094772, "learning_rate": 9.513069336361067e-06, "loss": 0.0025, "step": 55610 }, { "epoch": 0.9100875398838256, "grad_norm": 0.030279893428087234, "learning_rate": 9.512761968496062e-06, "loss": 0.0018, "step": 55620 }, { "epoch": 0.9102511658349014, "grad_norm": 0.006056909915059805, "learning_rate": 9.512454508619678e-06, "loss": 0.0027, "step": 55630 }, { "epoch": 0.9104147917859773, "grad_norm": 0.03224967420101166, "learning_rate": 9.512146956738184e-06, "loss": 0.0018, "step": 55640 }, { "epoch": 0.9105784177370531, "grad_norm": 0.06799876689910889, "learning_rate": 9.51183931285785e-06, "loss": 0.0022, "step": 55650 }, { "epoch": 0.9107420436881289, "grad_norm": 0.023028656840324402, "learning_rate": 9.51153157698495e-06, "loss": 0.0026, "step": 55660 }, { "epoch": 0.9109056696392048, "grad_norm": 0.05568560212850571, "learning_rate": 9.511223749125762e-06, "loss": 0.0024, "step": 55670 }, { "epoch": 0.9110692955902806, "grad_norm": 0.0655636414885521, "learning_rate": 9.510915829286553e-06, "loss": 0.0024, "step": 55680 }, { "epoch": 0.9112329215413565, "grad_norm": 0.09980279952287674, "learning_rate": 9.510607817473609e-06, "loss": 0.0039, "step": 55690 }, { "epoch": 0.9113965474924323, "grad_norm": 0.005017060320824385, "learning_rate": 9.510299713693209e-06, "loss": 0.0015, "step": 55700 }, { "epoch": 0.9115601734435081, "grad_norm": 0.021732298657298088, "learning_rate": 9.50999151795163e-06, "loss": 0.0019, "step": 55710 }, { "epoch": 0.911723799394584, "grad_norm": 0.062302377074956894, "learning_rate": 9.509683230255165e-06, "loss": 0.0015, "step": 55720 }, { "epoch": 0.9118874253456598, "grad_norm": 0.08582182228565216, "learning_rate": 9.509374850610089e-06, "loss": 0.0019, "step": 55730 }, { "epoch": 0.9120510512967357, "grad_norm": 0.1501898318529129, "learning_rate": 9.509066379022697e-06, "loss": 0.0033, "step": 55740 }, { "epoch": 0.9122146772478115, "grad_norm": 0.03497464209794998, "learning_rate": 9.508757815499278e-06, "loss": 0.0017, "step": 55750 }, { "epoch": 0.9123783031988874, "grad_norm": 0.24105845391750336, "learning_rate": 9.508449160046121e-06, "loss": 0.0028, "step": 55760 }, { "epoch": 0.9125419291499632, "grad_norm": 0.08751703053712845, "learning_rate": 9.508140412669518e-06, "loss": 0.0016, "step": 55770 }, { "epoch": 0.912705555101039, "grad_norm": 0.12122136354446411, "learning_rate": 9.507831573375767e-06, "loss": 0.0026, "step": 55780 }, { "epoch": 0.9128691810521149, "grad_norm": 0.017590884119272232, "learning_rate": 9.507522642171165e-06, "loss": 0.0014, "step": 55790 }, { "epoch": 0.9130328070031907, "grad_norm": 0.13939017057418823, "learning_rate": 9.507213619062008e-06, "loss": 0.0024, "step": 55800 }, { "epoch": 0.9131964329542666, "grad_norm": 0.13846023380756378, "learning_rate": 9.506904504054599e-06, "loss": 0.0033, "step": 55810 }, { "epoch": 0.9133600589053424, "grad_norm": 0.030942106619477272, "learning_rate": 9.50659529715524e-06, "loss": 0.0034, "step": 55820 }, { "epoch": 0.9135236848564182, "grad_norm": 0.06791126728057861, "learning_rate": 9.506285998370237e-06, "loss": 0.0024, "step": 55830 }, { "epoch": 0.9136873108074941, "grad_norm": 0.09498511999845505, "learning_rate": 9.505976607705895e-06, "loss": 0.0026, "step": 55840 }, { "epoch": 0.9138509367585699, "grad_norm": 0.12848225235939026, "learning_rate": 9.505667125168521e-06, "loss": 0.004, "step": 55850 }, { "epoch": 0.9140145627096458, "grad_norm": 0.10676459968090057, "learning_rate": 9.505357550764427e-06, "loss": 0.0026, "step": 55860 }, { "epoch": 0.9141781886607215, "grad_norm": 0.28717249631881714, "learning_rate": 9.505047884499923e-06, "loss": 0.0025, "step": 55870 }, { "epoch": 0.9143418146117974, "grad_norm": 0.04759836941957474, "learning_rate": 9.504738126381325e-06, "loss": 0.0029, "step": 55880 }, { "epoch": 0.9145054405628733, "grad_norm": 0.11902035027742386, "learning_rate": 9.504428276414948e-06, "loss": 0.0028, "step": 55890 }, { "epoch": 0.9146690665139491, "grad_norm": 0.14975422620773315, "learning_rate": 9.504118334607108e-06, "loss": 0.002, "step": 55900 }, { "epoch": 0.914832692465025, "grad_norm": 0.21592801809310913, "learning_rate": 9.503808300964127e-06, "loss": 0.003, "step": 55910 }, { "epoch": 0.9149963184161007, "grad_norm": 0.06941696256399155, "learning_rate": 9.503498175492326e-06, "loss": 0.0019, "step": 55920 }, { "epoch": 0.9151599443671766, "grad_norm": 0.0681343600153923, "learning_rate": 9.503187958198026e-06, "loss": 0.0018, "step": 55930 }, { "epoch": 0.9153235703182525, "grad_norm": 0.024679498746991158, "learning_rate": 9.502877649087552e-06, "loss": 0.0013, "step": 55940 }, { "epoch": 0.9154871962693283, "grad_norm": 0.08356201648712158, "learning_rate": 9.502567248167234e-06, "loss": 0.0025, "step": 55950 }, { "epoch": 0.9156508222204042, "grad_norm": 0.16532239317893982, "learning_rate": 9.5022567554434e-06, "loss": 0.0039, "step": 55960 }, { "epoch": 0.91581444817148, "grad_norm": 0.5656148195266724, "learning_rate": 9.501946170922377e-06, "loss": 0.0034, "step": 55970 }, { "epoch": 0.9159780741225558, "grad_norm": 0.09031599760055542, "learning_rate": 9.501635494610503e-06, "loss": 0.004, "step": 55980 }, { "epoch": 0.9161417000736317, "grad_norm": 0.10471747815608978, "learning_rate": 9.501324726514108e-06, "loss": 0.0033, "step": 55990 }, { "epoch": 0.9163053260247075, "grad_norm": 0.06813917309045792, "learning_rate": 9.50101386663953e-06, "loss": 0.0018, "step": 56000 }, { "epoch": 0.9164689519757834, "grad_norm": 0.24866588413715363, "learning_rate": 9.50070291499311e-06, "loss": 0.0035, "step": 56010 }, { "epoch": 0.9166325779268591, "grad_norm": 0.2393270581960678, "learning_rate": 9.500391871581182e-06, "loss": 0.0032, "step": 56020 }, { "epoch": 0.916796203877935, "grad_norm": 0.04666808992624283, "learning_rate": 9.500080736410093e-06, "loss": 0.002, "step": 56030 }, { "epoch": 0.9169598298290109, "grad_norm": 0.1210368350148201, "learning_rate": 9.499769509486185e-06, "loss": 0.0049, "step": 56040 }, { "epoch": 0.9171234557800867, "grad_norm": 0.13948290050029755, "learning_rate": 9.499458190815802e-06, "loss": 0.0028, "step": 56050 }, { "epoch": 0.9172870817311626, "grad_norm": 0.0882047712802887, "learning_rate": 9.499146780405294e-06, "loss": 0.0033, "step": 56060 }, { "epoch": 0.9174507076822384, "grad_norm": 0.11440379917621613, "learning_rate": 9.49883527826101e-06, "loss": 0.003, "step": 56070 }, { "epoch": 0.9176143336333142, "grad_norm": 0.0967336893081665, "learning_rate": 9.498523684389301e-06, "loss": 0.0026, "step": 56080 }, { "epoch": 0.9177779595843901, "grad_norm": 0.0886610671877861, "learning_rate": 9.49821199879652e-06, "loss": 0.0029, "step": 56090 }, { "epoch": 0.9179415855354659, "grad_norm": 0.07345239073038101, "learning_rate": 9.497900221489022e-06, "loss": 0.0021, "step": 56100 }, { "epoch": 0.9181052114865418, "grad_norm": 0.033308129757642746, "learning_rate": 9.497588352473163e-06, "loss": 0.0041, "step": 56110 }, { "epoch": 0.9182688374376176, "grad_norm": 0.11180134862661362, "learning_rate": 9.497276391755302e-06, "loss": 0.0019, "step": 56120 }, { "epoch": 0.9184324633886934, "grad_norm": 0.05917154252529144, "learning_rate": 9.496964339341801e-06, "loss": 0.0037, "step": 56130 }, { "epoch": 0.9185960893397693, "grad_norm": 0.06316974759101868, "learning_rate": 9.496652195239023e-06, "loss": 0.0021, "step": 56140 }, { "epoch": 0.9187597152908451, "grad_norm": 0.03767119348049164, "learning_rate": 9.49633995945333e-06, "loss": 0.0014, "step": 56150 }, { "epoch": 0.918923341241921, "grad_norm": 0.11493214219808578, "learning_rate": 9.496027631991086e-06, "loss": 0.002, "step": 56160 }, { "epoch": 0.9190869671929968, "grad_norm": 0.050223298370838165, "learning_rate": 9.495715212858667e-06, "loss": 0.0019, "step": 56170 }, { "epoch": 0.9192505931440726, "grad_norm": 0.0626193955540657, "learning_rate": 9.495402702062434e-06, "loss": 0.0034, "step": 56180 }, { "epoch": 0.9194142190951485, "grad_norm": 0.03982340171933174, "learning_rate": 9.495090099608768e-06, "loss": 0.0016, "step": 56190 }, { "epoch": 0.9195778450462243, "grad_norm": 0.03274843841791153, "learning_rate": 9.494777405504035e-06, "loss": 0.002, "step": 56200 }, { "epoch": 0.9197414709973002, "grad_norm": 0.018955720588564873, "learning_rate": 9.494464619754612e-06, "loss": 0.0026, "step": 56210 }, { "epoch": 0.919905096948376, "grad_norm": 0.18404324352741241, "learning_rate": 9.494151742366882e-06, "loss": 0.004, "step": 56220 }, { "epoch": 0.9200687228994519, "grad_norm": 0.2009442001581192, "learning_rate": 9.493838773347218e-06, "loss": 0.0034, "step": 56230 }, { "epoch": 0.9202323488505277, "grad_norm": 0.07173808664083481, "learning_rate": 9.493525712702e-06, "loss": 0.0021, "step": 56240 }, { "epoch": 0.9203959748016035, "grad_norm": 0.0769689753651619, "learning_rate": 9.493212560437619e-06, "loss": 0.0021, "step": 56250 }, { "epoch": 0.9205596007526794, "grad_norm": 0.050492387264966965, "learning_rate": 9.492899316560454e-06, "loss": 0.0018, "step": 56260 }, { "epoch": 0.9207232267037552, "grad_norm": 0.2268935889005661, "learning_rate": 9.492585981076892e-06, "loss": 0.0033, "step": 56270 }, { "epoch": 0.9208868526548311, "grad_norm": 0.046170834451913834, "learning_rate": 9.492272553993323e-06, "loss": 0.0022, "step": 56280 }, { "epoch": 0.9210504786059069, "grad_norm": 0.03678552433848381, "learning_rate": 9.491959035316137e-06, "loss": 0.0021, "step": 56290 }, { "epoch": 0.9212141045569827, "grad_norm": 0.11697898060083389, "learning_rate": 9.491645425051728e-06, "loss": 0.0024, "step": 56300 }, { "epoch": 0.9213777305080586, "grad_norm": 0.056644175201654434, "learning_rate": 9.491331723206486e-06, "loss": 0.0019, "step": 56310 }, { "epoch": 0.9215413564591344, "grad_norm": 0.19982823729515076, "learning_rate": 9.491017929786812e-06, "loss": 0.0034, "step": 56320 }, { "epoch": 0.9217049824102103, "grad_norm": 0.020374177023768425, "learning_rate": 9.490704044799101e-06, "loss": 0.0016, "step": 56330 }, { "epoch": 0.9218686083612861, "grad_norm": 0.13764651119709015, "learning_rate": 9.490390068249755e-06, "loss": 0.0026, "step": 56340 }, { "epoch": 0.9220322343123619, "grad_norm": 0.20850719511508942, "learning_rate": 9.490076000145175e-06, "loss": 0.0022, "step": 56350 }, { "epoch": 0.9221958602634378, "grad_norm": 0.12803031504154205, "learning_rate": 9.489761840491761e-06, "loss": 0.0027, "step": 56360 }, { "epoch": 0.9223594862145136, "grad_norm": 0.052389830350875854, "learning_rate": 9.489447589295924e-06, "loss": 0.0036, "step": 56370 }, { "epoch": 0.9225231121655895, "grad_norm": 0.0742652490735054, "learning_rate": 9.489133246564068e-06, "loss": 0.0021, "step": 56380 }, { "epoch": 0.9226867381166654, "grad_norm": 0.10430403798818588, "learning_rate": 9.488818812302603e-06, "loss": 0.003, "step": 56390 }, { "epoch": 0.9228503640677411, "grad_norm": 0.07706351578235626, "learning_rate": 9.488504286517939e-06, "loss": 0.0024, "step": 56400 }, { "epoch": 0.923013990018817, "grad_norm": 0.03790932893753052, "learning_rate": 9.488189669216491e-06, "loss": 0.0016, "step": 56410 }, { "epoch": 0.9231776159698928, "grad_norm": 0.04502084106206894, "learning_rate": 9.487874960404673e-06, "loss": 0.002, "step": 56420 }, { "epoch": 0.9233412419209687, "grad_norm": 0.14208202064037323, "learning_rate": 9.4875601600889e-06, "loss": 0.0013, "step": 56430 }, { "epoch": 0.9235048678720446, "grad_norm": 0.14498473703861237, "learning_rate": 9.487245268275593e-06, "loss": 0.0032, "step": 56440 }, { "epoch": 0.9236684938231203, "grad_norm": 0.09127403795719147, "learning_rate": 9.486930284971171e-06, "loss": 0.0019, "step": 56450 }, { "epoch": 0.9238321197741962, "grad_norm": 0.15553948283195496, "learning_rate": 9.486615210182057e-06, "loss": 0.0019, "step": 56460 }, { "epoch": 0.923995745725272, "grad_norm": 0.02406824193894863, "learning_rate": 9.486300043914673e-06, "loss": 0.0022, "step": 56470 }, { "epoch": 0.9241593716763479, "grad_norm": 0.055591318756341934, "learning_rate": 9.485984786175448e-06, "loss": 0.0031, "step": 56480 }, { "epoch": 0.9243229976274238, "grad_norm": 0.09256725758314133, "learning_rate": 9.48566943697081e-06, "loss": 0.0032, "step": 56490 }, { "epoch": 0.9244866235784995, "grad_norm": 0.020796146243810654, "learning_rate": 9.485353996307185e-06, "loss": 0.0024, "step": 56500 }, { "epoch": 0.9246502495295754, "grad_norm": 0.00531911151483655, "learning_rate": 9.485038464191005e-06, "loss": 0.0018, "step": 56510 }, { "epoch": 0.9248138754806512, "grad_norm": 0.03892528638243675, "learning_rate": 9.484722840628708e-06, "loss": 0.0022, "step": 56520 }, { "epoch": 0.9249775014317271, "grad_norm": 0.1351790577173233, "learning_rate": 9.484407125626725e-06, "loss": 0.0026, "step": 56530 }, { "epoch": 0.925141127382803, "grad_norm": 0.2015703320503235, "learning_rate": 9.484091319191495e-06, "loss": 0.0023, "step": 56540 }, { "epoch": 0.9253047533338787, "grad_norm": 0.013676916249096394, "learning_rate": 9.483775421329457e-06, "loss": 0.0027, "step": 56550 }, { "epoch": 0.9254683792849546, "grad_norm": 0.030954042449593544, "learning_rate": 9.48345943204705e-06, "loss": 0.0018, "step": 56560 }, { "epoch": 0.9256320052360304, "grad_norm": 0.15184125304222107, "learning_rate": 9.483143351350722e-06, "loss": 0.0026, "step": 56570 }, { "epoch": 0.9257956311871063, "grad_norm": 0.07632570713758469, "learning_rate": 9.482827179246909e-06, "loss": 0.0026, "step": 56580 }, { "epoch": 0.9259592571381822, "grad_norm": 0.0597643218934536, "learning_rate": 9.482510915742065e-06, "loss": 0.0025, "step": 56590 }, { "epoch": 0.9261228830892579, "grad_norm": 0.17775173485279083, "learning_rate": 9.482194560842635e-06, "loss": 0.0039, "step": 56600 }, { "epoch": 0.9262865090403338, "grad_norm": 0.07817158848047256, "learning_rate": 9.48187811455507e-06, "loss": 0.0027, "step": 56610 }, { "epoch": 0.9264501349914096, "grad_norm": 0.05574214458465576, "learning_rate": 9.481561576885821e-06, "loss": 0.0025, "step": 56620 }, { "epoch": 0.9266137609424855, "grad_norm": 0.05494425818324089, "learning_rate": 9.481244947841344e-06, "loss": 0.0018, "step": 56630 }, { "epoch": 0.9267773868935614, "grad_norm": 0.017942851409316063, "learning_rate": 9.480928227428095e-06, "loss": 0.0029, "step": 56640 }, { "epoch": 0.9269410128446371, "grad_norm": 0.12084273248910904, "learning_rate": 9.480611415652527e-06, "loss": 0.0035, "step": 56650 }, { "epoch": 0.927104638795713, "grad_norm": 0.11238863319158554, "learning_rate": 9.480294512521105e-06, "loss": 0.0019, "step": 56660 }, { "epoch": 0.9272682647467888, "grad_norm": 0.048400621861219406, "learning_rate": 9.479977518040288e-06, "loss": 0.0018, "step": 56670 }, { "epoch": 0.9274318906978647, "grad_norm": 0.0533057264983654, "learning_rate": 9.47966043221654e-06, "loss": 0.0021, "step": 56680 }, { "epoch": 0.9275955166489406, "grad_norm": 0.02890601195394993, "learning_rate": 9.479343255056323e-06, "loss": 0.0018, "step": 56690 }, { "epoch": 0.9277591426000164, "grad_norm": 0.0656982809305191, "learning_rate": 9.479025986566111e-06, "loss": 0.0021, "step": 56700 }, { "epoch": 0.9279227685510922, "grad_norm": 0.10621701925992966, "learning_rate": 9.478708626752364e-06, "loss": 0.0027, "step": 56710 }, { "epoch": 0.928086394502168, "grad_norm": 0.10634395480155945, "learning_rate": 9.47839117562156e-06, "loss": 0.0045, "step": 56720 }, { "epoch": 0.9282500204532439, "grad_norm": 0.07654057443141937, "learning_rate": 9.478073633180167e-06, "loss": 0.0039, "step": 56730 }, { "epoch": 0.9284136464043197, "grad_norm": 0.10819920897483826, "learning_rate": 9.477755999434662e-06, "loss": 0.0021, "step": 56740 }, { "epoch": 0.9285772723553956, "grad_norm": 0.17079044878482819, "learning_rate": 9.477438274391521e-06, "loss": 0.0021, "step": 56750 }, { "epoch": 0.9287408983064714, "grad_norm": 0.04137624055147171, "learning_rate": 9.477120458057221e-06, "loss": 0.0029, "step": 56760 }, { "epoch": 0.9289045242575472, "grad_norm": 0.11626918613910675, "learning_rate": 9.476802550438241e-06, "loss": 0.0032, "step": 56770 }, { "epoch": 0.9290681502086231, "grad_norm": 0.056360770016908646, "learning_rate": 9.476484551541065e-06, "loss": 0.0018, "step": 56780 }, { "epoch": 0.9292317761596989, "grad_norm": 0.030419673770666122, "learning_rate": 9.476166461372178e-06, "loss": 0.0028, "step": 56790 }, { "epoch": 0.9293954021107748, "grad_norm": 0.14146853983402252, "learning_rate": 9.475848279938063e-06, "loss": 0.0027, "step": 56800 }, { "epoch": 0.9295590280618506, "grad_norm": 0.1600349247455597, "learning_rate": 9.475530007245206e-06, "loss": 0.0026, "step": 56810 }, { "epoch": 0.9297226540129264, "grad_norm": 0.1058850884437561, "learning_rate": 9.475211643300099e-06, "loss": 0.002, "step": 56820 }, { "epoch": 0.9298862799640023, "grad_norm": 0.09164418280124664, "learning_rate": 9.474893188109235e-06, "loss": 0.002, "step": 56830 }, { "epoch": 0.9300499059150781, "grad_norm": 0.046092767268419266, "learning_rate": 9.474574641679102e-06, "loss": 0.0023, "step": 56840 }, { "epoch": 0.930213531866154, "grad_norm": 0.07372233271598816, "learning_rate": 9.4742560040162e-06, "loss": 0.0023, "step": 56850 }, { "epoch": 0.9303771578172298, "grad_norm": 0.09452763199806213, "learning_rate": 9.473937275127022e-06, "loss": 0.0032, "step": 56860 }, { "epoch": 0.9305407837683056, "grad_norm": 0.05264393612742424, "learning_rate": 9.473618455018067e-06, "loss": 0.0013, "step": 56870 }, { "epoch": 0.9307044097193815, "grad_norm": 0.08179201185703278, "learning_rate": 9.47329954369584e-06, "loss": 0.0017, "step": 56880 }, { "epoch": 0.9308680356704573, "grad_norm": 0.03889448940753937, "learning_rate": 9.472980541166837e-06, "loss": 0.0027, "step": 56890 }, { "epoch": 0.9310316616215332, "grad_norm": 0.014267885126173496, "learning_rate": 9.472661447437563e-06, "loss": 0.0008, "step": 56900 }, { "epoch": 0.931195287572609, "grad_norm": 0.09362617880105972, "learning_rate": 9.472342262514529e-06, "loss": 0.0034, "step": 56910 }, { "epoch": 0.9313589135236848, "grad_norm": 0.05370856076478958, "learning_rate": 9.472022986404238e-06, "loss": 0.003, "step": 56920 }, { "epoch": 0.9315225394747607, "grad_norm": 0.10939586162567139, "learning_rate": 9.471703619113203e-06, "loss": 0.0018, "step": 56930 }, { "epoch": 0.9316861654258365, "grad_norm": 0.28356489539146423, "learning_rate": 9.471384160647933e-06, "loss": 0.0031, "step": 56940 }, { "epoch": 0.9318497913769124, "grad_norm": 0.052440207451581955, "learning_rate": 9.471064611014945e-06, "loss": 0.0022, "step": 56950 }, { "epoch": 0.9320134173279883, "grad_norm": 0.04802414029836655, "learning_rate": 9.47074497022075e-06, "loss": 0.0023, "step": 56960 }, { "epoch": 0.932177043279064, "grad_norm": 0.06371615082025528, "learning_rate": 9.470425238271867e-06, "loss": 0.0032, "step": 56970 }, { "epoch": 0.9323406692301399, "grad_norm": 0.053812719881534576, "learning_rate": 9.470105415174817e-06, "loss": 0.0013, "step": 56980 }, { "epoch": 0.9325042951812157, "grad_norm": 0.2899244427680969, "learning_rate": 9.469785500936118e-06, "loss": 0.0033, "step": 56990 }, { "epoch": 0.9326679211322916, "grad_norm": 0.03364367410540581, "learning_rate": 9.469465495562293e-06, "loss": 0.0027, "step": 57000 }, { "epoch": 0.9328315470833675, "grad_norm": 0.017726397141814232, "learning_rate": 9.46914539905987e-06, "loss": 0.0033, "step": 57010 }, { "epoch": 0.9329951730344432, "grad_norm": 0.053840745240449905, "learning_rate": 9.468825211435373e-06, "loss": 0.0021, "step": 57020 }, { "epoch": 0.9331587989855191, "grad_norm": 0.148805633187294, "learning_rate": 9.468504932695329e-06, "loss": 0.0023, "step": 57030 }, { "epoch": 0.9333224249365949, "grad_norm": 0.023021452128887177, "learning_rate": 9.468184562846271e-06, "loss": 0.0016, "step": 57040 }, { "epoch": 0.9334860508876708, "grad_norm": 0.07152780890464783, "learning_rate": 9.467864101894729e-06, "loss": 0.0039, "step": 57050 }, { "epoch": 0.9336496768387467, "grad_norm": 0.10355637967586517, "learning_rate": 9.467543549847238e-06, "loss": 0.0039, "step": 57060 }, { "epoch": 0.9338133027898224, "grad_norm": 0.14437618851661682, "learning_rate": 9.467222906710333e-06, "loss": 0.0023, "step": 57070 }, { "epoch": 0.9339769287408983, "grad_norm": 0.0063718752935528755, "learning_rate": 9.466902172490555e-06, "loss": 0.0026, "step": 57080 }, { "epoch": 0.9341405546919741, "grad_norm": 0.008033337071537971, "learning_rate": 9.466581347194438e-06, "loss": 0.0028, "step": 57090 }, { "epoch": 0.93430418064305, "grad_norm": 0.043063778430223465, "learning_rate": 9.466260430828526e-06, "loss": 0.0021, "step": 57100 }, { "epoch": 0.9344678065941259, "grad_norm": 0.1288606822490692, "learning_rate": 9.465939423399363e-06, "loss": 0.0034, "step": 57110 }, { "epoch": 0.9346314325452016, "grad_norm": 0.21053865551948547, "learning_rate": 9.465618324913494e-06, "loss": 0.0033, "step": 57120 }, { "epoch": 0.9347950584962775, "grad_norm": 0.08964724093675613, "learning_rate": 9.465297135377463e-06, "loss": 0.0024, "step": 57130 }, { "epoch": 0.9349586844473533, "grad_norm": 0.0259945597499609, "learning_rate": 9.464975854797824e-06, "loss": 0.0019, "step": 57140 }, { "epoch": 0.9351223103984292, "grad_norm": 0.05965486168861389, "learning_rate": 9.464654483181122e-06, "loss": 0.0023, "step": 57150 }, { "epoch": 0.9352859363495051, "grad_norm": 0.16191570460796356, "learning_rate": 9.464333020533914e-06, "loss": 0.0034, "step": 57160 }, { "epoch": 0.9354495623005809, "grad_norm": 0.0996428057551384, "learning_rate": 9.46401146686275e-06, "loss": 0.0024, "step": 57170 }, { "epoch": 0.9356131882516567, "grad_norm": 0.1163264736533165, "learning_rate": 9.463689822174191e-06, "loss": 0.0018, "step": 57180 }, { "epoch": 0.9357768142027325, "grad_norm": 0.11327008157968521, "learning_rate": 9.463368086474792e-06, "loss": 0.0018, "step": 57190 }, { "epoch": 0.9359404401538084, "grad_norm": 0.05948421731591225, "learning_rate": 9.463046259771113e-06, "loss": 0.0023, "step": 57200 }, { "epoch": 0.9361040661048843, "grad_norm": 0.0193459652364254, "learning_rate": 9.462724342069718e-06, "loss": 0.0026, "step": 57210 }, { "epoch": 0.93626769205596, "grad_norm": 0.11364615708589554, "learning_rate": 9.462402333377169e-06, "loss": 0.0018, "step": 57220 }, { "epoch": 0.9364313180070359, "grad_norm": 0.06373066455125809, "learning_rate": 9.462080233700031e-06, "loss": 0.0018, "step": 57230 }, { "epoch": 0.9365949439581117, "grad_norm": 0.046993501484394073, "learning_rate": 9.461758043044871e-06, "loss": 0.0023, "step": 57240 }, { "epoch": 0.9367585699091876, "grad_norm": 0.060208383947610855, "learning_rate": 9.46143576141826e-06, "loss": 0.0025, "step": 57250 }, { "epoch": 0.9369221958602635, "grad_norm": 0.10607155412435532, "learning_rate": 9.461113388826768e-06, "loss": 0.0032, "step": 57260 }, { "epoch": 0.9370858218113393, "grad_norm": 0.14915424585342407, "learning_rate": 9.460790925276969e-06, "loss": 0.0037, "step": 57270 }, { "epoch": 0.9372494477624151, "grad_norm": 0.22396357357501984, "learning_rate": 9.460468370775435e-06, "loss": 0.0023, "step": 57280 }, { "epoch": 0.9374130737134909, "grad_norm": 0.14287075400352478, "learning_rate": 9.460145725328744e-06, "loss": 0.0018, "step": 57290 }, { "epoch": 0.9375766996645668, "grad_norm": 0.1826111227273941, "learning_rate": 9.459822988943477e-06, "loss": 0.0022, "step": 57300 }, { "epoch": 0.9377403256156427, "grad_norm": 0.06133579835295677, "learning_rate": 9.459500161626212e-06, "loss": 0.0029, "step": 57310 }, { "epoch": 0.9379039515667185, "grad_norm": 0.08047100901603699, "learning_rate": 9.459177243383531e-06, "loss": 0.0024, "step": 57320 }, { "epoch": 0.9380675775177943, "grad_norm": 0.07014890760183334, "learning_rate": 9.458854234222017e-06, "loss": 0.0014, "step": 57330 }, { "epoch": 0.9382312034688701, "grad_norm": 0.1185605600476265, "learning_rate": 9.45853113414826e-06, "loss": 0.0038, "step": 57340 }, { "epoch": 0.938394829419946, "grad_norm": 0.09670939296483994, "learning_rate": 9.458207943168842e-06, "loss": 0.0023, "step": 57350 }, { "epoch": 0.9385584553710219, "grad_norm": 0.02559746988117695, "learning_rate": 9.457884661290358e-06, "loss": 0.0025, "step": 57360 }, { "epoch": 0.9387220813220977, "grad_norm": 0.17444205284118652, "learning_rate": 9.457561288519394e-06, "loss": 0.0022, "step": 57370 }, { "epoch": 0.9388857072731736, "grad_norm": 0.10861969739198685, "learning_rate": 9.457237824862549e-06, "loss": 0.0027, "step": 57380 }, { "epoch": 0.9390493332242493, "grad_norm": 0.009416023269295692, "learning_rate": 9.456914270326416e-06, "loss": 0.0022, "step": 57390 }, { "epoch": 0.9392129591753252, "grad_norm": 0.0391194224357605, "learning_rate": 9.45659062491759e-06, "loss": 0.002, "step": 57400 }, { "epoch": 0.9393765851264011, "grad_norm": 0.09816915541887283, "learning_rate": 9.456266888642672e-06, "loss": 0.0022, "step": 57410 }, { "epoch": 0.9395402110774769, "grad_norm": 0.1846799999475479, "learning_rate": 9.455943061508261e-06, "loss": 0.0019, "step": 57420 }, { "epoch": 0.9397038370285528, "grad_norm": 0.23097293078899384, "learning_rate": 9.455619143520963e-06, "loss": 0.0027, "step": 57430 }, { "epoch": 0.9398674629796285, "grad_norm": 0.1275094896554947, "learning_rate": 9.455295134687377e-06, "loss": 0.003, "step": 57440 }, { "epoch": 0.9400310889307044, "grad_norm": 0.0931822657585144, "learning_rate": 9.454971035014113e-06, "loss": 0.003, "step": 57450 }, { "epoch": 0.9401947148817803, "grad_norm": 0.05246676877140999, "learning_rate": 9.45464684450778e-06, "loss": 0.0021, "step": 57460 }, { "epoch": 0.9403583408328561, "grad_norm": 0.13812164962291718, "learning_rate": 9.454322563174985e-06, "loss": 0.0056, "step": 57470 }, { "epoch": 0.940521966783932, "grad_norm": 0.2776726484298706, "learning_rate": 9.453998191022343e-06, "loss": 0.0044, "step": 57480 }, { "epoch": 0.9406855927350077, "grad_norm": 0.30362585186958313, "learning_rate": 9.453673728056463e-06, "loss": 0.0018, "step": 57490 }, { "epoch": 0.9408492186860836, "grad_norm": 0.07336888462305069, "learning_rate": 9.453349174283965e-06, "loss": 0.0025, "step": 57500 }, { "epoch": 0.9410128446371595, "grad_norm": 0.11749880760908127, "learning_rate": 9.453024529711466e-06, "loss": 0.0024, "step": 57510 }, { "epoch": 0.9411764705882353, "grad_norm": 0.14844687283039093, "learning_rate": 9.452699794345583e-06, "loss": 0.0024, "step": 57520 }, { "epoch": 0.9413400965393112, "grad_norm": 0.08617258071899414, "learning_rate": 9.452374968192936e-06, "loss": 0.0012, "step": 57530 }, { "epoch": 0.9415037224903869, "grad_norm": 0.111676886677742, "learning_rate": 9.452050051260152e-06, "loss": 0.0047, "step": 57540 }, { "epoch": 0.9416673484414628, "grad_norm": 0.042387016117572784, "learning_rate": 9.451725043553854e-06, "loss": 0.0027, "step": 57550 }, { "epoch": 0.9418309743925387, "grad_norm": 0.04397552087903023, "learning_rate": 9.451399945080668e-06, "loss": 0.0027, "step": 57560 }, { "epoch": 0.9419946003436145, "grad_norm": 0.141043558716774, "learning_rate": 9.451074755847224e-06, "loss": 0.0034, "step": 57570 }, { "epoch": 0.9421582262946904, "grad_norm": 0.07329373806715012, "learning_rate": 9.45074947586015e-06, "loss": 0.0029, "step": 57580 }, { "epoch": 0.9423218522457661, "grad_norm": 0.021321795880794525, "learning_rate": 9.45042410512608e-06, "loss": 0.002, "step": 57590 }, { "epoch": 0.942485478196842, "grad_norm": 0.05603791028261185, "learning_rate": 9.450098643651647e-06, "loss": 0.0024, "step": 57600 }, { "epoch": 0.9426491041479178, "grad_norm": 1.3001214265823364, "learning_rate": 9.449773091443487e-06, "loss": 0.0101, "step": 57610 }, { "epoch": 0.9428127300989937, "grad_norm": 0.23826691508293152, "learning_rate": 9.449447448508238e-06, "loss": 0.0029, "step": 57620 }, { "epoch": 0.9429763560500696, "grad_norm": 0.09418904036283493, "learning_rate": 9.449121714852541e-06, "loss": 0.0033, "step": 57630 }, { "epoch": 0.9431399820011453, "grad_norm": 0.010086756199598312, "learning_rate": 9.448795890483035e-06, "loss": 0.0022, "step": 57640 }, { "epoch": 0.9433036079522212, "grad_norm": 0.13322779536247253, "learning_rate": 9.448469975406365e-06, "loss": 0.0024, "step": 57650 }, { "epoch": 0.943467233903297, "grad_norm": 0.08473622053861618, "learning_rate": 9.448143969629175e-06, "loss": 0.0026, "step": 57660 }, { "epoch": 0.9436308598543729, "grad_norm": 0.1661970168352127, "learning_rate": 9.447817873158111e-06, "loss": 0.0021, "step": 57670 }, { "epoch": 0.9437944858054488, "grad_norm": 0.07182981073856354, "learning_rate": 9.447491685999825e-06, "loss": 0.0109, "step": 57680 }, { "epoch": 0.9439581117565246, "grad_norm": 0.06154948100447655, "learning_rate": 9.447165408160967e-06, "loss": 0.0022, "step": 57690 }, { "epoch": 0.9441217377076004, "grad_norm": 0.06604491919279099, "learning_rate": 9.446839039648187e-06, "loss": 0.002, "step": 57700 }, { "epoch": 0.9442853636586762, "grad_norm": 0.2119409590959549, "learning_rate": 9.446512580468141e-06, "loss": 0.0036, "step": 57710 }, { "epoch": 0.9444489896097521, "grad_norm": 0.048993684351444244, "learning_rate": 9.446186030627486e-06, "loss": 0.0042, "step": 57720 }, { "epoch": 0.944612615560828, "grad_norm": 0.5547046065330505, "learning_rate": 9.445859390132877e-06, "loss": 0.0019, "step": 57730 }, { "epoch": 0.9447762415119038, "grad_norm": 0.05244401842355728, "learning_rate": 9.44553265899098e-06, "loss": 0.0022, "step": 57740 }, { "epoch": 0.9449398674629796, "grad_norm": 0.11578496545553207, "learning_rate": 9.445205837208449e-06, "loss": 0.0026, "step": 57750 }, { "epoch": 0.9451034934140554, "grad_norm": 0.11608607321977615, "learning_rate": 9.444878924791952e-06, "loss": 0.0038, "step": 57760 }, { "epoch": 0.9452671193651313, "grad_norm": 0.22504957020282745, "learning_rate": 9.444551921748155e-06, "loss": 0.0037, "step": 57770 }, { "epoch": 0.9454307453162072, "grad_norm": 0.0909281075000763, "learning_rate": 9.444224828083725e-06, "loss": 0.0024, "step": 57780 }, { "epoch": 0.945594371267283, "grad_norm": 0.032253168523311615, "learning_rate": 9.443897643805328e-06, "loss": 0.0019, "step": 57790 }, { "epoch": 0.9457579972183588, "grad_norm": 0.036347564309835434, "learning_rate": 9.44357036891964e-06, "loss": 0.0016, "step": 57800 }, { "epoch": 0.9459216231694346, "grad_norm": 0.19950465857982635, "learning_rate": 9.443243003433332e-06, "loss": 0.0038, "step": 57810 }, { "epoch": 0.9460852491205105, "grad_norm": 0.15231943130493164, "learning_rate": 9.442915547353075e-06, "loss": 0.0017, "step": 57820 }, { "epoch": 0.9462488750715864, "grad_norm": 0.07416021078824997, "learning_rate": 9.44258800068555e-06, "loss": 0.0022, "step": 57830 }, { "epoch": 0.9464125010226622, "grad_norm": 0.1885964721441269, "learning_rate": 9.442260363437435e-06, "loss": 0.0025, "step": 57840 }, { "epoch": 0.946576126973738, "grad_norm": 0.033338867127895355, "learning_rate": 9.441932635615407e-06, "loss": 0.0024, "step": 57850 }, { "epoch": 0.9467397529248138, "grad_norm": 0.05057501420378685, "learning_rate": 9.441604817226151e-06, "loss": 0.0026, "step": 57860 }, { "epoch": 0.9469033788758897, "grad_norm": 0.21829725801944733, "learning_rate": 9.441276908276352e-06, "loss": 0.0026, "step": 57870 }, { "epoch": 0.9470670048269656, "grad_norm": 0.031973764300346375, "learning_rate": 9.440948908772692e-06, "loss": 0.0042, "step": 57880 }, { "epoch": 0.9472306307780414, "grad_norm": 0.047340091317892075, "learning_rate": 9.44062081872186e-06, "loss": 0.002, "step": 57890 }, { "epoch": 0.9473942567291173, "grad_norm": 0.06001614034175873, "learning_rate": 9.440292638130548e-06, "loss": 0.0014, "step": 57900 }, { "epoch": 0.947557882680193, "grad_norm": 0.03687538579106331, "learning_rate": 9.439964367005445e-06, "loss": 0.0041, "step": 57910 }, { "epoch": 0.9477215086312689, "grad_norm": 0.11121448874473572, "learning_rate": 9.439636005353244e-06, "loss": 0.0027, "step": 57920 }, { "epoch": 0.9478851345823448, "grad_norm": 0.07019926607608795, "learning_rate": 9.439307553180642e-06, "loss": 0.0026, "step": 57930 }, { "epoch": 0.9480487605334206, "grad_norm": 0.008688589558005333, "learning_rate": 9.438979010494333e-06, "loss": 0.0015, "step": 57940 }, { "epoch": 0.9482123864844965, "grad_norm": 0.13829414546489716, "learning_rate": 9.438650377301018e-06, "loss": 0.0016, "step": 57950 }, { "epoch": 0.9483760124355722, "grad_norm": 0.03137505427002907, "learning_rate": 9.438321653607397e-06, "loss": 0.003, "step": 57960 }, { "epoch": 0.9485396383866481, "grad_norm": 0.034697093069553375, "learning_rate": 9.43799283942017e-06, "loss": 0.0014, "step": 57970 }, { "epoch": 0.948703264337724, "grad_norm": 0.07317320257425308, "learning_rate": 9.437663934746045e-06, "loss": 0.0033, "step": 57980 }, { "epoch": 0.9488668902887998, "grad_norm": 0.08062900602817535, "learning_rate": 9.437334939591727e-06, "loss": 0.0018, "step": 57990 }, { "epoch": 0.9490305162398757, "grad_norm": 0.17392274737358093, "learning_rate": 9.437005853963923e-06, "loss": 0.0017, "step": 58000 }, { "epoch": 0.9491941421909514, "grad_norm": 0.14054661989212036, "learning_rate": 9.436676677869343e-06, "loss": 0.0021, "step": 58010 }, { "epoch": 0.9493577681420273, "grad_norm": 0.13518120348453522, "learning_rate": 9.436347411314698e-06, "loss": 0.0028, "step": 58020 }, { "epoch": 0.9495213940931032, "grad_norm": 0.09343806654214859, "learning_rate": 9.436018054306704e-06, "loss": 0.0016, "step": 58030 }, { "epoch": 0.949685020044179, "grad_norm": 0.03262775018811226, "learning_rate": 9.435688606852073e-06, "loss": 0.0022, "step": 58040 }, { "epoch": 0.9498486459952549, "grad_norm": 0.08901920914649963, "learning_rate": 9.435359068957524e-06, "loss": 0.0013, "step": 58050 }, { "epoch": 0.9500122719463306, "grad_norm": 0.09837601333856583, "learning_rate": 9.435029440629776e-06, "loss": 0.0013, "step": 58060 }, { "epoch": 0.9501758978974065, "grad_norm": 0.09787619113922119, "learning_rate": 9.43469972187555e-06, "loss": 0.0015, "step": 58070 }, { "epoch": 0.9503395238484824, "grad_norm": 0.05775364115834236, "learning_rate": 9.434369912701568e-06, "loss": 0.0031, "step": 58080 }, { "epoch": 0.9505031497995582, "grad_norm": 0.19965891540050507, "learning_rate": 9.434040013114554e-06, "loss": 0.0017, "step": 58090 }, { "epoch": 0.9506667757506341, "grad_norm": 0.11653802543878555, "learning_rate": 9.433710023121236e-06, "loss": 0.0047, "step": 58100 }, { "epoch": 0.9508304017017098, "grad_norm": 0.140816330909729, "learning_rate": 9.43337994272834e-06, "loss": 0.0017, "step": 58110 }, { "epoch": 0.9509940276527857, "grad_norm": 0.2073264867067337, "learning_rate": 9.433049771942599e-06, "loss": 0.0022, "step": 58120 }, { "epoch": 0.9511576536038616, "grad_norm": 0.03231818974018097, "learning_rate": 9.432719510770744e-06, "loss": 0.0025, "step": 58130 }, { "epoch": 0.9513212795549374, "grad_norm": 0.04122627153992653, "learning_rate": 9.432389159219507e-06, "loss": 0.0027, "step": 58140 }, { "epoch": 0.9514849055060133, "grad_norm": 0.0512322299182415, "learning_rate": 9.432058717295625e-06, "loss": 0.0023, "step": 58150 }, { "epoch": 0.951648531457089, "grad_norm": 0.028705177828669548, "learning_rate": 9.431728185005834e-06, "loss": 0.0018, "step": 58160 }, { "epoch": 0.9518121574081649, "grad_norm": 0.13111987709999084, "learning_rate": 9.431397562356876e-06, "loss": 0.0017, "step": 58170 }, { "epoch": 0.9519757833592408, "grad_norm": 0.08373081684112549, "learning_rate": 9.431066849355488e-06, "loss": 0.0022, "step": 58180 }, { "epoch": 0.9521394093103166, "grad_norm": 0.07631798833608627, "learning_rate": 9.430736046008417e-06, "loss": 0.0021, "step": 58190 }, { "epoch": 0.9523030352613925, "grad_norm": 0.06360931694507599, "learning_rate": 9.430405152322407e-06, "loss": 0.0021, "step": 58200 }, { "epoch": 0.9524666612124683, "grad_norm": 0.15624891221523285, "learning_rate": 9.430074168304203e-06, "loss": 0.0018, "step": 58210 }, { "epoch": 0.9526302871635441, "grad_norm": 0.04072519391775131, "learning_rate": 9.429743093960555e-06, "loss": 0.002, "step": 58220 }, { "epoch": 0.95279391311462, "grad_norm": 0.05828690901398659, "learning_rate": 9.429411929298212e-06, "loss": 0.0014, "step": 58230 }, { "epoch": 0.9529575390656958, "grad_norm": 0.09745912998914719, "learning_rate": 9.429080674323927e-06, "loss": 0.0022, "step": 58240 }, { "epoch": 0.9531211650167717, "grad_norm": 0.05096180737018585, "learning_rate": 9.428749329044454e-06, "loss": 0.0018, "step": 58250 }, { "epoch": 0.9532847909678475, "grad_norm": 0.07504450529813766, "learning_rate": 9.428417893466549e-06, "loss": 0.0026, "step": 58260 }, { "epoch": 0.9534484169189233, "grad_norm": 0.03636730834841728, "learning_rate": 9.428086367596968e-06, "loss": 0.0027, "step": 58270 }, { "epoch": 0.9536120428699992, "grad_norm": 0.09096529334783554, "learning_rate": 9.427754751442473e-06, "loss": 0.0016, "step": 58280 }, { "epoch": 0.953775668821075, "grad_norm": 0.05219779536128044, "learning_rate": 9.427423045009823e-06, "loss": 0.0015, "step": 58290 }, { "epoch": 0.9539392947721509, "grad_norm": 0.0955628901720047, "learning_rate": 9.427091248305784e-06, "loss": 0.003, "step": 58300 }, { "epoch": 0.9541029207232267, "grad_norm": 0.11374442279338837, "learning_rate": 9.426759361337119e-06, "loss": 0.0019, "step": 58310 }, { "epoch": 0.9542665466743026, "grad_norm": 0.24611367285251617, "learning_rate": 9.426427384110596e-06, "loss": 0.0022, "step": 58320 }, { "epoch": 0.9544301726253784, "grad_norm": 0.0968439057469368, "learning_rate": 9.426095316632983e-06, "loss": 0.0034, "step": 58330 }, { "epoch": 0.9545937985764542, "grad_norm": 0.042107004672288895, "learning_rate": 9.42576315891105e-06, "loss": 0.003, "step": 58340 }, { "epoch": 0.9547574245275301, "grad_norm": 0.09851545840501785, "learning_rate": 9.425430910951571e-06, "loss": 0.0033, "step": 58350 }, { "epoch": 0.9549210504786059, "grad_norm": 0.08076642453670502, "learning_rate": 9.42509857276132e-06, "loss": 0.0039, "step": 58360 }, { "epoch": 0.9550846764296818, "grad_norm": 0.15204784274101257, "learning_rate": 9.424766144347071e-06, "loss": 0.0034, "step": 58370 }, { "epoch": 0.9552483023807576, "grad_norm": 0.042435117065906525, "learning_rate": 9.424433625715602e-06, "loss": 0.0043, "step": 58380 }, { "epoch": 0.9554119283318334, "grad_norm": 0.05067948251962662, "learning_rate": 9.424101016873697e-06, "loss": 0.0028, "step": 58390 }, { "epoch": 0.9555755542829093, "grad_norm": 0.12186034768819809, "learning_rate": 9.423768317828134e-06, "loss": 0.002, "step": 58400 }, { "epoch": 0.9557391802339851, "grad_norm": 0.130891814827919, "learning_rate": 9.423435528585697e-06, "loss": 0.0011, "step": 58410 }, { "epoch": 0.955902806185061, "grad_norm": 0.02567070908844471, "learning_rate": 9.423102649153173e-06, "loss": 0.0033, "step": 58420 }, { "epoch": 0.9560664321361368, "grad_norm": 0.07939761877059937, "learning_rate": 9.422769679537346e-06, "loss": 0.0024, "step": 58430 }, { "epoch": 0.9562300580872126, "grad_norm": 0.18333324790000916, "learning_rate": 9.422436619745008e-06, "loss": 0.0021, "step": 58440 }, { "epoch": 0.9563936840382885, "grad_norm": 0.028996044769883156, "learning_rate": 9.422103469782949e-06, "loss": 0.0025, "step": 58450 }, { "epoch": 0.9565573099893643, "grad_norm": 0.07702628523111343, "learning_rate": 9.421770229657959e-06, "loss": 0.0018, "step": 58460 }, { "epoch": 0.9567209359404402, "grad_norm": 0.09808099269866943, "learning_rate": 9.421436899376835e-06, "loss": 0.003, "step": 58470 }, { "epoch": 0.9568845618915159, "grad_norm": 0.05474299192428589, "learning_rate": 9.421103478946374e-06, "loss": 0.0027, "step": 58480 }, { "epoch": 0.9570481878425918, "grad_norm": 0.10441876202821732, "learning_rate": 9.420769968373373e-06, "loss": 0.002, "step": 58490 }, { "epoch": 0.9572118137936677, "grad_norm": 0.04635070264339447, "learning_rate": 9.420436367664631e-06, "loss": 0.0028, "step": 58500 }, { "epoch": 0.9573754397447435, "grad_norm": 0.06783477216959, "learning_rate": 9.420102676826952e-06, "loss": 0.0027, "step": 58510 }, { "epoch": 0.9575390656958194, "grad_norm": 0.09749399870634079, "learning_rate": 9.41976889586714e-06, "loss": 0.0019, "step": 58520 }, { "epoch": 0.9577026916468951, "grad_norm": 0.06524252891540527, "learning_rate": 9.419435024791994e-06, "loss": 0.0032, "step": 58530 }, { "epoch": 0.957866317597971, "grad_norm": 0.05683908984065056, "learning_rate": 9.419101063608331e-06, "loss": 0.0017, "step": 58540 }, { "epoch": 0.9580299435490469, "grad_norm": 0.010639239102602005, "learning_rate": 9.418767012322954e-06, "loss": 0.0031, "step": 58550 }, { "epoch": 0.9581935695001227, "grad_norm": 0.29099005460739136, "learning_rate": 9.418432870942675e-06, "loss": 0.0017, "step": 58560 }, { "epoch": 0.9583571954511986, "grad_norm": 0.06396688520908356, "learning_rate": 9.418098639474308e-06, "loss": 0.0027, "step": 58570 }, { "epoch": 0.9585208214022743, "grad_norm": 0.12656669318675995, "learning_rate": 9.417764317924669e-06, "loss": 0.0041, "step": 58580 }, { "epoch": 0.9586844473533502, "grad_norm": 0.02310197241604328, "learning_rate": 9.417429906300571e-06, "loss": 0.0012, "step": 58590 }, { "epoch": 0.9588480733044261, "grad_norm": 0.06866530328989029, "learning_rate": 9.417095404608834e-06, "loss": 0.0016, "step": 58600 }, { "epoch": 0.9590116992555019, "grad_norm": 0.056834157556295395, "learning_rate": 9.416760812856279e-06, "loss": 0.0018, "step": 58610 }, { "epoch": 0.9591753252065778, "grad_norm": 0.07639967650175095, "learning_rate": 9.416426131049727e-06, "loss": 0.0021, "step": 58620 }, { "epoch": 0.9593389511576536, "grad_norm": 0.031442489475011826, "learning_rate": 9.416091359196002e-06, "loss": 0.0022, "step": 58630 }, { "epoch": 0.9595025771087294, "grad_norm": 0.04723643139004707, "learning_rate": 9.41575649730193e-06, "loss": 0.0017, "step": 58640 }, { "epoch": 0.9596662030598053, "grad_norm": 0.2162490040063858, "learning_rate": 9.415421545374339e-06, "loss": 0.0027, "step": 58650 }, { "epoch": 0.9598298290108811, "grad_norm": 0.25380808115005493, "learning_rate": 9.415086503420059e-06, "loss": 0.0049, "step": 58660 }, { "epoch": 0.959993454961957, "grad_norm": 0.042565252631902695, "learning_rate": 9.414751371445919e-06, "loss": 0.0023, "step": 58670 }, { "epoch": 0.9601570809130328, "grad_norm": 0.07301189005374908, "learning_rate": 9.414416149458751e-06, "loss": 0.0031, "step": 58680 }, { "epoch": 0.9603207068641086, "grad_norm": 0.05953185632824898, "learning_rate": 9.414080837465394e-06, "loss": 0.0024, "step": 58690 }, { "epoch": 0.9604843328151845, "grad_norm": 0.07232492417097092, "learning_rate": 9.413745435472683e-06, "loss": 0.0025, "step": 58700 }, { "epoch": 0.9606479587662603, "grad_norm": 0.142743781208992, "learning_rate": 9.413409943487455e-06, "loss": 0.0014, "step": 58710 }, { "epoch": 0.9608115847173362, "grad_norm": 0.02648179419338703, "learning_rate": 9.413074361516554e-06, "loss": 0.0031, "step": 58720 }, { "epoch": 0.960975210668412, "grad_norm": 0.07139645516872406, "learning_rate": 9.412738689566818e-06, "loss": 0.002, "step": 58730 }, { "epoch": 0.9611388366194878, "grad_norm": 0.0976395532488823, "learning_rate": 9.412402927645095e-06, "loss": 0.0025, "step": 58740 }, { "epoch": 0.9613024625705637, "grad_norm": 0.0497332327067852, "learning_rate": 9.412067075758226e-06, "loss": 0.0016, "step": 58750 }, { "epoch": 0.9614660885216395, "grad_norm": 0.07855155318975449, "learning_rate": 9.411731133913063e-06, "loss": 0.0034, "step": 58760 }, { "epoch": 0.9616297144727154, "grad_norm": 0.035570595413446426, "learning_rate": 9.411395102116455e-06, "loss": 0.0024, "step": 58770 }, { "epoch": 0.9617933404237912, "grad_norm": 0.13873648643493652, "learning_rate": 9.411058980375251e-06, "loss": 0.0023, "step": 58780 }, { "epoch": 0.961956966374867, "grad_norm": 0.028694087639451027, "learning_rate": 9.410722768696308e-06, "loss": 0.002, "step": 58790 }, { "epoch": 0.9621205923259429, "grad_norm": 0.08709688484668732, "learning_rate": 9.410386467086479e-06, "loss": 0.0023, "step": 58800 }, { "epoch": 0.9622842182770187, "grad_norm": 0.0496167354285717, "learning_rate": 9.41005007555262e-06, "loss": 0.0015, "step": 58810 }, { "epoch": 0.9624478442280946, "grad_norm": 0.020907754078507423, "learning_rate": 9.409713594101589e-06, "loss": 0.0017, "step": 58820 }, { "epoch": 0.9626114701791704, "grad_norm": 0.7191760540008545, "learning_rate": 9.40937702274025e-06, "loss": 0.003, "step": 58830 }, { "epoch": 0.9627750961302463, "grad_norm": 0.04605472460389137, "learning_rate": 9.409040361475463e-06, "loss": 0.0022, "step": 58840 }, { "epoch": 0.9629387220813221, "grad_norm": 0.0748826190829277, "learning_rate": 9.408703610314091e-06, "loss": 0.0046, "step": 58850 }, { "epoch": 0.9631023480323979, "grad_norm": 0.15941430628299713, "learning_rate": 9.408366769263005e-06, "loss": 0.0029, "step": 58860 }, { "epoch": 0.9632659739834738, "grad_norm": 0.10657966881990433, "learning_rate": 9.408029838329068e-06, "loss": 0.0018, "step": 58870 }, { "epoch": 0.9634295999345496, "grad_norm": 0.01642734743654728, "learning_rate": 9.407692817519151e-06, "loss": 0.0028, "step": 58880 }, { "epoch": 0.9635932258856255, "grad_norm": 0.03644726797938347, "learning_rate": 9.407355706840127e-06, "loss": 0.0021, "step": 58890 }, { "epoch": 0.9637568518367013, "grad_norm": 0.13872921466827393, "learning_rate": 9.407018506298868e-06, "loss": 0.0015, "step": 58900 }, { "epoch": 0.9639204777877771, "grad_norm": 0.08301213383674622, "learning_rate": 9.406681215902251e-06, "loss": 0.0021, "step": 58910 }, { "epoch": 0.964084103738853, "grad_norm": 0.10349325835704803, "learning_rate": 9.40634383565715e-06, "loss": 0.0013, "step": 58920 }, { "epoch": 0.9642477296899288, "grad_norm": 0.014419510960578918, "learning_rate": 9.406006365570447e-06, "loss": 0.0035, "step": 58930 }, { "epoch": 0.9644113556410047, "grad_norm": 0.029120881110429764, "learning_rate": 9.405668805649023e-06, "loss": 0.0022, "step": 58940 }, { "epoch": 0.9645749815920805, "grad_norm": 0.04258144646883011, "learning_rate": 9.405331155899756e-06, "loss": 0.0017, "step": 58950 }, { "epoch": 0.9647386075431563, "grad_norm": 0.10581524670124054, "learning_rate": 9.404993416329533e-06, "loss": 0.0021, "step": 58960 }, { "epoch": 0.9649022334942322, "grad_norm": 0.16358833014965057, "learning_rate": 9.404655586945242e-06, "loss": 0.0041, "step": 58970 }, { "epoch": 0.965065859445308, "grad_norm": 0.12687422335147858, "learning_rate": 9.40431766775377e-06, "loss": 0.0021, "step": 58980 }, { "epoch": 0.9652294853963839, "grad_norm": 0.035959772765636444, "learning_rate": 9.403979658762005e-06, "loss": 0.0024, "step": 58990 }, { "epoch": 0.9653931113474598, "grad_norm": 0.05705726519227028, "learning_rate": 9.40364155997684e-06, "loss": 0.0022, "step": 59000 }, { "epoch": 0.9655567372985355, "grad_norm": 0.18883077800273895, "learning_rate": 9.40330337140517e-06, "loss": 0.0023, "step": 59010 }, { "epoch": 0.9657203632496114, "grad_norm": 0.16986924409866333, "learning_rate": 9.402965093053888e-06, "loss": 0.002, "step": 59020 }, { "epoch": 0.9658839892006872, "grad_norm": 0.06009199097752571, "learning_rate": 9.402626724929895e-06, "loss": 0.0019, "step": 59030 }, { "epoch": 0.9660476151517631, "grad_norm": 0.05990465357899666, "learning_rate": 9.402288267040085e-06, "loss": 0.0036, "step": 59040 }, { "epoch": 0.966211241102839, "grad_norm": 0.1867128163576126, "learning_rate": 9.401949719391362e-06, "loss": 0.0024, "step": 59050 }, { "epoch": 0.9663748670539147, "grad_norm": 0.017606308683753014, "learning_rate": 9.401611081990628e-06, "loss": 0.0017, "step": 59060 }, { "epoch": 0.9665384930049906, "grad_norm": 0.0500749908387661, "learning_rate": 9.401272354844788e-06, "loss": 0.0026, "step": 59070 }, { "epoch": 0.9667021189560664, "grad_norm": 0.06789514422416687, "learning_rate": 9.400933537960745e-06, "loss": 0.0026, "step": 59080 }, { "epoch": 0.9668657449071423, "grad_norm": 0.05841313302516937, "learning_rate": 9.400594631345414e-06, "loss": 0.0018, "step": 59090 }, { "epoch": 0.9670293708582182, "grad_norm": 0.10518983006477356, "learning_rate": 9.400255635005699e-06, "loss": 0.0026, "step": 59100 }, { "epoch": 0.9671929968092939, "grad_norm": 0.05604898929595947, "learning_rate": 9.399916548948514e-06, "loss": 0.0012, "step": 59110 }, { "epoch": 0.9673566227603698, "grad_norm": 0.09373167157173157, "learning_rate": 9.399577373180773e-06, "loss": 0.0037, "step": 59120 }, { "epoch": 0.9675202487114456, "grad_norm": 0.10967706143856049, "learning_rate": 9.39923810770939e-06, "loss": 0.0018, "step": 59130 }, { "epoch": 0.9676838746625215, "grad_norm": 0.20696698129177094, "learning_rate": 9.398898752541284e-06, "loss": 0.0044, "step": 59140 }, { "epoch": 0.9678475006135974, "grad_norm": 0.042170822620391846, "learning_rate": 9.398559307683373e-06, "loss": 0.0027, "step": 59150 }, { "epoch": 0.9680111265646731, "grad_norm": 0.052718352526426315, "learning_rate": 9.39821977314258e-06, "loss": 0.0018, "step": 59160 }, { "epoch": 0.968174752515749, "grad_norm": 0.03554947301745415, "learning_rate": 9.397880148925825e-06, "loss": 0.0018, "step": 59170 }, { "epoch": 0.9683383784668248, "grad_norm": 0.035509251058101654, "learning_rate": 9.397540435040034e-06, "loss": 0.0018, "step": 59180 }, { "epoch": 0.9685020044179007, "grad_norm": 0.02136817015707493, "learning_rate": 9.397200631492134e-06, "loss": 0.0018, "step": 59190 }, { "epoch": 0.9686656303689766, "grad_norm": 0.059554968029260635, "learning_rate": 9.396860738289054e-06, "loss": 0.0029, "step": 59200 }, { "epoch": 0.9688292563200523, "grad_norm": 0.0817946046590805, "learning_rate": 9.396520755437722e-06, "loss": 0.0029, "step": 59210 }, { "epoch": 0.9689928822711282, "grad_norm": 0.1386914849281311, "learning_rate": 9.396180682945071e-06, "loss": 0.0034, "step": 59220 }, { "epoch": 0.969156508222204, "grad_norm": 0.15243680775165558, "learning_rate": 9.395840520818033e-06, "loss": 0.0033, "step": 59230 }, { "epoch": 0.9693201341732799, "grad_norm": 0.07763664424419403, "learning_rate": 9.395500269063546e-06, "loss": 0.002, "step": 59240 }, { "epoch": 0.9694837601243558, "grad_norm": 0.05854769051074982, "learning_rate": 9.395159927688548e-06, "loss": 0.0017, "step": 59250 }, { "epoch": 0.9696473860754315, "grad_norm": 0.30561816692352295, "learning_rate": 9.394819496699977e-06, "loss": 0.0034, "step": 59260 }, { "epoch": 0.9698110120265074, "grad_norm": 0.12019655108451843, "learning_rate": 9.394478976104774e-06, "loss": 0.0027, "step": 59270 }, { "epoch": 0.9699746379775832, "grad_norm": 0.03279072046279907, "learning_rate": 9.394138365909882e-06, "loss": 0.0055, "step": 59280 }, { "epoch": 0.9701382639286591, "grad_norm": 0.10708396136760712, "learning_rate": 9.393797666122246e-06, "loss": 0.002, "step": 59290 }, { "epoch": 0.970301889879735, "grad_norm": 0.09034889936447144, "learning_rate": 9.393456876748812e-06, "loss": 0.0014, "step": 59300 }, { "epoch": 0.9704655158308108, "grad_norm": 0.06554750353097916, "learning_rate": 9.393115997796527e-06, "loss": 0.0024, "step": 59310 }, { "epoch": 0.9706291417818866, "grad_norm": 0.015089423395693302, "learning_rate": 9.392775029272345e-06, "loss": 0.002, "step": 59320 }, { "epoch": 0.9707927677329624, "grad_norm": 0.10854396224021912, "learning_rate": 9.392433971183215e-06, "loss": 0.0044, "step": 59330 }, { "epoch": 0.9709563936840383, "grad_norm": 0.13071149587631226, "learning_rate": 9.392092823536095e-06, "loss": 0.0022, "step": 59340 }, { "epoch": 0.9711200196351142, "grad_norm": 0.07579005509614944, "learning_rate": 9.391751586337934e-06, "loss": 0.0027, "step": 59350 }, { "epoch": 0.97128364558619, "grad_norm": 0.1578858345746994, "learning_rate": 9.391410259595695e-06, "loss": 0.0038, "step": 59360 }, { "epoch": 0.9714472715372658, "grad_norm": 0.05021621659398079, "learning_rate": 9.391068843316336e-06, "loss": 0.0017, "step": 59370 }, { "epoch": 0.9716108974883416, "grad_norm": 0.03856870159506798, "learning_rate": 9.390727337506815e-06, "loss": 0.0024, "step": 59380 }, { "epoch": 0.9717745234394175, "grad_norm": 0.1385786235332489, "learning_rate": 9.3903857421741e-06, "loss": 0.0025, "step": 59390 }, { "epoch": 0.9719381493904933, "grad_norm": 0.041757531464099884, "learning_rate": 9.390044057325154e-06, "loss": 0.0016, "step": 59400 }, { "epoch": 0.9721017753415692, "grad_norm": 0.08493485301733017, "learning_rate": 9.389702282966941e-06, "loss": 0.0033, "step": 59410 }, { "epoch": 0.972265401292645, "grad_norm": 0.0686621367931366, "learning_rate": 9.389360419106435e-06, "loss": 0.003, "step": 59420 }, { "epoch": 0.9724290272437208, "grad_norm": 0.16846078634262085, "learning_rate": 9.3890184657506e-06, "loss": 0.0027, "step": 59430 }, { "epoch": 0.9725926531947967, "grad_norm": 0.10593196749687195, "learning_rate": 9.388676422906413e-06, "loss": 0.002, "step": 59440 }, { "epoch": 0.9727562791458725, "grad_norm": 0.04029037803411484, "learning_rate": 9.388334290580845e-06, "loss": 0.0029, "step": 59450 }, { "epoch": 0.9729199050969484, "grad_norm": 0.017334720119833946, "learning_rate": 9.387992068780876e-06, "loss": 0.0016, "step": 59460 }, { "epoch": 0.9730835310480243, "grad_norm": 0.06020909920334816, "learning_rate": 9.387649757513478e-06, "loss": 0.0028, "step": 59470 }, { "epoch": 0.9732471569991, "grad_norm": 0.15472665429115295, "learning_rate": 9.387307356785634e-06, "loss": 0.0031, "step": 59480 }, { "epoch": 0.9734107829501759, "grad_norm": 0.1773456484079361, "learning_rate": 9.386964866604323e-06, "loss": 0.0032, "step": 59490 }, { "epoch": 0.9735744089012517, "grad_norm": 0.04544470086693764, "learning_rate": 9.386622286976529e-06, "loss": 0.0023, "step": 59500 }, { "epoch": 0.9737380348523276, "grad_norm": 0.10733740031719208, "learning_rate": 9.386279617909237e-06, "loss": 0.0026, "step": 59510 }, { "epoch": 0.9739016608034035, "grad_norm": 0.11613000929355621, "learning_rate": 9.385936859409436e-06, "loss": 0.0024, "step": 59520 }, { "epoch": 0.9740652867544792, "grad_norm": 0.06578903645277023, "learning_rate": 9.385594011484112e-06, "loss": 0.0029, "step": 59530 }, { "epoch": 0.9742289127055551, "grad_norm": 0.08920370787382126, "learning_rate": 9.385251074140256e-06, "loss": 0.0022, "step": 59540 }, { "epoch": 0.9743925386566309, "grad_norm": 0.10824060440063477, "learning_rate": 9.38490804738486e-06, "loss": 0.0029, "step": 59550 }, { "epoch": 0.9745561646077068, "grad_norm": 0.08234572410583496, "learning_rate": 9.384564931224918e-06, "loss": 0.0023, "step": 59560 }, { "epoch": 0.9747197905587827, "grad_norm": 0.14229722321033478, "learning_rate": 9.384221725667425e-06, "loss": 0.0036, "step": 59570 }, { "epoch": 0.9748834165098584, "grad_norm": 0.06527045369148254, "learning_rate": 9.38387843071938e-06, "loss": 0.003, "step": 59580 }, { "epoch": 0.9750470424609343, "grad_norm": 0.06422606855630875, "learning_rate": 9.383535046387785e-06, "loss": 0.0018, "step": 59590 }, { "epoch": 0.9752106684120101, "grad_norm": 0.04868666082620621, "learning_rate": 9.383191572679635e-06, "loss": 0.0012, "step": 59600 }, { "epoch": 0.975374294363086, "grad_norm": 0.597815752029419, "learning_rate": 9.382848009601937e-06, "loss": 0.0037, "step": 59610 }, { "epoch": 0.9755379203141619, "grad_norm": 0.3672791123390198, "learning_rate": 9.382504357161698e-06, "loss": 0.0039, "step": 59620 }, { "epoch": 0.9757015462652376, "grad_norm": 0.2882446348667145, "learning_rate": 9.38216061536592e-06, "loss": 0.0033, "step": 59630 }, { "epoch": 0.9758651722163135, "grad_norm": 0.13354012370109558, "learning_rate": 9.381816784221613e-06, "loss": 0.0015, "step": 59640 }, { "epoch": 0.9760287981673893, "grad_norm": 0.053285934031009674, "learning_rate": 9.381472863735791e-06, "loss": 0.0026, "step": 59650 }, { "epoch": 0.9761924241184652, "grad_norm": 0.06669556349515915, "learning_rate": 9.381128853915462e-06, "loss": 0.0028, "step": 59660 }, { "epoch": 0.9763560500695411, "grad_norm": 0.01125655323266983, "learning_rate": 9.380784754767642e-06, "loss": 0.0017, "step": 59670 }, { "epoch": 0.9765196760206168, "grad_norm": 0.05730072781443596, "learning_rate": 9.380440566299347e-06, "loss": 0.0019, "step": 59680 }, { "epoch": 0.9766833019716927, "grad_norm": 0.07429566234350204, "learning_rate": 9.380096288517595e-06, "loss": 0.0047, "step": 59690 }, { "epoch": 0.9768469279227685, "grad_norm": 0.09967128187417984, "learning_rate": 9.379751921429403e-06, "loss": 0.0032, "step": 59700 }, { "epoch": 0.9770105538738444, "grad_norm": 0.021969735622406006, "learning_rate": 9.379407465041796e-06, "loss": 0.0021, "step": 59710 }, { "epoch": 0.9771741798249203, "grad_norm": 0.0919286385178566, "learning_rate": 9.379062919361794e-06, "loss": 0.0034, "step": 59720 }, { "epoch": 0.977337805775996, "grad_norm": 0.04027273505926132, "learning_rate": 9.378718284396424e-06, "loss": 0.0022, "step": 59730 }, { "epoch": 0.9775014317270719, "grad_norm": 0.14776067435741425, "learning_rate": 9.378373560152711e-06, "loss": 0.0018, "step": 59740 }, { "epoch": 0.9776650576781477, "grad_norm": 0.006730169057846069, "learning_rate": 9.378028746637687e-06, "loss": 0.0029, "step": 59750 }, { "epoch": 0.9778286836292236, "grad_norm": 0.170599102973938, "learning_rate": 9.377683843858378e-06, "loss": 0.003, "step": 59760 }, { "epoch": 0.9779923095802995, "grad_norm": 0.02189823053777218, "learning_rate": 9.37733885182182e-06, "loss": 0.0019, "step": 59770 }, { "epoch": 0.9781559355313753, "grad_norm": 0.15165144205093384, "learning_rate": 9.376993770535046e-06, "loss": 0.0023, "step": 59780 }, { "epoch": 0.9783195614824511, "grad_norm": 0.14875419437885284, "learning_rate": 9.37664860000509e-06, "loss": 0.0031, "step": 59790 }, { "epoch": 0.9784831874335269, "grad_norm": 0.16950346529483795, "learning_rate": 9.376303340238993e-06, "loss": 0.0015, "step": 59800 }, { "epoch": 0.9786468133846028, "grad_norm": 0.04467441886663437, "learning_rate": 9.37595799124379e-06, "loss": 0.0026, "step": 59810 }, { "epoch": 0.9788104393356787, "grad_norm": 0.032187845557928085, "learning_rate": 9.375612553026527e-06, "loss": 0.0024, "step": 59820 }, { "epoch": 0.9789740652867545, "grad_norm": 0.03101181983947754, "learning_rate": 9.375267025594245e-06, "loss": 0.0035, "step": 59830 }, { "epoch": 0.9791376912378303, "grad_norm": 0.0867827832698822, "learning_rate": 9.37492140895399e-06, "loss": 0.0029, "step": 59840 }, { "epoch": 0.9793013171889061, "grad_norm": 0.07778282463550568, "learning_rate": 9.374575703112808e-06, "loss": 0.0026, "step": 59850 }, { "epoch": 0.979464943139982, "grad_norm": 0.22539466619491577, "learning_rate": 9.374229908077748e-06, "loss": 0.0023, "step": 59860 }, { "epoch": 0.9796285690910579, "grad_norm": 0.048014361411333084, "learning_rate": 9.37388402385586e-06, "loss": 0.0019, "step": 59870 }, { "epoch": 0.9797921950421337, "grad_norm": 0.0290620606392622, "learning_rate": 9.373538050454197e-06, "loss": 0.0019, "step": 59880 }, { "epoch": 0.9799558209932095, "grad_norm": 0.15181989967823029, "learning_rate": 9.373191987879812e-06, "loss": 0.0027, "step": 59890 }, { "epoch": 0.9801194469442853, "grad_norm": 0.040208250284194946, "learning_rate": 9.37284583613976e-06, "loss": 0.0023, "step": 59900 }, { "epoch": 0.9802830728953612, "grad_norm": 0.07411646097898483, "learning_rate": 9.372499595241104e-06, "loss": 0.0024, "step": 59910 }, { "epoch": 0.9804466988464371, "grad_norm": 0.03000156581401825, "learning_rate": 9.372153265190898e-06, "loss": 0.002, "step": 59920 }, { "epoch": 0.9806103247975129, "grad_norm": 0.0597178153693676, "learning_rate": 9.371806845996205e-06, "loss": 0.0019, "step": 59930 }, { "epoch": 0.9807739507485888, "grad_norm": 0.007860665209591389, "learning_rate": 9.37146033766409e-06, "loss": 0.0029, "step": 59940 }, { "epoch": 0.9809375766996645, "grad_norm": 0.11553884297609329, "learning_rate": 9.371113740201614e-06, "loss": 0.0018, "step": 59950 }, { "epoch": 0.9811012026507404, "grad_norm": 0.02003282681107521, "learning_rate": 9.370767053615849e-06, "loss": 0.0023, "step": 59960 }, { "epoch": 0.9812648286018163, "grad_norm": 0.15192604064941406, "learning_rate": 9.37042027791386e-06, "loss": 0.0022, "step": 59970 }, { "epoch": 0.9814284545528921, "grad_norm": 0.05869186297059059, "learning_rate": 9.370073413102718e-06, "loss": 0.0022, "step": 59980 }, { "epoch": 0.981592080503968, "grad_norm": 0.12526053190231323, "learning_rate": 9.369726459189497e-06, "loss": 0.0041, "step": 59990 }, { "epoch": 0.9817557064550437, "grad_norm": 0.04395894333720207, "learning_rate": 9.369379416181268e-06, "loss": 0.0036, "step": 60000 }, { "epoch": 0.9819193324061196, "grad_norm": 0.3077646791934967, "learning_rate": 9.369032284085108e-06, "loss": 0.0038, "step": 60010 }, { "epoch": 0.9820829583571955, "grad_norm": 0.0740702897310257, "learning_rate": 9.368685062908098e-06, "loss": 0.0023, "step": 60020 }, { "epoch": 0.9822465843082713, "grad_norm": 0.028068503364920616, "learning_rate": 9.368337752657315e-06, "loss": 0.0023, "step": 60030 }, { "epoch": 0.9824102102593472, "grad_norm": 0.034559279680252075, "learning_rate": 9.367990353339838e-06, "loss": 0.0021, "step": 60040 }, { "epoch": 0.9825738362104229, "grad_norm": 0.13627946376800537, "learning_rate": 9.367642864962754e-06, "loss": 0.0017, "step": 60050 }, { "epoch": 0.9827374621614988, "grad_norm": 0.04937463626265526, "learning_rate": 9.367295287533148e-06, "loss": 0.0027, "step": 60060 }, { "epoch": 0.9829010881125747, "grad_norm": 0.05035670846700668, "learning_rate": 9.366947621058103e-06, "loss": 0.0025, "step": 60070 }, { "epoch": 0.9830647140636505, "grad_norm": 0.038908857852220535, "learning_rate": 9.36659986554471e-06, "loss": 0.0014, "step": 60080 }, { "epoch": 0.9832283400147264, "grad_norm": 0.1760895997285843, "learning_rate": 9.36625202100006e-06, "loss": 0.0041, "step": 60090 }, { "epoch": 0.9833919659658021, "grad_norm": 0.04922424629330635, "learning_rate": 9.365904087431246e-06, "loss": 0.0023, "step": 60100 }, { "epoch": 0.983555591916878, "grad_norm": 0.18811900913715363, "learning_rate": 9.36555606484536e-06, "loss": 0.0026, "step": 60110 }, { "epoch": 0.9837192178679539, "grad_norm": 0.4227871298789978, "learning_rate": 9.365207953249497e-06, "loss": 0.0031, "step": 60120 }, { "epoch": 0.9838828438190297, "grad_norm": 0.054773639887571335, "learning_rate": 9.364859752650758e-06, "loss": 0.0041, "step": 60130 }, { "epoch": 0.9840464697701056, "grad_norm": 0.13385649025440216, "learning_rate": 9.364511463056241e-06, "loss": 0.0042, "step": 60140 }, { "epoch": 0.9842100957211813, "grad_norm": 0.05040515959262848, "learning_rate": 9.364163084473045e-06, "loss": 0.0014, "step": 60150 }, { "epoch": 0.9843737216722572, "grad_norm": 0.15005896985530853, "learning_rate": 9.363814616908278e-06, "loss": 0.0039, "step": 60160 }, { "epoch": 0.9845373476233331, "grad_norm": 0.23939859867095947, "learning_rate": 9.363466060369039e-06, "loss": 0.0036, "step": 60170 }, { "epoch": 0.9847009735744089, "grad_norm": 0.022488852962851524, "learning_rate": 9.363117414862441e-06, "loss": 0.0015, "step": 60180 }, { "epoch": 0.9848645995254848, "grad_norm": 0.04257138818502426, "learning_rate": 9.362768680395588e-06, "loss": 0.0019, "step": 60190 }, { "epoch": 0.9850282254765605, "grad_norm": 0.00437391921877861, "learning_rate": 9.362419856975593e-06, "loss": 0.0015, "step": 60200 }, { "epoch": 0.9851918514276364, "grad_norm": 0.016400137916207314, "learning_rate": 9.362070944609567e-06, "loss": 0.0012, "step": 60210 }, { "epoch": 0.9853554773787123, "grad_norm": 0.1657567173242569, "learning_rate": 9.361721943304625e-06, "loss": 0.0014, "step": 60220 }, { "epoch": 0.9855191033297881, "grad_norm": 0.027916066348552704, "learning_rate": 9.36137285306788e-06, "loss": 0.0027, "step": 60230 }, { "epoch": 0.985682729280864, "grad_norm": 0.09854999929666519, "learning_rate": 9.361023673906454e-06, "loss": 0.0021, "step": 60240 }, { "epoch": 0.9858463552319398, "grad_norm": 0.2760668098926544, "learning_rate": 9.360674405827463e-06, "loss": 0.002, "step": 60250 }, { "epoch": 0.9860099811830156, "grad_norm": 0.06062379851937294, "learning_rate": 9.36032504883803e-06, "loss": 0.0021, "step": 60260 }, { "epoch": 0.9861736071340914, "grad_norm": 0.07205376029014587, "learning_rate": 9.359975602945277e-06, "loss": 0.0019, "step": 60270 }, { "epoch": 0.9863372330851673, "grad_norm": 0.11281125992536545, "learning_rate": 9.359626068156329e-06, "loss": 0.0016, "step": 60280 }, { "epoch": 0.9865008590362432, "grad_norm": 0.053505100309848785, "learning_rate": 9.359276444478313e-06, "loss": 0.0035, "step": 60290 }, { "epoch": 0.986664484987319, "grad_norm": 0.10846849530935287, "learning_rate": 9.35892673191836e-06, "loss": 0.0027, "step": 60300 }, { "epoch": 0.9868281109383948, "grad_norm": 0.012516388669610023, "learning_rate": 9.358576930483595e-06, "loss": 0.003, "step": 60310 }, { "epoch": 0.9869917368894706, "grad_norm": 0.45621994137763977, "learning_rate": 9.358227040181156e-06, "loss": 0.0035, "step": 60320 }, { "epoch": 0.9871553628405465, "grad_norm": 0.04821210354566574, "learning_rate": 9.357877061018174e-06, "loss": 0.0019, "step": 60330 }, { "epoch": 0.9873189887916224, "grad_norm": 0.20245786011219025, "learning_rate": 9.357526993001784e-06, "loss": 0.0023, "step": 60340 }, { "epoch": 0.9874826147426982, "grad_norm": 0.044612377882003784, "learning_rate": 9.357176836139123e-06, "loss": 0.0023, "step": 60350 }, { "epoch": 0.987646240693774, "grad_norm": 0.1946815401315689, "learning_rate": 9.356826590437334e-06, "loss": 0.0061, "step": 60360 }, { "epoch": 0.9878098666448498, "grad_norm": 0.10065727680921555, "learning_rate": 9.356476255903556e-06, "loss": 0.0022, "step": 60370 }, { "epoch": 0.9879734925959257, "grad_norm": 0.0729856863617897, "learning_rate": 9.356125832544932e-06, "loss": 0.0026, "step": 60380 }, { "epoch": 0.9881371185470016, "grad_norm": 0.026070188730955124, "learning_rate": 9.355775320368607e-06, "loss": 0.0016, "step": 60390 }, { "epoch": 0.9883007444980774, "grad_norm": 0.012297333218157291, "learning_rate": 9.355424719381729e-06, "loss": 0.0019, "step": 60400 }, { "epoch": 0.9884643704491533, "grad_norm": 0.0909871906042099, "learning_rate": 9.355074029591442e-06, "loss": 0.0025, "step": 60410 }, { "epoch": 0.988627996400229, "grad_norm": 0.12982827425003052, "learning_rate": 9.354723251004902e-06, "loss": 0.0024, "step": 60420 }, { "epoch": 0.9887916223513049, "grad_norm": 0.1468738317489624, "learning_rate": 9.354372383629258e-06, "loss": 0.0016, "step": 60430 }, { "epoch": 0.9889552483023808, "grad_norm": 0.038689229637384415, "learning_rate": 9.354021427471665e-06, "loss": 0.0026, "step": 60440 }, { "epoch": 0.9891188742534566, "grad_norm": 0.05555734410881996, "learning_rate": 9.353670382539276e-06, "loss": 0.0018, "step": 60450 }, { "epoch": 0.9892825002045325, "grad_norm": 0.036878395825624466, "learning_rate": 9.353319248839251e-06, "loss": 0.0016, "step": 60460 }, { "epoch": 0.9894461261556082, "grad_norm": 0.1170571818947792, "learning_rate": 9.35296802637875e-06, "loss": 0.002, "step": 60470 }, { "epoch": 0.9896097521066841, "grad_norm": 0.030798356980085373, "learning_rate": 9.352616715164933e-06, "loss": 0.0011, "step": 60480 }, { "epoch": 0.98977337805776, "grad_norm": 0.06276801973581314, "learning_rate": 9.352265315204962e-06, "loss": 0.0025, "step": 60490 }, { "epoch": 0.9899370040088358, "grad_norm": 0.07826773822307587, "learning_rate": 9.351913826506003e-06, "loss": 0.0027, "step": 60500 }, { "epoch": 0.9901006299599117, "grad_norm": 0.06611716747283936, "learning_rate": 9.351562249075223e-06, "loss": 0.0018, "step": 60510 }, { "epoch": 0.9902642559109874, "grad_norm": 0.07587224990129471, "learning_rate": 9.351210582919789e-06, "loss": 0.0023, "step": 60520 }, { "epoch": 0.9904278818620633, "grad_norm": 0.023770412430167198, "learning_rate": 9.350858828046872e-06, "loss": 0.0023, "step": 60530 }, { "epoch": 0.9905915078131392, "grad_norm": 0.020139126107096672, "learning_rate": 9.350506984463643e-06, "loss": 0.002, "step": 60540 }, { "epoch": 0.990755133764215, "grad_norm": 0.12416273355484009, "learning_rate": 9.350155052177276e-06, "loss": 0.003, "step": 60550 }, { "epoch": 0.9909187597152909, "grad_norm": 0.04802967607975006, "learning_rate": 9.34980303119495e-06, "loss": 0.002, "step": 60560 }, { "epoch": 0.9910823856663666, "grad_norm": 0.060007866472005844, "learning_rate": 9.349450921523838e-06, "loss": 0.0024, "step": 60570 }, { "epoch": 0.9912460116174425, "grad_norm": 0.07024600356817245, "learning_rate": 9.349098723171119e-06, "loss": 0.0022, "step": 60580 }, { "epoch": 0.9914096375685184, "grad_norm": 0.05218667909502983, "learning_rate": 9.348746436143978e-06, "loss": 0.0026, "step": 60590 }, { "epoch": 0.9915732635195942, "grad_norm": 0.20649020373821259, "learning_rate": 9.348394060449594e-06, "loss": 0.0043, "step": 60600 }, { "epoch": 0.9917368894706701, "grad_norm": 0.06463110446929932, "learning_rate": 9.348041596095155e-06, "loss": 0.0025, "step": 60610 }, { "epoch": 0.9919005154217458, "grad_norm": 0.1100788488984108, "learning_rate": 9.347689043087846e-06, "loss": 0.0023, "step": 60620 }, { "epoch": 0.9920641413728217, "grad_norm": 0.14205293357372284, "learning_rate": 9.347336401434855e-06, "loss": 0.0022, "step": 60630 }, { "epoch": 0.9922277673238976, "grad_norm": 0.07902803272008896, "learning_rate": 9.34698367114337e-06, "loss": 0.0026, "step": 60640 }, { "epoch": 0.9923913932749734, "grad_norm": 0.056057531386613846, "learning_rate": 9.346630852220587e-06, "loss": 0.0017, "step": 60650 }, { "epoch": 0.9925550192260493, "grad_norm": 0.09063239395618439, "learning_rate": 9.346277944673696e-06, "loss": 0.004, "step": 60660 }, { "epoch": 0.992718645177125, "grad_norm": 0.07490795105695724, "learning_rate": 9.345924948509895e-06, "loss": 0.002, "step": 60670 }, { "epoch": 0.9928822711282009, "grad_norm": 0.05313915014266968, "learning_rate": 9.34557186373638e-06, "loss": 0.0017, "step": 60680 }, { "epoch": 0.9930458970792768, "grad_norm": 0.10497010499238968, "learning_rate": 9.345218690360349e-06, "loss": 0.0032, "step": 60690 }, { "epoch": 0.9932095230303526, "grad_norm": 0.15680783987045288, "learning_rate": 9.344865428389007e-06, "loss": 0.0028, "step": 60700 }, { "epoch": 0.9933731489814285, "grad_norm": 0.0603984035551548, "learning_rate": 9.344512077829552e-06, "loss": 0.0026, "step": 60710 }, { "epoch": 0.9935367749325043, "grad_norm": 0.06623845547437668, "learning_rate": 9.344158638689193e-06, "loss": 0.0019, "step": 60720 }, { "epoch": 0.9937004008835801, "grad_norm": 0.06679930537939072, "learning_rate": 9.343805110975133e-06, "loss": 0.0026, "step": 60730 }, { "epoch": 0.993864026834656, "grad_norm": 0.07183536142110825, "learning_rate": 9.34345149469458e-06, "loss": 0.0047, "step": 60740 }, { "epoch": 0.9940276527857318, "grad_norm": 0.04963688179850578, "learning_rate": 9.343097789854744e-06, "loss": 0.0026, "step": 60750 }, { "epoch": 0.9941912787368077, "grad_norm": 0.11362797766923904, "learning_rate": 9.342743996462841e-06, "loss": 0.0032, "step": 60760 }, { "epoch": 0.9943549046878835, "grad_norm": 0.17177900671958923, "learning_rate": 9.34239011452608e-06, "loss": 0.0024, "step": 60770 }, { "epoch": 0.9945185306389593, "grad_norm": 0.13347700238227844, "learning_rate": 9.342036144051678e-06, "loss": 0.0012, "step": 60780 }, { "epoch": 0.9946821565900352, "grad_norm": 0.2341926395893097, "learning_rate": 9.341682085046851e-06, "loss": 0.0047, "step": 60790 }, { "epoch": 0.994845782541111, "grad_norm": 0.08644140511751175, "learning_rate": 9.341327937518818e-06, "loss": 0.0031, "step": 60800 }, { "epoch": 0.9950094084921869, "grad_norm": 0.1372774839401245, "learning_rate": 9.340973701474801e-06, "loss": 0.0033, "step": 60810 }, { "epoch": 0.9951730344432627, "grad_norm": 0.028111103922128677, "learning_rate": 9.340619376922023e-06, "loss": 0.001, "step": 60820 }, { "epoch": 0.9953366603943385, "grad_norm": 0.06416027992963791, "learning_rate": 9.340264963867708e-06, "loss": 0.003, "step": 60830 }, { "epoch": 0.9955002863454144, "grad_norm": 0.08177004754543304, "learning_rate": 9.33991046231908e-06, "loss": 0.0018, "step": 60840 }, { "epoch": 0.9956639122964902, "grad_norm": 0.17180781066417694, "learning_rate": 9.339555872283368e-06, "loss": 0.0023, "step": 60850 }, { "epoch": 0.9958275382475661, "grad_norm": 0.08466265350580215, "learning_rate": 9.339201193767804e-06, "loss": 0.0041, "step": 60860 }, { "epoch": 0.9959911641986419, "grad_norm": 0.16043125092983246, "learning_rate": 9.338846426779618e-06, "loss": 0.0021, "step": 60870 }, { "epoch": 0.9961547901497178, "grad_norm": 0.1249219998717308, "learning_rate": 9.338491571326043e-06, "loss": 0.002, "step": 60880 }, { "epoch": 0.9963184161007936, "grad_norm": 0.01866929419338703, "learning_rate": 9.338136627414316e-06, "loss": 0.0013, "step": 60890 }, { "epoch": 0.9964820420518694, "grad_norm": 0.1304594725370407, "learning_rate": 9.33778159505167e-06, "loss": 0.0018, "step": 60900 }, { "epoch": 0.9966456680029453, "grad_norm": 0.012876631692051888, "learning_rate": 9.337426474245347e-06, "loss": 0.0028, "step": 60910 }, { "epoch": 0.9968092939540211, "grad_norm": 0.009069924242794514, "learning_rate": 9.337071265002589e-06, "loss": 0.0021, "step": 60920 }, { "epoch": 0.996972919905097, "grad_norm": 0.19489061832427979, "learning_rate": 9.336715967330636e-06, "loss": 0.0034, "step": 60930 }, { "epoch": 0.9971365458561728, "grad_norm": 0.1880512237548828, "learning_rate": 9.33636058123673e-06, "loss": 0.0028, "step": 60940 }, { "epoch": 0.9973001718072486, "grad_norm": 0.08962858468294144, "learning_rate": 9.336005106728124e-06, "loss": 0.0025, "step": 60950 }, { "epoch": 0.9974637977583245, "grad_norm": 0.12022215127944946, "learning_rate": 9.335649543812057e-06, "loss": 0.0037, "step": 60960 }, { "epoch": 0.9976274237094003, "grad_norm": 0.17834928631782532, "learning_rate": 9.335293892495787e-06, "loss": 0.0014, "step": 60970 }, { "epoch": 0.9977910496604762, "grad_norm": 0.04308273270726204, "learning_rate": 9.33493815278656e-06, "loss": 0.003, "step": 60980 }, { "epoch": 0.997954675611552, "grad_norm": 0.060553960502147675, "learning_rate": 9.33458232469163e-06, "loss": 0.0043, "step": 60990 }, { "epoch": 0.9981183015626278, "grad_norm": 0.16674906015396118, "learning_rate": 9.334226408218253e-06, "loss": 0.0014, "step": 61000 }, { "epoch": 0.9981183015626278, "eval_loss": 0.0017024847911670804, "eval_runtime": 5.3422, "eval_samples_per_second": 37.438, "eval_steps_per_second": 9.359, "step": 61000 }, { "epoch": 0.9982819275137037, "grad_norm": 0.09251892566680908, "learning_rate": 9.333870403373686e-06, "loss": 0.0018, "step": 61010 }, { "epoch": 0.9984455534647795, "grad_norm": 0.10596659779548645, "learning_rate": 9.333514310165188e-06, "loss": 0.0021, "step": 61020 }, { "epoch": 0.9986091794158554, "grad_norm": 0.07345631718635559, "learning_rate": 9.333158128600017e-06, "loss": 0.0019, "step": 61030 }, { "epoch": 0.9987728053669312, "grad_norm": 0.04949796199798584, "learning_rate": 9.332801858685438e-06, "loss": 0.002, "step": 61040 }, { "epoch": 0.998936431318007, "grad_norm": 0.04834764450788498, "learning_rate": 9.332445500428716e-06, "loss": 0.0018, "step": 61050 }, { "epoch": 0.9991000572690829, "grad_norm": 0.08664409816265106, "learning_rate": 9.332089053837112e-06, "loss": 0.0013, "step": 61060 }, { "epoch": 0.9992636832201587, "grad_norm": 0.07373697310686111, "learning_rate": 9.331732518917898e-06, "loss": 0.0015, "step": 61070 }, { "epoch": 0.9994273091712346, "grad_norm": 0.04862679913640022, "learning_rate": 9.331375895678341e-06, "loss": 0.0028, "step": 61080 }, { "epoch": 0.9995909351223105, "grad_norm": 0.10175292938947678, "learning_rate": 9.331019184125714e-06, "loss": 0.0029, "step": 61090 }, { "epoch": 0.9997545610733862, "grad_norm": 0.05772218480706215, "learning_rate": 9.330662384267289e-06, "loss": 0.0016, "step": 61100 }, { "epoch": 0.9999181870244621, "grad_norm": 0.05063002184033394, "learning_rate": 9.33030549611034e-06, "loss": 0.002, "step": 61110 }, { "epoch": 1.000081812975538, "grad_norm": 0.0473678819835186, "learning_rate": 9.329948519662147e-06, "loss": 0.0018, "step": 61120 }, { "epoch": 1.0002454389266138, "grad_norm": 0.15688979625701904, "learning_rate": 9.329591454929983e-06, "loss": 0.0022, "step": 61130 }, { "epoch": 1.0004090648776895, "grad_norm": 0.10916554182767868, "learning_rate": 9.329234301921135e-06, "loss": 0.0018, "step": 61140 }, { "epoch": 1.0005726908287655, "grad_norm": 0.016647526994347572, "learning_rate": 9.328877060642878e-06, "loss": 0.0041, "step": 61150 }, { "epoch": 1.0007363167798413, "grad_norm": 0.07490374147891998, "learning_rate": 9.328519731102501e-06, "loss": 0.0012, "step": 61160 }, { "epoch": 1.000899942730917, "grad_norm": 0.05339968204498291, "learning_rate": 9.32816231330729e-06, "loss": 0.0019, "step": 61170 }, { "epoch": 1.0010635686819929, "grad_norm": 0.08965140581130981, "learning_rate": 9.327804807264528e-06, "loss": 0.0015, "step": 61180 }, { "epoch": 1.0012271946330689, "grad_norm": 0.2142292857170105, "learning_rate": 9.327447212981507e-06, "loss": 0.0022, "step": 61190 }, { "epoch": 1.0013908205841446, "grad_norm": 0.02406143583357334, "learning_rate": 9.327089530465517e-06, "loss": 0.0026, "step": 61200 }, { "epoch": 1.0015544465352204, "grad_norm": 0.10584107041358948, "learning_rate": 9.326731759723852e-06, "loss": 0.0032, "step": 61210 }, { "epoch": 1.0017180724862964, "grad_norm": 0.14630161225795746, "learning_rate": 9.326373900763807e-06, "loss": 0.002, "step": 61220 }, { "epoch": 1.0018816984373722, "grad_norm": 0.017551448196172714, "learning_rate": 9.326015953592678e-06, "loss": 0.0012, "step": 61230 }, { "epoch": 1.002045324388448, "grad_norm": 0.057531218975782394, "learning_rate": 9.32565791821776e-06, "loss": 0.0024, "step": 61240 }, { "epoch": 1.002208950339524, "grad_norm": 0.10203977674245834, "learning_rate": 9.32529979464636e-06, "loss": 0.003, "step": 61250 }, { "epoch": 1.0023725762905997, "grad_norm": 0.029759537428617477, "learning_rate": 9.324941582885772e-06, "loss": 0.0013, "step": 61260 }, { "epoch": 1.0025362022416755, "grad_norm": 0.002003355883061886, "learning_rate": 9.324583282943305e-06, "loss": 0.0009, "step": 61270 }, { "epoch": 1.0026998281927513, "grad_norm": 0.036466244608163834, "learning_rate": 9.324224894826263e-06, "loss": 0.0022, "step": 61280 }, { "epoch": 1.0028634541438273, "grad_norm": 0.029428943991661072, "learning_rate": 9.323866418541952e-06, "loss": 0.0026, "step": 61290 }, { "epoch": 1.003027080094903, "grad_norm": 0.16884973645210266, "learning_rate": 9.323507854097684e-06, "loss": 0.0033, "step": 61300 }, { "epoch": 1.0031907060459788, "grad_norm": 0.02091153711080551, "learning_rate": 9.323149201500766e-06, "loss": 0.0016, "step": 61310 }, { "epoch": 1.0033543319970548, "grad_norm": 0.043501000851392746, "learning_rate": 9.322790460758513e-06, "loss": 0.0021, "step": 61320 }, { "epoch": 1.0035179579481306, "grad_norm": 0.07870914787054062, "learning_rate": 9.32243163187824e-06, "loss": 0.0019, "step": 61330 }, { "epoch": 1.0036815838992064, "grad_norm": 0.05861688777804375, "learning_rate": 9.322072714867261e-06, "loss": 0.003, "step": 61340 }, { "epoch": 1.0038452098502824, "grad_norm": 0.029894715175032616, "learning_rate": 9.321713709732894e-06, "loss": 0.0032, "step": 61350 }, { "epoch": 1.0040088358013581, "grad_norm": 0.21398594975471497, "learning_rate": 9.32135461648246e-06, "loss": 0.0023, "step": 61360 }, { "epoch": 1.004172461752434, "grad_norm": 0.03223501890897751, "learning_rate": 9.320995435123284e-06, "loss": 0.0018, "step": 61370 }, { "epoch": 1.0043360877035097, "grad_norm": 0.10418049246072769, "learning_rate": 9.320636165662684e-06, "loss": 0.0019, "step": 61380 }, { "epoch": 1.0044997136545857, "grad_norm": 0.04419049993157387, "learning_rate": 9.320276808107986e-06, "loss": 0.0049, "step": 61390 }, { "epoch": 1.0046633396056615, "grad_norm": 0.06908238679170609, "learning_rate": 9.31991736246652e-06, "loss": 0.0023, "step": 61400 }, { "epoch": 1.0048269655567372, "grad_norm": 0.05684058740735054, "learning_rate": 9.319557828745611e-06, "loss": 0.002, "step": 61410 }, { "epoch": 1.0049905915078132, "grad_norm": 0.05661802366375923, "learning_rate": 9.319198206952592e-06, "loss": 0.002, "step": 61420 }, { "epoch": 1.005154217458889, "grad_norm": 0.14673535525798798, "learning_rate": 9.318838497094796e-06, "loss": 0.0023, "step": 61430 }, { "epoch": 1.0053178434099648, "grad_norm": 0.17467616498470306, "learning_rate": 9.318478699179555e-06, "loss": 0.0024, "step": 61440 }, { "epoch": 1.0054814693610408, "grad_norm": 0.035273268818855286, "learning_rate": 9.318118813214207e-06, "loss": 0.0036, "step": 61450 }, { "epoch": 1.0056450953121165, "grad_norm": 0.25822019577026367, "learning_rate": 9.317758839206088e-06, "loss": 0.0021, "step": 61460 }, { "epoch": 1.0058087212631923, "grad_norm": 0.07971999049186707, "learning_rate": 9.317398777162539e-06, "loss": 0.0013, "step": 61470 }, { "epoch": 1.005972347214268, "grad_norm": 0.13383787870407104, "learning_rate": 9.3170386270909e-06, "loss": 0.0015, "step": 61480 }, { "epoch": 1.006135973165344, "grad_norm": 0.10749956220388412, "learning_rate": 9.316678388998516e-06, "loss": 0.0016, "step": 61490 }, { "epoch": 1.0062995991164199, "grad_norm": 0.32485339045524597, "learning_rate": 9.31631806289273e-06, "loss": 0.0036, "step": 61500 }, { "epoch": 1.0064632250674956, "grad_norm": 0.038222454488277435, "learning_rate": 9.31595764878089e-06, "loss": 0.0025, "step": 61510 }, { "epoch": 1.0066268510185716, "grad_norm": 0.14815863966941833, "learning_rate": 9.315597146670343e-06, "loss": 0.0018, "step": 61520 }, { "epoch": 1.0067904769696474, "grad_norm": 0.05686727911233902, "learning_rate": 9.315236556568442e-06, "loss": 0.0014, "step": 61530 }, { "epoch": 1.0069541029207232, "grad_norm": 0.003563667181879282, "learning_rate": 9.314875878482536e-06, "loss": 0.0025, "step": 61540 }, { "epoch": 1.007117728871799, "grad_norm": 0.04711107164621353, "learning_rate": 9.314515112419983e-06, "loss": 0.0011, "step": 61550 }, { "epoch": 1.007281354822875, "grad_norm": 0.017903923988342285, "learning_rate": 9.314154258388135e-06, "loss": 0.0014, "step": 61560 }, { "epoch": 1.0074449807739507, "grad_norm": 0.028776343911886215, "learning_rate": 9.313793316394351e-06, "loss": 0.0027, "step": 61570 }, { "epoch": 1.0076086067250265, "grad_norm": 0.284142404794693, "learning_rate": 9.31343228644599e-06, "loss": 0.0031, "step": 61580 }, { "epoch": 1.0077722326761025, "grad_norm": 0.08782514929771423, "learning_rate": 9.313071168550412e-06, "loss": 0.0028, "step": 61590 }, { "epoch": 1.0079358586271783, "grad_norm": 0.06311526894569397, "learning_rate": 9.31270996271498e-06, "loss": 0.0012, "step": 61600 }, { "epoch": 1.008099484578254, "grad_norm": 0.7511153817176819, "learning_rate": 9.312348668947062e-06, "loss": 0.0028, "step": 61610 }, { "epoch": 1.00826311052933, "grad_norm": 0.05686328560113907, "learning_rate": 9.311987287254022e-06, "loss": 0.004, "step": 61620 }, { "epoch": 1.0084267364804058, "grad_norm": 0.03145642578601837, "learning_rate": 9.311625817643227e-06, "loss": 0.0014, "step": 61630 }, { "epoch": 1.0085903624314816, "grad_norm": 0.09478707611560822, "learning_rate": 9.311264260122051e-06, "loss": 0.0014, "step": 61640 }, { "epoch": 1.0087539883825574, "grad_norm": 0.23451468348503113, "learning_rate": 9.31090261469786e-06, "loss": 0.0044, "step": 61650 }, { "epoch": 1.0089176143336334, "grad_norm": 0.03853405639529228, "learning_rate": 9.310540881378034e-06, "loss": 0.0017, "step": 61660 }, { "epoch": 1.0090812402847091, "grad_norm": 0.014699294231832027, "learning_rate": 9.310179060169945e-06, "loss": 0.0028, "step": 61670 }, { "epoch": 1.009244866235785, "grad_norm": 0.06072486191987991, "learning_rate": 9.309817151080969e-06, "loss": 0.0014, "step": 61680 }, { "epoch": 1.009408492186861, "grad_norm": 0.1076788380742073, "learning_rate": 9.309455154118486e-06, "loss": 0.0015, "step": 61690 }, { "epoch": 1.0095721181379367, "grad_norm": 0.07454735785722733, "learning_rate": 9.30909306928988e-06, "loss": 0.0025, "step": 61700 }, { "epoch": 1.0097357440890125, "grad_norm": 0.05920223146677017, "learning_rate": 9.30873089660253e-06, "loss": 0.0024, "step": 61710 }, { "epoch": 1.0098993700400885, "grad_norm": 0.04751500114798546, "learning_rate": 9.30836863606382e-06, "loss": 0.0021, "step": 61720 }, { "epoch": 1.0100629959911642, "grad_norm": 0.12464810162782669, "learning_rate": 9.30800628768114e-06, "loss": 0.0027, "step": 61730 }, { "epoch": 1.01022662194224, "grad_norm": 0.06774920970201492, "learning_rate": 9.307643851461874e-06, "loss": 0.001, "step": 61740 }, { "epoch": 1.0103902478933158, "grad_norm": 0.06351277977228165, "learning_rate": 9.307281327413414e-06, "loss": 0.0023, "step": 61750 }, { "epoch": 1.0105538738443918, "grad_norm": 0.07307562977075577, "learning_rate": 9.306918715543152e-06, "loss": 0.0015, "step": 61760 }, { "epoch": 1.0107174997954675, "grad_norm": 0.015813587233424187, "learning_rate": 9.306556015858479e-06, "loss": 0.0019, "step": 61770 }, { "epoch": 1.0108811257465433, "grad_norm": 0.34819281101226807, "learning_rate": 9.306193228366791e-06, "loss": 0.0015, "step": 61780 }, { "epoch": 1.0110447516976193, "grad_norm": 0.21181908249855042, "learning_rate": 9.305830353075486e-06, "loss": 0.0016, "step": 61790 }, { "epoch": 1.011208377648695, "grad_norm": 0.1325477957725525, "learning_rate": 9.305467389991964e-06, "loss": 0.0014, "step": 61800 }, { "epoch": 1.0113720035997709, "grad_norm": 0.05025854706764221, "learning_rate": 9.305104339123621e-06, "loss": 0.0019, "step": 61810 }, { "epoch": 1.0115356295508469, "grad_norm": 0.1361219435930252, "learning_rate": 9.304741200477863e-06, "loss": 0.001, "step": 61820 }, { "epoch": 1.0116992555019226, "grad_norm": 0.011360155418515205, "learning_rate": 9.304377974062093e-06, "loss": 0.0013, "step": 61830 }, { "epoch": 1.0118628814529984, "grad_norm": 0.046155381947755814, "learning_rate": 9.304014659883716e-06, "loss": 0.0013, "step": 61840 }, { "epoch": 1.0120265074040742, "grad_norm": 0.19705262780189514, "learning_rate": 9.303651257950141e-06, "loss": 0.0017, "step": 61850 }, { "epoch": 1.0121901333551502, "grad_norm": 0.08687467128038406, "learning_rate": 9.303287768268775e-06, "loss": 0.0017, "step": 61860 }, { "epoch": 1.012353759306226, "grad_norm": 0.056184105575084686, "learning_rate": 9.302924190847034e-06, "loss": 0.0017, "step": 61870 }, { "epoch": 1.0125173852573017, "grad_norm": 0.04938114434480667, "learning_rate": 9.302560525692329e-06, "loss": 0.0024, "step": 61880 }, { "epoch": 1.0126810112083777, "grad_norm": 0.021533403545618057, "learning_rate": 9.302196772812071e-06, "loss": 0.0021, "step": 61890 }, { "epoch": 1.0128446371594535, "grad_norm": 0.05036887899041176, "learning_rate": 9.301832932213681e-06, "loss": 0.0022, "step": 61900 }, { "epoch": 1.0130082631105293, "grad_norm": 0.08971427381038666, "learning_rate": 9.301469003904577e-06, "loss": 0.0022, "step": 61910 }, { "epoch": 1.0131718890616053, "grad_norm": 0.033434443175792694, "learning_rate": 9.301104987892177e-06, "loss": 0.0015, "step": 61920 }, { "epoch": 1.013335515012681, "grad_norm": 0.04931390658020973, "learning_rate": 9.300740884183905e-06, "loss": 0.0015, "step": 61930 }, { "epoch": 1.0134991409637568, "grad_norm": 0.07462302595376968, "learning_rate": 9.300376692787185e-06, "loss": 0.0038, "step": 61940 }, { "epoch": 1.0136627669148326, "grad_norm": 0.016052698716521263, "learning_rate": 9.30001241370944e-06, "loss": 0.0012, "step": 61950 }, { "epoch": 1.0138263928659086, "grad_norm": 0.07270491123199463, "learning_rate": 9.299648046958098e-06, "loss": 0.0022, "step": 61960 }, { "epoch": 1.0139900188169844, "grad_norm": 0.058857861906290054, "learning_rate": 9.299283592540592e-06, "loss": 0.0039, "step": 61970 }, { "epoch": 1.0141536447680601, "grad_norm": 0.1336885690689087, "learning_rate": 9.298919050464348e-06, "loss": 0.0013, "step": 61980 }, { "epoch": 1.0143172707191361, "grad_norm": 0.03591775521636009, "learning_rate": 9.298554420736801e-06, "loss": 0.002, "step": 61990 }, { "epoch": 1.014480896670212, "grad_norm": 0.05401716008782387, "learning_rate": 9.298189703365385e-06, "loss": 0.0026, "step": 62000 }, { "epoch": 1.0146445226212877, "grad_norm": 0.17057207226753235, "learning_rate": 9.297824898357536e-06, "loss": 0.001, "step": 62010 }, { "epoch": 1.0148081485723637, "grad_norm": 0.10331284999847412, "learning_rate": 9.297460005720694e-06, "loss": 0.0022, "step": 62020 }, { "epoch": 1.0149717745234395, "grad_norm": 0.06402084231376648, "learning_rate": 9.297095025462296e-06, "loss": 0.0024, "step": 62030 }, { "epoch": 1.0151354004745152, "grad_norm": 0.05689209699630737, "learning_rate": 9.296729957589784e-06, "loss": 0.0016, "step": 62040 }, { "epoch": 1.015299026425591, "grad_norm": 0.055492598563432693, "learning_rate": 9.296364802110605e-06, "loss": 0.0015, "step": 62050 }, { "epoch": 1.015462652376667, "grad_norm": 0.0934985801577568, "learning_rate": 9.2959995590322e-06, "loss": 0.0023, "step": 62060 }, { "epoch": 1.0156262783277428, "grad_norm": 0.0501958467066288, "learning_rate": 9.295634228362018e-06, "loss": 0.0022, "step": 62070 }, { "epoch": 1.0157899042788185, "grad_norm": 0.12663021683692932, "learning_rate": 9.295268810107508e-06, "loss": 0.0015, "step": 62080 }, { "epoch": 1.0159535302298945, "grad_norm": 0.03617825731635094, "learning_rate": 9.29490330427612e-06, "loss": 0.0011, "step": 62090 }, { "epoch": 1.0161171561809703, "grad_norm": 0.13538557291030884, "learning_rate": 9.294537710875305e-06, "loss": 0.002, "step": 62100 }, { "epoch": 1.016280782132046, "grad_norm": 0.0720764696598053, "learning_rate": 9.294172029912521e-06, "loss": 0.0024, "step": 62110 }, { "epoch": 1.016444408083122, "grad_norm": 0.04499427601695061, "learning_rate": 9.293806261395218e-06, "loss": 0.0022, "step": 62120 }, { "epoch": 1.0166080340341979, "grad_norm": 0.14523977041244507, "learning_rate": 9.293440405330858e-06, "loss": 0.0012, "step": 62130 }, { "epoch": 1.0167716599852736, "grad_norm": 0.048497430980205536, "learning_rate": 9.2930744617269e-06, "loss": 0.0018, "step": 62140 }, { "epoch": 1.0169352859363494, "grad_norm": 0.03021843358874321, "learning_rate": 9.292708430590807e-06, "loss": 0.0025, "step": 62150 }, { "epoch": 1.0170989118874254, "grad_norm": 0.10147867351770401, "learning_rate": 9.292342311930038e-06, "loss": 0.0008, "step": 62160 }, { "epoch": 1.0172625378385012, "grad_norm": 0.12681958079338074, "learning_rate": 9.291976105752058e-06, "loss": 0.0017, "step": 62170 }, { "epoch": 1.017426163789577, "grad_norm": 0.14029741287231445, "learning_rate": 9.291609812064338e-06, "loss": 0.0017, "step": 62180 }, { "epoch": 1.017589789740653, "grad_norm": 0.09267131239175797, "learning_rate": 9.291243430874343e-06, "loss": 0.0018, "step": 62190 }, { "epoch": 1.0177534156917287, "grad_norm": 0.018236424773931503, "learning_rate": 9.290876962189543e-06, "loss": 0.0023, "step": 62200 }, { "epoch": 1.0179170416428045, "grad_norm": 0.00670363986864686, "learning_rate": 9.290510406017413e-06, "loss": 0.0032, "step": 62210 }, { "epoch": 1.0180806675938805, "grad_norm": 0.03749886900186539, "learning_rate": 9.290143762365423e-06, "loss": 0.0017, "step": 62220 }, { "epoch": 1.0182442935449563, "grad_norm": 0.09181356430053711, "learning_rate": 9.289777031241051e-06, "loss": 0.0027, "step": 62230 }, { "epoch": 1.018407919496032, "grad_norm": 0.03308061137795448, "learning_rate": 9.289410212651774e-06, "loss": 0.0021, "step": 62240 }, { "epoch": 1.0185715454471078, "grad_norm": 0.10101155191659927, "learning_rate": 9.28904330660507e-06, "loss": 0.0074, "step": 62250 }, { "epoch": 1.0187351713981838, "grad_norm": 0.009913098998367786, "learning_rate": 9.288676313108422e-06, "loss": 0.002, "step": 62260 }, { "epoch": 1.0188987973492596, "grad_norm": 0.08576328307390213, "learning_rate": 9.28830923216931e-06, "loss": 0.0022, "step": 62270 }, { "epoch": 1.0190624233003354, "grad_norm": 0.06479435414075851, "learning_rate": 9.28794206379522e-06, "loss": 0.0025, "step": 62280 }, { "epoch": 1.0192260492514114, "grad_norm": 0.10950502753257751, "learning_rate": 9.287574807993639e-06, "loss": 0.0025, "step": 62290 }, { "epoch": 1.0193896752024871, "grad_norm": 0.05064884200692177, "learning_rate": 9.287207464772054e-06, "loss": 0.0024, "step": 62300 }, { "epoch": 1.019553301153563, "grad_norm": 0.019223345443606377, "learning_rate": 9.286840034137954e-06, "loss": 0.0015, "step": 62310 }, { "epoch": 1.019716927104639, "grad_norm": 0.036392517387866974, "learning_rate": 9.286472516098832e-06, "loss": 0.0033, "step": 62320 }, { "epoch": 1.0198805530557147, "grad_norm": 0.05706041306257248, "learning_rate": 9.286104910662181e-06, "loss": 0.0024, "step": 62330 }, { "epoch": 1.0200441790067905, "grad_norm": 0.02115272730588913, "learning_rate": 9.285737217835496e-06, "loss": 0.0035, "step": 62340 }, { "epoch": 1.0202078049578662, "grad_norm": 0.1077798381447792, "learning_rate": 9.285369437626273e-06, "loss": 0.0019, "step": 62350 }, { "epoch": 1.0203714309089422, "grad_norm": 0.5761215090751648, "learning_rate": 9.285001570042014e-06, "loss": 0.0012, "step": 62360 }, { "epoch": 1.020535056860018, "grad_norm": 0.05048255994915962, "learning_rate": 9.284633615090215e-06, "loss": 0.002, "step": 62370 }, { "epoch": 1.0206986828110938, "grad_norm": 0.2519931197166443, "learning_rate": 9.284265572778381e-06, "loss": 0.0023, "step": 62380 }, { "epoch": 1.0208623087621698, "grad_norm": 0.13181249797344208, "learning_rate": 9.283897443114017e-06, "loss": 0.0025, "step": 62390 }, { "epoch": 1.0210259347132455, "grad_norm": 0.05920034646987915, "learning_rate": 9.283529226104626e-06, "loss": 0.003, "step": 62400 }, { "epoch": 1.0211895606643213, "grad_norm": 0.0257562305778265, "learning_rate": 9.283160921757718e-06, "loss": 0.0017, "step": 62410 }, { "epoch": 1.0213531866153973, "grad_norm": 0.009482457302510738, "learning_rate": 9.282792530080801e-06, "loss": 0.0027, "step": 62420 }, { "epoch": 1.021516812566473, "grad_norm": 0.07290036976337433, "learning_rate": 9.282424051081388e-06, "loss": 0.0035, "step": 62430 }, { "epoch": 1.0216804385175489, "grad_norm": 0.043114710599184036, "learning_rate": 9.28205548476699e-06, "loss": 0.0018, "step": 62440 }, { "epoch": 1.0218440644686246, "grad_norm": 0.15389874577522278, "learning_rate": 9.281686831145121e-06, "loss": 0.0028, "step": 62450 }, { "epoch": 1.0220076904197006, "grad_norm": 0.06127632409334183, "learning_rate": 9.2813180902233e-06, "loss": 0.0064, "step": 62460 }, { "epoch": 1.0221713163707764, "grad_norm": 0.20663292706012726, "learning_rate": 9.280949262009044e-06, "loss": 0.0025, "step": 62470 }, { "epoch": 1.0223349423218522, "grad_norm": 0.1333935260772705, "learning_rate": 9.280580346509873e-06, "loss": 0.0022, "step": 62480 }, { "epoch": 1.0224985682729282, "grad_norm": 0.017575398087501526, "learning_rate": 9.28021134373331e-06, "loss": 0.0022, "step": 62490 }, { "epoch": 1.022662194224004, "grad_norm": 0.0770731121301651, "learning_rate": 9.279842253686878e-06, "loss": 0.0013, "step": 62500 }, { "epoch": 1.0228258201750797, "grad_norm": 0.11944708973169327, "learning_rate": 9.279473076378102e-06, "loss": 0.0022, "step": 62510 }, { "epoch": 1.0229894461261555, "grad_norm": 0.026962194591760635, "learning_rate": 9.279103811814509e-06, "loss": 0.0019, "step": 62520 }, { "epoch": 1.0231530720772315, "grad_norm": 0.11368787288665771, "learning_rate": 9.278734460003629e-06, "loss": 0.0014, "step": 62530 }, { "epoch": 1.0233166980283073, "grad_norm": 0.10832122713327408, "learning_rate": 9.278365020952993e-06, "loss": 0.0021, "step": 62540 }, { "epoch": 1.023480323979383, "grad_norm": 0.024616006761789322, "learning_rate": 9.277995494670133e-06, "loss": 0.0012, "step": 62550 }, { "epoch": 1.023643949930459, "grad_norm": 0.22573445737361908, "learning_rate": 9.277625881162582e-06, "loss": 0.0021, "step": 62560 }, { "epoch": 1.0238075758815348, "grad_norm": 0.058213308453559875, "learning_rate": 9.277256180437877e-06, "loss": 0.0015, "step": 62570 }, { "epoch": 1.0239712018326106, "grad_norm": 0.10979881882667542, "learning_rate": 9.276886392503558e-06, "loss": 0.0028, "step": 62580 }, { "epoch": 1.0241348277836866, "grad_norm": 0.03481617197394371, "learning_rate": 9.27651651736716e-06, "loss": 0.0016, "step": 62590 }, { "epoch": 1.0242984537347624, "grad_norm": 0.0554765909910202, "learning_rate": 9.27614655503623e-06, "loss": 0.0023, "step": 62600 }, { "epoch": 1.0244620796858381, "grad_norm": 0.05459179729223251, "learning_rate": 9.275776505518307e-06, "loss": 0.0028, "step": 62610 }, { "epoch": 1.024625705636914, "grad_norm": 0.05538322404026985, "learning_rate": 9.275406368820938e-06, "loss": 0.0017, "step": 62620 }, { "epoch": 1.02478933158799, "grad_norm": 0.05171097069978714, "learning_rate": 9.27503614495167e-06, "loss": 0.001, "step": 62630 }, { "epoch": 1.0249529575390657, "grad_norm": 0.028707332909107208, "learning_rate": 9.274665833918049e-06, "loss": 0.0016, "step": 62640 }, { "epoch": 1.0251165834901415, "grad_norm": 0.04786459729075432, "learning_rate": 9.274295435727627e-06, "loss": 0.0018, "step": 62650 }, { "epoch": 1.0252802094412174, "grad_norm": 0.05466194078326225, "learning_rate": 9.273924950387958e-06, "loss": 0.0043, "step": 62660 }, { "epoch": 1.0254438353922932, "grad_norm": 0.014386789873242378, "learning_rate": 9.273554377906593e-06, "loss": 0.0026, "step": 62670 }, { "epoch": 1.025607461343369, "grad_norm": 0.03119182586669922, "learning_rate": 9.27318371829109e-06, "loss": 0.0015, "step": 62680 }, { "epoch": 1.025771087294445, "grad_norm": 0.04187420755624771, "learning_rate": 9.272812971549003e-06, "loss": 0.0012, "step": 62690 }, { "epoch": 1.0259347132455208, "grad_norm": 0.08008750528097153, "learning_rate": 9.272442137687895e-06, "loss": 0.0026, "step": 62700 }, { "epoch": 1.0260983391965965, "grad_norm": 0.10847261548042297, "learning_rate": 9.272071216715326e-06, "loss": 0.0022, "step": 62710 }, { "epoch": 1.0262619651476723, "grad_norm": 0.06784509867429733, "learning_rate": 9.271700208638856e-06, "loss": 0.0015, "step": 62720 }, { "epoch": 1.0264255910987483, "grad_norm": 0.043881505727767944, "learning_rate": 9.271329113466053e-06, "loss": 0.0026, "step": 62730 }, { "epoch": 1.026589217049824, "grad_norm": 0.05852636694908142, "learning_rate": 9.270957931204482e-06, "loss": 0.0017, "step": 62740 }, { "epoch": 1.0267528430008999, "grad_norm": 0.032690223306417465, "learning_rate": 9.270586661861712e-06, "loss": 0.0016, "step": 62750 }, { "epoch": 1.0269164689519759, "grad_norm": 0.18903903663158417, "learning_rate": 9.270215305445311e-06, "loss": 0.0021, "step": 62760 }, { "epoch": 1.0270800949030516, "grad_norm": 0.10049130767583847, "learning_rate": 9.26984386196285e-06, "loss": 0.0013, "step": 62770 }, { "epoch": 1.0272437208541274, "grad_norm": 0.021498991176486015, "learning_rate": 9.269472331421907e-06, "loss": 0.0033, "step": 62780 }, { "epoch": 1.0274073468052034, "grad_norm": 0.03256230801343918, "learning_rate": 9.269100713830053e-06, "loss": 0.0007, "step": 62790 }, { "epoch": 1.0275709727562792, "grad_norm": 0.026557376608252525, "learning_rate": 9.268729009194865e-06, "loss": 0.0019, "step": 62800 }, { "epoch": 1.027734598707355, "grad_norm": 0.036971915513277054, "learning_rate": 9.268357217523924e-06, "loss": 0.0032, "step": 62810 }, { "epoch": 1.0278982246584307, "grad_norm": 0.03602815419435501, "learning_rate": 9.26798533882481e-06, "loss": 0.0022, "step": 62820 }, { "epoch": 1.0280618506095067, "grad_norm": 0.11008908599615097, "learning_rate": 9.267613373105104e-06, "loss": 0.0017, "step": 62830 }, { "epoch": 1.0282254765605825, "grad_norm": 0.052703067660331726, "learning_rate": 9.267241320372391e-06, "loss": 0.0021, "step": 62840 }, { "epoch": 1.0283891025116583, "grad_norm": 0.035865988582372665, "learning_rate": 9.266869180634256e-06, "loss": 0.0009, "step": 62850 }, { "epoch": 1.0285527284627343, "grad_norm": 0.07268939167261124, "learning_rate": 9.266496953898289e-06, "loss": 0.0029, "step": 62860 }, { "epoch": 1.02871635441381, "grad_norm": 0.06735679507255554, "learning_rate": 9.266124640172075e-06, "loss": 0.0022, "step": 62870 }, { "epoch": 1.0288799803648858, "grad_norm": 0.045382436364889145, "learning_rate": 9.26575223946321e-06, "loss": 0.0011, "step": 62880 }, { "epoch": 1.0290436063159618, "grad_norm": 0.07382488995790482, "learning_rate": 9.265379751779285e-06, "loss": 0.002, "step": 62890 }, { "epoch": 1.0292072322670376, "grad_norm": 0.025516636669635773, "learning_rate": 9.265007177127894e-06, "loss": 0.0019, "step": 62900 }, { "epoch": 1.0293708582181134, "grad_norm": 0.07589970529079437, "learning_rate": 9.264634515516633e-06, "loss": 0.0019, "step": 62910 }, { "epoch": 1.0295344841691891, "grad_norm": 0.06228888779878616, "learning_rate": 9.264261766953101e-06, "loss": 0.002, "step": 62920 }, { "epoch": 1.0296981101202651, "grad_norm": 0.10739631205797195, "learning_rate": 9.263888931444898e-06, "loss": 0.0028, "step": 62930 }, { "epoch": 1.029861736071341, "grad_norm": 0.05957471951842308, "learning_rate": 9.263516008999628e-06, "loss": 0.0018, "step": 62940 }, { "epoch": 1.0300253620224167, "grad_norm": 0.010999472811818123, "learning_rate": 9.26314299962489e-06, "loss": 0.0013, "step": 62950 }, { "epoch": 1.0301889879734927, "grad_norm": 0.30671054124832153, "learning_rate": 9.262769903328294e-06, "loss": 0.0018, "step": 62960 }, { "epoch": 1.0303526139245685, "grad_norm": 0.040985554456710815, "learning_rate": 9.262396720117444e-06, "loss": 0.0018, "step": 62970 }, { "epoch": 1.0305162398756442, "grad_norm": 0.010178688913583755, "learning_rate": 9.262023449999951e-06, "loss": 0.0014, "step": 62980 }, { "epoch": 1.0306798658267202, "grad_norm": 0.04184887558221817, "learning_rate": 9.261650092983423e-06, "loss": 0.0022, "step": 62990 }, { "epoch": 1.030843491777796, "grad_norm": 0.14219936728477478, "learning_rate": 9.261276649075474e-06, "loss": 0.0016, "step": 63000 }, { "epoch": 1.0310071177288718, "grad_norm": 0.03750208392739296, "learning_rate": 9.260903118283719e-06, "loss": 0.0014, "step": 63010 }, { "epoch": 1.0311707436799475, "grad_norm": 0.08180815726518631, "learning_rate": 9.260529500615774e-06, "loss": 0.0028, "step": 63020 }, { "epoch": 1.0313343696310235, "grad_norm": 0.08954513072967529, "learning_rate": 9.260155796079255e-06, "loss": 0.0015, "step": 63030 }, { "epoch": 1.0314979955820993, "grad_norm": 0.454410582780838, "learning_rate": 9.259782004681783e-06, "loss": 0.0022, "step": 63040 }, { "epoch": 1.031661621533175, "grad_norm": 0.2835443615913391, "learning_rate": 9.259408126430977e-06, "loss": 0.0027, "step": 63050 }, { "epoch": 1.031825247484251, "grad_norm": 0.10159897059202194, "learning_rate": 9.259034161334464e-06, "loss": 0.0014, "step": 63060 }, { "epoch": 1.0319888734353269, "grad_norm": 0.023983903229236603, "learning_rate": 9.258660109399866e-06, "loss": 0.0015, "step": 63070 }, { "epoch": 1.0321524993864026, "grad_norm": 0.11300794035196304, "learning_rate": 9.25828597063481e-06, "loss": 0.005, "step": 63080 }, { "epoch": 1.0323161253374786, "grad_norm": 0.04699046164751053, "learning_rate": 9.257911745046924e-06, "loss": 0.0018, "step": 63090 }, { "epoch": 1.0324797512885544, "grad_norm": 0.10544373840093613, "learning_rate": 9.25753743264384e-06, "loss": 0.0017, "step": 63100 }, { "epoch": 1.0326433772396302, "grad_norm": 0.025078650563955307, "learning_rate": 9.257163033433187e-06, "loss": 0.0018, "step": 63110 }, { "epoch": 1.032807003190706, "grad_norm": 0.09279242902994156, "learning_rate": 9.256788547422601e-06, "loss": 0.0018, "step": 63120 }, { "epoch": 1.032970629141782, "grad_norm": 0.1307639330625534, "learning_rate": 9.256413974619718e-06, "loss": 0.0018, "step": 63130 }, { "epoch": 1.0331342550928577, "grad_norm": 0.10855966806411743, "learning_rate": 9.256039315032172e-06, "loss": 0.0018, "step": 63140 }, { "epoch": 1.0332978810439335, "grad_norm": 0.04002469778060913, "learning_rate": 9.255664568667605e-06, "loss": 0.002, "step": 63150 }, { "epoch": 1.0334615069950095, "grad_norm": 0.04228093847632408, "learning_rate": 9.255289735533656e-06, "loss": 0.003, "step": 63160 }, { "epoch": 1.0336251329460853, "grad_norm": 0.07610825449228287, "learning_rate": 9.25491481563797e-06, "loss": 0.0012, "step": 63170 }, { "epoch": 1.033788758897161, "grad_norm": 0.18874569237232208, "learning_rate": 9.254539808988189e-06, "loss": 0.0024, "step": 63180 }, { "epoch": 1.033952384848237, "grad_norm": 0.08838175237178802, "learning_rate": 9.254164715591959e-06, "loss": 0.0017, "step": 63190 }, { "epoch": 1.0341160107993128, "grad_norm": 0.10994619876146317, "learning_rate": 9.253789535456929e-06, "loss": 0.0018, "step": 63200 }, { "epoch": 1.0342796367503886, "grad_norm": 0.029415730386972427, "learning_rate": 9.253414268590748e-06, "loss": 0.0018, "step": 63210 }, { "epoch": 1.0344432627014644, "grad_norm": 0.04671817272901535, "learning_rate": 9.253038915001066e-06, "loss": 0.0016, "step": 63220 }, { "epoch": 1.0346068886525404, "grad_norm": 0.05935228243470192, "learning_rate": 9.25266347469554e-06, "loss": 0.0021, "step": 63230 }, { "epoch": 1.0347705146036161, "grad_norm": 0.10359574109315872, "learning_rate": 9.252287947681822e-06, "loss": 0.0019, "step": 63240 }, { "epoch": 1.034934140554692, "grad_norm": 0.04205641895532608, "learning_rate": 9.251912333967567e-06, "loss": 0.0016, "step": 63250 }, { "epoch": 1.035097766505768, "grad_norm": 0.037503186613321304, "learning_rate": 9.251536633560439e-06, "loss": 0.0018, "step": 63260 }, { "epoch": 1.0352613924568437, "grad_norm": 0.13158519566059113, "learning_rate": 9.251160846468093e-06, "loss": 0.0019, "step": 63270 }, { "epoch": 1.0354250184079195, "grad_norm": 0.07353059947490692, "learning_rate": 9.250784972698192e-06, "loss": 0.002, "step": 63280 }, { "epoch": 1.0355886443589952, "grad_norm": 0.11290540546178818, "learning_rate": 9.250409012258401e-06, "loss": 0.0024, "step": 63290 }, { "epoch": 1.0357522703100712, "grad_norm": 0.05592896789312363, "learning_rate": 9.250032965156386e-06, "loss": 0.0012, "step": 63300 }, { "epoch": 1.035915896261147, "grad_norm": 0.023617852479219437, "learning_rate": 9.249656831399814e-06, "loss": 0.0009, "step": 63310 }, { "epoch": 1.0360795222122228, "grad_norm": 0.034140799194574356, "learning_rate": 9.249280610996352e-06, "loss": 0.0023, "step": 63320 }, { "epoch": 1.0362431481632988, "grad_norm": 0.08920124173164368, "learning_rate": 9.248904303953672e-06, "loss": 0.002, "step": 63330 }, { "epoch": 1.0364067741143745, "grad_norm": 0.17879237234592438, "learning_rate": 9.248527910279447e-06, "loss": 0.002, "step": 63340 }, { "epoch": 1.0365704000654503, "grad_norm": 0.09365537762641907, "learning_rate": 9.248151429981353e-06, "loss": 0.002, "step": 63350 }, { "epoch": 1.0367340260165263, "grad_norm": 0.10512114316225052, "learning_rate": 9.247774863067063e-06, "loss": 0.0023, "step": 63360 }, { "epoch": 1.036897651967602, "grad_norm": 0.09406161308288574, "learning_rate": 9.247398209544255e-06, "loss": 0.0021, "step": 63370 }, { "epoch": 1.0370612779186779, "grad_norm": 0.022947320714592934, "learning_rate": 9.24702146942061e-06, "loss": 0.0012, "step": 63380 }, { "epoch": 1.0372249038697539, "grad_norm": 0.5647009611129761, "learning_rate": 9.24664464270381e-06, "loss": 0.0027, "step": 63390 }, { "epoch": 1.0373885298208296, "grad_norm": 0.11487042158842087, "learning_rate": 9.24626772940154e-06, "loss": 0.0025, "step": 63400 }, { "epoch": 1.0375521557719054, "grad_norm": 0.05939564108848572, "learning_rate": 9.245890729521478e-06, "loss": 0.0015, "step": 63410 }, { "epoch": 1.0377157817229812, "grad_norm": 0.03685497120022774, "learning_rate": 9.245513643071317e-06, "loss": 0.0025, "step": 63420 }, { "epoch": 1.0378794076740572, "grad_norm": 0.05458051711320877, "learning_rate": 9.245136470058743e-06, "loss": 0.0013, "step": 63430 }, { "epoch": 1.038043033625133, "grad_norm": 0.0038172940257936716, "learning_rate": 9.244759210491448e-06, "loss": 0.0014, "step": 63440 }, { "epoch": 1.0382066595762087, "grad_norm": 0.05386197939515114, "learning_rate": 9.244381864377121e-06, "loss": 0.0019, "step": 63450 }, { "epoch": 1.0383702855272847, "grad_norm": 0.017694367095828056, "learning_rate": 9.244004431723458e-06, "loss": 0.0021, "step": 63460 }, { "epoch": 1.0385339114783605, "grad_norm": 0.12909577786922455, "learning_rate": 9.243626912538154e-06, "loss": 0.0054, "step": 63470 }, { "epoch": 1.0386975374294363, "grad_norm": 0.05565285310149193, "learning_rate": 9.243249306828907e-06, "loss": 0.0012, "step": 63480 }, { "epoch": 1.038861163380512, "grad_norm": 0.024307239800691605, "learning_rate": 9.242871614603415e-06, "loss": 0.0019, "step": 63490 }, { "epoch": 1.039024789331588, "grad_norm": 0.1676136553287506, "learning_rate": 9.24249383586938e-06, "loss": 0.0043, "step": 63500 }, { "epoch": 1.0391884152826638, "grad_norm": 0.20547142624855042, "learning_rate": 9.242115970634502e-06, "loss": 0.0032, "step": 63510 }, { "epoch": 1.0393520412337396, "grad_norm": 0.0793842151761055, "learning_rate": 9.241738018906487e-06, "loss": 0.0013, "step": 63520 }, { "epoch": 1.0395156671848156, "grad_norm": 0.1551537960767746, "learning_rate": 9.241359980693043e-06, "loss": 0.0026, "step": 63530 }, { "epoch": 1.0396792931358914, "grad_norm": 0.060188598930835724, "learning_rate": 9.240981856001876e-06, "loss": 0.0023, "step": 63540 }, { "epoch": 1.0398429190869671, "grad_norm": 0.0672174021601677, "learning_rate": 9.240603644840695e-06, "loss": 0.001, "step": 63550 }, { "epoch": 1.0400065450380431, "grad_norm": 0.06962887197732925, "learning_rate": 9.240225347217213e-06, "loss": 0.0048, "step": 63560 }, { "epoch": 1.040170170989119, "grad_norm": 0.08981279283761978, "learning_rate": 9.239846963139141e-06, "loss": 0.0008, "step": 63570 }, { "epoch": 1.0403337969401947, "grad_norm": 0.055277902632951736, "learning_rate": 9.239468492614197e-06, "loss": 0.005, "step": 63580 }, { "epoch": 1.0404974228912705, "grad_norm": 0.09173452854156494, "learning_rate": 9.239089935650094e-06, "loss": 0.0028, "step": 63590 }, { "epoch": 1.0406610488423464, "grad_norm": 0.07282580435276031, "learning_rate": 9.238711292254553e-06, "loss": 0.0015, "step": 63600 }, { "epoch": 1.0408246747934222, "grad_norm": 0.027994144707918167, "learning_rate": 9.238332562435295e-06, "loss": 0.0012, "step": 63610 }, { "epoch": 1.040988300744498, "grad_norm": 0.03037797287106514, "learning_rate": 9.23795374620004e-06, "loss": 0.0014, "step": 63620 }, { "epoch": 1.041151926695574, "grad_norm": 0.04334964230656624, "learning_rate": 9.237574843556512e-06, "loss": 0.0018, "step": 63630 }, { "epoch": 1.0413155526466498, "grad_norm": 0.06752116978168488, "learning_rate": 9.237195854512436e-06, "loss": 0.0023, "step": 63640 }, { "epoch": 1.0414791785977255, "grad_norm": 0.04502108693122864, "learning_rate": 9.236816779075541e-06, "loss": 0.0025, "step": 63650 }, { "epoch": 1.0416428045488015, "grad_norm": 0.12341869622468948, "learning_rate": 9.236437617253556e-06, "loss": 0.0012, "step": 63660 }, { "epoch": 1.0418064304998773, "grad_norm": 0.0728251188993454, "learning_rate": 9.23605836905421e-06, "loss": 0.0032, "step": 63670 }, { "epoch": 1.041970056450953, "grad_norm": 0.03771638870239258, "learning_rate": 9.235679034485237e-06, "loss": 0.001, "step": 63680 }, { "epoch": 1.0421336824020289, "grad_norm": 0.15158256888389587, "learning_rate": 9.23529961355437e-06, "loss": 0.0029, "step": 63690 }, { "epoch": 1.0422973083531049, "grad_norm": 0.08546021580696106, "learning_rate": 9.234920106269346e-06, "loss": 0.0017, "step": 63700 }, { "epoch": 1.0424609343041806, "grad_norm": 0.13850091397762299, "learning_rate": 9.234540512637905e-06, "loss": 0.0033, "step": 63710 }, { "epoch": 1.0426245602552564, "grad_norm": 0.02540903352200985, "learning_rate": 9.234160832667781e-06, "loss": 0.0024, "step": 63720 }, { "epoch": 1.0427881862063324, "grad_norm": 0.0077177188359200954, "learning_rate": 9.23378106636672e-06, "loss": 0.0019, "step": 63730 }, { "epoch": 1.0429518121574082, "grad_norm": 0.016136493533849716, "learning_rate": 9.233401213742464e-06, "loss": 0.0025, "step": 63740 }, { "epoch": 1.043115438108484, "grad_norm": 0.04357876628637314, "learning_rate": 9.233021274802759e-06, "loss": 0.0021, "step": 63750 }, { "epoch": 1.04327906405956, "grad_norm": 0.057133276015520096, "learning_rate": 9.232641249555348e-06, "loss": 0.0018, "step": 63760 }, { "epoch": 1.0434426900106357, "grad_norm": 0.047275982797145844, "learning_rate": 9.23226113800798e-06, "loss": 0.0032, "step": 63770 }, { "epoch": 1.0436063159617115, "grad_norm": 0.02346358448266983, "learning_rate": 9.23188094016841e-06, "loss": 0.0023, "step": 63780 }, { "epoch": 1.0437699419127873, "grad_norm": 0.053049445152282715, "learning_rate": 9.231500656044387e-06, "loss": 0.0047, "step": 63790 }, { "epoch": 1.0439335678638633, "grad_norm": 0.04540481045842171, "learning_rate": 9.231120285643662e-06, "loss": 0.0017, "step": 63800 }, { "epoch": 1.044097193814939, "grad_norm": 0.06433603912591934, "learning_rate": 9.230739828973994e-06, "loss": 0.0018, "step": 63810 }, { "epoch": 1.0442608197660148, "grad_norm": 0.046505123376846313, "learning_rate": 9.23035928604314e-06, "loss": 0.001, "step": 63820 }, { "epoch": 1.0444244457170908, "grad_norm": 0.07851351052522659, "learning_rate": 9.229978656858856e-06, "loss": 0.0023, "step": 63830 }, { "epoch": 1.0445880716681666, "grad_norm": 0.036209143698215485, "learning_rate": 9.229597941428907e-06, "loss": 0.001, "step": 63840 }, { "epoch": 1.0447516976192424, "grad_norm": 0.03418751806020737, "learning_rate": 9.22921713976105e-06, "loss": 0.0015, "step": 63850 }, { "epoch": 1.0449153235703184, "grad_norm": 0.09652628749608994, "learning_rate": 9.228836251863055e-06, "loss": 0.0021, "step": 63860 }, { "epoch": 1.0450789495213941, "grad_norm": 0.05415923520922661, "learning_rate": 9.228455277742685e-06, "loss": 0.0025, "step": 63870 }, { "epoch": 1.04524257547247, "grad_norm": 0.03881541267037392, "learning_rate": 9.228074217407707e-06, "loss": 0.0027, "step": 63880 }, { "epoch": 1.0454062014235457, "grad_norm": 0.07290578633546829, "learning_rate": 9.227693070865892e-06, "loss": 0.0016, "step": 63890 }, { "epoch": 1.0455698273746217, "grad_norm": 0.06994597613811493, "learning_rate": 9.22731183812501e-06, "loss": 0.0016, "step": 63900 }, { "epoch": 1.0457334533256974, "grad_norm": 0.1583678275346756, "learning_rate": 9.226930519192836e-06, "loss": 0.0029, "step": 63910 }, { "epoch": 1.0458970792767732, "grad_norm": 0.03433655947446823, "learning_rate": 9.226549114077143e-06, "loss": 0.0016, "step": 63920 }, { "epoch": 1.0460607052278492, "grad_norm": 0.04526662826538086, "learning_rate": 9.226167622785709e-06, "loss": 0.0026, "step": 63930 }, { "epoch": 1.046224331178925, "grad_norm": 0.0406726635992527, "learning_rate": 9.225786045326311e-06, "loss": 0.0046, "step": 63940 }, { "epoch": 1.0463879571300008, "grad_norm": 0.048367422074079514, "learning_rate": 9.22540438170673e-06, "loss": 0.0025, "step": 63950 }, { "epoch": 1.0465515830810768, "grad_norm": 0.03693831339478493, "learning_rate": 9.225022631934745e-06, "loss": 0.0032, "step": 63960 }, { "epoch": 1.0467152090321525, "grad_norm": 0.056611318141222, "learning_rate": 9.224640796018144e-06, "loss": 0.0018, "step": 63970 }, { "epoch": 1.0468788349832283, "grad_norm": 0.05493976175785065, "learning_rate": 9.224258873964708e-06, "loss": 0.0021, "step": 63980 }, { "epoch": 1.047042460934304, "grad_norm": 0.05654580518603325, "learning_rate": 9.223876865782229e-06, "loss": 0.0022, "step": 63990 }, { "epoch": 1.04720608688538, "grad_norm": 0.027430761605501175, "learning_rate": 9.223494771478492e-06, "loss": 0.0027, "step": 64000 }, { "epoch": 1.0473697128364559, "grad_norm": 0.1330675482749939, "learning_rate": 9.223112591061287e-06, "loss": 0.0016, "step": 64010 }, { "epoch": 1.0475333387875316, "grad_norm": 0.08558131009340286, "learning_rate": 9.22273032453841e-06, "loss": 0.0031, "step": 64020 }, { "epoch": 1.0476969647386076, "grad_norm": 0.049288202077150345, "learning_rate": 9.222347971917652e-06, "loss": 0.0009, "step": 64030 }, { "epoch": 1.0478605906896834, "grad_norm": 0.09972430765628815, "learning_rate": 9.221965533206808e-06, "loss": 0.0017, "step": 64040 }, { "epoch": 1.0480242166407592, "grad_norm": 0.08666155487298965, "learning_rate": 9.221583008413682e-06, "loss": 0.0015, "step": 64050 }, { "epoch": 1.048187842591835, "grad_norm": 0.10525741428136826, "learning_rate": 9.221200397546065e-06, "loss": 0.0022, "step": 64060 }, { "epoch": 1.048351468542911, "grad_norm": 0.16131643950939178, "learning_rate": 9.220817700611762e-06, "loss": 0.0014, "step": 64070 }, { "epoch": 1.0485150944939867, "grad_norm": 0.10731440782546997, "learning_rate": 9.220434917618576e-06, "loss": 0.0023, "step": 64080 }, { "epoch": 1.0486787204450625, "grad_norm": 0.33253514766693115, "learning_rate": 9.220052048574313e-06, "loss": 0.0015, "step": 64090 }, { "epoch": 1.0488423463961385, "grad_norm": 0.1640753298997879, "learning_rate": 9.219669093486777e-06, "loss": 0.0039, "step": 64100 }, { "epoch": 1.0490059723472143, "grad_norm": 0.04951220005750656, "learning_rate": 9.219286052363777e-06, "loss": 0.0036, "step": 64110 }, { "epoch": 1.04916959829829, "grad_norm": 0.07228408008813858, "learning_rate": 9.218902925213122e-06, "loss": 0.0026, "step": 64120 }, { "epoch": 1.049333224249366, "grad_norm": 0.05793791264295578, "learning_rate": 9.218519712042625e-06, "loss": 0.0012, "step": 64130 }, { "epoch": 1.0494968502004418, "grad_norm": 0.10001615434885025, "learning_rate": 9.218136412860099e-06, "loss": 0.0018, "step": 64140 }, { "epoch": 1.0496604761515176, "grad_norm": 0.061796873807907104, "learning_rate": 9.217753027673359e-06, "loss": 0.0022, "step": 64150 }, { "epoch": 1.0498241021025936, "grad_norm": 0.04905644804239273, "learning_rate": 9.21736955649022e-06, "loss": 0.0016, "step": 64160 }, { "epoch": 1.0499877280536694, "grad_norm": 0.030251435935497284, "learning_rate": 9.216985999318504e-06, "loss": 0.0024, "step": 64170 }, { "epoch": 1.0501513540047451, "grad_norm": 0.0831340104341507, "learning_rate": 9.21660235616603e-06, "loss": 0.0024, "step": 64180 }, { "epoch": 1.050314979955821, "grad_norm": 0.11990511417388916, "learning_rate": 9.216218627040621e-06, "loss": 0.0019, "step": 64190 }, { "epoch": 1.050478605906897, "grad_norm": 0.023926081135869026, "learning_rate": 9.215834811950097e-06, "loss": 0.0027, "step": 64200 }, { "epoch": 1.0506422318579727, "grad_norm": 0.12977640330791473, "learning_rate": 9.21545091090229e-06, "loss": 0.0025, "step": 64210 }, { "epoch": 1.0508058578090484, "grad_norm": 0.22559620440006256, "learning_rate": 9.215066923905022e-06, "loss": 0.0028, "step": 64220 }, { "epoch": 1.0509694837601244, "grad_norm": 0.02161334827542305, "learning_rate": 9.214682850966125e-06, "loss": 0.0012, "step": 64230 }, { "epoch": 1.0511331097112002, "grad_norm": 0.014958570711314678, "learning_rate": 9.214298692093429e-06, "loss": 0.0016, "step": 64240 }, { "epoch": 1.051296735662276, "grad_norm": 0.07277438044548035, "learning_rate": 9.213914447294765e-06, "loss": 0.0017, "step": 64250 }, { "epoch": 1.0514603616133518, "grad_norm": 0.1155308336019516, "learning_rate": 9.21353011657797e-06, "loss": 0.0021, "step": 64260 }, { "epoch": 1.0516239875644278, "grad_norm": 0.015788644552230835, "learning_rate": 9.213145699950882e-06, "loss": 0.0026, "step": 64270 }, { "epoch": 1.0517876135155035, "grad_norm": 0.06607986241579056, "learning_rate": 9.212761197421334e-06, "loss": 0.0014, "step": 64280 }, { "epoch": 1.0519512394665793, "grad_norm": 0.1545177698135376, "learning_rate": 9.21237660899717e-06, "loss": 0.0015, "step": 64290 }, { "epoch": 1.0521148654176553, "grad_norm": 0.03331146016716957, "learning_rate": 9.211991934686227e-06, "loss": 0.0022, "step": 64300 }, { "epoch": 1.052278491368731, "grad_norm": 0.04420895129442215, "learning_rate": 9.21160717449635e-06, "loss": 0.0023, "step": 64310 }, { "epoch": 1.0524421173198069, "grad_norm": 0.020057348534464836, "learning_rate": 9.211222328435388e-06, "loss": 0.0016, "step": 64320 }, { "epoch": 1.0526057432708829, "grad_norm": 0.007954043336212635, "learning_rate": 9.210837396511182e-06, "loss": 0.0026, "step": 64330 }, { "epoch": 1.0527693692219586, "grad_norm": 0.055748481303453445, "learning_rate": 9.210452378731583e-06, "loss": 0.0013, "step": 64340 }, { "epoch": 1.0529329951730344, "grad_norm": 0.06351203471422195, "learning_rate": 9.210067275104441e-06, "loss": 0.0028, "step": 64350 }, { "epoch": 1.0530966211241102, "grad_norm": 0.13670538365840912, "learning_rate": 9.209682085637608e-06, "loss": 0.0019, "step": 64360 }, { "epoch": 1.0532602470751862, "grad_norm": 0.11048968881368637, "learning_rate": 9.209296810338937e-06, "loss": 0.002, "step": 64370 }, { "epoch": 1.053423873026262, "grad_norm": 0.04444071277976036, "learning_rate": 9.208911449216284e-06, "loss": 0.0013, "step": 64380 }, { "epoch": 1.0535874989773377, "grad_norm": 0.0015681232325732708, "learning_rate": 9.208526002277507e-06, "loss": 0.0016, "step": 64390 }, { "epoch": 1.0537511249284137, "grad_norm": 0.28409543633461, "learning_rate": 9.208140469530463e-06, "loss": 0.0022, "step": 64400 }, { "epoch": 1.0539147508794895, "grad_norm": 0.17556829750537872, "learning_rate": 9.207754850983014e-06, "loss": 0.0028, "step": 64410 }, { "epoch": 1.0540783768305653, "grad_norm": 0.20991091430187225, "learning_rate": 9.207369146643022e-06, "loss": 0.003, "step": 64420 }, { "epoch": 1.0542420027816413, "grad_norm": 0.08890116959810257, "learning_rate": 9.206983356518351e-06, "loss": 0.0019, "step": 64430 }, { "epoch": 1.054405628732717, "grad_norm": 0.08910475671291351, "learning_rate": 9.20659748061687e-06, "loss": 0.0016, "step": 64440 }, { "epoch": 1.0545692546837928, "grad_norm": 0.11096809059381485, "learning_rate": 9.206211518946441e-06, "loss": 0.0019, "step": 64450 }, { "epoch": 1.0547328806348686, "grad_norm": 0.1409253180027008, "learning_rate": 9.205825471514937e-06, "loss": 0.002, "step": 64460 }, { "epoch": 1.0548965065859446, "grad_norm": 0.0714377611875534, "learning_rate": 9.20543933833023e-06, "loss": 0.0018, "step": 64470 }, { "epoch": 1.0550601325370204, "grad_norm": 0.011670861393213272, "learning_rate": 9.20505311940019e-06, "loss": 0.0021, "step": 64480 }, { "epoch": 1.0552237584880961, "grad_norm": 0.1362626552581787, "learning_rate": 9.204666814732693e-06, "loss": 0.0013, "step": 64490 }, { "epoch": 1.0553873844391721, "grad_norm": 0.08568049967288971, "learning_rate": 9.204280424335615e-06, "loss": 0.0017, "step": 64500 }, { "epoch": 1.055551010390248, "grad_norm": 0.03760910779237747, "learning_rate": 9.203893948216836e-06, "loss": 0.0013, "step": 64510 }, { "epoch": 1.0557146363413237, "grad_norm": 0.04667765274643898, "learning_rate": 9.203507386384233e-06, "loss": 0.0035, "step": 64520 }, { "epoch": 1.0558782622923997, "grad_norm": 0.035437945276498795, "learning_rate": 9.203120738845693e-06, "loss": 0.002, "step": 64530 }, { "epoch": 1.0560418882434754, "grad_norm": 0.05943673476576805, "learning_rate": 9.202734005609092e-06, "loss": 0.0018, "step": 64540 }, { "epoch": 1.0562055141945512, "grad_norm": 0.0892714262008667, "learning_rate": 9.202347186682321e-06, "loss": 0.0017, "step": 64550 }, { "epoch": 1.056369140145627, "grad_norm": 0.021090198308229446, "learning_rate": 9.201960282073264e-06, "loss": 0.0021, "step": 64560 }, { "epoch": 1.056532766096703, "grad_norm": 0.24896621704101562, "learning_rate": 9.20157329178981e-06, "loss": 0.0023, "step": 64570 }, { "epoch": 1.0566963920477788, "grad_norm": 0.06578769534826279, "learning_rate": 9.20118621583985e-06, "loss": 0.0018, "step": 64580 }, { "epoch": 1.0568600179988545, "grad_norm": 0.10241097211837769, "learning_rate": 9.200799054231278e-06, "loss": 0.0019, "step": 64590 }, { "epoch": 1.0570236439499305, "grad_norm": 0.03811495006084442, "learning_rate": 9.200411806971985e-06, "loss": 0.0016, "step": 64600 }, { "epoch": 1.0571872699010063, "grad_norm": 0.08029206097126007, "learning_rate": 9.200024474069867e-06, "loss": 0.0014, "step": 64610 }, { "epoch": 1.057350895852082, "grad_norm": 0.06294998526573181, "learning_rate": 9.199637055532822e-06, "loss": 0.0021, "step": 64620 }, { "epoch": 1.057514521803158, "grad_norm": 0.029218073934316635, "learning_rate": 9.199249551368749e-06, "loss": 0.0025, "step": 64630 }, { "epoch": 1.0576781477542339, "grad_norm": 0.10057835280895233, "learning_rate": 9.198861961585548e-06, "loss": 0.0019, "step": 64640 }, { "epoch": 1.0578417737053096, "grad_norm": 0.03491555526852608, "learning_rate": 9.198474286191124e-06, "loss": 0.0019, "step": 64650 }, { "epoch": 1.0580053996563854, "grad_norm": 0.10534986108541489, "learning_rate": 9.198086525193381e-06, "loss": 0.0018, "step": 64660 }, { "epoch": 1.0581690256074614, "grad_norm": 0.054539408534765244, "learning_rate": 9.197698678600222e-06, "loss": 0.0016, "step": 64670 }, { "epoch": 1.0583326515585372, "grad_norm": 0.02798079513013363, "learning_rate": 9.197310746419558e-06, "loss": 0.002, "step": 64680 }, { "epoch": 1.058496277509613, "grad_norm": 0.12975333631038666, "learning_rate": 9.196922728659297e-06, "loss": 0.002, "step": 64690 }, { "epoch": 1.058659903460689, "grad_norm": 0.15665875375270844, "learning_rate": 9.19653462532735e-06, "loss": 0.0017, "step": 64700 }, { "epoch": 1.0588235294117647, "grad_norm": 0.03488144651055336, "learning_rate": 9.196146436431635e-06, "loss": 0.0014, "step": 64710 }, { "epoch": 1.0589871553628405, "grad_norm": 0.07134082168340683, "learning_rate": 9.19575816198006e-06, "loss": 0.0017, "step": 64720 }, { "epoch": 1.0591507813139165, "grad_norm": 0.05256078392267227, "learning_rate": 9.195369801980544e-06, "loss": 0.0014, "step": 64730 }, { "epoch": 1.0593144072649923, "grad_norm": 0.05586503446102142, "learning_rate": 9.194981356441006e-06, "loss": 0.0016, "step": 64740 }, { "epoch": 1.059478033216068, "grad_norm": 0.08956608921289444, "learning_rate": 9.194592825369366e-06, "loss": 0.003, "step": 64750 }, { "epoch": 1.0596416591671438, "grad_norm": 0.06286794692277908, "learning_rate": 9.194204208773547e-06, "loss": 0.0024, "step": 64760 }, { "epoch": 1.0598052851182198, "grad_norm": 0.05365390703082085, "learning_rate": 9.19381550666147e-06, "loss": 0.0017, "step": 64770 }, { "epoch": 1.0599689110692956, "grad_norm": 0.0074942223727703094, "learning_rate": 9.193426719041062e-06, "loss": 0.0025, "step": 64780 }, { "epoch": 1.0601325370203714, "grad_norm": 0.11494365334510803, "learning_rate": 9.193037845920249e-06, "loss": 0.0016, "step": 64790 }, { "epoch": 1.0602961629714474, "grad_norm": 0.024462318047881126, "learning_rate": 9.19264888730696e-06, "loss": 0.002, "step": 64800 }, { "epoch": 1.0604597889225231, "grad_norm": 0.035855188965797424, "learning_rate": 9.192259843209127e-06, "loss": 0.0025, "step": 64810 }, { "epoch": 1.060623414873599, "grad_norm": 0.0403343066573143, "learning_rate": 9.191870713634681e-06, "loss": 0.0034, "step": 64820 }, { "epoch": 1.060787040824675, "grad_norm": 0.09941089898347855, "learning_rate": 9.191481498591556e-06, "loss": 0.0018, "step": 64830 }, { "epoch": 1.0609506667757507, "grad_norm": 0.10645300149917603, "learning_rate": 9.191092198087688e-06, "loss": 0.0025, "step": 64840 }, { "epoch": 1.0611142927268264, "grad_norm": 0.05186714604496956, "learning_rate": 9.190702812131014e-06, "loss": 0.0028, "step": 64850 }, { "epoch": 1.0612779186779022, "grad_norm": 0.035144079476594925, "learning_rate": 9.190313340729474e-06, "loss": 0.0027, "step": 64860 }, { "epoch": 1.0614415446289782, "grad_norm": 0.04904953017830849, "learning_rate": 9.189923783891008e-06, "loss": 0.0015, "step": 64870 }, { "epoch": 1.061605170580054, "grad_norm": 0.07883857935667038, "learning_rate": 9.189534141623562e-06, "loss": 0.0013, "step": 64880 }, { "epoch": 1.0617687965311298, "grad_norm": 0.07569791376590729, "learning_rate": 9.189144413935074e-06, "loss": 0.0024, "step": 64890 }, { "epoch": 1.0619324224822058, "grad_norm": 0.04225289821624756, "learning_rate": 9.188754600833499e-06, "loss": 0.0015, "step": 64900 }, { "epoch": 1.0620960484332815, "grad_norm": 0.12240641564130783, "learning_rate": 9.188364702326775e-06, "loss": 0.0018, "step": 64910 }, { "epoch": 1.0622596743843573, "grad_norm": 0.0471763052046299, "learning_rate": 9.18797471842286e-06, "loss": 0.0017, "step": 64920 }, { "epoch": 1.0624233003354333, "grad_norm": 0.08607795089483261, "learning_rate": 9.187584649129702e-06, "loss": 0.0015, "step": 64930 }, { "epoch": 1.062586926286509, "grad_norm": 0.1052476018667221, "learning_rate": 9.187194494455254e-06, "loss": 0.0015, "step": 64940 }, { "epoch": 1.0627505522375849, "grad_norm": 0.07677216082811356, "learning_rate": 9.186804254407471e-06, "loss": 0.003, "step": 64950 }, { "epoch": 1.0629141781886606, "grad_norm": 0.03735024854540825, "learning_rate": 9.18641392899431e-06, "loss": 0.0024, "step": 64960 }, { "epoch": 1.0630778041397366, "grad_norm": 0.023053670302033424, "learning_rate": 9.18602351822373e-06, "loss": 0.0016, "step": 64970 }, { "epoch": 1.0632414300908124, "grad_norm": 0.024158068001270294, "learning_rate": 9.18563302210369e-06, "loss": 0.0021, "step": 64980 }, { "epoch": 1.0634050560418882, "grad_norm": 0.052337970584630966, "learning_rate": 9.185242440642153e-06, "loss": 0.0016, "step": 64990 }, { "epoch": 1.0635686819929642, "grad_norm": 0.03951907530426979, "learning_rate": 9.184851773847081e-06, "loss": 0.0022, "step": 65000 }, { "epoch": 1.06373230794404, "grad_norm": 0.0614396370947361, "learning_rate": 9.18446102172644e-06, "loss": 0.0027, "step": 65010 }, { "epoch": 1.0638959338951157, "grad_norm": 0.0776737704873085, "learning_rate": 9.1840701842882e-06, "loss": 0.003, "step": 65020 }, { "epoch": 1.0640595598461915, "grad_norm": 0.03240140527486801, "learning_rate": 9.183679261540326e-06, "loss": 0.0025, "step": 65030 }, { "epoch": 1.0642231857972675, "grad_norm": 0.039718274027109146, "learning_rate": 9.183288253490789e-06, "loss": 0.0034, "step": 65040 }, { "epoch": 1.0643868117483433, "grad_norm": 0.0479300394654274, "learning_rate": 9.182897160147564e-06, "loss": 0.0017, "step": 65050 }, { "epoch": 1.064550437699419, "grad_norm": 0.04788938909769058, "learning_rate": 9.182505981518622e-06, "loss": 0.0019, "step": 65060 }, { "epoch": 1.064714063650495, "grad_norm": 0.06924169510602951, "learning_rate": 9.18211471761194e-06, "loss": 0.0017, "step": 65070 }, { "epoch": 1.0648776896015708, "grad_norm": 0.04578614607453346, "learning_rate": 9.181723368435498e-06, "loss": 0.0021, "step": 65080 }, { "epoch": 1.0650413155526466, "grad_norm": 0.008558901958167553, "learning_rate": 9.18133193399727e-06, "loss": 0.0013, "step": 65090 }, { "epoch": 1.0652049415037226, "grad_norm": 0.2057529091835022, "learning_rate": 9.180940414305241e-06, "loss": 0.0023, "step": 65100 }, { "epoch": 1.0653685674547984, "grad_norm": 0.5022710561752319, "learning_rate": 9.180548809367394e-06, "loss": 0.0013, "step": 65110 }, { "epoch": 1.0655321934058741, "grad_norm": 0.09540531784296036, "learning_rate": 9.180157119191711e-06, "loss": 0.0036, "step": 65120 }, { "epoch": 1.0656958193569501, "grad_norm": 0.05749184638261795, "learning_rate": 9.17976534378618e-06, "loss": 0.0014, "step": 65130 }, { "epoch": 1.065859445308026, "grad_norm": 0.09613946080207825, "learning_rate": 9.179373483158788e-06, "loss": 0.0027, "step": 65140 }, { "epoch": 1.0660230712591017, "grad_norm": 0.03415205702185631, "learning_rate": 9.178981537317525e-06, "loss": 0.0029, "step": 65150 }, { "epoch": 1.0661866972101774, "grad_norm": 0.07084658741950989, "learning_rate": 9.178589506270382e-06, "loss": 0.0019, "step": 65160 }, { "epoch": 1.0663503231612534, "grad_norm": 0.08740316331386566, "learning_rate": 9.178197390025355e-06, "loss": 0.0016, "step": 65170 }, { "epoch": 1.0665139491123292, "grad_norm": 0.1211305484175682, "learning_rate": 9.177805188590435e-06, "loss": 0.0038, "step": 65180 }, { "epoch": 1.066677575063405, "grad_norm": 0.041046131402254105, "learning_rate": 9.17741290197362e-06, "loss": 0.0021, "step": 65190 }, { "epoch": 1.066841201014481, "grad_norm": 0.10995245724916458, "learning_rate": 9.177020530182908e-06, "loss": 0.0022, "step": 65200 }, { "epoch": 1.0670048269655568, "grad_norm": 0.09473247081041336, "learning_rate": 9.176628073226301e-06, "loss": 0.0014, "step": 65210 }, { "epoch": 1.0671684529166325, "grad_norm": 0.01583273522555828, "learning_rate": 9.176235531111799e-06, "loss": 0.0024, "step": 65220 }, { "epoch": 1.0673320788677083, "grad_norm": 0.047702811658382416, "learning_rate": 9.175842903847407e-06, "loss": 0.0025, "step": 65230 }, { "epoch": 1.0674957048187843, "grad_norm": 0.1509535312652588, "learning_rate": 9.17545019144113e-06, "loss": 0.0015, "step": 65240 }, { "epoch": 1.06765933076986, "grad_norm": 0.006422937382012606, "learning_rate": 9.175057393900973e-06, "loss": 0.0029, "step": 65250 }, { "epoch": 1.0678229567209359, "grad_norm": 0.06888054311275482, "learning_rate": 9.174664511234947e-06, "loss": 0.0026, "step": 65260 }, { "epoch": 1.0679865826720119, "grad_norm": 0.10274622589349747, "learning_rate": 9.174271543451064e-06, "loss": 0.0021, "step": 65270 }, { "epoch": 1.0681502086230876, "grad_norm": 0.08183176815509796, "learning_rate": 9.173878490557332e-06, "loss": 0.0022, "step": 65280 }, { "epoch": 1.0683138345741634, "grad_norm": 0.12926161289215088, "learning_rate": 9.173485352561769e-06, "loss": 0.0024, "step": 65290 }, { "epoch": 1.0684774605252394, "grad_norm": 0.05401905998587608, "learning_rate": 9.173092129472388e-06, "loss": 0.0027, "step": 65300 }, { "epoch": 1.0686410864763152, "grad_norm": 0.10921086370944977, "learning_rate": 9.172698821297207e-06, "loss": 0.0025, "step": 65310 }, { "epoch": 1.068804712427391, "grad_norm": 0.0916113406419754, "learning_rate": 9.172305428044248e-06, "loss": 0.0022, "step": 65320 }, { "epoch": 1.068968338378467, "grad_norm": 0.3080599904060364, "learning_rate": 9.171911949721527e-06, "loss": 0.0049, "step": 65330 }, { "epoch": 1.0691319643295427, "grad_norm": 0.05102291330695152, "learning_rate": 9.171518386337073e-06, "loss": 0.0038, "step": 65340 }, { "epoch": 1.0692955902806185, "grad_norm": 0.020815949887037277, "learning_rate": 9.171124737898905e-06, "loss": 0.0013, "step": 65350 }, { "epoch": 1.0694592162316943, "grad_norm": 0.03056205064058304, "learning_rate": 9.17073100441505e-06, "loss": 0.0017, "step": 65360 }, { "epoch": 1.0696228421827703, "grad_norm": 0.07323649525642395, "learning_rate": 9.170337185893539e-06, "loss": 0.0021, "step": 65370 }, { "epoch": 1.069786468133846, "grad_norm": 0.0536668561398983, "learning_rate": 9.169943282342398e-06, "loss": 0.0018, "step": 65380 }, { "epoch": 1.0699500940849218, "grad_norm": 0.10528131574392319, "learning_rate": 9.169549293769659e-06, "loss": 0.0023, "step": 65390 }, { "epoch": 1.0701137200359978, "grad_norm": 0.048951808363199234, "learning_rate": 9.169155220183357e-06, "loss": 0.0023, "step": 65400 }, { "epoch": 1.0702773459870736, "grad_norm": 0.19506753981113434, "learning_rate": 9.168761061591527e-06, "loss": 0.0021, "step": 65410 }, { "epoch": 1.0704409719381494, "grad_norm": 0.0519748330116272, "learning_rate": 9.168366818002203e-06, "loss": 0.0039, "step": 65420 }, { "epoch": 1.0706045978892251, "grad_norm": 0.0930592492222786, "learning_rate": 9.167972489423424e-06, "loss": 0.0025, "step": 65430 }, { "epoch": 1.0707682238403011, "grad_norm": 0.19436515867710114, "learning_rate": 9.167578075863232e-06, "loss": 0.002, "step": 65440 }, { "epoch": 1.070931849791377, "grad_norm": 0.05903230607509613, "learning_rate": 9.167183577329667e-06, "loss": 0.0016, "step": 65450 }, { "epoch": 1.0710954757424527, "grad_norm": 0.10172824561595917, "learning_rate": 9.166788993830773e-06, "loss": 0.002, "step": 65460 }, { "epoch": 1.0712591016935287, "grad_norm": 0.09358508139848709, "learning_rate": 9.166394325374594e-06, "loss": 0.0031, "step": 65470 }, { "epoch": 1.0714227276446044, "grad_norm": 0.04700016975402832, "learning_rate": 9.16599957196918e-06, "loss": 0.0016, "step": 65480 }, { "epoch": 1.0715863535956802, "grad_norm": 0.031762704253196716, "learning_rate": 9.165604733622578e-06, "loss": 0.0018, "step": 65490 }, { "epoch": 1.0717499795467562, "grad_norm": 0.13624463975429535, "learning_rate": 9.165209810342835e-06, "loss": 0.0029, "step": 65500 }, { "epoch": 1.071913605497832, "grad_norm": 0.04966851696372032, "learning_rate": 9.16481480213801e-06, "loss": 0.0029, "step": 65510 }, { "epoch": 1.0720772314489078, "grad_norm": 0.10457579046487808, "learning_rate": 9.16441970901615e-06, "loss": 0.0028, "step": 65520 }, { "epoch": 1.0722408573999835, "grad_norm": 0.05201493576169014, "learning_rate": 9.164024530985317e-06, "loss": 0.0016, "step": 65530 }, { "epoch": 1.0724044833510595, "grad_norm": 0.432621568441391, "learning_rate": 9.163629268053564e-06, "loss": 0.0021, "step": 65540 }, { "epoch": 1.0725681093021353, "grad_norm": 0.23283736407756805, "learning_rate": 9.16323392022895e-06, "loss": 0.0042, "step": 65550 }, { "epoch": 1.072731735253211, "grad_norm": 0.14298714697360992, "learning_rate": 9.162838487519539e-06, "loss": 0.0027, "step": 65560 }, { "epoch": 1.072895361204287, "grad_norm": 0.016429085284471512, "learning_rate": 9.16244296993339e-06, "loss": 0.0025, "step": 65570 }, { "epoch": 1.0730589871553629, "grad_norm": 0.10656705498695374, "learning_rate": 9.16204736747857e-06, "loss": 0.0013, "step": 65580 }, { "epoch": 1.0732226131064386, "grad_norm": 0.10664746165275574, "learning_rate": 9.161651680163144e-06, "loss": 0.0024, "step": 65590 }, { "epoch": 1.0733862390575146, "grad_norm": 0.1944837123155594, "learning_rate": 9.161255907995177e-06, "loss": 0.0016, "step": 65600 }, { "epoch": 1.0735498650085904, "grad_norm": 0.07402072101831436, "learning_rate": 9.160860050982743e-06, "loss": 0.0015, "step": 65610 }, { "epoch": 1.0737134909596662, "grad_norm": 0.17404785752296448, "learning_rate": 9.160464109133913e-06, "loss": 0.0026, "step": 65620 }, { "epoch": 1.073877116910742, "grad_norm": 0.043403804302215576, "learning_rate": 9.160068082456756e-06, "loss": 0.0028, "step": 65630 }, { "epoch": 1.074040742861818, "grad_norm": 0.0551396980881691, "learning_rate": 9.15967197095935e-06, "loss": 0.0033, "step": 65640 }, { "epoch": 1.0742043688128937, "grad_norm": 0.14470207691192627, "learning_rate": 9.15927577464977e-06, "loss": 0.0021, "step": 65650 }, { "epoch": 1.0743679947639695, "grad_norm": 0.03053428791463375, "learning_rate": 9.158879493536092e-06, "loss": 0.002, "step": 65660 }, { "epoch": 1.0745316207150455, "grad_norm": 0.11044459044933319, "learning_rate": 9.1584831276264e-06, "loss": 0.0025, "step": 65670 }, { "epoch": 1.0746952466661213, "grad_norm": 0.10763964056968689, "learning_rate": 9.158086676928773e-06, "loss": 0.0025, "step": 65680 }, { "epoch": 1.074858872617197, "grad_norm": 0.013788175769150257, "learning_rate": 9.157690141451297e-06, "loss": 0.0033, "step": 65690 }, { "epoch": 1.075022498568273, "grad_norm": 0.1001296266913414, "learning_rate": 9.157293521202053e-06, "loss": 0.0027, "step": 65700 }, { "epoch": 1.0751861245193488, "grad_norm": 0.19084681570529938, "learning_rate": 9.15689681618913e-06, "loss": 0.0013, "step": 65710 }, { "epoch": 1.0753497504704246, "grad_norm": 0.08400560915470123, "learning_rate": 9.156500026420616e-06, "loss": 0.0018, "step": 65720 }, { "epoch": 1.0755133764215004, "grad_norm": 0.052613597363233566, "learning_rate": 9.156103151904604e-06, "loss": 0.0021, "step": 65730 }, { "epoch": 1.0756770023725764, "grad_norm": 0.0722840279340744, "learning_rate": 9.155706192649181e-06, "loss": 0.0016, "step": 65740 }, { "epoch": 1.0758406283236521, "grad_norm": 0.17563919723033905, "learning_rate": 9.155309148662444e-06, "loss": 0.0016, "step": 65750 }, { "epoch": 1.076004254274728, "grad_norm": 0.09707590937614441, "learning_rate": 9.154912019952485e-06, "loss": 0.0014, "step": 65760 }, { "epoch": 1.076167880225804, "grad_norm": 0.09502574801445007, "learning_rate": 9.154514806527407e-06, "loss": 0.0017, "step": 65770 }, { "epoch": 1.0763315061768797, "grad_norm": 0.0367140956223011, "learning_rate": 9.154117508395303e-06, "loss": 0.0025, "step": 65780 }, { "epoch": 1.0764951321279554, "grad_norm": 0.07694802433252335, "learning_rate": 9.153720125564276e-06, "loss": 0.0019, "step": 65790 }, { "epoch": 1.0766587580790312, "grad_norm": 0.07691600173711777, "learning_rate": 9.153322658042429e-06, "loss": 0.0023, "step": 65800 }, { "epoch": 1.0768223840301072, "grad_norm": 0.054841771721839905, "learning_rate": 9.152925105837865e-06, "loss": 0.0014, "step": 65810 }, { "epoch": 1.076986009981183, "grad_norm": 0.3002072274684906, "learning_rate": 9.152527468958692e-06, "loss": 0.0044, "step": 65820 }, { "epoch": 1.0771496359322588, "grad_norm": 0.06011604890227318, "learning_rate": 9.152129747413012e-06, "loss": 0.0041, "step": 65830 }, { "epoch": 1.0773132618833348, "grad_norm": 0.018235135823488235, "learning_rate": 9.15173194120894e-06, "loss": 0.0024, "step": 65840 }, { "epoch": 1.0774768878344105, "grad_norm": 0.20800943672657013, "learning_rate": 9.151334050354584e-06, "loss": 0.0026, "step": 65850 }, { "epoch": 1.0776405137854863, "grad_norm": 0.003250368405133486, "learning_rate": 9.150936074858057e-06, "loss": 0.0013, "step": 65860 }, { "epoch": 1.0778041397365623, "grad_norm": 0.02953527122735977, "learning_rate": 9.150538014727474e-06, "loss": 0.001, "step": 65870 }, { "epoch": 1.077967765687638, "grad_norm": 0.11347854137420654, "learning_rate": 9.150139869970951e-06, "loss": 0.0038, "step": 65880 }, { "epoch": 1.0781313916387139, "grad_norm": 0.13857150077819824, "learning_rate": 9.149741640596605e-06, "loss": 0.002, "step": 65890 }, { "epoch": 1.0782950175897899, "grad_norm": 0.057403963059186935, "learning_rate": 9.149343326612557e-06, "loss": 0.0025, "step": 65900 }, { "epoch": 1.0784586435408656, "grad_norm": 0.055724915117025375, "learning_rate": 9.148944928026929e-06, "loss": 0.0018, "step": 65910 }, { "epoch": 1.0786222694919414, "grad_norm": 0.08658299595117569, "learning_rate": 9.14854644484784e-06, "loss": 0.0024, "step": 65920 }, { "epoch": 1.0787858954430172, "grad_norm": 0.05451343581080437, "learning_rate": 9.148147877083418e-06, "loss": 0.002, "step": 65930 }, { "epoch": 1.0789495213940932, "grad_norm": 0.04430605471134186, "learning_rate": 9.147749224741788e-06, "loss": 0.0038, "step": 65940 }, { "epoch": 1.079113147345169, "grad_norm": 0.037793971598148346, "learning_rate": 9.147350487831079e-06, "loss": 0.002, "step": 65950 }, { "epoch": 1.0792767732962447, "grad_norm": 0.10595103353261948, "learning_rate": 9.14695166635942e-06, "loss": 0.0045, "step": 65960 }, { "epoch": 1.0794403992473207, "grad_norm": 0.24198050796985626, "learning_rate": 9.146552760334946e-06, "loss": 0.0016, "step": 65970 }, { "epoch": 1.0796040251983965, "grad_norm": 0.042309004813432693, "learning_rate": 9.146153769765786e-06, "loss": 0.0014, "step": 65980 }, { "epoch": 1.0797676511494723, "grad_norm": 0.06947043538093567, "learning_rate": 9.145754694660076e-06, "loss": 0.0023, "step": 65990 }, { "epoch": 1.079931277100548, "grad_norm": 0.003528889501467347, "learning_rate": 9.145355535025955e-06, "loss": 0.0014, "step": 66000 }, { "epoch": 1.080094903051624, "grad_norm": 0.0629240870475769, "learning_rate": 9.14495629087156e-06, "loss": 0.0014, "step": 66010 }, { "epoch": 1.0802585290026998, "grad_norm": 0.13467128574848175, "learning_rate": 9.144556962205032e-06, "loss": 0.0026, "step": 66020 }, { "epoch": 1.0804221549537756, "grad_norm": 0.06062241271138191, "learning_rate": 9.14415754903451e-06, "loss": 0.0018, "step": 66030 }, { "epoch": 1.0805857809048516, "grad_norm": 0.1288267970085144, "learning_rate": 9.14375805136814e-06, "loss": 0.0023, "step": 66040 }, { "epoch": 1.0807494068559274, "grad_norm": 0.058903057128190994, "learning_rate": 9.143358469214072e-06, "loss": 0.0017, "step": 66050 }, { "epoch": 1.0809130328070031, "grad_norm": 0.07259523123502731, "learning_rate": 9.142958802580445e-06, "loss": 0.0013, "step": 66060 }, { "epoch": 1.0810766587580791, "grad_norm": 0.033370599150657654, "learning_rate": 9.142559051475413e-06, "loss": 0.0023, "step": 66070 }, { "epoch": 1.081240284709155, "grad_norm": 0.05820600315928459, "learning_rate": 9.142159215907123e-06, "loss": 0.0022, "step": 66080 }, { "epoch": 1.0814039106602307, "grad_norm": 0.13214093446731567, "learning_rate": 9.14175929588373e-06, "loss": 0.0036, "step": 66090 }, { "epoch": 1.0815675366113067, "grad_norm": 0.18257379531860352, "learning_rate": 9.14135929141339e-06, "loss": 0.0021, "step": 66100 }, { "epoch": 1.0817311625623824, "grad_norm": 0.038764357566833496, "learning_rate": 9.140959202504253e-06, "loss": 0.0018, "step": 66110 }, { "epoch": 1.0818947885134582, "grad_norm": 0.13342708349227905, "learning_rate": 9.140559029164479e-06, "loss": 0.0047, "step": 66120 }, { "epoch": 1.082058414464534, "grad_norm": 1.077258825302124, "learning_rate": 9.14015877140223e-06, "loss": 0.0017, "step": 66130 }, { "epoch": 1.08222204041561, "grad_norm": 0.12104656547307968, "learning_rate": 9.139758429225664e-06, "loss": 0.0016, "step": 66140 }, { "epoch": 1.0823856663666858, "grad_norm": 0.11780006438493729, "learning_rate": 9.139358002642945e-06, "loss": 0.0026, "step": 66150 }, { "epoch": 1.0825492923177615, "grad_norm": 0.28931817412376404, "learning_rate": 9.138957491662237e-06, "loss": 0.0037, "step": 66160 }, { "epoch": 1.0827129182688375, "grad_norm": 0.028573540970683098, "learning_rate": 9.138556896291705e-06, "loss": 0.0017, "step": 66170 }, { "epoch": 1.0828765442199133, "grad_norm": 0.10498062521219254, "learning_rate": 9.138156216539517e-06, "loss": 0.0033, "step": 66180 }, { "epoch": 1.083040170170989, "grad_norm": 0.015286959707736969, "learning_rate": 9.137755452413844e-06, "loss": 0.0024, "step": 66190 }, { "epoch": 1.0832037961220649, "grad_norm": 0.07273107767105103, "learning_rate": 9.137354603922857e-06, "loss": 0.0024, "step": 66200 }, { "epoch": 1.0833674220731409, "grad_norm": 0.0994812473654747, "learning_rate": 9.136953671074728e-06, "loss": 0.0022, "step": 66210 }, { "epoch": 1.0835310480242166, "grad_norm": 0.08603793382644653, "learning_rate": 9.136552653877631e-06, "loss": 0.0019, "step": 66220 }, { "epoch": 1.0836946739752924, "grad_norm": 0.16736677289009094, "learning_rate": 9.136151552339744e-06, "loss": 0.0022, "step": 66230 }, { "epoch": 1.0838582999263684, "grad_norm": 0.05801735073328018, "learning_rate": 9.135750366469245e-06, "loss": 0.0011, "step": 66240 }, { "epoch": 1.0840219258774442, "grad_norm": 0.22713062167167664, "learning_rate": 9.135349096274311e-06, "loss": 0.0021, "step": 66250 }, { "epoch": 1.08418555182852, "grad_norm": 0.10425695776939392, "learning_rate": 9.13494774176313e-06, "loss": 0.0023, "step": 66260 }, { "epoch": 1.084349177779596, "grad_norm": 0.10229244828224182, "learning_rate": 9.134546302943878e-06, "loss": 0.0019, "step": 66270 }, { "epoch": 1.0845128037306717, "grad_norm": 0.04463769495487213, "learning_rate": 9.134144779824744e-06, "loss": 0.0017, "step": 66280 }, { "epoch": 1.0846764296817475, "grad_norm": 0.009758549742400646, "learning_rate": 9.133743172413913e-06, "loss": 0.0026, "step": 66290 }, { "epoch": 1.0848400556328233, "grad_norm": 0.09451769292354584, "learning_rate": 9.133341480719575e-06, "loss": 0.0021, "step": 66300 }, { "epoch": 1.0850036815838993, "grad_norm": 0.0534018874168396, "learning_rate": 9.132939704749922e-06, "loss": 0.0016, "step": 66310 }, { "epoch": 1.085167307534975, "grad_norm": 0.22915251553058624, "learning_rate": 9.13253784451314e-06, "loss": 0.0048, "step": 66320 }, { "epoch": 1.0853309334860508, "grad_norm": 0.13512344658374786, "learning_rate": 9.132135900017427e-06, "loss": 0.0016, "step": 66330 }, { "epoch": 1.0854945594371268, "grad_norm": 0.07347933948040009, "learning_rate": 9.131733871270978e-06, "loss": 0.002, "step": 66340 }, { "epoch": 1.0856581853882026, "grad_norm": 0.0960719883441925, "learning_rate": 9.131331758281989e-06, "loss": 0.0022, "step": 66350 }, { "epoch": 1.0858218113392784, "grad_norm": 0.18109636008739471, "learning_rate": 9.130929561058658e-06, "loss": 0.0017, "step": 66360 }, { "epoch": 1.0859854372903543, "grad_norm": 0.040295667946338654, "learning_rate": 9.130527279609187e-06, "loss": 0.0036, "step": 66370 }, { "epoch": 1.0861490632414301, "grad_norm": 0.23843106627464294, "learning_rate": 9.130124913941779e-06, "loss": 0.0017, "step": 66380 }, { "epoch": 1.086312689192506, "grad_norm": 0.08680379390716553, "learning_rate": 9.129722464064634e-06, "loss": 0.0023, "step": 66390 }, { "epoch": 1.0864763151435817, "grad_norm": 0.09031941741704941, "learning_rate": 9.129319929985963e-06, "loss": 0.0016, "step": 66400 }, { "epoch": 1.0866399410946577, "grad_norm": 0.08294931054115295, "learning_rate": 9.128917311713967e-06, "loss": 0.0038, "step": 66410 }, { "epoch": 1.0868035670457334, "grad_norm": 0.023359816521406174, "learning_rate": 9.128514609256863e-06, "loss": 0.0031, "step": 66420 }, { "epoch": 1.0869671929968092, "grad_norm": 0.0819118320941925, "learning_rate": 9.128111822622854e-06, "loss": 0.0018, "step": 66430 }, { "epoch": 1.0871308189478852, "grad_norm": 0.10472564399242401, "learning_rate": 9.127708951820156e-06, "loss": 0.0025, "step": 66440 }, { "epoch": 1.087294444898961, "grad_norm": 0.03328560292720795, "learning_rate": 9.127305996856984e-06, "loss": 0.0024, "step": 66450 }, { "epoch": 1.0874580708500368, "grad_norm": 0.003478611819446087, "learning_rate": 9.126902957741552e-06, "loss": 0.0031, "step": 66460 }, { "epoch": 1.0876216968011128, "grad_norm": 0.06874654442071915, "learning_rate": 9.12649983448208e-06, "loss": 0.0025, "step": 66470 }, { "epoch": 1.0877853227521885, "grad_norm": 0.09478245675563812, "learning_rate": 9.126096627086785e-06, "loss": 0.0027, "step": 66480 }, { "epoch": 1.0879489487032643, "grad_norm": 0.07507172971963882, "learning_rate": 9.125693335563888e-06, "loss": 0.0021, "step": 66490 }, { "epoch": 1.08811257465434, "grad_norm": 0.06820246577262878, "learning_rate": 9.125289959921613e-06, "loss": 0.0019, "step": 66500 }, { "epoch": 1.088276200605416, "grad_norm": 0.037011753767728806, "learning_rate": 9.124886500168185e-06, "loss": 0.0023, "step": 66510 }, { "epoch": 1.0884398265564919, "grad_norm": 0.16354656219482422, "learning_rate": 9.124482956311828e-06, "loss": 0.0016, "step": 66520 }, { "epoch": 1.0886034525075676, "grad_norm": 0.06105493754148483, "learning_rate": 9.124079328360773e-06, "loss": 0.0019, "step": 66530 }, { "epoch": 1.0887670784586436, "grad_norm": 0.047013044357299805, "learning_rate": 9.123675616323246e-06, "loss": 0.0021, "step": 66540 }, { "epoch": 1.0889307044097194, "grad_norm": 0.08459862321615219, "learning_rate": 9.123271820207482e-06, "loss": 0.0015, "step": 66550 }, { "epoch": 1.0890943303607952, "grad_norm": 0.5100809931755066, "learning_rate": 9.122867940021712e-06, "loss": 0.0018, "step": 66560 }, { "epoch": 1.089257956311871, "grad_norm": 0.04994954913854599, "learning_rate": 9.12246397577417e-06, "loss": 0.0021, "step": 66570 }, { "epoch": 1.089421582262947, "grad_norm": 0.1375192254781723, "learning_rate": 9.122059927473094e-06, "loss": 0.0021, "step": 66580 }, { "epoch": 1.0895852082140227, "grad_norm": 0.04234788566827774, "learning_rate": 9.121655795126722e-06, "loss": 0.0014, "step": 66590 }, { "epoch": 1.0897488341650985, "grad_norm": 0.10674771666526794, "learning_rate": 9.121251578743294e-06, "loss": 0.0012, "step": 66600 }, { "epoch": 1.0899124601161745, "grad_norm": 0.04080842807888985, "learning_rate": 9.120847278331052e-06, "loss": 0.0025, "step": 66610 }, { "epoch": 1.0900760860672503, "grad_norm": 0.2512199580669403, "learning_rate": 9.120442893898237e-06, "loss": 0.0027, "step": 66620 }, { "epoch": 1.090239712018326, "grad_norm": 0.14808011054992676, "learning_rate": 9.120038425453096e-06, "loss": 0.0024, "step": 66630 }, { "epoch": 1.090403337969402, "grad_norm": 0.12930940091609955, "learning_rate": 9.119633873003875e-06, "loss": 0.0016, "step": 66640 }, { "epoch": 1.0905669639204778, "grad_norm": 0.027368484064936638, "learning_rate": 9.119229236558825e-06, "loss": 0.001, "step": 66650 }, { "epoch": 1.0907305898715536, "grad_norm": 0.10323580354452133, "learning_rate": 9.118824516126193e-06, "loss": 0.0021, "step": 66660 }, { "epoch": 1.0908942158226296, "grad_norm": 0.04532657936215401, "learning_rate": 9.11841971171423e-06, "loss": 0.002, "step": 66670 }, { "epoch": 1.0910578417737054, "grad_norm": 0.027856936678290367, "learning_rate": 9.118014823331193e-06, "loss": 0.0021, "step": 66680 }, { "epoch": 1.0912214677247811, "grad_norm": 0.058790840208530426, "learning_rate": 9.117609850985339e-06, "loss": 0.0017, "step": 66690 }, { "epoch": 1.091385093675857, "grad_norm": 0.03354340046644211, "learning_rate": 9.11720479468492e-06, "loss": 0.0025, "step": 66700 }, { "epoch": 1.091548719626933, "grad_norm": 0.10839113593101501, "learning_rate": 9.116799654438195e-06, "loss": 0.0016, "step": 66710 }, { "epoch": 1.0917123455780087, "grad_norm": 0.04712298884987831, "learning_rate": 9.116394430253428e-06, "loss": 0.0019, "step": 66720 }, { "epoch": 1.0918759715290844, "grad_norm": 0.06545422226190567, "learning_rate": 9.11598912213888e-06, "loss": 0.0012, "step": 66730 }, { "epoch": 1.0920395974801604, "grad_norm": 0.17744997143745422, "learning_rate": 9.115583730102813e-06, "loss": 0.0059, "step": 66740 }, { "epoch": 1.0922032234312362, "grad_norm": 0.01855878345668316, "learning_rate": 9.115178254153493e-06, "loss": 0.0018, "step": 66750 }, { "epoch": 1.092366849382312, "grad_norm": 0.17380668222904205, "learning_rate": 9.11477269429919e-06, "loss": 0.0026, "step": 66760 }, { "epoch": 1.0925304753333878, "grad_norm": 0.09012824296951294, "learning_rate": 9.114367050548172e-06, "loss": 0.0044, "step": 66770 }, { "epoch": 1.0926941012844638, "grad_norm": 0.0729655772447586, "learning_rate": 9.113961322908707e-06, "loss": 0.0019, "step": 66780 }, { "epoch": 1.0928577272355395, "grad_norm": 0.024743784219026566, "learning_rate": 9.113555511389071e-06, "loss": 0.0039, "step": 66790 }, { "epoch": 1.0930213531866153, "grad_norm": 0.2562086284160614, "learning_rate": 9.113149615997535e-06, "loss": 0.0037, "step": 66800 }, { "epoch": 1.0931849791376913, "grad_norm": 0.3638673722743988, "learning_rate": 9.112743636742379e-06, "loss": 0.0015, "step": 66810 }, { "epoch": 1.093348605088767, "grad_norm": 0.03065389208495617, "learning_rate": 9.112337573631875e-06, "loss": 0.0012, "step": 66820 }, { "epoch": 1.0935122310398429, "grad_norm": 0.3780148923397064, "learning_rate": 9.111931426674307e-06, "loss": 0.0019, "step": 66830 }, { "epoch": 1.0936758569909188, "grad_norm": 0.014395899139344692, "learning_rate": 9.111525195877952e-06, "loss": 0.0021, "step": 66840 }, { "epoch": 1.0938394829419946, "grad_norm": 0.045905400067567825, "learning_rate": 9.111118881251097e-06, "loss": 0.0012, "step": 66850 }, { "epoch": 1.0940031088930704, "grad_norm": 0.07552381604909897, "learning_rate": 9.110712482802026e-06, "loss": 0.0016, "step": 66860 }, { "epoch": 1.0941667348441464, "grad_norm": 0.058833032846450806, "learning_rate": 9.110306000539023e-06, "loss": 0.0023, "step": 66870 }, { "epoch": 1.0943303607952222, "grad_norm": 0.08238431066274643, "learning_rate": 9.109899434470373e-06, "loss": 0.0036, "step": 66880 }, { "epoch": 1.094493986746298, "grad_norm": 0.26507651805877686, "learning_rate": 9.10949278460437e-06, "loss": 0.0028, "step": 66890 }, { "epoch": 1.0946576126973737, "grad_norm": 0.12748844921588898, "learning_rate": 9.109086050949307e-06, "loss": 0.0027, "step": 66900 }, { "epoch": 1.0948212386484497, "grad_norm": 0.04707646369934082, "learning_rate": 9.10867923351347e-06, "loss": 0.0018, "step": 66910 }, { "epoch": 1.0949848645995255, "grad_norm": 0.08428768813610077, "learning_rate": 9.108272332305161e-06, "loss": 0.0015, "step": 66920 }, { "epoch": 1.0951484905506013, "grad_norm": 0.06033562868833542, "learning_rate": 9.107865347332672e-06, "loss": 0.0012, "step": 66930 }, { "epoch": 1.0953121165016773, "grad_norm": 0.12761156260967255, "learning_rate": 9.107458278604302e-06, "loss": 0.0018, "step": 66940 }, { "epoch": 1.095475742452753, "grad_norm": 0.008496387861669064, "learning_rate": 9.107051126128353e-06, "loss": 0.0012, "step": 66950 }, { "epoch": 1.0956393684038288, "grad_norm": 0.05298171937465668, "learning_rate": 9.106643889913122e-06, "loss": 0.0037, "step": 66960 }, { "epoch": 1.0958029943549046, "grad_norm": 0.045827582478523254, "learning_rate": 9.106236569966915e-06, "loss": 0.002, "step": 66970 }, { "epoch": 1.0959666203059806, "grad_norm": 0.04425172880291939, "learning_rate": 9.105829166298037e-06, "loss": 0.0019, "step": 66980 }, { "epoch": 1.0961302462570564, "grad_norm": 0.04895619675517082, "learning_rate": 9.105421678914795e-06, "loss": 0.0021, "step": 66990 }, { "epoch": 1.0962938722081321, "grad_norm": 0.06994932889938354, "learning_rate": 9.105014107825493e-06, "loss": 0.0023, "step": 67000 }, { "epoch": 1.0964574981592081, "grad_norm": 0.06861771643161774, "learning_rate": 9.104606453038449e-06, "loss": 0.0027, "step": 67010 }, { "epoch": 1.096621124110284, "grad_norm": 0.108479805290699, "learning_rate": 9.104198714561968e-06, "loss": 0.0023, "step": 67020 }, { "epoch": 1.0967847500613597, "grad_norm": 0.05120735615491867, "learning_rate": 9.103790892404365e-06, "loss": 0.0012, "step": 67030 }, { "epoch": 1.0969483760124357, "grad_norm": 0.059930555522441864, "learning_rate": 9.103382986573957e-06, "loss": 0.0028, "step": 67040 }, { "epoch": 1.0971120019635114, "grad_norm": 0.058499131351709366, "learning_rate": 9.10297499707906e-06, "loss": 0.0018, "step": 67050 }, { "epoch": 1.0972756279145872, "grad_norm": 0.07416712492704391, "learning_rate": 9.102566923927991e-06, "loss": 0.0018, "step": 67060 }, { "epoch": 1.0974392538656632, "grad_norm": 0.03172357752919197, "learning_rate": 9.102158767129074e-06, "loss": 0.0012, "step": 67070 }, { "epoch": 1.097602879816739, "grad_norm": 0.022670412436127663, "learning_rate": 9.101750526690626e-06, "loss": 0.004, "step": 67080 }, { "epoch": 1.0977665057678148, "grad_norm": 0.11766385287046432, "learning_rate": 9.101342202620974e-06, "loss": 0.0022, "step": 67090 }, { "epoch": 1.0979301317188905, "grad_norm": 0.06059037521481514, "learning_rate": 9.100933794928442e-06, "loss": 0.0012, "step": 67100 }, { "epoch": 1.0980937576699665, "grad_norm": 0.04528335481882095, "learning_rate": 9.10052530362136e-06, "loss": 0.0023, "step": 67110 }, { "epoch": 1.0982573836210423, "grad_norm": 0.1356668472290039, "learning_rate": 9.100116728708052e-06, "loss": 0.0011, "step": 67120 }, { "epoch": 1.098421009572118, "grad_norm": 0.04218030348420143, "learning_rate": 9.099708070196851e-06, "loss": 0.0023, "step": 67130 }, { "epoch": 1.098584635523194, "grad_norm": 0.061127517372369766, "learning_rate": 9.09929932809609e-06, "loss": 0.0018, "step": 67140 }, { "epoch": 1.0987482614742698, "grad_norm": 0.17670772969722748, "learning_rate": 9.098890502414102e-06, "loss": 0.003, "step": 67150 }, { "epoch": 1.0989118874253456, "grad_norm": 0.07107468694448471, "learning_rate": 9.098481593159223e-06, "loss": 0.0011, "step": 67160 }, { "epoch": 1.0990755133764214, "grad_norm": 0.09310024976730347, "learning_rate": 9.098072600339789e-06, "loss": 0.0015, "step": 67170 }, { "epoch": 1.0992391393274974, "grad_norm": 0.0827503576874733, "learning_rate": 9.09766352396414e-06, "loss": 0.0018, "step": 67180 }, { "epoch": 1.0994027652785732, "grad_norm": 0.09629655629396439, "learning_rate": 9.097254364040618e-06, "loss": 0.0015, "step": 67190 }, { "epoch": 1.099566391229649, "grad_norm": 0.06339531391859055, "learning_rate": 9.096845120577561e-06, "loss": 0.0028, "step": 67200 }, { "epoch": 1.099730017180725, "grad_norm": 0.051723089069128036, "learning_rate": 9.096435793583319e-06, "loss": 0.0014, "step": 67210 }, { "epoch": 1.0998936431318007, "grad_norm": 0.039584483951330185, "learning_rate": 9.096026383066234e-06, "loss": 0.0061, "step": 67220 }, { "epoch": 1.1000572690828765, "grad_norm": 0.037146106362342834, "learning_rate": 9.095616889034655e-06, "loss": 0.0023, "step": 67230 }, { "epoch": 1.1002208950339525, "grad_norm": 0.07341740280389786, "learning_rate": 9.095207311496932e-06, "loss": 0.0025, "step": 67240 }, { "epoch": 1.1003845209850283, "grad_norm": 0.04803255945444107, "learning_rate": 9.094797650461413e-06, "loss": 0.0028, "step": 67250 }, { "epoch": 1.100548146936104, "grad_norm": 0.06549502909183502, "learning_rate": 9.094387905936452e-06, "loss": 0.0023, "step": 67260 }, { "epoch": 1.1007117728871798, "grad_norm": 0.1067865863442421, "learning_rate": 9.093978077930405e-06, "loss": 0.0027, "step": 67270 }, { "epoch": 1.1008753988382558, "grad_norm": 0.05328942835330963, "learning_rate": 9.093568166451627e-06, "loss": 0.002, "step": 67280 }, { "epoch": 1.1010390247893316, "grad_norm": 0.06453101336956024, "learning_rate": 9.093158171508476e-06, "loss": 0.0019, "step": 67290 }, { "epoch": 1.1012026507404074, "grad_norm": 0.10930044949054718, "learning_rate": 9.092748093109309e-06, "loss": 0.002, "step": 67300 }, { "epoch": 1.1013662766914833, "grad_norm": 0.040221624076366425, "learning_rate": 9.09233793126249e-06, "loss": 0.001, "step": 67310 }, { "epoch": 1.1015299026425591, "grad_norm": 0.05062701180577278, "learning_rate": 9.091927685976382e-06, "loss": 0.0016, "step": 67320 }, { "epoch": 1.101693528593635, "grad_norm": 0.07860197126865387, "learning_rate": 9.091517357259347e-06, "loss": 0.0014, "step": 67330 }, { "epoch": 1.101857154544711, "grad_norm": 0.16704130172729492, "learning_rate": 9.091106945119754e-06, "loss": 0.0035, "step": 67340 }, { "epoch": 1.1020207804957867, "grad_norm": 0.03720107302069664, "learning_rate": 9.09069644956597e-06, "loss": 0.0011, "step": 67350 }, { "epoch": 1.1021844064468624, "grad_norm": 0.06850151717662811, "learning_rate": 9.090285870606363e-06, "loss": 0.0013, "step": 67360 }, { "epoch": 1.1023480323979382, "grad_norm": 0.0615425668656826, "learning_rate": 9.089875208249307e-06, "loss": 0.0022, "step": 67370 }, { "epoch": 1.1025116583490142, "grad_norm": 0.01809445023536682, "learning_rate": 9.089464462503172e-06, "loss": 0.0013, "step": 67380 }, { "epoch": 1.10267528430009, "grad_norm": 0.004661364480853081, "learning_rate": 9.089053633376337e-06, "loss": 0.0018, "step": 67390 }, { "epoch": 1.1028389102511658, "grad_norm": 0.020533684641122818, "learning_rate": 9.088642720877174e-06, "loss": 0.002, "step": 67400 }, { "epoch": 1.1030025362022418, "grad_norm": 0.05973532050848007, "learning_rate": 9.088231725014065e-06, "loss": 0.0017, "step": 67410 }, { "epoch": 1.1031661621533175, "grad_norm": 0.028564734384417534, "learning_rate": 9.087820645795386e-06, "loss": 0.0014, "step": 67420 }, { "epoch": 1.1033297881043933, "grad_norm": 0.09811490029096603, "learning_rate": 9.087409483229523e-06, "loss": 0.0015, "step": 67430 }, { "epoch": 1.1034934140554693, "grad_norm": 0.14248549938201904, "learning_rate": 9.086998237324855e-06, "loss": 0.0025, "step": 67440 }, { "epoch": 1.103657040006545, "grad_norm": 0.1002650260925293, "learning_rate": 9.08658690808977e-06, "loss": 0.0022, "step": 67450 }, { "epoch": 1.1038206659576209, "grad_norm": 0.13227017223834991, "learning_rate": 9.086175495532653e-06, "loss": 0.0026, "step": 67460 }, { "epoch": 1.1039842919086966, "grad_norm": 0.07354418933391571, "learning_rate": 9.085763999661892e-06, "loss": 0.0017, "step": 67470 }, { "epoch": 1.1041479178597726, "grad_norm": 0.02558077685534954, "learning_rate": 9.085352420485878e-06, "loss": 0.002, "step": 67480 }, { "epoch": 1.1043115438108484, "grad_norm": 0.02402508072555065, "learning_rate": 9.084940758013003e-06, "loss": 0.0016, "step": 67490 }, { "epoch": 1.1044751697619242, "grad_norm": 0.08383668214082718, "learning_rate": 9.084529012251661e-06, "loss": 0.002, "step": 67500 }, { "epoch": 1.1046387957130002, "grad_norm": 0.08528129756450653, "learning_rate": 9.084117183210246e-06, "loss": 0.0021, "step": 67510 }, { "epoch": 1.104802421664076, "grad_norm": 0.11970506608486176, "learning_rate": 9.083705270897153e-06, "loss": 0.0018, "step": 67520 }, { "epoch": 1.1049660476151517, "grad_norm": 0.07433865964412689, "learning_rate": 9.083293275320784e-06, "loss": 0.003, "step": 67530 }, { "epoch": 1.1051296735662275, "grad_norm": 0.08556734025478363, "learning_rate": 9.082881196489538e-06, "loss": 0.0029, "step": 67540 }, { "epoch": 1.1052932995173035, "grad_norm": 0.08107417821884155, "learning_rate": 9.082469034411816e-06, "loss": 0.0033, "step": 67550 }, { "epoch": 1.1054569254683793, "grad_norm": 0.0543704554438591, "learning_rate": 9.082056789096024e-06, "loss": 0.0018, "step": 67560 }, { "epoch": 1.105620551419455, "grad_norm": 0.062426112592220306, "learning_rate": 9.081644460550565e-06, "loss": 0.0019, "step": 67570 }, { "epoch": 1.105784177370531, "grad_norm": 0.12093596160411835, "learning_rate": 9.081232048783847e-06, "loss": 0.0015, "step": 67580 }, { "epoch": 1.1059478033216068, "grad_norm": 0.09553118795156479, "learning_rate": 9.080819553804278e-06, "loss": 0.0017, "step": 67590 }, { "epoch": 1.1061114292726826, "grad_norm": 0.021638400852680206, "learning_rate": 9.080406975620269e-06, "loss": 0.0023, "step": 67600 }, { "epoch": 1.1062750552237586, "grad_norm": 0.013625558465719223, "learning_rate": 9.079994314240231e-06, "loss": 0.0014, "step": 67610 }, { "epoch": 1.1064386811748343, "grad_norm": 0.07296217978000641, "learning_rate": 9.07958156967258e-06, "loss": 0.0043, "step": 67620 }, { "epoch": 1.1066023071259101, "grad_norm": 0.009541323408484459, "learning_rate": 9.07916874192573e-06, "loss": 0.0013, "step": 67630 }, { "epoch": 1.1067659330769861, "grad_norm": 0.02636280655860901, "learning_rate": 9.078755831008099e-06, "loss": 0.0014, "step": 67640 }, { "epoch": 1.106929559028062, "grad_norm": 0.05964317545294762, "learning_rate": 9.078342836928106e-06, "loss": 0.0019, "step": 67650 }, { "epoch": 1.1070931849791377, "grad_norm": 0.11934079229831696, "learning_rate": 9.077929759694171e-06, "loss": 0.0019, "step": 67660 }, { "epoch": 1.1072568109302134, "grad_norm": 0.01914595440030098, "learning_rate": 9.077516599314716e-06, "loss": 0.0043, "step": 67670 }, { "epoch": 1.1074204368812894, "grad_norm": 0.10245729237794876, "learning_rate": 9.077103355798163e-06, "loss": 0.0019, "step": 67680 }, { "epoch": 1.1075840628323652, "grad_norm": 0.07164263725280762, "learning_rate": 9.076690029152943e-06, "loss": 0.0021, "step": 67690 }, { "epoch": 1.107747688783441, "grad_norm": 0.1140742227435112, "learning_rate": 9.076276619387478e-06, "loss": 0.0019, "step": 67700 }, { "epoch": 1.107911314734517, "grad_norm": 0.055138085037469864, "learning_rate": 9.075863126510203e-06, "loss": 0.0026, "step": 67710 }, { "epoch": 1.1080749406855928, "grad_norm": 0.06281735748052597, "learning_rate": 9.075449550529542e-06, "loss": 0.0024, "step": 67720 }, { "epoch": 1.1082385666366685, "grad_norm": 0.0019961982034146786, "learning_rate": 9.075035891453931e-06, "loss": 0.0022, "step": 67730 }, { "epoch": 1.1084021925877443, "grad_norm": 0.05027841031551361, "learning_rate": 9.074622149291805e-06, "loss": 0.0029, "step": 67740 }, { "epoch": 1.1085658185388203, "grad_norm": 0.0895039439201355, "learning_rate": 9.074208324051599e-06, "loss": 0.0018, "step": 67750 }, { "epoch": 1.108729444489896, "grad_norm": 0.15588754415512085, "learning_rate": 9.07379441574175e-06, "loss": 0.0019, "step": 67760 }, { "epoch": 1.1088930704409719, "grad_norm": 0.09427383542060852, "learning_rate": 9.073380424370697e-06, "loss": 0.0027, "step": 67770 }, { "epoch": 1.1090566963920478, "grad_norm": 0.05543957278132439, "learning_rate": 9.072966349946881e-06, "loss": 0.0013, "step": 67780 }, { "epoch": 1.1092203223431236, "grad_norm": 0.055623408406972885, "learning_rate": 9.072552192478746e-06, "loss": 0.0033, "step": 67790 }, { "epoch": 1.1093839482941994, "grad_norm": 0.03527195006608963, "learning_rate": 9.072137951974736e-06, "loss": 0.0035, "step": 67800 }, { "epoch": 1.1095475742452754, "grad_norm": 0.06955483555793762, "learning_rate": 9.071723628443294e-06, "loss": 0.002, "step": 67810 }, { "epoch": 1.1097112001963512, "grad_norm": 0.07761432230472565, "learning_rate": 9.071309221892873e-06, "loss": 0.0029, "step": 67820 }, { "epoch": 1.109874826147427, "grad_norm": 0.0907166376709938, "learning_rate": 9.070894732331916e-06, "loss": 0.0018, "step": 67830 }, { "epoch": 1.110038452098503, "grad_norm": 0.14481472969055176, "learning_rate": 9.07048015976888e-06, "loss": 0.0019, "step": 67840 }, { "epoch": 1.1102020780495787, "grad_norm": 0.11426294595003128, "learning_rate": 9.070065504212216e-06, "loss": 0.0023, "step": 67850 }, { "epoch": 1.1103657040006545, "grad_norm": 0.058134835213422775, "learning_rate": 9.069650765670376e-06, "loss": 0.0023, "step": 67860 }, { "epoch": 1.1105293299517303, "grad_norm": 0.14620710909366608, "learning_rate": 9.069235944151819e-06, "loss": 0.0021, "step": 67870 }, { "epoch": 1.1106929559028063, "grad_norm": 0.04197992756962776, "learning_rate": 9.068821039665e-06, "loss": 0.0017, "step": 67880 }, { "epoch": 1.110856581853882, "grad_norm": 0.1084236204624176, "learning_rate": 9.068406052218384e-06, "loss": 0.0012, "step": 67890 }, { "epoch": 1.1110202078049578, "grad_norm": 0.017584890127182007, "learning_rate": 9.067990981820427e-06, "loss": 0.002, "step": 67900 }, { "epoch": 1.1111838337560338, "grad_norm": 0.030538199469447136, "learning_rate": 9.067575828479593e-06, "loss": 0.0013, "step": 67910 }, { "epoch": 1.1113474597071096, "grad_norm": 0.019997116178274155, "learning_rate": 9.067160592204346e-06, "loss": 0.0015, "step": 67920 }, { "epoch": 1.1115110856581853, "grad_norm": 0.05408622324466705, "learning_rate": 9.066745273003156e-06, "loss": 0.0013, "step": 67930 }, { "epoch": 1.1116747116092611, "grad_norm": 0.056237753480672836, "learning_rate": 9.066329870884488e-06, "loss": 0.0023, "step": 67940 }, { "epoch": 1.1118383375603371, "grad_norm": 0.13349540531635284, "learning_rate": 9.065914385856812e-06, "loss": 0.0011, "step": 67950 }, { "epoch": 1.112001963511413, "grad_norm": 0.043433669954538345, "learning_rate": 9.0654988179286e-06, "loss": 0.0021, "step": 67960 }, { "epoch": 1.1121655894624887, "grad_norm": 0.09532980620861053, "learning_rate": 9.065083167108324e-06, "loss": 0.0025, "step": 67970 }, { "epoch": 1.1123292154135647, "grad_norm": 0.16649086773395538, "learning_rate": 9.06466743340446e-06, "loss": 0.0033, "step": 67980 }, { "epoch": 1.1124928413646404, "grad_norm": 0.062269847840070724, "learning_rate": 9.064251616825482e-06, "loss": 0.0023, "step": 67990 }, { "epoch": 1.1126564673157162, "grad_norm": 0.03728662058711052, "learning_rate": 9.063835717379872e-06, "loss": 0.0018, "step": 68000 }, { "epoch": 1.1128200932667922, "grad_norm": 0.03488551080226898, "learning_rate": 9.063419735076107e-06, "loss": 0.0026, "step": 68010 }, { "epoch": 1.112983719217868, "grad_norm": 0.026387721300125122, "learning_rate": 9.06300366992267e-06, "loss": 0.0038, "step": 68020 }, { "epoch": 1.1131473451689438, "grad_norm": 0.20112663507461548, "learning_rate": 9.062587521928044e-06, "loss": 0.0034, "step": 68030 }, { "epoch": 1.1133109711200195, "grad_norm": 0.01513766124844551, "learning_rate": 9.062171291100713e-06, "loss": 0.0029, "step": 68040 }, { "epoch": 1.1134745970710955, "grad_norm": 0.10005407780408859, "learning_rate": 9.061754977449164e-06, "loss": 0.0023, "step": 68050 }, { "epoch": 1.1136382230221713, "grad_norm": 0.039009131491184235, "learning_rate": 9.061338580981884e-06, "loss": 0.002, "step": 68060 }, { "epoch": 1.113801848973247, "grad_norm": 0.058031532913446426, "learning_rate": 9.060922101707366e-06, "loss": 0.002, "step": 68070 }, { "epoch": 1.113965474924323, "grad_norm": 0.08482014387845993, "learning_rate": 9.0605055396341e-06, "loss": 0.002, "step": 68080 }, { "epoch": 1.1141291008753988, "grad_norm": 0.09348995238542557, "learning_rate": 9.06008889477058e-06, "loss": 0.0023, "step": 68090 }, { "epoch": 1.1142927268264746, "grad_norm": 0.07829553633928299, "learning_rate": 9.0596721671253e-06, "loss": 0.0018, "step": 68100 }, { "epoch": 1.1144563527775506, "grad_norm": 0.05501865968108177, "learning_rate": 9.059255356706757e-06, "loss": 0.0023, "step": 68110 }, { "epoch": 1.1146199787286264, "grad_norm": 0.06477465480566025, "learning_rate": 9.05883846352345e-06, "loss": 0.0019, "step": 68120 }, { "epoch": 1.1147836046797022, "grad_norm": 0.22018127143383026, "learning_rate": 9.058421487583878e-06, "loss": 0.0015, "step": 68130 }, { "epoch": 1.114947230630778, "grad_norm": 0.06051616370677948, "learning_rate": 9.058004428896544e-06, "loss": 0.0028, "step": 68140 }, { "epoch": 1.115110856581854, "grad_norm": 0.06046310439705849, "learning_rate": 9.057587287469949e-06, "loss": 0.0021, "step": 68150 }, { "epoch": 1.1152744825329297, "grad_norm": 0.1066422238945961, "learning_rate": 9.0571700633126e-06, "loss": 0.0012, "step": 68160 }, { "epoch": 1.1154381084840055, "grad_norm": 0.03848449885845184, "learning_rate": 9.056752756433007e-06, "loss": 0.002, "step": 68170 }, { "epoch": 1.1156017344350815, "grad_norm": 0.05941805988550186, "learning_rate": 9.056335366839674e-06, "loss": 0.0015, "step": 68180 }, { "epoch": 1.1157653603861573, "grad_norm": 0.10824012011289597, "learning_rate": 9.055917894541112e-06, "loss": 0.0014, "step": 68190 }, { "epoch": 1.115928986337233, "grad_norm": 0.04185949265956879, "learning_rate": 9.055500339545834e-06, "loss": 0.003, "step": 68200 }, { "epoch": 1.116092612288309, "grad_norm": 0.07149162143468857, "learning_rate": 9.055082701862354e-06, "loss": 0.0014, "step": 68210 }, { "epoch": 1.1162562382393848, "grad_norm": 0.06450047343969345, "learning_rate": 9.054664981499185e-06, "loss": 0.0016, "step": 68220 }, { "epoch": 1.1164198641904606, "grad_norm": 0.13175524771213531, "learning_rate": 9.054247178464845e-06, "loss": 0.0026, "step": 68230 }, { "epoch": 1.1165834901415364, "grad_norm": 0.03608955070376396, "learning_rate": 9.053829292767855e-06, "loss": 0.0015, "step": 68240 }, { "epoch": 1.1167471160926123, "grad_norm": 0.053104981780052185, "learning_rate": 9.053411324416731e-06, "loss": 0.0014, "step": 68250 }, { "epoch": 1.1169107420436881, "grad_norm": 0.05327589809894562, "learning_rate": 9.05299327342e-06, "loss": 0.0023, "step": 68260 }, { "epoch": 1.117074367994764, "grad_norm": 0.1915101408958435, "learning_rate": 9.052575139786181e-06, "loss": 0.0035, "step": 68270 }, { "epoch": 1.11723799394584, "grad_norm": 0.09252609312534332, "learning_rate": 9.052156923523803e-06, "loss": 0.0018, "step": 68280 }, { "epoch": 1.1174016198969157, "grad_norm": 0.04214799404144287, "learning_rate": 9.051738624641392e-06, "loss": 0.0018, "step": 68290 }, { "epoch": 1.1175652458479914, "grad_norm": 0.08564850687980652, "learning_rate": 9.051320243147474e-06, "loss": 0.002, "step": 68300 }, { "epoch": 1.1177288717990674, "grad_norm": 0.038068123161792755, "learning_rate": 9.050901779050585e-06, "loss": 0.001, "step": 68310 }, { "epoch": 1.1178924977501432, "grad_norm": 0.09062056243419647, "learning_rate": 9.050483232359251e-06, "loss": 0.0031, "step": 68320 }, { "epoch": 1.118056123701219, "grad_norm": 0.10031469166278839, "learning_rate": 9.05006460308201e-06, "loss": 0.0024, "step": 68330 }, { "epoch": 1.1182197496522948, "grad_norm": 0.15015903115272522, "learning_rate": 9.049645891227395e-06, "loss": 0.0019, "step": 68340 }, { "epoch": 1.1183833756033708, "grad_norm": 0.06053030490875244, "learning_rate": 9.049227096803945e-06, "loss": 0.0017, "step": 68350 }, { "epoch": 1.1185470015544465, "grad_norm": 0.004713630303740501, "learning_rate": 9.0488082198202e-06, "loss": 0.0029, "step": 68360 }, { "epoch": 1.1187106275055223, "grad_norm": 0.010351118631660938, "learning_rate": 9.048389260284696e-06, "loss": 0.0022, "step": 68370 }, { "epoch": 1.1188742534565983, "grad_norm": 0.028277643024921417, "learning_rate": 9.04797021820598e-06, "loss": 0.002, "step": 68380 }, { "epoch": 1.119037879407674, "grad_norm": 0.0892530009150505, "learning_rate": 9.047551093592594e-06, "loss": 0.002, "step": 68390 }, { "epoch": 1.1192015053587498, "grad_norm": 0.03798588737845421, "learning_rate": 9.047131886453081e-06, "loss": 0.0019, "step": 68400 }, { "epoch": 1.1193651313098258, "grad_norm": 0.05072322487831116, "learning_rate": 9.046712596795995e-06, "loss": 0.0018, "step": 68410 }, { "epoch": 1.1195287572609016, "grad_norm": 0.03516717255115509, "learning_rate": 9.046293224629876e-06, "loss": 0.0025, "step": 68420 }, { "epoch": 1.1196923832119774, "grad_norm": 0.032822441309690475, "learning_rate": 9.045873769963283e-06, "loss": 0.0019, "step": 68430 }, { "epoch": 1.1198560091630532, "grad_norm": 0.01445015799254179, "learning_rate": 9.045454232804764e-06, "loss": 0.0017, "step": 68440 }, { "epoch": 1.1200196351141292, "grad_norm": 0.10774804651737213, "learning_rate": 9.045034613162873e-06, "loss": 0.0018, "step": 68450 }, { "epoch": 1.120183261065205, "grad_norm": 0.08143427222967148, "learning_rate": 9.044614911046166e-06, "loss": 0.0021, "step": 68460 }, { "epoch": 1.1203468870162807, "grad_norm": 0.025465724989771843, "learning_rate": 9.044195126463202e-06, "loss": 0.0013, "step": 68470 }, { "epoch": 1.1205105129673567, "grad_norm": 0.010437553748488426, "learning_rate": 9.043775259422539e-06, "loss": 0.0017, "step": 68480 }, { "epoch": 1.1206741389184325, "grad_norm": 0.1572045087814331, "learning_rate": 9.043355309932736e-06, "loss": 0.0027, "step": 68490 }, { "epoch": 1.1208377648695083, "grad_norm": 0.08204343169927597, "learning_rate": 9.042935278002356e-06, "loss": 0.0019, "step": 68500 }, { "epoch": 1.121001390820584, "grad_norm": 0.051611561328172684, "learning_rate": 9.042515163639967e-06, "loss": 0.0014, "step": 68510 }, { "epoch": 1.12116501677166, "grad_norm": 0.058183129876852036, "learning_rate": 9.042094966854131e-06, "loss": 0.0013, "step": 68520 }, { "epoch": 1.1213286427227358, "grad_norm": 0.04343927651643753, "learning_rate": 9.041674687653414e-06, "loss": 0.0028, "step": 68530 }, { "epoch": 1.1214922686738116, "grad_norm": 0.0842527374625206, "learning_rate": 9.041254326046392e-06, "loss": 0.0021, "step": 68540 }, { "epoch": 1.1216558946248876, "grad_norm": 0.1217217892408371, "learning_rate": 9.040833882041627e-06, "loss": 0.0022, "step": 68550 }, { "epoch": 1.1218195205759633, "grad_norm": 0.12471466511487961, "learning_rate": 9.040413355647697e-06, "loss": 0.0026, "step": 68560 }, { "epoch": 1.1219831465270391, "grad_norm": 0.14241379499435425, "learning_rate": 9.039992746873176e-06, "loss": 0.0029, "step": 68570 }, { "epoch": 1.1221467724781151, "grad_norm": 0.07791817933320999, "learning_rate": 9.039572055726637e-06, "loss": 0.0018, "step": 68580 }, { "epoch": 1.122310398429191, "grad_norm": 0.036834895610809326, "learning_rate": 9.039151282216662e-06, "loss": 0.0036, "step": 68590 }, { "epoch": 1.1224740243802667, "grad_norm": 0.025972994044423103, "learning_rate": 9.038730426351826e-06, "loss": 0.0032, "step": 68600 }, { "epoch": 1.1226376503313427, "grad_norm": 0.2536979615688324, "learning_rate": 9.038309488140711e-06, "loss": 0.0028, "step": 68610 }, { "epoch": 1.1228012762824184, "grad_norm": 0.010920795612037182, "learning_rate": 9.0378884675919e-06, "loss": 0.0014, "step": 68620 }, { "epoch": 1.1229649022334942, "grad_norm": 0.008452143520116806, "learning_rate": 9.037467364713979e-06, "loss": 0.0018, "step": 68630 }, { "epoch": 1.12312852818457, "grad_norm": 0.0325760655105114, "learning_rate": 9.037046179515529e-06, "loss": 0.0026, "step": 68640 }, { "epoch": 1.123292154135646, "grad_norm": 0.08989023417234421, "learning_rate": 9.036624912005143e-06, "loss": 0.0021, "step": 68650 }, { "epoch": 1.1234557800867218, "grad_norm": 0.10149700939655304, "learning_rate": 9.036203562191408e-06, "loss": 0.0023, "step": 68660 }, { "epoch": 1.1236194060377975, "grad_norm": 0.048537127673625946, "learning_rate": 9.035782130082915e-06, "loss": 0.0015, "step": 68670 }, { "epoch": 1.1237830319888735, "grad_norm": 0.036714013665914536, "learning_rate": 9.035360615688255e-06, "loss": 0.001, "step": 68680 }, { "epoch": 1.1239466579399493, "grad_norm": 0.08625059574842453, "learning_rate": 9.034939019016027e-06, "loss": 0.0022, "step": 68690 }, { "epoch": 1.124110283891025, "grad_norm": 0.07663793861865997, "learning_rate": 9.034517340074822e-06, "loss": 0.0015, "step": 68700 }, { "epoch": 1.1242739098421008, "grad_norm": 0.02308559976518154, "learning_rate": 9.034095578873241e-06, "loss": 0.0013, "step": 68710 }, { "epoch": 1.1244375357931768, "grad_norm": 0.10175775736570358, "learning_rate": 9.03367373541988e-06, "loss": 0.0016, "step": 68720 }, { "epoch": 1.1246011617442526, "grad_norm": 0.011367067694664001, "learning_rate": 9.033251809723343e-06, "loss": 0.002, "step": 68730 }, { "epoch": 1.1247647876953284, "grad_norm": 0.01438133791089058, "learning_rate": 9.032829801792232e-06, "loss": 0.0018, "step": 68740 }, { "epoch": 1.1249284136464044, "grad_norm": 0.02399609424173832, "learning_rate": 9.032407711635152e-06, "loss": 0.0021, "step": 68750 }, { "epoch": 1.1250920395974802, "grad_norm": 0.07371282577514648, "learning_rate": 9.031985539260705e-06, "loss": 0.0028, "step": 68760 }, { "epoch": 1.125255665548556, "grad_norm": 0.011086700484156609, "learning_rate": 9.031563284677505e-06, "loss": 0.0019, "step": 68770 }, { "epoch": 1.125419291499632, "grad_norm": 0.011606118641793728, "learning_rate": 9.031140947894158e-06, "loss": 0.0017, "step": 68780 }, { "epoch": 1.1255829174507077, "grad_norm": 0.03794405609369278, "learning_rate": 9.030718528919276e-06, "loss": 0.0019, "step": 68790 }, { "epoch": 1.1257465434017835, "grad_norm": 0.05241428688168526, "learning_rate": 9.030296027761469e-06, "loss": 0.0036, "step": 68800 }, { "epoch": 1.1259101693528595, "grad_norm": 0.005590530578047037, "learning_rate": 9.029873444429355e-06, "loss": 0.0014, "step": 68810 }, { "epoch": 1.1260737953039353, "grad_norm": 0.0673234760761261, "learning_rate": 9.029450778931548e-06, "loss": 0.0017, "step": 68820 }, { "epoch": 1.126237421255011, "grad_norm": 0.2979619801044464, "learning_rate": 9.029028031276667e-06, "loss": 0.0021, "step": 68830 }, { "epoch": 1.1264010472060868, "grad_norm": 0.07240944355726242, "learning_rate": 9.028605201473331e-06, "loss": 0.0033, "step": 68840 }, { "epoch": 1.1265646731571628, "grad_norm": 0.00596068287268281, "learning_rate": 9.028182289530161e-06, "loss": 0.0017, "step": 68850 }, { "epoch": 1.1267282991082386, "grad_norm": 0.06878530979156494, "learning_rate": 9.02775929545578e-06, "loss": 0.0022, "step": 68860 }, { "epoch": 1.1268919250593143, "grad_norm": 0.039097417145967484, "learning_rate": 9.027336219258813e-06, "loss": 0.0016, "step": 68870 }, { "epoch": 1.1270555510103903, "grad_norm": 0.06206772103905678, "learning_rate": 9.026913060947887e-06, "loss": 0.0025, "step": 68880 }, { "epoch": 1.1272191769614661, "grad_norm": 0.19856609404087067, "learning_rate": 9.026489820531625e-06, "loss": 0.0015, "step": 68890 }, { "epoch": 1.127382802912542, "grad_norm": 0.052789490669965744, "learning_rate": 9.026066498018661e-06, "loss": 0.0019, "step": 68900 }, { "epoch": 1.1275464288636177, "grad_norm": 0.14164948463439941, "learning_rate": 9.025643093417627e-06, "loss": 0.0009, "step": 68910 }, { "epoch": 1.1277100548146937, "grad_norm": 0.11931169033050537, "learning_rate": 9.025219606737152e-06, "loss": 0.0017, "step": 68920 }, { "epoch": 1.1278736807657694, "grad_norm": 0.018332097679376602, "learning_rate": 9.024796037985875e-06, "loss": 0.002, "step": 68930 }, { "epoch": 1.1280373067168452, "grad_norm": 0.03278995677828789, "learning_rate": 9.024372387172426e-06, "loss": 0.0034, "step": 68940 }, { "epoch": 1.1282009326679212, "grad_norm": 0.15212123095989227, "learning_rate": 9.02394865430545e-06, "loss": 0.0016, "step": 68950 }, { "epoch": 1.128364558618997, "grad_norm": 0.1343289166688919, "learning_rate": 9.023524839393582e-06, "loss": 0.0029, "step": 68960 }, { "epoch": 1.1285281845700728, "grad_norm": 0.08904068917036057, "learning_rate": 9.023100942445464e-06, "loss": 0.0019, "step": 68970 }, { "epoch": 1.1286918105211488, "grad_norm": 0.2950946092605591, "learning_rate": 9.02267696346974e-06, "loss": 0.0028, "step": 68980 }, { "epoch": 1.1288554364722245, "grad_norm": 0.07903938740491867, "learning_rate": 9.022252902475052e-06, "loss": 0.0021, "step": 68990 }, { "epoch": 1.1290190624233003, "grad_norm": 0.10009576380252838, "learning_rate": 9.021828759470051e-06, "loss": 0.0015, "step": 69000 }, { "epoch": 1.1291826883743763, "grad_norm": 0.09115635603666306, "learning_rate": 9.02140453446338e-06, "loss": 0.0017, "step": 69010 }, { "epoch": 1.129346314325452, "grad_norm": 0.046060625463724136, "learning_rate": 9.020980227463692e-06, "loss": 0.0038, "step": 69020 }, { "epoch": 1.1295099402765278, "grad_norm": 0.23631112277507782, "learning_rate": 9.020555838479637e-06, "loss": 0.0035, "step": 69030 }, { "epoch": 1.1296735662276036, "grad_norm": 0.02016650326550007, "learning_rate": 9.020131367519866e-06, "loss": 0.0021, "step": 69040 }, { "epoch": 1.1298371921786796, "grad_norm": 0.026171572506427765, "learning_rate": 9.019706814593036e-06, "loss": 0.0019, "step": 69050 }, { "epoch": 1.1300008181297554, "grad_norm": 0.05513225495815277, "learning_rate": 9.019282179707805e-06, "loss": 0.0021, "step": 69060 }, { "epoch": 1.1301644440808312, "grad_norm": 0.026335086673498154, "learning_rate": 9.018857462872826e-06, "loss": 0.0016, "step": 69070 }, { "epoch": 1.130328070031907, "grad_norm": 0.047275010496377945, "learning_rate": 9.01843266409676e-06, "loss": 0.0015, "step": 69080 }, { "epoch": 1.130491695982983, "grad_norm": 0.03642553091049194, "learning_rate": 9.018007783388274e-06, "loss": 0.0022, "step": 69090 }, { "epoch": 1.1306553219340587, "grad_norm": 0.04731664061546326, "learning_rate": 9.017582820756024e-06, "loss": 0.002, "step": 69100 }, { "epoch": 1.1308189478851345, "grad_norm": 0.06435723602771759, "learning_rate": 9.017157776208677e-06, "loss": 0.0015, "step": 69110 }, { "epoch": 1.1309825738362105, "grad_norm": 0.07111003249883652, "learning_rate": 9.016732649754898e-06, "loss": 0.0023, "step": 69120 }, { "epoch": 1.1311461997872863, "grad_norm": 0.06097150593996048, "learning_rate": 9.01630744140336e-06, "loss": 0.0026, "step": 69130 }, { "epoch": 1.131309825738362, "grad_norm": 0.04525141045451164, "learning_rate": 9.015882151162727e-06, "loss": 0.0023, "step": 69140 }, { "epoch": 1.131473451689438, "grad_norm": 0.11825767159461975, "learning_rate": 9.015456779041672e-06, "loss": 0.0024, "step": 69150 }, { "epoch": 1.1316370776405138, "grad_norm": 0.03325272351503372, "learning_rate": 9.015031325048869e-06, "loss": 0.0014, "step": 69160 }, { "epoch": 1.1318007035915896, "grad_norm": 0.3143913745880127, "learning_rate": 9.014605789192991e-06, "loss": 0.0019, "step": 69170 }, { "epoch": 1.1319643295426656, "grad_norm": 0.08585474640130997, "learning_rate": 9.014180171482716e-06, "loss": 0.002, "step": 69180 }, { "epoch": 1.1321279554937413, "grad_norm": 0.037785228341817856, "learning_rate": 9.013754471926721e-06, "loss": 0.0024, "step": 69190 }, { "epoch": 1.1322915814448171, "grad_norm": 0.03489150479435921, "learning_rate": 9.013328690533685e-06, "loss": 0.001, "step": 69200 }, { "epoch": 1.132455207395893, "grad_norm": 0.04126887395977974, "learning_rate": 9.01290282731229e-06, "loss": 0.0028, "step": 69210 }, { "epoch": 1.132618833346969, "grad_norm": 0.18100981414318085, "learning_rate": 9.012476882271222e-06, "loss": 0.0018, "step": 69220 }, { "epoch": 1.1327824592980447, "grad_norm": 0.08322653919458389, "learning_rate": 9.01205085541916e-06, "loss": 0.0018, "step": 69230 }, { "epoch": 1.1329460852491204, "grad_norm": 0.05225270986557007, "learning_rate": 9.011624746764796e-06, "loss": 0.0016, "step": 69240 }, { "epoch": 1.1331097112001964, "grad_norm": 0.04902074858546257, "learning_rate": 9.011198556316814e-06, "loss": 0.002, "step": 69250 }, { "epoch": 1.1332733371512722, "grad_norm": 0.0550420843064785, "learning_rate": 9.010772284083904e-06, "loss": 0.0017, "step": 69260 }, { "epoch": 1.133436963102348, "grad_norm": 0.1963515430688858, "learning_rate": 9.010345930074757e-06, "loss": 0.0033, "step": 69270 }, { "epoch": 1.1336005890534238, "grad_norm": 0.09935883432626724, "learning_rate": 9.009919494298069e-06, "loss": 0.0021, "step": 69280 }, { "epoch": 1.1337642150044998, "grad_norm": 0.07857081294059753, "learning_rate": 9.009492976762534e-06, "loss": 0.0011, "step": 69290 }, { "epoch": 1.1339278409555755, "grad_norm": 0.11996868252754211, "learning_rate": 9.009066377476845e-06, "loss": 0.0013, "step": 69300 }, { "epoch": 1.1340914669066513, "grad_norm": 0.039828330278396606, "learning_rate": 9.008639696449702e-06, "loss": 0.0016, "step": 69310 }, { "epoch": 1.1342550928577273, "grad_norm": 0.097042016685009, "learning_rate": 9.008212933689806e-06, "loss": 0.0019, "step": 69320 }, { "epoch": 1.134418718808803, "grad_norm": 0.08341521769762039, "learning_rate": 9.007786089205856e-06, "loss": 0.0017, "step": 69330 }, { "epoch": 1.1345823447598788, "grad_norm": 0.028665216639637947, "learning_rate": 9.007359163006558e-06, "loss": 0.0018, "step": 69340 }, { "epoch": 1.1347459707109548, "grad_norm": 0.051694076508283615, "learning_rate": 9.006932155100614e-06, "loss": 0.0019, "step": 69350 }, { "epoch": 1.1349095966620306, "grad_norm": 0.11546874791383743, "learning_rate": 9.00650506549673e-06, "loss": 0.0019, "step": 69360 }, { "epoch": 1.1350732226131064, "grad_norm": 0.10446613281965256, "learning_rate": 9.006077894203618e-06, "loss": 0.0018, "step": 69370 }, { "epoch": 1.1352368485641824, "grad_norm": 0.12076250463724136, "learning_rate": 9.005650641229981e-06, "loss": 0.0016, "step": 69380 }, { "epoch": 1.1354004745152582, "grad_norm": 0.06657516956329346, "learning_rate": 9.005223306584537e-06, "loss": 0.0026, "step": 69390 }, { "epoch": 1.135564100466334, "grad_norm": 0.030552657321095467, "learning_rate": 9.004795890275995e-06, "loss": 0.0024, "step": 69400 }, { "epoch": 1.1357277264174097, "grad_norm": 0.04191623255610466, "learning_rate": 9.00436839231307e-06, "loss": 0.0031, "step": 69410 }, { "epoch": 1.1358913523684857, "grad_norm": 0.5435457229614258, "learning_rate": 9.00394081270448e-06, "loss": 0.0015, "step": 69420 }, { "epoch": 1.1360549783195615, "grad_norm": 0.23892243206501007, "learning_rate": 9.003513151458942e-06, "loss": 0.0025, "step": 69430 }, { "epoch": 1.1362186042706373, "grad_norm": 0.10380707681179047, "learning_rate": 9.003085408585176e-06, "loss": 0.0013, "step": 69440 }, { "epoch": 1.1363822302217133, "grad_norm": 0.15260009467601776, "learning_rate": 9.002657584091903e-06, "loss": 0.0018, "step": 69450 }, { "epoch": 1.136545856172789, "grad_norm": 0.161994069814682, "learning_rate": 9.002229677987845e-06, "loss": 0.0029, "step": 69460 }, { "epoch": 1.1367094821238648, "grad_norm": 0.1761971116065979, "learning_rate": 9.001801690281728e-06, "loss": 0.002, "step": 69470 }, { "epoch": 1.1368731080749406, "grad_norm": 0.03291692957282066, "learning_rate": 9.001373620982279e-06, "loss": 0.0016, "step": 69480 }, { "epoch": 1.1370367340260166, "grad_norm": 0.08654295653104782, "learning_rate": 9.000945470098224e-06, "loss": 0.003, "step": 69490 }, { "epoch": 1.1372003599770923, "grad_norm": 0.15578487515449524, "learning_rate": 9.000517237638293e-06, "loss": 0.0023, "step": 69500 }, { "epoch": 1.1373639859281681, "grad_norm": 0.03235594183206558, "learning_rate": 9.000088923611218e-06, "loss": 0.002, "step": 69510 }, { "epoch": 1.1375276118792441, "grad_norm": 0.12350495159626007, "learning_rate": 8.999660528025733e-06, "loss": 0.0026, "step": 69520 }, { "epoch": 1.13769123783032, "grad_norm": 0.0053862594068050385, "learning_rate": 8.999232050890572e-06, "loss": 0.0023, "step": 69530 }, { "epoch": 1.1378548637813957, "grad_norm": 0.04569831117987633, "learning_rate": 8.998803492214468e-06, "loss": 0.0019, "step": 69540 }, { "epoch": 1.1380184897324717, "grad_norm": 0.014292234554886818, "learning_rate": 8.998374852006165e-06, "loss": 0.0015, "step": 69550 }, { "epoch": 1.1381821156835474, "grad_norm": 0.05549948662519455, "learning_rate": 8.997946130274396e-06, "loss": 0.0012, "step": 69560 }, { "epoch": 1.1383457416346232, "grad_norm": 0.026406576856970787, "learning_rate": 8.997517327027908e-06, "loss": 0.0018, "step": 69570 }, { "epoch": 1.1385093675856992, "grad_norm": 0.017586110159754753, "learning_rate": 8.997088442275439e-06, "loss": 0.0018, "step": 69580 }, { "epoch": 1.138672993536775, "grad_norm": 0.0482463464140892, "learning_rate": 8.996659476025738e-06, "loss": 0.0023, "step": 69590 }, { "epoch": 1.1388366194878508, "grad_norm": 0.19235548377037048, "learning_rate": 8.996230428287548e-06, "loss": 0.001, "step": 69600 }, { "epoch": 1.1390002454389265, "grad_norm": 0.021972933784127235, "learning_rate": 8.99580129906962e-06, "loss": 0.0012, "step": 69610 }, { "epoch": 1.1391638713900025, "grad_norm": 0.025957399979233742, "learning_rate": 8.9953720883807e-06, "loss": 0.0016, "step": 69620 }, { "epoch": 1.1393274973410783, "grad_norm": 0.07871576398611069, "learning_rate": 8.994942796229542e-06, "loss": 0.0023, "step": 69630 }, { "epoch": 1.139491123292154, "grad_norm": 0.1123921200633049, "learning_rate": 8.994513422624898e-06, "loss": 0.0022, "step": 69640 }, { "epoch": 1.13965474924323, "grad_norm": 0.046600133180618286, "learning_rate": 8.994083967575522e-06, "loss": 0.0025, "step": 69650 }, { "epoch": 1.1398183751943058, "grad_norm": 0.024364430457353592, "learning_rate": 8.99365443109017e-06, "loss": 0.0015, "step": 69660 }, { "epoch": 1.1399820011453816, "grad_norm": 0.07026126980781555, "learning_rate": 8.993224813177604e-06, "loss": 0.0022, "step": 69670 }, { "epoch": 1.1401456270964574, "grad_norm": 0.052533745765686035, "learning_rate": 8.992795113846577e-06, "loss": 0.0029, "step": 69680 }, { "epoch": 1.1403092530475334, "grad_norm": 0.12508723139762878, "learning_rate": 8.992365333105855e-06, "loss": 0.0026, "step": 69690 }, { "epoch": 1.1404728789986092, "grad_norm": 0.10562694072723389, "learning_rate": 8.991935470964198e-06, "loss": 0.0021, "step": 69700 }, { "epoch": 1.140636504949685, "grad_norm": 0.11979402601718903, "learning_rate": 8.991505527430373e-06, "loss": 0.0019, "step": 69710 }, { "epoch": 1.140800130900761, "grad_norm": 0.08316656947135925, "learning_rate": 8.991075502513144e-06, "loss": 0.0021, "step": 69720 }, { "epoch": 1.1409637568518367, "grad_norm": 0.04986121132969856, "learning_rate": 8.99064539622128e-06, "loss": 0.0017, "step": 69730 }, { "epoch": 1.1411273828029125, "grad_norm": 0.023363707587122917, "learning_rate": 8.990215208563549e-06, "loss": 0.0014, "step": 69740 }, { "epoch": 1.1412910087539885, "grad_norm": 0.2703987956047058, "learning_rate": 8.989784939548726e-06, "loss": 0.003, "step": 69750 }, { "epoch": 1.1414546347050643, "grad_norm": 0.05498642474412918, "learning_rate": 8.98935458918558e-06, "loss": 0.0018, "step": 69760 }, { "epoch": 1.14161826065614, "grad_norm": 0.27791208028793335, "learning_rate": 8.988924157482887e-06, "loss": 0.0021, "step": 69770 }, { "epoch": 1.141781886607216, "grad_norm": 0.029876932501792908, "learning_rate": 8.988493644449424e-06, "loss": 0.0018, "step": 69780 }, { "epoch": 1.1419455125582918, "grad_norm": 0.05560513958334923, "learning_rate": 8.988063050093966e-06, "loss": 0.002, "step": 69790 }, { "epoch": 1.1421091385093676, "grad_norm": 0.09476712346076965, "learning_rate": 8.987632374425293e-06, "loss": 0.0013, "step": 69800 }, { "epoch": 1.1422727644604433, "grad_norm": 0.028556346893310547, "learning_rate": 8.987201617452189e-06, "loss": 0.0024, "step": 69810 }, { "epoch": 1.1424363904115193, "grad_norm": 0.6656917333602905, "learning_rate": 8.986770779183434e-06, "loss": 0.0023, "step": 69820 }, { "epoch": 1.1426000163625951, "grad_norm": 0.17389634251594543, "learning_rate": 8.986339859627815e-06, "loss": 0.002, "step": 69830 }, { "epoch": 1.142763642313671, "grad_norm": 0.04621324688196182, "learning_rate": 8.985908858794115e-06, "loss": 0.0021, "step": 69840 }, { "epoch": 1.1429272682647467, "grad_norm": 0.10963451117277145, "learning_rate": 8.985477776691124e-06, "loss": 0.0018, "step": 69850 }, { "epoch": 1.1430908942158227, "grad_norm": 0.1114107072353363, "learning_rate": 8.985046613327631e-06, "loss": 0.0015, "step": 69860 }, { "epoch": 1.1432545201668984, "grad_norm": 0.030338114127516747, "learning_rate": 8.984615368712425e-06, "loss": 0.0023, "step": 69870 }, { "epoch": 1.1434181461179742, "grad_norm": 0.07339799404144287, "learning_rate": 8.984184042854303e-06, "loss": 0.0018, "step": 69880 }, { "epoch": 1.1435817720690502, "grad_norm": 0.06696785241365433, "learning_rate": 8.983752635762057e-06, "loss": 0.0013, "step": 69890 }, { "epoch": 1.143745398020126, "grad_norm": 0.014948206953704357, "learning_rate": 8.98332114744448e-06, "loss": 0.0024, "step": 69900 }, { "epoch": 1.1439090239712018, "grad_norm": 0.08619976788759232, "learning_rate": 8.982889577910375e-06, "loss": 0.0032, "step": 69910 }, { "epoch": 1.1440726499222778, "grad_norm": 0.04636203125119209, "learning_rate": 8.982457927168537e-06, "loss": 0.0019, "step": 69920 }, { "epoch": 1.1442362758733535, "grad_norm": 0.14979985356330872, "learning_rate": 8.98202619522777e-06, "loss": 0.0042, "step": 69930 }, { "epoch": 1.1443999018244293, "grad_norm": 0.03847116231918335, "learning_rate": 8.981594382096875e-06, "loss": 0.0012, "step": 69940 }, { "epoch": 1.1445635277755053, "grad_norm": 0.11544715613126755, "learning_rate": 8.981162487784655e-06, "loss": 0.0019, "step": 69950 }, { "epoch": 1.144727153726581, "grad_norm": 0.11712008714675903, "learning_rate": 8.980730512299922e-06, "loss": 0.0022, "step": 69960 }, { "epoch": 1.1448907796776568, "grad_norm": 0.17244578897953033, "learning_rate": 8.980298455651476e-06, "loss": 0.0031, "step": 69970 }, { "epoch": 1.1450544056287328, "grad_norm": 0.04855690523982048, "learning_rate": 8.97986631784813e-06, "loss": 0.0017, "step": 69980 }, { "epoch": 1.1452180315798086, "grad_norm": 0.07339038699865341, "learning_rate": 8.979434098898697e-06, "loss": 0.0026, "step": 69990 }, { "epoch": 1.1453816575308844, "grad_norm": 0.054577071219682693, "learning_rate": 8.979001798811984e-06, "loss": 0.0011, "step": 70000 }, { "epoch": 1.1455452834819602, "grad_norm": 0.1533423811197281, "learning_rate": 8.97856941759681e-06, "loss": 0.0021, "step": 70010 }, { "epoch": 1.1457089094330362, "grad_norm": 0.29975536465644836, "learning_rate": 8.978136955261989e-06, "loss": 0.0015, "step": 70020 }, { "epoch": 1.145872535384112, "grad_norm": 0.035029761493206024, "learning_rate": 8.977704411816338e-06, "loss": 0.0017, "step": 70030 }, { "epoch": 1.1460361613351877, "grad_norm": 0.10776318609714508, "learning_rate": 8.977271787268678e-06, "loss": 0.0018, "step": 70040 }, { "epoch": 1.1461997872862635, "grad_norm": 0.004722067154943943, "learning_rate": 8.976839081627827e-06, "loss": 0.0018, "step": 70050 }, { "epoch": 1.1463634132373395, "grad_norm": 0.0651974156498909, "learning_rate": 8.97640629490261e-06, "loss": 0.0019, "step": 70060 }, { "epoch": 1.1465270391884153, "grad_norm": 0.05316061154007912, "learning_rate": 8.975973427101852e-06, "loss": 0.002, "step": 70070 }, { "epoch": 1.146690665139491, "grad_norm": 0.17428499460220337, "learning_rate": 8.975540478234377e-06, "loss": 0.0041, "step": 70080 }, { "epoch": 1.146854291090567, "grad_norm": 0.03859271854162216, "learning_rate": 8.97510744830901e-06, "loss": 0.003, "step": 70090 }, { "epoch": 1.1470179170416428, "grad_norm": 0.07261243462562561, "learning_rate": 8.974674337334586e-06, "loss": 0.0017, "step": 70100 }, { "epoch": 1.1471815429927186, "grad_norm": 0.026973234489560127, "learning_rate": 8.97424114531993e-06, "loss": 0.0022, "step": 70110 }, { "epoch": 1.1473451689437946, "grad_norm": 0.06496724486351013, "learning_rate": 8.97380787227388e-06, "loss": 0.0024, "step": 70120 }, { "epoch": 1.1475087948948703, "grad_norm": 0.09366229176521301, "learning_rate": 8.973374518205265e-06, "loss": 0.0016, "step": 70130 }, { "epoch": 1.1476724208459461, "grad_norm": 0.2473803013563156, "learning_rate": 8.972941083122923e-06, "loss": 0.0025, "step": 70140 }, { "epoch": 1.1478360467970221, "grad_norm": 0.06202621012926102, "learning_rate": 8.972507567035693e-06, "loss": 0.0014, "step": 70150 }, { "epoch": 1.1479996727480979, "grad_norm": 0.046330273151397705, "learning_rate": 8.972073969952411e-06, "loss": 0.002, "step": 70160 }, { "epoch": 1.1481632986991737, "grad_norm": 0.05772656202316284, "learning_rate": 8.971640291881918e-06, "loss": 0.0019, "step": 70170 }, { "epoch": 1.1483269246502494, "grad_norm": 0.08591804653406143, "learning_rate": 8.971206532833058e-06, "loss": 0.0018, "step": 70180 }, { "epoch": 1.1484905506013254, "grad_norm": 0.051280196756124496, "learning_rate": 8.970772692814674e-06, "loss": 0.0022, "step": 70190 }, { "epoch": 1.1486541765524012, "grad_norm": 0.0070269787684082985, "learning_rate": 8.970338771835612e-06, "loss": 0.0033, "step": 70200 }, { "epoch": 1.148817802503477, "grad_norm": 0.32452720403671265, "learning_rate": 8.969904769904719e-06, "loss": 0.0026, "step": 70210 }, { "epoch": 1.148981428454553, "grad_norm": 0.07648895680904388, "learning_rate": 8.969470687030843e-06, "loss": 0.0044, "step": 70220 }, { "epoch": 1.1491450544056288, "grad_norm": 0.17687660455703735, "learning_rate": 8.969036523222836e-06, "loss": 0.0021, "step": 70230 }, { "epoch": 1.1493086803567045, "grad_norm": 0.011663327924907207, "learning_rate": 8.96860227848955e-06, "loss": 0.0014, "step": 70240 }, { "epoch": 1.1494723063077803, "grad_norm": 0.11932702362537384, "learning_rate": 8.96816795283984e-06, "loss": 0.0027, "step": 70250 }, { "epoch": 1.1496359322588563, "grad_norm": 0.11281036585569382, "learning_rate": 8.96773354628256e-06, "loss": 0.0023, "step": 70260 }, { "epoch": 1.149799558209932, "grad_norm": 0.09508690983057022, "learning_rate": 8.967299058826566e-06, "loss": 0.0014, "step": 70270 }, { "epoch": 1.1499631841610078, "grad_norm": 0.04007675126194954, "learning_rate": 8.96686449048072e-06, "loss": 0.0018, "step": 70280 }, { "epoch": 1.1501268101120838, "grad_norm": 0.06732998788356781, "learning_rate": 8.966429841253879e-06, "loss": 0.0025, "step": 70290 }, { "epoch": 1.1502904360631596, "grad_norm": 0.1163821741938591, "learning_rate": 8.965995111154909e-06, "loss": 0.0028, "step": 70300 }, { "epoch": 1.1504540620142354, "grad_norm": 0.06016070395708084, "learning_rate": 8.96556030019267e-06, "loss": 0.0016, "step": 70310 }, { "epoch": 1.1506176879653114, "grad_norm": 0.022495770826935768, "learning_rate": 8.965125408376029e-06, "loss": 0.0015, "step": 70320 }, { "epoch": 1.1507813139163872, "grad_norm": 0.1646103411912918, "learning_rate": 8.964690435713853e-06, "loss": 0.0033, "step": 70330 }, { "epoch": 1.150944939867463, "grad_norm": 0.021104231476783752, "learning_rate": 8.964255382215012e-06, "loss": 0.0025, "step": 70340 }, { "epoch": 1.151108565818539, "grad_norm": 0.03730098530650139, "learning_rate": 8.963820247888375e-06, "loss": 0.0029, "step": 70350 }, { "epoch": 1.1512721917696147, "grad_norm": 0.04141266271471977, "learning_rate": 8.963385032742813e-06, "loss": 0.002, "step": 70360 }, { "epoch": 1.1514358177206905, "grad_norm": 0.11374370008707047, "learning_rate": 8.962949736787202e-06, "loss": 0.0017, "step": 70370 }, { "epoch": 1.1515994436717663, "grad_norm": 0.07010037451982498, "learning_rate": 8.962514360030417e-06, "loss": 0.0018, "step": 70380 }, { "epoch": 1.1517630696228423, "grad_norm": 0.0327409952878952, "learning_rate": 8.962078902481333e-06, "loss": 0.0011, "step": 70390 }, { "epoch": 1.151926695573918, "grad_norm": 0.021570907905697823, "learning_rate": 8.961643364148829e-06, "loss": 0.0024, "step": 70400 }, { "epoch": 1.1520903215249938, "grad_norm": 0.020730119198560715, "learning_rate": 8.961207745041787e-06, "loss": 0.0026, "step": 70410 }, { "epoch": 1.1522539474760698, "grad_norm": 0.04103189706802368, "learning_rate": 8.960772045169088e-06, "loss": 0.0013, "step": 70420 }, { "epoch": 1.1524175734271456, "grad_norm": 0.21556104719638824, "learning_rate": 8.960336264539616e-06, "loss": 0.0018, "step": 70430 }, { "epoch": 1.1525811993782213, "grad_norm": 0.05582258850336075, "learning_rate": 8.959900403162255e-06, "loss": 0.0026, "step": 70440 }, { "epoch": 1.1527448253292971, "grad_norm": 0.25971922278404236, "learning_rate": 8.959464461045892e-06, "loss": 0.0028, "step": 70450 }, { "epoch": 1.1529084512803731, "grad_norm": 0.24248771369457245, "learning_rate": 8.959028438199417e-06, "loss": 0.0022, "step": 70460 }, { "epoch": 1.1530720772314489, "grad_norm": 0.1037585511803627, "learning_rate": 8.95859233463172e-06, "loss": 0.0024, "step": 70470 }, { "epoch": 1.1532357031825247, "grad_norm": 0.09529253095388412, "learning_rate": 8.958156150351692e-06, "loss": 0.002, "step": 70480 }, { "epoch": 1.1533993291336007, "grad_norm": 0.12183483690023422, "learning_rate": 8.957719885368223e-06, "loss": 0.0011, "step": 70490 }, { "epoch": 1.1535629550846764, "grad_norm": 0.1352684199810028, "learning_rate": 8.957283539690215e-06, "loss": 0.0023, "step": 70500 }, { "epoch": 1.1537265810357522, "grad_norm": 0.4691754877567291, "learning_rate": 8.95684711332656e-06, "loss": 0.0022, "step": 70510 }, { "epoch": 1.1538902069868282, "grad_norm": 0.08708100765943527, "learning_rate": 8.956410606286157e-06, "loss": 0.0022, "step": 70520 }, { "epoch": 1.154053832937904, "grad_norm": 0.05489550530910492, "learning_rate": 8.955974018577908e-06, "loss": 0.0017, "step": 70530 }, { "epoch": 1.1542174588889798, "grad_norm": 0.08116576820611954, "learning_rate": 8.955537350210712e-06, "loss": 0.002, "step": 70540 }, { "epoch": 1.1543810848400557, "grad_norm": 0.11777094751596451, "learning_rate": 8.955100601193474e-06, "loss": 0.0022, "step": 70550 }, { "epoch": 1.1545447107911315, "grad_norm": 0.09702542424201965, "learning_rate": 8.954663771535098e-06, "loss": 0.0013, "step": 70560 }, { "epoch": 1.1547083367422073, "grad_norm": 0.2508227527141571, "learning_rate": 8.95422686124449e-06, "loss": 0.0016, "step": 70570 }, { "epoch": 1.154871962693283, "grad_norm": 0.003688486060127616, "learning_rate": 8.953789870330562e-06, "loss": 0.0022, "step": 70580 }, { "epoch": 1.155035588644359, "grad_norm": 0.041653528809547424, "learning_rate": 8.953352798802219e-06, "loss": 0.0019, "step": 70590 }, { "epoch": 1.1551992145954348, "grad_norm": 0.031936876475811005, "learning_rate": 8.952915646668376e-06, "loss": 0.0025, "step": 70600 }, { "epoch": 1.1553628405465106, "grad_norm": 0.12265530973672867, "learning_rate": 8.952478413937945e-06, "loss": 0.0034, "step": 70610 }, { "epoch": 1.1555264664975866, "grad_norm": 0.3226751983165741, "learning_rate": 8.952041100619837e-06, "loss": 0.0026, "step": 70620 }, { "epoch": 1.1556900924486624, "grad_norm": 0.035071004182100296, "learning_rate": 8.951603706722975e-06, "loss": 0.0021, "step": 70630 }, { "epoch": 1.1558537183997382, "grad_norm": 0.2190142720937729, "learning_rate": 8.951166232256273e-06, "loss": 0.0008, "step": 70640 }, { "epoch": 1.156017344350814, "grad_norm": 0.018702268600463867, "learning_rate": 8.950728677228654e-06, "loss": 0.0018, "step": 70650 }, { "epoch": 1.15618097030189, "grad_norm": 0.054287899285554886, "learning_rate": 8.950291041649037e-06, "loss": 0.0026, "step": 70660 }, { "epoch": 1.1563445962529657, "grad_norm": 0.04658515006303787, "learning_rate": 8.949853325526343e-06, "loss": 0.001, "step": 70670 }, { "epoch": 1.1565082222040415, "grad_norm": 0.08386875689029694, "learning_rate": 8.9494155288695e-06, "loss": 0.0015, "step": 70680 }, { "epoch": 1.1566718481551175, "grad_norm": 0.018258800730109215, "learning_rate": 8.948977651687433e-06, "loss": 0.0022, "step": 70690 }, { "epoch": 1.1568354741061933, "grad_norm": 0.19330696761608124, "learning_rate": 8.948539693989072e-06, "loss": 0.0039, "step": 70700 }, { "epoch": 1.156999100057269, "grad_norm": 0.07797151058912277, "learning_rate": 8.948101655783343e-06, "loss": 0.0042, "step": 70710 }, { "epoch": 1.157162726008345, "grad_norm": 0.013780644163489342, "learning_rate": 8.947663537079178e-06, "loss": 0.0014, "step": 70720 }, { "epoch": 1.1573263519594208, "grad_norm": 0.24384978413581848, "learning_rate": 8.947225337885513e-06, "loss": 0.0023, "step": 70730 }, { "epoch": 1.1574899779104966, "grad_norm": 0.03840942308306694, "learning_rate": 8.946787058211279e-06, "loss": 0.0031, "step": 70740 }, { "epoch": 1.1576536038615726, "grad_norm": 0.0586140938103199, "learning_rate": 8.946348698065414e-06, "loss": 0.0018, "step": 70750 }, { "epoch": 1.1578172298126483, "grad_norm": 0.05156877264380455, "learning_rate": 8.945910257456856e-06, "loss": 0.0046, "step": 70760 }, { "epoch": 1.1579808557637241, "grad_norm": 0.04851439967751503, "learning_rate": 8.945471736394542e-06, "loss": 0.002, "step": 70770 }, { "epoch": 1.1581444817148, "grad_norm": 0.15728691220283508, "learning_rate": 8.945033134887417e-06, "loss": 0.0027, "step": 70780 }, { "epoch": 1.1583081076658759, "grad_norm": 0.0879538282752037, "learning_rate": 8.94459445294442e-06, "loss": 0.0032, "step": 70790 }, { "epoch": 1.1584717336169517, "grad_norm": 0.11153075844049454, "learning_rate": 8.944155690574497e-06, "loss": 0.0027, "step": 70800 }, { "epoch": 1.1586353595680274, "grad_norm": 0.06889285892248154, "learning_rate": 8.943716847786594e-06, "loss": 0.0023, "step": 70810 }, { "epoch": 1.1587989855191032, "grad_norm": 0.027969403192400932, "learning_rate": 8.943277924589659e-06, "loss": 0.001, "step": 70820 }, { "epoch": 1.1589626114701792, "grad_norm": 0.057150550186634064, "learning_rate": 8.94283892099264e-06, "loss": 0.0011, "step": 70830 }, { "epoch": 1.159126237421255, "grad_norm": 0.0767960324883461, "learning_rate": 8.942399837004489e-06, "loss": 0.0015, "step": 70840 }, { "epoch": 1.1592898633723308, "grad_norm": 0.05492737516760826, "learning_rate": 8.941960672634158e-06, "loss": 0.0018, "step": 70850 }, { "epoch": 1.1594534893234067, "grad_norm": 0.05690428987145424, "learning_rate": 8.9415214278906e-06, "loss": 0.0012, "step": 70860 }, { "epoch": 1.1596171152744825, "grad_norm": 0.08667419105768204, "learning_rate": 8.941082102782775e-06, "loss": 0.0026, "step": 70870 }, { "epoch": 1.1597807412255583, "grad_norm": 0.036543503403663635, "learning_rate": 8.940642697319637e-06, "loss": 0.0021, "step": 70880 }, { "epoch": 1.1599443671766343, "grad_norm": 0.01717054843902588, "learning_rate": 8.940203211510145e-06, "loss": 0.0019, "step": 70890 }, { "epoch": 1.16010799312771, "grad_norm": 0.04857753962278366, "learning_rate": 8.939763645363262e-06, "loss": 0.0029, "step": 70900 }, { "epoch": 1.1602716190787858, "grad_norm": 0.036491744220256805, "learning_rate": 8.939323998887946e-06, "loss": 0.0014, "step": 70910 }, { "epoch": 1.1604352450298618, "grad_norm": 0.026774834841489792, "learning_rate": 8.938884272093166e-06, "loss": 0.0016, "step": 70920 }, { "epoch": 1.1605988709809376, "grad_norm": 0.043253034353256226, "learning_rate": 8.938444464987885e-06, "loss": 0.0034, "step": 70930 }, { "epoch": 1.1607624969320134, "grad_norm": 0.005137507803738117, "learning_rate": 8.938004577581071e-06, "loss": 0.0031, "step": 70940 }, { "epoch": 1.1609261228830892, "grad_norm": 0.05714041367173195, "learning_rate": 8.937564609881693e-06, "loss": 0.0018, "step": 70950 }, { "epoch": 1.1610897488341652, "grad_norm": 0.04391579329967499, "learning_rate": 8.937124561898723e-06, "loss": 0.0017, "step": 70960 }, { "epoch": 1.161253374785241, "grad_norm": 0.04809566214680672, "learning_rate": 8.93668443364113e-06, "loss": 0.0017, "step": 70970 }, { "epoch": 1.1614170007363167, "grad_norm": 0.03688882663846016, "learning_rate": 8.936244225117887e-06, "loss": 0.0016, "step": 70980 }, { "epoch": 1.1615806266873927, "grad_norm": 0.006928622256964445, "learning_rate": 8.935803936337975e-06, "loss": 0.0011, "step": 70990 }, { "epoch": 1.1617442526384685, "grad_norm": 0.0988544300198555, "learning_rate": 8.935363567310367e-06, "loss": 0.0013, "step": 71000 }, { "epoch": 1.1619078785895443, "grad_norm": 0.024960367009043694, "learning_rate": 8.934923118044044e-06, "loss": 0.0014, "step": 71010 }, { "epoch": 1.16207150454062, "grad_norm": 0.0696580559015274, "learning_rate": 8.934482588547983e-06, "loss": 0.002, "step": 71020 }, { "epoch": 1.162235130491696, "grad_norm": 0.05608615279197693, "learning_rate": 8.934041978831169e-06, "loss": 0.0014, "step": 71030 }, { "epoch": 1.1623987564427718, "grad_norm": 0.09341675788164139, "learning_rate": 8.933601288902587e-06, "loss": 0.0011, "step": 71040 }, { "epoch": 1.1625623823938476, "grad_norm": 0.039520263671875, "learning_rate": 8.933160518771217e-06, "loss": 0.0016, "step": 71050 }, { "epoch": 1.1627260083449236, "grad_norm": 0.30844852328300476, "learning_rate": 8.93271966844605e-06, "loss": 0.0013, "step": 71060 }, { "epoch": 1.1628896342959993, "grad_norm": 0.20154866576194763, "learning_rate": 8.932278737936075e-06, "loss": 0.0018, "step": 71070 }, { "epoch": 1.1630532602470751, "grad_norm": 0.13012512028217316, "learning_rate": 8.93183772725028e-06, "loss": 0.0022, "step": 71080 }, { "epoch": 1.1632168861981511, "grad_norm": 0.08196554332971573, "learning_rate": 8.931396636397657e-06, "loss": 0.002, "step": 71090 }, { "epoch": 1.1633805121492269, "grad_norm": 0.07098203897476196, "learning_rate": 8.930955465387201e-06, "loss": 0.0019, "step": 71100 }, { "epoch": 1.1635441381003027, "grad_norm": 0.041801001876592636, "learning_rate": 8.930514214227906e-06, "loss": 0.0023, "step": 71110 }, { "epoch": 1.1637077640513787, "grad_norm": 0.04181079566478729, "learning_rate": 8.930072882928768e-06, "loss": 0.0033, "step": 71120 }, { "epoch": 1.1638713900024544, "grad_norm": 0.050517257302999496, "learning_rate": 8.929631471498788e-06, "loss": 0.0028, "step": 71130 }, { "epoch": 1.1640350159535302, "grad_norm": 0.05733387544751167, "learning_rate": 8.929189979946964e-06, "loss": 0.0014, "step": 71140 }, { "epoch": 1.164198641904606, "grad_norm": 0.04432041943073273, "learning_rate": 8.928748408282297e-06, "loss": 0.0017, "step": 71150 }, { "epoch": 1.164362267855682, "grad_norm": 0.08680341392755508, "learning_rate": 8.92830675651379e-06, "loss": 0.0032, "step": 71160 }, { "epoch": 1.1645258938067578, "grad_norm": 0.07772308588027954, "learning_rate": 8.927865024650452e-06, "loss": 0.0013, "step": 71170 }, { "epoch": 1.1646895197578335, "grad_norm": 0.20308654010295868, "learning_rate": 8.927423212701284e-06, "loss": 0.0016, "step": 71180 }, { "epoch": 1.1648531457089095, "grad_norm": 0.08617744594812393, "learning_rate": 8.926981320675299e-06, "loss": 0.0021, "step": 71190 }, { "epoch": 1.1650167716599853, "grad_norm": 0.06717842817306519, "learning_rate": 8.926539348581505e-06, "loss": 0.0028, "step": 71200 }, { "epoch": 1.165180397611061, "grad_norm": 0.17935499548912048, "learning_rate": 8.926097296428912e-06, "loss": 0.0023, "step": 71210 }, { "epoch": 1.1653440235621368, "grad_norm": 0.01850162073969841, "learning_rate": 8.925655164226534e-06, "loss": 0.0009, "step": 71220 }, { "epoch": 1.1655076495132128, "grad_norm": 0.05111554637551308, "learning_rate": 8.925212951983386e-06, "loss": 0.0033, "step": 71230 }, { "epoch": 1.1656712754642886, "grad_norm": 0.008869693614542484, "learning_rate": 8.924770659708483e-06, "loss": 0.0012, "step": 71240 }, { "epoch": 1.1658349014153644, "grad_norm": 0.0968051552772522, "learning_rate": 8.924328287410846e-06, "loss": 0.0019, "step": 71250 }, { "epoch": 1.1659985273664404, "grad_norm": 0.018725795671343803, "learning_rate": 8.923885835099493e-06, "loss": 0.0022, "step": 71260 }, { "epoch": 1.1661621533175162, "grad_norm": 0.05000225454568863, "learning_rate": 8.923443302783443e-06, "loss": 0.0029, "step": 71270 }, { "epoch": 1.166325779268592, "grad_norm": 0.08318432420492172, "learning_rate": 8.923000690471723e-06, "loss": 0.0011, "step": 71280 }, { "epoch": 1.166489405219668, "grad_norm": 0.0616910420358181, "learning_rate": 8.922557998173353e-06, "loss": 0.0016, "step": 71290 }, { "epoch": 1.1666530311707437, "grad_norm": 0.0025029846001416445, "learning_rate": 8.922115225897363e-06, "loss": 0.0011, "step": 71300 }, { "epoch": 1.1668166571218195, "grad_norm": 0.07532427459955215, "learning_rate": 8.921672373652779e-06, "loss": 0.0021, "step": 71310 }, { "epoch": 1.1669802830728955, "grad_norm": 0.049737878143787384, "learning_rate": 8.921229441448632e-06, "loss": 0.0012, "step": 71320 }, { "epoch": 1.1671439090239712, "grad_norm": 0.4919170141220093, "learning_rate": 8.920786429293948e-06, "loss": 0.0021, "step": 71330 }, { "epoch": 1.167307534975047, "grad_norm": 0.08708576112985611, "learning_rate": 8.920343337197766e-06, "loss": 0.0016, "step": 71340 }, { "epoch": 1.1674711609261228, "grad_norm": 0.19092780351638794, "learning_rate": 8.919900165169116e-06, "loss": 0.0015, "step": 71350 }, { "epoch": 1.1676347868771988, "grad_norm": 0.03275628387928009, "learning_rate": 8.919456913217037e-06, "loss": 0.0038, "step": 71360 }, { "epoch": 1.1677984128282746, "grad_norm": 0.06073877215385437, "learning_rate": 8.919013581350565e-06, "loss": 0.0018, "step": 71370 }, { "epoch": 1.1679620387793503, "grad_norm": 0.09677965193986893, "learning_rate": 8.918570169578736e-06, "loss": 0.0017, "step": 71380 }, { "epoch": 1.1681256647304263, "grad_norm": 0.009066395461559296, "learning_rate": 8.918126677910596e-06, "loss": 0.0009, "step": 71390 }, { "epoch": 1.1682892906815021, "grad_norm": 0.046372245997190475, "learning_rate": 8.917683106355186e-06, "loss": 0.0018, "step": 71400 }, { "epoch": 1.1684529166325779, "grad_norm": 0.11190778017044067, "learning_rate": 8.917239454921548e-06, "loss": 0.0019, "step": 71410 }, { "epoch": 1.1686165425836537, "grad_norm": 0.06696871668100357, "learning_rate": 8.91679572361873e-06, "loss": 0.0028, "step": 71420 }, { "epoch": 1.1687801685347297, "grad_norm": 0.052881184965372086, "learning_rate": 8.916351912455777e-06, "loss": 0.0013, "step": 71430 }, { "epoch": 1.1689437944858054, "grad_norm": 0.06407517194747925, "learning_rate": 8.91590802144174e-06, "loss": 0.0029, "step": 71440 }, { "epoch": 1.1691074204368812, "grad_norm": 0.027614930644631386, "learning_rate": 8.91546405058567e-06, "loss": 0.0014, "step": 71450 }, { "epoch": 1.1692710463879572, "grad_norm": 0.08930761367082596, "learning_rate": 8.915019999896616e-06, "loss": 0.002, "step": 71460 }, { "epoch": 1.169434672339033, "grad_norm": 0.06320717930793762, "learning_rate": 8.914575869383635e-06, "loss": 0.003, "step": 71470 }, { "epoch": 1.1695982982901088, "grad_norm": 0.035674042999744415, "learning_rate": 8.91413165905578e-06, "loss": 0.0011, "step": 71480 }, { "epoch": 1.1697619242411847, "grad_norm": 0.04924258217215538, "learning_rate": 8.91368736892211e-06, "loss": 0.002, "step": 71490 }, { "epoch": 1.1699255501922605, "grad_norm": 0.04916336387395859, "learning_rate": 8.913242998991684e-06, "loss": 0.0021, "step": 71500 }, { "epoch": 1.1700891761433363, "grad_norm": 0.11865325272083282, "learning_rate": 8.91279854927356e-06, "loss": 0.0017, "step": 71510 }, { "epoch": 1.1702528020944123, "grad_norm": 0.10837219655513763, "learning_rate": 8.912354019776804e-06, "loss": 0.0048, "step": 71520 }, { "epoch": 1.170416428045488, "grad_norm": 0.1919778287410736, "learning_rate": 8.911909410510474e-06, "loss": 0.002, "step": 71530 }, { "epoch": 1.1705800539965638, "grad_norm": 0.09018212556838989, "learning_rate": 8.911464721483638e-06, "loss": 0.0027, "step": 71540 }, { "epoch": 1.1707436799476396, "grad_norm": 0.04181081801652908, "learning_rate": 8.911019952705364e-06, "loss": 0.0018, "step": 71550 }, { "epoch": 1.1709073058987156, "grad_norm": 0.05917549878358841, "learning_rate": 8.91057510418472e-06, "loss": 0.0016, "step": 71560 }, { "epoch": 1.1710709318497914, "grad_norm": 0.06257057934999466, "learning_rate": 8.910130175930773e-06, "loss": 0.002, "step": 71570 }, { "epoch": 1.1712345578008672, "grad_norm": 0.04500136524438858, "learning_rate": 8.9096851679526e-06, "loss": 0.0024, "step": 71580 }, { "epoch": 1.171398183751943, "grad_norm": 0.030405977740883827, "learning_rate": 8.909240080259269e-06, "loss": 0.003, "step": 71590 }, { "epoch": 1.171561809703019, "grad_norm": 0.14580655097961426, "learning_rate": 8.908794912859859e-06, "loss": 0.0014, "step": 71600 }, { "epoch": 1.1717254356540947, "grad_norm": 0.0467979796230793, "learning_rate": 8.908349665763444e-06, "loss": 0.0023, "step": 71610 }, { "epoch": 1.1718890616051705, "grad_norm": 0.07793543487787247, "learning_rate": 8.907904338979105e-06, "loss": 0.0015, "step": 71620 }, { "epoch": 1.1720526875562465, "grad_norm": 0.11909995228052139, "learning_rate": 8.90745893251592e-06, "loss": 0.0022, "step": 71630 }, { "epoch": 1.1722163135073222, "grad_norm": 0.11717238277196884, "learning_rate": 8.90701344638297e-06, "loss": 0.005, "step": 71640 }, { "epoch": 1.172379939458398, "grad_norm": 0.07532411813735962, "learning_rate": 8.906567880589337e-06, "loss": 0.002, "step": 71650 }, { "epoch": 1.172543565409474, "grad_norm": 0.03822420910000801, "learning_rate": 8.90612223514411e-06, "loss": 0.0014, "step": 71660 }, { "epoch": 1.1727071913605498, "grad_norm": 0.15605828166007996, "learning_rate": 8.90567651005637e-06, "loss": 0.0019, "step": 71670 }, { "epoch": 1.1728708173116256, "grad_norm": 0.04435122385621071, "learning_rate": 8.905230705335209e-06, "loss": 0.0014, "step": 71680 }, { "epoch": 1.1730344432627016, "grad_norm": 0.06672533601522446, "learning_rate": 8.904784820989716e-06, "loss": 0.0016, "step": 71690 }, { "epoch": 1.1731980692137773, "grad_norm": 0.02362622134387493, "learning_rate": 8.90433885702898e-06, "loss": 0.0018, "step": 71700 }, { "epoch": 1.1733616951648531, "grad_norm": 0.1637563556432724, "learning_rate": 8.903892813462097e-06, "loss": 0.0028, "step": 71710 }, { "epoch": 1.173525321115929, "grad_norm": 0.04799013212323189, "learning_rate": 8.903446690298156e-06, "loss": 0.002, "step": 71720 }, { "epoch": 1.1736889470670049, "grad_norm": 0.06015025079250336, "learning_rate": 8.90300048754626e-06, "loss": 0.0013, "step": 71730 }, { "epoch": 1.1738525730180807, "grad_norm": 0.054412297904491425, "learning_rate": 8.902554205215503e-06, "loss": 0.0024, "step": 71740 }, { "epoch": 1.1740161989691564, "grad_norm": 0.1975947916507721, "learning_rate": 8.902107843314984e-06, "loss": 0.0017, "step": 71750 }, { "epoch": 1.1741798249202324, "grad_norm": 0.03192982077598572, "learning_rate": 8.901661401853806e-06, "loss": 0.0024, "step": 71760 }, { "epoch": 1.1743434508713082, "grad_norm": 0.07052136957645416, "learning_rate": 8.901214880841068e-06, "loss": 0.0027, "step": 71770 }, { "epoch": 1.174507076822384, "grad_norm": 0.10716145485639572, "learning_rate": 8.900768280285878e-06, "loss": 0.0014, "step": 71780 }, { "epoch": 1.1746707027734598, "grad_norm": 0.05780303105711937, "learning_rate": 8.900321600197339e-06, "loss": 0.0023, "step": 71790 }, { "epoch": 1.1748343287245357, "grad_norm": 0.06006699055433273, "learning_rate": 8.899874840584561e-06, "loss": 0.0024, "step": 71800 }, { "epoch": 1.1749979546756115, "grad_norm": 0.07152523845434189, "learning_rate": 8.89942800145665e-06, "loss": 0.0032, "step": 71810 }, { "epoch": 1.1751615806266873, "grad_norm": 0.027260232716798782, "learning_rate": 8.898981082822723e-06, "loss": 0.0018, "step": 71820 }, { "epoch": 1.1753252065777633, "grad_norm": 0.03079555556178093, "learning_rate": 8.898534084691883e-06, "loss": 0.0044, "step": 71830 }, { "epoch": 1.175488832528839, "grad_norm": 0.12421001493930817, "learning_rate": 8.89808700707325e-06, "loss": 0.002, "step": 71840 }, { "epoch": 1.1756524584799148, "grad_norm": 0.021248919889330864, "learning_rate": 8.897639849975939e-06, "loss": 0.0017, "step": 71850 }, { "epoch": 1.1758160844309908, "grad_norm": 0.0674552470445633, "learning_rate": 8.897192613409065e-06, "loss": 0.0019, "step": 71860 }, { "epoch": 1.1759797103820666, "grad_norm": 0.06067625433206558, "learning_rate": 8.896745297381749e-06, "loss": 0.0024, "step": 71870 }, { "epoch": 1.1761433363331424, "grad_norm": 0.35098031163215637, "learning_rate": 8.896297901903108e-06, "loss": 0.0031, "step": 71880 }, { "epoch": 1.1763069622842184, "grad_norm": 0.14451055228710175, "learning_rate": 8.895850426982269e-06, "loss": 0.003, "step": 71890 }, { "epoch": 1.1764705882352942, "grad_norm": 0.03870411217212677, "learning_rate": 8.895402872628352e-06, "loss": 0.0026, "step": 71900 }, { "epoch": 1.17663421418637, "grad_norm": 0.05143825709819794, "learning_rate": 8.894955238850485e-06, "loss": 0.0021, "step": 71910 }, { "epoch": 1.1767978401374457, "grad_norm": 0.01958618499338627, "learning_rate": 8.894507525657792e-06, "loss": 0.0026, "step": 71920 }, { "epoch": 1.1769614660885217, "grad_norm": 0.02244170755147934, "learning_rate": 8.894059733059402e-06, "loss": 0.0019, "step": 71930 }, { "epoch": 1.1771250920395975, "grad_norm": 0.002416870091110468, "learning_rate": 8.893611861064447e-06, "loss": 0.0025, "step": 71940 }, { "epoch": 1.1772887179906733, "grad_norm": 0.012837540358304977, "learning_rate": 8.893163909682057e-06, "loss": 0.001, "step": 71950 }, { "epoch": 1.1774523439417492, "grad_norm": 0.0800638198852539, "learning_rate": 8.892715878921366e-06, "loss": 0.0021, "step": 71960 }, { "epoch": 1.177615969892825, "grad_norm": 0.05906900390982628, "learning_rate": 8.892267768791509e-06, "loss": 0.0022, "step": 71970 }, { "epoch": 1.1777795958439008, "grad_norm": 0.08714069426059723, "learning_rate": 8.891819579301623e-06, "loss": 0.0015, "step": 71980 }, { "epoch": 1.1779432217949766, "grad_norm": 0.045349422842264175, "learning_rate": 8.891371310460847e-06, "loss": 0.0022, "step": 71990 }, { "epoch": 1.1781068477460526, "grad_norm": 0.06770152598619461, "learning_rate": 8.890922962278315e-06, "loss": 0.0022, "step": 72000 }, { "epoch": 1.1782704736971283, "grad_norm": 0.11870700120925903, "learning_rate": 8.890474534763177e-06, "loss": 0.0028, "step": 72010 }, { "epoch": 1.1784340996482041, "grad_norm": 0.03854367509484291, "learning_rate": 8.89002602792457e-06, "loss": 0.002, "step": 72020 }, { "epoch": 1.17859772559928, "grad_norm": 0.0683712437748909, "learning_rate": 8.88957744177164e-06, "loss": 0.0017, "step": 72030 }, { "epoch": 1.1787613515503559, "grad_norm": 0.09648788720369339, "learning_rate": 8.889128776313536e-06, "loss": 0.0017, "step": 72040 }, { "epoch": 1.1789249775014317, "grad_norm": 0.02965831197798252, "learning_rate": 8.888680031559403e-06, "loss": 0.0017, "step": 72050 }, { "epoch": 1.1790886034525077, "grad_norm": 0.06156346946954727, "learning_rate": 8.88823120751839e-06, "loss": 0.0028, "step": 72060 }, { "epoch": 1.1792522294035834, "grad_norm": 0.09536924958229065, "learning_rate": 8.88778230419965e-06, "loss": 0.0022, "step": 72070 }, { "epoch": 1.1794158553546592, "grad_norm": 0.12155801802873611, "learning_rate": 8.887333321612338e-06, "loss": 0.0033, "step": 72080 }, { "epoch": 1.1795794813057352, "grad_norm": 0.029723044484853745, "learning_rate": 8.886884259765602e-06, "loss": 0.002, "step": 72090 }, { "epoch": 1.179743107256811, "grad_norm": 0.10072343796491623, "learning_rate": 8.886435118668604e-06, "loss": 0.0022, "step": 72100 }, { "epoch": 1.1799067332078867, "grad_norm": 0.14363622665405273, "learning_rate": 8.885985898330497e-06, "loss": 0.0019, "step": 72110 }, { "epoch": 1.1800703591589625, "grad_norm": 0.12079304456710815, "learning_rate": 8.885536598760445e-06, "loss": 0.0026, "step": 72120 }, { "epoch": 1.1802339851100385, "grad_norm": 0.013871977105736732, "learning_rate": 8.885087219967606e-06, "loss": 0.0017, "step": 72130 }, { "epoch": 1.1803976110611143, "grad_norm": 0.08006840199232101, "learning_rate": 8.884637761961142e-06, "loss": 0.0012, "step": 72140 }, { "epoch": 1.18056123701219, "grad_norm": 0.03574131429195404, "learning_rate": 8.884188224750217e-06, "loss": 0.0012, "step": 72150 }, { "epoch": 1.180724862963266, "grad_norm": 0.0699596256017685, "learning_rate": 8.883738608343997e-06, "loss": 0.0034, "step": 72160 }, { "epoch": 1.1808884889143418, "grad_norm": 0.027769653126597404, "learning_rate": 8.88328891275165e-06, "loss": 0.0019, "step": 72170 }, { "epoch": 1.1810521148654176, "grad_norm": 0.01549489051103592, "learning_rate": 8.882839137982347e-06, "loss": 0.0019, "step": 72180 }, { "epoch": 1.1812157408164934, "grad_norm": 0.10522101819515228, "learning_rate": 8.882389284045255e-06, "loss": 0.0016, "step": 72190 }, { "epoch": 1.1813793667675694, "grad_norm": 0.052218761295080185, "learning_rate": 8.881939350949547e-06, "loss": 0.0024, "step": 72200 }, { "epoch": 1.1815429927186452, "grad_norm": 0.007932720705866814, "learning_rate": 8.881489338704397e-06, "loss": 0.002, "step": 72210 }, { "epoch": 1.181706618669721, "grad_norm": 0.02375289611518383, "learning_rate": 8.88103924731898e-06, "loss": 0.0022, "step": 72220 }, { "epoch": 1.181870244620797, "grad_norm": 0.025538424029946327, "learning_rate": 8.880589076802476e-06, "loss": 0.0016, "step": 72230 }, { "epoch": 1.1820338705718727, "grad_norm": 0.029430221766233444, "learning_rate": 8.880138827164057e-06, "loss": 0.0029, "step": 72240 }, { "epoch": 1.1821974965229485, "grad_norm": 0.030596435070037842, "learning_rate": 8.87968849841291e-06, "loss": 0.0014, "step": 72250 }, { "epoch": 1.1823611224740245, "grad_norm": 0.01329885981976986, "learning_rate": 8.879238090558214e-06, "loss": 0.0022, "step": 72260 }, { "epoch": 1.1825247484251002, "grad_norm": 0.13094773888587952, "learning_rate": 8.878787603609153e-06, "loss": 0.0026, "step": 72270 }, { "epoch": 1.182688374376176, "grad_norm": 0.11684589833021164, "learning_rate": 8.87833703757491e-06, "loss": 0.0015, "step": 72280 }, { "epoch": 1.182852000327252, "grad_norm": 0.05225653573870659, "learning_rate": 8.877886392464675e-06, "loss": 0.0011, "step": 72290 }, { "epoch": 1.1830156262783278, "grad_norm": 0.05164182558655739, "learning_rate": 8.877435668287633e-06, "loss": 0.0018, "step": 72300 }, { "epoch": 1.1831792522294036, "grad_norm": 0.04933340474963188, "learning_rate": 8.876984865052976e-06, "loss": 0.0011, "step": 72310 }, { "epoch": 1.1833428781804793, "grad_norm": 0.02951749972999096, "learning_rate": 8.876533982769893e-06, "loss": 0.002, "step": 72320 }, { "epoch": 1.1835065041315553, "grad_norm": 0.07076246291399002, "learning_rate": 8.876083021447582e-06, "loss": 0.0028, "step": 72330 }, { "epoch": 1.1836701300826311, "grad_norm": 0.13658180832862854, "learning_rate": 8.875631981095232e-06, "loss": 0.0022, "step": 72340 }, { "epoch": 1.1838337560337069, "grad_norm": 0.050415802747011185, "learning_rate": 8.875180861722043e-06, "loss": 0.0019, "step": 72350 }, { "epoch": 1.1839973819847829, "grad_norm": 0.10256805270910263, "learning_rate": 8.874729663337213e-06, "loss": 0.0017, "step": 72360 }, { "epoch": 1.1841610079358587, "grad_norm": 0.06745364516973495, "learning_rate": 8.87427838594994e-06, "loss": 0.0023, "step": 72370 }, { "epoch": 1.1843246338869344, "grad_norm": 0.06308191269636154, "learning_rate": 8.873827029569424e-06, "loss": 0.0009, "step": 72380 }, { "epoch": 1.1844882598380102, "grad_norm": 0.03394214063882828, "learning_rate": 8.873375594204872e-06, "loss": 0.0014, "step": 72390 }, { "epoch": 1.1846518857890862, "grad_norm": 0.0032585866283625364, "learning_rate": 8.872924079865483e-06, "loss": 0.0018, "step": 72400 }, { "epoch": 1.184815511740162, "grad_norm": 0.07345162332057953, "learning_rate": 8.872472486560469e-06, "loss": 0.0022, "step": 72410 }, { "epoch": 1.1849791376912377, "grad_norm": 0.07935750484466553, "learning_rate": 8.87202081429903e-06, "loss": 0.0023, "step": 72420 }, { "epoch": 1.1851427636423137, "grad_norm": 0.054423559457063675, "learning_rate": 8.871569063090384e-06, "loss": 0.0011, "step": 72430 }, { "epoch": 1.1853063895933895, "grad_norm": 0.14077724516391754, "learning_rate": 8.871117232943737e-06, "loss": 0.0017, "step": 72440 }, { "epoch": 1.1854700155444653, "grad_norm": 0.11861390620470047, "learning_rate": 8.870665323868301e-06, "loss": 0.0031, "step": 72450 }, { "epoch": 1.1856336414955413, "grad_norm": 0.031539324671030045, "learning_rate": 8.87021333587329e-06, "loss": 0.0013, "step": 72460 }, { "epoch": 1.185797267446617, "grad_norm": 0.07319339364767075, "learning_rate": 8.86976126896792e-06, "loss": 0.0024, "step": 72470 }, { "epoch": 1.1859608933976928, "grad_norm": 0.05932572856545448, "learning_rate": 8.86930912316141e-06, "loss": 0.0025, "step": 72480 }, { "epoch": 1.1861245193487688, "grad_norm": 0.20732823014259338, "learning_rate": 8.868856898462979e-06, "loss": 0.0017, "step": 72490 }, { "epoch": 1.1862881452998446, "grad_norm": 0.03690950572490692, "learning_rate": 8.868404594881846e-06, "loss": 0.0016, "step": 72500 }, { "epoch": 1.1864517712509204, "grad_norm": 0.03408829867839813, "learning_rate": 8.867952212427231e-06, "loss": 0.0013, "step": 72510 }, { "epoch": 1.1866153972019962, "grad_norm": 0.09638100862503052, "learning_rate": 8.867499751108362e-06, "loss": 0.0013, "step": 72520 }, { "epoch": 1.1867790231530722, "grad_norm": 0.09095708280801773, "learning_rate": 8.86704721093446e-06, "loss": 0.0017, "step": 72530 }, { "epoch": 1.186942649104148, "grad_norm": 0.05441784858703613, "learning_rate": 8.866594591914756e-06, "loss": 0.0013, "step": 72540 }, { "epoch": 1.1871062750552237, "grad_norm": 0.037630483508110046, "learning_rate": 8.866141894058477e-06, "loss": 0.0025, "step": 72550 }, { "epoch": 1.1872699010062995, "grad_norm": 0.06696026772260666, "learning_rate": 8.865689117374853e-06, "loss": 0.002, "step": 72560 }, { "epoch": 1.1874335269573755, "grad_norm": 0.050587672740221024, "learning_rate": 8.865236261873115e-06, "loss": 0.0014, "step": 72570 }, { "epoch": 1.1875971529084512, "grad_norm": 0.02398059330880642, "learning_rate": 8.864783327562496e-06, "loss": 0.0025, "step": 72580 }, { "epoch": 1.187760778859527, "grad_norm": 0.07376964390277863, "learning_rate": 8.864330314452231e-06, "loss": 0.003, "step": 72590 }, { "epoch": 1.187924404810603, "grad_norm": 0.04068354517221451, "learning_rate": 8.86387722255156e-06, "loss": 0.0011, "step": 72600 }, { "epoch": 1.1880880307616788, "grad_norm": 0.30255067348480225, "learning_rate": 8.863424051869717e-06, "loss": 0.0029, "step": 72610 }, { "epoch": 1.1882516567127546, "grad_norm": 0.11279942840337753, "learning_rate": 8.862970802415945e-06, "loss": 0.0018, "step": 72620 }, { "epoch": 1.1884152826638306, "grad_norm": 0.008608336560428143, "learning_rate": 8.862517474199481e-06, "loss": 0.0013, "step": 72630 }, { "epoch": 1.1885789086149063, "grad_norm": 0.05074889212846756, "learning_rate": 8.862064067229573e-06, "loss": 0.0019, "step": 72640 }, { "epoch": 1.1887425345659821, "grad_norm": 0.156234472990036, "learning_rate": 8.861610581515463e-06, "loss": 0.0022, "step": 72650 }, { "epoch": 1.188906160517058, "grad_norm": 0.005193889141082764, "learning_rate": 8.861157017066396e-06, "loss": 0.002, "step": 72660 }, { "epoch": 1.1890697864681339, "grad_norm": 0.04509848728775978, "learning_rate": 8.860703373891622e-06, "loss": 0.0018, "step": 72670 }, { "epoch": 1.1892334124192097, "grad_norm": 0.027748890221118927, "learning_rate": 8.86024965200039e-06, "loss": 0.0016, "step": 72680 }, { "epoch": 1.1893970383702854, "grad_norm": 0.04792681336402893, "learning_rate": 8.85979585140195e-06, "loss": 0.0016, "step": 72690 }, { "epoch": 1.1895606643213614, "grad_norm": 0.1347064971923828, "learning_rate": 8.859341972105555e-06, "loss": 0.002, "step": 72700 }, { "epoch": 1.1897242902724372, "grad_norm": 0.06029074639081955, "learning_rate": 8.85888801412046e-06, "loss": 0.0022, "step": 72710 }, { "epoch": 1.189887916223513, "grad_norm": 0.10277736186981201, "learning_rate": 8.858433977455921e-06, "loss": 0.0029, "step": 72720 }, { "epoch": 1.190051542174589, "grad_norm": 0.06424213945865631, "learning_rate": 8.857979862121194e-06, "loss": 0.0021, "step": 72730 }, { "epoch": 1.1902151681256647, "grad_norm": 0.09908853471279144, "learning_rate": 8.857525668125538e-06, "loss": 0.0016, "step": 72740 }, { "epoch": 1.1903787940767405, "grad_norm": 0.0531662292778492, "learning_rate": 8.857071395478214e-06, "loss": 0.0033, "step": 72750 }, { "epoch": 1.1905424200278163, "grad_norm": 0.011055492796003819, "learning_rate": 8.856617044188486e-06, "loss": 0.0013, "step": 72760 }, { "epoch": 1.1907060459788923, "grad_norm": 0.12431849539279938, "learning_rate": 8.856162614265616e-06, "loss": 0.002, "step": 72770 }, { "epoch": 1.190869671929968, "grad_norm": 0.04701603949069977, "learning_rate": 8.855708105718869e-06, "loss": 0.0022, "step": 72780 }, { "epoch": 1.1910332978810438, "grad_norm": 0.052084069699048996, "learning_rate": 8.855253518557514e-06, "loss": 0.0014, "step": 72790 }, { "epoch": 1.1911969238321198, "grad_norm": 0.04094717279076576, "learning_rate": 8.854798852790818e-06, "loss": 0.0036, "step": 72800 }, { "epoch": 1.1913605497831956, "grad_norm": 0.14926788210868835, "learning_rate": 8.854344108428052e-06, "loss": 0.0019, "step": 72810 }, { "epoch": 1.1915241757342714, "grad_norm": 0.052239470183849335, "learning_rate": 8.853889285478487e-06, "loss": 0.0012, "step": 72820 }, { "epoch": 1.1916878016853474, "grad_norm": 0.019002385437488556, "learning_rate": 8.853434383951397e-06, "loss": 0.0023, "step": 72830 }, { "epoch": 1.1918514276364232, "grad_norm": 0.02613089419901371, "learning_rate": 8.852979403856059e-06, "loss": 0.0016, "step": 72840 }, { "epoch": 1.192015053587499, "grad_norm": 0.05116555467247963, "learning_rate": 8.852524345201745e-06, "loss": 0.0015, "step": 72850 }, { "epoch": 1.192178679538575, "grad_norm": 0.031165193766355515, "learning_rate": 8.852069207997737e-06, "loss": 0.0014, "step": 72860 }, { "epoch": 1.1923423054896507, "grad_norm": 0.07527168840169907, "learning_rate": 8.851613992253316e-06, "loss": 0.0014, "step": 72870 }, { "epoch": 1.1925059314407265, "grad_norm": 0.04865449666976929, "learning_rate": 8.85115869797776e-06, "loss": 0.001, "step": 72880 }, { "epoch": 1.1926695573918022, "grad_norm": 0.2496810406446457, "learning_rate": 8.850703325180353e-06, "loss": 0.0029, "step": 72890 }, { "epoch": 1.1928331833428782, "grad_norm": 0.05601845309138298, "learning_rate": 8.85024787387038e-06, "loss": 0.002, "step": 72900 }, { "epoch": 1.192996809293954, "grad_norm": 0.05422599986195564, "learning_rate": 8.849792344057127e-06, "loss": 0.0019, "step": 72910 }, { "epoch": 1.1931604352450298, "grad_norm": 0.04254503920674324, "learning_rate": 8.849336735749884e-06, "loss": 0.0016, "step": 72920 }, { "epoch": 1.1933240611961058, "grad_norm": 0.10248836129903793, "learning_rate": 8.848881048957937e-06, "loss": 0.0027, "step": 72930 }, { "epoch": 1.1934876871471816, "grad_norm": 0.09130644053220749, "learning_rate": 8.84842528369058e-06, "loss": 0.0016, "step": 72940 }, { "epoch": 1.1936513130982573, "grad_norm": 0.07497701793909073, "learning_rate": 8.847969439957104e-06, "loss": 0.005, "step": 72950 }, { "epoch": 1.1938149390493331, "grad_norm": 0.07964511215686798, "learning_rate": 8.847513517766804e-06, "loss": 0.0017, "step": 72960 }, { "epoch": 1.193978565000409, "grad_norm": 0.02900681272149086, "learning_rate": 8.847057517128974e-06, "loss": 0.0016, "step": 72970 }, { "epoch": 1.1941421909514849, "grad_norm": 0.10807224363088608, "learning_rate": 8.846601438052915e-06, "loss": 0.002, "step": 72980 }, { "epoch": 1.1943058169025607, "grad_norm": 0.06218891218304634, "learning_rate": 8.846145280547923e-06, "loss": 0.0015, "step": 72990 }, { "epoch": 1.1944694428536367, "grad_norm": 0.04874090850353241, "learning_rate": 8.8456890446233e-06, "loss": 0.0024, "step": 73000 }, { "epoch": 1.1946330688047124, "grad_norm": 0.0964570939540863, "learning_rate": 8.845232730288348e-06, "loss": 0.0019, "step": 73010 }, { "epoch": 1.1947966947557882, "grad_norm": 0.18132853507995605, "learning_rate": 8.844776337552372e-06, "loss": 0.0042, "step": 73020 }, { "epoch": 1.1949603207068642, "grad_norm": 0.08984702080488205, "learning_rate": 8.844319866424674e-06, "loss": 0.0016, "step": 73030 }, { "epoch": 1.19512394665794, "grad_norm": 0.016131578013300896, "learning_rate": 8.843863316914566e-06, "loss": 0.0017, "step": 73040 }, { "epoch": 1.1952875726090157, "grad_norm": 0.028661439195275307, "learning_rate": 8.843406689031352e-06, "loss": 0.0018, "step": 73050 }, { "epoch": 1.1954511985600917, "grad_norm": 0.08430697023868561, "learning_rate": 8.842949982784346e-06, "loss": 0.0028, "step": 73060 }, { "epoch": 1.1956148245111675, "grad_norm": 0.08678245544433594, "learning_rate": 8.842493198182855e-06, "loss": 0.0012, "step": 73070 }, { "epoch": 1.1957784504622433, "grad_norm": 0.13325031101703644, "learning_rate": 8.842036335236199e-06, "loss": 0.0008, "step": 73080 }, { "epoch": 1.195942076413319, "grad_norm": 0.06851695477962494, "learning_rate": 8.841579393953688e-06, "loss": 0.0015, "step": 73090 }, { "epoch": 1.196105702364395, "grad_norm": 0.04561896249651909, "learning_rate": 8.841122374344642e-06, "loss": 0.003, "step": 73100 }, { "epoch": 1.1962693283154708, "grad_norm": 0.04475175961852074, "learning_rate": 8.840665276418375e-06, "loss": 0.0011, "step": 73110 }, { "epoch": 1.1964329542665466, "grad_norm": 0.028002148494124413, "learning_rate": 8.840208100184212e-06, "loss": 0.0018, "step": 73120 }, { "epoch": 1.1965965802176226, "grad_norm": 0.048623230308294296, "learning_rate": 8.83975084565147e-06, "loss": 0.0028, "step": 73130 }, { "epoch": 1.1967602061686984, "grad_norm": 0.04422427713871002, "learning_rate": 8.839293512829474e-06, "loss": 0.0015, "step": 73140 }, { "epoch": 1.1969238321197742, "grad_norm": 0.07482912391424179, "learning_rate": 8.83883610172755e-06, "loss": 0.0021, "step": 73150 }, { "epoch": 1.19708745807085, "grad_norm": 0.0285961851477623, "learning_rate": 8.83837861235502e-06, "loss": 0.0039, "step": 73160 }, { "epoch": 1.197251084021926, "grad_norm": 0.13201288878917694, "learning_rate": 8.837921044721217e-06, "loss": 0.0021, "step": 73170 }, { "epoch": 1.1974147099730017, "grad_norm": 0.03423953428864479, "learning_rate": 8.837463398835467e-06, "loss": 0.0021, "step": 73180 }, { "epoch": 1.1975783359240775, "grad_norm": 0.09018529206514359, "learning_rate": 8.837005674707102e-06, "loss": 0.0015, "step": 73190 }, { "epoch": 1.1977419618751535, "grad_norm": 0.07367795705795288, "learning_rate": 8.836547872345454e-06, "loss": 0.003, "step": 73200 }, { "epoch": 1.1979055878262292, "grad_norm": 0.09012651443481445, "learning_rate": 8.836089991759858e-06, "loss": 0.0022, "step": 73210 }, { "epoch": 1.198069213777305, "grad_norm": 0.09317804127931595, "learning_rate": 8.835632032959651e-06, "loss": 0.0022, "step": 73220 }, { "epoch": 1.198232839728381, "grad_norm": 0.1074906662106514, "learning_rate": 8.835173995954168e-06, "loss": 0.0016, "step": 73230 }, { "epoch": 1.1983964656794568, "grad_norm": 0.0636029914021492, "learning_rate": 8.834715880752747e-06, "loss": 0.0021, "step": 73240 }, { "epoch": 1.1985600916305326, "grad_norm": 0.05850861966609955, "learning_rate": 8.834257687364733e-06, "loss": 0.0022, "step": 73250 }, { "epoch": 1.1987237175816086, "grad_norm": 0.09896481037139893, "learning_rate": 8.833799415799464e-06, "loss": 0.0013, "step": 73260 }, { "epoch": 1.1988873435326843, "grad_norm": 0.07864756137132645, "learning_rate": 8.833341066066287e-06, "loss": 0.0015, "step": 73270 }, { "epoch": 1.19905096948376, "grad_norm": 0.019310805946588516, "learning_rate": 8.832882638174545e-06, "loss": 0.0011, "step": 73280 }, { "epoch": 1.1992145954348359, "grad_norm": 0.11059537529945374, "learning_rate": 8.832424132133585e-06, "loss": 0.0025, "step": 73290 }, { "epoch": 1.1993782213859119, "grad_norm": 0.096697598695755, "learning_rate": 8.83196554795276e-06, "loss": 0.0034, "step": 73300 }, { "epoch": 1.1995418473369877, "grad_norm": 0.07791110873222351, "learning_rate": 8.831506885641412e-06, "loss": 0.0024, "step": 73310 }, { "epoch": 1.1997054732880634, "grad_norm": 0.14104826748371124, "learning_rate": 8.831048145208898e-06, "loss": 0.002, "step": 73320 }, { "epoch": 1.1998690992391392, "grad_norm": 0.32091501355171204, "learning_rate": 8.830589326664571e-06, "loss": 0.0026, "step": 73330 }, { "epoch": 1.2000327251902152, "grad_norm": 0.022833701223134995, "learning_rate": 8.830130430017788e-06, "loss": 0.003, "step": 73340 }, { "epoch": 1.200196351141291, "grad_norm": 0.014518325217068195, "learning_rate": 8.8296714552779e-06, "loss": 0.0013, "step": 73350 }, { "epoch": 1.2003599770923667, "grad_norm": 0.07765807956457138, "learning_rate": 8.829212402454269e-06, "loss": 0.0026, "step": 73360 }, { "epoch": 1.2005236030434427, "grad_norm": 0.030716780573129654, "learning_rate": 8.828753271556251e-06, "loss": 0.0015, "step": 73370 }, { "epoch": 1.2006872289945185, "grad_norm": 0.0437595508992672, "learning_rate": 8.828294062593214e-06, "loss": 0.0027, "step": 73380 }, { "epoch": 1.2008508549455943, "grad_norm": 0.08022785931825638, "learning_rate": 8.827834775574515e-06, "loss": 0.0015, "step": 73390 }, { "epoch": 1.2010144808966703, "grad_norm": 0.06842119246721268, "learning_rate": 8.827375410509522e-06, "loss": 0.002, "step": 73400 }, { "epoch": 1.201178106847746, "grad_norm": 0.052227653563022614, "learning_rate": 8.826915967407596e-06, "loss": 0.0021, "step": 73410 }, { "epoch": 1.2013417327988218, "grad_norm": 0.10145394504070282, "learning_rate": 8.826456446278111e-06, "loss": 0.0064, "step": 73420 }, { "epoch": 1.2015053587498978, "grad_norm": 0.04984059929847717, "learning_rate": 8.825996847130433e-06, "loss": 0.0021, "step": 73430 }, { "epoch": 1.2016689847009736, "grad_norm": 0.04320550337433815, "learning_rate": 8.825537169973932e-06, "loss": 0.0014, "step": 73440 }, { "epoch": 1.2018326106520494, "grad_norm": 0.0683826357126236, "learning_rate": 8.825077414817982e-06, "loss": 0.0012, "step": 73450 }, { "epoch": 1.2019962366031254, "grad_norm": 0.047349486500024796, "learning_rate": 8.824617581671958e-06, "loss": 0.0011, "step": 73460 }, { "epoch": 1.2021598625542012, "grad_norm": 0.0475376658141613, "learning_rate": 8.824157670545234e-06, "loss": 0.0019, "step": 73470 }, { "epoch": 1.202323488505277, "grad_norm": 0.06331076472997665, "learning_rate": 8.823697681447186e-06, "loss": 0.0022, "step": 73480 }, { "epoch": 1.2024871144563527, "grad_norm": 0.06255774945020676, "learning_rate": 8.823237614387195e-06, "loss": 0.0038, "step": 73490 }, { "epoch": 1.2026507404074287, "grad_norm": 0.12324684113264084, "learning_rate": 8.822777469374641e-06, "loss": 0.002, "step": 73500 }, { "epoch": 1.2028143663585045, "grad_norm": 0.01720973290503025, "learning_rate": 8.822317246418905e-06, "loss": 0.0016, "step": 73510 }, { "epoch": 1.2029779923095802, "grad_norm": 0.03634273633360863, "learning_rate": 8.821856945529371e-06, "loss": 0.0018, "step": 73520 }, { "epoch": 1.203141618260656, "grad_norm": 0.029111122712492943, "learning_rate": 8.821396566715424e-06, "loss": 0.0014, "step": 73530 }, { "epoch": 1.203305244211732, "grad_norm": 0.010075073689222336, "learning_rate": 8.820936109986451e-06, "loss": 0.0013, "step": 73540 }, { "epoch": 1.2034688701628078, "grad_norm": 0.07500747591257095, "learning_rate": 8.820475575351842e-06, "loss": 0.0024, "step": 73550 }, { "epoch": 1.2036324961138836, "grad_norm": 0.0384826585650444, "learning_rate": 8.820014962820984e-06, "loss": 0.0011, "step": 73560 }, { "epoch": 1.2037961220649596, "grad_norm": 0.03273945301771164, "learning_rate": 8.819554272403269e-06, "loss": 0.0021, "step": 73570 }, { "epoch": 1.2039597480160353, "grad_norm": 0.01024361327290535, "learning_rate": 8.819093504108092e-06, "loss": 0.0017, "step": 73580 }, { "epoch": 1.204123373967111, "grad_norm": 0.07553193718194962, "learning_rate": 8.818632657944846e-06, "loss": 0.0024, "step": 73590 }, { "epoch": 1.204286999918187, "grad_norm": 0.06120026856660843, "learning_rate": 8.818171733922927e-06, "loss": 0.0017, "step": 73600 }, { "epoch": 1.2044506258692629, "grad_norm": 0.10718384385108948, "learning_rate": 8.817710732051734e-06, "loss": 0.0017, "step": 73610 }, { "epoch": 1.2046142518203387, "grad_norm": 0.06168508902192116, "learning_rate": 8.817249652340667e-06, "loss": 0.0018, "step": 73620 }, { "epoch": 1.2047778777714147, "grad_norm": 0.008765246719121933, "learning_rate": 8.816788494799125e-06, "loss": 0.0011, "step": 73630 }, { "epoch": 1.2049415037224904, "grad_norm": 0.014652698300778866, "learning_rate": 8.816327259436513e-06, "loss": 0.0027, "step": 73640 }, { "epoch": 1.2051051296735662, "grad_norm": 0.03761477768421173, "learning_rate": 8.815865946262233e-06, "loss": 0.0005, "step": 73650 }, { "epoch": 1.205268755624642, "grad_norm": 0.1228545606136322, "learning_rate": 8.81540455528569e-06, "loss": 0.002, "step": 73660 }, { "epoch": 1.205432381575718, "grad_norm": 0.025270354002714157, "learning_rate": 8.814943086516294e-06, "loss": 0.0022, "step": 73670 }, { "epoch": 1.2055960075267937, "grad_norm": 0.05559875816106796, "learning_rate": 8.814481539963456e-06, "loss": 0.0018, "step": 73680 }, { "epoch": 1.2057596334778695, "grad_norm": 0.06110501289367676, "learning_rate": 8.814019915636581e-06, "loss": 0.003, "step": 73690 }, { "epoch": 1.2059232594289455, "grad_norm": 0.020968761295080185, "learning_rate": 8.813558213545082e-06, "loss": 0.0022, "step": 73700 }, { "epoch": 1.2060868853800213, "grad_norm": 0.007280793972313404, "learning_rate": 8.813096433698377e-06, "loss": 0.0014, "step": 73710 }, { "epoch": 1.206250511331097, "grad_norm": 0.12524037063121796, "learning_rate": 8.812634576105877e-06, "loss": 0.0027, "step": 73720 }, { "epoch": 1.2064141372821728, "grad_norm": 0.10710034519433975, "learning_rate": 8.812172640777001e-06, "loss": 0.0027, "step": 73730 }, { "epoch": 1.2065777632332488, "grad_norm": 0.008820896968245506, "learning_rate": 8.811710627721167e-06, "loss": 0.0016, "step": 73740 }, { "epoch": 1.2067413891843246, "grad_norm": 0.005418422631919384, "learning_rate": 8.811248536947795e-06, "loss": 0.0015, "step": 73750 }, { "epoch": 1.2069050151354004, "grad_norm": 0.12339776754379272, "learning_rate": 8.810786368466307e-06, "loss": 0.0014, "step": 73760 }, { "epoch": 1.2070686410864764, "grad_norm": 0.06493106484413147, "learning_rate": 8.810324122286125e-06, "loss": 0.0016, "step": 73770 }, { "epoch": 1.2072322670375522, "grad_norm": 0.050852540880441666, "learning_rate": 8.809861798416675e-06, "loss": 0.001, "step": 73780 }, { "epoch": 1.207395892988628, "grad_norm": 0.0953221321105957, "learning_rate": 8.809399396867384e-06, "loss": 0.0017, "step": 73790 }, { "epoch": 1.207559518939704, "grad_norm": 0.015691125765442848, "learning_rate": 8.808936917647678e-06, "loss": 0.0008, "step": 73800 }, { "epoch": 1.2077231448907797, "grad_norm": 0.020932042971253395, "learning_rate": 8.808474360766988e-06, "loss": 0.0013, "step": 73810 }, { "epoch": 1.2078867708418555, "grad_norm": 0.07496188580989838, "learning_rate": 8.808011726234746e-06, "loss": 0.0017, "step": 73820 }, { "epoch": 1.2080503967929315, "grad_norm": 0.14662525057792664, "learning_rate": 8.807549014060382e-06, "loss": 0.0028, "step": 73830 }, { "epoch": 1.2082140227440072, "grad_norm": 0.049506090581417084, "learning_rate": 8.807086224253332e-06, "loss": 0.0033, "step": 73840 }, { "epoch": 1.208377648695083, "grad_norm": 0.03194468468427658, "learning_rate": 8.806623356823032e-06, "loss": 0.0019, "step": 73850 }, { "epoch": 1.2085412746461588, "grad_norm": 0.13135653734207153, "learning_rate": 8.806160411778919e-06, "loss": 0.0013, "step": 73860 }, { "epoch": 1.2087049005972348, "grad_norm": 0.04421916604042053, "learning_rate": 8.805697389130432e-06, "loss": 0.0014, "step": 73870 }, { "epoch": 1.2088685265483106, "grad_norm": 0.010858963243663311, "learning_rate": 8.805234288887012e-06, "loss": 0.0018, "step": 73880 }, { "epoch": 1.2090321524993863, "grad_norm": 0.00853346474468708, "learning_rate": 8.804771111058103e-06, "loss": 0.0026, "step": 73890 }, { "epoch": 1.2091957784504623, "grad_norm": 0.003607161808758974, "learning_rate": 8.804307855653145e-06, "loss": 0.0009, "step": 73900 }, { "epoch": 1.209359404401538, "grad_norm": 0.14449089765548706, "learning_rate": 8.803844522681584e-06, "loss": 0.0037, "step": 73910 }, { "epoch": 1.2095230303526139, "grad_norm": 0.0647694543004036, "learning_rate": 8.80338111215287e-06, "loss": 0.0019, "step": 73920 }, { "epoch": 1.2096866563036897, "grad_norm": 0.08556487411260605, "learning_rate": 8.80291762407645e-06, "loss": 0.0017, "step": 73930 }, { "epoch": 1.2098502822547657, "grad_norm": 0.15589919686317444, "learning_rate": 8.802454058461774e-06, "loss": 0.002, "step": 73940 }, { "epoch": 1.2100139082058414, "grad_norm": 0.05364708602428436, "learning_rate": 8.801990415318293e-06, "loss": 0.0013, "step": 73950 }, { "epoch": 1.2101775341569172, "grad_norm": 0.09204890578985214, "learning_rate": 8.80152669465546e-06, "loss": 0.0023, "step": 73960 }, { "epoch": 1.2103411601079932, "grad_norm": 0.021671351045370102, "learning_rate": 8.801062896482733e-06, "loss": 0.0018, "step": 73970 }, { "epoch": 1.210504786059069, "grad_norm": 0.04151728004217148, "learning_rate": 8.800599020809566e-06, "loss": 0.0039, "step": 73980 }, { "epoch": 1.2106684120101447, "grad_norm": 0.0569857582449913, "learning_rate": 8.800135067645415e-06, "loss": 0.0029, "step": 73990 }, { "epoch": 1.2108320379612207, "grad_norm": 0.05066404119133949, "learning_rate": 8.799671036999743e-06, "loss": 0.0017, "step": 74000 }, { "epoch": 1.2109956639122965, "grad_norm": 0.02069631963968277, "learning_rate": 8.79920692888201e-06, "loss": 0.0017, "step": 74010 }, { "epoch": 1.2111592898633723, "grad_norm": 0.04496379569172859, "learning_rate": 8.79874274330168e-06, "loss": 0.0032, "step": 74020 }, { "epoch": 1.2113229158144483, "grad_norm": 0.11580649763345718, "learning_rate": 8.798278480268215e-06, "loss": 0.0015, "step": 74030 }, { "epoch": 1.211486541765524, "grad_norm": 0.07541324198246002, "learning_rate": 8.797814139791081e-06, "loss": 0.0019, "step": 74040 }, { "epoch": 1.2116501677165998, "grad_norm": 0.027969639748334885, "learning_rate": 8.797349721879749e-06, "loss": 0.0021, "step": 74050 }, { "epoch": 1.2118137936676756, "grad_norm": 0.1523810178041458, "learning_rate": 8.796885226543682e-06, "loss": 0.0024, "step": 74060 }, { "epoch": 1.2119774196187516, "grad_norm": 0.16077792644500732, "learning_rate": 8.796420653792359e-06, "loss": 0.0014, "step": 74070 }, { "epoch": 1.2121410455698274, "grad_norm": 0.08254176378250122, "learning_rate": 8.795956003635245e-06, "loss": 0.0017, "step": 74080 }, { "epoch": 1.2123046715209032, "grad_norm": 0.07417170703411102, "learning_rate": 8.795491276081818e-06, "loss": 0.0033, "step": 74090 }, { "epoch": 1.2124682974719792, "grad_norm": 0.14428596198558807, "learning_rate": 8.79502647114155e-06, "loss": 0.0023, "step": 74100 }, { "epoch": 1.212631923423055, "grad_norm": 0.09734809398651123, "learning_rate": 8.79456158882392e-06, "loss": 0.002, "step": 74110 }, { "epoch": 1.2127955493741307, "grad_norm": 0.103830985724926, "learning_rate": 8.794096629138407e-06, "loss": 0.0021, "step": 74120 }, { "epoch": 1.2129591753252065, "grad_norm": 0.06322076171636581, "learning_rate": 8.793631592094492e-06, "loss": 0.0024, "step": 74130 }, { "epoch": 1.2131228012762825, "grad_norm": 0.06458721309900284, "learning_rate": 8.793166477701653e-06, "loss": 0.0011, "step": 74140 }, { "epoch": 1.2132864272273582, "grad_norm": 0.054188940674066544, "learning_rate": 8.792701285969376e-06, "loss": 0.0014, "step": 74150 }, { "epoch": 1.213450053178434, "grad_norm": 0.02743646875023842, "learning_rate": 8.792236016907146e-06, "loss": 0.0019, "step": 74160 }, { "epoch": 1.21361367912951, "grad_norm": 0.14492148160934448, "learning_rate": 8.791770670524448e-06, "loss": 0.0038, "step": 74170 }, { "epoch": 1.2137773050805858, "grad_norm": 0.034757085144519806, "learning_rate": 8.79130524683077e-06, "loss": 0.0011, "step": 74180 }, { "epoch": 1.2139409310316616, "grad_norm": 0.04952031373977661, "learning_rate": 8.790839745835604e-06, "loss": 0.0013, "step": 74190 }, { "epoch": 1.2141045569827376, "grad_norm": 0.019880348816514015, "learning_rate": 8.790374167548438e-06, "loss": 0.0014, "step": 74200 }, { "epoch": 1.2142681829338133, "grad_norm": 0.06924626976251602, "learning_rate": 8.789908511978769e-06, "loss": 0.0018, "step": 74210 }, { "epoch": 1.214431808884889, "grad_norm": 0.14408650994300842, "learning_rate": 8.789442779136086e-06, "loss": 0.0022, "step": 74220 }, { "epoch": 1.214595434835965, "grad_norm": 0.06332434713840485, "learning_rate": 8.788976969029889e-06, "loss": 0.0017, "step": 74230 }, { "epoch": 1.2147590607870409, "grad_norm": 0.030079122632741928, "learning_rate": 8.788511081669676e-06, "loss": 0.0048, "step": 74240 }, { "epoch": 1.2149226867381167, "grad_norm": 0.04262183606624603, "learning_rate": 8.78804511706494e-06, "loss": 0.003, "step": 74250 }, { "epoch": 1.2150863126891924, "grad_norm": 0.10590803623199463, "learning_rate": 8.787579075225187e-06, "loss": 0.002, "step": 74260 }, { "epoch": 1.2152499386402684, "grad_norm": 0.025951744988560677, "learning_rate": 8.787112956159919e-06, "loss": 0.0025, "step": 74270 }, { "epoch": 1.2154135645913442, "grad_norm": 0.1237805038690567, "learning_rate": 8.786646759878638e-06, "loss": 0.0018, "step": 74280 }, { "epoch": 1.21557719054242, "grad_norm": 0.21523351967334747, "learning_rate": 8.78618048639085e-06, "loss": 0.0022, "step": 74290 }, { "epoch": 1.2157408164934957, "grad_norm": 0.09758035093545914, "learning_rate": 8.785714135706062e-06, "loss": 0.0014, "step": 74300 }, { "epoch": 1.2159044424445717, "grad_norm": 0.06799943000078201, "learning_rate": 8.785247707833783e-06, "loss": 0.002, "step": 74310 }, { "epoch": 1.2160680683956475, "grad_norm": 0.029857197776436806, "learning_rate": 8.784781202783522e-06, "loss": 0.002, "step": 74320 }, { "epoch": 1.2162316943467233, "grad_norm": 0.10155349224805832, "learning_rate": 8.784314620564789e-06, "loss": 0.0046, "step": 74330 }, { "epoch": 1.2163953202977993, "grad_norm": 0.06691360473632812, "learning_rate": 8.783847961187102e-06, "loss": 0.0012, "step": 74340 }, { "epoch": 1.216558946248875, "grad_norm": 0.2538083791732788, "learning_rate": 8.783381224659973e-06, "loss": 0.0013, "step": 74350 }, { "epoch": 1.2167225721999508, "grad_norm": 0.12512780725955963, "learning_rate": 8.782914410992917e-06, "loss": 0.0016, "step": 74360 }, { "epoch": 1.2168861981510268, "grad_norm": 0.07931777834892273, "learning_rate": 8.782447520195453e-06, "loss": 0.0019, "step": 74370 }, { "epoch": 1.2170498241021026, "grad_norm": 0.027043387293815613, "learning_rate": 8.781980552277103e-06, "loss": 0.0014, "step": 74380 }, { "epoch": 1.2172134500531784, "grad_norm": 0.09436113387346268, "learning_rate": 8.781513507247384e-06, "loss": 0.0015, "step": 74390 }, { "epoch": 1.2173770760042544, "grad_norm": 0.030254250392317772, "learning_rate": 8.781046385115822e-06, "loss": 0.0015, "step": 74400 }, { "epoch": 1.2175407019553302, "grad_norm": 0.10930409282445908, "learning_rate": 8.78057918589194e-06, "loss": 0.0018, "step": 74410 }, { "epoch": 1.217704327906406, "grad_norm": 0.08458217978477478, "learning_rate": 8.780111909585262e-06, "loss": 0.0019, "step": 74420 }, { "epoch": 1.2178679538574817, "grad_norm": 0.064837746322155, "learning_rate": 8.77964455620532e-06, "loss": 0.0017, "step": 74430 }, { "epoch": 1.2180315798085577, "grad_norm": 0.06945952028036118, "learning_rate": 8.779177125761637e-06, "loss": 0.0021, "step": 74440 }, { "epoch": 1.2181952057596335, "grad_norm": 0.0979158878326416, "learning_rate": 8.778709618263749e-06, "loss": 0.0025, "step": 74450 }, { "epoch": 1.2183588317107092, "grad_norm": 0.0679890587925911, "learning_rate": 8.778242033721185e-06, "loss": 0.0018, "step": 74460 }, { "epoch": 1.2185224576617852, "grad_norm": 0.03019220381975174, "learning_rate": 8.777774372143478e-06, "loss": 0.0017, "step": 74470 }, { "epoch": 1.218686083612861, "grad_norm": 0.05549831688404083, "learning_rate": 8.777306633540164e-06, "loss": 0.002, "step": 74480 }, { "epoch": 1.2188497095639368, "grad_norm": 0.11253856122493744, "learning_rate": 8.776838817920782e-06, "loss": 0.0018, "step": 74490 }, { "epoch": 1.2190133355150126, "grad_norm": 0.04143275320529938, "learning_rate": 8.776370925294867e-06, "loss": 0.003, "step": 74500 }, { "epoch": 1.2191769614660886, "grad_norm": 0.056609757244586945, "learning_rate": 8.775902955671962e-06, "loss": 0.0022, "step": 74510 }, { "epoch": 1.2193405874171643, "grad_norm": 0.06458990275859833, "learning_rate": 8.775434909061606e-06, "loss": 0.0013, "step": 74520 }, { "epoch": 1.21950421336824, "grad_norm": 0.05513772740960121, "learning_rate": 8.774966785473344e-06, "loss": 0.0031, "step": 74530 }, { "epoch": 1.219667839319316, "grad_norm": 0.17733044922351837, "learning_rate": 8.774498584916718e-06, "loss": 0.0012, "step": 74540 }, { "epoch": 1.2198314652703919, "grad_norm": 0.28803586959838867, "learning_rate": 8.774030307401279e-06, "loss": 0.0015, "step": 74550 }, { "epoch": 1.2199950912214677, "grad_norm": 0.03486178442835808, "learning_rate": 8.773561952936569e-06, "loss": 0.0016, "step": 74560 }, { "epoch": 1.2201587171725436, "grad_norm": 0.04698203131556511, "learning_rate": 8.773093521532141e-06, "loss": 0.0015, "step": 74570 }, { "epoch": 1.2203223431236194, "grad_norm": 0.12008313834667206, "learning_rate": 8.772625013197544e-06, "loss": 0.0033, "step": 74580 }, { "epoch": 1.2204859690746952, "grad_norm": 0.025358371436595917, "learning_rate": 8.772156427942333e-06, "loss": 0.003, "step": 74590 }, { "epoch": 1.2206495950257712, "grad_norm": 0.09735561162233353, "learning_rate": 8.77168776577606e-06, "loss": 0.0019, "step": 74600 }, { "epoch": 1.220813220976847, "grad_norm": 0.12011044472455978, "learning_rate": 8.77121902670828e-06, "loss": 0.0014, "step": 74610 }, { "epoch": 1.2209768469279227, "grad_norm": 0.05631161481142044, "learning_rate": 8.770750210748554e-06, "loss": 0.0031, "step": 74620 }, { "epoch": 1.2211404728789985, "grad_norm": 0.06336871534585953, "learning_rate": 8.770281317906435e-06, "loss": 0.002, "step": 74630 }, { "epoch": 1.2213040988300745, "grad_norm": 0.07017950713634491, "learning_rate": 8.769812348191489e-06, "loss": 0.0021, "step": 74640 }, { "epoch": 1.2214677247811503, "grad_norm": 0.07778137177228928, "learning_rate": 8.769343301613274e-06, "loss": 0.0021, "step": 74650 }, { "epoch": 1.221631350732226, "grad_norm": 0.2997229993343353, "learning_rate": 8.768874178181355e-06, "loss": 0.0025, "step": 74660 }, { "epoch": 1.221794976683302, "grad_norm": 0.04716581106185913, "learning_rate": 8.768404977905297e-06, "loss": 0.0011, "step": 74670 }, { "epoch": 1.2219586026343778, "grad_norm": 0.17038597166538239, "learning_rate": 8.767935700794666e-06, "loss": 0.0034, "step": 74680 }, { "epoch": 1.2221222285854536, "grad_norm": 0.10748691111803055, "learning_rate": 8.767466346859032e-06, "loss": 0.0025, "step": 74690 }, { "epoch": 1.2222858545365294, "grad_norm": 0.047815896570682526, "learning_rate": 8.766996916107962e-06, "loss": 0.0009, "step": 74700 }, { "epoch": 1.2224494804876054, "grad_norm": 0.06666827201843262, "learning_rate": 8.766527408551028e-06, "loss": 0.0027, "step": 74710 }, { "epoch": 1.2226131064386812, "grad_norm": 0.11897066980600357, "learning_rate": 8.766057824197804e-06, "loss": 0.0014, "step": 74720 }, { "epoch": 1.222776732389757, "grad_norm": 0.036791834980249405, "learning_rate": 8.765588163057866e-06, "loss": 0.0009, "step": 74730 }, { "epoch": 1.222940358340833, "grad_norm": 0.028227295726537704, "learning_rate": 8.765118425140784e-06, "loss": 0.002, "step": 74740 }, { "epoch": 1.2231039842919087, "grad_norm": 0.1595212072134018, "learning_rate": 8.764648610456142e-06, "loss": 0.0023, "step": 74750 }, { "epoch": 1.2232676102429845, "grad_norm": 0.13113871216773987, "learning_rate": 8.764178719013516e-06, "loss": 0.0047, "step": 74760 }, { "epoch": 1.2234312361940605, "grad_norm": 0.09017467498779297, "learning_rate": 8.763708750822487e-06, "loss": 0.0024, "step": 74770 }, { "epoch": 1.2235948621451362, "grad_norm": 0.049713000655174255, "learning_rate": 8.763238705892638e-06, "loss": 0.0016, "step": 74780 }, { "epoch": 1.223758488096212, "grad_norm": 0.016117015853524208, "learning_rate": 8.762768584233551e-06, "loss": 0.0028, "step": 74790 }, { "epoch": 1.223922114047288, "grad_norm": 0.04766162857413292, "learning_rate": 8.762298385854814e-06, "loss": 0.0012, "step": 74800 }, { "epoch": 1.2240857399983638, "grad_norm": 0.050139714032411575, "learning_rate": 8.761828110766014e-06, "loss": 0.0014, "step": 74810 }, { "epoch": 1.2242493659494396, "grad_norm": 0.005717010702937841, "learning_rate": 8.761357758976737e-06, "loss": 0.0015, "step": 74820 }, { "epoch": 1.2244129919005153, "grad_norm": 0.03825102746486664, "learning_rate": 8.760887330496573e-06, "loss": 0.0011, "step": 74830 }, { "epoch": 1.2245766178515913, "grad_norm": 0.002105529885739088, "learning_rate": 8.760416825335115e-06, "loss": 0.001, "step": 74840 }, { "epoch": 1.224740243802667, "grad_norm": 0.1419937014579773, "learning_rate": 8.759946243501957e-06, "loss": 0.0017, "step": 74850 }, { "epoch": 1.2249038697537429, "grad_norm": 0.051051005721092224, "learning_rate": 8.759475585006691e-06, "loss": 0.0011, "step": 74860 }, { "epoch": 1.2250674957048189, "grad_norm": 0.027183406054973602, "learning_rate": 8.759004849858918e-06, "loss": 0.0016, "step": 74870 }, { "epoch": 1.2252311216558947, "grad_norm": 0.039614252746105194, "learning_rate": 8.758534038068231e-06, "loss": 0.002, "step": 74880 }, { "epoch": 1.2253947476069704, "grad_norm": 0.19739028811454773, "learning_rate": 8.758063149644232e-06, "loss": 0.0025, "step": 74890 }, { "epoch": 1.2255583735580462, "grad_norm": 0.17652630805969238, "learning_rate": 8.757592184596522e-06, "loss": 0.0022, "step": 74900 }, { "epoch": 1.2257219995091222, "grad_norm": 0.020397275686264038, "learning_rate": 8.757121142934703e-06, "loss": 0.002, "step": 74910 }, { "epoch": 1.225885625460198, "grad_norm": 0.04414468631148338, "learning_rate": 8.756650024668378e-06, "loss": 0.0019, "step": 74920 }, { "epoch": 1.2260492514112737, "grad_norm": 0.07663978636264801, "learning_rate": 8.756178829807156e-06, "loss": 0.0016, "step": 74930 }, { "epoch": 1.2262128773623497, "grad_norm": 0.006580832414329052, "learning_rate": 8.75570755836064e-06, "loss": 0.0019, "step": 74940 }, { "epoch": 1.2263765033134255, "grad_norm": 0.013975769281387329, "learning_rate": 8.755236210338442e-06, "loss": 0.001, "step": 74950 }, { "epoch": 1.2265401292645013, "grad_norm": 0.060750801116228104, "learning_rate": 8.754764785750172e-06, "loss": 0.0023, "step": 74960 }, { "epoch": 1.2267037552155773, "grad_norm": 0.03562282770872116, "learning_rate": 8.754293284605441e-06, "loss": 0.001, "step": 74970 }, { "epoch": 1.226867381166653, "grad_norm": 0.180167555809021, "learning_rate": 8.753821706913862e-06, "loss": 0.0021, "step": 74980 }, { "epoch": 1.2270310071177288, "grad_norm": 0.0366571880877018, "learning_rate": 8.753350052685051e-06, "loss": 0.0031, "step": 74990 }, { "epoch": 1.2271946330688048, "grad_norm": 0.11993511766195297, "learning_rate": 8.752878321928626e-06, "loss": 0.0026, "step": 75000 }, { "epoch": 1.2273582590198806, "grad_norm": 0.11462995409965515, "learning_rate": 8.7524065146542e-06, "loss": 0.0017, "step": 75010 }, { "epoch": 1.2275218849709564, "grad_norm": 0.12276584655046463, "learning_rate": 8.751934630871401e-06, "loss": 0.0019, "step": 75020 }, { "epoch": 1.2276855109220322, "grad_norm": 0.03301592543721199, "learning_rate": 8.751462670589844e-06, "loss": 0.0015, "step": 75030 }, { "epoch": 1.2278491368731081, "grad_norm": 0.07263326644897461, "learning_rate": 8.750990633819155e-06, "loss": 0.0036, "step": 75040 }, { "epoch": 1.228012762824184, "grad_norm": 0.030115583911538124, "learning_rate": 8.750518520568954e-06, "loss": 0.0017, "step": 75050 }, { "epoch": 1.2281763887752597, "grad_norm": 0.029740046709775925, "learning_rate": 8.750046330848872e-06, "loss": 0.0025, "step": 75060 }, { "epoch": 1.2283400147263357, "grad_norm": 0.04408130422234535, "learning_rate": 8.749574064668535e-06, "loss": 0.0015, "step": 75070 }, { "epoch": 1.2285036406774115, "grad_norm": 0.06939876824617386, "learning_rate": 8.749101722037571e-06, "loss": 0.0018, "step": 75080 }, { "epoch": 1.2286672666284872, "grad_norm": 0.22061368823051453, "learning_rate": 8.748629302965612e-06, "loss": 0.0015, "step": 75090 }, { "epoch": 1.228830892579563, "grad_norm": 0.02691456861793995, "learning_rate": 8.748156807462289e-06, "loss": 0.0013, "step": 75100 }, { "epoch": 1.228994518530639, "grad_norm": 0.07283743470907211, "learning_rate": 8.747684235537236e-06, "loss": 0.0012, "step": 75110 }, { "epoch": 1.2291581444817148, "grad_norm": 0.04394949600100517, "learning_rate": 8.747211587200088e-06, "loss": 0.0016, "step": 75120 }, { "epoch": 1.2293217704327906, "grad_norm": 0.03931640088558197, "learning_rate": 8.746738862460485e-06, "loss": 0.0029, "step": 75130 }, { "epoch": 1.2294853963838666, "grad_norm": 0.03931419551372528, "learning_rate": 8.746266061328062e-06, "loss": 0.0022, "step": 75140 }, { "epoch": 1.2296490223349423, "grad_norm": 0.025415226817131042, "learning_rate": 8.74579318381246e-06, "loss": 0.0013, "step": 75150 }, { "epoch": 1.229812648286018, "grad_norm": 0.02864990010857582, "learning_rate": 8.74532022992332e-06, "loss": 0.0022, "step": 75160 }, { "epoch": 1.229976274237094, "grad_norm": 0.016696926206350327, "learning_rate": 8.744847199670287e-06, "loss": 0.0012, "step": 75170 }, { "epoch": 1.2301399001881699, "grad_norm": 0.09827528893947601, "learning_rate": 8.744374093063004e-06, "loss": 0.0015, "step": 75180 }, { "epoch": 1.2303035261392457, "grad_norm": 0.02489851787686348, "learning_rate": 8.743900910111118e-06, "loss": 0.0011, "step": 75190 }, { "epoch": 1.2304671520903216, "grad_norm": 0.162788987159729, "learning_rate": 8.743427650824276e-06, "loss": 0.0021, "step": 75200 }, { "epoch": 1.2306307780413974, "grad_norm": 0.07342688739299774, "learning_rate": 8.742954315212127e-06, "loss": 0.0013, "step": 75210 }, { "epoch": 1.2307944039924732, "grad_norm": 0.0743943378329277, "learning_rate": 8.742480903284326e-06, "loss": 0.0025, "step": 75220 }, { "epoch": 1.230958029943549, "grad_norm": 0.18022096157073975, "learning_rate": 8.742007415050518e-06, "loss": 0.002, "step": 75230 }, { "epoch": 1.231121655894625, "grad_norm": 0.00519242649897933, "learning_rate": 8.741533850520364e-06, "loss": 0.0012, "step": 75240 }, { "epoch": 1.2312852818457007, "grad_norm": 0.04183154180645943, "learning_rate": 8.741060209703515e-06, "loss": 0.0012, "step": 75250 }, { "epoch": 1.2314489077967765, "grad_norm": 0.12147953361272812, "learning_rate": 8.740586492609633e-06, "loss": 0.0027, "step": 75260 }, { "epoch": 1.2316125337478523, "grad_norm": 0.031106973066926003, "learning_rate": 8.74011269924837e-06, "loss": 0.0026, "step": 75270 }, { "epoch": 1.2317761596989283, "grad_norm": 0.03747696429491043, "learning_rate": 8.739638829629394e-06, "loss": 0.002, "step": 75280 }, { "epoch": 1.231939785650004, "grad_norm": 0.06885416060686111, "learning_rate": 8.739164883762358e-06, "loss": 0.0025, "step": 75290 }, { "epoch": 1.2321034116010798, "grad_norm": 0.04731731116771698, "learning_rate": 8.738690861656933e-06, "loss": 0.0016, "step": 75300 }, { "epoch": 1.2322670375521558, "grad_norm": 0.06719710677862167, "learning_rate": 8.738216763322781e-06, "loss": 0.0015, "step": 75310 }, { "epoch": 1.2324306635032316, "grad_norm": 0.06744828820228577, "learning_rate": 8.737742588769569e-06, "loss": 0.0036, "step": 75320 }, { "epoch": 1.2325942894543074, "grad_norm": 0.024281905964016914, "learning_rate": 8.737268338006962e-06, "loss": 0.0015, "step": 75330 }, { "epoch": 1.2327579154053834, "grad_norm": 0.08966421335935593, "learning_rate": 8.736794011044633e-06, "loss": 0.0015, "step": 75340 }, { "epoch": 1.2329215413564591, "grad_norm": 0.11703852564096451, "learning_rate": 8.736319607892255e-06, "loss": 0.002, "step": 75350 }, { "epoch": 1.233085167307535, "grad_norm": 0.04810117185115814, "learning_rate": 8.735845128559493e-06, "loss": 0.0012, "step": 75360 }, { "epoch": 1.233248793258611, "grad_norm": 0.08402729779481888, "learning_rate": 8.73537057305603e-06, "loss": 0.0016, "step": 75370 }, { "epoch": 1.2334124192096867, "grad_norm": 0.11834168434143066, "learning_rate": 8.734895941391535e-06, "loss": 0.0016, "step": 75380 }, { "epoch": 1.2335760451607625, "grad_norm": 0.04208146408200264, "learning_rate": 8.73442123357569e-06, "loss": 0.0062, "step": 75390 }, { "epoch": 1.2337396711118382, "grad_norm": 0.09679408371448517, "learning_rate": 8.733946449618172e-06, "loss": 0.0016, "step": 75400 }, { "epoch": 1.2339032970629142, "grad_norm": 0.09913047403097153, "learning_rate": 8.733471589528661e-06, "loss": 0.0016, "step": 75410 }, { "epoch": 1.23406692301399, "grad_norm": 0.00963651668280363, "learning_rate": 8.732996653316839e-06, "loss": 0.0016, "step": 75420 }, { "epoch": 1.2342305489650658, "grad_norm": 0.07294116169214249, "learning_rate": 8.73252164099239e-06, "loss": 0.0015, "step": 75430 }, { "epoch": 1.2343941749161418, "grad_norm": 0.026538247242569923, "learning_rate": 8.732046552565e-06, "loss": 0.0011, "step": 75440 }, { "epoch": 1.2345578008672176, "grad_norm": 0.012921475805342197, "learning_rate": 8.731571388044354e-06, "loss": 0.0013, "step": 75450 }, { "epoch": 1.2347214268182933, "grad_norm": 0.00819370523095131, "learning_rate": 8.731096147440141e-06, "loss": 0.0014, "step": 75460 }, { "epoch": 1.234885052769369, "grad_norm": 0.2725745737552643, "learning_rate": 8.730620830762053e-06, "loss": 0.0028, "step": 75470 }, { "epoch": 1.235048678720445, "grad_norm": 0.09809263795614243, "learning_rate": 8.730145438019776e-06, "loss": 0.0026, "step": 75480 }, { "epoch": 1.2352123046715209, "grad_norm": 0.03850580006837845, "learning_rate": 8.729669969223007e-06, "loss": 0.0017, "step": 75490 }, { "epoch": 1.2353759306225967, "grad_norm": 0.010635962709784508, "learning_rate": 8.72919442438144e-06, "loss": 0.0018, "step": 75500 }, { "epoch": 1.2355395565736726, "grad_norm": 0.03969942033290863, "learning_rate": 8.72871880350477e-06, "loss": 0.0024, "step": 75510 }, { "epoch": 1.2357031825247484, "grad_norm": 0.058856990188360214, "learning_rate": 8.728243106602694e-06, "loss": 0.0027, "step": 75520 }, { "epoch": 1.2358668084758242, "grad_norm": 0.08003389835357666, "learning_rate": 8.72776733368491e-06, "loss": 0.0017, "step": 75530 }, { "epoch": 1.2360304344269002, "grad_norm": 0.045158639550209045, "learning_rate": 8.727291484761124e-06, "loss": 0.0014, "step": 75540 }, { "epoch": 1.236194060377976, "grad_norm": 0.05608590319752693, "learning_rate": 8.726815559841035e-06, "loss": 0.0014, "step": 75550 }, { "epoch": 1.2363576863290517, "grad_norm": 0.04187047481536865, "learning_rate": 8.726339558934343e-06, "loss": 0.0013, "step": 75560 }, { "epoch": 1.2365213122801277, "grad_norm": 0.06300738453865051, "learning_rate": 8.725863482050758e-06, "loss": 0.0014, "step": 75570 }, { "epoch": 1.2366849382312035, "grad_norm": 0.10617884248495102, "learning_rate": 8.725387329199986e-06, "loss": 0.0024, "step": 75580 }, { "epoch": 1.2368485641822793, "grad_norm": 0.03172844648361206, "learning_rate": 8.724911100391733e-06, "loss": 0.0015, "step": 75590 }, { "epoch": 1.237012190133355, "grad_norm": 0.0706271231174469, "learning_rate": 8.724434795635712e-06, "loss": 0.0015, "step": 75600 }, { "epoch": 1.237175816084431, "grad_norm": 0.012056001462042332, "learning_rate": 8.723958414941634e-06, "loss": 0.0013, "step": 75610 }, { "epoch": 1.2373394420355068, "grad_norm": 0.10704883188009262, "learning_rate": 8.723481958319209e-06, "loss": 0.0013, "step": 75620 }, { "epoch": 1.2375030679865826, "grad_norm": 0.05322526395320892, "learning_rate": 8.723005425778155e-06, "loss": 0.0015, "step": 75630 }, { "epoch": 1.2376666939376586, "grad_norm": 0.05972364917397499, "learning_rate": 8.722528817328186e-06, "loss": 0.0016, "step": 75640 }, { "epoch": 1.2378303198887344, "grad_norm": 0.031377919018268585, "learning_rate": 8.72205213297902e-06, "loss": 0.0017, "step": 75650 }, { "epoch": 1.2379939458398102, "grad_norm": 0.1943347454071045, "learning_rate": 8.721575372740377e-06, "loss": 0.0035, "step": 75660 }, { "epoch": 1.238157571790886, "grad_norm": 0.03150485083460808, "learning_rate": 8.721098536621975e-06, "loss": 0.0011, "step": 75670 }, { "epoch": 1.238321197741962, "grad_norm": 0.03272932767868042, "learning_rate": 8.720621624633542e-06, "loss": 0.0017, "step": 75680 }, { "epoch": 1.2384848236930377, "grad_norm": 0.2795042097568512, "learning_rate": 8.720144636784796e-06, "loss": 0.0013, "step": 75690 }, { "epoch": 1.2386484496441135, "grad_norm": 0.0709453597664833, "learning_rate": 8.719667573085467e-06, "loss": 0.0025, "step": 75700 }, { "epoch": 1.2388120755951895, "grad_norm": 0.016088496893644333, "learning_rate": 8.719190433545276e-06, "loss": 0.0018, "step": 75710 }, { "epoch": 1.2389757015462652, "grad_norm": 0.042825847864151, "learning_rate": 8.718713218173958e-06, "loss": 0.0011, "step": 75720 }, { "epoch": 1.239139327497341, "grad_norm": 0.19395555555820465, "learning_rate": 8.71823592698124e-06, "loss": 0.0021, "step": 75730 }, { "epoch": 1.239302953448417, "grad_norm": 0.07232578098773956, "learning_rate": 8.717758559976853e-06, "loss": 0.0016, "step": 75740 }, { "epoch": 1.2394665793994928, "grad_norm": 0.014691956341266632, "learning_rate": 8.71728111717053e-06, "loss": 0.0011, "step": 75750 }, { "epoch": 1.2396302053505686, "grad_norm": 0.07408197969198227, "learning_rate": 8.716803598572008e-06, "loss": 0.0024, "step": 75760 }, { "epoch": 1.2397938313016446, "grad_norm": 0.05246606469154358, "learning_rate": 8.716326004191022e-06, "loss": 0.0029, "step": 75770 }, { "epoch": 1.2399574572527203, "grad_norm": 0.09673754870891571, "learning_rate": 8.715848334037307e-06, "loss": 0.0028, "step": 75780 }, { "epoch": 1.240121083203796, "grad_norm": 0.06974990665912628, "learning_rate": 8.715370588120607e-06, "loss": 0.0021, "step": 75790 }, { "epoch": 1.2402847091548719, "grad_norm": 0.058107901364564896, "learning_rate": 8.71489276645066e-06, "loss": 0.0014, "step": 75800 }, { "epoch": 1.2404483351059479, "grad_norm": 0.08627115190029144, "learning_rate": 8.714414869037209e-06, "loss": 0.0018, "step": 75810 }, { "epoch": 1.2406119610570236, "grad_norm": 0.05499175935983658, "learning_rate": 8.713936895889997e-06, "loss": 0.0012, "step": 75820 }, { "epoch": 1.2407755870080994, "grad_norm": 0.0612916462123394, "learning_rate": 8.71345884701877e-06, "loss": 0.0023, "step": 75830 }, { "epoch": 1.2409392129591754, "grad_norm": 0.15779739618301392, "learning_rate": 8.712980722433277e-06, "loss": 0.0019, "step": 75840 }, { "epoch": 1.2411028389102512, "grad_norm": 0.03836877644062042, "learning_rate": 8.712502522143264e-06, "loss": 0.0015, "step": 75850 }, { "epoch": 1.241266464861327, "grad_norm": 0.0926869660615921, "learning_rate": 8.712024246158482e-06, "loss": 0.0015, "step": 75860 }, { "epoch": 1.2414300908124027, "grad_norm": 0.09646128118038177, "learning_rate": 8.711545894488682e-06, "loss": 0.0017, "step": 75870 }, { "epoch": 1.2415937167634787, "grad_norm": 0.0858236700296402, "learning_rate": 8.711067467143617e-06, "loss": 0.0052, "step": 75880 }, { "epoch": 1.2417573427145545, "grad_norm": 0.07169157266616821, "learning_rate": 8.710588964133045e-06, "loss": 0.0027, "step": 75890 }, { "epoch": 1.2419209686656303, "grad_norm": 0.07202302664518356, "learning_rate": 8.710110385466717e-06, "loss": 0.0044, "step": 75900 }, { "epoch": 1.2420845946167063, "grad_norm": 0.036488957703113556, "learning_rate": 8.709631731154395e-06, "loss": 0.0021, "step": 75910 }, { "epoch": 1.242248220567782, "grad_norm": 0.12856771051883698, "learning_rate": 8.709153001205837e-06, "loss": 0.0019, "step": 75920 }, { "epoch": 1.2424118465188578, "grad_norm": 0.06917132437229156, "learning_rate": 8.708674195630805e-06, "loss": 0.0015, "step": 75930 }, { "epoch": 1.2425754724699338, "grad_norm": 0.011806569062173367, "learning_rate": 8.70819531443906e-06, "loss": 0.0019, "step": 75940 }, { "epoch": 1.2427390984210096, "grad_norm": 1.0186251401901245, "learning_rate": 8.707716357640363e-06, "loss": 0.0064, "step": 75950 }, { "epoch": 1.2429027243720854, "grad_norm": 0.02006656862795353, "learning_rate": 8.707237325244487e-06, "loss": 0.0015, "step": 75960 }, { "epoch": 1.2430663503231614, "grad_norm": 0.12216638028621674, "learning_rate": 8.70675821726119e-06, "loss": 0.0021, "step": 75970 }, { "epoch": 1.2432299762742371, "grad_norm": 0.042055729776620865, "learning_rate": 8.70627903370025e-06, "loss": 0.0025, "step": 75980 }, { "epoch": 1.243393602225313, "grad_norm": 0.2085392326116562, "learning_rate": 8.705799774571431e-06, "loss": 0.0026, "step": 75990 }, { "epoch": 1.2435572281763887, "grad_norm": 0.012148501351475716, "learning_rate": 8.705320439884505e-06, "loss": 0.0018, "step": 76000 }, { "epoch": 1.2437208541274647, "grad_norm": 0.011320688761770725, "learning_rate": 8.704841029649249e-06, "loss": 0.0016, "step": 76010 }, { "epoch": 1.2438844800785405, "grad_norm": 0.18356853723526, "learning_rate": 8.704361543875433e-06, "loss": 0.0021, "step": 76020 }, { "epoch": 1.2440481060296162, "grad_norm": 0.017837142571806908, "learning_rate": 8.703881982572836e-06, "loss": 0.0017, "step": 76030 }, { "epoch": 1.244211731980692, "grad_norm": 0.010370557196438313, "learning_rate": 8.703402345751237e-06, "loss": 0.0033, "step": 76040 }, { "epoch": 1.244375357931768, "grad_norm": 0.027000103145837784, "learning_rate": 8.702922633420412e-06, "loss": 0.0022, "step": 76050 }, { "epoch": 1.2445389838828438, "grad_norm": 0.02911590039730072, "learning_rate": 8.702442845590145e-06, "loss": 0.0018, "step": 76060 }, { "epoch": 1.2447026098339196, "grad_norm": 0.1347668170928955, "learning_rate": 8.701962982270216e-06, "loss": 0.0016, "step": 76070 }, { "epoch": 1.2448662357849956, "grad_norm": 0.05557018518447876, "learning_rate": 8.701483043470412e-06, "loss": 0.002, "step": 76080 }, { "epoch": 1.2450298617360713, "grad_norm": 0.010378395207226276, "learning_rate": 8.701003029200514e-06, "loss": 0.0016, "step": 76090 }, { "epoch": 1.245193487687147, "grad_norm": 0.06488751620054245, "learning_rate": 8.700522939470313e-06, "loss": 0.0018, "step": 76100 }, { "epoch": 1.245357113638223, "grad_norm": 0.09179822355508804, "learning_rate": 8.700042774289596e-06, "loss": 0.0025, "step": 76110 }, { "epoch": 1.2455207395892989, "grad_norm": 0.007916184142231941, "learning_rate": 8.699562533668155e-06, "loss": 0.0012, "step": 76120 }, { "epoch": 1.2456843655403746, "grad_norm": 0.0752815529704094, "learning_rate": 8.69908221761578e-06, "loss": 0.0015, "step": 76130 }, { "epoch": 1.2458479914914506, "grad_norm": 0.23923541605472565, "learning_rate": 8.698601826142265e-06, "loss": 0.0016, "step": 76140 }, { "epoch": 1.2460116174425264, "grad_norm": 0.2029772847890854, "learning_rate": 8.698121359257403e-06, "loss": 0.0013, "step": 76150 }, { "epoch": 1.2461752433936022, "grad_norm": 0.11279718577861786, "learning_rate": 8.697640816970993e-06, "loss": 0.001, "step": 76160 }, { "epoch": 1.2463388693446782, "grad_norm": 0.10722684860229492, "learning_rate": 8.69716019929283e-06, "loss": 0.0021, "step": 76170 }, { "epoch": 1.246502495295754, "grad_norm": 0.12818646430969238, "learning_rate": 8.696679506232714e-06, "loss": 0.0024, "step": 76180 }, { "epoch": 1.2466661212468297, "grad_norm": 0.06900682300329208, "learning_rate": 8.69619873780045e-06, "loss": 0.0038, "step": 76190 }, { "epoch": 1.2468297471979055, "grad_norm": 0.05835568532347679, "learning_rate": 8.695717894005836e-06, "loss": 0.0019, "step": 76200 }, { "epoch": 1.2469933731489815, "grad_norm": 0.24453505873680115, "learning_rate": 8.695236974858677e-06, "loss": 0.0014, "step": 76210 }, { "epoch": 1.2471569991000573, "grad_norm": 0.12088282406330109, "learning_rate": 8.694755980368778e-06, "loss": 0.0013, "step": 76220 }, { "epoch": 1.247320625051133, "grad_norm": 0.08328910917043686, "learning_rate": 8.694274910545947e-06, "loss": 0.0013, "step": 76230 }, { "epoch": 1.2474842510022088, "grad_norm": 0.04959578812122345, "learning_rate": 8.693793765399993e-06, "loss": 0.0009, "step": 76240 }, { "epoch": 1.2476478769532848, "grad_norm": 0.14766213297843933, "learning_rate": 8.693312544940725e-06, "loss": 0.0028, "step": 76250 }, { "epoch": 1.2478115029043606, "grad_norm": 0.1077360063791275, "learning_rate": 8.692831249177956e-06, "loss": 0.0016, "step": 76260 }, { "epoch": 1.2479751288554364, "grad_norm": 0.008973360992968082, "learning_rate": 8.692349878121498e-06, "loss": 0.0027, "step": 76270 }, { "epoch": 1.2481387548065124, "grad_norm": 0.06231240555644035, "learning_rate": 8.691868431781167e-06, "loss": 0.0014, "step": 76280 }, { "epoch": 1.2483023807575881, "grad_norm": 0.07380857318639755, "learning_rate": 8.691386910166776e-06, "loss": 0.0018, "step": 76290 }, { "epoch": 1.248466006708664, "grad_norm": 0.0651499554514885, "learning_rate": 8.69090531328815e-06, "loss": 0.0023, "step": 76300 }, { "epoch": 1.24862963265974, "grad_norm": 0.0871339812874794, "learning_rate": 8.6904236411551e-06, "loss": 0.0023, "step": 76310 }, { "epoch": 1.2487932586108157, "grad_norm": 0.04767316207289696, "learning_rate": 8.689941893777452e-06, "loss": 0.0027, "step": 76320 }, { "epoch": 1.2489568845618915, "grad_norm": 0.044882092624902725, "learning_rate": 8.689460071165026e-06, "loss": 0.0022, "step": 76330 }, { "epoch": 1.2491205105129675, "grad_norm": 0.08343739062547684, "learning_rate": 8.688978173327646e-06, "loss": 0.0017, "step": 76340 }, { "epoch": 1.2492841364640432, "grad_norm": 0.12629611790180206, "learning_rate": 8.68849620027514e-06, "loss": 0.0019, "step": 76350 }, { "epoch": 1.249447762415119, "grad_norm": 0.035439323633909225, "learning_rate": 8.688014152017332e-06, "loss": 0.0017, "step": 76360 }, { "epoch": 1.2496113883661948, "grad_norm": 0.037050761282444, "learning_rate": 8.687532028564054e-06, "loss": 0.0017, "step": 76370 }, { "epoch": 1.2497750143172708, "grad_norm": 0.09416184574365616, "learning_rate": 8.687049829925134e-06, "loss": 0.0037, "step": 76380 }, { "epoch": 1.2499386402683466, "grad_norm": 0.19351449608802795, "learning_rate": 8.686567556110402e-06, "loss": 0.0015, "step": 76390 }, { "epoch": 1.2501022662194223, "grad_norm": 0.09729155898094177, "learning_rate": 8.686085207129693e-06, "loss": 0.0017, "step": 76400 }, { "epoch": 1.250265892170498, "grad_norm": 0.21277952194213867, "learning_rate": 8.685602782992844e-06, "loss": 0.0014, "step": 76410 }, { "epoch": 1.250429518121574, "grad_norm": 0.11664614081382751, "learning_rate": 8.685120283709687e-06, "loss": 0.0013, "step": 76420 }, { "epoch": 1.2505931440726499, "grad_norm": 0.0276772603392601, "learning_rate": 8.68463770929006e-06, "loss": 0.0016, "step": 76430 }, { "epoch": 1.2507567700237257, "grad_norm": 0.05167677626013756, "learning_rate": 8.684155059743806e-06, "loss": 0.0025, "step": 76440 }, { "epoch": 1.2509203959748016, "grad_norm": 0.026932870969176292, "learning_rate": 8.683672335080762e-06, "loss": 0.0016, "step": 76450 }, { "epoch": 1.2510840219258774, "grad_norm": 0.06582721322774887, "learning_rate": 8.683189535310774e-06, "loss": 0.0008, "step": 76460 }, { "epoch": 1.2512476478769532, "grad_norm": 0.027993429452180862, "learning_rate": 8.682706660443682e-06, "loss": 0.001, "step": 76470 }, { "epoch": 1.2514112738280292, "grad_norm": 0.057259295135736465, "learning_rate": 8.682223710489333e-06, "loss": 0.0017, "step": 76480 }, { "epoch": 1.251574899779105, "grad_norm": 0.04753151535987854, "learning_rate": 8.681740685457575e-06, "loss": 0.002, "step": 76490 }, { "epoch": 1.2517385257301807, "grad_norm": 0.056612778455019, "learning_rate": 8.681257585358256e-06, "loss": 0.0012, "step": 76500 }, { "epoch": 1.2519021516812567, "grad_norm": 0.025252200663089752, "learning_rate": 8.680774410201225e-06, "loss": 0.001, "step": 76510 }, { "epoch": 1.2520657776323325, "grad_norm": 0.19275762140750885, "learning_rate": 8.680291159996334e-06, "loss": 0.002, "step": 76520 }, { "epoch": 1.2522294035834083, "grad_norm": 0.17066030204296112, "learning_rate": 8.679807834753438e-06, "loss": 0.002, "step": 76530 }, { "epoch": 1.2523930295344843, "grad_norm": 0.12389721721410751, "learning_rate": 8.679324434482388e-06, "loss": 0.0023, "step": 76540 }, { "epoch": 1.25255665548556, "grad_norm": 0.09339045733213425, "learning_rate": 8.678840959193042e-06, "loss": 0.0016, "step": 76550 }, { "epoch": 1.2527202814366358, "grad_norm": 0.08319197595119476, "learning_rate": 8.678357408895257e-06, "loss": 0.0017, "step": 76560 }, { "epoch": 1.2528839073877118, "grad_norm": 0.08488989621400833, "learning_rate": 8.677873783598895e-06, "loss": 0.0017, "step": 76570 }, { "epoch": 1.2530475333387876, "grad_norm": 0.050924502313137054, "learning_rate": 8.677390083313812e-06, "loss": 0.0015, "step": 76580 }, { "epoch": 1.2532111592898634, "grad_norm": 0.23560728132724762, "learning_rate": 8.676906308049875e-06, "loss": 0.0017, "step": 76590 }, { "epoch": 1.2533747852409391, "grad_norm": 0.10191647708415985, "learning_rate": 8.676422457816945e-06, "loss": 0.0023, "step": 76600 }, { "epoch": 1.253538411192015, "grad_norm": 0.38379254937171936, "learning_rate": 8.675938532624887e-06, "loss": 0.0019, "step": 76610 }, { "epoch": 1.253702037143091, "grad_norm": 0.23123204708099365, "learning_rate": 8.675454532483569e-06, "loss": 0.0026, "step": 76620 }, { "epoch": 1.2538656630941667, "grad_norm": 0.06593289971351624, "learning_rate": 8.674970457402859e-06, "loss": 0.0017, "step": 76630 }, { "epoch": 1.2540292890452425, "grad_norm": 0.3220798969268799, "learning_rate": 8.674486307392625e-06, "loss": 0.003, "step": 76640 }, { "epoch": 1.2541929149963185, "grad_norm": 0.052055541425943375, "learning_rate": 8.674002082462743e-06, "loss": 0.0019, "step": 76650 }, { "epoch": 1.2543565409473942, "grad_norm": 0.025026852265000343, "learning_rate": 8.67351778262308e-06, "loss": 0.0016, "step": 76660 }, { "epoch": 1.25452016689847, "grad_norm": 0.045580435544252396, "learning_rate": 8.673033407883517e-06, "loss": 0.002, "step": 76670 }, { "epoch": 1.254683792849546, "grad_norm": 0.04306081309914589, "learning_rate": 8.672548958253925e-06, "loss": 0.0027, "step": 76680 }, { "epoch": 1.2548474188006218, "grad_norm": 0.04201148822903633, "learning_rate": 8.672064433744182e-06, "loss": 0.0015, "step": 76690 }, { "epoch": 1.2550110447516976, "grad_norm": 0.06492362171411514, "learning_rate": 8.67157983436417e-06, "loss": 0.0013, "step": 76700 }, { "epoch": 1.2551746707027736, "grad_norm": 0.03235653042793274, "learning_rate": 8.671095160123768e-06, "loss": 0.0015, "step": 76710 }, { "epoch": 1.2553382966538493, "grad_norm": 0.009361272677779198, "learning_rate": 8.670610411032857e-06, "loss": 0.0027, "step": 76720 }, { "epoch": 1.255501922604925, "grad_norm": 0.054857805371284485, "learning_rate": 8.670125587101321e-06, "loss": 0.0019, "step": 76730 }, { "epoch": 1.255665548556001, "grad_norm": 0.11792723834514618, "learning_rate": 8.669640688339046e-06, "loss": 0.002, "step": 76740 }, { "epoch": 1.2558291745070769, "grad_norm": 0.05688590556383133, "learning_rate": 8.669155714755918e-06, "loss": 0.0024, "step": 76750 }, { "epoch": 1.2559928004581526, "grad_norm": 0.04952988028526306, "learning_rate": 8.668670666361828e-06, "loss": 0.0019, "step": 76760 }, { "epoch": 1.2561564264092284, "grad_norm": 0.1756308227777481, "learning_rate": 8.668185543166659e-06, "loss": 0.0028, "step": 76770 }, { "epoch": 1.2563200523603044, "grad_norm": 0.15459512174129486, "learning_rate": 8.667700345180309e-06, "loss": 0.001, "step": 76780 }, { "epoch": 1.2564836783113802, "grad_norm": 0.03290152922272682, "learning_rate": 8.667215072412667e-06, "loss": 0.0016, "step": 76790 }, { "epoch": 1.256647304262456, "grad_norm": 0.03934522718191147, "learning_rate": 8.66672972487363e-06, "loss": 0.002, "step": 76800 }, { "epoch": 1.2568109302135317, "grad_norm": 0.02399502694606781, "learning_rate": 8.666244302573092e-06, "loss": 0.0019, "step": 76810 }, { "epoch": 1.2569745561646077, "grad_norm": 0.03093862161040306, "learning_rate": 8.66575880552095e-06, "loss": 0.0013, "step": 76820 }, { "epoch": 1.2571381821156835, "grad_norm": 0.03681759163737297, "learning_rate": 8.665273233727105e-06, "loss": 0.0011, "step": 76830 }, { "epoch": 1.2573018080667593, "grad_norm": 0.024788374081254005, "learning_rate": 8.664787587201454e-06, "loss": 0.0022, "step": 76840 }, { "epoch": 1.2574654340178353, "grad_norm": 0.025767136365175247, "learning_rate": 8.664301865953902e-06, "loss": 0.0022, "step": 76850 }, { "epoch": 1.257629059968911, "grad_norm": 0.03997783735394478, "learning_rate": 8.663816069994351e-06, "loss": 0.0037, "step": 76860 }, { "epoch": 1.2577926859199868, "grad_norm": 0.03108169138431549, "learning_rate": 8.663330199332705e-06, "loss": 0.0016, "step": 76870 }, { "epoch": 1.2579563118710628, "grad_norm": 0.06428482383489609, "learning_rate": 8.662844253978873e-06, "loss": 0.0021, "step": 76880 }, { "epoch": 1.2581199378221386, "grad_norm": 0.07037154585123062, "learning_rate": 8.662358233942762e-06, "loss": 0.0022, "step": 76890 }, { "epoch": 1.2582835637732144, "grad_norm": 0.07668635994195938, "learning_rate": 8.661872139234282e-06, "loss": 0.0025, "step": 76900 }, { "epoch": 1.2584471897242904, "grad_norm": 0.2808360159397125, "learning_rate": 8.661385969863343e-06, "loss": 0.0018, "step": 76910 }, { "epoch": 1.2586108156753661, "grad_norm": 0.03235631063580513, "learning_rate": 8.660899725839857e-06, "loss": 0.001, "step": 76920 }, { "epoch": 1.258774441626442, "grad_norm": 0.104448601603508, "learning_rate": 8.66041340717374e-06, "loss": 0.0024, "step": 76930 }, { "epoch": 1.258938067577518, "grad_norm": 0.07909616827964783, "learning_rate": 8.659927013874907e-06, "loss": 0.002, "step": 76940 }, { "epoch": 1.2591016935285937, "grad_norm": 0.045962706208229065, "learning_rate": 8.659440545953275e-06, "loss": 0.0026, "step": 76950 }, { "epoch": 1.2592653194796695, "grad_norm": 0.010758241638541222, "learning_rate": 8.658954003418761e-06, "loss": 0.001, "step": 76960 }, { "epoch": 1.2594289454307452, "grad_norm": 0.0533389151096344, "learning_rate": 8.658467386281288e-06, "loss": 0.0017, "step": 76970 }, { "epoch": 1.2595925713818212, "grad_norm": 0.19015200436115265, "learning_rate": 8.657980694550777e-06, "loss": 0.0015, "step": 76980 }, { "epoch": 1.259756197332897, "grad_norm": 0.026806510984897614, "learning_rate": 8.657493928237148e-06, "loss": 0.0038, "step": 76990 }, { "epoch": 1.2599198232839728, "grad_norm": 0.07965224981307983, "learning_rate": 8.65700708735033e-06, "loss": 0.0032, "step": 77000 }, { "epoch": 1.2600834492350486, "grad_norm": 0.07183558493852615, "learning_rate": 8.656520171900247e-06, "loss": 0.0014, "step": 77010 }, { "epoch": 1.2602470751861246, "grad_norm": 0.11808524280786514, "learning_rate": 8.656033181896827e-06, "loss": 0.0019, "step": 77020 }, { "epoch": 1.2604107011372003, "grad_norm": 0.11986327916383743, "learning_rate": 8.655546117350002e-06, "loss": 0.002, "step": 77030 }, { "epoch": 1.260574327088276, "grad_norm": 0.09307223558425903, "learning_rate": 8.655058978269699e-06, "loss": 0.002, "step": 77040 }, { "epoch": 1.260737953039352, "grad_norm": 0.056172799319028854, "learning_rate": 8.654571764665853e-06, "loss": 0.0026, "step": 77050 }, { "epoch": 1.2609015789904279, "grad_norm": 0.05363352224230766, "learning_rate": 8.654084476548397e-06, "loss": 0.0022, "step": 77060 }, { "epoch": 1.2610652049415036, "grad_norm": 0.04858268424868584, "learning_rate": 8.653597113927264e-06, "loss": 0.0019, "step": 77070 }, { "epoch": 1.2612288308925796, "grad_norm": 0.047102659940719604, "learning_rate": 8.653109676812395e-06, "loss": 0.0017, "step": 77080 }, { "epoch": 1.2613924568436554, "grad_norm": 0.0017341122729703784, "learning_rate": 8.652622165213727e-06, "loss": 0.0016, "step": 77090 }, { "epoch": 1.2615560827947312, "grad_norm": 0.021400149911642075, "learning_rate": 8.652134579141198e-06, "loss": 0.0015, "step": 77100 }, { "epoch": 1.2617197087458072, "grad_norm": 0.12394706159830093, "learning_rate": 8.65164691860475e-06, "loss": 0.003, "step": 77110 }, { "epoch": 1.261883334696883, "grad_norm": 0.12874886393547058, "learning_rate": 8.651159183614331e-06, "loss": 0.0017, "step": 77120 }, { "epoch": 1.2620469606479587, "grad_norm": 0.056776173412799835, "learning_rate": 8.650671374179878e-06, "loss": 0.0024, "step": 77130 }, { "epoch": 1.2622105865990347, "grad_norm": 0.031202854588627815, "learning_rate": 8.650183490311341e-06, "loss": 0.0022, "step": 77140 }, { "epoch": 1.2623742125501105, "grad_norm": 0.09382439404726028, "learning_rate": 8.649695532018668e-06, "loss": 0.0017, "step": 77150 }, { "epoch": 1.2625378385011863, "grad_norm": 0.21161994338035583, "learning_rate": 8.649207499311805e-06, "loss": 0.0019, "step": 77160 }, { "epoch": 1.262701464452262, "grad_norm": 0.021838108077645302, "learning_rate": 8.648719392200706e-06, "loss": 0.0015, "step": 77170 }, { "epoch": 1.2628650904033378, "grad_norm": 0.09478498250246048, "learning_rate": 8.648231210695323e-06, "loss": 0.0023, "step": 77180 }, { "epoch": 1.2630287163544138, "grad_norm": 0.10853097587823868, "learning_rate": 8.647742954805606e-06, "loss": 0.0013, "step": 77190 }, { "epoch": 1.2631923423054896, "grad_norm": 0.12472327798604965, "learning_rate": 8.647254624541514e-06, "loss": 0.0039, "step": 77200 }, { "epoch": 1.2633559682565654, "grad_norm": 0.17706604301929474, "learning_rate": 8.646766219913003e-06, "loss": 0.0015, "step": 77210 }, { "epoch": 1.2635195942076414, "grad_norm": 0.07108800858259201, "learning_rate": 8.646277740930028e-06, "loss": 0.0024, "step": 77220 }, { "epoch": 1.2636832201587171, "grad_norm": 0.09686877578496933, "learning_rate": 8.64578918760255e-06, "loss": 0.0035, "step": 77230 }, { "epoch": 1.263846846109793, "grad_norm": 0.0508933961391449, "learning_rate": 8.645300559940535e-06, "loss": 0.002, "step": 77240 }, { "epoch": 1.264010472060869, "grad_norm": 0.04880641773343086, "learning_rate": 8.644811857953939e-06, "loss": 0.0012, "step": 77250 }, { "epoch": 1.2641740980119447, "grad_norm": 0.18240152299404144, "learning_rate": 8.64432308165273e-06, "loss": 0.0033, "step": 77260 }, { "epoch": 1.2643377239630205, "grad_norm": 0.11191031336784363, "learning_rate": 8.643834231046873e-06, "loss": 0.0024, "step": 77270 }, { "epoch": 1.2645013499140965, "grad_norm": 0.06383684277534485, "learning_rate": 8.643345306146334e-06, "loss": 0.0009, "step": 77280 }, { "epoch": 1.2646649758651722, "grad_norm": 0.05571810528635979, "learning_rate": 8.642856306961082e-06, "loss": 0.0013, "step": 77290 }, { "epoch": 1.264828601816248, "grad_norm": 0.20013701915740967, "learning_rate": 8.64236723350109e-06, "loss": 0.0064, "step": 77300 }, { "epoch": 1.264992227767324, "grad_norm": 0.05176997184753418, "learning_rate": 8.641878085776327e-06, "loss": 0.0014, "step": 77310 }, { "epoch": 1.2651558537183998, "grad_norm": 0.02704133279621601, "learning_rate": 8.641388863796767e-06, "loss": 0.0019, "step": 77320 }, { "epoch": 1.2653194796694756, "grad_norm": 0.11163891851902008, "learning_rate": 8.640899567572385e-06, "loss": 0.0017, "step": 77330 }, { "epoch": 1.2654831056205516, "grad_norm": 0.05666264146566391, "learning_rate": 8.640410197113158e-06, "loss": 0.0015, "step": 77340 }, { "epoch": 1.2656467315716273, "grad_norm": 0.03024539165198803, "learning_rate": 8.639920752429064e-06, "loss": 0.0014, "step": 77350 }, { "epoch": 1.265810357522703, "grad_norm": 0.07569063454866409, "learning_rate": 8.639431233530079e-06, "loss": 0.0016, "step": 77360 }, { "epoch": 1.2659739834737789, "grad_norm": 0.0467677041888237, "learning_rate": 8.638941640426187e-06, "loss": 0.0023, "step": 77370 }, { "epoch": 1.2661376094248546, "grad_norm": 0.019454944878816605, "learning_rate": 8.63845197312737e-06, "loss": 0.0039, "step": 77380 }, { "epoch": 1.2663012353759306, "grad_norm": 0.011508814990520477, "learning_rate": 8.637962231643613e-06, "loss": 0.002, "step": 77390 }, { "epoch": 1.2664648613270064, "grad_norm": 0.05841073393821716, "learning_rate": 8.637472415984897e-06, "loss": 0.0022, "step": 77400 }, { "epoch": 1.2666284872780822, "grad_norm": 0.17088767886161804, "learning_rate": 8.636982526161215e-06, "loss": 0.0025, "step": 77410 }, { "epoch": 1.2667921132291582, "grad_norm": 0.0066193630918860435, "learning_rate": 8.636492562182553e-06, "loss": 0.0009, "step": 77420 }, { "epoch": 1.266955739180234, "grad_norm": 0.10122372955083847, "learning_rate": 8.636002524058897e-06, "loss": 0.0019, "step": 77430 }, { "epoch": 1.2671193651313097, "grad_norm": 0.1770397126674652, "learning_rate": 8.635512411800245e-06, "loss": 0.0019, "step": 77440 }, { "epoch": 1.2672829910823857, "grad_norm": 0.043190423399209976, "learning_rate": 8.635022225416585e-06, "loss": 0.002, "step": 77450 }, { "epoch": 1.2674466170334615, "grad_norm": 0.06243487447500229, "learning_rate": 8.634531964917914e-06, "loss": 0.0015, "step": 77460 }, { "epoch": 1.2676102429845373, "grad_norm": 0.11001945286989212, "learning_rate": 8.634041630314227e-06, "loss": 0.0014, "step": 77470 }, { "epoch": 1.2677738689356133, "grad_norm": 0.19500237703323364, "learning_rate": 8.63355122161552e-06, "loss": 0.0046, "step": 77480 }, { "epoch": 1.267937494886689, "grad_norm": 0.08682650327682495, "learning_rate": 8.633060738831797e-06, "loss": 0.0023, "step": 77490 }, { "epoch": 1.2681011208377648, "grad_norm": 0.08259439468383789, "learning_rate": 8.632570181973054e-06, "loss": 0.0032, "step": 77500 }, { "epoch": 1.2682647467888408, "grad_norm": 0.06826230138540268, "learning_rate": 8.632079551049296e-06, "loss": 0.002, "step": 77510 }, { "epoch": 1.2684283727399166, "grad_norm": 0.10085809975862503, "learning_rate": 8.631588846070522e-06, "loss": 0.0009, "step": 77520 }, { "epoch": 1.2685919986909924, "grad_norm": 0.056762419641017914, "learning_rate": 8.631098067046741e-06, "loss": 0.0051, "step": 77530 }, { "epoch": 1.2687556246420684, "grad_norm": 0.03377929702401161, "learning_rate": 8.630607213987959e-06, "loss": 0.0017, "step": 77540 }, { "epoch": 1.2689192505931441, "grad_norm": 0.0639740601181984, "learning_rate": 8.630116286904184e-06, "loss": 0.0017, "step": 77550 }, { "epoch": 1.26908287654422, "grad_norm": 0.10874205082654953, "learning_rate": 8.629625285805425e-06, "loss": 0.0029, "step": 77560 }, { "epoch": 1.2692465024952957, "grad_norm": 0.07492777705192566, "learning_rate": 8.629134210701693e-06, "loss": 0.0014, "step": 77570 }, { "epoch": 1.2694101284463715, "grad_norm": 0.031850673258304596, "learning_rate": 8.628643061602999e-06, "loss": 0.0018, "step": 77580 }, { "epoch": 1.2695737543974475, "grad_norm": 0.14244742691516876, "learning_rate": 8.628151838519362e-06, "loss": 0.0014, "step": 77590 }, { "epoch": 1.2697373803485232, "grad_norm": 0.03544258698821068, "learning_rate": 8.627660541460795e-06, "loss": 0.0021, "step": 77600 }, { "epoch": 1.269901006299599, "grad_norm": 0.05995278060436249, "learning_rate": 8.627169170437311e-06, "loss": 0.0012, "step": 77610 }, { "epoch": 1.270064632250675, "grad_norm": 0.02812039665877819, "learning_rate": 8.626677725458935e-06, "loss": 0.0007, "step": 77620 }, { "epoch": 1.2702282582017508, "grad_norm": 0.0707254707813263, "learning_rate": 8.626186206535685e-06, "loss": 0.0021, "step": 77630 }, { "epoch": 1.2703918841528266, "grad_norm": 0.029347717761993408, "learning_rate": 8.625694613677582e-06, "loss": 0.0009, "step": 77640 }, { "epoch": 1.2705555101039026, "grad_norm": 0.12346512079238892, "learning_rate": 8.62520294689465e-06, "loss": 0.0019, "step": 77650 }, { "epoch": 1.2707191360549783, "grad_norm": 0.13303695619106293, "learning_rate": 8.624711206196909e-06, "loss": 0.0017, "step": 77660 }, { "epoch": 1.270882762006054, "grad_norm": 0.14553186297416687, "learning_rate": 8.624219391594394e-06, "loss": 0.0015, "step": 77670 }, { "epoch": 1.27104638795713, "grad_norm": 0.0070737199857831, "learning_rate": 8.623727503097126e-06, "loss": 0.0011, "step": 77680 }, { "epoch": 1.2712100139082059, "grad_norm": 0.10497470200061798, "learning_rate": 8.623235540715135e-06, "loss": 0.0019, "step": 77690 }, { "epoch": 1.2713736398592816, "grad_norm": 0.08100970089435577, "learning_rate": 8.622743504458455e-06, "loss": 0.0027, "step": 77700 }, { "epoch": 1.2715372658103576, "grad_norm": 0.0205882228910923, "learning_rate": 8.622251394337115e-06, "loss": 0.0008, "step": 77710 }, { "epoch": 1.2717008917614334, "grad_norm": 0.0849112868309021, "learning_rate": 8.62175921036115e-06, "loss": 0.0021, "step": 77720 }, { "epoch": 1.2718645177125092, "grad_norm": 0.06618082523345947, "learning_rate": 8.621266952540593e-06, "loss": 0.0019, "step": 77730 }, { "epoch": 1.272028143663585, "grad_norm": 0.04318990930914879, "learning_rate": 8.620774620885484e-06, "loss": 0.0015, "step": 77740 }, { "epoch": 1.272191769614661, "grad_norm": 0.059106484055519104, "learning_rate": 8.620282215405859e-06, "loss": 0.0016, "step": 77750 }, { "epoch": 1.2723553955657367, "grad_norm": 0.08182672411203384, "learning_rate": 8.619789736111762e-06, "loss": 0.0035, "step": 77760 }, { "epoch": 1.2725190215168125, "grad_norm": 0.019738340750336647, "learning_rate": 8.619297183013226e-06, "loss": 0.0012, "step": 77770 }, { "epoch": 1.2726826474678883, "grad_norm": 0.11256015300750732, "learning_rate": 8.618804556120302e-06, "loss": 0.0013, "step": 77780 }, { "epoch": 1.2728462734189643, "grad_norm": 0.13359226286411285, "learning_rate": 8.61831185544303e-06, "loss": 0.0013, "step": 77790 }, { "epoch": 1.27300989937004, "grad_norm": 0.1275704801082611, "learning_rate": 8.617819080991455e-06, "loss": 0.0019, "step": 77800 }, { "epoch": 1.2731735253211158, "grad_norm": 0.03292832896113396, "learning_rate": 8.617326232775627e-06, "loss": 0.0024, "step": 77810 }, { "epoch": 1.2733371512721918, "grad_norm": 0.08841615915298462, "learning_rate": 8.616833310805593e-06, "loss": 0.001, "step": 77820 }, { "epoch": 1.2735007772232676, "grad_norm": 0.011927049607038498, "learning_rate": 8.616340315091405e-06, "loss": 0.0011, "step": 77830 }, { "epoch": 1.2736644031743434, "grad_norm": 0.022918086498975754, "learning_rate": 8.615847245643112e-06, "loss": 0.0037, "step": 77840 }, { "epoch": 1.2738280291254194, "grad_norm": 0.04372696578502655, "learning_rate": 8.61535410247077e-06, "loss": 0.0012, "step": 77850 }, { "epoch": 1.2739916550764951, "grad_norm": 0.062226034700870514, "learning_rate": 8.614860885584432e-06, "loss": 0.0018, "step": 77860 }, { "epoch": 1.274155281027571, "grad_norm": 0.04147147014737129, "learning_rate": 8.614367594994157e-06, "loss": 0.0013, "step": 77870 }, { "epoch": 1.274318906978647, "grad_norm": 0.05865407735109329, "learning_rate": 8.613874230709997e-06, "loss": 0.0011, "step": 77880 }, { "epoch": 1.2744825329297227, "grad_norm": 0.09156455099582672, "learning_rate": 8.613380792742018e-06, "loss": 0.0025, "step": 77890 }, { "epoch": 1.2746461588807985, "grad_norm": 0.04061713069677353, "learning_rate": 8.612887281100277e-06, "loss": 0.0026, "step": 77900 }, { "epoch": 1.2748097848318745, "grad_norm": 0.12919408082962036, "learning_rate": 8.612393695794835e-06, "loss": 0.0014, "step": 77910 }, { "epoch": 1.2749734107829502, "grad_norm": 0.4597261846065521, "learning_rate": 8.61190003683576e-06, "loss": 0.001, "step": 77920 }, { "epoch": 1.275137036734026, "grad_norm": 0.039300624281167984, "learning_rate": 8.611406304233115e-06, "loss": 0.0012, "step": 77930 }, { "epoch": 1.2753006626851018, "grad_norm": 0.03198215365409851, "learning_rate": 8.610912497996965e-06, "loss": 0.0019, "step": 77940 }, { "epoch": 1.2754642886361778, "grad_norm": 0.11431149393320084, "learning_rate": 8.610418618137381e-06, "loss": 0.0021, "step": 77950 }, { "epoch": 1.2756279145872536, "grad_norm": 0.0689242035150528, "learning_rate": 8.609924664664432e-06, "loss": 0.0018, "step": 77960 }, { "epoch": 1.2757915405383293, "grad_norm": 0.12135417014360428, "learning_rate": 8.60943063758819e-06, "loss": 0.0022, "step": 77970 }, { "epoch": 1.275955166489405, "grad_norm": 0.045299407094717026, "learning_rate": 8.608936536918727e-06, "loss": 0.001, "step": 77980 }, { "epoch": 1.276118792440481, "grad_norm": 0.091968834400177, "learning_rate": 8.608442362666115e-06, "loss": 0.0025, "step": 77990 }, { "epoch": 1.2762824183915569, "grad_norm": 0.02820540964603424, "learning_rate": 8.607948114840435e-06, "loss": 0.0013, "step": 78000 }, { "epoch": 1.2764460443426326, "grad_norm": 0.019742924720048904, "learning_rate": 8.607453793451759e-06, "loss": 0.0014, "step": 78010 }, { "epoch": 1.2766096702937086, "grad_norm": 0.05424179881811142, "learning_rate": 8.606959398510169e-06, "loss": 0.0044, "step": 78020 }, { "epoch": 1.2767732962447844, "grad_norm": 0.06445953994989395, "learning_rate": 8.606464930025745e-06, "loss": 0.0015, "step": 78030 }, { "epoch": 1.2769369221958602, "grad_norm": 0.058756764978170395, "learning_rate": 8.60597038800857e-06, "loss": 0.0018, "step": 78040 }, { "epoch": 1.2771005481469362, "grad_norm": 0.0370054692029953, "learning_rate": 8.605475772468722e-06, "loss": 0.0019, "step": 78050 }, { "epoch": 1.277264174098012, "grad_norm": 0.04393920302391052, "learning_rate": 8.60498108341629e-06, "loss": 0.0012, "step": 78060 }, { "epoch": 1.2774278000490877, "grad_norm": 0.1134905144572258, "learning_rate": 8.604486320861361e-06, "loss": 0.0018, "step": 78070 }, { "epoch": 1.2775914260001637, "grad_norm": 0.09278996288776398, "learning_rate": 8.603991484814022e-06, "loss": 0.0013, "step": 78080 }, { "epoch": 1.2777550519512395, "grad_norm": 0.014358323998749256, "learning_rate": 8.60349657528436e-06, "loss": 0.0014, "step": 78090 }, { "epoch": 1.2779186779023153, "grad_norm": 0.0705438107252121, "learning_rate": 8.603001592282469e-06, "loss": 0.0017, "step": 78100 }, { "epoch": 1.2780823038533913, "grad_norm": 0.06569821387529373, "learning_rate": 8.602506535818439e-06, "loss": 0.0023, "step": 78110 }, { "epoch": 1.278245929804467, "grad_norm": 0.14566873013973236, "learning_rate": 8.602011405902364e-06, "loss": 0.0032, "step": 78120 }, { "epoch": 1.2784095557555428, "grad_norm": 0.04423107951879501, "learning_rate": 8.60151620254434e-06, "loss": 0.0019, "step": 78130 }, { "epoch": 1.2785731817066186, "grad_norm": 0.016807693988084793, "learning_rate": 8.601020925754464e-06, "loss": 0.0015, "step": 78140 }, { "epoch": 1.2787368076576944, "grad_norm": 0.06356599181890488, "learning_rate": 8.600525575542835e-06, "loss": 0.0016, "step": 78150 }, { "epoch": 1.2789004336087704, "grad_norm": 0.026511330157518387, "learning_rate": 8.60003015191955e-06, "loss": 0.002, "step": 78160 }, { "epoch": 1.2790640595598461, "grad_norm": 0.09689044952392578, "learning_rate": 8.599534654894712e-06, "loss": 0.0022, "step": 78170 }, { "epoch": 1.279227685510922, "grad_norm": 0.08507683128118515, "learning_rate": 8.599039084478425e-06, "loss": 0.001, "step": 78180 }, { "epoch": 1.279391311461998, "grad_norm": 0.07269373536109924, "learning_rate": 8.598543440680792e-06, "loss": 0.0014, "step": 78190 }, { "epoch": 1.2795549374130737, "grad_norm": 0.046533212065696716, "learning_rate": 8.598047723511916e-06, "loss": 0.001, "step": 78200 }, { "epoch": 1.2797185633641495, "grad_norm": 0.003925804514437914, "learning_rate": 8.597551932981909e-06, "loss": 0.001, "step": 78210 }, { "epoch": 1.2798821893152255, "grad_norm": 0.03758307173848152, "learning_rate": 8.597056069100877e-06, "loss": 0.0014, "step": 78220 }, { "epoch": 1.2800458152663012, "grad_norm": 0.15886592864990234, "learning_rate": 8.596560131878933e-06, "loss": 0.0021, "step": 78230 }, { "epoch": 1.280209441217377, "grad_norm": 0.026232974603772163, "learning_rate": 8.596064121326184e-06, "loss": 0.0029, "step": 78240 }, { "epoch": 1.280373067168453, "grad_norm": 0.07521108537912369, "learning_rate": 8.595568037452748e-06, "loss": 0.0014, "step": 78250 }, { "epoch": 1.2805366931195288, "grad_norm": 0.2656894624233246, "learning_rate": 8.595071880268735e-06, "loss": 0.0008, "step": 78260 }, { "epoch": 1.2807003190706046, "grad_norm": 0.19931304454803467, "learning_rate": 8.594575649784265e-06, "loss": 0.0017, "step": 78270 }, { "epoch": 1.2808639450216805, "grad_norm": 0.013412420637905598, "learning_rate": 8.594079346009456e-06, "loss": 0.0021, "step": 78280 }, { "epoch": 1.2810275709727563, "grad_norm": 0.09668263792991638, "learning_rate": 8.593582968954425e-06, "loss": 0.0031, "step": 78290 }, { "epoch": 1.281191196923832, "grad_norm": 0.04942458122968674, "learning_rate": 8.593086518629292e-06, "loss": 0.002, "step": 78300 }, { "epoch": 1.281354822874908, "grad_norm": 0.06829536706209183, "learning_rate": 8.592589995044182e-06, "loss": 0.0073, "step": 78310 }, { "epoch": 1.2815184488259839, "grad_norm": 0.03119976446032524, "learning_rate": 8.592093398209217e-06, "loss": 0.0019, "step": 78320 }, { "epoch": 1.2816820747770596, "grad_norm": 0.09957346320152283, "learning_rate": 8.591596728134524e-06, "loss": 0.0031, "step": 78330 }, { "epoch": 1.2818457007281354, "grad_norm": 0.051781851798295975, "learning_rate": 8.591099984830228e-06, "loss": 0.0021, "step": 78340 }, { "epoch": 1.2820093266792112, "grad_norm": 0.11936734616756439, "learning_rate": 8.590603168306457e-06, "loss": 0.002, "step": 78350 }, { "epoch": 1.2821729526302872, "grad_norm": 0.0545586496591568, "learning_rate": 8.59010627857334e-06, "loss": 0.0012, "step": 78360 }, { "epoch": 1.282336578581363, "grad_norm": 0.022370686754584312, "learning_rate": 8.58960931564101e-06, "loss": 0.0022, "step": 78370 }, { "epoch": 1.2825002045324387, "grad_norm": 0.004112221300601959, "learning_rate": 8.589112279519599e-06, "loss": 0.001, "step": 78380 }, { "epoch": 1.2826638304835147, "grad_norm": 0.031956471502780914, "learning_rate": 8.588615170219242e-06, "loss": 0.0014, "step": 78390 }, { "epoch": 1.2828274564345905, "grad_norm": 0.03925715386867523, "learning_rate": 8.588117987750072e-06, "loss": 0.0013, "step": 78400 }, { "epoch": 1.2829910823856663, "grad_norm": 0.03032500483095646, "learning_rate": 8.587620732122229e-06, "loss": 0.002, "step": 78410 }, { "epoch": 1.2831547083367423, "grad_norm": 0.028033535927534103, "learning_rate": 8.58712340334585e-06, "loss": 0.0013, "step": 78420 }, { "epoch": 1.283318334287818, "grad_norm": 0.03857472166419029, "learning_rate": 8.586626001431074e-06, "loss": 0.0012, "step": 78430 }, { "epoch": 1.2834819602388938, "grad_norm": 0.026612717658281326, "learning_rate": 8.586128526388046e-06, "loss": 0.0019, "step": 78440 }, { "epoch": 1.2836455861899698, "grad_norm": 0.06133998557925224, "learning_rate": 8.585630978226906e-06, "loss": 0.0023, "step": 78450 }, { "epoch": 1.2838092121410456, "grad_norm": 0.03583287447690964, "learning_rate": 8.585133356957802e-06, "loss": 0.0017, "step": 78460 }, { "epoch": 1.2839728380921214, "grad_norm": 0.07463372498750687, "learning_rate": 8.584635662590877e-06, "loss": 0.0032, "step": 78470 }, { "epoch": 1.2841364640431974, "grad_norm": 0.1611146330833435, "learning_rate": 8.584137895136278e-06, "loss": 0.0021, "step": 78480 }, { "epoch": 1.2843000899942731, "grad_norm": 0.07293737679719925, "learning_rate": 8.583640054604157e-06, "loss": 0.0015, "step": 78490 }, { "epoch": 1.284463715945349, "grad_norm": 0.07848577946424484, "learning_rate": 8.583142141004662e-06, "loss": 0.0021, "step": 78500 }, { "epoch": 1.2846273418964247, "grad_norm": 0.03565020114183426, "learning_rate": 8.582644154347948e-06, "loss": 0.0015, "step": 78510 }, { "epoch": 1.2847909678475007, "grad_norm": 0.030274150893092155, "learning_rate": 8.582146094644164e-06, "loss": 0.0031, "step": 78520 }, { "epoch": 1.2849545937985765, "grad_norm": 0.036408666521310806, "learning_rate": 8.581647961903469e-06, "loss": 0.0013, "step": 78530 }, { "epoch": 1.2851182197496522, "grad_norm": 0.08651316165924072, "learning_rate": 8.581149756136018e-06, "loss": 0.0026, "step": 78540 }, { "epoch": 1.285281845700728, "grad_norm": 0.07452166825532913, "learning_rate": 8.580651477351968e-06, "loss": 0.0013, "step": 78550 }, { "epoch": 1.285445471651804, "grad_norm": 0.046697258949279785, "learning_rate": 8.580153125561482e-06, "loss": 0.0014, "step": 78560 }, { "epoch": 1.2856090976028798, "grad_norm": 0.03326058015227318, "learning_rate": 8.579654700774717e-06, "loss": 0.0011, "step": 78570 }, { "epoch": 1.2857727235539556, "grad_norm": 0.07874537259340286, "learning_rate": 8.579156203001838e-06, "loss": 0.0017, "step": 78580 }, { "epoch": 1.2859363495050316, "grad_norm": 0.21733780205249786, "learning_rate": 8.578657632253007e-06, "loss": 0.0014, "step": 78590 }, { "epoch": 1.2860999754561073, "grad_norm": 0.24600517749786377, "learning_rate": 8.578158988538392e-06, "loss": 0.0018, "step": 78600 }, { "epoch": 1.286263601407183, "grad_norm": 0.04040563479065895, "learning_rate": 8.577660271868157e-06, "loss": 0.0019, "step": 78610 }, { "epoch": 1.286427227358259, "grad_norm": 0.12938830256462097, "learning_rate": 8.577161482252472e-06, "loss": 0.0026, "step": 78620 }, { "epoch": 1.2865908533093349, "grad_norm": 0.034256696701049805, "learning_rate": 8.576662619701506e-06, "loss": 0.0018, "step": 78630 }, { "epoch": 1.2867544792604106, "grad_norm": 0.039291076362133026, "learning_rate": 8.576163684225432e-06, "loss": 0.0024, "step": 78640 }, { "epoch": 1.2869181052114866, "grad_norm": 0.009385433048009872, "learning_rate": 8.57566467583442e-06, "loss": 0.0015, "step": 78650 }, { "epoch": 1.2870817311625624, "grad_norm": 0.09118425846099854, "learning_rate": 8.57516559453865e-06, "loss": 0.0024, "step": 78660 }, { "epoch": 1.2872453571136382, "grad_norm": 0.05232430249452591, "learning_rate": 8.574666440348292e-06, "loss": 0.0021, "step": 78670 }, { "epoch": 1.2874089830647142, "grad_norm": 0.07227593660354614, "learning_rate": 8.574167213273524e-06, "loss": 0.0034, "step": 78680 }, { "epoch": 1.28757260901579, "grad_norm": 0.05640187859535217, "learning_rate": 8.573667913324528e-06, "loss": 0.0026, "step": 78690 }, { "epoch": 1.2877362349668657, "grad_norm": 0.06281477957963943, "learning_rate": 8.573168540511482e-06, "loss": 0.0022, "step": 78700 }, { "epoch": 1.2878998609179415, "grad_norm": 0.02121046371757984, "learning_rate": 8.572669094844568e-06, "loss": 0.0025, "step": 78710 }, { "epoch": 1.2880634868690175, "grad_norm": 0.10742220282554626, "learning_rate": 8.572169576333971e-06, "loss": 0.0034, "step": 78720 }, { "epoch": 1.2882271128200933, "grad_norm": 0.10517984628677368, "learning_rate": 8.571669984989872e-06, "loss": 0.004, "step": 78730 }, { "epoch": 1.288390738771169, "grad_norm": 0.0833430364727974, "learning_rate": 8.571170320822461e-06, "loss": 0.002, "step": 78740 }, { "epoch": 1.2885543647222448, "grad_norm": 0.03191520273685455, "learning_rate": 8.570670583841925e-06, "loss": 0.0013, "step": 78750 }, { "epoch": 1.2887179906733208, "grad_norm": 0.03244727477431297, "learning_rate": 8.570170774058451e-06, "loss": 0.0021, "step": 78760 }, { "epoch": 1.2888816166243966, "grad_norm": 0.05712274834513664, "learning_rate": 8.569670891482234e-06, "loss": 0.0013, "step": 78770 }, { "epoch": 1.2890452425754724, "grad_norm": 0.14652210474014282, "learning_rate": 8.569170936123461e-06, "loss": 0.0018, "step": 78780 }, { "epoch": 1.2892088685265484, "grad_norm": 0.0016483762301504612, "learning_rate": 8.568670907992327e-06, "loss": 0.0021, "step": 78790 }, { "epoch": 1.2893724944776241, "grad_norm": 0.051516637206077576, "learning_rate": 8.56817080709903e-06, "loss": 0.0026, "step": 78800 }, { "epoch": 1.2895361204287, "grad_norm": 0.027263028547167778, "learning_rate": 8.567670633453765e-06, "loss": 0.001, "step": 78810 }, { "epoch": 1.289699746379776, "grad_norm": 0.02165922522544861, "learning_rate": 8.56717038706673e-06, "loss": 0.0019, "step": 78820 }, { "epoch": 1.2898633723308517, "grad_norm": 0.014103812165558338, "learning_rate": 8.566670067948124e-06, "loss": 0.0021, "step": 78830 }, { "epoch": 1.2900269982819275, "grad_norm": 0.016712302342057228, "learning_rate": 8.566169676108149e-06, "loss": 0.0074, "step": 78840 }, { "epoch": 1.2901906242330035, "grad_norm": 0.3699946999549866, "learning_rate": 8.565669211557009e-06, "loss": 0.0015, "step": 78850 }, { "epoch": 1.2903542501840792, "grad_norm": 0.06845702230930328, "learning_rate": 8.565168674304902e-06, "loss": 0.002, "step": 78860 }, { "epoch": 1.290517876135155, "grad_norm": 0.06485264748334885, "learning_rate": 8.564668064362041e-06, "loss": 0.0023, "step": 78870 }, { "epoch": 1.290681502086231, "grad_norm": 0.026777582243084908, "learning_rate": 8.564167381738628e-06, "loss": 0.004, "step": 78880 }, { "epoch": 1.2908451280373068, "grad_norm": 0.20939204096794128, "learning_rate": 8.563666626444876e-06, "loss": 0.0019, "step": 78890 }, { "epoch": 1.2910087539883826, "grad_norm": 0.11581704020500183, "learning_rate": 8.56316579849099e-06, "loss": 0.0022, "step": 78900 }, { "epoch": 1.2911723799394583, "grad_norm": 0.008656048215925694, "learning_rate": 8.562664897887184e-06, "loss": 0.0017, "step": 78910 }, { "epoch": 1.2913360058905343, "grad_norm": 0.032674215734004974, "learning_rate": 8.562163924643672e-06, "loss": 0.0017, "step": 78920 }, { "epoch": 1.29149963184161, "grad_norm": 0.1366446167230606, "learning_rate": 8.561662878770664e-06, "loss": 0.0022, "step": 78930 }, { "epoch": 1.2916632577926859, "grad_norm": 0.03468158841133118, "learning_rate": 8.561161760278383e-06, "loss": 0.0022, "step": 78940 }, { "epoch": 1.2918268837437616, "grad_norm": 0.11580320447683334, "learning_rate": 8.56066056917704e-06, "loss": 0.0023, "step": 78950 }, { "epoch": 1.2919905096948376, "grad_norm": 0.04989323019981384, "learning_rate": 8.560159305476857e-06, "loss": 0.0014, "step": 78960 }, { "epoch": 1.2921541356459134, "grad_norm": 0.0672837644815445, "learning_rate": 8.559657969188054e-06, "loss": 0.0012, "step": 78970 }, { "epoch": 1.2923177615969892, "grad_norm": 0.2108697146177292, "learning_rate": 8.55915656032085e-06, "loss": 0.0028, "step": 78980 }, { "epoch": 1.2924813875480652, "grad_norm": 0.06120941787958145, "learning_rate": 8.558655078885471e-06, "loss": 0.0016, "step": 78990 }, { "epoch": 1.292645013499141, "grad_norm": 0.08390291035175323, "learning_rate": 8.558153524892145e-06, "loss": 0.0013, "step": 79000 }, { "epoch": 1.2928086394502167, "grad_norm": 0.10053674876689911, "learning_rate": 8.557651898351091e-06, "loss": 0.0016, "step": 79010 }, { "epoch": 1.2929722654012927, "grad_norm": 0.06556560844182968, "learning_rate": 8.557150199272542e-06, "loss": 0.0015, "step": 79020 }, { "epoch": 1.2931358913523685, "grad_norm": 0.1244731917977333, "learning_rate": 8.556648427666727e-06, "loss": 0.0027, "step": 79030 }, { "epoch": 1.2932995173034443, "grad_norm": 0.06958924233913422, "learning_rate": 8.556146583543874e-06, "loss": 0.0021, "step": 79040 }, { "epoch": 1.2934631432545203, "grad_norm": 0.004718066193163395, "learning_rate": 8.555644666914215e-06, "loss": 0.0022, "step": 79050 }, { "epoch": 1.293626769205596, "grad_norm": 0.044462669640779495, "learning_rate": 8.555142677787987e-06, "loss": 0.0014, "step": 79060 }, { "epoch": 1.2937903951566718, "grad_norm": 0.08993500471115112, "learning_rate": 8.554640616175424e-06, "loss": 0.0019, "step": 79070 }, { "epoch": 1.2939540211077478, "grad_norm": 0.15881212055683136, "learning_rate": 8.55413848208676e-06, "loss": 0.0023, "step": 79080 }, { "epoch": 1.2941176470588236, "grad_norm": 0.026637891307473183, "learning_rate": 8.553636275532236e-06, "loss": 0.0011, "step": 79090 }, { "epoch": 1.2942812730098994, "grad_norm": 0.18290722370147705, "learning_rate": 8.553133996522092e-06, "loss": 0.0024, "step": 79100 }, { "epoch": 1.2944448989609751, "grad_norm": 0.046773333102464676, "learning_rate": 8.552631645066565e-06, "loss": 0.0017, "step": 79110 }, { "epoch": 1.294608524912051, "grad_norm": 0.011792654171586037, "learning_rate": 8.552129221175901e-06, "loss": 0.0018, "step": 79120 }, { "epoch": 1.294772150863127, "grad_norm": 0.017240850254893303, "learning_rate": 8.551626724860344e-06, "loss": 0.0015, "step": 79130 }, { "epoch": 1.2949357768142027, "grad_norm": 0.08368983864784241, "learning_rate": 8.55112415613014e-06, "loss": 0.0025, "step": 79140 }, { "epoch": 1.2950994027652785, "grad_norm": 0.10434108972549438, "learning_rate": 8.550621514995532e-06, "loss": 0.0031, "step": 79150 }, { "epoch": 1.2952630287163545, "grad_norm": 0.1629168540239334, "learning_rate": 8.550118801466773e-06, "loss": 0.0017, "step": 79160 }, { "epoch": 1.2954266546674302, "grad_norm": 0.01387413777410984, "learning_rate": 8.549616015554109e-06, "loss": 0.0013, "step": 79170 }, { "epoch": 1.295590280618506, "grad_norm": 0.0426555834710598, "learning_rate": 8.549113157267794e-06, "loss": 0.0032, "step": 79180 }, { "epoch": 1.295753906569582, "grad_norm": 0.04751654714345932, "learning_rate": 8.54861022661808e-06, "loss": 0.0021, "step": 79190 }, { "epoch": 1.2959175325206578, "grad_norm": 0.04086978733539581, "learning_rate": 8.548107223615224e-06, "loss": 0.0023, "step": 79200 }, { "epoch": 1.2960811584717336, "grad_norm": 0.04342479631304741, "learning_rate": 8.547604148269475e-06, "loss": 0.0023, "step": 79210 }, { "epoch": 1.2962447844228095, "grad_norm": 0.062000054866075516, "learning_rate": 8.547101000591096e-06, "loss": 0.0013, "step": 79220 }, { "epoch": 1.2964084103738853, "grad_norm": 0.03156150504946709, "learning_rate": 8.546597780590345e-06, "loss": 0.003, "step": 79230 }, { "epoch": 1.296572036324961, "grad_norm": 0.06596570461988449, "learning_rate": 8.546094488277482e-06, "loss": 0.0019, "step": 79240 }, { "epoch": 1.296735662276037, "grad_norm": 0.06716080009937286, "learning_rate": 8.545591123662767e-06, "loss": 0.001, "step": 79250 }, { "epoch": 1.2968992882271129, "grad_norm": 0.10306563973426819, "learning_rate": 8.545087686756467e-06, "loss": 0.0036, "step": 79260 }, { "epoch": 1.2970629141781886, "grad_norm": 0.10073012858629227, "learning_rate": 8.54458417756884e-06, "loss": 0.0012, "step": 79270 }, { "epoch": 1.2972265401292646, "grad_norm": 0.06875721365213394, "learning_rate": 8.544080596110159e-06, "loss": 0.0023, "step": 79280 }, { "epoch": 1.2973901660803404, "grad_norm": 0.08468332141637802, "learning_rate": 8.543576942390688e-06, "loss": 0.001, "step": 79290 }, { "epoch": 1.2975537920314162, "grad_norm": 0.04941927269101143, "learning_rate": 8.543073216420697e-06, "loss": 0.0018, "step": 79300 }, { "epoch": 1.297717417982492, "grad_norm": 0.15632380545139313, "learning_rate": 8.542569418210456e-06, "loss": 0.002, "step": 79310 }, { "epoch": 1.2978810439335677, "grad_norm": 0.02616730146110058, "learning_rate": 8.542065547770237e-06, "loss": 0.0022, "step": 79320 }, { "epoch": 1.2980446698846437, "grad_norm": 0.13326455652713776, "learning_rate": 8.541561605110315e-06, "loss": 0.0016, "step": 79330 }, { "epoch": 1.2982082958357195, "grad_norm": 0.06769983470439911, "learning_rate": 8.541057590240963e-06, "loss": 0.0019, "step": 79340 }, { "epoch": 1.2983719217867953, "grad_norm": 0.05765802413225174, "learning_rate": 8.540553503172459e-06, "loss": 0.0032, "step": 79350 }, { "epoch": 1.2985355477378713, "grad_norm": 0.026005392894148827, "learning_rate": 8.54004934391508e-06, "loss": 0.0016, "step": 79360 }, { "epoch": 1.298699173688947, "grad_norm": 0.10359671711921692, "learning_rate": 8.539545112479106e-06, "loss": 0.0018, "step": 79370 }, { "epoch": 1.2988627996400228, "grad_norm": 0.19111156463623047, "learning_rate": 8.539040808874816e-06, "loss": 0.0022, "step": 79380 }, { "epoch": 1.2990264255910988, "grad_norm": 0.10560057312250137, "learning_rate": 8.538536433112494e-06, "loss": 0.0016, "step": 79390 }, { "epoch": 1.2991900515421746, "grad_norm": 0.036339279264211655, "learning_rate": 8.538031985202424e-06, "loss": 0.0015, "step": 79400 }, { "epoch": 1.2993536774932504, "grad_norm": 0.23530925810337067, "learning_rate": 8.537527465154892e-06, "loss": 0.0029, "step": 79410 }, { "epoch": 1.2995173034443264, "grad_norm": 0.05009636655449867, "learning_rate": 8.537022872980184e-06, "loss": 0.0024, "step": 79420 }, { "epoch": 1.2996809293954021, "grad_norm": 0.05614149197936058, "learning_rate": 8.536518208688584e-06, "loss": 0.0019, "step": 79430 }, { "epoch": 1.299844555346478, "grad_norm": 0.05604657530784607, "learning_rate": 8.536013472290387e-06, "loss": 0.0016, "step": 79440 }, { "epoch": 1.300008181297554, "grad_norm": 0.02816956117749214, "learning_rate": 8.535508663795884e-06, "loss": 0.0009, "step": 79450 }, { "epoch": 1.3001718072486297, "grad_norm": 0.049245838075876236, "learning_rate": 8.535003783215366e-06, "loss": 0.0013, "step": 79460 }, { "epoch": 1.3003354331997055, "grad_norm": 0.32088714838027954, "learning_rate": 8.534498830559128e-06, "loss": 0.0021, "step": 79470 }, { "epoch": 1.3004990591507812, "grad_norm": 0.05760052055120468, "learning_rate": 8.533993805837463e-06, "loss": 0.0011, "step": 79480 }, { "epoch": 1.3006626851018572, "grad_norm": 0.0724153071641922, "learning_rate": 8.53348870906067e-06, "loss": 0.0038, "step": 79490 }, { "epoch": 1.300826311052933, "grad_norm": 0.025424841791391373, "learning_rate": 8.532983540239048e-06, "loss": 0.0028, "step": 79500 }, { "epoch": 1.3009899370040088, "grad_norm": 0.09447524696588516, "learning_rate": 8.532478299382896e-06, "loss": 0.0015, "step": 79510 }, { "epoch": 1.3011535629550846, "grad_norm": 0.10019547492265701, "learning_rate": 8.531972986502517e-06, "loss": 0.0021, "step": 79520 }, { "epoch": 1.3013171889061605, "grad_norm": 0.10668069124221802, "learning_rate": 8.53146760160821e-06, "loss": 0.0022, "step": 79530 }, { "epoch": 1.3014808148572363, "grad_norm": 0.02215401455760002, "learning_rate": 8.530962144710285e-06, "loss": 0.002, "step": 79540 }, { "epoch": 1.301644440808312, "grad_norm": 0.013551775366067886, "learning_rate": 8.530456615819043e-06, "loss": 0.0019, "step": 79550 }, { "epoch": 1.301808066759388, "grad_norm": 0.03742603957653046, "learning_rate": 8.529951014944792e-06, "loss": 0.0007, "step": 79560 }, { "epoch": 1.3019716927104639, "grad_norm": 0.03123163804411888, "learning_rate": 8.529445342097844e-06, "loss": 0.009, "step": 79570 }, { "epoch": 1.3021353186615396, "grad_norm": 0.07319080084562302, "learning_rate": 8.528939597288506e-06, "loss": 0.0012, "step": 79580 }, { "epoch": 1.3022989446126156, "grad_norm": 0.008544465526938438, "learning_rate": 8.528433780527093e-06, "loss": 0.0024, "step": 79590 }, { "epoch": 1.3024625705636914, "grad_norm": 0.1424582600593567, "learning_rate": 8.527927891823913e-06, "loss": 0.0034, "step": 79600 }, { "epoch": 1.3026261965147672, "grad_norm": 0.03400289639830589, "learning_rate": 8.527421931189286e-06, "loss": 0.0017, "step": 79610 }, { "epoch": 1.3027898224658432, "grad_norm": 0.06682585179805756, "learning_rate": 8.526915898633524e-06, "loss": 0.0023, "step": 79620 }, { "epoch": 1.302953448416919, "grad_norm": 0.11156055331230164, "learning_rate": 8.526409794166948e-06, "loss": 0.0014, "step": 79630 }, { "epoch": 1.3031170743679947, "grad_norm": 0.0625109076499939, "learning_rate": 8.525903617799874e-06, "loss": 0.0017, "step": 79640 }, { "epoch": 1.3032807003190707, "grad_norm": 0.04894212633371353, "learning_rate": 8.525397369542624e-06, "loss": 0.0011, "step": 79650 }, { "epoch": 1.3034443262701465, "grad_norm": 0.07780783623456955, "learning_rate": 8.524891049405522e-06, "loss": 0.0024, "step": 79660 }, { "epoch": 1.3036079522212223, "grad_norm": 0.2913656234741211, "learning_rate": 8.524384657398888e-06, "loss": 0.0032, "step": 79670 }, { "epoch": 1.303771578172298, "grad_norm": 0.008364393375813961, "learning_rate": 8.523878193533047e-06, "loss": 0.0018, "step": 79680 }, { "epoch": 1.303935204123374, "grad_norm": 0.05456334352493286, "learning_rate": 8.523371657818327e-06, "loss": 0.0011, "step": 79690 }, { "epoch": 1.3040988300744498, "grad_norm": 0.0809636190533638, "learning_rate": 8.522865050265056e-06, "loss": 0.004, "step": 79700 }, { "epoch": 1.3042624560255256, "grad_norm": 0.05754462257027626, "learning_rate": 8.522358370883564e-06, "loss": 0.0022, "step": 79710 }, { "epoch": 1.3044260819766014, "grad_norm": 0.01774998940527439, "learning_rate": 8.521851619684178e-06, "loss": 0.001, "step": 79720 }, { "epoch": 1.3045897079276774, "grad_norm": 0.08332259207963943, "learning_rate": 8.521344796677236e-06, "loss": 0.0019, "step": 79730 }, { "epoch": 1.3047533338787531, "grad_norm": 0.03572879731655121, "learning_rate": 8.520837901873065e-06, "loss": 0.0016, "step": 79740 }, { "epoch": 1.304916959829829, "grad_norm": 0.09340070933103561, "learning_rate": 8.520330935282006e-06, "loss": 0.0018, "step": 79750 }, { "epoch": 1.305080585780905, "grad_norm": 0.052956778556108475, "learning_rate": 8.519823896914391e-06, "loss": 0.002, "step": 79760 }, { "epoch": 1.3052442117319807, "grad_norm": 0.06022334471344948, "learning_rate": 8.519316786780562e-06, "loss": 0.0022, "step": 79770 }, { "epoch": 1.3054078376830565, "grad_norm": 0.06673215329647064, "learning_rate": 8.518809604890856e-06, "loss": 0.0025, "step": 79780 }, { "epoch": 1.3055714636341325, "grad_norm": 0.1496211290359497, "learning_rate": 8.518302351255613e-06, "loss": 0.0022, "step": 79790 }, { "epoch": 1.3057350895852082, "grad_norm": 0.009155592881143093, "learning_rate": 8.51779502588518e-06, "loss": 0.002, "step": 79800 }, { "epoch": 1.305898715536284, "grad_norm": 0.10461550951004028, "learning_rate": 8.517287628789899e-06, "loss": 0.0016, "step": 79810 }, { "epoch": 1.30606234148736, "grad_norm": 0.09453170746564865, "learning_rate": 8.516780159980112e-06, "loss": 0.0019, "step": 79820 }, { "epoch": 1.3062259674384358, "grad_norm": 0.04802963510155678, "learning_rate": 8.516272619466169e-06, "loss": 0.001, "step": 79830 }, { "epoch": 1.3063895933895115, "grad_norm": 0.05569612979888916, "learning_rate": 8.515765007258418e-06, "loss": 0.0021, "step": 79840 }, { "epoch": 1.3065532193405875, "grad_norm": 0.030103718861937523, "learning_rate": 8.51525732336721e-06, "loss": 0.0009, "step": 79850 }, { "epoch": 1.3067168452916633, "grad_norm": 0.12821531295776367, "learning_rate": 8.514749567802892e-06, "loss": 0.0012, "step": 79860 }, { "epoch": 1.306880471242739, "grad_norm": 0.023344093933701515, "learning_rate": 8.514241740575821e-06, "loss": 0.0031, "step": 79870 }, { "epoch": 1.3070440971938149, "grad_norm": 0.07735724747180939, "learning_rate": 8.51373384169635e-06, "loss": 0.0019, "step": 79880 }, { "epoch": 1.3072077231448906, "grad_norm": 0.05525326356291771, "learning_rate": 8.513225871174833e-06, "loss": 0.003, "step": 79890 }, { "epoch": 1.3073713490959666, "grad_norm": 0.024820871651172638, "learning_rate": 8.51271782902163e-06, "loss": 0.0021, "step": 79900 }, { "epoch": 1.3075349750470424, "grad_norm": 0.06921489536762238, "learning_rate": 8.512209715247097e-06, "loss": 0.0014, "step": 79910 }, { "epoch": 1.3076986009981182, "grad_norm": 0.05046744644641876, "learning_rate": 8.511701529861595e-06, "loss": 0.002, "step": 79920 }, { "epoch": 1.3078622269491942, "grad_norm": 0.0661848857998848, "learning_rate": 8.511193272875486e-06, "loss": 0.0015, "step": 79930 }, { "epoch": 1.30802585290027, "grad_norm": 0.029043471440672874, "learning_rate": 8.510684944299132e-06, "loss": 0.0017, "step": 79940 }, { "epoch": 1.3081894788513457, "grad_norm": 0.42711561918258667, "learning_rate": 8.510176544142898e-06, "loss": 0.0015, "step": 79950 }, { "epoch": 1.3083531048024217, "grad_norm": 0.06877776235342026, "learning_rate": 8.50966807241715e-06, "loss": 0.0014, "step": 79960 }, { "epoch": 1.3085167307534975, "grad_norm": 0.014344267547130585, "learning_rate": 8.509159529132254e-06, "loss": 0.0011, "step": 79970 }, { "epoch": 1.3086803567045733, "grad_norm": 0.03465251997113228, "learning_rate": 8.50865091429858e-06, "loss": 0.0029, "step": 79980 }, { "epoch": 1.3088439826556493, "grad_norm": 0.03998319432139397, "learning_rate": 8.508142227926497e-06, "loss": 0.0031, "step": 79990 }, { "epoch": 1.309007608606725, "grad_norm": 0.03161986172199249, "learning_rate": 8.507633470026377e-06, "loss": 0.0022, "step": 80000 }, { "epoch": 1.3091712345578008, "grad_norm": 0.07799658924341202, "learning_rate": 8.507124640608595e-06, "loss": 0.0044, "step": 80010 }, { "epoch": 1.3093348605088768, "grad_norm": 0.0552963949739933, "learning_rate": 8.506615739683524e-06, "loss": 0.002, "step": 80020 }, { "epoch": 1.3094984864599526, "grad_norm": 0.10290983319282532, "learning_rate": 8.50610676726154e-06, "loss": 0.002, "step": 80030 }, { "epoch": 1.3096621124110284, "grad_norm": 0.040889207273721695, "learning_rate": 8.505597723353022e-06, "loss": 0.0021, "step": 80040 }, { "epoch": 1.3098257383621044, "grad_norm": 0.028436550870537758, "learning_rate": 8.505088607968347e-06, "loss": 0.0019, "step": 80050 }, { "epoch": 1.3099893643131801, "grad_norm": 0.057926204055547714, "learning_rate": 8.504579421117896e-06, "loss": 0.0032, "step": 80060 }, { "epoch": 1.310152990264256, "grad_norm": 0.12233589589595795, "learning_rate": 8.504070162812052e-06, "loss": 0.0017, "step": 80070 }, { "epoch": 1.3103166162153317, "grad_norm": 0.12577562034130096, "learning_rate": 8.503560833061196e-06, "loss": 0.0021, "step": 80080 }, { "epoch": 1.3104802421664075, "grad_norm": 0.06403519958257675, "learning_rate": 8.503051431875716e-06, "loss": 0.0014, "step": 80090 }, { "epoch": 1.3106438681174835, "grad_norm": 0.13451944291591644, "learning_rate": 8.502541959265996e-06, "loss": 0.0017, "step": 80100 }, { "epoch": 1.3108074940685592, "grad_norm": 0.06745319068431854, "learning_rate": 8.502032415242424e-06, "loss": 0.0017, "step": 80110 }, { "epoch": 1.310971120019635, "grad_norm": 0.13153748214244843, "learning_rate": 8.501522799815389e-06, "loss": 0.0019, "step": 80120 }, { "epoch": 1.311134745970711, "grad_norm": 0.024858780205249786, "learning_rate": 8.501013112995282e-06, "loss": 0.0013, "step": 80130 }, { "epoch": 1.3112983719217868, "grad_norm": 0.0745086744427681, "learning_rate": 8.500503354792497e-06, "loss": 0.002, "step": 80140 }, { "epoch": 1.3114619978728626, "grad_norm": 0.015336193144321442, "learning_rate": 8.499993525217423e-06, "loss": 0.0012, "step": 80150 }, { "epoch": 1.3116256238239385, "grad_norm": 0.026163777336478233, "learning_rate": 8.49948362428046e-06, "loss": 0.0019, "step": 80160 }, { "epoch": 1.3117892497750143, "grad_norm": 0.12666241824626923, "learning_rate": 8.498973651992e-06, "loss": 0.0016, "step": 80170 }, { "epoch": 1.31195287572609, "grad_norm": 0.06520118564367294, "learning_rate": 8.498463608362445e-06, "loss": 0.0016, "step": 80180 }, { "epoch": 1.312116501677166, "grad_norm": 0.014170105569064617, "learning_rate": 8.497953493402191e-06, "loss": 0.0018, "step": 80190 }, { "epoch": 1.3122801276282419, "grad_norm": 0.02403724379837513, "learning_rate": 8.497443307121641e-06, "loss": 0.0022, "step": 80200 }, { "epoch": 1.3124437535793176, "grad_norm": 0.00898815505206585, "learning_rate": 8.496933049531195e-06, "loss": 0.002, "step": 80210 }, { "epoch": 1.3126073795303936, "grad_norm": 0.0833255797624588, "learning_rate": 8.49642272064126e-06, "loss": 0.002, "step": 80220 }, { "epoch": 1.3127710054814694, "grad_norm": 0.019270911812782288, "learning_rate": 8.495912320462238e-06, "loss": 0.0024, "step": 80230 }, { "epoch": 1.3129346314325452, "grad_norm": 0.060607388615608215, "learning_rate": 8.49540184900454e-06, "loss": 0.0017, "step": 80240 }, { "epoch": 1.313098257383621, "grad_norm": 0.06410450488328934, "learning_rate": 8.494891306278566e-06, "loss": 0.0022, "step": 80250 }, { "epoch": 1.313261883334697, "grad_norm": 0.056128572672605515, "learning_rate": 8.494380692294734e-06, "loss": 0.0016, "step": 80260 }, { "epoch": 1.3134255092857727, "grad_norm": 0.03954307734966278, "learning_rate": 8.49387000706345e-06, "loss": 0.0008, "step": 80270 }, { "epoch": 1.3135891352368485, "grad_norm": 0.06805521249771118, "learning_rate": 8.49335925059513e-06, "loss": 0.0014, "step": 80280 }, { "epoch": 1.3137527611879243, "grad_norm": 0.10724730789661407, "learning_rate": 8.492848422900183e-06, "loss": 0.0015, "step": 80290 }, { "epoch": 1.3139163871390003, "grad_norm": 0.06141386926174164, "learning_rate": 8.49233752398903e-06, "loss": 0.0016, "step": 80300 }, { "epoch": 1.314080013090076, "grad_norm": 0.059034813195466995, "learning_rate": 8.491826553872083e-06, "loss": 0.0015, "step": 80310 }, { "epoch": 1.3142436390411518, "grad_norm": 0.023469818755984306, "learning_rate": 8.491315512559764e-06, "loss": 0.0016, "step": 80320 }, { "epoch": 1.3144072649922278, "grad_norm": 0.04777003079652786, "learning_rate": 8.490804400062488e-06, "loss": 0.0018, "step": 80330 }, { "epoch": 1.3145708909433036, "grad_norm": 0.12226912379264832, "learning_rate": 8.490293216390682e-06, "loss": 0.0017, "step": 80340 }, { "epoch": 1.3147345168943794, "grad_norm": 0.10281695425510406, "learning_rate": 8.489781961554764e-06, "loss": 0.0026, "step": 80350 }, { "epoch": 1.3148981428454554, "grad_norm": 0.08342410624027252, "learning_rate": 8.489270635565161e-06, "loss": 0.0016, "step": 80360 }, { "epoch": 1.3150617687965311, "grad_norm": 0.304252028465271, "learning_rate": 8.488759238432297e-06, "loss": 0.0027, "step": 80370 }, { "epoch": 1.315225394747607, "grad_norm": 0.025333650410175323, "learning_rate": 8.4882477701666e-06, "loss": 0.0025, "step": 80380 }, { "epoch": 1.315389020698683, "grad_norm": 0.03447440266609192, "learning_rate": 8.487736230778496e-06, "loss": 0.0018, "step": 80390 }, { "epoch": 1.3155526466497587, "grad_norm": 0.028517862781882286, "learning_rate": 8.487224620278415e-06, "loss": 0.0021, "step": 80400 }, { "epoch": 1.3157162726008345, "grad_norm": 0.1126151978969574, "learning_rate": 8.486712938676793e-06, "loss": 0.0029, "step": 80410 }, { "epoch": 1.3158798985519105, "grad_norm": 0.0259290412068367, "learning_rate": 8.486201185984059e-06, "loss": 0.0026, "step": 80420 }, { "epoch": 1.3160435245029862, "grad_norm": 0.03787507861852646, "learning_rate": 8.485689362210647e-06, "loss": 0.0008, "step": 80430 }, { "epoch": 1.316207150454062, "grad_norm": 0.05979831889271736, "learning_rate": 8.485177467366993e-06, "loss": 0.0013, "step": 80440 }, { "epoch": 1.3163707764051378, "grad_norm": 0.0505683571100235, "learning_rate": 8.484665501463537e-06, "loss": 0.0017, "step": 80450 }, { "epoch": 1.3165344023562138, "grad_norm": 0.022979311645030975, "learning_rate": 8.484153464510712e-06, "loss": 0.0008, "step": 80460 }, { "epoch": 1.3166980283072895, "grad_norm": 0.012883502058684826, "learning_rate": 8.483641356518961e-06, "loss": 0.0019, "step": 80470 }, { "epoch": 1.3168616542583653, "grad_norm": 0.06471913307905197, "learning_rate": 8.483129177498727e-06, "loss": 0.0016, "step": 80480 }, { "epoch": 1.317025280209441, "grad_norm": 0.023513421416282654, "learning_rate": 8.48261692746045e-06, "loss": 0.0013, "step": 80490 }, { "epoch": 1.317188906160517, "grad_norm": 0.10976769775152206, "learning_rate": 8.482104606414576e-06, "loss": 0.0025, "step": 80500 }, { "epoch": 1.3173525321115929, "grad_norm": 0.15443411469459534, "learning_rate": 8.481592214371552e-06, "loss": 0.0026, "step": 80510 }, { "epoch": 1.3175161580626686, "grad_norm": 0.03232494741678238, "learning_rate": 8.481079751341823e-06, "loss": 0.0009, "step": 80520 }, { "epoch": 1.3176797840137446, "grad_norm": 0.06455060094594955, "learning_rate": 8.480567217335838e-06, "loss": 0.0016, "step": 80530 }, { "epoch": 1.3178434099648204, "grad_norm": 0.0682133287191391, "learning_rate": 8.480054612364048e-06, "loss": 0.0015, "step": 80540 }, { "epoch": 1.3180070359158962, "grad_norm": 0.0695497989654541, "learning_rate": 8.479541936436903e-06, "loss": 0.0018, "step": 80550 }, { "epoch": 1.3181706618669722, "grad_norm": 0.03488348796963692, "learning_rate": 8.479029189564858e-06, "loss": 0.003, "step": 80560 }, { "epoch": 1.318334287818048, "grad_norm": 0.07015686482191086, "learning_rate": 8.478516371758367e-06, "loss": 0.0014, "step": 80570 }, { "epoch": 1.3184979137691237, "grad_norm": 0.031699951738119125, "learning_rate": 8.478003483027886e-06, "loss": 0.0021, "step": 80580 }, { "epoch": 1.3186615397201997, "grad_norm": 0.14806270599365234, "learning_rate": 8.477490523383871e-06, "loss": 0.0024, "step": 80590 }, { "epoch": 1.3188251656712755, "grad_norm": 0.024625401943922043, "learning_rate": 8.476977492836782e-06, "loss": 0.0013, "step": 80600 }, { "epoch": 1.3189887916223513, "grad_norm": 0.033763740211725235, "learning_rate": 8.476464391397078e-06, "loss": 0.0007, "step": 80610 }, { "epoch": 1.3191524175734273, "grad_norm": 0.08697792887687683, "learning_rate": 8.475951219075222e-06, "loss": 0.002, "step": 80620 }, { "epoch": 1.319316043524503, "grad_norm": 0.06859993189573288, "learning_rate": 8.475437975881678e-06, "loss": 0.0011, "step": 80630 }, { "epoch": 1.3194796694755788, "grad_norm": 0.12321880459785461, "learning_rate": 8.47492466182691e-06, "loss": 0.0018, "step": 80640 }, { "epoch": 1.3196432954266546, "grad_norm": 0.06638049334287643, "learning_rate": 8.47441127692138e-06, "loss": 0.0013, "step": 80650 }, { "epoch": 1.3198069213777306, "grad_norm": 0.14198362827301025, "learning_rate": 8.473897821175563e-06, "loss": 0.0028, "step": 80660 }, { "epoch": 1.3199705473288064, "grad_norm": 0.05044763907790184, "learning_rate": 8.473384294599921e-06, "loss": 0.0026, "step": 80670 }, { "epoch": 1.3201341732798821, "grad_norm": 0.043167296797037125, "learning_rate": 8.47287069720493e-06, "loss": 0.0029, "step": 80680 }, { "epoch": 1.320297799230958, "grad_norm": 0.08402055501937866, "learning_rate": 8.472357029001056e-06, "loss": 0.0017, "step": 80690 }, { "epoch": 1.320461425182034, "grad_norm": 0.014949745498597622, "learning_rate": 8.471843289998777e-06, "loss": 0.0022, "step": 80700 }, { "epoch": 1.3206250511331097, "grad_norm": 0.053316131234169006, "learning_rate": 8.471329480208566e-06, "loss": 0.0015, "step": 80710 }, { "epoch": 1.3207886770841855, "grad_norm": 0.03154044225811958, "learning_rate": 8.470815599640898e-06, "loss": 0.0015, "step": 80720 }, { "epoch": 1.3209523030352615, "grad_norm": 0.008465672843158245, "learning_rate": 8.470301648306253e-06, "loss": 0.0026, "step": 80730 }, { "epoch": 1.3211159289863372, "grad_norm": 0.09463571012020111, "learning_rate": 8.46978762621511e-06, "loss": 0.0013, "step": 80740 }, { "epoch": 1.321279554937413, "grad_norm": 0.11021050065755844, "learning_rate": 8.469273533377944e-06, "loss": 0.003, "step": 80750 }, { "epoch": 1.321443180888489, "grad_norm": 0.1797020584344864, "learning_rate": 8.468759369805244e-06, "loss": 0.0037, "step": 80760 }, { "epoch": 1.3216068068395648, "grad_norm": 0.11677510291337967, "learning_rate": 8.468245135507492e-06, "loss": 0.0011, "step": 80770 }, { "epoch": 1.3217704327906405, "grad_norm": 0.15180885791778564, "learning_rate": 8.467730830495168e-06, "loss": 0.0013, "step": 80780 }, { "epoch": 1.3219340587417165, "grad_norm": 0.37013447284698486, "learning_rate": 8.467216454778764e-06, "loss": 0.0017, "step": 80790 }, { "epoch": 1.3220976846927923, "grad_norm": 0.06172894686460495, "learning_rate": 8.466702008368765e-06, "loss": 0.0015, "step": 80800 }, { "epoch": 1.322261310643868, "grad_norm": 0.06098209694027901, "learning_rate": 8.466187491275658e-06, "loss": 0.0021, "step": 80810 }, { "epoch": 1.322424936594944, "grad_norm": 0.03443707898259163, "learning_rate": 8.465672903509939e-06, "loss": 0.0019, "step": 80820 }, { "epoch": 1.3225885625460199, "grad_norm": 0.018717052415013313, "learning_rate": 8.465158245082095e-06, "loss": 0.0015, "step": 80830 }, { "epoch": 1.3227521884970956, "grad_norm": 0.06946135312318802, "learning_rate": 8.464643516002623e-06, "loss": 0.0026, "step": 80840 }, { "epoch": 1.3229158144481714, "grad_norm": 0.04063395410776138, "learning_rate": 8.464128716282017e-06, "loss": 0.0017, "step": 80850 }, { "epoch": 1.3230794403992472, "grad_norm": 0.14322450757026672, "learning_rate": 8.46361384593077e-06, "loss": 0.0025, "step": 80860 }, { "epoch": 1.3232430663503232, "grad_norm": 0.16890661418437958, "learning_rate": 8.463098904959382e-06, "loss": 0.0012, "step": 80870 }, { "epoch": 1.323406692301399, "grad_norm": 0.029746361076831818, "learning_rate": 8.462583893378356e-06, "loss": 0.0013, "step": 80880 }, { "epoch": 1.3235703182524747, "grad_norm": 0.07570669800043106, "learning_rate": 8.462068811198186e-06, "loss": 0.0021, "step": 80890 }, { "epoch": 1.3237339442035507, "grad_norm": 0.01509146485477686, "learning_rate": 8.461553658429378e-06, "loss": 0.0011, "step": 80900 }, { "epoch": 1.3238975701546265, "grad_norm": 0.05152295529842377, "learning_rate": 8.461038435082436e-06, "loss": 0.0018, "step": 80910 }, { "epoch": 1.3240611961057023, "grad_norm": 0.22951722145080566, "learning_rate": 8.460523141167862e-06, "loss": 0.0012, "step": 80920 }, { "epoch": 1.3242248220567783, "grad_norm": 0.10155807435512543, "learning_rate": 8.460007776696166e-06, "loss": 0.0012, "step": 80930 }, { "epoch": 1.324388448007854, "grad_norm": 0.1692497879266739, "learning_rate": 8.459492341677853e-06, "loss": 0.0022, "step": 80940 }, { "epoch": 1.3245520739589298, "grad_norm": 0.12931501865386963, "learning_rate": 8.458976836123435e-06, "loss": 0.0021, "step": 80950 }, { "epoch": 1.3247156999100058, "grad_norm": 0.09329165518283844, "learning_rate": 8.45846126004342e-06, "loss": 0.0019, "step": 80960 }, { "epoch": 1.3248793258610816, "grad_norm": 0.04432225972414017, "learning_rate": 8.45794561344832e-06, "loss": 0.001, "step": 80970 }, { "epoch": 1.3250429518121574, "grad_norm": 0.1003001257777214, "learning_rate": 8.457429896348653e-06, "loss": 0.0022, "step": 80980 }, { "epoch": 1.3252065777632334, "grad_norm": 0.06503021717071533, "learning_rate": 8.456914108754928e-06, "loss": 0.0014, "step": 80990 }, { "epoch": 1.3253702037143091, "grad_norm": 0.02680092677474022, "learning_rate": 8.456398250677665e-06, "loss": 0.0014, "step": 81000 }, { "epoch": 1.325533829665385, "grad_norm": 0.09770604968070984, "learning_rate": 8.455882322127381e-06, "loss": 0.0019, "step": 81010 }, { "epoch": 1.325697455616461, "grad_norm": 0.07513240724802017, "learning_rate": 8.455366323114596e-06, "loss": 0.0035, "step": 81020 }, { "epoch": 1.3258610815675367, "grad_norm": 0.05602920427918434, "learning_rate": 8.45485025364983e-06, "loss": 0.0023, "step": 81030 }, { "epoch": 1.3260247075186125, "grad_norm": 0.11492210626602173, "learning_rate": 8.454334113743605e-06, "loss": 0.0045, "step": 81040 }, { "epoch": 1.3261883334696882, "grad_norm": 0.047935813665390015, "learning_rate": 8.453817903406448e-06, "loss": 0.0017, "step": 81050 }, { "epoch": 1.326351959420764, "grad_norm": 0.039451342076063156, "learning_rate": 8.453301622648878e-06, "loss": 0.0011, "step": 81060 }, { "epoch": 1.32651558537184, "grad_norm": 0.036385420709848404, "learning_rate": 8.452785271481428e-06, "loss": 0.0016, "step": 81070 }, { "epoch": 1.3266792113229158, "grad_norm": 0.023682311177253723, "learning_rate": 8.45226884991462e-06, "loss": 0.0011, "step": 81080 }, { "epoch": 1.3268428372739915, "grad_norm": 0.13700604438781738, "learning_rate": 8.451752357958986e-06, "loss": 0.0023, "step": 81090 }, { "epoch": 1.3270064632250675, "grad_norm": 0.10246316343545914, "learning_rate": 8.45123579562506e-06, "loss": 0.0029, "step": 81100 }, { "epoch": 1.3271700891761433, "grad_norm": 0.17513881623744965, "learning_rate": 8.45071916292337e-06, "loss": 0.0012, "step": 81110 }, { "epoch": 1.327333715127219, "grad_norm": 0.16292697191238403, "learning_rate": 8.45020245986445e-06, "loss": 0.001, "step": 81120 }, { "epoch": 1.327497341078295, "grad_norm": 0.12625956535339355, "learning_rate": 8.449685686458837e-06, "loss": 0.0015, "step": 81130 }, { "epoch": 1.3276609670293709, "grad_norm": 0.18133175373077393, "learning_rate": 8.449168842717067e-06, "loss": 0.0025, "step": 81140 }, { "epoch": 1.3278245929804466, "grad_norm": 0.0639411136507988, "learning_rate": 8.448651928649676e-06, "loss": 0.001, "step": 81150 }, { "epoch": 1.3279882189315226, "grad_norm": 0.06151815876364708, "learning_rate": 8.448134944267206e-06, "loss": 0.0015, "step": 81160 }, { "epoch": 1.3281518448825984, "grad_norm": 0.2879434823989868, "learning_rate": 8.447617889580198e-06, "loss": 0.0027, "step": 81170 }, { "epoch": 1.3283154708336742, "grad_norm": 0.12227308750152588, "learning_rate": 8.447100764599192e-06, "loss": 0.0017, "step": 81180 }, { "epoch": 1.3284790967847502, "grad_norm": 0.0020495522767305374, "learning_rate": 8.446583569334734e-06, "loss": 0.0017, "step": 81190 }, { "epoch": 1.328642722735826, "grad_norm": 0.0646853968501091, "learning_rate": 8.446066303797366e-06, "loss": 0.0015, "step": 81200 }, { "epoch": 1.3288063486869017, "grad_norm": 0.031478554010391235, "learning_rate": 8.445548967997641e-06, "loss": 0.0018, "step": 81210 }, { "epoch": 1.3289699746379775, "grad_norm": 0.03462730720639229, "learning_rate": 8.4450315619461e-06, "loss": 0.0014, "step": 81220 }, { "epoch": 1.3291336005890535, "grad_norm": 0.03218173608183861, "learning_rate": 8.444514085653297e-06, "loss": 0.0012, "step": 81230 }, { "epoch": 1.3292972265401293, "grad_norm": 0.14204572141170502, "learning_rate": 8.44399653912978e-06, "loss": 0.002, "step": 81240 }, { "epoch": 1.329460852491205, "grad_norm": 0.1136992946267128, "learning_rate": 8.4434789223861e-06, "loss": 0.0018, "step": 81250 }, { "epoch": 1.3296244784422808, "grad_norm": 0.019463790580630302, "learning_rate": 8.442961235432818e-06, "loss": 0.0024, "step": 81260 }, { "epoch": 1.3297881043933568, "grad_norm": 0.020692283287644386, "learning_rate": 8.442443478280482e-06, "loss": 0.0011, "step": 81270 }, { "epoch": 1.3299517303444326, "grad_norm": 0.25536444783210754, "learning_rate": 8.441925650939653e-06, "loss": 0.0031, "step": 81280 }, { "epoch": 1.3301153562955084, "grad_norm": 0.08693768829107285, "learning_rate": 8.441407753420885e-06, "loss": 0.0038, "step": 81290 }, { "epoch": 1.3302789822465844, "grad_norm": 0.0640816017985344, "learning_rate": 8.440889785734742e-06, "loss": 0.0014, "step": 81300 }, { "epoch": 1.3304426081976601, "grad_norm": 0.11435449868440628, "learning_rate": 8.44037174789178e-06, "loss": 0.0017, "step": 81310 }, { "epoch": 1.330606234148736, "grad_norm": 0.04126932471990585, "learning_rate": 8.439853639902566e-06, "loss": 0.0019, "step": 81320 }, { "epoch": 1.330769860099812, "grad_norm": 0.015851864591240883, "learning_rate": 8.439335461777661e-06, "loss": 0.0016, "step": 81330 }, { "epoch": 1.3309334860508877, "grad_norm": 0.12898385524749756, "learning_rate": 8.438817213527632e-06, "loss": 0.0013, "step": 81340 }, { "epoch": 1.3310971120019635, "grad_norm": 0.12078318744897842, "learning_rate": 8.438298895163045e-06, "loss": 0.0016, "step": 81350 }, { "epoch": 1.3312607379530395, "grad_norm": 0.13624030351638794, "learning_rate": 8.437780506694465e-06, "loss": 0.0019, "step": 81360 }, { "epoch": 1.3314243639041152, "grad_norm": 0.11751702427864075, "learning_rate": 8.437262048132467e-06, "loss": 0.0021, "step": 81370 }, { "epoch": 1.331587989855191, "grad_norm": 0.019158102571964264, "learning_rate": 8.436743519487615e-06, "loss": 0.0017, "step": 81380 }, { "epoch": 1.331751615806267, "grad_norm": 0.08901119232177734, "learning_rate": 8.43622492077049e-06, "loss": 0.002, "step": 81390 }, { "epoch": 1.3319152417573428, "grad_norm": 0.0765245333313942, "learning_rate": 8.43570625199166e-06, "loss": 0.0017, "step": 81400 }, { "epoch": 1.3320788677084185, "grad_norm": 0.009231406264007092, "learning_rate": 8.4351875131617e-06, "loss": 0.0019, "step": 81410 }, { "epoch": 1.3322424936594943, "grad_norm": 0.019459137693047523, "learning_rate": 8.43466870429119e-06, "loss": 0.001, "step": 81420 }, { "epoch": 1.3324061196105703, "grad_norm": 0.3370632827281952, "learning_rate": 8.434149825390705e-06, "loss": 0.001, "step": 81430 }, { "epoch": 1.332569745561646, "grad_norm": 0.01758437044918537, "learning_rate": 8.433630876470826e-06, "loss": 0.0013, "step": 81440 }, { "epoch": 1.3327333715127219, "grad_norm": 0.04735654965043068, "learning_rate": 8.433111857542135e-06, "loss": 0.0022, "step": 81450 }, { "epoch": 1.3328969974637976, "grad_norm": 0.019772838801145554, "learning_rate": 8.43259276861521e-06, "loss": 0.0013, "step": 81460 }, { "epoch": 1.3330606234148736, "grad_norm": 0.08973066508769989, "learning_rate": 8.43207360970064e-06, "loss": 0.0022, "step": 81470 }, { "epoch": 1.3332242493659494, "grad_norm": 0.05673031508922577, "learning_rate": 8.431554380809008e-06, "loss": 0.0022, "step": 81480 }, { "epoch": 1.3333878753170252, "grad_norm": 0.01843651756644249, "learning_rate": 8.4310350819509e-06, "loss": 0.0011, "step": 81490 }, { "epoch": 1.3335515012681012, "grad_norm": 0.02606235072016716, "learning_rate": 8.430515713136904e-06, "loss": 0.0024, "step": 81500 }, { "epoch": 1.333715127219177, "grad_norm": 0.09767841547727585, "learning_rate": 8.429996274377612e-06, "loss": 0.0024, "step": 81510 }, { "epoch": 1.3338787531702527, "grad_norm": 0.030536381527781487, "learning_rate": 8.429476765683612e-06, "loss": 0.0029, "step": 81520 }, { "epoch": 1.3340423791213287, "grad_norm": 0.042406197637319565, "learning_rate": 8.428957187065496e-06, "loss": 0.0019, "step": 81530 }, { "epoch": 1.3342060050724045, "grad_norm": 0.002667035209015012, "learning_rate": 8.428437538533861e-06, "loss": 0.0021, "step": 81540 }, { "epoch": 1.3343696310234803, "grad_norm": 0.028336800634860992, "learning_rate": 8.427917820099301e-06, "loss": 0.0019, "step": 81550 }, { "epoch": 1.3345332569745563, "grad_norm": 0.03304385021328926, "learning_rate": 8.42739803177241e-06, "loss": 0.0017, "step": 81560 }, { "epoch": 1.334696882925632, "grad_norm": 0.06755725294351578, "learning_rate": 8.42687817356379e-06, "loss": 0.0028, "step": 81570 }, { "epoch": 1.3348605088767078, "grad_norm": 0.04635929316282272, "learning_rate": 8.426358245484036e-06, "loss": 0.0021, "step": 81580 }, { "epoch": 1.3350241348277838, "grad_norm": 0.22265447676181793, "learning_rate": 8.425838247543755e-06, "loss": 0.0014, "step": 81590 }, { "epoch": 1.3351877607788596, "grad_norm": 0.1310240924358368, "learning_rate": 8.425318179753542e-06, "loss": 0.0027, "step": 81600 }, { "epoch": 1.3353513867299354, "grad_norm": 0.07465910166501999, "learning_rate": 8.424798042124006e-06, "loss": 0.0025, "step": 81610 }, { "epoch": 1.3355150126810111, "grad_norm": 0.028617100790143013, "learning_rate": 8.42427783466575e-06, "loss": 0.0013, "step": 81620 }, { "epoch": 1.335678638632087, "grad_norm": 0.19794981181621552, "learning_rate": 8.423757557389383e-06, "loss": 0.0028, "step": 81630 }, { "epoch": 1.335842264583163, "grad_norm": 0.01908002607524395, "learning_rate": 8.42323721030551e-06, "loss": 0.0015, "step": 81640 }, { "epoch": 1.3360058905342387, "grad_norm": 0.06209902837872505, "learning_rate": 8.42271679342474e-06, "loss": 0.0017, "step": 81650 }, { "epoch": 1.3361695164853145, "grad_norm": 0.0470794215798378, "learning_rate": 8.422196306757689e-06, "loss": 0.0033, "step": 81660 }, { "epoch": 1.3363331424363905, "grad_norm": 0.0381462536752224, "learning_rate": 8.421675750314964e-06, "loss": 0.0021, "step": 81670 }, { "epoch": 1.3364967683874662, "grad_norm": 0.08970743417739868, "learning_rate": 8.42115512410718e-06, "loss": 0.0012, "step": 81680 }, { "epoch": 1.336660394338542, "grad_norm": 0.03504388779401779, "learning_rate": 8.420634428144953e-06, "loss": 0.0022, "step": 81690 }, { "epoch": 1.336824020289618, "grad_norm": 0.09269638359546661, "learning_rate": 8.420113662438898e-06, "loss": 0.001, "step": 81700 }, { "epoch": 1.3369876462406938, "grad_norm": 0.03669722005724907, "learning_rate": 8.419592826999637e-06, "loss": 0.002, "step": 81710 }, { "epoch": 1.3371512721917695, "grad_norm": 0.09657824784517288, "learning_rate": 8.419071921837784e-06, "loss": 0.0021, "step": 81720 }, { "epoch": 1.3373148981428455, "grad_norm": 0.05288926139473915, "learning_rate": 8.418550946963962e-06, "loss": 0.0018, "step": 81730 }, { "epoch": 1.3374785240939213, "grad_norm": 0.1271684765815735, "learning_rate": 8.418029902388796e-06, "loss": 0.0036, "step": 81740 }, { "epoch": 1.337642150044997, "grad_norm": 0.1514531672000885, "learning_rate": 8.417508788122906e-06, "loss": 0.0019, "step": 81750 }, { "epoch": 1.337805775996073, "grad_norm": 0.11756335198879242, "learning_rate": 8.416987604176918e-06, "loss": 0.0022, "step": 81760 }, { "epoch": 1.3379694019471489, "grad_norm": 0.033767808228731155, "learning_rate": 8.416466350561458e-06, "loss": 0.0022, "step": 81770 }, { "epoch": 1.3381330278982246, "grad_norm": 0.0026398582849651575, "learning_rate": 8.415945027287156e-06, "loss": 0.0014, "step": 81780 }, { "epoch": 1.3382966538493006, "grad_norm": 0.025422319769859314, "learning_rate": 8.41542363436464e-06, "loss": 0.001, "step": 81790 }, { "epoch": 1.3384602798003764, "grad_norm": 0.015165633521974087, "learning_rate": 8.414902171804542e-06, "loss": 0.0012, "step": 81800 }, { "epoch": 1.3386239057514522, "grad_norm": 0.08865697681903839, "learning_rate": 8.41438063961749e-06, "loss": 0.0015, "step": 81810 }, { "epoch": 1.338787531702528, "grad_norm": 0.2451353371143341, "learning_rate": 8.413859037814123e-06, "loss": 0.0032, "step": 81820 }, { "epoch": 1.3389511576536037, "grad_norm": 0.08514862507581711, "learning_rate": 8.413337366405073e-06, "loss": 0.0019, "step": 81830 }, { "epoch": 1.3391147836046797, "grad_norm": 0.06735774129629135, "learning_rate": 8.412815625400976e-06, "loss": 0.0016, "step": 81840 }, { "epoch": 1.3392784095557555, "grad_norm": 0.056210871785879135, "learning_rate": 8.412293814812471e-06, "loss": 0.0007, "step": 81850 }, { "epoch": 1.3394420355068313, "grad_norm": 0.09498138725757599, "learning_rate": 8.4117719346502e-06, "loss": 0.0011, "step": 81860 }, { "epoch": 1.3396056614579073, "grad_norm": 0.26585912704467773, "learning_rate": 8.411249984924797e-06, "loss": 0.0015, "step": 81870 }, { "epoch": 1.339769287408983, "grad_norm": 0.18199118971824646, "learning_rate": 8.410727965646909e-06, "loss": 0.0014, "step": 81880 }, { "epoch": 1.3399329133600588, "grad_norm": 0.0744047462940216, "learning_rate": 8.41020587682718e-06, "loss": 0.0017, "step": 81890 }, { "epoch": 1.3400965393111348, "grad_norm": 0.01612238585948944, "learning_rate": 8.409683718476253e-06, "loss": 0.0012, "step": 81900 }, { "epoch": 1.3402601652622106, "grad_norm": 0.015707900747656822, "learning_rate": 8.409161490604774e-06, "loss": 0.0011, "step": 81910 }, { "epoch": 1.3404237912132864, "grad_norm": 0.10248570889234543, "learning_rate": 8.408639193223392e-06, "loss": 0.0029, "step": 81920 }, { "epoch": 1.3405874171643624, "grad_norm": 0.14930205047130585, "learning_rate": 8.408116826342755e-06, "loss": 0.0025, "step": 81930 }, { "epoch": 1.3407510431154381, "grad_norm": 0.07672658562660217, "learning_rate": 8.407594389973517e-06, "loss": 0.0033, "step": 81940 }, { "epoch": 1.340914669066514, "grad_norm": 0.04548202455043793, "learning_rate": 8.407071884126326e-06, "loss": 0.002, "step": 81950 }, { "epoch": 1.34107829501759, "grad_norm": 0.048482201993465424, "learning_rate": 8.406549308811835e-06, "loss": 0.0031, "step": 81960 }, { "epoch": 1.3412419209686657, "grad_norm": 0.20060469210147858, "learning_rate": 8.406026664040703e-06, "loss": 0.0025, "step": 81970 }, { "epoch": 1.3414055469197415, "grad_norm": 0.04539802297949791, "learning_rate": 8.405503949823583e-06, "loss": 0.0013, "step": 81980 }, { "epoch": 1.3415691728708175, "grad_norm": 0.02800576388835907, "learning_rate": 8.404981166171135e-06, "loss": 0.002, "step": 81990 }, { "epoch": 1.3417327988218932, "grad_norm": 0.16304771602153778, "learning_rate": 8.404458313094015e-06, "loss": 0.0027, "step": 82000 }, { "epoch": 1.341896424772969, "grad_norm": 0.0721987783908844, "learning_rate": 8.403935390602886e-06, "loss": 0.002, "step": 82010 }, { "epoch": 1.3420600507240448, "grad_norm": 0.18163567781448364, "learning_rate": 8.403412398708411e-06, "loss": 0.0026, "step": 82020 }, { "epoch": 1.3422236766751205, "grad_norm": 0.12358362227678299, "learning_rate": 8.40288933742125e-06, "loss": 0.0011, "step": 82030 }, { "epoch": 1.3423873026261965, "grad_norm": 0.07966704666614532, "learning_rate": 8.40236620675207e-06, "loss": 0.0026, "step": 82040 }, { "epoch": 1.3425509285772723, "grad_norm": 0.08221210539340973, "learning_rate": 8.401843006711536e-06, "loss": 0.0026, "step": 82050 }, { "epoch": 1.342714554528348, "grad_norm": 0.1037357822060585, "learning_rate": 8.401319737310318e-06, "loss": 0.0024, "step": 82060 }, { "epoch": 1.342878180479424, "grad_norm": 0.09949656575918198, "learning_rate": 8.400796398559082e-06, "loss": 0.002, "step": 82070 }, { "epoch": 1.3430418064304999, "grad_norm": 0.10142141580581665, "learning_rate": 8.400272990468499e-06, "loss": 0.0013, "step": 82080 }, { "epoch": 1.3432054323815756, "grad_norm": 0.06434213370084763, "learning_rate": 8.399749513049243e-06, "loss": 0.0047, "step": 82090 }, { "epoch": 1.3433690583326516, "grad_norm": 0.07143118232488632, "learning_rate": 8.399225966311984e-06, "loss": 0.0016, "step": 82100 }, { "epoch": 1.3435326842837274, "grad_norm": 0.041630588471889496, "learning_rate": 8.3987023502674e-06, "loss": 0.0019, "step": 82110 }, { "epoch": 1.3436963102348032, "grad_norm": 0.022830409929156303, "learning_rate": 8.398178664926164e-06, "loss": 0.0022, "step": 82120 }, { "epoch": 1.3438599361858792, "grad_norm": 0.0620516873896122, "learning_rate": 8.397654910298956e-06, "loss": 0.0022, "step": 82130 }, { "epoch": 1.344023562136955, "grad_norm": 0.05269624665379524, "learning_rate": 8.397131086396455e-06, "loss": 0.0011, "step": 82140 }, { "epoch": 1.3441871880880307, "grad_norm": 0.0993862971663475, "learning_rate": 8.39660719322934e-06, "loss": 0.0035, "step": 82150 }, { "epoch": 1.3443508140391067, "grad_norm": 0.09038801491260529, "learning_rate": 8.396083230808292e-06, "loss": 0.0016, "step": 82160 }, { "epoch": 1.3445144399901825, "grad_norm": 0.04465785250067711, "learning_rate": 8.395559199143996e-06, "loss": 0.0021, "step": 82170 }, { "epoch": 1.3446780659412583, "grad_norm": 0.06708484143018723, "learning_rate": 8.395035098247136e-06, "loss": 0.0011, "step": 82180 }, { "epoch": 1.344841691892334, "grad_norm": 0.0077681634575128555, "learning_rate": 8.394510928128396e-06, "loss": 0.0027, "step": 82190 }, { "epoch": 1.34500531784341, "grad_norm": 0.0010217423550784588, "learning_rate": 8.393986688798468e-06, "loss": 0.0022, "step": 82200 }, { "epoch": 1.3451689437944858, "grad_norm": 0.01394124049693346, "learning_rate": 8.393462380268036e-06, "loss": 0.0014, "step": 82210 }, { "epoch": 1.3453325697455616, "grad_norm": 0.09940624237060547, "learning_rate": 8.392938002547793e-06, "loss": 0.0022, "step": 82220 }, { "epoch": 1.3454961956966374, "grad_norm": 0.07468611001968384, "learning_rate": 8.392413555648428e-06, "loss": 0.0066, "step": 82230 }, { "epoch": 1.3456598216477134, "grad_norm": 0.061204977333545685, "learning_rate": 8.391889039580637e-06, "loss": 0.0021, "step": 82240 }, { "epoch": 1.3458234475987891, "grad_norm": 0.02483614906668663, "learning_rate": 8.391364454355112e-06, "loss": 0.002, "step": 82250 }, { "epoch": 1.345987073549865, "grad_norm": 0.17554126679897308, "learning_rate": 8.390839799982552e-06, "loss": 0.0015, "step": 82260 }, { "epoch": 1.346150699500941, "grad_norm": 0.0074843172915279865, "learning_rate": 8.390315076473652e-06, "loss": 0.0017, "step": 82270 }, { "epoch": 1.3463143254520167, "grad_norm": 0.0890878438949585, "learning_rate": 8.389790283839109e-06, "loss": 0.0017, "step": 82280 }, { "epoch": 1.3464779514030925, "grad_norm": 0.04335111007094383, "learning_rate": 8.389265422089627e-06, "loss": 0.001, "step": 82290 }, { "epoch": 1.3466415773541685, "grad_norm": 0.014910552650690079, "learning_rate": 8.388740491235905e-06, "loss": 0.0029, "step": 82300 }, { "epoch": 1.3468052033052442, "grad_norm": 0.0178558100014925, "learning_rate": 8.388215491288645e-06, "loss": 0.0031, "step": 82310 }, { "epoch": 1.34696882925632, "grad_norm": 0.0568411722779274, "learning_rate": 8.387690422258554e-06, "loss": 0.0023, "step": 82320 }, { "epoch": 1.347132455207396, "grad_norm": 0.06904259324073792, "learning_rate": 8.387165284156337e-06, "loss": 0.003, "step": 82330 }, { "epoch": 1.3472960811584718, "grad_norm": 0.043462980538606644, "learning_rate": 8.386640076992701e-06, "loss": 0.0017, "step": 82340 }, { "epoch": 1.3474597071095475, "grad_norm": 0.04479461535811424, "learning_rate": 8.38611480077835e-06, "loss": 0.0018, "step": 82350 }, { "epoch": 1.3476233330606235, "grad_norm": 0.04067343473434448, "learning_rate": 8.385589455524002e-06, "loss": 0.0013, "step": 82360 }, { "epoch": 1.3477869590116993, "grad_norm": 0.04551088437438011, "learning_rate": 8.385064041240362e-06, "loss": 0.0023, "step": 82370 }, { "epoch": 1.347950584962775, "grad_norm": 0.0383855476975441, "learning_rate": 8.384538557938147e-06, "loss": 0.0018, "step": 82380 }, { "epoch": 1.3481142109138509, "grad_norm": 0.1333918422460556, "learning_rate": 8.384013005628068e-06, "loss": 0.0031, "step": 82390 }, { "epoch": 1.3482778368649269, "grad_norm": 0.08675937354564667, "learning_rate": 8.383487384320844e-06, "loss": 0.0022, "step": 82400 }, { "epoch": 1.3484414628160026, "grad_norm": 0.07121489197015762, "learning_rate": 8.382961694027187e-06, "loss": 0.0016, "step": 82410 }, { "epoch": 1.3486050887670784, "grad_norm": 0.040899187326431274, "learning_rate": 8.38243593475782e-06, "loss": 0.0015, "step": 82420 }, { "epoch": 1.3487687147181542, "grad_norm": 0.08739755302667618, "learning_rate": 8.38191010652346e-06, "loss": 0.0026, "step": 82430 }, { "epoch": 1.3489323406692302, "grad_norm": 0.025585658848285675, "learning_rate": 8.38138420933483e-06, "loss": 0.0011, "step": 82440 }, { "epoch": 1.349095966620306, "grad_norm": 0.05434189364314079, "learning_rate": 8.380858243202653e-06, "loss": 0.002, "step": 82450 }, { "epoch": 1.3492595925713817, "grad_norm": 0.3682818114757538, "learning_rate": 8.380332208137648e-06, "loss": 0.0031, "step": 82460 }, { "epoch": 1.3494232185224577, "grad_norm": 0.11299610137939453, "learning_rate": 8.379806104150546e-06, "loss": 0.0016, "step": 82470 }, { "epoch": 1.3495868444735335, "grad_norm": 0.07201839983463287, "learning_rate": 8.379279931252072e-06, "loss": 0.0015, "step": 82480 }, { "epoch": 1.3497504704246093, "grad_norm": 0.06631486862897873, "learning_rate": 8.378753689452954e-06, "loss": 0.0017, "step": 82490 }, { "epoch": 1.3499140963756853, "grad_norm": 0.035372450947761536, "learning_rate": 8.378227378763923e-06, "loss": 0.0034, "step": 82500 }, { "epoch": 1.350077722326761, "grad_norm": 0.02663424424827099, "learning_rate": 8.377700999195707e-06, "loss": 0.0013, "step": 82510 }, { "epoch": 1.3502413482778368, "grad_norm": 0.05635925754904747, "learning_rate": 8.377174550759043e-06, "loss": 0.0023, "step": 82520 }, { "epoch": 1.3504049742289128, "grad_norm": 0.0417541079223156, "learning_rate": 8.376648033464659e-06, "loss": 0.003, "step": 82530 }, { "epoch": 1.3505686001799886, "grad_norm": 0.27435922622680664, "learning_rate": 8.376121447323294e-06, "loss": 0.0037, "step": 82540 }, { "epoch": 1.3507322261310644, "grad_norm": 0.019706591963768005, "learning_rate": 8.375594792345685e-06, "loss": 0.0016, "step": 82550 }, { "epoch": 1.3508958520821404, "grad_norm": 0.18248942494392395, "learning_rate": 8.375068068542568e-06, "loss": 0.0012, "step": 82560 }, { "epoch": 1.3510594780332161, "grad_norm": 0.04129261150956154, "learning_rate": 8.374541275924684e-06, "loss": 0.0023, "step": 82570 }, { "epoch": 1.351223103984292, "grad_norm": 0.08431877940893173, "learning_rate": 8.374014414502774e-06, "loss": 0.0024, "step": 82580 }, { "epoch": 1.3513867299353677, "grad_norm": 0.06714943051338196, "learning_rate": 8.373487484287577e-06, "loss": 0.0014, "step": 82590 }, { "epoch": 1.3515503558864435, "grad_norm": 0.01794588193297386, "learning_rate": 8.372960485289843e-06, "loss": 0.002, "step": 82600 }, { "epoch": 1.3517139818375195, "grad_norm": 0.12829793989658356, "learning_rate": 8.372433417520312e-06, "loss": 0.0019, "step": 82610 }, { "epoch": 1.3518776077885952, "grad_norm": 0.013311557471752167, "learning_rate": 8.37190628098973e-06, "loss": 0.0031, "step": 82620 }, { "epoch": 1.352041233739671, "grad_norm": 0.034221332520246506, "learning_rate": 8.371379075708847e-06, "loss": 0.0008, "step": 82630 }, { "epoch": 1.352204859690747, "grad_norm": 0.055220529437065125, "learning_rate": 8.370851801688413e-06, "loss": 0.0023, "step": 82640 }, { "epoch": 1.3523684856418228, "grad_norm": 0.08334865421056747, "learning_rate": 8.370324458939177e-06, "loss": 0.0022, "step": 82650 }, { "epoch": 1.3525321115928985, "grad_norm": 0.13797034323215485, "learning_rate": 8.369797047471893e-06, "loss": 0.0017, "step": 82660 }, { "epoch": 1.3526957375439745, "grad_norm": 0.05831380560994148, "learning_rate": 8.369269567297311e-06, "loss": 0.0013, "step": 82670 }, { "epoch": 1.3528593634950503, "grad_norm": 0.12490341812372208, "learning_rate": 8.36874201842619e-06, "loss": 0.0017, "step": 82680 }, { "epoch": 1.353022989446126, "grad_norm": 0.05225791782140732, "learning_rate": 8.368214400869283e-06, "loss": 0.005, "step": 82690 }, { "epoch": 1.353186615397202, "grad_norm": 0.017807986587285995, "learning_rate": 8.367686714637348e-06, "loss": 0.0039, "step": 82700 }, { "epoch": 1.3533502413482779, "grad_norm": 0.10569663345813751, "learning_rate": 8.367158959741147e-06, "loss": 0.0018, "step": 82710 }, { "epoch": 1.3535138672993536, "grad_norm": 0.1016215831041336, "learning_rate": 8.36663113619144e-06, "loss": 0.0017, "step": 82720 }, { "epoch": 1.3536774932504296, "grad_norm": 0.022139092907309532, "learning_rate": 8.366103243998983e-06, "loss": 0.0029, "step": 82730 }, { "epoch": 1.3538411192015054, "grad_norm": 0.06685173511505127, "learning_rate": 8.365575283174546e-06, "loss": 0.0017, "step": 82740 }, { "epoch": 1.3540047451525812, "grad_norm": 0.06670430302619934, "learning_rate": 8.365047253728891e-06, "loss": 0.0027, "step": 82750 }, { "epoch": 1.3541683711036572, "grad_norm": 0.019497768953442574, "learning_rate": 8.364519155672783e-06, "loss": 0.002, "step": 82760 }, { "epoch": 1.354331997054733, "grad_norm": 0.01990545354783535, "learning_rate": 8.363990989016993e-06, "loss": 0.0013, "step": 82770 }, { "epoch": 1.3544956230058087, "grad_norm": 0.14576856791973114, "learning_rate": 8.363462753772287e-06, "loss": 0.0016, "step": 82780 }, { "epoch": 1.3546592489568845, "grad_norm": 0.08281613886356354, "learning_rate": 8.362934449949437e-06, "loss": 0.0017, "step": 82790 }, { "epoch": 1.3548228749079603, "grad_norm": 0.008353314362466335, "learning_rate": 8.362406077559212e-06, "loss": 0.001, "step": 82800 }, { "epoch": 1.3549865008590363, "grad_norm": 0.07283806055784225, "learning_rate": 8.361877636612386e-06, "loss": 0.0033, "step": 82810 }, { "epoch": 1.355150126810112, "grad_norm": 0.07985186576843262, "learning_rate": 8.361349127119735e-06, "loss": 0.0026, "step": 82820 }, { "epoch": 1.3553137527611878, "grad_norm": 0.012383552268147469, "learning_rate": 8.360820549092035e-06, "loss": 0.003, "step": 82830 }, { "epoch": 1.3554773787122638, "grad_norm": 0.07304152101278305, "learning_rate": 8.360291902540062e-06, "loss": 0.0015, "step": 82840 }, { "epoch": 1.3556410046633396, "grad_norm": 0.023742644116282463, "learning_rate": 8.359763187474595e-06, "loss": 0.0015, "step": 82850 }, { "epoch": 1.3558046306144154, "grad_norm": 0.056631628423929214, "learning_rate": 8.359234403906413e-06, "loss": 0.0017, "step": 82860 }, { "epoch": 1.3559682565654914, "grad_norm": 0.07229973375797272, "learning_rate": 8.3587055518463e-06, "loss": 0.0012, "step": 82870 }, { "epoch": 1.3561318825165671, "grad_norm": 0.09687382727861404, "learning_rate": 8.358176631305036e-06, "loss": 0.0016, "step": 82880 }, { "epoch": 1.356295508467643, "grad_norm": 0.06749277561903, "learning_rate": 8.357647642293408e-06, "loss": 0.0018, "step": 82890 }, { "epoch": 1.356459134418719, "grad_norm": 0.07605363428592682, "learning_rate": 8.357118584822197e-06, "loss": 0.0022, "step": 82900 }, { "epoch": 1.3566227603697947, "grad_norm": 0.03284405544400215, "learning_rate": 8.356589458902195e-06, "loss": 0.0032, "step": 82910 }, { "epoch": 1.3567863863208705, "grad_norm": 0.04315026476979256, "learning_rate": 8.35606026454419e-06, "loss": 0.0013, "step": 82920 }, { "epoch": 1.3569500122719464, "grad_norm": 0.08775345236063004, "learning_rate": 8.355531001758969e-06, "loss": 0.0018, "step": 82930 }, { "epoch": 1.3571136382230222, "grad_norm": 0.04019992798566818, "learning_rate": 8.355001670557324e-06, "loss": 0.0016, "step": 82940 }, { "epoch": 1.357277264174098, "grad_norm": 0.12267586588859558, "learning_rate": 8.354472270950049e-06, "loss": 0.0015, "step": 82950 }, { "epoch": 1.3574408901251738, "grad_norm": 0.0593767911195755, "learning_rate": 8.353942802947938e-06, "loss": 0.0015, "step": 82960 }, { "epoch": 1.3576045160762498, "grad_norm": 0.110775887966156, "learning_rate": 8.353413266561784e-06, "loss": 0.0018, "step": 82970 }, { "epoch": 1.3577681420273255, "grad_norm": 0.26335927844047546, "learning_rate": 8.352883661802388e-06, "loss": 0.0021, "step": 82980 }, { "epoch": 1.3579317679784013, "grad_norm": 0.049374427646398544, "learning_rate": 8.352353988680544e-06, "loss": 0.0022, "step": 82990 }, { "epoch": 1.358095393929477, "grad_norm": 0.22851930558681488, "learning_rate": 8.351824247207053e-06, "loss": 0.0045, "step": 83000 }, { "epoch": 1.358259019880553, "grad_norm": 0.017566639930009842, "learning_rate": 8.351294437392716e-06, "loss": 0.0015, "step": 83010 }, { "epoch": 1.3584226458316289, "grad_norm": 0.011990956030786037, "learning_rate": 8.350764559248336e-06, "loss": 0.0021, "step": 83020 }, { "epoch": 1.3585862717827046, "grad_norm": 0.08286643773317337, "learning_rate": 8.350234612784718e-06, "loss": 0.0012, "step": 83030 }, { "epoch": 1.3587498977337806, "grad_norm": 0.030376087874174118, "learning_rate": 8.349704598012664e-06, "loss": 0.0017, "step": 83040 }, { "epoch": 1.3589135236848564, "grad_norm": 0.05962565541267395, "learning_rate": 8.34917451494298e-06, "loss": 0.002, "step": 83050 }, { "epoch": 1.3590771496359322, "grad_norm": 0.03910797834396362, "learning_rate": 8.34864436358648e-06, "loss": 0.0014, "step": 83060 }, { "epoch": 1.3592407755870082, "grad_norm": 0.06926596164703369, "learning_rate": 8.348114143953967e-06, "loss": 0.0033, "step": 83070 }, { "epoch": 1.359404401538084, "grad_norm": 0.049377258867025375, "learning_rate": 8.347583856056255e-06, "loss": 0.0014, "step": 83080 }, { "epoch": 1.3595680274891597, "grad_norm": 0.07751760631799698, "learning_rate": 8.347053499904156e-06, "loss": 0.0016, "step": 83090 }, { "epoch": 1.3597316534402357, "grad_norm": 0.06400879472494125, "learning_rate": 8.346523075508481e-06, "loss": 0.0017, "step": 83100 }, { "epoch": 1.3598952793913115, "grad_norm": 0.17089076340198517, "learning_rate": 8.345992582880047e-06, "loss": 0.0019, "step": 83110 }, { "epoch": 1.3600589053423873, "grad_norm": 0.04422549158334732, "learning_rate": 8.34546202202967e-06, "loss": 0.0027, "step": 83120 }, { "epoch": 1.3602225312934633, "grad_norm": 0.06586378812789917, "learning_rate": 8.344931392968168e-06, "loss": 0.0021, "step": 83130 }, { "epoch": 1.360386157244539, "grad_norm": 0.0806720182299614, "learning_rate": 8.344400695706358e-06, "loss": 0.002, "step": 83140 }, { "epoch": 1.3605497831956148, "grad_norm": 0.03822488710284233, "learning_rate": 8.343869930255064e-06, "loss": 0.0014, "step": 83150 }, { "epoch": 1.3607134091466906, "grad_norm": 0.07760681957006454, "learning_rate": 8.343339096625104e-06, "loss": 0.0021, "step": 83160 }, { "epoch": 1.3608770350977666, "grad_norm": 0.036846477538347244, "learning_rate": 8.342808194827306e-06, "loss": 0.001, "step": 83170 }, { "epoch": 1.3610406610488424, "grad_norm": 0.03489351645112038, "learning_rate": 8.34227722487249e-06, "loss": 0.0018, "step": 83180 }, { "epoch": 1.3612042869999181, "grad_norm": 0.0028465145733207464, "learning_rate": 8.341746186771484e-06, "loss": 0.0014, "step": 83190 }, { "epoch": 1.361367912950994, "grad_norm": 0.015042927116155624, "learning_rate": 8.341215080535117e-06, "loss": 0.0018, "step": 83200 }, { "epoch": 1.36153153890207, "grad_norm": 0.03350342437624931, "learning_rate": 8.340683906174214e-06, "loss": 0.0031, "step": 83210 }, { "epoch": 1.3616951648531457, "grad_norm": 0.010758471675217152, "learning_rate": 8.340152663699607e-06, "loss": 0.0018, "step": 83220 }, { "epoch": 1.3618587908042215, "grad_norm": 0.02171928435564041, "learning_rate": 8.33962135312213e-06, "loss": 0.0011, "step": 83230 }, { "epoch": 1.3620224167552974, "grad_norm": 0.18250896036624908, "learning_rate": 8.339089974452613e-06, "loss": 0.0021, "step": 83240 }, { "epoch": 1.3621860427063732, "grad_norm": 0.15097659826278687, "learning_rate": 8.33855852770189e-06, "loss": 0.0027, "step": 83250 }, { "epoch": 1.362349668657449, "grad_norm": 0.09617456048727036, "learning_rate": 8.3380270128808e-06, "loss": 0.0048, "step": 83260 }, { "epoch": 1.362513294608525, "grad_norm": 0.01084696315228939, "learning_rate": 8.337495430000178e-06, "loss": 0.002, "step": 83270 }, { "epoch": 1.3626769205596008, "grad_norm": 0.009824035689234734, "learning_rate": 8.336963779070861e-06, "loss": 0.002, "step": 83280 }, { "epoch": 1.3628405465106765, "grad_norm": 0.005880730226635933, "learning_rate": 8.336432060103692e-06, "loss": 0.0014, "step": 83290 }, { "epoch": 1.3630041724617525, "grad_norm": 0.10739075392484665, "learning_rate": 8.33590027310951e-06, "loss": 0.002, "step": 83300 }, { "epoch": 1.3631677984128283, "grad_norm": 0.053448691964149475, "learning_rate": 8.335368418099158e-06, "loss": 0.003, "step": 83310 }, { "epoch": 1.363331424363904, "grad_norm": 0.11812595278024673, "learning_rate": 8.33483649508348e-06, "loss": 0.0019, "step": 83320 }, { "epoch": 1.36349505031498, "grad_norm": 0.0836363360285759, "learning_rate": 8.334304504073325e-06, "loss": 0.0028, "step": 83330 }, { "epoch": 1.3636586762660559, "grad_norm": 0.09773733466863632, "learning_rate": 8.333772445079533e-06, "loss": 0.0023, "step": 83340 }, { "epoch": 1.3638223022171316, "grad_norm": 0.10959619283676147, "learning_rate": 8.33324031811296e-06, "loss": 0.0035, "step": 83350 }, { "epoch": 1.3639859281682074, "grad_norm": 0.001682510832324624, "learning_rate": 8.33270812318445e-06, "loss": 0.0011, "step": 83360 }, { "epoch": 1.3641495541192832, "grad_norm": 0.1330232173204422, "learning_rate": 8.332175860304855e-06, "loss": 0.0015, "step": 83370 }, { "epoch": 1.3643131800703592, "grad_norm": 0.035174768418073654, "learning_rate": 8.33164352948503e-06, "loss": 0.0012, "step": 83380 }, { "epoch": 1.364476806021435, "grad_norm": 0.10841120779514313, "learning_rate": 8.331111130735824e-06, "loss": 0.0013, "step": 83390 }, { "epoch": 1.3646404319725107, "grad_norm": 0.016846273094415665, "learning_rate": 8.330578664068097e-06, "loss": 0.0018, "step": 83400 }, { "epoch": 1.3648040579235867, "grad_norm": 0.20695029199123383, "learning_rate": 8.330046129492704e-06, "loss": 0.0021, "step": 83410 }, { "epoch": 1.3649676838746625, "grad_norm": 0.052226148545742035, "learning_rate": 8.3295135270205e-06, "loss": 0.0011, "step": 83420 }, { "epoch": 1.3651313098257383, "grad_norm": 0.003929703030735254, "learning_rate": 8.32898085666235e-06, "loss": 0.002, "step": 83430 }, { "epoch": 1.3652949357768143, "grad_norm": 0.07914528250694275, "learning_rate": 8.32844811842911e-06, "loss": 0.0014, "step": 83440 }, { "epoch": 1.36545856172789, "grad_norm": 0.0038506099954247475, "learning_rate": 8.327915312331646e-06, "loss": 0.0015, "step": 83450 }, { "epoch": 1.3656221876789658, "grad_norm": 0.016254665330052376, "learning_rate": 8.327382438380816e-06, "loss": 0.002, "step": 83460 }, { "epoch": 1.3657858136300418, "grad_norm": 0.17953211069107056, "learning_rate": 8.326849496587487e-06, "loss": 0.0015, "step": 83470 }, { "epoch": 1.3659494395811176, "grad_norm": 0.043870072811841965, "learning_rate": 8.326316486962529e-06, "loss": 0.0018, "step": 83480 }, { "epoch": 1.3661130655321934, "grad_norm": 0.11929380148649216, "learning_rate": 8.325783409516806e-06, "loss": 0.0017, "step": 83490 }, { "epoch": 1.3662766914832694, "grad_norm": 0.07310284674167633, "learning_rate": 8.325250264261187e-06, "loss": 0.0018, "step": 83500 }, { "epoch": 1.3664403174343451, "grad_norm": 0.03463859483599663, "learning_rate": 8.324717051206543e-06, "loss": 0.0022, "step": 83510 }, { "epoch": 1.366603943385421, "grad_norm": 0.03111531212925911, "learning_rate": 8.324183770363747e-06, "loss": 0.002, "step": 83520 }, { "epoch": 1.366767569336497, "grad_norm": 0.05577366054058075, "learning_rate": 8.323650421743672e-06, "loss": 0.0021, "step": 83530 }, { "epoch": 1.3669311952875727, "grad_norm": 0.02931862697005272, "learning_rate": 8.323117005357188e-06, "loss": 0.0025, "step": 83540 }, { "epoch": 1.3670948212386484, "grad_norm": 0.109883151948452, "learning_rate": 8.322583521215177e-06, "loss": 0.002, "step": 83550 }, { "epoch": 1.3672584471897242, "grad_norm": 0.10552117228507996, "learning_rate": 8.322049969328515e-06, "loss": 0.0027, "step": 83560 }, { "epoch": 1.3674220731408, "grad_norm": 0.0533643402159214, "learning_rate": 8.321516349708076e-06, "loss": 0.0018, "step": 83570 }, { "epoch": 1.367585699091876, "grad_norm": 0.04200919345021248, "learning_rate": 8.320982662364746e-06, "loss": 0.0018, "step": 83580 }, { "epoch": 1.3677493250429518, "grad_norm": 0.008127731271088123, "learning_rate": 8.320448907309405e-06, "loss": 0.001, "step": 83590 }, { "epoch": 1.3679129509940275, "grad_norm": 0.1287589818239212, "learning_rate": 8.319915084552932e-06, "loss": 0.0021, "step": 83600 }, { "epoch": 1.3680765769451035, "grad_norm": 0.05522772669792175, "learning_rate": 8.319381194106214e-06, "loss": 0.0011, "step": 83610 }, { "epoch": 1.3682402028961793, "grad_norm": 0.014168796129524708, "learning_rate": 8.318847235980138e-06, "loss": 0.001, "step": 83620 }, { "epoch": 1.368403828847255, "grad_norm": 0.02642037905752659, "learning_rate": 8.318313210185588e-06, "loss": 0.002, "step": 83630 }, { "epoch": 1.368567454798331, "grad_norm": 0.13152889907360077, "learning_rate": 8.317779116733455e-06, "loss": 0.003, "step": 83640 }, { "epoch": 1.3687310807494069, "grad_norm": 0.04444776475429535, "learning_rate": 8.317244955634627e-06, "loss": 0.0017, "step": 83650 }, { "epoch": 1.3688947067004826, "grad_norm": 0.06838791817426682, "learning_rate": 8.316710726899994e-06, "loss": 0.0014, "step": 83660 }, { "epoch": 1.3690583326515586, "grad_norm": 0.1087026372551918, "learning_rate": 8.316176430540453e-06, "loss": 0.0017, "step": 83670 }, { "epoch": 1.3692219586026344, "grad_norm": 0.044564057141542435, "learning_rate": 8.315642066566893e-06, "loss": 0.0021, "step": 83680 }, { "epoch": 1.3693855845537102, "grad_norm": 0.024873869493603706, "learning_rate": 8.315107634990211e-06, "loss": 0.0021, "step": 83690 }, { "epoch": 1.3695492105047862, "grad_norm": 0.09703509509563446, "learning_rate": 8.314573135821304e-06, "loss": 0.0014, "step": 83700 }, { "epoch": 1.369712836455862, "grad_norm": 0.08340300619602203, "learning_rate": 8.314038569071072e-06, "loss": 0.0014, "step": 83710 }, { "epoch": 1.3698764624069377, "grad_norm": 0.024968991056084633, "learning_rate": 8.31350393475041e-06, "loss": 0.0019, "step": 83720 }, { "epoch": 1.3700400883580137, "grad_norm": 0.030250493437051773, "learning_rate": 8.31296923287022e-06, "loss": 0.0022, "step": 83730 }, { "epoch": 1.3702037143090895, "grad_norm": 0.05956198647618294, "learning_rate": 8.312434463441405e-06, "loss": 0.0016, "step": 83740 }, { "epoch": 1.3703673402601653, "grad_norm": 0.04404841363430023, "learning_rate": 8.311899626474871e-06, "loss": 0.0025, "step": 83750 }, { "epoch": 1.370530966211241, "grad_norm": 0.03941219672560692, "learning_rate": 8.311364721981517e-06, "loss": 0.0011, "step": 83760 }, { "epoch": 1.3706945921623168, "grad_norm": 0.08032143861055374, "learning_rate": 8.310829749972254e-06, "loss": 0.0023, "step": 83770 }, { "epoch": 1.3708582181133928, "grad_norm": 0.16431625187397003, "learning_rate": 8.31029471045799e-06, "loss": 0.0029, "step": 83780 }, { "epoch": 1.3710218440644686, "grad_norm": 0.04794597625732422, "learning_rate": 8.30975960344963e-06, "loss": 0.0037, "step": 83790 }, { "epoch": 1.3711854700155444, "grad_norm": 0.07986754179000854, "learning_rate": 8.309224428958087e-06, "loss": 0.0023, "step": 83800 }, { "epoch": 1.3713490959666204, "grad_norm": 0.23188211023807526, "learning_rate": 8.308689186994272e-06, "loss": 0.005, "step": 83810 }, { "epoch": 1.3715127219176961, "grad_norm": 0.0958874523639679, "learning_rate": 8.3081538775691e-06, "loss": 0.0015, "step": 83820 }, { "epoch": 1.371676347868772, "grad_norm": 0.03017180599272251, "learning_rate": 8.30761850069348e-06, "loss": 0.0013, "step": 83830 }, { "epoch": 1.371839973819848, "grad_norm": 0.03906385600566864, "learning_rate": 8.307083056378336e-06, "loss": 0.0015, "step": 83840 }, { "epoch": 1.3720035997709237, "grad_norm": 0.051623739302158356, "learning_rate": 8.306547544634578e-06, "loss": 0.0011, "step": 83850 }, { "epoch": 1.3721672257219995, "grad_norm": 0.10702141374349594, "learning_rate": 8.306011965473129e-06, "loss": 0.0018, "step": 83860 }, { "epoch": 1.3723308516730754, "grad_norm": 0.019821299239993095, "learning_rate": 8.305476318904908e-06, "loss": 0.0012, "step": 83870 }, { "epoch": 1.3724944776241512, "grad_norm": 0.11168746650218964, "learning_rate": 8.304940604940836e-06, "loss": 0.0022, "step": 83880 }, { "epoch": 1.372658103575227, "grad_norm": 0.06022688001394272, "learning_rate": 8.304404823591836e-06, "loss": 0.0024, "step": 83890 }, { "epoch": 1.372821729526303, "grad_norm": 0.014144851826131344, "learning_rate": 8.303868974868831e-06, "loss": 0.0015, "step": 83900 }, { "epoch": 1.3729853554773788, "grad_norm": 0.04052388295531273, "learning_rate": 8.303333058782747e-06, "loss": 0.0016, "step": 83910 }, { "epoch": 1.3731489814284545, "grad_norm": 0.03047829121351242, "learning_rate": 8.302797075344514e-06, "loss": 0.0029, "step": 83920 }, { "epoch": 1.3733126073795303, "grad_norm": 0.042534343898296356, "learning_rate": 8.302261024565055e-06, "loss": 0.0022, "step": 83930 }, { "epoch": 1.3734762333306063, "grad_norm": 0.04086651653051376, "learning_rate": 8.301724906455305e-06, "loss": 0.004, "step": 83940 }, { "epoch": 1.373639859281682, "grad_norm": 0.10935360193252563, "learning_rate": 8.301188721026188e-06, "loss": 0.0014, "step": 83950 }, { "epoch": 1.3738034852327579, "grad_norm": 0.02599087730050087, "learning_rate": 8.300652468288643e-06, "loss": 0.0022, "step": 83960 }, { "epoch": 1.3739671111838336, "grad_norm": 0.045303456485271454, "learning_rate": 8.300116148253601e-06, "loss": 0.0017, "step": 83970 }, { "epoch": 1.3741307371349096, "grad_norm": 0.07494262605905533, "learning_rate": 8.299579760931998e-06, "loss": 0.0012, "step": 83980 }, { "epoch": 1.3742943630859854, "grad_norm": 0.1538262814283371, "learning_rate": 8.299043306334768e-06, "loss": 0.0018, "step": 83990 }, { "epoch": 1.3744579890370612, "grad_norm": 0.10152026265859604, "learning_rate": 8.298506784472852e-06, "loss": 0.0021, "step": 84000 }, { "epoch": 1.3746216149881372, "grad_norm": 0.08454479277133942, "learning_rate": 8.297970195357189e-06, "loss": 0.0021, "step": 84010 }, { "epoch": 1.374785240939213, "grad_norm": 0.104360431432724, "learning_rate": 8.297433538998718e-06, "loss": 0.001, "step": 84020 }, { "epoch": 1.3749488668902887, "grad_norm": 0.03399254009127617, "learning_rate": 8.29689681540838e-06, "loss": 0.0014, "step": 84030 }, { "epoch": 1.3751124928413647, "grad_norm": 0.05746503174304962, "learning_rate": 8.296360024597122e-06, "loss": 0.0043, "step": 84040 }, { "epoch": 1.3752761187924405, "grad_norm": 0.04036037251353264, "learning_rate": 8.295823166575886e-06, "loss": 0.0028, "step": 84050 }, { "epoch": 1.3754397447435163, "grad_norm": 0.061715900897979736, "learning_rate": 8.295286241355616e-06, "loss": 0.0019, "step": 84060 }, { "epoch": 1.3756033706945923, "grad_norm": 0.0879318118095398, "learning_rate": 8.294749248947265e-06, "loss": 0.0014, "step": 84070 }, { "epoch": 1.375766996645668, "grad_norm": 0.13148580491542816, "learning_rate": 8.294212189361778e-06, "loss": 0.0022, "step": 84080 }, { "epoch": 1.3759306225967438, "grad_norm": 0.10657060891389847, "learning_rate": 8.293675062610105e-06, "loss": 0.0021, "step": 84090 }, { "epoch": 1.3760942485478198, "grad_norm": 0.13791923224925995, "learning_rate": 8.2931378687032e-06, "loss": 0.0032, "step": 84100 }, { "epoch": 1.3762578744988956, "grad_norm": 0.05140616372227669, "learning_rate": 8.292600607652015e-06, "loss": 0.0017, "step": 84110 }, { "epoch": 1.3764215004499714, "grad_norm": 0.15264207124710083, "learning_rate": 8.292063279467503e-06, "loss": 0.0015, "step": 84120 }, { "epoch": 1.3765851264010471, "grad_norm": 0.20059964060783386, "learning_rate": 8.29152588416062e-06, "loss": 0.0028, "step": 84130 }, { "epoch": 1.3767487523521231, "grad_norm": 0.15003424882888794, "learning_rate": 8.290988421742325e-06, "loss": 0.0023, "step": 84140 }, { "epoch": 1.376912378303199, "grad_norm": 0.04598694667220116, "learning_rate": 8.290450892223574e-06, "loss": 0.0021, "step": 84150 }, { "epoch": 1.3770760042542747, "grad_norm": 0.17800405621528625, "learning_rate": 8.289913295615328e-06, "loss": 0.0041, "step": 84160 }, { "epoch": 1.3772396302053505, "grad_norm": 0.18481266498565674, "learning_rate": 8.28937563192855e-06, "loss": 0.0021, "step": 84170 }, { "epoch": 1.3774032561564264, "grad_norm": 0.07227969914674759, "learning_rate": 8.288837901174198e-06, "loss": 0.0018, "step": 84180 }, { "epoch": 1.3775668821075022, "grad_norm": 0.044101204723119736, "learning_rate": 8.288300103363239e-06, "loss": 0.001, "step": 84190 }, { "epoch": 1.377730508058578, "grad_norm": 0.05815744027495384, "learning_rate": 8.287762238506636e-06, "loss": 0.0021, "step": 84200 }, { "epoch": 1.377894134009654, "grad_norm": 0.015506272204220295, "learning_rate": 8.287224306615357e-06, "loss": 0.0016, "step": 84210 }, { "epoch": 1.3780577599607298, "grad_norm": 0.020869407802820206, "learning_rate": 8.286686307700371e-06, "loss": 0.003, "step": 84220 }, { "epoch": 1.3782213859118055, "grad_norm": 0.06392698734998703, "learning_rate": 8.286148241772648e-06, "loss": 0.0012, "step": 84230 }, { "epoch": 1.3783850118628815, "grad_norm": 0.03342386335134506, "learning_rate": 8.285610108843156e-06, "loss": 0.0011, "step": 84240 }, { "epoch": 1.3785486378139573, "grad_norm": 0.08799833059310913, "learning_rate": 8.285071908922867e-06, "loss": 0.001, "step": 84250 }, { "epoch": 1.378712263765033, "grad_norm": 0.1244782879948616, "learning_rate": 8.284533642022756e-06, "loss": 0.0016, "step": 84260 }, { "epoch": 1.378875889716109, "grad_norm": 0.08687452971935272, "learning_rate": 8.283995308153798e-06, "loss": 0.0009, "step": 84270 }, { "epoch": 1.3790395156671849, "grad_norm": 0.03778639808297157, "learning_rate": 8.28345690732697e-06, "loss": 0.0019, "step": 84280 }, { "epoch": 1.3792031416182606, "grad_norm": 0.10436270385980606, "learning_rate": 8.282918439553244e-06, "loss": 0.0014, "step": 84290 }, { "epoch": 1.3793667675693366, "grad_norm": 0.06285719573497772, "learning_rate": 8.282379904843606e-06, "loss": 0.0022, "step": 84300 }, { "epoch": 1.3795303935204124, "grad_norm": 0.03959321603178978, "learning_rate": 8.281841303209035e-06, "loss": 0.0023, "step": 84310 }, { "epoch": 1.3796940194714882, "grad_norm": 0.12161765247583389, "learning_rate": 8.281302634660509e-06, "loss": 0.0019, "step": 84320 }, { "epoch": 1.379857645422564, "grad_norm": 0.05895822122693062, "learning_rate": 8.280763899209012e-06, "loss": 0.0014, "step": 84330 }, { "epoch": 1.3800212713736397, "grad_norm": 0.13658924400806427, "learning_rate": 8.28022509686553e-06, "loss": 0.0023, "step": 84340 }, { "epoch": 1.3801848973247157, "grad_norm": 0.12981297075748444, "learning_rate": 8.279686227641048e-06, "loss": 0.0029, "step": 84350 }, { "epoch": 1.3803485232757915, "grad_norm": 0.021820586174726486, "learning_rate": 8.279147291546554e-06, "loss": 0.0015, "step": 84360 }, { "epoch": 1.3805121492268673, "grad_norm": 0.0155732911080122, "learning_rate": 8.278608288593033e-06, "loss": 0.0012, "step": 84370 }, { "epoch": 1.3806757751779433, "grad_norm": 0.0901559442281723, "learning_rate": 8.27806921879148e-06, "loss": 0.0021, "step": 84380 }, { "epoch": 1.380839401129019, "grad_norm": 0.08389588445425034, "learning_rate": 8.277530082152882e-06, "loss": 0.0016, "step": 84390 }, { "epoch": 1.3810030270800948, "grad_norm": 0.06358399987220764, "learning_rate": 8.276990878688235e-06, "loss": 0.0016, "step": 84400 }, { "epoch": 1.3811666530311708, "grad_norm": 0.06967510282993317, "learning_rate": 8.276451608408528e-06, "loss": 0.0018, "step": 84410 }, { "epoch": 1.3813302789822466, "grad_norm": 0.08703718334436417, "learning_rate": 8.275912271324763e-06, "loss": 0.0028, "step": 84420 }, { "epoch": 1.3814939049333224, "grad_norm": 0.08442842215299606, "learning_rate": 8.275372867447931e-06, "loss": 0.0056, "step": 84430 }, { "epoch": 1.3816575308843984, "grad_norm": 0.04262305051088333, "learning_rate": 8.274833396789031e-06, "loss": 0.0024, "step": 84440 }, { "epoch": 1.3818211568354741, "grad_norm": 0.05877410247921944, "learning_rate": 8.274293859359064e-06, "loss": 0.0022, "step": 84450 }, { "epoch": 1.38198478278655, "grad_norm": 0.0587134025990963, "learning_rate": 8.27375425516903e-06, "loss": 0.0017, "step": 84460 }, { "epoch": 1.382148408737626, "grad_norm": 0.04951620474457741, "learning_rate": 8.27321458422993e-06, "loss": 0.0029, "step": 84470 }, { "epoch": 1.3823120346887017, "grad_norm": 0.05085930973291397, "learning_rate": 8.27267484655277e-06, "loss": 0.0011, "step": 84480 }, { "epoch": 1.3824756606397774, "grad_norm": 0.007250741124153137, "learning_rate": 8.272135042148553e-06, "loss": 0.0017, "step": 84490 }, { "epoch": 1.3826392865908534, "grad_norm": 0.05588969588279724, "learning_rate": 8.271595171028283e-06, "loss": 0.0013, "step": 84500 }, { "epoch": 1.3828029125419292, "grad_norm": 0.08957484364509583, "learning_rate": 8.271055233202971e-06, "loss": 0.0016, "step": 84510 }, { "epoch": 1.382966538493005, "grad_norm": 0.040992312133312225, "learning_rate": 8.270515228683626e-06, "loss": 0.0016, "step": 84520 }, { "epoch": 1.3831301644440808, "grad_norm": 0.10888328403234482, "learning_rate": 8.269975157481255e-06, "loss": 0.0015, "step": 84530 }, { "epoch": 1.3832937903951565, "grad_norm": 0.04706072434782982, "learning_rate": 8.269435019606875e-06, "loss": 0.0023, "step": 84540 }, { "epoch": 1.3834574163462325, "grad_norm": 0.041850849986076355, "learning_rate": 8.268894815071492e-06, "loss": 0.0024, "step": 84550 }, { "epoch": 1.3836210422973083, "grad_norm": 0.30448901653289795, "learning_rate": 8.268354543886126e-06, "loss": 0.0018, "step": 84560 }, { "epoch": 1.383784668248384, "grad_norm": 0.14749623835086823, "learning_rate": 8.267814206061788e-06, "loss": 0.0017, "step": 84570 }, { "epoch": 1.38394829419946, "grad_norm": 0.011189352720975876, "learning_rate": 8.2672738016095e-06, "loss": 0.001, "step": 84580 }, { "epoch": 1.3841119201505359, "grad_norm": 0.011571753770112991, "learning_rate": 8.266733330540278e-06, "loss": 0.0011, "step": 84590 }, { "epoch": 1.3842755461016116, "grad_norm": 0.1206396296620369, "learning_rate": 8.26619279286514e-06, "loss": 0.0012, "step": 84600 }, { "epoch": 1.3844391720526876, "grad_norm": 0.08492659777402878, "learning_rate": 8.265652188595108e-06, "loss": 0.0015, "step": 84610 }, { "epoch": 1.3846027980037634, "grad_norm": 0.00883982703089714, "learning_rate": 8.265111517741206e-06, "loss": 0.0018, "step": 84620 }, { "epoch": 1.3847664239548392, "grad_norm": 0.021983684971928596, "learning_rate": 8.264570780314458e-06, "loss": 0.002, "step": 84630 }, { "epoch": 1.3849300499059152, "grad_norm": 0.050639525055885315, "learning_rate": 8.26402997632589e-06, "loss": 0.0017, "step": 84640 }, { "epoch": 1.385093675856991, "grad_norm": 0.062431205064058304, "learning_rate": 8.263489105786524e-06, "loss": 0.0028, "step": 84650 }, { "epoch": 1.3852573018080667, "grad_norm": 0.022109996527433395, "learning_rate": 8.26294816870739e-06, "loss": 0.0014, "step": 84660 }, { "epoch": 1.3854209277591427, "grad_norm": 0.3256184458732605, "learning_rate": 8.26240716509952e-06, "loss": 0.0014, "step": 84670 }, { "epoch": 1.3855845537102185, "grad_norm": 0.10933385789394379, "learning_rate": 8.26186609497394e-06, "loss": 0.0021, "step": 84680 }, { "epoch": 1.3857481796612943, "grad_norm": 0.03530566766858101, "learning_rate": 8.261324958341686e-06, "loss": 0.0013, "step": 84690 }, { "epoch": 1.38591180561237, "grad_norm": 0.04505179449915886, "learning_rate": 8.26078375521379e-06, "loss": 0.0085, "step": 84700 }, { "epoch": 1.386075431563446, "grad_norm": 0.043144263327121735, "learning_rate": 8.260242485601286e-06, "loss": 0.0024, "step": 84710 }, { "epoch": 1.3862390575145218, "grad_norm": 0.3117769658565521, "learning_rate": 8.259701149515211e-06, "loss": 0.0022, "step": 84720 }, { "epoch": 1.3864026834655976, "grad_norm": 0.2254185974597931, "learning_rate": 8.259159746966601e-06, "loss": 0.0022, "step": 84730 }, { "epoch": 1.3865663094166734, "grad_norm": 0.0006562051712535322, "learning_rate": 8.258618277966498e-06, "loss": 0.0012, "step": 84740 }, { "epoch": 1.3867299353677494, "grad_norm": 0.0367521196603775, "learning_rate": 8.258076742525935e-06, "loss": 0.0019, "step": 84750 }, { "epoch": 1.3868935613188251, "grad_norm": 0.0878819078207016, "learning_rate": 8.257535140655961e-06, "loss": 0.0014, "step": 84760 }, { "epoch": 1.387057187269901, "grad_norm": 0.07470402866601944, "learning_rate": 8.256993472367616e-06, "loss": 0.0013, "step": 84770 }, { "epoch": 1.387220813220977, "grad_norm": 0.12001055479049683, "learning_rate": 8.256451737671945e-06, "loss": 0.0026, "step": 84780 }, { "epoch": 1.3873844391720527, "grad_norm": 0.04070049151778221, "learning_rate": 8.255909936579991e-06, "loss": 0.0013, "step": 84790 }, { "epoch": 1.3875480651231284, "grad_norm": 0.1716008186340332, "learning_rate": 8.255368069102803e-06, "loss": 0.0018, "step": 84800 }, { "epoch": 1.3877116910742044, "grad_norm": 0.06332981586456299, "learning_rate": 8.254826135251428e-06, "loss": 0.0011, "step": 84810 }, { "epoch": 1.3878753170252802, "grad_norm": 0.05287964269518852, "learning_rate": 8.254284135036916e-06, "loss": 0.002, "step": 84820 }, { "epoch": 1.388038942976356, "grad_norm": 0.011061828583478928, "learning_rate": 8.25374206847032e-06, "loss": 0.0011, "step": 84830 }, { "epoch": 1.388202568927432, "grad_norm": 0.001770665985532105, "learning_rate": 8.25319993556269e-06, "loss": 0.003, "step": 84840 }, { "epoch": 1.3883661948785078, "grad_norm": 0.014196856878697872, "learning_rate": 8.252657736325077e-06, "loss": 0.001, "step": 84850 }, { "epoch": 1.3885298208295835, "grad_norm": 0.008903002366423607, "learning_rate": 8.252115470768543e-06, "loss": 0.0012, "step": 84860 }, { "epoch": 1.3886934467806595, "grad_norm": 0.05734460428357124, "learning_rate": 8.251573138904139e-06, "loss": 0.0038, "step": 84870 }, { "epoch": 1.3888570727317353, "grad_norm": 0.2454068958759308, "learning_rate": 8.251030740742923e-06, "loss": 0.0011, "step": 84880 }, { "epoch": 1.389020698682811, "grad_norm": 0.01381621789187193, "learning_rate": 8.250488276295956e-06, "loss": 0.0027, "step": 84890 }, { "epoch": 1.3891843246338869, "grad_norm": 0.06249716505408287, "learning_rate": 8.249945745574297e-06, "loss": 0.0012, "step": 84900 }, { "epoch": 1.3893479505849629, "grad_norm": 0.05240399390459061, "learning_rate": 8.249403148589009e-06, "loss": 0.0014, "step": 84910 }, { "epoch": 1.3895115765360386, "grad_norm": 0.04492214322090149, "learning_rate": 8.248860485351155e-06, "loss": 0.0011, "step": 84920 }, { "epoch": 1.3896752024871144, "grad_norm": 0.13296620547771454, "learning_rate": 8.248317755871797e-06, "loss": 0.0015, "step": 84930 }, { "epoch": 1.3898388284381902, "grad_norm": 0.03895588591694832, "learning_rate": 8.247774960162003e-06, "loss": 0.001, "step": 84940 }, { "epoch": 1.3900024543892662, "grad_norm": 0.04336373507976532, "learning_rate": 8.24723209823284e-06, "loss": 0.002, "step": 84950 }, { "epoch": 1.390166080340342, "grad_norm": 0.05046654865145683, "learning_rate": 8.246689170095376e-06, "loss": 0.0015, "step": 84960 }, { "epoch": 1.3903297062914177, "grad_norm": 0.033213019371032715, "learning_rate": 8.246146175760681e-06, "loss": 0.0014, "step": 84970 }, { "epoch": 1.3904933322424937, "grad_norm": 0.12314862012863159, "learning_rate": 8.245603115239826e-06, "loss": 0.0019, "step": 84980 }, { "epoch": 1.3906569581935695, "grad_norm": 0.836834192276001, "learning_rate": 8.245059988543885e-06, "loss": 0.0023, "step": 84990 }, { "epoch": 1.3908205841446453, "grad_norm": 0.027123888954520226, "learning_rate": 8.24451679568393e-06, "loss": 0.0015, "step": 85000 }, { "epoch": 1.3909842100957213, "grad_norm": 0.012415563687682152, "learning_rate": 8.243973536671036e-06, "loss": 0.0025, "step": 85010 }, { "epoch": 1.391147836046797, "grad_norm": 0.014098512940108776, "learning_rate": 8.243430211516282e-06, "loss": 0.0015, "step": 85020 }, { "epoch": 1.3913114619978728, "grad_norm": 0.1654808670282364, "learning_rate": 8.242886820230745e-06, "loss": 0.0013, "step": 85030 }, { "epoch": 1.3914750879489488, "grad_norm": 0.11375343799591064, "learning_rate": 8.2423433628255e-06, "loss": 0.0027, "step": 85040 }, { "epoch": 1.3916387139000246, "grad_norm": 0.04883484169840813, "learning_rate": 8.241799839311636e-06, "loss": 0.0013, "step": 85050 }, { "epoch": 1.3918023398511004, "grad_norm": 0.027821864932775497, "learning_rate": 8.241256249700227e-06, "loss": 0.0017, "step": 85060 }, { "epoch": 1.3919659658021764, "grad_norm": 0.025622084736824036, "learning_rate": 8.240712594002361e-06, "loss": 0.0011, "step": 85070 }, { "epoch": 1.3921295917532521, "grad_norm": 0.0412280298769474, "learning_rate": 8.24016887222912e-06, "loss": 0.0012, "step": 85080 }, { "epoch": 1.392293217704328, "grad_norm": 0.10488954931497574, "learning_rate": 8.239625084391593e-06, "loss": 0.0018, "step": 85090 }, { "epoch": 1.3924568436554037, "grad_norm": 0.06807077676057816, "learning_rate": 8.239081230500865e-06, "loss": 0.0014, "step": 85100 }, { "epoch": 1.3926204696064794, "grad_norm": 0.12468797713518143, "learning_rate": 8.238537310568027e-06, "loss": 0.0016, "step": 85110 }, { "epoch": 1.3927840955575554, "grad_norm": 0.04593636468052864, "learning_rate": 8.237993324604166e-06, "loss": 0.0011, "step": 85120 }, { "epoch": 1.3929477215086312, "grad_norm": 0.09028539806604385, "learning_rate": 8.237449272620376e-06, "loss": 0.0023, "step": 85130 }, { "epoch": 1.393111347459707, "grad_norm": 0.038293156772851944, "learning_rate": 8.236905154627746e-06, "loss": 0.0011, "step": 85140 }, { "epoch": 1.393274973410783, "grad_norm": 0.2811160683631897, "learning_rate": 8.236360970637374e-06, "loss": 0.0017, "step": 85150 }, { "epoch": 1.3934385993618588, "grad_norm": 0.11244338750839233, "learning_rate": 8.235816720660355e-06, "loss": 0.0017, "step": 85160 }, { "epoch": 1.3936022253129345, "grad_norm": 0.004961560945957899, "learning_rate": 8.235272404707784e-06, "loss": 0.0008, "step": 85170 }, { "epoch": 1.3937658512640105, "grad_norm": 0.0400683619081974, "learning_rate": 8.23472802279076e-06, "loss": 0.0017, "step": 85180 }, { "epoch": 1.3939294772150863, "grad_norm": 0.010890142060816288, "learning_rate": 8.234183574920384e-06, "loss": 0.0019, "step": 85190 }, { "epoch": 1.394093103166162, "grad_norm": 0.04339621216058731, "learning_rate": 8.233639061107757e-06, "loss": 0.0018, "step": 85200 }, { "epoch": 1.394256729117238, "grad_norm": 0.0620986633002758, "learning_rate": 8.233094481363977e-06, "loss": 0.0014, "step": 85210 }, { "epoch": 1.3944203550683139, "grad_norm": 0.06438495963811874, "learning_rate": 8.23254983570015e-06, "loss": 0.0027, "step": 85220 }, { "epoch": 1.3945839810193896, "grad_norm": 0.06436100602149963, "learning_rate": 8.232005124127382e-06, "loss": 0.0016, "step": 85230 }, { "epoch": 1.3947476069704656, "grad_norm": 0.13490046560764313, "learning_rate": 8.231460346656778e-06, "loss": 0.001, "step": 85240 }, { "epoch": 1.3949112329215414, "grad_norm": 0.02479497902095318, "learning_rate": 8.230915503299446e-06, "loss": 0.0013, "step": 85250 }, { "epoch": 1.3950748588726172, "grad_norm": 0.06576614081859589, "learning_rate": 8.230370594066494e-06, "loss": 0.0009, "step": 85260 }, { "epoch": 1.3952384848236932, "grad_norm": 0.05531184375286102, "learning_rate": 8.229825618969035e-06, "loss": 0.0011, "step": 85270 }, { "epoch": 1.395402110774769, "grad_norm": 0.00900990143418312, "learning_rate": 8.229280578018178e-06, "loss": 0.0019, "step": 85280 }, { "epoch": 1.3955657367258447, "grad_norm": 0.011500924825668335, "learning_rate": 8.228735471225034e-06, "loss": 0.0016, "step": 85290 }, { "epoch": 1.3957293626769205, "grad_norm": 0.0016764284810051322, "learning_rate": 8.228190298600723e-06, "loss": 0.0019, "step": 85300 }, { "epoch": 1.3958929886279963, "grad_norm": 0.08344411104917526, "learning_rate": 8.227645060156353e-06, "loss": 0.0015, "step": 85310 }, { "epoch": 1.3960566145790723, "grad_norm": 0.014452359639108181, "learning_rate": 8.22709975590305e-06, "loss": 0.0012, "step": 85320 }, { "epoch": 1.396220240530148, "grad_norm": 0.062364593148231506, "learning_rate": 8.226554385851923e-06, "loss": 0.0019, "step": 85330 }, { "epoch": 1.3963838664812238, "grad_norm": 0.022791588678956032, "learning_rate": 8.226008950014099e-06, "loss": 0.001, "step": 85340 }, { "epoch": 1.3965474924322998, "grad_norm": 0.02725742571055889, "learning_rate": 8.225463448400696e-06, "loss": 0.0014, "step": 85350 }, { "epoch": 1.3967111183833756, "grad_norm": 0.0720963105559349, "learning_rate": 8.224917881022836e-06, "loss": 0.0018, "step": 85360 }, { "epoch": 1.3968747443344514, "grad_norm": 0.056665487587451935, "learning_rate": 8.224372247891643e-06, "loss": 0.0042, "step": 85370 }, { "epoch": 1.3970383702855274, "grad_norm": 0.040643978863954544, "learning_rate": 8.22382654901824e-06, "loss": 0.0014, "step": 85380 }, { "epoch": 1.3972019962366031, "grad_norm": 0.04232388734817505, "learning_rate": 8.223280784413758e-06, "loss": 0.0012, "step": 85390 }, { "epoch": 1.397365622187679, "grad_norm": 0.04045966640114784, "learning_rate": 8.222734954089323e-06, "loss": 0.0013, "step": 85400 }, { "epoch": 1.397529248138755, "grad_norm": 0.07560890167951584, "learning_rate": 8.222189058056062e-06, "loss": 0.0008, "step": 85410 }, { "epoch": 1.3976928740898307, "grad_norm": 0.07168114930391312, "learning_rate": 8.221643096325105e-06, "loss": 0.0015, "step": 85420 }, { "epoch": 1.3978565000409064, "grad_norm": 0.03171724081039429, "learning_rate": 8.221097068907589e-06, "loss": 0.001, "step": 85430 }, { "epoch": 1.3980201259919824, "grad_norm": 0.06926529109477997, "learning_rate": 8.22055097581464e-06, "loss": 0.0011, "step": 85440 }, { "epoch": 1.3981837519430582, "grad_norm": 0.1805880069732666, "learning_rate": 8.220004817057397e-06, "loss": 0.0024, "step": 85450 }, { "epoch": 1.398347377894134, "grad_norm": 0.03902384638786316, "learning_rate": 8.219458592646991e-06, "loss": 0.0018, "step": 85460 }, { "epoch": 1.39851100384521, "grad_norm": 0.0611749105155468, "learning_rate": 8.218912302594566e-06, "loss": 0.001, "step": 85470 }, { "epoch": 1.3986746297962858, "grad_norm": 0.0291975736618042, "learning_rate": 8.218365946911256e-06, "loss": 0.0015, "step": 85480 }, { "epoch": 1.3988382557473615, "grad_norm": 0.07924709469079971, "learning_rate": 8.217819525608201e-06, "loss": 0.0025, "step": 85490 }, { "epoch": 1.3990018816984373, "grad_norm": 0.15075138211250305, "learning_rate": 8.217273038696542e-06, "loss": 0.0017, "step": 85500 }, { "epoch": 1.399165507649513, "grad_norm": 0.029555534943938255, "learning_rate": 8.216726486187423e-06, "loss": 0.0014, "step": 85510 }, { "epoch": 1.399329133600589, "grad_norm": 0.15720826387405396, "learning_rate": 8.216179868091987e-06, "loss": 0.0017, "step": 85520 }, { "epoch": 1.3994927595516649, "grad_norm": 0.043863147497177124, "learning_rate": 8.215633184421377e-06, "loss": 0.0016, "step": 85530 }, { "epoch": 1.3996563855027406, "grad_norm": 0.2264384776353836, "learning_rate": 8.215086435186743e-06, "loss": 0.0026, "step": 85540 }, { "epoch": 1.3998200114538166, "grad_norm": 0.034213047474622726, "learning_rate": 8.214539620399229e-06, "loss": 0.0014, "step": 85550 }, { "epoch": 1.3999836374048924, "grad_norm": 0.019672662019729614, "learning_rate": 8.213992740069987e-06, "loss": 0.0011, "step": 85560 }, { "epoch": 1.4001472633559682, "grad_norm": 0.02174626849591732, "learning_rate": 8.213445794210167e-06, "loss": 0.0014, "step": 85570 }, { "epoch": 1.4003108893070442, "grad_norm": 0.07954105734825134, "learning_rate": 8.21289878283092e-06, "loss": 0.0017, "step": 85580 }, { "epoch": 1.40047451525812, "grad_norm": 0.08840786665678024, "learning_rate": 8.212351705943398e-06, "loss": 0.0016, "step": 85590 }, { "epoch": 1.4006381412091957, "grad_norm": 0.06759927421808243, "learning_rate": 8.21180456355876e-06, "loss": 0.0028, "step": 85600 }, { "epoch": 1.4008017671602717, "grad_norm": 0.04595992714166641, "learning_rate": 8.211257355688155e-06, "loss": 0.0007, "step": 85610 }, { "epoch": 1.4009653931113475, "grad_norm": 0.11282442510128021, "learning_rate": 8.210710082342744e-06, "loss": 0.0018, "step": 85620 }, { "epoch": 1.4011290190624233, "grad_norm": 0.08033892512321472, "learning_rate": 8.210162743533686e-06, "loss": 0.0015, "step": 85630 }, { "epoch": 1.4012926450134993, "grad_norm": 0.013663397170603275, "learning_rate": 8.20961533927214e-06, "loss": 0.0013, "step": 85640 }, { "epoch": 1.401456270964575, "grad_norm": 0.10717205703258514, "learning_rate": 8.209067869569267e-06, "loss": 0.0019, "step": 85650 }, { "epoch": 1.4016198969156508, "grad_norm": 0.021849509328603745, "learning_rate": 8.208520334436228e-06, "loss": 0.001, "step": 85660 }, { "epoch": 1.4017835228667266, "grad_norm": 0.11894243210554123, "learning_rate": 8.207972733884189e-06, "loss": 0.0017, "step": 85670 }, { "epoch": 1.4019471488178026, "grad_norm": 0.0320327989757061, "learning_rate": 8.207425067924314e-06, "loss": 0.0015, "step": 85680 }, { "epoch": 1.4021107747688784, "grad_norm": 0.09802786260843277, "learning_rate": 8.20687733656777e-06, "loss": 0.0045, "step": 85690 }, { "epoch": 1.4022744007199541, "grad_norm": 0.016743306070566177, "learning_rate": 8.206329539825724e-06, "loss": 0.0014, "step": 85700 }, { "epoch": 1.40243802667103, "grad_norm": 0.1568026840686798, "learning_rate": 8.205781677709347e-06, "loss": 0.0038, "step": 85710 }, { "epoch": 1.402601652622106, "grad_norm": 0.06770619004964828, "learning_rate": 8.205233750229806e-06, "loss": 0.0024, "step": 85720 }, { "epoch": 1.4027652785731817, "grad_norm": 0.021262871101498604, "learning_rate": 8.204685757398276e-06, "loss": 0.0017, "step": 85730 }, { "epoch": 1.4029289045242574, "grad_norm": 0.025582343339920044, "learning_rate": 8.20413769922593e-06, "loss": 0.0015, "step": 85740 }, { "epoch": 1.4030925304753334, "grad_norm": 0.05053820088505745, "learning_rate": 8.203589575723941e-06, "loss": 0.0012, "step": 85750 }, { "epoch": 1.4032561564264092, "grad_norm": 0.08275462687015533, "learning_rate": 8.203041386903483e-06, "loss": 0.0019, "step": 85760 }, { "epoch": 1.403419782377485, "grad_norm": 0.06468658149242401, "learning_rate": 8.202493132775736e-06, "loss": 0.0013, "step": 85770 }, { "epoch": 1.403583408328561, "grad_norm": 0.0523967407643795, "learning_rate": 8.201944813351879e-06, "loss": 0.0023, "step": 85780 }, { "epoch": 1.4037470342796368, "grad_norm": 0.17889747023582458, "learning_rate": 8.20139642864309e-06, "loss": 0.0022, "step": 85790 }, { "epoch": 1.4039106602307125, "grad_norm": 0.11283553391695023, "learning_rate": 8.200847978660549e-06, "loss": 0.0013, "step": 85800 }, { "epoch": 1.4040742861817885, "grad_norm": 0.12711483240127563, "learning_rate": 8.20029946341544e-06, "loss": 0.0035, "step": 85810 }, { "epoch": 1.4042379121328643, "grad_norm": 0.04611670970916748, "learning_rate": 8.199750882918947e-06, "loss": 0.002, "step": 85820 }, { "epoch": 1.40440153808394, "grad_norm": 0.02830475941300392, "learning_rate": 8.199202237182256e-06, "loss": 0.0012, "step": 85830 }, { "epoch": 1.404565164035016, "grad_norm": 0.09817026555538177, "learning_rate": 8.198653526216552e-06, "loss": 0.0015, "step": 85840 }, { "epoch": 1.4047287899860919, "grad_norm": 0.26381629705429077, "learning_rate": 8.198104750033022e-06, "loss": 0.0025, "step": 85850 }, { "epoch": 1.4048924159371676, "grad_norm": 0.016162671148777008, "learning_rate": 8.197555908642857e-06, "loss": 0.0014, "step": 85860 }, { "epoch": 1.4050560418882434, "grad_norm": 0.11308854818344116, "learning_rate": 8.197007002057245e-06, "loss": 0.0026, "step": 85870 }, { "epoch": 1.4052196678393194, "grad_norm": 0.07511643320322037, "learning_rate": 8.196458030287381e-06, "loss": 0.002, "step": 85880 }, { "epoch": 1.4053832937903952, "grad_norm": 0.0028694872744381428, "learning_rate": 8.195908993344454e-06, "loss": 0.0012, "step": 85890 }, { "epoch": 1.405546919741471, "grad_norm": 0.045142270624637604, "learning_rate": 8.195359891239662e-06, "loss": 0.0021, "step": 85900 }, { "epoch": 1.4057105456925467, "grad_norm": 0.037996213883161545, "learning_rate": 8.1948107239842e-06, "loss": 0.0015, "step": 85910 }, { "epoch": 1.4058741716436227, "grad_norm": 0.04562247917056084, "learning_rate": 8.194261491589265e-06, "loss": 0.0012, "step": 85920 }, { "epoch": 1.4060377975946985, "grad_norm": 0.056615930050611496, "learning_rate": 8.193712194066052e-06, "loss": 0.0014, "step": 85930 }, { "epoch": 1.4062014235457743, "grad_norm": 0.015473724342882633, "learning_rate": 8.193162831425766e-06, "loss": 0.0016, "step": 85940 }, { "epoch": 1.4063650494968503, "grad_norm": 0.07412919402122498, "learning_rate": 8.192613403679606e-06, "loss": 0.0013, "step": 85950 }, { "epoch": 1.406528675447926, "grad_norm": 0.04837470129132271, "learning_rate": 8.192063910838771e-06, "loss": 0.0025, "step": 85960 }, { "epoch": 1.4066923013990018, "grad_norm": 0.03220709413290024, "learning_rate": 8.19151435291447e-06, "loss": 0.002, "step": 85970 }, { "epoch": 1.4068559273500778, "grad_norm": 0.05156199634075165, "learning_rate": 8.190964729917904e-06, "loss": 0.0036, "step": 85980 }, { "epoch": 1.4070195533011536, "grad_norm": 0.08160155266523361, "learning_rate": 8.190415041860282e-06, "loss": 0.0016, "step": 85990 }, { "epoch": 1.4071831792522294, "grad_norm": 0.044558871537446976, "learning_rate": 8.189865288752812e-06, "loss": 0.0025, "step": 86000 }, { "epoch": 1.4073468052033054, "grad_norm": 0.06467525660991669, "learning_rate": 8.1893154706067e-06, "loss": 0.0019, "step": 86010 }, { "epoch": 1.4075104311543811, "grad_norm": 0.06674395501613617, "learning_rate": 8.188765587433157e-06, "loss": 0.0014, "step": 86020 }, { "epoch": 1.407674057105457, "grad_norm": 0.08188920468091965, "learning_rate": 8.188215639243397e-06, "loss": 0.0022, "step": 86030 }, { "epoch": 1.407837683056533, "grad_norm": 0.11206168681383133, "learning_rate": 8.187665626048632e-06, "loss": 0.0021, "step": 86040 }, { "epoch": 1.4080013090076087, "grad_norm": 0.14717446267604828, "learning_rate": 8.187115547860074e-06, "loss": 0.0038, "step": 86050 }, { "epoch": 1.4081649349586844, "grad_norm": 0.0455758199095726, "learning_rate": 8.186565404688942e-06, "loss": 0.0017, "step": 86060 }, { "epoch": 1.4083285609097602, "grad_norm": 0.0021257689222693443, "learning_rate": 8.186015196546454e-06, "loss": 0.0015, "step": 86070 }, { "epoch": 1.408492186860836, "grad_norm": 0.026790112257003784, "learning_rate": 8.185464923443823e-06, "loss": 0.001, "step": 86080 }, { "epoch": 1.408655812811912, "grad_norm": 0.02806266024708748, "learning_rate": 8.18491458539227e-06, "loss": 0.0011, "step": 86090 }, { "epoch": 1.4088194387629878, "grad_norm": 0.08007021248340607, "learning_rate": 8.184364182403018e-06, "loss": 0.0011, "step": 86100 }, { "epoch": 1.4089830647140635, "grad_norm": 0.06965044885873795, "learning_rate": 8.18381371448729e-06, "loss": 0.0016, "step": 86110 }, { "epoch": 1.4091466906651395, "grad_norm": 0.033687982708215714, "learning_rate": 8.183263181656308e-06, "loss": 0.0014, "step": 86120 }, { "epoch": 1.4093103166162153, "grad_norm": 0.054546255618333817, "learning_rate": 8.182712583921295e-06, "loss": 0.0021, "step": 86130 }, { "epoch": 1.409473942567291, "grad_norm": 0.012294416315853596, "learning_rate": 8.18216192129348e-06, "loss": 0.002, "step": 86140 }, { "epoch": 1.409637568518367, "grad_norm": 0.09605805575847626, "learning_rate": 8.18161119378409e-06, "loss": 0.0018, "step": 86150 }, { "epoch": 1.4098011944694429, "grad_norm": 0.181951105594635, "learning_rate": 8.181060401404354e-06, "loss": 0.002, "step": 86160 }, { "epoch": 1.4099648204205186, "grad_norm": 0.051558397710323334, "learning_rate": 8.180509544165502e-06, "loss": 0.0015, "step": 86170 }, { "epoch": 1.4101284463715946, "grad_norm": 0.006931046489626169, "learning_rate": 8.179958622078765e-06, "loss": 0.0018, "step": 86180 }, { "epoch": 1.4102920723226704, "grad_norm": 0.04546690359711647, "learning_rate": 8.179407635155376e-06, "loss": 0.0012, "step": 86190 }, { "epoch": 1.4104556982737462, "grad_norm": 0.039440467953681946, "learning_rate": 8.178856583406569e-06, "loss": 0.0009, "step": 86200 }, { "epoch": 1.4106193242248222, "grad_norm": 0.22247092425823212, "learning_rate": 8.17830546684358e-06, "loss": 0.0032, "step": 86210 }, { "epoch": 1.410782950175898, "grad_norm": 0.11549188196659088, "learning_rate": 8.177754285477646e-06, "loss": 0.0015, "step": 86220 }, { "epoch": 1.4109465761269737, "grad_norm": 0.07344572991132736, "learning_rate": 8.177203039320005e-06, "loss": 0.0023, "step": 86230 }, { "epoch": 1.4111102020780497, "grad_norm": 0.008797932416200638, "learning_rate": 8.176651728381895e-06, "loss": 0.0019, "step": 86240 }, { "epoch": 1.4112738280291255, "grad_norm": 0.05913636460900307, "learning_rate": 8.17610035267456e-06, "loss": 0.0012, "step": 86250 }, { "epoch": 1.4114374539802013, "grad_norm": 0.06865714490413666, "learning_rate": 8.175548912209239e-06, "loss": 0.0023, "step": 86260 }, { "epoch": 1.411601079931277, "grad_norm": 0.023853128775954247, "learning_rate": 8.174997406997175e-06, "loss": 0.0015, "step": 86270 }, { "epoch": 1.4117647058823528, "grad_norm": 0.051786597818136215, "learning_rate": 8.174445837049614e-06, "loss": 0.0012, "step": 86280 }, { "epoch": 1.4119283318334288, "grad_norm": 0.05271969735622406, "learning_rate": 8.173894202377803e-06, "loss": 0.0019, "step": 86290 }, { "epoch": 1.4120919577845046, "grad_norm": 0.03360547497868538, "learning_rate": 8.173342502992989e-06, "loss": 0.0017, "step": 86300 }, { "epoch": 1.4122555837355804, "grad_norm": 0.06189177930355072, "learning_rate": 8.17279073890642e-06, "loss": 0.0023, "step": 86310 }, { "epoch": 1.4124192096866564, "grad_norm": 0.0432027168571949, "learning_rate": 8.172238910129346e-06, "loss": 0.0017, "step": 86320 }, { "epoch": 1.4125828356377321, "grad_norm": 0.059149377048015594, "learning_rate": 8.17168701667302e-06, "loss": 0.0014, "step": 86330 }, { "epoch": 1.412746461588808, "grad_norm": 0.07163205742835999, "learning_rate": 8.171135058548692e-06, "loss": 0.0018, "step": 86340 }, { "epoch": 1.412910087539884, "grad_norm": 0.01703125238418579, "learning_rate": 8.170583035767616e-06, "loss": 0.005, "step": 86350 }, { "epoch": 1.4130737134909597, "grad_norm": 0.11839171499013901, "learning_rate": 8.170030948341049e-06, "loss": 0.0032, "step": 86360 }, { "epoch": 1.4132373394420354, "grad_norm": 0.018654173240065575, "learning_rate": 8.169478796280248e-06, "loss": 0.0016, "step": 86370 }, { "epoch": 1.4134009653931114, "grad_norm": 0.06727386265993118, "learning_rate": 8.168926579596469e-06, "loss": 0.0019, "step": 86380 }, { "epoch": 1.4135645913441872, "grad_norm": 0.04134442284703255, "learning_rate": 8.168374298300972e-06, "loss": 0.0026, "step": 86390 }, { "epoch": 1.413728217295263, "grad_norm": 0.12582345306873322, "learning_rate": 8.167821952405017e-06, "loss": 0.0013, "step": 86400 }, { "epoch": 1.413891843246339, "grad_norm": 0.07427453249692917, "learning_rate": 8.16726954191987e-06, "loss": 0.0012, "step": 86410 }, { "epoch": 1.4140554691974148, "grad_norm": 0.03545006737112999, "learning_rate": 8.166717066856786e-06, "loss": 0.0019, "step": 86420 }, { "epoch": 1.4142190951484905, "grad_norm": 0.0959739163517952, "learning_rate": 8.166164527227037e-06, "loss": 0.0019, "step": 86430 }, { "epoch": 1.4143827210995663, "grad_norm": 0.04592728614807129, "learning_rate": 8.165611923041884e-06, "loss": 0.0017, "step": 86440 }, { "epoch": 1.4145463470506423, "grad_norm": 0.014627465046942234, "learning_rate": 8.1650592543126e-06, "loss": 0.0035, "step": 86450 }, { "epoch": 1.414709973001718, "grad_norm": 0.020459698513150215, "learning_rate": 8.164506521050446e-06, "loss": 0.001, "step": 86460 }, { "epoch": 1.4148735989527939, "grad_norm": 0.15603390336036682, "learning_rate": 8.163953723266695e-06, "loss": 0.0016, "step": 86470 }, { "epoch": 1.4150372249038696, "grad_norm": 0.08350764214992523, "learning_rate": 8.163400860972621e-06, "loss": 0.002, "step": 86480 }, { "epoch": 1.4152008508549456, "grad_norm": 0.03874802589416504, "learning_rate": 8.162847934179493e-06, "loss": 0.0023, "step": 86490 }, { "epoch": 1.4153644768060214, "grad_norm": 0.05094561353325844, "learning_rate": 8.162294942898586e-06, "loss": 0.0019, "step": 86500 }, { "epoch": 1.4155281027570972, "grad_norm": 0.08197402209043503, "learning_rate": 8.161741887141173e-06, "loss": 0.0013, "step": 86510 }, { "epoch": 1.4156917287081732, "grad_norm": 0.06004476174712181, "learning_rate": 8.161188766918533e-06, "loss": 0.0013, "step": 86520 }, { "epoch": 1.415855354659249, "grad_norm": 0.055103935301303864, "learning_rate": 8.160635582241942e-06, "loss": 0.0024, "step": 86530 }, { "epoch": 1.4160189806103247, "grad_norm": 0.23720720410346985, "learning_rate": 8.160082333122679e-06, "loss": 0.0025, "step": 86540 }, { "epoch": 1.4161826065614007, "grad_norm": 0.019834967330098152, "learning_rate": 8.159529019572026e-06, "loss": 0.0017, "step": 86550 }, { "epoch": 1.4163462325124765, "grad_norm": 0.1081312820315361, "learning_rate": 8.158975641601263e-06, "loss": 0.0034, "step": 86560 }, { "epoch": 1.4165098584635523, "grad_norm": 0.07273285835981369, "learning_rate": 8.158422199221673e-06, "loss": 0.0021, "step": 86570 }, { "epoch": 1.4166734844146283, "grad_norm": 0.07118023931980133, "learning_rate": 8.157868692444541e-06, "loss": 0.0035, "step": 86580 }, { "epoch": 1.416837110365704, "grad_norm": 0.08688318729400635, "learning_rate": 8.15731512128115e-06, "loss": 0.0029, "step": 86590 }, { "epoch": 1.4170007363167798, "grad_norm": 0.02815968357026577, "learning_rate": 8.156761485742791e-06, "loss": 0.0036, "step": 86600 }, { "epoch": 1.4171643622678558, "grad_norm": 0.030659910291433334, "learning_rate": 8.15620778584075e-06, "loss": 0.0012, "step": 86610 }, { "epoch": 1.4173279882189316, "grad_norm": 0.07467029243707657, "learning_rate": 8.155654021586316e-06, "loss": 0.0014, "step": 86620 }, { "epoch": 1.4174916141700074, "grad_norm": 0.013359046541154385, "learning_rate": 8.15510019299078e-06, "loss": 0.0011, "step": 86630 }, { "epoch": 1.4176552401210831, "grad_norm": 0.010714574716985226, "learning_rate": 8.154546300065435e-06, "loss": 0.0019, "step": 86640 }, { "epoch": 1.4178188660721591, "grad_norm": 0.046726588159799576, "learning_rate": 8.153992342821574e-06, "loss": 0.0017, "step": 86650 }, { "epoch": 1.417982492023235, "grad_norm": 0.11305667459964752, "learning_rate": 8.15343832127049e-06, "loss": 0.0019, "step": 86660 }, { "epoch": 1.4181461179743107, "grad_norm": 0.31488731503486633, "learning_rate": 8.152884235423481e-06, "loss": 0.0018, "step": 86670 }, { "epoch": 1.4183097439253864, "grad_norm": 0.04014440253376961, "learning_rate": 8.152330085291846e-06, "loss": 0.0017, "step": 86680 }, { "epoch": 1.4184733698764624, "grad_norm": 0.11816905438899994, "learning_rate": 8.15177587088688e-06, "loss": 0.0035, "step": 86690 }, { "epoch": 1.4186369958275382, "grad_norm": 0.04124940186738968, "learning_rate": 8.151221592219885e-06, "loss": 0.0013, "step": 86700 }, { "epoch": 1.418800621778614, "grad_norm": 0.04281597211956978, "learning_rate": 8.15066724930216e-06, "loss": 0.0023, "step": 86710 }, { "epoch": 1.41896424772969, "grad_norm": 0.01586921513080597, "learning_rate": 8.150112842145011e-06, "loss": 0.002, "step": 86720 }, { "epoch": 1.4191278736807658, "grad_norm": 0.037790652364492416, "learning_rate": 8.14955837075974e-06, "loss": 0.0022, "step": 86730 }, { "epoch": 1.4192914996318415, "grad_norm": 0.01918884366750717, "learning_rate": 8.149003835157651e-06, "loss": 0.0012, "step": 86740 }, { "epoch": 1.4194551255829175, "grad_norm": 0.0880705937743187, "learning_rate": 8.148449235350055e-06, "loss": 0.0025, "step": 86750 }, { "epoch": 1.4196187515339933, "grad_norm": 0.33318060636520386, "learning_rate": 8.147894571348254e-06, "loss": 0.0021, "step": 86760 }, { "epoch": 1.419782377485069, "grad_norm": 0.10535962134599686, "learning_rate": 8.147339843163561e-06, "loss": 0.0013, "step": 86770 }, { "epoch": 1.419946003436145, "grad_norm": 0.07826575636863708, "learning_rate": 8.146785050807285e-06, "loss": 0.0014, "step": 86780 }, { "epoch": 1.4201096293872209, "grad_norm": 0.046689290553331375, "learning_rate": 8.146230194290739e-06, "loss": 0.0015, "step": 86790 }, { "epoch": 1.4202732553382966, "grad_norm": 0.062029968947172165, "learning_rate": 8.145675273625232e-06, "loss": 0.002, "step": 86800 }, { "epoch": 1.4204368812893726, "grad_norm": 0.028672389686107635, "learning_rate": 8.145120288822086e-06, "loss": 0.0021, "step": 86810 }, { "epoch": 1.4206005072404484, "grad_norm": 0.11127030849456787, "learning_rate": 8.144565239892608e-06, "loss": 0.0028, "step": 86820 }, { "epoch": 1.4207641331915242, "grad_norm": 0.07011177390813828, "learning_rate": 8.144010126848121e-06, "loss": 0.0017, "step": 86830 }, { "epoch": 1.4209277591426, "grad_norm": 0.11445792019367218, "learning_rate": 8.143454949699942e-06, "loss": 0.0014, "step": 86840 }, { "epoch": 1.4210913850936757, "grad_norm": 0.023528361693024635, "learning_rate": 8.142899708459389e-06, "loss": 0.0029, "step": 86850 }, { "epoch": 1.4212550110447517, "grad_norm": 0.11159677803516388, "learning_rate": 8.142344403137785e-06, "loss": 0.0043, "step": 86860 }, { "epoch": 1.4214186369958275, "grad_norm": 0.010012522339820862, "learning_rate": 8.14178903374645e-06, "loss": 0.0021, "step": 86870 }, { "epoch": 1.4215822629469033, "grad_norm": 0.385044664144516, "learning_rate": 8.14123360029671e-06, "loss": 0.002, "step": 86880 }, { "epoch": 1.4217458888979793, "grad_norm": 0.1438375562429428, "learning_rate": 8.140678102799886e-06, "loss": 0.0023, "step": 86890 }, { "epoch": 1.421909514849055, "grad_norm": 0.02692410536110401, "learning_rate": 8.140122541267308e-06, "loss": 0.0018, "step": 86900 }, { "epoch": 1.4220731408001308, "grad_norm": 0.06630314141511917, "learning_rate": 8.139566915710302e-06, "loss": 0.0024, "step": 86910 }, { "epoch": 1.4222367667512068, "grad_norm": 0.04141812026500702, "learning_rate": 8.139011226140197e-06, "loss": 0.0015, "step": 86920 }, { "epoch": 1.4224003927022826, "grad_norm": 0.08557868748903275, "learning_rate": 8.138455472568321e-06, "loss": 0.0017, "step": 86930 }, { "epoch": 1.4225640186533584, "grad_norm": 0.030390935018658638, "learning_rate": 8.13789965500601e-06, "loss": 0.0024, "step": 86940 }, { "epoch": 1.4227276446044343, "grad_norm": 0.08096308261156082, "learning_rate": 8.13734377346459e-06, "loss": 0.0015, "step": 86950 }, { "epoch": 1.4228912705555101, "grad_norm": 0.004069402348250151, "learning_rate": 8.136787827955401e-06, "loss": 0.0017, "step": 86960 }, { "epoch": 1.423054896506586, "grad_norm": 0.027260039001703262, "learning_rate": 8.136231818489777e-06, "loss": 0.0009, "step": 86970 }, { "epoch": 1.423218522457662, "grad_norm": 0.05655611306428909, "learning_rate": 8.135675745079053e-06, "loss": 0.0014, "step": 86980 }, { "epoch": 1.4233821484087377, "grad_norm": 0.12350815534591675, "learning_rate": 8.135119607734566e-06, "loss": 0.0016, "step": 86990 }, { "epoch": 1.4235457743598134, "grad_norm": 0.026846570894122124, "learning_rate": 8.134563406467656e-06, "loss": 0.0017, "step": 87000 }, { "epoch": 1.4237094003108894, "grad_norm": 0.026594577357172966, "learning_rate": 8.134007141289666e-06, "loss": 0.0025, "step": 87010 }, { "epoch": 1.4238730262619652, "grad_norm": 0.09432078152894974, "learning_rate": 8.133450812211934e-06, "loss": 0.0025, "step": 87020 }, { "epoch": 1.424036652213041, "grad_norm": 0.08865708112716675, "learning_rate": 8.132894419245805e-06, "loss": 0.0025, "step": 87030 }, { "epoch": 1.4242002781641168, "grad_norm": 0.06814311444759369, "learning_rate": 8.132337962402626e-06, "loss": 0.0029, "step": 87040 }, { "epoch": 1.4243639041151925, "grad_norm": 0.0353669710457325, "learning_rate": 8.131781441693737e-06, "loss": 0.0015, "step": 87050 }, { "epoch": 1.4245275300662685, "grad_norm": 0.05837947130203247, "learning_rate": 8.131224857130488e-06, "loss": 0.0016, "step": 87060 }, { "epoch": 1.4246911560173443, "grad_norm": 0.12059807032346725, "learning_rate": 8.130668208724228e-06, "loss": 0.0011, "step": 87070 }, { "epoch": 1.42485478196842, "grad_norm": 0.040729742497205734, "learning_rate": 8.130111496486306e-06, "loss": 0.0015, "step": 87080 }, { "epoch": 1.425018407919496, "grad_norm": 0.06883910298347473, "learning_rate": 8.129554720428073e-06, "loss": 0.0023, "step": 87090 }, { "epoch": 1.4251820338705719, "grad_norm": 0.03881394490599632, "learning_rate": 8.128997880560878e-06, "loss": 0.0014, "step": 87100 }, { "epoch": 1.4253456598216476, "grad_norm": 0.07819048315286636, "learning_rate": 8.12844097689608e-06, "loss": 0.0009, "step": 87110 }, { "epoch": 1.4255092857727236, "grad_norm": 0.031271349638700485, "learning_rate": 8.12788400944503e-06, "loss": 0.002, "step": 87120 }, { "epoch": 1.4256729117237994, "grad_norm": 0.03436679393053055, "learning_rate": 8.127326978219086e-06, "loss": 0.0013, "step": 87130 }, { "epoch": 1.4258365376748752, "grad_norm": 0.12935753166675568, "learning_rate": 8.126769883229604e-06, "loss": 0.0019, "step": 87140 }, { "epoch": 1.4260001636259512, "grad_norm": 0.033385246992111206, "learning_rate": 8.126212724487945e-06, "loss": 0.0016, "step": 87150 }, { "epoch": 1.426163789577027, "grad_norm": 0.010064649395644665, "learning_rate": 8.125655502005465e-06, "loss": 0.0008, "step": 87160 }, { "epoch": 1.4263274155281027, "grad_norm": 0.051320530474185944, "learning_rate": 8.12509821579353e-06, "loss": 0.0016, "step": 87170 }, { "epoch": 1.4264910414791787, "grad_norm": 0.13620422780513763, "learning_rate": 8.124540865863497e-06, "loss": 0.001, "step": 87180 }, { "epoch": 1.4266546674302545, "grad_norm": 0.02491678297519684, "learning_rate": 8.123983452226736e-06, "loss": 0.0013, "step": 87190 }, { "epoch": 1.4268182933813303, "grad_norm": 0.0062821595929563046, "learning_rate": 8.123425974894608e-06, "loss": 0.0025, "step": 87200 }, { "epoch": 1.4269819193324063, "grad_norm": 0.0675419494509697, "learning_rate": 8.12286843387848e-06, "loss": 0.0021, "step": 87210 }, { "epoch": 1.427145545283482, "grad_norm": 0.06756063550710678, "learning_rate": 8.122310829189721e-06, "loss": 0.0015, "step": 87220 }, { "epoch": 1.4273091712345578, "grad_norm": 0.12638789415359497, "learning_rate": 8.1217531608397e-06, "loss": 0.0028, "step": 87230 }, { "epoch": 1.4274727971856336, "grad_norm": 0.18564410507678986, "learning_rate": 8.121195428839787e-06, "loss": 0.0022, "step": 87240 }, { "epoch": 1.4276364231367094, "grad_norm": 0.04881761968135834, "learning_rate": 8.120637633201354e-06, "loss": 0.0017, "step": 87250 }, { "epoch": 1.4278000490877854, "grad_norm": 0.05406583473086357, "learning_rate": 8.120079773935774e-06, "loss": 0.0019, "step": 87260 }, { "epoch": 1.4279636750388611, "grad_norm": 0.08884704113006592, "learning_rate": 8.119521851054419e-06, "loss": 0.0023, "step": 87270 }, { "epoch": 1.428127300989937, "grad_norm": 0.02096165157854557, "learning_rate": 8.118963864568669e-06, "loss": 0.0028, "step": 87280 }, { "epoch": 1.428290926941013, "grad_norm": 0.10657638311386108, "learning_rate": 8.118405814489898e-06, "loss": 0.0016, "step": 87290 }, { "epoch": 1.4284545528920887, "grad_norm": 0.00566820940002799, "learning_rate": 8.117847700829483e-06, "loss": 0.0017, "step": 87300 }, { "epoch": 1.4286181788431644, "grad_norm": 0.01593773066997528, "learning_rate": 8.117289523598806e-06, "loss": 0.0016, "step": 87310 }, { "epoch": 1.4287818047942404, "grad_norm": 0.08832352608442307, "learning_rate": 8.116731282809245e-06, "loss": 0.001, "step": 87320 }, { "epoch": 1.4289454307453162, "grad_norm": 0.05696431174874306, "learning_rate": 8.116172978472185e-06, "loss": 0.0014, "step": 87330 }, { "epoch": 1.429109056696392, "grad_norm": 0.002298606326803565, "learning_rate": 8.11561461059901e-06, "loss": 0.0007, "step": 87340 }, { "epoch": 1.429272682647468, "grad_norm": 0.12751102447509766, "learning_rate": 8.115056179201102e-06, "loss": 0.002, "step": 87350 }, { "epoch": 1.4294363085985438, "grad_norm": 0.0598100945353508, "learning_rate": 8.114497684289846e-06, "loss": 0.0028, "step": 87360 }, { "epoch": 1.4295999345496195, "grad_norm": 0.08738629519939423, "learning_rate": 8.113939125876633e-06, "loss": 0.0018, "step": 87370 }, { "epoch": 1.4297635605006955, "grad_norm": 0.03913773223757744, "learning_rate": 8.113380503972848e-06, "loss": 0.002, "step": 87380 }, { "epoch": 1.4299271864517713, "grad_norm": 0.049211643636226654, "learning_rate": 8.112821818589883e-06, "loss": 0.0019, "step": 87390 }, { "epoch": 1.430090812402847, "grad_norm": 0.06628332287073135, "learning_rate": 8.112263069739129e-06, "loss": 0.0015, "step": 87400 }, { "epoch": 1.4302544383539229, "grad_norm": 0.01592564396560192, "learning_rate": 8.111704257431978e-06, "loss": 0.0011, "step": 87410 }, { "epoch": 1.4304180643049988, "grad_norm": 0.017824172973632812, "learning_rate": 8.111145381679821e-06, "loss": 0.0009, "step": 87420 }, { "epoch": 1.4305816902560746, "grad_norm": 0.03734183683991432, "learning_rate": 8.110586442494059e-06, "loss": 0.0025, "step": 87430 }, { "epoch": 1.4307453162071504, "grad_norm": 0.03457338362932205, "learning_rate": 8.110027439886084e-06, "loss": 0.001, "step": 87440 }, { "epoch": 1.4309089421582262, "grad_norm": 0.018053093925118446, "learning_rate": 8.109468373867294e-06, "loss": 0.0025, "step": 87450 }, { "epoch": 1.4310725681093022, "grad_norm": 0.0014696402940899134, "learning_rate": 8.108909244449088e-06, "loss": 0.0017, "step": 87460 }, { "epoch": 1.431236194060378, "grad_norm": 0.046066537499427795, "learning_rate": 8.108350051642865e-06, "loss": 0.0016, "step": 87470 }, { "epoch": 1.4313998200114537, "grad_norm": 0.11195012181997299, "learning_rate": 8.107790795460032e-06, "loss": 0.0014, "step": 87480 }, { "epoch": 1.4315634459625297, "grad_norm": 0.0700952485203743, "learning_rate": 8.107231475911986e-06, "loss": 0.0016, "step": 87490 }, { "epoch": 1.4317270719136055, "grad_norm": 0.06493076682090759, "learning_rate": 8.10667209301013e-06, "loss": 0.0014, "step": 87500 }, { "epoch": 1.4318906978646813, "grad_norm": 0.07041225582361221, "learning_rate": 8.106112646765876e-06, "loss": 0.0017, "step": 87510 }, { "epoch": 1.4320543238157573, "grad_norm": 0.08171770721673965, "learning_rate": 8.105553137190625e-06, "loss": 0.0011, "step": 87520 }, { "epoch": 1.432217949766833, "grad_norm": 0.03177402913570404, "learning_rate": 8.104993564295786e-06, "loss": 0.0012, "step": 87530 }, { "epoch": 1.4323815757179088, "grad_norm": 0.002381363185122609, "learning_rate": 8.104433928092772e-06, "loss": 0.0017, "step": 87540 }, { "epoch": 1.4325452016689848, "grad_norm": 0.14247509837150574, "learning_rate": 8.103874228592987e-06, "loss": 0.0011, "step": 87550 }, { "epoch": 1.4327088276200606, "grad_norm": 0.14177890121936798, "learning_rate": 8.103314465807847e-06, "loss": 0.0011, "step": 87560 }, { "epoch": 1.4328724535711364, "grad_norm": 0.027440574020147324, "learning_rate": 8.102754639748766e-06, "loss": 0.0011, "step": 87570 }, { "epoch": 1.4330360795222123, "grad_norm": 0.11472063511610031, "learning_rate": 8.102194750427156e-06, "loss": 0.0023, "step": 87580 }, { "epoch": 1.4331997054732881, "grad_norm": 0.08697143197059631, "learning_rate": 8.101634797854433e-06, "loss": 0.0014, "step": 87590 }, { "epoch": 1.433363331424364, "grad_norm": 0.002411076333373785, "learning_rate": 8.101074782042015e-06, "loss": 0.0011, "step": 87600 }, { "epoch": 1.4335269573754397, "grad_norm": 0.05727458745241165, "learning_rate": 8.100514703001318e-06, "loss": 0.0021, "step": 87610 }, { "epoch": 1.4336905833265157, "grad_norm": 0.07849786430597305, "learning_rate": 8.099954560743764e-06, "loss": 0.0014, "step": 87620 }, { "epoch": 1.4338542092775914, "grad_norm": 0.020609278231859207, "learning_rate": 8.099394355280774e-06, "loss": 0.0013, "step": 87630 }, { "epoch": 1.4340178352286672, "grad_norm": 0.04958502575755119, "learning_rate": 8.098834086623768e-06, "loss": 0.0032, "step": 87640 }, { "epoch": 1.434181461179743, "grad_norm": 0.07530534267425537, "learning_rate": 8.098273754784169e-06, "loss": 0.0007, "step": 87650 }, { "epoch": 1.434345087130819, "grad_norm": 0.014120771549642086, "learning_rate": 8.097713359773405e-06, "loss": 0.0019, "step": 87660 }, { "epoch": 1.4345087130818948, "grad_norm": 0.05322503671050072, "learning_rate": 8.0971529016029e-06, "loss": 0.0012, "step": 87670 }, { "epoch": 1.4346723390329705, "grad_norm": 0.061401285231113434, "learning_rate": 8.096592380284083e-06, "loss": 0.0013, "step": 87680 }, { "epoch": 1.4348359649840465, "grad_norm": 0.05594056099653244, "learning_rate": 8.096031795828377e-06, "loss": 0.0021, "step": 87690 }, { "epoch": 1.4349995909351223, "grad_norm": 0.03292938321828842, "learning_rate": 8.095471148247218e-06, "loss": 0.002, "step": 87700 }, { "epoch": 1.435163216886198, "grad_norm": 0.10764303058385849, "learning_rate": 8.094910437552035e-06, "loss": 0.0017, "step": 87710 }, { "epoch": 1.435326842837274, "grad_norm": 0.011055572889745235, "learning_rate": 8.094349663754259e-06, "loss": 0.0015, "step": 87720 }, { "epoch": 1.4354904687883498, "grad_norm": 0.08367232978343964, "learning_rate": 8.093788826865326e-06, "loss": 0.0027, "step": 87730 }, { "epoch": 1.4356540947394256, "grad_norm": 0.09636068344116211, "learning_rate": 8.093227926896671e-06, "loss": 0.0034, "step": 87740 }, { "epoch": 1.4358177206905016, "grad_norm": 0.05079222097992897, "learning_rate": 8.092666963859728e-06, "loss": 0.0014, "step": 87750 }, { "epoch": 1.4359813466415774, "grad_norm": 0.11230233311653137, "learning_rate": 8.092105937765934e-06, "loss": 0.0025, "step": 87760 }, { "epoch": 1.4361449725926532, "grad_norm": 0.003880574367940426, "learning_rate": 8.091544848626734e-06, "loss": 0.0023, "step": 87770 }, { "epoch": 1.4363085985437292, "grad_norm": 0.053959719836711884, "learning_rate": 8.09098369645356e-06, "loss": 0.0018, "step": 87780 }, { "epoch": 1.436472224494805, "grad_norm": 0.08432050794363022, "learning_rate": 8.09042248125786e-06, "loss": 0.002, "step": 87790 }, { "epoch": 1.4366358504458807, "grad_norm": 0.053326938301324844, "learning_rate": 8.089861203051072e-06, "loss": 0.0015, "step": 87800 }, { "epoch": 1.4367994763969565, "grad_norm": 0.0512537881731987, "learning_rate": 8.089299861844643e-06, "loss": 0.0011, "step": 87810 }, { "epoch": 1.4369631023480323, "grad_norm": 0.07134948670864105, "learning_rate": 8.088738457650016e-06, "loss": 0.0016, "step": 87820 }, { "epoch": 1.4371267282991083, "grad_norm": 0.025154300034046173, "learning_rate": 8.08817699047864e-06, "loss": 0.001, "step": 87830 }, { "epoch": 1.437290354250184, "grad_norm": 0.0370798222720623, "learning_rate": 8.08761546034196e-06, "loss": 0.003, "step": 87840 }, { "epoch": 1.4374539802012598, "grad_norm": 0.08355949074029922, "learning_rate": 8.08705386725143e-06, "loss": 0.002, "step": 87850 }, { "epoch": 1.4376176061523358, "grad_norm": 0.03477158024907112, "learning_rate": 8.086492211218493e-06, "loss": 0.002, "step": 87860 }, { "epoch": 1.4377812321034116, "grad_norm": 0.06191577762365341, "learning_rate": 8.085930492254607e-06, "loss": 0.0013, "step": 87870 }, { "epoch": 1.4379448580544874, "grad_norm": 0.026977747678756714, "learning_rate": 8.085368710371221e-06, "loss": 0.0012, "step": 87880 }, { "epoch": 1.4381084840055633, "grad_norm": 0.032396428287029266, "learning_rate": 8.084806865579793e-06, "loss": 0.0011, "step": 87890 }, { "epoch": 1.4382721099566391, "grad_norm": 0.011736291460692883, "learning_rate": 8.084244957891776e-06, "loss": 0.0012, "step": 87900 }, { "epoch": 1.438435735907715, "grad_norm": 0.14656144380569458, "learning_rate": 8.083682987318628e-06, "loss": 0.0025, "step": 87910 }, { "epoch": 1.438599361858791, "grad_norm": 0.07041492313146591, "learning_rate": 8.083120953871805e-06, "loss": 0.0015, "step": 87920 }, { "epoch": 1.4387629878098667, "grad_norm": 0.04531499370932579, "learning_rate": 8.082558857562767e-06, "loss": 0.0015, "step": 87930 }, { "epoch": 1.4389266137609424, "grad_norm": 0.06615522503852844, "learning_rate": 8.081996698402979e-06, "loss": 0.0019, "step": 87940 }, { "epoch": 1.4390902397120184, "grad_norm": 0.15318423509597778, "learning_rate": 8.081434476403898e-06, "loss": 0.0022, "step": 87950 }, { "epoch": 1.4392538656630942, "grad_norm": 0.013561123050749302, "learning_rate": 8.080872191576987e-06, "loss": 0.0023, "step": 87960 }, { "epoch": 1.43941749161417, "grad_norm": 0.04760092496871948, "learning_rate": 8.080309843933714e-06, "loss": 0.0022, "step": 87970 }, { "epoch": 1.439581117565246, "grad_norm": 0.0327962189912796, "learning_rate": 8.079747433485542e-06, "loss": 0.0008, "step": 87980 }, { "epoch": 1.4397447435163218, "grad_norm": 0.15630948543548584, "learning_rate": 8.07918496024394e-06, "loss": 0.0016, "step": 87990 }, { "epoch": 1.4399083694673975, "grad_norm": 0.039964865893125534, "learning_rate": 8.078622424220374e-06, "loss": 0.0015, "step": 88000 }, { "epoch": 1.4400719954184733, "grad_norm": 0.09298530966043472, "learning_rate": 8.078059825426317e-06, "loss": 0.0016, "step": 88010 }, { "epoch": 1.440235621369549, "grad_norm": 0.007121716160327196, "learning_rate": 8.077497163873237e-06, "loss": 0.002, "step": 88020 }, { "epoch": 1.440399247320625, "grad_norm": 0.10856568068265915, "learning_rate": 8.076934439572607e-06, "loss": 0.0013, "step": 88030 }, { "epoch": 1.4405628732717008, "grad_norm": 0.07874457538127899, "learning_rate": 8.0763716525359e-06, "loss": 0.0019, "step": 88040 }, { "epoch": 1.4407264992227766, "grad_norm": 0.0405912771821022, "learning_rate": 8.075808802774592e-06, "loss": 0.001, "step": 88050 }, { "epoch": 1.4408901251738526, "grad_norm": 0.020086845383048058, "learning_rate": 8.07524589030016e-06, "loss": 0.0021, "step": 88060 }, { "epoch": 1.4410537511249284, "grad_norm": 0.10715054720640182, "learning_rate": 8.074682915124078e-06, "loss": 0.0034, "step": 88070 }, { "epoch": 1.4412173770760042, "grad_norm": 0.007371082901954651, "learning_rate": 8.074119877257824e-06, "loss": 0.0015, "step": 88080 }, { "epoch": 1.4413810030270802, "grad_norm": 0.06857823580503464, "learning_rate": 8.073556776712883e-06, "loss": 0.0015, "step": 88090 }, { "epoch": 1.441544628978156, "grad_norm": 0.06123557314276695, "learning_rate": 8.072993613500734e-06, "loss": 0.001, "step": 88100 }, { "epoch": 1.4417082549292317, "grad_norm": 0.04252896457910538, "learning_rate": 8.072430387632857e-06, "loss": 0.0022, "step": 88110 }, { "epoch": 1.4418718808803077, "grad_norm": 0.04831987991929054, "learning_rate": 8.071867099120739e-06, "loss": 0.0021, "step": 88120 }, { "epoch": 1.4420355068313835, "grad_norm": 0.058172132819890976, "learning_rate": 8.071303747975861e-06, "loss": 0.0027, "step": 88130 }, { "epoch": 1.4421991327824593, "grad_norm": 0.08792924880981445, "learning_rate": 8.070740334209712e-06, "loss": 0.0021, "step": 88140 }, { "epoch": 1.4423627587335353, "grad_norm": 0.09092686325311661, "learning_rate": 8.070176857833782e-06, "loss": 0.0023, "step": 88150 }, { "epoch": 1.442526384684611, "grad_norm": 0.059450626373291016, "learning_rate": 8.069613318859555e-06, "loss": 0.002, "step": 88160 }, { "epoch": 1.4426900106356868, "grad_norm": 0.049400098621845245, "learning_rate": 8.069049717298522e-06, "loss": 0.0014, "step": 88170 }, { "epoch": 1.4428536365867626, "grad_norm": 0.039475731551647186, "learning_rate": 8.068486053162177e-06, "loss": 0.0048, "step": 88180 }, { "epoch": 1.4430172625378386, "grad_norm": 0.11152265220880508, "learning_rate": 8.06792232646201e-06, "loss": 0.0028, "step": 88190 }, { "epoch": 1.4431808884889143, "grad_norm": 0.022507760673761368, "learning_rate": 8.067358537209517e-06, "loss": 0.0011, "step": 88200 }, { "epoch": 1.4433445144399901, "grad_norm": 0.011659185402095318, "learning_rate": 8.066794685416192e-06, "loss": 0.0014, "step": 88210 }, { "epoch": 1.443508140391066, "grad_norm": 0.11818519979715347, "learning_rate": 8.066230771093531e-06, "loss": 0.0019, "step": 88220 }, { "epoch": 1.443671766342142, "grad_norm": 0.12649664282798767, "learning_rate": 8.065666794253032e-06, "loss": 0.0019, "step": 88230 }, { "epoch": 1.4438353922932177, "grad_norm": 0.09506122022867203, "learning_rate": 8.065102754906195e-06, "loss": 0.0033, "step": 88240 }, { "epoch": 1.4439990182442934, "grad_norm": 0.04342591390013695, "learning_rate": 8.06453865306452e-06, "loss": 0.0023, "step": 88250 }, { "epoch": 1.4441626441953694, "grad_norm": 0.055975284427404404, "learning_rate": 8.063974488739506e-06, "loss": 0.0011, "step": 88260 }, { "epoch": 1.4443262701464452, "grad_norm": 0.10332855582237244, "learning_rate": 8.063410261942661e-06, "loss": 0.0018, "step": 88270 }, { "epoch": 1.444489896097521, "grad_norm": 0.10131540149450302, "learning_rate": 8.062845972685484e-06, "loss": 0.0015, "step": 88280 }, { "epoch": 1.444653522048597, "grad_norm": 0.11468470096588135, "learning_rate": 8.062281620979483e-06, "loss": 0.0023, "step": 88290 }, { "epoch": 1.4448171479996728, "grad_norm": 0.1757725030183792, "learning_rate": 8.061717206836163e-06, "loss": 0.0033, "step": 88300 }, { "epoch": 1.4449807739507485, "grad_norm": 0.0344943068921566, "learning_rate": 8.061152730267035e-06, "loss": 0.001, "step": 88310 }, { "epoch": 1.4451443999018245, "grad_norm": 0.08995692431926727, "learning_rate": 8.060588191283607e-06, "loss": 0.0024, "step": 88320 }, { "epoch": 1.4453080258529003, "grad_norm": 0.014005608856678009, "learning_rate": 8.060023589897387e-06, "loss": 0.0011, "step": 88330 }, { "epoch": 1.445471651803976, "grad_norm": 0.07483003288507462, "learning_rate": 8.059458926119888e-06, "loss": 0.0012, "step": 88340 }, { "epoch": 1.445635277755052, "grad_norm": 0.07057053595781326, "learning_rate": 8.058894199962624e-06, "loss": 0.0018, "step": 88350 }, { "epoch": 1.4457989037061278, "grad_norm": 0.025117559358477592, "learning_rate": 8.05832941143711e-06, "loss": 0.0017, "step": 88360 }, { "epoch": 1.4459625296572036, "grad_norm": 0.04688750207424164, "learning_rate": 8.057764560554859e-06, "loss": 0.002, "step": 88370 }, { "epoch": 1.4461261556082794, "grad_norm": 0.21378332376480103, "learning_rate": 8.05719964732739e-06, "loss": 0.0027, "step": 88380 }, { "epoch": 1.4462897815593554, "grad_norm": 0.062016185373067856, "learning_rate": 8.05663467176622e-06, "loss": 0.002, "step": 88390 }, { "epoch": 1.4464534075104312, "grad_norm": 0.016850249841809273, "learning_rate": 8.056069633882868e-06, "loss": 0.0015, "step": 88400 }, { "epoch": 1.446617033461507, "grad_norm": 0.04969257861375809, "learning_rate": 8.055504533688855e-06, "loss": 0.0024, "step": 88410 }, { "epoch": 1.4467806594125827, "grad_norm": 0.01355340052396059, "learning_rate": 8.054939371195704e-06, "loss": 0.0019, "step": 88420 }, { "epoch": 1.4469442853636587, "grad_norm": 0.02232198603451252, "learning_rate": 8.054374146414936e-06, "loss": 0.0019, "step": 88430 }, { "epoch": 1.4471079113147345, "grad_norm": 0.1414230316877365, "learning_rate": 8.053808859358079e-06, "loss": 0.0032, "step": 88440 }, { "epoch": 1.4472715372658103, "grad_norm": 0.08685953915119171, "learning_rate": 8.053243510036654e-06, "loss": 0.0016, "step": 88450 }, { "epoch": 1.4474351632168863, "grad_norm": 0.038031432777643204, "learning_rate": 8.052678098462193e-06, "loss": 0.0013, "step": 88460 }, { "epoch": 1.447598789167962, "grad_norm": 0.046047456562519073, "learning_rate": 8.05211262464622e-06, "loss": 0.002, "step": 88470 }, { "epoch": 1.4477624151190378, "grad_norm": 0.005691900849342346, "learning_rate": 8.051547088600267e-06, "loss": 0.0023, "step": 88480 }, { "epoch": 1.4479260410701138, "grad_norm": 0.01985158585011959, "learning_rate": 8.050981490335865e-06, "loss": 0.0018, "step": 88490 }, { "epoch": 1.4480896670211896, "grad_norm": 0.05411594733595848, "learning_rate": 8.050415829864544e-06, "loss": 0.0018, "step": 88500 }, { "epoch": 1.4482532929722653, "grad_norm": 0.06661323457956314, "learning_rate": 8.049850107197839e-06, "loss": 0.0021, "step": 88510 }, { "epoch": 1.4484169189233413, "grad_norm": 0.10109138488769531, "learning_rate": 8.049284322347284e-06, "loss": 0.0021, "step": 88520 }, { "epoch": 1.4485805448744171, "grad_norm": 0.014180307276546955, "learning_rate": 8.048718475324414e-06, "loss": 0.0031, "step": 88530 }, { "epoch": 1.448744170825493, "grad_norm": 0.043766241520643234, "learning_rate": 8.048152566140768e-06, "loss": 0.0014, "step": 88540 }, { "epoch": 1.448907796776569, "grad_norm": 0.06364922225475311, "learning_rate": 8.047586594807883e-06, "loss": 0.0015, "step": 88550 }, { "epoch": 1.4490714227276447, "grad_norm": 0.05437728762626648, "learning_rate": 8.0470205613373e-06, "loss": 0.0012, "step": 88560 }, { "epoch": 1.4492350486787204, "grad_norm": 0.04416601359844208, "learning_rate": 8.04645446574056e-06, "loss": 0.0018, "step": 88570 }, { "epoch": 1.4493986746297962, "grad_norm": 0.07854390144348145, "learning_rate": 8.045888308029204e-06, "loss": 0.0019, "step": 88580 }, { "epoch": 1.4495623005808722, "grad_norm": 0.051488324999809265, "learning_rate": 8.045322088214774e-06, "loss": 0.0012, "step": 88590 }, { "epoch": 1.449725926531948, "grad_norm": 0.0016302771400660276, "learning_rate": 8.044755806308818e-06, "loss": 0.0017, "step": 88600 }, { "epoch": 1.4498895524830238, "grad_norm": 0.020245110616087914, "learning_rate": 8.04418946232288e-06, "loss": 0.0024, "step": 88610 }, { "epoch": 1.4500531784340995, "grad_norm": 0.04111722111701965, "learning_rate": 8.043623056268509e-06, "loss": 0.001, "step": 88620 }, { "epoch": 1.4502168043851755, "grad_norm": 0.1183180883526802, "learning_rate": 8.043056588157251e-06, "loss": 0.0046, "step": 88630 }, { "epoch": 1.4503804303362513, "grad_norm": 0.18319883942604065, "learning_rate": 8.042490058000658e-06, "loss": 0.0023, "step": 88640 }, { "epoch": 1.450544056287327, "grad_norm": 0.11228092759847641, "learning_rate": 8.04192346581028e-06, "loss": 0.0023, "step": 88650 }, { "epoch": 1.450707682238403, "grad_norm": 0.06723596900701523, "learning_rate": 8.04135681159767e-06, "loss": 0.0016, "step": 88660 }, { "epoch": 1.4508713081894788, "grad_norm": 0.04019910842180252, "learning_rate": 8.04079009537438e-06, "loss": 0.0018, "step": 88670 }, { "epoch": 1.4510349341405546, "grad_norm": 0.03208707645535469, "learning_rate": 8.040223317151967e-06, "loss": 0.0021, "step": 88680 }, { "epoch": 1.4511985600916306, "grad_norm": 0.0381612554192543, "learning_rate": 8.039656476941986e-06, "loss": 0.0016, "step": 88690 }, { "epoch": 1.4513621860427064, "grad_norm": 0.05983539670705795, "learning_rate": 8.039089574755996e-06, "loss": 0.0023, "step": 88700 }, { "epoch": 1.4515258119937822, "grad_norm": 0.026381008327007294, "learning_rate": 8.038522610605553e-06, "loss": 0.0014, "step": 88710 }, { "epoch": 1.4516894379448582, "grad_norm": 0.04930445924401283, "learning_rate": 8.037955584502217e-06, "loss": 0.0012, "step": 88720 }, { "epoch": 1.451853063895934, "grad_norm": 0.10779973119497299, "learning_rate": 8.037388496457553e-06, "loss": 0.0024, "step": 88730 }, { "epoch": 1.4520166898470097, "grad_norm": 0.07489132136106491, "learning_rate": 8.036821346483116e-06, "loss": 0.002, "step": 88740 }, { "epoch": 1.4521803157980857, "grad_norm": 0.08242703229188919, "learning_rate": 8.036254134590478e-06, "loss": 0.0019, "step": 88750 }, { "epoch": 1.4523439417491615, "grad_norm": 0.017415540292859077, "learning_rate": 8.0356868607912e-06, "loss": 0.0024, "step": 88760 }, { "epoch": 1.4525075677002373, "grad_norm": 0.03667505830526352, "learning_rate": 8.03511952509685e-06, "loss": 0.0013, "step": 88770 }, { "epoch": 1.452671193651313, "grad_norm": 0.0315861701965332, "learning_rate": 8.034552127518994e-06, "loss": 0.0015, "step": 88780 }, { "epoch": 1.4528348196023888, "grad_norm": 0.10804776102304459, "learning_rate": 8.033984668069198e-06, "loss": 0.0012, "step": 88790 }, { "epoch": 1.4529984455534648, "grad_norm": 0.08428821712732315, "learning_rate": 8.033417146759037e-06, "loss": 0.0019, "step": 88800 }, { "epoch": 1.4531620715045406, "grad_norm": 0.02847917564213276, "learning_rate": 8.03284956360008e-06, "loss": 0.0008, "step": 88810 }, { "epoch": 1.4533256974556163, "grad_norm": 0.033524975180625916, "learning_rate": 8.0322819186039e-06, "loss": 0.0014, "step": 88820 }, { "epoch": 1.4534893234066923, "grad_norm": 0.0686885342001915, "learning_rate": 8.031714211782069e-06, "loss": 0.002, "step": 88830 }, { "epoch": 1.4536529493577681, "grad_norm": 0.030316483229398727, "learning_rate": 8.031146443146165e-06, "loss": 0.0027, "step": 88840 }, { "epoch": 1.453816575308844, "grad_norm": 0.04183785244822502, "learning_rate": 8.030578612707763e-06, "loss": 0.0024, "step": 88850 }, { "epoch": 1.45398020125992, "grad_norm": 0.0524381585419178, "learning_rate": 8.030010720478439e-06, "loss": 0.0021, "step": 88860 }, { "epoch": 1.4541438272109957, "grad_norm": 0.04690251499414444, "learning_rate": 8.029442766469775e-06, "loss": 0.0014, "step": 88870 }, { "epoch": 1.4543074531620714, "grad_norm": 0.0696665421128273, "learning_rate": 8.02887475069335e-06, "loss": 0.0021, "step": 88880 }, { "epoch": 1.4544710791131474, "grad_norm": 0.11572691053152084, "learning_rate": 8.028306673160743e-06, "loss": 0.0025, "step": 88890 }, { "epoch": 1.4546347050642232, "grad_norm": 0.05450555309653282, "learning_rate": 8.027738533883539e-06, "loss": 0.0019, "step": 88900 }, { "epoch": 1.454798331015299, "grad_norm": 0.07526401430368423, "learning_rate": 8.027170332873323e-06, "loss": 0.0018, "step": 88910 }, { "epoch": 1.454961956966375, "grad_norm": 0.05829628184437752, "learning_rate": 8.026602070141676e-06, "loss": 0.0015, "step": 88920 }, { "epoch": 1.4551255829174508, "grad_norm": 0.12112385034561157, "learning_rate": 8.026033745700188e-06, "loss": 0.0022, "step": 88930 }, { "epoch": 1.4552892088685265, "grad_norm": 0.019666770473122597, "learning_rate": 8.025465359560445e-06, "loss": 0.0022, "step": 88940 }, { "epoch": 1.4554528348196025, "grad_norm": 0.06329824775457382, "learning_rate": 8.024896911734036e-06, "loss": 0.0019, "step": 88950 }, { "epoch": 1.4556164607706783, "grad_norm": 0.07140061259269714, "learning_rate": 8.024328402232552e-06, "loss": 0.0017, "step": 88960 }, { "epoch": 1.455780086721754, "grad_norm": 0.0353868305683136, "learning_rate": 8.023759831067583e-06, "loss": 0.0015, "step": 88970 }, { "epoch": 1.4559437126728298, "grad_norm": 0.049538932740688324, "learning_rate": 8.023191198250724e-06, "loss": 0.002, "step": 88980 }, { "epoch": 1.4561073386239056, "grad_norm": 0.03503125160932541, "learning_rate": 8.022622503793569e-06, "loss": 0.0022, "step": 88990 }, { "epoch": 1.4562709645749816, "grad_norm": 0.07651124149560928, "learning_rate": 8.02205374770771e-06, "loss": 0.0019, "step": 89000 }, { "epoch": 1.4564345905260574, "grad_norm": 0.07074853777885437, "learning_rate": 8.021484930004745e-06, "loss": 0.0012, "step": 89010 }, { "epoch": 1.4565982164771332, "grad_norm": 0.07129286974668503, "learning_rate": 8.020916050696272e-06, "loss": 0.0015, "step": 89020 }, { "epoch": 1.4567618424282092, "grad_norm": 0.03850165382027626, "learning_rate": 8.020347109793892e-06, "loss": 0.001, "step": 89030 }, { "epoch": 1.456925468379285, "grad_norm": 0.08118460327386856, "learning_rate": 8.0197781073092e-06, "loss": 0.0029, "step": 89040 }, { "epoch": 1.4570890943303607, "grad_norm": 0.0478372797369957, "learning_rate": 8.019209043253803e-06, "loss": 0.0013, "step": 89050 }, { "epoch": 1.4572527202814367, "grad_norm": 0.05711648613214493, "learning_rate": 8.0186399176393e-06, "loss": 0.0061, "step": 89060 }, { "epoch": 1.4574163462325125, "grad_norm": 0.039049725979566574, "learning_rate": 8.018070730477297e-06, "loss": 0.0014, "step": 89070 }, { "epoch": 1.4575799721835883, "grad_norm": 0.04863554984331131, "learning_rate": 8.017501481779398e-06, "loss": 0.0023, "step": 89080 }, { "epoch": 1.4577435981346643, "grad_norm": 0.06155328080058098, "learning_rate": 8.01693217155721e-06, "loss": 0.0021, "step": 89090 }, { "epoch": 1.45790722408574, "grad_norm": 0.041242264211177826, "learning_rate": 8.016362799822342e-06, "loss": 0.0015, "step": 89100 }, { "epoch": 1.4580708500368158, "grad_norm": 0.2622426748275757, "learning_rate": 8.0157933665864e-06, "loss": 0.0018, "step": 89110 }, { "epoch": 1.4582344759878918, "grad_norm": 0.12571081519126892, "learning_rate": 8.015223871860998e-06, "loss": 0.0022, "step": 89120 }, { "epoch": 1.4583981019389676, "grad_norm": 0.0472026988863945, "learning_rate": 8.014654315657743e-06, "loss": 0.0012, "step": 89130 }, { "epoch": 1.4585617278900433, "grad_norm": 0.05946741998195648, "learning_rate": 8.014084697988254e-06, "loss": 0.0011, "step": 89140 }, { "epoch": 1.4587253538411191, "grad_norm": 0.03636970743536949, "learning_rate": 8.013515018864138e-06, "loss": 0.0014, "step": 89150 }, { "epoch": 1.4588889797921951, "grad_norm": 0.12262728065252304, "learning_rate": 8.012945278297016e-06, "loss": 0.0024, "step": 89160 }, { "epoch": 1.459052605743271, "grad_norm": 0.09003550559282303, "learning_rate": 8.012375476298501e-06, "loss": 0.0036, "step": 89170 }, { "epoch": 1.4592162316943467, "grad_norm": 0.019813835620880127, "learning_rate": 8.011805612880212e-06, "loss": 0.0015, "step": 89180 }, { "epoch": 1.4593798576454224, "grad_norm": 0.0885537639260292, "learning_rate": 8.011235688053768e-06, "loss": 0.0025, "step": 89190 }, { "epoch": 1.4595434835964984, "grad_norm": 0.381199449300766, "learning_rate": 8.01066570183079e-06, "loss": 0.002, "step": 89200 }, { "epoch": 1.4597071095475742, "grad_norm": 0.05675028637051582, "learning_rate": 8.0100956542229e-06, "loss": 0.0013, "step": 89210 }, { "epoch": 1.45987073549865, "grad_norm": 0.019064536318182945, "learning_rate": 8.009525545241717e-06, "loss": 0.0025, "step": 89220 }, { "epoch": 1.460034361449726, "grad_norm": 0.04065026715397835, "learning_rate": 8.008955374898869e-06, "loss": 0.0016, "step": 89230 }, { "epoch": 1.4601979874008018, "grad_norm": 0.06156119331717491, "learning_rate": 8.008385143205978e-06, "loss": 0.0008, "step": 89240 }, { "epoch": 1.4603616133518775, "grad_norm": 0.06902032345533371, "learning_rate": 8.007814850174675e-06, "loss": 0.0024, "step": 89250 }, { "epoch": 1.4605252393029535, "grad_norm": 0.015139404684305191, "learning_rate": 8.007244495816586e-06, "loss": 0.0015, "step": 89260 }, { "epoch": 1.4606888652540293, "grad_norm": 0.02190936915576458, "learning_rate": 8.006674080143337e-06, "loss": 0.002, "step": 89270 }, { "epoch": 1.460852491205105, "grad_norm": 0.09794832766056061, "learning_rate": 8.00610360316656e-06, "loss": 0.0014, "step": 89280 }, { "epoch": 1.461016117156181, "grad_norm": 0.1683366447687149, "learning_rate": 8.005533064897889e-06, "loss": 0.0024, "step": 89290 }, { "epoch": 1.4611797431072568, "grad_norm": 0.03971565142273903, "learning_rate": 8.004962465348956e-06, "loss": 0.0014, "step": 89300 }, { "epoch": 1.4613433690583326, "grad_norm": 0.05605550855398178, "learning_rate": 8.00439180453139e-06, "loss": 0.0024, "step": 89310 }, { "epoch": 1.4615069950094086, "grad_norm": 0.040574971586465836, "learning_rate": 8.003821082456835e-06, "loss": 0.0014, "step": 89320 }, { "epoch": 1.4616706209604844, "grad_norm": 0.08042531460523605, "learning_rate": 8.00325029913692e-06, "loss": 0.0008, "step": 89330 }, { "epoch": 1.4618342469115602, "grad_norm": 0.002910205628722906, "learning_rate": 8.002679454583285e-06, "loss": 0.0012, "step": 89340 }, { "epoch": 1.461997872862636, "grad_norm": 0.08651532232761383, "learning_rate": 8.00210854880757e-06, "loss": 0.0023, "step": 89350 }, { "epoch": 1.462161498813712, "grad_norm": 0.026236848905682564, "learning_rate": 8.001537581821416e-06, "loss": 0.0024, "step": 89360 }, { "epoch": 1.4623251247647877, "grad_norm": 0.027541056275367737, "learning_rate": 8.000966553636463e-06, "loss": 0.001, "step": 89370 }, { "epoch": 1.4624887507158635, "grad_norm": 0.05656750127673149, "learning_rate": 8.000395464264354e-06, "loss": 0.0011, "step": 89380 }, { "epoch": 1.4626523766669393, "grad_norm": 0.03274872899055481, "learning_rate": 7.999824313716733e-06, "loss": 0.0034, "step": 89390 }, { "epoch": 1.4628160026180153, "grad_norm": 0.10582852363586426, "learning_rate": 7.999253102005245e-06, "loss": 0.0013, "step": 89400 }, { "epoch": 1.462979628569091, "grad_norm": 0.14259764552116394, "learning_rate": 7.99868182914154e-06, "loss": 0.0021, "step": 89410 }, { "epoch": 1.4631432545201668, "grad_norm": 0.05579623579978943, "learning_rate": 7.99811049513726e-06, "loss": 0.0019, "step": 89420 }, { "epoch": 1.4633068804712428, "grad_norm": 0.05822779983282089, "learning_rate": 7.997539100004056e-06, "loss": 0.0023, "step": 89430 }, { "epoch": 1.4634705064223186, "grad_norm": 0.03152143582701683, "learning_rate": 7.996967643753581e-06, "loss": 0.0018, "step": 89440 }, { "epoch": 1.4636341323733943, "grad_norm": 0.0439387746155262, "learning_rate": 7.996396126397483e-06, "loss": 0.0025, "step": 89450 }, { "epoch": 1.4637977583244703, "grad_norm": 0.11891940236091614, "learning_rate": 7.995824547947419e-06, "loss": 0.0016, "step": 89460 }, { "epoch": 1.4639613842755461, "grad_norm": 0.1839444488286972, "learning_rate": 7.995252908415038e-06, "loss": 0.0028, "step": 89470 }, { "epoch": 1.464125010226622, "grad_norm": 0.02474386803805828, "learning_rate": 7.994681207811998e-06, "loss": 0.0034, "step": 89480 }, { "epoch": 1.4642886361776979, "grad_norm": 0.15091946721076965, "learning_rate": 7.994109446149956e-06, "loss": 0.0024, "step": 89490 }, { "epoch": 1.4644522621287737, "grad_norm": 0.4493570327758789, "learning_rate": 7.993537623440568e-06, "loss": 0.0008, "step": 89500 }, { "epoch": 1.4646158880798494, "grad_norm": 0.02878584899008274, "learning_rate": 7.992965739695493e-06, "loss": 0.0014, "step": 89510 }, { "epoch": 1.4647795140309254, "grad_norm": 0.025629563257098198, "learning_rate": 7.992393794926393e-06, "loss": 0.0011, "step": 89520 }, { "epoch": 1.4649431399820012, "grad_norm": 0.0386253260076046, "learning_rate": 7.99182178914493e-06, "loss": 0.001, "step": 89530 }, { "epoch": 1.465106765933077, "grad_norm": 0.3171965479850769, "learning_rate": 7.991249722362762e-06, "loss": 0.0015, "step": 89540 }, { "epoch": 1.4652703918841528, "grad_norm": 0.14258885383605957, "learning_rate": 7.990677594591559e-06, "loss": 0.001, "step": 89550 }, { "epoch": 1.4654340178352285, "grad_norm": 0.07608826458454132, "learning_rate": 7.99010540584298e-06, "loss": 0.0022, "step": 89560 }, { "epoch": 1.4655976437863045, "grad_norm": 0.04194560647010803, "learning_rate": 7.989533156128696e-06, "loss": 0.0009, "step": 89570 }, { "epoch": 1.4657612697373803, "grad_norm": 0.05801656097173691, "learning_rate": 7.988960845460375e-06, "loss": 0.0012, "step": 89580 }, { "epoch": 1.465924895688456, "grad_norm": 0.019615372642874718, "learning_rate": 7.988388473849684e-06, "loss": 0.0013, "step": 89590 }, { "epoch": 1.466088521639532, "grad_norm": 0.012659566476941109, "learning_rate": 7.987816041308293e-06, "loss": 0.0009, "step": 89600 }, { "epoch": 1.4662521475906078, "grad_norm": 0.014209723100066185, "learning_rate": 7.987243547847874e-06, "loss": 0.002, "step": 89610 }, { "epoch": 1.4664157735416836, "grad_norm": 0.048308584839105606, "learning_rate": 7.9866709934801e-06, "loss": 0.0014, "step": 89620 }, { "epoch": 1.4665793994927596, "grad_norm": 0.035589709877967834, "learning_rate": 7.986098378216644e-06, "loss": 0.0018, "step": 89630 }, { "epoch": 1.4667430254438354, "grad_norm": 0.06948991864919662, "learning_rate": 7.985525702069183e-06, "loss": 0.0012, "step": 89640 }, { "epoch": 1.4669066513949112, "grad_norm": 0.024343790486454964, "learning_rate": 7.984952965049391e-06, "loss": 0.0009, "step": 89650 }, { "epoch": 1.4670702773459872, "grad_norm": 0.10863609611988068, "learning_rate": 7.984380167168947e-06, "loss": 0.0017, "step": 89660 }, { "epoch": 1.467233903297063, "grad_norm": 0.017112713307142258, "learning_rate": 7.983807308439527e-06, "loss": 0.0018, "step": 89670 }, { "epoch": 1.4673975292481387, "grad_norm": 0.04722965136170387, "learning_rate": 7.983234388872816e-06, "loss": 0.0031, "step": 89680 }, { "epoch": 1.4675611551992147, "grad_norm": 0.1828913390636444, "learning_rate": 7.982661408480493e-06, "loss": 0.0013, "step": 89690 }, { "epoch": 1.4677247811502905, "grad_norm": 0.03170527517795563, "learning_rate": 7.98208836727424e-06, "loss": 0.002, "step": 89700 }, { "epoch": 1.4678884071013663, "grad_norm": 0.13871437311172485, "learning_rate": 7.981515265265741e-06, "loss": 0.0023, "step": 89710 }, { "epoch": 1.4680520330524423, "grad_norm": 0.051139891147613525, "learning_rate": 7.980942102466682e-06, "loss": 0.0014, "step": 89720 }, { "epoch": 1.468215659003518, "grad_norm": 0.0653398334980011, "learning_rate": 7.980368878888748e-06, "loss": 0.0015, "step": 89730 }, { "epoch": 1.4683792849545938, "grad_norm": 0.10349863022565842, "learning_rate": 7.97979559454363e-06, "loss": 0.0019, "step": 89740 }, { "epoch": 1.4685429109056696, "grad_norm": 0.1536974459886551, "learning_rate": 7.97922224944301e-06, "loss": 0.0016, "step": 89750 }, { "epoch": 1.4687065368567453, "grad_norm": 0.02766246162354946, "learning_rate": 7.978648843598586e-06, "loss": 0.0015, "step": 89760 }, { "epoch": 1.4688701628078213, "grad_norm": 0.07257676124572754, "learning_rate": 7.978075377022043e-06, "loss": 0.0018, "step": 89770 }, { "epoch": 1.4690337887588971, "grad_norm": 0.060533203184604645, "learning_rate": 7.977501849725076e-06, "loss": 0.0027, "step": 89780 }, { "epoch": 1.469197414709973, "grad_norm": 0.05046839267015457, "learning_rate": 7.976928261719379e-06, "loss": 0.0022, "step": 89790 }, { "epoch": 1.469361040661049, "grad_norm": 0.1349511593580246, "learning_rate": 7.976354613016646e-06, "loss": 0.0063, "step": 89800 }, { "epoch": 1.4695246666121247, "grad_norm": 0.06685233116149902, "learning_rate": 7.975780903628574e-06, "loss": 0.0014, "step": 89810 }, { "epoch": 1.4696882925632004, "grad_norm": 0.10659744590520859, "learning_rate": 7.97520713356686e-06, "loss": 0.0017, "step": 89820 }, { "epoch": 1.4698519185142764, "grad_norm": 0.057401467114686966, "learning_rate": 7.974633302843203e-06, "loss": 0.0008, "step": 89830 }, { "epoch": 1.4700155444653522, "grad_norm": 0.09935004264116287, "learning_rate": 7.974059411469304e-06, "loss": 0.0011, "step": 89840 }, { "epoch": 1.470179170416428, "grad_norm": 0.11456397920846939, "learning_rate": 7.97348545945686e-06, "loss": 0.0011, "step": 89850 }, { "epoch": 1.470342796367504, "grad_norm": 0.055334173142910004, "learning_rate": 7.972911446817579e-06, "loss": 0.0026, "step": 89860 }, { "epoch": 1.4705064223185798, "grad_norm": 0.17681476473808289, "learning_rate": 7.97233737356316e-06, "loss": 0.0015, "step": 89870 }, { "epoch": 1.4706700482696555, "grad_norm": 0.11800109595060349, "learning_rate": 7.971763239705312e-06, "loss": 0.0023, "step": 89880 }, { "epoch": 1.4708336742207315, "grad_norm": 0.0817442238330841, "learning_rate": 7.971189045255738e-06, "loss": 0.0017, "step": 89890 }, { "epoch": 1.4709973001718073, "grad_norm": 0.00430133705958724, "learning_rate": 7.970614790226147e-06, "loss": 0.0023, "step": 89900 }, { "epoch": 1.471160926122883, "grad_norm": 0.06564731895923615, "learning_rate": 7.970040474628245e-06, "loss": 0.0014, "step": 89910 }, { "epoch": 1.4713245520739588, "grad_norm": 0.0525529608130455, "learning_rate": 7.969466098473746e-06, "loss": 0.0015, "step": 89920 }, { "epoch": 1.4714881780250348, "grad_norm": 0.021234385669231415, "learning_rate": 7.968891661774358e-06, "loss": 0.001, "step": 89930 }, { "epoch": 1.4716518039761106, "grad_norm": 0.030038151890039444, "learning_rate": 7.968317164541792e-06, "loss": 0.0033, "step": 89940 }, { "epoch": 1.4718154299271864, "grad_norm": 0.03533266484737396, "learning_rate": 7.967742606787766e-06, "loss": 0.0024, "step": 89950 }, { "epoch": 1.4719790558782622, "grad_norm": 0.11015431582927704, "learning_rate": 7.96716798852399e-06, "loss": 0.0008, "step": 89960 }, { "epoch": 1.4721426818293382, "grad_norm": 0.029017608612775803, "learning_rate": 7.966593309762183e-06, "loss": 0.0025, "step": 89970 }, { "epoch": 1.472306307780414, "grad_norm": 0.03724600374698639, "learning_rate": 7.966018570514063e-06, "loss": 0.0013, "step": 89980 }, { "epoch": 1.4724699337314897, "grad_norm": 0.016723359003663063, "learning_rate": 7.965443770791345e-06, "loss": 0.0016, "step": 89990 }, { "epoch": 1.4726335596825657, "grad_norm": 0.0849829614162445, "learning_rate": 7.964868910605751e-06, "loss": 0.0012, "step": 90000 }, { "epoch": 1.4727971856336415, "grad_norm": 0.04567443206906319, "learning_rate": 7.964293989969002e-06, "loss": 0.003, "step": 90010 }, { "epoch": 1.4729608115847173, "grad_norm": 0.11049305647611618, "learning_rate": 7.963719008892818e-06, "loss": 0.0019, "step": 90020 }, { "epoch": 1.4731244375357933, "grad_norm": 0.10319826751947403, "learning_rate": 7.963143967388927e-06, "loss": 0.0022, "step": 90030 }, { "epoch": 1.473288063486869, "grad_norm": 0.04962516576051712, "learning_rate": 7.962568865469048e-06, "loss": 0.0013, "step": 90040 }, { "epoch": 1.4734516894379448, "grad_norm": 0.024980463087558746, "learning_rate": 7.96199370314491e-06, "loss": 0.0012, "step": 90050 }, { "epoch": 1.4736153153890208, "grad_norm": 0.05388210341334343, "learning_rate": 7.961418480428239e-06, "loss": 0.0022, "step": 90060 }, { "epoch": 1.4737789413400966, "grad_norm": 0.04056275635957718, "learning_rate": 7.960843197330765e-06, "loss": 0.0037, "step": 90070 }, { "epoch": 1.4739425672911723, "grad_norm": 0.07224096357822418, "learning_rate": 7.960267853864216e-06, "loss": 0.0014, "step": 90080 }, { "epoch": 1.4741061932422483, "grad_norm": 0.14966921508312225, "learning_rate": 7.959692450040322e-06, "loss": 0.0025, "step": 90090 }, { "epoch": 1.4742698191933241, "grad_norm": 0.08013840019702911, "learning_rate": 7.959116985870816e-06, "loss": 0.0026, "step": 90100 }, { "epoch": 1.4744334451444, "grad_norm": 0.02261880412697792, "learning_rate": 7.958541461367432e-06, "loss": 0.0008, "step": 90110 }, { "epoch": 1.4745970710954757, "grad_norm": 0.10631977766752243, "learning_rate": 7.957965876541905e-06, "loss": 0.0027, "step": 90120 }, { "epoch": 1.4747606970465517, "grad_norm": 0.0242417361587286, "learning_rate": 7.957390231405968e-06, "loss": 0.0015, "step": 90130 }, { "epoch": 1.4749243229976274, "grad_norm": 0.03378242254257202, "learning_rate": 7.95681452597136e-06, "loss": 0.0027, "step": 90140 }, { "epoch": 1.4750879489487032, "grad_norm": 0.11130809783935547, "learning_rate": 7.956238760249818e-06, "loss": 0.0023, "step": 90150 }, { "epoch": 1.475251574899779, "grad_norm": 0.1454525887966156, "learning_rate": 7.955662934253082e-06, "loss": 0.0015, "step": 90160 }, { "epoch": 1.475415200850855, "grad_norm": 0.04227616637945175, "learning_rate": 7.955087047992892e-06, "loss": 0.0015, "step": 90170 }, { "epoch": 1.4755788268019308, "grad_norm": 0.0649259090423584, "learning_rate": 7.954511101480991e-06, "loss": 0.0018, "step": 90180 }, { "epoch": 1.4757424527530065, "grad_norm": 0.08504209667444229, "learning_rate": 7.953935094729122e-06, "loss": 0.0014, "step": 90190 }, { "epoch": 1.4759060787040825, "grad_norm": 0.030054941773414612, "learning_rate": 7.953359027749028e-06, "loss": 0.0013, "step": 90200 }, { "epoch": 1.4760697046551583, "grad_norm": 0.08231766521930695, "learning_rate": 7.952782900552453e-06, "loss": 0.0019, "step": 90210 }, { "epoch": 1.476233330606234, "grad_norm": 0.04522545263171196, "learning_rate": 7.95220671315115e-06, "loss": 0.0018, "step": 90220 }, { "epoch": 1.47639695655731, "grad_norm": 0.06878785789012909, "learning_rate": 7.95163046555686e-06, "loss": 0.002, "step": 90230 }, { "epoch": 1.4765605825083858, "grad_norm": 0.2595647871494293, "learning_rate": 7.951054157781335e-06, "loss": 0.0028, "step": 90240 }, { "epoch": 1.4767242084594616, "grad_norm": 0.12276510149240494, "learning_rate": 7.950477789836326e-06, "loss": 0.0022, "step": 90250 }, { "epoch": 1.4768878344105376, "grad_norm": 0.06425286084413528, "learning_rate": 7.949901361733585e-06, "loss": 0.0024, "step": 90260 }, { "epoch": 1.4770514603616134, "grad_norm": 0.055714670568704605, "learning_rate": 7.949324873484864e-06, "loss": 0.0028, "step": 90270 }, { "epoch": 1.4772150863126892, "grad_norm": 0.061443667858839035, "learning_rate": 7.948748325101916e-06, "loss": 0.0027, "step": 90280 }, { "epoch": 1.4773787122637652, "grad_norm": 0.030612042173743248, "learning_rate": 7.948171716596497e-06, "loss": 0.0015, "step": 90290 }, { "epoch": 1.477542338214841, "grad_norm": 0.0893501415848732, "learning_rate": 7.947595047980367e-06, "loss": 0.0011, "step": 90300 }, { "epoch": 1.4777059641659167, "grad_norm": 0.09269126504659653, "learning_rate": 7.947018319265278e-06, "loss": 0.0014, "step": 90310 }, { "epoch": 1.4778695901169925, "grad_norm": 0.00485334312543273, "learning_rate": 7.946441530462994e-06, "loss": 0.0012, "step": 90320 }, { "epoch": 1.4780332160680685, "grad_norm": 0.10709493607282639, "learning_rate": 7.94586468158527e-06, "loss": 0.0015, "step": 90330 }, { "epoch": 1.4781968420191443, "grad_norm": 0.0681118294596672, "learning_rate": 7.945287772643872e-06, "loss": 0.002, "step": 90340 }, { "epoch": 1.47836046797022, "grad_norm": 0.03720761463046074, "learning_rate": 7.944710803650563e-06, "loss": 0.0017, "step": 90350 }, { "epoch": 1.4785240939212958, "grad_norm": 0.05549487844109535, "learning_rate": 7.944133774617105e-06, "loss": 0.0007, "step": 90360 }, { "epoch": 1.4786877198723718, "grad_norm": 0.03714008256793022, "learning_rate": 7.94355668555526e-06, "loss": 0.0045, "step": 90370 }, { "epoch": 1.4788513458234476, "grad_norm": 0.06434638798236847, "learning_rate": 7.9429795364768e-06, "loss": 0.0014, "step": 90380 }, { "epoch": 1.4790149717745233, "grad_norm": 0.1290341019630432, "learning_rate": 7.94240232739349e-06, "loss": 0.0009, "step": 90390 }, { "epoch": 1.4791785977255993, "grad_norm": 0.013750975020229816, "learning_rate": 7.941825058317098e-06, "loss": 0.0024, "step": 90400 }, { "epoch": 1.4793422236766751, "grad_norm": 0.04026090353727341, "learning_rate": 7.941247729259395e-06, "loss": 0.0015, "step": 90410 }, { "epoch": 1.479505849627751, "grad_norm": 0.11577226221561432, "learning_rate": 7.940670340232154e-06, "loss": 0.0029, "step": 90420 }, { "epoch": 1.4796694755788269, "grad_norm": 0.05530195310711861, "learning_rate": 7.940092891247144e-06, "loss": 0.0021, "step": 90430 }, { "epoch": 1.4798331015299027, "grad_norm": 0.059360258281230927, "learning_rate": 7.93951538231614e-06, "loss": 0.0014, "step": 90440 }, { "epoch": 1.4799967274809784, "grad_norm": 0.03255594149231911, "learning_rate": 7.938937813450919e-06, "loss": 0.0028, "step": 90450 }, { "epoch": 1.4801603534320544, "grad_norm": 0.14413660764694214, "learning_rate": 7.938360184663254e-06, "loss": 0.0017, "step": 90460 }, { "epoch": 1.4803239793831302, "grad_norm": 0.014871650375425816, "learning_rate": 7.937782495964926e-06, "loss": 0.001, "step": 90470 }, { "epoch": 1.480487605334206, "grad_norm": 0.028493380174040794, "learning_rate": 7.93720474736771e-06, "loss": 0.0019, "step": 90480 }, { "epoch": 1.480651231285282, "grad_norm": 0.04865841194987297, "learning_rate": 7.936626938883386e-06, "loss": 0.004, "step": 90490 }, { "epoch": 1.4808148572363578, "grad_norm": 0.16827498376369476, "learning_rate": 7.936049070523739e-06, "loss": 0.0014, "step": 90500 }, { "epoch": 1.4809784831874335, "grad_norm": 0.2540033161640167, "learning_rate": 7.935471142300545e-06, "loss": 0.001, "step": 90510 }, { "epoch": 1.4811421091385093, "grad_norm": 0.02196020632982254, "learning_rate": 7.934893154225594e-06, "loss": 0.001, "step": 90520 }, { "epoch": 1.481305735089585, "grad_norm": 0.019636673852801323, "learning_rate": 7.934315106310665e-06, "loss": 0.0016, "step": 90530 }, { "epoch": 1.481469361040661, "grad_norm": 0.0898507609963417, "learning_rate": 7.93373699856755e-06, "loss": 0.0014, "step": 90540 }, { "epoch": 1.4816329869917368, "grad_norm": 0.20606444776058197, "learning_rate": 7.93315883100803e-06, "loss": 0.003, "step": 90550 }, { "epoch": 1.4817966129428126, "grad_norm": 0.04809416085481644, "learning_rate": 7.932580603643896e-06, "loss": 0.0008, "step": 90560 }, { "epoch": 1.4819602388938886, "grad_norm": 0.06693452596664429, "learning_rate": 7.932002316486939e-06, "loss": 0.0021, "step": 90570 }, { "epoch": 1.4821238648449644, "grad_norm": 0.04150101915001869, "learning_rate": 7.931423969548948e-06, "loss": 0.0028, "step": 90580 }, { "epoch": 1.4822874907960402, "grad_norm": 0.030287450179457664, "learning_rate": 7.930845562841713e-06, "loss": 0.0012, "step": 90590 }, { "epoch": 1.4824511167471162, "grad_norm": 0.1366679072380066, "learning_rate": 7.930267096377032e-06, "loss": 0.0017, "step": 90600 }, { "epoch": 1.482614742698192, "grad_norm": 0.06565532833337784, "learning_rate": 7.929688570166696e-06, "loss": 0.001, "step": 90610 }, { "epoch": 1.4827783686492677, "grad_norm": 0.02880273200571537, "learning_rate": 7.929109984222503e-06, "loss": 0.0015, "step": 90620 }, { "epoch": 1.4829419946003437, "grad_norm": 0.04726841673254967, "learning_rate": 7.928531338556247e-06, "loss": 0.0019, "step": 90630 }, { "epoch": 1.4831056205514195, "grad_norm": 0.042054999619722366, "learning_rate": 7.927952633179728e-06, "loss": 0.0024, "step": 90640 }, { "epoch": 1.4832692465024953, "grad_norm": 0.049199867993593216, "learning_rate": 7.927373868104746e-06, "loss": 0.0015, "step": 90650 }, { "epoch": 1.4834328724535712, "grad_norm": 0.04032820835709572, "learning_rate": 7.926795043343099e-06, "loss": 0.0017, "step": 90660 }, { "epoch": 1.483596498404647, "grad_norm": 0.1940896213054657, "learning_rate": 7.92621615890659e-06, "loss": 0.0009, "step": 90670 }, { "epoch": 1.4837601243557228, "grad_norm": 0.029323885217308998, "learning_rate": 7.925637214807026e-06, "loss": 0.0012, "step": 90680 }, { "epoch": 1.4839237503067988, "grad_norm": 0.06520579755306244, "learning_rate": 7.925058211056202e-06, "loss": 0.0012, "step": 90690 }, { "epoch": 1.4840873762578746, "grad_norm": 0.04360515996813774, "learning_rate": 7.924479147665931e-06, "loss": 0.0009, "step": 90700 }, { "epoch": 1.4842510022089503, "grad_norm": 0.045208100229501724, "learning_rate": 7.923900024648017e-06, "loss": 0.0032, "step": 90710 }, { "epoch": 1.4844146281600261, "grad_norm": 0.06877396255731583, "learning_rate": 7.92332084201427e-06, "loss": 0.0012, "step": 90720 }, { "epoch": 1.484578254111102, "grad_norm": 0.08591903746128082, "learning_rate": 7.922741599776495e-06, "loss": 0.0015, "step": 90730 }, { "epoch": 1.4847418800621779, "grad_norm": 0.06813142448663712, "learning_rate": 7.922162297946507e-06, "loss": 0.0017, "step": 90740 }, { "epoch": 1.4849055060132537, "grad_norm": 0.03952178359031677, "learning_rate": 7.921582936536113e-06, "loss": 0.001, "step": 90750 }, { "epoch": 1.4850691319643294, "grad_norm": 0.04571244493126869, "learning_rate": 7.92100351555713e-06, "loss": 0.002, "step": 90760 }, { "epoch": 1.4852327579154054, "grad_norm": 0.033422328531742096, "learning_rate": 7.920424035021365e-06, "loss": 0.002, "step": 90770 }, { "epoch": 1.4853963838664812, "grad_norm": 0.08507746458053589, "learning_rate": 7.91984449494064e-06, "loss": 0.0014, "step": 90780 }, { "epoch": 1.485560009817557, "grad_norm": 0.007751322817057371, "learning_rate": 7.91926489532677e-06, "loss": 0.003, "step": 90790 }, { "epoch": 1.485723635768633, "grad_norm": 0.18288201093673706, "learning_rate": 7.91868523619157e-06, "loss": 0.0033, "step": 90800 }, { "epoch": 1.4858872617197088, "grad_norm": 0.070297010242939, "learning_rate": 7.918105517546861e-06, "loss": 0.0023, "step": 90810 }, { "epoch": 1.4860508876707845, "grad_norm": 0.08091577142477036, "learning_rate": 7.917525739404464e-06, "loss": 0.0018, "step": 90820 }, { "epoch": 1.4862145136218605, "grad_norm": 0.046729736030101776, "learning_rate": 7.916945901776195e-06, "loss": 0.0008, "step": 90830 }, { "epoch": 1.4863781395729363, "grad_norm": 0.04366743192076683, "learning_rate": 7.916366004673882e-06, "loss": 0.0019, "step": 90840 }, { "epoch": 1.486541765524012, "grad_norm": 0.04125697538256645, "learning_rate": 7.915786048109347e-06, "loss": 0.0011, "step": 90850 }, { "epoch": 1.486705391475088, "grad_norm": 0.0675581619143486, "learning_rate": 7.915206032094412e-06, "loss": 0.0015, "step": 90860 }, { "epoch": 1.4868690174261638, "grad_norm": 0.0376104861497879, "learning_rate": 7.914625956640907e-06, "loss": 0.0011, "step": 90870 }, { "epoch": 1.4870326433772396, "grad_norm": 0.029858112335205078, "learning_rate": 7.914045821760658e-06, "loss": 0.0011, "step": 90880 }, { "epoch": 1.4871962693283154, "grad_norm": 0.1019463911652565, "learning_rate": 7.91346562746549e-06, "loss": 0.0018, "step": 90890 }, { "epoch": 1.4873598952793914, "grad_norm": 0.07736611366271973, "learning_rate": 7.912885373767238e-06, "loss": 0.0012, "step": 90900 }, { "epoch": 1.4875235212304672, "grad_norm": 0.06272589415311813, "learning_rate": 7.912305060677731e-06, "loss": 0.0011, "step": 90910 }, { "epoch": 1.487687147181543, "grad_norm": 0.1118847206234932, "learning_rate": 7.911724688208801e-06, "loss": 0.0025, "step": 90920 }, { "epoch": 1.4878507731326187, "grad_norm": 0.05664315074682236, "learning_rate": 7.91114425637228e-06, "loss": 0.002, "step": 90930 }, { "epoch": 1.4880143990836947, "grad_norm": 0.07823187112808228, "learning_rate": 7.910563765180002e-06, "loss": 0.0011, "step": 90940 }, { "epoch": 1.4881780250347705, "grad_norm": 0.06416172534227371, "learning_rate": 7.909983214643806e-06, "loss": 0.0013, "step": 90950 }, { "epoch": 1.4883416509858463, "grad_norm": 0.07697953283786774, "learning_rate": 7.909402604775527e-06, "loss": 0.0014, "step": 90960 }, { "epoch": 1.4885052769369223, "grad_norm": 0.10316766053438187, "learning_rate": 7.908821935587003e-06, "loss": 0.0016, "step": 90970 }, { "epoch": 1.488668902887998, "grad_norm": 0.04428387060761452, "learning_rate": 7.908241207090073e-06, "loss": 0.0009, "step": 90980 }, { "epoch": 1.4888325288390738, "grad_norm": 0.01470914576202631, "learning_rate": 7.907660419296579e-06, "loss": 0.0013, "step": 90990 }, { "epoch": 1.4889961547901498, "grad_norm": 0.06591139733791351, "learning_rate": 7.907079572218361e-06, "loss": 0.0009, "step": 91000 }, { "epoch": 1.4891597807412256, "grad_norm": 0.13105107843875885, "learning_rate": 7.906498665867264e-06, "loss": 0.0009, "step": 91010 }, { "epoch": 1.4893234066923013, "grad_norm": 0.05364183709025383, "learning_rate": 7.90591770025513e-06, "loss": 0.0021, "step": 91020 }, { "epoch": 1.4894870326433773, "grad_norm": 0.0507393553853035, "learning_rate": 7.905336675393806e-06, "loss": 0.0012, "step": 91030 }, { "epoch": 1.4896506585944531, "grad_norm": 0.2137073427438736, "learning_rate": 7.90475559129514e-06, "loss": 0.0021, "step": 91040 }, { "epoch": 1.4898142845455289, "grad_norm": 0.0665067806839943, "learning_rate": 7.904174447970976e-06, "loss": 0.0011, "step": 91050 }, { "epoch": 1.4899779104966049, "grad_norm": 0.010415741242468357, "learning_rate": 7.903593245433162e-06, "loss": 0.0014, "step": 91060 }, { "epoch": 1.4901415364476807, "grad_norm": 0.053889065980911255, "learning_rate": 7.903011983693555e-06, "loss": 0.0019, "step": 91070 }, { "epoch": 1.4903051623987564, "grad_norm": 0.09556637704372406, "learning_rate": 7.902430662764002e-06, "loss": 0.0009, "step": 91080 }, { "epoch": 1.4904687883498322, "grad_norm": 0.06805505603551865, "learning_rate": 7.901849282656357e-06, "loss": 0.0019, "step": 91090 }, { "epoch": 1.4906324143009082, "grad_norm": 0.06040697172284126, "learning_rate": 7.901267843382472e-06, "loss": 0.0009, "step": 91100 }, { "epoch": 1.490796040251984, "grad_norm": 0.013686120510101318, "learning_rate": 7.900686344954202e-06, "loss": 0.0018, "step": 91110 }, { "epoch": 1.4909596662030598, "grad_norm": 0.032405175268650055, "learning_rate": 7.900104787383407e-06, "loss": 0.001, "step": 91120 }, { "epoch": 1.4911232921541355, "grad_norm": 0.050149597227573395, "learning_rate": 7.899523170681939e-06, "loss": 0.0017, "step": 91130 }, { "epoch": 1.4912869181052115, "grad_norm": 0.07034562528133392, "learning_rate": 7.898941494861661e-06, "loss": 0.0026, "step": 91140 }, { "epoch": 1.4914505440562873, "grad_norm": 0.008503036573529243, "learning_rate": 7.898359759934431e-06, "loss": 0.0035, "step": 91150 }, { "epoch": 1.491614170007363, "grad_norm": 0.1635364294052124, "learning_rate": 7.89777796591211e-06, "loss": 0.002, "step": 91160 }, { "epoch": 1.491777795958439, "grad_norm": 0.40166687965393066, "learning_rate": 7.897196112806562e-06, "loss": 0.0038, "step": 91170 }, { "epoch": 1.4919414219095148, "grad_norm": 0.031975600868463516, "learning_rate": 7.896614200629648e-06, "loss": 0.0014, "step": 91180 }, { "epoch": 1.4921050478605906, "grad_norm": 0.03573009744286537, "learning_rate": 7.896032229393235e-06, "loss": 0.0012, "step": 91190 }, { "epoch": 1.4922686738116666, "grad_norm": 0.04101209715008736, "learning_rate": 7.895450199109186e-06, "loss": 0.0016, "step": 91200 }, { "epoch": 1.4924322997627424, "grad_norm": 0.006261303089559078, "learning_rate": 7.894868109789374e-06, "loss": 0.0016, "step": 91210 }, { "epoch": 1.4925959257138182, "grad_norm": 0.028075747191905975, "learning_rate": 7.89428596144566e-06, "loss": 0.0013, "step": 91220 }, { "epoch": 1.4927595516648942, "grad_norm": 0.0374009944498539, "learning_rate": 7.893703754089916e-06, "loss": 0.002, "step": 91230 }, { "epoch": 1.49292317761597, "grad_norm": 0.012723361141979694, "learning_rate": 7.893121487734015e-06, "loss": 0.0016, "step": 91240 }, { "epoch": 1.4930868035670457, "grad_norm": 0.10167280584573746, "learning_rate": 7.892539162389827e-06, "loss": 0.0013, "step": 91250 }, { "epoch": 1.4932504295181217, "grad_norm": 0.043277669697999954, "learning_rate": 7.891956778069227e-06, "loss": 0.0016, "step": 91260 }, { "epoch": 1.4934140554691975, "grad_norm": 0.08927137404680252, "learning_rate": 7.891374334784085e-06, "loss": 0.0024, "step": 91270 }, { "epoch": 1.4935776814202733, "grad_norm": 0.02100794017314911, "learning_rate": 7.890791832546281e-06, "loss": 0.002, "step": 91280 }, { "epoch": 1.493741307371349, "grad_norm": 0.02814546413719654, "learning_rate": 7.89020927136769e-06, "loss": 0.002, "step": 91290 }, { "epoch": 1.4939049333224248, "grad_norm": 0.06517739593982697, "learning_rate": 7.88962665126019e-06, "loss": 0.0015, "step": 91300 }, { "epoch": 1.4940685592735008, "grad_norm": 0.042362675070762634, "learning_rate": 7.88904397223566e-06, "loss": 0.0015, "step": 91310 }, { "epoch": 1.4942321852245766, "grad_norm": 0.026513269171118736, "learning_rate": 7.88846123430598e-06, "loss": 0.0006, "step": 91320 }, { "epoch": 1.4943958111756523, "grad_norm": 0.056038063019514084, "learning_rate": 7.887878437483033e-06, "loss": 0.0017, "step": 91330 }, { "epoch": 1.4945594371267283, "grad_norm": 0.09990955144166946, "learning_rate": 7.8872955817787e-06, "loss": 0.0027, "step": 91340 }, { "epoch": 1.4947230630778041, "grad_norm": 0.09063870459794998, "learning_rate": 7.886712667204866e-06, "loss": 0.0011, "step": 91350 }, { "epoch": 1.49488668902888, "grad_norm": 0.06189780682325363, "learning_rate": 7.886129693773416e-06, "loss": 0.0009, "step": 91360 }, { "epoch": 1.4950503149799559, "grad_norm": 0.20324236154556274, "learning_rate": 7.885546661496235e-06, "loss": 0.0015, "step": 91370 }, { "epoch": 1.4952139409310317, "grad_norm": 0.15218515694141388, "learning_rate": 7.884963570385214e-06, "loss": 0.0015, "step": 91380 }, { "epoch": 1.4953775668821074, "grad_norm": 0.0644054040312767, "learning_rate": 7.884380420452236e-06, "loss": 0.0013, "step": 91390 }, { "epoch": 1.4955411928331834, "grad_norm": 0.016700852662324905, "learning_rate": 7.883797211709196e-06, "loss": 0.0019, "step": 91400 }, { "epoch": 1.4957048187842592, "grad_norm": 0.04524700716137886, "learning_rate": 7.883213944167984e-06, "loss": 0.0016, "step": 91410 }, { "epoch": 1.495868444735335, "grad_norm": 0.1819811910390854, "learning_rate": 7.88263061784049e-06, "loss": 0.0014, "step": 91420 }, { "epoch": 1.496032070686411, "grad_norm": 0.06974012404680252, "learning_rate": 7.882047232738611e-06, "loss": 0.0015, "step": 91430 }, { "epoch": 1.4961956966374867, "grad_norm": 0.37790051102638245, "learning_rate": 7.88146378887424e-06, "loss": 0.0028, "step": 91440 }, { "epoch": 1.4963593225885625, "grad_norm": 0.07222484797239304, "learning_rate": 7.880880286259272e-06, "loss": 0.0014, "step": 91450 }, { "epoch": 1.4965229485396385, "grad_norm": 0.16387520730495453, "learning_rate": 7.880296724905605e-06, "loss": 0.0018, "step": 91460 }, { "epoch": 1.4966865744907143, "grad_norm": 0.0642009750008583, "learning_rate": 7.879713104825139e-06, "loss": 0.0013, "step": 91470 }, { "epoch": 1.49685020044179, "grad_norm": 0.024794984608888626, "learning_rate": 7.879129426029771e-06, "loss": 0.0011, "step": 91480 }, { "epoch": 1.4970138263928658, "grad_norm": 0.10316440463066101, "learning_rate": 7.878545688531401e-06, "loss": 0.002, "step": 91490 }, { "epoch": 1.4971774523439416, "grad_norm": 0.09527518600225449, "learning_rate": 7.877961892341934e-06, "loss": 0.0028, "step": 91500 }, { "epoch": 1.4973410782950176, "grad_norm": 0.02827058546245098, "learning_rate": 7.877378037473271e-06, "loss": 0.0017, "step": 91510 }, { "epoch": 1.4975047042460934, "grad_norm": 0.09212218970060349, "learning_rate": 7.876794123937318e-06, "loss": 0.0027, "step": 91520 }, { "epoch": 1.4976683301971692, "grad_norm": 0.021330492570996284, "learning_rate": 7.87621015174598e-06, "loss": 0.0017, "step": 91530 }, { "epoch": 1.4978319561482452, "grad_norm": 0.0159298088401556, "learning_rate": 7.875626120911162e-06, "loss": 0.0017, "step": 91540 }, { "epoch": 1.497995582099321, "grad_norm": 0.08136018365621567, "learning_rate": 7.875042031444773e-06, "loss": 0.0025, "step": 91550 }, { "epoch": 1.4981592080503967, "grad_norm": 0.05409404635429382, "learning_rate": 7.874457883358722e-06, "loss": 0.0011, "step": 91560 }, { "epoch": 1.4983228340014727, "grad_norm": 0.08611982315778732, "learning_rate": 7.873873676664918e-06, "loss": 0.0018, "step": 91570 }, { "epoch": 1.4984864599525485, "grad_norm": 0.04431384429335594, "learning_rate": 7.873289411375276e-06, "loss": 0.0016, "step": 91580 }, { "epoch": 1.4986500859036243, "grad_norm": 0.18684428930282593, "learning_rate": 7.872705087501706e-06, "loss": 0.0027, "step": 91590 }, { "epoch": 1.4988137118547002, "grad_norm": 0.03789890930056572, "learning_rate": 7.872120705056124e-06, "loss": 0.0013, "step": 91600 }, { "epoch": 1.498977337805776, "grad_norm": 0.06741201877593994, "learning_rate": 7.871536264050442e-06, "loss": 0.0018, "step": 91610 }, { "epoch": 1.4991409637568518, "grad_norm": 0.058735623955726624, "learning_rate": 7.870951764496575e-06, "loss": 0.0013, "step": 91620 }, { "epoch": 1.4993045897079278, "grad_norm": 0.0502336286008358, "learning_rate": 7.870367206406447e-06, "loss": 0.0009, "step": 91630 }, { "epoch": 1.4994682156590036, "grad_norm": 0.09769315272569656, "learning_rate": 7.869782589791971e-06, "loss": 0.0013, "step": 91640 }, { "epoch": 1.4996318416100793, "grad_norm": 0.12105381488800049, "learning_rate": 7.86919791466507e-06, "loss": 0.0014, "step": 91650 }, { "epoch": 1.4997954675611553, "grad_norm": 0.02927481010556221, "learning_rate": 7.868613181037663e-06, "loss": 0.0016, "step": 91660 }, { "epoch": 1.4999590935122311, "grad_norm": 0.06138233467936516, "learning_rate": 7.868028388921673e-06, "loss": 0.0014, "step": 91670 }, { "epoch": 1.5001227194633069, "grad_norm": 0.04463763162493706, "learning_rate": 7.867443538329023e-06, "loss": 0.0017, "step": 91680 }, { "epoch": 1.5002863454143829, "grad_norm": 0.043669939041137695, "learning_rate": 7.866858629271638e-06, "loss": 0.0015, "step": 91690 }, { "epoch": 1.5004499713654584, "grad_norm": 0.03356056660413742, "learning_rate": 7.866273661761445e-06, "loss": 0.0008, "step": 91700 }, { "epoch": 1.5006135973165344, "grad_norm": 0.020143533125519753, "learning_rate": 7.865688635810368e-06, "loss": 0.0017, "step": 91710 }, { "epoch": 1.5007772232676102, "grad_norm": 0.004381553735584021, "learning_rate": 7.865103551430338e-06, "loss": 0.0011, "step": 91720 }, { "epoch": 1.500940849218686, "grad_norm": 0.1291174590587616, "learning_rate": 7.864518408633281e-06, "loss": 0.0013, "step": 91730 }, { "epoch": 1.501104475169762, "grad_norm": 0.026398897171020508, "learning_rate": 7.863933207431133e-06, "loss": 0.0011, "step": 91740 }, { "epoch": 1.5012681011208378, "grad_norm": 0.050252556800842285, "learning_rate": 7.86334794783582e-06, "loss": 0.0023, "step": 91750 }, { "epoch": 1.5014317270719135, "grad_norm": 0.14102870225906372, "learning_rate": 7.86276262985928e-06, "loss": 0.0023, "step": 91760 }, { "epoch": 1.5015953530229895, "grad_norm": 0.010733316652476788, "learning_rate": 7.862177253513444e-06, "loss": 0.0035, "step": 91770 }, { "epoch": 1.5017589789740653, "grad_norm": 0.05353912338614464, "learning_rate": 7.861591818810246e-06, "loss": 0.0012, "step": 91780 }, { "epoch": 1.501922604925141, "grad_norm": 0.023169174790382385, "learning_rate": 7.861006325761627e-06, "loss": 0.0024, "step": 91790 }, { "epoch": 1.502086230876217, "grad_norm": 0.04764796420931816, "learning_rate": 7.86042077437952e-06, "loss": 0.0009, "step": 91800 }, { "epoch": 1.5022498568272928, "grad_norm": 0.05868222564458847, "learning_rate": 7.859835164675866e-06, "loss": 0.0012, "step": 91810 }, { "epoch": 1.5024134827783686, "grad_norm": 0.10006286948919296, "learning_rate": 7.859249496662607e-06, "loss": 0.0032, "step": 91820 }, { "epoch": 1.5025771087294446, "grad_norm": 0.06078581511974335, "learning_rate": 7.858663770351682e-06, "loss": 0.0015, "step": 91830 }, { "epoch": 1.5027407346805204, "grad_norm": 0.03602749854326248, "learning_rate": 7.858077985755034e-06, "loss": 0.0018, "step": 91840 }, { "epoch": 1.5029043606315962, "grad_norm": 0.06645442545413971, "learning_rate": 7.857492142884606e-06, "loss": 0.0018, "step": 91850 }, { "epoch": 1.5030679865826722, "grad_norm": 0.011145813390612602, "learning_rate": 7.856906241752345e-06, "loss": 0.0012, "step": 91860 }, { "epoch": 1.5032316125337477, "grad_norm": 0.011002888903021812, "learning_rate": 7.856320282370194e-06, "loss": 0.0011, "step": 91870 }, { "epoch": 1.5033952384848237, "grad_norm": 0.09825784713029861, "learning_rate": 7.855734264750102e-06, "loss": 0.0017, "step": 91880 }, { "epoch": 1.5035588644358997, "grad_norm": 0.031673699617385864, "learning_rate": 7.855148188904016e-06, "loss": 0.0008, "step": 91890 }, { "epoch": 1.5037224903869753, "grad_norm": 0.02828347682952881, "learning_rate": 7.854562054843888e-06, "loss": 0.0014, "step": 91900 }, { "epoch": 1.5038861163380512, "grad_norm": 0.20610825717449188, "learning_rate": 7.853975862581666e-06, "loss": 0.001, "step": 91910 }, { "epoch": 1.504049742289127, "grad_norm": 0.07650528848171234, "learning_rate": 7.853389612129305e-06, "loss": 0.0023, "step": 91920 }, { "epoch": 1.5042133682402028, "grad_norm": 0.041111357510089874, "learning_rate": 7.852803303498756e-06, "loss": 0.0015, "step": 91930 }, { "epoch": 1.5043769941912788, "grad_norm": 0.0029548094607889652, "learning_rate": 7.852216936701972e-06, "loss": 0.0018, "step": 91940 }, { "epoch": 1.5045406201423546, "grad_norm": 0.032725874334573746, "learning_rate": 7.851630511750915e-06, "loss": 0.0016, "step": 91950 }, { "epoch": 1.5047042460934303, "grad_norm": 0.05136667564511299, "learning_rate": 7.851044028657532e-06, "loss": 0.0007, "step": 91960 }, { "epoch": 1.5048678720445063, "grad_norm": 0.030387381091713905, "learning_rate": 7.85045748743379e-06, "loss": 0.0009, "step": 91970 }, { "epoch": 1.5050314979955821, "grad_norm": 0.13217230141162872, "learning_rate": 7.849870888091642e-06, "loss": 0.0014, "step": 91980 }, { "epoch": 1.5051951239466579, "grad_norm": 0.06238849461078644, "learning_rate": 7.84928423064305e-06, "loss": 0.002, "step": 91990 }, { "epoch": 1.5053587498977339, "grad_norm": 0.09066459536552429, "learning_rate": 7.848697515099976e-06, "loss": 0.0022, "step": 92000 }, { "epoch": 1.5055223758488097, "grad_norm": 0.05727618187665939, "learning_rate": 7.848110741474383e-06, "loss": 0.0014, "step": 92010 }, { "epoch": 1.5056860017998854, "grad_norm": 0.0155140720307827, "learning_rate": 7.847523909778234e-06, "loss": 0.0009, "step": 92020 }, { "epoch": 1.5058496277509614, "grad_norm": 0.051895298063755035, "learning_rate": 7.846937020023495e-06, "loss": 0.0013, "step": 92030 }, { "epoch": 1.506013253702037, "grad_norm": 0.038460761308670044, "learning_rate": 7.84635007222213e-06, "loss": 0.003, "step": 92040 }, { "epoch": 1.506176879653113, "grad_norm": 0.025319106876850128, "learning_rate": 7.845763066386108e-06, "loss": 0.0009, "step": 92050 }, { "epoch": 1.506340505604189, "grad_norm": 0.1095888540148735, "learning_rate": 7.845176002527399e-06, "loss": 0.0016, "step": 92060 }, { "epoch": 1.5065041315552645, "grad_norm": 0.004712133668363094, "learning_rate": 7.844588880657968e-06, "loss": 0.0011, "step": 92070 }, { "epoch": 1.5066677575063405, "grad_norm": 0.03035750426352024, "learning_rate": 7.844001700789791e-06, "loss": 0.0015, "step": 92080 }, { "epoch": 1.5068313834574163, "grad_norm": 0.09898818284273148, "learning_rate": 7.843414462934838e-06, "loss": 0.0032, "step": 92090 }, { "epoch": 1.506995009408492, "grad_norm": 0.03697250038385391, "learning_rate": 7.842827167105083e-06, "loss": 0.0025, "step": 92100 }, { "epoch": 1.507158635359568, "grad_norm": 0.1550431102514267, "learning_rate": 7.842239813312497e-06, "loss": 0.0018, "step": 92110 }, { "epoch": 1.5073222613106438, "grad_norm": 0.017590712755918503, "learning_rate": 7.841652401569062e-06, "loss": 0.0007, "step": 92120 }, { "epoch": 1.5074858872617196, "grad_norm": 0.08404555171728134, "learning_rate": 7.841064931886748e-06, "loss": 0.0018, "step": 92130 }, { "epoch": 1.5076495132127956, "grad_norm": 0.03656144440174103, "learning_rate": 7.84047740427754e-06, "loss": 0.0017, "step": 92140 }, { "epoch": 1.5078131391638714, "grad_norm": 0.07455088943243027, "learning_rate": 7.83988981875341e-06, "loss": 0.0015, "step": 92150 }, { "epoch": 1.5079767651149472, "grad_norm": 0.034647755324840546, "learning_rate": 7.839302175326344e-06, "loss": 0.0013, "step": 92160 }, { "epoch": 1.5081403910660232, "grad_norm": 0.04392668977379799, "learning_rate": 7.83871447400832e-06, "loss": 0.0015, "step": 92170 }, { "epoch": 1.508304017017099, "grad_norm": 0.019125517457723618, "learning_rate": 7.838126714811323e-06, "loss": 0.0018, "step": 92180 }, { "epoch": 1.5084676429681747, "grad_norm": 0.07586568593978882, "learning_rate": 7.837538897747334e-06, "loss": 0.0015, "step": 92190 }, { "epoch": 1.5086312689192507, "grad_norm": 0.014317124150693417, "learning_rate": 7.836951022828343e-06, "loss": 0.0015, "step": 92200 }, { "epoch": 1.5087948948703265, "grad_norm": 0.03085220605134964, "learning_rate": 7.836363090066332e-06, "loss": 0.0016, "step": 92210 }, { "epoch": 1.5089585208214022, "grad_norm": 0.08095445483922958, "learning_rate": 7.83577509947329e-06, "loss": 0.0018, "step": 92220 }, { "epoch": 1.5091221467724782, "grad_norm": 0.07258553057909012, "learning_rate": 7.835187051061206e-06, "loss": 0.0016, "step": 92230 }, { "epoch": 1.5092857727235538, "grad_norm": 0.05398806557059288, "learning_rate": 7.834598944842068e-06, "loss": 0.001, "step": 92240 }, { "epoch": 1.5094493986746298, "grad_norm": 0.0666961818933487, "learning_rate": 7.83401078082787e-06, "loss": 0.0031, "step": 92250 }, { "epoch": 1.5096130246257058, "grad_norm": 0.07757312059402466, "learning_rate": 7.833422559030602e-06, "loss": 0.002, "step": 92260 }, { "epoch": 1.5097766505767813, "grad_norm": 0.07760588079690933, "learning_rate": 7.832834279462257e-06, "loss": 0.0032, "step": 92270 }, { "epoch": 1.5099402765278573, "grad_norm": 0.11233413219451904, "learning_rate": 7.83224594213483e-06, "loss": 0.0021, "step": 92280 }, { "epoch": 1.5101039024789331, "grad_norm": 0.04756970703601837, "learning_rate": 7.831657547060316e-06, "loss": 0.0017, "step": 92290 }, { "epoch": 1.5102675284300089, "grad_norm": 0.039770424365997314, "learning_rate": 7.831069094250715e-06, "loss": 0.0008, "step": 92300 }, { "epoch": 1.5104311543810849, "grad_norm": 0.12664794921875, "learning_rate": 7.830480583718022e-06, "loss": 0.0025, "step": 92310 }, { "epoch": 1.5105947803321607, "grad_norm": 0.05072101578116417, "learning_rate": 7.829892015474237e-06, "loss": 0.0023, "step": 92320 }, { "epoch": 1.5107584062832364, "grad_norm": 0.02968592569231987, "learning_rate": 7.829303389531362e-06, "loss": 0.0011, "step": 92330 }, { "epoch": 1.5109220322343124, "grad_norm": 0.21015779674053192, "learning_rate": 7.828714705901395e-06, "loss": 0.0018, "step": 92340 }, { "epoch": 1.5110856581853882, "grad_norm": 0.055915310978889465, "learning_rate": 7.828125964596342e-06, "loss": 0.0021, "step": 92350 }, { "epoch": 1.511249284136464, "grad_norm": 0.05129760876297951, "learning_rate": 7.827537165628206e-06, "loss": 0.0013, "step": 92360 }, { "epoch": 1.51141291008754, "grad_norm": 0.04591250792145729, "learning_rate": 7.826948309008992e-06, "loss": 0.0014, "step": 92370 }, { "epoch": 1.5115765360386157, "grad_norm": 0.09412539005279541, "learning_rate": 7.826359394750706e-06, "loss": 0.0015, "step": 92380 }, { "epoch": 1.5117401619896915, "grad_norm": 0.04553578421473503, "learning_rate": 7.825770422865354e-06, "loss": 0.0014, "step": 92390 }, { "epoch": 1.5119037879407675, "grad_norm": 0.06481429189443588, "learning_rate": 7.825181393364949e-06, "loss": 0.0014, "step": 92400 }, { "epoch": 1.5120674138918433, "grad_norm": 0.12191089987754822, "learning_rate": 7.824592306261497e-06, "loss": 0.0025, "step": 92410 }, { "epoch": 1.512231039842919, "grad_norm": 0.03174940124154091, "learning_rate": 7.82400316156701e-06, "loss": 0.0008, "step": 92420 }, { "epoch": 1.512394665793995, "grad_norm": 0.056936342269182205, "learning_rate": 7.8234139592935e-06, "loss": 0.0013, "step": 92430 }, { "epoch": 1.5125582917450706, "grad_norm": 0.02676558680832386, "learning_rate": 7.82282469945298e-06, "loss": 0.0029, "step": 92440 }, { "epoch": 1.5127219176961466, "grad_norm": 0.0403161495923996, "learning_rate": 7.822235382057467e-06, "loss": 0.0013, "step": 92450 }, { "epoch": 1.5128855436472226, "grad_norm": 0.062493517994880676, "learning_rate": 7.821646007118974e-06, "loss": 0.0011, "step": 92460 }, { "epoch": 1.5130491695982982, "grad_norm": 0.10197070240974426, "learning_rate": 7.821056574649518e-06, "loss": 0.0019, "step": 92470 }, { "epoch": 1.5132127955493742, "grad_norm": 0.11486933380365372, "learning_rate": 7.820467084661118e-06, "loss": 0.0014, "step": 92480 }, { "epoch": 1.51337642150045, "grad_norm": 0.02587779425084591, "learning_rate": 7.819877537165793e-06, "loss": 0.0019, "step": 92490 }, { "epoch": 1.5135400474515257, "grad_norm": 0.0320187546312809, "learning_rate": 7.819287932175563e-06, "loss": 0.0035, "step": 92500 }, { "epoch": 1.5137036734026017, "grad_norm": 0.057902608066797256, "learning_rate": 7.818698269702451e-06, "loss": 0.0017, "step": 92510 }, { "epoch": 1.5138672993536775, "grad_norm": 0.09161176532506943, "learning_rate": 7.818108549758477e-06, "loss": 0.0025, "step": 92520 }, { "epoch": 1.5140309253047533, "grad_norm": 0.08542528003454208, "learning_rate": 7.817518772355667e-06, "loss": 0.001, "step": 92530 }, { "epoch": 1.5141945512558292, "grad_norm": 0.044841304421424866, "learning_rate": 7.816928937506045e-06, "loss": 0.0066, "step": 92540 }, { "epoch": 1.514358177206905, "grad_norm": 0.0988459438085556, "learning_rate": 7.816339045221637e-06, "loss": 0.0011, "step": 92550 }, { "epoch": 1.5145218031579808, "grad_norm": 0.059487320482730865, "learning_rate": 7.815749095514474e-06, "loss": 0.0084, "step": 92560 }, { "epoch": 1.5146854291090568, "grad_norm": 0.022597843781113625, "learning_rate": 7.815159088396579e-06, "loss": 0.0021, "step": 92570 }, { "epoch": 1.5148490550601326, "grad_norm": 0.08298558741807938, "learning_rate": 7.814569023879985e-06, "loss": 0.003, "step": 92580 }, { "epoch": 1.5150126810112083, "grad_norm": 0.053207531571388245, "learning_rate": 7.813978901976722e-06, "loss": 0.0015, "step": 92590 }, { "epoch": 1.5151763069622843, "grad_norm": 0.039539653807878494, "learning_rate": 7.813388722698823e-06, "loss": 0.001, "step": 92600 }, { "epoch": 1.51533993291336, "grad_norm": 0.04759596660733223, "learning_rate": 7.81279848605832e-06, "loss": 0.0022, "step": 92610 }, { "epoch": 1.5155035588644359, "grad_norm": 0.039984021335840225, "learning_rate": 7.81220819206725e-06, "loss": 0.0021, "step": 92620 }, { "epoch": 1.5156671848155119, "grad_norm": 0.02409449592232704, "learning_rate": 7.811617840737644e-06, "loss": 0.0012, "step": 92630 }, { "epoch": 1.5158308107665874, "grad_norm": 0.1203232929110527, "learning_rate": 7.811027432081543e-06, "loss": 0.0019, "step": 92640 }, { "epoch": 1.5159944367176634, "grad_norm": 0.011016885749995708, "learning_rate": 7.810436966110983e-06, "loss": 0.0017, "step": 92650 }, { "epoch": 1.5161580626687394, "grad_norm": 0.060002490878105164, "learning_rate": 7.809846442838003e-06, "loss": 0.002, "step": 92660 }, { "epoch": 1.516321688619815, "grad_norm": 0.10056222975254059, "learning_rate": 7.809255862274645e-06, "loss": 0.002, "step": 92670 }, { "epoch": 1.516485314570891, "grad_norm": 0.035996317863464355, "learning_rate": 7.808665224432948e-06, "loss": 0.0009, "step": 92680 }, { "epoch": 1.5166489405219667, "grad_norm": 0.05650579556822777, "learning_rate": 7.808074529324956e-06, "loss": 0.0019, "step": 92690 }, { "epoch": 1.5168125664730425, "grad_norm": 0.11553927510976791, "learning_rate": 7.807483776962712e-06, "loss": 0.0012, "step": 92700 }, { "epoch": 1.5169761924241185, "grad_norm": 0.18173199892044067, "learning_rate": 7.806892967358263e-06, "loss": 0.0029, "step": 92710 }, { "epoch": 1.5171398183751943, "grad_norm": 0.05138034373521805, "learning_rate": 7.806302100523653e-06, "loss": 0.001, "step": 92720 }, { "epoch": 1.51730344432627, "grad_norm": 0.1495794951915741, "learning_rate": 7.80571117647093e-06, "loss": 0.0026, "step": 92730 }, { "epoch": 1.517467070277346, "grad_norm": 0.1600273698568344, "learning_rate": 7.80512019521214e-06, "loss": 0.0017, "step": 92740 }, { "epoch": 1.5176306962284218, "grad_norm": 0.07083771377801895, "learning_rate": 7.804529156759338e-06, "loss": 0.001, "step": 92750 }, { "epoch": 1.5177943221794976, "grad_norm": 0.015001297928392887, "learning_rate": 7.80393806112457e-06, "loss": 0.0022, "step": 92760 }, { "epoch": 1.5179579481305736, "grad_norm": 0.012372524477541447, "learning_rate": 7.803346908319889e-06, "loss": 0.0011, "step": 92770 }, { "epoch": 1.5181215740816494, "grad_norm": 0.019645443186163902, "learning_rate": 7.80275569835735e-06, "loss": 0.0013, "step": 92780 }, { "epoch": 1.5182852000327252, "grad_norm": 0.05230730399489403, "learning_rate": 7.802164431249005e-06, "loss": 0.0009, "step": 92790 }, { "epoch": 1.5184488259838012, "grad_norm": 0.016104940325021744, "learning_rate": 7.801573107006912e-06, "loss": 0.0017, "step": 92800 }, { "epoch": 1.5186124519348767, "grad_norm": 0.0755678042769432, "learning_rate": 7.800981725643124e-06, "loss": 0.0012, "step": 92810 }, { "epoch": 1.5187760778859527, "grad_norm": 0.033865638077259064, "learning_rate": 7.8003902871697e-06, "loss": 0.0012, "step": 92820 }, { "epoch": 1.5189397038370287, "grad_norm": 0.045852478593587875, "learning_rate": 7.799798791598702e-06, "loss": 0.0011, "step": 92830 }, { "epoch": 1.5191033297881043, "grad_norm": 0.13485026359558105, "learning_rate": 7.799207238942188e-06, "loss": 0.0019, "step": 92840 }, { "epoch": 1.5192669557391802, "grad_norm": 0.11437754333019257, "learning_rate": 7.798615629212219e-06, "loss": 0.0017, "step": 92850 }, { "epoch": 1.519430581690256, "grad_norm": 0.024181438609957695, "learning_rate": 7.798023962420856e-06, "loss": 0.0028, "step": 92860 }, { "epoch": 1.5195942076413318, "grad_norm": 0.058550313115119934, "learning_rate": 7.797432238580165e-06, "loss": 0.0006, "step": 92870 }, { "epoch": 1.5197578335924078, "grad_norm": 0.10621500760316849, "learning_rate": 7.79684045770221e-06, "loss": 0.0011, "step": 92880 }, { "epoch": 1.5199214595434836, "grad_norm": 0.05655359849333763, "learning_rate": 7.796248619799056e-06, "loss": 0.0019, "step": 92890 }, { "epoch": 1.5200850854945593, "grad_norm": 0.05131177604198456, "learning_rate": 7.79565672488277e-06, "loss": 0.0015, "step": 92900 }, { "epoch": 1.5202487114456353, "grad_norm": 0.013208159245550632, "learning_rate": 7.795064772965423e-06, "loss": 0.0009, "step": 92910 }, { "epoch": 1.5204123373967111, "grad_norm": 0.018521204590797424, "learning_rate": 7.794472764059082e-06, "loss": 0.002, "step": 92920 }, { "epoch": 1.5205759633477869, "grad_norm": 0.02633097767829895, "learning_rate": 7.793880698175816e-06, "loss": 0.0012, "step": 92930 }, { "epoch": 1.5207395892988629, "grad_norm": 0.013576476834714413, "learning_rate": 7.7932885753277e-06, "loss": 0.0017, "step": 92940 }, { "epoch": 1.5209032152499387, "grad_norm": 0.01149922888725996, "learning_rate": 7.792696395526807e-06, "loss": 0.0023, "step": 92950 }, { "epoch": 1.5210668412010144, "grad_norm": 0.09787238389253616, "learning_rate": 7.792104158785208e-06, "loss": 0.0027, "step": 92960 }, { "epoch": 1.5212304671520904, "grad_norm": 0.03567834943532944, "learning_rate": 7.79151186511498e-06, "loss": 0.0013, "step": 92970 }, { "epoch": 1.5213940931031662, "grad_norm": 0.13000251352787018, "learning_rate": 7.7909195145282e-06, "loss": 0.0016, "step": 92980 }, { "epoch": 1.521557719054242, "grad_norm": 0.07946809381246567, "learning_rate": 7.790327107036942e-06, "loss": 0.002, "step": 92990 }, { "epoch": 1.521721345005318, "grad_norm": 0.002805691910907626, "learning_rate": 7.789734642653291e-06, "loss": 0.0012, "step": 93000 }, { "epoch": 1.5218849709563935, "grad_norm": 0.11739125102758408, "learning_rate": 7.78914212138932e-06, "loss": 0.0023, "step": 93010 }, { "epoch": 1.5220485969074695, "grad_norm": 0.06629146635532379, "learning_rate": 7.788549543257116e-06, "loss": 0.0016, "step": 93020 }, { "epoch": 1.5222122228585455, "grad_norm": 0.05911160260438919, "learning_rate": 7.787956908268757e-06, "loss": 0.0023, "step": 93030 }, { "epoch": 1.522375848809621, "grad_norm": 0.037359148263931274, "learning_rate": 7.787364216436326e-06, "loss": 0.0019, "step": 93040 }, { "epoch": 1.522539474760697, "grad_norm": 0.1179179698228836, "learning_rate": 7.786771467771913e-06, "loss": 0.0028, "step": 93050 }, { "epoch": 1.5227031007117728, "grad_norm": 0.058885734528303146, "learning_rate": 7.786178662287596e-06, "loss": 0.002, "step": 93060 }, { "epoch": 1.5228667266628486, "grad_norm": 0.029980482533574104, "learning_rate": 7.785585799995467e-06, "loss": 0.0015, "step": 93070 }, { "epoch": 1.5230303526139246, "grad_norm": 0.08554088324308395, "learning_rate": 7.784992880907613e-06, "loss": 0.0022, "step": 93080 }, { "epoch": 1.5231939785650004, "grad_norm": 0.13785605132579803, "learning_rate": 7.784399905036124e-06, "loss": 0.0019, "step": 93090 }, { "epoch": 1.5233576045160762, "grad_norm": 0.07108182460069656, "learning_rate": 7.783806872393086e-06, "loss": 0.0021, "step": 93100 }, { "epoch": 1.5235212304671522, "grad_norm": 0.02123451419174671, "learning_rate": 7.783213782990596e-06, "loss": 0.0023, "step": 93110 }, { "epoch": 1.523684856418228, "grad_norm": 0.13213656842708588, "learning_rate": 7.782620636840743e-06, "loss": 0.0016, "step": 93120 }, { "epoch": 1.5238484823693037, "grad_norm": 0.22347159683704376, "learning_rate": 7.782027433955623e-06, "loss": 0.0015, "step": 93130 }, { "epoch": 1.5240121083203797, "grad_norm": 0.018235869705677032, "learning_rate": 7.781434174347327e-06, "loss": 0.0007, "step": 93140 }, { "epoch": 1.5241757342714555, "grad_norm": 0.04336737096309662, "learning_rate": 7.780840858027956e-06, "loss": 0.0034, "step": 93150 }, { "epoch": 1.5243393602225312, "grad_norm": 0.007075407542288303, "learning_rate": 7.780247485009604e-06, "loss": 0.0013, "step": 93160 }, { "epoch": 1.5245029861736072, "grad_norm": 0.045361507683992386, "learning_rate": 7.77965405530437e-06, "loss": 0.002, "step": 93170 }, { "epoch": 1.524666612124683, "grad_norm": 0.10014241188764572, "learning_rate": 7.779060568924355e-06, "loss": 0.0017, "step": 93180 }, { "epoch": 1.5248302380757588, "grad_norm": 0.016102198511362076, "learning_rate": 7.778467025881657e-06, "loss": 0.0014, "step": 93190 }, { "epoch": 1.5249938640268348, "grad_norm": 0.0876363217830658, "learning_rate": 7.77787342618838e-06, "loss": 0.0022, "step": 93200 }, { "epoch": 1.5251574899779103, "grad_norm": 0.1427038311958313, "learning_rate": 7.777279769856628e-06, "loss": 0.0025, "step": 93210 }, { "epoch": 1.5253211159289863, "grad_norm": 0.06405994296073914, "learning_rate": 7.776686056898501e-06, "loss": 0.0016, "step": 93220 }, { "epoch": 1.5254847418800623, "grad_norm": 0.02745913341641426, "learning_rate": 7.776092287326108e-06, "loss": 0.0013, "step": 93230 }, { "epoch": 1.5256483678311379, "grad_norm": 0.043983809649944305, "learning_rate": 7.775498461151553e-06, "loss": 0.0028, "step": 93240 }, { "epoch": 1.5258119937822139, "grad_norm": 0.038523778319358826, "learning_rate": 7.774904578386946e-06, "loss": 0.0012, "step": 93250 }, { "epoch": 1.5259756197332897, "grad_norm": 0.059129249304533005, "learning_rate": 7.774310639044395e-06, "loss": 0.0014, "step": 93260 }, { "epoch": 1.5261392456843654, "grad_norm": 0.05069431662559509, "learning_rate": 7.773716643136008e-06, "loss": 0.0012, "step": 93270 }, { "epoch": 1.5263028716354414, "grad_norm": 0.06535604596138, "learning_rate": 7.7731225906739e-06, "loss": 0.0014, "step": 93280 }, { "epoch": 1.5264664975865172, "grad_norm": 0.07446514815092087, "learning_rate": 7.772528481670178e-06, "loss": 0.0013, "step": 93290 }, { "epoch": 1.526630123537593, "grad_norm": 0.13768643140792847, "learning_rate": 7.77193431613696e-06, "loss": 0.0019, "step": 93300 }, { "epoch": 1.526793749488669, "grad_norm": 0.214603990316391, "learning_rate": 7.771340094086359e-06, "loss": 0.0031, "step": 93310 }, { "epoch": 1.5269573754397447, "grad_norm": 0.01101094949990511, "learning_rate": 7.77074581553049e-06, "loss": 0.0027, "step": 93320 }, { "epoch": 1.5271210013908205, "grad_norm": 0.049427330493927, "learning_rate": 7.77015148048147e-06, "loss": 0.0077, "step": 93330 }, { "epoch": 1.5272846273418965, "grad_norm": 0.032521817833185196, "learning_rate": 7.769557088951419e-06, "loss": 0.0018, "step": 93340 }, { "epoch": 1.5274482532929723, "grad_norm": 0.06289082020521164, "learning_rate": 7.768962640952451e-06, "loss": 0.0021, "step": 93350 }, { "epoch": 1.527611879244048, "grad_norm": 0.048975635319948196, "learning_rate": 7.76836813649669e-06, "loss": 0.0012, "step": 93360 }, { "epoch": 1.527775505195124, "grad_norm": 0.26441138982772827, "learning_rate": 7.76777357559626e-06, "loss": 0.0019, "step": 93370 }, { "epoch": 1.5279391311461998, "grad_norm": 0.05599047988653183, "learning_rate": 7.76717895826328e-06, "loss": 0.0014, "step": 93380 }, { "epoch": 1.5281027570972756, "grad_norm": 0.013066095300018787, "learning_rate": 7.766584284509872e-06, "loss": 0.0022, "step": 93390 }, { "epoch": 1.5282663830483516, "grad_norm": 0.06602037698030472, "learning_rate": 7.765989554348166e-06, "loss": 0.0012, "step": 93400 }, { "epoch": 1.5284300089994272, "grad_norm": 0.08403456956148148, "learning_rate": 7.765394767790285e-06, "loss": 0.002, "step": 93410 }, { "epoch": 1.5285936349505032, "grad_norm": 0.06702597439289093, "learning_rate": 7.764799924848354e-06, "loss": 0.0015, "step": 93420 }, { "epoch": 1.5287572609015792, "grad_norm": 0.10284309834241867, "learning_rate": 7.764205025534506e-06, "loss": 0.001, "step": 93430 }, { "epoch": 1.5289208868526547, "grad_norm": 0.04154201224446297, "learning_rate": 7.763610069860869e-06, "loss": 0.0015, "step": 93440 }, { "epoch": 1.5290845128037307, "grad_norm": 0.046979136765003204, "learning_rate": 7.763015057839571e-06, "loss": 0.0018, "step": 93450 }, { "epoch": 1.5292481387548065, "grad_norm": 0.42486369609832764, "learning_rate": 7.762419989482748e-06, "loss": 0.0027, "step": 93460 }, { "epoch": 1.5294117647058822, "grad_norm": 0.0269173514097929, "learning_rate": 7.76182486480253e-06, "loss": 0.0015, "step": 93470 }, { "epoch": 1.5295753906569582, "grad_norm": 0.01550376322120428, "learning_rate": 7.76122968381105e-06, "loss": 0.0014, "step": 93480 }, { "epoch": 1.529739016608034, "grad_norm": 0.03390613943338394, "learning_rate": 7.760634446520448e-06, "loss": 0.0025, "step": 93490 }, { "epoch": 1.5299026425591098, "grad_norm": 0.11964254826307297, "learning_rate": 7.760039152942856e-06, "loss": 0.0022, "step": 93500 }, { "epoch": 1.5300662685101858, "grad_norm": 0.05856117606163025, "learning_rate": 7.759443803090414e-06, "loss": 0.0012, "step": 93510 }, { "epoch": 1.5302298944612616, "grad_norm": 0.04811723902821541, "learning_rate": 7.758848396975258e-06, "loss": 0.0013, "step": 93520 }, { "epoch": 1.5303935204123373, "grad_norm": 0.10150565207004547, "learning_rate": 7.758252934609532e-06, "loss": 0.0023, "step": 93530 }, { "epoch": 1.5305571463634133, "grad_norm": 0.045371685177087784, "learning_rate": 7.757657416005373e-06, "loss": 0.0012, "step": 93540 }, { "epoch": 1.530720772314489, "grad_norm": 0.11376247555017471, "learning_rate": 7.757061841174924e-06, "loss": 0.003, "step": 93550 }, { "epoch": 1.5308843982655649, "grad_norm": 0.13673394918441772, "learning_rate": 7.756466210130329e-06, "loss": 0.0033, "step": 93560 }, { "epoch": 1.5310480242166409, "grad_norm": 0.12203549593687057, "learning_rate": 7.755870522883732e-06, "loss": 0.0014, "step": 93570 }, { "epoch": 1.5312116501677167, "grad_norm": 0.0539158433675766, "learning_rate": 7.75527477944728e-06, "loss": 0.0011, "step": 93580 }, { "epoch": 1.5313752761187924, "grad_norm": 0.07093073427677155, "learning_rate": 7.754678979833118e-06, "loss": 0.0012, "step": 93590 }, { "epoch": 1.5315389020698684, "grad_norm": 0.11029540747404099, "learning_rate": 7.754083124053394e-06, "loss": 0.0023, "step": 93600 }, { "epoch": 1.531702528020944, "grad_norm": 0.030714010819792747, "learning_rate": 7.753487212120258e-06, "loss": 0.0027, "step": 93610 }, { "epoch": 1.53186615397202, "grad_norm": 0.06172260269522667, "learning_rate": 7.752891244045859e-06, "loss": 0.0012, "step": 93620 }, { "epoch": 1.532029779923096, "grad_norm": 0.03526344522833824, "learning_rate": 7.75229521984235e-06, "loss": 0.0019, "step": 93630 }, { "epoch": 1.5321934058741715, "grad_norm": 0.028737753629684448, "learning_rate": 7.75169913952188e-06, "loss": 0.0018, "step": 93640 }, { "epoch": 1.5323570318252475, "grad_norm": 0.11571808159351349, "learning_rate": 7.751103003096608e-06, "loss": 0.0013, "step": 93650 }, { "epoch": 1.5325206577763233, "grad_norm": 0.043877072632312775, "learning_rate": 7.750506810578682e-06, "loss": 0.0021, "step": 93660 }, { "epoch": 1.532684283727399, "grad_norm": 0.02701399475336075, "learning_rate": 7.749910561980263e-06, "loss": 0.0024, "step": 93670 }, { "epoch": 1.532847909678475, "grad_norm": 0.038616981357336044, "learning_rate": 7.749314257313506e-06, "loss": 0.0009, "step": 93680 }, { "epoch": 1.5330115356295508, "grad_norm": 0.03400932624936104, "learning_rate": 7.748717896590569e-06, "loss": 0.0016, "step": 93690 }, { "epoch": 1.5331751615806266, "grad_norm": 0.10926592350006104, "learning_rate": 7.748121479823614e-06, "loss": 0.0018, "step": 93700 }, { "epoch": 1.5333387875317026, "grad_norm": 0.025459859520196915, "learning_rate": 7.747525007024797e-06, "loss": 0.0016, "step": 93710 }, { "epoch": 1.5335024134827784, "grad_norm": 0.058298103511333466, "learning_rate": 7.746928478206283e-06, "loss": 0.0017, "step": 93720 }, { "epoch": 1.5336660394338542, "grad_norm": 0.01735171303153038, "learning_rate": 7.746331893380232e-06, "loss": 0.002, "step": 93730 }, { "epoch": 1.5338296653849302, "grad_norm": 0.0650179535150528, "learning_rate": 7.745735252558811e-06, "loss": 0.0014, "step": 93740 }, { "epoch": 1.533993291336006, "grad_norm": 0.09811414033174515, "learning_rate": 7.745138555754182e-06, "loss": 0.0017, "step": 93750 }, { "epoch": 1.5341569172870817, "grad_norm": 0.07228904962539673, "learning_rate": 7.744541802978514e-06, "loss": 0.0023, "step": 93760 }, { "epoch": 1.5343205432381577, "grad_norm": 0.04679113253951073, "learning_rate": 7.743944994243971e-06, "loss": 0.0012, "step": 93770 }, { "epoch": 1.5344841691892332, "grad_norm": 0.07354404032230377, "learning_rate": 7.743348129562724e-06, "loss": 0.0013, "step": 93780 }, { "epoch": 1.5346477951403092, "grad_norm": 0.12102992832660675, "learning_rate": 7.742751208946942e-06, "loss": 0.0017, "step": 93790 }, { "epoch": 1.5348114210913852, "grad_norm": 0.046720899641513824, "learning_rate": 7.742154232408796e-06, "loss": 0.0014, "step": 93800 }, { "epoch": 1.5349750470424608, "grad_norm": 0.09730877727270126, "learning_rate": 7.741557199960457e-06, "loss": 0.0026, "step": 93810 }, { "epoch": 1.5351386729935368, "grad_norm": 0.028378024697303772, "learning_rate": 7.740960111614097e-06, "loss": 0.002, "step": 93820 }, { "epoch": 1.5353022989446126, "grad_norm": 0.047946829348802567, "learning_rate": 7.740362967381893e-06, "loss": 0.0022, "step": 93830 }, { "epoch": 1.5354659248956883, "grad_norm": 0.19108378887176514, "learning_rate": 7.73976576727602e-06, "loss": 0.0012, "step": 93840 }, { "epoch": 1.5356295508467643, "grad_norm": 0.021794864907860756, "learning_rate": 7.73916851130865e-06, "loss": 0.0013, "step": 93850 }, { "epoch": 1.53579317679784, "grad_norm": 0.056179262697696686, "learning_rate": 7.738571199491965e-06, "loss": 0.0034, "step": 93860 }, { "epoch": 1.5359568027489159, "grad_norm": 0.015256480313837528, "learning_rate": 7.73797383183814e-06, "loss": 0.0011, "step": 93870 }, { "epoch": 1.5361204286999919, "grad_norm": 0.057024504989385605, "learning_rate": 7.737376408359362e-06, "loss": 0.001, "step": 93880 }, { "epoch": 1.5362840546510677, "grad_norm": 0.02208586223423481, "learning_rate": 7.736778929067804e-06, "loss": 0.0028, "step": 93890 }, { "epoch": 1.5364476806021434, "grad_norm": 0.010442906990647316, "learning_rate": 7.736181393975653e-06, "loss": 0.0016, "step": 93900 }, { "epoch": 1.5366113065532194, "grad_norm": 0.08820034563541412, "learning_rate": 7.735583803095088e-06, "loss": 0.0013, "step": 93910 }, { "epoch": 1.5367749325042952, "grad_norm": 0.03330162540078163, "learning_rate": 7.734986156438296e-06, "loss": 0.0014, "step": 93920 }, { "epoch": 1.536938558455371, "grad_norm": 0.039124760776758194, "learning_rate": 7.734388454017463e-06, "loss": 0.0126, "step": 93930 }, { "epoch": 1.537102184406447, "grad_norm": 0.09567532688379288, "learning_rate": 7.733790695844776e-06, "loss": 0.0011, "step": 93940 }, { "epoch": 1.5372658103575227, "grad_norm": 0.04627474024891853, "learning_rate": 7.733192881932421e-06, "loss": 0.0015, "step": 93950 }, { "epoch": 1.5374294363085985, "grad_norm": 0.034613557159900665, "learning_rate": 7.732595012292587e-06, "loss": 0.001, "step": 93960 }, { "epoch": 1.5375930622596745, "grad_norm": 0.02414688467979431, "learning_rate": 7.731997086937466e-06, "loss": 0.001, "step": 93970 }, { "epoch": 1.53775668821075, "grad_norm": 0.07709834724664688, "learning_rate": 7.731399105879246e-06, "loss": 0.0017, "step": 93980 }, { "epoch": 1.537920314161826, "grad_norm": 0.08050071448087692, "learning_rate": 7.730801069130123e-06, "loss": 0.002, "step": 93990 }, { "epoch": 1.538083940112902, "grad_norm": 0.16395682096481323, "learning_rate": 7.730202976702288e-06, "loss": 0.0031, "step": 94000 }, { "epoch": 1.5382475660639776, "grad_norm": 0.09704656153917313, "learning_rate": 7.729604828607935e-06, "loss": 0.0022, "step": 94010 }, { "epoch": 1.5384111920150536, "grad_norm": 0.05278277024626732, "learning_rate": 7.729006624859263e-06, "loss": 0.003, "step": 94020 }, { "epoch": 1.5385748179661294, "grad_norm": 0.030834274366497993, "learning_rate": 7.728408365468466e-06, "loss": 0.0026, "step": 94030 }, { "epoch": 1.5387384439172052, "grad_norm": 0.049989696592092514, "learning_rate": 7.727810050447742e-06, "loss": 0.0015, "step": 94040 }, { "epoch": 1.5389020698682812, "grad_norm": 0.028057513758540154, "learning_rate": 7.727211679809292e-06, "loss": 0.0014, "step": 94050 }, { "epoch": 1.539065695819357, "grad_norm": 0.12275896221399307, "learning_rate": 7.726613253565315e-06, "loss": 0.003, "step": 94060 }, { "epoch": 1.5392293217704327, "grad_norm": 0.06397764384746552, "learning_rate": 7.726014771728014e-06, "loss": 0.0017, "step": 94070 }, { "epoch": 1.5393929477215087, "grad_norm": 0.05805019661784172, "learning_rate": 7.725416234309589e-06, "loss": 0.0017, "step": 94080 }, { "epoch": 1.5395565736725845, "grad_norm": 0.040418725460767746, "learning_rate": 7.724817641322244e-06, "loss": 0.0024, "step": 94090 }, { "epoch": 1.5397201996236602, "grad_norm": 0.10134690254926682, "learning_rate": 7.724218992778185e-06, "loss": 0.0017, "step": 94100 }, { "epoch": 1.5398838255747362, "grad_norm": 0.16786016523838043, "learning_rate": 7.72362028868962e-06, "loss": 0.0018, "step": 94110 }, { "epoch": 1.540047451525812, "grad_norm": 0.019320914521813393, "learning_rate": 7.72302152906875e-06, "loss": 0.0014, "step": 94120 }, { "epoch": 1.5402110774768878, "grad_norm": 0.06292970478534698, "learning_rate": 7.722422713927789e-06, "loss": 0.0019, "step": 94130 }, { "epoch": 1.5403747034279638, "grad_norm": 0.2912803292274475, "learning_rate": 7.721823843278944e-06, "loss": 0.0025, "step": 94140 }, { "epoch": 1.5405383293790396, "grad_norm": 0.07295631617307663, "learning_rate": 7.721224917134424e-06, "loss": 0.0012, "step": 94150 }, { "epoch": 1.5407019553301153, "grad_norm": 0.09545114636421204, "learning_rate": 7.720625935506445e-06, "loss": 0.0026, "step": 94160 }, { "epoch": 1.5408655812811913, "grad_norm": 0.0383169986307621, "learning_rate": 7.720026898407215e-06, "loss": 0.0014, "step": 94170 }, { "epoch": 1.5410292072322669, "grad_norm": 0.030350062996149063, "learning_rate": 7.719427805848952e-06, "loss": 0.0029, "step": 94180 }, { "epoch": 1.5411928331833429, "grad_norm": 0.04438008368015289, "learning_rate": 7.718828657843866e-06, "loss": 0.0036, "step": 94190 }, { "epoch": 1.5413564591344189, "grad_norm": 0.0714246854186058, "learning_rate": 7.718229454404178e-06, "loss": 0.0012, "step": 94200 }, { "epoch": 1.5415200850854944, "grad_norm": 0.03005394898355007, "learning_rate": 7.717630195542104e-06, "loss": 0.0012, "step": 94210 }, { "epoch": 1.5416837110365704, "grad_norm": 0.00890421587973833, "learning_rate": 7.717030881269858e-06, "loss": 0.0018, "step": 94220 }, { "epoch": 1.5418473369876462, "grad_norm": 0.04932582750916481, "learning_rate": 7.716431511599666e-06, "loss": 0.0012, "step": 94230 }, { "epoch": 1.542010962938722, "grad_norm": 0.04558493569493294, "learning_rate": 7.715832086543744e-06, "loss": 0.0012, "step": 94240 }, { "epoch": 1.542174588889798, "grad_norm": 0.03884853422641754, "learning_rate": 7.715232606114317e-06, "loss": 0.0011, "step": 94250 }, { "epoch": 1.5423382148408737, "grad_norm": 0.04142134636640549, "learning_rate": 7.714633070323607e-06, "loss": 0.0017, "step": 94260 }, { "epoch": 1.5425018407919495, "grad_norm": 0.04885726049542427, "learning_rate": 7.714033479183836e-06, "loss": 0.0017, "step": 94270 }, { "epoch": 1.5426654667430255, "grad_norm": 0.07813854515552521, "learning_rate": 7.71343383270723e-06, "loss": 0.0019, "step": 94280 }, { "epoch": 1.5428290926941013, "grad_norm": 0.03360377252101898, "learning_rate": 7.712834130906018e-06, "loss": 0.0016, "step": 94290 }, { "epoch": 1.542992718645177, "grad_norm": 0.046063382178545, "learning_rate": 7.712234373792423e-06, "loss": 0.0017, "step": 94300 }, { "epoch": 1.543156344596253, "grad_norm": 0.06124385446310043, "learning_rate": 7.711634561378676e-06, "loss": 0.0013, "step": 94310 }, { "epoch": 1.5433199705473288, "grad_norm": 0.12254296243190765, "learning_rate": 7.711034693677008e-06, "loss": 0.0016, "step": 94320 }, { "epoch": 1.5434835964984046, "grad_norm": 0.1372957080602646, "learning_rate": 7.710434770699647e-06, "loss": 0.0018, "step": 94330 }, { "epoch": 1.5436472224494806, "grad_norm": 0.031282685697078705, "learning_rate": 7.709834792458826e-06, "loss": 0.0018, "step": 94340 }, { "epoch": 1.5438108484005564, "grad_norm": 0.07550206035375595, "learning_rate": 7.70923475896678e-06, "loss": 0.0016, "step": 94350 }, { "epoch": 1.5439744743516322, "grad_norm": 0.006063178647309542, "learning_rate": 7.70863467023574e-06, "loss": 0.0014, "step": 94360 }, { "epoch": 1.5441381003027081, "grad_norm": 0.005758981220424175, "learning_rate": 7.708034526277943e-06, "loss": 0.0021, "step": 94370 }, { "epoch": 1.5443017262537837, "grad_norm": 0.08807338774204254, "learning_rate": 7.707434327105625e-06, "loss": 0.002, "step": 94380 }, { "epoch": 1.5444653522048597, "grad_norm": 0.00961596705019474, "learning_rate": 7.706834072731021e-06, "loss": 0.001, "step": 94390 }, { "epoch": 1.5446289781559357, "grad_norm": 0.1088404729962349, "learning_rate": 7.706233763166377e-06, "loss": 0.002, "step": 94400 }, { "epoch": 1.5447926041070112, "grad_norm": 0.0641908198595047, "learning_rate": 7.705633398423925e-06, "loss": 0.0028, "step": 94410 }, { "epoch": 1.5449562300580872, "grad_norm": 0.019839268177747726, "learning_rate": 7.70503297851591e-06, "loss": 0.0013, "step": 94420 }, { "epoch": 1.545119856009163, "grad_norm": 0.053487423807382584, "learning_rate": 7.704432503454572e-06, "loss": 0.0049, "step": 94430 }, { "epoch": 1.5452834819602388, "grad_norm": 0.24566827714443207, "learning_rate": 7.703831973252158e-06, "loss": 0.0025, "step": 94440 }, { "epoch": 1.5454471079113148, "grad_norm": 0.040807705372571945, "learning_rate": 7.70323138792091e-06, "loss": 0.0019, "step": 94450 }, { "epoch": 1.5456107338623906, "grad_norm": 0.039976924657821655, "learning_rate": 7.70263074747307e-06, "loss": 0.0037, "step": 94460 }, { "epoch": 1.5457743598134663, "grad_norm": 0.0666600912809372, "learning_rate": 7.702030051920887e-06, "loss": 0.0019, "step": 94470 }, { "epoch": 1.5459379857645423, "grad_norm": 0.13798412680625916, "learning_rate": 7.701429301276612e-06, "loss": 0.0022, "step": 94480 }, { "epoch": 1.546101611715618, "grad_norm": 0.050319135189056396, "learning_rate": 7.70082849555249e-06, "loss": 0.0027, "step": 94490 }, { "epoch": 1.5462652376666939, "grad_norm": 0.08840583264827728, "learning_rate": 7.700227634760773e-06, "loss": 0.0026, "step": 94500 }, { "epoch": 1.5464288636177699, "grad_norm": 0.16266296803951263, "learning_rate": 7.69962671891371e-06, "loss": 0.0028, "step": 94510 }, { "epoch": 1.5465924895688457, "grad_norm": 0.057421181350946426, "learning_rate": 7.699025748023553e-06, "loss": 0.0011, "step": 94520 }, { "epoch": 1.5467561155199214, "grad_norm": 0.04220881313085556, "learning_rate": 7.698424722102558e-06, "loss": 0.0008, "step": 94530 }, { "epoch": 1.5469197414709974, "grad_norm": 0.1432119607925415, "learning_rate": 7.697823641162978e-06, "loss": 0.0011, "step": 94540 }, { "epoch": 1.547083367422073, "grad_norm": 0.04307156801223755, "learning_rate": 7.697222505217065e-06, "loss": 0.0016, "step": 94550 }, { "epoch": 1.547246993373149, "grad_norm": 0.03296162560582161, "learning_rate": 7.696621314277083e-06, "loss": 0.0019, "step": 94560 }, { "epoch": 1.547410619324225, "grad_norm": 0.014621433801949024, "learning_rate": 7.696020068355285e-06, "loss": 0.0023, "step": 94570 }, { "epoch": 1.5475742452753005, "grad_norm": 0.07736492902040482, "learning_rate": 7.695418767463931e-06, "loss": 0.0015, "step": 94580 }, { "epoch": 1.5477378712263765, "grad_norm": 0.19660060107707977, "learning_rate": 7.694817411615279e-06, "loss": 0.0018, "step": 94590 }, { "epoch": 1.5479014971774523, "grad_norm": 0.07001695781946182, "learning_rate": 7.694216000821592e-06, "loss": 0.0014, "step": 94600 }, { "epoch": 1.548065123128528, "grad_norm": 0.19812771677970886, "learning_rate": 7.693614535095133e-06, "loss": 0.0021, "step": 94610 }, { "epoch": 1.548228749079604, "grad_norm": 0.04761344566941261, "learning_rate": 7.693013014448166e-06, "loss": 0.0013, "step": 94620 }, { "epoch": 1.5483923750306798, "grad_norm": 0.1871534138917923, "learning_rate": 7.69241143889295e-06, "loss": 0.0043, "step": 94630 }, { "epoch": 1.5485560009817556, "grad_norm": 0.016032101586461067, "learning_rate": 7.691809808441758e-06, "loss": 0.001, "step": 94640 }, { "epoch": 1.5487196269328316, "grad_norm": 0.06465773284435272, "learning_rate": 7.691208123106852e-06, "loss": 0.0016, "step": 94650 }, { "epoch": 1.5488832528839074, "grad_norm": 0.09080789983272552, "learning_rate": 7.6906063829005e-06, "loss": 0.0023, "step": 94660 }, { "epoch": 1.5490468788349832, "grad_norm": 0.04587014019489288, "learning_rate": 7.690004587834974e-06, "loss": 0.0007, "step": 94670 }, { "epoch": 1.5492105047860592, "grad_norm": 0.012280844151973724, "learning_rate": 7.689402737922542e-06, "loss": 0.0012, "step": 94680 }, { "epoch": 1.549374130737135, "grad_norm": 0.018742302432656288, "learning_rate": 7.688800833175476e-06, "loss": 0.0017, "step": 94690 }, { "epoch": 1.5495377566882107, "grad_norm": 0.20574426651000977, "learning_rate": 7.688198873606046e-06, "loss": 0.0017, "step": 94700 }, { "epoch": 1.5497013826392867, "grad_norm": 0.05346616730093956, "learning_rate": 7.687596859226528e-06, "loss": 0.0009, "step": 94710 }, { "epoch": 1.5498650085903625, "grad_norm": 0.02134278230369091, "learning_rate": 7.686994790049197e-06, "loss": 0.0011, "step": 94720 }, { "epoch": 1.5500286345414382, "grad_norm": 0.28118082880973816, "learning_rate": 7.686392666086325e-06, "loss": 0.0016, "step": 94730 }, { "epoch": 1.5501922604925142, "grad_norm": 0.03732013329863548, "learning_rate": 7.685790487350194e-06, "loss": 0.001, "step": 94740 }, { "epoch": 1.5503558864435898, "grad_norm": 0.05782457813620567, "learning_rate": 7.685188253853077e-06, "loss": 0.0016, "step": 94750 }, { "epoch": 1.5505195123946658, "grad_norm": 0.03530120477080345, "learning_rate": 7.684585965607255e-06, "loss": 0.0011, "step": 94760 }, { "epoch": 1.5506831383457418, "grad_norm": 0.05049153417348862, "learning_rate": 7.68398362262501e-06, "loss": 0.0013, "step": 94770 }, { "epoch": 1.5508467642968173, "grad_norm": 0.045431897044181824, "learning_rate": 7.68338122491862e-06, "loss": 0.0027, "step": 94780 }, { "epoch": 1.5510103902478933, "grad_norm": 0.060159243643283844, "learning_rate": 7.68277877250037e-06, "loss": 0.0014, "step": 94790 }, { "epoch": 1.551174016198969, "grad_norm": 0.07681483775377274, "learning_rate": 7.682176265382541e-06, "loss": 0.0023, "step": 94800 }, { "epoch": 1.5513376421500449, "grad_norm": 0.22321976721286774, "learning_rate": 7.681573703577422e-06, "loss": 0.0018, "step": 94810 }, { "epoch": 1.5515012681011209, "grad_norm": 0.01603671722114086, "learning_rate": 7.680971087097293e-06, "loss": 0.001, "step": 94820 }, { "epoch": 1.5516648940521967, "grad_norm": 0.027682511135935783, "learning_rate": 7.680368415954444e-06, "loss": 0.0015, "step": 94830 }, { "epoch": 1.5518285200032724, "grad_norm": 0.010590538382530212, "learning_rate": 7.679765690161165e-06, "loss": 0.0016, "step": 94840 }, { "epoch": 1.5519921459543484, "grad_norm": 0.061615586280822754, "learning_rate": 7.67916290972974e-06, "loss": 0.0018, "step": 94850 }, { "epoch": 1.5521557719054242, "grad_norm": 0.030306680127978325, "learning_rate": 7.678560074672461e-06, "loss": 0.0016, "step": 94860 }, { "epoch": 1.5523193978565, "grad_norm": 0.07297494262456894, "learning_rate": 7.677957185001624e-06, "loss": 0.0027, "step": 94870 }, { "epoch": 1.552483023807576, "grad_norm": 0.0470055527985096, "learning_rate": 7.677354240729514e-06, "loss": 0.002, "step": 94880 }, { "epoch": 1.5526466497586517, "grad_norm": 0.0757136195898056, "learning_rate": 7.67675124186843e-06, "loss": 0.0011, "step": 94890 }, { "epoch": 1.5528102757097275, "grad_norm": 0.06269910931587219, "learning_rate": 7.676148188430664e-06, "loss": 0.0031, "step": 94900 }, { "epoch": 1.5529739016608035, "grad_norm": 0.05475131794810295, "learning_rate": 7.675545080428512e-06, "loss": 0.0023, "step": 94910 }, { "epoch": 1.5531375276118793, "grad_norm": 0.20788739621639252, "learning_rate": 7.67494191787427e-06, "loss": 0.0021, "step": 94920 }, { "epoch": 1.553301153562955, "grad_norm": 0.053077392280101776, "learning_rate": 7.67433870078024e-06, "loss": 0.001, "step": 94930 }, { "epoch": 1.553464779514031, "grad_norm": 0.0700356736779213, "learning_rate": 7.673735429158717e-06, "loss": 0.0012, "step": 94940 }, { "epoch": 1.5536284054651066, "grad_norm": 0.01644502952694893, "learning_rate": 7.673132103022004e-06, "loss": 0.0016, "step": 94950 }, { "epoch": 1.5537920314161826, "grad_norm": 0.04911402612924576, "learning_rate": 7.672528722382398e-06, "loss": 0.0011, "step": 94960 }, { "epoch": 1.5539556573672586, "grad_norm": 0.11293618381023407, "learning_rate": 7.671925287252207e-06, "loss": 0.0014, "step": 94970 }, { "epoch": 1.5541192833183342, "grad_norm": 0.02110135927796364, "learning_rate": 7.67132179764373e-06, "loss": 0.0014, "step": 94980 }, { "epoch": 1.5542829092694102, "grad_norm": 0.026496225968003273, "learning_rate": 7.670718253569274e-06, "loss": 0.0023, "step": 94990 }, { "epoch": 1.554446535220486, "grad_norm": 0.08040380477905273, "learning_rate": 7.670114655041144e-06, "loss": 0.0014, "step": 95000 }, { "epoch": 1.5546101611715617, "grad_norm": 0.09811513125896454, "learning_rate": 7.669511002071648e-06, "loss": 0.0017, "step": 95010 }, { "epoch": 1.5547737871226377, "grad_norm": 0.08608756214380264, "learning_rate": 7.668907294673092e-06, "loss": 0.0014, "step": 95020 }, { "epoch": 1.5549374130737135, "grad_norm": 0.023106861859560013, "learning_rate": 7.668303532857786e-06, "loss": 0.002, "step": 95030 }, { "epoch": 1.5551010390247892, "grad_norm": 0.06787912547588348, "learning_rate": 7.66769971663804e-06, "loss": 0.0011, "step": 95040 }, { "epoch": 1.5552646649758652, "grad_norm": 0.06239595264196396, "learning_rate": 7.667095846026167e-06, "loss": 0.002, "step": 95050 }, { "epoch": 1.555428290926941, "grad_norm": 0.06543776392936707, "learning_rate": 7.666491921034478e-06, "loss": 0.0012, "step": 95060 }, { "epoch": 1.5555919168780168, "grad_norm": 0.06617656350135803, "learning_rate": 7.665887941675286e-06, "loss": 0.0012, "step": 95070 }, { "epoch": 1.5557555428290928, "grad_norm": 0.06516584753990173, "learning_rate": 7.665283907960906e-06, "loss": 0.0032, "step": 95080 }, { "epoch": 1.5559191687801686, "grad_norm": 0.23765267431735992, "learning_rate": 7.664679819903653e-06, "loss": 0.0014, "step": 95090 }, { "epoch": 1.5560827947312443, "grad_norm": 0.02622072398662567, "learning_rate": 7.664075677515845e-06, "loss": 0.0017, "step": 95100 }, { "epoch": 1.5562464206823203, "grad_norm": 0.03308173269033432, "learning_rate": 7.6634714808098e-06, "loss": 0.0014, "step": 95110 }, { "epoch": 1.556410046633396, "grad_norm": 0.056911688297986984, "learning_rate": 7.662867229797837e-06, "loss": 0.0017, "step": 95120 }, { "epoch": 1.5565736725844719, "grad_norm": 0.0777679905295372, "learning_rate": 7.662262924492274e-06, "loss": 0.0011, "step": 95130 }, { "epoch": 1.5567372985355479, "grad_norm": 0.008404085412621498, "learning_rate": 7.661658564905437e-06, "loss": 0.0005, "step": 95140 }, { "epoch": 1.5569009244866234, "grad_norm": 0.0668143630027771, "learning_rate": 7.661054151049645e-06, "loss": 0.001, "step": 95150 }, { "epoch": 1.5570645504376994, "grad_norm": 0.03306376188993454, "learning_rate": 7.660449682937222e-06, "loss": 0.0015, "step": 95160 }, { "epoch": 1.5572281763887754, "grad_norm": 0.1524168998003006, "learning_rate": 7.65984516058049e-06, "loss": 0.0021, "step": 95170 }, { "epoch": 1.557391802339851, "grad_norm": 0.00873512215912342, "learning_rate": 7.65924058399178e-06, "loss": 0.0014, "step": 95180 }, { "epoch": 1.557555428290927, "grad_norm": 0.034942325204610825, "learning_rate": 7.658635953183415e-06, "loss": 0.0014, "step": 95190 }, { "epoch": 1.5577190542420027, "grad_norm": 0.1385362446308136, "learning_rate": 7.658031268167724e-06, "loss": 0.0012, "step": 95200 }, { "epoch": 1.5578826801930785, "grad_norm": 0.058620937168598175, "learning_rate": 7.657426528957037e-06, "loss": 0.0017, "step": 95210 }, { "epoch": 1.5580463061441545, "grad_norm": 0.04665398597717285, "learning_rate": 7.656821735563683e-06, "loss": 0.0017, "step": 95220 }, { "epoch": 1.5582099320952303, "grad_norm": 0.03632679954171181, "learning_rate": 7.656216887999992e-06, "loss": 0.0016, "step": 95230 }, { "epoch": 1.558373558046306, "grad_norm": 0.05030184984207153, "learning_rate": 7.655611986278298e-06, "loss": 0.0014, "step": 95240 }, { "epoch": 1.558537183997382, "grad_norm": 0.035657986998558044, "learning_rate": 7.655007030410935e-06, "loss": 0.0008, "step": 95250 }, { "epoch": 1.5587008099484578, "grad_norm": 0.0658717006444931, "learning_rate": 7.654402020410236e-06, "loss": 0.0011, "step": 95260 }, { "epoch": 1.5588644358995336, "grad_norm": 0.02868214063346386, "learning_rate": 7.653796956288537e-06, "loss": 0.0019, "step": 95270 }, { "epoch": 1.5590280618506096, "grad_norm": 0.08511567860841751, "learning_rate": 7.653191838058176e-06, "loss": 0.0012, "step": 95280 }, { "epoch": 1.5591916878016854, "grad_norm": 0.0516081340610981, "learning_rate": 7.65258666573149e-06, "loss": 0.001, "step": 95290 }, { "epoch": 1.5593553137527612, "grad_norm": 0.02750057354569435, "learning_rate": 7.651981439320816e-06, "loss": 0.0019, "step": 95300 }, { "epoch": 1.5595189397038371, "grad_norm": 0.055199723690748215, "learning_rate": 7.651376158838497e-06, "loss": 0.0015, "step": 95310 }, { "epoch": 1.559682565654913, "grad_norm": 0.06460028886795044, "learning_rate": 7.650770824296873e-06, "loss": 0.0011, "step": 95320 }, { "epoch": 1.5598461916059887, "grad_norm": 0.08816245943307877, "learning_rate": 7.650165435708287e-06, "loss": 0.0014, "step": 95330 }, { "epoch": 1.5600098175570647, "grad_norm": 0.1214061826467514, "learning_rate": 7.649559993085083e-06, "loss": 0.0021, "step": 95340 }, { "epoch": 1.5601734435081402, "grad_norm": 0.056300099939107895, "learning_rate": 7.6489544964396e-06, "loss": 0.0027, "step": 95350 }, { "epoch": 1.5603370694592162, "grad_norm": 0.12485233694314957, "learning_rate": 7.648348945784193e-06, "loss": 0.0014, "step": 95360 }, { "epoch": 1.5605006954102922, "grad_norm": 0.05881715193390846, "learning_rate": 7.647743341131197e-06, "loss": 0.0015, "step": 95370 }, { "epoch": 1.5606643213613678, "grad_norm": 0.03536097705364227, "learning_rate": 7.647137682492972e-06, "loss": 0.0021, "step": 95380 }, { "epoch": 1.5608279473124438, "grad_norm": 0.01662662997841835, "learning_rate": 7.646531969881856e-06, "loss": 0.0007, "step": 95390 }, { "epoch": 1.5609915732635196, "grad_norm": 0.06053727865219116, "learning_rate": 7.645926203310208e-06, "loss": 0.0008, "step": 95400 }, { "epoch": 1.5611551992145953, "grad_norm": 0.026222707703709602, "learning_rate": 7.645320382790373e-06, "loss": 0.0024, "step": 95410 }, { "epoch": 1.5613188251656713, "grad_norm": 0.11342639476060867, "learning_rate": 7.644714508334705e-06, "loss": 0.0019, "step": 95420 }, { "epoch": 1.561482451116747, "grad_norm": 0.06323763728141785, "learning_rate": 7.644108579955558e-06, "loss": 0.0015, "step": 95430 }, { "epoch": 1.5616460770678229, "grad_norm": 0.2431422621011734, "learning_rate": 7.643502597665285e-06, "loss": 0.002, "step": 95440 }, { "epoch": 1.5618097030188989, "grad_norm": 0.0292951799929142, "learning_rate": 7.642896561476243e-06, "loss": 0.0013, "step": 95450 }, { "epoch": 1.5619733289699747, "grad_norm": 0.015974892303347588, "learning_rate": 7.642290471400788e-06, "loss": 0.0012, "step": 95460 }, { "epoch": 1.5621369549210504, "grad_norm": 0.02449287660419941, "learning_rate": 7.641684327451277e-06, "loss": 0.0011, "step": 95470 }, { "epoch": 1.5623005808721264, "grad_norm": 0.03630131483078003, "learning_rate": 7.64107812964007e-06, "loss": 0.0015, "step": 95480 }, { "epoch": 1.5624642068232022, "grad_norm": 0.19347059726715088, "learning_rate": 7.640471877979524e-06, "loss": 0.0016, "step": 95490 }, { "epoch": 1.562627832774278, "grad_norm": 0.030013257637619972, "learning_rate": 7.639865572482004e-06, "loss": 0.0018, "step": 95500 }, { "epoch": 1.562791458725354, "grad_norm": 0.07313106954097748, "learning_rate": 7.63925921315987e-06, "loss": 0.0027, "step": 95510 }, { "epoch": 1.5629550846764295, "grad_norm": 0.006453657988458872, "learning_rate": 7.638652800025484e-06, "loss": 0.0015, "step": 95520 }, { "epoch": 1.5631187106275055, "grad_norm": 0.05946727842092514, "learning_rate": 7.638046333091213e-06, "loss": 0.005, "step": 95530 }, { "epoch": 1.5632823365785815, "grad_norm": 0.033790722489356995, "learning_rate": 7.63743981236942e-06, "loss": 0.0016, "step": 95540 }, { "epoch": 1.563445962529657, "grad_norm": 0.025018930435180664, "learning_rate": 7.636833237872472e-06, "loss": 0.0015, "step": 95550 }, { "epoch": 1.563609588480733, "grad_norm": 0.04538646340370178, "learning_rate": 7.636226609612739e-06, "loss": 0.0018, "step": 95560 }, { "epoch": 1.5637732144318088, "grad_norm": 0.06899354606866837, "learning_rate": 7.635619927602586e-06, "loss": 0.001, "step": 95570 }, { "epoch": 1.5639368403828846, "grad_norm": 0.03022640570998192, "learning_rate": 7.635013191854383e-06, "loss": 0.0008, "step": 95580 }, { "epoch": 1.5641004663339606, "grad_norm": 0.03280191496014595, "learning_rate": 7.634406402380504e-06, "loss": 0.0015, "step": 95590 }, { "epoch": 1.5642640922850364, "grad_norm": 0.02505890280008316, "learning_rate": 7.633799559193317e-06, "loss": 0.0019, "step": 95600 }, { "epoch": 1.5644277182361122, "grad_norm": 0.027498511597514153, "learning_rate": 7.633192662305199e-06, "loss": 0.0008, "step": 95610 }, { "epoch": 1.5645913441871881, "grad_norm": 0.05454948544502258, "learning_rate": 7.63258571172852e-06, "loss": 0.0011, "step": 95620 }, { "epoch": 1.564754970138264, "grad_norm": 0.029083600267767906, "learning_rate": 7.631978707475657e-06, "loss": 0.0017, "step": 95630 }, { "epoch": 1.5649185960893397, "grad_norm": 0.09550821781158447, "learning_rate": 7.631371649558988e-06, "loss": 0.0016, "step": 95640 }, { "epoch": 1.5650822220404157, "grad_norm": 0.09289596229791641, "learning_rate": 7.63076453799089e-06, "loss": 0.002, "step": 95650 }, { "epoch": 1.5652458479914915, "grad_norm": 0.022032108157873154, "learning_rate": 7.630157372783738e-06, "loss": 0.002, "step": 95660 }, { "epoch": 1.5654094739425672, "grad_norm": 0.06356532126665115, "learning_rate": 7.629550153949916e-06, "loss": 0.0011, "step": 95670 }, { "epoch": 1.5655730998936432, "grad_norm": 0.004992218222469091, "learning_rate": 7.628942881501802e-06, "loss": 0.0013, "step": 95680 }, { "epoch": 1.565736725844719, "grad_norm": 0.06797034293413162, "learning_rate": 7.628335555451779e-06, "loss": 0.0022, "step": 95690 }, { "epoch": 1.5659003517957948, "grad_norm": 0.05350487679243088, "learning_rate": 7.627728175812228e-06, "loss": 0.0013, "step": 95700 }, { "epoch": 1.5660639777468708, "grad_norm": 0.09268709272146225, "learning_rate": 7.627120742595535e-06, "loss": 0.0018, "step": 95710 }, { "epoch": 1.5662276036979463, "grad_norm": 0.019940614700317383, "learning_rate": 7.626513255814085e-06, "loss": 0.002, "step": 95720 }, { "epoch": 1.5663912296490223, "grad_norm": 0.10306815803050995, "learning_rate": 7.625905715480263e-06, "loss": 0.0024, "step": 95730 }, { "epoch": 1.5665548556000983, "grad_norm": 0.051205866038799286, "learning_rate": 7.625298121606457e-06, "loss": 0.0011, "step": 95740 }, { "epoch": 1.5667184815511739, "grad_norm": 0.04914378374814987, "learning_rate": 7.624690474205054e-06, "loss": 0.0013, "step": 95750 }, { "epoch": 1.5668821075022499, "grad_norm": 0.05056005343794823, "learning_rate": 7.624082773288446e-06, "loss": 0.0012, "step": 95760 }, { "epoch": 1.5670457334533257, "grad_norm": 0.003144304035231471, "learning_rate": 7.62347501886902e-06, "loss": 0.0013, "step": 95770 }, { "epoch": 1.5672093594044014, "grad_norm": 0.04980071261525154, "learning_rate": 7.622867210959171e-06, "loss": 0.0023, "step": 95780 }, { "epoch": 1.5673729853554774, "grad_norm": 0.04100455716252327, "learning_rate": 7.62225934957129e-06, "loss": 0.0019, "step": 95790 }, { "epoch": 1.5675366113065532, "grad_norm": 0.06226328760385513, "learning_rate": 7.6216514347177715e-06, "loss": 0.0011, "step": 95800 }, { "epoch": 1.567700237257629, "grad_norm": 0.009659329429268837, "learning_rate": 7.621043466411009e-06, "loss": 0.0017, "step": 95810 }, { "epoch": 1.567863863208705, "grad_norm": 0.11078331619501114, "learning_rate": 7.6204354446634e-06, "loss": 0.0017, "step": 95820 }, { "epoch": 1.5680274891597807, "grad_norm": 0.010740530677139759, "learning_rate": 7.61982736948734e-06, "loss": 0.0017, "step": 95830 }, { "epoch": 1.5681911151108565, "grad_norm": 0.043703317642211914, "learning_rate": 7.6192192408952284e-06, "loss": 0.0013, "step": 95840 }, { "epoch": 1.5683547410619325, "grad_norm": 0.07610243558883667, "learning_rate": 7.618611058899463e-06, "loss": 0.0028, "step": 95850 }, { "epoch": 1.5685183670130083, "grad_norm": 0.04996560513973236, "learning_rate": 7.618002823512447e-06, "loss": 0.003, "step": 95860 }, { "epoch": 1.568681992964084, "grad_norm": 0.0985526517033577, "learning_rate": 7.617394534746579e-06, "loss": 0.001, "step": 95870 }, { "epoch": 1.56884561891516, "grad_norm": 0.02669636718928814, "learning_rate": 7.616786192614264e-06, "loss": 0.0019, "step": 95880 }, { "epoch": 1.5690092448662358, "grad_norm": 0.035400837659835815, "learning_rate": 7.616177797127902e-06, "loss": 0.001, "step": 95890 }, { "epoch": 1.5691728708173116, "grad_norm": 0.039998799562454224, "learning_rate": 7.6155693482999e-06, "loss": 0.0015, "step": 95900 }, { "epoch": 1.5693364967683876, "grad_norm": 0.08561044931411743, "learning_rate": 7.614960846142665e-06, "loss": 0.002, "step": 95910 }, { "epoch": 1.5695001227194632, "grad_norm": 0.10250250995159149, "learning_rate": 7.614352290668601e-06, "loss": 0.0014, "step": 95920 }, { "epoch": 1.5696637486705391, "grad_norm": 0.055771131068468094, "learning_rate": 7.613743681890118e-06, "loss": 0.0022, "step": 95930 }, { "epoch": 1.5698273746216151, "grad_norm": 0.0323370024561882, "learning_rate": 7.613135019819623e-06, "loss": 0.0019, "step": 95940 }, { "epoch": 1.5699910005726907, "grad_norm": 0.045190148055553436, "learning_rate": 7.612526304469527e-06, "loss": 0.001, "step": 95950 }, { "epoch": 1.5701546265237667, "grad_norm": 0.13067862391471863, "learning_rate": 7.6119175358522436e-06, "loss": 0.002, "step": 95960 }, { "epoch": 1.5703182524748425, "grad_norm": 0.005714109167456627, "learning_rate": 7.6113087139801815e-06, "loss": 0.0025, "step": 95970 }, { "epoch": 1.5704818784259182, "grad_norm": 0.038285985589027405, "learning_rate": 7.610699838865756e-06, "loss": 0.0019, "step": 95980 }, { "epoch": 1.5706455043769942, "grad_norm": 0.05877935513854027, "learning_rate": 7.6100909105213814e-06, "loss": 0.0007, "step": 95990 }, { "epoch": 1.57080913032807, "grad_norm": 0.062130458652973175, "learning_rate": 7.609481928959473e-06, "loss": 0.0013, "step": 96000 }, { "epoch": 1.5709727562791458, "grad_norm": 0.044137947261333466, "learning_rate": 7.608872894192446e-06, "loss": 0.0024, "step": 96010 }, { "epoch": 1.5711363822302218, "grad_norm": 0.05075130984187126, "learning_rate": 7.6082638062327205e-06, "loss": 0.0021, "step": 96020 }, { "epoch": 1.5713000081812976, "grad_norm": 0.07550398260354996, "learning_rate": 7.607654665092714e-06, "loss": 0.0042, "step": 96030 }, { "epoch": 1.5714636341323733, "grad_norm": 0.011026551946997643, "learning_rate": 7.607045470784847e-06, "loss": 0.0015, "step": 96040 }, { "epoch": 1.5716272600834493, "grad_norm": 0.029042119160294533, "learning_rate": 7.606436223321541e-06, "loss": 0.0009, "step": 96050 }, { "epoch": 1.571790886034525, "grad_norm": 0.05910797789692879, "learning_rate": 7.605826922715216e-06, "loss": 0.0012, "step": 96060 }, { "epoch": 1.5719545119856009, "grad_norm": 0.009520908817648888, "learning_rate": 7.605217568978296e-06, "loss": 0.0014, "step": 96070 }, { "epoch": 1.5721181379366769, "grad_norm": 0.1223711222410202, "learning_rate": 7.604608162123205e-06, "loss": 0.0029, "step": 96080 }, { "epoch": 1.5722817638877526, "grad_norm": 0.0409405380487442, "learning_rate": 7.6039987021623696e-06, "loss": 0.0012, "step": 96090 }, { "epoch": 1.5724453898388284, "grad_norm": 0.017062589526176453, "learning_rate": 7.603389189108215e-06, "loss": 0.0008, "step": 96100 }, { "epoch": 1.5726090157899044, "grad_norm": 0.027661586180329323, "learning_rate": 7.602779622973169e-06, "loss": 0.0015, "step": 96110 }, { "epoch": 1.57277264174098, "grad_norm": 0.029355905950069427, "learning_rate": 7.60217000376966e-06, "loss": 0.0011, "step": 96120 }, { "epoch": 1.572936267692056, "grad_norm": 0.08017967641353607, "learning_rate": 7.601560331510118e-06, "loss": 0.0025, "step": 96130 }, { "epoch": 1.573099893643132, "grad_norm": 0.044624339789152145, "learning_rate": 7.600950606206971e-06, "loss": 0.0015, "step": 96140 }, { "epoch": 1.5732635195942075, "grad_norm": 0.04555397108197212, "learning_rate": 7.600340827872656e-06, "loss": 0.001, "step": 96150 }, { "epoch": 1.5734271455452835, "grad_norm": 0.01820327714085579, "learning_rate": 7.5997309965196035e-06, "loss": 0.0022, "step": 96160 }, { "epoch": 1.5735907714963593, "grad_norm": 0.1528116762638092, "learning_rate": 7.599121112160246e-06, "loss": 0.0018, "step": 96170 }, { "epoch": 1.573754397447435, "grad_norm": 0.14122474193572998, "learning_rate": 7.5985111748070185e-06, "loss": 0.002, "step": 96180 }, { "epoch": 1.573918023398511, "grad_norm": 0.00632710475474596, "learning_rate": 7.597901184472358e-06, "loss": 0.0015, "step": 96190 }, { "epoch": 1.5740816493495868, "grad_norm": 0.0634409561753273, "learning_rate": 7.597291141168704e-06, "loss": 0.003, "step": 96200 }, { "epoch": 1.5742452753006626, "grad_norm": 0.08886983245611191, "learning_rate": 7.596681044908492e-06, "loss": 0.0021, "step": 96210 }, { "epoch": 1.5744089012517386, "grad_norm": 0.02427397109568119, "learning_rate": 7.59607089570416e-06, "loss": 0.0016, "step": 96220 }, { "epoch": 1.5745725272028144, "grad_norm": 0.03700042888522148, "learning_rate": 7.595460693568151e-06, "loss": 0.0016, "step": 96230 }, { "epoch": 1.5747361531538902, "grad_norm": 0.11368055641651154, "learning_rate": 7.594850438512906e-06, "loss": 0.0017, "step": 96240 }, { "epoch": 1.5748997791049661, "grad_norm": 0.07211928814649582, "learning_rate": 7.594240130550868e-06, "loss": 0.0015, "step": 96250 }, { "epoch": 1.575063405056042, "grad_norm": 0.003500535385683179, "learning_rate": 7.593629769694479e-06, "loss": 0.0015, "step": 96260 }, { "epoch": 1.5752270310071177, "grad_norm": 0.0956205502152443, "learning_rate": 7.593019355956186e-06, "loss": 0.0016, "step": 96270 }, { "epoch": 1.5753906569581937, "grad_norm": 0.119269460439682, "learning_rate": 7.592408889348433e-06, "loss": 0.0015, "step": 96280 }, { "epoch": 1.5755542829092692, "grad_norm": 0.09693964570760727, "learning_rate": 7.591798369883667e-06, "loss": 0.002, "step": 96290 }, { "epoch": 1.5757179088603452, "grad_norm": 0.05186380073428154, "learning_rate": 7.591187797574337e-06, "loss": 0.0023, "step": 96300 }, { "epoch": 1.5758815348114212, "grad_norm": 0.09334087371826172, "learning_rate": 7.590577172432893e-06, "loss": 0.0017, "step": 96310 }, { "epoch": 1.5760451607624968, "grad_norm": 0.032498862594366074, "learning_rate": 7.589966494471781e-06, "loss": 0.0036, "step": 96320 }, { "epoch": 1.5762087867135728, "grad_norm": 0.043655168265104294, "learning_rate": 7.589355763703457e-06, "loss": 0.0007, "step": 96330 }, { "epoch": 1.5763724126646486, "grad_norm": 0.04537096992135048, "learning_rate": 7.58874498014037e-06, "loss": 0.0013, "step": 96340 }, { "epoch": 1.5765360386157243, "grad_norm": 0.11837664246559143, "learning_rate": 7.588134143794974e-06, "loss": 0.0015, "step": 96350 }, { "epoch": 1.5766996645668003, "grad_norm": 0.09860668331384659, "learning_rate": 7.587523254679725e-06, "loss": 0.0013, "step": 96360 }, { "epoch": 1.576863290517876, "grad_norm": 0.03669103607535362, "learning_rate": 7.5869123128070775e-06, "loss": 0.0022, "step": 96370 }, { "epoch": 1.5770269164689519, "grad_norm": 0.10440675169229507, "learning_rate": 7.5863013181894885e-06, "loss": 0.0028, "step": 96380 }, { "epoch": 1.5771905424200279, "grad_norm": 0.045767512172460556, "learning_rate": 7.5856902708394144e-06, "loss": 0.0011, "step": 96390 }, { "epoch": 1.5773541683711036, "grad_norm": 0.11971075087785721, "learning_rate": 7.5850791707693135e-06, "loss": 0.002, "step": 96400 }, { "epoch": 1.5775177943221794, "grad_norm": 0.4229120910167694, "learning_rate": 7.584468017991649e-06, "loss": 0.0022, "step": 96410 }, { "epoch": 1.5776814202732554, "grad_norm": 0.06951602548360825, "learning_rate": 7.58385681251888e-06, "loss": 0.0021, "step": 96420 }, { "epoch": 1.5778450462243312, "grad_norm": 0.02388356439769268, "learning_rate": 7.583245554363466e-06, "loss": 0.0047, "step": 96430 }, { "epoch": 1.578008672175407, "grad_norm": 0.03697299212217331, "learning_rate": 7.582634243537872e-06, "loss": 0.0019, "step": 96440 }, { "epoch": 1.578172298126483, "grad_norm": 0.0026176422834396362, "learning_rate": 7.582022880054563e-06, "loss": 0.0011, "step": 96450 }, { "epoch": 1.5783359240775587, "grad_norm": 0.0881384089589119, "learning_rate": 7.581411463926004e-06, "loss": 0.0018, "step": 96460 }, { "epoch": 1.5784995500286345, "grad_norm": 0.014557965099811554, "learning_rate": 7.580799995164659e-06, "loss": 0.0021, "step": 96470 }, { "epoch": 1.5786631759797105, "grad_norm": 0.0663168877363205, "learning_rate": 7.580188473782999e-06, "loss": 0.0015, "step": 96480 }, { "epoch": 1.578826801930786, "grad_norm": 0.131452739238739, "learning_rate": 7.579576899793489e-06, "loss": 0.0025, "step": 96490 }, { "epoch": 1.578990427881862, "grad_norm": 0.035973306745290756, "learning_rate": 7.578965273208601e-06, "loss": 0.0012, "step": 96500 }, { "epoch": 1.579154053832938, "grad_norm": 0.05295603349804878, "learning_rate": 7.5783535940408034e-06, "loss": 0.0017, "step": 96510 }, { "epoch": 1.5793176797840136, "grad_norm": 0.02661389671266079, "learning_rate": 7.57774186230257e-06, "loss": 0.0014, "step": 96520 }, { "epoch": 1.5794813057350896, "grad_norm": 0.030932430177927017, "learning_rate": 7.577130078006371e-06, "loss": 0.0012, "step": 96530 }, { "epoch": 1.5796449316861654, "grad_norm": 0.03770974278450012, "learning_rate": 7.576518241164683e-06, "loss": 0.002, "step": 96540 }, { "epoch": 1.5798085576372412, "grad_norm": 0.05094987899065018, "learning_rate": 7.575906351789976e-06, "loss": 0.0014, "step": 96550 }, { "epoch": 1.5799721835883171, "grad_norm": 0.058016274124383926, "learning_rate": 7.575294409894733e-06, "loss": 0.0012, "step": 96560 }, { "epoch": 1.580135809539393, "grad_norm": 0.2191055417060852, "learning_rate": 7.5746824154914256e-06, "loss": 0.0015, "step": 96570 }, { "epoch": 1.5802994354904687, "grad_norm": 0.04098830372095108, "learning_rate": 7.574070368592534e-06, "loss": 0.0014, "step": 96580 }, { "epoch": 1.5804630614415447, "grad_norm": 0.0688880980014801, "learning_rate": 7.5734582692105365e-06, "loss": 0.0015, "step": 96590 }, { "epoch": 1.5806266873926205, "grad_norm": 0.0797303169965744, "learning_rate": 7.572846117357914e-06, "loss": 0.0007, "step": 96600 }, { "epoch": 1.5807903133436962, "grad_norm": 0.14297746121883392, "learning_rate": 7.572233913047146e-06, "loss": 0.0016, "step": 96610 }, { "epoch": 1.5809539392947722, "grad_norm": 0.2402307391166687, "learning_rate": 7.571621656290717e-06, "loss": 0.0015, "step": 96620 }, { "epoch": 1.581117565245848, "grad_norm": 0.054886750876903534, "learning_rate": 7.571009347101108e-06, "loss": 0.0011, "step": 96630 }, { "epoch": 1.5812811911969238, "grad_norm": 0.06332045793533325, "learning_rate": 7.570396985490808e-06, "loss": 0.0012, "step": 96640 }, { "epoch": 1.5814448171479998, "grad_norm": 0.04010794684290886, "learning_rate": 7.569784571472297e-06, "loss": 0.0013, "step": 96650 }, { "epoch": 1.5816084430990756, "grad_norm": 0.026637444272637367, "learning_rate": 7.569172105058064e-06, "loss": 0.0013, "step": 96660 }, { "epoch": 1.5817720690501513, "grad_norm": 0.09926632046699524, "learning_rate": 7.5685595862605986e-06, "loss": 0.0017, "step": 96670 }, { "epoch": 1.5819356950012273, "grad_norm": 0.027483142912387848, "learning_rate": 7.5679470150923876e-06, "loss": 0.0028, "step": 96680 }, { "epoch": 1.5820993209523029, "grad_norm": 0.10061636567115784, "learning_rate": 7.56733439156592e-06, "loss": 0.0035, "step": 96690 }, { "epoch": 1.5822629469033789, "grad_norm": 0.038149893283843994, "learning_rate": 7.566721715693688e-06, "loss": 0.0015, "step": 96700 }, { "epoch": 1.5824265728544549, "grad_norm": 0.046306680887937546, "learning_rate": 7.566108987488183e-06, "loss": 0.0017, "step": 96710 }, { "epoch": 1.5825901988055304, "grad_norm": 0.04563567787408829, "learning_rate": 7.565496206961897e-06, "loss": 0.0029, "step": 96720 }, { "epoch": 1.5827538247566064, "grad_norm": 0.047138869762420654, "learning_rate": 7.564883374127328e-06, "loss": 0.0011, "step": 96730 }, { "epoch": 1.5829174507076822, "grad_norm": 0.11733153462409973, "learning_rate": 7.564270488996966e-06, "loss": 0.0019, "step": 96740 }, { "epoch": 1.583081076658758, "grad_norm": 0.07621563225984573, "learning_rate": 7.56365755158331e-06, "loss": 0.001, "step": 96750 }, { "epoch": 1.583244702609834, "grad_norm": 0.06529466807842255, "learning_rate": 7.5630445618988566e-06, "loss": 0.0016, "step": 96760 }, { "epoch": 1.5834083285609097, "grad_norm": 0.06150686740875244, "learning_rate": 7.562431519956106e-06, "loss": 0.0011, "step": 96770 }, { "epoch": 1.5835719545119855, "grad_norm": 0.060907963663339615, "learning_rate": 7.561818425767553e-06, "loss": 0.001, "step": 96780 }, { "epoch": 1.5837355804630615, "grad_norm": 0.015234087593853474, "learning_rate": 7.561205279345704e-06, "loss": 0.0015, "step": 96790 }, { "epoch": 1.5838992064141373, "grad_norm": 0.047553371638059616, "learning_rate": 7.560592080703055e-06, "loss": 0.0016, "step": 96800 }, { "epoch": 1.584062832365213, "grad_norm": 0.00795611459761858, "learning_rate": 7.559978829852114e-06, "loss": 0.0016, "step": 96810 }, { "epoch": 1.584226458316289, "grad_norm": 0.03940999507904053, "learning_rate": 7.55936552680538e-06, "loss": 0.0019, "step": 96820 }, { "epoch": 1.5843900842673648, "grad_norm": 0.05226863548159599, "learning_rate": 7.558752171575361e-06, "loss": 0.0008, "step": 96830 }, { "epoch": 1.5845537102184406, "grad_norm": 0.050951309502124786, "learning_rate": 7.558138764174558e-06, "loss": 0.0015, "step": 96840 }, { "epoch": 1.5847173361695166, "grad_norm": 0.030449477955698967, "learning_rate": 7.557525304615485e-06, "loss": 0.0014, "step": 96850 }, { "epoch": 1.5848809621205924, "grad_norm": 0.03242103382945061, "learning_rate": 7.556911792910644e-06, "loss": 0.0008, "step": 96860 }, { "epoch": 1.5850445880716681, "grad_norm": 0.11120728403329849, "learning_rate": 7.556298229072547e-06, "loss": 0.0017, "step": 96870 }, { "epoch": 1.5852082140227441, "grad_norm": 0.07419583946466446, "learning_rate": 7.555684613113703e-06, "loss": 0.0013, "step": 96880 }, { "epoch": 1.5853718399738197, "grad_norm": 0.0312582291662693, "learning_rate": 7.555070945046623e-06, "loss": 0.0013, "step": 96890 }, { "epoch": 1.5855354659248957, "grad_norm": 0.038740549236536026, "learning_rate": 7.55445722488382e-06, "loss": 0.0009, "step": 96900 }, { "epoch": 1.5856990918759717, "grad_norm": 0.050400711596012115, "learning_rate": 7.5538434526378066e-06, "loss": 0.0019, "step": 96910 }, { "epoch": 1.5858627178270472, "grad_norm": 0.02395164966583252, "learning_rate": 7.553229628321097e-06, "loss": 0.0009, "step": 96920 }, { "epoch": 1.5860263437781232, "grad_norm": 0.06549564749002457, "learning_rate": 7.5526157519462065e-06, "loss": 0.0023, "step": 96930 }, { "epoch": 1.586189969729199, "grad_norm": 0.09685604274272919, "learning_rate": 7.552001823525652e-06, "loss": 0.0013, "step": 96940 }, { "epoch": 1.5863535956802748, "grad_norm": 0.171199232339859, "learning_rate": 7.5513878430719515e-06, "loss": 0.0016, "step": 96950 }, { "epoch": 1.5865172216313508, "grad_norm": 0.03943144157528877, "learning_rate": 7.55077381059762e-06, "loss": 0.0009, "step": 96960 }, { "epoch": 1.5866808475824266, "grad_norm": 0.04320184141397476, "learning_rate": 7.550159726115183e-06, "loss": 0.0015, "step": 96970 }, { "epoch": 1.5868444735335023, "grad_norm": 0.032532691955566406, "learning_rate": 7.549545589637156e-06, "loss": 0.0018, "step": 96980 }, { "epoch": 1.5870080994845783, "grad_norm": 0.2647916078567505, "learning_rate": 7.548931401176064e-06, "loss": 0.0058, "step": 96990 }, { "epoch": 1.587171725435654, "grad_norm": 0.06998138129711151, "learning_rate": 7.548317160744427e-06, "loss": 0.0016, "step": 97000 }, { "epoch": 1.5873353513867299, "grad_norm": 0.07784351706504822, "learning_rate": 7.5477028683547725e-06, "loss": 0.0016, "step": 97010 }, { "epoch": 1.5874989773378059, "grad_norm": 0.03436139598488808, "learning_rate": 7.547088524019622e-06, "loss": 0.0019, "step": 97020 }, { "epoch": 1.5876626032888816, "grad_norm": 0.12133406102657318, "learning_rate": 7.546474127751503e-06, "loss": 0.0019, "step": 97030 }, { "epoch": 1.5878262292399574, "grad_norm": 0.09672055393457413, "learning_rate": 7.545859679562942e-06, "loss": 0.0015, "step": 97040 }, { "epoch": 1.5879898551910334, "grad_norm": 0.01696709543466568, "learning_rate": 7.545245179466467e-06, "loss": 0.0013, "step": 97050 }, { "epoch": 1.5881534811421092, "grad_norm": 0.016474004834890366, "learning_rate": 7.544630627474609e-06, "loss": 0.0018, "step": 97060 }, { "epoch": 1.588317107093185, "grad_norm": 0.07022275030612946, "learning_rate": 7.544016023599894e-06, "loss": 0.0018, "step": 97070 }, { "epoch": 1.588480733044261, "grad_norm": 0.03573715314269066, "learning_rate": 7.543401367854859e-06, "loss": 0.0018, "step": 97080 }, { "epoch": 1.5886443589953365, "grad_norm": 0.0432613231241703, "learning_rate": 7.542786660252031e-06, "loss": 0.0024, "step": 97090 }, { "epoch": 1.5888079849464125, "grad_norm": 0.027387361973524094, "learning_rate": 7.542171900803945e-06, "loss": 0.0011, "step": 97100 }, { "epoch": 1.5889716108974885, "grad_norm": 0.12895727157592773, "learning_rate": 7.541557089523137e-06, "loss": 0.0029, "step": 97110 }, { "epoch": 1.589135236848564, "grad_norm": 0.06346653401851654, "learning_rate": 7.540942226422143e-06, "loss": 0.0012, "step": 97120 }, { "epoch": 1.58929886279964, "grad_norm": 0.06495916098356247, "learning_rate": 7.540327311513495e-06, "loss": 0.0017, "step": 97130 }, { "epoch": 1.5894624887507158, "grad_norm": 0.04263191297650337, "learning_rate": 7.5397123448097354e-06, "loss": 0.0019, "step": 97140 }, { "epoch": 1.5896261147017916, "grad_norm": 0.08372174203395844, "learning_rate": 7.539097326323402e-06, "loss": 0.001, "step": 97150 }, { "epoch": 1.5897897406528676, "grad_norm": 0.05393011495471001, "learning_rate": 7.538482256067032e-06, "loss": 0.0014, "step": 97160 }, { "epoch": 1.5899533666039434, "grad_norm": 0.04073631763458252, "learning_rate": 7.537867134053167e-06, "loss": 0.0021, "step": 97170 }, { "epoch": 1.5901169925550191, "grad_norm": 0.05929417163133621, "learning_rate": 7.537251960294352e-06, "loss": 0.0017, "step": 97180 }, { "epoch": 1.5902806185060951, "grad_norm": 0.05434208735823631, "learning_rate": 7.536636734803124e-06, "loss": 0.0017, "step": 97190 }, { "epoch": 1.590444244457171, "grad_norm": 0.04036393761634827, "learning_rate": 7.536021457592033e-06, "loss": 0.002, "step": 97200 }, { "epoch": 1.5906078704082467, "grad_norm": 0.03686368092894554, "learning_rate": 7.53540612867362e-06, "loss": 0.0013, "step": 97210 }, { "epoch": 1.5907714963593227, "grad_norm": 0.003402507631108165, "learning_rate": 7.534790748060434e-06, "loss": 0.0013, "step": 97220 }, { "epoch": 1.5909351223103985, "grad_norm": 0.057339172810316086, "learning_rate": 7.5341753157650174e-06, "loss": 0.0061, "step": 97230 }, { "epoch": 1.5910987482614742, "grad_norm": 0.09347930550575256, "learning_rate": 7.533559831799925e-06, "loss": 0.0015, "step": 97240 }, { "epoch": 1.5912623742125502, "grad_norm": 0.09675119817256927, "learning_rate": 7.5329442961777e-06, "loss": 0.0013, "step": 97250 }, { "epoch": 1.5914260001636258, "grad_norm": 0.0859195739030838, "learning_rate": 7.532328708910897e-06, "loss": 0.0013, "step": 97260 }, { "epoch": 1.5915896261147018, "grad_norm": 0.05596604570746422, "learning_rate": 7.531713070012062e-06, "loss": 0.0029, "step": 97270 }, { "epoch": 1.5917532520657778, "grad_norm": 0.035631418228149414, "learning_rate": 7.531097379493752e-06, "loss": 0.0011, "step": 97280 }, { "epoch": 1.5919168780168533, "grad_norm": 0.03365952894091606, "learning_rate": 7.530481637368521e-06, "loss": 0.001, "step": 97290 }, { "epoch": 1.5920805039679293, "grad_norm": 0.05644739046692848, "learning_rate": 7.52986584364892e-06, "loss": 0.0018, "step": 97300 }, { "epoch": 1.592244129919005, "grad_norm": 0.09474650025367737, "learning_rate": 7.529249998347506e-06, "loss": 0.0017, "step": 97310 }, { "epoch": 1.5924077558700809, "grad_norm": 0.02294088341295719, "learning_rate": 7.528634101476835e-06, "loss": 0.0021, "step": 97320 }, { "epoch": 1.5925713818211569, "grad_norm": 0.05151808261871338, "learning_rate": 7.528018153049466e-06, "loss": 0.0017, "step": 97330 }, { "epoch": 1.5927350077722326, "grad_norm": 0.09321659803390503, "learning_rate": 7.5274021530779565e-06, "loss": 0.0022, "step": 97340 }, { "epoch": 1.5928986337233084, "grad_norm": 0.02271154150366783, "learning_rate": 7.526786101574868e-06, "loss": 0.0014, "step": 97350 }, { "epoch": 1.5930622596743844, "grad_norm": 0.015305254608392715, "learning_rate": 7.526169998552757e-06, "loss": 0.0012, "step": 97360 }, { "epoch": 1.5932258856254602, "grad_norm": 0.024423880502581596, "learning_rate": 7.5255538440241915e-06, "loss": 0.0007, "step": 97370 }, { "epoch": 1.593389511576536, "grad_norm": 0.053025297820568085, "learning_rate": 7.524937638001728e-06, "loss": 0.0014, "step": 97380 }, { "epoch": 1.593553137527612, "grad_norm": 0.06446052342653275, "learning_rate": 7.524321380497935e-06, "loss": 0.0008, "step": 97390 }, { "epoch": 1.5937167634786877, "grad_norm": 0.015600229613482952, "learning_rate": 7.5237050715253755e-06, "loss": 0.0015, "step": 97400 }, { "epoch": 1.5938803894297635, "grad_norm": 0.03927970305085182, "learning_rate": 7.523088711096617e-06, "loss": 0.0018, "step": 97410 }, { "epoch": 1.5940440153808395, "grad_norm": 0.09979335963726044, "learning_rate": 7.522472299224224e-06, "loss": 0.001, "step": 97420 }, { "epoch": 1.5942076413319153, "grad_norm": 0.09032167494297028, "learning_rate": 7.521855835920767e-06, "loss": 0.0015, "step": 97430 }, { "epoch": 1.594371267282991, "grad_norm": 0.05092831701040268, "learning_rate": 7.521239321198813e-06, "loss": 0.0039, "step": 97440 }, { "epoch": 1.594534893234067, "grad_norm": 0.03987281396985054, "learning_rate": 7.520622755070936e-06, "loss": 0.0012, "step": 97450 }, { "epoch": 1.5946985191851426, "grad_norm": 0.013599294237792492, "learning_rate": 7.520006137549702e-06, "loss": 0.0013, "step": 97460 }, { "epoch": 1.5948621451362186, "grad_norm": 0.17364448308944702, "learning_rate": 7.519389468647688e-06, "loss": 0.0022, "step": 97470 }, { "epoch": 1.5950257710872946, "grad_norm": 0.017796987667679787, "learning_rate": 7.518772748377463e-06, "loss": 0.0017, "step": 97480 }, { "epoch": 1.5951893970383701, "grad_norm": 0.04090600833296776, "learning_rate": 7.518155976751607e-06, "loss": 0.0009, "step": 97490 }, { "epoch": 1.5953530229894461, "grad_norm": 0.027468515560030937, "learning_rate": 7.5175391537826894e-06, "loss": 0.0012, "step": 97500 }, { "epoch": 1.595516648940522, "grad_norm": 0.06710810959339142, "learning_rate": 7.516922279483292e-06, "loss": 0.0016, "step": 97510 }, { "epoch": 1.5956802748915977, "grad_norm": 0.21976561844348907, "learning_rate": 7.516305353865988e-06, "loss": 0.0014, "step": 97520 }, { "epoch": 1.5958439008426737, "grad_norm": 0.07232493907213211, "learning_rate": 7.5156883769433585e-06, "loss": 0.003, "step": 97530 }, { "epoch": 1.5960075267937495, "grad_norm": 0.030652115121483803, "learning_rate": 7.5150713487279826e-06, "loss": 0.0012, "step": 97540 }, { "epoch": 1.5961711527448252, "grad_norm": 0.06631026417016983, "learning_rate": 7.514454269232441e-06, "loss": 0.0031, "step": 97550 }, { "epoch": 1.5963347786959012, "grad_norm": 0.11902723461389542, "learning_rate": 7.513837138469315e-06, "loss": 0.0017, "step": 97560 }, { "epoch": 1.596498404646977, "grad_norm": 0.08044162392616272, "learning_rate": 7.513219956451188e-06, "loss": 0.0011, "step": 97570 }, { "epoch": 1.5966620305980528, "grad_norm": 0.07868382334709167, "learning_rate": 7.512602723190643e-06, "loss": 0.0016, "step": 97580 }, { "epoch": 1.5968256565491288, "grad_norm": 0.04440293833613396, "learning_rate": 7.511985438700267e-06, "loss": 0.0019, "step": 97590 }, { "epoch": 1.5969892825002046, "grad_norm": 0.05265267193317413, "learning_rate": 7.5113681029926425e-06, "loss": 0.0016, "step": 97600 }, { "epoch": 1.5971529084512803, "grad_norm": 0.11392497271299362, "learning_rate": 7.5107507160803585e-06, "loss": 0.0028, "step": 97610 }, { "epoch": 1.5973165344023563, "grad_norm": 0.062086135149002075, "learning_rate": 7.510133277976002e-06, "loss": 0.0028, "step": 97620 }, { "epoch": 1.597480160353432, "grad_norm": 0.1317416876554489, "learning_rate": 7.509515788692166e-06, "loss": 0.0014, "step": 97630 }, { "epoch": 1.5976437863045079, "grad_norm": 0.059186406433582306, "learning_rate": 7.508898248241433e-06, "loss": 0.0017, "step": 97640 }, { "epoch": 1.5978074122555839, "grad_norm": 0.05603529140353203, "learning_rate": 7.508280656636402e-06, "loss": 0.0019, "step": 97650 }, { "epoch": 1.5979710382066594, "grad_norm": 0.0849769115447998, "learning_rate": 7.50766301388966e-06, "loss": 0.0024, "step": 97660 }, { "epoch": 1.5981346641577354, "grad_norm": 0.09577538073062897, "learning_rate": 7.507045320013802e-06, "loss": 0.002, "step": 97670 }, { "epoch": 1.5982982901088114, "grad_norm": 0.03894062340259552, "learning_rate": 7.506427575021422e-06, "loss": 0.0017, "step": 97680 }, { "epoch": 1.598461916059887, "grad_norm": 0.010592583566904068, "learning_rate": 7.505809778925115e-06, "loss": 0.0014, "step": 97690 }, { "epoch": 1.598625542010963, "grad_norm": 0.020324500277638435, "learning_rate": 7.505191931737479e-06, "loss": 0.0014, "step": 97700 }, { "epoch": 1.5987891679620387, "grad_norm": 0.11384397000074387, "learning_rate": 7.504574033471108e-06, "loss": 0.0024, "step": 97710 }, { "epoch": 1.5989527939131145, "grad_norm": 0.033329375088214874, "learning_rate": 7.503956084138604e-06, "loss": 0.0006, "step": 97720 }, { "epoch": 1.5991164198641905, "grad_norm": 0.09079685807228088, "learning_rate": 7.503338083752565e-06, "loss": 0.0014, "step": 97730 }, { "epoch": 1.5992800458152663, "grad_norm": 0.03950981795787811, "learning_rate": 7.502720032325592e-06, "loss": 0.0012, "step": 97740 }, { "epoch": 1.599443671766342, "grad_norm": 0.07293063402175903, "learning_rate": 7.502101929870284e-06, "loss": 0.0019, "step": 97750 }, { "epoch": 1.599607297717418, "grad_norm": 0.05209546908736229, "learning_rate": 7.501483776399248e-06, "loss": 0.0019, "step": 97760 }, { "epoch": 1.5997709236684938, "grad_norm": 0.13909591734409332, "learning_rate": 7.500865571925083e-06, "loss": 0.0022, "step": 97770 }, { "epoch": 1.5999345496195696, "grad_norm": 0.04464314877986908, "learning_rate": 7.500247316460399e-06, "loss": 0.0026, "step": 97780 }, { "epoch": 1.6000981755706456, "grad_norm": 0.10413773357868195, "learning_rate": 7.499629010017795e-06, "loss": 0.0013, "step": 97790 }, { "epoch": 1.6002618015217214, "grad_norm": 0.03155042976140976, "learning_rate": 7.499010652609884e-06, "loss": 0.0027, "step": 97800 }, { "epoch": 1.6004254274727971, "grad_norm": 0.07640814781188965, "learning_rate": 7.4983922442492686e-06, "loss": 0.002, "step": 97810 }, { "epoch": 1.6005890534238731, "grad_norm": 0.05439772084355354, "learning_rate": 7.497773784948562e-06, "loss": 0.0012, "step": 97820 }, { "epoch": 1.600752679374949, "grad_norm": 0.03836065158247948, "learning_rate": 7.497155274720371e-06, "loss": 0.0007, "step": 97830 }, { "epoch": 1.6009163053260247, "grad_norm": 0.03657205030322075, "learning_rate": 7.4965367135773095e-06, "loss": 0.0017, "step": 97840 }, { "epoch": 1.6010799312771007, "grad_norm": 0.0022615292109549046, "learning_rate": 7.495918101531987e-06, "loss": 0.0018, "step": 97850 }, { "epoch": 1.6012435572281762, "grad_norm": 0.05259415879845619, "learning_rate": 7.495299438597017e-06, "loss": 0.0015, "step": 97860 }, { "epoch": 1.6014071831792522, "grad_norm": 0.10789497196674347, "learning_rate": 7.494680724785013e-06, "loss": 0.0013, "step": 97870 }, { "epoch": 1.6015708091303282, "grad_norm": 0.02784569188952446, "learning_rate": 7.494061960108591e-06, "loss": 0.0016, "step": 97880 }, { "epoch": 1.6017344350814038, "grad_norm": 0.15667174756526947, "learning_rate": 7.4934431445803666e-06, "loss": 0.0019, "step": 97890 }, { "epoch": 1.6018980610324798, "grad_norm": 0.04010492563247681, "learning_rate": 7.4928242782129575e-06, "loss": 0.0026, "step": 97900 }, { "epoch": 1.6020616869835556, "grad_norm": 0.04074040800333023, "learning_rate": 7.492205361018981e-06, "loss": 0.0011, "step": 97910 }, { "epoch": 1.6022253129346313, "grad_norm": 0.018376655876636505, "learning_rate": 7.491586393011058e-06, "loss": 0.0017, "step": 97920 }, { "epoch": 1.6023889388857073, "grad_norm": 0.08320482075214386, "learning_rate": 7.490967374201805e-06, "loss": 0.0027, "step": 97930 }, { "epoch": 1.602552564836783, "grad_norm": 0.10502516478300095, "learning_rate": 7.490348304603848e-06, "loss": 0.0017, "step": 97940 }, { "epoch": 1.6027161907878589, "grad_norm": 0.08820626139640808, "learning_rate": 7.489729184229806e-06, "loss": 0.001, "step": 97950 }, { "epoch": 1.6028798167389349, "grad_norm": 0.15561166405677795, "learning_rate": 7.489110013092304e-06, "loss": 0.002, "step": 97960 }, { "epoch": 1.6030434426900106, "grad_norm": 0.07334393262863159, "learning_rate": 7.488490791203965e-06, "loss": 0.0012, "step": 97970 }, { "epoch": 1.6032070686410864, "grad_norm": 0.060561325401067734, "learning_rate": 7.487871518577417e-06, "loss": 0.0016, "step": 97980 }, { "epoch": 1.6033706945921624, "grad_norm": 0.07218711078166962, "learning_rate": 7.487252195225283e-06, "loss": 0.0045, "step": 97990 }, { "epoch": 1.6035343205432382, "grad_norm": 0.26336154341697693, "learning_rate": 7.486632821160192e-06, "loss": 0.0017, "step": 98000 }, { "epoch": 1.603697946494314, "grad_norm": 0.04951619356870651, "learning_rate": 7.486013396394772e-06, "loss": 0.0011, "step": 98010 }, { "epoch": 1.60386157244539, "grad_norm": 0.05844757333397865, "learning_rate": 7.485393920941654e-06, "loss": 0.003, "step": 98020 }, { "epoch": 1.6040251983964657, "grad_norm": 0.07719859480857849, "learning_rate": 7.4847743948134675e-06, "loss": 0.0011, "step": 98030 }, { "epoch": 1.6041888243475415, "grad_norm": 0.09140767902135849, "learning_rate": 7.484154818022844e-06, "loss": 0.0019, "step": 98040 }, { "epoch": 1.6043524502986175, "grad_norm": 0.050362810492515564, "learning_rate": 7.483535190582417e-06, "loss": 0.0013, "step": 98050 }, { "epoch": 1.604516076249693, "grad_norm": 0.08443980664014816, "learning_rate": 7.48291551250482e-06, "loss": 0.0013, "step": 98060 }, { "epoch": 1.604679702200769, "grad_norm": 0.1297031193971634, "learning_rate": 7.4822957838026865e-06, "loss": 0.0021, "step": 98070 }, { "epoch": 1.6048433281518448, "grad_norm": 0.019830184057354927, "learning_rate": 7.481676004488654e-06, "loss": 0.0009, "step": 98080 }, { "epoch": 1.6050069541029206, "grad_norm": 0.057336464524269104, "learning_rate": 7.481056174575357e-06, "loss": 0.0022, "step": 98090 }, { "epoch": 1.6051705800539966, "grad_norm": 0.013721815310418606, "learning_rate": 7.480436294075437e-06, "loss": 0.0016, "step": 98100 }, { "epoch": 1.6053342060050724, "grad_norm": 0.09373671561479568, "learning_rate": 7.479816363001528e-06, "loss": 0.0019, "step": 98110 }, { "epoch": 1.6054978319561481, "grad_norm": 0.13021855056285858, "learning_rate": 7.479196381366274e-06, "loss": 0.0015, "step": 98120 }, { "epoch": 1.6056614579072241, "grad_norm": 0.020709039643406868, "learning_rate": 7.478576349182314e-06, "loss": 0.002, "step": 98130 }, { "epoch": 1.6058250838583, "grad_norm": 0.00387133052572608, "learning_rate": 7.477956266462289e-06, "loss": 0.0012, "step": 98140 }, { "epoch": 1.6059887098093757, "grad_norm": 0.1661771982908249, "learning_rate": 7.477336133218845e-06, "loss": 0.0021, "step": 98150 }, { "epoch": 1.6061523357604517, "grad_norm": 0.029419196769595146, "learning_rate": 7.476715949464621e-06, "loss": 0.0016, "step": 98160 }, { "epoch": 1.6063159617115275, "grad_norm": 0.078350730240345, "learning_rate": 7.476095715212268e-06, "loss": 0.002, "step": 98170 }, { "epoch": 1.6064795876626032, "grad_norm": 0.08319169282913208, "learning_rate": 7.475475430474428e-06, "loss": 0.0016, "step": 98180 }, { "epoch": 1.6066432136136792, "grad_norm": 0.048313699662685394, "learning_rate": 7.474855095263751e-06, "loss": 0.0016, "step": 98190 }, { "epoch": 1.606806839564755, "grad_norm": 0.04137891158461571, "learning_rate": 7.4742347095928815e-06, "loss": 0.0017, "step": 98200 }, { "epoch": 1.6069704655158308, "grad_norm": 0.10837650299072266, "learning_rate": 7.473614273474472e-06, "loss": 0.0025, "step": 98210 }, { "epoch": 1.6071340914669068, "grad_norm": 0.04449353367090225, "learning_rate": 7.47299378692117e-06, "loss": 0.0006, "step": 98220 }, { "epoch": 1.6072977174179823, "grad_norm": 0.10643704235553741, "learning_rate": 7.472373249945628e-06, "loss": 0.0015, "step": 98230 }, { "epoch": 1.6074613433690583, "grad_norm": 0.05762813612818718, "learning_rate": 7.4717526625604984e-06, "loss": 0.001, "step": 98240 }, { "epoch": 1.6076249693201343, "grad_norm": 0.14971721172332764, "learning_rate": 7.471132024778434e-06, "loss": 0.0018, "step": 98250 }, { "epoch": 1.6077885952712099, "grad_norm": 0.03752509877085686, "learning_rate": 7.470511336612089e-06, "loss": 0.0021, "step": 98260 }, { "epoch": 1.6079522212222859, "grad_norm": 0.014623367227613926, "learning_rate": 7.46989059807412e-06, "loss": 0.0012, "step": 98270 }, { "epoch": 1.6081158471733616, "grad_norm": 0.01994745247066021, "learning_rate": 7.4692698091771805e-06, "loss": 0.0032, "step": 98280 }, { "epoch": 1.6082794731244374, "grad_norm": 0.026176495477557182, "learning_rate": 7.468648969933932e-06, "loss": 0.001, "step": 98290 }, { "epoch": 1.6084430990755134, "grad_norm": 0.022185277193784714, "learning_rate": 7.468028080357028e-06, "loss": 0.0018, "step": 98300 }, { "epoch": 1.6086067250265892, "grad_norm": 0.04538439214229584, "learning_rate": 7.467407140459133e-06, "loss": 0.0018, "step": 98310 }, { "epoch": 1.608770350977665, "grad_norm": 0.0745648518204689, "learning_rate": 7.466786150252903e-06, "loss": 0.0014, "step": 98320 }, { "epoch": 1.608933976928741, "grad_norm": 0.07164046168327332, "learning_rate": 7.4661651097510024e-06, "loss": 0.0009, "step": 98330 }, { "epoch": 1.6090976028798167, "grad_norm": 0.034459855407476425, "learning_rate": 7.465544018966091e-06, "loss": 0.0022, "step": 98340 }, { "epoch": 1.6092612288308925, "grad_norm": 0.08598585426807404, "learning_rate": 7.464922877910835e-06, "loss": 0.0015, "step": 98350 }, { "epoch": 1.6094248547819685, "grad_norm": 0.05755342170596123, "learning_rate": 7.464301686597898e-06, "loss": 0.0025, "step": 98360 }, { "epoch": 1.6095884807330443, "grad_norm": 0.06305579096078873, "learning_rate": 7.463680445039945e-06, "loss": 0.0012, "step": 98370 }, { "epoch": 1.60975210668412, "grad_norm": 0.043391354382038116, "learning_rate": 7.463059153249644e-06, "loss": 0.0009, "step": 98380 }, { "epoch": 1.609915732635196, "grad_norm": 0.004356960766017437, "learning_rate": 7.46243781123966e-06, "loss": 0.0011, "step": 98390 }, { "epoch": 1.6100793585862718, "grad_norm": 0.06220957636833191, "learning_rate": 7.461816419022664e-06, "loss": 0.0023, "step": 98400 }, { "epoch": 1.6102429845373476, "grad_norm": 0.07236689329147339, "learning_rate": 7.461194976611326e-06, "loss": 0.0022, "step": 98410 }, { "epoch": 1.6104066104884236, "grad_norm": 0.07854359596967697, "learning_rate": 7.460573484018314e-06, "loss": 0.0009, "step": 98420 }, { "epoch": 1.6105702364394991, "grad_norm": 0.2002670019865036, "learning_rate": 7.459951941256303e-06, "loss": 0.0016, "step": 98430 }, { "epoch": 1.6107338623905751, "grad_norm": 0.06855029612779617, "learning_rate": 7.459330348337963e-06, "loss": 0.0023, "step": 98440 }, { "epoch": 1.6108974883416511, "grad_norm": 0.026224976405501366, "learning_rate": 7.4587087052759675e-06, "loss": 0.0009, "step": 98450 }, { "epoch": 1.6110611142927267, "grad_norm": 0.04982713237404823, "learning_rate": 7.458087012082995e-06, "loss": 0.0014, "step": 98460 }, { "epoch": 1.6112247402438027, "grad_norm": 0.009353130124509335, "learning_rate": 7.4574652687717164e-06, "loss": 0.0011, "step": 98470 }, { "epoch": 1.6113883661948785, "grad_norm": 0.05399448052048683, "learning_rate": 7.456843475354813e-06, "loss": 0.0019, "step": 98480 }, { "epoch": 1.6115519921459542, "grad_norm": 0.1439037322998047, "learning_rate": 7.4562216318449595e-06, "loss": 0.0017, "step": 98490 }, { "epoch": 1.6117156180970302, "grad_norm": 0.1301651895046234, "learning_rate": 7.455599738254837e-06, "loss": 0.0018, "step": 98500 }, { "epoch": 1.611879244048106, "grad_norm": 0.052181243896484375, "learning_rate": 7.454977794597122e-06, "loss": 0.0029, "step": 98510 }, { "epoch": 1.6120428699991818, "grad_norm": 0.18218684196472168, "learning_rate": 7.4543558008845005e-06, "loss": 0.0015, "step": 98520 }, { "epoch": 1.6122064959502578, "grad_norm": 0.054925426840782166, "learning_rate": 7.45373375712965e-06, "loss": 0.0013, "step": 98530 }, { "epoch": 1.6123701219013336, "grad_norm": 0.03303828462958336, "learning_rate": 7.453111663345255e-06, "loss": 0.0015, "step": 98540 }, { "epoch": 1.6125337478524093, "grad_norm": 0.03691629320383072, "learning_rate": 7.452489519543999e-06, "loss": 0.0009, "step": 98550 }, { "epoch": 1.6126973738034853, "grad_norm": 0.05042186379432678, "learning_rate": 7.451867325738568e-06, "loss": 0.0011, "step": 98560 }, { "epoch": 1.612860999754561, "grad_norm": 0.12067291885614395, "learning_rate": 7.4512450819416485e-06, "loss": 0.0016, "step": 98570 }, { "epoch": 1.6130246257056369, "grad_norm": 0.0024502777960151434, "learning_rate": 7.450622788165926e-06, "loss": 0.0011, "step": 98580 }, { "epoch": 1.6131882516567129, "grad_norm": 0.022821133956313133, "learning_rate": 7.450000444424088e-06, "loss": 0.0011, "step": 98590 }, { "epoch": 1.6133518776077886, "grad_norm": 0.07431645691394806, "learning_rate": 7.449378050728826e-06, "loss": 0.0018, "step": 98600 }, { "epoch": 1.6135155035588644, "grad_norm": 0.037558916956186295, "learning_rate": 7.4487556070928265e-06, "loss": 0.0017, "step": 98610 }, { "epoch": 1.6136791295099404, "grad_norm": 0.04977196827530861, "learning_rate": 7.448133113528785e-06, "loss": 0.0017, "step": 98620 }, { "epoch": 1.613842755461016, "grad_norm": 0.14405842125415802, "learning_rate": 7.44751057004939e-06, "loss": 0.0015, "step": 98630 }, { "epoch": 1.614006381412092, "grad_norm": 0.09232313185930252, "learning_rate": 7.446887976667338e-06, "loss": 0.0012, "step": 98640 }, { "epoch": 1.614170007363168, "grad_norm": 0.007531360257416964, "learning_rate": 7.4462653333953184e-06, "loss": 0.002, "step": 98650 }, { "epoch": 1.6143336333142435, "grad_norm": 0.04131360352039337, "learning_rate": 7.445642640246032e-06, "loss": 0.0022, "step": 98660 }, { "epoch": 1.6144972592653195, "grad_norm": 0.1215822845697403, "learning_rate": 7.445019897232169e-06, "loss": 0.0015, "step": 98670 }, { "epoch": 1.6146608852163953, "grad_norm": 0.08304113894701004, "learning_rate": 7.444397104366432e-06, "loss": 0.0014, "step": 98680 }, { "epoch": 1.614824511167471, "grad_norm": 0.010378577746450901, "learning_rate": 7.443774261661515e-06, "loss": 0.0014, "step": 98690 }, { "epoch": 1.614988137118547, "grad_norm": 0.008497321046888828, "learning_rate": 7.4431513691301215e-06, "loss": 0.0032, "step": 98700 }, { "epoch": 1.6151517630696228, "grad_norm": 0.054241254925727844, "learning_rate": 7.442528426784947e-06, "loss": 0.0013, "step": 98710 }, { "epoch": 1.6153153890206986, "grad_norm": 0.03602859377861023, "learning_rate": 7.441905434638697e-06, "loss": 0.0024, "step": 98720 }, { "epoch": 1.6154790149717746, "grad_norm": 0.05911391228437424, "learning_rate": 7.4412823927040715e-06, "loss": 0.0011, "step": 98730 }, { "epoch": 1.6156426409228504, "grad_norm": 0.37296533584594727, "learning_rate": 7.4406593009937746e-06, "loss": 0.0028, "step": 98740 }, { "epoch": 1.6158062668739261, "grad_norm": 0.06103935465216637, "learning_rate": 7.44003615952051e-06, "loss": 0.0018, "step": 98750 }, { "epoch": 1.6159698928250021, "grad_norm": 0.06902666389942169, "learning_rate": 7.4394129682969815e-06, "loss": 0.0019, "step": 98760 }, { "epoch": 1.616133518776078, "grad_norm": 0.03352069854736328, "learning_rate": 7.4387897273359e-06, "loss": 0.0013, "step": 98770 }, { "epoch": 1.6162971447271537, "grad_norm": 0.04291786998510361, "learning_rate": 7.438166436649968e-06, "loss": 0.0008, "step": 98780 }, { "epoch": 1.6164607706782297, "grad_norm": 0.03118802048265934, "learning_rate": 7.437543096251898e-06, "loss": 0.001, "step": 98790 }, { "epoch": 1.6166243966293055, "grad_norm": 0.09489654004573822, "learning_rate": 7.4369197061543955e-06, "loss": 0.0016, "step": 98800 }, { "epoch": 1.6167880225803812, "grad_norm": 0.014625906944274902, "learning_rate": 7.436296266370173e-06, "loss": 0.0007, "step": 98810 }, { "epoch": 1.6169516485314572, "grad_norm": 0.10406310856342316, "learning_rate": 7.435672776911942e-06, "loss": 0.0017, "step": 98820 }, { "epoch": 1.6171152744825328, "grad_norm": 0.10965224355459213, "learning_rate": 7.435049237792415e-06, "loss": 0.0013, "step": 98830 }, { "epoch": 1.6172789004336088, "grad_norm": 0.2025170922279358, "learning_rate": 7.434425649024304e-06, "loss": 0.0015, "step": 98840 }, { "epoch": 1.6174425263846848, "grad_norm": 0.13034510612487793, "learning_rate": 7.433802010620326e-06, "loss": 0.0019, "step": 98850 }, { "epoch": 1.6176061523357603, "grad_norm": 0.09389202296733856, "learning_rate": 7.4331783225931934e-06, "loss": 0.002, "step": 98860 }, { "epoch": 1.6177697782868363, "grad_norm": 0.038213733583688736, "learning_rate": 7.4325545849556245e-06, "loss": 0.0029, "step": 98870 }, { "epoch": 1.617933404237912, "grad_norm": 0.016478557139635086, "learning_rate": 7.431930797720336e-06, "loss": 0.0013, "step": 98880 }, { "epoch": 1.6180970301889879, "grad_norm": 0.04322148114442825, "learning_rate": 7.431306960900048e-06, "loss": 0.0058, "step": 98890 }, { "epoch": 1.6182606561400639, "grad_norm": 0.11494992673397064, "learning_rate": 7.430683074507478e-06, "loss": 0.0024, "step": 98900 }, { "epoch": 1.6184242820911396, "grad_norm": 0.0400385707616806, "learning_rate": 7.430059138555348e-06, "loss": 0.0013, "step": 98910 }, { "epoch": 1.6185879080422154, "grad_norm": 0.0811285600066185, "learning_rate": 7.429435153056377e-06, "loss": 0.0018, "step": 98920 }, { "epoch": 1.6187515339932914, "grad_norm": 0.044487204402685165, "learning_rate": 7.4288111180232915e-06, "loss": 0.0012, "step": 98930 }, { "epoch": 1.6189151599443672, "grad_norm": 0.07258113473653793, "learning_rate": 7.428187033468811e-06, "loss": 0.0009, "step": 98940 }, { "epoch": 1.619078785895443, "grad_norm": 0.0005270187393762171, "learning_rate": 7.427562899405664e-06, "loss": 0.0012, "step": 98950 }, { "epoch": 1.619242411846519, "grad_norm": 0.036414485424757004, "learning_rate": 7.426938715846572e-06, "loss": 0.0017, "step": 98960 }, { "epoch": 1.6194060377975947, "grad_norm": 0.079173244535923, "learning_rate": 7.426314482804266e-06, "loss": 0.0011, "step": 98970 }, { "epoch": 1.6195696637486705, "grad_norm": 0.025795312598347664, "learning_rate": 7.425690200291469e-06, "loss": 0.0015, "step": 98980 }, { "epoch": 1.6197332896997465, "grad_norm": 0.06961794942617416, "learning_rate": 7.425065868320913e-06, "loss": 0.0014, "step": 98990 }, { "epoch": 1.619896915650822, "grad_norm": 0.04506934434175491, "learning_rate": 7.424441486905326e-06, "loss": 0.0023, "step": 99000 }, { "epoch": 1.620060541601898, "grad_norm": 0.07526300102472305, "learning_rate": 7.42381705605744e-06, "loss": 0.0009, "step": 99010 }, { "epoch": 1.620224167552974, "grad_norm": 0.04929833114147186, "learning_rate": 7.423192575789984e-06, "loss": 0.003, "step": 99020 }, { "epoch": 1.6203877935040496, "grad_norm": 0.1168147549033165, "learning_rate": 7.422568046115693e-06, "loss": 0.0024, "step": 99030 }, { "epoch": 1.6205514194551256, "grad_norm": 0.014927929267287254, "learning_rate": 7.4219434670473e-06, "loss": 0.0013, "step": 99040 }, { "epoch": 1.6207150454062014, "grad_norm": 0.03229750692844391, "learning_rate": 7.42131883859754e-06, "loss": 0.0023, "step": 99050 }, { "epoch": 1.6208786713572771, "grad_norm": 0.09730228036642075, "learning_rate": 7.4206941607791474e-06, "loss": 0.0024, "step": 99060 }, { "epoch": 1.6210422973083531, "grad_norm": 0.04613430052995682, "learning_rate": 7.420069433604859e-06, "loss": 0.0008, "step": 99070 }, { "epoch": 1.621205923259429, "grad_norm": 0.06413130462169647, "learning_rate": 7.419444657087413e-06, "loss": 0.0016, "step": 99080 }, { "epoch": 1.6213695492105047, "grad_norm": 0.03514116629958153, "learning_rate": 7.41881983123955e-06, "loss": 0.0015, "step": 99090 }, { "epoch": 1.6215331751615807, "grad_norm": 0.031732216477394104, "learning_rate": 7.418194956074007e-06, "loss": 0.0017, "step": 99100 }, { "epoch": 1.6216968011126565, "grad_norm": 0.051984574645757675, "learning_rate": 7.417570031603525e-06, "loss": 0.0017, "step": 99110 }, { "epoch": 1.6218604270637322, "grad_norm": 0.22698462009429932, "learning_rate": 7.416945057840847e-06, "loss": 0.0019, "step": 99120 }, { "epoch": 1.6220240530148082, "grad_norm": 0.042750101536512375, "learning_rate": 7.416320034798713e-06, "loss": 0.0011, "step": 99130 }, { "epoch": 1.622187678965884, "grad_norm": 0.041687943041324615, "learning_rate": 7.41569496248987e-06, "loss": 0.0013, "step": 99140 }, { "epoch": 1.6223513049169598, "grad_norm": 0.0037747237365692854, "learning_rate": 7.415069840927061e-06, "loss": 0.0008, "step": 99150 }, { "epoch": 1.6225149308680358, "grad_norm": 0.006445981562137604, "learning_rate": 7.414444670123031e-06, "loss": 0.0013, "step": 99160 }, { "epoch": 1.6226785568191116, "grad_norm": 0.164654940366745, "learning_rate": 7.413819450090527e-06, "loss": 0.0011, "step": 99170 }, { "epoch": 1.6228421827701873, "grad_norm": 0.08207637071609497, "learning_rate": 7.413194180842299e-06, "loss": 0.0016, "step": 99180 }, { "epoch": 1.6230058087212633, "grad_norm": 0.05668092891573906, "learning_rate": 7.4125688623910945e-06, "loss": 0.0018, "step": 99190 }, { "epoch": 1.6231694346723389, "grad_norm": 0.005453913006931543, "learning_rate": 7.41194349474966e-06, "loss": 0.0013, "step": 99200 }, { "epoch": 1.6233330606234149, "grad_norm": 0.25081855058670044, "learning_rate": 7.411318077930751e-06, "loss": 0.0013, "step": 99210 }, { "epoch": 1.6234966865744909, "grad_norm": 0.055974289774894714, "learning_rate": 7.410692611947117e-06, "loss": 0.0015, "step": 99220 }, { "epoch": 1.6236603125255664, "grad_norm": 0.06748989969491959, "learning_rate": 7.4100670968115106e-06, "loss": 0.0016, "step": 99230 }, { "epoch": 1.6238239384766424, "grad_norm": 0.00562110124155879, "learning_rate": 7.409441532536686e-06, "loss": 0.0008, "step": 99240 }, { "epoch": 1.6239875644277182, "grad_norm": 0.14977310597896576, "learning_rate": 7.408815919135397e-06, "loss": 0.0012, "step": 99250 }, { "epoch": 1.624151190378794, "grad_norm": 0.027807550504803658, "learning_rate": 7.4081902566204015e-06, "loss": 0.0024, "step": 99260 }, { "epoch": 1.62431481632987, "grad_norm": 0.08475083112716675, "learning_rate": 7.407564545004453e-06, "loss": 0.0011, "step": 99270 }, { "epoch": 1.6244784422809457, "grad_norm": 0.05055151507258415, "learning_rate": 7.406938784300312e-06, "loss": 0.0025, "step": 99280 }, { "epoch": 1.6246420682320215, "grad_norm": 0.016771268099546432, "learning_rate": 7.406312974520735e-06, "loss": 0.0014, "step": 99290 }, { "epoch": 1.6248056941830975, "grad_norm": 0.03938732668757439, "learning_rate": 7.4056871156784844e-06, "loss": 0.0009, "step": 99300 }, { "epoch": 1.6249693201341733, "grad_norm": 0.060104016214609146, "learning_rate": 7.405061207786319e-06, "loss": 0.002, "step": 99310 }, { "epoch": 1.625132946085249, "grad_norm": 0.13139458000659943, "learning_rate": 7.404435250857002e-06, "loss": 0.0015, "step": 99320 }, { "epoch": 1.625296572036325, "grad_norm": 0.06462104618549347, "learning_rate": 7.403809244903293e-06, "loss": 0.0018, "step": 99330 }, { "epoch": 1.6254601979874008, "grad_norm": 0.07042374461889267, "learning_rate": 7.403183189937959e-06, "loss": 0.0008, "step": 99340 }, { "epoch": 1.6256238239384766, "grad_norm": 0.029185360297560692, "learning_rate": 7.402557085973764e-06, "loss": 0.0026, "step": 99350 }, { "epoch": 1.6257874498895526, "grad_norm": 0.11242368817329407, "learning_rate": 7.4019309330234735e-06, "loss": 0.0014, "step": 99360 }, { "epoch": 1.6259510758406284, "grad_norm": 0.01770659349858761, "learning_rate": 7.401304731099852e-06, "loss": 0.0016, "step": 99370 }, { "epoch": 1.6261147017917041, "grad_norm": 0.014126712456345558, "learning_rate": 7.4006784802156715e-06, "loss": 0.0007, "step": 99380 }, { "epoch": 1.6262783277427801, "grad_norm": 0.07345698028802872, "learning_rate": 7.400052180383698e-06, "loss": 0.0009, "step": 99390 }, { "epoch": 1.6264419536938557, "grad_norm": 0.034991856664419174, "learning_rate": 7.399425831616701e-06, "loss": 0.0021, "step": 99400 }, { "epoch": 1.6266055796449317, "grad_norm": 0.03604298084974289, "learning_rate": 7.398799433927453e-06, "loss": 0.0014, "step": 99410 }, { "epoch": 1.6267692055960077, "grad_norm": 0.03720438852906227, "learning_rate": 7.3981729873287234e-06, "loss": 0.0026, "step": 99420 }, { "epoch": 1.6269328315470832, "grad_norm": 0.15676535665988922, "learning_rate": 7.397546491833287e-06, "loss": 0.0015, "step": 99430 }, { "epoch": 1.6270964574981592, "grad_norm": 0.10392594337463379, "learning_rate": 7.396919947453916e-06, "loss": 0.0014, "step": 99440 }, { "epoch": 1.627260083449235, "grad_norm": 0.07405377179384232, "learning_rate": 7.396293354203388e-06, "loss": 0.0036, "step": 99450 }, { "epoch": 1.6274237094003108, "grad_norm": 0.09258932620286942, "learning_rate": 7.3956667120944735e-06, "loss": 0.0018, "step": 99460 }, { "epoch": 1.6275873353513868, "grad_norm": 0.0127404211089015, "learning_rate": 7.395040021139954e-06, "loss": 0.0012, "step": 99470 }, { "epoch": 1.6277509613024626, "grad_norm": 0.09768541157245636, "learning_rate": 7.394413281352604e-06, "loss": 0.0016, "step": 99480 }, { "epoch": 1.6279145872535383, "grad_norm": 0.06636276096105576, "learning_rate": 7.393786492745205e-06, "loss": 0.0015, "step": 99490 }, { "epoch": 1.6280782132046143, "grad_norm": 0.11083022505044937, "learning_rate": 7.393159655330534e-06, "loss": 0.0017, "step": 99500 }, { "epoch": 1.62824183915569, "grad_norm": 0.035407327115535736, "learning_rate": 7.3925327691213744e-06, "loss": 0.0018, "step": 99510 }, { "epoch": 1.6284054651067659, "grad_norm": 0.06660429388284683, "learning_rate": 7.391905834130504e-06, "loss": 0.0015, "step": 99520 }, { "epoch": 1.6285690910578419, "grad_norm": 0.08485773950815201, "learning_rate": 7.39127885037071e-06, "loss": 0.0012, "step": 99530 }, { "epoch": 1.6287327170089176, "grad_norm": 0.055254288017749786, "learning_rate": 7.3906518178547725e-06, "loss": 0.0039, "step": 99540 }, { "epoch": 1.6288963429599934, "grad_norm": 0.05437983572483063, "learning_rate": 7.390024736595481e-06, "loss": 0.0018, "step": 99550 }, { "epoch": 1.6290599689110694, "grad_norm": 0.07922807335853577, "learning_rate": 7.389397606605612e-06, "loss": 0.0015, "step": 99560 }, { "epoch": 1.6292235948621452, "grad_norm": 0.002190712606534362, "learning_rate": 7.388770427897963e-06, "loss": 0.001, "step": 99570 }, { "epoch": 1.629387220813221, "grad_norm": 0.037567559629678726, "learning_rate": 7.388143200485314e-06, "loss": 0.0007, "step": 99580 }, { "epoch": 1.629550846764297, "grad_norm": 0.08029983937740326, "learning_rate": 7.387515924380457e-06, "loss": 0.0015, "step": 99590 }, { "epoch": 1.6297144727153725, "grad_norm": 0.09983482211828232, "learning_rate": 7.38688859959618e-06, "loss": 0.0018, "step": 99600 }, { "epoch": 1.6298780986664485, "grad_norm": 0.07510628551244736, "learning_rate": 7.386261226145277e-06, "loss": 0.002, "step": 99610 }, { "epoch": 1.6300417246175245, "grad_norm": 0.04404643923044205, "learning_rate": 7.385633804040534e-06, "loss": 0.0013, "step": 99620 }, { "epoch": 1.6302053505686, "grad_norm": 0.005121331661939621, "learning_rate": 7.3850063332947484e-06, "loss": 0.0018, "step": 99630 }, { "epoch": 1.630368976519676, "grad_norm": 0.06081508845090866, "learning_rate": 7.3843788139207105e-06, "loss": 0.0021, "step": 99640 }, { "epoch": 1.6305326024707518, "grad_norm": 0.24447375535964966, "learning_rate": 7.383751245931219e-06, "loss": 0.0016, "step": 99650 }, { "epoch": 1.6306962284218276, "grad_norm": 0.1591046303510666, "learning_rate": 7.383123629339064e-06, "loss": 0.0054, "step": 99660 }, { "epoch": 1.6308598543729036, "grad_norm": 0.04624222218990326, "learning_rate": 7.382495964157047e-06, "loss": 0.0011, "step": 99670 }, { "epoch": 1.6310234803239794, "grad_norm": 0.09611093252897263, "learning_rate": 7.3818682503979626e-06, "loss": 0.0037, "step": 99680 }, { "epoch": 1.6311871062750551, "grad_norm": 0.042438313364982605, "learning_rate": 7.38124048807461e-06, "loss": 0.0013, "step": 99690 }, { "epoch": 1.6313507322261311, "grad_norm": 0.0629676878452301, "learning_rate": 7.38061267719979e-06, "loss": 0.0008, "step": 99700 }, { "epoch": 1.631514358177207, "grad_norm": 0.042327411472797394, "learning_rate": 7.379984817786301e-06, "loss": 0.0015, "step": 99710 }, { "epoch": 1.6316779841282827, "grad_norm": 0.07500994205474854, "learning_rate": 7.379356909846946e-06, "loss": 0.0012, "step": 99720 }, { "epoch": 1.6318416100793587, "grad_norm": 0.05551949515938759, "learning_rate": 7.378728953394528e-06, "loss": 0.0017, "step": 99730 }, { "epoch": 1.6320052360304345, "grad_norm": 0.13285714387893677, "learning_rate": 7.37810094844185e-06, "loss": 0.0018, "step": 99740 }, { "epoch": 1.6321688619815102, "grad_norm": 0.001422732719220221, "learning_rate": 7.377472895001715e-06, "loss": 0.0017, "step": 99750 }, { "epoch": 1.6323324879325862, "grad_norm": 0.08280148357152939, "learning_rate": 7.3768447930869306e-06, "loss": 0.0024, "step": 99760 }, { "epoch": 1.632496113883662, "grad_norm": 0.05470971390604973, "learning_rate": 7.376216642710301e-06, "loss": 0.0023, "step": 99770 }, { "epoch": 1.6326597398347378, "grad_norm": 0.03797735273838043, "learning_rate": 7.375588443884636e-06, "loss": 0.0012, "step": 99780 }, { "epoch": 1.6328233657858138, "grad_norm": 0.06510048359632492, "learning_rate": 7.374960196622743e-06, "loss": 0.001, "step": 99790 }, { "epoch": 1.6329869917368893, "grad_norm": 0.022689199075102806, "learning_rate": 7.37433190093743e-06, "loss": 0.0017, "step": 99800 }, { "epoch": 1.6331506176879653, "grad_norm": 0.11079506576061249, "learning_rate": 7.3737035568415094e-06, "loss": 0.0022, "step": 99810 }, { "epoch": 1.633314243639041, "grad_norm": 0.01739976741373539, "learning_rate": 7.373075164347794e-06, "loss": 0.0008, "step": 99820 }, { "epoch": 1.6334778695901169, "grad_norm": 0.12702660262584686, "learning_rate": 7.372446723469092e-06, "loss": 0.0019, "step": 99830 }, { "epoch": 1.6336414955411929, "grad_norm": 0.05213851481676102, "learning_rate": 7.3718182342182196e-06, "loss": 0.0017, "step": 99840 }, { "epoch": 1.6338051214922686, "grad_norm": 0.019381210207939148, "learning_rate": 7.371189696607991e-06, "loss": 0.0013, "step": 99850 }, { "epoch": 1.6339687474433444, "grad_norm": 0.11266452074050903, "learning_rate": 7.370561110651221e-06, "loss": 0.0018, "step": 99860 }, { "epoch": 1.6341323733944204, "grad_norm": 0.02784690447151661, "learning_rate": 7.369932476360725e-06, "loss": 0.0019, "step": 99870 }, { "epoch": 1.6342959993454962, "grad_norm": 0.10003897547721863, "learning_rate": 7.369303793749323e-06, "loss": 0.0014, "step": 99880 }, { "epoch": 1.634459625296572, "grad_norm": 0.011508649215102196, "learning_rate": 7.368675062829831e-06, "loss": 0.001, "step": 99890 }, { "epoch": 1.634623251247648, "grad_norm": 0.06960906833410263, "learning_rate": 7.368046283615069e-06, "loss": 0.0016, "step": 99900 }, { "epoch": 1.6347868771987237, "grad_norm": 0.06117837876081467, "learning_rate": 7.367417456117856e-06, "loss": 0.0018, "step": 99910 }, { "epoch": 1.6349505031497995, "grad_norm": 0.09999518096446991, "learning_rate": 7.366788580351016e-06, "loss": 0.0016, "step": 99920 }, { "epoch": 1.6351141291008755, "grad_norm": 0.08859653025865555, "learning_rate": 7.36615965632737e-06, "loss": 0.0014, "step": 99930 }, { "epoch": 1.6352777550519513, "grad_norm": 0.03636933118104935, "learning_rate": 7.3655306840597395e-06, "loss": 0.0012, "step": 99940 }, { "epoch": 1.635441381003027, "grad_norm": 0.22999189794063568, "learning_rate": 7.36490166356095e-06, "loss": 0.0025, "step": 99950 }, { "epoch": 1.635605006954103, "grad_norm": 0.017478913068771362, "learning_rate": 7.364272594843829e-06, "loss": 0.001, "step": 99960 }, { "epoch": 1.6357686329051786, "grad_norm": 0.03594844415783882, "learning_rate": 7.3636434779211985e-06, "loss": 0.0038, "step": 99970 }, { "epoch": 1.6359322588562546, "grad_norm": 0.10184531658887863, "learning_rate": 7.36301431280589e-06, "loss": 0.0018, "step": 99980 }, { "epoch": 1.6360958848073306, "grad_norm": 0.0563923716545105, "learning_rate": 7.362385099510727e-06, "loss": 0.0017, "step": 99990 }, { "epoch": 1.6362595107584061, "grad_norm": 0.04771708697080612, "learning_rate": 7.361755838048542e-06, "loss": 0.0006, "step": 100000 }, { "epoch": 1.6364231367094821, "grad_norm": 0.05845789983868599, "learning_rate": 7.361126528432163e-06, "loss": 0.0012, "step": 100010 }, { "epoch": 1.636586762660558, "grad_norm": 0.08642744272947311, "learning_rate": 7.3604971706744235e-06, "loss": 0.0013, "step": 100020 }, { "epoch": 1.6367503886116337, "grad_norm": 0.030177731066942215, "learning_rate": 7.359867764788153e-06, "loss": 0.0015, "step": 100030 }, { "epoch": 1.6369140145627097, "grad_norm": 0.032490916550159454, "learning_rate": 7.359238310786187e-06, "loss": 0.0014, "step": 100040 }, { "epoch": 1.6370776405137855, "grad_norm": 0.10349056869745255, "learning_rate": 7.3586088086813575e-06, "loss": 0.0017, "step": 100050 }, { "epoch": 1.6372412664648612, "grad_norm": 0.09051650762557983, "learning_rate": 7.357979258486501e-06, "loss": 0.0019, "step": 100060 }, { "epoch": 1.6374048924159372, "grad_norm": 0.06951526552438736, "learning_rate": 7.3573496602144535e-06, "loss": 0.0018, "step": 100070 }, { "epoch": 1.637568518367013, "grad_norm": 0.04659709334373474, "learning_rate": 7.3567200138780495e-06, "loss": 0.0015, "step": 100080 }, { "epoch": 1.6377321443180888, "grad_norm": 0.10708335787057877, "learning_rate": 7.3560903194901304e-06, "loss": 0.0012, "step": 100090 }, { "epoch": 1.6378957702691648, "grad_norm": 0.05217113345861435, "learning_rate": 7.3554605770635315e-06, "loss": 0.0018, "step": 100100 }, { "epoch": 1.6380593962202405, "grad_norm": 0.02114526368677616, "learning_rate": 7.354830786611096e-06, "loss": 0.0024, "step": 100110 }, { "epoch": 1.6382230221713163, "grad_norm": 0.041843146085739136, "learning_rate": 7.354200948145662e-06, "loss": 0.003, "step": 100120 }, { "epoch": 1.6383866481223923, "grad_norm": 0.1545921117067337, "learning_rate": 7.3535710616800744e-06, "loss": 0.002, "step": 100130 }, { "epoch": 1.638550274073468, "grad_norm": 0.016839755699038506, "learning_rate": 7.352941127227174e-06, "loss": 0.0011, "step": 100140 }, { "epoch": 1.6387139000245439, "grad_norm": 0.22923074662685394, "learning_rate": 7.352311144799806e-06, "loss": 0.0033, "step": 100150 }, { "epoch": 1.6388775259756199, "grad_norm": 0.01960967481136322, "learning_rate": 7.351681114410814e-06, "loss": 0.0012, "step": 100160 }, { "epoch": 1.6390411519266954, "grad_norm": 0.03905986621975899, "learning_rate": 7.3510510360730435e-06, "loss": 0.0013, "step": 100170 }, { "epoch": 1.6392047778777714, "grad_norm": 0.05031822249293327, "learning_rate": 7.35042090979934e-06, "loss": 0.0012, "step": 100180 }, { "epoch": 1.6393684038288474, "grad_norm": 0.003884856356307864, "learning_rate": 7.349790735602557e-06, "loss": 0.0026, "step": 100190 }, { "epoch": 1.639532029779923, "grad_norm": 0.04146142303943634, "learning_rate": 7.349160513495537e-06, "loss": 0.0022, "step": 100200 }, { "epoch": 1.639695655730999, "grad_norm": 0.010762088932096958, "learning_rate": 7.348530243491133e-06, "loss": 0.0009, "step": 100210 }, { "epoch": 1.6398592816820747, "grad_norm": 0.06311865895986557, "learning_rate": 7.347899925602193e-06, "loss": 0.0018, "step": 100220 }, { "epoch": 1.6400229076331505, "grad_norm": 0.04305097088217735, "learning_rate": 7.347269559841572e-06, "loss": 0.0014, "step": 100230 }, { "epoch": 1.6401865335842265, "grad_norm": 0.030692143365740776, "learning_rate": 7.346639146222121e-06, "loss": 0.0013, "step": 100240 }, { "epoch": 1.6403501595353023, "grad_norm": 0.06957350671291351, "learning_rate": 7.346008684756694e-06, "loss": 0.0012, "step": 100250 }, { "epoch": 1.640513785486378, "grad_norm": 0.00722523732110858, "learning_rate": 7.3453781754581425e-06, "loss": 0.0014, "step": 100260 }, { "epoch": 1.640677411437454, "grad_norm": 0.06896369159221649, "learning_rate": 7.344747618339327e-06, "loss": 0.0014, "step": 100270 }, { "epoch": 1.6408410373885298, "grad_norm": 0.01380205899477005, "learning_rate": 7.3441170134131e-06, "loss": 0.0019, "step": 100280 }, { "epoch": 1.6410046633396056, "grad_norm": 0.020061321556568146, "learning_rate": 7.343486360692322e-06, "loss": 0.0008, "step": 100290 }, { "epoch": 1.6411682892906816, "grad_norm": 0.1674642562866211, "learning_rate": 7.342855660189849e-06, "loss": 0.0028, "step": 100300 }, { "epoch": 1.6413319152417574, "grad_norm": 0.03238382562994957, "learning_rate": 7.342224911918544e-06, "loss": 0.0019, "step": 100310 }, { "epoch": 1.6414955411928331, "grad_norm": 0.0349264033138752, "learning_rate": 7.3415941158912625e-06, "loss": 0.0016, "step": 100320 }, { "epoch": 1.6416591671439091, "grad_norm": 0.03005646914243698, "learning_rate": 7.34096327212087e-06, "loss": 0.0011, "step": 100330 }, { "epoch": 1.641822793094985, "grad_norm": 0.04024549201130867, "learning_rate": 7.340332380620226e-06, "loss": 0.0015, "step": 100340 }, { "epoch": 1.6419864190460607, "grad_norm": 0.10797762125730515, "learning_rate": 7.339701441402197e-06, "loss": 0.0023, "step": 100350 }, { "epoch": 1.6421500449971367, "grad_norm": 0.07801230996847153, "learning_rate": 7.339070454479645e-06, "loss": 0.0019, "step": 100360 }, { "epoch": 1.6423136709482122, "grad_norm": 0.13199694454669952, "learning_rate": 7.338439419865434e-06, "loss": 0.0015, "step": 100370 }, { "epoch": 1.6424772968992882, "grad_norm": 0.058924831449985504, "learning_rate": 7.337808337572434e-06, "loss": 0.0012, "step": 100380 }, { "epoch": 1.6426409228503642, "grad_norm": 0.026357822120189667, "learning_rate": 7.337177207613508e-06, "loss": 0.0006, "step": 100390 }, { "epoch": 1.6428045488014398, "grad_norm": 0.07109837979078293, "learning_rate": 7.336546030001529e-06, "loss": 0.0016, "step": 100400 }, { "epoch": 1.6429681747525158, "grad_norm": 0.029550019651651382, "learning_rate": 7.335914804749362e-06, "loss": 0.0018, "step": 100410 }, { "epoch": 1.6431318007035915, "grad_norm": 0.07188637554645538, "learning_rate": 7.335283531869879e-06, "loss": 0.0022, "step": 100420 }, { "epoch": 1.6432954266546673, "grad_norm": 0.34504249691963196, "learning_rate": 7.334652211375951e-06, "loss": 0.002, "step": 100430 }, { "epoch": 1.6434590526057433, "grad_norm": 0.013034996576607227, "learning_rate": 7.334020843280451e-06, "loss": 0.0022, "step": 100440 }, { "epoch": 1.643622678556819, "grad_norm": 0.017551854252815247, "learning_rate": 7.333389427596249e-06, "loss": 0.0033, "step": 100450 }, { "epoch": 1.6437863045078949, "grad_norm": 0.06309592723846436, "learning_rate": 7.332757964336222e-06, "loss": 0.001, "step": 100460 }, { "epoch": 1.6439499304589709, "grad_norm": 0.06063041090965271, "learning_rate": 7.332126453513244e-06, "loss": 0.0014, "step": 100470 }, { "epoch": 1.6441135564100466, "grad_norm": 0.07614735513925552, "learning_rate": 7.331494895140192e-06, "loss": 0.0023, "step": 100480 }, { "epoch": 1.6442771823611224, "grad_norm": 0.01910034939646721, "learning_rate": 7.33086328922994e-06, "loss": 0.0012, "step": 100490 }, { "epoch": 1.6444408083121984, "grad_norm": 0.08828983455896378, "learning_rate": 7.330231635795369e-06, "loss": 0.0019, "step": 100500 }, { "epoch": 1.6446044342632742, "grad_norm": 0.07797668874263763, "learning_rate": 7.3295999348493564e-06, "loss": 0.0016, "step": 100510 }, { "epoch": 1.64476806021435, "grad_norm": 0.015019861049950123, "learning_rate": 7.3289681864047835e-06, "loss": 0.0009, "step": 100520 }, { "epoch": 1.644931686165426, "grad_norm": 0.1614726334810257, "learning_rate": 7.328336390474529e-06, "loss": 0.0017, "step": 100530 }, { "epoch": 1.6450953121165017, "grad_norm": 0.1239028200507164, "learning_rate": 7.327704547071476e-06, "loss": 0.0017, "step": 100540 }, { "epoch": 1.6452589380675775, "grad_norm": 0.07276678085327148, "learning_rate": 7.327072656208506e-06, "loss": 0.0028, "step": 100550 }, { "epoch": 1.6454225640186535, "grad_norm": 0.04590804502367973, "learning_rate": 7.3264407178985055e-06, "loss": 0.0012, "step": 100560 }, { "epoch": 1.645586189969729, "grad_norm": 0.06148697808384895, "learning_rate": 7.325808732154356e-06, "loss": 0.001, "step": 100570 }, { "epoch": 1.645749815920805, "grad_norm": 0.046202972531318665, "learning_rate": 7.325176698988945e-06, "loss": 0.0006, "step": 100580 }, { "epoch": 1.645913441871881, "grad_norm": 0.052195336669683456, "learning_rate": 7.324544618415157e-06, "loss": 0.0024, "step": 100590 }, { "epoch": 1.6460770678229566, "grad_norm": 0.059218525886535645, "learning_rate": 7.323912490445884e-06, "loss": 0.0013, "step": 100600 }, { "epoch": 1.6462406937740326, "grad_norm": 0.06493990868330002, "learning_rate": 7.323280315094009e-06, "loss": 0.002, "step": 100610 }, { "epoch": 1.6464043197251084, "grad_norm": 0.12280164659023285, "learning_rate": 7.322648092372426e-06, "loss": 0.0015, "step": 100620 }, { "epoch": 1.6465679456761841, "grad_norm": 0.0021489716600626707, "learning_rate": 7.322015822294022e-06, "loss": 0.0013, "step": 100630 }, { "epoch": 1.6467315716272601, "grad_norm": 0.049451299011707306, "learning_rate": 7.321383504871692e-06, "loss": 0.0016, "step": 100640 }, { "epoch": 1.646895197578336, "grad_norm": 0.003148362273350358, "learning_rate": 7.320751140118325e-06, "loss": 0.0014, "step": 100650 }, { "epoch": 1.6470588235294117, "grad_norm": 0.032490476965904236, "learning_rate": 7.320118728046818e-06, "loss": 0.0013, "step": 100660 }, { "epoch": 1.6472224494804877, "grad_norm": 0.0011689908569678664, "learning_rate": 7.319486268670062e-06, "loss": 0.0017, "step": 100670 }, { "epoch": 1.6473860754315635, "grad_norm": 0.1099620908498764, "learning_rate": 7.318853762000955e-06, "loss": 0.0032, "step": 100680 }, { "epoch": 1.6475497013826392, "grad_norm": 0.1322086602449417, "learning_rate": 7.318221208052389e-06, "loss": 0.0009, "step": 100690 }, { "epoch": 1.6477133273337152, "grad_norm": 0.06388381123542786, "learning_rate": 7.317588606837267e-06, "loss": 0.0024, "step": 100700 }, { "epoch": 1.647876953284791, "grad_norm": 0.07097798585891724, "learning_rate": 7.316955958368483e-06, "loss": 0.0018, "step": 100710 }, { "epoch": 1.6480405792358668, "grad_norm": 0.2233915477991104, "learning_rate": 7.316323262658939e-06, "loss": 0.0024, "step": 100720 }, { "epoch": 1.6482042051869428, "grad_norm": 0.07791968435049057, "learning_rate": 7.3156905197215324e-06, "loss": 0.0023, "step": 100730 }, { "epoch": 1.6483678311380183, "grad_norm": 0.05111980065703392, "learning_rate": 7.3150577295691646e-06, "loss": 0.0031, "step": 100740 }, { "epoch": 1.6485314570890943, "grad_norm": 0.045101288706064224, "learning_rate": 7.31442489221474e-06, "loss": 0.0008, "step": 100750 }, { "epoch": 1.6486950830401703, "grad_norm": 0.024166390299797058, "learning_rate": 7.31379200767116e-06, "loss": 0.0021, "step": 100760 }, { "epoch": 1.6488587089912459, "grad_norm": 0.14793042838573456, "learning_rate": 7.3131590759513286e-06, "loss": 0.0013, "step": 100770 }, { "epoch": 1.6490223349423219, "grad_norm": 0.09345638751983643, "learning_rate": 7.312526097068151e-06, "loss": 0.0017, "step": 100780 }, { "epoch": 1.6491859608933976, "grad_norm": 0.016167260706424713, "learning_rate": 7.311893071034533e-06, "loss": 0.001, "step": 100790 }, { "epoch": 1.6493495868444734, "grad_norm": 0.04797253757715225, "learning_rate": 7.31125999786338e-06, "loss": 0.0009, "step": 100800 }, { "epoch": 1.6495132127955494, "grad_norm": 0.007611151318997145, "learning_rate": 7.3106268775676035e-06, "loss": 0.0018, "step": 100810 }, { "epoch": 1.6496768387466252, "grad_norm": 0.10032690316438675, "learning_rate": 7.309993710160109e-06, "loss": 0.0017, "step": 100820 }, { "epoch": 1.649840464697701, "grad_norm": 0.017537208274006844, "learning_rate": 7.309360495653808e-06, "loss": 0.0016, "step": 100830 }, { "epoch": 1.650004090648777, "grad_norm": 0.013645618222653866, "learning_rate": 7.30872723406161e-06, "loss": 0.0016, "step": 100840 }, { "epoch": 1.6501677165998527, "grad_norm": 0.032411571592092514, "learning_rate": 7.308093925396427e-06, "loss": 0.0012, "step": 100850 }, { "epoch": 1.6503313425509285, "grad_norm": 0.0751567929983139, "learning_rate": 7.307460569671172e-06, "loss": 0.0011, "step": 100860 }, { "epoch": 1.6504949685020045, "grad_norm": 0.026018667966127396, "learning_rate": 7.306827166898759e-06, "loss": 0.0015, "step": 100870 }, { "epoch": 1.6506585944530803, "grad_norm": 0.12846526503562927, "learning_rate": 7.306193717092101e-06, "loss": 0.0019, "step": 100880 }, { "epoch": 1.650822220404156, "grad_norm": 0.025230279192328453, "learning_rate": 7.305560220264115e-06, "loss": 0.0014, "step": 100890 }, { "epoch": 1.650985846355232, "grad_norm": 0.08344145864248276, "learning_rate": 7.304926676427717e-06, "loss": 0.0029, "step": 100900 }, { "epoch": 1.6511494723063078, "grad_norm": 0.14763100445270538, "learning_rate": 7.304293085595824e-06, "loss": 0.0021, "step": 100910 }, { "epoch": 1.6513130982573836, "grad_norm": 0.23541006445884705, "learning_rate": 7.303659447781355e-06, "loss": 0.0027, "step": 100920 }, { "epoch": 1.6514767242084596, "grad_norm": 0.05932603403925896, "learning_rate": 7.303025762997229e-06, "loss": 0.001, "step": 100930 }, { "epoch": 1.6516403501595351, "grad_norm": 0.11770252138376236, "learning_rate": 7.3023920312563665e-06, "loss": 0.003, "step": 100940 }, { "epoch": 1.6518039761106111, "grad_norm": 0.07041673362255096, "learning_rate": 7.301758252571689e-06, "loss": 0.0013, "step": 100950 }, { "epoch": 1.6519676020616871, "grad_norm": 0.031235376372933388, "learning_rate": 7.301124426956117e-06, "loss": 0.0017, "step": 100960 }, { "epoch": 1.6521312280127627, "grad_norm": 0.029568541795015335, "learning_rate": 7.3004905544225755e-06, "loss": 0.0013, "step": 100970 }, { "epoch": 1.6522948539638387, "grad_norm": 0.08486157655715942, "learning_rate": 7.299856634983988e-06, "loss": 0.0015, "step": 100980 }, { "epoch": 1.6524584799149145, "grad_norm": 0.02538946270942688, "learning_rate": 7.299222668653281e-06, "loss": 0.0018, "step": 100990 }, { "epoch": 1.6526221058659902, "grad_norm": 0.03677397221326828, "learning_rate": 7.298588655443377e-06, "loss": 0.0012, "step": 101000 }, { "epoch": 1.6527857318170662, "grad_norm": 0.014986048452556133, "learning_rate": 7.297954595367207e-06, "loss": 0.0014, "step": 101010 }, { "epoch": 1.652949357768142, "grad_norm": 0.011416024528443813, "learning_rate": 7.297320488437697e-06, "loss": 0.0022, "step": 101020 }, { "epoch": 1.6531129837192178, "grad_norm": 0.035167448222637177, "learning_rate": 7.296686334667775e-06, "loss": 0.0014, "step": 101030 }, { "epoch": 1.6532766096702938, "grad_norm": 0.08473289012908936, "learning_rate": 7.296052134070373e-06, "loss": 0.0022, "step": 101040 }, { "epoch": 1.6534402356213695, "grad_norm": 0.010432890616357327, "learning_rate": 7.295417886658419e-06, "loss": 0.0015, "step": 101050 }, { "epoch": 1.6536038615724453, "grad_norm": 0.048307280987501144, "learning_rate": 7.294783592444849e-06, "loss": 0.0012, "step": 101060 }, { "epoch": 1.6537674875235213, "grad_norm": 0.015407691709697247, "learning_rate": 7.29414925144259e-06, "loss": 0.0015, "step": 101070 }, { "epoch": 1.653931113474597, "grad_norm": 0.014108225703239441, "learning_rate": 7.293514863664581e-06, "loss": 0.0004, "step": 101080 }, { "epoch": 1.6540947394256729, "grad_norm": 0.15072429180145264, "learning_rate": 7.292880429123754e-06, "loss": 0.0012, "step": 101090 }, { "epoch": 1.6542583653767489, "grad_norm": 0.009456695057451725, "learning_rate": 7.292245947833047e-06, "loss": 0.0019, "step": 101100 }, { "epoch": 1.6544219913278246, "grad_norm": 0.0034439063165336847, "learning_rate": 7.291611419805391e-06, "loss": 0.0014, "step": 101110 }, { "epoch": 1.6545856172789004, "grad_norm": 0.15090501308441162, "learning_rate": 7.29097684505373e-06, "loss": 0.0021, "step": 101120 }, { "epoch": 1.6547492432299764, "grad_norm": 0.04838437959551811, "learning_rate": 7.2903422235909985e-06, "loss": 0.0017, "step": 101130 }, { "epoch": 1.654912869181052, "grad_norm": 0.03755245357751846, "learning_rate": 7.289707555430136e-06, "loss": 0.0024, "step": 101140 }, { "epoch": 1.655076495132128, "grad_norm": 0.03455277532339096, "learning_rate": 7.289072840584085e-06, "loss": 0.0013, "step": 101150 }, { "epoch": 1.655240121083204, "grad_norm": 0.015921708196401596, "learning_rate": 7.288438079065786e-06, "loss": 0.0012, "step": 101160 }, { "epoch": 1.6554037470342795, "grad_norm": 0.06733939796686172, "learning_rate": 7.287803270888179e-06, "loss": 0.0022, "step": 101170 }, { "epoch": 1.6555673729853555, "grad_norm": 0.024022817611694336, "learning_rate": 7.287168416064211e-06, "loss": 0.0011, "step": 101180 }, { "epoch": 1.6557309989364313, "grad_norm": 0.08015142381191254, "learning_rate": 7.286533514606823e-06, "loss": 0.0018, "step": 101190 }, { "epoch": 1.655894624887507, "grad_norm": 0.04711209982633591, "learning_rate": 7.285898566528962e-06, "loss": 0.0011, "step": 101200 }, { "epoch": 1.656058250838583, "grad_norm": 0.029322123154997826, "learning_rate": 7.2852635718435724e-06, "loss": 0.0008, "step": 101210 }, { "epoch": 1.6562218767896588, "grad_norm": 0.06200619786977768, "learning_rate": 7.284628530563602e-06, "loss": 0.0008, "step": 101220 }, { "epoch": 1.6563855027407346, "grad_norm": 0.20434968173503876, "learning_rate": 7.2839934427019995e-06, "loss": 0.0013, "step": 101230 }, { "epoch": 1.6565491286918106, "grad_norm": 0.052322983741760254, "learning_rate": 7.283358308271713e-06, "loss": 0.0018, "step": 101240 }, { "epoch": 1.6567127546428864, "grad_norm": 0.11672504991292953, "learning_rate": 7.282723127285693e-06, "loss": 0.0018, "step": 101250 }, { "epoch": 1.6568763805939621, "grad_norm": 0.015510003082454205, "learning_rate": 7.282087899756889e-06, "loss": 0.0013, "step": 101260 }, { "epoch": 1.6570400065450381, "grad_norm": 0.014893121086061, "learning_rate": 7.281452625698254e-06, "loss": 0.0014, "step": 101270 }, { "epoch": 1.657203632496114, "grad_norm": 0.040926966816186905, "learning_rate": 7.28081730512274e-06, "loss": 0.0021, "step": 101280 }, { "epoch": 1.6573672584471897, "grad_norm": 0.07589961588382721, "learning_rate": 7.280181938043302e-06, "loss": 0.0025, "step": 101290 }, { "epoch": 1.6575308843982657, "grad_norm": 0.03295258432626724, "learning_rate": 7.279546524472893e-06, "loss": 0.001, "step": 101300 }, { "epoch": 1.6576945103493415, "grad_norm": 0.011349596083164215, "learning_rate": 7.278911064424468e-06, "loss": 0.0014, "step": 101310 }, { "epoch": 1.6578581363004172, "grad_norm": 0.4080277979373932, "learning_rate": 7.278275557910985e-06, "loss": 0.001, "step": 101320 }, { "epoch": 1.6580217622514932, "grad_norm": 0.04352203756570816, "learning_rate": 7.277640004945401e-06, "loss": 0.0032, "step": 101330 }, { "epoch": 1.6581853882025688, "grad_norm": 0.08412224054336548, "learning_rate": 7.277004405540675e-06, "loss": 0.0012, "step": 101340 }, { "epoch": 1.6583490141536448, "grad_norm": 0.08031485974788666, "learning_rate": 7.276368759709766e-06, "loss": 0.0017, "step": 101350 }, { "epoch": 1.6585126401047208, "grad_norm": 0.0613316148519516, "learning_rate": 7.275733067465634e-06, "loss": 0.0014, "step": 101360 }, { "epoch": 1.6586762660557963, "grad_norm": 0.04265898838639259, "learning_rate": 7.275097328821239e-06, "loss": 0.0015, "step": 101370 }, { "epoch": 1.6588398920068723, "grad_norm": 0.0524805448949337, "learning_rate": 7.2744615437895435e-06, "loss": 0.0017, "step": 101380 }, { "epoch": 1.659003517957948, "grad_norm": 0.04641233757138252, "learning_rate": 7.273825712383514e-06, "loss": 0.0013, "step": 101390 }, { "epoch": 1.6591671439090239, "grad_norm": 0.08346845209598541, "learning_rate": 7.27318983461611e-06, "loss": 0.002, "step": 101400 }, { "epoch": 1.6593307698600999, "grad_norm": 0.05447343364357948, "learning_rate": 7.2725539105002995e-06, "loss": 0.0012, "step": 101410 }, { "epoch": 1.6594943958111756, "grad_norm": 0.023423125967383385, "learning_rate": 7.271917940049048e-06, "loss": 0.0019, "step": 101420 }, { "epoch": 1.6596580217622514, "grad_norm": 0.043076783418655396, "learning_rate": 7.271281923275322e-06, "loss": 0.0021, "step": 101430 }, { "epoch": 1.6598216477133274, "grad_norm": 0.016837103292346, "learning_rate": 7.270645860192087e-06, "loss": 0.0031, "step": 101440 }, { "epoch": 1.6599852736644032, "grad_norm": 0.11963218450546265, "learning_rate": 7.270009750812317e-06, "loss": 0.0082, "step": 101450 }, { "epoch": 1.660148899615479, "grad_norm": 0.06557774543762207, "learning_rate": 7.269373595148976e-06, "loss": 0.0015, "step": 101460 }, { "epoch": 1.660312525566555, "grad_norm": 0.10065004974603653, "learning_rate": 7.26873739321504e-06, "loss": 0.0009, "step": 101470 }, { "epoch": 1.6604761515176307, "grad_norm": 0.007653196342289448, "learning_rate": 7.2681011450234765e-06, "loss": 0.0009, "step": 101480 }, { "epoch": 1.6606397774687065, "grad_norm": 0.047723934054374695, "learning_rate": 7.26746485058726e-06, "loss": 0.0016, "step": 101490 }, { "epoch": 1.6608034034197825, "grad_norm": 0.04685061797499657, "learning_rate": 7.266828509919362e-06, "loss": 0.0022, "step": 101500 }, { "epoch": 1.6609670293708583, "grad_norm": 0.07286231964826584, "learning_rate": 7.266192123032762e-06, "loss": 0.0013, "step": 101510 }, { "epoch": 1.661130655321934, "grad_norm": 0.035257793962955475, "learning_rate": 7.2655556899404285e-06, "loss": 0.0012, "step": 101520 }, { "epoch": 1.66129428127301, "grad_norm": 0.06788406521081924, "learning_rate": 7.264919210655343e-06, "loss": 0.0016, "step": 101530 }, { "epoch": 1.6614579072240856, "grad_norm": 0.053179092705249786, "learning_rate": 7.2642826851904814e-06, "loss": 0.0016, "step": 101540 }, { "epoch": 1.6616215331751616, "grad_norm": 0.04361078143119812, "learning_rate": 7.263646113558821e-06, "loss": 0.0019, "step": 101550 }, { "epoch": 1.6617851591262376, "grad_norm": 0.11017477512359619, "learning_rate": 7.263009495773341e-06, "loss": 0.0022, "step": 101560 }, { "epoch": 1.6619487850773131, "grad_norm": 0.13689225912094116, "learning_rate": 7.262372831847023e-06, "loss": 0.0016, "step": 101570 }, { "epoch": 1.6621124110283891, "grad_norm": 0.01548644807189703, "learning_rate": 7.261736121792846e-06, "loss": 0.0014, "step": 101580 }, { "epoch": 1.662276036979465, "grad_norm": 0.02995683252811432, "learning_rate": 7.261099365623794e-06, "loss": 0.0008, "step": 101590 }, { "epoch": 1.6624396629305407, "grad_norm": 0.08432107418775558, "learning_rate": 7.260462563352848e-06, "loss": 0.0015, "step": 101600 }, { "epoch": 1.6626032888816167, "grad_norm": 0.053025729954242706, "learning_rate": 7.259825714992994e-06, "loss": 0.0015, "step": 101610 }, { "epoch": 1.6627669148326925, "grad_norm": 0.07528650015592575, "learning_rate": 7.259188820557213e-06, "loss": 0.0011, "step": 101620 }, { "epoch": 1.6629305407837682, "grad_norm": 0.06874921172857285, "learning_rate": 7.258551880058496e-06, "loss": 0.0013, "step": 101630 }, { "epoch": 1.6630941667348442, "grad_norm": 0.02753896825015545, "learning_rate": 7.257914893509827e-06, "loss": 0.0013, "step": 101640 }, { "epoch": 1.66325779268592, "grad_norm": 0.14800146222114563, "learning_rate": 7.257277860924192e-06, "loss": 0.0015, "step": 101650 }, { "epoch": 1.6634214186369958, "grad_norm": 0.1733073741197586, "learning_rate": 7.256640782314581e-06, "loss": 0.0018, "step": 101660 }, { "epoch": 1.6635850445880718, "grad_norm": 0.0369713269174099, "learning_rate": 7.2560036576939865e-06, "loss": 0.001, "step": 101670 }, { "epoch": 1.6637486705391475, "grad_norm": 0.08525059372186661, "learning_rate": 7.255366487075394e-06, "loss": 0.0016, "step": 101680 }, { "epoch": 1.6639122964902233, "grad_norm": 0.056994397193193436, "learning_rate": 7.254729270471796e-06, "loss": 0.0025, "step": 101690 }, { "epoch": 1.6640759224412993, "grad_norm": 0.019522882997989655, "learning_rate": 7.254092007896187e-06, "loss": 0.0007, "step": 101700 }, { "epoch": 1.6642395483923749, "grad_norm": 0.1409287303686142, "learning_rate": 7.25345469936156e-06, "loss": 0.0015, "step": 101710 }, { "epoch": 1.6644031743434509, "grad_norm": 0.06796388328075409, "learning_rate": 7.252817344880907e-06, "loss": 0.0025, "step": 101720 }, { "epoch": 1.6645668002945269, "grad_norm": 0.04898270592093468, "learning_rate": 7.252179944467223e-06, "loss": 0.0024, "step": 101730 }, { "epoch": 1.6647304262456024, "grad_norm": 0.025501547381281853, "learning_rate": 7.251542498133508e-06, "loss": 0.0009, "step": 101740 }, { "epoch": 1.6648940521966784, "grad_norm": 0.040918491780757904, "learning_rate": 7.250905005892755e-06, "loss": 0.0018, "step": 101750 }, { "epoch": 1.6650576781477542, "grad_norm": 0.004044698551297188, "learning_rate": 7.250267467757963e-06, "loss": 0.0009, "step": 101760 }, { "epoch": 1.66522130409883, "grad_norm": 0.06088124215602875, "learning_rate": 7.2496298837421306e-06, "loss": 0.0017, "step": 101770 }, { "epoch": 1.665384930049906, "grad_norm": 0.0028151606675237417, "learning_rate": 7.2489922538582605e-06, "loss": 0.0015, "step": 101780 }, { "epoch": 1.6655485560009817, "grad_norm": 0.010668314062058926, "learning_rate": 7.248354578119348e-06, "loss": 0.0022, "step": 101790 }, { "epoch": 1.6657121819520575, "grad_norm": 0.05228869244456291, "learning_rate": 7.247716856538399e-06, "loss": 0.0012, "step": 101800 }, { "epoch": 1.6658758079031335, "grad_norm": 0.05869488790631294, "learning_rate": 7.247079089128415e-06, "loss": 0.0018, "step": 101810 }, { "epoch": 1.6660394338542093, "grad_norm": 0.08513905107975006, "learning_rate": 7.246441275902401e-06, "loss": 0.0013, "step": 101820 }, { "epoch": 1.666203059805285, "grad_norm": 0.03893130645155907, "learning_rate": 7.245803416873357e-06, "loss": 0.0014, "step": 101830 }, { "epoch": 1.666366685756361, "grad_norm": 0.036403920501470566, "learning_rate": 7.245165512054294e-06, "loss": 0.0016, "step": 101840 }, { "epoch": 1.6665303117074368, "grad_norm": 0.0527021586894989, "learning_rate": 7.244527561458214e-06, "loss": 0.0011, "step": 101850 }, { "epoch": 1.6666939376585126, "grad_norm": 0.04546986147761345, "learning_rate": 7.243889565098127e-06, "loss": 0.0011, "step": 101860 }, { "epoch": 1.6668575636095886, "grad_norm": 0.02808457426726818, "learning_rate": 7.243251522987038e-06, "loss": 0.0007, "step": 101870 }, { "epoch": 1.6670211895606644, "grad_norm": 0.03530687093734741, "learning_rate": 7.242613435137961e-06, "loss": 0.0016, "step": 101880 }, { "epoch": 1.6671848155117401, "grad_norm": 0.027286389842629433, "learning_rate": 7.2419753015639015e-06, "loss": 0.0012, "step": 101890 }, { "epoch": 1.6673484414628161, "grad_norm": 0.032380033284425735, "learning_rate": 7.241337122277874e-06, "loss": 0.0018, "step": 101900 }, { "epoch": 1.6675120674138917, "grad_norm": 0.1438898742198944, "learning_rate": 7.240698897292886e-06, "loss": 0.0016, "step": 101910 }, { "epoch": 1.6676756933649677, "grad_norm": 0.049850501120090485, "learning_rate": 7.2400606266219564e-06, "loss": 0.0008, "step": 101920 }, { "epoch": 1.6678393193160437, "grad_norm": 0.012970827519893646, "learning_rate": 7.239422310278092e-06, "loss": 0.0009, "step": 101930 }, { "epoch": 1.6680029452671192, "grad_norm": 0.049842722713947296, "learning_rate": 7.238783948274315e-06, "loss": 0.0013, "step": 101940 }, { "epoch": 1.6681665712181952, "grad_norm": 0.053950581699609756, "learning_rate": 7.2381455406236354e-06, "loss": 0.0028, "step": 101950 }, { "epoch": 1.668330197169271, "grad_norm": 0.013066639192402363, "learning_rate": 7.237507087339073e-06, "loss": 0.001, "step": 101960 }, { "epoch": 1.6684938231203468, "grad_norm": 0.04046615958213806, "learning_rate": 7.236868588433642e-06, "loss": 0.001, "step": 101970 }, { "epoch": 1.6686574490714228, "grad_norm": 0.11720210313796997, "learning_rate": 7.236230043920365e-06, "loss": 0.0017, "step": 101980 }, { "epoch": 1.6688210750224985, "grad_norm": 0.07497010380029678, "learning_rate": 7.235591453812258e-06, "loss": 0.0018, "step": 101990 }, { "epoch": 1.6689847009735743, "grad_norm": 0.05423549562692642, "learning_rate": 7.234952818122343e-06, "loss": 0.001, "step": 102000 }, { "epoch": 1.6691483269246503, "grad_norm": 0.020595289766788483, "learning_rate": 7.234314136863641e-06, "loss": 0.0012, "step": 102010 }, { "epoch": 1.669311952875726, "grad_norm": 0.011020096018910408, "learning_rate": 7.233675410049172e-06, "loss": 0.0027, "step": 102020 }, { "epoch": 1.6694755788268019, "grad_norm": 0.04626615718007088, "learning_rate": 7.233036637691964e-06, "loss": 0.0019, "step": 102030 }, { "epoch": 1.6696392047778779, "grad_norm": 0.03185964375734329, "learning_rate": 7.232397819805035e-06, "loss": 0.0007, "step": 102040 }, { "epoch": 1.6698028307289536, "grad_norm": 0.21498289704322815, "learning_rate": 7.231758956401415e-06, "loss": 0.0025, "step": 102050 }, { "epoch": 1.6699664566800294, "grad_norm": 0.13030818104743958, "learning_rate": 7.231120047494128e-06, "loss": 0.0016, "step": 102060 }, { "epoch": 1.6701300826311054, "grad_norm": 0.05935397371649742, "learning_rate": 7.2304810930962e-06, "loss": 0.0012, "step": 102070 }, { "epoch": 1.6702937085821812, "grad_norm": 0.07265820354223251, "learning_rate": 7.2298420932206605e-06, "loss": 0.0014, "step": 102080 }, { "epoch": 1.670457334533257, "grad_norm": 0.13691243529319763, "learning_rate": 7.229203047880536e-06, "loss": 0.002, "step": 102090 }, { "epoch": 1.670620960484333, "grad_norm": 0.0642358660697937, "learning_rate": 7.228563957088858e-06, "loss": 0.0015, "step": 102100 }, { "epoch": 1.6707845864354085, "grad_norm": 0.03644563630223274, "learning_rate": 7.2279248208586565e-06, "loss": 0.0015, "step": 102110 }, { "epoch": 1.6709482123864845, "grad_norm": 0.10100129246711731, "learning_rate": 7.227285639202961e-06, "loss": 0.0014, "step": 102120 }, { "epoch": 1.6711118383375605, "grad_norm": 0.09121351689100266, "learning_rate": 7.226646412134807e-06, "loss": 0.0018, "step": 102130 }, { "epoch": 1.671275464288636, "grad_norm": 0.028511231765151024, "learning_rate": 7.226007139667226e-06, "loss": 0.0014, "step": 102140 }, { "epoch": 1.671439090239712, "grad_norm": 0.012813701294362545, "learning_rate": 7.225367821813253e-06, "loss": 0.0012, "step": 102150 }, { "epoch": 1.6716027161907878, "grad_norm": 0.115474171936512, "learning_rate": 7.224728458585921e-06, "loss": 0.002, "step": 102160 }, { "epoch": 1.6717663421418636, "grad_norm": 0.0598750039935112, "learning_rate": 7.224089049998269e-06, "loss": 0.0012, "step": 102170 }, { "epoch": 1.6719299680929396, "grad_norm": 0.022440722212195396, "learning_rate": 7.223449596063334e-06, "loss": 0.0012, "step": 102180 }, { "epoch": 1.6720935940440154, "grad_norm": 0.02938864193856716, "learning_rate": 7.222810096794152e-06, "loss": 0.001, "step": 102190 }, { "epoch": 1.6722572199950911, "grad_norm": 0.004681746941059828, "learning_rate": 7.222170552203761e-06, "loss": 0.0011, "step": 102200 }, { "epoch": 1.6724208459461671, "grad_norm": 0.06946520507335663, "learning_rate": 7.2215309623052034e-06, "loss": 0.0019, "step": 102210 }, { "epoch": 1.672584471897243, "grad_norm": 0.06486685574054718, "learning_rate": 7.220891327111519e-06, "loss": 0.0022, "step": 102220 }, { "epoch": 1.6727480978483187, "grad_norm": 0.016430964693427086, "learning_rate": 7.220251646635749e-06, "loss": 0.0012, "step": 102230 }, { "epoch": 1.6729117237993947, "grad_norm": 0.20066121220588684, "learning_rate": 7.219611920890937e-06, "loss": 0.0019, "step": 102240 }, { "epoch": 1.6730753497504705, "grad_norm": 0.07751838862895966, "learning_rate": 7.218972149890124e-06, "loss": 0.0009, "step": 102250 }, { "epoch": 1.6732389757015462, "grad_norm": 0.007567335851490498, "learning_rate": 7.2183323336463576e-06, "loss": 0.0009, "step": 102260 }, { "epoch": 1.6734026016526222, "grad_norm": 0.034599047154188156, "learning_rate": 7.217692472172681e-06, "loss": 0.0007, "step": 102270 }, { "epoch": 1.673566227603698, "grad_norm": 0.037313856184482574, "learning_rate": 7.217052565482141e-06, "loss": 0.0018, "step": 102280 }, { "epoch": 1.6737298535547738, "grad_norm": 0.028277253732085228, "learning_rate": 7.216412613587784e-06, "loss": 0.0011, "step": 102290 }, { "epoch": 1.6738934795058498, "grad_norm": 0.04978024214506149, "learning_rate": 7.21577261650266e-06, "loss": 0.0013, "step": 102300 }, { "epoch": 1.6740571054569253, "grad_norm": 0.04651666432619095, "learning_rate": 7.215132574239816e-06, "loss": 0.0016, "step": 102310 }, { "epoch": 1.6742207314080013, "grad_norm": 0.06196007877588272, "learning_rate": 7.214492486812303e-06, "loss": 0.0017, "step": 102320 }, { "epoch": 1.6743843573590773, "grad_norm": 0.044528596103191376, "learning_rate": 7.2138523542331715e-06, "loss": 0.0014, "step": 102330 }, { "epoch": 1.6745479833101529, "grad_norm": 0.04723607376217842, "learning_rate": 7.213212176515474e-06, "loss": 0.0011, "step": 102340 }, { "epoch": 1.6747116092612289, "grad_norm": 0.06003262847661972, "learning_rate": 7.212571953672262e-06, "loss": 0.0012, "step": 102350 }, { "epoch": 1.6748752352123046, "grad_norm": 0.0546569898724556, "learning_rate": 7.211931685716589e-06, "loss": 0.0013, "step": 102360 }, { "epoch": 1.6750388611633804, "grad_norm": 0.029747677966952324, "learning_rate": 7.211291372661511e-06, "loss": 0.0024, "step": 102370 }, { "epoch": 1.6752024871144564, "grad_norm": 0.02273760549724102, "learning_rate": 7.2106510145200814e-06, "loss": 0.0015, "step": 102380 }, { "epoch": 1.6753661130655322, "grad_norm": 0.046275414526462555, "learning_rate": 7.210010611305358e-06, "loss": 0.0019, "step": 102390 }, { "epoch": 1.675529739016608, "grad_norm": 0.0648576095700264, "learning_rate": 7.209370163030398e-06, "loss": 0.0015, "step": 102400 }, { "epoch": 1.675693364967684, "grad_norm": 0.015689268708229065, "learning_rate": 7.20872966970826e-06, "loss": 0.001, "step": 102410 }, { "epoch": 1.6758569909187597, "grad_norm": 0.05077148601412773, "learning_rate": 7.208089131352002e-06, "loss": 0.0016, "step": 102420 }, { "epoch": 1.6760206168698355, "grad_norm": 0.074223592877388, "learning_rate": 7.207448547974685e-06, "loss": 0.0013, "step": 102430 }, { "epoch": 1.6761842428209115, "grad_norm": 0.007689600344747305, "learning_rate": 7.2068079195893696e-06, "loss": 0.0007, "step": 102440 }, { "epoch": 1.6763478687719873, "grad_norm": 0.029315786436200142, "learning_rate": 7.206167246209116e-06, "loss": 0.0012, "step": 102450 }, { "epoch": 1.676511494723063, "grad_norm": 0.09348084777593613, "learning_rate": 7.20552652784699e-06, "loss": 0.0016, "step": 102460 }, { "epoch": 1.676675120674139, "grad_norm": 0.06347664445638657, "learning_rate": 7.204885764516052e-06, "loss": 0.0012, "step": 102470 }, { "epoch": 1.6768387466252146, "grad_norm": 0.06882255524396896, "learning_rate": 7.20424495622937e-06, "loss": 0.0013, "step": 102480 }, { "epoch": 1.6770023725762906, "grad_norm": 0.0395415797829628, "learning_rate": 7.203604103000008e-06, "loss": 0.0026, "step": 102490 }, { "epoch": 1.6771659985273666, "grad_norm": 0.2149331271648407, "learning_rate": 7.202963204841033e-06, "loss": 0.0023, "step": 102500 }, { "epoch": 1.6773296244784421, "grad_norm": 0.028183847665786743, "learning_rate": 7.20232226176551e-06, "loss": 0.0006, "step": 102510 }, { "epoch": 1.6774932504295181, "grad_norm": 0.03428283706307411, "learning_rate": 7.201681273786511e-06, "loss": 0.0011, "step": 102520 }, { "epoch": 1.677656876380594, "grad_norm": 0.04054995998740196, "learning_rate": 7.201040240917101e-06, "loss": 0.0015, "step": 102530 }, { "epoch": 1.6778205023316697, "grad_norm": 0.022050023078918457, "learning_rate": 7.2003991631703554e-06, "loss": 0.0008, "step": 102540 }, { "epoch": 1.6779841282827457, "grad_norm": 0.10638316720724106, "learning_rate": 7.199758040559339e-06, "loss": 0.0015, "step": 102550 }, { "epoch": 1.6781477542338215, "grad_norm": 0.045363880693912506, "learning_rate": 7.199116873097129e-06, "loss": 0.0011, "step": 102560 }, { "epoch": 1.6783113801848972, "grad_norm": 0.01193652767688036, "learning_rate": 7.198475660796795e-06, "loss": 0.0016, "step": 102570 }, { "epoch": 1.6784750061359732, "grad_norm": 0.017078083008527756, "learning_rate": 7.1978344036714145e-06, "loss": 0.0012, "step": 102580 }, { "epoch": 1.678638632087049, "grad_norm": 0.05584794282913208, "learning_rate": 7.197193101734057e-06, "loss": 0.0019, "step": 102590 }, { "epoch": 1.6788022580381248, "grad_norm": 0.0125889265909791, "learning_rate": 7.196551754997802e-06, "loss": 0.001, "step": 102600 }, { "epoch": 1.6789658839892008, "grad_norm": 0.08713118731975555, "learning_rate": 7.195910363475725e-06, "loss": 0.0077, "step": 102610 }, { "epoch": 1.6791295099402765, "grad_norm": 0.22575795650482178, "learning_rate": 7.1952689271809025e-06, "loss": 0.0018, "step": 102620 }, { "epoch": 1.6792931358913523, "grad_norm": 0.002584181958809495, "learning_rate": 7.194627446126414e-06, "loss": 0.0013, "step": 102630 }, { "epoch": 1.6794567618424283, "grad_norm": 0.03682943806052208, "learning_rate": 7.193985920325339e-06, "loss": 0.0011, "step": 102640 }, { "epoch": 1.679620387793504, "grad_norm": 0.032361965626478195, "learning_rate": 7.193344349790756e-06, "loss": 0.0016, "step": 102650 }, { "epoch": 1.6797840137445799, "grad_norm": 0.030108729377388954, "learning_rate": 7.192702734535747e-06, "loss": 0.0013, "step": 102660 }, { "epoch": 1.6799476396956559, "grad_norm": 0.1309002935886383, "learning_rate": 7.192061074573396e-06, "loss": 0.0013, "step": 102670 }, { "epoch": 1.6801112656467314, "grad_norm": 0.01033733505755663, "learning_rate": 7.191419369916781e-06, "loss": 0.0026, "step": 102680 }, { "epoch": 1.6802748915978074, "grad_norm": 0.020969601348042488, "learning_rate": 7.190777620578991e-06, "loss": 0.001, "step": 102690 }, { "epoch": 1.6804385175488834, "grad_norm": 0.028345271944999695, "learning_rate": 7.190135826573107e-06, "loss": 0.0008, "step": 102700 }, { "epoch": 1.680602143499959, "grad_norm": 0.031182732433080673, "learning_rate": 7.189493987912217e-06, "loss": 0.0016, "step": 102710 }, { "epoch": 1.680765769451035, "grad_norm": 0.02775944210588932, "learning_rate": 7.188852104609406e-06, "loss": 0.0019, "step": 102720 }, { "epoch": 1.6809293954021107, "grad_norm": 0.053510475903749466, "learning_rate": 7.188210176677763e-06, "loss": 0.0012, "step": 102730 }, { "epoch": 1.6810930213531865, "grad_norm": 0.05883682891726494, "learning_rate": 7.187568204130375e-06, "loss": 0.0018, "step": 102740 }, { "epoch": 1.6812566473042625, "grad_norm": 0.08468199521303177, "learning_rate": 7.186926186980333e-06, "loss": 0.002, "step": 102750 }, { "epoch": 1.6814202732553383, "grad_norm": 0.21756570041179657, "learning_rate": 7.186284125240724e-06, "loss": 0.0014, "step": 102760 }, { "epoch": 1.681583899206414, "grad_norm": 0.38003844022750854, "learning_rate": 7.185642018924643e-06, "loss": 0.0015, "step": 102770 }, { "epoch": 1.68174752515749, "grad_norm": 0.047671280801296234, "learning_rate": 7.184999868045178e-06, "loss": 0.0017, "step": 102780 }, { "epoch": 1.6819111511085658, "grad_norm": 0.07649466395378113, "learning_rate": 7.184357672615426e-06, "loss": 0.0016, "step": 102790 }, { "epoch": 1.6820747770596416, "grad_norm": 0.022571666166186333, "learning_rate": 7.183715432648477e-06, "loss": 0.0019, "step": 102800 }, { "epoch": 1.6822384030107176, "grad_norm": 0.11310169845819473, "learning_rate": 7.183073148157429e-06, "loss": 0.0013, "step": 102810 }, { "epoch": 1.6824020289617934, "grad_norm": 0.01647929847240448, "learning_rate": 7.182430819155375e-06, "loss": 0.0015, "step": 102820 }, { "epoch": 1.6825656549128691, "grad_norm": 0.0511794351041317, "learning_rate": 7.181788445655413e-06, "loss": 0.0015, "step": 102830 }, { "epoch": 1.6827292808639451, "grad_norm": 0.012281819246709347, "learning_rate": 7.18114602767064e-06, "loss": 0.0011, "step": 102840 }, { "epoch": 1.682892906815021, "grad_norm": 0.08178556710481644, "learning_rate": 7.1805035652141556e-06, "loss": 0.0012, "step": 102850 }, { "epoch": 1.6830565327660967, "grad_norm": 0.04631522670388222, "learning_rate": 7.1798610582990565e-06, "loss": 0.0011, "step": 102860 }, { "epoch": 1.6832201587171727, "grad_norm": 0.018380092456936836, "learning_rate": 7.179218506938445e-06, "loss": 0.001, "step": 102870 }, { "epoch": 1.6833837846682482, "grad_norm": 0.11745163053274155, "learning_rate": 7.178575911145421e-06, "loss": 0.0016, "step": 102880 }, { "epoch": 1.6835474106193242, "grad_norm": 0.015270231291651726, "learning_rate": 7.177933270933088e-06, "loss": 0.0017, "step": 102890 }, { "epoch": 1.6837110365704002, "grad_norm": 0.010296421125531197, "learning_rate": 7.177290586314547e-06, "loss": 0.0011, "step": 102900 }, { "epoch": 1.6838746625214758, "grad_norm": 0.0750584676861763, "learning_rate": 7.176647857302903e-06, "loss": 0.0013, "step": 102910 }, { "epoch": 1.6840382884725518, "grad_norm": 0.02526324801146984, "learning_rate": 7.17600508391126e-06, "loss": 0.0011, "step": 102920 }, { "epoch": 1.6842019144236275, "grad_norm": 0.05933697149157524, "learning_rate": 7.175362266152725e-06, "loss": 0.0013, "step": 102930 }, { "epoch": 1.6843655403747033, "grad_norm": 0.05247284471988678, "learning_rate": 7.174719404040402e-06, "loss": 0.0072, "step": 102940 }, { "epoch": 1.6845291663257793, "grad_norm": 0.003347681602463126, "learning_rate": 7.1740764975874015e-06, "loss": 0.0016, "step": 102950 }, { "epoch": 1.684692792276855, "grad_norm": 0.046639375388622284, "learning_rate": 7.1734335468068295e-06, "loss": 0.0018, "step": 102960 }, { "epoch": 1.6848564182279309, "grad_norm": 0.03427392244338989, "learning_rate": 7.172790551711796e-06, "loss": 0.0012, "step": 102970 }, { "epoch": 1.6850200441790069, "grad_norm": 0.26539960503578186, "learning_rate": 7.172147512315411e-06, "loss": 0.0028, "step": 102980 }, { "epoch": 1.6851836701300826, "grad_norm": 0.07021181285381317, "learning_rate": 7.171504428630785e-06, "loss": 0.0023, "step": 102990 }, { "epoch": 1.6853472960811584, "grad_norm": 0.06758652627468109, "learning_rate": 7.170861300671031e-06, "loss": 0.0017, "step": 103000 }, { "epoch": 1.6855109220322344, "grad_norm": 0.15315701067447662, "learning_rate": 7.170218128449261e-06, "loss": 0.0011, "step": 103010 }, { "epoch": 1.6856745479833102, "grad_norm": 0.020727258175611496, "learning_rate": 7.16957491197859e-06, "loss": 0.002, "step": 103020 }, { "epoch": 1.685838173934386, "grad_norm": 0.03030742146074772, "learning_rate": 7.168931651272131e-06, "loss": 0.0018, "step": 103030 }, { "epoch": 1.686001799885462, "grad_norm": 0.12442433089017868, "learning_rate": 7.1682883463429995e-06, "loss": 0.0019, "step": 103040 }, { "epoch": 1.6861654258365377, "grad_norm": 0.04505766183137894, "learning_rate": 7.167644997204314e-06, "loss": 0.0013, "step": 103050 }, { "epoch": 1.6863290517876135, "grad_norm": 0.1453160047531128, "learning_rate": 7.167001603869191e-06, "loss": 0.0017, "step": 103060 }, { "epoch": 1.6864926777386895, "grad_norm": 0.03574560582637787, "learning_rate": 7.166358166350747e-06, "loss": 0.0013, "step": 103070 }, { "epoch": 1.686656303689765, "grad_norm": 0.004779661074280739, "learning_rate": 7.165714684662102e-06, "loss": 0.0024, "step": 103080 }, { "epoch": 1.686819929640841, "grad_norm": 0.04949164018034935, "learning_rate": 7.165071158816377e-06, "loss": 0.0027, "step": 103090 }, { "epoch": 1.686983555591917, "grad_norm": 0.010383925400674343, "learning_rate": 7.164427588826692e-06, "loss": 0.0008, "step": 103100 }, { "epoch": 1.6871471815429926, "grad_norm": 0.03252293914556503, "learning_rate": 7.16378397470617e-06, "loss": 0.0026, "step": 103110 }, { "epoch": 1.6873108074940686, "grad_norm": 0.04451068118214607, "learning_rate": 7.1631403164679326e-06, "loss": 0.0023, "step": 103120 }, { "epoch": 1.6874744334451444, "grad_norm": 0.008633768185973167, "learning_rate": 7.162496614125104e-06, "loss": 0.0021, "step": 103130 }, { "epoch": 1.6876380593962201, "grad_norm": 0.04684577137231827, "learning_rate": 7.161852867690808e-06, "loss": 0.0017, "step": 103140 }, { "epoch": 1.6878016853472961, "grad_norm": 0.176906019449234, "learning_rate": 7.161209077178171e-06, "loss": 0.0012, "step": 103150 }, { "epoch": 1.687965311298372, "grad_norm": 0.14892277121543884, "learning_rate": 7.160565242600319e-06, "loss": 0.0013, "step": 103160 }, { "epoch": 1.6881289372494477, "grad_norm": 0.08931376039981842, "learning_rate": 7.159921363970379e-06, "loss": 0.0022, "step": 103170 }, { "epoch": 1.6882925632005237, "grad_norm": 0.038526806980371475, "learning_rate": 7.159277441301481e-06, "loss": 0.0012, "step": 103180 }, { "epoch": 1.6884561891515995, "grad_norm": 0.023711513727903366, "learning_rate": 7.15863347460675e-06, "loss": 0.002, "step": 103190 }, { "epoch": 1.6886198151026752, "grad_norm": 0.018059035763144493, "learning_rate": 7.15798946389932e-06, "loss": 0.0008, "step": 103200 }, { "epoch": 1.6887834410537512, "grad_norm": 0.06450273841619492, "learning_rate": 7.157345409192319e-06, "loss": 0.0016, "step": 103210 }, { "epoch": 1.688947067004827, "grad_norm": 0.021571442484855652, "learning_rate": 7.15670131049888e-06, "loss": 0.0025, "step": 103220 }, { "epoch": 1.6891106929559028, "grad_norm": 0.027638893574476242, "learning_rate": 7.1560571678321355e-06, "loss": 0.0017, "step": 103230 }, { "epoch": 1.6892743189069788, "grad_norm": 0.0395815372467041, "learning_rate": 7.15541298120522e-06, "loss": 0.0019, "step": 103240 }, { "epoch": 1.6894379448580545, "grad_norm": 0.03017597459256649, "learning_rate": 7.154768750631266e-06, "loss": 0.0018, "step": 103250 }, { "epoch": 1.6896015708091303, "grad_norm": 0.08559595048427582, "learning_rate": 7.154124476123409e-06, "loss": 0.0009, "step": 103260 }, { "epoch": 1.6897651967602063, "grad_norm": 0.01708637923002243, "learning_rate": 7.153480157694787e-06, "loss": 0.0008, "step": 103270 }, { "epoch": 1.6899288227112819, "grad_norm": 0.027763502672314644, "learning_rate": 7.152835795358537e-06, "loss": 0.0009, "step": 103280 }, { "epoch": 1.6900924486623579, "grad_norm": 0.052683793008327484, "learning_rate": 7.152191389127796e-06, "loss": 0.0028, "step": 103290 }, { "epoch": 1.6902560746134339, "grad_norm": 0.058236055076122284, "learning_rate": 7.151546939015701e-06, "loss": 0.0018, "step": 103300 }, { "epoch": 1.6904197005645094, "grad_norm": 0.05201776325702667, "learning_rate": 7.150902445035395e-06, "loss": 0.0008, "step": 103310 }, { "epoch": 1.6905833265155854, "grad_norm": 0.06813503056764603, "learning_rate": 7.1502579072000165e-06, "loss": 0.0018, "step": 103320 }, { "epoch": 1.6907469524666612, "grad_norm": 0.034071363508701324, "learning_rate": 7.149613325522709e-06, "loss": 0.0009, "step": 103330 }, { "epoch": 1.690910578417737, "grad_norm": 0.23053209483623505, "learning_rate": 7.148968700016614e-06, "loss": 0.0029, "step": 103340 }, { "epoch": 1.691074204368813, "grad_norm": 0.041743237525224686, "learning_rate": 7.1483240306948744e-06, "loss": 0.0011, "step": 103350 }, { "epoch": 1.6912378303198887, "grad_norm": 0.01799536682665348, "learning_rate": 7.147679317570635e-06, "loss": 0.0011, "step": 103360 }, { "epoch": 1.6914014562709645, "grad_norm": 0.021002288907766342, "learning_rate": 7.147034560657041e-06, "loss": 0.002, "step": 103370 }, { "epoch": 1.6915650822220405, "grad_norm": 0.16935627162456512, "learning_rate": 7.146389759967237e-06, "loss": 0.0019, "step": 103380 }, { "epoch": 1.6917287081731163, "grad_norm": 0.053594835102558136, "learning_rate": 7.145744915514373e-06, "loss": 0.002, "step": 103390 }, { "epoch": 1.691892334124192, "grad_norm": 0.08300173282623291, "learning_rate": 7.145100027311594e-06, "loss": 0.0016, "step": 103400 }, { "epoch": 1.692055960075268, "grad_norm": 0.046689290553331375, "learning_rate": 7.14445509537205e-06, "loss": 0.0007, "step": 103410 }, { "epoch": 1.6922195860263438, "grad_norm": 0.09650639444589615, "learning_rate": 7.14381011970889e-06, "loss": 0.0019, "step": 103420 }, { "epoch": 1.6923832119774196, "grad_norm": 0.03264879807829857, "learning_rate": 7.143165100335266e-06, "loss": 0.0028, "step": 103430 }, { "epoch": 1.6925468379284956, "grad_norm": 0.1052589938044548, "learning_rate": 7.142520037264328e-06, "loss": 0.0014, "step": 103440 }, { "epoch": 1.6927104638795711, "grad_norm": 0.013837005943059921, "learning_rate": 7.141874930509229e-06, "loss": 0.001, "step": 103450 }, { "epoch": 1.6928740898306471, "grad_norm": 0.024793749675154686, "learning_rate": 7.1412297800831206e-06, "loss": 0.0013, "step": 103460 }, { "epoch": 1.6930377157817231, "grad_norm": 0.011547005735337734, "learning_rate": 7.140584585999159e-06, "loss": 0.0016, "step": 103470 }, { "epoch": 1.6932013417327987, "grad_norm": 0.0677604004740715, "learning_rate": 7.1399393482704975e-06, "loss": 0.0021, "step": 103480 }, { "epoch": 1.6933649676838747, "grad_norm": 0.052013885229825974, "learning_rate": 7.139294066910293e-06, "loss": 0.0021, "step": 103490 }, { "epoch": 1.6935285936349505, "grad_norm": 0.020984267815947533, "learning_rate": 7.138648741931702e-06, "loss": 0.0009, "step": 103500 }, { "epoch": 1.6936922195860262, "grad_norm": 0.09034120291471481, "learning_rate": 7.138003373347882e-06, "loss": 0.0016, "step": 103510 }, { "epoch": 1.6938558455371022, "grad_norm": 0.1166573092341423, "learning_rate": 7.13735796117199e-06, "loss": 0.0012, "step": 103520 }, { "epoch": 1.694019471488178, "grad_norm": 0.001379474881105125, "learning_rate": 7.13671250541719e-06, "loss": 0.0008, "step": 103530 }, { "epoch": 1.6941830974392538, "grad_norm": 0.00989519152790308, "learning_rate": 7.136067006096636e-06, "loss": 0.0011, "step": 103540 }, { "epoch": 1.6943467233903298, "grad_norm": 0.05426046997308731, "learning_rate": 7.135421463223494e-06, "loss": 0.0007, "step": 103550 }, { "epoch": 1.6945103493414055, "grad_norm": 0.028256407007575035, "learning_rate": 7.134775876810924e-06, "loss": 0.0012, "step": 103560 }, { "epoch": 1.6946739752924813, "grad_norm": 0.026690511032938957, "learning_rate": 7.1341302468720906e-06, "loss": 0.0018, "step": 103570 }, { "epoch": 1.6948376012435573, "grad_norm": 0.04520764574408531, "learning_rate": 7.133484573420155e-06, "loss": 0.0028, "step": 103580 }, { "epoch": 1.695001227194633, "grad_norm": 0.02080061472952366, "learning_rate": 7.132838856468285e-06, "loss": 0.0022, "step": 103590 }, { "epoch": 1.6951648531457089, "grad_norm": 0.10823390632867813, "learning_rate": 7.132193096029644e-06, "loss": 0.0019, "step": 103600 }, { "epoch": 1.6953284790967849, "grad_norm": 0.03820024058222771, "learning_rate": 7.131547292117398e-06, "loss": 0.0022, "step": 103610 }, { "epoch": 1.6954921050478606, "grad_norm": 0.004931232891976833, "learning_rate": 7.130901444744717e-06, "loss": 0.0009, "step": 103620 }, { "epoch": 1.6956557309989364, "grad_norm": 0.210389643907547, "learning_rate": 7.130255553924766e-06, "loss": 0.0018, "step": 103630 }, { "epoch": 1.6958193569500124, "grad_norm": 0.04559078812599182, "learning_rate": 7.1296096196707185e-06, "loss": 0.0013, "step": 103640 }, { "epoch": 1.695982982901088, "grad_norm": 0.031402185559272766, "learning_rate": 7.12896364199574e-06, "loss": 0.0016, "step": 103650 }, { "epoch": 1.696146608852164, "grad_norm": 0.10293040424585342, "learning_rate": 7.128317620913004e-06, "loss": 0.0024, "step": 103660 }, { "epoch": 1.69631023480324, "grad_norm": 0.0010115274926647544, "learning_rate": 7.1276715564356825e-06, "loss": 0.0009, "step": 103670 }, { "epoch": 1.6964738607543155, "grad_norm": 0.06629026681184769, "learning_rate": 7.127025448576948e-06, "loss": 0.0019, "step": 103680 }, { "epoch": 1.6966374867053915, "grad_norm": 0.01973135955631733, "learning_rate": 7.12637929734997e-06, "loss": 0.0011, "step": 103690 }, { "epoch": 1.6968011126564673, "grad_norm": 0.02077295631170273, "learning_rate": 7.12573310276793e-06, "loss": 0.0022, "step": 103700 }, { "epoch": 1.696964738607543, "grad_norm": 0.03707075119018555, "learning_rate": 7.125086864843998e-06, "loss": 0.0018, "step": 103710 }, { "epoch": 1.697128364558619, "grad_norm": 0.004294439218938351, "learning_rate": 7.124440583591355e-06, "loss": 0.002, "step": 103720 }, { "epoch": 1.6972919905096948, "grad_norm": 0.1765844225883484, "learning_rate": 7.123794259023172e-06, "loss": 0.0021, "step": 103730 }, { "epoch": 1.6974556164607706, "grad_norm": 0.040959443897008896, "learning_rate": 7.123147891152632e-06, "loss": 0.0035, "step": 103740 }, { "epoch": 1.6976192424118466, "grad_norm": 0.09630906581878662, "learning_rate": 7.1225014799929115e-06, "loss": 0.0018, "step": 103750 }, { "epoch": 1.6977828683629224, "grad_norm": 0.07298273593187332, "learning_rate": 7.121855025557192e-06, "loss": 0.0013, "step": 103760 }, { "epoch": 1.6979464943139981, "grad_norm": 0.18379350006580353, "learning_rate": 7.1212085278586515e-06, "loss": 0.0015, "step": 103770 }, { "epoch": 1.6981101202650741, "grad_norm": 0.010446595959365368, "learning_rate": 7.120561986910475e-06, "loss": 0.0017, "step": 103780 }, { "epoch": 1.69827374621615, "grad_norm": 0.03242595121264458, "learning_rate": 7.119915402725841e-06, "loss": 0.0008, "step": 103790 }, { "epoch": 1.6984373721672257, "grad_norm": 0.05437513813376427, "learning_rate": 7.119268775317936e-06, "loss": 0.0026, "step": 103800 }, { "epoch": 1.6986009981183017, "grad_norm": 0.1554172933101654, "learning_rate": 7.118622104699943e-06, "loss": 0.0022, "step": 103810 }, { "epoch": 1.6987646240693774, "grad_norm": 0.004066268913447857, "learning_rate": 7.117975390885048e-06, "loss": 0.0019, "step": 103820 }, { "epoch": 1.6989282500204532, "grad_norm": 0.0034072911366820335, "learning_rate": 7.117328633886433e-06, "loss": 0.0009, "step": 103830 }, { "epoch": 1.6990918759715292, "grad_norm": 0.03676136955618858, "learning_rate": 7.116681833717292e-06, "loss": 0.0014, "step": 103840 }, { "epoch": 1.6992555019226048, "grad_norm": 0.1301039308309555, "learning_rate": 7.116034990390805e-06, "loss": 0.0016, "step": 103850 }, { "epoch": 1.6994191278736808, "grad_norm": 0.011665447615087032, "learning_rate": 7.115388103920167e-06, "loss": 0.0008, "step": 103860 }, { "epoch": 1.6995827538247568, "grad_norm": 0.09138178080320358, "learning_rate": 7.114741174318563e-06, "loss": 0.002, "step": 103870 }, { "epoch": 1.6997463797758323, "grad_norm": 0.25102347135543823, "learning_rate": 7.114094201599185e-06, "loss": 0.0012, "step": 103880 }, { "epoch": 1.6999100057269083, "grad_norm": 0.11377573758363724, "learning_rate": 7.113447185775224e-06, "loss": 0.0014, "step": 103890 }, { "epoch": 1.700073631677984, "grad_norm": 0.06965705007314682, "learning_rate": 7.112800126859874e-06, "loss": 0.0011, "step": 103900 }, { "epoch": 1.7002372576290599, "grad_norm": 0.026122912764549255, "learning_rate": 7.112153024866324e-06, "loss": 0.0026, "step": 103910 }, { "epoch": 1.7004008835801359, "grad_norm": 0.02869570627808571, "learning_rate": 7.1115058798077715e-06, "loss": 0.0006, "step": 103920 }, { "epoch": 1.7005645095312116, "grad_norm": 0.062143635004758835, "learning_rate": 7.110858691697411e-06, "loss": 0.0014, "step": 103930 }, { "epoch": 1.7007281354822874, "grad_norm": 0.04115520417690277, "learning_rate": 7.110211460548435e-06, "loss": 0.0013, "step": 103940 }, { "epoch": 1.7008917614333634, "grad_norm": 0.07453031092882156, "learning_rate": 7.109564186374043e-06, "loss": 0.0043, "step": 103950 }, { "epoch": 1.7010553873844392, "grad_norm": 0.061380572617053986, "learning_rate": 7.10891686918743e-06, "loss": 0.0018, "step": 103960 }, { "epoch": 1.701219013335515, "grad_norm": 0.03138990327715874, "learning_rate": 7.108269509001797e-06, "loss": 0.0021, "step": 103970 }, { "epoch": 1.701382639286591, "grad_norm": 0.005745321046561003, "learning_rate": 7.107622105830342e-06, "loss": 0.0011, "step": 103980 }, { "epoch": 1.7015462652376667, "grad_norm": 0.07944558560848236, "learning_rate": 7.106974659686264e-06, "loss": 0.0019, "step": 103990 }, { "epoch": 1.7017098911887425, "grad_norm": 0.02996879629790783, "learning_rate": 7.106327170582764e-06, "loss": 0.0012, "step": 104000 }, { "epoch": 1.7018735171398185, "grad_norm": 0.01629439927637577, "learning_rate": 7.105679638533046e-06, "loss": 0.0038, "step": 104010 }, { "epoch": 1.7020371430908943, "grad_norm": 0.007294696290045977, "learning_rate": 7.105032063550309e-06, "loss": 0.002, "step": 104020 }, { "epoch": 1.70220076904197, "grad_norm": 0.006453908514231443, "learning_rate": 7.10438444564776e-06, "loss": 0.0009, "step": 104030 }, { "epoch": 1.702364394993046, "grad_norm": 0.023892536759376526, "learning_rate": 7.1037367848386015e-06, "loss": 0.001, "step": 104040 }, { "epoch": 1.7025280209441216, "grad_norm": 0.024456337094306946, "learning_rate": 7.103089081136039e-06, "loss": 0.0016, "step": 104050 }, { "epoch": 1.7026916468951976, "grad_norm": 0.03332405164837837, "learning_rate": 7.102441334553279e-06, "loss": 0.0013, "step": 104060 }, { "epoch": 1.7028552728462736, "grad_norm": 0.03364037722349167, "learning_rate": 7.101793545103528e-06, "loss": 0.0006, "step": 104070 }, { "epoch": 1.7030188987973491, "grad_norm": 0.0968012809753418, "learning_rate": 7.101145712799994e-06, "loss": 0.0023, "step": 104080 }, { "epoch": 1.7031825247484251, "grad_norm": 0.05592881515622139, "learning_rate": 7.100497837655886e-06, "loss": 0.0015, "step": 104090 }, { "epoch": 1.703346150699501, "grad_norm": 0.057801250368356705, "learning_rate": 7.0998499196844125e-06, "loss": 0.0009, "step": 104100 }, { "epoch": 1.7035097766505767, "grad_norm": 0.020404966548085213, "learning_rate": 7.099201958898787e-06, "loss": 0.0012, "step": 104110 }, { "epoch": 1.7036734026016527, "grad_norm": 0.07698772847652435, "learning_rate": 7.098553955312217e-06, "loss": 0.0022, "step": 104120 }, { "epoch": 1.7038370285527284, "grad_norm": 0.06792163848876953, "learning_rate": 7.097905908937919e-06, "loss": 0.001, "step": 104130 }, { "epoch": 1.7040006545038042, "grad_norm": 0.00906484853476286, "learning_rate": 7.0972578197891e-06, "loss": 0.0014, "step": 104140 }, { "epoch": 1.7041642804548802, "grad_norm": 0.004886595066636801, "learning_rate": 7.096609687878981e-06, "loss": 0.0011, "step": 104150 }, { "epoch": 1.704327906405956, "grad_norm": 0.01738443411886692, "learning_rate": 7.095961513220772e-06, "loss": 0.0026, "step": 104160 }, { "epoch": 1.7044915323570318, "grad_norm": 0.010890117846429348, "learning_rate": 7.095313295827692e-06, "loss": 0.0017, "step": 104170 }, { "epoch": 1.7046551583081078, "grad_norm": 0.0051125711761415005, "learning_rate": 7.094665035712954e-06, "loss": 0.0012, "step": 104180 }, { "epoch": 1.7048187842591835, "grad_norm": 0.05653606355190277, "learning_rate": 7.094016732889779e-06, "loss": 0.001, "step": 104190 }, { "epoch": 1.7049824102102593, "grad_norm": 0.026521451771259308, "learning_rate": 7.093368387371382e-06, "loss": 0.0016, "step": 104200 }, { "epoch": 1.7051460361613353, "grad_norm": 0.05263958126306534, "learning_rate": 7.092719999170985e-06, "loss": 0.0017, "step": 104210 }, { "epoch": 1.7053096621124109, "grad_norm": 0.028932036831974983, "learning_rate": 7.092071568301806e-06, "loss": 0.0018, "step": 104220 }, { "epoch": 1.7054732880634869, "grad_norm": 0.03061746247112751, "learning_rate": 7.091423094777068e-06, "loss": 0.0006, "step": 104230 }, { "epoch": 1.7056369140145629, "grad_norm": 0.07508138567209244, "learning_rate": 7.0907745786099915e-06, "loss": 0.0011, "step": 104240 }, { "epoch": 1.7058005399656384, "grad_norm": 0.025557266548275948, "learning_rate": 7.0901260198138e-06, "loss": 0.0023, "step": 104250 }, { "epoch": 1.7059641659167144, "grad_norm": 0.08593142777681351, "learning_rate": 7.089477418401716e-06, "loss": 0.0019, "step": 104260 }, { "epoch": 1.7061277918677902, "grad_norm": 0.03287171199917793, "learning_rate": 7.088828774386965e-06, "loss": 0.0014, "step": 104270 }, { "epoch": 1.706291417818866, "grad_norm": 0.03172576427459717, "learning_rate": 7.088180087782771e-06, "loss": 0.0016, "step": 104280 }, { "epoch": 1.706455043769942, "grad_norm": 0.045596275478601456, "learning_rate": 7.087531358602361e-06, "loss": 0.0015, "step": 104290 }, { "epoch": 1.7066186697210177, "grad_norm": 0.034286946058273315, "learning_rate": 7.086882586858962e-06, "loss": 0.0017, "step": 104300 }, { "epoch": 1.7067822956720935, "grad_norm": 0.0542944073677063, "learning_rate": 7.086233772565801e-06, "loss": 0.0019, "step": 104310 }, { "epoch": 1.7069459216231695, "grad_norm": 0.11853388696908951, "learning_rate": 7.08558491573611e-06, "loss": 0.0023, "step": 104320 }, { "epoch": 1.7071095475742453, "grad_norm": 0.05501984432339668, "learning_rate": 7.084936016383115e-06, "loss": 0.0009, "step": 104330 }, { "epoch": 1.707273173525321, "grad_norm": 0.013568326830863953, "learning_rate": 7.084287074520047e-06, "loss": 0.0011, "step": 104340 }, { "epoch": 1.707436799476397, "grad_norm": 0.050235819071531296, "learning_rate": 7.0836380901601385e-06, "loss": 0.0016, "step": 104350 }, { "epoch": 1.7076004254274728, "grad_norm": 0.05272424593567848, "learning_rate": 7.082989063316623e-06, "loss": 0.0012, "step": 104360 }, { "epoch": 1.7077640513785486, "grad_norm": 0.024718429893255234, "learning_rate": 7.082339994002731e-06, "loss": 0.0009, "step": 104370 }, { "epoch": 1.7079276773296246, "grad_norm": 0.08013209700584412, "learning_rate": 7.0816908822316975e-06, "loss": 0.0022, "step": 104380 }, { "epoch": 1.7080913032807004, "grad_norm": 0.058272797614336014, "learning_rate": 7.081041728016756e-06, "loss": 0.0013, "step": 104390 }, { "epoch": 1.7082549292317761, "grad_norm": 0.07932526618242264, "learning_rate": 7.080392531371145e-06, "loss": 0.0019, "step": 104400 }, { "epoch": 1.7084185551828521, "grad_norm": 0.033774327486753464, "learning_rate": 7.079743292308099e-06, "loss": 0.0012, "step": 104410 }, { "epoch": 1.7085821811339277, "grad_norm": 0.03074910305440426, "learning_rate": 7.079094010840857e-06, "loss": 0.0014, "step": 104420 }, { "epoch": 1.7087458070850037, "grad_norm": 0.02529175952076912, "learning_rate": 7.078444686982655e-06, "loss": 0.0013, "step": 104430 }, { "epoch": 1.7089094330360797, "grad_norm": 0.01655268669128418, "learning_rate": 7.077795320746735e-06, "loss": 0.0011, "step": 104440 }, { "epoch": 1.7090730589871552, "grad_norm": 0.08906667679548264, "learning_rate": 7.077145912146335e-06, "loss": 0.0017, "step": 104450 }, { "epoch": 1.7092366849382312, "grad_norm": 0.054719191044569016, "learning_rate": 7.076496461194696e-06, "loss": 0.0008, "step": 104460 }, { "epoch": 1.709400310889307, "grad_norm": 0.019515013322234154, "learning_rate": 7.07584696790506e-06, "loss": 0.0012, "step": 104470 }, { "epoch": 1.7095639368403828, "grad_norm": 0.018529264256358147, "learning_rate": 7.07519743229067e-06, "loss": 0.0021, "step": 104480 }, { "epoch": 1.7097275627914588, "grad_norm": 0.04177301749587059, "learning_rate": 7.074547854364768e-06, "loss": 0.0018, "step": 104490 }, { "epoch": 1.7098911887425345, "grad_norm": 0.06180957332253456, "learning_rate": 7.073898234140602e-06, "loss": 0.0008, "step": 104500 }, { "epoch": 1.7100548146936103, "grad_norm": 0.051801517605781555, "learning_rate": 7.0732485716314135e-06, "loss": 0.0011, "step": 104510 }, { "epoch": 1.7102184406446863, "grad_norm": 0.04275397211313248, "learning_rate": 7.0725988668504505e-06, "loss": 0.0015, "step": 104520 }, { "epoch": 1.710382066595762, "grad_norm": 0.10116671025753021, "learning_rate": 7.071949119810958e-06, "loss": 0.0019, "step": 104530 }, { "epoch": 1.7105456925468379, "grad_norm": 0.023375779390335083, "learning_rate": 7.0712993305261865e-06, "loss": 0.0008, "step": 104540 }, { "epoch": 1.7107093184979139, "grad_norm": 0.017345327883958817, "learning_rate": 7.070649499009382e-06, "loss": 0.0009, "step": 104550 }, { "epoch": 1.7108729444489896, "grad_norm": 0.02953844889998436, "learning_rate": 7.069999625273796e-06, "loss": 0.0012, "step": 104560 }, { "epoch": 1.7110365704000654, "grad_norm": 0.0844954252243042, "learning_rate": 7.069349709332679e-06, "loss": 0.0013, "step": 104570 }, { "epoch": 1.7112001963511414, "grad_norm": 0.0393495187163353, "learning_rate": 7.068699751199279e-06, "loss": 0.0012, "step": 104580 }, { "epoch": 1.7113638223022172, "grad_norm": 0.04957215115427971, "learning_rate": 7.068049750886853e-06, "loss": 0.0011, "step": 104590 }, { "epoch": 1.711527448253293, "grad_norm": 0.011843300424516201, "learning_rate": 7.067399708408649e-06, "loss": 0.0018, "step": 104600 }, { "epoch": 1.711691074204369, "grad_norm": 0.061587512493133545, "learning_rate": 7.066749623777926e-06, "loss": 0.0014, "step": 104610 }, { "epoch": 1.7118547001554445, "grad_norm": 0.29071131348609924, "learning_rate": 7.066099497007932e-06, "loss": 0.0014, "step": 104620 }, { "epoch": 1.7120183261065205, "grad_norm": 0.07364480197429657, "learning_rate": 7.06544932811193e-06, "loss": 0.002, "step": 104630 }, { "epoch": 1.7121819520575965, "grad_norm": 0.030398139730095863, "learning_rate": 7.06479911710317e-06, "loss": 0.001, "step": 104640 }, { "epoch": 1.712345578008672, "grad_norm": 0.023602470755577087, "learning_rate": 7.064148863994915e-06, "loss": 0.0025, "step": 104650 }, { "epoch": 1.712509203959748, "grad_norm": 0.06124921888113022, "learning_rate": 7.0634985688004174e-06, "loss": 0.0015, "step": 104660 }, { "epoch": 1.7126728299108238, "grad_norm": 0.045758605003356934, "learning_rate": 7.062848231532942e-06, "loss": 0.0016, "step": 104670 }, { "epoch": 1.7128364558618996, "grad_norm": 0.016435083001852036, "learning_rate": 7.062197852205742e-06, "loss": 0.0016, "step": 104680 }, { "epoch": 1.7130000818129756, "grad_norm": 0.05817462131381035, "learning_rate": 7.061547430832084e-06, "loss": 0.0017, "step": 104690 }, { "epoch": 1.7131637077640514, "grad_norm": 0.0642758309841156, "learning_rate": 7.060896967425226e-06, "loss": 0.0011, "step": 104700 }, { "epoch": 1.7133273337151271, "grad_norm": 0.1896945983171463, "learning_rate": 7.060246461998432e-06, "loss": 0.0017, "step": 104710 }, { "epoch": 1.7134909596662031, "grad_norm": 0.007102771196514368, "learning_rate": 7.059595914564965e-06, "loss": 0.0019, "step": 104720 }, { "epoch": 1.713654585617279, "grad_norm": 0.030701717361807823, "learning_rate": 7.058945325138089e-06, "loss": 0.0009, "step": 104730 }, { "epoch": 1.7138182115683547, "grad_norm": 0.08858716487884521, "learning_rate": 7.058294693731068e-06, "loss": 0.0013, "step": 104740 }, { "epoch": 1.7139818375194307, "grad_norm": 0.08099167048931122, "learning_rate": 7.057644020357171e-06, "loss": 0.0016, "step": 104750 }, { "epoch": 1.7141454634705064, "grad_norm": 0.1415717601776123, "learning_rate": 7.05699330502966e-06, "loss": 0.0012, "step": 104760 }, { "epoch": 1.7143090894215822, "grad_norm": 0.10229115188121796, "learning_rate": 7.056342547761805e-06, "loss": 0.0018, "step": 104770 }, { "epoch": 1.7144727153726582, "grad_norm": 0.05930432304739952, "learning_rate": 7.055691748566875e-06, "loss": 0.0018, "step": 104780 }, { "epoch": 1.714636341323734, "grad_norm": 0.049520134925842285, "learning_rate": 7.0550409074581394e-06, "loss": 0.0009, "step": 104790 }, { "epoch": 1.7147999672748098, "grad_norm": 0.06477637588977814, "learning_rate": 7.0543900244488665e-06, "loss": 0.0014, "step": 104800 }, { "epoch": 1.7149635932258858, "grad_norm": 0.023266462609171867, "learning_rate": 7.053739099552328e-06, "loss": 0.002, "step": 104810 }, { "epoch": 1.7151272191769613, "grad_norm": 0.05296730622649193, "learning_rate": 7.053088132781795e-06, "loss": 0.0015, "step": 104820 }, { "epoch": 1.7152908451280373, "grad_norm": 0.060941483825445175, "learning_rate": 7.052437124150544e-06, "loss": 0.0009, "step": 104830 }, { "epoch": 1.7154544710791133, "grad_norm": 0.013200799934566021, "learning_rate": 7.051786073671843e-06, "loss": 0.0095, "step": 104840 }, { "epoch": 1.7156180970301889, "grad_norm": 0.015537217259407043, "learning_rate": 7.05113498135897e-06, "loss": 0.0015, "step": 104850 }, { "epoch": 1.7157817229812649, "grad_norm": 0.03754744306206703, "learning_rate": 7.050483847225199e-06, "loss": 0.0015, "step": 104860 }, { "epoch": 1.7159453489323406, "grad_norm": 0.04672963544726372, "learning_rate": 7.049832671283807e-06, "loss": 0.0013, "step": 104870 }, { "epoch": 1.7161089748834164, "grad_norm": 0.03624200448393822, "learning_rate": 7.049181453548068e-06, "loss": 0.0011, "step": 104880 }, { "epoch": 1.7162726008344924, "grad_norm": 0.002681654877960682, "learning_rate": 7.048530194031264e-06, "loss": 0.0019, "step": 104890 }, { "epoch": 1.7164362267855682, "grad_norm": 0.05358370393514633, "learning_rate": 7.047878892746671e-06, "loss": 0.0025, "step": 104900 }, { "epoch": 1.716599852736644, "grad_norm": 0.04571151360869408, "learning_rate": 7.047227549707569e-06, "loss": 0.0016, "step": 104910 }, { "epoch": 1.71676347868772, "grad_norm": 0.029490841552615166, "learning_rate": 7.046576164927239e-06, "loss": 0.0013, "step": 104920 }, { "epoch": 1.7169271046387957, "grad_norm": 0.0035952047910541296, "learning_rate": 7.04592473841896e-06, "loss": 0.0008, "step": 104930 }, { "epoch": 1.7170907305898715, "grad_norm": 0.1795932948589325, "learning_rate": 7.045273270196017e-06, "loss": 0.0021, "step": 104940 }, { "epoch": 1.7172543565409475, "grad_norm": 0.020458709448575974, "learning_rate": 7.0446217602716905e-06, "loss": 0.0004, "step": 104950 }, { "epoch": 1.7174179824920233, "grad_norm": 0.07781112939119339, "learning_rate": 7.043970208659265e-06, "loss": 0.0014, "step": 104960 }, { "epoch": 1.717581608443099, "grad_norm": 0.02630995213985443, "learning_rate": 7.043318615372026e-06, "loss": 0.0013, "step": 104970 }, { "epoch": 1.717745234394175, "grad_norm": 0.10450643301010132, "learning_rate": 7.0426669804232586e-06, "loss": 0.0011, "step": 104980 }, { "epoch": 1.7179088603452508, "grad_norm": 0.05696291849017143, "learning_rate": 7.042015303826247e-06, "loss": 0.0008, "step": 104990 }, { "epoch": 1.7180724862963266, "grad_norm": 0.006633178796619177, "learning_rate": 7.041363585594282e-06, "loss": 0.0024, "step": 105000 }, { "epoch": 1.7182361122474026, "grad_norm": 0.023979108780622482, "learning_rate": 7.040711825740648e-06, "loss": 0.0019, "step": 105010 }, { "epoch": 1.7183997381984781, "grad_norm": 0.18095241487026215, "learning_rate": 7.040060024278636e-06, "loss": 0.0026, "step": 105020 }, { "epoch": 1.7185633641495541, "grad_norm": 0.022224267944693565, "learning_rate": 7.039408181221535e-06, "loss": 0.0011, "step": 105030 }, { "epoch": 1.7187269901006301, "grad_norm": 0.07280789315700531, "learning_rate": 7.038756296582638e-06, "loss": 0.0011, "step": 105040 }, { "epoch": 1.7188906160517057, "grad_norm": 0.17215994000434875, "learning_rate": 7.0381043703752305e-06, "loss": 0.0017, "step": 105050 }, { "epoch": 1.7190542420027817, "grad_norm": 0.012546819634735584, "learning_rate": 7.03745240261261e-06, "loss": 0.0017, "step": 105060 }, { "epoch": 1.7192178679538574, "grad_norm": 0.0554032027721405, "learning_rate": 7.036800393308068e-06, "loss": 0.002, "step": 105070 }, { "epoch": 1.7193814939049332, "grad_norm": 0.07795459032058716, "learning_rate": 7.036148342474898e-06, "loss": 0.0024, "step": 105080 }, { "epoch": 1.7195451198560092, "grad_norm": 0.07506353408098221, "learning_rate": 7.035496250126394e-06, "loss": 0.0016, "step": 105090 }, { "epoch": 1.719708745807085, "grad_norm": 0.03249547630548477, "learning_rate": 7.034844116275853e-06, "loss": 0.0035, "step": 105100 }, { "epoch": 1.7198723717581608, "grad_norm": 0.02833534963428974, "learning_rate": 7.0341919409365714e-06, "loss": 0.0025, "step": 105110 }, { "epoch": 1.7200359977092368, "grad_norm": 0.029294339939951897, "learning_rate": 7.033539724121846e-06, "loss": 0.0014, "step": 105120 }, { "epoch": 1.7201996236603125, "grad_norm": 0.013210630044341087, "learning_rate": 7.0328874658449744e-06, "loss": 0.0018, "step": 105130 }, { "epoch": 1.7203632496113883, "grad_norm": 0.13905026018619537, "learning_rate": 7.032235166119258e-06, "loss": 0.0024, "step": 105140 }, { "epoch": 1.7205268755624643, "grad_norm": 0.2469211220741272, "learning_rate": 7.0315828249579935e-06, "loss": 0.0032, "step": 105150 }, { "epoch": 1.72069050151354, "grad_norm": 0.07745424658060074, "learning_rate": 7.030930442374484e-06, "loss": 0.0014, "step": 105160 }, { "epoch": 1.7208541274646159, "grad_norm": 0.008835259824991226, "learning_rate": 7.030278018382029e-06, "loss": 0.0016, "step": 105170 }, { "epoch": 1.7210177534156919, "grad_norm": 0.03574865311384201, "learning_rate": 7.029625552993933e-06, "loss": 0.0015, "step": 105180 }, { "epoch": 1.7211813793667674, "grad_norm": 0.0494944304227829, "learning_rate": 7.028973046223496e-06, "loss": 0.0008, "step": 105190 }, { "epoch": 1.7213450053178434, "grad_norm": 0.031133556738495827, "learning_rate": 7.028320498084027e-06, "loss": 0.0023, "step": 105200 }, { "epoch": 1.7215086312689194, "grad_norm": 0.04704384133219719, "learning_rate": 7.027667908588826e-06, "loss": 0.0021, "step": 105210 }, { "epoch": 1.721672257219995, "grad_norm": 0.01595958322286606, "learning_rate": 7.027015277751201e-06, "loss": 0.0039, "step": 105220 }, { "epoch": 1.721835883171071, "grad_norm": 0.049043454229831696, "learning_rate": 7.02636260558446e-06, "loss": 0.0017, "step": 105230 }, { "epoch": 1.7219995091221467, "grad_norm": 0.020872151479125023, "learning_rate": 7.025709892101908e-06, "loss": 0.0018, "step": 105240 }, { "epoch": 1.7221631350732225, "grad_norm": 0.09741479903459549, "learning_rate": 7.025057137316853e-06, "loss": 0.0015, "step": 105250 }, { "epoch": 1.7223267610242985, "grad_norm": 0.026101358234882355, "learning_rate": 7.024404341242606e-06, "loss": 0.0016, "step": 105260 }, { "epoch": 1.7224903869753743, "grad_norm": 0.07292997092008591, "learning_rate": 7.023751503892477e-06, "loss": 0.0018, "step": 105270 }, { "epoch": 1.72265401292645, "grad_norm": 0.04740697145462036, "learning_rate": 7.023098625279774e-06, "loss": 0.0011, "step": 105280 }, { "epoch": 1.722817638877526, "grad_norm": 0.009998463094234467, "learning_rate": 7.022445705417813e-06, "loss": 0.0013, "step": 105290 }, { "epoch": 1.7229812648286018, "grad_norm": 0.042709801346063614, "learning_rate": 7.021792744319902e-06, "loss": 0.0011, "step": 105300 }, { "epoch": 1.7231448907796776, "grad_norm": 0.03205155208706856, "learning_rate": 7.021139741999358e-06, "loss": 0.0012, "step": 105310 }, { "epoch": 1.7233085167307536, "grad_norm": 0.22353298962116241, "learning_rate": 7.020486698469492e-06, "loss": 0.0021, "step": 105320 }, { "epoch": 1.7234721426818294, "grad_norm": 0.0064379191026091576, "learning_rate": 7.019833613743623e-06, "loss": 0.002, "step": 105330 }, { "epoch": 1.7236357686329051, "grad_norm": 0.030382905155420303, "learning_rate": 7.019180487835062e-06, "loss": 0.0033, "step": 105340 }, { "epoch": 1.7237993945839811, "grad_norm": 0.06191784143447876, "learning_rate": 7.0185273207571295e-06, "loss": 0.0024, "step": 105350 }, { "epoch": 1.723963020535057, "grad_norm": 0.07825969159603119, "learning_rate": 7.01787411252314e-06, "loss": 0.0015, "step": 105360 }, { "epoch": 1.7241266464861327, "grad_norm": 0.03816290572285652, "learning_rate": 7.017220863146416e-06, "loss": 0.0011, "step": 105370 }, { "epoch": 1.7242902724372087, "grad_norm": 0.07257892191410065, "learning_rate": 7.016567572640272e-06, "loss": 0.0018, "step": 105380 }, { "epoch": 1.7244538983882842, "grad_norm": 0.08256115019321442, "learning_rate": 7.015914241018031e-06, "loss": 0.0011, "step": 105390 }, { "epoch": 1.7246175243393602, "grad_norm": 0.007434937171638012, "learning_rate": 7.015260868293013e-06, "loss": 0.0013, "step": 105400 }, { "epoch": 1.7247811502904362, "grad_norm": 0.05260475352406502, "learning_rate": 7.0146074544785404e-06, "loss": 0.0018, "step": 105410 }, { "epoch": 1.7249447762415118, "grad_norm": 0.07367666065692902, "learning_rate": 7.013953999587935e-06, "loss": 0.0016, "step": 105420 }, { "epoch": 1.7251084021925878, "grad_norm": 0.021074572578072548, "learning_rate": 7.013300503634519e-06, "loss": 0.0007, "step": 105430 }, { "epoch": 1.7252720281436635, "grad_norm": 0.07345517724752426, "learning_rate": 7.01264696663162e-06, "loss": 0.0013, "step": 105440 }, { "epoch": 1.7254356540947393, "grad_norm": 0.12464892119169235, "learning_rate": 7.01199338859256e-06, "loss": 0.0014, "step": 105450 }, { "epoch": 1.7255992800458153, "grad_norm": 0.07831023633480072, "learning_rate": 7.011339769530665e-06, "loss": 0.0007, "step": 105460 }, { "epoch": 1.725762905996891, "grad_norm": 0.02398955635726452, "learning_rate": 7.010686109459265e-06, "loss": 0.0012, "step": 105470 }, { "epoch": 1.7259265319479669, "grad_norm": 0.16436031460762024, "learning_rate": 7.010032408391683e-06, "loss": 0.0021, "step": 105480 }, { "epoch": 1.7260901578990429, "grad_norm": 0.24264945089817047, "learning_rate": 7.0093786663412514e-06, "loss": 0.0024, "step": 105490 }, { "epoch": 1.7262537838501186, "grad_norm": 0.041498880833387375, "learning_rate": 7.008724883321297e-06, "loss": 0.001, "step": 105500 }, { "epoch": 1.7264174098011944, "grad_norm": 0.036223869770765305, "learning_rate": 7.008071059345151e-06, "loss": 0.0019, "step": 105510 }, { "epoch": 1.7265810357522704, "grad_norm": 0.0660877600312233, "learning_rate": 7.007417194426143e-06, "loss": 0.001, "step": 105520 }, { "epoch": 1.7267446617033462, "grad_norm": 0.05317332223057747, "learning_rate": 7.006763288577606e-06, "loss": 0.0019, "step": 105530 }, { "epoch": 1.726908287654422, "grad_norm": 0.012065144255757332, "learning_rate": 7.0061093418128726e-06, "loss": 0.0012, "step": 105540 }, { "epoch": 1.727071913605498, "grad_norm": 0.12088248878717422, "learning_rate": 7.005455354145275e-06, "loss": 0.0023, "step": 105550 }, { "epoch": 1.7272355395565737, "grad_norm": 0.03345303609967232, "learning_rate": 7.00480132558815e-06, "loss": 0.0007, "step": 105560 }, { "epoch": 1.7273991655076495, "grad_norm": 0.1438400000333786, "learning_rate": 7.00414725615483e-06, "loss": 0.0019, "step": 105570 }, { "epoch": 1.7275627914587255, "grad_norm": 0.1386096179485321, "learning_rate": 7.003493145858651e-06, "loss": 0.0016, "step": 105580 }, { "epoch": 1.727726417409801, "grad_norm": 0.06962073594331741, "learning_rate": 7.00283899471295e-06, "loss": 0.0012, "step": 105590 }, { "epoch": 1.727890043360877, "grad_norm": 0.046329565346241, "learning_rate": 7.002184802731069e-06, "loss": 0.0013, "step": 105600 }, { "epoch": 1.728053669311953, "grad_norm": 0.05247405916452408, "learning_rate": 7.001530569926339e-06, "loss": 0.0018, "step": 105610 }, { "epoch": 1.7282172952630286, "grad_norm": 0.14702565968036652, "learning_rate": 7.000876296312104e-06, "loss": 0.0019, "step": 105620 }, { "epoch": 1.7283809212141046, "grad_norm": 0.09147395193576813, "learning_rate": 7.000221981901703e-06, "loss": 0.0015, "step": 105630 }, { "epoch": 1.7285445471651804, "grad_norm": 0.02313242293894291, "learning_rate": 6.999567626708479e-06, "loss": 0.0017, "step": 105640 }, { "epoch": 1.7287081731162561, "grad_norm": 0.12120743840932846, "learning_rate": 6.998913230745769e-06, "loss": 0.0012, "step": 105650 }, { "epoch": 1.7288717990673321, "grad_norm": 0.11384135484695435, "learning_rate": 6.998258794026919e-06, "loss": 0.001, "step": 105660 }, { "epoch": 1.729035425018408, "grad_norm": 0.03341713175177574, "learning_rate": 6.997604316565272e-06, "loss": 0.0016, "step": 105670 }, { "epoch": 1.7291990509694837, "grad_norm": 0.05499788373708725, "learning_rate": 6.996949798374172e-06, "loss": 0.0023, "step": 105680 }, { "epoch": 1.7293626769205597, "grad_norm": 0.0898444727063179, "learning_rate": 6.996295239466964e-06, "loss": 0.002, "step": 105690 }, { "epoch": 1.7295263028716354, "grad_norm": 0.09240201860666275, "learning_rate": 6.995640639856993e-06, "loss": 0.0012, "step": 105700 }, { "epoch": 1.7296899288227112, "grad_norm": 0.07682747393846512, "learning_rate": 6.9949859995576065e-06, "loss": 0.0018, "step": 105710 }, { "epoch": 1.7298535547737872, "grad_norm": 0.061969559639692307, "learning_rate": 6.9943313185821535e-06, "loss": 0.0017, "step": 105720 }, { "epoch": 1.730017180724863, "grad_norm": 0.04497687146067619, "learning_rate": 6.993676596943979e-06, "loss": 0.0015, "step": 105730 }, { "epoch": 1.7301808066759388, "grad_norm": 0.0570676214993, "learning_rate": 6.993021834656437e-06, "loss": 0.0008, "step": 105740 }, { "epoch": 1.7303444326270148, "grad_norm": 0.0009125859360210598, "learning_rate": 6.992367031732872e-06, "loss": 0.001, "step": 105750 }, { "epoch": 1.7305080585780905, "grad_norm": 0.0335814468562603, "learning_rate": 6.99171218818664e-06, "loss": 0.002, "step": 105760 }, { "epoch": 1.7306716845291663, "grad_norm": 0.006543698254972696, "learning_rate": 6.991057304031088e-06, "loss": 0.0013, "step": 105770 }, { "epoch": 1.7308353104802423, "grad_norm": 0.06194464862346649, "learning_rate": 6.990402379279573e-06, "loss": 0.0013, "step": 105780 }, { "epoch": 1.7309989364313179, "grad_norm": 0.020253609865903854, "learning_rate": 6.989747413945446e-06, "loss": 0.0008, "step": 105790 }, { "epoch": 1.7311625623823939, "grad_norm": 0.08326154202222824, "learning_rate": 6.98909240804206e-06, "loss": 0.0008, "step": 105800 }, { "epoch": 1.7313261883334699, "grad_norm": 0.0237790048122406, "learning_rate": 6.988437361582772e-06, "loss": 0.0013, "step": 105810 }, { "epoch": 1.7314898142845454, "grad_norm": 0.05840442702174187, "learning_rate": 6.9877822745809385e-06, "loss": 0.0012, "step": 105820 }, { "epoch": 1.7316534402356214, "grad_norm": 0.03485627472400665, "learning_rate": 6.987127147049915e-06, "loss": 0.0018, "step": 105830 }, { "epoch": 1.7318170661866972, "grad_norm": 0.08668190240859985, "learning_rate": 6.986471979003058e-06, "loss": 0.0013, "step": 105840 }, { "epoch": 1.731980692137773, "grad_norm": 0.06459915637969971, "learning_rate": 6.9858167704537265e-06, "loss": 0.0017, "step": 105850 }, { "epoch": 1.732144318088849, "grad_norm": 0.057849179953336716, "learning_rate": 6.98516152141528e-06, "loss": 0.0014, "step": 105860 }, { "epoch": 1.7323079440399247, "grad_norm": 0.17658764123916626, "learning_rate": 6.9845062319010805e-06, "loss": 0.0013, "step": 105870 }, { "epoch": 1.7324715699910005, "grad_norm": 0.029165130108594894, "learning_rate": 6.983850901924484e-06, "loss": 0.0011, "step": 105880 }, { "epoch": 1.7326351959420765, "grad_norm": 0.07731308788061142, "learning_rate": 6.983195531498856e-06, "loss": 0.0016, "step": 105890 }, { "epoch": 1.7327988218931523, "grad_norm": 0.02628009207546711, "learning_rate": 6.982540120637558e-06, "loss": 0.0015, "step": 105900 }, { "epoch": 1.732962447844228, "grad_norm": 0.02581878751516342, "learning_rate": 6.981884669353952e-06, "loss": 0.0013, "step": 105910 }, { "epoch": 1.733126073795304, "grad_norm": 0.23674140870571136, "learning_rate": 6.981229177661403e-06, "loss": 0.0011, "step": 105920 }, { "epoch": 1.7332896997463798, "grad_norm": 0.04256238415837288, "learning_rate": 6.9805736455732775e-06, "loss": 0.0013, "step": 105930 }, { "epoch": 1.7334533256974556, "grad_norm": 0.08771298080682755, "learning_rate": 6.979918073102938e-06, "loss": 0.0014, "step": 105940 }, { "epoch": 1.7336169516485316, "grad_norm": 0.10737545043230057, "learning_rate": 6.979262460263755e-06, "loss": 0.0009, "step": 105950 }, { "epoch": 1.7337805775996074, "grad_norm": 0.02823605015873909, "learning_rate": 6.978606807069092e-06, "loss": 0.0018, "step": 105960 }, { "epoch": 1.7339442035506831, "grad_norm": 0.03783890977501869, "learning_rate": 6.9779511135323195e-06, "loss": 0.0018, "step": 105970 }, { "epoch": 1.7341078295017591, "grad_norm": 0.06800531595945358, "learning_rate": 6.977295379666807e-06, "loss": 0.0011, "step": 105980 }, { "epoch": 1.7342714554528347, "grad_norm": 0.07583371549844742, "learning_rate": 6.976639605485922e-06, "loss": 0.0012, "step": 105990 }, { "epoch": 1.7344350814039107, "grad_norm": 0.027715347707271576, "learning_rate": 6.975983791003037e-06, "loss": 0.0008, "step": 106000 }, { "epoch": 1.7345987073549864, "grad_norm": 0.016776621341705322, "learning_rate": 6.975327936231524e-06, "loss": 0.0011, "step": 106010 }, { "epoch": 1.7347623333060622, "grad_norm": 0.05385077744722366, "learning_rate": 6.974672041184752e-06, "loss": 0.0014, "step": 106020 }, { "epoch": 1.7349259592571382, "grad_norm": 0.2762097418308258, "learning_rate": 6.974016105876098e-06, "loss": 0.0016, "step": 106030 }, { "epoch": 1.735089585208214, "grad_norm": 0.08060509711503983, "learning_rate": 6.973360130318933e-06, "loss": 0.001, "step": 106040 }, { "epoch": 1.7352532111592898, "grad_norm": 0.04465053975582123, "learning_rate": 6.9727041145266364e-06, "loss": 0.002, "step": 106050 }, { "epoch": 1.7354168371103658, "grad_norm": 0.01779773086309433, "learning_rate": 6.972048058512577e-06, "loss": 0.001, "step": 106060 }, { "epoch": 1.7355804630614415, "grad_norm": 0.08685337752103806, "learning_rate": 6.971391962290137e-06, "loss": 0.0015, "step": 106070 }, { "epoch": 1.7357440890125173, "grad_norm": 0.042648836970329285, "learning_rate": 6.970735825872691e-06, "loss": 0.0012, "step": 106080 }, { "epoch": 1.7359077149635933, "grad_norm": 0.026115484535694122, "learning_rate": 6.970079649273618e-06, "loss": 0.0009, "step": 106090 }, { "epoch": 1.736071340914669, "grad_norm": 0.013151869177818298, "learning_rate": 6.9694234325062945e-06, "loss": 0.0008, "step": 106100 }, { "epoch": 1.7362349668657449, "grad_norm": 0.009662076830863953, "learning_rate": 6.968767175584105e-06, "loss": 0.001, "step": 106110 }, { "epoch": 1.7363985928168209, "grad_norm": 0.05237213522195816, "learning_rate": 6.968110878520426e-06, "loss": 0.0012, "step": 106120 }, { "epoch": 1.7365622187678966, "grad_norm": 1.294887661933899, "learning_rate": 6.96745454132864e-06, "loss": 0.0028, "step": 106130 }, { "epoch": 1.7367258447189724, "grad_norm": 0.07423088699579239, "learning_rate": 6.966798164022128e-06, "loss": 0.0014, "step": 106140 }, { "epoch": 1.7368894706700484, "grad_norm": 0.11334321647882462, "learning_rate": 6.966141746614276e-06, "loss": 0.0011, "step": 106150 }, { "epoch": 1.737053096621124, "grad_norm": 0.029991794377565384, "learning_rate": 6.965485289118464e-06, "loss": 0.0018, "step": 106160 }, { "epoch": 1.7372167225722, "grad_norm": 0.009270381182432175, "learning_rate": 6.964828791548079e-06, "loss": 0.0014, "step": 106170 }, { "epoch": 1.737380348523276, "grad_norm": 0.04584416374564171, "learning_rate": 6.964172253916507e-06, "loss": 0.0011, "step": 106180 }, { "epoch": 1.7375439744743515, "grad_norm": 0.16642406582832336, "learning_rate": 6.963515676237134e-06, "loss": 0.0017, "step": 106190 }, { "epoch": 1.7377076004254275, "grad_norm": 0.02655792236328125, "learning_rate": 6.962859058523344e-06, "loss": 0.0016, "step": 106200 }, { "epoch": 1.7378712263765033, "grad_norm": 0.06242433190345764, "learning_rate": 6.962202400788528e-06, "loss": 0.0009, "step": 106210 }, { "epoch": 1.738034852327579, "grad_norm": 0.12525038421154022, "learning_rate": 6.961545703046073e-06, "loss": 0.0025, "step": 106220 }, { "epoch": 1.738198478278655, "grad_norm": 0.04117996618151665, "learning_rate": 6.960888965309371e-06, "loss": 0.0012, "step": 106230 }, { "epoch": 1.7383621042297308, "grad_norm": 0.15994927287101746, "learning_rate": 6.96023218759181e-06, "loss": 0.0032, "step": 106240 }, { "epoch": 1.7385257301808066, "grad_norm": 0.07437343895435333, "learning_rate": 6.959575369906783e-06, "loss": 0.0009, "step": 106250 }, { "epoch": 1.7386893561318826, "grad_norm": 0.027391815558075905, "learning_rate": 6.958918512267678e-06, "loss": 0.0016, "step": 106260 }, { "epoch": 1.7388529820829584, "grad_norm": 0.08369435369968414, "learning_rate": 6.958261614687893e-06, "loss": 0.0015, "step": 106270 }, { "epoch": 1.7390166080340341, "grad_norm": 0.015832826495170593, "learning_rate": 6.957604677180818e-06, "loss": 0.0013, "step": 106280 }, { "epoch": 1.7391802339851101, "grad_norm": 0.19731497764587402, "learning_rate": 6.9569476997598484e-06, "loss": 0.0014, "step": 106290 }, { "epoch": 1.739343859936186, "grad_norm": 0.16175390779972076, "learning_rate": 6.95629068243838e-06, "loss": 0.0028, "step": 106300 }, { "epoch": 1.7395074858872617, "grad_norm": 0.02509847842156887, "learning_rate": 6.955633625229809e-06, "loss": 0.0009, "step": 106310 }, { "epoch": 1.7396711118383377, "grad_norm": 0.08855531364679337, "learning_rate": 6.954976528147531e-06, "loss": 0.0008, "step": 106320 }, { "epoch": 1.7398347377894134, "grad_norm": 0.07409320026636124, "learning_rate": 6.954319391204943e-06, "loss": 0.0013, "step": 106330 }, { "epoch": 1.7399983637404892, "grad_norm": 0.01939237304031849, "learning_rate": 6.953662214415446e-06, "loss": 0.0011, "step": 106340 }, { "epoch": 1.7401619896915652, "grad_norm": 0.06289539486169815, "learning_rate": 6.953004997792439e-06, "loss": 0.0014, "step": 106350 }, { "epoch": 1.7403256156426408, "grad_norm": 0.0302079189568758, "learning_rate": 6.952347741349319e-06, "loss": 0.0008, "step": 106360 }, { "epoch": 1.7404892415937168, "grad_norm": 0.04088163748383522, "learning_rate": 6.951690445099489e-06, "loss": 0.0005, "step": 106370 }, { "epoch": 1.7406528675447928, "grad_norm": 0.046773336827754974, "learning_rate": 6.951033109056353e-06, "loss": 0.0014, "step": 106380 }, { "epoch": 1.7408164934958683, "grad_norm": 0.08107298612594604, "learning_rate": 6.950375733233309e-06, "loss": 0.0017, "step": 106390 }, { "epoch": 1.7409801194469443, "grad_norm": 0.024163072928786278, "learning_rate": 6.949718317643764e-06, "loss": 0.0012, "step": 106400 }, { "epoch": 1.74114374539802, "grad_norm": 0.0336419977247715, "learning_rate": 6.949060862301119e-06, "loss": 0.0017, "step": 106410 }, { "epoch": 1.7413073713490959, "grad_norm": 0.03330640867352486, "learning_rate": 6.948403367218781e-06, "loss": 0.0022, "step": 106420 }, { "epoch": 1.7414709973001719, "grad_norm": 0.030625667423009872, "learning_rate": 6.947745832410157e-06, "loss": 0.0018, "step": 106430 }, { "epoch": 1.7416346232512476, "grad_norm": 0.05258497968316078, "learning_rate": 6.94708825788865e-06, "loss": 0.0014, "step": 106440 }, { "epoch": 1.7417982492023234, "grad_norm": 0.023726586252450943, "learning_rate": 6.94643064366767e-06, "loss": 0.0019, "step": 106450 }, { "epoch": 1.7419618751533994, "grad_norm": 0.04910953715443611, "learning_rate": 6.945772989760626e-06, "loss": 0.0018, "step": 106460 }, { "epoch": 1.7421255011044752, "grad_norm": 0.05473499745130539, "learning_rate": 6.945115296180923e-06, "loss": 0.0021, "step": 106470 }, { "epoch": 1.742289127055551, "grad_norm": 0.10702459514141083, "learning_rate": 6.944457562941975e-06, "loss": 0.0014, "step": 106480 }, { "epoch": 1.742452753006627, "grad_norm": 0.01999497413635254, "learning_rate": 6.943799790057191e-06, "loss": 0.0015, "step": 106490 }, { "epoch": 1.7426163789577027, "grad_norm": 0.1657571643590927, "learning_rate": 6.943141977539982e-06, "loss": 0.0022, "step": 106500 }, { "epoch": 1.7427800049087785, "grad_norm": 0.04981048032641411, "learning_rate": 6.942484125403761e-06, "loss": 0.0007, "step": 106510 }, { "epoch": 1.7429436308598545, "grad_norm": 0.01625736430287361, "learning_rate": 6.941826233661941e-06, "loss": 0.0008, "step": 106520 }, { "epoch": 1.7431072568109303, "grad_norm": 0.03243494778871536, "learning_rate": 6.941168302327934e-06, "loss": 0.0017, "step": 106530 }, { "epoch": 1.743270882762006, "grad_norm": 0.05957464873790741, "learning_rate": 6.940510331415158e-06, "loss": 0.0015, "step": 106540 }, { "epoch": 1.743434508713082, "grad_norm": 0.036173637956380844, "learning_rate": 6.939852320937026e-06, "loss": 0.0008, "step": 106550 }, { "epoch": 1.7435981346641576, "grad_norm": 0.07671251893043518, "learning_rate": 6.939194270906955e-06, "loss": 0.0014, "step": 106560 }, { "epoch": 1.7437617606152336, "grad_norm": 0.01574897952377796, "learning_rate": 6.938536181338361e-06, "loss": 0.001, "step": 106570 }, { "epoch": 1.7439253865663096, "grad_norm": 0.1061110645532608, "learning_rate": 6.937878052244664e-06, "loss": 0.0025, "step": 106580 }, { "epoch": 1.7440890125173851, "grad_norm": 0.030664905905723572, "learning_rate": 6.937219883639281e-06, "loss": 0.0055, "step": 106590 }, { "epoch": 1.7442526384684611, "grad_norm": 0.08034719526767731, "learning_rate": 6.936561675535634e-06, "loss": 0.0013, "step": 106600 }, { "epoch": 1.744416264419537, "grad_norm": 0.05381302163004875, "learning_rate": 6.935903427947139e-06, "loss": 0.0014, "step": 106610 }, { "epoch": 1.7445798903706127, "grad_norm": 0.024324998259544373, "learning_rate": 6.935245140887221e-06, "loss": 0.0012, "step": 106620 }, { "epoch": 1.7447435163216887, "grad_norm": 0.021137388423085213, "learning_rate": 6.9345868143693e-06, "loss": 0.001, "step": 106630 }, { "epoch": 1.7449071422727644, "grad_norm": 0.07523952424526215, "learning_rate": 6.9339284484068e-06, "loss": 0.0013, "step": 106640 }, { "epoch": 1.7450707682238402, "grad_norm": 0.05251080542802811, "learning_rate": 6.933270043013143e-06, "loss": 0.002, "step": 106650 }, { "epoch": 1.7452343941749162, "grad_norm": 0.041030459105968475, "learning_rate": 6.932611598201754e-06, "loss": 0.0013, "step": 106660 }, { "epoch": 1.745398020125992, "grad_norm": 0.08251962065696716, "learning_rate": 6.9319531139860585e-06, "loss": 0.0022, "step": 106670 }, { "epoch": 1.7455616460770678, "grad_norm": 0.03304721787571907, "learning_rate": 6.93129459037948e-06, "loss": 0.0018, "step": 106680 }, { "epoch": 1.7457252720281438, "grad_norm": 0.0495140478014946, "learning_rate": 6.93063602739545e-06, "loss": 0.0018, "step": 106690 }, { "epoch": 1.7458888979792195, "grad_norm": 0.036541711539030075, "learning_rate": 6.929977425047392e-06, "loss": 0.0008, "step": 106700 }, { "epoch": 1.7460525239302953, "grad_norm": 0.07962270826101303, "learning_rate": 6.929318783348736e-06, "loss": 0.0015, "step": 106710 }, { "epoch": 1.7462161498813713, "grad_norm": 0.06847725063562393, "learning_rate": 6.928660102312911e-06, "loss": 0.0012, "step": 106720 }, { "epoch": 1.746379775832447, "grad_norm": 0.03872443735599518, "learning_rate": 6.928001381953347e-06, "loss": 0.0015, "step": 106730 }, { "epoch": 1.7465434017835229, "grad_norm": 0.03455096855759621, "learning_rate": 6.9273426222834735e-06, "loss": 0.0029, "step": 106740 }, { "epoch": 1.7467070277345988, "grad_norm": 0.05769239366054535, "learning_rate": 6.926683823316725e-06, "loss": 0.0016, "step": 106750 }, { "epoch": 1.7468706536856744, "grad_norm": 0.11133593320846558, "learning_rate": 6.92602498506653e-06, "loss": 0.0011, "step": 106760 }, { "epoch": 1.7470342796367504, "grad_norm": 0.05094987154006958, "learning_rate": 6.925366107546324e-06, "loss": 0.0016, "step": 106770 }, { "epoch": 1.7471979055878264, "grad_norm": 0.08216430246829987, "learning_rate": 6.92470719076954e-06, "loss": 0.0016, "step": 106780 }, { "epoch": 1.747361531538902, "grad_norm": 0.017470832914114, "learning_rate": 6.924048234749614e-06, "loss": 0.0018, "step": 106790 }, { "epoch": 1.747525157489978, "grad_norm": 0.25535690784454346, "learning_rate": 6.923389239499982e-06, "loss": 0.0019, "step": 106800 }, { "epoch": 1.7476887834410537, "grad_norm": 0.028016727417707443, "learning_rate": 6.922730205034077e-06, "loss": 0.001, "step": 106810 }, { "epoch": 1.7478524093921295, "grad_norm": 0.022677741944789886, "learning_rate": 6.92207113136534e-06, "loss": 0.0009, "step": 106820 }, { "epoch": 1.7480160353432055, "grad_norm": 0.02477477304637432, "learning_rate": 6.921412018507206e-06, "loss": 0.0014, "step": 106830 }, { "epoch": 1.7481796612942813, "grad_norm": 0.08113706856966019, "learning_rate": 6.920752866473116e-06, "loss": 0.0027, "step": 106840 }, { "epoch": 1.748343287245357, "grad_norm": 0.03695002570748329, "learning_rate": 6.920093675276509e-06, "loss": 0.0015, "step": 106850 }, { "epoch": 1.748506913196433, "grad_norm": 0.0748840942978859, "learning_rate": 6.919434444930823e-06, "loss": 0.0026, "step": 106860 }, { "epoch": 1.7486705391475088, "grad_norm": 0.06683019548654556, "learning_rate": 6.9187751754495034e-06, "loss": 0.0011, "step": 106870 }, { "epoch": 1.7488341650985846, "grad_norm": 0.042707059532403946, "learning_rate": 6.918115866845988e-06, "loss": 0.0017, "step": 106880 }, { "epoch": 1.7489977910496606, "grad_norm": 0.04769822955131531, "learning_rate": 6.9174565191337226e-06, "loss": 0.0009, "step": 106890 }, { "epoch": 1.7491614170007364, "grad_norm": 0.043457332998514175, "learning_rate": 6.916797132326148e-06, "loss": 0.0004, "step": 106900 }, { "epoch": 1.7493250429518121, "grad_norm": 0.03539445996284485, "learning_rate": 6.9161377064367115e-06, "loss": 0.0021, "step": 106910 }, { "epoch": 1.7494886689028881, "grad_norm": 0.02117108181118965, "learning_rate": 6.915478241478857e-06, "loss": 0.0009, "step": 106920 }, { "epoch": 1.7496522948539637, "grad_norm": 0.0918683335185051, "learning_rate": 6.914818737466029e-06, "loss": 0.0015, "step": 106930 }, { "epoch": 1.7498159208050397, "grad_norm": 0.05135972425341606, "learning_rate": 6.914159194411676e-06, "loss": 0.0015, "step": 106940 }, { "epoch": 1.7499795467561157, "grad_norm": 0.006166053470224142, "learning_rate": 6.913499612329247e-06, "loss": 0.0014, "step": 106950 }, { "epoch": 1.7501431727071912, "grad_norm": 0.03512187674641609, "learning_rate": 6.9128399912321845e-06, "loss": 0.0015, "step": 106960 }, { "epoch": 1.7503067986582672, "grad_norm": 0.13610942661762238, "learning_rate": 6.912180331133944e-06, "loss": 0.0016, "step": 106970 }, { "epoch": 1.750470424609343, "grad_norm": 0.025666620582342148, "learning_rate": 6.911520632047972e-06, "loss": 0.0014, "step": 106980 }, { "epoch": 1.7506340505604188, "grad_norm": 0.047162484377622604, "learning_rate": 6.91086089398772e-06, "loss": 0.0007, "step": 106990 }, { "epoch": 1.7507976765114948, "grad_norm": 0.039821475744247437, "learning_rate": 6.91020111696664e-06, "loss": 0.0015, "step": 107000 }, { "epoch": 1.7509613024625705, "grad_norm": 0.004426253493875265, "learning_rate": 6.9095413009981826e-06, "loss": 0.0014, "step": 107010 }, { "epoch": 1.7511249284136463, "grad_norm": 0.001365557429380715, "learning_rate": 6.9088814460958035e-06, "loss": 0.0016, "step": 107020 }, { "epoch": 1.7512885543647223, "grad_norm": 0.047957926988601685, "learning_rate": 6.9082215522729534e-06, "loss": 0.0013, "step": 107030 }, { "epoch": 1.751452180315798, "grad_norm": 0.011832693591713905, "learning_rate": 6.907561619543091e-06, "loss": 0.0015, "step": 107040 }, { "epoch": 1.7516158062668739, "grad_norm": 0.458221971988678, "learning_rate": 6.906901647919667e-06, "loss": 0.0009, "step": 107050 }, { "epoch": 1.7517794322179499, "grad_norm": 0.011715072207152843, "learning_rate": 6.906241637416142e-06, "loss": 0.0027, "step": 107060 }, { "epoch": 1.7519430581690256, "grad_norm": 0.03355957195162773, "learning_rate": 6.9055815880459695e-06, "loss": 0.0019, "step": 107070 }, { "epoch": 1.7521066841201014, "grad_norm": 0.10419507324695587, "learning_rate": 6.904921499822611e-06, "loss": 0.0018, "step": 107080 }, { "epoch": 1.7522703100711774, "grad_norm": 0.11579752713441849, "learning_rate": 6.904261372759522e-06, "loss": 0.0017, "step": 107090 }, { "epoch": 1.7524339360222532, "grad_norm": 0.08631109446287155, "learning_rate": 6.903601206870164e-06, "loss": 0.0011, "step": 107100 }, { "epoch": 1.752597561973329, "grad_norm": 0.037390273064374924, "learning_rate": 6.902941002167995e-06, "loss": 0.0014, "step": 107110 }, { "epoch": 1.752761187924405, "grad_norm": 0.042208779603242874, "learning_rate": 6.90228075866648e-06, "loss": 0.0019, "step": 107120 }, { "epoch": 1.7529248138754805, "grad_norm": 0.11616631597280502, "learning_rate": 6.9016204763790755e-06, "loss": 0.0014, "step": 107130 }, { "epoch": 1.7530884398265565, "grad_norm": 0.05087413266301155, "learning_rate": 6.900960155319248e-06, "loss": 0.004, "step": 107140 }, { "epoch": 1.7532520657776325, "grad_norm": 0.028958125039935112, "learning_rate": 6.9002997955004584e-06, "loss": 0.0016, "step": 107150 }, { "epoch": 1.753415691728708, "grad_norm": 0.03099628910422325, "learning_rate": 6.899639396936173e-06, "loss": 0.0013, "step": 107160 }, { "epoch": 1.753579317679784, "grad_norm": 0.1572871208190918, "learning_rate": 6.898978959639856e-06, "loss": 0.0015, "step": 107170 }, { "epoch": 1.7537429436308598, "grad_norm": 0.041906971484422684, "learning_rate": 6.898318483624973e-06, "loss": 0.0012, "step": 107180 }, { "epoch": 1.7539065695819356, "grad_norm": 0.028544573113322258, "learning_rate": 6.89765796890499e-06, "loss": 0.0022, "step": 107190 }, { "epoch": 1.7540701955330116, "grad_norm": 0.01491416897624731, "learning_rate": 6.896997415493376e-06, "loss": 0.0006, "step": 107200 }, { "epoch": 1.7542338214840874, "grad_norm": 0.025302225723862648, "learning_rate": 6.896336823403597e-06, "loss": 0.0024, "step": 107210 }, { "epoch": 1.7543974474351631, "grad_norm": 0.07990511506795883, "learning_rate": 6.895676192649124e-06, "loss": 0.0021, "step": 107220 }, { "epoch": 1.7545610733862391, "grad_norm": 0.06048288941383362, "learning_rate": 6.895015523243424e-06, "loss": 0.0023, "step": 107230 }, { "epoch": 1.754724699337315, "grad_norm": 0.09192521870136261, "learning_rate": 6.894354815199971e-06, "loss": 0.0024, "step": 107240 }, { "epoch": 1.7548883252883907, "grad_norm": 0.0073741041123867035, "learning_rate": 6.893694068532233e-06, "loss": 0.0007, "step": 107250 }, { "epoch": 1.7550519512394667, "grad_norm": 0.03977558761835098, "learning_rate": 6.893033283253685e-06, "loss": 0.0024, "step": 107260 }, { "epoch": 1.7552155771905424, "grad_norm": 0.046216197311878204, "learning_rate": 6.892372459377797e-06, "loss": 0.0012, "step": 107270 }, { "epoch": 1.7553792031416182, "grad_norm": 0.07400636374950409, "learning_rate": 6.8917115969180445e-06, "loss": 0.0019, "step": 107280 }, { "epoch": 1.7555428290926942, "grad_norm": 0.03975580632686615, "learning_rate": 6.891050695887902e-06, "loss": 0.0019, "step": 107290 }, { "epoch": 1.75570645504377, "grad_norm": 0.048455096781253815, "learning_rate": 6.890389756300842e-06, "loss": 0.0011, "step": 107300 }, { "epoch": 1.7558700809948458, "grad_norm": 0.06911356747150421, "learning_rate": 6.8897287781703445e-06, "loss": 0.0014, "step": 107310 }, { "epoch": 1.7560337069459218, "grad_norm": 0.06722444295883179, "learning_rate": 6.8890677615098834e-06, "loss": 0.0018, "step": 107320 }, { "epoch": 1.7561973328969973, "grad_norm": 0.026390915736556053, "learning_rate": 6.888406706332938e-06, "loss": 0.0011, "step": 107330 }, { "epoch": 1.7563609588480733, "grad_norm": 0.14463607966899872, "learning_rate": 6.887745612652985e-06, "loss": 0.0015, "step": 107340 }, { "epoch": 1.7565245847991493, "grad_norm": 0.027621731162071228, "learning_rate": 6.887084480483506e-06, "loss": 0.0006, "step": 107350 }, { "epoch": 1.7566882107502249, "grad_norm": 0.06904059648513794, "learning_rate": 6.886423309837978e-06, "loss": 0.0008, "step": 107360 }, { "epoch": 1.7568518367013009, "grad_norm": 0.010113488882780075, "learning_rate": 6.885762100729883e-06, "loss": 0.0011, "step": 107370 }, { "epoch": 1.7570154626523766, "grad_norm": 0.026212386786937714, "learning_rate": 6.885100853172702e-06, "loss": 0.0019, "step": 107380 }, { "epoch": 1.7571790886034524, "grad_norm": 0.03087315894663334, "learning_rate": 6.884439567179919e-06, "loss": 0.001, "step": 107390 }, { "epoch": 1.7573427145545284, "grad_norm": 0.043972935527563095, "learning_rate": 6.883778242765016e-06, "loss": 0.0012, "step": 107400 }, { "epoch": 1.7575063405056042, "grad_norm": 0.2333526462316513, "learning_rate": 6.883116879941476e-06, "loss": 0.0015, "step": 107410 }, { "epoch": 1.75766996645668, "grad_norm": 0.011484364978969097, "learning_rate": 6.8824554787227845e-06, "loss": 0.0013, "step": 107420 }, { "epoch": 1.757833592407756, "grad_norm": 0.026227863505482674, "learning_rate": 6.881794039122427e-06, "loss": 0.0007, "step": 107430 }, { "epoch": 1.7579972183588317, "grad_norm": 0.28204435110092163, "learning_rate": 6.881132561153889e-06, "loss": 0.0022, "step": 107440 }, { "epoch": 1.7581608443099075, "grad_norm": 0.009938361123204231, "learning_rate": 6.880471044830659e-06, "loss": 0.0012, "step": 107450 }, { "epoch": 1.7583244702609835, "grad_norm": 0.04905107244849205, "learning_rate": 6.879809490166223e-06, "loss": 0.002, "step": 107460 }, { "epoch": 1.7584880962120593, "grad_norm": 0.05351666361093521, "learning_rate": 6.87914789717407e-06, "loss": 0.0008, "step": 107470 }, { "epoch": 1.758651722163135, "grad_norm": 0.08999839425086975, "learning_rate": 6.8784862658676895e-06, "loss": 0.0033, "step": 107480 }, { "epoch": 1.758815348114211, "grad_norm": 0.0060929059982299805, "learning_rate": 6.877824596260571e-06, "loss": 0.001, "step": 107490 }, { "epoch": 1.7589789740652868, "grad_norm": 0.07083297520875931, "learning_rate": 6.877162888366208e-06, "loss": 0.0023, "step": 107500 }, { "epoch": 1.7591426000163626, "grad_norm": 0.05055364593863487, "learning_rate": 6.876501142198089e-06, "loss": 0.001, "step": 107510 }, { "epoch": 1.7593062259674386, "grad_norm": 0.03735458478331566, "learning_rate": 6.875839357769707e-06, "loss": 0.0015, "step": 107520 }, { "epoch": 1.7594698519185141, "grad_norm": 0.0031108304392546415, "learning_rate": 6.875177535094557e-06, "loss": 0.0012, "step": 107530 }, { "epoch": 1.7596334778695901, "grad_norm": 0.07065120339393616, "learning_rate": 6.874515674186131e-06, "loss": 0.0012, "step": 107540 }, { "epoch": 1.7597971038206661, "grad_norm": 0.03702051565051079, "learning_rate": 6.873853775057926e-06, "loss": 0.0021, "step": 107550 }, { "epoch": 1.7599607297717417, "grad_norm": 0.0793071761727333, "learning_rate": 6.873191837723434e-06, "loss": 0.0017, "step": 107560 }, { "epoch": 1.7601243557228177, "grad_norm": 0.014803657308220863, "learning_rate": 6.872529862196156e-06, "loss": 0.0012, "step": 107570 }, { "epoch": 1.7602879816738934, "grad_norm": 0.00657362537458539, "learning_rate": 6.871867848489584e-06, "loss": 0.0016, "step": 107580 }, { "epoch": 1.7604516076249692, "grad_norm": 0.046529438346624374, "learning_rate": 6.871205796617221e-06, "loss": 0.0011, "step": 107590 }, { "epoch": 1.7606152335760452, "grad_norm": 0.1618199497461319, "learning_rate": 6.870543706592561e-06, "loss": 0.0019, "step": 107600 }, { "epoch": 1.760778859527121, "grad_norm": 0.04886746034026146, "learning_rate": 6.869881578429108e-06, "loss": 0.0011, "step": 107610 }, { "epoch": 1.7609424854781968, "grad_norm": 0.19256331026554108, "learning_rate": 6.869219412140359e-06, "loss": 0.0014, "step": 107620 }, { "epoch": 1.7611061114292728, "grad_norm": 0.020153138786554337, "learning_rate": 6.868557207739815e-06, "loss": 0.0009, "step": 107630 }, { "epoch": 1.7612697373803485, "grad_norm": 0.233740895986557, "learning_rate": 6.8678949652409785e-06, "loss": 0.0015, "step": 107640 }, { "epoch": 1.7614333633314243, "grad_norm": 0.16433480381965637, "learning_rate": 6.867232684657353e-06, "loss": 0.0028, "step": 107650 }, { "epoch": 1.7615969892825003, "grad_norm": 0.060912635177373886, "learning_rate": 6.866570366002442e-06, "loss": 0.0019, "step": 107660 }, { "epoch": 1.761760615233576, "grad_norm": 0.12018071860074997, "learning_rate": 6.865908009289747e-06, "loss": 0.0018, "step": 107670 }, { "epoch": 1.7619242411846519, "grad_norm": 0.06023476645350456, "learning_rate": 6.8652456145327765e-06, "loss": 0.0018, "step": 107680 }, { "epoch": 1.7620878671357278, "grad_norm": 0.024518955498933792, "learning_rate": 6.864583181745032e-06, "loss": 0.0009, "step": 107690 }, { "epoch": 1.7622514930868036, "grad_norm": 0.025074796751141548, "learning_rate": 6.863920710940023e-06, "loss": 0.0013, "step": 107700 }, { "epoch": 1.7624151190378794, "grad_norm": 0.030015822499990463, "learning_rate": 6.863258202131255e-06, "loss": 0.0018, "step": 107710 }, { "epoch": 1.7625787449889554, "grad_norm": 0.10037550330162048, "learning_rate": 6.8625956553322395e-06, "loss": 0.0016, "step": 107720 }, { "epoch": 1.762742370940031, "grad_norm": 0.03630734235048294, "learning_rate": 6.861933070556481e-06, "loss": 0.0019, "step": 107730 }, { "epoch": 1.762905996891107, "grad_norm": 0.011409761384129524, "learning_rate": 6.861270447817492e-06, "loss": 0.0023, "step": 107740 }, { "epoch": 1.7630696228421827, "grad_norm": 0.02337935008108616, "learning_rate": 6.860607787128779e-06, "loss": 0.0012, "step": 107750 }, { "epoch": 1.7632332487932585, "grad_norm": 0.08101265877485275, "learning_rate": 6.859945088503858e-06, "loss": 0.0009, "step": 107760 }, { "epoch": 1.7633968747443345, "grad_norm": 0.08120698481798172, "learning_rate": 6.859282351956236e-06, "loss": 0.0013, "step": 107770 }, { "epoch": 1.7635605006954103, "grad_norm": 0.07661749422550201, "learning_rate": 6.858619577499431e-06, "loss": 0.001, "step": 107780 }, { "epoch": 1.763724126646486, "grad_norm": 0.044006992131471634, "learning_rate": 6.857956765146952e-06, "loss": 0.0015, "step": 107790 }, { "epoch": 1.763887752597562, "grad_norm": 0.03648419305682182, "learning_rate": 6.857293914912316e-06, "loss": 0.001, "step": 107800 }, { "epoch": 1.7640513785486378, "grad_norm": 0.04898051172494888, "learning_rate": 6.8566310268090344e-06, "loss": 0.0012, "step": 107810 }, { "epoch": 1.7642150044997136, "grad_norm": 0.022203141823410988, "learning_rate": 6.855968100850626e-06, "loss": 0.0017, "step": 107820 }, { "epoch": 1.7643786304507896, "grad_norm": 0.029791390523314476, "learning_rate": 6.855305137050607e-06, "loss": 0.0011, "step": 107830 }, { "epoch": 1.7645422564018653, "grad_norm": 0.010238309390842915, "learning_rate": 6.8546421354224946e-06, "loss": 0.004, "step": 107840 }, { "epoch": 1.7647058823529411, "grad_norm": 0.1265590786933899, "learning_rate": 6.8539790959798045e-06, "loss": 0.003, "step": 107850 }, { "epoch": 1.7648695083040171, "grad_norm": 0.05432509630918503, "learning_rate": 6.853316018736059e-06, "loss": 0.0036, "step": 107860 }, { "epoch": 1.765033134255093, "grad_norm": 0.01961560733616352, "learning_rate": 6.852652903704775e-06, "loss": 0.0007, "step": 107870 }, { "epoch": 1.7651967602061687, "grad_norm": 0.1225176453590393, "learning_rate": 6.8519897508994745e-06, "loss": 0.0014, "step": 107880 }, { "epoch": 1.7653603861572447, "grad_norm": 0.04219582676887512, "learning_rate": 6.851326560333679e-06, "loss": 0.0013, "step": 107890 }, { "epoch": 1.7655240121083202, "grad_norm": 0.07172752171754837, "learning_rate": 6.850663332020908e-06, "loss": 0.0015, "step": 107900 }, { "epoch": 1.7656876380593962, "grad_norm": 0.0904170349240303, "learning_rate": 6.850000065974686e-06, "loss": 0.0017, "step": 107910 }, { "epoch": 1.7658512640104722, "grad_norm": 0.06150229275226593, "learning_rate": 6.849336762208536e-06, "loss": 0.0026, "step": 107920 }, { "epoch": 1.7660148899615478, "grad_norm": 0.06567308306694031, "learning_rate": 6.848673420735983e-06, "loss": 0.0016, "step": 107930 }, { "epoch": 1.7661785159126238, "grad_norm": 0.06636751443147659, "learning_rate": 6.848010041570551e-06, "loss": 0.0018, "step": 107940 }, { "epoch": 1.7663421418636995, "grad_norm": 0.04910203069448471, "learning_rate": 6.847346624725766e-06, "loss": 0.0018, "step": 107950 }, { "epoch": 1.7665057678147753, "grad_norm": 0.03626154735684395, "learning_rate": 6.846683170215153e-06, "loss": 0.0014, "step": 107960 }, { "epoch": 1.7666693937658513, "grad_norm": 0.0036560113076120615, "learning_rate": 6.846019678052242e-06, "loss": 0.0047, "step": 107970 }, { "epoch": 1.766833019716927, "grad_norm": 0.1623310148715973, "learning_rate": 6.8453561482505594e-06, "loss": 0.0018, "step": 107980 }, { "epoch": 1.7669966456680029, "grad_norm": 0.02200945094227791, "learning_rate": 6.844692580823634e-06, "loss": 0.0013, "step": 107990 }, { "epoch": 1.7671602716190788, "grad_norm": 0.04357696697115898, "learning_rate": 6.844028975784995e-06, "loss": 0.0009, "step": 108000 }, { "epoch": 1.7673238975701546, "grad_norm": 0.08434666693210602, "learning_rate": 6.843365333148176e-06, "loss": 0.0027, "step": 108010 }, { "epoch": 1.7674875235212304, "grad_norm": 0.039948634803295135, "learning_rate": 6.842701652926703e-06, "loss": 0.0015, "step": 108020 }, { "epoch": 1.7676511494723064, "grad_norm": 0.07802166044712067, "learning_rate": 6.842037935134112e-06, "loss": 0.0017, "step": 108030 }, { "epoch": 1.7678147754233822, "grad_norm": 0.06458370387554169, "learning_rate": 6.841374179783934e-06, "loss": 0.0016, "step": 108040 }, { "epoch": 1.767978401374458, "grad_norm": 0.051985982805490494, "learning_rate": 6.840710386889702e-06, "loss": 0.0011, "step": 108050 }, { "epoch": 1.768142027325534, "grad_norm": 0.00892580021172762, "learning_rate": 6.840046556464951e-06, "loss": 0.0012, "step": 108060 }, { "epoch": 1.7683056532766097, "grad_norm": 0.1172146126627922, "learning_rate": 6.839382688523215e-06, "loss": 0.0015, "step": 108070 }, { "epoch": 1.7684692792276855, "grad_norm": 0.02359403856098652, "learning_rate": 6.838718783078031e-06, "loss": 0.0016, "step": 108080 }, { "epoch": 1.7686329051787615, "grad_norm": 0.03807912766933441, "learning_rate": 6.838054840142935e-06, "loss": 0.0006, "step": 108090 }, { "epoch": 1.768796531129837, "grad_norm": 0.057681698352098465, "learning_rate": 6.837390859731463e-06, "loss": 0.0011, "step": 108100 }, { "epoch": 1.768960157080913, "grad_norm": 0.08385086059570312, "learning_rate": 6.836726841857155e-06, "loss": 0.0016, "step": 108110 }, { "epoch": 1.769123783031989, "grad_norm": 0.16694743931293488, "learning_rate": 6.8360627865335486e-06, "loss": 0.0018, "step": 108120 }, { "epoch": 1.7692874089830646, "grad_norm": 0.027288351207971573, "learning_rate": 6.835398693774183e-06, "loss": 0.0014, "step": 108130 }, { "epoch": 1.7694510349341406, "grad_norm": 0.018415888771414757, "learning_rate": 6.834734563592599e-06, "loss": 0.0007, "step": 108140 }, { "epoch": 1.7696146608852164, "grad_norm": 0.15198524296283722, "learning_rate": 6.834070396002339e-06, "loss": 0.002, "step": 108150 }, { "epoch": 1.7697782868362921, "grad_norm": 0.05282733961939812, "learning_rate": 6.833406191016941e-06, "loss": 0.0017, "step": 108160 }, { "epoch": 1.7699419127873681, "grad_norm": 0.07961073517799377, "learning_rate": 6.832741948649952e-06, "loss": 0.0016, "step": 108170 }, { "epoch": 1.770105538738444, "grad_norm": 0.004411840811371803, "learning_rate": 6.8320776689149126e-06, "loss": 0.001, "step": 108180 }, { "epoch": 1.7702691646895197, "grad_norm": 0.12215033918619156, "learning_rate": 6.831413351825368e-06, "loss": 0.0011, "step": 108190 }, { "epoch": 1.7704327906405957, "grad_norm": 0.015391300432384014, "learning_rate": 6.8307489973948625e-06, "loss": 0.0011, "step": 108200 }, { "epoch": 1.7705964165916714, "grad_norm": 0.04485857114195824, "learning_rate": 6.830084605636942e-06, "loss": 0.0011, "step": 108210 }, { "epoch": 1.7707600425427472, "grad_norm": 0.008179917000234127, "learning_rate": 6.829420176565153e-06, "loss": 0.001, "step": 108220 }, { "epoch": 1.7709236684938232, "grad_norm": 0.17386071383953094, "learning_rate": 6.828755710193042e-06, "loss": 0.0022, "step": 108230 }, { "epoch": 1.771087294444899, "grad_norm": 0.07000856101512909, "learning_rate": 6.828091206534157e-06, "loss": 0.0019, "step": 108240 }, { "epoch": 1.7712509203959748, "grad_norm": 0.10510426759719849, "learning_rate": 6.827426665602048e-06, "loss": 0.0023, "step": 108250 }, { "epoch": 1.7714145463470508, "grad_norm": 0.022699343040585518, "learning_rate": 6.826762087410264e-06, "loss": 0.0011, "step": 108260 }, { "epoch": 1.7715781722981265, "grad_norm": 0.07545548677444458, "learning_rate": 6.826097471972353e-06, "loss": 0.0013, "step": 108270 }, { "epoch": 1.7717417982492023, "grad_norm": 0.014641174115240574, "learning_rate": 6.8254328193018695e-06, "loss": 0.001, "step": 108280 }, { "epoch": 1.7719054242002783, "grad_norm": 0.04669724032282829, "learning_rate": 6.824768129412361e-06, "loss": 0.0018, "step": 108290 }, { "epoch": 1.7720690501513539, "grad_norm": 0.05747746303677559, "learning_rate": 6.824103402317384e-06, "loss": 0.002, "step": 108300 }, { "epoch": 1.7722326761024298, "grad_norm": 0.19811737537384033, "learning_rate": 6.8234386380304885e-06, "loss": 0.0022, "step": 108310 }, { "epoch": 1.7723963020535058, "grad_norm": 0.014738243073225021, "learning_rate": 6.8227738365652315e-06, "loss": 0.0019, "step": 108320 }, { "epoch": 1.7725599280045814, "grad_norm": 0.03711644187569618, "learning_rate": 6.822108997935165e-06, "loss": 0.0013, "step": 108330 }, { "epoch": 1.7727235539556574, "grad_norm": 0.03668210655450821, "learning_rate": 6.821444122153846e-06, "loss": 0.0031, "step": 108340 }, { "epoch": 1.7728871799067332, "grad_norm": 0.1082221195101738, "learning_rate": 6.820779209234832e-06, "loss": 0.0012, "step": 108350 }, { "epoch": 1.773050805857809, "grad_norm": 0.07482797652482986, "learning_rate": 6.820114259191675e-06, "loss": 0.0015, "step": 108360 }, { "epoch": 1.773214431808885, "grad_norm": 0.0150266969576478, "learning_rate": 6.819449272037939e-06, "loss": 0.001, "step": 108370 }, { "epoch": 1.7733780577599607, "grad_norm": 0.0566699281334877, "learning_rate": 6.818784247787179e-06, "loss": 0.0021, "step": 108380 }, { "epoch": 1.7735416837110365, "grad_norm": 0.023646356537938118, "learning_rate": 6.818119186452955e-06, "loss": 0.0014, "step": 108390 }, { "epoch": 1.7737053096621125, "grad_norm": 0.27761638164520264, "learning_rate": 6.817454088048827e-06, "loss": 0.0015, "step": 108400 }, { "epoch": 1.7738689356131883, "grad_norm": 0.008251999504864216, "learning_rate": 6.816788952588356e-06, "loss": 0.003, "step": 108410 }, { "epoch": 1.774032561564264, "grad_norm": 0.039203789085149765, "learning_rate": 6.816123780085103e-06, "loss": 0.0014, "step": 108420 }, { "epoch": 1.77419618751534, "grad_norm": 0.039915286004543304, "learning_rate": 6.815458570552631e-06, "loss": 0.0014, "step": 108430 }, { "epoch": 1.7743598134664158, "grad_norm": 0.1213294044137001, "learning_rate": 6.814793324004503e-06, "loss": 0.0021, "step": 108440 }, { "epoch": 1.7745234394174916, "grad_norm": 0.0852871760725975, "learning_rate": 6.814128040454282e-06, "loss": 0.0011, "step": 108450 }, { "epoch": 1.7746870653685676, "grad_norm": 0.06563583761453629, "learning_rate": 6.813462719915533e-06, "loss": 0.0015, "step": 108460 }, { "epoch": 1.7748506913196433, "grad_norm": 0.02801262028515339, "learning_rate": 6.812797362401822e-06, "loss": 0.0016, "step": 108470 }, { "epoch": 1.7750143172707191, "grad_norm": 0.0034076874144375324, "learning_rate": 6.812131967926714e-06, "loss": 0.0012, "step": 108480 }, { "epoch": 1.7751779432217951, "grad_norm": 0.050476543605327606, "learning_rate": 6.811466536503777e-06, "loss": 0.0012, "step": 108490 }, { "epoch": 1.7753415691728707, "grad_norm": 0.026644153520464897, "learning_rate": 6.8108010681465775e-06, "loss": 0.0005, "step": 108500 }, { "epoch": 1.7755051951239467, "grad_norm": 0.06163833290338516, "learning_rate": 6.810135562868684e-06, "loss": 0.0038, "step": 108510 }, { "epoch": 1.7756688210750227, "grad_norm": 0.04968169704079628, "learning_rate": 6.809470020683666e-06, "loss": 0.0023, "step": 108520 }, { "epoch": 1.7758324470260982, "grad_norm": 0.0364549495279789, "learning_rate": 6.808804441605093e-06, "loss": 0.002, "step": 108530 }, { "epoch": 1.7759960729771742, "grad_norm": 0.1066678836941719, "learning_rate": 6.808138825646537e-06, "loss": 0.0015, "step": 108540 }, { "epoch": 1.77615969892825, "grad_norm": 0.0035201753489673138, "learning_rate": 6.807473172821566e-06, "loss": 0.0006, "step": 108550 }, { "epoch": 1.7763233248793258, "grad_norm": 0.10105164349079132, "learning_rate": 6.806807483143756e-06, "loss": 0.0015, "step": 108560 }, { "epoch": 1.7764869508304018, "grad_norm": 0.01844046451151371, "learning_rate": 6.8061417566266775e-06, "loss": 0.0011, "step": 108570 }, { "epoch": 1.7766505767814775, "grad_norm": 0.07854162901639938, "learning_rate": 6.805475993283904e-06, "loss": 0.0014, "step": 108580 }, { "epoch": 1.7768142027325533, "grad_norm": 0.015456932596862316, "learning_rate": 6.804810193129012e-06, "loss": 0.0018, "step": 108590 }, { "epoch": 1.7769778286836293, "grad_norm": 0.0450010746717453, "learning_rate": 6.804144356175574e-06, "loss": 0.0013, "step": 108600 }, { "epoch": 1.777141454634705, "grad_norm": 0.028939293697476387, "learning_rate": 6.803478482437167e-06, "loss": 0.0011, "step": 108610 }, { "epoch": 1.7773050805857808, "grad_norm": 0.18345321714878082, "learning_rate": 6.8028125719273675e-06, "loss": 0.0025, "step": 108620 }, { "epoch": 1.7774687065368568, "grad_norm": 0.05056355893611908, "learning_rate": 6.802146624659753e-06, "loss": 0.0011, "step": 108630 }, { "epoch": 1.7776323324879326, "grad_norm": 0.04446853697299957, "learning_rate": 6.8014806406479015e-06, "loss": 0.0009, "step": 108640 }, { "epoch": 1.7777959584390084, "grad_norm": 0.04898419231176376, "learning_rate": 6.800814619905393e-06, "loss": 0.0013, "step": 108650 }, { "epoch": 1.7779595843900844, "grad_norm": 0.033834319561719894, "learning_rate": 6.800148562445804e-06, "loss": 0.0009, "step": 108660 }, { "epoch": 1.77812321034116, "grad_norm": 0.05148136243224144, "learning_rate": 6.799482468282719e-06, "loss": 0.0011, "step": 108670 }, { "epoch": 1.778286836292236, "grad_norm": 0.06852688640356064, "learning_rate": 6.7988163374297145e-06, "loss": 0.0015, "step": 108680 }, { "epoch": 1.778450462243312, "grad_norm": 0.0334387868642807, "learning_rate": 6.798150169900376e-06, "loss": 0.0013, "step": 108690 }, { "epoch": 1.7786140881943875, "grad_norm": 0.14482003450393677, "learning_rate": 6.797483965708284e-06, "loss": 0.0011, "step": 108700 }, { "epoch": 1.7787777141454635, "grad_norm": 0.03656134009361267, "learning_rate": 6.796817724867023e-06, "loss": 0.0019, "step": 108710 }, { "epoch": 1.7789413400965393, "grad_norm": 0.035631489008665085, "learning_rate": 6.796151447390177e-06, "loss": 0.0017, "step": 108720 }, { "epoch": 1.779104966047615, "grad_norm": 0.030485883355140686, "learning_rate": 6.795485133291331e-06, "loss": 0.0016, "step": 108730 }, { "epoch": 1.779268591998691, "grad_norm": 0.040446992963552475, "learning_rate": 6.794818782584069e-06, "loss": 0.0014, "step": 108740 }, { "epoch": 1.7794322179497668, "grad_norm": 0.055587656795978546, "learning_rate": 6.7941523952819784e-06, "loss": 0.0015, "step": 108750 }, { "epoch": 1.7795958439008426, "grad_norm": 0.010567354038357735, "learning_rate": 6.793485971398648e-06, "loss": 0.0038, "step": 108760 }, { "epoch": 1.7797594698519186, "grad_norm": 0.03585613891482353, "learning_rate": 6.792819510947663e-06, "loss": 0.0008, "step": 108770 }, { "epoch": 1.7799230958029943, "grad_norm": 0.007003211881965399, "learning_rate": 6.792153013942613e-06, "loss": 0.0015, "step": 108780 }, { "epoch": 1.7800867217540701, "grad_norm": 0.12091436237096786, "learning_rate": 6.791486480397087e-06, "loss": 0.0014, "step": 108790 }, { "epoch": 1.7802503477051461, "grad_norm": 0.03475462272763252, "learning_rate": 6.790819910324675e-06, "loss": 0.0012, "step": 108800 }, { "epoch": 1.780413973656222, "grad_norm": 0.04386192187666893, "learning_rate": 6.790153303738969e-06, "loss": 0.0019, "step": 108810 }, { "epoch": 1.7805775996072977, "grad_norm": 0.009293737821280956, "learning_rate": 6.789486660653558e-06, "loss": 0.0009, "step": 108820 }, { "epoch": 1.7807412255583737, "grad_norm": 0.023077497258782387, "learning_rate": 6.788819981082037e-06, "loss": 0.0012, "step": 108830 }, { "epoch": 1.7809048515094494, "grad_norm": 0.03814025968313217, "learning_rate": 6.788153265037998e-06, "loss": 0.0011, "step": 108840 }, { "epoch": 1.7810684774605252, "grad_norm": 0.0768059566617012, "learning_rate": 6.787486512535035e-06, "loss": 0.0021, "step": 108850 }, { "epoch": 1.7812321034116012, "grad_norm": 0.044272974133491516, "learning_rate": 6.78681972358674e-06, "loss": 0.0008, "step": 108860 }, { "epoch": 1.7813957293626768, "grad_norm": 0.019778313115239143, "learning_rate": 6.786152898206714e-06, "loss": 0.0014, "step": 108870 }, { "epoch": 1.7815593553137528, "grad_norm": 0.1856951266527176, "learning_rate": 6.785486036408546e-06, "loss": 0.0022, "step": 108880 }, { "epoch": 1.7817229812648288, "grad_norm": 0.09534719586372375, "learning_rate": 6.78481913820584e-06, "loss": 0.0016, "step": 108890 }, { "epoch": 1.7818866072159043, "grad_norm": 0.03269411250948906, "learning_rate": 6.784152203612188e-06, "loss": 0.0054, "step": 108900 }, { "epoch": 1.7820502331669803, "grad_norm": 0.2903829514980316, "learning_rate": 6.783485232641188e-06, "loss": 0.0015, "step": 108910 }, { "epoch": 1.782213859118056, "grad_norm": 0.08344890922307968, "learning_rate": 6.782818225306443e-06, "loss": 0.0016, "step": 108920 }, { "epoch": 1.7823774850691319, "grad_norm": 0.019733497872948647, "learning_rate": 6.78215118162155e-06, "loss": 0.0015, "step": 108930 }, { "epoch": 1.7825411110202078, "grad_norm": 0.03369942680001259, "learning_rate": 6.781484101600111e-06, "loss": 0.0016, "step": 108940 }, { "epoch": 1.7827047369712836, "grad_norm": 0.1534830927848816, "learning_rate": 6.780816985255725e-06, "loss": 0.0023, "step": 108950 }, { "epoch": 1.7828683629223594, "grad_norm": 0.06638512760400772, "learning_rate": 6.780149832601996e-06, "loss": 0.0014, "step": 108960 }, { "epoch": 1.7830319888734354, "grad_norm": 0.03568095341324806, "learning_rate": 6.779482643652524e-06, "loss": 0.0012, "step": 108970 }, { "epoch": 1.7831956148245112, "grad_norm": 0.036263205111026764, "learning_rate": 6.778815418420916e-06, "loss": 0.0012, "step": 108980 }, { "epoch": 1.783359240775587, "grad_norm": 0.1243572011590004, "learning_rate": 6.778148156920774e-06, "loss": 0.0019, "step": 108990 }, { "epoch": 1.783522866726663, "grad_norm": 0.04737140238285065, "learning_rate": 6.777480859165703e-06, "loss": 0.0019, "step": 109000 }, { "epoch": 1.7836864926777387, "grad_norm": 0.039544302970170975, "learning_rate": 6.77681352516931e-06, "loss": 0.0009, "step": 109010 }, { "epoch": 1.7838501186288145, "grad_norm": 0.09085015207529068, "learning_rate": 6.7761461549452e-06, "loss": 0.0052, "step": 109020 }, { "epoch": 1.7840137445798905, "grad_norm": 0.021648969501256943, "learning_rate": 6.775478748506981e-06, "loss": 0.001, "step": 109030 }, { "epoch": 1.7841773705309663, "grad_norm": 0.0927124097943306, "learning_rate": 6.774811305868259e-06, "loss": 0.0022, "step": 109040 }, { "epoch": 1.784340996482042, "grad_norm": 0.16868281364440918, "learning_rate": 6.7741438270426455e-06, "loss": 0.002, "step": 109050 }, { "epoch": 1.784504622433118, "grad_norm": 0.08816517144441605, "learning_rate": 6.773476312043748e-06, "loss": 0.0012, "step": 109060 }, { "epoch": 1.7846682483841936, "grad_norm": 0.03959045931696892, "learning_rate": 6.772808760885174e-06, "loss": 0.0007, "step": 109070 }, { "epoch": 1.7848318743352696, "grad_norm": 0.0806620717048645, "learning_rate": 6.7721411735805395e-06, "loss": 0.0016, "step": 109080 }, { "epoch": 1.7849955002863456, "grad_norm": 0.04272640496492386, "learning_rate": 6.771473550143454e-06, "loss": 0.0015, "step": 109090 }, { "epoch": 1.7851591262374211, "grad_norm": 0.10557691007852554, "learning_rate": 6.770805890587529e-06, "loss": 0.0012, "step": 109100 }, { "epoch": 1.7853227521884971, "grad_norm": 0.19772836565971375, "learning_rate": 6.770138194926377e-06, "loss": 0.0023, "step": 109110 }, { "epoch": 1.785486378139573, "grad_norm": 0.11781859397888184, "learning_rate": 6.769470463173613e-06, "loss": 0.0023, "step": 109120 }, { "epoch": 1.7856500040906487, "grad_norm": 0.03143687546253204, "learning_rate": 6.768802695342851e-06, "loss": 0.0007, "step": 109130 }, { "epoch": 1.7858136300417247, "grad_norm": 0.15556639432907104, "learning_rate": 6.768134891447708e-06, "loss": 0.0017, "step": 109140 }, { "epoch": 1.7859772559928004, "grad_norm": 0.06533191353082657, "learning_rate": 6.767467051501796e-06, "loss": 0.0013, "step": 109150 }, { "epoch": 1.7861408819438762, "grad_norm": 0.027433982118964195, "learning_rate": 6.766799175518736e-06, "loss": 0.0016, "step": 109160 }, { "epoch": 1.7863045078949522, "grad_norm": 0.09002053737640381, "learning_rate": 6.7661312635121414e-06, "loss": 0.0013, "step": 109170 }, { "epoch": 1.786468133846028, "grad_norm": 0.021716628223657608, "learning_rate": 6.7654633154956336e-06, "loss": 0.0015, "step": 109180 }, { "epoch": 1.7866317597971038, "grad_norm": 0.09023863077163696, "learning_rate": 6.764795331482829e-06, "loss": 0.002, "step": 109190 }, { "epoch": 1.7867953857481798, "grad_norm": 0.03968828171491623, "learning_rate": 6.76412731148735e-06, "loss": 0.0013, "step": 109200 }, { "epoch": 1.7869590116992555, "grad_norm": 0.017260627821087837, "learning_rate": 6.763459255522815e-06, "loss": 0.0007, "step": 109210 }, { "epoch": 1.7871226376503313, "grad_norm": 0.03197481483221054, "learning_rate": 6.762791163602846e-06, "loss": 0.0029, "step": 109220 }, { "epoch": 1.7872862636014073, "grad_norm": 0.043132100254297256, "learning_rate": 6.762123035741065e-06, "loss": 0.0014, "step": 109230 }, { "epoch": 1.787449889552483, "grad_norm": 0.006177300587296486, "learning_rate": 6.761454871951092e-06, "loss": 0.0018, "step": 109240 }, { "epoch": 1.7876135155035588, "grad_norm": 0.009340504184365273, "learning_rate": 6.760786672246554e-06, "loss": 0.0014, "step": 109250 }, { "epoch": 1.7877771414546348, "grad_norm": 0.08718208223581314, "learning_rate": 6.760118436641072e-06, "loss": 0.0008, "step": 109260 }, { "epoch": 1.7879407674057104, "grad_norm": 0.053703851997852325, "learning_rate": 6.759450165148272e-06, "loss": 0.003, "step": 109270 }, { "epoch": 1.7881043933567864, "grad_norm": 0.03303331509232521, "learning_rate": 6.75878185778178e-06, "loss": 0.0008, "step": 109280 }, { "epoch": 1.7882680193078624, "grad_norm": 0.09793052822351456, "learning_rate": 6.758113514555223e-06, "loss": 0.0027, "step": 109290 }, { "epoch": 1.788431645258938, "grad_norm": 0.0659894272685051, "learning_rate": 6.757445135482223e-06, "loss": 0.0012, "step": 109300 }, { "epoch": 1.788595271210014, "grad_norm": 0.00994095765054226, "learning_rate": 6.756776720576415e-06, "loss": 0.0015, "step": 109310 }, { "epoch": 1.7887588971610897, "grad_norm": 0.02149931527674198, "learning_rate": 6.756108269851421e-06, "loss": 0.0013, "step": 109320 }, { "epoch": 1.7889225231121655, "grad_norm": 0.02767838165163994, "learning_rate": 6.755439783320875e-06, "loss": 0.0008, "step": 109330 }, { "epoch": 1.7890861490632415, "grad_norm": 0.31709450483322144, "learning_rate": 6.754771260998404e-06, "loss": 0.0019, "step": 109340 }, { "epoch": 1.7892497750143173, "grad_norm": 0.08983433246612549, "learning_rate": 6.75410270289764e-06, "loss": 0.0012, "step": 109350 }, { "epoch": 1.789413400965393, "grad_norm": 0.002022230764850974, "learning_rate": 6.753434109032212e-06, "loss": 0.0014, "step": 109360 }, { "epoch": 1.789577026916469, "grad_norm": 0.2159547060728073, "learning_rate": 6.752765479415755e-06, "loss": 0.0015, "step": 109370 }, { "epoch": 1.7897406528675448, "grad_norm": 0.04438220337033272, "learning_rate": 6.7520968140619e-06, "loss": 0.0012, "step": 109380 }, { "epoch": 1.7899042788186206, "grad_norm": 0.022558433935046196, "learning_rate": 6.751428112984282e-06, "loss": 0.0013, "step": 109390 }, { "epoch": 1.7900679047696966, "grad_norm": 0.11707032471895218, "learning_rate": 6.750759376196534e-06, "loss": 0.002, "step": 109400 }, { "epoch": 1.7902315307207723, "grad_norm": 0.09876656532287598, "learning_rate": 6.750090603712292e-06, "loss": 0.002, "step": 109410 }, { "epoch": 1.7903951566718481, "grad_norm": 0.013116689398884773, "learning_rate": 6.749421795545188e-06, "loss": 0.0018, "step": 109420 }, { "epoch": 1.7905587826229241, "grad_norm": 0.0272822342813015, "learning_rate": 6.748752951708865e-06, "loss": 0.0021, "step": 109430 }, { "epoch": 1.790722408574, "grad_norm": 0.04172922298312187, "learning_rate": 6.7480840722169536e-06, "loss": 0.0012, "step": 109440 }, { "epoch": 1.7908860345250757, "grad_norm": 0.14452162384986877, "learning_rate": 6.747415157083097e-06, "loss": 0.0016, "step": 109450 }, { "epoch": 1.7910496604761517, "grad_norm": 0.008305506780743599, "learning_rate": 6.74674620632093e-06, "loss": 0.0016, "step": 109460 }, { "epoch": 1.7912132864272272, "grad_norm": 0.13852910697460175, "learning_rate": 6.746077219944094e-06, "loss": 0.0012, "step": 109470 }, { "epoch": 1.7913769123783032, "grad_norm": 0.10799295455217361, "learning_rate": 6.745408197966228e-06, "loss": 0.003, "step": 109480 }, { "epoch": 1.791540538329379, "grad_norm": 0.07366310060024261, "learning_rate": 6.7447391404009745e-06, "loss": 0.0008, "step": 109490 }, { "epoch": 1.7917041642804548, "grad_norm": 0.03593514487147331, "learning_rate": 6.744070047261973e-06, "loss": 0.0021, "step": 109500 }, { "epoch": 1.7918677902315308, "grad_norm": 0.07841930538415909, "learning_rate": 6.743400918562866e-06, "loss": 0.0015, "step": 109510 }, { "epoch": 1.7920314161826065, "grad_norm": 0.024425188079476357, "learning_rate": 6.742731754317297e-06, "loss": 0.0018, "step": 109520 }, { "epoch": 1.7921950421336823, "grad_norm": 0.1637423187494278, "learning_rate": 6.742062554538911e-06, "loss": 0.0018, "step": 109530 }, { "epoch": 1.7923586680847583, "grad_norm": 0.02022472396492958, "learning_rate": 6.74139331924135e-06, "loss": 0.0023, "step": 109540 }, { "epoch": 1.792522294035834, "grad_norm": 0.014985338784754276, "learning_rate": 6.7407240484382604e-06, "loss": 0.0022, "step": 109550 }, { "epoch": 1.7926859199869098, "grad_norm": 0.03726683184504509, "learning_rate": 6.740054742143288e-06, "loss": 0.0052, "step": 109560 }, { "epoch": 1.7928495459379858, "grad_norm": 0.02053188718855381, "learning_rate": 6.739385400370079e-06, "loss": 0.0009, "step": 109570 }, { "epoch": 1.7930131718890616, "grad_norm": 0.05671931058168411, "learning_rate": 6.738716023132281e-06, "loss": 0.0018, "step": 109580 }, { "epoch": 1.7931767978401374, "grad_norm": 0.019128523766994476, "learning_rate": 6.738046610443543e-06, "loss": 0.0008, "step": 109590 }, { "epoch": 1.7933404237912134, "grad_norm": 0.20108169317245483, "learning_rate": 6.737377162317511e-06, "loss": 0.0032, "step": 109600 }, { "epoch": 1.7935040497422892, "grad_norm": 0.030798498541116714, "learning_rate": 6.736707678767837e-06, "loss": 0.0021, "step": 109610 }, { "epoch": 1.793667675693365, "grad_norm": 0.0969134271144867, "learning_rate": 6.7360381598081715e-06, "loss": 0.0014, "step": 109620 }, { "epoch": 1.793831301644441, "grad_norm": 0.10191790759563446, "learning_rate": 6.7353686054521626e-06, "loss": 0.0015, "step": 109630 }, { "epoch": 1.7939949275955165, "grad_norm": 0.05821782350540161, "learning_rate": 6.7346990157134664e-06, "loss": 0.0012, "step": 109640 }, { "epoch": 1.7941585535465925, "grad_norm": 0.01302441954612732, "learning_rate": 6.73402939060573e-06, "loss": 0.0012, "step": 109650 }, { "epoch": 1.7943221794976685, "grad_norm": 0.010318618267774582, "learning_rate": 6.73335973014261e-06, "loss": 0.0043, "step": 109660 }, { "epoch": 1.794485805448744, "grad_norm": 0.017022930085659027, "learning_rate": 6.732690034337759e-06, "loss": 0.0019, "step": 109670 }, { "epoch": 1.79464943139982, "grad_norm": 0.031012339517474174, "learning_rate": 6.732020303204832e-06, "loss": 0.0013, "step": 109680 }, { "epoch": 1.7948130573508958, "grad_norm": 0.011705342680215836, "learning_rate": 6.7313505367574835e-06, "loss": 0.0013, "step": 109690 }, { "epoch": 1.7949766833019716, "grad_norm": 0.04397737234830856, "learning_rate": 6.730680735009371e-06, "loss": 0.0014, "step": 109700 }, { "epoch": 1.7951403092530476, "grad_norm": 0.14419756829738617, "learning_rate": 6.730010897974149e-06, "loss": 0.0017, "step": 109710 }, { "epoch": 1.7953039352041233, "grad_norm": 0.031112857162952423, "learning_rate": 6.729341025665477e-06, "loss": 0.0012, "step": 109720 }, { "epoch": 1.7954675611551991, "grad_norm": 0.02375543676316738, "learning_rate": 6.7286711180970125e-06, "loss": 0.0021, "step": 109730 }, { "epoch": 1.7956311871062751, "grad_norm": 0.041756898164749146, "learning_rate": 6.728001175282414e-06, "loss": 0.0011, "step": 109740 }, { "epoch": 1.795794813057351, "grad_norm": 0.02079571597278118, "learning_rate": 6.727331197235339e-06, "loss": 0.0011, "step": 109750 }, { "epoch": 1.7959584390084267, "grad_norm": 0.05897747352719307, "learning_rate": 6.726661183969453e-06, "loss": 0.0007, "step": 109760 }, { "epoch": 1.7961220649595027, "grad_norm": 0.054673194885253906, "learning_rate": 6.725991135498412e-06, "loss": 0.0011, "step": 109770 }, { "epoch": 1.7962856909105784, "grad_norm": 0.0017767091048881412, "learning_rate": 6.72532105183588e-06, "loss": 0.0012, "step": 109780 }, { "epoch": 1.7964493168616542, "grad_norm": 0.02608896605670452, "learning_rate": 6.72465093299552e-06, "loss": 0.001, "step": 109790 }, { "epoch": 1.7966129428127302, "grad_norm": 0.026921231299638748, "learning_rate": 6.723980778990994e-06, "loss": 0.0012, "step": 109800 }, { "epoch": 1.796776568763806, "grad_norm": 0.03969479724764824, "learning_rate": 6.7233105898359655e-06, "loss": 0.0013, "step": 109810 }, { "epoch": 1.7969401947148818, "grad_norm": 0.04735242947936058, "learning_rate": 6.7226403655441e-06, "loss": 0.0029, "step": 109820 }, { "epoch": 1.7971038206659578, "grad_norm": 0.19207336008548737, "learning_rate": 6.721970106129062e-06, "loss": 0.0041, "step": 109830 }, { "epoch": 1.7972674466170333, "grad_norm": 0.07175588607788086, "learning_rate": 6.721299811604518e-06, "loss": 0.0027, "step": 109840 }, { "epoch": 1.7974310725681093, "grad_norm": 0.049215372651815414, "learning_rate": 6.720629481984136e-06, "loss": 0.0016, "step": 109850 }, { "epoch": 1.7975946985191853, "grad_norm": 0.05026363208889961, "learning_rate": 6.7199591172815816e-06, "loss": 0.0011, "step": 109860 }, { "epoch": 1.7977583244702608, "grad_norm": 0.04765968397259712, "learning_rate": 6.7192887175105235e-06, "loss": 0.0008, "step": 109870 }, { "epoch": 1.7979219504213368, "grad_norm": 0.05155918374657631, "learning_rate": 6.71861828268463e-06, "loss": 0.0015, "step": 109880 }, { "epoch": 1.7980855763724126, "grad_norm": 0.026249753311276436, "learning_rate": 6.717947812817572e-06, "loss": 0.0008, "step": 109890 }, { "epoch": 1.7982492023234884, "grad_norm": 0.091555655002594, "learning_rate": 6.717277307923019e-06, "loss": 0.0031, "step": 109900 }, { "epoch": 1.7984128282745644, "grad_norm": 0.045142579823732376, "learning_rate": 6.716606768014643e-06, "loss": 0.0011, "step": 109910 }, { "epoch": 1.7985764542256402, "grad_norm": 0.043651603162288666, "learning_rate": 6.715936193106113e-06, "loss": 0.0014, "step": 109920 }, { "epoch": 1.798740080176716, "grad_norm": 0.09895902872085571, "learning_rate": 6.715265583211106e-06, "loss": 0.0013, "step": 109930 }, { "epoch": 1.798903706127792, "grad_norm": 0.09756592661142349, "learning_rate": 6.71459493834329e-06, "loss": 0.0016, "step": 109940 }, { "epoch": 1.7990673320788677, "grad_norm": 0.09922512620687485, "learning_rate": 6.713924258516342e-06, "loss": 0.0018, "step": 109950 }, { "epoch": 1.7992309580299435, "grad_norm": 0.03224080801010132, "learning_rate": 6.713253543743936e-06, "loss": 0.0018, "step": 109960 }, { "epoch": 1.7993945839810195, "grad_norm": 0.04234858229756355, "learning_rate": 6.7125827940397495e-06, "loss": 0.0015, "step": 109970 }, { "epoch": 1.7995582099320953, "grad_norm": 0.2340266853570938, "learning_rate": 6.711912009417454e-06, "loss": 0.0018, "step": 109980 }, { "epoch": 1.799721835883171, "grad_norm": 0.07366763055324554, "learning_rate": 6.711241189890729e-06, "loss": 0.0018, "step": 109990 }, { "epoch": 1.799885461834247, "grad_norm": 0.04627513885498047, "learning_rate": 6.710570335473252e-06, "loss": 0.0005, "step": 110000 }, { "epoch": 1.8000490877853228, "grad_norm": 0.045523811131715775, "learning_rate": 6.7098994461787e-06, "loss": 0.0011, "step": 110010 }, { "epoch": 1.8002127137363986, "grad_norm": 0.05166612192988396, "learning_rate": 6.709228522020752e-06, "loss": 0.0019, "step": 110020 }, { "epoch": 1.8003763396874746, "grad_norm": 0.07576767355203629, "learning_rate": 6.708557563013089e-06, "loss": 0.0007, "step": 110030 }, { "epoch": 1.8005399656385501, "grad_norm": 0.030799031257629395, "learning_rate": 6.707886569169391e-06, "loss": 0.0014, "step": 110040 }, { "epoch": 1.8007035915896261, "grad_norm": 0.049767788499593735, "learning_rate": 6.707215540503338e-06, "loss": 0.0029, "step": 110050 }, { "epoch": 1.8008672175407021, "grad_norm": 0.10202526301145554, "learning_rate": 6.7065444770286115e-06, "loss": 0.0032, "step": 110060 }, { "epoch": 1.8010308434917777, "grad_norm": 0.09180304408073425, "learning_rate": 6.705873378758895e-06, "loss": 0.0017, "step": 110070 }, { "epoch": 1.8011944694428537, "grad_norm": 0.03500739485025406, "learning_rate": 6.70520224570787e-06, "loss": 0.0007, "step": 110080 }, { "epoch": 1.8013580953939294, "grad_norm": 0.035036321729421616, "learning_rate": 6.704531077889223e-06, "loss": 0.0014, "step": 110090 }, { "epoch": 1.8015217213450052, "grad_norm": 0.0556439608335495, "learning_rate": 6.703859875316635e-06, "loss": 0.001, "step": 110100 }, { "epoch": 1.8016853472960812, "grad_norm": 0.06321204453706741, "learning_rate": 6.703188638003795e-06, "loss": 0.0016, "step": 110110 }, { "epoch": 1.801848973247157, "grad_norm": 0.0015897557605057955, "learning_rate": 6.702517365964386e-06, "loss": 0.0016, "step": 110120 }, { "epoch": 1.8020125991982328, "grad_norm": 0.06622970849275589, "learning_rate": 6.701846059212097e-06, "loss": 0.0014, "step": 110130 }, { "epoch": 1.8021762251493088, "grad_norm": 0.01980341225862503, "learning_rate": 6.7011747177606135e-06, "loss": 0.0013, "step": 110140 }, { "epoch": 1.8023398511003845, "grad_norm": 0.011586155742406845, "learning_rate": 6.700503341623625e-06, "loss": 0.0011, "step": 110150 }, { "epoch": 1.8025034770514603, "grad_norm": 0.07018989324569702, "learning_rate": 6.699831930814819e-06, "loss": 0.0072, "step": 110160 }, { "epoch": 1.8026671030025363, "grad_norm": 0.06599203497171402, "learning_rate": 6.699160485347886e-06, "loss": 0.0012, "step": 110170 }, { "epoch": 1.802830728953612, "grad_norm": 0.02355051413178444, "learning_rate": 6.698489005236515e-06, "loss": 0.0009, "step": 110180 }, { "epoch": 1.8029943549046878, "grad_norm": 0.03420333191752434, "learning_rate": 6.697817490494398e-06, "loss": 0.0013, "step": 110190 }, { "epoch": 1.8031579808557638, "grad_norm": 0.017708301544189453, "learning_rate": 6.697145941135227e-06, "loss": 0.0018, "step": 110200 }, { "epoch": 1.8033216068068396, "grad_norm": 0.023546122014522552, "learning_rate": 6.696474357172693e-06, "loss": 0.0015, "step": 110210 }, { "epoch": 1.8034852327579154, "grad_norm": 0.05180859565734863, "learning_rate": 6.69580273862049e-06, "loss": 0.0011, "step": 110220 }, { "epoch": 1.8036488587089914, "grad_norm": 0.05448821932077408, "learning_rate": 6.6951310854923104e-06, "loss": 0.0019, "step": 110230 }, { "epoch": 1.803812484660067, "grad_norm": 0.020031549036502838, "learning_rate": 6.694459397801851e-06, "loss": 0.0019, "step": 110240 }, { "epoch": 1.803976110611143, "grad_norm": 0.021664973348379135, "learning_rate": 6.693787675562804e-06, "loss": 0.0021, "step": 110250 }, { "epoch": 1.804139736562219, "grad_norm": 0.06148085743188858, "learning_rate": 6.693115918788869e-06, "loss": 0.0011, "step": 110260 }, { "epoch": 1.8043033625132945, "grad_norm": 0.014130481518805027, "learning_rate": 6.692444127493738e-06, "loss": 0.001, "step": 110270 }, { "epoch": 1.8044669884643705, "grad_norm": 0.33114156126976013, "learning_rate": 6.691772301691113e-06, "loss": 0.0014, "step": 110280 }, { "epoch": 1.8046306144154463, "grad_norm": 0.048854924738407135, "learning_rate": 6.6911004413946864e-06, "loss": 0.0023, "step": 110290 }, { "epoch": 1.804794240366522, "grad_norm": 0.021556904539465904, "learning_rate": 6.6904285466181625e-06, "loss": 0.0011, "step": 110300 }, { "epoch": 1.804957866317598, "grad_norm": 0.01004905253648758, "learning_rate": 6.6897566173752385e-06, "loss": 0.0023, "step": 110310 }, { "epoch": 1.8051214922686738, "grad_norm": 0.014158941805362701, "learning_rate": 6.689084653679613e-06, "loss": 0.0026, "step": 110320 }, { "epoch": 1.8052851182197496, "grad_norm": 0.03247445821762085, "learning_rate": 6.688412655544987e-06, "loss": 0.0013, "step": 110330 }, { "epoch": 1.8054487441708256, "grad_norm": 0.04133544862270355, "learning_rate": 6.687740622985065e-06, "loss": 0.001, "step": 110340 }, { "epoch": 1.8056123701219013, "grad_norm": 0.10780951380729675, "learning_rate": 6.6870685560135455e-06, "loss": 0.001, "step": 110350 }, { "epoch": 1.8057759960729771, "grad_norm": 0.06247051805257797, "learning_rate": 6.686396454644134e-06, "loss": 0.0007, "step": 110360 }, { "epoch": 1.8059396220240531, "grad_norm": 0.1589779555797577, "learning_rate": 6.685724318890532e-06, "loss": 0.0013, "step": 110370 }, { "epoch": 1.806103247975129, "grad_norm": 0.017074501141905785, "learning_rate": 6.6850521487664465e-06, "loss": 0.0012, "step": 110380 }, { "epoch": 1.8062668739262047, "grad_norm": 0.07145863771438599, "learning_rate": 6.684379944285578e-06, "loss": 0.0021, "step": 110390 }, { "epoch": 1.8064304998772807, "grad_norm": 0.046360500156879425, "learning_rate": 6.6837077054616376e-06, "loss": 0.0018, "step": 110400 }, { "epoch": 1.8065941258283562, "grad_norm": 0.03481677547097206, "learning_rate": 6.683035432308328e-06, "loss": 0.0012, "step": 110410 }, { "epoch": 1.8067577517794322, "grad_norm": 0.027222568169236183, "learning_rate": 6.682363124839357e-06, "loss": 0.0009, "step": 110420 }, { "epoch": 1.8069213777305082, "grad_norm": 0.06546434015035629, "learning_rate": 6.681690783068433e-06, "loss": 0.0015, "step": 110430 }, { "epoch": 1.8070850036815838, "grad_norm": 0.023905744776129723, "learning_rate": 6.681018407009265e-06, "loss": 0.002, "step": 110440 }, { "epoch": 1.8072486296326598, "grad_norm": 0.14430618286132812, "learning_rate": 6.680345996675559e-06, "loss": 0.0014, "step": 110450 }, { "epoch": 1.8074122555837355, "grad_norm": 0.23826836049556732, "learning_rate": 6.679673552081029e-06, "loss": 0.0018, "step": 110460 }, { "epoch": 1.8075758815348113, "grad_norm": 0.04643605649471283, "learning_rate": 6.679001073239384e-06, "loss": 0.0008, "step": 110470 }, { "epoch": 1.8077395074858873, "grad_norm": 0.023999907076358795, "learning_rate": 6.678328560164336e-06, "loss": 0.001, "step": 110480 }, { "epoch": 1.807903133436963, "grad_norm": 0.045412637293338776, "learning_rate": 6.677656012869593e-06, "loss": 0.0014, "step": 110490 }, { "epoch": 1.8080667593880388, "grad_norm": 0.02981303632259369, "learning_rate": 6.676983431368873e-06, "loss": 0.0016, "step": 110500 }, { "epoch": 1.8082303853391148, "grad_norm": 0.04052702337503433, "learning_rate": 6.6763108156758864e-06, "loss": 0.0016, "step": 110510 }, { "epoch": 1.8083940112901906, "grad_norm": 0.02121530845761299, "learning_rate": 6.675638165804348e-06, "loss": 0.0006, "step": 110520 }, { "epoch": 1.8085576372412664, "grad_norm": 0.07415112853050232, "learning_rate": 6.674965481767972e-06, "loss": 0.0013, "step": 110530 }, { "epoch": 1.8087212631923424, "grad_norm": 0.05468817055225372, "learning_rate": 6.674292763580474e-06, "loss": 0.0018, "step": 110540 }, { "epoch": 1.8088848891434182, "grad_norm": 0.027739156037569046, "learning_rate": 6.6736200112555725e-06, "loss": 0.0009, "step": 110550 }, { "epoch": 1.809048515094494, "grad_norm": 0.10857971012592316, "learning_rate": 6.6729472248069804e-06, "loss": 0.0014, "step": 110560 }, { "epoch": 1.80921214104557, "grad_norm": 0.10206150263547897, "learning_rate": 6.672274404248417e-06, "loss": 0.002, "step": 110570 }, { "epoch": 1.8093757669966457, "grad_norm": 0.05693456903100014, "learning_rate": 6.671601549593603e-06, "loss": 0.0013, "step": 110580 }, { "epoch": 1.8095393929477215, "grad_norm": 0.06237233057618141, "learning_rate": 6.670928660856253e-06, "loss": 0.0023, "step": 110590 }, { "epoch": 1.8097030188987975, "grad_norm": 0.06406567990779877, "learning_rate": 6.670255738050089e-06, "loss": 0.0011, "step": 110600 }, { "epoch": 1.809866644849873, "grad_norm": 0.007007698994129896, "learning_rate": 6.669582781188831e-06, "loss": 0.002, "step": 110610 }, { "epoch": 1.810030270800949, "grad_norm": 0.07085368037223816, "learning_rate": 6.6689097902862e-06, "loss": 0.0015, "step": 110620 }, { "epoch": 1.810193896752025, "grad_norm": 0.06430697441101074, "learning_rate": 6.668236765355918e-06, "loss": 0.0017, "step": 110630 }, { "epoch": 1.8103575227031006, "grad_norm": 0.026942212134599686, "learning_rate": 6.667563706411707e-06, "loss": 0.0012, "step": 110640 }, { "epoch": 1.8105211486541766, "grad_norm": 0.02646411769092083, "learning_rate": 6.666890613467291e-06, "loss": 0.0011, "step": 110650 }, { "epoch": 1.8106847746052523, "grad_norm": 0.0021650532726198435, "learning_rate": 6.666217486536393e-06, "loss": 0.0015, "step": 110660 }, { "epoch": 1.8108484005563281, "grad_norm": 0.022405359894037247, "learning_rate": 6.665544325632737e-06, "loss": 0.0025, "step": 110670 }, { "epoch": 1.8110120265074041, "grad_norm": 0.06825248897075653, "learning_rate": 6.664871130770047e-06, "loss": 0.0018, "step": 110680 }, { "epoch": 1.81117565245848, "grad_norm": 0.016141511499881744, "learning_rate": 6.664197901962053e-06, "loss": 0.0014, "step": 110690 }, { "epoch": 1.8113392784095557, "grad_norm": 0.032910920679569244, "learning_rate": 6.663524639222478e-06, "loss": 0.0016, "step": 110700 }, { "epoch": 1.8115029043606317, "grad_norm": 0.03725455701351166, "learning_rate": 6.662851342565051e-06, "loss": 0.0026, "step": 110710 }, { "epoch": 1.8116665303117074, "grad_norm": 0.08168468624353409, "learning_rate": 6.662178012003498e-06, "loss": 0.0013, "step": 110720 }, { "epoch": 1.8118301562627832, "grad_norm": 0.0738372728228569, "learning_rate": 6.661504647551551e-06, "loss": 0.0013, "step": 110730 }, { "epoch": 1.8119937822138592, "grad_norm": 0.05048363283276558, "learning_rate": 6.660831249222936e-06, "loss": 0.0022, "step": 110740 }, { "epoch": 1.812157408164935, "grad_norm": 0.047484930604696274, "learning_rate": 6.660157817031385e-06, "loss": 0.0029, "step": 110750 }, { "epoch": 1.8123210341160108, "grad_norm": 0.03896041214466095, "learning_rate": 6.659484350990627e-06, "loss": 0.0031, "step": 110760 }, { "epoch": 1.8124846600670868, "grad_norm": 0.03568293899297714, "learning_rate": 6.658810851114395e-06, "loss": 0.0009, "step": 110770 }, { "epoch": 1.8126482860181625, "grad_norm": 0.03466687723994255, "learning_rate": 6.65813731741642e-06, "loss": 0.0016, "step": 110780 }, { "epoch": 1.8128119119692383, "grad_norm": 0.04109685495495796, "learning_rate": 6.657463749910437e-06, "loss": 0.0008, "step": 110790 }, { "epoch": 1.8129755379203143, "grad_norm": 0.05295223742723465, "learning_rate": 6.656790148610176e-06, "loss": 0.001, "step": 110800 }, { "epoch": 1.8131391638713898, "grad_norm": 0.05265174061059952, "learning_rate": 6.656116513529373e-06, "loss": 0.0017, "step": 110810 }, { "epoch": 1.8133027898224658, "grad_norm": 0.05019636079668999, "learning_rate": 6.655442844681763e-06, "loss": 0.0017, "step": 110820 }, { "epoch": 1.8134664157735418, "grad_norm": 0.006164264399558306, "learning_rate": 6.654769142081083e-06, "loss": 0.001, "step": 110830 }, { "epoch": 1.8136300417246174, "grad_norm": 0.0035572745837271214, "learning_rate": 6.654095405741067e-06, "loss": 0.0004, "step": 110840 }, { "epoch": 1.8137936676756934, "grad_norm": 0.06521579623222351, "learning_rate": 6.65342163567545e-06, "loss": 0.0009, "step": 110850 }, { "epoch": 1.8139572936267692, "grad_norm": 0.05968748778104782, "learning_rate": 6.652747831897975e-06, "loss": 0.0014, "step": 110860 }, { "epoch": 1.814120919577845, "grad_norm": 0.04888618364930153, "learning_rate": 6.652073994422374e-06, "loss": 0.0025, "step": 110870 }, { "epoch": 1.814284545528921, "grad_norm": 0.0799802765250206, "learning_rate": 6.651400123262392e-06, "loss": 0.0014, "step": 110880 }, { "epoch": 1.8144481714799967, "grad_norm": 0.03404858335852623, "learning_rate": 6.6507262184317655e-06, "loss": 0.0013, "step": 110890 }, { "epoch": 1.8146117974310725, "grad_norm": 0.03494372218847275, "learning_rate": 6.6500522799442345e-06, "loss": 0.0006, "step": 110900 }, { "epoch": 1.8147754233821485, "grad_norm": 0.07554125785827637, "learning_rate": 6.649378307813541e-06, "loss": 0.0009, "step": 110910 }, { "epoch": 1.8149390493332243, "grad_norm": 0.031247273087501526, "learning_rate": 6.648704302053428e-06, "loss": 0.0018, "step": 110920 }, { "epoch": 1.8151026752843, "grad_norm": 0.06341877579689026, "learning_rate": 6.6480302626776336e-06, "loss": 0.0019, "step": 110930 }, { "epoch": 1.815266301235376, "grad_norm": 0.016034595668315887, "learning_rate": 6.647356189699907e-06, "loss": 0.0008, "step": 110940 }, { "epoch": 1.8154299271864518, "grad_norm": 0.06413327157497406, "learning_rate": 6.646682083133986e-06, "loss": 0.0015, "step": 110950 }, { "epoch": 1.8155935531375276, "grad_norm": 0.04577947407960892, "learning_rate": 6.646007942993619e-06, "loss": 0.0009, "step": 110960 }, { "epoch": 1.8157571790886036, "grad_norm": 0.017821716144680977, "learning_rate": 6.6453337692925505e-06, "loss": 0.0017, "step": 110970 }, { "epoch": 1.8159208050396793, "grad_norm": 0.06511768698692322, "learning_rate": 6.644659562044526e-06, "loss": 0.0013, "step": 110980 }, { "epoch": 1.8160844309907551, "grad_norm": 0.07797002047300339, "learning_rate": 6.64398532126329e-06, "loss": 0.0018, "step": 110990 }, { "epoch": 1.8162480569418311, "grad_norm": 0.04042145982384682, "learning_rate": 6.643311046962593e-06, "loss": 0.0011, "step": 111000 }, { "epoch": 1.8164116828929067, "grad_norm": 0.03662225604057312, "learning_rate": 6.6426367391561794e-06, "loss": 0.0014, "step": 111010 }, { "epoch": 1.8165753088439827, "grad_norm": 0.0035887828562408686, "learning_rate": 6.641962397857802e-06, "loss": 0.0012, "step": 111020 }, { "epoch": 1.8167389347950587, "grad_norm": 0.08401373028755188, "learning_rate": 6.641288023081206e-06, "loss": 0.001, "step": 111030 }, { "epoch": 1.8169025607461342, "grad_norm": 0.045983340591192245, "learning_rate": 6.640613614840143e-06, "loss": 0.0017, "step": 111040 }, { "epoch": 1.8170661866972102, "grad_norm": 0.039880864322185516, "learning_rate": 6.639939173148364e-06, "loss": 0.0013, "step": 111050 }, { "epoch": 1.817229812648286, "grad_norm": 0.08438082784414291, "learning_rate": 6.639264698019622e-06, "loss": 0.0011, "step": 111060 }, { "epoch": 1.8173934385993618, "grad_norm": 0.24404877424240112, "learning_rate": 6.638590189467664e-06, "loss": 0.0018, "step": 111070 }, { "epoch": 1.8175570645504378, "grad_norm": 0.029787536710500717, "learning_rate": 6.637915647506248e-06, "loss": 0.0009, "step": 111080 }, { "epoch": 1.8177206905015135, "grad_norm": 0.008498032577335835, "learning_rate": 6.637241072149123e-06, "loss": 0.0014, "step": 111090 }, { "epoch": 1.8178843164525893, "grad_norm": 0.020767666399478912, "learning_rate": 6.636566463410047e-06, "loss": 0.0007, "step": 111100 }, { "epoch": 1.8180479424036653, "grad_norm": 0.06655687093734741, "learning_rate": 6.635891821302771e-06, "loss": 0.0021, "step": 111110 }, { "epoch": 1.818211568354741, "grad_norm": 0.03837389126420021, "learning_rate": 6.635217145841053e-06, "loss": 0.0012, "step": 111120 }, { "epoch": 1.8183751943058168, "grad_norm": 0.022371722385287285, "learning_rate": 6.634542437038647e-06, "loss": 0.0008, "step": 111130 }, { "epoch": 1.8185388202568928, "grad_norm": 0.03782705217599869, "learning_rate": 6.633867694909313e-06, "loss": 0.0015, "step": 111140 }, { "epoch": 1.8187024462079686, "grad_norm": 0.03830079361796379, "learning_rate": 6.633192919466803e-06, "loss": 0.0008, "step": 111150 }, { "epoch": 1.8188660721590444, "grad_norm": 0.04849506914615631, "learning_rate": 6.632518110724882e-06, "loss": 0.001, "step": 111160 }, { "epoch": 1.8190296981101204, "grad_norm": 0.24909280240535736, "learning_rate": 6.631843268697303e-06, "loss": 0.0012, "step": 111170 }, { "epoch": 1.8191933240611962, "grad_norm": 0.03732369840145111, "learning_rate": 6.631168393397827e-06, "loss": 0.0032, "step": 111180 }, { "epoch": 1.819356950012272, "grad_norm": 0.06453453004360199, "learning_rate": 6.630493484840217e-06, "loss": 0.0021, "step": 111190 }, { "epoch": 1.819520575963348, "grad_norm": 0.03371567651629448, "learning_rate": 6.629818543038229e-06, "loss": 0.0019, "step": 111200 }, { "epoch": 1.8196842019144235, "grad_norm": 0.07707808911800385, "learning_rate": 6.629143568005628e-06, "loss": 0.0016, "step": 111210 }, { "epoch": 1.8198478278654995, "grad_norm": 0.15493327379226685, "learning_rate": 6.628468559756175e-06, "loss": 0.0018, "step": 111220 }, { "epoch": 1.8200114538165755, "grad_norm": 0.013982895761728287, "learning_rate": 6.627793518303633e-06, "loss": 0.0012, "step": 111230 }, { "epoch": 1.820175079767651, "grad_norm": 0.06393011659383774, "learning_rate": 6.627118443661766e-06, "loss": 0.001, "step": 111240 }, { "epoch": 1.820338705718727, "grad_norm": 0.026966454461216927, "learning_rate": 6.626443335844339e-06, "loss": 0.0011, "step": 111250 }, { "epoch": 1.8205023316698028, "grad_norm": 0.10260537266731262, "learning_rate": 6.6257681948651135e-06, "loss": 0.0037, "step": 111260 }, { "epoch": 1.8206659576208786, "grad_norm": 0.09834121912717819, "learning_rate": 6.625093020737858e-06, "loss": 0.002, "step": 111270 }, { "epoch": 1.8208295835719546, "grad_norm": 0.023893393576145172, "learning_rate": 6.624417813476338e-06, "loss": 0.0011, "step": 111280 }, { "epoch": 1.8209932095230303, "grad_norm": 0.024747053161263466, "learning_rate": 6.6237425730943205e-06, "loss": 0.0014, "step": 111290 }, { "epoch": 1.8211568354741061, "grad_norm": 0.05735292658209801, "learning_rate": 6.623067299605572e-06, "loss": 0.0012, "step": 111300 }, { "epoch": 1.8213204614251821, "grad_norm": 0.062871553003788, "learning_rate": 6.622391993023862e-06, "loss": 0.0018, "step": 111310 }, { "epoch": 1.8214840873762579, "grad_norm": 0.06321808695793152, "learning_rate": 6.621716653362959e-06, "loss": 0.0023, "step": 111320 }, { "epoch": 1.8216477133273337, "grad_norm": 0.12675543129444122, "learning_rate": 6.621041280636633e-06, "loss": 0.0016, "step": 111330 }, { "epoch": 1.8218113392784097, "grad_norm": 0.09044851362705231, "learning_rate": 6.620365874858654e-06, "loss": 0.0021, "step": 111340 }, { "epoch": 1.8219749652294854, "grad_norm": 0.02707103081047535, "learning_rate": 6.619690436042793e-06, "loss": 0.0021, "step": 111350 }, { "epoch": 1.8221385911805612, "grad_norm": 0.13540001213550568, "learning_rate": 6.6190149642028205e-06, "loss": 0.0013, "step": 111360 }, { "epoch": 1.8223022171316372, "grad_norm": 0.033624812960624695, "learning_rate": 6.6183394593525104e-06, "loss": 0.0007, "step": 111370 }, { "epoch": 1.8224658430827128, "grad_norm": 0.054779455065727234, "learning_rate": 6.617663921505635e-06, "loss": 0.0025, "step": 111380 }, { "epoch": 1.8226294690337888, "grad_norm": 0.07687123119831085, "learning_rate": 6.616988350675969e-06, "loss": 0.0014, "step": 111390 }, { "epoch": 1.8227930949848647, "grad_norm": 0.00875958614051342, "learning_rate": 6.616312746877285e-06, "loss": 0.0025, "step": 111400 }, { "epoch": 1.8229567209359403, "grad_norm": 0.05694053694605827, "learning_rate": 6.6156371101233585e-06, "loss": 0.0005, "step": 111410 }, { "epoch": 1.8231203468870163, "grad_norm": 0.08822426199913025, "learning_rate": 6.614961440427965e-06, "loss": 0.0016, "step": 111420 }, { "epoch": 1.823283972838092, "grad_norm": 0.05850430950522423, "learning_rate": 6.614285737804882e-06, "loss": 0.0015, "step": 111430 }, { "epoch": 1.8234475987891678, "grad_norm": 0.0299849733710289, "learning_rate": 6.613610002267885e-06, "loss": 0.0017, "step": 111440 }, { "epoch": 1.8236112247402438, "grad_norm": 0.01444945763796568, "learning_rate": 6.612934233830754e-06, "loss": 0.0015, "step": 111450 }, { "epoch": 1.8237748506913196, "grad_norm": 0.00446460722014308, "learning_rate": 6.612258432507264e-06, "loss": 0.002, "step": 111460 }, { "epoch": 1.8239384766423954, "grad_norm": 0.05672220140695572, "learning_rate": 6.611582598311196e-06, "loss": 0.0008, "step": 111470 }, { "epoch": 1.8241021025934714, "grad_norm": 0.06629370898008347, "learning_rate": 6.6109067312563304e-06, "loss": 0.0022, "step": 111480 }, { "epoch": 1.8242657285445472, "grad_norm": 0.028613055124878883, "learning_rate": 6.610230831356445e-06, "loss": 0.0013, "step": 111490 }, { "epoch": 1.824429354495623, "grad_norm": 0.01628170907497406, "learning_rate": 6.609554898625324e-06, "loss": 0.0008, "step": 111500 }, { "epoch": 1.824592980446699, "grad_norm": 0.025747640058398247, "learning_rate": 6.6088789330767455e-06, "loss": 0.0016, "step": 111510 }, { "epoch": 1.8247566063977747, "grad_norm": 0.017149539664387703, "learning_rate": 6.608202934724496e-06, "loss": 0.0012, "step": 111520 }, { "epoch": 1.8249202323488505, "grad_norm": 0.029621975496411324, "learning_rate": 6.607526903582355e-06, "loss": 0.0012, "step": 111530 }, { "epoch": 1.8250838582999265, "grad_norm": 0.1613551676273346, "learning_rate": 6.606850839664109e-06, "loss": 0.0012, "step": 111540 }, { "epoch": 1.8252474842510023, "grad_norm": 0.034094735980033875, "learning_rate": 6.606174742983539e-06, "loss": 0.0016, "step": 111550 }, { "epoch": 1.825411110202078, "grad_norm": 0.07294294983148575, "learning_rate": 6.605498613554433e-06, "loss": 0.0011, "step": 111560 }, { "epoch": 1.825574736153154, "grad_norm": 0.01338218990713358, "learning_rate": 6.6048224513905754e-06, "loss": 0.001, "step": 111570 }, { "epoch": 1.8257383621042296, "grad_norm": 0.03394569829106331, "learning_rate": 6.604146256505755e-06, "loss": 0.0014, "step": 111580 }, { "epoch": 1.8259019880553056, "grad_norm": 0.027149952948093414, "learning_rate": 6.603470028913753e-06, "loss": 0.0011, "step": 111590 }, { "epoch": 1.8260656140063816, "grad_norm": 0.030340779572725296, "learning_rate": 6.602793768628364e-06, "loss": 0.0018, "step": 111600 }, { "epoch": 1.8262292399574571, "grad_norm": 0.13563822209835052, "learning_rate": 6.602117475663371e-06, "loss": 0.0014, "step": 111610 }, { "epoch": 1.8263928659085331, "grad_norm": 0.03010241873562336, "learning_rate": 6.601441150032566e-06, "loss": 0.0009, "step": 111620 }, { "epoch": 1.8265564918596089, "grad_norm": 0.04394758120179176, "learning_rate": 6.600764791749739e-06, "loss": 0.0021, "step": 111630 }, { "epoch": 1.8267201178106847, "grad_norm": 0.07204029709100723, "learning_rate": 6.600088400828678e-06, "loss": 0.0014, "step": 111640 }, { "epoch": 1.8268837437617607, "grad_norm": 0.01734449900686741, "learning_rate": 6.5994119772831746e-06, "loss": 0.0014, "step": 111650 }, { "epoch": 1.8270473697128364, "grad_norm": 0.05255594849586487, "learning_rate": 6.598735521127023e-06, "loss": 0.0023, "step": 111660 }, { "epoch": 1.8272109956639122, "grad_norm": 0.10484502464532852, "learning_rate": 6.598059032374013e-06, "loss": 0.0019, "step": 111670 }, { "epoch": 1.8273746216149882, "grad_norm": 0.1207885667681694, "learning_rate": 6.597382511037939e-06, "loss": 0.003, "step": 111680 }, { "epoch": 1.827538247566064, "grad_norm": 0.08408387750387192, "learning_rate": 6.596705957132593e-06, "loss": 0.0012, "step": 111690 }, { "epoch": 1.8277018735171398, "grad_norm": 0.06593617796897888, "learning_rate": 6.596029370671771e-06, "loss": 0.0009, "step": 111700 }, { "epoch": 1.8278654994682157, "grad_norm": 0.08748337626457214, "learning_rate": 6.5953527516692685e-06, "loss": 0.0012, "step": 111710 }, { "epoch": 1.8280291254192915, "grad_norm": 0.07678984105587006, "learning_rate": 6.594676100138879e-06, "loss": 0.0013, "step": 111720 }, { "epoch": 1.8281927513703673, "grad_norm": 0.04381922632455826, "learning_rate": 6.5939994160944e-06, "loss": 0.0013, "step": 111730 }, { "epoch": 1.8283563773214433, "grad_norm": 0.08476757258176804, "learning_rate": 6.593322699549629e-06, "loss": 0.0015, "step": 111740 }, { "epoch": 1.828520003272519, "grad_norm": 0.041565097868442535, "learning_rate": 6.592645950518364e-06, "loss": 0.0013, "step": 111750 }, { "epoch": 1.8286836292235948, "grad_norm": 0.2274903506040573, "learning_rate": 6.591969169014401e-06, "loss": 0.0023, "step": 111760 }, { "epoch": 1.8288472551746708, "grad_norm": 0.13286250829696655, "learning_rate": 6.591292355051541e-06, "loss": 0.0016, "step": 111770 }, { "epoch": 1.8290108811257464, "grad_norm": 0.08695773780345917, "learning_rate": 6.590615508643584e-06, "loss": 0.0017, "step": 111780 }, { "epoch": 1.8291745070768224, "grad_norm": 0.04894595220685005, "learning_rate": 6.5899386298043286e-06, "loss": 0.0017, "step": 111790 }, { "epoch": 1.8293381330278984, "grad_norm": 0.04019671306014061, "learning_rate": 6.589261718547577e-06, "loss": 0.0011, "step": 111800 }, { "epoch": 1.829501758978974, "grad_norm": 0.016490750014781952, "learning_rate": 6.588584774887132e-06, "loss": 0.001, "step": 111810 }, { "epoch": 1.82966538493005, "grad_norm": 0.05742994323372841, "learning_rate": 6.587907798836792e-06, "loss": 0.0014, "step": 111820 }, { "epoch": 1.8298290108811257, "grad_norm": 0.03432959318161011, "learning_rate": 6.587230790410365e-06, "loss": 0.0012, "step": 111830 }, { "epoch": 1.8299926368322015, "grad_norm": 0.017482725903391838, "learning_rate": 6.586553749621651e-06, "loss": 0.0018, "step": 111840 }, { "epoch": 1.8301562627832775, "grad_norm": 0.06078844144940376, "learning_rate": 6.585876676484455e-06, "loss": 0.0011, "step": 111850 }, { "epoch": 1.8303198887343533, "grad_norm": 0.014617910608649254, "learning_rate": 6.585199571012581e-06, "loss": 0.0016, "step": 111860 }, { "epoch": 1.830483514685429, "grad_norm": 0.08484090119600296, "learning_rate": 6.584522433219839e-06, "loss": 0.0017, "step": 111870 }, { "epoch": 1.830647140636505, "grad_norm": 0.021130703389644623, "learning_rate": 6.5838452631200304e-06, "loss": 0.0011, "step": 111880 }, { "epoch": 1.8308107665875808, "grad_norm": 0.028431707993149757, "learning_rate": 6.583168060726964e-06, "loss": 0.0013, "step": 111890 }, { "epoch": 1.8309743925386566, "grad_norm": 0.007489405106753111, "learning_rate": 6.582490826054446e-06, "loss": 0.0018, "step": 111900 }, { "epoch": 1.8311380184897326, "grad_norm": 0.08836304396390915, "learning_rate": 6.581813559116288e-06, "loss": 0.0012, "step": 111910 }, { "epoch": 1.8313016444408083, "grad_norm": 0.04966489598155022, "learning_rate": 6.5811362599262965e-06, "loss": 0.0014, "step": 111920 }, { "epoch": 1.8314652703918841, "grad_norm": 0.026364373043179512, "learning_rate": 6.5804589284982816e-06, "loss": 0.0012, "step": 111930 }, { "epoch": 1.8316288963429601, "grad_norm": 0.039211153984069824, "learning_rate": 6.579781564846052e-06, "loss": 0.0011, "step": 111940 }, { "epoch": 1.8317925222940359, "grad_norm": 0.09929897636175156, "learning_rate": 6.57910416898342e-06, "loss": 0.0023, "step": 111950 }, { "epoch": 1.8319561482451117, "grad_norm": 0.0230252742767334, "learning_rate": 6.578426740924197e-06, "loss": 0.0015, "step": 111960 }, { "epoch": 1.8321197741961877, "grad_norm": 0.1596827208995819, "learning_rate": 6.577749280682197e-06, "loss": 0.0016, "step": 111970 }, { "epoch": 1.8322834001472632, "grad_norm": 0.09635336697101593, "learning_rate": 6.577071788271229e-06, "loss": 0.0022, "step": 111980 }, { "epoch": 1.8324470260983392, "grad_norm": 0.030889179557561874, "learning_rate": 6.5763942637051095e-06, "loss": 0.0014, "step": 111990 }, { "epoch": 1.8326106520494152, "grad_norm": 0.04218324273824692, "learning_rate": 6.5757167069976505e-06, "loss": 0.0012, "step": 112000 }, { "epoch": 1.8327742780004908, "grad_norm": 0.04818880558013916, "learning_rate": 6.57503911816267e-06, "loss": 0.0028, "step": 112010 }, { "epoch": 1.8329379039515667, "grad_norm": 0.028977947309613228, "learning_rate": 6.574361497213978e-06, "loss": 0.0013, "step": 112020 }, { "epoch": 1.8331015299026425, "grad_norm": 0.047259621322155, "learning_rate": 6.573683844165397e-06, "loss": 0.0019, "step": 112030 }, { "epoch": 1.8332651558537183, "grad_norm": 0.07000336050987244, "learning_rate": 6.573006159030739e-06, "loss": 0.0018, "step": 112040 }, { "epoch": 1.8334287818047943, "grad_norm": 0.03534364700317383, "learning_rate": 6.572328441823825e-06, "loss": 0.001, "step": 112050 }, { "epoch": 1.83359240775587, "grad_norm": 0.06893318146467209, "learning_rate": 6.571650692558469e-06, "loss": 0.0013, "step": 112060 }, { "epoch": 1.8337560337069458, "grad_norm": 0.02380603924393654, "learning_rate": 6.570972911248494e-06, "loss": 0.0018, "step": 112070 }, { "epoch": 1.8339196596580218, "grad_norm": 0.07879145443439484, "learning_rate": 6.570295097907718e-06, "loss": 0.0013, "step": 112080 }, { "epoch": 1.8340832856090976, "grad_norm": 0.026367761194705963, "learning_rate": 6.569617252549958e-06, "loss": 0.0015, "step": 112090 }, { "epoch": 1.8342469115601734, "grad_norm": 0.057152289897203445, "learning_rate": 6.568939375189038e-06, "loss": 0.0012, "step": 112100 }, { "epoch": 1.8344105375112494, "grad_norm": 0.06189114600419998, "learning_rate": 6.568261465838779e-06, "loss": 0.0012, "step": 112110 }, { "epoch": 1.8345741634623252, "grad_norm": 0.11264533549547195, "learning_rate": 6.5675835245130025e-06, "loss": 0.0029, "step": 112120 }, { "epoch": 1.834737789413401, "grad_norm": 0.06500635296106339, "learning_rate": 6.566905551225531e-06, "loss": 0.0017, "step": 112130 }, { "epoch": 1.834901415364477, "grad_norm": 0.03854741156101227, "learning_rate": 6.566227545990189e-06, "loss": 0.0011, "step": 112140 }, { "epoch": 1.8350650413155525, "grad_norm": 0.06894407421350479, "learning_rate": 6.5655495088207975e-06, "loss": 0.0014, "step": 112150 }, { "epoch": 1.8352286672666285, "grad_norm": 0.04053189232945442, "learning_rate": 6.564871439731184e-06, "loss": 0.0013, "step": 112160 }, { "epoch": 1.8353922932177045, "grad_norm": 0.04947401583194733, "learning_rate": 6.564193338735172e-06, "loss": 0.0014, "step": 112170 }, { "epoch": 1.83555591916878, "grad_norm": 0.05834561586380005, "learning_rate": 6.563515205846589e-06, "loss": 0.0014, "step": 112180 }, { "epoch": 1.835719545119856, "grad_norm": 0.0901034027338028, "learning_rate": 6.5628370410792595e-06, "loss": 0.0016, "step": 112190 }, { "epoch": 1.8358831710709318, "grad_norm": 0.05594683066010475, "learning_rate": 6.562158844447013e-06, "loss": 0.0007, "step": 112200 }, { "epoch": 1.8360467970220076, "grad_norm": 0.13189208507537842, "learning_rate": 6.561480615963676e-06, "loss": 0.0017, "step": 112210 }, { "epoch": 1.8362104229730836, "grad_norm": 0.0035627929028123617, "learning_rate": 6.560802355643077e-06, "loss": 0.0009, "step": 112220 }, { "epoch": 1.8363740489241593, "grad_norm": 0.022662121802568436, "learning_rate": 6.5601240634990445e-06, "loss": 0.0011, "step": 112230 }, { "epoch": 1.8365376748752351, "grad_norm": 0.03540605306625366, "learning_rate": 6.5594457395454115e-06, "loss": 0.0015, "step": 112240 }, { "epoch": 1.8367013008263111, "grad_norm": 0.04112672060728073, "learning_rate": 6.558767383796003e-06, "loss": 0.0011, "step": 112250 }, { "epoch": 1.8368649267773869, "grad_norm": 0.031173883005976677, "learning_rate": 6.558088996264656e-06, "loss": 0.0016, "step": 112260 }, { "epoch": 1.8370285527284627, "grad_norm": 0.07004079222679138, "learning_rate": 6.557410576965198e-06, "loss": 0.0009, "step": 112270 }, { "epoch": 1.8371921786795387, "grad_norm": 0.037445761263370514, "learning_rate": 6.556732125911463e-06, "loss": 0.0026, "step": 112280 }, { "epoch": 1.8373558046306144, "grad_norm": 0.04015830531716347, "learning_rate": 6.556053643117284e-06, "loss": 0.0015, "step": 112290 }, { "epoch": 1.8375194305816902, "grad_norm": 0.02721831202507019, "learning_rate": 6.555375128596495e-06, "loss": 0.0013, "step": 112300 }, { "epoch": 1.8376830565327662, "grad_norm": 0.287875771522522, "learning_rate": 6.55469658236293e-06, "loss": 0.0015, "step": 112310 }, { "epoch": 1.837846682483842, "grad_norm": 0.18613390624523163, "learning_rate": 6.554018004430424e-06, "loss": 0.0016, "step": 112320 }, { "epoch": 1.8380103084349178, "grad_norm": 0.12739026546478271, "learning_rate": 6.553339394812812e-06, "loss": 0.0012, "step": 112330 }, { "epoch": 1.8381739343859937, "grad_norm": 0.05665707588195801, "learning_rate": 6.552660753523931e-06, "loss": 0.0013, "step": 112340 }, { "epoch": 1.8383375603370693, "grad_norm": 0.029643118381500244, "learning_rate": 6.551982080577617e-06, "loss": 0.0008, "step": 112350 }, { "epoch": 1.8385011862881453, "grad_norm": 0.07010482251644135, "learning_rate": 6.55130337598771e-06, "loss": 0.0016, "step": 112360 }, { "epoch": 1.8386648122392213, "grad_norm": 0.02721446193754673, "learning_rate": 6.550624639768044e-06, "loss": 0.0008, "step": 112370 }, { "epoch": 1.8388284381902968, "grad_norm": 0.023786988109350204, "learning_rate": 6.549945871932463e-06, "loss": 0.0011, "step": 112380 }, { "epoch": 1.8389920641413728, "grad_norm": 0.033925849944353104, "learning_rate": 6.549267072494803e-06, "loss": 0.0008, "step": 112390 }, { "epoch": 1.8391556900924486, "grad_norm": 0.11234590411186218, "learning_rate": 6.548588241468904e-06, "loss": 0.0019, "step": 112400 }, { "epoch": 1.8393193160435244, "grad_norm": 0.03683744743466377, "learning_rate": 6.547909378868609e-06, "loss": 0.0012, "step": 112410 }, { "epoch": 1.8394829419946004, "grad_norm": 0.04414168372750282, "learning_rate": 6.547230484707758e-06, "loss": 0.0014, "step": 112420 }, { "epoch": 1.8396465679456762, "grad_norm": 0.0017156221438199282, "learning_rate": 6.546551559000192e-06, "loss": 0.0018, "step": 112430 }, { "epoch": 1.839810193896752, "grad_norm": 0.05216014385223389, "learning_rate": 6.545872601759756e-06, "loss": 0.0018, "step": 112440 }, { "epoch": 1.839973819847828, "grad_norm": 0.05301665514707565, "learning_rate": 6.545193613000293e-06, "loss": 0.0012, "step": 112450 }, { "epoch": 1.8401374457989037, "grad_norm": 0.01450265385210514, "learning_rate": 6.544514592735645e-06, "loss": 0.0013, "step": 112460 }, { "epoch": 1.8403010717499795, "grad_norm": 0.01280900277197361, "learning_rate": 6.543835540979659e-06, "loss": 0.0007, "step": 112470 }, { "epoch": 1.8404646977010555, "grad_norm": 0.10972993820905685, "learning_rate": 6.5431564577461795e-06, "loss": 0.0017, "step": 112480 }, { "epoch": 1.8406283236521312, "grad_norm": 0.014976118691265583, "learning_rate": 6.542477343049052e-06, "loss": 0.0012, "step": 112490 }, { "epoch": 1.840791949603207, "grad_norm": 0.05655437335371971, "learning_rate": 6.541798196902123e-06, "loss": 0.001, "step": 112500 }, { "epoch": 1.840955575554283, "grad_norm": 0.030603278428316116, "learning_rate": 6.5411190193192395e-06, "loss": 0.0008, "step": 112510 }, { "epoch": 1.8411192015053588, "grad_norm": 0.013317020609974861, "learning_rate": 6.5404398103142495e-06, "loss": 0.0017, "step": 112520 }, { "epoch": 1.8412828274564346, "grad_norm": 0.022098982706665993, "learning_rate": 6.539760569901003e-06, "loss": 0.0015, "step": 112530 }, { "epoch": 1.8414464534075106, "grad_norm": 0.058128632605075836, "learning_rate": 6.5390812980933485e-06, "loss": 0.0028, "step": 112540 }, { "epoch": 1.8416100793585861, "grad_norm": 0.005098348017781973, "learning_rate": 6.5384019949051345e-06, "loss": 0.0012, "step": 112550 }, { "epoch": 1.8417737053096621, "grad_norm": 0.048018019646406174, "learning_rate": 6.537722660350212e-06, "loss": 0.0011, "step": 112560 }, { "epoch": 1.841937331260738, "grad_norm": 0.09628043323755264, "learning_rate": 6.537043294442433e-06, "loss": 0.001, "step": 112570 }, { "epoch": 1.8421009572118137, "grad_norm": 0.03543010726571083, "learning_rate": 6.536363897195648e-06, "loss": 0.0007, "step": 112580 }, { "epoch": 1.8422645831628897, "grad_norm": 0.0245963241904974, "learning_rate": 6.53568446862371e-06, "loss": 0.0024, "step": 112590 }, { "epoch": 1.8424282091139654, "grad_norm": 0.012904289178550243, "learning_rate": 6.535005008740472e-06, "loss": 0.0016, "step": 112600 }, { "epoch": 1.8425918350650412, "grad_norm": 0.08937376737594604, "learning_rate": 6.534325517559787e-06, "loss": 0.0024, "step": 112610 }, { "epoch": 1.8427554610161172, "grad_norm": 0.06743168085813522, "learning_rate": 6.533645995095508e-06, "loss": 0.0018, "step": 112620 }, { "epoch": 1.842919086967193, "grad_norm": 0.03750006854534149, "learning_rate": 6.532966441361494e-06, "loss": 0.0014, "step": 112630 }, { "epoch": 1.8430827129182688, "grad_norm": 0.08730506151914597, "learning_rate": 6.532286856371596e-06, "loss": 0.0008, "step": 112640 }, { "epoch": 1.8432463388693447, "grad_norm": 0.03965878114104271, "learning_rate": 6.5316072401396735e-06, "loss": 0.0017, "step": 112650 }, { "epoch": 1.8434099648204205, "grad_norm": 0.01091597881168127, "learning_rate": 6.530927592679581e-06, "loss": 0.0018, "step": 112660 }, { "epoch": 1.8435735907714963, "grad_norm": 0.06569281965494156, "learning_rate": 6.530247914005178e-06, "loss": 0.0012, "step": 112670 }, { "epoch": 1.8437372167225723, "grad_norm": 0.0828692615032196, "learning_rate": 6.52956820413032e-06, "loss": 0.0022, "step": 112680 }, { "epoch": 1.843900842673648, "grad_norm": 0.002205453347414732, "learning_rate": 6.5288884630688685e-06, "loss": 0.0017, "step": 112690 }, { "epoch": 1.8440644686247238, "grad_norm": 0.07079224288463593, "learning_rate": 6.528208690834681e-06, "loss": 0.0024, "step": 112700 }, { "epoch": 1.8442280945757998, "grad_norm": 0.029837429523468018, "learning_rate": 6.5275288874416185e-06, "loss": 0.0014, "step": 112710 }, { "epoch": 1.8443917205268756, "grad_norm": 0.040403906255960464, "learning_rate": 6.52684905290354e-06, "loss": 0.0006, "step": 112720 }, { "epoch": 1.8445553464779514, "grad_norm": 0.0075782788917422295, "learning_rate": 6.526169187234309e-06, "loss": 0.0013, "step": 112730 }, { "epoch": 1.8447189724290274, "grad_norm": 0.042963430285453796, "learning_rate": 6.525489290447785e-06, "loss": 0.0018, "step": 112740 }, { "epoch": 1.844882598380103, "grad_norm": 0.05081202834844589, "learning_rate": 6.5248093625578335e-06, "loss": 0.0009, "step": 112750 }, { "epoch": 1.845046224331179, "grad_norm": 0.06200045347213745, "learning_rate": 6.524129403578314e-06, "loss": 0.0012, "step": 112760 }, { "epoch": 1.845209850282255, "grad_norm": 0.029644181951880455, "learning_rate": 6.523449413523093e-06, "loss": 0.0016, "step": 112770 }, { "epoch": 1.8453734762333305, "grad_norm": 0.047032106667757034, "learning_rate": 6.522769392406035e-06, "loss": 0.0006, "step": 112780 }, { "epoch": 1.8455371021844065, "grad_norm": 0.19231395423412323, "learning_rate": 6.5220893402410025e-06, "loss": 0.0014, "step": 112790 }, { "epoch": 1.8457007281354822, "grad_norm": 0.07914836704730988, "learning_rate": 6.521409257041864e-06, "loss": 0.0025, "step": 112800 }, { "epoch": 1.845864354086558, "grad_norm": 0.15845991671085358, "learning_rate": 6.5207291428224835e-06, "loss": 0.001, "step": 112810 }, { "epoch": 1.846027980037634, "grad_norm": 0.09808018058538437, "learning_rate": 6.520048997596732e-06, "loss": 0.0013, "step": 112820 }, { "epoch": 1.8461916059887098, "grad_norm": 0.03519142419099808, "learning_rate": 6.519368821378471e-06, "loss": 0.001, "step": 112830 }, { "epoch": 1.8463552319397856, "grad_norm": 0.08143886923789978, "learning_rate": 6.518688614181575e-06, "loss": 0.0018, "step": 112840 }, { "epoch": 1.8465188578908616, "grad_norm": 0.02558835782110691, "learning_rate": 6.518008376019906e-06, "loss": 0.0012, "step": 112850 }, { "epoch": 1.8466824838419373, "grad_norm": 0.03789612278342247, "learning_rate": 6.517328106907339e-06, "loss": 0.0006, "step": 112860 }, { "epoch": 1.8468461097930131, "grad_norm": 0.17904923856258392, "learning_rate": 6.516647806857742e-06, "loss": 0.0021, "step": 112870 }, { "epoch": 1.847009735744089, "grad_norm": 0.19414229691028595, "learning_rate": 6.5159674758849875e-06, "loss": 0.0013, "step": 112880 }, { "epoch": 1.8471733616951649, "grad_norm": 0.0717141330242157, "learning_rate": 6.515287114002944e-06, "loss": 0.0012, "step": 112890 }, { "epoch": 1.8473369876462407, "grad_norm": 0.0944838672876358, "learning_rate": 6.514606721225485e-06, "loss": 0.0015, "step": 112900 }, { "epoch": 1.8475006135973167, "grad_norm": 0.05483235791325569, "learning_rate": 6.513926297566483e-06, "loss": 0.0019, "step": 112910 }, { "epoch": 1.8476642395483924, "grad_norm": 0.021950552240014076, "learning_rate": 6.513245843039812e-06, "loss": 0.0014, "step": 112920 }, { "epoch": 1.8478278654994682, "grad_norm": 0.016694027930498123, "learning_rate": 6.512565357659345e-06, "loss": 0.0018, "step": 112930 }, { "epoch": 1.8479914914505442, "grad_norm": 0.05832232907414436, "learning_rate": 6.511884841438958e-06, "loss": 0.0013, "step": 112940 }, { "epoch": 1.8481551174016198, "grad_norm": 0.002658274257555604, "learning_rate": 6.511204294392523e-06, "loss": 0.0008, "step": 112950 }, { "epoch": 1.8483187433526957, "grad_norm": 0.03251804783940315, "learning_rate": 6.5105237165339186e-06, "loss": 0.0013, "step": 112960 }, { "epoch": 1.8484823693037717, "grad_norm": 0.03395070508122444, "learning_rate": 6.5098431078770184e-06, "loss": 0.0009, "step": 112970 }, { "epoch": 1.8486459952548473, "grad_norm": 0.11354569345712662, "learning_rate": 6.509162468435704e-06, "loss": 0.0018, "step": 112980 }, { "epoch": 1.8488096212059233, "grad_norm": 0.09856243431568146, "learning_rate": 6.5084817982238495e-06, "loss": 0.001, "step": 112990 }, { "epoch": 1.848973247156999, "grad_norm": 0.035839494317770004, "learning_rate": 6.507801097255334e-06, "loss": 0.001, "step": 113000 }, { "epoch": 1.8491368731080748, "grad_norm": 0.1803169697523117, "learning_rate": 6.507120365544035e-06, "loss": 0.0028, "step": 113010 }, { "epoch": 1.8493004990591508, "grad_norm": 0.031192122027277946, "learning_rate": 6.506439603103836e-06, "loss": 0.0012, "step": 113020 }, { "epoch": 1.8494641250102266, "grad_norm": 0.052567023783922195, "learning_rate": 6.505758809948612e-06, "loss": 0.001, "step": 113030 }, { "epoch": 1.8496277509613024, "grad_norm": 0.017661431804299355, "learning_rate": 6.50507798609225e-06, "loss": 0.0008, "step": 113040 }, { "epoch": 1.8497913769123784, "grad_norm": 0.034625038504600525, "learning_rate": 6.504397131548624e-06, "loss": 0.0016, "step": 113050 }, { "epoch": 1.8499550028634542, "grad_norm": 0.029816679656505585, "learning_rate": 6.503716246331623e-06, "loss": 0.0017, "step": 113060 }, { "epoch": 1.85011862881453, "grad_norm": 0.020382720977067947, "learning_rate": 6.503035330455124e-06, "loss": 0.0013, "step": 113070 }, { "epoch": 1.850282254765606, "grad_norm": 0.1003430038690567, "learning_rate": 6.502354383933015e-06, "loss": 0.0014, "step": 113080 }, { "epoch": 1.8504458807166817, "grad_norm": 0.05602677911520004, "learning_rate": 6.5016734067791765e-06, "loss": 0.0018, "step": 113090 }, { "epoch": 1.8506095066677575, "grad_norm": 0.0138311255723238, "learning_rate": 6.5009923990074956e-06, "loss": 0.0022, "step": 113100 }, { "epoch": 1.8507731326188335, "grad_norm": 0.07020757347345352, "learning_rate": 6.500311360631857e-06, "loss": 0.0018, "step": 113110 }, { "epoch": 1.850936758569909, "grad_norm": 0.19629938900470734, "learning_rate": 6.499630291666143e-06, "loss": 0.0019, "step": 113120 }, { "epoch": 1.851100384520985, "grad_norm": 0.02506330795586109, "learning_rate": 6.498949192124244e-06, "loss": 0.0012, "step": 113130 }, { "epoch": 1.851264010472061, "grad_norm": 0.009804431349039078, "learning_rate": 6.4982680620200455e-06, "loss": 0.0016, "step": 113140 }, { "epoch": 1.8514276364231366, "grad_norm": 0.02432212047278881, "learning_rate": 6.497586901367436e-06, "loss": 0.001, "step": 113150 }, { "epoch": 1.8515912623742126, "grad_norm": 0.034424543380737305, "learning_rate": 6.496905710180304e-06, "loss": 0.0006, "step": 113160 }, { "epoch": 1.8517548883252883, "grad_norm": 0.04392316937446594, "learning_rate": 6.496224488472536e-06, "loss": 0.0015, "step": 113170 }, { "epoch": 1.8519185142763641, "grad_norm": 0.03802569583058357, "learning_rate": 6.495543236258024e-06, "loss": 0.0024, "step": 113180 }, { "epoch": 1.85208214022744, "grad_norm": 0.05418698117136955, "learning_rate": 6.494861953550658e-06, "loss": 0.0007, "step": 113190 }, { "epoch": 1.8522457661785159, "grad_norm": 0.0895879790186882, "learning_rate": 6.494180640364326e-06, "loss": 0.0012, "step": 113200 }, { "epoch": 1.8524093921295917, "grad_norm": 0.024831155315041542, "learning_rate": 6.493499296712923e-06, "loss": 0.0014, "step": 113210 }, { "epoch": 1.8525730180806677, "grad_norm": 0.07503622770309448, "learning_rate": 6.492817922610339e-06, "loss": 0.0011, "step": 113220 }, { "epoch": 1.8527366440317434, "grad_norm": 0.19531793892383575, "learning_rate": 6.492136518070468e-06, "loss": 0.0022, "step": 113230 }, { "epoch": 1.8529002699828192, "grad_norm": 0.06261177361011505, "learning_rate": 6.491455083107201e-06, "loss": 0.0016, "step": 113240 }, { "epoch": 1.8530638959338952, "grad_norm": 0.04635102301836014, "learning_rate": 6.490773617734435e-06, "loss": 0.0018, "step": 113250 }, { "epoch": 1.853227521884971, "grad_norm": 0.02166881412267685, "learning_rate": 6.490092121966061e-06, "loss": 0.0013, "step": 113260 }, { "epoch": 1.8533911478360467, "grad_norm": 0.021767161786556244, "learning_rate": 6.489410595815977e-06, "loss": 0.0015, "step": 113270 }, { "epoch": 1.8535547737871227, "grad_norm": 0.10925807058811188, "learning_rate": 6.488729039298077e-06, "loss": 0.0017, "step": 113280 }, { "epoch": 1.8537183997381985, "grad_norm": 0.10010676831007004, "learning_rate": 6.488047452426259e-06, "loss": 0.0013, "step": 113290 }, { "epoch": 1.8538820256892743, "grad_norm": 0.18253697454929352, "learning_rate": 6.4873658352144185e-06, "loss": 0.0015, "step": 113300 }, { "epoch": 1.8540456516403503, "grad_norm": 0.27932310104370117, "learning_rate": 6.486684187676453e-06, "loss": 0.0015, "step": 113310 }, { "epoch": 1.8542092775914258, "grad_norm": 0.02496466413140297, "learning_rate": 6.486002509826261e-06, "loss": 0.0008, "step": 113320 }, { "epoch": 1.8543729035425018, "grad_norm": 0.028245534747838974, "learning_rate": 6.485320801677743e-06, "loss": 0.0021, "step": 113330 }, { "epoch": 1.8545365294935778, "grad_norm": 0.05314521864056587, "learning_rate": 6.484639063244797e-06, "loss": 0.0013, "step": 113340 }, { "epoch": 1.8547001554446534, "grad_norm": 0.045941222459077835, "learning_rate": 6.483957294541323e-06, "loss": 0.0014, "step": 113350 }, { "epoch": 1.8548637813957294, "grad_norm": 0.058222346007823944, "learning_rate": 6.4832754955812204e-06, "loss": 0.0011, "step": 113360 }, { "epoch": 1.8550274073468052, "grad_norm": 0.048324134200811386, "learning_rate": 6.482593666378394e-06, "loss": 0.001, "step": 113370 }, { "epoch": 1.855191033297881, "grad_norm": 0.0232540350407362, "learning_rate": 6.481911806946743e-06, "loss": 0.0007, "step": 113380 }, { "epoch": 1.855354659248957, "grad_norm": 0.007895734161138535, "learning_rate": 6.4812299173001706e-06, "loss": 0.0009, "step": 113390 }, { "epoch": 1.8555182852000327, "grad_norm": 0.03634113073348999, "learning_rate": 6.4805479974525786e-06, "loss": 0.0006, "step": 113400 }, { "epoch": 1.8556819111511085, "grad_norm": 0.04541604220867157, "learning_rate": 6.479866047417873e-06, "loss": 0.0007, "step": 113410 }, { "epoch": 1.8558455371021845, "grad_norm": 0.0177626870572567, "learning_rate": 6.479184067209958e-06, "loss": 0.0023, "step": 113420 }, { "epoch": 1.8560091630532602, "grad_norm": 0.014378293417394161, "learning_rate": 6.478502056842737e-06, "loss": 0.0012, "step": 113430 }, { "epoch": 1.856172789004336, "grad_norm": 0.07827039062976837, "learning_rate": 6.477820016330117e-06, "loss": 0.0014, "step": 113440 }, { "epoch": 1.856336414955412, "grad_norm": 0.04745528846979141, "learning_rate": 6.477137945686003e-06, "loss": 0.0015, "step": 113450 }, { "epoch": 1.8565000409064878, "grad_norm": 0.014323465526103973, "learning_rate": 6.476455844924303e-06, "loss": 0.0007, "step": 113460 }, { "epoch": 1.8566636668575636, "grad_norm": 0.07568442076444626, "learning_rate": 6.475773714058924e-06, "loss": 0.0014, "step": 113470 }, { "epoch": 1.8568272928086396, "grad_norm": 0.027373919263482094, "learning_rate": 6.475091553103774e-06, "loss": 0.0017, "step": 113480 }, { "epoch": 1.8569909187597153, "grad_norm": 0.054491929709911346, "learning_rate": 6.474409362072762e-06, "loss": 0.0019, "step": 113490 }, { "epoch": 1.857154544710791, "grad_norm": 0.0757937803864479, "learning_rate": 6.473727140979798e-06, "loss": 0.0014, "step": 113500 }, { "epoch": 1.857318170661867, "grad_norm": 0.05465291813015938, "learning_rate": 6.473044889838789e-06, "loss": 0.001, "step": 113510 }, { "epoch": 1.8574817966129427, "grad_norm": 0.012683062814176083, "learning_rate": 6.4723626086636486e-06, "loss": 0.0015, "step": 113520 }, { "epoch": 1.8576454225640187, "grad_norm": 0.20037733018398285, "learning_rate": 6.471680297468286e-06, "loss": 0.0011, "step": 113530 }, { "epoch": 1.8578090485150947, "grad_norm": 0.019184352830052376, "learning_rate": 6.470997956266614e-06, "loss": 0.0008, "step": 113540 }, { "epoch": 1.8579726744661702, "grad_norm": 0.0317043736577034, "learning_rate": 6.470315585072544e-06, "loss": 0.0015, "step": 113550 }, { "epoch": 1.8581363004172462, "grad_norm": 0.01851126179099083, "learning_rate": 6.469633183899992e-06, "loss": 0.0015, "step": 113560 }, { "epoch": 1.858299926368322, "grad_norm": 0.0308688897639513, "learning_rate": 6.468950752762867e-06, "loss": 0.0009, "step": 113570 }, { "epoch": 1.8584635523193977, "grad_norm": 0.0017401312943547964, "learning_rate": 6.468268291675086e-06, "loss": 0.0013, "step": 113580 }, { "epoch": 1.8586271782704737, "grad_norm": 0.14430272579193115, "learning_rate": 6.467585800650564e-06, "loss": 0.0008, "step": 113590 }, { "epoch": 1.8587908042215495, "grad_norm": 0.05990436673164368, "learning_rate": 6.466903279703215e-06, "loss": 0.0014, "step": 113600 }, { "epoch": 1.8589544301726253, "grad_norm": 0.059653811156749725, "learning_rate": 6.466220728846956e-06, "loss": 0.0016, "step": 113610 }, { "epoch": 1.8591180561237013, "grad_norm": 0.031213365495204926, "learning_rate": 6.465538148095704e-06, "loss": 0.0012, "step": 113620 }, { "epoch": 1.859281682074777, "grad_norm": 0.09215909987688065, "learning_rate": 6.464855537463373e-06, "loss": 0.0012, "step": 113630 }, { "epoch": 1.8594453080258528, "grad_norm": 0.02275950275361538, "learning_rate": 6.464172896963886e-06, "loss": 0.0011, "step": 113640 }, { "epoch": 1.8596089339769288, "grad_norm": 0.045355428010225296, "learning_rate": 6.463490226611158e-06, "loss": 0.0008, "step": 113650 }, { "epoch": 1.8597725599280046, "grad_norm": 0.09175878763198853, "learning_rate": 6.462807526419109e-06, "loss": 0.0018, "step": 113660 }, { "epoch": 1.8599361858790804, "grad_norm": 0.05660561844706535, "learning_rate": 6.462124796401659e-06, "loss": 0.0009, "step": 113670 }, { "epoch": 1.8600998118301564, "grad_norm": 0.03247513622045517, "learning_rate": 6.461442036572727e-06, "loss": 0.0012, "step": 113680 }, { "epoch": 1.8602634377812322, "grad_norm": 0.023840265348553658, "learning_rate": 6.4607592469462356e-06, "loss": 0.0013, "step": 113690 }, { "epoch": 1.860427063732308, "grad_norm": 0.2602851390838623, "learning_rate": 6.460076427536105e-06, "loss": 0.0012, "step": 113700 }, { "epoch": 1.860590689683384, "grad_norm": 0.031016068533062935, "learning_rate": 6.4593935783562575e-06, "loss": 0.0011, "step": 113710 }, { "epoch": 1.8607543156344595, "grad_norm": 0.01979067735373974, "learning_rate": 6.4587106994206176e-06, "loss": 0.002, "step": 113720 }, { "epoch": 1.8609179415855355, "grad_norm": 0.029249407351017, "learning_rate": 6.4580277907431054e-06, "loss": 0.0014, "step": 113730 }, { "epoch": 1.8610815675366115, "grad_norm": 0.026170741766691208, "learning_rate": 6.457344852337648e-06, "loss": 0.0017, "step": 113740 }, { "epoch": 1.861245193487687, "grad_norm": 0.05505536124110222, "learning_rate": 6.4566618842181675e-06, "loss": 0.0016, "step": 113750 }, { "epoch": 1.861408819438763, "grad_norm": 0.02145390212535858, "learning_rate": 6.45597888639859e-06, "loss": 0.0007, "step": 113760 }, { "epoch": 1.8615724453898388, "grad_norm": 0.035385485738515854, "learning_rate": 6.455295858892841e-06, "loss": 0.0015, "step": 113770 }, { "epoch": 1.8617360713409146, "grad_norm": 0.062343042343854904, "learning_rate": 6.454612801714848e-06, "loss": 0.001, "step": 113780 }, { "epoch": 1.8618996972919906, "grad_norm": 0.03812574967741966, "learning_rate": 6.453929714878538e-06, "loss": 0.002, "step": 113790 }, { "epoch": 1.8620633232430663, "grad_norm": 0.062270697206258774, "learning_rate": 6.4532465983978354e-06, "loss": 0.0032, "step": 113800 }, { "epoch": 1.8622269491941421, "grad_norm": 0.04403532296419144, "learning_rate": 6.452563452286672e-06, "loss": 0.0009, "step": 113810 }, { "epoch": 1.862390575145218, "grad_norm": 0.04493014141917229, "learning_rate": 6.451880276558974e-06, "loss": 0.001, "step": 113820 }, { "epoch": 1.8625542010962939, "grad_norm": 0.04386144503951073, "learning_rate": 6.451197071228673e-06, "loss": 0.0009, "step": 113830 }, { "epoch": 1.8627178270473697, "grad_norm": 0.14845293760299683, "learning_rate": 6.450513836309697e-06, "loss": 0.0018, "step": 113840 }, { "epoch": 1.8628814529984457, "grad_norm": 0.1742793470621109, "learning_rate": 6.449830571815978e-06, "loss": 0.0016, "step": 113850 }, { "epoch": 1.8630450789495214, "grad_norm": 0.05786088854074478, "learning_rate": 6.449147277761447e-06, "loss": 0.0007, "step": 113860 }, { "epoch": 1.8632087049005972, "grad_norm": 0.061199259012937546, "learning_rate": 6.4484639541600355e-06, "loss": 0.0012, "step": 113870 }, { "epoch": 1.8633723308516732, "grad_norm": 0.0040442668832838535, "learning_rate": 6.447780601025676e-06, "loss": 0.0009, "step": 113880 }, { "epoch": 1.8635359568027487, "grad_norm": 0.0703231617808342, "learning_rate": 6.447097218372303e-06, "loss": 0.001, "step": 113890 }, { "epoch": 1.8636995827538247, "grad_norm": 0.3046713173389435, "learning_rate": 6.446413806213845e-06, "loss": 0.0019, "step": 113900 }, { "epoch": 1.8638632087049007, "grad_norm": 0.0695275217294693, "learning_rate": 6.445730364564241e-06, "loss": 0.0011, "step": 113910 }, { "epoch": 1.8640268346559763, "grad_norm": 0.021251752972602844, "learning_rate": 6.445046893437423e-06, "loss": 0.0026, "step": 113920 }, { "epoch": 1.8641904606070523, "grad_norm": 0.07766327261924744, "learning_rate": 6.44436339284733e-06, "loss": 0.002, "step": 113930 }, { "epoch": 1.864354086558128, "grad_norm": 0.011497252620756626, "learning_rate": 6.443679862807895e-06, "loss": 0.0011, "step": 113940 }, { "epoch": 1.8645177125092038, "grad_norm": 0.024780910462141037, "learning_rate": 6.442996303333056e-06, "loss": 0.0017, "step": 113950 }, { "epoch": 1.8646813384602798, "grad_norm": 0.03349556028842926, "learning_rate": 6.442312714436748e-06, "loss": 0.0007, "step": 113960 }, { "epoch": 1.8648449644113556, "grad_norm": 0.036917101591825485, "learning_rate": 6.441629096132912e-06, "loss": 0.001, "step": 113970 }, { "epoch": 1.8650085903624314, "grad_norm": 0.06886055320501328, "learning_rate": 6.4409454484354835e-06, "loss": 0.0012, "step": 113980 }, { "epoch": 1.8651722163135074, "grad_norm": 0.0036520480643957853, "learning_rate": 6.4402617713584045e-06, "loss": 0.0014, "step": 113990 }, { "epoch": 1.8653358422645832, "grad_norm": 0.1163535863161087, "learning_rate": 6.4395780649156115e-06, "loss": 0.0014, "step": 114000 }, { "epoch": 1.865499468215659, "grad_norm": 0.044601041823625565, "learning_rate": 6.438894329121047e-06, "loss": 0.0022, "step": 114010 }, { "epoch": 1.865663094166735, "grad_norm": 0.0055495742708444595, "learning_rate": 6.438210563988649e-06, "loss": 0.0011, "step": 114020 }, { "epoch": 1.8658267201178107, "grad_norm": 0.018279720097780228, "learning_rate": 6.4375267695323625e-06, "loss": 0.0023, "step": 114030 }, { "epoch": 1.8659903460688865, "grad_norm": 0.03573724254965782, "learning_rate": 6.436842945766127e-06, "loss": 0.0009, "step": 114040 }, { "epoch": 1.8661539720199625, "grad_norm": 0.06943731755018234, "learning_rate": 6.436159092703887e-06, "loss": 0.0013, "step": 114050 }, { "epoch": 1.8663175979710382, "grad_norm": 0.0028849889058619738, "learning_rate": 6.435475210359583e-06, "loss": 0.0018, "step": 114060 }, { "epoch": 1.866481223922114, "grad_norm": 0.051848482340574265, "learning_rate": 6.434791298747162e-06, "loss": 0.0015, "step": 114070 }, { "epoch": 1.86664484987319, "grad_norm": 0.06573983281850815, "learning_rate": 6.434107357880565e-06, "loss": 0.0013, "step": 114080 }, { "epoch": 1.8668084758242656, "grad_norm": 0.04160178825259209, "learning_rate": 6.4334233877737395e-06, "loss": 0.0018, "step": 114090 }, { "epoch": 1.8669721017753416, "grad_norm": 0.019269103184342384, "learning_rate": 6.4327393884406295e-06, "loss": 0.0022, "step": 114100 }, { "epoch": 1.8671357277264176, "grad_norm": 0.027719944715499878, "learning_rate": 6.432055359895182e-06, "loss": 0.001, "step": 114110 }, { "epoch": 1.8672993536774931, "grad_norm": 0.04059600457549095, "learning_rate": 6.431371302151344e-06, "loss": 0.0009, "step": 114120 }, { "epoch": 1.867462979628569, "grad_norm": 0.11129605770111084, "learning_rate": 6.430687215223063e-06, "loss": 0.0015, "step": 114130 }, { "epoch": 1.8676266055796449, "grad_norm": 0.0047219195403158665, "learning_rate": 6.430003099124285e-06, "loss": 0.002, "step": 114140 }, { "epoch": 1.8677902315307207, "grad_norm": 0.045032914727926254, "learning_rate": 6.429318953868962e-06, "loss": 0.0014, "step": 114150 }, { "epoch": 1.8679538574817967, "grad_norm": 0.005488367285579443, "learning_rate": 6.428634779471039e-06, "loss": 0.001, "step": 114160 }, { "epoch": 1.8681174834328724, "grad_norm": 0.03728976473212242, "learning_rate": 6.42795057594447e-06, "loss": 0.0026, "step": 114170 }, { "epoch": 1.8682811093839482, "grad_norm": 0.03891729190945625, "learning_rate": 6.4272663433032e-06, "loss": 0.0009, "step": 114180 }, { "epoch": 1.8684447353350242, "grad_norm": 0.05545050650835037, "learning_rate": 6.426582081561186e-06, "loss": 0.0011, "step": 114190 }, { "epoch": 1.8686083612861, "grad_norm": 0.020968563854694366, "learning_rate": 6.4258977907323764e-06, "loss": 0.0014, "step": 114200 }, { "epoch": 1.8687719872371757, "grad_norm": 0.11776348203420639, "learning_rate": 6.425213470830722e-06, "loss": 0.0013, "step": 114210 }, { "epoch": 1.8689356131882517, "grad_norm": 0.042979542165994644, "learning_rate": 6.424529121870179e-06, "loss": 0.0021, "step": 114220 }, { "epoch": 1.8690992391393275, "grad_norm": 0.1599726378917694, "learning_rate": 6.4238447438646966e-06, "loss": 0.0014, "step": 114230 }, { "epoch": 1.8692628650904033, "grad_norm": 0.10072427242994308, "learning_rate": 6.423160336828232e-06, "loss": 0.0019, "step": 114240 }, { "epoch": 1.8694264910414793, "grad_norm": 0.03480834141373634, "learning_rate": 6.422475900774738e-06, "loss": 0.0011, "step": 114250 }, { "epoch": 1.869590116992555, "grad_norm": 0.04209292680025101, "learning_rate": 6.421791435718171e-06, "loss": 0.001, "step": 114260 }, { "epoch": 1.8697537429436308, "grad_norm": 0.1010698676109314, "learning_rate": 6.421106941672485e-06, "loss": 0.0019, "step": 114270 }, { "epoch": 1.8699173688947068, "grad_norm": 0.0021953927353024483, "learning_rate": 6.420422418651637e-06, "loss": 0.0006, "step": 114280 }, { "epoch": 1.8700809948457824, "grad_norm": 0.06425735354423523, "learning_rate": 6.419737866669584e-06, "loss": 0.0029, "step": 114290 }, { "epoch": 1.8702446207968584, "grad_norm": 0.004551838152110577, "learning_rate": 6.419053285740285e-06, "loss": 0.0013, "step": 114300 }, { "epoch": 1.8704082467479344, "grad_norm": 0.005389090161770582, "learning_rate": 6.418368675877694e-06, "loss": 0.0007, "step": 114310 }, { "epoch": 1.87057187269901, "grad_norm": 0.055788617581129074, "learning_rate": 6.417684037095774e-06, "loss": 0.0036, "step": 114320 }, { "epoch": 1.870735498650086, "grad_norm": 0.040131524205207825, "learning_rate": 6.416999369408479e-06, "loss": 0.0013, "step": 114330 }, { "epoch": 1.8708991246011617, "grad_norm": 0.038666196167469025, "learning_rate": 6.416314672829775e-06, "loss": 0.0013, "step": 114340 }, { "epoch": 1.8710627505522375, "grad_norm": 0.0432373508810997, "learning_rate": 6.415629947373618e-06, "loss": 0.0008, "step": 114350 }, { "epoch": 1.8712263765033135, "grad_norm": 0.04711610823869705, "learning_rate": 6.414945193053972e-06, "loss": 0.0008, "step": 114360 }, { "epoch": 1.8713900024543892, "grad_norm": 0.03250957280397415, "learning_rate": 6.414260409884795e-06, "loss": 0.0016, "step": 114370 }, { "epoch": 1.871553628405465, "grad_norm": 0.04380489140748978, "learning_rate": 6.413575597880052e-06, "loss": 0.0015, "step": 114380 }, { "epoch": 1.871717254356541, "grad_norm": 0.07480750232934952, "learning_rate": 6.412890757053705e-06, "loss": 0.0015, "step": 114390 }, { "epoch": 1.8718808803076168, "grad_norm": 0.035141073167324066, "learning_rate": 6.412205887419716e-06, "loss": 0.0012, "step": 114400 }, { "epoch": 1.8720445062586926, "grad_norm": 0.08925417810678482, "learning_rate": 6.411520988992051e-06, "loss": 0.0016, "step": 114410 }, { "epoch": 1.8722081322097686, "grad_norm": 0.15507082641124725, "learning_rate": 6.4108360617846735e-06, "loss": 0.0021, "step": 114420 }, { "epoch": 1.8723717581608443, "grad_norm": 0.10824127495288849, "learning_rate": 6.4101511058115465e-06, "loss": 0.0015, "step": 114430 }, { "epoch": 1.87253538411192, "grad_norm": 0.007497861515730619, "learning_rate": 6.40946612108664e-06, "loss": 0.0005, "step": 114440 }, { "epoch": 1.872699010062996, "grad_norm": 0.01620221510529518, "learning_rate": 6.408781107623917e-06, "loss": 0.004, "step": 114450 }, { "epoch": 1.8728626360140719, "grad_norm": 0.1838141232728958, "learning_rate": 6.408096065437346e-06, "loss": 0.0014, "step": 114460 }, { "epoch": 1.8730262619651477, "grad_norm": 0.021606771275401115, "learning_rate": 6.407410994540892e-06, "loss": 0.001, "step": 114470 }, { "epoch": 1.8731898879162237, "grad_norm": 0.0369165800511837, "learning_rate": 6.406725894948528e-06, "loss": 0.0027, "step": 114480 }, { "epoch": 1.8733535138672992, "grad_norm": 0.07535174489021301, "learning_rate": 6.406040766674216e-06, "loss": 0.0016, "step": 114490 }, { "epoch": 1.8735171398183752, "grad_norm": 0.04521791264414787, "learning_rate": 6.405355609731931e-06, "loss": 0.0063, "step": 114500 }, { "epoch": 1.8736807657694512, "grad_norm": 0.03453431278467178, "learning_rate": 6.40467042413564e-06, "loss": 0.0007, "step": 114510 }, { "epoch": 1.8738443917205267, "grad_norm": 0.03420819714665413, "learning_rate": 6.403985209899313e-06, "loss": 0.0011, "step": 114520 }, { "epoch": 1.8740080176716027, "grad_norm": 0.011232340708374977, "learning_rate": 6.403299967036922e-06, "loss": 0.0011, "step": 114530 }, { "epoch": 1.8741716436226785, "grad_norm": 0.1108151525259018, "learning_rate": 6.402614695562437e-06, "loss": 0.0029, "step": 114540 }, { "epoch": 1.8743352695737543, "grad_norm": 0.019919827580451965, "learning_rate": 6.401929395489833e-06, "loss": 0.0025, "step": 114550 }, { "epoch": 1.8744988955248303, "grad_norm": 0.04790327325463295, "learning_rate": 6.40124406683308e-06, "loss": 0.001, "step": 114560 }, { "epoch": 1.874662521475906, "grad_norm": 0.011226259171962738, "learning_rate": 6.400558709606153e-06, "loss": 0.0011, "step": 114570 }, { "epoch": 1.8748261474269818, "grad_norm": 0.012693059630692005, "learning_rate": 6.399873323823022e-06, "loss": 0.0014, "step": 114580 }, { "epoch": 1.8749897733780578, "grad_norm": 0.03990272060036659, "learning_rate": 6.399187909497668e-06, "loss": 0.0012, "step": 114590 }, { "epoch": 1.8751533993291336, "grad_norm": 0.04787773638963699, "learning_rate": 6.398502466644061e-06, "loss": 0.0015, "step": 114600 }, { "epoch": 1.8753170252802094, "grad_norm": 0.07030639797449112, "learning_rate": 6.397816995276179e-06, "loss": 0.0014, "step": 114610 }, { "epoch": 1.8754806512312854, "grad_norm": 0.02979014813899994, "learning_rate": 6.397131495407997e-06, "loss": 0.0013, "step": 114620 }, { "epoch": 1.8756442771823612, "grad_norm": 0.009502149187028408, "learning_rate": 6.396445967053492e-06, "loss": 0.0019, "step": 114630 }, { "epoch": 1.875807903133437, "grad_norm": 0.10112470388412476, "learning_rate": 6.395760410226641e-06, "loss": 0.0013, "step": 114640 }, { "epoch": 1.875971529084513, "grad_norm": 0.008663794957101345, "learning_rate": 6.395074824941422e-06, "loss": 0.0011, "step": 114650 }, { "epoch": 1.8761351550355887, "grad_norm": 0.020543767139315605, "learning_rate": 6.394389211211813e-06, "loss": 0.0007, "step": 114660 }, { "epoch": 1.8762987809866645, "grad_norm": 0.04779677838087082, "learning_rate": 6.393703569051796e-06, "loss": 0.0006, "step": 114670 }, { "epoch": 1.8764624069377405, "grad_norm": 0.04993719235062599, "learning_rate": 6.393017898475346e-06, "loss": 0.0015, "step": 114680 }, { "epoch": 1.876626032888816, "grad_norm": 0.03211997449398041, "learning_rate": 6.39233219949645e-06, "loss": 0.0013, "step": 114690 }, { "epoch": 1.876789658839892, "grad_norm": 0.14431969821453094, "learning_rate": 6.39164647212908e-06, "loss": 0.0018, "step": 114700 }, { "epoch": 1.876953284790968, "grad_norm": 0.05632403865456581, "learning_rate": 6.390960716387224e-06, "loss": 0.0015, "step": 114710 }, { "epoch": 1.8771169107420436, "grad_norm": 0.016072947531938553, "learning_rate": 6.390274932284861e-06, "loss": 0.001, "step": 114720 }, { "epoch": 1.8772805366931196, "grad_norm": 0.039805613458156586, "learning_rate": 6.389589119835977e-06, "loss": 0.0012, "step": 114730 }, { "epoch": 1.8774441626441953, "grad_norm": 0.038642317056655884, "learning_rate": 6.38890327905455e-06, "loss": 0.0012, "step": 114740 }, { "epoch": 1.877607788595271, "grad_norm": 0.022261913865804672, "learning_rate": 6.388217409954568e-06, "loss": 0.0012, "step": 114750 }, { "epoch": 1.877771414546347, "grad_norm": 0.022167926654219627, "learning_rate": 6.387531512550013e-06, "loss": 0.0008, "step": 114760 }, { "epoch": 1.8779350404974229, "grad_norm": 0.07045488059520721, "learning_rate": 6.386845586854871e-06, "loss": 0.0024, "step": 114770 }, { "epoch": 1.8780986664484987, "grad_norm": 0.040036700665950775, "learning_rate": 6.386159632883125e-06, "loss": 0.0013, "step": 114780 }, { "epoch": 1.8782622923995747, "grad_norm": 0.045302752405405045, "learning_rate": 6.3854736506487654e-06, "loss": 0.0013, "step": 114790 }, { "epoch": 1.8784259183506504, "grad_norm": 0.06820569187402725, "learning_rate": 6.384787640165775e-06, "loss": 0.0014, "step": 114800 }, { "epoch": 1.8785895443017262, "grad_norm": 0.1464105099439621, "learning_rate": 6.384101601448143e-06, "loss": 0.0017, "step": 114810 }, { "epoch": 1.8787531702528022, "grad_norm": 0.05786789208650589, "learning_rate": 6.383415534509856e-06, "loss": 0.0012, "step": 114820 }, { "epoch": 1.878916796203878, "grad_norm": 0.017705678939819336, "learning_rate": 6.382729439364903e-06, "loss": 0.0012, "step": 114830 }, { "epoch": 1.8790804221549537, "grad_norm": 0.009453458711504936, "learning_rate": 6.382043316027272e-06, "loss": 0.0016, "step": 114840 }, { "epoch": 1.8792440481060297, "grad_norm": 0.0494491383433342, "learning_rate": 6.381357164510955e-06, "loss": 0.0019, "step": 114850 }, { "epoch": 1.8794076740571053, "grad_norm": 0.037836477160453796, "learning_rate": 6.380670984829939e-06, "loss": 0.0014, "step": 114860 }, { "epoch": 1.8795713000081813, "grad_norm": 0.27165114879608154, "learning_rate": 6.379984776998216e-06, "loss": 0.0028, "step": 114870 }, { "epoch": 1.8797349259592573, "grad_norm": 0.021620867773890495, "learning_rate": 6.379298541029777e-06, "loss": 0.0016, "step": 114880 }, { "epoch": 1.8798985519103328, "grad_norm": 0.08273636549711227, "learning_rate": 6.378612276938614e-06, "loss": 0.0013, "step": 114890 }, { "epoch": 1.8800621778614088, "grad_norm": 0.03491542860865593, "learning_rate": 6.377925984738718e-06, "loss": 0.0013, "step": 114900 }, { "epoch": 1.8802258038124846, "grad_norm": 0.015166858211159706, "learning_rate": 6.377239664444084e-06, "loss": 0.0009, "step": 114910 }, { "epoch": 1.8803894297635604, "grad_norm": 0.05168034881353378, "learning_rate": 6.376553316068705e-06, "loss": 0.0007, "step": 114920 }, { "epoch": 1.8805530557146364, "grad_norm": 0.07806628942489624, "learning_rate": 6.375866939626574e-06, "loss": 0.001, "step": 114930 }, { "epoch": 1.8807166816657122, "grad_norm": 0.1282086968421936, "learning_rate": 6.375180535131686e-06, "loss": 0.0023, "step": 114940 }, { "epoch": 1.880880307616788, "grad_norm": 0.023322520777583122, "learning_rate": 6.374494102598036e-06, "loss": 0.001, "step": 114950 }, { "epoch": 1.881043933567864, "grad_norm": 0.02674379199743271, "learning_rate": 6.373807642039622e-06, "loss": 0.0006, "step": 114960 }, { "epoch": 1.8812075595189397, "grad_norm": 0.052580442279577255, "learning_rate": 6.373121153470437e-06, "loss": 0.0011, "step": 114970 }, { "epoch": 1.8813711854700155, "grad_norm": 0.038103632628917694, "learning_rate": 6.372434636904481e-06, "loss": 0.0014, "step": 114980 }, { "epoch": 1.8815348114210915, "grad_norm": 0.006355026736855507, "learning_rate": 6.371748092355749e-06, "loss": 0.0012, "step": 114990 }, { "epoch": 1.8816984373721672, "grad_norm": 0.06266943365335464, "learning_rate": 6.37106151983824e-06, "loss": 0.0011, "step": 115000 }, { "epoch": 1.881862063323243, "grad_norm": 0.14861205220222473, "learning_rate": 6.370374919365952e-06, "loss": 0.001, "step": 115010 }, { "epoch": 1.882025689274319, "grad_norm": 0.05520108342170715, "learning_rate": 6.3696882909528865e-06, "loss": 0.002, "step": 115020 }, { "epoch": 1.8821893152253948, "grad_norm": 0.03640350326895714, "learning_rate": 6.36900163461304e-06, "loss": 0.0014, "step": 115030 }, { "epoch": 1.8823529411764706, "grad_norm": 0.004920425359159708, "learning_rate": 6.368314950360416e-06, "loss": 0.0017, "step": 115040 }, { "epoch": 1.8825165671275466, "grad_norm": 0.013360130600631237, "learning_rate": 6.367628238209011e-06, "loss": 0.0009, "step": 115050 }, { "epoch": 1.882680193078622, "grad_norm": 0.05173852667212486, "learning_rate": 6.366941498172833e-06, "loss": 0.0018, "step": 115060 }, { "epoch": 1.882843819029698, "grad_norm": 0.09316015243530273, "learning_rate": 6.366254730265878e-06, "loss": 0.003, "step": 115070 }, { "epoch": 1.883007444980774, "grad_norm": 0.13563016057014465, "learning_rate": 6.365567934502153e-06, "loss": 0.0007, "step": 115080 }, { "epoch": 1.8831710709318497, "grad_norm": 0.03494580462574959, "learning_rate": 6.364881110895656e-06, "loss": 0.0008, "step": 115090 }, { "epoch": 1.8833346968829257, "grad_norm": 0.03251999244093895, "learning_rate": 6.364194259460397e-06, "loss": 0.0014, "step": 115100 }, { "epoch": 1.8834983228340014, "grad_norm": 0.08429354429244995, "learning_rate": 6.363507380210376e-06, "loss": 0.0022, "step": 115110 }, { "epoch": 1.8836619487850772, "grad_norm": 0.021545762196183205, "learning_rate": 6.3628204731596e-06, "loss": 0.0017, "step": 115120 }, { "epoch": 1.8838255747361532, "grad_norm": 0.010755563154816628, "learning_rate": 6.3621335383220715e-06, "loss": 0.0014, "step": 115130 }, { "epoch": 1.883989200687229, "grad_norm": 0.036439985036849976, "learning_rate": 6.361446575711801e-06, "loss": 0.0021, "step": 115140 }, { "epoch": 1.8841528266383047, "grad_norm": 0.05979278311133385, "learning_rate": 6.36075958534279e-06, "loss": 0.0012, "step": 115150 }, { "epoch": 1.8843164525893807, "grad_norm": 0.13326381146907806, "learning_rate": 6.360072567229052e-06, "loss": 0.0015, "step": 115160 }, { "epoch": 1.8844800785404565, "grad_norm": 0.0541546605527401, "learning_rate": 6.359385521384589e-06, "loss": 0.0015, "step": 115170 }, { "epoch": 1.8846437044915323, "grad_norm": 0.00768026290461421, "learning_rate": 6.358698447823411e-06, "loss": 0.0005, "step": 115180 }, { "epoch": 1.8848073304426083, "grad_norm": 0.044219013303518295, "learning_rate": 6.358011346559528e-06, "loss": 0.0008, "step": 115190 }, { "epoch": 1.884970956393684, "grad_norm": 0.044721122831106186, "learning_rate": 6.357324217606948e-06, "loss": 0.0013, "step": 115200 }, { "epoch": 1.8851345823447598, "grad_norm": 0.043520353734493256, "learning_rate": 6.356637060979682e-06, "loss": 0.0018, "step": 115210 }, { "epoch": 1.8852982082958358, "grad_norm": 0.04735809937119484, "learning_rate": 6.355949876691739e-06, "loss": 0.003, "step": 115220 }, { "epoch": 1.8854618342469116, "grad_norm": 0.35763972997665405, "learning_rate": 6.355262664757134e-06, "loss": 0.0018, "step": 115230 }, { "epoch": 1.8856254601979874, "grad_norm": 0.02772861160337925, "learning_rate": 6.354575425189873e-06, "loss": 0.0014, "step": 115240 }, { "epoch": 1.8857890861490634, "grad_norm": 0.1254786103963852, "learning_rate": 6.353888158003972e-06, "loss": 0.0019, "step": 115250 }, { "epoch": 1.885952712100139, "grad_norm": 0.3198104500770569, "learning_rate": 6.353200863213442e-06, "loss": 0.0017, "step": 115260 }, { "epoch": 1.886116338051215, "grad_norm": 0.032054755836725235, "learning_rate": 6.352513540832299e-06, "loss": 0.0018, "step": 115270 }, { "epoch": 1.886279964002291, "grad_norm": 0.03239808976650238, "learning_rate": 6.351826190874554e-06, "loss": 0.0014, "step": 115280 }, { "epoch": 1.8864435899533665, "grad_norm": 0.027296515181660652, "learning_rate": 6.3511388133542225e-06, "loss": 0.0024, "step": 115290 }, { "epoch": 1.8866072159044425, "grad_norm": 0.01131098996847868, "learning_rate": 6.350451408285321e-06, "loss": 0.0012, "step": 115300 }, { "epoch": 1.8867708418555182, "grad_norm": 0.04994446039199829, "learning_rate": 6.349763975681863e-06, "loss": 0.0008, "step": 115310 }, { "epoch": 1.886934467806594, "grad_norm": 0.04433007165789604, "learning_rate": 6.349076515557865e-06, "loss": 0.0013, "step": 115320 }, { "epoch": 1.88709809375767, "grad_norm": 0.017164725810289383, "learning_rate": 6.348389027927345e-06, "loss": 0.0014, "step": 115330 }, { "epoch": 1.8872617197087458, "grad_norm": 0.006907330360263586, "learning_rate": 6.3477015128043186e-06, "loss": 0.0008, "step": 115340 }, { "epoch": 1.8874253456598216, "grad_norm": 0.015295621007680893, "learning_rate": 6.347013970202805e-06, "loss": 0.0009, "step": 115350 }, { "epoch": 1.8875889716108976, "grad_norm": 0.004749813582748175, "learning_rate": 6.346326400136822e-06, "loss": 0.0008, "step": 115360 }, { "epoch": 1.8877525975619733, "grad_norm": 0.09317850321531296, "learning_rate": 6.34563880262039e-06, "loss": 0.0012, "step": 115370 }, { "epoch": 1.887916223513049, "grad_norm": 0.11671481281518936, "learning_rate": 6.344951177667525e-06, "loss": 0.0009, "step": 115380 }, { "epoch": 1.888079849464125, "grad_norm": 0.03567185997962952, "learning_rate": 6.344263525292251e-06, "loss": 0.0012, "step": 115390 }, { "epoch": 1.8882434754152009, "grad_norm": 0.1024574488401413, "learning_rate": 6.3435758455085875e-06, "loss": 0.002, "step": 115400 }, { "epoch": 1.8884071013662767, "grad_norm": 0.02872089110314846, "learning_rate": 6.3428881383305544e-06, "loss": 0.0016, "step": 115410 }, { "epoch": 1.8885707273173526, "grad_norm": 0.05649649724364281, "learning_rate": 6.342200403772173e-06, "loss": 0.001, "step": 115420 }, { "epoch": 1.8887343532684284, "grad_norm": 0.2812473177909851, "learning_rate": 6.341512641847468e-06, "loss": 0.0013, "step": 115430 }, { "epoch": 1.8888979792195042, "grad_norm": 0.2397480309009552, "learning_rate": 6.340824852570461e-06, "loss": 0.0018, "step": 115440 }, { "epoch": 1.8890616051705802, "grad_norm": 0.007679019123315811, "learning_rate": 6.3401370359551765e-06, "loss": 0.0008, "step": 115450 }, { "epoch": 1.8892252311216557, "grad_norm": 0.09153502434492111, "learning_rate": 6.339449192015636e-06, "loss": 0.0012, "step": 115460 }, { "epoch": 1.8893888570727317, "grad_norm": 0.05044927820563316, "learning_rate": 6.338761320765866e-06, "loss": 0.001, "step": 115470 }, { "epoch": 1.8895524830238077, "grad_norm": 0.0512239895761013, "learning_rate": 6.338073422219891e-06, "loss": 0.0018, "step": 115480 }, { "epoch": 1.8897161089748833, "grad_norm": 0.05369527265429497, "learning_rate": 6.337385496391738e-06, "loss": 0.0032, "step": 115490 }, { "epoch": 1.8898797349259593, "grad_norm": 0.03701943904161453, "learning_rate": 6.336697543295432e-06, "loss": 0.0011, "step": 115500 }, { "epoch": 1.890043360877035, "grad_norm": 0.00914072711020708, "learning_rate": 6.3360095629450005e-06, "loss": 0.0012, "step": 115510 }, { "epoch": 1.8902069868281108, "grad_norm": 0.026114571839571, "learning_rate": 6.3353215553544686e-06, "loss": 0.0011, "step": 115520 }, { "epoch": 1.8903706127791868, "grad_norm": 0.04018435627222061, "learning_rate": 6.334633520537866e-06, "loss": 0.001, "step": 115530 }, { "epoch": 1.8905342387302626, "grad_norm": 0.019849522039294243, "learning_rate": 6.333945458509222e-06, "loss": 0.0012, "step": 115540 }, { "epoch": 1.8906978646813384, "grad_norm": 0.043886635452508926, "learning_rate": 6.333257369282564e-06, "loss": 0.0014, "step": 115550 }, { "epoch": 1.8908614906324144, "grad_norm": 0.015712998807430267, "learning_rate": 6.332569252871923e-06, "loss": 0.0019, "step": 115560 }, { "epoch": 1.8910251165834902, "grad_norm": 0.01911265403032303, "learning_rate": 6.331881109291327e-06, "loss": 0.001, "step": 115570 }, { "epoch": 1.891188742534566, "grad_norm": 0.045935019850730896, "learning_rate": 6.331192938554809e-06, "loss": 0.0008, "step": 115580 }, { "epoch": 1.891352368485642, "grad_norm": 0.04630069062113762, "learning_rate": 6.330504740676398e-06, "loss": 0.0027, "step": 115590 }, { "epoch": 1.8915159944367177, "grad_norm": 0.063604436814785, "learning_rate": 6.329816515670127e-06, "loss": 0.0014, "step": 115600 }, { "epoch": 1.8916796203877935, "grad_norm": 0.016560887917876244, "learning_rate": 6.329128263550028e-06, "loss": 0.0011, "step": 115610 }, { "epoch": 1.8918432463388695, "grad_norm": 0.07203450798988342, "learning_rate": 6.328439984330136e-06, "loss": 0.0016, "step": 115620 }, { "epoch": 1.8920068722899452, "grad_norm": 0.06929733604192734, "learning_rate": 6.327751678024481e-06, "loss": 0.0014, "step": 115630 }, { "epoch": 1.892170498241021, "grad_norm": 0.06434408575296402, "learning_rate": 6.327063344647098e-06, "loss": 0.0022, "step": 115640 }, { "epoch": 1.892334124192097, "grad_norm": 0.042124129831790924, "learning_rate": 6.3263749842120225e-06, "loss": 0.0013, "step": 115650 }, { "epoch": 1.8924977501431726, "grad_norm": 0.06356802582740784, "learning_rate": 6.32568659673329e-06, "loss": 0.0016, "step": 115660 }, { "epoch": 1.8926613760942486, "grad_norm": 0.04707297310233116, "learning_rate": 6.324998182224934e-06, "loss": 0.0007, "step": 115670 }, { "epoch": 1.8928250020453243, "grad_norm": 0.04834985360503197, "learning_rate": 6.324309740700993e-06, "loss": 0.0017, "step": 115680 }, { "epoch": 1.8929886279964, "grad_norm": 0.06312521547079086, "learning_rate": 6.323621272175502e-06, "loss": 0.0012, "step": 115690 }, { "epoch": 1.893152253947476, "grad_norm": 0.14975771307945251, "learning_rate": 6.3229327766624996e-06, "loss": 0.0008, "step": 115700 }, { "epoch": 1.8933158798985519, "grad_norm": 0.0014825622783973813, "learning_rate": 6.322244254176022e-06, "loss": 0.0011, "step": 115710 }, { "epoch": 1.8934795058496277, "grad_norm": 0.016476692631840706, "learning_rate": 6.321555704730109e-06, "loss": 0.0018, "step": 115720 }, { "epoch": 1.8936431318007036, "grad_norm": 0.1980966329574585, "learning_rate": 6.320867128338799e-06, "loss": 0.0021, "step": 115730 }, { "epoch": 1.8938067577517794, "grad_norm": 0.10830225050449371, "learning_rate": 6.320178525016133e-06, "loss": 0.0016, "step": 115740 }, { "epoch": 1.8939703837028552, "grad_norm": 0.02639567106962204, "learning_rate": 6.3194898947761475e-06, "loss": 0.0009, "step": 115750 }, { "epoch": 1.8941340096539312, "grad_norm": 0.03301612660288811, "learning_rate": 6.318801237632887e-06, "loss": 0.0014, "step": 115760 }, { "epoch": 1.894297635605007, "grad_norm": 0.043692074716091156, "learning_rate": 6.3181125536003895e-06, "loss": 0.0015, "step": 115770 }, { "epoch": 1.8944612615560827, "grad_norm": 0.049599505960941315, "learning_rate": 6.317423842692699e-06, "loss": 0.0009, "step": 115780 }, { "epoch": 1.8946248875071587, "grad_norm": 0.04681338369846344, "learning_rate": 6.316735104923856e-06, "loss": 0.0016, "step": 115790 }, { "epoch": 1.8947885134582345, "grad_norm": 0.06790671497583389, "learning_rate": 6.316046340307905e-06, "loss": 0.0011, "step": 115800 }, { "epoch": 1.8949521394093103, "grad_norm": 0.02683304436504841, "learning_rate": 6.315357548858888e-06, "loss": 0.0011, "step": 115810 }, { "epoch": 1.8951157653603863, "grad_norm": 0.03740342706441879, "learning_rate": 6.314668730590849e-06, "loss": 0.0013, "step": 115820 }, { "epoch": 1.8952793913114618, "grad_norm": 0.12274408340454102, "learning_rate": 6.313979885517833e-06, "loss": 0.0008, "step": 115830 }, { "epoch": 1.8954430172625378, "grad_norm": 0.036694031208753586, "learning_rate": 6.313291013653884e-06, "loss": 0.0009, "step": 115840 }, { "epoch": 1.8956066432136138, "grad_norm": 0.05139205977320671, "learning_rate": 6.312602115013049e-06, "loss": 0.001, "step": 115850 }, { "epoch": 1.8957702691646894, "grad_norm": 0.5078861117362976, "learning_rate": 6.311913189609372e-06, "loss": 0.0019, "step": 115860 }, { "epoch": 1.8959338951157654, "grad_norm": 0.05784669890999794, "learning_rate": 6.3112242374569035e-06, "loss": 0.0017, "step": 115870 }, { "epoch": 1.8960975210668412, "grad_norm": 0.11676786839962006, "learning_rate": 6.3105352585696845e-06, "loss": 0.0011, "step": 115880 }, { "epoch": 1.896261147017917, "grad_norm": 0.06844828277826309, "learning_rate": 6.309846252961768e-06, "loss": 0.0012, "step": 115890 }, { "epoch": 1.896424772968993, "grad_norm": 0.04129399359226227, "learning_rate": 6.3091572206472e-06, "loss": 0.0015, "step": 115900 }, { "epoch": 1.8965883989200687, "grad_norm": 0.009806904941797256, "learning_rate": 6.30846816164003e-06, "loss": 0.0017, "step": 115910 }, { "epoch": 1.8967520248711445, "grad_norm": 0.14353269338607788, "learning_rate": 6.307779075954307e-06, "loss": 0.0035, "step": 115920 }, { "epoch": 1.8969156508222205, "grad_norm": 0.027207357808947563, "learning_rate": 6.3070899636040804e-06, "loss": 0.0014, "step": 115930 }, { "epoch": 1.8970792767732962, "grad_norm": 0.01955990679562092, "learning_rate": 6.3064008246034e-06, "loss": 0.001, "step": 115940 }, { "epoch": 1.897242902724372, "grad_norm": 0.03622856363654137, "learning_rate": 6.30571165896632e-06, "loss": 0.0012, "step": 115950 }, { "epoch": 1.897406528675448, "grad_norm": 0.03990836441516876, "learning_rate": 6.305022466706889e-06, "loss": 0.0009, "step": 115960 }, { "epoch": 1.8975701546265238, "grad_norm": 0.017243022099137306, "learning_rate": 6.3043332478391606e-06, "loss": 0.0006, "step": 115970 }, { "epoch": 1.8977337805775996, "grad_norm": 0.012160161510109901, "learning_rate": 6.303644002377185e-06, "loss": 0.0011, "step": 115980 }, { "epoch": 1.8978974065286756, "grad_norm": 0.036072202026844025, "learning_rate": 6.302954730335018e-06, "loss": 0.0011, "step": 115990 }, { "epoch": 1.8980610324797513, "grad_norm": 0.04936667159199715, "learning_rate": 6.3022654317267105e-06, "loss": 0.0017, "step": 116000 }, { "epoch": 1.898224658430827, "grad_norm": 0.02475091814994812, "learning_rate": 6.301576106566321e-06, "loss": 0.0044, "step": 116010 }, { "epoch": 1.898388284381903, "grad_norm": 0.08020435273647308, "learning_rate": 6.300886754867899e-06, "loss": 0.0017, "step": 116020 }, { "epoch": 1.8985519103329787, "grad_norm": 0.013753058388829231, "learning_rate": 6.3001973766455035e-06, "loss": 0.0014, "step": 116030 }, { "epoch": 1.8987155362840547, "grad_norm": 0.045551251620054245, "learning_rate": 6.29950797191319e-06, "loss": 0.0019, "step": 116040 }, { "epoch": 1.8988791622351306, "grad_norm": 0.03667276352643967, "learning_rate": 6.298818540685013e-06, "loss": 0.0032, "step": 116050 }, { "epoch": 1.8990427881862062, "grad_norm": 0.057859957218170166, "learning_rate": 6.298129082975031e-06, "loss": 0.0036, "step": 116060 }, { "epoch": 1.8992064141372822, "grad_norm": 0.08026796579360962, "learning_rate": 6.297439598797301e-06, "loss": 0.0022, "step": 116070 }, { "epoch": 1.899370040088358, "grad_norm": 0.04688871651887894, "learning_rate": 6.296750088165882e-06, "loss": 0.0036, "step": 116080 }, { "epoch": 1.8995336660394337, "grad_norm": 2.838284730911255, "learning_rate": 6.296060551094832e-06, "loss": 0.0016, "step": 116090 }, { "epoch": 1.8996972919905097, "grad_norm": 0.04093659669160843, "learning_rate": 6.295370987598206e-06, "loss": 0.0024, "step": 116100 }, { "epoch": 1.8998609179415855, "grad_norm": 0.08340257406234741, "learning_rate": 6.29468139769007e-06, "loss": 0.0022, "step": 116110 }, { "epoch": 1.9000245438926613, "grad_norm": 0.04814256727695465, "learning_rate": 6.293991781384481e-06, "loss": 0.0018, "step": 116120 }, { "epoch": 1.9001881698437373, "grad_norm": 0.1072133406996727, "learning_rate": 6.293302138695501e-06, "loss": 0.0019, "step": 116130 }, { "epoch": 1.900351795794813, "grad_norm": 0.028147026896476746, "learning_rate": 6.292612469637189e-06, "loss": 0.001, "step": 116140 }, { "epoch": 1.9005154217458888, "grad_norm": 0.0014686795184388757, "learning_rate": 6.291922774223609e-06, "loss": 0.0013, "step": 116150 }, { "epoch": 1.9006790476969648, "grad_norm": 0.08722487837076187, "learning_rate": 6.291233052468822e-06, "loss": 0.0018, "step": 116160 }, { "epoch": 1.9008426736480406, "grad_norm": 0.022459276020526886, "learning_rate": 6.2905433043868905e-06, "loss": 0.0016, "step": 116170 }, { "epoch": 1.9010062995991164, "grad_norm": 0.03376904875040054, "learning_rate": 6.28985352999188e-06, "loss": 0.0013, "step": 116180 }, { "epoch": 1.9011699255501924, "grad_norm": 0.06014271453022957, "learning_rate": 6.289163729297851e-06, "loss": 0.0013, "step": 116190 }, { "epoch": 1.9013335515012681, "grad_norm": 0.054265886545181274, "learning_rate": 6.288473902318871e-06, "loss": 0.0009, "step": 116200 }, { "epoch": 1.901497177452344, "grad_norm": 0.06405841559171677, "learning_rate": 6.287784049069004e-06, "loss": 0.0028, "step": 116210 }, { "epoch": 1.90166080340342, "grad_norm": 0.01094901841133833, "learning_rate": 6.287094169562315e-06, "loss": 0.0006, "step": 116220 }, { "epoch": 1.9018244293544955, "grad_norm": 0.09592213481664658, "learning_rate": 6.286404263812871e-06, "loss": 0.0015, "step": 116230 }, { "epoch": 1.9019880553055715, "grad_norm": 0.016800211742520332, "learning_rate": 6.285714331834739e-06, "loss": 0.0064, "step": 116240 }, { "epoch": 1.9021516812566475, "grad_norm": 0.027808818966150284, "learning_rate": 6.285024373641983e-06, "loss": 0.0009, "step": 116250 }, { "epoch": 1.902315307207723, "grad_norm": 0.02655695006251335, "learning_rate": 6.2843343892486756e-06, "loss": 0.0014, "step": 116260 }, { "epoch": 1.902478933158799, "grad_norm": 0.011617420241236687, "learning_rate": 6.283644378668881e-06, "loss": 0.0009, "step": 116270 }, { "epoch": 1.9026425591098748, "grad_norm": 0.050128016620874405, "learning_rate": 6.2829543419166685e-06, "loss": 0.0008, "step": 116280 }, { "epoch": 1.9028061850609506, "grad_norm": 0.07532954216003418, "learning_rate": 6.2822642790061085e-06, "loss": 0.0019, "step": 116290 }, { "epoch": 1.9029698110120266, "grad_norm": 0.03608565032482147, "learning_rate": 6.281574189951271e-06, "loss": 0.0009, "step": 116300 }, { "epoch": 1.9031334369631023, "grad_norm": 0.08841672539710999, "learning_rate": 6.280884074766225e-06, "loss": 0.0016, "step": 116310 }, { "epoch": 1.903297062914178, "grad_norm": 0.008032144978642464, "learning_rate": 6.280193933465042e-06, "loss": 0.0011, "step": 116320 }, { "epoch": 1.903460688865254, "grad_norm": 0.05667736381292343, "learning_rate": 6.2795037660617945e-06, "loss": 0.0028, "step": 116330 }, { "epoch": 1.9036243148163299, "grad_norm": 0.015322903171181679, "learning_rate": 6.2788135725705525e-06, "loss": 0.0011, "step": 116340 }, { "epoch": 1.9037879407674057, "grad_norm": 0.08856234699487686, "learning_rate": 6.278123353005388e-06, "loss": 0.001, "step": 116350 }, { "epoch": 1.9039515667184816, "grad_norm": 0.12385960668325424, "learning_rate": 6.277433107380378e-06, "loss": 0.0018, "step": 116360 }, { "epoch": 1.9041151926695574, "grad_norm": 0.022247489541769028, "learning_rate": 6.2767428357095926e-06, "loss": 0.0009, "step": 116370 }, { "epoch": 1.9042788186206332, "grad_norm": 0.06765469163656235, "learning_rate": 6.276052538007107e-06, "loss": 0.0008, "step": 116380 }, { "epoch": 1.9044424445717092, "grad_norm": 0.009680033661425114, "learning_rate": 6.275362214286993e-06, "loss": 0.001, "step": 116390 }, { "epoch": 1.904606070522785, "grad_norm": 0.01891959458589554, "learning_rate": 6.274671864563331e-06, "loss": 0.0022, "step": 116400 }, { "epoch": 1.9047696964738607, "grad_norm": 0.051937755197286606, "learning_rate": 6.273981488850192e-06, "loss": 0.001, "step": 116410 }, { "epoch": 1.9049333224249367, "grad_norm": 0.018585244193673134, "learning_rate": 6.273291087161655e-06, "loss": 0.0008, "step": 116420 }, { "epoch": 1.9050969483760123, "grad_norm": 0.045751288533210754, "learning_rate": 6.272600659511794e-06, "loss": 0.0016, "step": 116430 }, { "epoch": 1.9052605743270883, "grad_norm": 0.008538858965039253, "learning_rate": 6.271910205914689e-06, "loss": 0.0011, "step": 116440 }, { "epoch": 1.9054242002781643, "grad_norm": 0.04220716282725334, "learning_rate": 6.2712197263844166e-06, "loss": 0.0016, "step": 116450 }, { "epoch": 1.9055878262292398, "grad_norm": 0.04348449409008026, "learning_rate": 6.270529220935056e-06, "loss": 0.0017, "step": 116460 }, { "epoch": 1.9057514521803158, "grad_norm": 0.08083783835172653, "learning_rate": 6.269838689580684e-06, "loss": 0.0018, "step": 116470 }, { "epoch": 1.9059150781313916, "grad_norm": 0.1841062754392624, "learning_rate": 6.2691481323353805e-06, "loss": 0.0029, "step": 116480 }, { "epoch": 1.9060787040824674, "grad_norm": 0.033557623624801636, "learning_rate": 6.268457549213228e-06, "loss": 0.001, "step": 116490 }, { "epoch": 1.9062423300335434, "grad_norm": 0.08358156681060791, "learning_rate": 6.267766940228303e-06, "loss": 0.0014, "step": 116500 }, { "epoch": 1.9064059559846191, "grad_norm": 0.04813264310359955, "learning_rate": 6.267076305394689e-06, "loss": 0.0014, "step": 116510 }, { "epoch": 1.906569581935695, "grad_norm": 0.040434300899505615, "learning_rate": 6.266385644726466e-06, "loss": 0.0005, "step": 116520 }, { "epoch": 1.906733207886771, "grad_norm": 0.0865260437130928, "learning_rate": 6.265694958237718e-06, "loss": 0.0013, "step": 116530 }, { "epoch": 1.9068968338378467, "grad_norm": 0.09247193485498428, "learning_rate": 6.265004245942525e-06, "loss": 0.0009, "step": 116540 }, { "epoch": 1.9070604597889225, "grad_norm": 0.04626036435365677, "learning_rate": 6.2643135078549725e-06, "loss": 0.0021, "step": 116550 }, { "epoch": 1.9072240857399985, "grad_norm": 0.008539445698261261, "learning_rate": 6.263622743989142e-06, "loss": 0.0014, "step": 116560 }, { "epoch": 1.9073877116910742, "grad_norm": 0.017700718715786934, "learning_rate": 6.262931954359119e-06, "loss": 0.0004, "step": 116570 }, { "epoch": 1.90755133764215, "grad_norm": 0.005421953741461039, "learning_rate": 6.262241138978986e-06, "loss": 0.0009, "step": 116580 }, { "epoch": 1.907714963593226, "grad_norm": 0.002429450396448374, "learning_rate": 6.261550297862832e-06, "loss": 0.0011, "step": 116590 }, { "epoch": 1.9078785895443016, "grad_norm": 0.050509389489889145, "learning_rate": 6.260859431024738e-06, "loss": 0.0027, "step": 116600 }, { "epoch": 1.9080422154953776, "grad_norm": 0.09488483518362045, "learning_rate": 6.2601685384787945e-06, "loss": 0.0021, "step": 116610 }, { "epoch": 1.9082058414464536, "grad_norm": 0.04147520661354065, "learning_rate": 6.259477620239085e-06, "loss": 0.0009, "step": 116620 }, { "epoch": 1.908369467397529, "grad_norm": 0.1389937847852707, "learning_rate": 6.2587866763196995e-06, "loss": 0.0035, "step": 116630 }, { "epoch": 1.908533093348605, "grad_norm": 0.06750273704528809, "learning_rate": 6.258095706734721e-06, "loss": 0.0015, "step": 116640 }, { "epoch": 1.9086967192996809, "grad_norm": 0.24934940040111542, "learning_rate": 6.257404711498245e-06, "loss": 0.0013, "step": 116650 }, { "epoch": 1.9088603452507567, "grad_norm": 0.029955200850963593, "learning_rate": 6.256713690624353e-06, "loss": 0.0014, "step": 116660 }, { "epoch": 1.9090239712018326, "grad_norm": 0.05066942423582077, "learning_rate": 6.25602264412714e-06, "loss": 0.0014, "step": 116670 }, { "epoch": 1.9091875971529084, "grad_norm": 0.06979629397392273, "learning_rate": 6.255331572020692e-06, "loss": 0.0014, "step": 116680 }, { "epoch": 1.9093512231039842, "grad_norm": 0.03992693871259689, "learning_rate": 6.254640474319102e-06, "loss": 0.0021, "step": 116690 }, { "epoch": 1.9095148490550602, "grad_norm": 0.009022719226777554, "learning_rate": 6.253949351036459e-06, "loss": 0.0008, "step": 116700 }, { "epoch": 1.909678475006136, "grad_norm": 0.03266141191124916, "learning_rate": 6.253258202186856e-06, "loss": 0.001, "step": 116710 }, { "epoch": 1.9098421009572117, "grad_norm": 0.029746955260634422, "learning_rate": 6.252567027784382e-06, "loss": 0.0016, "step": 116720 }, { "epoch": 1.9100057269082877, "grad_norm": 0.022747203707695007, "learning_rate": 6.251875827843135e-06, "loss": 0.0011, "step": 116730 }, { "epoch": 1.9101693528593635, "grad_norm": 0.045174453407526016, "learning_rate": 6.251184602377202e-06, "loss": 0.003, "step": 116740 }, { "epoch": 1.9103329788104393, "grad_norm": 0.029559098184108734, "learning_rate": 6.2504933514006794e-06, "loss": 0.0007, "step": 116750 }, { "epoch": 1.9104966047615153, "grad_norm": 0.08992127329111099, "learning_rate": 6.24980207492766e-06, "loss": 0.0009, "step": 116760 }, { "epoch": 1.910660230712591, "grad_norm": 0.06725527346134186, "learning_rate": 6.249110772972241e-06, "loss": 0.0011, "step": 116770 }, { "epoch": 1.9108238566636668, "grad_norm": 0.019611340016126633, "learning_rate": 6.248419445548516e-06, "loss": 0.0008, "step": 116780 }, { "epoch": 1.9109874826147428, "grad_norm": 0.06482728570699692, "learning_rate": 6.247728092670578e-06, "loss": 0.0018, "step": 116790 }, { "epoch": 1.9111511085658184, "grad_norm": 0.019909603521227837, "learning_rate": 6.247036714352528e-06, "loss": 0.0015, "step": 116800 }, { "epoch": 1.9113147345168944, "grad_norm": 0.07600054889917374, "learning_rate": 6.2463453106084586e-06, "loss": 0.0018, "step": 116810 }, { "epoch": 1.9114783604679704, "grad_norm": 0.021997425705194473, "learning_rate": 6.245653881452468e-06, "loss": 0.0009, "step": 116820 }, { "epoch": 1.911641986419046, "grad_norm": 0.023141009733080864, "learning_rate": 6.244962426898654e-06, "loss": 0.0017, "step": 116830 }, { "epoch": 1.911805612370122, "grad_norm": 0.0629562959074974, "learning_rate": 6.244270946961116e-06, "loss": 0.0015, "step": 116840 }, { "epoch": 1.9119692383211977, "grad_norm": 0.0875762477517128, "learning_rate": 6.243579441653951e-06, "loss": 0.0008, "step": 116850 }, { "epoch": 1.9121328642722735, "grad_norm": 0.0532524511218071, "learning_rate": 6.2428879109912585e-06, "loss": 0.0019, "step": 116860 }, { "epoch": 1.9122964902233495, "grad_norm": 0.05662708356976509, "learning_rate": 6.2421963549871375e-06, "loss": 0.0019, "step": 116870 }, { "epoch": 1.9124601161744252, "grad_norm": 0.04294246807694435, "learning_rate": 6.241504773655692e-06, "loss": 0.001, "step": 116880 }, { "epoch": 1.912623742125501, "grad_norm": 0.007077959831804037, "learning_rate": 6.240813167011019e-06, "loss": 0.0014, "step": 116890 }, { "epoch": 1.912787368076577, "grad_norm": 0.029983825981616974, "learning_rate": 6.240121535067219e-06, "loss": 0.0013, "step": 116900 }, { "epoch": 1.9129509940276528, "grad_norm": 0.039538390934467316, "learning_rate": 6.239429877838398e-06, "loss": 0.0012, "step": 116910 }, { "epoch": 1.9131146199787286, "grad_norm": 0.05175456404685974, "learning_rate": 6.238738195338655e-06, "loss": 0.0012, "step": 116920 }, { "epoch": 1.9132782459298046, "grad_norm": 0.031881798058748245, "learning_rate": 6.238046487582092e-06, "loss": 0.0019, "step": 116930 }, { "epoch": 1.9134418718808803, "grad_norm": 0.016610438004136086, "learning_rate": 6.237354754582817e-06, "loss": 0.0007, "step": 116940 }, { "epoch": 1.913605497831956, "grad_norm": 0.05232591927051544, "learning_rate": 6.236662996354929e-06, "loss": 0.0008, "step": 116950 }, { "epoch": 1.913769123783032, "grad_norm": 0.019808607175946236, "learning_rate": 6.235971212912535e-06, "loss": 0.0006, "step": 116960 }, { "epoch": 1.9139327497341079, "grad_norm": 0.06992368400096893, "learning_rate": 6.235279404269738e-06, "loss": 0.0012, "step": 116970 }, { "epoch": 1.9140963756851836, "grad_norm": 0.055094070732593536, "learning_rate": 6.234587570440647e-06, "loss": 0.0012, "step": 116980 }, { "epoch": 1.9142600016362596, "grad_norm": 0.06005999818444252, "learning_rate": 6.233895711439363e-06, "loss": 0.0013, "step": 116990 }, { "epoch": 1.9144236275873352, "grad_norm": 0.017986653372645378, "learning_rate": 6.2332038272799955e-06, "loss": 0.0009, "step": 117000 }, { "epoch": 1.9145872535384112, "grad_norm": 0.04299338907003403, "learning_rate": 6.23251191797665e-06, "loss": 0.0017, "step": 117010 }, { "epoch": 1.9147508794894872, "grad_norm": 0.01969030313193798, "learning_rate": 6.231819983543436e-06, "loss": 0.0011, "step": 117020 }, { "epoch": 1.9149145054405627, "grad_norm": 0.0683872327208519, "learning_rate": 6.231128023994459e-06, "loss": 0.0012, "step": 117030 }, { "epoch": 1.9150781313916387, "grad_norm": 0.042427971959114075, "learning_rate": 6.2304360393438315e-06, "loss": 0.0011, "step": 117040 }, { "epoch": 1.9152417573427145, "grad_norm": 0.04792046174407005, "learning_rate": 6.229744029605657e-06, "loss": 0.0024, "step": 117050 }, { "epoch": 1.9154053832937903, "grad_norm": 0.07452265173196793, "learning_rate": 6.229051994794047e-06, "loss": 0.0017, "step": 117060 }, { "epoch": 1.9155690092448663, "grad_norm": 0.08753026276826859, "learning_rate": 6.2283599349231135e-06, "loss": 0.0037, "step": 117070 }, { "epoch": 1.915732635195942, "grad_norm": 0.045287709683179855, "learning_rate": 6.227667850006967e-06, "loss": 0.0025, "step": 117080 }, { "epoch": 1.9158962611470178, "grad_norm": 0.09413715451955795, "learning_rate": 6.2269757400597145e-06, "loss": 0.0021, "step": 117090 }, { "epoch": 1.9160598870980938, "grad_norm": 0.04876013472676277, "learning_rate": 6.226283605095471e-06, "loss": 0.0011, "step": 117100 }, { "epoch": 1.9162235130491696, "grad_norm": 0.08294740319252014, "learning_rate": 6.225591445128349e-06, "loss": 0.0014, "step": 117110 }, { "epoch": 1.9163871390002454, "grad_norm": 0.040865279734134674, "learning_rate": 6.224899260172458e-06, "loss": 0.0013, "step": 117120 }, { "epoch": 1.9165507649513214, "grad_norm": 0.07282593101263046, "learning_rate": 6.224207050241913e-06, "loss": 0.0009, "step": 117130 }, { "epoch": 1.9167143909023971, "grad_norm": 0.03755568712949753, "learning_rate": 6.223514815350827e-06, "loss": 0.0014, "step": 117140 }, { "epoch": 1.916878016853473, "grad_norm": 0.021435867995023727, "learning_rate": 6.222822555513316e-06, "loss": 0.0009, "step": 117150 }, { "epoch": 1.917041642804549, "grad_norm": 0.025241538882255554, "learning_rate": 6.222130270743492e-06, "loss": 0.001, "step": 117160 }, { "epoch": 1.9172052687556247, "grad_norm": 0.051542218774557114, "learning_rate": 6.221437961055472e-06, "loss": 0.0016, "step": 117170 }, { "epoch": 1.9173688947067005, "grad_norm": 0.03815788775682449, "learning_rate": 6.22074562646337e-06, "loss": 0.0006, "step": 117180 }, { "epoch": 1.9175325206577765, "grad_norm": 0.013118752278387547, "learning_rate": 6.2200532669813025e-06, "loss": 0.0015, "step": 117190 }, { "epoch": 1.917696146608852, "grad_norm": 0.027323339134454727, "learning_rate": 6.219360882623388e-06, "loss": 0.0012, "step": 117200 }, { "epoch": 1.917859772559928, "grad_norm": 0.02540556713938713, "learning_rate": 6.218668473403741e-06, "loss": 0.0014, "step": 117210 }, { "epoch": 1.918023398511004, "grad_norm": 0.020059069618582726, "learning_rate": 6.217976039336481e-06, "loss": 0.0008, "step": 117220 }, { "epoch": 1.9181870244620796, "grad_norm": 0.1233687698841095, "learning_rate": 6.217283580435725e-06, "loss": 0.0016, "step": 117230 }, { "epoch": 1.9183506504131556, "grad_norm": 0.055823151022195816, "learning_rate": 6.216591096715592e-06, "loss": 0.0013, "step": 117240 }, { "epoch": 1.9185142763642313, "grad_norm": 0.08128265291452408, "learning_rate": 6.215898588190201e-06, "loss": 0.0022, "step": 117250 }, { "epoch": 1.918677902315307, "grad_norm": 0.01140593271702528, "learning_rate": 6.215206054873672e-06, "loss": 0.001, "step": 117260 }, { "epoch": 1.918841528266383, "grad_norm": 0.02923998050391674, "learning_rate": 6.214513496780126e-06, "loss": 0.0009, "step": 117270 }, { "epoch": 1.9190051542174589, "grad_norm": 0.006421959493309259, "learning_rate": 6.213820913923681e-06, "loss": 0.0018, "step": 117280 }, { "epoch": 1.9191687801685346, "grad_norm": 0.06911318004131317, "learning_rate": 6.213128306318461e-06, "loss": 0.0015, "step": 117290 }, { "epoch": 1.9193324061196106, "grad_norm": 0.019122201949357986, "learning_rate": 6.212435673978587e-06, "loss": 0.0007, "step": 117300 }, { "epoch": 1.9194960320706864, "grad_norm": 0.020951054990291595, "learning_rate": 6.211743016918179e-06, "loss": 0.001, "step": 117310 }, { "epoch": 1.9196596580217622, "grad_norm": 0.0524008609354496, "learning_rate": 6.211050335151363e-06, "loss": 0.0012, "step": 117320 }, { "epoch": 1.9198232839728382, "grad_norm": 0.004270805045962334, "learning_rate": 6.21035762869226e-06, "loss": 0.0011, "step": 117330 }, { "epoch": 1.919986909923914, "grad_norm": 0.03800695762038231, "learning_rate": 6.209664897554995e-06, "loss": 0.0016, "step": 117340 }, { "epoch": 1.9201505358749897, "grad_norm": 0.07370657473802567, "learning_rate": 6.208972141753691e-06, "loss": 0.001, "step": 117350 }, { "epoch": 1.9203141618260657, "grad_norm": 0.007636485155671835, "learning_rate": 6.2082793613024716e-06, "loss": 0.0007, "step": 117360 }, { "epoch": 1.9204777877771415, "grad_norm": 0.020243607461452484, "learning_rate": 6.207586556215466e-06, "loss": 0.0013, "step": 117370 }, { "epoch": 1.9206414137282173, "grad_norm": 0.28582963347435, "learning_rate": 6.206893726506796e-06, "loss": 0.0074, "step": 117380 }, { "epoch": 1.9208050396792933, "grad_norm": 0.016866492107510567, "learning_rate": 6.206200872190591e-06, "loss": 0.0005, "step": 117390 }, { "epoch": 1.9209686656303688, "grad_norm": 0.02172408066689968, "learning_rate": 6.205507993280975e-06, "loss": 0.0012, "step": 117400 }, { "epoch": 1.9211322915814448, "grad_norm": 0.06948840618133545, "learning_rate": 6.204815089792077e-06, "loss": 0.0014, "step": 117410 }, { "epoch": 1.9212959175325206, "grad_norm": 0.13537366688251495, "learning_rate": 6.204122161738022e-06, "loss": 0.0014, "step": 117420 }, { "epoch": 1.9214595434835964, "grad_norm": 0.04846704751253128, "learning_rate": 6.203429209132943e-06, "loss": 0.0006, "step": 117430 }, { "epoch": 1.9216231694346724, "grad_norm": 0.10861245542764664, "learning_rate": 6.202736231990965e-06, "loss": 0.0012, "step": 117440 }, { "epoch": 1.9217867953857481, "grad_norm": 0.16896504163742065, "learning_rate": 6.2020432303262165e-06, "loss": 0.0024, "step": 117450 }, { "epoch": 1.921950421336824, "grad_norm": 0.11656534671783447, "learning_rate": 6.201350204152831e-06, "loss": 0.0014, "step": 117460 }, { "epoch": 1.9221140472879, "grad_norm": 0.012587035074830055, "learning_rate": 6.2006571534849345e-06, "loss": 0.0012, "step": 117470 }, { "epoch": 1.9222776732389757, "grad_norm": 0.09946872293949127, "learning_rate": 6.199964078336661e-06, "loss": 0.0016, "step": 117480 }, { "epoch": 1.9224412991900515, "grad_norm": 0.040327493101358414, "learning_rate": 6.199270978722139e-06, "loss": 0.0012, "step": 117490 }, { "epoch": 1.9226049251411275, "grad_norm": 0.004797131288796663, "learning_rate": 6.198577854655504e-06, "loss": 0.0011, "step": 117500 }, { "epoch": 1.9227685510922032, "grad_norm": 0.04960168898105621, "learning_rate": 6.197884706150884e-06, "loss": 0.0014, "step": 117510 }, { "epoch": 1.922932177043279, "grad_norm": 0.008972249925136566, "learning_rate": 6.197191533222415e-06, "loss": 0.0008, "step": 117520 }, { "epoch": 1.923095802994355, "grad_norm": 0.09370189905166626, "learning_rate": 6.196498335884228e-06, "loss": 0.0012, "step": 117530 }, { "epoch": 1.9232594289454308, "grad_norm": 0.002865150570869446, "learning_rate": 6.195805114150458e-06, "loss": 0.001, "step": 117540 }, { "epoch": 1.9234230548965066, "grad_norm": 0.07479730248451233, "learning_rate": 6.195111868035239e-06, "loss": 0.0024, "step": 117550 }, { "epoch": 1.9235866808475826, "grad_norm": 0.06303602457046509, "learning_rate": 6.194418597552705e-06, "loss": 0.0017, "step": 117560 }, { "epoch": 1.923750306798658, "grad_norm": 0.04165343940258026, "learning_rate": 6.193725302716991e-06, "loss": 0.0012, "step": 117570 }, { "epoch": 1.923913932749734, "grad_norm": 0.01873505860567093, "learning_rate": 6.1930319835422336e-06, "loss": 0.0017, "step": 117580 }, { "epoch": 1.92407755870081, "grad_norm": 0.007651142310351133, "learning_rate": 6.192338640042569e-06, "loss": 0.001, "step": 117590 }, { "epoch": 1.9242411846518856, "grad_norm": 0.044182341545820236, "learning_rate": 6.191645272232134e-06, "loss": 0.0013, "step": 117600 }, { "epoch": 1.9244048106029616, "grad_norm": 0.0046302806586027145, "learning_rate": 6.190951880125063e-06, "loss": 0.001, "step": 117610 }, { "epoch": 1.9245684365540374, "grad_norm": 0.003783752443268895, "learning_rate": 6.190258463735499e-06, "loss": 0.0021, "step": 117620 }, { "epoch": 1.9247320625051132, "grad_norm": 0.040209680795669556, "learning_rate": 6.189565023077576e-06, "loss": 0.0034, "step": 117630 }, { "epoch": 1.9248956884561892, "grad_norm": 0.047952525317668915, "learning_rate": 6.1888715581654345e-06, "loss": 0.0011, "step": 117640 }, { "epoch": 1.925059314407265, "grad_norm": 0.21416862308979034, "learning_rate": 6.188178069013212e-06, "loss": 0.0012, "step": 117650 }, { "epoch": 1.9252229403583407, "grad_norm": 0.1288350522518158, "learning_rate": 6.187484555635049e-06, "loss": 0.0011, "step": 117660 }, { "epoch": 1.9253865663094167, "grad_norm": 0.008839355781674385, "learning_rate": 6.186791018045086e-06, "loss": 0.0016, "step": 117670 }, { "epoch": 1.9255501922604925, "grad_norm": 0.07214859127998352, "learning_rate": 6.186097456257465e-06, "loss": 0.0018, "step": 117680 }, { "epoch": 1.9257138182115683, "grad_norm": 0.020649190992116928, "learning_rate": 6.185403870286324e-06, "loss": 0.0009, "step": 117690 }, { "epoch": 1.9258774441626443, "grad_norm": 0.02840014360845089, "learning_rate": 6.184710260145807e-06, "loss": 0.0011, "step": 117700 }, { "epoch": 1.92604107011372, "grad_norm": 0.003183106193318963, "learning_rate": 6.184016625850055e-06, "loss": 0.0018, "step": 117710 }, { "epoch": 1.9262046960647958, "grad_norm": 0.031074360013008118, "learning_rate": 6.183322967413212e-06, "loss": 0.0036, "step": 117720 }, { "epoch": 1.9263683220158718, "grad_norm": 0.01478617824614048, "learning_rate": 6.182629284849419e-06, "loss": 0.0015, "step": 117730 }, { "epoch": 1.9265319479669476, "grad_norm": 0.031958963721990585, "learning_rate": 6.181935578172821e-06, "loss": 0.001, "step": 117740 }, { "epoch": 1.9266955739180234, "grad_norm": 0.062027040868997574, "learning_rate": 6.181241847397562e-06, "loss": 0.0032, "step": 117750 }, { "epoch": 1.9268591998690994, "grad_norm": 0.06047171354293823, "learning_rate": 6.180548092537786e-06, "loss": 0.0013, "step": 117760 }, { "epoch": 1.927022825820175, "grad_norm": 0.0773591473698616, "learning_rate": 6.179854313607638e-06, "loss": 0.0011, "step": 117770 }, { "epoch": 1.927186451771251, "grad_norm": 0.20607854425907135, "learning_rate": 6.179160510621264e-06, "loss": 0.0016, "step": 117780 }, { "epoch": 1.927350077722327, "grad_norm": 0.040451738983392715, "learning_rate": 6.178466683592811e-06, "loss": 0.0018, "step": 117790 }, { "epoch": 1.9275137036734025, "grad_norm": 0.03292211890220642, "learning_rate": 6.177772832536423e-06, "loss": 0.0019, "step": 117800 }, { "epoch": 1.9276773296244785, "grad_norm": 0.12923869490623474, "learning_rate": 6.1770789574662505e-06, "loss": 0.0013, "step": 117810 }, { "epoch": 1.9278409555755542, "grad_norm": 0.057012882083654404, "learning_rate": 6.1763850583964365e-06, "loss": 0.0012, "step": 117820 }, { "epoch": 1.92800458152663, "grad_norm": 0.0453040786087513, "learning_rate": 6.1756911353411355e-06, "loss": 0.0007, "step": 117830 }, { "epoch": 1.928168207477706, "grad_norm": 0.03468862548470497, "learning_rate": 6.174997188314489e-06, "loss": 0.0016, "step": 117840 }, { "epoch": 1.9283318334287818, "grad_norm": 0.08176030963659286, "learning_rate": 6.174303217330651e-06, "loss": 0.0018, "step": 117850 }, { "epoch": 1.9284954593798576, "grad_norm": 0.052685435861349106, "learning_rate": 6.173609222403767e-06, "loss": 0.0018, "step": 117860 }, { "epoch": 1.9286590853309336, "grad_norm": 0.025135941803455353, "learning_rate": 6.172915203547991e-06, "loss": 0.0005, "step": 117870 }, { "epoch": 1.9288227112820093, "grad_norm": 0.02392028272151947, "learning_rate": 6.172221160777469e-06, "loss": 0.0016, "step": 117880 }, { "epoch": 1.928986337233085, "grad_norm": 0.0710330531001091, "learning_rate": 6.171527094106358e-06, "loss": 0.0015, "step": 117890 }, { "epoch": 1.929149963184161, "grad_norm": 0.09466784447431564, "learning_rate": 6.170833003548803e-06, "loss": 0.0019, "step": 117900 }, { "epoch": 1.9293135891352369, "grad_norm": 0.001489592483267188, "learning_rate": 6.17013888911896e-06, "loss": 0.0004, "step": 117910 }, { "epoch": 1.9294772150863126, "grad_norm": 0.06147840991616249, "learning_rate": 6.1694447508309805e-06, "loss": 0.0011, "step": 117920 }, { "epoch": 1.9296408410373886, "grad_norm": 0.0382123664021492, "learning_rate": 6.168750588699017e-06, "loss": 0.001, "step": 117930 }, { "epoch": 1.9298044669884644, "grad_norm": 0.004053120501339436, "learning_rate": 6.168056402737222e-06, "loss": 0.0012, "step": 117940 }, { "epoch": 1.9299680929395402, "grad_norm": 0.16087211668491364, "learning_rate": 6.167362192959752e-06, "loss": 0.0021, "step": 117950 }, { "epoch": 1.9301317188906162, "grad_norm": 0.0631430596113205, "learning_rate": 6.166667959380759e-06, "loss": 0.0019, "step": 117960 }, { "epoch": 1.9302953448416917, "grad_norm": 0.09733365476131439, "learning_rate": 6.165973702014398e-06, "loss": 0.0015, "step": 117970 }, { "epoch": 1.9304589707927677, "grad_norm": 0.47413530945777893, "learning_rate": 6.1652794208748245e-06, "loss": 0.0014, "step": 117980 }, { "epoch": 1.9306225967438437, "grad_norm": 0.05025734007358551, "learning_rate": 6.164585115976196e-06, "loss": 0.0013, "step": 117990 }, { "epoch": 1.9307862226949193, "grad_norm": 0.02546517364680767, "learning_rate": 6.163890787332667e-06, "loss": 0.0012, "step": 118000 }, { "epoch": 1.9309498486459953, "grad_norm": 0.052029676735401154, "learning_rate": 6.163196434958395e-06, "loss": 0.0018, "step": 118010 }, { "epoch": 1.931113474597071, "grad_norm": 0.07788275182247162, "learning_rate": 6.162502058867536e-06, "loss": 0.0014, "step": 118020 }, { "epoch": 1.9312771005481468, "grad_norm": 0.030521459877490997, "learning_rate": 6.161807659074249e-06, "loss": 0.0013, "step": 118030 }, { "epoch": 1.9314407264992228, "grad_norm": 0.09160740673542023, "learning_rate": 6.161113235592692e-06, "loss": 0.0012, "step": 118040 }, { "epoch": 1.9316043524502986, "grad_norm": 0.008437631651759148, "learning_rate": 6.1604187884370236e-06, "loss": 0.0006, "step": 118050 }, { "epoch": 1.9317679784013744, "grad_norm": 0.07921970635652542, "learning_rate": 6.1597243176214025e-06, "loss": 0.0011, "step": 118060 }, { "epoch": 1.9319316043524504, "grad_norm": 0.0684942975640297, "learning_rate": 6.15902982315999e-06, "loss": 0.0009, "step": 118070 }, { "epoch": 1.9320952303035261, "grad_norm": 0.03998180851340294, "learning_rate": 6.158335305066945e-06, "loss": 0.0016, "step": 118080 }, { "epoch": 1.932258856254602, "grad_norm": 0.023357072845101357, "learning_rate": 6.157640763356427e-06, "loss": 0.0019, "step": 118090 }, { "epoch": 1.932422482205678, "grad_norm": 0.038074374198913574, "learning_rate": 6.1569461980426e-06, "loss": 0.0036, "step": 118100 }, { "epoch": 1.9325861081567537, "grad_norm": 0.07781840115785599, "learning_rate": 6.156251609139623e-06, "loss": 0.0024, "step": 118110 }, { "epoch": 1.9327497341078295, "grad_norm": 0.018070466816425323, "learning_rate": 6.155556996661659e-06, "loss": 0.0008, "step": 118120 }, { "epoch": 1.9329133600589055, "grad_norm": 0.025971967726945877, "learning_rate": 6.154862360622871e-06, "loss": 0.0009, "step": 118130 }, { "epoch": 1.9330769860099812, "grad_norm": 0.03925008699297905, "learning_rate": 6.154167701037421e-06, "loss": 0.0012, "step": 118140 }, { "epoch": 1.933240611961057, "grad_norm": 0.16911296546459198, "learning_rate": 6.153473017919473e-06, "loss": 0.002, "step": 118150 }, { "epoch": 1.933404237912133, "grad_norm": 0.05068056657910347, "learning_rate": 6.152778311283193e-06, "loss": 0.0015, "step": 118160 }, { "epoch": 1.9335678638632086, "grad_norm": 0.009455600753426552, "learning_rate": 6.1520835811427416e-06, "loss": 0.0014, "step": 118170 }, { "epoch": 1.9337314898142846, "grad_norm": 0.05380374193191528, "learning_rate": 6.151388827512287e-06, "loss": 0.0015, "step": 118180 }, { "epoch": 1.9338951157653606, "grad_norm": 0.03655407577753067, "learning_rate": 6.150694050405993e-06, "loss": 0.0011, "step": 118190 }, { "epoch": 1.934058741716436, "grad_norm": 0.011913926340639591, "learning_rate": 6.149999249838025e-06, "loss": 0.0008, "step": 118200 }, { "epoch": 1.934222367667512, "grad_norm": 0.01242611650377512, "learning_rate": 6.149304425822552e-06, "loss": 0.001, "step": 118210 }, { "epoch": 1.9343859936185879, "grad_norm": 0.06837023794651031, "learning_rate": 6.1486095783737386e-06, "loss": 0.0018, "step": 118220 }, { "epoch": 1.9345496195696636, "grad_norm": 0.05885119363665581, "learning_rate": 6.1479147075057525e-06, "loss": 0.001, "step": 118230 }, { "epoch": 1.9347132455207396, "grad_norm": 0.03515050560235977, "learning_rate": 6.147219813232762e-06, "loss": 0.0005, "step": 118240 }, { "epoch": 1.9348768714718154, "grad_norm": 0.13358670473098755, "learning_rate": 6.146524895568935e-06, "loss": 0.0012, "step": 118250 }, { "epoch": 1.9350404974228912, "grad_norm": 0.006638607941567898, "learning_rate": 6.145829954528441e-06, "loss": 0.0016, "step": 118260 }, { "epoch": 1.9352041233739672, "grad_norm": 0.03446587175130844, "learning_rate": 6.145134990125447e-06, "loss": 0.0017, "step": 118270 }, { "epoch": 1.935367749325043, "grad_norm": 0.03723165765404701, "learning_rate": 6.144440002374127e-06, "loss": 0.0009, "step": 118280 }, { "epoch": 1.9355313752761187, "grad_norm": 0.06019928306341171, "learning_rate": 6.143744991288647e-06, "loss": 0.0034, "step": 118290 }, { "epoch": 1.9356950012271947, "grad_norm": 0.09850374609231949, "learning_rate": 6.14304995688318e-06, "loss": 0.0015, "step": 118300 }, { "epoch": 1.9358586271782705, "grad_norm": 0.005140484776347876, "learning_rate": 6.142354899171896e-06, "loss": 0.0009, "step": 118310 }, { "epoch": 1.9360222531293463, "grad_norm": 0.031659722328186035, "learning_rate": 6.1416598181689675e-06, "loss": 0.0018, "step": 118320 }, { "epoch": 1.9361858790804223, "grad_norm": 0.06874864548444748, "learning_rate": 6.140964713888567e-06, "loss": 0.0014, "step": 118330 }, { "epoch": 1.9363495050314978, "grad_norm": 0.0171416737139225, "learning_rate": 6.140269586344864e-06, "loss": 0.0031, "step": 118340 }, { "epoch": 1.9365131309825738, "grad_norm": 0.07310586422681808, "learning_rate": 6.139574435552036e-06, "loss": 0.0017, "step": 118350 }, { "epoch": 1.9366767569336498, "grad_norm": 0.008323591202497482, "learning_rate": 6.138879261524254e-06, "loss": 0.0014, "step": 118360 }, { "epoch": 1.9368403828847254, "grad_norm": 0.021702520549297333, "learning_rate": 6.138184064275692e-06, "loss": 0.0016, "step": 118370 }, { "epoch": 1.9370040088358014, "grad_norm": 0.022383147850632668, "learning_rate": 6.137488843820526e-06, "loss": 0.0011, "step": 118380 }, { "epoch": 1.9371676347868771, "grad_norm": 0.04893776774406433, "learning_rate": 6.136793600172928e-06, "loss": 0.0013, "step": 118390 }, { "epoch": 1.937331260737953, "grad_norm": 0.0228461604565382, "learning_rate": 6.136098333347077e-06, "loss": 0.001, "step": 118400 }, { "epoch": 1.937494886689029, "grad_norm": 0.03914283588528633, "learning_rate": 6.135403043357148e-06, "loss": 0.0015, "step": 118410 }, { "epoch": 1.9376585126401047, "grad_norm": 0.024769404903054237, "learning_rate": 6.1347077302173145e-06, "loss": 0.0012, "step": 118420 }, { "epoch": 1.9378221385911805, "grad_norm": 0.07912629842758179, "learning_rate": 6.134012393941756e-06, "loss": 0.0021, "step": 118430 }, { "epoch": 1.9379857645422565, "grad_norm": 0.04113065078854561, "learning_rate": 6.133317034544649e-06, "loss": 0.0023, "step": 118440 }, { "epoch": 1.9381493904933322, "grad_norm": 0.0318610779941082, "learning_rate": 6.132621652040172e-06, "loss": 0.0028, "step": 118450 }, { "epoch": 1.938313016444408, "grad_norm": 0.08953453600406647, "learning_rate": 6.131926246442502e-06, "loss": 0.0016, "step": 118460 }, { "epoch": 1.938476642395484, "grad_norm": 0.02938358299434185, "learning_rate": 6.1312308177658196e-06, "loss": 0.0015, "step": 118470 }, { "epoch": 1.9386402683465598, "grad_norm": 0.14038024842739105, "learning_rate": 6.130535366024302e-06, "loss": 0.0014, "step": 118480 }, { "epoch": 1.9388038942976356, "grad_norm": 0.10161426663398743, "learning_rate": 6.12983989123213e-06, "loss": 0.0012, "step": 118490 }, { "epoch": 1.9389675202487116, "grad_norm": 0.011795098893344402, "learning_rate": 6.129144393403483e-06, "loss": 0.0013, "step": 118500 }, { "epoch": 1.9391311461997873, "grad_norm": 0.010330092161893845, "learning_rate": 6.128448872552544e-06, "loss": 0.0019, "step": 118510 }, { "epoch": 1.939294772150863, "grad_norm": 0.004644056316465139, "learning_rate": 6.1277533286934906e-06, "loss": 0.0013, "step": 118520 }, { "epoch": 1.939458398101939, "grad_norm": 0.008897819556295872, "learning_rate": 6.127057761840507e-06, "loss": 0.0012, "step": 118530 }, { "epoch": 1.9396220240530146, "grad_norm": 0.0269397534430027, "learning_rate": 6.126362172007772e-06, "loss": 0.0007, "step": 118540 }, { "epoch": 1.9397856500040906, "grad_norm": 0.03839292377233505, "learning_rate": 6.125666559209473e-06, "loss": 0.0013, "step": 118550 }, { "epoch": 1.9399492759551666, "grad_norm": 0.01652940921485424, "learning_rate": 6.1249709234597884e-06, "loss": 0.0024, "step": 118560 }, { "epoch": 1.9401129019062422, "grad_norm": 0.043022241443395615, "learning_rate": 6.124275264772904e-06, "loss": 0.0015, "step": 118570 }, { "epoch": 1.9402765278573182, "grad_norm": 0.09923496842384338, "learning_rate": 6.123579583163003e-06, "loss": 0.0012, "step": 118580 }, { "epoch": 1.940440153808394, "grad_norm": 0.04602548107504845, "learning_rate": 6.12288387864427e-06, "loss": 0.001, "step": 118590 }, { "epoch": 1.9406037797594697, "grad_norm": 0.06064559519290924, "learning_rate": 6.12218815123089e-06, "loss": 0.0011, "step": 118600 }, { "epoch": 1.9407674057105457, "grad_norm": 0.019266832619905472, "learning_rate": 6.121492400937048e-06, "loss": 0.0027, "step": 118610 }, { "epoch": 1.9409310316616215, "grad_norm": 0.02585272490978241, "learning_rate": 6.120796627776928e-06, "loss": 0.0009, "step": 118620 }, { "epoch": 1.9410946576126973, "grad_norm": 0.0688747689127922, "learning_rate": 6.12010083176472e-06, "loss": 0.0017, "step": 118630 }, { "epoch": 1.9412582835637733, "grad_norm": 0.014731621369719505, "learning_rate": 6.119405012914608e-06, "loss": 0.0015, "step": 118640 }, { "epoch": 1.941421909514849, "grad_norm": 0.005429956596344709, "learning_rate": 6.11870917124078e-06, "loss": 0.0011, "step": 118650 }, { "epoch": 1.9415855354659248, "grad_norm": 0.06628963351249695, "learning_rate": 6.118013306757423e-06, "loss": 0.0015, "step": 118660 }, { "epoch": 1.9417491614170008, "grad_norm": 0.03311660885810852, "learning_rate": 6.1173174194787266e-06, "loss": 0.0017, "step": 118670 }, { "epoch": 1.9419127873680766, "grad_norm": 0.051923539489507675, "learning_rate": 6.1166215094188764e-06, "loss": 0.0015, "step": 118680 }, { "epoch": 1.9420764133191524, "grad_norm": 0.043122243136167526, "learning_rate": 6.115925576592065e-06, "loss": 0.001, "step": 118690 }, { "epoch": 1.9422400392702284, "grad_norm": 0.028520511463284492, "learning_rate": 6.115229621012479e-06, "loss": 0.002, "step": 118700 }, { "epoch": 1.9424036652213041, "grad_norm": 0.07293510437011719, "learning_rate": 6.11453364269431e-06, "loss": 0.0018, "step": 118710 }, { "epoch": 1.94256729117238, "grad_norm": 0.03483474999666214, "learning_rate": 6.113837641651749e-06, "loss": 0.0014, "step": 118720 }, { "epoch": 1.942730917123456, "grad_norm": 0.026431811973452568, "learning_rate": 6.113141617898984e-06, "loss": 0.0008, "step": 118730 }, { "epoch": 1.9428945430745315, "grad_norm": 0.04345131665468216, "learning_rate": 6.1124455714502085e-06, "loss": 0.0008, "step": 118740 }, { "epoch": 1.9430581690256075, "grad_norm": 0.04004210606217384, "learning_rate": 6.111749502319614e-06, "loss": 0.001, "step": 118750 }, { "epoch": 1.9432217949766835, "grad_norm": 0.06451477110385895, "learning_rate": 6.111053410521394e-06, "loss": 0.0009, "step": 118760 }, { "epoch": 1.943385420927759, "grad_norm": 0.022690029814839363, "learning_rate": 6.110357296069738e-06, "loss": 0.0013, "step": 118770 }, { "epoch": 1.943549046878835, "grad_norm": 0.00284174969419837, "learning_rate": 6.1096611589788415e-06, "loss": 0.0015, "step": 118780 }, { "epoch": 1.9437126728299108, "grad_norm": 0.010375400073826313, "learning_rate": 6.108964999262897e-06, "loss": 0.001, "step": 118790 }, { "epoch": 1.9438762987809866, "grad_norm": 0.03513306751847267, "learning_rate": 6.108268816936102e-06, "loss": 0.0009, "step": 118800 }, { "epoch": 1.9440399247320626, "grad_norm": 0.017811156809329987, "learning_rate": 6.107572612012645e-06, "loss": 0.001, "step": 118810 }, { "epoch": 1.9442035506831383, "grad_norm": 0.045858416706323624, "learning_rate": 6.106876384506727e-06, "loss": 0.0009, "step": 118820 }, { "epoch": 1.944367176634214, "grad_norm": 0.019092276692390442, "learning_rate": 6.106180134432539e-06, "loss": 0.0014, "step": 118830 }, { "epoch": 1.94453080258529, "grad_norm": 0.36621546745300293, "learning_rate": 6.10548386180428e-06, "loss": 0.0011, "step": 118840 }, { "epoch": 1.9446944285363659, "grad_norm": 0.014136096462607384, "learning_rate": 6.104787566636144e-06, "loss": 0.0008, "step": 118850 }, { "epoch": 1.9448580544874416, "grad_norm": 0.003796932753175497, "learning_rate": 6.104091248942331e-06, "loss": 0.0016, "step": 118860 }, { "epoch": 1.9450216804385176, "grad_norm": 0.044914938509464264, "learning_rate": 6.103394908737035e-06, "loss": 0.0023, "step": 118870 }, { "epoch": 1.9451853063895934, "grad_norm": 0.13112995028495789, "learning_rate": 6.102698546034456e-06, "loss": 0.0028, "step": 118880 }, { "epoch": 1.9453489323406692, "grad_norm": 0.017208995297551155, "learning_rate": 6.102002160848791e-06, "loss": 0.0011, "step": 118890 }, { "epoch": 1.9455125582917452, "grad_norm": 0.02665264904499054, "learning_rate": 6.10130575319424e-06, "loss": 0.0013, "step": 118900 }, { "epoch": 1.945676184242821, "grad_norm": 0.08625130355358124, "learning_rate": 6.1006093230849995e-06, "loss": 0.0014, "step": 118910 }, { "epoch": 1.9458398101938967, "grad_norm": 0.10868486016988754, "learning_rate": 6.0999128705352724e-06, "loss": 0.0019, "step": 118920 }, { "epoch": 1.9460034361449727, "grad_norm": 0.03313758224248886, "learning_rate": 6.099216395559256e-06, "loss": 0.0014, "step": 118930 }, { "epoch": 1.9461670620960483, "grad_norm": 0.12774190306663513, "learning_rate": 6.098519898171155e-06, "loss": 0.0014, "step": 118940 }, { "epoch": 1.9463306880471243, "grad_norm": 0.02289140224456787, "learning_rate": 6.097823378385165e-06, "loss": 0.0013, "step": 118950 }, { "epoch": 1.9464943139982003, "grad_norm": 0.009346865117549896, "learning_rate": 6.097126836215491e-06, "loss": 0.001, "step": 118960 }, { "epoch": 1.9466579399492758, "grad_norm": 0.09549041092395782, "learning_rate": 6.096430271676332e-06, "loss": 0.0014, "step": 118970 }, { "epoch": 1.9468215659003518, "grad_norm": 0.0468902513384819, "learning_rate": 6.095733684781895e-06, "loss": 0.001, "step": 118980 }, { "epoch": 1.9469851918514276, "grad_norm": 0.06475846469402313, "learning_rate": 6.095037075546379e-06, "loss": 0.0013, "step": 118990 }, { "epoch": 1.9471488178025034, "grad_norm": 0.04643094167113304, "learning_rate": 6.0943404439839885e-06, "loss": 0.0017, "step": 119000 }, { "epoch": 1.9473124437535794, "grad_norm": 0.11105461418628693, "learning_rate": 6.093643790108925e-06, "loss": 0.0038, "step": 119010 }, { "epoch": 1.9474760697046551, "grad_norm": 0.014546798542141914, "learning_rate": 6.092947113935397e-06, "loss": 0.001, "step": 119020 }, { "epoch": 1.947639695655731, "grad_norm": 0.007287310902029276, "learning_rate": 6.092250415477607e-06, "loss": 0.0007, "step": 119030 }, { "epoch": 1.947803321606807, "grad_norm": 0.05269814655184746, "learning_rate": 6.091553694749759e-06, "loss": 0.0019, "step": 119040 }, { "epoch": 1.9479669475578827, "grad_norm": 0.03625969961285591, "learning_rate": 6.09085695176606e-06, "loss": 0.0015, "step": 119050 }, { "epoch": 1.9481305735089585, "grad_norm": 0.01896703988313675, "learning_rate": 6.0901601865407144e-06, "loss": 0.0013, "step": 119060 }, { "epoch": 1.9482941994600345, "grad_norm": 0.02988276071846485, "learning_rate": 6.089463399087931e-06, "loss": 0.0016, "step": 119070 }, { "epoch": 1.9484578254111102, "grad_norm": 0.03091036155819893, "learning_rate": 6.088766589421915e-06, "loss": 0.0016, "step": 119080 }, { "epoch": 1.948621451362186, "grad_norm": 0.0022295129019767046, "learning_rate": 6.088069757556875e-06, "loss": 0.0012, "step": 119090 }, { "epoch": 1.948785077313262, "grad_norm": 0.07226409018039703, "learning_rate": 6.087372903507016e-06, "loss": 0.002, "step": 119100 }, { "epoch": 1.9489487032643378, "grad_norm": 0.030575450509786606, "learning_rate": 6.086676027286551e-06, "loss": 0.0025, "step": 119110 }, { "epoch": 1.9491123292154136, "grad_norm": 0.03141757845878601, "learning_rate": 6.085979128909684e-06, "loss": 0.0012, "step": 119120 }, { "epoch": 1.9492759551664895, "grad_norm": 0.030193526297807693, "learning_rate": 6.085282208390626e-06, "loss": 0.0012, "step": 119130 }, { "epoch": 1.949439581117565, "grad_norm": 0.059515099972486496, "learning_rate": 6.084585265743588e-06, "loss": 0.0018, "step": 119140 }, { "epoch": 1.949603207068641, "grad_norm": 0.035271383821964264, "learning_rate": 6.083888300982779e-06, "loss": 0.0006, "step": 119150 }, { "epoch": 1.9497668330197169, "grad_norm": 0.024444013833999634, "learning_rate": 6.083191314122407e-06, "loss": 0.002, "step": 119160 }, { "epoch": 1.9499304589707926, "grad_norm": 0.07298582047224045, "learning_rate": 6.0824943051766875e-06, "loss": 0.0011, "step": 119170 }, { "epoch": 1.9500940849218686, "grad_norm": 0.022182315587997437, "learning_rate": 6.081797274159828e-06, "loss": 0.0012, "step": 119180 }, { "epoch": 1.9502577108729444, "grad_norm": 0.06204352155327797, "learning_rate": 6.081100221086045e-06, "loss": 0.0017, "step": 119190 }, { "epoch": 1.9504213368240202, "grad_norm": 0.019729726016521454, "learning_rate": 6.080403145969545e-06, "loss": 0.0026, "step": 119200 }, { "epoch": 1.9505849627750962, "grad_norm": 0.07906555384397507, "learning_rate": 6.079706048824546e-06, "loss": 0.0013, "step": 119210 }, { "epoch": 1.950748588726172, "grad_norm": 0.08100667595863342, "learning_rate": 6.079008929665257e-06, "loss": 0.0016, "step": 119220 }, { "epoch": 1.9509122146772477, "grad_norm": 0.049713384360075, "learning_rate": 6.078311788505895e-06, "loss": 0.0028, "step": 119230 }, { "epoch": 1.9510758406283237, "grad_norm": 0.04871242120862007, "learning_rate": 6.077614625360672e-06, "loss": 0.0024, "step": 119240 }, { "epoch": 1.9512394665793995, "grad_norm": 0.026498012244701385, "learning_rate": 6.076917440243804e-06, "loss": 0.0016, "step": 119250 }, { "epoch": 1.9514030925304753, "grad_norm": 0.0456148236989975, "learning_rate": 6.076220233169504e-06, "loss": 0.0014, "step": 119260 }, { "epoch": 1.9515667184815513, "grad_norm": 0.021109307184815407, "learning_rate": 6.07552300415199e-06, "loss": 0.0015, "step": 119270 }, { "epoch": 1.951730344432627, "grad_norm": 0.041412029415369034, "learning_rate": 6.074825753205475e-06, "loss": 0.0012, "step": 119280 }, { "epoch": 1.9518939703837028, "grad_norm": 0.009609627537429333, "learning_rate": 6.074128480344179e-06, "loss": 0.0006, "step": 119290 }, { "epoch": 1.9520575963347788, "grad_norm": 0.04334276169538498, "learning_rate": 6.073431185582315e-06, "loss": 0.0011, "step": 119300 }, { "epoch": 1.9522212222858544, "grad_norm": 0.056369420140981674, "learning_rate": 6.0727338689341045e-06, "loss": 0.0009, "step": 119310 }, { "epoch": 1.9523848482369304, "grad_norm": 0.016989052295684814, "learning_rate": 6.072036530413759e-06, "loss": 0.0004, "step": 119320 }, { "epoch": 1.9525484741880064, "grad_norm": 0.05167118459939957, "learning_rate": 6.071339170035504e-06, "loss": 0.0009, "step": 119330 }, { "epoch": 1.952712100139082, "grad_norm": 0.03151416406035423, "learning_rate": 6.070641787813552e-06, "loss": 0.0039, "step": 119340 }, { "epoch": 1.952875726090158, "grad_norm": 0.03395336866378784, "learning_rate": 6.069944383762126e-06, "loss": 0.0019, "step": 119350 }, { "epoch": 1.9530393520412337, "grad_norm": 0.052283499389886856, "learning_rate": 6.0692469578954445e-06, "loss": 0.0007, "step": 119360 }, { "epoch": 1.9532029779923095, "grad_norm": 0.04098045453429222, "learning_rate": 6.068549510227725e-06, "loss": 0.0008, "step": 119370 }, { "epoch": 1.9533666039433855, "grad_norm": 0.08850008994340897, "learning_rate": 6.067852040773191e-06, "loss": 0.0088, "step": 119380 }, { "epoch": 1.9535302298944612, "grad_norm": 0.03149624541401863, "learning_rate": 6.067154549546061e-06, "loss": 0.0021, "step": 119390 }, { "epoch": 1.953693855845537, "grad_norm": 0.02862219139933586, "learning_rate": 6.0664570365605595e-06, "loss": 0.0012, "step": 119400 }, { "epoch": 1.953857481796613, "grad_norm": 0.015992142260074615, "learning_rate": 6.065759501830904e-06, "loss": 0.0014, "step": 119410 }, { "epoch": 1.9540211077476888, "grad_norm": 0.028695162385702133, "learning_rate": 6.065061945371319e-06, "loss": 0.001, "step": 119420 }, { "epoch": 1.9541847336987646, "grad_norm": 0.034263331443071365, "learning_rate": 6.064364367196026e-06, "loss": 0.0017, "step": 119430 }, { "epoch": 1.9543483596498405, "grad_norm": 0.051756601780653, "learning_rate": 6.0636667673192504e-06, "loss": 0.0014, "step": 119440 }, { "epoch": 1.9545119856009163, "grad_norm": 0.031576551496982574, "learning_rate": 6.062969145755213e-06, "loss": 0.0011, "step": 119450 }, { "epoch": 1.954675611551992, "grad_norm": 0.10313902795314789, "learning_rate": 6.06227150251814e-06, "loss": 0.0012, "step": 119460 }, { "epoch": 1.954839237503068, "grad_norm": 0.031585484743118286, "learning_rate": 6.061573837622252e-06, "loss": 0.0024, "step": 119470 }, { "epoch": 1.9550028634541439, "grad_norm": 0.005143970251083374, "learning_rate": 6.06087615108178e-06, "loss": 0.001, "step": 119480 }, { "epoch": 1.9551664894052196, "grad_norm": 0.02497096359729767, "learning_rate": 6.060178442910942e-06, "loss": 0.0008, "step": 119490 }, { "epoch": 1.9553301153562956, "grad_norm": 0.02893294394016266, "learning_rate": 6.059480713123968e-06, "loss": 0.002, "step": 119500 }, { "epoch": 1.9554937413073712, "grad_norm": 0.05937973037362099, "learning_rate": 6.058782961735085e-06, "loss": 0.0014, "step": 119510 }, { "epoch": 1.9556573672584472, "grad_norm": 0.057451147586107254, "learning_rate": 6.058085188758517e-06, "loss": 0.0025, "step": 119520 }, { "epoch": 1.9558209932095232, "grad_norm": 0.04169692099094391, "learning_rate": 6.057387394208491e-06, "loss": 0.0017, "step": 119530 }, { "epoch": 1.9559846191605987, "grad_norm": 0.04406777396798134, "learning_rate": 6.056689578099236e-06, "loss": 0.0014, "step": 119540 }, { "epoch": 1.9561482451116747, "grad_norm": 0.08059296011924744, "learning_rate": 6.055991740444979e-06, "loss": 0.0018, "step": 119550 }, { "epoch": 1.9563118710627505, "grad_norm": 0.12154093384742737, "learning_rate": 6.055293881259948e-06, "loss": 0.0015, "step": 119560 }, { "epoch": 1.9564754970138263, "grad_norm": 0.03166789188981056, "learning_rate": 6.054596000558373e-06, "loss": 0.0016, "step": 119570 }, { "epoch": 1.9566391229649023, "grad_norm": 0.07111836224794388, "learning_rate": 6.053898098354483e-06, "loss": 0.0007, "step": 119580 }, { "epoch": 1.956802748915978, "grad_norm": 0.03958609327673912, "learning_rate": 6.053200174662506e-06, "loss": 0.0009, "step": 119590 }, { "epoch": 1.9569663748670538, "grad_norm": 0.21469347178936005, "learning_rate": 6.052502229496674e-06, "loss": 0.0012, "step": 119600 }, { "epoch": 1.9571300008181298, "grad_norm": 0.06747271120548248, "learning_rate": 6.051804262871215e-06, "loss": 0.0028, "step": 119610 }, { "epoch": 1.9572936267692056, "grad_norm": 0.10579819232225418, "learning_rate": 6.051106274800363e-06, "loss": 0.0015, "step": 119620 }, { "epoch": 1.9574572527202814, "grad_norm": 0.012511544860899448, "learning_rate": 6.050408265298347e-06, "loss": 0.0007, "step": 119630 }, { "epoch": 1.9576208786713574, "grad_norm": 0.02535507269203663, "learning_rate": 6.0497102343794025e-06, "loss": 0.0008, "step": 119640 }, { "epoch": 1.9577845046224331, "grad_norm": 0.04503259062767029, "learning_rate": 6.049012182057756e-06, "loss": 0.002, "step": 119650 }, { "epoch": 1.957948130573509, "grad_norm": 0.011289082467556, "learning_rate": 6.0483141083476445e-06, "loss": 0.0013, "step": 119660 }, { "epoch": 1.958111756524585, "grad_norm": 0.010366929695010185, "learning_rate": 6.0476160132633e-06, "loss": 0.0039, "step": 119670 }, { "epoch": 1.9582753824756607, "grad_norm": 0.05073757842183113, "learning_rate": 6.046917896818956e-06, "loss": 0.0013, "step": 119680 }, { "epoch": 1.9584390084267365, "grad_norm": 0.03268340602517128, "learning_rate": 6.0462197590288466e-06, "loss": 0.0004, "step": 119690 }, { "epoch": 1.9586026343778125, "grad_norm": 0.05734584480524063, "learning_rate": 6.045521599907206e-06, "loss": 0.0018, "step": 119700 }, { "epoch": 1.958766260328888, "grad_norm": 0.04009668156504631, "learning_rate": 6.044823419468268e-06, "loss": 0.0012, "step": 119710 }, { "epoch": 1.958929886279964, "grad_norm": 0.15444017946720123, "learning_rate": 6.0441252177262695e-06, "loss": 0.001, "step": 119720 }, { "epoch": 1.95909351223104, "grad_norm": 0.0032901533413678408, "learning_rate": 6.043426994695447e-06, "loss": 0.001, "step": 119730 }, { "epoch": 1.9592571381821156, "grad_norm": 0.015472210012376308, "learning_rate": 6.042728750390034e-06, "loss": 0.001, "step": 119740 }, { "epoch": 1.9594207641331916, "grad_norm": 0.15823563933372498, "learning_rate": 6.042030484824271e-06, "loss": 0.0012, "step": 119750 }, { "epoch": 1.9595843900842673, "grad_norm": 0.06124548241496086, "learning_rate": 6.04133219801239e-06, "loss": 0.0012, "step": 119760 }, { "epoch": 1.959748016035343, "grad_norm": 0.021792365238070488, "learning_rate": 6.040633889968632e-06, "loss": 0.0007, "step": 119770 }, { "epoch": 1.959911641986419, "grad_norm": 0.040056534111499786, "learning_rate": 6.039935560707234e-06, "loss": 0.0014, "step": 119780 }, { "epoch": 1.9600752679374949, "grad_norm": 0.043373264372348785, "learning_rate": 6.039237210242434e-06, "loss": 0.0011, "step": 119790 }, { "epoch": 1.9602388938885706, "grad_norm": 0.01861015520989895, "learning_rate": 6.038538838588472e-06, "loss": 0.0012, "step": 119800 }, { "epoch": 1.9604025198396466, "grad_norm": 0.03554641827940941, "learning_rate": 6.037840445759586e-06, "loss": 0.0018, "step": 119810 }, { "epoch": 1.9605661457907224, "grad_norm": 0.0881902426481247, "learning_rate": 6.0371420317700155e-06, "loss": 0.0009, "step": 119820 }, { "epoch": 1.9607297717417982, "grad_norm": 0.03179938346147537, "learning_rate": 6.036443596634002e-06, "loss": 0.0009, "step": 119830 }, { "epoch": 1.9608933976928742, "grad_norm": 0.11756028980016708, "learning_rate": 6.035745140365784e-06, "loss": 0.0019, "step": 119840 }, { "epoch": 1.96105702364395, "grad_norm": 0.07029756903648376, "learning_rate": 6.0350466629796045e-06, "loss": 0.0009, "step": 119850 }, { "epoch": 1.9612206495950257, "grad_norm": 0.05022771283984184, "learning_rate": 6.0343481644897026e-06, "loss": 0.0014, "step": 119860 }, { "epoch": 1.9613842755461017, "grad_norm": 0.0022081451024860144, "learning_rate": 6.033649644910322e-06, "loss": 0.0011, "step": 119870 }, { "epoch": 1.9615479014971775, "grad_norm": 0.01335524395108223, "learning_rate": 6.032951104255704e-06, "loss": 0.0012, "step": 119880 }, { "epoch": 1.9617115274482533, "grad_norm": 0.09796490520238876, "learning_rate": 6.032252542540093e-06, "loss": 0.0012, "step": 119890 }, { "epoch": 1.9618751533993293, "grad_norm": 0.009334666654467583, "learning_rate": 6.031553959777727e-06, "loss": 0.0013, "step": 119900 }, { "epoch": 1.9620387793504048, "grad_norm": 0.05950700491666794, "learning_rate": 6.0308553559828565e-06, "loss": 0.0014, "step": 119910 }, { "epoch": 1.9622024053014808, "grad_norm": 0.02048266865313053, "learning_rate": 6.030156731169719e-06, "loss": 0.0006, "step": 119920 }, { "epoch": 1.9623660312525568, "grad_norm": 0.1322682648897171, "learning_rate": 6.0294580853525655e-06, "loss": 0.0007, "step": 119930 }, { "epoch": 1.9625296572036324, "grad_norm": 0.0402773879468441, "learning_rate": 6.028759418545635e-06, "loss": 0.0016, "step": 119940 }, { "epoch": 1.9626932831547084, "grad_norm": 0.04436475783586502, "learning_rate": 6.0280607307631746e-06, "loss": 0.0013, "step": 119950 }, { "epoch": 1.9628569091057841, "grad_norm": 0.033960118889808655, "learning_rate": 6.027362022019431e-06, "loss": 0.0016, "step": 119960 }, { "epoch": 1.96302053505686, "grad_norm": 0.020776916295289993, "learning_rate": 6.026663292328649e-06, "loss": 0.0008, "step": 119970 }, { "epoch": 1.963184161007936, "grad_norm": 0.05244554951786995, "learning_rate": 6.025964541705076e-06, "loss": 0.0012, "step": 119980 }, { "epoch": 1.9633477869590117, "grad_norm": 0.03702375292778015, "learning_rate": 6.025265770162959e-06, "loss": 0.0015, "step": 119990 }, { "epoch": 1.9635114129100875, "grad_norm": 0.014515979215502739, "learning_rate": 6.024566977716544e-06, "loss": 0.001, "step": 120000 }, { "epoch": 1.9636750388611635, "grad_norm": 0.018521299585700035, "learning_rate": 6.02386816438008e-06, "loss": 0.001, "step": 120010 }, { "epoch": 1.9638386648122392, "grad_norm": 0.07115820050239563, "learning_rate": 6.023169330167815e-06, "loss": 0.0019, "step": 120020 }, { "epoch": 1.964002290763315, "grad_norm": 0.0027043228037655354, "learning_rate": 6.022470475093998e-06, "loss": 0.001, "step": 120030 }, { "epoch": 1.964165916714391, "grad_norm": 0.028324143961071968, "learning_rate": 6.0217715991728766e-06, "loss": 0.0005, "step": 120040 }, { "epoch": 1.9643295426654668, "grad_norm": 0.11278217285871506, "learning_rate": 6.021072702418702e-06, "loss": 0.0017, "step": 120050 }, { "epoch": 1.9644931686165426, "grad_norm": 0.002024602610617876, "learning_rate": 6.020373784845723e-06, "loss": 0.0008, "step": 120060 }, { "epoch": 1.9646567945676185, "grad_norm": 0.01443014107644558, "learning_rate": 6.01967484646819e-06, "loss": 0.0009, "step": 120070 }, { "epoch": 1.964820420518694, "grad_norm": 0.039996057748794556, "learning_rate": 6.018975887300356e-06, "loss": 0.0013, "step": 120080 }, { "epoch": 1.96498404646977, "grad_norm": 0.022381484508514404, "learning_rate": 6.018276907356468e-06, "loss": 0.0026, "step": 120090 }, { "epoch": 1.965147672420846, "grad_norm": 0.08181717246770859, "learning_rate": 6.01757790665078e-06, "loss": 0.0009, "step": 120100 }, { "epoch": 1.9653112983719216, "grad_norm": 0.05858665704727173, "learning_rate": 6.0168788851975445e-06, "loss": 0.0011, "step": 120110 }, { "epoch": 1.9654749243229976, "grad_norm": 0.07952732592821121, "learning_rate": 6.016179843011014e-06, "loss": 0.0016, "step": 120120 }, { "epoch": 1.9656385502740734, "grad_norm": 0.0410461351275444, "learning_rate": 6.015480780105439e-06, "loss": 0.0017, "step": 120130 }, { "epoch": 1.9658021762251492, "grad_norm": 0.10876015573740005, "learning_rate": 6.0147816964950764e-06, "loss": 0.0012, "step": 120140 }, { "epoch": 1.9659658021762252, "grad_norm": 0.06501609832048416, "learning_rate": 6.014082592194177e-06, "loss": 0.0018, "step": 120150 }, { "epoch": 1.966129428127301, "grad_norm": 0.0024795220233500004, "learning_rate": 6.0133834672169964e-06, "loss": 0.0013, "step": 120160 }, { "epoch": 1.9662930540783767, "grad_norm": 0.10373988002538681, "learning_rate": 6.0126843215777885e-06, "loss": 0.0015, "step": 120170 }, { "epoch": 1.9664566800294527, "grad_norm": 0.01795678772032261, "learning_rate": 6.011985155290809e-06, "loss": 0.0007, "step": 120180 }, { "epoch": 1.9666203059805285, "grad_norm": 0.05856075882911682, "learning_rate": 6.011285968370313e-06, "loss": 0.0009, "step": 120190 }, { "epoch": 1.9667839319316043, "grad_norm": 0.06335438042879105, "learning_rate": 6.0105867608305555e-06, "loss": 0.0018, "step": 120200 }, { "epoch": 1.9669475578826803, "grad_norm": 0.09197258204221725, "learning_rate": 6.009887532685793e-06, "loss": 0.0018, "step": 120210 }, { "epoch": 1.967111183833756, "grad_norm": 0.03387967497110367, "learning_rate": 6.0091882839502845e-06, "loss": 0.0012, "step": 120220 }, { "epoch": 1.9672748097848318, "grad_norm": 0.006151356268674135, "learning_rate": 6.008489014638284e-06, "loss": 0.0008, "step": 120230 }, { "epoch": 1.9674384357359078, "grad_norm": 0.02344529516994953, "learning_rate": 6.007789724764049e-06, "loss": 0.0012, "step": 120240 }, { "epoch": 1.9676020616869836, "grad_norm": 0.0664188489317894, "learning_rate": 6.007090414341839e-06, "loss": 0.001, "step": 120250 }, { "epoch": 1.9677656876380594, "grad_norm": 0.16511133313179016, "learning_rate": 6.006391083385914e-06, "loss": 0.0022, "step": 120260 }, { "epoch": 1.9679293135891354, "grad_norm": 0.03032892756164074, "learning_rate": 6.005691731910529e-06, "loss": 0.0013, "step": 120270 }, { "epoch": 1.968092939540211, "grad_norm": 0.012369050644338131, "learning_rate": 6.004992359929946e-06, "loss": 0.0006, "step": 120280 }, { "epoch": 1.968256565491287, "grad_norm": 0.03612591326236725, "learning_rate": 6.004292967458422e-06, "loss": 0.0012, "step": 120290 }, { "epoch": 1.968420191442363, "grad_norm": 0.022598236799240112, "learning_rate": 6.00359355451022e-06, "loss": 0.0006, "step": 120300 }, { "epoch": 1.9685838173934385, "grad_norm": 0.028234757483005524, "learning_rate": 6.0028941210995974e-06, "loss": 0.0013, "step": 120310 }, { "epoch": 1.9687474433445145, "grad_norm": 0.0360042005777359, "learning_rate": 6.002194667240818e-06, "loss": 0.0008, "step": 120320 }, { "epoch": 1.9689110692955902, "grad_norm": 0.07282338291406631, "learning_rate": 6.001495192948141e-06, "loss": 0.0012, "step": 120330 }, { "epoch": 1.969074695246666, "grad_norm": 0.03464047238230705, "learning_rate": 6.000795698235828e-06, "loss": 0.0005, "step": 120340 }, { "epoch": 1.969238321197742, "grad_norm": 0.017345814034342766, "learning_rate": 6.000096183118144e-06, "loss": 0.0019, "step": 120350 }, { "epoch": 1.9694019471488178, "grad_norm": 0.21935173869132996, "learning_rate": 5.999396647609348e-06, "loss": 0.0022, "step": 120360 }, { "epoch": 1.9695655730998936, "grad_norm": 0.10554875433444977, "learning_rate": 5.9986970917237045e-06, "loss": 0.0014, "step": 120370 }, { "epoch": 1.9697291990509695, "grad_norm": 0.012680991552770138, "learning_rate": 5.997997515475476e-06, "loss": 0.0025, "step": 120380 }, { "epoch": 1.9698928250020453, "grad_norm": 0.14218463003635406, "learning_rate": 5.997297918878928e-06, "loss": 0.0013, "step": 120390 }, { "epoch": 1.970056450953121, "grad_norm": 0.013385574333369732, "learning_rate": 5.996598301948322e-06, "loss": 0.0021, "step": 120400 }, { "epoch": 1.970220076904197, "grad_norm": 0.05590427666902542, "learning_rate": 5.995898664697927e-06, "loss": 0.0015, "step": 120410 }, { "epoch": 1.9703837028552729, "grad_norm": 0.0032802452333271503, "learning_rate": 5.995199007142002e-06, "loss": 0.0013, "step": 120420 }, { "epoch": 1.9705473288063486, "grad_norm": 0.025091970339417458, "learning_rate": 5.994499329294819e-06, "loss": 0.0012, "step": 120430 }, { "epoch": 1.9707109547574246, "grad_norm": 0.046225327998399734, "learning_rate": 5.993799631170637e-06, "loss": 0.0016, "step": 120440 }, { "epoch": 1.9708745807085004, "grad_norm": 0.052718814462423325, "learning_rate": 5.993099912783727e-06, "loss": 0.0019, "step": 120450 }, { "epoch": 1.9710382066595762, "grad_norm": 0.07267333567142487, "learning_rate": 5.992400174148355e-06, "loss": 0.0011, "step": 120460 }, { "epoch": 1.9712018326106522, "grad_norm": 0.09835490584373474, "learning_rate": 5.991700415278787e-06, "loss": 0.0016, "step": 120470 }, { "epoch": 1.9713654585617277, "grad_norm": 0.018855765461921692, "learning_rate": 5.991000636189292e-06, "loss": 0.0007, "step": 120480 }, { "epoch": 1.9715290845128037, "grad_norm": 0.012952865101397038, "learning_rate": 5.9903008368941366e-06, "loss": 0.0008, "step": 120490 }, { "epoch": 1.9716927104638797, "grad_norm": 0.0431370735168457, "learning_rate": 5.989601017407587e-06, "loss": 0.0015, "step": 120500 }, { "epoch": 1.9718563364149553, "grad_norm": 0.0333053357899189, "learning_rate": 5.988901177743917e-06, "loss": 0.0013, "step": 120510 }, { "epoch": 1.9720199623660313, "grad_norm": 0.03059707023203373, "learning_rate": 5.988201317917392e-06, "loss": 0.0014, "step": 120520 }, { "epoch": 1.972183588317107, "grad_norm": 0.04811149090528488, "learning_rate": 5.9875014379422825e-06, "loss": 0.0014, "step": 120530 }, { "epoch": 1.9723472142681828, "grad_norm": 0.07083598524332047, "learning_rate": 5.986801537832858e-06, "loss": 0.0011, "step": 120540 }, { "epoch": 1.9725108402192588, "grad_norm": 0.0039057198446244, "learning_rate": 5.98610161760339e-06, "loss": 0.0009, "step": 120550 }, { "epoch": 1.9726744661703346, "grad_norm": 0.010278336703777313, "learning_rate": 5.985401677268148e-06, "loss": 0.0013, "step": 120560 }, { "epoch": 1.9728380921214104, "grad_norm": 0.15436801314353943, "learning_rate": 5.984701716841406e-06, "loss": 0.0021, "step": 120570 }, { "epoch": 1.9730017180724864, "grad_norm": 0.031567636877298355, "learning_rate": 5.98400173633743e-06, "loss": 0.001, "step": 120580 }, { "epoch": 1.9731653440235621, "grad_norm": 0.09013398736715317, "learning_rate": 5.983301735770499e-06, "loss": 0.0017, "step": 120590 }, { "epoch": 1.973328969974638, "grad_norm": 0.03134392574429512, "learning_rate": 5.982601715154879e-06, "loss": 0.0025, "step": 120600 }, { "epoch": 1.973492595925714, "grad_norm": 0.03778289631009102, "learning_rate": 5.981901674504847e-06, "loss": 0.0019, "step": 120610 }, { "epoch": 1.9736562218767897, "grad_norm": 0.009876053780317307, "learning_rate": 5.981201613834675e-06, "loss": 0.0003, "step": 120620 }, { "epoch": 1.9738198478278655, "grad_norm": 0.07754971086978912, "learning_rate": 5.980501533158637e-06, "loss": 0.002, "step": 120630 }, { "epoch": 1.9739834737789415, "grad_norm": 0.05478062108159065, "learning_rate": 5.9798014324910055e-06, "loss": 0.001, "step": 120640 }, { "epoch": 1.9741470997300172, "grad_norm": 0.015378721989691257, "learning_rate": 5.9791013118460574e-06, "loss": 0.0014, "step": 120650 }, { "epoch": 1.974310725681093, "grad_norm": 0.031114870682358742, "learning_rate": 5.9784011712380666e-06, "loss": 0.0014, "step": 120660 }, { "epoch": 1.974474351632169, "grad_norm": 0.08300197124481201, "learning_rate": 5.977701010681306e-06, "loss": 0.0012, "step": 120670 }, { "epoch": 1.9746379775832446, "grad_norm": 0.030437540262937546, "learning_rate": 5.977000830190055e-06, "loss": 0.0023, "step": 120680 }, { "epoch": 1.9748016035343205, "grad_norm": 0.0328933447599411, "learning_rate": 5.976300629778586e-06, "loss": 0.002, "step": 120690 }, { "epoch": 1.9749652294853965, "grad_norm": 0.0599110908806324, "learning_rate": 5.97560040946118e-06, "loss": 0.001, "step": 120700 }, { "epoch": 1.975128855436472, "grad_norm": 0.07167226821184158, "learning_rate": 5.9749001692521094e-06, "loss": 0.0015, "step": 120710 }, { "epoch": 1.975292481387548, "grad_norm": 0.01218449417501688, "learning_rate": 5.974199909165654e-06, "loss": 0.0028, "step": 120720 }, { "epoch": 1.9754561073386239, "grad_norm": 0.08212718367576599, "learning_rate": 5.973499629216091e-06, "loss": 0.0004, "step": 120730 }, { "epoch": 1.9756197332896996, "grad_norm": 0.02850750833749771, "learning_rate": 5.972799329417699e-06, "loss": 0.0016, "step": 120740 }, { "epoch": 1.9757833592407756, "grad_norm": 0.04229884222149849, "learning_rate": 5.972099009784755e-06, "loss": 0.0018, "step": 120750 }, { "epoch": 1.9759469851918514, "grad_norm": 0.04720968380570412, "learning_rate": 5.971398670331541e-06, "loss": 0.0009, "step": 120760 }, { "epoch": 1.9761106111429272, "grad_norm": 0.08107700943946838, "learning_rate": 5.970698311072331e-06, "loss": 0.0007, "step": 120770 }, { "epoch": 1.9762742370940032, "grad_norm": 0.04697823151946068, "learning_rate": 5.969997932021411e-06, "loss": 0.001, "step": 120780 }, { "epoch": 1.976437863045079, "grad_norm": 0.09553470462560654, "learning_rate": 5.969297533193057e-06, "loss": 0.001, "step": 120790 }, { "epoch": 1.9766014889961547, "grad_norm": 0.027567528188228607, "learning_rate": 5.968597114601551e-06, "loss": 0.0011, "step": 120800 }, { "epoch": 1.9767651149472307, "grad_norm": 0.08192907273769379, "learning_rate": 5.967896676261172e-06, "loss": 0.0024, "step": 120810 }, { "epoch": 1.9769287408983065, "grad_norm": 0.042459528893232346, "learning_rate": 5.967196218186204e-06, "loss": 0.001, "step": 120820 }, { "epoch": 1.9770923668493823, "grad_norm": 0.06774356216192245, "learning_rate": 5.966495740390928e-06, "loss": 0.002, "step": 120830 }, { "epoch": 1.9772559928004583, "grad_norm": 0.043876100331544876, "learning_rate": 5.965795242889626e-06, "loss": 0.0017, "step": 120840 }, { "epoch": 1.977419618751534, "grad_norm": 0.019719643518328667, "learning_rate": 5.96509472569658e-06, "loss": 0.0013, "step": 120850 }, { "epoch": 1.9775832447026098, "grad_norm": 0.05998596176505089, "learning_rate": 5.964394188826074e-06, "loss": 0.0028, "step": 120860 }, { "epoch": 1.9777468706536858, "grad_norm": 0.04889371618628502, "learning_rate": 5.963693632292389e-06, "loss": 0.0016, "step": 120870 }, { "epoch": 1.9779104966047614, "grad_norm": 0.031017443165183067, "learning_rate": 5.962993056109812e-06, "loss": 0.001, "step": 120880 }, { "epoch": 1.9780741225558374, "grad_norm": 0.06076476350426674, "learning_rate": 5.962292460292624e-06, "loss": 0.0018, "step": 120890 }, { "epoch": 1.9782377485069134, "grad_norm": 0.16194531321525574, "learning_rate": 5.961591844855112e-06, "loss": 0.0009, "step": 120900 }, { "epoch": 1.978401374457989, "grad_norm": 0.16374348104000092, "learning_rate": 5.9608912098115615e-06, "loss": 0.0012, "step": 120910 }, { "epoch": 1.978565000409065, "grad_norm": 0.06888694316148758, "learning_rate": 5.960190555176255e-06, "loss": 0.001, "step": 120920 }, { "epoch": 1.9787286263601407, "grad_norm": 0.053899798542261124, "learning_rate": 5.95948988096348e-06, "loss": 0.0015, "step": 120930 }, { "epoch": 1.9788922523112165, "grad_norm": 0.012800579890608788, "learning_rate": 5.9587891871875215e-06, "loss": 0.0011, "step": 120940 }, { "epoch": 1.9790558782622925, "grad_norm": 0.04272180423140526, "learning_rate": 5.958088473862667e-06, "loss": 0.0009, "step": 120950 }, { "epoch": 1.9792195042133682, "grad_norm": 0.014459795318543911, "learning_rate": 5.957387741003205e-06, "loss": 0.0014, "step": 120960 }, { "epoch": 1.979383130164444, "grad_norm": 0.05283423140645027, "learning_rate": 5.956686988623419e-06, "loss": 0.0021, "step": 120970 }, { "epoch": 1.97954675611552, "grad_norm": 0.05149313807487488, "learning_rate": 5.955986216737602e-06, "loss": 0.0018, "step": 120980 }, { "epoch": 1.9797103820665958, "grad_norm": 0.015637967735528946, "learning_rate": 5.955285425360037e-06, "loss": 0.0008, "step": 120990 }, { "epoch": 1.9798740080176715, "grad_norm": 0.032946180552244186, "learning_rate": 5.954584614505015e-06, "loss": 0.0025, "step": 121000 }, { "epoch": 1.9800376339687475, "grad_norm": 0.12150261551141739, "learning_rate": 5.9538837841868245e-06, "loss": 0.0018, "step": 121010 }, { "epoch": 1.9802012599198233, "grad_norm": 0.24339458346366882, "learning_rate": 5.953182934419753e-06, "loss": 0.003, "step": 121020 }, { "epoch": 1.980364885870899, "grad_norm": 0.014527656137943268, "learning_rate": 5.952482065218095e-06, "loss": 0.0007, "step": 121030 }, { "epoch": 1.980528511821975, "grad_norm": 0.028303440660238266, "learning_rate": 5.9517811765961365e-06, "loss": 0.0007, "step": 121040 }, { "epoch": 1.9806921377730506, "grad_norm": 0.25868773460388184, "learning_rate": 5.95108026856817e-06, "loss": 0.0011, "step": 121050 }, { "epoch": 1.9808557637241266, "grad_norm": 0.01966514065861702, "learning_rate": 5.950379341148484e-06, "loss": 0.0021, "step": 121060 }, { "epoch": 1.9810193896752026, "grad_norm": 0.04184694215655327, "learning_rate": 5.949678394351372e-06, "loss": 0.0012, "step": 121070 }, { "epoch": 1.9811830156262782, "grad_norm": 0.06740278750658035, "learning_rate": 5.948977428191126e-06, "loss": 0.0009, "step": 121080 }, { "epoch": 1.9813466415773542, "grad_norm": 0.05503866448998451, "learning_rate": 5.948276442682038e-06, "loss": 0.0019, "step": 121090 }, { "epoch": 1.98151026752843, "grad_norm": 0.03882739320397377, "learning_rate": 5.9475754378383985e-06, "loss": 0.001, "step": 121100 }, { "epoch": 1.9816738934795057, "grad_norm": 0.011673104017972946, "learning_rate": 5.9468744136745025e-06, "loss": 0.001, "step": 121110 }, { "epoch": 1.9818375194305817, "grad_norm": 0.03173340484499931, "learning_rate": 5.946173370204642e-06, "loss": 0.001, "step": 121120 }, { "epoch": 1.9820011453816575, "grad_norm": 0.07315175980329514, "learning_rate": 5.9454723074431106e-06, "loss": 0.0011, "step": 121130 }, { "epoch": 1.9821647713327333, "grad_norm": 0.07349228858947754, "learning_rate": 5.944771225404204e-06, "loss": 0.0016, "step": 121140 }, { "epoch": 1.9823283972838093, "grad_norm": 0.022708510980010033, "learning_rate": 5.944070124102216e-06, "loss": 0.0011, "step": 121150 }, { "epoch": 1.982492023234885, "grad_norm": 0.062282830476760864, "learning_rate": 5.943369003551439e-06, "loss": 0.0011, "step": 121160 }, { "epoch": 1.9826556491859608, "grad_norm": 0.10886427015066147, "learning_rate": 5.9426678637661725e-06, "loss": 0.0016, "step": 121170 }, { "epoch": 1.9828192751370368, "grad_norm": 0.0418018139898777, "learning_rate": 5.941966704760709e-06, "loss": 0.0005, "step": 121180 }, { "epoch": 1.9829829010881126, "grad_norm": 0.04348471760749817, "learning_rate": 5.941265526549346e-06, "loss": 0.0018, "step": 121190 }, { "epoch": 1.9831465270391884, "grad_norm": 0.042085789144039154, "learning_rate": 5.9405643291463775e-06, "loss": 0.0022, "step": 121200 }, { "epoch": 1.9833101529902644, "grad_norm": 0.02687765099108219, "learning_rate": 5.939863112566104e-06, "loss": 0.0016, "step": 121210 }, { "epoch": 1.9834737789413401, "grad_norm": 0.04347970336675644, "learning_rate": 5.939161876822821e-06, "loss": 0.0017, "step": 121220 }, { "epoch": 1.983637404892416, "grad_norm": 0.006333755794912577, "learning_rate": 5.938460621930826e-06, "loss": 0.0022, "step": 121230 }, { "epoch": 1.983801030843492, "grad_norm": 0.024189293384552002, "learning_rate": 5.937759347904417e-06, "loss": 0.0011, "step": 121240 }, { "epoch": 1.9839646567945675, "grad_norm": 0.060464683920145035, "learning_rate": 5.937058054757893e-06, "loss": 0.0012, "step": 121250 }, { "epoch": 1.9841282827456435, "grad_norm": 0.008257892914116383, "learning_rate": 5.93635674250555e-06, "loss": 0.0018, "step": 121260 }, { "epoch": 1.9842919086967195, "grad_norm": 0.2678913474082947, "learning_rate": 5.935655411161692e-06, "loss": 0.0025, "step": 121270 }, { "epoch": 1.984455534647795, "grad_norm": 0.06186923012137413, "learning_rate": 5.934954060740614e-06, "loss": 0.003, "step": 121280 }, { "epoch": 1.984619160598871, "grad_norm": 0.020404962822794914, "learning_rate": 5.934252691256619e-06, "loss": 0.0011, "step": 121290 }, { "epoch": 1.9847827865499468, "grad_norm": 0.1068066954612732, "learning_rate": 5.9335513027240065e-06, "loss": 0.0029, "step": 121300 }, { "epoch": 1.9849464125010226, "grad_norm": 0.02288714423775673, "learning_rate": 5.932849895157076e-06, "loss": 0.0008, "step": 121310 }, { "epoch": 1.9851100384520985, "grad_norm": 0.11028242111206055, "learning_rate": 5.932148468570129e-06, "loss": 0.0014, "step": 121320 }, { "epoch": 1.9852736644031743, "grad_norm": 0.03514254838228226, "learning_rate": 5.931447022977468e-06, "loss": 0.0024, "step": 121330 }, { "epoch": 1.98543729035425, "grad_norm": 0.03993914648890495, "learning_rate": 5.930745558393396e-06, "loss": 0.0008, "step": 121340 }, { "epoch": 1.985600916305326, "grad_norm": 0.02114560827612877, "learning_rate": 5.930044074832211e-06, "loss": 0.0009, "step": 121350 }, { "epoch": 1.9857645422564019, "grad_norm": 0.0732840970158577, "learning_rate": 5.92934257230822e-06, "loss": 0.0022, "step": 121360 }, { "epoch": 1.9859281682074776, "grad_norm": 0.037652090191841125, "learning_rate": 5.9286410508357224e-06, "loss": 0.0007, "step": 121370 }, { "epoch": 1.9860917941585536, "grad_norm": 0.04145767167210579, "learning_rate": 5.927939510429026e-06, "loss": 0.0012, "step": 121380 }, { "epoch": 1.9862554201096294, "grad_norm": 0.011120164766907692, "learning_rate": 5.927237951102432e-06, "loss": 0.0011, "step": 121390 }, { "epoch": 1.9864190460607052, "grad_norm": 0.04396144300699234, "learning_rate": 5.926536372870243e-06, "loss": 0.0012, "step": 121400 }, { "epoch": 1.9865826720117812, "grad_norm": 0.016230957582592964, "learning_rate": 5.925834775746766e-06, "loss": 0.0009, "step": 121410 }, { "epoch": 1.986746297962857, "grad_norm": 0.05336958169937134, "learning_rate": 5.925133159746305e-06, "loss": 0.0018, "step": 121420 }, { "epoch": 1.9869099239139327, "grad_norm": 0.14094124734401703, "learning_rate": 5.9244315248831665e-06, "loss": 0.0018, "step": 121430 }, { "epoch": 1.9870735498650087, "grad_norm": 0.008504750207066536, "learning_rate": 5.9237298711716565e-06, "loss": 0.0011, "step": 121440 }, { "epoch": 1.9872371758160843, "grad_norm": 0.0056702932342886925, "learning_rate": 5.9230281986260775e-06, "loss": 0.001, "step": 121450 }, { "epoch": 1.9874008017671603, "grad_norm": 0.05062565952539444, "learning_rate": 5.92232650726074e-06, "loss": 0.0009, "step": 121460 }, { "epoch": 1.9875644277182363, "grad_norm": 0.1001640036702156, "learning_rate": 5.921624797089948e-06, "loss": 0.0007, "step": 121470 }, { "epoch": 1.9877280536693118, "grad_norm": 0.07670793682336807, "learning_rate": 5.920923068128013e-06, "loss": 0.0015, "step": 121480 }, { "epoch": 1.9878916796203878, "grad_norm": 0.028555814176797867, "learning_rate": 5.920221320389237e-06, "loss": 0.0012, "step": 121490 }, { "epoch": 1.9880553055714636, "grad_norm": 0.06886592507362366, "learning_rate": 5.919519553887933e-06, "loss": 0.001, "step": 121500 }, { "epoch": 1.9882189315225394, "grad_norm": 0.044380418956279755, "learning_rate": 5.918817768638405e-06, "loss": 0.0013, "step": 121510 }, { "epoch": 1.9883825574736154, "grad_norm": 0.05111803114414215, "learning_rate": 5.9181159646549666e-06, "loss": 0.0011, "step": 121520 }, { "epoch": 1.9885461834246911, "grad_norm": 0.042232319712638855, "learning_rate": 5.9174141419519226e-06, "loss": 0.0007, "step": 121530 }, { "epoch": 1.988709809375767, "grad_norm": 0.033283501863479614, "learning_rate": 5.916712300543586e-06, "loss": 0.001, "step": 121540 }, { "epoch": 1.988873435326843, "grad_norm": 0.0849536880850792, "learning_rate": 5.916010440444264e-06, "loss": 0.0008, "step": 121550 }, { "epoch": 1.9890370612779187, "grad_norm": 0.05733587220311165, "learning_rate": 5.915308561668269e-06, "loss": 0.0016, "step": 121560 }, { "epoch": 1.9892006872289945, "grad_norm": 0.017530418932437897, "learning_rate": 5.914606664229911e-06, "loss": 0.0024, "step": 121570 }, { "epoch": 1.9893643131800705, "grad_norm": 0.07754581421613693, "learning_rate": 5.913904748143501e-06, "loss": 0.0012, "step": 121580 }, { "epoch": 1.9895279391311462, "grad_norm": 0.04351446032524109, "learning_rate": 5.913202813423349e-06, "loss": 0.0012, "step": 121590 }, { "epoch": 1.989691565082222, "grad_norm": 0.0796041488647461, "learning_rate": 5.91250086008377e-06, "loss": 0.0016, "step": 121600 }, { "epoch": 1.989855191033298, "grad_norm": 0.010440152138471603, "learning_rate": 5.911798888139073e-06, "loss": 0.0012, "step": 121610 }, { "epoch": 1.9900188169843738, "grad_norm": 0.05852888152003288, "learning_rate": 5.911096897603573e-06, "loss": 0.0013, "step": 121620 }, { "epoch": 1.9901824429354495, "grad_norm": 0.037827491760253906, "learning_rate": 5.910394888491583e-06, "loss": 0.0017, "step": 121630 }, { "epoch": 1.9903460688865255, "grad_norm": 0.0026948763988912106, "learning_rate": 5.909692860817413e-06, "loss": 0.002, "step": 121640 }, { "epoch": 1.990509694837601, "grad_norm": 0.023798203095793724, "learning_rate": 5.9089908145953824e-06, "loss": 0.001, "step": 121650 }, { "epoch": 1.990673320788677, "grad_norm": 0.05089031532406807, "learning_rate": 5.908288749839799e-06, "loss": 0.0012, "step": 121660 }, { "epoch": 1.990836946739753, "grad_norm": 0.349528968334198, "learning_rate": 5.9075866665649825e-06, "loss": 0.0006, "step": 121670 }, { "epoch": 1.9910005726908286, "grad_norm": 0.05820062384009361, "learning_rate": 5.906884564785245e-06, "loss": 0.0013, "step": 121680 }, { "epoch": 1.9911641986419046, "grad_norm": 0.07424446940422058, "learning_rate": 5.906182444514903e-06, "loss": 0.0017, "step": 121690 }, { "epoch": 1.9913278245929804, "grad_norm": 0.03107602149248123, "learning_rate": 5.90548030576827e-06, "loss": 0.0007, "step": 121700 }, { "epoch": 1.9914914505440562, "grad_norm": 0.026145290583372116, "learning_rate": 5.9047781485596666e-06, "loss": 0.0013, "step": 121710 }, { "epoch": 1.9916550764951322, "grad_norm": 0.019694194197654724, "learning_rate": 5.9040759729034034e-06, "loss": 0.0014, "step": 121720 }, { "epoch": 1.991818702446208, "grad_norm": 0.032699864357709885, "learning_rate": 5.903373778813803e-06, "loss": 0.0013, "step": 121730 }, { "epoch": 1.9919823283972837, "grad_norm": 0.05439327284693718, "learning_rate": 5.902671566305177e-06, "loss": 0.0009, "step": 121740 }, { "epoch": 1.9921459543483597, "grad_norm": 0.02685769833624363, "learning_rate": 5.901969335391846e-06, "loss": 0.0028, "step": 121750 }, { "epoch": 1.9923095802994355, "grad_norm": 0.0984416976571083, "learning_rate": 5.901267086088127e-06, "loss": 0.0015, "step": 121760 }, { "epoch": 1.9924732062505113, "grad_norm": 0.022787591442465782, "learning_rate": 5.90056481840834e-06, "loss": 0.0016, "step": 121770 }, { "epoch": 1.9926368322015873, "grad_norm": 0.03630421683192253, "learning_rate": 5.899862532366801e-06, "loss": 0.0016, "step": 121780 }, { "epoch": 1.992800458152663, "grad_norm": 0.008872362785041332, "learning_rate": 5.899160227977832e-06, "loss": 0.0015, "step": 121790 }, { "epoch": 1.9929640841037388, "grad_norm": 0.01441967487335205, "learning_rate": 5.8984579052557485e-06, "loss": 0.0006, "step": 121800 }, { "epoch": 1.9931277100548148, "grad_norm": 0.06691376864910126, "learning_rate": 5.897755564214874e-06, "loss": 0.0015, "step": 121810 }, { "epoch": 1.9932913360058904, "grad_norm": 0.037066295742988586, "learning_rate": 5.897053204869526e-06, "loss": 0.0018, "step": 121820 }, { "epoch": 1.9934549619569664, "grad_norm": 0.17607396841049194, "learning_rate": 5.896350827234027e-06, "loss": 0.0013, "step": 121830 }, { "epoch": 1.9936185879080424, "grad_norm": 0.03949511796236038, "learning_rate": 5.895648431322696e-06, "loss": 0.0006, "step": 121840 }, { "epoch": 1.993782213859118, "grad_norm": 0.05142595246434212, "learning_rate": 5.894946017149856e-06, "loss": 0.0025, "step": 121850 }, { "epoch": 1.993945839810194, "grad_norm": 0.02285982109606266, "learning_rate": 5.894243584729827e-06, "loss": 0.0011, "step": 121860 }, { "epoch": 1.9941094657612697, "grad_norm": 0.038655467331409454, "learning_rate": 5.8935411340769325e-06, "loss": 0.0013, "step": 121870 }, { "epoch": 1.9942730917123455, "grad_norm": 0.04350844398140907, "learning_rate": 5.892838665205493e-06, "loss": 0.0016, "step": 121880 }, { "epoch": 1.9944367176634215, "grad_norm": 0.046397630125284195, "learning_rate": 5.8921361781298345e-06, "loss": 0.0016, "step": 121890 }, { "epoch": 1.9946003436144972, "grad_norm": 0.11642362177371979, "learning_rate": 5.891433672864276e-06, "loss": 0.0019, "step": 121900 }, { "epoch": 1.994763969565573, "grad_norm": 0.06329454481601715, "learning_rate": 5.890731149423144e-06, "loss": 0.0015, "step": 121910 }, { "epoch": 1.994927595516649, "grad_norm": 0.03363262489438057, "learning_rate": 5.890028607820761e-06, "loss": 0.0012, "step": 121920 }, { "epoch": 1.9950912214677248, "grad_norm": 0.11763675510883331, "learning_rate": 5.889326048071453e-06, "loss": 0.0016, "step": 121930 }, { "epoch": 1.9952548474188005, "grad_norm": 0.07185167074203491, "learning_rate": 5.888623470189542e-06, "loss": 0.0013, "step": 121940 }, { "epoch": 1.9954184733698765, "grad_norm": 0.04546946659684181, "learning_rate": 5.887920874189354e-06, "loss": 0.0013, "step": 121950 }, { "epoch": 1.9955820993209523, "grad_norm": 0.05979829281568527, "learning_rate": 5.887218260085216e-06, "loss": 0.002, "step": 121960 }, { "epoch": 1.995745725272028, "grad_norm": 0.01613023318350315, "learning_rate": 5.886515627891452e-06, "loss": 0.0005, "step": 121970 }, { "epoch": 1.995909351223104, "grad_norm": 0.042127300053834915, "learning_rate": 5.885812977622386e-06, "loss": 0.001, "step": 121980 }, { "epoch": 1.9960729771741799, "grad_norm": 0.03596227616071701, "learning_rate": 5.885110309292349e-06, "loss": 0.0008, "step": 121990 }, { "epoch": 1.9962366031252556, "grad_norm": 0.1402198076248169, "learning_rate": 5.8844076229156645e-06, "loss": 0.0014, "step": 122000 }, { "epoch": 1.9962366031252556, "eval_loss": 0.0009761345572769642, "eval_runtime": 5.3351, "eval_samples_per_second": 37.488, "eval_steps_per_second": 9.372, "step": 122000 }, { "epoch": 1.9964002290763316, "grad_norm": 0.06061146408319473, "learning_rate": 5.883704918506661e-06, "loss": 0.0015, "step": 122010 }, { "epoch": 1.9965638550274072, "grad_norm": 0.04062331095337868, "learning_rate": 5.883002196079665e-06, "loss": 0.0014, "step": 122020 }, { "epoch": 1.9967274809784832, "grad_norm": 0.0760769248008728, "learning_rate": 5.8822994556490075e-06, "loss": 0.0018, "step": 122030 }, { "epoch": 1.9968911069295592, "grad_norm": 0.054715465754270554, "learning_rate": 5.881596697229011e-06, "loss": 0.0019, "step": 122040 }, { "epoch": 1.9970547328806347, "grad_norm": 0.12097571790218353, "learning_rate": 5.880893920834011e-06, "loss": 0.002, "step": 122050 }, { "epoch": 1.9972183588317107, "grad_norm": 0.0037964265793561935, "learning_rate": 5.880191126478332e-06, "loss": 0.001, "step": 122060 }, { "epoch": 1.9973819847827865, "grad_norm": 0.05333339050412178, "learning_rate": 5.879488314176304e-06, "loss": 0.0031, "step": 122070 }, { "epoch": 1.9975456107338623, "grad_norm": 0.01003317628055811, "learning_rate": 5.8787854839422576e-06, "loss": 0.0014, "step": 122080 }, { "epoch": 1.9977092366849383, "grad_norm": 0.12223179638385773, "learning_rate": 5.878082635790523e-06, "loss": 0.0012, "step": 122090 }, { "epoch": 1.997872862636014, "grad_norm": 0.10575570166110992, "learning_rate": 5.877379769735429e-06, "loss": 0.0012, "step": 122100 }, { "epoch": 1.9980364885870898, "grad_norm": 0.06177161633968353, "learning_rate": 5.876676885791308e-06, "loss": 0.0008, "step": 122110 }, { "epoch": 1.9982001145381658, "grad_norm": 0.03858451172709465, "learning_rate": 5.8759739839724916e-06, "loss": 0.0012, "step": 122120 }, { "epoch": 1.9983637404892416, "grad_norm": 0.013639613054692745, "learning_rate": 5.875271064293311e-06, "loss": 0.001, "step": 122130 }, { "epoch": 1.9985273664403174, "grad_norm": 0.06919161230325699, "learning_rate": 5.8745681267680965e-06, "loss": 0.0011, "step": 122140 }, { "epoch": 1.9986909923913934, "grad_norm": 0.06315713375806808, "learning_rate": 5.873865171411182e-06, "loss": 0.0013, "step": 122150 }, { "epoch": 1.9988546183424691, "grad_norm": 0.04355355724692345, "learning_rate": 5.873162198236901e-06, "loss": 0.0015, "step": 122160 }, { "epoch": 1.999018244293545, "grad_norm": 0.020160697400569916, "learning_rate": 5.872459207259584e-06, "loss": 0.0017, "step": 122170 }, { "epoch": 1.999181870244621, "grad_norm": 0.03933744505047798, "learning_rate": 5.871756198493567e-06, "loss": 0.0023, "step": 122180 }, { "epoch": 1.9993454961956967, "grad_norm": 0.04582236334681511, "learning_rate": 5.871053171953181e-06, "loss": 0.001, "step": 122190 }, { "epoch": 1.9995091221467725, "grad_norm": 0.07160431146621704, "learning_rate": 5.870350127652763e-06, "loss": 0.0013, "step": 122200 }, { "epoch": 1.9996727480978485, "grad_norm": 0.033832523971796036, "learning_rate": 5.8696470656066455e-06, "loss": 0.0014, "step": 122210 }, { "epoch": 1.999836374048924, "grad_norm": 0.02695462666451931, "learning_rate": 5.868943985829163e-06, "loss": 0.0005, "step": 122220 }, { "epoch": 2.0, "grad_norm": 0.004032541066408157, "learning_rate": 5.8682408883346535e-06, "loss": 0.0006, "step": 122230 }, { "epoch": 2.000163625951076, "grad_norm": 0.01209216471761465, "learning_rate": 5.86753777313745e-06, "loss": 0.0006, "step": 122240 }, { "epoch": 2.0003272519021515, "grad_norm": 0.028562862426042557, "learning_rate": 5.866834640251889e-06, "loss": 0.0013, "step": 122250 }, { "epoch": 2.0004908778532275, "grad_norm": 0.039652373641729355, "learning_rate": 5.866131489692306e-06, "loss": 0.001, "step": 122260 }, { "epoch": 2.0006545038043035, "grad_norm": 0.02617492526769638, "learning_rate": 5.865428321473039e-06, "loss": 0.0014, "step": 122270 }, { "epoch": 2.000818129755379, "grad_norm": 0.042269911617040634, "learning_rate": 5.864725135608426e-06, "loss": 0.0017, "step": 122280 }, { "epoch": 2.000981755706455, "grad_norm": 0.05553917586803436, "learning_rate": 5.8640219321128e-06, "loss": 0.001, "step": 122290 }, { "epoch": 2.001145381657531, "grad_norm": 0.0012463605962693691, "learning_rate": 5.863318711000505e-06, "loss": 0.0013, "step": 122300 }, { "epoch": 2.0013090076086066, "grad_norm": 0.06778084486722946, "learning_rate": 5.862615472285873e-06, "loss": 0.0012, "step": 122310 }, { "epoch": 2.0014726335596826, "grad_norm": 0.008668642491102219, "learning_rate": 5.861912215983247e-06, "loss": 0.0011, "step": 122320 }, { "epoch": 2.001636259510758, "grad_norm": 0.03888960927724838, "learning_rate": 5.861208942106963e-06, "loss": 0.0032, "step": 122330 }, { "epoch": 2.001799885461834, "grad_norm": 0.021158607676625252, "learning_rate": 5.860505650671362e-06, "loss": 0.0009, "step": 122340 }, { "epoch": 2.00196351141291, "grad_norm": 0.1024215817451477, "learning_rate": 5.859802341690781e-06, "loss": 0.0011, "step": 122350 }, { "epoch": 2.0021271373639857, "grad_norm": 0.0013140636729076505, "learning_rate": 5.859099015179563e-06, "loss": 0.0006, "step": 122360 }, { "epoch": 2.0022907633150617, "grad_norm": 0.054518815129995346, "learning_rate": 5.858395671152045e-06, "loss": 0.0014, "step": 122370 }, { "epoch": 2.0024543892661377, "grad_norm": 0.04088488593697548, "learning_rate": 5.8576923096225715e-06, "loss": 0.0006, "step": 122380 }, { "epoch": 2.0026180152172133, "grad_norm": 0.05119670554995537, "learning_rate": 5.85698893060548e-06, "loss": 0.001, "step": 122390 }, { "epoch": 2.0027816411682893, "grad_norm": 0.036075517535209656, "learning_rate": 5.856285534115114e-06, "loss": 0.0015, "step": 122400 }, { "epoch": 2.0029452671193653, "grad_norm": 0.0666375383734703, "learning_rate": 5.855582120165813e-06, "loss": 0.0007, "step": 122410 }, { "epoch": 2.003108893070441, "grad_norm": 0.012219875119626522, "learning_rate": 5.854878688771921e-06, "loss": 0.0005, "step": 122420 }, { "epoch": 2.003272519021517, "grad_norm": 0.025291454046964645, "learning_rate": 5.854175239947779e-06, "loss": 0.0008, "step": 122430 }, { "epoch": 2.003436144972593, "grad_norm": 0.06012621521949768, "learning_rate": 5.853471773707731e-06, "loss": 0.0007, "step": 122440 }, { "epoch": 2.0035997709236684, "grad_norm": 0.011514756828546524, "learning_rate": 5.852768290066119e-06, "loss": 0.001, "step": 122450 }, { "epoch": 2.0037633968747444, "grad_norm": 0.03853663429617882, "learning_rate": 5.852064789037286e-06, "loss": 0.0017, "step": 122460 }, { "epoch": 2.0039270228258204, "grad_norm": 0.0010727037442848086, "learning_rate": 5.851361270635578e-06, "loss": 0.001, "step": 122470 }, { "epoch": 2.004090648776896, "grad_norm": 0.013119262643158436, "learning_rate": 5.8506577348753365e-06, "loss": 0.0016, "step": 122480 }, { "epoch": 2.004254274727972, "grad_norm": 0.008640491403639317, "learning_rate": 5.849954181770908e-06, "loss": 0.0007, "step": 122490 }, { "epoch": 2.004417900679048, "grad_norm": 0.04031805694103241, "learning_rate": 5.849250611336637e-06, "loss": 0.0006, "step": 122500 }, { "epoch": 2.0045815266301235, "grad_norm": 0.04919197037816048, "learning_rate": 5.848547023586868e-06, "loss": 0.0018, "step": 122510 }, { "epoch": 2.0047451525811995, "grad_norm": 0.008489163592457771, "learning_rate": 5.847843418535946e-06, "loss": 0.0009, "step": 122520 }, { "epoch": 2.004908778532275, "grad_norm": 0.10990001261234283, "learning_rate": 5.847139796198219e-06, "loss": 0.002, "step": 122530 }, { "epoch": 2.005072404483351, "grad_norm": 0.012465324252843857, "learning_rate": 5.846436156588031e-06, "loss": 0.0007, "step": 122540 }, { "epoch": 2.005236030434427, "grad_norm": 0.014378277584910393, "learning_rate": 5.845732499719731e-06, "loss": 0.0012, "step": 122550 }, { "epoch": 2.0053996563855025, "grad_norm": 0.05551069229841232, "learning_rate": 5.845028825607663e-06, "loss": 0.0007, "step": 122560 }, { "epoch": 2.0055632823365785, "grad_norm": 0.0034286065492779016, "learning_rate": 5.844325134266177e-06, "loss": 0.0006, "step": 122570 }, { "epoch": 2.0057269082876545, "grad_norm": 0.2014462947845459, "learning_rate": 5.843621425709618e-06, "loss": 0.0011, "step": 122580 }, { "epoch": 2.00589053423873, "grad_norm": 0.03697923198342323, "learning_rate": 5.842917699952337e-06, "loss": 0.002, "step": 122590 }, { "epoch": 2.006054160189806, "grad_norm": 0.022050898522138596, "learning_rate": 5.84221395700868e-06, "loss": 0.0012, "step": 122600 }, { "epoch": 2.006217786140882, "grad_norm": 0.05537615716457367, "learning_rate": 5.841510196892999e-06, "loss": 0.0011, "step": 122610 }, { "epoch": 2.0063814120919576, "grad_norm": 0.0056890519335865974, "learning_rate": 5.840806419619638e-06, "loss": 0.0007, "step": 122620 }, { "epoch": 2.0065450380430336, "grad_norm": 0.09133002907037735, "learning_rate": 5.8401026252029515e-06, "loss": 0.0006, "step": 122630 }, { "epoch": 2.0067086639941096, "grad_norm": 0.016698282212018967, "learning_rate": 5.839398813657284e-06, "loss": 0.0007, "step": 122640 }, { "epoch": 2.006872289945185, "grad_norm": 0.06941226869821548, "learning_rate": 5.838694984996992e-06, "loss": 0.0008, "step": 122650 }, { "epoch": 2.007035915896261, "grad_norm": 0.0023062778636813164, "learning_rate": 5.83799113923642e-06, "loss": 0.0009, "step": 122660 }, { "epoch": 2.007199541847337, "grad_norm": 0.08168697357177734, "learning_rate": 5.837287276389922e-06, "loss": 0.0007, "step": 122670 }, { "epoch": 2.0073631677984127, "grad_norm": 0.2522999346256256, "learning_rate": 5.836583396471849e-06, "loss": 0.0022, "step": 122680 }, { "epoch": 2.0075267937494887, "grad_norm": 0.015178371220827103, "learning_rate": 5.8358794994965525e-06, "loss": 0.0006, "step": 122690 }, { "epoch": 2.0076904197005647, "grad_norm": 0.00826078001409769, "learning_rate": 5.835175585478383e-06, "loss": 0.0007, "step": 122700 }, { "epoch": 2.0078540456516403, "grad_norm": 0.06699009239673615, "learning_rate": 5.834471654431693e-06, "loss": 0.0009, "step": 122710 }, { "epoch": 2.0080176716027163, "grad_norm": 0.1166653260588646, "learning_rate": 5.8337677063708364e-06, "loss": 0.001, "step": 122720 }, { "epoch": 2.008181297553792, "grad_norm": 0.01755117066204548, "learning_rate": 5.833063741310166e-06, "loss": 0.0006, "step": 122730 }, { "epoch": 2.008344923504868, "grad_norm": 0.05072310194373131, "learning_rate": 5.832359759264034e-06, "loss": 0.0007, "step": 122740 }, { "epoch": 2.008508549455944, "grad_norm": 0.02108931727707386, "learning_rate": 5.831655760246795e-06, "loss": 0.0011, "step": 122750 }, { "epoch": 2.0086721754070194, "grad_norm": 0.06415762007236481, "learning_rate": 5.830951744272801e-06, "loss": 0.001, "step": 122760 }, { "epoch": 2.0088358013580954, "grad_norm": 0.00296503072604537, "learning_rate": 5.830247711356408e-06, "loss": 0.0006, "step": 122770 }, { "epoch": 2.0089994273091714, "grad_norm": 0.029032496735453606, "learning_rate": 5.829543661511972e-06, "loss": 0.0007, "step": 122780 }, { "epoch": 2.009163053260247, "grad_norm": 0.05128752440214157, "learning_rate": 5.8288395947538435e-06, "loss": 0.0024, "step": 122790 }, { "epoch": 2.009326679211323, "grad_norm": 0.02161572128534317, "learning_rate": 5.828135511096382e-06, "loss": 0.0007, "step": 122800 }, { "epoch": 2.009490305162399, "grad_norm": 0.04583371803164482, "learning_rate": 5.827431410553942e-06, "loss": 0.0017, "step": 122810 }, { "epoch": 2.0096539311134745, "grad_norm": 0.015587142668664455, "learning_rate": 5.826727293140879e-06, "loss": 0.0011, "step": 122820 }, { "epoch": 2.0098175570645505, "grad_norm": 0.04189873859286308, "learning_rate": 5.826023158871549e-06, "loss": 0.0007, "step": 122830 }, { "epoch": 2.0099811830156264, "grad_norm": 0.01407738495618105, "learning_rate": 5.82531900776031e-06, "loss": 0.001, "step": 122840 }, { "epoch": 2.010144808966702, "grad_norm": 0.019575590267777443, "learning_rate": 5.824614839821517e-06, "loss": 0.0006, "step": 122850 }, { "epoch": 2.010308434917778, "grad_norm": 0.0896303728222847, "learning_rate": 5.823910655069531e-06, "loss": 0.0007, "step": 122860 }, { "epoch": 2.010472060868854, "grad_norm": 0.08632038533687592, "learning_rate": 5.823206453518706e-06, "loss": 0.001, "step": 122870 }, { "epoch": 2.0106356868199295, "grad_norm": 0.051003046333789825, "learning_rate": 5.822502235183402e-06, "loss": 0.0008, "step": 122880 }, { "epoch": 2.0107993127710055, "grad_norm": 0.11879287660121918, "learning_rate": 5.821798000077976e-06, "loss": 0.0014, "step": 122890 }, { "epoch": 2.0109629387220815, "grad_norm": 0.009473982267081738, "learning_rate": 5.821093748216789e-06, "loss": 0.0009, "step": 122900 }, { "epoch": 2.011126564673157, "grad_norm": 0.02348344400525093, "learning_rate": 5.820389479614198e-06, "loss": 0.001, "step": 122910 }, { "epoch": 2.011290190624233, "grad_norm": 0.123201884329319, "learning_rate": 5.819685194284563e-06, "loss": 0.0019, "step": 122920 }, { "epoch": 2.0114538165753086, "grad_norm": 0.3114264905452728, "learning_rate": 5.818980892242244e-06, "loss": 0.0166, "step": 122930 }, { "epoch": 2.0116174425263846, "grad_norm": 0.02818281389772892, "learning_rate": 5.818276573501602e-06, "loss": 0.001, "step": 122940 }, { "epoch": 2.0117810684774606, "grad_norm": 0.040008388459682465, "learning_rate": 5.8175722380769964e-06, "loss": 0.0031, "step": 122950 }, { "epoch": 2.011944694428536, "grad_norm": 0.0508110448718071, "learning_rate": 5.816867885982787e-06, "loss": 0.0007, "step": 122960 }, { "epoch": 2.012108320379612, "grad_norm": 0.04766111448407173, "learning_rate": 5.816163517233337e-06, "loss": 0.0009, "step": 122970 }, { "epoch": 2.012271946330688, "grad_norm": 0.06910490989685059, "learning_rate": 5.815459131843008e-06, "loss": 0.001, "step": 122980 }, { "epoch": 2.0124355722817637, "grad_norm": 0.052790284156799316, "learning_rate": 5.81475472982616e-06, "loss": 0.0017, "step": 122990 }, { "epoch": 2.0125991982328397, "grad_norm": 0.013515307568013668, "learning_rate": 5.814050311197156e-06, "loss": 0.0006, "step": 123000 }, { "epoch": 2.0127628241839157, "grad_norm": 0.020221002399921417, "learning_rate": 5.813345875970358e-06, "loss": 0.0013, "step": 123010 }, { "epoch": 2.0129264501349913, "grad_norm": 0.005036111455410719, "learning_rate": 5.81264142416013e-06, "loss": 0.0005, "step": 123020 }, { "epoch": 2.0130900760860673, "grad_norm": 0.012942143715918064, "learning_rate": 5.811936955780834e-06, "loss": 0.0006, "step": 123030 }, { "epoch": 2.0132537020371433, "grad_norm": 0.023368241265416145, "learning_rate": 5.811232470846834e-06, "loss": 0.0008, "step": 123040 }, { "epoch": 2.013417327988219, "grad_norm": 0.24185527861118317, "learning_rate": 5.810527969372494e-06, "loss": 0.001, "step": 123050 }, { "epoch": 2.013580953939295, "grad_norm": 0.011277283541858196, "learning_rate": 5.809823451372178e-06, "loss": 0.0011, "step": 123060 }, { "epoch": 2.013744579890371, "grad_norm": 0.037130557000637054, "learning_rate": 5.809118916860251e-06, "loss": 0.0008, "step": 123070 }, { "epoch": 2.0139082058414464, "grad_norm": 0.029541604220867157, "learning_rate": 5.808414365851076e-06, "loss": 0.0008, "step": 123080 }, { "epoch": 2.0140718317925224, "grad_norm": 0.040692392736673355, "learning_rate": 5.807709798359021e-06, "loss": 0.0012, "step": 123090 }, { "epoch": 2.014235457743598, "grad_norm": 0.06997204571962357, "learning_rate": 5.80700521439845e-06, "loss": 0.0009, "step": 123100 }, { "epoch": 2.014399083694674, "grad_norm": 0.0018488302594050765, "learning_rate": 5.806300613983728e-06, "loss": 0.0013, "step": 123110 }, { "epoch": 2.01456270964575, "grad_norm": 0.04050973802804947, "learning_rate": 5.805595997129223e-06, "loss": 0.001, "step": 123120 }, { "epoch": 2.0147263355968255, "grad_norm": 0.032119020819664, "learning_rate": 5.804891363849302e-06, "loss": 0.0007, "step": 123130 }, { "epoch": 2.0148899615479015, "grad_norm": 0.0066593908704817295, "learning_rate": 5.804186714158328e-06, "loss": 0.0003, "step": 123140 }, { "epoch": 2.0150535874989774, "grad_norm": 0.0911034420132637, "learning_rate": 5.8034820480706736e-06, "loss": 0.0009, "step": 123150 }, { "epoch": 2.015217213450053, "grad_norm": 0.06186807155609131, "learning_rate": 5.802777365600701e-06, "loss": 0.001, "step": 123160 }, { "epoch": 2.015380839401129, "grad_norm": 0.028937771916389465, "learning_rate": 5.8020726667627824e-06, "loss": 0.0009, "step": 123170 }, { "epoch": 2.015544465352205, "grad_norm": 0.13856421411037445, "learning_rate": 5.801367951571283e-06, "loss": 0.0016, "step": 123180 }, { "epoch": 2.0157080913032805, "grad_norm": 0.12211834639310837, "learning_rate": 5.800663220040575e-06, "loss": 0.0022, "step": 123190 }, { "epoch": 2.0158717172543565, "grad_norm": 0.021558530628681183, "learning_rate": 5.799958472185021e-06, "loss": 0.0006, "step": 123200 }, { "epoch": 2.0160353432054325, "grad_norm": 0.1018480584025383, "learning_rate": 5.799253708018997e-06, "loss": 0.0014, "step": 123210 }, { "epoch": 2.016198969156508, "grad_norm": 0.07821517437696457, "learning_rate": 5.798548927556869e-06, "loss": 0.0008, "step": 123220 }, { "epoch": 2.016362595107584, "grad_norm": 0.17954936623573303, "learning_rate": 5.797844130813007e-06, "loss": 0.0016, "step": 123230 }, { "epoch": 2.01652622105866, "grad_norm": 0.013683894649147987, "learning_rate": 5.797139317801782e-06, "loss": 0.0008, "step": 123240 }, { "epoch": 2.0166898470097356, "grad_norm": 0.06114475801587105, "learning_rate": 5.796434488537566e-06, "loss": 0.0009, "step": 123250 }, { "epoch": 2.0168534729608116, "grad_norm": 0.06848076730966568, "learning_rate": 5.795729643034725e-06, "loss": 0.0007, "step": 123260 }, { "epoch": 2.0170170989118876, "grad_norm": 0.029893437400460243, "learning_rate": 5.795024781307635e-06, "loss": 0.0011, "step": 123270 }, { "epoch": 2.017180724862963, "grad_norm": 0.08100614696741104, "learning_rate": 5.7943199033706655e-06, "loss": 0.0013, "step": 123280 }, { "epoch": 2.017344350814039, "grad_norm": 0.06708745658397675, "learning_rate": 5.7936150092381895e-06, "loss": 0.0013, "step": 123290 }, { "epoch": 2.0175079767651147, "grad_norm": 0.02853184938430786, "learning_rate": 5.792910098924577e-06, "loss": 0.0021, "step": 123300 }, { "epoch": 2.0176716027161907, "grad_norm": 0.03339894488453865, "learning_rate": 5.792205172444203e-06, "loss": 0.0007, "step": 123310 }, { "epoch": 2.0178352286672667, "grad_norm": 0.02979980781674385, "learning_rate": 5.791500229811438e-06, "loss": 0.0007, "step": 123320 }, { "epoch": 2.0179988546183423, "grad_norm": 0.5069419145584106, "learning_rate": 5.790795271040659e-06, "loss": 0.0011, "step": 123330 }, { "epoch": 2.0181624805694183, "grad_norm": 0.07607215642929077, "learning_rate": 5.7900902961462345e-06, "loss": 0.0009, "step": 123340 }, { "epoch": 2.0183261065204943, "grad_norm": 0.058537814766168594, "learning_rate": 5.789385305142543e-06, "loss": 0.0012, "step": 123350 }, { "epoch": 2.01848973247157, "grad_norm": 0.1260153204202652, "learning_rate": 5.788680298043954e-06, "loss": 0.0011, "step": 123360 }, { "epoch": 2.018653358422646, "grad_norm": 0.19586466252803802, "learning_rate": 5.787975274864846e-06, "loss": 0.0007, "step": 123370 }, { "epoch": 2.018816984373722, "grad_norm": 0.06720031797885895, "learning_rate": 5.787270235619593e-06, "loss": 0.0006, "step": 123380 }, { "epoch": 2.0189806103247974, "grad_norm": 0.02362598106265068, "learning_rate": 5.786565180322567e-06, "loss": 0.001, "step": 123390 }, { "epoch": 2.0191442362758734, "grad_norm": 0.09611867368221283, "learning_rate": 5.785860108988148e-06, "loss": 0.0014, "step": 123400 }, { "epoch": 2.0193078622269494, "grad_norm": 0.046436816453933716, "learning_rate": 5.7851550216307095e-06, "loss": 0.0011, "step": 123410 }, { "epoch": 2.019471488178025, "grad_norm": 0.12885111570358276, "learning_rate": 5.7844499182646276e-06, "loss": 0.0018, "step": 123420 }, { "epoch": 2.019635114129101, "grad_norm": 0.06819497048854828, "learning_rate": 5.783744798904278e-06, "loss": 0.0009, "step": 123430 }, { "epoch": 2.019798740080177, "grad_norm": 0.03121306747198105, "learning_rate": 5.783039663564041e-06, "loss": 0.0006, "step": 123440 }, { "epoch": 2.0199623660312525, "grad_norm": 0.09220661222934723, "learning_rate": 5.782334512258289e-06, "loss": 0.0009, "step": 123450 }, { "epoch": 2.0201259919823285, "grad_norm": 0.001534288632683456, "learning_rate": 5.781629345001402e-06, "loss": 0.001, "step": 123460 }, { "epoch": 2.0202896179334044, "grad_norm": 0.04409559443593025, "learning_rate": 5.780924161807758e-06, "loss": 0.0017, "step": 123470 }, { "epoch": 2.02045324388448, "grad_norm": 0.0026548609603196383, "learning_rate": 5.780218962691734e-06, "loss": 0.0011, "step": 123480 }, { "epoch": 2.020616869835556, "grad_norm": 0.04210636764764786, "learning_rate": 5.779513747667708e-06, "loss": 0.0011, "step": 123490 }, { "epoch": 2.0207804957866315, "grad_norm": 0.041659481823444366, "learning_rate": 5.778808516750061e-06, "loss": 0.0014, "step": 123500 }, { "epoch": 2.0209441217377075, "grad_norm": 0.018219826743006706, "learning_rate": 5.77810326995317e-06, "loss": 0.001, "step": 123510 }, { "epoch": 2.0211077476887835, "grad_norm": 0.027882639318704605, "learning_rate": 5.777398007291416e-06, "loss": 0.0022, "step": 123520 }, { "epoch": 2.021271373639859, "grad_norm": 0.06312672793865204, "learning_rate": 5.776692728779176e-06, "loss": 0.0004, "step": 123530 }, { "epoch": 2.021434999590935, "grad_norm": 0.08556745201349258, "learning_rate": 5.775987434430834e-06, "loss": 0.0012, "step": 123540 }, { "epoch": 2.021598625542011, "grad_norm": 0.023607825860381126, "learning_rate": 5.775282124260767e-06, "loss": 0.0008, "step": 123550 }, { "epoch": 2.0217622514930866, "grad_norm": 0.09046496450901031, "learning_rate": 5.7745767982833576e-06, "loss": 0.0007, "step": 123560 }, { "epoch": 2.0219258774441626, "grad_norm": 0.06343670934438705, "learning_rate": 5.773871456512985e-06, "loss": 0.0007, "step": 123570 }, { "epoch": 2.0220895033952386, "grad_norm": 0.042992182075977325, "learning_rate": 5.7731660989640324e-06, "loss": 0.001, "step": 123580 }, { "epoch": 2.022253129346314, "grad_norm": 0.057709068059921265, "learning_rate": 5.7724607256508795e-06, "loss": 0.0011, "step": 123590 }, { "epoch": 2.02241675529739, "grad_norm": 0.00759494211524725, "learning_rate": 5.7717553365879105e-06, "loss": 0.001, "step": 123600 }, { "epoch": 2.022580381248466, "grad_norm": 0.058368559926748276, "learning_rate": 5.771049931789505e-06, "loss": 0.0005, "step": 123610 }, { "epoch": 2.0227440071995417, "grad_norm": 0.12100210040807724, "learning_rate": 5.770344511270049e-06, "loss": 0.0012, "step": 123620 }, { "epoch": 2.0229076331506177, "grad_norm": 0.004559995606541634, "learning_rate": 5.769639075043922e-06, "loss": 0.0006, "step": 123630 }, { "epoch": 2.0230712591016937, "grad_norm": 0.052047308534383774, "learning_rate": 5.768933623125511e-06, "loss": 0.0009, "step": 123640 }, { "epoch": 2.0232348850527693, "grad_norm": 0.004008663352578878, "learning_rate": 5.7682281555291955e-06, "loss": 0.001, "step": 123650 }, { "epoch": 2.0233985110038453, "grad_norm": 0.020433707162737846, "learning_rate": 5.767522672269362e-06, "loss": 0.0011, "step": 123660 }, { "epoch": 2.0235621369549213, "grad_norm": 0.0262891948223114, "learning_rate": 5.766817173360393e-06, "loss": 0.0008, "step": 123670 }, { "epoch": 2.023725762905997, "grad_norm": 0.06331874430179596, "learning_rate": 5.766111658816676e-06, "loss": 0.0016, "step": 123680 }, { "epoch": 2.023889388857073, "grad_norm": 0.02953602559864521, "learning_rate": 5.765406128652593e-06, "loss": 0.0009, "step": 123690 }, { "epoch": 2.0240530148081484, "grad_norm": 0.03243729844689369, "learning_rate": 5.7647005828825285e-06, "loss": 0.0007, "step": 123700 }, { "epoch": 2.0242166407592244, "grad_norm": 0.0445563867688179, "learning_rate": 5.763995021520872e-06, "loss": 0.0006, "step": 123710 }, { "epoch": 2.0243802667103004, "grad_norm": 0.052624065428972244, "learning_rate": 5.763289444582005e-06, "loss": 0.0013, "step": 123720 }, { "epoch": 2.024543892661376, "grad_norm": 0.053481489419937134, "learning_rate": 5.762583852080316e-06, "loss": 0.0008, "step": 123730 }, { "epoch": 2.024707518612452, "grad_norm": 0.016321880742907524, "learning_rate": 5.761878244030189e-06, "loss": 0.0011, "step": 123740 }, { "epoch": 2.024871144563528, "grad_norm": 0.02614610083401203, "learning_rate": 5.761172620446015e-06, "loss": 0.0009, "step": 123750 }, { "epoch": 2.0250347705146035, "grad_norm": 0.045975152403116226, "learning_rate": 5.7604669813421765e-06, "loss": 0.0004, "step": 123760 }, { "epoch": 2.0251983964656795, "grad_norm": 0.029098695144057274, "learning_rate": 5.7597613267330655e-06, "loss": 0.0007, "step": 123770 }, { "epoch": 2.0253620224167554, "grad_norm": 0.05481715872883797, "learning_rate": 5.7590556566330645e-06, "loss": 0.0009, "step": 123780 }, { "epoch": 2.025525648367831, "grad_norm": 0.11022494733333588, "learning_rate": 5.758349971056567e-06, "loss": 0.0007, "step": 123790 }, { "epoch": 2.025689274318907, "grad_norm": 0.15609650313854218, "learning_rate": 5.757644270017956e-06, "loss": 0.0011, "step": 123800 }, { "epoch": 2.025852900269983, "grad_norm": 0.016567008569836617, "learning_rate": 5.7569385535316244e-06, "loss": 0.0006, "step": 123810 }, { "epoch": 2.0260165262210585, "grad_norm": 0.032359376549720764, "learning_rate": 5.756232821611958e-06, "loss": 0.0012, "step": 123820 }, { "epoch": 2.0261801521721345, "grad_norm": 0.014817731454968452, "learning_rate": 5.755527074273349e-06, "loss": 0.0005, "step": 123830 }, { "epoch": 2.0263437781232105, "grad_norm": 0.0468674935400486, "learning_rate": 5.7548213115301845e-06, "loss": 0.0005, "step": 123840 }, { "epoch": 2.026507404074286, "grad_norm": 0.00806101132184267, "learning_rate": 5.754115533396857e-06, "loss": 0.0018, "step": 123850 }, { "epoch": 2.026671030025362, "grad_norm": 0.02413199283182621, "learning_rate": 5.753409739887753e-06, "loss": 0.0008, "step": 123860 }, { "epoch": 2.026834655976438, "grad_norm": 0.056515082716941833, "learning_rate": 5.7527039310172676e-06, "loss": 0.0011, "step": 123870 }, { "epoch": 2.0269982819275136, "grad_norm": 0.03387651592493057, "learning_rate": 5.751998106799786e-06, "loss": 0.001, "step": 123880 }, { "epoch": 2.0271619078785896, "grad_norm": 0.09243346005678177, "learning_rate": 5.751292267249705e-06, "loss": 0.0011, "step": 123890 }, { "epoch": 2.027325533829665, "grad_norm": 0.011900302954018116, "learning_rate": 5.750586412381413e-06, "loss": 0.0007, "step": 123900 }, { "epoch": 2.027489159780741, "grad_norm": 0.05328201875090599, "learning_rate": 5.7498805422093016e-06, "loss": 0.0015, "step": 123910 }, { "epoch": 2.027652785731817, "grad_norm": 0.05321149155497551, "learning_rate": 5.749174656747764e-06, "loss": 0.0007, "step": 123920 }, { "epoch": 2.0278164116828927, "grad_norm": 0.1467210054397583, "learning_rate": 5.748468756011194e-06, "loss": 0.0011, "step": 123930 }, { "epoch": 2.0279800376339687, "grad_norm": 0.02277764119207859, "learning_rate": 5.7477628400139796e-06, "loss": 0.0007, "step": 123940 }, { "epoch": 2.0281436635850447, "grad_norm": 0.035381123423576355, "learning_rate": 5.74705690877052e-06, "loss": 0.0014, "step": 123950 }, { "epoch": 2.0283072895361203, "grad_norm": 0.025819256901741028, "learning_rate": 5.746350962295203e-06, "loss": 0.0007, "step": 123960 }, { "epoch": 2.0284709154871963, "grad_norm": 0.056958671659231186, "learning_rate": 5.745645000602425e-06, "loss": 0.0014, "step": 123970 }, { "epoch": 2.0286345414382723, "grad_norm": 0.045148953795433044, "learning_rate": 5.74493902370658e-06, "loss": 0.0006, "step": 123980 }, { "epoch": 2.028798167389348, "grad_norm": 0.05487125739455223, "learning_rate": 5.744233031622063e-06, "loss": 0.0015, "step": 123990 }, { "epoch": 2.028961793340424, "grad_norm": 0.019840169697999954, "learning_rate": 5.743527024363266e-06, "loss": 0.0006, "step": 124000 }, { "epoch": 2.0291254192915, "grad_norm": 0.003431517630815506, "learning_rate": 5.742821001944586e-06, "loss": 0.0007, "step": 124010 }, { "epoch": 2.0292890452425754, "grad_norm": 0.0018975331913679838, "learning_rate": 5.742114964380418e-06, "loss": 0.0004, "step": 124020 }, { "epoch": 2.0294526711936514, "grad_norm": 0.07424990087747574, "learning_rate": 5.741408911685156e-06, "loss": 0.0013, "step": 124030 }, { "epoch": 2.0296162971447274, "grad_norm": 0.012926973402500153, "learning_rate": 5.740702843873197e-06, "loss": 0.002, "step": 124040 }, { "epoch": 2.029779923095803, "grad_norm": 0.0874800980091095, "learning_rate": 5.739996760958937e-06, "loss": 0.0011, "step": 124050 }, { "epoch": 2.029943549046879, "grad_norm": 0.035612497478723526, "learning_rate": 5.7392906629567725e-06, "loss": 0.0007, "step": 124060 }, { "epoch": 2.0301071749979545, "grad_norm": 0.03587617352604866, "learning_rate": 5.7385845498811e-06, "loss": 0.0008, "step": 124070 }, { "epoch": 2.0302708009490305, "grad_norm": 0.0641164779663086, "learning_rate": 5.737878421746317e-06, "loss": 0.0011, "step": 124080 }, { "epoch": 2.0304344269001064, "grad_norm": 0.02568477764725685, "learning_rate": 5.7371722785668195e-06, "loss": 0.0008, "step": 124090 }, { "epoch": 2.030598052851182, "grad_norm": 0.03791767358779907, "learning_rate": 5.736466120357008e-06, "loss": 0.0009, "step": 124100 }, { "epoch": 2.030761678802258, "grad_norm": 0.013966050930321217, "learning_rate": 5.735759947131276e-06, "loss": 0.0006, "step": 124110 }, { "epoch": 2.030925304753334, "grad_norm": 0.019890660420060158, "learning_rate": 5.735053758904028e-06, "loss": 0.0008, "step": 124120 }, { "epoch": 2.0310889307044095, "grad_norm": 0.00234611495397985, "learning_rate": 5.734347555689655e-06, "loss": 0.001, "step": 124130 }, { "epoch": 2.0312525566554855, "grad_norm": 0.07182000577449799, "learning_rate": 5.733641337502563e-06, "loss": 0.0008, "step": 124140 }, { "epoch": 2.0314161826065615, "grad_norm": 0.02416510321199894, "learning_rate": 5.732935104357146e-06, "loss": 0.001, "step": 124150 }, { "epoch": 2.031579808557637, "grad_norm": 0.02965659275650978, "learning_rate": 5.732228856267808e-06, "loss": 0.0006, "step": 124160 }, { "epoch": 2.031743434508713, "grad_norm": 0.07008717954158783, "learning_rate": 5.731522593248944e-06, "loss": 0.001, "step": 124170 }, { "epoch": 2.031907060459789, "grad_norm": 0.021772265434265137, "learning_rate": 5.730816315314958e-06, "loss": 0.0009, "step": 124180 }, { "epoch": 2.0320706864108646, "grad_norm": 0.030107226222753525, "learning_rate": 5.730110022480248e-06, "loss": 0.0011, "step": 124190 }, { "epoch": 2.0322343123619406, "grad_norm": 0.015858056023716927, "learning_rate": 5.729403714759216e-06, "loss": 0.0007, "step": 124200 }, { "epoch": 2.0323979383130166, "grad_norm": 0.03212274610996246, "learning_rate": 5.728697392166261e-06, "loss": 0.001, "step": 124210 }, { "epoch": 2.032561564264092, "grad_norm": 0.007674095220863819, "learning_rate": 5.727991054715789e-06, "loss": 0.0008, "step": 124220 }, { "epoch": 2.032725190215168, "grad_norm": 0.003547247964888811, "learning_rate": 5.727284702422196e-06, "loss": 0.001, "step": 124230 }, { "epoch": 2.032888816166244, "grad_norm": 0.05909010022878647, "learning_rate": 5.726578335299887e-06, "loss": 0.0012, "step": 124240 }, { "epoch": 2.0330524421173197, "grad_norm": 0.006230009254068136, "learning_rate": 5.725871953363264e-06, "loss": 0.0011, "step": 124250 }, { "epoch": 2.0332160680683957, "grad_norm": 0.20164914429187775, "learning_rate": 5.72516555662673e-06, "loss": 0.0008, "step": 124260 }, { "epoch": 2.0333796940194713, "grad_norm": 0.15534348785877228, "learning_rate": 5.724459145104685e-06, "loss": 0.0008, "step": 124270 }, { "epoch": 2.0335433199705473, "grad_norm": 0.02506355755031109, "learning_rate": 5.723752718811536e-06, "loss": 0.0013, "step": 124280 }, { "epoch": 2.0337069459216233, "grad_norm": 0.014953047968447208, "learning_rate": 5.7230462777616835e-06, "loss": 0.0016, "step": 124290 }, { "epoch": 2.033870571872699, "grad_norm": 0.0358613096177578, "learning_rate": 5.722339821969534e-06, "loss": 0.001, "step": 124300 }, { "epoch": 2.034034197823775, "grad_norm": 0.005480636842548847, "learning_rate": 5.721633351449487e-06, "loss": 0.0008, "step": 124310 }, { "epoch": 2.034197823774851, "grad_norm": 0.0077994465827941895, "learning_rate": 5.720926866215952e-06, "loss": 0.0007, "step": 124320 }, { "epoch": 2.0343614497259264, "grad_norm": 0.007929123006761074, "learning_rate": 5.72022036628333e-06, "loss": 0.0008, "step": 124330 }, { "epoch": 2.0345250756770024, "grad_norm": 0.05674552544951439, "learning_rate": 5.719513851666028e-06, "loss": 0.0009, "step": 124340 }, { "epoch": 2.0346887016280784, "grad_norm": 0.16010251641273499, "learning_rate": 5.718807322378451e-06, "loss": 0.0012, "step": 124350 }, { "epoch": 2.034852327579154, "grad_norm": 0.04301229864358902, "learning_rate": 5.718100778435004e-06, "loss": 0.0011, "step": 124360 }, { "epoch": 2.03501595353023, "grad_norm": 0.010216427966952324, "learning_rate": 5.717394219850092e-06, "loss": 0.001, "step": 124370 }, { "epoch": 2.035179579481306, "grad_norm": 0.026567094027996063, "learning_rate": 5.716687646638122e-06, "loss": 0.0007, "step": 124380 }, { "epoch": 2.0353432054323815, "grad_norm": 0.014878498390316963, "learning_rate": 5.715981058813501e-06, "loss": 0.001, "step": 124390 }, { "epoch": 2.0355068313834574, "grad_norm": 0.02516949363052845, "learning_rate": 5.7152744563906345e-06, "loss": 0.0009, "step": 124400 }, { "epoch": 2.0356704573345334, "grad_norm": 0.010888887569308281, "learning_rate": 5.714567839383931e-06, "loss": 0.0008, "step": 124410 }, { "epoch": 2.035834083285609, "grad_norm": 0.02120453305542469, "learning_rate": 5.713861207807796e-06, "loss": 0.002, "step": 124420 }, { "epoch": 2.035997709236685, "grad_norm": 0.024745874106884003, "learning_rate": 5.713154561676638e-06, "loss": 0.0008, "step": 124430 }, { "epoch": 2.036161335187761, "grad_norm": 0.029110927134752274, "learning_rate": 5.712447901004865e-06, "loss": 0.0012, "step": 124440 }, { "epoch": 2.0363249611388365, "grad_norm": 0.08133193105459213, "learning_rate": 5.711741225806887e-06, "loss": 0.0008, "step": 124450 }, { "epoch": 2.0364885870899125, "grad_norm": 0.00351930339820683, "learning_rate": 5.711034536097109e-06, "loss": 0.0005, "step": 124460 }, { "epoch": 2.036652213040988, "grad_norm": 0.18661271035671234, "learning_rate": 5.710327831889943e-06, "loss": 0.0017, "step": 124470 }, { "epoch": 2.036815838992064, "grad_norm": 0.020964492112398148, "learning_rate": 5.709621113199795e-06, "loss": 0.0009, "step": 124480 }, { "epoch": 2.03697946494314, "grad_norm": 0.026590334251523018, "learning_rate": 5.708914380041077e-06, "loss": 0.0011, "step": 124490 }, { "epoch": 2.0371430908942156, "grad_norm": 0.0028209169395267963, "learning_rate": 5.708207632428195e-06, "loss": 0.001, "step": 124500 }, { "epoch": 2.0373067168452916, "grad_norm": 0.047690633684396744, "learning_rate": 5.707500870375565e-06, "loss": 0.0005, "step": 124510 }, { "epoch": 2.0374703427963676, "grad_norm": 0.007298252545297146, "learning_rate": 5.706794093897593e-06, "loss": 0.0009, "step": 124520 }, { "epoch": 2.037633968747443, "grad_norm": 0.029687458649277687, "learning_rate": 5.70608730300869e-06, "loss": 0.0007, "step": 124530 }, { "epoch": 2.037797594698519, "grad_norm": 0.34806349873542786, "learning_rate": 5.705380497723268e-06, "loss": 0.0008, "step": 124540 }, { "epoch": 2.037961220649595, "grad_norm": 0.047569721937179565, "learning_rate": 5.7046736780557375e-06, "loss": 0.0011, "step": 124550 }, { "epoch": 2.0381248466006707, "grad_norm": 0.014218444935977459, "learning_rate": 5.7039668440205096e-06, "loss": 0.0011, "step": 124560 }, { "epoch": 2.0382884725517467, "grad_norm": 0.023789620026946068, "learning_rate": 5.7032599956319975e-06, "loss": 0.0006, "step": 124570 }, { "epoch": 2.0384520985028227, "grad_norm": 0.032649215310811996, "learning_rate": 5.702553132904611e-06, "loss": 0.001, "step": 124580 }, { "epoch": 2.0386157244538983, "grad_norm": 0.017308153212070465, "learning_rate": 5.701846255852764e-06, "loss": 0.0007, "step": 124590 }, { "epoch": 2.0387793504049743, "grad_norm": 0.08141926676034927, "learning_rate": 5.701139364490869e-06, "loss": 0.0022, "step": 124600 }, { "epoch": 2.0389429763560503, "grad_norm": 0.030755136162042618, "learning_rate": 5.700432458833339e-06, "loss": 0.001, "step": 124610 }, { "epoch": 2.039106602307126, "grad_norm": 0.00880351010710001, "learning_rate": 5.699725538894586e-06, "loss": 0.0006, "step": 124620 }, { "epoch": 2.039270228258202, "grad_norm": 0.020051661878824234, "learning_rate": 5.699018604689025e-06, "loss": 0.0006, "step": 124630 }, { "epoch": 2.039433854209278, "grad_norm": 0.023381909355521202, "learning_rate": 5.698311656231068e-06, "loss": 0.0005, "step": 124640 }, { "epoch": 2.0395974801603534, "grad_norm": 0.03758268430829048, "learning_rate": 5.697604693535131e-06, "loss": 0.0016, "step": 124650 }, { "epoch": 2.0397611061114294, "grad_norm": 0.048512157052755356, "learning_rate": 5.696897716615629e-06, "loss": 0.0006, "step": 124660 }, { "epoch": 2.039924732062505, "grad_norm": 0.0292520672082901, "learning_rate": 5.696190725486973e-06, "loss": 0.0006, "step": 124670 }, { "epoch": 2.040088358013581, "grad_norm": 0.06023142486810684, "learning_rate": 5.695483720163582e-06, "loss": 0.0011, "step": 124680 }, { "epoch": 2.040251983964657, "grad_norm": 0.017268039286136627, "learning_rate": 5.694776700659867e-06, "loss": 0.0007, "step": 124690 }, { "epoch": 2.0404156099157325, "grad_norm": 0.016490530222654343, "learning_rate": 5.694069666990249e-06, "loss": 0.0013, "step": 124700 }, { "epoch": 2.0405792358668084, "grad_norm": 0.001761007122695446, "learning_rate": 5.6933626191691385e-06, "loss": 0.0004, "step": 124710 }, { "epoch": 2.0407428618178844, "grad_norm": 0.038056857883930206, "learning_rate": 5.6926555572109554e-06, "loss": 0.0006, "step": 124720 }, { "epoch": 2.04090648776896, "grad_norm": 0.00963135901838541, "learning_rate": 5.691948481130113e-06, "loss": 0.0008, "step": 124730 }, { "epoch": 2.041070113720036, "grad_norm": 0.22121980786323547, "learning_rate": 5.691241390941031e-06, "loss": 0.0008, "step": 124740 }, { "epoch": 2.041233739671112, "grad_norm": 0.025389976799488068, "learning_rate": 5.6905342866581235e-06, "loss": 0.0004, "step": 124750 }, { "epoch": 2.0413973656221875, "grad_norm": 0.06620761752128601, "learning_rate": 5.689827168295811e-06, "loss": 0.0018, "step": 124760 }, { "epoch": 2.0415609915732635, "grad_norm": 0.03223131597042084, "learning_rate": 5.689120035868507e-06, "loss": 0.0008, "step": 124770 }, { "epoch": 2.0417246175243395, "grad_norm": 0.06073355674743652, "learning_rate": 5.688412889390633e-06, "loss": 0.0014, "step": 124780 }, { "epoch": 2.041888243475415, "grad_norm": 0.15890659391880035, "learning_rate": 5.687705728876606e-06, "loss": 0.0019, "step": 124790 }, { "epoch": 2.042051869426491, "grad_norm": 0.10773017257452011, "learning_rate": 5.686998554340843e-06, "loss": 0.0011, "step": 124800 }, { "epoch": 2.042215495377567, "grad_norm": 0.035403233021497726, "learning_rate": 5.686291365797764e-06, "loss": 0.0022, "step": 124810 }, { "epoch": 2.0423791213286426, "grad_norm": 0.0027364129200577736, "learning_rate": 5.685584163261788e-06, "loss": 0.0004, "step": 124820 }, { "epoch": 2.0425427472797186, "grad_norm": 0.039556220173835754, "learning_rate": 5.684876946747334e-06, "loss": 0.0009, "step": 124830 }, { "epoch": 2.0427063732307946, "grad_norm": 0.022852644324302673, "learning_rate": 5.6841697162688216e-06, "loss": 0.001, "step": 124840 }, { "epoch": 2.04286999918187, "grad_norm": 0.06728991121053696, "learning_rate": 5.683462471840671e-06, "loss": 0.0012, "step": 124850 }, { "epoch": 2.043033625132946, "grad_norm": 0.04303060472011566, "learning_rate": 5.6827552134773015e-06, "loss": 0.0011, "step": 124860 }, { "epoch": 2.0431972510840217, "grad_norm": 0.012269360944628716, "learning_rate": 5.682047941193132e-06, "loss": 0.0006, "step": 124870 }, { "epoch": 2.0433608770350977, "grad_norm": 0.11550545692443848, "learning_rate": 5.681340655002588e-06, "loss": 0.0012, "step": 124880 }, { "epoch": 2.0435245029861737, "grad_norm": 0.07113753259181976, "learning_rate": 5.680633354920085e-06, "loss": 0.0011, "step": 124890 }, { "epoch": 2.0436881289372493, "grad_norm": 0.01777237467467785, "learning_rate": 5.679926040960048e-06, "loss": 0.0015, "step": 124900 }, { "epoch": 2.0438517548883253, "grad_norm": 0.03022056631743908, "learning_rate": 5.679218713136897e-06, "loss": 0.0005, "step": 124910 }, { "epoch": 2.0440153808394013, "grad_norm": 0.012240900658071041, "learning_rate": 5.678511371465056e-06, "loss": 0.0008, "step": 124920 }, { "epoch": 2.044179006790477, "grad_norm": 0.056650735437870026, "learning_rate": 5.677804015958942e-06, "loss": 0.0006, "step": 124930 }, { "epoch": 2.044342632741553, "grad_norm": 0.09123074263334274, "learning_rate": 5.677096646632983e-06, "loss": 0.0006, "step": 124940 }, { "epoch": 2.044506258692629, "grad_norm": 0.07323645055294037, "learning_rate": 5.676389263501597e-06, "loss": 0.0009, "step": 124950 }, { "epoch": 2.0446698846437044, "grad_norm": 0.1890130490064621, "learning_rate": 5.675681866579211e-06, "loss": 0.001, "step": 124960 }, { "epoch": 2.0448335105947804, "grad_norm": 0.018108943477272987, "learning_rate": 5.674974455880245e-06, "loss": 0.0006, "step": 124970 }, { "epoch": 2.0449971365458564, "grad_norm": 0.05029362067580223, "learning_rate": 5.674267031419125e-06, "loss": 0.0008, "step": 124980 }, { "epoch": 2.045160762496932, "grad_norm": 0.05510476604104042, "learning_rate": 5.673559593210274e-06, "loss": 0.0008, "step": 124990 }, { "epoch": 2.045324388448008, "grad_norm": 0.0019000069005414844, "learning_rate": 5.672852141268115e-06, "loss": 0.0011, "step": 125000 }, { "epoch": 2.045488014399084, "grad_norm": 0.06416638195514679, "learning_rate": 5.672144675607073e-06, "loss": 0.0015, "step": 125010 }, { "epoch": 2.0456516403501595, "grad_norm": 0.07372941076755524, "learning_rate": 5.671437196241572e-06, "loss": 0.0014, "step": 125020 }, { "epoch": 2.0458152663012354, "grad_norm": 0.05934053286910057, "learning_rate": 5.670729703186039e-06, "loss": 0.0011, "step": 125030 }, { "epoch": 2.045978892252311, "grad_norm": 0.0051507200114429, "learning_rate": 5.670022196454898e-06, "loss": 0.0011, "step": 125040 }, { "epoch": 2.046142518203387, "grad_norm": 0.01723831705749035, "learning_rate": 5.669314676062574e-06, "loss": 0.0011, "step": 125050 }, { "epoch": 2.046306144154463, "grad_norm": 0.0007227336755022407, "learning_rate": 5.668607142023491e-06, "loss": 0.0013, "step": 125060 }, { "epoch": 2.0464697701055385, "grad_norm": 0.035050541162490845, "learning_rate": 5.66789959435208e-06, "loss": 0.0008, "step": 125070 }, { "epoch": 2.0466333960566145, "grad_norm": 0.007413554470986128, "learning_rate": 5.667192033062762e-06, "loss": 0.0008, "step": 125080 }, { "epoch": 2.0467970220076905, "grad_norm": 0.09566394984722137, "learning_rate": 5.666484458169967e-06, "loss": 0.0006, "step": 125090 }, { "epoch": 2.046960647958766, "grad_norm": 0.018905198201537132, "learning_rate": 5.66577686968812e-06, "loss": 0.0009, "step": 125100 }, { "epoch": 2.047124273909842, "grad_norm": 0.03642895817756653, "learning_rate": 5.66506926763165e-06, "loss": 0.0008, "step": 125110 }, { "epoch": 2.047287899860918, "grad_norm": 0.12292728573083878, "learning_rate": 5.664361652014981e-06, "loss": 0.001, "step": 125120 }, { "epoch": 2.0474515258119936, "grad_norm": 0.10239500552415848, "learning_rate": 5.6636540228525445e-06, "loss": 0.0011, "step": 125130 }, { "epoch": 2.0476151517630696, "grad_norm": 0.09077306091785431, "learning_rate": 5.662946380158767e-06, "loss": 0.0017, "step": 125140 }, { "epoch": 2.0477787777141456, "grad_norm": 0.023670615628361702, "learning_rate": 5.662238723948077e-06, "loss": 0.0009, "step": 125150 }, { "epoch": 2.047942403665221, "grad_norm": 0.08370963484048843, "learning_rate": 5.661531054234901e-06, "loss": 0.0011, "step": 125160 }, { "epoch": 2.048106029616297, "grad_norm": 0.013854516670107841, "learning_rate": 5.6608233710336715e-06, "loss": 0.003, "step": 125170 }, { "epoch": 2.048269655567373, "grad_norm": 0.03965938463807106, "learning_rate": 5.6601156743588135e-06, "loss": 0.0011, "step": 125180 }, { "epoch": 2.0484332815184487, "grad_norm": 0.06956108659505844, "learning_rate": 5.65940796422476e-06, "loss": 0.0009, "step": 125190 }, { "epoch": 2.0485969074695247, "grad_norm": 0.027044974267482758, "learning_rate": 5.658700240645938e-06, "loss": 0.0005, "step": 125200 }, { "epoch": 2.0487605334206007, "grad_norm": 0.02519354410469532, "learning_rate": 5.6579925036367785e-06, "loss": 0.0013, "step": 125210 }, { "epoch": 2.0489241593716763, "grad_norm": 0.0073639629408717155, "learning_rate": 5.657284753211712e-06, "loss": 0.0013, "step": 125220 }, { "epoch": 2.0490877853227523, "grad_norm": 0.0016433421988040209, "learning_rate": 5.6565769893851675e-06, "loss": 0.0006, "step": 125230 }, { "epoch": 2.049251411273828, "grad_norm": 0.05897844582796097, "learning_rate": 5.655869212171577e-06, "loss": 0.001, "step": 125240 }, { "epoch": 2.049415037224904, "grad_norm": 0.049651842564344406, "learning_rate": 5.655161421585371e-06, "loss": 0.001, "step": 125250 }, { "epoch": 2.04957866317598, "grad_norm": 0.07324064522981644, "learning_rate": 5.65445361764098e-06, "loss": 0.0009, "step": 125260 }, { "epoch": 2.0497422891270554, "grad_norm": 0.05796436592936516, "learning_rate": 5.653745800352838e-06, "loss": 0.0006, "step": 125270 }, { "epoch": 2.0499059150781314, "grad_norm": 0.0026803482323884964, "learning_rate": 5.6530379697353736e-06, "loss": 0.0008, "step": 125280 }, { "epoch": 2.0500695410292074, "grad_norm": 0.0642707496881485, "learning_rate": 5.652330125803022e-06, "loss": 0.0008, "step": 125290 }, { "epoch": 2.050233166980283, "grad_norm": 0.02582157962024212, "learning_rate": 5.651622268570212e-06, "loss": 0.0007, "step": 125300 }, { "epoch": 2.050396792931359, "grad_norm": 0.049030717462301254, "learning_rate": 5.650914398051379e-06, "loss": 0.0009, "step": 125310 }, { "epoch": 2.050560418882435, "grad_norm": 0.059373125433921814, "learning_rate": 5.650206514260957e-06, "loss": 0.0012, "step": 125320 }, { "epoch": 2.0507240448335105, "grad_norm": 0.04444790631532669, "learning_rate": 5.649498617213374e-06, "loss": 0.0007, "step": 125330 }, { "epoch": 2.0508876707845864, "grad_norm": 0.016188334673643112, "learning_rate": 5.6487907069230685e-06, "loss": 0.0008, "step": 125340 }, { "epoch": 2.0510512967356624, "grad_norm": 0.05534360930323601, "learning_rate": 5.64808278340447e-06, "loss": 0.0008, "step": 125350 }, { "epoch": 2.051214922686738, "grad_norm": 0.08003396540880203, "learning_rate": 5.647374846672018e-06, "loss": 0.0018, "step": 125360 }, { "epoch": 2.051378548637814, "grad_norm": 0.14823785424232483, "learning_rate": 5.646666896740141e-06, "loss": 0.0021, "step": 125370 }, { "epoch": 2.05154217458889, "grad_norm": 0.030598226934671402, "learning_rate": 5.645958933623277e-06, "loss": 0.001, "step": 125380 }, { "epoch": 2.0517058005399655, "grad_norm": 0.045618459582328796, "learning_rate": 5.645250957335859e-06, "loss": 0.0008, "step": 125390 }, { "epoch": 2.0518694264910415, "grad_norm": 0.011297231540083885, "learning_rate": 5.644542967892323e-06, "loss": 0.0008, "step": 125400 }, { "epoch": 2.0520330524421175, "grad_norm": 0.026881881058216095, "learning_rate": 5.643834965307104e-06, "loss": 0.0016, "step": 125410 }, { "epoch": 2.052196678393193, "grad_norm": 0.03291483595967293, "learning_rate": 5.643126949594639e-06, "loss": 0.0006, "step": 125420 }, { "epoch": 2.052360304344269, "grad_norm": 0.08018305897712708, "learning_rate": 5.6424189207693604e-06, "loss": 0.0007, "step": 125430 }, { "epoch": 2.0525239302953446, "grad_norm": 0.03160637617111206, "learning_rate": 5.641710878845708e-06, "loss": 0.0007, "step": 125440 }, { "epoch": 2.0526875562464206, "grad_norm": 0.03933035954833031, "learning_rate": 5.641002823838115e-06, "loss": 0.0006, "step": 125450 }, { "epoch": 2.0528511821974966, "grad_norm": 0.009737718850374222, "learning_rate": 5.640294755761021e-06, "loss": 0.0037, "step": 125460 }, { "epoch": 2.053014808148572, "grad_norm": 0.06412266939878464, "learning_rate": 5.63958667462886e-06, "loss": 0.0011, "step": 125470 }, { "epoch": 2.053178434099648, "grad_norm": 0.028196601197123528, "learning_rate": 5.638878580456072e-06, "loss": 0.0013, "step": 125480 }, { "epoch": 2.053342060050724, "grad_norm": 0.054382022470235825, "learning_rate": 5.638170473257092e-06, "loss": 0.001, "step": 125490 }, { "epoch": 2.0535056860017997, "grad_norm": 0.023668156936764717, "learning_rate": 5.63746235304636e-06, "loss": 0.0006, "step": 125500 }, { "epoch": 2.0536693119528757, "grad_norm": 0.025397470220923424, "learning_rate": 5.636754219838312e-06, "loss": 0.0009, "step": 125510 }, { "epoch": 2.0538329379039517, "grad_norm": 0.03468424454331398, "learning_rate": 5.636046073647388e-06, "loss": 0.0008, "step": 125520 }, { "epoch": 2.0539965638550273, "grad_norm": 0.056118838489055634, "learning_rate": 5.635337914488024e-06, "loss": 0.0013, "step": 125530 }, { "epoch": 2.0541601898061033, "grad_norm": 0.19330385327339172, "learning_rate": 5.634629742374661e-06, "loss": 0.0009, "step": 125540 }, { "epoch": 2.0543238157571793, "grad_norm": 0.06806580722332001, "learning_rate": 5.633921557321737e-06, "loss": 0.0018, "step": 125550 }, { "epoch": 2.054487441708255, "grad_norm": 0.07404562830924988, "learning_rate": 5.6332133593436925e-06, "loss": 0.0011, "step": 125560 }, { "epoch": 2.054651067659331, "grad_norm": 0.02622327208518982, "learning_rate": 5.632505148454965e-06, "loss": 0.0008, "step": 125570 }, { "epoch": 2.054814693610407, "grad_norm": 0.03767002001404762, "learning_rate": 5.631796924669996e-06, "loss": 0.0018, "step": 125580 }, { "epoch": 2.0549783195614824, "grad_norm": 0.03649899363517761, "learning_rate": 5.631088688003224e-06, "loss": 0.0006, "step": 125590 }, { "epoch": 2.0551419455125584, "grad_norm": 0.014396289363503456, "learning_rate": 5.630380438469092e-06, "loss": 0.0015, "step": 125600 }, { "epoch": 2.0553055714636344, "grad_norm": 0.019934022799134254, "learning_rate": 5.629672176082037e-06, "loss": 0.001, "step": 125610 }, { "epoch": 2.05546919741471, "grad_norm": 0.1302247941493988, "learning_rate": 5.628963900856503e-06, "loss": 0.0009, "step": 125620 }, { "epoch": 2.055632823365786, "grad_norm": 0.014869173988699913, "learning_rate": 5.628255612806931e-06, "loss": 0.0007, "step": 125630 }, { "epoch": 2.0557964493168615, "grad_norm": 0.01349637471139431, "learning_rate": 5.62754731194776e-06, "loss": 0.0007, "step": 125640 }, { "epoch": 2.0559600752679374, "grad_norm": 0.03882092610001564, "learning_rate": 5.626838998293434e-06, "loss": 0.0008, "step": 125650 }, { "epoch": 2.0561237012190134, "grad_norm": 0.04041079804301262, "learning_rate": 5.626130671858393e-06, "loss": 0.0009, "step": 125660 }, { "epoch": 2.056287327170089, "grad_norm": 0.036832500249147415, "learning_rate": 5.625422332657081e-06, "loss": 0.0008, "step": 125670 }, { "epoch": 2.056450953121165, "grad_norm": 0.03693995624780655, "learning_rate": 5.624713980703939e-06, "loss": 0.0014, "step": 125680 }, { "epoch": 2.056614579072241, "grad_norm": 0.06407913565635681, "learning_rate": 5.62400561601341e-06, "loss": 0.0016, "step": 125690 }, { "epoch": 2.0567782050233165, "grad_norm": 0.01161846611648798, "learning_rate": 5.623297238599936e-06, "loss": 0.0007, "step": 125700 }, { "epoch": 2.0569418309743925, "grad_norm": 0.029450466856360435, "learning_rate": 5.622588848477964e-06, "loss": 0.0012, "step": 125710 }, { "epoch": 2.0571054569254685, "grad_norm": 0.025810571387410164, "learning_rate": 5.621880445661934e-06, "loss": 0.0012, "step": 125720 }, { "epoch": 2.057269082876544, "grad_norm": 0.060586925595998764, "learning_rate": 5.6211720301662895e-06, "loss": 0.001, "step": 125730 }, { "epoch": 2.05743270882762, "grad_norm": 0.05953025817871094, "learning_rate": 5.620463602005476e-06, "loss": 0.0006, "step": 125740 }, { "epoch": 2.057596334778696, "grad_norm": 0.024456297978758812, "learning_rate": 5.619755161193939e-06, "loss": 0.0006, "step": 125750 }, { "epoch": 2.0577599607297716, "grad_norm": 0.021157745271921158, "learning_rate": 5.619046707746118e-06, "loss": 0.0008, "step": 125760 }, { "epoch": 2.0579235866808476, "grad_norm": 0.06086505949497223, "learning_rate": 5.618338241676465e-06, "loss": 0.0008, "step": 125770 }, { "epoch": 2.0580872126319236, "grad_norm": 0.040453627705574036, "learning_rate": 5.617629762999419e-06, "loss": 0.001, "step": 125780 }, { "epoch": 2.058250838582999, "grad_norm": 0.019519813358783722, "learning_rate": 5.616921271729428e-06, "loss": 0.0006, "step": 125790 }, { "epoch": 2.058414464534075, "grad_norm": 0.019026324152946472, "learning_rate": 5.616212767880936e-06, "loss": 0.0017, "step": 125800 }, { "epoch": 2.0585780904851507, "grad_norm": 0.059005845338106155, "learning_rate": 5.615504251468391e-06, "loss": 0.001, "step": 125810 }, { "epoch": 2.0587417164362267, "grad_norm": 0.0059578618966042995, "learning_rate": 5.614795722506236e-06, "loss": 0.0016, "step": 125820 }, { "epoch": 2.0589053423873027, "grad_norm": 0.045362718403339386, "learning_rate": 5.614087181008921e-06, "loss": 0.0008, "step": 125830 }, { "epoch": 2.0590689683383783, "grad_norm": 0.02235020324587822, "learning_rate": 5.613378626990889e-06, "loss": 0.0005, "step": 125840 }, { "epoch": 2.0592325942894543, "grad_norm": 0.02485421858727932, "learning_rate": 5.6126700604665905e-06, "loss": 0.0014, "step": 125850 }, { "epoch": 2.0593962202405303, "grad_norm": 0.053242914378643036, "learning_rate": 5.611961481450468e-06, "loss": 0.0012, "step": 125860 }, { "epoch": 2.059559846191606, "grad_norm": 0.03967425972223282, "learning_rate": 5.611252889956974e-06, "loss": 0.001, "step": 125870 }, { "epoch": 2.059723472142682, "grad_norm": 0.026126576587557793, "learning_rate": 5.610544286000552e-06, "loss": 0.0007, "step": 125880 }, { "epoch": 2.059887098093758, "grad_norm": 0.10060082376003265, "learning_rate": 5.609835669595651e-06, "loss": 0.0013, "step": 125890 }, { "epoch": 2.0600507240448334, "grad_norm": 0.038795121014118195, "learning_rate": 5.60912704075672e-06, "loss": 0.0011, "step": 125900 }, { "epoch": 2.0602143499959094, "grad_norm": 0.030308036133646965, "learning_rate": 5.608418399498207e-06, "loss": 0.0008, "step": 125910 }, { "epoch": 2.0603779759469854, "grad_norm": 0.04685911536216736, "learning_rate": 5.6077097458345585e-06, "loss": 0.0007, "step": 125920 }, { "epoch": 2.060541601898061, "grad_norm": 0.011732998304069042, "learning_rate": 5.6070010797802275e-06, "loss": 0.0008, "step": 125930 }, { "epoch": 2.060705227849137, "grad_norm": 0.004961515311151743, "learning_rate": 5.606292401349659e-06, "loss": 0.0009, "step": 125940 }, { "epoch": 2.060868853800213, "grad_norm": 0.026676513254642487, "learning_rate": 5.605583710557305e-06, "loss": 0.001, "step": 125950 }, { "epoch": 2.0610324797512884, "grad_norm": 0.01645735837519169, "learning_rate": 5.604875007417614e-06, "loss": 0.0014, "step": 125960 }, { "epoch": 2.0611961057023644, "grad_norm": 0.03997056558728218, "learning_rate": 5.604166291945036e-06, "loss": 0.0016, "step": 125970 }, { "epoch": 2.0613597316534404, "grad_norm": 0.06698976457118988, "learning_rate": 5.6034575641540215e-06, "loss": 0.0007, "step": 125980 }, { "epoch": 2.061523357604516, "grad_norm": 0.019128359854221344, "learning_rate": 5.6027488240590185e-06, "loss": 0.0009, "step": 125990 }, { "epoch": 2.061686983555592, "grad_norm": 0.0309952050447464, "learning_rate": 5.602040071674481e-06, "loss": 0.0014, "step": 126000 }, { "epoch": 2.0618506095066675, "grad_norm": 0.030629321932792664, "learning_rate": 5.601331307014859e-06, "loss": 0.0014, "step": 126010 }, { "epoch": 2.0620142354577435, "grad_norm": 0.04046066477894783, "learning_rate": 5.600622530094603e-06, "loss": 0.001, "step": 126020 }, { "epoch": 2.0621778614088195, "grad_norm": 0.07176393270492554, "learning_rate": 5.599913740928163e-06, "loss": 0.001, "step": 126030 }, { "epoch": 2.062341487359895, "grad_norm": 0.010868691839277744, "learning_rate": 5.599204939529994e-06, "loss": 0.0007, "step": 126040 }, { "epoch": 2.062505113310971, "grad_norm": 0.03339247778058052, "learning_rate": 5.598496125914544e-06, "loss": 0.001, "step": 126050 }, { "epoch": 2.062668739262047, "grad_norm": 0.035059213638305664, "learning_rate": 5.597787300096269e-06, "loss": 0.0017, "step": 126060 }, { "epoch": 2.0628323652131226, "grad_norm": 0.11543473601341248, "learning_rate": 5.597078462089618e-06, "loss": 0.0008, "step": 126070 }, { "epoch": 2.0629959911641986, "grad_norm": 0.00632654270157218, "learning_rate": 5.5963696119090445e-06, "loss": 0.0006, "step": 126080 }, { "epoch": 2.0631596171152746, "grad_norm": 0.039960600435733795, "learning_rate": 5.595660749569003e-06, "loss": 0.001, "step": 126090 }, { "epoch": 2.06332324306635, "grad_norm": 0.12612053751945496, "learning_rate": 5.594951875083945e-06, "loss": 0.0016, "step": 126100 }, { "epoch": 2.063486869017426, "grad_norm": 0.010120428167283535, "learning_rate": 5.594242988468324e-06, "loss": 0.0018, "step": 126110 }, { "epoch": 2.063650494968502, "grad_norm": 0.11553063243627548, "learning_rate": 5.5935340897365945e-06, "loss": 0.0011, "step": 126120 }, { "epoch": 2.0638141209195777, "grad_norm": 0.10140273720026016, "learning_rate": 5.592825178903209e-06, "loss": 0.0017, "step": 126130 }, { "epoch": 2.0639777468706537, "grad_norm": 0.08989692479372025, "learning_rate": 5.592116255982622e-06, "loss": 0.0016, "step": 126140 }, { "epoch": 2.0641413728217297, "grad_norm": 0.04119957610964775, "learning_rate": 5.5914073209892886e-06, "loss": 0.0008, "step": 126150 }, { "epoch": 2.0643049987728053, "grad_norm": 0.09927903115749359, "learning_rate": 5.590698373937663e-06, "loss": 0.0008, "step": 126160 }, { "epoch": 2.0644686247238813, "grad_norm": 0.0666622668504715, "learning_rate": 5.5899894148421995e-06, "loss": 0.0004, "step": 126170 }, { "epoch": 2.0646322506749573, "grad_norm": 0.0046473280526697636, "learning_rate": 5.589280443717354e-06, "loss": 0.0008, "step": 126180 }, { "epoch": 2.064795876626033, "grad_norm": 0.06273087114095688, "learning_rate": 5.588571460577581e-06, "loss": 0.0008, "step": 126190 }, { "epoch": 2.064959502577109, "grad_norm": 0.006211153231561184, "learning_rate": 5.587862465437338e-06, "loss": 0.0002, "step": 126200 }, { "epoch": 2.0651231285281844, "grad_norm": 0.05705232173204422, "learning_rate": 5.587153458311078e-06, "loss": 0.0015, "step": 126210 }, { "epoch": 2.0652867544792604, "grad_norm": 0.0006146291270852089, "learning_rate": 5.586444439213259e-06, "loss": 0.0016, "step": 126220 }, { "epoch": 2.0654503804303364, "grad_norm": 0.01663467474281788, "learning_rate": 5.5857354081583344e-06, "loss": 0.0006, "step": 126230 }, { "epoch": 2.065614006381412, "grad_norm": 0.0028204310219734907, "learning_rate": 5.585026365160766e-06, "loss": 0.001, "step": 126240 }, { "epoch": 2.065777632332488, "grad_norm": 0.05972966551780701, "learning_rate": 5.5843173102350054e-06, "loss": 0.0012, "step": 126250 }, { "epoch": 2.065941258283564, "grad_norm": 0.06363020092248917, "learning_rate": 5.583608243395513e-06, "loss": 0.001, "step": 126260 }, { "epoch": 2.0661048842346394, "grad_norm": 0.022857587784528732, "learning_rate": 5.582899164656746e-06, "loss": 0.0006, "step": 126270 }, { "epoch": 2.0662685101857154, "grad_norm": 0.00438783410936594, "learning_rate": 5.582190074033158e-06, "loss": 0.0048, "step": 126280 }, { "epoch": 2.0664321361367914, "grad_norm": 0.01888585463166237, "learning_rate": 5.581480971539211e-06, "loss": 0.0013, "step": 126290 }, { "epoch": 2.066595762087867, "grad_norm": 0.0023753135465085506, "learning_rate": 5.580771857189361e-06, "loss": 0.001, "step": 126300 }, { "epoch": 2.066759388038943, "grad_norm": 0.043583955615758896, "learning_rate": 5.5800627309980685e-06, "loss": 0.0012, "step": 126310 }, { "epoch": 2.066923013990019, "grad_norm": 0.04572613164782524, "learning_rate": 5.579353592979787e-06, "loss": 0.0024, "step": 126320 }, { "epoch": 2.0670866399410945, "grad_norm": 0.10316348820924759, "learning_rate": 5.578644443148982e-06, "loss": 0.0014, "step": 126330 }, { "epoch": 2.0672502658921705, "grad_norm": 0.0019180913222953677, "learning_rate": 5.577935281520106e-06, "loss": 0.0015, "step": 126340 }, { "epoch": 2.0674138918432465, "grad_norm": 0.01167813315987587, "learning_rate": 5.577226108107623e-06, "loss": 0.0009, "step": 126350 }, { "epoch": 2.067577517794322, "grad_norm": 0.0381314679980278, "learning_rate": 5.576516922925988e-06, "loss": 0.0016, "step": 126360 }, { "epoch": 2.067741143745398, "grad_norm": 0.0345546118915081, "learning_rate": 5.575807725989667e-06, "loss": 0.003, "step": 126370 }, { "epoch": 2.067904769696474, "grad_norm": 0.06248912215232849, "learning_rate": 5.575098517313113e-06, "loss": 0.0006, "step": 126380 }, { "epoch": 2.0680683956475496, "grad_norm": 0.08647521585226059, "learning_rate": 5.574389296910791e-06, "loss": 0.0008, "step": 126390 }, { "epoch": 2.0682320215986256, "grad_norm": 0.0032159145921468735, "learning_rate": 5.57368006479716e-06, "loss": 0.0007, "step": 126400 }, { "epoch": 2.068395647549701, "grad_norm": 0.12594473361968994, "learning_rate": 5.572970820986679e-06, "loss": 0.0009, "step": 126410 }, { "epoch": 2.068559273500777, "grad_norm": 0.042944878339767456, "learning_rate": 5.572261565493812e-06, "loss": 0.0009, "step": 126420 }, { "epoch": 2.068722899451853, "grad_norm": 0.0065295882523059845, "learning_rate": 5.5715522983330175e-06, "loss": 0.0011, "step": 126430 }, { "epoch": 2.0688865254029287, "grad_norm": 0.0458095520734787, "learning_rate": 5.570843019518757e-06, "loss": 0.0005, "step": 126440 }, { "epoch": 2.0690501513540047, "grad_norm": 0.1559317260980606, "learning_rate": 5.5701337290654935e-06, "loss": 0.0018, "step": 126450 }, { "epoch": 2.0692137773050807, "grad_norm": 0.15765973925590515, "learning_rate": 5.569424426987688e-06, "loss": 0.0011, "step": 126460 }, { "epoch": 2.0693774032561563, "grad_norm": 0.05849795043468475, "learning_rate": 5.568715113299803e-06, "loss": 0.0009, "step": 126470 }, { "epoch": 2.0695410292072323, "grad_norm": 0.019625496119260788, "learning_rate": 5.5680057880163e-06, "loss": 0.0009, "step": 126480 }, { "epoch": 2.0697046551583083, "grad_norm": 0.03308209404349327, "learning_rate": 5.567296451151643e-06, "loss": 0.0012, "step": 126490 }, { "epoch": 2.069868281109384, "grad_norm": 0.07449325174093246, "learning_rate": 5.5665871027202925e-06, "loss": 0.001, "step": 126500 }, { "epoch": 2.07003190706046, "grad_norm": 0.07789013534784317, "learning_rate": 5.565877742736714e-06, "loss": 0.0015, "step": 126510 }, { "epoch": 2.070195533011536, "grad_norm": 0.062038879841566086, "learning_rate": 5.5651683712153685e-06, "loss": 0.0011, "step": 126520 }, { "epoch": 2.0703591589626114, "grad_norm": 0.0527057945728302, "learning_rate": 5.564458988170723e-06, "loss": 0.0019, "step": 126530 }, { "epoch": 2.0705227849136874, "grad_norm": 0.05685804784297943, "learning_rate": 5.563749593617235e-06, "loss": 0.0008, "step": 126540 }, { "epoch": 2.0706864108647633, "grad_norm": 0.06928563863039017, "learning_rate": 5.563040187569375e-06, "loss": 0.001, "step": 126550 }, { "epoch": 2.070850036815839, "grad_norm": 0.05049328878521919, "learning_rate": 5.5623307700416026e-06, "loss": 0.0012, "step": 126560 }, { "epoch": 2.071013662766915, "grad_norm": 0.06238316372036934, "learning_rate": 5.5616213410483855e-06, "loss": 0.0006, "step": 126570 }, { "epoch": 2.0711772887179905, "grad_norm": 0.04858449101448059, "learning_rate": 5.560911900604187e-06, "loss": 0.0011, "step": 126580 }, { "epoch": 2.0713409146690664, "grad_norm": 0.008856616914272308, "learning_rate": 5.5602024487234705e-06, "loss": 0.0008, "step": 126590 }, { "epoch": 2.0715045406201424, "grad_norm": 0.012909521348774433, "learning_rate": 5.5594929854207045e-06, "loss": 0.0011, "step": 126600 }, { "epoch": 2.071668166571218, "grad_norm": 0.02146795764565468, "learning_rate": 5.55878351071035e-06, "loss": 0.0013, "step": 126610 }, { "epoch": 2.071831792522294, "grad_norm": 0.0256649237126112, "learning_rate": 5.558074024606878e-06, "loss": 0.0005, "step": 126620 }, { "epoch": 2.07199541847337, "grad_norm": 0.029334230348467827, "learning_rate": 5.557364527124749e-06, "loss": 0.0006, "step": 126630 }, { "epoch": 2.0721590444244455, "grad_norm": 0.00552686071023345, "learning_rate": 5.556655018278432e-06, "loss": 0.0009, "step": 126640 }, { "epoch": 2.0723226703755215, "grad_norm": 0.06485817581415176, "learning_rate": 5.555945498082391e-06, "loss": 0.0012, "step": 126650 }, { "epoch": 2.0724862963265975, "grad_norm": 0.0007553459727205336, "learning_rate": 5.555235966551097e-06, "loss": 0.0004, "step": 126660 }, { "epoch": 2.072649922277673, "grad_norm": 0.012117542326450348, "learning_rate": 5.554526423699011e-06, "loss": 0.0008, "step": 126670 }, { "epoch": 2.072813548228749, "grad_norm": 0.027743559330701828, "learning_rate": 5.5538168695406046e-06, "loss": 0.0006, "step": 126680 }, { "epoch": 2.072977174179825, "grad_norm": 0.17866480350494385, "learning_rate": 5.5531073040903425e-06, "loss": 0.0026, "step": 126690 }, { "epoch": 2.0731408001309006, "grad_norm": 0.02159965969622135, "learning_rate": 5.552397727362694e-06, "loss": 0.0011, "step": 126700 }, { "epoch": 2.0733044260819766, "grad_norm": 0.06973577290773392, "learning_rate": 5.551688139372124e-06, "loss": 0.0012, "step": 126710 }, { "epoch": 2.0734680520330526, "grad_norm": 0.01876549981534481, "learning_rate": 5.550978540133103e-06, "loss": 0.001, "step": 126720 }, { "epoch": 2.073631677984128, "grad_norm": 0.10894888639450073, "learning_rate": 5.550268929660097e-06, "loss": 0.0012, "step": 126730 }, { "epoch": 2.073795303935204, "grad_norm": 0.023779410868883133, "learning_rate": 5.549559307967578e-06, "loss": 0.0014, "step": 126740 }, { "epoch": 2.07395892988628, "grad_norm": 0.05924034118652344, "learning_rate": 5.548849675070009e-06, "loss": 0.0004, "step": 126750 }, { "epoch": 2.0741225558373557, "grad_norm": 0.01231676060706377, "learning_rate": 5.5481400309818645e-06, "loss": 0.0028, "step": 126760 }, { "epoch": 2.0742861817884317, "grad_norm": 0.009542014449834824, "learning_rate": 5.54743037571761e-06, "loss": 0.0006, "step": 126770 }, { "epoch": 2.0744498077395077, "grad_norm": 0.03539556264877319, "learning_rate": 5.546720709291715e-06, "loss": 0.0007, "step": 126780 }, { "epoch": 2.0746134336905833, "grad_norm": 0.03514469787478447, "learning_rate": 5.54601103171865e-06, "loss": 0.0012, "step": 126790 }, { "epoch": 2.0747770596416593, "grad_norm": 0.03221013769507408, "learning_rate": 5.5453013430128855e-06, "loss": 0.001, "step": 126800 }, { "epoch": 2.074940685592735, "grad_norm": 0.037754859775304794, "learning_rate": 5.544591643188889e-06, "loss": 0.0007, "step": 126810 }, { "epoch": 2.075104311543811, "grad_norm": 0.09447064995765686, "learning_rate": 5.543881932261134e-06, "loss": 0.0014, "step": 126820 }, { "epoch": 2.075267937494887, "grad_norm": 0.08090722560882568, "learning_rate": 5.5431722102440864e-06, "loss": 0.0009, "step": 126830 }, { "epoch": 2.0754315634459624, "grad_norm": 0.11773242801427841, "learning_rate": 5.542462477152222e-06, "loss": 0.0011, "step": 126840 }, { "epoch": 2.0755951893970384, "grad_norm": 0.03545616194605827, "learning_rate": 5.541752733000006e-06, "loss": 0.0008, "step": 126850 }, { "epoch": 2.0757588153481144, "grad_norm": 0.020160820335149765, "learning_rate": 5.541042977801915e-06, "loss": 0.0015, "step": 126860 }, { "epoch": 2.07592244129919, "grad_norm": 0.002868418814614415, "learning_rate": 5.540333211572418e-06, "loss": 0.0029, "step": 126870 }, { "epoch": 2.076086067250266, "grad_norm": 0.01968287117779255, "learning_rate": 5.539623434325984e-06, "loss": 0.001, "step": 126880 }, { "epoch": 2.076249693201342, "grad_norm": 0.01817791908979416, "learning_rate": 5.538913646077089e-06, "loss": 0.0007, "step": 126890 }, { "epoch": 2.0764133191524174, "grad_norm": 0.08099914342164993, "learning_rate": 5.538203846840202e-06, "loss": 0.0009, "step": 126900 }, { "epoch": 2.0765769451034934, "grad_norm": 0.04886912927031517, "learning_rate": 5.537494036629797e-06, "loss": 0.0018, "step": 126910 }, { "epoch": 2.0767405710545694, "grad_norm": 0.08538492023944855, "learning_rate": 5.5367842154603456e-06, "loss": 0.0012, "step": 126920 }, { "epoch": 2.076904197005645, "grad_norm": 0.025165975093841553, "learning_rate": 5.5360743833463214e-06, "loss": 0.0007, "step": 126930 }, { "epoch": 2.077067822956721, "grad_norm": 0.018273578956723213, "learning_rate": 5.535364540302195e-06, "loss": 0.0009, "step": 126940 }, { "epoch": 2.077231448907797, "grad_norm": 0.021661080420017242, "learning_rate": 5.534654686342443e-06, "loss": 0.001, "step": 126950 }, { "epoch": 2.0773950748588725, "grad_norm": 0.18048600852489471, "learning_rate": 5.5339448214815335e-06, "loss": 0.0008, "step": 126960 }, { "epoch": 2.0775587008099485, "grad_norm": 0.03857841715216637, "learning_rate": 5.533234945733945e-06, "loss": 0.0007, "step": 126970 }, { "epoch": 2.077722326761024, "grad_norm": 0.035724248737096786, "learning_rate": 5.53252505911415e-06, "loss": 0.0007, "step": 126980 }, { "epoch": 2.0778859527121, "grad_norm": 0.08578412979841232, "learning_rate": 5.531815161636621e-06, "loss": 0.0008, "step": 126990 }, { "epoch": 2.078049578663176, "grad_norm": 0.11418414115905762, "learning_rate": 5.531105253315833e-06, "loss": 0.0006, "step": 127000 }, { "epoch": 2.0782132046142516, "grad_norm": 0.020469481125473976, "learning_rate": 5.530395334166262e-06, "loss": 0.0007, "step": 127010 }, { "epoch": 2.0783768305653276, "grad_norm": 0.006163117941468954, "learning_rate": 5.52968540420238e-06, "loss": 0.001, "step": 127020 }, { "epoch": 2.0785404565164036, "grad_norm": 0.051462844014167786, "learning_rate": 5.528975463438663e-06, "loss": 0.0007, "step": 127030 }, { "epoch": 2.078704082467479, "grad_norm": 0.08771901577711105, "learning_rate": 5.528265511889585e-06, "loss": 0.0014, "step": 127040 }, { "epoch": 2.078867708418555, "grad_norm": 0.09041114151477814, "learning_rate": 5.527555549569625e-06, "loss": 0.0011, "step": 127050 }, { "epoch": 2.079031334369631, "grad_norm": 0.015558835119009018, "learning_rate": 5.526845576493255e-06, "loss": 0.0008, "step": 127060 }, { "epoch": 2.0791949603207067, "grad_norm": 0.011388033628463745, "learning_rate": 5.526135592674953e-06, "loss": 0.0007, "step": 127070 }, { "epoch": 2.0793585862717827, "grad_norm": 0.029925981536507607, "learning_rate": 5.525425598129191e-06, "loss": 0.001, "step": 127080 }, { "epoch": 2.0795222122228587, "grad_norm": 0.017690591514110565, "learning_rate": 5.52471559287045e-06, "loss": 0.0006, "step": 127090 }, { "epoch": 2.0796858381739343, "grad_norm": 0.08585383743047714, "learning_rate": 5.524005576913203e-06, "loss": 0.0008, "step": 127100 }, { "epoch": 2.0798494641250103, "grad_norm": 0.06295497715473175, "learning_rate": 5.523295550271928e-06, "loss": 0.0007, "step": 127110 }, { "epoch": 2.0800130900760863, "grad_norm": 0.05521595850586891, "learning_rate": 5.522585512961103e-06, "loss": 0.0013, "step": 127120 }, { "epoch": 2.080176716027162, "grad_norm": 0.03146205097436905, "learning_rate": 5.521875464995203e-06, "loss": 0.0015, "step": 127130 }, { "epoch": 2.080340341978238, "grad_norm": 0.010804425925016403, "learning_rate": 5.521165406388707e-06, "loss": 0.001, "step": 127140 }, { "epoch": 2.080503967929314, "grad_norm": 0.07480696588754654, "learning_rate": 5.520455337156091e-06, "loss": 0.0005, "step": 127150 }, { "epoch": 2.0806675938803894, "grad_norm": 0.04743222892284393, "learning_rate": 5.519745257311833e-06, "loss": 0.0008, "step": 127160 }, { "epoch": 2.0808312198314654, "grad_norm": 0.03364475816488266, "learning_rate": 5.519035166870413e-06, "loss": 0.0005, "step": 127170 }, { "epoch": 2.080994845782541, "grad_norm": 0.03827837482094765, "learning_rate": 5.518325065846304e-06, "loss": 0.001, "step": 127180 }, { "epoch": 2.081158471733617, "grad_norm": 0.05607517436146736, "learning_rate": 5.517614954253991e-06, "loss": 0.0006, "step": 127190 }, { "epoch": 2.081322097684693, "grad_norm": 0.021949106827378273, "learning_rate": 5.516904832107947e-06, "loss": 0.0009, "step": 127200 }, { "epoch": 2.0814857236357684, "grad_norm": 0.031204361468553543, "learning_rate": 5.516194699422654e-06, "loss": 0.0069, "step": 127210 }, { "epoch": 2.0816493495868444, "grad_norm": 0.04875165969133377, "learning_rate": 5.5154845562125894e-06, "loss": 0.0008, "step": 127220 }, { "epoch": 2.0818129755379204, "grad_norm": 0.06671786308288574, "learning_rate": 5.514774402492235e-06, "loss": 0.001, "step": 127230 }, { "epoch": 2.081976601488996, "grad_norm": 0.0016335243126377463, "learning_rate": 5.514064238276067e-06, "loss": 0.0013, "step": 127240 }, { "epoch": 2.082140227440072, "grad_norm": 0.05579192191362381, "learning_rate": 5.513354063578565e-06, "loss": 0.0006, "step": 127250 }, { "epoch": 2.082303853391148, "grad_norm": 0.10695833712816238, "learning_rate": 5.512643878414213e-06, "loss": 0.0015, "step": 127260 }, { "epoch": 2.0824674793422235, "grad_norm": 0.08716979622840881, "learning_rate": 5.511933682797487e-06, "loss": 0.0014, "step": 127270 }, { "epoch": 2.0826311052932995, "grad_norm": 0.053068600594997406, "learning_rate": 5.511223476742868e-06, "loss": 0.0011, "step": 127280 }, { "epoch": 2.0827947312443755, "grad_norm": 0.04701203852891922, "learning_rate": 5.510513260264838e-06, "loss": 0.0006, "step": 127290 }, { "epoch": 2.082958357195451, "grad_norm": 0.012136290781199932, "learning_rate": 5.509803033377876e-06, "loss": 0.001, "step": 127300 }, { "epoch": 2.083121983146527, "grad_norm": 0.10490881651639938, "learning_rate": 5.509092796096464e-06, "loss": 0.001, "step": 127310 }, { "epoch": 2.083285609097603, "grad_norm": 0.07015959173440933, "learning_rate": 5.508382548435084e-06, "loss": 0.0013, "step": 127320 }, { "epoch": 2.0834492350486786, "grad_norm": 0.056542858481407166, "learning_rate": 5.507672290408215e-06, "loss": 0.0009, "step": 127330 }, { "epoch": 2.0836128609997546, "grad_norm": 0.04648370295763016, "learning_rate": 5.5069620220303395e-06, "loss": 0.0011, "step": 127340 }, { "epoch": 2.08377648695083, "grad_norm": 0.10938386619091034, "learning_rate": 5.50625174331594e-06, "loss": 0.0008, "step": 127350 }, { "epoch": 2.083940112901906, "grad_norm": 0.03417222201824188, "learning_rate": 5.505541454279498e-06, "loss": 0.001, "step": 127360 }, { "epoch": 2.084103738852982, "grad_norm": 0.04907946288585663, "learning_rate": 5.504831154935496e-06, "loss": 0.0014, "step": 127370 }, { "epoch": 2.0842673648040577, "grad_norm": 0.009581728838384151, "learning_rate": 5.504120845298416e-06, "loss": 0.0011, "step": 127380 }, { "epoch": 2.0844309907551337, "grad_norm": 0.02711651660501957, "learning_rate": 5.503410525382741e-06, "loss": 0.0011, "step": 127390 }, { "epoch": 2.0845946167062097, "grad_norm": 0.005724904593080282, "learning_rate": 5.502700195202952e-06, "loss": 0.0007, "step": 127400 }, { "epoch": 2.0847582426572853, "grad_norm": 0.006302583031356335, "learning_rate": 5.501989854773535e-06, "loss": 0.0028, "step": 127410 }, { "epoch": 2.0849218686083613, "grad_norm": 0.07038763910531998, "learning_rate": 5.501279504108972e-06, "loss": 0.0008, "step": 127420 }, { "epoch": 2.0850854945594373, "grad_norm": 0.29055702686309814, "learning_rate": 5.500569143223745e-06, "loss": 0.0011, "step": 127430 }, { "epoch": 2.085249120510513, "grad_norm": 0.0331367589533329, "learning_rate": 5.49985877213234e-06, "loss": 0.0004, "step": 127440 }, { "epoch": 2.085412746461589, "grad_norm": 0.03355489298701286, "learning_rate": 5.499148390849239e-06, "loss": 0.0011, "step": 127450 }, { "epoch": 2.085576372412665, "grad_norm": 0.07194185256958008, "learning_rate": 5.498437999388927e-06, "loss": 0.0009, "step": 127460 }, { "epoch": 2.0857399983637404, "grad_norm": 0.04514938220381737, "learning_rate": 5.497727597765888e-06, "loss": 0.0007, "step": 127470 }, { "epoch": 2.0859036243148164, "grad_norm": 0.0450536273419857, "learning_rate": 5.497017185994607e-06, "loss": 0.0015, "step": 127480 }, { "epoch": 2.0860672502658923, "grad_norm": 0.015093316324055195, "learning_rate": 5.4963067640895676e-06, "loss": 0.0011, "step": 127490 }, { "epoch": 2.086230876216968, "grad_norm": 0.2687315344810486, "learning_rate": 5.4955963320652565e-06, "loss": 0.0006, "step": 127500 }, { "epoch": 2.086394502168044, "grad_norm": 0.0703972652554512, "learning_rate": 5.4948858899361565e-06, "loss": 0.0018, "step": 127510 }, { "epoch": 2.08655812811912, "grad_norm": 0.020363228395581245, "learning_rate": 5.4941754377167555e-06, "loss": 0.0011, "step": 127520 }, { "epoch": 2.0867217540701954, "grad_norm": 0.0436420701444149, "learning_rate": 5.493464975421537e-06, "loss": 0.0009, "step": 127530 }, { "epoch": 2.0868853800212714, "grad_norm": 0.032921139150857925, "learning_rate": 5.492754503064987e-06, "loss": 0.0009, "step": 127540 }, { "epoch": 2.0870490059723474, "grad_norm": 0.047293927520513535, "learning_rate": 5.492044020661591e-06, "loss": 0.0004, "step": 127550 }, { "epoch": 2.087212631923423, "grad_norm": 0.037558458745479584, "learning_rate": 5.4913335282258375e-06, "loss": 0.0008, "step": 127560 }, { "epoch": 2.087376257874499, "grad_norm": 0.05816853046417236, "learning_rate": 5.490623025772211e-06, "loss": 0.0008, "step": 127570 }, { "epoch": 2.0875398838255745, "grad_norm": 0.01794358901679516, "learning_rate": 5.489912513315197e-06, "loss": 0.0013, "step": 127580 }, { "epoch": 2.0877035097766505, "grad_norm": 0.013770426623523235, "learning_rate": 5.489201990869284e-06, "loss": 0.0016, "step": 127590 }, { "epoch": 2.0878671357277265, "grad_norm": 0.04773096367716789, "learning_rate": 5.488491458448958e-06, "loss": 0.0009, "step": 127600 }, { "epoch": 2.088030761678802, "grad_norm": 0.03080865927040577, "learning_rate": 5.487780916068708e-06, "loss": 0.001, "step": 127610 }, { "epoch": 2.088194387629878, "grad_norm": 0.025234008207917213, "learning_rate": 5.487070363743019e-06, "loss": 0.0005, "step": 127620 }, { "epoch": 2.088358013580954, "grad_norm": 0.0036649247631430626, "learning_rate": 5.48635980148638e-06, "loss": 0.001, "step": 127630 }, { "epoch": 2.0885216395320296, "grad_norm": 0.03009183146059513, "learning_rate": 5.485649229313277e-06, "loss": 0.0005, "step": 127640 }, { "epoch": 2.0886852654831056, "grad_norm": 0.03840583562850952, "learning_rate": 5.484938647238202e-06, "loss": 0.0016, "step": 127650 }, { "epoch": 2.0888488914341816, "grad_norm": 0.015472621656954288, "learning_rate": 5.484228055275638e-06, "loss": 0.001, "step": 127660 }, { "epoch": 2.089012517385257, "grad_norm": 0.0361701101064682, "learning_rate": 5.483517453440076e-06, "loss": 0.0004, "step": 127670 }, { "epoch": 2.089176143336333, "grad_norm": 0.01107774768024683, "learning_rate": 5.482806841746005e-06, "loss": 0.0011, "step": 127680 }, { "epoch": 2.089339769287409, "grad_norm": 0.02606785297393799, "learning_rate": 5.482096220207913e-06, "loss": 0.0007, "step": 127690 }, { "epoch": 2.0895033952384847, "grad_norm": 0.06091620400547981, "learning_rate": 5.4813855888402876e-06, "loss": 0.0013, "step": 127700 }, { "epoch": 2.0896670211895607, "grad_norm": 0.0531572662293911, "learning_rate": 5.480674947657622e-06, "loss": 0.0009, "step": 127710 }, { "epoch": 2.0898306471406367, "grad_norm": 0.03172849118709564, "learning_rate": 5.479964296674402e-06, "loss": 0.0011, "step": 127720 }, { "epoch": 2.0899942730917123, "grad_norm": 0.200875923037529, "learning_rate": 5.479253635905117e-06, "loss": 0.0009, "step": 127730 }, { "epoch": 2.0901578990427883, "grad_norm": 0.0009728423901833594, "learning_rate": 5.47854296536426e-06, "loss": 0.0012, "step": 127740 }, { "epoch": 2.090321524993864, "grad_norm": 0.07105766981840134, "learning_rate": 5.477832285066318e-06, "loss": 0.0007, "step": 127750 }, { "epoch": 2.09048515094494, "grad_norm": 0.03161156177520752, "learning_rate": 5.4771215950257804e-06, "loss": 0.0015, "step": 127760 }, { "epoch": 2.090648776896016, "grad_norm": 0.07549428939819336, "learning_rate": 5.4764108952571414e-06, "loss": 0.0013, "step": 127770 }, { "epoch": 2.0908124028470914, "grad_norm": 0.10856516659259796, "learning_rate": 5.4757001857748895e-06, "loss": 0.002, "step": 127780 }, { "epoch": 2.0909760287981674, "grad_norm": 0.008837908506393433, "learning_rate": 5.474989466593514e-06, "loss": 0.0011, "step": 127790 }, { "epoch": 2.0911396547492433, "grad_norm": 0.026460031047463417, "learning_rate": 5.474278737727508e-06, "loss": 0.0009, "step": 127800 }, { "epoch": 2.091303280700319, "grad_norm": 0.021960735321044922, "learning_rate": 5.473567999191363e-06, "loss": 0.0007, "step": 127810 }, { "epoch": 2.091466906651395, "grad_norm": 0.03951607644557953, "learning_rate": 5.472857250999567e-06, "loss": 0.0011, "step": 127820 }, { "epoch": 2.091630532602471, "grad_norm": 0.03023754619061947, "learning_rate": 5.472146493166616e-06, "loss": 0.0008, "step": 127830 }, { "epoch": 2.0917941585535464, "grad_norm": 0.025942059233784676, "learning_rate": 5.471435725706997e-06, "loss": 0.0008, "step": 127840 }, { "epoch": 2.0919577845046224, "grad_norm": 0.08151586353778839, "learning_rate": 5.470724948635206e-06, "loss": 0.0016, "step": 127850 }, { "epoch": 2.0921214104556984, "grad_norm": 0.02109985426068306, "learning_rate": 5.470014161965735e-06, "loss": 0.0014, "step": 127860 }, { "epoch": 2.092285036406774, "grad_norm": 0.027180980890989304, "learning_rate": 5.4693033657130735e-06, "loss": 0.0011, "step": 127870 }, { "epoch": 2.09244866235785, "grad_norm": 0.06020648032426834, "learning_rate": 5.468592559891714e-06, "loss": 0.0011, "step": 127880 }, { "epoch": 2.092612288308926, "grad_norm": 0.14823676645755768, "learning_rate": 5.4678817445161535e-06, "loss": 0.0012, "step": 127890 }, { "epoch": 2.0927759142600015, "grad_norm": 0.03964037448167801, "learning_rate": 5.467170919600882e-06, "loss": 0.001, "step": 127900 }, { "epoch": 2.0929395402110775, "grad_norm": 0.0011675918940454721, "learning_rate": 5.46646008516039e-06, "loss": 0.0007, "step": 127910 }, { "epoch": 2.0931031661621535, "grad_norm": 0.01185520924627781, "learning_rate": 5.465749241209176e-06, "loss": 0.0006, "step": 127920 }, { "epoch": 2.093266792113229, "grad_norm": 0.005160016473382711, "learning_rate": 5.46503838776173e-06, "loss": 0.0006, "step": 127930 }, { "epoch": 2.093430418064305, "grad_norm": 0.006387898698449135, "learning_rate": 5.464327524832548e-06, "loss": 0.001, "step": 127940 }, { "epoch": 2.0935940440153806, "grad_norm": 0.04681560397148132, "learning_rate": 5.4636166524361215e-06, "loss": 0.0011, "step": 127950 }, { "epoch": 2.0937576699664566, "grad_norm": 0.02931833826005459, "learning_rate": 5.462905770586946e-06, "loss": 0.0007, "step": 127960 }, { "epoch": 2.0939212959175326, "grad_norm": 0.08033258467912674, "learning_rate": 5.462194879299515e-06, "loss": 0.0032, "step": 127970 }, { "epoch": 2.094084921868608, "grad_norm": 0.033607352524995804, "learning_rate": 5.461483978588325e-06, "loss": 0.0011, "step": 127980 }, { "epoch": 2.094248547819684, "grad_norm": 0.03679050877690315, "learning_rate": 5.4607730684678676e-06, "loss": 0.0005, "step": 127990 }, { "epoch": 2.09441217377076, "grad_norm": 0.01229835394769907, "learning_rate": 5.460062148952641e-06, "loss": 0.0005, "step": 128000 }, { "epoch": 2.0945757997218357, "grad_norm": 0.017433805391192436, "learning_rate": 5.459351220057136e-06, "loss": 0.0008, "step": 128010 }, { "epoch": 2.0947394256729117, "grad_norm": 0.005076334346085787, "learning_rate": 5.4586402817958514e-06, "loss": 0.0008, "step": 128020 }, { "epoch": 2.0949030516239877, "grad_norm": 0.04750005528330803, "learning_rate": 5.457929334183282e-06, "loss": 0.0005, "step": 128030 }, { "epoch": 2.0950666775750633, "grad_norm": 0.03877115622162819, "learning_rate": 5.457218377233923e-06, "loss": 0.0006, "step": 128040 }, { "epoch": 2.0952303035261393, "grad_norm": 0.019528213888406754, "learning_rate": 5.45650741096227e-06, "loss": 0.001, "step": 128050 }, { "epoch": 2.0953939294772153, "grad_norm": 0.03886176645755768, "learning_rate": 5.4557964353828185e-06, "loss": 0.0013, "step": 128060 }, { "epoch": 2.095557555428291, "grad_norm": 0.034218449145555496, "learning_rate": 5.455085450510065e-06, "loss": 0.0011, "step": 128070 }, { "epoch": 2.095721181379367, "grad_norm": 0.0508282408118248, "learning_rate": 5.454374456358508e-06, "loss": 0.0018, "step": 128080 }, { "epoch": 2.095884807330443, "grad_norm": 0.054158978164196014, "learning_rate": 5.45366345294264e-06, "loss": 0.0013, "step": 128090 }, { "epoch": 2.0960484332815184, "grad_norm": 0.05748018994927406, "learning_rate": 5.4529524402769615e-06, "loss": 0.0008, "step": 128100 }, { "epoch": 2.0962120592325943, "grad_norm": 0.07453984022140503, "learning_rate": 5.452241418375968e-06, "loss": 0.0005, "step": 128110 }, { "epoch": 2.09637568518367, "grad_norm": 0.05112658441066742, "learning_rate": 5.451530387254157e-06, "loss": 0.0006, "step": 128120 }, { "epoch": 2.096539311134746, "grad_norm": 0.006029472686350346, "learning_rate": 5.450819346926024e-06, "loss": 0.0012, "step": 128130 }, { "epoch": 2.096702937085822, "grad_norm": 0.049987610429525375, "learning_rate": 5.450108297406069e-06, "loss": 0.001, "step": 128140 }, { "epoch": 2.0968665630368974, "grad_norm": 0.026073476299643517, "learning_rate": 5.449397238708786e-06, "loss": 0.0008, "step": 128150 }, { "epoch": 2.0970301889879734, "grad_norm": 0.10895001143217087, "learning_rate": 5.448686170848679e-06, "loss": 0.0014, "step": 128160 }, { "epoch": 2.0971938149390494, "grad_norm": 0.059170953929424286, "learning_rate": 5.447975093840241e-06, "loss": 0.0012, "step": 128170 }, { "epoch": 2.097357440890125, "grad_norm": 0.07669810205698013, "learning_rate": 5.447264007697973e-06, "loss": 0.0008, "step": 128180 }, { "epoch": 2.097521066841201, "grad_norm": 0.06210273504257202, "learning_rate": 5.446552912436371e-06, "loss": 0.0007, "step": 128190 }, { "epoch": 2.097684692792277, "grad_norm": 0.027957452461123466, "learning_rate": 5.445841808069936e-06, "loss": 0.0011, "step": 128200 }, { "epoch": 2.0978483187433525, "grad_norm": 0.01657315157353878, "learning_rate": 5.445130694613166e-06, "loss": 0.0008, "step": 128210 }, { "epoch": 2.0980119446944285, "grad_norm": 0.053585492074489594, "learning_rate": 5.444419572080557e-06, "loss": 0.001, "step": 128220 }, { "epoch": 2.0981755706455045, "grad_norm": 0.1811482161283493, "learning_rate": 5.443708440486614e-06, "loss": 0.0014, "step": 128230 }, { "epoch": 2.09833919659658, "grad_norm": 0.05013137310743332, "learning_rate": 5.442997299845832e-06, "loss": 0.001, "step": 128240 }, { "epoch": 2.098502822547656, "grad_norm": 0.020505372434854507, "learning_rate": 5.442286150172711e-06, "loss": 0.0011, "step": 128250 }, { "epoch": 2.098666448498732, "grad_norm": 0.006534064654260874, "learning_rate": 5.441574991481752e-06, "loss": 0.0008, "step": 128260 }, { "epoch": 2.0988300744498076, "grad_norm": 0.0710241049528122, "learning_rate": 5.440863823787455e-06, "loss": 0.0008, "step": 128270 }, { "epoch": 2.0989937004008836, "grad_norm": 0.043916989117860794, "learning_rate": 5.440152647104319e-06, "loss": 0.0009, "step": 128280 }, { "epoch": 2.0991573263519596, "grad_norm": 0.11252643913030624, "learning_rate": 5.439441461446845e-06, "loss": 0.0017, "step": 128290 }, { "epoch": 2.099320952303035, "grad_norm": 0.035514164716005325, "learning_rate": 5.438730266829533e-06, "loss": 0.001, "step": 128300 }, { "epoch": 2.099484578254111, "grad_norm": 0.003507747082039714, "learning_rate": 5.438019063266885e-06, "loss": 0.0004, "step": 128310 }, { "epoch": 2.099648204205187, "grad_norm": 0.06512708216905594, "learning_rate": 5.4373078507734e-06, "loss": 0.0011, "step": 128320 }, { "epoch": 2.0998118301562627, "grad_norm": 0.04155729338526726, "learning_rate": 5.4365966293635805e-06, "loss": 0.0007, "step": 128330 }, { "epoch": 2.0999754561073387, "grad_norm": 0.08366266638040543, "learning_rate": 5.435885399051926e-06, "loss": 0.0008, "step": 128340 }, { "epoch": 2.1001390820584143, "grad_norm": 0.05120432376861572, "learning_rate": 5.43517415985294e-06, "loss": 0.0009, "step": 128350 }, { "epoch": 2.1003027080094903, "grad_norm": 0.06429214775562286, "learning_rate": 5.434462911781122e-06, "loss": 0.001, "step": 128360 }, { "epoch": 2.1004663339605663, "grad_norm": 0.04152296483516693, "learning_rate": 5.4337516548509764e-06, "loss": 0.0011, "step": 128370 }, { "epoch": 2.100629959911642, "grad_norm": 0.015728864818811417, "learning_rate": 5.433040389077001e-06, "loss": 0.0012, "step": 128380 }, { "epoch": 2.100793585862718, "grad_norm": 0.011106511577963829, "learning_rate": 5.432329114473702e-06, "loss": 0.0007, "step": 128390 }, { "epoch": 2.100957211813794, "grad_norm": 0.002668684581294656, "learning_rate": 5.43161783105558e-06, "loss": 0.001, "step": 128400 }, { "epoch": 2.1011208377648694, "grad_norm": 0.03396306186914444, "learning_rate": 5.430906538837138e-06, "loss": 0.0007, "step": 128410 }, { "epoch": 2.1012844637159453, "grad_norm": 0.034778330475091934, "learning_rate": 5.430195237832876e-06, "loss": 0.002, "step": 128420 }, { "epoch": 2.1014480896670213, "grad_norm": 0.020169131457805634, "learning_rate": 5.4294839280573005e-06, "loss": 0.0007, "step": 128430 }, { "epoch": 2.101611715618097, "grad_norm": 0.02032463625073433, "learning_rate": 5.428772609524913e-06, "loss": 0.0009, "step": 128440 }, { "epoch": 2.101775341569173, "grad_norm": 0.09976786375045776, "learning_rate": 5.428061282250217e-06, "loss": 0.001, "step": 128450 }, { "epoch": 2.101938967520249, "grad_norm": 0.021337129175662994, "learning_rate": 5.427349946247714e-06, "loss": 0.0011, "step": 128460 }, { "epoch": 2.1021025934713244, "grad_norm": 0.03314554691314697, "learning_rate": 5.426638601531911e-06, "loss": 0.0004, "step": 128470 }, { "epoch": 2.1022662194224004, "grad_norm": 0.040230851620435715, "learning_rate": 5.425927248117308e-06, "loss": 0.0012, "step": 128480 }, { "epoch": 2.1024298453734764, "grad_norm": 0.13855083286762238, "learning_rate": 5.4252158860184115e-06, "loss": 0.0014, "step": 128490 }, { "epoch": 2.102593471324552, "grad_norm": 0.07660362869501114, "learning_rate": 5.424504515249725e-06, "loss": 0.0013, "step": 128500 }, { "epoch": 2.102757097275628, "grad_norm": 0.03207476809620857, "learning_rate": 5.423793135825752e-06, "loss": 0.0004, "step": 128510 }, { "epoch": 2.1029207232267035, "grad_norm": 0.04486614465713501, "learning_rate": 5.423081747760996e-06, "loss": 0.0009, "step": 128520 }, { "epoch": 2.1030843491777795, "grad_norm": 0.0033042393624782562, "learning_rate": 5.4223703510699654e-06, "loss": 0.001, "step": 128530 }, { "epoch": 2.1032479751288555, "grad_norm": 0.021685002371668816, "learning_rate": 5.421658945767163e-06, "loss": 0.0019, "step": 128540 }, { "epoch": 2.103411601079931, "grad_norm": 0.05071347951889038, "learning_rate": 5.420947531867091e-06, "loss": 0.0011, "step": 128550 }, { "epoch": 2.103575227031007, "grad_norm": 0.024000020697712898, "learning_rate": 5.420236109384258e-06, "loss": 0.0011, "step": 128560 }, { "epoch": 2.103738852982083, "grad_norm": 0.05308258906006813, "learning_rate": 5.419524678333168e-06, "loss": 0.0016, "step": 128570 }, { "epoch": 2.1039024789331586, "grad_norm": 0.0318782813847065, "learning_rate": 5.418813238728327e-06, "loss": 0.0012, "step": 128580 }, { "epoch": 2.1040661048842346, "grad_norm": 0.19338677823543549, "learning_rate": 5.418101790584238e-06, "loss": 0.001, "step": 128590 }, { "epoch": 2.1042297308353106, "grad_norm": 0.051501739770174026, "learning_rate": 5.417390333915412e-06, "loss": 0.0011, "step": 128600 }, { "epoch": 2.104393356786386, "grad_norm": 0.015313917770981789, "learning_rate": 5.4166788687363505e-06, "loss": 0.0012, "step": 128610 }, { "epoch": 2.104556982737462, "grad_norm": 0.14873723685741425, "learning_rate": 5.415967395061562e-06, "loss": 0.001, "step": 128620 }, { "epoch": 2.104720608688538, "grad_norm": 0.052041273564100266, "learning_rate": 5.415255912905552e-06, "loss": 0.0008, "step": 128630 }, { "epoch": 2.1048842346396137, "grad_norm": 0.005774533376097679, "learning_rate": 5.414544422282826e-06, "loss": 0.0005, "step": 128640 }, { "epoch": 2.1050478605906897, "grad_norm": 0.03869450464844704, "learning_rate": 5.413832923207892e-06, "loss": 0.0008, "step": 128650 }, { "epoch": 2.1052114865417657, "grad_norm": 0.042450129985809326, "learning_rate": 5.413121415695258e-06, "loss": 0.0016, "step": 128660 }, { "epoch": 2.1053751124928413, "grad_norm": 0.020771289244294167, "learning_rate": 5.412409899759429e-06, "loss": 0.0011, "step": 128670 }, { "epoch": 2.1055387384439173, "grad_norm": 0.06832046061754227, "learning_rate": 5.411698375414913e-06, "loss": 0.001, "step": 128680 }, { "epoch": 2.1057023643949933, "grad_norm": 0.21682244539260864, "learning_rate": 5.410986842676216e-06, "loss": 0.0012, "step": 128690 }, { "epoch": 2.105865990346069, "grad_norm": 0.00772886723279953, "learning_rate": 5.410275301557849e-06, "loss": 0.0005, "step": 128700 }, { "epoch": 2.106029616297145, "grad_norm": 0.013005086220800877, "learning_rate": 5.409563752074316e-06, "loss": 0.0007, "step": 128710 }, { "epoch": 2.1061932422482204, "grad_norm": 0.07774049043655396, "learning_rate": 5.408852194240127e-06, "loss": 0.0012, "step": 128720 }, { "epoch": 2.1063568681992964, "grad_norm": 0.1634555459022522, "learning_rate": 5.408140628069789e-06, "loss": 0.0013, "step": 128730 }, { "epoch": 2.1065204941503723, "grad_norm": 0.039952151477336884, "learning_rate": 5.407429053577811e-06, "loss": 0.0007, "step": 128740 }, { "epoch": 2.106684120101448, "grad_norm": 0.04738624766469002, "learning_rate": 5.4067174707787006e-06, "loss": 0.001, "step": 128750 }, { "epoch": 2.106847746052524, "grad_norm": 0.10790464282035828, "learning_rate": 5.406005879686968e-06, "loss": 0.0011, "step": 128760 }, { "epoch": 2.1070113720036, "grad_norm": 0.0958368331193924, "learning_rate": 5.405294280317119e-06, "loss": 0.0012, "step": 128770 }, { "epoch": 2.1071749979546754, "grad_norm": 0.04208781570196152, "learning_rate": 5.404582672683667e-06, "loss": 0.001, "step": 128780 }, { "epoch": 2.1073386239057514, "grad_norm": 0.04373541474342346, "learning_rate": 5.403871056801115e-06, "loss": 0.0016, "step": 128790 }, { "epoch": 2.1075022498568274, "grad_norm": 0.00957710575312376, "learning_rate": 5.403159432683979e-06, "loss": 0.0008, "step": 128800 }, { "epoch": 2.107665875807903, "grad_norm": 0.00452103465795517, "learning_rate": 5.4024478003467625e-06, "loss": 0.0016, "step": 128810 }, { "epoch": 2.107829501758979, "grad_norm": 0.021282238885760307, "learning_rate": 5.40173615980398e-06, "loss": 0.0007, "step": 128820 }, { "epoch": 2.107993127710055, "grad_norm": 0.008213472552597523, "learning_rate": 5.401024511070136e-06, "loss": 0.0007, "step": 128830 }, { "epoch": 2.1081567536611305, "grad_norm": 0.036547258496284485, "learning_rate": 5.400312854159746e-06, "loss": 0.001, "step": 128840 }, { "epoch": 2.1083203796122065, "grad_norm": 0.006087661255151033, "learning_rate": 5.399601189087317e-06, "loss": 0.0008, "step": 128850 }, { "epoch": 2.1084840055632825, "grad_norm": 0.06327825784683228, "learning_rate": 5.398889515867358e-06, "loss": 0.0013, "step": 128860 }, { "epoch": 2.108647631514358, "grad_norm": 0.1145077794790268, "learning_rate": 5.398177834514382e-06, "loss": 0.0013, "step": 128870 }, { "epoch": 2.108811257465434, "grad_norm": 0.030956553295254707, "learning_rate": 5.397466145042898e-06, "loss": 0.0008, "step": 128880 }, { "epoch": 2.10897488341651, "grad_norm": 0.015219282358884811, "learning_rate": 5.396754447467418e-06, "loss": 0.0007, "step": 128890 }, { "epoch": 2.1091385093675856, "grad_norm": 0.05790552496910095, "learning_rate": 5.3960427418024515e-06, "loss": 0.0007, "step": 128900 }, { "epoch": 2.1093021353186616, "grad_norm": 0.06269259750843048, "learning_rate": 5.395331028062511e-06, "loss": 0.0008, "step": 128910 }, { "epoch": 2.109465761269737, "grad_norm": 0.05463624373078346, "learning_rate": 5.394619306262106e-06, "loss": 0.0015, "step": 128920 }, { "epoch": 2.109629387220813, "grad_norm": 0.028963671997189522, "learning_rate": 5.393907576415751e-06, "loss": 0.0016, "step": 128930 }, { "epoch": 2.109793013171889, "grad_norm": 0.09018687158823013, "learning_rate": 5.393195838537954e-06, "loss": 0.0007, "step": 128940 }, { "epoch": 2.1099566391229647, "grad_norm": 0.09030677378177643, "learning_rate": 5.392484092643228e-06, "loss": 0.0015, "step": 128950 }, { "epoch": 2.1101202650740407, "grad_norm": 0.04395951330661774, "learning_rate": 5.391772338746086e-06, "loss": 0.0008, "step": 128960 }, { "epoch": 2.1102838910251167, "grad_norm": 0.016099167987704277, "learning_rate": 5.391060576861039e-06, "loss": 0.0006, "step": 128970 }, { "epoch": 2.1104475169761923, "grad_norm": 0.03294757753610611, "learning_rate": 5.390348807002599e-06, "loss": 0.0007, "step": 128980 }, { "epoch": 2.1106111429272683, "grad_norm": 0.054866448044776917, "learning_rate": 5.3896370291852796e-06, "loss": 0.0012, "step": 128990 }, { "epoch": 2.1107747688783443, "grad_norm": 0.026474185287952423, "learning_rate": 5.388925243423591e-06, "loss": 0.001, "step": 129000 }, { "epoch": 2.11093839482942, "grad_norm": 0.09316133707761765, "learning_rate": 5.388213449732049e-06, "loss": 0.0013, "step": 129010 }, { "epoch": 2.111102020780496, "grad_norm": 0.03991600498557091, "learning_rate": 5.387501648125165e-06, "loss": 0.0014, "step": 129020 }, { "epoch": 2.111265646731572, "grad_norm": 0.06469649821519852, "learning_rate": 5.386789838617452e-06, "loss": 0.001, "step": 129030 }, { "epoch": 2.1114292726826474, "grad_norm": 0.00869617611169815, "learning_rate": 5.386078021223422e-06, "loss": 0.0005, "step": 129040 }, { "epoch": 2.1115928986337233, "grad_norm": 0.013891749083995819, "learning_rate": 5.38536619595759e-06, "loss": 0.0008, "step": 129050 }, { "epoch": 2.1117565245847993, "grad_norm": 0.018288522958755493, "learning_rate": 5.384654362834469e-06, "loss": 0.0011, "step": 129060 }, { "epoch": 2.111920150535875, "grad_norm": 0.0957750603556633, "learning_rate": 5.383942521868573e-06, "loss": 0.0008, "step": 129070 }, { "epoch": 2.112083776486951, "grad_norm": 0.08211557567119598, "learning_rate": 5.3832306730744146e-06, "loss": 0.0005, "step": 129080 }, { "epoch": 2.112247402438027, "grad_norm": 0.08216419070959091, "learning_rate": 5.3825188164665095e-06, "loss": 0.0014, "step": 129090 }, { "epoch": 2.1124110283891024, "grad_norm": 0.046954210847616196, "learning_rate": 5.381806952059371e-06, "loss": 0.0012, "step": 129100 }, { "epoch": 2.1125746543401784, "grad_norm": 0.05451008304953575, "learning_rate": 5.381095079867513e-06, "loss": 0.0012, "step": 129110 }, { "epoch": 2.112738280291254, "grad_norm": 0.04332735016942024, "learning_rate": 5.380383199905451e-06, "loss": 0.0017, "step": 129120 }, { "epoch": 2.11290190624233, "grad_norm": 0.030689332634210587, "learning_rate": 5.3796713121877e-06, "loss": 0.0009, "step": 129130 }, { "epoch": 2.113065532193406, "grad_norm": 0.021522553637623787, "learning_rate": 5.378959416728772e-06, "loss": 0.0021, "step": 129140 }, { "epoch": 2.1132291581444815, "grad_norm": 0.03237219899892807, "learning_rate": 5.378247513543185e-06, "loss": 0.0008, "step": 129150 }, { "epoch": 2.1133927840955575, "grad_norm": 0.08639468997716904, "learning_rate": 5.377535602645452e-06, "loss": 0.001, "step": 129160 }, { "epoch": 2.1135564100466335, "grad_norm": 0.04379066824913025, "learning_rate": 5.37682368405009e-06, "loss": 0.001, "step": 129170 }, { "epoch": 2.113720035997709, "grad_norm": 0.008758671581745148, "learning_rate": 5.3761117577716126e-06, "loss": 0.0009, "step": 129180 }, { "epoch": 2.113883661948785, "grad_norm": 0.02812991663813591, "learning_rate": 5.375399823824537e-06, "loss": 0.0007, "step": 129190 }, { "epoch": 2.114047287899861, "grad_norm": 0.06792179495096207, "learning_rate": 5.374687882223378e-06, "loss": 0.0012, "step": 129200 }, { "epoch": 2.1142109138509366, "grad_norm": 0.019751688465476036, "learning_rate": 5.3739759329826515e-06, "loss": 0.0006, "step": 129210 }, { "epoch": 2.1143745398020126, "grad_norm": 0.0017937865341082215, "learning_rate": 5.373263976116875e-06, "loss": 0.001, "step": 129220 }, { "epoch": 2.1145381657530886, "grad_norm": 0.024622635915875435, "learning_rate": 5.372552011640562e-06, "loss": 0.0008, "step": 129230 }, { "epoch": 2.114701791704164, "grad_norm": 0.016784492880105972, "learning_rate": 5.371840039568231e-06, "loss": 0.0007, "step": 129240 }, { "epoch": 2.11486541765524, "grad_norm": 0.0550917387008667, "learning_rate": 5.371128059914398e-06, "loss": 0.0005, "step": 129250 }, { "epoch": 2.115029043606316, "grad_norm": 0.00821860320866108, "learning_rate": 5.3704160726935795e-06, "loss": 0.0006, "step": 129260 }, { "epoch": 2.1151926695573917, "grad_norm": 0.024402398616075516, "learning_rate": 5.369704077920291e-06, "loss": 0.0006, "step": 129270 }, { "epoch": 2.1153562955084677, "grad_norm": 0.03564322367310524, "learning_rate": 5.368992075609052e-06, "loss": 0.0011, "step": 129280 }, { "epoch": 2.1155199214595433, "grad_norm": 0.534134030342102, "learning_rate": 5.368280065774379e-06, "loss": 0.0012, "step": 129290 }, { "epoch": 2.1156835474106193, "grad_norm": 0.005957210902124643, "learning_rate": 5.367568048430787e-06, "loss": 0.0007, "step": 129300 }, { "epoch": 2.1158471733616953, "grad_norm": 0.04043357074260712, "learning_rate": 5.366856023592796e-06, "loss": 0.0008, "step": 129310 }, { "epoch": 2.116010799312771, "grad_norm": 0.0024069107603281736, "learning_rate": 5.3661439912749225e-06, "loss": 0.0009, "step": 129320 }, { "epoch": 2.116174425263847, "grad_norm": 0.01956222951412201, "learning_rate": 5.365431951491684e-06, "loss": 0.0017, "step": 129330 }, { "epoch": 2.116338051214923, "grad_norm": 0.004906907677650452, "learning_rate": 5.3647199042576e-06, "loss": 0.0008, "step": 129340 }, { "epoch": 2.1165016771659984, "grad_norm": 0.05126886069774628, "learning_rate": 5.364007849587186e-06, "loss": 0.0008, "step": 129350 }, { "epoch": 2.1166653031170743, "grad_norm": 0.005901460535824299, "learning_rate": 5.363295787494963e-06, "loss": 0.0014, "step": 129360 }, { "epoch": 2.1168289290681503, "grad_norm": 0.06286363303661346, "learning_rate": 5.362583717995447e-06, "loss": 0.0012, "step": 129370 }, { "epoch": 2.116992555019226, "grad_norm": 0.06628582626581192, "learning_rate": 5.361871641103158e-06, "loss": 0.0009, "step": 129380 }, { "epoch": 2.117156180970302, "grad_norm": 0.04684653505682945, "learning_rate": 5.361159556832612e-06, "loss": 0.0008, "step": 129390 }, { "epoch": 2.117319806921378, "grad_norm": 0.05028873682022095, "learning_rate": 5.360447465198332e-06, "loss": 0.0007, "step": 129400 }, { "epoch": 2.1174834328724534, "grad_norm": 0.024502957239747047, "learning_rate": 5.359735366214833e-06, "loss": 0.001, "step": 129410 }, { "epoch": 2.1176470588235294, "grad_norm": 0.0598117858171463, "learning_rate": 5.359023259896638e-06, "loss": 0.0015, "step": 129420 }, { "epoch": 2.1178106847746054, "grad_norm": 0.07173280417919159, "learning_rate": 5.3583111462582625e-06, "loss": 0.0011, "step": 129430 }, { "epoch": 2.117974310725681, "grad_norm": 0.046607889235019684, "learning_rate": 5.357599025314228e-06, "loss": 0.0013, "step": 129440 }, { "epoch": 2.118137936676757, "grad_norm": 0.018439145758748055, "learning_rate": 5.356886897079053e-06, "loss": 0.0006, "step": 129450 }, { "epoch": 2.118301562627833, "grad_norm": 0.03717545047402382, "learning_rate": 5.356174761567259e-06, "loss": 0.0005, "step": 129460 }, { "epoch": 2.1184651885789085, "grad_norm": 0.021857429295778275, "learning_rate": 5.355462618793362e-06, "loss": 0.0012, "step": 129470 }, { "epoch": 2.1186288145299845, "grad_norm": 0.04251977056264877, "learning_rate": 5.354750468771886e-06, "loss": 0.001, "step": 129480 }, { "epoch": 2.1187924404810605, "grad_norm": 0.009001165628433228, "learning_rate": 5.35403831151735e-06, "loss": 0.0007, "step": 129490 }, { "epoch": 2.118956066432136, "grad_norm": 0.05415630713105202, "learning_rate": 5.3533261470442746e-06, "loss": 0.0013, "step": 129500 }, { "epoch": 2.119119692383212, "grad_norm": 0.03642025962471962, "learning_rate": 5.352613975367179e-06, "loss": 0.0006, "step": 129510 }, { "epoch": 2.1192833183342876, "grad_norm": 0.03718504309654236, "learning_rate": 5.351901796500583e-06, "loss": 0.001, "step": 129520 }, { "epoch": 2.1194469442853636, "grad_norm": 0.018629835918545723, "learning_rate": 5.35118961045901e-06, "loss": 0.0008, "step": 129530 }, { "epoch": 2.1196105702364396, "grad_norm": 0.05823684483766556, "learning_rate": 5.350477417256978e-06, "loss": 0.0008, "step": 129540 }, { "epoch": 2.119774196187515, "grad_norm": 0.0021217523608356714, "learning_rate": 5.349765216909012e-06, "loss": 0.0011, "step": 129550 }, { "epoch": 2.119937822138591, "grad_norm": 0.0419219508767128, "learning_rate": 5.349053009429629e-06, "loss": 0.0015, "step": 129560 }, { "epoch": 2.120101448089667, "grad_norm": 0.027748316526412964, "learning_rate": 5.348340794833352e-06, "loss": 0.0004, "step": 129570 }, { "epoch": 2.1202650740407427, "grad_norm": 0.011854033917188644, "learning_rate": 5.3476285731347024e-06, "loss": 0.0009, "step": 129580 }, { "epoch": 2.1204286999918187, "grad_norm": 0.01117202453315258, "learning_rate": 5.346916344348203e-06, "loss": 0.0006, "step": 129590 }, { "epoch": 2.1205923259428947, "grad_norm": 0.029115479439496994, "learning_rate": 5.346204108488373e-06, "loss": 0.0012, "step": 129600 }, { "epoch": 2.1207559518939703, "grad_norm": 0.012176436372101307, "learning_rate": 5.345491865569737e-06, "loss": 0.0005, "step": 129610 }, { "epoch": 2.1209195778450463, "grad_norm": 0.003124251728877425, "learning_rate": 5.344779615606815e-06, "loss": 0.0016, "step": 129620 }, { "epoch": 2.1210832037961223, "grad_norm": 0.04505382478237152, "learning_rate": 5.344067358614129e-06, "loss": 0.0012, "step": 129630 }, { "epoch": 2.121246829747198, "grad_norm": 0.008172710426151752, "learning_rate": 5.343355094606203e-06, "loss": 0.0005, "step": 129640 }, { "epoch": 2.121410455698274, "grad_norm": 0.009493296034634113, "learning_rate": 5.342642823597559e-06, "loss": 0.0011, "step": 129650 }, { "epoch": 2.12157408164935, "grad_norm": 0.04829761013388634, "learning_rate": 5.341930545602718e-06, "loss": 0.0008, "step": 129660 }, { "epoch": 2.1217377076004253, "grad_norm": 0.023199399933218956, "learning_rate": 5.341218260636205e-06, "loss": 0.001, "step": 129670 }, { "epoch": 2.1219013335515013, "grad_norm": 0.04010861739516258, "learning_rate": 5.3405059687125415e-06, "loss": 0.0008, "step": 129680 }, { "epoch": 2.122064959502577, "grad_norm": 0.04834034666419029, "learning_rate": 5.339793669846252e-06, "loss": 0.001, "step": 129690 }, { "epoch": 2.122228585453653, "grad_norm": 0.06072433292865753, "learning_rate": 5.339081364051857e-06, "loss": 0.0012, "step": 129700 }, { "epoch": 2.122392211404729, "grad_norm": 0.03965955227613449, "learning_rate": 5.338369051343884e-06, "loss": 0.0006, "step": 129710 }, { "epoch": 2.1225558373558044, "grad_norm": 0.04208150878548622, "learning_rate": 5.337656731736851e-06, "loss": 0.0012, "step": 129720 }, { "epoch": 2.1227194633068804, "grad_norm": 0.02989141456782818, "learning_rate": 5.336944405245286e-06, "loss": 0.0009, "step": 129730 }, { "epoch": 2.1228830892579564, "grad_norm": 0.03804560378193855, "learning_rate": 5.336232071883709e-06, "loss": 0.0008, "step": 129740 }, { "epoch": 2.123046715209032, "grad_norm": 0.0019249145407229662, "learning_rate": 5.335519731666648e-06, "loss": 0.0003, "step": 129750 }, { "epoch": 2.123210341160108, "grad_norm": 0.049140091985464096, "learning_rate": 5.3348073846086235e-06, "loss": 0.0018, "step": 129760 }, { "epoch": 2.123373967111184, "grad_norm": 0.055944815278053284, "learning_rate": 5.334095030724163e-06, "loss": 0.0007, "step": 129770 }, { "epoch": 2.1235375930622595, "grad_norm": 0.016953863203525543, "learning_rate": 5.3333826700277866e-06, "loss": 0.001, "step": 129780 }, { "epoch": 2.1237012190133355, "grad_norm": 0.017854493111371994, "learning_rate": 5.332670302534023e-06, "loss": 0.001, "step": 129790 }, { "epoch": 2.1238648449644115, "grad_norm": 0.018906498327851295, "learning_rate": 5.3319579282573945e-06, "loss": 0.0005, "step": 129800 }, { "epoch": 2.124028470915487, "grad_norm": 0.007138936314731836, "learning_rate": 5.331245547212427e-06, "loss": 0.0007, "step": 129810 }, { "epoch": 2.124192096866563, "grad_norm": 0.05078347027301788, "learning_rate": 5.330533159413642e-06, "loss": 0.001, "step": 129820 }, { "epoch": 2.124355722817639, "grad_norm": 0.19167941808700562, "learning_rate": 5.329820764875569e-06, "loss": 0.0008, "step": 129830 }, { "epoch": 2.1245193487687146, "grad_norm": 0.003386640688404441, "learning_rate": 5.32910836361273e-06, "loss": 0.0007, "step": 129840 }, { "epoch": 2.1246829747197906, "grad_norm": 0.009739565663039684, "learning_rate": 5.3283959556396525e-06, "loss": 0.0006, "step": 129850 }, { "epoch": 2.1248466006708666, "grad_norm": 0.02075538970530033, "learning_rate": 5.3276835409708605e-06, "loss": 0.0007, "step": 129860 }, { "epoch": 2.125010226621942, "grad_norm": 0.03974950686097145, "learning_rate": 5.326971119620879e-06, "loss": 0.0009, "step": 129870 }, { "epoch": 2.125173852573018, "grad_norm": 0.12099061906337738, "learning_rate": 5.326258691604235e-06, "loss": 0.0019, "step": 129880 }, { "epoch": 2.1253374785240937, "grad_norm": 0.047795042395591736, "learning_rate": 5.325546256935454e-06, "loss": 0.0011, "step": 129890 }, { "epoch": 2.1255011044751697, "grad_norm": 0.06397563219070435, "learning_rate": 5.3248338156290615e-06, "loss": 0.0012, "step": 129900 }, { "epoch": 2.1256647304262457, "grad_norm": 0.09876074641942978, "learning_rate": 5.324121367699584e-06, "loss": 0.001, "step": 129910 }, { "epoch": 2.1258283563773213, "grad_norm": 0.06970450282096863, "learning_rate": 5.323408913161546e-06, "loss": 0.0008, "step": 129920 }, { "epoch": 2.1259919823283973, "grad_norm": 0.05393698439002037, "learning_rate": 5.322696452029478e-06, "loss": 0.0008, "step": 129930 }, { "epoch": 2.1261556082794733, "grad_norm": 0.03757818415760994, "learning_rate": 5.321983984317901e-06, "loss": 0.0007, "step": 129940 }, { "epoch": 2.126319234230549, "grad_norm": 0.08815275877714157, "learning_rate": 5.321271510041346e-06, "loss": 0.0013, "step": 129950 }, { "epoch": 2.126482860181625, "grad_norm": 0.013355598784983158, "learning_rate": 5.320559029214338e-06, "loss": 0.0012, "step": 129960 }, { "epoch": 2.126646486132701, "grad_norm": 0.019976625218987465, "learning_rate": 5.319846541851403e-06, "loss": 0.001, "step": 129970 }, { "epoch": 2.1268101120837763, "grad_norm": 0.13280293345451355, "learning_rate": 5.319134047967072e-06, "loss": 0.0015, "step": 129980 }, { "epoch": 2.1269737380348523, "grad_norm": 0.021376628428697586, "learning_rate": 5.318421547575865e-06, "loss": 0.0003, "step": 129990 }, { "epoch": 2.1271373639859283, "grad_norm": 0.013368922285735607, "learning_rate": 5.317709040692316e-06, "loss": 0.0011, "step": 130000 }, { "epoch": 2.127300989937004, "grad_norm": 0.0034727919846773148, "learning_rate": 5.316996527330948e-06, "loss": 0.0004, "step": 130010 }, { "epoch": 2.12746461588808, "grad_norm": 0.03470517694950104, "learning_rate": 5.316284007506294e-06, "loss": 0.0007, "step": 130020 }, { "epoch": 2.127628241839156, "grad_norm": 0.05521965026855469, "learning_rate": 5.315571481232875e-06, "loss": 0.0006, "step": 130030 }, { "epoch": 2.1277918677902314, "grad_norm": 0.06448023021221161, "learning_rate": 5.314858948525223e-06, "loss": 0.0009, "step": 130040 }, { "epoch": 2.1279554937413074, "grad_norm": 0.05319356918334961, "learning_rate": 5.314146409397863e-06, "loss": 0.0009, "step": 130050 }, { "epoch": 2.128119119692383, "grad_norm": 0.03503270447254181, "learning_rate": 5.313433863865328e-06, "loss": 0.0011, "step": 130060 }, { "epoch": 2.128282745643459, "grad_norm": 0.06261865049600601, "learning_rate": 5.31272131194214e-06, "loss": 0.0009, "step": 130070 }, { "epoch": 2.128446371594535, "grad_norm": 0.016067953780293465, "learning_rate": 5.312008753642834e-06, "loss": 0.0007, "step": 130080 }, { "epoch": 2.1286099975456105, "grad_norm": 0.0480027012526989, "learning_rate": 5.311296188981932e-06, "loss": 0.0014, "step": 130090 }, { "epoch": 2.1287736234966865, "grad_norm": 0.031282126903533936, "learning_rate": 5.310583617973966e-06, "loss": 0.0011, "step": 130100 }, { "epoch": 2.1289372494477625, "grad_norm": 0.023497184738516808, "learning_rate": 5.309871040633464e-06, "loss": 0.0006, "step": 130110 }, { "epoch": 2.129100875398838, "grad_norm": 0.012691405601799488, "learning_rate": 5.309158456974955e-06, "loss": 0.0012, "step": 130120 }, { "epoch": 2.129264501349914, "grad_norm": 0.022999603301286697, "learning_rate": 5.3084458670129676e-06, "loss": 0.0005, "step": 130130 }, { "epoch": 2.12942812730099, "grad_norm": 0.0340847410261631, "learning_rate": 5.307733270762033e-06, "loss": 0.0006, "step": 130140 }, { "epoch": 2.1295917532520656, "grad_norm": 0.003112656995654106, "learning_rate": 5.307020668236677e-06, "loss": 0.0013, "step": 130150 }, { "epoch": 2.1297553792031416, "grad_norm": 0.007306591607630253, "learning_rate": 5.306308059451433e-06, "loss": 0.0009, "step": 130160 }, { "epoch": 2.1299190051542176, "grad_norm": 0.03640034422278404, "learning_rate": 5.305595444420826e-06, "loss": 0.0005, "step": 130170 }, { "epoch": 2.130082631105293, "grad_norm": 0.03545380383729935, "learning_rate": 5.304882823159389e-06, "loss": 0.001, "step": 130180 }, { "epoch": 2.130246257056369, "grad_norm": 0.0415743812918663, "learning_rate": 5.304170195681649e-06, "loss": 0.0008, "step": 130190 }, { "epoch": 2.130409883007445, "grad_norm": 0.01031255442649126, "learning_rate": 5.303457562002139e-06, "loss": 0.0011, "step": 130200 }, { "epoch": 2.1305735089585207, "grad_norm": 0.037761591374874115, "learning_rate": 5.3027449221353865e-06, "loss": 0.0006, "step": 130210 }, { "epoch": 2.1307371349095967, "grad_norm": 0.034005410969257355, "learning_rate": 5.302032276095923e-06, "loss": 0.0016, "step": 130220 }, { "epoch": 2.1309007608606727, "grad_norm": 0.04912678897380829, "learning_rate": 5.301319623898278e-06, "loss": 0.0012, "step": 130230 }, { "epoch": 2.1310643868117483, "grad_norm": 0.06344347447156906, "learning_rate": 5.300606965556983e-06, "loss": 0.0009, "step": 130240 }, { "epoch": 2.1312280127628243, "grad_norm": 0.028982026502490044, "learning_rate": 5.299894301086565e-06, "loss": 0.0007, "step": 130250 }, { "epoch": 2.1313916387139002, "grad_norm": 0.004801479633897543, "learning_rate": 5.29918163050156e-06, "loss": 0.0007, "step": 130260 }, { "epoch": 2.131555264664976, "grad_norm": 0.03920265659689903, "learning_rate": 5.2984689538164945e-06, "loss": 0.0006, "step": 130270 }, { "epoch": 2.131718890616052, "grad_norm": 0.045925430953502655, "learning_rate": 5.297756271045902e-06, "loss": 0.0006, "step": 130280 }, { "epoch": 2.1318825165671274, "grad_norm": 0.06356952339410782, "learning_rate": 5.297043582204312e-06, "loss": 0.0019, "step": 130290 }, { "epoch": 2.1320461425182033, "grad_norm": 0.031928736716508865, "learning_rate": 5.296330887306256e-06, "loss": 0.0011, "step": 130300 }, { "epoch": 2.1322097684692793, "grad_norm": 0.05612688139081001, "learning_rate": 5.295618186366265e-06, "loss": 0.0008, "step": 130310 }, { "epoch": 2.132373394420355, "grad_norm": 0.014232456684112549, "learning_rate": 5.2949054793988695e-06, "loss": 0.0007, "step": 130320 }, { "epoch": 2.132537020371431, "grad_norm": 0.006215374451130629, "learning_rate": 5.294192766418604e-06, "loss": 0.0009, "step": 130330 }, { "epoch": 2.132700646322507, "grad_norm": 0.15468010306358337, "learning_rate": 5.2934800474399976e-06, "loss": 0.0016, "step": 130340 }, { "epoch": 2.1328642722735824, "grad_norm": 0.024466222152113914, "learning_rate": 5.292767322477582e-06, "loss": 0.0013, "step": 130350 }, { "epoch": 2.1330278982246584, "grad_norm": 0.09748148173093796, "learning_rate": 5.29205459154589e-06, "loss": 0.0013, "step": 130360 }, { "epoch": 2.1331915241757344, "grad_norm": 0.014625366777181625, "learning_rate": 5.291341854659453e-06, "loss": 0.0008, "step": 130370 }, { "epoch": 2.13335515012681, "grad_norm": 0.019417569041252136, "learning_rate": 5.290629111832804e-06, "loss": 0.0007, "step": 130380 }, { "epoch": 2.133518776077886, "grad_norm": 0.03474685922265053, "learning_rate": 5.289916363080474e-06, "loss": 0.0013, "step": 130390 }, { "epoch": 2.133682402028962, "grad_norm": 0.022362008690834045, "learning_rate": 5.2892036084169965e-06, "loss": 0.0013, "step": 130400 }, { "epoch": 2.1338460279800375, "grad_norm": 0.08029027283191681, "learning_rate": 5.288490847856903e-06, "loss": 0.0011, "step": 130410 }, { "epoch": 2.1340096539311135, "grad_norm": 0.025793028995394707, "learning_rate": 5.287778081414726e-06, "loss": 0.0012, "step": 130420 }, { "epoch": 2.1341732798821895, "grad_norm": 0.024118058383464813, "learning_rate": 5.287065309105e-06, "loss": 0.0007, "step": 130430 }, { "epoch": 2.134336905833265, "grad_norm": 0.018330005928874016, "learning_rate": 5.286352530942255e-06, "loss": 0.0022, "step": 130440 }, { "epoch": 2.134500531784341, "grad_norm": 0.05640363693237305, "learning_rate": 5.285639746941027e-06, "loss": 0.0015, "step": 130450 }, { "epoch": 2.1346641577354166, "grad_norm": 0.0494956448674202, "learning_rate": 5.284926957115846e-06, "loss": 0.0011, "step": 130460 }, { "epoch": 2.1348277836864926, "grad_norm": 0.02282514050602913, "learning_rate": 5.284214161481248e-06, "loss": 0.0005, "step": 130470 }, { "epoch": 2.1349914096375686, "grad_norm": 0.045750342309474945, "learning_rate": 5.283501360051764e-06, "loss": 0.0004, "step": 130480 }, { "epoch": 2.135155035588644, "grad_norm": 0.024734018370509148, "learning_rate": 5.282788552841928e-06, "loss": 0.0005, "step": 130490 }, { "epoch": 2.13531866153972, "grad_norm": 0.035429731011390686, "learning_rate": 5.282075739866275e-06, "loss": 0.0005, "step": 130500 }, { "epoch": 2.135482287490796, "grad_norm": 0.020834079012274742, "learning_rate": 5.2813629211393365e-06, "loss": 0.0008, "step": 130510 }, { "epoch": 2.1356459134418717, "grad_norm": 0.04906206578016281, "learning_rate": 5.280650096675648e-06, "loss": 0.0011, "step": 130520 }, { "epoch": 2.1358095393929477, "grad_norm": 0.06923519819974899, "learning_rate": 5.279937266489743e-06, "loss": 0.0007, "step": 130530 }, { "epoch": 2.1359731653440237, "grad_norm": 0.15300767123699188, "learning_rate": 5.279224430596155e-06, "loss": 0.0014, "step": 130540 }, { "epoch": 2.1361367912950993, "grad_norm": 0.011935463175177574, "learning_rate": 5.278511589009418e-06, "loss": 0.0011, "step": 130550 }, { "epoch": 2.1363004172461753, "grad_norm": 0.10168244689702988, "learning_rate": 5.277798741744066e-06, "loss": 0.001, "step": 130560 }, { "epoch": 2.1364640431972513, "grad_norm": 0.10552670061588287, "learning_rate": 5.277085888814636e-06, "loss": 0.0011, "step": 130570 }, { "epoch": 2.136627669148327, "grad_norm": 0.050991497933864594, "learning_rate": 5.276373030235658e-06, "loss": 0.0012, "step": 130580 }, { "epoch": 2.136791295099403, "grad_norm": 0.006043323781341314, "learning_rate": 5.27566016602167e-06, "loss": 0.0018, "step": 130590 }, { "epoch": 2.136954921050479, "grad_norm": 0.030017323791980743, "learning_rate": 5.274947296187205e-06, "loss": 0.0011, "step": 130600 }, { "epoch": 2.1371185470015543, "grad_norm": 0.07391556352376938, "learning_rate": 5.2742344207468e-06, "loss": 0.0009, "step": 130610 }, { "epoch": 2.1372821729526303, "grad_norm": 0.06511707603931427, "learning_rate": 5.273521539714989e-06, "loss": 0.0011, "step": 130620 }, { "epoch": 2.1374457989037063, "grad_norm": 0.1141660287976265, "learning_rate": 5.272808653106304e-06, "loss": 0.0009, "step": 130630 }, { "epoch": 2.137609424854782, "grad_norm": 0.0502573624253273, "learning_rate": 5.272095760935284e-06, "loss": 0.0006, "step": 130640 }, { "epoch": 2.137773050805858, "grad_norm": 0.06888173520565033, "learning_rate": 5.271382863216461e-06, "loss": 0.0009, "step": 130650 }, { "epoch": 2.137936676756934, "grad_norm": 0.03341297432780266, "learning_rate": 5.270669959964374e-06, "loss": 0.0005, "step": 130660 }, { "epoch": 2.1381003027080094, "grad_norm": 0.03564421460032463, "learning_rate": 5.269957051193556e-06, "loss": 0.0006, "step": 130670 }, { "epoch": 2.1382639286590854, "grad_norm": 0.0280919186770916, "learning_rate": 5.269244136918544e-06, "loss": 0.0009, "step": 130680 }, { "epoch": 2.138427554610161, "grad_norm": 0.0409596785902977, "learning_rate": 5.2685312171538725e-06, "loss": 0.0009, "step": 130690 }, { "epoch": 2.138591180561237, "grad_norm": 0.07359869033098221, "learning_rate": 5.267818291914078e-06, "loss": 0.0009, "step": 130700 }, { "epoch": 2.138754806512313, "grad_norm": 0.01179539691656828, "learning_rate": 5.2671053612136965e-06, "loss": 0.0015, "step": 130710 }, { "epoch": 2.1389184324633885, "grad_norm": 0.1284937709569931, "learning_rate": 5.266392425067264e-06, "loss": 0.0015, "step": 130720 }, { "epoch": 2.1390820584144645, "grad_norm": 0.024208204820752144, "learning_rate": 5.265679483489315e-06, "loss": 0.001, "step": 130730 }, { "epoch": 2.1392456843655405, "grad_norm": 0.06745673716068268, "learning_rate": 5.2649665364943896e-06, "loss": 0.0017, "step": 130740 }, { "epoch": 2.139409310316616, "grad_norm": 0.019333714619278908, "learning_rate": 5.26425358409702e-06, "loss": 0.0006, "step": 130750 }, { "epoch": 2.139572936267692, "grad_norm": 0.057479847222566605, "learning_rate": 5.2635406263117465e-06, "loss": 0.0008, "step": 130760 }, { "epoch": 2.139736562218768, "grad_norm": 0.04267285391688347, "learning_rate": 5.262827663153102e-06, "loss": 0.0007, "step": 130770 }, { "epoch": 2.1399001881698436, "grad_norm": 0.021848302334547043, "learning_rate": 5.262114694635625e-06, "loss": 0.0005, "step": 130780 }, { "epoch": 2.1400638141209196, "grad_norm": 0.04148397967219353, "learning_rate": 5.261401720773853e-06, "loss": 0.0009, "step": 130790 }, { "epoch": 2.1402274400719956, "grad_norm": 0.00517183868214488, "learning_rate": 5.2606887415823225e-06, "loss": 0.0007, "step": 130800 }, { "epoch": 2.140391066023071, "grad_norm": 0.04040100425481796, "learning_rate": 5.259975757075569e-06, "loss": 0.0006, "step": 130810 }, { "epoch": 2.140554691974147, "grad_norm": 0.025784984230995178, "learning_rate": 5.259262767268132e-06, "loss": 0.0007, "step": 130820 }, { "epoch": 2.1407183179252227, "grad_norm": 0.0024113692343235016, "learning_rate": 5.2585497721745475e-06, "loss": 0.0006, "step": 130830 }, { "epoch": 2.1408819438762987, "grad_norm": 0.04443611949682236, "learning_rate": 5.257836771809352e-06, "loss": 0.0005, "step": 130840 }, { "epoch": 2.1410455698273747, "grad_norm": 0.004580482840538025, "learning_rate": 5.257123766187086e-06, "loss": 0.0014, "step": 130850 }, { "epoch": 2.1412091957784503, "grad_norm": 0.022253509610891342, "learning_rate": 5.2564107553222845e-06, "loss": 0.0008, "step": 130860 }, { "epoch": 2.1413728217295263, "grad_norm": 0.03969595208764076, "learning_rate": 5.255697739229484e-06, "loss": 0.001, "step": 130870 }, { "epoch": 2.1415364476806023, "grad_norm": 0.010752481408417225, "learning_rate": 5.254984717923226e-06, "loss": 0.0005, "step": 130880 }, { "epoch": 2.141700073631678, "grad_norm": 0.05658755078911781, "learning_rate": 5.254271691418046e-06, "loss": 0.0005, "step": 130890 }, { "epoch": 2.141863699582754, "grad_norm": 0.03210613131523132, "learning_rate": 5.253558659728482e-06, "loss": 0.0014, "step": 130900 }, { "epoch": 2.14202732553383, "grad_norm": 0.0016502741491422057, "learning_rate": 5.252845622869072e-06, "loss": 0.0003, "step": 130910 }, { "epoch": 2.1421909514849053, "grad_norm": 0.09646958857774734, "learning_rate": 5.2521325808543565e-06, "loss": 0.001, "step": 130920 }, { "epoch": 2.1423545774359813, "grad_norm": 0.1595163196325302, "learning_rate": 5.2514195336988715e-06, "loss": 0.0014, "step": 130930 }, { "epoch": 2.1425182033870573, "grad_norm": 0.026851836591959, "learning_rate": 5.250706481417156e-06, "loss": 0.0004, "step": 130940 }, { "epoch": 2.142681829338133, "grad_norm": 0.014901410788297653, "learning_rate": 5.249993424023748e-06, "loss": 0.0009, "step": 130950 }, { "epoch": 2.142845455289209, "grad_norm": 0.046922467648983, "learning_rate": 5.249280361533187e-06, "loss": 0.001, "step": 130960 }, { "epoch": 2.143009081240285, "grad_norm": 0.01013029832392931, "learning_rate": 5.248567293960012e-06, "loss": 0.0008, "step": 130970 }, { "epoch": 2.1431727071913604, "grad_norm": 0.12095099687576294, "learning_rate": 5.24785422131876e-06, "loss": 0.0009, "step": 130980 }, { "epoch": 2.1433363331424364, "grad_norm": 0.06877543032169342, "learning_rate": 5.247141143623973e-06, "loss": 0.0014, "step": 130990 }, { "epoch": 2.1434999590935124, "grad_norm": 0.0248185433447361, "learning_rate": 5.246428060890186e-06, "loss": 0.0013, "step": 131000 }, { "epoch": 2.143663585044588, "grad_norm": 0.06270477175712585, "learning_rate": 5.245714973131942e-06, "loss": 0.0015, "step": 131010 }, { "epoch": 2.143827210995664, "grad_norm": 0.002328480128198862, "learning_rate": 5.2450018803637784e-06, "loss": 0.0016, "step": 131020 }, { "epoch": 2.14399083694674, "grad_norm": 0.08520756661891937, "learning_rate": 5.244288782600236e-06, "loss": 0.001, "step": 131030 }, { "epoch": 2.1441544628978155, "grad_norm": 0.09134002774953842, "learning_rate": 5.24357567985585e-06, "loss": 0.0011, "step": 131040 }, { "epoch": 2.1443180888488915, "grad_norm": 0.01516138669103384, "learning_rate": 5.242862572145166e-06, "loss": 0.0008, "step": 131050 }, { "epoch": 2.144481714799967, "grad_norm": 0.03203577920794487, "learning_rate": 5.242149459482718e-06, "loss": 0.001, "step": 131060 }, { "epoch": 2.144645340751043, "grad_norm": 0.02473059855401516, "learning_rate": 5.241436341883051e-06, "loss": 0.0011, "step": 131070 }, { "epoch": 2.144808966702119, "grad_norm": 0.22028937935829163, "learning_rate": 5.240723219360701e-06, "loss": 0.0008, "step": 131080 }, { "epoch": 2.1449725926531946, "grad_norm": 0.036953553557395935, "learning_rate": 5.24001009193021e-06, "loss": 0.001, "step": 131090 }, { "epoch": 2.1451362186042706, "grad_norm": 0.028811555355787277, "learning_rate": 5.239296959606116e-06, "loss": 0.0015, "step": 131100 }, { "epoch": 2.1452998445553466, "grad_norm": 0.037954892963171005, "learning_rate": 5.2385838224029625e-06, "loss": 0.0013, "step": 131110 }, { "epoch": 2.145463470506422, "grad_norm": 0.029111353680491447, "learning_rate": 5.237870680335286e-06, "loss": 0.0008, "step": 131120 }, { "epoch": 2.145627096457498, "grad_norm": 0.023605475202202797, "learning_rate": 5.237157533417631e-06, "loss": 0.0006, "step": 131130 }, { "epoch": 2.145790722408574, "grad_norm": 0.008182669058442116, "learning_rate": 5.236444381664532e-06, "loss": 0.0013, "step": 131140 }, { "epoch": 2.1459543483596497, "grad_norm": 0.004822615534067154, "learning_rate": 5.235731225090536e-06, "loss": 0.0012, "step": 131150 }, { "epoch": 2.1461179743107257, "grad_norm": 0.023052264004945755, "learning_rate": 5.23501806371018e-06, "loss": 0.0008, "step": 131160 }, { "epoch": 2.1462816002618017, "grad_norm": 0.004947153851389885, "learning_rate": 5.234304897538005e-06, "loss": 0.0007, "step": 131170 }, { "epoch": 2.1464452262128773, "grad_norm": 0.09121137857437134, "learning_rate": 5.233591726588552e-06, "loss": 0.0016, "step": 131180 }, { "epoch": 2.1466088521639533, "grad_norm": 0.0017293813871219754, "learning_rate": 5.232878550876365e-06, "loss": 0.0011, "step": 131190 }, { "epoch": 2.1467724781150292, "grad_norm": 0.017006298527121544, "learning_rate": 5.232165370415979e-06, "loss": 0.0012, "step": 131200 }, { "epoch": 2.146936104066105, "grad_norm": 0.07409606128931046, "learning_rate": 5.231452185221942e-06, "loss": 0.001, "step": 131210 }, { "epoch": 2.147099730017181, "grad_norm": 0.03693300485610962, "learning_rate": 5.230738995308788e-06, "loss": 0.0007, "step": 131220 }, { "epoch": 2.1472633559682563, "grad_norm": 0.007763877511024475, "learning_rate": 5.230025800691066e-06, "loss": 0.0006, "step": 131230 }, { "epoch": 2.1474269819193323, "grad_norm": 0.02531830593943596, "learning_rate": 5.229312601383311e-06, "loss": 0.0005, "step": 131240 }, { "epoch": 2.1475906078704083, "grad_norm": 0.06319885700941086, "learning_rate": 5.2285993974000685e-06, "loss": 0.001, "step": 131250 }, { "epoch": 2.147754233821484, "grad_norm": 0.04939579218626022, "learning_rate": 5.227886188755878e-06, "loss": 0.0009, "step": 131260 }, { "epoch": 2.14791785977256, "grad_norm": 0.03748093172907829, "learning_rate": 5.227172975465281e-06, "loss": 0.0007, "step": 131270 }, { "epoch": 2.148081485723636, "grad_norm": 0.09774491935968399, "learning_rate": 5.226459757542822e-06, "loss": 0.0013, "step": 131280 }, { "epoch": 2.1482451116747114, "grad_norm": 0.09450750797986984, "learning_rate": 5.225746535003039e-06, "loss": 0.0017, "step": 131290 }, { "epoch": 2.1484087376257874, "grad_norm": 0.02158893086016178, "learning_rate": 5.2250333078604785e-06, "loss": 0.0007, "step": 131300 }, { "epoch": 2.1485723635768634, "grad_norm": 0.2529752552509308, "learning_rate": 5.224320076129679e-06, "loss": 0.0012, "step": 131310 }, { "epoch": 2.148735989527939, "grad_norm": 0.1581340879201889, "learning_rate": 5.223606839825184e-06, "loss": 0.0012, "step": 131320 }, { "epoch": 2.148899615479015, "grad_norm": 0.0018618417670950294, "learning_rate": 5.222893598961535e-06, "loss": 0.0006, "step": 131330 }, { "epoch": 2.149063241430091, "grad_norm": 0.05844445526599884, "learning_rate": 5.222180353553277e-06, "loss": 0.0008, "step": 131340 }, { "epoch": 2.1492268673811665, "grad_norm": 0.010028311982750893, "learning_rate": 5.221467103614948e-06, "loss": 0.0005, "step": 131350 }, { "epoch": 2.1493904933322425, "grad_norm": 0.0443674698472023, "learning_rate": 5.220753849161095e-06, "loss": 0.001, "step": 131360 }, { "epoch": 2.1495541192833185, "grad_norm": 0.0024992350954562426, "learning_rate": 5.220040590206258e-06, "loss": 0.0011, "step": 131370 }, { "epoch": 2.149717745234394, "grad_norm": 0.3711538314819336, "learning_rate": 5.219327326764981e-06, "loss": 0.0007, "step": 131380 }, { "epoch": 2.14988137118547, "grad_norm": 0.03949323669075966, "learning_rate": 5.218614058851805e-06, "loss": 0.0005, "step": 131390 }, { "epoch": 2.150044997136546, "grad_norm": 0.016515251249074936, "learning_rate": 5.217900786481277e-06, "loss": 0.0006, "step": 131400 }, { "epoch": 2.1502086230876216, "grad_norm": 0.02815687656402588, "learning_rate": 5.217187509667935e-06, "loss": 0.0015, "step": 131410 }, { "epoch": 2.1503722490386976, "grad_norm": 0.023872796446084976, "learning_rate": 5.2164742284263244e-06, "loss": 0.0006, "step": 131420 }, { "epoch": 2.1505358749897736, "grad_norm": 0.0414656363427639, "learning_rate": 5.215760942770989e-06, "loss": 0.0007, "step": 131430 }, { "epoch": 2.150699500940849, "grad_norm": 0.0038417833857238293, "learning_rate": 5.215047652716473e-06, "loss": 0.0009, "step": 131440 }, { "epoch": 2.150863126891925, "grad_norm": 0.037373628467321396, "learning_rate": 5.214334358277316e-06, "loss": 0.0013, "step": 131450 }, { "epoch": 2.1510267528430007, "grad_norm": 0.07512473315000534, "learning_rate": 5.213621059468065e-06, "loss": 0.0011, "step": 131460 }, { "epoch": 2.1511903787940767, "grad_norm": 0.04484357312321663, "learning_rate": 5.21290775630326e-06, "loss": 0.0009, "step": 131470 }, { "epoch": 2.1513540047451527, "grad_norm": 0.003340332768857479, "learning_rate": 5.212194448797449e-06, "loss": 0.0007, "step": 131480 }, { "epoch": 2.1515176306962283, "grad_norm": 0.053547453135252, "learning_rate": 5.211481136965174e-06, "loss": 0.0014, "step": 131490 }, { "epoch": 2.1516812566473043, "grad_norm": 0.004217978101223707, "learning_rate": 5.210767820820978e-06, "loss": 0.0011, "step": 131500 }, { "epoch": 2.1518448825983802, "grad_norm": 0.035796813666820526, "learning_rate": 5.210054500379405e-06, "loss": 0.0007, "step": 131510 }, { "epoch": 2.152008508549456, "grad_norm": 0.00391038041561842, "learning_rate": 5.209341175655e-06, "loss": 0.0011, "step": 131520 }, { "epoch": 2.152172134500532, "grad_norm": 0.009036663919687271, "learning_rate": 5.208627846662305e-06, "loss": 0.0008, "step": 131530 }, { "epoch": 2.152335760451608, "grad_norm": 0.028337445110082626, "learning_rate": 5.207914513415867e-06, "loss": 0.0007, "step": 131540 }, { "epoch": 2.1524993864026833, "grad_norm": 0.03241876885294914, "learning_rate": 5.207201175930229e-06, "loss": 0.0007, "step": 131550 }, { "epoch": 2.1526630123537593, "grad_norm": 0.07593461871147156, "learning_rate": 5.206487834219934e-06, "loss": 0.0008, "step": 131560 }, { "epoch": 2.1528266383048353, "grad_norm": 0.056191835552453995, "learning_rate": 5.2057744882995295e-06, "loss": 0.0012, "step": 131570 }, { "epoch": 2.152990264255911, "grad_norm": 0.10566790401935577, "learning_rate": 5.205061138183557e-06, "loss": 0.0013, "step": 131580 }, { "epoch": 2.153153890206987, "grad_norm": 0.20859529078006744, "learning_rate": 5.204347783886563e-06, "loss": 0.0019, "step": 131590 }, { "epoch": 2.1533175161580624, "grad_norm": 0.04539180546998978, "learning_rate": 5.20363442542309e-06, "loss": 0.0008, "step": 131600 }, { "epoch": 2.1534811421091384, "grad_norm": 0.0005642681499011815, "learning_rate": 5.202921062807685e-06, "loss": 0.0011, "step": 131610 }, { "epoch": 2.1536447680602144, "grad_norm": 0.05782206729054451, "learning_rate": 5.202207696054893e-06, "loss": 0.0009, "step": 131620 }, { "epoch": 2.15380839401129, "grad_norm": 0.044027797877788544, "learning_rate": 5.201494325179257e-06, "loss": 0.0006, "step": 131630 }, { "epoch": 2.153972019962366, "grad_norm": 0.0772567167878151, "learning_rate": 5.200780950195323e-06, "loss": 0.0006, "step": 131640 }, { "epoch": 2.154135645913442, "grad_norm": 0.004712289199233055, "learning_rate": 5.200067571117637e-06, "loss": 0.0015, "step": 131650 }, { "epoch": 2.1542992718645175, "grad_norm": 0.03873579204082489, "learning_rate": 5.1993541879607425e-06, "loss": 0.001, "step": 131660 }, { "epoch": 2.1544628978155935, "grad_norm": 0.20174147188663483, "learning_rate": 5.1986408007391864e-06, "loss": 0.0007, "step": 131670 }, { "epoch": 2.1546265237666695, "grad_norm": 0.01206426415592432, "learning_rate": 5.197927409467513e-06, "loss": 0.0005, "step": 131680 }, { "epoch": 2.154790149717745, "grad_norm": 0.021798307076096535, "learning_rate": 5.197214014160269e-06, "loss": 0.0007, "step": 131690 }, { "epoch": 2.154953775668821, "grad_norm": 0.05981908366084099, "learning_rate": 5.196500614831997e-06, "loss": 0.0004, "step": 131700 }, { "epoch": 2.155117401619897, "grad_norm": 0.030789880082011223, "learning_rate": 5.195787211497246e-06, "loss": 0.0008, "step": 131710 }, { "epoch": 2.1552810275709726, "grad_norm": 0.013004160486161709, "learning_rate": 5.195073804170559e-06, "loss": 0.0006, "step": 131720 }, { "epoch": 2.1554446535220486, "grad_norm": 0.36853957176208496, "learning_rate": 5.194360392866485e-06, "loss": 0.0011, "step": 131730 }, { "epoch": 2.1556082794731246, "grad_norm": 0.0356261320412159, "learning_rate": 5.193646977599566e-06, "loss": 0.0012, "step": 131740 }, { "epoch": 2.1557719054242, "grad_norm": 0.008618801832199097, "learning_rate": 5.192933558384351e-06, "loss": 0.0012, "step": 131750 }, { "epoch": 2.155935531375276, "grad_norm": 0.003272030269727111, "learning_rate": 5.1922201352353844e-06, "loss": 0.0008, "step": 131760 }, { "epoch": 2.156099157326352, "grad_norm": 0.0217570997774601, "learning_rate": 5.191506708167213e-06, "loss": 0.0007, "step": 131770 }, { "epoch": 2.1562627832774277, "grad_norm": 0.05000786855816841, "learning_rate": 5.190793277194383e-06, "loss": 0.0006, "step": 131780 }, { "epoch": 2.1564264092285037, "grad_norm": 0.058396607637405396, "learning_rate": 5.190079842331439e-06, "loss": 0.0011, "step": 131790 }, { "epoch": 2.1565900351795797, "grad_norm": 0.014092272147536278, "learning_rate": 5.189366403592929e-06, "loss": 0.0006, "step": 131800 }, { "epoch": 2.1567536611306553, "grad_norm": 0.10233177989721298, "learning_rate": 5.188652960993401e-06, "loss": 0.0007, "step": 131810 }, { "epoch": 2.1569172870817312, "grad_norm": 0.13708704710006714, "learning_rate": 5.1879395145473975e-06, "loss": 0.0009, "step": 131820 }, { "epoch": 2.157080913032807, "grad_norm": 0.04340042918920517, "learning_rate": 5.187226064269468e-06, "loss": 0.001, "step": 131830 }, { "epoch": 2.157244538983883, "grad_norm": 0.009770174510776997, "learning_rate": 5.186512610174157e-06, "loss": 0.0008, "step": 131840 }, { "epoch": 2.157408164934959, "grad_norm": 0.028034307062625885, "learning_rate": 5.185799152276014e-06, "loss": 0.0004, "step": 131850 }, { "epoch": 2.1575717908860343, "grad_norm": 0.0720149576663971, "learning_rate": 5.185085690589584e-06, "loss": 0.0011, "step": 131860 }, { "epoch": 2.1577354168371103, "grad_norm": 0.014210633002221584, "learning_rate": 5.184372225129415e-06, "loss": 0.0004, "step": 131870 }, { "epoch": 2.1578990427881863, "grad_norm": 0.05159645900130272, "learning_rate": 5.183658755910052e-06, "loss": 0.0007, "step": 131880 }, { "epoch": 2.158062668739262, "grad_norm": 0.010722938925027847, "learning_rate": 5.182945282946042e-06, "loss": 0.001, "step": 131890 }, { "epoch": 2.158226294690338, "grad_norm": 0.03662394732236862, "learning_rate": 5.182231806251936e-06, "loss": 0.001, "step": 131900 }, { "epoch": 2.158389920641414, "grad_norm": 0.0011895766947418451, "learning_rate": 5.181518325842277e-06, "loss": 0.0006, "step": 131910 }, { "epoch": 2.1585535465924894, "grad_norm": 0.04749349504709244, "learning_rate": 5.180804841731613e-06, "loss": 0.0007, "step": 131920 }, { "epoch": 2.1587171725435654, "grad_norm": 0.03131093457341194, "learning_rate": 5.180091353934493e-06, "loss": 0.0007, "step": 131930 }, { "epoch": 2.1588807984946414, "grad_norm": 0.0947180837392807, "learning_rate": 5.179377862465464e-06, "loss": 0.0011, "step": 131940 }, { "epoch": 2.159044424445717, "grad_norm": 0.0033844660501927137, "learning_rate": 5.178664367339072e-06, "loss": 0.001, "step": 131950 }, { "epoch": 2.159208050396793, "grad_norm": 0.09117348492145538, "learning_rate": 5.1779508685698654e-06, "loss": 0.0011, "step": 131960 }, { "epoch": 2.159371676347869, "grad_norm": 0.24991627037525177, "learning_rate": 5.177237366172392e-06, "loss": 0.0018, "step": 131970 }, { "epoch": 2.1595353022989445, "grad_norm": 0.02996715158224106, "learning_rate": 5.1765238601612e-06, "loss": 0.0022, "step": 131980 }, { "epoch": 2.1596989282500205, "grad_norm": 0.06241634115576744, "learning_rate": 5.175810350550836e-06, "loss": 0.0007, "step": 131990 }, { "epoch": 2.159862554201096, "grad_norm": 0.07654788345098495, "learning_rate": 5.17509683735585e-06, "loss": 0.0008, "step": 132000 }, { "epoch": 2.160026180152172, "grad_norm": 0.02889319881796837, "learning_rate": 5.174383320590786e-06, "loss": 0.0018, "step": 132010 }, { "epoch": 2.160189806103248, "grad_norm": 0.03564263880252838, "learning_rate": 5.173669800270196e-06, "loss": 0.0007, "step": 132020 }, { "epoch": 2.1603534320543236, "grad_norm": 0.049488615244627, "learning_rate": 5.172956276408626e-06, "loss": 0.0012, "step": 132030 }, { "epoch": 2.1605170580053996, "grad_norm": 0.03769952431321144, "learning_rate": 5.1722427490206254e-06, "loss": 0.0014, "step": 132040 }, { "epoch": 2.1606806839564756, "grad_norm": 0.06861207634210587, "learning_rate": 5.17152921812074e-06, "loss": 0.0007, "step": 132050 }, { "epoch": 2.160844309907551, "grad_norm": 0.0013049488188698888, "learning_rate": 5.170815683723521e-06, "loss": 0.0008, "step": 132060 }, { "epoch": 2.161007935858627, "grad_norm": 0.06284059584140778, "learning_rate": 5.170102145843514e-06, "loss": 0.0008, "step": 132070 }, { "epoch": 2.161171561809703, "grad_norm": 0.014020560309290886, "learning_rate": 5.169388604495271e-06, "loss": 0.0011, "step": 132080 }, { "epoch": 2.1613351877607787, "grad_norm": 0.027024727314710617, "learning_rate": 5.168675059693337e-06, "loss": 0.0009, "step": 132090 }, { "epoch": 2.1614988137118547, "grad_norm": 0.015122836455702782, "learning_rate": 5.167961511452263e-06, "loss": 0.0013, "step": 132100 }, { "epoch": 2.1616624396629307, "grad_norm": 0.03483263775706291, "learning_rate": 5.167247959786596e-06, "loss": 0.0006, "step": 132110 }, { "epoch": 2.1618260656140063, "grad_norm": 0.013220278546214104, "learning_rate": 5.166534404710885e-06, "loss": 0.0007, "step": 132120 }, { "epoch": 2.1619896915650822, "grad_norm": 0.0242755189538002, "learning_rate": 5.16582084623968e-06, "loss": 0.0005, "step": 132130 }, { "epoch": 2.1621533175161582, "grad_norm": 0.10978028178215027, "learning_rate": 5.165107284387528e-06, "loss": 0.0013, "step": 132140 }, { "epoch": 2.162316943467234, "grad_norm": 0.08660856634378433, "learning_rate": 5.164393719168979e-06, "loss": 0.0009, "step": 132150 }, { "epoch": 2.16248056941831, "grad_norm": 0.02735201083123684, "learning_rate": 5.163680150598583e-06, "loss": 0.0011, "step": 132160 }, { "epoch": 2.162644195369386, "grad_norm": 0.017108801752328873, "learning_rate": 5.162966578690887e-06, "loss": 0.0004, "step": 132170 }, { "epoch": 2.1628078213204613, "grad_norm": 0.06684187054634094, "learning_rate": 5.162253003460442e-06, "loss": 0.0011, "step": 132180 }, { "epoch": 2.1629714472715373, "grad_norm": 0.02526683360338211, "learning_rate": 5.161539424921794e-06, "loss": 0.0006, "step": 132190 }, { "epoch": 2.1631350732226133, "grad_norm": 0.06021735817193985, "learning_rate": 5.160825843089496e-06, "loss": 0.0016, "step": 132200 }, { "epoch": 2.163298699173689, "grad_norm": 0.04800250008702278, "learning_rate": 5.160112257978096e-06, "loss": 0.0005, "step": 132210 }, { "epoch": 2.163462325124765, "grad_norm": 0.015002584084868431, "learning_rate": 5.159398669602143e-06, "loss": 0.0007, "step": 132220 }, { "epoch": 2.1636259510758404, "grad_norm": 0.059495363384485245, "learning_rate": 5.158685077976188e-06, "loss": 0.0011, "step": 132230 }, { "epoch": 2.1637895770269164, "grad_norm": 0.035305771976709366, "learning_rate": 5.157971483114778e-06, "loss": 0.0018, "step": 132240 }, { "epoch": 2.1639532029779924, "grad_norm": 0.1670788824558258, "learning_rate": 5.1572578850324636e-06, "loss": 0.0016, "step": 132250 }, { "epoch": 2.164116828929068, "grad_norm": 0.10063536465167999, "learning_rate": 5.156544283743794e-06, "loss": 0.0009, "step": 132260 }, { "epoch": 2.164280454880144, "grad_norm": 0.061924345791339874, "learning_rate": 5.155830679263322e-06, "loss": 0.001, "step": 132270 }, { "epoch": 2.16444408083122, "grad_norm": 0.03143107146024704, "learning_rate": 5.155117071605592e-06, "loss": 0.0008, "step": 132280 }, { "epoch": 2.1646077067822955, "grad_norm": 0.06386993080377579, "learning_rate": 5.154403460785159e-06, "loss": 0.0007, "step": 132290 }, { "epoch": 2.1647713327333715, "grad_norm": 0.02148699201643467, "learning_rate": 5.1536898468165695e-06, "loss": 0.0006, "step": 132300 }, { "epoch": 2.1649349586844475, "grad_norm": 0.054101210087537766, "learning_rate": 5.152976229714376e-06, "loss": 0.0009, "step": 132310 }, { "epoch": 2.165098584635523, "grad_norm": 0.05230865254998207, "learning_rate": 5.1522626094931265e-06, "loss": 0.0004, "step": 132320 }, { "epoch": 2.165262210586599, "grad_norm": 0.0369931161403656, "learning_rate": 5.151548986167371e-06, "loss": 0.001, "step": 132330 }, { "epoch": 2.165425836537675, "grad_norm": 0.08612873405218124, "learning_rate": 5.150835359751661e-06, "loss": 0.0007, "step": 132340 }, { "epoch": 2.1655894624887506, "grad_norm": 0.03307497501373291, "learning_rate": 5.150121730260548e-06, "loss": 0.0009, "step": 132350 }, { "epoch": 2.1657530884398266, "grad_norm": 0.08045225590467453, "learning_rate": 5.149408097708578e-06, "loss": 0.0011, "step": 132360 }, { "epoch": 2.165916714390902, "grad_norm": 0.0744347795844078, "learning_rate": 5.1486944621103065e-06, "loss": 0.0008, "step": 132370 }, { "epoch": 2.166080340341978, "grad_norm": 0.0025595289189368486, "learning_rate": 5.14798082348028e-06, "loss": 0.0008, "step": 132380 }, { "epoch": 2.166243966293054, "grad_norm": 0.023133233189582825, "learning_rate": 5.147267181833051e-06, "loss": 0.0018, "step": 132390 }, { "epoch": 2.1664075922441297, "grad_norm": 0.10562750697135925, "learning_rate": 5.146553537183168e-06, "loss": 0.001, "step": 132400 }, { "epoch": 2.1665712181952057, "grad_norm": 0.022705424576997757, "learning_rate": 5.145839889545184e-06, "loss": 0.001, "step": 132410 }, { "epoch": 2.1667348441462817, "grad_norm": 0.017431247979402542, "learning_rate": 5.145126238933649e-06, "loss": 0.0009, "step": 132420 }, { "epoch": 2.1668984700973573, "grad_norm": 0.035757310688495636, "learning_rate": 5.144412585363113e-06, "loss": 0.0009, "step": 132430 }, { "epoch": 2.1670620960484333, "grad_norm": 0.039828818291425705, "learning_rate": 5.143698928848126e-06, "loss": 0.0006, "step": 132440 }, { "epoch": 2.1672257219995092, "grad_norm": 0.015247441828250885, "learning_rate": 5.142985269403243e-06, "loss": 0.0004, "step": 132450 }, { "epoch": 2.167389347950585, "grad_norm": 0.029897678643465042, "learning_rate": 5.14227160704301e-06, "loss": 0.0012, "step": 132460 }, { "epoch": 2.167552973901661, "grad_norm": 0.016736013814806938, "learning_rate": 5.141557941781981e-06, "loss": 0.0014, "step": 132470 }, { "epoch": 2.167716599852737, "grad_norm": 0.07268859446048737, "learning_rate": 5.140844273634704e-06, "loss": 0.0008, "step": 132480 }, { "epoch": 2.1678802258038123, "grad_norm": 0.003824051935225725, "learning_rate": 5.140130602615734e-06, "loss": 0.0007, "step": 132490 }, { "epoch": 2.1680438517548883, "grad_norm": 0.07306405156850815, "learning_rate": 5.139416928739619e-06, "loss": 0.0008, "step": 132500 }, { "epoch": 2.1682074777059643, "grad_norm": 0.010060307569801807, "learning_rate": 5.138703252020912e-06, "loss": 0.0009, "step": 132510 }, { "epoch": 2.16837110365704, "grad_norm": 0.03490087017416954, "learning_rate": 5.137989572474163e-06, "loss": 0.0009, "step": 132520 }, { "epoch": 2.168534729608116, "grad_norm": 0.09043283015489578, "learning_rate": 5.1372758901139254e-06, "loss": 0.0005, "step": 132530 }, { "epoch": 2.168698355559192, "grad_norm": 0.06616868078708649, "learning_rate": 5.136562204954748e-06, "loss": 0.0013, "step": 132540 }, { "epoch": 2.1688619815102674, "grad_norm": 0.0623522512614727, "learning_rate": 5.135848517011183e-06, "loss": 0.0013, "step": 132550 }, { "epoch": 2.1690256074613434, "grad_norm": 0.061029162257909775, "learning_rate": 5.135134826297783e-06, "loss": 0.0009, "step": 132560 }, { "epoch": 2.1691892334124194, "grad_norm": 0.01192642655223608, "learning_rate": 5.134421132829098e-06, "loss": 0.0007, "step": 132570 }, { "epoch": 2.169352859363495, "grad_norm": 0.07095392048358917, "learning_rate": 5.1337074366196825e-06, "loss": 0.0011, "step": 132580 }, { "epoch": 2.169516485314571, "grad_norm": 0.012128788977861404, "learning_rate": 5.1329937376840836e-06, "loss": 0.0006, "step": 132590 }, { "epoch": 2.1696801112656465, "grad_norm": 0.04423276707530022, "learning_rate": 5.132280036036858e-06, "loss": 0.0006, "step": 132600 }, { "epoch": 2.1698437372167225, "grad_norm": 0.03811779245734215, "learning_rate": 5.131566331692552e-06, "loss": 0.0012, "step": 132610 }, { "epoch": 2.1700073631677985, "grad_norm": 0.008911468088626862, "learning_rate": 5.130852624665723e-06, "loss": 0.0007, "step": 132620 }, { "epoch": 2.170170989118874, "grad_norm": 0.050865642726421356, "learning_rate": 5.130138914970918e-06, "loss": 0.0017, "step": 132630 }, { "epoch": 2.17033461506995, "grad_norm": 0.1268516182899475, "learning_rate": 5.129425202622693e-06, "loss": 0.0006, "step": 132640 }, { "epoch": 2.170498241021026, "grad_norm": 0.1378186196088791, "learning_rate": 5.128711487635597e-06, "loss": 0.0014, "step": 132650 }, { "epoch": 2.1706618669721016, "grad_norm": 0.012193955481052399, "learning_rate": 5.1279977700241845e-06, "loss": 0.001, "step": 132660 }, { "epoch": 2.1708254929231776, "grad_norm": 0.04963427409529686, "learning_rate": 5.127284049803005e-06, "loss": 0.0006, "step": 132670 }, { "epoch": 2.1709891188742536, "grad_norm": 0.015357007272541523, "learning_rate": 5.126570326986613e-06, "loss": 0.0011, "step": 132680 }, { "epoch": 2.171152744825329, "grad_norm": 0.01745641976594925, "learning_rate": 5.125856601589559e-06, "loss": 0.0003, "step": 132690 }, { "epoch": 2.171316370776405, "grad_norm": 0.0022422708570957184, "learning_rate": 5.125142873626396e-06, "loss": 0.0008, "step": 132700 }, { "epoch": 2.171479996727481, "grad_norm": 0.04071100428700447, "learning_rate": 5.124429143111677e-06, "loss": 0.0006, "step": 132710 }, { "epoch": 2.1716436226785567, "grad_norm": 0.047575704753398895, "learning_rate": 5.1237154100599525e-06, "loss": 0.0009, "step": 132720 }, { "epoch": 2.1718072486296327, "grad_norm": 0.040776513516902924, "learning_rate": 5.123001674485777e-06, "loss": 0.0007, "step": 132730 }, { "epoch": 2.1719708745807087, "grad_norm": 0.06302850693464279, "learning_rate": 5.122287936403701e-06, "loss": 0.0014, "step": 132740 }, { "epoch": 2.1721345005317843, "grad_norm": 0.025603322312235832, "learning_rate": 5.121574195828279e-06, "loss": 0.0012, "step": 132750 }, { "epoch": 2.1722981264828602, "grad_norm": 0.05228763446211815, "learning_rate": 5.120860452774062e-06, "loss": 0.001, "step": 132760 }, { "epoch": 2.172461752433936, "grad_norm": 0.037727732211351395, "learning_rate": 5.120146707255604e-06, "loss": 0.0006, "step": 132770 }, { "epoch": 2.172625378385012, "grad_norm": 0.013554349541664124, "learning_rate": 5.119432959287456e-06, "loss": 0.0008, "step": 132780 }, { "epoch": 2.172789004336088, "grad_norm": 0.028516927734017372, "learning_rate": 5.118719208884171e-06, "loss": 0.001, "step": 132790 }, { "epoch": 2.1729526302871633, "grad_norm": 0.16297942399978638, "learning_rate": 5.118005456060303e-06, "loss": 0.0014, "step": 132800 }, { "epoch": 2.1731162562382393, "grad_norm": 0.07093736529350281, "learning_rate": 5.117291700830403e-06, "loss": 0.0012, "step": 132810 }, { "epoch": 2.1732798821893153, "grad_norm": 0.010588572360575199, "learning_rate": 5.116577943209027e-06, "loss": 0.0005, "step": 132820 }, { "epoch": 2.173443508140391, "grad_norm": 0.14723478257656097, "learning_rate": 5.115864183210724e-06, "loss": 0.0012, "step": 132830 }, { "epoch": 2.173607134091467, "grad_norm": 0.14543306827545166, "learning_rate": 5.11515042085005e-06, "loss": 0.001, "step": 132840 }, { "epoch": 2.173770760042543, "grad_norm": 0.019255148246884346, "learning_rate": 5.114436656141556e-06, "loss": 0.0007, "step": 132850 }, { "epoch": 2.1739343859936184, "grad_norm": 0.006870268378406763, "learning_rate": 5.113722889099797e-06, "loss": 0.0007, "step": 132860 }, { "epoch": 2.1740980119446944, "grad_norm": 0.09855835884809494, "learning_rate": 5.113009119739325e-06, "loss": 0.0003, "step": 132870 }, { "epoch": 2.1742616378957704, "grad_norm": 0.015596297569572926, "learning_rate": 5.112295348074692e-06, "loss": 0.0004, "step": 132880 }, { "epoch": 2.174425263846846, "grad_norm": 0.03842705115675926, "learning_rate": 5.111581574120453e-06, "loss": 0.0007, "step": 132890 }, { "epoch": 2.174588889797922, "grad_norm": 0.00872009713202715, "learning_rate": 5.1108677978911604e-06, "loss": 0.0004, "step": 132900 }, { "epoch": 2.174752515748998, "grad_norm": 0.005999826360493898, "learning_rate": 5.110154019401369e-06, "loss": 0.0007, "step": 132910 }, { "epoch": 2.1749161417000735, "grad_norm": 0.021204102784395218, "learning_rate": 5.109440238665628e-06, "loss": 0.0005, "step": 132920 }, { "epoch": 2.1750797676511495, "grad_norm": 0.02100161835551262, "learning_rate": 5.108726455698496e-06, "loss": 0.0007, "step": 132930 }, { "epoch": 2.1752433936022255, "grad_norm": 0.04055703058838844, "learning_rate": 5.108012670514523e-06, "loss": 0.0013, "step": 132940 }, { "epoch": 2.175407019553301, "grad_norm": 0.027376040816307068, "learning_rate": 5.107298883128264e-06, "loss": 0.0018, "step": 132950 }, { "epoch": 2.175570645504377, "grad_norm": 0.10890685766935349, "learning_rate": 5.106585093554272e-06, "loss": 0.001, "step": 132960 }, { "epoch": 2.175734271455453, "grad_norm": 0.02454194985330105, "learning_rate": 5.1058713018071e-06, "loss": 0.0015, "step": 132970 }, { "epoch": 2.1758978974065286, "grad_norm": 0.024043474346399307, "learning_rate": 5.105157507901302e-06, "loss": 0.0006, "step": 132980 }, { "epoch": 2.1760615233576046, "grad_norm": 0.022545553743839264, "learning_rate": 5.104443711851434e-06, "loss": 0.0005, "step": 132990 }, { "epoch": 2.17622514930868, "grad_norm": 0.03366660326719284, "learning_rate": 5.103729913672046e-06, "loss": 0.0007, "step": 133000 }, { "epoch": 2.176388775259756, "grad_norm": 0.0030721276998519897, "learning_rate": 5.103016113377693e-06, "loss": 0.0011, "step": 133010 }, { "epoch": 2.176552401210832, "grad_norm": 0.036428701132535934, "learning_rate": 5.102302310982929e-06, "loss": 0.0007, "step": 133020 }, { "epoch": 2.1767160271619077, "grad_norm": 0.07777273654937744, "learning_rate": 5.101588506502308e-06, "loss": 0.0008, "step": 133030 }, { "epoch": 2.1768796531129837, "grad_norm": 0.04170243442058563, "learning_rate": 5.100874699950384e-06, "loss": 0.0011, "step": 133040 }, { "epoch": 2.1770432790640597, "grad_norm": 0.003547205124050379, "learning_rate": 5.100160891341711e-06, "loss": 0.0005, "step": 133050 }, { "epoch": 2.1772069050151353, "grad_norm": 0.021404877305030823, "learning_rate": 5.099447080690842e-06, "loss": 0.0011, "step": 133060 }, { "epoch": 2.1773705309662112, "grad_norm": 0.05524975806474686, "learning_rate": 5.098733268012333e-06, "loss": 0.0006, "step": 133070 }, { "epoch": 2.1775341569172872, "grad_norm": 0.0016337076667696238, "learning_rate": 5.0980194533207336e-06, "loss": 0.0008, "step": 133080 }, { "epoch": 2.177697782868363, "grad_norm": 0.013846287503838539, "learning_rate": 5.097305636630604e-06, "loss": 0.0013, "step": 133090 }, { "epoch": 2.177861408819439, "grad_norm": 0.06165917217731476, "learning_rate": 5.096591817956493e-06, "loss": 0.0008, "step": 133100 }, { "epoch": 2.178025034770515, "grad_norm": 0.021243825554847717, "learning_rate": 5.095877997312959e-06, "loss": 0.0005, "step": 133110 }, { "epoch": 2.1781886607215903, "grad_norm": 0.011383337900042534, "learning_rate": 5.095164174714553e-06, "loss": 0.0015, "step": 133120 }, { "epoch": 2.1783522866726663, "grad_norm": 0.07124808430671692, "learning_rate": 5.094450350175831e-06, "loss": 0.001, "step": 133130 }, { "epoch": 2.178515912623742, "grad_norm": 0.03493935614824295, "learning_rate": 5.0937365237113455e-06, "loss": 0.0009, "step": 133140 }, { "epoch": 2.178679538574818, "grad_norm": 0.03630709648132324, "learning_rate": 5.093022695335653e-06, "loss": 0.0014, "step": 133150 }, { "epoch": 2.178843164525894, "grad_norm": 0.16153298318386078, "learning_rate": 5.092308865063307e-06, "loss": 0.0016, "step": 133160 }, { "epoch": 2.1790067904769694, "grad_norm": 0.060174234211444855, "learning_rate": 5.0915950329088616e-06, "loss": 0.0013, "step": 133170 }, { "epoch": 2.1791704164280454, "grad_norm": 0.15814559161663055, "learning_rate": 5.09088119888687e-06, "loss": 0.0015, "step": 133180 }, { "epoch": 2.1793340423791214, "grad_norm": 0.045118480920791626, "learning_rate": 5.09016736301189e-06, "loss": 0.0007, "step": 133190 }, { "epoch": 2.179497668330197, "grad_norm": 0.0595676489174366, "learning_rate": 5.089453525298474e-06, "loss": 0.0013, "step": 133200 }, { "epoch": 2.179661294281273, "grad_norm": 0.028509343042969704, "learning_rate": 5.088739685761176e-06, "loss": 0.001, "step": 133210 }, { "epoch": 2.179824920232349, "grad_norm": 0.04259336367249489, "learning_rate": 5.088025844414551e-06, "loss": 0.0047, "step": 133220 }, { "epoch": 2.1799885461834245, "grad_norm": 0.024862762540578842, "learning_rate": 5.087312001273154e-06, "loss": 0.0009, "step": 133230 }, { "epoch": 2.1801521721345005, "grad_norm": 0.06844490766525269, "learning_rate": 5.086598156351541e-06, "loss": 0.0008, "step": 133240 }, { "epoch": 2.1803157980855765, "grad_norm": 0.062349770218133926, "learning_rate": 5.085884309664263e-06, "loss": 0.0016, "step": 133250 }, { "epoch": 2.180479424036652, "grad_norm": 0.03903033584356308, "learning_rate": 5.085170461225879e-06, "loss": 0.0011, "step": 133260 }, { "epoch": 2.180643049987728, "grad_norm": 0.013420801609754562, "learning_rate": 5.084456611050939e-06, "loss": 0.0013, "step": 133270 }, { "epoch": 2.180806675938804, "grad_norm": 0.011357001028954983, "learning_rate": 5.083742759154003e-06, "loss": 0.0013, "step": 133280 }, { "epoch": 2.1809703018898796, "grad_norm": 0.05441857874393463, "learning_rate": 5.083028905549623e-06, "loss": 0.0014, "step": 133290 }, { "epoch": 2.1811339278409556, "grad_norm": 0.09027910232543945, "learning_rate": 5.082315050252355e-06, "loss": 0.0007, "step": 133300 }, { "epoch": 2.1812975537920316, "grad_norm": 0.07436022907495499, "learning_rate": 5.08160119327675e-06, "loss": 0.0008, "step": 133310 }, { "epoch": 2.181461179743107, "grad_norm": 0.014663045294582844, "learning_rate": 5.0808873346373685e-06, "loss": 0.0009, "step": 133320 }, { "epoch": 2.181624805694183, "grad_norm": 0.0034611723385751247, "learning_rate": 5.080173474348762e-06, "loss": 0.0022, "step": 133330 }, { "epoch": 2.181788431645259, "grad_norm": 0.014016291126608849, "learning_rate": 5.079459612425488e-06, "loss": 0.0009, "step": 133340 }, { "epoch": 2.1819520575963347, "grad_norm": 0.05073550343513489, "learning_rate": 5.078745748882099e-06, "loss": 0.0007, "step": 133350 }, { "epoch": 2.1821156835474107, "grad_norm": 0.29995468258857727, "learning_rate": 5.078031883733152e-06, "loss": 0.0015, "step": 133360 }, { "epoch": 2.1822793094984863, "grad_norm": 0.08799844235181808, "learning_rate": 5.0773180169932e-06, "loss": 0.0014, "step": 133370 }, { "epoch": 2.1824429354495622, "grad_norm": 0.008153904229402542, "learning_rate": 5.0766041486768005e-06, "loss": 0.0004, "step": 133380 }, { "epoch": 2.1826065614006382, "grad_norm": 0.09970121085643768, "learning_rate": 5.075890278798506e-06, "loss": 0.0014, "step": 133390 }, { "epoch": 2.182770187351714, "grad_norm": 0.004192731808871031, "learning_rate": 5.075176407372875e-06, "loss": 0.0024, "step": 133400 }, { "epoch": 2.18293381330279, "grad_norm": 0.02286132052540779, "learning_rate": 5.074462534414459e-06, "loss": 0.001, "step": 133410 }, { "epoch": 2.183097439253866, "grad_norm": 0.03621543571352959, "learning_rate": 5.073748659937819e-06, "loss": 0.0006, "step": 133420 }, { "epoch": 2.1832610652049413, "grad_norm": 0.03288775309920311, "learning_rate": 5.073034783957502e-06, "loss": 0.0008, "step": 133430 }, { "epoch": 2.1834246911560173, "grad_norm": 0.031202517449855804, "learning_rate": 5.0723209064880705e-06, "loss": 0.0006, "step": 133440 }, { "epoch": 2.1835883171070933, "grad_norm": 0.039356913417577744, "learning_rate": 5.071607027544076e-06, "loss": 0.0009, "step": 133450 }, { "epoch": 2.183751943058169, "grad_norm": 0.06485525518655777, "learning_rate": 5.070893147140077e-06, "loss": 0.0011, "step": 133460 }, { "epoch": 2.183915569009245, "grad_norm": 0.03090715780854225, "learning_rate": 5.070179265290625e-06, "loss": 0.001, "step": 133470 }, { "epoch": 2.184079194960321, "grad_norm": 0.1975259780883789, "learning_rate": 5.069465382010279e-06, "loss": 0.001, "step": 133480 }, { "epoch": 2.1842428209113964, "grad_norm": 0.005412629339843988, "learning_rate": 5.068751497313592e-06, "loss": 0.0005, "step": 133490 }, { "epoch": 2.1844064468624724, "grad_norm": 0.1312752366065979, "learning_rate": 5.068037611215121e-06, "loss": 0.0009, "step": 133500 }, { "epoch": 2.1845700728135484, "grad_norm": 0.07029860466718674, "learning_rate": 5.067323723729421e-06, "loss": 0.001, "step": 133510 }, { "epoch": 2.184733698764624, "grad_norm": 0.04959215223789215, "learning_rate": 5.066609834871047e-06, "loss": 0.0009, "step": 133520 }, { "epoch": 2.1848973247157, "grad_norm": 0.046235620975494385, "learning_rate": 5.0658959446545555e-06, "loss": 0.0005, "step": 133530 }, { "epoch": 2.1850609506667755, "grad_norm": 0.0022209377493709326, "learning_rate": 5.0651820530945e-06, "loss": 0.001, "step": 133540 }, { "epoch": 2.1852245766178515, "grad_norm": 0.0027011074125766754, "learning_rate": 5.064468160205441e-06, "loss": 0.0008, "step": 133550 }, { "epoch": 2.1853882025689275, "grad_norm": 0.03359526768326759, "learning_rate": 5.063754266001929e-06, "loss": 0.0014, "step": 133560 }, { "epoch": 2.185551828520003, "grad_norm": 0.0693717747926712, "learning_rate": 5.063040370498522e-06, "loss": 0.0007, "step": 133570 }, { "epoch": 2.185715454471079, "grad_norm": 0.02481403574347496, "learning_rate": 5.062326473709775e-06, "loss": 0.0005, "step": 133580 }, { "epoch": 2.185879080422155, "grad_norm": 0.025182776153087616, "learning_rate": 5.0616125756502455e-06, "loss": 0.0006, "step": 133590 }, { "epoch": 2.1860427063732306, "grad_norm": 0.032918162643909454, "learning_rate": 5.060898676334487e-06, "loss": 0.0009, "step": 133600 }, { "epoch": 2.1862063323243066, "grad_norm": 0.002902632812038064, "learning_rate": 5.060184775777058e-06, "loss": 0.0011, "step": 133610 }, { "epoch": 2.1863699582753826, "grad_norm": 0.044303566217422485, "learning_rate": 5.05947087399251e-06, "loss": 0.001, "step": 133620 }, { "epoch": 2.186533584226458, "grad_norm": 0.043209511786699295, "learning_rate": 5.058756970995404e-06, "loss": 0.0004, "step": 133630 }, { "epoch": 2.186697210177534, "grad_norm": 0.0523659773170948, "learning_rate": 5.058043066800291e-06, "loss": 0.001, "step": 133640 }, { "epoch": 2.18686083612861, "grad_norm": 0.02538548782467842, "learning_rate": 5.05732916142173e-06, "loss": 0.0007, "step": 133650 }, { "epoch": 2.1870244620796857, "grad_norm": 0.0039671421982347965, "learning_rate": 5.0566152548742766e-06, "loss": 0.0006, "step": 133660 }, { "epoch": 2.1871880880307617, "grad_norm": 0.06253079324960709, "learning_rate": 5.055901347172486e-06, "loss": 0.001, "step": 133670 }, { "epoch": 2.1873517139818377, "grad_norm": 0.00617243442684412, "learning_rate": 5.0551874383309145e-06, "loss": 0.0006, "step": 133680 }, { "epoch": 2.1875153399329132, "grad_norm": 0.030873240903019905, "learning_rate": 5.0544735283641185e-06, "loss": 0.001, "step": 133690 }, { "epoch": 2.1876789658839892, "grad_norm": 0.0217451062053442, "learning_rate": 5.053759617286652e-06, "loss": 0.0006, "step": 133700 }, { "epoch": 2.1878425918350652, "grad_norm": 0.02554520033299923, "learning_rate": 5.053045705113075e-06, "loss": 0.001, "step": 133710 }, { "epoch": 2.188006217786141, "grad_norm": 0.04368767887353897, "learning_rate": 5.05233179185794e-06, "loss": 0.0004, "step": 133720 }, { "epoch": 2.188169843737217, "grad_norm": 0.065827876329422, "learning_rate": 5.051617877535804e-06, "loss": 0.0007, "step": 133730 }, { "epoch": 2.188333469688293, "grad_norm": 0.08965959399938583, "learning_rate": 5.050903962161223e-06, "loss": 0.0017, "step": 133740 }, { "epoch": 2.1884970956393683, "grad_norm": 0.056246012449264526, "learning_rate": 5.050190045748755e-06, "loss": 0.0015, "step": 133750 }, { "epoch": 2.1886607215904443, "grad_norm": 0.0020647458732128143, "learning_rate": 5.049476128312954e-06, "loss": 0.0006, "step": 133760 }, { "epoch": 2.18882434754152, "grad_norm": 0.04372989386320114, "learning_rate": 5.048762209868377e-06, "loss": 0.001, "step": 133770 }, { "epoch": 2.188987973492596, "grad_norm": 0.15670925378799438, "learning_rate": 5.048048290429579e-06, "loss": 0.0009, "step": 133780 }, { "epoch": 2.189151599443672, "grad_norm": 0.0033765234984457493, "learning_rate": 5.047334370011118e-06, "loss": 0.0004, "step": 133790 }, { "epoch": 2.1893152253947474, "grad_norm": 0.04234068840742111, "learning_rate": 5.0466204486275495e-06, "loss": 0.0006, "step": 133800 }, { "epoch": 2.1894788513458234, "grad_norm": 0.08041514456272125, "learning_rate": 5.04590652629343e-06, "loss": 0.002, "step": 133810 }, { "epoch": 2.1896424772968994, "grad_norm": 0.06522290408611298, "learning_rate": 5.0451926030233145e-06, "loss": 0.0007, "step": 133820 }, { "epoch": 2.189806103247975, "grad_norm": 0.0013587516732513905, "learning_rate": 5.0444786788317615e-06, "loss": 0.0008, "step": 133830 }, { "epoch": 2.189969729199051, "grad_norm": 0.021369660273194313, "learning_rate": 5.043764753733326e-06, "loss": 0.0007, "step": 133840 }, { "epoch": 2.190133355150127, "grad_norm": 0.040330950170755386, "learning_rate": 5.043050827742563e-06, "loss": 0.0014, "step": 133850 }, { "epoch": 2.1902969811012025, "grad_norm": 0.01845158450305462, "learning_rate": 5.042336900874031e-06, "loss": 0.0009, "step": 133860 }, { "epoch": 2.1904606070522785, "grad_norm": 0.04459642246365547, "learning_rate": 5.041622973142285e-06, "loss": 0.001, "step": 133870 }, { "epoch": 2.1906242330033545, "grad_norm": 0.05638842657208443, "learning_rate": 5.040909044561882e-06, "loss": 0.0016, "step": 133880 }, { "epoch": 2.19078785895443, "grad_norm": 0.06678682565689087, "learning_rate": 5.040195115147379e-06, "loss": 0.0014, "step": 133890 }, { "epoch": 2.190951484905506, "grad_norm": 0.039045076817274094, "learning_rate": 5.039481184913333e-06, "loss": 0.0008, "step": 133900 }, { "epoch": 2.191115110856582, "grad_norm": 0.041053514927625656, "learning_rate": 5.038767253874297e-06, "loss": 0.0011, "step": 133910 }, { "epoch": 2.1912787368076576, "grad_norm": 0.04937037080526352, "learning_rate": 5.038053322044832e-06, "loss": 0.0009, "step": 133920 }, { "epoch": 2.1914423627587336, "grad_norm": 0.039465777575969696, "learning_rate": 5.0373393894394895e-06, "loss": 0.0013, "step": 133930 }, { "epoch": 2.191605988709809, "grad_norm": 0.0005525247543118894, "learning_rate": 5.03662545607283e-06, "loss": 0.0005, "step": 133940 }, { "epoch": 2.191769614660885, "grad_norm": 0.03335276618599892, "learning_rate": 5.035911521959408e-06, "loss": 0.0016, "step": 133950 }, { "epoch": 2.191933240611961, "grad_norm": 0.04523013159632683, "learning_rate": 5.0351975871137814e-06, "loss": 0.0008, "step": 133960 }, { "epoch": 2.1920968665630367, "grad_norm": 0.12009347230195999, "learning_rate": 5.034483651550504e-06, "loss": 0.0007, "step": 133970 }, { "epoch": 2.1922604925141127, "grad_norm": 0.12426785379648209, "learning_rate": 5.033769715284137e-06, "loss": 0.0007, "step": 133980 }, { "epoch": 2.1924241184651887, "grad_norm": 0.039637695997953415, "learning_rate": 5.033055778329234e-06, "loss": 0.0007, "step": 133990 }, { "epoch": 2.1925877444162643, "grad_norm": 0.04236973822116852, "learning_rate": 5.03234184070035e-06, "loss": 0.0006, "step": 134000 }, { "epoch": 2.1927513703673402, "grad_norm": 0.04901636019349098, "learning_rate": 5.031627902412044e-06, "loss": 0.0005, "step": 134010 }, { "epoch": 2.1929149963184162, "grad_norm": 0.06156611815094948, "learning_rate": 5.030913963478873e-06, "loss": 0.0009, "step": 134020 }, { "epoch": 2.193078622269492, "grad_norm": 0.00590739818289876, "learning_rate": 5.0302000239153905e-06, "loss": 0.0006, "step": 134030 }, { "epoch": 2.193242248220568, "grad_norm": 0.03786858916282654, "learning_rate": 5.029486083736157e-06, "loss": 0.0009, "step": 134040 }, { "epoch": 2.193405874171644, "grad_norm": 0.04943416267633438, "learning_rate": 5.028772142955727e-06, "loss": 0.0014, "step": 134050 }, { "epoch": 2.1935695001227193, "grad_norm": 0.045386798679828644, "learning_rate": 5.028058201588657e-06, "loss": 0.0006, "step": 134060 }, { "epoch": 2.1937331260737953, "grad_norm": 0.008895941078662872, "learning_rate": 5.027344259649504e-06, "loss": 0.0012, "step": 134070 }, { "epoch": 2.1938967520248713, "grad_norm": 0.2271803319454193, "learning_rate": 5.026630317152826e-06, "loss": 0.0016, "step": 134080 }, { "epoch": 2.194060377975947, "grad_norm": 0.06289710104465485, "learning_rate": 5.025916374113177e-06, "loss": 0.002, "step": 134090 }, { "epoch": 2.194224003927023, "grad_norm": 0.05548407509922981, "learning_rate": 5.025202430545116e-06, "loss": 0.0005, "step": 134100 }, { "epoch": 2.194387629878099, "grad_norm": 0.02851760759949684, "learning_rate": 5.024488486463198e-06, "loss": 0.001, "step": 134110 }, { "epoch": 2.1945512558291744, "grad_norm": 0.1312255859375, "learning_rate": 5.023774541881983e-06, "loss": 0.0009, "step": 134120 }, { "epoch": 2.1947148817802504, "grad_norm": 0.10321973264217377, "learning_rate": 5.023060596816023e-06, "loss": 0.0008, "step": 134130 }, { "epoch": 2.1948785077313264, "grad_norm": 0.03090348280966282, "learning_rate": 5.022346651279878e-06, "loss": 0.0009, "step": 134140 }, { "epoch": 2.195042133682402, "grad_norm": 0.004360199440270662, "learning_rate": 5.0216327052881046e-06, "loss": 0.0022, "step": 134150 }, { "epoch": 2.195205759633478, "grad_norm": 0.039616137742996216, "learning_rate": 5.020918758855257e-06, "loss": 0.0014, "step": 134160 }, { "epoch": 2.1953693855845535, "grad_norm": 0.06525007635354996, "learning_rate": 5.0202048119958935e-06, "loss": 0.0007, "step": 134170 }, { "epoch": 2.1955330115356295, "grad_norm": 0.0726492628455162, "learning_rate": 5.019490864724572e-06, "loss": 0.0011, "step": 134180 }, { "epoch": 2.1956966374867055, "grad_norm": 0.11462720483541489, "learning_rate": 5.018776917055848e-06, "loss": 0.001, "step": 134190 }, { "epoch": 2.195860263437781, "grad_norm": 0.04300875961780548, "learning_rate": 5.018062969004279e-06, "loss": 0.0007, "step": 134200 }, { "epoch": 2.196023889388857, "grad_norm": 0.07552115619182587, "learning_rate": 5.017349020584421e-06, "loss": 0.0009, "step": 134210 }, { "epoch": 2.196187515339933, "grad_norm": 0.015088643878698349, "learning_rate": 5.016635071810832e-06, "loss": 0.0005, "step": 134220 }, { "epoch": 2.1963511412910086, "grad_norm": 0.029525546357035637, "learning_rate": 5.015921122698068e-06, "loss": 0.002, "step": 134230 }, { "epoch": 2.1965147672420846, "grad_norm": 0.014646231196820736, "learning_rate": 5.015207173260684e-06, "loss": 0.0003, "step": 134240 }, { "epoch": 2.1966783931931606, "grad_norm": 0.08247936517000198, "learning_rate": 5.014493223513241e-06, "loss": 0.0013, "step": 134250 }, { "epoch": 2.196842019144236, "grad_norm": 0.04168146103620529, "learning_rate": 5.013779273470292e-06, "loss": 0.0021, "step": 134260 }, { "epoch": 2.197005645095312, "grad_norm": 0.017100045457482338, "learning_rate": 5.013065323146397e-06, "loss": 0.0008, "step": 134270 }, { "epoch": 2.197169271046388, "grad_norm": 0.06062128022313118, "learning_rate": 5.012351372556109e-06, "loss": 0.0008, "step": 134280 }, { "epoch": 2.1973328969974637, "grad_norm": 0.055368438363075256, "learning_rate": 5.011637421713988e-06, "loss": 0.0009, "step": 134290 }, { "epoch": 2.1974965229485397, "grad_norm": 0.026301488280296326, "learning_rate": 5.01092347063459e-06, "loss": 0.0007, "step": 134300 }, { "epoch": 2.1976601488996153, "grad_norm": 0.0661306381225586, "learning_rate": 5.010209519332472e-06, "loss": 0.0009, "step": 134310 }, { "epoch": 2.1978237748506912, "grad_norm": 0.007232753559947014, "learning_rate": 5.009495567822189e-06, "loss": 0.0005, "step": 134320 }, { "epoch": 2.1979874008017672, "grad_norm": 0.017698731273412704, "learning_rate": 5.008781616118302e-06, "loss": 0.0007, "step": 134330 }, { "epoch": 2.198151026752843, "grad_norm": 0.03146834671497345, "learning_rate": 5.008067664235363e-06, "loss": 0.0009, "step": 134340 }, { "epoch": 2.198314652703919, "grad_norm": 0.02815254218876362, "learning_rate": 5.007353712187933e-06, "loss": 0.0008, "step": 134350 }, { "epoch": 2.198478278654995, "grad_norm": 0.042273618280887604, "learning_rate": 5.006639759990566e-06, "loss": 0.0011, "step": 134360 }, { "epoch": 2.1986419046060703, "grad_norm": 0.10136000066995621, "learning_rate": 5.00592580765782e-06, "loss": 0.0009, "step": 134370 }, { "epoch": 2.1988055305571463, "grad_norm": 0.01906038075685501, "learning_rate": 5.005211855204251e-06, "loss": 0.001, "step": 134380 }, { "epoch": 2.1989691565082223, "grad_norm": 0.026970237493515015, "learning_rate": 5.004497902644419e-06, "loss": 0.0007, "step": 134390 }, { "epoch": 2.199132782459298, "grad_norm": 0.020539671182632446, "learning_rate": 5.003783949992876e-06, "loss": 0.0008, "step": 134400 }, { "epoch": 2.199296408410374, "grad_norm": 0.028324419632554054, "learning_rate": 5.0030699972641836e-06, "loss": 0.0005, "step": 134410 }, { "epoch": 2.19946003436145, "grad_norm": 0.0812520906329155, "learning_rate": 5.002356044472896e-06, "loss": 0.0013, "step": 134420 }, { "epoch": 2.1996236603125254, "grad_norm": 0.019305819645524025, "learning_rate": 5.0016420916335705e-06, "loss": 0.0009, "step": 134430 }, { "epoch": 2.1997872862636014, "grad_norm": 0.08007299900054932, "learning_rate": 5.0009281387607635e-06, "loss": 0.0005, "step": 134440 }, { "epoch": 2.1999509122146774, "grad_norm": 0.01840873807668686, "learning_rate": 5.000214185869033e-06, "loss": 0.0006, "step": 134450 }, { "epoch": 2.200114538165753, "grad_norm": 0.044533710926771164, "learning_rate": 4.999500232972936e-06, "loss": 0.0007, "step": 134460 }, { "epoch": 2.200278164116829, "grad_norm": 0.1473854035139084, "learning_rate": 4.99878628008703e-06, "loss": 0.0005, "step": 134470 }, { "epoch": 2.200441790067905, "grad_norm": 0.041377656161785126, "learning_rate": 4.998072327225868e-06, "loss": 0.0004, "step": 134480 }, { "epoch": 2.2006054160189805, "grad_norm": 0.07973428815603256, "learning_rate": 4.997358374404012e-06, "loss": 0.0012, "step": 134490 }, { "epoch": 2.2007690419700565, "grad_norm": 0.04790882393717766, "learning_rate": 4.996644421636014e-06, "loss": 0.0013, "step": 134500 }, { "epoch": 2.2009326679211325, "grad_norm": 0.04772808402776718, "learning_rate": 4.9959304689364355e-06, "loss": 0.0006, "step": 134510 }, { "epoch": 2.201096293872208, "grad_norm": 0.024572094902396202, "learning_rate": 4.99521651631983e-06, "loss": 0.0006, "step": 134520 }, { "epoch": 2.201259919823284, "grad_norm": 0.04496666416525841, "learning_rate": 4.9945025638007564e-06, "loss": 0.0005, "step": 134530 }, { "epoch": 2.2014235457743596, "grad_norm": 0.11788041144609451, "learning_rate": 4.993788611393769e-06, "loss": 0.0013, "step": 134540 }, { "epoch": 2.2015871717254356, "grad_norm": 0.00399753637611866, "learning_rate": 4.993074659113429e-06, "loss": 0.0007, "step": 134550 }, { "epoch": 2.2017507976765116, "grad_norm": 0.07214364409446716, "learning_rate": 4.992360706974289e-06, "loss": 0.0012, "step": 134560 }, { "epoch": 2.201914423627587, "grad_norm": 0.07662001997232437, "learning_rate": 4.991646754990909e-06, "loss": 0.0005, "step": 134570 }, { "epoch": 2.202078049578663, "grad_norm": 0.021875429898500443, "learning_rate": 4.9909328031778435e-06, "loss": 0.0004, "step": 134580 }, { "epoch": 2.202241675529739, "grad_norm": 0.0572219118475914, "learning_rate": 4.9902188515496515e-06, "loss": 0.0009, "step": 134590 }, { "epoch": 2.2024053014808147, "grad_norm": 0.010412391275167465, "learning_rate": 4.9895049001208875e-06, "loss": 0.0016, "step": 134600 }, { "epoch": 2.2025689274318907, "grad_norm": 0.04098214581608772, "learning_rate": 4.988790948906111e-06, "loss": 0.0009, "step": 134610 }, { "epoch": 2.2027325533829667, "grad_norm": 0.15842577815055847, "learning_rate": 4.988076997919877e-06, "loss": 0.0011, "step": 134620 }, { "epoch": 2.2028961793340422, "grad_norm": 0.06622492522001266, "learning_rate": 4.987363047176743e-06, "loss": 0.0013, "step": 134630 }, { "epoch": 2.2030598052851182, "grad_norm": 0.017093366011977196, "learning_rate": 4.986649096691265e-06, "loss": 0.0009, "step": 134640 }, { "epoch": 2.2032234312361942, "grad_norm": 0.02174472063779831, "learning_rate": 4.985935146478001e-06, "loss": 0.0013, "step": 134650 }, { "epoch": 2.20338705718727, "grad_norm": 0.029918214306235313, "learning_rate": 4.985221196551508e-06, "loss": 0.0007, "step": 134660 }, { "epoch": 2.203550683138346, "grad_norm": 0.046462949365377426, "learning_rate": 4.984507246926342e-06, "loss": 0.002, "step": 134670 }, { "epoch": 2.203714309089422, "grad_norm": 0.02724730409681797, "learning_rate": 4.9837932976170594e-06, "loss": 0.0009, "step": 134680 }, { "epoch": 2.2038779350404973, "grad_norm": 0.07216127216815948, "learning_rate": 4.983079348638219e-06, "loss": 0.0009, "step": 134690 }, { "epoch": 2.2040415609915733, "grad_norm": 0.005547720473259687, "learning_rate": 4.982365400004374e-06, "loss": 0.0014, "step": 134700 }, { "epoch": 2.204205186942649, "grad_norm": 0.027677949517965317, "learning_rate": 4.981651451730086e-06, "loss": 0.001, "step": 134710 }, { "epoch": 2.204368812893725, "grad_norm": 0.025072269141674042, "learning_rate": 4.980937503829907e-06, "loss": 0.0007, "step": 134720 }, { "epoch": 2.204532438844801, "grad_norm": 0.04724979028105736, "learning_rate": 4.9802235563183985e-06, "loss": 0.0007, "step": 134730 }, { "epoch": 2.2046960647958764, "grad_norm": 0.013891017064452171, "learning_rate": 4.979509609210112e-06, "loss": 0.0008, "step": 134740 }, { "epoch": 2.2048596907469524, "grad_norm": 0.09439883381128311, "learning_rate": 4.97879566251961e-06, "loss": 0.001, "step": 134750 }, { "epoch": 2.2050233166980284, "grad_norm": 0.052819475531578064, "learning_rate": 4.978081716261445e-06, "loss": 0.0009, "step": 134760 }, { "epoch": 2.205186942649104, "grad_norm": 0.1775171011686325, "learning_rate": 4.977367770450176e-06, "loss": 0.0007, "step": 134770 }, { "epoch": 2.20535056860018, "grad_norm": 0.018475642427802086, "learning_rate": 4.976653825100357e-06, "loss": 0.0006, "step": 134780 }, { "epoch": 2.205514194551256, "grad_norm": 0.056880462914705276, "learning_rate": 4.975939880226548e-06, "loss": 0.0008, "step": 134790 }, { "epoch": 2.2056778205023315, "grad_norm": 0.02156502939760685, "learning_rate": 4.975225935843304e-06, "loss": 0.0013, "step": 134800 }, { "epoch": 2.2058414464534075, "grad_norm": 0.027907708659768105, "learning_rate": 4.9745119919651826e-06, "loss": 0.0008, "step": 134810 }, { "epoch": 2.2060050724044835, "grad_norm": 0.036292336881160736, "learning_rate": 4.97379804860674e-06, "loss": 0.0008, "step": 134820 }, { "epoch": 2.206168698355559, "grad_norm": 0.16955789923667908, "learning_rate": 4.973084105782533e-06, "loss": 0.0009, "step": 134830 }, { "epoch": 2.206332324306635, "grad_norm": 0.04172381013631821, "learning_rate": 4.972370163507117e-06, "loss": 0.0006, "step": 134840 }, { "epoch": 2.206495950257711, "grad_norm": 0.02368123084306717, "learning_rate": 4.971656221795051e-06, "loss": 0.0012, "step": 134850 }, { "epoch": 2.2066595762087866, "grad_norm": 0.0713518038392067, "learning_rate": 4.97094228066089e-06, "loss": 0.0008, "step": 134860 }, { "epoch": 2.2068232021598626, "grad_norm": 0.009430157020688057, "learning_rate": 4.970228340119192e-06, "loss": 0.001, "step": 134870 }, { "epoch": 2.2069868281109386, "grad_norm": 0.07916197180747986, "learning_rate": 4.969514400184512e-06, "loss": 0.0023, "step": 134880 }, { "epoch": 2.207150454062014, "grad_norm": 0.00507709663361311, "learning_rate": 4.968800460871408e-06, "loss": 0.0005, "step": 134890 }, { "epoch": 2.20731408001309, "grad_norm": 0.003556142095476389, "learning_rate": 4.9680865221944355e-06, "loss": 0.0008, "step": 134900 }, { "epoch": 2.207477705964166, "grad_norm": 0.034075163304805756, "learning_rate": 4.967372584168153e-06, "loss": 0.0007, "step": 134910 }, { "epoch": 2.2076413319152417, "grad_norm": 0.05971343815326691, "learning_rate": 4.9666586468071135e-06, "loss": 0.0009, "step": 134920 }, { "epoch": 2.2078049578663177, "grad_norm": 0.08883824944496155, "learning_rate": 4.9659447101258775e-06, "loss": 0.0008, "step": 134930 }, { "epoch": 2.2079685838173932, "grad_norm": 0.001362925278954208, "learning_rate": 4.965230774138999e-06, "loss": 0.0008, "step": 134940 }, { "epoch": 2.2081322097684692, "grad_norm": 0.0057477811351418495, "learning_rate": 4.964516838861038e-06, "loss": 0.0008, "step": 134950 }, { "epoch": 2.2082958357195452, "grad_norm": 0.022669587284326553, "learning_rate": 4.963802904306545e-06, "loss": 0.0012, "step": 134960 }, { "epoch": 2.208459461670621, "grad_norm": 0.013272490352392197, "learning_rate": 4.963088970490083e-06, "loss": 0.0042, "step": 134970 }, { "epoch": 2.208623087621697, "grad_norm": 0.02670869044959545, "learning_rate": 4.962375037426202e-06, "loss": 0.0005, "step": 134980 }, { "epoch": 2.208786713572773, "grad_norm": 0.043563760817050934, "learning_rate": 4.961661105129466e-06, "loss": 0.0008, "step": 134990 }, { "epoch": 2.2089503395238483, "grad_norm": 0.021103229373693466, "learning_rate": 4.9609471736144234e-06, "loss": 0.0003, "step": 135000 }, { "epoch": 2.2091139654749243, "grad_norm": 0.03933648392558098, "learning_rate": 4.960233242895638e-06, "loss": 0.0009, "step": 135010 }, { "epoch": 2.2092775914260003, "grad_norm": 0.014243829995393753, "learning_rate": 4.959519312987659e-06, "loss": 0.0005, "step": 135020 }, { "epoch": 2.209441217377076, "grad_norm": 0.006053130142390728, "learning_rate": 4.958805383905052e-06, "loss": 0.0009, "step": 135030 }, { "epoch": 2.209604843328152, "grad_norm": 0.003697823267430067, "learning_rate": 4.958091455662364e-06, "loss": 0.0006, "step": 135040 }, { "epoch": 2.209768469279228, "grad_norm": 0.00930149294435978, "learning_rate": 4.957377528274159e-06, "loss": 0.0004, "step": 135050 }, { "epoch": 2.2099320952303034, "grad_norm": 0.0433034673333168, "learning_rate": 4.956663601754987e-06, "loss": 0.0009, "step": 135060 }, { "epoch": 2.2100957211813794, "grad_norm": 0.014844703488051891, "learning_rate": 4.955949676119409e-06, "loss": 0.0002, "step": 135070 }, { "epoch": 2.210259347132455, "grad_norm": 0.02840907871723175, "learning_rate": 4.955235751381977e-06, "loss": 0.0006, "step": 135080 }, { "epoch": 2.210422973083531, "grad_norm": 0.17177845537662506, "learning_rate": 4.954521827557254e-06, "loss": 0.0011, "step": 135090 }, { "epoch": 2.210586599034607, "grad_norm": 0.058716703206300735, "learning_rate": 4.9538079046597885e-06, "loss": 0.0017, "step": 135100 }, { "epoch": 2.2107502249856825, "grad_norm": 0.02195795811712742, "learning_rate": 4.953093982704144e-06, "loss": 0.0011, "step": 135110 }, { "epoch": 2.2109138509367585, "grad_norm": 0.0005602446617558599, "learning_rate": 4.952380061704871e-06, "loss": 0.001, "step": 135120 }, { "epoch": 2.2110774768878345, "grad_norm": 0.003225443186238408, "learning_rate": 4.95166614167653e-06, "loss": 0.0009, "step": 135130 }, { "epoch": 2.21124110283891, "grad_norm": 0.08997116982936859, "learning_rate": 4.950952222633672e-06, "loss": 0.0006, "step": 135140 }, { "epoch": 2.211404728789986, "grad_norm": 0.07087930291891098, "learning_rate": 4.950238304590859e-06, "loss": 0.0011, "step": 135150 }, { "epoch": 2.211568354741062, "grad_norm": 0.08538839221000671, "learning_rate": 4.9495243875626444e-06, "loss": 0.0007, "step": 135160 }, { "epoch": 2.2117319806921376, "grad_norm": 0.013447572477161884, "learning_rate": 4.948810471563584e-06, "loss": 0.0006, "step": 135170 }, { "epoch": 2.2118956066432136, "grad_norm": 0.010830838233232498, "learning_rate": 4.948096556608235e-06, "loss": 0.0022, "step": 135180 }, { "epoch": 2.2120592325942896, "grad_norm": 0.009778796695172787, "learning_rate": 4.947382642711153e-06, "loss": 0.0009, "step": 135190 }, { "epoch": 2.212222858545365, "grad_norm": 0.014290392398834229, "learning_rate": 4.9466687298868934e-06, "loss": 0.0011, "step": 135200 }, { "epoch": 2.212386484496441, "grad_norm": 0.06360254436731339, "learning_rate": 4.9459548181500146e-06, "loss": 0.0008, "step": 135210 }, { "epoch": 2.212550110447517, "grad_norm": 0.03681422770023346, "learning_rate": 4.945240907515069e-06, "loss": 0.0012, "step": 135220 }, { "epoch": 2.2127137363985927, "grad_norm": 0.06488761305809021, "learning_rate": 4.944526997996617e-06, "loss": 0.0008, "step": 135230 }, { "epoch": 2.2128773623496687, "grad_norm": 0.026671195402741432, "learning_rate": 4.943813089609211e-06, "loss": 0.0008, "step": 135240 }, { "epoch": 2.2130409883007447, "grad_norm": 0.04481323063373566, "learning_rate": 4.943099182367408e-06, "loss": 0.0006, "step": 135250 }, { "epoch": 2.2132046142518202, "grad_norm": 0.002236276166513562, "learning_rate": 4.942385276285765e-06, "loss": 0.0009, "step": 135260 }, { "epoch": 2.2133682402028962, "grad_norm": 0.022836042568087578, "learning_rate": 4.941671371378837e-06, "loss": 0.0006, "step": 135270 }, { "epoch": 2.2135318661539722, "grad_norm": 0.03900330513715744, "learning_rate": 4.94095746766118e-06, "loss": 0.0008, "step": 135280 }, { "epoch": 2.213695492105048, "grad_norm": 0.012081734836101532, "learning_rate": 4.940243565147351e-06, "loss": 0.001, "step": 135290 }, { "epoch": 2.213859118056124, "grad_norm": 0.06008905544877052, "learning_rate": 4.939529663851903e-06, "loss": 0.0007, "step": 135300 }, { "epoch": 2.2140227440071993, "grad_norm": 0.08137839287519455, "learning_rate": 4.9388157637893955e-06, "loss": 0.0012, "step": 135310 }, { "epoch": 2.2141863699582753, "grad_norm": 0.003097549779340625, "learning_rate": 4.938101864974381e-06, "loss": 0.0013, "step": 135320 }, { "epoch": 2.2143499959093513, "grad_norm": 0.1123538464307785, "learning_rate": 4.937387967421419e-06, "loss": 0.0009, "step": 135330 }, { "epoch": 2.214513621860427, "grad_norm": 0.02141154557466507, "learning_rate": 4.9366740711450605e-06, "loss": 0.0026, "step": 135340 }, { "epoch": 2.214677247811503, "grad_norm": 0.07200434803962708, "learning_rate": 4.935960176159866e-06, "loss": 0.002, "step": 135350 }, { "epoch": 2.214840873762579, "grad_norm": 0.025110071524977684, "learning_rate": 4.935246282480388e-06, "loss": 0.001, "step": 135360 }, { "epoch": 2.2150044997136544, "grad_norm": 0.02399212121963501, "learning_rate": 4.934532390121183e-06, "loss": 0.0006, "step": 135370 }, { "epoch": 2.2151681256647304, "grad_norm": 0.01789342425763607, "learning_rate": 4.933818499096807e-06, "loss": 0.0008, "step": 135380 }, { "epoch": 2.2153317516158064, "grad_norm": 0.15035119652748108, "learning_rate": 4.933104609421817e-06, "loss": 0.0012, "step": 135390 }, { "epoch": 2.215495377566882, "grad_norm": 0.12143184244632721, "learning_rate": 4.932390721110765e-06, "loss": 0.0009, "step": 135400 }, { "epoch": 2.215659003517958, "grad_norm": 0.025419851765036583, "learning_rate": 4.93167683417821e-06, "loss": 0.0005, "step": 135410 }, { "epoch": 2.215822629469034, "grad_norm": 0.03053506836295128, "learning_rate": 4.930962948638705e-06, "loss": 0.0007, "step": 135420 }, { "epoch": 2.2159862554201095, "grad_norm": 0.027081558480858803, "learning_rate": 4.9302490645068085e-06, "loss": 0.0004, "step": 135430 }, { "epoch": 2.2161498813711855, "grad_norm": 0.016230115666985512, "learning_rate": 4.929535181797073e-06, "loss": 0.0008, "step": 135440 }, { "epoch": 2.2163135073222615, "grad_norm": 0.15312176942825317, "learning_rate": 4.928821300524056e-06, "loss": 0.0014, "step": 135450 }, { "epoch": 2.216477133273337, "grad_norm": 0.008419238030910492, "learning_rate": 4.92810742070231e-06, "loss": 0.0031, "step": 135460 }, { "epoch": 2.216640759224413, "grad_norm": 0.09770318865776062, "learning_rate": 4.927393542346395e-06, "loss": 0.0011, "step": 135470 }, { "epoch": 2.2168043851754886, "grad_norm": 0.22490908205509186, "learning_rate": 4.926679665470863e-06, "loss": 0.0007, "step": 135480 }, { "epoch": 2.2169680111265646, "grad_norm": 0.05447162687778473, "learning_rate": 4.925965790090271e-06, "loss": 0.0013, "step": 135490 }, { "epoch": 2.2171316370776406, "grad_norm": 0.0663597509264946, "learning_rate": 4.925251916219173e-06, "loss": 0.0013, "step": 135500 }, { "epoch": 2.217295263028716, "grad_norm": 0.08450702577829361, "learning_rate": 4.9245380438721255e-06, "loss": 0.0013, "step": 135510 }, { "epoch": 2.217458888979792, "grad_norm": 0.012310695834457874, "learning_rate": 4.923824173063681e-06, "loss": 0.0008, "step": 135520 }, { "epoch": 2.217622514930868, "grad_norm": 0.008670191280543804, "learning_rate": 4.923110303808399e-06, "loss": 0.0011, "step": 135530 }, { "epoch": 2.2177861408819437, "grad_norm": 0.06470473110675812, "learning_rate": 4.922396436120832e-06, "loss": 0.0005, "step": 135540 }, { "epoch": 2.2179497668330197, "grad_norm": 0.024450812488794327, "learning_rate": 4.921682570015535e-06, "loss": 0.0013, "step": 135550 }, { "epoch": 2.2181133927840957, "grad_norm": 0.051567044109106064, "learning_rate": 4.920968705507065e-06, "loss": 0.0009, "step": 135560 }, { "epoch": 2.2182770187351712, "grad_norm": 0.025189805775880814, "learning_rate": 4.920254842609975e-06, "loss": 0.0005, "step": 135570 }, { "epoch": 2.2184406446862472, "grad_norm": 0.04873934015631676, "learning_rate": 4.91954098133882e-06, "loss": 0.0013, "step": 135580 }, { "epoch": 2.2186042706373232, "grad_norm": 0.022029079496860504, "learning_rate": 4.918827121708159e-06, "loss": 0.001, "step": 135590 }, { "epoch": 2.218767896588399, "grad_norm": 0.061087772250175476, "learning_rate": 4.918113263732541e-06, "loss": 0.0015, "step": 135600 }, { "epoch": 2.218931522539475, "grad_norm": 0.0613812580704689, "learning_rate": 4.917399407426527e-06, "loss": 0.0011, "step": 135610 }, { "epoch": 2.219095148490551, "grad_norm": 0.05672299861907959, "learning_rate": 4.916685552804664e-06, "loss": 0.0008, "step": 135620 }, { "epoch": 2.2192587744416263, "grad_norm": 0.09995602816343307, "learning_rate": 4.9159716998815165e-06, "loss": 0.001, "step": 135630 }, { "epoch": 2.2194224003927023, "grad_norm": 0.01340377889573574, "learning_rate": 4.915257848671631e-06, "loss": 0.0009, "step": 135640 }, { "epoch": 2.2195860263437783, "grad_norm": 0.027505651116371155, "learning_rate": 4.914543999189569e-06, "loss": 0.0006, "step": 135650 }, { "epoch": 2.219749652294854, "grad_norm": 0.04771843180060387, "learning_rate": 4.913830151449879e-06, "loss": 0.0008, "step": 135660 }, { "epoch": 2.21991327824593, "grad_norm": 0.09249347448348999, "learning_rate": 4.913116305467122e-06, "loss": 0.001, "step": 135670 }, { "epoch": 2.220076904197006, "grad_norm": 0.057118240743875504, "learning_rate": 4.912402461255847e-06, "loss": 0.0008, "step": 135680 }, { "epoch": 2.2202405301480814, "grad_norm": 0.016187073662877083, "learning_rate": 4.911688618830614e-06, "loss": 0.0009, "step": 135690 }, { "epoch": 2.2204041560991574, "grad_norm": 0.025853684172034264, "learning_rate": 4.910974778205972e-06, "loss": 0.0005, "step": 135700 }, { "epoch": 2.220567782050233, "grad_norm": 0.0019744865130633116, "learning_rate": 4.910260939396481e-06, "loss": 0.0006, "step": 135710 }, { "epoch": 2.220731408001309, "grad_norm": 0.02814037911593914, "learning_rate": 4.90954710241669e-06, "loss": 0.0052, "step": 135720 }, { "epoch": 2.220895033952385, "grad_norm": 0.049398407340049744, "learning_rate": 4.9088332672811616e-06, "loss": 0.0011, "step": 135730 }, { "epoch": 2.2210586599034605, "grad_norm": 0.012196908704936504, "learning_rate": 4.908119434004441e-06, "loss": 0.0028, "step": 135740 }, { "epoch": 2.2212222858545365, "grad_norm": 0.025151187554001808, "learning_rate": 4.90740560260109e-06, "loss": 0.0019, "step": 135750 }, { "epoch": 2.2213859118056125, "grad_norm": 0.03184604272246361, "learning_rate": 4.906691773085657e-06, "loss": 0.0012, "step": 135760 }, { "epoch": 2.221549537756688, "grad_norm": 0.05211792513728142, "learning_rate": 4.905977945472702e-06, "loss": 0.0015, "step": 135770 }, { "epoch": 2.221713163707764, "grad_norm": 0.030694689601659775, "learning_rate": 4.905264119776775e-06, "loss": 0.0013, "step": 135780 }, { "epoch": 2.22187678965884, "grad_norm": 0.06385420262813568, "learning_rate": 4.904550296012433e-06, "loss": 0.0009, "step": 135790 }, { "epoch": 2.2220404156099156, "grad_norm": 0.025693150237202644, "learning_rate": 4.903836474194229e-06, "loss": 0.0008, "step": 135800 }, { "epoch": 2.2222040415609916, "grad_norm": 0.02226187475025654, "learning_rate": 4.903122654336718e-06, "loss": 0.0006, "step": 135810 }, { "epoch": 2.2223676675120676, "grad_norm": 0.07309363037347794, "learning_rate": 4.902408836454453e-06, "loss": 0.0006, "step": 135820 }, { "epoch": 2.222531293463143, "grad_norm": 0.012903676368296146, "learning_rate": 4.90169502056199e-06, "loss": 0.0011, "step": 135830 }, { "epoch": 2.222694919414219, "grad_norm": 0.016753556206822395, "learning_rate": 4.90098120667388e-06, "loss": 0.0006, "step": 135840 }, { "epoch": 2.2228585453652947, "grad_norm": 0.03989275544881821, "learning_rate": 4.90026739480468e-06, "loss": 0.0008, "step": 135850 }, { "epoch": 2.2230221713163707, "grad_norm": 0.0227245781570673, "learning_rate": 4.899553584968943e-06, "loss": 0.0012, "step": 135860 }, { "epoch": 2.2231857972674467, "grad_norm": 0.08162596821784973, "learning_rate": 4.898839777181224e-06, "loss": 0.0013, "step": 135870 }, { "epoch": 2.2233494232185222, "grad_norm": 0.01881076768040657, "learning_rate": 4.898125971456074e-06, "loss": 0.0013, "step": 135880 }, { "epoch": 2.2235130491695982, "grad_norm": 0.07620874792337418, "learning_rate": 4.89741216780805e-06, "loss": 0.0008, "step": 135890 }, { "epoch": 2.2236766751206742, "grad_norm": 0.001576818060129881, "learning_rate": 4.896698366251703e-06, "loss": 0.0009, "step": 135900 }, { "epoch": 2.22384030107175, "grad_norm": 0.022959688678383827, "learning_rate": 4.89598456680159e-06, "loss": 0.0009, "step": 135910 }, { "epoch": 2.224003927022826, "grad_norm": 0.051078010350465775, "learning_rate": 4.895270769472263e-06, "loss": 0.0021, "step": 135920 }, { "epoch": 2.224167552973902, "grad_norm": 0.10477802902460098, "learning_rate": 4.894556974278277e-06, "loss": 0.0009, "step": 135930 }, { "epoch": 2.2243311789249773, "grad_norm": 0.1004137247800827, "learning_rate": 4.893843181234182e-06, "loss": 0.0016, "step": 135940 }, { "epoch": 2.2244948048760533, "grad_norm": 0.010497752577066422, "learning_rate": 4.8931293903545365e-06, "loss": 0.0004, "step": 135950 }, { "epoch": 2.2246584308271293, "grad_norm": 0.07173757255077362, "learning_rate": 4.892415601653891e-06, "loss": 0.002, "step": 135960 }, { "epoch": 2.224822056778205, "grad_norm": 0.038229845464229584, "learning_rate": 4.8917018151468006e-06, "loss": 0.0008, "step": 135970 }, { "epoch": 2.224985682729281, "grad_norm": 0.00589515408501029, "learning_rate": 4.890988030847817e-06, "loss": 0.0006, "step": 135980 }, { "epoch": 2.225149308680357, "grad_norm": 0.1412743180990219, "learning_rate": 4.890274248771496e-06, "loss": 0.0011, "step": 135990 }, { "epoch": 2.2253129346314324, "grad_norm": 0.024792740121483803, "learning_rate": 4.8895604689323875e-06, "loss": 0.0016, "step": 136000 }, { "epoch": 2.2254765605825084, "grad_norm": 0.03297397494316101, "learning_rate": 4.88884669134505e-06, "loss": 0.001, "step": 136010 }, { "epoch": 2.2256401865335844, "grad_norm": 0.04641592502593994, "learning_rate": 4.888132916024031e-06, "loss": 0.0008, "step": 136020 }, { "epoch": 2.22580381248466, "grad_norm": 0.0679096058011055, "learning_rate": 4.887419142983889e-06, "loss": 0.0006, "step": 136030 }, { "epoch": 2.225967438435736, "grad_norm": 0.04353826493024826, "learning_rate": 4.886705372239174e-06, "loss": 0.0007, "step": 136040 }, { "epoch": 2.226131064386812, "grad_norm": 0.07977273315191269, "learning_rate": 4.88599160380444e-06, "loss": 0.0008, "step": 136050 }, { "epoch": 2.2262946903378875, "grad_norm": 0.057930607348680496, "learning_rate": 4.8852778376942405e-06, "loss": 0.0011, "step": 136060 }, { "epoch": 2.2264583162889635, "grad_norm": 0.11268320679664612, "learning_rate": 4.884564073923129e-06, "loss": 0.0011, "step": 136070 }, { "epoch": 2.226621942240039, "grad_norm": 0.06318831443786621, "learning_rate": 4.883850312505656e-06, "loss": 0.001, "step": 136080 }, { "epoch": 2.226785568191115, "grad_norm": 0.06829258799552917, "learning_rate": 4.883136553456379e-06, "loss": 0.0011, "step": 136090 }, { "epoch": 2.226949194142191, "grad_norm": 0.016489801928400993, "learning_rate": 4.882422796789846e-06, "loss": 0.001, "step": 136100 }, { "epoch": 2.2271128200932666, "grad_norm": 0.020804772153496742, "learning_rate": 4.881709042520613e-06, "loss": 0.001, "step": 136110 }, { "epoch": 2.2272764460443426, "grad_norm": 0.04441211000084877, "learning_rate": 4.880995290663231e-06, "loss": 0.0007, "step": 136120 }, { "epoch": 2.2274400719954186, "grad_norm": 0.03030546009540558, "learning_rate": 4.880281541232256e-06, "loss": 0.0011, "step": 136130 }, { "epoch": 2.227603697946494, "grad_norm": 0.05516219884157181, "learning_rate": 4.879567794242237e-06, "loss": 0.0017, "step": 136140 }, { "epoch": 2.22776732389757, "grad_norm": 0.023008499294519424, "learning_rate": 4.878854049707729e-06, "loss": 0.0012, "step": 136150 }, { "epoch": 2.227930949848646, "grad_norm": 0.07229474186897278, "learning_rate": 4.878140307643282e-06, "loss": 0.0014, "step": 136160 }, { "epoch": 2.2280945757997217, "grad_norm": 0.03444996848702431, "learning_rate": 4.877426568063454e-06, "loss": 0.0007, "step": 136170 }, { "epoch": 2.2282582017507977, "grad_norm": 0.08925872296094894, "learning_rate": 4.876712830982791e-06, "loss": 0.0009, "step": 136180 }, { "epoch": 2.2284218277018737, "grad_norm": 0.06056445837020874, "learning_rate": 4.87599909641585e-06, "loss": 0.001, "step": 136190 }, { "epoch": 2.2285854536529492, "grad_norm": 0.05294337123632431, "learning_rate": 4.875285364377181e-06, "loss": 0.0004, "step": 136200 }, { "epoch": 2.2287490796040252, "grad_norm": 0.003266116138547659, "learning_rate": 4.874571634881338e-06, "loss": 0.0006, "step": 136210 }, { "epoch": 2.2289127055551012, "grad_norm": 0.03397321328520775, "learning_rate": 4.873857907942872e-06, "loss": 0.0039, "step": 136220 }, { "epoch": 2.229076331506177, "grad_norm": 0.03611673787236214, "learning_rate": 4.873144183576338e-06, "loss": 0.0018, "step": 136230 }, { "epoch": 2.229239957457253, "grad_norm": 0.032876044511795044, "learning_rate": 4.872430461796283e-06, "loss": 0.0007, "step": 136240 }, { "epoch": 2.2294035834083283, "grad_norm": 0.05431939661502838, "learning_rate": 4.871716742617266e-06, "loss": 0.0011, "step": 136250 }, { "epoch": 2.2295672093594043, "grad_norm": 0.039801329374313354, "learning_rate": 4.8710030260538325e-06, "loss": 0.0009, "step": 136260 }, { "epoch": 2.2297308353104803, "grad_norm": 0.02323911339044571, "learning_rate": 4.87028931212054e-06, "loss": 0.001, "step": 136270 }, { "epoch": 2.229894461261556, "grad_norm": 0.01903504692018032, "learning_rate": 4.869575600831936e-06, "loss": 0.0006, "step": 136280 }, { "epoch": 2.230058087212632, "grad_norm": 0.04901636391878128, "learning_rate": 4.868861892202577e-06, "loss": 0.0009, "step": 136290 }, { "epoch": 2.230221713163708, "grad_norm": 0.004010640550404787, "learning_rate": 4.8681481862470085e-06, "loss": 0.0005, "step": 136300 }, { "epoch": 2.2303853391147834, "grad_norm": 0.012716557830572128, "learning_rate": 4.867434482979791e-06, "loss": 0.0008, "step": 136310 }, { "epoch": 2.2305489650658594, "grad_norm": 0.019696276634931564, "learning_rate": 4.866720782415467e-06, "loss": 0.0008, "step": 136320 }, { "epoch": 2.2307125910169354, "grad_norm": 0.010206971317529678, "learning_rate": 4.8660070845685975e-06, "loss": 0.0008, "step": 136330 }, { "epoch": 2.230876216968011, "grad_norm": 0.03523840010166168, "learning_rate": 4.865293389453725e-06, "loss": 0.0006, "step": 136340 }, { "epoch": 2.231039842919087, "grad_norm": 0.029572876170277596, "learning_rate": 4.86457969708541e-06, "loss": 0.0015, "step": 136350 }, { "epoch": 2.231203468870163, "grad_norm": 0.0369398407638073, "learning_rate": 4.863866007478197e-06, "loss": 0.0012, "step": 136360 }, { "epoch": 2.2313670948212385, "grad_norm": 0.01035348605364561, "learning_rate": 4.863152320646642e-06, "loss": 0.0007, "step": 136370 }, { "epoch": 2.2315307207723145, "grad_norm": 0.019981807097792625, "learning_rate": 4.8624386366052925e-06, "loss": 0.0009, "step": 136380 }, { "epoch": 2.2316943467233905, "grad_norm": 0.06241169199347496, "learning_rate": 4.861724955368706e-06, "loss": 0.0008, "step": 136390 }, { "epoch": 2.231857972674466, "grad_norm": 0.02910880744457245, "learning_rate": 4.861011276951426e-06, "loss": 0.0007, "step": 136400 }, { "epoch": 2.232021598625542, "grad_norm": 0.41979601979255676, "learning_rate": 4.860297601368011e-06, "loss": 0.001, "step": 136410 }, { "epoch": 2.232185224576618, "grad_norm": 0.03938192129135132, "learning_rate": 4.859583928633007e-06, "loss": 0.0011, "step": 136420 }, { "epoch": 2.2323488505276936, "grad_norm": 0.02317868359386921, "learning_rate": 4.858870258760969e-06, "loss": 0.0006, "step": 136430 }, { "epoch": 2.2325124764787696, "grad_norm": 0.01884492114186287, "learning_rate": 4.8581565917664455e-06, "loss": 0.0006, "step": 136440 }, { "epoch": 2.2326761024298456, "grad_norm": 0.016229625791311264, "learning_rate": 4.857442927663989e-06, "loss": 0.0011, "step": 136450 }, { "epoch": 2.232839728380921, "grad_norm": 0.14609397947788239, "learning_rate": 4.856729266468149e-06, "loss": 0.0013, "step": 136460 }, { "epoch": 2.233003354331997, "grad_norm": 0.0581982359290123, "learning_rate": 4.856015608193478e-06, "loss": 0.0009, "step": 136470 }, { "epoch": 2.2331669802830727, "grad_norm": 0.027580274268984795, "learning_rate": 4.855301952854525e-06, "loss": 0.0012, "step": 136480 }, { "epoch": 2.2333306062341487, "grad_norm": 0.012602800503373146, "learning_rate": 4.854588300465844e-06, "loss": 0.0011, "step": 136490 }, { "epoch": 2.2334942321852247, "grad_norm": 0.06348510831594467, "learning_rate": 4.853874651041983e-06, "loss": 0.0009, "step": 136500 }, { "epoch": 2.2336578581363002, "grad_norm": 0.0132713308557868, "learning_rate": 4.8531610045974945e-06, "loss": 0.0014, "step": 136510 }, { "epoch": 2.2338214840873762, "grad_norm": 0.008429369889199734, "learning_rate": 4.852447361146926e-06, "loss": 0.0012, "step": 136520 }, { "epoch": 2.2339851100384522, "grad_norm": 0.000992781831882894, "learning_rate": 4.851733720704832e-06, "loss": 0.0005, "step": 136530 }, { "epoch": 2.234148735989528, "grad_norm": 0.005896254908293486, "learning_rate": 4.851020083285761e-06, "loss": 0.0004, "step": 136540 }, { "epoch": 2.234312361940604, "grad_norm": 0.05347975343465805, "learning_rate": 4.850306448904265e-06, "loss": 0.0006, "step": 136550 }, { "epoch": 2.23447598789168, "grad_norm": 0.00858096219599247, "learning_rate": 4.84959281757489e-06, "loss": 0.001, "step": 136560 }, { "epoch": 2.2346396138427553, "grad_norm": 0.020868953317403793, "learning_rate": 4.8488791893121935e-06, "loss": 0.0006, "step": 136570 }, { "epoch": 2.2348032397938313, "grad_norm": 0.002544090151786804, "learning_rate": 4.848165564130719e-06, "loss": 0.0003, "step": 136580 }, { "epoch": 2.2349668657449073, "grad_norm": 0.025777719914913177, "learning_rate": 4.847451942045021e-06, "loss": 0.0015, "step": 136590 }, { "epoch": 2.235130491695983, "grad_norm": 0.0656517967581749, "learning_rate": 4.846738323069647e-06, "loss": 0.0012, "step": 136600 }, { "epoch": 2.235294117647059, "grad_norm": 0.05185987800359726, "learning_rate": 4.846024707219149e-06, "loss": 0.0006, "step": 136610 }, { "epoch": 2.235457743598135, "grad_norm": 0.030710870400071144, "learning_rate": 4.8453110945080764e-06, "loss": 0.0007, "step": 136620 }, { "epoch": 2.2356213695492104, "grad_norm": 0.005142407491803169, "learning_rate": 4.844597484950978e-06, "loss": 0.0013, "step": 136630 }, { "epoch": 2.2357849955002864, "grad_norm": 0.003574354574084282, "learning_rate": 4.843883878562406e-06, "loss": 0.0007, "step": 136640 }, { "epoch": 2.235948621451362, "grad_norm": 0.03961043059825897, "learning_rate": 4.843170275356909e-06, "loss": 0.0012, "step": 136650 }, { "epoch": 2.236112247402438, "grad_norm": 0.022183796390891075, "learning_rate": 4.8424566753490355e-06, "loss": 0.0009, "step": 136660 }, { "epoch": 2.236275873353514, "grad_norm": 0.017516983672976494, "learning_rate": 4.841743078553337e-06, "loss": 0.0006, "step": 136670 }, { "epoch": 2.2364394993045895, "grad_norm": 0.018794480711221695, "learning_rate": 4.841029484984362e-06, "loss": 0.0005, "step": 136680 }, { "epoch": 2.2366031252556655, "grad_norm": 0.01691107079386711, "learning_rate": 4.840315894656662e-06, "loss": 0.0011, "step": 136690 }, { "epoch": 2.2367667512067415, "grad_norm": 0.05239488184452057, "learning_rate": 4.839602307584783e-06, "loss": 0.0005, "step": 136700 }, { "epoch": 2.236930377157817, "grad_norm": 0.04883141070604324, "learning_rate": 4.838888723783279e-06, "loss": 0.001, "step": 136710 }, { "epoch": 2.237094003108893, "grad_norm": 0.024089312180876732, "learning_rate": 4.838175143266695e-06, "loss": 0.0007, "step": 136720 }, { "epoch": 2.237257629059969, "grad_norm": 0.1367446780204773, "learning_rate": 4.837461566049583e-06, "loss": 0.0016, "step": 136730 }, { "epoch": 2.2374212550110446, "grad_norm": 0.051648810505867004, "learning_rate": 4.836747992146491e-06, "loss": 0.0007, "step": 136740 }, { "epoch": 2.2375848809621206, "grad_norm": 0.06167486682534218, "learning_rate": 4.836034421571969e-06, "loss": 0.0011, "step": 136750 }, { "epoch": 2.2377485069131966, "grad_norm": 0.17730392515659332, "learning_rate": 4.835320854340565e-06, "loss": 0.0016, "step": 136760 }, { "epoch": 2.237912132864272, "grad_norm": 0.059962011873722076, "learning_rate": 4.834607290466831e-06, "loss": 0.0011, "step": 136770 }, { "epoch": 2.238075758815348, "grad_norm": 0.07671120762825012, "learning_rate": 4.833893729965311e-06, "loss": 0.0015, "step": 136780 }, { "epoch": 2.238239384766424, "grad_norm": 0.09387888759374619, "learning_rate": 4.8331801728505585e-06, "loss": 0.0017, "step": 136790 }, { "epoch": 2.2384030107174997, "grad_norm": 0.11376676708459854, "learning_rate": 4.832466619137119e-06, "loss": 0.0011, "step": 136800 }, { "epoch": 2.2385666366685757, "grad_norm": 0.024699803441762924, "learning_rate": 4.831753068839543e-06, "loss": 0.0011, "step": 136810 }, { "epoch": 2.2387302626196517, "grad_norm": 0.0372159443795681, "learning_rate": 4.831039521972379e-06, "loss": 0.0017, "step": 136820 }, { "epoch": 2.2388938885707272, "grad_norm": 0.023022806271910667, "learning_rate": 4.830325978550175e-06, "loss": 0.0009, "step": 136830 }, { "epoch": 2.2390575145218032, "grad_norm": 0.08016005903482437, "learning_rate": 4.829612438587481e-06, "loss": 0.0009, "step": 136840 }, { "epoch": 2.239221140472879, "grad_norm": 0.02491956576704979, "learning_rate": 4.828898902098844e-06, "loss": 0.0004, "step": 136850 }, { "epoch": 2.239384766423955, "grad_norm": 0.04345036298036575, "learning_rate": 4.828185369098813e-06, "loss": 0.001, "step": 136860 }, { "epoch": 2.239548392375031, "grad_norm": 0.04476727172732353, "learning_rate": 4.827471839601937e-06, "loss": 0.0006, "step": 136870 }, { "epoch": 2.2397120183261063, "grad_norm": 0.025731422007083893, "learning_rate": 4.826758313622761e-06, "loss": 0.0006, "step": 136880 }, { "epoch": 2.2398756442771823, "grad_norm": 0.05992526933550835, "learning_rate": 4.8260447911758385e-06, "loss": 0.0008, "step": 136890 }, { "epoch": 2.2400392702282583, "grad_norm": 0.10012935101985931, "learning_rate": 4.825331272275712e-06, "loss": 0.0009, "step": 136900 }, { "epoch": 2.240202896179334, "grad_norm": 0.042603347450494766, "learning_rate": 4.824617756936935e-06, "loss": 0.0007, "step": 136910 }, { "epoch": 2.24036652213041, "grad_norm": 0.02174074947834015, "learning_rate": 4.82390424517405e-06, "loss": 0.0004, "step": 136920 }, { "epoch": 2.240530148081486, "grad_norm": 0.0041737849824130535, "learning_rate": 4.823190737001611e-06, "loss": 0.0006, "step": 136930 }, { "epoch": 2.2406937740325614, "grad_norm": 0.20565065741539001, "learning_rate": 4.822477232434158e-06, "loss": 0.0008, "step": 136940 }, { "epoch": 2.2408573999836374, "grad_norm": 0.021674079820513725, "learning_rate": 4.821763731486248e-06, "loss": 0.0012, "step": 136950 }, { "epoch": 2.2410210259347134, "grad_norm": 0.005679299123585224, "learning_rate": 4.82105023417242e-06, "loss": 0.0005, "step": 136960 }, { "epoch": 2.241184651885789, "grad_norm": 0.11921368539333344, "learning_rate": 4.820336740507229e-06, "loss": 0.0006, "step": 136970 }, { "epoch": 2.241348277836865, "grad_norm": 0.026314599439501762, "learning_rate": 4.819623250505216e-06, "loss": 0.005, "step": 136980 }, { "epoch": 2.241511903787941, "grad_norm": 0.07375772297382355, "learning_rate": 4.818909764180935e-06, "loss": 0.0014, "step": 136990 }, { "epoch": 2.2416755297390165, "grad_norm": 0.041399579495191574, "learning_rate": 4.818196281548925e-06, "loss": 0.0006, "step": 137000 }, { "epoch": 2.2418391556900925, "grad_norm": 0.10114839673042297, "learning_rate": 4.817482802623743e-06, "loss": 0.0009, "step": 137010 }, { "epoch": 2.242002781641168, "grad_norm": 0.11268191784620285, "learning_rate": 4.816769327419928e-06, "loss": 0.0005, "step": 137020 }, { "epoch": 2.242166407592244, "grad_norm": 0.08767176419496536, "learning_rate": 4.816055855952034e-06, "loss": 0.0004, "step": 137030 }, { "epoch": 2.24233003354332, "grad_norm": 0.06323045492172241, "learning_rate": 4.8153423882346005e-06, "loss": 0.0018, "step": 137040 }, { "epoch": 2.2424936594943956, "grad_norm": 0.020325617864727974, "learning_rate": 4.814628924282182e-06, "loss": 0.0005, "step": 137050 }, { "epoch": 2.2426572854454716, "grad_norm": 0.0521891824901104, "learning_rate": 4.813915464109321e-06, "loss": 0.0007, "step": 137060 }, { "epoch": 2.2428209113965476, "grad_norm": 0.006553840357810259, "learning_rate": 4.813202007730566e-06, "loss": 0.0004, "step": 137070 }, { "epoch": 2.242984537347623, "grad_norm": 0.001363706192933023, "learning_rate": 4.812488555160461e-06, "loss": 0.0007, "step": 137080 }, { "epoch": 2.243148163298699, "grad_norm": 0.023358307778835297, "learning_rate": 4.8117751064135574e-06, "loss": 0.0006, "step": 137090 }, { "epoch": 2.243311789249775, "grad_norm": 0.04543708637356758, "learning_rate": 4.811061661504398e-06, "loss": 0.0008, "step": 137100 }, { "epoch": 2.2434754152008507, "grad_norm": 0.058586686849594116, "learning_rate": 4.810348220447532e-06, "loss": 0.0009, "step": 137110 }, { "epoch": 2.2436390411519267, "grad_norm": 0.23546025156974792, "learning_rate": 4.809634783257502e-06, "loss": 0.0009, "step": 137120 }, { "epoch": 2.2438026671030027, "grad_norm": 0.05789906531572342, "learning_rate": 4.80892134994886e-06, "loss": 0.0005, "step": 137130 }, { "epoch": 2.2439662930540782, "grad_norm": 0.04355677217245102, "learning_rate": 4.808207920536146e-06, "loss": 0.0009, "step": 137140 }, { "epoch": 2.2441299190051542, "grad_norm": 0.05667303130030632, "learning_rate": 4.807494495033912e-06, "loss": 0.0013, "step": 137150 }, { "epoch": 2.2442935449562302, "grad_norm": 0.04367038235068321, "learning_rate": 4.8067810734566996e-06, "loss": 0.0007, "step": 137160 }, { "epoch": 2.244457170907306, "grad_norm": 0.03545735403895378, "learning_rate": 4.8060676558190585e-06, "loss": 0.0008, "step": 137170 }, { "epoch": 2.244620796858382, "grad_norm": 0.019875597208738327, "learning_rate": 4.805354242135531e-06, "loss": 0.0005, "step": 137180 }, { "epoch": 2.244784422809458, "grad_norm": 0.1285896897315979, "learning_rate": 4.804640832420667e-06, "loss": 0.0022, "step": 137190 }, { "epoch": 2.2449480487605333, "grad_norm": 0.06753906607627869, "learning_rate": 4.803927426689009e-06, "loss": 0.0009, "step": 137200 }, { "epoch": 2.2451116747116093, "grad_norm": 0.08544394373893738, "learning_rate": 4.803214024955104e-06, "loss": 0.0009, "step": 137210 }, { "epoch": 2.2452753006626853, "grad_norm": 0.052545249462127686, "learning_rate": 4.802500627233498e-06, "loss": 0.0006, "step": 137220 }, { "epoch": 2.245438926613761, "grad_norm": 0.0023216051049530506, "learning_rate": 4.801787233538737e-06, "loss": 0.0012, "step": 137230 }, { "epoch": 2.245602552564837, "grad_norm": 0.01158390287309885, "learning_rate": 4.8010738438853635e-06, "loss": 0.0005, "step": 137240 }, { "epoch": 2.2457661785159124, "grad_norm": 0.08295346796512604, "learning_rate": 4.800360458287928e-06, "loss": 0.0009, "step": 137250 }, { "epoch": 2.2459298044669884, "grad_norm": 0.055965520441532135, "learning_rate": 4.79964707676097e-06, "loss": 0.0013, "step": 137260 }, { "epoch": 2.2460934304180644, "grad_norm": 0.02236192673444748, "learning_rate": 4.79893369931904e-06, "loss": 0.0015, "step": 137270 }, { "epoch": 2.24625705636914, "grad_norm": 0.10377421230077744, "learning_rate": 4.79822032597668e-06, "loss": 0.0011, "step": 137280 }, { "epoch": 2.246420682320216, "grad_norm": 0.026498612016439438, "learning_rate": 4.797506956748437e-06, "loss": 0.0011, "step": 137290 }, { "epoch": 2.246584308271292, "grad_norm": 0.02492314949631691, "learning_rate": 4.796793591648853e-06, "loss": 0.0011, "step": 137300 }, { "epoch": 2.2467479342223675, "grad_norm": 0.06940802186727524, "learning_rate": 4.796080230692477e-06, "loss": 0.0015, "step": 137310 }, { "epoch": 2.2469115601734435, "grad_norm": 0.06205128878355026, "learning_rate": 4.79536687389385e-06, "loss": 0.001, "step": 137320 }, { "epoch": 2.2470751861245195, "grad_norm": 0.02677728980779648, "learning_rate": 4.7946535212675196e-06, "loss": 0.0008, "step": 137330 }, { "epoch": 2.247238812075595, "grad_norm": 0.06873565167188644, "learning_rate": 4.793940172828028e-06, "loss": 0.001, "step": 137340 }, { "epoch": 2.247402438026671, "grad_norm": 0.0732707604765892, "learning_rate": 4.793226828589922e-06, "loss": 0.0009, "step": 137350 }, { "epoch": 2.247566063977747, "grad_norm": 0.03581288084387779, "learning_rate": 4.792513488567743e-06, "loss": 0.0008, "step": 137360 }, { "epoch": 2.2477296899288226, "grad_norm": 0.031709935516119, "learning_rate": 4.791800152776041e-06, "loss": 0.0004, "step": 137370 }, { "epoch": 2.2478933158798986, "grad_norm": 0.040510307997465134, "learning_rate": 4.791086821229355e-06, "loss": 0.0006, "step": 137380 }, { "epoch": 2.2480569418309746, "grad_norm": 0.057531047612428665, "learning_rate": 4.790373493942231e-06, "loss": 0.001, "step": 137390 }, { "epoch": 2.24822056778205, "grad_norm": 0.01916862465441227, "learning_rate": 4.789660170929213e-06, "loss": 0.0005, "step": 137400 }, { "epoch": 2.248384193733126, "grad_norm": 0.04279434308409691, "learning_rate": 4.788946852204846e-06, "loss": 0.0008, "step": 137410 }, { "epoch": 2.2485478196842017, "grad_norm": 0.08369458466768265, "learning_rate": 4.788233537783672e-06, "loss": 0.0011, "step": 137420 }, { "epoch": 2.2487114456352777, "grad_norm": 0.04263691604137421, "learning_rate": 4.787520227680237e-06, "loss": 0.0008, "step": 137430 }, { "epoch": 2.2488750715863537, "grad_norm": 0.07760177552700043, "learning_rate": 4.786806921909084e-06, "loss": 0.0009, "step": 137440 }, { "epoch": 2.2490386975374292, "grad_norm": 0.017290601506829262, "learning_rate": 4.786093620484756e-06, "loss": 0.0008, "step": 137450 }, { "epoch": 2.2492023234885052, "grad_norm": 0.09822747856378555, "learning_rate": 4.785380323421797e-06, "loss": 0.0007, "step": 137460 }, { "epoch": 2.2493659494395812, "grad_norm": 0.031062781810760498, "learning_rate": 4.784667030734751e-06, "loss": 0.0006, "step": 137470 }, { "epoch": 2.249529575390657, "grad_norm": 0.03921612724661827, "learning_rate": 4.783953742438161e-06, "loss": 0.0009, "step": 137480 }, { "epoch": 2.249693201341733, "grad_norm": 0.0023693866096436977, "learning_rate": 4.783240458546571e-06, "loss": 0.0013, "step": 137490 }, { "epoch": 2.249856827292809, "grad_norm": 0.05162276700139046, "learning_rate": 4.782527179074523e-06, "loss": 0.001, "step": 137500 }, { "epoch": 2.2500204532438843, "grad_norm": 0.06204346939921379, "learning_rate": 4.7818139040365615e-06, "loss": 0.002, "step": 137510 }, { "epoch": 2.2501840791949603, "grad_norm": 0.03228798881173134, "learning_rate": 4.781100633447228e-06, "loss": 0.0016, "step": 137520 }, { "epoch": 2.2503477051460363, "grad_norm": 0.06397654116153717, "learning_rate": 4.780387367321069e-06, "loss": 0.0007, "step": 137530 }, { "epoch": 2.250511331097112, "grad_norm": 0.001751047559082508, "learning_rate": 4.779674105672621e-06, "loss": 0.0006, "step": 137540 }, { "epoch": 2.250674957048188, "grad_norm": 0.022377820685505867, "learning_rate": 4.778960848516434e-06, "loss": 0.0014, "step": 137550 }, { "epoch": 2.250838582999264, "grad_norm": 0.07798672467470169, "learning_rate": 4.7782475958670435e-06, "loss": 0.0014, "step": 137560 }, { "epoch": 2.2510022089503394, "grad_norm": 0.010853045620024204, "learning_rate": 4.7775343477390004e-06, "loss": 0.0006, "step": 137570 }, { "epoch": 2.2511658349014154, "grad_norm": 0.029433539137244225, "learning_rate": 4.776821104146839e-06, "loss": 0.0008, "step": 137580 }, { "epoch": 2.2513294608524914, "grad_norm": 0.02967381477355957, "learning_rate": 4.776107865105108e-06, "loss": 0.002, "step": 137590 }, { "epoch": 2.251493086803567, "grad_norm": 0.018464982509613037, "learning_rate": 4.7753946306283446e-06, "loss": 0.0007, "step": 137600 }, { "epoch": 2.251656712754643, "grad_norm": 0.05829204246401787, "learning_rate": 4.7746814007310975e-06, "loss": 0.0008, "step": 137610 }, { "epoch": 2.251820338705719, "grad_norm": 0.05379010736942291, "learning_rate": 4.773968175427901e-06, "loss": 0.0009, "step": 137620 }, { "epoch": 2.2519839646567945, "grad_norm": 0.05923907458782196, "learning_rate": 4.773254954733305e-06, "loss": 0.0007, "step": 137630 }, { "epoch": 2.2521475906078705, "grad_norm": 0.07876089960336685, "learning_rate": 4.772541738661844e-06, "loss": 0.0009, "step": 137640 }, { "epoch": 2.252311216558946, "grad_norm": 0.01678679697215557, "learning_rate": 4.771828527228067e-06, "loss": 0.0012, "step": 137650 }, { "epoch": 2.252474842510022, "grad_norm": 0.04166676104068756, "learning_rate": 4.771115320446508e-06, "loss": 0.0007, "step": 137660 }, { "epoch": 2.252638468461098, "grad_norm": 0.07262758910655975, "learning_rate": 4.770402118331717e-06, "loss": 0.0011, "step": 137670 }, { "epoch": 2.2528020944121736, "grad_norm": 0.21515633165836334, "learning_rate": 4.7696889208982275e-06, "loss": 0.0009, "step": 137680 }, { "epoch": 2.2529657203632496, "grad_norm": 0.03343009203672409, "learning_rate": 4.7689757281605894e-06, "loss": 0.0009, "step": 137690 }, { "epoch": 2.2531293463143256, "grad_norm": 0.007578182499855757, "learning_rate": 4.768262540133337e-06, "loss": 0.0009, "step": 137700 }, { "epoch": 2.253292972265401, "grad_norm": 0.12735112011432648, "learning_rate": 4.767549356831015e-06, "loss": 0.0014, "step": 137710 }, { "epoch": 2.253456598216477, "grad_norm": 0.05894704908132553, "learning_rate": 4.766836178268163e-06, "loss": 0.001, "step": 137720 }, { "epoch": 2.253620224167553, "grad_norm": 0.06946303695440292, "learning_rate": 4.766123004459326e-06, "loss": 0.0009, "step": 137730 }, { "epoch": 2.2537838501186287, "grad_norm": 0.02698449417948723, "learning_rate": 4.765409835419039e-06, "loss": 0.0009, "step": 137740 }, { "epoch": 2.2539474760697047, "grad_norm": 0.012041737325489521, "learning_rate": 4.7646966711618474e-06, "loss": 0.001, "step": 137750 }, { "epoch": 2.2541111020207807, "grad_norm": 0.04818934202194214, "learning_rate": 4.76398351170229e-06, "loss": 0.0007, "step": 137760 }, { "epoch": 2.2542747279718562, "grad_norm": 0.06737493723630905, "learning_rate": 4.763270357054909e-06, "loss": 0.001, "step": 137770 }, { "epoch": 2.2544383539229322, "grad_norm": 0.023138169199228287, "learning_rate": 4.762557207234242e-06, "loss": 0.0009, "step": 137780 }, { "epoch": 2.254601979874008, "grad_norm": 0.026320571079850197, "learning_rate": 4.761844062254833e-06, "loss": 0.001, "step": 137790 }, { "epoch": 2.254765605825084, "grad_norm": 0.003280928824096918, "learning_rate": 4.761130922131221e-06, "loss": 0.0004, "step": 137800 }, { "epoch": 2.25492923177616, "grad_norm": 0.09396369755268097, "learning_rate": 4.760417786877946e-06, "loss": 0.0011, "step": 137810 }, { "epoch": 2.2550928577272353, "grad_norm": 0.031820930540561676, "learning_rate": 4.759704656509549e-06, "loss": 0.0008, "step": 137820 }, { "epoch": 2.2552564836783113, "grad_norm": 0.08496785908937454, "learning_rate": 4.75899153104057e-06, "loss": 0.001, "step": 137830 }, { "epoch": 2.2554201096293873, "grad_norm": 0.036943767219781876, "learning_rate": 4.758278410485547e-06, "loss": 0.001, "step": 137840 }, { "epoch": 2.255583735580463, "grad_norm": 0.0017876537749543786, "learning_rate": 4.757565294859024e-06, "loss": 0.0006, "step": 137850 }, { "epoch": 2.255747361531539, "grad_norm": 0.0891231819987297, "learning_rate": 4.756852184175537e-06, "loss": 0.0016, "step": 137860 }, { "epoch": 2.255910987482615, "grad_norm": 0.011304762214422226, "learning_rate": 4.756139078449628e-06, "loss": 0.0046, "step": 137870 }, { "epoch": 2.2560746134336904, "grad_norm": 0.023799926042556763, "learning_rate": 4.755425977695834e-06, "loss": 0.0011, "step": 137880 }, { "epoch": 2.2562382393847664, "grad_norm": 0.03461712226271629, "learning_rate": 4.754712881928699e-06, "loss": 0.0006, "step": 137890 }, { "epoch": 2.2564018653358424, "grad_norm": 0.11013227701187134, "learning_rate": 4.753999791162757e-06, "loss": 0.0015, "step": 137900 }, { "epoch": 2.256565491286918, "grad_norm": 0.014244548045098782, "learning_rate": 4.753286705412552e-06, "loss": 0.0018, "step": 137910 }, { "epoch": 2.256729117237994, "grad_norm": 0.01116438303142786, "learning_rate": 4.75257362469262e-06, "loss": 0.001, "step": 137920 }, { "epoch": 2.25689274318907, "grad_norm": 0.03783174604177475, "learning_rate": 4.751860549017503e-06, "loss": 0.0005, "step": 137930 }, { "epoch": 2.2570563691401455, "grad_norm": 0.024838900193572044, "learning_rate": 4.7511474784017365e-06, "loss": 0.0011, "step": 137940 }, { "epoch": 2.2572199950912215, "grad_norm": 0.06048784777522087, "learning_rate": 4.750434412859863e-06, "loss": 0.0009, "step": 137950 }, { "epoch": 2.2573836210422975, "grad_norm": 0.0058603142388165, "learning_rate": 4.749721352406418e-06, "loss": 0.0016, "step": 137960 }, { "epoch": 2.257547246993373, "grad_norm": 0.0650729313492775, "learning_rate": 4.749008297055943e-06, "loss": 0.0017, "step": 137970 }, { "epoch": 2.257710872944449, "grad_norm": 0.03696402534842491, "learning_rate": 4.7482952468229745e-06, "loss": 0.0008, "step": 137980 }, { "epoch": 2.257874498895525, "grad_norm": 0.03054935671389103, "learning_rate": 4.747582201722053e-06, "loss": 0.0009, "step": 137990 }, { "epoch": 2.2580381248466006, "grad_norm": 0.038469716906547546, "learning_rate": 4.746869161767714e-06, "loss": 0.0007, "step": 138000 }, { "epoch": 2.2582017507976766, "grad_norm": 0.04975133389234543, "learning_rate": 4.7461561269744986e-06, "loss": 0.0013, "step": 138010 }, { "epoch": 2.2583653767487526, "grad_norm": 0.03923806548118591, "learning_rate": 4.745443097356943e-06, "loss": 0.0011, "step": 138020 }, { "epoch": 2.258529002699828, "grad_norm": 0.0107218436896801, "learning_rate": 4.744730072929588e-06, "loss": 0.0005, "step": 138030 }, { "epoch": 2.258692628650904, "grad_norm": 0.021101906895637512, "learning_rate": 4.744017053706967e-06, "loss": 0.0007, "step": 138040 }, { "epoch": 2.2588562546019797, "grad_norm": 0.06558126956224442, "learning_rate": 4.743304039703623e-06, "loss": 0.0008, "step": 138050 }, { "epoch": 2.2590198805530557, "grad_norm": 0.05895255133509636, "learning_rate": 4.74259103093409e-06, "loss": 0.001, "step": 138060 }, { "epoch": 2.2591835065041317, "grad_norm": 0.05204280465841293, "learning_rate": 4.7418780274129085e-06, "loss": 0.0005, "step": 138070 }, { "epoch": 2.2593471324552072, "grad_norm": 0.02131231501698494, "learning_rate": 4.741165029154612e-06, "loss": 0.0008, "step": 138080 }, { "epoch": 2.2595107584062832, "grad_norm": 0.03802712261676788, "learning_rate": 4.740452036173742e-06, "loss": 0.0009, "step": 138090 }, { "epoch": 2.2596743843573592, "grad_norm": 0.053505949676036835, "learning_rate": 4.739739048484834e-06, "loss": 0.0006, "step": 138100 }, { "epoch": 2.259838010308435, "grad_norm": 0.07883110642433167, "learning_rate": 4.739026066102427e-06, "loss": 0.001, "step": 138110 }, { "epoch": 2.260001636259511, "grad_norm": 0.20982930064201355, "learning_rate": 4.7383130890410535e-06, "loss": 0.001, "step": 138120 }, { "epoch": 2.260165262210587, "grad_norm": 0.02627575770020485, "learning_rate": 4.737600117315256e-06, "loss": 0.0008, "step": 138130 }, { "epoch": 2.2603288881616623, "grad_norm": 0.005831717513501644, "learning_rate": 4.736887150939568e-06, "loss": 0.001, "step": 138140 }, { "epoch": 2.2604925141127383, "grad_norm": 0.0074350787326693535, "learning_rate": 4.736174189928528e-06, "loss": 0.0007, "step": 138150 }, { "epoch": 2.260656140063814, "grad_norm": 0.03882322087883949, "learning_rate": 4.73546123429667e-06, "loss": 0.0007, "step": 138160 }, { "epoch": 2.26081976601489, "grad_norm": 0.018625950440764427, "learning_rate": 4.734748284058536e-06, "loss": 0.0009, "step": 138170 }, { "epoch": 2.260983391965966, "grad_norm": 0.011961298063397408, "learning_rate": 4.734035339228655e-06, "loss": 0.0006, "step": 138180 }, { "epoch": 2.2611470179170414, "grad_norm": 0.05012591928243637, "learning_rate": 4.73332239982157e-06, "loss": 0.0012, "step": 138190 }, { "epoch": 2.2613106438681174, "grad_norm": 0.17867083847522736, "learning_rate": 4.732609465851812e-06, "loss": 0.0008, "step": 138200 }, { "epoch": 2.2614742698191934, "grad_norm": 0.008764644153416157, "learning_rate": 4.731896537333922e-06, "loss": 0.0008, "step": 138210 }, { "epoch": 2.261637895770269, "grad_norm": 0.003978605382144451, "learning_rate": 4.731183614282431e-06, "loss": 0.0005, "step": 138220 }, { "epoch": 2.261801521721345, "grad_norm": 0.008799129165709019, "learning_rate": 4.730470696711881e-06, "loss": 0.0007, "step": 138230 }, { "epoch": 2.261965147672421, "grad_norm": 0.04101108759641647, "learning_rate": 4.7297577846367996e-06, "loss": 0.001, "step": 138240 }, { "epoch": 2.2621287736234965, "grad_norm": 0.06197251006960869, "learning_rate": 4.729044878071732e-06, "loss": 0.0008, "step": 138250 }, { "epoch": 2.2622923995745725, "grad_norm": 0.016297971829771996, "learning_rate": 4.728331977031205e-06, "loss": 0.0008, "step": 138260 }, { "epoch": 2.2624560255256485, "grad_norm": 0.07796336710453033, "learning_rate": 4.727619081529761e-06, "loss": 0.0016, "step": 138270 }, { "epoch": 2.262619651476724, "grad_norm": 0.049543652683496475, "learning_rate": 4.726906191581929e-06, "loss": 0.0008, "step": 138280 }, { "epoch": 2.2627832774278, "grad_norm": 0.10022540390491486, "learning_rate": 4.726193307202251e-06, "loss": 0.0012, "step": 138290 }, { "epoch": 2.262946903378876, "grad_norm": 0.12135680764913559, "learning_rate": 4.725480428405255e-06, "loss": 0.0011, "step": 138300 }, { "epoch": 2.2631105293299516, "grad_norm": 0.01941358670592308, "learning_rate": 4.724767555205484e-06, "loss": 0.0006, "step": 138310 }, { "epoch": 2.2632741552810276, "grad_norm": 0.09635898470878601, "learning_rate": 4.724054687617464e-06, "loss": 0.0014, "step": 138320 }, { "epoch": 2.2634377812321036, "grad_norm": 0.1568843573331833, "learning_rate": 4.723341825655738e-06, "loss": 0.0014, "step": 138330 }, { "epoch": 2.263601407183179, "grad_norm": 0.026507794857025146, "learning_rate": 4.722628969334833e-06, "loss": 0.0008, "step": 138340 }, { "epoch": 2.263765033134255, "grad_norm": 0.059126563370227814, "learning_rate": 4.721916118669292e-06, "loss": 0.0011, "step": 138350 }, { "epoch": 2.263928659085331, "grad_norm": 0.005907712038606405, "learning_rate": 4.721203273673641e-06, "loss": 0.0005, "step": 138360 }, { "epoch": 2.2640922850364067, "grad_norm": 0.06215871870517731, "learning_rate": 4.720490434362421e-06, "loss": 0.0013, "step": 138370 }, { "epoch": 2.2642559109874827, "grad_norm": 0.0016415819991379976, "learning_rate": 4.7197776007501605e-06, "loss": 0.0013, "step": 138380 }, { "epoch": 2.2644195369385587, "grad_norm": 0.11663074791431427, "learning_rate": 4.7190647728513985e-06, "loss": 0.0008, "step": 138390 }, { "epoch": 2.2645831628896342, "grad_norm": 0.03299747779965401, "learning_rate": 4.7183519506806655e-06, "loss": 0.001, "step": 138400 }, { "epoch": 2.2647467888407102, "grad_norm": 0.025936752557754517, "learning_rate": 4.717639134252498e-06, "loss": 0.0009, "step": 138410 }, { "epoch": 2.264910414791786, "grad_norm": 0.03992857411503792, "learning_rate": 4.7169263235814275e-06, "loss": 0.001, "step": 138420 }, { "epoch": 2.265074040742862, "grad_norm": 0.07336580008268356, "learning_rate": 4.716213518681989e-06, "loss": 0.0012, "step": 138430 }, { "epoch": 2.265237666693938, "grad_norm": 0.026013411581516266, "learning_rate": 4.715500719568715e-06, "loss": 0.0009, "step": 138440 }, { "epoch": 2.2654012926450133, "grad_norm": 0.006119230762124062, "learning_rate": 4.71478792625614e-06, "loss": 0.001, "step": 138450 }, { "epoch": 2.2655649185960893, "grad_norm": 0.019350843504071236, "learning_rate": 4.7140751387587955e-06, "loss": 0.0008, "step": 138460 }, { "epoch": 2.2657285445471653, "grad_norm": 0.01690942980349064, "learning_rate": 4.713362357091217e-06, "loss": 0.0003, "step": 138470 }, { "epoch": 2.265892170498241, "grad_norm": 0.004700907506048679, "learning_rate": 4.712649581267935e-06, "loss": 0.0011, "step": 138480 }, { "epoch": 2.266055796449317, "grad_norm": 0.004650948103517294, "learning_rate": 4.7119368113034855e-06, "loss": 0.0011, "step": 138490 }, { "epoch": 2.266219422400393, "grad_norm": 0.07177475839853287, "learning_rate": 4.711224047212397e-06, "loss": 0.001, "step": 138500 }, { "epoch": 2.2663830483514684, "grad_norm": 0.015455415472388268, "learning_rate": 4.710511289009207e-06, "loss": 0.0009, "step": 138510 }, { "epoch": 2.2665466743025444, "grad_norm": 0.18353445827960968, "learning_rate": 4.709798536708444e-06, "loss": 0.0016, "step": 138520 }, { "epoch": 2.2667103002536204, "grad_norm": 0.025595523416996002, "learning_rate": 4.709085790324643e-06, "loss": 0.0009, "step": 138530 }, { "epoch": 2.266873926204696, "grad_norm": 0.05224046856164932, "learning_rate": 4.708373049872334e-06, "loss": 0.0013, "step": 138540 }, { "epoch": 2.267037552155772, "grad_norm": 0.014681038446724415, "learning_rate": 4.707660315366052e-06, "loss": 0.0013, "step": 138550 }, { "epoch": 2.2672011781068475, "grad_norm": 0.04625434800982475, "learning_rate": 4.706947586820327e-06, "loss": 0.0007, "step": 138560 }, { "epoch": 2.2673648040579235, "grad_norm": 0.02378273941576481, "learning_rate": 4.706234864249691e-06, "loss": 0.0006, "step": 138570 }, { "epoch": 2.2675284300089995, "grad_norm": 0.14270015060901642, "learning_rate": 4.705522147668677e-06, "loss": 0.0017, "step": 138580 }, { "epoch": 2.267692055960075, "grad_norm": 0.047076255083084106, "learning_rate": 4.7048094370918155e-06, "loss": 0.0008, "step": 138590 }, { "epoch": 2.267855681911151, "grad_norm": 0.031535252928733826, "learning_rate": 4.704096732533638e-06, "loss": 0.0007, "step": 138600 }, { "epoch": 2.268019307862227, "grad_norm": 0.0007313237874768674, "learning_rate": 4.703384034008679e-06, "loss": 0.0007, "step": 138610 }, { "epoch": 2.2681829338133026, "grad_norm": 0.0059571536257863045, "learning_rate": 4.702671341531464e-06, "loss": 0.0008, "step": 138620 }, { "epoch": 2.2683465597643786, "grad_norm": 0.0370248481631279, "learning_rate": 4.701958655116529e-06, "loss": 0.0008, "step": 138630 }, { "epoch": 2.2685101857154546, "grad_norm": 0.0134641844779253, "learning_rate": 4.701245974778403e-06, "loss": 0.0012, "step": 138640 }, { "epoch": 2.26867381166653, "grad_norm": 0.06713787466287613, "learning_rate": 4.700533300531619e-06, "loss": 0.0009, "step": 138650 }, { "epoch": 2.268837437617606, "grad_norm": 0.052910976111888885, "learning_rate": 4.699820632390705e-06, "loss": 0.0011, "step": 138660 }, { "epoch": 2.269001063568682, "grad_norm": 0.10418592393398285, "learning_rate": 4.699107970370193e-06, "loss": 0.0014, "step": 138670 }, { "epoch": 2.2691646895197577, "grad_norm": 0.015591630712151527, "learning_rate": 4.698395314484613e-06, "loss": 0.0008, "step": 138680 }, { "epoch": 2.2693283154708337, "grad_norm": 0.0424230583012104, "learning_rate": 4.697682664748498e-06, "loss": 0.0016, "step": 138690 }, { "epoch": 2.2694919414219097, "grad_norm": 0.0411476232111454, "learning_rate": 4.696970021176375e-06, "loss": 0.0018, "step": 138700 }, { "epoch": 2.2696555673729852, "grad_norm": 0.05236544832587242, "learning_rate": 4.6962573837827765e-06, "loss": 0.0006, "step": 138710 }, { "epoch": 2.2698191933240612, "grad_norm": 0.04772105813026428, "learning_rate": 4.69554475258223e-06, "loss": 0.0006, "step": 138720 }, { "epoch": 2.2699828192751372, "grad_norm": 0.017209630459547043, "learning_rate": 4.694832127589269e-06, "loss": 0.0011, "step": 138730 }, { "epoch": 2.270146445226213, "grad_norm": 0.02082819677889347, "learning_rate": 4.694119508818419e-06, "loss": 0.0008, "step": 138740 }, { "epoch": 2.270310071177289, "grad_norm": 0.013384549878537655, "learning_rate": 4.693406896284214e-06, "loss": 0.0006, "step": 138750 }, { "epoch": 2.2704736971283648, "grad_norm": 0.02236255072057247, "learning_rate": 4.692694290001181e-06, "loss": 0.0006, "step": 138760 }, { "epoch": 2.2706373230794403, "grad_norm": 0.19959032535552979, "learning_rate": 4.691981689983851e-06, "loss": 0.001, "step": 138770 }, { "epoch": 2.2708009490305163, "grad_norm": 0.0862797275185585, "learning_rate": 4.69126909624675e-06, "loss": 0.0007, "step": 138780 }, { "epoch": 2.2709645749815923, "grad_norm": 0.024399273097515106, "learning_rate": 4.690556508804412e-06, "loss": 0.0014, "step": 138790 }, { "epoch": 2.271128200932668, "grad_norm": 0.051316145807504654, "learning_rate": 4.689843927671362e-06, "loss": 0.0008, "step": 138800 }, { "epoch": 2.271291826883744, "grad_norm": 0.02405732311308384, "learning_rate": 4.689131352862133e-06, "loss": 0.0012, "step": 138810 }, { "epoch": 2.2714554528348194, "grad_norm": 0.014337684027850628, "learning_rate": 4.688418784391247e-06, "loss": 0.0008, "step": 138820 }, { "epoch": 2.2716190787858954, "grad_norm": 0.012004420161247253, "learning_rate": 4.687706222273242e-06, "loss": 0.0005, "step": 138830 }, { "epoch": 2.2717827047369714, "grad_norm": 0.009727055206894875, "learning_rate": 4.686993666522637e-06, "loss": 0.001, "step": 138840 }, { "epoch": 2.271946330688047, "grad_norm": 0.03106122836470604, "learning_rate": 4.686281117153969e-06, "loss": 0.0006, "step": 138850 }, { "epoch": 2.272109956639123, "grad_norm": 0.016351159662008286, "learning_rate": 4.685568574181758e-06, "loss": 0.0012, "step": 138860 }, { "epoch": 2.272273582590199, "grad_norm": 0.060370031744241714, "learning_rate": 4.684856037620541e-06, "loss": 0.0012, "step": 138870 }, { "epoch": 2.2724372085412745, "grad_norm": 0.03712896257638931, "learning_rate": 4.6841435074848376e-06, "loss": 0.0013, "step": 138880 }, { "epoch": 2.2726008344923505, "grad_norm": 0.10410293191671371, "learning_rate": 4.6834309837891825e-06, "loss": 0.001, "step": 138890 }, { "epoch": 2.2727644604434265, "grad_norm": 0.005907077342271805, "learning_rate": 4.682718466548096e-06, "loss": 0.0013, "step": 138900 }, { "epoch": 2.272928086394502, "grad_norm": 0.07523870468139648, "learning_rate": 4.682005955776115e-06, "loss": 0.0013, "step": 138910 }, { "epoch": 2.273091712345578, "grad_norm": 0.020626142621040344, "learning_rate": 4.6812934514877585e-06, "loss": 0.0006, "step": 138920 }, { "epoch": 2.2732553382966536, "grad_norm": 0.1182180643081665, "learning_rate": 4.680580953697561e-06, "loss": 0.0018, "step": 138930 }, { "epoch": 2.2734189642477296, "grad_norm": 0.025267573073506355, "learning_rate": 4.679868462420042e-06, "loss": 0.0012, "step": 138940 }, { "epoch": 2.2735825901988056, "grad_norm": 0.024696551263332367, "learning_rate": 4.679155977669736e-06, "loss": 0.0009, "step": 138950 }, { "epoch": 2.273746216149881, "grad_norm": 0.06291269510984421, "learning_rate": 4.678443499461164e-06, "loss": 0.001, "step": 138960 }, { "epoch": 2.273909842100957, "grad_norm": 0.02023273892700672, "learning_rate": 4.67773102780886e-06, "loss": 0.0009, "step": 138970 }, { "epoch": 2.274073468052033, "grad_norm": 0.020529164001345634, "learning_rate": 4.677018562727341e-06, "loss": 0.0006, "step": 138980 }, { "epoch": 2.2742370940031087, "grad_norm": 0.08771273493766785, "learning_rate": 4.676306104231143e-06, "loss": 0.0014, "step": 138990 }, { "epoch": 2.2744007199541847, "grad_norm": 0.04421643912792206, "learning_rate": 4.675593652334786e-06, "loss": 0.0012, "step": 139000 }, { "epoch": 2.2745643459052607, "grad_norm": 0.15575315058231354, "learning_rate": 4.6748812070528e-06, "loss": 0.0016, "step": 139010 }, { "epoch": 2.2747279718563362, "grad_norm": 0.016370194032788277, "learning_rate": 4.674168768399708e-06, "loss": 0.0009, "step": 139020 }, { "epoch": 2.2748915978074122, "grad_norm": 0.036673322319984436, "learning_rate": 4.6734563363900394e-06, "loss": 0.0006, "step": 139030 }, { "epoch": 2.2750552237584882, "grad_norm": 0.0191126000136137, "learning_rate": 4.672743911038316e-06, "loss": 0.0015, "step": 139040 }, { "epoch": 2.275218849709564, "grad_norm": 0.023447541519999504, "learning_rate": 4.672031492359069e-06, "loss": 0.002, "step": 139050 }, { "epoch": 2.27538247566064, "grad_norm": 0.005173043813556433, "learning_rate": 4.671319080366819e-06, "loss": 0.0006, "step": 139060 }, { "epoch": 2.2755461016117158, "grad_norm": 0.04514610394835472, "learning_rate": 4.670606675076094e-06, "loss": 0.001, "step": 139070 }, { "epoch": 2.2757097275627913, "grad_norm": 0.0585734099149704, "learning_rate": 4.669894276501418e-06, "loss": 0.0006, "step": 139080 }, { "epoch": 2.2758733535138673, "grad_norm": 0.02579583041369915, "learning_rate": 4.669181884657318e-06, "loss": 0.001, "step": 139090 }, { "epoch": 2.2760369794649433, "grad_norm": 0.05253830924630165, "learning_rate": 4.6684694995583165e-06, "loss": 0.0012, "step": 139100 }, { "epoch": 2.276200605416019, "grad_norm": 0.2178698033094406, "learning_rate": 4.667757121218942e-06, "loss": 0.0013, "step": 139110 }, { "epoch": 2.276364231367095, "grad_norm": 0.02018270269036293, "learning_rate": 4.6670447496537154e-06, "loss": 0.0012, "step": 139120 }, { "epoch": 2.276527857318171, "grad_norm": 0.020253902301192284, "learning_rate": 4.666332384877165e-06, "loss": 0.0013, "step": 139130 }, { "epoch": 2.2766914832692464, "grad_norm": 0.02138684317469597, "learning_rate": 4.6656200269038115e-06, "loss": 0.0005, "step": 139140 }, { "epoch": 2.2768551092203224, "grad_norm": 0.023861102759838104, "learning_rate": 4.6649076757481825e-06, "loss": 0.001, "step": 139150 }, { "epoch": 2.2770187351713984, "grad_norm": 0.03450924530625343, "learning_rate": 4.664195331424801e-06, "loss": 0.0013, "step": 139160 }, { "epoch": 2.277182361122474, "grad_norm": 0.02037501335144043, "learning_rate": 4.663482993948191e-06, "loss": 0.0006, "step": 139170 }, { "epoch": 2.27734598707355, "grad_norm": 0.004860075190663338, "learning_rate": 4.662770663332876e-06, "loss": 0.0008, "step": 139180 }, { "epoch": 2.2775096130246255, "grad_norm": 0.04532352089881897, "learning_rate": 4.662058339593381e-06, "loss": 0.0011, "step": 139190 }, { "epoch": 2.2776732389757015, "grad_norm": 0.03971247375011444, "learning_rate": 4.661346022744229e-06, "loss": 0.0012, "step": 139200 }, { "epoch": 2.2778368649267775, "grad_norm": 0.04618031159043312, "learning_rate": 4.660633712799943e-06, "loss": 0.0008, "step": 139210 }, { "epoch": 2.278000490877853, "grad_norm": 0.11007510870695114, "learning_rate": 4.659921409775047e-06, "loss": 0.001, "step": 139220 }, { "epoch": 2.278164116828929, "grad_norm": 0.10257569700479507, "learning_rate": 4.659209113684065e-06, "loss": 0.001, "step": 139230 }, { "epoch": 2.278327742780005, "grad_norm": 0.021219896152615547, "learning_rate": 4.658496824541518e-06, "loss": 0.0005, "step": 139240 }, { "epoch": 2.2784913687310806, "grad_norm": 0.018379367887973785, "learning_rate": 4.657784542361933e-06, "loss": 0.0004, "step": 139250 }, { "epoch": 2.2786549946821566, "grad_norm": 0.02604534476995468, "learning_rate": 4.657072267159828e-06, "loss": 0.0015, "step": 139260 }, { "epoch": 2.2788186206332326, "grad_norm": 0.044268544763326645, "learning_rate": 4.656359998949728e-06, "loss": 0.0007, "step": 139270 }, { "epoch": 2.278982246584308, "grad_norm": 0.01739954575896263, "learning_rate": 4.655647737746155e-06, "loss": 0.0007, "step": 139280 }, { "epoch": 2.279145872535384, "grad_norm": 0.018853243440389633, "learning_rate": 4.654935483563633e-06, "loss": 0.0007, "step": 139290 }, { "epoch": 2.27930949848646, "grad_norm": 0.026260454207658768, "learning_rate": 4.654223236416682e-06, "loss": 0.002, "step": 139300 }, { "epoch": 2.2794731244375357, "grad_norm": 0.04242595285177231, "learning_rate": 4.653510996319826e-06, "loss": 0.0005, "step": 139310 }, { "epoch": 2.2796367503886117, "grad_norm": 0.03763701021671295, "learning_rate": 4.652798763287585e-06, "loss": 0.0006, "step": 139320 }, { "epoch": 2.2798003763396872, "grad_norm": 0.025065883994102478, "learning_rate": 4.652086537334483e-06, "loss": 0.0009, "step": 139330 }, { "epoch": 2.2799640022907632, "grad_norm": 0.12787838280200958, "learning_rate": 4.65137431847504e-06, "loss": 0.0006, "step": 139340 }, { "epoch": 2.2801276282418392, "grad_norm": 0.0074137188494205475, "learning_rate": 4.650662106723778e-06, "loss": 0.0006, "step": 139350 }, { "epoch": 2.280291254192915, "grad_norm": 0.0787399560213089, "learning_rate": 4.6499499020952185e-06, "loss": 0.001, "step": 139360 }, { "epoch": 2.280454880143991, "grad_norm": 0.05493280664086342, "learning_rate": 4.6492377046038835e-06, "loss": 0.0009, "step": 139370 }, { "epoch": 2.2806185060950668, "grad_norm": 0.014804708771407604, "learning_rate": 4.648525514264293e-06, "loss": 0.0015, "step": 139380 }, { "epoch": 2.2807821320461423, "grad_norm": 0.00643426226451993, "learning_rate": 4.647813331090969e-06, "loss": 0.0004, "step": 139390 }, { "epoch": 2.2809457579972183, "grad_norm": 0.019825659692287445, "learning_rate": 4.64710115509843e-06, "loss": 0.0007, "step": 139400 }, { "epoch": 2.2811093839482943, "grad_norm": 0.02015482261776924, "learning_rate": 4.6463889863012004e-06, "loss": 0.0021, "step": 139410 }, { "epoch": 2.28127300989937, "grad_norm": 0.006451125256717205, "learning_rate": 4.645676824713797e-06, "loss": 0.0007, "step": 139420 }, { "epoch": 2.281436635850446, "grad_norm": 0.021886056289076805, "learning_rate": 4.644964670350743e-06, "loss": 0.0015, "step": 139430 }, { "epoch": 2.281600261801522, "grad_norm": 0.07532492280006409, "learning_rate": 4.644252523226556e-06, "loss": 0.0015, "step": 139440 }, { "epoch": 2.2817638877525974, "grad_norm": 0.01593787968158722, "learning_rate": 4.64354038335576e-06, "loss": 0.0006, "step": 139450 }, { "epoch": 2.2819275137036734, "grad_norm": 0.01230832003057003, "learning_rate": 4.64282825075287e-06, "loss": 0.0004, "step": 139460 }, { "epoch": 2.2820911396547494, "grad_norm": 0.016451481729745865, "learning_rate": 4.64211612543241e-06, "loss": 0.0013, "step": 139470 }, { "epoch": 2.282254765605825, "grad_norm": 0.04061181843280792, "learning_rate": 4.641404007408895e-06, "loss": 0.0009, "step": 139480 }, { "epoch": 2.282418391556901, "grad_norm": 0.012340959161520004, "learning_rate": 4.64069189669685e-06, "loss": 0.0009, "step": 139490 }, { "epoch": 2.282582017507977, "grad_norm": 0.06011228635907173, "learning_rate": 4.63997979331079e-06, "loss": 0.0006, "step": 139500 }, { "epoch": 2.2827456434590525, "grad_norm": 0.046247318387031555, "learning_rate": 4.639267697265238e-06, "loss": 0.0009, "step": 139510 }, { "epoch": 2.2829092694101285, "grad_norm": 0.10948976129293442, "learning_rate": 4.638555608574708e-06, "loss": 0.0015, "step": 139520 }, { "epoch": 2.2830728953612045, "grad_norm": 0.1617412567138672, "learning_rate": 4.637843527253724e-06, "loss": 0.002, "step": 139530 }, { "epoch": 2.28323652131228, "grad_norm": 0.0660766065120697, "learning_rate": 4.6371314533168e-06, "loss": 0.0015, "step": 139540 }, { "epoch": 2.283400147263356, "grad_norm": 0.028656402602791786, "learning_rate": 4.636419386778461e-06, "loss": 0.001, "step": 139550 }, { "epoch": 2.283563773214432, "grad_norm": 0.03687477111816406, "learning_rate": 4.635707327653218e-06, "loss": 0.0004, "step": 139560 }, { "epoch": 2.2837273991655076, "grad_norm": 0.04335309565067291, "learning_rate": 4.634995275955595e-06, "loss": 0.0004, "step": 139570 }, { "epoch": 2.2838910251165836, "grad_norm": 0.03399628400802612, "learning_rate": 4.6342832317001054e-06, "loss": 0.0007, "step": 139580 }, { "epoch": 2.284054651067659, "grad_norm": 0.03551675006747246, "learning_rate": 4.633571194901273e-06, "loss": 0.0009, "step": 139590 }, { "epoch": 2.284218277018735, "grad_norm": 0.04747453331947327, "learning_rate": 4.632859165573609e-06, "loss": 0.0007, "step": 139600 }, { "epoch": 2.284381902969811, "grad_norm": 0.013506532646715641, "learning_rate": 4.632147143731637e-06, "loss": 0.0011, "step": 139610 }, { "epoch": 2.2845455289208867, "grad_norm": 0.003775828517973423, "learning_rate": 4.631435129389869e-06, "loss": 0.0024, "step": 139620 }, { "epoch": 2.2847091548719627, "grad_norm": 0.01596767269074917, "learning_rate": 4.630723122562828e-06, "loss": 0.0004, "step": 139630 }, { "epoch": 2.2848727808230387, "grad_norm": 0.05570659786462784, "learning_rate": 4.630011123265028e-06, "loss": 0.0012, "step": 139640 }, { "epoch": 2.2850364067741142, "grad_norm": 0.006974375806748867, "learning_rate": 4.6292991315109855e-06, "loss": 0.0006, "step": 139650 }, { "epoch": 2.2852000327251902, "grad_norm": 0.08238665759563446, "learning_rate": 4.628587147315219e-06, "loss": 0.0013, "step": 139660 }, { "epoch": 2.2853636586762662, "grad_norm": 0.048042334616184235, "learning_rate": 4.627875170692245e-06, "loss": 0.0008, "step": 139670 }, { "epoch": 2.285527284627342, "grad_norm": 0.024370703846216202, "learning_rate": 4.627163201656579e-06, "loss": 0.0006, "step": 139680 }, { "epoch": 2.285690910578418, "grad_norm": 0.05329496040940285, "learning_rate": 4.626451240222739e-06, "loss": 0.0009, "step": 139690 }, { "epoch": 2.2858545365294933, "grad_norm": 0.049790266901254654, "learning_rate": 4.62573928640524e-06, "loss": 0.0011, "step": 139700 }, { "epoch": 2.2860181624805693, "grad_norm": 0.02745143696665764, "learning_rate": 4.625027340218599e-06, "loss": 0.0016, "step": 139710 }, { "epoch": 2.2861817884316453, "grad_norm": 0.02838124893605709, "learning_rate": 4.624315401677331e-06, "loss": 0.0009, "step": 139720 }, { "epoch": 2.286345414382721, "grad_norm": 0.05981391668319702, "learning_rate": 4.623603470795953e-06, "loss": 0.0012, "step": 139730 }, { "epoch": 2.286509040333797, "grad_norm": 0.09159982949495316, "learning_rate": 4.6228915475889795e-06, "loss": 0.0008, "step": 139740 }, { "epoch": 2.286672666284873, "grad_norm": 0.019049888476729393, "learning_rate": 4.622179632070928e-06, "loss": 0.0008, "step": 139750 }, { "epoch": 2.2868362922359484, "grad_norm": 0.0004165704594925046, "learning_rate": 4.621467724256311e-06, "loss": 0.0014, "step": 139760 }, { "epoch": 2.2869999181870244, "grad_norm": 0.08729510009288788, "learning_rate": 4.620755824159646e-06, "loss": 0.0007, "step": 139770 }, { "epoch": 2.2871635441381004, "grad_norm": 0.04623589292168617, "learning_rate": 4.620043931795446e-06, "loss": 0.0006, "step": 139780 }, { "epoch": 2.287327170089176, "grad_norm": 0.04892631992697716, "learning_rate": 4.619332047178229e-06, "loss": 0.0012, "step": 139790 }, { "epoch": 2.287490796040252, "grad_norm": 0.06980438530445099, "learning_rate": 4.618620170322506e-06, "loss": 0.0008, "step": 139800 }, { "epoch": 2.287654421991328, "grad_norm": 0.035479605197906494, "learning_rate": 4.617908301242796e-06, "loss": 0.0006, "step": 139810 }, { "epoch": 2.2878180479424035, "grad_norm": 0.03580670803785324, "learning_rate": 4.617196439953608e-06, "loss": 0.0002, "step": 139820 }, { "epoch": 2.2879816738934795, "grad_norm": 0.05095307528972626, "learning_rate": 4.616484586469461e-06, "loss": 0.0012, "step": 139830 }, { "epoch": 2.2881452998445555, "grad_norm": 0.031154435127973557, "learning_rate": 4.615772740804866e-06, "loss": 0.0004, "step": 139840 }, { "epoch": 2.288308925795631, "grad_norm": 0.23434235155582428, "learning_rate": 4.615060902974338e-06, "loss": 0.0029, "step": 139850 }, { "epoch": 2.288472551746707, "grad_norm": 0.03756482154130936, "learning_rate": 4.61434907299239e-06, "loss": 0.0005, "step": 139860 }, { "epoch": 2.288636177697783, "grad_norm": 0.054588235914707184, "learning_rate": 4.613637250873539e-06, "loss": 0.0005, "step": 139870 }, { "epoch": 2.2887998036488586, "grad_norm": 0.04006483778357506, "learning_rate": 4.612925436632293e-06, "loss": 0.0017, "step": 139880 }, { "epoch": 2.2889634295999346, "grad_norm": 0.05009271577000618, "learning_rate": 4.61221363028317e-06, "loss": 0.0006, "step": 139890 }, { "epoch": 2.2891270555510106, "grad_norm": 0.015216062776744366, "learning_rate": 4.61150183184068e-06, "loss": 0.0002, "step": 139900 }, { "epoch": 2.289290681502086, "grad_norm": 0.33762043714523315, "learning_rate": 4.610790041319339e-06, "loss": 0.0014, "step": 139910 }, { "epoch": 2.289454307453162, "grad_norm": 0.04152775555849075, "learning_rate": 4.6100782587336566e-06, "loss": 0.0005, "step": 139920 }, { "epoch": 2.289617933404238, "grad_norm": 0.03435197472572327, "learning_rate": 4.609366484098148e-06, "loss": 0.0006, "step": 139930 }, { "epoch": 2.2897815593553137, "grad_norm": 0.03857884183526039, "learning_rate": 4.608654717427323e-06, "loss": 0.0008, "step": 139940 }, { "epoch": 2.2899451853063897, "grad_norm": 0.05317791923880577, "learning_rate": 4.607942958735697e-06, "loss": 0.0008, "step": 139950 }, { "epoch": 2.2901088112574657, "grad_norm": 0.014749370515346527, "learning_rate": 4.607231208037779e-06, "loss": 0.001, "step": 139960 }, { "epoch": 2.2902724372085412, "grad_norm": 0.09578577429056168, "learning_rate": 4.6065194653480844e-06, "loss": 0.001, "step": 139970 }, { "epoch": 2.2904360631596172, "grad_norm": 0.054131414741277695, "learning_rate": 4.605807730681122e-06, "loss": 0.0009, "step": 139980 }, { "epoch": 2.290599689110693, "grad_norm": 0.08106668293476105, "learning_rate": 4.605096004051406e-06, "loss": 0.0012, "step": 139990 }, { "epoch": 2.290763315061769, "grad_norm": 0.050516046583652496, "learning_rate": 4.604384285473445e-06, "loss": 0.0008, "step": 140000 }, { "epoch": 2.2909269410128448, "grad_norm": 0.07138187438249588, "learning_rate": 4.603672574961754e-06, "loss": 0.0012, "step": 140010 }, { "epoch": 2.2910905669639203, "grad_norm": 0.0035009533166885376, "learning_rate": 4.602960872530839e-06, "loss": 0.0013, "step": 140020 }, { "epoch": 2.2912541929149963, "grad_norm": 0.0735802873969078, "learning_rate": 4.6022491781952174e-06, "loss": 0.0006, "step": 140030 }, { "epoch": 2.2914178188660723, "grad_norm": 0.09915518760681152, "learning_rate": 4.601537491969394e-06, "loss": 0.0011, "step": 140040 }, { "epoch": 2.291581444817148, "grad_norm": 0.06525250524282455, "learning_rate": 4.600825813867885e-06, "loss": 0.001, "step": 140050 }, { "epoch": 2.291745070768224, "grad_norm": 0.11066386103630066, "learning_rate": 4.600114143905196e-06, "loss": 0.0008, "step": 140060 }, { "epoch": 2.2919086967193, "grad_norm": 0.23323749005794525, "learning_rate": 4.599402482095841e-06, "loss": 0.0012, "step": 140070 }, { "epoch": 2.2920723226703754, "grad_norm": 0.0178990438580513, "learning_rate": 4.598690828454327e-06, "loss": 0.0006, "step": 140080 }, { "epoch": 2.2922359486214514, "grad_norm": 0.01675679348409176, "learning_rate": 4.597979182995168e-06, "loss": 0.0007, "step": 140090 }, { "epoch": 2.292399574572527, "grad_norm": 0.05855638161301613, "learning_rate": 4.597267545732869e-06, "loss": 0.001, "step": 140100 }, { "epoch": 2.292563200523603, "grad_norm": 0.023595912382006645, "learning_rate": 4.596555916681946e-06, "loss": 0.0008, "step": 140110 }, { "epoch": 2.292726826474679, "grad_norm": 0.0330306775867939, "learning_rate": 4.5958442958569e-06, "loss": 0.0004, "step": 140120 }, { "epoch": 2.2928904524257545, "grad_norm": 0.012972629629075527, "learning_rate": 4.5951326832722494e-06, "loss": 0.0013, "step": 140130 }, { "epoch": 2.2930540783768305, "grad_norm": 0.10067613422870636, "learning_rate": 4.594421078942496e-06, "loss": 0.0009, "step": 140140 }, { "epoch": 2.2932177043279065, "grad_norm": 0.0010858355090022087, "learning_rate": 4.593709482882154e-06, "loss": 0.0008, "step": 140150 }, { "epoch": 2.293381330278982, "grad_norm": 0.017670612782239914, "learning_rate": 4.592997895105728e-06, "loss": 0.0008, "step": 140160 }, { "epoch": 2.293544956230058, "grad_norm": 0.08135300129652023, "learning_rate": 4.5922863156277315e-06, "loss": 0.0007, "step": 140170 }, { "epoch": 2.293708582181134, "grad_norm": 0.11437839269638062, "learning_rate": 4.591574744462666e-06, "loss": 0.001, "step": 140180 }, { "epoch": 2.2938722081322096, "grad_norm": 0.01779148168861866, "learning_rate": 4.590863181625049e-06, "loss": 0.0009, "step": 140190 }, { "epoch": 2.2940358340832856, "grad_norm": 0.22207124531269073, "learning_rate": 4.59015162712938e-06, "loss": 0.0007, "step": 140200 }, { "epoch": 2.2941994600343616, "grad_norm": 0.040927935391664505, "learning_rate": 4.589440080990174e-06, "loss": 0.0012, "step": 140210 }, { "epoch": 2.294363085985437, "grad_norm": 0.028723804280161858, "learning_rate": 4.588728543221932e-06, "loss": 0.0011, "step": 140220 }, { "epoch": 2.294526711936513, "grad_norm": 0.11653237789869308, "learning_rate": 4.588017013839169e-06, "loss": 0.0006, "step": 140230 }, { "epoch": 2.294690337887589, "grad_norm": 0.08459381759166718, "learning_rate": 4.587305492856385e-06, "loss": 0.0005, "step": 140240 }, { "epoch": 2.2948539638386647, "grad_norm": 0.007862108759582043, "learning_rate": 4.5865939802880946e-06, "loss": 0.001, "step": 140250 }, { "epoch": 2.2950175897897407, "grad_norm": 0.044066864997148514, "learning_rate": 4.585882476148797e-06, "loss": 0.0008, "step": 140260 }, { "epoch": 2.2951812157408167, "grad_norm": 0.05935574322938919, "learning_rate": 4.585170980453008e-06, "loss": 0.0008, "step": 140270 }, { "epoch": 2.2953448416918922, "grad_norm": 0.017561424523591995, "learning_rate": 4.584459493215228e-06, "loss": 0.0009, "step": 140280 }, { "epoch": 2.2955084676429682, "grad_norm": 0.04551788419485092, "learning_rate": 4.583748014449965e-06, "loss": 0.0012, "step": 140290 }, { "epoch": 2.2956720935940442, "grad_norm": 0.07358012348413467, "learning_rate": 4.583036544171726e-06, "loss": 0.0006, "step": 140300 }, { "epoch": 2.29583571954512, "grad_norm": 0.0241440050303936, "learning_rate": 4.582325082395018e-06, "loss": 0.0013, "step": 140310 }, { "epoch": 2.2959993454961958, "grad_norm": 0.000994542264379561, "learning_rate": 4.581613629134346e-06, "loss": 0.0004, "step": 140320 }, { "epoch": 2.2961629714472718, "grad_norm": 0.11682666838169098, "learning_rate": 4.580902184404217e-06, "loss": 0.0014, "step": 140330 }, { "epoch": 2.2963265973983473, "grad_norm": 0.00879219826310873, "learning_rate": 4.580190748219135e-06, "loss": 0.0009, "step": 140340 }, { "epoch": 2.2964902233494233, "grad_norm": 0.06178050860762596, "learning_rate": 4.579479320593607e-06, "loss": 0.0014, "step": 140350 }, { "epoch": 2.296653849300499, "grad_norm": 0.04155758023262024, "learning_rate": 4.578767901542138e-06, "loss": 0.0011, "step": 140360 }, { "epoch": 2.296817475251575, "grad_norm": 0.017229771241545677, "learning_rate": 4.578056491079233e-06, "loss": 0.0012, "step": 140370 }, { "epoch": 2.296981101202651, "grad_norm": 0.05513696372509003, "learning_rate": 4.577345089219397e-06, "loss": 0.0012, "step": 140380 }, { "epoch": 2.2971447271537264, "grad_norm": 0.18793535232543945, "learning_rate": 4.576633695977137e-06, "loss": 0.0012, "step": 140390 }, { "epoch": 2.2973083531048024, "grad_norm": 0.02031848020851612, "learning_rate": 4.575922311366954e-06, "loss": 0.0008, "step": 140400 }, { "epoch": 2.2974719790558784, "grad_norm": 0.03720786049962044, "learning_rate": 4.575210935403356e-06, "loss": 0.0009, "step": 140410 }, { "epoch": 2.297635605006954, "grad_norm": 0.03744739294052124, "learning_rate": 4.574499568100843e-06, "loss": 0.0009, "step": 140420 }, { "epoch": 2.29779923095803, "grad_norm": 0.031923726201057434, "learning_rate": 4.573788209473924e-06, "loss": 0.0007, "step": 140430 }, { "epoch": 2.297962856909106, "grad_norm": 0.09398449957370758, "learning_rate": 4.5730768595371005e-06, "loss": 0.0013, "step": 140440 }, { "epoch": 2.2981264828601815, "grad_norm": 0.01785324141383171, "learning_rate": 4.572365518304878e-06, "loss": 0.0006, "step": 140450 }, { "epoch": 2.2982901088112575, "grad_norm": 0.02174960821866989, "learning_rate": 4.571654185791757e-06, "loss": 0.0008, "step": 140460 }, { "epoch": 2.2984537347623335, "grad_norm": 0.016939766705036163, "learning_rate": 4.5709428620122434e-06, "loss": 0.0006, "step": 140470 }, { "epoch": 2.298617360713409, "grad_norm": 0.040748387575149536, "learning_rate": 4.57023154698084e-06, "loss": 0.0006, "step": 140480 }, { "epoch": 2.298780986664485, "grad_norm": 0.061889927834272385, "learning_rate": 4.56952024071205e-06, "loss": 0.0009, "step": 140490 }, { "epoch": 2.2989446126155606, "grad_norm": 0.027774296700954437, "learning_rate": 4.568808943220376e-06, "loss": 0.0008, "step": 140500 }, { "epoch": 2.2991082385666366, "grad_norm": 0.02480895444750786, "learning_rate": 4.568097654520322e-06, "loss": 0.0007, "step": 140510 }, { "epoch": 2.2992718645177126, "grad_norm": 0.03321587294340134, "learning_rate": 4.567386374626388e-06, "loss": 0.0006, "step": 140520 }, { "epoch": 2.299435490468788, "grad_norm": 0.008128318004310131, "learning_rate": 4.5666751035530785e-06, "loss": 0.0009, "step": 140530 }, { "epoch": 2.299599116419864, "grad_norm": 0.0207186508923769, "learning_rate": 4.565963841314895e-06, "loss": 0.0012, "step": 140540 }, { "epoch": 2.29976274237094, "grad_norm": 0.079751156270504, "learning_rate": 4.56525258792634e-06, "loss": 0.0013, "step": 140550 }, { "epoch": 2.2999263683220157, "grad_norm": 0.0381140299141407, "learning_rate": 4.564541343401914e-06, "loss": 0.0008, "step": 140560 }, { "epoch": 2.3000899942730917, "grad_norm": 0.03604358062148094, "learning_rate": 4.563830107756121e-06, "loss": 0.0007, "step": 140570 }, { "epoch": 2.3002536202241677, "grad_norm": 0.1604817658662796, "learning_rate": 4.563118881003461e-06, "loss": 0.0017, "step": 140580 }, { "epoch": 2.3004172461752432, "grad_norm": 0.04248397424817085, "learning_rate": 4.562407663158434e-06, "loss": 0.001, "step": 140590 }, { "epoch": 2.3005808721263192, "grad_norm": 0.03261958062648773, "learning_rate": 4.561696454235544e-06, "loss": 0.0007, "step": 140600 }, { "epoch": 2.3007444980773952, "grad_norm": 0.11330734193325043, "learning_rate": 4.56098525424929e-06, "loss": 0.0008, "step": 140610 }, { "epoch": 2.300908124028471, "grad_norm": 0.05655575916171074, "learning_rate": 4.560274063214174e-06, "loss": 0.0012, "step": 140620 }, { "epoch": 2.3010717499795468, "grad_norm": 0.0014629375655204058, "learning_rate": 4.559562881144695e-06, "loss": 0.001, "step": 140630 }, { "epoch": 2.3012353759306228, "grad_norm": 0.0029182457365095615, "learning_rate": 4.558851708055355e-06, "loss": 0.0008, "step": 140640 }, { "epoch": 2.3013990018816983, "grad_norm": 0.01161411963403225, "learning_rate": 4.558140543960653e-06, "loss": 0.0006, "step": 140650 }, { "epoch": 2.3015626278327743, "grad_norm": 0.006135293748229742, "learning_rate": 4.557429388875089e-06, "loss": 0.0008, "step": 140660 }, { "epoch": 2.3017262537838503, "grad_norm": 0.020837537944316864, "learning_rate": 4.556718242813164e-06, "loss": 0.0011, "step": 140670 }, { "epoch": 2.301889879734926, "grad_norm": 0.007751280441880226, "learning_rate": 4.556007105789377e-06, "loss": 0.0016, "step": 140680 }, { "epoch": 2.302053505686002, "grad_norm": 0.05999525263905525, "learning_rate": 4.555295977818228e-06, "loss": 0.0011, "step": 140690 }, { "epoch": 2.302217131637078, "grad_norm": 0.022816801443696022, "learning_rate": 4.554584858914215e-06, "loss": 0.001, "step": 140700 }, { "epoch": 2.3023807575881534, "grad_norm": 0.021359503269195557, "learning_rate": 4.553873749091838e-06, "loss": 0.0013, "step": 140710 }, { "epoch": 2.3025443835392294, "grad_norm": 0.0069732083939015865, "learning_rate": 4.553162648365596e-06, "loss": 0.0004, "step": 140720 }, { "epoch": 2.3027080094903054, "grad_norm": 0.023124849423766136, "learning_rate": 4.5524515567499884e-06, "loss": 0.0011, "step": 140730 }, { "epoch": 2.302871635441381, "grad_norm": 0.06795581430196762, "learning_rate": 4.5517404742595115e-06, "loss": 0.0006, "step": 140740 }, { "epoch": 2.303035261392457, "grad_norm": 0.06392474472522736, "learning_rate": 4.551029400908667e-06, "loss": 0.0009, "step": 140750 }, { "epoch": 2.3031988873435325, "grad_norm": 0.019380131736397743, "learning_rate": 4.550318336711949e-06, "loss": 0.0007, "step": 140760 }, { "epoch": 2.3033625132946085, "grad_norm": 0.07359065860509872, "learning_rate": 4.54960728168386e-06, "loss": 0.0005, "step": 140770 }, { "epoch": 2.3035261392456845, "grad_norm": 0.04131990671157837, "learning_rate": 4.548896235838893e-06, "loss": 0.0015, "step": 140780 }, { "epoch": 2.30368976519676, "grad_norm": 0.0034846749622374773, "learning_rate": 4.548185199191551e-06, "loss": 0.0006, "step": 140790 }, { "epoch": 2.303853391147836, "grad_norm": 0.045587554574012756, "learning_rate": 4.547474171756324e-06, "loss": 0.0011, "step": 140800 }, { "epoch": 2.304017017098912, "grad_norm": 0.01113450713455677, "learning_rate": 4.5467631535477185e-06, "loss": 0.0016, "step": 140810 }, { "epoch": 2.3041806430499876, "grad_norm": 0.07310383021831512, "learning_rate": 4.546052144580224e-06, "loss": 0.001, "step": 140820 }, { "epoch": 2.3043442690010636, "grad_norm": 0.023867206647992134, "learning_rate": 4.545341144868342e-06, "loss": 0.0012, "step": 140830 }, { "epoch": 2.3045078949521396, "grad_norm": 0.011656973510980606, "learning_rate": 4.5446301544265645e-06, "loss": 0.0007, "step": 140840 }, { "epoch": 2.304671520903215, "grad_norm": 0.03636128455400467, "learning_rate": 4.543919173269395e-06, "loss": 0.001, "step": 140850 }, { "epoch": 2.304835146854291, "grad_norm": 0.09476085752248764, "learning_rate": 4.543208201411321e-06, "loss": 0.0008, "step": 140860 }, { "epoch": 2.3049987728053667, "grad_norm": 0.1157945841550827, "learning_rate": 4.542497238866848e-06, "loss": 0.0006, "step": 140870 }, { "epoch": 2.3051623987564427, "grad_norm": 0.007583112455904484, "learning_rate": 4.541786285650463e-06, "loss": 0.0003, "step": 140880 }, { "epoch": 2.3053260247075187, "grad_norm": 0.19871842861175537, "learning_rate": 4.54107534177667e-06, "loss": 0.0012, "step": 140890 }, { "epoch": 2.3054896506585942, "grad_norm": 0.0465315505862236, "learning_rate": 4.540364407259957e-06, "loss": 0.0009, "step": 140900 }, { "epoch": 2.3056532766096702, "grad_norm": 0.054133761674165726, "learning_rate": 4.539653482114826e-06, "loss": 0.0008, "step": 140910 }, { "epoch": 2.3058169025607462, "grad_norm": 0.06639733165502548, "learning_rate": 4.538942566355765e-06, "loss": 0.001, "step": 140920 }, { "epoch": 2.305980528511822, "grad_norm": 0.0031819669529795647, "learning_rate": 4.538231659997277e-06, "loss": 0.0008, "step": 140930 }, { "epoch": 2.3061441544628978, "grad_norm": 0.005360690876841545, "learning_rate": 4.53752076305385e-06, "loss": 0.0006, "step": 140940 }, { "epoch": 2.3063077804139738, "grad_norm": 0.05600108951330185, "learning_rate": 4.536809875539983e-06, "loss": 0.0022, "step": 140950 }, { "epoch": 2.3064714063650493, "grad_norm": 0.033593565225601196, "learning_rate": 4.536098997470168e-06, "loss": 0.0008, "step": 140960 }, { "epoch": 2.3066350323161253, "grad_norm": 0.04161834344267845, "learning_rate": 4.5353881288589e-06, "loss": 0.0014, "step": 140970 }, { "epoch": 2.3067986582672013, "grad_norm": 0.0101346829906106, "learning_rate": 4.534677269720672e-06, "loss": 0.0003, "step": 140980 }, { "epoch": 2.306962284218277, "grad_norm": 0.015066905878484249, "learning_rate": 4.53396642006998e-06, "loss": 0.0005, "step": 140990 }, { "epoch": 2.307125910169353, "grad_norm": 0.04244116321206093, "learning_rate": 4.533255579921315e-06, "loss": 0.001, "step": 141000 }, { "epoch": 2.307289536120429, "grad_norm": 0.04756949841976166, "learning_rate": 4.5325447492891735e-06, "loss": 0.0005, "step": 141010 }, { "epoch": 2.3074531620715044, "grad_norm": 0.15869715809822083, "learning_rate": 4.531833928188046e-06, "loss": 0.0014, "step": 141020 }, { "epoch": 2.3076167880225804, "grad_norm": 0.0786227285861969, "learning_rate": 4.531123116632427e-06, "loss": 0.0007, "step": 141030 }, { "epoch": 2.3077804139736564, "grad_norm": 0.020392606034874916, "learning_rate": 4.530412314636808e-06, "loss": 0.0007, "step": 141040 }, { "epoch": 2.307944039924732, "grad_norm": 0.04358493536710739, "learning_rate": 4.529701522215684e-06, "loss": 0.0011, "step": 141050 }, { "epoch": 2.308107665875808, "grad_norm": 0.04557153210043907, "learning_rate": 4.528990739383544e-06, "loss": 0.0008, "step": 141060 }, { "epoch": 2.308271291826884, "grad_norm": 0.09698963910341263, "learning_rate": 4.528279966154884e-06, "loss": 0.0014, "step": 141070 }, { "epoch": 2.3084349177779595, "grad_norm": 0.06903493404388428, "learning_rate": 4.527569202544193e-06, "loss": 0.0005, "step": 141080 }, { "epoch": 2.3085985437290355, "grad_norm": 0.07080480456352234, "learning_rate": 4.526858448565965e-06, "loss": 0.0006, "step": 141090 }, { "epoch": 2.3087621696801115, "grad_norm": 0.052209705114364624, "learning_rate": 4.526147704234691e-06, "loss": 0.0006, "step": 141100 }, { "epoch": 2.308925795631187, "grad_norm": 0.010546657256782055, "learning_rate": 4.525436969564862e-06, "loss": 0.0009, "step": 141110 }, { "epoch": 2.309089421582263, "grad_norm": 0.2036951780319214, "learning_rate": 4.524726244570969e-06, "loss": 0.0009, "step": 141120 }, { "epoch": 2.3092530475333386, "grad_norm": 0.10485866665840149, "learning_rate": 4.524015529267505e-06, "loss": 0.0008, "step": 141130 }, { "epoch": 2.3094166734844146, "grad_norm": 0.04085854813456535, "learning_rate": 4.5233048236689584e-06, "loss": 0.0004, "step": 141140 }, { "epoch": 2.3095802994354906, "grad_norm": 0.03144391253590584, "learning_rate": 4.522594127789822e-06, "loss": 0.0009, "step": 141150 }, { "epoch": 2.309743925386566, "grad_norm": 0.02752421610057354, "learning_rate": 4.521883441644583e-06, "loss": 0.0009, "step": 141160 }, { "epoch": 2.309907551337642, "grad_norm": 0.2026616334915161, "learning_rate": 4.521172765247737e-06, "loss": 0.0009, "step": 141170 }, { "epoch": 2.310071177288718, "grad_norm": 0.0073436349630355835, "learning_rate": 4.520462098613769e-06, "loss": 0.0006, "step": 141180 }, { "epoch": 2.3102348032397937, "grad_norm": 0.02770882658660412, "learning_rate": 4.519751441757173e-06, "loss": 0.0005, "step": 141190 }, { "epoch": 2.3103984291908697, "grad_norm": 0.02444012276828289, "learning_rate": 4.519040794692434e-06, "loss": 0.0006, "step": 141200 }, { "epoch": 2.3105620551419457, "grad_norm": 0.057564638555049896, "learning_rate": 4.518330157434046e-06, "loss": 0.0007, "step": 141210 }, { "epoch": 2.3107256810930212, "grad_norm": 0.009113338775932789, "learning_rate": 4.517619529996496e-06, "loss": 0.0005, "step": 141220 }, { "epoch": 2.3108893070440972, "grad_norm": 0.013905988074839115, "learning_rate": 4.5169089123942736e-06, "loss": 0.0007, "step": 141230 }, { "epoch": 2.3110529329951732, "grad_norm": 0.03861887753009796, "learning_rate": 4.516198304641867e-06, "loss": 0.0008, "step": 141240 }, { "epoch": 2.311216558946249, "grad_norm": 0.06793826073408127, "learning_rate": 4.515487706753767e-06, "loss": 0.0006, "step": 141250 }, { "epoch": 2.3113801848973248, "grad_norm": 0.03937405347824097, "learning_rate": 4.5147771187444595e-06, "loss": 0.0007, "step": 141260 }, { "epoch": 2.3115438108484003, "grad_norm": 0.04127608239650726, "learning_rate": 4.514066540628435e-06, "loss": 0.0013, "step": 141270 }, { "epoch": 2.3117074367994763, "grad_norm": 0.002749876817688346, "learning_rate": 4.513355972420178e-06, "loss": 0.0008, "step": 141280 }, { "epoch": 2.3118710627505523, "grad_norm": 0.03383681923151016, "learning_rate": 4.5126454141341814e-06, "loss": 0.001, "step": 141290 }, { "epoch": 2.312034688701628, "grad_norm": 0.04745142161846161, "learning_rate": 4.511934865784929e-06, "loss": 0.0014, "step": 141300 }, { "epoch": 2.312198314652704, "grad_norm": 0.010607732459902763, "learning_rate": 4.51122432738691e-06, "loss": 0.001, "step": 141310 }, { "epoch": 2.31236194060378, "grad_norm": 0.056734856218099594, "learning_rate": 4.510513798954611e-06, "loss": 0.0007, "step": 141320 }, { "epoch": 2.3125255665548554, "grad_norm": 0.005037798546254635, "learning_rate": 4.50980328050252e-06, "loss": 0.0019, "step": 141330 }, { "epoch": 2.3126891925059314, "grad_norm": 0.005884434096515179, "learning_rate": 4.5090927720451225e-06, "loss": 0.0019, "step": 141340 }, { "epoch": 2.3128528184570074, "grad_norm": 0.0024271043948829174, "learning_rate": 4.508382273596906e-06, "loss": 0.0012, "step": 141350 }, { "epoch": 2.313016444408083, "grad_norm": 0.0031976974569261074, "learning_rate": 4.5076717851723565e-06, "loss": 0.0007, "step": 141360 }, { "epoch": 2.313180070359159, "grad_norm": 0.05544734746217728, "learning_rate": 4.506961306785962e-06, "loss": 0.0009, "step": 141370 }, { "epoch": 2.313343696310235, "grad_norm": 0.12681540846824646, "learning_rate": 4.506250838452206e-06, "loss": 0.0017, "step": 141380 }, { "epoch": 2.3135073222613105, "grad_norm": 0.1358232945203781, "learning_rate": 4.505540380185577e-06, "loss": 0.0012, "step": 141390 }, { "epoch": 2.3136709482123865, "grad_norm": 0.06173717603087425, "learning_rate": 4.504829932000556e-06, "loss": 0.0009, "step": 141400 }, { "epoch": 2.3138345741634625, "grad_norm": 0.049906592816114426, "learning_rate": 4.504119493911635e-06, "loss": 0.001, "step": 141410 }, { "epoch": 2.313998200114538, "grad_norm": 0.009390922263264656, "learning_rate": 4.503409065933292e-06, "loss": 0.0013, "step": 141420 }, { "epoch": 2.314161826065614, "grad_norm": 0.006278768181800842, "learning_rate": 4.50269864808002e-06, "loss": 0.0007, "step": 141430 }, { "epoch": 2.31432545201669, "grad_norm": 0.01851160079240799, "learning_rate": 4.501988240366296e-06, "loss": 0.0007, "step": 141440 }, { "epoch": 2.3144890779677656, "grad_norm": 0.05374264717102051, "learning_rate": 4.501277842806611e-06, "loss": 0.0009, "step": 141450 }, { "epoch": 2.3146527039188416, "grad_norm": 0.027522804215550423, "learning_rate": 4.500567455415444e-06, "loss": 0.0007, "step": 141460 }, { "epoch": 2.3148163298699176, "grad_norm": 0.1782599836587906, "learning_rate": 4.4998570782072845e-06, "loss": 0.001, "step": 141470 }, { "epoch": 2.314979955820993, "grad_norm": 0.03798840194940567, "learning_rate": 4.499146711196611e-06, "loss": 0.0013, "step": 141480 }, { "epoch": 2.315143581772069, "grad_norm": 0.03445257619023323, "learning_rate": 4.4984363543979136e-06, "loss": 0.0011, "step": 141490 }, { "epoch": 2.315307207723145, "grad_norm": 0.012761056423187256, "learning_rate": 4.497726007825669e-06, "loss": 0.0006, "step": 141500 }, { "epoch": 2.3154708336742207, "grad_norm": 0.023880131542682648, "learning_rate": 4.497015671494367e-06, "loss": 0.0006, "step": 141510 }, { "epoch": 2.3156344596252967, "grad_norm": 0.02536904066801071, "learning_rate": 4.496305345418485e-06, "loss": 0.0014, "step": 141520 }, { "epoch": 2.3157980855763722, "grad_norm": 0.028741218149662018, "learning_rate": 4.495595029612511e-06, "loss": 0.0012, "step": 141530 }, { "epoch": 2.3159617115274482, "grad_norm": 0.10739660263061523, "learning_rate": 4.494884724090922e-06, "loss": 0.0007, "step": 141540 }, { "epoch": 2.3161253374785242, "grad_norm": 0.012289648875594139, "learning_rate": 4.494174428868208e-06, "loss": 0.0008, "step": 141550 }, { "epoch": 2.3162889634296, "grad_norm": 0.10328267514705658, "learning_rate": 4.493464143958843e-06, "loss": 0.0005, "step": 141560 }, { "epoch": 2.3164525893806758, "grad_norm": 0.022748008370399475, "learning_rate": 4.4927538693773174e-06, "loss": 0.0009, "step": 141570 }, { "epoch": 2.3166162153317518, "grad_norm": 0.030940599739551544, "learning_rate": 4.492043605138108e-06, "loss": 0.0008, "step": 141580 }, { "epoch": 2.3167798412828273, "grad_norm": 0.009690365754067898, "learning_rate": 4.4913333512556964e-06, "loss": 0.0007, "step": 141590 }, { "epoch": 2.3169434672339033, "grad_norm": 0.06266815215349197, "learning_rate": 4.490623107744566e-06, "loss": 0.0006, "step": 141600 }, { "epoch": 2.3171070931849793, "grad_norm": 0.11453786492347717, "learning_rate": 4.489912874619195e-06, "loss": 0.0034, "step": 141610 }, { "epoch": 2.317270719136055, "grad_norm": 0.04212502762675285, "learning_rate": 4.489202651894069e-06, "loss": 0.001, "step": 141620 }, { "epoch": 2.317434345087131, "grad_norm": 0.0031714029610157013, "learning_rate": 4.4884924395836655e-06, "loss": 0.0005, "step": 141630 }, { "epoch": 2.3175979710382064, "grad_norm": 0.017431581392884254, "learning_rate": 4.487782237702467e-06, "loss": 0.0007, "step": 141640 }, { "epoch": 2.3177615969892824, "grad_norm": 0.04067966714501381, "learning_rate": 4.487072046264952e-06, "loss": 0.0008, "step": 141650 }, { "epoch": 2.3179252229403584, "grad_norm": 0.027501456439495087, "learning_rate": 4.4863618652856026e-06, "loss": 0.0005, "step": 141660 }, { "epoch": 2.318088848891434, "grad_norm": 0.18088437616825104, "learning_rate": 4.485651694778897e-06, "loss": 0.0014, "step": 141670 }, { "epoch": 2.31825247484251, "grad_norm": 0.015477307140827179, "learning_rate": 4.484941534759317e-06, "loss": 0.0011, "step": 141680 }, { "epoch": 2.318416100793586, "grad_norm": 0.014327275566756725, "learning_rate": 4.48423138524134e-06, "loss": 0.0006, "step": 141690 }, { "epoch": 2.3185797267446615, "grad_norm": 0.00843789242208004, "learning_rate": 4.4835212462394475e-06, "loss": 0.0007, "step": 141700 }, { "epoch": 2.3187433526957375, "grad_norm": 0.08041774481534958, "learning_rate": 4.482811117768116e-06, "loss": 0.0015, "step": 141710 }, { "epoch": 2.3189069786468135, "grad_norm": 0.015439414419233799, "learning_rate": 4.482100999841828e-06, "loss": 0.0008, "step": 141720 }, { "epoch": 2.319070604597889, "grad_norm": 0.06131491810083389, "learning_rate": 4.481390892475058e-06, "loss": 0.0007, "step": 141730 }, { "epoch": 2.319234230548965, "grad_norm": 0.021514199674129486, "learning_rate": 4.4806807956822885e-06, "loss": 0.0009, "step": 141740 }, { "epoch": 2.319397856500041, "grad_norm": 0.020525243133306503, "learning_rate": 4.479970709477994e-06, "loss": 0.0006, "step": 141750 }, { "epoch": 2.3195614824511166, "grad_norm": 0.12020832300186157, "learning_rate": 4.4792606338766565e-06, "loss": 0.0013, "step": 141760 }, { "epoch": 2.3197251084021926, "grad_norm": 0.052710022777318954, "learning_rate": 4.478550568892751e-06, "loss": 0.0006, "step": 141770 }, { "epoch": 2.3198887343532686, "grad_norm": 0.04152975231409073, "learning_rate": 4.477840514540756e-06, "loss": 0.0013, "step": 141780 }, { "epoch": 2.320052360304344, "grad_norm": 0.06837081909179688, "learning_rate": 4.477130470835148e-06, "loss": 0.0009, "step": 141790 }, { "epoch": 2.32021598625542, "grad_norm": 0.042877692729234695, "learning_rate": 4.476420437790407e-06, "loss": 0.0009, "step": 141800 }, { "epoch": 2.320379612206496, "grad_norm": 0.02066047117114067, "learning_rate": 4.475710415421005e-06, "loss": 0.0009, "step": 141810 }, { "epoch": 2.3205432381575717, "grad_norm": 0.0008901625405997038, "learning_rate": 4.475000403741424e-06, "loss": 0.0006, "step": 141820 }, { "epoch": 2.3207068641086477, "grad_norm": 0.030172739177942276, "learning_rate": 4.474290402766137e-06, "loss": 0.0006, "step": 141830 }, { "epoch": 2.3208704900597237, "grad_norm": 0.021368814632296562, "learning_rate": 4.473580412509623e-06, "loss": 0.0011, "step": 141840 }, { "epoch": 2.3210341160107992, "grad_norm": 0.11157748848199844, "learning_rate": 4.472870432986356e-06, "loss": 0.0013, "step": 141850 }, { "epoch": 2.3211977419618752, "grad_norm": 0.010143434628844261, "learning_rate": 4.472160464210814e-06, "loss": 0.0009, "step": 141860 }, { "epoch": 2.3213613679129512, "grad_norm": 0.049412019550800323, "learning_rate": 4.471450506197467e-06, "loss": 0.0009, "step": 141870 }, { "epoch": 2.3215249938640268, "grad_norm": 0.0261375792324543, "learning_rate": 4.470740558960799e-06, "loss": 0.0012, "step": 141880 }, { "epoch": 2.3216886198151028, "grad_norm": 0.19146887958049774, "learning_rate": 4.470030622515278e-06, "loss": 0.0009, "step": 141890 }, { "epoch": 2.3218522457661783, "grad_norm": 0.00754582230001688, "learning_rate": 4.469320696875385e-06, "loss": 0.0007, "step": 141900 }, { "epoch": 2.3220158717172543, "grad_norm": 0.009274601005017757, "learning_rate": 4.468610782055588e-06, "loss": 0.0012, "step": 141910 }, { "epoch": 2.3221794976683303, "grad_norm": 0.12529544532299042, "learning_rate": 4.467900878070369e-06, "loss": 0.0015, "step": 141920 }, { "epoch": 2.322343123619406, "grad_norm": 0.06887828558683395, "learning_rate": 4.467190984934194e-06, "loss": 0.0008, "step": 141930 }, { "epoch": 2.322506749570482, "grad_norm": 0.01653459295630455, "learning_rate": 4.466481102661546e-06, "loss": 0.0006, "step": 141940 }, { "epoch": 2.322670375521558, "grad_norm": 0.07098782062530518, "learning_rate": 4.465771231266892e-06, "loss": 0.0008, "step": 141950 }, { "epoch": 2.3228340014726334, "grad_norm": 0.10088174045085907, "learning_rate": 4.465061370764711e-06, "loss": 0.001, "step": 141960 }, { "epoch": 2.3229976274237094, "grad_norm": 0.005236591212451458, "learning_rate": 4.464351521169469e-06, "loss": 0.0003, "step": 141970 }, { "epoch": 2.3231612533747854, "grad_norm": 0.014160172082483768, "learning_rate": 4.463641682495648e-06, "loss": 0.0008, "step": 141980 }, { "epoch": 2.323324879325861, "grad_norm": 0.016922788694500923, "learning_rate": 4.462931854757715e-06, "loss": 0.0007, "step": 141990 }, { "epoch": 2.323488505276937, "grad_norm": 0.054253388196229935, "learning_rate": 4.462222037970147e-06, "loss": 0.001, "step": 142000 }, { "epoch": 2.323652131228013, "grad_norm": 0.002121597295626998, "learning_rate": 4.461512232147411e-06, "loss": 0.0006, "step": 142010 }, { "epoch": 2.3238157571790885, "grad_norm": 0.08899153769016266, "learning_rate": 4.460802437303986e-06, "loss": 0.0012, "step": 142020 }, { "epoch": 2.3239793831301645, "grad_norm": 0.054922789335250854, "learning_rate": 4.4600926534543385e-06, "loss": 0.0012, "step": 142030 }, { "epoch": 2.32414300908124, "grad_norm": 0.04564875364303589, "learning_rate": 4.459382880612943e-06, "loss": 0.0011, "step": 142040 }, { "epoch": 2.324306635032316, "grad_norm": 0.03645686060190201, "learning_rate": 4.458673118794271e-06, "loss": 0.0006, "step": 142050 }, { "epoch": 2.324470260983392, "grad_norm": 0.08600547909736633, "learning_rate": 4.457963368012794e-06, "loss": 0.0005, "step": 142060 }, { "epoch": 2.3246338869344676, "grad_norm": 0.027717649936676025, "learning_rate": 4.457253628282982e-06, "loss": 0.0009, "step": 142070 }, { "epoch": 2.3247975128855436, "grad_norm": 0.0890464335680008, "learning_rate": 4.456543899619308e-06, "loss": 0.001, "step": 142080 }, { "epoch": 2.3249611388366196, "grad_norm": 0.0244058296084404, "learning_rate": 4.455834182036241e-06, "loss": 0.001, "step": 142090 }, { "epoch": 2.325124764787695, "grad_norm": 0.16228823363780975, "learning_rate": 4.455124475548253e-06, "loss": 0.0006, "step": 142100 }, { "epoch": 2.325288390738771, "grad_norm": 0.12300052493810654, "learning_rate": 4.454414780169812e-06, "loss": 0.0012, "step": 142110 }, { "epoch": 2.325452016689847, "grad_norm": 0.03078339621424675, "learning_rate": 4.453705095915391e-06, "loss": 0.0008, "step": 142120 }, { "epoch": 2.3256156426409227, "grad_norm": 0.0014482353581115603, "learning_rate": 4.452995422799458e-06, "loss": 0.0004, "step": 142130 }, { "epoch": 2.3257792685919987, "grad_norm": 0.034786563366651535, "learning_rate": 4.452285760836484e-06, "loss": 0.0011, "step": 142140 }, { "epoch": 2.3259428945430747, "grad_norm": 0.16903778910636902, "learning_rate": 4.451576110040935e-06, "loss": 0.0006, "step": 142150 }, { "epoch": 2.3261065204941502, "grad_norm": 0.13742168247699738, "learning_rate": 4.4508664704272855e-06, "loss": 0.0008, "step": 142160 }, { "epoch": 2.3262701464452262, "grad_norm": 0.04033748805522919, "learning_rate": 4.45015684201e-06, "loss": 0.0007, "step": 142170 }, { "epoch": 2.3264337723963022, "grad_norm": 0.020756682381033897, "learning_rate": 4.44944722480355e-06, "loss": 0.0007, "step": 142180 }, { "epoch": 2.3265973983473778, "grad_norm": 0.02823612652719021, "learning_rate": 4.448737618822402e-06, "loss": 0.0006, "step": 142190 }, { "epoch": 2.3267610242984538, "grad_norm": 0.03599498048424721, "learning_rate": 4.448028024081026e-06, "loss": 0.0006, "step": 142200 }, { "epoch": 2.3269246502495298, "grad_norm": 0.07927911728620529, "learning_rate": 4.4473184405938876e-06, "loss": 0.0007, "step": 142210 }, { "epoch": 2.3270882762006053, "grad_norm": 0.10495486110448837, "learning_rate": 4.446608868375458e-06, "loss": 0.0011, "step": 142220 }, { "epoch": 2.3272519021516813, "grad_norm": 0.016455180943012238, "learning_rate": 4.4458993074402015e-06, "loss": 0.0009, "step": 142230 }, { "epoch": 2.3274155281027573, "grad_norm": 0.04572524502873421, "learning_rate": 4.4451897578025885e-06, "loss": 0.0007, "step": 142240 }, { "epoch": 2.327579154053833, "grad_norm": 0.023306621238589287, "learning_rate": 4.444480219477083e-06, "loss": 0.001, "step": 142250 }, { "epoch": 2.327742780004909, "grad_norm": 0.08330176770687103, "learning_rate": 4.443770692478154e-06, "loss": 0.0011, "step": 142260 }, { "epoch": 2.327906405955985, "grad_norm": 0.05138396844267845, "learning_rate": 4.443061176820266e-06, "loss": 0.0011, "step": 142270 }, { "epoch": 2.3280700319070604, "grad_norm": 0.19411924481391907, "learning_rate": 4.4423516725178895e-06, "loss": 0.0006, "step": 142280 }, { "epoch": 2.3282336578581364, "grad_norm": 0.038917604833841324, "learning_rate": 4.441642179585487e-06, "loss": 0.0007, "step": 142290 }, { "epoch": 2.328397283809212, "grad_norm": 0.02971922606229782, "learning_rate": 4.440932698037525e-06, "loss": 0.0007, "step": 142300 }, { "epoch": 2.328560909760288, "grad_norm": 0.18718861043453217, "learning_rate": 4.44022322788847e-06, "loss": 0.0012, "step": 142310 }, { "epoch": 2.328724535711364, "grad_norm": 0.06474003195762634, "learning_rate": 4.439513769152788e-06, "loss": 0.0008, "step": 142320 }, { "epoch": 2.3288881616624395, "grad_norm": 0.0712849423289299, "learning_rate": 4.438804321844943e-06, "loss": 0.0007, "step": 142330 }, { "epoch": 2.3290517876135155, "grad_norm": 0.030123824253678322, "learning_rate": 4.438094885979401e-06, "loss": 0.0012, "step": 142340 }, { "epoch": 2.3292154135645915, "grad_norm": 0.053503137081861496, "learning_rate": 4.4373854615706255e-06, "loss": 0.0009, "step": 142350 }, { "epoch": 2.329379039515667, "grad_norm": 0.02420474775135517, "learning_rate": 4.436676048633083e-06, "loss": 0.0009, "step": 142360 }, { "epoch": 2.329542665466743, "grad_norm": 0.02437887340784073, "learning_rate": 4.4359666471812365e-06, "loss": 0.001, "step": 142370 }, { "epoch": 2.329706291417819, "grad_norm": 0.05334363505244255, "learning_rate": 4.43525725722955e-06, "loss": 0.001, "step": 142380 }, { "epoch": 2.3298699173688946, "grad_norm": 0.00806429609656334, "learning_rate": 4.434547878792488e-06, "loss": 0.0007, "step": 142390 }, { "epoch": 2.3300335433199706, "grad_norm": 0.04424981400370598, "learning_rate": 4.433838511884514e-06, "loss": 0.0007, "step": 142400 }, { "epoch": 2.330197169271046, "grad_norm": 0.1785760521888733, "learning_rate": 4.433129156520091e-06, "loss": 0.0012, "step": 142410 }, { "epoch": 2.330360795222122, "grad_norm": 0.041922178119421005, "learning_rate": 4.432419812713683e-06, "loss": 0.0012, "step": 142420 }, { "epoch": 2.330524421173198, "grad_norm": 0.11849691718816757, "learning_rate": 4.431710480479751e-06, "loss": 0.0014, "step": 142430 }, { "epoch": 2.3306880471242737, "grad_norm": 0.03297300636768341, "learning_rate": 4.4310011598327605e-06, "loss": 0.0005, "step": 142440 }, { "epoch": 2.3308516730753497, "grad_norm": 0.032962750643491745, "learning_rate": 4.430291850787171e-06, "loss": 0.0008, "step": 142450 }, { "epoch": 2.3310152990264257, "grad_norm": 0.06130113825201988, "learning_rate": 4.4295825533574475e-06, "loss": 0.0011, "step": 142460 }, { "epoch": 2.3311789249775012, "grad_norm": 0.08408099412918091, "learning_rate": 4.42887326755805e-06, "loss": 0.0012, "step": 142470 }, { "epoch": 2.3313425509285772, "grad_norm": 0.15728381276130676, "learning_rate": 4.428163993403441e-06, "loss": 0.0008, "step": 142480 }, { "epoch": 2.3315061768796532, "grad_norm": 0.022748000919818878, "learning_rate": 4.4274547309080814e-06, "loss": 0.0008, "step": 142490 }, { "epoch": 2.3316698028307288, "grad_norm": 0.06055579334497452, "learning_rate": 4.426745480086435e-06, "loss": 0.0013, "step": 142500 }, { "epoch": 2.3318334287818048, "grad_norm": 0.35778844356536865, "learning_rate": 4.426036240952957e-06, "loss": 0.0008, "step": 142510 }, { "epoch": 2.3319970547328808, "grad_norm": 0.01878097467124462, "learning_rate": 4.425327013522116e-06, "loss": 0.0008, "step": 142520 }, { "epoch": 2.3321606806839563, "grad_norm": 0.06600616872310638, "learning_rate": 4.424617797808363e-06, "loss": 0.0006, "step": 142530 }, { "epoch": 2.3323243066350323, "grad_norm": 0.027286598458886147, "learning_rate": 4.423908593826169e-06, "loss": 0.0008, "step": 142540 }, { "epoch": 2.3324879325861083, "grad_norm": 0.09353535622358322, "learning_rate": 4.423199401589986e-06, "loss": 0.0008, "step": 142550 }, { "epoch": 2.332651558537184, "grad_norm": 0.0737186074256897, "learning_rate": 4.422490221114279e-06, "loss": 0.0013, "step": 142560 }, { "epoch": 2.33281518448826, "grad_norm": 0.07105213403701782, "learning_rate": 4.421781052413502e-06, "loss": 0.0011, "step": 142570 }, { "epoch": 2.332978810439336, "grad_norm": 0.027070920914411545, "learning_rate": 4.421071895502121e-06, "loss": 0.0007, "step": 142580 }, { "epoch": 2.3331424363904114, "grad_norm": 0.04551715403795242, "learning_rate": 4.4203627503945884e-06, "loss": 0.0014, "step": 142590 }, { "epoch": 2.3333060623414874, "grad_norm": 0.03158954530954361, "learning_rate": 4.419653617105369e-06, "loss": 0.0005, "step": 142600 }, { "epoch": 2.3334696882925634, "grad_norm": 0.007294036448001862, "learning_rate": 4.418944495648917e-06, "loss": 0.0005, "step": 142610 }, { "epoch": 2.333633314243639, "grad_norm": 0.060232385993003845, "learning_rate": 4.418235386039695e-06, "loss": 0.0008, "step": 142620 }, { "epoch": 2.333796940194715, "grad_norm": 0.029330959543585777, "learning_rate": 4.417526288292155e-06, "loss": 0.0006, "step": 142630 }, { "epoch": 2.333960566145791, "grad_norm": 0.020636774599552155, "learning_rate": 4.416817202420762e-06, "loss": 0.0012, "step": 142640 }, { "epoch": 2.3341241920968665, "grad_norm": 0.03346136957406998, "learning_rate": 4.416108128439967e-06, "loss": 0.0015, "step": 142650 }, { "epoch": 2.3342878180479425, "grad_norm": 0.0740942731499672, "learning_rate": 4.415399066364235e-06, "loss": 0.0005, "step": 142660 }, { "epoch": 2.3344514439990185, "grad_norm": 0.04831238090991974, "learning_rate": 4.414690016208015e-06, "loss": 0.0013, "step": 142670 }, { "epoch": 2.334615069950094, "grad_norm": 0.037260301411151886, "learning_rate": 4.4139809779857705e-06, "loss": 0.0006, "step": 142680 }, { "epoch": 2.33477869590117, "grad_norm": 0.09755398333072662, "learning_rate": 4.413271951711953e-06, "loss": 0.0009, "step": 142690 }, { "epoch": 2.3349423218522456, "grad_norm": 0.024876797571778297, "learning_rate": 4.412562937401024e-06, "loss": 0.0009, "step": 142700 }, { "epoch": 2.3351059478033216, "grad_norm": 0.09346575289964676, "learning_rate": 4.411853935067435e-06, "loss": 0.0013, "step": 142710 }, { "epoch": 2.3352695737543976, "grad_norm": 0.010762454941868782, "learning_rate": 4.411144944725645e-06, "loss": 0.0005, "step": 142720 }, { "epoch": 2.335433199705473, "grad_norm": 0.0063763391226530075, "learning_rate": 4.410435966390108e-06, "loss": 0.0012, "step": 142730 }, { "epoch": 2.335596825656549, "grad_norm": 0.022731086239218712, "learning_rate": 4.409727000075281e-06, "loss": 0.0004, "step": 142740 }, { "epoch": 2.335760451607625, "grad_norm": 0.07802167534828186, "learning_rate": 4.409018045795619e-06, "loss": 0.0007, "step": 142750 }, { "epoch": 2.3359240775587007, "grad_norm": 0.020991405472159386, "learning_rate": 4.408309103565575e-06, "loss": 0.001, "step": 142760 }, { "epoch": 2.3360877035097767, "grad_norm": 0.22994914650917053, "learning_rate": 4.407600173399605e-06, "loss": 0.0008, "step": 142770 }, { "epoch": 2.3362513294608527, "grad_norm": 0.02743867039680481, "learning_rate": 4.406891255312166e-06, "loss": 0.0008, "step": 142780 }, { "epoch": 2.3364149554119282, "grad_norm": 0.0805526003241539, "learning_rate": 4.406182349317707e-06, "loss": 0.0028, "step": 142790 }, { "epoch": 2.3365785813630042, "grad_norm": 0.050994474440813065, "learning_rate": 4.405473455430687e-06, "loss": 0.0007, "step": 142800 }, { "epoch": 2.3367422073140798, "grad_norm": 0.032989468425512314, "learning_rate": 4.404764573665556e-06, "loss": 0.0004, "step": 142810 }, { "epoch": 2.3369058332651558, "grad_norm": 0.05239018797874451, "learning_rate": 4.404055704036771e-06, "loss": 0.001, "step": 142820 }, { "epoch": 2.3370694592162318, "grad_norm": 0.06778672337532043, "learning_rate": 4.403346846558782e-06, "loss": 0.0006, "step": 142830 }, { "epoch": 2.3372330851673073, "grad_norm": 0.030884671956300735, "learning_rate": 4.402638001246044e-06, "loss": 0.0005, "step": 142840 }, { "epoch": 2.3373967111183833, "grad_norm": 0.09109976887702942, "learning_rate": 4.401929168113009e-06, "loss": 0.0017, "step": 142850 }, { "epoch": 2.3375603370694593, "grad_norm": 0.0028230291791260242, "learning_rate": 4.401220347174129e-06, "loss": 0.0008, "step": 142860 }, { "epoch": 2.337723963020535, "grad_norm": 0.0737968236207962, "learning_rate": 4.400511538443857e-06, "loss": 0.0026, "step": 142870 }, { "epoch": 2.337887588971611, "grad_norm": 0.012124534696340561, "learning_rate": 4.399802741936646e-06, "loss": 0.0011, "step": 142880 }, { "epoch": 2.338051214922687, "grad_norm": 0.08816391229629517, "learning_rate": 4.399093957666945e-06, "loss": 0.0012, "step": 142890 }, { "epoch": 2.3382148408737624, "grad_norm": 0.02244524471461773, "learning_rate": 4.3983851856492084e-06, "loss": 0.0008, "step": 142900 }, { "epoch": 2.3383784668248384, "grad_norm": 0.016776828095316887, "learning_rate": 4.397676425897885e-06, "loss": 0.0006, "step": 142910 }, { "epoch": 2.3385420927759144, "grad_norm": 0.05938836559653282, "learning_rate": 4.396967678427428e-06, "loss": 0.0009, "step": 142920 }, { "epoch": 2.33870571872699, "grad_norm": 0.027695506811141968, "learning_rate": 4.3962589432522875e-06, "loss": 0.0007, "step": 142930 }, { "epoch": 2.338869344678066, "grad_norm": 0.04633859544992447, "learning_rate": 4.395550220386913e-06, "loss": 0.0009, "step": 142940 }, { "epoch": 2.339032970629142, "grad_norm": 0.02284647524356842, "learning_rate": 4.394841509845755e-06, "loss": 0.0021, "step": 142950 }, { "epoch": 2.3391965965802175, "grad_norm": 0.012107075192034245, "learning_rate": 4.394132811643266e-06, "loss": 0.0009, "step": 142960 }, { "epoch": 2.3393602225312935, "grad_norm": 0.08687449991703033, "learning_rate": 4.393424125793891e-06, "loss": 0.002, "step": 142970 }, { "epoch": 2.3395238484823695, "grad_norm": 0.02963964454829693, "learning_rate": 4.392715452312084e-06, "loss": 0.0005, "step": 142980 }, { "epoch": 2.339687474433445, "grad_norm": 0.0026947355363518, "learning_rate": 4.392006791212291e-06, "loss": 0.0007, "step": 142990 }, { "epoch": 2.339851100384521, "grad_norm": 0.02516433596611023, "learning_rate": 4.391298142508964e-06, "loss": 0.0006, "step": 143000 }, { "epoch": 2.340014726335597, "grad_norm": 0.004355539567768574, "learning_rate": 4.39058950621655e-06, "loss": 0.0008, "step": 143010 }, { "epoch": 2.3401783522866726, "grad_norm": 0.06583341211080551, "learning_rate": 4.389880882349497e-06, "loss": 0.0011, "step": 143020 }, { "epoch": 2.3403419782377486, "grad_norm": 0.0014920044923201203, "learning_rate": 4.3891722709222535e-06, "loss": 0.0007, "step": 143030 }, { "epoch": 2.3405056041888246, "grad_norm": 0.03345075622200966, "learning_rate": 4.3884636719492694e-06, "loss": 0.0006, "step": 143040 }, { "epoch": 2.3406692301399, "grad_norm": 0.07919789105653763, "learning_rate": 4.38775508544499e-06, "loss": 0.0009, "step": 143050 }, { "epoch": 2.340832856090976, "grad_norm": 0.10194171965122223, "learning_rate": 4.387046511423863e-06, "loss": 0.0006, "step": 143060 }, { "epoch": 2.3409964820420517, "grad_norm": 0.034123893827199936, "learning_rate": 4.386337949900337e-06, "loss": 0.001, "step": 143070 }, { "epoch": 2.3411601079931277, "grad_norm": 0.01211558934301138, "learning_rate": 4.385629400888859e-06, "loss": 0.0012, "step": 143080 }, { "epoch": 2.3413237339442037, "grad_norm": 0.01154989656060934, "learning_rate": 4.384920864403874e-06, "loss": 0.0008, "step": 143090 }, { "epoch": 2.3414873598952792, "grad_norm": 0.033034417778253555, "learning_rate": 4.384212340459831e-06, "loss": 0.0007, "step": 143100 }, { "epoch": 2.3416509858463552, "grad_norm": 0.05124310031533241, "learning_rate": 4.383503829071173e-06, "loss": 0.001, "step": 143110 }, { "epoch": 2.3418146117974312, "grad_norm": 0.016417868435382843, "learning_rate": 4.3827953302523485e-06, "loss": 0.0009, "step": 143120 }, { "epoch": 2.3419782377485068, "grad_norm": 0.005937636364251375, "learning_rate": 4.382086844017801e-06, "loss": 0.0009, "step": 143130 }, { "epoch": 2.3421418636995828, "grad_norm": 0.022661913186311722, "learning_rate": 4.38137837038198e-06, "loss": 0.0014, "step": 143140 }, { "epoch": 2.3423054896506588, "grad_norm": 0.07803632318973541, "learning_rate": 4.380669909359324e-06, "loss": 0.0006, "step": 143150 }, { "epoch": 2.3424691156017343, "grad_norm": 0.018128778785467148, "learning_rate": 4.379961460964287e-06, "loss": 0.0009, "step": 143160 }, { "epoch": 2.3426327415528103, "grad_norm": 0.09309232234954834, "learning_rate": 4.379253025211303e-06, "loss": 0.0006, "step": 143170 }, { "epoch": 2.342796367503886, "grad_norm": 0.06695307791233063, "learning_rate": 4.378544602114826e-06, "loss": 0.0013, "step": 143180 }, { "epoch": 2.342959993454962, "grad_norm": 0.06460007280111313, "learning_rate": 4.377836191689293e-06, "loss": 0.0009, "step": 143190 }, { "epoch": 2.343123619406038, "grad_norm": 0.03588186204433441, "learning_rate": 4.377127793949154e-06, "loss": 0.0009, "step": 143200 }, { "epoch": 2.3432872453571134, "grad_norm": 0.03918863832950592, "learning_rate": 4.376419408908846e-06, "loss": 0.0006, "step": 143210 }, { "epoch": 2.3434508713081894, "grad_norm": 0.012347022071480751, "learning_rate": 4.375711036582819e-06, "loss": 0.0005, "step": 143220 }, { "epoch": 2.3436144972592654, "grad_norm": 0.044482823461294174, "learning_rate": 4.3750026769855106e-06, "loss": 0.001, "step": 143230 }, { "epoch": 2.343778123210341, "grad_norm": 0.05348649248480797, "learning_rate": 4.374294330131369e-06, "loss": 0.0006, "step": 143240 }, { "epoch": 2.343941749161417, "grad_norm": 0.022388136014342308, "learning_rate": 4.37358599603483e-06, "loss": 0.0009, "step": 143250 }, { "epoch": 2.344105375112493, "grad_norm": 0.1092398539185524, "learning_rate": 4.372877674710344e-06, "loss": 0.0007, "step": 143260 }, { "epoch": 2.3442690010635685, "grad_norm": 0.06564076989889145, "learning_rate": 4.372169366172345e-06, "loss": 0.0012, "step": 143270 }, { "epoch": 2.3444326270146445, "grad_norm": 0.026466816663742065, "learning_rate": 4.371461070435283e-06, "loss": 0.0007, "step": 143280 }, { "epoch": 2.3445962529657205, "grad_norm": 0.06734713166952133, "learning_rate": 4.3707527875135905e-06, "loss": 0.0009, "step": 143290 }, { "epoch": 2.344759878916796, "grad_norm": 0.060522813349962234, "learning_rate": 4.3700445174217175e-06, "loss": 0.0011, "step": 143300 }, { "epoch": 2.344923504867872, "grad_norm": 0.11714698374271393, "learning_rate": 4.369336260174097e-06, "loss": 0.0007, "step": 143310 }, { "epoch": 2.345087130818948, "grad_norm": 0.017619363963603973, "learning_rate": 4.368628015785178e-06, "loss": 0.0008, "step": 143320 }, { "epoch": 2.3452507567700236, "grad_norm": 0.016513174399733543, "learning_rate": 4.3679197842693945e-06, "loss": 0.0009, "step": 143330 }, { "epoch": 2.3454143827210996, "grad_norm": 0.0011599685531109571, "learning_rate": 4.367211565641189e-06, "loss": 0.0008, "step": 143340 }, { "epoch": 2.3455780086721756, "grad_norm": 0.11573547124862671, "learning_rate": 4.366503359915002e-06, "loss": 0.001, "step": 143350 }, { "epoch": 2.345741634623251, "grad_norm": 0.012584676034748554, "learning_rate": 4.365795167105273e-06, "loss": 0.001, "step": 143360 }, { "epoch": 2.345905260574327, "grad_norm": 0.06950235366821289, "learning_rate": 4.36508698722644e-06, "loss": 0.0005, "step": 143370 }, { "epoch": 2.346068886525403, "grad_norm": 0.004808464553207159, "learning_rate": 4.3643788202929446e-06, "loss": 0.0008, "step": 143380 }, { "epoch": 2.3462325124764787, "grad_norm": 0.04583169147372246, "learning_rate": 4.363670666319223e-06, "loss": 0.0021, "step": 143390 }, { "epoch": 2.3463961384275547, "grad_norm": 0.03527078777551651, "learning_rate": 4.3629625253197176e-06, "loss": 0.0006, "step": 143400 }, { "epoch": 2.3465597643786307, "grad_norm": 0.040799617767333984, "learning_rate": 4.362254397308862e-06, "loss": 0.0005, "step": 143410 }, { "epoch": 2.3467233903297062, "grad_norm": 0.04751143604516983, "learning_rate": 4.361546282301099e-06, "loss": 0.0009, "step": 143420 }, { "epoch": 2.3468870162807822, "grad_norm": 0.05159417912364006, "learning_rate": 4.360838180310863e-06, "loss": 0.0007, "step": 143430 }, { "epoch": 2.347050642231858, "grad_norm": 0.026726791635155678, "learning_rate": 4.360130091352594e-06, "loss": 0.0011, "step": 143440 }, { "epoch": 2.3472142681829338, "grad_norm": 0.07103944569826126, "learning_rate": 4.3594220154407265e-06, "loss": 0.0008, "step": 143450 }, { "epoch": 2.3473778941340098, "grad_norm": 0.12620283663272858, "learning_rate": 4.358713952589702e-06, "loss": 0.0009, "step": 143460 }, { "epoch": 2.3475415200850853, "grad_norm": 0.040706127882003784, "learning_rate": 4.358005902813952e-06, "loss": 0.0007, "step": 143470 }, { "epoch": 2.3477051460361613, "grad_norm": 0.05799393728375435, "learning_rate": 4.357297866127917e-06, "loss": 0.0012, "step": 143480 }, { "epoch": 2.3478687719872373, "grad_norm": 0.011037529446184635, "learning_rate": 4.356589842546031e-06, "loss": 0.0011, "step": 143490 }, { "epoch": 2.348032397938313, "grad_norm": 0.04027289152145386, "learning_rate": 4.3558818320827325e-06, "loss": 0.0005, "step": 143500 }, { "epoch": 2.348196023889389, "grad_norm": 0.007692880462855101, "learning_rate": 4.355173834752453e-06, "loss": 0.0009, "step": 143510 }, { "epoch": 2.348359649840465, "grad_norm": 0.04277868941426277, "learning_rate": 4.354465850569634e-06, "loss": 0.0006, "step": 143520 }, { "epoch": 2.3485232757915404, "grad_norm": 0.12290660291910172, "learning_rate": 4.3537578795487045e-06, "loss": 0.0009, "step": 143530 }, { "epoch": 2.3486869017426164, "grad_norm": 0.06235234811902046, "learning_rate": 4.3530499217041036e-06, "loss": 0.0005, "step": 143540 }, { "epoch": 2.3488505276936924, "grad_norm": 0.09883380681276321, "learning_rate": 4.352341977050264e-06, "loss": 0.0012, "step": 143550 }, { "epoch": 2.349014153644768, "grad_norm": 0.05809694528579712, "learning_rate": 4.351634045601621e-06, "loss": 0.001, "step": 143560 }, { "epoch": 2.349177779595844, "grad_norm": 0.044340573251247406, "learning_rate": 4.350926127372608e-06, "loss": 0.0013, "step": 143570 }, { "epoch": 2.3493414055469195, "grad_norm": 0.005562294740229845, "learning_rate": 4.35021822237766e-06, "loss": 0.0007, "step": 143580 }, { "epoch": 2.3495050314979955, "grad_norm": 0.038584932684898376, "learning_rate": 4.3495103306312085e-06, "loss": 0.0009, "step": 143590 }, { "epoch": 2.3496686574490715, "grad_norm": 0.10919465124607086, "learning_rate": 4.348802452147689e-06, "loss": 0.0019, "step": 143600 }, { "epoch": 2.349832283400147, "grad_norm": 0.03887861967086792, "learning_rate": 4.3480945869415334e-06, "loss": 0.002, "step": 143610 }, { "epoch": 2.349995909351223, "grad_norm": 0.029854396358132362, "learning_rate": 4.347386735027176e-06, "loss": 0.002, "step": 143620 }, { "epoch": 2.350159535302299, "grad_norm": 0.018407931551337242, "learning_rate": 4.346678896419047e-06, "loss": 0.0007, "step": 143630 }, { "epoch": 2.3503231612533746, "grad_norm": 0.0035484209656715393, "learning_rate": 4.34597107113158e-06, "loss": 0.0012, "step": 143640 }, { "epoch": 2.3504867872044506, "grad_norm": 0.052092649042606354, "learning_rate": 4.3452632591792054e-06, "loss": 0.0006, "step": 143650 }, { "epoch": 2.3506504131555266, "grad_norm": 0.0011573941446840763, "learning_rate": 4.344555460576358e-06, "loss": 0.0011, "step": 143660 }, { "epoch": 2.350814039106602, "grad_norm": 0.05869236961007118, "learning_rate": 4.343847675337466e-06, "loss": 0.0013, "step": 143670 }, { "epoch": 2.350977665057678, "grad_norm": 0.018872885033488274, "learning_rate": 4.343139903476963e-06, "loss": 0.0007, "step": 143680 }, { "epoch": 2.351141291008754, "grad_norm": 0.006017865613102913, "learning_rate": 4.342432145009277e-06, "loss": 0.0011, "step": 143690 }, { "epoch": 2.3513049169598297, "grad_norm": 0.15893320739269257, "learning_rate": 4.341724399948842e-06, "loss": 0.0012, "step": 143700 }, { "epoch": 2.3514685429109057, "grad_norm": 0.021735573187470436, "learning_rate": 4.3410166683100856e-06, "loss": 0.0011, "step": 143710 }, { "epoch": 2.3516321688619817, "grad_norm": 0.05483744665980339, "learning_rate": 4.34030895010744e-06, "loss": 0.0006, "step": 143720 }, { "epoch": 2.3517957948130572, "grad_norm": 0.06995350867509842, "learning_rate": 4.339601245355332e-06, "loss": 0.001, "step": 143730 }, { "epoch": 2.3519594207641332, "grad_norm": 0.028410034254193306, "learning_rate": 4.338893554068195e-06, "loss": 0.0005, "step": 143740 }, { "epoch": 2.352123046715209, "grad_norm": 0.0403343141078949, "learning_rate": 4.338185876260454e-06, "loss": 0.0009, "step": 143750 }, { "epoch": 2.3522866726662848, "grad_norm": 0.04155813902616501, "learning_rate": 4.337478211946543e-06, "loss": 0.0011, "step": 143760 }, { "epoch": 2.3524502986173608, "grad_norm": 0.04131147637963295, "learning_rate": 4.336770561140885e-06, "loss": 0.0008, "step": 143770 }, { "epoch": 2.3526139245684368, "grad_norm": 0.004695537965744734, "learning_rate": 4.336062923857914e-06, "loss": 0.0005, "step": 143780 }, { "epoch": 2.3527775505195123, "grad_norm": 0.09056772291660309, "learning_rate": 4.3353553001120515e-06, "loss": 0.0015, "step": 143790 }, { "epoch": 2.3529411764705883, "grad_norm": 0.04156920686364174, "learning_rate": 4.334647689917734e-06, "loss": 0.001, "step": 143800 }, { "epoch": 2.3531048024216643, "grad_norm": 0.030366983264684677, "learning_rate": 4.333940093289379e-06, "loss": 0.001, "step": 143810 }, { "epoch": 2.35326842837274, "grad_norm": 0.0033431858755648136, "learning_rate": 4.333232510241424e-06, "loss": 0.0007, "step": 143820 }, { "epoch": 2.353432054323816, "grad_norm": 0.05598219484090805, "learning_rate": 4.332524940788287e-06, "loss": 0.0011, "step": 143830 }, { "epoch": 2.3535956802748914, "grad_norm": 0.1022983193397522, "learning_rate": 4.331817384944402e-06, "loss": 0.0009, "step": 143840 }, { "epoch": 2.3537593062259674, "grad_norm": 0.027846921235322952, "learning_rate": 4.331109842724189e-06, "loss": 0.0011, "step": 143850 }, { "epoch": 2.3539229321770434, "grad_norm": 0.03701205924153328, "learning_rate": 4.330402314142081e-06, "loss": 0.0009, "step": 143860 }, { "epoch": 2.354086558128119, "grad_norm": 0.018134912475943565, "learning_rate": 4.329694799212497e-06, "loss": 0.0022, "step": 143870 }, { "epoch": 2.354250184079195, "grad_norm": 0.03310570865869522, "learning_rate": 4.328987297949869e-06, "loss": 0.0012, "step": 143880 }, { "epoch": 2.354413810030271, "grad_norm": 0.11213578283786774, "learning_rate": 4.328279810368616e-06, "loss": 0.0017, "step": 143890 }, { "epoch": 2.3545774359813465, "grad_norm": 0.03146783635020256, "learning_rate": 4.32757233648317e-06, "loss": 0.0004, "step": 143900 }, { "epoch": 2.3547410619324225, "grad_norm": 0.013254611752927303, "learning_rate": 4.326864876307949e-06, "loss": 0.001, "step": 143910 }, { "epoch": 2.3549046878834985, "grad_norm": 0.0021542171016335487, "learning_rate": 4.3261574298573835e-06, "loss": 0.0004, "step": 143920 }, { "epoch": 2.355068313834574, "grad_norm": 0.03134103864431381, "learning_rate": 4.325449997145892e-06, "loss": 0.0005, "step": 143930 }, { "epoch": 2.35523193978565, "grad_norm": 0.01789662055671215, "learning_rate": 4.3247425781879045e-06, "loss": 0.0009, "step": 143940 }, { "epoch": 2.355395565736726, "grad_norm": 0.0038028177805244923, "learning_rate": 4.324035172997838e-06, "loss": 0.0011, "step": 143950 }, { "epoch": 2.3555591916878016, "grad_norm": 0.015408659353852272, "learning_rate": 4.323327781590123e-06, "loss": 0.0007, "step": 143960 }, { "epoch": 2.3557228176388776, "grad_norm": 0.05583018437027931, "learning_rate": 4.322620403979177e-06, "loss": 0.0011, "step": 143970 }, { "epoch": 2.355886443589953, "grad_norm": 0.018010705709457397, "learning_rate": 4.321913040179426e-06, "loss": 0.0016, "step": 143980 }, { "epoch": 2.356050069541029, "grad_norm": 0.006153735797852278, "learning_rate": 4.32120569020529e-06, "loss": 0.0007, "step": 143990 }, { "epoch": 2.356213695492105, "grad_norm": 0.031961020082235336, "learning_rate": 4.320498354071193e-06, "loss": 0.0007, "step": 144000 }, { "epoch": 2.3563773214431807, "grad_norm": 0.028095854446291924, "learning_rate": 4.319791031791557e-06, "loss": 0.0011, "step": 144010 }, { "epoch": 2.3565409473942567, "grad_norm": 0.03687870129942894, "learning_rate": 4.319083723380804e-06, "loss": 0.0014, "step": 144020 }, { "epoch": 2.3567045733453327, "grad_norm": 0.01006619818508625, "learning_rate": 4.318376428853354e-06, "loss": 0.0003, "step": 144030 }, { "epoch": 2.3568681992964082, "grad_norm": 0.0525582954287529, "learning_rate": 4.317669148223629e-06, "loss": 0.0006, "step": 144040 }, { "epoch": 2.3570318252474842, "grad_norm": 0.08033517748117447, "learning_rate": 4.31696188150605e-06, "loss": 0.0007, "step": 144050 }, { "epoch": 2.35719545119856, "grad_norm": 0.060666244477033615, "learning_rate": 4.316254628715038e-06, "loss": 0.001, "step": 144060 }, { "epoch": 2.3573590771496358, "grad_norm": 0.0540970116853714, "learning_rate": 4.315547389865011e-06, "loss": 0.0009, "step": 144070 }, { "epoch": 2.3575227031007118, "grad_norm": 0.06942979246377945, "learning_rate": 4.314840164970392e-06, "loss": 0.001, "step": 144080 }, { "epoch": 2.3576863290517878, "grad_norm": 0.03707389906048775, "learning_rate": 4.314132954045598e-06, "loss": 0.0007, "step": 144090 }, { "epoch": 2.3578499550028633, "grad_norm": 0.009067985229194164, "learning_rate": 4.313425757105051e-06, "loss": 0.0003, "step": 144100 }, { "epoch": 2.3580135809539393, "grad_norm": 0.018033768981695175, "learning_rate": 4.312718574163168e-06, "loss": 0.0006, "step": 144110 }, { "epoch": 2.3581772069050153, "grad_norm": 0.004741364158689976, "learning_rate": 4.31201140523437e-06, "loss": 0.0009, "step": 144120 }, { "epoch": 2.358340832856091, "grad_norm": 0.042480189353227615, "learning_rate": 4.3113042503330725e-06, "loss": 0.0007, "step": 144130 }, { "epoch": 2.358504458807167, "grad_norm": 0.04523114487528801, "learning_rate": 4.3105971094736975e-06, "loss": 0.001, "step": 144140 }, { "epoch": 2.358668084758243, "grad_norm": 0.06456886976957321, "learning_rate": 4.30988998267066e-06, "loss": 0.0006, "step": 144150 }, { "epoch": 2.3588317107093184, "grad_norm": 0.04172525182366371, "learning_rate": 4.309182869938379e-06, "loss": 0.0008, "step": 144160 }, { "epoch": 2.3589953366603944, "grad_norm": 0.08255679160356522, "learning_rate": 4.3084757712912716e-06, "loss": 0.0008, "step": 144170 }, { "epoch": 2.3591589626114704, "grad_norm": 0.2718735933303833, "learning_rate": 4.307768686743756e-06, "loss": 0.001, "step": 144180 }, { "epoch": 2.359322588562546, "grad_norm": 0.018518241122364998, "learning_rate": 4.3070616163102465e-06, "loss": 0.0007, "step": 144190 }, { "epoch": 2.359486214513622, "grad_norm": 0.056767597794532776, "learning_rate": 4.306354560005163e-06, "loss": 0.0008, "step": 144200 }, { "epoch": 2.359649840464698, "grad_norm": 0.05145205929875374, "learning_rate": 4.305647517842919e-06, "loss": 0.0006, "step": 144210 }, { "epoch": 2.3598134664157735, "grad_norm": 0.025852002203464508, "learning_rate": 4.304940489837932e-06, "loss": 0.0011, "step": 144220 }, { "epoch": 2.3599770923668495, "grad_norm": 0.07114867866039276, "learning_rate": 4.304233476004617e-06, "loss": 0.0009, "step": 144230 }, { "epoch": 2.360140718317925, "grad_norm": 0.0004314325633458793, "learning_rate": 4.303526476357391e-06, "loss": 0.0006, "step": 144240 }, { "epoch": 2.360304344269001, "grad_norm": 0.0289412010461092, "learning_rate": 4.302819490910665e-06, "loss": 0.003, "step": 144250 }, { "epoch": 2.360467970220077, "grad_norm": 0.03891318663954735, "learning_rate": 4.30211251967886e-06, "loss": 0.0012, "step": 144260 }, { "epoch": 2.3606315961711526, "grad_norm": 0.07609360665082932, "learning_rate": 4.301405562676385e-06, "loss": 0.0008, "step": 144270 }, { "epoch": 2.3607952221222286, "grad_norm": 0.014948444440960884, "learning_rate": 4.300698619917658e-06, "loss": 0.0004, "step": 144280 }, { "epoch": 2.3609588480733046, "grad_norm": 0.10616090893745422, "learning_rate": 4.29999169141709e-06, "loss": 0.0013, "step": 144290 }, { "epoch": 2.36112247402438, "grad_norm": 0.013492031022906303, "learning_rate": 4.2992847771890965e-06, "loss": 0.0005, "step": 144300 }, { "epoch": 2.361286099975456, "grad_norm": 0.02732423134148121, "learning_rate": 4.29857787724809e-06, "loss": 0.0012, "step": 144310 }, { "epoch": 2.361449725926532, "grad_norm": 0.02537713758647442, "learning_rate": 4.297870991608485e-06, "loss": 0.0008, "step": 144320 }, { "epoch": 2.3616133518776077, "grad_norm": 0.057100847363471985, "learning_rate": 4.297164120284693e-06, "loss": 0.0007, "step": 144330 }, { "epoch": 2.3617769778286837, "grad_norm": 0.012858434580266476, "learning_rate": 4.296457263291127e-06, "loss": 0.0006, "step": 144340 }, { "epoch": 2.3619406037797592, "grad_norm": 0.02594614028930664, "learning_rate": 4.295750420642198e-06, "loss": 0.0006, "step": 144350 }, { "epoch": 2.3621042297308352, "grad_norm": 0.023354388773441315, "learning_rate": 4.295043592352321e-06, "loss": 0.001, "step": 144360 }, { "epoch": 2.362267855681911, "grad_norm": 0.08812329918146133, "learning_rate": 4.294336778435903e-06, "loss": 0.001, "step": 144370 }, { "epoch": 2.3624314816329868, "grad_norm": 0.11375389248132706, "learning_rate": 4.29362997890736e-06, "loss": 0.001, "step": 144380 }, { "epoch": 2.3625951075840628, "grad_norm": 0.05770057439804077, "learning_rate": 4.2929231937811005e-06, "loss": 0.0007, "step": 144390 }, { "epoch": 2.3627587335351388, "grad_norm": 0.037583861500024796, "learning_rate": 4.2922164230715355e-06, "loss": 0.0008, "step": 144400 }, { "epoch": 2.3629223594862143, "grad_norm": 0.0033096028491854668, "learning_rate": 4.291509666793076e-06, "loss": 0.0011, "step": 144410 }, { "epoch": 2.3630859854372903, "grad_norm": 0.035132378339767456, "learning_rate": 4.290802924960132e-06, "loss": 0.0008, "step": 144420 }, { "epoch": 2.3632496113883663, "grad_norm": 0.024404561147093773, "learning_rate": 4.290096197587113e-06, "loss": 0.0009, "step": 144430 }, { "epoch": 2.363413237339442, "grad_norm": 0.04864124953746796, "learning_rate": 4.289389484688429e-06, "loss": 0.0008, "step": 144440 }, { "epoch": 2.363576863290518, "grad_norm": 0.0030366200953722, "learning_rate": 4.288682786278488e-06, "loss": 0.0006, "step": 144450 }, { "epoch": 2.363740489241594, "grad_norm": 0.04016510769724846, "learning_rate": 4.287976102371701e-06, "loss": 0.0006, "step": 144460 }, { "epoch": 2.3639041151926694, "grad_norm": 0.030188629403710365, "learning_rate": 4.287269432982474e-06, "loss": 0.0009, "step": 144470 }, { "epoch": 2.3640677411437454, "grad_norm": 0.08299907296895981, "learning_rate": 4.28656277812522e-06, "loss": 0.0003, "step": 144480 }, { "epoch": 2.3642313670948214, "grad_norm": 0.0014918441884219646, "learning_rate": 4.285856137814341e-06, "loss": 0.0003, "step": 144490 }, { "epoch": 2.364394993045897, "grad_norm": 0.03622402995824814, "learning_rate": 4.285149512064252e-06, "loss": 0.0012, "step": 144500 }, { "epoch": 2.364558618996973, "grad_norm": 0.00343385711312294, "learning_rate": 4.284442900889353e-06, "loss": 0.0006, "step": 144510 }, { "epoch": 2.364722244948049, "grad_norm": 0.048452336341142654, "learning_rate": 4.283736304304057e-06, "loss": 0.0014, "step": 144520 }, { "epoch": 2.3648858708991245, "grad_norm": 0.012855875305831432, "learning_rate": 4.283029722322766e-06, "loss": 0.0008, "step": 144530 }, { "epoch": 2.3650494968502005, "grad_norm": 0.02895098552107811, "learning_rate": 4.282323154959892e-06, "loss": 0.0011, "step": 144540 }, { "epoch": 2.3652131228012765, "grad_norm": 0.03521023690700531, "learning_rate": 4.281616602229835e-06, "loss": 0.0008, "step": 144550 }, { "epoch": 2.365376748752352, "grad_norm": 0.024703320115804672, "learning_rate": 4.280910064147009e-06, "loss": 0.0007, "step": 144560 }, { "epoch": 2.365540374703428, "grad_norm": 0.03048924170434475, "learning_rate": 4.280203540725811e-06, "loss": 0.0003, "step": 144570 }, { "epoch": 2.365704000654504, "grad_norm": 0.06608810275793076, "learning_rate": 4.279497031980654e-06, "loss": 0.0014, "step": 144580 }, { "epoch": 2.3658676266055796, "grad_norm": 0.13107582926750183, "learning_rate": 4.278790537925938e-06, "loss": 0.0009, "step": 144590 }, { "epoch": 2.3660312525566556, "grad_norm": 0.03750166669487953, "learning_rate": 4.278084058576071e-06, "loss": 0.0009, "step": 144600 }, { "epoch": 2.366194878507731, "grad_norm": 0.08172734826803207, "learning_rate": 4.277377593945454e-06, "loss": 0.0011, "step": 144610 }, { "epoch": 2.366358504458807, "grad_norm": 0.08420170843601227, "learning_rate": 4.276671144048495e-06, "loss": 0.0008, "step": 144620 }, { "epoch": 2.366522130409883, "grad_norm": 0.01946049928665161, "learning_rate": 4.275964708899594e-06, "loss": 0.0005, "step": 144630 }, { "epoch": 2.3666857563609587, "grad_norm": 0.05386173725128174, "learning_rate": 4.27525828851316e-06, "loss": 0.0006, "step": 144640 }, { "epoch": 2.3668493823120347, "grad_norm": 0.042074743658304214, "learning_rate": 4.2745518829035905e-06, "loss": 0.0011, "step": 144650 }, { "epoch": 2.3670130082631107, "grad_norm": 0.024461813271045685, "learning_rate": 4.273845492085293e-06, "loss": 0.0004, "step": 144660 }, { "epoch": 2.3671766342141862, "grad_norm": 0.004975370597094297, "learning_rate": 4.273139116072667e-06, "loss": 0.0005, "step": 144670 }, { "epoch": 2.3673402601652622, "grad_norm": 0.034304287284612656, "learning_rate": 4.272432754880117e-06, "loss": 0.0011, "step": 144680 }, { "epoch": 2.367503886116338, "grad_norm": 0.032842256128787994, "learning_rate": 4.2717264085220435e-06, "loss": 0.0008, "step": 144690 }, { "epoch": 2.3676675120674138, "grad_norm": 0.035438548773527145, "learning_rate": 4.27102007701285e-06, "loss": 0.0007, "step": 144700 }, { "epoch": 2.3678311380184898, "grad_norm": 0.019366448745131493, "learning_rate": 4.270313760366936e-06, "loss": 0.001, "step": 144710 }, { "epoch": 2.3679947639695658, "grad_norm": 0.01848200522363186, "learning_rate": 4.269607458598705e-06, "loss": 0.0008, "step": 144720 }, { "epoch": 2.3681583899206413, "grad_norm": 0.03850442171096802, "learning_rate": 4.2689011717225555e-06, "loss": 0.0008, "step": 144730 }, { "epoch": 2.3683220158717173, "grad_norm": 0.002571249380707741, "learning_rate": 4.268194899752891e-06, "loss": 0.0007, "step": 144740 }, { "epoch": 2.368485641822793, "grad_norm": 0.016433047130703926, "learning_rate": 4.267488642704109e-06, "loss": 0.0017, "step": 144750 }, { "epoch": 2.368649267773869, "grad_norm": 0.05571278557181358, "learning_rate": 4.2667824005906116e-06, "loss": 0.0004, "step": 144760 }, { "epoch": 2.368812893724945, "grad_norm": 0.025368018075823784, "learning_rate": 4.266076173426797e-06, "loss": 0.0009, "step": 144770 }, { "epoch": 2.3689765196760204, "grad_norm": 0.015509577468037605, "learning_rate": 4.265369961227065e-06, "loss": 0.0009, "step": 144780 }, { "epoch": 2.3691401456270964, "grad_norm": 0.07853769510984421, "learning_rate": 4.2646637640058135e-06, "loss": 0.0007, "step": 144790 }, { "epoch": 2.3693037715781724, "grad_norm": 0.03716465085744858, "learning_rate": 4.2639575817774445e-06, "loss": 0.0006, "step": 144800 }, { "epoch": 2.369467397529248, "grad_norm": 0.04198264703154564, "learning_rate": 4.263251414556353e-06, "loss": 0.0007, "step": 144810 }, { "epoch": 2.369631023480324, "grad_norm": 0.00967201218008995, "learning_rate": 4.262545262356939e-06, "loss": 0.001, "step": 144820 }, { "epoch": 2.3697946494314, "grad_norm": 0.029063280671834946, "learning_rate": 4.2618391251935995e-06, "loss": 0.0007, "step": 144830 }, { "epoch": 2.3699582753824755, "grad_norm": 0.012776548974215984, "learning_rate": 4.261133003080733e-06, "loss": 0.0007, "step": 144840 }, { "epoch": 2.3701219013335515, "grad_norm": 0.01727365516126156, "learning_rate": 4.260426896032736e-06, "loss": 0.0009, "step": 144850 }, { "epoch": 2.3702855272846275, "grad_norm": 0.07145776599645615, "learning_rate": 4.259720804064007e-06, "loss": 0.0005, "step": 144860 }, { "epoch": 2.370449153235703, "grad_norm": 0.030325328931212425, "learning_rate": 4.259014727188939e-06, "loss": 0.0012, "step": 144870 }, { "epoch": 2.370612779186779, "grad_norm": 0.07743062824010849, "learning_rate": 4.258308665421932e-06, "loss": 0.0009, "step": 144880 }, { "epoch": 2.370776405137855, "grad_norm": 0.06581143289804459, "learning_rate": 4.2576026187773785e-06, "loss": 0.0005, "step": 144890 }, { "epoch": 2.3709400310889306, "grad_norm": 0.040732938796281815, "learning_rate": 4.256896587269679e-06, "loss": 0.0007, "step": 144900 }, { "epoch": 2.3711036570400066, "grad_norm": 0.05045606940984726, "learning_rate": 4.256190570913224e-06, "loss": 0.0009, "step": 144910 }, { "epoch": 2.3712672829910826, "grad_norm": 0.048917192965745926, "learning_rate": 4.255484569722412e-06, "loss": 0.0008, "step": 144920 }, { "epoch": 2.371430908942158, "grad_norm": 0.041428737342357635, "learning_rate": 4.254778583711635e-06, "loss": 0.0012, "step": 144930 }, { "epoch": 2.371594534893234, "grad_norm": 0.017417684197425842, "learning_rate": 4.254072612895291e-06, "loss": 0.0012, "step": 144940 }, { "epoch": 2.37175816084431, "grad_norm": 0.048332568258047104, "learning_rate": 4.2533666572877696e-06, "loss": 0.0005, "step": 144950 }, { "epoch": 2.3719217867953857, "grad_norm": 0.04921181872487068, "learning_rate": 4.252660716903469e-06, "loss": 0.0008, "step": 144960 }, { "epoch": 2.3720854127464617, "grad_norm": 0.019533947110176086, "learning_rate": 4.251954791756779e-06, "loss": 0.0016, "step": 144970 }, { "epoch": 2.3722490386975377, "grad_norm": 0.018026040866971016, "learning_rate": 4.251248881862096e-06, "loss": 0.0009, "step": 144980 }, { "epoch": 2.3724126646486132, "grad_norm": 0.018946943804621696, "learning_rate": 4.250542987233812e-06, "loss": 0.0008, "step": 144990 }, { "epoch": 2.372576290599689, "grad_norm": 0.048198964446783066, "learning_rate": 4.249837107886318e-06, "loss": 0.0007, "step": 145000 }, { "epoch": 2.3727399165507648, "grad_norm": 0.010190469212830067, "learning_rate": 4.249131243834008e-06, "loss": 0.0007, "step": 145010 }, { "epoch": 2.3729035425018408, "grad_norm": 0.08974719792604446, "learning_rate": 4.248425395091273e-06, "loss": 0.0011, "step": 145020 }, { "epoch": 2.3730671684529168, "grad_norm": 0.015262089669704437, "learning_rate": 4.2477195616725045e-06, "loss": 0.0007, "step": 145030 }, { "epoch": 2.3732307944039923, "grad_norm": 0.0033396335784345865, "learning_rate": 4.247013743592095e-06, "loss": 0.0007, "step": 145040 }, { "epoch": 2.3733944203550683, "grad_norm": 0.004594051279127598, "learning_rate": 4.246307940864435e-06, "loss": 0.001, "step": 145050 }, { "epoch": 2.3735580463061443, "grad_norm": 0.022924404591321945, "learning_rate": 4.245602153503915e-06, "loss": 0.0014, "step": 145060 }, { "epoch": 2.37372167225722, "grad_norm": 0.010580492205917835, "learning_rate": 4.244896381524924e-06, "loss": 0.0006, "step": 145070 }, { "epoch": 2.373885298208296, "grad_norm": 0.010799615643918514, "learning_rate": 4.244190624941857e-06, "loss": 0.0011, "step": 145080 }, { "epoch": 2.374048924159372, "grad_norm": 0.041197270154953, "learning_rate": 4.243484883769097e-06, "loss": 0.0008, "step": 145090 }, { "epoch": 2.3742125501104474, "grad_norm": 0.048394978046417236, "learning_rate": 4.24277915802104e-06, "loss": 0.0007, "step": 145100 }, { "epoch": 2.3743761760615234, "grad_norm": 0.12991639971733093, "learning_rate": 4.242073447712069e-06, "loss": 0.002, "step": 145110 }, { "epoch": 2.374539802012599, "grad_norm": 0.07027889788150787, "learning_rate": 4.241367752856578e-06, "loss": 0.0006, "step": 145120 }, { "epoch": 2.374703427963675, "grad_norm": 0.004830446559935808, "learning_rate": 4.24066207346895e-06, "loss": 0.0007, "step": 145130 }, { "epoch": 2.374867053914751, "grad_norm": 0.017877083271741867, "learning_rate": 4.23995640956358e-06, "loss": 0.0011, "step": 145140 }, { "epoch": 2.3750306798658265, "grad_norm": 0.06057748943567276, "learning_rate": 4.23925076115485e-06, "loss": 0.0006, "step": 145150 }, { "epoch": 2.3751943058169025, "grad_norm": 0.04554426670074463, "learning_rate": 4.238545128257154e-06, "loss": 0.001, "step": 145160 }, { "epoch": 2.3753579317679785, "grad_norm": 0.007033512927591801, "learning_rate": 4.23783951088487e-06, "loss": 0.0006, "step": 145170 }, { "epoch": 2.375521557719054, "grad_norm": 0.04569651931524277, "learning_rate": 4.237133909052394e-06, "loss": 0.0007, "step": 145180 }, { "epoch": 2.37568518367013, "grad_norm": 0.01821427047252655, "learning_rate": 4.236428322774106e-06, "loss": 0.0006, "step": 145190 }, { "epoch": 2.375848809621206, "grad_norm": 0.16118383407592773, "learning_rate": 4.235722752064398e-06, "loss": 0.0008, "step": 145200 }, { "epoch": 2.3760124355722816, "grad_norm": 0.024583226069808006, "learning_rate": 4.23501719693765e-06, "loss": 0.0005, "step": 145210 }, { "epoch": 2.3761760615233576, "grad_norm": 0.14106538891792297, "learning_rate": 4.2343116574082535e-06, "loss": 0.001, "step": 145220 }, { "epoch": 2.3763396874744336, "grad_norm": 0.019151834771037102, "learning_rate": 4.2336061334905885e-06, "loss": 0.0008, "step": 145230 }, { "epoch": 2.376503313425509, "grad_norm": 0.014729492366313934, "learning_rate": 4.232900625199046e-06, "loss": 0.0006, "step": 145240 }, { "epoch": 2.376666939376585, "grad_norm": 0.0655680000782013, "learning_rate": 4.232195132548004e-06, "loss": 0.0014, "step": 145250 }, { "epoch": 2.376830565327661, "grad_norm": 0.019720597192645073, "learning_rate": 4.2314896555518525e-06, "loss": 0.0005, "step": 145260 }, { "epoch": 2.3769941912787367, "grad_norm": 0.09533333033323288, "learning_rate": 4.230784194224972e-06, "loss": 0.0005, "step": 145270 }, { "epoch": 2.3771578172298127, "grad_norm": 0.025408077985048294, "learning_rate": 4.230078748581749e-06, "loss": 0.0014, "step": 145280 }, { "epoch": 2.3773214431808887, "grad_norm": 0.026495005935430527, "learning_rate": 4.2293733186365636e-06, "loss": 0.0007, "step": 145290 }, { "epoch": 2.3774850691319642, "grad_norm": 0.002453201450407505, "learning_rate": 4.228667904403803e-06, "loss": 0.0007, "step": 145300 }, { "epoch": 2.37764869508304, "grad_norm": 0.049840908497571945, "learning_rate": 4.2279625058978454e-06, "loss": 0.0008, "step": 145310 }, { "epoch": 2.377812321034116, "grad_norm": 0.141469806432724, "learning_rate": 4.227257123133078e-06, "loss": 0.0011, "step": 145320 }, { "epoch": 2.3779759469851918, "grad_norm": 0.10426681488752365, "learning_rate": 4.226551756123879e-06, "loss": 0.0012, "step": 145330 }, { "epoch": 2.3781395729362678, "grad_norm": 0.011357774958014488, "learning_rate": 4.225846404884633e-06, "loss": 0.0016, "step": 145340 }, { "epoch": 2.3783031988873438, "grad_norm": 0.03168405964970589, "learning_rate": 4.2251410694297195e-06, "loss": 0.0013, "step": 145350 }, { "epoch": 2.3784668248384193, "grad_norm": 0.05868662893772125, "learning_rate": 4.224435749773522e-06, "loss": 0.0018, "step": 145360 }, { "epoch": 2.3786304507894953, "grad_norm": 0.005398761946707964, "learning_rate": 4.223730445930418e-06, "loss": 0.0005, "step": 145370 }, { "epoch": 2.378794076740571, "grad_norm": 0.036631640046834946, "learning_rate": 4.223025157914792e-06, "loss": 0.0006, "step": 145380 }, { "epoch": 2.378957702691647, "grad_norm": 0.04446527734398842, "learning_rate": 4.222319885741021e-06, "loss": 0.0004, "step": 145390 }, { "epoch": 2.379121328642723, "grad_norm": 0.014052147977054119, "learning_rate": 4.221614629423487e-06, "loss": 0.0006, "step": 145400 }, { "epoch": 2.3792849545937984, "grad_norm": 0.001772678573615849, "learning_rate": 4.2209093889765674e-06, "loss": 0.0007, "step": 145410 }, { "epoch": 2.3794485805448744, "grad_norm": 0.047495026141405106, "learning_rate": 4.220204164414644e-06, "loss": 0.001, "step": 145420 }, { "epoch": 2.3796122064959504, "grad_norm": 0.017590349540114403, "learning_rate": 4.219498955752093e-06, "loss": 0.0006, "step": 145430 }, { "epoch": 2.379775832447026, "grad_norm": 0.04710206389427185, "learning_rate": 4.218793763003296e-06, "loss": 0.0005, "step": 145440 }, { "epoch": 2.379939458398102, "grad_norm": 0.02323029935359955, "learning_rate": 4.218088586182629e-06, "loss": 0.0005, "step": 145450 }, { "epoch": 2.380103084349178, "grad_norm": 0.027849089354276657, "learning_rate": 4.217383425304472e-06, "loss": 0.001, "step": 145460 }, { "epoch": 2.3802667103002535, "grad_norm": 0.061781324446201324, "learning_rate": 4.216678280383199e-06, "loss": 0.001, "step": 145470 }, { "epoch": 2.3804303362513295, "grad_norm": 0.004714277107268572, "learning_rate": 4.215973151433193e-06, "loss": 0.0006, "step": 145480 }, { "epoch": 2.3805939622024055, "grad_norm": 0.03003733977675438, "learning_rate": 4.215268038468825e-06, "loss": 0.0008, "step": 145490 }, { "epoch": 2.380757588153481, "grad_norm": 0.06510300189256668, "learning_rate": 4.2145629415044754e-06, "loss": 0.0012, "step": 145500 }, { "epoch": 2.380921214104557, "grad_norm": 0.044987961649894714, "learning_rate": 4.213857860554517e-06, "loss": 0.0008, "step": 145510 }, { "epoch": 2.3810848400556326, "grad_norm": 0.047816675156354904, "learning_rate": 4.213152795633332e-06, "loss": 0.0011, "step": 145520 }, { "epoch": 2.3812484660067086, "grad_norm": 0.10673844069242477, "learning_rate": 4.21244774675529e-06, "loss": 0.0027, "step": 145530 }, { "epoch": 2.3814120919577846, "grad_norm": 0.026577798649668694, "learning_rate": 4.2117427139347696e-06, "loss": 0.0007, "step": 145540 }, { "epoch": 2.38157571790886, "grad_norm": 0.09556518495082855, "learning_rate": 4.211037697186145e-06, "loss": 0.0013, "step": 145550 }, { "epoch": 2.381739343859936, "grad_norm": 0.03400268033146858, "learning_rate": 4.21033269652379e-06, "loss": 0.0008, "step": 145560 }, { "epoch": 2.381902969811012, "grad_norm": 0.13966397941112518, "learning_rate": 4.209627711962079e-06, "loss": 0.0009, "step": 145570 }, { "epoch": 2.3820665957620877, "grad_norm": 0.0410550981760025, "learning_rate": 4.208922743515389e-06, "loss": 0.0007, "step": 145580 }, { "epoch": 2.3822302217131637, "grad_norm": 0.010867778211832047, "learning_rate": 4.208217791198089e-06, "loss": 0.0012, "step": 145590 }, { "epoch": 2.3823938476642397, "grad_norm": 0.026449590921401978, "learning_rate": 4.207512855024557e-06, "loss": 0.0006, "step": 145600 }, { "epoch": 2.3825574736153152, "grad_norm": 0.05850580707192421, "learning_rate": 4.206807935009163e-06, "loss": 0.0016, "step": 145610 }, { "epoch": 2.382721099566391, "grad_norm": 0.00494764931499958, "learning_rate": 4.206103031166281e-06, "loss": 0.0005, "step": 145620 }, { "epoch": 2.382884725517467, "grad_norm": 0.029482195153832436, "learning_rate": 4.205398143510282e-06, "loss": 0.0019, "step": 145630 }, { "epoch": 2.3830483514685428, "grad_norm": 0.0028361172880977392, "learning_rate": 4.20469327205554e-06, "loss": 0.0004, "step": 145640 }, { "epoch": 2.3832119774196188, "grad_norm": 0.0151938796043396, "learning_rate": 4.203988416816426e-06, "loss": 0.001, "step": 145650 }, { "epoch": 2.3833756033706948, "grad_norm": 0.09608354419469833, "learning_rate": 4.20328357780731e-06, "loss": 0.0012, "step": 145660 }, { "epoch": 2.3835392293217703, "grad_norm": 0.002281482797116041, "learning_rate": 4.202578755042565e-06, "loss": 0.0006, "step": 145670 }, { "epoch": 2.3837028552728463, "grad_norm": 0.017058784142136574, "learning_rate": 4.201873948536561e-06, "loss": 0.0004, "step": 145680 }, { "epoch": 2.3838664812239223, "grad_norm": 0.007544373627752066, "learning_rate": 4.201169158303668e-06, "loss": 0.0017, "step": 145690 }, { "epoch": 2.384030107174998, "grad_norm": 0.06902377307415009, "learning_rate": 4.200464384358257e-06, "loss": 0.0006, "step": 145700 }, { "epoch": 2.384193733126074, "grad_norm": 0.025375962257385254, "learning_rate": 4.199759626714697e-06, "loss": 0.0009, "step": 145710 }, { "epoch": 2.38435735907715, "grad_norm": 0.02460326999425888, "learning_rate": 4.199054885387359e-06, "loss": 0.0006, "step": 145720 }, { "epoch": 2.3845209850282254, "grad_norm": 0.027731290087103844, "learning_rate": 4.198350160390608e-06, "loss": 0.001, "step": 145730 }, { "epoch": 2.3846846109793014, "grad_norm": 0.002504519186913967, "learning_rate": 4.197645451738819e-06, "loss": 0.0015, "step": 145740 }, { "epoch": 2.3848482369303774, "grad_norm": 0.0060051740147173405, "learning_rate": 4.196940759446353e-06, "loss": 0.0011, "step": 145750 }, { "epoch": 2.385011862881453, "grad_norm": 0.009846005588769913, "learning_rate": 4.196236083527585e-06, "loss": 0.0015, "step": 145760 }, { "epoch": 2.385175488832529, "grad_norm": 0.0388077087700367, "learning_rate": 4.195531423996877e-06, "loss": 0.0012, "step": 145770 }, { "epoch": 2.3853391147836045, "grad_norm": 0.02221035771071911, "learning_rate": 4.194826780868602e-06, "loss": 0.0007, "step": 145780 }, { "epoch": 2.3855027407346805, "grad_norm": 0.001092107268050313, "learning_rate": 4.194122154157121e-06, "loss": 0.0005, "step": 145790 }, { "epoch": 2.3856663666857565, "grad_norm": 0.06756853312253952, "learning_rate": 4.193417543876806e-06, "loss": 0.0008, "step": 145800 }, { "epoch": 2.385829992636832, "grad_norm": 0.0075613451190292835, "learning_rate": 4.19271295004202e-06, "loss": 0.0007, "step": 145810 }, { "epoch": 2.385993618587908, "grad_norm": 0.04325985535979271, "learning_rate": 4.192008372667133e-06, "loss": 0.0006, "step": 145820 }, { "epoch": 2.386157244538984, "grad_norm": 0.016921471804380417, "learning_rate": 4.191303811766504e-06, "loss": 0.0007, "step": 145830 }, { "epoch": 2.3863208704900596, "grad_norm": 0.0020079819951206446, "learning_rate": 4.190599267354507e-06, "loss": 0.001, "step": 145840 }, { "epoch": 2.3864844964411356, "grad_norm": 0.0019934081938117743, "learning_rate": 4.189894739445499e-06, "loss": 0.0007, "step": 145850 }, { "epoch": 2.3866481223922116, "grad_norm": 0.03786010667681694, "learning_rate": 4.189190228053851e-06, "loss": 0.0016, "step": 145860 }, { "epoch": 2.386811748343287, "grad_norm": 0.11419138312339783, "learning_rate": 4.188485733193922e-06, "loss": 0.0015, "step": 145870 }, { "epoch": 2.386975374294363, "grad_norm": 0.06546318531036377, "learning_rate": 4.1877812548800816e-06, "loss": 0.0006, "step": 145880 }, { "epoch": 2.3871390002454387, "grad_norm": 0.05826917290687561, "learning_rate": 4.1870767931266874e-06, "loss": 0.0011, "step": 145890 }, { "epoch": 2.3873026261965147, "grad_norm": 0.03783240541815758, "learning_rate": 4.186372347948109e-06, "loss": 0.0015, "step": 145900 }, { "epoch": 2.3874662521475907, "grad_norm": 0.03633413836359978, "learning_rate": 4.185667919358705e-06, "loss": 0.0009, "step": 145910 }, { "epoch": 2.3876298780986662, "grad_norm": 0.0844312310218811, "learning_rate": 4.18496350737284e-06, "loss": 0.0008, "step": 145920 }, { "epoch": 2.387793504049742, "grad_norm": 0.09244171530008316, "learning_rate": 4.184259112004875e-06, "loss": 0.0005, "step": 145930 }, { "epoch": 2.387957130000818, "grad_norm": 0.062230419367551804, "learning_rate": 4.1835547332691744e-06, "loss": 0.0009, "step": 145940 }, { "epoch": 2.3881207559518938, "grad_norm": 0.0029631690122187138, "learning_rate": 4.182850371180097e-06, "loss": 0.0011, "step": 145950 }, { "epoch": 2.3882843819029698, "grad_norm": 0.023532673716545105, "learning_rate": 4.182146025752007e-06, "loss": 0.0015, "step": 145960 }, { "epoch": 2.3884480078540458, "grad_norm": 0.028864240273833275, "learning_rate": 4.1814416969992625e-06, "loss": 0.0007, "step": 145970 }, { "epoch": 2.3886116338051213, "grad_norm": 0.04710301756858826, "learning_rate": 4.180737384936227e-06, "loss": 0.0012, "step": 145980 }, { "epoch": 2.3887752597561973, "grad_norm": 0.008640728890895844, "learning_rate": 4.180033089577259e-06, "loss": 0.0007, "step": 145990 }, { "epoch": 2.3889388857072733, "grad_norm": 0.02905939519405365, "learning_rate": 4.1793288109367185e-06, "loss": 0.0008, "step": 146000 }, { "epoch": 2.389102511658349, "grad_norm": 0.03151736781001091, "learning_rate": 4.178624549028965e-06, "loss": 0.0007, "step": 146010 }, { "epoch": 2.389266137609425, "grad_norm": 0.0012737010838463902, "learning_rate": 4.17792030386836e-06, "loss": 0.0011, "step": 146020 }, { "epoch": 2.389429763560501, "grad_norm": 0.024020854383707047, "learning_rate": 4.17721607546926e-06, "loss": 0.0008, "step": 146030 }, { "epoch": 2.3895933895115764, "grad_norm": 0.04025198146700859, "learning_rate": 4.176511863846024e-06, "loss": 0.0005, "step": 146040 }, { "epoch": 2.3897570154626524, "grad_norm": 0.05993346497416496, "learning_rate": 4.175807669013011e-06, "loss": 0.0007, "step": 146050 }, { "epoch": 2.3899206414137284, "grad_norm": 0.06258746236562729, "learning_rate": 4.17510349098458e-06, "loss": 0.001, "step": 146060 }, { "epoch": 2.390084267364804, "grad_norm": 0.02836768887937069, "learning_rate": 4.174399329775086e-06, "loss": 0.0006, "step": 146070 }, { "epoch": 2.39024789331588, "grad_norm": 0.06905461847782135, "learning_rate": 4.1736951853988875e-06, "loss": 0.0006, "step": 146080 }, { "epoch": 2.390411519266956, "grad_norm": 0.00576144689694047, "learning_rate": 4.172991057870341e-06, "loss": 0.0009, "step": 146090 }, { "epoch": 2.3905751452180315, "grad_norm": 0.043815746903419495, "learning_rate": 4.1722869472038055e-06, "loss": 0.0006, "step": 146100 }, { "epoch": 2.3907387711691075, "grad_norm": 0.04308261349797249, "learning_rate": 4.171582853413633e-06, "loss": 0.0011, "step": 146110 }, { "epoch": 2.3909023971201835, "grad_norm": 0.030907826498150826, "learning_rate": 4.170878776514183e-06, "loss": 0.0005, "step": 146120 }, { "epoch": 2.391066023071259, "grad_norm": 0.02836962416768074, "learning_rate": 4.1701747165198085e-06, "loss": 0.0008, "step": 146130 }, { "epoch": 2.391229649022335, "grad_norm": 0.047660116106271744, "learning_rate": 4.169470673444867e-06, "loss": 0.0025, "step": 146140 }, { "epoch": 2.391393274973411, "grad_norm": 0.00034789624623954296, "learning_rate": 4.168766647303711e-06, "loss": 0.001, "step": 146150 }, { "epoch": 2.3915569009244866, "grad_norm": 0.005692503415048122, "learning_rate": 4.168062638110697e-06, "loss": 0.0007, "step": 146160 }, { "epoch": 2.3917205268755626, "grad_norm": 0.05811876431107521, "learning_rate": 4.167358645880177e-06, "loss": 0.0012, "step": 146170 }, { "epoch": 2.391884152826638, "grad_norm": 0.02168862707912922, "learning_rate": 4.1666546706265074e-06, "loss": 0.0005, "step": 146180 }, { "epoch": 2.392047778777714, "grad_norm": 0.0021373280324041843, "learning_rate": 4.16595071236404e-06, "loss": 0.0008, "step": 146190 }, { "epoch": 2.39221140472879, "grad_norm": 0.08672746270895004, "learning_rate": 4.165246771107128e-06, "loss": 0.002, "step": 146200 }, { "epoch": 2.3923750306798657, "grad_norm": 0.009250969626009464, "learning_rate": 4.164542846870124e-06, "loss": 0.0009, "step": 146210 }, { "epoch": 2.3925386566309417, "grad_norm": 0.015176656655967236, "learning_rate": 4.163838939667382e-06, "loss": 0.0009, "step": 146220 }, { "epoch": 2.3927022825820177, "grad_norm": 0.010646921582520008, "learning_rate": 4.1631350495132515e-06, "loss": 0.0006, "step": 146230 }, { "epoch": 2.3928659085330932, "grad_norm": 0.022036980837583542, "learning_rate": 4.162431176422087e-06, "loss": 0.0009, "step": 146240 }, { "epoch": 2.393029534484169, "grad_norm": 0.043176375329494476, "learning_rate": 4.161727320408237e-06, "loss": 0.001, "step": 146250 }, { "epoch": 2.393193160435245, "grad_norm": 0.033912207931280136, "learning_rate": 4.161023481486056e-06, "loss": 0.0007, "step": 146260 }, { "epoch": 2.3933567863863208, "grad_norm": 0.04435432329773903, "learning_rate": 4.1603196596698905e-06, "loss": 0.0007, "step": 146270 }, { "epoch": 2.3935204123373968, "grad_norm": 0.04255463555455208, "learning_rate": 4.159615854974095e-06, "loss": 0.001, "step": 146280 }, { "epoch": 2.3936840382884723, "grad_norm": 0.003687328426167369, "learning_rate": 4.158912067413016e-06, "loss": 0.0005, "step": 146290 }, { "epoch": 2.3938476642395483, "grad_norm": 0.002851062221452594, "learning_rate": 4.158208297001006e-06, "loss": 0.0006, "step": 146300 }, { "epoch": 2.3940112901906243, "grad_norm": 0.0725243091583252, "learning_rate": 4.1575045437524105e-06, "loss": 0.0015, "step": 146310 }, { "epoch": 2.3941749161417, "grad_norm": 0.11315174400806427, "learning_rate": 4.1568008076815835e-06, "loss": 0.0009, "step": 146320 }, { "epoch": 2.394338542092776, "grad_norm": 0.00585919339209795, "learning_rate": 4.15609708880287e-06, "loss": 0.0008, "step": 146330 }, { "epoch": 2.394502168043852, "grad_norm": 0.007595815230160952, "learning_rate": 4.155393387130618e-06, "loss": 0.0006, "step": 146340 }, { "epoch": 2.3946657939949274, "grad_norm": 0.04253964498639107, "learning_rate": 4.154689702679178e-06, "loss": 0.0004, "step": 146350 }, { "epoch": 2.3948294199460034, "grad_norm": 0.02070918306708336, "learning_rate": 4.1539860354628965e-06, "loss": 0.0006, "step": 146360 }, { "epoch": 2.3949930458970794, "grad_norm": 0.057805418968200684, "learning_rate": 4.153282385496118e-06, "loss": 0.001, "step": 146370 }, { "epoch": 2.395156671848155, "grad_norm": 0.013851486146450043, "learning_rate": 4.1525787527931945e-06, "loss": 0.0008, "step": 146380 }, { "epoch": 2.395320297799231, "grad_norm": 0.037726547569036484, "learning_rate": 4.151875137368466e-06, "loss": 0.0006, "step": 146390 }, { "epoch": 2.395483923750307, "grad_norm": 0.025154318660497665, "learning_rate": 4.151171539236286e-06, "loss": 0.0007, "step": 146400 }, { "epoch": 2.3956475497013825, "grad_norm": 0.08665670454502106, "learning_rate": 4.150467958410993e-06, "loss": 0.0011, "step": 146410 }, { "epoch": 2.3958111756524585, "grad_norm": 0.09421893954277039, "learning_rate": 4.149764394906938e-06, "loss": 0.001, "step": 146420 }, { "epoch": 2.3959748016035345, "grad_norm": 0.0024501371663063765, "learning_rate": 4.149060848738462e-06, "loss": 0.001, "step": 146430 }, { "epoch": 2.39613842755461, "grad_norm": 0.03700675442814827, "learning_rate": 4.148357319919915e-06, "loss": 0.0011, "step": 146440 }, { "epoch": 2.396302053505686, "grad_norm": 0.03706556558609009, "learning_rate": 4.147653808465633e-06, "loss": 0.0008, "step": 146450 }, { "epoch": 2.396465679456762, "grad_norm": 0.06507950276136398, "learning_rate": 4.146950314389969e-06, "loss": 0.0009, "step": 146460 }, { "epoch": 2.3966293054078376, "grad_norm": 0.027533454820513725, "learning_rate": 4.14624683770726e-06, "loss": 0.0009, "step": 146470 }, { "epoch": 2.3967929313589136, "grad_norm": 0.01966329663991928, "learning_rate": 4.145543378431855e-06, "loss": 0.0009, "step": 146480 }, { "epoch": 2.3969565573099896, "grad_norm": 0.0011110982159152627, "learning_rate": 4.14483993657809e-06, "loss": 0.0003, "step": 146490 }, { "epoch": 2.397120183261065, "grad_norm": 0.03127239644527435, "learning_rate": 4.144136512160315e-06, "loss": 0.0008, "step": 146500 }, { "epoch": 2.397283809212141, "grad_norm": 0.02774646133184433, "learning_rate": 4.143433105192865e-06, "loss": 0.0007, "step": 146510 }, { "epoch": 2.397447435163217, "grad_norm": 0.03466450050473213, "learning_rate": 4.142729715690089e-06, "loss": 0.0009, "step": 146520 }, { "epoch": 2.3976110611142927, "grad_norm": 0.011405671946704388, "learning_rate": 4.1420263436663226e-06, "loss": 0.0011, "step": 146530 }, { "epoch": 2.3977746870653687, "grad_norm": 0.1059623509645462, "learning_rate": 4.141322989135912e-06, "loss": 0.0006, "step": 146540 }, { "epoch": 2.3979383130164442, "grad_norm": 0.07460809499025345, "learning_rate": 4.140619652113193e-06, "loss": 0.0012, "step": 146550 }, { "epoch": 2.39810193896752, "grad_norm": 0.11788494139909744, "learning_rate": 4.139916332612509e-06, "loss": 0.0012, "step": 146560 }, { "epoch": 2.398265564918596, "grad_norm": 0.010466134175658226, "learning_rate": 4.139213030648199e-06, "loss": 0.0006, "step": 146570 }, { "epoch": 2.3984291908696718, "grad_norm": 0.08556206524372101, "learning_rate": 4.138509746234604e-06, "loss": 0.0013, "step": 146580 }, { "epoch": 2.3985928168207478, "grad_norm": 0.05156935751438141, "learning_rate": 4.137806479386062e-06, "loss": 0.0004, "step": 146590 }, { "epoch": 2.3987564427718238, "grad_norm": 0.10737697780132294, "learning_rate": 4.137103230116914e-06, "loss": 0.0017, "step": 146600 }, { "epoch": 2.3989200687228993, "grad_norm": 0.015113960020244122, "learning_rate": 4.136399998441496e-06, "loss": 0.0019, "step": 146610 }, { "epoch": 2.3990836946739753, "grad_norm": 0.08254607021808624, "learning_rate": 4.135696784374148e-06, "loss": 0.0014, "step": 146620 }, { "epoch": 2.3992473206250513, "grad_norm": 0.01696351356804371, "learning_rate": 4.134993587929207e-06, "loss": 0.001, "step": 146630 }, { "epoch": 2.399410946576127, "grad_norm": 0.013753910548985004, "learning_rate": 4.134290409121012e-06, "loss": 0.0006, "step": 146640 }, { "epoch": 2.399574572527203, "grad_norm": 0.0022576337214559317, "learning_rate": 4.133587247963898e-06, "loss": 0.0006, "step": 146650 }, { "epoch": 2.3997381984782784, "grad_norm": 0.007551929913461208, "learning_rate": 4.1328841044722046e-06, "loss": 0.0007, "step": 146660 }, { "epoch": 2.3999018244293544, "grad_norm": 0.0021957007702440023, "learning_rate": 4.132180978660265e-06, "loss": 0.0004, "step": 146670 }, { "epoch": 2.4000654503804304, "grad_norm": 0.028378838673233986, "learning_rate": 4.131477870542419e-06, "loss": 0.0008, "step": 146680 }, { "epoch": 2.400229076331506, "grad_norm": 0.02674955502152443, "learning_rate": 4.130774780132999e-06, "loss": 0.0011, "step": 146690 }, { "epoch": 2.400392702282582, "grad_norm": 0.1652206927537918, "learning_rate": 4.130071707446344e-06, "loss": 0.0009, "step": 146700 }, { "epoch": 2.400556328233658, "grad_norm": 0.03368297219276428, "learning_rate": 4.129368652496785e-06, "loss": 0.0008, "step": 146710 }, { "epoch": 2.4007199541847335, "grad_norm": 0.0645778477191925, "learning_rate": 4.128665615298661e-06, "loss": 0.001, "step": 146720 }, { "epoch": 2.4008835801358095, "grad_norm": 0.0009938528528437018, "learning_rate": 4.127962595866301e-06, "loss": 0.001, "step": 146730 }, { "epoch": 2.4010472060868855, "grad_norm": 0.06187603622674942, "learning_rate": 4.127259594214044e-06, "loss": 0.0009, "step": 146740 }, { "epoch": 2.401210832037961, "grad_norm": 0.009182535111904144, "learning_rate": 4.126556610356222e-06, "loss": 0.001, "step": 146750 }, { "epoch": 2.401374457989037, "grad_norm": 0.07387594878673553, "learning_rate": 4.125853644307167e-06, "loss": 0.0008, "step": 146760 }, { "epoch": 2.401538083940113, "grad_norm": 0.04637860879302025, "learning_rate": 4.1251506960812125e-06, "loss": 0.0017, "step": 146770 }, { "epoch": 2.4017017098911886, "grad_norm": 0.4540545642375946, "learning_rate": 4.124447765692693e-06, "loss": 0.0009, "step": 146780 }, { "epoch": 2.4018653358422646, "grad_norm": 0.050076305866241455, "learning_rate": 4.123744853155936e-06, "loss": 0.0006, "step": 146790 }, { "epoch": 2.4020289617933406, "grad_norm": 0.04234588146209717, "learning_rate": 4.12304195848528e-06, "loss": 0.0008, "step": 146800 }, { "epoch": 2.402192587744416, "grad_norm": 0.027492186054587364, "learning_rate": 4.122339081695049e-06, "loss": 0.001, "step": 146810 }, { "epoch": 2.402356213695492, "grad_norm": 0.039715953171253204, "learning_rate": 4.1216362227995796e-06, "loss": 0.0006, "step": 146820 }, { "epoch": 2.402519839646568, "grad_norm": 0.002670279471203685, "learning_rate": 4.1209333818132e-06, "loss": 0.0007, "step": 146830 }, { "epoch": 2.4026834655976437, "grad_norm": 0.08931277692317963, "learning_rate": 4.1202305587502425e-06, "loss": 0.0008, "step": 146840 }, { "epoch": 2.4028470915487197, "grad_norm": 0.09401024878025055, "learning_rate": 4.1195277536250335e-06, "loss": 0.0007, "step": 146850 }, { "epoch": 2.4030107174997957, "grad_norm": 0.022057097405195236, "learning_rate": 4.118824966451906e-06, "loss": 0.0007, "step": 146860 }, { "epoch": 2.403174343450871, "grad_norm": 0.0850737988948822, "learning_rate": 4.118122197245188e-06, "loss": 0.0005, "step": 146870 }, { "epoch": 2.403337969401947, "grad_norm": 0.002083584200590849, "learning_rate": 4.117419446019208e-06, "loss": 0.0007, "step": 146880 }, { "epoch": 2.403501595353023, "grad_norm": 0.019755933433771133, "learning_rate": 4.116716712788295e-06, "loss": 0.0008, "step": 146890 }, { "epoch": 2.4036652213040988, "grad_norm": 0.009106545709073544, "learning_rate": 4.116013997566778e-06, "loss": 0.0009, "step": 146900 }, { "epoch": 2.4038288472551748, "grad_norm": 0.1349049061536789, "learning_rate": 4.115311300368982e-06, "loss": 0.0013, "step": 146910 }, { "epoch": 2.4039924732062508, "grad_norm": 0.00275608291849494, "learning_rate": 4.114608621209238e-06, "loss": 0.0007, "step": 146920 }, { "epoch": 2.4041560991573263, "grad_norm": 0.0350952185690403, "learning_rate": 4.113905960101869e-06, "loss": 0.0011, "step": 146930 }, { "epoch": 2.4043197251084023, "grad_norm": 0.052596043795347214, "learning_rate": 4.1132033170612065e-06, "loss": 0.001, "step": 146940 }, { "epoch": 2.404483351059478, "grad_norm": 0.06780418008565903, "learning_rate": 4.112500692101572e-06, "loss": 0.001, "step": 146950 }, { "epoch": 2.404646977010554, "grad_norm": 0.009219875559210777, "learning_rate": 4.111798085237295e-06, "loss": 0.0005, "step": 146960 }, { "epoch": 2.40481060296163, "grad_norm": 0.02959110215306282, "learning_rate": 4.111095496482699e-06, "loss": 0.0026, "step": 146970 }, { "epoch": 2.4049742289127054, "grad_norm": 0.029625020921230316, "learning_rate": 4.11039292585211e-06, "loss": 0.0009, "step": 146980 }, { "epoch": 2.4051378548637814, "grad_norm": 0.03293364867568016, "learning_rate": 4.109690373359852e-06, "loss": 0.0005, "step": 146990 }, { "epoch": 2.4053014808148574, "grad_norm": 0.0114517230540514, "learning_rate": 4.108987839020252e-06, "loss": 0.0007, "step": 147000 }, { "epoch": 2.405465106765933, "grad_norm": 0.12769147753715515, "learning_rate": 4.10828532284763e-06, "loss": 0.0008, "step": 147010 }, { "epoch": 2.405628732717009, "grad_norm": 0.04716530442237854, "learning_rate": 4.1075828248563145e-06, "loss": 0.0007, "step": 147020 }, { "epoch": 2.405792358668085, "grad_norm": 0.11788532137870789, "learning_rate": 4.106880345060623e-06, "loss": 0.0014, "step": 147030 }, { "epoch": 2.4059559846191605, "grad_norm": 0.03842136263847351, "learning_rate": 4.106177883474885e-06, "loss": 0.0009, "step": 147040 }, { "epoch": 2.4061196105702365, "grad_norm": 0.008221525698900223, "learning_rate": 4.105475440113417e-06, "loss": 0.0012, "step": 147050 }, { "epoch": 2.406283236521312, "grad_norm": 0.03827422484755516, "learning_rate": 4.104773014990546e-06, "loss": 0.001, "step": 147060 }, { "epoch": 2.406446862472388, "grad_norm": 0.037954188883304596, "learning_rate": 4.104070608120589e-06, "loss": 0.0017, "step": 147070 }, { "epoch": 2.406610488423464, "grad_norm": 0.06737605482339859, "learning_rate": 4.103368219517874e-06, "loss": 0.0017, "step": 147080 }, { "epoch": 2.4067741143745396, "grad_norm": 0.025100449100136757, "learning_rate": 4.102665849196716e-06, "loss": 0.0014, "step": 147090 }, { "epoch": 2.4069377403256156, "grad_norm": 0.04748481884598732, "learning_rate": 4.101963497171439e-06, "loss": 0.001, "step": 147100 }, { "epoch": 2.4071013662766916, "grad_norm": 0.03609771281480789, "learning_rate": 4.101261163456362e-06, "loss": 0.0006, "step": 147110 }, { "epoch": 2.407264992227767, "grad_norm": 0.11947447061538696, "learning_rate": 4.100558848065807e-06, "loss": 0.0007, "step": 147120 }, { "epoch": 2.407428618178843, "grad_norm": 0.005800878629088402, "learning_rate": 4.0998565510140895e-06, "loss": 0.0006, "step": 147130 }, { "epoch": 2.407592244129919, "grad_norm": 0.033785052597522736, "learning_rate": 4.099154272315535e-06, "loss": 0.0005, "step": 147140 }, { "epoch": 2.4077558700809947, "grad_norm": 0.007222397718578577, "learning_rate": 4.098452011984455e-06, "loss": 0.0011, "step": 147150 }, { "epoch": 2.4079194960320707, "grad_norm": 0.008655103854835033, "learning_rate": 4.097749770035175e-06, "loss": 0.0008, "step": 147160 }, { "epoch": 2.4080831219831467, "grad_norm": 0.07565563172101974, "learning_rate": 4.097047546482006e-06, "loss": 0.0009, "step": 147170 }, { "epoch": 2.408246747934222, "grad_norm": 0.036072466522455215, "learning_rate": 4.096345341339274e-06, "loss": 0.001, "step": 147180 }, { "epoch": 2.408410373885298, "grad_norm": 0.013102875091135502, "learning_rate": 4.095643154621289e-06, "loss": 0.0016, "step": 147190 }, { "epoch": 2.408573999836374, "grad_norm": 0.059363897889852524, "learning_rate": 4.094940986342373e-06, "loss": 0.001, "step": 147200 }, { "epoch": 2.4087376257874498, "grad_norm": 0.032609451562166214, "learning_rate": 4.094238836516839e-06, "loss": 0.0012, "step": 147210 }, { "epoch": 2.4089012517385258, "grad_norm": 0.015046516433358192, "learning_rate": 4.093536705159005e-06, "loss": 0.0008, "step": 147220 }, { "epoch": 2.4090648776896018, "grad_norm": 0.031644757837057114, "learning_rate": 4.092834592283186e-06, "loss": 0.0003, "step": 147230 }, { "epoch": 2.4092285036406773, "grad_norm": 0.0433892086148262, "learning_rate": 4.0921324979037e-06, "loss": 0.0013, "step": 147240 }, { "epoch": 2.4093921295917533, "grad_norm": 0.02169622853398323, "learning_rate": 4.091430422034859e-06, "loss": 0.0008, "step": 147250 }, { "epoch": 2.4095557555428293, "grad_norm": 0.022861197590827942, "learning_rate": 4.0907283646909795e-06, "loss": 0.0006, "step": 147260 }, { "epoch": 2.409719381493905, "grad_norm": 0.159433513879776, "learning_rate": 4.0900263258863746e-06, "loss": 0.002, "step": 147270 }, { "epoch": 2.409883007444981, "grad_norm": 0.019631046801805496, "learning_rate": 4.08932430563536e-06, "loss": 0.0007, "step": 147280 }, { "epoch": 2.410046633396057, "grad_norm": 0.047832004725933075, "learning_rate": 4.088622303952247e-06, "loss": 0.0007, "step": 147290 }, { "epoch": 2.4102102593471324, "grad_norm": 0.018303662538528442, "learning_rate": 4.087920320851351e-06, "loss": 0.0007, "step": 147300 }, { "epoch": 2.4103738852982084, "grad_norm": 0.02535712718963623, "learning_rate": 4.087218356346983e-06, "loss": 0.0018, "step": 147310 }, { "epoch": 2.410537511249284, "grad_norm": 0.1565566509962082, "learning_rate": 4.086516410453458e-06, "loss": 0.001, "step": 147320 }, { "epoch": 2.41070113720036, "grad_norm": 0.019839538261294365, "learning_rate": 4.085814483185086e-06, "loss": 0.0008, "step": 147330 }, { "epoch": 2.410864763151436, "grad_norm": 0.005056059453636408, "learning_rate": 4.085112574556179e-06, "loss": 0.001, "step": 147340 }, { "epoch": 2.4110283891025115, "grad_norm": 0.025519099086523056, "learning_rate": 4.084410684581048e-06, "loss": 0.001, "step": 147350 }, { "epoch": 2.4111920150535875, "grad_norm": 0.015169931575655937, "learning_rate": 4.083708813274005e-06, "loss": 0.0008, "step": 147360 }, { "epoch": 2.4113556410046635, "grad_norm": 0.0036031315103173256, "learning_rate": 4.08300696064936e-06, "loss": 0.0008, "step": 147370 }, { "epoch": 2.411519266955739, "grad_norm": 0.02439696714282036, "learning_rate": 4.082305126721424e-06, "loss": 0.0007, "step": 147380 }, { "epoch": 2.411682892906815, "grad_norm": 0.06380024552345276, "learning_rate": 4.0816033115045045e-06, "loss": 0.0006, "step": 147390 }, { "epoch": 2.411846518857891, "grad_norm": 0.039083972573280334, "learning_rate": 4.080901515012914e-06, "loss": 0.0006, "step": 147400 }, { "epoch": 2.4120101448089666, "grad_norm": 0.03509699925780296, "learning_rate": 4.080199737260958e-06, "loss": 0.0009, "step": 147410 }, { "epoch": 2.4121737707600426, "grad_norm": 0.050613824278116226, "learning_rate": 4.079497978262948e-06, "loss": 0.0018, "step": 147420 }, { "epoch": 2.4123373967111186, "grad_norm": 0.02017759531736374, "learning_rate": 4.078796238033192e-06, "loss": 0.0008, "step": 147430 }, { "epoch": 2.412501022662194, "grad_norm": 0.003103602211922407, "learning_rate": 4.078094516585997e-06, "loss": 0.0005, "step": 147440 }, { "epoch": 2.41266464861327, "grad_norm": 0.00858713686466217, "learning_rate": 4.07739281393567e-06, "loss": 0.0005, "step": 147450 }, { "epoch": 2.4128282745643457, "grad_norm": 0.04562317579984665, "learning_rate": 4.0766911300965195e-06, "loss": 0.0011, "step": 147460 }, { "epoch": 2.4129919005154217, "grad_norm": 0.04628712683916092, "learning_rate": 4.0759894650828515e-06, "loss": 0.0003, "step": 147470 }, { "epoch": 2.4131555264664977, "grad_norm": 0.01762169413268566, "learning_rate": 4.0752878189089725e-06, "loss": 0.0008, "step": 147480 }, { "epoch": 2.413319152417573, "grad_norm": 0.046333540230989456, "learning_rate": 4.074586191589188e-06, "loss": 0.001, "step": 147490 }, { "epoch": 2.413482778368649, "grad_norm": 0.08737366646528244, "learning_rate": 4.073884583137805e-06, "loss": 0.0007, "step": 147500 }, { "epoch": 2.413646404319725, "grad_norm": 0.009432674385607243, "learning_rate": 4.073182993569126e-06, "loss": 0.0007, "step": 147510 }, { "epoch": 2.4138100302708008, "grad_norm": 0.034782811999320984, "learning_rate": 4.0724814228974595e-06, "loss": 0.0008, "step": 147520 }, { "epoch": 2.4139736562218768, "grad_norm": 0.0206584669649601, "learning_rate": 4.071779871137107e-06, "loss": 0.0013, "step": 147530 }, { "epoch": 2.4141372821729528, "grad_norm": 0.06838134676218033, "learning_rate": 4.071078338302374e-06, "loss": 0.0012, "step": 147540 }, { "epoch": 2.4143009081240283, "grad_norm": 0.06392958015203476, "learning_rate": 4.070376824407562e-06, "loss": 0.0012, "step": 147550 }, { "epoch": 2.4144645340751043, "grad_norm": 0.03379460424184799, "learning_rate": 4.0696753294669785e-06, "loss": 0.0004, "step": 147560 }, { "epoch": 2.4146281600261803, "grad_norm": 0.0917697623372078, "learning_rate": 4.068973853494922e-06, "loss": 0.0006, "step": 147570 }, { "epoch": 2.414791785977256, "grad_norm": 0.03694979473948479, "learning_rate": 4.068272396505697e-06, "loss": 0.0005, "step": 147580 }, { "epoch": 2.414955411928332, "grad_norm": 0.008545983582735062, "learning_rate": 4.067570958513606e-06, "loss": 0.0008, "step": 147590 }, { "epoch": 2.415119037879408, "grad_norm": 0.011864403262734413, "learning_rate": 4.06686953953295e-06, "loss": 0.0008, "step": 147600 }, { "epoch": 2.4152826638304834, "grad_norm": 0.004751914646476507, "learning_rate": 4.06616813957803e-06, "loss": 0.0008, "step": 147610 }, { "epoch": 2.4154462897815594, "grad_norm": 0.02301429957151413, "learning_rate": 4.065466758663148e-06, "loss": 0.0006, "step": 147620 }, { "epoch": 2.4156099157326354, "grad_norm": 0.04363631829619408, "learning_rate": 4.064765396802604e-06, "loss": 0.0025, "step": 147630 }, { "epoch": 2.415773541683711, "grad_norm": 0.07516923546791077, "learning_rate": 4.064064054010699e-06, "loss": 0.0008, "step": 147640 }, { "epoch": 2.415937167634787, "grad_norm": 0.016092289239168167, "learning_rate": 4.06336273030173e-06, "loss": 0.0007, "step": 147650 }, { "epoch": 2.416100793585863, "grad_norm": 0.014085622504353523, "learning_rate": 4.062661425690001e-06, "loss": 0.0004, "step": 147660 }, { "epoch": 2.4162644195369385, "grad_norm": 0.009544692933559418, "learning_rate": 4.061960140189805e-06, "loss": 0.0007, "step": 147670 }, { "epoch": 2.4164280454880145, "grad_norm": 0.08073137700557709, "learning_rate": 4.061258873815447e-06, "loss": 0.0009, "step": 147680 }, { "epoch": 2.4165916714390905, "grad_norm": 0.0034448953811079264, "learning_rate": 4.060557626581219e-06, "loss": 0.0013, "step": 147690 }, { "epoch": 2.416755297390166, "grad_norm": 0.11745734512805939, "learning_rate": 4.059856398501426e-06, "loss": 0.0009, "step": 147700 }, { "epoch": 2.416918923341242, "grad_norm": 0.08095382899045944, "learning_rate": 4.059155189590357e-06, "loss": 0.0015, "step": 147710 }, { "epoch": 2.4170825492923176, "grad_norm": 0.05566380172967911, "learning_rate": 4.0584539998623175e-06, "loss": 0.0008, "step": 147720 }, { "epoch": 2.4172461752433936, "grad_norm": 0.08117559552192688, "learning_rate": 4.057752829331597e-06, "loss": 0.0011, "step": 147730 }, { "epoch": 2.4174098011944696, "grad_norm": 0.017076827585697174, "learning_rate": 4.057051678012499e-06, "loss": 0.001, "step": 147740 }, { "epoch": 2.417573427145545, "grad_norm": 0.08369353413581848, "learning_rate": 4.056350545919311e-06, "loss": 0.0008, "step": 147750 }, { "epoch": 2.417737053096621, "grad_norm": 0.13180473446846008, "learning_rate": 4.055649433066336e-06, "loss": 0.0009, "step": 147760 }, { "epoch": 2.417900679047697, "grad_norm": 0.12627723813056946, "learning_rate": 4.054948339467863e-06, "loss": 0.0008, "step": 147770 }, { "epoch": 2.4180643049987727, "grad_norm": 0.03334576636552811, "learning_rate": 4.0542472651381925e-06, "loss": 0.0011, "step": 147780 }, { "epoch": 2.4182279309498487, "grad_norm": 0.08977006375789642, "learning_rate": 4.053546210091613e-06, "loss": 0.001, "step": 147790 }, { "epoch": 2.4183915569009247, "grad_norm": 0.007861998863518238, "learning_rate": 4.052845174342424e-06, "loss": 0.0009, "step": 147800 }, { "epoch": 2.418555182852, "grad_norm": 0.13064582645893097, "learning_rate": 4.052144157904915e-06, "loss": 0.0012, "step": 147810 }, { "epoch": 2.418718808803076, "grad_norm": 0.07755519449710846, "learning_rate": 4.051443160793382e-06, "loss": 0.0007, "step": 147820 }, { "epoch": 2.4188824347541518, "grad_norm": 0.014315729960799217, "learning_rate": 4.050742183022113e-06, "loss": 0.0007, "step": 147830 }, { "epoch": 2.4190460607052278, "grad_norm": 0.11924134939908981, "learning_rate": 4.050041224605408e-06, "loss": 0.0014, "step": 147840 }, { "epoch": 2.4192096866563038, "grad_norm": 0.031516432762145996, "learning_rate": 4.049340285557552e-06, "loss": 0.0008, "step": 147850 }, { "epoch": 2.4193733126073793, "grad_norm": 0.003307241713628173, "learning_rate": 4.048639365892839e-06, "loss": 0.0011, "step": 147860 }, { "epoch": 2.4195369385584553, "grad_norm": 0.03334670886397362, "learning_rate": 4.04793846562556e-06, "loss": 0.0012, "step": 147870 }, { "epoch": 2.4197005645095313, "grad_norm": 0.058554522693157196, "learning_rate": 4.047237584770007e-06, "loss": 0.0012, "step": 147880 }, { "epoch": 2.419864190460607, "grad_norm": 0.010944467969238758, "learning_rate": 4.046536723340469e-06, "loss": 0.0006, "step": 147890 }, { "epoch": 2.420027816411683, "grad_norm": 0.04809904843568802, "learning_rate": 4.045835881351235e-06, "loss": 0.0008, "step": 147900 }, { "epoch": 2.420191442362759, "grad_norm": 0.055776067078113556, "learning_rate": 4.045135058816597e-06, "loss": 0.0004, "step": 147910 }, { "epoch": 2.4203550683138344, "grad_norm": 0.029655367136001587, "learning_rate": 4.044434255750844e-06, "loss": 0.0005, "step": 147920 }, { "epoch": 2.4205186942649104, "grad_norm": 0.03250817954540253, "learning_rate": 4.043733472168262e-06, "loss": 0.0007, "step": 147930 }, { "epoch": 2.4206823202159864, "grad_norm": 0.04142424464225769, "learning_rate": 4.043032708083143e-06, "loss": 0.0012, "step": 147940 }, { "epoch": 2.420845946167062, "grad_norm": 0.0703406110405922, "learning_rate": 4.042331963509772e-06, "loss": 0.001, "step": 147950 }, { "epoch": 2.421009572118138, "grad_norm": 0.029762156307697296, "learning_rate": 4.041631238462438e-06, "loss": 0.0007, "step": 147960 }, { "epoch": 2.421173198069214, "grad_norm": 0.030865829437971115, "learning_rate": 4.0409305329554285e-06, "loss": 0.0005, "step": 147970 }, { "epoch": 2.4213368240202895, "grad_norm": 0.037248674780130386, "learning_rate": 4.040229847003029e-06, "loss": 0.0007, "step": 147980 }, { "epoch": 2.4215004499713655, "grad_norm": 0.0040017394348979, "learning_rate": 4.039529180619527e-06, "loss": 0.0009, "step": 147990 }, { "epoch": 2.4216640759224415, "grad_norm": 0.020519044250249863, "learning_rate": 4.038828533819209e-06, "loss": 0.0011, "step": 148000 }, { "epoch": 2.421827701873517, "grad_norm": 0.048821043223142624, "learning_rate": 4.0381279066163585e-06, "loss": 0.0021, "step": 148010 }, { "epoch": 2.421991327824593, "grad_norm": 0.06346671283245087, "learning_rate": 4.0374272990252625e-06, "loss": 0.0009, "step": 148020 }, { "epoch": 2.422154953775669, "grad_norm": 0.03721042349934578, "learning_rate": 4.036726711060205e-06, "loss": 0.0008, "step": 148030 }, { "epoch": 2.4223185797267446, "grad_norm": 0.05507481470704079, "learning_rate": 4.03602614273547e-06, "loss": 0.0006, "step": 148040 }, { "epoch": 2.4224822056778206, "grad_norm": 0.05272253602743149, "learning_rate": 4.035325594065343e-06, "loss": 0.0016, "step": 148050 }, { "epoch": 2.4226458316288966, "grad_norm": 0.04567726328969002, "learning_rate": 4.034625065064108e-06, "loss": 0.0011, "step": 148060 }, { "epoch": 2.422809457579972, "grad_norm": 0.0073843467980623245, "learning_rate": 4.033924555746045e-06, "loss": 0.0007, "step": 148070 }, { "epoch": 2.422973083531048, "grad_norm": 0.014473779127001762, "learning_rate": 4.033224066125439e-06, "loss": 0.0006, "step": 148080 }, { "epoch": 2.4231367094821237, "grad_norm": 0.04396287724375725, "learning_rate": 4.032523596216572e-06, "loss": 0.0013, "step": 148090 }, { "epoch": 2.4233003354331997, "grad_norm": 0.0531262531876564, "learning_rate": 4.031823146033727e-06, "loss": 0.0009, "step": 148100 }, { "epoch": 2.4234639613842757, "grad_norm": 0.05015639215707779, "learning_rate": 4.031122715591184e-06, "loss": 0.0005, "step": 148110 }, { "epoch": 2.423627587335351, "grad_norm": 0.01930811069905758, "learning_rate": 4.030422304903225e-06, "loss": 0.0005, "step": 148120 }, { "epoch": 2.423791213286427, "grad_norm": 0.15020954608917236, "learning_rate": 4.02972191398413e-06, "loss": 0.0005, "step": 148130 }, { "epoch": 2.423954839237503, "grad_norm": 0.011017093434929848, "learning_rate": 4.02902154284818e-06, "loss": 0.0005, "step": 148140 }, { "epoch": 2.4241184651885788, "grad_norm": 0.053608328104019165, "learning_rate": 4.028321191509655e-06, "loss": 0.0028, "step": 148150 }, { "epoch": 2.4242820911396548, "grad_norm": 0.033541299402713776, "learning_rate": 4.027620859982836e-06, "loss": 0.0012, "step": 148160 }, { "epoch": 2.4244457170907308, "grad_norm": 0.024030612781643867, "learning_rate": 4.026920548281999e-06, "loss": 0.0009, "step": 148170 }, { "epoch": 2.4246093430418063, "grad_norm": 0.057326026260852814, "learning_rate": 4.0262202564214255e-06, "loss": 0.0007, "step": 148180 }, { "epoch": 2.4247729689928823, "grad_norm": 0.03239580616354942, "learning_rate": 4.025519984415391e-06, "loss": 0.0013, "step": 148190 }, { "epoch": 2.4249365949439583, "grad_norm": 0.028511112555861473, "learning_rate": 4.024819732278178e-06, "loss": 0.0005, "step": 148200 }, { "epoch": 2.425100220895034, "grad_norm": 0.008561459369957447, "learning_rate": 4.024119500024059e-06, "loss": 0.0009, "step": 148210 }, { "epoch": 2.42526384684611, "grad_norm": 0.0995476245880127, "learning_rate": 4.023419287667315e-06, "loss": 0.0008, "step": 148220 }, { "epoch": 2.4254274727971854, "grad_norm": 0.050222575664520264, "learning_rate": 4.02271909522222e-06, "loss": 0.0023, "step": 148230 }, { "epoch": 2.4255910987482614, "grad_norm": 0.041171520948410034, "learning_rate": 4.022018922703052e-06, "loss": 0.0013, "step": 148240 }, { "epoch": 2.4257547246993374, "grad_norm": 0.06628189235925674, "learning_rate": 4.021318770124086e-06, "loss": 0.0009, "step": 148250 }, { "epoch": 2.425918350650413, "grad_norm": 0.07531756162643433, "learning_rate": 4.020618637499599e-06, "loss": 0.0006, "step": 148260 }, { "epoch": 2.426081976601489, "grad_norm": 0.014596967026591301, "learning_rate": 4.0199185248438635e-06, "loss": 0.0008, "step": 148270 }, { "epoch": 2.426245602552565, "grad_norm": 0.014368673786520958, "learning_rate": 4.019218432171158e-06, "loss": 0.0012, "step": 148280 }, { "epoch": 2.4264092285036405, "grad_norm": 0.0017859707586467266, "learning_rate": 4.018518359495752e-06, "loss": 0.0006, "step": 148290 }, { "epoch": 2.4265728544547165, "grad_norm": 0.03864646330475807, "learning_rate": 4.017818306831925e-06, "loss": 0.0015, "step": 148300 }, { "epoch": 2.4267364804057925, "grad_norm": 0.05404319614171982, "learning_rate": 4.017118274193944e-06, "loss": 0.0005, "step": 148310 }, { "epoch": 2.426900106356868, "grad_norm": 0.45453813672065735, "learning_rate": 4.016418261596089e-06, "loss": 0.0019, "step": 148320 }, { "epoch": 2.427063732307944, "grad_norm": 0.03322461247444153, "learning_rate": 4.015718269052625e-06, "loss": 0.0011, "step": 148330 }, { "epoch": 2.42722735825902, "grad_norm": 0.10177291184663773, "learning_rate": 4.015018296577832e-06, "loss": 0.0008, "step": 148340 }, { "epoch": 2.4273909842100956, "grad_norm": 0.06122678518295288, "learning_rate": 4.014318344185976e-06, "loss": 0.0008, "step": 148350 }, { "epoch": 2.4275546101611716, "grad_norm": 0.04889995604753494, "learning_rate": 4.013618411891333e-06, "loss": 0.001, "step": 148360 }, { "epoch": 2.4277182361122476, "grad_norm": 0.03140859678387642, "learning_rate": 4.0129184997081674e-06, "loss": 0.0009, "step": 148370 }, { "epoch": 2.427881862063323, "grad_norm": 0.009043307974934578, "learning_rate": 4.012218607650759e-06, "loss": 0.0007, "step": 148380 }, { "epoch": 2.428045488014399, "grad_norm": 0.04389815405011177, "learning_rate": 4.01151873573337e-06, "loss": 0.0008, "step": 148390 }, { "epoch": 2.428209113965475, "grad_norm": 0.021557485684752464, "learning_rate": 4.010818883970275e-06, "loss": 0.0015, "step": 148400 }, { "epoch": 2.4283727399165507, "grad_norm": 0.005635992623865604, "learning_rate": 4.01011905237574e-06, "loss": 0.0006, "step": 148410 }, { "epoch": 2.4285363658676267, "grad_norm": 0.02403765730559826, "learning_rate": 4.0094192409640374e-06, "loss": 0.0012, "step": 148420 }, { "epoch": 2.4286999918187027, "grad_norm": 0.044221971184015274, "learning_rate": 4.008719449749432e-06, "loss": 0.0012, "step": 148430 }, { "epoch": 2.428863617769778, "grad_norm": 0.01264910213649273, "learning_rate": 4.008019678746197e-06, "loss": 0.0006, "step": 148440 }, { "epoch": 2.429027243720854, "grad_norm": 0.045292019844055176, "learning_rate": 4.007319927968593e-06, "loss": 0.0014, "step": 148450 }, { "epoch": 2.42919086967193, "grad_norm": 0.053349029272794724, "learning_rate": 4.006620197430896e-06, "loss": 0.0008, "step": 148460 }, { "epoch": 2.4293544956230058, "grad_norm": 0.011919780634343624, "learning_rate": 4.005920487147364e-06, "loss": 0.0012, "step": 148470 }, { "epoch": 2.4295181215740818, "grad_norm": 0.09639975428581238, "learning_rate": 4.005220797132271e-06, "loss": 0.0006, "step": 148480 }, { "epoch": 2.4296817475251573, "grad_norm": 0.0837942436337471, "learning_rate": 4.004521127399879e-06, "loss": 0.0007, "step": 148490 }, { "epoch": 2.4298453734762333, "grad_norm": 0.05558514967560768, "learning_rate": 4.003821477964455e-06, "loss": 0.0006, "step": 148500 }, { "epoch": 2.4300089994273093, "grad_norm": 0.04416656866669655, "learning_rate": 4.003121848840262e-06, "loss": 0.0005, "step": 148510 }, { "epoch": 2.430172625378385, "grad_norm": 0.060288723558187485, "learning_rate": 4.002422240041569e-06, "loss": 0.001, "step": 148520 }, { "epoch": 2.430336251329461, "grad_norm": 0.052232787013053894, "learning_rate": 4.001722651582636e-06, "loss": 0.0016, "step": 148530 }, { "epoch": 2.430499877280537, "grad_norm": 0.09053073078393936, "learning_rate": 4.001023083477731e-06, "loss": 0.0008, "step": 148540 }, { "epoch": 2.4306635032316124, "grad_norm": 0.030474023893475533, "learning_rate": 4.000323535741115e-06, "loss": 0.0007, "step": 148550 }, { "epoch": 2.4308271291826884, "grad_norm": 0.011830596253275871, "learning_rate": 3.999624008387052e-06, "loss": 0.0007, "step": 148560 }, { "epoch": 2.4309907551337644, "grad_norm": 0.020800001919269562, "learning_rate": 3.998924501429804e-06, "loss": 0.0022, "step": 148570 }, { "epoch": 2.43115438108484, "grad_norm": 0.000966563995461911, "learning_rate": 3.9982250148836345e-06, "loss": 0.001, "step": 148580 }, { "epoch": 2.431318007035916, "grad_norm": 0.002346954308450222, "learning_rate": 3.997525548762804e-06, "loss": 0.0011, "step": 148590 }, { "epoch": 2.4314816329869915, "grad_norm": 0.042551979422569275, "learning_rate": 3.996826103081577e-06, "loss": 0.0005, "step": 148600 }, { "epoch": 2.4316452589380675, "grad_norm": 0.01460945699363947, "learning_rate": 3.996126677854211e-06, "loss": 0.0006, "step": 148610 }, { "epoch": 2.4318088848891435, "grad_norm": 0.023319249972701073, "learning_rate": 3.995427273094969e-06, "loss": 0.0012, "step": 148620 }, { "epoch": 2.431972510840219, "grad_norm": 0.027666190639138222, "learning_rate": 3.994727888818111e-06, "loss": 0.001, "step": 148630 }, { "epoch": 2.432136136791295, "grad_norm": 0.02335228957235813, "learning_rate": 3.994028525037896e-06, "loss": 0.001, "step": 148640 }, { "epoch": 2.432299762742371, "grad_norm": 0.016582559794187546, "learning_rate": 3.993329181768583e-06, "loss": 0.001, "step": 148650 }, { "epoch": 2.4324633886934466, "grad_norm": 0.028497418388724327, "learning_rate": 3.992629859024433e-06, "loss": 0.0009, "step": 148660 }, { "epoch": 2.4326270146445226, "grad_norm": 0.054950036108493805, "learning_rate": 3.991930556819702e-06, "loss": 0.0012, "step": 148670 }, { "epoch": 2.4327906405955986, "grad_norm": 0.039075739681720734, "learning_rate": 3.991231275168651e-06, "loss": 0.001, "step": 148680 }, { "epoch": 2.432954266546674, "grad_norm": 0.0273760873824358, "learning_rate": 3.9905320140855355e-06, "loss": 0.0008, "step": 148690 }, { "epoch": 2.43311789249775, "grad_norm": 0.012453868053853512, "learning_rate": 3.989832773584614e-06, "loss": 0.0009, "step": 148700 }, { "epoch": 2.433281518448826, "grad_norm": 0.021584611386060715, "learning_rate": 3.9891335536801434e-06, "loss": 0.0007, "step": 148710 }, { "epoch": 2.4334451443999017, "grad_norm": 0.06358496844768524, "learning_rate": 3.9884343543863805e-06, "loss": 0.0005, "step": 148720 }, { "epoch": 2.4336087703509777, "grad_norm": 0.02206256613135338, "learning_rate": 3.987735175717579e-06, "loss": 0.0006, "step": 148730 }, { "epoch": 2.4337723963020537, "grad_norm": 0.051690008491277695, "learning_rate": 3.987036017687998e-06, "loss": 0.0009, "step": 148740 }, { "epoch": 2.433936022253129, "grad_norm": 0.07544250786304474, "learning_rate": 3.9863368803118905e-06, "loss": 0.0007, "step": 148750 }, { "epoch": 2.434099648204205, "grad_norm": 0.09289561957120895, "learning_rate": 3.985637763603512e-06, "loss": 0.0012, "step": 148760 }, { "epoch": 2.434263274155281, "grad_norm": 0.0563623383641243, "learning_rate": 3.984938667577116e-06, "loss": 0.0006, "step": 148770 }, { "epoch": 2.4344269001063568, "grad_norm": 0.008377346210181713, "learning_rate": 3.984239592246959e-06, "loss": 0.0006, "step": 148780 }, { "epoch": 2.4345905260574328, "grad_norm": 0.022997448220849037, "learning_rate": 3.9835405376272915e-06, "loss": 0.0009, "step": 148790 }, { "epoch": 2.4347541520085088, "grad_norm": 0.014758425764739513, "learning_rate": 3.982841503732369e-06, "loss": 0.0011, "step": 148800 }, { "epoch": 2.4349177779595843, "grad_norm": 0.046782154589891434, "learning_rate": 3.982142490576442e-06, "loss": 0.0011, "step": 148810 }, { "epoch": 2.4350814039106603, "grad_norm": 0.02463490143418312, "learning_rate": 3.981443498173764e-06, "loss": 0.0005, "step": 148820 }, { "epoch": 2.4352450298617363, "grad_norm": 0.0069418917410075665, "learning_rate": 3.9807445265385865e-06, "loss": 0.0006, "step": 148830 }, { "epoch": 2.435408655812812, "grad_norm": 0.009419113397598267, "learning_rate": 3.980045575685163e-06, "loss": 0.0008, "step": 148840 }, { "epoch": 2.435572281763888, "grad_norm": 0.04220080375671387, "learning_rate": 3.979346645627742e-06, "loss": 0.0012, "step": 148850 }, { "epoch": 2.4357359077149634, "grad_norm": 0.17018160223960876, "learning_rate": 3.978647736380575e-06, "loss": 0.0007, "step": 148860 }, { "epoch": 2.4358995336660394, "grad_norm": 0.015274155884981155, "learning_rate": 3.977948847957911e-06, "loss": 0.0007, "step": 148870 }, { "epoch": 2.4360631596171154, "grad_norm": 0.12818630039691925, "learning_rate": 3.977249980374002e-06, "loss": 0.0011, "step": 148880 }, { "epoch": 2.436226785568191, "grad_norm": 0.03157830238342285, "learning_rate": 3.976551133643095e-06, "loss": 0.0004, "step": 148890 }, { "epoch": 2.436390411519267, "grad_norm": 0.037235163152217865, "learning_rate": 3.975852307779441e-06, "loss": 0.0006, "step": 148900 }, { "epoch": 2.436554037470343, "grad_norm": 0.0032816624734550714, "learning_rate": 3.9751535027972865e-06, "loss": 0.0009, "step": 148910 }, { "epoch": 2.4367176634214185, "grad_norm": 0.031428735703229904, "learning_rate": 3.974454718710882e-06, "loss": 0.0011, "step": 148920 }, { "epoch": 2.4368812893724945, "grad_norm": 0.005433597601950169, "learning_rate": 3.973755955534471e-06, "loss": 0.0007, "step": 148930 }, { "epoch": 2.4370449153235705, "grad_norm": 0.0013518515042960644, "learning_rate": 3.973057213282307e-06, "loss": 0.0005, "step": 148940 }, { "epoch": 2.437208541274646, "grad_norm": 0.02421480603516102, "learning_rate": 3.972358491968629e-06, "loss": 0.0008, "step": 148950 }, { "epoch": 2.437372167225722, "grad_norm": 0.04296385869383812, "learning_rate": 3.971659791607691e-06, "loss": 0.0008, "step": 148960 }, { "epoch": 2.437535793176798, "grad_norm": 0.009186485782265663, "learning_rate": 3.970961112213732e-06, "loss": 0.0004, "step": 148970 }, { "epoch": 2.4376994191278736, "grad_norm": 0.042835719883441925, "learning_rate": 3.9702624538010035e-06, "loss": 0.001, "step": 148980 }, { "epoch": 2.4378630450789496, "grad_norm": 0.05422279238700867, "learning_rate": 3.969563816383745e-06, "loss": 0.0004, "step": 148990 }, { "epoch": 2.438026671030025, "grad_norm": 0.025080794468522072, "learning_rate": 3.968865199976207e-06, "loss": 0.0008, "step": 149000 }, { "epoch": 2.438190296981101, "grad_norm": 0.016518203541636467, "learning_rate": 3.968166604592628e-06, "loss": 0.0007, "step": 149010 }, { "epoch": 2.438353922932177, "grad_norm": 0.009026875719428062, "learning_rate": 3.9674680302472565e-06, "loss": 0.0007, "step": 149020 }, { "epoch": 2.4385175488832527, "grad_norm": 0.021450169384479523, "learning_rate": 3.966769476954332e-06, "loss": 0.0009, "step": 149030 }, { "epoch": 2.4386811748343287, "grad_norm": 0.13895921409130096, "learning_rate": 3.966070944728101e-06, "loss": 0.0006, "step": 149040 }, { "epoch": 2.4388448007854047, "grad_norm": 0.049099262803792953, "learning_rate": 3.9653724335828015e-06, "loss": 0.0009, "step": 149050 }, { "epoch": 2.43900842673648, "grad_norm": 0.13102909922599792, "learning_rate": 3.964673943532682e-06, "loss": 0.0009, "step": 149060 }, { "epoch": 2.439172052687556, "grad_norm": 0.07113751769065857, "learning_rate": 3.9639754745919755e-06, "loss": 0.0009, "step": 149070 }, { "epoch": 2.439335678638632, "grad_norm": 0.01282584574073553, "learning_rate": 3.963277026774932e-06, "loss": 0.0005, "step": 149080 }, { "epoch": 2.4394993045897078, "grad_norm": 0.10147931426763535, "learning_rate": 3.962578600095785e-06, "loss": 0.0008, "step": 149090 }, { "epoch": 2.4396629305407838, "grad_norm": 0.03858553245663643, "learning_rate": 3.961880194568781e-06, "loss": 0.0006, "step": 149100 }, { "epoch": 2.4398265564918598, "grad_norm": 0.026906423270702362, "learning_rate": 3.961181810208154e-06, "loss": 0.0006, "step": 149110 }, { "epoch": 2.4399901824429353, "grad_norm": 0.16116183996200562, "learning_rate": 3.96048344702815e-06, "loss": 0.0017, "step": 149120 }, { "epoch": 2.4401538083940113, "grad_norm": 0.036030881106853485, "learning_rate": 3.959785105043002e-06, "loss": 0.0006, "step": 149130 }, { "epoch": 2.4403174343450873, "grad_norm": 0.02710048295557499, "learning_rate": 3.959086784266952e-06, "loss": 0.0003, "step": 149140 }, { "epoch": 2.440481060296163, "grad_norm": 0.029040370136499405, "learning_rate": 3.958388484714236e-06, "loss": 0.0004, "step": 149150 }, { "epoch": 2.440644686247239, "grad_norm": 0.050214655697345734, "learning_rate": 3.9576902063990945e-06, "loss": 0.0009, "step": 149160 }, { "epoch": 2.440808312198315, "grad_norm": 0.049917832016944885, "learning_rate": 3.956991949335762e-06, "loss": 0.0013, "step": 149170 }, { "epoch": 2.4409719381493904, "grad_norm": 0.012414957396686077, "learning_rate": 3.956293713538477e-06, "loss": 0.0005, "step": 149180 }, { "epoch": 2.4411355641004664, "grad_norm": 0.025513749569654465, "learning_rate": 3.955595499021475e-06, "loss": 0.0013, "step": 149190 }, { "epoch": 2.4412991900515424, "grad_norm": 0.0029737832956016064, "learning_rate": 3.954897305798993e-06, "loss": 0.0007, "step": 149200 }, { "epoch": 2.441462816002618, "grad_norm": 0.06561461091041565, "learning_rate": 3.954199133885267e-06, "loss": 0.0008, "step": 149210 }, { "epoch": 2.441626441953694, "grad_norm": 0.03387913852930069, "learning_rate": 3.9535009832945305e-06, "loss": 0.0007, "step": 149220 }, { "epoch": 2.44179006790477, "grad_norm": 0.02060861699283123, "learning_rate": 3.952802854041018e-06, "loss": 0.0011, "step": 149230 }, { "epoch": 2.4419536938558455, "grad_norm": 0.06473975628614426, "learning_rate": 3.952104746138966e-06, "loss": 0.0013, "step": 149240 }, { "epoch": 2.4421173198069215, "grad_norm": 0.034025419503450394, "learning_rate": 3.951406659602606e-06, "loss": 0.0008, "step": 149250 }, { "epoch": 2.442280945757997, "grad_norm": 0.015230637975037098, "learning_rate": 3.9507085944461735e-06, "loss": 0.0011, "step": 149260 }, { "epoch": 2.442444571709073, "grad_norm": 0.05834510177373886, "learning_rate": 3.950010550683899e-06, "loss": 0.0008, "step": 149270 }, { "epoch": 2.442608197660149, "grad_norm": 0.022828850895166397, "learning_rate": 3.949312528330019e-06, "loss": 0.0008, "step": 149280 }, { "epoch": 2.4427718236112246, "grad_norm": 0.023250829428434372, "learning_rate": 3.94861452739876e-06, "loss": 0.0007, "step": 149290 }, { "epoch": 2.4429354495623006, "grad_norm": 0.023240987211465836, "learning_rate": 3.947916547904358e-06, "loss": 0.0005, "step": 149300 }, { "epoch": 2.4430990755133766, "grad_norm": 0.01603192649781704, "learning_rate": 3.9472185898610425e-06, "loss": 0.0008, "step": 149310 }, { "epoch": 2.443262701464452, "grad_norm": 0.023664124310016632, "learning_rate": 3.946520653283045e-06, "loss": 0.0009, "step": 149320 }, { "epoch": 2.443426327415528, "grad_norm": 0.004069575108587742, "learning_rate": 3.945822738184594e-06, "loss": 0.0008, "step": 149330 }, { "epoch": 2.443589953366604, "grad_norm": 0.022419439628720284, "learning_rate": 3.9451248445799224e-06, "loss": 0.0005, "step": 149340 }, { "epoch": 2.4437535793176797, "grad_norm": 0.05919869616627693, "learning_rate": 3.944426972483258e-06, "loss": 0.0008, "step": 149350 }, { "epoch": 2.4439172052687557, "grad_norm": 0.009136136621236801, "learning_rate": 3.94372912190883e-06, "loss": 0.0003, "step": 149360 }, { "epoch": 2.444080831219831, "grad_norm": 0.05961159989237785, "learning_rate": 3.943031292870866e-06, "loss": 0.001, "step": 149370 }, { "epoch": 2.444244457170907, "grad_norm": 0.06453901529312134, "learning_rate": 3.9423334853835956e-06, "loss": 0.0003, "step": 149380 }, { "epoch": 2.444408083121983, "grad_norm": 0.041029367595911026, "learning_rate": 3.9416356994612455e-06, "loss": 0.0004, "step": 149390 }, { "epoch": 2.4445717090730588, "grad_norm": 0.04008742421865463, "learning_rate": 3.940937935118044e-06, "loss": 0.0007, "step": 149400 }, { "epoch": 2.4447353350241348, "grad_norm": 0.05744939297437668, "learning_rate": 3.940240192368216e-06, "loss": 0.002, "step": 149410 }, { "epoch": 2.4448989609752108, "grad_norm": 0.032397862523794174, "learning_rate": 3.9395424712259915e-06, "loss": 0.001, "step": 149420 }, { "epoch": 2.4450625869262863, "grad_norm": 0.0533674880862236, "learning_rate": 3.938844771705592e-06, "loss": 0.0008, "step": 149430 }, { "epoch": 2.4452262128773623, "grad_norm": 0.012170976027846336, "learning_rate": 3.938147093821246e-06, "loss": 0.0007, "step": 149440 }, { "epoch": 2.4453898388284383, "grad_norm": 0.06069180741906166, "learning_rate": 3.937449437587177e-06, "loss": 0.0006, "step": 149450 }, { "epoch": 2.445553464779514, "grad_norm": 0.0032063100952655077, "learning_rate": 3.936751803017612e-06, "loss": 0.0004, "step": 149460 }, { "epoch": 2.44571709073059, "grad_norm": 0.12993863224983215, "learning_rate": 3.9360541901267716e-06, "loss": 0.0015, "step": 149470 }, { "epoch": 2.445880716681666, "grad_norm": 0.005645650904625654, "learning_rate": 3.9353565989288825e-06, "loss": 0.0007, "step": 149480 }, { "epoch": 2.4460443426327414, "grad_norm": 0.05196567624807358, "learning_rate": 3.934659029438168e-06, "loss": 0.0007, "step": 149490 }, { "epoch": 2.4462079685838174, "grad_norm": 0.02726210281252861, "learning_rate": 3.933961481668847e-06, "loss": 0.0016, "step": 149500 }, { "epoch": 2.4463715945348934, "grad_norm": 0.034377165138721466, "learning_rate": 3.933263955635148e-06, "loss": 0.0021, "step": 149510 }, { "epoch": 2.446535220485969, "grad_norm": 0.03090793639421463, "learning_rate": 3.932566451351286e-06, "loss": 0.0008, "step": 149520 }, { "epoch": 2.446698846437045, "grad_norm": 0.08738775551319122, "learning_rate": 3.93186896883149e-06, "loss": 0.0014, "step": 149530 }, { "epoch": 2.446862472388121, "grad_norm": 0.07723113149404526, "learning_rate": 3.931171508089974e-06, "loss": 0.0005, "step": 149540 }, { "epoch": 2.4470260983391965, "grad_norm": 0.004021161701530218, "learning_rate": 3.930474069140965e-06, "loss": 0.0007, "step": 149550 }, { "epoch": 2.4471897242902725, "grad_norm": 0.013734049163758755, "learning_rate": 3.9297766519986766e-06, "loss": 0.0007, "step": 149560 }, { "epoch": 2.4473533502413485, "grad_norm": 0.055789995938539505, "learning_rate": 3.929079256677335e-06, "loss": 0.0007, "step": 149570 }, { "epoch": 2.447516976192424, "grad_norm": 0.04384993389248848, "learning_rate": 3.928381883191155e-06, "loss": 0.0009, "step": 149580 }, { "epoch": 2.4476806021435, "grad_norm": 0.033470313996076584, "learning_rate": 3.927684531554359e-06, "loss": 0.0004, "step": 149590 }, { "epoch": 2.447844228094576, "grad_norm": 0.04503292217850685, "learning_rate": 3.926987201781162e-06, "loss": 0.0015, "step": 149600 }, { "epoch": 2.4480078540456516, "grad_norm": 0.029304146766662598, "learning_rate": 3.926289893885783e-06, "loss": 0.0013, "step": 149610 }, { "epoch": 2.4481714799967276, "grad_norm": 0.033192239701747894, "learning_rate": 3.92559260788244e-06, "loss": 0.0007, "step": 149620 }, { "epoch": 2.4483351059478036, "grad_norm": 0.024088405072689056, "learning_rate": 3.924895343785351e-06, "loss": 0.0007, "step": 149630 }, { "epoch": 2.448498731898879, "grad_norm": 0.12167217582464218, "learning_rate": 3.924198101608731e-06, "loss": 0.0017, "step": 149640 }, { "epoch": 2.448662357849955, "grad_norm": 0.028920842334628105, "learning_rate": 3.9235008813667965e-06, "loss": 0.0007, "step": 149650 }, { "epoch": 2.4488259838010307, "grad_norm": 0.012005764991044998, "learning_rate": 3.922803683073763e-06, "loss": 0.0007, "step": 149660 }, { "epoch": 2.4489896097521067, "grad_norm": 0.06854907423257828, "learning_rate": 3.922106506743847e-06, "loss": 0.0008, "step": 149670 }, { "epoch": 2.4491532357031827, "grad_norm": 0.02374492585659027, "learning_rate": 3.921409352391262e-06, "loss": 0.0009, "step": 149680 }, { "epoch": 2.449316861654258, "grad_norm": 0.041165560483932495, "learning_rate": 3.920712220030224e-06, "loss": 0.0006, "step": 149690 }, { "epoch": 2.449480487605334, "grad_norm": 0.05923851206898689, "learning_rate": 3.920015109674946e-06, "loss": 0.0008, "step": 149700 }, { "epoch": 2.44964411355641, "grad_norm": 0.002078503603115678, "learning_rate": 3.919318021339641e-06, "loss": 0.0019, "step": 149710 }, { "epoch": 2.4498077395074858, "grad_norm": 0.06474234163761139, "learning_rate": 3.918620955038522e-06, "loss": 0.0014, "step": 149720 }, { "epoch": 2.4499713654585618, "grad_norm": 0.009766720235347748, "learning_rate": 3.9179239107858025e-06, "loss": 0.0015, "step": 149730 }, { "epoch": 2.4501349914096378, "grad_norm": 0.05371371656656265, "learning_rate": 3.917226888595693e-06, "loss": 0.0006, "step": 149740 }, { "epoch": 2.4502986173607133, "grad_norm": 0.03646279126405716, "learning_rate": 3.916529888482409e-06, "loss": 0.0008, "step": 149750 }, { "epoch": 2.4504622433117893, "grad_norm": 0.020492443814873695, "learning_rate": 3.915832910460157e-06, "loss": 0.0004, "step": 149760 }, { "epoch": 2.450625869262865, "grad_norm": 0.02992619201540947, "learning_rate": 3.91513595454315e-06, "loss": 0.0007, "step": 149770 }, { "epoch": 2.450789495213941, "grad_norm": 0.03678682819008827, "learning_rate": 3.914439020745599e-06, "loss": 0.0007, "step": 149780 }, { "epoch": 2.450953121165017, "grad_norm": 0.04633420705795288, "learning_rate": 3.913742109081713e-06, "loss": 0.0005, "step": 149790 }, { "epoch": 2.4511167471160924, "grad_norm": 0.022435147315263748, "learning_rate": 3.9130452195657006e-06, "loss": 0.0007, "step": 149800 }, { "epoch": 2.4512803730671684, "grad_norm": 0.034521061927080154, "learning_rate": 3.912348352211773e-06, "loss": 0.0006, "step": 149810 }, { "epoch": 2.4514439990182444, "grad_norm": 0.1666451245546341, "learning_rate": 3.911651507034137e-06, "loss": 0.0007, "step": 149820 }, { "epoch": 2.45160762496932, "grad_norm": 0.035535234957933426, "learning_rate": 3.910954684047002e-06, "loss": 0.0012, "step": 149830 }, { "epoch": 2.451771250920396, "grad_norm": 0.0010548284044489264, "learning_rate": 3.9102578832645736e-06, "loss": 0.0003, "step": 149840 }, { "epoch": 2.451934876871472, "grad_norm": 0.00913222786039114, "learning_rate": 3.9095611047010605e-06, "loss": 0.0006, "step": 149850 }, { "epoch": 2.4520985028225475, "grad_norm": 0.011261926963925362, "learning_rate": 3.908864348370669e-06, "loss": 0.0023, "step": 149860 }, { "epoch": 2.4522621287736235, "grad_norm": 0.033262211829423904, "learning_rate": 3.908167614287606e-06, "loss": 0.0007, "step": 149870 }, { "epoch": 2.4524257547246995, "grad_norm": 0.02528540790081024, "learning_rate": 3.9074709024660765e-06, "loss": 0.0012, "step": 149880 }, { "epoch": 2.452589380675775, "grad_norm": 0.011173881590366364, "learning_rate": 3.9067742129202865e-06, "loss": 0.0012, "step": 149890 }, { "epoch": 2.452753006626851, "grad_norm": 0.006532685365527868, "learning_rate": 3.906077545664439e-06, "loss": 0.0012, "step": 149900 }, { "epoch": 2.452916632577927, "grad_norm": 0.028413807973265648, "learning_rate": 3.905380900712742e-06, "loss": 0.0008, "step": 149910 }, { "epoch": 2.4530802585290026, "grad_norm": 0.0011403143871575594, "learning_rate": 3.904684278079396e-06, "loss": 0.0007, "step": 149920 }, { "epoch": 2.4532438844800786, "grad_norm": 0.09266354143619537, "learning_rate": 3.9039876777786066e-06, "loss": 0.0012, "step": 149930 }, { "epoch": 2.4534075104311546, "grad_norm": 0.0014889788581058383, "learning_rate": 3.903291099824576e-06, "loss": 0.0005, "step": 149940 }, { "epoch": 2.45357113638223, "grad_norm": 0.010337356477975845, "learning_rate": 3.902594544231508e-06, "loss": 0.0008, "step": 149950 }, { "epoch": 2.453734762333306, "grad_norm": 0.15964964032173157, "learning_rate": 3.901898011013603e-06, "loss": 0.0012, "step": 149960 }, { "epoch": 2.453898388284382, "grad_norm": 0.025364510715007782, "learning_rate": 3.901201500185065e-06, "loss": 0.0008, "step": 149970 }, { "epoch": 2.4540620142354577, "grad_norm": 0.037936415523290634, "learning_rate": 3.900505011760092e-06, "loss": 0.0005, "step": 149980 }, { "epoch": 2.4542256401865337, "grad_norm": 0.0028894376009702682, "learning_rate": 3.899808545752888e-06, "loss": 0.0008, "step": 149990 }, { "epoch": 2.4543892661376097, "grad_norm": 0.04215960577130318, "learning_rate": 3.899112102177651e-06, "loss": 0.0014, "step": 150000 }, { "epoch": 2.454552892088685, "grad_norm": 0.06630747020244598, "learning_rate": 3.898415681048584e-06, "loss": 0.0017, "step": 150010 }, { "epoch": 2.454716518039761, "grad_norm": 0.027254916727542877, "learning_rate": 3.897719282379883e-06, "loss": 0.0007, "step": 150020 }, { "epoch": 2.4548801439908368, "grad_norm": 0.03912115842103958, "learning_rate": 3.897022906185748e-06, "loss": 0.001, "step": 150030 }, { "epoch": 2.4550437699419128, "grad_norm": 0.025284549221396446, "learning_rate": 3.896326552480379e-06, "loss": 0.0008, "step": 150040 }, { "epoch": 2.4552073958929888, "grad_norm": 0.0988224521279335, "learning_rate": 3.895630221277973e-06, "loss": 0.0009, "step": 150050 }, { "epoch": 2.4553710218440643, "grad_norm": 0.019205335527658463, "learning_rate": 3.894933912592726e-06, "loss": 0.0008, "step": 150060 }, { "epoch": 2.4555346477951403, "grad_norm": 0.06480720639228821, "learning_rate": 3.894237626438839e-06, "loss": 0.0008, "step": 150070 }, { "epoch": 2.4556982737462163, "grad_norm": 0.06721252202987671, "learning_rate": 3.893541362830503e-06, "loss": 0.002, "step": 150080 }, { "epoch": 2.455861899697292, "grad_norm": 0.00300317769870162, "learning_rate": 3.8928451217819214e-06, "loss": 0.0007, "step": 150090 }, { "epoch": 2.456025525648368, "grad_norm": 0.03979577496647835, "learning_rate": 3.892148903307283e-06, "loss": 0.0014, "step": 150100 }, { "epoch": 2.456189151599444, "grad_norm": 0.048378270119428635, "learning_rate": 3.8914527074207895e-06, "loss": 0.0008, "step": 150110 }, { "epoch": 2.4563527775505194, "grad_norm": 0.04313550889492035, "learning_rate": 3.890756534136629e-06, "loss": 0.0012, "step": 150120 }, { "epoch": 2.4565164035015954, "grad_norm": 0.0358000248670578, "learning_rate": 3.890060383469002e-06, "loss": 0.001, "step": 150130 }, { "epoch": 2.4566800294526714, "grad_norm": 0.01434353832155466, "learning_rate": 3.889364255432098e-06, "loss": 0.001, "step": 150140 }, { "epoch": 2.456843655403747, "grad_norm": 0.05994269251823425, "learning_rate": 3.888668150040114e-06, "loss": 0.0005, "step": 150150 }, { "epoch": 2.457007281354823, "grad_norm": 0.07507793605327606, "learning_rate": 3.8879720673072395e-06, "loss": 0.0013, "step": 150160 }, { "epoch": 2.4571709073058985, "grad_norm": 0.013419094495475292, "learning_rate": 3.8872760072476715e-06, "loss": 0.0012, "step": 150170 }, { "epoch": 2.4573345332569745, "grad_norm": 0.19039419293403625, "learning_rate": 3.886579969875595e-06, "loss": 0.0017, "step": 150180 }, { "epoch": 2.4574981592080505, "grad_norm": 0.024912377819418907, "learning_rate": 3.8858839552052105e-06, "loss": 0.0009, "step": 150190 }, { "epoch": 2.457661785159126, "grad_norm": 0.0017188511556014419, "learning_rate": 3.885187963250701e-06, "loss": 0.0008, "step": 150200 }, { "epoch": 2.457825411110202, "grad_norm": 0.030016178265213966, "learning_rate": 3.884491994026264e-06, "loss": 0.0008, "step": 150210 }, { "epoch": 2.457989037061278, "grad_norm": 0.01408027671277523, "learning_rate": 3.8837960475460836e-06, "loss": 0.0006, "step": 150220 }, { "epoch": 2.4581526630123536, "grad_norm": 0.02572486735880375, "learning_rate": 3.883100123824355e-06, "loss": 0.0007, "step": 150230 }, { "epoch": 2.4583162889634296, "grad_norm": 0.028705716133117676, "learning_rate": 3.882404222875264e-06, "loss": 0.0007, "step": 150240 }, { "epoch": 2.4584799149145056, "grad_norm": 0.03387998417019844, "learning_rate": 3.881708344713001e-06, "loss": 0.0006, "step": 150250 }, { "epoch": 2.458643540865581, "grad_norm": 0.03931339830160141, "learning_rate": 3.881012489351752e-06, "loss": 0.0007, "step": 150260 }, { "epoch": 2.458807166816657, "grad_norm": 0.12351952493190765, "learning_rate": 3.880316656805709e-06, "loss": 0.0018, "step": 150270 }, { "epoch": 2.458970792767733, "grad_norm": 0.08373849093914032, "learning_rate": 3.879620847089055e-06, "loss": 0.0006, "step": 150280 }, { "epoch": 2.4591344187188087, "grad_norm": 0.08090164512395859, "learning_rate": 3.878925060215981e-06, "loss": 0.0009, "step": 150290 }, { "epoch": 2.4592980446698847, "grad_norm": 0.04372487589716911, "learning_rate": 3.87822929620067e-06, "loss": 0.0009, "step": 150300 }, { "epoch": 2.4594616706209607, "grad_norm": 0.01105529349297285, "learning_rate": 3.87753355505731e-06, "loss": 0.0005, "step": 150310 }, { "epoch": 2.459625296572036, "grad_norm": 0.03775753453373909, "learning_rate": 3.876837836800086e-06, "loss": 0.0012, "step": 150320 }, { "epoch": 2.459788922523112, "grad_norm": 0.22289374470710754, "learning_rate": 3.876142141443184e-06, "loss": 0.0024, "step": 150330 }, { "epoch": 2.459952548474188, "grad_norm": 0.0006960445316508412, "learning_rate": 3.875446469000788e-06, "loss": 0.0008, "step": 150340 }, { "epoch": 2.4601161744252638, "grad_norm": 0.003717527724802494, "learning_rate": 3.874750819487081e-06, "loss": 0.0009, "step": 150350 }, { "epoch": 2.4602798003763398, "grad_norm": 0.04855293780565262, "learning_rate": 3.874055192916248e-06, "loss": 0.0007, "step": 150360 }, { "epoch": 2.4604434263274158, "grad_norm": 0.037694092839956284, "learning_rate": 3.873359589302474e-06, "loss": 0.0012, "step": 150370 }, { "epoch": 2.4606070522784913, "grad_norm": 0.005968938115984201, "learning_rate": 3.872664008659938e-06, "loss": 0.0006, "step": 150380 }, { "epoch": 2.4607706782295673, "grad_norm": 0.1260668933391571, "learning_rate": 3.871968451002824e-06, "loss": 0.0013, "step": 150390 }, { "epoch": 2.4609343041806433, "grad_norm": 0.04627228155732155, "learning_rate": 3.871272916345314e-06, "loss": 0.0013, "step": 150400 }, { "epoch": 2.461097930131719, "grad_norm": 0.01198414247483015, "learning_rate": 3.87057740470159e-06, "loss": 0.0006, "step": 150410 }, { "epoch": 2.461261556082795, "grad_norm": 0.009334263391792774, "learning_rate": 3.869881916085832e-06, "loss": 0.0007, "step": 150420 }, { "epoch": 2.4614251820338704, "grad_norm": 0.021210310980677605, "learning_rate": 3.86918645051222e-06, "loss": 0.001, "step": 150430 }, { "epoch": 2.4615888079849464, "grad_norm": 0.05982343479990959, "learning_rate": 3.868491007994935e-06, "loss": 0.0006, "step": 150440 }, { "epoch": 2.4617524339360224, "grad_norm": 0.0641832947731018, "learning_rate": 3.867795588548157e-06, "loss": 0.0009, "step": 150450 }, { "epoch": 2.461916059887098, "grad_norm": 0.24870191514492035, "learning_rate": 3.8671001921860614e-06, "loss": 0.001, "step": 150460 }, { "epoch": 2.462079685838174, "grad_norm": 0.04824231564998627, "learning_rate": 3.866404818922832e-06, "loss": 0.0013, "step": 150470 }, { "epoch": 2.46224331178925, "grad_norm": 0.06102379038929939, "learning_rate": 3.8657094687726424e-06, "loss": 0.0011, "step": 150480 }, { "epoch": 2.4624069377403255, "grad_norm": 0.007166668307036161, "learning_rate": 3.865014141749674e-06, "loss": 0.0008, "step": 150490 }, { "epoch": 2.4625705636914015, "grad_norm": 0.03476399928331375, "learning_rate": 3.8643188378681e-06, "loss": 0.0008, "step": 150500 }, { "epoch": 2.4627341896424775, "grad_norm": 0.021728601306676865, "learning_rate": 3.863623557142101e-06, "loss": 0.0011, "step": 150510 }, { "epoch": 2.462897815593553, "grad_norm": 0.022494859993457794, "learning_rate": 3.86292829958585e-06, "loss": 0.0006, "step": 150520 }, { "epoch": 2.463061441544629, "grad_norm": 0.03425291180610657, "learning_rate": 3.862233065213524e-06, "loss": 0.0008, "step": 150530 }, { "epoch": 2.4632250674957046, "grad_norm": 0.032602980732917786, "learning_rate": 3.861537854039298e-06, "loss": 0.0021, "step": 150540 }, { "epoch": 2.4633886934467806, "grad_norm": 0.04245387017726898, "learning_rate": 3.860842666077348e-06, "loss": 0.0009, "step": 150550 }, { "epoch": 2.4635523193978566, "grad_norm": 0.03252383694052696, "learning_rate": 3.8601475013418455e-06, "loss": 0.0006, "step": 150560 }, { "epoch": 2.463715945348932, "grad_norm": 0.036053527146577835, "learning_rate": 3.8594523598469674e-06, "loss": 0.0005, "step": 150570 }, { "epoch": 2.463879571300008, "grad_norm": 0.023739773780107498, "learning_rate": 3.858757241606884e-06, "loss": 0.0013, "step": 150580 }, { "epoch": 2.464043197251084, "grad_norm": 0.023039907217025757, "learning_rate": 3.858062146635772e-06, "loss": 0.0005, "step": 150590 }, { "epoch": 2.4642068232021597, "grad_norm": 0.001466132584027946, "learning_rate": 3.8573670749478005e-06, "loss": 0.0005, "step": 150600 }, { "epoch": 2.4643704491532357, "grad_norm": 0.02234429493546486, "learning_rate": 3.856672026557143e-06, "loss": 0.0004, "step": 150610 }, { "epoch": 2.4645340751043117, "grad_norm": 0.006178558804094791, "learning_rate": 3.85597700147797e-06, "loss": 0.0008, "step": 150620 }, { "epoch": 2.464697701055387, "grad_norm": 0.04415585845708847, "learning_rate": 3.855281999724454e-06, "loss": 0.0008, "step": 150630 }, { "epoch": 2.464861327006463, "grad_norm": 0.08495564758777618, "learning_rate": 3.8545870213107626e-06, "loss": 0.0005, "step": 150640 }, { "epoch": 2.465024952957539, "grad_norm": 0.009712014347314835, "learning_rate": 3.85389206625107e-06, "loss": 0.001, "step": 150650 }, { "epoch": 2.4651885789086148, "grad_norm": 0.011834766715765, "learning_rate": 3.853197134559541e-06, "loss": 0.0008, "step": 150660 }, { "epoch": 2.4653522048596908, "grad_norm": 0.05994366109371185, "learning_rate": 3.85250222625035e-06, "loss": 0.0011, "step": 150670 }, { "epoch": 2.4655158308107668, "grad_norm": 0.03619290515780449, "learning_rate": 3.85180734133766e-06, "loss": 0.0007, "step": 150680 }, { "epoch": 2.4656794567618423, "grad_norm": 0.09952894598245621, "learning_rate": 3.851112479835643e-06, "loss": 0.002, "step": 150690 }, { "epoch": 2.4658430827129183, "grad_norm": 0.08963082730770111, "learning_rate": 3.850417641758466e-06, "loss": 0.0005, "step": 150700 }, { "epoch": 2.4660067086639943, "grad_norm": 0.00589312007650733, "learning_rate": 3.849722827120296e-06, "loss": 0.0009, "step": 150710 }, { "epoch": 2.46617033461507, "grad_norm": 0.04771585762500763, "learning_rate": 3.849028035935296e-06, "loss": 0.0009, "step": 150720 }, { "epoch": 2.466333960566146, "grad_norm": 0.016357069835066795, "learning_rate": 3.84833326821764e-06, "loss": 0.0005, "step": 150730 }, { "epoch": 2.466497586517222, "grad_norm": 0.003747503738850355, "learning_rate": 3.847638523981485e-06, "loss": 0.0012, "step": 150740 }, { "epoch": 2.4666612124682974, "grad_norm": 0.0031007423531264067, "learning_rate": 3.846943803241004e-06, "loss": 0.0003, "step": 150750 }, { "epoch": 2.4668248384193734, "grad_norm": 0.05772589147090912, "learning_rate": 3.846249106010355e-06, "loss": 0.0014, "step": 150760 }, { "epoch": 2.4669884643704494, "grad_norm": 0.028420034795999527, "learning_rate": 3.845554432303709e-06, "loss": 0.0006, "step": 150770 }, { "epoch": 2.467152090321525, "grad_norm": 0.030320987105369568, "learning_rate": 3.844859782135222e-06, "loss": 0.0004, "step": 150780 }, { "epoch": 2.467315716272601, "grad_norm": 0.06403105705976486, "learning_rate": 3.844165155519066e-06, "loss": 0.0011, "step": 150790 }, { "epoch": 2.4674793422236765, "grad_norm": 0.00739277433604002, "learning_rate": 3.843470552469396e-06, "loss": 0.0004, "step": 150800 }, { "epoch": 2.4676429681747525, "grad_norm": 0.010279043577611446, "learning_rate": 3.842775973000381e-06, "loss": 0.0007, "step": 150810 }, { "epoch": 2.4678065941258285, "grad_norm": 0.015239723958075047, "learning_rate": 3.842081417126176e-06, "loss": 0.0005, "step": 150820 }, { "epoch": 2.467970220076904, "grad_norm": 0.07988087087869644, "learning_rate": 3.84138688486095e-06, "loss": 0.0009, "step": 150830 }, { "epoch": 2.46813384602798, "grad_norm": 0.029171990230679512, "learning_rate": 3.840692376218856e-06, "loss": 0.0007, "step": 150840 }, { "epoch": 2.468297471979056, "grad_norm": 0.055349960923194885, "learning_rate": 3.839997891214063e-06, "loss": 0.0009, "step": 150850 }, { "epoch": 2.4684610979301316, "grad_norm": 0.1204259917140007, "learning_rate": 3.839303429860722e-06, "loss": 0.0009, "step": 150860 }, { "epoch": 2.4686247238812076, "grad_norm": 0.043716251850128174, "learning_rate": 3.838608992173001e-06, "loss": 0.0009, "step": 150870 }, { "epoch": 2.4687883498322836, "grad_norm": 0.00241474830545485, "learning_rate": 3.837914578165053e-06, "loss": 0.0007, "step": 150880 }, { "epoch": 2.468951975783359, "grad_norm": 0.10676749050617218, "learning_rate": 3.8372201878510385e-06, "loss": 0.001, "step": 150890 }, { "epoch": 2.469115601734435, "grad_norm": 0.025970669463276863, "learning_rate": 3.836525821245115e-06, "loss": 0.0004, "step": 150900 }, { "epoch": 2.469279227685511, "grad_norm": 0.032503873109817505, "learning_rate": 3.8358314783614415e-06, "loss": 0.0009, "step": 150910 }, { "epoch": 2.4694428536365867, "grad_norm": 0.017456937581300735, "learning_rate": 3.835137159214173e-06, "loss": 0.0003, "step": 150920 }, { "epoch": 2.4696064795876627, "grad_norm": 0.026635563001036644, "learning_rate": 3.834442863817468e-06, "loss": 0.0006, "step": 150930 }, { "epoch": 2.469770105538738, "grad_norm": 0.1383848786354065, "learning_rate": 3.833748592185481e-06, "loss": 0.001, "step": 150940 }, { "epoch": 2.469933731489814, "grad_norm": 0.022273650392889977, "learning_rate": 3.833054344332369e-06, "loss": 0.0004, "step": 150950 }, { "epoch": 2.47009735744089, "grad_norm": 0.030892780050635338, "learning_rate": 3.832360120272286e-06, "loss": 0.0006, "step": 150960 }, { "epoch": 2.4702609833919658, "grad_norm": 0.3077205717563629, "learning_rate": 3.831665920019388e-06, "loss": 0.0024, "step": 150970 }, { "epoch": 2.4704246093430418, "grad_norm": 0.030392801389098167, "learning_rate": 3.830971743587827e-06, "loss": 0.0008, "step": 150980 }, { "epoch": 2.4705882352941178, "grad_norm": 0.02726469747722149, "learning_rate": 3.8302775909917585e-06, "loss": 0.0005, "step": 150990 }, { "epoch": 2.4707518612451933, "grad_norm": 0.047679491341114044, "learning_rate": 3.829583462245335e-06, "loss": 0.0011, "step": 151000 }, { "epoch": 2.4709154871962693, "grad_norm": 0.12461323291063309, "learning_rate": 3.82888935736271e-06, "loss": 0.0007, "step": 151010 }, { "epoch": 2.4710791131473453, "grad_norm": 0.05449671298265457, "learning_rate": 3.8281952763580335e-06, "loss": 0.0004, "step": 151020 }, { "epoch": 2.471242739098421, "grad_norm": 0.02003658004105091, "learning_rate": 3.8275012192454605e-06, "loss": 0.0009, "step": 151030 }, { "epoch": 2.471406365049497, "grad_norm": 0.003741421038284898, "learning_rate": 3.826807186039139e-06, "loss": 0.0006, "step": 151040 }, { "epoch": 2.471569991000573, "grad_norm": 0.0031020038295537233, "learning_rate": 3.826113176753222e-06, "loss": 0.001, "step": 151050 }, { "epoch": 2.4717336169516484, "grad_norm": 0.02073070965707302, "learning_rate": 3.825419191401858e-06, "loss": 0.0009, "step": 151060 }, { "epoch": 2.4718972429027244, "grad_norm": 0.08472375571727753, "learning_rate": 3.824725229999199e-06, "loss": 0.0005, "step": 151070 }, { "epoch": 2.4720608688538004, "grad_norm": 0.0406053364276886, "learning_rate": 3.824031292559392e-06, "loss": 0.0011, "step": 151080 }, { "epoch": 2.472224494804876, "grad_norm": 0.002613699994981289, "learning_rate": 3.8233373790965885e-06, "loss": 0.0006, "step": 151090 }, { "epoch": 2.472388120755952, "grad_norm": 0.035503510385751724, "learning_rate": 3.8226434896249334e-06, "loss": 0.0003, "step": 151100 }, { "epoch": 2.472551746707028, "grad_norm": 0.08003028482198715, "learning_rate": 3.8219496241585775e-06, "loss": 0.0014, "step": 151110 }, { "epoch": 2.4727153726581035, "grad_norm": 0.0209655724465847, "learning_rate": 3.821255782711665e-06, "loss": 0.0009, "step": 151120 }, { "epoch": 2.4728789986091795, "grad_norm": 0.18958573043346405, "learning_rate": 3.8205619652983465e-06, "loss": 0.0016, "step": 151130 }, { "epoch": 2.4730426245602555, "grad_norm": 0.016589241102337837, "learning_rate": 3.819868171932765e-06, "loss": 0.0007, "step": 151140 }, { "epoch": 2.473206250511331, "grad_norm": 0.019307738170027733, "learning_rate": 3.819174402629069e-06, "loss": 0.0009, "step": 151150 }, { "epoch": 2.473369876462407, "grad_norm": 0.0549454465508461, "learning_rate": 3.8184806574014025e-06, "loss": 0.0007, "step": 151160 }, { "epoch": 2.473533502413483, "grad_norm": 0.024352040141820908, "learning_rate": 3.81778693626391e-06, "loss": 0.0006, "step": 151170 }, { "epoch": 2.4736971283645586, "grad_norm": 0.14320364594459534, "learning_rate": 3.817093239230737e-06, "loss": 0.0007, "step": 151180 }, { "epoch": 2.4738607543156346, "grad_norm": 0.04872163012623787, "learning_rate": 3.816399566316027e-06, "loss": 0.001, "step": 151190 }, { "epoch": 2.47402438026671, "grad_norm": 0.02875504083931446, "learning_rate": 3.815705917533922e-06, "loss": 0.0005, "step": 151200 }, { "epoch": 2.474188006217786, "grad_norm": 0.026243086904287338, "learning_rate": 3.815012292898567e-06, "loss": 0.0006, "step": 151210 }, { "epoch": 2.474351632168862, "grad_norm": 0.07170163840055466, "learning_rate": 3.814318692424103e-06, "loss": 0.0006, "step": 151220 }, { "epoch": 2.4745152581199377, "grad_norm": 0.029445834457874298, "learning_rate": 3.813625116124674e-06, "loss": 0.0007, "step": 151230 }, { "epoch": 2.4746788840710137, "grad_norm": 0.033624276518821716, "learning_rate": 3.8129315640144183e-06, "loss": 0.0008, "step": 151240 }, { "epoch": 2.4748425100220897, "grad_norm": 0.030050192028284073, "learning_rate": 3.812238036107479e-06, "loss": 0.0006, "step": 151250 }, { "epoch": 2.475006135973165, "grad_norm": 0.0470137856900692, "learning_rate": 3.8115445324179955e-06, "loss": 0.0008, "step": 151260 }, { "epoch": 2.475169761924241, "grad_norm": 0.07270277291536331, "learning_rate": 3.8108510529601094e-06, "loss": 0.0005, "step": 151270 }, { "epoch": 2.475333387875317, "grad_norm": 0.0028279228135943413, "learning_rate": 3.810157597747958e-06, "loss": 0.0008, "step": 151280 }, { "epoch": 2.4754970138263928, "grad_norm": 0.0508652962744236, "learning_rate": 3.8094641667956815e-06, "loss": 0.0012, "step": 151290 }, { "epoch": 2.4756606397774688, "grad_norm": 0.0439949631690979, "learning_rate": 3.808770760117418e-06, "loss": 0.0005, "step": 151300 }, { "epoch": 2.4758242657285443, "grad_norm": 0.15863211452960968, "learning_rate": 3.808077377727306e-06, "loss": 0.0014, "step": 151310 }, { "epoch": 2.4759878916796203, "grad_norm": 0.01772797852754593, "learning_rate": 3.807384019639482e-06, "loss": 0.0008, "step": 151320 }, { "epoch": 2.4761515176306963, "grad_norm": 0.008894248865544796, "learning_rate": 3.8066906858680846e-06, "loss": 0.0009, "step": 151330 }, { "epoch": 2.476315143581772, "grad_norm": 0.0041593704372644424, "learning_rate": 3.805997376427248e-06, "loss": 0.0008, "step": 151340 }, { "epoch": 2.476478769532848, "grad_norm": 0.09944803267717361, "learning_rate": 3.805304091331111e-06, "loss": 0.0012, "step": 151350 }, { "epoch": 2.476642395483924, "grad_norm": 0.0035856382455676794, "learning_rate": 3.8046108305938057e-06, "loss": 0.0008, "step": 151360 }, { "epoch": 2.4768060214349994, "grad_norm": 0.09392152726650238, "learning_rate": 3.8039175942294707e-06, "loss": 0.001, "step": 151370 }, { "epoch": 2.4769696473860754, "grad_norm": 0.053814347833395004, "learning_rate": 3.8032243822522362e-06, "loss": 0.0008, "step": 151380 }, { "epoch": 2.4771332733371514, "grad_norm": 0.01616041548550129, "learning_rate": 3.802531194676242e-06, "loss": 0.0009, "step": 151390 }, { "epoch": 2.477296899288227, "grad_norm": 0.05384358391165733, "learning_rate": 3.801838031515615e-06, "loss": 0.001, "step": 151400 }, { "epoch": 2.477460525239303, "grad_norm": 0.014174113050103188, "learning_rate": 3.8011448927844954e-06, "loss": 0.0002, "step": 151410 }, { "epoch": 2.477624151190379, "grad_norm": 0.21926674246788025, "learning_rate": 3.8004517784970072e-06, "loss": 0.0012, "step": 151420 }, { "epoch": 2.4777877771414545, "grad_norm": 0.07072693854570389, "learning_rate": 3.799758688667291e-06, "loss": 0.0011, "step": 151430 }, { "epoch": 2.4779514030925305, "grad_norm": 0.059458862990140915, "learning_rate": 3.7990656233094712e-06, "loss": 0.0007, "step": 151440 }, { "epoch": 2.4781150290436065, "grad_norm": 0.10889916121959686, "learning_rate": 3.7983725824376854e-06, "loss": 0.0008, "step": 151450 }, { "epoch": 2.478278654994682, "grad_norm": 0.05099314823746681, "learning_rate": 3.7976795660660574e-06, "loss": 0.0007, "step": 151460 }, { "epoch": 2.478442280945758, "grad_norm": 0.01595246233046055, "learning_rate": 3.7969865742087238e-06, "loss": 0.0007, "step": 151470 }, { "epoch": 2.478605906896834, "grad_norm": 0.07518425583839417, "learning_rate": 3.7962936068798075e-06, "loss": 0.0009, "step": 151480 }, { "epoch": 2.4787695328479096, "grad_norm": 0.09797260910272598, "learning_rate": 3.795600664093444e-06, "loss": 0.001, "step": 151490 }, { "epoch": 2.4789331587989856, "grad_norm": 0.05285646393895149, "learning_rate": 3.794907745863755e-06, "loss": 0.0015, "step": 151500 }, { "epoch": 2.4790967847500616, "grad_norm": 0.018572580069303513, "learning_rate": 3.794214852204876e-06, "loss": 0.0009, "step": 151510 }, { "epoch": 2.479260410701137, "grad_norm": 0.0184831190854311, "learning_rate": 3.793521983130927e-06, "loss": 0.0004, "step": 151520 }, { "epoch": 2.479424036652213, "grad_norm": 0.03734654188156128, "learning_rate": 3.7928291386560422e-06, "loss": 0.0016, "step": 151530 }, { "epoch": 2.479587662603289, "grad_norm": 0.06384141743183136, "learning_rate": 3.7921363187943427e-06, "loss": 0.0011, "step": 151540 }, { "epoch": 2.4797512885543647, "grad_norm": 0.016456186771392822, "learning_rate": 3.7914435235599572e-06, "loss": 0.0009, "step": 151550 }, { "epoch": 2.4799149145054407, "grad_norm": 0.07929002493619919, "learning_rate": 3.7907507529670095e-06, "loss": 0.0013, "step": 151560 }, { "epoch": 2.480078540456516, "grad_norm": 0.037976451218128204, "learning_rate": 3.790058007029626e-06, "loss": 0.0013, "step": 151570 }, { "epoch": 2.480242166407592, "grad_norm": 0.022426443174481392, "learning_rate": 3.78936528576193e-06, "loss": 0.0005, "step": 151580 }, { "epoch": 2.480405792358668, "grad_norm": 0.023532312363386154, "learning_rate": 3.7886725891780474e-06, "loss": 0.0007, "step": 151590 }, { "epoch": 2.4805694183097438, "grad_norm": 0.03768516331911087, "learning_rate": 3.7879799172920996e-06, "loss": 0.0007, "step": 151600 }, { "epoch": 2.4807330442608198, "grad_norm": 0.01480037346482277, "learning_rate": 3.787287270118212e-06, "loss": 0.0008, "step": 151610 }, { "epoch": 2.4808966702118957, "grad_norm": 0.013139540329575539, "learning_rate": 3.7865946476705033e-06, "loss": 0.0008, "step": 151620 }, { "epoch": 2.4810602961629713, "grad_norm": 0.06965295970439911, "learning_rate": 3.7859020499630995e-06, "loss": 0.0008, "step": 151630 }, { "epoch": 2.4812239221140473, "grad_norm": 0.020787913352251053, "learning_rate": 3.785209477010119e-06, "loss": 0.0007, "step": 151640 }, { "epoch": 2.4813875480651233, "grad_norm": 0.05768759176135063, "learning_rate": 3.7845169288256855e-06, "loss": 0.0009, "step": 151650 }, { "epoch": 2.481551174016199, "grad_norm": 0.020154956728219986, "learning_rate": 3.783824405423917e-06, "loss": 0.0009, "step": 151660 }, { "epoch": 2.481714799967275, "grad_norm": 0.09473145008087158, "learning_rate": 3.7831319068189353e-06, "loss": 0.0013, "step": 151670 }, { "epoch": 2.481878425918351, "grad_norm": 0.07397446036338806, "learning_rate": 3.782439433024859e-06, "loss": 0.0013, "step": 151680 }, { "epoch": 2.4820420518694264, "grad_norm": 0.01561837736517191, "learning_rate": 3.7817469840558067e-06, "loss": 0.0008, "step": 151690 }, { "epoch": 2.4822056778205024, "grad_norm": 0.11363254487514496, "learning_rate": 3.7810545599258973e-06, "loss": 0.0007, "step": 151700 }, { "epoch": 2.482369303771578, "grad_norm": 0.05949949100613594, "learning_rate": 3.7803621606492497e-06, "loss": 0.0012, "step": 151710 }, { "epoch": 2.482532929722654, "grad_norm": 0.12574787437915802, "learning_rate": 3.779669786239979e-06, "loss": 0.0011, "step": 151720 }, { "epoch": 2.48269655567373, "grad_norm": 0.011194161139428616, "learning_rate": 3.7789774367122053e-06, "loss": 0.0008, "step": 151730 }, { "epoch": 2.4828601816248055, "grad_norm": 0.047558270394802094, "learning_rate": 3.7782851120800413e-06, "loss": 0.0007, "step": 151740 }, { "epoch": 2.4830238075758815, "grad_norm": 0.10429278761148453, "learning_rate": 3.777592812357606e-06, "loss": 0.0009, "step": 151750 }, { "epoch": 2.4831874335269575, "grad_norm": 0.17985619604587555, "learning_rate": 3.776900537559013e-06, "loss": 0.0013, "step": 151760 }, { "epoch": 2.483351059478033, "grad_norm": 0.026901252567768097, "learning_rate": 3.7762082876983787e-06, "loss": 0.0014, "step": 151770 }, { "epoch": 2.483514685429109, "grad_norm": 0.02004210837185383, "learning_rate": 3.7755160627898156e-06, "loss": 0.001, "step": 151780 }, { "epoch": 2.483678311380185, "grad_norm": 0.040928471833467484, "learning_rate": 3.7748238628474393e-06, "loss": 0.001, "step": 151790 }, { "epoch": 2.4838419373312606, "grad_norm": 0.032806627452373505, "learning_rate": 3.774131687885362e-06, "loss": 0.0005, "step": 151800 }, { "epoch": 2.4840055632823366, "grad_norm": 0.051358338445425034, "learning_rate": 3.7734395379176974e-06, "loss": 0.0017, "step": 151810 }, { "epoch": 2.4841691892334126, "grad_norm": 0.04380761831998825, "learning_rate": 3.7727474129585575e-06, "loss": 0.0011, "step": 151820 }, { "epoch": 2.484332815184488, "grad_norm": 0.028974037617444992, "learning_rate": 3.772055313022054e-06, "loss": 0.0004, "step": 151830 }, { "epoch": 2.484496441135564, "grad_norm": 0.052193570882081985, "learning_rate": 3.7713632381222974e-06, "loss": 0.0003, "step": 151840 }, { "epoch": 2.48466006708664, "grad_norm": 0.045333825051784515, "learning_rate": 3.770671188273401e-06, "loss": 0.0005, "step": 151850 }, { "epoch": 2.4848236930377157, "grad_norm": 0.08205941319465637, "learning_rate": 3.769979163489472e-06, "loss": 0.0008, "step": 151860 }, { "epoch": 2.4849873189887917, "grad_norm": 0.006576797924935818, "learning_rate": 3.7692871637846233e-06, "loss": 0.0006, "step": 151870 }, { "epoch": 2.4851509449398677, "grad_norm": 0.05536860600113869, "learning_rate": 3.7685951891729616e-06, "loss": 0.0012, "step": 151880 }, { "epoch": 2.485314570890943, "grad_norm": 0.05259178951382637, "learning_rate": 3.7679032396685978e-06, "loss": 0.0005, "step": 151890 }, { "epoch": 2.485478196842019, "grad_norm": 0.030652059242129326, "learning_rate": 3.7672113152856383e-06, "loss": 0.0007, "step": 151900 }, { "epoch": 2.485641822793095, "grad_norm": 0.051156532019376755, "learning_rate": 3.766519416038192e-06, "loss": 0.0009, "step": 151910 }, { "epoch": 2.4858054487441708, "grad_norm": 0.015974372625350952, "learning_rate": 3.7658275419403644e-06, "loss": 0.001, "step": 151920 }, { "epoch": 2.4859690746952467, "grad_norm": 0.04290607571601868, "learning_rate": 3.7651356930062654e-06, "loss": 0.0012, "step": 151930 }, { "epoch": 2.4861327006463227, "grad_norm": 0.08478474617004395, "learning_rate": 3.7644438692499985e-06, "loss": 0.0008, "step": 151940 }, { "epoch": 2.4862963265973983, "grad_norm": 0.037134140729904175, "learning_rate": 3.7637520706856713e-06, "loss": 0.001, "step": 151950 }, { "epoch": 2.4864599525484743, "grad_norm": 0.06707217544317245, "learning_rate": 3.763060297327386e-06, "loss": 0.0006, "step": 151960 }, { "epoch": 2.48662357849955, "grad_norm": 0.009292224422097206, "learning_rate": 3.7623685491892516e-06, "loss": 0.0005, "step": 151970 }, { "epoch": 2.486787204450626, "grad_norm": 0.04188608005642891, "learning_rate": 3.7616768262853674e-06, "loss": 0.001, "step": 151980 }, { "epoch": 2.486950830401702, "grad_norm": 0.029628779739141464, "learning_rate": 3.7609851286298426e-06, "loss": 0.0005, "step": 151990 }, { "epoch": 2.4871144563527774, "grad_norm": 0.018811969086527824, "learning_rate": 3.7602934562367743e-06, "loss": 0.0006, "step": 152000 }, { "epoch": 2.4872780823038534, "grad_norm": 0.029387904331088066, "learning_rate": 3.759601809120271e-06, "loss": 0.001, "step": 152010 }, { "epoch": 2.4874417082549294, "grad_norm": 0.05854375660419464, "learning_rate": 3.758910187294429e-06, "loss": 0.0007, "step": 152020 }, { "epoch": 2.487605334206005, "grad_norm": 0.13472820818424225, "learning_rate": 3.7582185907733552e-06, "loss": 0.0007, "step": 152030 }, { "epoch": 2.487768960157081, "grad_norm": 0.06751329451799393, "learning_rate": 3.7575270195711453e-06, "loss": 0.0007, "step": 152040 }, { "epoch": 2.487932586108157, "grad_norm": 0.05524401739239693, "learning_rate": 3.7568354737019066e-06, "loss": 0.0007, "step": 152050 }, { "epoch": 2.4880962120592325, "grad_norm": 0.014311235398054123, "learning_rate": 3.7561439531797315e-06, "loss": 0.0005, "step": 152060 }, { "epoch": 2.4882598380103085, "grad_norm": 0.09982415288686752, "learning_rate": 3.7554524580187266e-06, "loss": 0.0013, "step": 152070 }, { "epoch": 2.488423463961384, "grad_norm": 0.06492825597524643, "learning_rate": 3.754760988232985e-06, "loss": 0.0006, "step": 152080 }, { "epoch": 2.48858708991246, "grad_norm": 0.08782383799552917, "learning_rate": 3.7540695438366106e-06, "loss": 0.0011, "step": 152090 }, { "epoch": 2.488750715863536, "grad_norm": 0.02861202135682106, "learning_rate": 3.753378124843695e-06, "loss": 0.0007, "step": 152100 }, { "epoch": 2.4889143418146116, "grad_norm": 0.032502952963113785, "learning_rate": 3.752686731268343e-06, "loss": 0.001, "step": 152110 }, { "epoch": 2.4890779677656876, "grad_norm": 0.04632508009672165, "learning_rate": 3.7519953631246443e-06, "loss": 0.001, "step": 152120 }, { "epoch": 2.4892415937167636, "grad_norm": 0.043235067278146744, "learning_rate": 3.7513040204267015e-06, "loss": 0.0013, "step": 152130 }, { "epoch": 2.489405219667839, "grad_norm": 0.03284946829080582, "learning_rate": 3.750612703188605e-06, "loss": 0.0013, "step": 152140 }, { "epoch": 2.489568845618915, "grad_norm": 0.05344406142830849, "learning_rate": 3.7499214114244564e-06, "loss": 0.0007, "step": 152150 }, { "epoch": 2.489732471569991, "grad_norm": 0.05180743336677551, "learning_rate": 3.749230145148344e-06, "loss": 0.0006, "step": 152160 }, { "epoch": 2.4898960975210667, "grad_norm": 0.044459473341703415, "learning_rate": 3.7485389043743677e-06, "loss": 0.0007, "step": 152170 }, { "epoch": 2.4900597234721427, "grad_norm": 0.04167299345135689, "learning_rate": 3.747847689116617e-06, "loss": 0.0009, "step": 152180 }, { "epoch": 2.4902233494232187, "grad_norm": 0.017656859010457993, "learning_rate": 3.747156499389188e-06, "loss": 0.0007, "step": 152190 }, { "epoch": 2.490386975374294, "grad_norm": 0.02330920286476612, "learning_rate": 3.7464653352061714e-06, "loss": 0.0007, "step": 152200 }, { "epoch": 2.49055060132537, "grad_norm": 0.03366296738386154, "learning_rate": 3.745774196581662e-06, "loss": 0.0005, "step": 152210 }, { "epoch": 2.490714227276446, "grad_norm": 0.023227551952004433, "learning_rate": 3.745083083529748e-06, "loss": 0.0009, "step": 152220 }, { "epoch": 2.4908778532275218, "grad_norm": 0.17391625046730042, "learning_rate": 3.7443919960645247e-06, "loss": 0.0016, "step": 152230 }, { "epoch": 2.4910414791785978, "grad_norm": 0.004544627387076616, "learning_rate": 3.743700934200079e-06, "loss": 0.0004, "step": 152240 }, { "epoch": 2.4912051051296737, "grad_norm": 0.09929653257131577, "learning_rate": 3.7430098979505046e-06, "loss": 0.0004, "step": 152250 }, { "epoch": 2.4913687310807493, "grad_norm": 0.016457511112093925, "learning_rate": 3.7423188873298877e-06, "loss": 0.0011, "step": 152260 }, { "epoch": 2.4915323570318253, "grad_norm": 0.07250604033470154, "learning_rate": 3.7416279023523205e-06, "loss": 0.0008, "step": 152270 }, { "epoch": 2.4916959829829013, "grad_norm": 0.10150414705276489, "learning_rate": 3.7409369430318886e-06, "loss": 0.001, "step": 152280 }, { "epoch": 2.491859608933977, "grad_norm": 0.0571080707013607, "learning_rate": 3.7402460093826833e-06, "loss": 0.0007, "step": 152290 }, { "epoch": 2.492023234885053, "grad_norm": 0.005372182931751013, "learning_rate": 3.739555101418789e-06, "loss": 0.0009, "step": 152300 }, { "epoch": 2.492186860836129, "grad_norm": 0.009854274801909924, "learning_rate": 3.7388642191542957e-06, "loss": 0.0007, "step": 152310 }, { "epoch": 2.4923504867872044, "grad_norm": 0.005789211485534906, "learning_rate": 3.7381733626032867e-06, "loss": 0.0008, "step": 152320 }, { "epoch": 2.4925141127382804, "grad_norm": 0.03836970403790474, "learning_rate": 3.7374825317798514e-06, "loss": 0.0007, "step": 152330 }, { "epoch": 2.4926777386893564, "grad_norm": 0.04476545378565788, "learning_rate": 3.736791726698072e-06, "loss": 0.0006, "step": 152340 }, { "epoch": 2.492841364640432, "grad_norm": 0.039484038949012756, "learning_rate": 3.736100947372036e-06, "loss": 0.0013, "step": 152350 }, { "epoch": 2.493004990591508, "grad_norm": 0.053069911897182465, "learning_rate": 3.735410193815826e-06, "loss": 0.0007, "step": 152360 }, { "epoch": 2.4931686165425835, "grad_norm": 0.050686903297901154, "learning_rate": 3.7347194660435274e-06, "loss": 0.0005, "step": 152370 }, { "epoch": 2.4933322424936595, "grad_norm": 0.039320096373558044, "learning_rate": 3.7340287640692224e-06, "loss": 0.0005, "step": 152380 }, { "epoch": 2.4934958684447355, "grad_norm": 0.033333610743284225, "learning_rate": 3.7333380879069945e-06, "loss": 0.0007, "step": 152390 }, { "epoch": 2.493659494395811, "grad_norm": 0.03701972961425781, "learning_rate": 3.7326474375709253e-06, "loss": 0.0004, "step": 152400 }, { "epoch": 2.493823120346887, "grad_norm": 0.03868063539266586, "learning_rate": 3.7319568130750983e-06, "loss": 0.0005, "step": 152410 }, { "epoch": 2.493986746297963, "grad_norm": 0.0017838068306446075, "learning_rate": 3.7312662144335913e-06, "loss": 0.0009, "step": 152420 }, { "epoch": 2.4941503722490386, "grad_norm": 0.023916402831673622, "learning_rate": 3.7305756416604887e-06, "loss": 0.0006, "step": 152430 }, { "epoch": 2.4943139982001146, "grad_norm": 0.022011112421751022, "learning_rate": 3.7298850947698682e-06, "loss": 0.0009, "step": 152440 }, { "epoch": 2.4944776241511906, "grad_norm": 0.045695580542087555, "learning_rate": 3.729194573775812e-06, "loss": 0.0009, "step": 152450 }, { "epoch": 2.494641250102266, "grad_norm": 0.013005943968892097, "learning_rate": 3.728504078692396e-06, "loss": 0.0007, "step": 152460 }, { "epoch": 2.494804876053342, "grad_norm": 0.013965762220323086, "learning_rate": 3.7278136095337013e-06, "loss": 0.0018, "step": 152470 }, { "epoch": 2.4949685020044177, "grad_norm": 0.01064684335142374, "learning_rate": 3.727123166313805e-06, "loss": 0.0009, "step": 152480 }, { "epoch": 2.4951321279554937, "grad_norm": 0.06621342152357101, "learning_rate": 3.7264327490467856e-06, "loss": 0.0008, "step": 152490 }, { "epoch": 2.4952957539065697, "grad_norm": 0.011288933455944061, "learning_rate": 3.725742357746718e-06, "loss": 0.0005, "step": 152500 }, { "epoch": 2.495459379857645, "grad_norm": 0.03345746546983719, "learning_rate": 3.7250519924276805e-06, "loss": 0.0008, "step": 152510 }, { "epoch": 2.495623005808721, "grad_norm": 0.0022906933445483446, "learning_rate": 3.724361653103748e-06, "loss": 0.0007, "step": 152520 }, { "epoch": 2.495786631759797, "grad_norm": 0.1808300018310547, "learning_rate": 3.723671339788998e-06, "loss": 0.0013, "step": 152530 }, { "epoch": 2.4959502577108728, "grad_norm": 0.04943521320819855, "learning_rate": 3.722981052497502e-06, "loss": 0.0004, "step": 152540 }, { "epoch": 2.4961138836619488, "grad_norm": 0.05104607343673706, "learning_rate": 3.7222907912433382e-06, "loss": 0.0009, "step": 152550 }, { "epoch": 2.4962775096130247, "grad_norm": 0.040960896760225296, "learning_rate": 3.7216005560405767e-06, "loss": 0.0012, "step": 152560 }, { "epoch": 2.4964411355641003, "grad_norm": 0.00038787961238995194, "learning_rate": 3.7209103469032936e-06, "loss": 0.0006, "step": 152570 }, { "epoch": 2.4966047615151763, "grad_norm": 0.018470462411642075, "learning_rate": 3.7202201638455603e-06, "loss": 0.0006, "step": 152580 }, { "epoch": 2.4967683874662523, "grad_norm": 0.024462860077619553, "learning_rate": 3.71953000688145e-06, "loss": 0.0008, "step": 152590 }, { "epoch": 2.496932013417328, "grad_norm": 0.013044699095189571, "learning_rate": 3.7188398760250333e-06, "loss": 0.0005, "step": 152600 }, { "epoch": 2.497095639368404, "grad_norm": 0.047523703426122665, "learning_rate": 3.7181497712903826e-06, "loss": 0.0008, "step": 152610 }, { "epoch": 2.49725926531948, "grad_norm": 0.042164359241724014, "learning_rate": 3.7174596926915676e-06, "loss": 0.0014, "step": 152620 }, { "epoch": 2.4974228912705554, "grad_norm": 0.02327102981507778, "learning_rate": 3.7167696402426602e-06, "loss": 0.0014, "step": 152630 }, { "epoch": 2.4975865172216314, "grad_norm": 0.005924841854721308, "learning_rate": 3.716079613957725e-06, "loss": 0.0007, "step": 152640 }, { "epoch": 2.4977501431727074, "grad_norm": 0.04079669341444969, "learning_rate": 3.715389613850838e-06, "loss": 0.0007, "step": 152650 }, { "epoch": 2.497913769123783, "grad_norm": 0.025701306760311127, "learning_rate": 3.7146996399360615e-06, "loss": 0.0007, "step": 152660 }, { "epoch": 2.498077395074859, "grad_norm": 0.06209730729460716, "learning_rate": 3.714009692227468e-06, "loss": 0.0007, "step": 152670 }, { "epoch": 2.498241021025935, "grad_norm": 0.03809286281466484, "learning_rate": 3.713319770739122e-06, "loss": 0.0008, "step": 152680 }, { "epoch": 2.4984046469770105, "grad_norm": 0.0027723454404622316, "learning_rate": 3.712629875485093e-06, "loss": 0.0005, "step": 152690 }, { "epoch": 2.4985682729280865, "grad_norm": 0.1374993920326233, "learning_rate": 3.711940006479444e-06, "loss": 0.0006, "step": 152700 }, { "epoch": 2.4987318988791625, "grad_norm": 0.0702613815665245, "learning_rate": 3.7112501637362454e-06, "loss": 0.001, "step": 152710 }, { "epoch": 2.498895524830238, "grad_norm": 0.004025573376566172, "learning_rate": 3.710560347269557e-06, "loss": 0.0006, "step": 152720 }, { "epoch": 2.499059150781314, "grad_norm": 0.11792077124118805, "learning_rate": 3.7098705570934494e-06, "loss": 0.002, "step": 152730 }, { "epoch": 2.4992227767323896, "grad_norm": 0.16491428017616272, "learning_rate": 3.7091807932219816e-06, "loss": 0.0019, "step": 152740 }, { "epoch": 2.4993864026834656, "grad_norm": 0.0031640955712646246, "learning_rate": 3.7084910556692217e-06, "loss": 0.0007, "step": 152750 }, { "epoch": 2.4995500286345416, "grad_norm": 0.014572865329682827, "learning_rate": 3.707801344449228e-06, "loss": 0.0007, "step": 152760 }, { "epoch": 2.499713654585617, "grad_norm": 0.003435952588915825, "learning_rate": 3.7071116595760693e-06, "loss": 0.0006, "step": 152770 }, { "epoch": 2.499877280536693, "grad_norm": 0.02677864208817482, "learning_rate": 3.706422001063801e-06, "loss": 0.0007, "step": 152780 }, { "epoch": 2.500040906487769, "grad_norm": 0.1733889877796173, "learning_rate": 3.7057323689264906e-06, "loss": 0.0018, "step": 152790 }, { "epoch": 2.5002045324388447, "grad_norm": 0.04876245930790901, "learning_rate": 3.7050427631781936e-06, "loss": 0.0013, "step": 152800 }, { "epoch": 2.5003681583899207, "grad_norm": 0.1295158863067627, "learning_rate": 3.704353183832977e-06, "loss": 0.0015, "step": 152810 }, { "epoch": 2.500531784340996, "grad_norm": 0.07525376975536346, "learning_rate": 3.7036636309048944e-06, "loss": 0.0008, "step": 152820 }, { "epoch": 2.500695410292072, "grad_norm": 0.03187021613121033, "learning_rate": 3.702974104408009e-06, "loss": 0.0009, "step": 152830 }, { "epoch": 2.500859036243148, "grad_norm": 0.06165613979101181, "learning_rate": 3.7022846043563774e-06, "loss": 0.0006, "step": 152840 }, { "epoch": 2.5010226621942238, "grad_norm": 0.05683921277523041, "learning_rate": 3.7015951307640607e-06, "loss": 0.0033, "step": 152850 }, { "epoch": 2.5011862881452998, "grad_norm": 0.029810283333063126, "learning_rate": 3.7009056836451142e-06, "loss": 0.002, "step": 152860 }, { "epoch": 2.5013499140963757, "grad_norm": 0.11722826212644577, "learning_rate": 3.700216263013596e-06, "loss": 0.0009, "step": 152870 }, { "epoch": 2.5015135400474513, "grad_norm": 0.00041372355190105736, "learning_rate": 3.6995268688835625e-06, "loss": 0.0011, "step": 152880 }, { "epoch": 2.5016771659985273, "grad_norm": 0.03330940380692482, "learning_rate": 3.6988375012690702e-06, "loss": 0.0005, "step": 152890 }, { "epoch": 2.5018407919496033, "grad_norm": 0.027297135442495346, "learning_rate": 3.6981481601841746e-06, "loss": 0.0006, "step": 152900 }, { "epoch": 2.502004417900679, "grad_norm": 0.027991635724902153, "learning_rate": 3.6974588456429317e-06, "loss": 0.0013, "step": 152910 }, { "epoch": 2.502168043851755, "grad_norm": 0.037469878792762756, "learning_rate": 3.696769557659394e-06, "loss": 0.0012, "step": 152920 }, { "epoch": 2.502331669802831, "grad_norm": 0.03941751644015312, "learning_rate": 3.696080296247617e-06, "loss": 0.0004, "step": 152930 }, { "epoch": 2.5024952957539064, "grad_norm": 0.024256402626633644, "learning_rate": 3.6953910614216537e-06, "loss": 0.0009, "step": 152940 }, { "epoch": 2.5026589217049824, "grad_norm": 0.02468583919107914, "learning_rate": 3.694701853195558e-06, "loss": 0.0007, "step": 152950 }, { "epoch": 2.5028225476560584, "grad_norm": 0.002030387055128813, "learning_rate": 3.69401267158338e-06, "loss": 0.0006, "step": 152960 }, { "epoch": 2.502986173607134, "grad_norm": 0.06683187931776047, "learning_rate": 3.693323516599174e-06, "loss": 0.0008, "step": 152970 }, { "epoch": 2.50314979955821, "grad_norm": 0.04874755069613457, "learning_rate": 3.692634388256989e-06, "loss": 0.0005, "step": 152980 }, { "epoch": 2.503313425509286, "grad_norm": 0.14582502841949463, "learning_rate": 3.6919452865708794e-06, "loss": 0.0015, "step": 152990 }, { "epoch": 2.5034770514603615, "grad_norm": 0.0041221315041184425, "learning_rate": 3.6912562115548906e-06, "loss": 0.0005, "step": 153000 }, { "epoch": 2.5036406774114375, "grad_norm": 0.04755428060889244, "learning_rate": 3.6905671632230753e-06, "loss": 0.0014, "step": 153010 }, { "epoch": 2.5038043033625135, "grad_norm": 0.02660432457923889, "learning_rate": 3.689878141589482e-06, "loss": 0.0006, "step": 153020 }, { "epoch": 2.503967929313589, "grad_norm": 0.07171571999788284, "learning_rate": 3.6891891466681595e-06, "loss": 0.0013, "step": 153030 }, { "epoch": 2.504131555264665, "grad_norm": 0.01679406687617302, "learning_rate": 3.6885001784731547e-06, "loss": 0.0005, "step": 153040 }, { "epoch": 2.504295181215741, "grad_norm": 0.040772728621959686, "learning_rate": 3.687811237018517e-06, "loss": 0.0006, "step": 153050 }, { "epoch": 2.5044588071668166, "grad_norm": 0.024170339107513428, "learning_rate": 3.6871223223182905e-06, "loss": 0.0011, "step": 153060 }, { "epoch": 2.5046224331178926, "grad_norm": 0.03489139303565025, "learning_rate": 3.6864334343865248e-06, "loss": 0.0005, "step": 153070 }, { "epoch": 2.5047860590689686, "grad_norm": 0.002443038858473301, "learning_rate": 3.685744573237263e-06, "loss": 0.0009, "step": 153080 }, { "epoch": 2.504949685020044, "grad_norm": 0.005975138396024704, "learning_rate": 3.685055738884553e-06, "loss": 0.0004, "step": 153090 }, { "epoch": 2.50511331097112, "grad_norm": 0.06089441478252411, "learning_rate": 3.6843669313424375e-06, "loss": 0.0007, "step": 153100 }, { "epoch": 2.505276936922196, "grad_norm": 0.03967242315411568, "learning_rate": 3.683678150624962e-06, "loss": 0.001, "step": 153110 }, { "epoch": 2.5054405628732717, "grad_norm": 0.006315918639302254, "learning_rate": 3.6829893967461685e-06, "loss": 0.0005, "step": 153120 }, { "epoch": 2.5056041888243477, "grad_norm": 0.0737459734082222, "learning_rate": 3.6823006697201018e-06, "loss": 0.001, "step": 153130 }, { "epoch": 2.5057678147754237, "grad_norm": 0.022061429917812347, "learning_rate": 3.6816119695608033e-06, "loss": 0.0006, "step": 153140 }, { "epoch": 2.505931440726499, "grad_norm": 0.046378400176763535, "learning_rate": 3.680923296282316e-06, "loss": 0.0009, "step": 153150 }, { "epoch": 2.506095066677575, "grad_norm": 0.03241518884897232, "learning_rate": 3.6802346498986797e-06, "loss": 0.0005, "step": 153160 }, { "epoch": 2.5062586926286508, "grad_norm": 0.04217643663287163, "learning_rate": 3.6795460304239383e-06, "loss": 0.001, "step": 153170 }, { "epoch": 2.5064223185797267, "grad_norm": 0.05719486624002457, "learning_rate": 3.6788574378721285e-06, "loss": 0.0016, "step": 153180 }, { "epoch": 2.5065859445308027, "grad_norm": 0.06537380069494247, "learning_rate": 3.6781688722572934e-06, "loss": 0.0008, "step": 153190 }, { "epoch": 2.5067495704818783, "grad_norm": 0.06623931229114532, "learning_rate": 3.6774803335934695e-06, "loss": 0.0008, "step": 153200 }, { "epoch": 2.5069131964329543, "grad_norm": 0.0935278832912445, "learning_rate": 3.6767918218946984e-06, "loss": 0.0012, "step": 153210 }, { "epoch": 2.50707682238403, "grad_norm": 0.08504905551671982, "learning_rate": 3.6761033371750156e-06, "loss": 0.0007, "step": 153220 }, { "epoch": 2.507240448335106, "grad_norm": 0.06585705280303955, "learning_rate": 3.67541487944846e-06, "loss": 0.001, "step": 153230 }, { "epoch": 2.507404074286182, "grad_norm": 0.011053196154534817, "learning_rate": 3.674726448729068e-06, "loss": 0.0006, "step": 153240 }, { "epoch": 2.5075677002372574, "grad_norm": 0.022531867027282715, "learning_rate": 3.674038045030878e-06, "loss": 0.0006, "step": 153250 }, { "epoch": 2.5077313261883334, "grad_norm": 0.05254030227661133, "learning_rate": 3.673349668367923e-06, "loss": 0.0013, "step": 153260 }, { "epoch": 2.5078949521394094, "grad_norm": 0.018674707040190697, "learning_rate": 3.6726613187542414e-06, "loss": 0.0007, "step": 153270 }, { "epoch": 2.508058578090485, "grad_norm": 0.009689213708043098, "learning_rate": 3.671972996203865e-06, "loss": 0.0005, "step": 153280 }, { "epoch": 2.508222204041561, "grad_norm": 0.014288049191236496, "learning_rate": 3.6712847007308326e-06, "loss": 0.0005, "step": 153290 }, { "epoch": 2.508385829992637, "grad_norm": 0.10923336446285248, "learning_rate": 3.670596432349172e-06, "loss": 0.001, "step": 153300 }, { "epoch": 2.5085494559437125, "grad_norm": 0.06620912998914719, "learning_rate": 3.6699081910729223e-06, "loss": 0.0007, "step": 153310 }, { "epoch": 2.5087130818947885, "grad_norm": 0.01772378198802471, "learning_rate": 3.66921997691611e-06, "loss": 0.0005, "step": 153320 }, { "epoch": 2.5088767078458645, "grad_norm": 0.025444380939006805, "learning_rate": 3.668531789892774e-06, "loss": 0.0009, "step": 153330 }, { "epoch": 2.50904033379694, "grad_norm": 0.029350360855460167, "learning_rate": 3.6678436300169397e-06, "loss": 0.0005, "step": 153340 }, { "epoch": 2.509203959748016, "grad_norm": 0.06696499139070511, "learning_rate": 3.6671554973026436e-06, "loss": 0.0007, "step": 153350 }, { "epoch": 2.509367585699092, "grad_norm": 0.004037285223603249, "learning_rate": 3.6664673917639106e-06, "loss": 0.0012, "step": 153360 }, { "epoch": 2.5095312116501676, "grad_norm": 0.04173517972230911, "learning_rate": 3.6657793134147755e-06, "loss": 0.0009, "step": 153370 }, { "epoch": 2.5096948376012436, "grad_norm": 0.02229713276028633, "learning_rate": 3.665091262269263e-06, "loss": 0.0005, "step": 153380 }, { "epoch": 2.5098584635523196, "grad_norm": 0.046832770109176636, "learning_rate": 3.6644032383414073e-06, "loss": 0.0007, "step": 153390 }, { "epoch": 2.510022089503395, "grad_norm": 0.05508207529783249, "learning_rate": 3.6637152416452306e-06, "loss": 0.0007, "step": 153400 }, { "epoch": 2.510185715454471, "grad_norm": 0.2232620120048523, "learning_rate": 3.663027272194767e-06, "loss": 0.0012, "step": 153410 }, { "epoch": 2.510349341405547, "grad_norm": 0.0010759718716144562, "learning_rate": 3.662339330004037e-06, "loss": 0.0005, "step": 153420 }, { "epoch": 2.5105129673566227, "grad_norm": 0.063326396048069, "learning_rate": 3.6616514150870742e-06, "loss": 0.0041, "step": 153430 }, { "epoch": 2.5106765933076987, "grad_norm": 0.005067439749836922, "learning_rate": 3.6609635274578964e-06, "loss": 0.0005, "step": 153440 }, { "epoch": 2.5108402192587747, "grad_norm": 0.030505962669849396, "learning_rate": 3.6602756671305374e-06, "loss": 0.0009, "step": 153450 }, { "epoch": 2.51100384520985, "grad_norm": 0.020767265930771828, "learning_rate": 3.659587834119016e-06, "loss": 0.0012, "step": 153460 }, { "epoch": 2.511167471160926, "grad_norm": 0.06281381100416183, "learning_rate": 3.65890002843736e-06, "loss": 0.0009, "step": 153470 }, { "epoch": 2.511331097112002, "grad_norm": 0.010246332734823227, "learning_rate": 3.6582122500995896e-06, "loss": 0.001, "step": 153480 }, { "epoch": 2.5114947230630777, "grad_norm": 0.041723787784576416, "learning_rate": 3.657524499119732e-06, "loss": 0.0011, "step": 153490 }, { "epoch": 2.5116583490141537, "grad_norm": 0.02793964371085167, "learning_rate": 3.6568367755118078e-06, "loss": 0.0007, "step": 153500 }, { "epoch": 2.5118219749652297, "grad_norm": 0.06214742362499237, "learning_rate": 3.65614907928984e-06, "loss": 0.0011, "step": 153510 }, { "epoch": 2.5119856009163053, "grad_norm": 0.021248968318104744, "learning_rate": 3.655461410467849e-06, "loss": 0.0008, "step": 153520 }, { "epoch": 2.5121492268673813, "grad_norm": 0.022962287068367004, "learning_rate": 3.654773769059856e-06, "loss": 0.0005, "step": 153530 }, { "epoch": 2.512312852818457, "grad_norm": 0.048084136098623276, "learning_rate": 3.6540861550798817e-06, "loss": 0.0013, "step": 153540 }, { "epoch": 2.512476478769533, "grad_norm": 0.021298319101333618, "learning_rate": 3.6533985685419473e-06, "loss": 0.0016, "step": 153550 }, { "epoch": 2.512640104720609, "grad_norm": 0.019936881959438324, "learning_rate": 3.652711009460069e-06, "loss": 0.0006, "step": 153560 }, { "epoch": 2.5128037306716844, "grad_norm": 0.02812562882900238, "learning_rate": 3.6520234778482686e-06, "loss": 0.0009, "step": 153570 }, { "epoch": 2.5129673566227604, "grad_norm": 0.08158423006534576, "learning_rate": 3.651335973720563e-06, "loss": 0.0009, "step": 153580 }, { "epoch": 2.513130982573836, "grad_norm": 0.0111308628693223, "learning_rate": 3.6506484970909704e-06, "loss": 0.0007, "step": 153590 }, { "epoch": 2.513294608524912, "grad_norm": 0.02834322303533554, "learning_rate": 3.649961047973506e-06, "loss": 0.0008, "step": 153600 }, { "epoch": 2.513458234475988, "grad_norm": 0.015463515184819698, "learning_rate": 3.649273626382189e-06, "loss": 0.0002, "step": 153610 }, { "epoch": 2.5136218604270635, "grad_norm": 0.017520297318696976, "learning_rate": 3.648586232331033e-06, "loss": 0.0009, "step": 153620 }, { "epoch": 2.5137854863781395, "grad_norm": 0.052320390939712524, "learning_rate": 3.6478988658340554e-06, "loss": 0.0008, "step": 153630 }, { "epoch": 2.5139491123292155, "grad_norm": 0.014621900394558907, "learning_rate": 3.6472115269052687e-06, "loss": 0.0009, "step": 153640 }, { "epoch": 2.514112738280291, "grad_norm": 0.08902517706155777, "learning_rate": 3.6465242155586894e-06, "loss": 0.0007, "step": 153650 }, { "epoch": 2.514276364231367, "grad_norm": 0.04763710871338844, "learning_rate": 3.6458369318083296e-06, "loss": 0.0006, "step": 153660 }, { "epoch": 2.514439990182443, "grad_norm": 0.11680256575345993, "learning_rate": 3.6451496756682035e-06, "loss": 0.0013, "step": 153670 }, { "epoch": 2.5146036161335186, "grad_norm": 0.02764865569770336, "learning_rate": 3.644462447152323e-06, "loss": 0.0004, "step": 153680 }, { "epoch": 2.5147672420845946, "grad_norm": 0.0038783284835517406, "learning_rate": 3.643775246274701e-06, "loss": 0.0009, "step": 153690 }, { "epoch": 2.5149308680356706, "grad_norm": 0.044086750596761703, "learning_rate": 3.643088073049348e-06, "loss": 0.0006, "step": 153700 }, { "epoch": 2.515094493986746, "grad_norm": 0.05960675701498985, "learning_rate": 3.642400927490275e-06, "loss": 0.0009, "step": 153710 }, { "epoch": 2.515258119937822, "grad_norm": 0.17543673515319824, "learning_rate": 3.6417138096114924e-06, "loss": 0.001, "step": 153720 }, { "epoch": 2.515421745888898, "grad_norm": 0.15913602709770203, "learning_rate": 3.6410267194270105e-06, "loss": 0.0015, "step": 153730 }, { "epoch": 2.5155853718399737, "grad_norm": 0.003446440678089857, "learning_rate": 3.6403396569508377e-06, "loss": 0.0006, "step": 153740 }, { "epoch": 2.5157489977910497, "grad_norm": 0.03877509385347366, "learning_rate": 3.639652622196983e-06, "loss": 0.0009, "step": 153750 }, { "epoch": 2.5159126237421257, "grad_norm": 0.07082076370716095, "learning_rate": 3.6389656151794538e-06, "loss": 0.0008, "step": 153760 }, { "epoch": 2.516076249693201, "grad_norm": 0.0032618495170027018, "learning_rate": 3.6382786359122592e-06, "loss": 0.0007, "step": 153770 }, { "epoch": 2.516239875644277, "grad_norm": 0.08622778207063675, "learning_rate": 3.637591684409404e-06, "loss": 0.0012, "step": 153780 }, { "epoch": 2.516403501595353, "grad_norm": 0.11502595245838165, "learning_rate": 3.6369047606848975e-06, "loss": 0.001, "step": 153790 }, { "epoch": 2.5165671275464288, "grad_norm": 0.057153504341840744, "learning_rate": 3.6362178647527413e-06, "loss": 0.0012, "step": 153800 }, { "epoch": 2.5167307534975047, "grad_norm": 0.04084737226366997, "learning_rate": 3.635530996626945e-06, "loss": 0.0025, "step": 153810 }, { "epoch": 2.5168943794485807, "grad_norm": 0.021045219153165817, "learning_rate": 3.63484415632151e-06, "loss": 0.0008, "step": 153820 }, { "epoch": 2.5170580053996563, "grad_norm": 0.03607975319027901, "learning_rate": 3.6341573438504425e-06, "loss": 0.0007, "step": 153830 }, { "epoch": 2.5172216313507323, "grad_norm": 0.05368160083889961, "learning_rate": 3.6334705592277443e-06, "loss": 0.0007, "step": 153840 }, { "epoch": 2.5173852573018083, "grad_norm": 0.00426101591438055, "learning_rate": 3.63278380246742e-06, "loss": 0.0003, "step": 153850 }, { "epoch": 2.517548883252884, "grad_norm": 0.047745682299137115, "learning_rate": 3.6320970735834704e-06, "loss": 0.0006, "step": 153860 }, { "epoch": 2.51771250920396, "grad_norm": 0.04639856889843941, "learning_rate": 3.6314103725898987e-06, "loss": 0.0004, "step": 153870 }, { "epoch": 2.517876135155036, "grad_norm": 0.11401758342981339, "learning_rate": 3.6307236995007052e-06, "loss": 0.001, "step": 153880 }, { "epoch": 2.5180397611061114, "grad_norm": 0.014663182199001312, "learning_rate": 3.6300370543298913e-06, "loss": 0.0008, "step": 153890 }, { "epoch": 2.5182033870571874, "grad_norm": 0.047717828303575516, "learning_rate": 3.6293504370914563e-06, "loss": 0.0011, "step": 153900 }, { "epoch": 2.5183670130082634, "grad_norm": 0.036890532821416855, "learning_rate": 3.628663847799401e-06, "loss": 0.0011, "step": 153910 }, { "epoch": 2.518530638959339, "grad_norm": 0.010070405900478363, "learning_rate": 3.627977286467722e-06, "loss": 0.0005, "step": 153920 }, { "epoch": 2.518694264910415, "grad_norm": 0.07688090205192566, "learning_rate": 3.627290753110422e-06, "loss": 0.0017, "step": 153930 }, { "epoch": 2.5188578908614905, "grad_norm": 0.05399307236075401, "learning_rate": 3.626604247741493e-06, "loss": 0.0008, "step": 153940 }, { "epoch": 2.5190215168125665, "grad_norm": 0.023858005180954933, "learning_rate": 3.625917770374938e-06, "loss": 0.0019, "step": 153950 }, { "epoch": 2.5191851427636425, "grad_norm": 0.03837679699063301, "learning_rate": 3.6252313210247482e-06, "loss": 0.0007, "step": 153960 }, { "epoch": 2.519348768714718, "grad_norm": 0.06653273850679398, "learning_rate": 3.624544899704926e-06, "loss": 0.0007, "step": 153970 }, { "epoch": 2.519512394665794, "grad_norm": 0.04122459515929222, "learning_rate": 3.6238585064294603e-06, "loss": 0.0004, "step": 153980 }, { "epoch": 2.5196760206168696, "grad_norm": 0.011678174138069153, "learning_rate": 3.6231721412123524e-06, "loss": 0.0008, "step": 153990 }, { "epoch": 2.5198396465679456, "grad_norm": 0.025104280561208725, "learning_rate": 3.6224858040675896e-06, "loss": 0.0009, "step": 154000 }, { "epoch": 2.5200032725190216, "grad_norm": 0.01598966121673584, "learning_rate": 3.621799495009174e-06, "loss": 0.0006, "step": 154010 }, { "epoch": 2.520166898470097, "grad_norm": 0.001307751052081585, "learning_rate": 3.6211132140510907e-06, "loss": 0.0005, "step": 154020 }, { "epoch": 2.520330524421173, "grad_norm": 0.028902344405651093, "learning_rate": 3.6204269612073397e-06, "loss": 0.0004, "step": 154030 }, { "epoch": 2.520494150372249, "grad_norm": 0.060630980879068375, "learning_rate": 3.619740736491907e-06, "loss": 0.0006, "step": 154040 }, { "epoch": 2.5206577763233247, "grad_norm": 0.004138193558901548, "learning_rate": 3.6190545399187895e-06, "loss": 0.0006, "step": 154050 }, { "epoch": 2.5208214022744007, "grad_norm": 0.06031571701169014, "learning_rate": 3.6183683715019717e-06, "loss": 0.0008, "step": 154060 }, { "epoch": 2.5209850282254767, "grad_norm": 0.07688019424676895, "learning_rate": 3.6176822312554517e-06, "loss": 0.0015, "step": 154070 }, { "epoch": 2.521148654176552, "grad_norm": 0.04708314314484596, "learning_rate": 3.616996119193212e-06, "loss": 0.0013, "step": 154080 }, { "epoch": 2.521312280127628, "grad_norm": 0.006344661582261324, "learning_rate": 3.616310035329248e-06, "loss": 0.0004, "step": 154090 }, { "epoch": 2.521475906078704, "grad_norm": 0.07343952357769012, "learning_rate": 3.615623979677543e-06, "loss": 0.0012, "step": 154100 }, { "epoch": 2.5216395320297798, "grad_norm": 0.03374044597148895, "learning_rate": 3.614937952252091e-06, "loss": 0.0016, "step": 154110 }, { "epoch": 2.5218031579808557, "grad_norm": 0.026001380756497383, "learning_rate": 3.614251953066874e-06, "loss": 0.0007, "step": 154120 }, { "epoch": 2.5219667839319317, "grad_norm": 0.030898455530405045, "learning_rate": 3.6135659821358815e-06, "loss": 0.0008, "step": 154130 }, { "epoch": 2.5221304098830073, "grad_norm": 0.12907451391220093, "learning_rate": 3.612880039473099e-06, "loss": 0.0011, "step": 154140 }, { "epoch": 2.5222940358340833, "grad_norm": 0.05313396453857422, "learning_rate": 3.6121941250925143e-06, "loss": 0.0011, "step": 154150 }, { "epoch": 2.5224576617851593, "grad_norm": 0.009555426426231861, "learning_rate": 3.611508239008109e-06, "loss": 0.0006, "step": 154160 }, { "epoch": 2.522621287736235, "grad_norm": 0.14752013981342316, "learning_rate": 3.6108223812338717e-06, "loss": 0.0013, "step": 154170 }, { "epoch": 2.522784913687311, "grad_norm": 0.051455896347761154, "learning_rate": 3.6101365517837826e-06, "loss": 0.0006, "step": 154180 }, { "epoch": 2.522948539638387, "grad_norm": 0.02784709259867668, "learning_rate": 3.609450750671829e-06, "loss": 0.0006, "step": 154190 }, { "epoch": 2.5231121655894624, "grad_norm": 0.05263075605034828, "learning_rate": 3.6087649779119915e-06, "loss": 0.0007, "step": 154200 }, { "epoch": 2.5232757915405384, "grad_norm": 0.03422782942652702, "learning_rate": 3.608079233518253e-06, "loss": 0.0009, "step": 154210 }, { "epoch": 2.5234394174916144, "grad_norm": 0.014267257414758205, "learning_rate": 3.6073935175045945e-06, "loss": 0.0006, "step": 154220 }, { "epoch": 2.52360304344269, "grad_norm": 0.03848368674516678, "learning_rate": 3.6067078298849987e-06, "loss": 0.0008, "step": 154230 }, { "epoch": 2.523766669393766, "grad_norm": 0.1321691870689392, "learning_rate": 3.6060221706734446e-06, "loss": 0.0009, "step": 154240 }, { "epoch": 2.523930295344842, "grad_norm": 0.031449515372514725, "learning_rate": 3.6053365398839137e-06, "loss": 0.0007, "step": 154250 }, { "epoch": 2.5240939212959175, "grad_norm": 0.05372105538845062, "learning_rate": 3.604650937530383e-06, "loss": 0.0016, "step": 154260 }, { "epoch": 2.5242575472469935, "grad_norm": 0.03495734557509422, "learning_rate": 3.6039653636268344e-06, "loss": 0.0006, "step": 154270 }, { "epoch": 2.5244211731980695, "grad_norm": 0.030470428988337517, "learning_rate": 3.603279818187244e-06, "loss": 0.0007, "step": 154280 }, { "epoch": 2.524584799149145, "grad_norm": 0.024335479363799095, "learning_rate": 3.602594301225591e-06, "loss": 0.0004, "step": 154290 }, { "epoch": 2.524748425100221, "grad_norm": 0.027817191556096077, "learning_rate": 3.601908812755851e-06, "loss": 0.001, "step": 154300 }, { "epoch": 2.524912051051297, "grad_norm": 0.1274981051683426, "learning_rate": 3.601223352792002e-06, "loss": 0.0007, "step": 154310 }, { "epoch": 2.5250756770023726, "grad_norm": 0.02957083098590374, "learning_rate": 3.6005379213480186e-06, "loss": 0.0008, "step": 154320 }, { "epoch": 2.5252393029534486, "grad_norm": 0.017204826697707176, "learning_rate": 3.5998525184378775e-06, "loss": 0.0008, "step": 154330 }, { "epoch": 2.525402928904524, "grad_norm": 0.018983494490385056, "learning_rate": 3.5991671440755516e-06, "loss": 0.0007, "step": 154340 }, { "epoch": 2.5255665548556, "grad_norm": 0.020192129537463188, "learning_rate": 3.5984817982750174e-06, "loss": 0.0004, "step": 154350 }, { "epoch": 2.5257301808066757, "grad_norm": 0.02493825927376747, "learning_rate": 3.5977964810502463e-06, "loss": 0.0008, "step": 154360 }, { "epoch": 2.5258938067577517, "grad_norm": 0.08640869706869125, "learning_rate": 3.5971111924152136e-06, "loss": 0.0014, "step": 154370 }, { "epoch": 2.5260574327088277, "grad_norm": 0.031105095520615578, "learning_rate": 3.5964259323838898e-06, "loss": 0.0007, "step": 154380 }, { "epoch": 2.526221058659903, "grad_norm": 0.020259736105799675, "learning_rate": 3.5957407009702483e-06, "loss": 0.0009, "step": 154390 }, { "epoch": 2.526384684610979, "grad_norm": 0.01778121292591095, "learning_rate": 3.5950554981882584e-06, "loss": 0.0014, "step": 154400 }, { "epoch": 2.526548310562055, "grad_norm": 0.04600261151790619, "learning_rate": 3.5943703240518937e-06, "loss": 0.0007, "step": 154410 }, { "epoch": 2.5267119365131308, "grad_norm": 0.10465473681688309, "learning_rate": 3.5936851785751204e-06, "loss": 0.0016, "step": 154420 }, { "epoch": 2.5268755624642067, "grad_norm": 0.011983387172222137, "learning_rate": 3.5930000617719124e-06, "loss": 0.0012, "step": 154430 }, { "epoch": 2.5270391884152827, "grad_norm": 0.0765000581741333, "learning_rate": 3.5923149736562353e-06, "loss": 0.001, "step": 154440 }, { "epoch": 2.5272028143663583, "grad_norm": 0.044438913464546204, "learning_rate": 3.5916299142420595e-06, "loss": 0.0006, "step": 154450 }, { "epoch": 2.5273664403174343, "grad_norm": 0.02062227949500084, "learning_rate": 3.5909448835433514e-06, "loss": 0.001, "step": 154460 }, { "epoch": 2.5275300662685103, "grad_norm": 0.0369647778570652, "learning_rate": 3.590259881574079e-06, "loss": 0.0008, "step": 154470 }, { "epoch": 2.527693692219586, "grad_norm": 0.002485171891748905, "learning_rate": 3.589574908348208e-06, "loss": 0.0015, "step": 154480 }, { "epoch": 2.527857318170662, "grad_norm": 0.07091446965932846, "learning_rate": 3.5888899638797066e-06, "loss": 0.0006, "step": 154490 }, { "epoch": 2.528020944121738, "grad_norm": 0.039859116077423096, "learning_rate": 3.588205048182537e-06, "loss": 0.0004, "step": 154500 }, { "epoch": 2.5281845700728134, "grad_norm": 0.006342704873532057, "learning_rate": 3.5875201612706674e-06, "loss": 0.0007, "step": 154510 }, { "epoch": 2.5283481960238894, "grad_norm": 0.010538669303059578, "learning_rate": 3.586835303158059e-06, "loss": 0.001, "step": 154520 }, { "epoch": 2.5285118219749654, "grad_norm": 0.02633039839565754, "learning_rate": 3.5861504738586783e-06, "loss": 0.001, "step": 154530 }, { "epoch": 2.528675447926041, "grad_norm": 0.0454259030520916, "learning_rate": 3.5854656733864856e-06, "loss": 0.0005, "step": 154540 }, { "epoch": 2.528839073877117, "grad_norm": 0.010062133893370628, "learning_rate": 3.584780901755446e-06, "loss": 0.0003, "step": 154550 }, { "epoch": 2.529002699828193, "grad_norm": 0.04125071316957474, "learning_rate": 3.5840961589795186e-06, "loss": 0.0007, "step": 154560 }, { "epoch": 2.5291663257792685, "grad_norm": 0.04618241637945175, "learning_rate": 3.5834114450726693e-06, "loss": 0.0009, "step": 154570 }, { "epoch": 2.5293299517303445, "grad_norm": 0.05832458287477493, "learning_rate": 3.5827267600488523e-06, "loss": 0.0007, "step": 154580 }, { "epoch": 2.5294935776814205, "grad_norm": 0.03552218899130821, "learning_rate": 3.5820421039220345e-06, "loss": 0.0006, "step": 154590 }, { "epoch": 2.529657203632496, "grad_norm": 0.008298840373754501, "learning_rate": 3.5813574767061693e-06, "loss": 0.0006, "step": 154600 }, { "epoch": 2.529820829583572, "grad_norm": 0.046780917793512344, "learning_rate": 3.580672878415222e-06, "loss": 0.0007, "step": 154610 }, { "epoch": 2.529984455534648, "grad_norm": 0.026745343580842018, "learning_rate": 3.5799883090631436e-06, "loss": 0.0008, "step": 154620 }, { "epoch": 2.5301480814857236, "grad_norm": 0.06899306923151016, "learning_rate": 3.5793037686638996e-06, "loss": 0.0021, "step": 154630 }, { "epoch": 2.5303117074367996, "grad_norm": 0.02569335512816906, "learning_rate": 3.5786192572314405e-06, "loss": 0.0012, "step": 154640 }, { "epoch": 2.5304753333878756, "grad_norm": 0.02575504779815674, "learning_rate": 3.5779347747797288e-06, "loss": 0.0013, "step": 154650 }, { "epoch": 2.530638959338951, "grad_norm": 0.03445049002766609, "learning_rate": 3.577250321322715e-06, "loss": 0.0012, "step": 154660 }, { "epoch": 2.530802585290027, "grad_norm": 0.04257968068122864, "learning_rate": 3.57656589687436e-06, "loss": 0.001, "step": 154670 }, { "epoch": 2.530966211241103, "grad_norm": 0.050708819180727005, "learning_rate": 3.575881501448612e-06, "loss": 0.002, "step": 154680 }, { "epoch": 2.5311298371921787, "grad_norm": 0.019105175510048866, "learning_rate": 3.575197135059433e-06, "loss": 0.001, "step": 154690 }, { "epoch": 2.5312934631432547, "grad_norm": 0.05340688303112984, "learning_rate": 3.5745127977207687e-06, "loss": 0.0006, "step": 154700 }, { "epoch": 2.53145708909433, "grad_norm": 0.0023759352043271065, "learning_rate": 3.5738284894465802e-06, "loss": 0.0005, "step": 154710 }, { "epoch": 2.531620715045406, "grad_norm": 0.10096611827611923, "learning_rate": 3.5731442102508118e-06, "loss": 0.0005, "step": 154720 }, { "epoch": 2.531784340996482, "grad_norm": 0.03829770162701607, "learning_rate": 3.572459960147423e-06, "loss": 0.0005, "step": 154730 }, { "epoch": 2.5319479669475577, "grad_norm": 0.13127030432224274, "learning_rate": 3.571775739150359e-06, "loss": 0.0019, "step": 154740 }, { "epoch": 2.5321115928986337, "grad_norm": 0.08038483560085297, "learning_rate": 3.571091547273575e-06, "loss": 0.0004, "step": 154750 }, { "epoch": 2.5322752188497093, "grad_norm": 0.012253639288246632, "learning_rate": 3.5704073845310183e-06, "loss": 0.0005, "step": 154760 }, { "epoch": 2.5324388448007853, "grad_norm": 0.004179561510682106, "learning_rate": 3.569723250936639e-06, "loss": 0.0007, "step": 154770 }, { "epoch": 2.5326024707518613, "grad_norm": 0.12234432995319366, "learning_rate": 3.569039146504386e-06, "loss": 0.0038, "step": 154780 }, { "epoch": 2.532766096702937, "grad_norm": 0.1514243185520172, "learning_rate": 3.5683550712482084e-06, "loss": 0.0012, "step": 154790 }, { "epoch": 2.532929722654013, "grad_norm": 0.04451322183012962, "learning_rate": 3.5676710251820524e-06, "loss": 0.0005, "step": 154800 }, { "epoch": 2.533093348605089, "grad_norm": 0.006964402738958597, "learning_rate": 3.566987008319868e-06, "loss": 0.0005, "step": 154810 }, { "epoch": 2.5332569745561644, "grad_norm": 0.0029705953784286976, "learning_rate": 3.5663030206755973e-06, "loss": 0.0007, "step": 154820 }, { "epoch": 2.5334206005072404, "grad_norm": 0.038306768983602524, "learning_rate": 3.56561906226319e-06, "loss": 0.0006, "step": 154830 }, { "epoch": 2.5335842264583164, "grad_norm": 0.024291636422276497, "learning_rate": 3.56493513309659e-06, "loss": 0.0007, "step": 154840 }, { "epoch": 2.533747852409392, "grad_norm": 0.01260542031377554, "learning_rate": 3.564251233189743e-06, "loss": 0.0005, "step": 154850 }, { "epoch": 2.533911478360468, "grad_norm": 0.06615322828292847, "learning_rate": 3.5635673625565905e-06, "loss": 0.0008, "step": 154860 }, { "epoch": 2.534075104311544, "grad_norm": 0.0009830683702602983, "learning_rate": 3.562883521211079e-06, "loss": 0.0003, "step": 154870 }, { "epoch": 2.5342387302626195, "grad_norm": 0.042849574238061905, "learning_rate": 3.5621997091671493e-06, "loss": 0.0011, "step": 154880 }, { "epoch": 2.5344023562136955, "grad_norm": 0.024150142446160316, "learning_rate": 3.561515926438746e-06, "loss": 0.0004, "step": 154890 }, { "epoch": 2.5345659821647715, "grad_norm": 0.055902957916259766, "learning_rate": 3.560832173039809e-06, "loss": 0.0018, "step": 154900 }, { "epoch": 2.534729608115847, "grad_norm": 0.024907629936933517, "learning_rate": 3.5601484489842802e-06, "loss": 0.0016, "step": 154910 }, { "epoch": 2.534893234066923, "grad_norm": 0.05110780894756317, "learning_rate": 3.5594647542860996e-06, "loss": 0.0011, "step": 154920 }, { "epoch": 2.535056860017999, "grad_norm": 0.026785342022776604, "learning_rate": 3.5587810889592078e-06, "loss": 0.0006, "step": 154930 }, { "epoch": 2.5352204859690746, "grad_norm": 0.028791939839720726, "learning_rate": 3.558097453017543e-06, "loss": 0.001, "step": 154940 }, { "epoch": 2.5353841119201506, "grad_norm": 0.061050958931446075, "learning_rate": 3.557413846475046e-06, "loss": 0.0004, "step": 154950 }, { "epoch": 2.5355477378712266, "grad_norm": 0.027851073071360588, "learning_rate": 3.5567302693456536e-06, "loss": 0.0012, "step": 154960 }, { "epoch": 2.535711363822302, "grad_norm": 0.003928446676582098, "learning_rate": 3.5560467216433036e-06, "loss": 0.0015, "step": 154970 }, { "epoch": 2.535874989773378, "grad_norm": 0.025589698925614357, "learning_rate": 3.5553632033819326e-06, "loss": 0.0011, "step": 154980 }, { "epoch": 2.536038615724454, "grad_norm": 0.04935379698872566, "learning_rate": 3.5546797145754775e-06, "loss": 0.0009, "step": 154990 }, { "epoch": 2.5362022416755297, "grad_norm": 0.03001677617430687, "learning_rate": 3.5539962552378734e-06, "loss": 0.0008, "step": 155000 }, { "epoch": 2.5363658676266057, "grad_norm": 0.014731623232364655, "learning_rate": 3.553312825383056e-06, "loss": 0.0003, "step": 155010 }, { "epoch": 2.5365294935776816, "grad_norm": 0.02131986804306507, "learning_rate": 3.55262942502496e-06, "loss": 0.0007, "step": 155020 }, { "epoch": 2.536693119528757, "grad_norm": 0.04620697349309921, "learning_rate": 3.5519460541775197e-06, "loss": 0.0006, "step": 155030 }, { "epoch": 2.536856745479833, "grad_norm": 0.012113229371607304, "learning_rate": 3.551262712854666e-06, "loss": 0.0005, "step": 155040 }, { "epoch": 2.537020371430909, "grad_norm": 0.04092741385102272, "learning_rate": 3.5505794010703356e-06, "loss": 0.0006, "step": 155050 }, { "epoch": 2.5371839973819847, "grad_norm": 0.03471275791525841, "learning_rate": 3.549896118838457e-06, "loss": 0.0014, "step": 155060 }, { "epoch": 2.5373476233330607, "grad_norm": 0.038876935839653015, "learning_rate": 3.5492128661729643e-06, "loss": 0.0007, "step": 155070 }, { "epoch": 2.5375112492841367, "grad_norm": 0.048598889261484146, "learning_rate": 3.5485296430877857e-06, "loss": 0.0009, "step": 155080 }, { "epoch": 2.5376748752352123, "grad_norm": 0.014780553057789803, "learning_rate": 3.547846449596856e-06, "loss": 0.0006, "step": 155090 }, { "epoch": 2.5378385011862883, "grad_norm": 0.12646906077861786, "learning_rate": 3.547163285714099e-06, "loss": 0.0009, "step": 155100 }, { "epoch": 2.538002127137364, "grad_norm": 0.09415265917778015, "learning_rate": 3.5464801514534497e-06, "loss": 0.0009, "step": 155110 }, { "epoch": 2.53816575308844, "grad_norm": 0.04923292621970177, "learning_rate": 3.545797046828832e-06, "loss": 0.0005, "step": 155120 }, { "epoch": 2.538329379039516, "grad_norm": 0.014480115845799446, "learning_rate": 3.5451139718541774e-06, "loss": 0.0005, "step": 155130 }, { "epoch": 2.5384930049905914, "grad_norm": 0.07106100767850876, "learning_rate": 3.5444309265434102e-06, "loss": 0.0006, "step": 155140 }, { "epoch": 2.5386566309416674, "grad_norm": 0.11280586570501328, "learning_rate": 3.5437479109104596e-06, "loss": 0.0006, "step": 155150 }, { "epoch": 2.538820256892743, "grad_norm": 0.005903830751776695, "learning_rate": 3.5430649249692495e-06, "loss": 0.0006, "step": 155160 }, { "epoch": 2.538983882843819, "grad_norm": 0.05487029254436493, "learning_rate": 3.542381968733708e-06, "loss": 0.0017, "step": 155170 }, { "epoch": 2.539147508794895, "grad_norm": 0.009706871584057808, "learning_rate": 3.5416990422177565e-06, "loss": 0.0005, "step": 155180 }, { "epoch": 2.5393111347459705, "grad_norm": 0.0007981673115864396, "learning_rate": 3.541016145435323e-06, "loss": 0.0005, "step": 155190 }, { "epoch": 2.5394747606970465, "grad_norm": 0.0161333866417408, "learning_rate": 3.5403332784003285e-06, "loss": 0.0008, "step": 155200 }, { "epoch": 2.5396383866481225, "grad_norm": 0.006840172689408064, "learning_rate": 3.5396504411266986e-06, "loss": 0.0008, "step": 155210 }, { "epoch": 2.539802012599198, "grad_norm": 0.042024072259664536, "learning_rate": 3.5389676336283516e-06, "loss": 0.0014, "step": 155220 }, { "epoch": 2.539965638550274, "grad_norm": 0.07077045738697052, "learning_rate": 3.5382848559192153e-06, "loss": 0.0007, "step": 155230 }, { "epoch": 2.54012926450135, "grad_norm": 0.016491122543811798, "learning_rate": 3.537602108013204e-06, "loss": 0.0005, "step": 155240 }, { "epoch": 2.5402928904524256, "grad_norm": 0.002430260181427002, "learning_rate": 3.536919389924245e-06, "loss": 0.0008, "step": 155250 }, { "epoch": 2.5404565164035016, "grad_norm": 0.016532914713025093, "learning_rate": 3.5362367016662526e-06, "loss": 0.0007, "step": 155260 }, { "epoch": 2.5406201423545776, "grad_norm": 0.030285106971859932, "learning_rate": 3.5355540432531522e-06, "loss": 0.0013, "step": 155270 }, { "epoch": 2.540783768305653, "grad_norm": 0.008719495497643948, "learning_rate": 3.534871414698856e-06, "loss": 0.001, "step": 155280 }, { "epoch": 2.540947394256729, "grad_norm": 0.03444495052099228, "learning_rate": 3.534188816017289e-06, "loss": 0.0014, "step": 155290 }, { "epoch": 2.541111020207805, "grad_norm": 0.0165153369307518, "learning_rate": 3.5335062472223614e-06, "loss": 0.0003, "step": 155300 }, { "epoch": 2.5412746461588807, "grad_norm": 0.02049689181149006, "learning_rate": 3.532823708327998e-06, "loss": 0.0005, "step": 155310 }, { "epoch": 2.5414382721099567, "grad_norm": 0.016284294426441193, "learning_rate": 3.5321411993481086e-06, "loss": 0.0005, "step": 155320 }, { "epoch": 2.5416018980610326, "grad_norm": 0.010798878036439419, "learning_rate": 3.5314587202966146e-06, "loss": 0.0008, "step": 155330 }, { "epoch": 2.541765524012108, "grad_norm": 0.06622608751058578, "learning_rate": 3.530776271187425e-06, "loss": 0.0008, "step": 155340 }, { "epoch": 2.541929149963184, "grad_norm": 0.03575843200087547, "learning_rate": 3.530093852034461e-06, "loss": 0.0007, "step": 155350 }, { "epoch": 2.54209277591426, "grad_norm": 0.06796368211507797, "learning_rate": 3.5294114628516305e-06, "loss": 0.0008, "step": 155360 }, { "epoch": 2.5422564018653357, "grad_norm": 0.040044769644737244, "learning_rate": 3.5287291036528528e-06, "loss": 0.0009, "step": 155370 }, { "epoch": 2.5424200278164117, "grad_norm": 0.004263023845851421, "learning_rate": 3.528046774452033e-06, "loss": 0.0004, "step": 155380 }, { "epoch": 2.5425836537674877, "grad_norm": 0.08759418874979019, "learning_rate": 3.527364475263092e-06, "loss": 0.0008, "step": 155390 }, { "epoch": 2.5427472797185633, "grad_norm": 0.05083611607551575, "learning_rate": 3.5266822060999343e-06, "loss": 0.0008, "step": 155400 }, { "epoch": 2.5429109056696393, "grad_norm": 0.023219618946313858, "learning_rate": 3.5259999669764743e-06, "loss": 0.0012, "step": 155410 }, { "epoch": 2.5430745316207153, "grad_norm": 0.019178232178092003, "learning_rate": 3.525317757906621e-06, "loss": 0.0009, "step": 155420 }, { "epoch": 2.543238157571791, "grad_norm": 0.0025480373296886683, "learning_rate": 3.5246355789042854e-06, "loss": 0.0014, "step": 155430 }, { "epoch": 2.543401783522867, "grad_norm": 0.04563030228018761, "learning_rate": 3.523953429983374e-06, "loss": 0.0012, "step": 155440 }, { "epoch": 2.543565409473943, "grad_norm": 0.016865450888872147, "learning_rate": 3.5232713111577976e-06, "loss": 0.0014, "step": 155450 }, { "epoch": 2.5437290354250184, "grad_norm": 0.01961255632340908, "learning_rate": 3.522589222441463e-06, "loss": 0.0008, "step": 155460 }, { "epoch": 2.5438926613760944, "grad_norm": 0.02856021374464035, "learning_rate": 3.5219071638482785e-06, "loss": 0.0005, "step": 155470 }, { "epoch": 2.54405628732717, "grad_norm": 0.031485866755247116, "learning_rate": 3.521225135392149e-06, "loss": 0.0008, "step": 155480 }, { "epoch": 2.544219913278246, "grad_norm": 0.14037592709064484, "learning_rate": 3.520543137086982e-06, "loss": 0.0011, "step": 155490 }, { "epoch": 2.544383539229322, "grad_norm": 0.05330238491296768, "learning_rate": 3.519861168946681e-06, "loss": 0.0008, "step": 155500 }, { "epoch": 2.5445471651803975, "grad_norm": 0.003014061599969864, "learning_rate": 3.5191792309851535e-06, "loss": 0.0007, "step": 155510 }, { "epoch": 2.5447107911314735, "grad_norm": 0.048802442848682404, "learning_rate": 3.5184973232163013e-06, "loss": 0.0005, "step": 155520 }, { "epoch": 2.544874417082549, "grad_norm": 0.011753559112548828, "learning_rate": 3.5178154456540293e-06, "loss": 0.0009, "step": 155530 }, { "epoch": 2.545038043033625, "grad_norm": 0.006031583994626999, "learning_rate": 3.517133598312238e-06, "loss": 0.0005, "step": 155540 }, { "epoch": 2.545201668984701, "grad_norm": 0.05585257709026337, "learning_rate": 3.5164517812048338e-06, "loss": 0.0009, "step": 155550 }, { "epoch": 2.5453652949357766, "grad_norm": 0.011995119974017143, "learning_rate": 3.5157699943457145e-06, "loss": 0.0005, "step": 155560 }, { "epoch": 2.5455289208868526, "grad_norm": 0.147762268781662, "learning_rate": 3.515088237748784e-06, "loss": 0.001, "step": 155570 }, { "epoch": 2.5456925468379286, "grad_norm": 0.004225740674883127, "learning_rate": 3.5144065114279394e-06, "loss": 0.0006, "step": 155580 }, { "epoch": 2.545856172789004, "grad_norm": 0.10490784049034119, "learning_rate": 3.513724815397085e-06, "loss": 0.0006, "step": 155590 }, { "epoch": 2.54601979874008, "grad_norm": 0.001532597467303276, "learning_rate": 3.5130431496701155e-06, "loss": 0.0005, "step": 155600 }, { "epoch": 2.546183424691156, "grad_norm": 0.004273237194865942, "learning_rate": 3.5123615142609336e-06, "loss": 0.0008, "step": 155610 }, { "epoch": 2.5463470506422317, "grad_norm": 0.00548368226736784, "learning_rate": 3.511679909183433e-06, "loss": 0.0007, "step": 155620 }, { "epoch": 2.5465106765933077, "grad_norm": 0.09682441502809525, "learning_rate": 3.510998334451515e-06, "loss": 0.0012, "step": 155630 }, { "epoch": 2.5466743025443837, "grad_norm": 0.10852310061454773, "learning_rate": 3.5103167900790737e-06, "loss": 0.0009, "step": 155640 }, { "epoch": 2.546837928495459, "grad_norm": 0.01089908741414547, "learning_rate": 3.509635276080007e-06, "loss": 0.0008, "step": 155650 }, { "epoch": 2.547001554446535, "grad_norm": 0.023239292204380035, "learning_rate": 3.5089537924682083e-06, "loss": 0.0007, "step": 155660 }, { "epoch": 2.547165180397611, "grad_norm": 0.10821156203746796, "learning_rate": 3.508272339257575e-06, "loss": 0.0005, "step": 155670 }, { "epoch": 2.5473288063486867, "grad_norm": 0.061035994440317154, "learning_rate": 3.507590916461999e-06, "loss": 0.001, "step": 155680 }, { "epoch": 2.5474924322997627, "grad_norm": 0.05230940505862236, "learning_rate": 3.506909524095376e-06, "loss": 0.0005, "step": 155690 }, { "epoch": 2.5476560582508387, "grad_norm": 0.01959083415567875, "learning_rate": 3.5062281621715965e-06, "loss": 0.0011, "step": 155700 }, { "epoch": 2.5478196842019143, "grad_norm": 0.029158689081668854, "learning_rate": 3.505546830704556e-06, "loss": 0.0005, "step": 155710 }, { "epoch": 2.5479833101529903, "grad_norm": 0.0028493355493992567, "learning_rate": 3.5048655297081436e-06, "loss": 0.0004, "step": 155720 }, { "epoch": 2.5481469361040663, "grad_norm": 0.05171186849474907, "learning_rate": 3.504184259196252e-06, "loss": 0.0008, "step": 155730 }, { "epoch": 2.548310562055142, "grad_norm": 0.0551241859793663, "learning_rate": 3.5035030191827703e-06, "loss": 0.0006, "step": 155740 }, { "epoch": 2.548474188006218, "grad_norm": 0.07664331793785095, "learning_rate": 3.5028218096815904e-06, "loss": 0.0017, "step": 155750 }, { "epoch": 2.548637813957294, "grad_norm": 0.031039239838719368, "learning_rate": 3.5021406307065996e-06, "loss": 0.0005, "step": 155760 }, { "epoch": 2.5488014399083694, "grad_norm": 0.030492762103676796, "learning_rate": 3.5014594822716884e-06, "loss": 0.0008, "step": 155770 }, { "epoch": 2.5489650658594454, "grad_norm": 0.03154410049319267, "learning_rate": 3.5007783643907434e-06, "loss": 0.0008, "step": 155780 }, { "epoch": 2.5491286918105214, "grad_norm": 0.06998840719461441, "learning_rate": 3.500097277077653e-06, "loss": 0.0008, "step": 155790 }, { "epoch": 2.549292317761597, "grad_norm": 0.008669944480061531, "learning_rate": 3.499416220346302e-06, "loss": 0.0006, "step": 155800 }, { "epoch": 2.549455943712673, "grad_norm": 0.04059765115380287, "learning_rate": 3.4987351942105796e-06, "loss": 0.0006, "step": 155810 }, { "epoch": 2.549619569663749, "grad_norm": 0.03775558993220329, "learning_rate": 3.4980541986843684e-06, "loss": 0.0004, "step": 155820 }, { "epoch": 2.5497831956148245, "grad_norm": 0.026322510093450546, "learning_rate": 3.497373233781557e-06, "loss": 0.0007, "step": 155830 }, { "epoch": 2.5499468215659005, "grad_norm": 0.004262750968337059, "learning_rate": 3.496692299516025e-06, "loss": 0.001, "step": 155840 }, { "epoch": 2.5501104475169765, "grad_norm": 0.018891286104917526, "learning_rate": 3.4960113959016607e-06, "loss": 0.0007, "step": 155850 }, { "epoch": 2.550274073468052, "grad_norm": 0.027648668736219406, "learning_rate": 3.4953305229523426e-06, "loss": 0.001, "step": 155860 }, { "epoch": 2.550437699419128, "grad_norm": 0.03756139799952507, "learning_rate": 3.4946496806819587e-06, "loss": 0.0008, "step": 155870 }, { "epoch": 2.5506013253702036, "grad_norm": 0.05730848386883736, "learning_rate": 3.4939688691043836e-06, "loss": 0.0007, "step": 155880 }, { "epoch": 2.5507649513212796, "grad_norm": 0.002704297425225377, "learning_rate": 3.4932880882335063e-06, "loss": 0.0003, "step": 155890 }, { "epoch": 2.5509285772723556, "grad_norm": 0.02576109580695629, "learning_rate": 3.4926073380831994e-06, "loss": 0.0007, "step": 155900 }, { "epoch": 2.551092203223431, "grad_norm": 0.05891149491071701, "learning_rate": 3.4919266186673508e-06, "loss": 0.0007, "step": 155910 }, { "epoch": 2.551255829174507, "grad_norm": 0.03261983022093773, "learning_rate": 3.4912459299998324e-06, "loss": 0.0011, "step": 155920 }, { "epoch": 2.5514194551255827, "grad_norm": 0.07009325176477432, "learning_rate": 3.490565272094529e-06, "loss": 0.0005, "step": 155930 }, { "epoch": 2.5515830810766587, "grad_norm": 0.027914224192500114, "learning_rate": 3.489884644965314e-06, "loss": 0.001, "step": 155940 }, { "epoch": 2.5517467070277347, "grad_norm": 0.011682286858558655, "learning_rate": 3.489204048626069e-06, "loss": 0.0006, "step": 155950 }, { "epoch": 2.55191033297881, "grad_norm": 0.12903010845184326, "learning_rate": 3.4885234830906663e-06, "loss": 0.0019, "step": 155960 }, { "epoch": 2.552073958929886, "grad_norm": 0.023473335430026054, "learning_rate": 3.487842948372987e-06, "loss": 0.0005, "step": 155970 }, { "epoch": 2.552237584880962, "grad_norm": 0.014645365998148918, "learning_rate": 3.4871624444869e-06, "loss": 0.0005, "step": 155980 }, { "epoch": 2.5524012108320377, "grad_norm": 0.029547706246376038, "learning_rate": 3.4864819714462885e-06, "loss": 0.0008, "step": 155990 }, { "epoch": 2.5525648367831137, "grad_norm": 0.037960391491651535, "learning_rate": 3.485801529265018e-06, "loss": 0.0012, "step": 156000 }, { "epoch": 2.5527284627341897, "grad_norm": 0.12060227245092392, "learning_rate": 3.485121117956971e-06, "loss": 0.0015, "step": 156010 }, { "epoch": 2.5528920886852653, "grad_norm": 0.017749857157468796, "learning_rate": 3.484440737536012e-06, "loss": 0.0008, "step": 156020 }, { "epoch": 2.5530557146363413, "grad_norm": 0.04311879351735115, "learning_rate": 3.483760388016021e-06, "loss": 0.0006, "step": 156030 }, { "epoch": 2.5532193405874173, "grad_norm": 0.0033219093456864357, "learning_rate": 3.483080069410864e-06, "loss": 0.0004, "step": 156040 }, { "epoch": 2.553382966538493, "grad_norm": 0.03718552365899086, "learning_rate": 3.4823997817344155e-06, "loss": 0.0014, "step": 156050 }, { "epoch": 2.553546592489569, "grad_norm": 0.07061880081892014, "learning_rate": 3.481719525000543e-06, "loss": 0.001, "step": 156060 }, { "epoch": 2.553710218440645, "grad_norm": 0.04559210687875748, "learning_rate": 3.48103929922312e-06, "loss": 0.001, "step": 156070 }, { "epoch": 2.5538738443917204, "grad_norm": 0.09137696027755737, "learning_rate": 3.4803591044160122e-06, "loss": 0.0019, "step": 156080 }, { "epoch": 2.5540374703427964, "grad_norm": 0.011784307658672333, "learning_rate": 3.479678940593092e-06, "loss": 0.0008, "step": 156090 }, { "epoch": 2.5542010962938724, "grad_norm": 0.06484616547822952, "learning_rate": 3.478998807768223e-06, "loss": 0.0012, "step": 156100 }, { "epoch": 2.554364722244948, "grad_norm": 0.0034737687092274427, "learning_rate": 3.478318705955276e-06, "loss": 0.0006, "step": 156110 }, { "epoch": 2.554528348196024, "grad_norm": 0.008703572675585747, "learning_rate": 3.4776386351681157e-06, "loss": 0.0007, "step": 156120 }, { "epoch": 2.5546919741471, "grad_norm": 0.03597158566117287, "learning_rate": 3.4769585954206096e-06, "loss": 0.0006, "step": 156130 }, { "epoch": 2.5548556000981755, "grad_norm": 0.0020502132829278708, "learning_rate": 3.476278586726622e-06, "loss": 0.0003, "step": 156140 }, { "epoch": 2.5550192260492515, "grad_norm": 0.21587726473808289, "learning_rate": 3.4755986091000187e-06, "loss": 0.0008, "step": 156150 }, { "epoch": 2.5551828520003275, "grad_norm": 0.06614086776971817, "learning_rate": 3.4749186625546627e-06, "loss": 0.0012, "step": 156160 }, { "epoch": 2.555346477951403, "grad_norm": 0.038901422172784805, "learning_rate": 3.474238747104418e-06, "loss": 0.0026, "step": 156170 }, { "epoch": 2.555510103902479, "grad_norm": 0.0037337292451411486, "learning_rate": 3.4735588627631476e-06, "loss": 0.0014, "step": 156180 }, { "epoch": 2.555673729853555, "grad_norm": 0.07874343544244766, "learning_rate": 3.4728790095447153e-06, "loss": 0.0008, "step": 156190 }, { "epoch": 2.5558373558046306, "grad_norm": 0.03183991089463234, "learning_rate": 3.472199187462979e-06, "loss": 0.0013, "step": 156200 }, { "epoch": 2.5560009817557066, "grad_norm": 0.07178599387407303, "learning_rate": 3.471519396531804e-06, "loss": 0.0009, "step": 156210 }, { "epoch": 2.5561646077067826, "grad_norm": 0.03478189557790756, "learning_rate": 3.4708396367650466e-06, "loss": 0.0007, "step": 156220 }, { "epoch": 2.556328233657858, "grad_norm": 0.002395372372120619, "learning_rate": 3.47015990817657e-06, "loss": 0.0008, "step": 156230 }, { "epoch": 2.556491859608934, "grad_norm": 0.010581253096461296, "learning_rate": 3.4694802107802304e-06, "loss": 0.0008, "step": 156240 }, { "epoch": 2.5566554855600097, "grad_norm": 0.06119757145643234, "learning_rate": 3.46880054458989e-06, "loss": 0.0019, "step": 156250 }, { "epoch": 2.5568191115110857, "grad_norm": 0.07341938465833664, "learning_rate": 3.468120909619402e-06, "loss": 0.0011, "step": 156260 }, { "epoch": 2.5569827374621616, "grad_norm": 0.06266360729932785, "learning_rate": 3.4674413058826272e-06, "loss": 0.0008, "step": 156270 }, { "epoch": 2.557146363413237, "grad_norm": 0.032503388822078705, "learning_rate": 3.466761733393419e-06, "loss": 0.0007, "step": 156280 }, { "epoch": 2.557309989364313, "grad_norm": 0.0381377637386322, "learning_rate": 3.4660821921656372e-06, "loss": 0.0009, "step": 156290 }, { "epoch": 2.5574736153153887, "grad_norm": 0.02051936835050583, "learning_rate": 3.4654026822131335e-06, "loss": 0.001, "step": 156300 }, { "epoch": 2.5576372412664647, "grad_norm": 0.10680248588323593, "learning_rate": 3.4647232035497647e-06, "loss": 0.0006, "step": 156310 }, { "epoch": 2.5578008672175407, "grad_norm": 0.04586634039878845, "learning_rate": 3.4640437561893836e-06, "loss": 0.001, "step": 156320 }, { "epoch": 2.5579644931686163, "grad_norm": 0.011363355442881584, "learning_rate": 3.463364340145845e-06, "loss": 0.0005, "step": 156330 }, { "epoch": 2.5581281191196923, "grad_norm": 0.07225383818149567, "learning_rate": 3.462684955432999e-06, "loss": 0.001, "step": 156340 }, { "epoch": 2.5582917450707683, "grad_norm": 0.030570069327950478, "learning_rate": 3.462005602064701e-06, "loss": 0.0005, "step": 156350 }, { "epoch": 2.558455371021844, "grad_norm": 0.0445258803665638, "learning_rate": 3.4613262800548e-06, "loss": 0.0008, "step": 156360 }, { "epoch": 2.55861899697292, "grad_norm": 0.07010691612958908, "learning_rate": 3.460646989417148e-06, "loss": 0.0006, "step": 156370 }, { "epoch": 2.558782622923996, "grad_norm": 0.038842249661684036, "learning_rate": 3.459967730165593e-06, "loss": 0.0005, "step": 156380 }, { "epoch": 2.5589462488750714, "grad_norm": 0.003404854563996196, "learning_rate": 3.4592885023139884e-06, "loss": 0.0005, "step": 156390 }, { "epoch": 2.5591098748261474, "grad_norm": 0.020853010937571526, "learning_rate": 3.4586093058761793e-06, "loss": 0.0005, "step": 156400 }, { "epoch": 2.5592735007772234, "grad_norm": 0.04782874509692192, "learning_rate": 3.457930140866017e-06, "loss": 0.0006, "step": 156410 }, { "epoch": 2.559437126728299, "grad_norm": 0.014162102714180946, "learning_rate": 3.4572510072973464e-06, "loss": 0.0009, "step": 156420 }, { "epoch": 2.559600752679375, "grad_norm": 0.06350173056125641, "learning_rate": 3.4565719051840163e-06, "loss": 0.0015, "step": 156430 }, { "epoch": 2.559764378630451, "grad_norm": 0.03668050840497017, "learning_rate": 3.4558928345398724e-06, "loss": 0.0005, "step": 156440 }, { "epoch": 2.5599280045815265, "grad_norm": 0.022807583212852478, "learning_rate": 3.4552137953787607e-06, "loss": 0.0011, "step": 156450 }, { "epoch": 2.5600916305326025, "grad_norm": 0.027117786929011345, "learning_rate": 3.454534787714525e-06, "loss": 0.0008, "step": 156460 }, { "epoch": 2.5602552564836785, "grad_norm": 0.028765177354216576, "learning_rate": 3.4538558115610115e-06, "loss": 0.0008, "step": 156470 }, { "epoch": 2.560418882434754, "grad_norm": 0.02627740427851677, "learning_rate": 3.453176866932062e-06, "loss": 0.0007, "step": 156480 }, { "epoch": 2.56058250838583, "grad_norm": 0.030667908489704132, "learning_rate": 3.4524979538415222e-06, "loss": 0.0003, "step": 156490 }, { "epoch": 2.560746134336906, "grad_norm": 0.04381543770432472, "learning_rate": 3.4518190723032317e-06, "loss": 0.0008, "step": 156500 }, { "epoch": 2.5609097602879816, "grad_norm": 0.028581634163856506, "learning_rate": 3.451140222331035e-06, "loss": 0.0006, "step": 156510 }, { "epoch": 2.5610733862390576, "grad_norm": 0.021193278953433037, "learning_rate": 3.4504614039387697e-06, "loss": 0.0008, "step": 156520 }, { "epoch": 2.5612370121901336, "grad_norm": 0.0010733207454904914, "learning_rate": 3.4497826171402817e-06, "loss": 0.0009, "step": 156530 }, { "epoch": 2.561400638141209, "grad_norm": 0.06268070638179779, "learning_rate": 3.4491038619494046e-06, "loss": 0.0009, "step": 156540 }, { "epoch": 2.561564264092285, "grad_norm": 0.003617148147895932, "learning_rate": 3.4484251383799834e-06, "loss": 0.0012, "step": 156550 }, { "epoch": 2.561727890043361, "grad_norm": 0.0009536351426504552, "learning_rate": 3.447746446445851e-06, "loss": 0.0004, "step": 156560 }, { "epoch": 2.5618915159944367, "grad_norm": 0.0008971596253104508, "learning_rate": 3.4470677861608514e-06, "loss": 0.0008, "step": 156570 }, { "epoch": 2.5620551419455126, "grad_norm": 0.019958391785621643, "learning_rate": 3.4463891575388164e-06, "loss": 0.0008, "step": 156580 }, { "epoch": 2.5622187678965886, "grad_norm": 0.024905450642108917, "learning_rate": 3.445710560593588e-06, "loss": 0.001, "step": 156590 }, { "epoch": 2.562382393847664, "grad_norm": 0.0035541777033358812, "learning_rate": 3.4450319953389967e-06, "loss": 0.0003, "step": 156600 }, { "epoch": 2.56254601979874, "grad_norm": 0.02959735319018364, "learning_rate": 3.444353461788883e-06, "loss": 0.0012, "step": 156610 }, { "epoch": 2.562709645749816, "grad_norm": 0.06399136036634445, "learning_rate": 3.443674959957076e-06, "loss": 0.0004, "step": 156620 }, { "epoch": 2.5628732717008917, "grad_norm": 0.07510251551866531, "learning_rate": 3.4429964898574165e-06, "loss": 0.0007, "step": 156630 }, { "epoch": 2.5630368976519677, "grad_norm": 0.057063519954681396, "learning_rate": 3.4423180515037313e-06, "loss": 0.0011, "step": 156640 }, { "epoch": 2.5632005236030433, "grad_norm": 0.023716500028967857, "learning_rate": 3.4416396449098587e-06, "loss": 0.0009, "step": 156650 }, { "epoch": 2.5633641495541193, "grad_norm": 0.0013581697130575776, "learning_rate": 3.440961270089626e-06, "loss": 0.0006, "step": 156660 }, { "epoch": 2.5635277755051953, "grad_norm": 0.00829810556024313, "learning_rate": 3.44028292705687e-06, "loss": 0.0007, "step": 156670 }, { "epoch": 2.563691401456271, "grad_norm": 0.00016792817041277885, "learning_rate": 3.439604615825416e-06, "loss": 0.0008, "step": 156680 }, { "epoch": 2.563855027407347, "grad_norm": 0.03306039422750473, "learning_rate": 3.438926336409099e-06, "loss": 0.0005, "step": 156690 }, { "epoch": 2.5640186533584224, "grad_norm": 0.0767318531870842, "learning_rate": 3.438248088821745e-06, "loss": 0.0008, "step": 156700 }, { "epoch": 2.5641822793094984, "grad_norm": 0.018030716106295586, "learning_rate": 3.4375698730771844e-06, "loss": 0.001, "step": 156710 }, { "epoch": 2.5643459052605744, "grad_norm": 0.021547790616750717, "learning_rate": 3.4368916891892457e-06, "loss": 0.0007, "step": 156720 }, { "epoch": 2.56450953121165, "grad_norm": 0.010201559402048588, "learning_rate": 3.436213537171756e-06, "loss": 0.0013, "step": 156730 }, { "epoch": 2.564673157162726, "grad_norm": 0.08355458825826645, "learning_rate": 3.4355354170385413e-06, "loss": 0.0013, "step": 156740 }, { "epoch": 2.564836783113802, "grad_norm": 0.010916578583419323, "learning_rate": 3.4348573288034304e-06, "loss": 0.0007, "step": 156750 }, { "epoch": 2.5650004090648775, "grad_norm": 0.06279832124710083, "learning_rate": 3.434179272480246e-06, "loss": 0.0007, "step": 156760 }, { "epoch": 2.5651640350159535, "grad_norm": 0.04385913163423538, "learning_rate": 3.4335012480828162e-06, "loss": 0.001, "step": 156770 }, { "epoch": 2.5653276609670295, "grad_norm": 0.015272158198058605, "learning_rate": 3.432823255624962e-06, "loss": 0.0006, "step": 156780 }, { "epoch": 2.565491286918105, "grad_norm": 0.02016819268465042, "learning_rate": 3.43214529512051e-06, "loss": 0.0008, "step": 156790 }, { "epoch": 2.565654912869181, "grad_norm": 0.07799045741558075, "learning_rate": 3.4314673665832814e-06, "loss": 0.0006, "step": 156800 }, { "epoch": 2.565818538820257, "grad_norm": 0.050520554184913635, "learning_rate": 3.4307894700271e-06, "loss": 0.001, "step": 156810 }, { "epoch": 2.5659821647713326, "grad_norm": 0.07365068793296814, "learning_rate": 3.4301116054657856e-06, "loss": 0.0013, "step": 156820 }, { "epoch": 2.5661457907224086, "grad_norm": 0.05136330798268318, "learning_rate": 3.4294337729131615e-06, "loss": 0.0005, "step": 156830 }, { "epoch": 2.5663094166734846, "grad_norm": 0.07425063103437424, "learning_rate": 3.428755972383046e-06, "loss": 0.001, "step": 156840 }, { "epoch": 2.56647304262456, "grad_norm": 0.05179853364825249, "learning_rate": 3.4280782038892614e-06, "loss": 0.001, "step": 156850 }, { "epoch": 2.566636668575636, "grad_norm": 0.01760115846991539, "learning_rate": 3.4274004674456233e-06, "loss": 0.0002, "step": 156860 }, { "epoch": 2.566800294526712, "grad_norm": 0.02280079945921898, "learning_rate": 3.426722763065954e-06, "loss": 0.0005, "step": 156870 }, { "epoch": 2.5669639204777877, "grad_norm": 0.0541527196764946, "learning_rate": 3.4260450907640684e-06, "loss": 0.0018, "step": 156880 }, { "epoch": 2.5671275464288636, "grad_norm": 0.034644052386283875, "learning_rate": 3.425367450553786e-06, "loss": 0.0007, "step": 156890 }, { "epoch": 2.5672911723799396, "grad_norm": 0.11106634140014648, "learning_rate": 3.4246898424489205e-06, "loss": 0.001, "step": 156900 }, { "epoch": 2.567454798331015, "grad_norm": 0.0049230302684009075, "learning_rate": 3.4240122664632906e-06, "loss": 0.0005, "step": 156910 }, { "epoch": 2.567618424282091, "grad_norm": 0.0412871278822422, "learning_rate": 3.4233347226107093e-06, "loss": 0.001, "step": 156920 }, { "epoch": 2.567782050233167, "grad_norm": 0.024426085874438286, "learning_rate": 3.422657210904994e-06, "loss": 0.0006, "step": 156930 }, { "epoch": 2.5679456761842427, "grad_norm": 0.018580403178930283, "learning_rate": 3.4219797313599546e-06, "loss": 0.0008, "step": 156940 }, { "epoch": 2.5681093021353187, "grad_norm": 0.09222965687513351, "learning_rate": 3.4213022839894074e-06, "loss": 0.0011, "step": 156950 }, { "epoch": 2.5682729280863947, "grad_norm": 0.03605073317885399, "learning_rate": 3.420624868807163e-06, "loss": 0.0008, "step": 156960 }, { "epoch": 2.5684365540374703, "grad_norm": 0.02080719545483589, "learning_rate": 3.419947485827036e-06, "loss": 0.0005, "step": 156970 }, { "epoch": 2.5686001799885463, "grad_norm": 0.023236917331814766, "learning_rate": 3.4192701350628344e-06, "loss": 0.0004, "step": 156980 }, { "epoch": 2.5687638059396223, "grad_norm": 0.03065578080713749, "learning_rate": 3.4185928165283723e-06, "loss": 0.0012, "step": 156990 }, { "epoch": 2.568927431890698, "grad_norm": 0.018898261711001396, "learning_rate": 3.4179155302374557e-06, "loss": 0.0006, "step": 157000 }, { "epoch": 2.569091057841774, "grad_norm": 0.008826748467981815, "learning_rate": 3.417238276203897e-06, "loss": 0.0004, "step": 157010 }, { "epoch": 2.5692546837928494, "grad_norm": 0.026285499334335327, "learning_rate": 3.416561054441503e-06, "loss": 0.0008, "step": 157020 }, { "epoch": 2.5694183097439254, "grad_norm": 0.05668025091290474, "learning_rate": 3.4158838649640837e-06, "loss": 0.0008, "step": 157030 }, { "epoch": 2.5695819356950014, "grad_norm": 0.008223298005759716, "learning_rate": 3.4152067077854433e-06, "loss": 0.0006, "step": 157040 }, { "epoch": 2.569745561646077, "grad_norm": 0.15519556403160095, "learning_rate": 3.4145295829193925e-06, "loss": 0.0009, "step": 157050 }, { "epoch": 2.569909187597153, "grad_norm": 0.02406831458210945, "learning_rate": 3.413852490379733e-06, "loss": 0.0009, "step": 157060 }, { "epoch": 2.5700728135482285, "grad_norm": 0.06633897870779037, "learning_rate": 3.413175430180274e-06, "loss": 0.0004, "step": 157070 }, { "epoch": 2.5702364394993045, "grad_norm": 0.02159939706325531, "learning_rate": 3.412498402334817e-06, "loss": 0.0004, "step": 157080 }, { "epoch": 2.5704000654503805, "grad_norm": 0.023533523082733154, "learning_rate": 3.4118214068571686e-06, "loss": 0.0009, "step": 157090 }, { "epoch": 2.570563691401456, "grad_norm": 0.06542139500379562, "learning_rate": 3.41114444376113e-06, "loss": 0.0006, "step": 157100 }, { "epoch": 2.570727317352532, "grad_norm": 0.05887893587350845, "learning_rate": 3.4104675130605057e-06, "loss": 0.0005, "step": 157110 }, { "epoch": 2.570890943303608, "grad_norm": 0.08314459770917892, "learning_rate": 3.409790614769096e-06, "loss": 0.0012, "step": 157120 }, { "epoch": 2.5710545692546836, "grad_norm": 0.002007189439609647, "learning_rate": 3.409113748900705e-06, "loss": 0.0005, "step": 157130 }, { "epoch": 2.5712181952057596, "grad_norm": 0.000473410909762606, "learning_rate": 3.40843691546913e-06, "loss": 0.0007, "step": 157140 }, { "epoch": 2.5713818211568356, "grad_norm": 0.0686216801404953, "learning_rate": 3.407760114488175e-06, "loss": 0.001, "step": 157150 }, { "epoch": 2.571545447107911, "grad_norm": 0.025339482352137566, "learning_rate": 3.407083345971634e-06, "loss": 0.001, "step": 157160 }, { "epoch": 2.571709073058987, "grad_norm": 0.01766258105635643, "learning_rate": 3.4064066099333116e-06, "loss": 0.0005, "step": 157170 }, { "epoch": 2.571872699010063, "grad_norm": 0.06514538824558258, "learning_rate": 3.405729906387001e-06, "loss": 0.0009, "step": 157180 }, { "epoch": 2.5720363249611387, "grad_norm": 0.041254766285419464, "learning_rate": 3.405053235346504e-06, "loss": 0.0009, "step": 157190 }, { "epoch": 2.5721999509122146, "grad_norm": 0.0225360207259655, "learning_rate": 3.4043765968256117e-06, "loss": 0.0014, "step": 157200 }, { "epoch": 2.5723635768632906, "grad_norm": 0.50986248254776, "learning_rate": 3.4036999908381275e-06, "loss": 0.0006, "step": 157210 }, { "epoch": 2.572527202814366, "grad_norm": 0.012552433647215366, "learning_rate": 3.4030234173978387e-06, "loss": 0.0006, "step": 157220 }, { "epoch": 2.572690828765442, "grad_norm": 0.01067019160836935, "learning_rate": 3.4023468765185487e-06, "loss": 0.0011, "step": 157230 }, { "epoch": 2.572854454716518, "grad_norm": 0.05095044896006584, "learning_rate": 3.4016703682140427e-06, "loss": 0.0007, "step": 157240 }, { "epoch": 2.5730180806675937, "grad_norm": 0.18578234314918518, "learning_rate": 3.4009938924981223e-06, "loss": 0.0026, "step": 157250 }, { "epoch": 2.5731817066186697, "grad_norm": 0.02256818301975727, "learning_rate": 3.4003174493845737e-06, "loss": 0.0011, "step": 157260 }, { "epoch": 2.5733453325697457, "grad_norm": 0.025090064853429794, "learning_rate": 3.399641038887194e-06, "loss": 0.0009, "step": 157270 }, { "epoch": 2.5735089585208213, "grad_norm": 0.014943055808544159, "learning_rate": 3.39896466101977e-06, "loss": 0.0006, "step": 157280 }, { "epoch": 2.5736725844718973, "grad_norm": 0.040379174053668976, "learning_rate": 3.398288315796098e-06, "loss": 0.0011, "step": 157290 }, { "epoch": 2.5738362104229733, "grad_norm": 0.07150556147098541, "learning_rate": 3.397612003229962e-06, "loss": 0.0015, "step": 157300 }, { "epoch": 2.573999836374049, "grad_norm": 0.07028814405202866, "learning_rate": 3.3969357233351575e-06, "loss": 0.0009, "step": 157310 }, { "epoch": 2.574163462325125, "grad_norm": 0.029448578134179115, "learning_rate": 3.3962594761254676e-06, "loss": 0.0006, "step": 157320 }, { "epoch": 2.574327088276201, "grad_norm": 0.0673682764172554, "learning_rate": 3.3955832616146853e-06, "loss": 0.0012, "step": 157330 }, { "epoch": 2.5744907142272764, "grad_norm": 0.04013373330235481, "learning_rate": 3.394907079816594e-06, "loss": 0.0009, "step": 157340 }, { "epoch": 2.5746543401783524, "grad_norm": 0.021513070911169052, "learning_rate": 3.394230930744984e-06, "loss": 0.0008, "step": 157350 }, { "epoch": 2.5748179661294284, "grad_norm": 0.09889757633209229, "learning_rate": 3.3935548144136377e-06, "loss": 0.001, "step": 157360 }, { "epoch": 2.574981592080504, "grad_norm": 0.011663763783872128, "learning_rate": 3.392878730836344e-06, "loss": 0.0004, "step": 157370 }, { "epoch": 2.57514521803158, "grad_norm": 0.004875568673014641, "learning_rate": 3.3922026800268847e-06, "loss": 0.0013, "step": 157380 }, { "epoch": 2.575308843982656, "grad_norm": 0.034691810607910156, "learning_rate": 3.391526661999047e-06, "loss": 0.0006, "step": 157390 }, { "epoch": 2.5754724699337315, "grad_norm": 0.04819081351161003, "learning_rate": 3.3908506767666116e-06, "loss": 0.0008, "step": 157400 }, { "epoch": 2.5756360958848075, "grad_norm": 0.010417142882943153, "learning_rate": 3.390174724343363e-06, "loss": 0.0004, "step": 157410 }, { "epoch": 2.575799721835883, "grad_norm": 0.02954253740608692, "learning_rate": 3.389498804743083e-06, "loss": 0.0006, "step": 157420 }, { "epoch": 2.575963347786959, "grad_norm": 0.06487741321325302, "learning_rate": 3.3888229179795518e-06, "loss": 0.0009, "step": 157430 }, { "epoch": 2.576126973738035, "grad_norm": 0.07638309895992279, "learning_rate": 3.3881470640665516e-06, "loss": 0.0008, "step": 157440 }, { "epoch": 2.5762905996891106, "grad_norm": 0.0389283150434494, "learning_rate": 3.3874712430178607e-06, "loss": 0.0003, "step": 157450 }, { "epoch": 2.5764542256401866, "grad_norm": 0.07186105847358704, "learning_rate": 3.3867954548472614e-06, "loss": 0.0008, "step": 157460 }, { "epoch": 2.576617851591262, "grad_norm": 0.10738056153059006, "learning_rate": 3.38611969956853e-06, "loss": 0.0004, "step": 157470 }, { "epoch": 2.576781477542338, "grad_norm": 0.03169473260641098, "learning_rate": 3.3854439771954452e-06, "loss": 0.001, "step": 157480 }, { "epoch": 2.576945103493414, "grad_norm": 0.0014467404689639807, "learning_rate": 3.384768287741784e-06, "loss": 0.0009, "step": 157490 }, { "epoch": 2.5771087294444897, "grad_norm": 0.04649551957845688, "learning_rate": 3.384092631221325e-06, "loss": 0.0008, "step": 157500 }, { "epoch": 2.5772723553955657, "grad_norm": 0.031050965189933777, "learning_rate": 3.383417007647842e-06, "loss": 0.0007, "step": 157510 }, { "epoch": 2.5774359813466416, "grad_norm": 0.02913885936141014, "learning_rate": 3.3827414170351125e-06, "loss": 0.0014, "step": 157520 }, { "epoch": 2.577599607297717, "grad_norm": 0.023262230679392815, "learning_rate": 3.382065859396909e-06, "loss": 0.0009, "step": 157530 }, { "epoch": 2.577763233248793, "grad_norm": 0.01650875248014927, "learning_rate": 3.381390334747008e-06, "loss": 0.0006, "step": 157540 }, { "epoch": 2.577926859199869, "grad_norm": 0.014173058792948723, "learning_rate": 3.3807148430991798e-06, "loss": 0.0007, "step": 157550 }, { "epoch": 2.5780904851509447, "grad_norm": 0.016393011435866356, "learning_rate": 3.3800393844671996e-06, "loss": 0.0009, "step": 157560 }, { "epoch": 2.5782541111020207, "grad_norm": 0.019639667123556137, "learning_rate": 3.3793639588648385e-06, "loss": 0.0005, "step": 157570 }, { "epoch": 2.5784177370530967, "grad_norm": 0.06607376784086227, "learning_rate": 3.378688566305869e-06, "loss": 0.001, "step": 157580 }, { "epoch": 2.5785813630041723, "grad_norm": 0.017213400453329086, "learning_rate": 3.3780132068040587e-06, "loss": 0.0005, "step": 157590 }, { "epoch": 2.5787449889552483, "grad_norm": 0.008348105475306511, "learning_rate": 3.3773378803731816e-06, "loss": 0.0005, "step": 157600 }, { "epoch": 2.5789086149063243, "grad_norm": 0.026355881243944168, "learning_rate": 3.376662587027003e-06, "loss": 0.001, "step": 157610 }, { "epoch": 2.5790722408574, "grad_norm": 0.1166093721985817, "learning_rate": 3.375987326779297e-06, "loss": 0.0016, "step": 157620 }, { "epoch": 2.579235866808476, "grad_norm": 0.08205459266901016, "learning_rate": 3.3753120996438243e-06, "loss": 0.0012, "step": 157630 }, { "epoch": 2.579399492759552, "grad_norm": 0.017083683982491493, "learning_rate": 3.3746369056343597e-06, "loss": 0.0003, "step": 157640 }, { "epoch": 2.5795631187106274, "grad_norm": 0.008056187070906162, "learning_rate": 3.373961744764663e-06, "loss": 0.0003, "step": 157650 }, { "epoch": 2.5797267446617034, "grad_norm": 0.02203085459768772, "learning_rate": 3.3732866170485066e-06, "loss": 0.0006, "step": 157660 }, { "epoch": 2.5798903706127794, "grad_norm": 0.025435173884034157, "learning_rate": 3.3726115224996493e-06, "loss": 0.0008, "step": 157670 }, { "epoch": 2.580053996563855, "grad_norm": 0.04770356044173241, "learning_rate": 3.371936461131862e-06, "loss": 0.0006, "step": 157680 }, { "epoch": 2.580217622514931, "grad_norm": 0.03735186532139778, "learning_rate": 3.3712614329589032e-06, "loss": 0.0009, "step": 157690 }, { "epoch": 2.580381248466007, "grad_norm": 0.04112925007939339, "learning_rate": 3.370586437994541e-06, "loss": 0.0008, "step": 157700 }, { "epoch": 2.5805448744170825, "grad_norm": 0.010151886381208897, "learning_rate": 3.369911476252533e-06, "loss": 0.0003, "step": 157710 }, { "epoch": 2.5807085003681585, "grad_norm": 0.0360533706843853, "learning_rate": 3.369236547746646e-06, "loss": 0.0004, "step": 157720 }, { "epoch": 2.5808721263192345, "grad_norm": 0.012467307969927788, "learning_rate": 3.3685616524906363e-06, "loss": 0.0006, "step": 157730 }, { "epoch": 2.58103575227031, "grad_norm": 0.03771194815635681, "learning_rate": 3.3678867904982702e-06, "loss": 0.0007, "step": 157740 }, { "epoch": 2.581199378221386, "grad_norm": 0.028694024309515953, "learning_rate": 3.367211961783301e-06, "loss": 0.0008, "step": 157750 }, { "epoch": 2.581363004172462, "grad_norm": 0.052467942237854004, "learning_rate": 3.366537166359495e-06, "loss": 0.0012, "step": 157760 }, { "epoch": 2.5815266301235376, "grad_norm": 0.004054373130202293, "learning_rate": 3.365862404240603e-06, "loss": 0.0004, "step": 157770 }, { "epoch": 2.5816902560746136, "grad_norm": 0.041726529598236084, "learning_rate": 3.3651876754403905e-06, "loss": 0.0005, "step": 157780 }, { "epoch": 2.5818538820256896, "grad_norm": 0.05667885020375252, "learning_rate": 3.3645129799726084e-06, "loss": 0.0007, "step": 157790 }, { "epoch": 2.582017507976765, "grad_norm": 0.12481068074703217, "learning_rate": 3.363838317851017e-06, "loss": 0.0008, "step": 157800 }, { "epoch": 2.582181133927841, "grad_norm": 0.05373399704694748, "learning_rate": 3.3631636890893706e-06, "loss": 0.0007, "step": 157810 }, { "epoch": 2.5823447598789167, "grad_norm": 0.004587383475154638, "learning_rate": 3.3624890937014243e-06, "loss": 0.0007, "step": 157820 }, { "epoch": 2.5825083858299926, "grad_norm": 0.0629727691411972, "learning_rate": 3.361814531700932e-06, "loss": 0.0011, "step": 157830 }, { "epoch": 2.5826720117810686, "grad_norm": 0.11971531063318253, "learning_rate": 3.3611400031016493e-06, "loss": 0.0014, "step": 157840 }, { "epoch": 2.582835637732144, "grad_norm": 0.0392579510807991, "learning_rate": 3.360465507917327e-06, "loss": 0.0016, "step": 157850 }, { "epoch": 2.58299926368322, "grad_norm": 0.03444840759038925, "learning_rate": 3.35979104616172e-06, "loss": 0.0008, "step": 157860 }, { "epoch": 2.5831628896342957, "grad_norm": 0.049059391021728516, "learning_rate": 3.359116617848577e-06, "loss": 0.0008, "step": 157870 }, { "epoch": 2.5833265155853717, "grad_norm": 0.06486990302801132, "learning_rate": 3.358442222991652e-06, "loss": 0.0007, "step": 157880 }, { "epoch": 2.5834901415364477, "grad_norm": 0.022487863898277283, "learning_rate": 3.3577678616046926e-06, "loss": 0.0012, "step": 157890 }, { "epoch": 2.5836537674875233, "grad_norm": 0.021492572501301765, "learning_rate": 3.3570935337014514e-06, "loss": 0.0005, "step": 157900 }, { "epoch": 2.5838173934385993, "grad_norm": 0.022156106308102608, "learning_rate": 3.3564192392956752e-06, "loss": 0.0006, "step": 157910 }, { "epoch": 2.5839810193896753, "grad_norm": 0.022072531282901764, "learning_rate": 3.355744978401113e-06, "loss": 0.0007, "step": 157920 }, { "epoch": 2.584144645340751, "grad_norm": 0.06856640428304672, "learning_rate": 3.3550707510315115e-06, "loss": 0.0009, "step": 157930 }, { "epoch": 2.584308271291827, "grad_norm": 0.13029427826404572, "learning_rate": 3.35439655720062e-06, "loss": 0.0006, "step": 157940 }, { "epoch": 2.584471897242903, "grad_norm": 0.03007562831044197, "learning_rate": 3.353722396922181e-06, "loss": 0.0006, "step": 157950 }, { "epoch": 2.5846355231939784, "grad_norm": 0.12350991368293762, "learning_rate": 3.353048270209944e-06, "loss": 0.0012, "step": 157960 }, { "epoch": 2.5847991491450544, "grad_norm": 0.03251833841204643, "learning_rate": 3.3523741770776506e-06, "loss": 0.0005, "step": 157970 }, { "epoch": 2.5849627750961304, "grad_norm": 0.004305060487240553, "learning_rate": 3.3517001175390483e-06, "loss": 0.0005, "step": 157980 }, { "epoch": 2.585126401047206, "grad_norm": 0.05417122691869736, "learning_rate": 3.3510260916078775e-06, "loss": 0.0005, "step": 157990 }, { "epoch": 2.585290026998282, "grad_norm": 0.034567009657621384, "learning_rate": 3.3503520992978834e-06, "loss": 0.0007, "step": 158000 }, { "epoch": 2.585453652949358, "grad_norm": 0.03673499822616577, "learning_rate": 3.3496781406228053e-06, "loss": 0.0009, "step": 158010 }, { "epoch": 2.5856172789004335, "grad_norm": 0.024749474599957466, "learning_rate": 3.3490042155963876e-06, "loss": 0.0008, "step": 158020 }, { "epoch": 2.5857809048515095, "grad_norm": 0.0751829743385315, "learning_rate": 3.3483303242323696e-06, "loss": 0.0012, "step": 158030 }, { "epoch": 2.5859445308025855, "grad_norm": 0.04896647483110428, "learning_rate": 3.3476564665444923e-06, "loss": 0.001, "step": 158040 }, { "epoch": 2.586108156753661, "grad_norm": 0.020130015909671783, "learning_rate": 3.3469826425464923e-06, "loss": 0.0006, "step": 158050 }, { "epoch": 2.586271782704737, "grad_norm": 0.007208617404103279, "learning_rate": 3.3463088522521126e-06, "loss": 0.0021, "step": 158060 }, { "epoch": 2.586435408655813, "grad_norm": 0.11060445010662079, "learning_rate": 3.345635095675087e-06, "loss": 0.0012, "step": 158070 }, { "epoch": 2.5865990346068886, "grad_norm": 0.021864071488380432, "learning_rate": 3.344961372829157e-06, "loss": 0.0004, "step": 158080 }, { "epoch": 2.5867626605579646, "grad_norm": 0.002735414542257786, "learning_rate": 3.344287683728055e-06, "loss": 0.0007, "step": 158090 }, { "epoch": 2.5869262865090406, "grad_norm": 0.040171489119529724, "learning_rate": 3.3436140283855212e-06, "loss": 0.0008, "step": 158100 }, { "epoch": 2.587089912460116, "grad_norm": 0.06894267350435257, "learning_rate": 3.3429404068152866e-06, "loss": 0.0006, "step": 158110 }, { "epoch": 2.587253538411192, "grad_norm": 0.014122482389211655, "learning_rate": 3.3422668190310893e-06, "loss": 0.0007, "step": 158120 }, { "epoch": 2.587417164362268, "grad_norm": 0.10956891626119614, "learning_rate": 3.34159326504666e-06, "loss": 0.0003, "step": 158130 }, { "epoch": 2.5875807903133436, "grad_norm": 0.009554039686918259, "learning_rate": 3.3409197448757356e-06, "loss": 0.001, "step": 158140 }, { "epoch": 2.5877444162644196, "grad_norm": 0.01353180967271328, "learning_rate": 3.3402462585320448e-06, "loss": 0.0004, "step": 158150 }, { "epoch": 2.5879080422154956, "grad_norm": 0.03690747171640396, "learning_rate": 3.3395728060293226e-06, "loss": 0.0007, "step": 158160 }, { "epoch": 2.588071668166571, "grad_norm": 0.0033409239258617163, "learning_rate": 3.338899387381298e-06, "loss": 0.0007, "step": 158170 }, { "epoch": 2.588235294117647, "grad_norm": 0.040539976209402084, "learning_rate": 3.3382260026017027e-06, "loss": 0.0006, "step": 158180 }, { "epoch": 2.5883989200687227, "grad_norm": 0.090871162712574, "learning_rate": 3.337552651704265e-06, "loss": 0.0008, "step": 158190 }, { "epoch": 2.5885625460197987, "grad_norm": 0.021905791014432907, "learning_rate": 3.3368793347027163e-06, "loss": 0.0005, "step": 158200 }, { "epoch": 2.5887261719708747, "grad_norm": 0.008966230787336826, "learning_rate": 3.336206051610782e-06, "loss": 0.0007, "step": 158210 }, { "epoch": 2.5888897979219503, "grad_norm": 0.11169939488172531, "learning_rate": 3.3355328024421926e-06, "loss": 0.0007, "step": 158220 }, { "epoch": 2.5890534238730263, "grad_norm": 0.001746173482388258, "learning_rate": 3.3348595872106724e-06, "loss": 0.0005, "step": 158230 }, { "epoch": 2.589217049824102, "grad_norm": 0.016212668269872665, "learning_rate": 3.33418640592995e-06, "loss": 0.0008, "step": 158240 }, { "epoch": 2.589380675775178, "grad_norm": 0.04115690290927887, "learning_rate": 3.33351325861375e-06, "loss": 0.0006, "step": 158250 }, { "epoch": 2.589544301726254, "grad_norm": 0.02514679543673992, "learning_rate": 3.3328401452757985e-06, "loss": 0.0005, "step": 158260 }, { "epoch": 2.5897079276773294, "grad_norm": 0.005748697556555271, "learning_rate": 3.3321670659298155e-06, "loss": 0.0008, "step": 158270 }, { "epoch": 2.5898715536284054, "grad_norm": 0.04558944329619408, "learning_rate": 3.3314940205895312e-06, "loss": 0.0009, "step": 158280 }, { "epoch": 2.5900351795794814, "grad_norm": 0.03039042465388775, "learning_rate": 3.3308210092686617e-06, "loss": 0.0007, "step": 158290 }, { "epoch": 2.590198805530557, "grad_norm": 0.05049572512507439, "learning_rate": 3.3301480319809353e-06, "loss": 0.0005, "step": 158300 }, { "epoch": 2.590362431481633, "grad_norm": 0.01520779263228178, "learning_rate": 3.3294750887400666e-06, "loss": 0.0008, "step": 158310 }, { "epoch": 2.590526057432709, "grad_norm": 0.04658330976963043, "learning_rate": 3.3288021795597835e-06, "loss": 0.0007, "step": 158320 }, { "epoch": 2.5906896833837845, "grad_norm": 0.03002905659377575, "learning_rate": 3.3281293044538e-06, "loss": 0.0005, "step": 158330 }, { "epoch": 2.5908533093348605, "grad_norm": 0.04292971268296242, "learning_rate": 3.327456463435841e-06, "loss": 0.0004, "step": 158340 }, { "epoch": 2.5910169352859365, "grad_norm": 0.06650032848119736, "learning_rate": 3.3267836565196188e-06, "loss": 0.001, "step": 158350 }, { "epoch": 2.591180561237012, "grad_norm": 0.0503162257373333, "learning_rate": 3.3261108837188582e-06, "loss": 0.0005, "step": 158360 }, { "epoch": 2.591344187188088, "grad_norm": 0.030095232650637627, "learning_rate": 3.3254381450472705e-06, "loss": 0.0009, "step": 158370 }, { "epoch": 2.591507813139164, "grad_norm": 0.11239448189735413, "learning_rate": 3.3247654405185776e-06, "loss": 0.0005, "step": 158380 }, { "epoch": 2.5916714390902396, "grad_norm": 0.01635642722249031, "learning_rate": 3.324092770146489e-06, "loss": 0.001, "step": 158390 }, { "epoch": 2.5918350650413156, "grad_norm": 0.0010538483038544655, "learning_rate": 3.323420133944727e-06, "loss": 0.0006, "step": 158400 }, { "epoch": 2.5919986909923916, "grad_norm": 0.042283762246370316, "learning_rate": 3.3227475319269998e-06, "loss": 0.0013, "step": 158410 }, { "epoch": 2.592162316943467, "grad_norm": 0.00366845796816051, "learning_rate": 3.322074964107027e-06, "loss": 0.0004, "step": 158420 }, { "epoch": 2.592325942894543, "grad_norm": 0.05091628432273865, "learning_rate": 3.3214024304985147e-06, "loss": 0.0007, "step": 158430 }, { "epoch": 2.592489568845619, "grad_norm": 0.01642388291656971, "learning_rate": 3.320729931115183e-06, "loss": 0.0004, "step": 158440 }, { "epoch": 2.5926531947966946, "grad_norm": 0.010148738510906696, "learning_rate": 3.3200574659707387e-06, "loss": 0.002, "step": 158450 }, { "epoch": 2.5928168207477706, "grad_norm": 0.029890624806284904, "learning_rate": 3.3193850350788935e-06, "loss": 0.001, "step": 158460 }, { "epoch": 2.5929804466988466, "grad_norm": 0.009435671381652355, "learning_rate": 3.3187126384533574e-06, "loss": 0.0006, "step": 158470 }, { "epoch": 2.593144072649922, "grad_norm": 0.08378943055868149, "learning_rate": 3.3180402761078424e-06, "loss": 0.0012, "step": 158480 }, { "epoch": 2.593307698600998, "grad_norm": 0.011454147286713123, "learning_rate": 3.317367948056054e-06, "loss": 0.0004, "step": 158490 }, { "epoch": 2.593471324552074, "grad_norm": 0.013145825825631618, "learning_rate": 3.3166956543117035e-06, "loss": 0.001, "step": 158500 }, { "epoch": 2.5936349505031497, "grad_norm": 0.06793717294931412, "learning_rate": 3.3160233948884957e-06, "loss": 0.0006, "step": 158510 }, { "epoch": 2.5937985764542257, "grad_norm": 0.010318613611161709, "learning_rate": 3.31535116980014e-06, "loss": 0.0004, "step": 158520 }, { "epoch": 2.5939622024053017, "grad_norm": 0.017078276723623276, "learning_rate": 3.31467897906034e-06, "loss": 0.0004, "step": 158530 }, { "epoch": 2.5941258283563773, "grad_norm": 0.03504364192485809, "learning_rate": 3.3140068226828038e-06, "loss": 0.0009, "step": 158540 }, { "epoch": 2.5942894543074533, "grad_norm": 0.022637037560343742, "learning_rate": 3.313334700681233e-06, "loss": 0.0006, "step": 158550 }, { "epoch": 2.5944530802585293, "grad_norm": 0.08565641939640045, "learning_rate": 3.312662613069334e-06, "loss": 0.0008, "step": 158560 }, { "epoch": 2.594616706209605, "grad_norm": 0.03724709153175354, "learning_rate": 3.3119905598608086e-06, "loss": 0.0009, "step": 158570 }, { "epoch": 2.594780332160681, "grad_norm": 0.04556117579340935, "learning_rate": 3.3113185410693616e-06, "loss": 0.0005, "step": 158580 }, { "epoch": 2.5949439581117564, "grad_norm": 0.12191247940063477, "learning_rate": 3.310646556708692e-06, "loss": 0.0006, "step": 158590 }, { "epoch": 2.5951075840628324, "grad_norm": 0.08643627911806107, "learning_rate": 3.309974606792503e-06, "loss": 0.0011, "step": 158600 }, { "epoch": 2.5952712100139084, "grad_norm": 0.0017720026662573218, "learning_rate": 3.309302691334494e-06, "loss": 0.0008, "step": 158610 }, { "epoch": 2.595434835964984, "grad_norm": 0.04225931689143181, "learning_rate": 3.3086308103483657e-06, "loss": 0.0008, "step": 158620 }, { "epoch": 2.59559846191606, "grad_norm": 0.20950381457805634, "learning_rate": 3.307958963847816e-06, "loss": 0.0007, "step": 158630 }, { "epoch": 2.5957620878671355, "grad_norm": 0.017190830782055855, "learning_rate": 3.3072871518465456e-06, "loss": 0.0004, "step": 158640 }, { "epoch": 2.5959257138182115, "grad_norm": 0.004670754540711641, "learning_rate": 3.3066153743582487e-06, "loss": 0.0006, "step": 158650 }, { "epoch": 2.5960893397692875, "grad_norm": 0.0171413142234087, "learning_rate": 3.3059436313966253e-06, "loss": 0.0012, "step": 158660 }, { "epoch": 2.596252965720363, "grad_norm": 0.017358683049678802, "learning_rate": 3.3052719229753694e-06, "loss": 0.0008, "step": 158670 }, { "epoch": 2.596416591671439, "grad_norm": 0.2083916962146759, "learning_rate": 3.3046002491081793e-06, "loss": 0.0018, "step": 158680 }, { "epoch": 2.596580217622515, "grad_norm": 0.06054559722542763, "learning_rate": 3.3039286098087464e-06, "loss": 0.0011, "step": 158690 }, { "epoch": 2.5967438435735906, "grad_norm": 0.05324682220816612, "learning_rate": 3.3032570050907685e-06, "loss": 0.0007, "step": 158700 }, { "epoch": 2.5969074695246666, "grad_norm": 0.04190618172287941, "learning_rate": 3.3025854349679353e-06, "loss": 0.0009, "step": 158710 }, { "epoch": 2.5970710954757426, "grad_norm": 0.027353649958968163, "learning_rate": 3.3019138994539434e-06, "loss": 0.0012, "step": 158720 }, { "epoch": 2.597234721426818, "grad_norm": 0.0006467396160587668, "learning_rate": 3.301242398562482e-06, "loss": 0.0013, "step": 158730 }, { "epoch": 2.597398347377894, "grad_norm": 0.05817082151770592, "learning_rate": 3.3005709323072436e-06, "loss": 0.0007, "step": 158740 }, { "epoch": 2.59756197332897, "grad_norm": 0.05208653211593628, "learning_rate": 3.299899500701918e-06, "loss": 0.0014, "step": 158750 }, { "epoch": 2.5977255992800456, "grad_norm": 0.02012687362730503, "learning_rate": 3.299228103760197e-06, "loss": 0.0005, "step": 158760 }, { "epoch": 2.5978892252311216, "grad_norm": 0.016033578664064407, "learning_rate": 3.2985567414957664e-06, "loss": 0.0004, "step": 158770 }, { "epoch": 2.5980528511821976, "grad_norm": 0.0023721239995211363, "learning_rate": 3.2978854139223186e-06, "loss": 0.0007, "step": 158780 }, { "epoch": 2.598216477133273, "grad_norm": 0.11192747950553894, "learning_rate": 3.297214121053539e-06, "loss": 0.0015, "step": 158790 }, { "epoch": 2.598380103084349, "grad_norm": 0.04336085543036461, "learning_rate": 3.296542862903116e-06, "loss": 0.0008, "step": 158800 }, { "epoch": 2.598543729035425, "grad_norm": 0.052565183490514755, "learning_rate": 3.2958716394847335e-06, "loss": 0.0005, "step": 158810 }, { "epoch": 2.5987073549865007, "grad_norm": 0.033791735768318176, "learning_rate": 3.2952004508120805e-06, "loss": 0.0008, "step": 158820 }, { "epoch": 2.5988709809375767, "grad_norm": 0.045484788715839386, "learning_rate": 3.2945292968988395e-06, "loss": 0.0008, "step": 158830 }, { "epoch": 2.5990346068886527, "grad_norm": 0.058269303292036057, "learning_rate": 3.2938581777586977e-06, "loss": 0.0004, "step": 158840 }, { "epoch": 2.5991982328397283, "grad_norm": 0.2502894997596741, "learning_rate": 3.293187093405334e-06, "loss": 0.001, "step": 158850 }, { "epoch": 2.5993618587908043, "grad_norm": 0.04001661390066147, "learning_rate": 3.2925160438524354e-06, "loss": 0.0007, "step": 158860 }, { "epoch": 2.5995254847418803, "grad_norm": 0.02960744872689247, "learning_rate": 3.2918450291136817e-06, "loss": 0.0014, "step": 158870 }, { "epoch": 2.599689110692956, "grad_norm": 0.09623171389102936, "learning_rate": 3.291174049202756e-06, "loss": 0.0007, "step": 158880 }, { "epoch": 2.599852736644032, "grad_norm": 0.14870291948318481, "learning_rate": 3.2905031041333373e-06, "loss": 0.0018, "step": 158890 }, { "epoch": 2.600016362595108, "grad_norm": 0.07081843912601471, "learning_rate": 3.289832193919108e-06, "loss": 0.0015, "step": 158900 }, { "epoch": 2.6001799885461834, "grad_norm": 0.028670504689216614, "learning_rate": 3.2891613185737436e-06, "loss": 0.0006, "step": 158910 }, { "epoch": 2.6003436144972594, "grad_norm": 0.07836366444826126, "learning_rate": 3.2884904781109273e-06, "loss": 0.0006, "step": 158920 }, { "epoch": 2.6005072404483354, "grad_norm": 0.044097255915403366, "learning_rate": 3.2878196725443316e-06, "loss": 0.0009, "step": 158930 }, { "epoch": 2.600670866399411, "grad_norm": 0.017262525856494904, "learning_rate": 3.2871489018876402e-06, "loss": 0.0008, "step": 158940 }, { "epoch": 2.600834492350487, "grad_norm": 0.04027764871716499, "learning_rate": 3.286478166154522e-06, "loss": 0.0011, "step": 158950 }, { "epoch": 2.6009981183015625, "grad_norm": 0.04041497781872749, "learning_rate": 3.2858074653586602e-06, "loss": 0.0004, "step": 158960 }, { "epoch": 2.6011617442526385, "grad_norm": 0.04245847091078758, "learning_rate": 3.2851367995137232e-06, "loss": 0.0009, "step": 158970 }, { "epoch": 2.6013253702037145, "grad_norm": 0.025458505377173424, "learning_rate": 3.284466168633392e-06, "loss": 0.0009, "step": 158980 }, { "epoch": 2.60148899615479, "grad_norm": 0.03631545975804329, "learning_rate": 3.283795572731333e-06, "loss": 0.0007, "step": 158990 }, { "epoch": 2.601652622105866, "grad_norm": 0.06170585751533508, "learning_rate": 3.2831250118212253e-06, "loss": 0.0009, "step": 159000 }, { "epoch": 2.6018162480569416, "grad_norm": 0.05235602334141731, "learning_rate": 3.282454485916736e-06, "loss": 0.0004, "step": 159010 }, { "epoch": 2.6019798740080176, "grad_norm": 0.06465861201286316, "learning_rate": 3.2817839950315424e-06, "loss": 0.0011, "step": 159020 }, { "epoch": 2.6021434999590936, "grad_norm": 0.018004612997174263, "learning_rate": 3.281113539179308e-06, "loss": 0.0005, "step": 159030 }, { "epoch": 2.602307125910169, "grad_norm": 0.043685268610715866, "learning_rate": 3.280443118373711e-06, "loss": 0.0007, "step": 159040 }, { "epoch": 2.602470751861245, "grad_norm": 0.04537879675626755, "learning_rate": 3.2797727326284117e-06, "loss": 0.0005, "step": 159050 }, { "epoch": 2.602634377812321, "grad_norm": 0.02321416139602661, "learning_rate": 3.2791023819570877e-06, "loss": 0.0004, "step": 159060 }, { "epoch": 2.6027980037633967, "grad_norm": 0.03240226209163666, "learning_rate": 3.2784320663733987e-06, "loss": 0.0014, "step": 159070 }, { "epoch": 2.6029616297144726, "grad_norm": 0.038664981722831726, "learning_rate": 3.2777617858910193e-06, "loss": 0.0005, "step": 159080 }, { "epoch": 2.6031252556655486, "grad_norm": 0.011113082990050316, "learning_rate": 3.2770915405236097e-06, "loss": 0.0008, "step": 159090 }, { "epoch": 2.603288881616624, "grad_norm": 0.21340051293373108, "learning_rate": 3.2764213302848404e-06, "loss": 0.0017, "step": 159100 }, { "epoch": 2.6034525075677, "grad_norm": 0.020566146820783615, "learning_rate": 3.2757511551883725e-06, "loss": 0.0009, "step": 159110 }, { "epoch": 2.603616133518776, "grad_norm": 0.06132696196436882, "learning_rate": 3.275081015247873e-06, "loss": 0.0012, "step": 159120 }, { "epoch": 2.6037797594698517, "grad_norm": 0.05943078175187111, "learning_rate": 3.274410910477004e-06, "loss": 0.0007, "step": 159130 }, { "epoch": 2.6039433854209277, "grad_norm": 0.02723180688917637, "learning_rate": 3.27374084088943e-06, "loss": 0.0006, "step": 159140 }, { "epoch": 2.6041070113720037, "grad_norm": 0.048158228397369385, "learning_rate": 3.2730708064988103e-06, "loss": 0.0012, "step": 159150 }, { "epoch": 2.6042706373230793, "grad_norm": 0.0657830759882927, "learning_rate": 3.2724008073188104e-06, "loss": 0.0009, "step": 159160 }, { "epoch": 2.6044342632741553, "grad_norm": 0.05329178273677826, "learning_rate": 3.2717308433630867e-06, "loss": 0.0007, "step": 159170 }, { "epoch": 2.6045978892252313, "grad_norm": 0.05319748818874359, "learning_rate": 3.2710609146453034e-06, "loss": 0.0008, "step": 159180 }, { "epoch": 2.604761515176307, "grad_norm": 0.08950410783290863, "learning_rate": 3.2703910211791157e-06, "loss": 0.0009, "step": 159190 }, { "epoch": 2.604925141127383, "grad_norm": 0.033564187586307526, "learning_rate": 3.2697211629781855e-06, "loss": 0.0005, "step": 159200 }, { "epoch": 2.605088767078459, "grad_norm": 0.02382448874413967, "learning_rate": 3.269051340056168e-06, "loss": 0.0006, "step": 159210 }, { "epoch": 2.6052523930295344, "grad_norm": 0.05594726651906967, "learning_rate": 3.268381552426723e-06, "loss": 0.001, "step": 159220 }, { "epoch": 2.6054160189806104, "grad_norm": 0.0026928316801786423, "learning_rate": 3.267711800103504e-06, "loss": 0.0007, "step": 159230 }, { "epoch": 2.6055796449316864, "grad_norm": 0.013293284922838211, "learning_rate": 3.2670420831001694e-06, "loss": 0.0005, "step": 159240 }, { "epoch": 2.605743270882762, "grad_norm": 0.08098294585943222, "learning_rate": 3.2663724014303714e-06, "loss": 0.0012, "step": 159250 }, { "epoch": 2.605906896833838, "grad_norm": 0.05017466843128204, "learning_rate": 3.2657027551077674e-06, "loss": 0.0006, "step": 159260 }, { "epoch": 2.606070522784914, "grad_norm": 0.07806222140789032, "learning_rate": 3.265033144146008e-06, "loss": 0.0006, "step": 159270 }, { "epoch": 2.6062341487359895, "grad_norm": 0.08918303996324539, "learning_rate": 3.2643635685587483e-06, "loss": 0.0011, "step": 159280 }, { "epoch": 2.6063977746870655, "grad_norm": 0.12952923774719238, "learning_rate": 3.2636940283596385e-06, "loss": 0.0011, "step": 159290 }, { "epoch": 2.6065614006381415, "grad_norm": 0.059123143553733826, "learning_rate": 3.2630245235623313e-06, "loss": 0.0006, "step": 159300 }, { "epoch": 2.606725026589217, "grad_norm": 0.025736885145306587, "learning_rate": 3.262355054180476e-06, "loss": 0.0006, "step": 159310 }, { "epoch": 2.606888652540293, "grad_norm": 0.045846614986658096, "learning_rate": 3.2616856202277248e-06, "loss": 0.0011, "step": 159320 }, { "epoch": 2.607052278491369, "grad_norm": 0.026132941246032715, "learning_rate": 3.2610162217177237e-06, "loss": 0.0009, "step": 159330 }, { "epoch": 2.6072159044424446, "grad_norm": 0.04714734107255936, "learning_rate": 3.260346858664124e-06, "loss": 0.0008, "step": 159340 }, { "epoch": 2.6073795303935206, "grad_norm": 0.015812654048204422, "learning_rate": 3.2596775310805717e-06, "loss": 0.0009, "step": 159350 }, { "epoch": 2.607543156344596, "grad_norm": 0.026324061676859856, "learning_rate": 3.2590082389807156e-06, "loss": 0.0005, "step": 159360 }, { "epoch": 2.607706782295672, "grad_norm": 0.0032764447387307882, "learning_rate": 3.2583389823781997e-06, "loss": 0.0006, "step": 159370 }, { "epoch": 2.607870408246748, "grad_norm": 0.0018459698185324669, "learning_rate": 3.2576697612866716e-06, "loss": 0.0005, "step": 159380 }, { "epoch": 2.6080340341978236, "grad_norm": 0.04055985435843468, "learning_rate": 3.2570005757197743e-06, "loss": 0.0009, "step": 159390 }, { "epoch": 2.6081976601488996, "grad_norm": 0.13122054934501648, "learning_rate": 3.256331425691154e-06, "loss": 0.001, "step": 159400 }, { "epoch": 2.608361286099975, "grad_norm": 0.0429835245013237, "learning_rate": 3.2556623112144516e-06, "loss": 0.0017, "step": 159410 }, { "epoch": 2.608524912051051, "grad_norm": 0.05811120569705963, "learning_rate": 3.2549932323033117e-06, "loss": 0.0007, "step": 159420 }, { "epoch": 2.608688538002127, "grad_norm": 0.03376048058271408, "learning_rate": 3.254324188971376e-06, "loss": 0.0008, "step": 159430 }, { "epoch": 2.6088521639532027, "grad_norm": 0.12987364828586578, "learning_rate": 3.253655181232286e-06, "loss": 0.001, "step": 159440 }, { "epoch": 2.6090157899042787, "grad_norm": 0.03444516658782959, "learning_rate": 3.25298620909968e-06, "loss": 0.0004, "step": 159450 }, { "epoch": 2.6091794158553547, "grad_norm": 0.02753458172082901, "learning_rate": 3.252317272587201e-06, "loss": 0.0006, "step": 159460 }, { "epoch": 2.6093430418064303, "grad_norm": 0.013268496841192245, "learning_rate": 3.2516483717084856e-06, "loss": 0.0006, "step": 159470 }, { "epoch": 2.6095066677575063, "grad_norm": 0.04730521887540817, "learning_rate": 3.2509795064771738e-06, "loss": 0.0006, "step": 159480 }, { "epoch": 2.6096702937085823, "grad_norm": 0.07778061181306839, "learning_rate": 3.2503106769069015e-06, "loss": 0.001, "step": 159490 }, { "epoch": 2.609833919659658, "grad_norm": 0.2156205028295517, "learning_rate": 3.249641883011307e-06, "loss": 0.001, "step": 159500 }, { "epoch": 2.609997545610734, "grad_norm": 0.018683303147554398, "learning_rate": 3.2489731248040253e-06, "loss": 0.0012, "step": 159510 }, { "epoch": 2.61016117156181, "grad_norm": 0.04085203632712364, "learning_rate": 3.248304402298693e-06, "loss": 0.0008, "step": 159520 }, { "epoch": 2.6103247975128854, "grad_norm": 0.00485370634123683, "learning_rate": 3.2476357155089432e-06, "loss": 0.0009, "step": 159530 }, { "epoch": 2.6104884234639614, "grad_norm": 0.06988374143838882, "learning_rate": 3.246967064448413e-06, "loss": 0.0012, "step": 159540 }, { "epoch": 2.6106520494150374, "grad_norm": 0.023071885108947754, "learning_rate": 3.246298449130731e-06, "loss": 0.0005, "step": 159550 }, { "epoch": 2.610815675366113, "grad_norm": 0.03270960599184036, "learning_rate": 3.2456298695695345e-06, "loss": 0.0007, "step": 159560 }, { "epoch": 2.610979301317189, "grad_norm": 0.0489141047000885, "learning_rate": 3.24496132577845e-06, "loss": 0.0005, "step": 159570 }, { "epoch": 2.611142927268265, "grad_norm": 0.03964439406991005, "learning_rate": 3.2442928177711148e-06, "loss": 0.0003, "step": 159580 }, { "epoch": 2.6113065532193405, "grad_norm": 0.012823655270040035, "learning_rate": 3.243624345561153e-06, "loss": 0.0006, "step": 159590 }, { "epoch": 2.6114701791704165, "grad_norm": 0.03553992137312889, "learning_rate": 3.2429559091621985e-06, "loss": 0.0006, "step": 159600 }, { "epoch": 2.6116338051214925, "grad_norm": 0.03114445134997368, "learning_rate": 3.2422875085878776e-06, "loss": 0.001, "step": 159610 }, { "epoch": 2.611797431072568, "grad_norm": 0.011454163119196892, "learning_rate": 3.2416191438518218e-06, "loss": 0.0009, "step": 159620 }, { "epoch": 2.611961057023644, "grad_norm": 0.029556451365351677, "learning_rate": 3.2409508149676526e-06, "loss": 0.0008, "step": 159630 }, { "epoch": 2.61212468297472, "grad_norm": 0.007619100157171488, "learning_rate": 3.240282521949004e-06, "loss": 0.0004, "step": 159640 }, { "epoch": 2.6122883089257956, "grad_norm": 0.02099202200770378, "learning_rate": 3.2396142648094954e-06, "loss": 0.0004, "step": 159650 }, { "epoch": 2.6124519348768716, "grad_norm": 0.009106948971748352, "learning_rate": 3.238946043562757e-06, "loss": 0.0008, "step": 159660 }, { "epoch": 2.6126155608279475, "grad_norm": 0.01915798708796501, "learning_rate": 3.2382778582224083e-06, "loss": 0.0006, "step": 159670 }, { "epoch": 2.612779186779023, "grad_norm": 0.01221378706395626, "learning_rate": 3.23760970880208e-06, "loss": 0.0008, "step": 159680 }, { "epoch": 2.612942812730099, "grad_norm": 0.035541608929634094, "learning_rate": 3.236941595315386e-06, "loss": 0.001, "step": 159690 }, { "epoch": 2.613106438681175, "grad_norm": 0.017594076693058014, "learning_rate": 3.2362735177759574e-06, "loss": 0.0006, "step": 159700 }, { "epoch": 2.6132700646322506, "grad_norm": 0.03427041321992874, "learning_rate": 3.2356054761974086e-06, "loss": 0.0003, "step": 159710 }, { "epoch": 2.6134336905833266, "grad_norm": 0.01025879941880703, "learning_rate": 3.234937470593367e-06, "loss": 0.0015, "step": 159720 }, { "epoch": 2.613597316534402, "grad_norm": 0.05187169089913368, "learning_rate": 3.234269500977447e-06, "loss": 0.0008, "step": 159730 }, { "epoch": 2.613760942485478, "grad_norm": 0.021063068881630898, "learning_rate": 3.2336015673632705e-06, "loss": 0.0004, "step": 159740 }, { "epoch": 2.613924568436554, "grad_norm": 0.033427488058805466, "learning_rate": 3.2329336697644553e-06, "loss": 0.0004, "step": 159750 }, { "epoch": 2.6140881943876297, "grad_norm": 0.036667946726083755, "learning_rate": 3.232265808194621e-06, "loss": 0.0007, "step": 159760 }, { "epoch": 2.6142518203387057, "grad_norm": 0.05451033264398575, "learning_rate": 3.231597982667382e-06, "loss": 0.0006, "step": 159770 }, { "epoch": 2.6144154462897813, "grad_norm": 0.0116246547549963, "learning_rate": 3.230930193196358e-06, "loss": 0.0005, "step": 159780 }, { "epoch": 2.6145790722408573, "grad_norm": 0.032031115144491196, "learning_rate": 3.230262439795161e-06, "loss": 0.0003, "step": 159790 }, { "epoch": 2.6147426981919333, "grad_norm": 0.05849166214466095, "learning_rate": 3.2295947224774086e-06, "loss": 0.0006, "step": 159800 }, { "epoch": 2.614906324143009, "grad_norm": 0.05267273262143135, "learning_rate": 3.2289270412567135e-06, "loss": 0.0017, "step": 159810 }, { "epoch": 2.615069950094085, "grad_norm": 0.0419389009475708, "learning_rate": 3.2282593961466915e-06, "loss": 0.0007, "step": 159820 }, { "epoch": 2.615233576045161, "grad_norm": 0.12691357731819153, "learning_rate": 3.2275917871609513e-06, "loss": 0.0017, "step": 159830 }, { "epoch": 2.6153972019962364, "grad_norm": 0.03369273245334625, "learning_rate": 3.2269242143131086e-06, "loss": 0.001, "step": 159840 }, { "epoch": 2.6155608279473124, "grad_norm": 0.06143258884549141, "learning_rate": 3.2262566776167723e-06, "loss": 0.0006, "step": 159850 }, { "epoch": 2.6157244538983884, "grad_norm": 0.08750570565462112, "learning_rate": 3.225589177085555e-06, "loss": 0.001, "step": 159860 }, { "epoch": 2.615888079849464, "grad_norm": 0.061509598046541214, "learning_rate": 3.2249217127330645e-06, "loss": 0.0011, "step": 159870 }, { "epoch": 2.61605170580054, "grad_norm": 0.06710506975650787, "learning_rate": 3.2242542845729106e-06, "loss": 0.0007, "step": 159880 }, { "epoch": 2.616215331751616, "grad_norm": 0.1204911544919014, "learning_rate": 3.223586892618702e-06, "loss": 0.0008, "step": 159890 }, { "epoch": 2.6163789577026915, "grad_norm": 0.012799963355064392, "learning_rate": 3.222919536884046e-06, "loss": 0.0003, "step": 159900 }, { "epoch": 2.6165425836537675, "grad_norm": 0.15202954411506653, "learning_rate": 3.2222522173825487e-06, "loss": 0.0009, "step": 159910 }, { "epoch": 2.6167062096048435, "grad_norm": 0.0019484222866594791, "learning_rate": 3.2215849341278172e-06, "loss": 0.0005, "step": 159920 }, { "epoch": 2.616869835555919, "grad_norm": 0.059933170676231384, "learning_rate": 3.220917687133456e-06, "loss": 0.0009, "step": 159930 }, { "epoch": 2.617033461506995, "grad_norm": 0.010631835088133812, "learning_rate": 3.220250476413071e-06, "loss": 0.0009, "step": 159940 }, { "epoch": 2.617197087458071, "grad_norm": 0.05163559690117836, "learning_rate": 3.2195833019802635e-06, "loss": 0.0008, "step": 159950 }, { "epoch": 2.6173607134091466, "grad_norm": 0.047219641506671906, "learning_rate": 3.21891616384864e-06, "loss": 0.0007, "step": 159960 }, { "epoch": 2.6175243393602226, "grad_norm": 0.010709602385759354, "learning_rate": 3.2182490620318004e-06, "loss": 0.0005, "step": 159970 }, { "epoch": 2.6176879653112985, "grad_norm": 0.21032652258872986, "learning_rate": 3.2175819965433473e-06, "loss": 0.002, "step": 159980 }, { "epoch": 2.617851591262374, "grad_norm": 0.0015176511369645596, "learning_rate": 3.2169149673968814e-06, "loss": 0.0007, "step": 159990 }, { "epoch": 2.61801521721345, "grad_norm": 0.05610127002000809, "learning_rate": 3.2162479746060025e-06, "loss": 0.0005, "step": 160000 }, { "epoch": 2.618178843164526, "grad_norm": 0.0321102999150753, "learning_rate": 3.2155810181843107e-06, "loss": 0.0009, "step": 160010 }, { "epoch": 2.6183424691156016, "grad_norm": 0.012910072691738605, "learning_rate": 3.2149140981454047e-06, "loss": 0.0009, "step": 160020 }, { "epoch": 2.6185060950666776, "grad_norm": 0.027064254507422447, "learning_rate": 3.2142472145028814e-06, "loss": 0.0009, "step": 160030 }, { "epoch": 2.6186697210177536, "grad_norm": 0.06898824125528336, "learning_rate": 3.213580367270339e-06, "loss": 0.0006, "step": 160040 }, { "epoch": 2.618833346968829, "grad_norm": 0.021354541182518005, "learning_rate": 3.2129135564613736e-06, "loss": 0.0007, "step": 160050 }, { "epoch": 2.618996972919905, "grad_norm": 0.01787680946290493, "learning_rate": 3.212246782089581e-06, "loss": 0.0007, "step": 160060 }, { "epoch": 2.619160598870981, "grad_norm": 0.03034539893269539, "learning_rate": 3.2115800441685552e-06, "loss": 0.0007, "step": 160070 }, { "epoch": 2.6193242248220567, "grad_norm": 0.03925119340419769, "learning_rate": 3.2109133427118926e-06, "loss": 0.001, "step": 160080 }, { "epoch": 2.6194878507731327, "grad_norm": 0.045867402106523514, "learning_rate": 3.210246677733184e-06, "loss": 0.0015, "step": 160090 }, { "epoch": 2.6196514767242087, "grad_norm": 0.05948508903384209, "learning_rate": 3.209580049246025e-06, "loss": 0.0006, "step": 160100 }, { "epoch": 2.6198151026752843, "grad_norm": 0.010634299367666245, "learning_rate": 3.208913457264005e-06, "loss": 0.0013, "step": 160110 }, { "epoch": 2.6199787286263603, "grad_norm": 0.05298440158367157, "learning_rate": 3.2082469018007174e-06, "loss": 0.003, "step": 160120 }, { "epoch": 2.620142354577436, "grad_norm": 0.03993292897939682, "learning_rate": 3.2075803828697506e-06, "loss": 0.0005, "step": 160130 }, { "epoch": 2.620305980528512, "grad_norm": 0.044701166450977325, "learning_rate": 3.206913900484696e-06, "loss": 0.0009, "step": 160140 }, { "epoch": 2.620469606479588, "grad_norm": 0.025880781933665276, "learning_rate": 3.2062474546591414e-06, "loss": 0.0007, "step": 160150 }, { "epoch": 2.6206332324306634, "grad_norm": 0.018952835351228714, "learning_rate": 3.2055810454066766e-06, "loss": 0.0008, "step": 160160 }, { "epoch": 2.6207968583817394, "grad_norm": 0.21760664880275726, "learning_rate": 3.204914672740887e-06, "loss": 0.001, "step": 160170 }, { "epoch": 2.620960484332815, "grad_norm": 0.01481202058494091, "learning_rate": 3.204248336675362e-06, "loss": 0.0003, "step": 160180 }, { "epoch": 2.621124110283891, "grad_norm": 0.006752702873200178, "learning_rate": 3.203582037223685e-06, "loss": 0.001, "step": 160190 }, { "epoch": 2.621287736234967, "grad_norm": 0.009521225467324257, "learning_rate": 3.2029157743994443e-06, "loss": 0.0006, "step": 160200 }, { "epoch": 2.6214513621860425, "grad_norm": 0.03103853575885296, "learning_rate": 3.2022495482162197e-06, "loss": 0.001, "step": 160210 }, { "epoch": 2.6216149881371185, "grad_norm": 0.06839359551668167, "learning_rate": 3.2015833586876006e-06, "loss": 0.0004, "step": 160220 }, { "epoch": 2.6217786140881945, "grad_norm": 0.0998929962515831, "learning_rate": 3.2009172058271643e-06, "loss": 0.0007, "step": 160230 }, { "epoch": 2.62194224003927, "grad_norm": 0.05088435858488083, "learning_rate": 3.2002510896484994e-06, "loss": 0.0004, "step": 160240 }, { "epoch": 2.622105865990346, "grad_norm": 0.0076099056750535965, "learning_rate": 3.1995850101651815e-06, "loss": 0.0008, "step": 160250 }, { "epoch": 2.622269491941422, "grad_norm": 0.04472475126385689, "learning_rate": 3.1989189673907963e-06, "loss": 0.0004, "step": 160260 }, { "epoch": 2.6224331178924976, "grad_norm": 0.024237139150500298, "learning_rate": 3.198252961338919e-06, "loss": 0.0006, "step": 160270 }, { "epoch": 2.6225967438435736, "grad_norm": 0.008610056713223457, "learning_rate": 3.1975869920231345e-06, "loss": 0.0005, "step": 160280 }, { "epoch": 2.6227603697946495, "grad_norm": 0.04083909094333649, "learning_rate": 3.1969210594570155e-06, "loss": 0.0004, "step": 160290 }, { "epoch": 2.622923995745725, "grad_norm": 0.06746545433998108, "learning_rate": 3.1962551636541452e-06, "loss": 0.0009, "step": 160300 }, { "epoch": 2.623087621696801, "grad_norm": 0.010425494983792305, "learning_rate": 3.195589304628095e-06, "loss": 0.0012, "step": 160310 }, { "epoch": 2.623251247647877, "grad_norm": 0.011741874739527702, "learning_rate": 3.194923482392448e-06, "loss": 0.0009, "step": 160320 }, { "epoch": 2.6234148735989526, "grad_norm": 0.056505825370550156, "learning_rate": 3.1942576969607724e-06, "loss": 0.0005, "step": 160330 }, { "epoch": 2.6235784995500286, "grad_norm": 0.016128212213516235, "learning_rate": 3.1935919483466497e-06, "loss": 0.0009, "step": 160340 }, { "epoch": 2.6237421255011046, "grad_norm": 0.033316727727651596, "learning_rate": 3.192926236563648e-06, "loss": 0.0009, "step": 160350 }, { "epoch": 2.62390575145218, "grad_norm": 0.05022198706865311, "learning_rate": 3.1922605616253465e-06, "loss": 0.0009, "step": 160360 }, { "epoch": 2.624069377403256, "grad_norm": 0.004524451680481434, "learning_rate": 3.1915949235453126e-06, "loss": 0.0004, "step": 160370 }, { "epoch": 2.624233003354332, "grad_norm": 0.03697807341814041, "learning_rate": 3.1909293223371217e-06, "loss": 0.0008, "step": 160380 }, { "epoch": 2.6243966293054077, "grad_norm": 0.06495725363492966, "learning_rate": 3.1902637580143413e-06, "loss": 0.0008, "step": 160390 }, { "epoch": 2.6245602552564837, "grad_norm": 0.02436954528093338, "learning_rate": 3.189598230590546e-06, "loss": 0.0008, "step": 160400 }, { "epoch": 2.6247238812075597, "grad_norm": 0.07762614637613297, "learning_rate": 3.188932740079301e-06, "loss": 0.0021, "step": 160410 }, { "epoch": 2.6248875071586353, "grad_norm": 0.054776161909103394, "learning_rate": 3.1882672864941787e-06, "loss": 0.0007, "step": 160420 }, { "epoch": 2.6250511331097113, "grad_norm": 0.022659486159682274, "learning_rate": 3.187601869848745e-06, "loss": 0.0006, "step": 160430 }, { "epoch": 2.6252147590607873, "grad_norm": 0.014002933166921139, "learning_rate": 3.186936490156568e-06, "loss": 0.0004, "step": 160440 }, { "epoch": 2.625378385011863, "grad_norm": 0.029376698657870293, "learning_rate": 3.1862711474312134e-06, "loss": 0.001, "step": 160450 }, { "epoch": 2.625542010962939, "grad_norm": 0.08382587879896164, "learning_rate": 3.185605841686249e-06, "loss": 0.0007, "step": 160460 }, { "epoch": 2.625705636914015, "grad_norm": 0.061147939413785934, "learning_rate": 3.1849405729352366e-06, "loss": 0.0006, "step": 160470 }, { "epoch": 2.6258692628650904, "grad_norm": 0.073320671916008, "learning_rate": 3.1842753411917437e-06, "loss": 0.0008, "step": 160480 }, { "epoch": 2.6260328888161664, "grad_norm": 0.01945413649082184, "learning_rate": 3.1836101464693313e-06, "loss": 0.0006, "step": 160490 }, { "epoch": 2.626196514767242, "grad_norm": 0.013871122151613235, "learning_rate": 3.1829449887815643e-06, "loss": 0.0011, "step": 160500 }, { "epoch": 2.626360140718318, "grad_norm": 0.04353705421090126, "learning_rate": 3.182279868142003e-06, "loss": 0.001, "step": 160510 }, { "epoch": 2.626523766669394, "grad_norm": 0.03428175300359726, "learning_rate": 3.18161478456421e-06, "loss": 0.0006, "step": 160520 }, { "epoch": 2.6266873926204695, "grad_norm": 0.07913212478160858, "learning_rate": 3.1809497380617437e-06, "loss": 0.0024, "step": 160530 }, { "epoch": 2.6268510185715455, "grad_norm": 0.09397857636213303, "learning_rate": 3.1802847286481665e-06, "loss": 0.0005, "step": 160540 }, { "epoch": 2.627014644522621, "grad_norm": 0.036560866981744766, "learning_rate": 3.1796197563370345e-06, "loss": 0.0006, "step": 160550 }, { "epoch": 2.627178270473697, "grad_norm": 0.09628056734800339, "learning_rate": 3.1789548211419092e-06, "loss": 0.0008, "step": 160560 }, { "epoch": 2.627341896424773, "grad_norm": 0.01131763681769371, "learning_rate": 3.1782899230763453e-06, "loss": 0.0007, "step": 160570 }, { "epoch": 2.6275055223758486, "grad_norm": 0.022868942469358444, "learning_rate": 3.1776250621539007e-06, "loss": 0.0004, "step": 160580 }, { "epoch": 2.6276691483269246, "grad_norm": 0.04644673317670822, "learning_rate": 3.1769602383881315e-06, "loss": 0.0005, "step": 160590 }, { "epoch": 2.6278327742780005, "grad_norm": 0.04148014634847641, "learning_rate": 3.1762954517925925e-06, "loss": 0.0007, "step": 160600 }, { "epoch": 2.627996400229076, "grad_norm": 0.02314424328505993, "learning_rate": 3.1756307023808374e-06, "loss": 0.0006, "step": 160610 }, { "epoch": 2.628160026180152, "grad_norm": 0.02898072451353073, "learning_rate": 3.1749659901664227e-06, "loss": 0.0009, "step": 160620 }, { "epoch": 2.628323652131228, "grad_norm": 0.04388423264026642, "learning_rate": 3.1743013151628976e-06, "loss": 0.0014, "step": 160630 }, { "epoch": 2.6284872780823036, "grad_norm": 0.0813218429684639, "learning_rate": 3.173636677383817e-06, "loss": 0.0018, "step": 160640 }, { "epoch": 2.6286509040333796, "grad_norm": 0.03565898910164833, "learning_rate": 3.1729720768427303e-06, "loss": 0.0011, "step": 160650 }, { "epoch": 2.6288145299844556, "grad_norm": 0.11347807198762894, "learning_rate": 3.17230751355319e-06, "loss": 0.0008, "step": 160660 }, { "epoch": 2.628978155935531, "grad_norm": 0.041541438549757004, "learning_rate": 3.1716429875287442e-06, "loss": 0.0006, "step": 160670 }, { "epoch": 2.629141781886607, "grad_norm": 0.043674446642398834, "learning_rate": 3.170978498782944e-06, "loss": 0.0007, "step": 160680 }, { "epoch": 2.629305407837683, "grad_norm": 0.08199942111968994, "learning_rate": 3.170314047329336e-06, "loss": 0.0012, "step": 160690 }, { "epoch": 2.6294690337887587, "grad_norm": 0.06770364195108414, "learning_rate": 3.1696496331814684e-06, "loss": 0.0008, "step": 160700 }, { "epoch": 2.6296326597398347, "grad_norm": 0.05166662111878395, "learning_rate": 3.168985256352888e-06, "loss": 0.0008, "step": 160710 }, { "epoch": 2.6297962856909107, "grad_norm": 0.053264129906892776, "learning_rate": 3.1683209168571417e-06, "loss": 0.0008, "step": 160720 }, { "epoch": 2.6299599116419863, "grad_norm": 0.04038612172007561, "learning_rate": 3.1676566147077737e-06, "loss": 0.0008, "step": 160730 }, { "epoch": 2.6301235375930623, "grad_norm": 0.04437314718961716, "learning_rate": 3.16699234991833e-06, "loss": 0.0006, "step": 160740 }, { "epoch": 2.6302871635441383, "grad_norm": 0.030925488099455833, "learning_rate": 3.1663281225023513e-06, "loss": 0.001, "step": 160750 }, { "epoch": 2.630450789495214, "grad_norm": 0.007188124116510153, "learning_rate": 3.165663932473384e-06, "loss": 0.0009, "step": 160760 }, { "epoch": 2.63061441544629, "grad_norm": 0.010249827988445759, "learning_rate": 3.1649997798449682e-06, "loss": 0.0006, "step": 160770 }, { "epoch": 2.630778041397366, "grad_norm": 0.02962801419198513, "learning_rate": 3.164335664630648e-06, "loss": 0.0008, "step": 160780 }, { "epoch": 2.6309416673484414, "grad_norm": 0.0212910994887352, "learning_rate": 3.1636715868439603e-06, "loss": 0.0011, "step": 160790 }, { "epoch": 2.6311052932995174, "grad_norm": 0.0881604552268982, "learning_rate": 3.163007546498449e-06, "loss": 0.001, "step": 160800 }, { "epoch": 2.6312689192505934, "grad_norm": 0.015584751963615417, "learning_rate": 3.1623435436076507e-06, "loss": 0.001, "step": 160810 }, { "epoch": 2.631432545201669, "grad_norm": 0.032241757959127426, "learning_rate": 3.161679578185105e-06, "loss": 0.0022, "step": 160820 }, { "epoch": 2.631596171152745, "grad_norm": 0.0015856745885685086, "learning_rate": 3.161015650244349e-06, "loss": 0.0008, "step": 160830 }, { "epoch": 2.631759797103821, "grad_norm": 0.002668787958100438, "learning_rate": 3.160351759798921e-06, "loss": 0.0009, "step": 160840 }, { "epoch": 2.6319234230548965, "grad_norm": 0.030727162957191467, "learning_rate": 3.159687906862353e-06, "loss": 0.001, "step": 160850 }, { "epoch": 2.6320870490059725, "grad_norm": 0.03764529526233673, "learning_rate": 3.159024091448187e-06, "loss": 0.0005, "step": 160860 }, { "epoch": 2.6322506749570485, "grad_norm": 0.02109582908451557, "learning_rate": 3.1583603135699516e-06, "loss": 0.0009, "step": 160870 }, { "epoch": 2.632414300908124, "grad_norm": 0.06354139745235443, "learning_rate": 3.157696573241186e-06, "loss": 0.0007, "step": 160880 }, { "epoch": 2.6325779268592, "grad_norm": 0.034159060567617416, "learning_rate": 3.157032870475416e-06, "loss": 0.0006, "step": 160890 }, { "epoch": 2.6327415528102756, "grad_norm": 0.06862873584032059, "learning_rate": 3.1563692052861827e-06, "loss": 0.001, "step": 160900 }, { "epoch": 2.6329051787613516, "grad_norm": 0.05143457278609276, "learning_rate": 3.1557055776870106e-06, "loss": 0.0005, "step": 160910 }, { "epoch": 2.6330688047124275, "grad_norm": 0.03845258057117462, "learning_rate": 3.155041987691435e-06, "loss": 0.0011, "step": 160920 }, { "epoch": 2.633232430663503, "grad_norm": 0.03474041074514389, "learning_rate": 3.154378435312982e-06, "loss": 0.0007, "step": 160930 }, { "epoch": 2.633396056614579, "grad_norm": 0.01717018149793148, "learning_rate": 3.1537149205651864e-06, "loss": 0.0004, "step": 160940 }, { "epoch": 2.6335596825656546, "grad_norm": 0.026211777701973915, "learning_rate": 3.1530514434615695e-06, "loss": 0.0007, "step": 160950 }, { "epoch": 2.6337233085167306, "grad_norm": 0.05147533491253853, "learning_rate": 3.152388004015666e-06, "loss": 0.0007, "step": 160960 }, { "epoch": 2.6338869344678066, "grad_norm": 0.0020140758715569973, "learning_rate": 3.1517246022409973e-06, "loss": 0.0004, "step": 160970 }, { "epoch": 2.634050560418882, "grad_norm": 0.03732433542609215, "learning_rate": 3.1510612381510954e-06, "loss": 0.0009, "step": 160980 }, { "epoch": 2.634214186369958, "grad_norm": 0.0003527702938299626, "learning_rate": 3.1503979117594786e-06, "loss": 0.0009, "step": 160990 }, { "epoch": 2.634377812321034, "grad_norm": 0.06591641157865524, "learning_rate": 3.1497346230796783e-06, "loss": 0.0004, "step": 161000 }, { "epoch": 2.6345414382721097, "grad_norm": 0.010792613960802555, "learning_rate": 3.1490713721252126e-06, "loss": 0.0013, "step": 161010 }, { "epoch": 2.6347050642231857, "grad_norm": 0.17207880318164825, "learning_rate": 3.1484081589096104e-06, "loss": 0.001, "step": 161020 }, { "epoch": 2.6348686901742617, "grad_norm": 0.10099245607852936, "learning_rate": 3.147744983446389e-06, "loss": 0.0011, "step": 161030 }, { "epoch": 2.6350323161253373, "grad_norm": 0.05708899721503258, "learning_rate": 3.1470818457490736e-06, "loss": 0.0018, "step": 161040 }, { "epoch": 2.6351959420764133, "grad_norm": 0.02989392913877964, "learning_rate": 3.146418745831182e-06, "loss": 0.0004, "step": 161050 }, { "epoch": 2.6353595680274893, "grad_norm": 0.0738377645611763, "learning_rate": 3.1457556837062373e-06, "loss": 0.0006, "step": 161060 }, { "epoch": 2.635523193978565, "grad_norm": 0.043839287012815475, "learning_rate": 3.1450926593877563e-06, "loss": 0.0013, "step": 161070 }, { "epoch": 2.635686819929641, "grad_norm": 0.01100244466215372, "learning_rate": 3.1444296728892588e-06, "loss": 0.0009, "step": 161080 }, { "epoch": 2.635850445880717, "grad_norm": 0.038030125200748444, "learning_rate": 3.143766724224262e-06, "loss": 0.0002, "step": 161090 }, { "epoch": 2.6360140718317924, "grad_norm": 0.00787853542715311, "learning_rate": 3.143103813406283e-06, "loss": 0.0007, "step": 161100 }, { "epoch": 2.6361776977828684, "grad_norm": 0.08342427760362625, "learning_rate": 3.1424409404488377e-06, "loss": 0.0011, "step": 161110 }, { "epoch": 2.6363413237339444, "grad_norm": 0.0574234314262867, "learning_rate": 3.141778105365443e-06, "loss": 0.0005, "step": 161120 }, { "epoch": 2.63650494968502, "grad_norm": 0.07129611074924469, "learning_rate": 3.141115308169611e-06, "loss": 0.0006, "step": 161130 }, { "epoch": 2.636668575636096, "grad_norm": 0.033502981066703796, "learning_rate": 3.1404525488748576e-06, "loss": 0.0005, "step": 161140 }, { "epoch": 2.636832201587172, "grad_norm": 0.026471544057130814, "learning_rate": 3.139789827494695e-06, "loss": 0.0007, "step": 161150 }, { "epoch": 2.6369958275382475, "grad_norm": 0.02532554790377617, "learning_rate": 3.139127144042636e-06, "loss": 0.0007, "step": 161160 }, { "epoch": 2.6371594534893235, "grad_norm": 0.05231475085020065, "learning_rate": 3.138464498532191e-06, "loss": 0.0007, "step": 161170 }, { "epoch": 2.6373230794403995, "grad_norm": 0.04078594222664833, "learning_rate": 3.137801890976873e-06, "loss": 0.0009, "step": 161180 }, { "epoch": 2.637486705391475, "grad_norm": 0.012594337575137615, "learning_rate": 3.13713932139019e-06, "loss": 0.0004, "step": 161190 }, { "epoch": 2.637650331342551, "grad_norm": 0.017115412279963493, "learning_rate": 3.1364767897856522e-06, "loss": 0.0005, "step": 161200 }, { "epoch": 2.637813957293627, "grad_norm": 0.07430422306060791, "learning_rate": 3.135814296176768e-06, "loss": 0.0009, "step": 161210 }, { "epoch": 2.6379775832447026, "grad_norm": 0.05725160241127014, "learning_rate": 3.1351518405770444e-06, "loss": 0.0003, "step": 161220 }, { "epoch": 2.6381412091957785, "grad_norm": 0.06149103119969368, "learning_rate": 3.134489422999989e-06, "loss": 0.0007, "step": 161230 }, { "epoch": 2.6383048351468545, "grad_norm": 0.0053898547776043415, "learning_rate": 3.133827043459108e-06, "loss": 0.0005, "step": 161240 }, { "epoch": 2.63846846109793, "grad_norm": 0.04060272127389908, "learning_rate": 3.1331647019679055e-06, "loss": 0.0007, "step": 161250 }, { "epoch": 2.638632087049006, "grad_norm": 0.07049289345741272, "learning_rate": 3.1325023985398883e-06, "loss": 0.0007, "step": 161260 }, { "epoch": 2.638795713000082, "grad_norm": 0.027296943590044975, "learning_rate": 3.131840133188557e-06, "loss": 0.0005, "step": 161270 }, { "epoch": 2.6389593389511576, "grad_norm": 0.02729487232863903, "learning_rate": 3.131177905927419e-06, "loss": 0.001, "step": 161280 }, { "epoch": 2.6391229649022336, "grad_norm": 0.17071959376335144, "learning_rate": 3.130515716769972e-06, "loss": 0.0017, "step": 161290 }, { "epoch": 2.639286590853309, "grad_norm": 0.035866521298885345, "learning_rate": 3.1298535657297213e-06, "loss": 0.0011, "step": 161300 }, { "epoch": 2.639450216804385, "grad_norm": 0.0777377337217331, "learning_rate": 3.129191452820165e-06, "loss": 0.0004, "step": 161310 }, { "epoch": 2.639613842755461, "grad_norm": 0.028757896274328232, "learning_rate": 3.128529378054804e-06, "loss": 0.0007, "step": 161320 }, { "epoch": 2.6397774687065367, "grad_norm": 0.06067243218421936, "learning_rate": 3.1278673414471374e-06, "loss": 0.0027, "step": 161330 }, { "epoch": 2.6399410946576127, "grad_norm": 0.030966835096478462, "learning_rate": 3.127205343010664e-06, "loss": 0.0012, "step": 161340 }, { "epoch": 2.6401047206086883, "grad_norm": 0.028663266450166702, "learning_rate": 3.1265433827588798e-06, "loss": 0.0008, "step": 161350 }, { "epoch": 2.6402683465597643, "grad_norm": 0.0032580799888819456, "learning_rate": 3.1258814607052833e-06, "loss": 0.0008, "step": 161360 }, { "epoch": 2.6404319725108403, "grad_norm": 0.0030924726743251085, "learning_rate": 3.1252195768633697e-06, "loss": 0.001, "step": 161370 }, { "epoch": 2.640595598461916, "grad_norm": 0.06316980719566345, "learning_rate": 3.124557731246636e-06, "loss": 0.001, "step": 161380 }, { "epoch": 2.640759224412992, "grad_norm": 0.045083001255989075, "learning_rate": 3.123895923868573e-06, "loss": 0.0006, "step": 161390 }, { "epoch": 2.640922850364068, "grad_norm": 0.12827704846858978, "learning_rate": 3.123234154742678e-06, "loss": 0.0009, "step": 161400 }, { "epoch": 2.6410864763151434, "grad_norm": 0.06300375610589981, "learning_rate": 3.122572423882441e-06, "loss": 0.0006, "step": 161410 }, { "epoch": 2.6412501022662194, "grad_norm": 0.08830621093511581, "learning_rate": 3.1219107313013576e-06, "loss": 0.0011, "step": 161420 }, { "epoch": 2.6414137282172954, "grad_norm": 0.0014743272913619876, "learning_rate": 3.1212490770129155e-06, "loss": 0.0006, "step": 161430 }, { "epoch": 2.641577354168371, "grad_norm": 0.012834037654101849, "learning_rate": 3.1205874610306075e-06, "loss": 0.0004, "step": 161440 }, { "epoch": 2.641740980119447, "grad_norm": 0.0760297179222107, "learning_rate": 3.1199258833679223e-06, "loss": 0.0011, "step": 161450 }, { "epoch": 2.641904606070523, "grad_norm": 0.018418217077851295, "learning_rate": 3.119264344038351e-06, "loss": 0.0003, "step": 161460 }, { "epoch": 2.6420682320215985, "grad_norm": 0.058673590421676636, "learning_rate": 3.1186028430553785e-06, "loss": 0.0012, "step": 161470 }, { "epoch": 2.6422318579726745, "grad_norm": 0.026258645579218864, "learning_rate": 3.1179413804324955e-06, "loss": 0.0004, "step": 161480 }, { "epoch": 2.6423954839237505, "grad_norm": 0.025941476225852966, "learning_rate": 3.1172799561831848e-06, "loss": 0.0007, "step": 161490 }, { "epoch": 2.642559109874826, "grad_norm": 0.021411394700407982, "learning_rate": 3.116618570320937e-06, "loss": 0.0003, "step": 161500 }, { "epoch": 2.642722735825902, "grad_norm": 0.012071702629327774, "learning_rate": 3.1159572228592323e-06, "loss": 0.0007, "step": 161510 }, { "epoch": 2.642886361776978, "grad_norm": 0.018175939098000526, "learning_rate": 3.11529591381156e-06, "loss": 0.0008, "step": 161520 }, { "epoch": 2.6430499877280536, "grad_norm": 0.012741026468575, "learning_rate": 3.1146346431913987e-06, "loss": 0.0005, "step": 161530 }, { "epoch": 2.6432136136791295, "grad_norm": 0.021470416337251663, "learning_rate": 3.1139734110122366e-06, "loss": 0.0005, "step": 161540 }, { "epoch": 2.6433772396302055, "grad_norm": 0.07223737984895706, "learning_rate": 3.1133122172875495e-06, "loss": 0.0009, "step": 161550 }, { "epoch": 2.643540865581281, "grad_norm": 0.05379154533147812, "learning_rate": 3.112651062030825e-06, "loss": 0.0003, "step": 161560 }, { "epoch": 2.643704491532357, "grad_norm": 0.0013109990395605564, "learning_rate": 3.111989945255537e-06, "loss": 0.0004, "step": 161570 }, { "epoch": 2.643868117483433, "grad_norm": 0.012828154489398003, "learning_rate": 3.1113288669751705e-06, "loss": 0.0004, "step": 161580 }, { "epoch": 2.6440317434345086, "grad_norm": 0.004995758179575205, "learning_rate": 3.1106678272032e-06, "loss": 0.0003, "step": 161590 }, { "epoch": 2.6441953693855846, "grad_norm": 0.0163167342543602, "learning_rate": 3.110006825953109e-06, "loss": 0.0004, "step": 161600 }, { "epoch": 2.6443589953366606, "grad_norm": 0.004409296903759241, "learning_rate": 3.109345863238368e-06, "loss": 0.0008, "step": 161610 }, { "epoch": 2.644522621287736, "grad_norm": 0.025749074295163155, "learning_rate": 3.1086849390724595e-06, "loss": 0.0005, "step": 161620 }, { "epoch": 2.644686247238812, "grad_norm": 0.09195730090141296, "learning_rate": 3.108024053468854e-06, "loss": 0.0006, "step": 161630 }, { "epoch": 2.644849873189888, "grad_norm": 0.02569514326751232, "learning_rate": 3.107363206441032e-06, "loss": 0.0004, "step": 161640 }, { "epoch": 2.6450134991409637, "grad_norm": 0.017211303114891052, "learning_rate": 3.1067023980024613e-06, "loss": 0.0009, "step": 161650 }, { "epoch": 2.6451771250920397, "grad_norm": 0.038220297545194626, "learning_rate": 3.1060416281666218e-06, "loss": 0.0009, "step": 161660 }, { "epoch": 2.6453407510431153, "grad_norm": 0.06599236279726028, "learning_rate": 3.1053808969469812e-06, "loss": 0.0011, "step": 161670 }, { "epoch": 2.6455043769941913, "grad_norm": 0.09849605709314346, "learning_rate": 3.104720204357013e-06, "loss": 0.0017, "step": 161680 }, { "epoch": 2.6456680029452673, "grad_norm": 0.05804932489991188, "learning_rate": 3.1040595504101872e-06, "loss": 0.0017, "step": 161690 }, { "epoch": 2.645831628896343, "grad_norm": 0.023229464888572693, "learning_rate": 3.1033989351199757e-06, "loss": 0.0004, "step": 161700 }, { "epoch": 2.645995254847419, "grad_norm": 0.010295792482793331, "learning_rate": 3.102738358499846e-06, "loss": 0.0008, "step": 161710 }, { "epoch": 2.6461588807984944, "grad_norm": 0.017999175935983658, "learning_rate": 3.1020778205632684e-06, "loss": 0.0004, "step": 161720 }, { "epoch": 2.6463225067495704, "grad_norm": 0.08967369794845581, "learning_rate": 3.101417321323709e-06, "loss": 0.0006, "step": 161730 }, { "epoch": 2.6464861327006464, "grad_norm": 0.0628538429737091, "learning_rate": 3.100756860794637e-06, "loss": 0.0016, "step": 161740 }, { "epoch": 2.646649758651722, "grad_norm": 0.03620447218418121, "learning_rate": 3.100096438989516e-06, "loss": 0.0009, "step": 161750 }, { "epoch": 2.646813384602798, "grad_norm": 0.10180577635765076, "learning_rate": 3.099436055921814e-06, "loss": 0.0005, "step": 161760 }, { "epoch": 2.646977010553874, "grad_norm": 0.006097619887441397, "learning_rate": 3.0987757116049937e-06, "loss": 0.0007, "step": 161770 }, { "epoch": 2.6471406365049495, "grad_norm": 0.04619421437382698, "learning_rate": 3.098115406052521e-06, "loss": 0.0008, "step": 161780 }, { "epoch": 2.6473042624560255, "grad_norm": 0.0686073899269104, "learning_rate": 3.0974551392778556e-06, "loss": 0.0013, "step": 161790 }, { "epoch": 2.6474678884071015, "grad_norm": 0.0662199854850769, "learning_rate": 3.0967949112944634e-06, "loss": 0.0007, "step": 161800 }, { "epoch": 2.647631514358177, "grad_norm": 0.008949004113674164, "learning_rate": 3.0961347221158032e-06, "loss": 0.0008, "step": 161810 }, { "epoch": 2.647795140309253, "grad_norm": 0.0062287659384310246, "learning_rate": 3.095474571755338e-06, "loss": 0.0006, "step": 161820 }, { "epoch": 2.647958766260329, "grad_norm": 0.0051177022978663445, "learning_rate": 3.0948144602265258e-06, "loss": 0.0009, "step": 161830 }, { "epoch": 2.6481223922114046, "grad_norm": 0.01892261579632759, "learning_rate": 3.094154387542827e-06, "loss": 0.0008, "step": 161840 }, { "epoch": 2.6482860181624805, "grad_norm": 0.018000174313783646, "learning_rate": 3.0934943537176985e-06, "loss": 0.0012, "step": 161850 }, { "epoch": 2.6484496441135565, "grad_norm": 0.011601028963923454, "learning_rate": 3.0928343587646e-06, "loss": 0.0005, "step": 161860 }, { "epoch": 2.648613270064632, "grad_norm": 0.049265362322330475, "learning_rate": 3.0921744026969856e-06, "loss": 0.0007, "step": 161870 }, { "epoch": 2.648776896015708, "grad_norm": 0.07668159157037735, "learning_rate": 3.0915144855283134e-06, "loss": 0.0006, "step": 161880 }, { "epoch": 2.648940521966784, "grad_norm": 0.07433749735355377, "learning_rate": 3.090854607272037e-06, "loss": 0.0005, "step": 161890 }, { "epoch": 2.6491041479178596, "grad_norm": 0.00819181278347969, "learning_rate": 3.090194767941612e-06, "loss": 0.0015, "step": 161900 }, { "epoch": 2.6492677738689356, "grad_norm": 0.03855130448937416, "learning_rate": 3.0895349675504915e-06, "loss": 0.0009, "step": 161910 }, { "epoch": 2.6494313998200116, "grad_norm": 0.03483477234840393, "learning_rate": 3.0888752061121284e-06, "loss": 0.001, "step": 161920 }, { "epoch": 2.649595025771087, "grad_norm": 0.0809592753648758, "learning_rate": 3.0882154836399735e-06, "loss": 0.0009, "step": 161930 }, { "epoch": 2.649758651722163, "grad_norm": 0.001214375370182097, "learning_rate": 3.0875558001474804e-06, "loss": 0.0007, "step": 161940 }, { "epoch": 2.649922277673239, "grad_norm": 0.06256186962127686, "learning_rate": 3.086896155648097e-06, "loss": 0.0006, "step": 161950 }, { "epoch": 2.6500859036243147, "grad_norm": 0.07008160650730133, "learning_rate": 3.0862365501552747e-06, "loss": 0.0008, "step": 161960 }, { "epoch": 2.6502495295753907, "grad_norm": 0.020084403455257416, "learning_rate": 3.085576983682461e-06, "loss": 0.0005, "step": 161970 }, { "epoch": 2.6504131555264667, "grad_norm": 0.02403631992638111, "learning_rate": 3.084917456243105e-06, "loss": 0.0005, "step": 161980 }, { "epoch": 2.6505767814775423, "grad_norm": 0.031606774777173996, "learning_rate": 3.0842579678506523e-06, "loss": 0.0006, "step": 161990 }, { "epoch": 2.6507404074286183, "grad_norm": 0.08440488576889038, "learning_rate": 3.0835985185185514e-06, "loss": 0.0013, "step": 162000 }, { "epoch": 2.6509040333796943, "grad_norm": 0.0017242662142962217, "learning_rate": 3.082939108260246e-06, "loss": 0.0006, "step": 162010 }, { "epoch": 2.65106765933077, "grad_norm": 0.0005832304595969617, "learning_rate": 3.0822797370891835e-06, "loss": 0.0007, "step": 162020 }, { "epoch": 2.651231285281846, "grad_norm": 0.018364788964390755, "learning_rate": 3.0816204050188038e-06, "loss": 0.001, "step": 162030 }, { "epoch": 2.651394911232922, "grad_norm": 0.0873674601316452, "learning_rate": 3.0809611120625544e-06, "loss": 0.0013, "step": 162040 }, { "epoch": 2.6515585371839974, "grad_norm": 0.08226965367794037, "learning_rate": 3.0803018582338744e-06, "loss": 0.0007, "step": 162050 }, { "epoch": 2.6517221631350734, "grad_norm": 0.012823255732655525, "learning_rate": 3.079642643546208e-06, "loss": 0.0006, "step": 162060 }, { "epoch": 2.651885789086149, "grad_norm": 0.0021352916955947876, "learning_rate": 3.0789834680129937e-06, "loss": 0.0011, "step": 162070 }, { "epoch": 2.652049415037225, "grad_norm": 0.09397493302822113, "learning_rate": 3.078324331647674e-06, "loss": 0.0017, "step": 162080 }, { "epoch": 2.652213040988301, "grad_norm": 0.01573173701763153, "learning_rate": 3.077665234463686e-06, "loss": 0.0007, "step": 162090 }, { "epoch": 2.6523766669393765, "grad_norm": 0.006756484974175692, "learning_rate": 3.07700617647447e-06, "loss": 0.0007, "step": 162100 }, { "epoch": 2.6525402928904525, "grad_norm": 0.035446759313344955, "learning_rate": 3.076347157693461e-06, "loss": 0.0008, "step": 162110 }, { "epoch": 2.652703918841528, "grad_norm": 0.1361808031797409, "learning_rate": 3.0756881781341007e-06, "loss": 0.0011, "step": 162120 }, { "epoch": 2.652867544792604, "grad_norm": 0.06383164972066879, "learning_rate": 3.075029237809818e-06, "loss": 0.0009, "step": 162130 }, { "epoch": 2.65303117074368, "grad_norm": 0.06612243503332138, "learning_rate": 3.074370336734055e-06, "loss": 0.0016, "step": 162140 }, { "epoch": 2.6531947966947556, "grad_norm": 0.0036031503695994616, "learning_rate": 3.07371147492024e-06, "loss": 0.001, "step": 162150 }, { "epoch": 2.6533584226458315, "grad_norm": 0.08030679821968079, "learning_rate": 3.0730526523818133e-06, "loss": 0.0007, "step": 162160 }, { "epoch": 2.6535220485969075, "grad_norm": 0.014330684207379818, "learning_rate": 3.0723938691322008e-06, "loss": 0.0007, "step": 162170 }, { "epoch": 2.653685674547983, "grad_norm": 0.027740174904465675, "learning_rate": 3.071735125184841e-06, "loss": 0.0009, "step": 162180 }, { "epoch": 2.653849300499059, "grad_norm": 0.03212345018982887, "learning_rate": 3.071076420553159e-06, "loss": 0.0008, "step": 162190 }, { "epoch": 2.654012926450135, "grad_norm": 0.04207668453454971, "learning_rate": 3.0704177552505913e-06, "loss": 0.0004, "step": 162200 }, { "epoch": 2.6541765524012106, "grad_norm": 0.09152814745903015, "learning_rate": 3.0697591292905616e-06, "loss": 0.0009, "step": 162210 }, { "epoch": 2.6543401783522866, "grad_norm": 0.04520801827311516, "learning_rate": 3.069100542686505e-06, "loss": 0.001, "step": 162220 }, { "epoch": 2.6545038043033626, "grad_norm": 0.037749551236629486, "learning_rate": 3.068441995451843e-06, "loss": 0.0006, "step": 162230 }, { "epoch": 2.654667430254438, "grad_norm": 0.01277028676122427, "learning_rate": 3.0677834876000085e-06, "loss": 0.0006, "step": 162240 }, { "epoch": 2.654831056205514, "grad_norm": 0.022160455584526062, "learning_rate": 3.0671250191444226e-06, "loss": 0.0007, "step": 162250 }, { "epoch": 2.65499468215659, "grad_norm": 0.019076194614171982, "learning_rate": 3.066466590098517e-06, "loss": 0.0011, "step": 162260 }, { "epoch": 2.6551583081076657, "grad_norm": 0.002896291669458151, "learning_rate": 3.06580820047571e-06, "loss": 0.0004, "step": 162270 }, { "epoch": 2.6553219340587417, "grad_norm": 0.03278978168964386, "learning_rate": 3.065149850289432e-06, "loss": 0.001, "step": 162280 }, { "epoch": 2.6554855600098177, "grad_norm": 0.28031235933303833, "learning_rate": 3.0644915395530993e-06, "loss": 0.0007, "step": 162290 }, { "epoch": 2.6556491859608933, "grad_norm": 0.14422543346881866, "learning_rate": 3.063833268280142e-06, "loss": 0.001, "step": 162300 }, { "epoch": 2.6558128119119693, "grad_norm": 0.033319223672151566, "learning_rate": 3.0631750364839754e-06, "loss": 0.0007, "step": 162310 }, { "epoch": 2.6559764378630453, "grad_norm": 0.03922286257147789, "learning_rate": 3.0625168441780235e-06, "loss": 0.0008, "step": 162320 }, { "epoch": 2.656140063814121, "grad_norm": 0.04597333073616028, "learning_rate": 3.061858691375704e-06, "loss": 0.0011, "step": 162330 }, { "epoch": 2.656303689765197, "grad_norm": 0.011832577176392078, "learning_rate": 3.061200578090439e-06, "loss": 0.0008, "step": 162340 }, { "epoch": 2.656467315716273, "grad_norm": 0.05440326780080795, "learning_rate": 3.060542504335644e-06, "loss": 0.0009, "step": 162350 }, { "epoch": 2.6566309416673484, "grad_norm": 0.033250875771045685, "learning_rate": 3.0598844701247395e-06, "loss": 0.0006, "step": 162360 }, { "epoch": 2.6567945676184244, "grad_norm": 0.006109187845140696, "learning_rate": 3.0592264754711387e-06, "loss": 0.0006, "step": 162370 }, { "epoch": 2.6569581935695004, "grad_norm": 0.027035899460315704, "learning_rate": 3.0585685203882617e-06, "loss": 0.0006, "step": 162380 }, { "epoch": 2.657121819520576, "grad_norm": 0.04064914584159851, "learning_rate": 3.057910604889519e-06, "loss": 0.0011, "step": 162390 }, { "epoch": 2.657285445471652, "grad_norm": 0.0384272038936615, "learning_rate": 3.057252728988329e-06, "loss": 0.0009, "step": 162400 }, { "epoch": 2.657449071422728, "grad_norm": 0.027098167687654495, "learning_rate": 3.056594892698103e-06, "loss": 0.0008, "step": 162410 }, { "epoch": 2.6576126973738035, "grad_norm": 0.019588515162467957, "learning_rate": 3.0559370960322556e-06, "loss": 0.0013, "step": 162420 }, { "epoch": 2.6577763233248795, "grad_norm": 0.04049665480852127, "learning_rate": 3.0552793390041957e-06, "loss": 0.0007, "step": 162430 }, { "epoch": 2.657939949275955, "grad_norm": 0.045909248292446136, "learning_rate": 3.054621621627337e-06, "loss": 0.0007, "step": 162440 }, { "epoch": 2.658103575227031, "grad_norm": 0.036165449768304825, "learning_rate": 3.0539639439150883e-06, "loss": 0.0005, "step": 162450 }, { "epoch": 2.658267201178107, "grad_norm": 0.002897100057452917, "learning_rate": 3.053306305880861e-06, "loss": 0.0008, "step": 162460 }, { "epoch": 2.6584308271291825, "grad_norm": 0.04414387419819832, "learning_rate": 3.052648707538061e-06, "loss": 0.0012, "step": 162470 }, { "epoch": 2.6585944530802585, "grad_norm": 0.040566034615039825, "learning_rate": 3.0519911489000985e-06, "loss": 0.0008, "step": 162480 }, { "epoch": 2.658758079031334, "grad_norm": 0.0020596261601895094, "learning_rate": 3.05133362998038e-06, "loss": 0.0013, "step": 162490 }, { "epoch": 2.65892170498241, "grad_norm": 0.019974179565906525, "learning_rate": 3.050676150792311e-06, "loss": 0.0007, "step": 162500 }, { "epoch": 2.659085330933486, "grad_norm": 0.0038914019241929054, "learning_rate": 3.0500187113492976e-06, "loss": 0.0006, "step": 162510 }, { "epoch": 2.6592489568845616, "grad_norm": 0.007375596556812525, "learning_rate": 3.0493613116647446e-06, "loss": 0.0003, "step": 162520 }, { "epoch": 2.6594125828356376, "grad_norm": 0.0045276181772351265, "learning_rate": 3.0487039517520545e-06, "loss": 0.0004, "step": 162530 }, { "epoch": 2.6595762087867136, "grad_norm": 0.05848367512226105, "learning_rate": 3.0480466316246327e-06, "loss": 0.0007, "step": 162540 }, { "epoch": 2.659739834737789, "grad_norm": 0.010172338224947453, "learning_rate": 3.047389351295879e-06, "loss": 0.0003, "step": 162550 }, { "epoch": 2.659903460688865, "grad_norm": 0.0023920137900859118, "learning_rate": 3.046732110779197e-06, "loss": 0.0005, "step": 162560 }, { "epoch": 2.660067086639941, "grad_norm": 0.047040387988090515, "learning_rate": 3.0460749100879845e-06, "loss": 0.0007, "step": 162570 }, { "epoch": 2.6602307125910167, "grad_norm": 0.032827481627464294, "learning_rate": 3.045417749235644e-06, "loss": 0.0007, "step": 162580 }, { "epoch": 2.6603943385420927, "grad_norm": 0.033477429300546646, "learning_rate": 3.0447606282355725e-06, "loss": 0.0005, "step": 162590 }, { "epoch": 2.6605579644931687, "grad_norm": 0.05627338960766792, "learning_rate": 3.0441035471011692e-06, "loss": 0.0009, "step": 162600 }, { "epoch": 2.6607215904442443, "grad_norm": 0.022236453369259834, "learning_rate": 3.0434465058458306e-06, "loss": 0.0003, "step": 162610 }, { "epoch": 2.6608852163953203, "grad_norm": 0.042935293167829514, "learning_rate": 3.042789504482955e-06, "loss": 0.0007, "step": 162620 }, { "epoch": 2.6610488423463963, "grad_norm": 0.060571182519197464, "learning_rate": 3.0421325430259354e-06, "loss": 0.001, "step": 162630 }, { "epoch": 2.661212468297472, "grad_norm": 0.006676092743873596, "learning_rate": 3.0414756214881696e-06, "loss": 0.0008, "step": 162640 }, { "epoch": 2.661376094248548, "grad_norm": 0.04320293664932251, "learning_rate": 3.0408187398830493e-06, "loss": 0.0009, "step": 162650 }, { "epoch": 2.661539720199624, "grad_norm": 0.06439736485481262, "learning_rate": 3.040161898223969e-06, "loss": 0.0011, "step": 162660 }, { "epoch": 2.6617033461506994, "grad_norm": 0.004045261535793543, "learning_rate": 3.03950509652432e-06, "loss": 0.0004, "step": 162670 }, { "epoch": 2.6618669721017754, "grad_norm": 0.007853190414607525, "learning_rate": 3.038848334797496e-06, "loss": 0.0013, "step": 162680 }, { "epoch": 2.6620305980528514, "grad_norm": 0.004403184168040752, "learning_rate": 3.0381916130568854e-06, "loss": 0.0008, "step": 162690 }, { "epoch": 2.662194224003927, "grad_norm": 0.009033712558448315, "learning_rate": 3.037534931315881e-06, "loss": 0.0008, "step": 162700 }, { "epoch": 2.662357849955003, "grad_norm": 0.04180406033992767, "learning_rate": 3.0368782895878684e-06, "loss": 0.0006, "step": 162710 }, { "epoch": 2.662521475906079, "grad_norm": 0.04987412691116333, "learning_rate": 3.0362216878862394e-06, "loss": 0.0012, "step": 162720 }, { "epoch": 2.6626851018571545, "grad_norm": 0.02704285830259323, "learning_rate": 3.035565126224379e-06, "loss": 0.0008, "step": 162730 }, { "epoch": 2.6628487278082305, "grad_norm": 0.005667391233146191, "learning_rate": 3.0349086046156763e-06, "loss": 0.0012, "step": 162740 }, { "epoch": 2.6630123537593064, "grad_norm": 0.16449770331382751, "learning_rate": 3.0342521230735144e-06, "loss": 0.0009, "step": 162750 }, { "epoch": 2.663175979710382, "grad_norm": 0.03146227449178696, "learning_rate": 3.033595681611281e-06, "loss": 0.001, "step": 162760 }, { "epoch": 2.663339605661458, "grad_norm": 0.037311553955078125, "learning_rate": 3.032939280242358e-06, "loss": 0.0007, "step": 162770 }, { "epoch": 2.663503231612534, "grad_norm": 0.02146373689174652, "learning_rate": 3.032282918980133e-06, "loss": 0.0007, "step": 162780 }, { "epoch": 2.6636668575636095, "grad_norm": 0.0025826359633356333, "learning_rate": 3.0316265978379817e-06, "loss": 0.0011, "step": 162790 }, { "epoch": 2.6638304835146855, "grad_norm": 0.036508820950984955, "learning_rate": 3.0309703168292937e-06, "loss": 0.0007, "step": 162800 }, { "epoch": 2.6639941094657615, "grad_norm": 0.04718393832445145, "learning_rate": 3.0303140759674433e-06, "loss": 0.0006, "step": 162810 }, { "epoch": 2.664157735416837, "grad_norm": 0.029807649552822113, "learning_rate": 3.029657875265817e-06, "loss": 0.0007, "step": 162820 }, { "epoch": 2.664321361367913, "grad_norm": 0.057197779417037964, "learning_rate": 3.029001714737787e-06, "loss": 0.001, "step": 162830 }, { "epoch": 2.6644849873189886, "grad_norm": 0.008562426082789898, "learning_rate": 3.0283455943967395e-06, "loss": 0.0004, "step": 162840 }, { "epoch": 2.6646486132700646, "grad_norm": 0.060174260288476944, "learning_rate": 3.0276895142560446e-06, "loss": 0.0014, "step": 162850 }, { "epoch": 2.6648122392211406, "grad_norm": 0.04495071992278099, "learning_rate": 3.0270334743290876e-06, "loss": 0.0005, "step": 162860 }, { "epoch": 2.664975865172216, "grad_norm": 0.06554047018289566, "learning_rate": 3.0263774746292364e-06, "loss": 0.0013, "step": 162870 }, { "epoch": 2.665139491123292, "grad_norm": 0.024218251928687096, "learning_rate": 3.0257215151698728e-06, "loss": 0.0007, "step": 162880 }, { "epoch": 2.6653031170743677, "grad_norm": 0.05660524591803551, "learning_rate": 3.0250655959643655e-06, "loss": 0.0009, "step": 162890 }, { "epoch": 2.6654667430254437, "grad_norm": 0.06864748895168304, "learning_rate": 3.0244097170260943e-06, "loss": 0.0008, "step": 162900 }, { "epoch": 2.6656303689765197, "grad_norm": 0.005950404331088066, "learning_rate": 3.023753878368425e-06, "loss": 0.0006, "step": 162910 }, { "epoch": 2.6657939949275953, "grad_norm": 0.002710176631808281, "learning_rate": 3.023098080004737e-06, "loss": 0.0007, "step": 162920 }, { "epoch": 2.6659576208786713, "grad_norm": 0.06944464892148972, "learning_rate": 3.0224423219483954e-06, "loss": 0.002, "step": 162930 }, { "epoch": 2.6661212468297473, "grad_norm": 0.03162287920713425, "learning_rate": 3.0217866042127752e-06, "loss": 0.0006, "step": 162940 }, { "epoch": 2.666284872780823, "grad_norm": 0.019546495750546455, "learning_rate": 3.021130926811242e-06, "loss": 0.0007, "step": 162950 }, { "epoch": 2.666448498731899, "grad_norm": 0.11149882525205612, "learning_rate": 3.0204752897571676e-06, "loss": 0.0009, "step": 162960 }, { "epoch": 2.666612124682975, "grad_norm": 0.03676782548427582, "learning_rate": 3.0198196930639167e-06, "loss": 0.0008, "step": 162970 }, { "epoch": 2.6667757506340504, "grad_norm": 0.01851705275475979, "learning_rate": 3.0191641367448598e-06, "loss": 0.0008, "step": 162980 }, { "epoch": 2.6669393765851264, "grad_norm": 0.03480890765786171, "learning_rate": 3.0185086208133608e-06, "loss": 0.0005, "step": 162990 }, { "epoch": 2.6671030025362024, "grad_norm": 0.022429170086979866, "learning_rate": 3.0178531452827863e-06, "loss": 0.0004, "step": 163000 }, { "epoch": 2.667266628487278, "grad_norm": 0.0012407767353579402, "learning_rate": 3.0171977101664996e-06, "loss": 0.0002, "step": 163010 }, { "epoch": 2.667430254438354, "grad_norm": 0.0024613901041448116, "learning_rate": 3.016542315477866e-06, "loss": 0.0006, "step": 163020 }, { "epoch": 2.66759388038943, "grad_norm": 0.028154218569397926, "learning_rate": 3.0158869612302467e-06, "loss": 0.0004, "step": 163030 }, { "epoch": 2.6677575063405055, "grad_norm": 0.031238261610269547, "learning_rate": 3.015231647437006e-06, "loss": 0.0014, "step": 163040 }, { "epoch": 2.6679211322915815, "grad_norm": 0.026049291715025902, "learning_rate": 3.014576374111503e-06, "loss": 0.0005, "step": 163050 }, { "epoch": 2.6680847582426575, "grad_norm": 0.034736268222332, "learning_rate": 3.0139211412671e-06, "loss": 0.0007, "step": 163060 }, { "epoch": 2.668248384193733, "grad_norm": 0.008239570073783398, "learning_rate": 3.0132659489171556e-06, "loss": 0.0005, "step": 163070 }, { "epoch": 2.668412010144809, "grad_norm": 0.03963981196284294, "learning_rate": 3.012610797075029e-06, "loss": 0.0008, "step": 163080 }, { "epoch": 2.668575636095885, "grad_norm": 0.009353777393698692, "learning_rate": 3.0119556857540776e-06, "loss": 0.0005, "step": 163090 }, { "epoch": 2.6687392620469605, "grad_norm": 0.009686528705060482, "learning_rate": 3.0113006149676595e-06, "loss": 0.0003, "step": 163100 }, { "epoch": 2.6689028879980365, "grad_norm": 0.029986849054694176, "learning_rate": 3.0106455847291296e-06, "loss": 0.0008, "step": 163110 }, { "epoch": 2.6690665139491125, "grad_norm": 0.027253905311226845, "learning_rate": 3.009990595051846e-06, "loss": 0.0005, "step": 163120 }, { "epoch": 2.669230139900188, "grad_norm": 0.04151954874396324, "learning_rate": 3.0093356459491594e-06, "loss": 0.0009, "step": 163130 }, { "epoch": 2.669393765851264, "grad_norm": 0.02611413039267063, "learning_rate": 3.0086807374344284e-06, "loss": 0.0005, "step": 163140 }, { "epoch": 2.66955739180234, "grad_norm": 0.053534992039203644, "learning_rate": 3.008025869521002e-06, "loss": 0.0009, "step": 163150 }, { "epoch": 2.6697210177534156, "grad_norm": 0.031659480184316635, "learning_rate": 3.0073710422222347e-06, "loss": 0.0006, "step": 163160 }, { "epoch": 2.6698846437044916, "grad_norm": 0.012235830537974834, "learning_rate": 3.0067162555514766e-06, "loss": 0.0007, "step": 163170 }, { "epoch": 2.6700482696555676, "grad_norm": 0.019455686211586, "learning_rate": 3.00606150952208e-06, "loss": 0.0005, "step": 163180 }, { "epoch": 2.670211895606643, "grad_norm": 0.037597671151161194, "learning_rate": 3.0054068041473915e-06, "loss": 0.0012, "step": 163190 }, { "epoch": 2.670375521557719, "grad_norm": 0.08969694375991821, "learning_rate": 3.0047521394407643e-06, "loss": 0.0007, "step": 163200 }, { "epoch": 2.6705391475087947, "grad_norm": 0.07946503162384033, "learning_rate": 3.0040975154155422e-06, "loss": 0.0005, "step": 163210 }, { "epoch": 2.6707027734598707, "grad_norm": 0.06927832216024399, "learning_rate": 3.0034429320850754e-06, "loss": 0.0007, "step": 163220 }, { "epoch": 2.6708663994109467, "grad_norm": 0.07346218079328537, "learning_rate": 3.0027883894627087e-06, "loss": 0.0009, "step": 163230 }, { "epoch": 2.6710300253620223, "grad_norm": 0.03700733929872513, "learning_rate": 3.0021338875617885e-06, "loss": 0.0006, "step": 163240 }, { "epoch": 2.6711936513130983, "grad_norm": 0.023511067032814026, "learning_rate": 3.001479426395658e-06, "loss": 0.0014, "step": 163250 }, { "epoch": 2.671357277264174, "grad_norm": 0.0024863039143383503, "learning_rate": 3.0008250059776646e-06, "loss": 0.0005, "step": 163260 }, { "epoch": 2.67152090321525, "grad_norm": 0.056298159062862396, "learning_rate": 3.0001706263211467e-06, "loss": 0.0017, "step": 163270 }, { "epoch": 2.671684529166326, "grad_norm": 0.00638155359774828, "learning_rate": 2.99951628743945e-06, "loss": 0.0007, "step": 163280 }, { "epoch": 2.6718481551174014, "grad_norm": 0.007004431448876858, "learning_rate": 2.9988619893459146e-06, "loss": 0.0005, "step": 163290 }, { "epoch": 2.6720117810684774, "grad_norm": 0.01897265948355198, "learning_rate": 2.9982077320538828e-06, "loss": 0.0005, "step": 163300 }, { "epoch": 2.6721754070195534, "grad_norm": 0.02926579862833023, "learning_rate": 2.997553515576691e-06, "loss": 0.0007, "step": 163310 }, { "epoch": 2.672339032970629, "grad_norm": 0.03567328304052353, "learning_rate": 2.996899339927681e-06, "loss": 0.0005, "step": 163320 }, { "epoch": 2.672502658921705, "grad_norm": 0.009700595401227474, "learning_rate": 2.9962452051201887e-06, "loss": 0.0009, "step": 163330 }, { "epoch": 2.672666284872781, "grad_norm": 0.03349621966481209, "learning_rate": 2.995591111167554e-06, "loss": 0.0008, "step": 163340 }, { "epoch": 2.6728299108238565, "grad_norm": 0.0026717553846538067, "learning_rate": 2.9949370580831103e-06, "loss": 0.0005, "step": 163350 }, { "epoch": 2.6729935367749325, "grad_norm": 0.019069213420152664, "learning_rate": 2.9942830458801965e-06, "loss": 0.0007, "step": 163360 }, { "epoch": 2.6731571627260085, "grad_norm": 0.09787153452634811, "learning_rate": 2.9936290745721443e-06, "loss": 0.0007, "step": 163370 }, { "epoch": 2.673320788677084, "grad_norm": 0.0676492229104042, "learning_rate": 2.99297514417229e-06, "loss": 0.0009, "step": 163380 }, { "epoch": 2.67348441462816, "grad_norm": 0.07875342667102814, "learning_rate": 2.9923212546939647e-06, "loss": 0.0007, "step": 163390 }, { "epoch": 2.673648040579236, "grad_norm": 0.02524031139910221, "learning_rate": 2.991667406150502e-06, "loss": 0.0003, "step": 163400 }, { "epoch": 2.6738116665303115, "grad_norm": 0.04077944532036781, "learning_rate": 2.9910135985552324e-06, "loss": 0.0009, "step": 163410 }, { "epoch": 2.6739752924813875, "grad_norm": 0.008859643712639809, "learning_rate": 2.9903598319214887e-06, "loss": 0.0008, "step": 163420 }, { "epoch": 2.6741389184324635, "grad_norm": 0.02783234603703022, "learning_rate": 2.9897061062625953e-06, "loss": 0.0021, "step": 163430 }, { "epoch": 2.674302544383539, "grad_norm": 0.008123842068016529, "learning_rate": 2.989052421591889e-06, "loss": 0.0007, "step": 163440 }, { "epoch": 2.674466170334615, "grad_norm": 0.08057883381843567, "learning_rate": 2.9883987779226897e-06, "loss": 0.0009, "step": 163450 }, { "epoch": 2.674629796285691, "grad_norm": 0.03015994466841221, "learning_rate": 2.987745175268332e-06, "loss": 0.0005, "step": 163460 }, { "epoch": 2.6747934222367666, "grad_norm": 0.03617357090115547, "learning_rate": 2.9870916136421354e-06, "loss": 0.0007, "step": 163470 }, { "epoch": 2.6749570481878426, "grad_norm": 0.030398843809962273, "learning_rate": 2.9864380930574317e-06, "loss": 0.0008, "step": 163480 }, { "epoch": 2.6751206741389186, "grad_norm": 0.005847258027642965, "learning_rate": 2.9857846135275404e-06, "loss": 0.0007, "step": 163490 }, { "epoch": 2.675284300089994, "grad_norm": 0.1508459597826004, "learning_rate": 2.98513117506579e-06, "loss": 0.0006, "step": 163500 }, { "epoch": 2.67544792604107, "grad_norm": 0.003086095442995429, "learning_rate": 2.9844777776854992e-06, "loss": 0.0004, "step": 163510 }, { "epoch": 2.675611551992146, "grad_norm": 0.17879246175289154, "learning_rate": 2.9838244213999945e-06, "loss": 0.0021, "step": 163520 }, { "epoch": 2.6757751779432217, "grad_norm": 0.021098667755723, "learning_rate": 2.983171106222593e-06, "loss": 0.0003, "step": 163530 }, { "epoch": 2.6759388038942977, "grad_norm": 0.053473033010959625, "learning_rate": 2.9825178321666204e-06, "loss": 0.0004, "step": 163540 }, { "epoch": 2.6761024298453737, "grad_norm": 0.004616378806531429, "learning_rate": 2.98186459924539e-06, "loss": 0.0013, "step": 163550 }, { "epoch": 2.6762660557964493, "grad_norm": 0.023267023265361786, "learning_rate": 2.981211407472227e-06, "loss": 0.0005, "step": 163560 }, { "epoch": 2.6764296817475253, "grad_norm": 0.0024195946753025055, "learning_rate": 2.9805582568604437e-06, "loss": 0.0009, "step": 163570 }, { "epoch": 2.6765933076986013, "grad_norm": 0.154166579246521, "learning_rate": 2.979905147423363e-06, "loss": 0.0009, "step": 163580 }, { "epoch": 2.676756933649677, "grad_norm": 0.02709219604730606, "learning_rate": 2.979252079174295e-06, "loss": 0.0006, "step": 163590 }, { "epoch": 2.676920559600753, "grad_norm": 0.008788411505520344, "learning_rate": 2.9785990521265617e-06, "loss": 0.0013, "step": 163600 }, { "epoch": 2.6770841855518284, "grad_norm": 0.058148257434368134, "learning_rate": 2.9779460662934727e-06, "loss": 0.0011, "step": 163610 }, { "epoch": 2.6772478115029044, "grad_norm": 0.05993300676345825, "learning_rate": 2.977293121688345e-06, "loss": 0.0008, "step": 163620 }, { "epoch": 2.6774114374539804, "grad_norm": 0.06525691598653793, "learning_rate": 2.976640218324489e-06, "loss": 0.0006, "step": 163630 }, { "epoch": 2.677575063405056, "grad_norm": 0.017809860408306122, "learning_rate": 2.975987356215219e-06, "loss": 0.0008, "step": 163640 }, { "epoch": 2.677738689356132, "grad_norm": 0.026320429518818855, "learning_rate": 2.975334535373845e-06, "loss": 0.0011, "step": 163650 }, { "epoch": 2.6779023153072075, "grad_norm": 0.006955885328352451, "learning_rate": 2.974681755813678e-06, "loss": 0.0005, "step": 163660 }, { "epoch": 2.6780659412582835, "grad_norm": 0.01299950759857893, "learning_rate": 2.974029017548027e-06, "loss": 0.0005, "step": 163670 }, { "epoch": 2.6782295672093595, "grad_norm": 0.033198874443769455, "learning_rate": 2.9733763205902022e-06, "loss": 0.0006, "step": 163680 }, { "epoch": 2.678393193160435, "grad_norm": 0.017162909731268883, "learning_rate": 2.9727236649535103e-06, "loss": 0.0005, "step": 163690 }, { "epoch": 2.678556819111511, "grad_norm": 0.017021365463733673, "learning_rate": 2.972071050651259e-06, "loss": 0.0007, "step": 163700 }, { "epoch": 2.678720445062587, "grad_norm": 0.03489486873149872, "learning_rate": 2.9714184776967537e-06, "loss": 0.0011, "step": 163710 }, { "epoch": 2.6788840710136625, "grad_norm": 0.06556104123592377, "learning_rate": 2.970765946103301e-06, "loss": 0.0013, "step": 163720 }, { "epoch": 2.6790476969647385, "grad_norm": 0.02820749394595623, "learning_rate": 2.9701134558842045e-06, "loss": 0.0008, "step": 163730 }, { "epoch": 2.6792113229158145, "grad_norm": 0.10611628741025925, "learning_rate": 2.9694610070527687e-06, "loss": 0.0005, "step": 163740 }, { "epoch": 2.67937494886689, "grad_norm": 0.012790380977094173, "learning_rate": 2.968808599622295e-06, "loss": 0.0005, "step": 163750 }, { "epoch": 2.679538574817966, "grad_norm": 0.03278224915266037, "learning_rate": 2.968156233606088e-06, "loss": 0.0007, "step": 163760 }, { "epoch": 2.679702200769042, "grad_norm": 0.07275401800870895, "learning_rate": 2.9675039090174456e-06, "loss": 0.001, "step": 163770 }, { "epoch": 2.6798658267201176, "grad_norm": 0.027617307379841805, "learning_rate": 2.9668516258696713e-06, "loss": 0.0005, "step": 163780 }, { "epoch": 2.6800294526711936, "grad_norm": 0.032421521842479706, "learning_rate": 2.966199384176063e-06, "loss": 0.0007, "step": 163790 }, { "epoch": 2.6801930786222696, "grad_norm": 0.07466503977775574, "learning_rate": 2.9655471839499195e-06, "loss": 0.0004, "step": 163800 }, { "epoch": 2.680356704573345, "grad_norm": 0.0461210198700428, "learning_rate": 2.964895025204538e-06, "loss": 0.0014, "step": 163810 }, { "epoch": 2.680520330524421, "grad_norm": 0.024356169626116753, "learning_rate": 2.964242907953217e-06, "loss": 0.0003, "step": 163820 }, { "epoch": 2.680683956475497, "grad_norm": 0.018818585202097893, "learning_rate": 2.963590832209251e-06, "loss": 0.0009, "step": 163830 }, { "epoch": 2.6808475824265727, "grad_norm": 0.07113000005483627, "learning_rate": 2.962938797985937e-06, "loss": 0.0008, "step": 163840 }, { "epoch": 2.6810112083776487, "grad_norm": 0.036134954541921616, "learning_rate": 2.962286805296567e-06, "loss": 0.0005, "step": 163850 }, { "epoch": 2.6811748343287247, "grad_norm": 0.004073712974786758, "learning_rate": 2.961634854154438e-06, "loss": 0.0006, "step": 163860 }, { "epoch": 2.6813384602798003, "grad_norm": 0.038884375244379044, "learning_rate": 2.9609829445728387e-06, "loss": 0.0009, "step": 163870 }, { "epoch": 2.6815020862308763, "grad_norm": 0.033929843455553055, "learning_rate": 2.960331076565065e-06, "loss": 0.0008, "step": 163880 }, { "epoch": 2.6816657121819523, "grad_norm": 0.0718521997332573, "learning_rate": 2.9596792501444044e-06, "loss": 0.0005, "step": 163890 }, { "epoch": 2.681829338133028, "grad_norm": 0.04105739668011665, "learning_rate": 2.9590274653241497e-06, "loss": 0.0007, "step": 163900 }, { "epoch": 2.681992964084104, "grad_norm": 0.022866031154990196, "learning_rate": 2.958375722117588e-06, "loss": 0.0012, "step": 163910 }, { "epoch": 2.68215659003518, "grad_norm": 0.017566299065947533, "learning_rate": 2.9577240205380107e-06, "loss": 0.001, "step": 163920 }, { "epoch": 2.6823202159862554, "grad_norm": 0.021244661882519722, "learning_rate": 2.9570723605987017e-06, "loss": 0.0007, "step": 163930 }, { "epoch": 2.6824838419373314, "grad_norm": 0.04740865156054497, "learning_rate": 2.9564207423129522e-06, "loss": 0.0007, "step": 163940 }, { "epoch": 2.6826474678884074, "grad_norm": 0.004084141459316015, "learning_rate": 2.9557691656940435e-06, "loss": 0.0004, "step": 163950 }, { "epoch": 2.682811093839483, "grad_norm": 0.029127879068255424, "learning_rate": 2.9551176307552642e-06, "loss": 0.0005, "step": 163960 }, { "epoch": 2.682974719790559, "grad_norm": 0.09016800671815872, "learning_rate": 2.954466137509897e-06, "loss": 0.0011, "step": 163970 }, { "epoch": 2.683138345741635, "grad_norm": 0.06637147068977356, "learning_rate": 2.953814685971226e-06, "loss": 0.0006, "step": 163980 }, { "epoch": 2.6833019716927105, "grad_norm": 0.05557726323604584, "learning_rate": 2.953163276152532e-06, "loss": 0.0008, "step": 163990 }, { "epoch": 2.6834655976437864, "grad_norm": 0.039374444633722305, "learning_rate": 2.9525119080670995e-06, "loss": 0.0009, "step": 164000 }, { "epoch": 2.683629223594862, "grad_norm": 0.11049165576696396, "learning_rate": 2.9518605817282075e-06, "loss": 0.0005, "step": 164010 }, { "epoch": 2.683792849545938, "grad_norm": 0.0006823497242294252, "learning_rate": 2.951209297149136e-06, "loss": 0.0003, "step": 164020 }, { "epoch": 2.6839564754970135, "grad_norm": 0.1452331244945526, "learning_rate": 2.9505580543431642e-06, "loss": 0.0012, "step": 164030 }, { "epoch": 2.6841201014480895, "grad_norm": 0.0013327004853636026, "learning_rate": 2.9499068533235718e-06, "loss": 0.0012, "step": 164040 }, { "epoch": 2.6842837273991655, "grad_norm": 0.06238269805908203, "learning_rate": 2.9492556941036344e-06, "loss": 0.0007, "step": 164050 }, { "epoch": 2.684447353350241, "grad_norm": 0.0434865839779377, "learning_rate": 2.948604576696631e-06, "loss": 0.0015, "step": 164060 }, { "epoch": 2.684610979301317, "grad_norm": 0.010050742886960506, "learning_rate": 2.9479535011158324e-06, "loss": 0.0009, "step": 164070 }, { "epoch": 2.684774605252393, "grad_norm": 0.007644369266927242, "learning_rate": 2.9473024673745197e-06, "loss": 0.0005, "step": 164080 }, { "epoch": 2.6849382312034686, "grad_norm": 0.01921002008020878, "learning_rate": 2.946651475485961e-06, "loss": 0.0006, "step": 164090 }, { "epoch": 2.6851018571545446, "grad_norm": 0.043826691806316376, "learning_rate": 2.9460005254634354e-06, "loss": 0.0004, "step": 164100 }, { "epoch": 2.6852654831056206, "grad_norm": 0.011659389361739159, "learning_rate": 2.9453496173202097e-06, "loss": 0.0008, "step": 164110 }, { "epoch": 2.685429109056696, "grad_norm": 0.012585148215293884, "learning_rate": 2.9446987510695602e-06, "loss": 0.0007, "step": 164120 }, { "epoch": 2.685592735007772, "grad_norm": 0.0038121133111417294, "learning_rate": 2.9440479267247524e-06, "loss": 0.001, "step": 164130 }, { "epoch": 2.685756360958848, "grad_norm": 0.03864398971199989, "learning_rate": 2.9433971442990618e-06, "loss": 0.001, "step": 164140 }, { "epoch": 2.6859199869099237, "grad_norm": 0.036153580993413925, "learning_rate": 2.9427464038057516e-06, "loss": 0.0005, "step": 164150 }, { "epoch": 2.6860836128609997, "grad_norm": 0.020376056432724, "learning_rate": 2.9420957052580958e-06, "loss": 0.001, "step": 164160 }, { "epoch": 2.6862472388120757, "grad_norm": 0.10579604655504227, "learning_rate": 2.9414450486693545e-06, "loss": 0.0005, "step": 164170 }, { "epoch": 2.6864108647631513, "grad_norm": 0.10437638312578201, "learning_rate": 2.9407944340528017e-06, "loss": 0.0012, "step": 164180 }, { "epoch": 2.6865744907142273, "grad_norm": 0.024587169289588928, "learning_rate": 2.940143861421696e-06, "loss": 0.0009, "step": 164190 }, { "epoch": 2.6867381166653033, "grad_norm": 0.028207028284668922, "learning_rate": 2.939493330789308e-06, "loss": 0.0016, "step": 164200 }, { "epoch": 2.686901742616379, "grad_norm": 0.052935000509023666, "learning_rate": 2.9388428421688953e-06, "loss": 0.0009, "step": 164210 }, { "epoch": 2.687065368567455, "grad_norm": 0.043738771229982376, "learning_rate": 2.9381923955737277e-06, "loss": 0.0005, "step": 164220 }, { "epoch": 2.687228994518531, "grad_norm": 0.006149246823042631, "learning_rate": 2.93754199101706e-06, "loss": 0.0005, "step": 164230 }, { "epoch": 2.6873926204696064, "grad_norm": 0.04263273626565933, "learning_rate": 2.9368916285121605e-06, "loss": 0.0007, "step": 164240 }, { "epoch": 2.6875562464206824, "grad_norm": 0.015228747390210629, "learning_rate": 2.936241308072284e-06, "loss": 0.0008, "step": 164250 }, { "epoch": 2.6877198723717584, "grad_norm": 0.02265186421573162, "learning_rate": 2.935591029710693e-06, "loss": 0.0004, "step": 164260 }, { "epoch": 2.687883498322834, "grad_norm": 0.011146222241222858, "learning_rate": 2.9349407934406444e-06, "loss": 0.0008, "step": 164270 }, { "epoch": 2.68804712427391, "grad_norm": 0.08474449813365936, "learning_rate": 2.934290599275398e-06, "loss": 0.0012, "step": 164280 }, { "epoch": 2.688210750224986, "grad_norm": 0.01346262451261282, "learning_rate": 2.9336404472282077e-06, "loss": 0.0004, "step": 164290 }, { "epoch": 2.6883743761760615, "grad_norm": 0.02029644139111042, "learning_rate": 2.9329903373123325e-06, "loss": 0.0011, "step": 164300 }, { "epoch": 2.6885380021271374, "grad_norm": 0.01473666075617075, "learning_rate": 2.932340269541025e-06, "loss": 0.0005, "step": 164310 }, { "epoch": 2.6887016280782134, "grad_norm": 0.012426448054611683, "learning_rate": 2.931690243927543e-06, "loss": 0.0004, "step": 164320 }, { "epoch": 2.688865254029289, "grad_norm": 0.10083427280187607, "learning_rate": 2.9310402604851363e-06, "loss": 0.0015, "step": 164330 }, { "epoch": 2.689028879980365, "grad_norm": 0.092314213514328, "learning_rate": 2.93039031922706e-06, "loss": 0.0007, "step": 164340 }, { "epoch": 2.689192505931441, "grad_norm": 0.06949101388454437, "learning_rate": 2.929740420166564e-06, "loss": 0.0007, "step": 164350 }, { "epoch": 2.6893561318825165, "grad_norm": 0.023393448442220688, "learning_rate": 2.9290905633169015e-06, "loss": 0.0005, "step": 164360 }, { "epoch": 2.6895197578335925, "grad_norm": 0.06564920395612717, "learning_rate": 2.9284407486913197e-06, "loss": 0.0005, "step": 164370 }, { "epoch": 2.689683383784668, "grad_norm": 0.03419800102710724, "learning_rate": 2.9277909763030705e-06, "loss": 0.0004, "step": 164380 }, { "epoch": 2.689847009735744, "grad_norm": 0.032719485461711884, "learning_rate": 2.9271412461654e-06, "loss": 0.0005, "step": 164390 }, { "epoch": 2.69001063568682, "grad_norm": 0.03814246505498886, "learning_rate": 2.9264915582915576e-06, "loss": 0.0005, "step": 164400 }, { "epoch": 2.6901742616378956, "grad_norm": 0.013769661076366901, "learning_rate": 2.9258419126947872e-06, "loss": 0.0007, "step": 164410 }, { "epoch": 2.6903378875889716, "grad_norm": 0.04855554923415184, "learning_rate": 2.9251923093883383e-06, "loss": 0.0007, "step": 164420 }, { "epoch": 2.690501513540047, "grad_norm": 0.08050039410591125, "learning_rate": 2.9245427483854514e-06, "loss": 0.0004, "step": 164430 }, { "epoch": 2.690665139491123, "grad_norm": 0.07795561850070953, "learning_rate": 2.9238932296993747e-06, "loss": 0.0008, "step": 164440 }, { "epoch": 2.690828765442199, "grad_norm": 0.018549203872680664, "learning_rate": 2.923243753343348e-06, "loss": 0.0007, "step": 164450 }, { "epoch": 2.6909923913932747, "grad_norm": 0.06503807008266449, "learning_rate": 2.922594319330616e-06, "loss": 0.0008, "step": 164460 }, { "epoch": 2.6911560173443507, "grad_norm": 0.014569628983736038, "learning_rate": 2.921944927674418e-06, "loss": 0.0004, "step": 164470 }, { "epoch": 2.6913196432954267, "grad_norm": 0.037114936858415604, "learning_rate": 2.921295578387997e-06, "loss": 0.0005, "step": 164480 }, { "epoch": 2.6914832692465023, "grad_norm": 0.07339192926883698, "learning_rate": 2.92064627148459e-06, "loss": 0.0013, "step": 164490 }, { "epoch": 2.6916468951975783, "grad_norm": 0.09450405091047287, "learning_rate": 2.919997006977438e-06, "loss": 0.0013, "step": 164500 }, { "epoch": 2.6918105211486543, "grad_norm": 0.005237665958702564, "learning_rate": 2.919347784879777e-06, "loss": 0.0005, "step": 164510 }, { "epoch": 2.69197414709973, "grad_norm": 0.030681753531098366, "learning_rate": 2.9186986052048473e-06, "loss": 0.0003, "step": 164520 }, { "epoch": 2.692137773050806, "grad_norm": 0.03571700677275658, "learning_rate": 2.9180494679658787e-06, "loss": 0.0005, "step": 164530 }, { "epoch": 2.692301399001882, "grad_norm": 0.020301779732108116, "learning_rate": 2.9174003731761157e-06, "loss": 0.0005, "step": 164540 }, { "epoch": 2.6924650249529574, "grad_norm": 0.05863901600241661, "learning_rate": 2.9167513208487852e-06, "loss": 0.0004, "step": 164550 }, { "epoch": 2.6926286509040334, "grad_norm": 0.055221524089574814, "learning_rate": 2.9161023109971255e-06, "loss": 0.0009, "step": 164560 }, { "epoch": 2.6927922768551094, "grad_norm": 0.023847611621022224, "learning_rate": 2.9154533436343656e-06, "loss": 0.0006, "step": 164570 }, { "epoch": 2.692955902806185, "grad_norm": 0.06476394832134247, "learning_rate": 2.914804418773739e-06, "loss": 0.0006, "step": 164580 }, { "epoch": 2.693119528757261, "grad_norm": 0.05969446152448654, "learning_rate": 2.9141555364284767e-06, "loss": 0.0008, "step": 164590 }, { "epoch": 2.693283154708337, "grad_norm": 0.12560325860977173, "learning_rate": 2.9135066966118113e-06, "loss": 0.0012, "step": 164600 }, { "epoch": 2.6934467806594125, "grad_norm": 0.014650905504822731, "learning_rate": 2.9128578993369683e-06, "loss": 0.0007, "step": 164610 }, { "epoch": 2.6936104066104885, "grad_norm": 0.010839940048754215, "learning_rate": 2.912209144617177e-06, "loss": 0.0009, "step": 164620 }, { "epoch": 2.6937740325615644, "grad_norm": 0.02575855702161789, "learning_rate": 2.9115604324656655e-06, "loss": 0.0006, "step": 164630 }, { "epoch": 2.69393765851264, "grad_norm": 0.030600205063819885, "learning_rate": 2.9109117628956633e-06, "loss": 0.0006, "step": 164640 }, { "epoch": 2.694101284463716, "grad_norm": 0.04214426875114441, "learning_rate": 2.9102631359203916e-06, "loss": 0.0011, "step": 164650 }, { "epoch": 2.694264910414792, "grad_norm": 0.1162208765745163, "learning_rate": 2.9096145515530765e-06, "loss": 0.0008, "step": 164660 }, { "epoch": 2.6944285363658675, "grad_norm": 0.007302908226847649, "learning_rate": 2.908966009806944e-06, "loss": 0.0004, "step": 164670 }, { "epoch": 2.6945921623169435, "grad_norm": 0.02425556257367134, "learning_rate": 2.908317510695217e-06, "loss": 0.0006, "step": 164680 }, { "epoch": 2.6947557882680195, "grad_norm": 0.050846755504608154, "learning_rate": 2.907669054231116e-06, "loss": 0.0005, "step": 164690 }, { "epoch": 2.694919414219095, "grad_norm": 0.05148063227534294, "learning_rate": 2.907020640427863e-06, "loss": 0.0011, "step": 164700 }, { "epoch": 2.695083040170171, "grad_norm": 0.029573258012533188, "learning_rate": 2.906372269298679e-06, "loss": 0.0004, "step": 164710 }, { "epoch": 2.695246666121247, "grad_norm": 0.02761581726372242, "learning_rate": 2.905723940856786e-06, "loss": 0.0004, "step": 164720 }, { "epoch": 2.6954102920723226, "grad_norm": 0.02260592021048069, "learning_rate": 2.905075655115399e-06, "loss": 0.0005, "step": 164730 }, { "epoch": 2.6955739180233986, "grad_norm": 0.05241803452372551, "learning_rate": 2.9044274120877374e-06, "loss": 0.0008, "step": 164740 }, { "epoch": 2.6957375439744746, "grad_norm": 0.04844943806529045, "learning_rate": 2.903779211787019e-06, "loss": 0.0006, "step": 164750 }, { "epoch": 2.69590116992555, "grad_norm": 0.001115773105993867, "learning_rate": 2.9031310542264614e-06, "loss": 0.0007, "step": 164760 }, { "epoch": 2.696064795876626, "grad_norm": 0.11646877974271774, "learning_rate": 2.9024829394192764e-06, "loss": 0.0011, "step": 164770 }, { "epoch": 2.6962284218277017, "grad_norm": 0.047942761331796646, "learning_rate": 2.9018348673786808e-06, "loss": 0.0006, "step": 164780 }, { "epoch": 2.6963920477787777, "grad_norm": 0.0036521167494356632, "learning_rate": 2.901186838117887e-06, "loss": 0.0006, "step": 164790 }, { "epoch": 2.6965556737298537, "grad_norm": 0.00886402279138565, "learning_rate": 2.900538851650111e-06, "loss": 0.0005, "step": 164800 }, { "epoch": 2.6967192996809293, "grad_norm": 0.07477031648159027, "learning_rate": 2.899890907988561e-06, "loss": 0.0011, "step": 164810 }, { "epoch": 2.6968829256320053, "grad_norm": 0.0037287031300365925, "learning_rate": 2.899243007146448e-06, "loss": 0.0006, "step": 164820 }, { "epoch": 2.697046551583081, "grad_norm": 0.030949687585234642, "learning_rate": 2.898595149136984e-06, "loss": 0.0011, "step": 164830 }, { "epoch": 2.697210177534157, "grad_norm": 0.0954134538769722, "learning_rate": 2.89794733397338e-06, "loss": 0.0008, "step": 164840 }, { "epoch": 2.697373803485233, "grad_norm": 0.01922152005136013, "learning_rate": 2.8972995616688394e-06, "loss": 0.0004, "step": 164850 }, { "epoch": 2.6975374294363084, "grad_norm": 0.06238219887018204, "learning_rate": 2.8966518322365732e-06, "loss": 0.0011, "step": 164860 }, { "epoch": 2.6977010553873844, "grad_norm": 0.022463267669081688, "learning_rate": 2.8960041456897868e-06, "loss": 0.0006, "step": 164870 }, { "epoch": 2.6978646813384604, "grad_norm": 0.03923700377345085, "learning_rate": 2.895356502041689e-06, "loss": 0.0004, "step": 164880 }, { "epoch": 2.698028307289536, "grad_norm": 0.09064501523971558, "learning_rate": 2.8947089013054795e-06, "loss": 0.001, "step": 164890 }, { "epoch": 2.698191933240612, "grad_norm": 0.057588979601860046, "learning_rate": 2.894061343494366e-06, "loss": 0.0006, "step": 164900 }, { "epoch": 2.698355559191688, "grad_norm": 0.020327210426330566, "learning_rate": 2.89341382862155e-06, "loss": 0.0005, "step": 164910 }, { "epoch": 2.6985191851427635, "grad_norm": 0.005909640341997147, "learning_rate": 2.8927663567002366e-06, "loss": 0.0007, "step": 164920 }, { "epoch": 2.6986828110938395, "grad_norm": 0.07184895873069763, "learning_rate": 2.8921189277436233e-06, "loss": 0.0017, "step": 164930 }, { "epoch": 2.6988464370449154, "grad_norm": 0.04551311209797859, "learning_rate": 2.891471541764912e-06, "loss": 0.0007, "step": 164940 }, { "epoch": 2.699010062995991, "grad_norm": 0.07727238535881042, "learning_rate": 2.890824198777303e-06, "loss": 0.0012, "step": 164950 }, { "epoch": 2.699173688947067, "grad_norm": 0.027327992022037506, "learning_rate": 2.890176898793996e-06, "loss": 0.0005, "step": 164960 }, { "epoch": 2.699337314898143, "grad_norm": 0.06140276417136192, "learning_rate": 2.8895296418281863e-06, "loss": 0.0009, "step": 164970 }, { "epoch": 2.6995009408492185, "grad_norm": 0.052106037735939026, "learning_rate": 2.888882427893073e-06, "loss": 0.0005, "step": 164980 }, { "epoch": 2.6996645668002945, "grad_norm": 0.07928349822759628, "learning_rate": 2.88823525700185e-06, "loss": 0.0009, "step": 164990 }, { "epoch": 2.6998281927513705, "grad_norm": 0.026867717504501343, "learning_rate": 2.8875881291677177e-06, "loss": 0.0006, "step": 165000 }, { "epoch": 2.699991818702446, "grad_norm": 0.12609755992889404, "learning_rate": 2.8869410444038637e-06, "loss": 0.0005, "step": 165010 }, { "epoch": 2.700155444653522, "grad_norm": 0.005043411161750555, "learning_rate": 2.8862940027234854e-06, "loss": 0.0006, "step": 165020 }, { "epoch": 2.700319070604598, "grad_norm": 0.04799190163612366, "learning_rate": 2.885647004139774e-06, "loss": 0.0009, "step": 165030 }, { "epoch": 2.7004826965556736, "grad_norm": 0.08652093261480331, "learning_rate": 2.885000048665925e-06, "loss": 0.0006, "step": 165040 }, { "epoch": 2.7006463225067496, "grad_norm": 0.0233281459659338, "learning_rate": 2.8843531363151243e-06, "loss": 0.0007, "step": 165050 }, { "epoch": 2.7008099484578256, "grad_norm": 0.032956186681985855, "learning_rate": 2.8837062671005657e-06, "loss": 0.0008, "step": 165060 }, { "epoch": 2.700973574408901, "grad_norm": 0.02263163961470127, "learning_rate": 2.883059441035432e-06, "loss": 0.0008, "step": 165070 }, { "epoch": 2.701137200359977, "grad_norm": 0.06689776480197906, "learning_rate": 2.882412658132919e-06, "loss": 0.0012, "step": 165080 }, { "epoch": 2.701300826311053, "grad_norm": 0.023819567635655403, "learning_rate": 2.8817659184062107e-06, "loss": 0.0006, "step": 165090 }, { "epoch": 2.7014644522621287, "grad_norm": 0.06719936430454254, "learning_rate": 2.8811192218684946e-06, "loss": 0.0005, "step": 165100 }, { "epoch": 2.7016280782132047, "grad_norm": 0.004079843871295452, "learning_rate": 2.8804725685329525e-06, "loss": 0.0004, "step": 165110 }, { "epoch": 2.7017917041642807, "grad_norm": 0.04936707019805908, "learning_rate": 2.879825958412776e-06, "loss": 0.0015, "step": 165120 }, { "epoch": 2.7019553301153563, "grad_norm": 0.04553413391113281, "learning_rate": 2.8791793915211423e-06, "loss": 0.0011, "step": 165130 }, { "epoch": 2.7021189560664323, "grad_norm": 0.05644327029585838, "learning_rate": 2.87853286787124e-06, "loss": 0.0012, "step": 165140 }, { "epoch": 2.702282582017508, "grad_norm": 0.020621763542294502, "learning_rate": 2.8778863874762437e-06, "loss": 0.0005, "step": 165150 }, { "epoch": 2.702446207968584, "grad_norm": 0.07644841820001602, "learning_rate": 2.877239950349343e-06, "loss": 0.0007, "step": 165160 }, { "epoch": 2.70260983391966, "grad_norm": 0.0077191186137497425, "learning_rate": 2.8765935565037127e-06, "loss": 0.0005, "step": 165170 }, { "epoch": 2.7027734598707354, "grad_norm": 0.21816390752792358, "learning_rate": 2.8759472059525363e-06, "loss": 0.0013, "step": 165180 }, { "epoch": 2.7029370858218114, "grad_norm": 0.0256621316075325, "learning_rate": 2.8753008987089855e-06, "loss": 0.0005, "step": 165190 }, { "epoch": 2.703100711772887, "grad_norm": 0.14016889035701752, "learning_rate": 2.874654634786246e-06, "loss": 0.0006, "step": 165200 }, { "epoch": 2.703264337723963, "grad_norm": 0.03444969654083252, "learning_rate": 2.8740084141974893e-06, "loss": 0.0013, "step": 165210 }, { "epoch": 2.703427963675039, "grad_norm": 0.014477293007075787, "learning_rate": 2.8733622369558954e-06, "loss": 0.0009, "step": 165220 }, { "epoch": 2.7035915896261145, "grad_norm": 0.02085966430604458, "learning_rate": 2.8727161030746343e-06, "loss": 0.0008, "step": 165230 }, { "epoch": 2.7037552155771905, "grad_norm": 0.006841568276286125, "learning_rate": 2.8720700125668828e-06, "loss": 0.0005, "step": 165240 }, { "epoch": 2.7039188415282664, "grad_norm": 0.027907729148864746, "learning_rate": 2.8714239654458144e-06, "loss": 0.0011, "step": 165250 }, { "epoch": 2.704082467479342, "grad_norm": 0.011680688709020615, "learning_rate": 2.8707779617246025e-06, "loss": 0.0006, "step": 165260 }, { "epoch": 2.704246093430418, "grad_norm": 0.030009590089321136, "learning_rate": 2.8701320014164147e-06, "loss": 0.0009, "step": 165270 }, { "epoch": 2.704409719381494, "grad_norm": 0.04639974609017372, "learning_rate": 2.8694860845344254e-06, "loss": 0.0007, "step": 165280 }, { "epoch": 2.7045733453325695, "grad_norm": 0.03663249686360359, "learning_rate": 2.8688402110918012e-06, "loss": 0.0007, "step": 165290 }, { "epoch": 2.7047369712836455, "grad_norm": 0.09284599125385284, "learning_rate": 2.8681943811017153e-06, "loss": 0.0014, "step": 165300 }, { "epoch": 2.7049005972347215, "grad_norm": 0.0024301172234117985, "learning_rate": 2.8675485945773317e-06, "loss": 0.0004, "step": 165310 }, { "epoch": 2.705064223185797, "grad_norm": 0.03133191168308258, "learning_rate": 2.866902851531818e-06, "loss": 0.0006, "step": 165320 }, { "epoch": 2.705227849136873, "grad_norm": 0.0210533756762743, "learning_rate": 2.8662571519783435e-06, "loss": 0.0006, "step": 165330 }, { "epoch": 2.705391475087949, "grad_norm": 0.046025924384593964, "learning_rate": 2.8656114959300685e-06, "loss": 0.0004, "step": 165340 }, { "epoch": 2.7055551010390246, "grad_norm": 0.039070334285497665, "learning_rate": 2.8649658834001597e-06, "loss": 0.0012, "step": 165350 }, { "epoch": 2.7057187269901006, "grad_norm": 0.014291632920503616, "learning_rate": 2.8643203144017812e-06, "loss": 0.001, "step": 165360 }, { "epoch": 2.7058823529411766, "grad_norm": 0.005677579902112484, "learning_rate": 2.863674788948097e-06, "loss": 0.0004, "step": 165370 }, { "epoch": 2.706045978892252, "grad_norm": 0.010292907245457172, "learning_rate": 2.8630293070522654e-06, "loss": 0.0014, "step": 165380 }, { "epoch": 2.706209604843328, "grad_norm": 0.010953434742987156, "learning_rate": 2.8623838687274474e-06, "loss": 0.0011, "step": 165390 }, { "epoch": 2.706373230794404, "grad_norm": 0.018817827105522156, "learning_rate": 2.861738473986806e-06, "loss": 0.0005, "step": 165400 }, { "epoch": 2.7065368567454797, "grad_norm": 0.046879176050424576, "learning_rate": 2.8610931228435e-06, "loss": 0.0005, "step": 165410 }, { "epoch": 2.7067004826965557, "grad_norm": 0.05797947198152542, "learning_rate": 2.860447815310684e-06, "loss": 0.0011, "step": 165420 }, { "epoch": 2.7068641086476317, "grad_norm": 0.026288054883480072, "learning_rate": 2.859802551401517e-06, "loss": 0.0005, "step": 165430 }, { "epoch": 2.7070277345987073, "grad_norm": 0.025134315714240074, "learning_rate": 2.8591573311291564e-06, "loss": 0.0007, "step": 165440 }, { "epoch": 2.7071913605497833, "grad_norm": 0.06621838361024857, "learning_rate": 2.858512154506759e-06, "loss": 0.0017, "step": 165450 }, { "epoch": 2.7073549865008593, "grad_norm": 0.02682272344827652, "learning_rate": 2.857867021547476e-06, "loss": 0.0014, "step": 165460 }, { "epoch": 2.707518612451935, "grad_norm": 0.02898399904370308, "learning_rate": 2.8572219322644613e-06, "loss": 0.0007, "step": 165470 }, { "epoch": 2.707682238403011, "grad_norm": 0.09131907671689987, "learning_rate": 2.8565768866708697e-06, "loss": 0.001, "step": 165480 }, { "epoch": 2.707845864354087, "grad_norm": 0.0061539215967059135, "learning_rate": 2.855931884779855e-06, "loss": 0.001, "step": 165490 }, { "epoch": 2.7080094903051624, "grad_norm": 0.06490159034729004, "learning_rate": 2.8552869266045636e-06, "loss": 0.0006, "step": 165500 }, { "epoch": 2.7081731162562384, "grad_norm": 0.02950110100209713, "learning_rate": 2.854642012158149e-06, "loss": 0.0009, "step": 165510 }, { "epoch": 2.7083367422073144, "grad_norm": 0.0337354950606823, "learning_rate": 2.8539971414537558e-06, "loss": 0.0006, "step": 165520 }, { "epoch": 2.70850036815839, "grad_norm": 0.016679251566529274, "learning_rate": 2.8533523145045395e-06, "loss": 0.0007, "step": 165530 }, { "epoch": 2.708663994109466, "grad_norm": 0.026797475293278694, "learning_rate": 2.8527075313236417e-06, "loss": 0.0013, "step": 165540 }, { "epoch": 2.7088276200605415, "grad_norm": 0.040530942380428314, "learning_rate": 2.8520627919242137e-06, "loss": 0.0005, "step": 165550 }, { "epoch": 2.7089912460116174, "grad_norm": 0.010718640871345997, "learning_rate": 2.8514180963193944e-06, "loss": 0.0006, "step": 165560 }, { "epoch": 2.7091548719626934, "grad_norm": 0.0032788242679089308, "learning_rate": 2.8507734445223365e-06, "loss": 0.0003, "step": 165570 }, { "epoch": 2.709318497913769, "grad_norm": 0.0076942783780395985, "learning_rate": 2.8501288365461787e-06, "loss": 0.0009, "step": 165580 }, { "epoch": 2.709482123864845, "grad_norm": 0.06793185323476791, "learning_rate": 2.849484272404067e-06, "loss": 0.0007, "step": 165590 }, { "epoch": 2.7096457498159205, "grad_norm": 0.007078592665493488, "learning_rate": 2.848839752109138e-06, "loss": 0.0003, "step": 165600 }, { "epoch": 2.7098093757669965, "grad_norm": 0.0702027752995491, "learning_rate": 2.848195275674542e-06, "loss": 0.001, "step": 165610 }, { "epoch": 2.7099730017180725, "grad_norm": 0.042302291840314865, "learning_rate": 2.8475508431134115e-06, "loss": 0.0007, "step": 165620 }, { "epoch": 2.710136627669148, "grad_norm": 0.03801511600613594, "learning_rate": 2.8469064544388902e-06, "loss": 0.0003, "step": 165630 }, { "epoch": 2.710300253620224, "grad_norm": 0.05435509607195854, "learning_rate": 2.8462621096641118e-06, "loss": 0.0012, "step": 165640 }, { "epoch": 2.7104638795713, "grad_norm": 0.05445283651351929, "learning_rate": 2.8456178088022212e-06, "loss": 0.0006, "step": 165650 }, { "epoch": 2.7106275055223756, "grad_norm": 0.05191740766167641, "learning_rate": 2.8449735518663497e-06, "loss": 0.0008, "step": 165660 }, { "epoch": 2.7107911314734516, "grad_norm": 0.03594503551721573, "learning_rate": 2.844329338869636e-06, "loss": 0.0004, "step": 165670 }, { "epoch": 2.7109547574245276, "grad_norm": 0.03290218114852905, "learning_rate": 2.843685169825212e-06, "loss": 0.001, "step": 165680 }, { "epoch": 2.711118383375603, "grad_norm": 0.08415141701698303, "learning_rate": 2.8430410447462135e-06, "loss": 0.0006, "step": 165690 }, { "epoch": 2.711282009326679, "grad_norm": 0.0365307554602623, "learning_rate": 2.842396963645774e-06, "loss": 0.0009, "step": 165700 }, { "epoch": 2.711445635277755, "grad_norm": 0.013340596109628677, "learning_rate": 2.841752926537027e-06, "loss": 0.0008, "step": 165710 }, { "epoch": 2.7116092612288307, "grad_norm": 0.006022294517606497, "learning_rate": 2.8411089334330997e-06, "loss": 0.0006, "step": 165720 }, { "epoch": 2.7117728871799067, "grad_norm": 0.04589848592877388, "learning_rate": 2.8404649843471266e-06, "loss": 0.0005, "step": 165730 }, { "epoch": 2.7119365131309827, "grad_norm": 0.044582951813936234, "learning_rate": 2.8398210792922355e-06, "loss": 0.0006, "step": 165740 }, { "epoch": 2.7121001390820583, "grad_norm": 0.0012172366259619594, "learning_rate": 2.8391772182815567e-06, "loss": 0.0007, "step": 165750 }, { "epoch": 2.7122637650331343, "grad_norm": 0.0007321885204873979, "learning_rate": 2.838533401328215e-06, "loss": 0.0018, "step": 165760 }, { "epoch": 2.7124273909842103, "grad_norm": 0.03069896437227726, "learning_rate": 2.837889628445339e-06, "loss": 0.0009, "step": 165770 }, { "epoch": 2.712591016935286, "grad_norm": 0.015872236341238022, "learning_rate": 2.837245899646055e-06, "loss": 0.0004, "step": 165780 }, { "epoch": 2.712754642886362, "grad_norm": 0.0059435442090034485, "learning_rate": 2.836602214943489e-06, "loss": 0.0012, "step": 165790 }, { "epoch": 2.712918268837438, "grad_norm": 0.011265150271356106, "learning_rate": 2.835958574350762e-06, "loss": 0.0009, "step": 165800 }, { "epoch": 2.7130818947885134, "grad_norm": 0.025661058723926544, "learning_rate": 2.835314977881e-06, "loss": 0.0009, "step": 165810 }, { "epoch": 2.7132455207395894, "grad_norm": 0.017597433179616928, "learning_rate": 2.8346714255473245e-06, "loss": 0.0004, "step": 165820 }, { "epoch": 2.7134091466906654, "grad_norm": 0.031454186886548996, "learning_rate": 2.8340279173628586e-06, "loss": 0.0013, "step": 165830 }, { "epoch": 2.713572772641741, "grad_norm": 0.009314668364822865, "learning_rate": 2.8333844533407195e-06, "loss": 0.0008, "step": 165840 }, { "epoch": 2.713736398592817, "grad_norm": 0.037518564611673355, "learning_rate": 2.832741033494029e-06, "loss": 0.0003, "step": 165850 }, { "epoch": 2.713900024543893, "grad_norm": 0.014619317837059498, "learning_rate": 2.8320976578359057e-06, "loss": 0.0023, "step": 165860 }, { "epoch": 2.7140636504949684, "grad_norm": 0.006948438473045826, "learning_rate": 2.831454326379469e-06, "loss": 0.0006, "step": 165870 }, { "epoch": 2.7142272764460444, "grad_norm": 0.08465239405632019, "learning_rate": 2.830811039137833e-06, "loss": 0.0017, "step": 165880 }, { "epoch": 2.7143909023971204, "grad_norm": 0.016431795433163643, "learning_rate": 2.830167796124114e-06, "loss": 0.0004, "step": 165890 }, { "epoch": 2.714554528348196, "grad_norm": 0.025787392631173134, "learning_rate": 2.829524597351429e-06, "loss": 0.0007, "step": 165900 }, { "epoch": 2.714718154299272, "grad_norm": 0.005130861885845661, "learning_rate": 2.828881442832894e-06, "loss": 0.0007, "step": 165910 }, { "epoch": 2.7148817802503475, "grad_norm": 0.040617819875478745, "learning_rate": 2.8282383325816165e-06, "loss": 0.0009, "step": 165920 }, { "epoch": 2.7150454062014235, "grad_norm": 0.00290691782720387, "learning_rate": 2.827595266610713e-06, "loss": 0.0007, "step": 165930 }, { "epoch": 2.7152090321524995, "grad_norm": 0.054643627256155014, "learning_rate": 2.8269522449332954e-06, "loss": 0.0004, "step": 165940 }, { "epoch": 2.715372658103575, "grad_norm": 0.043041884899139404, "learning_rate": 2.826309267562474e-06, "loss": 0.0005, "step": 165950 }, { "epoch": 2.715536284054651, "grad_norm": 0.024211211130023003, "learning_rate": 2.8256663345113567e-06, "loss": 0.0006, "step": 165960 }, { "epoch": 2.7156999100057266, "grad_norm": 0.04453933238983154, "learning_rate": 2.8250234457930535e-06, "loss": 0.0006, "step": 165970 }, { "epoch": 2.7158635359568026, "grad_norm": 0.0877954289317131, "learning_rate": 2.8243806014206732e-06, "loss": 0.001, "step": 165980 }, { "epoch": 2.7160271619078786, "grad_norm": 0.01661178283393383, "learning_rate": 2.823737801407323e-06, "loss": 0.0007, "step": 165990 }, { "epoch": 2.716190787858954, "grad_norm": 0.04375023767352104, "learning_rate": 2.8230950457661064e-06, "loss": 0.0006, "step": 166000 }, { "epoch": 2.71635441381003, "grad_norm": 0.049739256501197815, "learning_rate": 2.822452334510131e-06, "loss": 0.0009, "step": 166010 }, { "epoch": 2.716518039761106, "grad_norm": 0.10348812490701675, "learning_rate": 2.8218096676525002e-06, "loss": 0.0006, "step": 166020 }, { "epoch": 2.7166816657121817, "grad_norm": 0.003953330684453249, "learning_rate": 2.82116704520632e-06, "loss": 0.0004, "step": 166030 }, { "epoch": 2.7168452916632577, "grad_norm": 0.05237279087305069, "learning_rate": 2.8205244671846886e-06, "loss": 0.0007, "step": 166040 }, { "epoch": 2.7170089176143337, "grad_norm": 0.02475372515618801, "learning_rate": 2.81988193360071e-06, "loss": 0.0009, "step": 166050 }, { "epoch": 2.7171725435654093, "grad_norm": 0.029546793550252914, "learning_rate": 2.8192394444674842e-06, "loss": 0.0007, "step": 166060 }, { "epoch": 2.7173361695164853, "grad_norm": 0.09500499814748764, "learning_rate": 2.8185969997981132e-06, "loss": 0.001, "step": 166070 }, { "epoch": 2.7174997954675613, "grad_norm": 0.04419306665658951, "learning_rate": 2.817954599605693e-06, "loss": 0.0004, "step": 166080 }, { "epoch": 2.717663421418637, "grad_norm": 0.09062111377716064, "learning_rate": 2.817312243903322e-06, "loss": 0.0009, "step": 166090 }, { "epoch": 2.717827047369713, "grad_norm": 0.039961133152246475, "learning_rate": 2.816669932704098e-06, "loss": 0.0004, "step": 166100 }, { "epoch": 2.717990673320789, "grad_norm": 0.11677978932857513, "learning_rate": 2.81602766602112e-06, "loss": 0.0008, "step": 166110 }, { "epoch": 2.7181542992718644, "grad_norm": 0.023228881880640984, "learning_rate": 2.8153854438674778e-06, "loss": 0.0005, "step": 166120 }, { "epoch": 2.7183179252229404, "grad_norm": 0.029386451467871666, "learning_rate": 2.8147432662562715e-06, "loss": 0.0008, "step": 166130 }, { "epoch": 2.7184815511740164, "grad_norm": 0.031369537115097046, "learning_rate": 2.8141011332005864e-06, "loss": 0.001, "step": 166140 }, { "epoch": 2.718645177125092, "grad_norm": 0.051093630492687225, "learning_rate": 2.813459044713525e-06, "loss": 0.0008, "step": 166150 }, { "epoch": 2.718808803076168, "grad_norm": 0.12527452409267426, "learning_rate": 2.8128170008081705e-06, "loss": 0.001, "step": 166160 }, { "epoch": 2.718972429027244, "grad_norm": 0.054671596735715866, "learning_rate": 2.81217500149762e-06, "loss": 0.0006, "step": 166170 }, { "epoch": 2.7191360549783195, "grad_norm": 0.06870420277118683, "learning_rate": 2.8115330467949577e-06, "loss": 0.0009, "step": 166180 }, { "epoch": 2.7192996809293954, "grad_norm": 0.059778060764074326, "learning_rate": 2.8108911367132785e-06, "loss": 0.0005, "step": 166190 }, { "epoch": 2.7194633068804714, "grad_norm": 0.052928999066352844, "learning_rate": 2.810249271265665e-06, "loss": 0.0006, "step": 166200 }, { "epoch": 2.719626932831547, "grad_norm": 0.021389024332165718, "learning_rate": 2.8096074504652086e-06, "loss": 0.0004, "step": 166210 }, { "epoch": 2.719790558782623, "grad_norm": 0.02974018268287182, "learning_rate": 2.8089656743249895e-06, "loss": 0.0006, "step": 166220 }, { "epoch": 2.719954184733699, "grad_norm": 0.050773024559020996, "learning_rate": 2.8083239428581012e-06, "loss": 0.0012, "step": 166230 }, { "epoch": 2.7201178106847745, "grad_norm": 0.008187090046703815, "learning_rate": 2.807682256077622e-06, "loss": 0.0005, "step": 166240 }, { "epoch": 2.7202814366358505, "grad_norm": 0.10236234217882156, "learning_rate": 2.807040613996638e-06, "loss": 0.0012, "step": 166250 }, { "epoch": 2.7204450625869265, "grad_norm": 0.024538779631257057, "learning_rate": 2.806399016628227e-06, "loss": 0.0004, "step": 166260 }, { "epoch": 2.720608688538002, "grad_norm": 0.0229912418872118, "learning_rate": 2.8057574639854795e-06, "loss": 0.0009, "step": 166270 }, { "epoch": 2.720772314489078, "grad_norm": 0.023774446919560432, "learning_rate": 2.805115956081469e-06, "loss": 0.0008, "step": 166280 }, { "epoch": 2.720935940440154, "grad_norm": 0.0012641864595934749, "learning_rate": 2.804474492929279e-06, "loss": 0.0003, "step": 166290 }, { "epoch": 2.7210995663912296, "grad_norm": 0.011894380673766136, "learning_rate": 2.803833074541983e-06, "loss": 0.0019, "step": 166300 }, { "epoch": 2.7212631923423056, "grad_norm": 0.0562899224460125, "learning_rate": 2.8031917009326664e-06, "loss": 0.0004, "step": 166310 }, { "epoch": 2.721426818293381, "grad_norm": 0.014598535373806953, "learning_rate": 2.802550372114401e-06, "loss": 0.0012, "step": 166320 }, { "epoch": 2.721590444244457, "grad_norm": 0.04805971682071686, "learning_rate": 2.8019090881002664e-06, "loss": 0.0009, "step": 166330 }, { "epoch": 2.721754070195533, "grad_norm": 0.0021457308903336525, "learning_rate": 2.8012678489033352e-06, "loss": 0.001, "step": 166340 }, { "epoch": 2.7219176961466087, "grad_norm": 0.0843704417347908, "learning_rate": 2.800626654536681e-06, "loss": 0.0013, "step": 166350 }, { "epoch": 2.7220813220976847, "grad_norm": 0.037912026047706604, "learning_rate": 2.79998550501338e-06, "loss": 0.0005, "step": 166360 }, { "epoch": 2.7222449480487603, "grad_norm": 0.0031462376937270164, "learning_rate": 2.7993444003465052e-06, "loss": 0.0006, "step": 166370 }, { "epoch": 2.7224085739998363, "grad_norm": 0.025206100195646286, "learning_rate": 2.798703340549124e-06, "loss": 0.0008, "step": 166380 }, { "epoch": 2.7225721999509123, "grad_norm": 0.07733476907014847, "learning_rate": 2.7980623256343103e-06, "loss": 0.0012, "step": 166390 }, { "epoch": 2.722735825901988, "grad_norm": 0.03988654538989067, "learning_rate": 2.797421355615133e-06, "loss": 0.001, "step": 166400 }, { "epoch": 2.722899451853064, "grad_norm": 0.028709741309285164, "learning_rate": 2.7967804305046627e-06, "loss": 0.001, "step": 166410 }, { "epoch": 2.72306307780414, "grad_norm": 0.20386254787445068, "learning_rate": 2.796139550315964e-06, "loss": 0.0009, "step": 166420 }, { "epoch": 2.7232267037552154, "grad_norm": 0.02670806460082531, "learning_rate": 2.7954987150621047e-06, "loss": 0.0007, "step": 166430 }, { "epoch": 2.7233903297062914, "grad_norm": 0.04744323343038559, "learning_rate": 2.7948579247561526e-06, "loss": 0.0005, "step": 166440 }, { "epoch": 2.7235539556573674, "grad_norm": 0.02345854602754116, "learning_rate": 2.794217179411174e-06, "loss": 0.0012, "step": 166450 }, { "epoch": 2.723717581608443, "grad_norm": 0.0021632183343172073, "learning_rate": 2.7935764790402287e-06, "loss": 0.0006, "step": 166460 }, { "epoch": 2.723881207559519, "grad_norm": 0.035748206079006195, "learning_rate": 2.792935823656383e-06, "loss": 0.0012, "step": 166470 }, { "epoch": 2.724044833510595, "grad_norm": 0.00866817869246006, "learning_rate": 2.7922952132726983e-06, "loss": 0.0006, "step": 166480 }, { "epoch": 2.7242084594616705, "grad_norm": 0.0017256353748962283, "learning_rate": 2.791654647902239e-06, "loss": 0.0011, "step": 166490 }, { "epoch": 2.7243720854127464, "grad_norm": 0.03083924576640129, "learning_rate": 2.791014127558061e-06, "loss": 0.0005, "step": 166500 }, { "epoch": 2.7245357113638224, "grad_norm": 0.030339760705828667, "learning_rate": 2.7903736522532266e-06, "loss": 0.0004, "step": 166510 }, { "epoch": 2.724699337314898, "grad_norm": 0.0021111243404448032, "learning_rate": 2.789733222000794e-06, "loss": 0.0009, "step": 166520 }, { "epoch": 2.724862963265974, "grad_norm": 0.0026621809229254723, "learning_rate": 2.7890928368138233e-06, "loss": 0.0012, "step": 166530 }, { "epoch": 2.72502658921705, "grad_norm": 0.05105732008814812, "learning_rate": 2.7884524967053672e-06, "loss": 0.0011, "step": 166540 }, { "epoch": 2.7251902151681255, "grad_norm": 0.051150090992450714, "learning_rate": 2.7878122016884835e-06, "loss": 0.0017, "step": 166550 }, { "epoch": 2.7253538411192015, "grad_norm": 0.057660844177007675, "learning_rate": 2.7871719517762284e-06, "loss": 0.001, "step": 166560 }, { "epoch": 2.7255174670702775, "grad_norm": 0.011081589385867119, "learning_rate": 2.7865317469816566e-06, "loss": 0.0004, "step": 166570 }, { "epoch": 2.725681093021353, "grad_norm": 0.00431025680154562, "learning_rate": 2.785891587317818e-06, "loss": 0.0006, "step": 166580 }, { "epoch": 2.725844718972429, "grad_norm": 0.0779186561703682, "learning_rate": 2.7852514727977674e-06, "loss": 0.001, "step": 166590 }, { "epoch": 2.726008344923505, "grad_norm": 0.1086762472987175, "learning_rate": 2.7846114034345562e-06, "loss": 0.0008, "step": 166600 }, { "epoch": 2.7261719708745806, "grad_norm": 0.027817286550998688, "learning_rate": 2.7839713792412354e-06, "loss": 0.0011, "step": 166610 }, { "epoch": 2.7263355968256566, "grad_norm": 0.016896650195121765, "learning_rate": 2.783331400230852e-06, "loss": 0.0003, "step": 166620 }, { "epoch": 2.7264992227767326, "grad_norm": 0.008028214797377586, "learning_rate": 2.7826914664164554e-06, "loss": 0.0006, "step": 166630 }, { "epoch": 2.726662848727808, "grad_norm": 0.03235195204615593, "learning_rate": 2.7820515778110946e-06, "loss": 0.0004, "step": 166640 }, { "epoch": 2.726826474678884, "grad_norm": 0.0495058111846447, "learning_rate": 2.781411734427818e-06, "loss": 0.0006, "step": 166650 }, { "epoch": 2.72699010062996, "grad_norm": 0.06035605072975159, "learning_rate": 2.780771936279667e-06, "loss": 0.0004, "step": 166660 }, { "epoch": 2.7271537265810357, "grad_norm": 0.025401858612895012, "learning_rate": 2.780132183379689e-06, "loss": 0.001, "step": 166670 }, { "epoch": 2.7273173525321117, "grad_norm": 0.009452990256249905, "learning_rate": 2.7794924757409282e-06, "loss": 0.0005, "step": 166680 }, { "epoch": 2.7274809784831873, "grad_norm": 0.29473650455474854, "learning_rate": 2.7788528133764282e-06, "loss": 0.002, "step": 166690 }, { "epoch": 2.7276446044342633, "grad_norm": 0.02833358198404312, "learning_rate": 2.778213196299229e-06, "loss": 0.0008, "step": 166700 }, { "epoch": 2.7278082303853393, "grad_norm": 0.05508801341056824, "learning_rate": 2.777573624522374e-06, "loss": 0.0012, "step": 166710 }, { "epoch": 2.727971856336415, "grad_norm": 0.009261352010071278, "learning_rate": 2.776934098058901e-06, "loss": 0.0006, "step": 166720 }, { "epoch": 2.728135482287491, "grad_norm": 0.02036195807158947, "learning_rate": 2.7762946169218543e-06, "loss": 0.0008, "step": 166730 }, { "epoch": 2.7282991082385664, "grad_norm": 0.02804555371403694, "learning_rate": 2.7756551811242673e-06, "loss": 0.0006, "step": 166740 }, { "epoch": 2.7284627341896424, "grad_norm": 0.06515271216630936, "learning_rate": 2.7750157906791785e-06, "loss": 0.0007, "step": 166750 }, { "epoch": 2.7286263601407184, "grad_norm": 0.0763256698846817, "learning_rate": 2.7743764455996257e-06, "loss": 0.0029, "step": 166760 }, { "epoch": 2.728789986091794, "grad_norm": 0.06083330884575844, "learning_rate": 2.7737371458986467e-06, "loss": 0.001, "step": 166770 }, { "epoch": 2.72895361204287, "grad_norm": 0.02156876027584076, "learning_rate": 2.773097891589272e-06, "loss": 0.0004, "step": 166780 }, { "epoch": 2.729117237993946, "grad_norm": 0.06739268451929092, "learning_rate": 2.772458682684539e-06, "loss": 0.0006, "step": 166790 }, { "epoch": 2.7292808639450215, "grad_norm": 0.004175321198999882, "learning_rate": 2.771819519197475e-06, "loss": 0.0007, "step": 166800 }, { "epoch": 2.7294444898960974, "grad_norm": 0.056945960968732834, "learning_rate": 2.77118040114112e-06, "loss": 0.0008, "step": 166810 }, { "epoch": 2.7296081158471734, "grad_norm": 0.009639794006943703, "learning_rate": 2.7705413285285e-06, "loss": 0.0007, "step": 166820 }, { "epoch": 2.729771741798249, "grad_norm": 0.056350816041231155, "learning_rate": 2.769902301372647e-06, "loss": 0.0004, "step": 166830 }, { "epoch": 2.729935367749325, "grad_norm": 0.007933767512440681, "learning_rate": 2.7692633196865858e-06, "loss": 0.0008, "step": 166840 }, { "epoch": 2.730098993700401, "grad_norm": 0.0017755364533513784, "learning_rate": 2.7686243834833527e-06, "loss": 0.0005, "step": 166850 }, { "epoch": 2.7302626196514765, "grad_norm": 0.03141481801867485, "learning_rate": 2.7679854927759685e-06, "loss": 0.0009, "step": 166860 }, { "epoch": 2.7304262456025525, "grad_norm": 0.06567896157503128, "learning_rate": 2.7673466475774635e-06, "loss": 0.0009, "step": 166870 }, { "epoch": 2.7305898715536285, "grad_norm": 0.03861453756690025, "learning_rate": 2.766707847900858e-06, "loss": 0.0005, "step": 166880 }, { "epoch": 2.730753497504704, "grad_norm": 0.07808706909418106, "learning_rate": 2.766069093759184e-06, "loss": 0.001, "step": 166890 }, { "epoch": 2.73091712345578, "grad_norm": 0.04832301661372185, "learning_rate": 2.7654303851654597e-06, "loss": 0.0006, "step": 166900 }, { "epoch": 2.731080749406856, "grad_norm": 0.05171757563948631, "learning_rate": 2.764791722132711e-06, "loss": 0.0013, "step": 166910 }, { "epoch": 2.7312443753579316, "grad_norm": 0.012968882918357849, "learning_rate": 2.764153104673954e-06, "loss": 0.0006, "step": 166920 }, { "epoch": 2.7314080013090076, "grad_norm": 0.05304248258471489, "learning_rate": 2.7635145328022174e-06, "loss": 0.0008, "step": 166930 }, { "epoch": 2.7315716272600836, "grad_norm": 0.11094836890697479, "learning_rate": 2.762876006530516e-06, "loss": 0.0011, "step": 166940 }, { "epoch": 2.731735253211159, "grad_norm": 0.039190623909235, "learning_rate": 2.762237525871872e-06, "loss": 0.0007, "step": 166950 }, { "epoch": 2.731898879162235, "grad_norm": 0.03533707559108734, "learning_rate": 2.7615990908393e-06, "loss": 0.0005, "step": 166960 }, { "epoch": 2.732062505113311, "grad_norm": 0.04631258547306061, "learning_rate": 2.7609607014458184e-06, "loss": 0.0006, "step": 166970 }, { "epoch": 2.7322261310643867, "grad_norm": 0.0015136670554056764, "learning_rate": 2.7603223577044436e-06, "loss": 0.001, "step": 166980 }, { "epoch": 2.7323897570154627, "grad_norm": 0.003290256718173623, "learning_rate": 2.759684059628194e-06, "loss": 0.0006, "step": 166990 }, { "epoch": 2.7325533829665387, "grad_norm": 0.04585360363125801, "learning_rate": 2.7590458072300787e-06, "loss": 0.0008, "step": 167000 }, { "epoch": 2.7327170089176143, "grad_norm": 0.03209371492266655, "learning_rate": 2.758407600523113e-06, "loss": 0.0005, "step": 167010 }, { "epoch": 2.7328806348686903, "grad_norm": 0.03819003328680992, "learning_rate": 2.757769439520311e-06, "loss": 0.0007, "step": 167020 }, { "epoch": 2.7330442608197663, "grad_norm": 0.023456217721104622, "learning_rate": 2.7571313242346844e-06, "loss": 0.0006, "step": 167030 }, { "epoch": 2.733207886770842, "grad_norm": 0.012006373144686222, "learning_rate": 2.7564932546792413e-06, "loss": 0.0008, "step": 167040 }, { "epoch": 2.733371512721918, "grad_norm": 0.03393910452723503, "learning_rate": 2.7558552308669918e-06, "loss": 0.0004, "step": 167050 }, { "epoch": 2.733535138672994, "grad_norm": 0.03170425072312355, "learning_rate": 2.7552172528109458e-06, "loss": 0.0003, "step": 167060 }, { "epoch": 2.7336987646240694, "grad_norm": 0.020785696804523468, "learning_rate": 2.7545793205241133e-06, "loss": 0.0009, "step": 167070 }, { "epoch": 2.7338623905751454, "grad_norm": 0.0011006585555151105, "learning_rate": 2.7539414340194963e-06, "loss": 0.001, "step": 167080 }, { "epoch": 2.734026016526221, "grad_norm": 0.04160745441913605, "learning_rate": 2.7533035933101036e-06, "loss": 0.001, "step": 167090 }, { "epoch": 2.734189642477297, "grad_norm": 0.027462270110845566, "learning_rate": 2.75266579840894e-06, "loss": 0.0013, "step": 167100 }, { "epoch": 2.734353268428373, "grad_norm": 0.06136393919587135, "learning_rate": 2.752028049329011e-06, "loss": 0.0005, "step": 167110 }, { "epoch": 2.7345168943794484, "grad_norm": 0.055424727499485016, "learning_rate": 2.751390346083317e-06, "loss": 0.0004, "step": 167120 }, { "epoch": 2.7346805203305244, "grad_norm": 0.010362175293266773, "learning_rate": 2.7507526886848608e-06, "loss": 0.0034, "step": 167130 }, { "epoch": 2.7348441462816, "grad_norm": 0.06980781257152557, "learning_rate": 2.7501150771466444e-06, "loss": 0.0009, "step": 167140 }, { "epoch": 2.735007772232676, "grad_norm": 0.03248542547225952, "learning_rate": 2.7494775114816703e-06, "loss": 0.0003, "step": 167150 }, { "epoch": 2.735171398183752, "grad_norm": 0.017595607787370682, "learning_rate": 2.748839991702934e-06, "loss": 0.0007, "step": 167160 }, { "epoch": 2.7353350241348275, "grad_norm": 0.02637130580842495, "learning_rate": 2.748202517823435e-06, "loss": 0.0006, "step": 167170 }, { "epoch": 2.7354986500859035, "grad_norm": 0.03211935982108116, "learning_rate": 2.7475650898561725e-06, "loss": 0.0009, "step": 167180 }, { "epoch": 2.7356622760369795, "grad_norm": 0.021877672523260117, "learning_rate": 2.7469277078141432e-06, "loss": 0.0005, "step": 167190 }, { "epoch": 2.735825901988055, "grad_norm": 0.03087650239467621, "learning_rate": 2.7462903717103407e-06, "loss": 0.0007, "step": 167200 }, { "epoch": 2.735989527939131, "grad_norm": 0.004178194794803858, "learning_rate": 2.7456530815577604e-06, "loss": 0.0004, "step": 167210 }, { "epoch": 2.736153153890207, "grad_norm": 0.02539634145796299, "learning_rate": 2.7450158373693963e-06, "loss": 0.0008, "step": 167220 }, { "epoch": 2.7363167798412826, "grad_norm": 0.012742758728563786, "learning_rate": 2.744378639158244e-06, "loss": 0.0004, "step": 167230 }, { "epoch": 2.7364804057923586, "grad_norm": 0.0028902129270136356, "learning_rate": 2.7437414869372903e-06, "loss": 0.0009, "step": 167240 }, { "epoch": 2.7366440317434346, "grad_norm": 0.17072054743766785, "learning_rate": 2.743104380719529e-06, "loss": 0.0009, "step": 167250 }, { "epoch": 2.73680765769451, "grad_norm": 0.03177541121840477, "learning_rate": 2.7424673205179497e-06, "loss": 0.0011, "step": 167260 }, { "epoch": 2.736971283645586, "grad_norm": 0.012905343435704708, "learning_rate": 2.7418303063455443e-06, "loss": 0.0011, "step": 167270 }, { "epoch": 2.737134909596662, "grad_norm": 0.09665924310684204, "learning_rate": 2.7411933382152966e-06, "loss": 0.0011, "step": 167280 }, { "epoch": 2.7372985355477377, "grad_norm": 0.042998459190130234, "learning_rate": 2.7405564161401953e-06, "loss": 0.0008, "step": 167290 }, { "epoch": 2.7374621614988137, "grad_norm": 0.05583907291293144, "learning_rate": 2.739919540133227e-06, "loss": 0.0013, "step": 167300 }, { "epoch": 2.7376257874498897, "grad_norm": 0.01668401248753071, "learning_rate": 2.7392827102073794e-06, "loss": 0.0006, "step": 167310 }, { "epoch": 2.7377894134009653, "grad_norm": 0.07921160757541656, "learning_rate": 2.7386459263756327e-06, "loss": 0.0009, "step": 167320 }, { "epoch": 2.7379530393520413, "grad_norm": 0.02184726670384407, "learning_rate": 2.7380091886509717e-06, "loss": 0.0005, "step": 167330 }, { "epoch": 2.7381166653031173, "grad_norm": 0.14494521915912628, "learning_rate": 2.7373724970463805e-06, "loss": 0.0015, "step": 167340 }, { "epoch": 2.738280291254193, "grad_norm": 0.004872411955147982, "learning_rate": 2.736735851574841e-06, "loss": 0.0004, "step": 167350 }, { "epoch": 2.738443917205269, "grad_norm": 0.07120804488658905, "learning_rate": 2.736099252249331e-06, "loss": 0.0006, "step": 167360 }, { "epoch": 2.738607543156345, "grad_norm": 0.04336120933294296, "learning_rate": 2.7354626990828324e-06, "loss": 0.0005, "step": 167370 }, { "epoch": 2.7387711691074204, "grad_norm": 0.09462010860443115, "learning_rate": 2.734826192088323e-06, "loss": 0.0011, "step": 167380 }, { "epoch": 2.7389347950584964, "grad_norm": 0.04100972041487694, "learning_rate": 2.734189731278783e-06, "loss": 0.0006, "step": 167390 }, { "epoch": 2.7390984210095723, "grad_norm": 0.03783899545669556, "learning_rate": 2.733553316667186e-06, "loss": 0.0003, "step": 167400 }, { "epoch": 2.739262046960648, "grad_norm": 0.030759628862142563, "learning_rate": 2.732916948266508e-06, "loss": 0.0011, "step": 167410 }, { "epoch": 2.739425672911724, "grad_norm": 0.005691247526556253, "learning_rate": 2.732280626089726e-06, "loss": 0.0007, "step": 167420 }, { "epoch": 2.7395892988628, "grad_norm": 0.05959920957684517, "learning_rate": 2.731644350149816e-06, "loss": 0.0012, "step": 167430 }, { "epoch": 2.7397529248138754, "grad_norm": 0.009655575267970562, "learning_rate": 2.731008120459746e-06, "loss": 0.0008, "step": 167440 }, { "epoch": 2.7399165507649514, "grad_norm": 0.024534612894058228, "learning_rate": 2.730371937032492e-06, "loss": 0.0007, "step": 167450 }, { "epoch": 2.7400801767160274, "grad_norm": 0.061405472457408905, "learning_rate": 2.7297357998810212e-06, "loss": 0.0007, "step": 167460 }, { "epoch": 2.740243802667103, "grad_norm": 0.08272659033536911, "learning_rate": 2.7290997090183103e-06, "loss": 0.0003, "step": 167470 }, { "epoch": 2.740407428618179, "grad_norm": 0.002179832896217704, "learning_rate": 2.7284636644573225e-06, "loss": 0.0013, "step": 167480 }, { "epoch": 2.7405710545692545, "grad_norm": 0.03950531780719757, "learning_rate": 2.7278276662110316e-06, "loss": 0.001, "step": 167490 }, { "epoch": 2.7407346805203305, "grad_norm": 0.006880322005599737, "learning_rate": 2.7271917142923976e-06, "loss": 0.0006, "step": 167500 }, { "epoch": 2.7408983064714065, "grad_norm": 0.03283172473311424, "learning_rate": 2.726555808714395e-06, "loss": 0.0004, "step": 167510 }, { "epoch": 2.741061932422482, "grad_norm": 0.03058423474431038, "learning_rate": 2.7259199494899847e-06, "loss": 0.0005, "step": 167520 }, { "epoch": 2.741225558373558, "grad_norm": 0.1151931956410408, "learning_rate": 2.725284136632135e-06, "loss": 0.0009, "step": 167530 }, { "epoch": 2.7413891843246336, "grad_norm": 0.014016378670930862, "learning_rate": 2.724648370153802e-06, "loss": 0.0005, "step": 167540 }, { "epoch": 2.7415528102757096, "grad_norm": 0.0008836164488457143, "learning_rate": 2.7240126500679588e-06, "loss": 0.0006, "step": 167550 }, { "epoch": 2.7417164362267856, "grad_norm": 0.041357867419719696, "learning_rate": 2.7233769763875595e-06, "loss": 0.0004, "step": 167560 }, { "epoch": 2.741880062177861, "grad_norm": 0.03767373040318489, "learning_rate": 2.7227413491255694e-06, "loss": 0.0007, "step": 167570 }, { "epoch": 2.742043688128937, "grad_norm": 0.06461050361394882, "learning_rate": 2.7221057682949427e-06, "loss": 0.001, "step": 167580 }, { "epoch": 2.742207314080013, "grad_norm": 0.01289092842489481, "learning_rate": 2.7214702339086453e-06, "loss": 0.0009, "step": 167590 }, { "epoch": 2.7423709400310887, "grad_norm": 0.028622323647141457, "learning_rate": 2.7208347459796304e-06, "loss": 0.0009, "step": 167600 }, { "epoch": 2.7425345659821647, "grad_norm": 0.03528197482228279, "learning_rate": 2.720199304520859e-06, "loss": 0.0009, "step": 167610 }, { "epoch": 2.7426981919332407, "grad_norm": 0.05223884806036949, "learning_rate": 2.719563909545283e-06, "loss": 0.0009, "step": 167620 }, { "epoch": 2.7428618178843163, "grad_norm": 0.024151330813765526, "learning_rate": 2.718928561065859e-06, "loss": 0.001, "step": 167630 }, { "epoch": 2.7430254438353923, "grad_norm": 0.03856680914759636, "learning_rate": 2.718293259095542e-06, "loss": 0.0005, "step": 167640 }, { "epoch": 2.7431890697864683, "grad_norm": 0.03632349148392677, "learning_rate": 2.7176580036472865e-06, "loss": 0.0011, "step": 167650 }, { "epoch": 2.743352695737544, "grad_norm": 0.05167586728930473, "learning_rate": 2.7170227947340416e-06, "loss": 0.0007, "step": 167660 }, { "epoch": 2.74351632168862, "grad_norm": 0.03384684771299362, "learning_rate": 2.71638763236876e-06, "loss": 0.0004, "step": 167670 }, { "epoch": 2.743679947639696, "grad_norm": 0.001253089401870966, "learning_rate": 2.715752516564393e-06, "loss": 0.0007, "step": 167680 }, { "epoch": 2.7438435735907714, "grad_norm": 0.0639883279800415, "learning_rate": 2.7151174473338914e-06, "loss": 0.0007, "step": 167690 }, { "epoch": 2.7440071995418474, "grad_norm": 0.002099810168147087, "learning_rate": 2.714482424690199e-06, "loss": 0.0006, "step": 167700 }, { "epoch": 2.7441708254929233, "grad_norm": 0.02525625191628933, "learning_rate": 2.7138474486462674e-06, "loss": 0.0007, "step": 167710 }, { "epoch": 2.744334451443999, "grad_norm": 0.047144815325737, "learning_rate": 2.713212519215041e-06, "loss": 0.001, "step": 167720 }, { "epoch": 2.744498077395075, "grad_norm": 0.030819740146398544, "learning_rate": 2.712577636409469e-06, "loss": 0.0013, "step": 167730 }, { "epoch": 2.744661703346151, "grad_norm": 0.08149080723524094, "learning_rate": 2.711942800242492e-06, "loss": 0.0007, "step": 167740 }, { "epoch": 2.7448253292972264, "grad_norm": 0.041584283113479614, "learning_rate": 2.7113080107270544e-06, "loss": 0.0005, "step": 167750 }, { "epoch": 2.7449889552483024, "grad_norm": 0.04108448326587677, "learning_rate": 2.7106732678761004e-06, "loss": 0.0009, "step": 167760 }, { "epoch": 2.7451525811993784, "grad_norm": 0.045441266149282455, "learning_rate": 2.7100385717025733e-06, "loss": 0.0009, "step": 167770 }, { "epoch": 2.745316207150454, "grad_norm": 0.03252951800823212, "learning_rate": 2.7094039222194106e-06, "loss": 0.001, "step": 167780 }, { "epoch": 2.74547983310153, "grad_norm": 0.07995085418224335, "learning_rate": 2.7087693194395538e-06, "loss": 0.0007, "step": 167790 }, { "epoch": 2.745643459052606, "grad_norm": 0.06793417036533356, "learning_rate": 2.708134763375942e-06, "loss": 0.0006, "step": 167800 }, { "epoch": 2.7458070850036815, "grad_norm": 0.029026906937360764, "learning_rate": 2.707500254041515e-06, "loss": 0.0007, "step": 167810 }, { "epoch": 2.7459707109547575, "grad_norm": 0.040278054773807526, "learning_rate": 2.7068657914492057e-06, "loss": 0.0007, "step": 167820 }, { "epoch": 2.7461343369058335, "grad_norm": 0.01151775848120451, "learning_rate": 2.7062313756119534e-06, "loss": 0.0005, "step": 167830 }, { "epoch": 2.746297962856909, "grad_norm": 0.05978390946984291, "learning_rate": 2.7055970065426922e-06, "loss": 0.0006, "step": 167840 }, { "epoch": 2.746461588807985, "grad_norm": 0.030366644263267517, "learning_rate": 2.704962684254359e-06, "loss": 0.0005, "step": 167850 }, { "epoch": 2.7466252147590606, "grad_norm": 0.049475520849227905, "learning_rate": 2.7043284087598824e-06, "loss": 0.0008, "step": 167860 }, { "epoch": 2.7467888407101366, "grad_norm": 0.010718319565057755, "learning_rate": 2.703694180072198e-06, "loss": 0.0014, "step": 167870 }, { "epoch": 2.7469524666612126, "grad_norm": 0.0849786028265953, "learning_rate": 2.7030599982042357e-06, "loss": 0.0005, "step": 167880 }, { "epoch": 2.747116092612288, "grad_norm": 0.03895796090364456, "learning_rate": 2.702425863168928e-06, "loss": 0.001, "step": 167890 }, { "epoch": 2.747279718563364, "grad_norm": 0.008569559082388878, "learning_rate": 2.7017917749792016e-06, "loss": 0.0009, "step": 167900 }, { "epoch": 2.7474433445144397, "grad_norm": 0.08176606148481369, "learning_rate": 2.7011577336479865e-06, "loss": 0.0009, "step": 167910 }, { "epoch": 2.7476069704655157, "grad_norm": 0.008553748019039631, "learning_rate": 2.70052373918821e-06, "loss": 0.0004, "step": 167920 }, { "epoch": 2.7477705964165917, "grad_norm": 0.03813775256276131, "learning_rate": 2.699889791612801e-06, "loss": 0.0004, "step": 167930 }, { "epoch": 2.7479342223676673, "grad_norm": 0.018501583486795425, "learning_rate": 2.69925589093468e-06, "loss": 0.0013, "step": 167940 }, { "epoch": 2.7480978483187433, "grad_norm": 0.090116485953331, "learning_rate": 2.6986220371667754e-06, "loss": 0.0006, "step": 167950 }, { "epoch": 2.7482614742698193, "grad_norm": 0.06594321131706238, "learning_rate": 2.69798823032201e-06, "loss": 0.001, "step": 167960 }, { "epoch": 2.748425100220895, "grad_norm": 0.0018261494114995003, "learning_rate": 2.6973544704133093e-06, "loss": 0.0005, "step": 167970 }, { "epoch": 2.748588726171971, "grad_norm": 0.026513729244470596, "learning_rate": 2.69672075745359e-06, "loss": 0.0013, "step": 167980 }, { "epoch": 2.748752352123047, "grad_norm": 0.01176949217915535, "learning_rate": 2.696087091455776e-06, "loss": 0.0004, "step": 167990 }, { "epoch": 2.7489159780741224, "grad_norm": 0.043709054589271545, "learning_rate": 2.6954534724327865e-06, "loss": 0.0006, "step": 168000 }, { "epoch": 2.7490796040251984, "grad_norm": 0.01771932654082775, "learning_rate": 2.694819900397543e-06, "loss": 0.001, "step": 168010 }, { "epoch": 2.7492432299762743, "grad_norm": 0.09072281420230865, "learning_rate": 2.6941863753629593e-06, "loss": 0.0011, "step": 168020 }, { "epoch": 2.74940685592735, "grad_norm": 0.07370568066835403, "learning_rate": 2.693552897341954e-06, "loss": 0.0008, "step": 168030 }, { "epoch": 2.749570481878426, "grad_norm": 0.018820835277438164, "learning_rate": 2.6929194663474438e-06, "loss": 0.0009, "step": 168040 }, { "epoch": 2.749734107829502, "grad_norm": 0.00295297265984118, "learning_rate": 2.692286082392346e-06, "loss": 0.0004, "step": 168050 }, { "epoch": 2.7498977337805774, "grad_norm": 0.013194048777222633, "learning_rate": 2.691652745489569e-06, "loss": 0.0011, "step": 168060 }, { "epoch": 2.7500613597316534, "grad_norm": 0.027823908254504204, "learning_rate": 2.691019455652033e-06, "loss": 0.0005, "step": 168070 }, { "epoch": 2.7502249856827294, "grad_norm": 0.05017206072807312, "learning_rate": 2.6903862128926415e-06, "loss": 0.0008, "step": 168080 }, { "epoch": 2.750388611633805, "grad_norm": 0.05112279951572418, "learning_rate": 2.6897530172243153e-06, "loss": 0.0005, "step": 168090 }, { "epoch": 2.750552237584881, "grad_norm": 0.040586601942777634, "learning_rate": 2.6891198686599584e-06, "loss": 0.0009, "step": 168100 }, { "epoch": 2.750715863535957, "grad_norm": 0.0005864578415639699, "learning_rate": 2.688486767212484e-06, "loss": 0.0004, "step": 168110 }, { "epoch": 2.7508794894870325, "grad_norm": 0.005730462726205587, "learning_rate": 2.687853712894794e-06, "loss": 0.0005, "step": 168120 }, { "epoch": 2.7510431154381085, "grad_norm": 0.020875098183751106, "learning_rate": 2.6872207057198047e-06, "loss": 0.0007, "step": 168130 }, { "epoch": 2.7512067413891845, "grad_norm": 0.038874752819538116, "learning_rate": 2.6865877457004164e-06, "loss": 0.0006, "step": 168140 }, { "epoch": 2.75137036734026, "grad_norm": 0.0400838740170002, "learning_rate": 2.6859548328495383e-06, "loss": 0.0011, "step": 168150 }, { "epoch": 2.751533993291336, "grad_norm": 0.1101117804646492, "learning_rate": 2.685321967180069e-06, "loss": 0.0013, "step": 168160 }, { "epoch": 2.751697619242412, "grad_norm": 0.033400435000658035, "learning_rate": 2.684689148704921e-06, "loss": 0.0006, "step": 168170 }, { "epoch": 2.7518612451934876, "grad_norm": 0.02969910390675068, "learning_rate": 2.684056377436989e-06, "loss": 0.0005, "step": 168180 }, { "epoch": 2.7520248711445636, "grad_norm": 0.028031356632709503, "learning_rate": 2.6834236533891805e-06, "loss": 0.0005, "step": 168190 }, { "epoch": 2.7521884970956396, "grad_norm": 0.0289714764803648, "learning_rate": 2.68279097657439e-06, "loss": 0.0006, "step": 168200 }, { "epoch": 2.752352123046715, "grad_norm": 0.022031016647815704, "learning_rate": 2.6821583470055236e-06, "loss": 0.0004, "step": 168210 }, { "epoch": 2.752515748997791, "grad_norm": 0.04309408739209175, "learning_rate": 2.6815257646954762e-06, "loss": 0.0007, "step": 168220 }, { "epoch": 2.752679374948867, "grad_norm": 0.020453454926609993, "learning_rate": 2.6808932296571484e-06, "loss": 0.0003, "step": 168230 }, { "epoch": 2.7528430008999427, "grad_norm": 0.03630562871694565, "learning_rate": 2.680260741903431e-06, "loss": 0.0006, "step": 168240 }, { "epoch": 2.7530066268510187, "grad_norm": 0.025767652317881584, "learning_rate": 2.6796283014472285e-06, "loss": 0.0006, "step": 168250 }, { "epoch": 2.7531702528020943, "grad_norm": 0.10972534865140915, "learning_rate": 2.6789959083014295e-06, "loss": 0.0005, "step": 168260 }, { "epoch": 2.7533338787531703, "grad_norm": 0.00404918659478426, "learning_rate": 2.6783635624789317e-06, "loss": 0.0006, "step": 168270 }, { "epoch": 2.7534975047042463, "grad_norm": 0.04201507195830345, "learning_rate": 2.677731263992625e-06, "loss": 0.0004, "step": 168280 }, { "epoch": 2.753661130655322, "grad_norm": 0.0025390787050127983, "learning_rate": 2.6770990128554025e-06, "loss": 0.0007, "step": 168290 }, { "epoch": 2.753824756606398, "grad_norm": 0.07286616414785385, "learning_rate": 2.6764668090801555e-06, "loss": 0.0009, "step": 168300 }, { "epoch": 2.7539883825574734, "grad_norm": 0.07271336019039154, "learning_rate": 2.6758346526797763e-06, "loss": 0.0007, "step": 168310 }, { "epoch": 2.7541520085085494, "grad_norm": 0.010182997211813927, "learning_rate": 2.6752025436671497e-06, "loss": 0.0006, "step": 168320 }, { "epoch": 2.7543156344596254, "grad_norm": 0.036163415759801865, "learning_rate": 2.6745704820551665e-06, "loss": 0.0006, "step": 168330 }, { "epoch": 2.754479260410701, "grad_norm": 0.041492290794849396, "learning_rate": 2.6739384678567127e-06, "loss": 0.0004, "step": 168340 }, { "epoch": 2.754642886361777, "grad_norm": 0.006646294612437487, "learning_rate": 2.673306501084678e-06, "loss": 0.0005, "step": 168350 }, { "epoch": 2.754806512312853, "grad_norm": 0.024128464981913567, "learning_rate": 2.6726745817519428e-06, "loss": 0.0005, "step": 168360 }, { "epoch": 2.7549701382639284, "grad_norm": 0.02307283878326416, "learning_rate": 2.672042709871393e-06, "loss": 0.0009, "step": 168370 }, { "epoch": 2.7551337642150044, "grad_norm": 0.0898175984621048, "learning_rate": 2.671410885455913e-06, "loss": 0.0003, "step": 168380 }, { "epoch": 2.7552973901660804, "grad_norm": 0.0969274565577507, "learning_rate": 2.6707791085183866e-06, "loss": 0.0008, "step": 168390 }, { "epoch": 2.755461016117156, "grad_norm": 0.028624258935451508, "learning_rate": 2.6701473790716913e-06, "loss": 0.0006, "step": 168400 }, { "epoch": 2.755624642068232, "grad_norm": 0.0052168117836117744, "learning_rate": 2.6695156971287093e-06, "loss": 0.0003, "step": 168410 }, { "epoch": 2.755788268019308, "grad_norm": 0.08131970465183258, "learning_rate": 2.6688840627023203e-06, "loss": 0.001, "step": 168420 }, { "epoch": 2.7559518939703835, "grad_norm": 0.04585862159729004, "learning_rate": 2.668252475805405e-06, "loss": 0.0008, "step": 168430 }, { "epoch": 2.7561155199214595, "grad_norm": 0.04936068877577782, "learning_rate": 2.6676209364508365e-06, "loss": 0.001, "step": 168440 }, { "epoch": 2.7562791458725355, "grad_norm": 0.04272553324699402, "learning_rate": 2.666989444651493e-06, "loss": 0.0005, "step": 168450 }, { "epoch": 2.756442771823611, "grad_norm": 0.02367137186229229, "learning_rate": 2.6663580004202517e-06, "loss": 0.0003, "step": 168460 }, { "epoch": 2.756606397774687, "grad_norm": 0.032601796090602875, "learning_rate": 2.665726603769987e-06, "loss": 0.0008, "step": 168470 }, { "epoch": 2.756770023725763, "grad_norm": 0.06989048421382904, "learning_rate": 2.66509525471357e-06, "loss": 0.0015, "step": 168480 }, { "epoch": 2.7569336496768386, "grad_norm": 0.0018570565152913332, "learning_rate": 2.664463953263875e-06, "loss": 0.0004, "step": 168490 }, { "epoch": 2.7570972756279146, "grad_norm": 0.021674999967217445, "learning_rate": 2.6638326994337736e-06, "loss": 0.0007, "step": 168500 }, { "epoch": 2.7572609015789906, "grad_norm": 0.004941517021507025, "learning_rate": 2.6632014932361384e-06, "loss": 0.0004, "step": 168510 }, { "epoch": 2.757424527530066, "grad_norm": 0.0058451308868825436, "learning_rate": 2.6625703346838363e-06, "loss": 0.0006, "step": 168520 }, { "epoch": 2.757588153481142, "grad_norm": 0.0016044283984228969, "learning_rate": 2.6619392237897358e-06, "loss": 0.0008, "step": 168530 }, { "epoch": 2.757751779432218, "grad_norm": 0.016046814620494843, "learning_rate": 2.6613081605667068e-06, "loss": 0.0004, "step": 168540 }, { "epoch": 2.7579154053832937, "grad_norm": 0.041782230138778687, "learning_rate": 2.6606771450276177e-06, "loss": 0.0006, "step": 168550 }, { "epoch": 2.7580790313343697, "grad_norm": 0.07880429923534393, "learning_rate": 2.660046177185329e-06, "loss": 0.001, "step": 168560 }, { "epoch": 2.7582426572854457, "grad_norm": 0.020255237817764282, "learning_rate": 2.6594152570527097e-06, "loss": 0.0004, "step": 168570 }, { "epoch": 2.7584062832365213, "grad_norm": 0.06260441988706589, "learning_rate": 2.658784384642623e-06, "loss": 0.0009, "step": 168580 }, { "epoch": 2.7585699091875973, "grad_norm": 0.012365843169391155, "learning_rate": 2.658153559967933e-06, "loss": 0.0006, "step": 168590 }, { "epoch": 2.7587335351386733, "grad_norm": 0.08782736212015152, "learning_rate": 2.6575227830414984e-06, "loss": 0.0008, "step": 168600 }, { "epoch": 2.758897161089749, "grad_norm": 0.031325265765190125, "learning_rate": 2.6568920538761817e-06, "loss": 0.0005, "step": 168610 }, { "epoch": 2.759060787040825, "grad_norm": 0.007564898114651442, "learning_rate": 2.6562613724848437e-06, "loss": 0.0005, "step": 168620 }, { "epoch": 2.7592244129919004, "grad_norm": 0.028816036880016327, "learning_rate": 2.655630738880345e-06, "loss": 0.0005, "step": 168630 }, { "epoch": 2.7593880389429764, "grad_norm": 0.07931694388389587, "learning_rate": 2.6550001530755393e-06, "loss": 0.0013, "step": 168640 }, { "epoch": 2.7595516648940523, "grad_norm": 0.012997593730688095, "learning_rate": 2.6543696150832866e-06, "loss": 0.0008, "step": 168650 }, { "epoch": 2.759715290845128, "grad_norm": 0.04464425519108772, "learning_rate": 2.6537391249164424e-06, "loss": 0.0015, "step": 168660 }, { "epoch": 2.759878916796204, "grad_norm": 0.0032382411882281303, "learning_rate": 2.6531086825878637e-06, "loss": 0.0005, "step": 168670 }, { "epoch": 2.7600425427472794, "grad_norm": 0.0492442324757576, "learning_rate": 2.6524782881104015e-06, "loss": 0.0005, "step": 168680 }, { "epoch": 2.7602061686983554, "grad_norm": 0.022624919191002846, "learning_rate": 2.65184794149691e-06, "loss": 0.001, "step": 168690 }, { "epoch": 2.7603697946494314, "grad_norm": 0.0036011850461363792, "learning_rate": 2.651217642760242e-06, "loss": 0.0002, "step": 168700 }, { "epoch": 2.760533420600507, "grad_norm": 0.02240709215402603, "learning_rate": 2.650587391913251e-06, "loss": 0.0005, "step": 168710 }, { "epoch": 2.760697046551583, "grad_norm": 0.09075091034173965, "learning_rate": 2.6499571889687835e-06, "loss": 0.0013, "step": 168720 }, { "epoch": 2.760860672502659, "grad_norm": 0.03995969146490097, "learning_rate": 2.6493270339396915e-06, "loss": 0.0009, "step": 168730 }, { "epoch": 2.7610242984537345, "grad_norm": 0.01954427734017372, "learning_rate": 2.6486969268388186e-06, "loss": 0.0011, "step": 168740 }, { "epoch": 2.7611879244048105, "grad_norm": 0.03627542778849602, "learning_rate": 2.6480668676790188e-06, "loss": 0.0004, "step": 168750 }, { "epoch": 2.7613515503558865, "grad_norm": 0.030472885817289352, "learning_rate": 2.647436856473134e-06, "loss": 0.0006, "step": 168760 }, { "epoch": 2.761515176306962, "grad_norm": 0.07126400619745255, "learning_rate": 2.646806893234013e-06, "loss": 0.0007, "step": 168770 }, { "epoch": 2.761678802258038, "grad_norm": 0.047823548316955566, "learning_rate": 2.646176977974494e-06, "loss": 0.0003, "step": 168780 }, { "epoch": 2.761842428209114, "grad_norm": 0.03543458878993988, "learning_rate": 2.6455471107074283e-06, "loss": 0.0007, "step": 168790 }, { "epoch": 2.7620060541601896, "grad_norm": 0.010149581357836723, "learning_rate": 2.6449172914456524e-06, "loss": 0.0005, "step": 168800 }, { "epoch": 2.7621696801112656, "grad_norm": 0.043290555477142334, "learning_rate": 2.6442875202020126e-06, "loss": 0.0007, "step": 168810 }, { "epoch": 2.7623333060623416, "grad_norm": 0.0005620401352643967, "learning_rate": 2.643657796989343e-06, "loss": 0.0011, "step": 168820 }, { "epoch": 2.762496932013417, "grad_norm": 0.003690407145768404, "learning_rate": 2.64302812182049e-06, "loss": 0.0005, "step": 168830 }, { "epoch": 2.762660557964493, "grad_norm": 0.05436837673187256, "learning_rate": 2.6423984947082874e-06, "loss": 0.0003, "step": 168840 }, { "epoch": 2.762824183915569, "grad_norm": 0.03576609119772911, "learning_rate": 2.6417689156655767e-06, "loss": 0.0016, "step": 168850 }, { "epoch": 2.7629878098666447, "grad_norm": 0.08584592491388321, "learning_rate": 2.641139384705188e-06, "loss": 0.0006, "step": 168860 }, { "epoch": 2.7631514358177207, "grad_norm": 0.05094887316226959, "learning_rate": 2.6405099018399655e-06, "loss": 0.0007, "step": 168870 }, { "epoch": 2.7633150617687967, "grad_norm": 0.06439793854951859, "learning_rate": 2.6398804670827372e-06, "loss": 0.0007, "step": 168880 }, { "epoch": 2.7634786877198723, "grad_norm": 0.0589623861014843, "learning_rate": 2.6392510804463413e-06, "loss": 0.0015, "step": 168890 }, { "epoch": 2.7636423136709483, "grad_norm": 0.016850490123033524, "learning_rate": 2.6386217419436057e-06, "loss": 0.0006, "step": 168900 }, { "epoch": 2.7638059396220243, "grad_norm": 0.03727179393172264, "learning_rate": 2.6379924515873646e-06, "loss": 0.0008, "step": 168910 }, { "epoch": 2.7639695655731, "grad_norm": 0.04896881431341171, "learning_rate": 2.6373632093904477e-06, "loss": 0.0009, "step": 168920 }, { "epoch": 2.764133191524176, "grad_norm": 0.04070782661437988, "learning_rate": 2.6367340153656886e-06, "loss": 0.0006, "step": 168930 }, { "epoch": 2.764296817475252, "grad_norm": 0.03564491122961044, "learning_rate": 2.636104869525911e-06, "loss": 0.0006, "step": 168940 }, { "epoch": 2.7644604434263274, "grad_norm": 0.005553773138672113, "learning_rate": 2.635475771883945e-06, "loss": 0.0007, "step": 168950 }, { "epoch": 2.7646240693774033, "grad_norm": 0.024650126695632935, "learning_rate": 2.6348467224526163e-06, "loss": 0.0007, "step": 168960 }, { "epoch": 2.7647876953284793, "grad_norm": 0.006456955336034298, "learning_rate": 2.634217721244754e-06, "loss": 0.0007, "step": 168970 }, { "epoch": 2.764951321279555, "grad_norm": 0.02973599173128605, "learning_rate": 2.6335887682731774e-06, "loss": 0.0007, "step": 168980 }, { "epoch": 2.765114947230631, "grad_norm": 0.058735135942697525, "learning_rate": 2.632959863550715e-06, "loss": 0.0011, "step": 168990 }, { "epoch": 2.765278573181707, "grad_norm": 0.07408122718334198, "learning_rate": 2.6323310070901863e-06, "loss": 0.001, "step": 169000 }, { "epoch": 2.7654421991327824, "grad_norm": 0.03387506306171417, "learning_rate": 2.631702198904417e-06, "loss": 0.0014, "step": 169010 }, { "epoch": 2.7656058250838584, "grad_norm": 0.0799381285905838, "learning_rate": 2.6310734390062244e-06, "loss": 0.0009, "step": 169020 }, { "epoch": 2.765769451034934, "grad_norm": 0.029930317774415016, "learning_rate": 2.6304447274084287e-06, "loss": 0.0006, "step": 169030 }, { "epoch": 2.76593307698601, "grad_norm": 0.013925342820584774, "learning_rate": 2.6298160641238507e-06, "loss": 0.0011, "step": 169040 }, { "epoch": 2.766096702937086, "grad_norm": 0.027116265147924423, "learning_rate": 2.629187449165309e-06, "loss": 0.0011, "step": 169050 }, { "epoch": 2.7662603288881615, "grad_norm": 0.03386782854795456, "learning_rate": 2.6285588825456165e-06, "loss": 0.0005, "step": 169060 }, { "epoch": 2.7664239548392375, "grad_norm": 0.006525530945509672, "learning_rate": 2.627930364277592e-06, "loss": 0.0005, "step": 169070 }, { "epoch": 2.766587580790313, "grad_norm": 0.04633353278040886, "learning_rate": 2.6273018943740503e-06, "loss": 0.0006, "step": 169080 }, { "epoch": 2.766751206741389, "grad_norm": 0.03684278950095177, "learning_rate": 2.6266734728478072e-06, "loss": 0.0012, "step": 169090 }, { "epoch": 2.766914832692465, "grad_norm": 0.07935095578432083, "learning_rate": 2.6260450997116705e-06, "loss": 0.0006, "step": 169100 }, { "epoch": 2.7670784586435406, "grad_norm": 0.010260473005473614, "learning_rate": 2.625416774978456e-06, "loss": 0.0011, "step": 169110 }, { "epoch": 2.7672420845946166, "grad_norm": 0.04321916773915291, "learning_rate": 2.624788498660974e-06, "loss": 0.0011, "step": 169120 }, { "epoch": 2.7674057105456926, "grad_norm": 0.01768164336681366, "learning_rate": 2.624160270772036e-06, "loss": 0.0004, "step": 169130 }, { "epoch": 2.767569336496768, "grad_norm": 0.0347382090985775, "learning_rate": 2.623532091324448e-06, "loss": 0.0005, "step": 169140 }, { "epoch": 2.767732962447844, "grad_norm": 0.04018424078822136, "learning_rate": 2.622903960331019e-06, "loss": 0.0062, "step": 169150 }, { "epoch": 2.76789658839892, "grad_norm": 0.08780490607023239, "learning_rate": 2.622275877804557e-06, "loss": 0.0009, "step": 169160 }, { "epoch": 2.7680602143499957, "grad_norm": 0.07547362893819809, "learning_rate": 2.6216478437578696e-06, "loss": 0.0011, "step": 169170 }, { "epoch": 2.7682238403010717, "grad_norm": 0.02283739298582077, "learning_rate": 2.621019858203758e-06, "loss": 0.0009, "step": 169180 }, { "epoch": 2.7683874662521477, "grad_norm": 0.08462992310523987, "learning_rate": 2.6203919211550275e-06, "loss": 0.0004, "step": 169190 }, { "epoch": 2.7685510922032233, "grad_norm": 0.006401325576007366, "learning_rate": 2.6197640326244834e-06, "loss": 0.0004, "step": 169200 }, { "epoch": 2.7687147181542993, "grad_norm": 0.03938895836472511, "learning_rate": 2.619136192624927e-06, "loss": 0.0008, "step": 169210 }, { "epoch": 2.7688783441053753, "grad_norm": 0.037975944578647614, "learning_rate": 2.6185084011691574e-06, "loss": 0.0006, "step": 169220 }, { "epoch": 2.769041970056451, "grad_norm": 0.040083061903715134, "learning_rate": 2.617880658269976e-06, "loss": 0.0009, "step": 169230 }, { "epoch": 2.769205596007527, "grad_norm": 0.05483349412679672, "learning_rate": 2.617252963940182e-06, "loss": 0.0006, "step": 169240 }, { "epoch": 2.769369221958603, "grad_norm": 0.016034554690122604, "learning_rate": 2.616625318192576e-06, "loss": 0.0008, "step": 169250 }, { "epoch": 2.7695328479096784, "grad_norm": 0.11363878101110458, "learning_rate": 2.61599772103995e-06, "loss": 0.0011, "step": 169260 }, { "epoch": 2.7696964738607543, "grad_norm": 0.042509812861680984, "learning_rate": 2.615370172495104e-06, "loss": 0.0005, "step": 169270 }, { "epoch": 2.7698600998118303, "grad_norm": 0.030252842232584953, "learning_rate": 2.6147426725708315e-06, "loss": 0.0011, "step": 169280 }, { "epoch": 2.770023725762906, "grad_norm": 0.011411488987505436, "learning_rate": 2.6141152212799293e-06, "loss": 0.0007, "step": 169290 }, { "epoch": 2.770187351713982, "grad_norm": 0.04737001284956932, "learning_rate": 2.6134878186351863e-06, "loss": 0.0004, "step": 169300 }, { "epoch": 2.770350977665058, "grad_norm": 0.0012127016671001911, "learning_rate": 2.612860464649398e-06, "loss": 0.0008, "step": 169310 }, { "epoch": 2.7705146036161334, "grad_norm": 0.01438699197024107, "learning_rate": 2.6122331593353537e-06, "loss": 0.0003, "step": 169320 }, { "epoch": 2.7706782295672094, "grad_norm": 0.015394962392747402, "learning_rate": 2.611605902705847e-06, "loss": 0.0006, "step": 169330 }, { "epoch": 2.7708418555182854, "grad_norm": 0.06150966137647629, "learning_rate": 2.6109786947736626e-06, "loss": 0.0006, "step": 169340 }, { "epoch": 2.771005481469361, "grad_norm": 0.005677678622305393, "learning_rate": 2.610351535551591e-06, "loss": 0.0008, "step": 169350 }, { "epoch": 2.771169107420437, "grad_norm": 0.023101847618818283, "learning_rate": 2.609724425052419e-06, "loss": 0.0005, "step": 169360 }, { "epoch": 2.771332733371513, "grad_norm": 0.005577328149229288, "learning_rate": 2.609097363288935e-06, "loss": 0.0014, "step": 169370 }, { "epoch": 2.7714963593225885, "grad_norm": 0.022728201001882553, "learning_rate": 2.6084703502739206e-06, "loss": 0.0003, "step": 169380 }, { "epoch": 2.7716599852736645, "grad_norm": 0.008402526378631592, "learning_rate": 2.6078433860201635e-06, "loss": 0.0011, "step": 169390 }, { "epoch": 2.77182361122474, "grad_norm": 0.042768824845552444, "learning_rate": 2.607216470540442e-06, "loss": 0.0007, "step": 169400 }, { "epoch": 2.771987237175816, "grad_norm": 0.02923726662993431, "learning_rate": 2.6065896038475452e-06, "loss": 0.0006, "step": 169410 }, { "epoch": 2.772150863126892, "grad_norm": 0.059130385518074036, "learning_rate": 2.6059627859542492e-06, "loss": 0.001, "step": 169420 }, { "epoch": 2.7723144890779676, "grad_norm": 0.06636963039636612, "learning_rate": 2.6053360168733376e-06, "loss": 0.0008, "step": 169430 }, { "epoch": 2.7724781150290436, "grad_norm": 0.0017828235868364573, "learning_rate": 2.604709296617584e-06, "loss": 0.0003, "step": 169440 }, { "epoch": 2.772641740980119, "grad_norm": 0.008225995115935802, "learning_rate": 2.6040826251997746e-06, "loss": 0.0006, "step": 169450 }, { "epoch": 2.772805366931195, "grad_norm": 0.006951289717108011, "learning_rate": 2.6034560026326815e-06, "loss": 0.0013, "step": 169460 }, { "epoch": 2.772968992882271, "grad_norm": 0.04613218083977699, "learning_rate": 2.6028294289290832e-06, "loss": 0.0007, "step": 169470 }, { "epoch": 2.7731326188333467, "grad_norm": 0.03568341210484505, "learning_rate": 2.6022029041017517e-06, "loss": 0.0007, "step": 169480 }, { "epoch": 2.7732962447844227, "grad_norm": 0.0009709939477033913, "learning_rate": 2.6015764281634666e-06, "loss": 0.0005, "step": 169490 }, { "epoch": 2.7734598707354987, "grad_norm": 0.28632116317749023, "learning_rate": 2.600950001126996e-06, "loss": 0.0009, "step": 169500 }, { "epoch": 2.7736234966865743, "grad_norm": 0.005468974821269512, "learning_rate": 2.6003236230051175e-06, "loss": 0.0005, "step": 169510 }, { "epoch": 2.7737871226376503, "grad_norm": 0.0144029650837183, "learning_rate": 2.5996972938105945e-06, "loss": 0.0004, "step": 169520 }, { "epoch": 2.7739507485887263, "grad_norm": 0.024686461314558983, "learning_rate": 2.5990710135562076e-06, "loss": 0.001, "step": 169530 }, { "epoch": 2.774114374539802, "grad_norm": 0.035521894693374634, "learning_rate": 2.5984447822547176e-06, "loss": 0.0006, "step": 169540 }, { "epoch": 2.774278000490878, "grad_norm": 0.03405411168932915, "learning_rate": 2.5978185999188988e-06, "loss": 0.0005, "step": 169550 }, { "epoch": 2.774441626441954, "grad_norm": 0.03284066170454025, "learning_rate": 2.5971924665615133e-06, "loss": 0.001, "step": 169560 }, { "epoch": 2.7746052523930294, "grad_norm": 0.07560758292675018, "learning_rate": 2.59656638219533e-06, "loss": 0.0006, "step": 169570 }, { "epoch": 2.7747688783441053, "grad_norm": 0.052003201097249985, "learning_rate": 2.5959403468331145e-06, "loss": 0.0004, "step": 169580 }, { "epoch": 2.7749325042951813, "grad_norm": 0.024292035028338432, "learning_rate": 2.5953143604876326e-06, "loss": 0.0024, "step": 169590 }, { "epoch": 2.775096130246257, "grad_norm": 0.037587396800518036, "learning_rate": 2.5946884231716435e-06, "loss": 0.0007, "step": 169600 }, { "epoch": 2.775259756197333, "grad_norm": 0.007069471757858992, "learning_rate": 2.5940625348979122e-06, "loss": 0.0012, "step": 169610 }, { "epoch": 2.775423382148409, "grad_norm": 0.045266494154930115, "learning_rate": 2.593436695679199e-06, "loss": 0.0002, "step": 169620 }, { "epoch": 2.7755870080994844, "grad_norm": 0.02268105372786522, "learning_rate": 2.5928109055282674e-06, "loss": 0.0006, "step": 169630 }, { "epoch": 2.7757506340505604, "grad_norm": 0.08594696968793869, "learning_rate": 2.592185164457872e-06, "loss": 0.0017, "step": 169640 }, { "epoch": 2.7759142600016364, "grad_norm": 0.025128837674856186, "learning_rate": 2.5915594724807736e-06, "loss": 0.0004, "step": 169650 }, { "epoch": 2.776077885952712, "grad_norm": 0.06394593417644501, "learning_rate": 2.5909338296097285e-06, "loss": 0.0007, "step": 169660 }, { "epoch": 2.776241511903788, "grad_norm": 0.0698341429233551, "learning_rate": 2.590308235857497e-06, "loss": 0.0005, "step": 169670 }, { "epoch": 2.776405137854864, "grad_norm": 0.020412033423781395, "learning_rate": 2.5896826912368282e-06, "loss": 0.0005, "step": 169680 }, { "epoch": 2.7765687638059395, "grad_norm": 0.03878944739699364, "learning_rate": 2.58905719576048e-06, "loss": 0.0003, "step": 169690 }, { "epoch": 2.7767323897570155, "grad_norm": 0.10296111553907394, "learning_rate": 2.588431749441205e-06, "loss": 0.0007, "step": 169700 }, { "epoch": 2.7768960157080915, "grad_norm": 0.010599798522889614, "learning_rate": 2.5878063522917573e-06, "loss": 0.0006, "step": 169710 }, { "epoch": 2.777059641659167, "grad_norm": 0.03131521865725517, "learning_rate": 2.5871810043248847e-06, "loss": 0.0005, "step": 169720 }, { "epoch": 2.777223267610243, "grad_norm": 0.05277930945158005, "learning_rate": 2.5865557055533396e-06, "loss": 0.0015, "step": 169730 }, { "epoch": 2.777386893561319, "grad_norm": 0.06893699616193771, "learning_rate": 2.5859304559898712e-06, "loss": 0.0007, "step": 169740 }, { "epoch": 2.7775505195123946, "grad_norm": 0.017641304060816765, "learning_rate": 2.585305255647229e-06, "loss": 0.0004, "step": 169750 }, { "epoch": 2.7777141454634706, "grad_norm": 0.0045057255774736404, "learning_rate": 2.5846801045381576e-06, "loss": 0.0007, "step": 169760 }, { "epoch": 2.7778777714145466, "grad_norm": 0.026017311960458755, "learning_rate": 2.584055002675405e-06, "loss": 0.0007, "step": 169770 }, { "epoch": 2.778041397365622, "grad_norm": 0.016111770644783974, "learning_rate": 2.583429950071716e-06, "loss": 0.0017, "step": 169780 }, { "epoch": 2.778205023316698, "grad_norm": 0.007431491278111935, "learning_rate": 2.5828049467398363e-06, "loss": 0.0013, "step": 169790 }, { "epoch": 2.7783686492677737, "grad_norm": 0.024081634357571602, "learning_rate": 2.5821799926925073e-06, "loss": 0.0005, "step": 169800 }, { "epoch": 2.7785322752188497, "grad_norm": 0.001590655418112874, "learning_rate": 2.581555087942471e-06, "loss": 0.0007, "step": 169810 }, { "epoch": 2.7786959011699257, "grad_norm": 0.0029736387077718973, "learning_rate": 2.58093023250247e-06, "loss": 0.0005, "step": 169820 }, { "epoch": 2.7788595271210013, "grad_norm": 0.14882931113243103, "learning_rate": 2.5803054263852463e-06, "loss": 0.0008, "step": 169830 }, { "epoch": 2.7790231530720773, "grad_norm": 0.011989852413535118, "learning_rate": 2.579680669603535e-06, "loss": 0.0005, "step": 169840 }, { "epoch": 2.779186779023153, "grad_norm": 0.048046112060546875, "learning_rate": 2.5790559621700763e-06, "loss": 0.0006, "step": 169850 }, { "epoch": 2.779350404974229, "grad_norm": 0.02493913285434246, "learning_rate": 2.5784313040976076e-06, "loss": 0.001, "step": 169860 }, { "epoch": 2.779514030925305, "grad_norm": 0.0064604138024151325, "learning_rate": 2.5778066953988673e-06, "loss": 0.0008, "step": 169870 }, { "epoch": 2.7796776568763804, "grad_norm": 0.03135082870721817, "learning_rate": 2.5771821360865867e-06, "loss": 0.001, "step": 169880 }, { "epoch": 2.7798412828274564, "grad_norm": 0.015127250924706459, "learning_rate": 2.5765576261735014e-06, "loss": 0.0004, "step": 169890 }, { "epoch": 2.7800049087785323, "grad_norm": 0.012296373955905437, "learning_rate": 2.5759331656723452e-06, "loss": 0.0005, "step": 169900 }, { "epoch": 2.780168534729608, "grad_norm": 0.04874745011329651, "learning_rate": 2.575308754595852e-06, "loss": 0.0003, "step": 169910 }, { "epoch": 2.780332160680684, "grad_norm": 0.029414502903819084, "learning_rate": 2.574684392956749e-06, "loss": 0.0004, "step": 169920 }, { "epoch": 2.78049578663176, "grad_norm": 0.010781509801745415, "learning_rate": 2.574060080767769e-06, "loss": 0.0007, "step": 169930 }, { "epoch": 2.7806594125828354, "grad_norm": 0.0016949650598689914, "learning_rate": 2.573435818041641e-06, "loss": 0.0004, "step": 169940 }, { "epoch": 2.7808230385339114, "grad_norm": 0.022995227947831154, "learning_rate": 2.5728116047910944e-06, "loss": 0.0003, "step": 169950 }, { "epoch": 2.7809866644849874, "grad_norm": 0.005540385376662016, "learning_rate": 2.5721874410288527e-06, "loss": 0.0003, "step": 169960 }, { "epoch": 2.781150290436063, "grad_norm": 0.06327693909406662, "learning_rate": 2.571563326767645e-06, "loss": 0.0006, "step": 169970 }, { "epoch": 2.781313916387139, "grad_norm": 0.05313989520072937, "learning_rate": 2.5709392620201955e-06, "loss": 0.0004, "step": 169980 }, { "epoch": 2.781477542338215, "grad_norm": 0.0013942945515736938, "learning_rate": 2.5703152467992304e-06, "loss": 0.0007, "step": 169990 }, { "epoch": 2.7816411682892905, "grad_norm": 0.04075857624411583, "learning_rate": 2.569691281117469e-06, "loss": 0.0006, "step": 170000 }, { "epoch": 2.7818047942403665, "grad_norm": 0.022394699975848198, "learning_rate": 2.5690673649876373e-06, "loss": 0.0025, "step": 170010 }, { "epoch": 2.7819684201914425, "grad_norm": 0.026622163131833076, "learning_rate": 2.5684434984224516e-06, "loss": 0.0006, "step": 170020 }, { "epoch": 2.782132046142518, "grad_norm": 0.029651908203959465, "learning_rate": 2.567819681434638e-06, "loss": 0.0014, "step": 170030 }, { "epoch": 2.782295672093594, "grad_norm": 0.013041273690760136, "learning_rate": 2.567195914036911e-06, "loss": 0.0004, "step": 170040 }, { "epoch": 2.78245929804467, "grad_norm": 0.0751965269446373, "learning_rate": 2.566572196241992e-06, "loss": 0.0005, "step": 170050 }, { "epoch": 2.7826229239957456, "grad_norm": 0.0021420528646558523, "learning_rate": 2.565948528062593e-06, "loss": 0.0004, "step": 170060 }, { "epoch": 2.7827865499468216, "grad_norm": 0.029121126979589462, "learning_rate": 2.5653249095114375e-06, "loss": 0.0002, "step": 170070 }, { "epoch": 2.7829501758978976, "grad_norm": 0.05156789347529411, "learning_rate": 2.5647013406012333e-06, "loss": 0.0014, "step": 170080 }, { "epoch": 2.783113801848973, "grad_norm": 0.014467882923781872, "learning_rate": 2.5640778213447005e-06, "loss": 0.0007, "step": 170090 }, { "epoch": 2.783277427800049, "grad_norm": 0.06377498060464859, "learning_rate": 2.563454351754545e-06, "loss": 0.0014, "step": 170100 }, { "epoch": 2.783441053751125, "grad_norm": 0.05445689707994461, "learning_rate": 2.5628309318434876e-06, "loss": 0.0005, "step": 170110 }, { "epoch": 2.7836046797022007, "grad_norm": 0.06795668601989746, "learning_rate": 2.562207561624232e-06, "loss": 0.001, "step": 170120 }, { "epoch": 2.7837683056532767, "grad_norm": 0.0376427099108696, "learning_rate": 2.561584241109493e-06, "loss": 0.0016, "step": 170130 }, { "epoch": 2.7839319316043527, "grad_norm": 0.04668845236301422, "learning_rate": 2.5609609703119743e-06, "loss": 0.0006, "step": 170140 }, { "epoch": 2.7840955575554283, "grad_norm": 0.031032809987664223, "learning_rate": 2.5603377492443905e-06, "loss": 0.0003, "step": 170150 }, { "epoch": 2.7842591835065043, "grad_norm": 0.016332920640707016, "learning_rate": 2.5597145779194433e-06, "loss": 0.0009, "step": 170160 }, { "epoch": 2.78442280945758, "grad_norm": 0.0068149082362651825, "learning_rate": 2.559091456349842e-06, "loss": 0.0004, "step": 170170 }, { "epoch": 2.784586435408656, "grad_norm": 0.008527068421244621, "learning_rate": 2.5584683845482868e-06, "loss": 0.0006, "step": 170180 }, { "epoch": 2.784750061359732, "grad_norm": 0.03392186015844345, "learning_rate": 2.557845362527488e-06, "loss": 0.001, "step": 170190 }, { "epoch": 2.7849136873108074, "grad_norm": 0.011527815833687782, "learning_rate": 2.5572223903001435e-06, "loss": 0.0008, "step": 170200 }, { "epoch": 2.7850773132618833, "grad_norm": 0.008236001245677471, "learning_rate": 2.556599467878959e-06, "loss": 0.0007, "step": 170210 }, { "epoch": 2.785240939212959, "grad_norm": 0.005280183628201485, "learning_rate": 2.5559765952766313e-06, "loss": 0.0007, "step": 170220 }, { "epoch": 2.785404565164035, "grad_norm": 0.003156334161758423, "learning_rate": 2.555353772505863e-06, "loss": 0.0004, "step": 170230 }, { "epoch": 2.785568191115111, "grad_norm": 0.04875931888818741, "learning_rate": 2.5547309995793512e-06, "loss": 0.001, "step": 170240 }, { "epoch": 2.7857318170661864, "grad_norm": 0.04177398979663849, "learning_rate": 2.5541082765097967e-06, "loss": 0.0008, "step": 170250 }, { "epoch": 2.7858954430172624, "grad_norm": 0.05102243274450302, "learning_rate": 2.553485603309893e-06, "loss": 0.0008, "step": 170260 }, { "epoch": 2.7860590689683384, "grad_norm": 0.045889366418123245, "learning_rate": 2.5528629799923366e-06, "loss": 0.0006, "step": 170270 }, { "epoch": 2.786222694919414, "grad_norm": 0.018049390986561775, "learning_rate": 2.552240406569823e-06, "loss": 0.0006, "step": 170280 }, { "epoch": 2.78638632087049, "grad_norm": 0.00283536477945745, "learning_rate": 2.5516178830550474e-06, "loss": 0.0007, "step": 170290 }, { "epoch": 2.786549946821566, "grad_norm": 0.0515143945813179, "learning_rate": 2.550995409460699e-06, "loss": 0.0005, "step": 170300 }, { "epoch": 2.7867135727726415, "grad_norm": 0.03803908824920654, "learning_rate": 2.550372985799472e-06, "loss": 0.0013, "step": 170310 }, { "epoch": 2.7868771987237175, "grad_norm": 0.026956262066960335, "learning_rate": 2.5497506120840563e-06, "loss": 0.001, "step": 170320 }, { "epoch": 2.7870408246747935, "grad_norm": 0.05282515659928322, "learning_rate": 2.549128288327144e-06, "loss": 0.0005, "step": 170330 }, { "epoch": 2.787204450625869, "grad_norm": 0.07770778983831406, "learning_rate": 2.5485060145414186e-06, "loss": 0.001, "step": 170340 }, { "epoch": 2.787368076576945, "grad_norm": 0.01917203515768051, "learning_rate": 2.5478837907395716e-06, "loss": 0.0009, "step": 170350 }, { "epoch": 2.787531702528021, "grad_norm": 0.02091929502785206, "learning_rate": 2.5472616169342878e-06, "loss": 0.0004, "step": 170360 }, { "epoch": 2.7876953284790966, "grad_norm": 0.04747414588928223, "learning_rate": 2.546639493138256e-06, "loss": 0.0015, "step": 170370 }, { "epoch": 2.7878589544301726, "grad_norm": 0.16647900640964508, "learning_rate": 2.5460174193641564e-06, "loss": 0.0013, "step": 170380 }, { "epoch": 2.7880225803812486, "grad_norm": 0.04266669228672981, "learning_rate": 2.5453953956246747e-06, "loss": 0.0007, "step": 170390 }, { "epoch": 2.788186206332324, "grad_norm": 0.05447833985090256, "learning_rate": 2.5447734219324937e-06, "loss": 0.0008, "step": 170400 }, { "epoch": 2.7883498322834, "grad_norm": 0.013441966846585274, "learning_rate": 2.544151498300296e-06, "loss": 0.0004, "step": 170410 }, { "epoch": 2.788513458234476, "grad_norm": 0.015259510837495327, "learning_rate": 2.543529624740758e-06, "loss": 0.0005, "step": 170420 }, { "epoch": 2.7886770841855517, "grad_norm": 0.028820812702178955, "learning_rate": 2.542907801266563e-06, "loss": 0.0004, "step": 170430 }, { "epoch": 2.7888407101366277, "grad_norm": 0.08899670839309692, "learning_rate": 2.5422860278903884e-06, "loss": 0.0007, "step": 170440 }, { "epoch": 2.7890043360877037, "grad_norm": 0.050370898097753525, "learning_rate": 2.5416643046249124e-06, "loss": 0.0006, "step": 170450 }, { "epoch": 2.7891679620387793, "grad_norm": 0.032510604709386826, "learning_rate": 2.5410426314828084e-06, "loss": 0.0005, "step": 170460 }, { "epoch": 2.7893315879898553, "grad_norm": 0.04611050710082054, "learning_rate": 2.540421008476755e-06, "loss": 0.001, "step": 170470 }, { "epoch": 2.7894952139409313, "grad_norm": 0.07475680112838745, "learning_rate": 2.5397994356194246e-06, "loss": 0.0005, "step": 170480 }, { "epoch": 2.789658839892007, "grad_norm": 0.014122777618467808, "learning_rate": 2.539177912923494e-06, "loss": 0.0007, "step": 170490 }, { "epoch": 2.789822465843083, "grad_norm": 0.02671266719698906, "learning_rate": 2.5385564404016305e-06, "loss": 0.0008, "step": 170500 }, { "epoch": 2.789986091794159, "grad_norm": 0.009431748650968075, "learning_rate": 2.5379350180665074e-06, "loss": 0.0005, "step": 170510 }, { "epoch": 2.7901497177452343, "grad_norm": 0.0017035426571965218, "learning_rate": 2.537313645930796e-06, "loss": 0.001, "step": 170520 }, { "epoch": 2.7903133436963103, "grad_norm": 0.004972402937710285, "learning_rate": 2.5366923240071666e-06, "loss": 0.001, "step": 170530 }, { "epoch": 2.7904769696473863, "grad_norm": 0.022162998095154762, "learning_rate": 2.536071052308284e-06, "loss": 0.0012, "step": 170540 }, { "epoch": 2.790640595598462, "grad_norm": 0.016164487227797508, "learning_rate": 2.5354498308468166e-06, "loss": 0.0007, "step": 170550 }, { "epoch": 2.790804221549538, "grad_norm": 0.025021087378263474, "learning_rate": 2.534828659635432e-06, "loss": 0.0012, "step": 170560 }, { "epoch": 2.7909678475006134, "grad_norm": 0.02442312240600586, "learning_rate": 2.5342075386867957e-06, "loss": 0.0018, "step": 170570 }, { "epoch": 2.7911314734516894, "grad_norm": 0.06194029375910759, "learning_rate": 2.5335864680135693e-06, "loss": 0.0007, "step": 170580 }, { "epoch": 2.7912950994027654, "grad_norm": 0.05996868014335632, "learning_rate": 2.532965447628417e-06, "loss": 0.0005, "step": 170590 }, { "epoch": 2.791458725353841, "grad_norm": 0.011501895263791084, "learning_rate": 2.5323444775440007e-06, "loss": 0.0004, "step": 170600 }, { "epoch": 2.791622351304917, "grad_norm": 0.039447128772735596, "learning_rate": 2.5317235577729837e-06, "loss": 0.0004, "step": 170610 }, { "epoch": 2.7917859772559925, "grad_norm": 0.0019519079942256212, "learning_rate": 2.531102688328023e-06, "loss": 0.0003, "step": 170620 }, { "epoch": 2.7919496032070685, "grad_norm": 0.013933551497757435, "learning_rate": 2.530481869221778e-06, "loss": 0.0008, "step": 170630 }, { "epoch": 2.7921132291581445, "grad_norm": 0.021282341331243515, "learning_rate": 2.5298611004669083e-06, "loss": 0.0011, "step": 170640 }, { "epoch": 2.79227685510922, "grad_norm": 0.09690163284540176, "learning_rate": 2.529240382076071e-06, "loss": 0.0006, "step": 170650 }, { "epoch": 2.792440481060296, "grad_norm": 0.06034090742468834, "learning_rate": 2.5286197140619193e-06, "loss": 0.0015, "step": 170660 }, { "epoch": 2.792604107011372, "grad_norm": 0.08359861373901367, "learning_rate": 2.527999096437112e-06, "loss": 0.0007, "step": 170670 }, { "epoch": 2.7927677329624476, "grad_norm": 0.006276390980929136, "learning_rate": 2.5273785292142973e-06, "loss": 0.0006, "step": 170680 }, { "epoch": 2.7929313589135236, "grad_norm": 0.014874911867082119, "learning_rate": 2.5267580124061343e-06, "loss": 0.0008, "step": 170690 }, { "epoch": 2.7930949848645996, "grad_norm": 0.06836733222007751, "learning_rate": 2.5261375460252707e-06, "loss": 0.0008, "step": 170700 }, { "epoch": 2.793258610815675, "grad_norm": 0.04670276492834091, "learning_rate": 2.5255171300843606e-06, "loss": 0.0008, "step": 170710 }, { "epoch": 2.793422236766751, "grad_norm": 0.026936255395412445, "learning_rate": 2.5248967645960476e-06, "loss": 0.0006, "step": 170720 }, { "epoch": 2.793585862717827, "grad_norm": 0.046652987599372864, "learning_rate": 2.5242764495729883e-06, "loss": 0.0005, "step": 170730 }, { "epoch": 2.7937494886689027, "grad_norm": 0.028560981154441833, "learning_rate": 2.5236561850278253e-06, "loss": 0.0007, "step": 170740 }, { "epoch": 2.7939131146199787, "grad_norm": 0.008243936114013195, "learning_rate": 2.523035970973208e-06, "loss": 0.0006, "step": 170750 }, { "epoch": 2.7940767405710547, "grad_norm": 0.04731481894850731, "learning_rate": 2.522415807421777e-06, "loss": 0.0011, "step": 170760 }, { "epoch": 2.7942403665221303, "grad_norm": 0.03149331733584404, "learning_rate": 2.5217956943861843e-06, "loss": 0.0008, "step": 170770 }, { "epoch": 2.7944039924732063, "grad_norm": 0.04571808874607086, "learning_rate": 2.521175631879068e-06, "loss": 0.0007, "step": 170780 }, { "epoch": 2.7945676184242823, "grad_norm": 0.010841126553714275, "learning_rate": 2.520555619913074e-06, "loss": 0.0007, "step": 170790 }, { "epoch": 2.794731244375358, "grad_norm": 0.03104635886847973, "learning_rate": 2.5199356585008393e-06, "loss": 0.0005, "step": 170800 }, { "epoch": 2.794894870326434, "grad_norm": 0.010799898765981197, "learning_rate": 2.51931574765501e-06, "loss": 0.0007, "step": 170810 }, { "epoch": 2.79505849627751, "grad_norm": 0.02845386601984501, "learning_rate": 2.5186958873882216e-06, "loss": 0.0004, "step": 170820 }, { "epoch": 2.7952221222285853, "grad_norm": 0.00817562174052, "learning_rate": 2.518076077713116e-06, "loss": 0.0008, "step": 170830 }, { "epoch": 2.7953857481796613, "grad_norm": 0.03924437612295151, "learning_rate": 2.517456318642326e-06, "loss": 0.0007, "step": 170840 }, { "epoch": 2.7955493741307373, "grad_norm": 0.082326240837574, "learning_rate": 2.51683661018849e-06, "loss": 0.0009, "step": 170850 }, { "epoch": 2.795713000081813, "grad_norm": 0.05594516918063164, "learning_rate": 2.5162169523642443e-06, "loss": 0.0012, "step": 170860 }, { "epoch": 2.795876626032889, "grad_norm": 0.03775654733181, "learning_rate": 2.515597345182225e-06, "loss": 0.0008, "step": 170870 }, { "epoch": 2.796040251983965, "grad_norm": 0.061964698135852814, "learning_rate": 2.5149777886550597e-06, "loss": 0.0009, "step": 170880 }, { "epoch": 2.7962038779350404, "grad_norm": 0.026030590757727623, "learning_rate": 2.514358282795384e-06, "loss": 0.0007, "step": 170890 }, { "epoch": 2.7963675038861164, "grad_norm": 0.08464383333921432, "learning_rate": 2.513738827615829e-06, "loss": 0.0012, "step": 170900 }, { "epoch": 2.7965311298371924, "grad_norm": 0.013459638692438602, "learning_rate": 2.513119423129028e-06, "loss": 0.0009, "step": 170910 }, { "epoch": 2.796694755788268, "grad_norm": 0.015585284680128098, "learning_rate": 2.512500069347604e-06, "loss": 0.0007, "step": 170920 }, { "epoch": 2.796858381739344, "grad_norm": 0.00221878569573164, "learning_rate": 2.5118807662841884e-06, "loss": 0.0003, "step": 170930 }, { "epoch": 2.79702200769042, "grad_norm": 0.014334019273519516, "learning_rate": 2.5112615139514073e-06, "loss": 0.0002, "step": 170940 }, { "epoch": 2.7971856336414955, "grad_norm": 0.0328059084713459, "learning_rate": 2.5106423123618896e-06, "loss": 0.0006, "step": 170950 }, { "epoch": 2.7973492595925715, "grad_norm": 0.02219534106552601, "learning_rate": 2.5100231615282556e-06, "loss": 0.0009, "step": 170960 }, { "epoch": 2.797512885543647, "grad_norm": 0.004102366976439953, "learning_rate": 2.509404061463132e-06, "loss": 0.0012, "step": 170970 }, { "epoch": 2.797676511494723, "grad_norm": 0.032507747411727905, "learning_rate": 2.5087850121791422e-06, "loss": 0.0006, "step": 170980 }, { "epoch": 2.797840137445799, "grad_norm": 0.033862173557281494, "learning_rate": 2.5081660136889087e-06, "loss": 0.0009, "step": 170990 }, { "epoch": 2.7980037633968746, "grad_norm": 0.0044704643078148365, "learning_rate": 2.5075470660050487e-06, "loss": 0.0004, "step": 171000 }, { "epoch": 2.7981673893479506, "grad_norm": 0.029910171404480934, "learning_rate": 2.506928169140185e-06, "loss": 0.0011, "step": 171010 }, { "epoch": 2.798331015299026, "grad_norm": 0.05204833671450615, "learning_rate": 2.5063093231069346e-06, "loss": 0.0009, "step": 171020 }, { "epoch": 2.798494641250102, "grad_norm": 0.004502648487687111, "learning_rate": 2.505690527917919e-06, "loss": 0.0004, "step": 171030 }, { "epoch": 2.798658267201178, "grad_norm": 0.012279222719371319, "learning_rate": 2.5050717835857497e-06, "loss": 0.0006, "step": 171040 }, { "epoch": 2.7988218931522537, "grad_norm": 0.01439549308270216, "learning_rate": 2.5044530901230457e-06, "loss": 0.0007, "step": 171050 }, { "epoch": 2.7989855191033297, "grad_norm": 0.029060525819659233, "learning_rate": 2.50383444754242e-06, "loss": 0.0005, "step": 171060 }, { "epoch": 2.7991491450544057, "grad_norm": 0.03905748948454857, "learning_rate": 2.5032158558564897e-06, "loss": 0.0007, "step": 171070 }, { "epoch": 2.7993127710054813, "grad_norm": 0.03941565752029419, "learning_rate": 2.502597315077862e-06, "loss": 0.0004, "step": 171080 }, { "epoch": 2.7994763969565573, "grad_norm": 0.06257851421833038, "learning_rate": 2.5019788252191512e-06, "loss": 0.0006, "step": 171090 }, { "epoch": 2.7996400229076333, "grad_norm": 0.010083980858325958, "learning_rate": 2.501360386292968e-06, "loss": 0.0018, "step": 171100 }, { "epoch": 2.799803648858709, "grad_norm": 0.06873942166566849, "learning_rate": 2.500741998311923e-06, "loss": 0.001, "step": 171110 }, { "epoch": 2.799967274809785, "grad_norm": 0.15748721361160278, "learning_rate": 2.5001236612886216e-06, "loss": 0.0013, "step": 171120 }, { "epoch": 2.800130900760861, "grad_norm": 0.13346073031425476, "learning_rate": 2.499505375235673e-06, "loss": 0.0009, "step": 171130 }, { "epoch": 2.8002945267119363, "grad_norm": 0.05639443174004555, "learning_rate": 2.498887140165683e-06, "loss": 0.0011, "step": 171140 }, { "epoch": 2.8004581526630123, "grad_norm": 0.007167041767388582, "learning_rate": 2.4982689560912584e-06, "loss": 0.0005, "step": 171150 }, { "epoch": 2.8006217786140883, "grad_norm": 0.025853468105196953, "learning_rate": 2.497650823025001e-06, "loss": 0.0007, "step": 171160 }, { "epoch": 2.800785404565164, "grad_norm": 0.06912831962108612, "learning_rate": 2.497032740979514e-06, "loss": 0.001, "step": 171170 }, { "epoch": 2.80094903051624, "grad_norm": 0.012265415862202644, "learning_rate": 2.496414709967402e-06, "loss": 0.0004, "step": 171180 }, { "epoch": 2.801112656467316, "grad_norm": 0.07613777369260788, "learning_rate": 2.495796730001266e-06, "loss": 0.0005, "step": 171190 }, { "epoch": 2.8012762824183914, "grad_norm": 0.009249639697372913, "learning_rate": 2.495178801093703e-06, "loss": 0.0004, "step": 171200 }, { "epoch": 2.8014399083694674, "grad_norm": 0.0826243907213211, "learning_rate": 2.4945609232573142e-06, "loss": 0.0006, "step": 171210 }, { "epoch": 2.8016035343205434, "grad_norm": 0.04295646771788597, "learning_rate": 2.4939430965046974e-06, "loss": 0.0008, "step": 171220 }, { "epoch": 2.801767160271619, "grad_norm": 0.08097703754901886, "learning_rate": 2.4933253208484516e-06, "loss": 0.0007, "step": 171230 }, { "epoch": 2.801930786222695, "grad_norm": 0.001582779223099351, "learning_rate": 2.4927075963011684e-06, "loss": 0.0002, "step": 171240 }, { "epoch": 2.802094412173771, "grad_norm": 0.029857760295271873, "learning_rate": 2.4920899228754453e-06, "loss": 0.0007, "step": 171250 }, { "epoch": 2.8022580381248465, "grad_norm": 0.018969492986798286, "learning_rate": 2.491472300583875e-06, "loss": 0.0003, "step": 171260 }, { "epoch": 2.8024216640759225, "grad_norm": 0.07697910070419312, "learning_rate": 2.4908547294390538e-06, "loss": 0.0007, "step": 171270 }, { "epoch": 2.8025852900269985, "grad_norm": 0.007839112542569637, "learning_rate": 2.490237209453569e-06, "loss": 0.0008, "step": 171280 }, { "epoch": 2.802748915978074, "grad_norm": 0.047402769327163696, "learning_rate": 2.489619740640014e-06, "loss": 0.001, "step": 171290 }, { "epoch": 2.80291254192915, "grad_norm": 0.026254361495375633, "learning_rate": 2.4890023230109744e-06, "loss": 0.0007, "step": 171300 }, { "epoch": 2.803076167880226, "grad_norm": 0.020696213468909264, "learning_rate": 2.488384956579045e-06, "loss": 0.0006, "step": 171310 }, { "epoch": 2.8032397938313016, "grad_norm": 0.0149840721860528, "learning_rate": 2.487767641356809e-06, "loss": 0.0004, "step": 171320 }, { "epoch": 2.8034034197823776, "grad_norm": 0.02524605765938759, "learning_rate": 2.4871503773568562e-06, "loss": 0.0011, "step": 171330 }, { "epoch": 2.803567045733453, "grad_norm": 0.016876516863703728, "learning_rate": 2.4865331645917663e-06, "loss": 0.001, "step": 171340 }, { "epoch": 2.803730671684529, "grad_norm": 0.024437684565782547, "learning_rate": 2.485916003074131e-06, "loss": 0.0004, "step": 171350 }, { "epoch": 2.803894297635605, "grad_norm": 0.07366999238729477, "learning_rate": 2.4852988928165283e-06, "loss": 0.0007, "step": 171360 }, { "epoch": 2.8040579235866807, "grad_norm": 0.04550350457429886, "learning_rate": 2.4846818338315443e-06, "loss": 0.0006, "step": 171370 }, { "epoch": 2.8042215495377567, "grad_norm": 0.08129384368658066, "learning_rate": 2.4840648261317553e-06, "loss": 0.0005, "step": 171380 }, { "epoch": 2.8043851754888323, "grad_norm": 0.0034994210582226515, "learning_rate": 2.4834478697297482e-06, "loss": 0.0014, "step": 171390 }, { "epoch": 2.8045488014399083, "grad_norm": 0.007379517890512943, "learning_rate": 2.4828309646380964e-06, "loss": 0.0004, "step": 171400 }, { "epoch": 2.8047124273909843, "grad_norm": 0.06922423839569092, "learning_rate": 2.482214110869383e-06, "loss": 0.0011, "step": 171410 }, { "epoch": 2.80487605334206, "grad_norm": 0.031135745346546173, "learning_rate": 2.481597308436179e-06, "loss": 0.0005, "step": 171420 }, { "epoch": 2.805039679293136, "grad_norm": 0.016597559675574303, "learning_rate": 2.4809805573510678e-06, "loss": 0.0005, "step": 171430 }, { "epoch": 2.805203305244212, "grad_norm": 0.08873755484819412, "learning_rate": 2.480363857626618e-06, "loss": 0.0013, "step": 171440 }, { "epoch": 2.8053669311952873, "grad_norm": 0.11751890927553177, "learning_rate": 2.4797472092754087e-06, "loss": 0.0008, "step": 171450 }, { "epoch": 2.8055305571463633, "grad_norm": 0.004146525636315346, "learning_rate": 2.4791306123100066e-06, "loss": 0.0003, "step": 171460 }, { "epoch": 2.8056941830974393, "grad_norm": 0.010716836899518967, "learning_rate": 2.4785140667429915e-06, "loss": 0.0007, "step": 171470 }, { "epoch": 2.805857809048515, "grad_norm": 0.07834234088659286, "learning_rate": 2.4778975725869275e-06, "loss": 0.0008, "step": 171480 }, { "epoch": 2.806021434999591, "grad_norm": 0.028109148144721985, "learning_rate": 2.477281129854389e-06, "loss": 0.0004, "step": 171490 }, { "epoch": 2.806185060950667, "grad_norm": 0.037260089069604874, "learning_rate": 2.4766647385579416e-06, "loss": 0.0007, "step": 171500 }, { "epoch": 2.8063486869017424, "grad_norm": 0.019430480897426605, "learning_rate": 2.4760483987101534e-06, "loss": 0.0004, "step": 171510 }, { "epoch": 2.8065123128528184, "grad_norm": 0.0521080307662487, "learning_rate": 2.4754321103235923e-06, "loss": 0.0007, "step": 171520 }, { "epoch": 2.8066759388038944, "grad_norm": 0.01689154841005802, "learning_rate": 2.4748158734108253e-06, "loss": 0.0004, "step": 171530 }, { "epoch": 2.80683956475497, "grad_norm": 0.015357550233602524, "learning_rate": 2.4741996879844134e-06, "loss": 0.0002, "step": 171540 }, { "epoch": 2.807003190706046, "grad_norm": 0.02863493748009205, "learning_rate": 2.473583554056921e-06, "loss": 0.0008, "step": 171550 }, { "epoch": 2.807166816657122, "grad_norm": 0.03525058925151825, "learning_rate": 2.4729674716409114e-06, "loss": 0.0006, "step": 171560 }, { "epoch": 2.8073304426081975, "grad_norm": 0.0035973102785646915, "learning_rate": 2.472351440748948e-06, "loss": 0.0006, "step": 171570 }, { "epoch": 2.8074940685592735, "grad_norm": 0.04087791591882706, "learning_rate": 2.471735461393587e-06, "loss": 0.0008, "step": 171580 }, { "epoch": 2.8076576945103495, "grad_norm": 0.09029048681259155, "learning_rate": 2.4711195335873898e-06, "loss": 0.001, "step": 171590 }, { "epoch": 2.807821320461425, "grad_norm": 0.03174332156777382, "learning_rate": 2.4705036573429153e-06, "loss": 0.0005, "step": 171600 }, { "epoch": 2.807984946412501, "grad_norm": 0.03197062760591507, "learning_rate": 2.469887832672721e-06, "loss": 0.0012, "step": 171610 }, { "epoch": 2.808148572363577, "grad_norm": 0.030726101249456406, "learning_rate": 2.4692720595893606e-06, "loss": 0.0007, "step": 171620 }, { "epoch": 2.8083121983146526, "grad_norm": 0.06137372553348541, "learning_rate": 2.4686563381053904e-06, "loss": 0.0007, "step": 171630 }, { "epoch": 2.8084758242657286, "grad_norm": 0.004115391056984663, "learning_rate": 2.468040668233364e-06, "loss": 0.0011, "step": 171640 }, { "epoch": 2.8086394502168046, "grad_norm": 0.010387618094682693, "learning_rate": 2.4674250499858375e-06, "loss": 0.0006, "step": 171650 }, { "epoch": 2.80880307616788, "grad_norm": 0.20544452965259552, "learning_rate": 2.4668094833753584e-06, "loss": 0.0013, "step": 171660 }, { "epoch": 2.808966702118956, "grad_norm": 0.05062934756278992, "learning_rate": 2.46619396841448e-06, "loss": 0.0007, "step": 171670 }, { "epoch": 2.809130328070032, "grad_norm": 0.00661207502707839, "learning_rate": 2.465578505115751e-06, "loss": 0.0012, "step": 171680 }, { "epoch": 2.8092939540211077, "grad_norm": 0.043683528900146484, "learning_rate": 2.464963093491723e-06, "loss": 0.0004, "step": 171690 }, { "epoch": 2.8094575799721837, "grad_norm": 0.018102841451764107, "learning_rate": 2.4643477335549397e-06, "loss": 0.0006, "step": 171700 }, { "epoch": 2.8096212059232597, "grad_norm": 0.015890030190348625, "learning_rate": 2.4637324253179495e-06, "loss": 0.0006, "step": 171710 }, { "epoch": 2.8097848318743353, "grad_norm": 0.03185810521245003, "learning_rate": 2.4631171687932982e-06, "loss": 0.0009, "step": 171720 }, { "epoch": 2.8099484578254112, "grad_norm": 0.051307886838912964, "learning_rate": 2.462501963993532e-06, "loss": 0.0008, "step": 171730 }, { "epoch": 2.810112083776487, "grad_norm": 0.02771066129207611, "learning_rate": 2.461886810931191e-06, "loss": 0.0005, "step": 171740 }, { "epoch": 2.810275709727563, "grad_norm": 0.10876939445734024, "learning_rate": 2.461271709618819e-06, "loss": 0.0008, "step": 171750 }, { "epoch": 2.810439335678639, "grad_norm": 0.0681518167257309, "learning_rate": 2.4606566600689586e-06, "loss": 0.0005, "step": 171760 }, { "epoch": 2.8106029616297143, "grad_norm": 0.010804236866533756, "learning_rate": 2.4600416622941503e-06, "loss": 0.0006, "step": 171770 }, { "epoch": 2.8107665875807903, "grad_norm": 0.019073309376835823, "learning_rate": 2.459426716306931e-06, "loss": 0.0008, "step": 171780 }, { "epoch": 2.810930213531866, "grad_norm": 0.025080371648073196, "learning_rate": 2.45881182211984e-06, "loss": 0.0013, "step": 171790 }, { "epoch": 2.811093839482942, "grad_norm": 0.021883519366383553, "learning_rate": 2.4581969797454146e-06, "loss": 0.0005, "step": 171800 }, { "epoch": 2.811257465434018, "grad_norm": 0.043725162744522095, "learning_rate": 2.4575821891961932e-06, "loss": 0.0005, "step": 171810 }, { "epoch": 2.8114210913850934, "grad_norm": 0.04779953882098198, "learning_rate": 2.456967450484707e-06, "loss": 0.0008, "step": 171820 }, { "epoch": 2.8115847173361694, "grad_norm": 0.005467752926051617, "learning_rate": 2.4563527636234914e-06, "loss": 0.0007, "step": 171830 }, { "epoch": 2.8117483432872454, "grad_norm": 0.0013861904153600335, "learning_rate": 2.4557381286250793e-06, "loss": 0.0005, "step": 171840 }, { "epoch": 2.811911969238321, "grad_norm": 0.026505623012781143, "learning_rate": 2.455123545502005e-06, "loss": 0.0008, "step": 171850 }, { "epoch": 2.812075595189397, "grad_norm": 0.06253369152545929, "learning_rate": 2.454509014266796e-06, "loss": 0.0005, "step": 171860 }, { "epoch": 2.812239221140473, "grad_norm": 0.0112251415848732, "learning_rate": 2.4538945349319825e-06, "loss": 0.0006, "step": 171870 }, { "epoch": 2.8124028470915485, "grad_norm": 0.05051334574818611, "learning_rate": 2.453280107510095e-06, "loss": 0.0008, "step": 171880 }, { "epoch": 2.8125664730426245, "grad_norm": 0.03631296008825302, "learning_rate": 2.4526657320136614e-06, "loss": 0.0006, "step": 171890 }, { "epoch": 2.8127300989937005, "grad_norm": 0.02465015836060047, "learning_rate": 2.4520514084552054e-06, "loss": 0.0017, "step": 171900 }, { "epoch": 2.812893724944776, "grad_norm": 0.0011579354759305716, "learning_rate": 2.451437136847254e-06, "loss": 0.0003, "step": 171910 }, { "epoch": 2.813057350895852, "grad_norm": 0.002017262391746044, "learning_rate": 2.4508229172023324e-06, "loss": 0.0008, "step": 171920 }, { "epoch": 2.813220976846928, "grad_norm": 0.06399448961019516, "learning_rate": 2.450208749532965e-06, "loss": 0.0006, "step": 171930 }, { "epoch": 2.8133846027980036, "grad_norm": 0.060256268829107285, "learning_rate": 2.4495946338516706e-06, "loss": 0.0008, "step": 171940 }, { "epoch": 2.8135482287490796, "grad_norm": 0.04563569650053978, "learning_rate": 2.448980570170975e-06, "loss": 0.0007, "step": 171950 }, { "epoch": 2.8137118547001556, "grad_norm": 0.06300219148397446, "learning_rate": 2.4483665585033923e-06, "loss": 0.0007, "step": 171960 }, { "epoch": 2.813875480651231, "grad_norm": 0.10839757323265076, "learning_rate": 2.4477525988614485e-06, "loss": 0.0009, "step": 171970 }, { "epoch": 2.814039106602307, "grad_norm": 0.005765646230429411, "learning_rate": 2.4471386912576574e-06, "loss": 0.0005, "step": 171980 }, { "epoch": 2.814202732553383, "grad_norm": 0.05228367820382118, "learning_rate": 2.4465248357045385e-06, "loss": 0.0006, "step": 171990 }, { "epoch": 2.8143663585044587, "grad_norm": 0.035851266235113144, "learning_rate": 2.4459110322146034e-06, "loss": 0.0007, "step": 172000 }, { "epoch": 2.8145299844555347, "grad_norm": 0.03720936179161072, "learning_rate": 2.4452972808003733e-06, "loss": 0.002, "step": 172010 }, { "epoch": 2.8146936104066107, "grad_norm": 0.011332050897181034, "learning_rate": 2.4446835814743576e-06, "loss": 0.0003, "step": 172020 }, { "epoch": 2.8148572363576863, "grad_norm": 0.0436323881149292, "learning_rate": 2.444069934249072e-06, "loss": 0.0006, "step": 172030 }, { "epoch": 2.8150208623087623, "grad_norm": 0.0344279482960701, "learning_rate": 2.443456339137023e-06, "loss": 0.0004, "step": 172040 }, { "epoch": 2.8151844882598382, "grad_norm": 0.018075313419103622, "learning_rate": 2.442842796150729e-06, "loss": 0.0011, "step": 172050 }, { "epoch": 2.815348114210914, "grad_norm": 0.04170909523963928, "learning_rate": 2.4422293053026936e-06, "loss": 0.0006, "step": 172060 }, { "epoch": 2.81551174016199, "grad_norm": 0.02965141460299492, "learning_rate": 2.441615866605429e-06, "loss": 0.0023, "step": 172070 }, { "epoch": 2.815675366113066, "grad_norm": 0.0036748237907886505, "learning_rate": 2.4410024800714376e-06, "loss": 0.0004, "step": 172080 }, { "epoch": 2.8158389920641413, "grad_norm": 0.0503084771335125, "learning_rate": 2.440389145713233e-06, "loss": 0.0008, "step": 172090 }, { "epoch": 2.8160026180152173, "grad_norm": 0.04472985491156578, "learning_rate": 2.439775863543315e-06, "loss": 0.0004, "step": 172100 }, { "epoch": 2.816166243966293, "grad_norm": 0.05367692932486534, "learning_rate": 2.4391626335741915e-06, "loss": 0.001, "step": 172110 }, { "epoch": 2.816329869917369, "grad_norm": 0.048324547708034515, "learning_rate": 2.438549455818362e-06, "loss": 0.0011, "step": 172120 }, { "epoch": 2.816493495868445, "grad_norm": 0.01982387900352478, "learning_rate": 2.4379363302883307e-06, "loss": 0.0007, "step": 172130 }, { "epoch": 2.8166571218195204, "grad_norm": 0.016059990972280502, "learning_rate": 2.4373232569965987e-06, "loss": 0.0015, "step": 172140 }, { "epoch": 2.8168207477705964, "grad_norm": 0.042541466653347015, "learning_rate": 2.436710235955668e-06, "loss": 0.0009, "step": 172150 }, { "epoch": 2.816984373721672, "grad_norm": 0.030882038176059723, "learning_rate": 2.436097267178034e-06, "loss": 0.0006, "step": 172160 }, { "epoch": 2.817147999672748, "grad_norm": 0.01714320108294487, "learning_rate": 2.4354843506761953e-06, "loss": 0.0004, "step": 172170 }, { "epoch": 2.817311625623824, "grad_norm": 0.001336942077614367, "learning_rate": 2.4348714864626505e-06, "loss": 0.0004, "step": 172180 }, { "epoch": 2.8174752515748995, "grad_norm": 0.053931545466184616, "learning_rate": 2.4342586745498963e-06, "loss": 0.0006, "step": 172190 }, { "epoch": 2.8176388775259755, "grad_norm": 0.04476374387741089, "learning_rate": 2.4336459149504233e-06, "loss": 0.0006, "step": 172200 }, { "epoch": 2.8178025034770515, "grad_norm": 0.0818912610411644, "learning_rate": 2.4330332076767278e-06, "loss": 0.0007, "step": 172210 }, { "epoch": 2.817966129428127, "grad_norm": 0.0860990509390831, "learning_rate": 2.432420552741302e-06, "loss": 0.0011, "step": 172220 }, { "epoch": 2.818129755379203, "grad_norm": 0.039799321442842484, "learning_rate": 2.4318079501566393e-06, "loss": 0.0008, "step": 172230 }, { "epoch": 2.818293381330279, "grad_norm": 0.003128479467704892, "learning_rate": 2.4311953999352265e-06, "loss": 0.0004, "step": 172240 }, { "epoch": 2.8184570072813546, "grad_norm": 0.019013701006770134, "learning_rate": 2.4305829020895548e-06, "loss": 0.0005, "step": 172250 }, { "epoch": 2.8186206332324306, "grad_norm": 0.017027195543050766, "learning_rate": 2.4299704566321123e-06, "loss": 0.0006, "step": 172260 }, { "epoch": 2.8187842591835066, "grad_norm": 0.048037830740213394, "learning_rate": 2.4293580635753886e-06, "loss": 0.001, "step": 172270 }, { "epoch": 2.818947885134582, "grad_norm": 0.04521628096699715, "learning_rate": 2.428745722931866e-06, "loss": 0.0005, "step": 172280 }, { "epoch": 2.819111511085658, "grad_norm": 0.028171628713607788, "learning_rate": 2.428133434714032e-06, "loss": 0.001, "step": 172290 }, { "epoch": 2.819275137036734, "grad_norm": 0.028686536476016045, "learning_rate": 2.427521198934369e-06, "loss": 0.0005, "step": 172300 }, { "epoch": 2.8194387629878097, "grad_norm": 0.003885350190103054, "learning_rate": 2.426909015605363e-06, "loss": 0.0004, "step": 172310 }, { "epoch": 2.8196023889388857, "grad_norm": 0.05595583841204643, "learning_rate": 2.426296884739491e-06, "loss": 0.0007, "step": 172320 }, { "epoch": 2.8197660148899617, "grad_norm": 0.12640655040740967, "learning_rate": 2.425684806349238e-06, "loss": 0.0006, "step": 172330 }, { "epoch": 2.8199296408410373, "grad_norm": 0.04295692592859268, "learning_rate": 2.425072780447082e-06, "loss": 0.0004, "step": 172340 }, { "epoch": 2.8200932667921133, "grad_norm": 0.15900972485542297, "learning_rate": 2.424460807045504e-06, "loss": 0.0013, "step": 172350 }, { "epoch": 2.8202568927431892, "grad_norm": 0.017198771238327026, "learning_rate": 2.423848886156978e-06, "loss": 0.0007, "step": 172360 }, { "epoch": 2.820420518694265, "grad_norm": 0.14690804481506348, "learning_rate": 2.4232370177939823e-06, "loss": 0.0013, "step": 172370 }, { "epoch": 2.820584144645341, "grad_norm": 0.05105411633849144, "learning_rate": 2.4226252019689923e-06, "loss": 0.0008, "step": 172380 }, { "epoch": 2.820747770596417, "grad_norm": 0.005091603379696608, "learning_rate": 2.4220134386944837e-06, "loss": 0.0007, "step": 172390 }, { "epoch": 2.8209113965474923, "grad_norm": 0.043878793716430664, "learning_rate": 2.4214017279829273e-06, "loss": 0.0008, "step": 172400 }, { "epoch": 2.8210750224985683, "grad_norm": 0.00932342279702425, "learning_rate": 2.4207900698467967e-06, "loss": 0.0008, "step": 172410 }, { "epoch": 2.8212386484496443, "grad_norm": 0.02394486963748932, "learning_rate": 2.4201784642985625e-06, "loss": 0.003, "step": 172420 }, { "epoch": 2.82140227440072, "grad_norm": 0.06821834295988083, "learning_rate": 2.419566911350698e-06, "loss": 0.0009, "step": 172430 }, { "epoch": 2.821565900351796, "grad_norm": 0.0035656869877129793, "learning_rate": 2.418955411015667e-06, "loss": 0.0007, "step": 172440 }, { "epoch": 2.821729526302872, "grad_norm": 0.01684020459651947, "learning_rate": 2.418343963305941e-06, "loss": 0.0014, "step": 172450 }, { "epoch": 2.8218931522539474, "grad_norm": 0.033568352460861206, "learning_rate": 2.4177325682339854e-06, "loss": 0.0017, "step": 172460 }, { "epoch": 2.8220567782050234, "grad_norm": 0.03740110993385315, "learning_rate": 2.4171212258122685e-06, "loss": 0.0004, "step": 172470 }, { "epoch": 2.8222204041560994, "grad_norm": 0.028125783428549767, "learning_rate": 2.4165099360532505e-06, "loss": 0.0006, "step": 172480 }, { "epoch": 2.822384030107175, "grad_norm": 0.002397105796262622, "learning_rate": 2.415898698969399e-06, "loss": 0.0006, "step": 172490 }, { "epoch": 2.822547656058251, "grad_norm": 0.006718449294567108, "learning_rate": 2.4152875145731747e-06, "loss": 0.0009, "step": 172500 }, { "epoch": 2.8227112820093265, "grad_norm": 0.05964549258351326, "learning_rate": 2.4146763828770412e-06, "loss": 0.0008, "step": 172510 }, { "epoch": 2.8228749079604025, "grad_norm": 0.05450465530157089, "learning_rate": 2.414065303893456e-06, "loss": 0.0006, "step": 172520 }, { "epoch": 2.8230385339114785, "grad_norm": 0.02498820796608925, "learning_rate": 2.4134542776348807e-06, "loss": 0.0009, "step": 172530 }, { "epoch": 2.823202159862554, "grad_norm": 0.022544294595718384, "learning_rate": 2.412843304113772e-06, "loss": 0.0005, "step": 172540 }, { "epoch": 2.82336578581363, "grad_norm": 0.026962019503116608, "learning_rate": 2.4122323833425907e-06, "loss": 0.0003, "step": 172550 }, { "epoch": 2.8235294117647056, "grad_norm": 0.02509063109755516, "learning_rate": 2.411621515333788e-06, "loss": 0.0004, "step": 172560 }, { "epoch": 2.8236930377157816, "grad_norm": 0.11629891395568848, "learning_rate": 2.4110107000998213e-06, "loss": 0.0007, "step": 172570 }, { "epoch": 2.8238566636668576, "grad_norm": 0.04539649933576584, "learning_rate": 2.4103999376531445e-06, "loss": 0.0013, "step": 172580 }, { "epoch": 2.824020289617933, "grad_norm": 0.005978779401630163, "learning_rate": 2.4097892280062122e-06, "loss": 0.0006, "step": 172590 }, { "epoch": 2.824183915569009, "grad_norm": 0.007238056976348162, "learning_rate": 2.4091785711714734e-06, "loss": 0.0005, "step": 172600 }, { "epoch": 2.824347541520085, "grad_norm": 0.010519333183765411, "learning_rate": 2.4085679671613814e-06, "loss": 0.0003, "step": 172610 }, { "epoch": 2.8245111674711607, "grad_norm": 0.06235479563474655, "learning_rate": 2.407957415988381e-06, "loss": 0.0009, "step": 172620 }, { "epoch": 2.8246747934222367, "grad_norm": 0.022449491545557976, "learning_rate": 2.4073469176649285e-06, "loss": 0.0008, "step": 172630 }, { "epoch": 2.8248384193733127, "grad_norm": 0.002296994673088193, "learning_rate": 2.4067364722034653e-06, "loss": 0.0006, "step": 172640 }, { "epoch": 2.8250020453243883, "grad_norm": 0.005234942305833101, "learning_rate": 2.4061260796164413e-06, "loss": 0.0007, "step": 172650 }, { "epoch": 2.8251656712754643, "grad_norm": 0.06974460929632187, "learning_rate": 2.4055157399162975e-06, "loss": 0.0006, "step": 172660 }, { "epoch": 2.8253292972265402, "grad_norm": 0.0858679711818695, "learning_rate": 2.4049054531154847e-06, "loss": 0.0028, "step": 172670 }, { "epoch": 2.825492923177616, "grad_norm": 0.032265909016132355, "learning_rate": 2.4042952192264413e-06, "loss": 0.0007, "step": 172680 }, { "epoch": 2.825656549128692, "grad_norm": 0.006380024831742048, "learning_rate": 2.403685038261612e-06, "loss": 0.0005, "step": 172690 }, { "epoch": 2.825820175079768, "grad_norm": 0.034209124743938446, "learning_rate": 2.403074910233433e-06, "loss": 0.0004, "step": 172700 }, { "epoch": 2.8259838010308433, "grad_norm": 0.018432805314660072, "learning_rate": 2.402464835154352e-06, "loss": 0.0004, "step": 172710 }, { "epoch": 2.8261474269819193, "grad_norm": 0.0444454699754715, "learning_rate": 2.401854813036802e-06, "loss": 0.0008, "step": 172720 }, { "epoch": 2.8263110529329953, "grad_norm": 0.011547407135367393, "learning_rate": 2.4012448438932247e-06, "loss": 0.0006, "step": 172730 }, { "epoch": 2.826474678884071, "grad_norm": 0.09316733479499817, "learning_rate": 2.400634927736051e-06, "loss": 0.0005, "step": 172740 }, { "epoch": 2.826638304835147, "grad_norm": 0.06291438639163971, "learning_rate": 2.400025064577724e-06, "loss": 0.0005, "step": 172750 }, { "epoch": 2.826801930786223, "grad_norm": 0.06861477345228195, "learning_rate": 2.3994152544306725e-06, "loss": 0.0007, "step": 172760 }, { "epoch": 2.8269655567372984, "grad_norm": 0.11881538480520248, "learning_rate": 2.398805497307335e-06, "loss": 0.0014, "step": 172770 }, { "epoch": 2.8271291826883744, "grad_norm": 0.06825726479291916, "learning_rate": 2.398195793220139e-06, "loss": 0.0011, "step": 172780 }, { "epoch": 2.8272928086394504, "grad_norm": 0.012384144589304924, "learning_rate": 2.397586142181518e-06, "loss": 0.0008, "step": 172790 }, { "epoch": 2.827456434590526, "grad_norm": 0.0055611575953662395, "learning_rate": 2.3969765442039017e-06, "loss": 0.0013, "step": 172800 }, { "epoch": 2.827620060541602, "grad_norm": 0.03897547721862793, "learning_rate": 2.3963669992997225e-06, "loss": 0.0005, "step": 172810 }, { "epoch": 2.827783686492678, "grad_norm": 0.01700606755912304, "learning_rate": 2.3957575074814037e-06, "loss": 0.0005, "step": 172820 }, { "epoch": 2.8279473124437535, "grad_norm": 0.05671721696853638, "learning_rate": 2.3951480687613742e-06, "loss": 0.0007, "step": 172830 }, { "epoch": 2.8281109383948295, "grad_norm": 0.01746835559606552, "learning_rate": 2.3945386831520606e-06, "loss": 0.0006, "step": 172840 }, { "epoch": 2.8282745643459055, "grad_norm": 0.08894019573926926, "learning_rate": 2.3939293506658895e-06, "loss": 0.0015, "step": 172850 }, { "epoch": 2.828438190296981, "grad_norm": 0.05195419862866402, "learning_rate": 2.3933200713152806e-06, "loss": 0.0004, "step": 172860 }, { "epoch": 2.828601816248057, "grad_norm": 0.021539919078350067, "learning_rate": 2.3927108451126577e-06, "loss": 0.0005, "step": 172870 }, { "epoch": 2.8287654421991326, "grad_norm": 0.03239240497350693, "learning_rate": 2.3921016720704444e-06, "loss": 0.0003, "step": 172880 }, { "epoch": 2.8289290681502086, "grad_norm": 0.013350249268114567, "learning_rate": 2.3914925522010615e-06, "loss": 0.0006, "step": 172890 }, { "epoch": 2.8290926941012846, "grad_norm": 0.023157907649874687, "learning_rate": 2.390883485516925e-06, "loss": 0.0006, "step": 172900 }, { "epoch": 2.82925632005236, "grad_norm": 0.0012986623914912343, "learning_rate": 2.3902744720304554e-06, "loss": 0.0013, "step": 172910 }, { "epoch": 2.829419946003436, "grad_norm": 0.03631286695599556, "learning_rate": 2.3896655117540695e-06, "loss": 0.0004, "step": 172920 }, { "epoch": 2.8295835719545117, "grad_norm": 0.0038584975991398096, "learning_rate": 2.389056604700186e-06, "loss": 0.0005, "step": 172930 }, { "epoch": 2.8297471979055877, "grad_norm": 0.00498177157714963, "learning_rate": 2.388447750881216e-06, "loss": 0.0014, "step": 172940 }, { "epoch": 2.8299108238566637, "grad_norm": 0.00379383098334074, "learning_rate": 2.387838950309575e-06, "loss": 0.0005, "step": 172950 }, { "epoch": 2.8300744498077393, "grad_norm": 0.06211578845977783, "learning_rate": 2.3872302029976763e-06, "loss": 0.0011, "step": 172960 }, { "epoch": 2.8302380757588153, "grad_norm": 0.006153179332613945, "learning_rate": 2.3866215089579335e-06, "loss": 0.0005, "step": 172970 }, { "epoch": 2.8304017017098912, "grad_norm": 0.041789621114730835, "learning_rate": 2.386012868202754e-06, "loss": 0.0007, "step": 172980 }, { "epoch": 2.830565327660967, "grad_norm": 0.1279471218585968, "learning_rate": 2.3854042807445487e-06, "loss": 0.0008, "step": 172990 }, { "epoch": 2.830728953612043, "grad_norm": 0.012685068883001804, "learning_rate": 2.384795746595727e-06, "loss": 0.0008, "step": 173000 }, { "epoch": 2.830892579563119, "grad_norm": 0.013594155199825764, "learning_rate": 2.384187265768697e-06, "loss": 0.0005, "step": 173010 }, { "epoch": 2.8310562055141943, "grad_norm": 0.0658644437789917, "learning_rate": 2.383578838275863e-06, "loss": 0.0005, "step": 173020 }, { "epoch": 2.8312198314652703, "grad_norm": 0.04077823832631111, "learning_rate": 2.3829704641296302e-06, "loss": 0.0014, "step": 173030 }, { "epoch": 2.8313834574163463, "grad_norm": 0.032639745622873306, "learning_rate": 2.382362143342405e-06, "loss": 0.0007, "step": 173040 }, { "epoch": 2.831547083367422, "grad_norm": 0.038820408284664154, "learning_rate": 2.381753875926591e-06, "loss": 0.0008, "step": 173050 }, { "epoch": 2.831710709318498, "grad_norm": 0.001998211955651641, "learning_rate": 2.381145661894586e-06, "loss": 0.0006, "step": 173060 }, { "epoch": 2.831874335269574, "grad_norm": 0.06598009169101715, "learning_rate": 2.3805375012587943e-06, "loss": 0.0006, "step": 173070 }, { "epoch": 2.8320379612206494, "grad_norm": 0.023548899218440056, "learning_rate": 2.379929394031616e-06, "loss": 0.0004, "step": 173080 }, { "epoch": 2.8322015871717254, "grad_norm": 0.006905536632984877, "learning_rate": 2.37932134022545e-06, "loss": 0.0006, "step": 173090 }, { "epoch": 2.8323652131228014, "grad_norm": 0.004812413360923529, "learning_rate": 2.378713339852691e-06, "loss": 0.0003, "step": 173100 }, { "epoch": 2.832528839073877, "grad_norm": 0.04625322297215462, "learning_rate": 2.3781053929257386e-06, "loss": 0.0006, "step": 173110 }, { "epoch": 2.832692465024953, "grad_norm": 0.007973539642989635, "learning_rate": 2.3774974994569866e-06, "loss": 0.0009, "step": 173120 }, { "epoch": 2.832856090976029, "grad_norm": 0.02520475722849369, "learning_rate": 2.3768896594588326e-06, "loss": 0.0009, "step": 173130 }, { "epoch": 2.8330197169271045, "grad_norm": 0.007843615487217903, "learning_rate": 2.3762818729436653e-06, "loss": 0.0004, "step": 173140 }, { "epoch": 2.8331833428781805, "grad_norm": 0.0026996517553925514, "learning_rate": 2.375674139923879e-06, "loss": 0.0003, "step": 173150 }, { "epoch": 2.8333469688292565, "grad_norm": 0.011708817444741726, "learning_rate": 2.375066460411866e-06, "loss": 0.0002, "step": 173160 }, { "epoch": 2.833510594780332, "grad_norm": 0.01321214996278286, "learning_rate": 2.3744588344200164e-06, "loss": 0.0005, "step": 173170 }, { "epoch": 2.833674220731408, "grad_norm": 0.03258780017495155, "learning_rate": 2.373851261960717e-06, "loss": 0.0005, "step": 173180 }, { "epoch": 2.833837846682484, "grad_norm": 0.009391098283231258, "learning_rate": 2.373243743046357e-06, "loss": 0.0005, "step": 173190 }, { "epoch": 2.8340014726335596, "grad_norm": 0.01093746442347765, "learning_rate": 2.3726362776893223e-06, "loss": 0.0005, "step": 173200 }, { "epoch": 2.8341650985846356, "grad_norm": 0.12021974474191666, "learning_rate": 2.372028865902002e-06, "loss": 0.0007, "step": 173210 }, { "epoch": 2.8343287245357116, "grad_norm": 0.019739963114261627, "learning_rate": 2.371421507696776e-06, "loss": 0.0004, "step": 173220 }, { "epoch": 2.834492350486787, "grad_norm": 0.055235616862773895, "learning_rate": 2.3708142030860316e-06, "loss": 0.0004, "step": 173230 }, { "epoch": 2.834655976437863, "grad_norm": 0.07979182153940201, "learning_rate": 2.3702069520821482e-06, "loss": 0.0007, "step": 173240 }, { "epoch": 2.834819602388939, "grad_norm": 0.04794212058186531, "learning_rate": 2.3695997546975082e-06, "loss": 0.0007, "step": 173250 }, { "epoch": 2.8349832283400147, "grad_norm": 0.0594930574297905, "learning_rate": 2.368992610944494e-06, "loss": 0.0005, "step": 173260 }, { "epoch": 2.8351468542910907, "grad_norm": 0.11810389906167984, "learning_rate": 2.3683855208354807e-06, "loss": 0.0008, "step": 173270 }, { "epoch": 2.8353104802421663, "grad_norm": 0.016899829730391502, "learning_rate": 2.367778484382849e-06, "loss": 0.0004, "step": 173280 }, { "epoch": 2.8354741061932422, "grad_norm": 0.012750225141644478, "learning_rate": 2.367171501598975e-06, "loss": 0.0006, "step": 173290 }, { "epoch": 2.8356377321443182, "grad_norm": 0.025560054928064346, "learning_rate": 2.3665645724962365e-06, "loss": 0.0004, "step": 173300 }, { "epoch": 2.835801358095394, "grad_norm": 0.03752842918038368, "learning_rate": 2.3659576970870056e-06, "loss": 0.0003, "step": 173310 }, { "epoch": 2.83596498404647, "grad_norm": 0.017490828409790993, "learning_rate": 2.365350875383657e-06, "loss": 0.0006, "step": 173320 }, { "epoch": 2.8361286099975453, "grad_norm": 0.0005342302029021084, "learning_rate": 2.364744107398563e-06, "loss": 0.0006, "step": 173330 }, { "epoch": 2.8362922359486213, "grad_norm": 0.052889030426740646, "learning_rate": 2.3641373931440976e-06, "loss": 0.0008, "step": 173340 }, { "epoch": 2.8364558618996973, "grad_norm": 0.005514521151781082, "learning_rate": 2.363530732632627e-06, "loss": 0.0007, "step": 173350 }, { "epoch": 2.836619487850773, "grad_norm": 0.04364633187651634, "learning_rate": 2.3629241258765224e-06, "loss": 0.0005, "step": 173360 }, { "epoch": 2.836783113801849, "grad_norm": 0.03194793686270714, "learning_rate": 2.3623175728881526e-06, "loss": 0.0005, "step": 173370 }, { "epoch": 2.836946739752925, "grad_norm": 0.08494970947504044, "learning_rate": 2.3617110736798854e-06, "loss": 0.0006, "step": 173380 }, { "epoch": 2.8371103657040004, "grad_norm": 0.002064561704173684, "learning_rate": 2.361104628264084e-06, "loss": 0.0004, "step": 173390 }, { "epoch": 2.8372739916550764, "grad_norm": 0.07352980971336365, "learning_rate": 2.3604982366531154e-06, "loss": 0.0003, "step": 173400 }, { "epoch": 2.8374376176061524, "grad_norm": 0.024939943104982376, "learning_rate": 2.359891898859342e-06, "loss": 0.0008, "step": 173410 }, { "epoch": 2.837601243557228, "grad_norm": 0.08491193503141403, "learning_rate": 2.3592856148951306e-06, "loss": 0.0009, "step": 173420 }, { "epoch": 2.837764869508304, "grad_norm": 0.002444449346512556, "learning_rate": 2.358679384772837e-06, "loss": 0.0005, "step": 173430 }, { "epoch": 2.83792849545938, "grad_norm": 0.04713872820138931, "learning_rate": 2.358073208504824e-06, "loss": 0.0014, "step": 173440 }, { "epoch": 2.8380921214104555, "grad_norm": 0.11120638251304626, "learning_rate": 2.357467086103452e-06, "loss": 0.0006, "step": 173450 }, { "epoch": 2.8382557473615315, "grad_norm": 0.025054963305592537, "learning_rate": 2.356861017581081e-06, "loss": 0.0012, "step": 173460 }, { "epoch": 2.8384193733126075, "grad_norm": 0.04053247347474098, "learning_rate": 2.356255002950063e-06, "loss": 0.0004, "step": 173470 }, { "epoch": 2.838582999263683, "grad_norm": 0.026513241231441498, "learning_rate": 2.355649042222758e-06, "loss": 0.0003, "step": 173480 }, { "epoch": 2.838746625214759, "grad_norm": 0.01927364617586136, "learning_rate": 2.3550431354115193e-06, "loss": 0.0007, "step": 173490 }, { "epoch": 2.838910251165835, "grad_norm": 0.036867354065179825, "learning_rate": 2.354437282528704e-06, "loss": 0.0008, "step": 173500 }, { "epoch": 2.8390738771169106, "grad_norm": 0.013064506463706493, "learning_rate": 2.3538314835866603e-06, "loss": 0.0004, "step": 173510 }, { "epoch": 2.8392375030679866, "grad_norm": 0.029395779594779015, "learning_rate": 2.3532257385977433e-06, "loss": 0.0004, "step": 173520 }, { "epoch": 2.8394011290190626, "grad_norm": 0.01321030966937542, "learning_rate": 2.352620047574301e-06, "loss": 0.0017, "step": 173530 }, { "epoch": 2.839564754970138, "grad_norm": 0.020428936928510666, "learning_rate": 2.3520144105286867e-06, "loss": 0.0006, "step": 173540 }, { "epoch": 2.839728380921214, "grad_norm": 0.03390159457921982, "learning_rate": 2.351408827473245e-06, "loss": 0.0004, "step": 173550 }, { "epoch": 2.83989200687229, "grad_norm": 0.07934441417455673, "learning_rate": 2.3508032984203248e-06, "loss": 0.0004, "step": 173560 }, { "epoch": 2.8400556328233657, "grad_norm": 0.03527875989675522, "learning_rate": 2.350197823382272e-06, "loss": 0.001, "step": 173570 }, { "epoch": 2.8402192587744417, "grad_norm": 0.0012392830103635788, "learning_rate": 2.3495924023714338e-06, "loss": 0.0005, "step": 173580 }, { "epoch": 2.8403828847255177, "grad_norm": 0.04558633267879486, "learning_rate": 2.348987035400151e-06, "loss": 0.0005, "step": 173590 }, { "epoch": 2.8405465106765933, "grad_norm": 0.013756968080997467, "learning_rate": 2.3483817224807677e-06, "loss": 0.0007, "step": 173600 }, { "epoch": 2.8407101366276692, "grad_norm": 0.016948334872722626, "learning_rate": 2.3477764636256262e-06, "loss": 0.0004, "step": 173610 }, { "epoch": 2.8408737625787452, "grad_norm": 0.021216467022895813, "learning_rate": 2.3471712588470684e-06, "loss": 0.001, "step": 173620 }, { "epoch": 2.841037388529821, "grad_norm": 0.003504659980535507, "learning_rate": 2.346566108157431e-06, "loss": 0.0007, "step": 173630 }, { "epoch": 2.841201014480897, "grad_norm": 0.07029460370540619, "learning_rate": 2.3459610115690534e-06, "loss": 0.0015, "step": 173640 }, { "epoch": 2.841364640431973, "grad_norm": 0.11247646808624268, "learning_rate": 2.3453559690942738e-06, "loss": 0.0012, "step": 173650 }, { "epoch": 2.8415282663830483, "grad_norm": 0.043061185628175735, "learning_rate": 2.3447509807454305e-06, "loss": 0.0006, "step": 173660 }, { "epoch": 2.8416918923341243, "grad_norm": 0.04615260660648346, "learning_rate": 2.344146046534855e-06, "loss": 0.0007, "step": 173670 }, { "epoch": 2.8418555182852, "grad_norm": 0.0709724947810173, "learning_rate": 2.343541166474882e-06, "loss": 0.0009, "step": 173680 }, { "epoch": 2.842019144236276, "grad_norm": 0.024790601804852486, "learning_rate": 2.3429363405778456e-06, "loss": 0.0011, "step": 173690 }, { "epoch": 2.8421827701873514, "grad_norm": 1.248962163925171, "learning_rate": 2.3423315688560794e-06, "loss": 0.0007, "step": 173700 }, { "epoch": 2.8423463961384274, "grad_norm": 0.04165496677160263, "learning_rate": 2.3417268513219103e-06, "loss": 0.0006, "step": 173710 }, { "epoch": 2.8425100220895034, "grad_norm": 0.02192775532603264, "learning_rate": 2.3411221879876722e-06, "loss": 0.0006, "step": 173720 }, { "epoch": 2.842673648040579, "grad_norm": 0.07488153129816055, "learning_rate": 2.3405175788656875e-06, "loss": 0.0005, "step": 173730 }, { "epoch": 2.842837273991655, "grad_norm": 0.03130171075463295, "learning_rate": 2.339913023968291e-06, "loss": 0.0007, "step": 173740 }, { "epoch": 2.843000899942731, "grad_norm": 0.09436507523059845, "learning_rate": 2.339308523307804e-06, "loss": 0.0013, "step": 173750 }, { "epoch": 2.8431645258938065, "grad_norm": 0.05698055028915405, "learning_rate": 2.3387040768965553e-06, "loss": 0.0009, "step": 173760 }, { "epoch": 2.8433281518448825, "grad_norm": 0.02340175211429596, "learning_rate": 2.338099684746863e-06, "loss": 0.0013, "step": 173770 }, { "epoch": 2.8434917777959585, "grad_norm": 0.011783290654420853, "learning_rate": 2.337495346871058e-06, "loss": 0.0005, "step": 173780 }, { "epoch": 2.843655403747034, "grad_norm": 0.057525403797626495, "learning_rate": 2.3368910632814575e-06, "loss": 0.0011, "step": 173790 }, { "epoch": 2.84381902969811, "grad_norm": 0.045574087649583817, "learning_rate": 2.3362868339903843e-06, "loss": 0.0008, "step": 173800 }, { "epoch": 2.843982655649186, "grad_norm": 0.001156661193817854, "learning_rate": 2.335682659010154e-06, "loss": 0.0005, "step": 173810 }, { "epoch": 2.8441462816002616, "grad_norm": 0.022004876285791397, "learning_rate": 2.3350785383530918e-06, "loss": 0.0004, "step": 173820 }, { "epoch": 2.8443099075513376, "grad_norm": 0.048493023961782455, "learning_rate": 2.33447447203151e-06, "loss": 0.0006, "step": 173830 }, { "epoch": 2.8444735335024136, "grad_norm": 0.0032427157275378704, "learning_rate": 2.3338704600577285e-06, "loss": 0.0011, "step": 173840 }, { "epoch": 2.844637159453489, "grad_norm": 0.028595978394150734, "learning_rate": 2.333266502444057e-06, "loss": 0.0007, "step": 173850 }, { "epoch": 2.844800785404565, "grad_norm": 0.016906851902604103, "learning_rate": 2.3326625992028175e-06, "loss": 0.0002, "step": 173860 }, { "epoch": 2.844964411355641, "grad_norm": 0.04933382570743561, "learning_rate": 2.332058750346317e-06, "loss": 0.0004, "step": 173870 }, { "epoch": 2.8451280373067167, "grad_norm": 0.02152927964925766, "learning_rate": 2.3314549558868717e-06, "loss": 0.0005, "step": 173880 }, { "epoch": 2.8452916632577927, "grad_norm": 0.04579908028244972, "learning_rate": 2.330851215836788e-06, "loss": 0.0015, "step": 173890 }, { "epoch": 2.8454552892088687, "grad_norm": 0.048251960426568985, "learning_rate": 2.330247530208379e-06, "loss": 0.0006, "step": 173900 }, { "epoch": 2.8456189151599443, "grad_norm": 0.0167902372777462, "learning_rate": 2.3296438990139513e-06, "loss": 0.0006, "step": 173910 }, { "epoch": 2.8457825411110202, "grad_norm": 0.2885096073150635, "learning_rate": 2.329040322265816e-06, "loss": 0.0005, "step": 173920 }, { "epoch": 2.8459461670620962, "grad_norm": 0.04847702011466026, "learning_rate": 2.3284367999762743e-06, "loss": 0.0009, "step": 173930 }, { "epoch": 2.846109793013172, "grad_norm": 0.02873728983104229, "learning_rate": 2.3278333321576347e-06, "loss": 0.0006, "step": 173940 }, { "epoch": 2.846273418964248, "grad_norm": 0.023723280057311058, "learning_rate": 2.3272299188222007e-06, "loss": 0.0006, "step": 173950 }, { "epoch": 2.846437044915324, "grad_norm": 0.03886307403445244, "learning_rate": 2.326626559982278e-06, "loss": 0.0011, "step": 173960 }, { "epoch": 2.8466006708663993, "grad_norm": 0.029754851013422012, "learning_rate": 2.3260232556501635e-06, "loss": 0.0004, "step": 173970 }, { "epoch": 2.8467642968174753, "grad_norm": 0.021468183025717735, "learning_rate": 2.325420005838161e-06, "loss": 0.0006, "step": 173980 }, { "epoch": 2.8469279227685513, "grad_norm": 0.04543721303343773, "learning_rate": 2.32481681055857e-06, "loss": 0.0016, "step": 173990 }, { "epoch": 2.847091548719627, "grad_norm": 0.06142931059002876, "learning_rate": 2.3242136698236905e-06, "loss": 0.0005, "step": 174000 }, { "epoch": 2.847255174670703, "grad_norm": 0.03231051564216614, "learning_rate": 2.3236105836458173e-06, "loss": 0.0009, "step": 174010 }, { "epoch": 2.847418800621779, "grad_norm": 0.047320250421762466, "learning_rate": 2.3230075520372473e-06, "loss": 0.0008, "step": 174020 }, { "epoch": 2.8475824265728544, "grad_norm": 0.06868243962526321, "learning_rate": 2.3224045750102776e-06, "loss": 0.0019, "step": 174030 }, { "epoch": 2.8477460525239304, "grad_norm": 0.045330870896577835, "learning_rate": 2.321801652577203e-06, "loss": 0.0005, "step": 174040 }, { "epoch": 2.847909678475006, "grad_norm": 0.0017562631983309984, "learning_rate": 2.3211987847503125e-06, "loss": 0.0004, "step": 174050 }, { "epoch": 2.848073304426082, "grad_norm": 0.01767265796661377, "learning_rate": 2.3205959715419014e-06, "loss": 0.0008, "step": 174060 }, { "epoch": 2.848236930377158, "grad_norm": 0.01622261479496956, "learning_rate": 2.3199932129642595e-06, "loss": 0.0013, "step": 174070 }, { "epoch": 2.8484005563282335, "grad_norm": 0.037053823471069336, "learning_rate": 2.3193905090296783e-06, "loss": 0.0012, "step": 174080 }, { "epoch": 2.8485641822793095, "grad_norm": 0.01308171171694994, "learning_rate": 2.3187878597504426e-06, "loss": 0.0007, "step": 174090 }, { "epoch": 2.848727808230385, "grad_norm": 0.03033013641834259, "learning_rate": 2.3181852651388426e-06, "loss": 0.0009, "step": 174100 }, { "epoch": 2.848891434181461, "grad_norm": 0.011378521099686623, "learning_rate": 2.3175827252071643e-06, "loss": 0.0005, "step": 174110 }, { "epoch": 2.849055060132537, "grad_norm": 0.031412284821271896, "learning_rate": 2.316980239967695e-06, "loss": 0.0006, "step": 174120 }, { "epoch": 2.8492186860836126, "grad_norm": 0.032679129391908646, "learning_rate": 2.3163778094327142e-06, "loss": 0.0013, "step": 174130 }, { "epoch": 2.8493823120346886, "grad_norm": 0.018184101209044456, "learning_rate": 2.3157754336145085e-06, "loss": 0.0002, "step": 174140 }, { "epoch": 2.8495459379857646, "grad_norm": 0.03716189041733742, "learning_rate": 2.315173112525358e-06, "loss": 0.0004, "step": 174150 }, { "epoch": 2.84970956393684, "grad_norm": 0.06526254862546921, "learning_rate": 2.3145708461775467e-06, "loss": 0.0007, "step": 174160 }, { "epoch": 2.849873189887916, "grad_norm": 0.027591105550527573, "learning_rate": 2.31396863458335e-06, "loss": 0.001, "step": 174170 }, { "epoch": 2.850036815838992, "grad_norm": 0.03384597226977348, "learning_rate": 2.3133664777550486e-06, "loss": 0.0009, "step": 174180 }, { "epoch": 2.8502004417900677, "grad_norm": 0.03446284309029579, "learning_rate": 2.3127643757049196e-06, "loss": 0.0005, "step": 174190 }, { "epoch": 2.8503640677411437, "grad_norm": 0.006541563663631678, "learning_rate": 2.3121623284452414e-06, "loss": 0.0004, "step": 174200 }, { "epoch": 2.8505276936922197, "grad_norm": 0.026075327768921852, "learning_rate": 2.3115603359882853e-06, "loss": 0.0006, "step": 174210 }, { "epoch": 2.8506913196432953, "grad_norm": 0.029440173879265785, "learning_rate": 2.310958398346328e-06, "loss": 0.0003, "step": 174220 }, { "epoch": 2.8508549455943712, "grad_norm": 0.010573618113994598, "learning_rate": 2.310356515531642e-06, "loss": 0.0003, "step": 174230 }, { "epoch": 2.8510185715454472, "grad_norm": 0.1846601963043213, "learning_rate": 2.3097546875565006e-06, "loss": 0.0017, "step": 174240 }, { "epoch": 2.851182197496523, "grad_norm": 0.2698554992675781, "learning_rate": 2.309152914433172e-06, "loss": 0.0005, "step": 174250 }, { "epoch": 2.851345823447599, "grad_norm": 0.017568383365869522, "learning_rate": 2.3085511961739266e-06, "loss": 0.0004, "step": 174260 }, { "epoch": 2.851509449398675, "grad_norm": 0.13459351658821106, "learning_rate": 2.3079495327910327e-06, "loss": 0.0009, "step": 174270 }, { "epoch": 2.8516730753497503, "grad_norm": 0.08898796886205673, "learning_rate": 2.307347924296761e-06, "loss": 0.0005, "step": 174280 }, { "epoch": 2.8518367013008263, "grad_norm": 0.13908684253692627, "learning_rate": 2.3067463707033733e-06, "loss": 0.001, "step": 174290 }, { "epoch": 2.8520003272519023, "grad_norm": 0.0369507297873497, "learning_rate": 2.3061448720231362e-06, "loss": 0.0005, "step": 174300 }, { "epoch": 2.852163953202978, "grad_norm": 0.03200801461935043, "learning_rate": 2.3055434282683144e-06, "loss": 0.0006, "step": 174310 }, { "epoch": 2.852327579154054, "grad_norm": 0.024837248027324677, "learning_rate": 2.304942039451172e-06, "loss": 0.0002, "step": 174320 }, { "epoch": 2.85249120510513, "grad_norm": 0.013637878932058811, "learning_rate": 2.3043407055839672e-06, "loss": 0.0006, "step": 174330 }, { "epoch": 2.8526548310562054, "grad_norm": 0.03434496745467186, "learning_rate": 2.3037394266789652e-06, "loss": 0.0005, "step": 174340 }, { "epoch": 2.8528184570072814, "grad_norm": 0.035956352949142456, "learning_rate": 2.30313820274842e-06, "loss": 0.0011, "step": 174350 }, { "epoch": 2.8529820829583574, "grad_norm": 0.0015779188834130764, "learning_rate": 2.302537033804596e-06, "loss": 0.0007, "step": 174360 }, { "epoch": 2.853145708909433, "grad_norm": 0.019278181716799736, "learning_rate": 2.3019359198597464e-06, "loss": 0.0005, "step": 174370 }, { "epoch": 2.853309334860509, "grad_norm": 0.019276132807135582, "learning_rate": 2.3013348609261303e-06, "loss": 0.0012, "step": 174380 }, { "epoch": 2.853472960811585, "grad_norm": 0.09967219084501266, "learning_rate": 2.300733857015998e-06, "loss": 0.0006, "step": 174390 }, { "epoch": 2.8536365867626605, "grad_norm": 0.10925410687923431, "learning_rate": 2.3001329081416105e-06, "loss": 0.0013, "step": 174400 }, { "epoch": 2.8538002127137365, "grad_norm": 0.025548089295625687, "learning_rate": 2.2995320143152145e-06, "loss": 0.0005, "step": 174410 }, { "epoch": 2.8539638386648125, "grad_norm": 0.047986265271902084, "learning_rate": 2.298931175549066e-06, "loss": 0.0007, "step": 174420 }, { "epoch": 2.854127464615888, "grad_norm": 0.047625988721847534, "learning_rate": 2.2983303918554104e-06, "loss": 0.0004, "step": 174430 }, { "epoch": 2.854291090566964, "grad_norm": 0.04640411585569382, "learning_rate": 2.297729663246504e-06, "loss": 0.0006, "step": 174440 }, { "epoch": 2.8544547165180396, "grad_norm": 0.028615886345505714, "learning_rate": 2.2971289897345893e-06, "loss": 0.0008, "step": 174450 }, { "epoch": 2.8546183424691156, "grad_norm": 0.013398952782154083, "learning_rate": 2.296528371331918e-06, "loss": 0.0005, "step": 174460 }, { "epoch": 2.8547819684201916, "grad_norm": 0.06776116043329239, "learning_rate": 2.29592780805073e-06, "loss": 0.001, "step": 174470 }, { "epoch": 2.854945594371267, "grad_norm": 0.010676495730876923, "learning_rate": 2.2953272999032784e-06, "loss": 0.0006, "step": 174480 }, { "epoch": 2.855109220322343, "grad_norm": 0.014679946936666965, "learning_rate": 2.294726846901801e-06, "loss": 0.0006, "step": 174490 }, { "epoch": 2.8552728462734187, "grad_norm": 0.027730610221624374, "learning_rate": 2.2941264490585434e-06, "loss": 0.0014, "step": 174500 }, { "epoch": 2.8554364722244947, "grad_norm": 0.04090607166290283, "learning_rate": 2.2935261063857432e-06, "loss": 0.0007, "step": 174510 }, { "epoch": 2.8556000981755707, "grad_norm": 0.017909478396177292, "learning_rate": 2.2929258188956484e-06, "loss": 0.0004, "step": 174520 }, { "epoch": 2.8557637241266463, "grad_norm": 0.018551494926214218, "learning_rate": 2.292325586600491e-06, "loss": 0.0005, "step": 174530 }, { "epoch": 2.8559273500777222, "grad_norm": 0.1606140285730362, "learning_rate": 2.291725409512514e-06, "loss": 0.0005, "step": 174540 }, { "epoch": 2.8560909760287982, "grad_norm": 0.011323581449687481, "learning_rate": 2.291125287643951e-06, "loss": 0.0008, "step": 174550 }, { "epoch": 2.856254601979874, "grad_norm": 0.0017693180125206709, "learning_rate": 2.2905252210070395e-06, "loss": 0.0007, "step": 174560 }, { "epoch": 2.85641822793095, "grad_norm": 0.0012137953890487552, "learning_rate": 2.2899252096140143e-06, "loss": 0.0008, "step": 174570 }, { "epoch": 2.856581853882026, "grad_norm": 0.007465306203812361, "learning_rate": 2.2893252534771115e-06, "loss": 0.0003, "step": 174580 }, { "epoch": 2.8567454798331013, "grad_norm": 0.019055761396884918, "learning_rate": 2.288725352608559e-06, "loss": 0.0008, "step": 174590 }, { "epoch": 2.8569091057841773, "grad_norm": 0.010106937028467655, "learning_rate": 2.288125507020591e-06, "loss": 0.0006, "step": 174600 }, { "epoch": 2.8570727317352533, "grad_norm": 0.03133446350693703, "learning_rate": 2.287525716725438e-06, "loss": 0.001, "step": 174610 }, { "epoch": 2.857236357686329, "grad_norm": 0.010340171866118908, "learning_rate": 2.2869259817353302e-06, "loss": 0.0006, "step": 174620 }, { "epoch": 2.857399983637405, "grad_norm": 0.003062963718548417, "learning_rate": 2.286326302062492e-06, "loss": 0.0004, "step": 174630 }, { "epoch": 2.857563609588481, "grad_norm": 0.04267513379454613, "learning_rate": 2.2857266777191535e-06, "loss": 0.0007, "step": 174640 }, { "epoch": 2.8577272355395564, "grad_norm": 0.0007449171971529722, "learning_rate": 2.2851271087175386e-06, "loss": 0.0009, "step": 174650 }, { "epoch": 2.8578908614906324, "grad_norm": 0.011619857512414455, "learning_rate": 2.2845275950698753e-06, "loss": 0.0007, "step": 174660 }, { "epoch": 2.8580544874417084, "grad_norm": 0.08178664743900299, "learning_rate": 2.283928136788384e-06, "loss": 0.001, "step": 174670 }, { "epoch": 2.858218113392784, "grad_norm": 0.01612279750406742, "learning_rate": 2.283328733885287e-06, "loss": 0.0005, "step": 174680 }, { "epoch": 2.85838173934386, "grad_norm": 0.05442840978503227, "learning_rate": 2.2827293863728066e-06, "loss": 0.0006, "step": 174690 }, { "epoch": 2.858545365294936, "grad_norm": 0.026626987382769585, "learning_rate": 2.2821300942631653e-06, "loss": 0.0005, "step": 174700 }, { "epoch": 2.8587089912460115, "grad_norm": 0.03436988964676857, "learning_rate": 2.2815308575685778e-06, "loss": 0.0006, "step": 174710 }, { "epoch": 2.8588726171970875, "grad_norm": 0.0392470508813858, "learning_rate": 2.280931676301264e-06, "loss": 0.0012, "step": 174720 }, { "epoch": 2.8590362431481635, "grad_norm": 0.030494509264826775, "learning_rate": 2.2803325504734416e-06, "loss": 0.0005, "step": 174730 }, { "epoch": 2.859199869099239, "grad_norm": 0.06203418970108032, "learning_rate": 2.2797334800973265e-06, "loss": 0.0004, "step": 174740 }, { "epoch": 2.859363495050315, "grad_norm": 0.010004338808357716, "learning_rate": 2.2791344651851312e-06, "loss": 0.0004, "step": 174750 }, { "epoch": 2.859527121001391, "grad_norm": 0.027094438672065735, "learning_rate": 2.278535505749069e-06, "loss": 0.0006, "step": 174760 }, { "epoch": 2.8596907469524666, "grad_norm": 0.011343895457684994, "learning_rate": 2.2779366018013545e-06, "loss": 0.0008, "step": 174770 }, { "epoch": 2.8598543729035426, "grad_norm": 0.04124109819531441, "learning_rate": 2.277337753354199e-06, "loss": 0.0004, "step": 174780 }, { "epoch": 2.8600179988546186, "grad_norm": 0.003253738395869732, "learning_rate": 2.276738960419809e-06, "loss": 0.0007, "step": 174790 }, { "epoch": 2.860181624805694, "grad_norm": 0.01941099762916565, "learning_rate": 2.276140223010396e-06, "loss": 0.0004, "step": 174800 }, { "epoch": 2.86034525075677, "grad_norm": 0.09988400340080261, "learning_rate": 2.275541541138167e-06, "loss": 0.0009, "step": 174810 }, { "epoch": 2.8605088767078457, "grad_norm": 0.2590123414993286, "learning_rate": 2.274942914815331e-06, "loss": 0.0038, "step": 174820 }, { "epoch": 2.8606725026589217, "grad_norm": 0.02118588611483574, "learning_rate": 2.2743443440540897e-06, "loss": 0.0002, "step": 174830 }, { "epoch": 2.8608361286099977, "grad_norm": 0.0008085328154265881, "learning_rate": 2.2737458288666487e-06, "loss": 0.0003, "step": 174840 }, { "epoch": 2.8609997545610732, "grad_norm": 0.04413025081157684, "learning_rate": 2.2731473692652125e-06, "loss": 0.0006, "step": 174850 }, { "epoch": 2.8611633805121492, "grad_norm": 0.0017601181752979755, "learning_rate": 2.272548965261983e-06, "loss": 0.0003, "step": 174860 }, { "epoch": 2.861327006463225, "grad_norm": 0.020222920924425125, "learning_rate": 2.271950616869159e-06, "loss": 0.0005, "step": 174870 }, { "epoch": 2.861490632414301, "grad_norm": 0.0014279830502346158, "learning_rate": 2.271352324098942e-06, "loss": 0.0003, "step": 174880 }, { "epoch": 2.861654258365377, "grad_norm": 0.0012868272606283426, "learning_rate": 2.270754086963531e-06, "loss": 0.0005, "step": 174890 }, { "epoch": 2.8618178843164523, "grad_norm": 0.020704858005046844, "learning_rate": 2.2701559054751244e-06, "loss": 0.0014, "step": 174900 }, { "epoch": 2.8619815102675283, "grad_norm": 0.011527355760335922, "learning_rate": 2.2695577796459152e-06, "loss": 0.0004, "step": 174910 }, { "epoch": 2.8621451362186043, "grad_norm": 0.053227610886096954, "learning_rate": 2.2689597094881007e-06, "loss": 0.0006, "step": 174920 }, { "epoch": 2.86230876216968, "grad_norm": 0.040396880358457565, "learning_rate": 2.2683616950138755e-06, "loss": 0.0007, "step": 174930 }, { "epoch": 2.862472388120756, "grad_norm": 0.00951777957379818, "learning_rate": 2.267763736235433e-06, "loss": 0.001, "step": 174940 }, { "epoch": 2.862636014071832, "grad_norm": 0.033776603639125824, "learning_rate": 2.267165833164963e-06, "loss": 0.0008, "step": 174950 }, { "epoch": 2.8627996400229074, "grad_norm": 0.004679510835558176, "learning_rate": 2.266567985814658e-06, "loss": 0.0005, "step": 174960 }, { "epoch": 2.8629632659739834, "grad_norm": 0.018288597464561462, "learning_rate": 2.2659701941967065e-06, "loss": 0.0004, "step": 174970 }, { "epoch": 2.8631268919250594, "grad_norm": 0.01061110571026802, "learning_rate": 2.2653724583232995e-06, "loss": 0.0006, "step": 174980 }, { "epoch": 2.863290517876135, "grad_norm": 0.009474324062466621, "learning_rate": 2.264774778206621e-06, "loss": 0.0004, "step": 174990 }, { "epoch": 2.863454143827211, "grad_norm": 0.043305814266204834, "learning_rate": 2.2641771538588595e-06, "loss": 0.0003, "step": 175000 }, { "epoch": 2.863617769778287, "grad_norm": 0.07257575541734695, "learning_rate": 2.263579585292196e-06, "loss": 0.0009, "step": 175010 }, { "epoch": 2.8637813957293625, "grad_norm": 0.022039253264665604, "learning_rate": 2.262982072518821e-06, "loss": 0.0007, "step": 175020 }, { "epoch": 2.8639450216804385, "grad_norm": 0.06883624941110611, "learning_rate": 2.2623846155509123e-06, "loss": 0.0006, "step": 175030 }, { "epoch": 2.8641086476315145, "grad_norm": 0.006473545450717211, "learning_rate": 2.2617872144006543e-06, "loss": 0.0005, "step": 175040 }, { "epoch": 2.86427227358259, "grad_norm": 0.039158184081315994, "learning_rate": 2.2611898690802225e-06, "loss": 0.0007, "step": 175050 }, { "epoch": 2.864435899533666, "grad_norm": 0.03971344977617264, "learning_rate": 2.260592579601804e-06, "loss": 0.0015, "step": 175060 }, { "epoch": 2.864599525484742, "grad_norm": 0.08345098793506622, "learning_rate": 2.2599953459775712e-06, "loss": 0.0012, "step": 175070 }, { "epoch": 2.8647631514358176, "grad_norm": 0.0027943355962634087, "learning_rate": 2.2593981682197047e-06, "loss": 0.0008, "step": 175080 }, { "epoch": 2.8649267773868936, "grad_norm": 0.004059444181621075, "learning_rate": 2.258801046340375e-06, "loss": 0.0008, "step": 175090 }, { "epoch": 2.8650904033379696, "grad_norm": 0.05732345208525658, "learning_rate": 2.2582039803517646e-06, "loss": 0.0012, "step": 175100 }, { "epoch": 2.865254029289045, "grad_norm": 0.048109423369169235, "learning_rate": 2.2576069702660415e-06, "loss": 0.001, "step": 175110 }, { "epoch": 2.865417655240121, "grad_norm": 0.034847140312194824, "learning_rate": 2.257010016095382e-06, "loss": 0.0007, "step": 175120 }, { "epoch": 2.865581281191197, "grad_norm": 0.048322129994630814, "learning_rate": 2.256413117851951e-06, "loss": 0.0007, "step": 175130 }, { "epoch": 2.8657449071422727, "grad_norm": 0.006009256467223167, "learning_rate": 2.2558162755479273e-06, "loss": 0.0008, "step": 175140 }, { "epoch": 2.8659085330933487, "grad_norm": 0.02710801362991333, "learning_rate": 2.255219489195474e-06, "loss": 0.0007, "step": 175150 }, { "epoch": 2.8660721590444247, "grad_norm": 0.03965028002858162, "learning_rate": 2.2546227588067626e-06, "loss": 0.0013, "step": 175160 }, { "epoch": 2.8662357849955002, "grad_norm": 0.038779377937316895, "learning_rate": 2.2540260843939564e-06, "loss": 0.0009, "step": 175170 }, { "epoch": 2.8663994109465762, "grad_norm": 0.04364139959216118, "learning_rate": 2.2534294659692233e-06, "loss": 0.0006, "step": 175180 }, { "epoch": 2.8665630368976522, "grad_norm": 0.04585621878504753, "learning_rate": 2.2528329035447276e-06, "loss": 0.0012, "step": 175190 }, { "epoch": 2.866726662848728, "grad_norm": 0.015306498855352402, "learning_rate": 2.2522363971326344e-06, "loss": 0.0005, "step": 175200 }, { "epoch": 2.866890288799804, "grad_norm": 0.07417362183332443, "learning_rate": 2.2516399467451023e-06, "loss": 0.002, "step": 175210 }, { "epoch": 2.8670539147508793, "grad_norm": 0.06354357302188873, "learning_rate": 2.251043552394294e-06, "loss": 0.0008, "step": 175220 }, { "epoch": 2.8672175407019553, "grad_norm": 0.003958777990192175, "learning_rate": 2.25044721409237e-06, "loss": 0.0005, "step": 175230 }, { "epoch": 2.8673811666530313, "grad_norm": 0.0011055766372010112, "learning_rate": 2.2498509318514907e-06, "loss": 0.0018, "step": 175240 }, { "epoch": 2.867544792604107, "grad_norm": 0.04239998012781143, "learning_rate": 2.2492547056838102e-06, "loss": 0.0008, "step": 175250 }, { "epoch": 2.867708418555183, "grad_norm": 0.013224292546510696, "learning_rate": 2.2486585356014868e-06, "loss": 0.0005, "step": 175260 }, { "epoch": 2.8678720445062584, "grad_norm": 0.03243719041347504, "learning_rate": 2.2480624216166756e-06, "loss": 0.0006, "step": 175270 }, { "epoch": 2.8680356704573344, "grad_norm": 0.03441484645009041, "learning_rate": 2.2474663637415334e-06, "loss": 0.0009, "step": 175280 }, { "epoch": 2.8681992964084104, "grad_norm": 0.7978865504264832, "learning_rate": 2.246870361988209e-06, "loss": 0.0009, "step": 175290 }, { "epoch": 2.868362922359486, "grad_norm": 0.03794480860233307, "learning_rate": 2.246274416368857e-06, "loss": 0.0007, "step": 175300 }, { "epoch": 2.868526548310562, "grad_norm": 0.006940773688256741, "learning_rate": 2.245678526895627e-06, "loss": 0.0006, "step": 175310 }, { "epoch": 2.868690174261638, "grad_norm": 0.045093320310115814, "learning_rate": 2.2450826935806717e-06, "loss": 0.0006, "step": 175320 }, { "epoch": 2.8688538002127135, "grad_norm": 0.03698210418224335, "learning_rate": 2.2444869164361354e-06, "loss": 0.0003, "step": 175330 }, { "epoch": 2.8690174261637895, "grad_norm": 0.10928162187337875, "learning_rate": 2.243891195474167e-06, "loss": 0.0011, "step": 175340 }, { "epoch": 2.8691810521148655, "grad_norm": 0.1231827363371849, "learning_rate": 2.2432955307069136e-06, "loss": 0.0011, "step": 175350 }, { "epoch": 2.869344678065941, "grad_norm": 0.040801532566547394, "learning_rate": 2.242699922146522e-06, "loss": 0.0006, "step": 175360 }, { "epoch": 2.869508304017017, "grad_norm": 0.02608460560441017, "learning_rate": 2.2421043698051317e-06, "loss": 0.0008, "step": 175370 }, { "epoch": 2.869671929968093, "grad_norm": 0.0220428928732872, "learning_rate": 2.2415088736948876e-06, "loss": 0.0005, "step": 175380 }, { "epoch": 2.8698355559191686, "grad_norm": 0.06557993590831757, "learning_rate": 2.2409134338279314e-06, "loss": 0.0008, "step": 175390 }, { "epoch": 2.8699991818702446, "grad_norm": 0.005883175879716873, "learning_rate": 2.2403180502164063e-06, "loss": 0.0004, "step": 175400 }, { "epoch": 2.8701628078213206, "grad_norm": 0.07849841564893723, "learning_rate": 2.2397227228724464e-06, "loss": 0.0005, "step": 175410 }, { "epoch": 2.870326433772396, "grad_norm": 0.0036288704723119736, "learning_rate": 2.2391274518081936e-06, "loss": 0.0006, "step": 175420 }, { "epoch": 2.870490059723472, "grad_norm": 0.016283953562378883, "learning_rate": 2.238532237035783e-06, "loss": 0.0005, "step": 175430 }, { "epoch": 2.870653685674548, "grad_norm": 0.07091300189495087, "learning_rate": 2.2379370785673543e-06, "loss": 0.0011, "step": 175440 }, { "epoch": 2.8708173116256237, "grad_norm": 0.04566425457596779, "learning_rate": 2.2373419764150372e-06, "loss": 0.0007, "step": 175450 }, { "epoch": 2.8709809375766997, "grad_norm": 0.03665918856859207, "learning_rate": 2.236746930590968e-06, "loss": 0.0009, "step": 175460 }, { "epoch": 2.8711445635277757, "grad_norm": 0.09952978044748306, "learning_rate": 2.2361519411072783e-06, "loss": 0.0006, "step": 175470 }, { "epoch": 2.8713081894788512, "grad_norm": 0.03790082037448883, "learning_rate": 2.2355570079761024e-06, "loss": 0.0008, "step": 175480 }, { "epoch": 2.8714718154299272, "grad_norm": 0.04451390355825424, "learning_rate": 2.234962131209566e-06, "loss": 0.0008, "step": 175490 }, { "epoch": 2.8716354413810032, "grad_norm": 0.06040722504258156, "learning_rate": 2.2343673108198e-06, "loss": 0.0005, "step": 175500 }, { "epoch": 2.871799067332079, "grad_norm": 0.025613848119974136, "learning_rate": 2.233772546818932e-06, "loss": 0.0005, "step": 175510 }, { "epoch": 2.871962693283155, "grad_norm": 0.03326615318655968, "learning_rate": 2.2331778392190917e-06, "loss": 0.0007, "step": 175520 }, { "epoch": 2.872126319234231, "grad_norm": 0.09174437820911407, "learning_rate": 2.2325831880324e-06, "loss": 0.0019, "step": 175530 }, { "epoch": 2.8722899451853063, "grad_norm": 0.04758554697036743, "learning_rate": 2.231988593270984e-06, "loss": 0.0003, "step": 175540 }, { "epoch": 2.8724535711363823, "grad_norm": 0.009677677415311337, "learning_rate": 2.231394054946966e-06, "loss": 0.0004, "step": 175550 }, { "epoch": 2.8726171970874583, "grad_norm": 0.00322204758413136, "learning_rate": 2.2307995730724704e-06, "loss": 0.0003, "step": 175560 }, { "epoch": 2.872780823038534, "grad_norm": 0.03345223888754845, "learning_rate": 2.230205147659615e-06, "loss": 0.0003, "step": 175570 }, { "epoch": 2.87294444898961, "grad_norm": 0.05529598891735077, "learning_rate": 2.2296107787205202e-06, "loss": 0.0011, "step": 175580 }, { "epoch": 2.8731080749406854, "grad_norm": 0.029616044834256172, "learning_rate": 2.229016466267306e-06, "loss": 0.0004, "step": 175590 }, { "epoch": 2.8732717008917614, "grad_norm": 0.044271912425756454, "learning_rate": 2.228422210312091e-06, "loss": 0.0006, "step": 175600 }, { "epoch": 2.8734353268428374, "grad_norm": 0.01942901499569416, "learning_rate": 2.2278280108669885e-06, "loss": 0.0005, "step": 175610 }, { "epoch": 2.873598952793913, "grad_norm": 0.033136505633592606, "learning_rate": 2.2272338679441175e-06, "loss": 0.0006, "step": 175620 }, { "epoch": 2.873762578744989, "grad_norm": 0.0008867769502103329, "learning_rate": 2.226639781555585e-06, "loss": 0.001, "step": 175630 }, { "epoch": 2.8739262046960645, "grad_norm": 0.03511272370815277, "learning_rate": 2.226045751713513e-06, "loss": 0.0012, "step": 175640 }, { "epoch": 2.8740898306471405, "grad_norm": 0.018394554033875465, "learning_rate": 2.225451778430007e-06, "loss": 0.0009, "step": 175650 }, { "epoch": 2.8742534565982165, "grad_norm": 0.10835187137126923, "learning_rate": 2.224857861717182e-06, "loss": 0.0007, "step": 175660 }, { "epoch": 2.874417082549292, "grad_norm": 0.026466598734259605, "learning_rate": 2.22426400158714e-06, "loss": 0.0008, "step": 175670 }, { "epoch": 2.874580708500368, "grad_norm": 0.01566237024962902, "learning_rate": 2.2236701980519994e-06, "loss": 0.0004, "step": 175680 }, { "epoch": 2.874744334451444, "grad_norm": 0.017786692827939987, "learning_rate": 2.2230764511238597e-06, "loss": 0.0012, "step": 175690 }, { "epoch": 2.8749079604025196, "grad_norm": 0.030115611851215363, "learning_rate": 2.2224827608148313e-06, "loss": 0.0005, "step": 175700 }, { "epoch": 2.8750715863535956, "grad_norm": 0.01851591095328331, "learning_rate": 2.2218891271370143e-06, "loss": 0.0006, "step": 175710 }, { "epoch": 2.8752352123046716, "grad_norm": 0.07412593811750412, "learning_rate": 2.221295550102518e-06, "loss": 0.0003, "step": 175720 }, { "epoch": 2.875398838255747, "grad_norm": 0.03682740405201912, "learning_rate": 2.2207020297234405e-06, "loss": 0.0005, "step": 175730 }, { "epoch": 2.875562464206823, "grad_norm": 0.0819416418671608, "learning_rate": 2.220108566011887e-06, "loss": 0.0014, "step": 175740 }, { "epoch": 2.875726090157899, "grad_norm": 0.04786141589283943, "learning_rate": 2.2195151589799517e-06, "loss": 0.0008, "step": 175750 }, { "epoch": 2.8758897161089747, "grad_norm": 0.024236509576439857, "learning_rate": 2.218921808639742e-06, "loss": 0.0004, "step": 175760 }, { "epoch": 2.8760533420600507, "grad_norm": 0.05020748823881149, "learning_rate": 2.218328515003349e-06, "loss": 0.0009, "step": 175770 }, { "epoch": 2.8762169680111267, "grad_norm": 0.07561156898736954, "learning_rate": 2.2177352780828735e-06, "loss": 0.0009, "step": 175780 }, { "epoch": 2.8763805939622022, "grad_norm": 0.02699074149131775, "learning_rate": 2.217142097890406e-06, "loss": 0.0003, "step": 175790 }, { "epoch": 2.8765442199132782, "grad_norm": 0.039405860006809235, "learning_rate": 2.2165489744380485e-06, "loss": 0.0009, "step": 175800 }, { "epoch": 2.8767078458643542, "grad_norm": 0.016343625262379646, "learning_rate": 2.215955907737888e-06, "loss": 0.0003, "step": 175810 }, { "epoch": 2.87687147181543, "grad_norm": 0.05038519203662872, "learning_rate": 2.2153628978020203e-06, "loss": 0.0007, "step": 175820 }, { "epoch": 2.877035097766506, "grad_norm": 0.04019506648182869, "learning_rate": 2.2147699446425336e-06, "loss": 0.001, "step": 175830 }, { "epoch": 2.877198723717582, "grad_norm": 0.028700944036245346, "learning_rate": 2.2141770482715186e-06, "loss": 0.0006, "step": 175840 }, { "epoch": 2.8773623496686573, "grad_norm": 0.03360706567764282, "learning_rate": 2.2135842087010644e-06, "loss": 0.0009, "step": 175850 }, { "epoch": 2.8775259756197333, "grad_norm": 0.01729968935251236, "learning_rate": 2.21299142594326e-06, "loss": 0.0006, "step": 175860 }, { "epoch": 2.8776896015708093, "grad_norm": 0.006647580768913031, "learning_rate": 2.2123987000101887e-06, "loss": 0.0013, "step": 175870 }, { "epoch": 2.877853227521885, "grad_norm": 0.021263636648654938, "learning_rate": 2.2118060309139373e-06, "loss": 0.0004, "step": 175880 }, { "epoch": 2.878016853472961, "grad_norm": 0.05999389663338661, "learning_rate": 2.211213418666589e-06, "loss": 0.0003, "step": 175890 }, { "epoch": 2.878180479424037, "grad_norm": 0.0827220007777214, "learning_rate": 2.2106208632802295e-06, "loss": 0.0011, "step": 175900 }, { "epoch": 2.8783441053751124, "grad_norm": 0.07457917928695679, "learning_rate": 2.2100283647669367e-06, "loss": 0.0006, "step": 175910 }, { "epoch": 2.8785077313261884, "grad_norm": 0.03973839432001114, "learning_rate": 2.2094359231387922e-06, "loss": 0.0006, "step": 175920 }, { "epoch": 2.8786713572772644, "grad_norm": 0.14332294464111328, "learning_rate": 2.2088435384078765e-06, "loss": 0.0007, "step": 175930 }, { "epoch": 2.87883498322834, "grad_norm": 0.026491541415452957, "learning_rate": 2.208251210586269e-06, "loss": 0.0006, "step": 175940 }, { "epoch": 2.878998609179416, "grad_norm": 0.025097116827964783, "learning_rate": 2.207658939686043e-06, "loss": 0.0006, "step": 175950 }, { "epoch": 2.879162235130492, "grad_norm": 0.010207563638687134, "learning_rate": 2.207066725719277e-06, "loss": 0.0008, "step": 175960 }, { "epoch": 2.8793258610815675, "grad_norm": 0.0328621044754982, "learning_rate": 2.206474568698044e-06, "loss": 0.0005, "step": 175970 }, { "epoch": 2.8794894870326435, "grad_norm": 0.05989981070160866, "learning_rate": 2.2058824686344216e-06, "loss": 0.0009, "step": 175980 }, { "epoch": 2.879653112983719, "grad_norm": 0.02976989932358265, "learning_rate": 2.205290425540477e-06, "loss": 0.0007, "step": 175990 }, { "epoch": 2.879816738934795, "grad_norm": 0.010792920365929604, "learning_rate": 2.2046984394282838e-06, "loss": 0.0008, "step": 176000 }, { "epoch": 2.879980364885871, "grad_norm": 0.005182221997529268, "learning_rate": 2.2041065103099123e-06, "loss": 0.001, "step": 176010 }, { "epoch": 2.8801439908369466, "grad_norm": 0.008852160535752773, "learning_rate": 2.2035146381974333e-06, "loss": 0.0004, "step": 176020 }, { "epoch": 2.8803076167880226, "grad_norm": 0.07492401450872421, "learning_rate": 2.2029228231029103e-06, "loss": 0.0009, "step": 176030 }, { "epoch": 2.880471242739098, "grad_norm": 0.06248025596141815, "learning_rate": 2.202331065038412e-06, "loss": 0.0014, "step": 176040 }, { "epoch": 2.880634868690174, "grad_norm": 0.0027911767829209566, "learning_rate": 2.2017393640160044e-06, "loss": 0.0007, "step": 176050 }, { "epoch": 2.88079849464125, "grad_norm": 0.016068818047642708, "learning_rate": 2.2011477200477526e-06, "loss": 0.0007, "step": 176060 }, { "epoch": 2.8809621205923257, "grad_norm": 0.06993944197893143, "learning_rate": 2.200556133145717e-06, "loss": 0.0006, "step": 176070 }, { "epoch": 2.8811257465434017, "grad_norm": 0.06153310090303421, "learning_rate": 2.199964603321961e-06, "loss": 0.001, "step": 176080 }, { "epoch": 2.8812893724944777, "grad_norm": 0.05118010193109512, "learning_rate": 2.1993731305885454e-06, "loss": 0.0006, "step": 176090 }, { "epoch": 2.8814529984455532, "grad_norm": 0.03826426342129707, "learning_rate": 2.1987817149575313e-06, "loss": 0.0007, "step": 176100 }, { "epoch": 2.8816166243966292, "grad_norm": 0.02796855755150318, "learning_rate": 2.1981903564409744e-06, "loss": 0.0014, "step": 176110 }, { "epoch": 2.8817802503477052, "grad_norm": 0.043664976954460144, "learning_rate": 2.197599055050933e-06, "loss": 0.0004, "step": 176120 }, { "epoch": 2.881943876298781, "grad_norm": 0.03758490830659866, "learning_rate": 2.1970078107994636e-06, "loss": 0.0003, "step": 176130 }, { "epoch": 2.882107502249857, "grad_norm": 0.026299992576241493, "learning_rate": 2.1964166236986227e-06, "loss": 0.0009, "step": 176140 }, { "epoch": 2.882271128200933, "grad_norm": 0.09246676415205002, "learning_rate": 2.1958254937604605e-06, "loss": 0.0008, "step": 176150 }, { "epoch": 2.8824347541520083, "grad_norm": 0.08046997338533401, "learning_rate": 2.1952344209970324e-06, "loss": 0.0006, "step": 176160 }, { "epoch": 2.8825983801030843, "grad_norm": 0.035155631601810455, "learning_rate": 2.1946434054203887e-06, "loss": 0.0008, "step": 176170 }, { "epoch": 2.8827620060541603, "grad_norm": 0.047049082815647125, "learning_rate": 2.194052447042582e-06, "loss": 0.0009, "step": 176180 }, { "epoch": 2.882925632005236, "grad_norm": 0.12347239255905151, "learning_rate": 2.193461545875657e-06, "loss": 0.0005, "step": 176190 }, { "epoch": 2.883089257956312, "grad_norm": 0.012806322425603867, "learning_rate": 2.192870701931665e-06, "loss": 0.0011, "step": 176200 }, { "epoch": 2.883252883907388, "grad_norm": 0.034594886004924774, "learning_rate": 2.1922799152226513e-06, "loss": 0.0003, "step": 176210 }, { "epoch": 2.8834165098584634, "grad_norm": 0.009104767814278603, "learning_rate": 2.191689185760665e-06, "loss": 0.0005, "step": 176220 }, { "epoch": 2.8835801358095394, "grad_norm": 0.043227531015872955, "learning_rate": 2.191098513557745e-06, "loss": 0.0009, "step": 176230 }, { "epoch": 2.8837437617606154, "grad_norm": 0.014541574753820896, "learning_rate": 2.190507898625938e-06, "loss": 0.0004, "step": 176240 }, { "epoch": 2.883907387711691, "grad_norm": 0.022814488038420677, "learning_rate": 2.1899173409772856e-06, "loss": 0.0009, "step": 176250 }, { "epoch": 2.884071013662767, "grad_norm": 0.15309235453605652, "learning_rate": 2.1893268406238303e-06, "loss": 0.0007, "step": 176260 }, { "epoch": 2.884234639613843, "grad_norm": 0.02884880267083645, "learning_rate": 2.188736397577609e-06, "loss": 0.0007, "step": 176270 }, { "epoch": 2.8843982655649185, "grad_norm": 0.05864391475915909, "learning_rate": 2.1881460118506625e-06, "loss": 0.0012, "step": 176280 }, { "epoch": 2.8845618915159945, "grad_norm": 0.058259516954422, "learning_rate": 2.1875556834550243e-06, "loss": 0.0007, "step": 176290 }, { "epoch": 2.8847255174670705, "grad_norm": 0.008398306556046009, "learning_rate": 2.1869654124027378e-06, "loss": 0.0008, "step": 176300 }, { "epoch": 2.884889143418146, "grad_norm": 0.06623674184083939, "learning_rate": 2.1863751987058325e-06, "loss": 0.0007, "step": 176310 }, { "epoch": 2.885052769369222, "grad_norm": 0.10285434871912003, "learning_rate": 2.185785042376345e-06, "loss": 0.0008, "step": 176320 }, { "epoch": 2.885216395320298, "grad_norm": 0.012089026160538197, "learning_rate": 2.185194943426304e-06, "loss": 0.0003, "step": 176330 }, { "epoch": 2.8853800212713736, "grad_norm": 0.010302970185875893, "learning_rate": 2.1846049018677472e-06, "loss": 0.0004, "step": 176340 }, { "epoch": 2.8855436472224496, "grad_norm": 0.0244772806763649, "learning_rate": 2.184014917712701e-06, "loss": 0.0005, "step": 176350 }, { "epoch": 2.885707273173525, "grad_norm": 0.03246033936738968, "learning_rate": 2.1834249909731976e-06, "loss": 0.0012, "step": 176360 }, { "epoch": 2.885870899124601, "grad_norm": 0.0058803861029446125, "learning_rate": 2.1828351216612586e-06, "loss": 0.0004, "step": 176370 }, { "epoch": 2.886034525075677, "grad_norm": 0.025216329842805862, "learning_rate": 2.18224530978892e-06, "loss": 0.0005, "step": 176380 }, { "epoch": 2.8861981510267527, "grad_norm": 0.04280976206064224, "learning_rate": 2.181655555368201e-06, "loss": 0.0006, "step": 176390 }, { "epoch": 2.8863617769778287, "grad_norm": 0.0032853696029633284, "learning_rate": 2.1810658584111305e-06, "loss": 0.0007, "step": 176400 }, { "epoch": 2.8865254029289042, "grad_norm": 0.054160092025995255, "learning_rate": 2.180476218929726e-06, "loss": 0.0004, "step": 176410 }, { "epoch": 2.8866890288799802, "grad_norm": 0.004589283373206854, "learning_rate": 2.1798866369360166e-06, "loss": 0.0014, "step": 176420 }, { "epoch": 2.8868526548310562, "grad_norm": 0.05429664999246597, "learning_rate": 2.179297112442019e-06, "loss": 0.0009, "step": 176430 }, { "epoch": 2.887016280782132, "grad_norm": 0.0223104078322649, "learning_rate": 2.1787076454597556e-06, "loss": 0.0004, "step": 176440 }, { "epoch": 2.887179906733208, "grad_norm": 0.008667773567140102, "learning_rate": 2.1781182360012426e-06, "loss": 0.0009, "step": 176450 }, { "epoch": 2.887343532684284, "grad_norm": 0.07585791498422623, "learning_rate": 2.1775288840784984e-06, "loss": 0.0007, "step": 176460 }, { "epoch": 2.8875071586353593, "grad_norm": 0.04945812374353409, "learning_rate": 2.17693958970354e-06, "loss": 0.0009, "step": 176470 }, { "epoch": 2.8876707845864353, "grad_norm": 0.0991426482796669, "learning_rate": 2.1763503528883846e-06, "loss": 0.0008, "step": 176480 }, { "epoch": 2.8878344105375113, "grad_norm": 0.007054779678583145, "learning_rate": 2.175761173645042e-06, "loss": 0.0008, "step": 176490 }, { "epoch": 2.887998036488587, "grad_norm": 0.06172759830951691, "learning_rate": 2.1751720519855274e-06, "loss": 0.0006, "step": 176500 }, { "epoch": 2.888161662439663, "grad_norm": 0.05063409358263016, "learning_rate": 2.1745829879218515e-06, "loss": 0.0006, "step": 176510 }, { "epoch": 2.888325288390739, "grad_norm": 0.02903308905661106, "learning_rate": 2.173993981466028e-06, "loss": 0.001, "step": 176520 }, { "epoch": 2.8884889143418144, "grad_norm": 0.043390411883592606, "learning_rate": 2.173405032630062e-06, "loss": 0.0007, "step": 176530 }, { "epoch": 2.8886525402928904, "grad_norm": 0.06773832440376282, "learning_rate": 2.172816141425963e-06, "loss": 0.0005, "step": 176540 }, { "epoch": 2.8888161662439664, "grad_norm": 0.03101511113345623, "learning_rate": 2.1722273078657386e-06, "loss": 0.0006, "step": 176550 }, { "epoch": 2.888979792195042, "grad_norm": 0.008495462127029896, "learning_rate": 2.1716385319613966e-06, "loss": 0.0005, "step": 176560 }, { "epoch": 2.889143418146118, "grad_norm": 0.010346234776079655, "learning_rate": 2.1710498137249376e-06, "loss": 0.0012, "step": 176570 }, { "epoch": 2.889307044097194, "grad_norm": 0.08409739285707474, "learning_rate": 2.1704611531683667e-06, "loss": 0.0008, "step": 176580 }, { "epoch": 2.8894706700482695, "grad_norm": 0.05359337478876114, "learning_rate": 2.1698725503036866e-06, "loss": 0.0012, "step": 176590 }, { "epoch": 2.8896342959993455, "grad_norm": 0.02883017621934414, "learning_rate": 2.1692840051429e-06, "loss": 0.0007, "step": 176600 }, { "epoch": 2.8897979219504215, "grad_norm": 0.08894481509923935, "learning_rate": 2.1686955176980037e-06, "loss": 0.0009, "step": 176610 }, { "epoch": 2.889961547901497, "grad_norm": 0.03534023091197014, "learning_rate": 2.1681070879809967e-06, "loss": 0.0005, "step": 176620 }, { "epoch": 2.890125173852573, "grad_norm": 0.020865241065621376, "learning_rate": 2.167518716003879e-06, "loss": 0.0006, "step": 176630 }, { "epoch": 2.890288799803649, "grad_norm": 0.019721122458577156, "learning_rate": 2.166930401778647e-06, "loss": 0.0008, "step": 176640 }, { "epoch": 2.8904524257547246, "grad_norm": 0.0051161423325538635, "learning_rate": 2.166342145317293e-06, "loss": 0.0005, "step": 176650 }, { "epoch": 2.8906160517058006, "grad_norm": 0.061121392995119095, "learning_rate": 2.165753946631812e-06, "loss": 0.0007, "step": 176660 }, { "epoch": 2.8907796776568766, "grad_norm": 0.02670552395284176, "learning_rate": 2.165165805734199e-06, "loss": 0.0005, "step": 176670 }, { "epoch": 2.890943303607952, "grad_norm": 0.06566611677408218, "learning_rate": 2.164577722636445e-06, "loss": 0.0006, "step": 176680 }, { "epoch": 2.891106929559028, "grad_norm": 0.10093427449464798, "learning_rate": 2.163989697350539e-06, "loss": 0.0008, "step": 176690 }, { "epoch": 2.891270555510104, "grad_norm": 0.008687400259077549, "learning_rate": 2.1634017298884695e-06, "loss": 0.0004, "step": 176700 }, { "epoch": 2.8914341814611797, "grad_norm": 0.0419042743742466, "learning_rate": 2.1628138202622274e-06, "loss": 0.0005, "step": 176710 }, { "epoch": 2.8915978074122557, "grad_norm": 0.04604877531528473, "learning_rate": 2.1622259684838002e-06, "loss": 0.0003, "step": 176720 }, { "epoch": 2.8917614333633317, "grad_norm": 0.09046250581741333, "learning_rate": 2.1616381745651704e-06, "loss": 0.0006, "step": 176730 }, { "epoch": 2.8919250593144072, "grad_norm": 0.03923799842596054, "learning_rate": 2.161050438518324e-06, "loss": 0.0007, "step": 176740 }, { "epoch": 2.8920886852654832, "grad_norm": 0.13225719332695007, "learning_rate": 2.160462760355245e-06, "loss": 0.0012, "step": 176750 }, { "epoch": 2.892252311216559, "grad_norm": 0.019222626462578773, "learning_rate": 2.159875140087917e-06, "loss": 0.0009, "step": 176760 }, { "epoch": 2.892415937167635, "grad_norm": 0.03554864227771759, "learning_rate": 2.159287577728318e-06, "loss": 0.0008, "step": 176770 }, { "epoch": 2.892579563118711, "grad_norm": 0.038094259798526764, "learning_rate": 2.158700073288429e-06, "loss": 0.0005, "step": 176780 }, { "epoch": 2.8927431890697863, "grad_norm": 0.029695618897676468, "learning_rate": 2.1581126267802287e-06, "loss": 0.0005, "step": 176790 }, { "epoch": 2.8929068150208623, "grad_norm": 0.05426757037639618, "learning_rate": 2.1575252382156973e-06, "loss": 0.0004, "step": 176800 }, { "epoch": 2.893070440971938, "grad_norm": 0.03658448904752731, "learning_rate": 2.156937907606807e-06, "loss": 0.0008, "step": 176810 }, { "epoch": 2.893234066923014, "grad_norm": 0.016176342964172363, "learning_rate": 2.156350634965535e-06, "loss": 0.0006, "step": 176820 }, { "epoch": 2.89339769287409, "grad_norm": 0.0021190063562244177, "learning_rate": 2.1557634203038545e-06, "loss": 0.0006, "step": 176830 }, { "epoch": 2.8935613188251654, "grad_norm": 0.08120778203010559, "learning_rate": 2.1551762636337413e-06, "loss": 0.0007, "step": 176840 }, { "epoch": 2.8937249447762414, "grad_norm": 0.026758559048175812, "learning_rate": 2.154589164967163e-06, "loss": 0.0004, "step": 176850 }, { "epoch": 2.8938885707273174, "grad_norm": 0.057105693966150284, "learning_rate": 2.1540021243160912e-06, "loss": 0.0009, "step": 176860 }, { "epoch": 2.894052196678393, "grad_norm": 0.020008254796266556, "learning_rate": 2.1534151416924958e-06, "loss": 0.0015, "step": 176870 }, { "epoch": 2.894215822629469, "grad_norm": 0.014276805333793163, "learning_rate": 2.1528282171083462e-06, "loss": 0.0007, "step": 176880 }, { "epoch": 2.894379448580545, "grad_norm": 0.03832237422466278, "learning_rate": 2.152241350575607e-06, "loss": 0.0021, "step": 176890 }, { "epoch": 2.8945430745316205, "grad_norm": 0.0006634188466705382, "learning_rate": 2.1516545421062435e-06, "loss": 0.0003, "step": 176900 }, { "epoch": 2.8947067004826965, "grad_norm": 0.025810377672314644, "learning_rate": 2.1510677917122223e-06, "loss": 0.0003, "step": 176910 }, { "epoch": 2.8948703264337725, "grad_norm": 0.03340725228190422, "learning_rate": 2.1504810994055074e-06, "loss": 0.0004, "step": 176920 }, { "epoch": 2.895033952384848, "grad_norm": 0.0024445417802780867, "learning_rate": 2.1498944651980576e-06, "loss": 0.0006, "step": 176930 }, { "epoch": 2.895197578335924, "grad_norm": 0.006152563728392124, "learning_rate": 2.1493078891018375e-06, "loss": 0.0023, "step": 176940 }, { "epoch": 2.895361204287, "grad_norm": 0.04333128780126572, "learning_rate": 2.148721371128802e-06, "loss": 0.0005, "step": 176950 }, { "epoch": 2.8955248302380756, "grad_norm": 0.024983402341604233, "learning_rate": 2.148134911290916e-06, "loss": 0.0004, "step": 176960 }, { "epoch": 2.8956884561891516, "grad_norm": 0.022774716839194298, "learning_rate": 2.147548509600132e-06, "loss": 0.0005, "step": 176970 }, { "epoch": 2.8958520821402276, "grad_norm": 0.004558480344712734, "learning_rate": 2.14696216606841e-06, "loss": 0.0007, "step": 176980 }, { "epoch": 2.896015708091303, "grad_norm": 0.05513720214366913, "learning_rate": 2.146375880707699e-06, "loss": 0.0007, "step": 176990 }, { "epoch": 2.896179334042379, "grad_norm": 0.0801529586315155, "learning_rate": 2.1457896535299605e-06, "loss": 0.0008, "step": 177000 }, { "epoch": 2.896342959993455, "grad_norm": 0.0713096559047699, "learning_rate": 2.145203484547142e-06, "loss": 0.0008, "step": 177010 }, { "epoch": 2.8965065859445307, "grad_norm": 0.02646057866513729, "learning_rate": 2.1446173737711985e-06, "loss": 0.0006, "step": 177020 }, { "epoch": 2.8966702118956067, "grad_norm": 0.014352780766785145, "learning_rate": 2.144031321214074e-06, "loss": 0.0004, "step": 177030 }, { "epoch": 2.8968338378466827, "grad_norm": 0.04271852225065231, "learning_rate": 2.143445326887726e-06, "loss": 0.0004, "step": 177040 }, { "epoch": 2.8969974637977582, "grad_norm": 0.06880933791399002, "learning_rate": 2.142859390804097e-06, "loss": 0.0006, "step": 177050 }, { "epoch": 2.8971610897488342, "grad_norm": 0.031149305403232574, "learning_rate": 2.142273512975136e-06, "loss": 0.0006, "step": 177060 }, { "epoch": 2.8973247156999102, "grad_norm": 0.05253808945417404, "learning_rate": 2.1416876934127845e-06, "loss": 0.0011, "step": 177070 }, { "epoch": 2.897488341650986, "grad_norm": 0.03394263610243797, "learning_rate": 2.1411019321289937e-06, "loss": 0.0003, "step": 177080 }, { "epoch": 2.897651967602062, "grad_norm": 0.04975878819823265, "learning_rate": 2.1405162291357014e-06, "loss": 0.001, "step": 177090 }, { "epoch": 2.897815593553138, "grad_norm": 0.0002517167595215142, "learning_rate": 2.1399305844448533e-06, "loss": 0.0006, "step": 177100 }, { "epoch": 2.8979792195042133, "grad_norm": 0.07889974117279053, "learning_rate": 2.1393449980683856e-06, "loss": 0.0005, "step": 177110 }, { "epoch": 2.8981428454552893, "grad_norm": 0.019734574481844902, "learning_rate": 2.138759470018241e-06, "loss": 0.0007, "step": 177120 }, { "epoch": 2.8983064714063653, "grad_norm": 0.017215756699442863, "learning_rate": 2.1381740003063573e-06, "loss": 0.001, "step": 177130 }, { "epoch": 2.898470097357441, "grad_norm": 0.07857027649879456, "learning_rate": 2.1375885889446736e-06, "loss": 0.0008, "step": 177140 }, { "epoch": 2.898633723308517, "grad_norm": 0.025760024785995483, "learning_rate": 2.1370032359451228e-06, "loss": 0.0007, "step": 177150 }, { "epoch": 2.8987973492595924, "grad_norm": 0.00060412532184273, "learning_rate": 2.1364179413196408e-06, "loss": 0.0005, "step": 177160 }, { "epoch": 2.8989609752106684, "grad_norm": 0.025936255231499672, "learning_rate": 2.1358327050801614e-06, "loss": 0.0005, "step": 177170 }, { "epoch": 2.8991246011617444, "grad_norm": 0.0580892413854599, "learning_rate": 2.135247527238619e-06, "loss": 0.0003, "step": 177180 }, { "epoch": 2.89928822711282, "grad_norm": 0.040999025106430054, "learning_rate": 2.134662407806942e-06, "loss": 0.0007, "step": 177190 }, { "epoch": 2.899451853063896, "grad_norm": 0.008688162080943584, "learning_rate": 2.134077346797061e-06, "loss": 0.0009, "step": 177200 }, { "epoch": 2.8996154790149715, "grad_norm": 0.04018067941069603, "learning_rate": 2.1334923442209055e-06, "loss": 0.001, "step": 177210 }, { "epoch": 2.8997791049660475, "grad_norm": 0.23400314152240753, "learning_rate": 2.1329074000904056e-06, "loss": 0.0006, "step": 177220 }, { "epoch": 2.8999427309171235, "grad_norm": 0.0640736073255539, "learning_rate": 2.132322514417483e-06, "loss": 0.0009, "step": 177230 }, { "epoch": 2.900106356868199, "grad_norm": 0.009185069240629673, "learning_rate": 2.1317376872140654e-06, "loss": 0.0004, "step": 177240 }, { "epoch": 2.900269982819275, "grad_norm": 0.047750212252140045, "learning_rate": 2.1311529184920776e-06, "loss": 0.0005, "step": 177250 }, { "epoch": 2.900433608770351, "grad_norm": 0.023673921823501587, "learning_rate": 2.130568208263443e-06, "loss": 0.0005, "step": 177260 }, { "epoch": 2.9005972347214266, "grad_norm": 0.020437490195035934, "learning_rate": 2.1299835565400813e-06, "loss": 0.0005, "step": 177270 }, { "epoch": 2.9007608606725026, "grad_norm": 0.009430529549717903, "learning_rate": 2.1293989633339136e-06, "loss": 0.0003, "step": 177280 }, { "epoch": 2.9009244866235786, "grad_norm": 0.06898472458124161, "learning_rate": 2.1288144286568597e-06, "loss": 0.0007, "step": 177290 }, { "epoch": 2.901088112574654, "grad_norm": 0.044406477361917496, "learning_rate": 2.1282299525208393e-06, "loss": 0.0009, "step": 177300 }, { "epoch": 2.90125173852573, "grad_norm": 0.01224274467676878, "learning_rate": 2.1276455349377662e-06, "loss": 0.0006, "step": 177310 }, { "epoch": 2.901415364476806, "grad_norm": 0.004196258261799812, "learning_rate": 2.1270611759195583e-06, "loss": 0.0006, "step": 177320 }, { "epoch": 2.9015789904278817, "grad_norm": 0.059121984988451004, "learning_rate": 2.126476875478129e-06, "loss": 0.0005, "step": 177330 }, { "epoch": 2.9017426163789577, "grad_norm": 0.016118288040161133, "learning_rate": 2.125892633625394e-06, "loss": 0.0036, "step": 177340 }, { "epoch": 2.9019062423300337, "grad_norm": 0.03209742531180382, "learning_rate": 2.125308450373263e-06, "loss": 0.0007, "step": 177350 }, { "epoch": 2.9020698682811092, "grad_norm": 0.03722938522696495, "learning_rate": 2.124724325733647e-06, "loss": 0.0007, "step": 177360 }, { "epoch": 2.9022334942321852, "grad_norm": 0.058879923075437546, "learning_rate": 2.124140259718457e-06, "loss": 0.0009, "step": 177370 }, { "epoch": 2.9023971201832612, "grad_norm": 0.02010641060769558, "learning_rate": 2.123556252339603e-06, "loss": 0.0007, "step": 177380 }, { "epoch": 2.902560746134337, "grad_norm": 0.005260211415588856, "learning_rate": 2.122972303608989e-06, "loss": 0.0007, "step": 177390 }, { "epoch": 2.902724372085413, "grad_norm": 0.027751464396715164, "learning_rate": 2.1223884135385232e-06, "loss": 0.0008, "step": 177400 }, { "epoch": 2.902887998036489, "grad_norm": 0.006878254935145378, "learning_rate": 2.1218045821401095e-06, "loss": 0.0009, "step": 177410 }, { "epoch": 2.9030516239875643, "grad_norm": 0.022355912253260612, "learning_rate": 2.121220809425655e-06, "loss": 0.0007, "step": 177420 }, { "epoch": 2.9032152499386403, "grad_norm": 0.012008478865027428, "learning_rate": 2.120637095407058e-06, "loss": 0.0007, "step": 177430 }, { "epoch": 2.9033788758897163, "grad_norm": 0.0047128149308264256, "learning_rate": 2.1200534400962216e-06, "loss": 0.0004, "step": 177440 }, { "epoch": 2.903542501840792, "grad_norm": 0.040680479258298874, "learning_rate": 2.1194698435050465e-06, "loss": 0.0009, "step": 177450 }, { "epoch": 2.903706127791868, "grad_norm": 0.008104072883725166, "learning_rate": 2.1188863056454327e-06, "loss": 0.0005, "step": 177460 }, { "epoch": 2.903869753742944, "grad_norm": 0.028712989762425423, "learning_rate": 2.118302826529276e-06, "loss": 0.0006, "step": 177470 }, { "epoch": 2.9040333796940194, "grad_norm": 0.03017980232834816, "learning_rate": 2.117719406168473e-06, "loss": 0.0007, "step": 177480 }, { "epoch": 2.9041970056450954, "grad_norm": 0.10520725697278976, "learning_rate": 2.1171360445749206e-06, "loss": 0.0007, "step": 177490 }, { "epoch": 2.9043606315961714, "grad_norm": 0.008986033499240875, "learning_rate": 2.116552741760514e-06, "loss": 0.0005, "step": 177500 }, { "epoch": 2.904524257547247, "grad_norm": 0.014616502448916435, "learning_rate": 2.1159694977371426e-06, "loss": 0.0007, "step": 177510 }, { "epoch": 2.904687883498323, "grad_norm": 0.009717185981571674, "learning_rate": 2.1153863125167007e-06, "loss": 0.0003, "step": 177520 }, { "epoch": 2.9048515094493985, "grad_norm": 0.060335706919431686, "learning_rate": 2.114803186111079e-06, "loss": 0.0009, "step": 177530 }, { "epoch": 2.9050151354004745, "grad_norm": 0.03839008882641792, "learning_rate": 2.1142201185321674e-06, "loss": 0.0005, "step": 177540 }, { "epoch": 2.9051787613515505, "grad_norm": 0.028804121538996696, "learning_rate": 2.1136371097918524e-06, "loss": 0.0004, "step": 177550 }, { "epoch": 2.905342387302626, "grad_norm": 0.04255538433790207, "learning_rate": 2.1130541599020236e-06, "loss": 0.0006, "step": 177560 }, { "epoch": 2.905506013253702, "grad_norm": 0.048011861741542816, "learning_rate": 2.112471268874561e-06, "loss": 0.0005, "step": 177570 }, { "epoch": 2.9056696392047776, "grad_norm": 0.004070754628628492, "learning_rate": 2.1118884367213583e-06, "loss": 0.0004, "step": 177580 }, { "epoch": 2.9058332651558536, "grad_norm": 0.011813007295131683, "learning_rate": 2.111305663454291e-06, "loss": 0.0004, "step": 177590 }, { "epoch": 2.9059968911069296, "grad_norm": 0.024593552574515343, "learning_rate": 2.110722949085247e-06, "loss": 0.0034, "step": 177600 }, { "epoch": 2.906160517058005, "grad_norm": 0.024365978315472603, "learning_rate": 2.110140293626101e-06, "loss": 0.0005, "step": 177610 }, { "epoch": 2.906324143009081, "grad_norm": 0.07025423645973206, "learning_rate": 2.1095576970887405e-06, "loss": 0.0008, "step": 177620 }, { "epoch": 2.906487768960157, "grad_norm": 0.061987537890672684, "learning_rate": 2.1089751594850384e-06, "loss": 0.0004, "step": 177630 }, { "epoch": 2.9066513949112327, "grad_norm": 0.04424640163779259, "learning_rate": 2.108392680826876e-06, "loss": 0.0011, "step": 177640 }, { "epoch": 2.9068150208623087, "grad_norm": 0.053725890815258026, "learning_rate": 2.1078102611261234e-06, "loss": 0.0006, "step": 177650 }, { "epoch": 2.9069786468133847, "grad_norm": 0.09627298265695572, "learning_rate": 2.1072279003946643e-06, "loss": 0.0013, "step": 177660 }, { "epoch": 2.9071422727644602, "grad_norm": 0.07825011759996414, "learning_rate": 2.106645598644366e-06, "loss": 0.0008, "step": 177670 }, { "epoch": 2.9073058987155362, "grad_norm": 0.026661943644285202, "learning_rate": 2.1060633558871046e-06, "loss": 0.0006, "step": 177680 }, { "epoch": 2.9074695246666122, "grad_norm": 0.0016177112702280283, "learning_rate": 2.105481172134747e-06, "loss": 0.0006, "step": 177690 }, { "epoch": 2.907633150617688, "grad_norm": 0.03407486900687218, "learning_rate": 2.10489904739917e-06, "loss": 0.0007, "step": 177700 }, { "epoch": 2.907796776568764, "grad_norm": 0.010538170114159584, "learning_rate": 2.1043169816922377e-06, "loss": 0.0008, "step": 177710 }, { "epoch": 2.90796040251984, "grad_norm": 0.008609225042164326, "learning_rate": 2.103734975025821e-06, "loss": 0.0006, "step": 177720 }, { "epoch": 2.9081240284709153, "grad_norm": 0.004194254521280527, "learning_rate": 2.1031530274117817e-06, "loss": 0.002, "step": 177730 }, { "epoch": 2.9082876544219913, "grad_norm": 0.0019117441261187196, "learning_rate": 2.1025711388619917e-06, "loss": 0.0005, "step": 177740 }, { "epoch": 2.9084512803730673, "grad_norm": 0.021378476172685623, "learning_rate": 2.1019893093883104e-06, "loss": 0.0012, "step": 177750 }, { "epoch": 2.908614906324143, "grad_norm": 0.04305338114500046, "learning_rate": 2.101407539002604e-06, "loss": 0.0005, "step": 177760 }, { "epoch": 2.908778532275219, "grad_norm": 0.028381533920764923, "learning_rate": 2.1008258277167314e-06, "loss": 0.0003, "step": 177770 }, { "epoch": 2.908942158226295, "grad_norm": 0.07830385118722916, "learning_rate": 2.100244175542554e-06, "loss": 0.001, "step": 177780 }, { "epoch": 2.9091057841773704, "grad_norm": 0.08624427020549774, "learning_rate": 2.099662582491932e-06, "loss": 0.0005, "step": 177790 }, { "epoch": 2.9092694101284464, "grad_norm": 0.0005485974834300578, "learning_rate": 2.099081048576725e-06, "loss": 0.0007, "step": 177800 }, { "epoch": 2.9094330360795224, "grad_norm": 0.02550804242491722, "learning_rate": 2.0984995738087864e-06, "loss": 0.0006, "step": 177810 }, { "epoch": 2.909596662030598, "grad_norm": 0.0661039873957634, "learning_rate": 2.097918158199974e-06, "loss": 0.001, "step": 177820 }, { "epoch": 2.909760287981674, "grad_norm": 0.015012141317129135, "learning_rate": 2.0973368017621422e-06, "loss": 0.0027, "step": 177830 }, { "epoch": 2.90992391393275, "grad_norm": 0.043805163353681564, "learning_rate": 2.0967555045071465e-06, "loss": 0.0011, "step": 177840 }, { "epoch": 2.9100875398838255, "grad_norm": 0.009923041798174381, "learning_rate": 2.0961742664468344e-06, "loss": 0.0014, "step": 177850 }, { "epoch": 2.9102511658349015, "grad_norm": 0.0004909878480248153, "learning_rate": 2.0955930875930598e-06, "loss": 0.0006, "step": 177860 }, { "epoch": 2.9104147917859775, "grad_norm": 0.02479095570743084, "learning_rate": 2.095011967957672e-06, "loss": 0.001, "step": 177870 }, { "epoch": 2.910578417737053, "grad_norm": 0.0012234908062964678, "learning_rate": 2.0944309075525215e-06, "loss": 0.0006, "step": 177880 }, { "epoch": 2.910742043688129, "grad_norm": 0.06504473090171814, "learning_rate": 2.093849906389452e-06, "loss": 0.0004, "step": 177890 }, { "epoch": 2.910905669639205, "grad_norm": 0.008254610933363438, "learning_rate": 2.0932689644803105e-06, "loss": 0.001, "step": 177900 }, { "epoch": 2.9110692955902806, "grad_norm": 0.05375050753355026, "learning_rate": 2.0926880818369437e-06, "loss": 0.0006, "step": 177910 }, { "epoch": 2.9112329215413566, "grad_norm": 0.05751027911901474, "learning_rate": 2.0921072584711952e-06, "loss": 0.0015, "step": 177920 }, { "epoch": 2.911396547492432, "grad_norm": 0.012707212939858437, "learning_rate": 2.091526494394906e-06, "loss": 0.0005, "step": 177930 }, { "epoch": 2.911560173443508, "grad_norm": 0.041068632155656815, "learning_rate": 2.090945789619917e-06, "loss": 0.0004, "step": 177940 }, { "epoch": 2.911723799394584, "grad_norm": 0.019001565873622894, "learning_rate": 2.0903651441580696e-06, "loss": 0.0011, "step": 177950 }, { "epoch": 2.9118874253456597, "grad_norm": 0.015936588868498802, "learning_rate": 2.0897845580212046e-06, "loss": 0.0004, "step": 177960 }, { "epoch": 2.9120510512967357, "grad_norm": 0.011714170686900616, "learning_rate": 2.0892040312211554e-06, "loss": 0.0006, "step": 177970 }, { "epoch": 2.9122146772478112, "grad_norm": 0.02221500128507614, "learning_rate": 2.08862356376976e-06, "loss": 0.0022, "step": 177980 }, { "epoch": 2.9123783031988872, "grad_norm": 0.03171189874410629, "learning_rate": 2.0880431556788542e-06, "loss": 0.001, "step": 177990 }, { "epoch": 2.9125419291499632, "grad_norm": 0.04434116557240486, "learning_rate": 2.0874628069602744e-06, "loss": 0.001, "step": 178000 }, { "epoch": 2.912705555101039, "grad_norm": 0.07468567043542862, "learning_rate": 2.0868825176258484e-06, "loss": 0.0011, "step": 178010 }, { "epoch": 2.912869181052115, "grad_norm": 0.0009253888856619596, "learning_rate": 2.0863022876874106e-06, "loss": 0.0006, "step": 178020 }, { "epoch": 2.913032807003191, "grad_norm": 0.002483588410541415, "learning_rate": 2.085722117156791e-06, "loss": 0.0005, "step": 178030 }, { "epoch": 2.9131964329542663, "grad_norm": 0.0576290488243103, "learning_rate": 2.085142006045821e-06, "loss": 0.0005, "step": 178040 }, { "epoch": 2.9133600589053423, "grad_norm": 0.06163637340068817, "learning_rate": 2.084561954366325e-06, "loss": 0.0009, "step": 178050 }, { "epoch": 2.9135236848564183, "grad_norm": 0.024707332253456116, "learning_rate": 2.0839819621301304e-06, "loss": 0.0008, "step": 178060 }, { "epoch": 2.913687310807494, "grad_norm": 0.014098848216235638, "learning_rate": 2.0834020293490643e-06, "loss": 0.0004, "step": 178070 }, { "epoch": 2.91385093675857, "grad_norm": 0.018590368330478668, "learning_rate": 2.082822156034952e-06, "loss": 0.0006, "step": 178080 }, { "epoch": 2.914014562709646, "grad_norm": 0.048041313886642456, "learning_rate": 2.082242342199613e-06, "loss": 0.0005, "step": 178090 }, { "epoch": 2.9141781886607214, "grad_norm": 0.07023672759532928, "learning_rate": 2.0816625878548712e-06, "loss": 0.0003, "step": 178100 }, { "epoch": 2.9143418146117974, "grad_norm": 0.02161402627825737, "learning_rate": 2.081082893012547e-06, "loss": 0.0002, "step": 178110 }, { "epoch": 2.9145054405628734, "grad_norm": 0.05916627123951912, "learning_rate": 2.0805032576844624e-06, "loss": 0.0009, "step": 178120 }, { "epoch": 2.914669066513949, "grad_norm": 0.015917593613266945, "learning_rate": 2.0799236818824313e-06, "loss": 0.0006, "step": 178130 }, { "epoch": 2.914832692465025, "grad_norm": 0.031330302357673645, "learning_rate": 2.0793441656182732e-06, "loss": 0.0005, "step": 178140 }, { "epoch": 2.914996318416101, "grad_norm": 0.016229528933763504, "learning_rate": 2.078764708903804e-06, "loss": 0.0009, "step": 178150 }, { "epoch": 2.9151599443671765, "grad_norm": 0.057934392243623734, "learning_rate": 2.078185311750839e-06, "loss": 0.0011, "step": 178160 }, { "epoch": 2.9153235703182525, "grad_norm": 0.008562915027141571, "learning_rate": 2.077605974171189e-06, "loss": 0.0004, "step": 178170 }, { "epoch": 2.9154871962693285, "grad_norm": 0.04125946760177612, "learning_rate": 2.077026696176668e-06, "loss": 0.0009, "step": 178180 }, { "epoch": 2.915650822220404, "grad_norm": 0.021269772201776505, "learning_rate": 2.076447477779087e-06, "loss": 0.0006, "step": 178190 }, { "epoch": 2.91581444817148, "grad_norm": 0.05143075808882713, "learning_rate": 2.075868318990257e-06, "loss": 0.0006, "step": 178200 }, { "epoch": 2.915978074122556, "grad_norm": 0.05764470249414444, "learning_rate": 2.0752892198219826e-06, "loss": 0.0005, "step": 178210 }, { "epoch": 2.9161417000736316, "grad_norm": 0.01277630589902401, "learning_rate": 2.0747101802860764e-06, "loss": 0.0005, "step": 178220 }, { "epoch": 2.9163053260247076, "grad_norm": 0.05284089967608452, "learning_rate": 2.0741312003943377e-06, "loss": 0.0014, "step": 178230 }, { "epoch": 2.9164689519757836, "grad_norm": 0.011891908012330532, "learning_rate": 2.0735522801585794e-06, "loss": 0.0009, "step": 178240 }, { "epoch": 2.916632577926859, "grad_norm": 0.04407447203993797, "learning_rate": 2.0729734195905994e-06, "loss": 0.0006, "step": 178250 }, { "epoch": 2.916796203877935, "grad_norm": 0.027789276093244553, "learning_rate": 2.072394618702203e-06, "loss": 0.0008, "step": 178260 }, { "epoch": 2.916959829829011, "grad_norm": 0.05814977362751961, "learning_rate": 2.071815877505188e-06, "loss": 0.0006, "step": 178270 }, { "epoch": 2.9171234557800867, "grad_norm": 0.019177816808223724, "learning_rate": 2.0712371960113604e-06, "loss": 0.0005, "step": 178280 }, { "epoch": 2.9172870817311627, "grad_norm": 0.06326300650835037, "learning_rate": 2.070658574232513e-06, "loss": 0.0006, "step": 178290 }, { "epoch": 2.9174507076822382, "grad_norm": 0.05072472617030144, "learning_rate": 2.0700800121804483e-06, "loss": 0.0006, "step": 178300 }, { "epoch": 2.9176143336333142, "grad_norm": 0.009279899299144745, "learning_rate": 2.069501509866957e-06, "loss": 0.005, "step": 178310 }, { "epoch": 2.9177779595843902, "grad_norm": 0.02661130204796791, "learning_rate": 2.0689230673038406e-06, "loss": 0.0005, "step": 178320 }, { "epoch": 2.917941585535466, "grad_norm": 0.024994023144245148, "learning_rate": 2.068344684502888e-06, "loss": 0.001, "step": 178330 }, { "epoch": 2.918105211486542, "grad_norm": 0.012455440126359463, "learning_rate": 2.0677663614758968e-06, "loss": 0.0008, "step": 178340 }, { "epoch": 2.9182688374376173, "grad_norm": 0.011648819781839848, "learning_rate": 2.0671880982346515e-06, "loss": 0.0003, "step": 178350 }, { "epoch": 2.9184324633886933, "grad_norm": 0.01086385641247034, "learning_rate": 2.0666098947909504e-06, "loss": 0.0006, "step": 178360 }, { "epoch": 2.9185960893397693, "grad_norm": 0.05204188823699951, "learning_rate": 2.066031751156577e-06, "loss": 0.0004, "step": 178370 }, { "epoch": 2.918759715290845, "grad_norm": 0.019058680161833763, "learning_rate": 2.065453667343323e-06, "loss": 0.0004, "step": 178380 }, { "epoch": 2.918923341241921, "grad_norm": 0.0603271909058094, "learning_rate": 2.0648756433629703e-06, "loss": 0.0008, "step": 178390 }, { "epoch": 2.919086967192997, "grad_norm": 0.017851632088422775, "learning_rate": 2.0642976792273077e-06, "loss": 0.0006, "step": 178400 }, { "epoch": 2.9192505931440724, "grad_norm": 0.0718260258436203, "learning_rate": 2.0637197749481185e-06, "loss": 0.0011, "step": 178410 }, { "epoch": 2.9194142190951484, "grad_norm": 0.027267320081591606, "learning_rate": 2.063141930537188e-06, "loss": 0.0005, "step": 178420 }, { "epoch": 2.9195778450462244, "grad_norm": 0.08743027597665787, "learning_rate": 2.0625641460062936e-06, "loss": 0.0006, "step": 178430 }, { "epoch": 2.9197414709973, "grad_norm": 0.013292742893099785, "learning_rate": 2.061986421367218e-06, "loss": 0.0006, "step": 178440 }, { "epoch": 2.919905096948376, "grad_norm": 0.0168993528932333, "learning_rate": 2.0614087566317408e-06, "loss": 0.0009, "step": 178450 }, { "epoch": 2.920068722899452, "grad_norm": 0.023303259164094925, "learning_rate": 2.060831151811642e-06, "loss": 0.0004, "step": 178460 }, { "epoch": 2.9202323488505275, "grad_norm": 0.023398468270897865, "learning_rate": 2.060253606918694e-06, "loss": 0.0005, "step": 178470 }, { "epoch": 2.9203959748016035, "grad_norm": 0.006680995691567659, "learning_rate": 2.059676121964675e-06, "loss": 0.0005, "step": 178480 }, { "epoch": 2.9205596007526795, "grad_norm": 0.01594560220837593, "learning_rate": 2.059098696961359e-06, "loss": 0.0004, "step": 178490 }, { "epoch": 2.920723226703755, "grad_norm": 0.007730069104582071, "learning_rate": 2.0585213319205218e-06, "loss": 0.0007, "step": 178500 }, { "epoch": 2.920886852654831, "grad_norm": 0.06836860626935959, "learning_rate": 2.057944026853931e-06, "loss": 0.0006, "step": 178510 }, { "epoch": 2.921050478605907, "grad_norm": 0.009606817737221718, "learning_rate": 2.057366781773359e-06, "loss": 0.0008, "step": 178520 }, { "epoch": 2.9212141045569826, "grad_norm": 0.03760126605629921, "learning_rate": 2.056789596690576e-06, "loss": 0.0005, "step": 178530 }, { "epoch": 2.9213777305080586, "grad_norm": 0.021640025079250336, "learning_rate": 2.0562124716173514e-06, "loss": 0.0006, "step": 178540 }, { "epoch": 2.9215413564591346, "grad_norm": 0.03365211561322212, "learning_rate": 2.0556354065654493e-06, "loss": 0.0005, "step": 178550 }, { "epoch": 2.92170498241021, "grad_norm": 0.008294424042105675, "learning_rate": 2.0550584015466376e-06, "loss": 0.0015, "step": 178560 }, { "epoch": 2.921868608361286, "grad_norm": 0.03247595205903053, "learning_rate": 2.05448145657268e-06, "loss": 0.001, "step": 178570 }, { "epoch": 2.922032234312362, "grad_norm": 0.010036864317953587, "learning_rate": 2.0539045716553423e-06, "loss": 0.0005, "step": 178580 }, { "epoch": 2.9221958602634377, "grad_norm": 0.030042188242077827, "learning_rate": 2.0533277468063834e-06, "loss": 0.0011, "step": 178590 }, { "epoch": 2.9223594862145137, "grad_norm": 0.0548376627266407, "learning_rate": 2.052750982037565e-06, "loss": 0.0007, "step": 178600 }, { "epoch": 2.9225231121655897, "grad_norm": 0.08760268986225128, "learning_rate": 2.0521742773606486e-06, "loss": 0.0005, "step": 178610 }, { "epoch": 2.9226867381166652, "grad_norm": 0.00724099762737751, "learning_rate": 2.051597632787393e-06, "loss": 0.0007, "step": 178620 }, { "epoch": 2.9228503640677412, "grad_norm": 0.08327998965978622, "learning_rate": 2.0510210483295518e-06, "loss": 0.0008, "step": 178630 }, { "epoch": 2.9230139900188172, "grad_norm": 0.0891764760017395, "learning_rate": 2.0504445239988836e-06, "loss": 0.001, "step": 178640 }, { "epoch": 2.923177615969893, "grad_norm": 0.020297516137361526, "learning_rate": 2.049868059807143e-06, "loss": 0.0005, "step": 178650 }, { "epoch": 2.923341241920969, "grad_norm": 0.033513784408569336, "learning_rate": 2.0492916557660857e-06, "loss": 0.0005, "step": 178660 }, { "epoch": 2.9235048678720448, "grad_norm": 0.013414403423666954, "learning_rate": 2.0487153118874603e-06, "loss": 0.0009, "step": 178670 }, { "epoch": 2.9236684938231203, "grad_norm": 0.0620177686214447, "learning_rate": 2.0481390281830193e-06, "loss": 0.0008, "step": 178680 }, { "epoch": 2.9238321197741963, "grad_norm": 0.007701562251895666, "learning_rate": 2.0475628046645137e-06, "loss": 0.0006, "step": 178690 }, { "epoch": 2.923995745725272, "grad_norm": 0.029423782601952553, "learning_rate": 2.0469866413436927e-06, "loss": 0.0004, "step": 178700 }, { "epoch": 2.924159371676348, "grad_norm": 0.0055070193484425545, "learning_rate": 2.0464105382323012e-06, "loss": 0.0008, "step": 178710 }, { "epoch": 2.924322997627424, "grad_norm": 0.020379172638058662, "learning_rate": 2.0458344953420872e-06, "loss": 0.0004, "step": 178720 }, { "epoch": 2.9244866235784994, "grad_norm": 0.01911788433790207, "learning_rate": 2.045258512684795e-06, "loss": 0.0004, "step": 178730 }, { "epoch": 2.9246502495295754, "grad_norm": 0.01404265221208334, "learning_rate": 2.0446825902721706e-06, "loss": 0.0006, "step": 178740 }, { "epoch": 2.924813875480651, "grad_norm": 0.03571268171072006, "learning_rate": 2.044106728115953e-06, "loss": 0.0006, "step": 178750 }, { "epoch": 2.924977501431727, "grad_norm": 0.044639501720666885, "learning_rate": 2.043530926227886e-06, "loss": 0.0005, "step": 178760 }, { "epoch": 2.925141127382803, "grad_norm": 0.007736612111330032, "learning_rate": 2.0429551846197078e-06, "loss": 0.0005, "step": 178770 }, { "epoch": 2.9253047533338785, "grad_norm": 0.05654943361878395, "learning_rate": 2.0423795033031607e-06, "loss": 0.0004, "step": 178780 }, { "epoch": 2.9254683792849545, "grad_norm": 0.01725638657808304, "learning_rate": 2.0418038822899787e-06, "loss": 0.0007, "step": 178790 }, { "epoch": 2.9256320052360305, "grad_norm": 0.009367917664349079, "learning_rate": 2.0412283215919e-06, "loss": 0.0013, "step": 178800 }, { "epoch": 2.925795631187106, "grad_norm": 0.02921956777572632, "learning_rate": 2.040652821220659e-06, "loss": 0.0013, "step": 178810 }, { "epoch": 2.925959257138182, "grad_norm": 0.010343953967094421, "learning_rate": 2.0400773811879915e-06, "loss": 0.0005, "step": 178820 }, { "epoch": 2.926122883089258, "grad_norm": 0.032180577516555786, "learning_rate": 2.039502001505628e-06, "loss": 0.0008, "step": 178830 }, { "epoch": 2.9262865090403336, "grad_norm": 0.021323632448911667, "learning_rate": 2.0389266821853e-06, "loss": 0.0002, "step": 178840 }, { "epoch": 2.9264501349914096, "grad_norm": 0.04753128066658974, "learning_rate": 2.038351423238739e-06, "loss": 0.0008, "step": 178850 }, { "epoch": 2.9266137609424856, "grad_norm": 0.04616774618625641, "learning_rate": 2.0377762246776754e-06, "loss": 0.0005, "step": 178860 }, { "epoch": 2.926777386893561, "grad_norm": 0.06374722719192505, "learning_rate": 2.037201086513834e-06, "loss": 0.0006, "step": 178870 }, { "epoch": 2.926941012844637, "grad_norm": 0.04618437588214874, "learning_rate": 2.036626008758944e-06, "loss": 0.0011, "step": 178880 }, { "epoch": 2.927104638795713, "grad_norm": 0.020513813942670822, "learning_rate": 2.036050991424726e-06, "loss": 0.0006, "step": 178890 }, { "epoch": 2.9272682647467887, "grad_norm": 0.07128001749515533, "learning_rate": 2.0354760345229115e-06, "loss": 0.0006, "step": 178900 }, { "epoch": 2.9274318906978647, "grad_norm": 0.0070556700229644775, "learning_rate": 2.0349011380652168e-06, "loss": 0.0005, "step": 178910 }, { "epoch": 2.9275955166489407, "grad_norm": 0.02371249720454216, "learning_rate": 2.0343263020633687e-06, "loss": 0.001, "step": 178920 }, { "epoch": 2.9277591426000162, "grad_norm": 0.00426480220630765, "learning_rate": 2.033751526529081e-06, "loss": 0.0039, "step": 178930 }, { "epoch": 2.9279227685510922, "grad_norm": 0.02361888252198696, "learning_rate": 2.0331768114740807e-06, "loss": 0.0015, "step": 178940 }, { "epoch": 2.9280863945021682, "grad_norm": 0.052659668028354645, "learning_rate": 2.03260215691008e-06, "loss": 0.0009, "step": 178950 }, { "epoch": 2.928250020453244, "grad_norm": 0.05401821807026863, "learning_rate": 2.0320275628487994e-06, "loss": 0.0007, "step": 178960 }, { "epoch": 2.92841364640432, "grad_norm": 0.013549497351050377, "learning_rate": 2.0314530293019496e-06, "loss": 0.0006, "step": 178970 }, { "epoch": 2.9285772723553958, "grad_norm": 0.012768980115652084, "learning_rate": 2.030878556281251e-06, "loss": 0.0004, "step": 178980 }, { "epoch": 2.9287408983064713, "grad_norm": 0.010439430363476276, "learning_rate": 2.030304143798412e-06, "loss": 0.0003, "step": 178990 }, { "epoch": 2.9289045242575473, "grad_norm": 0.02934621460735798, "learning_rate": 2.0297297918651476e-06, "loss": 0.0004, "step": 179000 }, { "epoch": 2.9290681502086233, "grad_norm": 0.026047803461551666, "learning_rate": 2.029155500493163e-06, "loss": 0.0006, "step": 179010 }, { "epoch": 2.929231776159699, "grad_norm": 0.010873946361243725, "learning_rate": 2.028581269694175e-06, "loss": 0.0012, "step": 179020 }, { "epoch": 2.929395402110775, "grad_norm": 0.0019215800566598773, "learning_rate": 2.0280070994798856e-06, "loss": 0.0012, "step": 179030 }, { "epoch": 2.929559028061851, "grad_norm": 0.05048966035246849, "learning_rate": 2.0274329898620055e-06, "loss": 0.0007, "step": 179040 }, { "epoch": 2.9297226540129264, "grad_norm": 0.024648820981383324, "learning_rate": 2.0268589408522373e-06, "loss": 0.0003, "step": 179050 }, { "epoch": 2.9298862799640024, "grad_norm": 0.02726985700428486, "learning_rate": 2.0262849524622865e-06, "loss": 0.0006, "step": 179060 }, { "epoch": 2.930049905915078, "grad_norm": 0.044336654245853424, "learning_rate": 2.0257110247038565e-06, "loss": 0.0006, "step": 179070 }, { "epoch": 2.930213531866154, "grad_norm": 0.0367763489484787, "learning_rate": 2.025137157588651e-06, "loss": 0.0006, "step": 179080 }, { "epoch": 2.93037715781723, "grad_norm": 0.1305222362279892, "learning_rate": 2.0245633511283668e-06, "loss": 0.0013, "step": 179090 }, { "epoch": 2.9305407837683055, "grad_norm": 0.04487916827201843, "learning_rate": 2.023989605334706e-06, "loss": 0.0014, "step": 179100 }, { "epoch": 2.9307044097193815, "grad_norm": 0.04058770090341568, "learning_rate": 2.0234159202193656e-06, "loss": 0.0005, "step": 179110 }, { "epoch": 2.930868035670457, "grad_norm": 0.05105336755514145, "learning_rate": 2.022842295794045e-06, "loss": 0.0005, "step": 179120 }, { "epoch": 2.931031661621533, "grad_norm": 0.007476025726646185, "learning_rate": 2.022268732070436e-06, "loss": 0.0009, "step": 179130 }, { "epoch": 2.931195287572609, "grad_norm": 0.066892609000206, "learning_rate": 2.0216952290602355e-06, "loss": 0.0007, "step": 179140 }, { "epoch": 2.9313589135236846, "grad_norm": 0.020953554660081863, "learning_rate": 2.021121786775136e-06, "loss": 0.0008, "step": 179150 }, { "epoch": 2.9315225394747606, "grad_norm": 0.04402484372258186, "learning_rate": 2.0205484052268313e-06, "loss": 0.0006, "step": 179160 }, { "epoch": 2.9316861654258366, "grad_norm": 0.026890384033322334, "learning_rate": 2.01997508442701e-06, "loss": 0.0006, "step": 179170 }, { "epoch": 2.931849791376912, "grad_norm": 0.020921960473060608, "learning_rate": 2.0194018243873612e-06, "loss": 0.0016, "step": 179180 }, { "epoch": 2.932013417327988, "grad_norm": 0.016873599961400032, "learning_rate": 2.018828625119575e-06, "loss": 0.0008, "step": 179190 }, { "epoch": 2.932177043279064, "grad_norm": 0.015538069419562817, "learning_rate": 2.0182554866353394e-06, "loss": 0.0005, "step": 179200 }, { "epoch": 2.9323406692301397, "grad_norm": 0.012375953607261181, "learning_rate": 2.0176824089463364e-06, "loss": 0.0007, "step": 179210 }, { "epoch": 2.9325042951812157, "grad_norm": 0.2207474410533905, "learning_rate": 2.0171093920642524e-06, "loss": 0.0008, "step": 179220 }, { "epoch": 2.9326679211322917, "grad_norm": 0.02365906536579132, "learning_rate": 2.016536436000772e-06, "loss": 0.0004, "step": 179230 }, { "epoch": 2.9328315470833672, "grad_norm": 0.04387802630662918, "learning_rate": 2.0159635407675776e-06, "loss": 0.0006, "step": 179240 }, { "epoch": 2.9329951730344432, "grad_norm": 0.05069315806031227, "learning_rate": 2.015390706376347e-06, "loss": 0.0011, "step": 179250 }, { "epoch": 2.9331587989855192, "grad_norm": 0.0063433777540922165, "learning_rate": 2.0148179328387617e-06, "loss": 0.0004, "step": 179260 }, { "epoch": 2.933322424936595, "grad_norm": 0.04299296438694, "learning_rate": 2.0142452201665e-06, "loss": 0.0004, "step": 179270 }, { "epoch": 2.933486050887671, "grad_norm": 0.029629239812493324, "learning_rate": 2.0136725683712405e-06, "loss": 0.0005, "step": 179280 }, { "epoch": 2.933649676838747, "grad_norm": 0.10261934995651245, "learning_rate": 2.013099977464656e-06, "loss": 0.0008, "step": 179290 }, { "epoch": 2.9338133027898223, "grad_norm": 0.0024775287602096796, "learning_rate": 2.012527447458422e-06, "loss": 0.0003, "step": 179300 }, { "epoch": 2.9339769287408983, "grad_norm": 0.008538900874555111, "learning_rate": 2.0119549783642135e-06, "loss": 0.0007, "step": 179310 }, { "epoch": 2.9341405546919743, "grad_norm": 0.009138255380094051, "learning_rate": 2.0113825701937033e-06, "loss": 0.0006, "step": 179320 }, { "epoch": 2.93430418064305, "grad_norm": 0.05070646107196808, "learning_rate": 2.010810222958559e-06, "loss": 0.0009, "step": 179330 }, { "epoch": 2.934467806594126, "grad_norm": 0.041477322578430176, "learning_rate": 2.0102379366704518e-06, "loss": 0.0008, "step": 179340 }, { "epoch": 2.934631432545202, "grad_norm": 0.036768585443496704, "learning_rate": 2.0096657113410504e-06, "loss": 0.0008, "step": 179350 }, { "epoch": 2.9347950584962774, "grad_norm": 0.07371758669614792, "learning_rate": 2.009093546982024e-06, "loss": 0.0008, "step": 179360 }, { "epoch": 2.9349586844473534, "grad_norm": 0.02314952202141285, "learning_rate": 2.0085214436050342e-06, "loss": 0.0006, "step": 179370 }, { "epoch": 2.9351223103984294, "grad_norm": 0.002642572158947587, "learning_rate": 2.007949401221749e-06, "loss": 0.0007, "step": 179380 }, { "epoch": 2.935285936349505, "grad_norm": 0.1258433610200882, "learning_rate": 2.00737741984383e-06, "loss": 0.001, "step": 179390 }, { "epoch": 2.935449562300581, "grad_norm": 0.04603440687060356, "learning_rate": 2.0068054994829423e-06, "loss": 0.0005, "step": 179400 }, { "epoch": 2.935613188251657, "grad_norm": 0.03215238079428673, "learning_rate": 2.0062336401507426e-06, "loss": 0.0006, "step": 179410 }, { "epoch": 2.9357768142027325, "grad_norm": 0.02932177484035492, "learning_rate": 2.0056618418588936e-06, "loss": 0.0008, "step": 179420 }, { "epoch": 2.9359404401538085, "grad_norm": 0.030668053776025772, "learning_rate": 2.005090104619053e-06, "loss": 0.0004, "step": 179430 }, { "epoch": 2.9361040661048845, "grad_norm": 0.016643527895212173, "learning_rate": 2.004518428442879e-06, "loss": 0.0003, "step": 179440 }, { "epoch": 2.93626769205596, "grad_norm": 0.108202263712883, "learning_rate": 2.003946813342026e-06, "loss": 0.001, "step": 179450 }, { "epoch": 2.936431318007036, "grad_norm": 0.012368196621537209, "learning_rate": 2.0033752593281485e-06, "loss": 0.0004, "step": 179460 }, { "epoch": 2.9365949439581116, "grad_norm": 0.026053689420223236, "learning_rate": 2.002803766412902e-06, "loss": 0.0011, "step": 179470 }, { "epoch": 2.9367585699091876, "grad_norm": 0.0030558346770703793, "learning_rate": 2.0022323346079387e-06, "loss": 0.0006, "step": 179480 }, { "epoch": 2.9369221958602636, "grad_norm": 0.015573738142848015, "learning_rate": 2.001660963924907e-06, "loss": 0.0004, "step": 179490 }, { "epoch": 2.937085821811339, "grad_norm": 0.23011772334575653, "learning_rate": 2.00108965437546e-06, "loss": 0.0018, "step": 179500 }, { "epoch": 2.937249447762415, "grad_norm": 0.025816641747951508, "learning_rate": 2.0005184059712405e-06, "loss": 0.001, "step": 179510 }, { "epoch": 2.9374130737134907, "grad_norm": 0.12173450738191605, "learning_rate": 1.9999472187239037e-06, "loss": 0.0009, "step": 179520 }, { "epoch": 2.9375766996645667, "grad_norm": 0.007833899930119514, "learning_rate": 1.99937609264509e-06, "loss": 0.0005, "step": 179530 }, { "epoch": 2.9377403256156427, "grad_norm": 0.07480165362358093, "learning_rate": 1.9988050277464473e-06, "loss": 0.0008, "step": 179540 }, { "epoch": 2.9379039515667182, "grad_norm": 0.022855618968605995, "learning_rate": 1.998234024039614e-06, "loss": 0.0019, "step": 179550 }, { "epoch": 2.9380675775177942, "grad_norm": 0.00719665689393878, "learning_rate": 1.9976630815362402e-06, "loss": 0.0012, "step": 179560 }, { "epoch": 2.9382312034688702, "grad_norm": 0.00042468414176255465, "learning_rate": 1.9970922002479604e-06, "loss": 0.001, "step": 179570 }, { "epoch": 2.938394829419946, "grad_norm": 0.0465422160923481, "learning_rate": 1.996521380186419e-06, "loss": 0.0004, "step": 179580 }, { "epoch": 2.938558455371022, "grad_norm": 0.005737567786127329, "learning_rate": 1.9959506213632483e-06, "loss": 0.0008, "step": 179590 }, { "epoch": 2.938722081322098, "grad_norm": 0.022114204242825508, "learning_rate": 1.9953799237900934e-06, "loss": 0.0005, "step": 179600 }, { "epoch": 2.9388857072731733, "grad_norm": 0.033067502081394196, "learning_rate": 1.9948092874785845e-06, "loss": 0.0006, "step": 179610 }, { "epoch": 2.9390493332242493, "grad_norm": 0.021103374660015106, "learning_rate": 1.9942387124403607e-06, "loss": 0.0003, "step": 179620 }, { "epoch": 2.9392129591753253, "grad_norm": 0.009794916026294231, "learning_rate": 1.9936681986870493e-06, "loss": 0.0011, "step": 179630 }, { "epoch": 2.939376585126401, "grad_norm": 0.0712134912610054, "learning_rate": 1.99309774623029e-06, "loss": 0.0009, "step": 179640 }, { "epoch": 2.939540211077477, "grad_norm": 0.03182503953576088, "learning_rate": 1.9925273550817092e-06, "loss": 0.0011, "step": 179650 }, { "epoch": 2.939703837028553, "grad_norm": 0.015390646643936634, "learning_rate": 1.9919570252529393e-06, "loss": 0.0009, "step": 179660 }, { "epoch": 2.9398674629796284, "grad_norm": 0.054408375173807144, "learning_rate": 1.9913867567556033e-06, "loss": 0.0006, "step": 179670 }, { "epoch": 2.9400310889307044, "grad_norm": 0.03489431366324425, "learning_rate": 1.9908165496013365e-06, "loss": 0.0004, "step": 179680 }, { "epoch": 2.9401947148817804, "grad_norm": 0.004111845046281815, "learning_rate": 1.9902464038017594e-06, "loss": 0.0006, "step": 179690 }, { "epoch": 2.940358340832856, "grad_norm": 0.0700371116399765, "learning_rate": 1.9896763193685004e-06, "loss": 0.0004, "step": 179700 }, { "epoch": 2.940521966783932, "grad_norm": 0.03578393906354904, "learning_rate": 1.989106296313179e-06, "loss": 0.0005, "step": 179710 }, { "epoch": 2.940685592735008, "grad_norm": 0.0004564893606584519, "learning_rate": 1.9885363346474194e-06, "loss": 0.0004, "step": 179720 }, { "epoch": 2.9408492186860835, "grad_norm": 0.04875173419713974, "learning_rate": 1.987966434382843e-06, "loss": 0.0005, "step": 179730 }, { "epoch": 2.9410128446371595, "grad_norm": 0.04391999542713165, "learning_rate": 1.987396595531071e-06, "loss": 0.0005, "step": 179740 }, { "epoch": 2.9411764705882355, "grad_norm": 0.02000267244875431, "learning_rate": 1.9868268181037186e-06, "loss": 0.0004, "step": 179750 }, { "epoch": 2.941340096539311, "grad_norm": 0.0015610981499776244, "learning_rate": 1.986257102112405e-06, "loss": 0.0004, "step": 179760 }, { "epoch": 2.941503722490387, "grad_norm": 0.037697870284318924, "learning_rate": 1.985687447568746e-06, "loss": 0.0004, "step": 179770 }, { "epoch": 2.941667348441463, "grad_norm": 0.04442691057920456, "learning_rate": 1.9851178544843576e-06, "loss": 0.0008, "step": 179780 }, { "epoch": 2.9418309743925386, "grad_norm": 0.010954941622912884, "learning_rate": 1.9845483228708506e-06, "loss": 0.0004, "step": 179790 }, { "epoch": 2.9419946003436146, "grad_norm": 0.08172059059143066, "learning_rate": 1.983978852739839e-06, "loss": 0.0006, "step": 179800 }, { "epoch": 2.9421582262946906, "grad_norm": 0.00037079074536450207, "learning_rate": 1.9834094441029334e-06, "loss": 0.0007, "step": 179810 }, { "epoch": 2.942321852245766, "grad_norm": 0.024218060076236725, "learning_rate": 1.9828400969717446e-06, "loss": 0.001, "step": 179820 }, { "epoch": 2.942485478196842, "grad_norm": 0.017300114035606384, "learning_rate": 1.9822708113578796e-06, "loss": 0.0009, "step": 179830 }, { "epoch": 2.9426491041479177, "grad_norm": 0.016491558402776718, "learning_rate": 1.981701587272946e-06, "loss": 0.0007, "step": 179840 }, { "epoch": 2.9428127300989937, "grad_norm": 0.0037173444870859385, "learning_rate": 1.9811324247285503e-06, "loss": 0.0007, "step": 179850 }, { "epoch": 2.9429763560500697, "grad_norm": 0.014881127513945103, "learning_rate": 1.980563323736298e-06, "loss": 0.0016, "step": 179860 }, { "epoch": 2.9431399820011452, "grad_norm": 0.0357307605445385, "learning_rate": 1.9799942843077907e-06, "loss": 0.0006, "step": 179870 }, { "epoch": 2.9433036079522212, "grad_norm": 0.001042304327711463, "learning_rate": 1.979425306454631e-06, "loss": 0.0005, "step": 179880 }, { "epoch": 2.943467233903297, "grad_norm": 0.019136108458042145, "learning_rate": 1.9788563901884206e-06, "loss": 0.0004, "step": 179890 }, { "epoch": 2.943630859854373, "grad_norm": 0.05704640597105026, "learning_rate": 1.9782875355207605e-06, "loss": 0.0012, "step": 179900 }, { "epoch": 2.943794485805449, "grad_norm": 0.03721858188509941, "learning_rate": 1.977718742463246e-06, "loss": 0.0011, "step": 179910 }, { "epoch": 2.9439581117565243, "grad_norm": 0.01231155265122652, "learning_rate": 1.977150011027476e-06, "loss": 0.0005, "step": 179920 }, { "epoch": 2.9441217377076003, "grad_norm": 0.02594136819243431, "learning_rate": 1.9765813412250467e-06, "loss": 0.0006, "step": 179930 }, { "epoch": 2.9442853636586763, "grad_norm": 0.09068949520587921, "learning_rate": 1.976012733067554e-06, "loss": 0.0007, "step": 179940 }, { "epoch": 2.944448989609752, "grad_norm": 0.1355009227991104, "learning_rate": 1.975444186566588e-06, "loss": 0.0008, "step": 179950 }, { "epoch": 2.944612615560828, "grad_norm": 0.015261875465512276, "learning_rate": 1.9748757017337433e-06, "loss": 0.0009, "step": 179960 }, { "epoch": 2.944776241511904, "grad_norm": 0.02803780883550644, "learning_rate": 1.97430727858061e-06, "loss": 0.0007, "step": 179970 }, { "epoch": 2.9449398674629794, "grad_norm": 0.0003909205552190542, "learning_rate": 1.9737389171187793e-06, "loss": 0.0008, "step": 179980 }, { "epoch": 2.9451034934140554, "grad_norm": 0.0062677739188075066, "learning_rate": 1.973170617359837e-06, "loss": 0.0012, "step": 179990 }, { "epoch": 2.9452671193651314, "grad_norm": 0.00221143732778728, "learning_rate": 1.9726023793153714e-06, "loss": 0.0013, "step": 180000 }, { "epoch": 2.945430745316207, "grad_norm": 0.007921312935650349, "learning_rate": 1.9720342029969693e-06, "loss": 0.0003, "step": 180010 }, { "epoch": 2.945594371267283, "grad_norm": 0.010115386918187141, "learning_rate": 1.971466088416216e-06, "loss": 0.0004, "step": 180020 }, { "epoch": 2.945757997218359, "grad_norm": 0.008189888671040535, "learning_rate": 1.9708980355846914e-06, "loss": 0.0008, "step": 180030 }, { "epoch": 2.9459216231694345, "grad_norm": 0.055575620383024216, "learning_rate": 1.9703300445139793e-06, "loss": 0.0004, "step": 180040 }, { "epoch": 2.9460852491205105, "grad_norm": 0.0589853934943676, "learning_rate": 1.9697621152156614e-06, "loss": 0.0006, "step": 180050 }, { "epoch": 2.9462488750715865, "grad_norm": 0.0021216950844973326, "learning_rate": 1.969194247701319e-06, "loss": 0.0006, "step": 180060 }, { "epoch": 2.946412501022662, "grad_norm": 0.0006392739014700055, "learning_rate": 1.9686264419825262e-06, "loss": 0.001, "step": 180070 }, { "epoch": 2.946576126973738, "grad_norm": 0.009777200408279896, "learning_rate": 1.968058698070862e-06, "loss": 0.0011, "step": 180080 }, { "epoch": 2.946739752924814, "grad_norm": 0.01769094355404377, "learning_rate": 1.9674910159779024e-06, "loss": 0.0008, "step": 180090 }, { "epoch": 2.9469033788758896, "grad_norm": 0.018064169213175774, "learning_rate": 1.966923395715224e-06, "loss": 0.0008, "step": 180100 }, { "epoch": 2.9470670048269656, "grad_norm": 0.016961511224508286, "learning_rate": 1.9663558372943966e-06, "loss": 0.001, "step": 180110 }, { "epoch": 2.9472306307780416, "grad_norm": 0.033928025513887405, "learning_rate": 1.965788340726993e-06, "loss": 0.0007, "step": 180120 }, { "epoch": 2.947394256729117, "grad_norm": 0.04509386047720909, "learning_rate": 1.9652209060245844e-06, "loss": 0.0007, "step": 180130 }, { "epoch": 2.947557882680193, "grad_norm": 0.014369618147611618, "learning_rate": 1.9646535331987426e-06, "loss": 0.0009, "step": 180140 }, { "epoch": 2.947721508631269, "grad_norm": 0.023561878129839897, "learning_rate": 1.9640862222610324e-06, "loss": 0.0016, "step": 180150 }, { "epoch": 2.9478851345823447, "grad_norm": 0.026243576779961586, "learning_rate": 1.963518973223024e-06, "loss": 0.0007, "step": 180160 }, { "epoch": 2.9480487605334207, "grad_norm": 0.003539957106113434, "learning_rate": 1.9629517860962766e-06, "loss": 0.0009, "step": 180170 }, { "epoch": 2.9482123864844967, "grad_norm": 0.0025224988348782063, "learning_rate": 1.962384660892364e-06, "loss": 0.001, "step": 180180 }, { "epoch": 2.9483760124355722, "grad_norm": 0.19156871736049652, "learning_rate": 1.9618175976228426e-06, "loss": 0.0007, "step": 180190 }, { "epoch": 2.9485396383866482, "grad_norm": 0.04852854833006859, "learning_rate": 1.9612505962992785e-06, "loss": 0.0004, "step": 180200 }, { "epoch": 2.9487032643377242, "grad_norm": 0.02987154759466648, "learning_rate": 1.9606836569332267e-06, "loss": 0.0009, "step": 180210 }, { "epoch": 2.9488668902888, "grad_norm": 0.10853470861911774, "learning_rate": 1.9601167795362537e-06, "loss": 0.0004, "step": 180220 }, { "epoch": 2.9490305162398758, "grad_norm": 0.07439246028661728, "learning_rate": 1.9595499641199128e-06, "loss": 0.0007, "step": 180230 }, { "epoch": 2.9491941421909513, "grad_norm": 0.001962959999218583, "learning_rate": 1.9589832106957635e-06, "loss": 0.0005, "step": 180240 }, { "epoch": 2.9493577681420273, "grad_norm": 0.034519389271736145, "learning_rate": 1.9584165192753573e-06, "loss": 0.0006, "step": 180250 }, { "epoch": 2.9495213940931033, "grad_norm": 0.002737648319453001, "learning_rate": 1.9578498898702545e-06, "loss": 0.0008, "step": 180260 }, { "epoch": 2.949685020044179, "grad_norm": 0.11872908473014832, "learning_rate": 1.957283322492004e-06, "loss": 0.001, "step": 180270 }, { "epoch": 2.949848645995255, "grad_norm": 0.035123828798532486, "learning_rate": 1.9567168171521607e-06, "loss": 0.0005, "step": 180280 }, { "epoch": 2.9500122719463304, "grad_norm": 0.06992211192846298, "learning_rate": 1.956150373862269e-06, "loss": 0.0009, "step": 180290 }, { "epoch": 2.9501758978974064, "grad_norm": 0.02545127645134926, "learning_rate": 1.9555839926338863e-06, "loss": 0.0003, "step": 180300 }, { "epoch": 2.9503395238484824, "grad_norm": 0.006787133868783712, "learning_rate": 1.9550176734785554e-06, "loss": 0.0008, "step": 180310 }, { "epoch": 2.950503149799558, "grad_norm": 0.01759186200797558, "learning_rate": 1.954451416407826e-06, "loss": 0.0011, "step": 180320 }, { "epoch": 2.950666775750634, "grad_norm": 0.038318052887916565, "learning_rate": 1.9538852214332397e-06, "loss": 0.0007, "step": 180330 }, { "epoch": 2.95083040170171, "grad_norm": 0.009524969384074211, "learning_rate": 1.953319088566344e-06, "loss": 0.003, "step": 180340 }, { "epoch": 2.9509940276527855, "grad_norm": 0.02763064205646515, "learning_rate": 1.952753017818681e-06, "loss": 0.0006, "step": 180350 }, { "epoch": 2.9511576536038615, "grad_norm": 0.011833542957901955, "learning_rate": 1.9521870092017937e-06, "loss": 0.0006, "step": 180360 }, { "epoch": 2.9513212795549375, "grad_norm": 0.04564597085118294, "learning_rate": 1.9516210627272193e-06, "loss": 0.0013, "step": 180370 }, { "epoch": 2.951484905506013, "grad_norm": 0.0690382719039917, "learning_rate": 1.9510551784064992e-06, "loss": 0.0005, "step": 180380 }, { "epoch": 2.951648531457089, "grad_norm": 0.12147779017686844, "learning_rate": 1.9504893562511712e-06, "loss": 0.0007, "step": 180390 }, { "epoch": 2.951812157408165, "grad_norm": 0.07460761070251465, "learning_rate": 1.949923596272773e-06, "loss": 0.0006, "step": 180400 }, { "epoch": 2.9519757833592406, "grad_norm": 0.0210939459502697, "learning_rate": 1.949357898482837e-06, "loss": 0.0006, "step": 180410 }, { "epoch": 2.9521394093103166, "grad_norm": 0.056237343698740005, "learning_rate": 1.948792262892899e-06, "loss": 0.0009, "step": 180420 }, { "epoch": 2.9523030352613926, "grad_norm": 0.07107891142368317, "learning_rate": 1.948226689514492e-06, "loss": 0.0005, "step": 180430 }, { "epoch": 2.952466661212468, "grad_norm": 0.008649609051644802, "learning_rate": 1.9476611783591487e-06, "loss": 0.0009, "step": 180440 }, { "epoch": 2.952630287163544, "grad_norm": 0.04106126353144646, "learning_rate": 1.9470957294383968e-06, "loss": 0.0007, "step": 180450 }, { "epoch": 2.95279391311462, "grad_norm": 0.03693629428744316, "learning_rate": 1.946530342763766e-06, "loss": 0.0003, "step": 180460 }, { "epoch": 2.9529575390656957, "grad_norm": 0.013791024684906006, "learning_rate": 1.9459650183467855e-06, "loss": 0.0005, "step": 180470 }, { "epoch": 2.9531211650167717, "grad_norm": 0.06423473358154297, "learning_rate": 1.9453997561989816e-06, "loss": 0.0007, "step": 180480 }, { "epoch": 2.9532847909678477, "grad_norm": 0.05358632281422615, "learning_rate": 1.9448345563318777e-06, "loss": 0.0007, "step": 180490 }, { "epoch": 2.9534484169189232, "grad_norm": 0.004412706475704908, "learning_rate": 1.944269418756999e-06, "loss": 0.0005, "step": 180500 }, { "epoch": 2.9536120428699992, "grad_norm": 0.08906673640012741, "learning_rate": 1.9437043434858677e-06, "loss": 0.0007, "step": 180510 }, { "epoch": 2.9537756688210752, "grad_norm": 0.038213081657886505, "learning_rate": 1.943139330530008e-06, "loss": 0.0004, "step": 180520 }, { "epoch": 2.953939294772151, "grad_norm": 0.059887923300266266, "learning_rate": 1.942574379900935e-06, "loss": 0.0003, "step": 180530 }, { "epoch": 2.9541029207232268, "grad_norm": 0.012224329635500908, "learning_rate": 1.94200949161017e-06, "loss": 0.0005, "step": 180540 }, { "epoch": 2.9542665466743028, "grad_norm": 0.08117301762104034, "learning_rate": 1.9414446656692317e-06, "loss": 0.0009, "step": 180550 }, { "epoch": 2.9544301726253783, "grad_norm": 0.026969650760293007, "learning_rate": 1.9408799020896365e-06, "loss": 0.0013, "step": 180560 }, { "epoch": 2.9545937985764543, "grad_norm": 0.011169016361236572, "learning_rate": 1.9403152008828967e-06, "loss": 0.0005, "step": 180570 }, { "epoch": 2.9547574245275303, "grad_norm": 0.0019606545101851225, "learning_rate": 1.9397505620605278e-06, "loss": 0.0004, "step": 180580 }, { "epoch": 2.954921050478606, "grad_norm": 0.018687773495912552, "learning_rate": 1.939185985634042e-06, "loss": 0.0008, "step": 180590 }, { "epoch": 2.955084676429682, "grad_norm": 0.020117301493883133, "learning_rate": 1.9386214716149536e-06, "loss": 0.0005, "step": 180600 }, { "epoch": 2.955248302380758, "grad_norm": 0.0017091863555833697, "learning_rate": 1.9380570200147677e-06, "loss": 0.0011, "step": 180610 }, { "epoch": 2.9554119283318334, "grad_norm": 0.04291628673672676, "learning_rate": 1.9374926308449944e-06, "loss": 0.0006, "step": 180620 }, { "epoch": 2.9555755542829094, "grad_norm": 0.04192960634827614, "learning_rate": 1.9369283041171423e-06, "loss": 0.0005, "step": 180630 }, { "epoch": 2.955739180233985, "grad_norm": 0.026409095153212547, "learning_rate": 1.9363640398427185e-06, "loss": 0.0008, "step": 180640 }, { "epoch": 2.955902806185061, "grad_norm": 0.027174437418580055, "learning_rate": 1.935799838033225e-06, "loss": 0.0004, "step": 180650 }, { "epoch": 2.956066432136137, "grad_norm": 0.009535891003906727, "learning_rate": 1.9352356987001666e-06, "loss": 0.0006, "step": 180660 }, { "epoch": 2.9562300580872125, "grad_norm": 0.02027169056236744, "learning_rate": 1.9346716218550456e-06, "loss": 0.0006, "step": 180670 }, { "epoch": 2.9563936840382885, "grad_norm": 0.04828871041536331, "learning_rate": 1.934107607509365e-06, "loss": 0.0005, "step": 180680 }, { "epoch": 2.956557309989364, "grad_norm": 0.00693202530965209, "learning_rate": 1.9335436556746214e-06, "loss": 0.0005, "step": 180690 }, { "epoch": 2.95672093594044, "grad_norm": 0.00035483381361700594, "learning_rate": 1.9329797663623146e-06, "loss": 0.0017, "step": 180700 }, { "epoch": 2.956884561891516, "grad_norm": 0.03412143886089325, "learning_rate": 1.9324159395839422e-06, "loss": 0.0005, "step": 180710 }, { "epoch": 2.9570481878425916, "grad_norm": 0.020005105063319206, "learning_rate": 1.931852175351001e-06, "loss": 0.0007, "step": 180720 }, { "epoch": 2.9572118137936676, "grad_norm": 0.0031494495924562216, "learning_rate": 1.9312884736749837e-06, "loss": 0.0004, "step": 180730 }, { "epoch": 2.9573754397447436, "grad_norm": 0.1106061264872551, "learning_rate": 1.930724834567384e-06, "loss": 0.001, "step": 180740 }, { "epoch": 2.957539065695819, "grad_norm": 0.037522971630096436, "learning_rate": 1.930161258039695e-06, "loss": 0.0009, "step": 180750 }, { "epoch": 2.957702691646895, "grad_norm": 0.02497316710650921, "learning_rate": 1.9295977441034087e-06, "loss": 0.001, "step": 180760 }, { "epoch": 2.957866317597971, "grad_norm": 0.022134976461529732, "learning_rate": 1.9290342927700114e-06, "loss": 0.0007, "step": 180770 }, { "epoch": 2.9580299435490467, "grad_norm": 0.003368015866726637, "learning_rate": 1.928470904050994e-06, "loss": 0.0007, "step": 180780 }, { "epoch": 2.9581935695001227, "grad_norm": 0.06943302601575851, "learning_rate": 1.9279075779578406e-06, "loss": 0.0009, "step": 180790 }, { "epoch": 2.9583571954511987, "grad_norm": 0.0007510258001275361, "learning_rate": 1.9273443145020417e-06, "loss": 0.0004, "step": 180800 }, { "epoch": 2.9585208214022742, "grad_norm": 0.08185969293117523, "learning_rate": 1.926781113695077e-06, "loss": 0.0007, "step": 180810 }, { "epoch": 2.9586844473533502, "grad_norm": 0.014579552225768566, "learning_rate": 1.926217975548434e-06, "loss": 0.0003, "step": 180820 }, { "epoch": 2.9588480733044262, "grad_norm": 0.020854402333498, "learning_rate": 1.9256549000735885e-06, "loss": 0.0003, "step": 180830 }, { "epoch": 2.959011699255502, "grad_norm": 0.069631427526474, "learning_rate": 1.9250918872820284e-06, "loss": 0.0011, "step": 180840 }, { "epoch": 2.9591753252065778, "grad_norm": 0.07426455616950989, "learning_rate": 1.924528937185228e-06, "loss": 0.0006, "step": 180850 }, { "epoch": 2.9593389511576538, "grad_norm": 0.052973102778196335, "learning_rate": 1.9239660497946686e-06, "loss": 0.0008, "step": 180860 }, { "epoch": 2.9595025771087293, "grad_norm": 0.03220529854297638, "learning_rate": 1.9234032251218222e-06, "loss": 0.0008, "step": 180870 }, { "epoch": 2.9596662030598053, "grad_norm": 0.011187829077243805, "learning_rate": 1.9228404631781712e-06, "loss": 0.0007, "step": 180880 }, { "epoch": 2.9598298290108813, "grad_norm": 0.025246521458029747, "learning_rate": 1.9222777639751836e-06, "loss": 0.0007, "step": 180890 }, { "epoch": 2.959993454961957, "grad_norm": 0.05032828450202942, "learning_rate": 1.9217151275243368e-06, "loss": 0.0005, "step": 180900 }, { "epoch": 2.960157080913033, "grad_norm": 0.04255085438489914, "learning_rate": 1.9211525538370973e-06, "loss": 0.0005, "step": 180910 }, { "epoch": 2.960320706864109, "grad_norm": 0.037487830966711044, "learning_rate": 1.9205900429249426e-06, "loss": 0.001, "step": 180920 }, { "epoch": 2.9604843328151844, "grad_norm": 0.050229888409376144, "learning_rate": 1.920027594799336e-06, "loss": 0.001, "step": 180930 }, { "epoch": 2.9606479587662604, "grad_norm": 0.01948412135243416, "learning_rate": 1.9194652094717485e-06, "loss": 0.0004, "step": 180940 }, { "epoch": 2.9608115847173364, "grad_norm": 0.048443954437971115, "learning_rate": 1.9189028869536423e-06, "loss": 0.0007, "step": 180950 }, { "epoch": 2.960975210668412, "grad_norm": 0.002512963954359293, "learning_rate": 1.9183406272564896e-06, "loss": 0.0004, "step": 180960 }, { "epoch": 2.961138836619488, "grad_norm": 0.12863706052303314, "learning_rate": 1.917778430391749e-06, "loss": 0.0011, "step": 180970 }, { "epoch": 2.961302462570564, "grad_norm": 0.0293224286288023, "learning_rate": 1.9172162963708864e-06, "loss": 0.0009, "step": 180980 }, { "epoch": 2.9614660885216395, "grad_norm": 0.03076287917792797, "learning_rate": 1.91665422520536e-06, "loss": 0.0006, "step": 180990 }, { "epoch": 2.9616297144727155, "grad_norm": 0.02669382281601429, "learning_rate": 1.9160922169066313e-06, "loss": 0.0008, "step": 181000 }, { "epoch": 2.961793340423791, "grad_norm": 0.01836228370666504, "learning_rate": 1.91553027148616e-06, "loss": 0.0009, "step": 181010 }, { "epoch": 2.961956966374867, "grad_norm": 0.005850083660334349, "learning_rate": 1.914968388955405e-06, "loss": 0.0003, "step": 181020 }, { "epoch": 2.962120592325943, "grad_norm": 0.0362529456615448, "learning_rate": 1.9144065693258187e-06, "loss": 0.0008, "step": 181030 }, { "epoch": 2.9622842182770186, "grad_norm": 0.005143057554960251, "learning_rate": 1.9138448126088584e-06, "loss": 0.0008, "step": 181040 }, { "epoch": 2.9624478442280946, "grad_norm": 0.04485630244016647, "learning_rate": 1.9132831188159773e-06, "loss": 0.0004, "step": 181050 }, { "epoch": 2.96261147017917, "grad_norm": 0.0746675357222557, "learning_rate": 1.912721487958631e-06, "loss": 0.0006, "step": 181060 }, { "epoch": 2.962775096130246, "grad_norm": 0.038296643644571304, "learning_rate": 1.9121599200482656e-06, "loss": 0.0008, "step": 181070 }, { "epoch": 2.962938722081322, "grad_norm": 0.005855101626366377, "learning_rate": 1.911598415096333e-06, "loss": 0.001, "step": 181080 }, { "epoch": 2.9631023480323977, "grad_norm": 0.005519104655832052, "learning_rate": 1.9110369731142823e-06, "loss": 0.0003, "step": 181090 }, { "epoch": 2.9632659739834737, "grad_norm": 0.1148848682641983, "learning_rate": 1.910475594113563e-06, "loss": 0.001, "step": 181100 }, { "epoch": 2.9634295999345497, "grad_norm": 0.021857384592294693, "learning_rate": 1.909914278105616e-06, "loss": 0.0005, "step": 181110 }, { "epoch": 2.9635932258856252, "grad_norm": 0.017125269398093224, "learning_rate": 1.9093530251018892e-06, "loss": 0.0002, "step": 181120 }, { "epoch": 2.9637568518367012, "grad_norm": 0.018663665279746056, "learning_rate": 1.9087918351138264e-06, "loss": 0.0008, "step": 181130 }, { "epoch": 2.9639204777877772, "grad_norm": 0.031027309596538544, "learning_rate": 1.90823070815287e-06, "loss": 0.0006, "step": 181140 }, { "epoch": 2.964084103738853, "grad_norm": 0.018175462260842323, "learning_rate": 1.9076696442304586e-06, "loss": 0.0006, "step": 181150 }, { "epoch": 2.964247729689929, "grad_norm": 0.0673060491681099, "learning_rate": 1.9071086433580332e-06, "loss": 0.0006, "step": 181160 }, { "epoch": 2.9644113556410048, "grad_norm": 0.1133841872215271, "learning_rate": 1.9065477055470344e-06, "loss": 0.0004, "step": 181170 }, { "epoch": 2.9645749815920803, "grad_norm": 0.03479275479912758, "learning_rate": 1.9059868308088948e-06, "loss": 0.0007, "step": 181180 }, { "epoch": 2.9647386075431563, "grad_norm": 0.03789222240447998, "learning_rate": 1.9054260191550522e-06, "loss": 0.0003, "step": 181190 }, { "epoch": 2.9649022334942323, "grad_norm": 0.04386139661073685, "learning_rate": 1.9048652705969412e-06, "loss": 0.0006, "step": 181200 }, { "epoch": 2.965065859445308, "grad_norm": 0.003436374245211482, "learning_rate": 1.9043045851459973e-06, "loss": 0.0012, "step": 181210 }, { "epoch": 2.965229485396384, "grad_norm": 0.020431604236364365, "learning_rate": 1.9037439628136479e-06, "loss": 0.0012, "step": 181220 }, { "epoch": 2.96539311134746, "grad_norm": 0.040235985070466995, "learning_rate": 1.9031834036113256e-06, "loss": 0.0008, "step": 181230 }, { "epoch": 2.9655567372985354, "grad_norm": 0.05113435164093971, "learning_rate": 1.9026229075504604e-06, "loss": 0.0005, "step": 181240 }, { "epoch": 2.9657203632496114, "grad_norm": 0.029458757489919662, "learning_rate": 1.9020624746424815e-06, "loss": 0.0006, "step": 181250 }, { "epoch": 2.9658839892006874, "grad_norm": 0.005910452920943499, "learning_rate": 1.9015021048988124e-06, "loss": 0.001, "step": 181260 }, { "epoch": 2.966047615151763, "grad_norm": 0.008635308593511581, "learning_rate": 1.900941798330882e-06, "loss": 0.0014, "step": 181270 }, { "epoch": 2.966211241102839, "grad_norm": 0.01708533614873886, "learning_rate": 1.9003815549501093e-06, "loss": 0.0003, "step": 181280 }, { "epoch": 2.966374867053915, "grad_norm": 0.04600028321146965, "learning_rate": 1.8998213747679233e-06, "loss": 0.0007, "step": 181290 }, { "epoch": 2.9665384930049905, "grad_norm": 0.024472977966070175, "learning_rate": 1.899261257795742e-06, "loss": 0.0008, "step": 181300 }, { "epoch": 2.9667021189560665, "grad_norm": 0.13153938949108124, "learning_rate": 1.8987012040449877e-06, "loss": 0.0008, "step": 181310 }, { "epoch": 2.9668657449071425, "grad_norm": 0.0007835113792680204, "learning_rate": 1.898141213527075e-06, "loss": 0.0003, "step": 181320 }, { "epoch": 2.967029370858218, "grad_norm": 0.00968608632683754, "learning_rate": 1.8975812862534288e-06, "loss": 0.0006, "step": 181330 }, { "epoch": 2.967192996809294, "grad_norm": 0.007010368164628744, "learning_rate": 1.8970214222354593e-06, "loss": 0.0008, "step": 181340 }, { "epoch": 2.96735662276037, "grad_norm": 0.10961490869522095, "learning_rate": 1.896461621484586e-06, "loss": 0.0009, "step": 181350 }, { "epoch": 2.9675202487114456, "grad_norm": 0.022901922464370728, "learning_rate": 1.8959018840122174e-06, "loss": 0.0005, "step": 181360 }, { "epoch": 2.9676838746625216, "grad_norm": 0.10200650990009308, "learning_rate": 1.895342209829773e-06, "loss": 0.0005, "step": 181370 }, { "epoch": 2.9678475006135976, "grad_norm": 0.013007395900785923, "learning_rate": 1.8947825989486585e-06, "loss": 0.001, "step": 181380 }, { "epoch": 2.968011126564673, "grad_norm": 0.04836577922105789, "learning_rate": 1.8942230513802878e-06, "loss": 0.0007, "step": 181390 }, { "epoch": 2.968174752515749, "grad_norm": 0.013250875286757946, "learning_rate": 1.8936635671360642e-06, "loss": 0.0004, "step": 181400 }, { "epoch": 2.9683383784668247, "grad_norm": 0.004027748946100473, "learning_rate": 1.8931041462274018e-06, "loss": 0.0003, "step": 181410 }, { "epoch": 2.9685020044179007, "grad_norm": 0.03366398066282272, "learning_rate": 1.8925447886657022e-06, "loss": 0.0007, "step": 181420 }, { "epoch": 2.9686656303689767, "grad_norm": 0.00786176323890686, "learning_rate": 1.8919854944623734e-06, "loss": 0.0008, "step": 181430 }, { "epoch": 2.9688292563200522, "grad_norm": 0.03411861136555672, "learning_rate": 1.891426263628815e-06, "loss": 0.0009, "step": 181440 }, { "epoch": 2.9689928822711282, "grad_norm": 0.09516547620296478, "learning_rate": 1.8908670961764313e-06, "loss": 0.0009, "step": 181450 }, { "epoch": 2.969156508222204, "grad_norm": 0.0061918157152831554, "learning_rate": 1.890307992116624e-06, "loss": 0.0006, "step": 181460 }, { "epoch": 2.96932013417328, "grad_norm": 0.027990708127617836, "learning_rate": 1.889748951460793e-06, "loss": 0.0006, "step": 181470 }, { "epoch": 2.9694837601243558, "grad_norm": 0.026369858533143997, "learning_rate": 1.8891899742203346e-06, "loss": 0.0007, "step": 181480 }, { "epoch": 2.9696473860754313, "grad_norm": 0.028244897723197937, "learning_rate": 1.8886310604066466e-06, "loss": 0.0001, "step": 181490 }, { "epoch": 2.9698110120265073, "grad_norm": 0.024438554421067238, "learning_rate": 1.888072210031125e-06, "loss": 0.0005, "step": 181500 }, { "epoch": 2.9699746379775833, "grad_norm": 0.0064524272456765175, "learning_rate": 1.8875134231051661e-06, "loss": 0.0004, "step": 181510 }, { "epoch": 2.970138263928659, "grad_norm": 0.034254882484674454, "learning_rate": 1.88695469964016e-06, "loss": 0.0008, "step": 181520 }, { "epoch": 2.970301889879735, "grad_norm": 0.23342673480510712, "learning_rate": 1.8863960396475001e-06, "loss": 0.0007, "step": 181530 }, { "epoch": 2.970465515830811, "grad_norm": 0.03491093963384628, "learning_rate": 1.8858374431385767e-06, "loss": 0.0005, "step": 181540 }, { "epoch": 2.9706291417818864, "grad_norm": 0.061280231922864914, "learning_rate": 1.8852789101247814e-06, "loss": 0.0021, "step": 181550 }, { "epoch": 2.9707927677329624, "grad_norm": 0.0013410302344709635, "learning_rate": 1.884720440617498e-06, "loss": 0.0005, "step": 181560 }, { "epoch": 2.9709563936840384, "grad_norm": 0.0445261150598526, "learning_rate": 1.8841620346281154e-06, "loss": 0.0007, "step": 181570 }, { "epoch": 2.971120019635114, "grad_norm": 0.04003015533089638, "learning_rate": 1.883603692168019e-06, "loss": 0.0006, "step": 181580 }, { "epoch": 2.97128364558619, "grad_norm": 0.002335895551368594, "learning_rate": 1.8830454132485947e-06, "loss": 0.0005, "step": 181590 }, { "epoch": 2.971447271537266, "grad_norm": 0.06082199513912201, "learning_rate": 1.882487197881222e-06, "loss": 0.0009, "step": 181600 }, { "epoch": 2.9716108974883415, "grad_norm": 0.06091908738017082, "learning_rate": 1.8819290460772844e-06, "loss": 0.0006, "step": 181610 }, { "epoch": 2.9717745234394175, "grad_norm": 0.03621384873986244, "learning_rate": 1.8813709578481609e-06, "loss": 0.0017, "step": 181620 }, { "epoch": 2.9719381493904935, "grad_norm": 0.019302254542708397, "learning_rate": 1.8808129332052332e-06, "loss": 0.0008, "step": 181630 }, { "epoch": 2.972101775341569, "grad_norm": 0.01220126822590828, "learning_rate": 1.8802549721598757e-06, "loss": 0.0007, "step": 181640 }, { "epoch": 2.972265401292645, "grad_norm": 0.045111555606126785, "learning_rate": 1.8796970747234656e-06, "loss": 0.0007, "step": 181650 }, { "epoch": 2.972429027243721, "grad_norm": 0.0693468376994133, "learning_rate": 1.8791392409073783e-06, "loss": 0.0014, "step": 181660 }, { "epoch": 2.9725926531947966, "grad_norm": 0.0008574937819503248, "learning_rate": 1.8785814707229899e-06, "loss": 0.0007, "step": 181670 }, { "epoch": 2.9727562791458726, "grad_norm": 0.09449701011180878, "learning_rate": 1.878023764181669e-06, "loss": 0.0009, "step": 181680 }, { "epoch": 2.9729199050969486, "grad_norm": 0.0364382266998291, "learning_rate": 1.8774661212947876e-06, "loss": 0.0005, "step": 181690 }, { "epoch": 2.973083531048024, "grad_norm": 0.03169586881995201, "learning_rate": 1.8769085420737171e-06, "loss": 0.0006, "step": 181700 }, { "epoch": 2.9732471569991, "grad_norm": 0.002490698592737317, "learning_rate": 1.876351026529827e-06, "loss": 0.0008, "step": 181710 }, { "epoch": 2.973410782950176, "grad_norm": 0.028086429461836815, "learning_rate": 1.875793574674481e-06, "loss": 0.0002, "step": 181720 }, { "epoch": 2.9735744089012517, "grad_norm": 0.0030864973086863756, "learning_rate": 1.8752361865190466e-06, "loss": 0.0004, "step": 181730 }, { "epoch": 2.9737380348523277, "grad_norm": 0.019929181784391403, "learning_rate": 1.8746788620748896e-06, "loss": 0.0007, "step": 181740 }, { "epoch": 2.9739016608034037, "grad_norm": 0.05160602927207947, "learning_rate": 1.8741216013533736e-06, "loss": 0.0016, "step": 181750 }, { "epoch": 2.9740652867544792, "grad_norm": 0.010708438232541084, "learning_rate": 1.8735644043658585e-06, "loss": 0.0005, "step": 181760 }, { "epoch": 2.9742289127055552, "grad_norm": 0.0682608038187027, "learning_rate": 1.8730072711237063e-06, "loss": 0.0009, "step": 181770 }, { "epoch": 2.974392538656631, "grad_norm": 0.04144410043954849, "learning_rate": 1.8724502016382761e-06, "loss": 0.0005, "step": 181780 }, { "epoch": 2.9745561646077068, "grad_norm": 0.024607902392745018, "learning_rate": 1.871893195920928e-06, "loss": 0.0005, "step": 181790 }, { "epoch": 2.9747197905587828, "grad_norm": 0.06596670299768448, "learning_rate": 1.871336253983016e-06, "loss": 0.0007, "step": 181800 }, { "epoch": 2.9748834165098583, "grad_norm": 0.03231077268719673, "learning_rate": 1.8707793758358966e-06, "loss": 0.0014, "step": 181810 }, { "epoch": 2.9750470424609343, "grad_norm": 0.046888455748558044, "learning_rate": 1.8702225614909247e-06, "loss": 0.0013, "step": 181820 }, { "epoch": 2.97521066841201, "grad_norm": 0.004396927542984486, "learning_rate": 1.8696658109594545e-06, "loss": 0.001, "step": 181830 }, { "epoch": 2.975374294363086, "grad_norm": 0.08658561110496521, "learning_rate": 1.8691091242528341e-06, "loss": 0.0007, "step": 181840 }, { "epoch": 2.975537920314162, "grad_norm": 0.009640082716941833, "learning_rate": 1.868552501382417e-06, "loss": 0.001, "step": 181850 }, { "epoch": 2.9757015462652374, "grad_norm": 0.03197328373789787, "learning_rate": 1.86799594235955e-06, "loss": 0.0004, "step": 181860 }, { "epoch": 2.9758651722163134, "grad_norm": 0.002634972333908081, "learning_rate": 1.8674394471955847e-06, "loss": 0.0007, "step": 181870 }, { "epoch": 2.9760287981673894, "grad_norm": 0.06722450256347656, "learning_rate": 1.8668830159018624e-06, "loss": 0.0008, "step": 181880 }, { "epoch": 2.976192424118465, "grad_norm": 0.006923688109964132, "learning_rate": 1.866326648489733e-06, "loss": 0.0009, "step": 181890 }, { "epoch": 2.976356050069541, "grad_norm": 0.03115309402346611, "learning_rate": 1.865770344970535e-06, "loss": 0.0007, "step": 181900 }, { "epoch": 2.976519676020617, "grad_norm": 0.029913082718849182, "learning_rate": 1.8652141053556172e-06, "loss": 0.0006, "step": 181910 }, { "epoch": 2.9766833019716925, "grad_norm": 0.05362984165549278, "learning_rate": 1.8646579296563155e-06, "loss": 0.0008, "step": 181920 }, { "epoch": 2.9768469279227685, "grad_norm": 0.07379812747240067, "learning_rate": 1.8641018178839748e-06, "loss": 0.0006, "step": 181930 }, { "epoch": 2.9770105538738445, "grad_norm": 0.007586746942251921, "learning_rate": 1.8635457700499271e-06, "loss": 0.0008, "step": 181940 }, { "epoch": 2.97717417982492, "grad_norm": 0.022726647555828094, "learning_rate": 1.8629897861655172e-06, "loss": 0.0012, "step": 181950 }, { "epoch": 2.977337805775996, "grad_norm": 0.026071790605783463, "learning_rate": 1.8624338662420754e-06, "loss": 0.0015, "step": 181960 }, { "epoch": 2.977501431727072, "grad_norm": 0.04424740746617317, "learning_rate": 1.8618780102909412e-06, "loss": 0.0006, "step": 181970 }, { "epoch": 2.9776650576781476, "grad_norm": 0.2470429241657257, "learning_rate": 1.8613222183234414e-06, "loss": 0.0009, "step": 181980 }, { "epoch": 2.9778286836292236, "grad_norm": 0.07610169798135757, "learning_rate": 1.860766490350916e-06, "loss": 0.0007, "step": 181990 }, { "epoch": 2.9779923095802996, "grad_norm": 0.0014813847374171019, "learning_rate": 1.8602108263846903e-06, "loss": 0.0004, "step": 182000 }, { "epoch": 2.978155935531375, "grad_norm": 0.06392728537321091, "learning_rate": 1.8596552264360972e-06, "loss": 0.0006, "step": 182010 }, { "epoch": 2.978319561482451, "grad_norm": 0.024177148938179016, "learning_rate": 1.85909969051646e-06, "loss": 0.0006, "step": 182020 }, { "epoch": 2.978483187433527, "grad_norm": 0.005343405529856682, "learning_rate": 1.858544218637112e-06, "loss": 0.0012, "step": 182030 }, { "epoch": 2.9786468133846027, "grad_norm": 0.0748273953795433, "learning_rate": 1.8579888108093742e-06, "loss": 0.0009, "step": 182040 }, { "epoch": 2.9788104393356787, "grad_norm": 0.00753845414146781, "learning_rate": 1.8574334670445742e-06, "loss": 0.0004, "step": 182050 }, { "epoch": 2.9789740652867547, "grad_norm": 0.11055367439985275, "learning_rate": 1.8568781873540292e-06, "loss": 0.0004, "step": 182060 }, { "epoch": 2.9791376912378302, "grad_norm": 0.0037216353230178356, "learning_rate": 1.8563229717490688e-06, "loss": 0.0006, "step": 182070 }, { "epoch": 2.9793013171889062, "grad_norm": 0.025686580687761307, "learning_rate": 1.8557678202410074e-06, "loss": 0.0007, "step": 182080 }, { "epoch": 2.9794649431399822, "grad_norm": 0.01920386590063572, "learning_rate": 1.855212732841168e-06, "loss": 0.0002, "step": 182090 }, { "epoch": 2.9796285690910578, "grad_norm": 0.10406340658664703, "learning_rate": 1.8546577095608648e-06, "loss": 0.0005, "step": 182100 }, { "epoch": 2.9797921950421338, "grad_norm": 0.03668210282921791, "learning_rate": 1.8541027504114155e-06, "loss": 0.0006, "step": 182110 }, { "epoch": 2.9799558209932098, "grad_norm": 0.028323689475655556, "learning_rate": 1.853547855404136e-06, "loss": 0.0006, "step": 182120 }, { "epoch": 2.9801194469442853, "grad_norm": 0.04763267561793327, "learning_rate": 1.8529930245503408e-06, "loss": 0.0007, "step": 182130 }, { "epoch": 2.9802830728953613, "grad_norm": 0.006845399271696806, "learning_rate": 1.8524382578613404e-06, "loss": 0.0003, "step": 182140 }, { "epoch": 2.9804466988464373, "grad_norm": 0.061271313577890396, "learning_rate": 1.8518835553484464e-06, "loss": 0.0011, "step": 182150 }, { "epoch": 2.980610324797513, "grad_norm": 0.04293489083647728, "learning_rate": 1.8513289170229704e-06, "loss": 0.0007, "step": 182160 }, { "epoch": 2.980773950748589, "grad_norm": 0.022950122132897377, "learning_rate": 1.8507743428962204e-06, "loss": 0.0008, "step": 182170 }, { "epoch": 2.9809375766996644, "grad_norm": 0.035766758024692535, "learning_rate": 1.8502198329795024e-06, "loss": 0.0004, "step": 182180 }, { "epoch": 2.9811012026507404, "grad_norm": 0.013722353614866734, "learning_rate": 1.849665387284123e-06, "loss": 0.0008, "step": 182190 }, { "epoch": 2.9812648286018164, "grad_norm": 0.027213530614972115, "learning_rate": 1.8491110058213867e-06, "loss": 0.0007, "step": 182200 }, { "epoch": 2.981428454552892, "grad_norm": 0.07404805719852448, "learning_rate": 1.8485566886025991e-06, "loss": 0.0004, "step": 182210 }, { "epoch": 2.981592080503968, "grad_norm": 0.0903419628739357, "learning_rate": 1.8480024356390592e-06, "loss": 0.0003, "step": 182220 }, { "epoch": 2.9817557064550435, "grad_norm": 0.02505430579185486, "learning_rate": 1.8474482469420685e-06, "loss": 0.0006, "step": 182230 }, { "epoch": 2.9819193324061195, "grad_norm": 0.028737490996718407, "learning_rate": 1.846894122522927e-06, "loss": 0.0006, "step": 182240 }, { "epoch": 2.9820829583571955, "grad_norm": 0.037799738347530365, "learning_rate": 1.8463400623929344e-06, "loss": 0.0005, "step": 182250 }, { "epoch": 2.982246584308271, "grad_norm": 0.10432359576225281, "learning_rate": 1.8457860665633843e-06, "loss": 0.0009, "step": 182260 }, { "epoch": 2.982410210259347, "grad_norm": 0.06593824177980423, "learning_rate": 1.8452321350455733e-06, "loss": 0.0006, "step": 182270 }, { "epoch": 2.982573836210423, "grad_norm": 0.0026297287549823523, "learning_rate": 1.8446782678507962e-06, "loss": 0.0006, "step": 182280 }, { "epoch": 2.9827374621614986, "grad_norm": 0.03156590834259987, "learning_rate": 1.8441244649903478e-06, "loss": 0.0024, "step": 182290 }, { "epoch": 2.9829010881125746, "grad_norm": 0.03465491905808449, "learning_rate": 1.8435707264755153e-06, "loss": 0.001, "step": 182300 }, { "epoch": 2.9830647140636506, "grad_norm": 0.038033001124858856, "learning_rate": 1.8430170523175916e-06, "loss": 0.0004, "step": 182310 }, { "epoch": 2.983228340014726, "grad_norm": 0.08082054555416107, "learning_rate": 1.8424634425278653e-06, "loss": 0.0004, "step": 182320 }, { "epoch": 2.983391965965802, "grad_norm": 0.027515489608049393, "learning_rate": 1.8419098971176259e-06, "loss": 0.0006, "step": 182330 }, { "epoch": 2.983555591916878, "grad_norm": 0.07535496354103088, "learning_rate": 1.8413564160981562e-06, "loss": 0.0008, "step": 182340 }, { "epoch": 2.9837192178679537, "grad_norm": 0.07141576707363129, "learning_rate": 1.8408029994807424e-06, "loss": 0.0005, "step": 182350 }, { "epoch": 2.9838828438190297, "grad_norm": 0.025496847927570343, "learning_rate": 1.8402496472766685e-06, "loss": 0.0003, "step": 182360 }, { "epoch": 2.9840464697701057, "grad_norm": 0.043134093284606934, "learning_rate": 1.8396963594972195e-06, "loss": 0.0006, "step": 182370 }, { "epoch": 2.9842100957211812, "grad_norm": 0.05765671655535698, "learning_rate": 1.8391431361536716e-06, "loss": 0.0007, "step": 182380 }, { "epoch": 2.9843737216722572, "grad_norm": 0.18379591405391693, "learning_rate": 1.8385899772573068e-06, "loss": 0.0009, "step": 182390 }, { "epoch": 2.9845373476233332, "grad_norm": 0.02075238525867462, "learning_rate": 1.8380368828194044e-06, "loss": 0.0004, "step": 182400 }, { "epoch": 2.9847009735744088, "grad_norm": 0.042106565088033676, "learning_rate": 1.8374838528512417e-06, "loss": 0.0005, "step": 182410 }, { "epoch": 2.9848645995254848, "grad_norm": 0.06526210159063339, "learning_rate": 1.836930887364093e-06, "loss": 0.0007, "step": 182420 }, { "epoch": 2.9850282254765608, "grad_norm": 0.03638661280274391, "learning_rate": 1.8363779863692321e-06, "loss": 0.0007, "step": 182430 }, { "epoch": 2.9851918514276363, "grad_norm": 0.011637005023658276, "learning_rate": 1.8358251498779345e-06, "loss": 0.0005, "step": 182440 }, { "epoch": 2.9853554773787123, "grad_norm": 0.017523745074868202, "learning_rate": 1.8352723779014725e-06, "loss": 0.0007, "step": 182450 }, { "epoch": 2.9855191033297883, "grad_norm": 0.02354198880493641, "learning_rate": 1.8347196704511134e-06, "loss": 0.0035, "step": 182460 }, { "epoch": 2.985682729280864, "grad_norm": 0.10586342215538025, "learning_rate": 1.834167027538128e-06, "loss": 0.0011, "step": 182470 }, { "epoch": 2.98584635523194, "grad_norm": 0.0446423664689064, "learning_rate": 1.833614449173785e-06, "loss": 0.0003, "step": 182480 }, { "epoch": 2.986009981183016, "grad_norm": 0.05408425256609917, "learning_rate": 1.8330619353693518e-06, "loss": 0.001, "step": 182490 }, { "epoch": 2.9861736071340914, "grad_norm": 0.007862025871872902, "learning_rate": 1.832509486136091e-06, "loss": 0.0002, "step": 182500 }, { "epoch": 2.9863372330851674, "grad_norm": 0.06258926540613174, "learning_rate": 1.8319571014852676e-06, "loss": 0.0003, "step": 182510 }, { "epoch": 2.9865008590362434, "grad_norm": 0.007991787977516651, "learning_rate": 1.8314047814281443e-06, "loss": 0.0006, "step": 182520 }, { "epoch": 2.986664484987319, "grad_norm": 0.0017974867951124907, "learning_rate": 1.8308525259759852e-06, "loss": 0.0005, "step": 182530 }, { "epoch": 2.986828110938395, "grad_norm": 0.06378717720508575, "learning_rate": 1.8303003351400456e-06, "loss": 0.0006, "step": 182540 }, { "epoch": 2.9869917368894705, "grad_norm": 0.04024354740977287, "learning_rate": 1.8297482089315882e-06, "loss": 0.0008, "step": 182550 }, { "epoch": 2.9871553628405465, "grad_norm": 0.05151909589767456, "learning_rate": 1.8291961473618653e-06, "loss": 0.0017, "step": 182560 }, { "epoch": 2.9873189887916225, "grad_norm": 0.04798518121242523, "learning_rate": 1.8286441504421394e-06, "loss": 0.0011, "step": 182570 }, { "epoch": 2.987482614742698, "grad_norm": 0.02393922582268715, "learning_rate": 1.8280922181836607e-06, "loss": 0.0006, "step": 182580 }, { "epoch": 2.987646240693774, "grad_norm": 0.026064159348607063, "learning_rate": 1.8275403505976858e-06, "loss": 0.0009, "step": 182590 }, { "epoch": 2.9878098666448496, "grad_norm": 0.016376156359910965, "learning_rate": 1.8269885476954614e-06, "loss": 0.0005, "step": 182600 }, { "epoch": 2.9879734925959256, "grad_norm": 0.028503721579909325, "learning_rate": 1.8264368094882457e-06, "loss": 0.0006, "step": 182610 }, { "epoch": 2.9881371185470016, "grad_norm": 0.028549542650580406, "learning_rate": 1.8258851359872825e-06, "loss": 0.0008, "step": 182620 }, { "epoch": 2.988300744498077, "grad_norm": 0.03550053760409355, "learning_rate": 1.8253335272038235e-06, "loss": 0.0007, "step": 182630 }, { "epoch": 2.988464370449153, "grad_norm": 0.01177552156150341, "learning_rate": 1.8247819831491109e-06, "loss": 0.0009, "step": 182640 }, { "epoch": 2.988627996400229, "grad_norm": 0.08013855665922165, "learning_rate": 1.8242305038343965e-06, "loss": 0.0008, "step": 182650 }, { "epoch": 2.9887916223513047, "grad_norm": 0.04141882807016373, "learning_rate": 1.8236790892709193e-06, "loss": 0.0005, "step": 182660 }, { "epoch": 2.9889552483023807, "grad_norm": 0.017077570781111717, "learning_rate": 1.8231277394699266e-06, "loss": 0.0004, "step": 182670 }, { "epoch": 2.9891188742534567, "grad_norm": 0.2086656093597412, "learning_rate": 1.8225764544426533e-06, "loss": 0.0006, "step": 182680 }, { "epoch": 2.9892825002045322, "grad_norm": 0.20019742846488953, "learning_rate": 1.822025234200347e-06, "loss": 0.0009, "step": 182690 }, { "epoch": 2.9894461261556082, "grad_norm": 0.0048479135148227215, "learning_rate": 1.8214740787542422e-06, "loss": 0.0004, "step": 182700 }, { "epoch": 2.9896097521066842, "grad_norm": 0.023049864917993546, "learning_rate": 1.8209229881155793e-06, "loss": 0.0007, "step": 182710 }, { "epoch": 2.98977337805776, "grad_norm": 0.04342946410179138, "learning_rate": 1.8203719622955911e-06, "loss": 0.0009, "step": 182720 }, { "epoch": 2.9899370040088358, "grad_norm": 0.002564516384154558, "learning_rate": 1.8198210013055146e-06, "loss": 0.0004, "step": 182730 }, { "epoch": 2.9901006299599118, "grad_norm": 0.014100815169513226, "learning_rate": 1.8192701051565831e-06, "loss": 0.0006, "step": 182740 }, { "epoch": 2.9902642559109873, "grad_norm": 0.0044606770388782024, "learning_rate": 1.8187192738600306e-06, "loss": 0.0004, "step": 182750 }, { "epoch": 2.9904278818620633, "grad_norm": 0.009564783424139023, "learning_rate": 1.818168507427085e-06, "loss": 0.0005, "step": 182760 }, { "epoch": 2.9905915078131393, "grad_norm": 0.013441590592265129, "learning_rate": 1.8176178058689775e-06, "loss": 0.0003, "step": 182770 }, { "epoch": 2.990755133764215, "grad_norm": 0.13854949176311493, "learning_rate": 1.8170671691969366e-06, "loss": 0.001, "step": 182780 }, { "epoch": 2.990918759715291, "grad_norm": 0.04869641736149788, "learning_rate": 1.8165165974221905e-06, "loss": 0.0012, "step": 182790 }, { "epoch": 2.991082385666367, "grad_norm": 0.028743864968419075, "learning_rate": 1.8159660905559622e-06, "loss": 0.0004, "step": 182800 }, { "epoch": 2.9912460116174424, "grad_norm": 0.025375016033649445, "learning_rate": 1.815415648609477e-06, "loss": 0.0004, "step": 182810 }, { "epoch": 2.9914096375685184, "grad_norm": 0.02787848189473152, "learning_rate": 1.8148652715939585e-06, "loss": 0.0009, "step": 182820 }, { "epoch": 2.9915732635195944, "grad_norm": 0.03030681423842907, "learning_rate": 1.81431495952063e-06, "loss": 0.001, "step": 182830 }, { "epoch": 2.99173688947067, "grad_norm": 0.09124954044818878, "learning_rate": 1.8137647124007084e-06, "loss": 0.0004, "step": 182840 }, { "epoch": 2.991900515421746, "grad_norm": 0.01960175484418869, "learning_rate": 1.8132145302454147e-06, "loss": 0.0006, "step": 182850 }, { "epoch": 2.992064141372822, "grad_norm": 0.03752153739333153, "learning_rate": 1.8126644130659659e-06, "loss": 0.0004, "step": 182860 }, { "epoch": 2.9922277673238975, "grad_norm": 0.010926112532615662, "learning_rate": 1.8121143608735815e-06, "loss": 0.0003, "step": 182870 }, { "epoch": 2.9923913932749735, "grad_norm": 0.03774826228618622, "learning_rate": 1.8115643736794714e-06, "loss": 0.0006, "step": 182880 }, { "epoch": 2.9925550192260495, "grad_norm": 0.01470978558063507, "learning_rate": 1.811014451494853e-06, "loss": 0.0008, "step": 182890 }, { "epoch": 2.992718645177125, "grad_norm": 0.0025946630630642176, "learning_rate": 1.8104645943309374e-06, "loss": 0.0001, "step": 182900 }, { "epoch": 2.992882271128201, "grad_norm": 0.021318161860108376, "learning_rate": 1.8099148021989377e-06, "loss": 0.0003, "step": 182910 }, { "epoch": 2.993045897079277, "grad_norm": 0.016292212530970573, "learning_rate": 1.8093650751100605e-06, "loss": 0.0004, "step": 182920 }, { "epoch": 2.9932095230303526, "grad_norm": 0.20829997956752777, "learning_rate": 1.8088154130755158e-06, "loss": 0.0005, "step": 182930 }, { "epoch": 2.9933731489814286, "grad_norm": 0.005621627904474735, "learning_rate": 1.8082658161065108e-06, "loss": 0.0001, "step": 182940 }, { "epoch": 2.993536774932504, "grad_norm": 0.033861033618450165, "learning_rate": 1.8077162842142532e-06, "loss": 0.0003, "step": 182950 }, { "epoch": 2.99370040088358, "grad_norm": 0.052269674837589264, "learning_rate": 1.8071668174099439e-06, "loss": 0.0008, "step": 182960 }, { "epoch": 2.993864026834656, "grad_norm": 0.07638556510210037, "learning_rate": 1.8066174157047877e-06, "loss": 0.001, "step": 182970 }, { "epoch": 2.9940276527857317, "grad_norm": 0.01972554437816143, "learning_rate": 1.8060680791099867e-06, "loss": 0.0006, "step": 182980 }, { "epoch": 2.9941912787368077, "grad_norm": 0.11216797679662704, "learning_rate": 1.805518807636743e-06, "loss": 0.0006, "step": 182990 }, { "epoch": 2.9943549046878832, "grad_norm": 0.022289907559752464, "learning_rate": 1.804969601296252e-06, "loss": 0.0017, "step": 183000 }, { "epoch": 2.9943549046878832, "eval_loss": 0.0007746397168375552, "eval_runtime": 5.3353, "eval_samples_per_second": 37.486, "eval_steps_per_second": 9.372, "step": 183000 }, { "epoch": 2.9945185306389592, "grad_norm": 0.08634079247713089, "learning_rate": 1.804420460099714e-06, "loss": 0.0007, "step": 183010 }, { "epoch": 2.9946821565900352, "grad_norm": 0.09782540798187256, "learning_rate": 1.8038713840583243e-06, "loss": 0.0012, "step": 183020 }, { "epoch": 2.994845782541111, "grad_norm": 0.02598346583545208, "learning_rate": 1.803322373183281e-06, "loss": 0.0005, "step": 183030 }, { "epoch": 2.9950094084921868, "grad_norm": 0.03484237566590309, "learning_rate": 1.8027734274857744e-06, "loss": 0.0004, "step": 183040 }, { "epoch": 2.9951730344432628, "grad_norm": 0.004998216405510902, "learning_rate": 1.8022245469769983e-06, "loss": 0.0004, "step": 183050 }, { "epoch": 2.9953366603943383, "grad_norm": 0.02461124025285244, "learning_rate": 1.8016757316681438e-06, "loss": 0.0004, "step": 183060 }, { "epoch": 2.9955002863454143, "grad_norm": 0.039181530475616455, "learning_rate": 1.801126981570403e-06, "loss": 0.0007, "step": 183070 }, { "epoch": 2.9956639122964903, "grad_norm": 0.11799510568380356, "learning_rate": 1.8005782966949609e-06, "loss": 0.0006, "step": 183080 }, { "epoch": 2.995827538247566, "grad_norm": 0.02570088766515255, "learning_rate": 1.8000296770530062e-06, "loss": 0.0009, "step": 183090 }, { "epoch": 2.995991164198642, "grad_norm": 0.05039988458156586, "learning_rate": 1.7994811226557246e-06, "loss": 0.0004, "step": 183100 }, { "epoch": 2.996154790149718, "grad_norm": 0.0058308979496359825, "learning_rate": 1.798932633514303e-06, "loss": 0.0003, "step": 183110 }, { "epoch": 2.9963184161007934, "grad_norm": 0.0604340061545372, "learning_rate": 1.7983842096399211e-06, "loss": 0.0006, "step": 183120 }, { "epoch": 2.9964820420518694, "grad_norm": 0.09524007141590118, "learning_rate": 1.797835851043762e-06, "loss": 0.0006, "step": 183130 }, { "epoch": 2.9966456680029454, "grad_norm": 0.022593343630433083, "learning_rate": 1.7972875577370063e-06, "loss": 0.0006, "step": 183140 }, { "epoch": 2.996809293954021, "grad_norm": 0.04243558645248413, "learning_rate": 1.7967393297308356e-06, "loss": 0.0005, "step": 183150 }, { "epoch": 2.996972919905097, "grad_norm": 0.07898478955030441, "learning_rate": 1.796191167036424e-06, "loss": 0.0009, "step": 183160 }, { "epoch": 2.997136545856173, "grad_norm": 0.012574645690619946, "learning_rate": 1.7956430696649495e-06, "loss": 0.0005, "step": 183170 }, { "epoch": 2.9973001718072485, "grad_norm": 0.05378221347928047, "learning_rate": 1.7950950376275878e-06, "loss": 0.0002, "step": 183180 }, { "epoch": 2.9974637977583245, "grad_norm": 0.03758442401885986, "learning_rate": 1.794547070935514e-06, "loss": 0.0005, "step": 183190 }, { "epoch": 2.9976274237094005, "grad_norm": 0.030070148408412933, "learning_rate": 1.7939991695998981e-06, "loss": 0.001, "step": 183200 }, { "epoch": 2.997791049660476, "grad_norm": 0.07242496311664581, "learning_rate": 1.7934513336319137e-06, "loss": 0.0008, "step": 183210 }, { "epoch": 2.997954675611552, "grad_norm": 0.029445970430970192, "learning_rate": 1.7929035630427266e-06, "loss": 0.0012, "step": 183220 }, { "epoch": 2.998118301562628, "grad_norm": 0.06858508288860321, "learning_rate": 1.7923558578435112e-06, "loss": 0.0007, "step": 183230 }, { "epoch": 2.9982819275137036, "grad_norm": 0.0033391444012522697, "learning_rate": 1.7918082180454305e-06, "loss": 0.0006, "step": 183240 }, { "epoch": 2.9984455534647796, "grad_norm": 0.026658765971660614, "learning_rate": 1.7912606436596536e-06, "loss": 0.0006, "step": 183250 }, { "epoch": 2.9986091794158556, "grad_norm": 0.017218464985489845, "learning_rate": 1.7907131346973394e-06, "loss": 0.0005, "step": 183260 }, { "epoch": 2.998772805366931, "grad_norm": 0.027821211144328117, "learning_rate": 1.7901656911696592e-06, "loss": 0.0007, "step": 183270 }, { "epoch": 2.998936431318007, "grad_norm": 0.07287786155939102, "learning_rate": 1.789618313087768e-06, "loss": 0.0005, "step": 183280 }, { "epoch": 2.999100057269083, "grad_norm": 0.02085571177303791, "learning_rate": 1.7890710004628315e-06, "loss": 0.0007, "step": 183290 }, { "epoch": 2.9992636832201587, "grad_norm": 0.009938763454556465, "learning_rate": 1.7885237533060023e-06, "loss": 0.0006, "step": 183300 }, { "epoch": 2.9994273091712347, "grad_norm": 0.004234275780618191, "learning_rate": 1.7879765716284469e-06, "loss": 0.0003, "step": 183310 }, { "epoch": 2.9995909351223107, "grad_norm": 0.023842109367251396, "learning_rate": 1.7874294554413152e-06, "loss": 0.0006, "step": 183320 }, { "epoch": 2.9997545610733862, "grad_norm": 0.04054657742381096, "learning_rate": 1.7868824047557665e-06, "loss": 0.0005, "step": 183330 }, { "epoch": 2.9999181870244622, "grad_norm": 0.0352015346288681, "learning_rate": 1.7863354195829497e-06, "loss": 0.0004, "step": 183340 }, { "epoch": 3.0000818129755378, "grad_norm": 0.05609165132045746, "learning_rate": 1.7857884999340237e-06, "loss": 0.0008, "step": 183350 }, { "epoch": 3.0002454389266138, "grad_norm": 0.002111552283167839, "learning_rate": 1.7852416458201348e-06, "loss": 0.0002, "step": 183360 }, { "epoch": 3.0004090648776898, "grad_norm": 0.02575002983212471, "learning_rate": 1.7846948572524365e-06, "loss": 0.0006, "step": 183370 }, { "epoch": 3.0005726908287653, "grad_norm": 0.007146704476326704, "learning_rate": 1.784148134242074e-06, "loss": 0.0004, "step": 183380 }, { "epoch": 3.0007363167798413, "grad_norm": 0.003246007254347205, "learning_rate": 1.7836014768001959e-06, "loss": 0.0005, "step": 183390 }, { "epoch": 3.0008999427309173, "grad_norm": 0.027390189468860626, "learning_rate": 1.7830548849379481e-06, "loss": 0.0008, "step": 183400 }, { "epoch": 3.001063568681993, "grad_norm": 0.0009983307681977749, "learning_rate": 1.782508358666477e-06, "loss": 0.0006, "step": 183410 }, { "epoch": 3.001227194633069, "grad_norm": 0.021105360239744186, "learning_rate": 1.7819618979969228e-06, "loss": 0.0004, "step": 183420 }, { "epoch": 3.001390820584145, "grad_norm": 0.08653547614812851, "learning_rate": 1.7814155029404284e-06, "loss": 0.0009, "step": 183430 }, { "epoch": 3.0015544465352204, "grad_norm": 0.032043732702732086, "learning_rate": 1.7808691735081347e-06, "loss": 0.0003, "step": 183440 }, { "epoch": 3.0017180724862964, "grad_norm": 0.03974844142794609, "learning_rate": 1.7803229097111823e-06, "loss": 0.0005, "step": 183450 }, { "epoch": 3.001881698437372, "grad_norm": 0.009495556354522705, "learning_rate": 1.7797767115607062e-06, "loss": 0.0005, "step": 183460 }, { "epoch": 3.002045324388448, "grad_norm": 0.0871616080403328, "learning_rate": 1.7792305790678439e-06, "loss": 0.0004, "step": 183470 }, { "epoch": 3.002208950339524, "grad_norm": 0.03561103716492653, "learning_rate": 1.7786845122437307e-06, "loss": 0.0003, "step": 183480 }, { "epoch": 3.0023725762905995, "grad_norm": 0.016056902706623077, "learning_rate": 1.778138511099503e-06, "loss": 0.0002, "step": 183490 }, { "epoch": 3.0025362022416755, "grad_norm": 0.0341474674642086, "learning_rate": 1.7775925756462887e-06, "loss": 0.0005, "step": 183500 }, { "epoch": 3.0026998281927515, "grad_norm": 0.0029473977629095316, "learning_rate": 1.7770467058952217e-06, "loss": 0.0009, "step": 183510 }, { "epoch": 3.002863454143827, "grad_norm": 0.01679789274930954, "learning_rate": 1.7765009018574307e-06, "loss": 0.0004, "step": 183520 }, { "epoch": 3.003027080094903, "grad_norm": 0.005965977441519499, "learning_rate": 1.775955163544047e-06, "loss": 0.0003, "step": 183530 }, { "epoch": 3.003190706045979, "grad_norm": 0.036601290106773376, "learning_rate": 1.775409490966194e-06, "loss": 0.0002, "step": 183540 }, { "epoch": 3.0033543319970546, "grad_norm": 0.006696636788547039, "learning_rate": 1.7748638841349986e-06, "loss": 0.0004, "step": 183550 }, { "epoch": 3.0035179579481306, "grad_norm": 0.020934676751494408, "learning_rate": 1.7743183430615863e-06, "loss": 0.0004, "step": 183560 }, { "epoch": 3.0036815838992066, "grad_norm": 0.012969217263162136, "learning_rate": 1.7737728677570808e-06, "loss": 0.0006, "step": 183570 }, { "epoch": 3.003845209850282, "grad_norm": 0.06198042631149292, "learning_rate": 1.7732274582326015e-06, "loss": 0.0001, "step": 183580 }, { "epoch": 3.004008835801358, "grad_norm": 0.008639436215162277, "learning_rate": 1.7726821144992695e-06, "loss": 0.0004, "step": 183590 }, { "epoch": 3.004172461752434, "grad_norm": 0.0013093852903693914, "learning_rate": 1.7721368365682045e-06, "loss": 0.0004, "step": 183600 }, { "epoch": 3.0043360877035097, "grad_norm": 0.0018895046086981893, "learning_rate": 1.7715916244505267e-06, "loss": 0.0006, "step": 183610 }, { "epoch": 3.0044997136545857, "grad_norm": 0.0009994633728638291, "learning_rate": 1.7710464781573473e-06, "loss": 0.0002, "step": 183620 }, { "epoch": 3.0046633396056617, "grad_norm": 0.0034574992023408413, "learning_rate": 1.7705013976997849e-06, "loss": 0.0002, "step": 183630 }, { "epoch": 3.0048269655567372, "grad_norm": 0.08746056258678436, "learning_rate": 1.7699563830889517e-06, "loss": 0.0006, "step": 183640 }, { "epoch": 3.0049905915078132, "grad_norm": 0.0005416101193986833, "learning_rate": 1.769411434335963e-06, "loss": 0.0003, "step": 183650 }, { "epoch": 3.0051542174588888, "grad_norm": 0.030520008876919746, "learning_rate": 1.7688665514519253e-06, "loss": 0.0004, "step": 183660 }, { "epoch": 3.0053178434099648, "grad_norm": 0.014801633544266224, "learning_rate": 1.7683217344479514e-06, "loss": 0.0003, "step": 183670 }, { "epoch": 3.0054814693610408, "grad_norm": 0.02202477678656578, "learning_rate": 1.7677769833351488e-06, "loss": 0.0004, "step": 183680 }, { "epoch": 3.0056450953121163, "grad_norm": 0.012963635846972466, "learning_rate": 1.767232298124626e-06, "loss": 0.0002, "step": 183690 }, { "epoch": 3.0058087212631923, "grad_norm": 0.09271533787250519, "learning_rate": 1.7666876788274857e-06, "loss": 0.0002, "step": 183700 }, { "epoch": 3.0059723472142683, "grad_norm": 0.03568367660045624, "learning_rate": 1.766143125454834e-06, "loss": 0.0002, "step": 183710 }, { "epoch": 3.006135973165344, "grad_norm": 0.020219214260578156, "learning_rate": 1.765598638017773e-06, "loss": 0.0004, "step": 183720 }, { "epoch": 3.00629959911642, "grad_norm": 0.020803917199373245, "learning_rate": 1.765054216527407e-06, "loss": 0.0003, "step": 183730 }, { "epoch": 3.006463225067496, "grad_norm": 0.0022268551401793957, "learning_rate": 1.764509860994833e-06, "loss": 0.0003, "step": 183740 }, { "epoch": 3.0066268510185714, "grad_norm": 0.0610131211578846, "learning_rate": 1.763965571431151e-06, "loss": 0.0008, "step": 183750 }, { "epoch": 3.0067904769696474, "grad_norm": 0.006356098689138889, "learning_rate": 1.7634213478474588e-06, "loss": 0.0006, "step": 183760 }, { "epoch": 3.0069541029207234, "grad_norm": 0.06523042917251587, "learning_rate": 1.7628771902548548e-06, "loss": 0.0005, "step": 183770 }, { "epoch": 3.007117728871799, "grad_norm": 0.036096494644880295, "learning_rate": 1.7623330986644294e-06, "loss": 0.0003, "step": 183780 }, { "epoch": 3.007281354822875, "grad_norm": 0.00040302288834936917, "learning_rate": 1.7617890730872788e-06, "loss": 0.0002, "step": 183790 }, { "epoch": 3.007444980773951, "grad_norm": 0.008100268431007862, "learning_rate": 1.7612451135344954e-06, "loss": 0.0002, "step": 183800 }, { "epoch": 3.0076086067250265, "grad_norm": 0.04758862033486366, "learning_rate": 1.7607012200171714e-06, "loss": 0.0005, "step": 183810 }, { "epoch": 3.0077722326761025, "grad_norm": 0.025775600224733353, "learning_rate": 1.760157392546392e-06, "loss": 0.0002, "step": 183820 }, { "epoch": 3.007935858627178, "grad_norm": 0.06004369258880615, "learning_rate": 1.7596136311332506e-06, "loss": 0.0007, "step": 183830 }, { "epoch": 3.008099484578254, "grad_norm": 0.04240216314792633, "learning_rate": 1.759069935788828e-06, "loss": 0.0003, "step": 183840 }, { "epoch": 3.00826311052933, "grad_norm": 0.007252768147736788, "learning_rate": 1.7585263065242159e-06, "loss": 0.0002, "step": 183850 }, { "epoch": 3.0084267364804056, "grad_norm": 0.016708437353372574, "learning_rate": 1.7579827433504943e-06, "loss": 0.0003, "step": 183860 }, { "epoch": 3.0085903624314816, "grad_norm": 0.0450969822704792, "learning_rate": 1.7574392462787494e-06, "loss": 0.0005, "step": 183870 }, { "epoch": 3.0087539883825576, "grad_norm": 0.03176796808838844, "learning_rate": 1.7568958153200565e-06, "loss": 0.0002, "step": 183880 }, { "epoch": 3.008917614333633, "grad_norm": 0.0007243637228384614, "learning_rate": 1.7563524504855033e-06, "loss": 0.0004, "step": 183890 }, { "epoch": 3.009081240284709, "grad_norm": 0.018932079896330833, "learning_rate": 1.7558091517861637e-06, "loss": 0.0004, "step": 183900 }, { "epoch": 3.009244866235785, "grad_norm": 0.044122859835624695, "learning_rate": 1.7552659192331177e-06, "loss": 0.0004, "step": 183910 }, { "epoch": 3.0094084921868607, "grad_norm": 0.02151617407798767, "learning_rate": 1.7547227528374365e-06, "loss": 0.0004, "step": 183920 }, { "epoch": 3.0095721181379367, "grad_norm": 0.04780061915516853, "learning_rate": 1.7541796526102023e-06, "loss": 0.0003, "step": 183930 }, { "epoch": 3.0097357440890127, "grad_norm": 0.031723883002996445, "learning_rate": 1.7536366185624825e-06, "loss": 0.0003, "step": 183940 }, { "epoch": 3.0098993700400882, "grad_norm": 0.004315203055739403, "learning_rate": 1.7530936507053532e-06, "loss": 0.0003, "step": 183950 }, { "epoch": 3.0100629959911642, "grad_norm": 0.018936360254883766, "learning_rate": 1.752550749049879e-06, "loss": 0.0005, "step": 183960 }, { "epoch": 3.01022662194224, "grad_norm": 0.02191627211868763, "learning_rate": 1.7520079136071366e-06, "loss": 0.0002, "step": 183970 }, { "epoch": 3.0103902478933158, "grad_norm": 0.0029439148493111134, "learning_rate": 1.7514651443881891e-06, "loss": 0.0003, "step": 183980 }, { "epoch": 3.0105538738443918, "grad_norm": 0.047564711421728134, "learning_rate": 1.7509224414041065e-06, "loss": 0.0002, "step": 183990 }, { "epoch": 3.0107174997954678, "grad_norm": 0.011416831985116005, "learning_rate": 1.7503798046659481e-06, "loss": 0.0002, "step": 184000 }, { "epoch": 3.0108811257465433, "grad_norm": 0.012188400141894817, "learning_rate": 1.7498372341847852e-06, "loss": 0.0004, "step": 184010 }, { "epoch": 3.0110447516976193, "grad_norm": 0.015989789739251137, "learning_rate": 1.7492947299716755e-06, "loss": 0.0001, "step": 184020 }, { "epoch": 3.011208377648695, "grad_norm": 0.05138515308499336, "learning_rate": 1.7487522920376837e-06, "loss": 0.0001, "step": 184030 }, { "epoch": 3.011372003599771, "grad_norm": 0.019808774814009666, "learning_rate": 1.748209920393865e-06, "loss": 0.0002, "step": 184040 }, { "epoch": 3.011535629550847, "grad_norm": 0.02331681363284588, "learning_rate": 1.747667615051281e-06, "loss": 0.0002, "step": 184050 }, { "epoch": 3.0116992555019224, "grad_norm": 0.1076192557811737, "learning_rate": 1.7471253760209883e-06, "loss": 0.0003, "step": 184060 }, { "epoch": 3.0118628814529984, "grad_norm": 0.042229972779750824, "learning_rate": 1.7465832033140445e-06, "loss": 0.0004, "step": 184070 }, { "epoch": 3.0120265074040744, "grad_norm": 0.043910641223192215, "learning_rate": 1.7460410969415003e-06, "loss": 0.0003, "step": 184080 }, { "epoch": 3.01219013335515, "grad_norm": 0.08210381120443344, "learning_rate": 1.7454990569144108e-06, "loss": 0.0002, "step": 184090 }, { "epoch": 3.012353759306226, "grad_norm": 0.01603592373430729, "learning_rate": 1.7449570832438278e-06, "loss": 0.0003, "step": 184100 }, { "epoch": 3.012517385257302, "grad_norm": 0.00360855576582253, "learning_rate": 1.7444151759408034e-06, "loss": 0.0005, "step": 184110 }, { "epoch": 3.0126810112083775, "grad_norm": 0.004137253854423761, "learning_rate": 1.7438733350163828e-06, "loss": 0.0002, "step": 184120 }, { "epoch": 3.0128446371594535, "grad_norm": 0.03325393423438072, "learning_rate": 1.743331560481616e-06, "loss": 0.0002, "step": 184130 }, { "epoch": 3.0130082631105295, "grad_norm": 0.007705851458013058, "learning_rate": 1.7427898523475483e-06, "loss": 0.0002, "step": 184140 }, { "epoch": 3.013171889061605, "grad_norm": 0.038818564265966415, "learning_rate": 1.7422482106252276e-06, "loss": 0.0004, "step": 184150 }, { "epoch": 3.013335515012681, "grad_norm": 0.005837416276335716, "learning_rate": 1.7417066353256933e-06, "loss": 0.0002, "step": 184160 }, { "epoch": 3.013499140963757, "grad_norm": 0.031233197078108788, "learning_rate": 1.74116512645999e-06, "loss": 0.0001, "step": 184170 }, { "epoch": 3.0136627669148326, "grad_norm": 0.005647206213325262, "learning_rate": 1.7406236840391578e-06, "loss": 0.0004, "step": 184180 }, { "epoch": 3.0138263928659086, "grad_norm": 0.09555128961801529, "learning_rate": 1.7400823080742385e-06, "loss": 0.0003, "step": 184190 }, { "epoch": 3.0139900188169846, "grad_norm": 0.026790549978613853, "learning_rate": 1.7395409985762673e-06, "loss": 0.0004, "step": 184200 }, { "epoch": 3.01415364476806, "grad_norm": 0.012510547414422035, "learning_rate": 1.7389997555562815e-06, "loss": 0.0002, "step": 184210 }, { "epoch": 3.014317270719136, "grad_norm": 0.04977480322122574, "learning_rate": 1.7384585790253178e-06, "loss": 0.0004, "step": 184220 }, { "epoch": 3.0144808966702117, "grad_norm": 0.021893354132771492, "learning_rate": 1.7379174689944111e-06, "loss": 0.0001, "step": 184230 }, { "epoch": 3.0146445226212877, "grad_norm": 0.02389637753367424, "learning_rate": 1.7373764254745917e-06, "loss": 0.0002, "step": 184240 }, { "epoch": 3.0148081485723637, "grad_norm": 0.031070278957486153, "learning_rate": 1.7368354484768917e-06, "loss": 0.0002, "step": 184250 }, { "epoch": 3.0149717745234392, "grad_norm": 0.05810923874378204, "learning_rate": 1.736294538012342e-06, "loss": 0.0001, "step": 184260 }, { "epoch": 3.0151354004745152, "grad_norm": 0.0006147499079816043, "learning_rate": 1.7357536940919729e-06, "loss": 0.0002, "step": 184270 }, { "epoch": 3.0152990264255912, "grad_norm": 0.07881589978933334, "learning_rate": 1.7352129167268078e-06, "loss": 0.0002, "step": 184280 }, { "epoch": 3.0154626523766668, "grad_norm": 0.001716442871838808, "learning_rate": 1.7346722059278748e-06, "loss": 0.0002, "step": 184290 }, { "epoch": 3.0156262783277428, "grad_norm": 0.0804416611790657, "learning_rate": 1.7341315617061981e-06, "loss": 0.0003, "step": 184300 }, { "epoch": 3.0157899042788188, "grad_norm": 0.003919156268239021, "learning_rate": 1.7335909840728032e-06, "loss": 0.0002, "step": 184310 }, { "epoch": 3.0159535302298943, "grad_norm": 0.04791659116744995, "learning_rate": 1.7330504730387088e-06, "loss": 0.0002, "step": 184320 }, { "epoch": 3.0161171561809703, "grad_norm": 0.005126339849084616, "learning_rate": 1.7325100286149365e-06, "loss": 0.0003, "step": 184330 }, { "epoch": 3.0162807821320463, "grad_norm": 0.04811473935842514, "learning_rate": 1.731969650812506e-06, "loss": 0.0002, "step": 184340 }, { "epoch": 3.016444408083122, "grad_norm": 0.03430076315999031, "learning_rate": 1.7314293396424369e-06, "loss": 0.0004, "step": 184350 }, { "epoch": 3.016608034034198, "grad_norm": 0.0009875672403723001, "learning_rate": 1.7308890951157415e-06, "loss": 0.0005, "step": 184360 }, { "epoch": 3.016771659985274, "grad_norm": 0.007543600630015135, "learning_rate": 1.7303489172434373e-06, "loss": 0.0005, "step": 184370 }, { "epoch": 3.0169352859363494, "grad_norm": 0.011267089284956455, "learning_rate": 1.7298088060365382e-06, "loss": 0.0002, "step": 184380 }, { "epoch": 3.0170989118874254, "grad_norm": 0.14380168914794922, "learning_rate": 1.7292687615060583e-06, "loss": 0.0006, "step": 184390 }, { "epoch": 3.0172625378385014, "grad_norm": 0.04555386304855347, "learning_rate": 1.7287287836630046e-06, "loss": 0.0006, "step": 184400 }, { "epoch": 3.017426163789577, "grad_norm": 0.006289375480264425, "learning_rate": 1.7281888725183887e-06, "loss": 0.0003, "step": 184410 }, { "epoch": 3.017589789740653, "grad_norm": 0.0431644581258297, "learning_rate": 1.7276490280832192e-06, "loss": 0.0004, "step": 184420 }, { "epoch": 3.0177534156917285, "grad_norm": 0.007949981838464737, "learning_rate": 1.727109250368505e-06, "loss": 0.0002, "step": 184430 }, { "epoch": 3.0179170416428045, "grad_norm": 0.0026828020345419645, "learning_rate": 1.7265695393852478e-06, "loss": 0.0003, "step": 184440 }, { "epoch": 3.0180806675938805, "grad_norm": 0.03175204247236252, "learning_rate": 1.7260298951444538e-06, "loss": 0.0003, "step": 184450 }, { "epoch": 3.018244293544956, "grad_norm": 0.00039560857112519443, "learning_rate": 1.7254903176571258e-06, "loss": 0.0003, "step": 184460 }, { "epoch": 3.018407919496032, "grad_norm": 0.010108760558068752, "learning_rate": 1.7249508069342668e-06, "loss": 0.0003, "step": 184470 }, { "epoch": 3.018571545447108, "grad_norm": 0.004036684520542622, "learning_rate": 1.7244113629868742e-06, "loss": 0.0004, "step": 184480 }, { "epoch": 3.0187351713981836, "grad_norm": 0.05646049603819847, "learning_rate": 1.7238719858259495e-06, "loss": 0.0005, "step": 184490 }, { "epoch": 3.0188987973492596, "grad_norm": 0.013112888671457767, "learning_rate": 1.7233326754624852e-06, "loss": 0.0005, "step": 184500 }, { "epoch": 3.0190624233003356, "grad_norm": 0.12314802408218384, "learning_rate": 1.7227934319074847e-06, "loss": 0.0003, "step": 184510 }, { "epoch": 3.019226049251411, "grad_norm": 0.06145815551280975, "learning_rate": 1.7222542551719373e-06, "loss": 0.0003, "step": 184520 }, { "epoch": 3.019389675202487, "grad_norm": 0.005487004294991493, "learning_rate": 1.7217151452668397e-06, "loss": 0.0003, "step": 184530 }, { "epoch": 3.019553301153563, "grad_norm": 0.037046901881694794, "learning_rate": 1.7211761022031787e-06, "loss": 0.0004, "step": 184540 }, { "epoch": 3.0197169271046387, "grad_norm": 0.01940469816327095, "learning_rate": 1.7206371259919518e-06, "loss": 0.0005, "step": 184550 }, { "epoch": 3.0198805530557147, "grad_norm": 0.020228374749422073, "learning_rate": 1.7200982166441433e-06, "loss": 0.0004, "step": 184560 }, { "epoch": 3.0200441790067907, "grad_norm": 0.030596353113651276, "learning_rate": 1.7195593741707445e-06, "loss": 0.0002, "step": 184570 }, { "epoch": 3.0202078049578662, "grad_norm": 0.020196104422211647, "learning_rate": 1.7190205985827364e-06, "loss": 0.0004, "step": 184580 }, { "epoch": 3.0203714309089422, "grad_norm": 0.02096528187394142, "learning_rate": 1.7184818898911115e-06, "loss": 0.0002, "step": 184590 }, { "epoch": 3.0205350568600178, "grad_norm": 0.001044033095240593, "learning_rate": 1.7179432481068486e-06, "loss": 0.0006, "step": 184600 }, { "epoch": 3.0206986828110938, "grad_norm": 0.02504463866353035, "learning_rate": 1.7174046732409328e-06, "loss": 0.0004, "step": 184610 }, { "epoch": 3.0208623087621698, "grad_norm": 0.049080293625593185, "learning_rate": 1.716866165304341e-06, "loss": 0.0002, "step": 184620 }, { "epoch": 3.0210259347132453, "grad_norm": 0.031119074672460556, "learning_rate": 1.7163277243080589e-06, "loss": 0.0004, "step": 184630 }, { "epoch": 3.0211895606643213, "grad_norm": 0.010858382098376751, "learning_rate": 1.7157893502630608e-06, "loss": 0.0003, "step": 184640 }, { "epoch": 3.0213531866153973, "grad_norm": 0.015271593816578388, "learning_rate": 1.7152510431803254e-06, "loss": 0.0001, "step": 184650 }, { "epoch": 3.021516812566473, "grad_norm": 0.05822769179940224, "learning_rate": 1.7147128030708266e-06, "loss": 0.0005, "step": 184660 }, { "epoch": 3.021680438517549, "grad_norm": 0.03984641656279564, "learning_rate": 1.7141746299455397e-06, "loss": 0.0004, "step": 184670 }, { "epoch": 3.021844064468625, "grad_norm": 0.022294776514172554, "learning_rate": 1.7136365238154378e-06, "loss": 0.0003, "step": 184680 }, { "epoch": 3.0220076904197004, "grad_norm": 0.0011800421634688973, "learning_rate": 1.7130984846914933e-06, "loss": 0.0002, "step": 184690 }, { "epoch": 3.0221713163707764, "grad_norm": 0.08149606734514236, "learning_rate": 1.7125605125846738e-06, "loss": 0.0005, "step": 184700 }, { "epoch": 3.0223349423218524, "grad_norm": 0.027349062263965607, "learning_rate": 1.7120226075059498e-06, "loss": 0.0002, "step": 184710 }, { "epoch": 3.022498568272928, "grad_norm": 0.07845977693796158, "learning_rate": 1.7114847694662888e-06, "loss": 0.0008, "step": 184720 }, { "epoch": 3.022662194224004, "grad_norm": 0.018661340698599815, "learning_rate": 1.710946998476658e-06, "loss": 0.0001, "step": 184730 }, { "epoch": 3.02282582017508, "grad_norm": 0.05864453315734863, "learning_rate": 1.7104092945480189e-06, "loss": 0.0004, "step": 184740 }, { "epoch": 3.0229894461261555, "grad_norm": 0.007402791175991297, "learning_rate": 1.7098716576913366e-06, "loss": 0.0003, "step": 184750 }, { "epoch": 3.0231530720772315, "grad_norm": 0.008365877903997898, "learning_rate": 1.709334087917573e-06, "loss": 0.0005, "step": 184760 }, { "epoch": 3.0233166980283075, "grad_norm": 0.018408246338367462, "learning_rate": 1.7087965852376904e-06, "loss": 0.0004, "step": 184770 }, { "epoch": 3.023480323979383, "grad_norm": 0.013060673139989376, "learning_rate": 1.7082591496626444e-06, "loss": 0.0003, "step": 184780 }, { "epoch": 3.023643949930459, "grad_norm": 0.01021944172680378, "learning_rate": 1.7077217812033954e-06, "loss": 0.0002, "step": 184790 }, { "epoch": 3.0238075758815346, "grad_norm": 0.06562232971191406, "learning_rate": 1.7071844798708986e-06, "loss": 0.0004, "step": 184800 }, { "epoch": 3.0239712018326106, "grad_norm": 0.025467127561569214, "learning_rate": 1.706647245676112e-06, "loss": 0.0004, "step": 184810 }, { "epoch": 3.0241348277836866, "grad_norm": 0.010438840836286545, "learning_rate": 1.7061100786299856e-06, "loss": 0.0003, "step": 184820 }, { "epoch": 3.024298453734762, "grad_norm": 0.10846948623657227, "learning_rate": 1.7055729787434726e-06, "loss": 0.0004, "step": 184830 }, { "epoch": 3.024462079685838, "grad_norm": 0.03734609857201576, "learning_rate": 1.7050359460275257e-06, "loss": 0.0003, "step": 184840 }, { "epoch": 3.024625705636914, "grad_norm": 0.036341629922389984, "learning_rate": 1.704498980493095e-06, "loss": 0.0003, "step": 184850 }, { "epoch": 3.0247893315879897, "grad_norm": 0.029879160225391388, "learning_rate": 1.7039620821511255e-06, "loss": 0.0003, "step": 184860 }, { "epoch": 3.0249529575390657, "grad_norm": 0.08538937568664551, "learning_rate": 1.703425251012566e-06, "loss": 0.0004, "step": 184870 }, { "epoch": 3.0251165834901417, "grad_norm": 0.019988304004073143, "learning_rate": 1.7028884870883617e-06, "loss": 0.0004, "step": 184880 }, { "epoch": 3.0252802094412172, "grad_norm": 0.025261783972382545, "learning_rate": 1.7023517903894592e-06, "loss": 0.0002, "step": 184890 }, { "epoch": 3.0254438353922932, "grad_norm": 0.012234778143465519, "learning_rate": 1.7018151609267975e-06, "loss": 0.0003, "step": 184900 }, { "epoch": 3.025607461343369, "grad_norm": 0.05239463225007057, "learning_rate": 1.7012785987113194e-06, "loss": 0.0002, "step": 184910 }, { "epoch": 3.0257710872944448, "grad_norm": 0.04190749675035477, "learning_rate": 1.700742103753965e-06, "loss": 0.0004, "step": 184920 }, { "epoch": 3.0259347132455208, "grad_norm": 0.004036054480820894, "learning_rate": 1.7002056760656754e-06, "loss": 0.0002, "step": 184930 }, { "epoch": 3.0260983391965968, "grad_norm": 0.011693141423165798, "learning_rate": 1.6996693156573835e-06, "loss": 0.0002, "step": 184940 }, { "epoch": 3.0262619651476723, "grad_norm": 0.05404501035809517, "learning_rate": 1.6991330225400277e-06, "loss": 0.0003, "step": 184950 }, { "epoch": 3.0264255910987483, "grad_norm": 0.05394536256790161, "learning_rate": 1.6985967967245426e-06, "loss": 0.0003, "step": 184960 }, { "epoch": 3.0265892170498243, "grad_norm": 0.03172071650624275, "learning_rate": 1.6980606382218623e-06, "loss": 0.0005, "step": 184970 }, { "epoch": 3.0267528430009, "grad_norm": 0.0027518717106431723, "learning_rate": 1.6975245470429158e-06, "loss": 0.0009, "step": 184980 }, { "epoch": 3.026916468951976, "grad_norm": 0.02091214619576931, "learning_rate": 1.696988523198635e-06, "loss": 0.0002, "step": 184990 }, { "epoch": 3.0270800949030514, "grad_norm": 0.055226247757673264, "learning_rate": 1.6964525666999493e-06, "loss": 0.0006, "step": 185000 }, { "epoch": 3.0272437208541274, "grad_norm": 0.0013643569545820355, "learning_rate": 1.695916677557788e-06, "loss": 0.0003, "step": 185010 }, { "epoch": 3.0274073468052034, "grad_norm": 0.018811378628015518, "learning_rate": 1.6953808557830737e-06, "loss": 0.0002, "step": 185020 }, { "epoch": 3.027570972756279, "grad_norm": 0.01094093732535839, "learning_rate": 1.6948451013867335e-06, "loss": 0.0004, "step": 185030 }, { "epoch": 3.027734598707355, "grad_norm": 0.02265227772295475, "learning_rate": 1.6943094143796906e-06, "loss": 0.0002, "step": 185040 }, { "epoch": 3.027898224658431, "grad_norm": 0.017261363565921783, "learning_rate": 1.6937737947728689e-06, "loss": 0.0006, "step": 185050 }, { "epoch": 3.0280618506095065, "grad_norm": 0.021630151197314262, "learning_rate": 1.693238242577186e-06, "loss": 0.0002, "step": 185060 }, { "epoch": 3.0282254765605825, "grad_norm": 0.04304366186261177, "learning_rate": 1.6927027578035633e-06, "loss": 0.0003, "step": 185070 }, { "epoch": 3.0283891025116585, "grad_norm": 0.009561709128320217, "learning_rate": 1.6921673404629185e-06, "loss": 0.0003, "step": 185080 }, { "epoch": 3.028552728462734, "grad_norm": 0.005303670186549425, "learning_rate": 1.69163199056617e-06, "loss": 0.0004, "step": 185090 }, { "epoch": 3.02871635441381, "grad_norm": 0.021366693079471588, "learning_rate": 1.6910967081242296e-06, "loss": 0.0003, "step": 185100 }, { "epoch": 3.028879980364886, "grad_norm": 0.02946818433701992, "learning_rate": 1.690561493148014e-06, "loss": 0.0004, "step": 185110 }, { "epoch": 3.0290436063159616, "grad_norm": 0.0007869338733144104, "learning_rate": 1.6900263456484345e-06, "loss": 0.0005, "step": 185120 }, { "epoch": 3.0292072322670376, "grad_norm": 0.002943518338724971, "learning_rate": 1.6894912656364043e-06, "loss": 0.0019, "step": 185130 }, { "epoch": 3.0293708582181136, "grad_norm": 0.06552290171384811, "learning_rate": 1.6889562531228304e-06, "loss": 0.0002, "step": 185140 }, { "epoch": 3.029534484169189, "grad_norm": 0.015441596508026123, "learning_rate": 1.688421308118624e-06, "loss": 0.0005, "step": 185150 }, { "epoch": 3.029698110120265, "grad_norm": 0.03402986377477646, "learning_rate": 1.687886430634687e-06, "loss": 0.0003, "step": 185160 }, { "epoch": 3.029861736071341, "grad_norm": 0.002395795891061425, "learning_rate": 1.6873516206819329e-06, "loss": 0.0001, "step": 185170 }, { "epoch": 3.0300253620224167, "grad_norm": 0.03385026007890701, "learning_rate": 1.68681687827126e-06, "loss": 0.0005, "step": 185180 }, { "epoch": 3.0301889879734927, "grad_norm": 0.0002477781381458044, "learning_rate": 1.6862822034135746e-06, "loss": 0.0004, "step": 185190 }, { "epoch": 3.0303526139245682, "grad_norm": 0.06484032422304153, "learning_rate": 1.685747596119774e-06, "loss": 0.0003, "step": 185200 }, { "epoch": 3.0305162398756442, "grad_norm": 0.35491687059402466, "learning_rate": 1.685213056400764e-06, "loss": 0.0004, "step": 185210 }, { "epoch": 3.03067986582672, "grad_norm": 0.030663397163152695, "learning_rate": 1.6846785842674396e-06, "loss": 0.0022, "step": 185220 }, { "epoch": 3.0308434917777958, "grad_norm": 0.03933076560497284, "learning_rate": 1.6841441797307006e-06, "loss": 0.0004, "step": 185230 }, { "epoch": 3.0310071177288718, "grad_norm": 0.006093719974160194, "learning_rate": 1.6836098428014386e-06, "loss": 0.0004, "step": 185240 }, { "epoch": 3.0311707436799478, "grad_norm": 0.01864355243742466, "learning_rate": 1.6830755734905545e-06, "loss": 0.0004, "step": 185250 }, { "epoch": 3.0313343696310233, "grad_norm": 0.006987098604440689, "learning_rate": 1.6825413718089372e-06, "loss": 0.0002, "step": 185260 }, { "epoch": 3.0314979955820993, "grad_norm": 0.011979526840150356, "learning_rate": 1.6820072377674811e-06, "loss": 0.0002, "step": 185270 }, { "epoch": 3.0316616215331753, "grad_norm": 0.04836736619472504, "learning_rate": 1.6814731713770727e-06, "loss": 0.0005, "step": 185280 }, { "epoch": 3.031825247484251, "grad_norm": 0.0018524975748732686, "learning_rate": 1.6809391726486069e-06, "loss": 0.0003, "step": 185290 }, { "epoch": 3.031988873435327, "grad_norm": 0.003893628716468811, "learning_rate": 1.6804052415929672e-06, "loss": 0.0003, "step": 185300 }, { "epoch": 3.032152499386403, "grad_norm": 0.02319318987429142, "learning_rate": 1.6798713782210436e-06, "loss": 0.0002, "step": 185310 }, { "epoch": 3.0323161253374784, "grad_norm": 0.03970054164528847, "learning_rate": 1.6793375825437163e-06, "loss": 0.0004, "step": 185320 }, { "epoch": 3.0324797512885544, "grad_norm": 0.023296672850847244, "learning_rate": 1.678803854571872e-06, "loss": 0.0003, "step": 185330 }, { "epoch": 3.0326433772396304, "grad_norm": 0.005305591505020857, "learning_rate": 1.6782701943163926e-06, "loss": 0.0004, "step": 185340 }, { "epoch": 3.032807003190706, "grad_norm": 0.01697123982012272, "learning_rate": 1.6777366017881607e-06, "loss": 0.0003, "step": 185350 }, { "epoch": 3.032970629141782, "grad_norm": 0.0010211082408204675, "learning_rate": 1.6772030769980519e-06, "loss": 0.0003, "step": 185360 }, { "epoch": 3.033134255092858, "grad_norm": 0.038981132209300995, "learning_rate": 1.676669619956947e-06, "loss": 0.0002, "step": 185370 }, { "epoch": 3.0332978810439335, "grad_norm": 0.0863332450389862, "learning_rate": 1.6761362306757217e-06, "loss": 0.0003, "step": 185380 }, { "epoch": 3.0334615069950095, "grad_norm": 0.07607802003622055, "learning_rate": 1.6756029091652538e-06, "loss": 0.0002, "step": 185390 }, { "epoch": 3.033625132946085, "grad_norm": 0.0010016115847975016, "learning_rate": 1.6750696554364132e-06, "loss": 0.0005, "step": 185400 }, { "epoch": 3.033788758897161, "grad_norm": 0.0027606510557234287, "learning_rate": 1.6745364695000748e-06, "loss": 0.0003, "step": 185410 }, { "epoch": 3.033952384848237, "grad_norm": 0.024898847565054893, "learning_rate": 1.67400335136711e-06, "loss": 0.0005, "step": 185420 }, { "epoch": 3.0341160107993126, "grad_norm": 0.11896813660860062, "learning_rate": 1.6734703010483894e-06, "loss": 0.001, "step": 185430 }, { "epoch": 3.0342796367503886, "grad_norm": 0.0028620599769055843, "learning_rate": 1.6729373185547788e-06, "loss": 0.0006, "step": 185440 }, { "epoch": 3.0344432627014646, "grad_norm": 0.10487113147974014, "learning_rate": 1.672404403897146e-06, "loss": 0.0004, "step": 185450 }, { "epoch": 3.03460688865254, "grad_norm": 0.010365704074501991, "learning_rate": 1.671871557086358e-06, "loss": 0.0008, "step": 185460 }, { "epoch": 3.034770514603616, "grad_norm": 0.01630973443388939, "learning_rate": 1.6713387781332803e-06, "loss": 0.0004, "step": 185470 }, { "epoch": 3.034934140554692, "grad_norm": 0.0011461310787126422, "learning_rate": 1.670806067048772e-06, "loss": 0.0005, "step": 185480 }, { "epoch": 3.0350977665057677, "grad_norm": 0.0032284599728882313, "learning_rate": 1.6702734238436969e-06, "loss": 0.0001, "step": 185490 }, { "epoch": 3.0352613924568437, "grad_norm": 0.01799272932112217, "learning_rate": 1.6697408485289145e-06, "loss": 0.0005, "step": 185500 }, { "epoch": 3.0354250184079197, "grad_norm": 0.025082506239414215, "learning_rate": 1.669208341115286e-06, "loss": 0.0003, "step": 185510 }, { "epoch": 3.0355886443589952, "grad_norm": 0.004252503160387278, "learning_rate": 1.6686759016136645e-06, "loss": 0.0004, "step": 185520 }, { "epoch": 3.035752270310071, "grad_norm": 0.03661943972110748, "learning_rate": 1.6681435300349087e-06, "loss": 0.0004, "step": 185530 }, { "epoch": 3.035915896261147, "grad_norm": 0.03031335212290287, "learning_rate": 1.6676112263898725e-06, "loss": 0.0004, "step": 185540 }, { "epoch": 3.0360795222122228, "grad_norm": 0.02808194048702717, "learning_rate": 1.6670789906894108e-06, "loss": 0.0002, "step": 185550 }, { "epoch": 3.0362431481632988, "grad_norm": 0.016876008361577988, "learning_rate": 1.6665468229443721e-06, "loss": 0.0003, "step": 185560 }, { "epoch": 3.0364067741143748, "grad_norm": 0.04518656060099602, "learning_rate": 1.6660147231656093e-06, "loss": 0.0003, "step": 185570 }, { "epoch": 3.0365704000654503, "grad_norm": 0.03910220041871071, "learning_rate": 1.6654826913639705e-06, "loss": 0.0005, "step": 185580 }, { "epoch": 3.0367340260165263, "grad_norm": 0.07681147754192352, "learning_rate": 1.6649507275503053e-06, "loss": 0.0004, "step": 185590 }, { "epoch": 3.036897651967602, "grad_norm": 0.0390087328851223, "learning_rate": 1.664418831735457e-06, "loss": 0.0004, "step": 185600 }, { "epoch": 3.037061277918678, "grad_norm": 0.04564997926354408, "learning_rate": 1.6638870039302718e-06, "loss": 0.0003, "step": 185610 }, { "epoch": 3.037224903869754, "grad_norm": 0.00030666502425447106, "learning_rate": 1.663355244145593e-06, "loss": 0.0002, "step": 185620 }, { "epoch": 3.0373885298208294, "grad_norm": 0.03073839657008648, "learning_rate": 1.6628235523922649e-06, "loss": 0.0004, "step": 185630 }, { "epoch": 3.0375521557719054, "grad_norm": 0.06721389293670654, "learning_rate": 1.6622919286811251e-06, "loss": 0.0004, "step": 185640 }, { "epoch": 3.0377157817229814, "grad_norm": 0.0007633297354914248, "learning_rate": 1.6617603730230137e-06, "loss": 0.0002, "step": 185650 }, { "epoch": 3.037879407674057, "grad_norm": 0.012769737280905247, "learning_rate": 1.6612288854287695e-06, "loss": 0.0003, "step": 185660 }, { "epoch": 3.038043033625133, "grad_norm": 0.03250228241086006, "learning_rate": 1.6606974659092306e-06, "loss": 0.0003, "step": 185670 }, { "epoch": 3.038206659576209, "grad_norm": 0.00590570829808712, "learning_rate": 1.6601661144752284e-06, "loss": 0.0002, "step": 185680 }, { "epoch": 3.0383702855272845, "grad_norm": 0.016675548627972603, "learning_rate": 1.6596348311375992e-06, "loss": 0.0004, "step": 185690 }, { "epoch": 3.0385339114783605, "grad_norm": 0.02214682474732399, "learning_rate": 1.6591036159071744e-06, "loss": 0.0006, "step": 185700 }, { "epoch": 3.0386975374294365, "grad_norm": 0.016265802085399628, "learning_rate": 1.6585724687947875e-06, "loss": 0.0004, "step": 185710 }, { "epoch": 3.038861163380512, "grad_norm": 0.21200723946094513, "learning_rate": 1.6580413898112646e-06, "loss": 0.0003, "step": 185720 }, { "epoch": 3.039024789331588, "grad_norm": 0.045600682497024536, "learning_rate": 1.6575103789674353e-06, "loss": 0.0002, "step": 185730 }, { "epoch": 3.039188415282664, "grad_norm": 0.008355236612260342, "learning_rate": 1.6569794362741265e-06, "loss": 0.0003, "step": 185740 }, { "epoch": 3.0393520412337396, "grad_norm": 0.010816008783876896, "learning_rate": 1.656448561742166e-06, "loss": 0.0002, "step": 185750 }, { "epoch": 3.0395156671848156, "grad_norm": 0.03026445396244526, "learning_rate": 1.6559177553823736e-06, "loss": 0.0001, "step": 185760 }, { "epoch": 3.039679293135891, "grad_norm": 0.10436386615037918, "learning_rate": 1.6553870172055765e-06, "loss": 0.0004, "step": 185770 }, { "epoch": 3.039842919086967, "grad_norm": 0.0028673349879682064, "learning_rate": 1.6548563472225892e-06, "loss": 0.0002, "step": 185780 }, { "epoch": 3.040006545038043, "grad_norm": 0.03439177945256233, "learning_rate": 1.6543257454442403e-06, "loss": 0.0003, "step": 185790 }, { "epoch": 3.0401701709891187, "grad_norm": 0.012370780110359192, "learning_rate": 1.6537952118813423e-06, "loss": 0.0004, "step": 185800 }, { "epoch": 3.0403337969401947, "grad_norm": 0.0084277568385005, "learning_rate": 1.6532647465447154e-06, "loss": 0.0008, "step": 185810 }, { "epoch": 3.0404974228912707, "grad_norm": 0.009715371765196323, "learning_rate": 1.652734349445171e-06, "loss": 0.0006, "step": 185820 }, { "epoch": 3.0406610488423462, "grad_norm": 0.004290800075978041, "learning_rate": 1.6522040205935291e-06, "loss": 0.0006, "step": 185830 }, { "epoch": 3.0408246747934222, "grad_norm": 0.015586741268634796, "learning_rate": 1.6516737600005994e-06, "loss": 0.0002, "step": 185840 }, { "epoch": 3.040988300744498, "grad_norm": 0.007680861745029688, "learning_rate": 1.6511435676771947e-06, "loss": 0.0002, "step": 185850 }, { "epoch": 3.0411519266955738, "grad_norm": 0.0016997926868498325, "learning_rate": 1.650613443634122e-06, "loss": 0.0005, "step": 185860 }, { "epoch": 3.0413155526466498, "grad_norm": 0.1370595395565033, "learning_rate": 1.650083387882196e-06, "loss": 0.0005, "step": 185870 }, { "epoch": 3.0414791785977258, "grad_norm": 0.007176091894507408, "learning_rate": 1.6495534004322183e-06, "loss": 0.0016, "step": 185880 }, { "epoch": 3.0416428045488013, "grad_norm": 0.002095041563734412, "learning_rate": 1.6490234812949995e-06, "loss": 0.0004, "step": 185890 }, { "epoch": 3.0418064304998773, "grad_norm": 0.02588627114892006, "learning_rate": 1.6484936304813387e-06, "loss": 0.0003, "step": 185900 }, { "epoch": 3.0419700564509533, "grad_norm": 0.019836537539958954, "learning_rate": 1.647963848002046e-06, "loss": 0.0005, "step": 185910 }, { "epoch": 3.042133682402029, "grad_norm": 0.0067936270497739315, "learning_rate": 1.6474341338679173e-06, "loss": 0.0003, "step": 185920 }, { "epoch": 3.042297308353105, "grad_norm": 0.15698398649692535, "learning_rate": 1.6469044880897578e-06, "loss": 0.0005, "step": 185930 }, { "epoch": 3.042460934304181, "grad_norm": 0.038602765649557114, "learning_rate": 1.64637491067836e-06, "loss": 0.0008, "step": 185940 }, { "epoch": 3.0426245602552564, "grad_norm": 0.03146769106388092, "learning_rate": 1.6458454016445292e-06, "loss": 0.0003, "step": 185950 }, { "epoch": 3.0427881862063324, "grad_norm": 0.01798064447939396, "learning_rate": 1.6453159609990565e-06, "loss": 0.0003, "step": 185960 }, { "epoch": 3.042951812157408, "grad_norm": 0.09618185460567474, "learning_rate": 1.6447865887527397e-06, "loss": 0.0004, "step": 185970 }, { "epoch": 3.043115438108484, "grad_norm": 0.021439217031002045, "learning_rate": 1.6442572849163695e-06, "loss": 0.0002, "step": 185980 }, { "epoch": 3.04327906405956, "grad_norm": 0.0033397178631275892, "learning_rate": 1.6437280495007397e-06, "loss": 0.0004, "step": 185990 }, { "epoch": 3.0434426900106355, "grad_norm": 0.02193615585565567, "learning_rate": 1.64319888251664e-06, "loss": 0.0001, "step": 186000 }, { "epoch": 3.0436063159617115, "grad_norm": 0.027636319398880005, "learning_rate": 1.6426697839748618e-06, "loss": 0.0002, "step": 186010 }, { "epoch": 3.0437699419127875, "grad_norm": 0.041304782032966614, "learning_rate": 1.64214075388619e-06, "loss": 0.0003, "step": 186020 }, { "epoch": 3.043933567863863, "grad_norm": 0.0003842230944428593, "learning_rate": 1.6416117922614122e-06, "loss": 0.0002, "step": 186030 }, { "epoch": 3.044097193814939, "grad_norm": 0.014386137016117573, "learning_rate": 1.6410828991113143e-06, "loss": 0.0004, "step": 186040 }, { "epoch": 3.044260819766015, "grad_norm": 0.05509496480226517, "learning_rate": 1.6405540744466801e-06, "loss": 0.0005, "step": 186050 }, { "epoch": 3.0444244457170906, "grad_norm": 0.040185388177633286, "learning_rate": 1.6400253182782905e-06, "loss": 0.0003, "step": 186060 }, { "epoch": 3.0445880716681666, "grad_norm": 0.06312817335128784, "learning_rate": 1.639496630616927e-06, "loss": 0.001, "step": 186070 }, { "epoch": 3.0447516976192426, "grad_norm": 0.09576680511236191, "learning_rate": 1.6389680114733686e-06, "loss": 0.0003, "step": 186080 }, { "epoch": 3.044915323570318, "grad_norm": 0.03972389176487923, "learning_rate": 1.638439460858396e-06, "loss": 0.0003, "step": 186090 }, { "epoch": 3.045078949521394, "grad_norm": 0.017525488510727882, "learning_rate": 1.6379109787827824e-06, "loss": 0.0003, "step": 186100 }, { "epoch": 3.04524257547247, "grad_norm": 0.04664516821503639, "learning_rate": 1.6373825652573044e-06, "loss": 0.0003, "step": 186110 }, { "epoch": 3.0454062014235457, "grad_norm": 0.05322020873427391, "learning_rate": 1.6368542202927362e-06, "loss": 0.0002, "step": 186120 }, { "epoch": 3.0455698273746217, "grad_norm": 0.04919896647334099, "learning_rate": 1.636325943899852e-06, "loss": 0.0005, "step": 186130 }, { "epoch": 3.0457334533256977, "grad_norm": 0.04491907358169556, "learning_rate": 1.6357977360894188e-06, "loss": 0.0004, "step": 186140 }, { "epoch": 3.0458970792767732, "grad_norm": 0.05368174612522125, "learning_rate": 1.6352695968722094e-06, "loss": 0.0004, "step": 186150 }, { "epoch": 3.046060705227849, "grad_norm": 0.0567261278629303, "learning_rate": 1.6347415262589906e-06, "loss": 0.0004, "step": 186160 }, { "epoch": 3.0462243311789248, "grad_norm": 0.03565910458564758, "learning_rate": 1.6342135242605322e-06, "loss": 0.0003, "step": 186170 }, { "epoch": 3.0463879571300008, "grad_norm": 0.019255878403782845, "learning_rate": 1.6336855908875958e-06, "loss": 0.0003, "step": 186180 }, { "epoch": 3.0465515830810768, "grad_norm": 0.03501967713236809, "learning_rate": 1.6331577261509475e-06, "loss": 0.0007, "step": 186190 }, { "epoch": 3.0467152090321523, "grad_norm": 0.007700914982706308, "learning_rate": 1.632629930061349e-06, "loss": 0.0002, "step": 186200 }, { "epoch": 3.0468788349832283, "grad_norm": 0.0038152909837663174, "learning_rate": 1.6321022026295646e-06, "loss": 0.0003, "step": 186210 }, { "epoch": 3.0470424609343043, "grad_norm": 0.0480671189725399, "learning_rate": 1.63157454386635e-06, "loss": 0.0002, "step": 186220 }, { "epoch": 3.04720608688538, "grad_norm": 0.013987450860440731, "learning_rate": 1.6310469537824663e-06, "loss": 0.0003, "step": 186230 }, { "epoch": 3.047369712836456, "grad_norm": 0.0004062009684275836, "learning_rate": 1.6305194323886698e-06, "loss": 0.0003, "step": 186240 }, { "epoch": 3.047533338787532, "grad_norm": 0.09976331144571304, "learning_rate": 1.6299919796957176e-06, "loss": 0.0005, "step": 186250 }, { "epoch": 3.0476969647386074, "grad_norm": 0.025049589574337006, "learning_rate": 1.6294645957143618e-06, "loss": 0.0001, "step": 186260 }, { "epoch": 3.0478605906896834, "grad_norm": 0.0006539165624417365, "learning_rate": 1.6289372804553561e-06, "loss": 0.0003, "step": 186270 }, { "epoch": 3.0480242166407594, "grad_norm": 0.011453203856945038, "learning_rate": 1.6284100339294518e-06, "loss": 0.0001, "step": 186280 }, { "epoch": 3.048187842591835, "grad_norm": 0.014557461254298687, "learning_rate": 1.627882856147402e-06, "loss": 0.0004, "step": 186290 }, { "epoch": 3.048351468542911, "grad_norm": 0.09153107553720474, "learning_rate": 1.6273557471199508e-06, "loss": 0.0002, "step": 186300 }, { "epoch": 3.048515094493987, "grad_norm": 0.05113891884684563, "learning_rate": 1.6268287068578476e-06, "loss": 0.0004, "step": 186310 }, { "epoch": 3.0486787204450625, "grad_norm": 0.04599897190928459, "learning_rate": 1.626301735371838e-06, "loss": 0.0003, "step": 186320 }, { "epoch": 3.0488423463961385, "grad_norm": 0.009846501983702183, "learning_rate": 1.6257748326726686e-06, "loss": 0.0002, "step": 186330 }, { "epoch": 3.0490059723472145, "grad_norm": 0.012956011109054089, "learning_rate": 1.6252479987710795e-06, "loss": 0.0004, "step": 186340 }, { "epoch": 3.04916959829829, "grad_norm": 0.0933169573545456, "learning_rate": 1.6247212336778128e-06, "loss": 0.0004, "step": 186350 }, { "epoch": 3.049333224249366, "grad_norm": 0.03178418427705765, "learning_rate": 1.62419453740361e-06, "loss": 0.0004, "step": 186360 }, { "epoch": 3.0494968502004416, "grad_norm": 0.010679385624825954, "learning_rate": 1.6236679099592112e-06, "loss": 0.0011, "step": 186370 }, { "epoch": 3.0496604761515176, "grad_norm": 0.03730376064777374, "learning_rate": 1.6231413513553506e-06, "loss": 0.0002, "step": 186380 }, { "epoch": 3.0498241021025936, "grad_norm": 0.03270159289240837, "learning_rate": 1.622614861602766e-06, "loss": 0.0003, "step": 186390 }, { "epoch": 3.049987728053669, "grad_norm": 0.02516099438071251, "learning_rate": 1.6220884407121924e-06, "loss": 0.0006, "step": 186400 }, { "epoch": 3.050151354004745, "grad_norm": 0.012613322585821152, "learning_rate": 1.6215620886943633e-06, "loss": 0.0002, "step": 186410 }, { "epoch": 3.050314979955821, "grad_norm": 0.10829927027225494, "learning_rate": 1.6210358055600095e-06, "loss": 0.0007, "step": 186420 }, { "epoch": 3.0504786059068967, "grad_norm": 0.003118082880973816, "learning_rate": 1.6205095913198627e-06, "loss": 0.0003, "step": 186430 }, { "epoch": 3.0506422318579727, "grad_norm": 0.017486318945884705, "learning_rate": 1.619983445984648e-06, "loss": 0.0005, "step": 186440 }, { "epoch": 3.0508058578090487, "grad_norm": 0.01640160195529461, "learning_rate": 1.6194573695651e-06, "loss": 0.0002, "step": 186450 }, { "epoch": 3.0509694837601242, "grad_norm": 0.07353264838457108, "learning_rate": 1.6189313620719388e-06, "loss": 0.0003, "step": 186460 }, { "epoch": 3.0511331097112, "grad_norm": 0.0383480042219162, "learning_rate": 1.6184054235158942e-06, "loss": 0.0005, "step": 186470 }, { "epoch": 3.051296735662276, "grad_norm": 0.06912926584482193, "learning_rate": 1.617879553907683e-06, "loss": 0.0004, "step": 186480 }, { "epoch": 3.0514603616133518, "grad_norm": 0.12675148248672485, "learning_rate": 1.6173537532580346e-06, "loss": 0.0007, "step": 186490 }, { "epoch": 3.0516239875644278, "grad_norm": 0.03491789847612381, "learning_rate": 1.6168280215776649e-06, "loss": 0.0004, "step": 186500 }, { "epoch": 3.0517876135155038, "grad_norm": 0.05216995254158974, "learning_rate": 1.6163023588772963e-06, "loss": 0.0003, "step": 186510 }, { "epoch": 3.0519512394665793, "grad_norm": 0.009232431650161743, "learning_rate": 1.615776765167642e-06, "loss": 0.0004, "step": 186520 }, { "epoch": 3.0521148654176553, "grad_norm": 0.04049041494727135, "learning_rate": 1.6152512404594239e-06, "loss": 0.0004, "step": 186530 }, { "epoch": 3.052278491368731, "grad_norm": 0.007996642030775547, "learning_rate": 1.614725784763353e-06, "loss": 0.0006, "step": 186540 }, { "epoch": 3.052442117319807, "grad_norm": 0.012062082067131996, "learning_rate": 1.614200398090146e-06, "loss": 0.0013, "step": 186550 }, { "epoch": 3.052605743270883, "grad_norm": 0.09822995215654373, "learning_rate": 1.6136750804505096e-06, "loss": 0.0005, "step": 186560 }, { "epoch": 3.0527693692219584, "grad_norm": 0.0036691001150757074, "learning_rate": 1.613149831855162e-06, "loss": 0.0003, "step": 186570 }, { "epoch": 3.0529329951730344, "grad_norm": 0.03609241545200348, "learning_rate": 1.6126246523148075e-06, "loss": 0.0004, "step": 186580 }, { "epoch": 3.0530966211241104, "grad_norm": 0.02823146991431713, "learning_rate": 1.6120995418401569e-06, "loss": 0.0003, "step": 186590 }, { "epoch": 3.053260247075186, "grad_norm": 0.07909906655550003, "learning_rate": 1.6115745004419136e-06, "loss": 0.0003, "step": 186600 }, { "epoch": 3.053423873026262, "grad_norm": 0.12011956423521042, "learning_rate": 1.6110495281307842e-06, "loss": 0.0002, "step": 186610 }, { "epoch": 3.053587498977338, "grad_norm": 0.000592098047491163, "learning_rate": 1.6105246249174733e-06, "loss": 0.0004, "step": 186620 }, { "epoch": 3.0537511249284135, "grad_norm": 0.001305286306887865, "learning_rate": 1.6099997908126842e-06, "loss": 0.0003, "step": 186630 }, { "epoch": 3.0539147508794895, "grad_norm": 0.030391911044716835, "learning_rate": 1.609475025827114e-06, "loss": 0.0001, "step": 186640 }, { "epoch": 3.0540783768305655, "grad_norm": 0.04372820258140564, "learning_rate": 1.6089503299714654e-06, "loss": 0.0004, "step": 186650 }, { "epoch": 3.054242002781641, "grad_norm": 0.04249125346541405, "learning_rate": 1.6084257032564349e-06, "loss": 0.0004, "step": 186660 }, { "epoch": 3.054405628732717, "grad_norm": 0.033630672842264175, "learning_rate": 1.6079011456927219e-06, "loss": 0.0002, "step": 186670 }, { "epoch": 3.054569254683793, "grad_norm": 0.2213255614042282, "learning_rate": 1.6073766572910183e-06, "loss": 0.0006, "step": 186680 }, { "epoch": 3.0547328806348686, "grad_norm": 0.04330792278051376, "learning_rate": 1.6068522380620189e-06, "loss": 0.0003, "step": 186690 }, { "epoch": 3.0548965065859446, "grad_norm": 0.032329265028238297, "learning_rate": 1.6063278880164163e-06, "loss": 0.0004, "step": 186700 }, { "epoch": 3.0550601325370206, "grad_norm": 0.057162366807460785, "learning_rate": 1.6058036071649036e-06, "loss": 0.0003, "step": 186710 }, { "epoch": 3.055223758488096, "grad_norm": 0.042215533554553986, "learning_rate": 1.6052793955181673e-06, "loss": 0.0002, "step": 186720 }, { "epoch": 3.055387384439172, "grad_norm": 0.025642843917012215, "learning_rate": 1.604755253086897e-06, "loss": 0.0011, "step": 186730 }, { "epoch": 3.0555510103902477, "grad_norm": 0.006779700517654419, "learning_rate": 1.60423117988178e-06, "loss": 0.0012, "step": 186740 }, { "epoch": 3.0557146363413237, "grad_norm": 0.004413953982293606, "learning_rate": 1.603707175913502e-06, "loss": 0.0002, "step": 186750 }, { "epoch": 3.0558782622923997, "grad_norm": 0.005296788644045591, "learning_rate": 1.6031832411927451e-06, "loss": 0.0004, "step": 186760 }, { "epoch": 3.0560418882434752, "grad_norm": 0.009204911068081856, "learning_rate": 1.6026593757301928e-06, "loss": 0.0003, "step": 186770 }, { "epoch": 3.056205514194551, "grad_norm": 0.026162810623645782, "learning_rate": 1.6021355795365268e-06, "loss": 0.0005, "step": 186780 }, { "epoch": 3.056369140145627, "grad_norm": 0.00889889057725668, "learning_rate": 1.6016118526224278e-06, "loss": 0.0003, "step": 186790 }, { "epoch": 3.0565327660967028, "grad_norm": 0.0011675627902150154, "learning_rate": 1.6010881949985718e-06, "loss": 0.0006, "step": 186800 }, { "epoch": 3.0566963920477788, "grad_norm": 0.025025147944688797, "learning_rate": 1.6005646066756364e-06, "loss": 0.0008, "step": 186810 }, { "epoch": 3.0568600179988548, "grad_norm": 0.019945034757256508, "learning_rate": 1.600041087664298e-06, "loss": 0.0003, "step": 186820 }, { "epoch": 3.0570236439499303, "grad_norm": 0.010918604210019112, "learning_rate": 1.5995176379752314e-06, "loss": 0.0004, "step": 186830 }, { "epoch": 3.0571872699010063, "grad_norm": 0.07626216113567352, "learning_rate": 1.5989942576191064e-06, "loss": 0.0002, "step": 186840 }, { "epoch": 3.0573508958520823, "grad_norm": 0.07386848330497742, "learning_rate": 1.5984709466065968e-06, "loss": 0.0002, "step": 186850 }, { "epoch": 3.057514521803158, "grad_norm": 0.003914120141416788, "learning_rate": 1.597947704948371e-06, "loss": 0.0001, "step": 186860 }, { "epoch": 3.057678147754234, "grad_norm": 0.00016943410446401685, "learning_rate": 1.5974245326551001e-06, "loss": 0.0003, "step": 186870 }, { "epoch": 3.05784177370531, "grad_norm": 0.04403695464134216, "learning_rate": 1.5969014297374475e-06, "loss": 0.0002, "step": 186880 }, { "epoch": 3.0580053996563854, "grad_norm": 0.03571514040231705, "learning_rate": 1.5963783962060802e-06, "loss": 0.0004, "step": 186890 }, { "epoch": 3.0581690256074614, "grad_norm": 0.05571533739566803, "learning_rate": 1.595855432071663e-06, "loss": 0.0004, "step": 186900 }, { "epoch": 3.0583326515585374, "grad_norm": 0.04350828751921654, "learning_rate": 1.5953325373448603e-06, "loss": 0.0003, "step": 186910 }, { "epoch": 3.058496277509613, "grad_norm": 0.03334376960992813, "learning_rate": 1.59480971203633e-06, "loss": 0.0003, "step": 186920 }, { "epoch": 3.058659903460689, "grad_norm": 0.0974610298871994, "learning_rate": 1.5942869561567331e-06, "loss": 0.0005, "step": 186930 }, { "epoch": 3.0588235294117645, "grad_norm": 0.02810271643102169, "learning_rate": 1.5937642697167288e-06, "loss": 0.0005, "step": 186940 }, { "epoch": 3.0589871553628405, "grad_norm": 0.12456294149160385, "learning_rate": 1.5932416527269761e-06, "loss": 0.0004, "step": 186950 }, { "epoch": 3.0591507813139165, "grad_norm": 0.015305531211197376, "learning_rate": 1.5927191051981273e-06, "loss": 0.0003, "step": 186960 }, { "epoch": 3.059314407264992, "grad_norm": 0.024396615102887154, "learning_rate": 1.5921966271408374e-06, "loss": 0.0004, "step": 186970 }, { "epoch": 3.059478033216068, "grad_norm": 0.0008134908857755363, "learning_rate": 1.5916742185657608e-06, "loss": 0.0003, "step": 186980 }, { "epoch": 3.059641659167144, "grad_norm": 0.06354181468486786, "learning_rate": 1.5911518794835494e-06, "loss": 0.0002, "step": 186990 }, { "epoch": 3.0598052851182196, "grad_norm": 0.0002914668002631515, "learning_rate": 1.5906296099048508e-06, "loss": 0.0003, "step": 187000 }, { "epoch": 3.0599689110692956, "grad_norm": 0.011612040922045708, "learning_rate": 1.5901074098403142e-06, "loss": 0.0004, "step": 187010 }, { "epoch": 3.0601325370203716, "grad_norm": 0.036702997982501984, "learning_rate": 1.5895852793005883e-06, "loss": 0.0005, "step": 187020 }, { "epoch": 3.060296162971447, "grad_norm": 0.03519263118505478, "learning_rate": 1.589063218296319e-06, "loss": 0.0003, "step": 187030 }, { "epoch": 3.060459788922523, "grad_norm": 0.001011071726679802, "learning_rate": 1.588541226838149e-06, "loss": 0.0003, "step": 187040 }, { "epoch": 3.060623414873599, "grad_norm": 0.00580094289034605, "learning_rate": 1.5880193049367232e-06, "loss": 0.0003, "step": 187050 }, { "epoch": 3.0607870408246747, "grad_norm": 0.043450407683849335, "learning_rate": 1.5874974526026783e-06, "loss": 0.0007, "step": 187060 }, { "epoch": 3.0609506667757507, "grad_norm": 0.02263975888490677, "learning_rate": 1.586975669846662e-06, "loss": 0.0004, "step": 187070 }, { "epoch": 3.0611142927268267, "grad_norm": 0.01698991470038891, "learning_rate": 1.5864539566793075e-06, "loss": 0.0005, "step": 187080 }, { "epoch": 3.061277918677902, "grad_norm": 0.008189492858946323, "learning_rate": 1.5859323131112547e-06, "loss": 0.0003, "step": 187090 }, { "epoch": 3.061441544628978, "grad_norm": 0.0035396991297602654, "learning_rate": 1.585410739153136e-06, "loss": 0.0001, "step": 187100 }, { "epoch": 3.061605170580054, "grad_norm": 0.1217251718044281, "learning_rate": 1.5848892348155908e-06, "loss": 0.0005, "step": 187110 }, { "epoch": 3.0617687965311298, "grad_norm": 0.002386003965511918, "learning_rate": 1.584367800109249e-06, "loss": 0.0002, "step": 187120 }, { "epoch": 3.0619324224822058, "grad_norm": 0.03004358895123005, "learning_rate": 1.5838464350447436e-06, "loss": 0.0002, "step": 187130 }, { "epoch": 3.0620960484332813, "grad_norm": 0.027022836729884148, "learning_rate": 1.583325139632701e-06, "loss": 0.0002, "step": 187140 }, { "epoch": 3.0622596743843573, "grad_norm": 0.03955508768558502, "learning_rate": 1.5828039138837565e-06, "loss": 0.0004, "step": 187150 }, { "epoch": 3.0624233003354333, "grad_norm": 0.01694759912788868, "learning_rate": 1.5822827578085327e-06, "loss": 0.0002, "step": 187160 }, { "epoch": 3.062586926286509, "grad_norm": 0.010968535207211971, "learning_rate": 1.5817616714176576e-06, "loss": 0.0003, "step": 187170 }, { "epoch": 3.062750552237585, "grad_norm": 0.02332761138677597, "learning_rate": 1.5812406547217524e-06, "loss": 0.0003, "step": 187180 }, { "epoch": 3.062914178188661, "grad_norm": 0.026023784652352333, "learning_rate": 1.5807197077314451e-06, "loss": 0.0002, "step": 187190 }, { "epoch": 3.0630778041397364, "grad_norm": 0.054202061146497726, "learning_rate": 1.580198830457354e-06, "loss": 0.0005, "step": 187200 }, { "epoch": 3.0632414300908124, "grad_norm": 0.015110677108168602, "learning_rate": 1.5796780229101017e-06, "loss": 0.0001, "step": 187210 }, { "epoch": 3.0634050560418884, "grad_norm": 0.08136854320764542, "learning_rate": 1.579157285100303e-06, "loss": 0.0004, "step": 187220 }, { "epoch": 3.063568681992964, "grad_norm": 0.0019239394459873438, "learning_rate": 1.5786366170385802e-06, "loss": 0.0001, "step": 187230 }, { "epoch": 3.06373230794404, "grad_norm": 0.00939460564404726, "learning_rate": 1.578116018735546e-06, "loss": 0.0002, "step": 187240 }, { "epoch": 3.063895933895116, "grad_norm": 0.0010411603143438697, "learning_rate": 1.5775954902018176e-06, "loss": 0.0004, "step": 187250 }, { "epoch": 3.0640595598461915, "grad_norm": 0.07697215676307678, "learning_rate": 1.5770750314480043e-06, "loss": 0.0004, "step": 187260 }, { "epoch": 3.0642231857972675, "grad_norm": 0.0002371899608988315, "learning_rate": 1.5765546424847206e-06, "loss": 0.0002, "step": 187270 }, { "epoch": 3.0643868117483435, "grad_norm": 0.02008044719696045, "learning_rate": 1.5760343233225761e-06, "loss": 0.0003, "step": 187280 }, { "epoch": 3.064550437699419, "grad_norm": 0.02651407942175865, "learning_rate": 1.5755140739721813e-06, "loss": 0.0002, "step": 187290 }, { "epoch": 3.064714063650495, "grad_norm": 0.04290300980210304, "learning_rate": 1.5749938944441407e-06, "loss": 0.0002, "step": 187300 }, { "epoch": 3.0648776896015706, "grad_norm": 0.0064421058632433414, "learning_rate": 1.5744737847490616e-06, "loss": 0.0002, "step": 187310 }, { "epoch": 3.0650413155526466, "grad_norm": 0.025270257145166397, "learning_rate": 1.5739537448975483e-06, "loss": 0.0003, "step": 187320 }, { "epoch": 3.0652049415037226, "grad_norm": 0.03303638845682144, "learning_rate": 1.5734337749002065e-06, "loss": 0.0004, "step": 187330 }, { "epoch": 3.065368567454798, "grad_norm": 0.04233492910861969, "learning_rate": 1.5729138747676343e-06, "loss": 0.0002, "step": 187340 }, { "epoch": 3.065532193405874, "grad_norm": 0.042652860283851624, "learning_rate": 1.572394044510433e-06, "loss": 0.0003, "step": 187350 }, { "epoch": 3.06569581935695, "grad_norm": 0.05285966768860817, "learning_rate": 1.5718742841392027e-06, "loss": 0.0005, "step": 187360 }, { "epoch": 3.0658594453080257, "grad_norm": 0.02126757800579071, "learning_rate": 1.5713545936645414e-06, "loss": 0.0002, "step": 187370 }, { "epoch": 3.0660230712591017, "grad_norm": 0.08399434387683868, "learning_rate": 1.5708349730970424e-06, "loss": 0.0004, "step": 187380 }, { "epoch": 3.0661866972101777, "grad_norm": 0.002823627321049571, "learning_rate": 1.570315422447302e-06, "loss": 0.0003, "step": 187390 }, { "epoch": 3.066350323161253, "grad_norm": 0.09763513505458832, "learning_rate": 1.5697959417259134e-06, "loss": 0.0003, "step": 187400 }, { "epoch": 3.066513949112329, "grad_norm": 0.05133707821369171, "learning_rate": 1.56927653094347e-06, "loss": 0.0005, "step": 187410 }, { "epoch": 3.066677575063405, "grad_norm": 0.16479890048503876, "learning_rate": 1.5687571901105586e-06, "loss": 0.0006, "step": 187420 }, { "epoch": 3.0668412010144808, "grad_norm": 0.11078565567731857, "learning_rate": 1.56823791923777e-06, "loss": 0.0004, "step": 187430 }, { "epoch": 3.0670048269655568, "grad_norm": 0.02430933713912964, "learning_rate": 1.5677187183356912e-06, "loss": 0.0003, "step": 187440 }, { "epoch": 3.0671684529166328, "grad_norm": 0.034524206072092056, "learning_rate": 1.567199587414911e-06, "loss": 0.0003, "step": 187450 }, { "epoch": 3.0673320788677083, "grad_norm": 0.007668725214898586, "learning_rate": 1.5666805264860096e-06, "loss": 0.0001, "step": 187460 }, { "epoch": 3.0674957048187843, "grad_norm": 0.021762922406196594, "learning_rate": 1.5661615355595727e-06, "loss": 0.0001, "step": 187470 }, { "epoch": 3.0676593307698603, "grad_norm": 0.03994714468717575, "learning_rate": 1.5656426146461812e-06, "loss": 0.0003, "step": 187480 }, { "epoch": 3.067822956720936, "grad_norm": 0.017149338498711586, "learning_rate": 1.5651237637564182e-06, "loss": 0.0001, "step": 187490 }, { "epoch": 3.067986582672012, "grad_norm": 0.08386045694351196, "learning_rate": 1.564604982900858e-06, "loss": 0.0002, "step": 187500 }, { "epoch": 3.068150208623088, "grad_norm": 0.022809727117419243, "learning_rate": 1.5640862720900813e-06, "loss": 0.0002, "step": 187510 }, { "epoch": 3.0683138345741634, "grad_norm": 0.013837720267474651, "learning_rate": 1.5635676313346627e-06, "loss": 0.0004, "step": 187520 }, { "epoch": 3.0684774605252394, "grad_norm": 0.016604851931333542, "learning_rate": 1.56304906064518e-06, "loss": 0.0003, "step": 187530 }, { "epoch": 3.068641086476315, "grad_norm": 0.04398950934410095, "learning_rate": 1.5625305600322016e-06, "loss": 0.0003, "step": 187540 }, { "epoch": 3.068804712427391, "grad_norm": 0.001985887996852398, "learning_rate": 1.5620121295063018e-06, "loss": 0.0001, "step": 187550 }, { "epoch": 3.068968338378467, "grad_norm": 0.018406879156827927, "learning_rate": 1.5614937690780506e-06, "loss": 0.0004, "step": 187560 }, { "epoch": 3.0691319643295425, "grad_norm": 0.0030764674302190542, "learning_rate": 1.5609754787580188e-06, "loss": 0.0004, "step": 187570 }, { "epoch": 3.0692955902806185, "grad_norm": 0.04540547356009483, "learning_rate": 1.5604572585567707e-06, "loss": 0.0004, "step": 187580 }, { "epoch": 3.0694592162316945, "grad_norm": 0.007227668073028326, "learning_rate": 1.5599391084848741e-06, "loss": 0.0002, "step": 187590 }, { "epoch": 3.06962284218277, "grad_norm": 0.048201534897089005, "learning_rate": 1.5594210285528932e-06, "loss": 0.0002, "step": 187600 }, { "epoch": 3.069786468133846, "grad_norm": 0.0024214230943471193, "learning_rate": 1.5589030187713926e-06, "loss": 0.0004, "step": 187610 }, { "epoch": 3.069950094084922, "grad_norm": 0.04573618993163109, "learning_rate": 1.558385079150932e-06, "loss": 0.0003, "step": 187620 }, { "epoch": 3.0701137200359976, "grad_norm": 0.026006462052464485, "learning_rate": 1.557867209702072e-06, "loss": 0.0003, "step": 187630 }, { "epoch": 3.0702773459870736, "grad_norm": 0.036974456161260605, "learning_rate": 1.5573494104353726e-06, "loss": 0.0004, "step": 187640 }, { "epoch": 3.0704409719381496, "grad_norm": 0.04101819172501564, "learning_rate": 1.5568316813613927e-06, "loss": 0.0002, "step": 187650 }, { "epoch": 3.070604597889225, "grad_norm": 0.04495404660701752, "learning_rate": 1.5563140224906841e-06, "loss": 0.0002, "step": 187660 }, { "epoch": 3.070768223840301, "grad_norm": 0.00022635332425124943, "learning_rate": 1.5557964338338045e-06, "loss": 0.0002, "step": 187670 }, { "epoch": 3.070931849791377, "grad_norm": 0.005922073498368263, "learning_rate": 1.555278915401306e-06, "loss": 0.0001, "step": 187680 }, { "epoch": 3.0710954757424527, "grad_norm": 0.013387538492679596, "learning_rate": 1.5547614672037426e-06, "loss": 0.0004, "step": 187690 }, { "epoch": 3.0712591016935287, "grad_norm": 0.008980381302535534, "learning_rate": 1.5542440892516614e-06, "loss": 0.0001, "step": 187700 }, { "epoch": 3.0714227276446042, "grad_norm": 0.0522870197892189, "learning_rate": 1.553726781555614e-06, "loss": 0.0004, "step": 187710 }, { "epoch": 3.07158635359568, "grad_norm": 0.0022261175327003, "learning_rate": 1.5532095441261437e-06, "loss": 0.0005, "step": 187720 }, { "epoch": 3.071749979546756, "grad_norm": 0.000369785149814561, "learning_rate": 1.5526923769738023e-06, "loss": 0.0006, "step": 187730 }, { "epoch": 3.0719136054978318, "grad_norm": 0.0029902851674705744, "learning_rate": 1.55217528010913e-06, "loss": 0.0002, "step": 187740 }, { "epoch": 3.0720772314489078, "grad_norm": 0.003858335316181183, "learning_rate": 1.5516582535426732e-06, "loss": 0.0004, "step": 187750 }, { "epoch": 3.0722408573999838, "grad_norm": 0.0013195836218073964, "learning_rate": 1.5511412972849682e-06, "loss": 0.0002, "step": 187760 }, { "epoch": 3.0724044833510593, "grad_norm": 0.011380596086382866, "learning_rate": 1.5506244113465629e-06, "loss": 0.0004, "step": 187770 }, { "epoch": 3.0725681093021353, "grad_norm": 0.005501982290297747, "learning_rate": 1.5501075957379903e-06, "loss": 0.0005, "step": 187780 }, { "epoch": 3.0727317352532113, "grad_norm": 0.03912767022848129, "learning_rate": 1.5495908504697916e-06, "loss": 0.0005, "step": 187790 }, { "epoch": 3.072895361204287, "grad_norm": 0.013769948855042458, "learning_rate": 1.5490741755524973e-06, "loss": 0.0001, "step": 187800 }, { "epoch": 3.073058987155363, "grad_norm": 0.03190129995346069, "learning_rate": 1.5485575709966494e-06, "loss": 0.0003, "step": 187810 }, { "epoch": 3.073222613106439, "grad_norm": 0.050647515803575516, "learning_rate": 1.5480410368127752e-06, "loss": 0.0008, "step": 187820 }, { "epoch": 3.0733862390575144, "grad_norm": 0.03577246516942978, "learning_rate": 1.5475245730114103e-06, "loss": 0.0002, "step": 187830 }, { "epoch": 3.0735498650085904, "grad_norm": 0.004351228941231966, "learning_rate": 1.54700817960308e-06, "loss": 0.0001, "step": 187840 }, { "epoch": 3.0737134909596664, "grad_norm": 0.023375818505883217, "learning_rate": 1.5464918565983195e-06, "loss": 0.0005, "step": 187850 }, { "epoch": 3.073877116910742, "grad_norm": 0.014678320847451687, "learning_rate": 1.5459756040076512e-06, "loss": 0.0005, "step": 187860 }, { "epoch": 3.074040742861818, "grad_norm": 0.011711658909916878, "learning_rate": 1.5454594218416047e-06, "loss": 0.0001, "step": 187870 }, { "epoch": 3.074204368812894, "grad_norm": 0.025790518149733543, "learning_rate": 1.5449433101107002e-06, "loss": 0.0008, "step": 187880 }, { "epoch": 3.0743679947639695, "grad_norm": 0.004444895777851343, "learning_rate": 1.5444272688254641e-06, "loss": 0.0001, "step": 187890 }, { "epoch": 3.0745316207150455, "grad_norm": 0.10951781272888184, "learning_rate": 1.5439112979964167e-06, "loss": 0.0003, "step": 187900 }, { "epoch": 3.074695246666121, "grad_norm": 0.030739106237888336, "learning_rate": 1.5433953976340804e-06, "loss": 0.0003, "step": 187910 }, { "epoch": 3.074858872617197, "grad_norm": 0.04929833114147186, "learning_rate": 1.5428795677489704e-06, "loss": 0.0003, "step": 187920 }, { "epoch": 3.075022498568273, "grad_norm": 0.014913874678313732, "learning_rate": 1.5423638083516058e-06, "loss": 0.0002, "step": 187930 }, { "epoch": 3.0751861245193486, "grad_norm": 0.054998092353343964, "learning_rate": 1.5418481194525032e-06, "loss": 0.0004, "step": 187940 }, { "epoch": 3.0753497504704246, "grad_norm": 0.007119542919099331, "learning_rate": 1.5413325010621771e-06, "loss": 0.0003, "step": 187950 }, { "epoch": 3.0755133764215006, "grad_norm": 0.0021854666993021965, "learning_rate": 1.5408169531911388e-06, "loss": 0.0003, "step": 187960 }, { "epoch": 3.075677002372576, "grad_norm": 0.0059032379649579525, "learning_rate": 1.5403014758499008e-06, "loss": 0.0002, "step": 187970 }, { "epoch": 3.075840628323652, "grad_norm": 0.014301236718893051, "learning_rate": 1.539786069048973e-06, "loss": 0.0005, "step": 187980 }, { "epoch": 3.076004254274728, "grad_norm": 0.0024868359323590994, "learning_rate": 1.539270732798866e-06, "loss": 0.0004, "step": 187990 }, { "epoch": 3.0761678802258037, "grad_norm": 0.0018274380126968026, "learning_rate": 1.5387554671100841e-06, "loss": 0.0048, "step": 188000 }, { "epoch": 3.0763315061768797, "grad_norm": 0.03038865141570568, "learning_rate": 1.538240271993135e-06, "loss": 0.0003, "step": 188010 }, { "epoch": 3.0764951321279557, "grad_norm": 0.0017103629652410746, "learning_rate": 1.537725147458522e-06, "loss": 0.0006, "step": 188020 }, { "epoch": 3.076658758079031, "grad_norm": 0.0034383065067231655, "learning_rate": 1.5372100935167506e-06, "loss": 0.0004, "step": 188030 }, { "epoch": 3.076822384030107, "grad_norm": 0.039770741015672684, "learning_rate": 1.536695110178319e-06, "loss": 0.0004, "step": 188040 }, { "epoch": 3.076986009981183, "grad_norm": 0.009303991682827473, "learning_rate": 1.5361801974537283e-06, "loss": 0.0003, "step": 188050 }, { "epoch": 3.0771496359322588, "grad_norm": 0.041166797280311584, "learning_rate": 1.5356653553534772e-06, "loss": 0.0003, "step": 188060 }, { "epoch": 3.0773132618833348, "grad_norm": 0.022207865491509438, "learning_rate": 1.5351505838880655e-06, "loss": 0.0004, "step": 188070 }, { "epoch": 3.0774768878344103, "grad_norm": 0.03819059208035469, "learning_rate": 1.5346358830679847e-06, "loss": 0.0003, "step": 188080 }, { "epoch": 3.0776405137854863, "grad_norm": 0.026005003601312637, "learning_rate": 1.5341212529037309e-06, "loss": 0.0004, "step": 188090 }, { "epoch": 3.0778041397365623, "grad_norm": 0.03643222525715828, "learning_rate": 1.5336066934057974e-06, "loss": 0.0004, "step": 188100 }, { "epoch": 3.077967765687638, "grad_norm": 0.019288314506411552, "learning_rate": 1.5330922045846769e-06, "loss": 0.0003, "step": 188110 }, { "epoch": 3.078131391638714, "grad_norm": 0.05043886974453926, "learning_rate": 1.5325777864508562e-06, "loss": 0.0003, "step": 188120 }, { "epoch": 3.07829501758979, "grad_norm": 0.026467742398381233, "learning_rate": 1.532063439014826e-06, "loss": 0.0002, "step": 188130 }, { "epoch": 3.0784586435408654, "grad_norm": 0.0016745883040130138, "learning_rate": 1.531549162287072e-06, "loss": 0.0002, "step": 188140 }, { "epoch": 3.0786222694919414, "grad_norm": 0.002376996912062168, "learning_rate": 1.531034956278083e-06, "loss": 0.0002, "step": 188150 }, { "epoch": 3.0787858954430174, "grad_norm": 0.001053879503160715, "learning_rate": 1.530520820998339e-06, "loss": 0.0005, "step": 188160 }, { "epoch": 3.078949521394093, "grad_norm": 0.01560696866363287, "learning_rate": 1.5300067564583248e-06, "loss": 0.0004, "step": 188170 }, { "epoch": 3.079113147345169, "grad_norm": 0.027497781440615654, "learning_rate": 1.529492762668522e-06, "loss": 0.0002, "step": 188180 }, { "epoch": 3.079276773296245, "grad_norm": 0.07529104501008987, "learning_rate": 1.5289788396394117e-06, "loss": 0.0005, "step": 188190 }, { "epoch": 3.0794403992473205, "grad_norm": 0.011189579032361507, "learning_rate": 1.5284649873814693e-06, "loss": 0.0002, "step": 188200 }, { "epoch": 3.0796040251983965, "grad_norm": 0.1468864530324936, "learning_rate": 1.5279512059051732e-06, "loss": 0.0004, "step": 188210 }, { "epoch": 3.0797676511494725, "grad_norm": 0.01951281912624836, "learning_rate": 1.5274374952209997e-06, "loss": 0.0001, "step": 188220 }, { "epoch": 3.079931277100548, "grad_norm": 0.016082381829619408, "learning_rate": 1.5269238553394234e-06, "loss": 0.0003, "step": 188230 }, { "epoch": 3.080094903051624, "grad_norm": 0.1572897583246231, "learning_rate": 1.5264102862709145e-06, "loss": 0.0003, "step": 188240 }, { "epoch": 3.0802585290027, "grad_norm": 0.016943862661719322, "learning_rate": 1.5258967880259458e-06, "loss": 0.0004, "step": 188250 }, { "epoch": 3.0804221549537756, "grad_norm": 0.03676280751824379, "learning_rate": 1.5253833606149864e-06, "loss": 0.0001, "step": 188260 }, { "epoch": 3.0805857809048516, "grad_norm": 0.0005758671322837472, "learning_rate": 1.5248700040485077e-06, "loss": 0.0004, "step": 188270 }, { "epoch": 3.0807494068559276, "grad_norm": 0.0024634895380586386, "learning_rate": 1.5243567183369718e-06, "loss": 0.0003, "step": 188280 }, { "epoch": 3.080913032807003, "grad_norm": 0.01795320026576519, "learning_rate": 1.5238435034908466e-06, "loss": 0.0002, "step": 188290 }, { "epoch": 3.081076658758079, "grad_norm": 0.029296888038516045, "learning_rate": 1.5233303595205962e-06, "loss": 0.0004, "step": 188300 }, { "epoch": 3.0812402847091547, "grad_norm": 0.0041534085758030415, "learning_rate": 1.5228172864366847e-06, "loss": 0.0005, "step": 188310 }, { "epoch": 3.0814039106602307, "grad_norm": 0.00983771774917841, "learning_rate": 1.5223042842495695e-06, "loss": 0.0005, "step": 188320 }, { "epoch": 3.0815675366113067, "grad_norm": 0.025259394198656082, "learning_rate": 1.5217913529697125e-06, "loss": 0.0003, "step": 188330 }, { "epoch": 3.081731162562382, "grad_norm": 0.00023102560953702778, "learning_rate": 1.521278492607572e-06, "loss": 0.0002, "step": 188340 }, { "epoch": 3.081894788513458, "grad_norm": 0.016453292220830917, "learning_rate": 1.5207657031736057e-06, "loss": 0.0002, "step": 188350 }, { "epoch": 3.082058414464534, "grad_norm": 0.03715888783335686, "learning_rate": 1.520252984678266e-06, "loss": 0.0003, "step": 188360 }, { "epoch": 3.0822220404156098, "grad_norm": 0.013069459237158298, "learning_rate": 1.5197403371320102e-06, "loss": 0.0002, "step": 188370 }, { "epoch": 3.0823856663666858, "grad_norm": 0.008407861925661564, "learning_rate": 1.519227760545286e-06, "loss": 0.0001, "step": 188380 }, { "epoch": 3.0825492923177618, "grad_norm": 0.027416203171014786, "learning_rate": 1.518715254928551e-06, "loss": 0.0006, "step": 188390 }, { "epoch": 3.0827129182688373, "grad_norm": 0.023709310218691826, "learning_rate": 1.5182028202922494e-06, "loss": 0.0005, "step": 188400 }, { "epoch": 3.0828765442199133, "grad_norm": 0.060317084193229675, "learning_rate": 1.5176904566468325e-06, "loss": 0.0002, "step": 188410 }, { "epoch": 3.0830401701709893, "grad_norm": 0.017259834334254265, "learning_rate": 1.517178164002742e-06, "loss": 0.0002, "step": 188420 }, { "epoch": 3.083203796122065, "grad_norm": 0.015775315463542938, "learning_rate": 1.5166659423704312e-06, "loss": 0.0008, "step": 188430 }, { "epoch": 3.083367422073141, "grad_norm": 0.019187698140740395, "learning_rate": 1.5161537917603363e-06, "loss": 0.0001, "step": 188440 }, { "epoch": 3.083531048024217, "grad_norm": 0.009787016548216343, "learning_rate": 1.5156417121829054e-06, "loss": 0.0003, "step": 188450 }, { "epoch": 3.0836946739752924, "grad_norm": 0.006419474259018898, "learning_rate": 1.5151297036485729e-06, "loss": 0.0002, "step": 188460 }, { "epoch": 3.0838582999263684, "grad_norm": 0.09563368558883667, "learning_rate": 1.5146177661677853e-06, "loss": 0.0003, "step": 188470 }, { "epoch": 3.084021925877444, "grad_norm": 0.0011214668629691005, "learning_rate": 1.5141058997509755e-06, "loss": 0.0009, "step": 188480 }, { "epoch": 3.08418555182852, "grad_norm": 0.0006974576972424984, "learning_rate": 1.5135941044085833e-06, "loss": 0.0003, "step": 188490 }, { "epoch": 3.084349177779596, "grad_norm": 0.024300867691636086, "learning_rate": 1.5130823801510397e-06, "loss": 0.0004, "step": 188500 }, { "epoch": 3.0845128037306715, "grad_norm": 0.005391149781644344, "learning_rate": 1.5125707269887835e-06, "loss": 0.0005, "step": 188510 }, { "epoch": 3.0846764296817475, "grad_norm": 0.047676585614681244, "learning_rate": 1.5120591449322425e-06, "loss": 0.0003, "step": 188520 }, { "epoch": 3.0848400556328235, "grad_norm": 0.06524380296468735, "learning_rate": 1.5115476339918512e-06, "loss": 0.0006, "step": 188530 }, { "epoch": 3.085003681583899, "grad_norm": 0.012526151724159718, "learning_rate": 1.511036194178035e-06, "loss": 0.0005, "step": 188540 }, { "epoch": 3.085167307534975, "grad_norm": 0.02944774180650711, "learning_rate": 1.5105248255012234e-06, "loss": 0.0003, "step": 188550 }, { "epoch": 3.085330933486051, "grad_norm": 0.00045827540452592075, "learning_rate": 1.510013527971843e-06, "loss": 0.0005, "step": 188560 }, { "epoch": 3.0854945594371266, "grad_norm": 0.013629673980176449, "learning_rate": 1.50950230160032e-06, "loss": 0.0002, "step": 188570 }, { "epoch": 3.0856581853882026, "grad_norm": 0.012202304787933826, "learning_rate": 1.5089911463970752e-06, "loss": 0.0004, "step": 188580 }, { "epoch": 3.0858218113392786, "grad_norm": 0.07069150358438492, "learning_rate": 1.508480062372532e-06, "loss": 0.0004, "step": 188590 }, { "epoch": 3.085985437290354, "grad_norm": 0.0017315729055553675, "learning_rate": 1.5079690495371109e-06, "loss": 0.0002, "step": 188600 }, { "epoch": 3.08614906324143, "grad_norm": 0.07403362542390823, "learning_rate": 1.5074581079012324e-06, "loss": 0.001, "step": 188610 }, { "epoch": 3.086312689192506, "grad_norm": 0.03357670083642006, "learning_rate": 1.5069472374753113e-06, "loss": 0.0004, "step": 188620 }, { "epoch": 3.0864763151435817, "grad_norm": 0.003400350222364068, "learning_rate": 1.5064364382697649e-06, "loss": 0.0003, "step": 188630 }, { "epoch": 3.0866399410946577, "grad_norm": 0.01984255388379097, "learning_rate": 1.5059257102950092e-06, "loss": 0.0003, "step": 188640 }, { "epoch": 3.0868035670457337, "grad_norm": 0.003431657562032342, "learning_rate": 1.5054150535614576e-06, "loss": 0.0003, "step": 188650 }, { "epoch": 3.086967192996809, "grad_norm": 0.014608429744839668, "learning_rate": 1.5049044680795193e-06, "loss": 0.0003, "step": 188660 }, { "epoch": 3.087130818947885, "grad_norm": 0.013040738180279732, "learning_rate": 1.5043939538596069e-06, "loss": 0.0002, "step": 188670 }, { "epoch": 3.0872944448989608, "grad_norm": 0.02929580770432949, "learning_rate": 1.5038835109121285e-06, "loss": 0.0001, "step": 188680 }, { "epoch": 3.0874580708500368, "grad_norm": 0.0041139922104775906, "learning_rate": 1.5033731392474943e-06, "loss": 0.0005, "step": 188690 }, { "epoch": 3.0876216968011128, "grad_norm": 0.008059495128691196, "learning_rate": 1.5028628388761058e-06, "loss": 0.0002, "step": 188700 }, { "epoch": 3.0877853227521883, "grad_norm": 0.00018271016597282141, "learning_rate": 1.5023526098083696e-06, "loss": 0.0001, "step": 188710 }, { "epoch": 3.0879489487032643, "grad_norm": 0.007643700577318668, "learning_rate": 1.5018424520546893e-06, "loss": 0.0002, "step": 188720 }, { "epoch": 3.0881125746543403, "grad_norm": 0.022682998329401016, "learning_rate": 1.501332365625468e-06, "loss": 0.0002, "step": 188730 }, { "epoch": 3.088276200605416, "grad_norm": 0.005754530429840088, "learning_rate": 1.5008223505311026e-06, "loss": 0.0002, "step": 188740 }, { "epoch": 3.088439826556492, "grad_norm": 0.0010638464009389281, "learning_rate": 1.5003124067819935e-06, "loss": 0.0004, "step": 188750 }, { "epoch": 3.088603452507568, "grad_norm": 0.005244428291916847, "learning_rate": 1.4998025343885386e-06, "loss": 0.0003, "step": 188760 }, { "epoch": 3.0887670784586434, "grad_norm": 0.006176062859594822, "learning_rate": 1.499292733361134e-06, "loss": 0.0003, "step": 188770 }, { "epoch": 3.0889307044097194, "grad_norm": 0.0434490442276001, "learning_rate": 1.498783003710172e-06, "loss": 0.0004, "step": 188780 }, { "epoch": 3.0890943303607954, "grad_norm": 0.0015119794989004731, "learning_rate": 1.4982733454460468e-06, "loss": 0.0003, "step": 188790 }, { "epoch": 3.089257956311871, "grad_norm": 0.06503486633300781, "learning_rate": 1.4977637585791504e-06, "loss": 0.0001, "step": 188800 }, { "epoch": 3.089421582262947, "grad_norm": 0.011304088868200779, "learning_rate": 1.4972542431198733e-06, "loss": 0.0006, "step": 188810 }, { "epoch": 3.089585208214023, "grad_norm": 0.007449423428624868, "learning_rate": 1.496744799078602e-06, "loss": 0.0001, "step": 188820 }, { "epoch": 3.0897488341650985, "grad_norm": 0.001814432442188263, "learning_rate": 1.4962354264657253e-06, "loss": 0.0002, "step": 188830 }, { "epoch": 3.0899124601161745, "grad_norm": 0.0015922404127195477, "learning_rate": 1.4957261252916277e-06, "loss": 0.0005, "step": 188840 }, { "epoch": 3.09007608606725, "grad_norm": 0.084598109126091, "learning_rate": 1.4952168955666963e-06, "loss": 0.0004, "step": 188850 }, { "epoch": 3.090239712018326, "grad_norm": 0.022702239453792572, "learning_rate": 1.4947077373013096e-06, "loss": 0.0003, "step": 188860 }, { "epoch": 3.090403337969402, "grad_norm": 0.04253275692462921, "learning_rate": 1.4941986505058508e-06, "loss": 0.0001, "step": 188870 }, { "epoch": 3.0905669639204776, "grad_norm": 0.008113752119243145, "learning_rate": 1.4936896351907003e-06, "loss": 0.0002, "step": 188880 }, { "epoch": 3.0907305898715536, "grad_norm": 0.008648070506751537, "learning_rate": 1.4931806913662379e-06, "loss": 0.0003, "step": 188890 }, { "epoch": 3.0908942158226296, "grad_norm": 0.0038115368224680424, "learning_rate": 1.4926718190428369e-06, "loss": 0.0003, "step": 188900 }, { "epoch": 3.091057841773705, "grad_norm": 0.051994841545820236, "learning_rate": 1.492163018230875e-06, "loss": 0.0002, "step": 188910 }, { "epoch": 3.091221467724781, "grad_norm": 0.07846017926931381, "learning_rate": 1.4916542889407254e-06, "loss": 0.0003, "step": 188920 }, { "epoch": 3.091385093675857, "grad_norm": 0.00979782734066248, "learning_rate": 1.4911456311827626e-06, "loss": 0.0004, "step": 188930 }, { "epoch": 3.0915487196269327, "grad_norm": 0.07374344766139984, "learning_rate": 1.4906370449673551e-06, "loss": 0.0005, "step": 188940 }, { "epoch": 3.0917123455780087, "grad_norm": 0.018797950819134712, "learning_rate": 1.4901285303048735e-06, "loss": 0.0004, "step": 188950 }, { "epoch": 3.0918759715290847, "grad_norm": 0.008626874536275864, "learning_rate": 1.4896200872056859e-06, "loss": 0.0003, "step": 188960 }, { "epoch": 3.09203959748016, "grad_norm": 0.05445713549852371, "learning_rate": 1.489111715680161e-06, "loss": 0.0003, "step": 188970 }, { "epoch": 3.092203223431236, "grad_norm": 0.004080787766724825, "learning_rate": 1.4886034157386608e-06, "loss": 0.0006, "step": 188980 }, { "epoch": 3.092366849382312, "grad_norm": 0.004920313600450754, "learning_rate": 1.4880951873915522e-06, "loss": 0.0004, "step": 188990 }, { "epoch": 3.0925304753333878, "grad_norm": 0.012503483332693577, "learning_rate": 1.4875870306491929e-06, "loss": 0.0004, "step": 189000 }, { "epoch": 3.0926941012844638, "grad_norm": 0.023700527846813202, "learning_rate": 1.4870789455219498e-06, "loss": 0.0004, "step": 189010 }, { "epoch": 3.0928577272355398, "grad_norm": 0.00048492124187760055, "learning_rate": 1.4865709320201777e-06, "loss": 0.0006, "step": 189020 }, { "epoch": 3.0930213531866153, "grad_norm": 0.03010720945894718, "learning_rate": 1.4860629901542378e-06, "loss": 0.0004, "step": 189030 }, { "epoch": 3.0931849791376913, "grad_norm": 0.030544867739081383, "learning_rate": 1.485555119934482e-06, "loss": 0.0003, "step": 189040 }, { "epoch": 3.0933486050887673, "grad_norm": 0.04772606119513512, "learning_rate": 1.485047321371272e-06, "loss": 0.0003, "step": 189050 }, { "epoch": 3.093512231039843, "grad_norm": 0.016651378944516182, "learning_rate": 1.4845395944749553e-06, "loss": 0.0003, "step": 189060 }, { "epoch": 3.093675856990919, "grad_norm": 0.019116973504424095, "learning_rate": 1.4840319392558889e-06, "loss": 0.0002, "step": 189070 }, { "epoch": 3.0938394829419944, "grad_norm": 0.17816588282585144, "learning_rate": 1.4835243557244188e-06, "loss": 0.0012, "step": 189080 }, { "epoch": 3.0940031088930704, "grad_norm": 0.010667137801647186, "learning_rate": 1.4830168438908966e-06, "loss": 0.0004, "step": 189090 }, { "epoch": 3.0941667348441464, "grad_norm": 0.02649802342057228, "learning_rate": 1.4825094037656717e-06, "loss": 0.0001, "step": 189100 }, { "epoch": 3.094330360795222, "grad_norm": 0.05360260233283043, "learning_rate": 1.4820020353590863e-06, "loss": 0.0005, "step": 189110 }, { "epoch": 3.094493986746298, "grad_norm": 0.02856898121535778, "learning_rate": 1.481494738681488e-06, "loss": 0.0006, "step": 189120 }, { "epoch": 3.094657612697374, "grad_norm": 0.017062151804566383, "learning_rate": 1.4809875137432195e-06, "loss": 0.0002, "step": 189130 }, { "epoch": 3.0948212386484495, "grad_norm": 0.03496261686086655, "learning_rate": 1.4804803605546246e-06, "loss": 0.0004, "step": 189140 }, { "epoch": 3.0949848645995255, "grad_norm": 0.0005873999907635152, "learning_rate": 1.4799732791260402e-06, "loss": 0.0004, "step": 189150 }, { "epoch": 3.0951484905506015, "grad_norm": 0.04100077971816063, "learning_rate": 1.4794662694678069e-06, "loss": 0.0002, "step": 189160 }, { "epoch": 3.095312116501677, "grad_norm": 0.0024233972653746605, "learning_rate": 1.4789593315902622e-06, "loss": 0.0005, "step": 189170 }, { "epoch": 3.095475742452753, "grad_norm": 0.034085676074028015, "learning_rate": 1.4784524655037436e-06, "loss": 0.0004, "step": 189180 }, { "epoch": 3.095639368403829, "grad_norm": 0.03274950385093689, "learning_rate": 1.477945671218583e-06, "loss": 0.0003, "step": 189190 }, { "epoch": 3.0958029943549046, "grad_norm": 0.011120334267616272, "learning_rate": 1.4774389487451146e-06, "loss": 0.0003, "step": 189200 }, { "epoch": 3.0959666203059806, "grad_norm": 0.02045561745762825, "learning_rate": 1.47693229809367e-06, "loss": 0.0004, "step": 189210 }, { "epoch": 3.0961302462570566, "grad_norm": 0.04895121231675148, "learning_rate": 1.4764257192745818e-06, "loss": 0.0002, "step": 189220 }, { "epoch": 3.096293872208132, "grad_norm": 0.03872358053922653, "learning_rate": 1.4759192122981742e-06, "loss": 0.0003, "step": 189230 }, { "epoch": 3.096457498159208, "grad_norm": 0.003324157791212201, "learning_rate": 1.4754127771747772e-06, "loss": 0.0001, "step": 189240 }, { "epoch": 3.0966211241102837, "grad_norm": 0.011737755499780178, "learning_rate": 1.4749064139147157e-06, "loss": 0.0001, "step": 189250 }, { "epoch": 3.0967847500613597, "grad_norm": 0.04431100934743881, "learning_rate": 1.4744001225283167e-06, "loss": 0.0003, "step": 189260 }, { "epoch": 3.0969483760124357, "grad_norm": 0.0025734654627740383, "learning_rate": 1.4738939030258987e-06, "loss": 0.0002, "step": 189270 }, { "epoch": 3.097112001963511, "grad_norm": 0.001283534336835146, "learning_rate": 1.473387755417785e-06, "loss": 0.0002, "step": 189280 }, { "epoch": 3.097275627914587, "grad_norm": 0.0010763658210635185, "learning_rate": 1.472881679714296e-06, "loss": 0.0001, "step": 189290 }, { "epoch": 3.097439253865663, "grad_norm": 9.721909736981615e-05, "learning_rate": 1.4723756759257514e-06, "loss": 0.0002, "step": 189300 }, { "epoch": 3.0976028798167388, "grad_norm": 0.01696242019534111, "learning_rate": 1.4718697440624653e-06, "loss": 0.0006, "step": 189310 }, { "epoch": 3.0977665057678148, "grad_norm": 0.010958433151245117, "learning_rate": 1.471363884134754e-06, "loss": 0.0002, "step": 189320 }, { "epoch": 3.0979301317188908, "grad_norm": 0.027314692735671997, "learning_rate": 1.4708580961529322e-06, "loss": 0.0003, "step": 189330 }, { "epoch": 3.0980937576699663, "grad_norm": 0.0007442903588525951, "learning_rate": 1.4703523801273145e-06, "loss": 0.0003, "step": 189340 }, { "epoch": 3.0982573836210423, "grad_norm": 0.03230316936969757, "learning_rate": 1.4698467360682078e-06, "loss": 0.0005, "step": 189350 }, { "epoch": 3.0984210095721183, "grad_norm": 0.0006476756534539163, "learning_rate": 1.4693411639859239e-06, "loss": 0.0003, "step": 189360 }, { "epoch": 3.098584635523194, "grad_norm": 0.01956617832183838, "learning_rate": 1.4688356638907713e-06, "loss": 0.0005, "step": 189370 }, { "epoch": 3.09874826147427, "grad_norm": 0.051718637347221375, "learning_rate": 1.468330235793058e-06, "loss": 0.0009, "step": 189380 }, { "epoch": 3.098911887425346, "grad_norm": 0.1372860074043274, "learning_rate": 1.4678248797030853e-06, "loss": 0.0003, "step": 189390 }, { "epoch": 3.0990755133764214, "grad_norm": 0.02708788774907589, "learning_rate": 1.4673195956311598e-06, "loss": 0.0003, "step": 189400 }, { "epoch": 3.0992391393274974, "grad_norm": 0.1079825907945633, "learning_rate": 1.4668143835875836e-06, "loss": 0.0003, "step": 189410 }, { "epoch": 3.0994027652785734, "grad_norm": 0.008664936758577824, "learning_rate": 1.4663092435826587e-06, "loss": 0.0003, "step": 189420 }, { "epoch": 3.099566391229649, "grad_norm": 0.04290173202753067, "learning_rate": 1.4658041756266812e-06, "loss": 0.0002, "step": 189430 }, { "epoch": 3.099730017180725, "grad_norm": 0.010372761636972427, "learning_rate": 1.465299179729951e-06, "loss": 0.0002, "step": 189440 }, { "epoch": 3.0998936431318005, "grad_norm": 0.004892629571259022, "learning_rate": 1.4647942559027639e-06, "loss": 0.0003, "step": 189450 }, { "epoch": 3.1000572690828765, "grad_norm": 0.042012039572000504, "learning_rate": 1.4642894041554174e-06, "loss": 0.0005, "step": 189460 }, { "epoch": 3.1002208950339525, "grad_norm": 0.029734889045357704, "learning_rate": 1.4637846244982013e-06, "loss": 0.0002, "step": 189470 }, { "epoch": 3.100384520985028, "grad_norm": 0.025362513959407806, "learning_rate": 1.4632799169414109e-06, "loss": 0.0005, "step": 189480 }, { "epoch": 3.100548146936104, "grad_norm": 0.0015971052926033735, "learning_rate": 1.4627752814953317e-06, "loss": 0.0004, "step": 189490 }, { "epoch": 3.10071177288718, "grad_norm": 0.040857944637537, "learning_rate": 1.4622707181702594e-06, "loss": 0.0007, "step": 189500 }, { "epoch": 3.1008753988382556, "grad_norm": 0.06972144544124603, "learning_rate": 1.4617662269764772e-06, "loss": 0.0004, "step": 189510 }, { "epoch": 3.1010390247893316, "grad_norm": 0.024925440549850464, "learning_rate": 1.4612618079242736e-06, "loss": 0.0002, "step": 189520 }, { "epoch": 3.1012026507404076, "grad_norm": 0.0022797207348048687, "learning_rate": 1.460757461023929e-06, "loss": 0.0003, "step": 189530 }, { "epoch": 3.101366276691483, "grad_norm": 0.012797131203114986, "learning_rate": 1.4602531862857333e-06, "loss": 0.0005, "step": 189540 }, { "epoch": 3.101529902642559, "grad_norm": 0.02414371818304062, "learning_rate": 1.459748983719963e-06, "loss": 0.0002, "step": 189550 }, { "epoch": 3.101693528593635, "grad_norm": 0.09413725137710571, "learning_rate": 1.4592448533369013e-06, "loss": 0.0005, "step": 189560 }, { "epoch": 3.1018571545447107, "grad_norm": 0.01488436572253704, "learning_rate": 1.458740795146823e-06, "loss": 0.0001, "step": 189570 }, { "epoch": 3.1020207804957867, "grad_norm": 0.00039998407009989023, "learning_rate": 1.4582368091600107e-06, "loss": 0.0002, "step": 189580 }, { "epoch": 3.1021844064468627, "grad_norm": 0.035259947180747986, "learning_rate": 1.457732895386737e-06, "loss": 0.0001, "step": 189590 }, { "epoch": 3.102348032397938, "grad_norm": 0.023265674710273743, "learning_rate": 1.457229053837278e-06, "loss": 0.0001, "step": 189600 }, { "epoch": 3.102511658349014, "grad_norm": 0.05503314360976219, "learning_rate": 1.4567252845219027e-06, "loss": 0.0002, "step": 189610 }, { "epoch": 3.10267528430009, "grad_norm": 0.0006573201389983296, "learning_rate": 1.4562215874508883e-06, "loss": 0.0001, "step": 189620 }, { "epoch": 3.1028389102511658, "grad_norm": 0.04276279732584953, "learning_rate": 1.4557179626345003e-06, "loss": 0.0002, "step": 189630 }, { "epoch": 3.1030025362022418, "grad_norm": 0.06525147706270218, "learning_rate": 1.45521441008301e-06, "loss": 0.0005, "step": 189640 }, { "epoch": 3.1031661621533173, "grad_norm": 0.010638206265866756, "learning_rate": 1.454710929806682e-06, "loss": 0.0003, "step": 189650 }, { "epoch": 3.1033297881043933, "grad_norm": 0.04952049255371094, "learning_rate": 1.454207521815783e-06, "loss": 0.0002, "step": 189660 }, { "epoch": 3.1034934140554693, "grad_norm": 0.012632766738533974, "learning_rate": 1.453704186120577e-06, "loss": 0.0002, "step": 189670 }, { "epoch": 3.103657040006545, "grad_norm": 0.07126107811927795, "learning_rate": 1.4532009227313281e-06, "loss": 0.0005, "step": 189680 }, { "epoch": 3.103820665957621, "grad_norm": 0.023879777640104294, "learning_rate": 1.4526977316582947e-06, "loss": 0.0003, "step": 189690 }, { "epoch": 3.103984291908697, "grad_norm": 0.02795354276895523, "learning_rate": 1.452194612911737e-06, "loss": 0.0002, "step": 189700 }, { "epoch": 3.1041479178597724, "grad_norm": 0.02192198485136032, "learning_rate": 1.4516915665019144e-06, "loss": 0.0003, "step": 189710 }, { "epoch": 3.1043115438108484, "grad_norm": 0.0459250770509243, "learning_rate": 1.4511885924390844e-06, "loss": 0.0005, "step": 189720 }, { "epoch": 3.1044751697619244, "grad_norm": 0.02678990550339222, "learning_rate": 1.4506856907334993e-06, "loss": 0.0002, "step": 189730 }, { "epoch": 3.104638795713, "grad_norm": 0.01359642669558525, "learning_rate": 1.4501828613954143e-06, "loss": 0.0005, "step": 189740 }, { "epoch": 3.104802421664076, "grad_norm": 0.01162472553551197, "learning_rate": 1.4496801044350822e-06, "loss": 0.0004, "step": 189750 }, { "epoch": 3.104966047615152, "grad_norm": 0.006207987200468779, "learning_rate": 1.4491774198627545e-06, "loss": 0.0002, "step": 189760 }, { "epoch": 3.1051296735662275, "grad_norm": 0.0010028529213741422, "learning_rate": 1.448674807688678e-06, "loss": 0.0001, "step": 189770 }, { "epoch": 3.1052932995173035, "grad_norm": 0.01972210220992565, "learning_rate": 1.4481722679231014e-06, "loss": 0.0004, "step": 189780 }, { "epoch": 3.1054569254683795, "grad_norm": 0.038271836936473846, "learning_rate": 1.4476698005762717e-06, "loss": 0.0004, "step": 189790 }, { "epoch": 3.105620551419455, "grad_norm": 0.024549441412091255, "learning_rate": 1.4471674056584357e-06, "loss": 0.0003, "step": 189800 }, { "epoch": 3.105784177370531, "grad_norm": 0.04639993607997894, "learning_rate": 1.4466650831798329e-06, "loss": 0.0004, "step": 189810 }, { "epoch": 3.105947803321607, "grad_norm": 0.04032577946782112, "learning_rate": 1.4461628331507066e-06, "loss": 0.0003, "step": 189820 }, { "epoch": 3.1061114292726826, "grad_norm": 0.00043461550376378, "learning_rate": 1.445660655581298e-06, "loss": 0.0004, "step": 189830 }, { "epoch": 3.1062750552237586, "grad_norm": 0.05281398817896843, "learning_rate": 1.4451585504818478e-06, "loss": 0.0004, "step": 189840 }, { "epoch": 3.106438681174834, "grad_norm": 0.00020012175082229078, "learning_rate": 1.4446565178625894e-06, "loss": 0.0006, "step": 189850 }, { "epoch": 3.10660230712591, "grad_norm": 0.005299615208059549, "learning_rate": 1.4441545577337606e-06, "loss": 0.0003, "step": 189860 }, { "epoch": 3.106765933076986, "grad_norm": 0.013311665505170822, "learning_rate": 1.443652670105597e-06, "loss": 0.0004, "step": 189870 }, { "epoch": 3.1069295590280617, "grad_norm": 0.034517329186201096, "learning_rate": 1.4431508549883316e-06, "loss": 0.0004, "step": 189880 }, { "epoch": 3.1070931849791377, "grad_norm": 0.016520580276846886, "learning_rate": 1.4426491123921943e-06, "loss": 0.0005, "step": 189890 }, { "epoch": 3.1072568109302137, "grad_norm": 0.0755135789513588, "learning_rate": 1.4421474423274162e-06, "loss": 0.0004, "step": 189900 }, { "epoch": 3.107420436881289, "grad_norm": 0.008101998828351498, "learning_rate": 1.441645844804226e-06, "loss": 0.0002, "step": 189910 }, { "epoch": 3.107584062832365, "grad_norm": 0.057240989059209824, "learning_rate": 1.4411443198328517e-06, "loss": 0.0004, "step": 189920 }, { "epoch": 3.107747688783441, "grad_norm": 0.01590724289417267, "learning_rate": 1.4406428674235169e-06, "loss": 0.0006, "step": 189930 }, { "epoch": 3.1079113147345168, "grad_norm": 0.02123870514333248, "learning_rate": 1.4401414875864467e-06, "loss": 0.0003, "step": 189940 }, { "epoch": 3.1080749406855928, "grad_norm": 0.006088252644985914, "learning_rate": 1.4396401803318643e-06, "loss": 0.0006, "step": 189950 }, { "epoch": 3.1082385666366688, "grad_norm": 0.046320393681526184, "learning_rate": 1.4391389456699923e-06, "loss": 0.0002, "step": 189960 }, { "epoch": 3.1084021925877443, "grad_norm": 0.011118460446596146, "learning_rate": 1.4386377836110466e-06, "loss": 0.0002, "step": 189970 }, { "epoch": 3.1085658185388203, "grad_norm": 0.018183521926403046, "learning_rate": 1.438136694165248e-06, "loss": 0.0006, "step": 189980 }, { "epoch": 3.1087294444898963, "grad_norm": 0.04064752161502838, "learning_rate": 1.437635677342813e-06, "loss": 0.0001, "step": 189990 }, { "epoch": 3.108893070440972, "grad_norm": 0.010870019905269146, "learning_rate": 1.4371347331539587e-06, "loss": 0.0005, "step": 190000 }, { "epoch": 3.109056696392048, "grad_norm": 0.02792048640549183, "learning_rate": 1.4366338616088948e-06, "loss": 0.0003, "step": 190010 }, { "epoch": 3.1092203223431234, "grad_norm": 0.008608084172010422, "learning_rate": 1.4361330627178367e-06, "loss": 0.0003, "step": 190020 }, { "epoch": 3.1093839482941994, "grad_norm": 0.00022474986326415092, "learning_rate": 1.4356323364909941e-06, "loss": 0.0002, "step": 190030 }, { "epoch": 3.1095475742452754, "grad_norm": 0.015966275706887245, "learning_rate": 1.4351316829385785e-06, "loss": 0.0002, "step": 190040 }, { "epoch": 3.109711200196351, "grad_norm": 0.00019921762577723712, "learning_rate": 1.4346311020707943e-06, "loss": 0.0002, "step": 190050 }, { "epoch": 3.109874826147427, "grad_norm": 0.061348807066679, "learning_rate": 1.4341305938978501e-06, "loss": 0.0001, "step": 190060 }, { "epoch": 3.110038452098503, "grad_norm": 0.037569135427474976, "learning_rate": 1.43363015842995e-06, "loss": 0.0003, "step": 190070 }, { "epoch": 3.1102020780495785, "grad_norm": 0.015336764976382256, "learning_rate": 1.4331297956772995e-06, "loss": 0.0002, "step": 190080 }, { "epoch": 3.1103657040006545, "grad_norm": 0.028656890615820885, "learning_rate": 1.4326295056500972e-06, "loss": 0.0003, "step": 190090 }, { "epoch": 3.1105293299517305, "grad_norm": 0.005824097897857428, "learning_rate": 1.4321292883585475e-06, "loss": 0.0002, "step": 190100 }, { "epoch": 3.110692955902806, "grad_norm": 0.009453196078538895, "learning_rate": 1.4316291438128439e-06, "loss": 0.0004, "step": 190110 }, { "epoch": 3.110856581853882, "grad_norm": 0.03273741528391838, "learning_rate": 1.4311290720231902e-06, "loss": 0.0003, "step": 190120 }, { "epoch": 3.111020207804958, "grad_norm": 0.017861200496554375, "learning_rate": 1.4306290729997779e-06, "loss": 0.0007, "step": 190130 }, { "epoch": 3.1111838337560336, "grad_norm": 0.008159280754625797, "learning_rate": 1.430129146752805e-06, "loss": 0.0007, "step": 190140 }, { "epoch": 3.1113474597071096, "grad_norm": 0.05913849174976349, "learning_rate": 1.4296292932924594e-06, "loss": 0.0007, "step": 190150 }, { "epoch": 3.1115110856581856, "grad_norm": 0.005338755436241627, "learning_rate": 1.429129512628939e-06, "loss": 0.0003, "step": 190160 }, { "epoch": 3.111674711609261, "grad_norm": 0.039475083351135254, "learning_rate": 1.4286298047724295e-06, "loss": 0.0006, "step": 190170 }, { "epoch": 3.111838337560337, "grad_norm": 0.017847299575805664, "learning_rate": 1.4281301697331223e-06, "loss": 0.0005, "step": 190180 }, { "epoch": 3.112001963511413, "grad_norm": 0.002880642656236887, "learning_rate": 1.4276306075212004e-06, "loss": 0.0, "step": 190190 }, { "epoch": 3.1121655894624887, "grad_norm": 0.010839492082595825, "learning_rate": 1.4271311181468555e-06, "loss": 0.0003, "step": 190200 }, { "epoch": 3.1123292154135647, "grad_norm": 0.04703912511467934, "learning_rate": 1.4266317016202663e-06, "loss": 0.0005, "step": 190210 }, { "epoch": 3.11249284136464, "grad_norm": 0.011182349175214767, "learning_rate": 1.42613235795162e-06, "loss": 0.0003, "step": 190220 }, { "epoch": 3.112656467315716, "grad_norm": 0.09083131700754166, "learning_rate": 1.4256330871510914e-06, "loss": 0.0002, "step": 190230 }, { "epoch": 3.112820093266792, "grad_norm": 0.13854990899562836, "learning_rate": 1.4251338892288684e-06, "loss": 0.0004, "step": 190240 }, { "epoch": 3.1129837192178678, "grad_norm": 0.017097610980272293, "learning_rate": 1.4246347641951231e-06, "loss": 0.0001, "step": 190250 }, { "epoch": 3.1131473451689438, "grad_norm": 0.041322313249111176, "learning_rate": 1.4241357120600358e-06, "loss": 0.0008, "step": 190260 }, { "epoch": 3.1133109711200198, "grad_norm": 0.1277758777141571, "learning_rate": 1.423636732833777e-06, "loss": 0.0005, "step": 190270 }, { "epoch": 3.1134745970710953, "grad_norm": 0.0013877692399546504, "learning_rate": 1.423137826526526e-06, "loss": 0.0004, "step": 190280 }, { "epoch": 3.1136382230221713, "grad_norm": 0.03126411885023117, "learning_rate": 1.4226389931484525e-06, "loss": 0.0005, "step": 190290 }, { "epoch": 3.1138018489732473, "grad_norm": 0.007616202346980572, "learning_rate": 1.4221402327097284e-06, "loss": 0.0002, "step": 190300 }, { "epoch": 3.113965474924323, "grad_norm": 0.04394512623548508, "learning_rate": 1.42164154522052e-06, "loss": 0.0009, "step": 190310 }, { "epoch": 3.114129100875399, "grad_norm": 0.011397735215723515, "learning_rate": 1.421142930690998e-06, "loss": 0.0002, "step": 190320 }, { "epoch": 3.114292726826475, "grad_norm": 0.029740003868937492, "learning_rate": 1.4206443891313276e-06, "loss": 0.0002, "step": 190330 }, { "epoch": 3.1144563527775504, "grad_norm": 0.09302923828363419, "learning_rate": 1.4201459205516754e-06, "loss": 0.0002, "step": 190340 }, { "epoch": 3.1146199787286264, "grad_norm": 0.003373639192432165, "learning_rate": 1.419647524962202e-06, "loss": 0.0003, "step": 190350 }, { "epoch": 3.1147836046797024, "grad_norm": 0.04878968745470047, "learning_rate": 1.4191492023730706e-06, "loss": 0.0003, "step": 190360 }, { "epoch": 3.114947230630778, "grad_norm": 0.008040720596909523, "learning_rate": 1.418650952794442e-06, "loss": 0.0001, "step": 190370 }, { "epoch": 3.115110856581854, "grad_norm": 0.04490569978952408, "learning_rate": 1.4181527762364755e-06, "loss": 0.0006, "step": 190380 }, { "epoch": 3.11527448253293, "grad_norm": 0.026125341653823853, "learning_rate": 1.417654672709326e-06, "loss": 0.0003, "step": 190390 }, { "epoch": 3.1154381084840055, "grad_norm": 0.10691559314727783, "learning_rate": 1.4171566422231515e-06, "loss": 0.0004, "step": 190400 }, { "epoch": 3.1156017344350815, "grad_norm": 0.10116235911846161, "learning_rate": 1.416658684788106e-06, "loss": 0.0004, "step": 190410 }, { "epoch": 3.115765360386157, "grad_norm": 0.006486839149147272, "learning_rate": 1.4161608004143435e-06, "loss": 0.0003, "step": 190420 }, { "epoch": 3.115928986337233, "grad_norm": 0.044137656688690186, "learning_rate": 1.4156629891120133e-06, "loss": 0.0005, "step": 190430 }, { "epoch": 3.116092612288309, "grad_norm": 0.10543983429670334, "learning_rate": 1.415165250891266e-06, "loss": 0.0003, "step": 190440 }, { "epoch": 3.1162562382393846, "grad_norm": 0.036130886524915695, "learning_rate": 1.4146675857622505e-06, "loss": 0.0003, "step": 190450 }, { "epoch": 3.1164198641904606, "grad_norm": 0.040350597351789474, "learning_rate": 1.4141699937351156e-06, "loss": 0.0004, "step": 190460 }, { "epoch": 3.1165834901415366, "grad_norm": 0.06619403511285782, "learning_rate": 1.413672474820003e-06, "loss": 0.0002, "step": 190470 }, { "epoch": 3.116747116092612, "grad_norm": 0.02379615232348442, "learning_rate": 1.4131750290270585e-06, "loss": 0.0002, "step": 190480 }, { "epoch": 3.116910742043688, "grad_norm": 0.1560189574956894, "learning_rate": 1.4126776563664253e-06, "loss": 0.0006, "step": 190490 }, { "epoch": 3.117074367994764, "grad_norm": 0.02380264922976494, "learning_rate": 1.4121803568482445e-06, "loss": 0.0002, "step": 190500 }, { "epoch": 3.1172379939458397, "grad_norm": 0.020006712526082993, "learning_rate": 1.411683130482654e-06, "loss": 0.0006, "step": 190510 }, { "epoch": 3.1174016198969157, "grad_norm": 0.05980224162340164, "learning_rate": 1.411185977279792e-06, "loss": 0.0002, "step": 190520 }, { "epoch": 3.1175652458479917, "grad_norm": 0.025921663269400597, "learning_rate": 1.4106888972497961e-06, "loss": 0.0002, "step": 190530 }, { "epoch": 3.117728871799067, "grad_norm": 0.0038575062062591314, "learning_rate": 1.4101918904028028e-06, "loss": 0.0013, "step": 190540 }, { "epoch": 3.117892497750143, "grad_norm": 0.15808740258216858, "learning_rate": 1.409694956748942e-06, "loss": 0.0007, "step": 190550 }, { "epoch": 3.118056123701219, "grad_norm": 0.018400724977254868, "learning_rate": 1.4091980962983475e-06, "loss": 0.0004, "step": 190560 }, { "epoch": 3.1182197496522948, "grad_norm": 0.04449866712093353, "learning_rate": 1.40870130906115e-06, "loss": 0.0002, "step": 190570 }, { "epoch": 3.1183833756033708, "grad_norm": 0.13246165215969086, "learning_rate": 1.4082045950474804e-06, "loss": 0.0004, "step": 190580 }, { "epoch": 3.1185470015544468, "grad_norm": 0.019830485805869102, "learning_rate": 1.4077079542674627e-06, "loss": 0.0003, "step": 190590 }, { "epoch": 3.1187106275055223, "grad_norm": 0.04881377890706062, "learning_rate": 1.4072113867312243e-06, "loss": 0.0002, "step": 190600 }, { "epoch": 3.1188742534565983, "grad_norm": 0.023700516670942307, "learning_rate": 1.4067148924488904e-06, "loss": 0.0004, "step": 190610 }, { "epoch": 3.119037879407674, "grad_norm": 0.053554385900497437, "learning_rate": 1.406218471430586e-06, "loss": 0.0005, "step": 190620 }, { "epoch": 3.11920150535875, "grad_norm": 0.056476566940546036, "learning_rate": 1.4057221236864283e-06, "loss": 0.0002, "step": 190630 }, { "epoch": 3.119365131309826, "grad_norm": 0.02477899193763733, "learning_rate": 1.4052258492265398e-06, "loss": 0.0003, "step": 190640 }, { "epoch": 3.1195287572609014, "grad_norm": 0.008252806030213833, "learning_rate": 1.404729648061039e-06, "loss": 0.0003, "step": 190650 }, { "epoch": 3.1196923832119774, "grad_norm": 0.05569424107670784, "learning_rate": 1.4042335202000445e-06, "loss": 0.0005, "step": 190660 }, { "epoch": 3.1198560091630534, "grad_norm": 0.003829031717032194, "learning_rate": 1.403737465653669e-06, "loss": 0.0002, "step": 190670 }, { "epoch": 3.120019635114129, "grad_norm": 0.045118268579244614, "learning_rate": 1.403241484432028e-06, "loss": 0.0007, "step": 190680 }, { "epoch": 3.120183261065205, "grad_norm": 0.008752160705626011, "learning_rate": 1.4027455765452348e-06, "loss": 0.0002, "step": 190690 }, { "epoch": 3.120346887016281, "grad_norm": 0.07146770507097244, "learning_rate": 1.402249742003401e-06, "loss": 0.0003, "step": 190700 }, { "epoch": 3.1205105129673565, "grad_norm": 0.004742403049021959, "learning_rate": 1.4017539808166331e-06, "loss": 0.0003, "step": 190710 }, { "epoch": 3.1206741389184325, "grad_norm": 0.014095626771450043, "learning_rate": 1.401258292995042e-06, "loss": 0.0003, "step": 190720 }, { "epoch": 3.1208377648695085, "grad_norm": 0.011896900832653046, "learning_rate": 1.4007626785487333e-06, "loss": 0.0002, "step": 190730 }, { "epoch": 3.121001390820584, "grad_norm": 0.04586442932486534, "learning_rate": 1.4002671374878146e-06, "loss": 0.0002, "step": 190740 }, { "epoch": 3.12116501677166, "grad_norm": 0.015564952045679092, "learning_rate": 1.3997716698223857e-06, "loss": 0.0002, "step": 190750 }, { "epoch": 3.121328642722736, "grad_norm": 0.05535292997956276, "learning_rate": 1.3992762755625515e-06, "loss": 0.0002, "step": 190760 }, { "epoch": 3.1214922686738116, "grad_norm": 0.022298920899629593, "learning_rate": 1.3987809547184095e-06, "loss": 0.0003, "step": 190770 }, { "epoch": 3.1216558946248876, "grad_norm": 0.015017304569482803, "learning_rate": 1.398285707300064e-06, "loss": 0.0004, "step": 190780 }, { "epoch": 3.121819520575963, "grad_norm": 0.014330564998090267, "learning_rate": 1.3977905333176078e-06, "loss": 0.0003, "step": 190790 }, { "epoch": 3.121983146527039, "grad_norm": 0.04890112206339836, "learning_rate": 1.397295432781141e-06, "loss": 0.0002, "step": 190800 }, { "epoch": 3.122146772478115, "grad_norm": 0.048589080572128296, "learning_rate": 1.3968004057007533e-06, "loss": 0.0004, "step": 190810 }, { "epoch": 3.1223103984291907, "grad_norm": 0.007565023377537727, "learning_rate": 1.3963054520865439e-06, "loss": 0.0003, "step": 190820 }, { "epoch": 3.1224740243802667, "grad_norm": 0.0301694106310606, "learning_rate": 1.3958105719485998e-06, "loss": 0.0004, "step": 190830 }, { "epoch": 3.1226376503313427, "grad_norm": 0.0007646844605915248, "learning_rate": 1.395315765297015e-06, "loss": 0.0002, "step": 190840 }, { "epoch": 3.122801276282418, "grad_norm": 0.00039437355007976294, "learning_rate": 1.3948210321418725e-06, "loss": 0.0004, "step": 190850 }, { "epoch": 3.122964902233494, "grad_norm": 0.03404984250664711, "learning_rate": 1.394326372493266e-06, "loss": 0.0005, "step": 190860 }, { "epoch": 3.12312852818457, "grad_norm": 0.012731633149087429, "learning_rate": 1.3938317863612771e-06, "loss": 0.0001, "step": 190870 }, { "epoch": 3.1232921541356458, "grad_norm": 0.003204723820090294, "learning_rate": 1.3933372737559924e-06, "loss": 0.0004, "step": 190880 }, { "epoch": 3.1234557800867218, "grad_norm": 0.047317229211330414, "learning_rate": 1.3928428346874907e-06, "loss": 0.0004, "step": 190890 }, { "epoch": 3.1236194060377978, "grad_norm": 0.012304565869271755, "learning_rate": 1.3923484691658583e-06, "loss": 0.0002, "step": 190900 }, { "epoch": 3.1237830319888733, "grad_norm": 0.0007616381044499576, "learning_rate": 1.3918541772011712e-06, "loss": 0.0002, "step": 190910 }, { "epoch": 3.1239466579399493, "grad_norm": 0.0636158362030983, "learning_rate": 1.3913599588035098e-06, "loss": 0.0005, "step": 190920 }, { "epoch": 3.1241102838910253, "grad_norm": 0.04928533732891083, "learning_rate": 1.390865813982949e-06, "loss": 0.0003, "step": 190930 }, { "epoch": 3.124273909842101, "grad_norm": 0.024757344275712967, "learning_rate": 1.3903717427495645e-06, "loss": 0.0002, "step": 190940 }, { "epoch": 3.124437535793177, "grad_norm": 0.004533539526164532, "learning_rate": 1.3898777451134305e-06, "loss": 0.0005, "step": 190950 }, { "epoch": 3.124601161744253, "grad_norm": 0.08910632878541946, "learning_rate": 1.38938382108462e-06, "loss": 0.0002, "step": 190960 }, { "epoch": 3.1247647876953284, "grad_norm": 0.04979579895734787, "learning_rate": 1.3888899706732017e-06, "loss": 0.0007, "step": 190970 }, { "epoch": 3.1249284136464044, "grad_norm": 0.04295658692717552, "learning_rate": 1.3883961938892454e-06, "loss": 0.0002, "step": 190980 }, { "epoch": 3.1250920395974804, "grad_norm": 0.02480102889239788, "learning_rate": 1.3879024907428195e-06, "loss": 0.0002, "step": 190990 }, { "epoch": 3.125255665548556, "grad_norm": 0.10026423633098602, "learning_rate": 1.3874088612439911e-06, "loss": 0.0005, "step": 191000 }, { "epoch": 3.125419291499632, "grad_norm": 0.008286864496767521, "learning_rate": 1.3869153054028217e-06, "loss": 0.0003, "step": 191010 }, { "epoch": 3.1255829174507075, "grad_norm": 0.011780339293181896, "learning_rate": 1.386421823229377e-06, "loss": 0.0003, "step": 191020 }, { "epoch": 3.1257465434017835, "grad_norm": 0.059483177959918976, "learning_rate": 1.3859284147337182e-06, "loss": 0.0004, "step": 191030 }, { "epoch": 3.1259101693528595, "grad_norm": 0.12689651548862457, "learning_rate": 1.3854350799259065e-06, "loss": 0.0003, "step": 191040 }, { "epoch": 3.126073795303935, "grad_norm": 0.03304408863186836, "learning_rate": 1.384941818815998e-06, "loss": 0.0003, "step": 191050 }, { "epoch": 3.126237421255011, "grad_norm": 0.0036408721935003996, "learning_rate": 1.3844486314140515e-06, "loss": 0.0003, "step": 191060 }, { "epoch": 3.126401047206087, "grad_norm": 0.11371579766273499, "learning_rate": 1.3839555177301222e-06, "loss": 0.0003, "step": 191070 }, { "epoch": 3.1265646731571626, "grad_norm": 0.11044024676084518, "learning_rate": 1.3834624777742662e-06, "loss": 0.0004, "step": 191080 }, { "epoch": 3.1267282991082386, "grad_norm": 0.010593649931252003, "learning_rate": 1.3829695115565328e-06, "loss": 0.0002, "step": 191090 }, { "epoch": 3.1268919250593146, "grad_norm": 0.020962005481123924, "learning_rate": 1.3824766190869753e-06, "loss": 0.0003, "step": 191100 }, { "epoch": 3.12705555101039, "grad_norm": 0.002616656245663762, "learning_rate": 1.3819838003756426e-06, "loss": 0.0001, "step": 191110 }, { "epoch": 3.127219176961466, "grad_norm": 0.000214360625250265, "learning_rate": 1.381491055432585e-06, "loss": 0.0003, "step": 191120 }, { "epoch": 3.127382802912542, "grad_norm": 0.06870482116937637, "learning_rate": 1.3809983842678454e-06, "loss": 0.0006, "step": 191130 }, { "epoch": 3.1275464288636177, "grad_norm": 0.03605928272008896, "learning_rate": 1.3805057868914712e-06, "loss": 0.0004, "step": 191140 }, { "epoch": 3.1277100548146937, "grad_norm": 0.04174315556883812, "learning_rate": 1.380013263313506e-06, "loss": 0.0003, "step": 191150 }, { "epoch": 3.127873680765769, "grad_norm": 0.05493967980146408, "learning_rate": 1.379520813543993e-06, "loss": 0.0003, "step": 191160 }, { "epoch": 3.128037306716845, "grad_norm": 0.05447472259402275, "learning_rate": 1.3790284375929696e-06, "loss": 0.0004, "step": 191170 }, { "epoch": 3.128200932667921, "grad_norm": 0.018988700583577156, "learning_rate": 1.3785361354704774e-06, "loss": 0.0002, "step": 191180 }, { "epoch": 3.1283645586189968, "grad_norm": 0.041257210075855255, "learning_rate": 1.3780439071865532e-06, "loss": 0.0005, "step": 191190 }, { "epoch": 3.1285281845700728, "grad_norm": 0.13561108708381653, "learning_rate": 1.377551752751235e-06, "loss": 0.0005, "step": 191200 }, { "epoch": 3.1286918105211488, "grad_norm": 0.07921213656663895, "learning_rate": 1.3770596721745537e-06, "loss": 0.0006, "step": 191210 }, { "epoch": 3.1288554364722243, "grad_norm": 0.014280606061220169, "learning_rate": 1.376567665466545e-06, "loss": 0.0011, "step": 191220 }, { "epoch": 3.1290190624233003, "grad_norm": 0.005211647599935532, "learning_rate": 1.3760757326372397e-06, "loss": 0.0002, "step": 191230 }, { "epoch": 3.1291826883743763, "grad_norm": 0.006521865725517273, "learning_rate": 1.3755838736966703e-06, "loss": 0.0003, "step": 191240 }, { "epoch": 3.129346314325452, "grad_norm": 0.020391713827848434, "learning_rate": 1.3750920886548614e-06, "loss": 0.0003, "step": 191250 }, { "epoch": 3.129509940276528, "grad_norm": 0.03794635459780693, "learning_rate": 1.3746003775218415e-06, "loss": 0.0003, "step": 191260 }, { "epoch": 3.129673566227604, "grad_norm": 0.0008511111373081803, "learning_rate": 1.374108740307637e-06, "loss": 0.0002, "step": 191270 }, { "epoch": 3.1298371921786794, "grad_norm": 0.016773074865341187, "learning_rate": 1.3736171770222734e-06, "loss": 0.0002, "step": 191280 }, { "epoch": 3.1300008181297554, "grad_norm": 0.027003681287169456, "learning_rate": 1.3731256876757692e-06, "loss": 0.0003, "step": 191290 }, { "epoch": 3.1301644440808314, "grad_norm": 0.03597450256347656, "learning_rate": 1.3726342722781483e-06, "loss": 0.0008, "step": 191300 }, { "epoch": 3.130328070031907, "grad_norm": 0.011165093630552292, "learning_rate": 1.372142930839429e-06, "loss": 0.0002, "step": 191310 }, { "epoch": 3.130491695982983, "grad_norm": 0.008267897181212902, "learning_rate": 1.3716516633696319e-06, "loss": 0.0003, "step": 191320 }, { "epoch": 3.130655321934059, "grad_norm": 0.0007288549095392227, "learning_rate": 1.3711604698787706e-06, "loss": 0.0005, "step": 191330 }, { "epoch": 3.1308189478851345, "grad_norm": 0.0007647146703675389, "learning_rate": 1.3706693503768603e-06, "loss": 0.0003, "step": 191340 }, { "epoch": 3.1309825738362105, "grad_norm": 0.04146188497543335, "learning_rate": 1.3701783048739153e-06, "loss": 0.0002, "step": 191350 }, { "epoch": 3.1311461997872865, "grad_norm": 0.04600007086992264, "learning_rate": 1.3696873333799499e-06, "loss": 0.0005, "step": 191360 }, { "epoch": 3.131309825738362, "grad_norm": 0.060494668781757355, "learning_rate": 1.3691964359049698e-06, "loss": 0.0005, "step": 191370 }, { "epoch": 3.131473451689438, "grad_norm": 0.002655821153894067, "learning_rate": 1.3687056124589887e-06, "loss": 0.0002, "step": 191380 }, { "epoch": 3.1316370776405136, "grad_norm": 0.004411136731505394, "learning_rate": 1.3682148630520088e-06, "loss": 0.0001, "step": 191390 }, { "epoch": 3.1318007035915896, "grad_norm": 0.003959465306252241, "learning_rate": 1.3677241876940417e-06, "loss": 0.0005, "step": 191400 }, { "epoch": 3.1319643295426656, "grad_norm": 0.00200714566744864, "learning_rate": 1.367233586395088e-06, "loss": 0.0002, "step": 191410 }, { "epoch": 3.132127955493741, "grad_norm": 0.0005001283134333789, "learning_rate": 1.3667430591651532e-06, "loss": 0.0002, "step": 191420 }, { "epoch": 3.132291581444817, "grad_norm": 0.0323614664375782, "learning_rate": 1.366252606014235e-06, "loss": 0.0004, "step": 191430 }, { "epoch": 3.132455207395893, "grad_norm": 0.0242465827614069, "learning_rate": 1.3657622269523385e-06, "loss": 0.0002, "step": 191440 }, { "epoch": 3.1326188333469687, "grad_norm": 0.004471123684197664, "learning_rate": 1.3652719219894577e-06, "loss": 0.0003, "step": 191450 }, { "epoch": 3.1327824592980447, "grad_norm": 0.057641252875328064, "learning_rate": 1.3647816911355926e-06, "loss": 0.0002, "step": 191460 }, { "epoch": 3.1329460852491207, "grad_norm": 0.027306299656629562, "learning_rate": 1.3642915344007347e-06, "loss": 0.0003, "step": 191470 }, { "epoch": 3.133109711200196, "grad_norm": 0.09404770284891129, "learning_rate": 1.3638014517948829e-06, "loss": 0.0004, "step": 191480 }, { "epoch": 3.133273337151272, "grad_norm": 0.0008006193675100803, "learning_rate": 1.3633114433280254e-06, "loss": 0.0002, "step": 191490 }, { "epoch": 3.133436963102348, "grad_norm": 0.026144346222281456, "learning_rate": 1.3628215090101564e-06, "loss": 0.0004, "step": 191500 }, { "epoch": 3.1336005890534238, "grad_norm": 0.0023115365765988827, "learning_rate": 1.3623316488512606e-06, "loss": 0.0004, "step": 191510 }, { "epoch": 3.1337642150044998, "grad_norm": 0.026430336758494377, "learning_rate": 1.3618418628613318e-06, "loss": 0.0004, "step": 191520 }, { "epoch": 3.1339278409555757, "grad_norm": 0.00011150213686050847, "learning_rate": 1.3613521510503518e-06, "loss": 0.0002, "step": 191530 }, { "epoch": 3.1340914669066513, "grad_norm": 0.056835658848285675, "learning_rate": 1.3608625134283082e-06, "loss": 0.0004, "step": 191540 }, { "epoch": 3.1342550928577273, "grad_norm": 0.007465999573469162, "learning_rate": 1.3603729500051799e-06, "loss": 0.0003, "step": 191550 }, { "epoch": 3.134418718808803, "grad_norm": 0.05090808495879173, "learning_rate": 1.359883460790955e-06, "loss": 0.0002, "step": 191560 }, { "epoch": 3.134582344759879, "grad_norm": 0.0192077849060297, "learning_rate": 1.359394045795609e-06, "loss": 0.0003, "step": 191570 }, { "epoch": 3.134745970710955, "grad_norm": 0.05387311801314354, "learning_rate": 1.3589047050291238e-06, "loss": 0.0006, "step": 191580 }, { "epoch": 3.1349095966620304, "grad_norm": 0.0007621226250194013, "learning_rate": 1.358415438501473e-06, "loss": 0.0001, "step": 191590 }, { "epoch": 3.1350732226131064, "grad_norm": 0.03364712372422218, "learning_rate": 1.357926246222635e-06, "loss": 0.0005, "step": 191600 }, { "epoch": 3.1352368485641824, "grad_norm": 0.005298278294503689, "learning_rate": 1.3574371282025829e-06, "loss": 0.0003, "step": 191610 }, { "epoch": 3.135400474515258, "grad_norm": 0.017036180943250656, "learning_rate": 1.3569480844512918e-06, "loss": 0.0002, "step": 191620 }, { "epoch": 3.135564100466334, "grad_norm": 0.02469629794359207, "learning_rate": 1.3564591149787288e-06, "loss": 0.0003, "step": 191630 }, { "epoch": 3.13572772641741, "grad_norm": 0.012503737583756447, "learning_rate": 1.355970219794866e-06, "loss": 0.0001, "step": 191640 }, { "epoch": 3.1358913523684855, "grad_norm": 0.020585743710398674, "learning_rate": 1.3554813989096715e-06, "loss": 0.0002, "step": 191650 }, { "epoch": 3.1360549783195615, "grad_norm": 0.015805775299668312, "learning_rate": 1.354992652333113e-06, "loss": 0.0004, "step": 191660 }, { "epoch": 3.1362186042706375, "grad_norm": 0.1476503312587738, "learning_rate": 1.354503980075153e-06, "loss": 0.0004, "step": 191670 }, { "epoch": 3.136382230221713, "grad_norm": 0.0009152843849733472, "learning_rate": 1.3540153821457563e-06, "loss": 0.0001, "step": 191680 }, { "epoch": 3.136545856172789, "grad_norm": 0.07225023210048676, "learning_rate": 1.3535268585548856e-06, "loss": 0.0002, "step": 191690 }, { "epoch": 3.136709482123865, "grad_norm": 0.006651431787759066, "learning_rate": 1.3530384093125021e-06, "loss": 0.0001, "step": 191700 }, { "epoch": 3.1368731080749406, "grad_norm": 0.016110526397824287, "learning_rate": 1.3525500344285624e-06, "loss": 0.0002, "step": 191710 }, { "epoch": 3.1370367340260166, "grad_norm": 0.07123060524463654, "learning_rate": 1.3520617339130254e-06, "loss": 0.0003, "step": 191720 }, { "epoch": 3.1372003599770926, "grad_norm": 0.0010941398795694113, "learning_rate": 1.3515735077758475e-06, "loss": 0.0002, "step": 191730 }, { "epoch": 3.137363985928168, "grad_norm": 0.055662043392658234, "learning_rate": 1.3510853560269838e-06, "loss": 0.0004, "step": 191740 }, { "epoch": 3.137527611879244, "grad_norm": 0.040469471365213394, "learning_rate": 1.3505972786763854e-06, "loss": 0.0005, "step": 191750 }, { "epoch": 3.13769123783032, "grad_norm": 0.025846784934401512, "learning_rate": 1.3501092757340045e-06, "loss": 0.0003, "step": 191760 }, { "epoch": 3.1378548637813957, "grad_norm": 0.0006285830750130117, "learning_rate": 1.3496213472097919e-06, "loss": 0.0005, "step": 191770 }, { "epoch": 3.1380184897324717, "grad_norm": 0.05716405063867569, "learning_rate": 1.3491334931136962e-06, "loss": 0.0003, "step": 191780 }, { "epoch": 3.138182115683547, "grad_norm": 0.009906202554702759, "learning_rate": 1.3486457134556624e-06, "loss": 0.0003, "step": 191790 }, { "epoch": 3.138345741634623, "grad_norm": 0.06602025032043457, "learning_rate": 1.3481580082456374e-06, "loss": 0.0005, "step": 191800 }, { "epoch": 3.138509367585699, "grad_norm": 0.0029280437156558037, "learning_rate": 1.3476703774935647e-06, "loss": 0.0002, "step": 191810 }, { "epoch": 3.1386729935367748, "grad_norm": 0.0010000023758038878, "learning_rate": 1.3471828212093885e-06, "loss": 0.0002, "step": 191820 }, { "epoch": 3.1388366194878508, "grad_norm": 0.00022509833797812462, "learning_rate": 1.3466953394030463e-06, "loss": 0.0001, "step": 191830 }, { "epoch": 3.1390002454389268, "grad_norm": 0.03520812466740608, "learning_rate": 1.346207932084479e-06, "loss": 0.0001, "step": 191840 }, { "epoch": 3.1391638713900023, "grad_norm": 0.0221492201089859, "learning_rate": 1.3457205992636241e-06, "loss": 0.0003, "step": 191850 }, { "epoch": 3.1393274973410783, "grad_norm": 0.05319325253367424, "learning_rate": 1.3452333409504204e-06, "loss": 0.0005, "step": 191860 }, { "epoch": 3.1394911232921543, "grad_norm": 0.035189952701330185, "learning_rate": 1.3447461571547994e-06, "loss": 0.0004, "step": 191870 }, { "epoch": 3.13965474924323, "grad_norm": 0.018308520317077637, "learning_rate": 1.3442590478866952e-06, "loss": 0.0002, "step": 191880 }, { "epoch": 3.139818375194306, "grad_norm": 0.029137758538126945, "learning_rate": 1.3437720131560395e-06, "loss": 0.0005, "step": 191890 }, { "epoch": 3.139982001145382, "grad_norm": 0.05508272722363472, "learning_rate": 1.3432850529727647e-06, "loss": 0.0003, "step": 191900 }, { "epoch": 3.1401456270964574, "grad_norm": 0.024535438045859337, "learning_rate": 1.3427981673467966e-06, "loss": 0.0003, "step": 191910 }, { "epoch": 3.1403092530475334, "grad_norm": 0.01977258175611496, "learning_rate": 1.3423113562880636e-06, "loss": 0.0002, "step": 191920 }, { "epoch": 3.1404728789986094, "grad_norm": 0.06879094243049622, "learning_rate": 1.341824619806491e-06, "loss": 0.0004, "step": 191930 }, { "epoch": 3.140636504949685, "grad_norm": 0.004797609057277441, "learning_rate": 1.3413379579120045e-06, "loss": 0.0006, "step": 191940 }, { "epoch": 3.140800130900761, "grad_norm": 0.005150042474269867, "learning_rate": 1.3408513706145242e-06, "loss": 0.0001, "step": 191950 }, { "epoch": 3.1409637568518365, "grad_norm": 0.040411848574876785, "learning_rate": 1.3403648579239726e-06, "loss": 0.0003, "step": 191960 }, { "epoch": 3.1411273828029125, "grad_norm": 0.01893470622599125, "learning_rate": 1.339878419850269e-06, "loss": 0.0001, "step": 191970 }, { "epoch": 3.1412910087539885, "grad_norm": 0.012627643533051014, "learning_rate": 1.339392056403333e-06, "loss": 0.0004, "step": 191980 }, { "epoch": 3.141454634705064, "grad_norm": 0.2533954381942749, "learning_rate": 1.3389057675930783e-06, "loss": 0.0006, "step": 191990 }, { "epoch": 3.14161826065614, "grad_norm": 0.0007997126085683703, "learning_rate": 1.3384195534294214e-06, "loss": 0.0002, "step": 192000 }, { "epoch": 3.141781886607216, "grad_norm": 0.023759616538882256, "learning_rate": 1.337933413922276e-06, "loss": 0.0004, "step": 192010 }, { "epoch": 3.1419455125582916, "grad_norm": 0.003792915493249893, "learning_rate": 1.3374473490815549e-06, "loss": 0.0006, "step": 192020 }, { "epoch": 3.1421091385093676, "grad_norm": 0.03317626193165779, "learning_rate": 1.3369613589171659e-06, "loss": 0.0003, "step": 192030 }, { "epoch": 3.1422727644604436, "grad_norm": 0.08047201484441757, "learning_rate": 1.3364754434390214e-06, "loss": 0.0005, "step": 192040 }, { "epoch": 3.142436390411519, "grad_norm": 0.02514764852821827, "learning_rate": 1.3359896026570235e-06, "loss": 0.0004, "step": 192050 }, { "epoch": 3.142600016362595, "grad_norm": 0.022392118349671364, "learning_rate": 1.3355038365810846e-06, "loss": 0.0003, "step": 192060 }, { "epoch": 3.142763642313671, "grad_norm": 0.0020396201871335506, "learning_rate": 1.3350181452211048e-06, "loss": 0.0002, "step": 192070 }, { "epoch": 3.1429272682647467, "grad_norm": 0.014654744416475296, "learning_rate": 1.3345325285869893e-06, "loss": 0.0003, "step": 192080 }, { "epoch": 3.1430908942158227, "grad_norm": 0.02528189867734909, "learning_rate": 1.3340469866886353e-06, "loss": 0.0003, "step": 192090 }, { "epoch": 3.1432545201668987, "grad_norm": 0.003093202831223607, "learning_rate": 1.3335615195359481e-06, "loss": 0.0003, "step": 192100 }, { "epoch": 3.143418146117974, "grad_norm": 0.03134193271398544, "learning_rate": 1.333076127138822e-06, "loss": 0.0004, "step": 192110 }, { "epoch": 3.14358177206905, "grad_norm": 0.007231492083519697, "learning_rate": 1.3325908095071565e-06, "loss": 0.0002, "step": 192120 }, { "epoch": 3.143745398020126, "grad_norm": 0.015649424865841866, "learning_rate": 1.3321055666508425e-06, "loss": 0.0003, "step": 192130 }, { "epoch": 3.1439090239712018, "grad_norm": 0.00721958139911294, "learning_rate": 1.3316203985797794e-06, "loss": 0.0001, "step": 192140 }, { "epoch": 3.1440726499222778, "grad_norm": 0.07684601098299026, "learning_rate": 1.331135305303855e-06, "loss": 0.0005, "step": 192150 }, { "epoch": 3.1442362758733537, "grad_norm": 0.011798973195254803, "learning_rate": 1.3306502868329634e-06, "loss": 0.0002, "step": 192160 }, { "epoch": 3.1443999018244293, "grad_norm": 0.029337078332901, "learning_rate": 1.3301653431769883e-06, "loss": 0.0003, "step": 192170 }, { "epoch": 3.1445635277755053, "grad_norm": 0.005471241194754839, "learning_rate": 1.3296804743458241e-06, "loss": 0.0003, "step": 192180 }, { "epoch": 3.144727153726581, "grad_norm": 0.022296857088804245, "learning_rate": 1.3291956803493517e-06, "loss": 0.0002, "step": 192190 }, { "epoch": 3.144890779677657, "grad_norm": 0.007908749394118786, "learning_rate": 1.3287109611974592e-06, "loss": 0.0003, "step": 192200 }, { "epoch": 3.145054405628733, "grad_norm": 0.0215753260999918, "learning_rate": 1.3282263169000265e-06, "loss": 0.0006, "step": 192210 }, { "epoch": 3.1452180315798084, "grad_norm": 0.013337288983166218, "learning_rate": 1.3277417474669364e-06, "loss": 0.0004, "step": 192220 }, { "epoch": 3.1453816575308844, "grad_norm": 0.027780970558524132, "learning_rate": 1.3272572529080686e-06, "loss": 0.0006, "step": 192230 }, { "epoch": 3.1455452834819604, "grad_norm": 0.00045072566717863083, "learning_rate": 1.3267728332333035e-06, "loss": 0.0002, "step": 192240 }, { "epoch": 3.145708909433036, "grad_norm": 0.006132975220680237, "learning_rate": 1.3262884884525151e-06, "loss": 0.0001, "step": 192250 }, { "epoch": 3.145872535384112, "grad_norm": 0.0048222169280052185, "learning_rate": 1.3258042185755798e-06, "loss": 0.0003, "step": 192260 }, { "epoch": 3.146036161335188, "grad_norm": 0.008542223833501339, "learning_rate": 1.325320023612372e-06, "loss": 0.0003, "step": 192270 }, { "epoch": 3.1461997872862635, "grad_norm": 0.04575857147574425, "learning_rate": 1.3248359035727649e-06, "loss": 0.0007, "step": 192280 }, { "epoch": 3.1463634132373395, "grad_norm": 0.0014492126647382975, "learning_rate": 1.3243518584666265e-06, "loss": 0.0006, "step": 192290 }, { "epoch": 3.1465270391884155, "grad_norm": 0.00803923700004816, "learning_rate": 1.323867888303828e-06, "loss": 0.0002, "step": 192300 }, { "epoch": 3.146690665139491, "grad_norm": 0.019940156489610672, "learning_rate": 1.3233839930942366e-06, "loss": 0.0003, "step": 192310 }, { "epoch": 3.146854291090567, "grad_norm": 0.006609009113162756, "learning_rate": 1.3229001728477204e-06, "loss": 0.0004, "step": 192320 }, { "epoch": 3.1470179170416426, "grad_norm": 0.02123531699180603, "learning_rate": 1.3224164275741408e-06, "loss": 0.0001, "step": 192330 }, { "epoch": 3.1471815429927186, "grad_norm": 0.17582419514656067, "learning_rate": 1.321932757283363e-06, "loss": 0.0005, "step": 192340 }, { "epoch": 3.1473451689437946, "grad_norm": 0.000898286874871701, "learning_rate": 1.3214491619852476e-06, "loss": 0.0002, "step": 192350 }, { "epoch": 3.14750879489487, "grad_norm": 0.07386354357004166, "learning_rate": 1.320965641689657e-06, "loss": 0.0002, "step": 192360 }, { "epoch": 3.147672420845946, "grad_norm": 0.009843340143561363, "learning_rate": 1.3204821964064467e-06, "loss": 0.0003, "step": 192370 }, { "epoch": 3.147836046797022, "grad_norm": 0.008881079964339733, "learning_rate": 1.319998826145475e-06, "loss": 0.0006, "step": 192380 }, { "epoch": 3.1479996727480977, "grad_norm": 0.0018268122803419828, "learning_rate": 1.3195155309165975e-06, "loss": 0.0002, "step": 192390 }, { "epoch": 3.1481632986991737, "grad_norm": 0.011719417758286, "learning_rate": 1.3190323107296692e-06, "loss": 0.0004, "step": 192400 }, { "epoch": 3.1483269246502497, "grad_norm": 0.03453025594353676, "learning_rate": 1.318549165594541e-06, "loss": 0.0004, "step": 192410 }, { "epoch": 3.148490550601325, "grad_norm": 0.015893543139100075, "learning_rate": 1.3180660955210638e-06, "loss": 0.0001, "step": 192420 }, { "epoch": 3.148654176552401, "grad_norm": 0.1357981115579605, "learning_rate": 1.3175831005190876e-06, "loss": 0.0004, "step": 192430 }, { "epoch": 3.148817802503477, "grad_norm": 0.02215300127863884, "learning_rate": 1.3171001805984613e-06, "loss": 0.0007, "step": 192440 }, { "epoch": 3.1489814284545528, "grad_norm": 0.028042735531926155, "learning_rate": 1.3166173357690294e-06, "loss": 0.0002, "step": 192450 }, { "epoch": 3.1491450544056288, "grad_norm": 0.027980506420135498, "learning_rate": 1.3161345660406366e-06, "loss": 0.0002, "step": 192460 }, { "epoch": 3.1493086803567047, "grad_norm": 0.025380317121744156, "learning_rate": 1.3156518714231276e-06, "loss": 0.0004, "step": 192470 }, { "epoch": 3.1494723063077803, "grad_norm": 0.013816796243190765, "learning_rate": 1.3151692519263447e-06, "loss": 0.0002, "step": 192480 }, { "epoch": 3.1496359322588563, "grad_norm": 0.010637742467224598, "learning_rate": 1.3146867075601256e-06, "loss": 0.0001, "step": 192490 }, { "epoch": 3.1497995582099323, "grad_norm": 0.017125798389315605, "learning_rate": 1.3142042383343101e-06, "loss": 0.0001, "step": 192500 }, { "epoch": 3.149963184161008, "grad_norm": 0.003266632556915283, "learning_rate": 1.3137218442587358e-06, "loss": 0.0003, "step": 192510 }, { "epoch": 3.150126810112084, "grad_norm": 0.04285513982176781, "learning_rate": 1.3132395253432395e-06, "loss": 0.0004, "step": 192520 }, { "epoch": 3.15029043606316, "grad_norm": 0.00015024124877527356, "learning_rate": 1.312757281597652e-06, "loss": 0.0002, "step": 192530 }, { "epoch": 3.1504540620142354, "grad_norm": 0.2047746479511261, "learning_rate": 1.312275113031808e-06, "loss": 0.0007, "step": 192540 }, { "epoch": 3.1506176879653114, "grad_norm": 0.036054715514183044, "learning_rate": 1.311793019655538e-06, "loss": 0.0006, "step": 192550 }, { "epoch": 3.150781313916387, "grad_norm": 0.11952225118875504, "learning_rate": 1.3113110014786728e-06, "loss": 0.0004, "step": 192560 }, { "epoch": 3.150944939867463, "grad_norm": 0.040425799787044525, "learning_rate": 1.3108290585110384e-06, "loss": 0.0004, "step": 192570 }, { "epoch": 3.151108565818539, "grad_norm": 0.08879195153713226, "learning_rate": 1.310347190762461e-06, "loss": 0.0002, "step": 192580 }, { "epoch": 3.1512721917696145, "grad_norm": 0.0008886773721314967, "learning_rate": 1.3098653982427673e-06, "loss": 0.0004, "step": 192590 }, { "epoch": 3.1514358177206905, "grad_norm": 0.0906791165471077, "learning_rate": 1.3093836809617811e-06, "loss": 0.0002, "step": 192600 }, { "epoch": 3.1515994436717665, "grad_norm": 0.03422036021947861, "learning_rate": 1.3089020389293212e-06, "loss": 0.0008, "step": 192610 }, { "epoch": 3.151763069622842, "grad_norm": 0.001670965226367116, "learning_rate": 1.3084204721552097e-06, "loss": 0.0002, "step": 192620 }, { "epoch": 3.151926695573918, "grad_norm": 0.006958048790693283, "learning_rate": 1.3079389806492649e-06, "loss": 0.0002, "step": 192630 }, { "epoch": 3.152090321524994, "grad_norm": 0.019636303186416626, "learning_rate": 1.3074575644213061e-06, "loss": 0.0003, "step": 192640 }, { "epoch": 3.1522539474760696, "grad_norm": 0.004839437548071146, "learning_rate": 1.306976223481145e-06, "loss": 0.001, "step": 192650 }, { "epoch": 3.1524175734271456, "grad_norm": 0.0010052691213786602, "learning_rate": 1.3064949578385988e-06, "loss": 0.0004, "step": 192660 }, { "epoch": 3.1525811993782216, "grad_norm": 0.02824896015226841, "learning_rate": 1.306013767503479e-06, "loss": 0.0002, "step": 192670 }, { "epoch": 3.152744825329297, "grad_norm": 0.01426512561738491, "learning_rate": 1.3055326524855982e-06, "loss": 0.0003, "step": 192680 }, { "epoch": 3.152908451280373, "grad_norm": 0.017140084877610207, "learning_rate": 1.3050516127947626e-06, "loss": 0.0004, "step": 192690 }, { "epoch": 3.153072077231449, "grad_norm": 0.01313310768455267, "learning_rate": 1.3045706484407843e-06, "loss": 0.0002, "step": 192700 }, { "epoch": 3.1532357031825247, "grad_norm": 0.02736225351691246, "learning_rate": 1.3040897594334644e-06, "loss": 0.0002, "step": 192710 }, { "epoch": 3.1533993291336007, "grad_norm": 0.014453881420195103, "learning_rate": 1.3036089457826144e-06, "loss": 0.0003, "step": 192720 }, { "epoch": 3.153562955084676, "grad_norm": 0.004139354452490807, "learning_rate": 1.3031282074980327e-06, "loss": 0.0004, "step": 192730 }, { "epoch": 3.153726581035752, "grad_norm": 0.007935339584946632, "learning_rate": 1.3026475445895243e-06, "loss": 0.0003, "step": 192740 }, { "epoch": 3.153890206986828, "grad_norm": 0.005917209666222334, "learning_rate": 1.3021669570668844e-06, "loss": 0.0001, "step": 192750 }, { "epoch": 3.1540538329379038, "grad_norm": 0.009009227156639099, "learning_rate": 1.3016864449399191e-06, "loss": 0.0003, "step": 192760 }, { "epoch": 3.1542174588889798, "grad_norm": 0.000600668543484062, "learning_rate": 1.3012060082184196e-06, "loss": 0.0002, "step": 192770 }, { "epoch": 3.1543810848400557, "grad_norm": 0.010045439004898071, "learning_rate": 1.300725646912186e-06, "loss": 0.0003, "step": 192780 }, { "epoch": 3.1545447107911313, "grad_norm": 0.014902075752615929, "learning_rate": 1.3002453610310073e-06, "loss": 0.0004, "step": 192790 }, { "epoch": 3.1547083367422073, "grad_norm": 0.016510270535945892, "learning_rate": 1.299765150584682e-06, "loss": 0.0002, "step": 192800 }, { "epoch": 3.1548719626932833, "grad_norm": 0.08474776148796082, "learning_rate": 1.2992850155829966e-06, "loss": 0.0005, "step": 192810 }, { "epoch": 3.155035588644359, "grad_norm": 0.002881812397390604, "learning_rate": 1.2988049560357447e-06, "loss": 0.0002, "step": 192820 }, { "epoch": 3.155199214595435, "grad_norm": 0.11560498178005219, "learning_rate": 1.2983249719527081e-06, "loss": 0.0004, "step": 192830 }, { "epoch": 3.155362840546511, "grad_norm": 0.003247734857723117, "learning_rate": 1.2978450633436806e-06, "loss": 0.0002, "step": 192840 }, { "epoch": 3.1555264664975864, "grad_norm": 0.007566823158413172, "learning_rate": 1.2973652302184419e-06, "loss": 0.0002, "step": 192850 }, { "epoch": 3.1556900924486624, "grad_norm": 0.04836384207010269, "learning_rate": 1.2968854725867792e-06, "loss": 0.0003, "step": 192860 }, { "epoch": 3.1558537183997384, "grad_norm": 0.05179351195693016, "learning_rate": 1.2964057904584704e-06, "loss": 0.0003, "step": 192870 }, { "epoch": 3.156017344350814, "grad_norm": 0.0008360932115465403, "learning_rate": 1.295926183843298e-06, "loss": 0.001, "step": 192880 }, { "epoch": 3.15618097030189, "grad_norm": 0.008218053728342056, "learning_rate": 1.2954466527510401e-06, "loss": 0.0002, "step": 192890 }, { "epoch": 3.156344596252966, "grad_norm": 0.046726882457733154, "learning_rate": 1.2949671971914762e-06, "loss": 0.0003, "step": 192900 }, { "epoch": 3.1565082222040415, "grad_norm": 0.03173869848251343, "learning_rate": 1.2944878171743785e-06, "loss": 0.0005, "step": 192910 }, { "epoch": 3.1566718481551175, "grad_norm": 0.02015022002160549, "learning_rate": 1.294008512709523e-06, "loss": 0.0004, "step": 192920 }, { "epoch": 3.1568354741061935, "grad_norm": 0.030988488346338272, "learning_rate": 1.293529283806682e-06, "loss": 0.0003, "step": 192930 }, { "epoch": 3.156999100057269, "grad_norm": 0.034380264580249786, "learning_rate": 1.2930501304756277e-06, "loss": 0.0002, "step": 192940 }, { "epoch": 3.157162726008345, "grad_norm": 0.0007854413124732673, "learning_rate": 1.2925710527261276e-06, "loss": 0.0002, "step": 192950 }, { "epoch": 3.1573263519594206, "grad_norm": 0.02640381082892418, "learning_rate": 1.2920920505679507e-06, "loss": 0.0004, "step": 192960 }, { "epoch": 3.1574899779104966, "grad_norm": 0.006653541699051857, "learning_rate": 1.2916131240108632e-06, "loss": 0.0003, "step": 192970 }, { "epoch": 3.1576536038615726, "grad_norm": 0.04820684716105461, "learning_rate": 1.291134273064632e-06, "loss": 0.0002, "step": 192980 }, { "epoch": 3.157817229812648, "grad_norm": 0.004671916831284761, "learning_rate": 1.2906554977390168e-06, "loss": 0.0004, "step": 192990 }, { "epoch": 3.157980855763724, "grad_norm": 0.02313011884689331, "learning_rate": 1.2901767980437813e-06, "loss": 0.0002, "step": 193000 }, { "epoch": 3.1581444817148, "grad_norm": 0.012832686305046082, "learning_rate": 1.289698173988686e-06, "loss": 0.0003, "step": 193010 }, { "epoch": 3.1583081076658757, "grad_norm": 0.11751144379377365, "learning_rate": 1.289219625583491e-06, "loss": 0.0007, "step": 193020 }, { "epoch": 3.1584717336169517, "grad_norm": 0.03373342752456665, "learning_rate": 1.2887411528379507e-06, "loss": 0.0003, "step": 193030 }, { "epoch": 3.1586353595680277, "grad_norm": 0.00782471802085638, "learning_rate": 1.2882627557618215e-06, "loss": 0.0005, "step": 193040 }, { "epoch": 3.158798985519103, "grad_norm": 0.0024859244003891945, "learning_rate": 1.2877844343648582e-06, "loss": 0.0004, "step": 193050 }, { "epoch": 3.158962611470179, "grad_norm": 0.00032076117349788547, "learning_rate": 1.2873061886568145e-06, "loss": 0.0002, "step": 193060 }, { "epoch": 3.159126237421255, "grad_norm": 0.06519315391778946, "learning_rate": 1.2868280186474386e-06, "loss": 0.0003, "step": 193070 }, { "epoch": 3.1592898633723308, "grad_norm": 0.002144870115444064, "learning_rate": 1.2863499243464817e-06, "loss": 0.0005, "step": 193080 }, { "epoch": 3.1594534893234067, "grad_norm": 0.0032056106720119715, "learning_rate": 1.2858719057636916e-06, "loss": 0.0004, "step": 193090 }, { "epoch": 3.1596171152744823, "grad_norm": 0.0014570059720426798, "learning_rate": 1.2853939629088152e-06, "loss": 0.0005, "step": 193100 }, { "epoch": 3.1597807412255583, "grad_norm": 0.031178796663880348, "learning_rate": 1.2849160957915964e-06, "loss": 0.0004, "step": 193110 }, { "epoch": 3.1599443671766343, "grad_norm": 0.13926905393600464, "learning_rate": 1.284438304421778e-06, "loss": 0.0004, "step": 193120 }, { "epoch": 3.16010799312771, "grad_norm": 0.004573267884552479, "learning_rate": 1.283960588809103e-06, "loss": 0.0001, "step": 193130 }, { "epoch": 3.160271619078786, "grad_norm": 0.012429730966687202, "learning_rate": 1.2834829489633126e-06, "loss": 0.0003, "step": 193140 }, { "epoch": 3.160435245029862, "grad_norm": 0.012371250428259373, "learning_rate": 1.2830053848941427e-06, "loss": 0.0001, "step": 193150 }, { "epoch": 3.1605988709809374, "grad_norm": 0.01266890112310648, "learning_rate": 1.2825278966113314e-06, "loss": 0.0004, "step": 193160 }, { "epoch": 3.1607624969320134, "grad_norm": 0.02169405110180378, "learning_rate": 1.282050484124615e-06, "loss": 0.0002, "step": 193170 }, { "epoch": 3.1609261228830894, "grad_norm": 0.06523922085762024, "learning_rate": 1.2815731474437288e-06, "loss": 0.0004, "step": 193180 }, { "epoch": 3.161089748834165, "grad_norm": 0.0212715994566679, "learning_rate": 1.2810958865784024e-06, "loss": 0.0002, "step": 193190 }, { "epoch": 3.161253374785241, "grad_norm": 0.14825236797332764, "learning_rate": 1.2806187015383676e-06, "loss": 0.0004, "step": 193200 }, { "epoch": 3.161417000736317, "grad_norm": 0.007764883805066347, "learning_rate": 1.2801415923333544e-06, "loss": 0.0004, "step": 193210 }, { "epoch": 3.1615806266873925, "grad_norm": 0.0033086577896028757, "learning_rate": 1.2796645589730922e-06, "loss": 0.0002, "step": 193220 }, { "epoch": 3.1617442526384685, "grad_norm": 0.016543474048376083, "learning_rate": 1.2791876014673038e-06, "loss": 0.0003, "step": 193230 }, { "epoch": 3.1619078785895445, "grad_norm": 0.0020741235930472612, "learning_rate": 1.2787107198257158e-06, "loss": 0.0002, "step": 193240 }, { "epoch": 3.16207150454062, "grad_norm": 0.03935086727142334, "learning_rate": 1.278233914058052e-06, "loss": 0.0003, "step": 193250 }, { "epoch": 3.162235130491696, "grad_norm": 0.03993871808052063, "learning_rate": 1.2777571841740343e-06, "loss": 0.0004, "step": 193260 }, { "epoch": 3.162398756442772, "grad_norm": 0.008842573501169682, "learning_rate": 1.2772805301833807e-06, "loss": 0.0004, "step": 193270 }, { "epoch": 3.1625623823938476, "grad_norm": 0.08763362467288971, "learning_rate": 1.2768039520958108e-06, "loss": 0.0005, "step": 193280 }, { "epoch": 3.1627260083449236, "grad_norm": 0.09957855939865112, "learning_rate": 1.2763274499210421e-06, "loss": 0.0003, "step": 193290 }, { "epoch": 3.1628896342959996, "grad_norm": 0.04701830819249153, "learning_rate": 1.275851023668791e-06, "loss": 0.0002, "step": 193300 }, { "epoch": 3.163053260247075, "grad_norm": 0.019928954541683197, "learning_rate": 1.2753746733487692e-06, "loss": 0.0005, "step": 193310 }, { "epoch": 3.163216886198151, "grad_norm": 0.04815790057182312, "learning_rate": 1.2748983989706915e-06, "loss": 0.0007, "step": 193320 }, { "epoch": 3.1633805121492267, "grad_norm": 0.08649684488773346, "learning_rate": 1.274422200544264e-06, "loss": 0.0005, "step": 193330 }, { "epoch": 3.1635441381003027, "grad_norm": 0.053817592561244965, "learning_rate": 1.273946078079203e-06, "loss": 0.0002, "step": 193340 }, { "epoch": 3.1637077640513787, "grad_norm": 0.05426526069641113, "learning_rate": 1.2734700315852106e-06, "loss": 0.0002, "step": 193350 }, { "epoch": 3.163871390002454, "grad_norm": 0.058500271290540695, "learning_rate": 1.2729940610719966e-06, "loss": 0.0002, "step": 193360 }, { "epoch": 3.16403501595353, "grad_norm": 0.035671256482601166, "learning_rate": 1.2725181665492608e-06, "loss": 0.0006, "step": 193370 }, { "epoch": 3.164198641904606, "grad_norm": 0.06464660912752151, "learning_rate": 1.2720423480267124e-06, "loss": 0.0003, "step": 193380 }, { "epoch": 3.1643622678556818, "grad_norm": 0.011439252644777298, "learning_rate": 1.2715666055140492e-06, "loss": 0.0001, "step": 193390 }, { "epoch": 3.1645258938067578, "grad_norm": 0.00046688845031894743, "learning_rate": 1.271090939020973e-06, "loss": 0.0003, "step": 193400 }, { "epoch": 3.1646895197578337, "grad_norm": 0.0009035340626724064, "learning_rate": 1.2706153485571782e-06, "loss": 0.0007, "step": 193410 }, { "epoch": 3.1648531457089093, "grad_norm": 0.0011557996040210128, "learning_rate": 1.270139834132368e-06, "loss": 0.0003, "step": 193420 }, { "epoch": 3.1650167716599853, "grad_norm": 0.055583007633686066, "learning_rate": 1.269664395756232e-06, "loss": 0.0002, "step": 193430 }, { "epoch": 3.1651803976110613, "grad_norm": 0.05517687276005745, "learning_rate": 1.2691890334384687e-06, "loss": 0.0003, "step": 193440 }, { "epoch": 3.165344023562137, "grad_norm": 0.0059104738757014275, "learning_rate": 1.2687137471887651e-06, "loss": 0.0003, "step": 193450 }, { "epoch": 3.165507649513213, "grad_norm": 0.0007504879031330347, "learning_rate": 1.2682385370168182e-06, "loss": 0.0003, "step": 193460 }, { "epoch": 3.165671275464289, "grad_norm": 0.01526866015046835, "learning_rate": 1.2677634029323116e-06, "loss": 0.0002, "step": 193470 }, { "epoch": 3.1658349014153644, "grad_norm": 0.003690106328576803, "learning_rate": 1.2672883449449368e-06, "loss": 0.0005, "step": 193480 }, { "epoch": 3.1659985273664404, "grad_norm": 0.03424202650785446, "learning_rate": 1.2668133630643753e-06, "loss": 0.0004, "step": 193490 }, { "epoch": 3.166162153317516, "grad_norm": 0.003739716252312064, "learning_rate": 1.2663384573003167e-06, "loss": 0.0002, "step": 193500 }, { "epoch": 3.166325779268592, "grad_norm": 0.043931711465120316, "learning_rate": 1.2658636276624409e-06, "loss": 0.0002, "step": 193510 }, { "epoch": 3.166489405219668, "grad_norm": 0.030923109501600266, "learning_rate": 1.2653888741604309e-06, "loss": 0.0003, "step": 193520 }, { "epoch": 3.1666530311707435, "grad_norm": 0.027857616543769836, "learning_rate": 1.2649141968039647e-06, "loss": 0.0002, "step": 193530 }, { "epoch": 3.1668166571218195, "grad_norm": 0.058494649827480316, "learning_rate": 1.2644395956027206e-06, "loss": 0.0003, "step": 193540 }, { "epoch": 3.1669802830728955, "grad_norm": 0.01403110846877098, "learning_rate": 1.2639650705663769e-06, "loss": 0.0001, "step": 193550 }, { "epoch": 3.167143909023971, "grad_norm": 0.0034683479461818933, "learning_rate": 1.2634906217046088e-06, "loss": 0.0002, "step": 193560 }, { "epoch": 3.167307534975047, "grad_norm": 0.05880054086446762, "learning_rate": 1.2630162490270887e-06, "loss": 0.0005, "step": 193570 }, { "epoch": 3.167471160926123, "grad_norm": 0.029339734464883804, "learning_rate": 1.2625419525434878e-06, "loss": 0.0005, "step": 193580 }, { "epoch": 3.1676347868771986, "grad_norm": 0.022063808515667915, "learning_rate": 1.262067732263479e-06, "loss": 0.0016, "step": 193590 }, { "epoch": 3.1677984128282746, "grad_norm": 0.03331292048096657, "learning_rate": 1.2615935881967312e-06, "loss": 0.0002, "step": 193600 }, { "epoch": 3.1679620387793506, "grad_norm": 0.001345337717793882, "learning_rate": 1.2611195203529092e-06, "loss": 0.0002, "step": 193610 }, { "epoch": 3.168125664730426, "grad_norm": 0.030490193516016006, "learning_rate": 1.2606455287416803e-06, "loss": 0.0003, "step": 193620 }, { "epoch": 3.168289290681502, "grad_norm": 0.05864826589822769, "learning_rate": 1.2601716133727094e-06, "loss": 0.0005, "step": 193630 }, { "epoch": 3.168452916632578, "grad_norm": 0.025397147983312607, "learning_rate": 1.2596977742556593e-06, "loss": 0.0003, "step": 193640 }, { "epoch": 3.1686165425836537, "grad_norm": 0.016538778319954872, "learning_rate": 1.2592240114001897e-06, "loss": 0.0002, "step": 193650 }, { "epoch": 3.1687801685347297, "grad_norm": 0.00030421625706367195, "learning_rate": 1.258750324815961e-06, "loss": 0.0002, "step": 193660 }, { "epoch": 3.1689437944858057, "grad_norm": 0.10935863852500916, "learning_rate": 1.2582767145126312e-06, "loss": 0.0004, "step": 193670 }, { "epoch": 3.169107420436881, "grad_norm": 0.07429196685552597, "learning_rate": 1.2578031804998586e-06, "loss": 0.0003, "step": 193680 }, { "epoch": 3.169271046387957, "grad_norm": 0.007430291268974543, "learning_rate": 1.257329722787295e-06, "loss": 0.0003, "step": 193690 }, { "epoch": 3.169434672339033, "grad_norm": 0.0021610106341540813, "learning_rate": 1.2568563413845952e-06, "loss": 0.0006, "step": 193700 }, { "epoch": 3.1695982982901088, "grad_norm": 0.03144911676645279, "learning_rate": 1.2563830363014112e-06, "loss": 0.0002, "step": 193710 }, { "epoch": 3.1697619242411847, "grad_norm": 0.07064589112997055, "learning_rate": 1.2559098075473946e-06, "loss": 0.0002, "step": 193720 }, { "epoch": 3.1699255501922603, "grad_norm": 0.022937241941690445, "learning_rate": 1.2554366551321917e-06, "loss": 0.0002, "step": 193730 }, { "epoch": 3.1700891761433363, "grad_norm": 0.08929896354675293, "learning_rate": 1.2549635790654508e-06, "loss": 0.0002, "step": 193740 }, { "epoch": 3.1702528020944123, "grad_norm": 0.04057818278670311, "learning_rate": 1.2544905793568173e-06, "loss": 0.0003, "step": 193750 }, { "epoch": 3.170416428045488, "grad_norm": 0.22371497750282288, "learning_rate": 1.2540176560159368e-06, "loss": 0.0005, "step": 193760 }, { "epoch": 3.170580053996564, "grad_norm": 0.06319583207368851, "learning_rate": 1.2535448090524493e-06, "loss": 0.0001, "step": 193770 }, { "epoch": 3.17074367994764, "grad_norm": 0.01678716577589512, "learning_rate": 1.2530720384759964e-06, "loss": 0.0004, "step": 193780 }, { "epoch": 3.1709073058987154, "grad_norm": 0.012338857166469097, "learning_rate": 1.252599344296218e-06, "loss": 0.0002, "step": 193790 }, { "epoch": 3.1710709318497914, "grad_norm": 0.04063068702816963, "learning_rate": 1.2521267265227538e-06, "loss": 0.0003, "step": 193800 }, { "epoch": 3.1712345578008674, "grad_norm": 0.0053977956995368, "learning_rate": 1.251654185165237e-06, "loss": 0.0001, "step": 193810 }, { "epoch": 3.171398183751943, "grad_norm": 0.04266050457954407, "learning_rate": 1.2511817202333027e-06, "loss": 0.0006, "step": 193820 }, { "epoch": 3.171561809703019, "grad_norm": 0.013353340327739716, "learning_rate": 1.2507093317365854e-06, "loss": 0.0003, "step": 193830 }, { "epoch": 3.171725435654095, "grad_norm": 0.012823224067687988, "learning_rate": 1.2502370196847175e-06, "loss": 0.0005, "step": 193840 }, { "epoch": 3.1718890616051705, "grad_norm": 0.09861620515584946, "learning_rate": 1.2497647840873262e-06, "loss": 0.0004, "step": 193850 }, { "epoch": 3.1720526875562465, "grad_norm": 0.007301584351807833, "learning_rate": 1.2492926249540416e-06, "loss": 0.0003, "step": 193860 }, { "epoch": 3.172216313507322, "grad_norm": 0.018138855695724487, "learning_rate": 1.2488205422944905e-06, "loss": 0.0004, "step": 193870 }, { "epoch": 3.172379939458398, "grad_norm": 0.05429983511567116, "learning_rate": 1.2483485361182994e-06, "loss": 0.0004, "step": 193880 }, { "epoch": 3.172543565409474, "grad_norm": 0.002221528207883239, "learning_rate": 1.24787660643509e-06, "loss": 0.0002, "step": 193890 }, { "epoch": 3.1727071913605496, "grad_norm": 0.0005386277916841209, "learning_rate": 1.2474047532544852e-06, "loss": 0.0003, "step": 193900 }, { "epoch": 3.1728708173116256, "grad_norm": 0.026488834992051125, "learning_rate": 1.246932976586106e-06, "loss": 0.0004, "step": 193910 }, { "epoch": 3.1730344432627016, "grad_norm": 0.04692493751645088, "learning_rate": 1.2464612764395734e-06, "loss": 0.0005, "step": 193920 }, { "epoch": 3.173198069213777, "grad_norm": 0.10409291833639145, "learning_rate": 1.2459896528245014e-06, "loss": 0.0001, "step": 193930 }, { "epoch": 3.173361695164853, "grad_norm": 0.09836719930171967, "learning_rate": 1.2455181057505077e-06, "loss": 0.0004, "step": 193940 }, { "epoch": 3.173525321115929, "grad_norm": 0.0011087011080235243, "learning_rate": 1.2450466352272067e-06, "loss": 0.0004, "step": 193950 }, { "epoch": 3.1736889470670047, "grad_norm": 0.02697526291012764, "learning_rate": 1.2445752412642126e-06, "loss": 0.0014, "step": 193960 }, { "epoch": 3.1738525730180807, "grad_norm": 0.02112656645476818, "learning_rate": 1.2441039238711344e-06, "loss": 0.0003, "step": 193970 }, { "epoch": 3.1740161989691567, "grad_norm": 0.010056904517114162, "learning_rate": 1.243632683057584e-06, "loss": 0.0005, "step": 193980 }, { "epoch": 3.174179824920232, "grad_norm": 0.028858691453933716, "learning_rate": 1.243161518833166e-06, "loss": 0.0003, "step": 193990 }, { "epoch": 3.174343450871308, "grad_norm": 0.06614009290933609, "learning_rate": 1.2426904312074923e-06, "loss": 0.0005, "step": 194000 }, { "epoch": 3.174507076822384, "grad_norm": 0.08464004099369049, "learning_rate": 1.2422194201901638e-06, "loss": 0.0004, "step": 194010 }, { "epoch": 3.1746707027734598, "grad_norm": 0.016358718276023865, "learning_rate": 1.241748485790787e-06, "loss": 0.0001, "step": 194020 }, { "epoch": 3.1748343287245357, "grad_norm": 0.04545595496892929, "learning_rate": 1.2412776280189592e-06, "loss": 0.0004, "step": 194030 }, { "epoch": 3.1749979546756117, "grad_norm": 7.204540452221408e-05, "learning_rate": 1.2408068468842866e-06, "loss": 0.0004, "step": 194040 }, { "epoch": 3.1751615806266873, "grad_norm": 0.007013489492237568, "learning_rate": 1.2403361423963644e-06, "loss": 0.0007, "step": 194050 }, { "epoch": 3.1753252065777633, "grad_norm": 0.022951846942305565, "learning_rate": 1.239865514564792e-06, "loss": 0.0002, "step": 194060 }, { "epoch": 3.1754888325288393, "grad_norm": 0.010672827251255512, "learning_rate": 1.2393949633991609e-06, "loss": 0.0002, "step": 194070 }, { "epoch": 3.175652458479915, "grad_norm": 0.02429889515042305, "learning_rate": 1.2389244889090707e-06, "loss": 0.0003, "step": 194080 }, { "epoch": 3.175816084430991, "grad_norm": 0.001413162099197507, "learning_rate": 1.2384540911041104e-06, "loss": 0.0004, "step": 194090 }, { "epoch": 3.1759797103820664, "grad_norm": 0.04237030819058418, "learning_rate": 1.237983769993874e-06, "loss": 0.0001, "step": 194100 }, { "epoch": 3.1761433363331424, "grad_norm": 0.0013975114561617374, "learning_rate": 1.2375135255879455e-06, "loss": 0.0003, "step": 194110 }, { "epoch": 3.1763069622842184, "grad_norm": 0.003699510358273983, "learning_rate": 1.2370433578959195e-06, "loss": 0.0005, "step": 194120 }, { "epoch": 3.176470588235294, "grad_norm": 0.009233531542122364, "learning_rate": 1.2365732669273778e-06, "loss": 0.0003, "step": 194130 }, { "epoch": 3.17663421418637, "grad_norm": 0.007653672248125076, "learning_rate": 1.2361032526919076e-06, "loss": 0.0001, "step": 194140 }, { "epoch": 3.176797840137446, "grad_norm": 0.037245918065309525, "learning_rate": 1.2356333151990901e-06, "loss": 0.0002, "step": 194150 }, { "epoch": 3.1769614660885215, "grad_norm": 0.04231451451778412, "learning_rate": 1.2351634544585072e-06, "loss": 0.0005, "step": 194160 }, { "epoch": 3.1771250920395975, "grad_norm": 0.06660721451044083, "learning_rate": 1.2346936704797401e-06, "loss": 0.0004, "step": 194170 }, { "epoch": 3.1772887179906735, "grad_norm": 0.0037668952718377113, "learning_rate": 1.2342239632723679e-06, "loss": 0.0001, "step": 194180 }, { "epoch": 3.177452343941749, "grad_norm": 0.02988518588244915, "learning_rate": 1.2337543328459656e-06, "loss": 0.0003, "step": 194190 }, { "epoch": 3.177615969892825, "grad_norm": 0.11899235099554062, "learning_rate": 1.2332847792101087e-06, "loss": 0.0004, "step": 194200 }, { "epoch": 3.177779595843901, "grad_norm": 0.0022241848055273294, "learning_rate": 1.2328153023743717e-06, "loss": 0.0002, "step": 194210 }, { "epoch": 3.1779432217949766, "grad_norm": 0.06987566500902176, "learning_rate": 1.2323459023483286e-06, "loss": 0.0003, "step": 194220 }, { "epoch": 3.1781068477460526, "grad_norm": 0.11244960874319077, "learning_rate": 1.2318765791415465e-06, "loss": 0.0002, "step": 194230 }, { "epoch": 3.1782704736971286, "grad_norm": 0.00793109368532896, "learning_rate": 1.231407332763596e-06, "loss": 0.0003, "step": 194240 }, { "epoch": 3.178434099648204, "grad_norm": 0.005484805908054113, "learning_rate": 1.2309381632240457e-06, "loss": 0.0002, "step": 194250 }, { "epoch": 3.17859772559928, "grad_norm": 0.0055795456282794476, "learning_rate": 1.2304690705324618e-06, "loss": 0.0004, "step": 194260 }, { "epoch": 3.1787613515503557, "grad_norm": 0.10305727273225784, "learning_rate": 1.2300000546984065e-06, "loss": 0.0009, "step": 194270 }, { "epoch": 3.1789249775014317, "grad_norm": 0.042701657861471176, "learning_rate": 1.2295311157314432e-06, "loss": 0.0003, "step": 194280 }, { "epoch": 3.1790886034525077, "grad_norm": 0.04089350998401642, "learning_rate": 1.2290622536411345e-06, "loss": 0.0003, "step": 194290 }, { "epoch": 3.179252229403583, "grad_norm": 0.03123556450009346, "learning_rate": 1.2285934684370404e-06, "loss": 0.0002, "step": 194300 }, { "epoch": 3.179415855354659, "grad_norm": 0.02682182751595974, "learning_rate": 1.2281247601287166e-06, "loss": 0.0005, "step": 194310 }, { "epoch": 3.179579481305735, "grad_norm": 0.021891245618462563, "learning_rate": 1.2276561287257211e-06, "loss": 0.0002, "step": 194320 }, { "epoch": 3.1797431072568108, "grad_norm": 0.03941091522574425, "learning_rate": 1.2271875742376094e-06, "loss": 0.0006, "step": 194330 }, { "epoch": 3.1799067332078867, "grad_norm": 0.05157089978456497, "learning_rate": 1.226719096673935e-06, "loss": 0.0002, "step": 194340 }, { "epoch": 3.1800703591589627, "grad_norm": 0.017976773902773857, "learning_rate": 1.2262506960442483e-06, "loss": 0.0003, "step": 194350 }, { "epoch": 3.1802339851100383, "grad_norm": 0.04274388775229454, "learning_rate": 1.2257823723581003e-06, "loss": 0.0003, "step": 194360 }, { "epoch": 3.1803976110611143, "grad_norm": 0.029977191239595413, "learning_rate": 1.2253141256250394e-06, "loss": 0.0002, "step": 194370 }, { "epoch": 3.1805612370121903, "grad_norm": 0.09305620938539505, "learning_rate": 1.2248459558546154e-06, "loss": 0.0005, "step": 194380 }, { "epoch": 3.180724862963266, "grad_norm": 0.03635571151971817, "learning_rate": 1.2243778630563696e-06, "loss": 0.0003, "step": 194390 }, { "epoch": 3.180888488914342, "grad_norm": 0.011571737937629223, "learning_rate": 1.2239098472398487e-06, "loss": 0.0001, "step": 194400 }, { "epoch": 3.181052114865418, "grad_norm": 0.056044235825538635, "learning_rate": 1.2234419084145942e-06, "loss": 0.0005, "step": 194410 }, { "epoch": 3.1812157408164934, "grad_norm": 0.029420649632811546, "learning_rate": 1.222974046590149e-06, "loss": 0.0005, "step": 194420 }, { "epoch": 3.1813793667675694, "grad_norm": 0.0020625207107514143, "learning_rate": 1.2225062617760491e-06, "loss": 0.0006, "step": 194430 }, { "epoch": 3.1815429927186454, "grad_norm": 0.004946212284266949, "learning_rate": 1.2220385539818341e-06, "loss": 0.0001, "step": 194440 }, { "epoch": 3.181706618669721, "grad_norm": 0.017017396166920662, "learning_rate": 1.2215709232170397e-06, "loss": 0.0003, "step": 194450 }, { "epoch": 3.181870244620797, "grad_norm": 0.02018539048731327, "learning_rate": 1.2211033694912021e-06, "loss": 0.0003, "step": 194460 }, { "epoch": 3.182033870571873, "grad_norm": 0.03854047879576683, "learning_rate": 1.2206358928138517e-06, "loss": 0.0007, "step": 194470 }, { "epoch": 3.1821974965229485, "grad_norm": 0.02599225752055645, "learning_rate": 1.2201684931945213e-06, "loss": 0.0002, "step": 194480 }, { "epoch": 3.1823611224740245, "grad_norm": 0.0397331528365612, "learning_rate": 1.2197011706427408e-06, "loss": 0.0001, "step": 194490 }, { "epoch": 3.1825247484251, "grad_norm": 0.009373139590024948, "learning_rate": 1.2192339251680395e-06, "loss": 0.0004, "step": 194500 }, { "epoch": 3.182688374376176, "grad_norm": 0.0354442298412323, "learning_rate": 1.2187667567799422e-06, "loss": 0.0003, "step": 194510 }, { "epoch": 3.182852000327252, "grad_norm": 0.002441320801153779, "learning_rate": 1.2182996654879742e-06, "loss": 0.0003, "step": 194520 }, { "epoch": 3.1830156262783276, "grad_norm": 0.04678329452872276, "learning_rate": 1.2178326513016604e-06, "loss": 0.0001, "step": 194530 }, { "epoch": 3.1831792522294036, "grad_norm": 0.013965362682938576, "learning_rate": 1.2173657142305229e-06, "loss": 0.0003, "step": 194540 }, { "epoch": 3.1833428781804796, "grad_norm": 0.06368423253297806, "learning_rate": 1.2168988542840805e-06, "loss": 0.0003, "step": 194550 }, { "epoch": 3.183506504131555, "grad_norm": 0.05753912031650543, "learning_rate": 1.2164320714718531e-06, "loss": 0.0005, "step": 194560 }, { "epoch": 3.183670130082631, "grad_norm": 0.027476761490106583, "learning_rate": 1.2159653658033577e-06, "loss": 0.0003, "step": 194570 }, { "epoch": 3.183833756033707, "grad_norm": 0.0020815348252654076, "learning_rate": 1.2154987372881121e-06, "loss": 0.0001, "step": 194580 }, { "epoch": 3.1839973819847827, "grad_norm": 0.02018493041396141, "learning_rate": 1.2150321859356269e-06, "loss": 0.0008, "step": 194590 }, { "epoch": 3.1841610079358587, "grad_norm": 0.02434912696480751, "learning_rate": 1.2145657117554166e-06, "loss": 0.0001, "step": 194600 }, { "epoch": 3.1843246338869347, "grad_norm": 0.03547298535704613, "learning_rate": 1.2140993147569919e-06, "loss": 0.0007, "step": 194610 }, { "epoch": 3.18448825983801, "grad_norm": 0.0029675783589482307, "learning_rate": 1.2136329949498638e-06, "loss": 0.0001, "step": 194620 }, { "epoch": 3.184651885789086, "grad_norm": 0.017010528594255447, "learning_rate": 1.2131667523435375e-06, "loss": 0.0004, "step": 194630 }, { "epoch": 3.184815511740162, "grad_norm": 0.03759729117155075, "learning_rate": 1.2127005869475216e-06, "loss": 0.0014, "step": 194640 }, { "epoch": 3.1849791376912377, "grad_norm": 0.0029743886552751064, "learning_rate": 1.2122344987713175e-06, "loss": 0.0002, "step": 194650 }, { "epoch": 3.1851427636423137, "grad_norm": 0.06011025607585907, "learning_rate": 1.211768487824433e-06, "loss": 0.0003, "step": 194660 }, { "epoch": 3.1853063895933893, "grad_norm": 0.20243266224861145, "learning_rate": 1.211302554116366e-06, "loss": 0.0004, "step": 194670 }, { "epoch": 3.1854700155444653, "grad_norm": 0.0048103188164532185, "learning_rate": 1.2108366976566188e-06, "loss": 0.0004, "step": 194680 }, { "epoch": 3.1856336414955413, "grad_norm": 0.04396907985210419, "learning_rate": 1.2103709184546863e-06, "loss": 0.0004, "step": 194690 }, { "epoch": 3.185797267446617, "grad_norm": 0.027404660359025, "learning_rate": 1.209905216520071e-06, "loss": 0.0005, "step": 194700 }, { "epoch": 3.185960893397693, "grad_norm": 0.008357656188309193, "learning_rate": 1.2094395918622632e-06, "loss": 0.0008, "step": 194710 }, { "epoch": 3.186124519348769, "grad_norm": 0.10217677801847458, "learning_rate": 1.2089740444907593e-06, "loss": 0.0002, "step": 194720 }, { "epoch": 3.1862881452998444, "grad_norm": 0.0009086029604077339, "learning_rate": 1.2085085744150482e-06, "loss": 0.0003, "step": 194730 }, { "epoch": 3.1864517712509204, "grad_norm": 0.016586042940616608, "learning_rate": 1.2080431816446253e-06, "loss": 0.0003, "step": 194740 }, { "epoch": 3.1866153972019964, "grad_norm": 0.010992259718477726, "learning_rate": 1.2075778661889759e-06, "loss": 0.0002, "step": 194750 }, { "epoch": 3.186779023153072, "grad_norm": 0.0070226457901299, "learning_rate": 1.2071126280575895e-06, "loss": 0.0008, "step": 194760 }, { "epoch": 3.186942649104148, "grad_norm": 0.0017706335056573153, "learning_rate": 1.2066474672599482e-06, "loss": 0.0005, "step": 194770 }, { "epoch": 3.187106275055224, "grad_norm": 0.013128450140357018, "learning_rate": 1.206182383805542e-06, "loss": 0.0007, "step": 194780 }, { "epoch": 3.1872699010062995, "grad_norm": 0.02044645883142948, "learning_rate": 1.2057173777038484e-06, "loss": 0.0003, "step": 194790 }, { "epoch": 3.1874335269573755, "grad_norm": 0.00015433992666658014, "learning_rate": 1.2052524489643525e-06, "loss": 0.0003, "step": 194800 }, { "epoch": 3.1875971529084515, "grad_norm": 0.19145047664642334, "learning_rate": 1.20478759759653e-06, "loss": 0.0005, "step": 194810 }, { "epoch": 3.187760778859527, "grad_norm": 0.022132012993097305, "learning_rate": 1.2043228236098613e-06, "loss": 0.0007, "step": 194820 }, { "epoch": 3.187924404810603, "grad_norm": 0.06629137694835663, "learning_rate": 1.2038581270138217e-06, "loss": 0.0006, "step": 194830 }, { "epoch": 3.188088030761679, "grad_norm": 0.08251287788152695, "learning_rate": 1.2033935078178876e-06, "loss": 0.0003, "step": 194840 }, { "epoch": 3.1882516567127546, "grad_norm": 0.026549572125077248, "learning_rate": 1.2029289660315302e-06, "loss": 0.0004, "step": 194850 }, { "epoch": 3.1884152826638306, "grad_norm": 0.049246009439229965, "learning_rate": 1.2024645016642216e-06, "loss": 0.0002, "step": 194860 }, { "epoch": 3.1885789086149066, "grad_norm": 0.03284885361790657, "learning_rate": 1.2020001147254317e-06, "loss": 0.0001, "step": 194870 }, { "epoch": 3.188742534565982, "grad_norm": 0.03745295852422714, "learning_rate": 1.2015358052246313e-06, "loss": 0.0003, "step": 194880 }, { "epoch": 3.188906160517058, "grad_norm": 0.008865587413311005, "learning_rate": 1.2010715731712842e-06, "loss": 0.0009, "step": 194890 }, { "epoch": 3.1890697864681337, "grad_norm": 0.01422471459954977, "learning_rate": 1.2006074185748563e-06, "loss": 0.0002, "step": 194900 }, { "epoch": 3.1892334124192097, "grad_norm": 0.027507297694683075, "learning_rate": 1.200143341444812e-06, "loss": 0.0001, "step": 194910 }, { "epoch": 3.1893970383702857, "grad_norm": 0.0403134860098362, "learning_rate": 1.1996793417906145e-06, "loss": 0.0005, "step": 194920 }, { "epoch": 3.189560664321361, "grad_norm": 0.07320832461118698, "learning_rate": 1.199215419621722e-06, "loss": 0.0007, "step": 194930 }, { "epoch": 3.189724290272437, "grad_norm": 0.06463044136762619, "learning_rate": 1.1987515749475948e-06, "loss": 0.0005, "step": 194940 }, { "epoch": 3.189887916223513, "grad_norm": 0.030678480863571167, "learning_rate": 1.19828780777769e-06, "loss": 0.0002, "step": 194950 }, { "epoch": 3.1900515421745888, "grad_norm": 0.017267240211367607, "learning_rate": 1.197824118121465e-06, "loss": 0.0003, "step": 194960 }, { "epoch": 3.1902151681256647, "grad_norm": 0.018229402601718903, "learning_rate": 1.197360505988371e-06, "loss": 0.0003, "step": 194970 }, { "epoch": 3.1903787940767407, "grad_norm": 0.022469164803624153, "learning_rate": 1.1968969713878626e-06, "loss": 0.0006, "step": 194980 }, { "epoch": 3.1905424200278163, "grad_norm": 0.026258621364831924, "learning_rate": 1.1964335143293898e-06, "loss": 0.0003, "step": 194990 }, { "epoch": 3.1907060459788923, "grad_norm": 0.160951629281044, "learning_rate": 1.1959701348224052e-06, "loss": 0.0004, "step": 195000 }, { "epoch": 3.1908696719299683, "grad_norm": 0.00045030174078419805, "learning_rate": 1.1955068328763524e-06, "loss": 0.0003, "step": 195010 }, { "epoch": 3.191033297881044, "grad_norm": 0.03176800161600113, "learning_rate": 1.1950436085006795e-06, "loss": 0.0002, "step": 195020 }, { "epoch": 3.19119692383212, "grad_norm": 0.029734592884778976, "learning_rate": 1.1945804617048317e-06, "loss": 0.0005, "step": 195030 }, { "epoch": 3.1913605497831954, "grad_norm": 0.04400445520877838, "learning_rate": 1.1941173924982535e-06, "loss": 0.0003, "step": 195040 }, { "epoch": 3.1915241757342714, "grad_norm": 0.03538262099027634, "learning_rate": 1.1936544008903833e-06, "loss": 0.0013, "step": 195050 }, { "epoch": 3.1916878016853474, "grad_norm": 0.00022168378927744925, "learning_rate": 1.1931914868906624e-06, "loss": 0.0006, "step": 195060 }, { "epoch": 3.191851427636423, "grad_norm": 0.030790871009230614, "learning_rate": 1.1927286505085296e-06, "loss": 0.0003, "step": 195070 }, { "epoch": 3.192015053587499, "grad_norm": 0.0012860047863796353, "learning_rate": 1.1922658917534236e-06, "loss": 0.0003, "step": 195080 }, { "epoch": 3.192178679538575, "grad_norm": 0.00943128764629364, "learning_rate": 1.1918032106347753e-06, "loss": 0.0002, "step": 195090 }, { "epoch": 3.1923423054896505, "grad_norm": 0.04234081506729126, "learning_rate": 1.1913406071620215e-06, "loss": 0.0006, "step": 195100 }, { "epoch": 3.1925059314407265, "grad_norm": 0.010912633500993252, "learning_rate": 1.1908780813445937e-06, "loss": 0.0002, "step": 195110 }, { "epoch": 3.1926695573918025, "grad_norm": 0.042725421488285065, "learning_rate": 1.1904156331919232e-06, "loss": 0.0002, "step": 195120 }, { "epoch": 3.192833183342878, "grad_norm": 0.02539142407476902, "learning_rate": 1.189953262713437e-06, "loss": 0.0003, "step": 195130 }, { "epoch": 3.192996809293954, "grad_norm": 0.0110963499173522, "learning_rate": 1.1894909699185631e-06, "loss": 0.0002, "step": 195140 }, { "epoch": 3.19316043524503, "grad_norm": 0.02629123069345951, "learning_rate": 1.1890287548167279e-06, "loss": 0.0003, "step": 195150 }, { "epoch": 3.1933240611961056, "grad_norm": 0.033233895897865295, "learning_rate": 1.1885666174173565e-06, "loss": 0.0008, "step": 195160 }, { "epoch": 3.1934876871471816, "grad_norm": 0.0023286358918994665, "learning_rate": 1.1881045577298688e-06, "loss": 0.0002, "step": 195170 }, { "epoch": 3.1936513130982576, "grad_norm": 0.08486637473106384, "learning_rate": 1.1876425757636877e-06, "loss": 0.0003, "step": 195180 }, { "epoch": 3.193814939049333, "grad_norm": 0.00018726444977801293, "learning_rate": 1.1871806715282319e-06, "loss": 0.0001, "step": 195190 }, { "epoch": 3.193978565000409, "grad_norm": 0.0692695751786232, "learning_rate": 1.1867188450329209e-06, "loss": 0.0004, "step": 195200 }, { "epoch": 3.194142190951485, "grad_norm": 0.025610676035284996, "learning_rate": 1.1862570962871679e-06, "loss": 0.0005, "step": 195210 }, { "epoch": 3.1943058169025607, "grad_norm": 0.009985147975385189, "learning_rate": 1.1857954253003894e-06, "loss": 0.0003, "step": 195220 }, { "epoch": 3.1944694428536367, "grad_norm": 0.38518792390823364, "learning_rate": 1.1853338320819985e-06, "loss": 0.0007, "step": 195230 }, { "epoch": 3.1946330688047127, "grad_norm": 0.024243665859103203, "learning_rate": 1.1848723166414077e-06, "loss": 0.0002, "step": 195240 }, { "epoch": 3.194796694755788, "grad_norm": 0.04761156067252159, "learning_rate": 1.1844108789880242e-06, "loss": 0.0002, "step": 195250 }, { "epoch": 3.194960320706864, "grad_norm": 0.023277193307876587, "learning_rate": 1.18394951913126e-06, "loss": 0.0004, "step": 195260 }, { "epoch": 3.1951239466579398, "grad_norm": 0.0019279236439615488, "learning_rate": 1.1834882370805162e-06, "loss": 0.0002, "step": 195270 }, { "epoch": 3.1952875726090157, "grad_norm": 0.000490981328766793, "learning_rate": 1.1830270328452042e-06, "loss": 0.0002, "step": 195280 }, { "epoch": 3.1954511985600917, "grad_norm": 0.044132642447948456, "learning_rate": 1.1825659064347233e-06, "loss": 0.0004, "step": 195290 }, { "epoch": 3.1956148245111673, "grad_norm": 0.004758263938128948, "learning_rate": 1.1821048578584787e-06, "loss": 0.0001, "step": 195300 }, { "epoch": 3.1957784504622433, "grad_norm": 0.020781394094228745, "learning_rate": 1.1816438871258662e-06, "loss": 0.0004, "step": 195310 }, { "epoch": 3.1959420764133193, "grad_norm": 0.027230406180024147, "learning_rate": 1.1811829942462904e-06, "loss": 0.0003, "step": 195320 }, { "epoch": 3.196105702364395, "grad_norm": 0.05301428213715553, "learning_rate": 1.180722179229144e-06, "loss": 0.0002, "step": 195330 }, { "epoch": 3.196269328315471, "grad_norm": 0.08381112664937973, "learning_rate": 1.1802614420838254e-06, "loss": 0.0006, "step": 195340 }, { "epoch": 3.196432954266547, "grad_norm": 0.01365706231445074, "learning_rate": 1.1798007828197244e-06, "loss": 0.0003, "step": 195350 }, { "epoch": 3.1965965802176224, "grad_norm": 0.04294325038790703, "learning_rate": 1.1793402014462395e-06, "loss": 0.0006, "step": 195360 }, { "epoch": 3.1967602061686984, "grad_norm": 0.05623805522918701, "learning_rate": 1.1788796979727574e-06, "loss": 0.0003, "step": 195370 }, { "epoch": 3.1969238321197744, "grad_norm": 0.02489914372563362, "learning_rate": 1.1784192724086701e-06, "loss": 0.0003, "step": 195380 }, { "epoch": 3.19708745807085, "grad_norm": 0.047158777713775635, "learning_rate": 1.1779589247633606e-06, "loss": 0.0003, "step": 195390 }, { "epoch": 3.197251084021926, "grad_norm": 0.045824889093637466, "learning_rate": 1.177498655046221e-06, "loss": 0.0002, "step": 195400 }, { "epoch": 3.197414709973002, "grad_norm": 0.0001998302759602666, "learning_rate": 1.177038463266631e-06, "loss": 0.0003, "step": 195410 }, { "epoch": 3.1975783359240775, "grad_norm": 0.03446882590651512, "learning_rate": 1.1765783494339771e-06, "loss": 0.0002, "step": 195420 }, { "epoch": 3.1977419618751535, "grad_norm": 0.03698503226041794, "learning_rate": 1.1761183135576364e-06, "loss": 0.0003, "step": 195430 }, { "epoch": 3.197905587826229, "grad_norm": 0.026964055374264717, "learning_rate": 1.1756583556469937e-06, "loss": 0.0002, "step": 195440 }, { "epoch": 3.198069213777305, "grad_norm": 0.054680924862623215, "learning_rate": 1.1751984757114233e-06, "loss": 0.0002, "step": 195450 }, { "epoch": 3.198232839728381, "grad_norm": 0.029134906828403473, "learning_rate": 1.174738673760305e-06, "loss": 0.0004, "step": 195460 }, { "epoch": 3.1983964656794566, "grad_norm": 0.011691045947372913, "learning_rate": 1.17427894980301e-06, "loss": 0.0002, "step": 195470 }, { "epoch": 3.1985600916305326, "grad_norm": 0.002990872599184513, "learning_rate": 1.1738193038489133e-06, "loss": 0.0006, "step": 195480 }, { "epoch": 3.1987237175816086, "grad_norm": 0.0012051622616127133, "learning_rate": 1.1733597359073872e-06, "loss": 0.0001, "step": 195490 }, { "epoch": 3.198887343532684, "grad_norm": 0.0351952500641346, "learning_rate": 1.172900245987803e-06, "loss": 0.0004, "step": 195500 }, { "epoch": 3.19905096948376, "grad_norm": 0.006420432589948177, "learning_rate": 1.1724408340995269e-06, "loss": 0.0002, "step": 195510 }, { "epoch": 3.199214595434836, "grad_norm": 0.0024088756181299686, "learning_rate": 1.1719815002519268e-06, "loss": 0.0001, "step": 195520 }, { "epoch": 3.1993782213859117, "grad_norm": 0.05369970574975014, "learning_rate": 1.1715222444543683e-06, "loss": 0.0003, "step": 195530 }, { "epoch": 3.1995418473369877, "grad_norm": 0.08777157962322235, "learning_rate": 1.1710630667162165e-06, "loss": 0.0003, "step": 195540 }, { "epoch": 3.1997054732880637, "grad_norm": 0.026002438738942146, "learning_rate": 1.1706039670468311e-06, "loss": 0.0003, "step": 195550 }, { "epoch": 3.199869099239139, "grad_norm": 0.029073823243379593, "learning_rate": 1.1701449454555736e-06, "loss": 0.0003, "step": 195560 }, { "epoch": 3.200032725190215, "grad_norm": 0.06002594158053398, "learning_rate": 1.169686001951804e-06, "loss": 0.0003, "step": 195570 }, { "epoch": 3.200196351141291, "grad_norm": 0.056572094559669495, "learning_rate": 1.1692271365448803e-06, "loss": 0.0005, "step": 195580 }, { "epoch": 3.2003599770923667, "grad_norm": 0.003215261735022068, "learning_rate": 1.168768349244156e-06, "loss": 0.0003, "step": 195590 }, { "epoch": 3.2005236030434427, "grad_norm": 0.0063358270563185215, "learning_rate": 1.168309640058987e-06, "loss": 0.0002, "step": 195600 }, { "epoch": 3.2006872289945187, "grad_norm": 0.0372551791369915, "learning_rate": 1.1678510089987256e-06, "loss": 0.0002, "step": 195610 }, { "epoch": 3.2008508549455943, "grad_norm": 0.010794843547046185, "learning_rate": 1.1673924560727245e-06, "loss": 0.0002, "step": 195620 }, { "epoch": 3.2010144808966703, "grad_norm": 0.017049161717295647, "learning_rate": 1.1669339812903302e-06, "loss": 0.0003, "step": 195630 }, { "epoch": 3.2011781068477463, "grad_norm": 0.017066672444343567, "learning_rate": 1.1664755846608917e-06, "loss": 0.0006, "step": 195640 }, { "epoch": 3.201341732798822, "grad_norm": 0.006641680374741554, "learning_rate": 1.1660172661937563e-06, "loss": 0.0003, "step": 195650 }, { "epoch": 3.201505358749898, "grad_norm": 0.033245597034692764, "learning_rate": 1.1655590258982691e-06, "loss": 0.0002, "step": 195660 }, { "epoch": 3.2016689847009734, "grad_norm": 0.04638093709945679, "learning_rate": 1.1651008637837714e-06, "loss": 0.0003, "step": 195670 }, { "epoch": 3.2018326106520494, "grad_norm": 0.01207799557596445, "learning_rate": 1.164642779859605e-06, "loss": 0.0004, "step": 195680 }, { "epoch": 3.2019962366031254, "grad_norm": 0.061930134892463684, "learning_rate": 1.164184774135111e-06, "loss": 0.0001, "step": 195690 }, { "epoch": 3.202159862554201, "grad_norm": 0.0003753162454813719, "learning_rate": 1.1637268466196283e-06, "loss": 0.0006, "step": 195700 }, { "epoch": 3.202323488505277, "grad_norm": 0.027038278058171272, "learning_rate": 1.1632689973224915e-06, "loss": 0.0004, "step": 195710 }, { "epoch": 3.202487114456353, "grad_norm": 0.010928408242762089, "learning_rate": 1.1628112262530367e-06, "loss": 0.0003, "step": 195720 }, { "epoch": 3.2026507404074285, "grad_norm": 0.0008701082551851869, "learning_rate": 1.1623535334205972e-06, "loss": 0.0003, "step": 195730 }, { "epoch": 3.2028143663585045, "grad_norm": 0.041293807327747345, "learning_rate": 1.1618959188345069e-06, "loss": 0.0003, "step": 195740 }, { "epoch": 3.2029779923095805, "grad_norm": 0.014142287895083427, "learning_rate": 1.1614383825040938e-06, "loss": 0.0003, "step": 195750 }, { "epoch": 3.203141618260656, "grad_norm": 0.010642712004482746, "learning_rate": 1.160980924438687e-06, "loss": 0.0006, "step": 195760 }, { "epoch": 3.203305244211732, "grad_norm": 0.0012242867378517985, "learning_rate": 1.1605235446476143e-06, "loss": 0.0003, "step": 195770 }, { "epoch": 3.203468870162808, "grad_norm": 0.030497301369905472, "learning_rate": 1.1600662431402028e-06, "loss": 0.0002, "step": 195780 }, { "epoch": 3.2036324961138836, "grad_norm": 0.08164279907941818, "learning_rate": 1.1596090199257726e-06, "loss": 0.0005, "step": 195790 }, { "epoch": 3.2037961220649596, "grad_norm": 0.09631805866956711, "learning_rate": 1.1591518750136494e-06, "loss": 0.0004, "step": 195800 }, { "epoch": 3.203959748016035, "grad_norm": 0.027728402987122536, "learning_rate": 1.1586948084131522e-06, "loss": 0.0003, "step": 195810 }, { "epoch": 3.204123373967111, "grad_norm": 0.007403665687888861, "learning_rate": 1.1582378201336025e-06, "loss": 0.0007, "step": 195820 }, { "epoch": 3.204286999918187, "grad_norm": 0.030744438990950584, "learning_rate": 1.1577809101843152e-06, "loss": 0.0003, "step": 195830 }, { "epoch": 3.2044506258692627, "grad_norm": 0.005737545900046825, "learning_rate": 1.157324078574607e-06, "loss": 0.0003, "step": 195840 }, { "epoch": 3.2046142518203387, "grad_norm": 0.10176354646682739, "learning_rate": 1.1568673253137936e-06, "loss": 0.0003, "step": 195850 }, { "epoch": 3.2047778777714147, "grad_norm": 0.09395482391119003, "learning_rate": 1.1564106504111876e-06, "loss": 0.0003, "step": 195860 }, { "epoch": 3.20494150372249, "grad_norm": 0.02126496285200119, "learning_rate": 1.1559540538760984e-06, "loss": 0.0005, "step": 195870 }, { "epoch": 3.205105129673566, "grad_norm": 0.0024098383728414774, "learning_rate": 1.1554975357178365e-06, "loss": 0.0002, "step": 195880 }, { "epoch": 3.205268755624642, "grad_norm": 0.003240377176553011, "learning_rate": 1.1550410959457109e-06, "loss": 0.0003, "step": 195890 }, { "epoch": 3.2054323815757177, "grad_norm": 0.03783790022134781, "learning_rate": 1.154584734569028e-06, "loss": 0.0005, "step": 195900 }, { "epoch": 3.2055960075267937, "grad_norm": 0.017801053822040558, "learning_rate": 1.1541284515970914e-06, "loss": 0.0002, "step": 195910 }, { "epoch": 3.2057596334778697, "grad_norm": 0.027277357876300812, "learning_rate": 1.1536722470392059e-06, "loss": 0.0001, "step": 195920 }, { "epoch": 3.2059232594289453, "grad_norm": 0.032006435096263885, "learning_rate": 1.153216120904669e-06, "loss": 0.0002, "step": 195930 }, { "epoch": 3.2060868853800213, "grad_norm": 0.057788025587797165, "learning_rate": 1.1527600732027866e-06, "loss": 0.0001, "step": 195940 }, { "epoch": 3.2062505113310973, "grad_norm": 0.008048279210925102, "learning_rate": 1.1523041039428535e-06, "loss": 0.0002, "step": 195950 }, { "epoch": 3.206414137282173, "grad_norm": 0.013978268019855022, "learning_rate": 1.1518482131341685e-06, "loss": 0.0002, "step": 195960 }, { "epoch": 3.206577763233249, "grad_norm": 0.007668199483305216, "learning_rate": 1.1513924007860227e-06, "loss": 0.0004, "step": 195970 }, { "epoch": 3.206741389184325, "grad_norm": 0.0546359121799469, "learning_rate": 1.1509366669077155e-06, "loss": 0.0001, "step": 195980 }, { "epoch": 3.2069050151354004, "grad_norm": 0.048120513558387756, "learning_rate": 1.1504810115085352e-06, "loss": 0.0004, "step": 195990 }, { "epoch": 3.2070686410864764, "grad_norm": 0.0008913712226785719, "learning_rate": 1.150025434597774e-06, "loss": 0.0002, "step": 196000 }, { "epoch": 3.2072322670375524, "grad_norm": 0.0020577488467097282, "learning_rate": 1.1495699361847173e-06, "loss": 0.0003, "step": 196010 }, { "epoch": 3.207395892988628, "grad_norm": 0.0818093791604042, "learning_rate": 1.1491145162786582e-06, "loss": 0.0004, "step": 196020 }, { "epoch": 3.207559518939704, "grad_norm": 0.03257383033633232, "learning_rate": 1.148659174888877e-06, "loss": 0.0003, "step": 196030 }, { "epoch": 3.2077231448907795, "grad_norm": 0.06863649934530258, "learning_rate": 1.1482039120246612e-06, "loss": 0.0003, "step": 196040 }, { "epoch": 3.2078867708418555, "grad_norm": 0.08263914287090302, "learning_rate": 1.1477487276952898e-06, "loss": 0.0002, "step": 196050 }, { "epoch": 3.2080503967929315, "grad_norm": 0.0003055808483622968, "learning_rate": 1.1472936219100478e-06, "loss": 0.0002, "step": 196060 }, { "epoch": 3.208214022744007, "grad_norm": 0.09669903665781021, "learning_rate": 1.1468385946782106e-06, "loss": 0.0004, "step": 196070 }, { "epoch": 3.208377648695083, "grad_norm": 0.07321848720312119, "learning_rate": 1.1463836460090594e-06, "loss": 0.0002, "step": 196080 }, { "epoch": 3.208541274646159, "grad_norm": 0.0002536409592721611, "learning_rate": 1.1459287759118665e-06, "loss": 0.0007, "step": 196090 }, { "epoch": 3.2087049005972346, "grad_norm": 0.04217373579740524, "learning_rate": 1.1454739843959084e-06, "loss": 0.0002, "step": 196100 }, { "epoch": 3.2088685265483106, "grad_norm": 0.012638727203011513, "learning_rate": 1.1450192714704573e-06, "loss": 0.0003, "step": 196110 }, { "epoch": 3.2090321524993866, "grad_norm": 0.10886891186237335, "learning_rate": 1.144564637144786e-06, "loss": 0.0005, "step": 196120 }, { "epoch": 3.209195778450462, "grad_norm": 0.064507395029068, "learning_rate": 1.1441100814281618e-06, "loss": 0.0003, "step": 196130 }, { "epoch": 3.209359404401538, "grad_norm": 0.003772893687710166, "learning_rate": 1.1436556043298535e-06, "loss": 0.0005, "step": 196140 }, { "epoch": 3.209523030352614, "grad_norm": 0.04202594980597496, "learning_rate": 1.1432012058591279e-06, "loss": 0.0003, "step": 196150 }, { "epoch": 3.2096866563036897, "grad_norm": 0.022535499185323715, "learning_rate": 1.142746886025251e-06, "loss": 0.0003, "step": 196160 }, { "epoch": 3.2098502822547657, "grad_norm": 0.013862261548638344, "learning_rate": 1.1422926448374833e-06, "loss": 0.0004, "step": 196170 }, { "epoch": 3.2100139082058416, "grad_norm": 0.04813162609934807, "learning_rate": 1.1418384823050871e-06, "loss": 0.0004, "step": 196180 }, { "epoch": 3.210177534156917, "grad_norm": 0.041405998170375824, "learning_rate": 1.1413843984373235e-06, "loss": 0.0006, "step": 196190 }, { "epoch": 3.210341160107993, "grad_norm": 0.005350063089281321, "learning_rate": 1.1409303932434519e-06, "loss": 0.0004, "step": 196200 }, { "epoch": 3.2105047860590687, "grad_norm": 0.07562971115112305, "learning_rate": 1.1404764667327262e-06, "loss": 0.0003, "step": 196210 }, { "epoch": 3.2106684120101447, "grad_norm": 0.024525344371795654, "learning_rate": 1.1400226189144025e-06, "loss": 0.0009, "step": 196220 }, { "epoch": 3.2108320379612207, "grad_norm": 0.046247515827417374, "learning_rate": 1.1395688497977352e-06, "loss": 0.0003, "step": 196230 }, { "epoch": 3.2109956639122963, "grad_norm": 0.011395609006285667, "learning_rate": 1.1391151593919768e-06, "loss": 0.0002, "step": 196240 }, { "epoch": 3.2111592898633723, "grad_norm": 0.0012784821446985006, "learning_rate": 1.1386615477063757e-06, "loss": 0.0004, "step": 196250 }, { "epoch": 3.2113229158144483, "grad_norm": 0.011622060090303421, "learning_rate": 1.1382080147501817e-06, "loss": 0.0002, "step": 196260 }, { "epoch": 3.211486541765524, "grad_norm": 0.040223076939582825, "learning_rate": 1.1377545605326417e-06, "loss": 0.0002, "step": 196270 }, { "epoch": 3.2116501677166, "grad_norm": 0.003639867063611746, "learning_rate": 1.1373011850630034e-06, "loss": 0.0001, "step": 196280 }, { "epoch": 3.211813793667676, "grad_norm": 0.07587603479623795, "learning_rate": 1.1368478883505068e-06, "loss": 0.0004, "step": 196290 }, { "epoch": 3.2119774196187514, "grad_norm": 0.004283803515136242, "learning_rate": 1.1363946704043966e-06, "loss": 0.0004, "step": 196300 }, { "epoch": 3.2121410455698274, "grad_norm": 0.03833172842860222, "learning_rate": 1.1359415312339134e-06, "loss": 0.0001, "step": 196310 }, { "epoch": 3.2123046715209034, "grad_norm": 0.12747034430503845, "learning_rate": 1.1354884708482972e-06, "loss": 0.0004, "step": 196320 }, { "epoch": 3.212468297471979, "grad_norm": 0.06877943873405457, "learning_rate": 1.1350354892567834e-06, "loss": 0.0003, "step": 196330 }, { "epoch": 3.212631923423055, "grad_norm": 0.08111510425806046, "learning_rate": 1.1345825864686083e-06, "loss": 0.0004, "step": 196340 }, { "epoch": 3.212795549374131, "grad_norm": 0.007306294050067663, "learning_rate": 1.1341297624930069e-06, "loss": 0.0009, "step": 196350 }, { "epoch": 3.2129591753252065, "grad_norm": 0.011214678175747395, "learning_rate": 1.1336770173392136e-06, "loss": 0.0004, "step": 196360 }, { "epoch": 3.2131228012762825, "grad_norm": 0.03845429793000221, "learning_rate": 1.1332243510164565e-06, "loss": 0.0003, "step": 196370 }, { "epoch": 3.2132864272273585, "grad_norm": 0.0220949687063694, "learning_rate": 1.1327717635339657e-06, "loss": 0.0002, "step": 196380 }, { "epoch": 3.213450053178434, "grad_norm": 0.010919306427240372, "learning_rate": 1.1323192549009704e-06, "loss": 0.0003, "step": 196390 }, { "epoch": 3.21361367912951, "grad_norm": 0.04467693343758583, "learning_rate": 1.1318668251266972e-06, "loss": 0.0005, "step": 196400 }, { "epoch": 3.213777305080586, "grad_norm": 0.0003057683352380991, "learning_rate": 1.1314144742203682e-06, "loss": 0.0002, "step": 196410 }, { "epoch": 3.2139409310316616, "grad_norm": 0.011446675285696983, "learning_rate": 1.1309622021912081e-06, "loss": 0.0002, "step": 196420 }, { "epoch": 3.2141045569827376, "grad_norm": 0.00031610450241714716, "learning_rate": 1.1305100090484382e-06, "loss": 0.0003, "step": 196430 }, { "epoch": 3.214268182933813, "grad_norm": 0.027572348713874817, "learning_rate": 1.1300578948012798e-06, "loss": 0.0003, "step": 196440 }, { "epoch": 3.214431808884889, "grad_norm": 0.010861240327358246, "learning_rate": 1.1296058594589476e-06, "loss": 0.0005, "step": 196450 }, { "epoch": 3.214595434835965, "grad_norm": 0.0227730143815279, "learning_rate": 1.1291539030306608e-06, "loss": 0.0003, "step": 196460 }, { "epoch": 3.2147590607870407, "grad_norm": 0.04301472753286362, "learning_rate": 1.1287020255256343e-06, "loss": 0.0004, "step": 196470 }, { "epoch": 3.2149226867381167, "grad_norm": 0.02246171236038208, "learning_rate": 1.128250226953082e-06, "loss": 0.0004, "step": 196480 }, { "epoch": 3.2150863126891926, "grad_norm": 0.038231510668992996, "learning_rate": 1.127798507322213e-06, "loss": 0.0005, "step": 196490 }, { "epoch": 3.215249938640268, "grad_norm": 0.0008122918079607189, "learning_rate": 1.1273468666422398e-06, "loss": 0.0004, "step": 196500 }, { "epoch": 3.215413564591344, "grad_norm": 0.020554501563310623, "learning_rate": 1.1268953049223707e-06, "loss": 0.0002, "step": 196510 }, { "epoch": 3.21557719054242, "grad_norm": 0.008360868319869041, "learning_rate": 1.1264438221718133e-06, "loss": 0.0003, "step": 196520 }, { "epoch": 3.2157408164934957, "grad_norm": 0.03296222537755966, "learning_rate": 1.1259924183997707e-06, "loss": 0.0002, "step": 196530 }, { "epoch": 3.2159044424445717, "grad_norm": 0.005983112845569849, "learning_rate": 1.1255410936154492e-06, "loss": 0.0003, "step": 196540 }, { "epoch": 3.2160680683956477, "grad_norm": 0.005991276353597641, "learning_rate": 1.125089847828047e-06, "loss": 0.0004, "step": 196550 }, { "epoch": 3.2162316943467233, "grad_norm": 0.00033306083059869707, "learning_rate": 1.1246386810467702e-06, "loss": 0.0001, "step": 196560 }, { "epoch": 3.2163953202977993, "grad_norm": 0.04979437217116356, "learning_rate": 1.1241875932808132e-06, "loss": 0.0002, "step": 196570 }, { "epoch": 3.216558946248875, "grad_norm": 0.0004853673162870109, "learning_rate": 1.1237365845393767e-06, "loss": 0.0002, "step": 196580 }, { "epoch": 3.216722572199951, "grad_norm": 0.01417562272399664, "learning_rate": 1.1232856548316518e-06, "loss": 0.0001, "step": 196590 }, { "epoch": 3.216886198151027, "grad_norm": 0.0006094201817177236, "learning_rate": 1.122834804166838e-06, "loss": 0.0002, "step": 196600 }, { "epoch": 3.2170498241021024, "grad_norm": 0.000558196275960654, "learning_rate": 1.1223840325541236e-06, "loss": 0.0002, "step": 196610 }, { "epoch": 3.2172134500531784, "grad_norm": 0.009462954476475716, "learning_rate": 1.121933340002702e-06, "loss": 0.0002, "step": 196620 }, { "epoch": 3.2173770760042544, "grad_norm": 0.016528069972991943, "learning_rate": 1.1214827265217592e-06, "loss": 0.0003, "step": 196630 }, { "epoch": 3.21754070195533, "grad_norm": 0.0022287487518042326, "learning_rate": 1.1210321921204881e-06, "loss": 0.0003, "step": 196640 }, { "epoch": 3.217704327906406, "grad_norm": 0.004233099054545164, "learning_rate": 1.1205817368080696e-06, "loss": 0.0001, "step": 196650 }, { "epoch": 3.217867953857482, "grad_norm": 0.0001607486919965595, "learning_rate": 1.120131360593692e-06, "loss": 0.0002, "step": 196660 }, { "epoch": 3.2180315798085575, "grad_norm": 0.006024772766977549, "learning_rate": 1.1196810634865335e-06, "loss": 0.0001, "step": 196670 }, { "epoch": 3.2181952057596335, "grad_norm": 0.04204423725605011, "learning_rate": 1.11923084549578e-06, "loss": 0.0001, "step": 196680 }, { "epoch": 3.2183588317107095, "grad_norm": 0.0004289193020667881, "learning_rate": 1.118780706630609e-06, "loss": 0.0009, "step": 196690 }, { "epoch": 3.218522457661785, "grad_norm": 0.010153880342841148, "learning_rate": 1.1183306469001986e-06, "loss": 0.0003, "step": 196700 }, { "epoch": 3.218686083612861, "grad_norm": 0.019789157435297966, "learning_rate": 1.117880666313723e-06, "loss": 0.0003, "step": 196710 }, { "epoch": 3.218849709563937, "grad_norm": 0.021666666492819786, "learning_rate": 1.1174307648803617e-06, "loss": 0.0002, "step": 196720 }, { "epoch": 3.2190133355150126, "grad_norm": 0.003746897215023637, "learning_rate": 1.1169809426092837e-06, "loss": 0.0007, "step": 196730 }, { "epoch": 3.2191769614660886, "grad_norm": 0.001286681741476059, "learning_rate": 1.1165311995096633e-06, "loss": 0.0002, "step": 196740 }, { "epoch": 3.2193405874171646, "grad_norm": 0.09740448743104935, "learning_rate": 1.1160815355906679e-06, "loss": 0.0004, "step": 196750 }, { "epoch": 3.21950421336824, "grad_norm": 0.002595350844785571, "learning_rate": 1.1156319508614666e-06, "loss": 0.0001, "step": 196760 }, { "epoch": 3.219667839319316, "grad_norm": 0.030263982713222504, "learning_rate": 1.115182445331226e-06, "loss": 0.0004, "step": 196770 }, { "epoch": 3.219831465270392, "grad_norm": 0.022717520594596863, "learning_rate": 1.1147330190091127e-06, "loss": 0.0002, "step": 196780 }, { "epoch": 3.2199950912214677, "grad_norm": 0.0024777392391115427, "learning_rate": 1.1142836719042877e-06, "loss": 0.0002, "step": 196790 }, { "epoch": 3.2201587171725436, "grad_norm": 0.04884007200598717, "learning_rate": 1.1138344040259142e-06, "loss": 0.0004, "step": 196800 }, { "epoch": 3.220322343123619, "grad_norm": 0.006201480980962515, "learning_rate": 1.113385215383152e-06, "loss": 0.0004, "step": 196810 }, { "epoch": 3.220485969074695, "grad_norm": 0.08169138431549072, "learning_rate": 1.1129361059851611e-06, "loss": 0.0008, "step": 196820 }, { "epoch": 3.220649595025771, "grad_norm": 0.02997354418039322, "learning_rate": 1.1124870758410954e-06, "loss": 0.0004, "step": 196830 }, { "epoch": 3.2208132209768467, "grad_norm": 0.05029186233878136, "learning_rate": 1.1120381249601126e-06, "loss": 0.0003, "step": 196840 }, { "epoch": 3.2209768469279227, "grad_norm": 0.046254709362983704, "learning_rate": 1.1115892533513655e-06, "loss": 0.0005, "step": 196850 }, { "epoch": 3.2211404728789987, "grad_norm": 0.01600034534931183, "learning_rate": 1.1111404610240085e-06, "loss": 0.0007, "step": 196860 }, { "epoch": 3.2213040988300743, "grad_norm": 0.006635275669395924, "learning_rate": 1.1106917479871881e-06, "loss": 0.0002, "step": 196870 }, { "epoch": 3.2214677247811503, "grad_norm": 0.05125761777162552, "learning_rate": 1.1102431142500559e-06, "loss": 0.0002, "step": 196880 }, { "epoch": 3.2216313507322263, "grad_norm": 0.01377555076032877, "learning_rate": 1.109794559821758e-06, "loss": 0.0004, "step": 196890 }, { "epoch": 3.221794976683302, "grad_norm": 0.041464123874902725, "learning_rate": 1.1093460847114418e-06, "loss": 0.0002, "step": 196900 }, { "epoch": 3.221958602634378, "grad_norm": 0.0013490903656929731, "learning_rate": 1.1088976889282488e-06, "loss": 0.0003, "step": 196910 }, { "epoch": 3.222122228585454, "grad_norm": 0.0062011172994971275, "learning_rate": 1.1084493724813229e-06, "loss": 0.0005, "step": 196920 }, { "epoch": 3.2222858545365294, "grad_norm": 0.05168970674276352, "learning_rate": 1.1080011353798043e-06, "loss": 0.0004, "step": 196930 }, { "epoch": 3.2224494804876054, "grad_norm": 0.005304677411913872, "learning_rate": 1.1075529776328335e-06, "loss": 0.0038, "step": 196940 }, { "epoch": 3.2226131064386814, "grad_norm": 0.0030567690264433622, "learning_rate": 1.1071048992495465e-06, "loss": 0.0002, "step": 196950 }, { "epoch": 3.222776732389757, "grad_norm": 0.0006158221513032913, "learning_rate": 1.1066569002390798e-06, "loss": 0.0002, "step": 196960 }, { "epoch": 3.222940358340833, "grad_norm": 0.10756181180477142, "learning_rate": 1.1062089806105675e-06, "loss": 0.0008, "step": 196970 }, { "epoch": 3.2231039842919085, "grad_norm": 0.012846231460571289, "learning_rate": 1.1057611403731438e-06, "loss": 0.0005, "step": 196980 }, { "epoch": 3.2232676102429845, "grad_norm": 0.001235002651810646, "learning_rate": 1.1053133795359372e-06, "loss": 0.0007, "step": 196990 }, { "epoch": 3.2234312361940605, "grad_norm": 0.016470227390527725, "learning_rate": 1.1048656981080785e-06, "loss": 0.0001, "step": 197000 }, { "epoch": 3.223594862145136, "grad_norm": 0.04470830038189888, "learning_rate": 1.1044180960986967e-06, "loss": 0.0002, "step": 197010 }, { "epoch": 3.223758488096212, "grad_norm": 0.0022329571656882763, "learning_rate": 1.1039705735169159e-06, "loss": 0.0003, "step": 197020 }, { "epoch": 3.223922114047288, "grad_norm": 0.17143385112285614, "learning_rate": 1.103523130371863e-06, "loss": 0.0003, "step": 197030 }, { "epoch": 3.2240857399983636, "grad_norm": 0.05411915108561516, "learning_rate": 1.1030757666726567e-06, "loss": 0.0002, "step": 197040 }, { "epoch": 3.2242493659494396, "grad_norm": 0.0053024692460894585, "learning_rate": 1.102628482428424e-06, "loss": 0.0006, "step": 197050 }, { "epoch": 3.2244129919005156, "grad_norm": 0.017816517502069473, "learning_rate": 1.1021812776482804e-06, "loss": 0.0003, "step": 197060 }, { "epoch": 3.224576617851591, "grad_norm": 0.027220919728279114, "learning_rate": 1.1017341523413473e-06, "loss": 0.0003, "step": 197070 }, { "epoch": 3.224740243802667, "grad_norm": 0.026892488822340965, "learning_rate": 1.1012871065167362e-06, "loss": 0.0002, "step": 197080 }, { "epoch": 3.224903869753743, "grad_norm": 0.08262401074171066, "learning_rate": 1.1008401401835688e-06, "loss": 0.0004, "step": 197090 }, { "epoch": 3.2250674957048187, "grad_norm": 0.01479349471628666, "learning_rate": 1.1003932533509531e-06, "loss": 0.0001, "step": 197100 }, { "epoch": 3.2252311216558947, "grad_norm": 0.0010117971105501056, "learning_rate": 1.0999464460280035e-06, "loss": 0.0003, "step": 197110 }, { "epoch": 3.2253947476069706, "grad_norm": 0.0484158918261528, "learning_rate": 1.0994997182238266e-06, "loss": 0.0003, "step": 197120 }, { "epoch": 3.225558373558046, "grad_norm": 0.0032766370568424463, "learning_rate": 1.099053069947536e-06, "loss": 0.0002, "step": 197130 }, { "epoch": 3.225721999509122, "grad_norm": 0.036250900477170944, "learning_rate": 1.0986065012082343e-06, "loss": 0.0002, "step": 197140 }, { "epoch": 3.225885625460198, "grad_norm": 0.02191382646560669, "learning_rate": 1.0981600120150292e-06, "loss": 0.0004, "step": 197150 }, { "epoch": 3.2260492514112737, "grad_norm": 0.0006253679166547954, "learning_rate": 1.0977136023770202e-06, "loss": 0.0002, "step": 197160 }, { "epoch": 3.2262128773623497, "grad_norm": 0.07692009955644608, "learning_rate": 1.097267272303315e-06, "loss": 0.0003, "step": 197170 }, { "epoch": 3.2263765033134257, "grad_norm": 0.0014337999746203423, "learning_rate": 1.0968210218030096e-06, "loss": 0.0002, "step": 197180 }, { "epoch": 3.2265401292645013, "grad_norm": 0.029646284878253937, "learning_rate": 1.0963748508852056e-06, "loss": 0.0002, "step": 197190 }, { "epoch": 3.2267037552155773, "grad_norm": 0.0007524396060034633, "learning_rate": 1.0959287595589968e-06, "loss": 0.0003, "step": 197200 }, { "epoch": 3.226867381166653, "grad_norm": 0.03375280275940895, "learning_rate": 1.0954827478334806e-06, "loss": 0.0001, "step": 197210 }, { "epoch": 3.227031007117729, "grad_norm": 0.013867117464542389, "learning_rate": 1.0950368157177498e-06, "loss": 0.0003, "step": 197220 }, { "epoch": 3.227194633068805, "grad_norm": 0.011994581669569016, "learning_rate": 1.0945909632208994e-06, "loss": 0.0004, "step": 197230 }, { "epoch": 3.2273582590198804, "grad_norm": 0.011411908082664013, "learning_rate": 1.0941451903520162e-06, "loss": 0.0005, "step": 197240 }, { "epoch": 3.2275218849709564, "grad_norm": 0.008207104168832302, "learning_rate": 1.0936994971201908e-06, "loss": 0.0002, "step": 197250 }, { "epoch": 3.2276855109220324, "grad_norm": 0.02395281381905079, "learning_rate": 1.0932538835345103e-06, "loss": 0.0003, "step": 197260 }, { "epoch": 3.227849136873108, "grad_norm": 0.0017529281321913004, "learning_rate": 1.092808349604062e-06, "loss": 0.0002, "step": 197270 }, { "epoch": 3.228012762824184, "grad_norm": 0.05345335602760315, "learning_rate": 1.0923628953379273e-06, "loss": 0.0004, "step": 197280 }, { "epoch": 3.22817638877526, "grad_norm": 0.0189724862575531, "learning_rate": 1.0919175207451898e-06, "loss": 0.0003, "step": 197290 }, { "epoch": 3.2283400147263355, "grad_norm": 0.03172054886817932, "learning_rate": 1.0914722258349303e-06, "loss": 0.0002, "step": 197300 }, { "epoch": 3.2285036406774115, "grad_norm": 0.027423322200775146, "learning_rate": 1.0910270106162291e-06, "loss": 0.0005, "step": 197310 }, { "epoch": 3.2286672666284875, "grad_norm": 0.01351303979754448, "learning_rate": 1.0905818750981618e-06, "loss": 0.0002, "step": 197320 }, { "epoch": 3.228830892579563, "grad_norm": 0.004833804443478584, "learning_rate": 1.090136819289805e-06, "loss": 0.0002, "step": 197330 }, { "epoch": 3.228994518530639, "grad_norm": 0.01306384801864624, "learning_rate": 1.0896918432002334e-06, "loss": 0.0003, "step": 197340 }, { "epoch": 3.2291581444817146, "grad_norm": 0.06999698281288147, "learning_rate": 1.0892469468385202e-06, "loss": 0.0003, "step": 197350 }, { "epoch": 3.2293217704327906, "grad_norm": 0.04822006821632385, "learning_rate": 1.0888021302137352e-06, "loss": 0.0003, "step": 197360 }, { "epoch": 3.2294853963838666, "grad_norm": 0.016204824671149254, "learning_rate": 1.0883573933349483e-06, "loss": 0.0004, "step": 197370 }, { "epoch": 3.229649022334942, "grad_norm": 0.016826212406158447, "learning_rate": 1.0879127362112274e-06, "loss": 0.0002, "step": 197380 }, { "epoch": 3.229812648286018, "grad_norm": 0.02210424467921257, "learning_rate": 1.0874681588516394e-06, "loss": 0.0005, "step": 197390 }, { "epoch": 3.229976274237094, "grad_norm": 0.025486692786216736, "learning_rate": 1.0870236612652474e-06, "loss": 0.0003, "step": 197400 }, { "epoch": 3.2301399001881697, "grad_norm": 0.000850641808938235, "learning_rate": 1.086579243461115e-06, "loss": 0.0003, "step": 197410 }, { "epoch": 3.2303035261392457, "grad_norm": 0.00304751587100327, "learning_rate": 1.086134905448304e-06, "loss": 0.0002, "step": 197420 }, { "epoch": 3.2304671520903216, "grad_norm": 0.024298258125782013, "learning_rate": 1.0856906472358742e-06, "loss": 0.0002, "step": 197430 }, { "epoch": 3.230630778041397, "grad_norm": 0.040268491953611374, "learning_rate": 1.0852464688328824e-06, "loss": 0.0003, "step": 197440 }, { "epoch": 3.230794403992473, "grad_norm": 0.027709897607564926, "learning_rate": 1.084802370248385e-06, "loss": 0.0004, "step": 197450 }, { "epoch": 3.230958029943549, "grad_norm": 0.001396377687342465, "learning_rate": 1.0843583514914385e-06, "loss": 0.0002, "step": 197460 }, { "epoch": 3.2311216558946247, "grad_norm": 0.00261281942948699, "learning_rate": 1.083914412571096e-06, "loss": 0.0003, "step": 197470 }, { "epoch": 3.2312852818457007, "grad_norm": 0.04198453202843666, "learning_rate": 1.0834705534964063e-06, "loss": 0.0006, "step": 197480 }, { "epoch": 3.2314489077967767, "grad_norm": 0.002361343940719962, "learning_rate": 1.0830267742764218e-06, "loss": 0.0003, "step": 197490 }, { "epoch": 3.2316125337478523, "grad_norm": 0.05382835865020752, "learning_rate": 1.0825830749201893e-06, "loss": 0.0004, "step": 197500 }, { "epoch": 3.2317761596989283, "grad_norm": 0.03459644317626953, "learning_rate": 1.0821394554367582e-06, "loss": 0.0002, "step": 197510 }, { "epoch": 3.2319397856500043, "grad_norm": 0.04683860391378403, "learning_rate": 1.0816959158351703e-06, "loss": 0.0003, "step": 197520 }, { "epoch": 3.23210341160108, "grad_norm": 0.007608460262417793, "learning_rate": 1.0812524561244703e-06, "loss": 0.0004, "step": 197530 }, { "epoch": 3.232267037552156, "grad_norm": 0.019077638164162636, "learning_rate": 1.0808090763136997e-06, "loss": 0.0006, "step": 197540 }, { "epoch": 3.232430663503232, "grad_norm": 0.009054862894117832, "learning_rate": 1.0803657764119003e-06, "loss": 0.0007, "step": 197550 }, { "epoch": 3.2325942894543074, "grad_norm": 0.033913593739271164, "learning_rate": 1.079922556428108e-06, "loss": 0.0001, "step": 197560 }, { "epoch": 3.2327579154053834, "grad_norm": 0.003026421181857586, "learning_rate": 1.0794794163713607e-06, "loss": 0.0005, "step": 197570 }, { "epoch": 3.232921541356459, "grad_norm": 0.0013883940409868956, "learning_rate": 1.0790363562506933e-06, "loss": 0.0002, "step": 197580 }, { "epoch": 3.233085167307535, "grad_norm": 0.03889048844575882, "learning_rate": 1.0785933760751422e-06, "loss": 0.0003, "step": 197590 }, { "epoch": 3.233248793258611, "grad_norm": 0.028316382318735123, "learning_rate": 1.0781504758537352e-06, "loss": 0.0002, "step": 197600 }, { "epoch": 3.2334124192096865, "grad_norm": 0.028362229466438293, "learning_rate": 1.0777076555955046e-06, "loss": 0.0003, "step": 197610 }, { "epoch": 3.2335760451607625, "grad_norm": 0.0020395475439727306, "learning_rate": 1.0772649153094793e-06, "loss": 0.0007, "step": 197620 }, { "epoch": 3.2337396711118385, "grad_norm": 0.005373879801481962, "learning_rate": 1.0768222550046874e-06, "loss": 0.0002, "step": 197630 }, { "epoch": 3.233903297062914, "grad_norm": 0.01853400655090809, "learning_rate": 1.0763796746901522e-06, "loss": 0.0004, "step": 197640 }, { "epoch": 3.23406692301399, "grad_norm": 0.050118789076805115, "learning_rate": 1.0759371743748999e-06, "loss": 0.0003, "step": 197650 }, { "epoch": 3.234230548965066, "grad_norm": 0.04621562361717224, "learning_rate": 1.0754947540679483e-06, "loss": 0.0008, "step": 197660 }, { "epoch": 3.2343941749161416, "grad_norm": 0.032283615320920944, "learning_rate": 1.0750524137783235e-06, "loss": 0.0001, "step": 197670 }, { "epoch": 3.2345578008672176, "grad_norm": 0.023478223010897636, "learning_rate": 1.074610153515041e-06, "loss": 0.0001, "step": 197680 }, { "epoch": 3.2347214268182936, "grad_norm": 0.01642177253961563, "learning_rate": 1.0741679732871202e-06, "loss": 0.0002, "step": 197690 }, { "epoch": 3.234885052769369, "grad_norm": 0.060931988060474396, "learning_rate": 1.0737258731035728e-06, "loss": 0.0004, "step": 197700 }, { "epoch": 3.235048678720445, "grad_norm": 0.01602240651845932, "learning_rate": 1.0732838529734186e-06, "loss": 0.0003, "step": 197710 }, { "epoch": 3.235212304671521, "grad_norm": 0.007900819182395935, "learning_rate": 1.0728419129056655e-06, "loss": 0.0002, "step": 197720 }, { "epoch": 3.2353759306225967, "grad_norm": 0.03619620203971863, "learning_rate": 1.0724000529093276e-06, "loss": 0.0003, "step": 197730 }, { "epoch": 3.2355395565736726, "grad_norm": 0.016095463186502457, "learning_rate": 1.0719582729934092e-06, "loss": 0.0003, "step": 197740 }, { "epoch": 3.235703182524748, "grad_norm": 0.001540232333354652, "learning_rate": 1.071516573166924e-06, "loss": 0.0008, "step": 197750 }, { "epoch": 3.235866808475824, "grad_norm": 0.07033023983240128, "learning_rate": 1.071074953438873e-06, "loss": 0.0004, "step": 197760 }, { "epoch": 3.2360304344269, "grad_norm": 0.014878200367093086, "learning_rate": 1.0706334138182638e-06, "loss": 0.0003, "step": 197770 }, { "epoch": 3.2361940603779757, "grad_norm": 0.003633445128798485, "learning_rate": 1.0701919543140955e-06, "loss": 0.0002, "step": 197780 }, { "epoch": 3.2363576863290517, "grad_norm": 0.08981707692146301, "learning_rate": 1.0697505749353725e-06, "loss": 0.0007, "step": 197790 }, { "epoch": 3.2365213122801277, "grad_norm": 0.08488619327545166, "learning_rate": 1.0693092756910922e-06, "loss": 0.0003, "step": 197800 }, { "epoch": 3.2366849382312033, "grad_norm": 0.012774087488651276, "learning_rate": 1.0688680565902543e-06, "loss": 0.0004, "step": 197810 }, { "epoch": 3.2368485641822793, "grad_norm": 0.009980987757444382, "learning_rate": 1.0684269176418504e-06, "loss": 0.0003, "step": 197820 }, { "epoch": 3.2370121901333553, "grad_norm": 0.006778918672353029, "learning_rate": 1.0679858588548808e-06, "loss": 0.0002, "step": 197830 }, { "epoch": 3.237175816084431, "grad_norm": 0.01869155652821064, "learning_rate": 1.0675448802383336e-06, "loss": 0.0003, "step": 197840 }, { "epoch": 3.237339442035507, "grad_norm": 0.015640171244740486, "learning_rate": 1.067103981801204e-06, "loss": 0.0001, "step": 197850 }, { "epoch": 3.237503067986583, "grad_norm": 0.000493639730848372, "learning_rate": 1.0666631635524776e-06, "loss": 0.0004, "step": 197860 }, { "epoch": 3.2376666939376584, "grad_norm": 0.005684906616806984, "learning_rate": 1.0662224255011439e-06, "loss": 0.001, "step": 197870 }, { "epoch": 3.2378303198887344, "grad_norm": 0.026116805151104927, "learning_rate": 1.06578176765619e-06, "loss": 0.0003, "step": 197880 }, { "epoch": 3.2379939458398104, "grad_norm": 0.019295938313007355, "learning_rate": 1.0653411900266008e-06, "loss": 0.0002, "step": 197890 }, { "epoch": 3.238157571790886, "grad_norm": 0.01883538067340851, "learning_rate": 1.0649006926213573e-06, "loss": 0.0001, "step": 197900 }, { "epoch": 3.238321197741962, "grad_norm": 0.23055976629257202, "learning_rate": 1.0644602754494422e-06, "loss": 0.0004, "step": 197910 }, { "epoch": 3.238484823693038, "grad_norm": 0.0003276877978350967, "learning_rate": 1.0640199385198347e-06, "loss": 0.0002, "step": 197920 }, { "epoch": 3.2386484496441135, "grad_norm": 0.00048605294432491064, "learning_rate": 1.0635796818415145e-06, "loss": 0.0006, "step": 197930 }, { "epoch": 3.2388120755951895, "grad_norm": 0.006382189691066742, "learning_rate": 1.0631395054234556e-06, "loss": 0.0001, "step": 197940 }, { "epoch": 3.2389757015462655, "grad_norm": 0.0008085363078862429, "learning_rate": 1.062699409274634e-06, "loss": 0.0002, "step": 197950 }, { "epoch": 3.239139327497341, "grad_norm": 0.040657058358192444, "learning_rate": 1.0622593934040226e-06, "loss": 0.0005, "step": 197960 }, { "epoch": 3.239302953448417, "grad_norm": 0.07170287519693375, "learning_rate": 1.0618194578205948e-06, "loss": 0.0004, "step": 197970 }, { "epoch": 3.2394665793994926, "grad_norm": 0.06536975502967834, "learning_rate": 1.0613796025333178e-06, "loss": 0.0002, "step": 197980 }, { "epoch": 3.2396302053505686, "grad_norm": 0.04975023493170738, "learning_rate": 1.0609398275511606e-06, "loss": 0.0002, "step": 197990 }, { "epoch": 3.2397938313016446, "grad_norm": 0.036243900656700134, "learning_rate": 1.060500132883091e-06, "loss": 0.0004, "step": 198000 }, { "epoch": 3.23995745725272, "grad_norm": 0.0399915836751461, "learning_rate": 1.060060518538074e-06, "loss": 0.0004, "step": 198010 }, { "epoch": 3.240121083203796, "grad_norm": 0.10251663625240326, "learning_rate": 1.059620984525071e-06, "loss": 0.0004, "step": 198020 }, { "epoch": 3.240284709154872, "grad_norm": 0.020900024101138115, "learning_rate": 1.059181530853045e-06, "loss": 0.0003, "step": 198030 }, { "epoch": 3.2404483351059477, "grad_norm": 0.03708701208233833, "learning_rate": 1.0587421575309559e-06, "loss": 0.0002, "step": 198040 }, { "epoch": 3.2406119610570236, "grad_norm": 0.0458720400929451, "learning_rate": 1.058302864567764e-06, "loss": 0.0002, "step": 198050 }, { "epoch": 3.2407755870080996, "grad_norm": 0.004263092298060656, "learning_rate": 1.0578636519724229e-06, "loss": 0.0002, "step": 198060 }, { "epoch": 3.240939212959175, "grad_norm": 0.0693117156624794, "learning_rate": 1.0574245197538891e-06, "loss": 0.0003, "step": 198070 }, { "epoch": 3.241102838910251, "grad_norm": 0.006373284850269556, "learning_rate": 1.056985467921116e-06, "loss": 0.0005, "step": 198080 }, { "epoch": 3.241266464861327, "grad_norm": 0.0072723072953522205, "learning_rate": 1.0565464964830579e-06, "loss": 0.0002, "step": 198090 }, { "epoch": 3.2414300908124027, "grad_norm": 0.009705345146358013, "learning_rate": 1.0561076054486614e-06, "loss": 0.0004, "step": 198100 }, { "epoch": 3.2415937167634787, "grad_norm": 0.03956371173262596, "learning_rate": 1.0556687948268763e-06, "loss": 0.0001, "step": 198110 }, { "epoch": 3.2417573427145547, "grad_norm": 0.005005877465009689, "learning_rate": 1.05523006462665e-06, "loss": 0.0003, "step": 198120 }, { "epoch": 3.2419209686656303, "grad_norm": 0.07112842798233032, "learning_rate": 1.0547914148569293e-06, "loss": 0.0003, "step": 198130 }, { "epoch": 3.2420845946167063, "grad_norm": 0.023456234484910965, "learning_rate": 1.054352845526655e-06, "loss": 0.0004, "step": 198140 }, { "epoch": 3.242248220567782, "grad_norm": 0.05180398374795914, "learning_rate": 1.0539143566447708e-06, "loss": 0.0003, "step": 198150 }, { "epoch": 3.242411846518858, "grad_norm": 0.010699340142309666, "learning_rate": 1.0534759482202162e-06, "loss": 0.0004, "step": 198160 }, { "epoch": 3.242575472469934, "grad_norm": 0.033975426107645035, "learning_rate": 1.0530376202619325e-06, "loss": 0.0006, "step": 198170 }, { "epoch": 3.2427390984210094, "grad_norm": 0.0011867316206917167, "learning_rate": 1.0525993727788535e-06, "loss": 0.0002, "step": 198180 }, { "epoch": 3.2429027243720854, "grad_norm": 7.725873001618311e-05, "learning_rate": 1.0521612057799158e-06, "loss": 0.0003, "step": 198190 }, { "epoch": 3.2430663503231614, "grad_norm": 0.012853848747909069, "learning_rate": 1.051723119274054e-06, "loss": 0.0005, "step": 198200 }, { "epoch": 3.243229976274237, "grad_norm": 0.01906820386648178, "learning_rate": 1.0512851132702013e-06, "loss": 0.0002, "step": 198210 }, { "epoch": 3.243393602225313, "grad_norm": 0.1350700855255127, "learning_rate": 1.0508471877772852e-06, "loss": 0.0008, "step": 198220 }, { "epoch": 3.243557228176389, "grad_norm": 0.03384329378604889, "learning_rate": 1.0504093428042368e-06, "loss": 0.0007, "step": 198230 }, { "epoch": 3.2437208541274645, "grad_norm": 0.0845336765050888, "learning_rate": 1.0499715783599824e-06, "loss": 0.0002, "step": 198240 }, { "epoch": 3.2438844800785405, "grad_norm": 0.006285959854722023, "learning_rate": 1.0495338944534493e-06, "loss": 0.0002, "step": 198250 }, { "epoch": 3.2440481060296165, "grad_norm": 0.01329015102237463, "learning_rate": 1.04909629109356e-06, "loss": 0.0002, "step": 198260 }, { "epoch": 3.244211731980692, "grad_norm": 0.0014580675633624196, "learning_rate": 1.0486587682892363e-06, "loss": 0.0004, "step": 198270 }, { "epoch": 3.244375357931768, "grad_norm": 0.07843455672264099, "learning_rate": 1.0482213260494e-06, "loss": 0.0004, "step": 198280 }, { "epoch": 3.244538983882844, "grad_norm": 0.008235645480453968, "learning_rate": 1.0477839643829713e-06, "loss": 0.0002, "step": 198290 }, { "epoch": 3.2447026098339196, "grad_norm": 0.0178545992821455, "learning_rate": 1.047346683298865e-06, "loss": 0.0001, "step": 198300 }, { "epoch": 3.2448662357849956, "grad_norm": 0.0014162688748911023, "learning_rate": 1.0469094828059995e-06, "loss": 0.0002, "step": 198310 }, { "epoch": 3.2450298617360716, "grad_norm": 0.05714212730526924, "learning_rate": 1.046472362913285e-06, "loss": 0.0003, "step": 198320 }, { "epoch": 3.245193487687147, "grad_norm": 0.011147513054311275, "learning_rate": 1.04603532362964e-06, "loss": 0.0002, "step": 198330 }, { "epoch": 3.245357113638223, "grad_norm": 0.06940095126628876, "learning_rate": 1.04559836496397e-06, "loss": 0.0006, "step": 198340 }, { "epoch": 3.245520739589299, "grad_norm": 0.008754687383770943, "learning_rate": 1.0451614869251874e-06, "loss": 0.0002, "step": 198350 }, { "epoch": 3.2456843655403746, "grad_norm": 0.07545728981494904, "learning_rate": 1.0447246895221963e-06, "loss": 0.0003, "step": 198360 }, { "epoch": 3.2458479914914506, "grad_norm": 0.07757912576198578, "learning_rate": 1.0442879727639078e-06, "loss": 0.0004, "step": 198370 }, { "epoch": 3.246011617442526, "grad_norm": 0.03563973680138588, "learning_rate": 1.0438513366592218e-06, "loss": 0.0002, "step": 198380 }, { "epoch": 3.246175243393602, "grad_norm": 0.008706695400178432, "learning_rate": 1.0434147812170437e-06, "loss": 0.0003, "step": 198390 }, { "epoch": 3.246338869344678, "grad_norm": 0.05015262961387634, "learning_rate": 1.0429783064462706e-06, "loss": 0.0002, "step": 198400 }, { "epoch": 3.2465024952957537, "grad_norm": 0.01494014635682106, "learning_rate": 1.0425419123558072e-06, "loss": 0.0005, "step": 198410 }, { "epoch": 3.2466661212468297, "grad_norm": 0.0020597411785274744, "learning_rate": 1.0421055989545476e-06, "loss": 0.0001, "step": 198420 }, { "epoch": 3.2468297471979057, "grad_norm": 0.021776199340820312, "learning_rate": 1.0416693662513894e-06, "loss": 0.0004, "step": 198430 }, { "epoch": 3.2469933731489813, "grad_norm": 0.0033375730272382498, "learning_rate": 1.041233214255224e-06, "loss": 0.0003, "step": 198440 }, { "epoch": 3.2471569991000573, "grad_norm": 0.00027678324840962887, "learning_rate": 1.0407971429749492e-06, "loss": 0.0002, "step": 198450 }, { "epoch": 3.2473206250511333, "grad_norm": 0.09121376276016235, "learning_rate": 1.0403611524194523e-06, "loss": 0.0008, "step": 198460 }, { "epoch": 3.247484251002209, "grad_norm": 0.017694327980279922, "learning_rate": 1.0399252425976252e-06, "loss": 0.0003, "step": 198470 }, { "epoch": 3.247647876953285, "grad_norm": 0.009904254227876663, "learning_rate": 1.039489413518353e-06, "loss": 0.0003, "step": 198480 }, { "epoch": 3.247811502904361, "grad_norm": 0.01028918381780386, "learning_rate": 1.0390536651905236e-06, "loss": 0.0003, "step": 198490 }, { "epoch": 3.2479751288554364, "grad_norm": 0.021616125479340553, "learning_rate": 1.038617997623021e-06, "loss": 0.0008, "step": 198500 }, { "epoch": 3.2481387548065124, "grad_norm": 0.0013763048918917775, "learning_rate": 1.0381824108247301e-06, "loss": 0.0003, "step": 198510 }, { "epoch": 3.248302380757588, "grad_norm": 0.17592915892601013, "learning_rate": 1.0377469048045285e-06, "loss": 0.0005, "step": 198520 }, { "epoch": 3.248466006708664, "grad_norm": 0.0017597401747480035, "learning_rate": 1.0373114795712986e-06, "loss": 0.0004, "step": 198530 }, { "epoch": 3.24862963265974, "grad_norm": 0.011367768049240112, "learning_rate": 1.0368761351339169e-06, "loss": 0.0002, "step": 198540 }, { "epoch": 3.2487932586108155, "grad_norm": 0.008275204338133335, "learning_rate": 1.0364408715012615e-06, "loss": 0.0002, "step": 198550 }, { "epoch": 3.2489568845618915, "grad_norm": 0.022981947287917137, "learning_rate": 1.036005688682205e-06, "loss": 0.0002, "step": 198560 }, { "epoch": 3.2491205105129675, "grad_norm": 0.013232741504907608, "learning_rate": 1.0355705866856208e-06, "loss": 0.0003, "step": 198570 }, { "epoch": 3.249284136464043, "grad_norm": 0.03645257651805878, "learning_rate": 1.035135565520381e-06, "loss": 0.0002, "step": 198580 }, { "epoch": 3.249447762415119, "grad_norm": 0.0006363103166222572, "learning_rate": 1.0347006251953561e-06, "loss": 0.0005, "step": 198590 }, { "epoch": 3.249611388366195, "grad_norm": 0.03562892600893974, "learning_rate": 1.0342657657194122e-06, "loss": 0.0002, "step": 198600 }, { "epoch": 3.2497750143172706, "grad_norm": 0.012261960655450821, "learning_rate": 1.0338309871014163e-06, "loss": 0.0002, "step": 198610 }, { "epoch": 3.2499386402683466, "grad_norm": 0.04375096410512924, "learning_rate": 1.0333962893502336e-06, "loss": 0.0004, "step": 198620 }, { "epoch": 3.2501022662194226, "grad_norm": 0.01136272307485342, "learning_rate": 1.0329616724747283e-06, "loss": 0.0002, "step": 198630 }, { "epoch": 3.250265892170498, "grad_norm": 0.04637575522065163, "learning_rate": 1.0325271364837592e-06, "loss": 0.0002, "step": 198640 }, { "epoch": 3.250429518121574, "grad_norm": 0.0073854015208780766, "learning_rate": 1.0320926813861876e-06, "loss": 0.0002, "step": 198650 }, { "epoch": 3.25059314407265, "grad_norm": 0.033848851919174194, "learning_rate": 1.031658307190872e-06, "loss": 0.0002, "step": 198660 }, { "epoch": 3.2507567700237257, "grad_norm": 0.005249810870736837, "learning_rate": 1.0312240139066693e-06, "loss": 0.0006, "step": 198670 }, { "epoch": 3.2509203959748016, "grad_norm": 0.004207277670502663, "learning_rate": 1.0307898015424328e-06, "loss": 0.0002, "step": 198680 }, { "epoch": 3.2510840219258776, "grad_norm": 0.03388801962137222, "learning_rate": 1.0303556701070167e-06, "loss": 0.0003, "step": 198690 }, { "epoch": 3.251247647876953, "grad_norm": 0.009010663256049156, "learning_rate": 1.0299216196092721e-06, "loss": 0.0001, "step": 198700 }, { "epoch": 3.251411273828029, "grad_norm": 0.04138711467385292, "learning_rate": 1.0294876500580503e-06, "loss": 0.0004, "step": 198710 }, { "epoch": 3.251574899779105, "grad_norm": 0.008930970914661884, "learning_rate": 1.0290537614621976e-06, "loss": 0.001, "step": 198720 }, { "epoch": 3.2517385257301807, "grad_norm": 0.0014884925913065672, "learning_rate": 1.0286199538305614e-06, "loss": 0.0003, "step": 198730 }, { "epoch": 3.2519021516812567, "grad_norm": 0.031796179711818695, "learning_rate": 1.0281862271719867e-06, "loss": 0.0005, "step": 198740 }, { "epoch": 3.2520657776323327, "grad_norm": 0.03007081151008606, "learning_rate": 1.0277525814953182e-06, "loss": 0.0003, "step": 198750 }, { "epoch": 3.2522294035834083, "grad_norm": 0.01129111833870411, "learning_rate": 1.0273190168093955e-06, "loss": 0.0003, "step": 198760 }, { "epoch": 3.2523930295344843, "grad_norm": 0.01070715393871069, "learning_rate": 1.0268855331230586e-06, "loss": 0.0002, "step": 198770 }, { "epoch": 3.25255665548556, "grad_norm": 0.031625084578990936, "learning_rate": 1.0264521304451474e-06, "loss": 0.0001, "step": 198780 }, { "epoch": 3.252720281436636, "grad_norm": 0.03848472982645035, "learning_rate": 1.026018808784499e-06, "loss": 0.0003, "step": 198790 }, { "epoch": 3.252883907387712, "grad_norm": 0.0963163748383522, "learning_rate": 1.025585568149946e-06, "loss": 0.0002, "step": 198800 }, { "epoch": 3.2530475333387874, "grad_norm": 0.030872948467731476, "learning_rate": 1.025152408550323e-06, "loss": 0.0018, "step": 198810 }, { "epoch": 3.2532111592898634, "grad_norm": 0.0007825649809092283, "learning_rate": 1.0247193299944618e-06, "loss": 0.0001, "step": 198820 }, { "epoch": 3.2533747852409394, "grad_norm": 0.018060164526104927, "learning_rate": 1.0242863324911945e-06, "loss": 0.0003, "step": 198830 }, { "epoch": 3.253538411192015, "grad_norm": 0.0003497111611068249, "learning_rate": 1.0238534160493457e-06, "loss": 0.0002, "step": 198840 }, { "epoch": 3.253702037143091, "grad_norm": 0.004724165890365839, "learning_rate": 1.0234205806777442e-06, "loss": 0.0004, "step": 198850 }, { "epoch": 3.253865663094167, "grad_norm": 0.037042517215013504, "learning_rate": 1.0229878263852156e-06, "loss": 0.0002, "step": 198860 }, { "epoch": 3.2540292890452425, "grad_norm": 0.018005264922976494, "learning_rate": 1.0225551531805838e-06, "loss": 0.0001, "step": 198870 }, { "epoch": 3.2541929149963185, "grad_norm": 0.024133535102009773, "learning_rate": 1.0221225610726688e-06, "loss": 0.0002, "step": 198880 }, { "epoch": 3.254356540947394, "grad_norm": 0.049558375030756, "learning_rate": 1.0216900500702908e-06, "loss": 0.0003, "step": 198890 }, { "epoch": 3.25452016689847, "grad_norm": 0.01986829936504364, "learning_rate": 1.02125762018227e-06, "loss": 0.0004, "step": 198900 }, { "epoch": 3.254683792849546, "grad_norm": 0.01781156286597252, "learning_rate": 1.0208252714174239e-06, "loss": 0.0004, "step": 198910 }, { "epoch": 3.2548474188006216, "grad_norm": 0.03118024207651615, "learning_rate": 1.020393003784565e-06, "loss": 0.0001, "step": 198920 }, { "epoch": 3.2550110447516976, "grad_norm": 0.01637953333556652, "learning_rate": 1.019960817292508e-06, "loss": 0.0002, "step": 198930 }, { "epoch": 3.2551746707027736, "grad_norm": 0.02061248943209648, "learning_rate": 1.0195287119500652e-06, "loss": 0.0002, "step": 198940 }, { "epoch": 3.255338296653849, "grad_norm": 0.012109951116144657, "learning_rate": 1.019096687766048e-06, "loss": 0.0003, "step": 198950 }, { "epoch": 3.255501922604925, "grad_norm": 0.039192646741867065, "learning_rate": 1.018664744749262e-06, "loss": 0.0005, "step": 198960 }, { "epoch": 3.255665548556001, "grad_norm": 0.0241034384816885, "learning_rate": 1.0182328829085175e-06, "loss": 0.0002, "step": 198970 }, { "epoch": 3.2558291745070767, "grad_norm": 0.0008696443983353674, "learning_rate": 1.0178011022526157e-06, "loss": 0.0002, "step": 198980 }, { "epoch": 3.2559928004581526, "grad_norm": 0.0852382555603981, "learning_rate": 1.0173694027903647e-06, "loss": 0.0003, "step": 198990 }, { "epoch": 3.2561564264092286, "grad_norm": 0.003490095492452383, "learning_rate": 1.0169377845305639e-06, "loss": 0.0002, "step": 199000 }, { "epoch": 3.256320052360304, "grad_norm": 0.000728648214135319, "learning_rate": 1.0165062474820147e-06, "loss": 0.0002, "step": 199010 }, { "epoch": 3.25648367831138, "grad_norm": 0.046436842530965805, "learning_rate": 1.0160747916535135e-06, "loss": 0.0005, "step": 199020 }, { "epoch": 3.256647304262456, "grad_norm": 0.008100063540041447, "learning_rate": 1.0156434170538615e-06, "loss": 0.0002, "step": 199030 }, { "epoch": 3.2568109302135317, "grad_norm": 0.03751488775014877, "learning_rate": 1.01521212369185e-06, "loss": 0.0002, "step": 199040 }, { "epoch": 3.2569745561646077, "grad_norm": 0.026265960186719894, "learning_rate": 1.0147809115762758e-06, "loss": 0.0004, "step": 199050 }, { "epoch": 3.2571381821156837, "grad_norm": 0.016688650473952293, "learning_rate": 1.014349780715927e-06, "loss": 0.0003, "step": 199060 }, { "epoch": 3.2573018080667593, "grad_norm": 0.019391747191548347, "learning_rate": 1.0139187311195992e-06, "loss": 0.0003, "step": 199070 }, { "epoch": 3.2574654340178353, "grad_norm": 0.000732638465706259, "learning_rate": 1.0134877627960765e-06, "loss": 0.0003, "step": 199080 }, { "epoch": 3.2576290599689113, "grad_norm": 0.06218458339571953, "learning_rate": 1.01305687575415e-06, "loss": 0.0003, "step": 199090 }, { "epoch": 3.257792685919987, "grad_norm": 0.009681308642029762, "learning_rate": 1.0126260700025997e-06, "loss": 0.0003, "step": 199100 }, { "epoch": 3.257956311871063, "grad_norm": 0.004594170954078436, "learning_rate": 1.0121953455502158e-06, "loss": 0.0003, "step": 199110 }, { "epoch": 3.258119937822139, "grad_norm": 0.03785251826047897, "learning_rate": 1.0117647024057759e-06, "loss": 0.0003, "step": 199120 }, { "epoch": 3.2582835637732144, "grad_norm": 0.036020468920469284, "learning_rate": 1.0113341405780635e-06, "loss": 0.0001, "step": 199130 }, { "epoch": 3.2584471897242904, "grad_norm": 0.0235128216445446, "learning_rate": 1.0109036600758538e-06, "loss": 0.0001, "step": 199140 }, { "epoch": 3.258610815675366, "grad_norm": 0.0002664361381903291, "learning_rate": 1.010473260907926e-06, "loss": 0.0001, "step": 199150 }, { "epoch": 3.258774441626442, "grad_norm": 0.0030341073870658875, "learning_rate": 1.0100429430830561e-06, "loss": 0.0004, "step": 199160 }, { "epoch": 3.258938067577518, "grad_norm": 0.01929277367889881, "learning_rate": 1.0096127066100176e-06, "loss": 0.0004, "step": 199170 }, { "epoch": 3.2591016935285935, "grad_norm": 0.01137225329875946, "learning_rate": 1.0091825514975818e-06, "loss": 0.0004, "step": 199180 }, { "epoch": 3.2592653194796695, "grad_norm": 0.02099543809890747, "learning_rate": 1.0087524777545194e-06, "loss": 0.0001, "step": 199190 }, { "epoch": 3.2594289454307455, "grad_norm": 0.38396382331848145, "learning_rate": 1.008322485389599e-06, "loss": 0.0003, "step": 199200 }, { "epoch": 3.259592571381821, "grad_norm": 0.05850442871451378, "learning_rate": 1.0078925744115903e-06, "loss": 0.0001, "step": 199210 }, { "epoch": 3.259756197332897, "grad_norm": 0.114360012114048, "learning_rate": 1.0074627448292557e-06, "loss": 0.0003, "step": 199220 }, { "epoch": 3.259919823283973, "grad_norm": 0.003600507276132703, "learning_rate": 1.0070329966513598e-06, "loss": 0.0002, "step": 199230 }, { "epoch": 3.2600834492350486, "grad_norm": 0.08813467621803284, "learning_rate": 1.0066033298866651e-06, "loss": 0.0002, "step": 199240 }, { "epoch": 3.2602470751861246, "grad_norm": 0.0026843291707336903, "learning_rate": 1.0061737445439335e-06, "loss": 0.0003, "step": 199250 }, { "epoch": 3.2604107011372006, "grad_norm": 0.08552662283182144, "learning_rate": 1.0057442406319219e-06, "loss": 0.0003, "step": 199260 }, { "epoch": 3.260574327088276, "grad_norm": 0.006809710990637541, "learning_rate": 1.0053148181593876e-06, "loss": 0.0003, "step": 199270 }, { "epoch": 3.260737953039352, "grad_norm": 0.005213567987084389, "learning_rate": 1.004885477135087e-06, "loss": 0.0001, "step": 199280 }, { "epoch": 3.2609015789904277, "grad_norm": 0.10560201108455658, "learning_rate": 1.0044562175677753e-06, "loss": 0.0006, "step": 199290 }, { "epoch": 3.2610652049415036, "grad_norm": 0.06765715032815933, "learning_rate": 1.0040270394662016e-06, "loss": 0.0003, "step": 199300 }, { "epoch": 3.2612288308925796, "grad_norm": 0.022708209231495857, "learning_rate": 1.0035979428391184e-06, "loss": 0.0002, "step": 199310 }, { "epoch": 3.261392456843655, "grad_norm": 0.027283605188131332, "learning_rate": 1.0031689276952743e-06, "loss": 0.0001, "step": 199320 }, { "epoch": 3.261556082794731, "grad_norm": 0.09199240058660507, "learning_rate": 1.002739994043418e-06, "loss": 0.0009, "step": 199330 }, { "epoch": 3.261719708745807, "grad_norm": 0.050306808203458786, "learning_rate": 1.002311141892292e-06, "loss": 0.0002, "step": 199340 }, { "epoch": 3.2618833346968827, "grad_norm": 0.05700178071856499, "learning_rate": 1.0018823712506421e-06, "loss": 0.0006, "step": 199350 }, { "epoch": 3.2620469606479587, "grad_norm": 0.0028593086171895266, "learning_rate": 1.0014536821272103e-06, "loss": 0.0003, "step": 199360 }, { "epoch": 3.2622105865990347, "grad_norm": 0.07504639774560928, "learning_rate": 1.0010250745307387e-06, "loss": 0.0003, "step": 199370 }, { "epoch": 3.2623742125501103, "grad_norm": 0.023083437234163284, "learning_rate": 1.0005965484699637e-06, "loss": 0.0004, "step": 199380 }, { "epoch": 3.2625378385011863, "grad_norm": 0.004111272748559713, "learning_rate": 1.0001681039536231e-06, "loss": 0.0003, "step": 199390 }, { "epoch": 3.2627014644522623, "grad_norm": 0.02751588076353073, "learning_rate": 9.997397409904535e-07, "loss": 0.0002, "step": 199400 }, { "epoch": 3.262865090403338, "grad_norm": 0.0007924679084680974, "learning_rate": 9.9931145958919e-07, "loss": 0.0004, "step": 199410 }, { "epoch": 3.263028716354414, "grad_norm": 0.024667054414749146, "learning_rate": 9.988832597585619e-07, "loss": 0.0002, "step": 199420 }, { "epoch": 3.26319234230549, "grad_norm": 0.000973626971244812, "learning_rate": 9.984551415073012e-07, "loss": 0.0003, "step": 199430 }, { "epoch": 3.2633559682565654, "grad_norm": 0.054253142327070236, "learning_rate": 9.98027104844137e-07, "loss": 0.0002, "step": 199440 }, { "epoch": 3.2635195942076414, "grad_norm": 0.0007905387319624424, "learning_rate": 9.975991497777981e-07, "loss": 0.0002, "step": 199450 }, { "epoch": 3.2636832201587174, "grad_norm": 0.02561030350625515, "learning_rate": 9.971712763170071e-07, "loss": 0.0003, "step": 199460 }, { "epoch": 3.263846846109793, "grad_norm": 0.01450368668884039, "learning_rate": 9.967434844704892e-07, "loss": 0.0005, "step": 199470 }, { "epoch": 3.264010472060869, "grad_norm": 0.04469381272792816, "learning_rate": 9.963157742469676e-07, "loss": 0.0001, "step": 199480 }, { "epoch": 3.264174098011945, "grad_norm": 0.06007380038499832, "learning_rate": 9.958881456551634e-07, "loss": 0.0002, "step": 199490 }, { "epoch": 3.2643377239630205, "grad_norm": 0.0006894905236549675, "learning_rate": 9.95460598703793e-07, "loss": 0.0006, "step": 199500 }, { "epoch": 3.2645013499140965, "grad_norm": 0.00019741890719160438, "learning_rate": 9.950331334015756e-07, "loss": 0.0001, "step": 199510 }, { "epoch": 3.2646649758651725, "grad_norm": 0.01347492728382349, "learning_rate": 9.946057497572265e-07, "loss": 0.0002, "step": 199520 }, { "epoch": 3.264828601816248, "grad_norm": 0.012850916013121605, "learning_rate": 9.94178447779461e-07, "loss": 0.0004, "step": 199530 }, { "epoch": 3.264992227767324, "grad_norm": 0.01711721159517765, "learning_rate": 9.937512274769883e-07, "loss": 0.0003, "step": 199540 }, { "epoch": 3.2651558537183996, "grad_norm": 0.0004824037605430931, "learning_rate": 9.933240888585217e-07, "loss": 0.0002, "step": 199550 }, { "epoch": 3.2653194796694756, "grad_norm": 0.0033155966084450483, "learning_rate": 9.928970319327685e-07, "loss": 0.0006, "step": 199560 }, { "epoch": 3.2654831056205516, "grad_norm": 0.013834052719175816, "learning_rate": 9.92470056708439e-07, "loss": 0.0001, "step": 199570 }, { "epoch": 3.265646731571627, "grad_norm": 0.06605332344770432, "learning_rate": 9.92043163194235e-07, "loss": 0.0004, "step": 199580 }, { "epoch": 3.265810357522703, "grad_norm": 0.431721568107605, "learning_rate": 9.916163513988637e-07, "loss": 0.0004, "step": 199590 }, { "epoch": 3.265973983473779, "grad_norm": 0.04124408960342407, "learning_rate": 9.911896213310235e-07, "loss": 0.0002, "step": 199600 }, { "epoch": 3.2661376094248546, "grad_norm": 0.00827087089419365, "learning_rate": 9.9076297299942e-07, "loss": 0.0002, "step": 199610 }, { "epoch": 3.2663012353759306, "grad_norm": 0.10291264206171036, "learning_rate": 9.90336406412748e-07, "loss": 0.0004, "step": 199620 }, { "epoch": 3.2664648613270066, "grad_norm": 0.017784666270017624, "learning_rate": 9.899099215797086e-07, "loss": 0.0002, "step": 199630 }, { "epoch": 3.266628487278082, "grad_norm": 0.00043360181734897196, "learning_rate": 9.894835185089924e-07, "loss": 0.0005, "step": 199640 }, { "epoch": 3.266792113229158, "grad_norm": 0.012765401042997837, "learning_rate": 9.890571972092993e-07, "loss": 0.0002, "step": 199650 }, { "epoch": 3.2669557391802337, "grad_norm": 0.013642384670674801, "learning_rate": 9.886309576893172e-07, "loss": 0.0004, "step": 199660 }, { "epoch": 3.2671193651313097, "grad_norm": 0.011434071697294712, "learning_rate": 9.882047999577398e-07, "loss": 0.0001, "step": 199670 }, { "epoch": 3.2672829910823857, "grad_norm": 0.002590825315564871, "learning_rate": 9.877787240232523e-07, "loss": 0.0003, "step": 199680 }, { "epoch": 3.2674466170334613, "grad_norm": 0.03251444920897484, "learning_rate": 9.873527298945468e-07, "loss": 0.0003, "step": 199690 }, { "epoch": 3.2676102429845373, "grad_norm": 0.003272962523624301, "learning_rate": 9.869268175803048e-07, "loss": 0.0002, "step": 199700 }, { "epoch": 3.2677738689356133, "grad_norm": 0.0009879912249743938, "learning_rate": 9.865009870892134e-07, "loss": 0.0005, "step": 199710 }, { "epoch": 3.267937494886689, "grad_norm": 0.001595141482539475, "learning_rate": 9.86075238429951e-07, "loss": 0.0001, "step": 199720 }, { "epoch": 3.268101120837765, "grad_norm": 0.015920981764793396, "learning_rate": 9.856495716112036e-07, "loss": 0.0003, "step": 199730 }, { "epoch": 3.268264746788841, "grad_norm": 0.025421010330319405, "learning_rate": 9.852239866416458e-07, "loss": 0.0001, "step": 199740 }, { "epoch": 3.2684283727399164, "grad_norm": 0.05262823775410652, "learning_rate": 9.847984835299578e-07, "loss": 0.0003, "step": 199750 }, { "epoch": 3.2685919986909924, "grad_norm": 0.015334313735365868, "learning_rate": 9.843730622848114e-07, "loss": 0.0002, "step": 199760 }, { "epoch": 3.2687556246420684, "grad_norm": 0.0006937335128895938, "learning_rate": 9.83947722914886e-07, "loss": 0.0001, "step": 199770 }, { "epoch": 3.268919250593144, "grad_norm": 0.0005935802473686635, "learning_rate": 9.835224654288494e-07, "loss": 0.0002, "step": 199780 }, { "epoch": 3.26908287654422, "grad_norm": 0.07910711318254471, "learning_rate": 9.830972898353746e-07, "loss": 0.0003, "step": 199790 }, { "epoch": 3.269246502495296, "grad_norm": 0.01954091154038906, "learning_rate": 9.82672196143129e-07, "loss": 0.0002, "step": 199800 }, { "epoch": 3.2694101284463715, "grad_norm": 0.01663377694785595, "learning_rate": 9.822471843607811e-07, "loss": 0.0002, "step": 199810 }, { "epoch": 3.2695737543974475, "grad_norm": 0.055277273058891296, "learning_rate": 9.818222544969959e-07, "loss": 0.0007, "step": 199820 }, { "epoch": 3.2697373803485235, "grad_norm": 0.0015187731478363276, "learning_rate": 9.813974065604387e-07, "loss": 0.0003, "step": 199830 }, { "epoch": 3.269901006299599, "grad_norm": 0.01599843241274357, "learning_rate": 9.809726405597692e-07, "loss": 0.0001, "step": 199840 }, { "epoch": 3.270064632250675, "grad_norm": 0.001681110356003046, "learning_rate": 9.805479565036497e-07, "loss": 0.0002, "step": 199850 }, { "epoch": 3.270228258201751, "grad_norm": 0.019979028031229973, "learning_rate": 9.801233544007389e-07, "loss": 0.0003, "step": 199860 }, { "epoch": 3.2703918841528266, "grad_norm": 0.0014634691178798676, "learning_rate": 9.796988342596958e-07, "loss": 0.0003, "step": 199870 }, { "epoch": 3.2705555101039026, "grad_norm": 0.015536532737314701, "learning_rate": 9.792743960891726e-07, "loss": 0.0003, "step": 199880 }, { "epoch": 3.2707191360549785, "grad_norm": 0.011456558480858803, "learning_rate": 9.78850039897825e-07, "loss": 0.0004, "step": 199890 }, { "epoch": 3.270882762006054, "grad_norm": 0.057311635464429855, "learning_rate": 9.784257656943053e-07, "loss": 0.0002, "step": 199900 }, { "epoch": 3.27104638795713, "grad_norm": 0.007678625173866749, "learning_rate": 9.78001573487265e-07, "loss": 0.0002, "step": 199910 }, { "epoch": 3.2712100139082056, "grad_norm": 0.015809064731001854, "learning_rate": 9.775774632853513e-07, "loss": 0.0003, "step": 199920 }, { "epoch": 3.2713736398592816, "grad_norm": 0.18398268520832062, "learning_rate": 9.771534350972118e-07, "loss": 0.0003, "step": 199930 }, { "epoch": 3.2715372658103576, "grad_norm": 0.038344606757164, "learning_rate": 9.767294889314926e-07, "loss": 0.0004, "step": 199940 }, { "epoch": 3.271700891761433, "grad_norm": 0.029686598107218742, "learning_rate": 9.76305624796839e-07, "loss": 0.0002, "step": 199950 }, { "epoch": 3.271864517712509, "grad_norm": 0.01441190019249916, "learning_rate": 9.758818427018897e-07, "loss": 0.0002, "step": 199960 }, { "epoch": 3.272028143663585, "grad_norm": 0.07719980925321579, "learning_rate": 9.754581426552878e-07, "loss": 0.0003, "step": 199970 }, { "epoch": 3.2721917696146607, "grad_norm": 0.023791396990418434, "learning_rate": 9.750345246656712e-07, "loss": 0.0003, "step": 199980 }, { "epoch": 3.2723553955657367, "grad_norm": 0.025566112250089645, "learning_rate": 9.746109887416794e-07, "loss": 0.0002, "step": 199990 }, { "epoch": 3.2725190215168127, "grad_norm": 0.015753930434584618, "learning_rate": 9.741875348919443e-07, "loss": 0.0003, "step": 200000 }, { "epoch": 3.2726826474678883, "grad_norm": 0.030980156734585762, "learning_rate": 9.737641631251015e-07, "loss": 0.0007, "step": 200010 }, { "epoch": 3.2728462734189643, "grad_norm": 0.03791609779000282, "learning_rate": 9.733408734497834e-07, "loss": 0.0005, "step": 200020 }, { "epoch": 3.2730098993700403, "grad_norm": 0.05202249065041542, "learning_rate": 9.72917665874622e-07, "loss": 0.0004, "step": 200030 }, { "epoch": 3.273173525321116, "grad_norm": 0.018257947638630867, "learning_rate": 9.724945404082425e-07, "loss": 0.0004, "step": 200040 }, { "epoch": 3.273337151272192, "grad_norm": 0.004124751780182123, "learning_rate": 9.720714970592743e-07, "loss": 0.0002, "step": 200050 }, { "epoch": 3.2735007772232674, "grad_norm": 0.04367455095052719, "learning_rate": 9.71648535836343e-07, "loss": 0.0003, "step": 200060 }, { "epoch": 3.2736644031743434, "grad_norm": 0.0009475418482907116, "learning_rate": 9.712256567480733e-07, "loss": 0.0001, "step": 200070 }, { "epoch": 3.2738280291254194, "grad_norm": 0.02367684803903103, "learning_rate": 9.708028598030844e-07, "loss": 0.0002, "step": 200080 }, { "epoch": 3.273991655076495, "grad_norm": 0.06964170187711716, "learning_rate": 9.703801450099987e-07, "loss": 0.0003, "step": 200090 }, { "epoch": 3.274155281027571, "grad_norm": 0.01042515691369772, "learning_rate": 9.699575123774346e-07, "loss": 0.0002, "step": 200100 }, { "epoch": 3.274318906978647, "grad_norm": 0.05811207741498947, "learning_rate": 9.695349619140109e-07, "loss": 0.0004, "step": 200110 }, { "epoch": 3.2744825329297225, "grad_norm": 0.04467709735035896, "learning_rate": 9.691124936283403e-07, "loss": 0.0003, "step": 200120 }, { "epoch": 3.2746461588807985, "grad_norm": 0.043146517127752304, "learning_rate": 9.686901075290372e-07, "loss": 0.0003, "step": 200130 }, { "epoch": 3.2748097848318745, "grad_norm": 0.03718460723757744, "learning_rate": 9.68267803624715e-07, "loss": 0.0002, "step": 200140 }, { "epoch": 3.27497341078295, "grad_norm": 0.004782631061971188, "learning_rate": 9.678455819239846e-07, "loss": 0.0003, "step": 200150 }, { "epoch": 3.275137036734026, "grad_norm": 0.0017415977781638503, "learning_rate": 9.674234424354522e-07, "loss": 0.0004, "step": 200160 }, { "epoch": 3.275300662685102, "grad_norm": 0.013359288685023785, "learning_rate": 9.670013851677261e-07, "loss": 0.0007, "step": 200170 }, { "epoch": 3.2754642886361776, "grad_norm": 0.04415007680654526, "learning_rate": 9.665794101294119e-07, "loss": 0.0003, "step": 200180 }, { "epoch": 3.2756279145872536, "grad_norm": 0.09341999143362045, "learning_rate": 9.661575173291143e-07, "loss": 0.0004, "step": 200190 }, { "epoch": 3.2757915405383295, "grad_norm": 0.018787477165460587, "learning_rate": 9.657357067754335e-07, "loss": 0.0002, "step": 200200 }, { "epoch": 3.275955166489405, "grad_norm": 0.04701843485236168, "learning_rate": 9.653139784769706e-07, "loss": 0.0003, "step": 200210 }, { "epoch": 3.276118792440481, "grad_norm": 0.022692810744047165, "learning_rate": 9.648923324423238e-07, "loss": 0.0004, "step": 200220 }, { "epoch": 3.276282418391557, "grad_norm": 0.020876947790384293, "learning_rate": 9.64470768680092e-07, "loss": 0.0001, "step": 200230 }, { "epoch": 3.2764460443426326, "grad_norm": 0.01680208183825016, "learning_rate": 9.640492871988683e-07, "loss": 0.0003, "step": 200240 }, { "epoch": 3.2766096702937086, "grad_norm": 0.039963431656360626, "learning_rate": 9.63627888007248e-07, "loss": 0.0004, "step": 200250 }, { "epoch": 3.2767732962447846, "grad_norm": 0.00042187594226561487, "learning_rate": 9.632065711138194e-07, "loss": 0.0003, "step": 200260 }, { "epoch": 3.27693692219586, "grad_norm": 0.001963474089279771, "learning_rate": 9.627853365271788e-07, "loss": 0.0003, "step": 200270 }, { "epoch": 3.277100548146936, "grad_norm": 0.07403164356946945, "learning_rate": 9.6236418425591e-07, "loss": 0.0007, "step": 200280 }, { "epoch": 3.277264174098012, "grad_norm": 0.09468286484479904, "learning_rate": 9.619431143086033e-07, "loss": 0.0001, "step": 200290 }, { "epoch": 3.2774278000490877, "grad_norm": 0.028220606967806816, "learning_rate": 9.615221266938397e-07, "loss": 0.0003, "step": 200300 }, { "epoch": 3.2775914260001637, "grad_norm": 0.06655927747488022, "learning_rate": 9.611012214202076e-07, "loss": 0.0003, "step": 200310 }, { "epoch": 3.2777550519512393, "grad_norm": 0.01883041113615036, "learning_rate": 9.606803984962853e-07, "loss": 0.0003, "step": 200320 }, { "epoch": 3.2779186779023153, "grad_norm": 0.008203214034438133, "learning_rate": 9.602596579306557e-07, "loss": 0.0002, "step": 200330 }, { "epoch": 3.2780823038533913, "grad_norm": 0.0156958419829607, "learning_rate": 9.598389997318934e-07, "loss": 0.0002, "step": 200340 }, { "epoch": 3.278245929804467, "grad_norm": 0.003947739489376545, "learning_rate": 9.594184239085808e-07, "loss": 0.0004, "step": 200350 }, { "epoch": 3.278409555755543, "grad_norm": 0.056332677602767944, "learning_rate": 9.589979304692882e-07, "loss": 0.0006, "step": 200360 }, { "epoch": 3.278573181706619, "grad_norm": 0.0017934823408722878, "learning_rate": 9.585775194225927e-07, "loss": 0.0002, "step": 200370 }, { "epoch": 3.2787368076576944, "grad_norm": 0.0034304987639188766, "learning_rate": 9.58157190777062e-07, "loss": 0.0001, "step": 200380 }, { "epoch": 3.2789004336087704, "grad_norm": 0.0016490567941218615, "learning_rate": 9.577369445412716e-07, "loss": 0.0004, "step": 200390 }, { "epoch": 3.2790640595598464, "grad_norm": 0.028625622391700745, "learning_rate": 9.573167807237848e-07, "loss": 0.0003, "step": 200400 }, { "epoch": 3.279227685510922, "grad_norm": 0.005725675728172064, "learning_rate": 9.568966993331729e-07, "loss": 0.0005, "step": 200410 }, { "epoch": 3.279391311461998, "grad_norm": 0.06867311149835587, "learning_rate": 9.564767003779974e-07, "loss": 0.0003, "step": 200420 }, { "epoch": 3.2795549374130735, "grad_norm": 0.026974430307745934, "learning_rate": 9.560567838668223e-07, "loss": 0.0004, "step": 200430 }, { "epoch": 3.2797185633641495, "grad_norm": 0.025277547538280487, "learning_rate": 9.556369498082108e-07, "loss": 0.0003, "step": 200440 }, { "epoch": 3.2798821893152255, "grad_norm": 0.015125883743166924, "learning_rate": 9.552171982107233e-07, "loss": 0.0003, "step": 200450 }, { "epoch": 3.280045815266301, "grad_norm": 0.000987453036941588, "learning_rate": 9.547975290829158e-07, "loss": 0.0002, "step": 200460 }, { "epoch": 3.280209441217377, "grad_norm": 0.06392742693424225, "learning_rate": 9.543779424333467e-07, "loss": 0.0001, "step": 200470 }, { "epoch": 3.280373067168453, "grad_norm": 0.017867160961031914, "learning_rate": 9.539584382705708e-07, "loss": 0.0003, "step": 200480 }, { "epoch": 3.2805366931195286, "grad_norm": 0.05609971657395363, "learning_rate": 9.535390166031422e-07, "loss": 0.0003, "step": 200490 }, { "epoch": 3.2807003190706046, "grad_norm": 0.0005528361070901155, "learning_rate": 9.531196774396107e-07, "loss": 0.0004, "step": 200500 }, { "epoch": 3.2808639450216805, "grad_norm": 0.00781580712646246, "learning_rate": 9.527004207885265e-07, "loss": 0.0002, "step": 200510 }, { "epoch": 3.281027570972756, "grad_norm": 0.011359576135873795, "learning_rate": 9.522812466584392e-07, "loss": 0.0003, "step": 200520 }, { "epoch": 3.281191196923832, "grad_norm": 0.07497256249189377, "learning_rate": 9.518621550578961e-07, "loss": 0.0003, "step": 200530 }, { "epoch": 3.281354822874908, "grad_norm": 0.012353925965726376, "learning_rate": 9.514431459954398e-07, "loss": 0.0002, "step": 200540 }, { "epoch": 3.2815184488259836, "grad_norm": 0.015920888632535934, "learning_rate": 9.510242194796144e-07, "loss": 0.0003, "step": 200550 }, { "epoch": 3.2816820747770596, "grad_norm": 0.08429495990276337, "learning_rate": 9.50605375518961e-07, "loss": 0.0002, "step": 200560 }, { "epoch": 3.2818457007281356, "grad_norm": 0.02800856903195381, "learning_rate": 9.501866141220223e-07, "loss": 0.0006, "step": 200570 }, { "epoch": 3.282009326679211, "grad_norm": 0.0007053965236991644, "learning_rate": 9.497679352973321e-07, "loss": 0.0003, "step": 200580 }, { "epoch": 3.282172952630287, "grad_norm": 0.03472466021776199, "learning_rate": 9.493493390534298e-07, "loss": 0.0003, "step": 200590 }, { "epoch": 3.282336578581363, "grad_norm": 0.027712102979421616, "learning_rate": 9.489308253988494e-07, "loss": 0.0001, "step": 200600 }, { "epoch": 3.2825002045324387, "grad_norm": 0.014721873216331005, "learning_rate": 9.485123943421254e-07, "loss": 0.0001, "step": 200610 }, { "epoch": 3.2826638304835147, "grad_norm": 0.0199537705630064, "learning_rate": 9.480940458917864e-07, "loss": 0.0001, "step": 200620 }, { "epoch": 3.2828274564345907, "grad_norm": 0.0003324691788293421, "learning_rate": 9.476757800563641e-07, "loss": 0.0003, "step": 200630 }, { "epoch": 3.2829910823856663, "grad_norm": 0.03877231478691101, "learning_rate": 9.472575968443864e-07, "loss": 0.0002, "step": 200640 }, { "epoch": 3.2831547083367423, "grad_norm": 0.051473189145326614, "learning_rate": 9.468394962643807e-07, "loss": 0.0005, "step": 200650 }, { "epoch": 3.2833183342878183, "grad_norm": 0.08252160996198654, "learning_rate": 9.464214783248693e-07, "loss": 0.001, "step": 200660 }, { "epoch": 3.283481960238894, "grad_norm": 0.015787385404109955, "learning_rate": 9.460035430343767e-07, "loss": 0.0003, "step": 200670 }, { "epoch": 3.28364558618997, "grad_norm": 0.0013406705111265182, "learning_rate": 9.455856904014238e-07, "loss": 0.0002, "step": 200680 }, { "epoch": 3.283809212141046, "grad_norm": 0.036101534962654114, "learning_rate": 9.451679204345321e-07, "loss": 0.0006, "step": 200690 }, { "epoch": 3.2839728380921214, "grad_norm": 0.0018434457015246153, "learning_rate": 9.447502331422159e-07, "loss": 0.0002, "step": 200700 }, { "epoch": 3.2841364640431974, "grad_norm": 0.010973901487886906, "learning_rate": 9.443326285329945e-07, "loss": 0.0002, "step": 200710 }, { "epoch": 3.284300089994273, "grad_norm": 0.006865539588034153, "learning_rate": 9.439151066153812e-07, "loss": 0.0003, "step": 200720 }, { "epoch": 3.284463715945349, "grad_norm": 0.000440118252299726, "learning_rate": 9.434976673978907e-07, "loss": 0.0003, "step": 200730 }, { "epoch": 3.284627341896425, "grad_norm": 0.029615668579936028, "learning_rate": 9.430803108890313e-07, "loss": 0.0003, "step": 200740 }, { "epoch": 3.2847909678475005, "grad_norm": 0.011547314934432507, "learning_rate": 9.426630370973144e-07, "loss": 0.0005, "step": 200750 }, { "epoch": 3.2849545937985765, "grad_norm": 0.09843946248292923, "learning_rate": 9.422458460312473e-07, "loss": 0.0003, "step": 200760 }, { "epoch": 3.2851182197496525, "grad_norm": 0.07435351610183716, "learning_rate": 9.418287376993379e-07, "loss": 0.0002, "step": 200770 }, { "epoch": 3.285281845700728, "grad_norm": 0.0613178014755249, "learning_rate": 9.414117121100874e-07, "loss": 0.0004, "step": 200780 }, { "epoch": 3.285445471651804, "grad_norm": 0.038646139204502106, "learning_rate": 9.409947692720012e-07, "loss": 0.0003, "step": 200790 }, { "epoch": 3.28560909760288, "grad_norm": 0.05915097892284393, "learning_rate": 9.405779091935791e-07, "loss": 0.0003, "step": 200800 }, { "epoch": 3.2857727235539556, "grad_norm": 0.018951479345560074, "learning_rate": 9.40161131883322e-07, "loss": 0.0004, "step": 200810 }, { "epoch": 3.2859363495050316, "grad_norm": 0.007791012525558472, "learning_rate": 9.397444373497255e-07, "loss": 0.0009, "step": 200820 }, { "epoch": 3.286099975456107, "grad_norm": 0.014316320419311523, "learning_rate": 9.393278256012872e-07, "loss": 0.0003, "step": 200830 }, { "epoch": 3.286263601407183, "grad_norm": 0.07731378823518753, "learning_rate": 9.389112966465008e-07, "loss": 0.0004, "step": 200840 }, { "epoch": 3.286427227358259, "grad_norm": 0.05469399318099022, "learning_rate": 9.384948504938607e-07, "loss": 0.0002, "step": 200850 }, { "epoch": 3.2865908533093346, "grad_norm": 0.042835041880607605, "learning_rate": 9.38078487151855e-07, "loss": 0.0004, "step": 200860 }, { "epoch": 3.2867544792604106, "grad_norm": 0.005463927984237671, "learning_rate": 9.376622066289742e-07, "loss": 0.0004, "step": 200870 }, { "epoch": 3.2869181052114866, "grad_norm": 0.0006293910555541515, "learning_rate": 9.372460089337066e-07, "loss": 0.0009, "step": 200880 }, { "epoch": 3.287081731162562, "grad_norm": 0.10745623707771301, "learning_rate": 9.368298940745385e-07, "loss": 0.0008, "step": 200890 }, { "epoch": 3.287245357113638, "grad_norm": 0.052699606865644455, "learning_rate": 9.364138620599522e-07, "loss": 0.0001, "step": 200900 }, { "epoch": 3.287408983064714, "grad_norm": 0.05395015701651573, "learning_rate": 9.359979128984326e-07, "loss": 0.0003, "step": 200910 }, { "epoch": 3.2875726090157897, "grad_norm": 0.03210518881678581, "learning_rate": 9.355820465984566e-07, "loss": 0.0002, "step": 200920 }, { "epoch": 3.2877362349668657, "grad_norm": 0.0688377395272255, "learning_rate": 9.351662631685082e-07, "loss": 0.0005, "step": 200930 }, { "epoch": 3.2878998609179417, "grad_norm": 0.04242047294974327, "learning_rate": 9.347505626170622e-07, "loss": 0.0002, "step": 200940 }, { "epoch": 3.2880634868690173, "grad_norm": 0.009892809204757214, "learning_rate": 9.343349449525957e-07, "loss": 0.0003, "step": 200950 }, { "epoch": 3.2882271128200933, "grad_norm": 0.007860788144171238, "learning_rate": 9.339194101835791e-07, "loss": 0.0006, "step": 200960 }, { "epoch": 3.2883907387711693, "grad_norm": 0.09448640048503876, "learning_rate": 9.335039583184907e-07, "loss": 0.0006, "step": 200970 }, { "epoch": 3.288554364722245, "grad_norm": 0.0013226555893197656, "learning_rate": 9.330885893657965e-07, "loss": 0.0004, "step": 200980 }, { "epoch": 3.288717990673321, "grad_norm": 0.005134375300258398, "learning_rate": 9.326733033339686e-07, "loss": 0.0005, "step": 200990 }, { "epoch": 3.288881616624397, "grad_norm": 0.042566731572151184, "learning_rate": 9.322581002314701e-07, "loss": 0.0004, "step": 201000 }, { "epoch": 3.2890452425754724, "grad_norm": 0.006333881989121437, "learning_rate": 9.318429800667722e-07, "loss": 0.0002, "step": 201010 }, { "epoch": 3.2892088685265484, "grad_norm": 0.017409762367606163, "learning_rate": 9.314279428483353e-07, "loss": 0.0002, "step": 201020 }, { "epoch": 3.2893724944776244, "grad_norm": 0.0019621839746832848, "learning_rate": 9.31012988584623e-07, "loss": 0.0005, "step": 201030 }, { "epoch": 3.2895361204287, "grad_norm": 0.05502589792013168, "learning_rate": 9.305981172840933e-07, "loss": 0.0003, "step": 201040 }, { "epoch": 3.289699746379776, "grad_norm": 0.013038898818194866, "learning_rate": 9.301833289552092e-07, "loss": 0.0002, "step": 201050 }, { "epoch": 3.289863372330852, "grad_norm": 0.02281132899224758, "learning_rate": 9.297686236064246e-07, "loss": 0.0003, "step": 201060 }, { "epoch": 3.2900269982819275, "grad_norm": 0.0424552783370018, "learning_rate": 9.293540012461977e-07, "loss": 0.0006, "step": 201070 }, { "epoch": 3.2901906242330035, "grad_norm": 0.04482399672269821, "learning_rate": 9.289394618829794e-07, "loss": 0.0003, "step": 201080 }, { "epoch": 3.290354250184079, "grad_norm": 0.00827814731746912, "learning_rate": 9.285250055252232e-07, "loss": 0.0002, "step": 201090 }, { "epoch": 3.290517876135155, "grad_norm": 0.04053470119833946, "learning_rate": 9.281106321813793e-07, "loss": 0.0004, "step": 201100 }, { "epoch": 3.290681502086231, "grad_norm": 0.003387963632121682, "learning_rate": 9.276963418598978e-07, "loss": 0.0004, "step": 201110 }, { "epoch": 3.2908451280373066, "grad_norm": 0.0013636482181027532, "learning_rate": 9.272821345692235e-07, "loss": 0.0003, "step": 201120 }, { "epoch": 3.2910087539883826, "grad_norm": 0.02593993954360485, "learning_rate": 9.268680103178024e-07, "loss": 0.0002, "step": 201130 }, { "epoch": 3.2911723799394585, "grad_norm": 0.13668662309646606, "learning_rate": 9.26453969114079e-07, "loss": 0.0006, "step": 201140 }, { "epoch": 3.291336005890534, "grad_norm": 0.019808493554592133, "learning_rate": 9.260400109664952e-07, "loss": 0.0002, "step": 201150 }, { "epoch": 3.29149963184161, "grad_norm": 0.0018727934220805764, "learning_rate": 9.2562613588349e-07, "loss": 0.0002, "step": 201160 }, { "epoch": 3.291663257792686, "grad_norm": 0.01917283982038498, "learning_rate": 9.252123438735028e-07, "loss": 0.0003, "step": 201170 }, { "epoch": 3.2918268837437616, "grad_norm": 0.043407343327999115, "learning_rate": 9.247986349449705e-07, "loss": 0.0003, "step": 201180 }, { "epoch": 3.2919905096948376, "grad_norm": 0.012220972217619419, "learning_rate": 9.243850091063295e-07, "loss": 0.0003, "step": 201190 }, { "epoch": 3.292154135645913, "grad_norm": 0.056126710027456284, "learning_rate": 9.239714663660105e-07, "loss": 0.0004, "step": 201200 }, { "epoch": 3.292317761596989, "grad_norm": 0.004669009242206812, "learning_rate": 9.235580067324468e-07, "loss": 0.0001, "step": 201210 }, { "epoch": 3.292481387548065, "grad_norm": 0.016623258590698242, "learning_rate": 9.231446302140684e-07, "loss": 0.0005, "step": 201220 }, { "epoch": 3.2926450134991407, "grad_norm": 0.001018986222334206, "learning_rate": 9.227313368193053e-07, "loss": 0.0002, "step": 201230 }, { "epoch": 3.2928086394502167, "grad_norm": 0.002619966398924589, "learning_rate": 9.223181265565812e-07, "loss": 0.0002, "step": 201240 }, { "epoch": 3.2929722654012927, "grad_norm": 0.0007557397475466132, "learning_rate": 9.219049994343226e-07, "loss": 0.0001, "step": 201250 }, { "epoch": 3.2931358913523683, "grad_norm": 0.0454275980591774, "learning_rate": 9.21491955460953e-07, "loss": 0.0005, "step": 201260 }, { "epoch": 3.2932995173034443, "grad_norm": 0.025484636425971985, "learning_rate": 9.210789946448945e-07, "loss": 0.0003, "step": 201270 }, { "epoch": 3.2934631432545203, "grad_norm": 0.1485198438167572, "learning_rate": 9.206661169945652e-07, "loss": 0.0005, "step": 201280 }, { "epoch": 3.293626769205596, "grad_norm": 0.009494914673268795, "learning_rate": 9.202533225183846e-07, "loss": 0.0002, "step": 201290 }, { "epoch": 3.293790395156672, "grad_norm": 0.0038053581956773996, "learning_rate": 9.198406112247693e-07, "loss": 0.0003, "step": 201300 }, { "epoch": 3.293954021107748, "grad_norm": 0.0352964922785759, "learning_rate": 9.194279831221347e-07, "loss": 0.0004, "step": 201310 }, { "epoch": 3.2941176470588234, "grad_norm": 0.04054992273449898, "learning_rate": 9.190154382188921e-07, "loss": 0.0012, "step": 201320 }, { "epoch": 3.2942812730098994, "grad_norm": 0.04517276957631111, "learning_rate": 9.186029765234538e-07, "loss": 0.0003, "step": 201330 }, { "epoch": 3.2944448989609754, "grad_norm": 0.011655110865831375, "learning_rate": 9.181905980442296e-07, "loss": 0.0003, "step": 201340 }, { "epoch": 3.294608524912051, "grad_norm": 0.04120515659451485, "learning_rate": 9.177783027896292e-07, "loss": 0.0002, "step": 201350 }, { "epoch": 3.294772150863127, "grad_norm": 0.007643811404705048, "learning_rate": 9.173660907680559e-07, "loss": 0.0004, "step": 201360 }, { "epoch": 3.294935776814203, "grad_norm": 0.006447784602642059, "learning_rate": 9.169539619879158e-07, "loss": 0.0003, "step": 201370 }, { "epoch": 3.2950994027652785, "grad_norm": 0.06916386634111404, "learning_rate": 9.165419164576122e-07, "loss": 0.0004, "step": 201380 }, { "epoch": 3.2952630287163545, "grad_norm": 0.037599701434373856, "learning_rate": 9.161299541855473e-07, "loss": 0.0001, "step": 201390 }, { "epoch": 3.2954266546674305, "grad_norm": 0.012658587656915188, "learning_rate": 9.157180751801182e-07, "loss": 0.0001, "step": 201400 }, { "epoch": 3.295590280618506, "grad_norm": 0.027299165725708008, "learning_rate": 9.153062794497236e-07, "loss": 0.0005, "step": 201410 }, { "epoch": 3.295753906569582, "grad_norm": 0.024486837908625603, "learning_rate": 9.148945670027604e-07, "loss": 0.0003, "step": 201420 }, { "epoch": 3.295917532520658, "grad_norm": 0.011776300147175789, "learning_rate": 9.144829378476239e-07, "loss": 0.0003, "step": 201430 }, { "epoch": 3.2960811584717336, "grad_norm": 0.0603807158768177, "learning_rate": 9.140713919927047e-07, "loss": 0.0004, "step": 201440 }, { "epoch": 3.2962447844228095, "grad_norm": 0.059041813015937805, "learning_rate": 9.136599294463944e-07, "loss": 0.0004, "step": 201450 }, { "epoch": 3.2964084103738855, "grad_norm": 0.0007508861599490047, "learning_rate": 9.132485502170829e-07, "loss": 0.0002, "step": 201460 }, { "epoch": 3.296572036324961, "grad_norm": 0.08940261602401733, "learning_rate": 9.128372543131592e-07, "loss": 0.0003, "step": 201470 }, { "epoch": 3.296735662276037, "grad_norm": 0.03585825860500336, "learning_rate": 9.12426041743007e-07, "loss": 0.0002, "step": 201480 }, { "epoch": 3.2968992882271126, "grad_norm": 0.031200269237160683, "learning_rate": 9.120149125150108e-07, "loss": 0.0003, "step": 201490 }, { "epoch": 3.2970629141781886, "grad_norm": 0.10067532956600189, "learning_rate": 9.116038666375543e-07, "loss": 0.0005, "step": 201500 }, { "epoch": 3.2972265401292646, "grad_norm": 0.0247307438403368, "learning_rate": 9.111929041190188e-07, "loss": 0.0004, "step": 201510 }, { "epoch": 3.29739016608034, "grad_norm": 0.013109545223414898, "learning_rate": 9.107820249677812e-07, "loss": 0.0004, "step": 201520 }, { "epoch": 3.297553792031416, "grad_norm": 0.024923209100961685, "learning_rate": 9.103712291922223e-07, "loss": 0.0005, "step": 201530 }, { "epoch": 3.297717417982492, "grad_norm": 0.00019321605213917792, "learning_rate": 9.09960516800713e-07, "loss": 0.0004, "step": 201540 }, { "epoch": 3.2978810439335677, "grad_norm": 0.002682384569197893, "learning_rate": 9.095498878016329e-07, "loss": 0.0002, "step": 201550 }, { "epoch": 3.2980446698846437, "grad_norm": 0.023154491558670998, "learning_rate": 9.091393422033501e-07, "loss": 0.0003, "step": 201560 }, { "epoch": 3.2982082958357197, "grad_norm": 0.0028044860810041428, "learning_rate": 9.087288800142385e-07, "loss": 0.0002, "step": 201570 }, { "epoch": 3.2983719217867953, "grad_norm": 0.06562749296426773, "learning_rate": 9.083185012426632e-07, "loss": 0.0002, "step": 201580 }, { "epoch": 3.2985355477378713, "grad_norm": 0.046601973474025726, "learning_rate": 9.079082058969963e-07, "loss": 0.0002, "step": 201590 }, { "epoch": 3.298699173688947, "grad_norm": 0.008281830698251724, "learning_rate": 9.074979939855993e-07, "loss": 0.0003, "step": 201600 }, { "epoch": 3.298862799640023, "grad_norm": 0.0038654168602079153, "learning_rate": 9.070878655168392e-07, "loss": 0.0003, "step": 201610 }, { "epoch": 3.299026425591099, "grad_norm": 0.028217609971761703, "learning_rate": 9.066778204990745e-07, "loss": 0.0007, "step": 201620 }, { "epoch": 3.2991900515421744, "grad_norm": 0.07001553475856781, "learning_rate": 9.062678589406698e-07, "loss": 0.0003, "step": 201630 }, { "epoch": 3.2993536774932504, "grad_norm": 0.0240879375487566, "learning_rate": 9.058579808499806e-07, "loss": 0.0002, "step": 201640 }, { "epoch": 3.2995173034443264, "grad_norm": 0.06860871613025665, "learning_rate": 9.054481862353665e-07, "loss": 0.0004, "step": 201650 }, { "epoch": 3.299680929395402, "grad_norm": 0.02890871837735176, "learning_rate": 9.050384751051789e-07, "loss": 0.0004, "step": 201660 }, { "epoch": 3.299844555346478, "grad_norm": 0.0008068291936069727, "learning_rate": 9.046288474677772e-07, "loss": 0.0002, "step": 201670 }, { "epoch": 3.300008181297554, "grad_norm": 0.06615956872701645, "learning_rate": 9.042193033315084e-07, "loss": 0.0001, "step": 201680 }, { "epoch": 3.3001718072486295, "grad_norm": 0.010694256983697414, "learning_rate": 9.038098427047265e-07, "loss": 0.0002, "step": 201690 }, { "epoch": 3.3003354331997055, "grad_norm": 0.027973564341664314, "learning_rate": 9.034004655957767e-07, "loss": 0.0003, "step": 201700 }, { "epoch": 3.3004990591507815, "grad_norm": 0.03164348751306534, "learning_rate": 9.029911720130075e-07, "loss": 0.0003, "step": 201710 }, { "epoch": 3.300662685101857, "grad_norm": 0.011937648989260197, "learning_rate": 9.025819619647636e-07, "loss": 0.0003, "step": 201720 }, { "epoch": 3.300826311052933, "grad_norm": 0.00017808342818170786, "learning_rate": 9.021728354593906e-07, "loss": 0.0003, "step": 201730 }, { "epoch": 3.300989937004009, "grad_norm": 0.08543913066387177, "learning_rate": 9.017637925052264e-07, "loss": 0.0005, "step": 201740 }, { "epoch": 3.3011535629550846, "grad_norm": 0.01801675744354725, "learning_rate": 9.013548331106136e-07, "loss": 0.0002, "step": 201750 }, { "epoch": 3.3013171889061605, "grad_norm": 0.038991622626781464, "learning_rate": 9.009459572838896e-07, "loss": 0.0003, "step": 201760 }, { "epoch": 3.3014808148572365, "grad_norm": 0.0010570732410997152, "learning_rate": 9.005371650333932e-07, "loss": 0.0003, "step": 201770 }, { "epoch": 3.301644440808312, "grad_norm": 0.0072220019064843655, "learning_rate": 9.001284563674562e-07, "loss": 0.0002, "step": 201780 }, { "epoch": 3.301808066759388, "grad_norm": 0.09562043845653534, "learning_rate": 8.997198312944128e-07, "loss": 0.0004, "step": 201790 }, { "epoch": 3.301971692710464, "grad_norm": 0.02773110568523407, "learning_rate": 8.993112898225948e-07, "loss": 0.0005, "step": 201800 }, { "epoch": 3.3021353186615396, "grad_norm": 0.05130105838179588, "learning_rate": 8.989028319603338e-07, "loss": 0.0002, "step": 201810 }, { "epoch": 3.3022989446126156, "grad_norm": 0.041575174778699875, "learning_rate": 8.984944577159549e-07, "loss": 0.0005, "step": 201820 }, { "epoch": 3.3024625705636916, "grad_norm": 0.000616672623436898, "learning_rate": 8.980861670977853e-07, "loss": 0.0003, "step": 201830 }, { "epoch": 3.302626196514767, "grad_norm": 0.0894264504313469, "learning_rate": 8.976779601141505e-07, "loss": 0.0005, "step": 201840 }, { "epoch": 3.302789822465843, "grad_norm": 0.05817392095923424, "learning_rate": 8.972698367733746e-07, "loss": 0.0004, "step": 201850 }, { "epoch": 3.3029534484169187, "grad_norm": 0.0151675371453166, "learning_rate": 8.968617970837762e-07, "loss": 0.0004, "step": 201860 }, { "epoch": 3.3031170743679947, "grad_norm": 0.002212886931374669, "learning_rate": 8.96453841053676e-07, "loss": 0.0002, "step": 201870 }, { "epoch": 3.3032807003190707, "grad_norm": 0.015775911509990692, "learning_rate": 8.960459686913919e-07, "loss": 0.0003, "step": 201880 }, { "epoch": 3.3034443262701463, "grad_norm": 0.02075069397687912, "learning_rate": 8.956381800052416e-07, "loss": 0.0002, "step": 201890 }, { "epoch": 3.3036079522212223, "grad_norm": 0.0276668518781662, "learning_rate": 8.952304750035373e-07, "loss": 0.0003, "step": 201900 }, { "epoch": 3.3037715781722983, "grad_norm": 0.0037532399874180555, "learning_rate": 8.948228536945919e-07, "loss": 0.0003, "step": 201910 }, { "epoch": 3.303935204123374, "grad_norm": 9.775438229553401e-05, "learning_rate": 8.94415316086718e-07, "loss": 0.0001, "step": 201920 }, { "epoch": 3.30409883007445, "grad_norm": 0.018190065398812294, "learning_rate": 8.940078621882248e-07, "loss": 0.0003, "step": 201930 }, { "epoch": 3.304262456025526, "grad_norm": 0.02207602933049202, "learning_rate": 8.936004920074182e-07, "loss": 0.0001, "step": 201940 }, { "epoch": 3.3044260819766014, "grad_norm": 0.00904550775885582, "learning_rate": 8.931932055526049e-07, "loss": 0.0003, "step": 201950 }, { "epoch": 3.3045897079276774, "grad_norm": 0.0014556406531482935, "learning_rate": 8.9278600283209e-07, "loss": 0.0009, "step": 201960 }, { "epoch": 3.3047533338787534, "grad_norm": 0.030771251767873764, "learning_rate": 8.923788838541764e-07, "loss": 0.0003, "step": 201970 }, { "epoch": 3.304916959829829, "grad_norm": 0.009450080804526806, "learning_rate": 8.919718486271628e-07, "loss": 0.0003, "step": 201980 }, { "epoch": 3.305080585780905, "grad_norm": 0.042763981968164444, "learning_rate": 8.915648971593493e-07, "loss": 0.0004, "step": 201990 }, { "epoch": 3.3052442117319805, "grad_norm": 0.0028714679647237062, "learning_rate": 8.911580294590332e-07, "loss": 0.0005, "step": 202000 }, { "epoch": 3.3054078376830565, "grad_norm": 0.03765484690666199, "learning_rate": 8.907512455345124e-07, "loss": 0.0004, "step": 202010 }, { "epoch": 3.3055714636341325, "grad_norm": 0.06682929396629333, "learning_rate": 8.903445453940773e-07, "loss": 0.0004, "step": 202020 }, { "epoch": 3.305735089585208, "grad_norm": 0.009784546680748463, "learning_rate": 8.899379290460214e-07, "loss": 0.0003, "step": 202030 }, { "epoch": 3.305898715536284, "grad_norm": 0.012119034305214882, "learning_rate": 8.895313964986363e-07, "loss": 0.0002, "step": 202040 }, { "epoch": 3.30606234148736, "grad_norm": 0.012794350273907185, "learning_rate": 8.89124947760211e-07, "loss": 0.0001, "step": 202050 }, { "epoch": 3.3062259674384356, "grad_norm": 0.015180638991296291, "learning_rate": 8.887185828390304e-07, "loss": 0.0002, "step": 202060 }, { "epoch": 3.3063895933895115, "grad_norm": 0.04964468255639076, "learning_rate": 8.88312301743382e-07, "loss": 0.0005, "step": 202070 }, { "epoch": 3.3065532193405875, "grad_norm": 0.00012749161396641284, "learning_rate": 8.879061044815484e-07, "loss": 0.0002, "step": 202080 }, { "epoch": 3.306716845291663, "grad_norm": 0.0024067291524261236, "learning_rate": 8.874999910618138e-07, "loss": 0.0001, "step": 202090 }, { "epoch": 3.306880471242739, "grad_norm": 0.0031048262026160955, "learning_rate": 8.870939614924551e-07, "loss": 0.0004, "step": 202100 }, { "epoch": 3.307044097193815, "grad_norm": 0.0374889001250267, "learning_rate": 8.866880157817531e-07, "loss": 0.0002, "step": 202110 }, { "epoch": 3.3072077231448906, "grad_norm": 0.015542914159595966, "learning_rate": 8.86282153937984e-07, "loss": 0.0004, "step": 202120 }, { "epoch": 3.3073713490959666, "grad_norm": 0.012408850714564323, "learning_rate": 8.858763759694245e-07, "loss": 0.0004, "step": 202130 }, { "epoch": 3.3075349750470426, "grad_norm": 0.025225527584552765, "learning_rate": 8.854706818843456e-07, "loss": 0.0003, "step": 202140 }, { "epoch": 3.307698600998118, "grad_norm": 0.026761021465063095, "learning_rate": 8.850650716910198e-07, "loss": 0.001, "step": 202150 }, { "epoch": 3.307862226949194, "grad_norm": 0.00046802256838418543, "learning_rate": 8.84659545397718e-07, "loss": 0.0001, "step": 202160 }, { "epoch": 3.30802585290027, "grad_norm": 0.004396763630211353, "learning_rate": 8.842541030127089e-07, "loss": 0.0002, "step": 202170 }, { "epoch": 3.3081894788513457, "grad_norm": 0.10322929918766022, "learning_rate": 8.838487445442573e-07, "loss": 0.0005, "step": 202180 }, { "epoch": 3.3083531048024217, "grad_norm": 0.0541209913790226, "learning_rate": 8.834434700006305e-07, "loss": 0.0003, "step": 202190 }, { "epoch": 3.3085167307534977, "grad_norm": 0.014425420202314854, "learning_rate": 8.830382793900877e-07, "loss": 0.0003, "step": 202200 }, { "epoch": 3.3086803567045733, "grad_norm": 0.026195155456662178, "learning_rate": 8.826331727208958e-07, "loss": 0.0003, "step": 202210 }, { "epoch": 3.3088439826556493, "grad_norm": 0.015944743528962135, "learning_rate": 8.822281500013102e-07, "loss": 0.0001, "step": 202220 }, { "epoch": 3.3090076086067253, "grad_norm": 0.0268503800034523, "learning_rate": 8.818232112395919e-07, "loss": 0.0003, "step": 202230 }, { "epoch": 3.309171234557801, "grad_norm": 0.028812041506171227, "learning_rate": 8.814183564439937e-07, "loss": 0.0002, "step": 202240 }, { "epoch": 3.309334860508877, "grad_norm": 0.0018317836802452803, "learning_rate": 8.810135856227747e-07, "loss": 0.0002, "step": 202250 }, { "epoch": 3.3094984864599524, "grad_norm": 0.030175460502505302, "learning_rate": 8.806088987841843e-07, "loss": 0.0004, "step": 202260 }, { "epoch": 3.3096621124110284, "grad_norm": 0.017741801217198372, "learning_rate": 8.802042959364765e-07, "loss": 0.0004, "step": 202270 }, { "epoch": 3.3098257383621044, "grad_norm": 0.018704941496253014, "learning_rate": 8.797997770878969e-07, "loss": 0.0004, "step": 202280 }, { "epoch": 3.30998936431318, "grad_norm": 0.03439503535628319, "learning_rate": 8.793953422466983e-07, "loss": 0.0003, "step": 202290 }, { "epoch": 3.310152990264256, "grad_norm": 0.013875382021069527, "learning_rate": 8.789909914211226e-07, "loss": 0.0019, "step": 202300 }, { "epoch": 3.310316616215332, "grad_norm": 0.04728320986032486, "learning_rate": 8.785867246194174e-07, "loss": 0.0005, "step": 202310 }, { "epoch": 3.3104802421664075, "grad_norm": 0.01735064946115017, "learning_rate": 8.781825418498213e-07, "loss": 0.0001, "step": 202320 }, { "epoch": 3.3106438681174835, "grad_norm": 0.03951627388596535, "learning_rate": 8.777784431205799e-07, "loss": 0.0002, "step": 202330 }, { "epoch": 3.3108074940685595, "grad_norm": 0.00893163401633501, "learning_rate": 8.773744284399288e-07, "loss": 0.0003, "step": 202340 }, { "epoch": 3.310971120019635, "grad_norm": 0.0015337415970861912, "learning_rate": 8.769704978161087e-07, "loss": 0.0002, "step": 202350 }, { "epoch": 3.311134745970711, "grad_norm": 0.018341049551963806, "learning_rate": 8.765666512573523e-07, "loss": 0.0003, "step": 202360 }, { "epoch": 3.3112983719217866, "grad_norm": 0.00020032766042277217, "learning_rate": 8.761628887718948e-07, "loss": 0.0002, "step": 202370 }, { "epoch": 3.3114619978728626, "grad_norm": 0.0008756174938753247, "learning_rate": 8.757592103679685e-07, "loss": 0.0004, "step": 202380 }, { "epoch": 3.3116256238239385, "grad_norm": 0.043243035674095154, "learning_rate": 8.753556160538062e-07, "loss": 0.0004, "step": 202390 }, { "epoch": 3.311789249775014, "grad_norm": 0.34350547194480896, "learning_rate": 8.749521058376336e-07, "loss": 0.0001, "step": 202400 }, { "epoch": 3.31195287572609, "grad_norm": 0.024540042504668236, "learning_rate": 8.745486797276792e-07, "loss": 0.0004, "step": 202410 }, { "epoch": 3.312116501677166, "grad_norm": 0.1425616443157196, "learning_rate": 8.741453377321685e-07, "loss": 0.0002, "step": 202420 }, { "epoch": 3.3122801276282416, "grad_norm": 0.0023607127368450165, "learning_rate": 8.737420798593272e-07, "loss": 0.0002, "step": 202430 }, { "epoch": 3.3124437535793176, "grad_norm": 0.073853999376297, "learning_rate": 8.733389061173742e-07, "loss": 0.0004, "step": 202440 }, { "epoch": 3.3126073795303936, "grad_norm": 0.018828244879841805, "learning_rate": 8.729358165145313e-07, "loss": 0.0001, "step": 202450 }, { "epoch": 3.312771005481469, "grad_norm": 0.048812881112098694, "learning_rate": 8.72532811059017e-07, "loss": 0.0003, "step": 202460 }, { "epoch": 3.312934631432545, "grad_norm": 0.15299133956432343, "learning_rate": 8.721298897590497e-07, "loss": 0.0003, "step": 202470 }, { "epoch": 3.313098257383621, "grad_norm": 0.0002699453616514802, "learning_rate": 8.717270526228422e-07, "loss": 0.0004, "step": 202480 }, { "epoch": 3.3132618833346967, "grad_norm": 0.008739093318581581, "learning_rate": 8.713242996586085e-07, "loss": 0.0002, "step": 202490 }, { "epoch": 3.3134255092857727, "grad_norm": 0.0007828496163710952, "learning_rate": 8.709216308745616e-07, "loss": 0.0002, "step": 202500 }, { "epoch": 3.3135891352368487, "grad_norm": 0.0018548958469182253, "learning_rate": 8.705190462789126e-07, "loss": 0.0005, "step": 202510 }, { "epoch": 3.3137527611879243, "grad_norm": 0.004977921023964882, "learning_rate": 8.701165458798666e-07, "loss": 0.0003, "step": 202520 }, { "epoch": 3.3139163871390003, "grad_norm": 0.015326968394219875, "learning_rate": 8.697141296856315e-07, "loss": 0.0003, "step": 202530 }, { "epoch": 3.3140800130900763, "grad_norm": 0.02327699400484562, "learning_rate": 8.69311797704413e-07, "loss": 0.0004, "step": 202540 }, { "epoch": 3.314243639041152, "grad_norm": 0.0005428699078038335, "learning_rate": 8.68909549944415e-07, "loss": 0.0002, "step": 202550 }, { "epoch": 3.314407264992228, "grad_norm": 0.042291104793548584, "learning_rate": 8.685073864138371e-07, "loss": 0.0002, "step": 202560 }, { "epoch": 3.314570890943304, "grad_norm": 0.0026172492653131485, "learning_rate": 8.681053071208795e-07, "loss": 0.0004, "step": 202570 }, { "epoch": 3.3147345168943794, "grad_norm": 0.02659497782588005, "learning_rate": 8.677033120737405e-07, "loss": 0.0001, "step": 202580 }, { "epoch": 3.3148981428454554, "grad_norm": 0.003512175055220723, "learning_rate": 8.67301401280618e-07, "loss": 0.0003, "step": 202590 }, { "epoch": 3.3150617687965314, "grad_norm": 0.0011921070981770754, "learning_rate": 8.668995747497045e-07, "loss": 0.0003, "step": 202600 }, { "epoch": 3.315225394747607, "grad_norm": 0.00169904506765306, "learning_rate": 8.664978324891926e-07, "loss": 0.0006, "step": 202610 }, { "epoch": 3.315389020698683, "grad_norm": 0.001052405685186386, "learning_rate": 8.660961745072755e-07, "loss": 0.0001, "step": 202620 }, { "epoch": 3.3155526466497585, "grad_norm": 0.012421795167028904, "learning_rate": 8.65694600812142e-07, "loss": 0.0001, "step": 202630 }, { "epoch": 3.3157162726008345, "grad_norm": 0.023768432438373566, "learning_rate": 8.652931114119789e-07, "loss": 0.0004, "step": 202640 }, { "epoch": 3.3158798985519105, "grad_norm": 0.02712174504995346, "learning_rate": 8.648917063149725e-07, "loss": 0.0002, "step": 202650 }, { "epoch": 3.316043524502986, "grad_norm": 0.02970075234770775, "learning_rate": 8.644903855293074e-07, "loss": 0.0004, "step": 202660 }, { "epoch": 3.316207150454062, "grad_norm": 0.03478776663541794, "learning_rate": 8.640891490631675e-07, "loss": 0.0002, "step": 202670 }, { "epoch": 3.316370776405138, "grad_norm": 0.05644052103161812, "learning_rate": 8.636879969247308e-07, "loss": 0.0007, "step": 202680 }, { "epoch": 3.3165344023562136, "grad_norm": 0.01074935868382454, "learning_rate": 8.632869291221784e-07, "loss": 0.0002, "step": 202690 }, { "epoch": 3.3166980283072895, "grad_norm": 0.0003173806762788445, "learning_rate": 8.628859456636873e-07, "loss": 0.0002, "step": 202700 }, { "epoch": 3.3168616542583655, "grad_norm": 0.0006027915515005589, "learning_rate": 8.624850465574347e-07, "loss": 0.0003, "step": 202710 }, { "epoch": 3.317025280209441, "grad_norm": 0.00040583027293905616, "learning_rate": 8.620842318115919e-07, "loss": 0.0003, "step": 202720 }, { "epoch": 3.317188906160517, "grad_norm": 0.02736521139740944, "learning_rate": 8.616835014343323e-07, "loss": 0.0002, "step": 202730 }, { "epoch": 3.317352532111593, "grad_norm": 0.012715638615190983, "learning_rate": 8.612828554338265e-07, "loss": 0.0002, "step": 202740 }, { "epoch": 3.3175161580626686, "grad_norm": 0.13055938482284546, "learning_rate": 8.608822938182449e-07, "loss": 0.0011, "step": 202750 }, { "epoch": 3.3176797840137446, "grad_norm": 0.0015963517362251878, "learning_rate": 8.604818165957518e-07, "loss": 0.0002, "step": 202760 }, { "epoch": 3.31784340996482, "grad_norm": 0.014566909521818161, "learning_rate": 8.600814237745137e-07, "loss": 0.0003, "step": 202770 }, { "epoch": 3.318007035915896, "grad_norm": 0.0234698299318552, "learning_rate": 8.59681115362695e-07, "loss": 0.0001, "step": 202780 }, { "epoch": 3.318170661866972, "grad_norm": 0.03303258493542671, "learning_rate": 8.592808913684581e-07, "loss": 0.0005, "step": 202790 }, { "epoch": 3.3183342878180477, "grad_norm": 0.08987531065940857, "learning_rate": 8.588807517999609e-07, "loss": 0.0002, "step": 202800 }, { "epoch": 3.3184979137691237, "grad_norm": 0.09138669818639755, "learning_rate": 8.584806966653647e-07, "loss": 0.0002, "step": 202810 }, { "epoch": 3.3186615397201997, "grad_norm": 0.02509045973420143, "learning_rate": 8.580807259728225e-07, "loss": 0.0002, "step": 202820 }, { "epoch": 3.3188251656712753, "grad_norm": 0.032665103673934937, "learning_rate": 8.576808397304936e-07, "loss": 0.0001, "step": 202830 }, { "epoch": 3.3189887916223513, "grad_norm": 0.018011903390288353, "learning_rate": 8.572810379465291e-07, "loss": 0.0008, "step": 202840 }, { "epoch": 3.3191524175734273, "grad_norm": 0.039010658860206604, "learning_rate": 8.568813206290816e-07, "loss": 0.0002, "step": 202850 }, { "epoch": 3.319316043524503, "grad_norm": 0.004908367060124874, "learning_rate": 8.564816877862981e-07, "loss": 0.0004, "step": 202860 }, { "epoch": 3.319479669475579, "grad_norm": 0.10353599488735199, "learning_rate": 8.560821394263313e-07, "loss": 0.0003, "step": 202870 }, { "epoch": 3.319643295426655, "grad_norm": 0.0021155867725610733, "learning_rate": 8.556826755573245e-07, "loss": 0.0001, "step": 202880 }, { "epoch": 3.3198069213777304, "grad_norm": 0.07538340240716934, "learning_rate": 8.552832961874241e-07, "loss": 0.0003, "step": 202890 }, { "epoch": 3.3199705473288064, "grad_norm": 0.0007040809723548591, "learning_rate": 8.548840013247706e-07, "loss": 0.0003, "step": 202900 }, { "epoch": 3.3201341732798824, "grad_norm": 0.037800271064043045, "learning_rate": 8.544847909775089e-07, "loss": 0.0001, "step": 202910 }, { "epoch": 3.320297799230958, "grad_norm": 0.03309149295091629, "learning_rate": 8.540856651537754e-07, "loss": 0.0003, "step": 202920 }, { "epoch": 3.320461425182034, "grad_norm": 0.002920881612226367, "learning_rate": 8.536866238617109e-07, "loss": 0.0002, "step": 202930 }, { "epoch": 3.32062505113311, "grad_norm": 2.451630098221358e-05, "learning_rate": 8.53287667109447e-07, "loss": 0.0002, "step": 202940 }, { "epoch": 3.3207886770841855, "grad_norm": 0.04429883137345314, "learning_rate": 8.528887949051234e-07, "loss": 0.0002, "step": 202950 }, { "epoch": 3.3209523030352615, "grad_norm": 0.02681531012058258, "learning_rate": 8.524900072568693e-07, "loss": 0.0002, "step": 202960 }, { "epoch": 3.3211159289863375, "grad_norm": 0.0018928565550595522, "learning_rate": 8.520913041728173e-07, "loss": 0.0005, "step": 202970 }, { "epoch": 3.321279554937413, "grad_norm": 0.130858913064003, "learning_rate": 8.51692685661094e-07, "loss": 0.0003, "step": 202980 }, { "epoch": 3.321443180888489, "grad_norm": 0.03358824551105499, "learning_rate": 8.512941517298312e-07, "loss": 0.0004, "step": 202990 }, { "epoch": 3.321606806839565, "grad_norm": 0.12532071769237518, "learning_rate": 8.508957023871512e-07, "loss": 0.0003, "step": 203000 }, { "epoch": 3.3217704327906405, "grad_norm": 0.08069305866956711, "learning_rate": 8.504973376411801e-07, "loss": 0.0004, "step": 203010 }, { "epoch": 3.3219340587417165, "grad_norm": 0.09346172213554382, "learning_rate": 8.500990575000384e-07, "loss": 0.0004, "step": 203020 }, { "epoch": 3.322097684692792, "grad_norm": 0.06194024160504341, "learning_rate": 8.497008619718472e-07, "loss": 0.0004, "step": 203030 }, { "epoch": 3.322261310643868, "grad_norm": 0.0012917992426082492, "learning_rate": 8.493027510647256e-07, "loss": 0.0002, "step": 203040 }, { "epoch": 3.322424936594944, "grad_norm": 0.015523037873208523, "learning_rate": 8.489047247867921e-07, "loss": 0.0002, "step": 203050 }, { "epoch": 3.3225885625460196, "grad_norm": 0.06034570559859276, "learning_rate": 8.485067831461602e-07, "loss": 0.0004, "step": 203060 }, { "epoch": 3.3227521884970956, "grad_norm": 0.009892752394080162, "learning_rate": 8.48108926150944e-07, "loss": 0.0002, "step": 203070 }, { "epoch": 3.3229158144481716, "grad_norm": 0.05840574577450752, "learning_rate": 8.477111538092552e-07, "loss": 0.0008, "step": 203080 }, { "epoch": 3.323079440399247, "grad_norm": 0.009326769039034843, "learning_rate": 8.473134661292065e-07, "loss": 0.0002, "step": 203090 }, { "epoch": 3.323243066350323, "grad_norm": 0.06417401880025864, "learning_rate": 8.469158631189028e-07, "loss": 0.0002, "step": 203100 }, { "epoch": 3.323406692301399, "grad_norm": 0.024791864678263664, "learning_rate": 8.465183447864534e-07, "loss": 0.0002, "step": 203110 }, { "epoch": 3.3235703182524747, "grad_norm": 0.012108663097023964, "learning_rate": 8.461209111399615e-07, "loss": 0.0001, "step": 203120 }, { "epoch": 3.3237339442035507, "grad_norm": 0.12368850409984589, "learning_rate": 8.457235621875331e-07, "loss": 0.0005, "step": 203130 }, { "epoch": 3.3238975701546263, "grad_norm": 0.10710987448692322, "learning_rate": 8.453262979372672e-07, "loss": 0.0002, "step": 203140 }, { "epoch": 3.3240611961057023, "grad_norm": 0.05719543993473053, "learning_rate": 8.449291183972646e-07, "loss": 0.0004, "step": 203150 }, { "epoch": 3.3242248220567783, "grad_norm": 0.03208883851766586, "learning_rate": 8.445320235756233e-07, "loss": 0.0003, "step": 203160 }, { "epoch": 3.324388448007854, "grad_norm": 0.01248969230800867, "learning_rate": 8.441350134804411e-07, "loss": 0.0001, "step": 203170 }, { "epoch": 3.32455207395893, "grad_norm": 0.08495111763477325, "learning_rate": 8.437380881198104e-07, "loss": 0.0004, "step": 203180 }, { "epoch": 3.324715699910006, "grad_norm": 0.02527693100273609, "learning_rate": 8.433412475018255e-07, "loss": 0.0004, "step": 203190 }, { "epoch": 3.3248793258610814, "grad_norm": 0.026038628071546555, "learning_rate": 8.429444916345774e-07, "loss": 0.0003, "step": 203200 }, { "epoch": 3.3250429518121574, "grad_norm": 0.0021028476767241955, "learning_rate": 8.425478205261567e-07, "loss": 0.0001, "step": 203210 }, { "epoch": 3.3252065777632334, "grad_norm": 0.013275831937789917, "learning_rate": 8.421512341846494e-07, "loss": 0.0002, "step": 203220 }, { "epoch": 3.325370203714309, "grad_norm": 0.0005594792310148478, "learning_rate": 8.417547326181419e-07, "loss": 0.0006, "step": 203230 }, { "epoch": 3.325533829665385, "grad_norm": 0.00024640592164359987, "learning_rate": 8.413583158347189e-07, "loss": 0.0006, "step": 203240 }, { "epoch": 3.325697455616461, "grad_norm": 0.005752052646130323, "learning_rate": 8.409619838424643e-07, "loss": 0.0002, "step": 203250 }, { "epoch": 3.3258610815675365, "grad_norm": 0.0013666526647284627, "learning_rate": 8.405657366494563e-07, "loss": 0.0005, "step": 203260 }, { "epoch": 3.3260247075186125, "grad_norm": 0.00446021743118763, "learning_rate": 8.401695742637761e-07, "loss": 0.0003, "step": 203270 }, { "epoch": 3.3261883334696885, "grad_norm": 0.006009560544043779, "learning_rate": 8.397734966935001e-07, "loss": 0.0002, "step": 203280 }, { "epoch": 3.326351959420764, "grad_norm": 0.01155808661133051, "learning_rate": 8.393775039467056e-07, "loss": 0.0002, "step": 203290 }, { "epoch": 3.32651558537184, "grad_norm": 0.0010995297925546765, "learning_rate": 8.38981596031464e-07, "loss": 0.0002, "step": 203300 }, { "epoch": 3.326679211322916, "grad_norm": 0.008991382084786892, "learning_rate": 8.385857729558489e-07, "loss": 0.0001, "step": 203310 }, { "epoch": 3.3268428372739915, "grad_norm": 0.45717480778694153, "learning_rate": 8.381900347279304e-07, "loss": 0.0003, "step": 203320 }, { "epoch": 3.3270064632250675, "grad_norm": 0.0065412321127951145, "learning_rate": 8.377943813557792e-07, "loss": 0.0002, "step": 203330 }, { "epoch": 3.3271700891761435, "grad_norm": 0.023329250514507294, "learning_rate": 8.373988128474597e-07, "loss": 0.0002, "step": 203340 }, { "epoch": 3.327333715127219, "grad_norm": 0.0021751727908849716, "learning_rate": 8.37003329211038e-07, "loss": 0.0004, "step": 203350 }, { "epoch": 3.327497341078295, "grad_norm": 0.0006074716220609844, "learning_rate": 8.366079304545776e-07, "loss": 0.0003, "step": 203360 }, { "epoch": 3.327660967029371, "grad_norm": 0.04994983598589897, "learning_rate": 8.362126165861423e-07, "loss": 0.0004, "step": 203370 }, { "epoch": 3.3278245929804466, "grad_norm": 0.0019546453841030598, "learning_rate": 8.358173876137893e-07, "loss": 0.0004, "step": 203380 }, { "epoch": 3.3279882189315226, "grad_norm": 0.01867240108549595, "learning_rate": 8.354222435455784e-07, "loss": 0.0003, "step": 203390 }, { "epoch": 3.328151844882598, "grad_norm": 0.0750032439827919, "learning_rate": 8.350271843895658e-07, "loss": 0.0003, "step": 203400 }, { "epoch": 3.328315470833674, "grad_norm": 0.012885499745607376, "learning_rate": 8.346322101538079e-07, "loss": 0.0002, "step": 203410 }, { "epoch": 3.32847909678475, "grad_norm": 0.0007292600348591805, "learning_rate": 8.342373208463556e-07, "loss": 0.0002, "step": 203420 }, { "epoch": 3.3286427227358257, "grad_norm": 0.11240704357624054, "learning_rate": 8.338425164752617e-07, "loss": 0.0008, "step": 203430 }, { "epoch": 3.3288063486869017, "grad_norm": 0.0004033622390124947, "learning_rate": 8.334477970485755e-07, "loss": 0.0002, "step": 203440 }, { "epoch": 3.3289699746379777, "grad_norm": 0.02359806001186371, "learning_rate": 8.330531625743465e-07, "loss": 0.0002, "step": 203450 }, { "epoch": 3.3291336005890533, "grad_norm": 0.017358070239424706, "learning_rate": 8.326586130606184e-07, "loss": 0.0001, "step": 203460 }, { "epoch": 3.3292972265401293, "grad_norm": 0.01982010528445244, "learning_rate": 8.322641485154381e-07, "loss": 0.0001, "step": 203470 }, { "epoch": 3.3294608524912053, "grad_norm": 0.046113334596157074, "learning_rate": 8.318697689468447e-07, "loss": 0.0004, "step": 203480 }, { "epoch": 3.329624478442281, "grad_norm": 0.006535796448588371, "learning_rate": 8.31475474362885e-07, "loss": 0.0002, "step": 203490 }, { "epoch": 3.329788104393357, "grad_norm": 0.101967953145504, "learning_rate": 8.310812647715932e-07, "loss": 0.0005, "step": 203500 }, { "epoch": 3.329951730344433, "grad_norm": 0.0106681939214468, "learning_rate": 8.306871401810102e-07, "loss": 0.0004, "step": 203510 }, { "epoch": 3.3301153562955084, "grad_norm": 0.0013628732413053513, "learning_rate": 8.302931005991682e-07, "loss": 0.0003, "step": 203520 }, { "epoch": 3.3302789822465844, "grad_norm": 0.003119498025625944, "learning_rate": 8.298991460341066e-07, "loss": 0.001, "step": 203530 }, { "epoch": 3.33044260819766, "grad_norm": 0.0014949181349948049, "learning_rate": 8.295052764938528e-07, "loss": 0.0001, "step": 203540 }, { "epoch": 3.330606234148736, "grad_norm": 0.06268849223852158, "learning_rate": 8.291114919864412e-07, "loss": 0.0005, "step": 203550 }, { "epoch": 3.330769860099812, "grad_norm": 0.07223843783140182, "learning_rate": 8.287177925198969e-07, "loss": 0.0004, "step": 203560 }, { "epoch": 3.3309334860508875, "grad_norm": 0.018396155908703804, "learning_rate": 8.283241781022511e-07, "loss": 0.0002, "step": 203570 }, { "epoch": 3.3310971120019635, "grad_norm": 0.045229408890008926, "learning_rate": 8.279306487415267e-07, "loss": 0.0004, "step": 203580 }, { "epoch": 3.3312607379530395, "grad_norm": 0.00010038473556051031, "learning_rate": 8.275372044457492e-07, "loss": 0.0002, "step": 203590 }, { "epoch": 3.331424363904115, "grad_norm": 0.07340171933174133, "learning_rate": 8.271438452229369e-07, "loss": 0.0003, "step": 203600 }, { "epoch": 3.331587989855191, "grad_norm": 0.005945628043264151, "learning_rate": 8.267505710811152e-07, "loss": 0.0003, "step": 203610 }, { "epoch": 3.331751615806267, "grad_norm": 0.07030507177114487, "learning_rate": 8.263573820282994e-07, "loss": 0.0002, "step": 203620 }, { "epoch": 3.3319152417573425, "grad_norm": 0.008563622832298279, "learning_rate": 8.259642780725074e-07, "loss": 0.0002, "step": 203630 }, { "epoch": 3.3320788677084185, "grad_norm": 0.0015222758520394564, "learning_rate": 8.255712592217536e-07, "loss": 0.0002, "step": 203640 }, { "epoch": 3.3322424936594945, "grad_norm": 0.012622588314116001, "learning_rate": 8.251783254840512e-07, "loss": 0.0002, "step": 203650 }, { "epoch": 3.33240611961057, "grad_norm": 0.013545242138206959, "learning_rate": 8.247854768674119e-07, "loss": 0.0001, "step": 203660 }, { "epoch": 3.332569745561646, "grad_norm": 0.06328161060810089, "learning_rate": 8.243927133798474e-07, "loss": 0.0002, "step": 203670 }, { "epoch": 3.332733371512722, "grad_norm": 0.04303823783993721, "learning_rate": 8.240000350293625e-07, "loss": 0.0004, "step": 203680 }, { "epoch": 3.3328969974637976, "grad_norm": 0.05590217933058739, "learning_rate": 8.236074418239659e-07, "loss": 0.0003, "step": 203690 }, { "epoch": 3.3330606234148736, "grad_norm": 0.09189247339963913, "learning_rate": 8.232149337716616e-07, "loss": 0.0003, "step": 203700 }, { "epoch": 3.3332242493659496, "grad_norm": 0.011259629391133785, "learning_rate": 8.228225108804538e-07, "loss": 0.0001, "step": 203710 }, { "epoch": 3.333387875317025, "grad_norm": 0.0031062248162925243, "learning_rate": 8.224301731583412e-07, "loss": 0.0004, "step": 203720 }, { "epoch": 3.333551501268101, "grad_norm": 0.007977353408932686, "learning_rate": 8.220379206133244e-07, "loss": 0.0001, "step": 203730 }, { "epoch": 3.333715127219177, "grad_norm": 0.003019626485183835, "learning_rate": 8.216457532534011e-07, "loss": 0.0003, "step": 203740 }, { "epoch": 3.3338787531702527, "grad_norm": 0.014702609740197659, "learning_rate": 8.212536710865687e-07, "loss": 0.0001, "step": 203750 }, { "epoch": 3.3340423791213287, "grad_norm": 0.0330062173306942, "learning_rate": 8.208616741208186e-07, "loss": 0.0004, "step": 203760 }, { "epoch": 3.3342060050724047, "grad_norm": 0.002240107860416174, "learning_rate": 8.20469762364145e-07, "loss": 0.0003, "step": 203770 }, { "epoch": 3.3343696310234803, "grad_norm": 0.004283652175217867, "learning_rate": 8.200779358245381e-07, "loss": 0.0005, "step": 203780 }, { "epoch": 3.3345332569745563, "grad_norm": 0.0006080136517994106, "learning_rate": 8.196861945099882e-07, "loss": 0.0004, "step": 203790 }, { "epoch": 3.334696882925632, "grad_norm": 0.013583126477897167, "learning_rate": 8.192945384284812e-07, "loss": 0.0003, "step": 203800 }, { "epoch": 3.334860508876708, "grad_norm": 0.0016862970078364015, "learning_rate": 8.189029675880023e-07, "loss": 0.0003, "step": 203810 }, { "epoch": 3.335024134827784, "grad_norm": 0.002446940867230296, "learning_rate": 8.185114819965362e-07, "loss": 0.0001, "step": 203820 }, { "epoch": 3.3351877607788594, "grad_norm": 0.020461535081267357, "learning_rate": 8.18120081662066e-07, "loss": 0.0004, "step": 203830 }, { "epoch": 3.3353513867299354, "grad_norm": 0.043227825313806534, "learning_rate": 8.177287665925693e-07, "loss": 0.0004, "step": 203840 }, { "epoch": 3.3355150126810114, "grad_norm": 0.06740036606788635, "learning_rate": 8.173375367960267e-07, "loss": 0.0007, "step": 203850 }, { "epoch": 3.335678638632087, "grad_norm": 0.014182789251208305, "learning_rate": 8.169463922804139e-07, "loss": 0.0002, "step": 203860 }, { "epoch": 3.335842264583163, "grad_norm": 0.013481165282428265, "learning_rate": 8.165553330537085e-07, "loss": 0.0001, "step": 203870 }, { "epoch": 3.336005890534239, "grad_norm": 0.022430701181292534, "learning_rate": 8.1616435912388e-07, "loss": 0.0001, "step": 203880 }, { "epoch": 3.3361695164853145, "grad_norm": 0.0011415258049964905, "learning_rate": 8.157734704989023e-07, "loss": 0.0003, "step": 203890 }, { "epoch": 3.3363331424363905, "grad_norm": 0.015211210586130619, "learning_rate": 8.153826671867454e-07, "loss": 0.0005, "step": 203900 }, { "epoch": 3.336496768387466, "grad_norm": 0.021819042041897774, "learning_rate": 8.149919491953778e-07, "loss": 0.0003, "step": 203910 }, { "epoch": 3.336660394338542, "grad_norm": 0.07330044358968735, "learning_rate": 8.146013165327637e-07, "loss": 0.0006, "step": 203920 }, { "epoch": 3.336824020289618, "grad_norm": 0.08899962902069092, "learning_rate": 8.142107692068696e-07, "loss": 0.0004, "step": 203930 }, { "epoch": 3.3369876462406936, "grad_norm": 0.015519808977842331, "learning_rate": 8.138203072256579e-07, "loss": 0.0002, "step": 203940 }, { "epoch": 3.3371512721917695, "grad_norm": 0.0173316840082407, "learning_rate": 8.134299305970911e-07, "loss": 0.0002, "step": 203950 }, { "epoch": 3.3373148981428455, "grad_norm": 0.05760210379958153, "learning_rate": 8.130396393291262e-07, "loss": 0.0003, "step": 203960 }, { "epoch": 3.337478524093921, "grad_norm": 0.05747615173459053, "learning_rate": 8.126494334297225e-07, "loss": 0.0008, "step": 203970 }, { "epoch": 3.337642150044997, "grad_norm": 0.010798947885632515, "learning_rate": 8.122593129068351e-07, "loss": 0.0002, "step": 203980 }, { "epoch": 3.337805775996073, "grad_norm": 0.04713267460465431, "learning_rate": 8.118692777684206e-07, "loss": 0.0002, "step": 203990 }, { "epoch": 3.3379694019471486, "grad_norm": 0.028967831283807755, "learning_rate": 8.114793280224281e-07, "loss": 0.0002, "step": 204000 }, { "epoch": 3.3381330278982246, "grad_norm": 0.0004504612006712705, "learning_rate": 8.110894636768097e-07, "loss": 0.0003, "step": 204010 }, { "epoch": 3.3382966538493006, "grad_norm": 0.0025459493044763803, "learning_rate": 8.106996847395149e-07, "loss": 0.0001, "step": 204020 }, { "epoch": 3.338460279800376, "grad_norm": 0.011475452221930027, "learning_rate": 8.10309991218492e-07, "loss": 0.0001, "step": 204030 }, { "epoch": 3.338623905751452, "grad_norm": 0.06562215089797974, "learning_rate": 8.099203831216834e-07, "loss": 0.0005, "step": 204040 }, { "epoch": 3.338787531702528, "grad_norm": 0.0018113335827365518, "learning_rate": 8.095308604570351e-07, "loss": 0.0001, "step": 204050 }, { "epoch": 3.3389511576536037, "grad_norm": 0.0024869435001164675, "learning_rate": 8.091414232324884e-07, "loss": 0.0001, "step": 204060 }, { "epoch": 3.3391147836046797, "grad_norm": 0.0159308984875679, "learning_rate": 8.087520714559855e-07, "loss": 0.0002, "step": 204070 }, { "epoch": 3.3392784095557557, "grad_norm": 0.026490766555070877, "learning_rate": 8.08362805135462e-07, "loss": 0.0004, "step": 204080 }, { "epoch": 3.3394420355068313, "grad_norm": 0.2183283567428589, "learning_rate": 8.07973624278856e-07, "loss": 0.0003, "step": 204090 }, { "epoch": 3.3396056614579073, "grad_norm": 0.015190055593848228, "learning_rate": 8.075845288941026e-07, "loss": 0.0003, "step": 204100 }, { "epoch": 3.3397692874089833, "grad_norm": 0.023826370015740395, "learning_rate": 8.071955189891362e-07, "loss": 0.0006, "step": 204110 }, { "epoch": 3.339932913360059, "grad_norm": 0.03522494062781334, "learning_rate": 8.068065945718862e-07, "loss": 0.0002, "step": 204120 }, { "epoch": 3.340096539311135, "grad_norm": 0.0334312841296196, "learning_rate": 8.064177556502845e-07, "loss": 0.0004, "step": 204130 }, { "epoch": 3.340260165262211, "grad_norm": 0.010153097100555897, "learning_rate": 8.060290022322565e-07, "loss": 0.0002, "step": 204140 }, { "epoch": 3.3404237912132864, "grad_norm": 0.016865763813257217, "learning_rate": 8.056403343257319e-07, "loss": 0.0002, "step": 204150 }, { "epoch": 3.3405874171643624, "grad_norm": 0.0009050791268236935, "learning_rate": 8.052517519386332e-07, "loss": 0.0006, "step": 204160 }, { "epoch": 3.3407510431154384, "grad_norm": 0.007533211726695299, "learning_rate": 8.048632550788848e-07, "loss": 0.0003, "step": 204170 }, { "epoch": 3.340914669066514, "grad_norm": 0.02825973369181156, "learning_rate": 8.044748437544042e-07, "loss": 0.0004, "step": 204180 }, { "epoch": 3.34107829501759, "grad_norm": 0.01871383748948574, "learning_rate": 8.040865179731156e-07, "loss": 0.0006, "step": 204190 }, { "epoch": 3.3412419209686655, "grad_norm": 0.04449094086885452, "learning_rate": 8.036982777429336e-07, "loss": 0.0003, "step": 204200 }, { "epoch": 3.3414055469197415, "grad_norm": 0.045510321855545044, "learning_rate": 8.033101230717761e-07, "loss": 0.0003, "step": 204210 }, { "epoch": 3.3415691728708175, "grad_norm": 0.026647275313735008, "learning_rate": 8.029220539675531e-07, "loss": 0.0003, "step": 204220 }, { "epoch": 3.341732798821893, "grad_norm": 0.023884965106844902, "learning_rate": 8.025340704381829e-07, "loss": 0.0002, "step": 204230 }, { "epoch": 3.341896424772969, "grad_norm": 0.009261497296392918, "learning_rate": 8.021461724915719e-07, "loss": 0.0004, "step": 204240 }, { "epoch": 3.342060050724045, "grad_norm": 0.00348899164237082, "learning_rate": 8.017583601356316e-07, "loss": 0.0002, "step": 204250 }, { "epoch": 3.3422236766751205, "grad_norm": 0.003502336796373129, "learning_rate": 8.013706333782651e-07, "loss": 0.0004, "step": 204260 }, { "epoch": 3.3423873026261965, "grad_norm": 0.0006771512562409043, "learning_rate": 8.009829922273832e-07, "loss": 0.0004, "step": 204270 }, { "epoch": 3.3425509285772725, "grad_norm": 0.003569534048438072, "learning_rate": 8.005954366908853e-07, "loss": 0.0001, "step": 204280 }, { "epoch": 3.342714554528348, "grad_norm": 0.04940301179885864, "learning_rate": 8.002079667766766e-07, "loss": 0.0004, "step": 204290 }, { "epoch": 3.342878180479424, "grad_norm": 0.018029138445854187, "learning_rate": 7.998205824926547e-07, "loss": 0.0002, "step": 204300 }, { "epoch": 3.3430418064304996, "grad_norm": 0.04466418921947479, "learning_rate": 7.994332838467183e-07, "loss": 0.0003, "step": 204310 }, { "epoch": 3.3432054323815756, "grad_norm": 0.11054480075836182, "learning_rate": 7.990460708467656e-07, "loss": 0.0003, "step": 204320 }, { "epoch": 3.3433690583326516, "grad_norm": 0.004314032848924398, "learning_rate": 7.986589435006913e-07, "loss": 0.0004, "step": 204330 }, { "epoch": 3.343532684283727, "grad_norm": 0.05016011744737625, "learning_rate": 7.982719018163871e-07, "loss": 0.0003, "step": 204340 }, { "epoch": 3.343696310234803, "grad_norm": 0.15016497671604156, "learning_rate": 7.978849458017457e-07, "loss": 0.0007, "step": 204350 }, { "epoch": 3.343859936185879, "grad_norm": 0.013560372404754162, "learning_rate": 7.974980754646561e-07, "loss": 0.0004, "step": 204360 }, { "epoch": 3.3440235621369547, "grad_norm": 0.00998004898428917, "learning_rate": 7.971112908130075e-07, "loss": 0.0002, "step": 204370 }, { "epoch": 3.3441871880880307, "grad_norm": 0.023478832095861435, "learning_rate": 7.967245918546845e-07, "loss": 0.0002, "step": 204380 }, { "epoch": 3.3443508140391067, "grad_norm": 0.007993555627763271, "learning_rate": 7.963379785975722e-07, "loss": 0.0001, "step": 204390 }, { "epoch": 3.3445144399901823, "grad_norm": 0.003026796504855156, "learning_rate": 7.959514510495536e-07, "loss": 0.0003, "step": 204400 }, { "epoch": 3.3446780659412583, "grad_norm": 0.0076701040379703045, "learning_rate": 7.955650092185108e-07, "loss": 0.0004, "step": 204410 }, { "epoch": 3.3448416918923343, "grad_norm": 0.02094223164021969, "learning_rate": 7.951786531123202e-07, "loss": 0.0004, "step": 204420 }, { "epoch": 3.34500531784341, "grad_norm": 0.001616873312741518, "learning_rate": 7.947923827388609e-07, "loss": 0.0004, "step": 204430 }, { "epoch": 3.345168943794486, "grad_norm": 0.0014056793879717588, "learning_rate": 7.944061981060086e-07, "loss": 0.0002, "step": 204440 }, { "epoch": 3.345332569745562, "grad_norm": 0.010937981307506561, "learning_rate": 7.940200992216379e-07, "loss": 0.0004, "step": 204450 }, { "epoch": 3.3454961956966374, "grad_norm": 0.004922199063003063, "learning_rate": 7.936340860936198e-07, "loss": 0.0005, "step": 204460 }, { "epoch": 3.3456598216477134, "grad_norm": 0.04099202901124954, "learning_rate": 7.932481587298246e-07, "loss": 0.0003, "step": 204470 }, { "epoch": 3.3458234475987894, "grad_norm": 0.0013156785862520337, "learning_rate": 7.928623171381222e-07, "loss": 0.0002, "step": 204480 }, { "epoch": 3.345987073549865, "grad_norm": 0.0007602601544931531, "learning_rate": 7.924765613263808e-07, "loss": 0.0002, "step": 204490 }, { "epoch": 3.346150699500941, "grad_norm": 0.09574282914400101, "learning_rate": 7.920908913024616e-07, "loss": 0.0005, "step": 204500 }, { "epoch": 3.346314325452017, "grad_norm": 0.04324238374829292, "learning_rate": 7.917053070742314e-07, "loss": 0.0003, "step": 204510 }, { "epoch": 3.3464779514030925, "grad_norm": 0.004542191978543997, "learning_rate": 7.913198086495505e-07, "loss": 0.0002, "step": 204520 }, { "epoch": 3.3466415773541685, "grad_norm": 0.07447001338005066, "learning_rate": 7.909343960362808e-07, "loss": 0.0005, "step": 204530 }, { "epoch": 3.3468052033052444, "grad_norm": 0.003889739979058504, "learning_rate": 7.905490692422774e-07, "loss": 0.0, "step": 204540 }, { "epoch": 3.34696882925632, "grad_norm": 0.0417555570602417, "learning_rate": 7.901638282753987e-07, "loss": 0.0005, "step": 204550 }, { "epoch": 3.347132455207396, "grad_norm": 0.003906527534127235, "learning_rate": 7.897786731434992e-07, "loss": 0.0003, "step": 204560 }, { "epoch": 3.3472960811584715, "grad_norm": 0.008076038211584091, "learning_rate": 7.893936038544337e-07, "loss": 0.0002, "step": 204570 }, { "epoch": 3.3474597071095475, "grad_norm": 0.038653332740068436, "learning_rate": 7.890086204160497e-07, "loss": 0.0003, "step": 204580 }, { "epoch": 3.3476233330606235, "grad_norm": 0.003028538776561618, "learning_rate": 7.886237228361987e-07, "loss": 0.0003, "step": 204590 }, { "epoch": 3.347786959011699, "grad_norm": 0.0924612283706665, "learning_rate": 7.882389111227284e-07, "loss": 0.0002, "step": 204600 }, { "epoch": 3.347950584962775, "grad_norm": 0.01610282063484192, "learning_rate": 7.878541852834859e-07, "loss": 0.0003, "step": 204610 }, { "epoch": 3.348114210913851, "grad_norm": 0.0074065751396119595, "learning_rate": 7.874695453263131e-07, "loss": 0.0001, "step": 204620 }, { "epoch": 3.3482778368649266, "grad_norm": 0.011377948336303234, "learning_rate": 7.870849912590539e-07, "loss": 0.0004, "step": 204630 }, { "epoch": 3.3484414628160026, "grad_norm": 0.015195793472230434, "learning_rate": 7.867005230895485e-07, "loss": 0.0001, "step": 204640 }, { "epoch": 3.3486050887670786, "grad_norm": 0.05275267735123634, "learning_rate": 7.863161408256375e-07, "loss": 0.0002, "step": 204650 }, { "epoch": 3.348768714718154, "grad_norm": 0.0765700563788414, "learning_rate": 7.859318444751557e-07, "loss": 0.0004, "step": 204660 }, { "epoch": 3.34893234066923, "grad_norm": 0.0029993578791618347, "learning_rate": 7.85547634045939e-07, "loss": 0.0004, "step": 204670 }, { "epoch": 3.3490959666203057, "grad_norm": 0.0009344872669316828, "learning_rate": 7.851635095458226e-07, "loss": 0.0003, "step": 204680 }, { "epoch": 3.3492595925713817, "grad_norm": 0.0006420532008633018, "learning_rate": 7.847794709826384e-07, "loss": 0.0002, "step": 204690 }, { "epoch": 3.3494232185224577, "grad_norm": 0.025779535993933678, "learning_rate": 7.843955183642143e-07, "loss": 0.0004, "step": 204700 }, { "epoch": 3.3495868444735333, "grad_norm": 0.07442481815814972, "learning_rate": 7.840116516983809e-07, "loss": 0.0004, "step": 204710 }, { "epoch": 3.3497504704246093, "grad_norm": 9.63551428867504e-05, "learning_rate": 7.836278709929645e-07, "loss": 0.0002, "step": 204720 }, { "epoch": 3.3499140963756853, "grad_norm": 0.024650439620018005, "learning_rate": 7.832441762557907e-07, "loss": 0.0002, "step": 204730 }, { "epoch": 3.350077722326761, "grad_norm": 0.008329354226589203, "learning_rate": 7.82860567494681e-07, "loss": 0.0001, "step": 204740 }, { "epoch": 3.350241348277837, "grad_norm": 0.05905326083302498, "learning_rate": 7.82477044717459e-07, "loss": 0.0004, "step": 204750 }, { "epoch": 3.350404974228913, "grad_norm": 0.004857782740145922, "learning_rate": 7.820936079319403e-07, "loss": 0.0002, "step": 204760 }, { "epoch": 3.3505686001799884, "grad_norm": 0.0016537508927285671, "learning_rate": 7.817102571459484e-07, "loss": 0.0001, "step": 204770 }, { "epoch": 3.3507322261310644, "grad_norm": 0.0012762381229549646, "learning_rate": 7.813269923672956e-07, "loss": 0.0001, "step": 204780 }, { "epoch": 3.3508958520821404, "grad_norm": 0.016140243038535118, "learning_rate": 7.809438136037989e-07, "loss": 0.0001, "step": 204790 }, { "epoch": 3.351059478033216, "grad_norm": 0.0704498216509819, "learning_rate": 7.805607208632676e-07, "loss": 0.0003, "step": 204800 }, { "epoch": 3.351223103984292, "grad_norm": 0.025340598076581955, "learning_rate": 7.801777141535166e-07, "loss": 0.0002, "step": 204810 }, { "epoch": 3.351386729935368, "grad_norm": 0.003277356969192624, "learning_rate": 7.797947934823519e-07, "loss": 0.0003, "step": 204820 }, { "epoch": 3.3515503558864435, "grad_norm": 0.06431891024112701, "learning_rate": 7.794119588575838e-07, "loss": 0.0004, "step": 204830 }, { "epoch": 3.3517139818375195, "grad_norm": 0.07660531252622604, "learning_rate": 7.79029210287014e-07, "loss": 0.0007, "step": 204840 }, { "epoch": 3.3518776077885954, "grad_norm": 0.0011112312786281109, "learning_rate": 7.786465477784505e-07, "loss": 0.0002, "step": 204850 }, { "epoch": 3.352041233739671, "grad_norm": 0.006716209463775158, "learning_rate": 7.782639713396933e-07, "loss": 0.0002, "step": 204860 }, { "epoch": 3.352204859690747, "grad_norm": 0.014867905527353287, "learning_rate": 7.77881480978544e-07, "loss": 0.0002, "step": 204870 }, { "epoch": 3.352368485641823, "grad_norm": 0.013095684349536896, "learning_rate": 7.774990767027985e-07, "loss": 0.0003, "step": 204880 }, { "epoch": 3.3525321115928985, "grad_norm": 0.00046168308472260833, "learning_rate": 7.771167585202576e-07, "loss": 0.0004, "step": 204890 }, { "epoch": 3.3526957375439745, "grad_norm": 0.0011733691208064556, "learning_rate": 7.767345264387155e-07, "loss": 0.0003, "step": 204900 }, { "epoch": 3.3528593634950505, "grad_norm": 0.04258018359541893, "learning_rate": 7.763523804659628e-07, "loss": 0.0003, "step": 204910 }, { "epoch": 3.353022989446126, "grad_norm": 0.007528218906372786, "learning_rate": 7.75970320609793e-07, "loss": 0.0003, "step": 204920 }, { "epoch": 3.353186615397202, "grad_norm": 0.029978018254041672, "learning_rate": 7.755883468779957e-07, "loss": 0.0002, "step": 204930 }, { "epoch": 3.353350241348278, "grad_norm": 0.0018982978072017431, "learning_rate": 7.752064592783609e-07, "loss": 0.0002, "step": 204940 }, { "epoch": 3.3535138672993536, "grad_norm": 0.01228137407451868, "learning_rate": 7.748246578186719e-07, "loss": 0.0001, "step": 204950 }, { "epoch": 3.3536774932504296, "grad_norm": 0.040701624006032944, "learning_rate": 7.744429425067152e-07, "loss": 0.0001, "step": 204960 }, { "epoch": 3.353841119201505, "grad_norm": 0.02948361448943615, "learning_rate": 7.740613133502728e-07, "loss": 0.0002, "step": 204970 }, { "epoch": 3.354004745152581, "grad_norm": 0.03164420649409294, "learning_rate": 7.73679770357127e-07, "loss": 0.0002, "step": 204980 }, { "epoch": 3.354168371103657, "grad_norm": 0.006027398630976677, "learning_rate": 7.732983135350558e-07, "loss": 0.0002, "step": 204990 }, { "epoch": 3.3543319970547327, "grad_norm": 0.038955800235271454, "learning_rate": 7.729169428918371e-07, "loss": 0.0003, "step": 205000 }, { "epoch": 3.3544956230058087, "grad_norm": 0.009889287874102592, "learning_rate": 7.725356584352467e-07, "loss": 0.0002, "step": 205010 }, { "epoch": 3.3546592489568847, "grad_norm": 0.024875221773982048, "learning_rate": 7.721544601730601e-07, "loss": 0.0002, "step": 205020 }, { "epoch": 3.3548228749079603, "grad_norm": 0.19824135303497314, "learning_rate": 7.717733481130474e-07, "loss": 0.0002, "step": 205030 }, { "epoch": 3.3549865008590363, "grad_norm": 0.00023785160738043487, "learning_rate": 7.713923222629805e-07, "loss": 0.0002, "step": 205040 }, { "epoch": 3.3551501268101123, "grad_norm": 0.0008422060054726899, "learning_rate": 7.71011382630627e-07, "loss": 0.0001, "step": 205050 }, { "epoch": 3.355313752761188, "grad_norm": 0.0317264087498188, "learning_rate": 7.706305292237564e-07, "loss": 0.0006, "step": 205060 }, { "epoch": 3.355477378712264, "grad_norm": 0.06857869029045105, "learning_rate": 7.702497620501315e-07, "loss": 0.0003, "step": 205070 }, { "epoch": 3.3556410046633394, "grad_norm": 0.0514199361205101, "learning_rate": 7.698690811175164e-07, "loss": 0.0004, "step": 205080 }, { "epoch": 3.3558046306144154, "grad_norm": 0.03562597185373306, "learning_rate": 7.694884864336732e-07, "loss": 0.0002, "step": 205090 }, { "epoch": 3.3559682565654914, "grad_norm": 0.008784102275967598, "learning_rate": 7.691079780063631e-07, "loss": 0.0002, "step": 205100 }, { "epoch": 3.356131882516567, "grad_norm": 0.0005806004628539085, "learning_rate": 7.687275558433415e-07, "loss": 0.0003, "step": 205110 }, { "epoch": 3.356295508467643, "grad_norm": 0.051866885274648666, "learning_rate": 7.683472199523667e-07, "loss": 0.0004, "step": 205120 }, { "epoch": 3.356459134418719, "grad_norm": 0.0011039376258850098, "learning_rate": 7.679669703411935e-07, "loss": 0.0001, "step": 205130 }, { "epoch": 3.3566227603697945, "grad_norm": 0.002898995066061616, "learning_rate": 7.675868070175752e-07, "loss": 0.0001, "step": 205140 }, { "epoch": 3.3567863863208705, "grad_norm": 0.009416790679097176, "learning_rate": 7.672067299892616e-07, "loss": 0.0002, "step": 205150 }, { "epoch": 3.3569500122719464, "grad_norm": 0.06533171981573105, "learning_rate": 7.668267392640027e-07, "loss": 0.0008, "step": 205160 }, { "epoch": 3.357113638223022, "grad_norm": 0.030524611473083496, "learning_rate": 7.664468348495469e-07, "loss": 0.0001, "step": 205170 }, { "epoch": 3.357277264174098, "grad_norm": 0.006523956079035997, "learning_rate": 7.660670167536399e-07, "loss": 0.0006, "step": 205180 }, { "epoch": 3.357440890125174, "grad_norm": 0.0016150096198543906, "learning_rate": 7.656872849840252e-07, "loss": 0.0003, "step": 205190 }, { "epoch": 3.3576045160762495, "grad_norm": 0.018820870667696, "learning_rate": 7.653076395484454e-07, "loss": 0.0005, "step": 205200 }, { "epoch": 3.3577681420273255, "grad_norm": 0.006309648044407368, "learning_rate": 7.649280804546411e-07, "loss": 0.0006, "step": 205210 }, { "epoch": 3.3579317679784015, "grad_norm": 0.05984663963317871, "learning_rate": 7.645486077103525e-07, "loss": 0.0004, "step": 205220 }, { "epoch": 3.358095393929477, "grad_norm": 0.0343095138669014, "learning_rate": 7.641692213233149e-07, "loss": 0.0004, "step": 205230 }, { "epoch": 3.358259019880553, "grad_norm": 0.034910161048173904, "learning_rate": 7.637899213012657e-07, "loss": 0.0003, "step": 205240 }, { "epoch": 3.358422645831629, "grad_norm": 0.00955383200198412, "learning_rate": 7.634107076519342e-07, "loss": 0.0002, "step": 205250 }, { "epoch": 3.3585862717827046, "grad_norm": 0.04831041395664215, "learning_rate": 7.630315803830585e-07, "loss": 0.0003, "step": 205260 }, { "epoch": 3.3587498977337806, "grad_norm": 0.03386993333697319, "learning_rate": 7.626525395023637e-07, "loss": 0.0001, "step": 205270 }, { "epoch": 3.3589135236848566, "grad_norm": 0.06898234784603119, "learning_rate": 7.622735850175811e-07, "loss": 0.0004, "step": 205280 }, { "epoch": 3.359077149635932, "grad_norm": 0.0004875630256719887, "learning_rate": 7.618947169364338e-07, "loss": 0.0003, "step": 205290 }, { "epoch": 3.359240775587008, "grad_norm": 0.0035474570468068123, "learning_rate": 7.615159352666512e-07, "loss": 0.0002, "step": 205300 }, { "epoch": 3.359404401538084, "grad_norm": 0.05010276660323143, "learning_rate": 7.611372400159533e-07, "loss": 0.0002, "step": 205310 }, { "epoch": 3.3595680274891597, "grad_norm": 0.01842544414103031, "learning_rate": 7.607586311920622e-07, "loss": 0.0003, "step": 205320 }, { "epoch": 3.3597316534402357, "grad_norm": 0.07014957815408707, "learning_rate": 7.603801088026958e-07, "loss": 0.0004, "step": 205330 }, { "epoch": 3.3598952793913113, "grad_norm": 0.007386763580143452, "learning_rate": 7.600016728555753e-07, "loss": 0.0002, "step": 205340 }, { "epoch": 3.3600589053423873, "grad_norm": 0.006445162929594517, "learning_rate": 7.596233233584138e-07, "loss": 0.0003, "step": 205350 }, { "epoch": 3.3602225312934633, "grad_norm": 0.0029574420768767595, "learning_rate": 7.592450603189272e-07, "loss": 0.0004, "step": 205360 }, { "epoch": 3.360386157244539, "grad_norm": 0.005700475070625544, "learning_rate": 7.588668837448249e-07, "loss": 0.0001, "step": 205370 }, { "epoch": 3.360549783195615, "grad_norm": 0.048414427787065506, "learning_rate": 7.584887936438224e-07, "loss": 0.0002, "step": 205380 }, { "epoch": 3.360713409146691, "grad_norm": 0.013787888921797276, "learning_rate": 7.581107900236251e-07, "loss": 0.0006, "step": 205390 }, { "epoch": 3.3608770350977664, "grad_norm": 0.02201549895107746, "learning_rate": 7.57732872891942e-07, "loss": 0.0002, "step": 205400 }, { "epoch": 3.3610406610488424, "grad_norm": 0.0009088019724003971, "learning_rate": 7.573550422564768e-07, "loss": 0.0001, "step": 205410 }, { "epoch": 3.3612042869999184, "grad_norm": 0.005044294986873865, "learning_rate": 7.569772981249335e-07, "loss": 0.0002, "step": 205420 }, { "epoch": 3.361367912950994, "grad_norm": 0.13428157567977905, "learning_rate": 7.565996405050152e-07, "loss": 0.0004, "step": 205430 }, { "epoch": 3.36153153890207, "grad_norm": 0.017015768215060234, "learning_rate": 7.56222069404422e-07, "loss": 0.0003, "step": 205440 }, { "epoch": 3.361695164853146, "grad_norm": 0.0008770682034082711, "learning_rate": 7.55844584830851e-07, "loss": 0.0001, "step": 205450 }, { "epoch": 3.3618587908042215, "grad_norm": 0.12396299093961716, "learning_rate": 7.554671867919989e-07, "loss": 0.0006, "step": 205460 }, { "epoch": 3.3620224167552974, "grad_norm": 0.0032213223166763783, "learning_rate": 7.550898752955615e-07, "loss": 0.0002, "step": 205470 }, { "epoch": 3.362186042706373, "grad_norm": 0.01553070917725563, "learning_rate": 7.547126503492319e-07, "loss": 0.0004, "step": 205480 }, { "epoch": 3.362349668657449, "grad_norm": 0.019430480897426605, "learning_rate": 7.543355119607004e-07, "loss": 0.0002, "step": 205490 }, { "epoch": 3.362513294608525, "grad_norm": 5.926673475187272e-05, "learning_rate": 7.539584601376565e-07, "loss": 0.0004, "step": 205500 }, { "epoch": 3.3626769205596005, "grad_norm": 0.010818368755280972, "learning_rate": 7.535814948877884e-07, "loss": 0.0002, "step": 205510 }, { "epoch": 3.3628405465106765, "grad_norm": 0.04071848839521408, "learning_rate": 7.532046162187839e-07, "loss": 0.0003, "step": 205520 }, { "epoch": 3.3630041724617525, "grad_norm": 0.030878284946084023, "learning_rate": 7.528278241383241e-07, "loss": 0.0004, "step": 205530 }, { "epoch": 3.363167798412828, "grad_norm": 0.060571637004613876, "learning_rate": 7.524511186540928e-07, "loss": 0.0004, "step": 205540 }, { "epoch": 3.363331424363904, "grad_norm": 0.044755108654499054, "learning_rate": 7.520744997737705e-07, "loss": 0.0002, "step": 205550 }, { "epoch": 3.36349505031498, "grad_norm": 0.029031863436102867, "learning_rate": 7.516979675050379e-07, "loss": 0.0004, "step": 205560 }, { "epoch": 3.3636586762660556, "grad_norm": 0.0367068387567997, "learning_rate": 7.513215218555697e-07, "loss": 0.0003, "step": 205570 }, { "epoch": 3.3638223022171316, "grad_norm": 0.08595560491085052, "learning_rate": 7.509451628330417e-07, "loss": 0.0005, "step": 205580 }, { "epoch": 3.3639859281682076, "grad_norm": 0.015535344369709492, "learning_rate": 7.505688904451286e-07, "loss": 0.0004, "step": 205590 }, { "epoch": 3.364149554119283, "grad_norm": 0.02094102092087269, "learning_rate": 7.501927046995028e-07, "loss": 0.0002, "step": 205600 }, { "epoch": 3.364313180070359, "grad_norm": 0.16931107640266418, "learning_rate": 7.498166056038325e-07, "loss": 0.0006, "step": 205610 }, { "epoch": 3.364476806021435, "grad_norm": 0.11159206181764603, "learning_rate": 7.494405931657866e-07, "loss": 0.0002, "step": 205620 }, { "epoch": 3.3646404319725107, "grad_norm": 0.0002861351240426302, "learning_rate": 7.490646673930324e-07, "loss": 0.0004, "step": 205630 }, { "epoch": 3.3648040579235867, "grad_norm": 0.0036038928665220737, "learning_rate": 7.486888282932353e-07, "loss": 0.0002, "step": 205640 }, { "epoch": 3.3649676838746627, "grad_norm": 0.047556329518556595, "learning_rate": 7.483130758740565e-07, "loss": 0.0001, "step": 205650 }, { "epoch": 3.3651313098257383, "grad_norm": 0.08225519955158234, "learning_rate": 7.479374101431575e-07, "loss": 0.0002, "step": 205660 }, { "epoch": 3.3652949357768143, "grad_norm": 0.0015055258991196752, "learning_rate": 7.475618311081989e-07, "loss": 0.0003, "step": 205670 }, { "epoch": 3.3654585617278903, "grad_norm": 0.043509386479854584, "learning_rate": 7.471863387768391e-07, "loss": 0.0003, "step": 205680 }, { "epoch": 3.365622187678966, "grad_norm": 0.03262147679924965, "learning_rate": 7.46810933156732e-07, "loss": 0.0004, "step": 205690 }, { "epoch": 3.365785813630042, "grad_norm": 0.0016424540663138032, "learning_rate": 7.464356142555323e-07, "loss": 0.0006, "step": 205700 }, { "epoch": 3.365949439581118, "grad_norm": 0.012554293498396873, "learning_rate": 7.460603820808937e-07, "loss": 0.0008, "step": 205710 }, { "epoch": 3.3661130655321934, "grad_norm": 0.01000199280679226, "learning_rate": 7.456852366404665e-07, "loss": 0.0006, "step": 205720 }, { "epoch": 3.3662766914832694, "grad_norm": 0.0012717351783066988, "learning_rate": 7.45310177941898e-07, "loss": 0.0003, "step": 205730 }, { "epoch": 3.366440317434345, "grad_norm": 0.06819737702608109, "learning_rate": 7.449352059928366e-07, "loss": 0.0003, "step": 205740 }, { "epoch": 3.366603943385421, "grad_norm": 0.019662532955408096, "learning_rate": 7.445603208009278e-07, "loss": 0.0007, "step": 205750 }, { "epoch": 3.366767569336497, "grad_norm": 0.007802649866789579, "learning_rate": 7.441855223738159e-07, "loss": 0.0005, "step": 205760 }, { "epoch": 3.3669311952875725, "grad_norm": 0.04106037691235542, "learning_rate": 7.438108107191405e-07, "loss": 0.0002, "step": 205770 }, { "epoch": 3.3670948212386484, "grad_norm": 0.011309819296002388, "learning_rate": 7.434361858445432e-07, "loss": 0.0001, "step": 205780 }, { "epoch": 3.3672584471897244, "grad_norm": 0.0019985907711088657, "learning_rate": 7.430616477576619e-07, "loss": 0.0002, "step": 205790 }, { "epoch": 3.3674220731408, "grad_norm": 0.0027499280404299498, "learning_rate": 7.426871964661342e-07, "loss": 0.0002, "step": 205800 }, { "epoch": 3.367585699091876, "grad_norm": 0.0021847463212907314, "learning_rate": 7.423128319775929e-07, "loss": 0.0006, "step": 205810 }, { "epoch": 3.367749325042952, "grad_norm": 0.01348755694925785, "learning_rate": 7.419385542996721e-07, "loss": 0.0003, "step": 205820 }, { "epoch": 3.3679129509940275, "grad_norm": 0.03938642516732216, "learning_rate": 7.415643634400022e-07, "loss": 0.0003, "step": 205830 }, { "epoch": 3.3680765769451035, "grad_norm": 0.004109678789973259, "learning_rate": 7.411902594062148e-07, "loss": 0.0004, "step": 205840 }, { "epoch": 3.368240202896179, "grad_norm": 0.04422616586089134, "learning_rate": 7.408162422059351e-07, "loss": 0.0002, "step": 205850 }, { "epoch": 3.368403828847255, "grad_norm": 0.07499025017023087, "learning_rate": 7.404423118467907e-07, "loss": 0.0002, "step": 205860 }, { "epoch": 3.368567454798331, "grad_norm": 0.06988642364740372, "learning_rate": 7.400684683364029e-07, "loss": 0.0002, "step": 205870 }, { "epoch": 3.3687310807494066, "grad_norm": 0.03164265677332878, "learning_rate": 7.396947116823983e-07, "loss": 0.0005, "step": 205880 }, { "epoch": 3.3688947067004826, "grad_norm": 0.027262436226010323, "learning_rate": 7.393210418923935e-07, "loss": 0.0003, "step": 205890 }, { "epoch": 3.3690583326515586, "grad_norm": 0.00667025800794363, "learning_rate": 7.389474589740108e-07, "loss": 0.0005, "step": 205900 }, { "epoch": 3.369221958602634, "grad_norm": 0.06551990658044815, "learning_rate": 7.385739629348632e-07, "loss": 0.0002, "step": 205910 }, { "epoch": 3.36938558455371, "grad_norm": 0.01137707382440567, "learning_rate": 7.38200553782571e-07, "loss": 0.0003, "step": 205920 }, { "epoch": 3.369549210504786, "grad_norm": 0.014865553006529808, "learning_rate": 7.378272315247431e-07, "loss": 0.0003, "step": 205930 }, { "epoch": 3.3697128364558617, "grad_norm": 0.023493904620409012, "learning_rate": 7.374539961689947e-07, "loss": 0.0002, "step": 205940 }, { "epoch": 3.3698764624069377, "grad_norm": 0.002489967504516244, "learning_rate": 7.370808477229319e-07, "loss": 0.0001, "step": 205950 }, { "epoch": 3.3700400883580137, "grad_norm": 0.020773107185959816, "learning_rate": 7.367077861941668e-07, "loss": 0.0004, "step": 205960 }, { "epoch": 3.3702037143090893, "grad_norm": 0.017129793763160706, "learning_rate": 7.363348115903035e-07, "loss": 0.0004, "step": 205970 }, { "epoch": 3.3703673402601653, "grad_norm": 0.012600945308804512, "learning_rate": 7.359619239189486e-07, "loss": 0.0002, "step": 205980 }, { "epoch": 3.3705309662112413, "grad_norm": 0.07697080820798874, "learning_rate": 7.35589123187701e-07, "loss": 0.0004, "step": 205990 }, { "epoch": 3.370694592162317, "grad_norm": 0.0008141077705658972, "learning_rate": 7.352164094041669e-07, "loss": 0.0002, "step": 206000 }, { "epoch": 3.370858218113393, "grad_norm": 0.0049086157232522964, "learning_rate": 7.348437825759419e-07, "loss": 0.0002, "step": 206010 }, { "epoch": 3.371021844064469, "grad_norm": 0.04871480539441109, "learning_rate": 7.34471242710626e-07, "loss": 0.0006, "step": 206020 }, { "epoch": 3.3711854700155444, "grad_norm": 0.0002580246073193848, "learning_rate": 7.340987898158114e-07, "loss": 0.0003, "step": 206030 }, { "epoch": 3.3713490959666204, "grad_norm": 0.0036066165193915367, "learning_rate": 7.337264238990966e-07, "loss": 0.0002, "step": 206040 }, { "epoch": 3.3715127219176964, "grad_norm": 0.022434066981077194, "learning_rate": 7.333541449680703e-07, "loss": 0.0003, "step": 206050 }, { "epoch": 3.371676347868772, "grad_norm": 0.03174472227692604, "learning_rate": 7.329819530303262e-07, "loss": 0.0005, "step": 206060 }, { "epoch": 3.371839973819848, "grad_norm": 0.034571573138237, "learning_rate": 7.326098480934496e-07, "loss": 0.0005, "step": 206070 }, { "epoch": 3.372003599770924, "grad_norm": 0.008091001771390438, "learning_rate": 7.32237830165029e-07, "loss": 0.0005, "step": 206080 }, { "epoch": 3.3721672257219995, "grad_norm": 0.0235417652875185, "learning_rate": 7.318658992526495e-07, "loss": 0.0005, "step": 206090 }, { "epoch": 3.3723308516730754, "grad_norm": 0.0034666010178625584, "learning_rate": 7.314940553638956e-07, "loss": 0.0001, "step": 206100 }, { "epoch": 3.372494477624151, "grad_norm": 0.0002593260433059186, "learning_rate": 7.31122298506346e-07, "loss": 0.0002, "step": 206110 }, { "epoch": 3.372658103575227, "grad_norm": 0.016616009175777435, "learning_rate": 7.307506286875826e-07, "loss": 0.0002, "step": 206120 }, { "epoch": 3.372821729526303, "grad_norm": 0.053518570959568024, "learning_rate": 7.303790459151827e-07, "loss": 0.0006, "step": 206130 }, { "epoch": 3.3729853554773785, "grad_norm": 0.043306123465299606, "learning_rate": 7.300075501967246e-07, "loss": 0.0002, "step": 206140 }, { "epoch": 3.3731489814284545, "grad_norm": 0.002017963444814086, "learning_rate": 7.296361415397796e-07, "loss": 0.0004, "step": 206150 }, { "epoch": 3.3733126073795305, "grad_norm": 0.002027276437729597, "learning_rate": 7.292648199519215e-07, "loss": 0.0003, "step": 206160 }, { "epoch": 3.373476233330606, "grad_norm": 0.007865549065172672, "learning_rate": 7.28893585440722e-07, "loss": 0.0005, "step": 206170 }, { "epoch": 3.373639859281682, "grad_norm": 0.053322453051805496, "learning_rate": 7.285224380137512e-07, "loss": 0.0004, "step": 206180 }, { "epoch": 3.373803485232758, "grad_norm": 0.000676339550409466, "learning_rate": 7.281513776785736e-07, "loss": 0.0006, "step": 206190 }, { "epoch": 3.3739671111838336, "grad_norm": 0.0024885558523237705, "learning_rate": 7.277804044427561e-07, "loss": 0.0011, "step": 206200 }, { "epoch": 3.3741307371349096, "grad_norm": 0.02751031704246998, "learning_rate": 7.274095183138635e-07, "loss": 0.0003, "step": 206210 }, { "epoch": 3.3742943630859856, "grad_norm": 0.01812381111085415, "learning_rate": 7.270387192994576e-07, "loss": 0.0003, "step": 206220 }, { "epoch": 3.374457989037061, "grad_norm": 0.02714492194354534, "learning_rate": 7.266680074070975e-07, "loss": 0.0001, "step": 206230 }, { "epoch": 3.374621614988137, "grad_norm": 0.007529216818511486, "learning_rate": 7.262973826443421e-07, "loss": 0.0002, "step": 206240 }, { "epoch": 3.3747852409392127, "grad_norm": 0.0007159230881370604, "learning_rate": 7.259268450187484e-07, "loss": 0.0001, "step": 206250 }, { "epoch": 3.3749488668902887, "grad_norm": 0.007400400936603546, "learning_rate": 7.25556394537873e-07, "loss": 0.0001, "step": 206260 }, { "epoch": 3.3751124928413647, "grad_norm": 0.0641312301158905, "learning_rate": 7.251860312092656e-07, "loss": 0.0004, "step": 206270 }, { "epoch": 3.3752761187924403, "grad_norm": 0.0341065339744091, "learning_rate": 7.248157550404799e-07, "loss": 0.0004, "step": 206280 }, { "epoch": 3.3754397447435163, "grad_norm": 0.014375886879861355, "learning_rate": 7.244455660390648e-07, "loss": 0.0006, "step": 206290 }, { "epoch": 3.3756033706945923, "grad_norm": 0.04197582229971886, "learning_rate": 7.240754642125697e-07, "loss": 0.0002, "step": 206300 }, { "epoch": 3.375766996645668, "grad_norm": 0.00027633164427243173, "learning_rate": 7.237054495685381e-07, "loss": 0.0003, "step": 206310 }, { "epoch": 3.375930622596744, "grad_norm": 0.05152283236384392, "learning_rate": 7.233355221145155e-07, "loss": 0.0002, "step": 206320 }, { "epoch": 3.37609424854782, "grad_norm": 0.03151853382587433, "learning_rate": 7.229656818580444e-07, "loss": 0.0002, "step": 206330 }, { "epoch": 3.3762578744988954, "grad_norm": 0.00019316040561534464, "learning_rate": 7.225959288066665e-07, "loss": 0.0001, "step": 206340 }, { "epoch": 3.3764215004499714, "grad_norm": 0.061233267188072205, "learning_rate": 7.222262629679189e-07, "loss": 0.0001, "step": 206350 }, { "epoch": 3.3765851264010474, "grad_norm": 0.00036061019636690617, "learning_rate": 7.218566843493396e-07, "loss": 0.0002, "step": 206360 }, { "epoch": 3.376748752352123, "grad_norm": 0.03744673356413841, "learning_rate": 7.214871929584644e-07, "loss": 0.0002, "step": 206370 }, { "epoch": 3.376912378303199, "grad_norm": 0.025421909987926483, "learning_rate": 7.211177888028276e-07, "loss": 0.0003, "step": 206380 }, { "epoch": 3.377076004254275, "grad_norm": 0.009847791865468025, "learning_rate": 7.207484718899588e-07, "loss": 0.0002, "step": 206390 }, { "epoch": 3.3772396302053505, "grad_norm": 0.10056880861520767, "learning_rate": 7.203792422273892e-07, "loss": 0.0003, "step": 206400 }, { "epoch": 3.3774032561564264, "grad_norm": 0.0011150100035592914, "learning_rate": 7.200100998226473e-07, "loss": 0.0003, "step": 206410 }, { "epoch": 3.3775668821075024, "grad_norm": 0.03741928189992905, "learning_rate": 7.196410446832603e-07, "loss": 0.0002, "step": 206420 }, { "epoch": 3.377730508058578, "grad_norm": 0.0023930168244987726, "learning_rate": 7.192720768167511e-07, "loss": 0.0002, "step": 206430 }, { "epoch": 3.377894134009654, "grad_norm": 0.06758014857769012, "learning_rate": 7.189031962306436e-07, "loss": 0.0002, "step": 206440 }, { "epoch": 3.37805775996073, "grad_norm": 0.06871042400598526, "learning_rate": 7.185344029324592e-07, "loss": 0.0005, "step": 206450 }, { "epoch": 3.3782213859118055, "grad_norm": 0.0012992346892133355, "learning_rate": 7.181656969297185e-07, "loss": 0.0004, "step": 206460 }, { "epoch": 3.3783850118628815, "grad_norm": 0.009705353528261185, "learning_rate": 7.177970782299359e-07, "loss": 0.0002, "step": 206470 }, { "epoch": 3.3785486378139575, "grad_norm": 0.037454262375831604, "learning_rate": 7.174285468406289e-07, "loss": 0.0003, "step": 206480 }, { "epoch": 3.378712263765033, "grad_norm": 0.0523965023458004, "learning_rate": 7.170601027693125e-07, "loss": 0.0003, "step": 206490 }, { "epoch": 3.378875889716109, "grad_norm": 0.04991881549358368, "learning_rate": 7.166917460234985e-07, "loss": 0.0004, "step": 206500 }, { "epoch": 3.3790395156671846, "grad_norm": 0.03383791446685791, "learning_rate": 7.163234766106958e-07, "loss": 0.0003, "step": 206510 }, { "epoch": 3.3792031416182606, "grad_norm": 0.014448484405875206, "learning_rate": 7.159552945384157e-07, "loss": 0.0005, "step": 206520 }, { "epoch": 3.3793667675693366, "grad_norm": 0.0004013422585558146, "learning_rate": 7.155871998141617e-07, "loss": 0.0003, "step": 206530 }, { "epoch": 3.379530393520412, "grad_norm": 0.02938288263976574, "learning_rate": 7.152191924454427e-07, "loss": 0.0002, "step": 206540 }, { "epoch": 3.379694019471488, "grad_norm": 0.026813002303242683, "learning_rate": 7.148512724397594e-07, "loss": 0.0003, "step": 206550 }, { "epoch": 3.379857645422564, "grad_norm": 0.008907566778361797, "learning_rate": 7.144834398046152e-07, "loss": 0.0005, "step": 206560 }, { "epoch": 3.3800212713736397, "grad_norm": 0.01253176387399435, "learning_rate": 7.14115694547507e-07, "loss": 0.0003, "step": 206570 }, { "epoch": 3.3801848973247157, "grad_norm": 0.0002820710651576519, "learning_rate": 7.137480366759364e-07, "loss": 0.0001, "step": 206580 }, { "epoch": 3.3803485232757917, "grad_norm": 0.012350297532975674, "learning_rate": 7.133804661973975e-07, "loss": 0.0003, "step": 206590 }, { "epoch": 3.3805121492268673, "grad_norm": 0.040218375623226166, "learning_rate": 7.130129831193861e-07, "loss": 0.0004, "step": 206600 }, { "epoch": 3.3806757751779433, "grad_norm": 0.0030168043449521065, "learning_rate": 7.12645587449392e-07, "loss": 0.0003, "step": 206610 }, { "epoch": 3.380839401129019, "grad_norm": 0.03468312695622444, "learning_rate": 7.122782791949101e-07, "loss": 0.0002, "step": 206620 }, { "epoch": 3.381003027080095, "grad_norm": 0.008901768364012241, "learning_rate": 7.11911058363427e-07, "loss": 0.0003, "step": 206630 }, { "epoch": 3.381166653031171, "grad_norm": 0.043836791068315506, "learning_rate": 7.115439249624312e-07, "loss": 0.0004, "step": 206640 }, { "epoch": 3.3813302789822464, "grad_norm": 0.021408112719655037, "learning_rate": 7.11176878999405e-07, "loss": 0.0002, "step": 206650 }, { "epoch": 3.3814939049333224, "grad_norm": 0.06839541345834732, "learning_rate": 7.108099204818375e-07, "loss": 0.0002, "step": 206660 }, { "epoch": 3.3816575308843984, "grad_norm": 0.006657908204942942, "learning_rate": 7.104430494172065e-07, "loss": 0.0004, "step": 206670 }, { "epoch": 3.381821156835474, "grad_norm": 0.012845270335674286, "learning_rate": 7.100762658129945e-07, "loss": 0.0002, "step": 206680 }, { "epoch": 3.38198478278655, "grad_norm": 0.01491253636777401, "learning_rate": 7.097095696766781e-07, "loss": 0.0002, "step": 206690 }, { "epoch": 3.382148408737626, "grad_norm": 0.05145774409174919, "learning_rate": 7.093429610157348e-07, "loss": 0.0003, "step": 206700 }, { "epoch": 3.3823120346887015, "grad_norm": 0.020066969096660614, "learning_rate": 7.089764398376398e-07, "loss": 0.0003, "step": 206710 }, { "epoch": 3.3824756606397774, "grad_norm": 0.0007327523198910058, "learning_rate": 7.086100061498663e-07, "loss": 0.0002, "step": 206720 }, { "epoch": 3.3826392865908534, "grad_norm": 0.02538713626563549, "learning_rate": 7.082436599598847e-07, "loss": 0.0002, "step": 206730 }, { "epoch": 3.382802912541929, "grad_norm": 0.016591167077422142, "learning_rate": 7.078774012751644e-07, "loss": 0.0005, "step": 206740 }, { "epoch": 3.382966538493005, "grad_norm": 0.0587722472846508, "learning_rate": 7.075112301031734e-07, "loss": 0.0003, "step": 206750 }, { "epoch": 3.383130164444081, "grad_norm": 0.0026390645653009415, "learning_rate": 7.07145146451379e-07, "loss": 0.0007, "step": 206760 }, { "epoch": 3.3832937903951565, "grad_norm": 0.01342260092496872, "learning_rate": 7.06779150327243e-07, "loss": 0.0004, "step": 206770 }, { "epoch": 3.3834574163462325, "grad_norm": 0.04788888618350029, "learning_rate": 7.064132417382291e-07, "loss": 0.0003, "step": 206780 }, { "epoch": 3.3836210422973085, "grad_norm": 0.2274760901927948, "learning_rate": 7.060474206917978e-07, "loss": 0.0003, "step": 206790 }, { "epoch": 3.383784668248384, "grad_norm": 0.008923227898776531, "learning_rate": 7.056816871954086e-07, "loss": 0.0002, "step": 206800 }, { "epoch": 3.38394829419946, "grad_norm": 0.04140017181634903, "learning_rate": 7.053160412565163e-07, "loss": 0.0004, "step": 206810 }, { "epoch": 3.384111920150536, "grad_norm": 0.04128917679190636, "learning_rate": 7.049504828825776e-07, "loss": 0.0003, "step": 206820 }, { "epoch": 3.3842755461016116, "grad_norm": 0.004656681790947914, "learning_rate": 7.045850120810454e-07, "loss": 0.0002, "step": 206830 }, { "epoch": 3.3844391720526876, "grad_norm": 0.009393084794282913, "learning_rate": 7.042196288593734e-07, "loss": 0.0001, "step": 206840 }, { "epoch": 3.3846027980037636, "grad_norm": 0.013300691731274128, "learning_rate": 7.03854333225008e-07, "loss": 0.0002, "step": 206850 }, { "epoch": 3.384766423954839, "grad_norm": 0.0022075437009334564, "learning_rate": 7.034891251853998e-07, "loss": 0.0002, "step": 206860 }, { "epoch": 3.384930049905915, "grad_norm": 0.0016472551506012678, "learning_rate": 7.031240047479937e-07, "loss": 0.0001, "step": 206870 }, { "epoch": 3.3850936758569907, "grad_norm": 0.0719062015414238, "learning_rate": 7.02758971920236e-07, "loss": 0.0003, "step": 206880 }, { "epoch": 3.3852573018080667, "grad_norm": 0.04470895975828171, "learning_rate": 7.023940267095669e-07, "loss": 0.0002, "step": 206890 }, { "epoch": 3.3854209277591427, "grad_norm": 0.029246561229228973, "learning_rate": 7.020291691234287e-07, "loss": 0.0001, "step": 206900 }, { "epoch": 3.3855845537102183, "grad_norm": 0.0460650660097599, "learning_rate": 7.016643991692607e-07, "loss": 0.0004, "step": 206910 }, { "epoch": 3.3857481796612943, "grad_norm": 0.0042206523939967155, "learning_rate": 7.01299716854501e-07, "loss": 0.0003, "step": 206920 }, { "epoch": 3.3859118056123703, "grad_norm": 0.002302682725712657, "learning_rate": 7.009351221865829e-07, "loss": 0.0002, "step": 206930 }, { "epoch": 3.386075431563446, "grad_norm": 0.0056176264770329, "learning_rate": 7.005706151729419e-07, "loss": 0.0004, "step": 206940 }, { "epoch": 3.386239057514522, "grad_norm": 0.008717983029782772, "learning_rate": 7.002061958210088e-07, "loss": 0.0003, "step": 206950 }, { "epoch": 3.386402683465598, "grad_norm": 0.07315169274806976, "learning_rate": 6.998418641382165e-07, "loss": 0.0002, "step": 206960 }, { "epoch": 3.3865663094166734, "grad_norm": 0.029590517282485962, "learning_rate": 6.994776201319897e-07, "loss": 0.0007, "step": 206970 }, { "epoch": 3.3867299353677494, "grad_norm": 0.030045023187994957, "learning_rate": 6.991134638097569e-07, "loss": 0.0003, "step": 206980 }, { "epoch": 3.3868935613188254, "grad_norm": 0.02030840888619423, "learning_rate": 6.987493951789426e-07, "loss": 0.0005, "step": 206990 }, { "epoch": 3.387057187269901, "grad_norm": 0.003791691968217492, "learning_rate": 6.983854142469715e-07, "loss": 0.0002, "step": 207000 }, { "epoch": 3.387220813220977, "grad_norm": 0.01928037218749523, "learning_rate": 6.980215210212621e-07, "loss": 0.0002, "step": 207010 }, { "epoch": 3.3873844391720525, "grad_norm": 0.019932271912693977, "learning_rate": 6.976577155092351e-07, "loss": 0.0002, "step": 207020 }, { "epoch": 3.3875480651231284, "grad_norm": 0.025191286578774452, "learning_rate": 6.972939977183085e-07, "loss": 0.0004, "step": 207030 }, { "epoch": 3.3877116910742044, "grad_norm": 0.0487813837826252, "learning_rate": 6.969303676558992e-07, "loss": 0.0005, "step": 207040 }, { "epoch": 3.38787531702528, "grad_norm": 0.015163745731115341, "learning_rate": 6.965668253294189e-07, "loss": 0.0002, "step": 207050 }, { "epoch": 3.388038942976356, "grad_norm": 0.06430620700120926, "learning_rate": 6.962033707462807e-07, "loss": 0.0003, "step": 207060 }, { "epoch": 3.388202568927432, "grad_norm": 0.060734014958143234, "learning_rate": 6.958400039138957e-07, "loss": 0.0003, "step": 207070 }, { "epoch": 3.3883661948785075, "grad_norm": 0.042946599423885345, "learning_rate": 6.954767248396737e-07, "loss": 0.0004, "step": 207080 }, { "epoch": 3.3885298208295835, "grad_norm": 0.07825828343629837, "learning_rate": 6.951135335310194e-07, "loss": 0.0002, "step": 207090 }, { "epoch": 3.3886934467806595, "grad_norm": 0.015742486342787743, "learning_rate": 6.94750429995339e-07, "loss": 0.0002, "step": 207100 }, { "epoch": 3.388857072731735, "grad_norm": 0.041687287390232086, "learning_rate": 6.943874142400353e-07, "loss": 0.0003, "step": 207110 }, { "epoch": 3.389020698682811, "grad_norm": 0.002461356343701482, "learning_rate": 6.94024486272512e-07, "loss": 0.0002, "step": 207120 }, { "epoch": 3.389184324633887, "grad_norm": 0.029201611876487732, "learning_rate": 6.936616461001661e-07, "loss": 0.0003, "step": 207130 }, { "epoch": 3.3893479505849626, "grad_norm": 0.0077058859169483185, "learning_rate": 6.932988937303981e-07, "loss": 0.0003, "step": 207140 }, { "epoch": 3.3895115765360386, "grad_norm": 0.09750429540872574, "learning_rate": 6.929362291706005e-07, "loss": 0.0001, "step": 207150 }, { "epoch": 3.3896752024871146, "grad_norm": 0.01692037098109722, "learning_rate": 6.925736524281723e-07, "loss": 0.0004, "step": 207160 }, { "epoch": 3.38983882843819, "grad_norm": 0.0012518307194113731, "learning_rate": 6.922111635105033e-07, "loss": 0.0003, "step": 207170 }, { "epoch": 3.390002454389266, "grad_norm": 0.01329110749065876, "learning_rate": 6.918487624249859e-07, "loss": 0.0004, "step": 207180 }, { "epoch": 3.390166080340342, "grad_norm": 0.0002066507440758869, "learning_rate": 6.914864491790057e-07, "loss": 0.0001, "step": 207190 }, { "epoch": 3.3903297062914177, "grad_norm": 0.051650695502758026, "learning_rate": 6.911242237799554e-07, "loss": 0.0003, "step": 207200 }, { "epoch": 3.3904933322424937, "grad_norm": 0.009355640038847923, "learning_rate": 6.90762086235216e-07, "loss": 0.0004, "step": 207210 }, { "epoch": 3.3906569581935697, "grad_norm": 0.008548630401492119, "learning_rate": 6.90400036552174e-07, "loss": 0.0006, "step": 207220 }, { "epoch": 3.3908205841446453, "grad_norm": 0.11903493106365204, "learning_rate": 6.900380747382085e-07, "loss": 0.0004, "step": 207230 }, { "epoch": 3.3909842100957213, "grad_norm": 0.07072007656097412, "learning_rate": 6.896762008007024e-07, "loss": 0.0004, "step": 207240 }, { "epoch": 3.3911478360467973, "grad_norm": 0.008169951848685741, "learning_rate": 6.893144147470321e-07, "loss": 0.0003, "step": 207250 }, { "epoch": 3.391311461997873, "grad_norm": 0.037270884960889816, "learning_rate": 6.889527165845761e-07, "loss": 0.0005, "step": 207260 }, { "epoch": 3.391475087948949, "grad_norm": 0.02739628590643406, "learning_rate": 6.885911063207051e-07, "loss": 0.0003, "step": 207270 }, { "epoch": 3.3916387139000244, "grad_norm": 0.013835842721164227, "learning_rate": 6.882295839627972e-07, "loss": 0.0001, "step": 207280 }, { "epoch": 3.3918023398511004, "grad_norm": 0.04763156175613403, "learning_rate": 6.878681495182205e-07, "loss": 0.0004, "step": 207290 }, { "epoch": 3.3919659658021764, "grad_norm": 0.0012034706305712461, "learning_rate": 6.875068029943449e-07, "loss": 0.0006, "step": 207300 }, { "epoch": 3.392129591753252, "grad_norm": 0.006074997130781412, "learning_rate": 6.871455443985364e-07, "loss": 0.0001, "step": 207310 }, { "epoch": 3.392293217704328, "grad_norm": 0.07719346135854721, "learning_rate": 6.867843737381646e-07, "loss": 0.0004, "step": 207320 }, { "epoch": 3.392456843655404, "grad_norm": 0.009007332846522331, "learning_rate": 6.864232910205898e-07, "loss": 0.0003, "step": 207330 }, { "epoch": 3.3926204696064794, "grad_norm": 0.003737980267032981, "learning_rate": 6.860622962531766e-07, "loss": 0.0001, "step": 207340 }, { "epoch": 3.3927840955575554, "grad_norm": 0.004996432922780514, "learning_rate": 6.857013894432835e-07, "loss": 0.0001, "step": 207350 }, { "epoch": 3.3929477215086314, "grad_norm": 0.018914861604571342, "learning_rate": 6.853405705982697e-07, "loss": 0.0002, "step": 207360 }, { "epoch": 3.393111347459707, "grad_norm": 0.005450249649584293, "learning_rate": 6.84979839725492e-07, "loss": 0.0002, "step": 207370 }, { "epoch": 3.393274973410783, "grad_norm": 0.004673720337450504, "learning_rate": 6.846191968323068e-07, "loss": 0.0001, "step": 207380 }, { "epoch": 3.3934385993618585, "grad_norm": 0.005934200715273619, "learning_rate": 6.84258641926065e-07, "loss": 0.0002, "step": 207390 }, { "epoch": 3.3936022253129345, "grad_norm": 0.019992055371403694, "learning_rate": 6.838981750141188e-07, "loss": 0.0002, "step": 207400 }, { "epoch": 3.3937658512640105, "grad_norm": 0.15704511106014252, "learning_rate": 6.835377961038187e-07, "loss": 0.0001, "step": 207410 }, { "epoch": 3.393929477215086, "grad_norm": 0.0072652967646718025, "learning_rate": 6.831775052025125e-07, "loss": 0.0006, "step": 207420 }, { "epoch": 3.394093103166162, "grad_norm": 0.006098312325775623, "learning_rate": 6.82817302317545e-07, "loss": 0.0003, "step": 207430 }, { "epoch": 3.394256729117238, "grad_norm": 0.026103444397449493, "learning_rate": 6.824571874562608e-07, "loss": 0.0002, "step": 207440 }, { "epoch": 3.3944203550683136, "grad_norm": 0.043519750237464905, "learning_rate": 6.820971606260024e-07, "loss": 0.0005, "step": 207450 }, { "epoch": 3.3945839810193896, "grad_norm": 0.0010252578649669886, "learning_rate": 6.817372218341123e-07, "loss": 0.0003, "step": 207460 }, { "epoch": 3.3947476069704656, "grad_norm": 0.1146286204457283, "learning_rate": 6.813773710879263e-07, "loss": 0.0003, "step": 207470 }, { "epoch": 3.394911232921541, "grad_norm": 0.017639368772506714, "learning_rate": 6.810176083947828e-07, "loss": 0.0001, "step": 207480 }, { "epoch": 3.395074858872617, "grad_norm": 0.041492920368909836, "learning_rate": 6.806579337620168e-07, "loss": 0.0002, "step": 207490 }, { "epoch": 3.395238484823693, "grad_norm": 0.05753032863140106, "learning_rate": 6.802983471969638e-07, "loss": 0.0005, "step": 207500 }, { "epoch": 3.3954021107747687, "grad_norm": 0.06320371478796005, "learning_rate": 6.79938848706952e-07, "loss": 0.0003, "step": 207510 }, { "epoch": 3.3955657367258447, "grad_norm": 0.006833259016275406, "learning_rate": 6.795794382993137e-07, "loss": 0.0002, "step": 207520 }, { "epoch": 3.3957293626769207, "grad_norm": 0.006414588075131178, "learning_rate": 6.792201159813755e-07, "loss": 0.0002, "step": 207530 }, { "epoch": 3.3958929886279963, "grad_norm": 0.08906685560941696, "learning_rate": 6.788608817604658e-07, "loss": 0.0004, "step": 207540 }, { "epoch": 3.3960566145790723, "grad_norm": 0.04041318967938423, "learning_rate": 6.785017356439072e-07, "loss": 0.0012, "step": 207550 }, { "epoch": 3.3962202405301483, "grad_norm": 0.11837950348854065, "learning_rate": 6.781426776390221e-07, "loss": 0.0007, "step": 207560 }, { "epoch": 3.396383866481224, "grad_norm": 0.00020800081256311387, "learning_rate": 6.77783707753133e-07, "loss": 0.0003, "step": 207570 }, { "epoch": 3.3965474924323, "grad_norm": 0.005921336822211742, "learning_rate": 6.774248259935584e-07, "loss": 0.0004, "step": 207580 }, { "epoch": 3.396711118383376, "grad_norm": 0.009474861435592175, "learning_rate": 6.770660323676148e-07, "loss": 0.0002, "step": 207590 }, { "epoch": 3.3968747443344514, "grad_norm": 0.031661976128816605, "learning_rate": 6.767073268826185e-07, "loss": 0.0002, "step": 207600 }, { "epoch": 3.3970383702855274, "grad_norm": 0.002546421717852354, "learning_rate": 6.763487095458826e-07, "loss": 0.0001, "step": 207610 }, { "epoch": 3.3972019962366033, "grad_norm": 0.003037391696125269, "learning_rate": 6.759901803647207e-07, "loss": 0.0003, "step": 207620 }, { "epoch": 3.397365622187679, "grad_norm": 0.009707508608698845, "learning_rate": 6.756317393464402e-07, "loss": 0.0001, "step": 207630 }, { "epoch": 3.397529248138755, "grad_norm": 0.02585786022245884, "learning_rate": 6.752733864983507e-07, "loss": 0.0007, "step": 207640 }, { "epoch": 3.397692874089831, "grad_norm": 6.949465023353696e-05, "learning_rate": 6.749151218277589e-07, "loss": 0.0004, "step": 207650 }, { "epoch": 3.3978565000409064, "grad_norm": 0.018020853400230408, "learning_rate": 6.745569453419709e-07, "loss": 0.0001, "step": 207660 }, { "epoch": 3.3980201259919824, "grad_norm": 0.09016421437263489, "learning_rate": 6.741988570482866e-07, "loss": 0.0008, "step": 207670 }, { "epoch": 3.398183751943058, "grad_norm": 0.022634776309132576, "learning_rate": 6.73840856954009e-07, "loss": 0.0002, "step": 207680 }, { "epoch": 3.398347377894134, "grad_norm": 0.03498126566410065, "learning_rate": 6.734829450664366e-07, "loss": 0.0005, "step": 207690 }, { "epoch": 3.39851100384521, "grad_norm": 0.023220762610435486, "learning_rate": 6.731251213928685e-07, "loss": 0.0002, "step": 207700 }, { "epoch": 3.3986746297962855, "grad_norm": 0.0826532319188118, "learning_rate": 6.727673859405986e-07, "loss": 0.0009, "step": 207710 }, { "epoch": 3.3988382557473615, "grad_norm": 0.000886200403328985, "learning_rate": 6.724097387169215e-07, "loss": 0.0002, "step": 207720 }, { "epoch": 3.3990018816984375, "grad_norm": 0.041572730988264084, "learning_rate": 6.720521797291291e-07, "loss": 0.0003, "step": 207730 }, { "epoch": 3.399165507649513, "grad_norm": 0.03621917590498924, "learning_rate": 6.716947089845133e-07, "loss": 0.0003, "step": 207740 }, { "epoch": 3.399329133600589, "grad_norm": 0.04641333594918251, "learning_rate": 6.7133732649036e-07, "loss": 0.0002, "step": 207750 }, { "epoch": 3.399492759551665, "grad_norm": 0.0016381466994062066, "learning_rate": 6.709800322539572e-07, "loss": 0.0002, "step": 207760 }, { "epoch": 3.3996563855027406, "grad_norm": 0.036209121346473694, "learning_rate": 6.706228262825898e-07, "loss": 0.0003, "step": 207770 }, { "epoch": 3.3998200114538166, "grad_norm": 0.030503802001476288, "learning_rate": 6.702657085835424e-07, "loss": 0.0006, "step": 207780 }, { "epoch": 3.399983637404892, "grad_norm": 0.001068775774911046, "learning_rate": 6.699086791640935e-07, "loss": 0.0001, "step": 207790 }, { "epoch": 3.400147263355968, "grad_norm": 0.036512214690446854, "learning_rate": 6.695517380315253e-07, "loss": 0.0003, "step": 207800 }, { "epoch": 3.400310889307044, "grad_norm": 0.05396750196814537, "learning_rate": 6.691948851931118e-07, "loss": 0.0003, "step": 207810 }, { "epoch": 3.4004745152581197, "grad_norm": 0.01564166508615017, "learning_rate": 6.688381206561339e-07, "loss": 0.0004, "step": 207820 }, { "epoch": 3.4006381412091957, "grad_norm": 0.001112543512135744, "learning_rate": 6.68481444427862e-07, "loss": 0.0006, "step": 207830 }, { "epoch": 3.4008017671602717, "grad_norm": 0.025589387863874435, "learning_rate": 6.681248565155707e-07, "loss": 0.0007, "step": 207840 }, { "epoch": 3.4009653931113473, "grad_norm": 0.0523044727742672, "learning_rate": 6.67768356926527e-07, "loss": 0.0003, "step": 207850 }, { "epoch": 3.4011290190624233, "grad_norm": 0.01650090143084526, "learning_rate": 6.674119456680045e-07, "loss": 0.0002, "step": 207860 }, { "epoch": 3.4012926450134993, "grad_norm": 0.009631246328353882, "learning_rate": 6.670556227472669e-07, "loss": 0.0002, "step": 207870 }, { "epoch": 3.401456270964575, "grad_norm": 0.007534893695265055, "learning_rate": 6.666993881715805e-07, "loss": 0.0002, "step": 207880 }, { "epoch": 3.401619896915651, "grad_norm": 0.0014139249688014388, "learning_rate": 6.663432419482063e-07, "loss": 0.0003, "step": 207890 }, { "epoch": 3.401783522866727, "grad_norm": 0.023879727348685265, "learning_rate": 6.659871840844101e-07, "loss": 0.0004, "step": 207900 }, { "epoch": 3.4019471488178024, "grad_norm": 0.011769242584705353, "learning_rate": 6.656312145874482e-07, "loss": 0.0005, "step": 207910 }, { "epoch": 3.4021107747688784, "grad_norm": 0.06876669824123383, "learning_rate": 6.652753334645806e-07, "loss": 0.0001, "step": 207920 }, { "epoch": 3.4022744007199544, "grad_norm": 0.07353337109088898, "learning_rate": 6.649195407230602e-07, "loss": 0.0001, "step": 207930 }, { "epoch": 3.40243802667103, "grad_norm": 0.03368352726101875, "learning_rate": 6.645638363701463e-07, "loss": 0.0003, "step": 207940 }, { "epoch": 3.402601652622106, "grad_norm": 0.006716514006257057, "learning_rate": 6.642082204130867e-07, "loss": 0.0003, "step": 207950 }, { "epoch": 3.402765278573182, "grad_norm": 0.017331616953015327, "learning_rate": 6.638526928591355e-07, "loss": 0.0003, "step": 207960 }, { "epoch": 3.4029289045242574, "grad_norm": 0.003603314980864525, "learning_rate": 6.634972537155388e-07, "loss": 0.0002, "step": 207970 }, { "epoch": 3.4030925304753334, "grad_norm": 0.038480065762996674, "learning_rate": 6.631419029895453e-07, "loss": 0.0017, "step": 207980 }, { "epoch": 3.4032561564264094, "grad_norm": 0.003014979651197791, "learning_rate": 6.627866406883998e-07, "loss": 0.0006, "step": 207990 }, { "epoch": 3.403419782377485, "grad_norm": 0.06066305190324783, "learning_rate": 6.624314668193472e-07, "loss": 0.0004, "step": 208000 }, { "epoch": 3.403583408328561, "grad_norm": 0.0007523203967139125, "learning_rate": 6.62076381389627e-07, "loss": 0.0001, "step": 208010 }, { "epoch": 3.403747034279637, "grad_norm": 0.013237097300589085, "learning_rate": 6.617213844064796e-07, "loss": 0.0003, "step": 208020 }, { "epoch": 3.4039106602307125, "grad_norm": 0.004130130168050528, "learning_rate": 6.613664758771443e-07, "loss": 0.0002, "step": 208030 }, { "epoch": 3.4040742861817885, "grad_norm": 0.009394148364663124, "learning_rate": 6.610116558088575e-07, "loss": 0.0002, "step": 208040 }, { "epoch": 3.404237912132864, "grad_norm": 0.06412658840417862, "learning_rate": 6.606569242088517e-07, "loss": 0.0003, "step": 208050 }, { "epoch": 3.40440153808394, "grad_norm": 0.08457817882299423, "learning_rate": 6.603022810843606e-07, "loss": 0.0003, "step": 208060 }, { "epoch": 3.404565164035016, "grad_norm": 0.0021832797210663557, "learning_rate": 6.599477264426157e-07, "loss": 0.0002, "step": 208070 }, { "epoch": 3.4047287899860916, "grad_norm": 0.00026405215612612665, "learning_rate": 6.595932602908461e-07, "loss": 0.0001, "step": 208080 }, { "epoch": 3.4048924159371676, "grad_norm": 0.02991885505616665, "learning_rate": 6.592388826362778e-07, "loss": 0.0004, "step": 208090 }, { "epoch": 3.4050560418882436, "grad_norm": 0.0453643761575222, "learning_rate": 6.588845934861371e-07, "loss": 0.0004, "step": 208100 }, { "epoch": 3.405219667839319, "grad_norm": 0.06596308946609497, "learning_rate": 6.585303928476472e-07, "loss": 0.0004, "step": 208110 }, { "epoch": 3.405383293790395, "grad_norm": 0.027267195284366608, "learning_rate": 6.581762807280317e-07, "loss": 0.0003, "step": 208120 }, { "epoch": 3.405546919741471, "grad_norm": 0.004663147032260895, "learning_rate": 6.578222571345083e-07, "loss": 0.0001, "step": 208130 }, { "epoch": 3.4057105456925467, "grad_norm": 0.02062254212796688, "learning_rate": 6.574683220742956e-07, "loss": 0.0002, "step": 208140 }, { "epoch": 3.4058741716436227, "grad_norm": 0.0484507717192173, "learning_rate": 6.57114475554611e-07, "loss": 0.0002, "step": 208150 }, { "epoch": 3.4060377975946987, "grad_norm": 0.019125407561659813, "learning_rate": 6.567607175826702e-07, "loss": 0.0003, "step": 208160 }, { "epoch": 3.4062014235457743, "grad_norm": 0.018960652872920036, "learning_rate": 6.564070481656825e-07, "loss": 0.0001, "step": 208170 }, { "epoch": 3.4063650494968503, "grad_norm": 0.0010489750420674682, "learning_rate": 6.560534673108621e-07, "loss": 0.0003, "step": 208180 }, { "epoch": 3.406528675447926, "grad_norm": 0.046454671770334244, "learning_rate": 6.556999750254161e-07, "loss": 0.0004, "step": 208190 }, { "epoch": 3.406692301399002, "grad_norm": 0.013347183354198933, "learning_rate": 6.553465713165547e-07, "loss": 0.0001, "step": 208200 }, { "epoch": 3.406855927350078, "grad_norm": 0.0008579976274631917, "learning_rate": 6.549932561914807e-07, "loss": 0.0001, "step": 208210 }, { "epoch": 3.4070195533011534, "grad_norm": 0.0035555728245526552, "learning_rate": 6.546400296573985e-07, "loss": 0.0002, "step": 208220 }, { "epoch": 3.4071831792522294, "grad_norm": 0.003010835498571396, "learning_rate": 6.542868917215106e-07, "loss": 0.0003, "step": 208230 }, { "epoch": 3.4073468052033054, "grad_norm": 0.04210580512881279, "learning_rate": 6.53933842391018e-07, "loss": 0.0002, "step": 208240 }, { "epoch": 3.407510431154381, "grad_norm": 0.011298040859401226, "learning_rate": 6.535808816731176e-07, "loss": 0.0002, "step": 208250 }, { "epoch": 3.407674057105457, "grad_norm": 0.002712102374061942, "learning_rate": 6.53228009575006e-07, "loss": 0.0002, "step": 208260 }, { "epoch": 3.407837683056533, "grad_norm": 0.030865471810102463, "learning_rate": 6.528752261038785e-07, "loss": 0.0003, "step": 208270 }, { "epoch": 3.4080013090076084, "grad_norm": 0.008541759103536606, "learning_rate": 6.525225312669292e-07, "loss": 0.0003, "step": 208280 }, { "epoch": 3.4081649349586844, "grad_norm": 0.07694410532712936, "learning_rate": 6.521699250713475e-07, "loss": 0.0004, "step": 208290 }, { "epoch": 3.4083285609097604, "grad_norm": 0.001080931513570249, "learning_rate": 6.518174075243228e-07, "loss": 0.0002, "step": 208300 }, { "epoch": 3.408492186860836, "grad_norm": 0.05197646841406822, "learning_rate": 6.514649786330435e-07, "loss": 0.0003, "step": 208310 }, { "epoch": 3.408655812811912, "grad_norm": 0.004429819993674755, "learning_rate": 6.511126384046956e-07, "loss": 0.0, "step": 208320 }, { "epoch": 3.408819438762988, "grad_norm": 0.0004147144209127873, "learning_rate": 6.507603868464618e-07, "loss": 0.0003, "step": 208330 }, { "epoch": 3.4089830647140635, "grad_norm": 0.20270538330078125, "learning_rate": 6.504082239655246e-07, "loss": 0.0012, "step": 208340 }, { "epoch": 3.4091466906651395, "grad_norm": 0.04635882377624512, "learning_rate": 6.500561497690644e-07, "loss": 0.0004, "step": 208350 }, { "epoch": 3.4093103166162155, "grad_norm": 0.1990021914243698, "learning_rate": 6.49704164264261e-07, "loss": 0.0006, "step": 208360 }, { "epoch": 3.409473942567291, "grad_norm": 0.014143608510494232, "learning_rate": 6.493522674582886e-07, "loss": 0.0001, "step": 208370 }, { "epoch": 3.409637568518367, "grad_norm": 0.002261075656861067, "learning_rate": 6.490004593583238e-07, "loss": 0.0002, "step": 208380 }, { "epoch": 3.409801194469443, "grad_norm": 0.002574541140347719, "learning_rate": 6.486487399715391e-07, "loss": 0.0003, "step": 208390 }, { "epoch": 3.4099648204205186, "grad_norm": 0.04355282336473465, "learning_rate": 6.482971093051072e-07, "loss": 0.0005, "step": 208400 }, { "epoch": 3.4101284463715946, "grad_norm": 0.005211963783949614, "learning_rate": 6.479455673661944e-07, "loss": 0.0002, "step": 208410 }, { "epoch": 3.4102920723226706, "grad_norm": 0.07482168823480606, "learning_rate": 6.475941141619707e-07, "loss": 0.0002, "step": 208420 }, { "epoch": 3.410455698273746, "grad_norm": 0.0028122863732278347, "learning_rate": 6.472427496996014e-07, "loss": 0.0001, "step": 208430 }, { "epoch": 3.410619324224822, "grad_norm": 0.11700528115034103, "learning_rate": 6.46891473986252e-07, "loss": 0.0004, "step": 208440 }, { "epoch": 3.4107829501758977, "grad_norm": 0.07623554021120071, "learning_rate": 6.465402870290816e-07, "loss": 0.0003, "step": 208450 }, { "epoch": 3.4109465761269737, "grad_norm": 0.0070502907037734985, "learning_rate": 6.461891888352539e-07, "loss": 0.0004, "step": 208460 }, { "epoch": 3.4111102020780497, "grad_norm": 0.13999132812023163, "learning_rate": 6.458381794119234e-07, "loss": 0.0003, "step": 208470 }, { "epoch": 3.4112738280291253, "grad_norm": 0.0007630570326000452, "learning_rate": 6.454872587662514e-07, "loss": 0.0002, "step": 208480 }, { "epoch": 3.4114374539802013, "grad_norm": 0.0051378230564296246, "learning_rate": 6.451364269053895e-07, "loss": 0.0003, "step": 208490 }, { "epoch": 3.4116010799312773, "grad_norm": 0.034798089414834976, "learning_rate": 6.447856838364935e-07, "loss": 0.0004, "step": 208500 }, { "epoch": 3.411764705882353, "grad_norm": 0.04749589040875435, "learning_rate": 6.444350295667112e-07, "loss": 0.0001, "step": 208510 }, { "epoch": 3.411928331833429, "grad_norm": 0.06947597116231918, "learning_rate": 6.440844641031969e-07, "loss": 0.0003, "step": 208520 }, { "epoch": 3.412091957784505, "grad_norm": 0.024557478725910187, "learning_rate": 6.437339874530946e-07, "loss": 0.0004, "step": 208530 }, { "epoch": 3.4122555837355804, "grad_norm": 0.03658308461308479, "learning_rate": 6.433835996235522e-07, "loss": 0.0003, "step": 208540 }, { "epoch": 3.4124192096866564, "grad_norm": 0.061402350664138794, "learning_rate": 6.43033300621711e-07, "loss": 0.0003, "step": 208550 }, { "epoch": 3.412582835637732, "grad_norm": 0.016443772241473198, "learning_rate": 6.426830904547176e-07, "loss": 0.0002, "step": 208560 }, { "epoch": 3.412746461588808, "grad_norm": 0.012903752736747265, "learning_rate": 6.423329691297093e-07, "loss": 0.0002, "step": 208570 }, { "epoch": 3.412910087539884, "grad_norm": 0.06387815624475479, "learning_rate": 6.419829366538266e-07, "loss": 0.0003, "step": 208580 }, { "epoch": 3.4130737134909594, "grad_norm": 0.0024811073672026396, "learning_rate": 6.41632993034203e-07, "loss": 0.0002, "step": 208590 }, { "epoch": 3.4132373394420354, "grad_norm": 0.04988335818052292, "learning_rate": 6.412831382779789e-07, "loss": 0.0004, "step": 208600 }, { "epoch": 3.4134009653931114, "grad_norm": 0.02783939428627491, "learning_rate": 6.409333723922828e-07, "loss": 0.0003, "step": 208610 }, { "epoch": 3.413564591344187, "grad_norm": 0.0009878751588985324, "learning_rate": 6.405836953842493e-07, "loss": 0.0004, "step": 208620 }, { "epoch": 3.413728217295263, "grad_norm": 0.029932767152786255, "learning_rate": 6.402341072610058e-07, "loss": 0.0003, "step": 208630 }, { "epoch": 3.413891843246339, "grad_norm": 0.023744283244013786, "learning_rate": 6.398846080296811e-07, "loss": 0.0004, "step": 208640 }, { "epoch": 3.4140554691974145, "grad_norm": 0.007909273728728294, "learning_rate": 6.395351976974007e-07, "loss": 0.0006, "step": 208650 }, { "epoch": 3.4142190951484905, "grad_norm": 0.024588268250226974, "learning_rate": 6.391858762712904e-07, "loss": 0.0002, "step": 208660 }, { "epoch": 3.4143827210995665, "grad_norm": 0.2513173222541809, "learning_rate": 6.388366437584703e-07, "loss": 0.0003, "step": 208670 }, { "epoch": 3.414546347050642, "grad_norm": 0.013222146779298782, "learning_rate": 6.384875001660623e-07, "loss": 0.0003, "step": 208680 }, { "epoch": 3.414709973001718, "grad_norm": 0.005338403396308422, "learning_rate": 6.381384455011841e-07, "loss": 0.0003, "step": 208690 }, { "epoch": 3.414873598952794, "grad_norm": 0.03186944127082825, "learning_rate": 6.37789479770955e-07, "loss": 0.0003, "step": 208700 }, { "epoch": 3.4150372249038696, "grad_norm": 0.04508795216679573, "learning_rate": 6.374406029824875e-07, "loss": 0.0003, "step": 208710 }, { "epoch": 3.4152008508549456, "grad_norm": 0.03724639490246773, "learning_rate": 6.370918151428957e-07, "loss": 0.0004, "step": 208720 }, { "epoch": 3.4153644768060216, "grad_norm": 0.07004209607839584, "learning_rate": 6.367431162592913e-07, "loss": 0.0003, "step": 208730 }, { "epoch": 3.415528102757097, "grad_norm": 0.08837256580591202, "learning_rate": 6.363945063387849e-07, "loss": 0.0004, "step": 208740 }, { "epoch": 3.415691728708173, "grad_norm": 0.043323151767253876, "learning_rate": 6.360459853884821e-07, "loss": 0.0002, "step": 208750 }, { "epoch": 3.415855354659249, "grad_norm": 0.022048799321055412, "learning_rate": 6.356975534154908e-07, "loss": 0.0001, "step": 208760 }, { "epoch": 3.4160189806103247, "grad_norm": 0.0051438757218420506, "learning_rate": 6.353492104269143e-07, "loss": 0.0001, "step": 208770 }, { "epoch": 3.4161826065614007, "grad_norm": 0.0028209900483489037, "learning_rate": 6.350009564298564e-07, "loss": 0.0002, "step": 208780 }, { "epoch": 3.4163462325124767, "grad_norm": 0.053951047360897064, "learning_rate": 6.34652791431416e-07, "loss": 0.0002, "step": 208790 }, { "epoch": 3.4165098584635523, "grad_norm": 0.0004542370734270662, "learning_rate": 6.343047154386927e-07, "loss": 0.0002, "step": 208800 }, { "epoch": 3.4166734844146283, "grad_norm": 0.015511926263570786, "learning_rate": 6.339567284587827e-07, "loss": 0.0001, "step": 208810 }, { "epoch": 3.416837110365704, "grad_norm": 0.15166126191616058, "learning_rate": 6.336088304987836e-07, "loss": 0.0004, "step": 208820 }, { "epoch": 3.41700073631678, "grad_norm": 0.011283544823527336, "learning_rate": 6.332610215657853e-07, "loss": 0.0002, "step": 208830 }, { "epoch": 3.417164362267856, "grad_norm": 0.02857678011059761, "learning_rate": 6.329133016668815e-07, "loss": 0.0003, "step": 208840 }, { "epoch": 3.4173279882189314, "grad_norm": 0.019820019602775574, "learning_rate": 6.32565670809161e-07, "loss": 0.0003, "step": 208850 }, { "epoch": 3.4174916141700074, "grad_norm": 0.024465007707476616, "learning_rate": 6.322181289997131e-07, "loss": 0.0005, "step": 208860 }, { "epoch": 3.4176552401210833, "grad_norm": 0.01902233622968197, "learning_rate": 6.318706762456223e-07, "loss": 0.0002, "step": 208870 }, { "epoch": 3.417818866072159, "grad_norm": 0.028260691091418266, "learning_rate": 6.315233125539733e-07, "loss": 0.0002, "step": 208880 }, { "epoch": 3.417982492023235, "grad_norm": 0.009858756326138973, "learning_rate": 6.311760379318483e-07, "loss": 0.0001, "step": 208890 }, { "epoch": 3.418146117974311, "grad_norm": 0.0392763689160347, "learning_rate": 6.308288523863298e-07, "loss": 0.0004, "step": 208900 }, { "epoch": 3.4183097439253864, "grad_norm": 0.025180809199810028, "learning_rate": 6.304817559244941e-07, "loss": 0.0002, "step": 208910 }, { "epoch": 3.4184733698764624, "grad_norm": 0.004567994736135006, "learning_rate": 6.301347485534193e-07, "loss": 0.0004, "step": 208920 }, { "epoch": 3.4186369958275384, "grad_norm": 0.010188722051680088, "learning_rate": 6.297878302801802e-07, "loss": 0.0001, "step": 208930 }, { "epoch": 3.418800621778614, "grad_norm": 0.009594942443072796, "learning_rate": 6.294410011118518e-07, "loss": 0.0007, "step": 208940 }, { "epoch": 3.41896424772969, "grad_norm": 0.03571736067533493, "learning_rate": 6.290942610555034e-07, "loss": 0.0002, "step": 208950 }, { "epoch": 3.4191278736807655, "grad_norm": 0.027623120695352554, "learning_rate": 6.287476101182055e-07, "loss": 0.0002, "step": 208960 }, { "epoch": 3.4192914996318415, "grad_norm": 0.031905218958854675, "learning_rate": 6.284010483070263e-07, "loss": 0.0002, "step": 208970 }, { "epoch": 3.4194551255829175, "grad_norm": 0.018799616023898125, "learning_rate": 6.280545756290323e-07, "loss": 0.0006, "step": 208980 }, { "epoch": 3.419618751533993, "grad_norm": 0.0005333409062586725, "learning_rate": 6.27708192091287e-07, "loss": 0.0006, "step": 208990 }, { "epoch": 3.419782377485069, "grad_norm": 0.054635196924209595, "learning_rate": 6.273618977008527e-07, "loss": 0.0022, "step": 209000 }, { "epoch": 3.419946003436145, "grad_norm": 0.006548465695232153, "learning_rate": 6.270156924647908e-07, "loss": 0.0023, "step": 209010 }, { "epoch": 3.4201096293872206, "grad_norm": 0.06303326785564423, "learning_rate": 6.266695763901609e-07, "loss": 0.0003, "step": 209020 }, { "epoch": 3.4202732553382966, "grad_norm": 0.04876498878002167, "learning_rate": 6.26323549484018e-07, "loss": 0.0002, "step": 209030 }, { "epoch": 3.4204368812893726, "grad_norm": 0.020977353677153587, "learning_rate": 6.259776117534177e-07, "loss": 0.0002, "step": 209040 }, { "epoch": 3.420600507240448, "grad_norm": 0.0018887465121224523, "learning_rate": 6.256317632054148e-07, "loss": 0.0004, "step": 209050 }, { "epoch": 3.420764133191524, "grad_norm": 0.01848825253546238, "learning_rate": 6.252860038470604e-07, "loss": 0.0002, "step": 209060 }, { "epoch": 3.4209277591426, "grad_norm": 0.03946329280734062, "learning_rate": 6.249403336854027e-07, "loss": 0.0004, "step": 209070 }, { "epoch": 3.4210913850936757, "grad_norm": 0.04684070125222206, "learning_rate": 6.245947527274921e-07, "loss": 0.0006, "step": 209080 }, { "epoch": 3.4212550110447517, "grad_norm": 0.0006143232458271086, "learning_rate": 6.242492609803713e-07, "loss": 0.0002, "step": 209090 }, { "epoch": 3.4214186369958277, "grad_norm": 0.00394106050953269, "learning_rate": 6.239038584510887e-07, "loss": 0.0002, "step": 209100 }, { "epoch": 3.4215822629469033, "grad_norm": 0.021571578457951546, "learning_rate": 6.23558545146683e-07, "loss": 0.0002, "step": 209110 }, { "epoch": 3.4217458888979793, "grad_norm": 0.036584414541721344, "learning_rate": 6.232133210741987e-07, "loss": 0.0001, "step": 209120 }, { "epoch": 3.4219095148490553, "grad_norm": 0.005855824798345566, "learning_rate": 6.228681862406694e-07, "loss": 0.0003, "step": 209130 }, { "epoch": 3.422073140800131, "grad_norm": 0.010801836848258972, "learning_rate": 6.22523140653138e-07, "loss": 0.0003, "step": 209140 }, { "epoch": 3.422236766751207, "grad_norm": 0.004894720856100321, "learning_rate": 6.221781843186353e-07, "loss": 0.0001, "step": 209150 }, { "epoch": 3.422400392702283, "grad_norm": 0.0702986866235733, "learning_rate": 6.21833317244197e-07, "loss": 0.0003, "step": 209160 }, { "epoch": 3.4225640186533584, "grad_norm": 0.013255827128887177, "learning_rate": 6.214885394368525e-07, "loss": 0.0001, "step": 209170 }, { "epoch": 3.4227276446044343, "grad_norm": 0.026504067704081535, "learning_rate": 6.211438509036343e-07, "loss": 0.0002, "step": 209180 }, { "epoch": 3.4228912705555103, "grad_norm": 0.02346608228981495, "learning_rate": 6.207992516515682e-07, "loss": 0.0002, "step": 209190 }, { "epoch": 3.423054896506586, "grad_norm": 0.010226362384855747, "learning_rate": 6.204547416876822e-07, "loss": 0.0002, "step": 209200 }, { "epoch": 3.423218522457662, "grad_norm": 0.006007370539009571, "learning_rate": 6.201103210189968e-07, "loss": 0.0005, "step": 209210 }, { "epoch": 3.4233821484087374, "grad_norm": 0.00017931235197465867, "learning_rate": 6.197659896525393e-07, "loss": 0.0001, "step": 209220 }, { "epoch": 3.4235457743598134, "grad_norm": 0.10084355622529984, "learning_rate": 6.194217475953273e-07, "loss": 0.0006, "step": 209230 }, { "epoch": 3.4237094003108894, "grad_norm": 0.03500344231724739, "learning_rate": 6.190775948543809e-07, "loss": 0.0005, "step": 209240 }, { "epoch": 3.423873026261965, "grad_norm": 0.06204846501350403, "learning_rate": 6.187335314367149e-07, "loss": 0.0003, "step": 209250 }, { "epoch": 3.424036652213041, "grad_norm": 0.021096210926771164, "learning_rate": 6.183895573493476e-07, "loss": 0.0004, "step": 209260 }, { "epoch": 3.424200278164117, "grad_norm": 0.024706825613975525, "learning_rate": 6.1804567259929e-07, "loss": 0.0001, "step": 209270 }, { "epoch": 3.4243639041151925, "grad_norm": 0.06365619599819183, "learning_rate": 6.177018771935556e-07, "loss": 0.0003, "step": 209280 }, { "epoch": 3.4245275300662685, "grad_norm": 0.0030868526082485914, "learning_rate": 6.173581711391518e-07, "loss": 0.0006, "step": 209290 }, { "epoch": 3.4246911560173445, "grad_norm": 0.03625478222966194, "learning_rate": 6.170145544430872e-07, "loss": 0.0003, "step": 209300 }, { "epoch": 3.42485478196842, "grad_norm": 0.0003485912748146802, "learning_rate": 6.166710271123688e-07, "loss": 0.0002, "step": 209310 }, { "epoch": 3.425018407919496, "grad_norm": 0.05375928804278374, "learning_rate": 6.16327589154001e-07, "loss": 0.0003, "step": 209320 }, { "epoch": 3.4251820338705716, "grad_norm": 0.04318246990442276, "learning_rate": 6.159842405749844e-07, "loss": 0.0003, "step": 209330 }, { "epoch": 3.4253456598216476, "grad_norm": 0.03920181095600128, "learning_rate": 6.156409813823211e-07, "loss": 0.0002, "step": 209340 }, { "epoch": 3.4255092857727236, "grad_norm": 0.02863253839313984, "learning_rate": 6.152978115830088e-07, "loss": 0.0009, "step": 209350 }, { "epoch": 3.425672911723799, "grad_norm": 0.10040168464183807, "learning_rate": 6.149547311840464e-07, "loss": 0.0002, "step": 209360 }, { "epoch": 3.425836537674875, "grad_norm": 0.021468371152877808, "learning_rate": 6.146117401924268e-07, "loss": 0.0002, "step": 209370 }, { "epoch": 3.426000163625951, "grad_norm": 0.04603024199604988, "learning_rate": 6.14268838615144e-07, "loss": 0.0002, "step": 209380 }, { "epoch": 3.4261637895770267, "grad_norm": 0.003881183685734868, "learning_rate": 6.139260264591895e-07, "loss": 0.0004, "step": 209390 }, { "epoch": 3.4263274155281027, "grad_norm": 0.007136349566280842, "learning_rate": 6.135833037315541e-07, "loss": 0.0001, "step": 209400 }, { "epoch": 3.4264910414791787, "grad_norm": 0.015431234613060951, "learning_rate": 6.132406704392241e-07, "loss": 0.0001, "step": 209410 }, { "epoch": 3.4266546674302543, "grad_norm": 0.0009760551620274782, "learning_rate": 6.128981265891859e-07, "loss": 0.0003, "step": 209420 }, { "epoch": 3.4268182933813303, "grad_norm": 0.0609733946621418, "learning_rate": 6.125556721884235e-07, "loss": 0.0003, "step": 209430 }, { "epoch": 3.4269819193324063, "grad_norm": 0.03240757808089256, "learning_rate": 6.122133072439207e-07, "loss": 0.0001, "step": 209440 }, { "epoch": 3.427145545283482, "grad_norm": 0.0032745408825576305, "learning_rate": 6.118710317626558e-07, "loss": 0.0004, "step": 209450 }, { "epoch": 3.427309171234558, "grad_norm": 0.05827983468770981, "learning_rate": 6.115288457516089e-07, "loss": 0.0005, "step": 209460 }, { "epoch": 3.427472797185634, "grad_norm": 0.00035333988489583135, "learning_rate": 6.111867492177559e-07, "loss": 0.0004, "step": 209470 }, { "epoch": 3.4276364231367094, "grad_norm": 0.0182361900806427, "learning_rate": 6.108447421680741e-07, "loss": 0.0004, "step": 209480 }, { "epoch": 3.4278000490877854, "grad_norm": 0.039596933871507645, "learning_rate": 6.10502824609534e-07, "loss": 0.0006, "step": 209490 }, { "epoch": 3.4279636750388613, "grad_norm": 0.004511293955147266, "learning_rate": 6.101609965491084e-07, "loss": 0.0003, "step": 209500 }, { "epoch": 3.428127300989937, "grad_norm": 0.09963662177324295, "learning_rate": 6.098192579937662e-07, "loss": 0.0004, "step": 209510 }, { "epoch": 3.428290926941013, "grad_norm": 0.0207669660449028, "learning_rate": 6.094776089504773e-07, "loss": 0.0005, "step": 209520 }, { "epoch": 3.428454552892089, "grad_norm": 0.09692542254924774, "learning_rate": 6.091360494262045e-07, "loss": 0.0004, "step": 209530 }, { "epoch": 3.4286181788431644, "grad_norm": 0.04917026311159134, "learning_rate": 6.087945794279132e-07, "loss": 0.0002, "step": 209540 }, { "epoch": 3.4287818047942404, "grad_norm": 0.0007584269624203444, "learning_rate": 6.084531989625658e-07, "loss": 0.0002, "step": 209550 }, { "epoch": 3.4289454307453164, "grad_norm": 0.0015211295103654265, "learning_rate": 6.081119080371239e-07, "loss": 0.0002, "step": 209560 }, { "epoch": 3.429109056696392, "grad_norm": 0.04477450251579285, "learning_rate": 6.077707066585443e-07, "loss": 0.0003, "step": 209570 }, { "epoch": 3.429272682647468, "grad_norm": 0.022352920845150948, "learning_rate": 6.074295948337844e-07, "loss": 0.0002, "step": 209580 }, { "epoch": 3.4294363085985435, "grad_norm": 0.031172601506114006, "learning_rate": 6.070885725697995e-07, "loss": 0.0003, "step": 209590 }, { "epoch": 3.4295999345496195, "grad_norm": 0.05115106701850891, "learning_rate": 6.067476398735433e-07, "loss": 0.0004, "step": 209600 }, { "epoch": 3.4297635605006955, "grad_norm": 0.04213026165962219, "learning_rate": 6.064067967519655e-07, "loss": 0.0003, "step": 209610 }, { "epoch": 3.429927186451771, "grad_norm": 5.6316112022614107e-05, "learning_rate": 6.060660432120163e-07, "loss": 0.0002, "step": 209620 }, { "epoch": 3.430090812402847, "grad_norm": 0.009098827838897705, "learning_rate": 6.057253792606438e-07, "loss": 0.0001, "step": 209630 }, { "epoch": 3.430254438353923, "grad_norm": 0.08564113825559616, "learning_rate": 6.053848049047944e-07, "loss": 0.0002, "step": 209640 }, { "epoch": 3.4304180643049986, "grad_norm": 0.014425351284444332, "learning_rate": 6.050443201514106e-07, "loss": 0.0001, "step": 209650 }, { "epoch": 3.4305816902560746, "grad_norm": 0.02707282267510891, "learning_rate": 6.047039250074354e-07, "loss": 0.0001, "step": 209660 }, { "epoch": 3.4307453162071506, "grad_norm": 0.02386518195271492, "learning_rate": 6.043636194798086e-07, "loss": 0.0004, "step": 209670 }, { "epoch": 3.430908942158226, "grad_norm": 0.00035339841269887984, "learning_rate": 6.040234035754705e-07, "loss": 0.0002, "step": 209680 }, { "epoch": 3.431072568109302, "grad_norm": 0.03861990571022034, "learning_rate": 6.036832773013556e-07, "loss": 0.0003, "step": 209690 }, { "epoch": 3.431236194060378, "grad_norm": 0.012351381592452526, "learning_rate": 6.033432406644002e-07, "loss": 0.0003, "step": 209700 }, { "epoch": 3.4313998200114537, "grad_norm": 0.002739050192758441, "learning_rate": 6.030032936715363e-07, "loss": 0.0005, "step": 209710 }, { "epoch": 3.4315634459625297, "grad_norm": 0.04431096464395523, "learning_rate": 6.026634363296968e-07, "loss": 0.0002, "step": 209720 }, { "epoch": 3.4317270719136053, "grad_norm": 0.012202462181448936, "learning_rate": 6.023236686458095e-07, "loss": 0.0002, "step": 209730 }, { "epoch": 3.4318906978646813, "grad_norm": 0.0006576546002179384, "learning_rate": 6.019839906268026e-07, "loss": 0.0003, "step": 209740 }, { "epoch": 3.4320543238157573, "grad_norm": 0.00871169101446867, "learning_rate": 6.016444022796003e-07, "loss": 0.0005, "step": 209750 }, { "epoch": 3.432217949766833, "grad_norm": 0.07234332710504532, "learning_rate": 6.013049036111301e-07, "loss": 0.0006, "step": 209760 }, { "epoch": 3.432381575717909, "grad_norm": 0.008651980198919773, "learning_rate": 6.009654946283111e-07, "loss": 0.0002, "step": 209770 }, { "epoch": 3.432545201668985, "grad_norm": 0.00026091665495187044, "learning_rate": 6.006261753380649e-07, "loss": 0.0005, "step": 209780 }, { "epoch": 3.4327088276200604, "grad_norm": 0.004479028284549713, "learning_rate": 6.002869457473076e-07, "loss": 0.0001, "step": 209790 }, { "epoch": 3.4328724535711364, "grad_norm": 0.013368639163672924, "learning_rate": 5.999478058629599e-07, "loss": 0.0001, "step": 209800 }, { "epoch": 3.4330360795222123, "grad_norm": 0.009465069510042667, "learning_rate": 5.996087556919333e-07, "loss": 0.0005, "step": 209810 }, { "epoch": 3.433199705473288, "grad_norm": 0.0003950783866457641, "learning_rate": 5.992697952411425e-07, "loss": 0.0004, "step": 209820 }, { "epoch": 3.433363331424364, "grad_norm": 0.04112354293465614, "learning_rate": 5.989309245174962e-07, "loss": 0.0001, "step": 209830 }, { "epoch": 3.43352695737544, "grad_norm": 0.10654153674840927, "learning_rate": 5.985921435279069e-07, "loss": 0.0004, "step": 209840 }, { "epoch": 3.4336905833265154, "grad_norm": 0.13447105884552002, "learning_rate": 5.982534522792799e-07, "loss": 0.0003, "step": 209850 }, { "epoch": 3.4338542092775914, "grad_norm": 0.004639375489205122, "learning_rate": 5.979148507785226e-07, "loss": 0.0004, "step": 209860 }, { "epoch": 3.4340178352286674, "grad_norm": 0.001391557278111577, "learning_rate": 5.975763390325351e-07, "loss": 0.0002, "step": 209870 }, { "epoch": 3.434181461179743, "grad_norm": 0.02431938610970974, "learning_rate": 5.972379170482245e-07, "loss": 0.0004, "step": 209880 }, { "epoch": 3.434345087130819, "grad_norm": 0.034717023372650146, "learning_rate": 5.968995848324871e-07, "loss": 0.0002, "step": 209890 }, { "epoch": 3.434508713081895, "grad_norm": 0.02490292489528656, "learning_rate": 5.965613423922234e-07, "loss": 0.0002, "step": 209900 }, { "epoch": 3.4346723390329705, "grad_norm": 0.0885215699672699, "learning_rate": 5.962231897343285e-07, "loss": 0.0004, "step": 209910 }, { "epoch": 3.4348359649840465, "grad_norm": 0.0012924027396366, "learning_rate": 5.958851268656968e-07, "loss": 0.0001, "step": 209920 }, { "epoch": 3.4349995909351225, "grad_norm": 0.05892471596598625, "learning_rate": 5.955471537932217e-07, "loss": 0.0004, "step": 209930 }, { "epoch": 3.435163216886198, "grad_norm": 0.0007673886138945818, "learning_rate": 5.952092705237961e-07, "loss": 0.0001, "step": 209940 }, { "epoch": 3.435326842837274, "grad_norm": 0.00690271845087409, "learning_rate": 5.948714770643055e-07, "loss": 0.0001, "step": 209950 }, { "epoch": 3.43549046878835, "grad_norm": 0.024210302159190178, "learning_rate": 5.945337734216394e-07, "loss": 0.0002, "step": 209960 }, { "epoch": 3.4356540947394256, "grad_norm": 0.04745052754878998, "learning_rate": 5.94196159602683e-07, "loss": 0.0004, "step": 209970 }, { "epoch": 3.4358177206905016, "grad_norm": 0.0395248681306839, "learning_rate": 5.938586356143205e-07, "loss": 0.0003, "step": 209980 }, { "epoch": 3.435981346641577, "grad_norm": 0.009749878197908401, "learning_rate": 5.935212014634328e-07, "loss": 0.0004, "step": 209990 }, { "epoch": 3.436144972592653, "grad_norm": 0.05597398430109024, "learning_rate": 5.931838571568998e-07, "loss": 0.0004, "step": 210000 }, { "epoch": 3.436308598543729, "grad_norm": 0.03207285329699516, "learning_rate": 5.928466027016e-07, "loss": 0.0001, "step": 210010 }, { "epoch": 3.4364722244948047, "grad_norm": 0.00430568540468812, "learning_rate": 5.925094381044111e-07, "loss": 0.0001, "step": 210020 }, { "epoch": 3.4366358504458807, "grad_norm": 0.06530380994081497, "learning_rate": 5.921723633722054e-07, "loss": 0.0006, "step": 210030 }, { "epoch": 3.4367994763969567, "grad_norm": 0.0004700414719991386, "learning_rate": 5.918353785118564e-07, "loss": 0.0002, "step": 210040 }, { "epoch": 3.4369631023480323, "grad_norm": 0.052105411887168884, "learning_rate": 5.914984835302346e-07, "loss": 0.0005, "step": 210050 }, { "epoch": 3.4371267282991083, "grad_norm": 0.009087633341550827, "learning_rate": 5.911616784342112e-07, "loss": 0.0001, "step": 210060 }, { "epoch": 3.4372903542501843, "grad_norm": 0.09024600684642792, "learning_rate": 5.908249632306501e-07, "loss": 0.0004, "step": 210070 }, { "epoch": 3.43745398020126, "grad_norm": 0.027964293956756592, "learning_rate": 5.904883379264181e-07, "loss": 0.0003, "step": 210080 }, { "epoch": 3.437617606152336, "grad_norm": 0.00954179372638464, "learning_rate": 5.901518025283787e-07, "loss": 0.0003, "step": 210090 }, { "epoch": 3.4377812321034114, "grad_norm": 0.03989096358418465, "learning_rate": 5.898153570433945e-07, "loss": 0.0004, "step": 210100 }, { "epoch": 3.4379448580544874, "grad_norm": 0.13851036131381989, "learning_rate": 5.894790014783236e-07, "loss": 0.0002, "step": 210110 }, { "epoch": 3.4381084840055633, "grad_norm": 0.040422309190034866, "learning_rate": 5.891427358400248e-07, "loss": 0.0003, "step": 210120 }, { "epoch": 3.438272109956639, "grad_norm": 0.23746412992477417, "learning_rate": 5.888065601353549e-07, "loss": 0.0006, "step": 210130 }, { "epoch": 3.438435735907715, "grad_norm": 0.010969935916364193, "learning_rate": 5.884704743711678e-07, "loss": 0.0002, "step": 210140 }, { "epoch": 3.438599361858791, "grad_norm": 0.004495489411056042, "learning_rate": 5.881344785543153e-07, "loss": 0.0003, "step": 210150 }, { "epoch": 3.4387629878098664, "grad_norm": 0.009002789855003357, "learning_rate": 5.877985726916485e-07, "loss": 0.0003, "step": 210160 }, { "epoch": 3.4389266137609424, "grad_norm": 0.1402270793914795, "learning_rate": 5.874627567900165e-07, "loss": 0.0003, "step": 210170 }, { "epoch": 3.4390902397120184, "grad_norm": 0.0007082195370458066, "learning_rate": 5.87127030856267e-07, "loss": 0.0001, "step": 210180 }, { "epoch": 3.439253865663094, "grad_norm": 0.012576931156218052, "learning_rate": 5.867913948972436e-07, "loss": 0.0003, "step": 210190 }, { "epoch": 3.43941749161417, "grad_norm": 0.0034643434919416904, "learning_rate": 5.864558489197903e-07, "loss": 0.0003, "step": 210200 }, { "epoch": 3.439581117565246, "grad_norm": 0.03601335734128952, "learning_rate": 5.861203929307486e-07, "loss": 0.0003, "step": 210210 }, { "epoch": 3.4397447435163215, "grad_norm": 0.05267283692955971, "learning_rate": 5.857850269369586e-07, "loss": 0.0004, "step": 210220 }, { "epoch": 3.4399083694673975, "grad_norm": 0.05117533728480339, "learning_rate": 5.854497509452572e-07, "loss": 0.0004, "step": 210230 }, { "epoch": 3.4400719954184735, "grad_norm": 0.04277999326586723, "learning_rate": 5.851145649624812e-07, "loss": 0.0002, "step": 210240 }, { "epoch": 3.440235621369549, "grad_norm": 0.008886545896530151, "learning_rate": 5.847794689954644e-07, "loss": 0.0002, "step": 210250 }, { "epoch": 3.440399247320625, "grad_norm": 0.01298919040709734, "learning_rate": 5.844444630510404e-07, "loss": 0.0001, "step": 210260 }, { "epoch": 3.440562873271701, "grad_norm": 0.050064168870449066, "learning_rate": 5.841095471360369e-07, "loss": 0.0002, "step": 210270 }, { "epoch": 3.4407264992227766, "grad_norm": 0.04658360406756401, "learning_rate": 5.837747212572847e-07, "loss": 0.0004, "step": 210280 }, { "epoch": 3.4408901251738526, "grad_norm": 0.019180364906787872, "learning_rate": 5.8343998542161e-07, "loss": 0.0002, "step": 210290 }, { "epoch": 3.4410537511249286, "grad_norm": 0.013723431155085564, "learning_rate": 5.831053396358383e-07, "loss": 0.0005, "step": 210300 }, { "epoch": 3.441217377076004, "grad_norm": 0.016402684152126312, "learning_rate": 5.82770783906792e-07, "loss": 0.0003, "step": 210310 }, { "epoch": 3.44138100302708, "grad_norm": 0.010837053880095482, "learning_rate": 5.824363182412923e-07, "loss": 0.0006, "step": 210320 }, { "epoch": 3.441544628978156, "grad_norm": 0.03632795438170433, "learning_rate": 5.821019426461588e-07, "loss": 0.0008, "step": 210330 }, { "epoch": 3.4417082549292317, "grad_norm": 0.00016763029270805418, "learning_rate": 5.81767657128211e-07, "loss": 0.0002, "step": 210340 }, { "epoch": 3.4418718808803077, "grad_norm": 0.04968835413455963, "learning_rate": 5.814334616942619e-07, "loss": 0.0002, "step": 210350 }, { "epoch": 3.4420355068313837, "grad_norm": 0.00011367411934770644, "learning_rate": 5.810993563511263e-07, "loss": 0.0003, "step": 210360 }, { "epoch": 3.4421991327824593, "grad_norm": 0.006906391587108374, "learning_rate": 5.807653411056175e-07, "loss": 0.0001, "step": 210370 }, { "epoch": 3.4423627587335353, "grad_norm": 0.07721784710884094, "learning_rate": 5.804314159645452e-07, "loss": 0.0005, "step": 210380 }, { "epoch": 3.442526384684611, "grad_norm": 0.002146328566595912, "learning_rate": 5.80097580934717e-07, "loss": 0.0001, "step": 210390 }, { "epoch": 3.442690010635687, "grad_norm": 0.038973644375801086, "learning_rate": 5.797638360229413e-07, "loss": 0.0002, "step": 210400 }, { "epoch": 3.442853636586763, "grad_norm": 0.045535799115896225, "learning_rate": 5.794301812360193e-07, "loss": 0.0002, "step": 210410 }, { "epoch": 3.4430172625378384, "grad_norm": 0.03439902886748314, "learning_rate": 5.790966165807582e-07, "loss": 0.0006, "step": 210420 }, { "epoch": 3.4431808884889143, "grad_norm": 0.017297344282269478, "learning_rate": 5.787631420639567e-07, "loss": 0.0001, "step": 210430 }, { "epoch": 3.4433445144399903, "grad_norm": 0.020657969638705254, "learning_rate": 5.784297576924152e-07, "loss": 0.0001, "step": 210440 }, { "epoch": 3.443508140391066, "grad_norm": 0.027389859780669212, "learning_rate": 5.780964634729291e-07, "loss": 0.0003, "step": 210450 }, { "epoch": 3.443671766342142, "grad_norm": 0.010628700256347656, "learning_rate": 5.777632594122967e-07, "loss": 0.0002, "step": 210460 }, { "epoch": 3.443835392293218, "grad_norm": 0.054454151540994644, "learning_rate": 5.774301455173098e-07, "loss": 0.0003, "step": 210470 }, { "epoch": 3.4439990182442934, "grad_norm": 0.009929743595421314, "learning_rate": 5.770971217947624e-07, "loss": 0.0003, "step": 210480 }, { "epoch": 3.4441626441953694, "grad_norm": 0.08972752094268799, "learning_rate": 5.767641882514407e-07, "loss": 0.0003, "step": 210490 }, { "epoch": 3.444326270146445, "grad_norm": 0.005530097987502813, "learning_rate": 5.764313448941372e-07, "loss": 0.0003, "step": 210500 }, { "epoch": 3.444489896097521, "grad_norm": 0.0006015151739120483, "learning_rate": 5.760985917296358e-07, "loss": 0.0002, "step": 210510 }, { "epoch": 3.444653522048597, "grad_norm": 0.05393563210964203, "learning_rate": 5.757659287647227e-07, "loss": 0.0002, "step": 210520 }, { "epoch": 3.4448171479996725, "grad_norm": 0.011563285253942013, "learning_rate": 5.754333560061775e-07, "loss": 0.0003, "step": 210530 }, { "epoch": 3.4449807739507485, "grad_norm": 0.03670826554298401, "learning_rate": 5.751008734607855e-07, "loss": 0.0003, "step": 210540 }, { "epoch": 3.4451443999018245, "grad_norm": 0.05091220512986183, "learning_rate": 5.747684811353221e-07, "loss": 0.0002, "step": 210550 }, { "epoch": 3.4453080258529, "grad_norm": 0.036818280816078186, "learning_rate": 5.744361790365665e-07, "loss": 0.0002, "step": 210560 }, { "epoch": 3.445471651803976, "grad_norm": 0.0769772157073021, "learning_rate": 5.741039671712934e-07, "loss": 0.0004, "step": 210570 }, { "epoch": 3.445635277755052, "grad_norm": 0.0014631547965109348, "learning_rate": 5.737718455462754e-07, "loss": 0.0003, "step": 210580 }, { "epoch": 3.4457989037061276, "grad_norm": 0.008351802825927734, "learning_rate": 5.734398141682856e-07, "loss": 0.0002, "step": 210590 }, { "epoch": 3.4459625296572036, "grad_norm": 0.06380392611026764, "learning_rate": 5.731078730440942e-07, "loss": 0.0001, "step": 210600 }, { "epoch": 3.4461261556082796, "grad_norm": 0.03142261132597923, "learning_rate": 5.727760221804674e-07, "loss": 0.0004, "step": 210610 }, { "epoch": 3.446289781559355, "grad_norm": 0.023216452449560165, "learning_rate": 5.724442615841719e-07, "loss": 0.0002, "step": 210620 }, { "epoch": 3.446453407510431, "grad_norm": 0.006398578640073538, "learning_rate": 5.721125912619729e-07, "loss": 0.0004, "step": 210630 }, { "epoch": 3.446617033461507, "grad_norm": 0.03274314105510712, "learning_rate": 5.717810112206335e-07, "loss": 0.0003, "step": 210640 }, { "epoch": 3.4467806594125827, "grad_norm": 0.034092411398887634, "learning_rate": 5.714495214669119e-07, "loss": 0.0002, "step": 210650 }, { "epoch": 3.4469442853636587, "grad_norm": 0.013747761957347393, "learning_rate": 5.711181220075679e-07, "loss": 0.0007, "step": 210660 }, { "epoch": 3.4471079113147347, "grad_norm": 0.027068009600043297, "learning_rate": 5.707868128493593e-07, "loss": 0.0003, "step": 210670 }, { "epoch": 3.4472715372658103, "grad_norm": 0.0009227448608726263, "learning_rate": 5.704555939990414e-07, "loss": 0.0004, "step": 210680 }, { "epoch": 3.4474351632168863, "grad_norm": 0.06760137528181076, "learning_rate": 5.701244654633658e-07, "loss": 0.0003, "step": 210690 }, { "epoch": 3.4475987891679623, "grad_norm": 0.034592192620038986, "learning_rate": 5.697934272490852e-07, "loss": 0.0005, "step": 210700 }, { "epoch": 3.447762415119038, "grad_norm": 0.017870476469397545, "learning_rate": 5.694624793629483e-07, "loss": 0.0004, "step": 210710 }, { "epoch": 3.447926041070114, "grad_norm": 0.012628138065338135, "learning_rate": 5.691316218117049e-07, "loss": 0.0002, "step": 210720 }, { "epoch": 3.44808966702119, "grad_norm": 0.04727257415652275, "learning_rate": 5.688008546020979e-07, "loss": 0.0002, "step": 210730 }, { "epoch": 3.4482532929722653, "grad_norm": 0.05999002978205681, "learning_rate": 5.68470177740873e-07, "loss": 0.0002, "step": 210740 }, { "epoch": 3.4484169189233413, "grad_norm": 0.0009215184836648405, "learning_rate": 5.681395912347726e-07, "loss": 0.0004, "step": 210750 }, { "epoch": 3.448580544874417, "grad_norm": 0.014416991733014584, "learning_rate": 5.678090950905379e-07, "loss": 0.0002, "step": 210760 }, { "epoch": 3.448744170825493, "grad_norm": 0.02166523039340973, "learning_rate": 5.674786893149048e-07, "loss": 0.0003, "step": 210770 }, { "epoch": 3.448907796776569, "grad_norm": 0.04144709184765816, "learning_rate": 5.671483739146116e-07, "loss": 0.0003, "step": 210780 }, { "epoch": 3.4490714227276444, "grad_norm": 0.062345944344997406, "learning_rate": 5.66818148896393e-07, "loss": 0.0006, "step": 210790 }, { "epoch": 3.4492350486787204, "grad_norm": 0.0007056629401631653, "learning_rate": 5.664880142669838e-07, "loss": 0.0002, "step": 210800 }, { "epoch": 3.4493986746297964, "grad_norm": 0.017795342952013016, "learning_rate": 5.661579700331116e-07, "loss": 0.0001, "step": 210810 }, { "epoch": 3.449562300580872, "grad_norm": 0.06795110553503036, "learning_rate": 5.658280162015078e-07, "loss": 0.0005, "step": 210820 }, { "epoch": 3.449725926531948, "grad_norm": 0.055311642587184906, "learning_rate": 5.654981527788994e-07, "loss": 0.0005, "step": 210830 }, { "epoch": 3.449889552483024, "grad_norm": 0.019208980724215508, "learning_rate": 5.651683797720136e-07, "loss": 0.0002, "step": 210840 }, { "epoch": 3.4500531784340995, "grad_norm": 0.018939776346087456, "learning_rate": 5.648386971875719e-07, "loss": 0.0003, "step": 210850 }, { "epoch": 3.4502168043851755, "grad_norm": 0.01102500781416893, "learning_rate": 5.645091050322971e-07, "loss": 0.0001, "step": 210860 }, { "epoch": 3.450380430336251, "grad_norm": 0.01774534210562706, "learning_rate": 5.641796033129094e-07, "loss": 0.0002, "step": 210870 }, { "epoch": 3.450544056287327, "grad_norm": 0.05550270900130272, "learning_rate": 5.638501920361278e-07, "loss": 0.0003, "step": 210880 }, { "epoch": 3.450707682238403, "grad_norm": 0.032161418348550797, "learning_rate": 5.635208712086671e-07, "loss": 0.0004, "step": 210890 }, { "epoch": 3.4508713081894786, "grad_norm": 0.11101003736257553, "learning_rate": 5.631916408372429e-07, "loss": 0.0002, "step": 210900 }, { "epoch": 3.4510349341405546, "grad_norm": 0.03168974071741104, "learning_rate": 5.628625009285677e-07, "loss": 0.0005, "step": 210910 }, { "epoch": 3.4511985600916306, "grad_norm": 0.05079776793718338, "learning_rate": 5.625334514893532e-07, "loss": 0.0003, "step": 210920 }, { "epoch": 3.451362186042706, "grad_norm": 0.06296057254076004, "learning_rate": 5.622044925263071e-07, "loss": 0.0002, "step": 210930 }, { "epoch": 3.451525811993782, "grad_norm": 0.03325612470507622, "learning_rate": 5.618756240461371e-07, "loss": 0.0002, "step": 210940 }, { "epoch": 3.451689437944858, "grad_norm": 0.013155153021216393, "learning_rate": 5.615468460555485e-07, "loss": 0.0002, "step": 210950 }, { "epoch": 3.4518530638959337, "grad_norm": 0.003345748409628868, "learning_rate": 5.612181585612459e-07, "loss": 0.0006, "step": 210960 }, { "epoch": 3.4520166898470097, "grad_norm": 0.0686645582318306, "learning_rate": 5.608895615699289e-07, "loss": 0.0004, "step": 210970 }, { "epoch": 3.4521803157980857, "grad_norm": 0.026050379499793053, "learning_rate": 5.605610550882989e-07, "loss": 0.0002, "step": 210980 }, { "epoch": 3.4523439417491613, "grad_norm": 0.08035518229007721, "learning_rate": 5.602326391230528e-07, "loss": 0.0001, "step": 210990 }, { "epoch": 3.4525075677002373, "grad_norm": 0.002204400021582842, "learning_rate": 5.59904313680889e-07, "loss": 0.0, "step": 211000 }, { "epoch": 3.4526711936513133, "grad_norm": 0.016341185197234154, "learning_rate": 5.595760787684984e-07, "loss": 0.0002, "step": 211010 }, { "epoch": 3.452834819602389, "grad_norm": 0.00932271033525467, "learning_rate": 5.592479343925766e-07, "loss": 0.0002, "step": 211020 }, { "epoch": 3.452998445553465, "grad_norm": 0.051973868161439896, "learning_rate": 5.589198805598101e-07, "loss": 0.0003, "step": 211030 }, { "epoch": 3.453162071504541, "grad_norm": 0.050416652113199234, "learning_rate": 5.585919172768928e-07, "loss": 0.0022, "step": 211040 }, { "epoch": 3.4533256974556163, "grad_norm": 0.00979676190763712, "learning_rate": 5.582640445505077e-07, "loss": 0.0004, "step": 211050 }, { "epoch": 3.4534893234066923, "grad_norm": 0.040039803832769394, "learning_rate": 5.579362623873425e-07, "loss": 0.0005, "step": 211060 }, { "epoch": 3.4536529493577683, "grad_norm": 0.04682520776987076, "learning_rate": 5.576085707940765e-07, "loss": 0.0003, "step": 211070 }, { "epoch": 3.453816575308844, "grad_norm": 0.009415283799171448, "learning_rate": 5.572809697773962e-07, "loss": 0.0002, "step": 211080 }, { "epoch": 3.45398020125992, "grad_norm": 0.003919039852917194, "learning_rate": 5.569534593439774e-07, "loss": 0.0002, "step": 211090 }, { "epoch": 3.454143827210996, "grad_norm": 0.013394911773502827, "learning_rate": 5.566260395004996e-07, "loss": 0.0001, "step": 211100 }, { "epoch": 3.4543074531620714, "grad_norm": 0.017815493047237396, "learning_rate": 5.56298710253636e-07, "loss": 0.0004, "step": 211110 }, { "epoch": 3.4544710791131474, "grad_norm": 0.06118544191122055, "learning_rate": 5.559714716100645e-07, "loss": 0.0004, "step": 211120 }, { "epoch": 3.4546347050642234, "grad_norm": 0.026866890490055084, "learning_rate": 5.556443235764547e-07, "loss": 0.0004, "step": 211130 }, { "epoch": 3.454798331015299, "grad_norm": 0.04615696519613266, "learning_rate": 5.553172661594774e-07, "loss": 0.0003, "step": 211140 }, { "epoch": 3.454961956966375, "grad_norm": 0.007045072969049215, "learning_rate": 5.549902993657996e-07, "loss": 0.0007, "step": 211150 }, { "epoch": 3.4551255829174505, "grad_norm": 0.00027640999178402126, "learning_rate": 5.546634232020914e-07, "loss": 0.0001, "step": 211160 }, { "epoch": 3.4552892088685265, "grad_norm": 0.0036959985736757517, "learning_rate": 5.543366376750143e-07, "loss": 0.0003, "step": 211170 }, { "epoch": 3.4554528348196025, "grad_norm": 0.08955541253089905, "learning_rate": 5.540099427912326e-07, "loss": 0.0003, "step": 211180 }, { "epoch": 3.455616460770678, "grad_norm": 0.03530428186058998, "learning_rate": 5.536833385574058e-07, "loss": 0.0006, "step": 211190 }, { "epoch": 3.455780086721754, "grad_norm": 7.440004264935851e-05, "learning_rate": 5.533568249801957e-07, "loss": 0.0002, "step": 211200 }, { "epoch": 3.45594371267283, "grad_norm": 0.01834939420223236, "learning_rate": 5.530304020662575e-07, "loss": 0.0003, "step": 211210 }, { "epoch": 3.4561073386239056, "grad_norm": 0.0020580252166837454, "learning_rate": 5.52704069822248e-07, "loss": 0.0003, "step": 211220 }, { "epoch": 3.4562709645749816, "grad_norm": 0.00429062033072114, "learning_rate": 5.523778282548197e-07, "loss": 0.0002, "step": 211230 }, { "epoch": 3.4564345905260576, "grad_norm": 0.019683314487338066, "learning_rate": 5.520516773706247e-07, "loss": 0.0002, "step": 211240 }, { "epoch": 3.456598216477133, "grad_norm": 0.0009452888043597341, "learning_rate": 5.517256171763135e-07, "loss": 0.0006, "step": 211250 }, { "epoch": 3.456761842428209, "grad_norm": 0.034577466547489166, "learning_rate": 5.513996476785344e-07, "loss": 0.0002, "step": 211260 }, { "epoch": 3.4569254683792847, "grad_norm": 0.04842616617679596, "learning_rate": 5.510737688839324e-07, "loss": 0.0003, "step": 211270 }, { "epoch": 3.4570890943303607, "grad_norm": 0.051721032708883286, "learning_rate": 5.507479807991522e-07, "loss": 0.0003, "step": 211280 }, { "epoch": 3.4572527202814367, "grad_norm": 0.02356911636888981, "learning_rate": 5.504222834308371e-07, "loss": 0.0004, "step": 211290 }, { "epoch": 3.4574163462325123, "grad_norm": 0.0013153757899999619, "learning_rate": 5.500966767856281e-07, "loss": 0.0003, "step": 211300 }, { "epoch": 3.4575799721835883, "grad_norm": 0.0015471301740035415, "learning_rate": 5.497711608701622e-07, "loss": 0.0002, "step": 211310 }, { "epoch": 3.4577435981346643, "grad_norm": 0.02342045120894909, "learning_rate": 5.494457356910782e-07, "loss": 0.0002, "step": 211320 }, { "epoch": 3.45790722408574, "grad_norm": 0.011834085918962955, "learning_rate": 5.4912040125501e-07, "loss": 0.0002, "step": 211330 }, { "epoch": 3.458070850036816, "grad_norm": 0.04800507798790932, "learning_rate": 5.487951575685929e-07, "loss": 0.0002, "step": 211340 }, { "epoch": 3.458234475987892, "grad_norm": 0.0261856522411108, "learning_rate": 5.484700046384556e-07, "loss": 0.0003, "step": 211350 }, { "epoch": 3.4583981019389674, "grad_norm": 0.022334830835461617, "learning_rate": 5.481449424712293e-07, "loss": 0.0002, "step": 211360 }, { "epoch": 3.4585617278900433, "grad_norm": 0.029799383133649826, "learning_rate": 5.47819971073541e-07, "loss": 0.0004, "step": 211370 }, { "epoch": 3.4587253538411193, "grad_norm": 0.026124637573957443, "learning_rate": 5.474950904520187e-07, "loss": 0.0001, "step": 211380 }, { "epoch": 3.458888979792195, "grad_norm": 0.017214177176356316, "learning_rate": 5.471703006132834e-07, "loss": 0.0002, "step": 211390 }, { "epoch": 3.459052605743271, "grad_norm": 0.0039091347716748714, "learning_rate": 5.468456015639584e-07, "loss": 0.0003, "step": 211400 }, { "epoch": 3.459216231694347, "grad_norm": 0.0016224345890805125, "learning_rate": 5.465209933106647e-07, "loss": 0.0002, "step": 211410 }, { "epoch": 3.4593798576454224, "grad_norm": 0.004902217537164688, "learning_rate": 5.461964758600213e-07, "loss": 0.0001, "step": 211420 }, { "epoch": 3.4595434835964984, "grad_norm": 0.009332200512290001, "learning_rate": 5.458720492186431e-07, "loss": 0.0002, "step": 211430 }, { "epoch": 3.4597071095475744, "grad_norm": 0.06865382194519043, "learning_rate": 5.455477133931459e-07, "loss": 0.0002, "step": 211440 }, { "epoch": 3.45987073549865, "grad_norm": 0.04037686064839363, "learning_rate": 5.45223468390142e-07, "loss": 0.0002, "step": 211450 }, { "epoch": 3.460034361449726, "grad_norm": 0.04729591682553291, "learning_rate": 5.44899314216244e-07, "loss": 0.0002, "step": 211460 }, { "epoch": 3.460197987400802, "grad_norm": 0.03442462533712387, "learning_rate": 5.445752508780589e-07, "loss": 0.0003, "step": 211470 }, { "epoch": 3.4603616133518775, "grad_norm": 0.0004774231347255409, "learning_rate": 5.442512783821957e-07, "loss": 0.0003, "step": 211480 }, { "epoch": 3.4605252393029535, "grad_norm": 0.006311782635748386, "learning_rate": 5.439273967352587e-07, "loss": 0.0003, "step": 211490 }, { "epoch": 3.4606888652540295, "grad_norm": 0.0018423659494146705, "learning_rate": 5.436036059438543e-07, "loss": 0.0001, "step": 211500 }, { "epoch": 3.460852491205105, "grad_norm": 0.0072029829025268555, "learning_rate": 5.432799060145805e-07, "loss": 0.0001, "step": 211510 }, { "epoch": 3.461016117156181, "grad_norm": 0.005475779063999653, "learning_rate": 5.429562969540398e-07, "loss": 0.0003, "step": 211520 }, { "epoch": 3.4611797431072566, "grad_norm": 0.0033373262267559767, "learning_rate": 5.42632778768829e-07, "loss": 0.0002, "step": 211530 }, { "epoch": 3.4613433690583326, "grad_norm": 0.04071800783276558, "learning_rate": 5.423093514655464e-07, "loss": 0.0003, "step": 211540 }, { "epoch": 3.4615069950094086, "grad_norm": 0.013495358638465405, "learning_rate": 5.419860150507838e-07, "loss": 0.0003, "step": 211550 }, { "epoch": 3.461670620960484, "grad_norm": 0.09132865071296692, "learning_rate": 5.41662769531135e-07, "loss": 0.0007, "step": 211560 }, { "epoch": 3.46183424691156, "grad_norm": 0.03438470885157585, "learning_rate": 5.413396149131905e-07, "loss": 0.0005, "step": 211570 }, { "epoch": 3.461997872862636, "grad_norm": 0.04998163878917694, "learning_rate": 5.41016551203541e-07, "loss": 0.0006, "step": 211580 }, { "epoch": 3.4621614988137117, "grad_norm": 0.007298088166862726, "learning_rate": 5.406935784087702e-07, "loss": 0.0001, "step": 211590 }, { "epoch": 3.4623251247647877, "grad_norm": 0.00131647614762187, "learning_rate": 5.40370696535465e-07, "loss": 0.0005, "step": 211600 }, { "epoch": 3.4624887507158637, "grad_norm": 0.03185822442173958, "learning_rate": 5.40047905590208e-07, "loss": 0.0002, "step": 211610 }, { "epoch": 3.4626523766669393, "grad_norm": 0.004600665532052517, "learning_rate": 5.397252055795827e-07, "loss": 0.0001, "step": 211620 }, { "epoch": 3.4628160026180153, "grad_norm": 0.0024205869995057583, "learning_rate": 5.394025965101657e-07, "loss": 0.0001, "step": 211630 }, { "epoch": 3.4629796285690913, "grad_norm": 0.033041179180145264, "learning_rate": 5.39080078388537e-07, "loss": 0.0005, "step": 211640 }, { "epoch": 3.463143254520167, "grad_norm": 0.011460470035672188, "learning_rate": 5.387576512212706e-07, "loss": 0.0003, "step": 211650 }, { "epoch": 3.463306880471243, "grad_norm": 0.03525212034583092, "learning_rate": 5.384353150149429e-07, "loss": 0.0001, "step": 211660 }, { "epoch": 3.4634705064223184, "grad_norm": 0.09128976613283157, "learning_rate": 5.381130697761239e-07, "loss": 0.0005, "step": 211670 }, { "epoch": 3.4636341323733943, "grad_norm": 0.007660438306629658, "learning_rate": 5.377909155113858e-07, "loss": 0.0002, "step": 211680 }, { "epoch": 3.4637977583244703, "grad_norm": 0.0118141183629632, "learning_rate": 5.374688522272936e-07, "loss": 0.0003, "step": 211690 }, { "epoch": 3.463961384275546, "grad_norm": 0.0764298066496849, "learning_rate": 5.371468799304186e-07, "loss": 0.0003, "step": 211700 }, { "epoch": 3.464125010226622, "grad_norm": 0.0006624922389164567, "learning_rate": 5.368249986273222e-07, "loss": 0.0003, "step": 211710 }, { "epoch": 3.464288636177698, "grad_norm": 0.0008416775381192565, "learning_rate": 5.365032083245691e-07, "loss": 0.0002, "step": 211720 }, { "epoch": 3.4644522621287734, "grad_norm": 0.00030740202055312693, "learning_rate": 5.361815090287176e-07, "loss": 0.0003, "step": 211730 }, { "epoch": 3.4646158880798494, "grad_norm": 0.023819969967007637, "learning_rate": 5.358599007463305e-07, "loss": 0.0003, "step": 211740 }, { "epoch": 3.4647795140309254, "grad_norm": 0.05797642841935158, "learning_rate": 5.355383834839628e-07, "loss": 0.0002, "step": 211750 }, { "epoch": 3.464943139982001, "grad_norm": 0.06081022694706917, "learning_rate": 5.352169572481713e-07, "loss": 0.0006, "step": 211760 }, { "epoch": 3.465106765933077, "grad_norm": 0.030837900936603546, "learning_rate": 5.34895622045507e-07, "loss": 0.0003, "step": 211770 }, { "epoch": 3.465270391884153, "grad_norm": 0.04878269135951996, "learning_rate": 5.345743778825258e-07, "loss": 0.0001, "step": 211780 }, { "epoch": 3.4654340178352285, "grad_norm": 0.03189634904265404, "learning_rate": 5.342532247657739e-07, "loss": 0.0002, "step": 211790 }, { "epoch": 3.4655976437863045, "grad_norm": 0.08218718320131302, "learning_rate": 5.339321627018018e-07, "loss": 0.0003, "step": 211800 }, { "epoch": 3.4657612697373805, "grad_norm": 0.006104086060076952, "learning_rate": 5.336111916971526e-07, "loss": 0.0, "step": 211810 }, { "epoch": 3.465924895688456, "grad_norm": 0.027976106852293015, "learning_rate": 5.332903117583749e-07, "loss": 0.0003, "step": 211820 }, { "epoch": 3.466088521639532, "grad_norm": 0.002129489090293646, "learning_rate": 5.329695228920079e-07, "loss": 0.0001, "step": 211830 }, { "epoch": 3.466252147590608, "grad_norm": 0.05307262763381004, "learning_rate": 5.326488251045941e-07, "loss": 0.0002, "step": 211840 }, { "epoch": 3.4664157735416836, "grad_norm": 0.06144668906927109, "learning_rate": 5.323282184026707e-07, "loss": 0.0004, "step": 211850 }, { "epoch": 3.4665793994927596, "grad_norm": 0.004333742428570986, "learning_rate": 5.32007702792775e-07, "loss": 0.0003, "step": 211860 }, { "epoch": 3.4667430254438356, "grad_norm": 0.0011788535630330443, "learning_rate": 5.316872782814425e-07, "loss": 0.0001, "step": 211870 }, { "epoch": 3.466906651394911, "grad_norm": 0.003036986570805311, "learning_rate": 5.313669448752073e-07, "loss": 0.0001, "step": 211880 }, { "epoch": 3.467070277345987, "grad_norm": 0.08284957706928253, "learning_rate": 5.310467025805987e-07, "loss": 0.0002, "step": 211890 }, { "epoch": 3.467233903297063, "grad_norm": 0.18641771376132965, "learning_rate": 5.307265514041471e-07, "loss": 0.0002, "step": 211900 }, { "epoch": 3.4673975292481387, "grad_norm": 0.014453309588134289, "learning_rate": 5.304064913523799e-07, "loss": 0.0001, "step": 211910 }, { "epoch": 3.4675611551992147, "grad_norm": 0.000650445930659771, "learning_rate": 5.300865224318241e-07, "loss": 0.0003, "step": 211920 }, { "epoch": 3.4677247811502903, "grad_norm": 0.11911773681640625, "learning_rate": 5.297666446490018e-07, "loss": 0.0008, "step": 211930 }, { "epoch": 3.4678884071013663, "grad_norm": 0.010756704024970531, "learning_rate": 5.294468580104356e-07, "loss": 0.0003, "step": 211940 }, { "epoch": 3.4680520330524423, "grad_norm": 0.07452954351902008, "learning_rate": 5.291271625226463e-07, "loss": 0.0003, "step": 211950 }, { "epoch": 3.468215659003518, "grad_norm": 0.0552913099527359, "learning_rate": 5.288075581921526e-07, "loss": 0.0005, "step": 211960 }, { "epoch": 3.468379284954594, "grad_norm": 0.00013328671047929674, "learning_rate": 5.284880450254687e-07, "loss": 0.0002, "step": 211970 }, { "epoch": 3.46854291090567, "grad_norm": 0.0012583821080625057, "learning_rate": 5.281686230291111e-07, "loss": 0.0002, "step": 211980 }, { "epoch": 3.4687065368567453, "grad_norm": 0.0041373856365680695, "learning_rate": 5.278492922095924e-07, "loss": 0.0002, "step": 211990 }, { "epoch": 3.4688701628078213, "grad_norm": 0.10453013330698013, "learning_rate": 5.275300525734239e-07, "loss": 0.0005, "step": 212000 }, { "epoch": 3.4690337887588973, "grad_norm": 0.0034768700134009123, "learning_rate": 5.272109041271128e-07, "loss": 0.0003, "step": 212010 }, { "epoch": 3.469197414709973, "grad_norm": 0.006292905192822218, "learning_rate": 5.268918468771678e-07, "loss": 0.0001, "step": 212020 }, { "epoch": 3.469361040661049, "grad_norm": 0.023509079590439796, "learning_rate": 5.265728808300935e-07, "loss": 0.0003, "step": 212030 }, { "epoch": 3.4695246666121244, "grad_norm": 0.0071974340826272964, "learning_rate": 5.262540059923949e-07, "loss": 0.0002, "step": 212040 }, { "epoch": 3.4696882925632004, "grad_norm": 0.036761753261089325, "learning_rate": 5.259352223705711e-07, "loss": 0.0002, "step": 212050 }, { "epoch": 3.4698519185142764, "grad_norm": 0.020676372572779655, "learning_rate": 5.25616529971123e-07, "loss": 0.0003, "step": 212060 }, { "epoch": 3.470015544465352, "grad_norm": 0.03546781465411186, "learning_rate": 5.252979288005483e-07, "loss": 0.0001, "step": 212070 }, { "epoch": 3.470179170416428, "grad_norm": 0.0027009279001504183, "learning_rate": 5.249794188653446e-07, "loss": 0.0005, "step": 212080 }, { "epoch": 3.470342796367504, "grad_norm": 0.03611007332801819, "learning_rate": 5.246610001720037e-07, "loss": 0.0004, "step": 212090 }, { "epoch": 3.4705064223185795, "grad_norm": 0.013826992362737656, "learning_rate": 5.24342672727019e-07, "loss": 0.0002, "step": 212100 }, { "epoch": 3.4706700482696555, "grad_norm": 0.00963637139648199, "learning_rate": 5.240244365368802e-07, "loss": 0.0003, "step": 212110 }, { "epoch": 3.4708336742207315, "grad_norm": 0.06086818501353264, "learning_rate": 5.237062916080782e-07, "loss": 0.0003, "step": 212120 }, { "epoch": 3.470997300171807, "grad_norm": 0.03336796537041664, "learning_rate": 5.233882379470967e-07, "loss": 0.0003, "step": 212130 }, { "epoch": 3.471160926122883, "grad_norm": 0.01724601536989212, "learning_rate": 5.230702755604216e-07, "loss": 0.0002, "step": 212140 }, { "epoch": 3.471324552073959, "grad_norm": 0.09933177381753922, "learning_rate": 5.227524044545362e-07, "loss": 0.0003, "step": 212150 }, { "epoch": 3.4714881780250346, "grad_norm": 0.05658682435750961, "learning_rate": 5.224346246359225e-07, "loss": 0.0006, "step": 212160 }, { "epoch": 3.4716518039761106, "grad_norm": 0.05981430038809776, "learning_rate": 5.221169361110578e-07, "loss": 0.0002, "step": 212170 }, { "epoch": 3.4718154299271866, "grad_norm": 0.03219384327530861, "learning_rate": 5.2179933888642e-07, "loss": 0.0002, "step": 212180 }, { "epoch": 3.471979055878262, "grad_norm": 0.01735318824648857, "learning_rate": 5.214818329684857e-07, "loss": 0.0003, "step": 212190 }, { "epoch": 3.472142681829338, "grad_norm": 0.012494699098169804, "learning_rate": 5.211644183637287e-07, "loss": 0.0003, "step": 212200 }, { "epoch": 3.472306307780414, "grad_norm": 0.004930426832288504, "learning_rate": 5.208470950786193e-07, "loss": 0.0003, "step": 212210 }, { "epoch": 3.4724699337314897, "grad_norm": 0.01354595273733139, "learning_rate": 5.205298631196287e-07, "loss": 0.0003, "step": 212220 }, { "epoch": 3.4726335596825657, "grad_norm": 0.0035261239390820265, "learning_rate": 5.202127224932235e-07, "loss": 0.0003, "step": 212230 }, { "epoch": 3.4727971856336417, "grad_norm": 0.01371942088007927, "learning_rate": 5.19895673205873e-07, "loss": 0.0002, "step": 212240 }, { "epoch": 3.4729608115847173, "grad_norm": 0.016596782952547073, "learning_rate": 5.19578715264038e-07, "loss": 0.0003, "step": 212250 }, { "epoch": 3.4731244375357933, "grad_norm": 0.050734709948301315, "learning_rate": 5.192618486741824e-07, "loss": 0.0003, "step": 212260 }, { "epoch": 3.4732880634868692, "grad_norm": 0.07375026494264603, "learning_rate": 5.18945073442767e-07, "loss": 0.0003, "step": 212270 }, { "epoch": 3.473451689437945, "grad_norm": 0.0104065565392375, "learning_rate": 5.186283895762517e-07, "loss": 0.0003, "step": 212280 }, { "epoch": 3.473615315389021, "grad_norm": 0.011609407141804695, "learning_rate": 5.183117970810913e-07, "loss": 0.0003, "step": 212290 }, { "epoch": 3.4737789413400963, "grad_norm": 0.04537360742688179, "learning_rate": 5.17995295963743e-07, "loss": 0.0003, "step": 212300 }, { "epoch": 3.4739425672911723, "grad_norm": 0.01107197254896164, "learning_rate": 5.176788862306564e-07, "loss": 0.0003, "step": 212310 }, { "epoch": 3.4741061932422483, "grad_norm": 0.028338739648461342, "learning_rate": 5.173625678882871e-07, "loss": 0.0001, "step": 212320 }, { "epoch": 3.474269819193324, "grad_norm": 0.014513738453388214, "learning_rate": 5.170463409430821e-07, "loss": 0.0003, "step": 212330 }, { "epoch": 3.4744334451444, "grad_norm": 0.0032797600142657757, "learning_rate": 5.167302054014906e-07, "loss": 0.0002, "step": 212340 }, { "epoch": 3.474597071095476, "grad_norm": 0.04545975849032402, "learning_rate": 5.164141612699553e-07, "loss": 0.0004, "step": 212350 }, { "epoch": 3.4747606970465514, "grad_norm": 0.0420084185898304, "learning_rate": 5.160982085549238e-07, "loss": 0.0004, "step": 212360 }, { "epoch": 3.4749243229976274, "grad_norm": 0.020371340215206146, "learning_rate": 5.157823472628359e-07, "loss": 0.0002, "step": 212370 }, { "epoch": 3.4750879489487034, "grad_norm": 0.008345184847712517, "learning_rate": 5.154665774001327e-07, "loss": 0.0009, "step": 212380 }, { "epoch": 3.475251574899779, "grad_norm": 0.004859854932874441, "learning_rate": 5.151508989732506e-07, "loss": 0.0003, "step": 212390 }, { "epoch": 3.475415200850855, "grad_norm": 0.05220484361052513, "learning_rate": 5.14835311988629e-07, "loss": 0.0003, "step": 212400 }, { "epoch": 3.475578826801931, "grad_norm": 0.02425549365580082, "learning_rate": 5.145198164526999e-07, "loss": 0.0004, "step": 212410 }, { "epoch": 3.4757424527530065, "grad_norm": 1.5735435226815753e-05, "learning_rate": 5.142044123718981e-07, "loss": 0.0003, "step": 212420 }, { "epoch": 3.4759060787040825, "grad_norm": 0.045983780175447464, "learning_rate": 5.138890997526514e-07, "loss": 0.0003, "step": 212430 }, { "epoch": 3.476069704655158, "grad_norm": 0.03122561424970627, "learning_rate": 5.135738786013927e-07, "loss": 0.0002, "step": 212440 }, { "epoch": 3.476233330606234, "grad_norm": 0.031047023832798004, "learning_rate": 5.132587489245461e-07, "loss": 0.0005, "step": 212450 }, { "epoch": 3.47639695655731, "grad_norm": 0.021413961425423622, "learning_rate": 5.129437107285385e-07, "loss": 0.0002, "step": 212460 }, { "epoch": 3.4765605825083856, "grad_norm": 0.042746152728796005, "learning_rate": 5.126287640197907e-07, "loss": 0.0003, "step": 212470 }, { "epoch": 3.4767242084594616, "grad_norm": 0.03687288612127304, "learning_rate": 5.12313908804728e-07, "loss": 0.0005, "step": 212480 }, { "epoch": 3.4768878344105376, "grad_norm": 0.014902932569384575, "learning_rate": 5.119991450897665e-07, "loss": 0.0005, "step": 212490 }, { "epoch": 3.477051460361613, "grad_norm": 0.0014517649542540312, "learning_rate": 5.116844728813275e-07, "loss": 0.0001, "step": 212500 }, { "epoch": 3.477215086312689, "grad_norm": 0.03917462378740311, "learning_rate": 5.113698921858234e-07, "loss": 0.0006, "step": 212510 }, { "epoch": 3.477378712263765, "grad_norm": 0.03815577179193497, "learning_rate": 5.110554030096698e-07, "loss": 0.0003, "step": 212520 }, { "epoch": 3.4775423382148407, "grad_norm": 0.028162531554698944, "learning_rate": 5.107410053592787e-07, "loss": 0.0003, "step": 212530 }, { "epoch": 3.4777059641659167, "grad_norm": 0.005997353233397007, "learning_rate": 5.104266992410616e-07, "loss": 0.0002, "step": 212540 }, { "epoch": 3.4778695901169927, "grad_norm": 0.02803533338010311, "learning_rate": 5.101124846614247e-07, "loss": 0.0004, "step": 212550 }, { "epoch": 3.4780332160680683, "grad_norm": 0.014857945032417774, "learning_rate": 5.09798361626776e-07, "loss": 0.0001, "step": 212560 }, { "epoch": 3.4781968420191443, "grad_norm": 0.002619342412799597, "learning_rate": 5.094843301435199e-07, "loss": 0.0002, "step": 212570 }, { "epoch": 3.4783604679702202, "grad_norm": 0.01020092610269785, "learning_rate": 5.091703902180606e-07, "loss": 0.0002, "step": 212580 }, { "epoch": 3.478524093921296, "grad_norm": 0.036277856677770615, "learning_rate": 5.088565418567964e-07, "loss": 0.0003, "step": 212590 }, { "epoch": 3.478687719872372, "grad_norm": 0.004851757548749447, "learning_rate": 5.085427850661273e-07, "loss": 0.0004, "step": 212600 }, { "epoch": 3.478851345823448, "grad_norm": 0.04835718125104904, "learning_rate": 5.082291198524513e-07, "loss": 0.0006, "step": 212610 }, { "epoch": 3.4790149717745233, "grad_norm": 0.008252330124378204, "learning_rate": 5.079155462221647e-07, "loss": 0.0002, "step": 212620 }, { "epoch": 3.4791785977255993, "grad_norm": 0.01425324846059084, "learning_rate": 5.076020641816587e-07, "loss": 0.0003, "step": 212630 }, { "epoch": 3.4793422236766753, "grad_norm": 0.03900543972849846, "learning_rate": 5.072886737373255e-07, "loss": 0.0003, "step": 212640 }, { "epoch": 3.479505849627751, "grad_norm": 0.014694270677864552, "learning_rate": 5.069753748955558e-07, "loss": 0.0001, "step": 212650 }, { "epoch": 3.479669475578827, "grad_norm": 0.013415318913757801, "learning_rate": 5.066621676627376e-07, "loss": 0.0002, "step": 212660 }, { "epoch": 3.479833101529903, "grad_norm": 0.006280930712819099, "learning_rate": 5.063490520452558e-07, "loss": 0.0001, "step": 212670 }, { "epoch": 3.4799967274809784, "grad_norm": 0.05495976656675339, "learning_rate": 5.060360280494947e-07, "loss": 0.0005, "step": 212680 }, { "epoch": 3.4801603534320544, "grad_norm": 0.030321018770337105, "learning_rate": 5.057230956818376e-07, "loss": 0.0003, "step": 212690 }, { "epoch": 3.48032397938313, "grad_norm": 0.02193065732717514, "learning_rate": 5.054102549486645e-07, "loss": 0.0004, "step": 212700 }, { "epoch": 3.480487605334206, "grad_norm": 0.07157812267541885, "learning_rate": 5.050975058563534e-07, "loss": 0.0003, "step": 212710 }, { "epoch": 3.480651231285282, "grad_norm": 0.00047628552420064807, "learning_rate": 5.047848484112811e-07, "loss": 0.0005, "step": 212720 }, { "epoch": 3.4808148572363575, "grad_norm": 0.0019468916580080986, "learning_rate": 5.044722826198229e-07, "loss": 0.0004, "step": 212730 }, { "epoch": 3.4809784831874335, "grad_norm": 0.014197650365531445, "learning_rate": 5.041598084883526e-07, "loss": 0.0004, "step": 212740 }, { "epoch": 3.4811421091385095, "grad_norm": 0.004219801630824804, "learning_rate": 5.03847426023239e-07, "loss": 0.0002, "step": 212750 }, { "epoch": 3.481305735089585, "grad_norm": 0.04356072098016739, "learning_rate": 5.035351352308526e-07, "loss": 0.0003, "step": 212760 }, { "epoch": 3.481469361040661, "grad_norm": 0.025682494044303894, "learning_rate": 5.032229361175605e-07, "loss": 0.0002, "step": 212770 }, { "epoch": 3.481632986991737, "grad_norm": 0.01782524213194847, "learning_rate": 5.029108286897299e-07, "loss": 0.0001, "step": 212780 }, { "epoch": 3.4817966129428126, "grad_norm": 0.004403345286846161, "learning_rate": 5.025988129537213e-07, "loss": 0.0004, "step": 212790 }, { "epoch": 3.4819602388938886, "grad_norm": 0.04218781366944313, "learning_rate": 5.022868889158982e-07, "loss": 0.0003, "step": 212800 }, { "epoch": 3.482123864844964, "grad_norm": 0.028927452862262726, "learning_rate": 5.019750565826215e-07, "loss": 0.0003, "step": 212810 }, { "epoch": 3.48228749079604, "grad_norm": 0.031216509640216827, "learning_rate": 5.016633159602469e-07, "loss": 0.0002, "step": 212820 }, { "epoch": 3.482451116747116, "grad_norm": 0.07785847038030624, "learning_rate": 5.013516670551322e-07, "loss": 0.0003, "step": 212830 }, { "epoch": 3.4826147426981917, "grad_norm": 0.015931444242596626, "learning_rate": 5.010401098736295e-07, "loss": 0.0004, "step": 212840 }, { "epoch": 3.4827783686492677, "grad_norm": 0.038646429777145386, "learning_rate": 5.007286444220944e-07, "loss": 0.0002, "step": 212850 }, { "epoch": 3.4829419946003437, "grad_norm": 0.03466450795531273, "learning_rate": 5.004172707068749e-07, "loss": 0.0002, "step": 212860 }, { "epoch": 3.4831056205514193, "grad_norm": 0.026893874630331993, "learning_rate": 5.001059887343218e-07, "loss": 0.0003, "step": 212870 }, { "epoch": 3.4832692465024953, "grad_norm": 0.0004773183318320662, "learning_rate": 4.997947985107782e-07, "loss": 0.0003, "step": 212880 }, { "epoch": 3.4834328724535712, "grad_norm": 0.00707553094252944, "learning_rate": 4.994837000425934e-07, "loss": 0.0004, "step": 212890 }, { "epoch": 3.483596498404647, "grad_norm": 0.057981815189123154, "learning_rate": 4.991726933361074e-07, "loss": 0.0004, "step": 212900 }, { "epoch": 3.483760124355723, "grad_norm": 0.04199368506669998, "learning_rate": 4.988617783976635e-07, "loss": 0.0002, "step": 212910 }, { "epoch": 3.483923750306799, "grad_norm": 0.19399188458919525, "learning_rate": 4.985509552335977e-07, "loss": 0.0003, "step": 212920 }, { "epoch": 3.4840873762578743, "grad_norm": 0.03123348578810692, "learning_rate": 4.982402238502515e-07, "loss": 0.0004, "step": 212930 }, { "epoch": 3.4842510022089503, "grad_norm": 0.0013164496049284935, "learning_rate": 4.979295842539577e-07, "loss": 0.0008, "step": 212940 }, { "epoch": 3.4844146281600263, "grad_norm": 0.014071791432797909, "learning_rate": 4.976190364510513e-07, "loss": 0.0002, "step": 212950 }, { "epoch": 3.484578254111102, "grad_norm": 0.027486668899655342, "learning_rate": 4.973085804478634e-07, "loss": 0.0003, "step": 212960 }, { "epoch": 3.484741880062178, "grad_norm": 0.007253268733620644, "learning_rate": 4.969982162507236e-07, "loss": 0.0001, "step": 212970 }, { "epoch": 3.484905506013254, "grad_norm": 0.00024997995933517814, "learning_rate": 4.966879438659611e-07, "loss": 0.0003, "step": 212980 }, { "epoch": 3.4850691319643294, "grad_norm": 0.0008770794374868274, "learning_rate": 4.963777632999017e-07, "loss": 0.0001, "step": 212990 }, { "epoch": 3.4852327579154054, "grad_norm": 0.009179777465760708, "learning_rate": 4.960676745588694e-07, "loss": 0.0005, "step": 213000 }, { "epoch": 3.4853963838664814, "grad_norm": 0.07185684889554977, "learning_rate": 4.957576776491863e-07, "loss": 0.0003, "step": 213010 }, { "epoch": 3.485560009817557, "grad_norm": 0.00016976130427792668, "learning_rate": 4.954477725771734e-07, "loss": 0.0003, "step": 213020 }, { "epoch": 3.485723635768633, "grad_norm": 0.007480171043425798, "learning_rate": 4.951379593491501e-07, "loss": 0.0002, "step": 213030 }, { "epoch": 3.485887261719709, "grad_norm": 0.07033878564834595, "learning_rate": 4.94828237971432e-07, "loss": 0.0006, "step": 213040 }, { "epoch": 3.4860508876707845, "grad_norm": 0.03461983799934387, "learning_rate": 4.945186084503345e-07, "loss": 0.0004, "step": 213050 }, { "epoch": 3.4862145136218605, "grad_norm": 0.06279153376817703, "learning_rate": 4.94209070792171e-07, "loss": 0.0004, "step": 213060 }, { "epoch": 3.486378139572936, "grad_norm": 0.06285121291875839, "learning_rate": 4.93899625003253e-07, "loss": 0.0005, "step": 213070 }, { "epoch": 3.486541765524012, "grad_norm": 0.023137206211686134, "learning_rate": 4.935902710898888e-07, "loss": 0.0005, "step": 213080 }, { "epoch": 3.486705391475088, "grad_norm": 0.008386642672121525, "learning_rate": 4.932810090583861e-07, "loss": 0.0002, "step": 213090 }, { "epoch": 3.4868690174261636, "grad_norm": 0.06034737825393677, "learning_rate": 4.929718389150512e-07, "loss": 0.0004, "step": 213100 }, { "epoch": 3.4870326433772396, "grad_norm": 0.03324514999985695, "learning_rate": 4.926627606661882e-07, "loss": 0.0003, "step": 213110 }, { "epoch": 3.4871962693283156, "grad_norm": 0.022029951214790344, "learning_rate": 4.923537743180973e-07, "loss": 0.0001, "step": 213120 }, { "epoch": 3.487359895279391, "grad_norm": 0.008153736591339111, "learning_rate": 4.920448798770789e-07, "loss": 0.0003, "step": 213130 }, { "epoch": 3.487523521230467, "grad_norm": 0.009854139760136604, "learning_rate": 4.917360773494318e-07, "loss": 0.0002, "step": 213140 }, { "epoch": 3.487687147181543, "grad_norm": 0.04888466000556946, "learning_rate": 4.914273667414532e-07, "loss": 0.0004, "step": 213150 }, { "epoch": 3.4878507731326187, "grad_norm": 0.13323469460010529, "learning_rate": 4.911187480594348e-07, "loss": 0.0007, "step": 213160 }, { "epoch": 3.4880143990836947, "grad_norm": 0.09407972544431686, "learning_rate": 4.908102213096705e-07, "loss": 0.0002, "step": 213170 }, { "epoch": 3.4881780250347707, "grad_norm": 0.05878284573554993, "learning_rate": 4.905017864984513e-07, "loss": 0.0002, "step": 213180 }, { "epoch": 3.4883416509858463, "grad_norm": 0.07844063639640808, "learning_rate": 4.90193443632066e-07, "loss": 0.0004, "step": 213190 }, { "epoch": 3.4885052769369223, "grad_norm": 0.0014614167157560587, "learning_rate": 4.898851927167997e-07, "loss": 0.0002, "step": 213200 }, { "epoch": 3.488668902887998, "grad_norm": 0.026078984141349792, "learning_rate": 4.895770337589389e-07, "loss": 0.0003, "step": 213210 }, { "epoch": 3.488832528839074, "grad_norm": 0.005329280160367489, "learning_rate": 4.892689667647666e-07, "loss": 0.0003, "step": 213220 }, { "epoch": 3.48899615479015, "grad_norm": 0.054281409829854965, "learning_rate": 4.889609917405647e-07, "loss": 0.0001, "step": 213230 }, { "epoch": 3.4891597807412253, "grad_norm": 0.042828820645809174, "learning_rate": 4.88653108692611e-07, "loss": 0.0004, "step": 213240 }, { "epoch": 3.4893234066923013, "grad_norm": 0.0038490549195557833, "learning_rate": 4.883453176271829e-07, "loss": 0.0009, "step": 213250 }, { "epoch": 3.4894870326433773, "grad_norm": 0.008601978421211243, "learning_rate": 4.880376185505576e-07, "loss": 0.0007, "step": 213260 }, { "epoch": 3.489650658594453, "grad_norm": 0.01182270236313343, "learning_rate": 4.877300114690081e-07, "loss": 0.0006, "step": 213270 }, { "epoch": 3.489814284545529, "grad_norm": 0.006395323667675257, "learning_rate": 4.874224963888058e-07, "loss": 0.0002, "step": 213280 }, { "epoch": 3.489977910496605, "grad_norm": 0.015864353626966476, "learning_rate": 4.871150733162206e-07, "loss": 0.0003, "step": 213290 }, { "epoch": 3.4901415364476804, "grad_norm": 0.06374531984329224, "learning_rate": 4.868077422575213e-07, "loss": 0.0004, "step": 213300 }, { "epoch": 3.4903051623987564, "grad_norm": 0.0001445369707653299, "learning_rate": 4.865005032189746e-07, "loss": 0.0005, "step": 213310 }, { "epoch": 3.4904687883498324, "grad_norm": 0.04617449641227722, "learning_rate": 4.861933562068433e-07, "loss": 0.0003, "step": 213320 }, { "epoch": 3.490632414300908, "grad_norm": 0.013814697973430157, "learning_rate": 4.858863012273901e-07, "loss": 0.0004, "step": 213330 }, { "epoch": 3.490796040251984, "grad_norm": 0.03391192480921745, "learning_rate": 4.855793382868768e-07, "loss": 0.0001, "step": 213340 }, { "epoch": 3.49095966620306, "grad_norm": 0.014871422201395035, "learning_rate": 4.852724673915615e-07, "loss": 0.0003, "step": 213350 }, { "epoch": 3.4911232921541355, "grad_norm": 0.0002437025832477957, "learning_rate": 4.849656885477006e-07, "loss": 0.0002, "step": 213360 }, { "epoch": 3.4912869181052115, "grad_norm": 0.014950369484722614, "learning_rate": 4.846590017615494e-07, "loss": 0.0004, "step": 213370 }, { "epoch": 3.4914505440562875, "grad_norm": 0.00783984549343586, "learning_rate": 4.843524070393607e-07, "loss": 0.0009, "step": 213380 }, { "epoch": 3.491614170007363, "grad_norm": 0.03431730344891548, "learning_rate": 4.840459043873875e-07, "loss": 0.0003, "step": 213390 }, { "epoch": 3.491777795958439, "grad_norm": 0.007561329286545515, "learning_rate": 4.837394938118761e-07, "loss": 0.0002, "step": 213400 }, { "epoch": 3.491941421909515, "grad_norm": 0.023527976125478745, "learning_rate": 4.834331753190768e-07, "loss": 0.0001, "step": 213410 }, { "epoch": 3.4921050478605906, "grad_norm": 0.03022470697760582, "learning_rate": 4.831269489152318e-07, "loss": 0.0003, "step": 213420 }, { "epoch": 3.4922686738116666, "grad_norm": 0.0009436738328076899, "learning_rate": 4.828208146065888e-07, "loss": 0.0002, "step": 213430 }, { "epoch": 3.4924322997627426, "grad_norm": 0.006310638505965471, "learning_rate": 4.825147723993867e-07, "loss": 0.0002, "step": 213440 }, { "epoch": 3.492595925713818, "grad_norm": 0.03191375732421875, "learning_rate": 4.822088222998672e-07, "loss": 0.0002, "step": 213450 }, { "epoch": 3.492759551664894, "grad_norm": 0.022364243865013123, "learning_rate": 4.819029643142659e-07, "loss": 0.0002, "step": 213460 }, { "epoch": 3.4929231776159697, "grad_norm": 0.05342153087258339, "learning_rate": 4.815971984488221e-07, "loss": 0.0003, "step": 213470 }, { "epoch": 3.4930868035670457, "grad_norm": 0.030240872874855995, "learning_rate": 4.812915247097683e-07, "loss": 0.0002, "step": 213480 }, { "epoch": 3.4932504295181217, "grad_norm": 0.005515790544450283, "learning_rate": 4.809859431033376e-07, "loss": 0.0002, "step": 213490 }, { "epoch": 3.4934140554691973, "grad_norm": 0.01518369559198618, "learning_rate": 4.80680453635759e-07, "loss": 0.0003, "step": 213500 }, { "epoch": 3.4935776814202733, "grad_norm": 0.0025122391525655985, "learning_rate": 4.803750563132642e-07, "loss": 0.0003, "step": 213510 }, { "epoch": 3.4937413073713492, "grad_norm": 0.07439948618412018, "learning_rate": 4.800697511420771e-07, "loss": 0.0004, "step": 213520 }, { "epoch": 3.493904933322425, "grad_norm": 0.0007553612231276929, "learning_rate": 4.797645381284244e-07, "loss": 0.0001, "step": 213530 }, { "epoch": 3.494068559273501, "grad_norm": 0.03912946954369545, "learning_rate": 4.794594172785267e-07, "loss": 0.0002, "step": 213540 }, { "epoch": 3.494232185224577, "grad_norm": 0.07164519280195236, "learning_rate": 4.791543885986088e-07, "loss": 0.0003, "step": 213550 }, { "epoch": 3.4943958111756523, "grad_norm": 0.045581866055727005, "learning_rate": 4.788494520948872e-07, "loss": 0.0003, "step": 213560 }, { "epoch": 3.4945594371267283, "grad_norm": 0.020013626664876938, "learning_rate": 4.785446077735811e-07, "loss": 0.0003, "step": 213570 }, { "epoch": 3.494723063077804, "grad_norm": 0.0010733017697930336, "learning_rate": 4.782398556409034e-07, "loss": 0.0002, "step": 213580 }, { "epoch": 3.49488668902888, "grad_norm": 0.0804298147559166, "learning_rate": 4.779351957030714e-07, "loss": 0.0002, "step": 213590 }, { "epoch": 3.495050314979956, "grad_norm": 0.03877793997526169, "learning_rate": 4.776306279662934e-07, "loss": 0.0003, "step": 213600 }, { "epoch": 3.4952139409310314, "grad_norm": 0.0007055071182549, "learning_rate": 4.773261524367822e-07, "loss": 0.0002, "step": 213610 }, { "epoch": 3.4953775668821074, "grad_norm": 0.0002510537742637098, "learning_rate": 4.77021769120743e-07, "loss": 0.0002, "step": 213620 }, { "epoch": 3.4955411928331834, "grad_norm": 0.06617656350135803, "learning_rate": 4.767174780243833e-07, "loss": 0.0002, "step": 213630 }, { "epoch": 3.495704818784259, "grad_norm": 0.0028148030396550894, "learning_rate": 4.764132791539078e-07, "loss": 0.0001, "step": 213640 }, { "epoch": 3.495868444735335, "grad_norm": 0.008422385901212692, "learning_rate": 4.761091725155187e-07, "loss": 0.0002, "step": 213650 }, { "epoch": 3.496032070686411, "grad_norm": 0.026202509179711342, "learning_rate": 4.758051581154155e-07, "loss": 0.0006, "step": 213660 }, { "epoch": 3.4961956966374865, "grad_norm": 0.006737092975527048, "learning_rate": 4.7550123595979767e-07, "loss": 0.0003, "step": 213670 }, { "epoch": 3.4963593225885625, "grad_norm": 0.037130121141672134, "learning_rate": 4.751974060548614e-07, "loss": 0.0003, "step": 213680 }, { "epoch": 3.4965229485396385, "grad_norm": 0.00805947370827198, "learning_rate": 4.748936684068028e-07, "loss": 0.0002, "step": 213690 }, { "epoch": 3.496686574490714, "grad_norm": 0.04790101200342178, "learning_rate": 4.745900230218126e-07, "loss": 0.0006, "step": 213700 }, { "epoch": 3.49685020044179, "grad_norm": 0.0009474946418777108, "learning_rate": 4.742864699060834e-07, "loss": 0.0003, "step": 213710 }, { "epoch": 3.497013826392866, "grad_norm": 0.06831742078065872, "learning_rate": 4.739830090658043e-07, "loss": 0.0002, "step": 213720 }, { "epoch": 3.4971774523439416, "grad_norm": 0.0030216434970498085, "learning_rate": 4.736796405071631e-07, "loss": 0.0005, "step": 213730 }, { "epoch": 3.4973410782950176, "grad_norm": 0.02082277275621891, "learning_rate": 4.733763642363437e-07, "loss": 0.0002, "step": 213740 }, { "epoch": 3.4975047042460936, "grad_norm": 0.035439129918813705, "learning_rate": 4.730731802595301e-07, "loss": 0.0005, "step": 213750 }, { "epoch": 3.497668330197169, "grad_norm": 0.02547547221183777, "learning_rate": 4.7277008858290506e-07, "loss": 0.0002, "step": 213760 }, { "epoch": 3.497831956148245, "grad_norm": 0.16159220039844513, "learning_rate": 4.724670892126482e-07, "loss": 0.0004, "step": 213770 }, { "epoch": 3.497995582099321, "grad_norm": 0.0007308662170544267, "learning_rate": 4.7216418215493555e-07, "loss": 0.0001, "step": 213780 }, { "epoch": 3.4981592080503967, "grad_norm": 0.042190201580524445, "learning_rate": 4.7186136741594503e-07, "loss": 0.0002, "step": 213790 }, { "epoch": 3.4983228340014727, "grad_norm": 0.04992168769240379, "learning_rate": 4.7155864500185e-07, "loss": 0.0004, "step": 213800 }, { "epoch": 3.4984864599525487, "grad_norm": 0.04093103110790253, "learning_rate": 4.712560149188233e-07, "loss": 0.0005, "step": 213810 }, { "epoch": 3.4986500859036243, "grad_norm": 0.03894557058811188, "learning_rate": 4.709534771730345e-07, "loss": 0.0004, "step": 213820 }, { "epoch": 3.4988137118547002, "grad_norm": 0.03380170837044716, "learning_rate": 4.706510317706525e-07, "loss": 0.0002, "step": 213830 }, { "epoch": 3.4989773378057762, "grad_norm": 0.04653288424015045, "learning_rate": 4.7034867871784406e-07, "loss": 0.0003, "step": 213840 }, { "epoch": 3.499140963756852, "grad_norm": 0.007184270303696394, "learning_rate": 4.700464180207742e-07, "loss": 0.0004, "step": 213850 }, { "epoch": 3.499304589707928, "grad_norm": 0.0031843201722949743, "learning_rate": 4.6974424968560475e-07, "loss": 0.0004, "step": 213860 }, { "epoch": 3.4994682156590033, "grad_norm": 0.018634382635354996, "learning_rate": 4.6944217371849675e-07, "loss": 0.0002, "step": 213870 }, { "epoch": 3.4996318416100793, "grad_norm": 0.00034854913246817887, "learning_rate": 4.691401901256104e-07, "loss": 0.0006, "step": 213880 }, { "epoch": 3.4997954675611553, "grad_norm": 0.013337591663002968, "learning_rate": 4.688382989131024e-07, "loss": 0.0002, "step": 213890 }, { "epoch": 3.499959093512231, "grad_norm": 0.02988244965672493, "learning_rate": 4.6853650008712723e-07, "loss": 0.0002, "step": 213900 }, { "epoch": 3.500122719463307, "grad_norm": 0.002412705682218075, "learning_rate": 4.6823479365383895e-07, "loss": 0.0004, "step": 213910 }, { "epoch": 3.500286345414383, "grad_norm": 0.0035041470546275377, "learning_rate": 4.679331796193892e-07, "loss": 0.0003, "step": 213920 }, { "epoch": 3.5004499713654584, "grad_norm": 0.041725825518369675, "learning_rate": 4.676316579899287e-07, "loss": 0.0003, "step": 213930 }, { "epoch": 3.5006135973165344, "grad_norm": 0.023881666362285614, "learning_rate": 4.673302287716025e-07, "loss": 0.0002, "step": 213940 }, { "epoch": 3.50077722326761, "grad_norm": 0.005211166571825743, "learning_rate": 4.6702889197055843e-07, "loss": 0.0001, "step": 213950 }, { "epoch": 3.500940849218686, "grad_norm": 0.018091542646288872, "learning_rate": 4.6672764759294e-07, "loss": 0.0002, "step": 213960 }, { "epoch": 3.501104475169762, "grad_norm": 0.02260947972536087, "learning_rate": 4.664264956448905e-07, "loss": 0.0002, "step": 213970 }, { "epoch": 3.5012681011208375, "grad_norm": 0.011475194245576859, "learning_rate": 4.661254361325479e-07, "loss": 0.0004, "step": 213980 }, { "epoch": 3.5014317270719135, "grad_norm": 0.007499185856431723, "learning_rate": 4.6582446906205224e-07, "loss": 0.0004, "step": 213990 }, { "epoch": 3.5015953530229895, "grad_norm": 0.013431745581328869, "learning_rate": 4.6552359443953923e-07, "loss": 0.0001, "step": 214000 }, { "epoch": 3.501758978974065, "grad_norm": 0.023802822455763817, "learning_rate": 4.652228122711444e-07, "loss": 0.0002, "step": 214010 }, { "epoch": 3.501922604925141, "grad_norm": 0.022386742755770683, "learning_rate": 4.6492212256299906e-07, "loss": 0.0001, "step": 214020 }, { "epoch": 3.502086230876217, "grad_norm": 0.03573571890592575, "learning_rate": 4.646215253212344e-07, "loss": 0.0001, "step": 214030 }, { "epoch": 3.5022498568272926, "grad_norm": 0.030504833906888962, "learning_rate": 4.6432102055197994e-07, "loss": 0.0003, "step": 214040 }, { "epoch": 3.5024134827783686, "grad_norm": 0.0017463566036894917, "learning_rate": 4.6402060826136295e-07, "loss": 0.0003, "step": 214050 }, { "epoch": 3.5025771087294446, "grad_norm": 0.0012970779789611697, "learning_rate": 4.6372028845550754e-07, "loss": 0.0001, "step": 214060 }, { "epoch": 3.50274073468052, "grad_norm": 0.07593691349029541, "learning_rate": 4.634200611405382e-07, "loss": 0.0003, "step": 214070 }, { "epoch": 3.502904360631596, "grad_norm": 0.04455888271331787, "learning_rate": 4.6311992632257384e-07, "loss": 0.0044, "step": 214080 }, { "epoch": 3.503067986582672, "grad_norm": 0.041757360100746155, "learning_rate": 4.6281988400773745e-07, "loss": 0.0002, "step": 214090 }, { "epoch": 3.5032316125337477, "grad_norm": 0.004772932268679142, "learning_rate": 4.6251993420214416e-07, "loss": 0.0002, "step": 214100 }, { "epoch": 3.5033952384848237, "grad_norm": 0.053573932498693466, "learning_rate": 4.622200769119112e-07, "loss": 0.0003, "step": 214110 }, { "epoch": 3.5035588644358997, "grad_norm": 0.007299461867660284, "learning_rate": 4.619203121431498e-07, "loss": 0.0001, "step": 214120 }, { "epoch": 3.5037224903869753, "grad_norm": 0.006043209694325924, "learning_rate": 4.616206399019751e-07, "loss": 0.0002, "step": 214130 }, { "epoch": 3.5038861163380512, "grad_norm": 0.013942509889602661, "learning_rate": 4.6132106019449564e-07, "loss": 0.0002, "step": 214140 }, { "epoch": 3.5040497422891272, "grad_norm": 0.0005016251234337687, "learning_rate": 4.6102157302682026e-07, "loss": 0.0003, "step": 214150 }, { "epoch": 3.504213368240203, "grad_norm": 0.029760848730802536, "learning_rate": 4.607221784050531e-07, "loss": 0.0002, "step": 214160 }, { "epoch": 3.504376994191279, "grad_norm": 0.0030106583144515753, "learning_rate": 4.604228763353019e-07, "loss": 0.0004, "step": 214170 }, { "epoch": 3.504540620142355, "grad_norm": 0.005808803252875805, "learning_rate": 4.6012366682366636e-07, "loss": 0.0001, "step": 214180 }, { "epoch": 3.5047042460934303, "grad_norm": 0.0013312731171026826, "learning_rate": 4.598245498762499e-07, "loss": 0.0002, "step": 214190 }, { "epoch": 3.5048678720445063, "grad_norm": 0.015437955036759377, "learning_rate": 4.595255254991471e-07, "loss": 0.0004, "step": 214200 }, { "epoch": 3.5050314979955823, "grad_norm": 0.002468904945999384, "learning_rate": 4.592265936984591e-07, "loss": 0.0004, "step": 214210 }, { "epoch": 3.505195123946658, "grad_norm": 0.09168095886707306, "learning_rate": 4.589277544802784e-07, "loss": 0.0001, "step": 214220 }, { "epoch": 3.505358749897734, "grad_norm": 0.013095813803374767, "learning_rate": 4.586290078507e-07, "loss": 0.0001, "step": 214230 }, { "epoch": 3.50552237584881, "grad_norm": 0.044715460389852524, "learning_rate": 4.583303538158124e-07, "loss": 0.0002, "step": 214240 }, { "epoch": 3.5056860017998854, "grad_norm": 0.018500160425901413, "learning_rate": 4.5803179238170626e-07, "loss": 0.0001, "step": 214250 }, { "epoch": 3.5058496277509614, "grad_norm": 0.0029536124784499407, "learning_rate": 4.577333235544695e-07, "loss": 0.0003, "step": 214260 }, { "epoch": 3.506013253702037, "grad_norm": 0.005244308616966009, "learning_rate": 4.574349473401879e-07, "loss": 0.0003, "step": 214270 }, { "epoch": 3.506176879653113, "grad_norm": 0.03303037956357002, "learning_rate": 4.5713666374494313e-07, "loss": 0.0002, "step": 214280 }, { "epoch": 3.506340505604189, "grad_norm": 0.03477994352579117, "learning_rate": 4.5683847277481875e-07, "loss": 0.0002, "step": 214290 }, { "epoch": 3.5065041315552645, "grad_norm": 0.014320864342153072, "learning_rate": 4.565403744358937e-07, "loss": 0.0002, "step": 214300 }, { "epoch": 3.5066677575063405, "grad_norm": 0.0398407019674778, "learning_rate": 4.562423687342477e-07, "loss": 0.0002, "step": 214310 }, { "epoch": 3.506831383457416, "grad_norm": 0.01634579710662365, "learning_rate": 4.55944455675954e-07, "loss": 0.0001, "step": 214320 }, { "epoch": 3.506995009408492, "grad_norm": 0.0010995810152962804, "learning_rate": 4.556466352670891e-07, "loss": 0.0001, "step": 214330 }, { "epoch": 3.507158635359568, "grad_norm": 0.024793347343802452, "learning_rate": 4.5534890751372354e-07, "loss": 0.0003, "step": 214340 }, { "epoch": 3.5073222613106436, "grad_norm": 0.04465140402317047, "learning_rate": 4.5505127242193024e-07, "loss": 0.0003, "step": 214350 }, { "epoch": 3.5074858872617196, "grad_norm": 0.09065975248813629, "learning_rate": 4.54753729997775e-07, "loss": 0.0005, "step": 214360 }, { "epoch": 3.5076495132127956, "grad_norm": 0.041539885103702545, "learning_rate": 4.544562802473257e-07, "loss": 0.0004, "step": 214370 }, { "epoch": 3.507813139163871, "grad_norm": 0.0007365712081082165, "learning_rate": 4.541589231766469e-07, "loss": 0.0003, "step": 214380 }, { "epoch": 3.507976765114947, "grad_norm": 0.01193312183022499, "learning_rate": 4.538616587918027e-07, "loss": 0.0001, "step": 214390 }, { "epoch": 3.508140391066023, "grad_norm": 0.01065569557249546, "learning_rate": 4.5356448709885146e-07, "loss": 0.0001, "step": 214400 }, { "epoch": 3.5083040170170987, "grad_norm": 0.017859630286693573, "learning_rate": 4.532674081038541e-07, "loss": 0.0003, "step": 214410 }, { "epoch": 3.5084676429681747, "grad_norm": 0.0038606186863034964, "learning_rate": 4.529704218128672e-07, "loss": 0.0003, "step": 214420 }, { "epoch": 3.5086312689192507, "grad_norm": 0.0028633903712034225, "learning_rate": 4.5267352823194777e-07, "loss": 0.0004, "step": 214430 }, { "epoch": 3.5087948948703263, "grad_norm": 0.004766879603266716, "learning_rate": 4.5237672736714645e-07, "loss": 0.0001, "step": 214440 }, { "epoch": 3.5089585208214022, "grad_norm": 0.04265652596950531, "learning_rate": 4.5208001922451617e-07, "loss": 0.0003, "step": 214450 }, { "epoch": 3.5091221467724782, "grad_norm": 0.020823918282985687, "learning_rate": 4.51783403810106e-07, "loss": 0.0003, "step": 214460 }, { "epoch": 3.509285772723554, "grad_norm": 0.00024562087492085993, "learning_rate": 4.5148688112996555e-07, "loss": 0.0002, "step": 214470 }, { "epoch": 3.50944939867463, "grad_norm": 0.0762178897857666, "learning_rate": 4.511904511901377e-07, "loss": 0.0003, "step": 214480 }, { "epoch": 3.509613024625706, "grad_norm": 0.031894903630018234, "learning_rate": 4.508941139966683e-07, "loss": 0.0002, "step": 214490 }, { "epoch": 3.5097766505767813, "grad_norm": 0.007862691767513752, "learning_rate": 4.5059786955559906e-07, "loss": 0.0002, "step": 214500 }, { "epoch": 3.5099402765278573, "grad_norm": 0.0020919220987707376, "learning_rate": 4.503017178729713e-07, "loss": 0.0003, "step": 214510 }, { "epoch": 3.5101039024789333, "grad_norm": 0.05917292833328247, "learning_rate": 4.500056589548207e-07, "loss": 0.0002, "step": 214520 }, { "epoch": 3.510267528430009, "grad_norm": 0.00416059885174036, "learning_rate": 4.4970969280718536e-07, "loss": 0.0006, "step": 214530 }, { "epoch": 3.510431154381085, "grad_norm": 0.000903684354852885, "learning_rate": 4.4941381943609977e-07, "loss": 0.0001, "step": 214540 }, { "epoch": 3.510594780332161, "grad_norm": 0.03634863719344139, "learning_rate": 4.491180388475969e-07, "loss": 0.0005, "step": 214550 }, { "epoch": 3.5107584062832364, "grad_norm": 0.01465563103556633, "learning_rate": 4.4882235104770577e-07, "loss": 0.0004, "step": 214560 }, { "epoch": 3.5109220322343124, "grad_norm": 0.016475535929203033, "learning_rate": 4.4852675604245667e-07, "loss": 0.0001, "step": 214570 }, { "epoch": 3.5110856581853884, "grad_norm": 0.011936530470848083, "learning_rate": 4.482312538378758e-07, "loss": 0.0, "step": 214580 }, { "epoch": 3.511249284136464, "grad_norm": 0.0019667395390570164, "learning_rate": 4.479358444399895e-07, "loss": 0.0003, "step": 214590 }, { "epoch": 3.51141291008754, "grad_norm": 0.03187684342265129, "learning_rate": 4.4764052785481904e-07, "loss": 0.0005, "step": 214600 }, { "epoch": 3.511576536038616, "grad_norm": 0.005652896128594875, "learning_rate": 4.473453040883863e-07, "loss": 0.0004, "step": 214610 }, { "epoch": 3.5117401619896915, "grad_norm": 0.03010551445186138, "learning_rate": 4.470501731467114e-07, "loss": 0.0002, "step": 214620 }, { "epoch": 3.5119037879407675, "grad_norm": 0.0020128313917666674, "learning_rate": 4.467551350358118e-07, "loss": 0.0001, "step": 214630 }, { "epoch": 3.5120674138918435, "grad_norm": 0.13508789241313934, "learning_rate": 4.4646018976170215e-07, "loss": 0.0002, "step": 214640 }, { "epoch": 3.512231039842919, "grad_norm": 0.0026530364993959665, "learning_rate": 4.461653373303959e-07, "loss": 0.0002, "step": 214650 }, { "epoch": 3.512394665793995, "grad_norm": 0.0013550581643357873, "learning_rate": 4.458705777479061e-07, "loss": 0.0001, "step": 214660 }, { "epoch": 3.5125582917450706, "grad_norm": 0.002059143502265215, "learning_rate": 4.455759110202429e-07, "loss": 0.0003, "step": 214670 }, { "epoch": 3.5127219176961466, "grad_norm": 0.029059533029794693, "learning_rate": 4.452813371534126e-07, "loss": 0.0003, "step": 214680 }, { "epoch": 3.5128855436472226, "grad_norm": 0.0017531351186335087, "learning_rate": 4.4498685615342207e-07, "loss": 0.0002, "step": 214690 }, { "epoch": 3.513049169598298, "grad_norm": 0.008548610843718052, "learning_rate": 4.4469246802627544e-07, "loss": 0.0002, "step": 214700 }, { "epoch": 3.513212795549374, "grad_norm": 0.06874313950538635, "learning_rate": 4.443981727779767e-07, "loss": 0.0002, "step": 214710 }, { "epoch": 3.5133764215004497, "grad_norm": 0.0067220404744148254, "learning_rate": 4.441039704145234e-07, "loss": 0.0002, "step": 214720 }, { "epoch": 3.5135400474515257, "grad_norm": 0.0009732830803841352, "learning_rate": 4.438098609419167e-07, "loss": 0.0003, "step": 214730 }, { "epoch": 3.5137036734026017, "grad_norm": 0.07146397233009338, "learning_rate": 4.4351584436615027e-07, "loss": 0.0002, "step": 214740 }, { "epoch": 3.5138672993536773, "grad_norm": 0.005985892377793789, "learning_rate": 4.4322192069322203e-07, "loss": 0.0004, "step": 214750 }, { "epoch": 3.5140309253047533, "grad_norm": 0.10352523624897003, "learning_rate": 4.4292808992912273e-07, "loss": 0.0007, "step": 214760 }, { "epoch": 3.5141945512558292, "grad_norm": 0.006889062933623791, "learning_rate": 4.426343520798448e-07, "loss": 0.0004, "step": 214770 }, { "epoch": 3.514358177206905, "grad_norm": 0.12383050471544266, "learning_rate": 4.423407071513752e-07, "loss": 0.0001, "step": 214780 }, { "epoch": 3.514521803157981, "grad_norm": 0.07257520407438278, "learning_rate": 4.420471551497035e-07, "loss": 0.0002, "step": 214790 }, { "epoch": 3.514685429109057, "grad_norm": 0.016601532697677612, "learning_rate": 4.417536960808139e-07, "loss": 0.0006, "step": 214800 }, { "epoch": 3.5148490550601323, "grad_norm": 0.004411717876791954, "learning_rate": 4.414603299506898e-07, "loss": 0.0002, "step": 214810 }, { "epoch": 3.5150126810112083, "grad_norm": 0.003499407786875963, "learning_rate": 4.4116705676531146e-07, "loss": 0.0002, "step": 214820 }, { "epoch": 3.5151763069622843, "grad_norm": 0.00629925774410367, "learning_rate": 4.4087387653066136e-07, "loss": 0.0001, "step": 214830 }, { "epoch": 3.51533993291336, "grad_norm": 0.03674633800983429, "learning_rate": 4.4058078925271476e-07, "loss": 0.0002, "step": 214840 }, { "epoch": 3.515503558864436, "grad_norm": 0.06417110562324524, "learning_rate": 4.4028779493744845e-07, "loss": 0.0005, "step": 214850 }, { "epoch": 3.515667184815512, "grad_norm": 0.04377258941531181, "learning_rate": 4.3999489359083494e-07, "loss": 0.0003, "step": 214860 }, { "epoch": 3.5158308107665874, "grad_norm": 0.00421293918043375, "learning_rate": 4.3970208521884885e-07, "loss": 0.0004, "step": 214870 }, { "epoch": 3.5159944367176634, "grad_norm": 0.1603328287601471, "learning_rate": 4.394093698274582e-07, "loss": 0.0003, "step": 214880 }, { "epoch": 3.5161580626687394, "grad_norm": 0.0005533632938750088, "learning_rate": 4.391167474226327e-07, "loss": 0.0001, "step": 214890 }, { "epoch": 3.516321688619815, "grad_norm": 0.02602485753595829, "learning_rate": 4.3882421801033693e-07, "loss": 0.0003, "step": 214900 }, { "epoch": 3.516485314570891, "grad_norm": 0.05123726278543472, "learning_rate": 4.385317815965362e-07, "loss": 0.0004, "step": 214910 }, { "epoch": 3.516648940521967, "grad_norm": 0.015296638943254948, "learning_rate": 4.382394381871935e-07, "loss": 0.0006, "step": 214920 }, { "epoch": 3.5168125664730425, "grad_norm": 0.008411383256316185, "learning_rate": 4.3794718778826907e-07, "loss": 0.0001, "step": 214930 }, { "epoch": 3.5169761924241185, "grad_norm": 0.026190949603915215, "learning_rate": 4.3765503040572146e-07, "loss": 0.0003, "step": 214940 }, { "epoch": 3.5171398183751945, "grad_norm": 0.0008577308035455644, "learning_rate": 4.37362966045507e-07, "loss": 0.0001, "step": 214950 }, { "epoch": 3.51730344432627, "grad_norm": 0.02410872094333172, "learning_rate": 4.370709947135815e-07, "loss": 0.0002, "step": 214960 }, { "epoch": 3.517467070277346, "grad_norm": 0.07426934689283371, "learning_rate": 4.367791164158991e-07, "loss": 0.0001, "step": 214970 }, { "epoch": 3.517630696228422, "grad_norm": 0.03249650448560715, "learning_rate": 4.3648733115840836e-07, "loss": 0.0003, "step": 214980 }, { "epoch": 3.5177943221794976, "grad_norm": 0.008973855525255203, "learning_rate": 4.361956389470595e-07, "loss": 0.0002, "step": 214990 }, { "epoch": 3.5179579481305736, "grad_norm": 0.02512860670685768, "learning_rate": 4.359040397878006e-07, "loss": 0.0005, "step": 215000 }, { "epoch": 3.5181215740816496, "grad_norm": 0.0010749420616775751, "learning_rate": 4.3561253368657743e-07, "loss": 0.0001, "step": 215010 }, { "epoch": 3.518285200032725, "grad_norm": 0.05400378257036209, "learning_rate": 4.353211206493324e-07, "loss": 0.0002, "step": 215020 }, { "epoch": 3.518448825983801, "grad_norm": 0.009904909878969193, "learning_rate": 4.3502980068200696e-07, "loss": 0.0001, "step": 215030 }, { "epoch": 3.5186124519348767, "grad_norm": 0.016967441886663437, "learning_rate": 4.3473857379054183e-07, "loss": 0.0003, "step": 215040 }, { "epoch": 3.5187760778859527, "grad_norm": 0.018081078305840492, "learning_rate": 4.344474399808751e-07, "loss": 0.0003, "step": 215050 }, { "epoch": 3.5189397038370287, "grad_norm": 0.008498142473399639, "learning_rate": 4.3415639925894203e-07, "loss": 0.0002, "step": 215060 }, { "epoch": 3.5191033297881043, "grad_norm": 0.005203763488680124, "learning_rate": 4.338654516306762e-07, "loss": 0.0002, "step": 215070 }, { "epoch": 3.5192669557391802, "grad_norm": 0.028918027877807617, "learning_rate": 4.335745971020111e-07, "loss": 0.0002, "step": 215080 }, { "epoch": 3.519430581690256, "grad_norm": 0.0071930233389139175, "learning_rate": 4.3328383567887654e-07, "loss": 0.0002, "step": 215090 }, { "epoch": 3.519594207641332, "grad_norm": 0.002062887419015169, "learning_rate": 4.3299316736719997e-07, "loss": 0.0003, "step": 215100 }, { "epoch": 3.519757833592408, "grad_norm": 0.026352746412158012, "learning_rate": 4.327025921729089e-07, "loss": 0.0002, "step": 215110 }, { "epoch": 3.5199214595434833, "grad_norm": 0.023818159475922585, "learning_rate": 4.324121101019274e-07, "loss": 0.0004, "step": 215120 }, { "epoch": 3.5200850854945593, "grad_norm": 0.025201037526130676, "learning_rate": 4.321217211601797e-07, "loss": 0.0003, "step": 215130 }, { "epoch": 3.5202487114456353, "grad_norm": 0.01891115866601467, "learning_rate": 4.318314253535838e-07, "loss": 0.0003, "step": 215140 }, { "epoch": 3.520412337396711, "grad_norm": 0.010590484365820885, "learning_rate": 4.3154122268806053e-07, "loss": 0.0001, "step": 215150 }, { "epoch": 3.520575963347787, "grad_norm": 0.00022928327962290496, "learning_rate": 4.312511131695263e-07, "loss": 0.0001, "step": 215160 }, { "epoch": 3.520739589298863, "grad_norm": 0.015043245628476143, "learning_rate": 4.3096109680389684e-07, "loss": 0.0004, "step": 215170 }, { "epoch": 3.5209032152499384, "grad_norm": 0.02152710035443306, "learning_rate": 4.306711735970842e-07, "loss": 0.0002, "step": 215180 }, { "epoch": 3.5210668412010144, "grad_norm": 0.07278916984796524, "learning_rate": 4.3038134355500025e-07, "loss": 0.0004, "step": 215190 }, { "epoch": 3.5212304671520904, "grad_norm": 0.005619295872747898, "learning_rate": 4.300916066835542e-07, "loss": 0.0004, "step": 215200 }, { "epoch": 3.521394093103166, "grad_norm": 0.013059843331575394, "learning_rate": 4.298019629886546e-07, "loss": 0.0006, "step": 215210 }, { "epoch": 3.521557719054242, "grad_norm": 0.0026604989543557167, "learning_rate": 4.295124124762057e-07, "loss": 0.0003, "step": 215220 }, { "epoch": 3.521721345005318, "grad_norm": 0.018150614574551582, "learning_rate": 4.2922295515211097e-07, "loss": 0.0003, "step": 215230 }, { "epoch": 3.5218849709563935, "grad_norm": 0.0053949817083776, "learning_rate": 4.289335910222736e-07, "loss": 0.0002, "step": 215240 }, { "epoch": 3.5220485969074695, "grad_norm": 0.03307148441672325, "learning_rate": 4.2864432009259325e-07, "loss": 0.0003, "step": 215250 }, { "epoch": 3.5222122228585455, "grad_norm": 0.0006495397537946701, "learning_rate": 4.2835514236896635e-07, "loss": 0.0001, "step": 215260 }, { "epoch": 3.522375848809621, "grad_norm": 0.03571642190217972, "learning_rate": 4.280660578572904e-07, "loss": 0.0002, "step": 215270 }, { "epoch": 3.522539474760697, "grad_norm": 0.0026793975848704576, "learning_rate": 4.2777706656345895e-07, "loss": 0.0002, "step": 215280 }, { "epoch": 3.522703100711773, "grad_norm": 0.023457273840904236, "learning_rate": 4.274881684933657e-07, "loss": 0.0003, "step": 215290 }, { "epoch": 3.5228667266628486, "grad_norm": 0.059548959136009216, "learning_rate": 4.271993636528993e-07, "loss": 0.0003, "step": 215300 }, { "epoch": 3.5230303526139246, "grad_norm": 0.0029258262366056442, "learning_rate": 4.2691065204794824e-07, "loss": 0.0002, "step": 215310 }, { "epoch": 3.5231939785650006, "grad_norm": 0.028560958802700043, "learning_rate": 4.2662203368440014e-07, "loss": 0.0006, "step": 215320 }, { "epoch": 3.523357604516076, "grad_norm": 0.01432044617831707, "learning_rate": 4.263335085681397e-07, "loss": 0.0002, "step": 215330 }, { "epoch": 3.523521230467152, "grad_norm": 0.046106886118650436, "learning_rate": 4.260450767050489e-07, "loss": 0.0002, "step": 215340 }, { "epoch": 3.523684856418228, "grad_norm": 0.012803059071302414, "learning_rate": 4.257567381010097e-07, "loss": 0.0003, "step": 215350 }, { "epoch": 3.5238484823693037, "grad_norm": 0.05045823007822037, "learning_rate": 4.254684927618985e-07, "loss": 0.0003, "step": 215360 }, { "epoch": 3.5240121083203797, "grad_norm": 0.020280759781599045, "learning_rate": 4.251803406935956e-07, "loss": 0.0004, "step": 215370 }, { "epoch": 3.5241757342714557, "grad_norm": 0.019470464438199997, "learning_rate": 4.2489228190197463e-07, "loss": 0.0002, "step": 215380 }, { "epoch": 3.5243393602225312, "grad_norm": 0.04677274823188782, "learning_rate": 4.246043163929092e-07, "loss": 0.0001, "step": 215390 }, { "epoch": 3.5245029861736072, "grad_norm": 0.033394791185855865, "learning_rate": 4.2431644417226913e-07, "loss": 0.0003, "step": 215400 }, { "epoch": 3.5246666121246832, "grad_norm": 0.01906268298625946, "learning_rate": 4.2402866524592743e-07, "loss": 0.0006, "step": 215410 }, { "epoch": 3.524830238075759, "grad_norm": 0.053679514676332474, "learning_rate": 4.2374097961974834e-07, "loss": 0.0004, "step": 215420 }, { "epoch": 3.524993864026835, "grad_norm": 0.00010091160947922617, "learning_rate": 4.234533872995994e-07, "loss": 0.0002, "step": 215430 }, { "epoch": 3.5251574899779103, "grad_norm": 0.0051349676214158535, "learning_rate": 4.2316588829134197e-07, "loss": 0.0004, "step": 215440 }, { "epoch": 3.5253211159289863, "grad_norm": 0.044566236436367035, "learning_rate": 4.2287848260084143e-07, "loss": 0.0003, "step": 215450 }, { "epoch": 3.5254847418800623, "grad_norm": 0.08626312762498856, "learning_rate": 4.2259117023395525e-07, "loss": 0.0004, "step": 215460 }, { "epoch": 3.525648367831138, "grad_norm": 0.036553043872117996, "learning_rate": 4.2230395119654267e-07, "loss": 0.0001, "step": 215470 }, { "epoch": 3.525811993782214, "grad_norm": 0.019678371027112007, "learning_rate": 4.220168254944579e-07, "loss": 0.0002, "step": 215480 }, { "epoch": 3.5259756197332894, "grad_norm": 0.041342951357364655, "learning_rate": 4.2172979313355846e-07, "loss": 0.0003, "step": 215490 }, { "epoch": 3.5261392456843654, "grad_norm": 0.020687635987997055, "learning_rate": 4.2144285411969356e-07, "loss": 0.0, "step": 215500 }, { "epoch": 3.5263028716354414, "grad_norm": 0.014345495961606503, "learning_rate": 4.2115600845871627e-07, "loss": 0.0003, "step": 215510 }, { "epoch": 3.526466497586517, "grad_norm": 0.01960095576941967, "learning_rate": 4.2086925615647143e-07, "loss": 0.0002, "step": 215520 }, { "epoch": 3.526630123537593, "grad_norm": 0.040990784764289856, "learning_rate": 4.2058259721881044e-07, "loss": 0.0001, "step": 215530 }, { "epoch": 3.526793749488669, "grad_norm": 0.015080425888299942, "learning_rate": 4.202960316515747e-07, "loss": 0.0002, "step": 215540 }, { "epoch": 3.5269573754397445, "grad_norm": 0.0034677612129598856, "learning_rate": 4.200095594606085e-07, "loss": 0.0001, "step": 215550 }, { "epoch": 3.5271210013908205, "grad_norm": 0.019140074029564857, "learning_rate": 4.1972318065175156e-07, "loss": 0.0001, "step": 215560 }, { "epoch": 3.5272846273418965, "grad_norm": 0.013363776728510857, "learning_rate": 4.1943689523084365e-07, "loss": 0.0003, "step": 215570 }, { "epoch": 3.527448253292972, "grad_norm": 0.060231857001781464, "learning_rate": 4.1915070320372175e-07, "loss": 0.0003, "step": 215580 }, { "epoch": 3.527611879244048, "grad_norm": 0.05332499369978905, "learning_rate": 4.188646045762218e-07, "loss": 0.0002, "step": 215590 }, { "epoch": 3.527775505195124, "grad_norm": 0.01903146132826805, "learning_rate": 4.185785993541758e-07, "loss": 0.0002, "step": 215600 }, { "epoch": 3.5279391311461996, "grad_norm": 0.0008013157057575881, "learning_rate": 4.1829268754341566e-07, "loss": 0.0002, "step": 215610 }, { "epoch": 3.5281027570972756, "grad_norm": 0.0011676111025735736, "learning_rate": 4.180068691497713e-07, "loss": 0.0002, "step": 215620 }, { "epoch": 3.5282663830483516, "grad_norm": 6.678573845420033e-05, "learning_rate": 4.1772114417907074e-07, "loss": 0.0002, "step": 215630 }, { "epoch": 3.528430008999427, "grad_norm": 0.0019092047587037086, "learning_rate": 4.174355126371382e-07, "loss": 0.0002, "step": 215640 }, { "epoch": 3.528593634950503, "grad_norm": 0.012724940665066242, "learning_rate": 4.17149974529798e-07, "loss": 0.0004, "step": 215650 }, { "epoch": 3.528757260901579, "grad_norm": 0.0032230529468506575, "learning_rate": 4.1686452986287205e-07, "loss": 0.0002, "step": 215660 }, { "epoch": 3.5289208868526547, "grad_norm": 0.04484075680375099, "learning_rate": 4.165791786421819e-07, "loss": 0.0003, "step": 215670 }, { "epoch": 3.5290845128037307, "grad_norm": 0.024478185921907425, "learning_rate": 4.162939208735428e-07, "loss": 0.0001, "step": 215680 }, { "epoch": 3.5292481387548067, "grad_norm": 0.024554841220378876, "learning_rate": 4.1600875656277294e-07, "loss": 0.0003, "step": 215690 }, { "epoch": 3.5294117647058822, "grad_norm": 0.0003324239805806428, "learning_rate": 4.15723685715686e-07, "loss": 0.0002, "step": 215700 }, { "epoch": 3.5295753906569582, "grad_norm": 0.005521126091480255, "learning_rate": 4.154387083380945e-07, "loss": 0.0001, "step": 215710 }, { "epoch": 3.5297390166080342, "grad_norm": 0.02385130152106285, "learning_rate": 4.151538244358083e-07, "loss": 0.0003, "step": 215720 }, { "epoch": 3.52990264255911, "grad_norm": 0.04241269826889038, "learning_rate": 4.148690340146366e-07, "loss": 0.0003, "step": 215730 }, { "epoch": 3.530066268510186, "grad_norm": 0.03556814789772034, "learning_rate": 4.145843370803854e-07, "loss": 0.0001, "step": 215740 }, { "epoch": 3.530229894461262, "grad_norm": 0.0011437456123530865, "learning_rate": 4.14299733638861e-07, "loss": 0.0004, "step": 215750 }, { "epoch": 3.5303935204123373, "grad_norm": 0.016779063269495964, "learning_rate": 4.140152236958639e-07, "loss": 0.0001, "step": 215760 }, { "epoch": 3.5305571463634133, "grad_norm": 0.029900657013058662, "learning_rate": 4.1373080725719605e-07, "loss": 0.0003, "step": 215770 }, { "epoch": 3.5307207723144893, "grad_norm": 0.0028952634893357754, "learning_rate": 4.1344648432865677e-07, "loss": 0.0002, "step": 215780 }, { "epoch": 3.530884398265565, "grad_norm": 0.01657460443675518, "learning_rate": 4.131622549160436e-07, "loss": 0.0002, "step": 215790 }, { "epoch": 3.531048024216641, "grad_norm": 0.049463167786598206, "learning_rate": 4.128781190251502e-07, "loss": 0.0001, "step": 215800 }, { "epoch": 3.531211650167717, "grad_norm": 0.11887083947658539, "learning_rate": 4.125940766617714e-07, "loss": 0.0002, "step": 215810 }, { "epoch": 3.5313752761187924, "grad_norm": 0.007516348268836737, "learning_rate": 4.123101278316971e-07, "loss": 0.0002, "step": 215820 }, { "epoch": 3.5315389020698684, "grad_norm": 0.0007383335614576936, "learning_rate": 4.120262725407187e-07, "loss": 0.0002, "step": 215830 }, { "epoch": 3.531702528020944, "grad_norm": 0.001630987273529172, "learning_rate": 4.1174251079462215e-07, "loss": 0.0003, "step": 215840 }, { "epoch": 3.53186615397202, "grad_norm": 0.003209742484614253, "learning_rate": 4.114588425991933e-07, "loss": 0.0002, "step": 215850 }, { "epoch": 3.532029779923096, "grad_norm": 0.0014113730285316706, "learning_rate": 4.1117526796021655e-07, "loss": 0.0002, "step": 215860 }, { "epoch": 3.5321934058741715, "grad_norm": 0.013080277480185032, "learning_rate": 4.1089178688347386e-07, "loss": 0.0004, "step": 215870 }, { "epoch": 3.5323570318252475, "grad_norm": 0.008914799429476261, "learning_rate": 4.1060839937474396e-07, "loss": 0.0002, "step": 215880 }, { "epoch": 3.532520657776323, "grad_norm": 0.001589714316651225, "learning_rate": 4.1032510543980606e-07, "loss": 0.0004, "step": 215890 }, { "epoch": 3.532684283727399, "grad_norm": 0.009632579050958157, "learning_rate": 4.1004190508443564e-07, "loss": 0.0003, "step": 215900 }, { "epoch": 3.532847909678475, "grad_norm": 0.0479317232966423, "learning_rate": 4.09758798314408e-07, "loss": 0.0005, "step": 215910 }, { "epoch": 3.5330115356295506, "grad_norm": 0.015169399790465832, "learning_rate": 4.0947578513549356e-07, "loss": 0.0002, "step": 215920 }, { "epoch": 3.5331751615806266, "grad_norm": 0.08976619690656662, "learning_rate": 4.091928655534638e-07, "loss": 0.0004, "step": 215930 }, { "epoch": 3.5333387875317026, "grad_norm": 0.00021065844339318573, "learning_rate": 4.089100395740875e-07, "loss": 0.0003, "step": 215940 }, { "epoch": 3.533502413482778, "grad_norm": 0.06309055536985397, "learning_rate": 4.086273072031316e-07, "loss": 0.0003, "step": 215950 }, { "epoch": 3.533666039433854, "grad_norm": 0.02916092425584793, "learning_rate": 4.083446684463593e-07, "loss": 0.0002, "step": 215960 }, { "epoch": 3.53382966538493, "grad_norm": 0.0007907498511485755, "learning_rate": 4.080621233095339e-07, "loss": 0.0001, "step": 215970 }, { "epoch": 3.5339932913360057, "grad_norm": 0.00039486901368945837, "learning_rate": 4.077796717984167e-07, "loss": 0.0003, "step": 215980 }, { "epoch": 3.5341569172870817, "grad_norm": 0.007465478498488665, "learning_rate": 4.074973139187671e-07, "loss": 0.0004, "step": 215990 }, { "epoch": 3.5343205432381577, "grad_norm": 0.04131713882088661, "learning_rate": 4.072150496763411e-07, "loss": 0.0002, "step": 216000 }, { "epoch": 3.5344841691892332, "grad_norm": 0.10793514549732208, "learning_rate": 4.069328790768945e-07, "loss": 0.0003, "step": 216010 }, { "epoch": 3.5346477951403092, "grad_norm": 0.057394761592149734, "learning_rate": 4.066508021261789e-07, "loss": 0.0002, "step": 216020 }, { "epoch": 3.5348114210913852, "grad_norm": 0.01246943511068821, "learning_rate": 4.063688188299486e-07, "loss": 0.0002, "step": 216030 }, { "epoch": 3.534975047042461, "grad_norm": 0.04247612878680229, "learning_rate": 4.060869291939501e-07, "loss": 0.0002, "step": 216040 }, { "epoch": 3.535138672993537, "grad_norm": 0.00563997495919466, "learning_rate": 4.058051332239338e-07, "loss": 0.0002, "step": 216050 }, { "epoch": 3.535302298944613, "grad_norm": 0.00047797596198506653, "learning_rate": 4.0552343092564115e-07, "loss": 0.0002, "step": 216060 }, { "epoch": 3.5354659248956883, "grad_norm": 0.021742159500718117, "learning_rate": 4.0524182230482046e-07, "loss": 0.0002, "step": 216070 }, { "epoch": 3.5356295508467643, "grad_norm": 0.023918164893984795, "learning_rate": 4.049603073672098e-07, "loss": 0.0003, "step": 216080 }, { "epoch": 3.5357931767978403, "grad_norm": 0.008372410200536251, "learning_rate": 4.0467888611855187e-07, "loss": 0.0004, "step": 216090 }, { "epoch": 3.535956802748916, "grad_norm": 0.0009009564528241754, "learning_rate": 4.0439755856458153e-07, "loss": 0.0001, "step": 216100 }, { "epoch": 3.536120428699992, "grad_norm": 0.023796197026968002, "learning_rate": 4.041163247110386e-07, "loss": 0.0002, "step": 216110 }, { "epoch": 3.536284054651068, "grad_norm": 0.00014997433754615486, "learning_rate": 4.0383518456365346e-07, "loss": 0.0007, "step": 216120 }, { "epoch": 3.5364476806021434, "grad_norm": 0.0007433267892338336, "learning_rate": 4.0355413812816155e-07, "loss": 0.0004, "step": 216130 }, { "epoch": 3.5366113065532194, "grad_norm": 0.02827180176973343, "learning_rate": 4.0327318541028947e-07, "loss": 0.0003, "step": 216140 }, { "epoch": 3.5367749325042954, "grad_norm": 0.028769154101610184, "learning_rate": 4.029923264157698e-07, "loss": 0.0002, "step": 216150 }, { "epoch": 3.536938558455371, "grad_norm": 0.010946427471935749, "learning_rate": 4.0271156115032625e-07, "loss": 0.0002, "step": 216160 }, { "epoch": 3.537102184406447, "grad_norm": 0.07757820188999176, "learning_rate": 4.024308896196849e-07, "loss": 0.0003, "step": 216170 }, { "epoch": 3.537265810357523, "grad_norm": 0.0010004934156313539, "learning_rate": 4.0215031182956665e-07, "loss": 0.0004, "step": 216180 }, { "epoch": 3.5374294363085985, "grad_norm": 0.00837447214871645, "learning_rate": 4.0186982778569304e-07, "loss": 0.0001, "step": 216190 }, { "epoch": 3.5375930622596745, "grad_norm": 0.0017423003446310759, "learning_rate": 4.015894374937829e-07, "loss": 0.0002, "step": 216200 }, { "epoch": 3.53775668821075, "grad_norm": 0.06378038972616196, "learning_rate": 4.0130914095955486e-07, "loss": 0.0002, "step": 216210 }, { "epoch": 3.537920314161826, "grad_norm": 0.022798560559749603, "learning_rate": 4.010289381887211e-07, "loss": 0.0003, "step": 216220 }, { "epoch": 3.538083940112902, "grad_norm": 0.00027165646315552294, "learning_rate": 4.00748829186996e-07, "loss": 0.0002, "step": 216230 }, { "epoch": 3.5382475660639776, "grad_norm": 0.004613944329321384, "learning_rate": 4.0046881396009096e-07, "loss": 0.0001, "step": 216240 }, { "epoch": 3.5384111920150536, "grad_norm": 0.02765141986310482, "learning_rate": 4.001888925137154e-07, "loss": 0.0002, "step": 216250 }, { "epoch": 3.538574817966129, "grad_norm": 0.004503559321165085, "learning_rate": 3.9990906485357583e-07, "loss": 0.0003, "step": 216260 }, { "epoch": 3.538738443917205, "grad_norm": 0.007174454629421234, "learning_rate": 3.9962933098537825e-07, "loss": 0.0004, "step": 216270 }, { "epoch": 3.538902069868281, "grad_norm": 0.05483933538198471, "learning_rate": 3.9934969091482647e-07, "loss": 0.0004, "step": 216280 }, { "epoch": 3.5390656958193567, "grad_norm": 0.013119247741997242, "learning_rate": 3.9907014464762194e-07, "loss": 0.0003, "step": 216290 }, { "epoch": 3.5392293217704327, "grad_norm": 0.02354021929204464, "learning_rate": 3.9879069218946354e-07, "loss": 0.0002, "step": 216300 }, { "epoch": 3.5393929477215087, "grad_norm": 0.041345398873090744, "learning_rate": 3.9851133354604996e-07, "loss": 0.0002, "step": 216310 }, { "epoch": 3.5395565736725842, "grad_norm": 0.0005942419520579278, "learning_rate": 3.9823206872307674e-07, "loss": 0.0002, "step": 216320 }, { "epoch": 3.5397201996236602, "grad_norm": 0.06468115746974945, "learning_rate": 3.979528977262387e-07, "loss": 0.0003, "step": 216330 }, { "epoch": 3.5398838255747362, "grad_norm": 0.06807846575975418, "learning_rate": 3.9767382056122685e-07, "loss": 0.0005, "step": 216340 }, { "epoch": 3.540047451525812, "grad_norm": 0.02225060760974884, "learning_rate": 3.973948372337311e-07, "loss": 0.0002, "step": 216350 }, { "epoch": 3.540211077476888, "grad_norm": 0.024910537526011467, "learning_rate": 3.9711594774944015e-07, "loss": 0.0006, "step": 216360 }, { "epoch": 3.540374703427964, "grad_norm": 6.576843588845804e-05, "learning_rate": 3.9683715211404186e-07, "loss": 0.0002, "step": 216370 }, { "epoch": 3.5405383293790393, "grad_norm": 0.04459504410624504, "learning_rate": 3.9655845033321814e-07, "loss": 0.0002, "step": 216380 }, { "epoch": 3.5407019553301153, "grad_norm": 0.003975185099989176, "learning_rate": 3.962798424126524e-07, "loss": 0.0001, "step": 216390 }, { "epoch": 3.5408655812811913, "grad_norm": 0.04442526772618294, "learning_rate": 3.9600132835802494e-07, "loss": 0.0004, "step": 216400 }, { "epoch": 3.541029207232267, "grad_norm": 0.010061725974082947, "learning_rate": 3.9572290817501634e-07, "loss": 0.0006, "step": 216410 }, { "epoch": 3.541192833183343, "grad_norm": 0.0007871570414863527, "learning_rate": 3.954445818693009e-07, "loss": 0.0003, "step": 216420 }, { "epoch": 3.541356459134419, "grad_norm": 0.019940445199608803, "learning_rate": 3.951663494465541e-07, "loss": 0.0003, "step": 216430 }, { "epoch": 3.5415200850854944, "grad_norm": 0.05407560616731644, "learning_rate": 3.948882109124491e-07, "loss": 0.0002, "step": 216440 }, { "epoch": 3.5416837110365704, "grad_norm": 0.08028872311115265, "learning_rate": 3.9461016627265814e-07, "loss": 0.0005, "step": 216450 }, { "epoch": 3.5418473369876464, "grad_norm": 0.00030924740713089705, "learning_rate": 3.943322155328483e-07, "loss": 0.0002, "step": 216460 }, { "epoch": 3.542010962938722, "grad_norm": 0.001323656877502799, "learning_rate": 3.9405435869868724e-07, "loss": 0.0002, "step": 216470 }, { "epoch": 3.542174588889798, "grad_norm": 0.05367785692214966, "learning_rate": 3.9377659577584105e-07, "loss": 0.0003, "step": 216480 }, { "epoch": 3.542338214840874, "grad_norm": 0.00029197437106631696, "learning_rate": 3.934989267699735e-07, "loss": 0.0003, "step": 216490 }, { "epoch": 3.5425018407919495, "grad_norm": 0.004028506577014923, "learning_rate": 3.9322135168674393e-07, "loss": 0.0003, "step": 216500 }, { "epoch": 3.5426654667430255, "grad_norm": 0.01796489767730236, "learning_rate": 3.9294387053181284e-07, "loss": 0.0001, "step": 216510 }, { "epoch": 3.5428290926941015, "grad_norm": 0.1913430392742157, "learning_rate": 3.9266648331083847e-07, "loss": 0.0004, "step": 216520 }, { "epoch": 3.542992718645177, "grad_norm": 0.12513013184070587, "learning_rate": 3.9238919002947686e-07, "loss": 0.0002, "step": 216530 }, { "epoch": 3.543156344596253, "grad_norm": 0.00897456519305706, "learning_rate": 3.921119906933807e-07, "loss": 0.0003, "step": 216540 }, { "epoch": 3.543319970547329, "grad_norm": 0.06942382454872131, "learning_rate": 3.918348853082016e-07, "loss": 0.0003, "step": 216550 }, { "epoch": 3.5434835964984046, "grad_norm": 0.01881391741335392, "learning_rate": 3.9155787387959e-07, "loss": 0.0002, "step": 216560 }, { "epoch": 3.5436472224494806, "grad_norm": 0.0027428697794675827, "learning_rate": 3.912809564131953e-07, "loss": 0.0002, "step": 216570 }, { "epoch": 3.5438108484005566, "grad_norm": 0.004028898198157549, "learning_rate": 3.9100413291466133e-07, "loss": 0.0002, "step": 216580 }, { "epoch": 3.543974474351632, "grad_norm": 0.04047404229640961, "learning_rate": 3.90727403389633e-07, "loss": 0.0008, "step": 216590 }, { "epoch": 3.544138100302708, "grad_norm": 0.011440388858318329, "learning_rate": 3.90450767843753e-07, "loss": 0.0001, "step": 216600 }, { "epoch": 3.5443017262537837, "grad_norm": 0.0010043795919045806, "learning_rate": 3.9017422628266245e-07, "loss": 0.0003, "step": 216610 }, { "epoch": 3.5444653522048597, "grad_norm": 0.04644084721803665, "learning_rate": 3.8989777871199843e-07, "loss": 0.0001, "step": 216620 }, { "epoch": 3.5446289781559357, "grad_norm": 0.0002301383501617238, "learning_rate": 3.896214251373981e-07, "loss": 0.0002, "step": 216630 }, { "epoch": 3.5447926041070112, "grad_norm": 0.014492087997496128, "learning_rate": 3.8934516556449474e-07, "loss": 0.0003, "step": 216640 }, { "epoch": 3.5449562300580872, "grad_norm": 3.9210444811033085e-05, "learning_rate": 3.890689999989239e-07, "loss": 0.0002, "step": 216650 }, { "epoch": 3.545119856009163, "grad_norm": 0.024685220792889595, "learning_rate": 3.887929284463138e-07, "loss": 0.0005, "step": 216660 }, { "epoch": 3.545283481960239, "grad_norm": 0.0490608811378479, "learning_rate": 3.885169509122949e-07, "loss": 0.0002, "step": 216670 }, { "epoch": 3.545447107911315, "grad_norm": 0.08608243614435196, "learning_rate": 3.882410674024917e-07, "loss": 0.0002, "step": 216680 }, { "epoch": 3.5456107338623903, "grad_norm": 0.03502362594008446, "learning_rate": 3.8796527792253293e-07, "loss": 0.0002, "step": 216690 }, { "epoch": 3.5457743598134663, "grad_norm": 0.008822974748909473, "learning_rate": 3.8768958247803857e-07, "loss": 0.0001, "step": 216700 }, { "epoch": 3.5459379857645423, "grad_norm": 0.002094449009746313, "learning_rate": 3.874139810746319e-07, "loss": 0.0003, "step": 216710 }, { "epoch": 3.546101611715618, "grad_norm": 0.0019721167627722025, "learning_rate": 3.8713847371792954e-07, "loss": 0.0001, "step": 216720 }, { "epoch": 3.546265237666694, "grad_norm": 0.021950753405690193, "learning_rate": 3.868630604135526e-07, "loss": 0.0001, "step": 216730 }, { "epoch": 3.54642886361777, "grad_norm": 0.005251083057373762, "learning_rate": 3.865877411671137e-07, "loss": 0.0002, "step": 216740 }, { "epoch": 3.5465924895688454, "grad_norm": 0.010564872995018959, "learning_rate": 3.863125159842279e-07, "loss": 0.0004, "step": 216750 }, { "epoch": 3.5467561155199214, "grad_norm": 0.0016803601756691933, "learning_rate": 3.8603738487050457e-07, "loss": 0.0001, "step": 216760 }, { "epoch": 3.5469197414709974, "grad_norm": 0.0020147268660366535, "learning_rate": 3.857623478315564e-07, "loss": 0.0002, "step": 216770 }, { "epoch": 3.547083367422073, "grad_norm": 0.011834284290671349, "learning_rate": 3.8548740487298895e-07, "loss": 0.0002, "step": 216780 }, { "epoch": 3.547246993373149, "grad_norm": 0.0005450216121971607, "learning_rate": 3.8521255600040944e-07, "loss": 0.0001, "step": 216790 }, { "epoch": 3.547410619324225, "grad_norm": 1.812165282899514e-05, "learning_rate": 3.8493780121941935e-07, "loss": 0.0004, "step": 216800 }, { "epoch": 3.5475742452753005, "grad_norm": 0.04345245659351349, "learning_rate": 3.8466314053562495e-07, "loss": 0.0003, "step": 216810 }, { "epoch": 3.5477378712263765, "grad_norm": 0.008363474160432816, "learning_rate": 3.843885739546227e-07, "loss": 0.0002, "step": 216820 }, { "epoch": 3.5479014971774525, "grad_norm": 0.030983032658696175, "learning_rate": 3.841141014820127e-07, "loss": 0.0001, "step": 216830 }, { "epoch": 3.548065123128528, "grad_norm": 0.018794113770127296, "learning_rate": 3.838397231233898e-07, "loss": 0.0001, "step": 216840 }, { "epoch": 3.548228749079604, "grad_norm": 0.040520139038562775, "learning_rate": 3.835654388843485e-07, "loss": 0.0002, "step": 216850 }, { "epoch": 3.54839237503068, "grad_norm": 0.0038775019347667694, "learning_rate": 3.832912487704826e-07, "loss": 0.0002, "step": 216860 }, { "epoch": 3.5485560009817556, "grad_norm": 0.005229289643466473, "learning_rate": 3.830171527873822e-07, "loss": 0.0002, "step": 216870 }, { "epoch": 3.5487196269328316, "grad_norm": 0.15319187939167023, "learning_rate": 3.827431509406343e-07, "loss": 0.0006, "step": 216880 }, { "epoch": 3.5488832528839076, "grad_norm": 0.0024291244335472584, "learning_rate": 3.8246924323582733e-07, "loss": 0.0002, "step": 216890 }, { "epoch": 3.549046878834983, "grad_norm": 0.04642331227660179, "learning_rate": 3.821954296785452e-07, "loss": 0.0002, "step": 216900 }, { "epoch": 3.549210504786059, "grad_norm": 0.05263778194785118, "learning_rate": 3.8192171027437163e-07, "loss": 0.0004, "step": 216910 }, { "epoch": 3.549374130737135, "grad_norm": 0.008172739297151566, "learning_rate": 3.816480850288862e-07, "loss": 0.0001, "step": 216920 }, { "epoch": 3.5495377566882107, "grad_norm": 0.05520511418581009, "learning_rate": 3.813745539476682e-07, "loss": 0.0003, "step": 216930 }, { "epoch": 3.5497013826392867, "grad_norm": 0.012372137978672981, "learning_rate": 3.811011170362955e-07, "loss": 0.0002, "step": 216940 }, { "epoch": 3.5498650085903627, "grad_norm": 0.005406411364674568, "learning_rate": 3.8082777430034356e-07, "loss": 0.0008, "step": 216950 }, { "epoch": 3.5500286345414382, "grad_norm": 0.044761523604393005, "learning_rate": 3.805545257453841e-07, "loss": 0.0002, "step": 216960 }, { "epoch": 3.5501922604925142, "grad_norm": 0.004698908422142267, "learning_rate": 3.8028137137698875e-07, "loss": 0.0002, "step": 216970 }, { "epoch": 3.55035588644359, "grad_norm": 0.13810069859027863, "learning_rate": 3.8000831120072747e-07, "loss": 0.0003, "step": 216980 }, { "epoch": 3.550519512394666, "grad_norm": 0.0427342988550663, "learning_rate": 3.797353452221686e-07, "loss": 0.0002, "step": 216990 }, { "epoch": 3.550683138345742, "grad_norm": 0.0005482430569827557, "learning_rate": 3.794624734468755e-07, "loss": 0.0002, "step": 217000 }, { "epoch": 3.5508467642968173, "grad_norm": 0.06521736085414886, "learning_rate": 3.791896958804131e-07, "loss": 0.0002, "step": 217010 }, { "epoch": 3.5510103902478933, "grad_norm": 0.03342504799365997, "learning_rate": 3.789170125283431e-07, "loss": 0.0004, "step": 217020 }, { "epoch": 3.551174016198969, "grad_norm": 0.03518366813659668, "learning_rate": 3.7864442339622556e-07, "loss": 0.0002, "step": 217030 }, { "epoch": 3.551337642150045, "grad_norm": 0.0011166025651618838, "learning_rate": 3.7837192848961757e-07, "loss": 0.0001, "step": 217040 }, { "epoch": 3.551501268101121, "grad_norm": 0.0035820333287119865, "learning_rate": 3.780995278140748e-07, "loss": 0.0002, "step": 217050 }, { "epoch": 3.5516648940521964, "grad_norm": 0.05932708829641342, "learning_rate": 3.7782722137515213e-07, "loss": 0.0002, "step": 217060 }, { "epoch": 3.5518285200032724, "grad_norm": 0.04122702404856682, "learning_rate": 3.775550091784025e-07, "loss": 0.0002, "step": 217070 }, { "epoch": 3.5519921459543484, "grad_norm": 0.022375404834747314, "learning_rate": 3.772828912293741e-07, "loss": 0.0004, "step": 217080 }, { "epoch": 3.552155771905424, "grad_norm": 0.038946691900491714, "learning_rate": 3.770108675336159e-07, "loss": 0.0002, "step": 217090 }, { "epoch": 3.5523193978565, "grad_norm": 0.014711924828588963, "learning_rate": 3.767389380966746e-07, "loss": 0.0004, "step": 217100 }, { "epoch": 3.552483023807576, "grad_norm": 0.005257236771285534, "learning_rate": 3.7646710292409463e-07, "loss": 0.0002, "step": 217110 }, { "epoch": 3.5526466497586515, "grad_norm": 0.06583445519208908, "learning_rate": 3.7619536202141813e-07, "loss": 0.0003, "step": 217120 }, { "epoch": 3.5528102757097275, "grad_norm": 0.022677699103951454, "learning_rate": 3.759237153941853e-07, "loss": 0.0002, "step": 217130 }, { "epoch": 3.5529739016608035, "grad_norm": 0.0020441475789994, "learning_rate": 3.7565216304793594e-07, "loss": 0.0002, "step": 217140 }, { "epoch": 3.553137527611879, "grad_norm": 0.04178673401474953, "learning_rate": 3.753807049882063e-07, "loss": 0.0001, "step": 217150 }, { "epoch": 3.553301153562955, "grad_norm": 0.009881909005343914, "learning_rate": 3.7510934122053034e-07, "loss": 0.0002, "step": 217160 }, { "epoch": 3.553464779514031, "grad_norm": 0.00042053768993355334, "learning_rate": 3.7483807175044184e-07, "loss": 0.0004, "step": 217170 }, { "epoch": 3.5536284054651066, "grad_norm": 0.035308003425598145, "learning_rate": 3.745668965834709e-07, "loss": 0.0002, "step": 217180 }, { "epoch": 3.5537920314161826, "grad_norm": 0.004697485361248255, "learning_rate": 3.7429581572514863e-07, "loss": 0.0002, "step": 217190 }, { "epoch": 3.5539556573672586, "grad_norm": 0.0008168789208866656, "learning_rate": 3.7402482918099946e-07, "loss": 0.0001, "step": 217200 }, { "epoch": 3.554119283318334, "grad_norm": 0.004080076236277819, "learning_rate": 3.737539369565496e-07, "loss": 0.0001, "step": 217210 }, { "epoch": 3.55428290926941, "grad_norm": 0.0017225544434040785, "learning_rate": 3.7348313905732294e-07, "loss": 0.0002, "step": 217220 }, { "epoch": 3.554446535220486, "grad_norm": 0.00019463345233816653, "learning_rate": 3.732124354888411e-07, "loss": 0.0013, "step": 217230 }, { "epoch": 3.5546101611715617, "grad_norm": 0.0063975197263062, "learning_rate": 3.7294182625662145e-07, "loss": 0.0002, "step": 217240 }, { "epoch": 3.5547737871226377, "grad_norm": 0.028400786221027374, "learning_rate": 3.726713113661834e-07, "loss": 0.0001, "step": 217250 }, { "epoch": 3.5549374130737137, "grad_norm": 0.0006264901021495461, "learning_rate": 3.7240089082304145e-07, "loss": 0.0001, "step": 217260 }, { "epoch": 3.5551010390247892, "grad_norm": 0.06403999775648117, "learning_rate": 3.7213056463271113e-07, "loss": 0.0001, "step": 217270 }, { "epoch": 3.5552646649758652, "grad_norm": 0.014691372402012348, "learning_rate": 3.7186033280070145e-07, "loss": 0.0004, "step": 217280 }, { "epoch": 3.5554282909269412, "grad_norm": 0.020447932183742523, "learning_rate": 3.715901953325246e-07, "loss": 0.0001, "step": 217290 }, { "epoch": 3.555591916878017, "grad_norm": 0.05414560064673424, "learning_rate": 3.713201522336851e-07, "loss": 0.0003, "step": 217300 }, { "epoch": 3.555755542829093, "grad_norm": 0.046699438244104385, "learning_rate": 3.710502035096936e-07, "loss": 0.0003, "step": 217310 }, { "epoch": 3.555919168780169, "grad_norm": 0.032360855489969254, "learning_rate": 3.7078034916605e-07, "loss": 0.0004, "step": 217320 }, { "epoch": 3.5560827947312443, "grad_norm": 0.0012636749306693673, "learning_rate": 3.7051058920825943e-07, "loss": 0.0002, "step": 217330 }, { "epoch": 3.5562464206823203, "grad_norm": 0.11685267835855484, "learning_rate": 3.702409236418192e-07, "loss": 0.0008, "step": 217340 }, { "epoch": 3.5564100466333963, "grad_norm": 0.028324194252490997, "learning_rate": 3.6997135247223035e-07, "loss": 0.0003, "step": 217350 }, { "epoch": 3.556573672584472, "grad_norm": 0.05744652450084686, "learning_rate": 3.6970187570498696e-07, "loss": 0.0004, "step": 217360 }, { "epoch": 3.556737298535548, "grad_norm": 0.0042003002017736435, "learning_rate": 3.694324933455856e-07, "loss": 0.0003, "step": 217370 }, { "epoch": 3.5569009244866234, "grad_norm": 0.0039032294880598783, "learning_rate": 3.691632053995159e-07, "loss": 0.0004, "step": 217380 }, { "epoch": 3.5570645504376994, "grad_norm": 0.012822470627725124, "learning_rate": 3.688940118722717e-07, "loss": 0.0004, "step": 217390 }, { "epoch": 3.5572281763887754, "grad_norm": 0.03572256863117218, "learning_rate": 3.6862491276933864e-07, "loss": 0.0003, "step": 217400 }, { "epoch": 3.557391802339851, "grad_norm": 0.06516546756029129, "learning_rate": 3.6835590809620624e-07, "loss": 0.0004, "step": 217410 }, { "epoch": 3.557555428290927, "grad_norm": 0.009016621857881546, "learning_rate": 3.680869978583557e-07, "loss": 0.0004, "step": 217420 }, { "epoch": 3.5577190542420025, "grad_norm": 0.010672484524548054, "learning_rate": 3.678181820612742e-07, "loss": 0.0003, "step": 217430 }, { "epoch": 3.5578826801930785, "grad_norm": 7.558158540632576e-05, "learning_rate": 3.6754946071043907e-07, "loss": 0.0002, "step": 217440 }, { "epoch": 3.5580463061441545, "grad_norm": 0.03521640598773956, "learning_rate": 3.672808338113315e-07, "loss": 0.0001, "step": 217450 }, { "epoch": 3.55820993209523, "grad_norm": 0.0018449359340593219, "learning_rate": 3.670123013694271e-07, "loss": 0.0003, "step": 217460 }, { "epoch": 3.558373558046306, "grad_norm": 0.018947770819067955, "learning_rate": 3.667438633902021e-07, "loss": 0.0001, "step": 217470 }, { "epoch": 3.558537183997382, "grad_norm": 0.04961947724223137, "learning_rate": 3.6647551987912865e-07, "loss": 0.0002, "step": 217480 }, { "epoch": 3.5587008099484576, "grad_norm": 0.023324841633439064, "learning_rate": 3.6620727084167974e-07, "loss": 0.0002, "step": 217490 }, { "epoch": 3.5588644358995336, "grad_norm": 0.0011620724108070135, "learning_rate": 3.6593911628332254e-07, "loss": 0.0002, "step": 217500 }, { "epoch": 3.5590280618506096, "grad_norm": 0.010905135422945023, "learning_rate": 3.6567105620952613e-07, "loss": 0.0002, "step": 217510 }, { "epoch": 3.559191687801685, "grad_norm": 0.00030964004690758884, "learning_rate": 3.65403090625755e-07, "loss": 0.0001, "step": 217520 }, { "epoch": 3.559355313752761, "grad_norm": 0.03806227818131447, "learning_rate": 3.6513521953747467e-07, "loss": 0.0005, "step": 217530 }, { "epoch": 3.559518939703837, "grad_norm": 0.02592477947473526, "learning_rate": 3.648674429501442e-07, "loss": 0.0001, "step": 217540 }, { "epoch": 3.5596825656549127, "grad_norm": 0.08618870377540588, "learning_rate": 3.645997608692248e-07, "loss": 0.0006, "step": 217550 }, { "epoch": 3.5598461916059887, "grad_norm": 0.01949167065322399, "learning_rate": 3.643321733001737e-07, "loss": 0.0002, "step": 217560 }, { "epoch": 3.5600098175570647, "grad_norm": 0.02180432714521885, "learning_rate": 3.640646802484482e-07, "loss": 0.0002, "step": 217570 }, { "epoch": 3.5601734435081402, "grad_norm": 0.01358653325587511, "learning_rate": 3.637972817195001e-07, "loss": 0.0003, "step": 217580 }, { "epoch": 3.5603370694592162, "grad_norm": 0.042856015264987946, "learning_rate": 3.635299777187823e-07, "loss": 0.0003, "step": 217590 }, { "epoch": 3.5605006954102922, "grad_norm": 0.009114561602473259, "learning_rate": 3.6326276825174536e-07, "loss": 0.0002, "step": 217600 }, { "epoch": 3.560664321361368, "grad_norm": 0.09084678441286087, "learning_rate": 3.6299565332383823e-07, "loss": 0.0003, "step": 217610 }, { "epoch": 3.560827947312444, "grad_norm": 0.05225306376814842, "learning_rate": 3.6272863294050497e-07, "loss": 0.0002, "step": 217620 }, { "epoch": 3.56099157326352, "grad_norm": 0.03291582316160202, "learning_rate": 3.624617071071912e-07, "loss": 0.0003, "step": 217630 }, { "epoch": 3.5611551992145953, "grad_norm": 0.030802637338638306, "learning_rate": 3.6219487582933863e-07, "loss": 0.0004, "step": 217640 }, { "epoch": 3.5613188251656713, "grad_norm": 0.02562759630382061, "learning_rate": 3.6192813911238966e-07, "loss": 0.0002, "step": 217650 }, { "epoch": 3.5614824511167473, "grad_norm": 0.0249533299356699, "learning_rate": 3.6166149696178045e-07, "loss": 0.0004, "step": 217660 }, { "epoch": 3.561646077067823, "grad_norm": 0.0005957194371148944, "learning_rate": 3.613949493829483e-07, "loss": 0.0001, "step": 217670 }, { "epoch": 3.561809703018899, "grad_norm": 0.006870143581181765, "learning_rate": 3.611284963813283e-07, "loss": 0.0003, "step": 217680 }, { "epoch": 3.561973328969975, "grad_norm": 0.08325424045324326, "learning_rate": 3.6086213796235393e-07, "loss": 0.0003, "step": 217690 }, { "epoch": 3.5621369549210504, "grad_norm": 0.13628168404102325, "learning_rate": 3.605958741314541e-07, "loss": 0.0006, "step": 217700 }, { "epoch": 3.5623005808721264, "grad_norm": 0.11880140006542206, "learning_rate": 3.603297048940585e-07, "loss": 0.0002, "step": 217710 }, { "epoch": 3.5624642068232024, "grad_norm": 0.025369735434651375, "learning_rate": 3.600636302555949e-07, "loss": 0.0003, "step": 217720 }, { "epoch": 3.562627832774278, "grad_norm": 0.09651242941617966, "learning_rate": 3.597976502214884e-07, "loss": 0.0003, "step": 217730 }, { "epoch": 3.562791458725354, "grad_norm": 0.011989498510956764, "learning_rate": 3.595317647971602e-07, "loss": 0.0003, "step": 217740 }, { "epoch": 3.5629550846764295, "grad_norm": 0.0009898929856717587, "learning_rate": 3.592659739880333e-07, "loss": 0.0003, "step": 217750 }, { "epoch": 3.5631187106275055, "grad_norm": 0.004707247018814087, "learning_rate": 3.5900027779952607e-07, "loss": 0.0001, "step": 217760 }, { "epoch": 3.5632823365785815, "grad_norm": 0.04091743007302284, "learning_rate": 3.587346762370575e-07, "loss": 0.0004, "step": 217770 }, { "epoch": 3.563445962529657, "grad_norm": 0.00131022441200912, "learning_rate": 3.584691693060399e-07, "loss": 0.0004, "step": 217780 }, { "epoch": 3.563609588480733, "grad_norm": 0.03290257975459099, "learning_rate": 3.5820375701188905e-07, "loss": 0.0004, "step": 217790 }, { "epoch": 3.5637732144318086, "grad_norm": 0.0324813574552536, "learning_rate": 3.5793843936001604e-07, "loss": 0.0002, "step": 217800 }, { "epoch": 3.5639368403828846, "grad_norm": 0.0005035446956753731, "learning_rate": 3.576732163558311e-07, "loss": 0.0002, "step": 217810 }, { "epoch": 3.5641004663339606, "grad_norm": 0.12793752551078796, "learning_rate": 3.574080880047398e-07, "loss": 0.0005, "step": 217820 }, { "epoch": 3.564264092285036, "grad_norm": 0.023770883679389954, "learning_rate": 3.571430543121496e-07, "loss": 0.0002, "step": 217830 }, { "epoch": 3.564427718236112, "grad_norm": 0.010323427617549896, "learning_rate": 3.5687811528346386e-07, "loss": 0.0002, "step": 217840 }, { "epoch": 3.564591344187188, "grad_norm": 0.011715898290276527, "learning_rate": 3.566132709240849e-07, "loss": 0.0001, "step": 217850 }, { "epoch": 3.5647549701382637, "grad_norm": 0.009993953630328178, "learning_rate": 3.5634852123941187e-07, "loss": 0.0004, "step": 217860 }, { "epoch": 3.5649185960893397, "grad_norm": 0.036244411021471024, "learning_rate": 3.5608386623484316e-07, "loss": 0.0001, "step": 217870 }, { "epoch": 3.5650822220404157, "grad_norm": 0.0036140496376901865, "learning_rate": 3.5581930591577495e-07, "loss": 0.0001, "step": 217880 }, { "epoch": 3.5652458479914912, "grad_norm": 0.03786827251315117, "learning_rate": 3.555548402876025e-07, "loss": 0.0003, "step": 217890 }, { "epoch": 3.5654094739425672, "grad_norm": 0.006403864827007055, "learning_rate": 3.552904693557158e-07, "loss": 0.0001, "step": 217900 }, { "epoch": 3.5655730998936432, "grad_norm": 0.015000035986304283, "learning_rate": 3.550261931255061e-07, "loss": 0.0003, "step": 217910 }, { "epoch": 3.565736725844719, "grad_norm": 0.0002582675660960376, "learning_rate": 3.54762011602362e-07, "loss": 0.0002, "step": 217920 }, { "epoch": 3.565900351795795, "grad_norm": 0.08964661508798599, "learning_rate": 3.5449792479167076e-07, "loss": 0.0003, "step": 217930 }, { "epoch": 3.566063977746871, "grad_norm": 0.002500242320820689, "learning_rate": 3.5423393269881477e-07, "loss": 0.0005, "step": 217940 }, { "epoch": 3.5662276036979463, "grad_norm": 0.053904835134744644, "learning_rate": 3.539700353291792e-07, "loss": 0.0002, "step": 217950 }, { "epoch": 3.5663912296490223, "grad_norm": 0.04794420674443245, "learning_rate": 3.537062326881413e-07, "loss": 0.0008, "step": 217960 }, { "epoch": 3.5665548556000983, "grad_norm": 0.0399046391248703, "learning_rate": 3.5344252478108356e-07, "loss": 0.0003, "step": 217970 }, { "epoch": 3.566718481551174, "grad_norm": 0.010266410186886787, "learning_rate": 3.531789116133799e-07, "loss": 0.0002, "step": 217980 }, { "epoch": 3.56688210750225, "grad_norm": 0.003952900413423777, "learning_rate": 3.5291539319040727e-07, "loss": 0.0003, "step": 217990 }, { "epoch": 3.567045733453326, "grad_norm": 0.03905739635229111, "learning_rate": 3.526519695175362e-07, "loss": 0.0003, "step": 218000 }, { "epoch": 3.5672093594044014, "grad_norm": 0.01996331289410591, "learning_rate": 3.5238864060014035e-07, "loss": 0.0002, "step": 218010 }, { "epoch": 3.5673729853554774, "grad_norm": 0.013088582083582878, "learning_rate": 3.5212540644358695e-07, "loss": 0.0002, "step": 218020 }, { "epoch": 3.5675366113065534, "grad_norm": 0.0027925516478717327, "learning_rate": 3.51862267053244e-07, "loss": 0.0003, "step": 218030 }, { "epoch": 3.567700237257629, "grad_norm": 0.037014804780483246, "learning_rate": 3.51599222434475e-07, "loss": 0.0007, "step": 218040 }, { "epoch": 3.567863863208705, "grad_norm": 0.04534823074936867, "learning_rate": 3.5133627259264613e-07, "loss": 0.0004, "step": 218050 }, { "epoch": 3.568027489159781, "grad_norm": 0.011865975335240364, "learning_rate": 3.5107341753311596e-07, "loss": 0.0004, "step": 218060 }, { "epoch": 3.5681911151108565, "grad_norm": 0.01711133122444153, "learning_rate": 3.508106572612463e-07, "loss": 0.0003, "step": 218070 }, { "epoch": 3.5683547410619325, "grad_norm": 0.005687564145773649, "learning_rate": 3.505479917823917e-07, "loss": 0.0003, "step": 218080 }, { "epoch": 3.5685183670130085, "grad_norm": 0.03258822113275528, "learning_rate": 3.502854211019108e-07, "loss": 0.0002, "step": 218090 }, { "epoch": 3.568681992964084, "grad_norm": 0.03022460825741291, "learning_rate": 3.500229452251547e-07, "loss": 0.0002, "step": 218100 }, { "epoch": 3.56884561891516, "grad_norm": 0.04688918963074684, "learning_rate": 3.497605641574775e-07, "loss": 0.0003, "step": 218110 }, { "epoch": 3.569009244866236, "grad_norm": 0.0019983493257313967, "learning_rate": 3.494982779042261e-07, "loss": 0.0001, "step": 218120 }, { "epoch": 3.5691728708173116, "grad_norm": 0.014492136426270008, "learning_rate": 3.4923608647075e-07, "loss": 0.0002, "step": 218130 }, { "epoch": 3.5693364967683876, "grad_norm": 0.04498031362891197, "learning_rate": 3.489739898623945e-07, "loss": 0.0003, "step": 218140 }, { "epoch": 3.569500122719463, "grad_norm": 0.004357778467237949, "learning_rate": 3.4871198808450526e-07, "loss": 0.0001, "step": 218150 }, { "epoch": 3.569663748670539, "grad_norm": 0.0007265252061188221, "learning_rate": 3.484500811424213e-07, "loss": 0.0002, "step": 218160 }, { "epoch": 3.569827374621615, "grad_norm": 0.0007579543744213879, "learning_rate": 3.481882690414845e-07, "loss": 0.0003, "step": 218170 }, { "epoch": 3.5699910005726907, "grad_norm": 0.00012901674199383706, "learning_rate": 3.4792655178703226e-07, "loss": 0.0004, "step": 218180 }, { "epoch": 3.5701546265237667, "grad_norm": 0.020962826907634735, "learning_rate": 3.476649293844026e-07, "loss": 0.0005, "step": 218190 }, { "epoch": 3.5703182524748422, "grad_norm": 0.01619771309196949, "learning_rate": 3.4740340183892716e-07, "loss": 0.0003, "step": 218200 }, { "epoch": 3.5704818784259182, "grad_norm": 0.0025057275779545307, "learning_rate": 3.471419691559397e-07, "loss": 0.0002, "step": 218210 }, { "epoch": 3.5706455043769942, "grad_norm": 0.06078442558646202, "learning_rate": 3.468806313407702e-07, "loss": 0.0005, "step": 218220 }, { "epoch": 3.57080913032807, "grad_norm": 0.055863503366708755, "learning_rate": 3.4661938839874787e-07, "loss": 0.0002, "step": 218230 }, { "epoch": 3.570972756279146, "grad_norm": 0.03025759756565094, "learning_rate": 3.4635824033519837e-07, "loss": 0.0004, "step": 218240 }, { "epoch": 3.571136382230222, "grad_norm": 0.06366134434938431, "learning_rate": 3.460971871554464e-07, "loss": 0.0003, "step": 218250 }, { "epoch": 3.5713000081812973, "grad_norm": 0.00636298255994916, "learning_rate": 3.458362288648148e-07, "loss": 0.0001, "step": 218260 }, { "epoch": 3.5714636341323733, "grad_norm": 0.002949213609099388, "learning_rate": 3.45575365468625e-07, "loss": 0.0003, "step": 218270 }, { "epoch": 3.5716272600834493, "grad_norm": 0.022218739613890648, "learning_rate": 3.4531459697219383e-07, "loss": 0.0003, "step": 218280 }, { "epoch": 3.571790886034525, "grad_norm": 0.1196468248963356, "learning_rate": 3.4505392338083975e-07, "loss": 0.0002, "step": 218290 }, { "epoch": 3.571954511985601, "grad_norm": 0.02244383655488491, "learning_rate": 3.4479334469987745e-07, "loss": 0.0001, "step": 218300 }, { "epoch": 3.572118137936677, "grad_norm": 0.00038023621891625226, "learning_rate": 3.4453286093462047e-07, "loss": 0.0002, "step": 218310 }, { "epoch": 3.5722817638877524, "grad_norm": 0.0382935032248497, "learning_rate": 3.442724720903784e-07, "loss": 0.0003, "step": 218320 }, { "epoch": 3.5724453898388284, "grad_norm": 0.0003837582771666348, "learning_rate": 3.4401217817246034e-07, "loss": 0.0001, "step": 218330 }, { "epoch": 3.5726090157899044, "grad_norm": 0.03749135881662369, "learning_rate": 3.4375197918617485e-07, "loss": 0.0005, "step": 218340 }, { "epoch": 3.57277264174098, "grad_norm": 0.06580426543951035, "learning_rate": 3.4349187513682715e-07, "loss": 0.0002, "step": 218350 }, { "epoch": 3.572936267692056, "grad_norm": 0.013798224739730358, "learning_rate": 3.432318660297196e-07, "loss": 0.0002, "step": 218360 }, { "epoch": 3.573099893643132, "grad_norm": 0.003896749345585704, "learning_rate": 3.4297195187015297e-07, "loss": 0.0001, "step": 218370 }, { "epoch": 3.5732635195942075, "grad_norm": 0.03152944892644882, "learning_rate": 3.4271213266342806e-07, "loss": 0.0002, "step": 218380 }, { "epoch": 3.5734271455452835, "grad_norm": 0.0018855424132198095, "learning_rate": 3.4245240841484286e-07, "loss": 0.001, "step": 218390 }, { "epoch": 3.5735907714963595, "grad_norm": 0.24465559422969818, "learning_rate": 3.4219277912969085e-07, "loss": 0.0006, "step": 218400 }, { "epoch": 3.573754397447435, "grad_norm": 0.011448496021330357, "learning_rate": 3.4193324481326673e-07, "loss": 0.0005, "step": 218410 }, { "epoch": 3.573918023398511, "grad_norm": 0.0017110839253291488, "learning_rate": 3.416738054708624e-07, "loss": 0.0002, "step": 218420 }, { "epoch": 3.574081649349587, "grad_norm": 0.015734028071165085, "learning_rate": 3.4141446110776797e-07, "loss": 0.0003, "step": 218430 }, { "epoch": 3.5742452753006626, "grad_norm": 0.08176490664482117, "learning_rate": 3.4115521172927044e-07, "loss": 0.0005, "step": 218440 }, { "epoch": 3.5744089012517386, "grad_norm": 0.0761839896440506, "learning_rate": 3.4089605734065556e-07, "loss": 0.0002, "step": 218450 }, { "epoch": 3.5745725272028146, "grad_norm": 0.001336537185125053, "learning_rate": 3.406369979472074e-07, "loss": 0.0002, "step": 218460 }, { "epoch": 3.57473615315389, "grad_norm": 0.0438665971159935, "learning_rate": 3.403780335542095e-07, "loss": 0.0002, "step": 218470 }, { "epoch": 3.574899779104966, "grad_norm": 0.010632546618580818, "learning_rate": 3.4011916416693933e-07, "loss": 0.0, "step": 218480 }, { "epoch": 3.575063405056042, "grad_norm": 0.02571401745080948, "learning_rate": 3.3986038979067657e-07, "loss": 0.0001, "step": 218490 }, { "epoch": 3.5752270310071177, "grad_norm": 0.04381737485527992, "learning_rate": 3.39601710430697e-07, "loss": 0.0004, "step": 218500 }, { "epoch": 3.5753906569581937, "grad_norm": 0.026878733187913895, "learning_rate": 3.393431260922764e-07, "loss": 0.0006, "step": 218510 }, { "epoch": 3.5755542829092692, "grad_norm": 0.025891771540045738, "learning_rate": 3.3908463678068435e-07, "loss": 0.0007, "step": 218520 }, { "epoch": 3.5757179088603452, "grad_norm": 0.012395746074616909, "learning_rate": 3.388262425011929e-07, "loss": 0.0003, "step": 218530 }, { "epoch": 3.5758815348114212, "grad_norm": 0.003717887680977583, "learning_rate": 3.3856794325906995e-07, "loss": 0.0001, "step": 218540 }, { "epoch": 3.576045160762497, "grad_norm": 0.040786270052194595, "learning_rate": 3.3830973905958297e-07, "loss": 0.0006, "step": 218550 }, { "epoch": 3.576208786713573, "grad_norm": 0.01989026553928852, "learning_rate": 3.38051629907995e-07, "loss": 0.0003, "step": 218560 }, { "epoch": 3.5763724126646483, "grad_norm": 0.012671484611928463, "learning_rate": 3.3779361580957016e-07, "loss": 0.0003, "step": 218570 }, { "epoch": 3.5765360386157243, "grad_norm": 0.058577071875333786, "learning_rate": 3.37535696769567e-07, "loss": 0.0002, "step": 218580 }, { "epoch": 3.5766996645668003, "grad_norm": 0.0026457849889993668, "learning_rate": 3.3727787279324744e-07, "loss": 0.0002, "step": 218590 }, { "epoch": 3.576863290517876, "grad_norm": 0.041400566697120667, "learning_rate": 3.3702014388586557e-07, "loss": 0.0005, "step": 218600 }, { "epoch": 3.577026916468952, "grad_norm": 0.005695435218513012, "learning_rate": 3.3676251005267836e-07, "loss": 0.0002, "step": 218610 }, { "epoch": 3.577190542420028, "grad_norm": 0.0007282505393959582, "learning_rate": 3.3650497129893546e-07, "loss": 0.0002, "step": 218620 }, { "epoch": 3.5773541683711034, "grad_norm": 0.002865969203412533, "learning_rate": 3.362475276298921e-07, "loss": 0.0002, "step": 218630 }, { "epoch": 3.5775177943221794, "grad_norm": 0.07628335803747177, "learning_rate": 3.359901790507941e-07, "loss": 0.0003, "step": 218640 }, { "epoch": 3.5776814202732554, "grad_norm": 0.02768641896545887, "learning_rate": 3.3573292556689117e-07, "loss": 0.0003, "step": 218650 }, { "epoch": 3.577845046224331, "grad_norm": 0.004173723515123129, "learning_rate": 3.3547576718342465e-07, "loss": 0.0002, "step": 218660 }, { "epoch": 3.578008672175407, "grad_norm": 0.009646927006542683, "learning_rate": 3.352187039056426e-07, "loss": 0.0003, "step": 218670 }, { "epoch": 3.578172298126483, "grad_norm": 0.0004984399420209229, "learning_rate": 3.349617357387824e-07, "loss": 0.0002, "step": 218680 }, { "epoch": 3.5783359240775585, "grad_norm": 0.00038513742038048804, "learning_rate": 3.347048626880861e-07, "loss": 0.0002, "step": 218690 }, { "epoch": 3.5784995500286345, "grad_norm": 0.029957497492432594, "learning_rate": 3.3444808475878833e-07, "loss": 0.0004, "step": 218700 }, { "epoch": 3.5786631759797105, "grad_norm": 0.015039661899209023, "learning_rate": 3.341914019561282e-07, "loss": 0.0004, "step": 218710 }, { "epoch": 3.578826801930786, "grad_norm": 0.001836154144257307, "learning_rate": 3.3393481428533604e-07, "loss": 0.0001, "step": 218720 }, { "epoch": 3.578990427881862, "grad_norm": 0.0015664510428905487, "learning_rate": 3.336783217516454e-07, "loss": 0.0004, "step": 218730 }, { "epoch": 3.579154053832938, "grad_norm": 0.03805035725235939, "learning_rate": 3.3342192436028374e-07, "loss": 0.0002, "step": 218740 }, { "epoch": 3.5793176797840136, "grad_norm": 0.015432269312441349, "learning_rate": 3.3316562211648197e-07, "loss": 0.0003, "step": 218750 }, { "epoch": 3.5794813057350896, "grad_norm": 0.0008458077209070325, "learning_rate": 3.3290941502546306e-07, "loss": 0.0002, "step": 218760 }, { "epoch": 3.5796449316861656, "grad_norm": 0.0006859638378955424, "learning_rate": 3.326533030924528e-07, "loss": 0.0003, "step": 218770 }, { "epoch": 3.579808557637241, "grad_norm": 0.025482958182692528, "learning_rate": 3.323972863226715e-07, "loss": 0.0001, "step": 218780 }, { "epoch": 3.579972183588317, "grad_norm": 0.0009319792152382433, "learning_rate": 3.3214136472134007e-07, "loss": 0.0002, "step": 218790 }, { "epoch": 3.580135809539393, "grad_norm": 0.011864238418638706, "learning_rate": 3.3188553829367644e-07, "loss": 0.0004, "step": 218800 }, { "epoch": 3.5802994354904687, "grad_norm": 0.03174973651766777, "learning_rate": 3.31629807044897e-07, "loss": 0.0003, "step": 218810 }, { "epoch": 3.5804630614415447, "grad_norm": 0.029963061213493347, "learning_rate": 3.3137417098021487e-07, "loss": 0.0007, "step": 218820 }, { "epoch": 3.5806266873926207, "grad_norm": 0.008253171108663082, "learning_rate": 3.31118630104843e-07, "loss": 0.0002, "step": 218830 }, { "epoch": 3.5807903133436962, "grad_norm": 0.0014384968671947718, "learning_rate": 3.3086318442399123e-07, "loss": 0.0005, "step": 218840 }, { "epoch": 3.5809539392947722, "grad_norm": 0.02453407272696495, "learning_rate": 3.306078339428692e-07, "loss": 0.0002, "step": 218850 }, { "epoch": 3.5811175652458482, "grad_norm": 0.002204206306487322, "learning_rate": 3.3035257866668113e-07, "loss": 0.0002, "step": 218860 }, { "epoch": 3.581281191196924, "grad_norm": 0.044420983642339706, "learning_rate": 3.3009741860063284e-07, "loss": 0.0004, "step": 218870 }, { "epoch": 3.581444817148, "grad_norm": 0.07052581012248993, "learning_rate": 3.2984235374992677e-07, "loss": 0.0002, "step": 218880 }, { "epoch": 3.581608443099076, "grad_norm": 0.0012104250490665436, "learning_rate": 3.2958738411976386e-07, "loss": 0.0002, "step": 218890 }, { "epoch": 3.5817720690501513, "grad_norm": 0.004618376959115267, "learning_rate": 3.293325097153416e-07, "loss": 0.0001, "step": 218900 }, { "epoch": 3.5819356950012273, "grad_norm": 0.01564343459904194, "learning_rate": 3.290777305418569e-07, "loss": 0.0003, "step": 218910 }, { "epoch": 3.582099320952303, "grad_norm": 0.004194656386971474, "learning_rate": 3.288230466045045e-07, "loss": 0.0003, "step": 218920 }, { "epoch": 3.582262946903379, "grad_norm": 0.00989146064966917, "learning_rate": 3.2856845790847915e-07, "loss": 0.0002, "step": 218930 }, { "epoch": 3.582426572854455, "grad_norm": 0.0003150594711769372, "learning_rate": 3.2831396445896834e-07, "loss": 0.0002, "step": 218940 }, { "epoch": 3.5825901988055304, "grad_norm": 0.08414019644260406, "learning_rate": 3.2805956626116297e-07, "loss": 0.0004, "step": 218950 }, { "epoch": 3.5827538247566064, "grad_norm": 0.00018103142792824656, "learning_rate": 3.2780526332024997e-07, "loss": 0.0001, "step": 218960 }, { "epoch": 3.582917450707682, "grad_norm": 0.01714230515062809, "learning_rate": 3.2755105564141465e-07, "loss": 0.0002, "step": 218970 }, { "epoch": 3.583081076658758, "grad_norm": 0.00011143511073896661, "learning_rate": 3.2729694322983895e-07, "loss": 0.0003, "step": 218980 }, { "epoch": 3.583244702609834, "grad_norm": 0.004994621966034174, "learning_rate": 3.2704292609070433e-07, "loss": 0.0004, "step": 218990 }, { "epoch": 3.5834083285609095, "grad_norm": 0.0003581656201276928, "learning_rate": 3.2678900422919046e-07, "loss": 0.0002, "step": 219000 }, { "epoch": 3.5835719545119855, "grad_norm": 0.0022422655019909143, "learning_rate": 3.265351776504749e-07, "loss": 0.0001, "step": 219010 }, { "epoch": 3.5837355804630615, "grad_norm": 0.06568591296672821, "learning_rate": 3.262814463597319e-07, "loss": 0.0002, "step": 219020 }, { "epoch": 3.583899206414137, "grad_norm": 0.04076892137527466, "learning_rate": 3.260278103621356e-07, "loss": 0.0003, "step": 219030 }, { "epoch": 3.584062832365213, "grad_norm": 0.02906392142176628, "learning_rate": 3.2577426966285687e-07, "loss": 0.0002, "step": 219040 }, { "epoch": 3.584226458316289, "grad_norm": 0.02383168786764145, "learning_rate": 3.25520824267066e-07, "loss": 0.0002, "step": 219050 }, { "epoch": 3.5843900842673646, "grad_norm": 0.0002067859168164432, "learning_rate": 3.2526747417992943e-07, "loss": 0.0001, "step": 219060 }, { "epoch": 3.5845537102184406, "grad_norm": 0.02690574899315834, "learning_rate": 3.250142194066136e-07, "loss": 0.0002, "step": 219070 }, { "epoch": 3.5847173361695166, "grad_norm": 0.039102938026189804, "learning_rate": 3.247610599522816e-07, "loss": 0.0002, "step": 219080 }, { "epoch": 3.584880962120592, "grad_norm": 0.021852821111679077, "learning_rate": 3.245079958220959e-07, "loss": 0.0009, "step": 219090 }, { "epoch": 3.585044588071668, "grad_norm": 0.01482396386563778, "learning_rate": 3.242550270212158e-07, "loss": 0.0005, "step": 219100 }, { "epoch": 3.585208214022744, "grad_norm": 0.006681601982563734, "learning_rate": 3.240021535547988e-07, "loss": 0.0001, "step": 219110 }, { "epoch": 3.5853718399738197, "grad_norm": 0.0019441407639533281, "learning_rate": 3.2374937542800075e-07, "loss": 0.0007, "step": 219120 }, { "epoch": 3.5855354659248957, "grad_norm": 0.07990431785583496, "learning_rate": 3.2349669264597705e-07, "loss": 0.0004, "step": 219130 }, { "epoch": 3.5856990918759717, "grad_norm": 0.03896738216280937, "learning_rate": 3.232441052138779e-07, "loss": 0.0003, "step": 219140 }, { "epoch": 3.5858627178270472, "grad_norm": 0.02319234609603882, "learning_rate": 3.229916131368538e-07, "loss": 0.0001, "step": 219150 }, { "epoch": 3.5860263437781232, "grad_norm": 0.12194161862134933, "learning_rate": 3.227392164200532e-07, "loss": 0.0002, "step": 219160 }, { "epoch": 3.5861899697291992, "grad_norm": 0.011909473687410355, "learning_rate": 3.2248691506862275e-07, "loss": 0.0001, "step": 219170 }, { "epoch": 3.586353595680275, "grad_norm": 0.015827620401978493, "learning_rate": 3.2223470908770537e-07, "loss": 0.0004, "step": 219180 }, { "epoch": 3.586517221631351, "grad_norm": 0.10464300960302353, "learning_rate": 3.2198259848244374e-07, "loss": 0.0004, "step": 219190 }, { "epoch": 3.586680847582427, "grad_norm": 0.0387900173664093, "learning_rate": 3.2173058325797866e-07, "loss": 0.0004, "step": 219200 }, { "epoch": 3.5868444735335023, "grad_norm": 0.004243787843734026, "learning_rate": 3.214786634194489e-07, "loss": 0.0001, "step": 219210 }, { "epoch": 3.5870080994845783, "grad_norm": 0.017501741647720337, "learning_rate": 3.2122683897198967e-07, "loss": 0.0003, "step": 219220 }, { "epoch": 3.5871717254356543, "grad_norm": 0.022522522136569023, "learning_rate": 3.2097510992073646e-07, "loss": 0.0001, "step": 219230 }, { "epoch": 3.58733535138673, "grad_norm": 0.02874683402478695, "learning_rate": 3.2072347627082003e-07, "loss": 0.0008, "step": 219240 }, { "epoch": 3.587498977337806, "grad_norm": 0.00019299220002721995, "learning_rate": 3.204719380273741e-07, "loss": 0.0001, "step": 219250 }, { "epoch": 3.587662603288882, "grad_norm": 0.012771319597959518, "learning_rate": 3.2022049519552457e-07, "loss": 0.0001, "step": 219260 }, { "epoch": 3.5878262292399574, "grad_norm": 0.0007076021865941584, "learning_rate": 3.1996914778039957e-07, "loss": 0.0001, "step": 219270 }, { "epoch": 3.5879898551910334, "grad_norm": 0.013402387499809265, "learning_rate": 3.197178957871222e-07, "loss": 0.0002, "step": 219280 }, { "epoch": 3.5881534811421094, "grad_norm": 0.2465803176164627, "learning_rate": 3.194667392208184e-07, "loss": 0.0008, "step": 219290 }, { "epoch": 3.588317107093185, "grad_norm": 0.005889517720788717, "learning_rate": 3.192156780866057e-07, "loss": 0.0003, "step": 219300 }, { "epoch": 3.588480733044261, "grad_norm": 0.0005051901680417359, "learning_rate": 3.189647123896056e-07, "loss": 0.0002, "step": 219310 }, { "epoch": 3.5886443589953365, "grad_norm": 0.0015152633422985673, "learning_rate": 3.187138421349323e-07, "loss": 0.0005, "step": 219320 }, { "epoch": 3.5888079849464125, "grad_norm": 0.026420319452881813, "learning_rate": 3.1846306732770446e-07, "loss": 0.0002, "step": 219330 }, { "epoch": 3.5889716108974885, "grad_norm": 0.03744376823306084, "learning_rate": 3.18212387973032e-07, "loss": 0.0001, "step": 219340 }, { "epoch": 3.589135236848564, "grad_norm": 0.000794712221249938, "learning_rate": 3.1796180407602793e-07, "loss": 0.0003, "step": 219350 }, { "epoch": 3.58929886279964, "grad_norm": 8.741497731534764e-05, "learning_rate": 3.1771131564179936e-07, "loss": 0.0003, "step": 219360 }, { "epoch": 3.5894624887507156, "grad_norm": 0.04055776447057724, "learning_rate": 3.1746092267545604e-07, "loss": 0.0003, "step": 219370 }, { "epoch": 3.5896261147017916, "grad_norm": 0.00773809477686882, "learning_rate": 3.1721062518210166e-07, "loss": 0.0006, "step": 219380 }, { "epoch": 3.5897897406528676, "grad_norm": 0.06666206568479538, "learning_rate": 3.1696042316684103e-07, "loss": 0.0002, "step": 219390 }, { "epoch": 3.589953366603943, "grad_norm": 0.010917583480477333, "learning_rate": 3.167103166347735e-07, "loss": 0.0001, "step": 219400 }, { "epoch": 3.590116992555019, "grad_norm": 0.0008212049142457545, "learning_rate": 3.1646030559099983e-07, "loss": 0.0001, "step": 219410 }, { "epoch": 3.590280618506095, "grad_norm": 0.020136840641498566, "learning_rate": 3.162103900406177e-07, "loss": 0.0002, "step": 219420 }, { "epoch": 3.5904442444571707, "grad_norm": 0.011826482601463795, "learning_rate": 3.159605699887225e-07, "loss": 0.0001, "step": 219430 }, { "epoch": 3.5906078704082467, "grad_norm": 0.030904436483979225, "learning_rate": 3.1571084544040674e-07, "loss": 0.0001, "step": 219440 }, { "epoch": 3.5907714963593227, "grad_norm": 0.02868117392063141, "learning_rate": 3.1546121640076364e-07, "loss": 0.0001, "step": 219450 }, { "epoch": 3.5909351223103982, "grad_norm": 0.0006347285816445947, "learning_rate": 3.152116828748819e-07, "loss": 0.0004, "step": 219460 }, { "epoch": 3.5910987482614742, "grad_norm": 0.00020494103955570608, "learning_rate": 3.149622448678502e-07, "loss": 0.0001, "step": 219470 }, { "epoch": 3.5912623742125502, "grad_norm": 0.003836296731606126, "learning_rate": 3.1471290238475337e-07, "loss": 0.0003, "step": 219480 }, { "epoch": 3.591426000163626, "grad_norm": 0.01615661196410656, "learning_rate": 3.144636554306757e-07, "loss": 0.0002, "step": 219490 }, { "epoch": 3.591589626114702, "grad_norm": 0.003474622266367078, "learning_rate": 3.1421450401069976e-07, "loss": 0.0004, "step": 219500 }, { "epoch": 3.591753252065778, "grad_norm": 0.025241080671548843, "learning_rate": 3.1396544812990484e-07, "loss": 0.0001, "step": 219510 }, { "epoch": 3.5919168780168533, "grad_norm": 0.04194605350494385, "learning_rate": 3.1371648779336906e-07, "loss": 0.0002, "step": 219520 }, { "epoch": 3.5920805039679293, "grad_norm": 0.009254083037376404, "learning_rate": 3.1346762300616785e-07, "loss": 0.0003, "step": 219530 }, { "epoch": 3.5922441299190053, "grad_norm": 0.0002658599696587771, "learning_rate": 3.1321885377337657e-07, "loss": 0.0005, "step": 219540 }, { "epoch": 3.592407755870081, "grad_norm": 0.014936676248908043, "learning_rate": 3.129701801000673e-07, "loss": 0.0001, "step": 219550 }, { "epoch": 3.592571381821157, "grad_norm": 0.0014347353717312217, "learning_rate": 3.127216019913093e-07, "loss": 0.0002, "step": 219560 }, { "epoch": 3.592735007772233, "grad_norm": 0.039397165179252625, "learning_rate": 3.1247311945217184e-07, "loss": 0.0003, "step": 219570 }, { "epoch": 3.5928986337233084, "grad_norm": 0.007777830585837364, "learning_rate": 3.1222473248772034e-07, "loss": 0.0002, "step": 219580 }, { "epoch": 3.5930622596743844, "grad_norm": 0.11456139385700226, "learning_rate": 3.119764411030207e-07, "loss": 0.0006, "step": 219590 }, { "epoch": 3.5932258856254604, "grad_norm": 0.08698941022157669, "learning_rate": 3.117282453031334e-07, "loss": 0.0002, "step": 219600 }, { "epoch": 3.593389511576536, "grad_norm": 0.026688728481531143, "learning_rate": 3.114801450931204e-07, "loss": 0.0002, "step": 219610 }, { "epoch": 3.593553137527612, "grad_norm": 0.0029746515210717916, "learning_rate": 3.112321404780394e-07, "loss": 0.0003, "step": 219620 }, { "epoch": 3.593716763478688, "grad_norm": 0.00017297329031862319, "learning_rate": 3.1098423146294855e-07, "loss": 0.0001, "step": 219630 }, { "epoch": 3.5938803894297635, "grad_norm": 0.028195949271321297, "learning_rate": 3.1073641805289987e-07, "loss": 0.0003, "step": 219640 }, { "epoch": 3.5940440153808395, "grad_norm": 0.004955345764756203, "learning_rate": 3.104887002529483e-07, "loss": 0.0002, "step": 219650 }, { "epoch": 3.5942076413319155, "grad_norm": 0.01878032088279724, "learning_rate": 3.1024107806814364e-07, "loss": 0.0004, "step": 219660 }, { "epoch": 3.594371267282991, "grad_norm": 0.01239298190921545, "learning_rate": 3.0999355150353573e-07, "loss": 0.0003, "step": 219670 }, { "epoch": 3.594534893234067, "grad_norm": 0.003112176200374961, "learning_rate": 3.097461205641694e-07, "loss": 0.0004, "step": 219680 }, { "epoch": 3.5946985191851426, "grad_norm": 0.137481227517128, "learning_rate": 3.094987852550912e-07, "loss": 0.0005, "step": 219690 }, { "epoch": 3.5948621451362186, "grad_norm": 0.028267664834856987, "learning_rate": 3.092515455813433e-07, "loss": 0.0002, "step": 219700 }, { "epoch": 3.5950257710872946, "grad_norm": 0.02968171238899231, "learning_rate": 3.0900440154796817e-07, "loss": 0.0002, "step": 219710 }, { "epoch": 3.59518939703837, "grad_norm": 0.04415919631719589, "learning_rate": 3.08757353160003e-07, "loss": 0.0004, "step": 219720 }, { "epoch": 3.595353022989446, "grad_norm": 0.05091855302453041, "learning_rate": 3.085104004224854e-07, "loss": 0.0004, "step": 219730 }, { "epoch": 3.5955166489405217, "grad_norm": 0.0003430507204029709, "learning_rate": 3.0826354334045073e-07, "loss": 0.0002, "step": 219740 }, { "epoch": 3.5956802748915977, "grad_norm": 0.0057800509966909885, "learning_rate": 3.080167819189328e-07, "loss": 0.0003, "step": 219750 }, { "epoch": 3.5958439008426737, "grad_norm": 0.019404688850045204, "learning_rate": 3.0777011616296205e-07, "loss": 0.0002, "step": 219760 }, { "epoch": 3.5960075267937492, "grad_norm": 0.023857641965150833, "learning_rate": 3.075235460775677e-07, "loss": 0.0003, "step": 219770 }, { "epoch": 3.5961711527448252, "grad_norm": 0.0029617920517921448, "learning_rate": 3.0727707166777753e-07, "loss": 0.0001, "step": 219780 }, { "epoch": 3.5963347786959012, "grad_norm": 0.0008356551988981664, "learning_rate": 3.07030692938618e-07, "loss": 0.0003, "step": 219790 }, { "epoch": 3.596498404646977, "grad_norm": 0.009548194706439972, "learning_rate": 3.0678440989511006e-07, "loss": 0.0003, "step": 219800 }, { "epoch": 3.596662030598053, "grad_norm": 0.09966674447059631, "learning_rate": 3.0653822254227695e-07, "loss": 0.0002, "step": 219810 }, { "epoch": 3.596825656549129, "grad_norm": 0.0028140402864664793, "learning_rate": 3.0629213088513743e-07, "loss": 0.0002, "step": 219820 }, { "epoch": 3.5969892825002043, "grad_norm": 0.016142867505550385, "learning_rate": 3.060461349287103e-07, "loss": 0.0002, "step": 219830 }, { "epoch": 3.5971529084512803, "grad_norm": 0.03874214366078377, "learning_rate": 3.0580023467801033e-07, "loss": 0.0002, "step": 219840 }, { "epoch": 3.5973165344023563, "grad_norm": 0.0269952230155468, "learning_rate": 3.055544301380503e-07, "loss": 0.0004, "step": 219850 }, { "epoch": 3.597480160353432, "grad_norm": 0.05365068092942238, "learning_rate": 3.053087213138439e-07, "loss": 0.0001, "step": 219860 }, { "epoch": 3.597643786304508, "grad_norm": 0.013092191889882088, "learning_rate": 3.0506310821039997e-07, "loss": 0.0002, "step": 219870 }, { "epoch": 3.597807412255584, "grad_norm": 0.0962938517332077, "learning_rate": 3.0481759083272613e-07, "loss": 0.0002, "step": 219880 }, { "epoch": 3.5979710382066594, "grad_norm": 0.006921404041349888, "learning_rate": 3.045721691858289e-07, "loss": 0.0001, "step": 219890 }, { "epoch": 3.5981346641577354, "grad_norm": 0.06469372659921646, "learning_rate": 3.0432684327471095e-07, "loss": 0.0003, "step": 219900 }, { "epoch": 3.5982982901088114, "grad_norm": 0.05584818869829178, "learning_rate": 3.040816131043761e-07, "loss": 0.0002, "step": 219910 }, { "epoch": 3.598461916059887, "grad_norm": 0.01110587827861309, "learning_rate": 3.038364786798226e-07, "loss": 0.0002, "step": 219920 }, { "epoch": 3.598625542010963, "grad_norm": 0.05518568307161331, "learning_rate": 3.035914400060502e-07, "loss": 0.0002, "step": 219930 }, { "epoch": 3.598789167962039, "grad_norm": 0.04691607877612114, "learning_rate": 3.0334649708805286e-07, "loss": 0.0002, "step": 219940 }, { "epoch": 3.5989527939131145, "grad_norm": 0.0710555762052536, "learning_rate": 3.031016499308276e-07, "loss": 0.0003, "step": 219950 }, { "epoch": 3.5991164198641905, "grad_norm": 0.04590560495853424, "learning_rate": 3.028568985393643e-07, "loss": 0.0002, "step": 219960 }, { "epoch": 3.5992800458152665, "grad_norm": 0.003984127193689346, "learning_rate": 3.026122429186545e-07, "loss": 0.0002, "step": 219970 }, { "epoch": 3.599443671766342, "grad_norm": 0.0060983323492109776, "learning_rate": 3.023676830736849e-07, "loss": 0.0002, "step": 219980 }, { "epoch": 3.599607297717418, "grad_norm": 0.023152252659201622, "learning_rate": 3.021232190094447e-07, "loss": 0.0002, "step": 219990 }, { "epoch": 3.599770923668494, "grad_norm": 0.006770946551114321, "learning_rate": 3.018788507309156e-07, "loss": 0.0002, "step": 220000 }, { "epoch": 3.5999345496195696, "grad_norm": 0.034987714141607285, "learning_rate": 3.0163457824308185e-07, "loss": 0.0003, "step": 220010 }, { "epoch": 3.6000981755706456, "grad_norm": 0.00970907136797905, "learning_rate": 3.0139040155092225e-07, "loss": 0.0002, "step": 220020 }, { "epoch": 3.6002618015217216, "grad_norm": 0.00045390689047053456, "learning_rate": 3.011463206594173e-07, "loss": 0.0002, "step": 220030 }, { "epoch": 3.600425427472797, "grad_norm": 0.07443641126155853, "learning_rate": 3.009023355735419e-07, "loss": 0.0002, "step": 220040 }, { "epoch": 3.600589053423873, "grad_norm": 0.044445257633924484, "learning_rate": 3.0065844629827256e-07, "loss": 0.0003, "step": 220050 }, { "epoch": 3.600752679374949, "grad_norm": 0.011594045907258987, "learning_rate": 3.0041465283857984e-07, "loss": 0.0002, "step": 220060 }, { "epoch": 3.6009163053260247, "grad_norm": 0.05305451899766922, "learning_rate": 3.001709551994353e-07, "loss": 0.0002, "step": 220070 }, { "epoch": 3.6010799312771007, "grad_norm": 0.025444012135267258, "learning_rate": 2.9992735338580825e-07, "loss": 0.0002, "step": 220080 }, { "epoch": 3.6012435572281762, "grad_norm": 0.010871756821870804, "learning_rate": 2.996838474026659e-07, "loss": 0.0002, "step": 220090 }, { "epoch": 3.6014071831792522, "grad_norm": 0.00020900683011859655, "learning_rate": 2.9944043725497194e-07, "loss": 0.0002, "step": 220100 }, { "epoch": 3.6015708091303282, "grad_norm": 0.03183828666806221, "learning_rate": 2.9919712294768923e-07, "loss": 0.0004, "step": 220110 }, { "epoch": 3.601734435081404, "grad_norm": 0.044023022055625916, "learning_rate": 2.989539044857798e-07, "loss": 0.0002, "step": 220120 }, { "epoch": 3.60189806103248, "grad_norm": 0.01479207444936037, "learning_rate": 2.9871078187420246e-07, "loss": 0.0003, "step": 220130 }, { "epoch": 3.6020616869835553, "grad_norm": 0.0007854839786887169, "learning_rate": 2.9846775511791324e-07, "loss": 0.0002, "step": 220140 }, { "epoch": 3.6022253129346313, "grad_norm": 0.03433051332831383, "learning_rate": 2.9822482422186827e-07, "loss": 0.0005, "step": 220150 }, { "epoch": 3.6023889388857073, "grad_norm": 0.031126240268349648, "learning_rate": 2.9798198919102016e-07, "loss": 0.0002, "step": 220160 }, { "epoch": 3.602552564836783, "grad_norm": 0.0005129494820721447, "learning_rate": 2.9773925003032166e-07, "loss": 0.0003, "step": 220170 }, { "epoch": 3.602716190787859, "grad_norm": 0.018832409754395485, "learning_rate": 2.9749660674471934e-07, "loss": 0.0003, "step": 220180 }, { "epoch": 3.602879816738935, "grad_norm": 0.0006754686473868787, "learning_rate": 2.9725405933916205e-07, "loss": 0.0002, "step": 220190 }, { "epoch": 3.6030434426900104, "grad_norm": 0.0011801030486822128, "learning_rate": 2.970116078185953e-07, "loss": 0.0005, "step": 220200 }, { "epoch": 3.6032070686410864, "grad_norm": 0.048521071672439575, "learning_rate": 2.9676925218796224e-07, "loss": 0.0002, "step": 220210 }, { "epoch": 3.6033706945921624, "grad_norm": 0.1935586780309677, "learning_rate": 2.9652699245220407e-07, "loss": 0.0008, "step": 220220 }, { "epoch": 3.603534320543238, "grad_norm": 0.009772706776857376, "learning_rate": 2.962848286162601e-07, "loss": 0.0003, "step": 220230 }, { "epoch": 3.603697946494314, "grad_norm": 0.06215304508805275, "learning_rate": 2.9604276068506797e-07, "loss": 0.0002, "step": 220240 }, { "epoch": 3.60386157244539, "grad_norm": 0.015503753907978535, "learning_rate": 2.9580078866356385e-07, "loss": 0.0004, "step": 220250 }, { "epoch": 3.6040251983964655, "grad_norm": 0.00015964251360855997, "learning_rate": 2.95558912556681e-07, "loss": 0.0002, "step": 220260 }, { "epoch": 3.6041888243475415, "grad_norm": 0.015508691780269146, "learning_rate": 2.9531713236935046e-07, "loss": 0.0002, "step": 220270 }, { "epoch": 3.6043524502986175, "grad_norm": 0.04261058196425438, "learning_rate": 2.9507544810650214e-07, "loss": 0.0005, "step": 220280 }, { "epoch": 3.604516076249693, "grad_norm": 0.057048771530389786, "learning_rate": 2.9483385977306487e-07, "loss": 0.0003, "step": 220290 }, { "epoch": 3.604679702200769, "grad_norm": 0.3726601004600525, "learning_rate": 2.9459236737396314e-07, "loss": 0.0011, "step": 220300 }, { "epoch": 3.604843328151845, "grad_norm": 0.08737389743328094, "learning_rate": 2.9435097091412125e-07, "loss": 0.0003, "step": 220310 }, { "epoch": 3.6050069541029206, "grad_norm": 0.0007111378945410252, "learning_rate": 2.941096703984608e-07, "loss": 0.0002, "step": 220320 }, { "epoch": 3.6051705800539966, "grad_norm": 0.007350974250584841, "learning_rate": 2.938684658319024e-07, "loss": 0.0002, "step": 220330 }, { "epoch": 3.6053342060050726, "grad_norm": 0.06378155946731567, "learning_rate": 2.9362735721936376e-07, "loss": 0.0004, "step": 220340 }, { "epoch": 3.605497831956148, "grad_norm": 0.047018762677907944, "learning_rate": 2.9338634456575976e-07, "loss": 0.0001, "step": 220350 }, { "epoch": 3.605661457907224, "grad_norm": 0.0007377950823865831, "learning_rate": 2.9314542787600596e-07, "loss": 0.0003, "step": 220360 }, { "epoch": 3.6058250838583, "grad_norm": 0.09469293802976608, "learning_rate": 2.92904607155014e-07, "loss": 0.0005, "step": 220370 }, { "epoch": 3.6059887098093757, "grad_norm": 0.015917519107460976, "learning_rate": 2.926638824076938e-07, "loss": 0.0001, "step": 220380 }, { "epoch": 3.6061523357604517, "grad_norm": 0.014120233245193958, "learning_rate": 2.924232536389532e-07, "loss": 0.0002, "step": 220390 }, { "epoch": 3.6063159617115277, "grad_norm": 0.0103109460324049, "learning_rate": 2.921827208536987e-07, "loss": 0.0001, "step": 220400 }, { "epoch": 3.6064795876626032, "grad_norm": 0.0017836537444964051, "learning_rate": 2.919422840568359e-07, "loss": 0.0003, "step": 220410 }, { "epoch": 3.6066432136136792, "grad_norm": 0.023532135412096977, "learning_rate": 2.917019432532647e-07, "loss": 0.0003, "step": 220420 }, { "epoch": 3.6068068395647552, "grad_norm": 0.013092206791043282, "learning_rate": 2.9146169844788684e-07, "loss": 0.0006, "step": 220430 }, { "epoch": 3.606970465515831, "grad_norm": 0.13157568871974945, "learning_rate": 2.9122154964560003e-07, "loss": 0.0004, "step": 220440 }, { "epoch": 3.6071340914669068, "grad_norm": 0.07349144667387009, "learning_rate": 2.909814968513025e-07, "loss": 0.0003, "step": 220450 }, { "epoch": 3.6072977174179823, "grad_norm": 0.03520024195313454, "learning_rate": 2.907415400698865e-07, "loss": 0.0003, "step": 220460 }, { "epoch": 3.6074613433690583, "grad_norm": 0.04580257087945938, "learning_rate": 2.9050167930624483e-07, "loss": 0.0004, "step": 220470 }, { "epoch": 3.6076249693201343, "grad_norm": 0.031146328896284103, "learning_rate": 2.902619145652691e-07, "loss": 0.0002, "step": 220480 }, { "epoch": 3.60778859527121, "grad_norm": 0.060079533606767654, "learning_rate": 2.900222458518481e-07, "loss": 0.0002, "step": 220490 }, { "epoch": 3.607952221222286, "grad_norm": 0.042006682604551315, "learning_rate": 2.897826731708675e-07, "loss": 0.0002, "step": 220500 }, { "epoch": 3.6081158471733614, "grad_norm": 0.020397428423166275, "learning_rate": 2.895431965272127e-07, "loss": 0.0005, "step": 220510 }, { "epoch": 3.6082794731244374, "grad_norm": 0.026404915377497673, "learning_rate": 2.893038159257644e-07, "loss": 0.0002, "step": 220520 }, { "epoch": 3.6084430990755134, "grad_norm": 0.0016416978323832154, "learning_rate": 2.890645313714069e-07, "loss": 0.0003, "step": 220530 }, { "epoch": 3.608606725026589, "grad_norm": 0.023890964686870575, "learning_rate": 2.8882534286901575e-07, "loss": 0.0004, "step": 220540 }, { "epoch": 3.608770350977665, "grad_norm": 0.0010817834408953786, "learning_rate": 2.8858625042347044e-07, "loss": 0.0002, "step": 220550 }, { "epoch": 3.608933976928741, "grad_norm": 0.10603424906730652, "learning_rate": 2.883472540396426e-07, "loss": 0.0006, "step": 220560 }, { "epoch": 3.6090976028798165, "grad_norm": 0.045296840369701385, "learning_rate": 2.8810835372240884e-07, "loss": 0.0001, "step": 220570 }, { "epoch": 3.6092612288308925, "grad_norm": 0.04928361251950264, "learning_rate": 2.8786954947663756e-07, "loss": 0.0003, "step": 220580 }, { "epoch": 3.6094248547819685, "grad_norm": 0.03738677501678467, "learning_rate": 2.8763084130719987e-07, "loss": 0.0002, "step": 220590 }, { "epoch": 3.609588480733044, "grad_norm": 0.021297980099916458, "learning_rate": 2.8739222921895957e-07, "loss": 0.0001, "step": 220600 }, { "epoch": 3.60975210668412, "grad_norm": 0.1019585132598877, "learning_rate": 2.8715371321678564e-07, "loss": 0.0003, "step": 220610 }, { "epoch": 3.609915732635196, "grad_norm": 0.00035291205858811736, "learning_rate": 2.86915293305538e-07, "loss": 0.0002, "step": 220620 }, { "epoch": 3.6100793585862716, "grad_norm": 0.04481744021177292, "learning_rate": 2.866769694900806e-07, "loss": 0.0001, "step": 220630 }, { "epoch": 3.6102429845373476, "grad_norm": 0.029564224183559418, "learning_rate": 2.864387417752695e-07, "loss": 0.0003, "step": 220640 }, { "epoch": 3.6104066104884236, "grad_norm": 0.06685692071914673, "learning_rate": 2.8620061016596536e-07, "loss": 0.0004, "step": 220650 }, { "epoch": 3.610570236439499, "grad_norm": 0.09295706450939178, "learning_rate": 2.8596257466702084e-07, "loss": 0.0003, "step": 220660 }, { "epoch": 3.610733862390575, "grad_norm": 0.036567866802215576, "learning_rate": 2.857246352832915e-07, "loss": 0.0002, "step": 220670 }, { "epoch": 3.610897488341651, "grad_norm": 0.014942199923098087, "learning_rate": 2.854867920196253e-07, "loss": 0.0002, "step": 220680 }, { "epoch": 3.6110611142927267, "grad_norm": 0.023124556988477707, "learning_rate": 2.852490448808759e-07, "loss": 0.0001, "step": 220690 }, { "epoch": 3.6112247402438027, "grad_norm": 0.03536678105592728, "learning_rate": 2.850113938718879e-07, "loss": 0.0005, "step": 220700 }, { "epoch": 3.6113883661948787, "grad_norm": 0.033125218003988266, "learning_rate": 2.84773838997508e-07, "loss": 0.0002, "step": 220710 }, { "epoch": 3.6115519921459542, "grad_norm": 0.025697264820337296, "learning_rate": 2.845363802625789e-07, "loss": 0.0004, "step": 220720 }, { "epoch": 3.6117156180970302, "grad_norm": 0.013629911467432976, "learning_rate": 2.842990176719429e-07, "loss": 0.0001, "step": 220730 }, { "epoch": 3.6118792440481062, "grad_norm": 0.015881016850471497, "learning_rate": 2.8406175123044e-07, "loss": 0.0003, "step": 220740 }, { "epoch": 3.612042869999182, "grad_norm": 0.0005643429467454553, "learning_rate": 2.8382458094290633e-07, "loss": 0.0004, "step": 220750 }, { "epoch": 3.612206495950258, "grad_norm": 0.059916894882917404, "learning_rate": 2.8358750681417857e-07, "loss": 0.0001, "step": 220760 }, { "epoch": 3.6123701219013338, "grad_norm": 0.11478573828935623, "learning_rate": 2.833505288490901e-07, "loss": 0.0007, "step": 220770 }, { "epoch": 3.6125337478524093, "grad_norm": 0.08076436817646027, "learning_rate": 2.831136470524742e-07, "loss": 0.0005, "step": 220780 }, { "epoch": 3.6126973738034853, "grad_norm": 0.0035243690945208073, "learning_rate": 2.828768614291583e-07, "loss": 0.0002, "step": 220790 }, { "epoch": 3.6128609997545613, "grad_norm": 0.03178674355149269, "learning_rate": 2.8264017198397163e-07, "loss": 0.0001, "step": 220800 }, { "epoch": 3.613024625705637, "grad_norm": 0.042269233614206314, "learning_rate": 2.8240357872173994e-07, "loss": 0.0009, "step": 220810 }, { "epoch": 3.613188251656713, "grad_norm": 0.04777984321117401, "learning_rate": 2.821670816472877e-07, "loss": 0.0003, "step": 220820 }, { "epoch": 3.613351877607789, "grad_norm": 0.000402790930820629, "learning_rate": 2.819306807654354e-07, "loss": 0.0002, "step": 220830 }, { "epoch": 3.6135155035588644, "grad_norm": 0.019628949463367462, "learning_rate": 2.8169437608100427e-07, "loss": 0.0001, "step": 220840 }, { "epoch": 3.6136791295099404, "grad_norm": 0.07953719794750214, "learning_rate": 2.8145816759881206e-07, "loss": 0.0002, "step": 220850 }, { "epoch": 3.613842755461016, "grad_norm": 0.07297419756650925, "learning_rate": 2.812220553236755e-07, "loss": 0.0005, "step": 220860 }, { "epoch": 3.614006381412092, "grad_norm": 0.026305418461561203, "learning_rate": 2.809860392604069e-07, "loss": 0.0002, "step": 220870 }, { "epoch": 3.614170007363168, "grad_norm": 0.011357258073985577, "learning_rate": 2.8075011941382015e-07, "loss": 0.0002, "step": 220880 }, { "epoch": 3.6143336333142435, "grad_norm": 0.031193621456623077, "learning_rate": 2.805142957887247e-07, "loss": 0.001, "step": 220890 }, { "epoch": 3.6144972592653195, "grad_norm": 0.048659201711416245, "learning_rate": 2.802785683899295e-07, "loss": 0.0004, "step": 220900 }, { "epoch": 3.614660885216395, "grad_norm": 0.017737507820129395, "learning_rate": 2.800429372222396e-07, "loss": 0.0002, "step": 220910 }, { "epoch": 3.614824511167471, "grad_norm": 0.014002195559442043, "learning_rate": 2.7980740229046064e-07, "loss": 0.0005, "step": 220920 }, { "epoch": 3.614988137118547, "grad_norm": 0.023537030443549156, "learning_rate": 2.7957196359939374e-07, "loss": 0.0004, "step": 220930 }, { "epoch": 3.6151517630696226, "grad_norm": 0.03452753275632858, "learning_rate": 2.7933662115384065e-07, "loss": 0.0002, "step": 220940 }, { "epoch": 3.6153153890206986, "grad_norm": 0.06516700983047485, "learning_rate": 2.791013749585986e-07, "loss": 0.0002, "step": 220950 }, { "epoch": 3.6154790149717746, "grad_norm": 0.02086370810866356, "learning_rate": 2.788662250184648e-07, "loss": 0.0002, "step": 220960 }, { "epoch": 3.61564264092285, "grad_norm": 0.02740313485264778, "learning_rate": 2.786311713382328e-07, "loss": 0.0003, "step": 220970 }, { "epoch": 3.615806266873926, "grad_norm": 0.014599812217056751, "learning_rate": 2.78396213922697e-07, "loss": 0.0001, "step": 220980 }, { "epoch": 3.615969892825002, "grad_norm": 0.04094075784087181, "learning_rate": 2.781613527766458e-07, "loss": 0.0003, "step": 220990 }, { "epoch": 3.6161335187760777, "grad_norm": 0.0014758682809770107, "learning_rate": 2.779265879048698e-07, "loss": 0.0005, "step": 221000 }, { "epoch": 3.6162971447271537, "grad_norm": 0.011066725477576256, "learning_rate": 2.776919193121536e-07, "loss": 0.0001, "step": 221010 }, { "epoch": 3.6164607706782297, "grad_norm": 0.04633498936891556, "learning_rate": 2.7745734700328374e-07, "loss": 0.0002, "step": 221020 }, { "epoch": 3.6166243966293052, "grad_norm": 0.03758566081523895, "learning_rate": 2.772228709830421e-07, "loss": 0.0005, "step": 221030 }, { "epoch": 3.6167880225803812, "grad_norm": 0.010043524205684662, "learning_rate": 2.7698849125620986e-07, "loss": 0.0002, "step": 221040 }, { "epoch": 3.6169516485314572, "grad_norm": 0.021070824936032295, "learning_rate": 2.767542078275642e-07, "loss": 0.0003, "step": 221050 }, { "epoch": 3.617115274482533, "grad_norm": 0.009903559461236, "learning_rate": 2.7652002070188476e-07, "loss": 0.0001, "step": 221060 }, { "epoch": 3.617278900433609, "grad_norm": 0.048661090433597565, "learning_rate": 2.762859298839438e-07, "loss": 0.0003, "step": 221070 }, { "epoch": 3.6174425263846848, "grad_norm": 0.01802935265004635, "learning_rate": 2.760519353785163e-07, "loss": 0.0001, "step": 221080 }, { "epoch": 3.6176061523357603, "grad_norm": 0.026346033439040184, "learning_rate": 2.7581803719037125e-07, "loss": 0.0005, "step": 221090 }, { "epoch": 3.6177697782868363, "grad_norm": 0.03072691150009632, "learning_rate": 2.7558423532427936e-07, "loss": 0.0004, "step": 221100 }, { "epoch": 3.6179334042379123, "grad_norm": 0.047646183520555496, "learning_rate": 2.7535052978500674e-07, "loss": 0.0001, "step": 221110 }, { "epoch": 3.618097030188988, "grad_norm": 0.001304120640270412, "learning_rate": 2.751169205773191e-07, "loss": 0.0002, "step": 221120 }, { "epoch": 3.618260656140064, "grad_norm": 0.004851606208831072, "learning_rate": 2.748834077059781e-07, "loss": 0.0003, "step": 221130 }, { "epoch": 3.61842428209114, "grad_norm": 0.0004974418552592397, "learning_rate": 2.7464999117574664e-07, "loss": 0.0003, "step": 221140 }, { "epoch": 3.6185879080422154, "grad_norm": 0.0023540998809039593, "learning_rate": 2.744166709913826e-07, "loss": 0.0001, "step": 221150 }, { "epoch": 3.6187515339932914, "grad_norm": 0.06862397491931915, "learning_rate": 2.741834471576449e-07, "loss": 0.0004, "step": 221160 }, { "epoch": 3.6189151599443674, "grad_norm": 0.002355284756049514, "learning_rate": 2.739503196792864e-07, "loss": 0.0002, "step": 221170 }, { "epoch": 3.619078785895443, "grad_norm": 0.005086830351501703, "learning_rate": 2.737172885610612e-07, "loss": 0.0002, "step": 221180 }, { "epoch": 3.619242411846519, "grad_norm": 0.01036439836025238, "learning_rate": 2.734843538077214e-07, "loss": 0.0003, "step": 221190 }, { "epoch": 3.619406037797595, "grad_norm": 0.041497062891721725, "learning_rate": 2.7325151542401673e-07, "loss": 0.0003, "step": 221200 }, { "epoch": 3.6195696637486705, "grad_norm": 0.012556852772831917, "learning_rate": 2.730187734146922e-07, "loss": 0.0002, "step": 221210 }, { "epoch": 3.6197332896997465, "grad_norm": 0.11705506592988968, "learning_rate": 2.7278612778449563e-07, "loss": 0.0003, "step": 221220 }, { "epoch": 3.619896915650822, "grad_norm": 0.007624284364283085, "learning_rate": 2.725535785381689e-07, "loss": 0.0003, "step": 221230 }, { "epoch": 3.620060541601898, "grad_norm": 0.004360627382993698, "learning_rate": 2.7232112568045534e-07, "loss": 0.0002, "step": 221240 }, { "epoch": 3.620224167552974, "grad_norm": 0.05223497748374939, "learning_rate": 2.7208876921609227e-07, "loss": 0.0001, "step": 221250 }, { "epoch": 3.6203877935040496, "grad_norm": 0.07860083878040314, "learning_rate": 2.718565091498182e-07, "loss": 0.0002, "step": 221260 }, { "epoch": 3.6205514194551256, "grad_norm": 0.006239154376089573, "learning_rate": 2.716243454863693e-07, "loss": 0.0002, "step": 221270 }, { "epoch": 3.620715045406201, "grad_norm": 0.043036483228206635, "learning_rate": 2.713922782304784e-07, "loss": 0.0002, "step": 221280 }, { "epoch": 3.620878671357277, "grad_norm": 0.08612391352653503, "learning_rate": 2.711603073868774e-07, "loss": 0.0007, "step": 221290 }, { "epoch": 3.621042297308353, "grad_norm": 0.001414684928022325, "learning_rate": 2.7092843296029627e-07, "loss": 0.0002, "step": 221300 }, { "epoch": 3.6212059232594287, "grad_norm": 0.04543793201446533, "learning_rate": 2.7069665495546184e-07, "loss": 0.0002, "step": 221310 }, { "epoch": 3.6213695492105047, "grad_norm": 0.06004747003316879, "learning_rate": 2.7046497337710144e-07, "loss": 0.0002, "step": 221320 }, { "epoch": 3.6215331751615807, "grad_norm": 0.0018310948507860303, "learning_rate": 2.702333882299374e-07, "loss": 0.0002, "step": 221330 }, { "epoch": 3.6216968011126562, "grad_norm": 0.008180365897715092, "learning_rate": 2.700018995186915e-07, "loss": 0.0001, "step": 221340 }, { "epoch": 3.6218604270637322, "grad_norm": 0.11164918541908264, "learning_rate": 2.697705072480844e-07, "loss": 0.001, "step": 221350 }, { "epoch": 3.6220240530148082, "grad_norm": 0.00768665736541152, "learning_rate": 2.6953921142283457e-07, "loss": 0.0003, "step": 221360 }, { "epoch": 3.622187678965884, "grad_norm": 0.00463187275454402, "learning_rate": 2.6930801204765657e-07, "loss": 0.0002, "step": 221370 }, { "epoch": 3.62235130491696, "grad_norm": 0.021261828020215034, "learning_rate": 2.6907690912726435e-07, "loss": 0.0004, "step": 221380 }, { "epoch": 3.6225149308680358, "grad_norm": 0.008761519566178322, "learning_rate": 2.688459026663709e-07, "loss": 0.0003, "step": 221390 }, { "epoch": 3.6226785568191113, "grad_norm": 0.06593596935272217, "learning_rate": 2.686149926696863e-07, "loss": 0.0003, "step": 221400 }, { "epoch": 3.6228421827701873, "grad_norm": 0.04048804193735123, "learning_rate": 2.683841791419178e-07, "loss": 0.0003, "step": 221410 }, { "epoch": 3.6230058087212633, "grad_norm": 0.039934683591127396, "learning_rate": 2.681534620877713e-07, "loss": 0.0004, "step": 221420 }, { "epoch": 3.623169434672339, "grad_norm": 0.21571668982505798, "learning_rate": 2.679228415119517e-07, "loss": 0.0003, "step": 221430 }, { "epoch": 3.623333060623415, "grad_norm": 0.0165594182908535, "learning_rate": 2.676923174191615e-07, "loss": 0.0004, "step": 221440 }, { "epoch": 3.623496686574491, "grad_norm": 0.013223931193351746, "learning_rate": 2.674618898140996e-07, "loss": 0.0001, "step": 221450 }, { "epoch": 3.6236603125255664, "grad_norm": 0.006486556027084589, "learning_rate": 2.6723155870146454e-07, "loss": 0.0001, "step": 221460 }, { "epoch": 3.6238239384766424, "grad_norm": 0.012204650789499283, "learning_rate": 2.670013240859531e-07, "loss": 0.0002, "step": 221470 }, { "epoch": 3.6239875644277184, "grad_norm": 0.011113543063402176, "learning_rate": 2.667711859722605e-07, "loss": 0.0004, "step": 221480 }, { "epoch": 3.624151190378794, "grad_norm": 0.01940695010125637, "learning_rate": 2.6654114436507673e-07, "loss": 0.0002, "step": 221490 }, { "epoch": 3.62431481632987, "grad_norm": 0.015484541654586792, "learning_rate": 2.6631119926909365e-07, "loss": 0.0003, "step": 221500 }, { "epoch": 3.624478442280946, "grad_norm": 0.0057901423424482346, "learning_rate": 2.660813506889992e-07, "loss": 0.0003, "step": 221510 }, { "epoch": 3.6246420682320215, "grad_norm": 0.05928637459874153, "learning_rate": 2.6585159862948073e-07, "loss": 0.0002, "step": 221520 }, { "epoch": 3.6248056941830975, "grad_norm": 0.01084611564874649, "learning_rate": 2.6562194309522114e-07, "loss": 0.0001, "step": 221530 }, { "epoch": 3.6249693201341735, "grad_norm": 0.06323028355836868, "learning_rate": 2.6539238409090393e-07, "loss": 0.0003, "step": 221540 }, { "epoch": 3.625132946085249, "grad_norm": 0.05434966832399368, "learning_rate": 2.651629216212093e-07, "loss": 0.0002, "step": 221550 }, { "epoch": 3.625296572036325, "grad_norm": 0.06752750277519226, "learning_rate": 2.6493355569081615e-07, "loss": 0.0002, "step": 221560 }, { "epoch": 3.625460197987401, "grad_norm": 0.07895198464393616, "learning_rate": 2.6470428630440026e-07, "loss": 0.0003, "step": 221570 }, { "epoch": 3.6256238239384766, "grad_norm": 0.007606458850204945, "learning_rate": 2.6447511346663734e-07, "loss": 0.0004, "step": 221580 }, { "epoch": 3.6257874498895526, "grad_norm": 0.0002166902122553438, "learning_rate": 2.6424603718219866e-07, "loss": 0.0003, "step": 221590 }, { "epoch": 3.6259510758406286, "grad_norm": 0.014801227487623692, "learning_rate": 2.640170574557566e-07, "loss": 0.0002, "step": 221600 }, { "epoch": 3.626114701791704, "grad_norm": 0.0015739407390356064, "learning_rate": 2.6378817429197845e-07, "loss": 0.0001, "step": 221610 }, { "epoch": 3.62627832774278, "grad_norm": 0.00474692415446043, "learning_rate": 2.635593876955322e-07, "loss": 0.0005, "step": 221620 }, { "epoch": 3.6264419536938557, "grad_norm": 0.0163098257035017, "learning_rate": 2.6333069767107975e-07, "loss": 0.0003, "step": 221630 }, { "epoch": 3.6266055796449317, "grad_norm": 0.05949067696928978, "learning_rate": 2.6310210422328787e-07, "loss": 0.0003, "step": 221640 }, { "epoch": 3.6267692055960077, "grad_norm": 0.0231317225843668, "learning_rate": 2.62873607356815e-07, "loss": 0.0005, "step": 221650 }, { "epoch": 3.6269328315470832, "grad_norm": 0.10164646804332733, "learning_rate": 2.626452070763208e-07, "loss": 0.0004, "step": 221660 }, { "epoch": 3.6270964574981592, "grad_norm": 0.05209675431251526, "learning_rate": 2.62416903386461e-07, "loss": 0.0001, "step": 221670 }, { "epoch": 3.627260083449235, "grad_norm": 0.004109958652406931, "learning_rate": 2.6218869629189235e-07, "loss": 0.0004, "step": 221680 }, { "epoch": 3.627423709400311, "grad_norm": 0.03910335525870323, "learning_rate": 2.619605857972662e-07, "loss": 0.0002, "step": 221690 }, { "epoch": 3.6275873353513868, "grad_norm": 0.07686314731836319, "learning_rate": 2.6173257190723445e-07, "loss": 0.0002, "step": 221700 }, { "epoch": 3.6277509613024623, "grad_norm": 0.012399506755173206, "learning_rate": 2.615046546264449e-07, "loss": 0.0004, "step": 221710 }, { "epoch": 3.6279145872535383, "grad_norm": 0.06363947689533234, "learning_rate": 2.6127683395954674e-07, "loss": 0.0004, "step": 221720 }, { "epoch": 3.6280782132046143, "grad_norm": 0.05279615521430969, "learning_rate": 2.610491099111834e-07, "loss": 0.0004, "step": 221730 }, { "epoch": 3.62824183915569, "grad_norm": 0.01435253769159317, "learning_rate": 2.608214824859984e-07, "loss": 0.0003, "step": 221740 }, { "epoch": 3.628405465106766, "grad_norm": 0.041237153112888336, "learning_rate": 2.605939516886319e-07, "loss": 0.0002, "step": 221750 }, { "epoch": 3.628569091057842, "grad_norm": 0.0011238270672038198, "learning_rate": 2.603665175237252e-07, "loss": 0.0003, "step": 221760 }, { "epoch": 3.6287327170089174, "grad_norm": 0.00017956407100427896, "learning_rate": 2.601391799959141e-07, "loss": 0.0002, "step": 221770 }, { "epoch": 3.6288963429599934, "grad_norm": 0.0035401079803705215, "learning_rate": 2.599119391098343e-07, "loss": 0.0003, "step": 221780 }, { "epoch": 3.6290599689110694, "grad_norm": 0.022117653861641884, "learning_rate": 2.596847948701181e-07, "loss": 0.0002, "step": 221790 }, { "epoch": 3.629223594862145, "grad_norm": 0.0030634570866823196, "learning_rate": 2.594577472813975e-07, "loss": 0.0001, "step": 221800 }, { "epoch": 3.629387220813221, "grad_norm": 4.695123789133504e-05, "learning_rate": 2.592307963483015e-07, "loss": 0.0, "step": 221810 }, { "epoch": 3.629550846764297, "grad_norm": 0.011206506751477718, "learning_rate": 2.5900394207545865e-07, "loss": 0.0002, "step": 221820 }, { "epoch": 3.6297144727153725, "grad_norm": 0.05152999237179756, "learning_rate": 2.5877718446749246e-07, "loss": 0.0002, "step": 221830 }, { "epoch": 3.6298780986664485, "grad_norm": 0.024221302941441536, "learning_rate": 2.5855052352902754e-07, "loss": 0.0004, "step": 221840 }, { "epoch": 3.6300417246175245, "grad_norm": 0.16471216082572937, "learning_rate": 2.583239592646847e-07, "loss": 0.0007, "step": 221850 }, { "epoch": 3.6302053505686, "grad_norm": 0.06823086738586426, "learning_rate": 2.580974916790846e-07, "loss": 0.0002, "step": 221860 }, { "epoch": 3.630368976519676, "grad_norm": 0.0014128488255664706, "learning_rate": 2.578711207768425e-07, "loss": 0.0005, "step": 221870 }, { "epoch": 3.630532602470752, "grad_norm": 0.023637404665350914, "learning_rate": 2.576448465625758e-07, "loss": 0.0002, "step": 221880 }, { "epoch": 3.6306962284218276, "grad_norm": 0.016232099384069443, "learning_rate": 2.5741866904089753e-07, "loss": 0.0002, "step": 221890 }, { "epoch": 3.6308598543729036, "grad_norm": 0.05810902267694473, "learning_rate": 2.5719258821641947e-07, "loss": 0.0004, "step": 221900 }, { "epoch": 3.6310234803239796, "grad_norm": 0.0061738574877381325, "learning_rate": 2.569666040937502e-07, "loss": 0.0001, "step": 221910 }, { "epoch": 3.631187106275055, "grad_norm": 0.0013912237482145429, "learning_rate": 2.567407166774988e-07, "loss": 0.0002, "step": 221920 }, { "epoch": 3.631350732226131, "grad_norm": 0.007655891589820385, "learning_rate": 2.5651492597226935e-07, "loss": 0.0001, "step": 221930 }, { "epoch": 3.631514358177207, "grad_norm": 0.026471586897969246, "learning_rate": 2.5628923198266764e-07, "loss": 0.0003, "step": 221940 }, { "epoch": 3.6316779841282827, "grad_norm": 0.005631709471344948, "learning_rate": 2.560636347132928e-07, "loss": 0.0004, "step": 221950 }, { "epoch": 3.6318416100793587, "grad_norm": 0.0023052552714943886, "learning_rate": 2.558381341687466e-07, "loss": 0.0004, "step": 221960 }, { "epoch": 3.6320052360304347, "grad_norm": 0.08785330504179001, "learning_rate": 2.5561273035362546e-07, "loss": 0.0002, "step": 221970 }, { "epoch": 3.6321688619815102, "grad_norm": 0.023137811571359634, "learning_rate": 2.5538742327252674e-07, "loss": 0.0002, "step": 221980 }, { "epoch": 3.6323324879325862, "grad_norm": 0.06532911956310272, "learning_rate": 2.551622129300424e-07, "loss": 0.0002, "step": 221990 }, { "epoch": 3.6324961138836622, "grad_norm": 0.008326050825417042, "learning_rate": 2.5493709933076593e-07, "loss": 0.0, "step": 222000 }, { "epoch": 3.6326597398347378, "grad_norm": 0.043210480362176895, "learning_rate": 2.5471208247928593e-07, "loss": 0.0002, "step": 222010 }, { "epoch": 3.6328233657858138, "grad_norm": 0.014023893512785435, "learning_rate": 2.5448716238019145e-07, "loss": 0.0002, "step": 222020 }, { "epoch": 3.6329869917368893, "grad_norm": 0.03639298304915428, "learning_rate": 2.5426233903806716e-07, "loss": 0.0003, "step": 222030 }, { "epoch": 3.6331506176879653, "grad_norm": 0.0009640603675507009, "learning_rate": 2.5403761245749783e-07, "loss": 0.0002, "step": 222040 }, { "epoch": 3.633314243639041, "grad_norm": 0.018544552847743034, "learning_rate": 2.538129826430652e-07, "loss": 0.0002, "step": 222050 }, { "epoch": 3.633477869590117, "grad_norm": 0.005245222244411707, "learning_rate": 2.5358844959935016e-07, "loss": 0.0004, "step": 222060 }, { "epoch": 3.633641495541193, "grad_norm": 0.009561249054968357, "learning_rate": 2.53364013330929e-07, "loss": 0.0005, "step": 222070 }, { "epoch": 3.6338051214922684, "grad_norm": 0.0737030953168869, "learning_rate": 2.531396738423791e-07, "loss": 0.0002, "step": 222080 }, { "epoch": 3.6339687474433444, "grad_norm": 0.019319036975502968, "learning_rate": 2.5291543113827364e-07, "loss": 0.0003, "step": 222090 }, { "epoch": 3.6341323733944204, "grad_norm": 0.05251697823405266, "learning_rate": 2.5269128522318664e-07, "loss": 0.0004, "step": 222100 }, { "epoch": 3.634295999345496, "grad_norm": 0.00024244106316473335, "learning_rate": 2.5246723610168553e-07, "loss": 0.0004, "step": 222110 }, { "epoch": 3.634459625296572, "grad_norm": 0.0020961337722837925, "learning_rate": 2.522432837783406e-07, "loss": 0.0002, "step": 222120 }, { "epoch": 3.634623251247648, "grad_norm": 0.02453775890171528, "learning_rate": 2.5201942825771654e-07, "loss": 0.0001, "step": 222130 }, { "epoch": 3.6347868771987235, "grad_norm": 0.0026497209910303354, "learning_rate": 2.5179566954437964e-07, "loss": 0.0002, "step": 222140 }, { "epoch": 3.6349505031497995, "grad_norm": 0.016599224880337715, "learning_rate": 2.515720076428896e-07, "loss": 0.0002, "step": 222150 }, { "epoch": 3.6351141291008755, "grad_norm": 0.03367682918906212, "learning_rate": 2.513484425578083e-07, "loss": 0.0003, "step": 222160 }, { "epoch": 3.635277755051951, "grad_norm": 0.0757489874958992, "learning_rate": 2.511249742936939e-07, "loss": 0.0001, "step": 222170 }, { "epoch": 3.635441381003027, "grad_norm": 0.053739383816719055, "learning_rate": 2.5090160285510257e-07, "loss": 0.0004, "step": 222180 }, { "epoch": 3.635605006954103, "grad_norm": 0.05908568575978279, "learning_rate": 2.5067832824658854e-07, "loss": 0.0003, "step": 222190 }, { "epoch": 3.6357686329051786, "grad_norm": 0.00845622643828392, "learning_rate": 2.5045515047270375e-07, "loss": 0.0002, "step": 222200 }, { "epoch": 3.6359322588562546, "grad_norm": 0.03877120092511177, "learning_rate": 2.5023206953799893e-07, "loss": 0.0002, "step": 222210 }, { "epoch": 3.6360958848073306, "grad_norm": 0.036680955439805984, "learning_rate": 2.500090854470238e-07, "loss": 0.0004, "step": 222220 }, { "epoch": 3.636259510758406, "grad_norm": 0.003286132588982582, "learning_rate": 2.4978619820432313e-07, "loss": 0.0001, "step": 222230 }, { "epoch": 3.636423136709482, "grad_norm": 0.14565710723400116, "learning_rate": 2.4956340781444154e-07, "loss": 0.0002, "step": 222240 }, { "epoch": 3.636586762660558, "grad_norm": 0.05380932614207268, "learning_rate": 2.493407142819221e-07, "loss": 0.0002, "step": 222250 }, { "epoch": 3.6367503886116337, "grad_norm": 9.831043280428275e-05, "learning_rate": 2.491181176113061e-07, "loss": 0.0001, "step": 222260 }, { "epoch": 3.6369140145627097, "grad_norm": 0.006782910320907831, "learning_rate": 2.4889561780713054e-07, "loss": 0.0002, "step": 222270 }, { "epoch": 3.6370776405137857, "grad_norm": 0.0013882059138268232, "learning_rate": 2.4867321487393294e-07, "loss": 0.0001, "step": 222280 }, { "epoch": 3.6372412664648612, "grad_norm": 0.010778030380606651, "learning_rate": 2.484509088162468e-07, "loss": 0.0003, "step": 222290 }, { "epoch": 3.6374048924159372, "grad_norm": 0.01011816319078207, "learning_rate": 2.482286996386063e-07, "loss": 0.0002, "step": 222300 }, { "epoch": 3.6375685183670132, "grad_norm": 0.006870212033390999, "learning_rate": 2.480065873455412e-07, "loss": 0.0002, "step": 222310 }, { "epoch": 3.637732144318089, "grad_norm": 0.0002218058070866391, "learning_rate": 2.477845719415811e-07, "loss": 0.0001, "step": 222320 }, { "epoch": 3.6378957702691648, "grad_norm": 0.02319510653614998, "learning_rate": 2.475626534312503e-07, "loss": 0.0001, "step": 222330 }, { "epoch": 3.6380593962202408, "grad_norm": 0.05109257251024246, "learning_rate": 2.4734083181907677e-07, "loss": 0.0001, "step": 222340 }, { "epoch": 3.6382230221713163, "grad_norm": 0.04057138413190842, "learning_rate": 2.4711910710958075e-07, "loss": 0.0004, "step": 222350 }, { "epoch": 3.6383866481223923, "grad_norm": 0.0414644293487072, "learning_rate": 2.468974793072848e-07, "loss": 0.0002, "step": 222360 }, { "epoch": 3.6385502740734683, "grad_norm": 0.0006175526650622487, "learning_rate": 2.4667594841670527e-07, "loss": 0.0004, "step": 222370 }, { "epoch": 3.638713900024544, "grad_norm": 0.060443926602602005, "learning_rate": 2.464545144423625e-07, "loss": 0.0002, "step": 222380 }, { "epoch": 3.63887752597562, "grad_norm": 0.05499015003442764, "learning_rate": 2.462331773887683e-07, "loss": 0.0004, "step": 222390 }, { "epoch": 3.6390411519266954, "grad_norm": 0.016445674002170563, "learning_rate": 2.460119372604375e-07, "loss": 0.0002, "step": 222400 }, { "epoch": 3.6392047778777714, "grad_norm": 0.036451637744903564, "learning_rate": 2.4579079406187867e-07, "loss": 0.0001, "step": 222410 }, { "epoch": 3.6393684038288474, "grad_norm": 0.021354783326387405, "learning_rate": 2.455697477976038e-07, "loss": 0.0002, "step": 222420 }, { "epoch": 3.639532029779923, "grad_norm": 0.08973333984613419, "learning_rate": 2.453487984721176e-07, "loss": 0.0003, "step": 222430 }, { "epoch": 3.639695655730999, "grad_norm": 0.027174049988389015, "learning_rate": 2.451279460899264e-07, "loss": 0.0003, "step": 222440 }, { "epoch": 3.6398592816820745, "grad_norm": 0.03102390468120575, "learning_rate": 2.4490719065553173e-07, "loss": 0.0004, "step": 222450 }, { "epoch": 3.6400229076331505, "grad_norm": 0.0006685377447865903, "learning_rate": 2.446865321734354e-07, "loss": 0.0003, "step": 222460 }, { "epoch": 3.6401865335842265, "grad_norm": 0.0057618399150669575, "learning_rate": 2.444659706481362e-07, "loss": 0.0004, "step": 222470 }, { "epoch": 3.640350159535302, "grad_norm": 0.04371798038482666, "learning_rate": 2.4424550608413254e-07, "loss": 0.0001, "step": 222480 }, { "epoch": 3.640513785486378, "grad_norm": 0.000291552918497473, "learning_rate": 2.440251384859171e-07, "loss": 0.0004, "step": 222490 }, { "epoch": 3.640677411437454, "grad_norm": 0.005980882793664932, "learning_rate": 2.4380486785798516e-07, "loss": 0.0008, "step": 222500 }, { "epoch": 3.6408410373885296, "grad_norm": 0.0010658357059583068, "learning_rate": 2.4358469420482643e-07, "loss": 0.0004, "step": 222510 }, { "epoch": 3.6410046633396056, "grad_norm": 0.0030470641795545816, "learning_rate": 2.433646175309312e-07, "loss": 0.0004, "step": 222520 }, { "epoch": 3.6411682892906816, "grad_norm": 0.014152798801660538, "learning_rate": 2.4314463784078537e-07, "loss": 0.0002, "step": 222530 }, { "epoch": 3.641331915241757, "grad_norm": 0.0007580578094348311, "learning_rate": 2.4292475513887523e-07, "loss": 0.0003, "step": 222540 }, { "epoch": 3.641495541192833, "grad_norm": 0.005594814196228981, "learning_rate": 2.427049694296829e-07, "loss": 0.0001, "step": 222550 }, { "epoch": 3.641659167143909, "grad_norm": 0.06924402713775635, "learning_rate": 2.424852807176914e-07, "loss": 0.0002, "step": 222560 }, { "epoch": 3.6418227930949847, "grad_norm": 0.057679757475852966, "learning_rate": 2.4226568900737823e-07, "loss": 0.0004, "step": 222570 }, { "epoch": 3.6419864190460607, "grad_norm": 0.1741264909505844, "learning_rate": 2.4204619430322153e-07, "loss": 0.0004, "step": 222580 }, { "epoch": 3.6421500449971367, "grad_norm": 0.10359813272953033, "learning_rate": 2.4182679660969597e-07, "loss": 0.0004, "step": 222590 }, { "epoch": 3.6423136709482122, "grad_norm": 0.015121842734515667, "learning_rate": 2.4160749593127643e-07, "loss": 0.0001, "step": 222600 }, { "epoch": 3.6424772968992882, "grad_norm": 0.0006785944569855928, "learning_rate": 2.4138829227243255e-07, "loss": 0.0006, "step": 222610 }, { "epoch": 3.6426409228503642, "grad_norm": 0.0931183472275734, "learning_rate": 2.4116918563763416e-07, "loss": 0.0003, "step": 222620 }, { "epoch": 3.64280454880144, "grad_norm": 0.0009202767978422344, "learning_rate": 2.4095017603134874e-07, "loss": 0.0002, "step": 222630 }, { "epoch": 3.6429681747525158, "grad_norm": 0.25765594840049744, "learning_rate": 2.407312634580422e-07, "loss": 0.0004, "step": 222640 }, { "epoch": 3.6431318007035918, "grad_norm": 0.012778124772012234, "learning_rate": 2.4051244792217764e-07, "loss": 0.0001, "step": 222650 }, { "epoch": 3.6432954266546673, "grad_norm": 0.004803916439414024, "learning_rate": 2.4029372942821594e-07, "loss": 0.0003, "step": 222660 }, { "epoch": 3.6434590526057433, "grad_norm": 0.025197722017765045, "learning_rate": 2.400751079806174e-07, "loss": 0.0004, "step": 222670 }, { "epoch": 3.6436226785568193, "grad_norm": 0.02022751048207283, "learning_rate": 2.3985658358383956e-07, "loss": 0.0002, "step": 222680 }, { "epoch": 3.643786304507895, "grad_norm": 0.04673852398991585, "learning_rate": 2.396381562423372e-07, "loss": 0.0004, "step": 222690 }, { "epoch": 3.643949930458971, "grad_norm": 0.016548829153180122, "learning_rate": 2.39419825960564e-07, "loss": 0.0002, "step": 222700 }, { "epoch": 3.644113556410047, "grad_norm": 0.034154534339904785, "learning_rate": 2.392015927429725e-07, "loss": 0.0003, "step": 222710 }, { "epoch": 3.6442771823611224, "grad_norm": 0.0022249529138207436, "learning_rate": 2.389834565940119e-07, "loss": 0.0003, "step": 222720 }, { "epoch": 3.6444408083121984, "grad_norm": 0.01364803034812212, "learning_rate": 2.3876541751812864e-07, "loss": 0.0001, "step": 222730 }, { "epoch": 3.6446044342632744, "grad_norm": 0.03228280320763588, "learning_rate": 2.385474755197692e-07, "loss": 0.0003, "step": 222740 }, { "epoch": 3.64476806021435, "grad_norm": 0.021811723709106445, "learning_rate": 2.3832963060337778e-07, "loss": 0.0002, "step": 222750 }, { "epoch": 3.644931686165426, "grad_norm": 0.004728931933641434, "learning_rate": 2.381118827733958e-07, "loss": 0.0002, "step": 222760 }, { "epoch": 3.645095312116502, "grad_norm": 0.029722336679697037, "learning_rate": 2.3789423203426197e-07, "loss": 0.0001, "step": 222770 }, { "epoch": 3.6452589380675775, "grad_norm": 0.04955332726240158, "learning_rate": 2.3767667839041497e-07, "loss": 0.0001, "step": 222780 }, { "epoch": 3.6454225640186535, "grad_norm": 0.006591263227164745, "learning_rate": 2.3745922184629012e-07, "loss": 0.0001, "step": 222790 }, { "epoch": 3.645586189969729, "grad_norm": 0.011604624800384045, "learning_rate": 2.3724186240632218e-07, "loss": 0.0005, "step": 222800 }, { "epoch": 3.645749815920805, "grad_norm": 0.0025527547113597393, "learning_rate": 2.37024600074941e-07, "loss": 0.0002, "step": 222810 }, { "epoch": 3.645913441871881, "grad_norm": 0.015286874026060104, "learning_rate": 2.3680743485657797e-07, "loss": 0.0003, "step": 222820 }, { "epoch": 3.6460770678229566, "grad_norm": 0.016737457364797592, "learning_rate": 2.3659036675566015e-07, "loss": 0.0003, "step": 222830 }, { "epoch": 3.6462406937740326, "grad_norm": 0.09065087884664536, "learning_rate": 2.3637339577661454e-07, "loss": 0.0001, "step": 222840 }, { "epoch": 3.646404319725108, "grad_norm": 0.028349727392196655, "learning_rate": 2.3615652192386317e-07, "loss": 0.0001, "step": 222850 }, { "epoch": 3.646567945676184, "grad_norm": 0.004210163839161396, "learning_rate": 2.359397452018286e-07, "loss": 0.0004, "step": 222860 }, { "epoch": 3.64673157162726, "grad_norm": 0.02070852369070053, "learning_rate": 2.357230656149312e-07, "loss": 0.0003, "step": 222870 }, { "epoch": 3.6468951975783357, "grad_norm": 0.0013595735654234886, "learning_rate": 2.3550648316758906e-07, "loss": 0.0001, "step": 222880 }, { "epoch": 3.6470588235294117, "grad_norm": 0.05638625845313072, "learning_rate": 2.3528999786421758e-07, "loss": 0.0008, "step": 222890 }, { "epoch": 3.6472224494804877, "grad_norm": 0.005445600021630526, "learning_rate": 2.35073609709231e-07, "loss": 0.0002, "step": 222900 }, { "epoch": 3.6473860754315632, "grad_norm": 0.06505828350782394, "learning_rate": 2.3485731870703964e-07, "loss": 0.0004, "step": 222910 }, { "epoch": 3.6475497013826392, "grad_norm": 0.049980562180280685, "learning_rate": 2.346411248620567e-07, "loss": 0.0002, "step": 222920 }, { "epoch": 3.6477133273337152, "grad_norm": 0.02423831820487976, "learning_rate": 2.344250281786875e-07, "loss": 0.0001, "step": 222930 }, { "epoch": 3.647876953284791, "grad_norm": 0.048476871103048325, "learning_rate": 2.3420902866133966e-07, "loss": 0.0003, "step": 222940 }, { "epoch": 3.6480405792358668, "grad_norm": 0.05554354563355446, "learning_rate": 2.3399312631441517e-07, "loss": 0.0007, "step": 222950 }, { "epoch": 3.6482042051869428, "grad_norm": 0.056216273456811905, "learning_rate": 2.337773211423189e-07, "loss": 0.0003, "step": 222960 }, { "epoch": 3.6483678311380183, "grad_norm": 0.034305036067962646, "learning_rate": 2.3356161314944891e-07, "loss": 0.0001, "step": 222970 }, { "epoch": 3.6485314570890943, "grad_norm": 0.003954124636948109, "learning_rate": 2.3334600234020455e-07, "loss": 0.0002, "step": 222980 }, { "epoch": 3.6486950830401703, "grad_norm": 0.023791123181581497, "learning_rate": 2.3313048871898004e-07, "loss": 0.0005, "step": 222990 }, { "epoch": 3.648858708991246, "grad_norm": 0.026336761191487312, "learning_rate": 2.329150722901724e-07, "loss": 0.0004, "step": 223000 }, { "epoch": 3.649022334942322, "grad_norm": 0.05763920769095421, "learning_rate": 2.3269975305817093e-07, "loss": 0.0006, "step": 223010 }, { "epoch": 3.649185960893398, "grad_norm": 0.0024084732867777348, "learning_rate": 2.3248453102736824e-07, "loss": 0.0003, "step": 223020 }, { "epoch": 3.6493495868444734, "grad_norm": 0.0015352348564192653, "learning_rate": 2.3226940620214967e-07, "loss": 0.0001, "step": 223030 }, { "epoch": 3.6495132127955494, "grad_norm": 0.005127063952386379, "learning_rate": 2.3205437858690449e-07, "loss": 0.0003, "step": 223040 }, { "epoch": 3.6496768387466254, "grad_norm": 0.00045310912537388504, "learning_rate": 2.318394481860148e-07, "loss": 0.0001, "step": 223050 }, { "epoch": 3.649840464697701, "grad_norm": 0.0009803699795156717, "learning_rate": 2.3162461500386423e-07, "loss": 0.0003, "step": 223060 }, { "epoch": 3.650004090648777, "grad_norm": 0.05335928872227669, "learning_rate": 2.3140987904483104e-07, "loss": 0.0003, "step": 223070 }, { "epoch": 3.650167716599853, "grad_norm": 0.035663291811943054, "learning_rate": 2.3119524031329665e-07, "loss": 0.0004, "step": 223080 }, { "epoch": 3.6503313425509285, "grad_norm": 0.022760750725865364, "learning_rate": 2.3098069881363426e-07, "loss": 0.0002, "step": 223090 }, { "epoch": 3.6504949685020045, "grad_norm": 0.046590302139520645, "learning_rate": 2.3076625455022094e-07, "loss": 0.0004, "step": 223100 }, { "epoch": 3.6506585944530805, "grad_norm": 0.037932563573122025, "learning_rate": 2.305519075274265e-07, "loss": 0.0004, "step": 223110 }, { "epoch": 3.650822220404156, "grad_norm": 0.007590991444885731, "learning_rate": 2.3033765774962247e-07, "loss": 0.0002, "step": 223120 }, { "epoch": 3.650985846355232, "grad_norm": 0.02291824482381344, "learning_rate": 2.3012350522117699e-07, "loss": 0.0002, "step": 223130 }, { "epoch": 3.651149472306308, "grad_norm": 0.03159964084625244, "learning_rate": 2.2990944994645713e-07, "loss": 0.0004, "step": 223140 }, { "epoch": 3.6513130982573836, "grad_norm": 0.0036177916917949915, "learning_rate": 2.2969549192982666e-07, "loss": 0.0004, "step": 223150 }, { "epoch": 3.6514767242084596, "grad_norm": 0.0024434616789221764, "learning_rate": 2.2948163117564758e-07, "loss": 0.0001, "step": 223160 }, { "epoch": 3.651640350159535, "grad_norm": 0.05207856371998787, "learning_rate": 2.2926786768828146e-07, "loss": 0.0005, "step": 223170 }, { "epoch": 3.651803976110611, "grad_norm": 0.051499366760253906, "learning_rate": 2.2905420147208645e-07, "loss": 0.0004, "step": 223180 }, { "epoch": 3.651967602061687, "grad_norm": 0.0026784634683281183, "learning_rate": 2.2884063253141798e-07, "loss": 0.0002, "step": 223190 }, { "epoch": 3.6521312280127627, "grad_norm": 0.015557069331407547, "learning_rate": 2.2862716087063142e-07, "loss": 0.0001, "step": 223200 }, { "epoch": 3.6522948539638387, "grad_norm": 0.021824071183800697, "learning_rate": 2.2841378649407886e-07, "loss": 0.0002, "step": 223210 }, { "epoch": 3.6524584799149142, "grad_norm": 0.023173972964286804, "learning_rate": 2.2820050940611184e-07, "loss": 0.0002, "step": 223220 }, { "epoch": 3.6526221058659902, "grad_norm": 0.07528048753738403, "learning_rate": 2.2798732961107793e-07, "loss": 0.0004, "step": 223230 }, { "epoch": 3.6527857318170662, "grad_norm": 0.00860598310828209, "learning_rate": 2.2777424711332375e-07, "loss": 0.0004, "step": 223240 }, { "epoch": 3.652949357768142, "grad_norm": 0.007318541407585144, "learning_rate": 2.275612619171935e-07, "loss": 0.0002, "step": 223250 }, { "epoch": 3.6531129837192178, "grad_norm": 0.009732317179441452, "learning_rate": 2.2734837402703159e-07, "loss": 0.0004, "step": 223260 }, { "epoch": 3.6532766096702938, "grad_norm": 0.04360820725560188, "learning_rate": 2.2713558344717667e-07, "loss": 0.0001, "step": 223270 }, { "epoch": 3.6534402356213693, "grad_norm": 0.0348820723593235, "learning_rate": 2.2692289018196756e-07, "loss": 0.0003, "step": 223280 }, { "epoch": 3.6536038615724453, "grad_norm": 0.02512947842478752, "learning_rate": 2.2671029423574187e-07, "loss": 0.0005, "step": 223290 }, { "epoch": 3.6537674875235213, "grad_norm": 0.04783888906240463, "learning_rate": 2.2649779561283391e-07, "loss": 0.0002, "step": 223300 }, { "epoch": 3.653931113474597, "grad_norm": 0.04555639997124672, "learning_rate": 2.2628539431757578e-07, "loss": 0.0004, "step": 223310 }, { "epoch": 3.654094739425673, "grad_norm": 0.0006142130587249994, "learning_rate": 2.2607309035429847e-07, "loss": 0.0004, "step": 223320 }, { "epoch": 3.654258365376749, "grad_norm": 0.0016901030903682113, "learning_rate": 2.2586088372733127e-07, "loss": 0.0003, "step": 223330 }, { "epoch": 3.6544219913278244, "grad_norm": 0.00016652216436341405, "learning_rate": 2.2564877444100075e-07, "loss": 0.0002, "step": 223340 }, { "epoch": 3.6545856172789004, "grad_norm": 0.04121949523687363, "learning_rate": 2.2543676249963008e-07, "loss": 0.0002, "step": 223350 }, { "epoch": 3.6547492432299764, "grad_norm": 0.0027063358575105667, "learning_rate": 2.252248479075436e-07, "loss": 0.0001, "step": 223360 }, { "epoch": 3.654912869181052, "grad_norm": 0.0006130837718956172, "learning_rate": 2.2501303066906178e-07, "loss": 0.0004, "step": 223370 }, { "epoch": 3.655076495132128, "grad_norm": 0.0037655190099030733, "learning_rate": 2.2480131078850386e-07, "loss": 0.0002, "step": 223380 }, { "epoch": 3.655240121083204, "grad_norm": 0.001165079534985125, "learning_rate": 2.2458968827018536e-07, "loss": 0.0003, "step": 223390 }, { "epoch": 3.6554037470342795, "grad_norm": 0.006552441045641899, "learning_rate": 2.243781631184222e-07, "loss": 0.0002, "step": 223400 }, { "epoch": 3.6555673729853555, "grad_norm": 0.011884589679539204, "learning_rate": 2.2416673533752597e-07, "loss": 0.0004, "step": 223410 }, { "epoch": 3.6557309989364315, "grad_norm": 0.0029862301889806986, "learning_rate": 2.2395540493180989e-07, "loss": 0.0002, "step": 223420 }, { "epoch": 3.655894624887507, "grad_norm": 0.0003093531122431159, "learning_rate": 2.2374417190557995e-07, "loss": 0.0001, "step": 223430 }, { "epoch": 3.656058250838583, "grad_norm": 0.015231412835419178, "learning_rate": 2.2353303626314438e-07, "loss": 0.0002, "step": 223440 }, { "epoch": 3.656221876789659, "grad_norm": 0.008553998544812202, "learning_rate": 2.2332199800880804e-07, "loss": 0.0004, "step": 223450 }, { "epoch": 3.6563855027407346, "grad_norm": 0.006397528108209372, "learning_rate": 2.231110571468742e-07, "loss": 0.0004, "step": 223460 }, { "epoch": 3.6565491286918106, "grad_norm": 0.033021047711372375, "learning_rate": 2.2290021368164272e-07, "loss": 0.0004, "step": 223470 }, { "epoch": 3.6567127546428866, "grad_norm": 0.163517028093338, "learning_rate": 2.226894676174135e-07, "loss": 0.0004, "step": 223480 }, { "epoch": 3.656876380593962, "grad_norm": 0.03084159456193447, "learning_rate": 2.2247881895848257e-07, "loss": 0.0002, "step": 223490 }, { "epoch": 3.657040006545038, "grad_norm": 0.04151791334152222, "learning_rate": 2.222682677091459e-07, "loss": 0.0003, "step": 223500 }, { "epoch": 3.657203632496114, "grad_norm": 0.008614029735326767, "learning_rate": 2.2205781387369564e-07, "loss": 0.0001, "step": 223510 }, { "epoch": 3.6573672584471897, "grad_norm": 0.014290066435933113, "learning_rate": 2.2184745745642334e-07, "loss": 0.0, "step": 223520 }, { "epoch": 3.6575308843982657, "grad_norm": 0.07315405458211899, "learning_rate": 2.2163719846161725e-07, "loss": 0.0008, "step": 223530 }, { "epoch": 3.6576945103493417, "grad_norm": 0.0008647298673167825, "learning_rate": 2.214270368935656e-07, "loss": 0.0001, "step": 223540 }, { "epoch": 3.6578581363004172, "grad_norm": 0.0008103896980173886, "learning_rate": 2.212169727565522e-07, "loss": 0.0001, "step": 223550 }, { "epoch": 3.6580217622514932, "grad_norm": 0.0013179603265598416, "learning_rate": 2.210070060548608e-07, "loss": 0.0002, "step": 223560 }, { "epoch": 3.6581853882025688, "grad_norm": 0.000597736332565546, "learning_rate": 2.207971367927708e-07, "loss": 0.0006, "step": 223570 }, { "epoch": 3.6583490141536448, "grad_norm": 0.03552775830030441, "learning_rate": 2.2058736497456436e-07, "loss": 0.0004, "step": 223580 }, { "epoch": 3.6585126401047208, "grad_norm": 0.0024957142304629087, "learning_rate": 2.203776906045163e-07, "loss": 0.0002, "step": 223590 }, { "epoch": 3.6586762660557963, "grad_norm": 0.03174237906932831, "learning_rate": 2.2016811368690217e-07, "loss": 0.0002, "step": 223600 }, { "epoch": 3.6588398920068723, "grad_norm": 0.026384666562080383, "learning_rate": 2.1995863422599407e-07, "loss": 0.0002, "step": 223610 }, { "epoch": 3.659003517957948, "grad_norm": 0.002245464129373431, "learning_rate": 2.197492522260658e-07, "loss": 0.0002, "step": 223620 }, { "epoch": 3.659167143909024, "grad_norm": 0.0029103418346494436, "learning_rate": 2.1953996769138342e-07, "loss": 0.0001, "step": 223630 }, { "epoch": 3.6593307698601, "grad_norm": 0.00020632527593988925, "learning_rate": 2.1933078062621682e-07, "loss": 0.0002, "step": 223640 }, { "epoch": 3.6594943958111754, "grad_norm": 0.03384645655751228, "learning_rate": 2.1912169103482815e-07, "loss": 0.0001, "step": 223650 }, { "epoch": 3.6596580217622514, "grad_norm": 0.03024369291961193, "learning_rate": 2.1891269892148348e-07, "loss": 0.0002, "step": 223660 }, { "epoch": 3.6598216477133274, "grad_norm": 0.02480211853981018, "learning_rate": 2.1870380429044215e-07, "loss": 0.0003, "step": 223670 }, { "epoch": 3.659985273664403, "grad_norm": 0.029749276116490364, "learning_rate": 2.184950071459646e-07, "loss": 0.0003, "step": 223680 }, { "epoch": 3.660148899615479, "grad_norm": 0.005379140842705965, "learning_rate": 2.182863074923064e-07, "loss": 0.0003, "step": 223690 }, { "epoch": 3.660312525566555, "grad_norm": 0.04828926548361778, "learning_rate": 2.1807770533372462e-07, "loss": 0.0003, "step": 223700 }, { "epoch": 3.6604761515176305, "grad_norm": 0.07247430086135864, "learning_rate": 2.1786920067447094e-07, "loss": 0.0002, "step": 223710 }, { "epoch": 3.6606397774687065, "grad_norm": 0.017550652846693993, "learning_rate": 2.1766079351879743e-07, "loss": 0.0006, "step": 223720 }, { "epoch": 3.6608034034197825, "grad_norm": 0.0004990443121641874, "learning_rate": 2.1745248387095296e-07, "loss": 0.0002, "step": 223730 }, { "epoch": 3.660967029370858, "grad_norm": 0.0007377996225841343, "learning_rate": 2.172442717351847e-07, "loss": 0.0003, "step": 223740 }, { "epoch": 3.661130655321934, "grad_norm": 0.008913499303162098, "learning_rate": 2.170361571157381e-07, "loss": 0.0001, "step": 223750 }, { "epoch": 3.66129428127301, "grad_norm": 0.03959730640053749, "learning_rate": 2.168281400168576e-07, "loss": 0.0002, "step": 223760 }, { "epoch": 3.6614579072240856, "grad_norm": 0.025140179321169853, "learning_rate": 2.1662022044278197e-07, "loss": 0.0003, "step": 223770 }, { "epoch": 3.6616215331751616, "grad_norm": 0.00794900767505169, "learning_rate": 2.1641239839775286e-07, "loss": 0.0005, "step": 223780 }, { "epoch": 3.6617851591262376, "grad_norm": 0.03685695677995682, "learning_rate": 2.1620467388600575e-07, "loss": 0.0003, "step": 223790 }, { "epoch": 3.661948785077313, "grad_norm": 0.0404885932803154, "learning_rate": 2.159970469117778e-07, "loss": 0.0003, "step": 223800 }, { "epoch": 3.662112411028389, "grad_norm": 0.030124299228191376, "learning_rate": 2.1578951747930121e-07, "loss": 0.0003, "step": 223810 }, { "epoch": 3.662276036979465, "grad_norm": 0.003683381946757436, "learning_rate": 2.1558208559280701e-07, "loss": 0.0005, "step": 223820 }, { "epoch": 3.6624396629305407, "grad_norm": 0.009710224345326424, "learning_rate": 2.1537475125652518e-07, "loss": 0.0003, "step": 223830 }, { "epoch": 3.6626032888816167, "grad_norm": 0.019023610278964043, "learning_rate": 2.1516751447468398e-07, "loss": 0.0001, "step": 223840 }, { "epoch": 3.6627669148326927, "grad_norm": 0.027980562299489975, "learning_rate": 2.149603752515067e-07, "loss": 0.0003, "step": 223850 }, { "epoch": 3.6629305407837682, "grad_norm": 0.002171497093513608, "learning_rate": 2.147533335912183e-07, "loss": 0.0002, "step": 223860 }, { "epoch": 3.6630941667348442, "grad_norm": 0.006353230215609074, "learning_rate": 2.145463894980393e-07, "loss": 0.0002, "step": 223870 }, { "epoch": 3.6632577926859202, "grad_norm": 0.009615246206521988, "learning_rate": 2.1433954297619076e-07, "loss": 0.0001, "step": 223880 }, { "epoch": 3.6634214186369958, "grad_norm": 0.02289809286594391, "learning_rate": 2.1413279402988763e-07, "loss": 0.0003, "step": 223890 }, { "epoch": 3.6635850445880718, "grad_norm": 0.05358545854687691, "learning_rate": 2.139261426633471e-07, "loss": 0.0005, "step": 223900 }, { "epoch": 3.6637486705391478, "grad_norm": 0.007723763585090637, "learning_rate": 2.1371958888078136e-07, "loss": 0.0004, "step": 223910 }, { "epoch": 3.6639122964902233, "grad_norm": 0.0377817302942276, "learning_rate": 2.1351313268640373e-07, "loss": 0.0003, "step": 223920 }, { "epoch": 3.6640759224412993, "grad_norm": 0.057673804461956024, "learning_rate": 2.1330677408442191e-07, "loss": 0.0004, "step": 223930 }, { "epoch": 3.664239548392375, "grad_norm": 0.010668080300092697, "learning_rate": 2.1310051307904422e-07, "loss": 0.0001, "step": 223940 }, { "epoch": 3.664403174343451, "grad_norm": 0.0022955930326133966, "learning_rate": 2.1289434967447563e-07, "loss": 0.0004, "step": 223950 }, { "epoch": 3.664566800294527, "grad_norm": 0.057870618999004364, "learning_rate": 2.1268828387492114e-07, "loss": 0.0007, "step": 223960 }, { "epoch": 3.6647304262456024, "grad_norm": 0.11625471711158752, "learning_rate": 2.1248231568457956e-07, "loss": 0.0005, "step": 223970 }, { "epoch": 3.6648940521966784, "grad_norm": 0.029946699738502502, "learning_rate": 2.1227644510765255e-07, "loss": 0.0004, "step": 223980 }, { "epoch": 3.665057678147754, "grad_norm": 0.030289726331830025, "learning_rate": 2.1207067214833677e-07, "loss": 0.0005, "step": 223990 }, { "epoch": 3.66522130409883, "grad_norm": 0.005498743150383234, "learning_rate": 2.1186499681082884e-07, "loss": 0.0001, "step": 224000 }, { "epoch": 3.665384930049906, "grad_norm": 0.00048453384079039097, "learning_rate": 2.1165941909932098e-07, "loss": 0.0003, "step": 224010 }, { "epoch": 3.6655485560009815, "grad_norm": 0.06925543397665024, "learning_rate": 2.1145393901800482e-07, "loss": 0.0002, "step": 224020 }, { "epoch": 3.6657121819520575, "grad_norm": 0.007346542552113533, "learning_rate": 2.1124855657107035e-07, "loss": 0.0002, "step": 224030 }, { "epoch": 3.6658758079031335, "grad_norm": 0.022516941651701927, "learning_rate": 2.110432717627059e-07, "loss": 0.0002, "step": 224040 }, { "epoch": 3.666039433854209, "grad_norm": 0.2475382685661316, "learning_rate": 2.1083808459709588e-07, "loss": 0.001, "step": 224050 }, { "epoch": 3.666203059805285, "grad_norm": 0.005958435125648975, "learning_rate": 2.1063299507842362e-07, "loss": 0.0002, "step": 224060 }, { "epoch": 3.666366685756361, "grad_norm": 0.03719722107052803, "learning_rate": 2.1042800321087186e-07, "loss": 0.0007, "step": 224070 }, { "epoch": 3.6665303117074366, "grad_norm": 0.05232291668653488, "learning_rate": 2.1022310899862065e-07, "loss": 0.0003, "step": 224080 }, { "epoch": 3.6666939376585126, "grad_norm": 0.008709120564162731, "learning_rate": 2.1001831244584547e-07, "loss": 0.0003, "step": 224090 }, { "epoch": 3.6668575636095886, "grad_norm": 0.005069750361144543, "learning_rate": 2.0981361355672358e-07, "loss": 0.0001, "step": 224100 }, { "epoch": 3.667021189560664, "grad_norm": 0.0010914114536717534, "learning_rate": 2.0960901233542775e-07, "loss": 0.0001, "step": 224110 }, { "epoch": 3.66718481551174, "grad_norm": 0.09415298700332642, "learning_rate": 2.0940450878613073e-07, "loss": 0.0004, "step": 224120 }, { "epoch": 3.667348441462816, "grad_norm": 0.0024666115641593933, "learning_rate": 2.092001029130014e-07, "loss": 0.0001, "step": 224130 }, { "epoch": 3.6675120674138917, "grad_norm": 0.015519184060394764, "learning_rate": 2.0899579472020704e-07, "loss": 0.0002, "step": 224140 }, { "epoch": 3.6676756933649677, "grad_norm": 0.05251674726605415, "learning_rate": 2.0879158421191426e-07, "loss": 0.0003, "step": 224150 }, { "epoch": 3.6678393193160437, "grad_norm": 0.0022912295535206795, "learning_rate": 2.0858747139228698e-07, "loss": 0.0007, "step": 224160 }, { "epoch": 3.6680029452671192, "grad_norm": 0.022236987948417664, "learning_rate": 2.0838345626548517e-07, "loss": 0.0003, "step": 224170 }, { "epoch": 3.6681665712181952, "grad_norm": 0.001887085265479982, "learning_rate": 2.0817953883566944e-07, "loss": 0.0003, "step": 224180 }, { "epoch": 3.6683301971692712, "grad_norm": 0.05288973078131676, "learning_rate": 2.079757191069981e-07, "loss": 0.0002, "step": 224190 }, { "epoch": 3.6684938231203468, "grad_norm": 0.030842037871479988, "learning_rate": 2.0777199708362672e-07, "loss": 0.0002, "step": 224200 }, { "epoch": 3.6686574490714228, "grad_norm": 0.039250992238521576, "learning_rate": 2.075683727697081e-07, "loss": 0.0002, "step": 224210 }, { "epoch": 3.6688210750224988, "grad_norm": 0.03556126728653908, "learning_rate": 2.07364846169395e-07, "loss": 0.0004, "step": 224220 }, { "epoch": 3.6689847009735743, "grad_norm": 0.0009344244608655572, "learning_rate": 2.071614172868358e-07, "loss": 0.0003, "step": 224230 }, { "epoch": 3.6691483269246503, "grad_norm": 0.012156287208199501, "learning_rate": 2.069580861261805e-07, "loss": 0.0002, "step": 224240 }, { "epoch": 3.6693119528757263, "grad_norm": 0.038361839950084686, "learning_rate": 2.0675485269157248e-07, "loss": 0.0002, "step": 224250 }, { "epoch": 3.669475578826802, "grad_norm": 0.041224405169487, "learning_rate": 2.0655171698715725e-07, "loss": 0.0003, "step": 224260 }, { "epoch": 3.669639204777878, "grad_norm": 0.019870955497026443, "learning_rate": 2.0634867901707435e-07, "loss": 0.0, "step": 224270 }, { "epoch": 3.669802830728954, "grad_norm": 0.0008749638800509274, "learning_rate": 2.0614573878546652e-07, "loss": 0.0001, "step": 224280 }, { "epoch": 3.6699664566800294, "grad_norm": 0.0011442812392488122, "learning_rate": 2.0594289629646934e-07, "loss": 0.0002, "step": 224290 }, { "epoch": 3.6701300826311054, "grad_norm": 0.0009800121188163757, "learning_rate": 2.0574015155422012e-07, "loss": 0.0001, "step": 224300 }, { "epoch": 3.6702937085821814, "grad_norm": 0.005116192623972893, "learning_rate": 2.0553750456285048e-07, "loss": 0.0008, "step": 224310 }, { "epoch": 3.670457334533257, "grad_norm": 0.0955733135342598, "learning_rate": 2.0533495532649494e-07, "loss": 0.0002, "step": 224320 }, { "epoch": 3.670620960484333, "grad_norm": 0.027510248124599457, "learning_rate": 2.0513250384928075e-07, "loss": 0.0003, "step": 224330 }, { "epoch": 3.6707845864354085, "grad_norm": 0.05546894669532776, "learning_rate": 2.0493015013533846e-07, "loss": 0.0002, "step": 224340 }, { "epoch": 3.6709482123864845, "grad_norm": 0.02439967356622219, "learning_rate": 2.0472789418879035e-07, "loss": 0.0002, "step": 224350 }, { "epoch": 3.6711118383375605, "grad_norm": 0.24186937510967255, "learning_rate": 2.0452573601376369e-07, "loss": 0.0007, "step": 224360 }, { "epoch": 3.671275464288636, "grad_norm": 0.0445609949529171, "learning_rate": 2.043236756143785e-07, "loss": 0.0009, "step": 224370 }, { "epoch": 3.671439090239712, "grad_norm": 0.04285275191068649, "learning_rate": 2.0412171299475535e-07, "loss": 0.0002, "step": 224380 }, { "epoch": 3.6716027161907876, "grad_norm": 0.0368671640753746, "learning_rate": 2.03919848159011e-07, "loss": 0.0002, "step": 224390 }, { "epoch": 3.6717663421418636, "grad_norm": 0.05388995260000229, "learning_rate": 2.0371808111126212e-07, "loss": 0.0002, "step": 224400 }, { "epoch": 3.6719299680929396, "grad_norm": 0.000875636818818748, "learning_rate": 2.0351641185562264e-07, "loss": 0.0002, "step": 224410 }, { "epoch": 3.672093594044015, "grad_norm": 0.04183351993560791, "learning_rate": 2.0331484039620487e-07, "loss": 0.0002, "step": 224420 }, { "epoch": 3.672257219995091, "grad_norm": 0.07863053679466248, "learning_rate": 2.0311336673711767e-07, "loss": 0.0002, "step": 224430 }, { "epoch": 3.672420845946167, "grad_norm": 0.0005673730047419667, "learning_rate": 2.0291199088246894e-07, "loss": 0.0002, "step": 224440 }, { "epoch": 3.6725844718972427, "grad_norm": 0.038655221462249756, "learning_rate": 2.027107128363648e-07, "loss": 0.0002, "step": 224450 }, { "epoch": 3.6727480978483187, "grad_norm": 0.005975736305117607, "learning_rate": 2.025095326029103e-07, "loss": 0.0001, "step": 224460 }, { "epoch": 3.6729117237993947, "grad_norm": 0.010383501648902893, "learning_rate": 2.0230845018620548e-07, "loss": 0.0002, "step": 224470 }, { "epoch": 3.6730753497504702, "grad_norm": 0.02358400635421276, "learning_rate": 2.0210746559035099e-07, "loss": 0.0002, "step": 224480 }, { "epoch": 3.6732389757015462, "grad_norm": 0.06332036852836609, "learning_rate": 2.019065788194452e-07, "loss": 0.0003, "step": 224490 }, { "epoch": 3.6734026016526222, "grad_norm": 0.0058388845063745975, "learning_rate": 2.017057898775837e-07, "loss": 0.0004, "step": 224500 }, { "epoch": 3.6735662276036978, "grad_norm": 0.00042793588363565505, "learning_rate": 2.0150509876885992e-07, "loss": 0.0001, "step": 224510 }, { "epoch": 3.6737298535547738, "grad_norm": 0.0027511543594300747, "learning_rate": 2.0130450549736668e-07, "loss": 0.0005, "step": 224520 }, { "epoch": 3.6738934795058498, "grad_norm": 0.027761518955230713, "learning_rate": 2.0110401006719294e-07, "loss": 0.0004, "step": 224530 }, { "epoch": 3.6740571054569253, "grad_norm": 0.00221710535697639, "learning_rate": 2.0090361248242762e-07, "loss": 0.0002, "step": 224540 }, { "epoch": 3.6742207314080013, "grad_norm": 0.018120765686035156, "learning_rate": 2.007033127471558e-07, "loss": 0.0002, "step": 224550 }, { "epoch": 3.6743843573590773, "grad_norm": 0.007500549778342247, "learning_rate": 2.0050311086546148e-07, "loss": 0.0002, "step": 224560 }, { "epoch": 3.674547983310153, "grad_norm": 0.029735274612903595, "learning_rate": 2.0030300684142745e-07, "loss": 0.0005, "step": 224570 }, { "epoch": 3.674711609261229, "grad_norm": 0.028520308434963226, "learning_rate": 2.0010300067913323e-07, "loss": 0.0002, "step": 224580 }, { "epoch": 3.674875235212305, "grad_norm": 0.047377876937389374, "learning_rate": 1.9990309238265614e-07, "loss": 0.0006, "step": 224590 }, { "epoch": 3.6750388611633804, "grad_norm": 0.0031049465760588646, "learning_rate": 1.997032819560729e-07, "loss": 0.0002, "step": 224600 }, { "epoch": 3.6752024871144564, "grad_norm": 0.0033735528122633696, "learning_rate": 1.9950356940345694e-07, "loss": 0.0001, "step": 224610 }, { "epoch": 3.6753661130655324, "grad_norm": 0.023983612656593323, "learning_rate": 1.9930395472888163e-07, "loss": 0.0002, "step": 224620 }, { "epoch": 3.675529739016608, "grad_norm": 0.021695489063858986, "learning_rate": 1.9910443793641489e-07, "loss": 0.0001, "step": 224630 }, { "epoch": 3.675693364967684, "grad_norm": 0.025400109589099884, "learning_rate": 1.989050190301256e-07, "loss": 0.0006, "step": 224640 }, { "epoch": 3.67585699091876, "grad_norm": 0.013275683857500553, "learning_rate": 1.9870569801408e-07, "loss": 0.0001, "step": 224650 }, { "epoch": 3.6760206168698355, "grad_norm": 0.05683034658432007, "learning_rate": 1.985064748923421e-07, "loss": 0.0001, "step": 224660 }, { "epoch": 3.6761842428209115, "grad_norm": 0.028387250378727913, "learning_rate": 1.9830734966897302e-07, "loss": 0.0002, "step": 224670 }, { "epoch": 3.6763478687719875, "grad_norm": 0.03130973130464554, "learning_rate": 1.98108322348034e-07, "loss": 0.0005, "step": 224680 }, { "epoch": 3.676511494723063, "grad_norm": 0.0019519282504916191, "learning_rate": 1.979093929335818e-07, "loss": 0.0003, "step": 224690 }, { "epoch": 3.676675120674139, "grad_norm": 0.0008069977047853172, "learning_rate": 1.977105614296737e-07, "loss": 0.0002, "step": 224700 }, { "epoch": 3.6768387466252146, "grad_norm": 0.02621155045926571, "learning_rate": 1.9751182784036315e-07, "loss": 0.0004, "step": 224710 }, { "epoch": 3.6770023725762906, "grad_norm": 0.005702508147805929, "learning_rate": 1.9731319216970135e-07, "loss": 0.0003, "step": 224720 }, { "epoch": 3.6771659985273666, "grad_norm": 0.03097398579120636, "learning_rate": 1.9711465442173894e-07, "loss": 0.0002, "step": 224730 }, { "epoch": 3.677329624478442, "grad_norm": 0.010002704337239265, "learning_rate": 1.969162146005249e-07, "loss": 0.0002, "step": 224740 }, { "epoch": 3.677493250429518, "grad_norm": 0.029602807015180588, "learning_rate": 1.9671787271010379e-07, "loss": 0.0002, "step": 224750 }, { "epoch": 3.6776568763805937, "grad_norm": 0.03577108308672905, "learning_rate": 1.9651962875452069e-07, "loss": 0.0002, "step": 224760 }, { "epoch": 3.6778205023316697, "grad_norm": 0.04453574866056442, "learning_rate": 1.9632148273781626e-07, "loss": 0.0003, "step": 224770 }, { "epoch": 3.6779841282827457, "grad_norm": 0.003120914800092578, "learning_rate": 1.9612343466403227e-07, "loss": 0.0002, "step": 224780 }, { "epoch": 3.6781477542338212, "grad_norm": 0.0017977968091145158, "learning_rate": 1.959254845372055e-07, "loss": 0.0003, "step": 224790 }, { "epoch": 3.6783113801848972, "grad_norm": 0.02225528098642826, "learning_rate": 1.9572763236137272e-07, "loss": 0.0004, "step": 224800 }, { "epoch": 3.6784750061359732, "grad_norm": 0.012809978798031807, "learning_rate": 1.955298781405668e-07, "loss": 0.0003, "step": 224810 }, { "epoch": 3.6786386320870488, "grad_norm": 0.03769908845424652, "learning_rate": 1.9533222187882172e-07, "loss": 0.0004, "step": 224820 }, { "epoch": 3.6788022580381248, "grad_norm": 0.037920113652944565, "learning_rate": 1.9513466358016598e-07, "loss": 0.0004, "step": 224830 }, { "epoch": 3.6789658839892008, "grad_norm": 0.010638969019055367, "learning_rate": 1.9493720324862854e-07, "loss": 0.0003, "step": 224840 }, { "epoch": 3.6791295099402763, "grad_norm": 0.10767169296741486, "learning_rate": 1.9473984088823338e-07, "loss": 0.0002, "step": 224850 }, { "epoch": 3.6792931358913523, "grad_norm": 0.03185885772109032, "learning_rate": 1.9454257650300733e-07, "loss": 0.0003, "step": 224860 }, { "epoch": 3.6794567618424283, "grad_norm": 0.0387774333357811, "learning_rate": 1.9434541009697105e-07, "loss": 0.0004, "step": 224870 }, { "epoch": 3.679620387793504, "grad_norm": 0.017079602926969528, "learning_rate": 1.941483416741452e-07, "loss": 0.0002, "step": 224880 }, { "epoch": 3.67978401374458, "grad_norm": 0.03793046995997429, "learning_rate": 1.9395137123854656e-07, "loss": 0.0002, "step": 224890 }, { "epoch": 3.679947639695656, "grad_norm": 0.013928724452853203, "learning_rate": 1.9375449879419306e-07, "loss": 0.0003, "step": 224900 }, { "epoch": 3.6801112656467314, "grad_norm": 0.009413233026862144, "learning_rate": 1.9355772434509702e-07, "loss": 0.0002, "step": 224910 }, { "epoch": 3.6802748915978074, "grad_norm": 0.011628407053649426, "learning_rate": 1.933610478952719e-07, "loss": 0.0005, "step": 224920 }, { "epoch": 3.6804385175488834, "grad_norm": 0.031155990436673164, "learning_rate": 1.9316446944872558e-07, "loss": 0.0004, "step": 224930 }, { "epoch": 3.680602143499959, "grad_norm": 0.012927750125527382, "learning_rate": 1.9296798900946935e-07, "loss": 0.0001, "step": 224940 }, { "epoch": 3.680765769451035, "grad_norm": 0.07043320685625076, "learning_rate": 1.9277160658150662e-07, "loss": 0.0004, "step": 224950 }, { "epoch": 3.680929395402111, "grad_norm": 0.012949627824127674, "learning_rate": 1.9257532216884368e-07, "loss": 0.0001, "step": 224960 }, { "epoch": 3.6810930213531865, "grad_norm": 0.0026265361811965704, "learning_rate": 1.923791357754795e-07, "loss": 0.0003, "step": 224970 }, { "epoch": 3.6812566473042625, "grad_norm": 0.10110542923212051, "learning_rate": 1.921830474054176e-07, "loss": 0.0004, "step": 224980 }, { "epoch": 3.6814202732553385, "grad_norm": 0.02653605118393898, "learning_rate": 1.9198705706265363e-07, "loss": 0.0002, "step": 224990 }, { "epoch": 3.681583899206414, "grad_norm": 0.013652574270963669, "learning_rate": 1.91791164751185e-07, "loss": 0.0002, "step": 225000 }, { "epoch": 3.68174752515749, "grad_norm": 0.0181294996291399, "learning_rate": 1.9159537047500454e-07, "loss": 0.0001, "step": 225010 }, { "epoch": 3.681911151108566, "grad_norm": 0.012735758908092976, "learning_rate": 1.913996742381058e-07, "loss": 0.0001, "step": 225020 }, { "epoch": 3.6820747770596416, "grad_norm": 0.00022493646247312427, "learning_rate": 1.9120407604447776e-07, "loss": 0.0003, "step": 225030 }, { "epoch": 3.6822384030107176, "grad_norm": 0.05551691725850105, "learning_rate": 1.9100857589810952e-07, "loss": 0.0002, "step": 225040 }, { "epoch": 3.6824020289617936, "grad_norm": 0.002484729513525963, "learning_rate": 1.908131738029856e-07, "loss": 0.0003, "step": 225050 }, { "epoch": 3.682565654912869, "grad_norm": 0.00792468711733818, "learning_rate": 1.9061786976309116e-07, "loss": 0.0002, "step": 225060 }, { "epoch": 3.682729280863945, "grad_norm": 0.002581672975793481, "learning_rate": 1.904226637824086e-07, "loss": 0.0002, "step": 225070 }, { "epoch": 3.682892906815021, "grad_norm": 0.0014945428119972348, "learning_rate": 1.9022755586491747e-07, "loss": 0.0003, "step": 225080 }, { "epoch": 3.6830565327660967, "grad_norm": 0.11258947849273682, "learning_rate": 1.9003254601459577e-07, "loss": 0.0003, "step": 225090 }, { "epoch": 3.6832201587171727, "grad_norm": 0.0009849534835666418, "learning_rate": 1.8983763423541913e-07, "loss": 0.0003, "step": 225100 }, { "epoch": 3.6833837846682482, "grad_norm": 0.010158960707485676, "learning_rate": 1.8964282053136273e-07, "loss": 0.0002, "step": 225110 }, { "epoch": 3.6835474106193242, "grad_norm": 0.01217112597078085, "learning_rate": 1.8944810490639897e-07, "loss": 0.0003, "step": 225120 }, { "epoch": 3.6837110365704, "grad_norm": 0.09986032545566559, "learning_rate": 1.8925348736449577e-07, "loss": 0.0001, "step": 225130 }, { "epoch": 3.6838746625214758, "grad_norm": 0.03613521531224251, "learning_rate": 1.890589679096233e-07, "loss": 0.0003, "step": 225140 }, { "epoch": 3.6840382884725518, "grad_norm": 0.0005031516193412244, "learning_rate": 1.888645465457467e-07, "loss": 0.0002, "step": 225150 }, { "epoch": 3.6842019144236273, "grad_norm": 0.0022174702025949955, "learning_rate": 1.886702232768306e-07, "loss": 0.0004, "step": 225160 }, { "epoch": 3.6843655403747033, "grad_norm": 0.006444720551371574, "learning_rate": 1.8847599810683626e-07, "loss": 0.0001, "step": 225170 }, { "epoch": 3.6845291663257793, "grad_norm": 0.0881890133023262, "learning_rate": 1.882818710397244e-07, "loss": 0.0003, "step": 225180 }, { "epoch": 3.684692792276855, "grad_norm": 0.08531233668327332, "learning_rate": 1.8808784207945295e-07, "loss": 0.0004, "step": 225190 }, { "epoch": 3.684856418227931, "grad_norm": 0.0031530545093119144, "learning_rate": 1.8789391122997879e-07, "loss": 0.0001, "step": 225200 }, { "epoch": 3.685020044179007, "grad_norm": 0.00026989670004695654, "learning_rate": 1.877000784952543e-07, "loss": 0.0001, "step": 225210 }, { "epoch": 3.6851836701300824, "grad_norm": 0.006896248087286949, "learning_rate": 1.875063438792324e-07, "loss": 0.0001, "step": 225220 }, { "epoch": 3.6853472960811584, "grad_norm": 0.08017553389072418, "learning_rate": 1.8731270738586382e-07, "loss": 0.0005, "step": 225230 }, { "epoch": 3.6855109220322344, "grad_norm": 0.01404279563575983, "learning_rate": 1.87119169019096e-07, "loss": 0.0002, "step": 225240 }, { "epoch": 3.68567454798331, "grad_norm": 0.003923621494323015, "learning_rate": 1.8692572878287519e-07, "loss": 0.0001, "step": 225250 }, { "epoch": 3.685838173934386, "grad_norm": 0.0004167342558503151, "learning_rate": 1.867323866811449e-07, "loss": 0.0003, "step": 225260 }, { "epoch": 3.686001799885462, "grad_norm": 0.0017833082238212228, "learning_rate": 1.8653914271784813e-07, "loss": 0.0001, "step": 225270 }, { "epoch": 3.6861654258365375, "grad_norm": 0.01679542288184166, "learning_rate": 1.8634599689692502e-07, "loss": 0.0003, "step": 225280 }, { "epoch": 3.6863290517876135, "grad_norm": 0.028078848496079445, "learning_rate": 1.861529492223124e-07, "loss": 0.0002, "step": 225290 }, { "epoch": 3.6864926777386895, "grad_norm": 0.0032142414711415768, "learning_rate": 1.859599996979472e-07, "loss": 0.0004, "step": 225300 }, { "epoch": 3.686656303689765, "grad_norm": 0.03818904235959053, "learning_rate": 1.8576714832776287e-07, "loss": 0.0005, "step": 225310 }, { "epoch": 3.686819929640841, "grad_norm": 0.022243190556764603, "learning_rate": 1.8557439511569353e-07, "loss": 0.0006, "step": 225320 }, { "epoch": 3.686983555591917, "grad_norm": 0.005756955593824387, "learning_rate": 1.8538174006566657e-07, "loss": 0.0003, "step": 225330 }, { "epoch": 3.6871471815429926, "grad_norm": 0.025332903489470482, "learning_rate": 1.851891831816116e-07, "loss": 0.0003, "step": 225340 }, { "epoch": 3.6873108074940686, "grad_norm": 0.03929131478071213, "learning_rate": 1.8499672446745388e-07, "loss": 0.0002, "step": 225350 }, { "epoch": 3.6874744334451446, "grad_norm": 0.006257663946598768, "learning_rate": 1.848043639271191e-07, "loss": 0.0001, "step": 225360 }, { "epoch": 3.68763805939622, "grad_norm": 0.003416887717321515, "learning_rate": 1.846121015645269e-07, "loss": 0.0002, "step": 225370 }, { "epoch": 3.687801685347296, "grad_norm": 0.0012500769225880504, "learning_rate": 1.844199373835992e-07, "loss": 0.0002, "step": 225380 }, { "epoch": 3.687965311298372, "grad_norm": 0.027936192229390144, "learning_rate": 1.842278713882534e-07, "loss": 0.0002, "step": 225390 }, { "epoch": 3.6881289372494477, "grad_norm": 0.008672156371176243, "learning_rate": 1.8403590358240576e-07, "loss": 0.0003, "step": 225400 }, { "epoch": 3.6882925632005237, "grad_norm": 0.010034026578068733, "learning_rate": 1.838440339699704e-07, "loss": 0.0002, "step": 225410 }, { "epoch": 3.6884561891515997, "grad_norm": 0.027200615033507347, "learning_rate": 1.8365226255485868e-07, "loss": 0.0004, "step": 225420 }, { "epoch": 3.6886198151026752, "grad_norm": 0.03498778119683266, "learning_rate": 1.8346058934098131e-07, "loss": 0.0001, "step": 225430 }, { "epoch": 3.688783441053751, "grad_norm": 0.0009556656586937606, "learning_rate": 1.8326901433224631e-07, "loss": 0.0003, "step": 225440 }, { "epoch": 3.688947067004827, "grad_norm": 0.007340965326875448, "learning_rate": 1.830775375325594e-07, "loss": 0.0003, "step": 225450 }, { "epoch": 3.6891106929559028, "grad_norm": 0.044486068189144135, "learning_rate": 1.828861589458253e-07, "loss": 0.0002, "step": 225460 }, { "epoch": 3.6892743189069788, "grad_norm": 0.025438837707042694, "learning_rate": 1.8269487857594526e-07, "loss": 0.0004, "step": 225470 }, { "epoch": 3.6894379448580548, "grad_norm": 0.017934419214725494, "learning_rate": 1.8250369642682007e-07, "loss": 0.0002, "step": 225480 }, { "epoch": 3.6896015708091303, "grad_norm": 0.02019430696964264, "learning_rate": 1.8231261250234667e-07, "loss": 0.0002, "step": 225490 }, { "epoch": 3.6897651967602063, "grad_norm": 0.0004494196909945458, "learning_rate": 1.8212162680642297e-07, "loss": 0.0004, "step": 225500 }, { "epoch": 3.689928822711282, "grad_norm": 0.007000397890806198, "learning_rate": 1.8193073934294037e-07, "loss": 0.0004, "step": 225510 }, { "epoch": 3.690092448662358, "grad_norm": 0.03734799101948738, "learning_rate": 1.817399501157935e-07, "loss": 0.0001, "step": 225520 }, { "epoch": 3.690256074613434, "grad_norm": 0.006407733075320721, "learning_rate": 1.8154925912887034e-07, "loss": 0.0002, "step": 225530 }, { "epoch": 3.6904197005645094, "grad_norm": 0.010397421196103096, "learning_rate": 1.8135866638606114e-07, "loss": 0.0003, "step": 225540 }, { "epoch": 3.6905833265155854, "grad_norm": 0.004343076609075069, "learning_rate": 1.8116817189124892e-07, "loss": 0.0001, "step": 225550 }, { "epoch": 3.690746952466661, "grad_norm": 0.1143331378698349, "learning_rate": 1.8097777564832053e-07, "loss": 0.0004, "step": 225560 }, { "epoch": 3.690910578417737, "grad_norm": 0.04041124880313873, "learning_rate": 1.8078747766115678e-07, "loss": 0.0006, "step": 225570 }, { "epoch": 3.691074204368813, "grad_norm": 0.001838890602812171, "learning_rate": 1.805972779336379e-07, "loss": 0.0002, "step": 225580 }, { "epoch": 3.6912378303198885, "grad_norm": 0.030845297500491142, "learning_rate": 1.804071764696408e-07, "loss": 0.0005, "step": 225590 }, { "epoch": 3.6914014562709645, "grad_norm": 0.015826130285859108, "learning_rate": 1.80217173273044e-07, "loss": 0.0002, "step": 225600 }, { "epoch": 3.6915650822220405, "grad_norm": 0.018039187416434288, "learning_rate": 1.8002726834771888e-07, "loss": 0.0002, "step": 225610 }, { "epoch": 3.691728708173116, "grad_norm": 0.011779425665736198, "learning_rate": 1.7983746169753902e-07, "loss": 0.0005, "step": 225620 }, { "epoch": 3.691892334124192, "grad_norm": 0.000815345614682883, "learning_rate": 1.7964775332637296e-07, "loss": 0.0002, "step": 225630 }, { "epoch": 3.692055960075268, "grad_norm": 0.001953249564394355, "learning_rate": 1.7945814323809096e-07, "loss": 0.0002, "step": 225640 }, { "epoch": 3.6922195860263436, "grad_norm": 0.06161058694124222, "learning_rate": 1.792686314365566e-07, "loss": 0.0002, "step": 225650 }, { "epoch": 3.6923832119774196, "grad_norm": 0.005643610842525959, "learning_rate": 1.790792179256362e-07, "loss": 0.0002, "step": 225660 }, { "epoch": 3.6925468379284956, "grad_norm": 0.009083697572350502, "learning_rate": 1.7888990270918948e-07, "loss": 0.0004, "step": 225670 }, { "epoch": 3.692710463879571, "grad_norm": 0.022028062492609024, "learning_rate": 1.7870068579107724e-07, "loss": 0.0002, "step": 225680 }, { "epoch": 3.692874089830647, "grad_norm": 0.005775100085884333, "learning_rate": 1.7851156717515804e-07, "loss": 0.0002, "step": 225690 }, { "epoch": 3.693037715781723, "grad_norm": 0.1047673225402832, "learning_rate": 1.7832254686528828e-07, "loss": 0.0005, "step": 225700 }, { "epoch": 3.6932013417327987, "grad_norm": 0.03078736737370491, "learning_rate": 1.7813362486531982e-07, "loss": 0.0001, "step": 225710 }, { "epoch": 3.6933649676838747, "grad_norm": 0.051164984703063965, "learning_rate": 1.779448011791063e-07, "loss": 0.0003, "step": 225720 }, { "epoch": 3.6935285936349507, "grad_norm": 0.01344112865626812, "learning_rate": 1.7775607581049737e-07, "loss": 0.0002, "step": 225730 }, { "epoch": 3.6936922195860262, "grad_norm": 0.01892692595720291, "learning_rate": 1.7756744876334165e-07, "loss": 0.0002, "step": 225740 }, { "epoch": 3.6938558455371022, "grad_norm": 0.0039270310662686825, "learning_rate": 1.7737892004148327e-07, "loss": 0.0004, "step": 225750 }, { "epoch": 3.694019471488178, "grad_norm": 0.03689466789364815, "learning_rate": 1.771904896487675e-07, "loss": 0.0001, "step": 225760 }, { "epoch": 3.6941830974392538, "grad_norm": 0.018329178914427757, "learning_rate": 1.770021575890357e-07, "loss": 0.0002, "step": 225770 }, { "epoch": 3.6943467233903298, "grad_norm": 0.015313392505049706, "learning_rate": 1.7681392386612872e-07, "loss": 0.0003, "step": 225780 }, { "epoch": 3.6945103493414058, "grad_norm": 0.046510372310876846, "learning_rate": 1.7662578848388346e-07, "loss": 0.0002, "step": 225790 }, { "epoch": 3.6946739752924813, "grad_norm": 0.016168218106031418, "learning_rate": 1.7643775144613628e-07, "loss": 0.0002, "step": 225800 }, { "epoch": 3.6948376012435573, "grad_norm": 0.02697771042585373, "learning_rate": 1.7624981275672082e-07, "loss": 0.0003, "step": 225810 }, { "epoch": 3.6950012271946333, "grad_norm": 0.004399414639919996, "learning_rate": 1.760619724194701e-07, "loss": 0.0001, "step": 225820 }, { "epoch": 3.695164853145709, "grad_norm": 0.006670479662716389, "learning_rate": 1.7587423043821217e-07, "loss": 0.0001, "step": 225830 }, { "epoch": 3.695328479096785, "grad_norm": 0.020169652998447418, "learning_rate": 1.756865868167762e-07, "loss": 0.0002, "step": 225840 }, { "epoch": 3.695492105047861, "grad_norm": 0.01928306370973587, "learning_rate": 1.75499041558988e-07, "loss": 0.0003, "step": 225850 }, { "epoch": 3.6956557309989364, "grad_norm": 0.059027910232543945, "learning_rate": 1.753115946686712e-07, "loss": 0.0006, "step": 225860 }, { "epoch": 3.6958193569500124, "grad_norm": 0.00992906279861927, "learning_rate": 1.7512424614964774e-07, "loss": 0.0001, "step": 225870 }, { "epoch": 3.695982982901088, "grad_norm": 0.040929023176431656, "learning_rate": 1.7493699600573734e-07, "loss": 0.0002, "step": 225880 }, { "epoch": 3.696146608852164, "grad_norm": 0.016435518860816956, "learning_rate": 1.7474984424075801e-07, "loss": 0.0002, "step": 225890 }, { "epoch": 3.69631023480324, "grad_norm": 0.0019449824467301369, "learning_rate": 1.7456279085852624e-07, "loss": 0.0002, "step": 225900 }, { "epoch": 3.6964738607543155, "grad_norm": 0.0018824724247679114, "learning_rate": 1.7437583586285445e-07, "loss": 0.0002, "step": 225910 }, { "epoch": 3.6966374867053915, "grad_norm": 0.10743264853954315, "learning_rate": 1.741889792575552e-07, "loss": 0.0002, "step": 225920 }, { "epoch": 3.696801112656467, "grad_norm": 0.003244529478251934, "learning_rate": 1.7400222104643872e-07, "loss": 0.0001, "step": 225930 }, { "epoch": 3.696964738607543, "grad_norm": 0.004046555142849684, "learning_rate": 1.7381556123331366e-07, "loss": 0.0003, "step": 225940 }, { "epoch": 3.697128364558619, "grad_norm": 0.1059577539563179, "learning_rate": 1.7362899982198366e-07, "loss": 0.0003, "step": 225950 }, { "epoch": 3.6972919905096946, "grad_norm": 0.0017672352259978652, "learning_rate": 1.7344253681625345e-07, "loss": 0.0004, "step": 225960 }, { "epoch": 3.6974556164607706, "grad_norm": 0.010917318053543568, "learning_rate": 1.7325617221992553e-07, "loss": 0.0002, "step": 225970 }, { "epoch": 3.6976192424118466, "grad_norm": 0.013855230063199997, "learning_rate": 1.7306990603679964e-07, "loss": 0.0004, "step": 225980 }, { "epoch": 3.697782868362922, "grad_norm": 0.0003502371837384999, "learning_rate": 1.7288373827067272e-07, "loss": 0.0001, "step": 225990 }, { "epoch": 3.697946494313998, "grad_norm": 0.011984609067440033, "learning_rate": 1.7269766892534124e-07, "loss": 0.0003, "step": 226000 }, { "epoch": 3.698110120265074, "grad_norm": 0.04566227272152901, "learning_rate": 1.7251169800459822e-07, "loss": 0.0001, "step": 226010 }, { "epoch": 3.6982737462161497, "grad_norm": 0.017704477533698082, "learning_rate": 1.723258255122373e-07, "loss": 0.0002, "step": 226020 }, { "epoch": 3.6984373721672257, "grad_norm": 0.0004949375870637596, "learning_rate": 1.7214005145204603e-07, "loss": 0.0002, "step": 226030 }, { "epoch": 3.6986009981183017, "grad_norm": 0.013280256651341915, "learning_rate": 1.7195437582781304e-07, "loss": 0.0002, "step": 226040 }, { "epoch": 3.6987646240693772, "grad_norm": 0.08496445417404175, "learning_rate": 1.7176879864332474e-07, "loss": 0.0003, "step": 226050 }, { "epoch": 3.6989282500204532, "grad_norm": 0.05479265749454498, "learning_rate": 1.715833199023642e-07, "loss": 0.0004, "step": 226060 }, { "epoch": 3.699091875971529, "grad_norm": 0.007651739288121462, "learning_rate": 1.7139793960871343e-07, "loss": 0.0001, "step": 226070 }, { "epoch": 3.6992555019226048, "grad_norm": 0.025194382295012474, "learning_rate": 1.7121265776615214e-07, "loss": 0.0004, "step": 226080 }, { "epoch": 3.6994191278736808, "grad_norm": 0.004732448607683182, "learning_rate": 1.7102747437845735e-07, "loss": 0.0003, "step": 226090 }, { "epoch": 3.6995827538247568, "grad_norm": 0.0010701771825551987, "learning_rate": 1.7084238944940656e-07, "loss": 0.0005, "step": 226100 }, { "epoch": 3.6997463797758323, "grad_norm": 0.08881375193595886, "learning_rate": 1.7065740298277177e-07, "loss": 0.0003, "step": 226110 }, { "epoch": 3.6999100057269083, "grad_norm": 0.010193383321166039, "learning_rate": 1.7047251498232553e-07, "loss": 0.0003, "step": 226120 }, { "epoch": 3.7000736316779843, "grad_norm": 0.012256584130227566, "learning_rate": 1.7028772545183702e-07, "loss": 0.0002, "step": 226130 }, { "epoch": 3.70023725762906, "grad_norm": 0.007705213967710733, "learning_rate": 1.7010303439507492e-07, "loss": 0.0002, "step": 226140 }, { "epoch": 3.700400883580136, "grad_norm": 0.013692844659090042, "learning_rate": 1.699184418158034e-07, "loss": 0.0003, "step": 226150 }, { "epoch": 3.700564509531212, "grad_norm": 0.0007633285713382065, "learning_rate": 1.6973394771778784e-07, "loss": 0.0002, "step": 226160 }, { "epoch": 3.7007281354822874, "grad_norm": 0.0732831135392189, "learning_rate": 1.6954955210478795e-07, "loss": 0.0002, "step": 226170 }, { "epoch": 3.7008917614333634, "grad_norm": 0.005907568149268627, "learning_rate": 1.6936525498056576e-07, "loss": 0.0001, "step": 226180 }, { "epoch": 3.7010553873844394, "grad_norm": 0.00501435250043869, "learning_rate": 1.6918105634887715e-07, "loss": 0.0003, "step": 226190 }, { "epoch": 3.701219013335515, "grad_norm": 0.012050727382302284, "learning_rate": 1.689969562134791e-07, "loss": 0.0002, "step": 226200 }, { "epoch": 3.701382639286591, "grad_norm": 0.019852114841341972, "learning_rate": 1.6881295457812308e-07, "loss": 0.0002, "step": 226210 }, { "epoch": 3.701546265237667, "grad_norm": 0.002334103686735034, "learning_rate": 1.6862905144656328e-07, "loss": 0.0001, "step": 226220 }, { "epoch": 3.7017098911887425, "grad_norm": 0.022825315594673157, "learning_rate": 1.6844524682254725e-07, "loss": 0.0002, "step": 226230 }, { "epoch": 3.7018735171398185, "grad_norm": 0.0006135934963822365, "learning_rate": 1.682615407098248e-07, "loss": 0.0002, "step": 226240 }, { "epoch": 3.7020371430908945, "grad_norm": 0.001062547555193305, "learning_rate": 1.6807793311213848e-07, "loss": 0.0005, "step": 226250 }, { "epoch": 3.70220076904197, "grad_norm": 0.0023954641073942184, "learning_rate": 1.678944240332353e-07, "loss": 0.0002, "step": 226260 }, { "epoch": 3.702364394993046, "grad_norm": 0.0008985106251202524, "learning_rate": 1.6771101347685504e-07, "loss": 0.0001, "step": 226270 }, { "epoch": 3.7025280209441216, "grad_norm": 0.050045039504766464, "learning_rate": 1.6752770144673748e-07, "loss": 0.0004, "step": 226280 }, { "epoch": 3.7026916468951976, "grad_norm": 0.021048860624432564, "learning_rate": 1.6734448794661907e-07, "loss": 0.0001, "step": 226290 }, { "epoch": 3.7028552728462736, "grad_norm": 0.0004364947380963713, "learning_rate": 1.6716137298023793e-07, "loss": 0.0002, "step": 226300 }, { "epoch": 3.703018898797349, "grad_norm": 0.01438947580754757, "learning_rate": 1.6697835655132556e-07, "loss": 0.0004, "step": 226310 }, { "epoch": 3.703182524748425, "grad_norm": 0.007889116182923317, "learning_rate": 1.6679543866361448e-07, "loss": 0.0002, "step": 226320 }, { "epoch": 3.7033461506995007, "grad_norm": 0.04681392014026642, "learning_rate": 1.666126193208334e-07, "loss": 0.0005, "step": 226330 }, { "epoch": 3.7035097766505767, "grad_norm": 0.038455188274383545, "learning_rate": 1.66429898526711e-07, "loss": 0.0002, "step": 226340 }, { "epoch": 3.7036734026016527, "grad_norm": 0.02271806076169014, "learning_rate": 1.662472762849715e-07, "loss": 0.0001, "step": 226350 }, { "epoch": 3.7038370285527282, "grad_norm": 0.053014058619737625, "learning_rate": 1.6606475259933973e-07, "loss": 0.0002, "step": 226360 }, { "epoch": 3.7040006545038042, "grad_norm": 0.019426696002483368, "learning_rate": 1.6588232747353604e-07, "loss": 0.0002, "step": 226370 }, { "epoch": 3.70416428045488, "grad_norm": 0.0028647787403315306, "learning_rate": 1.657000009112808e-07, "loss": 0.0001, "step": 226380 }, { "epoch": 3.7043279064059558, "grad_norm": 0.05303007364273071, "learning_rate": 1.6551777291629044e-07, "loss": 0.0002, "step": 226390 }, { "epoch": 3.7044915323570318, "grad_norm": 0.01994863897562027, "learning_rate": 1.6533564349228204e-07, "loss": 0.0001, "step": 226400 }, { "epoch": 3.7046551583081078, "grad_norm": 0.0037728745955973864, "learning_rate": 1.651536126429676e-07, "loss": 0.0004, "step": 226410 }, { "epoch": 3.7048187842591833, "grad_norm": 0.004827897530049086, "learning_rate": 1.6497168037205912e-07, "loss": 0.0002, "step": 226420 }, { "epoch": 3.7049824102102593, "grad_norm": 0.015992827713489532, "learning_rate": 1.647898466832659e-07, "loss": 0.0004, "step": 226430 }, { "epoch": 3.7051460361613353, "grad_norm": 0.01380086224526167, "learning_rate": 1.646081115802961e-07, "loss": 0.0002, "step": 226440 }, { "epoch": 3.705309662112411, "grad_norm": 0.002609078772366047, "learning_rate": 1.6442647506685395e-07, "loss": 0.0003, "step": 226450 }, { "epoch": 3.705473288063487, "grad_norm": 0.04053870588541031, "learning_rate": 1.642449371466437e-07, "loss": 0.0002, "step": 226460 }, { "epoch": 3.705636914014563, "grad_norm": 0.004774339031428099, "learning_rate": 1.6406349782336627e-07, "loss": 0.0003, "step": 226470 }, { "epoch": 3.7058005399656384, "grad_norm": 0.0010064007947221398, "learning_rate": 1.638821571007221e-07, "loss": 0.0001, "step": 226480 }, { "epoch": 3.7059641659167144, "grad_norm": 0.001158671104349196, "learning_rate": 1.6370091498240702e-07, "loss": 0.0002, "step": 226490 }, { "epoch": 3.7061277918677904, "grad_norm": 0.08455604314804077, "learning_rate": 1.6351977147211707e-07, "loss": 0.0006, "step": 226500 }, { "epoch": 3.706291417818866, "grad_norm": 0.03038167953491211, "learning_rate": 1.6333872657354588e-07, "loss": 0.0003, "step": 226510 }, { "epoch": 3.706455043769942, "grad_norm": 0.016929367557168007, "learning_rate": 1.6315778029038498e-07, "loss": 0.0003, "step": 226520 }, { "epoch": 3.706618669721018, "grad_norm": 0.0007240367704071105, "learning_rate": 1.629769326263231e-07, "loss": 0.0014, "step": 226530 }, { "epoch": 3.7067822956720935, "grad_norm": 0.04273104667663574, "learning_rate": 1.6279618358504777e-07, "loss": 0.0003, "step": 226540 }, { "epoch": 3.7069459216231695, "grad_norm": 0.003067198907956481, "learning_rate": 1.6261553317024447e-07, "loss": 0.0001, "step": 226550 }, { "epoch": 3.7071095475742455, "grad_norm": 0.08035743981599808, "learning_rate": 1.624349813855963e-07, "loss": 0.0007, "step": 226560 }, { "epoch": 3.707273173525321, "grad_norm": 0.00574712548404932, "learning_rate": 1.6225452823478483e-07, "loss": 0.0002, "step": 226570 }, { "epoch": 3.707436799476397, "grad_norm": 0.07786620408296585, "learning_rate": 1.620741737214887e-07, "loss": 0.0006, "step": 226580 }, { "epoch": 3.707600425427473, "grad_norm": 0.0012170778354629874, "learning_rate": 1.6189391784938613e-07, "loss": 0.0002, "step": 226590 }, { "epoch": 3.7077640513785486, "grad_norm": 0.002701834309846163, "learning_rate": 1.6171376062215194e-07, "loss": 0.0002, "step": 226600 }, { "epoch": 3.7079276773296246, "grad_norm": 0.06713485717773438, "learning_rate": 1.615337020434593e-07, "loss": 0.0003, "step": 226610 }, { "epoch": 3.7080913032807006, "grad_norm": 0.0005051522166468203, "learning_rate": 1.613537421169792e-07, "loss": 0.0002, "step": 226620 }, { "epoch": 3.708254929231776, "grad_norm": 0.0026971090119332075, "learning_rate": 1.6117388084638142e-07, "loss": 0.0003, "step": 226630 }, { "epoch": 3.708418555182852, "grad_norm": 0.026333743706345558, "learning_rate": 1.6099411823533307e-07, "loss": 0.0002, "step": 226640 }, { "epoch": 3.7085821811339277, "grad_norm": 0.009565813466906548, "learning_rate": 1.6081445428749898e-07, "loss": 0.0001, "step": 226650 }, { "epoch": 3.7087458070850037, "grad_norm": 0.0031730453483760357, "learning_rate": 1.6063488900654233e-07, "loss": 0.0008, "step": 226660 }, { "epoch": 3.7089094330360797, "grad_norm": 0.04511921852827072, "learning_rate": 1.6045542239612521e-07, "loss": 0.0005, "step": 226670 }, { "epoch": 3.7090730589871552, "grad_norm": 0.043338462710380554, "learning_rate": 1.6027605445990579e-07, "loss": 0.0002, "step": 226680 }, { "epoch": 3.709236684938231, "grad_norm": 0.018954597413539886, "learning_rate": 1.6009678520154172e-07, "loss": 0.0003, "step": 226690 }, { "epoch": 3.7094003108893068, "grad_norm": 0.03073546476662159, "learning_rate": 1.599176146246878e-07, "loss": 0.0001, "step": 226700 }, { "epoch": 3.7095639368403828, "grad_norm": 0.027617692947387695, "learning_rate": 1.5973854273299728e-07, "loss": 0.0001, "step": 226710 }, { "epoch": 3.7097275627914588, "grad_norm": 0.057533375918865204, "learning_rate": 1.595595695301222e-07, "loss": 0.0005, "step": 226720 }, { "epoch": 3.7098911887425343, "grad_norm": 0.003212417010217905, "learning_rate": 1.5938069501970966e-07, "loss": 0.0003, "step": 226730 }, { "epoch": 3.7100548146936103, "grad_norm": 0.016369586810469627, "learning_rate": 1.592019192054084e-07, "loss": 0.0, "step": 226740 }, { "epoch": 3.7102184406446863, "grad_norm": 0.0025975466705858707, "learning_rate": 1.5902324209086327e-07, "loss": 0.0003, "step": 226750 }, { "epoch": 3.710382066595762, "grad_norm": 0.0003990604018326849, "learning_rate": 1.588446636797175e-07, "loss": 0.0002, "step": 226760 }, { "epoch": 3.710545692546838, "grad_norm": 0.07867252826690674, "learning_rate": 1.5866618397561096e-07, "loss": 0.0002, "step": 226770 }, { "epoch": 3.710709318497914, "grad_norm": 0.02365599386394024, "learning_rate": 1.5848780298218458e-07, "loss": 0.0002, "step": 226780 }, { "epoch": 3.7108729444489894, "grad_norm": 0.006167967338114977, "learning_rate": 1.583095207030727e-07, "loss": 0.0002, "step": 226790 }, { "epoch": 3.7110365704000654, "grad_norm": 0.058375343680381775, "learning_rate": 1.5813133714191352e-07, "loss": 0.0007, "step": 226800 }, { "epoch": 3.7112001963511414, "grad_norm": 0.00027498352574184537, "learning_rate": 1.5795325230233748e-07, "loss": 0.0002, "step": 226810 }, { "epoch": 3.711363822302217, "grad_norm": 0.02546597272157669, "learning_rate": 1.577752661879778e-07, "loss": 0.0004, "step": 226820 }, { "epoch": 3.711527448253293, "grad_norm": 0.004183898214250803, "learning_rate": 1.5759737880246096e-07, "loss": 0.0004, "step": 226830 }, { "epoch": 3.711691074204369, "grad_norm": 0.027941260486841202, "learning_rate": 1.5741959014941578e-07, "loss": 0.0009, "step": 226840 }, { "epoch": 3.7118547001554445, "grad_norm": 0.018369080498814583, "learning_rate": 1.5724190023246656e-07, "loss": 0.0001, "step": 226850 }, { "epoch": 3.7120183261065205, "grad_norm": 0.00024235199089162052, "learning_rate": 1.5706430905523708e-07, "loss": 0.0003, "step": 226860 }, { "epoch": 3.7121819520575965, "grad_norm": 0.0613597072660923, "learning_rate": 1.5688681662134664e-07, "loss": 0.0002, "step": 226870 }, { "epoch": 3.712345578008672, "grad_norm": 0.06533525884151459, "learning_rate": 1.5670942293441572e-07, "loss": 0.0002, "step": 226880 }, { "epoch": 3.712509203959748, "grad_norm": 0.009269236586987972, "learning_rate": 1.5653212799806027e-07, "loss": 0.0002, "step": 226890 }, { "epoch": 3.712672829910824, "grad_norm": 0.002292020246386528, "learning_rate": 1.563549318158958e-07, "loss": 0.0, "step": 226900 }, { "epoch": 3.7128364558618996, "grad_norm": 0.007582899183034897, "learning_rate": 1.5617783439153433e-07, "loss": 0.0003, "step": 226910 }, { "epoch": 3.7130000818129756, "grad_norm": 0.015678245574235916, "learning_rate": 1.5600083572858805e-07, "loss": 0.0003, "step": 226920 }, { "epoch": 3.7131637077640516, "grad_norm": 0.05699897184967995, "learning_rate": 1.5582393583066457e-07, "loss": 0.0003, "step": 226930 }, { "epoch": 3.713327333715127, "grad_norm": 0.0015879679704084992, "learning_rate": 1.556471347013716e-07, "loss": 0.0004, "step": 226940 }, { "epoch": 3.713490959666203, "grad_norm": 0.011309964582324028, "learning_rate": 1.5547043234431236e-07, "loss": 0.0002, "step": 226950 }, { "epoch": 3.713654585617279, "grad_norm": 0.0031724385917186737, "learning_rate": 1.552938287630923e-07, "loss": 0.0001, "step": 226960 }, { "epoch": 3.7138182115683547, "grad_norm": 0.0005193018005229533, "learning_rate": 1.551173239613102e-07, "loss": 0.0003, "step": 226970 }, { "epoch": 3.7139818375194307, "grad_norm": 0.002081681741401553, "learning_rate": 1.5494091794256593e-07, "loss": 0.0001, "step": 226980 }, { "epoch": 3.7141454634705067, "grad_norm": 0.027532383799552917, "learning_rate": 1.54764610710455e-07, "loss": 0.0003, "step": 226990 }, { "epoch": 3.714309089421582, "grad_norm": 0.011068534106016159, "learning_rate": 1.545884022685734e-07, "loss": 0.0003, "step": 227000 }, { "epoch": 3.714472715372658, "grad_norm": 0.007651918102055788, "learning_rate": 1.5441229262051328e-07, "loss": 0.0004, "step": 227010 }, { "epoch": 3.714636341323734, "grad_norm": 0.00932006724178791, "learning_rate": 1.542362817698656e-07, "loss": 0.0003, "step": 227020 }, { "epoch": 3.7147999672748098, "grad_norm": 0.00016557981143705547, "learning_rate": 1.540603697202192e-07, "loss": 0.0002, "step": 227030 }, { "epoch": 3.7149635932258858, "grad_norm": 0.015236378647387028, "learning_rate": 1.538845564751601e-07, "loss": 0.0003, "step": 227040 }, { "epoch": 3.7151272191769613, "grad_norm": 0.03635022044181824, "learning_rate": 1.5370884203827318e-07, "loss": 0.0002, "step": 227050 }, { "epoch": 3.7152908451280373, "grad_norm": 0.04057444632053375, "learning_rate": 1.5353322641314228e-07, "loss": 0.0004, "step": 227060 }, { "epoch": 3.7154544710791133, "grad_norm": 0.008399909362196922, "learning_rate": 1.5335770960334617e-07, "loss": 0.0002, "step": 227070 }, { "epoch": 3.715618097030189, "grad_norm": 0.0003157334867864847, "learning_rate": 1.5318229161246477e-07, "loss": 0.0001, "step": 227080 }, { "epoch": 3.715781722981265, "grad_norm": 0.036627188324928284, "learning_rate": 1.5300697244407414e-07, "loss": 0.0002, "step": 227090 }, { "epoch": 3.7159453489323404, "grad_norm": 0.09146486967802048, "learning_rate": 1.5283175210174973e-07, "loss": 0.0003, "step": 227100 }, { "epoch": 3.7161089748834164, "grad_norm": 0.004137728363275528, "learning_rate": 1.526566305890631e-07, "loss": 0.0001, "step": 227110 }, { "epoch": 3.7162726008344924, "grad_norm": 0.028242778033018112, "learning_rate": 1.5248160790958533e-07, "loss": 0.0005, "step": 227120 }, { "epoch": 3.716436226785568, "grad_norm": 0.03348454460501671, "learning_rate": 1.5230668406688464e-07, "loss": 0.0002, "step": 227130 }, { "epoch": 3.716599852736644, "grad_norm": 0.007594242691993713, "learning_rate": 1.5213185906452876e-07, "loss": 0.0003, "step": 227140 }, { "epoch": 3.71676347868772, "grad_norm": 0.016170844435691833, "learning_rate": 1.5195713290608037e-07, "loss": 0.0002, "step": 227150 }, { "epoch": 3.7169271046387955, "grad_norm": 0.002724084537476301, "learning_rate": 1.5178250559510333e-07, "loss": 0.0001, "step": 227160 }, { "epoch": 3.7170907305898715, "grad_norm": 0.007143994327634573, "learning_rate": 1.516079771351575e-07, "loss": 0.0001, "step": 227170 }, { "epoch": 3.7172543565409475, "grad_norm": 0.00012672119191847742, "learning_rate": 1.5143354752980233e-07, "loss": 0.0002, "step": 227180 }, { "epoch": 3.717417982492023, "grad_norm": 0.09239823371171951, "learning_rate": 1.512592167825927e-07, "loss": 0.0003, "step": 227190 }, { "epoch": 3.717581608443099, "grad_norm": 0.0014173374511301517, "learning_rate": 1.5108498489708412e-07, "loss": 0.0002, "step": 227200 }, { "epoch": 3.717745234394175, "grad_norm": 0.046063438057899475, "learning_rate": 1.5091085187682874e-07, "loss": 0.0001, "step": 227210 }, { "epoch": 3.7179088603452506, "grad_norm": 0.030573250725865364, "learning_rate": 1.5073681772537763e-07, "loss": 0.0002, "step": 227220 }, { "epoch": 3.7180724862963266, "grad_norm": 0.05257280543446541, "learning_rate": 1.5056288244627793e-07, "loss": 0.0002, "step": 227230 }, { "epoch": 3.7182361122474026, "grad_norm": 0.004603130742907524, "learning_rate": 1.5038904604307736e-07, "loss": 0.0002, "step": 227240 }, { "epoch": 3.718399738198478, "grad_norm": 0.0017088456079363823, "learning_rate": 1.502153085193192e-07, "loss": 0.0002, "step": 227250 }, { "epoch": 3.718563364149554, "grad_norm": 0.014903116039931774, "learning_rate": 1.500416698785473e-07, "loss": 0.0002, "step": 227260 }, { "epoch": 3.71872699010063, "grad_norm": 0.004406645894050598, "learning_rate": 1.498681301242999e-07, "loss": 0.0003, "step": 227270 }, { "epoch": 3.7188906160517057, "grad_norm": 0.025977537035942078, "learning_rate": 1.49694689260117e-07, "loss": 0.0002, "step": 227280 }, { "epoch": 3.7190542420027817, "grad_norm": 0.026754293590784073, "learning_rate": 1.4952134728953404e-07, "loss": 0.0004, "step": 227290 }, { "epoch": 3.7192178679538577, "grad_norm": 0.00019570083532016724, "learning_rate": 1.4934810421608602e-07, "loss": 0.0002, "step": 227300 }, { "epoch": 3.7193814939049332, "grad_norm": 0.001044111093506217, "learning_rate": 1.4917496004330456e-07, "loss": 0.0008, "step": 227310 }, { "epoch": 3.719545119856009, "grad_norm": 0.0002177652932005003, "learning_rate": 1.4900191477472014e-07, "loss": 0.0002, "step": 227320 }, { "epoch": 3.719708745807085, "grad_norm": 0.027601638808846474, "learning_rate": 1.4882896841386051e-07, "loss": 0.0001, "step": 227330 }, { "epoch": 3.7198723717581608, "grad_norm": 0.11476419866085052, "learning_rate": 1.486561209642534e-07, "loss": 0.0002, "step": 227340 }, { "epoch": 3.7200359977092368, "grad_norm": 0.013898489065468311, "learning_rate": 1.4848337242942156e-07, "loss": 0.0003, "step": 227350 }, { "epoch": 3.7201996236603128, "grad_norm": 0.0005014913040213287, "learning_rate": 1.483107228128877e-07, "loss": 0.0004, "step": 227360 }, { "epoch": 3.7203632496113883, "grad_norm": 0.024276647716760635, "learning_rate": 1.481381721181724e-07, "loss": 0.0003, "step": 227370 }, { "epoch": 3.7205268755624643, "grad_norm": 0.01433608029037714, "learning_rate": 1.479657203487933e-07, "loss": 0.0002, "step": 227380 }, { "epoch": 3.7206905015135403, "grad_norm": 0.055789392441511154, "learning_rate": 1.4779336750826655e-07, "loss": 0.0002, "step": 227390 }, { "epoch": 3.720854127464616, "grad_norm": 0.045513369143009186, "learning_rate": 1.47621113600106e-07, "loss": 0.0011, "step": 227400 }, { "epoch": 3.721017753415692, "grad_norm": 0.005804733373224735, "learning_rate": 1.4744895862782439e-07, "loss": 0.0003, "step": 227410 }, { "epoch": 3.7211813793667674, "grad_norm": 0.049946025013923645, "learning_rate": 1.4727690259493222e-07, "loss": 0.0003, "step": 227420 }, { "epoch": 3.7213450053178434, "grad_norm": 0.019657397642731667, "learning_rate": 1.4710494550493614e-07, "loss": 0.0002, "step": 227430 }, { "epoch": 3.7215086312689194, "grad_norm": 0.022485477849841118, "learning_rate": 1.4693308736134336e-07, "loss": 0.0002, "step": 227440 }, { "epoch": 3.721672257219995, "grad_norm": 0.07274895161390305, "learning_rate": 1.4676132816765664e-07, "loss": 0.0001, "step": 227450 }, { "epoch": 3.721835883171071, "grad_norm": 0.0038229709025472403, "learning_rate": 1.4658966792738038e-07, "loss": 0.0001, "step": 227460 }, { "epoch": 3.7219995091221465, "grad_norm": 0.11612258106470108, "learning_rate": 1.4641810664401235e-07, "loss": 0.0002, "step": 227470 }, { "epoch": 3.7221631350732225, "grad_norm": 0.05190582573413849, "learning_rate": 1.462466443210514e-07, "loss": 0.0005, "step": 227480 }, { "epoch": 3.7223267610242985, "grad_norm": 0.02100791037082672, "learning_rate": 1.4607528096199252e-07, "loss": 0.0002, "step": 227490 }, { "epoch": 3.722490386975374, "grad_norm": 0.02487950026988983, "learning_rate": 1.459040165703318e-07, "loss": 0.0004, "step": 227500 }, { "epoch": 3.72265401292645, "grad_norm": 0.012800473719835281, "learning_rate": 1.4573285114955927e-07, "loss": 0.0002, "step": 227510 }, { "epoch": 3.722817638877526, "grad_norm": 0.0006376555538736284, "learning_rate": 1.4556178470316652e-07, "loss": 0.0002, "step": 227520 }, { "epoch": 3.7229812648286016, "grad_norm": 0.029848594218492508, "learning_rate": 1.4539081723463856e-07, "loss": 0.0004, "step": 227530 }, { "epoch": 3.7231448907796776, "grad_norm": 0.02785961888730526, "learning_rate": 1.452199487474648e-07, "loss": 0.0002, "step": 227540 }, { "epoch": 3.7233085167307536, "grad_norm": 0.03903603553771973, "learning_rate": 1.4504917924512642e-07, "loss": 0.0003, "step": 227550 }, { "epoch": 3.723472142681829, "grad_norm": 0.0070230416022241116, "learning_rate": 1.448785087311072e-07, "loss": 0.0005, "step": 227560 }, { "epoch": 3.723635768632905, "grad_norm": 0.02121744304895401, "learning_rate": 1.4470793720888498e-07, "loss": 0.0004, "step": 227570 }, { "epoch": 3.723799394583981, "grad_norm": 0.057843051850795746, "learning_rate": 1.4453746468193975e-07, "loss": 0.0005, "step": 227580 }, { "epoch": 3.7239630205350567, "grad_norm": 0.008910465054214, "learning_rate": 1.4436709115374537e-07, "loss": 0.0001, "step": 227590 }, { "epoch": 3.7241266464861327, "grad_norm": 0.009588100016117096, "learning_rate": 1.4419681662777685e-07, "loss": 0.0002, "step": 227600 }, { "epoch": 3.7242902724372087, "grad_norm": 0.01915670931339264, "learning_rate": 1.440266411075053e-07, "loss": 0.0002, "step": 227610 }, { "epoch": 3.7244538983882842, "grad_norm": 0.0012918781721964478, "learning_rate": 1.4385656459640074e-07, "loss": 0.0003, "step": 227620 }, { "epoch": 3.72461752433936, "grad_norm": 0.028734976425766945, "learning_rate": 1.4368658709793092e-07, "loss": 0.0002, "step": 227630 }, { "epoch": 3.724781150290436, "grad_norm": 0.006262549199163914, "learning_rate": 1.4351670861556145e-07, "loss": 0.0003, "step": 227640 }, { "epoch": 3.7249447762415118, "grad_norm": 0.04338809847831726, "learning_rate": 1.4334692915275617e-07, "loss": 0.0002, "step": 227650 }, { "epoch": 3.7251084021925878, "grad_norm": 0.007513225078582764, "learning_rate": 1.4317724871297623e-07, "loss": 0.0001, "step": 227660 }, { "epoch": 3.7252720281436638, "grad_norm": 0.047767288982868195, "learning_rate": 1.4300766729968162e-07, "loss": 0.0002, "step": 227670 }, { "epoch": 3.7254356540947393, "grad_norm": 0.001120103057473898, "learning_rate": 1.4283818491633017e-07, "loss": 0.0003, "step": 227680 }, { "epoch": 3.7255992800458153, "grad_norm": 0.017063267529010773, "learning_rate": 1.4266880156637685e-07, "loss": 0.0001, "step": 227690 }, { "epoch": 3.7257629059968913, "grad_norm": 0.04719940200448036, "learning_rate": 1.4249951725327615e-07, "loss": 0.0002, "step": 227700 }, { "epoch": 3.725926531947967, "grad_norm": 0.03068070113658905, "learning_rate": 1.4233033198047863e-07, "loss": 0.0004, "step": 227710 }, { "epoch": 3.726090157899043, "grad_norm": 0.01262261625379324, "learning_rate": 1.4216124575143486e-07, "loss": 0.0003, "step": 227720 }, { "epoch": 3.726253783850119, "grad_norm": 0.009172170422971249, "learning_rate": 1.4199225856959154e-07, "loss": 0.0002, "step": 227730 }, { "epoch": 3.7264174098011944, "grad_norm": 0.002816695487126708, "learning_rate": 1.4182337043839477e-07, "loss": 0.0002, "step": 227740 }, { "epoch": 3.7265810357522704, "grad_norm": 0.02828122489154339, "learning_rate": 1.416545813612874e-07, "loss": 0.0003, "step": 227750 }, { "epoch": 3.7267446617033464, "grad_norm": 0.012791590765118599, "learning_rate": 1.4148589134171165e-07, "loss": 0.0003, "step": 227760 }, { "epoch": 3.726908287654422, "grad_norm": 0.026009781286120415, "learning_rate": 1.4131730038310588e-07, "loss": 0.0004, "step": 227770 }, { "epoch": 3.727071913605498, "grad_norm": 0.028016412630677223, "learning_rate": 1.4114880848890843e-07, "loss": 0.0004, "step": 227780 }, { "epoch": 3.727235539556574, "grad_norm": 0.08489461243152618, "learning_rate": 1.4098041566255495e-07, "loss": 0.0005, "step": 227790 }, { "epoch": 3.7273991655076495, "grad_norm": 0.04255087673664093, "learning_rate": 1.4081212190747816e-07, "loss": 0.0003, "step": 227800 }, { "epoch": 3.7275627914587255, "grad_norm": 0.040854644030332565, "learning_rate": 1.4064392722710928e-07, "loss": 0.0001, "step": 227810 }, { "epoch": 3.727726417409801, "grad_norm": 0.014251804910600185, "learning_rate": 1.4047583162487832e-07, "loss": 0.0003, "step": 227820 }, { "epoch": 3.727890043360877, "grad_norm": 0.033618707209825516, "learning_rate": 1.4030783510421197e-07, "loss": 0.0003, "step": 227830 }, { "epoch": 3.728053669311953, "grad_norm": 0.02594420127570629, "learning_rate": 1.401399376685364e-07, "loss": 0.0002, "step": 227840 }, { "epoch": 3.7282172952630286, "grad_norm": 0.022338692098855972, "learning_rate": 1.3997213932127386e-07, "loss": 0.0002, "step": 227850 }, { "epoch": 3.7283809212141046, "grad_norm": 0.015987038612365723, "learning_rate": 1.3980444006584605e-07, "loss": 0.0002, "step": 227860 }, { "epoch": 3.72854454716518, "grad_norm": 0.00039611535612493753, "learning_rate": 1.396368399056719e-07, "loss": 0.0003, "step": 227870 }, { "epoch": 3.728708173116256, "grad_norm": 0.01448910217732191, "learning_rate": 1.3946933884417034e-07, "loss": 0.0003, "step": 227880 }, { "epoch": 3.728871799067332, "grad_norm": 0.01734807714819908, "learning_rate": 1.3930193688475423e-07, "loss": 0.0001, "step": 227890 }, { "epoch": 3.7290354250184077, "grad_norm": 0.019177673384547234, "learning_rate": 1.3913463403083748e-07, "loss": 0.0005, "step": 227900 }, { "epoch": 3.7291990509694837, "grad_norm": 0.000974956143181771, "learning_rate": 1.3896743028583182e-07, "loss": 0.0002, "step": 227910 }, { "epoch": 3.7293626769205597, "grad_norm": 0.001745595713146031, "learning_rate": 1.3880032565314672e-07, "loss": 0.0003, "step": 227920 }, { "epoch": 3.7295263028716352, "grad_norm": 0.0011787808034569025, "learning_rate": 1.386333201361878e-07, "loss": 0.0001, "step": 227930 }, { "epoch": 3.729689928822711, "grad_norm": 0.14988024532794952, "learning_rate": 1.3846641373836122e-07, "loss": 0.0003, "step": 227940 }, { "epoch": 3.729853554773787, "grad_norm": 0.002898303559049964, "learning_rate": 1.3829960646307039e-07, "loss": 0.0003, "step": 227950 }, { "epoch": 3.7300171807248628, "grad_norm": 0.04067634791135788, "learning_rate": 1.381328983137159e-07, "loss": 0.0001, "step": 227960 }, { "epoch": 3.7301808066759388, "grad_norm": 0.020949382334947586, "learning_rate": 1.3796628929369615e-07, "loss": 0.0002, "step": 227970 }, { "epoch": 3.7303444326270148, "grad_norm": 0.03102518618106842, "learning_rate": 1.37799779406409e-07, "loss": 0.0003, "step": 227980 }, { "epoch": 3.7305080585780903, "grad_norm": 0.00010167326399823651, "learning_rate": 1.376333686552489e-07, "loss": 0.0003, "step": 227990 }, { "epoch": 3.7306716845291663, "grad_norm": 0.018866131082177162, "learning_rate": 1.3746705704360985e-07, "loss": 0.0003, "step": 228000 }, { "epoch": 3.7308353104802423, "grad_norm": 0.01078523974865675, "learning_rate": 1.3730084457488135e-07, "loss": 0.0002, "step": 228010 }, { "epoch": 3.730998936431318, "grad_norm": 0.028644109144806862, "learning_rate": 1.3713473125245348e-07, "loss": 0.0002, "step": 228020 }, { "epoch": 3.731162562382394, "grad_norm": 0.022414853796362877, "learning_rate": 1.3696871707971238e-07, "loss": 0.0002, "step": 228030 }, { "epoch": 3.73132618833347, "grad_norm": 0.045921601355075836, "learning_rate": 1.368028020600437e-07, "loss": 0.0002, "step": 228040 }, { "epoch": 3.7314898142845454, "grad_norm": 0.024127857759594917, "learning_rate": 1.3663698619682974e-07, "loss": 0.0003, "step": 228050 }, { "epoch": 3.7316534402356214, "grad_norm": 0.006384809967130423, "learning_rate": 1.364712694934517e-07, "loss": 0.0005, "step": 228060 }, { "epoch": 3.7318170661866974, "grad_norm": 0.005318458657711744, "learning_rate": 1.3630565195328683e-07, "loss": 0.0003, "step": 228070 }, { "epoch": 3.731980692137773, "grad_norm": 0.0026460590306669474, "learning_rate": 1.3614013357971523e-07, "loss": 0.0003, "step": 228080 }, { "epoch": 3.732144318088849, "grad_norm": 0.0019654345232993364, "learning_rate": 1.3597471437610864e-07, "loss": 0.0002, "step": 228090 }, { "epoch": 3.732307944039925, "grad_norm": 0.010246035642921925, "learning_rate": 1.3580939434584162e-07, "loss": 0.0003, "step": 228100 }, { "epoch": 3.7324715699910005, "grad_norm": 0.004220716655254364, "learning_rate": 1.356441734922831e-07, "loss": 0.0001, "step": 228110 }, { "epoch": 3.7326351959420765, "grad_norm": 0.01363549567759037, "learning_rate": 1.354790518188037e-07, "loss": 0.0003, "step": 228120 }, { "epoch": 3.7327988218931525, "grad_norm": 0.028982941061258316, "learning_rate": 1.3531402932876913e-07, "loss": 0.0005, "step": 228130 }, { "epoch": 3.732962447844228, "grad_norm": 0.029516998678445816, "learning_rate": 1.3514910602554442e-07, "loss": 0.0002, "step": 228140 }, { "epoch": 3.733126073795304, "grad_norm": 0.00022471227566711605, "learning_rate": 1.3498428191249136e-07, "loss": 0.0001, "step": 228150 }, { "epoch": 3.73328969974638, "grad_norm": 0.003626833902671933, "learning_rate": 1.3481955699297221e-07, "loss": 0.0001, "step": 228160 }, { "epoch": 3.7334533256974556, "grad_norm": 0.03455125540494919, "learning_rate": 1.346549312703438e-07, "loss": 0.0002, "step": 228170 }, { "epoch": 3.7336169516485316, "grad_norm": 0.03113538585603237, "learning_rate": 1.3449040474796394e-07, "loss": 0.0003, "step": 228180 }, { "epoch": 3.7337805775996076, "grad_norm": 0.024149859324097633, "learning_rate": 1.3432597742918607e-07, "loss": 0.0004, "step": 228190 }, { "epoch": 3.733944203550683, "grad_norm": 0.00992712751030922, "learning_rate": 1.3416164931736475e-07, "loss": 0.0007, "step": 228200 }, { "epoch": 3.734107829501759, "grad_norm": 0.036808889359235764, "learning_rate": 1.3399742041584785e-07, "loss": 0.0005, "step": 228210 }, { "epoch": 3.7342714554528347, "grad_norm": 0.01728750579059124, "learning_rate": 1.3383329072798657e-07, "loss": 0.0004, "step": 228220 }, { "epoch": 3.7344350814039107, "grad_norm": 0.03605188801884651, "learning_rate": 1.3366926025712434e-07, "loss": 0.0004, "step": 228230 }, { "epoch": 3.7345987073549862, "grad_norm": 0.018420889973640442, "learning_rate": 1.335053290066085e-07, "loss": 0.0003, "step": 228240 }, { "epoch": 3.734762333306062, "grad_norm": 0.0009355013025924563, "learning_rate": 1.3334149697977916e-07, "loss": 0.0002, "step": 228250 }, { "epoch": 3.734925959257138, "grad_norm": 0.018382640555500984, "learning_rate": 1.3317776417997864e-07, "loss": 0.0004, "step": 228260 }, { "epoch": 3.7350895852082138, "grad_norm": 0.019095297902822495, "learning_rate": 1.3301413061054426e-07, "loss": 0.0004, "step": 228270 }, { "epoch": 3.7352532111592898, "grad_norm": 0.0002564009337220341, "learning_rate": 1.3285059627481224e-07, "loss": 0.0001, "step": 228280 }, { "epoch": 3.7354168371103658, "grad_norm": 0.10888082534074783, "learning_rate": 1.3268716117611713e-07, "loss": 0.0003, "step": 228290 }, { "epoch": 3.7355804630614413, "grad_norm": 0.000970742607023567, "learning_rate": 1.3252382531779184e-07, "loss": 0.0002, "step": 228300 }, { "epoch": 3.7357440890125173, "grad_norm": 0.11944998800754547, "learning_rate": 1.3236058870316537e-07, "loss": 0.0004, "step": 228310 }, { "epoch": 3.7359077149635933, "grad_norm": 0.0016750257927924395, "learning_rate": 1.3219745133556727e-07, "loss": 0.0004, "step": 228320 }, { "epoch": 3.736071340914669, "grad_norm": 0.029954873025417328, "learning_rate": 1.3203441321832266e-07, "loss": 0.0003, "step": 228330 }, { "epoch": 3.736234966865745, "grad_norm": 0.06584570556879044, "learning_rate": 1.3187147435475723e-07, "loss": 0.0006, "step": 228340 }, { "epoch": 3.736398592816821, "grad_norm": 0.020862700417637825, "learning_rate": 1.3170863474819161e-07, "loss": 0.0003, "step": 228350 }, { "epoch": 3.7365622187678964, "grad_norm": 0.01888030208647251, "learning_rate": 1.3154589440194653e-07, "loss": 0.0003, "step": 228360 }, { "epoch": 3.7367258447189724, "grad_norm": 0.03614683076739311, "learning_rate": 1.3138325331934044e-07, "loss": 0.0004, "step": 228370 }, { "epoch": 3.7368894706700484, "grad_norm": 0.03820711374282837, "learning_rate": 1.3122071150368898e-07, "loss": 0.0002, "step": 228380 }, { "epoch": 3.737053096621124, "grad_norm": 0.04231440648436546, "learning_rate": 1.3105826895830677e-07, "loss": 0.0003, "step": 228390 }, { "epoch": 3.7372167225722, "grad_norm": 0.025893043726682663, "learning_rate": 1.3089592568650499e-07, "loss": 0.0003, "step": 228400 }, { "epoch": 3.737380348523276, "grad_norm": 0.0018289396539330482, "learning_rate": 1.3073368169159496e-07, "loss": 0.0005, "step": 228410 }, { "epoch": 3.7375439744743515, "grad_norm": 0.010030550882220268, "learning_rate": 1.3057153697688397e-07, "loss": 0.0002, "step": 228420 }, { "epoch": 3.7377076004254275, "grad_norm": 0.0636286810040474, "learning_rate": 1.3040949154567773e-07, "loss": 0.0002, "step": 228430 }, { "epoch": 3.7378712263765035, "grad_norm": 0.008020684123039246, "learning_rate": 1.3024754540128025e-07, "loss": 0.0001, "step": 228440 }, { "epoch": 3.738034852327579, "grad_norm": 0.02821088396012783, "learning_rate": 1.3008569854699393e-07, "loss": 0.0001, "step": 228450 }, { "epoch": 3.738198478278655, "grad_norm": 0.01665656641125679, "learning_rate": 1.2992395098611888e-07, "loss": 0.0001, "step": 228460 }, { "epoch": 3.738362104229731, "grad_norm": 0.008223404176533222, "learning_rate": 1.2976230272195244e-07, "loss": 0.0003, "step": 228470 }, { "epoch": 3.7385257301808066, "grad_norm": 0.031016705557703972, "learning_rate": 1.2960075375779035e-07, "loss": 0.0004, "step": 228480 }, { "epoch": 3.7386893561318826, "grad_norm": 0.024902060627937317, "learning_rate": 1.294393040969272e-07, "loss": 0.0002, "step": 228490 }, { "epoch": 3.7388529820829586, "grad_norm": 0.0018631648272275925, "learning_rate": 1.2927795374265418e-07, "loss": 0.0005, "step": 228500 }, { "epoch": 3.739016608034034, "grad_norm": 0.01934938319027424, "learning_rate": 1.291167026982615e-07, "loss": 0.0002, "step": 228510 }, { "epoch": 3.73918023398511, "grad_norm": 0.004517135675996542, "learning_rate": 1.2895555096703593e-07, "loss": 0.0003, "step": 228520 }, { "epoch": 3.739343859936186, "grad_norm": 0.005869982298463583, "learning_rate": 1.287944985522649e-07, "loss": 0.0003, "step": 228530 }, { "epoch": 3.7395074858872617, "grad_norm": 0.00017048230802174658, "learning_rate": 1.2863354545723072e-07, "loss": 0.0001, "step": 228540 }, { "epoch": 3.7396711118383377, "grad_norm": 0.0029040800873190165, "learning_rate": 1.284726916852158e-07, "loss": 0.0002, "step": 228550 }, { "epoch": 3.7398347377894137, "grad_norm": 0.05888636037707329, "learning_rate": 1.2831193723949976e-07, "loss": 0.0003, "step": 228560 }, { "epoch": 3.739998363740489, "grad_norm": 0.00736662745475769, "learning_rate": 1.2815128212335937e-07, "loss": 0.0003, "step": 228570 }, { "epoch": 3.740161989691565, "grad_norm": 0.00374852423556149, "learning_rate": 1.2799072634007204e-07, "loss": 0.0003, "step": 228580 }, { "epoch": 3.7403256156426408, "grad_norm": 0.005735885351896286, "learning_rate": 1.278302698929096e-07, "loss": 0.0002, "step": 228590 }, { "epoch": 3.7404892415937168, "grad_norm": 0.05346367508172989, "learning_rate": 1.276699127851444e-07, "loss": 0.0003, "step": 228600 }, { "epoch": 3.7406528675447928, "grad_norm": 0.0009102777694351971, "learning_rate": 1.275096550200461e-07, "loss": 0.0003, "step": 228610 }, { "epoch": 3.7408164934958683, "grad_norm": 0.00014872400788590312, "learning_rate": 1.273494966008826e-07, "loss": 0.0001, "step": 228620 }, { "epoch": 3.7409801194469443, "grad_norm": 0.014071757905185223, "learning_rate": 1.2718943753091796e-07, "loss": 0.0001, "step": 228630 }, { "epoch": 3.74114374539802, "grad_norm": 0.0035967964213341475, "learning_rate": 1.2702947781341622e-07, "loss": 0.0003, "step": 228640 }, { "epoch": 3.741307371349096, "grad_norm": 0.0038529690355062485, "learning_rate": 1.268696174516404e-07, "loss": 0.0004, "step": 228650 }, { "epoch": 3.741470997300172, "grad_norm": 0.0014823023229837418, "learning_rate": 1.2670985644884782e-07, "loss": 0.0002, "step": 228660 }, { "epoch": 3.7416346232512474, "grad_norm": 0.011446170508861542, "learning_rate": 1.26550194808297e-07, "loss": 0.0006, "step": 228670 }, { "epoch": 3.7417982492023234, "grad_norm": 0.002321208594366908, "learning_rate": 1.2639063253324202e-07, "loss": 0.0003, "step": 228680 }, { "epoch": 3.7419618751533994, "grad_norm": 0.030209289863705635, "learning_rate": 1.2623116962693804e-07, "loss": 0.0002, "step": 228690 }, { "epoch": 3.742125501104475, "grad_norm": 0.008149335160851479, "learning_rate": 1.2607180609263526e-07, "loss": 0.0002, "step": 228700 }, { "epoch": 3.742289127055551, "grad_norm": 0.0010939171770587564, "learning_rate": 1.2591254193358383e-07, "loss": 0.0003, "step": 228710 }, { "epoch": 3.742452753006627, "grad_norm": 0.02942955493927002, "learning_rate": 1.257533771530295e-07, "loss": 0.0002, "step": 228720 }, { "epoch": 3.7426163789577025, "grad_norm": 0.06111418083310127, "learning_rate": 1.255943117542191e-07, "loss": 0.0002, "step": 228730 }, { "epoch": 3.7427800049087785, "grad_norm": 0.026048943400382996, "learning_rate": 1.254353457403945e-07, "loss": 0.0002, "step": 228740 }, { "epoch": 3.7429436308598545, "grad_norm": 0.08590498566627502, "learning_rate": 1.2527647911479812e-07, "loss": 0.0006, "step": 228750 }, { "epoch": 3.74310725681093, "grad_norm": 0.048844192177057266, "learning_rate": 1.2511771188066845e-07, "loss": 0.0002, "step": 228760 }, { "epoch": 3.743270882762006, "grad_norm": 0.001737757003866136, "learning_rate": 1.2495904404124238e-07, "loss": 0.0002, "step": 228770 }, { "epoch": 3.743434508713082, "grad_norm": 0.03982735052704811, "learning_rate": 1.248004755997556e-07, "loss": 0.0002, "step": 228780 }, { "epoch": 3.7435981346641576, "grad_norm": 0.0007199293468147516, "learning_rate": 1.2464200655944115e-07, "loss": 0.0003, "step": 228790 }, { "epoch": 3.7437617606152336, "grad_norm": 0.0003145617956761271, "learning_rate": 1.2448363692352915e-07, "loss": 0.0005, "step": 228800 }, { "epoch": 3.7439253865663096, "grad_norm": 0.007215557619929314, "learning_rate": 1.243253666952493e-07, "loss": 0.0003, "step": 228810 }, { "epoch": 3.744089012517385, "grad_norm": 0.01150728389620781, "learning_rate": 1.24167195877829e-07, "loss": 0.0002, "step": 228820 }, { "epoch": 3.744252638468461, "grad_norm": 0.0049144551157951355, "learning_rate": 1.240091244744934e-07, "loss": 0.0002, "step": 228830 }, { "epoch": 3.744416264419537, "grad_norm": 8.872078615240753e-05, "learning_rate": 1.238511524884639e-07, "loss": 0.0002, "step": 228840 }, { "epoch": 3.7445798903706127, "grad_norm": 0.031291063874959946, "learning_rate": 1.2369327992296286e-07, "loss": 0.0002, "step": 228850 }, { "epoch": 3.7447435163216887, "grad_norm": 0.01436540950089693, "learning_rate": 1.2353550678120828e-07, "loss": 0.0002, "step": 228860 }, { "epoch": 3.7449071422727647, "grad_norm": 0.0439780093729496, "learning_rate": 1.2337783306641814e-07, "loss": 0.0002, "step": 228870 }, { "epoch": 3.74507076822384, "grad_norm": 0.07003703713417053, "learning_rate": 1.2322025878180598e-07, "loss": 0.0002, "step": 228880 }, { "epoch": 3.745234394174916, "grad_norm": 0.016734911128878593, "learning_rate": 1.2306278393058534e-07, "loss": 0.0003, "step": 228890 }, { "epoch": 3.745398020125992, "grad_norm": 0.041540976613759995, "learning_rate": 1.22905408515967e-07, "loss": 0.0003, "step": 228900 }, { "epoch": 3.7455616460770678, "grad_norm": 0.01933755725622177, "learning_rate": 1.2274813254115946e-07, "loss": 0.0004, "step": 228910 }, { "epoch": 3.7457252720281438, "grad_norm": 0.06801405549049377, "learning_rate": 1.2259095600936966e-07, "loss": 0.0003, "step": 228920 }, { "epoch": 3.7458888979792198, "grad_norm": 0.0004698759876191616, "learning_rate": 1.2243387892380166e-07, "loss": 0.0005, "step": 228930 }, { "epoch": 3.7460525239302953, "grad_norm": 0.008215613663196564, "learning_rate": 1.2227690128765902e-07, "loss": 0.0002, "step": 228940 }, { "epoch": 3.7462161498813713, "grad_norm": 0.0018040611175820231, "learning_rate": 1.221200231041425e-07, "loss": 0.0003, "step": 228950 }, { "epoch": 3.7463797758324473, "grad_norm": 0.17369911074638367, "learning_rate": 1.2196324437644958e-07, "loss": 0.0004, "step": 228960 }, { "epoch": 3.746543401783523, "grad_norm": 0.015888968482613564, "learning_rate": 1.2180656510777767e-07, "loss": 0.0001, "step": 228970 }, { "epoch": 3.746707027734599, "grad_norm": 0.02470017969608307, "learning_rate": 1.2164998530132089e-07, "loss": 0.0002, "step": 228980 }, { "epoch": 3.7468706536856744, "grad_norm": 0.0383036732673645, "learning_rate": 1.214935049602728e-07, "loss": 0.0002, "step": 228990 }, { "epoch": 3.7470342796367504, "grad_norm": 0.010552740655839443, "learning_rate": 1.213371240878225e-07, "loss": 0.0001, "step": 229000 }, { "epoch": 3.7471979055878264, "grad_norm": 0.05801253393292427, "learning_rate": 1.2118084268715912e-07, "loss": 0.0004, "step": 229010 }, { "epoch": 3.747361531538902, "grad_norm": 0.0702643096446991, "learning_rate": 1.21024660761469e-07, "loss": 0.0002, "step": 229020 }, { "epoch": 3.747525157489978, "grad_norm": 0.028549326583743095, "learning_rate": 1.2086857831393683e-07, "loss": 0.0005, "step": 229030 }, { "epoch": 3.7476887834410535, "grad_norm": 0.020094463601708412, "learning_rate": 1.207125953477445e-07, "loss": 0.0001, "step": 229040 }, { "epoch": 3.7478524093921295, "grad_norm": 0.022613734006881714, "learning_rate": 1.2055671186607276e-07, "loss": 0.0004, "step": 229050 }, { "epoch": 3.7480160353432055, "grad_norm": 0.03825066238641739, "learning_rate": 1.2040092787210022e-07, "loss": 0.0002, "step": 229060 }, { "epoch": 3.748179661294281, "grad_norm": 0.08471325039863586, "learning_rate": 1.2024524336900266e-07, "loss": 0.0004, "step": 229070 }, { "epoch": 3.748343287245357, "grad_norm": 0.02670622058212757, "learning_rate": 1.2008965835995424e-07, "loss": 0.0002, "step": 229080 }, { "epoch": 3.748506913196433, "grad_norm": 0.015417011454701424, "learning_rate": 1.1993417284812735e-07, "loss": 0.0001, "step": 229090 }, { "epoch": 3.7486705391475086, "grad_norm": 7.003657083259895e-05, "learning_rate": 1.1977878683669287e-07, "loss": 0.0006, "step": 229100 }, { "epoch": 3.7488341650985846, "grad_norm": 0.010848692618310452, "learning_rate": 1.196235003288182e-07, "loss": 0.0005, "step": 229110 }, { "epoch": 3.7489977910496606, "grad_norm": 0.043180324137210846, "learning_rate": 1.1946831332766974e-07, "loss": 0.0002, "step": 229120 }, { "epoch": 3.749161417000736, "grad_norm": 0.011956169269979, "learning_rate": 1.1931322583641158e-07, "loss": 0.0002, "step": 229130 }, { "epoch": 3.749325042951812, "grad_norm": 0.0010346358176320791, "learning_rate": 1.1915823785820568e-07, "loss": 0.0001, "step": 229140 }, { "epoch": 3.749488668902888, "grad_norm": 0.05125183239579201, "learning_rate": 1.1900334939621284e-07, "loss": 0.0004, "step": 229150 }, { "epoch": 3.7496522948539637, "grad_norm": 0.0419071801006794, "learning_rate": 1.188485604535905e-07, "loss": 0.0002, "step": 229160 }, { "epoch": 3.7498159208050397, "grad_norm": 0.0006306182476691902, "learning_rate": 1.186938710334945e-07, "loss": 0.0002, "step": 229170 }, { "epoch": 3.7499795467561157, "grad_norm": 0.032949697226285934, "learning_rate": 1.1853928113907842e-07, "loss": 0.0003, "step": 229180 }, { "epoch": 3.750143172707191, "grad_norm": 0.04984831437468529, "learning_rate": 1.1838479077349585e-07, "loss": 0.0003, "step": 229190 }, { "epoch": 3.750306798658267, "grad_norm": 0.018120935186743736, "learning_rate": 1.1823039993989538e-07, "loss": 0.0002, "step": 229200 }, { "epoch": 3.750470424609343, "grad_norm": 0.009320844896137714, "learning_rate": 1.180761086414256e-07, "loss": 0.0002, "step": 229210 }, { "epoch": 3.7506340505604188, "grad_norm": 0.006354834884405136, "learning_rate": 1.1792191688123123e-07, "loss": 0.0003, "step": 229220 }, { "epoch": 3.7507976765114948, "grad_norm": 0.00027383994893170893, "learning_rate": 1.1776782466245806e-07, "loss": 0.0004, "step": 229230 }, { "epoch": 3.7509613024625708, "grad_norm": 0.020052751526236534, "learning_rate": 1.1761383198824583e-07, "loss": 0.0002, "step": 229240 }, { "epoch": 3.7511249284136463, "grad_norm": 0.00652963574975729, "learning_rate": 1.174599388617359e-07, "loss": 0.0003, "step": 229250 }, { "epoch": 3.7512885543647223, "grad_norm": 0.16124503314495087, "learning_rate": 1.173061452860641e-07, "loss": 0.001, "step": 229260 }, { "epoch": 3.7514521803157983, "grad_norm": 0.030671698972582817, "learning_rate": 1.1715245126436903e-07, "loss": 0.0002, "step": 229270 }, { "epoch": 3.751615806266874, "grad_norm": 0.003352042054757476, "learning_rate": 1.1699885679978151e-07, "loss": 0.0005, "step": 229280 }, { "epoch": 3.75177943221795, "grad_norm": 0.00030974895344115794, "learning_rate": 1.1684536189543572e-07, "loss": 0.0002, "step": 229290 }, { "epoch": 3.751943058169026, "grad_norm": 0.026140442118048668, "learning_rate": 1.1669196655445858e-07, "loss": 0.0003, "step": 229300 }, { "epoch": 3.7521066841201014, "grad_norm": 0.02392476424574852, "learning_rate": 1.1653867077997982e-07, "loss": 0.0001, "step": 229310 }, { "epoch": 3.7522703100711774, "grad_norm": 0.0397169403731823, "learning_rate": 1.1638547457512417e-07, "loss": 0.0003, "step": 229320 }, { "epoch": 3.7524339360222534, "grad_norm": 0.006706717889755964, "learning_rate": 1.1623237794301633e-07, "loss": 0.0001, "step": 229330 }, { "epoch": 3.752597561973329, "grad_norm": 0.03118951804935932, "learning_rate": 1.1607938088677494e-07, "loss": 0.0001, "step": 229340 }, { "epoch": 3.752761187924405, "grad_norm": 0.000877431477420032, "learning_rate": 1.1592648340952306e-07, "loss": 0.0003, "step": 229350 }, { "epoch": 3.7529248138754805, "grad_norm": 0.027121651917696, "learning_rate": 1.157736855143754e-07, "loss": 0.0004, "step": 229360 }, { "epoch": 3.7530884398265565, "grad_norm": 0.08136075735092163, "learning_rate": 1.1562098720444947e-07, "loss": 0.0001, "step": 229370 }, { "epoch": 3.7532520657776325, "grad_norm": 0.12120132148265839, "learning_rate": 1.1546838848285668e-07, "loss": 0.0001, "step": 229380 }, { "epoch": 3.753415691728708, "grad_norm": 0.011388061568140984, "learning_rate": 1.1531588935270899e-07, "loss": 0.0002, "step": 229390 }, { "epoch": 3.753579317679784, "grad_norm": 0.0031985866371542215, "learning_rate": 1.1516348981711667e-07, "loss": 0.0002, "step": 229400 }, { "epoch": 3.7537429436308596, "grad_norm": 0.03020288608968258, "learning_rate": 1.150111898791867e-07, "loss": 0.0002, "step": 229410 }, { "epoch": 3.7539065695819356, "grad_norm": 0.0010250958148390055, "learning_rate": 1.1485898954202323e-07, "loss": 0.0003, "step": 229420 }, { "epoch": 3.7540701955330116, "grad_norm": 0.04870040342211723, "learning_rate": 1.1470688880873104e-07, "loss": 0.0003, "step": 229430 }, { "epoch": 3.754233821484087, "grad_norm": 0.01919219270348549, "learning_rate": 1.1455488768240985e-07, "loss": 0.0003, "step": 229440 }, { "epoch": 3.754397447435163, "grad_norm": 0.00010551152081461623, "learning_rate": 1.1440298616616052e-07, "loss": 0.0002, "step": 229450 }, { "epoch": 3.754561073386239, "grad_norm": 0.0009702965617179871, "learning_rate": 1.1425118426307835e-07, "loss": 0.0003, "step": 229460 }, { "epoch": 3.7547246993373147, "grad_norm": 0.004283412825316191, "learning_rate": 1.1409948197625975e-07, "loss": 0.0001, "step": 229470 }, { "epoch": 3.7548883252883907, "grad_norm": 0.015976034104824066, "learning_rate": 1.1394787930879725e-07, "loss": 0.0003, "step": 229480 }, { "epoch": 3.7550519512394667, "grad_norm": 0.012423262931406498, "learning_rate": 1.1379637626378226e-07, "loss": 0.0002, "step": 229490 }, { "epoch": 3.755215577190542, "grad_norm": 0.030859949067234993, "learning_rate": 1.1364497284430342e-07, "loss": 0.0002, "step": 229500 }, { "epoch": 3.755379203141618, "grad_norm": 0.038179781287908554, "learning_rate": 1.1349366905344771e-07, "loss": 0.0004, "step": 229510 }, { "epoch": 3.755542829092694, "grad_norm": 0.04358275979757309, "learning_rate": 1.1334246489430046e-07, "loss": 0.0003, "step": 229520 }, { "epoch": 3.7557064550437698, "grad_norm": 0.0010200385004281998, "learning_rate": 1.1319136036994472e-07, "loss": 0.0003, "step": 229530 }, { "epoch": 3.7558700809948458, "grad_norm": 0.006183219607919455, "learning_rate": 1.1304035548346081e-07, "loss": 0.0002, "step": 229540 }, { "epoch": 3.7560337069459218, "grad_norm": 0.0004509396094363183, "learning_rate": 1.1288945023792797e-07, "loss": 0.0002, "step": 229550 }, { "epoch": 3.7561973328969973, "grad_norm": 0.02804328314960003, "learning_rate": 1.1273864463642259e-07, "loss": 0.0002, "step": 229560 }, { "epoch": 3.7563609588480733, "grad_norm": 0.0006682368693873286, "learning_rate": 1.1258793868202e-07, "loss": 0.0002, "step": 229570 }, { "epoch": 3.7565245847991493, "grad_norm": 0.005518913734704256, "learning_rate": 1.1243733237779275e-07, "loss": 0.0003, "step": 229580 }, { "epoch": 3.756688210750225, "grad_norm": 0.00013028418470639735, "learning_rate": 1.1228682572681116e-07, "loss": 0.0002, "step": 229590 }, { "epoch": 3.756851836701301, "grad_norm": 0.06853540986776352, "learning_rate": 1.1213641873214443e-07, "loss": 0.0004, "step": 229600 }, { "epoch": 3.757015462652377, "grad_norm": 0.003225046908482909, "learning_rate": 1.1198611139686011e-07, "loss": 0.0002, "step": 229610 }, { "epoch": 3.7571790886034524, "grad_norm": 4.273962622391991e-05, "learning_rate": 1.1183590372402075e-07, "loss": 0.0004, "step": 229620 }, { "epoch": 3.7573427145545284, "grad_norm": 0.009220748208463192, "learning_rate": 1.1168579571669059e-07, "loss": 0.0001, "step": 229630 }, { "epoch": 3.7575063405056044, "grad_norm": 0.016506165266036987, "learning_rate": 1.1153578737792937e-07, "loss": 0.0002, "step": 229640 }, { "epoch": 3.75766996645668, "grad_norm": 0.030049821361899376, "learning_rate": 1.1138587871079631e-07, "loss": 0.0002, "step": 229650 }, { "epoch": 3.757833592407756, "grad_norm": 0.00010189450404141098, "learning_rate": 1.1123606971834732e-07, "loss": 0.0003, "step": 229660 }, { "epoch": 3.757997218358832, "grad_norm": 0.020838553085923195, "learning_rate": 1.1108636040363718e-07, "loss": 0.0002, "step": 229670 }, { "epoch": 3.7581608443099075, "grad_norm": 0.022988442331552505, "learning_rate": 1.1093675076971844e-07, "loss": 0.0002, "step": 229680 }, { "epoch": 3.7583244702609835, "grad_norm": 0.0016145906411111355, "learning_rate": 1.1078724081964143e-07, "loss": 0.0002, "step": 229690 }, { "epoch": 3.7584880962120595, "grad_norm": 0.19279597699642181, "learning_rate": 1.1063783055645428e-07, "loss": 0.0003, "step": 229700 }, { "epoch": 3.758651722163135, "grad_norm": 8.866122516337782e-05, "learning_rate": 1.1048851998320343e-07, "loss": 0.0001, "step": 229710 }, { "epoch": 3.758815348114211, "grad_norm": 0.01719116047024727, "learning_rate": 1.1033930910293311e-07, "loss": 0.0002, "step": 229720 }, { "epoch": 3.758978974065287, "grad_norm": 0.07658778131008148, "learning_rate": 1.1019019791868646e-07, "loss": 0.0002, "step": 229730 }, { "epoch": 3.7591426000163626, "grad_norm": 0.009381746873259544, "learning_rate": 1.1004118643350214e-07, "loss": 0.0002, "step": 229740 }, { "epoch": 3.7593062259674386, "grad_norm": 0.022495577111840248, "learning_rate": 1.0989227465041997e-07, "loss": 0.0002, "step": 229750 }, { "epoch": 3.759469851918514, "grad_norm": 0.033072393387556076, "learning_rate": 1.097434625724747e-07, "loss": 0.0002, "step": 229760 }, { "epoch": 3.75963347786959, "grad_norm": 0.018452757969498634, "learning_rate": 1.0959475020270227e-07, "loss": 0.0003, "step": 229770 }, { "epoch": 3.759797103820666, "grad_norm": 0.024094045162200928, "learning_rate": 1.0944613754413302e-07, "loss": 0.0004, "step": 229780 }, { "epoch": 3.7599607297717417, "grad_norm": 0.038265153765678406, "learning_rate": 1.0929762459979786e-07, "loss": 0.0004, "step": 229790 }, { "epoch": 3.7601243557228177, "grad_norm": 0.012648671865463257, "learning_rate": 1.0914921137272439e-07, "loss": 0.0001, "step": 229800 }, { "epoch": 3.760287981673893, "grad_norm": 0.011707935482263565, "learning_rate": 1.0900089786594015e-07, "loss": 0.0001, "step": 229810 }, { "epoch": 3.760451607624969, "grad_norm": 0.005844694096595049, "learning_rate": 1.0885268408246663e-07, "loss": 0.0004, "step": 229820 }, { "epoch": 3.760615233576045, "grad_norm": 0.009903378784656525, "learning_rate": 1.0870457002532808e-07, "loss": 0.0005, "step": 229830 }, { "epoch": 3.7607788595271208, "grad_norm": 0.01878104917705059, "learning_rate": 1.0855655569754264e-07, "loss": 0.0001, "step": 229840 }, { "epoch": 3.7609424854781968, "grad_norm": 0.015439596027135849, "learning_rate": 1.0840864110212956e-07, "loss": 0.0003, "step": 229850 }, { "epoch": 3.7611061114292728, "grad_norm": 0.008146064355969429, "learning_rate": 1.0826082624210421e-07, "loss": 0.0002, "step": 229860 }, { "epoch": 3.7612697373803483, "grad_norm": 0.017682824283838272, "learning_rate": 1.0811311112048029e-07, "loss": 0.0001, "step": 229870 }, { "epoch": 3.7614333633314243, "grad_norm": 0.019063521176576614, "learning_rate": 1.0796549574026872e-07, "loss": 0.0002, "step": 229880 }, { "epoch": 3.7615969892825003, "grad_norm": 0.03261654078960419, "learning_rate": 1.0781798010448153e-07, "loss": 0.0002, "step": 229890 }, { "epoch": 3.761760615233576, "grad_norm": 0.028815820813179016, "learning_rate": 1.076705642161241e-07, "loss": 0.0002, "step": 229900 }, { "epoch": 3.761924241184652, "grad_norm": 0.015648096799850464, "learning_rate": 1.0752324807820403e-07, "loss": 0.0002, "step": 229910 }, { "epoch": 3.762087867135728, "grad_norm": 0.01018570177257061, "learning_rate": 1.0737603169372279e-07, "loss": 0.0004, "step": 229920 }, { "epoch": 3.7622514930868034, "grad_norm": 0.02123810164630413, "learning_rate": 1.0722891506568412e-07, "loss": 0.0004, "step": 229930 }, { "epoch": 3.7624151190378794, "grad_norm": 0.007566848304122686, "learning_rate": 1.0708189819708614e-07, "loss": 0.0003, "step": 229940 }, { "epoch": 3.7625787449889554, "grad_norm": 0.06601417809724808, "learning_rate": 1.0693498109092759e-07, "loss": 0.0002, "step": 229950 }, { "epoch": 3.762742370940031, "grad_norm": 0.010164470411837101, "learning_rate": 1.0678816375020273e-07, "loss": 0.0002, "step": 229960 }, { "epoch": 3.762905996891107, "grad_norm": 0.03425804153084755, "learning_rate": 1.0664144617790639e-07, "loss": 0.0003, "step": 229970 }, { "epoch": 3.763069622842183, "grad_norm": 0.004400037694722414, "learning_rate": 1.0649482837702896e-07, "loss": 0.0002, "step": 229980 }, { "epoch": 3.7632332487932585, "grad_norm": 0.011756121180951595, "learning_rate": 1.0634831035056026e-07, "loss": 0.0002, "step": 229990 }, { "epoch": 3.7633968747443345, "grad_norm": 0.03317946940660477, "learning_rate": 1.0620189210148735e-07, "loss": 0.0003, "step": 230000 }, { "epoch": 3.7635605006954105, "grad_norm": 0.015978077426552773, "learning_rate": 1.0605557363279617e-07, "loss": 0.0001, "step": 230010 }, { "epoch": 3.763724126646486, "grad_norm": 0.003134384984150529, "learning_rate": 1.0590935494746935e-07, "loss": 0.0003, "step": 230020 }, { "epoch": 3.763887752597562, "grad_norm": 0.06165801361203194, "learning_rate": 1.0576323604848893e-07, "loss": 0.0003, "step": 230030 }, { "epoch": 3.764051378548638, "grad_norm": 0.016950296238064766, "learning_rate": 1.0561721693883364e-07, "loss": 0.0002, "step": 230040 }, { "epoch": 3.7642150044997136, "grad_norm": 0.04219474643468857, "learning_rate": 1.0547129762148056e-07, "loss": 0.0002, "step": 230050 }, { "epoch": 3.7643786304507896, "grad_norm": 0.0022071904968470335, "learning_rate": 1.0532547809940507e-07, "loss": 0.0008, "step": 230060 }, { "epoch": 3.7645422564018656, "grad_norm": 0.00017666710482444614, "learning_rate": 1.0517975837558037e-07, "loss": 0.0004, "step": 230070 }, { "epoch": 3.764705882352941, "grad_norm": 0.21941693127155304, "learning_rate": 1.0503413845297739e-07, "loss": 0.0009, "step": 230080 }, { "epoch": 3.764869508304017, "grad_norm": 0.15043771266937256, "learning_rate": 1.0488861833456543e-07, "loss": 0.0004, "step": 230090 }, { "epoch": 3.765033134255093, "grad_norm": 0.006494206376373768, "learning_rate": 1.0474319802331157e-07, "loss": 0.0004, "step": 230100 }, { "epoch": 3.7651967602061687, "grad_norm": 0.006154426373541355, "learning_rate": 1.0459787752218064e-07, "loss": 0.0002, "step": 230110 }, { "epoch": 3.7653603861572447, "grad_norm": 0.0009022806189022958, "learning_rate": 1.0445265683413531e-07, "loss": 0.0002, "step": 230120 }, { "epoch": 3.76552401210832, "grad_norm": 0.03286191448569298, "learning_rate": 1.0430753596213705e-07, "loss": 0.0003, "step": 230130 }, { "epoch": 3.765687638059396, "grad_norm": 0.01089558843523264, "learning_rate": 1.0416251490914409e-07, "loss": 0.0004, "step": 230140 }, { "epoch": 3.765851264010472, "grad_norm": 0.14418962597846985, "learning_rate": 1.040175936781146e-07, "loss": 0.0002, "step": 230150 }, { "epoch": 3.7660148899615478, "grad_norm": 0.026239093393087387, "learning_rate": 1.0387277227200176e-07, "loss": 0.0002, "step": 230160 }, { "epoch": 3.7661785159126238, "grad_norm": 0.022988615557551384, "learning_rate": 1.037280506937588e-07, "loss": 0.0002, "step": 230170 }, { "epoch": 3.7663421418636993, "grad_norm": 0.0002856631763279438, "learning_rate": 1.0358342894633722e-07, "loss": 0.0001, "step": 230180 }, { "epoch": 3.7665057678147753, "grad_norm": 0.011779005639255047, "learning_rate": 1.0343890703268523e-07, "loss": 0.0001, "step": 230190 }, { "epoch": 3.7666693937658513, "grad_norm": 0.07703487575054169, "learning_rate": 1.0329448495574934e-07, "loss": 0.0002, "step": 230200 }, { "epoch": 3.766833019716927, "grad_norm": 0.045828577131032944, "learning_rate": 1.0315016271847444e-07, "loss": 0.0003, "step": 230210 }, { "epoch": 3.766996645668003, "grad_norm": 0.0004170121974311769, "learning_rate": 1.0300594032380318e-07, "loss": 0.0001, "step": 230220 }, { "epoch": 3.767160271619079, "grad_norm": 0.02981688268482685, "learning_rate": 1.0286181777467652e-07, "loss": 0.0002, "step": 230230 }, { "epoch": 3.7673238975701544, "grad_norm": 0.023606279864907265, "learning_rate": 1.0271779507403213e-07, "loss": 0.0002, "step": 230240 }, { "epoch": 3.7674875235212304, "grad_norm": 0.0015155529836192727, "learning_rate": 1.0257387222480653e-07, "loss": 0.0003, "step": 230250 }, { "epoch": 3.7676511494723064, "grad_norm": 0.015143110416829586, "learning_rate": 1.0243004922993516e-07, "loss": 0.0004, "step": 230260 }, { "epoch": 3.767814775423382, "grad_norm": 0.10166559368371964, "learning_rate": 1.0228632609234957e-07, "loss": 0.0003, "step": 230270 }, { "epoch": 3.767978401374458, "grad_norm": 0.12444323301315308, "learning_rate": 1.0214270281498074e-07, "loss": 0.0004, "step": 230280 }, { "epoch": 3.768142027325534, "grad_norm": 0.01272834837436676, "learning_rate": 1.0199917940075631e-07, "loss": 0.0001, "step": 230290 }, { "epoch": 3.7683056532766095, "grad_norm": 0.02250533178448677, "learning_rate": 1.0185575585260288e-07, "loss": 0.0004, "step": 230300 }, { "epoch": 3.7684692792276855, "grad_norm": 0.01916183903813362, "learning_rate": 1.0171243217344583e-07, "loss": 0.0003, "step": 230310 }, { "epoch": 3.7686329051787615, "grad_norm": 0.06086881458759308, "learning_rate": 1.0156920836620566e-07, "loss": 0.0003, "step": 230320 }, { "epoch": 3.768796531129837, "grad_norm": 0.00017278680752497166, "learning_rate": 1.014260844338033e-07, "loss": 0.0002, "step": 230330 }, { "epoch": 3.768960157080913, "grad_norm": 0.04661509767174721, "learning_rate": 1.0128306037915703e-07, "loss": 0.0004, "step": 230340 }, { "epoch": 3.769123783031989, "grad_norm": 0.0005515561788342893, "learning_rate": 1.0114013620518338e-07, "loss": 0.0003, "step": 230350 }, { "epoch": 3.7692874089830646, "grad_norm": 0.013348554261028767, "learning_rate": 1.0099731191479556e-07, "loss": 0.0003, "step": 230360 }, { "epoch": 3.7694510349341406, "grad_norm": 0.022482069209218025, "learning_rate": 1.0085458751090626e-07, "loss": 0.0005, "step": 230370 }, { "epoch": 3.7696146608852166, "grad_norm": 0.02454916574060917, "learning_rate": 1.0071196299642483e-07, "loss": 0.0005, "step": 230380 }, { "epoch": 3.769778286836292, "grad_norm": 0.0011993865482509136, "learning_rate": 1.0056943837426114e-07, "loss": 0.0003, "step": 230390 }, { "epoch": 3.769941912787368, "grad_norm": 4.8286954552168027e-05, "learning_rate": 1.0042701364731844e-07, "loss": 0.0002, "step": 230400 }, { "epoch": 3.770105538738444, "grad_norm": 0.03269389271736145, "learning_rate": 1.0028468881850273e-07, "loss": 0.0002, "step": 230410 }, { "epoch": 3.7702691646895197, "grad_norm": 0.012508995831012726, "learning_rate": 1.0014246389071447e-07, "loss": 0.0003, "step": 230420 }, { "epoch": 3.7704327906405957, "grad_norm": 0.03949466347694397, "learning_rate": 1.0000033886685523e-07, "loss": 0.0002, "step": 230430 }, { "epoch": 3.7705964165916717, "grad_norm": 0.00382249872200191, "learning_rate": 9.985831374982102e-08, "loss": 0.0004, "step": 230440 }, { "epoch": 3.770760042542747, "grad_norm": 0.0025203239638358355, "learning_rate": 9.971638854250843e-08, "loss": 0.0002, "step": 230450 }, { "epoch": 3.770923668493823, "grad_norm": 0.0155729278922081, "learning_rate": 9.957456324781123e-08, "loss": 0.0002, "step": 230460 }, { "epoch": 3.771087294444899, "grad_norm": 0.017999226227402687, "learning_rate": 9.943283786862102e-08, "loss": 0.0003, "step": 230470 }, { "epoch": 3.7712509203959748, "grad_norm": 0.028450123965740204, "learning_rate": 9.929121240782769e-08, "loss": 0.0001, "step": 230480 }, { "epoch": 3.7714145463470508, "grad_norm": 0.0030916451942175627, "learning_rate": 9.914968686831894e-08, "loss": 0.0002, "step": 230490 }, { "epoch": 3.7715781722981268, "grad_norm": 0.020155608654022217, "learning_rate": 9.900826125297858e-08, "loss": 0.0002, "step": 230500 }, { "epoch": 3.7717417982492023, "grad_norm": 0.045230936259031296, "learning_rate": 9.886693556469317e-08, "loss": 0.0003, "step": 230510 }, { "epoch": 3.7719054242002783, "grad_norm": 0.011577301658689976, "learning_rate": 9.87257098063421e-08, "loss": 0.0001, "step": 230520 }, { "epoch": 3.772069050151354, "grad_norm": 0.06015946343541145, "learning_rate": 9.858458398080584e-08, "loss": 0.0003, "step": 230530 }, { "epoch": 3.77223267610243, "grad_norm": 0.016663815826177597, "learning_rate": 9.84435580909604e-08, "loss": 0.0002, "step": 230540 }, { "epoch": 3.772396302053506, "grad_norm": 0.034223563969135284, "learning_rate": 9.830263213968294e-08, "loss": 0.0003, "step": 230550 }, { "epoch": 3.7725599280045814, "grad_norm": 0.03709347918629646, "learning_rate": 9.816180612984561e-08, "loss": 0.0004, "step": 230560 }, { "epoch": 3.7727235539556574, "grad_norm": 0.0012807268649339676, "learning_rate": 9.802108006432109e-08, "loss": 0.0002, "step": 230570 }, { "epoch": 3.772887179906733, "grad_norm": 0.01217977236956358, "learning_rate": 9.788045394597656e-08, "loss": 0.0002, "step": 230580 }, { "epoch": 3.773050805857809, "grad_norm": 0.06559693813323975, "learning_rate": 9.773992777768137e-08, "loss": 0.0003, "step": 230590 }, { "epoch": 3.773214431808885, "grad_norm": 0.07540278136730194, "learning_rate": 9.759950156229936e-08, "loss": 0.0005, "step": 230600 }, { "epoch": 3.7733780577599605, "grad_norm": 0.039662230759859085, "learning_rate": 9.745917530269489e-08, "loss": 0.0003, "step": 230610 }, { "epoch": 3.7735416837110365, "grad_norm": 0.03586586192250252, "learning_rate": 9.731894900172734e-08, "loss": 0.0004, "step": 230620 }, { "epoch": 3.7737053096621125, "grad_norm": 0.04319726303219795, "learning_rate": 9.71788226622572e-08, "loss": 0.0003, "step": 230630 }, { "epoch": 3.773868935613188, "grad_norm": 0.002766441321000457, "learning_rate": 9.703879628714163e-08, "loss": 0.0002, "step": 230640 }, { "epoch": 3.774032561564264, "grad_norm": 0.004154320806264877, "learning_rate": 9.689886987923502e-08, "loss": 0.0002, "step": 230650 }, { "epoch": 3.77419618751534, "grad_norm": 0.014294327236711979, "learning_rate": 9.675904344139008e-08, "loss": 0.0001, "step": 230660 }, { "epoch": 3.7743598134664156, "grad_norm": 0.0140670882537961, "learning_rate": 9.661931697645844e-08, "loss": 0.0002, "step": 230670 }, { "epoch": 3.7745234394174916, "grad_norm": 0.04939091205596924, "learning_rate": 9.647969048728888e-08, "loss": 0.0003, "step": 230680 }, { "epoch": 3.7746870653685676, "grad_norm": 0.0004894527373835444, "learning_rate": 9.634016397672807e-08, "loss": 0.0001, "step": 230690 }, { "epoch": 3.774850691319643, "grad_norm": 0.023218154907226562, "learning_rate": 9.620073744762093e-08, "loss": 0.0002, "step": 230700 }, { "epoch": 3.775014317270719, "grad_norm": 0.10879167169332504, "learning_rate": 9.606141090280962e-08, "loss": 0.0004, "step": 230710 }, { "epoch": 3.775177943221795, "grad_norm": 0.02031555213034153, "learning_rate": 9.592218434513634e-08, "loss": 0.0003, "step": 230720 }, { "epoch": 3.7753415691728707, "grad_norm": 0.06939562410116196, "learning_rate": 9.578305777743879e-08, "loss": 0.0006, "step": 230730 }, { "epoch": 3.7755051951239467, "grad_norm": 0.026003845036029816, "learning_rate": 9.564403120255361e-08, "loss": 0.0001, "step": 230740 }, { "epoch": 3.7756688210750227, "grad_norm": 0.012502556666731834, "learning_rate": 9.550510462331519e-08, "loss": 0.0001, "step": 230750 }, { "epoch": 3.775832447026098, "grad_norm": 0.0016033927677199244, "learning_rate": 9.536627804255738e-08, "loss": 0.0002, "step": 230760 }, { "epoch": 3.775996072977174, "grad_norm": 0.03283066675066948, "learning_rate": 9.522755146310958e-08, "loss": 0.0002, "step": 230770 }, { "epoch": 3.77615969892825, "grad_norm": 0.2630903422832489, "learning_rate": 9.508892488780064e-08, "loss": 0.0004, "step": 230780 }, { "epoch": 3.7763233248793258, "grad_norm": 0.02856048382818699, "learning_rate": 9.495039831945718e-08, "loss": 0.0003, "step": 230790 }, { "epoch": 3.7764869508304018, "grad_norm": 0.01103216502815485, "learning_rate": 9.481197176090307e-08, "loss": 0.0002, "step": 230800 }, { "epoch": 3.7766505767814778, "grad_norm": 0.017127884551882744, "learning_rate": 9.467364521496158e-08, "loss": 0.0001, "step": 230810 }, { "epoch": 3.7768142027325533, "grad_norm": 0.014849645085632801, "learning_rate": 9.45354186844527e-08, "loss": 0.0001, "step": 230820 }, { "epoch": 3.7769778286836293, "grad_norm": 0.02088957279920578, "learning_rate": 9.439729217219418e-08, "loss": 0.0001, "step": 230830 }, { "epoch": 3.7771414546347053, "grad_norm": 0.015260945074260235, "learning_rate": 9.425926568100319e-08, "loss": 0.0002, "step": 230840 }, { "epoch": 3.777305080585781, "grad_norm": 0.014246596023440361, "learning_rate": 9.41213392136936e-08, "loss": 0.0001, "step": 230850 }, { "epoch": 3.777468706536857, "grad_norm": 0.002821593778207898, "learning_rate": 9.398351277307761e-08, "loss": 0.0003, "step": 230860 }, { "epoch": 3.777632332487933, "grad_norm": 0.0010893485741689801, "learning_rate": 9.384578636196518e-08, "loss": 0.0001, "step": 230870 }, { "epoch": 3.7777959584390084, "grad_norm": 0.01435165572911501, "learning_rate": 9.370815998316463e-08, "loss": 0.0004, "step": 230880 }, { "epoch": 3.7779595843900844, "grad_norm": 0.04814400523900986, "learning_rate": 9.357063363948205e-08, "loss": 0.0002, "step": 230890 }, { "epoch": 3.77812321034116, "grad_norm": 0.003842715173959732, "learning_rate": 9.343320733372186e-08, "loss": 0.0004, "step": 230900 }, { "epoch": 3.778286836292236, "grad_norm": 0.03560378775000572, "learning_rate": 9.329588106868514e-08, "loss": 0.0002, "step": 230910 }, { "epoch": 3.778450462243312, "grad_norm": 0.017166318371891975, "learning_rate": 9.315865484717245e-08, "loss": 0.0001, "step": 230920 }, { "epoch": 3.7786140881943875, "grad_norm": 0.0626198872923851, "learning_rate": 9.302152867198211e-08, "loss": 0.0002, "step": 230930 }, { "epoch": 3.7787777141454635, "grad_norm": 0.009075548499822617, "learning_rate": 9.288450254590909e-08, "loss": 0.0005, "step": 230940 }, { "epoch": 3.778941340096539, "grad_norm": 0.011035403236746788, "learning_rate": 9.274757647174837e-08, "loss": 0.0005, "step": 230950 }, { "epoch": 3.779104966047615, "grad_norm": 0.0845511332154274, "learning_rate": 9.261075045229051e-08, "loss": 0.001, "step": 230960 }, { "epoch": 3.779268591998691, "grad_norm": 0.025389457121491432, "learning_rate": 9.247402449032606e-08, "loss": 0.0004, "step": 230970 }, { "epoch": 3.7794322179497666, "grad_norm": 0.007538375910371542, "learning_rate": 9.233739858864277e-08, "loss": 0.0003, "step": 230980 }, { "epoch": 3.7795958439008426, "grad_norm": 0.004006000701338053, "learning_rate": 9.220087275002564e-08, "loss": 0.0001, "step": 230990 }, { "epoch": 3.7797594698519186, "grad_norm": 0.001545339822769165, "learning_rate": 9.206444697725858e-08, "loss": 0.0001, "step": 231000 }, { "epoch": 3.779923095802994, "grad_norm": 0.0002175591653212905, "learning_rate": 9.192812127312433e-08, "loss": 0.0002, "step": 231010 }, { "epoch": 3.78008672175407, "grad_norm": 0.05647727847099304, "learning_rate": 9.17918956404007e-08, "loss": 0.0002, "step": 231020 }, { "epoch": 3.780250347705146, "grad_norm": 0.0382552295923233, "learning_rate": 9.165577008186655e-08, "loss": 0.0003, "step": 231030 }, { "epoch": 3.7804139736562217, "grad_norm": 0.03763643279671669, "learning_rate": 9.151974460029634e-08, "loss": 0.0002, "step": 231040 }, { "epoch": 3.7805775996072977, "grad_norm": 0.048377733677625656, "learning_rate": 9.13838191984645e-08, "loss": 0.0002, "step": 231050 }, { "epoch": 3.7807412255583737, "grad_norm": 0.01014506071805954, "learning_rate": 9.124799387914162e-08, "loss": 0.0003, "step": 231060 }, { "epoch": 3.780904851509449, "grad_norm": 0.04559725150465965, "learning_rate": 9.111226864509771e-08, "loss": 0.0001, "step": 231070 }, { "epoch": 3.781068477460525, "grad_norm": 0.00476654339581728, "learning_rate": 9.09766434990994e-08, "loss": 0.0001, "step": 231080 }, { "epoch": 3.781232103411601, "grad_norm": 0.02008981630206108, "learning_rate": 9.084111844391286e-08, "loss": 0.0005, "step": 231090 }, { "epoch": 3.7813957293626768, "grad_norm": 0.022261526435613632, "learning_rate": 9.070569348230085e-08, "loss": 0.0002, "step": 231100 }, { "epoch": 3.7815593553137528, "grad_norm": 0.01253675576299429, "learning_rate": 9.057036861702451e-08, "loss": 0.0002, "step": 231110 }, { "epoch": 3.7817229812648288, "grad_norm": 0.04300680384039879, "learning_rate": 9.043514385084218e-08, "loss": 0.0002, "step": 231120 }, { "epoch": 3.7818866072159043, "grad_norm": 0.060444049537181854, "learning_rate": 9.030001918651276e-08, "loss": 0.0004, "step": 231130 }, { "epoch": 3.7820502331669803, "grad_norm": 0.02020401880145073, "learning_rate": 9.016499462679019e-08, "loss": 0.0002, "step": 231140 }, { "epoch": 3.7822138591180563, "grad_norm": 0.05564092844724655, "learning_rate": 9.003007017442777e-08, "loss": 0.0003, "step": 231150 }, { "epoch": 3.782377485069132, "grad_norm": 0.027967868372797966, "learning_rate": 8.989524583217557e-08, "loss": 0.0002, "step": 231160 }, { "epoch": 3.782541111020208, "grad_norm": 0.17478832602500916, "learning_rate": 8.97605216027847e-08, "loss": 0.0019, "step": 231170 }, { "epoch": 3.782704736971284, "grad_norm": 0.09673842787742615, "learning_rate": 8.962589748899963e-08, "loss": 0.0003, "step": 231180 }, { "epoch": 3.7828683629223594, "grad_norm": 0.040496379137039185, "learning_rate": 8.949137349356763e-08, "loss": 0.0003, "step": 231190 }, { "epoch": 3.7830319888734354, "grad_norm": 0.04096417501568794, "learning_rate": 8.935694961922869e-08, "loss": 0.0004, "step": 231200 }, { "epoch": 3.7831956148245114, "grad_norm": 0.007521204184740782, "learning_rate": 8.922262586872677e-08, "loss": 0.0003, "step": 231210 }, { "epoch": 3.783359240775587, "grad_norm": 0.02491609938442707, "learning_rate": 8.908840224479797e-08, "loss": 0.0003, "step": 231220 }, { "epoch": 3.783522866726663, "grad_norm": 0.003751585027202964, "learning_rate": 8.895427875018126e-08, "loss": 0.0001, "step": 231230 }, { "epoch": 3.783686492677739, "grad_norm": 0.0065505835227668285, "learning_rate": 8.88202553876083e-08, "loss": 0.0002, "step": 231240 }, { "epoch": 3.7838501186288145, "grad_norm": 6.646291149081662e-05, "learning_rate": 8.868633215981526e-08, "loss": 0.0002, "step": 231250 }, { "epoch": 3.7840137445798905, "grad_norm": 0.09864336997270584, "learning_rate": 8.855250906952995e-08, "loss": 0.0003, "step": 231260 }, { "epoch": 3.7841773705309665, "grad_norm": 0.030208932235836983, "learning_rate": 8.84187861194824e-08, "loss": 0.0004, "step": 231270 }, { "epoch": 3.784340996482042, "grad_norm": 0.022923069074749947, "learning_rate": 8.828516331239823e-08, "loss": 0.0002, "step": 231280 }, { "epoch": 3.784504622433118, "grad_norm": 0.15179628133773804, "learning_rate": 8.815164065100301e-08, "loss": 0.0001, "step": 231290 }, { "epoch": 3.7846682483841936, "grad_norm": 0.06298604607582092, "learning_rate": 8.801821813801792e-08, "loss": 0.0004, "step": 231300 }, { "epoch": 3.7848318743352696, "grad_norm": 0.04897647351026535, "learning_rate": 8.788489577616411e-08, "loss": 0.0002, "step": 231310 }, { "epoch": 3.7849955002863456, "grad_norm": 0.028854260221123695, "learning_rate": 8.775167356815939e-08, "loss": 0.0004, "step": 231320 }, { "epoch": 3.785159126237421, "grad_norm": 0.05924156680703163, "learning_rate": 8.76185515167205e-08, "loss": 0.0005, "step": 231330 }, { "epoch": 3.785322752188497, "grad_norm": 0.012829780578613281, "learning_rate": 8.748552962456136e-08, "loss": 0.0003, "step": 231340 }, { "epoch": 3.7854863781395727, "grad_norm": 0.0007772128446958959, "learning_rate": 8.735260789439481e-08, "loss": 0.0004, "step": 231350 }, { "epoch": 3.7856500040906487, "grad_norm": 0.00230172136798501, "learning_rate": 8.721978632893036e-08, "loss": 0.0003, "step": 231360 }, { "epoch": 3.7858136300417247, "grad_norm": 0.03673519566655159, "learning_rate": 8.708706493087581e-08, "loss": 0.0002, "step": 231370 }, { "epoch": 3.7859772559928, "grad_norm": 0.03186556324362755, "learning_rate": 8.695444370293793e-08, "loss": 0.0002, "step": 231380 }, { "epoch": 3.786140881943876, "grad_norm": 0.01425672136247158, "learning_rate": 8.682192264782119e-08, "loss": 0.0001, "step": 231390 }, { "epoch": 3.786304507894952, "grad_norm": 0.12949389219284058, "learning_rate": 8.668950176822621e-08, "loss": 0.0001, "step": 231400 }, { "epoch": 3.7864681338460278, "grad_norm": 0.055670611560344696, "learning_rate": 8.655718106685417e-08, "loss": 0.0006, "step": 231410 }, { "epoch": 3.7866317597971038, "grad_norm": 0.01667688600718975, "learning_rate": 8.642496054640181e-08, "loss": 0.0004, "step": 231420 }, { "epoch": 3.7867953857481798, "grad_norm": 0.06095891445875168, "learning_rate": 8.629284020956641e-08, "loss": 0.0002, "step": 231430 }, { "epoch": 3.7869590116992553, "grad_norm": 0.03598764166235924, "learning_rate": 8.616082005904026e-08, "loss": 0.0002, "step": 231440 }, { "epoch": 3.7871226376503313, "grad_norm": 0.013113250024616718, "learning_rate": 8.602890009751675e-08, "loss": 0.0004, "step": 231450 }, { "epoch": 3.7872862636014073, "grad_norm": 0.00010357026621932164, "learning_rate": 8.589708032768374e-08, "loss": 0.0003, "step": 231460 }, { "epoch": 3.787449889552483, "grad_norm": 0.038434144109487534, "learning_rate": 8.57653607522313e-08, "loss": 0.0002, "step": 231470 }, { "epoch": 3.787613515503559, "grad_norm": 0.031166592612862587, "learning_rate": 8.563374137384283e-08, "loss": 0.0002, "step": 231480 }, { "epoch": 3.787777141454635, "grad_norm": 0.0018130786484107375, "learning_rate": 8.550222219520288e-08, "loss": 0.0004, "step": 231490 }, { "epoch": 3.7879407674057104, "grad_norm": 0.009439527057111263, "learning_rate": 8.537080321899316e-08, "loss": 0.0002, "step": 231500 }, { "epoch": 3.7881043933567864, "grad_norm": 0.0064724585972726345, "learning_rate": 8.52394844478932e-08, "loss": 0.0005, "step": 231510 }, { "epoch": 3.7882680193078624, "grad_norm": 0.031200001016259193, "learning_rate": 8.510826588458032e-08, "loss": 0.0002, "step": 231520 }, { "epoch": 3.788431645258938, "grad_norm": 0.014310574159026146, "learning_rate": 8.497714753172959e-08, "loss": 0.0002, "step": 231530 }, { "epoch": 3.788595271210014, "grad_norm": 0.007077084854245186, "learning_rate": 8.484612939201441e-08, "loss": 0.0002, "step": 231540 }, { "epoch": 3.78875889716109, "grad_norm": 0.047255728393793106, "learning_rate": 8.471521146810769e-08, "loss": 0.0004, "step": 231550 }, { "epoch": 3.7889225231121655, "grad_norm": 0.025888660922646523, "learning_rate": 8.458439376267669e-08, "loss": 0.0001, "step": 231560 }, { "epoch": 3.7890861490632415, "grad_norm": 0.01471870020031929, "learning_rate": 8.44536762783893e-08, "loss": 0.0002, "step": 231570 }, { "epoch": 3.7892497750143175, "grad_norm": 0.023873664438724518, "learning_rate": 8.432305901791116e-08, "loss": 0.0002, "step": 231580 }, { "epoch": 3.789413400965393, "grad_norm": 0.015413123182952404, "learning_rate": 8.419254198390514e-08, "loss": 0.0002, "step": 231590 }, { "epoch": 3.789577026916469, "grad_norm": 0.008393206633627415, "learning_rate": 8.406212517903245e-08, "loss": 0.0, "step": 231600 }, { "epoch": 3.789740652867545, "grad_norm": 0.019316017627716064, "learning_rate": 8.393180860595262e-08, "loss": 0.0002, "step": 231610 }, { "epoch": 3.7899042788186206, "grad_norm": 0.0019814735278487206, "learning_rate": 8.380159226732132e-08, "loss": 0.0002, "step": 231620 }, { "epoch": 3.7900679047696966, "grad_norm": 0.03875036537647247, "learning_rate": 8.367147616579585e-08, "loss": 0.0004, "step": 231630 }, { "epoch": 3.7902315307207726, "grad_norm": 0.05320524051785469, "learning_rate": 8.354146030402688e-08, "loss": 0.0001, "step": 231640 }, { "epoch": 3.790395156671848, "grad_norm": 0.10047191381454468, "learning_rate": 8.341154468466617e-08, "loss": 0.0004, "step": 231650 }, { "epoch": 3.790558782622924, "grad_norm": 0.009769144468009472, "learning_rate": 8.328172931036272e-08, "loss": 0.0003, "step": 231660 }, { "epoch": 3.790722408574, "grad_norm": 0.03124522604048252, "learning_rate": 8.315201418376384e-08, "loss": 0.0001, "step": 231670 }, { "epoch": 3.7908860345250757, "grad_norm": 0.04680245742201805, "learning_rate": 8.302239930751299e-08, "loss": 0.0002, "step": 231680 }, { "epoch": 3.7910496604761517, "grad_norm": 0.011323564685881138, "learning_rate": 8.289288468425415e-08, "loss": 0.0002, "step": 231690 }, { "epoch": 3.791213286427227, "grad_norm": 0.09773209691047668, "learning_rate": 8.276347031662691e-08, "loss": 0.0006, "step": 231700 }, { "epoch": 3.791376912378303, "grad_norm": 0.06313443183898926, "learning_rate": 8.263415620727134e-08, "loss": 0.0004, "step": 231710 }, { "epoch": 3.7915405383293788, "grad_norm": 0.058963462710380554, "learning_rate": 8.250494235882311e-08, "loss": 0.0004, "step": 231720 }, { "epoch": 3.7917041642804548, "grad_norm": 0.0045481291599571705, "learning_rate": 8.23758287739168e-08, "loss": 0.0003, "step": 231730 }, { "epoch": 3.7918677902315308, "grad_norm": 0.04701860994100571, "learning_rate": 8.224681545518476e-08, "loss": 0.0002, "step": 231740 }, { "epoch": 3.7920314161826063, "grad_norm": 0.007129038218408823, "learning_rate": 8.211790240525819e-08, "loss": 0.0005, "step": 231750 }, { "epoch": 3.7921950421336823, "grad_norm": 0.02397509478032589, "learning_rate": 8.198908962676499e-08, "loss": 0.0001, "step": 231760 }, { "epoch": 3.7923586680847583, "grad_norm": 0.010474376380443573, "learning_rate": 8.186037712233196e-08, "loss": 0.0001, "step": 231770 }, { "epoch": 3.792522294035834, "grad_norm": 0.024610048159956932, "learning_rate": 8.173176489458201e-08, "loss": 0.0002, "step": 231780 }, { "epoch": 3.79268591998691, "grad_norm": 0.0003221376391593367, "learning_rate": 8.160325294613969e-08, "loss": 0.0003, "step": 231790 }, { "epoch": 3.792849545937986, "grad_norm": 0.009787330403923988, "learning_rate": 8.147484127962347e-08, "loss": 0.0004, "step": 231800 }, { "epoch": 3.7930131718890614, "grad_norm": 0.014209561049938202, "learning_rate": 8.13465298976529e-08, "loss": 0.0002, "step": 231810 }, { "epoch": 3.7931767978401374, "grad_norm": 0.032125379890203476, "learning_rate": 8.121831880284203e-08, "loss": 0.0001, "step": 231820 }, { "epoch": 3.7933404237912134, "grad_norm": 0.14504899084568024, "learning_rate": 8.109020799780765e-08, "loss": 0.0004, "step": 231830 }, { "epoch": 3.793504049742289, "grad_norm": 0.000267522205831483, "learning_rate": 8.096219748515987e-08, "loss": 0.0001, "step": 231840 }, { "epoch": 3.793667675693365, "grad_norm": 0.03861680254340172, "learning_rate": 8.083428726750998e-08, "loss": 0.0004, "step": 231850 }, { "epoch": 3.793831301644441, "grad_norm": 0.00038469937862828374, "learning_rate": 8.070647734746473e-08, "loss": 0.0003, "step": 231860 }, { "epoch": 3.7939949275955165, "grad_norm": 0.011755772866308689, "learning_rate": 8.057876772763096e-08, "loss": 0.0022, "step": 231870 }, { "epoch": 3.7941585535465925, "grad_norm": 0.045268986374139786, "learning_rate": 8.045115841061268e-08, "loss": 0.0005, "step": 231880 }, { "epoch": 3.7943221794976685, "grad_norm": 0.00636486429721117, "learning_rate": 8.032364939901116e-08, "loss": 0.0, "step": 231890 }, { "epoch": 3.794485805448744, "grad_norm": 0.03142883628606796, "learning_rate": 8.019624069542541e-08, "loss": 0.0007, "step": 231900 }, { "epoch": 3.79464943139982, "grad_norm": 0.060911376029253006, "learning_rate": 8.00689323024556e-08, "loss": 0.0004, "step": 231910 }, { "epoch": 3.794813057350896, "grad_norm": 0.004609587136656046, "learning_rate": 7.99417242226952e-08, "loss": 0.0004, "step": 231920 }, { "epoch": 3.7949766833019716, "grad_norm": 0.03356783837080002, "learning_rate": 7.981461645873933e-08, "loss": 0.0001, "step": 231930 }, { "epoch": 3.7951403092530476, "grad_norm": 0.18255212903022766, "learning_rate": 7.968760901317818e-08, "loss": 0.0008, "step": 231940 }, { "epoch": 3.7953039352041236, "grad_norm": 0.021620100364089012, "learning_rate": 7.956070188860243e-08, "loss": 0.0002, "step": 231950 }, { "epoch": 3.795467561155199, "grad_norm": 0.026108017191290855, "learning_rate": 7.943389508759947e-08, "loss": 0.0002, "step": 231960 }, { "epoch": 3.795631187106275, "grad_norm": 0.04650769382715225, "learning_rate": 7.930718861275444e-08, "loss": 0.0003, "step": 231970 }, { "epoch": 3.795794813057351, "grad_norm": 0.0203146543353796, "learning_rate": 7.918058246665084e-08, "loss": 0.0004, "step": 231980 }, { "epoch": 3.7959584390084267, "grad_norm": 0.021994762122631073, "learning_rate": 7.905407665187048e-08, "loss": 0.0005, "step": 231990 }, { "epoch": 3.7961220649595027, "grad_norm": 0.00872511975467205, "learning_rate": 7.892767117099187e-08, "loss": 0.0002, "step": 232000 }, { "epoch": 3.7962856909105787, "grad_norm": 0.042825642973184586, "learning_rate": 7.88013660265935e-08, "loss": 0.0005, "step": 232010 }, { "epoch": 3.796449316861654, "grad_norm": 0.01385131198912859, "learning_rate": 7.867516122124941e-08, "loss": 0.0002, "step": 232020 }, { "epoch": 3.79661294281273, "grad_norm": 0.004393429029732943, "learning_rate": 7.85490567575331e-08, "loss": 0.0001, "step": 232030 }, { "epoch": 3.796776568763806, "grad_norm": 0.007532848976552486, "learning_rate": 7.842305263801641e-08, "loss": 0.0002, "step": 232040 }, { "epoch": 3.7969401947148818, "grad_norm": 0.03504055738449097, "learning_rate": 7.82971488652684e-08, "loss": 0.0002, "step": 232050 }, { "epoch": 3.7971038206659578, "grad_norm": 0.000862400745972991, "learning_rate": 7.817134544185535e-08, "loss": 0.0001, "step": 232060 }, { "epoch": 3.7972674466170333, "grad_norm": 0.030164439231157303, "learning_rate": 7.80456423703424e-08, "loss": 0.0003, "step": 232070 }, { "epoch": 3.7974310725681093, "grad_norm": 0.0388847216963768, "learning_rate": 7.79200396532931e-08, "loss": 0.0002, "step": 232080 }, { "epoch": 3.7975946985191853, "grad_norm": 0.003722786670550704, "learning_rate": 7.779453729326813e-08, "loss": 0.0002, "step": 232090 }, { "epoch": 3.797758324470261, "grad_norm": 0.0015812052879482508, "learning_rate": 7.76691352928266e-08, "loss": 0.0001, "step": 232100 }, { "epoch": 3.797921950421337, "grad_norm": 0.0018011750653386116, "learning_rate": 7.754383365452478e-08, "loss": 0.0003, "step": 232110 }, { "epoch": 3.7980855763724124, "grad_norm": 0.0002354718162678182, "learning_rate": 7.741863238091785e-08, "loss": 0.0006, "step": 232120 }, { "epoch": 3.7982492023234884, "grad_norm": 0.00032069499138742685, "learning_rate": 7.729353147455875e-08, "loss": 0.0003, "step": 232130 }, { "epoch": 3.7984128282745644, "grad_norm": 0.02770596370100975, "learning_rate": 7.71685309379977e-08, "loss": 0.0001, "step": 232140 }, { "epoch": 3.79857645422564, "grad_norm": 0.0062593515031039715, "learning_rate": 7.704363077378319e-08, "loss": 0.0001, "step": 232150 }, { "epoch": 3.798740080176716, "grad_norm": 0.012736932374536991, "learning_rate": 7.691883098446262e-08, "loss": 0.0003, "step": 232160 }, { "epoch": 3.798903706127792, "grad_norm": 0.05332467705011368, "learning_rate": 7.679413157258065e-08, "loss": 0.0003, "step": 232170 }, { "epoch": 3.7990673320788675, "grad_norm": 0.0032111098989844322, "learning_rate": 7.666953254067855e-08, "loss": 0.0001, "step": 232180 }, { "epoch": 3.7992309580299435, "grad_norm": 0.010955343954265118, "learning_rate": 7.654503389129763e-08, "loss": 0.0002, "step": 232190 }, { "epoch": 3.7993945839810195, "grad_norm": 0.05188930034637451, "learning_rate": 7.642063562697644e-08, "loss": 0.0003, "step": 232200 }, { "epoch": 3.799558209932095, "grad_norm": 0.019167279824614525, "learning_rate": 7.629633775025124e-08, "loss": 0.0001, "step": 232210 }, { "epoch": 3.799721835883171, "grad_norm": 0.06584112346172333, "learning_rate": 7.617214026365616e-08, "loss": 0.0002, "step": 232220 }, { "epoch": 3.799885461834247, "grad_norm": 0.036455631256103516, "learning_rate": 7.604804316972358e-08, "loss": 0.0002, "step": 232230 }, { "epoch": 3.8000490877853226, "grad_norm": 0.008142962120473385, "learning_rate": 7.592404647098317e-08, "loss": 0.0003, "step": 232240 }, { "epoch": 3.8002127137363986, "grad_norm": 0.01260942593216896, "learning_rate": 7.580015016996456e-08, "loss": 0.0001, "step": 232250 }, { "epoch": 3.8003763396874746, "grad_norm": 0.02449032850563526, "learning_rate": 7.56763542691924e-08, "loss": 0.0002, "step": 232260 }, { "epoch": 3.80053996563855, "grad_norm": 0.01880807615816593, "learning_rate": 7.555265877119189e-08, "loss": 0.0002, "step": 232270 }, { "epoch": 3.800703591589626, "grad_norm": 0.006190719082951546, "learning_rate": 7.542906367848435e-08, "loss": 0.0005, "step": 232280 }, { "epoch": 3.800867217540702, "grad_norm": 0.004902289714664221, "learning_rate": 7.530556899358998e-08, "loss": 0.0002, "step": 232290 }, { "epoch": 3.8010308434917777, "grad_norm": 0.029462376609444618, "learning_rate": 7.518217471902677e-08, "loss": 0.0004, "step": 232300 }, { "epoch": 3.8011944694428537, "grad_norm": 0.0012754095951095223, "learning_rate": 7.50588808573105e-08, "loss": 0.0001, "step": 232310 }, { "epoch": 3.8013580953939297, "grad_norm": 0.08132106065750122, "learning_rate": 7.493568741095469e-08, "loss": 0.0002, "step": 232320 }, { "epoch": 3.801521721345005, "grad_norm": 0.010911224409937859, "learning_rate": 7.48125943824729e-08, "loss": 0.0002, "step": 232330 }, { "epoch": 3.801685347296081, "grad_norm": 0.031944602727890015, "learning_rate": 7.468960177437257e-08, "loss": 0.0002, "step": 232340 }, { "epoch": 3.801848973247157, "grad_norm": 0.007444322109222412, "learning_rate": 7.456670958916224e-08, "loss": 0.0002, "step": 232350 }, { "epoch": 3.8020125991982328, "grad_norm": 0.0690184012055397, "learning_rate": 7.444391782934823e-08, "loss": 0.0002, "step": 232360 }, { "epoch": 3.8021762251493088, "grad_norm": 0.06187097728252411, "learning_rate": 7.432122649743412e-08, "loss": 0.0006, "step": 232370 }, { "epoch": 3.8023398511003847, "grad_norm": 0.010053768754005432, "learning_rate": 7.419863559592066e-08, "loss": 0.0002, "step": 232380 }, { "epoch": 3.8025034770514603, "grad_norm": 0.0008718224707990885, "learning_rate": 7.407614512730755e-08, "loss": 0.0002, "step": 232390 }, { "epoch": 3.8026671030025363, "grad_norm": 0.015646042302250862, "learning_rate": 7.395375509409219e-08, "loss": 0.0004, "step": 232400 }, { "epoch": 3.8028307289536123, "grad_norm": 0.005364570766687393, "learning_rate": 7.383146549877151e-08, "loss": 0.0008, "step": 232410 }, { "epoch": 3.802994354904688, "grad_norm": 0.0014883170370012522, "learning_rate": 7.370927634383685e-08, "loss": 0.0002, "step": 232420 }, { "epoch": 3.803157980855764, "grad_norm": 0.033452294766902924, "learning_rate": 7.358718763178063e-08, "loss": 0.0004, "step": 232430 }, { "epoch": 3.80332160680684, "grad_norm": 0.006851024925708771, "learning_rate": 7.346519936509145e-08, "loss": 0.0001, "step": 232440 }, { "epoch": 3.8034852327579154, "grad_norm": 9.657959890319034e-05, "learning_rate": 7.334331154625729e-08, "loss": 0.0002, "step": 232450 }, { "epoch": 3.8036488587089914, "grad_norm": 0.00014234981790650636, "learning_rate": 7.322152417776285e-08, "loss": 0.0002, "step": 232460 }, { "epoch": 3.803812484660067, "grad_norm": 0.012129362672567368, "learning_rate": 7.309983726209225e-08, "loss": 0.0003, "step": 232470 }, { "epoch": 3.803976110611143, "grad_norm": 0.040866393595933914, "learning_rate": 7.29782508017246e-08, "loss": 0.0002, "step": 232480 }, { "epoch": 3.804139736562219, "grad_norm": 0.02194795571267605, "learning_rate": 7.285676479914127e-08, "loss": 0.0003, "step": 232490 }, { "epoch": 3.8043033625132945, "grad_norm": 0.02880783937871456, "learning_rate": 7.273537925681751e-08, "loss": 0.0005, "step": 232500 }, { "epoch": 3.8044669884643705, "grad_norm": 0.09818008542060852, "learning_rate": 7.261409417722909e-08, "loss": 0.0006, "step": 232510 }, { "epoch": 3.804630614415446, "grad_norm": 0.007969762198626995, "learning_rate": 7.249290956284794e-08, "loss": 0.0003, "step": 232520 }, { "epoch": 3.804794240366522, "grad_norm": 0.01585097797214985, "learning_rate": 7.237182541614651e-08, "loss": 0.0003, "step": 232530 }, { "epoch": 3.804957866317598, "grad_norm": 0.03378874063491821, "learning_rate": 7.225084173959285e-08, "loss": 0.0003, "step": 232540 }, { "epoch": 3.8051214922686736, "grad_norm": 0.019765805453062057, "learning_rate": 7.21299585356533e-08, "loss": 0.0002, "step": 232550 }, { "epoch": 3.8052851182197496, "grad_norm": 0.050344448536634445, "learning_rate": 7.200917580679256e-08, "loss": 0.0003, "step": 232560 }, { "epoch": 3.8054487441708256, "grad_norm": 0.002192220650613308, "learning_rate": 7.188849355547423e-08, "loss": 0.0004, "step": 232570 }, { "epoch": 3.805612370121901, "grad_norm": 0.005856642033904791, "learning_rate": 7.176791178415799e-08, "loss": 0.0002, "step": 232580 }, { "epoch": 3.805775996072977, "grad_norm": 0.04847966507077217, "learning_rate": 7.164743049530298e-08, "loss": 0.0003, "step": 232590 }, { "epoch": 3.805939622024053, "grad_norm": 0.0002639127487782389, "learning_rate": 7.152704969136504e-08, "loss": 0.0002, "step": 232600 }, { "epoch": 3.8061032479751287, "grad_norm": 0.0019028570968657732, "learning_rate": 7.14067693747994e-08, "loss": 0.0001, "step": 232610 }, { "epoch": 3.8062668739262047, "grad_norm": 0.010060274973511696, "learning_rate": 7.128658954805745e-08, "loss": 0.0002, "step": 232620 }, { "epoch": 3.8064304998772807, "grad_norm": 0.05425138399004936, "learning_rate": 7.116651021359111e-08, "loss": 0.0002, "step": 232630 }, { "epoch": 3.806594125828356, "grad_norm": 0.04976722225546837, "learning_rate": 7.104653137384732e-08, "loss": 0.0003, "step": 232640 }, { "epoch": 3.806757751779432, "grad_norm": 0.020944438874721527, "learning_rate": 7.0926653031273e-08, "loss": 0.0003, "step": 232650 }, { "epoch": 3.806921377730508, "grad_norm": 0.0005654380074702203, "learning_rate": 7.080687518831175e-08, "loss": 0.0003, "step": 232660 }, { "epoch": 3.8070850036815838, "grad_norm": 0.0004235291271470487, "learning_rate": 7.068719784740718e-08, "loss": 0.0004, "step": 232670 }, { "epoch": 3.8072486296326598, "grad_norm": 0.06201142817735672, "learning_rate": 7.05676210109979e-08, "loss": 0.0004, "step": 232680 }, { "epoch": 3.8074122555837357, "grad_norm": 0.08660507947206497, "learning_rate": 7.044814468152194e-08, "loss": 0.0002, "step": 232690 }, { "epoch": 3.8075758815348113, "grad_norm": 0.030611388385295868, "learning_rate": 7.032876886141627e-08, "loss": 0.0004, "step": 232700 }, { "epoch": 3.8077395074858873, "grad_norm": 0.03294803574681282, "learning_rate": 7.020949355311502e-08, "loss": 0.0002, "step": 232710 }, { "epoch": 3.8079031334369633, "grad_norm": 0.051645684987306595, "learning_rate": 7.009031875904903e-08, "loss": 0.0002, "step": 232720 }, { "epoch": 3.808066759388039, "grad_norm": 0.00842302106320858, "learning_rate": 6.99712444816486e-08, "loss": 0.0002, "step": 232730 }, { "epoch": 3.808230385339115, "grad_norm": 0.012112047523260117, "learning_rate": 6.985227072334178e-08, "loss": 0.0001, "step": 232740 }, { "epoch": 3.808394011290191, "grad_norm": 0.019259626045823097, "learning_rate": 6.973339748655439e-08, "loss": 0.0003, "step": 232750 }, { "epoch": 3.8085576372412664, "grad_norm": 0.02181416191160679, "learning_rate": 6.96146247737095e-08, "loss": 0.0001, "step": 232760 }, { "epoch": 3.8087212631923424, "grad_norm": 0.02660098485648632, "learning_rate": 6.949595258722963e-08, "loss": 0.0001, "step": 232770 }, { "epoch": 3.8088848891434184, "grad_norm": 0.0005168615025468171, "learning_rate": 6.937738092953394e-08, "loss": 0.0004, "step": 232780 }, { "epoch": 3.809048515094494, "grad_norm": 0.03633192181587219, "learning_rate": 6.92589098030405e-08, "loss": 0.0003, "step": 232790 }, { "epoch": 3.80921214104557, "grad_norm": 0.00724498787894845, "learning_rate": 6.914053921016406e-08, "loss": 0.0003, "step": 232800 }, { "epoch": 3.809375766996646, "grad_norm": 0.06300099939107895, "learning_rate": 6.902226915331822e-08, "loss": 0.0004, "step": 232810 }, { "epoch": 3.8095393929477215, "grad_norm": 0.06249717250466347, "learning_rate": 6.890409963491495e-08, "loss": 0.0002, "step": 232820 }, { "epoch": 3.8097030188987975, "grad_norm": 0.010445484891533852, "learning_rate": 6.8786030657364e-08, "loss": 0.0003, "step": 232830 }, { "epoch": 3.809866644849873, "grad_norm": 0.06335531920194626, "learning_rate": 6.866806222307121e-08, "loss": 0.0003, "step": 232840 }, { "epoch": 3.810030270800949, "grad_norm": 0.051224008202552795, "learning_rate": 6.855019433444243e-08, "loss": 0.0003, "step": 232850 }, { "epoch": 3.810193896752025, "grad_norm": 0.0021193602588027716, "learning_rate": 6.843242699388187e-08, "loss": 0.0003, "step": 232860 }, { "epoch": 3.8103575227031006, "grad_norm": 0.005812444724142551, "learning_rate": 6.83147602037898e-08, "loss": 0.0002, "step": 232870 }, { "epoch": 3.8105211486541766, "grad_norm": 0.031186392530798912, "learning_rate": 6.819719396656544e-08, "loss": 0.0003, "step": 232880 }, { "epoch": 3.810684774605252, "grad_norm": 0.002180119976401329, "learning_rate": 6.807972828460573e-08, "loss": 0.0002, "step": 232890 }, { "epoch": 3.810848400556328, "grad_norm": 0.026130765676498413, "learning_rate": 6.796236316030602e-08, "loss": 0.0001, "step": 232900 }, { "epoch": 3.811012026507404, "grad_norm": 0.0005051686312071979, "learning_rate": 6.784509859605937e-08, "loss": 0.0003, "step": 232910 }, { "epoch": 3.8111756524584797, "grad_norm": 0.00023465644335374236, "learning_rate": 6.772793459425665e-08, "loss": 0.0001, "step": 232920 }, { "epoch": 3.8113392784095557, "grad_norm": 0.00113893102388829, "learning_rate": 6.761087115728648e-08, "loss": 0.0003, "step": 232930 }, { "epoch": 3.8115029043606317, "grad_norm": 0.002477531787008047, "learning_rate": 6.749390828753533e-08, "loss": 0.0003, "step": 232940 }, { "epoch": 3.811666530311707, "grad_norm": 0.009943443350493908, "learning_rate": 6.737704598738903e-08, "loss": 0.0002, "step": 232950 }, { "epoch": 3.811830156262783, "grad_norm": 0.001790891052223742, "learning_rate": 6.72602842592296e-08, "loss": 0.0003, "step": 232960 }, { "epoch": 3.811993782213859, "grad_norm": 0.026584573090076447, "learning_rate": 6.714362310543732e-08, "loss": 0.0003, "step": 232970 }, { "epoch": 3.8121574081649348, "grad_norm": 0.021509315818548203, "learning_rate": 6.702706252839197e-08, "loss": 0.0003, "step": 232980 }, { "epoch": 3.8123210341160108, "grad_norm": 0.07359650731086731, "learning_rate": 6.691060253046944e-08, "loss": 0.0004, "step": 232990 }, { "epoch": 3.8124846600670868, "grad_norm": 0.0312582328915596, "learning_rate": 6.679424311404392e-08, "loss": 0.0001, "step": 233000 }, { "epoch": 3.8126482860181623, "grad_norm": 0.018939146772027016, "learning_rate": 6.667798428148798e-08, "loss": 0.0005, "step": 233010 }, { "epoch": 3.8128119119692383, "grad_norm": 0.004341980908066034, "learning_rate": 6.656182603517303e-08, "loss": 0.0001, "step": 233020 }, { "epoch": 3.8129755379203143, "grad_norm": 0.021315498277544975, "learning_rate": 6.644576837746663e-08, "loss": 0.0002, "step": 233030 }, { "epoch": 3.81313916387139, "grad_norm": 0.08030238002538681, "learning_rate": 6.632981131073469e-08, "loss": 0.0002, "step": 233040 }, { "epoch": 3.813302789822466, "grad_norm": 0.062012430280447006, "learning_rate": 6.62139548373425e-08, "loss": 0.0002, "step": 233050 }, { "epoch": 3.813466415773542, "grad_norm": 0.0014361540088430047, "learning_rate": 6.609819895965097e-08, "loss": 0.0003, "step": 233060 }, { "epoch": 3.8136300417246174, "grad_norm": 0.06116388738155365, "learning_rate": 6.598254368002155e-08, "loss": 0.0003, "step": 233070 }, { "epoch": 3.8137936676756934, "grad_norm": 0.04365900531411171, "learning_rate": 6.586698900081179e-08, "loss": 0.0002, "step": 233080 }, { "epoch": 3.8139572936267694, "grad_norm": 0.03644154220819473, "learning_rate": 6.575153492437813e-08, "loss": 0.0003, "step": 233090 }, { "epoch": 3.814120919577845, "grad_norm": 0.03357277065515518, "learning_rate": 6.563618145307316e-08, "loss": 0.0002, "step": 233100 }, { "epoch": 3.814284545528921, "grad_norm": 0.05915634334087372, "learning_rate": 6.552092858925108e-08, "loss": 0.0004, "step": 233110 }, { "epoch": 3.814448171479997, "grad_norm": 0.02519446425139904, "learning_rate": 6.540577633526001e-08, "loss": 0.0002, "step": 233120 }, { "epoch": 3.8146117974310725, "grad_norm": 0.0015456774272024632, "learning_rate": 6.529072469344921e-08, "loss": 0.0004, "step": 233130 }, { "epoch": 3.8147754233821485, "grad_norm": 0.03556786850094795, "learning_rate": 6.51757736661629e-08, "loss": 0.0002, "step": 233140 }, { "epoch": 3.8149390493332245, "grad_norm": 0.011308843269944191, "learning_rate": 6.506092325574586e-08, "loss": 0.0001, "step": 233150 }, { "epoch": 3.8151026752843, "grad_norm": 0.011424418538808823, "learning_rate": 6.494617346453957e-08, "loss": 0.0002, "step": 233160 }, { "epoch": 3.815266301235376, "grad_norm": 0.001094555831514299, "learning_rate": 6.483152429488382e-08, "loss": 0.0001, "step": 233170 }, { "epoch": 3.815429927186452, "grad_norm": 0.04208850488066673, "learning_rate": 6.471697574911562e-08, "loss": 0.0004, "step": 233180 }, { "epoch": 3.8155935531375276, "grad_norm": 0.0003039324947167188, "learning_rate": 6.460252782957199e-08, "loss": 0.0001, "step": 233190 }, { "epoch": 3.8157571790886036, "grad_norm": 0.04142163321375847, "learning_rate": 6.448818053858441e-08, "loss": 0.0002, "step": 233200 }, { "epoch": 3.8159208050396796, "grad_norm": 0.0689089447259903, "learning_rate": 6.437393387848601e-08, "loss": 0.001, "step": 233210 }, { "epoch": 3.816084430990755, "grad_norm": 0.035396724939346313, "learning_rate": 6.425978785160492e-08, "loss": 0.0002, "step": 233220 }, { "epoch": 3.816248056941831, "grad_norm": 0.000458333088317886, "learning_rate": 6.414574246026927e-08, "loss": 0.0003, "step": 233230 }, { "epoch": 3.8164116828929067, "grad_norm": 0.08275055885314941, "learning_rate": 6.403179770680445e-08, "loss": 0.0004, "step": 233240 }, { "epoch": 3.8165753088439827, "grad_norm": 0.06579848378896713, "learning_rate": 6.391795359353304e-08, "loss": 0.0002, "step": 233250 }, { "epoch": 3.8167389347950587, "grad_norm": 0.015268409624695778, "learning_rate": 6.380421012277649e-08, "loss": 0.0001, "step": 233260 }, { "epoch": 3.816902560746134, "grad_norm": 0.0726628303527832, "learning_rate": 6.369056729685408e-08, "loss": 0.0006, "step": 233270 }, { "epoch": 3.81706618669721, "grad_norm": 0.042424917221069336, "learning_rate": 6.357702511808284e-08, "loss": 0.0002, "step": 233280 }, { "epoch": 3.8172298126482858, "grad_norm": 0.013723086565732956, "learning_rate": 6.346358358877813e-08, "loss": 0.0003, "step": 233290 }, { "epoch": 3.8173934385993618, "grad_norm": 0.004410591907799244, "learning_rate": 6.335024271125256e-08, "loss": 0.0002, "step": 233300 }, { "epoch": 3.8175570645504378, "grad_norm": 0.03335077688097954, "learning_rate": 6.32370024878165e-08, "loss": 0.0002, "step": 233310 }, { "epoch": 3.8177206905015133, "grad_norm": 0.07638564705848694, "learning_rate": 6.312386292077977e-08, "loss": 0.0004, "step": 233320 }, { "epoch": 3.8178843164525893, "grad_norm": 0.012170383706688881, "learning_rate": 6.301082401244885e-08, "loss": 0.0003, "step": 233330 }, { "epoch": 3.8180479424036653, "grad_norm": 0.007338371593505144, "learning_rate": 6.289788576512856e-08, "loss": 0.0003, "step": 233340 }, { "epoch": 3.818211568354741, "grad_norm": 0.0005732271238230169, "learning_rate": 6.278504818112097e-08, "loss": 0.0001, "step": 233350 }, { "epoch": 3.818375194305817, "grad_norm": 0.012801793403923512, "learning_rate": 6.267231126272754e-08, "loss": 0.0002, "step": 233360 }, { "epoch": 3.818538820256893, "grad_norm": 0.0021026229951530695, "learning_rate": 6.255967501224702e-08, "loss": 0.0002, "step": 233370 }, { "epoch": 3.8187024462079684, "grad_norm": 0.0002771488798316568, "learning_rate": 6.244713943197533e-08, "loss": 0.0002, "step": 233380 }, { "epoch": 3.8188660721590444, "grad_norm": 0.03345119208097458, "learning_rate": 6.233470452420731e-08, "loss": 0.0002, "step": 233390 }, { "epoch": 3.8190296981101204, "grad_norm": 0.04568612948060036, "learning_rate": 6.222237029123501e-08, "loss": 0.0004, "step": 233400 }, { "epoch": 3.819193324061196, "grad_norm": 0.032131265848875046, "learning_rate": 6.211013673534939e-08, "loss": 0.0006, "step": 233410 }, { "epoch": 3.819356950012272, "grad_norm": 0.06264649331569672, "learning_rate": 6.199800385883858e-08, "loss": 0.0004, "step": 233420 }, { "epoch": 3.819520575963348, "grad_norm": 0.011174771934747696, "learning_rate": 6.188597166398857e-08, "loss": 0.0002, "step": 233430 }, { "epoch": 3.8196842019144235, "grad_norm": 0.020869983360171318, "learning_rate": 6.177404015308419e-08, "loss": 0.0006, "step": 233440 }, { "epoch": 3.8198478278654995, "grad_norm": 0.08901209384202957, "learning_rate": 6.16622093284075e-08, "loss": 0.0003, "step": 233450 }, { "epoch": 3.8200114538165755, "grad_norm": 0.05058719590306282, "learning_rate": 6.155047919223833e-08, "loss": 0.0003, "step": 233460 }, { "epoch": 3.820175079767651, "grad_norm": 0.01864541880786419, "learning_rate": 6.143884974685488e-08, "loss": 0.0003, "step": 233470 }, { "epoch": 3.820338705718727, "grad_norm": 0.060861632227897644, "learning_rate": 6.132732099453254e-08, "loss": 0.0002, "step": 233480 }, { "epoch": 3.820502331669803, "grad_norm": 0.008390108123421669, "learning_rate": 6.12158929375467e-08, "loss": 0.0001, "step": 233490 }, { "epoch": 3.8206659576208786, "grad_norm": 0.0290097463876009, "learning_rate": 6.110456557816835e-08, "loss": 0.0002, "step": 233500 }, { "epoch": 3.8208295835719546, "grad_norm": 0.003086577169597149, "learning_rate": 6.099333891866732e-08, "loss": 0.0005, "step": 233510 }, { "epoch": 3.8209932095230306, "grad_norm": 0.026631029322743416, "learning_rate": 6.08822129613118e-08, "loss": 0.0002, "step": 233520 }, { "epoch": 3.821156835474106, "grad_norm": 0.0306314118206501, "learning_rate": 6.077118770836776e-08, "loss": 0.0006, "step": 233530 }, { "epoch": 3.821320461425182, "grad_norm": 0.10499487817287445, "learning_rate": 6.066026316209784e-08, "loss": 0.0004, "step": 233540 }, { "epoch": 3.821484087376258, "grad_norm": 0.048879578709602356, "learning_rate": 6.054943932476465e-08, "loss": 0.0002, "step": 233550 }, { "epoch": 3.8216477133273337, "grad_norm": 0.018085051327943802, "learning_rate": 6.043871619862751e-08, "loss": 0.0001, "step": 233560 }, { "epoch": 3.8218113392784097, "grad_norm": 0.11577393114566803, "learning_rate": 6.032809378594462e-08, "loss": 0.0002, "step": 233570 }, { "epoch": 3.8219749652294857, "grad_norm": 0.04436212778091431, "learning_rate": 6.021757208897028e-08, "loss": 0.0001, "step": 233580 }, { "epoch": 3.822138591180561, "grad_norm": 0.0011794234160333872, "learning_rate": 6.010715110995825e-08, "loss": 0.0006, "step": 233590 }, { "epoch": 3.822302217131637, "grad_norm": 0.003529251553118229, "learning_rate": 5.999683085116059e-08, "loss": 0.0002, "step": 233600 }, { "epoch": 3.8224658430827128, "grad_norm": 0.019598601385951042, "learning_rate": 5.98866113148261e-08, "loss": 0.0005, "step": 233610 }, { "epoch": 3.8226294690337888, "grad_norm": 0.003584187012165785, "learning_rate": 5.977649250320239e-08, "loss": 0.0002, "step": 233620 }, { "epoch": 3.8227930949848647, "grad_norm": 0.03336330130696297, "learning_rate": 5.966647441853434e-08, "loss": 0.0002, "step": 233630 }, { "epoch": 3.8229567209359403, "grad_norm": 0.0015844169538468122, "learning_rate": 5.955655706306518e-08, "loss": 0.0002, "step": 233640 }, { "epoch": 3.8231203468870163, "grad_norm": 0.03852821886539459, "learning_rate": 5.944674043903642e-08, "loss": 0.0002, "step": 233650 }, { "epoch": 3.823283972838092, "grad_norm": 0.0001988812000490725, "learning_rate": 5.933702454868628e-08, "loss": 0.0002, "step": 233660 }, { "epoch": 3.823447598789168, "grad_norm": 0.005774191580712795, "learning_rate": 5.9227409394252425e-08, "loss": 0.0002, "step": 233670 }, { "epoch": 3.823611224740244, "grad_norm": 0.013283731415867805, "learning_rate": 5.911789497797027e-08, "loss": 0.0001, "step": 233680 }, { "epoch": 3.8237748506913194, "grad_norm": 0.007811259012669325, "learning_rate": 5.9008481302071926e-08, "loss": 0.0002, "step": 233690 }, { "epoch": 3.8239384766423954, "grad_norm": 0.05704915523529053, "learning_rate": 5.8899168368788395e-08, "loss": 0.0002, "step": 233700 }, { "epoch": 3.8241021025934714, "grad_norm": 0.029197243973612785, "learning_rate": 5.8789956180348996e-08, "loss": 0.0003, "step": 233710 }, { "epoch": 3.824265728544547, "grad_norm": 0.030223151668906212, "learning_rate": 5.868084473897917e-08, "loss": 0.0003, "step": 233720 }, { "epoch": 3.824429354495623, "grad_norm": 0.007052435539662838, "learning_rate": 5.8571834046904916e-08, "loss": 0.0001, "step": 233730 }, { "epoch": 3.824592980446699, "grad_norm": 0.036502886563539505, "learning_rate": 5.846292410634835e-08, "loss": 0.0002, "step": 233740 }, { "epoch": 3.8247566063977745, "grad_norm": 0.030954239889979362, "learning_rate": 5.835411491953047e-08, "loss": 0.0002, "step": 233750 }, { "epoch": 3.8249202323488505, "grad_norm": 0.02837541699409485, "learning_rate": 5.8245406488668944e-08, "loss": 0.0002, "step": 233760 }, { "epoch": 3.8250838582999265, "grad_norm": 0.015012871474027634, "learning_rate": 5.813679881598144e-08, "loss": 0.0003, "step": 233770 }, { "epoch": 3.825247484251002, "grad_norm": 0.0004395744181238115, "learning_rate": 5.80282919036812e-08, "loss": 0.0002, "step": 233780 }, { "epoch": 3.825411110202078, "grad_norm": 0.005182893481105566, "learning_rate": 5.791988575398144e-08, "loss": 0.0004, "step": 233790 }, { "epoch": 3.825574736153154, "grad_norm": 0.014507177285850048, "learning_rate": 5.781158036909096e-08, "loss": 0.0005, "step": 233800 }, { "epoch": 3.8257383621042296, "grad_norm": 0.05085531994700432, "learning_rate": 5.770337575122076e-08, "loss": 0.0002, "step": 233810 }, { "epoch": 3.8259019880553056, "grad_norm": 0.0036773199681192636, "learning_rate": 5.7595271902574636e-08, "loss": 0.0003, "step": 233820 }, { "epoch": 3.8260656140063816, "grad_norm": 0.005808459594845772, "learning_rate": 5.7487268825358046e-08, "loss": 0.0006, "step": 233830 }, { "epoch": 3.826229239957457, "grad_norm": 0.0018467084737494588, "learning_rate": 5.737936652177145e-08, "loss": 0.0004, "step": 233840 }, { "epoch": 3.826392865908533, "grad_norm": 0.08537857979536057, "learning_rate": 5.7271564994017534e-08, "loss": 0.0002, "step": 233850 }, { "epoch": 3.826556491859609, "grad_norm": 0.047033198177814484, "learning_rate": 5.7163864244291764e-08, "loss": 0.0002, "step": 233860 }, { "epoch": 3.8267201178106847, "grad_norm": 0.004055826459079981, "learning_rate": 5.705626427479183e-08, "loss": 0.0001, "step": 233870 }, { "epoch": 3.8268837437617607, "grad_norm": 0.008887662552297115, "learning_rate": 5.6948765087709855e-08, "loss": 0.0001, "step": 233880 }, { "epoch": 3.8270473697128367, "grad_norm": 0.01449511107057333, "learning_rate": 5.6841366685239096e-08, "loss": 0.0009, "step": 233890 }, { "epoch": 3.827210995663912, "grad_norm": 0.07585658878087997, "learning_rate": 5.673406906956891e-08, "loss": 0.0003, "step": 233900 }, { "epoch": 3.827374621614988, "grad_norm": 0.0065934364683926105, "learning_rate": 5.6626872242887545e-08, "loss": 0.0001, "step": 233910 }, { "epoch": 3.827538247566064, "grad_norm": 0.0005693370476365089, "learning_rate": 5.651977620737936e-08, "loss": 0.0001, "step": 233920 }, { "epoch": 3.8277018735171398, "grad_norm": 9.046894410857931e-05, "learning_rate": 5.641278096522873e-08, "loss": 0.0005, "step": 233930 }, { "epoch": 3.8278654994682157, "grad_norm": 0.17758281528949738, "learning_rate": 5.6305886518617235e-08, "loss": 0.0004, "step": 233940 }, { "epoch": 3.8280291254192917, "grad_norm": 0.03820066154003143, "learning_rate": 5.619909286972425e-08, "loss": 0.0006, "step": 233950 }, { "epoch": 3.8281927513703673, "grad_norm": 0.03226963430643082, "learning_rate": 5.609240002072691e-08, "loss": 0.0002, "step": 233960 }, { "epoch": 3.8283563773214433, "grad_norm": 0.0410495288670063, "learning_rate": 5.5985807973801265e-08, "loss": 0.0001, "step": 233970 }, { "epoch": 3.8285200032725193, "grad_norm": 0.011594688519835472, "learning_rate": 5.587931673112002e-08, "loss": 0.0004, "step": 233980 }, { "epoch": 3.828683629223595, "grad_norm": 0.09201325476169586, "learning_rate": 5.5772926294855314e-08, "loss": 0.0003, "step": 233990 }, { "epoch": 3.828847255174671, "grad_norm": 0.0033453013747930527, "learning_rate": 5.5666636667174866e-08, "loss": 0.0021, "step": 234000 }, { "epoch": 3.8290108811257464, "grad_norm": 0.1673714518547058, "learning_rate": 5.5560447850246944e-08, "loss": 0.0003, "step": 234010 }, { "epoch": 3.8291745070768224, "grad_norm": 0.00017118104733526707, "learning_rate": 5.5454359846236485e-08, "loss": 0.0002, "step": 234020 }, { "epoch": 3.8293381330278984, "grad_norm": 0.045451272279024124, "learning_rate": 5.5348372657306745e-08, "loss": 0.0002, "step": 234030 }, { "epoch": 3.829501758978974, "grad_norm": 0.11362308263778687, "learning_rate": 5.5242486285618235e-08, "loss": 0.0007, "step": 234040 }, { "epoch": 3.82966538493005, "grad_norm": 0.029530789703130722, "learning_rate": 5.5136700733329775e-08, "loss": 0.001, "step": 234050 }, { "epoch": 3.8298290108811255, "grad_norm": 0.012948507443070412, "learning_rate": 5.503101600259853e-08, "loss": 0.0003, "step": 234060 }, { "epoch": 3.8299926368322015, "grad_norm": 0.01552500855177641, "learning_rate": 5.492543209557999e-08, "loss": 0.0002, "step": 234070 }, { "epoch": 3.8301562627832775, "grad_norm": 0.0005992981605231762, "learning_rate": 5.4819949014425776e-08, "loss": 0.0003, "step": 234080 }, { "epoch": 3.830319888734353, "grad_norm": 0.007876093499362469, "learning_rate": 5.4714566761286946e-08, "loss": 0.0003, "step": 234090 }, { "epoch": 3.830483514685429, "grad_norm": 0.02849663980305195, "learning_rate": 5.4609285338312335e-08, "loss": 0.0001, "step": 234100 }, { "epoch": 3.830647140636505, "grad_norm": 0.037814151495695114, "learning_rate": 5.4504104747648556e-08, "loss": 0.0003, "step": 234110 }, { "epoch": 3.8308107665875806, "grad_norm": 0.00010610848403302953, "learning_rate": 5.4399024991440005e-08, "loss": 0.0003, "step": 234120 }, { "epoch": 3.8309743925386566, "grad_norm": 0.07730753719806671, "learning_rate": 5.429404607182942e-08, "loss": 0.0001, "step": 234130 }, { "epoch": 3.8311380184897326, "grad_norm": 0.05797190964221954, "learning_rate": 5.418916799095675e-08, "loss": 0.0004, "step": 234140 }, { "epoch": 3.831301644440808, "grad_norm": 0.010292117483913898, "learning_rate": 5.4084390750961393e-08, "loss": 0.0001, "step": 234150 }, { "epoch": 3.831465270391884, "grad_norm": 0.005556423682719469, "learning_rate": 5.397971435397831e-08, "loss": 0.0004, "step": 234160 }, { "epoch": 3.83162889634296, "grad_norm": 0.015001432038843632, "learning_rate": 5.387513880214246e-08, "loss": 0.0002, "step": 234170 }, { "epoch": 3.8317925222940357, "grad_norm": 0.046788305044174194, "learning_rate": 5.377066409758602e-08, "loss": 0.0006, "step": 234180 }, { "epoch": 3.8319561482451117, "grad_norm": 0.03502902388572693, "learning_rate": 5.366629024243952e-08, "loss": 0.0003, "step": 234190 }, { "epoch": 3.8321197741961877, "grad_norm": 0.0007285664323717356, "learning_rate": 5.3562017238830147e-08, "loss": 0.0002, "step": 234200 }, { "epoch": 3.832283400147263, "grad_norm": 0.06114911660552025, "learning_rate": 5.345784508888507e-08, "loss": 0.0002, "step": 234210 }, { "epoch": 3.832447026098339, "grad_norm": 0.03387473523616791, "learning_rate": 5.335377379472706e-08, "loss": 0.0002, "step": 234220 }, { "epoch": 3.832610652049415, "grad_norm": 0.04812155291438103, "learning_rate": 5.3249803358478846e-08, "loss": 0.0004, "step": 234230 }, { "epoch": 3.8327742780004908, "grad_norm": 0.019187496975064278, "learning_rate": 5.314593378225985e-08, "loss": 0.0003, "step": 234240 }, { "epoch": 3.8329379039515667, "grad_norm": 0.026747750118374825, "learning_rate": 5.304216506818838e-08, "loss": 0.0004, "step": 234250 }, { "epoch": 3.8331015299026427, "grad_norm": 0.047677166759967804, "learning_rate": 5.293849721837996e-08, "loss": 0.0003, "step": 234260 }, { "epoch": 3.8332651558537183, "grad_norm": 0.06997208297252655, "learning_rate": 5.283493023494846e-08, "loss": 0.0004, "step": 234270 }, { "epoch": 3.8334287818047943, "grad_norm": 0.012478964403271675, "learning_rate": 5.273146412000496e-08, "loss": 0.0001, "step": 234280 }, { "epoch": 3.8335924077558703, "grad_norm": 0.01919412799179554, "learning_rate": 5.262809887565945e-08, "loss": 0.0003, "step": 234290 }, { "epoch": 3.833756033706946, "grad_norm": 0.013206827454268932, "learning_rate": 5.252483450401913e-08, "loss": 0.0002, "step": 234300 }, { "epoch": 3.833919659658022, "grad_norm": 0.000988520449027419, "learning_rate": 5.242167100719009e-08, "loss": 0.0002, "step": 234310 }, { "epoch": 3.834083285609098, "grad_norm": 0.01362193375825882, "learning_rate": 5.231860838727565e-08, "loss": 0.0002, "step": 234320 }, { "epoch": 3.8342469115601734, "grad_norm": 0.0014745796797797084, "learning_rate": 5.2215646646376906e-08, "loss": 0.0002, "step": 234330 }, { "epoch": 3.8344105375112494, "grad_norm": 0.03838697075843811, "learning_rate": 5.211278578659273e-08, "loss": 0.0003, "step": 234340 }, { "epoch": 3.8345741634623254, "grad_norm": 0.03082980401813984, "learning_rate": 5.20100258100209e-08, "loss": 0.0003, "step": 234350 }, { "epoch": 3.834737789413401, "grad_norm": 0.001189488684758544, "learning_rate": 5.1907366718756955e-08, "loss": 0.0002, "step": 234360 }, { "epoch": 3.834901415364477, "grad_norm": 0.03960356116294861, "learning_rate": 5.1804808514893115e-08, "loss": 0.0002, "step": 234370 }, { "epoch": 3.8350650413155525, "grad_norm": 0.011379989795386791, "learning_rate": 5.1702351200521026e-08, "loss": 0.0002, "step": 234380 }, { "epoch": 3.8352286672666285, "grad_norm": 0.0006245823460631073, "learning_rate": 5.1599994777730144e-08, "loss": 0.0002, "step": 234390 }, { "epoch": 3.8353922932177045, "grad_norm": 0.016853248700499535, "learning_rate": 5.149773924860602e-08, "loss": 0.0007, "step": 234400 }, { "epoch": 3.83555591916878, "grad_norm": 0.0002368368732277304, "learning_rate": 5.1395584615235306e-08, "loss": 0.0003, "step": 234410 }, { "epoch": 3.835719545119856, "grad_norm": 0.04768236726522446, "learning_rate": 5.129353087969913e-08, "loss": 0.0003, "step": 234420 }, { "epoch": 3.8358831710709316, "grad_norm": 0.00591812701895833, "learning_rate": 5.1191578044079704e-08, "loss": 0.0003, "step": 234430 }, { "epoch": 3.8360467970220076, "grad_norm": 0.019789405167102814, "learning_rate": 5.1089726110455374e-08, "loss": 0.0004, "step": 234440 }, { "epoch": 3.8362104229730836, "grad_norm": 0.10672655701637268, "learning_rate": 5.098797508090225e-08, "loss": 0.0002, "step": 234450 }, { "epoch": 3.836374048924159, "grad_norm": 0.00011457250366220251, "learning_rate": 5.088632495749535e-08, "loss": 0.0002, "step": 234460 }, { "epoch": 3.836537674875235, "grad_norm": 0.03642500936985016, "learning_rate": 5.078477574230745e-08, "loss": 0.0004, "step": 234470 }, { "epoch": 3.836701300826311, "grad_norm": 0.022190803661942482, "learning_rate": 5.068332743740856e-08, "loss": 0.0005, "step": 234480 }, { "epoch": 3.8368649267773867, "grad_norm": 0.021127570420503616, "learning_rate": 5.058198004486758e-08, "loss": 0.0003, "step": 234490 }, { "epoch": 3.8370285527284627, "grad_norm": 0.030476059764623642, "learning_rate": 5.0480733566750094e-08, "loss": 0.0005, "step": 234500 }, { "epoch": 3.8371921786795387, "grad_norm": 0.025961626321077347, "learning_rate": 5.037958800512221e-08, "loss": 0.0002, "step": 234510 }, { "epoch": 3.837355804630614, "grad_norm": 0.005100763402879238, "learning_rate": 5.0278543362043964e-08, "loss": 0.0003, "step": 234520 }, { "epoch": 3.83751943058169, "grad_norm": 0.00957468431442976, "learning_rate": 5.017759963957758e-08, "loss": 0.0003, "step": 234530 }, { "epoch": 3.837683056532766, "grad_norm": 0.04428458213806152, "learning_rate": 5.0076756839779196e-08, "loss": 0.0001, "step": 234540 }, { "epoch": 3.8378466824838418, "grad_norm": 0.015723882243037224, "learning_rate": 4.997601496470661e-08, "loss": 0.0002, "step": 234550 }, { "epoch": 3.8380103084349178, "grad_norm": 0.018393956124782562, "learning_rate": 4.987537401641318e-08, "loss": 0.0002, "step": 234560 }, { "epoch": 3.8381739343859937, "grad_norm": 0.012669961899518967, "learning_rate": 4.977483399695115e-08, "loss": 0.0001, "step": 234570 }, { "epoch": 3.8383375603370693, "grad_norm": 0.02437266893684864, "learning_rate": 4.967439490837e-08, "loss": 0.0002, "step": 234580 }, { "epoch": 3.8385011862881453, "grad_norm": 0.02526186592876911, "learning_rate": 4.957405675271754e-08, "loss": 0.0001, "step": 234590 }, { "epoch": 3.8386648122392213, "grad_norm": 0.0046725794672966, "learning_rate": 4.947381953203989e-08, "loss": 0.0001, "step": 234600 }, { "epoch": 3.838828438190297, "grad_norm": 0.01175262313336134, "learning_rate": 4.9373683248381543e-08, "loss": 0.0003, "step": 234610 }, { "epoch": 3.838992064141373, "grad_norm": 0.11360718309879303, "learning_rate": 4.927364790378253e-08, "loss": 0.0004, "step": 234620 }, { "epoch": 3.839155690092449, "grad_norm": 0.0525348037481308, "learning_rate": 4.9173713500283995e-08, "loss": 0.0003, "step": 234630 }, { "epoch": 3.8393193160435244, "grad_norm": 0.008064016699790955, "learning_rate": 4.9073880039922636e-08, "loss": 0.0002, "step": 234640 }, { "epoch": 3.8394829419946004, "grad_norm": 0.005983187351375818, "learning_rate": 4.897414752473462e-08, "loss": 0.0001, "step": 234650 }, { "epoch": 3.8396465679456764, "grad_norm": 0.07121299207210541, "learning_rate": 4.8874515956752745e-08, "loss": 0.0003, "step": 234660 }, { "epoch": 3.839810193896752, "grad_norm": 0.0035067033022642136, "learning_rate": 4.8774985338008726e-08, "loss": 0.0004, "step": 234670 }, { "epoch": 3.839973819847828, "grad_norm": 0.011394069530069828, "learning_rate": 4.86755556705315e-08, "loss": 0.0002, "step": 234680 }, { "epoch": 3.840137445798904, "grad_norm": 0.02819088287651539, "learning_rate": 4.857622695634945e-08, "loss": 0.0003, "step": 234690 }, { "epoch": 3.8403010717499795, "grad_norm": 0.02958821691572666, "learning_rate": 4.8476999197486494e-08, "loss": 0.0004, "step": 234700 }, { "epoch": 3.8404646977010555, "grad_norm": 0.014903008006513119, "learning_rate": 4.8377872395966585e-08, "loss": 0.0002, "step": 234710 }, { "epoch": 3.8406283236521315, "grad_norm": 0.00029304149211384356, "learning_rate": 4.8278846553810876e-08, "loss": 0.0002, "step": 234720 }, { "epoch": 3.840791949603207, "grad_norm": 0.009487394243478775, "learning_rate": 4.817992167303831e-08, "loss": 0.0004, "step": 234730 }, { "epoch": 3.840955575554283, "grad_norm": 0.04417652264237404, "learning_rate": 4.808109775566561e-08, "loss": 0.0002, "step": 234740 }, { "epoch": 3.841119201505359, "grad_norm": 0.04113645851612091, "learning_rate": 4.798237480370782e-08, "loss": 0.0003, "step": 234750 }, { "epoch": 3.8412828274564346, "grad_norm": 0.003955107647925615, "learning_rate": 4.7883752819177787e-08, "loss": 0.0, "step": 234760 }, { "epoch": 3.8414464534075106, "grad_norm": 0.02053542248904705, "learning_rate": 4.7785231804086674e-08, "loss": 0.0004, "step": 234770 }, { "epoch": 3.841610079358586, "grad_norm": 0.00634709931910038, "learning_rate": 4.768681176044232e-08, "loss": 0.0001, "step": 234780 }, { "epoch": 3.841773705309662, "grad_norm": 0.07064823806285858, "learning_rate": 4.758849269025256e-08, "loss": 0.0003, "step": 234790 }, { "epoch": 3.841937331260738, "grad_norm": 0.02062802202999592, "learning_rate": 4.7490274595521355e-08, "loss": 0.0002, "step": 234800 }, { "epoch": 3.8421009572118137, "grad_norm": 0.026863811537623405, "learning_rate": 4.73921574782521e-08, "loss": 0.0004, "step": 234810 }, { "epoch": 3.8422645831628897, "grad_norm": 0.007742343004792929, "learning_rate": 4.72941413404443e-08, "loss": 0.0003, "step": 234820 }, { "epoch": 3.842428209113965, "grad_norm": 0.06558939069509506, "learning_rate": 4.719622618409636e-08, "loss": 0.0002, "step": 234830 }, { "epoch": 3.842591835065041, "grad_norm": 0.0400533489882946, "learning_rate": 4.709841201120557e-08, "loss": 0.0004, "step": 234840 }, { "epoch": 3.842755461016117, "grad_norm": 0.020396705716848373, "learning_rate": 4.7000698823766456e-08, "loss": 0.0001, "step": 234850 }, { "epoch": 3.8429190869671928, "grad_norm": 0.00016965223767329007, "learning_rate": 4.690308662377019e-08, "loss": 0.0002, "step": 234860 }, { "epoch": 3.8430827129182688, "grad_norm": 0.03732188791036606, "learning_rate": 4.6805575413207404e-08, "loss": 0.0002, "step": 234870 }, { "epoch": 3.8432463388693447, "grad_norm": 0.010378445498645306, "learning_rate": 4.670816519406651e-08, "loss": 0.0002, "step": 234880 }, { "epoch": 3.8434099648204203, "grad_norm": 0.05016191676259041, "learning_rate": 4.66108559683337e-08, "loss": 0.0003, "step": 234890 }, { "epoch": 3.8435735907714963, "grad_norm": 0.0005379404756240547, "learning_rate": 4.651364773799238e-08, "loss": 0.0003, "step": 234900 }, { "epoch": 3.8437372167225723, "grad_norm": 0.07518358528614044, "learning_rate": 4.641654050502486e-08, "loss": 0.0004, "step": 234910 }, { "epoch": 3.843900842673648, "grad_norm": 0.07152844220399857, "learning_rate": 4.631953427141178e-08, "loss": 0.0004, "step": 234920 }, { "epoch": 3.844064468624724, "grad_norm": 0.06032605469226837, "learning_rate": 4.622262903913044e-08, "loss": 0.0003, "step": 234930 }, { "epoch": 3.8442280945758, "grad_norm": 0.004207140300422907, "learning_rate": 4.6125824810156484e-08, "loss": 0.0003, "step": 234940 }, { "epoch": 3.8443917205268754, "grad_norm": 0.011994215659797192, "learning_rate": 4.6029121586463885e-08, "loss": 0.0003, "step": 234950 }, { "epoch": 3.8445553464779514, "grad_norm": 0.024897025898098946, "learning_rate": 4.5932519370023855e-08, "loss": 0.0001, "step": 234960 }, { "epoch": 3.8447189724290274, "grad_norm": 0.058966245502233505, "learning_rate": 4.5836018162807025e-08, "loss": 0.0002, "step": 234970 }, { "epoch": 3.844882598380103, "grad_norm": 0.002279289998114109, "learning_rate": 4.573961796678017e-08, "loss": 0.0002, "step": 234980 }, { "epoch": 3.845046224331179, "grad_norm": 0.0032314585987478495, "learning_rate": 4.564331878390949e-08, "loss": 0.0004, "step": 234990 }, { "epoch": 3.845209850282255, "grad_norm": 0.05040430277585983, "learning_rate": 4.554712061615729e-08, "loss": 0.0005, "step": 235000 }, { "epoch": 3.8453734762333305, "grad_norm": 0.010899334214627743, "learning_rate": 4.5451023465486465e-08, "loss": 0.0005, "step": 235010 }, { "epoch": 3.8455371021844065, "grad_norm": 0.016963884234428406, "learning_rate": 4.535502733385488e-08, "loss": 0.0002, "step": 235020 }, { "epoch": 3.8457007281354825, "grad_norm": 0.02434944361448288, "learning_rate": 4.525913222322098e-08, "loss": 0.0006, "step": 235030 }, { "epoch": 3.845864354086558, "grad_norm": 0.0023185363970696926, "learning_rate": 4.5163338135538745e-08, "loss": 0.0005, "step": 235040 }, { "epoch": 3.846027980037634, "grad_norm": 0.024734649807214737, "learning_rate": 4.5067645072763286e-08, "loss": 0.0004, "step": 235050 }, { "epoch": 3.84619160598871, "grad_norm": 0.06330228596925735, "learning_rate": 4.49720530368436e-08, "loss": 0.0003, "step": 235060 }, { "epoch": 3.8463552319397856, "grad_norm": 0.0577508844435215, "learning_rate": 4.487656202973034e-08, "loss": 0.0003, "step": 235070 }, { "epoch": 3.8465188578908616, "grad_norm": 0.008987323381006718, "learning_rate": 4.478117205336918e-08, "loss": 0.0005, "step": 235080 }, { "epoch": 3.8466824838419376, "grad_norm": 0.004200365860015154, "learning_rate": 4.4685883109706344e-08, "loss": 0.0002, "step": 235090 }, { "epoch": 3.846846109793013, "grad_norm": 0.038501545786857605, "learning_rate": 4.459069520068304e-08, "loss": 0.0003, "step": 235100 }, { "epoch": 3.847009735744089, "grad_norm": 0.0006256556953303516, "learning_rate": 4.449560832824218e-08, "loss": 0.0001, "step": 235110 }, { "epoch": 3.847173361695165, "grad_norm": 0.011018837802112103, "learning_rate": 4.440062249431998e-08, "loss": 0.0005, "step": 235120 }, { "epoch": 3.8473369876462407, "grad_norm": 0.02013695240020752, "learning_rate": 4.4305737700856e-08, "loss": 0.0001, "step": 235130 }, { "epoch": 3.8475006135973167, "grad_norm": 0.014128380455076694, "learning_rate": 4.421095394978259e-08, "loss": 0.0003, "step": 235140 }, { "epoch": 3.8476642395483927, "grad_norm": 0.0071447547525167465, "learning_rate": 4.4116271243033194e-08, "loss": 0.0002, "step": 235150 }, { "epoch": 3.847827865499468, "grad_norm": 0.0006327113951556385, "learning_rate": 4.402168958253794e-08, "loss": 0.0001, "step": 235160 }, { "epoch": 3.847991491450544, "grad_norm": 0.04115954041481018, "learning_rate": 4.392720897022584e-08, "loss": 0.0003, "step": 235170 }, { "epoch": 3.8481551174016198, "grad_norm": 0.006259161047637463, "learning_rate": 4.383282940802258e-08, "loss": 0.0001, "step": 235180 }, { "epoch": 3.8483187433526957, "grad_norm": 0.020837970077991486, "learning_rate": 4.3738550897853284e-08, "loss": 0.0002, "step": 235190 }, { "epoch": 3.8484823693037717, "grad_norm": 0.0039770156145095825, "learning_rate": 4.364437344163974e-08, "loss": 0.0005, "step": 235200 }, { "epoch": 3.8486459952548473, "grad_norm": 0.010036502964794636, "learning_rate": 4.355029704130209e-08, "loss": 0.0003, "step": 235210 }, { "epoch": 3.8488096212059233, "grad_norm": 0.05592415854334831, "learning_rate": 4.3456321698758244e-08, "loss": 0.0008, "step": 235220 }, { "epoch": 3.848973247156999, "grad_norm": 0.0810285434126854, "learning_rate": 4.3362447415925545e-08, "loss": 0.0001, "step": 235230 }, { "epoch": 3.849136873108075, "grad_norm": 0.0008189269574359059, "learning_rate": 4.326867419471637e-08, "loss": 0.001, "step": 235240 }, { "epoch": 3.849300499059151, "grad_norm": 0.0010654411744326353, "learning_rate": 4.3175002037043055e-08, "loss": 0.0003, "step": 235250 }, { "epoch": 3.8494641250102264, "grad_norm": 0.0203153807669878, "learning_rate": 4.3081430944816315e-08, "loss": 0.0002, "step": 235260 }, { "epoch": 3.8496277509613024, "grad_norm": 0.07666973024606705, "learning_rate": 4.2987960919943507e-08, "loss": 0.0002, "step": 235270 }, { "epoch": 3.8497913769123784, "grad_norm": 0.04097326099872589, "learning_rate": 4.289459196433032e-08, "loss": 0.0003, "step": 235280 }, { "epoch": 3.849955002863454, "grad_norm": 0.034386828541755676, "learning_rate": 4.280132407988025e-08, "loss": 0.0002, "step": 235290 }, { "epoch": 3.85011862881453, "grad_norm": 2.8725638912874274e-05, "learning_rate": 4.270815726849564e-08, "loss": 0.0002, "step": 235300 }, { "epoch": 3.850282254765606, "grad_norm": 0.005414658226072788, "learning_rate": 4.2615091532075545e-08, "loss": 0.0003, "step": 235310 }, { "epoch": 3.8504458807166815, "grad_norm": 0.01443235483020544, "learning_rate": 4.252212687251789e-08, "loss": 0.0002, "step": 235320 }, { "epoch": 3.8506095066677575, "grad_norm": 0.050723183900117874, "learning_rate": 4.242926329171726e-08, "loss": 0.0007, "step": 235330 }, { "epoch": 3.8507731326188335, "grad_norm": 0.01935432106256485, "learning_rate": 4.233650079156826e-08, "loss": 0.0001, "step": 235340 }, { "epoch": 3.850936758569909, "grad_norm": 0.0008856425993144512, "learning_rate": 4.2243839373961594e-08, "loss": 0.0002, "step": 235350 }, { "epoch": 3.851100384520985, "grad_norm": 0.04132036119699478, "learning_rate": 4.215127904078631e-08, "loss": 0.0004, "step": 235360 }, { "epoch": 3.851264010472061, "grad_norm": 0.06263656169176102, "learning_rate": 4.205881979393034e-08, "loss": 0.0003, "step": 235370 }, { "epoch": 3.8514276364231366, "grad_norm": 0.035442251712083817, "learning_rate": 4.196646163527773e-08, "loss": 0.0003, "step": 235380 }, { "epoch": 3.8515912623742126, "grad_norm": 0.0647418424487114, "learning_rate": 4.18742045667131e-08, "loss": 0.0002, "step": 235390 }, { "epoch": 3.8517548883252886, "grad_norm": 0.03853793814778328, "learning_rate": 4.1782048590116583e-08, "loss": 0.0003, "step": 235400 }, { "epoch": 3.851918514276364, "grad_norm": 0.04617507383227348, "learning_rate": 4.16899937073667e-08, "loss": 0.0002, "step": 235410 }, { "epoch": 3.85208214022744, "grad_norm": 0.03439110890030861, "learning_rate": 4.1598039920341394e-08, "loss": 0.0002, "step": 235420 }, { "epoch": 3.852245766178516, "grad_norm": 0.021791063249111176, "learning_rate": 4.150618723091526e-08, "loss": 0.0001, "step": 235430 }, { "epoch": 3.8524093921295917, "grad_norm": 0.01876239851117134, "learning_rate": 4.14144356409607e-08, "loss": 0.0003, "step": 235440 }, { "epoch": 3.8525730180806677, "grad_norm": 0.07372787594795227, "learning_rate": 4.1322785152348444e-08, "loss": 0.0004, "step": 235450 }, { "epoch": 3.8527366440317437, "grad_norm": 0.06917180866003036, "learning_rate": 4.123123576694754e-08, "loss": 0.0003, "step": 235460 }, { "epoch": 3.852900269982819, "grad_norm": 0.10864042490720749, "learning_rate": 4.113978748662484e-08, "loss": 0.0003, "step": 235470 }, { "epoch": 3.853063895933895, "grad_norm": 0.07449523359537125, "learning_rate": 4.10484403132444e-08, "loss": 0.0001, "step": 235480 }, { "epoch": 3.853227521884971, "grad_norm": 0.0001268977503059432, "learning_rate": 4.095719424866862e-08, "loss": 0.0003, "step": 235490 }, { "epoch": 3.8533911478360467, "grad_norm": 0.006286648567765951, "learning_rate": 4.086604929475824e-08, "loss": 0.0001, "step": 235500 }, { "epoch": 3.8535547737871227, "grad_norm": 0.02885100245475769, "learning_rate": 4.0775005453371206e-08, "loss": 0.0003, "step": 235510 }, { "epoch": 3.8537183997381987, "grad_norm": 0.02231215313076973, "learning_rate": 4.0684062726364384e-08, "loss": 0.0006, "step": 235520 }, { "epoch": 3.8538820256892743, "grad_norm": 0.040248073637485504, "learning_rate": 4.059322111559183e-08, "loss": 0.0002, "step": 235530 }, { "epoch": 3.8540456516403503, "grad_norm": 0.022857096046209335, "learning_rate": 4.0502480622905404e-08, "loss": 0.0002, "step": 235540 }, { "epoch": 3.854209277591426, "grad_norm": 0.00819538813084364, "learning_rate": 4.0411841250155845e-08, "loss": 0.0002, "step": 235550 }, { "epoch": 3.854372903542502, "grad_norm": 0.012035451829433441, "learning_rate": 4.0321302999191125e-08, "loss": 0.0003, "step": 235560 }, { "epoch": 3.854536529493578, "grad_norm": 0.05425310507416725, "learning_rate": 4.023086587185643e-08, "loss": 0.0004, "step": 235570 }, { "epoch": 3.8547001554446534, "grad_norm": 0.02693183533847332, "learning_rate": 4.0140529869996395e-08, "loss": 0.0001, "step": 235580 }, { "epoch": 3.8548637813957294, "grad_norm": 0.00772582134231925, "learning_rate": 4.005029499545288e-08, "loss": 0.0002, "step": 235590 }, { "epoch": 3.855027407346805, "grad_norm": 0.08602418005466461, "learning_rate": 3.9960161250065524e-08, "loss": 0.0002, "step": 235600 }, { "epoch": 3.855191033297881, "grad_norm": 0.11572452634572983, "learning_rate": 3.9870128635671744e-08, "loss": 0.0005, "step": 235610 }, { "epoch": 3.855354659248957, "grad_norm": 0.03286358714103699, "learning_rate": 3.978019715410786e-08, "loss": 0.0005, "step": 235620 }, { "epoch": 3.8555182852000325, "grad_norm": 0.0026614742819219828, "learning_rate": 3.9690366807207395e-08, "loss": 0.0002, "step": 235630 }, { "epoch": 3.8556819111511085, "grad_norm": 0.008025290444493294, "learning_rate": 3.960063759680166e-08, "loss": 0.0003, "step": 235640 }, { "epoch": 3.8558455371021845, "grad_norm": 0.0705045536160469, "learning_rate": 3.951100952472031e-08, "loss": 0.0002, "step": 235650 }, { "epoch": 3.85600916305326, "grad_norm": 0.05324886739253998, "learning_rate": 3.942148259279022e-08, "loss": 0.0002, "step": 235660 }, { "epoch": 3.856172789004336, "grad_norm": 0.022331304848194122, "learning_rate": 3.9332056802838245e-08, "loss": 0.0003, "step": 235670 }, { "epoch": 3.856336414955412, "grad_norm": 0.024776197969913483, "learning_rate": 3.924273215668628e-08, "loss": 0.0002, "step": 235680 }, { "epoch": 3.8565000409064876, "grad_norm": 0.0517563670873642, "learning_rate": 3.915350865615564e-08, "loss": 0.0005, "step": 235690 }, { "epoch": 3.8566636668575636, "grad_norm": 0.0024044285528361797, "learning_rate": 3.906438630306597e-08, "loss": 0.0003, "step": 235700 }, { "epoch": 3.8568272928086396, "grad_norm": 0.01995321549475193, "learning_rate": 3.897536509923472e-08, "loss": 0.0003, "step": 235710 }, { "epoch": 3.856990918759715, "grad_norm": 0.0031188945285975933, "learning_rate": 3.888644504647654e-08, "loss": 0.0002, "step": 235720 }, { "epoch": 3.857154544710791, "grad_norm": 0.06458479166030884, "learning_rate": 3.879762614660443e-08, "loss": 0.0003, "step": 235730 }, { "epoch": 3.857318170661867, "grad_norm": 0.003568336134776473, "learning_rate": 3.870890840142916e-08, "loss": 0.0002, "step": 235740 }, { "epoch": 3.8574817966129427, "grad_norm": 0.04300002381205559, "learning_rate": 3.8620291812760394e-08, "loss": 0.0003, "step": 235750 }, { "epoch": 3.8576454225640187, "grad_norm": 0.002678551245480776, "learning_rate": 3.853177638240391e-08, "loss": 0.0006, "step": 235760 }, { "epoch": 3.8578090485150947, "grad_norm": 0.015124070458114147, "learning_rate": 3.844336211216493e-08, "loss": 0.0005, "step": 235770 }, { "epoch": 3.85797267446617, "grad_norm": 0.014979032799601555, "learning_rate": 3.83550490038459e-08, "loss": 0.0003, "step": 235780 }, { "epoch": 3.858136300417246, "grad_norm": 0.004006974399089813, "learning_rate": 3.826683705924816e-08, "loss": 0.0003, "step": 235790 }, { "epoch": 3.858299926368322, "grad_norm": 0.00521910609677434, "learning_rate": 3.817872628016972e-08, "loss": 0.0001, "step": 235800 }, { "epoch": 3.8584635523193977, "grad_norm": 0.013236995786428452, "learning_rate": 3.8090716668407465e-08, "loss": 0.0003, "step": 235810 }, { "epoch": 3.8586271782704737, "grad_norm": 0.07156693190336227, "learning_rate": 3.800280822575497e-08, "loss": 0.0004, "step": 235820 }, { "epoch": 3.8587908042215497, "grad_norm": 0.0006514648557640612, "learning_rate": 3.791500095400524e-08, "loss": 0.0002, "step": 235830 }, { "epoch": 3.8589544301726253, "grad_norm": 0.0293620266020298, "learning_rate": 3.7827294854948516e-08, "loss": 0.0002, "step": 235840 }, { "epoch": 3.8591180561237013, "grad_norm": 0.0009675708133727312, "learning_rate": 3.7739689930373355e-08, "loss": 0.0003, "step": 235850 }, { "epoch": 3.8592816820747773, "grad_norm": 0.002930242568254471, "learning_rate": 3.765218618206556e-08, "loss": 0.0001, "step": 235860 }, { "epoch": 3.859445308025853, "grad_norm": 0.0043029580265283585, "learning_rate": 3.7564783611809264e-08, "loss": 0.0011, "step": 235870 }, { "epoch": 3.859608933976929, "grad_norm": 0.0010626042494550347, "learning_rate": 3.747748222138692e-08, "loss": 0.0003, "step": 235880 }, { "epoch": 3.859772559928005, "grad_norm": 0.07161464542150497, "learning_rate": 3.739028201257822e-08, "loss": 0.0002, "step": 235890 }, { "epoch": 3.8599361858790804, "grad_norm": 0.013599840924143791, "learning_rate": 3.7303182987160624e-08, "loss": 0.0001, "step": 235900 }, { "epoch": 3.8600998118301564, "grad_norm": 0.024762572720646858, "learning_rate": 3.721618514691105e-08, "loss": 0.0002, "step": 235910 }, { "epoch": 3.8602634377812324, "grad_norm": 0.06961410492658615, "learning_rate": 3.7129288493602534e-08, "loss": 0.0002, "step": 235920 }, { "epoch": 3.860427063732308, "grad_norm": 0.0062926094979047775, "learning_rate": 3.704249302900698e-08, "loss": 0.0003, "step": 235930 }, { "epoch": 3.860590689683384, "grad_norm": 0.004938486032187939, "learning_rate": 3.695579875489408e-08, "loss": 0.0002, "step": 235940 }, { "epoch": 3.8607543156344595, "grad_norm": 0.009158717468380928, "learning_rate": 3.686920567303187e-08, "loss": 0.0003, "step": 235950 }, { "epoch": 3.8609179415855355, "grad_norm": 0.00035313187981955707, "learning_rate": 3.6782713785185054e-08, "loss": 0.0001, "step": 235960 }, { "epoch": 3.8610815675366115, "grad_norm": 0.026064902544021606, "learning_rate": 3.669632309311832e-08, "loss": 0.0003, "step": 235970 }, { "epoch": 3.861245193487687, "grad_norm": 0.006704560946673155, "learning_rate": 3.6610033598591946e-08, "loss": 0.0002, "step": 235980 }, { "epoch": 3.861408819438763, "grad_norm": 0.0012308645527809858, "learning_rate": 3.652384530336617e-08, "loss": 0.0003, "step": 235990 }, { "epoch": 3.8615724453898386, "grad_norm": 0.03227057307958603, "learning_rate": 3.643775820919737e-08, "loss": 0.0002, "step": 236000 }, { "epoch": 3.8617360713409146, "grad_norm": 0.000269075739197433, "learning_rate": 3.635177231784193e-08, "loss": 0.0002, "step": 236010 }, { "epoch": 3.8618996972919906, "grad_norm": 0.0004998752847313881, "learning_rate": 3.6265887631051764e-08, "loss": 0.0001, "step": 236020 }, { "epoch": 3.862063323243066, "grad_norm": 0.002489787060767412, "learning_rate": 3.618010415057882e-08, "loss": 0.0005, "step": 236030 }, { "epoch": 3.862226949194142, "grad_norm": 0.0010596559150144458, "learning_rate": 3.6094421878172245e-08, "loss": 0.0001, "step": 236040 }, { "epoch": 3.862390575145218, "grad_norm": 0.03692234307527542, "learning_rate": 3.600884081557843e-08, "loss": 0.0004, "step": 236050 }, { "epoch": 3.8625542010962937, "grad_norm": 0.030153216794133186, "learning_rate": 3.5923360964543184e-08, "loss": 0.0002, "step": 236060 }, { "epoch": 3.8627178270473697, "grad_norm": 0.16813015937805176, "learning_rate": 3.583798232680791e-08, "loss": 0.0003, "step": 236070 }, { "epoch": 3.8628814529984457, "grad_norm": 0.00355535838752985, "learning_rate": 3.575270490411509e-08, "loss": 0.0002, "step": 236080 }, { "epoch": 3.863045078949521, "grad_norm": 0.005843442864716053, "learning_rate": 3.5667528698202225e-08, "loss": 0.0001, "step": 236090 }, { "epoch": 3.863208704900597, "grad_norm": 0.007735988590866327, "learning_rate": 3.558245371080682e-08, "loss": 0.0002, "step": 236100 }, { "epoch": 3.863372330851673, "grad_norm": 0.0766625851392746, "learning_rate": 3.5497479943663036e-08, "loss": 0.0003, "step": 236110 }, { "epoch": 3.8635359568027487, "grad_norm": 0.03454893082380295, "learning_rate": 3.5412607398503384e-08, "loss": 0.0003, "step": 236120 }, { "epoch": 3.8636995827538247, "grad_norm": 0.02686571143567562, "learning_rate": 3.5327836077058145e-08, "loss": 0.0002, "step": 236130 }, { "epoch": 3.8638632087049007, "grad_norm": 0.013332336209714413, "learning_rate": 3.524316598105648e-08, "loss": 0.0002, "step": 236140 }, { "epoch": 3.8640268346559763, "grad_norm": 0.025238430127501488, "learning_rate": 3.515859711222369e-08, "loss": 0.0002, "step": 236150 }, { "epoch": 3.8641904606070523, "grad_norm": 0.013445364311337471, "learning_rate": 3.5074129472285056e-08, "loss": 0.0002, "step": 236160 }, { "epoch": 3.8643540865581283, "grad_norm": 0.03613954409956932, "learning_rate": 3.498976306296253e-08, "loss": 0.0005, "step": 236170 }, { "epoch": 3.864517712509204, "grad_norm": 0.004401141777634621, "learning_rate": 3.4905497885976415e-08, "loss": 0.0001, "step": 236180 }, { "epoch": 3.86468133846028, "grad_norm": 0.0036129646468907595, "learning_rate": 3.482133394304366e-08, "loss": 0.0002, "step": 236190 }, { "epoch": 3.864844964411356, "grad_norm": 0.0052380007691681385, "learning_rate": 3.4737271235881774e-08, "loss": 0.0002, "step": 236200 }, { "epoch": 3.8650085903624314, "grad_norm": 0.0021429513581097126, "learning_rate": 3.46533097662044e-08, "loss": 0.0001, "step": 236210 }, { "epoch": 3.8651722163135074, "grad_norm": 0.02430974319577217, "learning_rate": 3.456944953572239e-08, "loss": 0.0004, "step": 236220 }, { "epoch": 3.8653358422645834, "grad_norm": 0.01897863857448101, "learning_rate": 3.448569054614659e-08, "loss": 0.0001, "step": 236230 }, { "epoch": 3.865499468215659, "grad_norm": 0.04013795405626297, "learning_rate": 3.440203279918508e-08, "loss": 0.0004, "step": 236240 }, { "epoch": 3.865663094166735, "grad_norm": 0.0004324852197896689, "learning_rate": 3.431847629654261e-08, "loss": 0.0002, "step": 236250 }, { "epoch": 3.865826720117811, "grad_norm": 0.048673588782548904, "learning_rate": 3.4235021039923375e-08, "loss": 0.0004, "step": 236260 }, { "epoch": 3.8659903460688865, "grad_norm": 0.051649387925863266, "learning_rate": 3.415166703102879e-08, "loss": 0.0003, "step": 236270 }, { "epoch": 3.8661539720199625, "grad_norm": 0.035008326172828674, "learning_rate": 3.4068414271558046e-08, "loss": 0.0003, "step": 236280 }, { "epoch": 3.8663175979710385, "grad_norm": 0.055218588560819626, "learning_rate": 3.398526276320924e-08, "loss": 0.0003, "step": 236290 }, { "epoch": 3.866481223922114, "grad_norm": 0.026182042434811592, "learning_rate": 3.390221250767767e-08, "loss": 0.0001, "step": 236300 }, { "epoch": 3.86664484987319, "grad_norm": 0.08108111470937729, "learning_rate": 3.381926350665643e-08, "loss": 0.0004, "step": 236310 }, { "epoch": 3.8668084758242656, "grad_norm": 0.007323147729039192, "learning_rate": 3.3736415761836397e-08, "loss": 0.0001, "step": 236320 }, { "epoch": 3.8669721017753416, "grad_norm": 0.021504903212189674, "learning_rate": 3.3653669274907873e-08, "loss": 0.0001, "step": 236330 }, { "epoch": 3.8671357277264176, "grad_norm": 0.019717087969183922, "learning_rate": 3.357102404755674e-08, "loss": 0.0002, "step": 236340 }, { "epoch": 3.867299353677493, "grad_norm": 0.03959518298506737, "learning_rate": 3.3488480081468876e-08, "loss": 0.0003, "step": 236350 }, { "epoch": 3.867462979628569, "grad_norm": 0.03926029056310654, "learning_rate": 3.340603737832682e-08, "loss": 0.0018, "step": 236360 }, { "epoch": 3.8676266055796447, "grad_norm": 0.006584597751498222, "learning_rate": 3.332369593981199e-08, "loss": 0.0002, "step": 236370 }, { "epoch": 3.8677902315307207, "grad_norm": 0.03900545835494995, "learning_rate": 3.324145576760307e-08, "loss": 0.0002, "step": 236380 }, { "epoch": 3.8679538574817967, "grad_norm": 0.08772199600934982, "learning_rate": 3.3159316863376475e-08, "loss": 0.0007, "step": 236390 }, { "epoch": 3.868117483432872, "grad_norm": 0.05244597792625427, "learning_rate": 3.307727922880699e-08, "loss": 0.0003, "step": 236400 }, { "epoch": 3.868281109383948, "grad_norm": 0.006861210335046053, "learning_rate": 3.2995342865568267e-08, "loss": 0.0002, "step": 236410 }, { "epoch": 3.868444735335024, "grad_norm": 0.05078267678618431, "learning_rate": 3.291350777533009e-08, "loss": 0.0003, "step": 236420 }, { "epoch": 3.8686083612860998, "grad_norm": 0.028180787339806557, "learning_rate": 3.283177395976112e-08, "loss": 0.0004, "step": 236430 }, { "epoch": 3.8687719872371757, "grad_norm": 0.002296655671671033, "learning_rate": 3.2750141420527236e-08, "loss": 0.0001, "step": 236440 }, { "epoch": 3.8689356131882517, "grad_norm": 0.01580778695642948, "learning_rate": 3.266861015929435e-08, "loss": 0.0004, "step": 236450 }, { "epoch": 3.8690992391393273, "grad_norm": 0.013571570627391338, "learning_rate": 3.258718017772333e-08, "loss": 0.0002, "step": 236460 }, { "epoch": 3.8692628650904033, "grad_norm": 0.03607358783483505, "learning_rate": 3.2505851477475095e-08, "loss": 0.0005, "step": 236470 }, { "epoch": 3.8694264910414793, "grad_norm": 0.04723463952541351, "learning_rate": 3.24246240602083e-08, "loss": 0.0002, "step": 236480 }, { "epoch": 3.869590116992555, "grad_norm": 0.0002808806602843106, "learning_rate": 3.234349792757829e-08, "loss": 0.0003, "step": 236490 }, { "epoch": 3.869753742943631, "grad_norm": 0.012973297387361526, "learning_rate": 3.226247308123931e-08, "loss": 0.0, "step": 236500 }, { "epoch": 3.869917368894707, "grad_norm": 0.024242766201496124, "learning_rate": 3.218154952284391e-08, "loss": 0.0002, "step": 236510 }, { "epoch": 3.8700809948457824, "grad_norm": 0.0011817002668976784, "learning_rate": 3.2100727254041896e-08, "loss": 0.0002, "step": 236520 }, { "epoch": 3.8702446207968584, "grad_norm": 0.025617387145757675, "learning_rate": 3.202000627648083e-08, "loss": 0.0001, "step": 236530 }, { "epoch": 3.8704082467479344, "grad_norm": 0.026310821995139122, "learning_rate": 3.1939386591806624e-08, "loss": 0.0002, "step": 236540 }, { "epoch": 3.87057187269901, "grad_norm": 0.19146262109279633, "learning_rate": 3.185886820166351e-08, "loss": 0.0022, "step": 236550 }, { "epoch": 3.870735498650086, "grad_norm": 0.01676228642463684, "learning_rate": 3.17784511076924e-08, "loss": 0.0006, "step": 236560 }, { "epoch": 3.870899124601162, "grad_norm": 0.0006320069660432637, "learning_rate": 3.16981353115331e-08, "loss": 0.0002, "step": 236570 }, { "epoch": 3.8710627505522375, "grad_norm": 0.011573409661650658, "learning_rate": 3.161792081482429e-08, "loss": 0.0002, "step": 236580 }, { "epoch": 3.8712263765033135, "grad_norm": 0.0744239017367363, "learning_rate": 3.1537807619200224e-08, "loss": 0.0004, "step": 236590 }, { "epoch": 3.8713900024543895, "grad_norm": 0.0852121040225029, "learning_rate": 3.1457795726294595e-08, "loss": 0.0003, "step": 236600 }, { "epoch": 3.871553628405465, "grad_norm": 0.07447576522827148, "learning_rate": 3.137788513773887e-08, "loss": 0.0002, "step": 236610 }, { "epoch": 3.871717254356541, "grad_norm": 0.015139587223529816, "learning_rate": 3.1298075855162314e-08, "loss": 0.0002, "step": 236620 }, { "epoch": 3.871880880307617, "grad_norm": 0.011670967563986778, "learning_rate": 3.1218367880192504e-08, "loss": 0.0002, "step": 236630 }, { "epoch": 3.8720445062586926, "grad_norm": 0.03979293256998062, "learning_rate": 3.113876121445425e-08, "loss": 0.0002, "step": 236640 }, { "epoch": 3.8722081322097686, "grad_norm": 0.0006278313812799752, "learning_rate": 3.10592558595707e-08, "loss": 0.0001, "step": 236650 }, { "epoch": 3.8723717581608446, "grad_norm": 0.02677827887237072, "learning_rate": 3.097985181716334e-08, "loss": 0.0002, "step": 236660 }, { "epoch": 3.87253538411192, "grad_norm": 0.05600820854306221, "learning_rate": 3.0900549088850876e-08, "loss": 0.0003, "step": 236670 }, { "epoch": 3.872699010062996, "grad_norm": 0.10574330389499664, "learning_rate": 3.082134767624978e-08, "loss": 0.0005, "step": 236680 }, { "epoch": 3.872862636014072, "grad_norm": 0.012006794102489948, "learning_rate": 3.074224758097544e-08, "loss": 0.0002, "step": 236690 }, { "epoch": 3.8730262619651477, "grad_norm": 0.00022476541926153004, "learning_rate": 3.0663248804640445e-08, "loss": 0.0003, "step": 236700 }, { "epoch": 3.8731898879162237, "grad_norm": 0.002615094417706132, "learning_rate": 3.0584351348855734e-08, "loss": 0.0002, "step": 236710 }, { "epoch": 3.873353513867299, "grad_norm": 0.001133812591433525, "learning_rate": 3.0505555215229466e-08, "loss": 0.0002, "step": 236720 }, { "epoch": 3.873517139818375, "grad_norm": 0.022808998823165894, "learning_rate": 3.042686040536813e-08, "loss": 0.0002, "step": 236730 }, { "epoch": 3.873680765769451, "grad_norm": 0.010014674626290798, "learning_rate": 3.034826692087767e-08, "loss": 0.0005, "step": 236740 }, { "epoch": 3.8738443917205267, "grad_norm": 0.015810266137123108, "learning_rate": 3.026977476335846e-08, "loss": 0.0001, "step": 236750 }, { "epoch": 3.8740080176716027, "grad_norm": 0.11161427944898605, "learning_rate": 3.0191383934412563e-08, "loss": 0.0006, "step": 236760 }, { "epoch": 3.8741716436226783, "grad_norm": 0.05398492142558098, "learning_rate": 3.011309443563759e-08, "loss": 0.0001, "step": 236770 }, { "epoch": 3.8743352695737543, "grad_norm": 0.02633439563214779, "learning_rate": 3.0034906268630034e-08, "loss": 0.0001, "step": 236780 }, { "epoch": 3.8744988955248303, "grad_norm": 0.021832449361681938, "learning_rate": 2.995681943498363e-08, "loss": 0.0001, "step": 236790 }, { "epoch": 3.874662521475906, "grad_norm": 0.05446575582027435, "learning_rate": 2.9878833936290985e-08, "loss": 0.0003, "step": 236800 }, { "epoch": 3.874826147426982, "grad_norm": 0.0038950666785240173, "learning_rate": 2.9800949774141385e-08, "loss": 0.0005, "step": 236810 }, { "epoch": 3.874989773378058, "grad_norm": 0.035145070403814316, "learning_rate": 2.9723166950124117e-08, "loss": 0.0003, "step": 236820 }, { "epoch": 3.8751533993291334, "grad_norm": 0.040675438940525055, "learning_rate": 2.9645485465824576e-08, "loss": 0.0002, "step": 236830 }, { "epoch": 3.8753170252802094, "grad_norm": 0.016346236690878868, "learning_rate": 2.956790532282594e-08, "loss": 0.0002, "step": 236840 }, { "epoch": 3.8754806512312854, "grad_norm": 0.017448125407099724, "learning_rate": 2.949042652271028e-08, "loss": 0.0003, "step": 236850 }, { "epoch": 3.875644277182361, "grad_norm": 0.07257537543773651, "learning_rate": 2.9413049067058552e-08, "loss": 0.0004, "step": 236860 }, { "epoch": 3.875807903133437, "grad_norm": 0.04807526245713234, "learning_rate": 2.9335772957446718e-08, "loss": 0.0004, "step": 236870 }, { "epoch": 3.875971529084513, "grad_norm": 0.0071371193043887615, "learning_rate": 2.9258598195451293e-08, "loss": 0.0001, "step": 236880 }, { "epoch": 3.8761351550355885, "grad_norm": 0.05407977104187012, "learning_rate": 2.918152478264491e-08, "loss": 0.0005, "step": 236890 }, { "epoch": 3.8762987809866645, "grad_norm": 0.015359394252300262, "learning_rate": 2.9104552720600754e-08, "loss": 0.0003, "step": 236900 }, { "epoch": 3.8764624069377405, "grad_norm": 0.007969011552631855, "learning_rate": 2.902768201088646e-08, "loss": 0.0001, "step": 236910 }, { "epoch": 3.876626032888816, "grad_norm": 0.0001997453364310786, "learning_rate": 2.895091265507022e-08, "loss": 0.0002, "step": 236920 }, { "epoch": 3.876789658839892, "grad_norm": 0.007644559722393751, "learning_rate": 2.887424465471744e-08, "loss": 0.0002, "step": 236930 }, { "epoch": 3.876953284790968, "grad_norm": 0.012687585316598415, "learning_rate": 2.8797678011390774e-08, "loss": 0.0001, "step": 236940 }, { "epoch": 3.8771169107420436, "grad_norm": 0.04639498144388199, "learning_rate": 2.8721212726651183e-08, "loss": 0.0002, "step": 236950 }, { "epoch": 3.8772805366931196, "grad_norm": 0.008712598122656345, "learning_rate": 2.8644848802059088e-08, "loss": 0.0003, "step": 236960 }, { "epoch": 3.8774441626441956, "grad_norm": 0.005726561415940523, "learning_rate": 2.8568586239169917e-08, "loss": 0.0001, "step": 236970 }, { "epoch": 3.877607788595271, "grad_norm": 0.015498843044042587, "learning_rate": 2.8492425039539085e-08, "loss": 0.0002, "step": 236980 }, { "epoch": 3.877771414546347, "grad_norm": 0.019977588206529617, "learning_rate": 2.84163652047198e-08, "loss": 0.0003, "step": 236990 }, { "epoch": 3.877935040497423, "grad_norm": 0.01808888465166092, "learning_rate": 2.834040673626304e-08, "loss": 0.0003, "step": 237000 }, { "epoch": 3.8780986664484987, "grad_norm": 0.015248188748955727, "learning_rate": 2.8264549635716455e-08, "loss": 0.0003, "step": 237010 }, { "epoch": 3.8782622923995747, "grad_norm": 0.031686194241046906, "learning_rate": 2.81887939046277e-08, "loss": 0.0003, "step": 237020 }, { "epoch": 3.8784259183506506, "grad_norm": 0.03288871794939041, "learning_rate": 2.811313954454109e-08, "loss": 0.0002, "step": 237030 }, { "epoch": 3.878589544301726, "grad_norm": 0.004963105544447899, "learning_rate": 2.8037586556999287e-08, "loss": 0.0002, "step": 237040 }, { "epoch": 3.878753170252802, "grad_norm": 0.09386824816465378, "learning_rate": 2.796213494354272e-08, "loss": 0.0005, "step": 237050 }, { "epoch": 3.878916796203878, "grad_norm": 0.021510105580091476, "learning_rate": 2.7886784705709048e-08, "loss": 0.0004, "step": 237060 }, { "epoch": 3.8790804221549537, "grad_norm": 0.0087575139477849, "learning_rate": 2.7811535845035374e-08, "loss": 0.0002, "step": 237070 }, { "epoch": 3.8792440481060297, "grad_norm": 0.0010757872369140387, "learning_rate": 2.7736388363056588e-08, "loss": 0.0002, "step": 237080 }, { "epoch": 3.8794076740571053, "grad_norm": 0.017424171790480614, "learning_rate": 2.7661342261303125e-08, "loss": 0.0003, "step": 237090 }, { "epoch": 3.8795713000081813, "grad_norm": 0.0659777969121933, "learning_rate": 2.7586397541306543e-08, "loss": 0.0003, "step": 237100 }, { "epoch": 3.8797349259592573, "grad_norm": 0.035705044865608215, "learning_rate": 2.7511554204593948e-08, "loss": 0.0002, "step": 237110 }, { "epoch": 3.879898551910333, "grad_norm": 0.003506164764985442, "learning_rate": 2.7436812252692456e-08, "loss": 0.0016, "step": 237120 }, { "epoch": 3.880062177861409, "grad_norm": 0.018691401928663254, "learning_rate": 2.736217168712474e-08, "loss": 0.0002, "step": 237130 }, { "epoch": 3.8802258038124844, "grad_norm": 0.05218884348869324, "learning_rate": 2.7287632509413464e-08, "loss": 0.0003, "step": 237140 }, { "epoch": 3.8803894297635604, "grad_norm": 0.01262248307466507, "learning_rate": 2.721319472107853e-08, "loss": 0.0003, "step": 237150 }, { "epoch": 3.8805530557146364, "grad_norm": 0.0005690624238923192, "learning_rate": 2.7138858323637052e-08, "loss": 0.0004, "step": 237160 }, { "epoch": 3.880716681665712, "grad_norm": 0.017142439261078835, "learning_rate": 2.7064623318605044e-08, "loss": 0.0001, "step": 237170 }, { "epoch": 3.880880307616788, "grad_norm": 0.03781977668404579, "learning_rate": 2.6990489707496292e-08, "loss": 0.0002, "step": 237180 }, { "epoch": 3.881043933567864, "grad_norm": 0.004210958257317543, "learning_rate": 2.6916457491821814e-08, "loss": 0.0003, "step": 237190 }, { "epoch": 3.8812075595189395, "grad_norm": 0.0009826213354244828, "learning_rate": 2.6842526673091508e-08, "loss": 0.0002, "step": 237200 }, { "epoch": 3.8813711854700155, "grad_norm": 0.029912205412983894, "learning_rate": 2.6768697252811948e-08, "loss": 0.0003, "step": 237210 }, { "epoch": 3.8815348114210915, "grad_norm": 0.018680380657315254, "learning_rate": 2.6694969232489708e-08, "loss": 0.0002, "step": 237220 }, { "epoch": 3.881698437372167, "grad_norm": 0.08431864529848099, "learning_rate": 2.662134261362692e-08, "loss": 0.0003, "step": 237230 }, { "epoch": 3.881862063323243, "grad_norm": 0.008182181976735592, "learning_rate": 2.654781739772572e-08, "loss": 0.0003, "step": 237240 }, { "epoch": 3.882025689274319, "grad_norm": 0.015386118553578854, "learning_rate": 2.6474393586284342e-08, "loss": 0.0002, "step": 237250 }, { "epoch": 3.8821893152253946, "grad_norm": 0.0014927511801943183, "learning_rate": 2.6401071180800485e-08, "loss": 0.0002, "step": 237260 }, { "epoch": 3.8823529411764706, "grad_norm": 0.0042606014758348465, "learning_rate": 2.6327850182769065e-08, "loss": 0.0, "step": 237270 }, { "epoch": 3.8825165671275466, "grad_norm": 0.025195572525262833, "learning_rate": 2.6254730593682775e-08, "loss": 0.0004, "step": 237280 }, { "epoch": 3.882680193078622, "grad_norm": 0.0574348010122776, "learning_rate": 2.618171241503209e-08, "loss": 0.0005, "step": 237290 }, { "epoch": 3.882843819029698, "grad_norm": 0.0005901502445340157, "learning_rate": 2.6108795648306372e-08, "loss": 0.0001, "step": 237300 }, { "epoch": 3.883007444980774, "grad_norm": 0.044227235019207, "learning_rate": 2.6035980294992213e-08, "loss": 0.0003, "step": 237310 }, { "epoch": 3.8831710709318497, "grad_norm": 0.02961721643805504, "learning_rate": 2.5963266356574534e-08, "loss": 0.0004, "step": 237320 }, { "epoch": 3.8833346968829257, "grad_norm": 0.030391011387109756, "learning_rate": 2.5890653834535484e-08, "loss": 0.0003, "step": 237330 }, { "epoch": 3.8834983228340016, "grad_norm": 0.000365789863280952, "learning_rate": 2.5818142730355544e-08, "loss": 0.0001, "step": 237340 }, { "epoch": 3.883661948785077, "grad_norm": 0.013431197963654995, "learning_rate": 2.5745733045512978e-08, "loss": 0.0002, "step": 237350 }, { "epoch": 3.883825574736153, "grad_norm": 0.00928861927241087, "learning_rate": 2.5673424781484936e-08, "loss": 0.0002, "step": 237360 }, { "epoch": 3.883989200687229, "grad_norm": 0.0467151403427124, "learning_rate": 2.5601217939745238e-08, "loss": 0.0001, "step": 237370 }, { "epoch": 3.8841528266383047, "grad_norm": 0.031907301396131516, "learning_rate": 2.552911252176604e-08, "loss": 0.0001, "step": 237380 }, { "epoch": 3.8843164525893807, "grad_norm": 0.011904630810022354, "learning_rate": 2.545710852901728e-08, "loss": 0.0002, "step": 237390 }, { "epoch": 3.8844800785404567, "grad_norm": 0.02873574011027813, "learning_rate": 2.5385205962967784e-08, "loss": 0.0001, "step": 237400 }, { "epoch": 3.8846437044915323, "grad_norm": 2.8114920496591367e-05, "learning_rate": 2.5313404825082488e-08, "loss": 0.0003, "step": 237410 }, { "epoch": 3.8848073304426083, "grad_norm": 0.0017985956510528922, "learning_rate": 2.5241705116826887e-08, "loss": 0.0002, "step": 237420 }, { "epoch": 3.8849709563936843, "grad_norm": 0.05127028003334999, "learning_rate": 2.5170106839661478e-08, "loss": 0.0004, "step": 237430 }, { "epoch": 3.88513458234476, "grad_norm": 0.002112203510478139, "learning_rate": 2.509860999504732e-08, "loss": 0.0001, "step": 237440 }, { "epoch": 3.885298208295836, "grad_norm": 0.051695916801691055, "learning_rate": 2.5027214584441016e-08, "loss": 0.0001, "step": 237450 }, { "epoch": 3.885461834246912, "grad_norm": 0.029934056103229523, "learning_rate": 2.4955920609298635e-08, "loss": 0.0003, "step": 237460 }, { "epoch": 3.8856254601979874, "grad_norm": 0.17300748825073242, "learning_rate": 2.4884728071074005e-08, "loss": 0.0006, "step": 237470 }, { "epoch": 3.8857890861490634, "grad_norm": 0.07183337211608887, "learning_rate": 2.4813636971218747e-08, "loss": 0.0003, "step": 237480 }, { "epoch": 3.885952712100139, "grad_norm": 0.07222805172204971, "learning_rate": 2.4742647311182255e-08, "loss": 0.0005, "step": 237490 }, { "epoch": 3.886116338051215, "grad_norm": 0.005140623543411493, "learning_rate": 2.4671759092411708e-08, "loss": 0.0002, "step": 237500 }, { "epoch": 3.886279964002291, "grad_norm": 0.048621971160173416, "learning_rate": 2.4600972316352612e-08, "loss": 0.0004, "step": 237510 }, { "epoch": 3.8864435899533665, "grad_norm": 0.07440219819545746, "learning_rate": 2.453028698444826e-08, "loss": 0.0002, "step": 237520 }, { "epoch": 3.8866072159044425, "grad_norm": 0.005861829966306686, "learning_rate": 2.445970309813972e-08, "loss": 0.0004, "step": 237530 }, { "epoch": 3.886770841855518, "grad_norm": 0.006997311022132635, "learning_rate": 2.4389220658866952e-08, "loss": 0.0002, "step": 237540 }, { "epoch": 3.886934467806594, "grad_norm": 0.03011159412562847, "learning_rate": 2.4318839668065476e-08, "loss": 0.0001, "step": 237550 }, { "epoch": 3.88709809375767, "grad_norm": 0.061636507511138916, "learning_rate": 2.4248560127171915e-08, "loss": 0.0003, "step": 237560 }, { "epoch": 3.8872617197087456, "grad_norm": 0.0005989936180412769, "learning_rate": 2.417838203761791e-08, "loss": 0.0001, "step": 237570 }, { "epoch": 3.8874253456598216, "grad_norm": 0.026471486315131187, "learning_rate": 2.4108305400835086e-08, "loss": 0.0002, "step": 237580 }, { "epoch": 3.8875889716108976, "grad_norm": 0.012404912151396275, "learning_rate": 2.4038330218252305e-08, "loss": 0.0003, "step": 237590 }, { "epoch": 3.887752597561973, "grad_norm": 0.3600476384162903, "learning_rate": 2.3968456491295643e-08, "loss": 0.0002, "step": 237600 }, { "epoch": 3.887916223513049, "grad_norm": 0.01784813404083252, "learning_rate": 2.3898684221390632e-08, "loss": 0.0002, "step": 237610 }, { "epoch": 3.888079849464125, "grad_norm": 0.0019294655648991466, "learning_rate": 2.382901340995947e-08, "loss": 0.0003, "step": 237620 }, { "epoch": 3.8882434754152007, "grad_norm": 0.00013398771989159286, "learning_rate": 2.3759444058422675e-08, "loss": 0.0001, "step": 237630 }, { "epoch": 3.8884071013662767, "grad_norm": 0.004263852257281542, "learning_rate": 2.3689976168198014e-08, "loss": 0.0002, "step": 237640 }, { "epoch": 3.8885707273173526, "grad_norm": 0.18286508321762085, "learning_rate": 2.362060974070324e-08, "loss": 0.0002, "step": 237650 }, { "epoch": 3.888734353268428, "grad_norm": 0.0012528423685580492, "learning_rate": 2.355134477735166e-08, "loss": 0.0004, "step": 237660 }, { "epoch": 3.888897979219504, "grad_norm": 0.015618981793522835, "learning_rate": 2.3482181279556037e-08, "loss": 0.0002, "step": 237670 }, { "epoch": 3.88906160517058, "grad_norm": 0.07350242137908936, "learning_rate": 2.3413119248725803e-08, "loss": 0.0002, "step": 237680 }, { "epoch": 3.8892252311216557, "grad_norm": 0.017008699476718903, "learning_rate": 2.334415868627038e-08, "loss": 0.0003, "step": 237690 }, { "epoch": 3.8893888570727317, "grad_norm": 0.044517528265714645, "learning_rate": 2.3275299593594758e-08, "loss": 0.0001, "step": 237700 }, { "epoch": 3.8895524830238077, "grad_norm": 0.03227957710623741, "learning_rate": 2.3206541972102815e-08, "loss": 0.0004, "step": 237710 }, { "epoch": 3.8897161089748833, "grad_norm": 0.014945144765079021, "learning_rate": 2.3137885823197316e-08, "loss": 0.0001, "step": 237720 }, { "epoch": 3.8898797349259593, "grad_norm": 0.025535566732287407, "learning_rate": 2.3069331148277696e-08, "loss": 0.0002, "step": 237730 }, { "epoch": 3.8900433608770353, "grad_norm": 0.06988061964511871, "learning_rate": 2.3000877948741728e-08, "loss": 0.0004, "step": 237740 }, { "epoch": 3.890206986828111, "grad_norm": 0.0009574806317687035, "learning_rate": 2.2932526225984965e-08, "loss": 0.0001, "step": 237750 }, { "epoch": 3.890370612779187, "grad_norm": 0.1213378757238388, "learning_rate": 2.286427598140073e-08, "loss": 0.0008, "step": 237760 }, { "epoch": 3.890534238730263, "grad_norm": 0.002026155823841691, "learning_rate": 2.279612721638125e-08, "loss": 0.0002, "step": 237770 }, { "epoch": 3.8906978646813384, "grad_norm": 0.016191957518458366, "learning_rate": 2.2728079932316515e-08, "loss": 0.0006, "step": 237780 }, { "epoch": 3.8908614906324144, "grad_norm": 0.04338680952787399, "learning_rate": 2.2660134130592094e-08, "loss": 0.0003, "step": 237790 }, { "epoch": 3.8910251165834904, "grad_norm": 0.054481420665979385, "learning_rate": 2.2592289812595202e-08, "loss": 0.0004, "step": 237800 }, { "epoch": 3.891188742534566, "grad_norm": 0.1479754000902176, "learning_rate": 2.2524546979708074e-08, "loss": 0.0005, "step": 237810 }, { "epoch": 3.891352368485642, "grad_norm": 0.0019910542760044336, "learning_rate": 2.2456905633312377e-08, "loss": 0.0002, "step": 237820 }, { "epoch": 3.891515994436718, "grad_norm": 0.013386833481490612, "learning_rate": 2.238936577478701e-08, "loss": 0.0003, "step": 237830 }, { "epoch": 3.8916796203877935, "grad_norm": 0.0005289442487992346, "learning_rate": 2.23219274055092e-08, "loss": 0.0004, "step": 237840 }, { "epoch": 3.8918432463388695, "grad_norm": 0.03333495929837227, "learning_rate": 2.225459052685397e-08, "loss": 0.0001, "step": 237850 }, { "epoch": 3.8920068722899455, "grad_norm": 0.02133624255657196, "learning_rate": 2.21873551401941e-08, "loss": 0.0002, "step": 237860 }, { "epoch": 3.892170498241021, "grad_norm": 0.009116358123719692, "learning_rate": 2.212022124690072e-08, "loss": 0.0003, "step": 237870 }, { "epoch": 3.892334124192097, "grad_norm": 0.04745730012655258, "learning_rate": 2.205318884834273e-08, "loss": 0.0003, "step": 237880 }, { "epoch": 3.8924977501431726, "grad_norm": 0.00013873928401153535, "learning_rate": 2.198625794588627e-08, "loss": 0.0003, "step": 237890 }, { "epoch": 3.8926613760942486, "grad_norm": 0.025598011910915375, "learning_rate": 2.1919428540896347e-08, "loss": 0.0002, "step": 237900 }, { "epoch": 3.892825002045324, "grad_norm": 0.06727661937475204, "learning_rate": 2.185270063473577e-08, "loss": 0.0003, "step": 237910 }, { "epoch": 3.8929886279964, "grad_norm": 0.08285490423440933, "learning_rate": 2.1786074228764552e-08, "loss": 0.0003, "step": 237920 }, { "epoch": 3.893152253947476, "grad_norm": 0.009581012651324272, "learning_rate": 2.1719549324341615e-08, "loss": 0.0002, "step": 237930 }, { "epoch": 3.8933158798985517, "grad_norm": 0.005427568685263395, "learning_rate": 2.1653125922823648e-08, "loss": 0.0002, "step": 237940 }, { "epoch": 3.8934795058496277, "grad_norm": 0.001690391101874411, "learning_rate": 2.158680402556401e-08, "loss": 0.0003, "step": 237950 }, { "epoch": 3.8936431318007036, "grad_norm": 0.04948057234287262, "learning_rate": 2.1520583633915514e-08, "loss": 0.0002, "step": 237960 }, { "epoch": 3.893806757751779, "grad_norm": 0.0024050227366387844, "learning_rate": 2.1454464749228187e-08, "loss": 0.0002, "step": 237970 }, { "epoch": 3.893970383702855, "grad_norm": 0.07101365178823471, "learning_rate": 2.1388447372850397e-08, "loss": 0.0003, "step": 237980 }, { "epoch": 3.894134009653931, "grad_norm": 0.023585112765431404, "learning_rate": 2.1322531506128286e-08, "loss": 0.0002, "step": 237990 }, { "epoch": 3.8942976356050067, "grad_norm": 0.021000085398554802, "learning_rate": 2.125671715040467e-08, "loss": 0.0002, "step": 238000 }, { "epoch": 3.8944612615560827, "grad_norm": 0.06159278377890587, "learning_rate": 2.1191004307022923e-08, "loss": 0.0003, "step": 238010 }, { "epoch": 3.8946248875071587, "grad_norm": 0.0035767792724072933, "learning_rate": 2.1125392977322524e-08, "loss": 0.0008, "step": 238020 }, { "epoch": 3.8947885134582343, "grad_norm": 0.0020886335987597704, "learning_rate": 2.105988316264074e-08, "loss": 0.0002, "step": 238030 }, { "epoch": 3.8949521394093103, "grad_norm": 0.054693736135959625, "learning_rate": 2.0994474864313718e-08, "loss": 0.0004, "step": 238040 }, { "epoch": 3.8951157653603863, "grad_norm": 0.030459502711892128, "learning_rate": 2.092916808367429e-08, "loss": 0.0006, "step": 238050 }, { "epoch": 3.895279391311462, "grad_norm": 0.004165296908468008, "learning_rate": 2.086396282205472e-08, "loss": 0.0004, "step": 238060 }, { "epoch": 3.895443017262538, "grad_norm": 0.028845541179180145, "learning_rate": 2.0798859080784495e-08, "loss": 0.0002, "step": 238070 }, { "epoch": 3.895606643213614, "grad_norm": 0.021011903882026672, "learning_rate": 2.0733856861190894e-08, "loss": 0.0002, "step": 238080 }, { "epoch": 3.8957702691646894, "grad_norm": 0.026535863056778908, "learning_rate": 2.0668956164598963e-08, "loss": 0.0003, "step": 238090 }, { "epoch": 3.8959338951157654, "grad_norm": 0.04060932993888855, "learning_rate": 2.0604156992332645e-08, "loss": 0.0002, "step": 238100 }, { "epoch": 3.8960975210668414, "grad_norm": 0.05273643136024475, "learning_rate": 2.0539459345712554e-08, "loss": 0.0004, "step": 238110 }, { "epoch": 3.896261147017917, "grad_norm": 0.0019492896972224116, "learning_rate": 2.0474863226058184e-08, "loss": 0.0001, "step": 238120 }, { "epoch": 3.896424772968993, "grad_norm": 0.0004058518970850855, "learning_rate": 2.041036863468626e-08, "loss": 0.0001, "step": 238130 }, { "epoch": 3.896588398920069, "grad_norm": 0.0037659932859241962, "learning_rate": 2.0345975572911848e-08, "loss": 0.0002, "step": 238140 }, { "epoch": 3.8967520248711445, "grad_norm": 0.0007285527535714209, "learning_rate": 2.0281684042048335e-08, "loss": 0.0002, "step": 238150 }, { "epoch": 3.8969156508222205, "grad_norm": 0.010068034753203392, "learning_rate": 2.0217494043405783e-08, "loss": 0.0002, "step": 238160 }, { "epoch": 3.8970792767732965, "grad_norm": 0.002146301092579961, "learning_rate": 2.015340557829315e-08, "loss": 0.0001, "step": 238170 }, { "epoch": 3.897242902724372, "grad_norm": 0.009704994037747383, "learning_rate": 2.0089418648018278e-08, "loss": 0.0002, "step": 238180 }, { "epoch": 3.897406528675448, "grad_norm": 0.04274354875087738, "learning_rate": 2.002553325388401e-08, "loss": 0.0003, "step": 238190 }, { "epoch": 3.897570154626524, "grad_norm": 0.0006202268996275961, "learning_rate": 1.996174939719431e-08, "loss": 0.0002, "step": 238200 }, { "epoch": 3.8977337805775996, "grad_norm": 0.0012952688848599792, "learning_rate": 1.989806707924924e-08, "loss": 0.0004, "step": 238210 }, { "epoch": 3.8978974065286756, "grad_norm": 0.007279532961547375, "learning_rate": 1.9834486301346655e-08, "loss": 0.0001, "step": 238220 }, { "epoch": 3.8980610324797516, "grad_norm": 0.008895144797861576, "learning_rate": 1.9771007064783298e-08, "loss": 0.0004, "step": 238230 }, { "epoch": 3.898224658430827, "grad_norm": 0.20646162331104279, "learning_rate": 1.9707629370854243e-08, "loss": 0.0003, "step": 238240 }, { "epoch": 3.898388284381903, "grad_norm": 0.01955949328839779, "learning_rate": 1.9644353220850676e-08, "loss": 0.0002, "step": 238250 }, { "epoch": 3.8985519103329787, "grad_norm": 0.015400471165776253, "learning_rate": 1.9581178616063235e-08, "loss": 0.0007, "step": 238260 }, { "epoch": 3.8987155362840547, "grad_norm": 0.012732917442917824, "learning_rate": 1.9518105557779775e-08, "loss": 0.0001, "step": 238270 }, { "epoch": 3.8988791622351306, "grad_norm": 0.02558112144470215, "learning_rate": 1.9455134047286495e-08, "loss": 0.0001, "step": 238280 }, { "epoch": 3.899042788186206, "grad_norm": 0.00426999619230628, "learning_rate": 1.9392264085867363e-08, "loss": 0.0002, "step": 238290 }, { "epoch": 3.899206414137282, "grad_norm": 0.0011718057794496417, "learning_rate": 1.9329495674803577e-08, "loss": 0.0002, "step": 238300 }, { "epoch": 3.8993700400883577, "grad_norm": 0.04579935222864151, "learning_rate": 1.9266828815375784e-08, "loss": 0.0002, "step": 238310 }, { "epoch": 3.8995336660394337, "grad_norm": 0.06735184043645859, "learning_rate": 1.920426350886129e-08, "loss": 0.0004, "step": 238320 }, { "epoch": 3.8996972919905097, "grad_norm": 0.007256234530359507, "learning_rate": 1.9141799756536294e-08, "loss": 0.0002, "step": 238330 }, { "epoch": 3.8998609179415853, "grad_norm": 0.016392385587096214, "learning_rate": 1.9079437559673673e-08, "loss": 0.0014, "step": 238340 }, { "epoch": 3.9000245438926613, "grad_norm": 0.08419141918420792, "learning_rate": 1.9017176919544633e-08, "loss": 0.0002, "step": 238350 }, { "epoch": 3.9001881698437373, "grad_norm": 0.06243232637643814, "learning_rate": 1.895501783741982e-08, "loss": 0.0005, "step": 238360 }, { "epoch": 3.900351795794813, "grad_norm": 0.0300761666148901, "learning_rate": 1.8892960314566e-08, "loss": 0.0002, "step": 238370 }, { "epoch": 3.900515421745889, "grad_norm": 3.6340567021397874e-05, "learning_rate": 1.8831004352248273e-08, "loss": 0.0001, "step": 238380 }, { "epoch": 3.900679047696965, "grad_norm": 0.007210540119558573, "learning_rate": 1.876914995173007e-08, "loss": 0.0001, "step": 238390 }, { "epoch": 3.9008426736480404, "grad_norm": 0.037287481129169464, "learning_rate": 1.8707397114272607e-08, "loss": 0.0002, "step": 238400 }, { "epoch": 3.9010062995991164, "grad_norm": 0.004606505855917931, "learning_rate": 1.864574584113432e-08, "loss": 0.0001, "step": 238410 }, { "epoch": 3.9011699255501924, "grad_norm": 0.0013000908074900508, "learning_rate": 1.8584196133573095e-08, "loss": 0.0002, "step": 238420 }, { "epoch": 3.901333551501268, "grad_norm": 0.016175830736756325, "learning_rate": 1.8522747992844036e-08, "loss": 0.0005, "step": 238430 }, { "epoch": 3.901497177452344, "grad_norm": 0.005969323683530092, "learning_rate": 1.846140142019892e-08, "loss": 0.0004, "step": 238440 }, { "epoch": 3.90166080340342, "grad_norm": 0.00461941584944725, "learning_rate": 1.8400156416889525e-08, "loss": 0.0002, "step": 238450 }, { "epoch": 3.9018244293544955, "grad_norm": 0.06847900152206421, "learning_rate": 1.8339012984164296e-08, "loss": 0.0003, "step": 238460 }, { "epoch": 3.9019880553055715, "grad_norm": 0.07023894786834717, "learning_rate": 1.8277971123269457e-08, "loss": 0.0002, "step": 238470 }, { "epoch": 3.9021516812566475, "grad_norm": 0.07092085480690002, "learning_rate": 1.8217030835450678e-08, "loss": 0.0004, "step": 238480 }, { "epoch": 3.902315307207723, "grad_norm": 0.007486239075660706, "learning_rate": 1.815619212194919e-08, "loss": 0.0002, "step": 238490 }, { "epoch": 3.902478933158799, "grad_norm": 0.03805208206176758, "learning_rate": 1.8095454984006222e-08, "loss": 0.0003, "step": 238500 }, { "epoch": 3.902642559109875, "grad_norm": 0.01784871146082878, "learning_rate": 1.8034819422859673e-08, "loss": 0.0002, "step": 238510 }, { "epoch": 3.9028061850609506, "grad_norm": 0.0052723875269293785, "learning_rate": 1.797428543974633e-08, "loss": 0.0002, "step": 238520 }, { "epoch": 3.9029698110120266, "grad_norm": 0.007690985221415758, "learning_rate": 1.7913853035900764e-08, "loss": 0.0003, "step": 238530 }, { "epoch": 3.9031334369631026, "grad_norm": 0.0014453879557549953, "learning_rate": 1.7853522212553652e-08, "loss": 0.0003, "step": 238540 }, { "epoch": 3.903297062914178, "grad_norm": 0.0019150417065247893, "learning_rate": 1.7793292970936794e-08, "loss": 0.0003, "step": 238550 }, { "epoch": 3.903460688865254, "grad_norm": 0.02038702741265297, "learning_rate": 1.7733165312277533e-08, "loss": 0.0002, "step": 238560 }, { "epoch": 3.90362431481633, "grad_norm": 2.438792034809012e-05, "learning_rate": 1.7673139237801562e-08, "loss": 0.0002, "step": 238570 }, { "epoch": 3.9037879407674057, "grad_norm": 0.004936219658702612, "learning_rate": 1.7613214748732344e-08, "loss": 0.0004, "step": 238580 }, { "epoch": 3.9039515667184816, "grad_norm": 0.04954860359430313, "learning_rate": 1.7553391846293343e-08, "loss": 0.0005, "step": 238590 }, { "epoch": 3.9041151926695576, "grad_norm": 0.012010662816464901, "learning_rate": 1.7493670531702478e-08, "loss": 0.0004, "step": 238600 }, { "epoch": 3.904278818620633, "grad_norm": 0.04183853790163994, "learning_rate": 1.743405080617877e-08, "loss": 0.0002, "step": 238610 }, { "epoch": 3.904442444571709, "grad_norm": 0.011030294001102448, "learning_rate": 1.7374532670936804e-08, "loss": 0.0002, "step": 238620 }, { "epoch": 3.904606070522785, "grad_norm": 0.00018793035997077823, "learning_rate": 1.7315116127190613e-08, "loss": 0.0005, "step": 238630 }, { "epoch": 3.9047696964738607, "grad_norm": 0.009689376689493656, "learning_rate": 1.7255801176151998e-08, "loss": 0.0007, "step": 238640 }, { "epoch": 3.9049333224249367, "grad_norm": 0.024125630035996437, "learning_rate": 1.719658781902944e-08, "loss": 0.0003, "step": 238650 }, { "epoch": 3.9050969483760123, "grad_norm": 0.011207393370568752, "learning_rate": 1.7137476057031423e-08, "loss": 0.0002, "step": 238660 }, { "epoch": 3.9052605743270883, "grad_norm": 0.07622464001178741, "learning_rate": 1.707846589136142e-08, "loss": 0.0006, "step": 238670 }, { "epoch": 3.9054242002781643, "grad_norm": 0.004608075600117445, "learning_rate": 1.7019557323224577e-08, "loss": 0.0001, "step": 238680 }, { "epoch": 3.90558782622924, "grad_norm": 0.0032805863302201033, "learning_rate": 1.69607503538205e-08, "loss": 0.0004, "step": 238690 }, { "epoch": 3.905751452180316, "grad_norm": 0.050852514803409576, "learning_rate": 1.6902044984349332e-08, "loss": 0.0003, "step": 238700 }, { "epoch": 3.9059150781313914, "grad_norm": 0.00019880411855410784, "learning_rate": 1.6843441216006784e-08, "loss": 0.0004, "step": 238710 }, { "epoch": 3.9060787040824674, "grad_norm": 0.03192745894193649, "learning_rate": 1.6784939049989123e-08, "loss": 0.0002, "step": 238720 }, { "epoch": 3.9062423300335434, "grad_norm": 0.004131961613893509, "learning_rate": 1.6726538487488177e-08, "loss": 0.0002, "step": 238730 }, { "epoch": 3.906405955984619, "grad_norm": 0.008680365979671478, "learning_rate": 1.6668239529695208e-08, "loss": 0.0001, "step": 238740 }, { "epoch": 3.906569581935695, "grad_norm": 0.0006149066030047834, "learning_rate": 1.661004217779816e-08, "loss": 0.0004, "step": 238750 }, { "epoch": 3.906733207886771, "grad_norm": 0.0017361198551952839, "learning_rate": 1.6551946432984966e-08, "loss": 0.0002, "step": 238760 }, { "epoch": 3.9068968338378465, "grad_norm": 0.06418469548225403, "learning_rate": 1.6493952296438576e-08, "loss": 0.0003, "step": 238770 }, { "epoch": 3.9070604597889225, "grad_norm": 0.0372718945145607, "learning_rate": 1.643605976934248e-08, "loss": 0.0002, "step": 238780 }, { "epoch": 3.9072240857399985, "grad_norm": 0.029314151033759117, "learning_rate": 1.6378268852876855e-08, "loss": 0.0002, "step": 238790 }, { "epoch": 3.907387711691074, "grad_norm": 0.006731295958161354, "learning_rate": 1.6320579548219638e-08, "loss": 0.0001, "step": 238800 }, { "epoch": 3.90755133764215, "grad_norm": 0.011432940140366554, "learning_rate": 1.626299185654767e-08, "loss": 0.0002, "step": 238810 }, { "epoch": 3.907714963593226, "grad_norm": 0.10367309302091599, "learning_rate": 1.620550577903446e-08, "loss": 0.0006, "step": 238820 }, { "epoch": 3.9078785895443016, "grad_norm": 0.03340691700577736, "learning_rate": 1.6148121316852393e-08, "loss": 0.0002, "step": 238830 }, { "epoch": 3.9080422154953776, "grad_norm": 0.05367486923933029, "learning_rate": 1.6090838471171655e-08, "loss": 0.0002, "step": 238840 }, { "epoch": 3.9082058414464536, "grad_norm": 0.061002347618341446, "learning_rate": 1.6033657243160193e-08, "loss": 0.0003, "step": 238850 }, { "epoch": 3.908369467397529, "grad_norm": 0.008035501465201378, "learning_rate": 1.597657763398375e-08, "loss": 0.0005, "step": 238860 }, { "epoch": 3.908533093348605, "grad_norm": 0.04882388189435005, "learning_rate": 1.5919599644805827e-08, "loss": 0.0001, "step": 238870 }, { "epoch": 3.908696719299681, "grad_norm": 0.0017900826642289758, "learning_rate": 1.586272327678884e-08, "loss": 0.0002, "step": 238880 }, { "epoch": 3.9088603452507567, "grad_norm": 0.066990926861763, "learning_rate": 1.580594853109185e-08, "loss": 0.0001, "step": 238890 }, { "epoch": 3.9090239712018326, "grad_norm": 0.06844396144151688, "learning_rate": 1.574927540887283e-08, "loss": 0.0003, "step": 238900 }, { "epoch": 3.9091875971529086, "grad_norm": 7.461700442945585e-05, "learning_rate": 1.5692703911286966e-08, "loss": 0.0003, "step": 238910 }, { "epoch": 3.909351223103984, "grad_norm": 0.0012350588804110885, "learning_rate": 1.5636234039487773e-08, "loss": 0.0003, "step": 238920 }, { "epoch": 3.90951484905506, "grad_norm": 0.009923795238137245, "learning_rate": 1.557986579462656e-08, "loss": 0.0004, "step": 238930 }, { "epoch": 3.909678475006136, "grad_norm": 0.014713004231452942, "learning_rate": 1.552359917785351e-08, "loss": 0.0002, "step": 238940 }, { "epoch": 3.9098421009572117, "grad_norm": 0.07383006811141968, "learning_rate": 1.5467434190314378e-08, "loss": 0.0003, "step": 238950 }, { "epoch": 3.9100057269082877, "grad_norm": 0.06710195541381836, "learning_rate": 1.541137083315547e-08, "loss": 0.0004, "step": 238960 }, { "epoch": 3.9101693528593637, "grad_norm": 0.02951279655098915, "learning_rate": 1.5355409107519205e-08, "loss": 0.0003, "step": 238970 }, { "epoch": 3.9103329788104393, "grad_norm": 0.0015975331189110875, "learning_rate": 1.5299549014546887e-08, "loss": 0.0002, "step": 238980 }, { "epoch": 3.9104966047615153, "grad_norm": 0.05391312018036842, "learning_rate": 1.5243790555377613e-08, "loss": 0.0003, "step": 238990 }, { "epoch": 3.9106602307125913, "grad_norm": 0.001745543791912496, "learning_rate": 1.5188133731148248e-08, "loss": 0.0001, "step": 239000 }, { "epoch": 3.910823856663667, "grad_norm": 0.02851029671728611, "learning_rate": 1.513257854299288e-08, "loss": 0.0019, "step": 239010 }, { "epoch": 3.910987482614743, "grad_norm": 0.02616926282644272, "learning_rate": 1.5077124992045057e-08, "loss": 0.0003, "step": 239020 }, { "epoch": 3.9111511085658184, "grad_norm": 0.011484251357614994, "learning_rate": 1.5021773079434976e-08, "loss": 0.0002, "step": 239030 }, { "epoch": 3.9113147345168944, "grad_norm": 0.05769815295934677, "learning_rate": 1.496652280629174e-08, "loss": 0.0003, "step": 239040 }, { "epoch": 3.9114783604679704, "grad_norm": 0.021050184965133667, "learning_rate": 1.4911374173741112e-08, "loss": 0.0004, "step": 239050 }, { "epoch": 3.911641986419046, "grad_norm": 0.07175266742706299, "learning_rate": 1.4856327182908304e-08, "loss": 0.0003, "step": 239060 }, { "epoch": 3.911805612370122, "grad_norm": 0.00019271051860414445, "learning_rate": 1.4801381834915197e-08, "loss": 0.0001, "step": 239070 }, { "epoch": 3.9119692383211975, "grad_norm": 0.00022516238095704466, "learning_rate": 1.474653813088145e-08, "loss": 0.0004, "step": 239080 }, { "epoch": 3.9121328642722735, "grad_norm": 0.026200000196695328, "learning_rate": 1.4691796071926722e-08, "loss": 0.0003, "step": 239090 }, { "epoch": 3.9122964902233495, "grad_norm": 0.014970309101045132, "learning_rate": 1.4637155659166236e-08, "loss": 0.0004, "step": 239100 }, { "epoch": 3.912460116174425, "grad_norm": 0.007793343160301447, "learning_rate": 1.4582616893714097e-08, "loss": 0.0007, "step": 239110 }, { "epoch": 3.912623742125501, "grad_norm": 0.001057296758517623, "learning_rate": 1.4528179776682749e-08, "loss": 0.0003, "step": 239120 }, { "epoch": 3.912787368076577, "grad_norm": 0.010774458758533001, "learning_rate": 1.4473844309181306e-08, "loss": 0.0001, "step": 239130 }, { "epoch": 3.9129509940276526, "grad_norm": 0.039380572736263275, "learning_rate": 1.4419610492318881e-08, "loss": 0.0003, "step": 239140 }, { "epoch": 3.9131146199787286, "grad_norm": 0.04208040237426758, "learning_rate": 1.4365478327200144e-08, "loss": 0.0001, "step": 239150 }, { "epoch": 3.9132782459298046, "grad_norm": 0.011350961402058601, "learning_rate": 1.4311447814928658e-08, "loss": 0.0002, "step": 239160 }, { "epoch": 3.91344187188088, "grad_norm": 0.0243475791066885, "learning_rate": 1.425751895660743e-08, "loss": 0.0003, "step": 239170 }, { "epoch": 3.913605497831956, "grad_norm": 0.012413657270371914, "learning_rate": 1.4203691753335025e-08, "loss": 0.0002, "step": 239180 }, { "epoch": 3.913769123783032, "grad_norm": 0.07841198146343231, "learning_rate": 1.4149966206208343e-08, "loss": 0.0003, "step": 239190 }, { "epoch": 3.9139327497341077, "grad_norm": 0.018180115148425102, "learning_rate": 1.4096342316324285e-08, "loss": 0.0002, "step": 239200 }, { "epoch": 3.9140963756851836, "grad_norm": 0.06769634038209915, "learning_rate": 1.404282008477531e-08, "loss": 0.0003, "step": 239210 }, { "epoch": 3.9142600016362596, "grad_norm": 0.09055343270301819, "learning_rate": 1.3989399512652768e-08, "loss": 0.0003, "step": 239220 }, { "epoch": 3.914423627587335, "grad_norm": 0.03532414138317108, "learning_rate": 1.3936080601045787e-08, "loss": 0.0002, "step": 239230 }, { "epoch": 3.914587253538411, "grad_norm": 0.003941170871257782, "learning_rate": 1.3882863351041276e-08, "loss": 0.0003, "step": 239240 }, { "epoch": 3.914750879489487, "grad_norm": 0.0063653006218373775, "learning_rate": 1.3829747763725033e-08, "loss": 0.0002, "step": 239250 }, { "epoch": 3.9149145054405627, "grad_norm": 0.012968947179615498, "learning_rate": 1.377673384018008e-08, "loss": 0.0006, "step": 239260 }, { "epoch": 3.9150781313916387, "grad_norm": 0.12776458263397217, "learning_rate": 1.3723821581486663e-08, "loss": 0.0003, "step": 239270 }, { "epoch": 3.9152417573427147, "grad_norm": 6.701869278913364e-05, "learning_rate": 1.3671010988723365e-08, "loss": 0.0005, "step": 239280 }, { "epoch": 3.9154053832937903, "grad_norm": 0.019480325281620026, "learning_rate": 1.3618302062967659e-08, "loss": 0.0001, "step": 239290 }, { "epoch": 3.9155690092448663, "grad_norm": 0.00206645461730659, "learning_rate": 1.3565694805294238e-08, "loss": 0.0002, "step": 239300 }, { "epoch": 3.9157326351959423, "grad_norm": 0.001766085741110146, "learning_rate": 1.351318921677558e-08, "loss": 0.0003, "step": 239310 }, { "epoch": 3.915896261147018, "grad_norm": 0.01058371365070343, "learning_rate": 1.3460785298481938e-08, "loss": 0.0002, "step": 239320 }, { "epoch": 3.916059887098094, "grad_norm": 0.006733427289873362, "learning_rate": 1.3408483051481902e-08, "loss": 0.0002, "step": 239330 }, { "epoch": 3.91622351304917, "grad_norm": 0.003792551113292575, "learning_rate": 1.3356282476841843e-08, "loss": 0.0006, "step": 239340 }, { "epoch": 3.9163871390002454, "grad_norm": 0.016304809600114822, "learning_rate": 1.3304183575626462e-08, "loss": 0.0006, "step": 239350 }, { "epoch": 3.9165507649513214, "grad_norm": 0.009912345558404922, "learning_rate": 1.3252186348897688e-08, "loss": 0.0002, "step": 239360 }, { "epoch": 3.9167143909023974, "grad_norm": 0.0021702430676668882, "learning_rate": 1.3200290797715786e-08, "loss": 0.0001, "step": 239370 }, { "epoch": 3.916878016853473, "grad_norm": 0.014573084190487862, "learning_rate": 1.3148496923138799e-08, "loss": 0.0002, "step": 239380 }, { "epoch": 3.917041642804549, "grad_norm": 7.910677959444001e-05, "learning_rate": 1.3096804726222546e-08, "loss": 0.0002, "step": 239390 }, { "epoch": 3.917205268755625, "grad_norm": 0.0008652630494907498, "learning_rate": 1.3045214208021739e-08, "loss": 0.0002, "step": 239400 }, { "epoch": 3.9173688947067005, "grad_norm": 0.0015646327519789338, "learning_rate": 1.2993725369587206e-08, "loss": 0.0004, "step": 239410 }, { "epoch": 3.9175325206577765, "grad_norm": 0.0612456314265728, "learning_rate": 1.2942338211969775e-08, "loss": 0.0002, "step": 239420 }, { "epoch": 3.917696146608852, "grad_norm": 0.014236465096473694, "learning_rate": 1.289105273621638e-08, "loss": 0.0001, "step": 239430 }, { "epoch": 3.917859772559928, "grad_norm": 0.020115245133638382, "learning_rate": 1.2839868943373413e-08, "loss": 0.0001, "step": 239440 }, { "epoch": 3.918023398511004, "grad_norm": 0.0007291788351722062, "learning_rate": 1.2788786834483369e-08, "loss": 0.0001, "step": 239450 }, { "epoch": 3.9181870244620796, "grad_norm": 0.03198063746094704, "learning_rate": 1.2737806410589304e-08, "loss": 0.0002, "step": 239460 }, { "epoch": 3.9183506504131556, "grad_norm": 0.08057805895805359, "learning_rate": 1.2686927672729276e-08, "loss": 0.0003, "step": 239470 }, { "epoch": 3.918514276364231, "grad_norm": 0.01836514100432396, "learning_rate": 1.26361506219419e-08, "loss": 0.0001, "step": 239480 }, { "epoch": 3.918677902315307, "grad_norm": 0.09265690296888351, "learning_rate": 1.2585475259261349e-08, "loss": 0.0003, "step": 239490 }, { "epoch": 3.918841528266383, "grad_norm": 0.0638127252459526, "learning_rate": 1.2534901585721238e-08, "loss": 0.0002, "step": 239500 }, { "epoch": 3.9190051542174587, "grad_norm": 0.0178532674908638, "learning_rate": 1.2484429602352966e-08, "loss": 0.0002, "step": 239510 }, { "epoch": 3.9191687801685346, "grad_norm": 0.013784430921077728, "learning_rate": 1.2434059310185708e-08, "loss": 0.0003, "step": 239520 }, { "epoch": 3.9193324061196106, "grad_norm": 0.04150955751538277, "learning_rate": 1.2383790710245869e-08, "loss": 0.0002, "step": 239530 }, { "epoch": 3.919496032070686, "grad_norm": 0.0033370957244187593, "learning_rate": 1.2333623803558737e-08, "loss": 0.0006, "step": 239540 }, { "epoch": 3.919659658021762, "grad_norm": 0.06723150610923767, "learning_rate": 1.2283558591146827e-08, "loss": 0.0004, "step": 239550 }, { "epoch": 3.919823283972838, "grad_norm": 0.0031864328775554895, "learning_rate": 1.2233595074031545e-08, "loss": 0.0001, "step": 239560 }, { "epoch": 3.9199869099239137, "grad_norm": 0.0064297690987586975, "learning_rate": 1.2183733253231522e-08, "loss": 0.0001, "step": 239570 }, { "epoch": 3.9201505358749897, "grad_norm": 0.021993374451994896, "learning_rate": 1.2133973129763165e-08, "loss": 0.0003, "step": 239580 }, { "epoch": 3.9203141618260657, "grad_norm": 0.009673917666077614, "learning_rate": 1.2084314704640665e-08, "loss": 0.0003, "step": 239590 }, { "epoch": 3.9204777877771413, "grad_norm": 0.03776511177420616, "learning_rate": 1.2034757978877654e-08, "loss": 0.0002, "step": 239600 }, { "epoch": 3.9206414137282173, "grad_norm": 0.0017022830434143543, "learning_rate": 1.1985302953483325e-08, "loss": 0.0003, "step": 239610 }, { "epoch": 3.9208050396792933, "grad_norm": 0.02796991728246212, "learning_rate": 1.1935949629466315e-08, "loss": 0.0005, "step": 239620 }, { "epoch": 3.920968665630369, "grad_norm": 0.02951311692595482, "learning_rate": 1.1886698007833596e-08, "loss": 0.0001, "step": 239630 }, { "epoch": 3.921132291581445, "grad_norm": 0.03264161944389343, "learning_rate": 1.1837548089588258e-08, "loss": 0.0004, "step": 239640 }, { "epoch": 3.921295917532521, "grad_norm": 0.07695106416940689, "learning_rate": 1.1788499875733383e-08, "loss": 0.0002, "step": 239650 }, { "epoch": 3.9214595434835964, "grad_norm": 0.02618435025215149, "learning_rate": 1.173955336726873e-08, "loss": 0.0003, "step": 239660 }, { "epoch": 3.9216231694346724, "grad_norm": 0.012612329795956612, "learning_rate": 1.1690708565192389e-08, "loss": 0.0001, "step": 239670 }, { "epoch": 3.9217867953857484, "grad_norm": 0.031266238540410995, "learning_rate": 1.1641965470499672e-08, "loss": 0.0002, "step": 239680 }, { "epoch": 3.921950421336824, "grad_norm": 0.0044199381954967976, "learning_rate": 1.1593324084185343e-08, "loss": 0.0002, "step": 239690 }, { "epoch": 3.9221140472879, "grad_norm": 0.005642554257065058, "learning_rate": 1.1544784407240273e-08, "loss": 0.0002, "step": 239700 }, { "epoch": 3.922277673238976, "grad_norm": 0.014525358565151691, "learning_rate": 1.1496346440654227e-08, "loss": 0.0004, "step": 239710 }, { "epoch": 3.9224412991900515, "grad_norm": 0.0022447090595960617, "learning_rate": 1.1448010185415859e-08, "loss": 0.0002, "step": 239720 }, { "epoch": 3.9226049251411275, "grad_norm": 0.015538156032562256, "learning_rate": 1.1399775642509382e-08, "loss": 0.0001, "step": 239730 }, { "epoch": 3.9227685510922035, "grad_norm": 0.03983365744352341, "learning_rate": 1.1351642812919006e-08, "loss": 0.0002, "step": 239740 }, { "epoch": 3.922932177043279, "grad_norm": 0.051346637308597565, "learning_rate": 1.1303611697625616e-08, "loss": 0.0006, "step": 239750 }, { "epoch": 3.923095802994355, "grad_norm": 0.01809394732117653, "learning_rate": 1.1255682297609539e-08, "loss": 0.0003, "step": 239760 }, { "epoch": 3.923259428945431, "grad_norm": 0.002477278932929039, "learning_rate": 1.1207854613846658e-08, "loss": 0.0007, "step": 239770 }, { "epoch": 3.9234230548965066, "grad_norm": 0.02301436848938465, "learning_rate": 1.1160128647313417e-08, "loss": 0.0002, "step": 239780 }, { "epoch": 3.9235866808475826, "grad_norm": 0.056959811598062515, "learning_rate": 1.1112504398981816e-08, "loss": 0.0001, "step": 239790 }, { "epoch": 3.923750306798658, "grad_norm": 0.018591511994600296, "learning_rate": 1.1064981869823855e-08, "loss": 0.0002, "step": 239800 }, { "epoch": 3.923913932749734, "grad_norm": 0.0004671601054724306, "learning_rate": 1.1017561060807647e-08, "loss": 0.0007, "step": 239810 }, { "epoch": 3.92407755870081, "grad_norm": 0.03534054756164551, "learning_rate": 1.0970241972900198e-08, "loss": 0.0002, "step": 239820 }, { "epoch": 3.9242411846518856, "grad_norm": 0.09848622232675552, "learning_rate": 1.092302460706629e-08, "loss": 0.0002, "step": 239830 }, { "epoch": 3.9244048106029616, "grad_norm": 0.0001689536584308371, "learning_rate": 1.08759089642696e-08, "loss": 0.0005, "step": 239840 }, { "epoch": 3.924568436554037, "grad_norm": 0.004212838131934404, "learning_rate": 1.0828895045469912e-08, "loss": 0.0003, "step": 239850 }, { "epoch": 3.924732062505113, "grad_norm": 0.028401995077729225, "learning_rate": 1.078198285162535e-08, "loss": 0.0002, "step": 239860 }, { "epoch": 3.924895688456189, "grad_norm": 0.0028031172696501017, "learning_rate": 1.0735172383693481e-08, "loss": 0.0001, "step": 239870 }, { "epoch": 3.9250593144072647, "grad_norm": 0.03321869671344757, "learning_rate": 1.0688463642627989e-08, "loss": 0.0001, "step": 239880 }, { "epoch": 3.9252229403583407, "grad_norm": 0.02616139128804207, "learning_rate": 1.0641856629381441e-08, "loss": 0.0001, "step": 239890 }, { "epoch": 3.9253865663094167, "grad_norm": 0.0015369143802672625, "learning_rate": 1.0595351344904192e-08, "loss": 0.0001, "step": 239900 }, { "epoch": 3.9255501922604923, "grad_norm": 0.01366428006440401, "learning_rate": 1.054894779014437e-08, "loss": 0.0006, "step": 239910 }, { "epoch": 3.9257138182115683, "grad_norm": 0.02996682934463024, "learning_rate": 1.050264596604844e-08, "loss": 0.0008, "step": 239920 }, { "epoch": 3.9258774441626443, "grad_norm": 0.061530329287052155, "learning_rate": 1.045644587355954e-08, "loss": 0.0002, "step": 239930 }, { "epoch": 3.92604107011372, "grad_norm": 0.009178130887448788, "learning_rate": 1.04103475136208e-08, "loss": 0.0001, "step": 239940 }, { "epoch": 3.926204696064796, "grad_norm": 0.006089415866881609, "learning_rate": 1.036435088717147e-08, "loss": 0.0002, "step": 239950 }, { "epoch": 3.926368322015872, "grad_norm": 0.06348525732755661, "learning_rate": 1.0318455995149689e-08, "loss": 0.0002, "step": 239960 }, { "epoch": 3.9265319479669474, "grad_norm": 0.0278983972966671, "learning_rate": 1.0272662838490266e-08, "loss": 0.0003, "step": 239970 }, { "epoch": 3.9266955739180234, "grad_norm": 0.013494380749762058, "learning_rate": 1.0226971418128562e-08, "loss": 0.0001, "step": 239980 }, { "epoch": 3.9268591998690994, "grad_norm": 0.01398968044668436, "learning_rate": 1.0181381734994388e-08, "loss": 0.0003, "step": 239990 }, { "epoch": 3.927022825820175, "grad_norm": 0.004865814000368118, "learning_rate": 1.013589379001867e-08, "loss": 0.0001, "step": 240000 }, { "epoch": 3.927186451771251, "grad_norm": 0.010585293173789978, "learning_rate": 1.009050758412844e-08, "loss": 0.0001, "step": 240010 }, { "epoch": 3.927350077722327, "grad_norm": 0.0015467077028006315, "learning_rate": 1.0045223118248514e-08, "loss": 0.0002, "step": 240020 }, { "epoch": 3.9275137036734025, "grad_norm": 0.03717828914523125, "learning_rate": 1.0000040393303157e-08, "loss": 0.0002, "step": 240030 }, { "epoch": 3.9276773296244785, "grad_norm": 0.05113723501563072, "learning_rate": 9.954959410212738e-09, "loss": 0.0002, "step": 240040 }, { "epoch": 3.9278409555755545, "grad_norm": 0.0030257939361035824, "learning_rate": 9.909980169897082e-09, "loss": 0.0001, "step": 240050 }, { "epoch": 3.92800458152663, "grad_norm": 0.0019092436414211988, "learning_rate": 9.865102673273231e-09, "loss": 0.0002, "step": 240060 }, { "epoch": 3.928168207477706, "grad_norm": 0.002320458646863699, "learning_rate": 9.820326921255453e-09, "loss": 0.0002, "step": 240070 }, { "epoch": 3.928331833428782, "grad_norm": 0.012748131528496742, "learning_rate": 9.77565291475746e-09, "loss": 0.0003, "step": 240080 }, { "epoch": 3.9284954593798576, "grad_norm": 0.010789240710437298, "learning_rate": 9.731080654690195e-09, "loss": 0.0002, "step": 240090 }, { "epoch": 3.9286590853309336, "grad_norm": 0.029318105429410934, "learning_rate": 9.686610141961816e-09, "loss": 0.0003, "step": 240100 }, { "epoch": 3.9288227112820095, "grad_norm": 0.0733247920870781, "learning_rate": 9.642241377478822e-09, "loss": 0.0003, "step": 240110 }, { "epoch": 3.928986337233085, "grad_norm": 0.14922116696834564, "learning_rate": 9.597974362147156e-09, "loss": 0.0004, "step": 240120 }, { "epoch": 3.929149963184161, "grad_norm": 0.022347452118992805, "learning_rate": 9.553809096868317e-09, "loss": 0.0001, "step": 240130 }, { "epoch": 3.929313589135237, "grad_norm": 0.014856248162686825, "learning_rate": 9.50974558254325e-09, "loss": 0.0003, "step": 240140 }, { "epoch": 3.9294772150863126, "grad_norm": 0.005241875536739826, "learning_rate": 9.465783820070129e-09, "loss": 0.0002, "step": 240150 }, { "epoch": 3.9296408410373886, "grad_norm": 0.003978188149631023, "learning_rate": 9.421923810345457e-09, "loss": 0.0001, "step": 240160 }, { "epoch": 3.9298044669884646, "grad_norm": 0.0012060787994414568, "learning_rate": 9.378165554262963e-09, "loss": 0.0002, "step": 240170 }, { "epoch": 3.92996809293954, "grad_norm": 0.008388794027268887, "learning_rate": 9.334509052715823e-09, "loss": 0.0002, "step": 240180 }, { "epoch": 3.930131718890616, "grad_norm": 0.014080148190259933, "learning_rate": 9.290954306593325e-09, "loss": 0.0003, "step": 240190 }, { "epoch": 3.9302953448416917, "grad_norm": 0.036560095846652985, "learning_rate": 9.247501316784202e-09, "loss": 0.0002, "step": 240200 }, { "epoch": 3.9304589707927677, "grad_norm": 0.024525314569473267, "learning_rate": 9.204150084173303e-09, "loss": 0.0001, "step": 240210 }, { "epoch": 3.9306225967438437, "grad_norm": 0.000440202682511881, "learning_rate": 9.16090060964603e-09, "loss": 0.0001, "step": 240220 }, { "epoch": 3.9307862226949193, "grad_norm": 0.030854329466819763, "learning_rate": 9.117752894083342e-09, "loss": 0.0001, "step": 240230 }, { "epoch": 3.9309498486459953, "grad_norm": 0.005794018507003784, "learning_rate": 9.074706938364542e-09, "loss": 0.0002, "step": 240240 }, { "epoch": 3.931113474597071, "grad_norm": 0.0006826702738180757, "learning_rate": 9.031762743368366e-09, "loss": 0.0003, "step": 240250 }, { "epoch": 3.931277100548147, "grad_norm": 0.0007484280504286289, "learning_rate": 8.988920309969673e-09, "loss": 0.0004, "step": 240260 }, { "epoch": 3.931440726499223, "grad_norm": 0.0014066274743527174, "learning_rate": 8.946179639042207e-09, "loss": 0.0002, "step": 240270 }, { "epoch": 3.9316043524502984, "grad_norm": 0.020002756267786026, "learning_rate": 8.903540731457494e-09, "loss": 0.0002, "step": 240280 }, { "epoch": 3.9317679784013744, "grad_norm": 0.07064570486545563, "learning_rate": 8.861003588084837e-09, "loss": 0.0005, "step": 240290 }, { "epoch": 3.9319316043524504, "grad_norm": 0.0793895423412323, "learning_rate": 8.818568209791323e-09, "loss": 0.0002, "step": 240300 }, { "epoch": 3.932095230303526, "grad_norm": 0.021312285214662552, "learning_rate": 8.776234597442368e-09, "loss": 0.0003, "step": 240310 }, { "epoch": 3.932258856254602, "grad_norm": 0.01903383433818817, "learning_rate": 8.734002751901171e-09, "loss": 0.0005, "step": 240320 }, { "epoch": 3.932422482205678, "grad_norm": 0.01267272513359785, "learning_rate": 8.691872674028711e-09, "loss": 0.0002, "step": 240330 }, { "epoch": 3.9325861081567535, "grad_norm": 0.04201356694102287, "learning_rate": 8.649844364684301e-09, "loss": 0.0002, "step": 240340 }, { "epoch": 3.9327497341078295, "grad_norm": 0.004652464762330055, "learning_rate": 8.607917824723921e-09, "loss": 0.0002, "step": 240350 }, { "epoch": 3.9329133600589055, "grad_norm": 0.0007865845109336078, "learning_rate": 8.566093055003555e-09, "loss": 0.0002, "step": 240360 }, { "epoch": 3.933076986009981, "grad_norm": 0.007195964455604553, "learning_rate": 8.524370056375298e-09, "loss": 0.0001, "step": 240370 }, { "epoch": 3.933240611961057, "grad_norm": 0.018870579078793526, "learning_rate": 8.482748829690134e-09, "loss": 0.0002, "step": 240380 }, { "epoch": 3.933404237912133, "grad_norm": 0.028043869882822037, "learning_rate": 8.441229375796834e-09, "loss": 0.0002, "step": 240390 }, { "epoch": 3.9335678638632086, "grad_norm": 0.016046389937400818, "learning_rate": 8.399811695541383e-09, "loss": 0.0002, "step": 240400 }, { "epoch": 3.9337314898142846, "grad_norm": 0.016020549461245537, "learning_rate": 8.358495789768661e-09, "loss": 0.0002, "step": 240410 }, { "epoch": 3.9338951157653606, "grad_norm": 0.0187423974275589, "learning_rate": 8.317281659320774e-09, "loss": 0.0003, "step": 240420 }, { "epoch": 3.934058741716436, "grad_norm": 0.02581082656979561, "learning_rate": 8.276169305038162e-09, "loss": 0.0003, "step": 240430 }, { "epoch": 3.934222367667512, "grad_norm": 0.008621889166533947, "learning_rate": 8.235158727759596e-09, "loss": 0.0002, "step": 240440 }, { "epoch": 3.934385993618588, "grad_norm": 0.047083739191293716, "learning_rate": 8.194249928320519e-09, "loss": 0.0003, "step": 240450 }, { "epoch": 3.9345496195696636, "grad_norm": 0.013607215136289597, "learning_rate": 8.15344290755582e-09, "loss": 0.0001, "step": 240460 }, { "epoch": 3.9347132455207396, "grad_norm": 0.03906885161995888, "learning_rate": 8.1127376662965e-09, "loss": 0.0003, "step": 240470 }, { "epoch": 3.9348768714718156, "grad_norm": 0.00526681262999773, "learning_rate": 8.072134205373005e-09, "loss": 0.0003, "step": 240480 }, { "epoch": 3.935040497422891, "grad_norm": 0.08785201609134674, "learning_rate": 8.03163252561301e-09, "loss": 0.0002, "step": 240490 }, { "epoch": 3.935204123373967, "grad_norm": 0.032215360552072525, "learning_rate": 7.991232627843071e-09, "loss": 0.0003, "step": 240500 }, { "epoch": 3.935367749325043, "grad_norm": 0.021200431510806084, "learning_rate": 7.950934512885866e-09, "loss": 0.0001, "step": 240510 }, { "epoch": 3.9355313752761187, "grad_norm": 0.007573321461677551, "learning_rate": 7.910738181563515e-09, "loss": 0.0002, "step": 240520 }, { "epoch": 3.9356950012271947, "grad_norm": 0.03473378345370293, "learning_rate": 7.870643634695917e-09, "loss": 0.0001, "step": 240530 }, { "epoch": 3.9358586271782707, "grad_norm": 0.03587587922811508, "learning_rate": 7.830650873100198e-09, "loss": 0.0004, "step": 240540 }, { "epoch": 3.9360222531293463, "grad_norm": 0.017835453152656555, "learning_rate": 7.790759897591815e-09, "loss": 0.0003, "step": 240550 }, { "epoch": 3.9361858790804223, "grad_norm": 0.007064711302518845, "learning_rate": 7.750970708984007e-09, "loss": 0.0003, "step": 240560 }, { "epoch": 3.936349505031498, "grad_norm": 0.0004094468313269317, "learning_rate": 7.711283308088347e-09, "loss": 0.0003, "step": 240570 }, { "epoch": 3.936513130982574, "grad_norm": 0.12155253440141678, "learning_rate": 7.671697695713632e-09, "loss": 0.0005, "step": 240580 }, { "epoch": 3.93667675693365, "grad_norm": 0.002461057621985674, "learning_rate": 7.632213872666993e-09, "loss": 0.0003, "step": 240590 }, { "epoch": 3.9368403828847254, "grad_norm": 0.008876132778823376, "learning_rate": 7.5928318397539e-09, "loss": 0.0002, "step": 240600 }, { "epoch": 3.9370040088358014, "grad_norm": 0.00032502817339263856, "learning_rate": 7.553551597777042e-09, "loss": 0.0002, "step": 240610 }, { "epoch": 3.937167634786877, "grad_norm": 0.004771659150719643, "learning_rate": 7.514373147537446e-09, "loss": 0.0001, "step": 240620 }, { "epoch": 3.937331260737953, "grad_norm": 0.0061235264874994755, "learning_rate": 7.475296489833917e-09, "loss": 0.0001, "step": 240630 }, { "epoch": 3.937494886689029, "grad_norm": 0.0024344732519239187, "learning_rate": 7.436321625463039e-09, "loss": 0.0002, "step": 240640 }, { "epoch": 3.9376585126401045, "grad_norm": 0.07699984312057495, "learning_rate": 7.397448555219178e-09, "loss": 0.0003, "step": 240650 }, { "epoch": 3.9378221385911805, "grad_norm": 0.025945577770471573, "learning_rate": 7.3586772798955874e-09, "loss": 0.0003, "step": 240660 }, { "epoch": 3.9379857645422565, "grad_norm": 0.04789771884679794, "learning_rate": 7.320007800282191e-09, "loss": 0.0002, "step": 240670 }, { "epoch": 3.938149390493332, "grad_norm": 0.013042842969298363, "learning_rate": 7.281440117168359e-09, "loss": 0.0002, "step": 240680 }, { "epoch": 3.938313016444408, "grad_norm": 0.06642214208841324, "learning_rate": 7.242974231339017e-09, "loss": 0.0003, "step": 240690 }, { "epoch": 3.938476642395484, "grad_norm": 0.019327929243445396, "learning_rate": 7.204610143579649e-09, "loss": 0.0002, "step": 240700 }, { "epoch": 3.9386402683465596, "grad_norm": 0.05189405381679535, "learning_rate": 7.166347854672406e-09, "loss": 0.0004, "step": 240710 }, { "epoch": 3.9388038942976356, "grad_norm": 0.010367552749812603, "learning_rate": 7.128187365396666e-09, "loss": 0.0003, "step": 240720 }, { "epoch": 3.9389675202487116, "grad_norm": 0.0030353630427271128, "learning_rate": 7.090128676530695e-09, "loss": 0.0002, "step": 240730 }, { "epoch": 3.939131146199787, "grad_norm": 0.026491492986679077, "learning_rate": 7.052171788851092e-09, "loss": 0.0002, "step": 240740 }, { "epoch": 3.939294772150863, "grad_norm": 0.022942710667848587, "learning_rate": 7.01431670313113e-09, "loss": 0.0002, "step": 240750 }, { "epoch": 3.939458398101939, "grad_norm": 0.10117331892251968, "learning_rate": 6.976563420142968e-09, "loss": 0.0003, "step": 240760 }, { "epoch": 3.9396220240530146, "grad_norm": 0.008046030066907406, "learning_rate": 6.93891194065599e-09, "loss": 0.0002, "step": 240770 }, { "epoch": 3.9397856500040906, "grad_norm": 0.02432638220489025, "learning_rate": 6.901362265438471e-09, "loss": 0.0001, "step": 240780 }, { "epoch": 3.9399492759551666, "grad_norm": 0.03619849681854248, "learning_rate": 6.863914395255355e-09, "loss": 0.0004, "step": 240790 }, { "epoch": 3.940112901906242, "grad_norm": 0.03164102882146835, "learning_rate": 6.826568330871031e-09, "loss": 0.0002, "step": 240800 }, { "epoch": 3.940276527857318, "grad_norm": 0.020911352708935738, "learning_rate": 6.789324073046e-09, "loss": 0.0001, "step": 240810 }, { "epoch": 3.940440153808394, "grad_norm": 0.01554705761373043, "learning_rate": 6.75218162254021e-09, "loss": 0.0002, "step": 240820 }, { "epoch": 3.9406037797594697, "grad_norm": 0.03947353735566139, "learning_rate": 6.715140980110835e-09, "loss": 0.0002, "step": 240830 }, { "epoch": 3.9407674057105457, "grad_norm": 0.036361612379550934, "learning_rate": 6.678202146513379e-09, "loss": 0.0002, "step": 240840 }, { "epoch": 3.9409310316616217, "grad_norm": 0.010960244573652744, "learning_rate": 6.641365122500576e-09, "loss": 0.0002, "step": 240850 }, { "epoch": 3.9410946576126973, "grad_norm": 0.05794225633144379, "learning_rate": 6.60462990882349e-09, "loss": 0.0003, "step": 240860 }, { "epoch": 3.9412582835637733, "grad_norm": 0.028376342728734016, "learning_rate": 6.567996506230967e-09, "loss": 0.0003, "step": 240870 }, { "epoch": 3.9414219095148493, "grad_norm": 0.012053677812218666, "learning_rate": 6.5314649154707425e-09, "loss": 0.0003, "step": 240880 }, { "epoch": 3.941585535465925, "grad_norm": 0.018567312508821487, "learning_rate": 6.49503513728722e-09, "loss": 0.0001, "step": 240890 }, { "epoch": 3.941749161417001, "grad_norm": 0.027212264016270638, "learning_rate": 6.4587071724225845e-09, "loss": 0.0002, "step": 240900 }, { "epoch": 3.941912787368077, "grad_norm": 0.029053233563899994, "learning_rate": 6.422481021618465e-09, "loss": 0.0002, "step": 240910 }, { "epoch": 3.9420764133191524, "grad_norm": 0.10076586157083511, "learning_rate": 6.38635668561316e-09, "loss": 0.0003, "step": 240920 }, { "epoch": 3.9422400392702284, "grad_norm": 0.16747310757637024, "learning_rate": 6.350334165142747e-09, "loss": 0.0009, "step": 240930 }, { "epoch": 3.9424036652213044, "grad_norm": 0.015636300668120384, "learning_rate": 6.314413460942192e-09, "loss": 0.0003, "step": 240940 }, { "epoch": 3.94256729117238, "grad_norm": 0.007081839721649885, "learning_rate": 6.27859457374369e-09, "loss": 0.0005, "step": 240950 }, { "epoch": 3.942730917123456, "grad_norm": 0.005346972495317459, "learning_rate": 6.242877504278322e-09, "loss": 0.0004, "step": 240960 }, { "epoch": 3.9428945430745315, "grad_norm": 0.027424577623605728, "learning_rate": 6.2072622532727275e-09, "loss": 0.0003, "step": 240970 }, { "epoch": 3.9430581690256075, "grad_norm": 0.001593883614987135, "learning_rate": 6.17174882145466e-09, "loss": 0.0003, "step": 240980 }, { "epoch": 3.9432217949766835, "grad_norm": 0.033254653215408325, "learning_rate": 6.136337209546872e-09, "loss": 0.0003, "step": 240990 }, { "epoch": 3.943385420927759, "grad_norm": 0.001319816685281694, "learning_rate": 6.101027418272676e-09, "loss": 0.0001, "step": 241000 }, { "epoch": 3.943549046878835, "grad_norm": 0.010389864444732666, "learning_rate": 6.06581944835094e-09, "loss": 0.0003, "step": 241010 }, { "epoch": 3.9437126728299106, "grad_norm": 0.03958556801080704, "learning_rate": 6.030713300499979e-09, "loss": 0.0004, "step": 241020 }, { "epoch": 3.9438762987809866, "grad_norm": 0.0002790322178043425, "learning_rate": 5.995708975435888e-09, "loss": 0.0002, "step": 241030 }, { "epoch": 3.9440399247320626, "grad_norm": 0.001826975028961897, "learning_rate": 5.960806473871983e-09, "loss": 0.0003, "step": 241040 }, { "epoch": 3.944203550683138, "grad_norm": 0.00029178219847381115, "learning_rate": 5.926005796519363e-09, "loss": 0.0002, "step": 241050 }, { "epoch": 3.944367176634214, "grad_norm": 0.021745964884757996, "learning_rate": 5.891306944088571e-09, "loss": 0.0002, "step": 241060 }, { "epoch": 3.94453080258529, "grad_norm": 0.009020708501338959, "learning_rate": 5.8567099172868184e-09, "loss": 0.0005, "step": 241070 }, { "epoch": 3.9446944285363656, "grad_norm": 0.03835200145840645, "learning_rate": 5.822214716819652e-09, "loss": 0.0005, "step": 241080 }, { "epoch": 3.9448580544874416, "grad_norm": 0.003101389156654477, "learning_rate": 5.787821343389843e-09, "loss": 0.0002, "step": 241090 }, { "epoch": 3.9450216804385176, "grad_norm": 0.05355137959122658, "learning_rate": 5.753529797698499e-09, "loss": 0.0002, "step": 241100 }, { "epoch": 3.945185306389593, "grad_norm": 0.008579631336033344, "learning_rate": 5.719340080445612e-09, "loss": 0.0004, "step": 241110 }, { "epoch": 3.945348932340669, "grad_norm": 0.03079427406191826, "learning_rate": 5.6852521923278505e-09, "loss": 0.0002, "step": 241120 }, { "epoch": 3.945512558291745, "grad_norm": 0.0010589113226160407, "learning_rate": 5.651266134040767e-09, "loss": 0.001, "step": 241130 }, { "epoch": 3.9456761842428207, "grad_norm": 0.0019165565026924014, "learning_rate": 5.617381906276031e-09, "loss": 0.0001, "step": 241140 }, { "epoch": 3.9458398101938967, "grad_norm": 0.006493162363767624, "learning_rate": 5.5835995097258675e-09, "loss": 0.0004, "step": 241150 }, { "epoch": 3.9460034361449727, "grad_norm": 0.00041775027057155967, "learning_rate": 5.5499189450780585e-09, "loss": 0.0002, "step": 241160 }, { "epoch": 3.9461670620960483, "grad_norm": 0.01388845406472683, "learning_rate": 5.516340213020388e-09, "loss": 0.0001, "step": 241170 }, { "epoch": 3.9463306880471243, "grad_norm": 0.06933871656656265, "learning_rate": 5.482863314236197e-09, "loss": 0.0002, "step": 241180 }, { "epoch": 3.9464943139982003, "grad_norm": 0.009470684453845024, "learning_rate": 5.449488249409385e-09, "loss": 0.0003, "step": 241190 }, { "epoch": 3.946657939949276, "grad_norm": 0.0037283708807080984, "learning_rate": 5.416215019219406e-09, "loss": 0.0002, "step": 241200 }, { "epoch": 3.946821565900352, "grad_norm": 0.009296927601099014, "learning_rate": 5.383043624345164e-09, "loss": 0.0004, "step": 241210 }, { "epoch": 3.946985191851428, "grad_norm": 0.0046508111990988255, "learning_rate": 5.349974065462782e-09, "loss": 0.0001, "step": 241220 }, { "epoch": 3.9471488178025034, "grad_norm": 0.03748248144984245, "learning_rate": 5.317006343246722e-09, "loss": 0.0003, "step": 241230 }, { "epoch": 3.9473124437535794, "grad_norm": 0.0203055702149868, "learning_rate": 5.2841404583692245e-09, "loss": 0.0003, "step": 241240 }, { "epoch": 3.9474760697046554, "grad_norm": 0.0004892916767857969, "learning_rate": 5.251376411499754e-09, "loss": 0.0002, "step": 241250 }, { "epoch": 3.947639695655731, "grad_norm": 0.0019817501306533813, "learning_rate": 5.218714203307218e-09, "loss": 0.0002, "step": 241260 }, { "epoch": 3.947803321606807, "grad_norm": 0.07009931653738022, "learning_rate": 5.186153834457197e-09, "loss": 0.0002, "step": 241270 }, { "epoch": 3.947966947557883, "grad_norm": 0.0004757447459269315, "learning_rate": 5.153695305613604e-09, "loss": 0.0002, "step": 241280 }, { "epoch": 3.9481305735089585, "grad_norm": 0.09605037420988083, "learning_rate": 5.1213386174381316e-09, "loss": 0.0002, "step": 241290 }, { "epoch": 3.9482941994600345, "grad_norm": 0.016326773911714554, "learning_rate": 5.089083770590253e-09, "loss": 0.0013, "step": 241300 }, { "epoch": 3.9484578254111105, "grad_norm": 0.016073603183031082, "learning_rate": 5.0569307657283295e-09, "loss": 0.0002, "step": 241310 }, { "epoch": 3.948621451362186, "grad_norm": 0.014839928597211838, "learning_rate": 5.024879603507393e-09, "loss": 0.0002, "step": 241320 }, { "epoch": 3.948785077313262, "grad_norm": 0.07064076513051987, "learning_rate": 4.9929302845813655e-09, "loss": 0.0005, "step": 241330 }, { "epoch": 3.948948703264338, "grad_norm": 0.06182431802153587, "learning_rate": 4.961082809600837e-09, "loss": 0.0001, "step": 241340 }, { "epoch": 3.9491123292154136, "grad_norm": 0.009572107344865799, "learning_rate": 4.929337179216398e-09, "loss": 0.0003, "step": 241350 }, { "epoch": 3.9492759551664895, "grad_norm": 0.0036354733165353537, "learning_rate": 4.897693394074199e-09, "loss": 0.0001, "step": 241360 }, { "epoch": 3.949439581117565, "grad_norm": 0.012425081804394722, "learning_rate": 4.866151454819834e-09, "loss": 0.0001, "step": 241370 }, { "epoch": 3.949603207068641, "grad_norm": 0.012224550358951092, "learning_rate": 4.834711362096678e-09, "loss": 0.0003, "step": 241380 }, { "epoch": 3.9497668330197166, "grad_norm": 0.0031121750362217426, "learning_rate": 4.8033731165447735e-09, "loss": 0.0001, "step": 241390 }, { "epoch": 3.9499304589707926, "grad_norm": 4.603762135957368e-05, "learning_rate": 4.772136718804721e-09, "loss": 0.0001, "step": 241400 }, { "epoch": 3.9500940849218686, "grad_norm": 0.05269702896475792, "learning_rate": 4.741002169512121e-09, "loss": 0.0001, "step": 241410 }, { "epoch": 3.950257710872944, "grad_norm": 0.03125518560409546, "learning_rate": 4.709969469302023e-09, "loss": 0.0002, "step": 241420 }, { "epoch": 3.95042133682402, "grad_norm": 0.002900310792028904, "learning_rate": 4.679038618807807e-09, "loss": 0.0001, "step": 241430 }, { "epoch": 3.950584962775096, "grad_norm": 0.020559577271342278, "learning_rate": 4.648209618658972e-09, "loss": 0.0001, "step": 241440 }, { "epoch": 3.9507485887261717, "grad_norm": 0.06922642141580582, "learning_rate": 4.617482469485568e-09, "loss": 0.0004, "step": 241450 }, { "epoch": 3.9509122146772477, "grad_norm": 0.0001818863966036588, "learning_rate": 4.586857171912651e-09, "loss": 0.0002, "step": 241460 }, { "epoch": 3.9510758406283237, "grad_norm": 0.07028012722730637, "learning_rate": 4.556333726565276e-09, "loss": 0.0005, "step": 241470 }, { "epoch": 3.9512394665793993, "grad_norm": 0.004828637931495905, "learning_rate": 4.525912134066279e-09, "loss": 0.0002, "step": 241480 }, { "epoch": 3.9514030925304753, "grad_norm": 0.02106226608157158, "learning_rate": 4.4955923950346094e-09, "loss": 0.0006, "step": 241490 }, { "epoch": 3.9515667184815513, "grad_norm": 0.017598705366253853, "learning_rate": 4.4653745100903255e-09, "loss": 0.0003, "step": 241500 }, { "epoch": 3.951730344432627, "grad_norm": 0.01202763244509697, "learning_rate": 4.435258479847937e-09, "loss": 0.0001, "step": 241510 }, { "epoch": 3.951893970383703, "grad_norm": 0.0005190340452827513, "learning_rate": 4.405244304921952e-09, "loss": 0.0001, "step": 241520 }, { "epoch": 3.952057596334779, "grad_norm": 0.008353804238140583, "learning_rate": 4.375331985924103e-09, "loss": 0.0003, "step": 241530 }, { "epoch": 3.9522212222858544, "grad_norm": 0.007467310875654221, "learning_rate": 4.345521523465568e-09, "loss": 0.0003, "step": 241540 }, { "epoch": 3.9523848482369304, "grad_norm": 0.03721633553504944, "learning_rate": 4.3158129181525285e-09, "loss": 0.0005, "step": 241550 }, { "epoch": 3.9525484741880064, "grad_norm": 0.001585818245075643, "learning_rate": 4.286206170591167e-09, "loss": 0.0004, "step": 241560 }, { "epoch": 3.952712100139082, "grad_norm": 0.0051811146549880505, "learning_rate": 4.256701281386e-09, "loss": 0.0002, "step": 241570 }, { "epoch": 3.952875726090158, "grad_norm": 0.0036832340992987156, "learning_rate": 4.227298251137657e-09, "loss": 0.0003, "step": 241580 }, { "epoch": 3.953039352041234, "grad_norm": 0.0044623310677707195, "learning_rate": 4.197997080446214e-09, "loss": 0.0002, "step": 241590 }, { "epoch": 3.9532029779923095, "grad_norm": 0.032492995262145996, "learning_rate": 4.168797769908417e-09, "loss": 0.0006, "step": 241600 }, { "epoch": 3.9533666039433855, "grad_norm": 0.0008544790325686336, "learning_rate": 4.139700320120454e-09, "loss": 0.0003, "step": 241610 }, { "epoch": 3.9535302298944615, "grad_norm": 0.013395563699305058, "learning_rate": 4.1107047316746305e-09, "loss": 0.0002, "step": 241620 }, { "epoch": 3.953693855845537, "grad_norm": 0.08131637424230576, "learning_rate": 4.08181100516325e-09, "loss": 0.0005, "step": 241630 }, { "epoch": 3.953857481796613, "grad_norm": 0.010509035550057888, "learning_rate": 4.05301914117473e-09, "loss": 0.0003, "step": 241640 }, { "epoch": 3.954021107747689, "grad_norm": 0.043314263224601746, "learning_rate": 4.024329140295824e-09, "loss": 0.0003, "step": 241650 }, { "epoch": 3.9541847336987646, "grad_norm": 0.07945915311574936, "learning_rate": 3.9957410031121744e-09, "loss": 0.0005, "step": 241660 }, { "epoch": 3.9543483596498405, "grad_norm": 0.0010261160787194967, "learning_rate": 3.967254730206649e-09, "loss": 0.0002, "step": 241670 }, { "epoch": 3.9545119856009165, "grad_norm": 0.0022427430376410484, "learning_rate": 3.938870322159339e-09, "loss": 0.0002, "step": 241680 }, { "epoch": 3.954675611551992, "grad_norm": 0.002813339000567794, "learning_rate": 3.9105877795497795e-09, "loss": 0.0002, "step": 241690 }, { "epoch": 3.954839237503068, "grad_norm": 0.0003425136674195528, "learning_rate": 3.8824071029547325e-09, "loss": 0.0002, "step": 241700 }, { "epoch": 3.955002863454144, "grad_norm": 0.004931437782943249, "learning_rate": 3.854328292947629e-09, "loss": 0.0002, "step": 241710 }, { "epoch": 3.9551664894052196, "grad_norm": 0.0066382745280861855, "learning_rate": 3.826351350102453e-09, "loss": 0.0003, "step": 241720 }, { "epoch": 3.9553301153562956, "grad_norm": 0.03162849694490433, "learning_rate": 3.798476274988194e-09, "loss": 0.0001, "step": 241730 }, { "epoch": 3.955493741307371, "grad_norm": 0.0037457039579749107, "learning_rate": 3.770703068174397e-09, "loss": 0.0004, "step": 241740 }, { "epoch": 3.955657367258447, "grad_norm": 0.01792590320110321, "learning_rate": 3.743031730226166e-09, "loss": 0.0001, "step": 241750 }, { "epoch": 3.955820993209523, "grad_norm": 0.05541263520717621, "learning_rate": 3.7154622617086024e-09, "loss": 0.0003, "step": 241760 }, { "epoch": 3.9559846191605987, "grad_norm": 0.08822593837976456, "learning_rate": 3.6879946631840356e-09, "loss": 0.0005, "step": 241770 }, { "epoch": 3.9561482451116747, "grad_norm": 0.062432911247015, "learning_rate": 3.6606289352114633e-09, "loss": 0.0006, "step": 241780 }, { "epoch": 3.9563118710627503, "grad_norm": 0.022567156702280045, "learning_rate": 3.633365078349882e-09, "loss": 0.0001, "step": 241790 }, { "epoch": 3.9564754970138263, "grad_norm": 0.03213062509894371, "learning_rate": 3.6062030931544034e-09, "loss": 0.0002, "step": 241800 }, { "epoch": 3.9566391229649023, "grad_norm": 0.0006230603321455419, "learning_rate": 3.579142980179584e-09, "loss": 0.0002, "step": 241810 }, { "epoch": 3.956802748915978, "grad_norm": 0.006677290890365839, "learning_rate": 3.5521847399766496e-09, "loss": 0.0002, "step": 241820 }, { "epoch": 3.956966374867054, "grad_norm": 0.013240446336567402, "learning_rate": 3.5253283730946053e-09, "loss": 0.0004, "step": 241830 }, { "epoch": 3.95713000081813, "grad_norm": 0.052890751510858536, "learning_rate": 3.4985738800824564e-09, "loss": 0.0003, "step": 241840 }, { "epoch": 3.9572936267692054, "grad_norm": 0.0485103577375412, "learning_rate": 3.4719212614847677e-09, "loss": 0.0004, "step": 241850 }, { "epoch": 3.9574572527202814, "grad_norm": 0.058142781257629395, "learning_rate": 3.4453705178455476e-09, "loss": 0.0005, "step": 241860 }, { "epoch": 3.9576208786713574, "grad_norm": 0.01865670084953308, "learning_rate": 3.4189216497054754e-09, "loss": 0.0004, "step": 241870 }, { "epoch": 3.957784504622433, "grad_norm": 0.04344009980559349, "learning_rate": 3.39257465760412e-09, "loss": 0.0002, "step": 241880 }, { "epoch": 3.957948130573509, "grad_norm": 0.02689848653972149, "learning_rate": 3.3663295420788276e-09, "loss": 0.0003, "step": 241890 }, { "epoch": 3.958111756524585, "grad_norm": 0.07494902610778809, "learning_rate": 3.3401863036647276e-09, "loss": 0.0005, "step": 241900 }, { "epoch": 3.9582753824756605, "grad_norm": 0.0040658386424183846, "learning_rate": 3.314144942894726e-09, "loss": 0.0001, "step": 241910 }, { "epoch": 3.9584390084267365, "grad_norm": 0.07417669147253036, "learning_rate": 3.2882054602995095e-09, "loss": 0.0003, "step": 241920 }, { "epoch": 3.9586026343778125, "grad_norm": 0.07616268843412399, "learning_rate": 3.262367856408655e-09, "loss": 0.0002, "step": 241930 }, { "epoch": 3.958766260328888, "grad_norm": 0.02831202559173107, "learning_rate": 3.236632131747852e-09, "loss": 0.0002, "step": 241940 }, { "epoch": 3.958929886279964, "grad_norm": 0.011285112239420414, "learning_rate": 3.2109982868433475e-09, "loss": 0.0002, "step": 241950 }, { "epoch": 3.95909351223104, "grad_norm": 0.0007155928760766983, "learning_rate": 3.1854663222163907e-09, "loss": 0.0002, "step": 241960 }, { "epoch": 3.9592571381821156, "grad_norm": 0.010949399322271347, "learning_rate": 3.160036238388231e-09, "loss": 0.0003, "step": 241970 }, { "epoch": 3.9594207641331916, "grad_norm": 0.1277139037847519, "learning_rate": 3.1347080358773428e-09, "loss": 0.0006, "step": 241980 }, { "epoch": 3.9595843900842675, "grad_norm": 0.02182963863015175, "learning_rate": 3.10948171519998e-09, "loss": 0.0003, "step": 241990 }, { "epoch": 3.959748016035343, "grad_norm": 0.004545552656054497, "learning_rate": 3.084357276870731e-09, "loss": 0.0003, "step": 242000 }, { "epoch": 3.959911641986419, "grad_norm": 0.001025934237986803, "learning_rate": 3.059334721401408e-09, "loss": 0.0001, "step": 242010 }, { "epoch": 3.960075267937495, "grad_norm": 0.03929255157709122, "learning_rate": 3.034414049303269e-09, "loss": 0.0003, "step": 242020 }, { "epoch": 3.9602388938885706, "grad_norm": 0.0005185323534533381, "learning_rate": 3.0095952610831314e-09, "loss": 0.0002, "step": 242030 }, { "epoch": 3.9604025198396466, "grad_norm": 0.07514916360378265, "learning_rate": 2.984878357247256e-09, "loss": 0.0003, "step": 242040 }, { "epoch": 3.9605661457907226, "grad_norm": 0.0010144641855731606, "learning_rate": 2.96026333830024e-09, "loss": 0.0002, "step": 242050 }, { "epoch": 3.960729771741798, "grad_norm": 0.035715196281671524, "learning_rate": 2.9357502047439035e-09, "loss": 0.0002, "step": 242060 }, { "epoch": 3.960893397692874, "grad_norm": 0.004051567986607552, "learning_rate": 2.911338957077292e-09, "loss": 0.0003, "step": 242070 }, { "epoch": 3.96105702364395, "grad_norm": 0.08156687021255493, "learning_rate": 2.887029595798896e-09, "loss": 0.0002, "step": 242080 }, { "epoch": 3.9612206495950257, "grad_norm": 0.007238594815135002, "learning_rate": 2.862822121403874e-09, "loss": 0.0003, "step": 242090 }, { "epoch": 3.9613842755461017, "grad_norm": 0.009525814093649387, "learning_rate": 2.8387165343862765e-09, "loss": 0.0002, "step": 242100 }, { "epoch": 3.9615479014971777, "grad_norm": 0.005817629862576723, "learning_rate": 2.8147128352368214e-09, "loss": 0.0002, "step": 242110 }, { "epoch": 3.9617115274482533, "grad_norm": 0.0115665877237916, "learning_rate": 2.790811024445672e-09, "loss": 0.0002, "step": 242120 }, { "epoch": 3.9618751533993293, "grad_norm": 0.00042970007052645087, "learning_rate": 2.7670111025002166e-09, "loss": 0.0002, "step": 242130 }, { "epoch": 3.962038779350405, "grad_norm": 0.013818780891597271, "learning_rate": 2.7433130698850673e-09, "loss": 0.0002, "step": 242140 }, { "epoch": 3.962202405301481, "grad_norm": 0.0004391397815197706, "learning_rate": 2.7197169270837263e-09, "loss": 0.0002, "step": 242150 }, { "epoch": 3.962366031252557, "grad_norm": 0.11306162178516388, "learning_rate": 2.6962226745774754e-09, "loss": 0.0001, "step": 242160 }, { "epoch": 3.9625296572036324, "grad_norm": 0.0365520715713501, "learning_rate": 2.6728303128453757e-09, "loss": 0.0004, "step": 242170 }, { "epoch": 3.9626932831547084, "grad_norm": 0.0245289858430624, "learning_rate": 2.649539842363713e-09, "loss": 0.0003, "step": 242180 }, { "epoch": 3.962856909105784, "grad_norm": 0.0012715845368802547, "learning_rate": 2.6263512636082176e-09, "loss": 0.0002, "step": 242190 }, { "epoch": 3.96302053505686, "grad_norm": 0.0986972376704216, "learning_rate": 2.60326457705129e-09, "loss": 0.0001, "step": 242200 }, { "epoch": 3.963184161007936, "grad_norm": 0.19039075076580048, "learning_rate": 2.580279783163109e-09, "loss": 0.0006, "step": 242210 }, { "epoch": 3.9633477869590115, "grad_norm": 0.008380274288356304, "learning_rate": 2.557396882413299e-09, "loss": 0.0003, "step": 242220 }, { "epoch": 3.9635114129100875, "grad_norm": 0.0041811238043010235, "learning_rate": 2.5346158752681538e-09, "loss": 0.0003, "step": 242230 }, { "epoch": 3.9636750388611635, "grad_norm": 0.004717675969004631, "learning_rate": 2.511936762191747e-09, "loss": 0.0001, "step": 242240 }, { "epoch": 3.963838664812239, "grad_norm": 0.035370323807001114, "learning_rate": 2.4893595436470408e-09, "loss": 0.0002, "step": 242250 }, { "epoch": 3.964002290763315, "grad_norm": 0.01733201928436756, "learning_rate": 2.466884220093668e-09, "loss": 0.0004, "step": 242260 }, { "epoch": 3.964165916714391, "grad_norm": 0.008368738926947117, "learning_rate": 2.4445107919907064e-09, "loss": 0.0002, "step": 242270 }, { "epoch": 3.9643295426654666, "grad_norm": 0.0005484704161062837, "learning_rate": 2.4222392597933464e-09, "loss": 0.0003, "step": 242280 }, { "epoch": 3.9644931686165426, "grad_norm": 0.024872953072190285, "learning_rate": 2.40006962395678e-09, "loss": 0.0003, "step": 242290 }, { "epoch": 3.9646567945676185, "grad_norm": 0.06614062935113907, "learning_rate": 2.378001884932313e-09, "loss": 0.0002, "step": 242300 }, { "epoch": 3.964820420518694, "grad_norm": 0.030836278572678566, "learning_rate": 2.35603604317014e-09, "loss": 0.0002, "step": 242310 }, { "epoch": 3.96498404646977, "grad_norm": 0.011122869327664375, "learning_rate": 2.334172099117682e-09, "loss": 0.0002, "step": 242320 }, { "epoch": 3.965147672420846, "grad_norm": 0.10118700563907623, "learning_rate": 2.312410053221248e-09, "loss": 0.0003, "step": 242330 }, { "epoch": 3.9653112983719216, "grad_norm": 0.03381729498505592, "learning_rate": 2.290749905924372e-09, "loss": 0.0003, "step": 242340 }, { "epoch": 3.9654749243229976, "grad_norm": 0.0002498764079064131, "learning_rate": 2.269191657668923e-09, "loss": 0.0002, "step": 242350 }, { "epoch": 3.9656385502740736, "grad_norm": 0.024567466229200363, "learning_rate": 2.2477353088945496e-09, "loss": 0.0001, "step": 242360 }, { "epoch": 3.965802176225149, "grad_norm": 0.046708349138498306, "learning_rate": 2.2263808600375693e-09, "loss": 0.0, "step": 242370 }, { "epoch": 3.965965802176225, "grad_norm": 0.01195519883185625, "learning_rate": 2.2051283115348542e-09, "loss": 0.0002, "step": 242380 }, { "epoch": 3.966129428127301, "grad_norm": 0.026321256533265114, "learning_rate": 2.183977663818837e-09, "loss": 0.0002, "step": 242390 }, { "epoch": 3.9662930540783767, "grad_norm": 0.00021938236022833735, "learning_rate": 2.1629289173213942e-09, "loss": 0.0002, "step": 242400 }, { "epoch": 3.9664566800294527, "grad_norm": 0.000226393633056432, "learning_rate": 2.1419820724705163e-09, "loss": 0.0002, "step": 242410 }, { "epoch": 3.9666203059805287, "grad_norm": 0.05674148350954056, "learning_rate": 2.1211371296947503e-09, "loss": 0.0001, "step": 242420 }, { "epoch": 3.9667839319316043, "grad_norm": 0.0052926079370081425, "learning_rate": 2.1003940894182007e-09, "loss": 0.0004, "step": 242430 }, { "epoch": 3.9669475578826803, "grad_norm": 0.0668359026312828, "learning_rate": 2.079752952064418e-09, "loss": 0.0004, "step": 242440 }, { "epoch": 3.9671111838337563, "grad_norm": 0.013307338580489159, "learning_rate": 2.059213718053621e-09, "loss": 0.0001, "step": 242450 }, { "epoch": 3.967274809784832, "grad_norm": 0.010697194375097752, "learning_rate": 2.038776387804919e-09, "loss": 0.0002, "step": 242460 }, { "epoch": 3.967438435735908, "grad_norm": 0.03722424432635307, "learning_rate": 2.0184409617346466e-09, "loss": 0.0002, "step": 242470 }, { "epoch": 3.967602061686984, "grad_norm": 0.023575061932206154, "learning_rate": 1.9982074402580266e-09, "loss": 0.0004, "step": 242480 }, { "epoch": 3.9677656876380594, "grad_norm": 0.0010993767064064741, "learning_rate": 1.9780758237875064e-09, "loss": 0.0002, "step": 242490 }, { "epoch": 3.9679293135891354, "grad_norm": 0.010733261704444885, "learning_rate": 1.9580461127327588e-09, "loss": 0.0004, "step": 242500 }, { "epoch": 3.968092939540211, "grad_norm": 0.01824861764907837, "learning_rate": 1.938118307503456e-09, "loss": 0.0002, "step": 242510 }, { "epoch": 3.968256565491287, "grad_norm": 0.025189070031046867, "learning_rate": 1.9182924085048295e-09, "loss": 0.0002, "step": 242520 }, { "epoch": 3.968420191442363, "grad_norm": 0.000297642924124375, "learning_rate": 1.8985684161415552e-09, "loss": 0.0001, "step": 242530 }, { "epoch": 3.9685838173934385, "grad_norm": 0.0015953800175338984, "learning_rate": 1.8789463308160894e-09, "loss": 0.0003, "step": 242540 }, { "epoch": 3.9687474433445145, "grad_norm": 0.10107693821191788, "learning_rate": 1.859426152927557e-09, "loss": 0.0003, "step": 242550 }, { "epoch": 3.96891106929559, "grad_norm": 0.003915309906005859, "learning_rate": 1.8400078828745283e-09, "loss": 0.0001, "step": 242560 }, { "epoch": 3.969074695246666, "grad_norm": 0.03134828805923462, "learning_rate": 1.8206915210533527e-09, "loss": 0.0002, "step": 242570 }, { "epoch": 3.969238321197742, "grad_norm": 0.049074117094278336, "learning_rate": 1.8014770678576043e-09, "loss": 0.0004, "step": 242580 }, { "epoch": 3.9694019471488176, "grad_norm": 0.0027956829871982336, "learning_rate": 1.782364523678637e-09, "loss": 0.0004, "step": 242590 }, { "epoch": 3.9695655730998936, "grad_norm": 0.017050787806510925, "learning_rate": 1.7633538889066937e-09, "loss": 0.0002, "step": 242600 }, { "epoch": 3.9697291990509695, "grad_norm": 0.009416555985808372, "learning_rate": 1.7444451639286876e-09, "loss": 0.0003, "step": 242610 }, { "epoch": 3.969892825002045, "grad_norm": 0.040275003761053085, "learning_rate": 1.7256383491309758e-09, "loss": 0.0002, "step": 242620 }, { "epoch": 3.970056450953121, "grad_norm": 0.0007636773516424, "learning_rate": 1.7069334448960306e-09, "loss": 0.0003, "step": 242630 }, { "epoch": 3.970220076904197, "grad_norm": 0.016639849171042442, "learning_rate": 1.6883304516063238e-09, "loss": 0.0001, "step": 242640 }, { "epoch": 3.9703837028552726, "grad_norm": 0.007554064504802227, "learning_rate": 1.669829369640441e-09, "loss": 0.0004, "step": 242650 }, { "epoch": 3.9705473288063486, "grad_norm": 0.007550583221018314, "learning_rate": 1.6514301993764138e-09, "loss": 0.0002, "step": 242660 }, { "epoch": 3.9707109547574246, "grad_norm": 0.11822967231273651, "learning_rate": 1.6331329411878316e-09, "loss": 0.0004, "step": 242670 }, { "epoch": 3.9708745807085, "grad_norm": 0.00210783164948225, "learning_rate": 1.6149375954493952e-09, "loss": 0.0003, "step": 242680 }, { "epoch": 3.971038206659576, "grad_norm": 0.01584850437939167, "learning_rate": 1.5968441625313635e-09, "loss": 0.0002, "step": 242690 }, { "epoch": 3.971201832610652, "grad_norm": 0.0016372987302020192, "learning_rate": 1.578852642802331e-09, "loss": 0.0003, "step": 242700 }, { "epoch": 3.9713654585617277, "grad_norm": 0.025488020852208138, "learning_rate": 1.5609630366297813e-09, "loss": 0.0002, "step": 242710 }, { "epoch": 3.9715290845128037, "grad_norm": 0.055389098823070526, "learning_rate": 1.5431753443778674e-09, "loss": 0.0004, "step": 242720 }, { "epoch": 3.9716927104638797, "grad_norm": 0.10989999026060104, "learning_rate": 1.5254895664096325e-09, "loss": 0.0004, "step": 242730 }, { "epoch": 3.9718563364149553, "grad_norm": 0.011835395358502865, "learning_rate": 1.507905703085899e-09, "loss": 0.0004, "step": 242740 }, { "epoch": 3.9720199623660313, "grad_norm": 0.00579506391659379, "learning_rate": 1.4904237547647138e-09, "loss": 0.0001, "step": 242750 }, { "epoch": 3.9721835883171073, "grad_norm": 0.0007396414293907583, "learning_rate": 1.4730437218030136e-09, "loss": 0.0003, "step": 242760 }, { "epoch": 3.972347214268183, "grad_norm": 0.006998383905738592, "learning_rate": 1.4557656045544044e-09, "loss": 0.0001, "step": 242770 }, { "epoch": 3.972510840219259, "grad_norm": 0.014428617432713509, "learning_rate": 1.4385894033719372e-09, "loss": 0.0002, "step": 242780 }, { "epoch": 3.972674466170335, "grad_norm": 0.0475534051656723, "learning_rate": 1.4215151186053322e-09, "loss": 0.0003, "step": 242790 }, { "epoch": 3.9728380921214104, "grad_norm": 0.043817855417728424, "learning_rate": 1.4045427506026443e-09, "loss": 0.0002, "step": 242800 }, { "epoch": 3.9730017180724864, "grad_norm": 0.02545195072889328, "learning_rate": 1.3876722997108182e-09, "loss": 0.0002, "step": 242810 }, { "epoch": 3.9731653440235624, "grad_norm": 0.01985970139503479, "learning_rate": 1.3709037662729129e-09, "loss": 0.0003, "step": 242820 }, { "epoch": 3.973328969974638, "grad_norm": 0.01823471486568451, "learning_rate": 1.3542371506308772e-09, "loss": 0.0004, "step": 242830 }, { "epoch": 3.973492595925714, "grad_norm": 0.03805399686098099, "learning_rate": 1.337672453124994e-09, "loss": 0.0005, "step": 242840 }, { "epoch": 3.97365622187679, "grad_norm": 0.11505287140607834, "learning_rate": 1.3212096740927716e-09, "loss": 0.0005, "step": 242850 }, { "epoch": 3.9738198478278655, "grad_norm": 0.004932223353534937, "learning_rate": 1.3048488138694971e-09, "loss": 0.0002, "step": 242860 }, { "epoch": 3.9739834737789415, "grad_norm": 0.006032584700733423, "learning_rate": 1.2885898727893476e-09, "loss": 0.0005, "step": 242870 }, { "epoch": 3.9741470997300175, "grad_norm": 0.049084603786468506, "learning_rate": 1.2724328511837248e-09, "loss": 0.0007, "step": 242880 }, { "epoch": 3.974310725681093, "grad_norm": 0.006837539374828339, "learning_rate": 1.2563777493818097e-09, "loss": 0.0003, "step": 242890 }, { "epoch": 3.974474351632169, "grad_norm": 0.001386832445859909, "learning_rate": 1.2404245677111183e-09, "loss": 0.0002, "step": 242900 }, { "epoch": 3.9746379775832446, "grad_norm": 0.0265278909355402, "learning_rate": 1.2245733064969456e-09, "loss": 0.0001, "step": 242910 }, { "epoch": 3.9748016035343205, "grad_norm": 0.05833207815885544, "learning_rate": 1.2088239660623668e-09, "loss": 0.0004, "step": 242920 }, { "epoch": 3.9749652294853965, "grad_norm": 0.025683119893074036, "learning_rate": 1.1931765467282363e-09, "loss": 0.0002, "step": 242930 }, { "epoch": 3.975128855436472, "grad_norm": 0.004348804242908955, "learning_rate": 1.1776310488142983e-09, "loss": 0.0003, "step": 242940 }, { "epoch": 3.975292481387548, "grad_norm": 0.026803798973560333, "learning_rate": 1.1621874726369664e-09, "loss": 0.0003, "step": 242950 }, { "epoch": 3.9754561073386236, "grad_norm": 0.011595462448894978, "learning_rate": 1.146845818511544e-09, "loss": 0.0002, "step": 242960 }, { "epoch": 3.9756197332896996, "grad_norm": 0.015873048454523087, "learning_rate": 1.1316060867500034e-09, "loss": 0.0001, "step": 242970 }, { "epoch": 3.9757833592407756, "grad_norm": 0.031630825251340866, "learning_rate": 1.1164682776637625e-09, "loss": 0.0002, "step": 242980 }, { "epoch": 3.975946985191851, "grad_norm": 0.02632499299943447, "learning_rate": 1.1014323915620184e-09, "loss": 0.0003, "step": 242990 }, { "epoch": 3.976110611142927, "grad_norm": 0.02320859022438526, "learning_rate": 1.0864984287500823e-09, "loss": 0.0002, "step": 243000 }, { "epoch": 3.976274237094003, "grad_norm": 0.06132917478680611, "learning_rate": 1.0716663895327107e-09, "loss": 0.0005, "step": 243010 }, { "epoch": 3.9764378630450787, "grad_norm": 0.0010873955907300115, "learning_rate": 1.056936274212994e-09, "loss": 0.0002, "step": 243020 }, { "epoch": 3.9766014889961547, "grad_norm": 0.021007491275668144, "learning_rate": 1.042308083091248e-09, "loss": 0.0001, "step": 243030 }, { "epoch": 3.9767651149472307, "grad_norm": 0.00548027316108346, "learning_rate": 1.0277818164650122e-09, "loss": 0.0002, "step": 243040 }, { "epoch": 3.9769287408983063, "grad_norm": 0.05492765083909035, "learning_rate": 1.0133574746307163e-09, "loss": 0.0002, "step": 243050 }, { "epoch": 3.9770923668493823, "grad_norm": 0.17410893738269806, "learning_rate": 9.990350578825692e-10, "loss": 0.0003, "step": 243060 }, { "epoch": 3.9772559928004583, "grad_norm": 0.02122414857149124, "learning_rate": 9.848145665131148e-10, "loss": 0.0001, "step": 243070 }, { "epoch": 3.977419618751534, "grad_norm": 0.0856589525938034, "learning_rate": 9.70696000811011e-10, "loss": 0.0003, "step": 243080 }, { "epoch": 3.97758324470261, "grad_norm": 0.010458648204803467, "learning_rate": 9.566793610654712e-10, "loss": 0.0003, "step": 243090 }, { "epoch": 3.977746870653686, "grad_norm": 0.04944673180580139, "learning_rate": 9.427646475618225e-10, "loss": 0.0005, "step": 243100 }, { "epoch": 3.9779104966047614, "grad_norm": 0.09928541630506516, "learning_rate": 9.289518605831715e-10, "loss": 0.0004, "step": 243110 }, { "epoch": 3.9780741225558374, "grad_norm": 0.006354372948408127, "learning_rate": 9.152410004120704e-10, "loss": 0.0005, "step": 243120 }, { "epoch": 3.9782377485069134, "grad_norm": 0.12180842459201813, "learning_rate": 9.01632067327185e-10, "loss": 0.0002, "step": 243130 }, { "epoch": 3.978401374457989, "grad_norm": 0.037099629640579224, "learning_rate": 8.881250616066261e-10, "loss": 0.0003, "step": 243140 }, { "epoch": 3.978565000409065, "grad_norm": 0.0012689741561189294, "learning_rate": 8.747199835257292e-10, "loss": 0.0002, "step": 243150 }, { "epoch": 3.978728626360141, "grad_norm": 0.018833698704838753, "learning_rate": 8.61416833357609e-10, "loss": 0.0002, "step": 243160 }, { "epoch": 3.9788922523112165, "grad_norm": 0.004146496299654245, "learning_rate": 8.482156113731599e-10, "loss": 0.0002, "step": 243170 }, { "epoch": 3.9790558782622925, "grad_norm": 0.000886140507645905, "learning_rate": 8.351163178427213e-10, "loss": 0.0001, "step": 243180 }, { "epoch": 3.9792195042133685, "grad_norm": 0.011928497813642025, "learning_rate": 8.221189530321916e-10, "loss": 0.0004, "step": 243190 }, { "epoch": 3.979383130164444, "grad_norm": 0.000748147489503026, "learning_rate": 8.09223517206914e-10, "loss": 0.0001, "step": 243200 }, { "epoch": 3.97954675611552, "grad_norm": 0.06004941463470459, "learning_rate": 7.964300106300116e-10, "loss": 0.0004, "step": 243210 }, { "epoch": 3.979710382066596, "grad_norm": 0.004838142544031143, "learning_rate": 7.837384335623865e-10, "loss": 0.0003, "step": 243220 }, { "epoch": 3.9798740080176715, "grad_norm": 0.01140690129250288, "learning_rate": 7.711487862627209e-10, "loss": 0.0002, "step": 243230 }, { "epoch": 3.9800376339687475, "grad_norm": 0.005961902905255556, "learning_rate": 7.586610689874763e-10, "loss": 0.0001, "step": 243240 }, { "epoch": 3.9802012599198235, "grad_norm": 0.055002786219120026, "learning_rate": 7.462752819914487e-10, "loss": 0.0001, "step": 243250 }, { "epoch": 3.980364885870899, "grad_norm": 0.004033748991787434, "learning_rate": 7.33991425527214e-10, "loss": 0.0001, "step": 243260 }, { "epoch": 3.980528511821975, "grad_norm": 0.058032553642988205, "learning_rate": 7.218094998451274e-10, "loss": 0.0002, "step": 243270 }, { "epoch": 3.9806921377730506, "grad_norm": 0.0011107303434982896, "learning_rate": 7.09729505193324e-10, "loss": 0.0001, "step": 243280 }, { "epoch": 3.9808557637241266, "grad_norm": 0.03731793910264969, "learning_rate": 6.97751441818828e-10, "loss": 0.0002, "step": 243290 }, { "epoch": 3.9810193896752026, "grad_norm": 0.0011762568028643727, "learning_rate": 6.858753099653337e-10, "loss": 0.0004, "step": 243300 }, { "epoch": 3.981183015626278, "grad_norm": 0.07040699571371078, "learning_rate": 6.741011098748696e-10, "loss": 0.0003, "step": 243310 }, { "epoch": 3.981346641577354, "grad_norm": 0.009817279875278473, "learning_rate": 6.624288417877989e-10, "loss": 0.0003, "step": 243320 }, { "epoch": 3.9815102675284297, "grad_norm": 0.012686763890087605, "learning_rate": 6.508585059422645e-10, "loss": 0.0006, "step": 243330 }, { "epoch": 3.9816738934795057, "grad_norm": 0.03547653928399086, "learning_rate": 6.393901025736337e-10, "loss": 0.0003, "step": 243340 }, { "epoch": 3.9818375194305817, "grad_norm": 0.0506691113114357, "learning_rate": 6.280236319161637e-10, "loss": 0.0003, "step": 243350 }, { "epoch": 3.9820011453816573, "grad_norm": 0.0028357377741485834, "learning_rate": 6.167590942013358e-10, "loss": 0.0002, "step": 243360 }, { "epoch": 3.9821647713327333, "grad_norm": 0.002271948615089059, "learning_rate": 6.055964896589661e-10, "loss": 0.0001, "step": 243370 }, { "epoch": 3.9823283972838093, "grad_norm": 0.000336849014274776, "learning_rate": 5.945358185172056e-10, "loss": 0.0003, "step": 243380 }, { "epoch": 3.982492023234885, "grad_norm": 0.005478637758642435, "learning_rate": 5.835770810003194e-10, "loss": 0.0002, "step": 243390 }, { "epoch": 3.982655649185961, "grad_norm": 0.019413074478507042, "learning_rate": 5.727202773325724e-10, "loss": 0.0003, "step": 243400 }, { "epoch": 3.982819275137037, "grad_norm": 0.006269759964197874, "learning_rate": 5.619654077354542e-10, "loss": 0.0002, "step": 243410 }, { "epoch": 3.9829829010881124, "grad_norm": 0.00640812236815691, "learning_rate": 5.513124724276786e-10, "loss": 0.0003, "step": 243420 }, { "epoch": 3.9831465270391884, "grad_norm": 0.002195351757109165, "learning_rate": 5.407614716274046e-10, "loss": 0.0002, "step": 243430 }, { "epoch": 3.9833101529902644, "grad_norm": 0.06291834264993668, "learning_rate": 5.3031240554835e-10, "loss": 0.0003, "step": 243440 }, { "epoch": 3.98347377894134, "grad_norm": 0.045496948063373566, "learning_rate": 5.199652744047879e-10, "loss": 0.0002, "step": 243450 }, { "epoch": 3.983637404892416, "grad_norm": 0.025207646191120148, "learning_rate": 5.097200784071054e-10, "loss": 0.0002, "step": 243460 }, { "epoch": 3.983801030843492, "grad_norm": 0.02587968297302723, "learning_rate": 4.995768177640248e-10, "loss": 0.0001, "step": 243470 }, { "epoch": 3.9839646567945675, "grad_norm": 0.00029696195269934833, "learning_rate": 4.895354926831575e-10, "loss": 0.0002, "step": 243480 }, { "epoch": 3.9841282827456435, "grad_norm": 0.027130281552672386, "learning_rate": 4.795961033682295e-10, "loss": 0.0003, "step": 243490 }, { "epoch": 3.9842919086967195, "grad_norm": 0.03859890624880791, "learning_rate": 4.697586500229667e-10, "loss": 0.0002, "step": 243500 }, { "epoch": 3.984455534647795, "grad_norm": 0.028537284582853317, "learning_rate": 4.6002313284720935e-10, "loss": 0.0002, "step": 243510 }, { "epoch": 3.984619160598871, "grad_norm": 0.0004017931642010808, "learning_rate": 4.5038955203913214e-10, "loss": 0.0002, "step": 243520 }, { "epoch": 3.984782786549947, "grad_norm": 0.11853707581758499, "learning_rate": 4.408579077963548e-10, "loss": 0.0004, "step": 243530 }, { "epoch": 3.9849464125010226, "grad_norm": 0.04253354296088219, "learning_rate": 4.3142820031205624e-10, "loss": 0.0002, "step": 243540 }, { "epoch": 3.9851100384520985, "grad_norm": 0.0672665685415268, "learning_rate": 4.2210042977886e-10, "loss": 0.0004, "step": 243550 }, { "epoch": 3.9852736644031745, "grad_norm": 0.0007297037518583238, "learning_rate": 4.128745963871694e-10, "loss": 0.0002, "step": 243560 }, { "epoch": 3.98543729035425, "grad_norm": 0.10478443652391434, "learning_rate": 4.037507003251673e-10, "loss": 0.0006, "step": 243570 }, { "epoch": 3.985600916305326, "grad_norm": 0.028355171903967857, "learning_rate": 3.9472874177881594e-10, "loss": 0.0002, "step": 243580 }, { "epoch": 3.985764542256402, "grad_norm": 0.00037184858229011297, "learning_rate": 3.8580872093130215e-10, "loss": 0.0006, "step": 243590 }, { "epoch": 3.9859281682074776, "grad_norm": 0.028477348387241364, "learning_rate": 3.769906379658128e-10, "loss": 0.0002, "step": 243600 }, { "epoch": 3.9860917941585536, "grad_norm": 0.0008203476900234818, "learning_rate": 3.6827449306109374e-10, "loss": 0.0002, "step": 243610 }, { "epoch": 3.9862554201096296, "grad_norm": 0.06564685702323914, "learning_rate": 3.5966028639533577e-10, "loss": 0.0003, "step": 243620 }, { "epoch": 3.986419046060705, "grad_norm": 0.0006637396872974932, "learning_rate": 3.5114801814395414e-10, "loss": 0.0002, "step": 243630 }, { "epoch": 3.986582672011781, "grad_norm": 0.007248829584568739, "learning_rate": 3.4273768848069875e-10, "loss": 0.0008, "step": 243640 }, { "epoch": 3.986746297962857, "grad_norm": 0.02344871684908867, "learning_rate": 3.3442929757654396e-10, "loss": 0.0002, "step": 243650 }, { "epoch": 3.9869099239139327, "grad_norm": 0.006966602988541126, "learning_rate": 3.26222845601909e-10, "loss": 0.0, "step": 243660 }, { "epoch": 3.9870735498650087, "grad_norm": 0.03237536922097206, "learning_rate": 3.1811833272332724e-10, "loss": 0.0002, "step": 243670 }, { "epoch": 3.9872371758160843, "grad_norm": 0.11829535663127899, "learning_rate": 3.1011575910622204e-10, "loss": 0.0007, "step": 243680 }, { "epoch": 3.9874008017671603, "grad_norm": 0.0033282190561294556, "learning_rate": 3.022151249137961e-10, "loss": 0.0002, "step": 243690 }, { "epoch": 3.9875644277182363, "grad_norm": 0.065312959253788, "learning_rate": 2.9441643030703184e-10, "loss": 0.0004, "step": 243700 }, { "epoch": 3.987728053669312, "grad_norm": 0.01994680054485798, "learning_rate": 2.8671967544524617e-10, "loss": 0.0003, "step": 243710 }, { "epoch": 3.987891679620388, "grad_norm": 0.0358918234705925, "learning_rate": 2.7912486048498054e-10, "loss": 0.0004, "step": 243720 }, { "epoch": 3.9880553055714634, "grad_norm": 0.001573287765495479, "learning_rate": 2.7163198558111116e-10, "loss": 0.0002, "step": 243730 }, { "epoch": 3.9882189315225394, "grad_norm": 0.03498284891247749, "learning_rate": 2.642410508868487e-10, "loss": 0.0004, "step": 243740 }, { "epoch": 3.9883825574736154, "grad_norm": 0.0192811731249094, "learning_rate": 2.5695205655207334e-10, "loss": 0.0003, "step": 243750 }, { "epoch": 3.988546183424691, "grad_norm": 0.03451351448893547, "learning_rate": 2.4976500272611005e-10, "loss": 0.0002, "step": 243760 }, { "epoch": 3.988709809375767, "grad_norm": 0.003629684215411544, "learning_rate": 2.4267988955550825e-10, "loss": 0.0004, "step": 243770 }, { "epoch": 3.988873435326843, "grad_norm": 0.029761267825961113, "learning_rate": 2.3569671718459694e-10, "loss": 0.0002, "step": 243780 }, { "epoch": 3.9890370612779185, "grad_norm": 0.0006021169829182327, "learning_rate": 2.2881548575548472e-10, "loss": 0.0002, "step": 243790 }, { "epoch": 3.9892006872289945, "grad_norm": 0.005953791085630655, "learning_rate": 2.2203619540861476e-10, "loss": 0.0001, "step": 243800 }, { "epoch": 3.9893643131800705, "grad_norm": 0.03431292995810509, "learning_rate": 2.1535884628276493e-10, "loss": 0.0005, "step": 243810 }, { "epoch": 3.989527939131146, "grad_norm": 0.010612842626869678, "learning_rate": 2.0878343851282733e-10, "loss": 0.0005, "step": 243820 }, { "epoch": 3.989691565082222, "grad_norm": 0.025403108447790146, "learning_rate": 2.023099722342492e-10, "loss": 0.0002, "step": 243830 }, { "epoch": 3.989855191033298, "grad_norm": 0.08986813575029373, "learning_rate": 1.9593844757803682e-10, "loss": 0.0002, "step": 243840 }, { "epoch": 3.9900188169843736, "grad_norm": 0.03387116640806198, "learning_rate": 1.8966886467464142e-10, "loss": 0.0002, "step": 243850 }, { "epoch": 3.9901824429354495, "grad_norm": 0.0011752824066206813, "learning_rate": 1.835012236517386e-10, "loss": 0.0001, "step": 243860 }, { "epoch": 3.9903460688865255, "grad_norm": 0.00015669589629396796, "learning_rate": 1.7743552463478363e-10, "loss": 0.0003, "step": 243870 }, { "epoch": 3.990509694837601, "grad_norm": 0.013308105058968067, "learning_rate": 1.7147176774812148e-10, "loss": 0.0004, "step": 243880 }, { "epoch": 3.990673320788677, "grad_norm": 0.002703318605199456, "learning_rate": 1.656099531127664e-10, "loss": 0.0003, "step": 243890 }, { "epoch": 3.990836946739753, "grad_norm": 0.012864634394645691, "learning_rate": 1.598500808480674e-10, "loss": 0.0001, "step": 243900 }, { "epoch": 3.9910005726908286, "grad_norm": 0.08055228739976883, "learning_rate": 1.5419215107226327e-10, "loss": 0.0002, "step": 243910 }, { "epoch": 3.9911641986419046, "grad_norm": 0.005415713880211115, "learning_rate": 1.4863616390026202e-10, "loss": 0.0001, "step": 243920 }, { "epoch": 3.9913278245929806, "grad_norm": 0.0005025346181355417, "learning_rate": 1.4318211944530646e-10, "loss": 0.0001, "step": 243930 }, { "epoch": 3.991491450544056, "grad_norm": 0.0225909985601902, "learning_rate": 1.3783001781841886e-10, "loss": 0.0001, "step": 243940 }, { "epoch": 3.991655076495132, "grad_norm": 0.0008216250571422279, "learning_rate": 1.3257985912895622e-10, "loss": 0.0004, "step": 243950 }, { "epoch": 3.991818702446208, "grad_norm": 0.0021426579914987087, "learning_rate": 1.2743164348405502e-10, "loss": 0.0003, "step": 243960 }, { "epoch": 3.9919823283972837, "grad_norm": 0.0007057894836179912, "learning_rate": 1.2238537098863136e-10, "loss": 0.0004, "step": 243970 }, { "epoch": 3.9921459543483597, "grad_norm": 0.021324673667550087, "learning_rate": 1.1744104174538084e-10, "loss": 0.0004, "step": 243980 }, { "epoch": 3.9923095802994357, "grad_norm": 0.016321830451488495, "learning_rate": 1.125986558553338e-10, "loss": 0.0003, "step": 243990 }, { "epoch": 3.9924732062505113, "grad_norm": 0.008008908480405807, "learning_rate": 1.0785821341730007e-10, "loss": 0.0003, "step": 244000 }, { "epoch": 3.9924732062505113, "eval_loss": 0.0008601969457231462, "eval_runtime": 5.3358, "eval_samples_per_second": 37.483, "eval_steps_per_second": 9.371, "step": 244000 }, { "epoch": 3.9926368322015873, "grad_norm": 0.026650747284293175, "learning_rate": 1.0321971452786905e-10, "loss": 0.0002, "step": 244010 }, { "epoch": 3.9928004581526633, "grad_norm": 0.010422932915389538, "learning_rate": 9.86831592814097e-11, "loss": 0.0007, "step": 244020 }, { "epoch": 3.992964084103739, "grad_norm": 0.07313695549964905, "learning_rate": 9.424854777007053e-11, "loss": 0.0001, "step": 244030 }, { "epoch": 3.993127710054815, "grad_norm": 0.008727041073143482, "learning_rate": 8.991588008544494e-11, "loss": 0.0002, "step": 244040 }, { "epoch": 3.9932913360058904, "grad_norm": 0.03752489760518074, "learning_rate": 8.568515631468543e-11, "loss": 0.0003, "step": 244050 }, { "epoch": 3.9934549619569664, "grad_norm": 0.0013455359730869532, "learning_rate": 8.155637654494452e-11, "loss": 0.0003, "step": 244060 }, { "epoch": 3.9936185879080424, "grad_norm": 0.016207801178097725, "learning_rate": 7.752954085948894e-11, "loss": 0.0001, "step": 244070 }, { "epoch": 3.993782213859118, "grad_norm": 0.0008094719378277659, "learning_rate": 7.360464934103029e-11, "loss": 0.0003, "step": 244080 }, { "epoch": 3.993945839810194, "grad_norm": 0.03817819431424141, "learning_rate": 6.978170207005975e-11, "loss": 0.0002, "step": 244090 }, { "epoch": 3.9941094657612695, "grad_norm": 0.0007480971980839968, "learning_rate": 6.606069912318269e-11, "loss": 0.0004, "step": 244100 }, { "epoch": 3.9942730917123455, "grad_norm": 0.03004487417638302, "learning_rate": 6.244164057755964e-11, "loss": 0.0002, "step": 244110 }, { "epoch": 3.9944367176634215, "grad_norm": 0.0069770426489412785, "learning_rate": 5.892452650591018e-11, "loss": 0.0003, "step": 244120 }, { "epoch": 3.994600343614497, "grad_norm": 0.0013345228508114815, "learning_rate": 5.550935698095394e-11, "loss": 0.0004, "step": 244130 }, { "epoch": 3.994763969565573, "grad_norm": 0.039497148245573044, "learning_rate": 5.2196132072079854e-11, "loss": 0.0002, "step": 244140 }, { "epoch": 3.994927595516649, "grad_norm": 0.064512237906456, "learning_rate": 4.898485184590129e-11, "loss": 0.0003, "step": 244150 }, { "epoch": 3.9950912214677246, "grad_norm": 0.08065709471702576, "learning_rate": 4.587551636903165e-11, "loss": 0.0002, "step": 244160 }, { "epoch": 3.9952548474188005, "grad_norm": 0.06596402078866959, "learning_rate": 4.286812570475363e-11, "loss": 0.0009, "step": 244170 }, { "epoch": 3.9954184733698765, "grad_norm": 0.009247119538486004, "learning_rate": 3.9962679913574386e-11, "loss": 0.0002, "step": 244180 }, { "epoch": 3.995582099320952, "grad_norm": 0.0016095158644020557, "learning_rate": 3.715917905544597e-11, "loss": 0.0001, "step": 244190 }, { "epoch": 3.995745725272028, "grad_norm": 0.009042998775839806, "learning_rate": 3.4457623187544865e-11, "loss": 0.0001, "step": 244200 }, { "epoch": 3.995909351223104, "grad_norm": 0.0025997310876846313, "learning_rate": 3.1858012364272e-11, "loss": 0.0003, "step": 244210 }, { "epoch": 3.9960729771741796, "grad_norm": 0.0003900259907823056, "learning_rate": 2.936034663947318e-11, "loss": 0.0004, "step": 244220 }, { "epoch": 3.9962366031252556, "grad_norm": 0.0025825893972069025, "learning_rate": 2.696462606366357e-11, "loss": 0.0001, "step": 244230 }, { "epoch": 3.9964002290763316, "grad_norm": 0.00562541838735342, "learning_rate": 2.467085068513786e-11, "loss": 0.0002, "step": 244240 }, { "epoch": 3.996563855027407, "grad_norm": 0.004040143918246031, "learning_rate": 2.2479020552190757e-11, "loss": 0.0009, "step": 244250 }, { "epoch": 3.996727480978483, "grad_norm": 0.00849115289747715, "learning_rate": 2.0389135708120954e-11, "loss": 0.0003, "step": 244260 }, { "epoch": 3.996891106929559, "grad_norm": 0.0014346996322274208, "learning_rate": 1.8401196195672044e-11, "loss": 0.0002, "step": 244270 }, { "epoch": 3.9970547328806347, "grad_norm": 0.00041910685831680894, "learning_rate": 1.6515202055922273e-11, "loss": 0.0002, "step": 244280 }, { "epoch": 3.9972183588317107, "grad_norm": 0.007005359977483749, "learning_rate": 1.473115332717434e-11, "loss": 0.0003, "step": 244290 }, { "epoch": 3.9973819847827867, "grad_norm": 0.014783843420445919, "learning_rate": 1.3049050045510492e-11, "loss": 0.0005, "step": 244300 }, { "epoch": 3.9975456107338623, "grad_norm": 0.05877729132771492, "learning_rate": 1.1468892245902752e-11, "loss": 0.0003, "step": 244310 }, { "epoch": 3.9977092366849383, "grad_norm": 0.010746563784778118, "learning_rate": 9.990679959992478e-12, "loss": 0.0002, "step": 244320 }, { "epoch": 3.9978728626360143, "grad_norm": 0.0069087049923837185, "learning_rate": 8.61441321775569e-12, "loss": 0.0001, "step": 244330 }, { "epoch": 3.99803648858709, "grad_norm": 0.050299908965826035, "learning_rate": 7.340092047503078e-12, "loss": 0.0006, "step": 244340 }, { "epoch": 3.998200114538166, "grad_norm": 0.002179265022277832, "learning_rate": 6.167716475324881e-12, "loss": 0.0004, "step": 244350 }, { "epoch": 3.998363740489242, "grad_norm": 0.014341484755277634, "learning_rate": 5.097286525090894e-12, "loss": 0.0004, "step": 244360 }, { "epoch": 3.9985273664403174, "grad_norm": 0.0012595306616276503, "learning_rate": 4.1288022184504675e-12, "loss": 0.0004, "step": 244370 }, { "epoch": 3.9986909923913934, "grad_norm": 0.0005419797380454838, "learning_rate": 3.2622635759427257e-12, "loss": 0.0002, "step": 244380 }, { "epoch": 3.9988546183424694, "grad_norm": 0.022302644327282906, "learning_rate": 2.4976706136659036e-12, "loss": 0.0004, "step": 244390 }, { "epoch": 3.999018244293545, "grad_norm": 0.002750069834291935, "learning_rate": 1.8350233488284573e-12, "loss": 0.0001, "step": 244400 }, { "epoch": 3.999181870244621, "grad_norm": 0.08859279751777649, "learning_rate": 1.274321793642841e-12, "loss": 0.0003, "step": 244410 }, { "epoch": 3.999345496195697, "grad_norm": 0.039956219494342804, "learning_rate": 8.155659603215072e-13, "loss": 0.0002, "step": 244420 }, { "epoch": 3.9995091221467725, "grad_norm": 0.0011573543306440115, "learning_rate": 4.587558583013519e-13, "loss": 0.0004, "step": 244430 }, { "epoch": 3.9996727480978485, "grad_norm": 0.0012487383792176843, "learning_rate": 2.0389149424371314e-13, "loss": 0.0003, "step": 244440 }, { "epoch": 3.999836374048924, "grad_norm": 0.0016772615490481257, "learning_rate": 5.097287369970616e-14, "loss": 0.0002, "step": 244450 }, { "epoch": 4.0, "grad_norm": 0.01218071673065424, "learning_rate": 0.0, "loss": 0.0003, "step": 244460 }, { "epoch": 4.0, "step": 244460, "total_flos": 2.849849363030396e+19, "train_loss": 0.002720164652114683, "train_runtime": 243138.9431, "train_samples_per_second": 8.043, "train_steps_per_second": 1.005 } ], "logging_steps": 10, "max_steps": 244460, "num_input_tokens_seen": 0, "num_train_epochs": 4, "save_steps": 62000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 2.849849363030396e+19, "train_batch_size": 1, "trial_name": null, "trial_params": null }