{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 54979, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00018188762982229579, "grad_norm": 5.18735408782959, "learning_rate": 5.454545454545455e-08, "loss": 0.1145, "step": 10 }, { "epoch": 0.00036377525964459157, "grad_norm": 5.220719337463379, "learning_rate": 1.1515151515151516e-07, "loss": 0.1171, "step": 20 }, { "epoch": 0.0005456628894668873, "grad_norm": 3.1081440448760986, "learning_rate": 1.7575757575757576e-07, "loss": 0.1147, "step": 30 }, { "epoch": 0.0007275505192891831, "grad_norm": 3.4854378700256348, "learning_rate": 2.3636363636363637e-07, "loss": 0.1179, "step": 40 }, { "epoch": 0.000909438149111479, "grad_norm": 2.5699403285980225, "learning_rate": 2.96969696969697e-07, "loss": 0.1228, "step": 50 }, { "epoch": 0.0010913257789337747, "grad_norm": 2.332740545272827, "learning_rate": 3.5757575757575764e-07, "loss": 0.122, "step": 60 }, { "epoch": 0.0012732134087560705, "grad_norm": 2.556152582168579, "learning_rate": 4.181818181818182e-07, "loss": 0.1162, "step": 70 }, { "epoch": 0.0014551010385783663, "grad_norm": 2.0353899002075195, "learning_rate": 4.787878787878789e-07, "loss": 0.1163, "step": 80 }, { "epoch": 0.001636988668400662, "grad_norm": 2.6888883113861084, "learning_rate": 5.393939393939395e-07, "loss": 0.1137, "step": 90 }, { "epoch": 0.001818876298222958, "grad_norm": 2.19909930229187, "learning_rate": 6.000000000000001e-07, "loss": 0.1141, "step": 100 }, { "epoch": 0.0020007639280452535, "grad_norm": 2.2978103160858154, "learning_rate": 6.606060606060606e-07, "loss": 0.1143, "step": 110 }, { "epoch": 0.0021826515578675493, "grad_norm": 2.215224504470825, "learning_rate": 7.212121212121213e-07, "loss": 0.1159, "step": 120 }, { "epoch": 0.002364539187689845, "grad_norm": 2.409761428833008, "learning_rate": 7.818181818181819e-07, "loss": 0.1121, "step": 130 }, { "epoch": 0.002546426817512141, "grad_norm": 1.8680859804153442, "learning_rate": 8.424242424242425e-07, "loss": 0.1135, "step": 140 }, { "epoch": 0.0027283144473344368, "grad_norm": 2.9763786792755127, "learning_rate": 9.030303030303031e-07, "loss": 0.116, "step": 150 }, { "epoch": 0.0029102020771567326, "grad_norm": 2.1952145099639893, "learning_rate": 9.636363636363636e-07, "loss": 0.1177, "step": 160 }, { "epoch": 0.0030920897069790284, "grad_norm": 2.4027280807495117, "learning_rate": 1.0242424242424242e-06, "loss": 0.1151, "step": 170 }, { "epoch": 0.003273977336801324, "grad_norm": 3.690617561340332, "learning_rate": 1.084848484848485e-06, "loss": 0.1217, "step": 180 }, { "epoch": 0.00345586496662362, "grad_norm": 2.6723642349243164, "learning_rate": 1.1454545454545457e-06, "loss": 0.1178, "step": 190 }, { "epoch": 0.003637752596445916, "grad_norm": 3.0747361183166504, "learning_rate": 1.206060606060606e-06, "loss": 0.1183, "step": 200 }, { "epoch": 0.0038196402262682116, "grad_norm": 2.259082555770874, "learning_rate": 1.2666666666666669e-06, "loss": 0.1153, "step": 210 }, { "epoch": 0.004001527856090507, "grad_norm": 1.9443049430847168, "learning_rate": 1.3272727272727273e-06, "loss": 0.115, "step": 220 }, { "epoch": 0.004183415485912803, "grad_norm": 1.9824458360671997, "learning_rate": 1.3878787878787881e-06, "loss": 0.1126, "step": 230 }, { "epoch": 0.004365303115735099, "grad_norm": 2.2282392978668213, "learning_rate": 1.4484848484848485e-06, "loss": 0.1179, "step": 240 }, { "epoch": 0.004547190745557395, "grad_norm": 2.8338794708251953, "learning_rate": 1.5090909090909091e-06, "loss": 0.1198, "step": 250 }, { "epoch": 0.00472907837537969, "grad_norm": 2.1649763584136963, "learning_rate": 1.56969696969697e-06, "loss": 0.116, "step": 260 }, { "epoch": 0.0049109660052019865, "grad_norm": 2.326118230819702, "learning_rate": 1.6303030303030303e-06, "loss": 0.1203, "step": 270 }, { "epoch": 0.005092853635024282, "grad_norm": 2.5504038333892822, "learning_rate": 1.6909090909090912e-06, "loss": 0.1175, "step": 280 }, { "epoch": 0.005274741264846578, "grad_norm": 2.558213710784912, "learning_rate": 1.7515151515151516e-06, "loss": 0.1221, "step": 290 }, { "epoch": 0.0054566288946688735, "grad_norm": 1.8130780458450317, "learning_rate": 1.8121212121212124e-06, "loss": 0.1162, "step": 300 }, { "epoch": 0.00563851652449117, "grad_norm": 3.928727149963379, "learning_rate": 1.872727272727273e-06, "loss": 0.122, "step": 310 }, { "epoch": 0.005820404154313465, "grad_norm": 2.4525771141052246, "learning_rate": 1.9333333333333336e-06, "loss": 0.1177, "step": 320 }, { "epoch": 0.0060022917841357605, "grad_norm": 1.919840931892395, "learning_rate": 1.993939393939394e-06, "loss": 0.1236, "step": 330 }, { "epoch": 0.006184179413958057, "grad_norm": 2.3171262741088867, "learning_rate": 2.054545454545455e-06, "loss": 0.1218, "step": 340 }, { "epoch": 0.006366067043780352, "grad_norm": 3.3264098167419434, "learning_rate": 2.1151515151515152e-06, "loss": 0.1194, "step": 350 }, { "epoch": 0.006547954673602648, "grad_norm": 2.5736546516418457, "learning_rate": 2.175757575757576e-06, "loss": 0.123, "step": 360 }, { "epoch": 0.006729842303424944, "grad_norm": 2.199043035507202, "learning_rate": 2.2363636363636364e-06, "loss": 0.1221, "step": 370 }, { "epoch": 0.00691172993324724, "grad_norm": 2.095303773880005, "learning_rate": 2.2969696969696973e-06, "loss": 0.1246, "step": 380 }, { "epoch": 0.007093617563069535, "grad_norm": 2.0196070671081543, "learning_rate": 2.3575757575757577e-06, "loss": 0.1216, "step": 390 }, { "epoch": 0.007275505192891832, "grad_norm": 2.2052719593048096, "learning_rate": 2.4181818181818185e-06, "loss": 0.1259, "step": 400 }, { "epoch": 0.007457392822714127, "grad_norm": 1.9543758630752563, "learning_rate": 2.478787878787879e-06, "loss": 0.1227, "step": 410 }, { "epoch": 0.007639280452536423, "grad_norm": 1.8035253286361694, "learning_rate": 2.5393939393939397e-06, "loss": 0.1232, "step": 420 }, { "epoch": 0.007821168082358719, "grad_norm": 3.0883185863494873, "learning_rate": 2.6e-06, "loss": 0.1244, "step": 430 }, { "epoch": 0.008003055712181014, "grad_norm": 6.68180513381958, "learning_rate": 2.660606060606061e-06, "loss": 0.126, "step": 440 }, { "epoch": 0.008184943342003311, "grad_norm": 2.4740474224090576, "learning_rate": 2.7212121212121213e-06, "loss": 0.1285, "step": 450 }, { "epoch": 0.008366830971825607, "grad_norm": 2.0251426696777344, "learning_rate": 2.781818181818182e-06, "loss": 0.1223, "step": 460 }, { "epoch": 0.008548718601647902, "grad_norm": 2.886218547821045, "learning_rate": 2.8424242424242425e-06, "loss": 0.1234, "step": 470 }, { "epoch": 0.008730606231470197, "grad_norm": 2.072406768798828, "learning_rate": 2.903030303030303e-06, "loss": 0.1264, "step": 480 }, { "epoch": 0.008912493861292493, "grad_norm": 2.0442657470703125, "learning_rate": 2.963636363636364e-06, "loss": 0.1277, "step": 490 }, { "epoch": 0.00909438149111479, "grad_norm": 2.058281421661377, "learning_rate": 3.0242424242424246e-06, "loss": 0.1281, "step": 500 }, { "epoch": 0.009276269120937085, "grad_norm": 2.6912970542907715, "learning_rate": 3.084848484848485e-06, "loss": 0.13, "step": 510 }, { "epoch": 0.00945815675075938, "grad_norm": 2.1201064586639404, "learning_rate": 3.145454545454546e-06, "loss": 0.1282, "step": 520 }, { "epoch": 0.009640044380581676, "grad_norm": 1.9643760919570923, "learning_rate": 3.2060606060606066e-06, "loss": 0.1301, "step": 530 }, { "epoch": 0.009821932010403973, "grad_norm": 1.681581974029541, "learning_rate": 3.266666666666667e-06, "loss": 0.1267, "step": 540 }, { "epoch": 0.010003819640226268, "grad_norm": 1.9110606908798218, "learning_rate": 3.3272727272727274e-06, "loss": 0.13, "step": 550 }, { "epoch": 0.010185707270048564, "grad_norm": 2.081155776977539, "learning_rate": 3.3878787878787882e-06, "loss": 0.1331, "step": 560 }, { "epoch": 0.01036759489987086, "grad_norm": 2.167654275894165, "learning_rate": 3.4484848484848486e-06, "loss": 0.1307, "step": 570 }, { "epoch": 0.010549482529693156, "grad_norm": 2.2313344478607178, "learning_rate": 3.509090909090909e-06, "loss": 0.1291, "step": 580 }, { "epoch": 0.010731370159515452, "grad_norm": 1.7115669250488281, "learning_rate": 3.5696969696969703e-06, "loss": 0.1298, "step": 590 }, { "epoch": 0.010913257789337747, "grad_norm": 2.271777391433716, "learning_rate": 3.6303030303030307e-06, "loss": 0.1306, "step": 600 }, { "epoch": 0.011095145419160042, "grad_norm": 1.9834517240524292, "learning_rate": 3.690909090909091e-06, "loss": 0.137, "step": 610 }, { "epoch": 0.01127703304898234, "grad_norm": 1.9907361268997192, "learning_rate": 3.7515151515151515e-06, "loss": 0.134, "step": 620 }, { "epoch": 0.011458920678804635, "grad_norm": 1.567321538925171, "learning_rate": 3.8121212121212127e-06, "loss": 0.1316, "step": 630 }, { "epoch": 0.01164080830862693, "grad_norm": 1.7078834772109985, "learning_rate": 3.872727272727273e-06, "loss": 0.1323, "step": 640 }, { "epoch": 0.011822695938449226, "grad_norm": 2.111600399017334, "learning_rate": 3.9333333333333335e-06, "loss": 0.1372, "step": 650 }, { "epoch": 0.012004583568271521, "grad_norm": 2.176180362701416, "learning_rate": 3.993939393939394e-06, "loss": 0.1397, "step": 660 }, { "epoch": 0.012186471198093818, "grad_norm": 1.845686435699463, "learning_rate": 4.054545454545455e-06, "loss": 0.1348, "step": 670 }, { "epoch": 0.012368358827916114, "grad_norm": 1.973986029624939, "learning_rate": 4.115151515151515e-06, "loss": 0.1353, "step": 680 }, { "epoch": 0.012550246457738409, "grad_norm": 1.8619797229766846, "learning_rate": 4.175757575757576e-06, "loss": 0.135, "step": 690 }, { "epoch": 0.012732134087560704, "grad_norm": 1.6131970882415771, "learning_rate": 4.236363636363637e-06, "loss": 0.1353, "step": 700 }, { "epoch": 0.012914021717383001, "grad_norm": 1.9592031240463257, "learning_rate": 4.296969696969698e-06, "loss": 0.1391, "step": 710 }, { "epoch": 0.013095909347205297, "grad_norm": 2.0887081623077393, "learning_rate": 4.3575757575757576e-06, "loss": 0.1368, "step": 720 }, { "epoch": 0.013277796977027592, "grad_norm": 1.7590745687484741, "learning_rate": 4.418181818181818e-06, "loss": 0.1439, "step": 730 }, { "epoch": 0.013459684606849888, "grad_norm": 1.8608827590942383, "learning_rate": 4.478787878787879e-06, "loss": 0.1445, "step": 740 }, { "epoch": 0.013641572236672185, "grad_norm": 1.9522223472595215, "learning_rate": 4.539393939393939e-06, "loss": 0.1436, "step": 750 }, { "epoch": 0.01382345986649448, "grad_norm": 1.641202449798584, "learning_rate": 4.600000000000001e-06, "loss": 0.1419, "step": 760 }, { "epoch": 0.014005347496316775, "grad_norm": 1.7941749095916748, "learning_rate": 4.660606060606061e-06, "loss": 0.1423, "step": 770 }, { "epoch": 0.01418723512613907, "grad_norm": 1.5595533847808838, "learning_rate": 4.721212121212122e-06, "loss": 0.1414, "step": 780 }, { "epoch": 0.014369122755961366, "grad_norm": 2.0197200775146484, "learning_rate": 4.7818181818181825e-06, "loss": 0.1426, "step": 790 }, { "epoch": 0.014551010385783663, "grad_norm": 2.1169183254241943, "learning_rate": 4.842424242424243e-06, "loss": 0.1445, "step": 800 }, { "epoch": 0.014732898015605959, "grad_norm": 2.0281026363372803, "learning_rate": 4.903030303030303e-06, "loss": 0.1457, "step": 810 }, { "epoch": 0.014914785645428254, "grad_norm": 1.930567979812622, "learning_rate": 4.963636363636364e-06, "loss": 0.1416, "step": 820 }, { "epoch": 0.01509667327525055, "grad_norm": 1.7667593955993652, "learning_rate": 5.024242424242425e-06, "loss": 0.1451, "step": 830 }, { "epoch": 0.015278560905072847, "grad_norm": 1.8319926261901855, "learning_rate": 5.084848484848486e-06, "loss": 0.1466, "step": 840 }, { "epoch": 0.015460448534895142, "grad_norm": 1.95608389377594, "learning_rate": 5.145454545454546e-06, "loss": 0.1493, "step": 850 }, { "epoch": 0.015642336164717437, "grad_norm": 1.8963232040405273, "learning_rate": 5.2060606060606065e-06, "loss": 0.1447, "step": 860 }, { "epoch": 0.015824223794539734, "grad_norm": 1.7617801427841187, "learning_rate": 5.2666666666666665e-06, "loss": 0.1444, "step": 870 }, { "epoch": 0.016006111424362028, "grad_norm": 2.2385365962982178, "learning_rate": 5.327272727272727e-06, "loss": 0.1444, "step": 880 }, { "epoch": 0.016187999054184325, "grad_norm": 1.8855175971984863, "learning_rate": 5.387878787878789e-06, "loss": 0.1483, "step": 890 }, { "epoch": 0.016369886684006622, "grad_norm": 2.088718891143799, "learning_rate": 5.448484848484848e-06, "loss": 0.1521, "step": 900 }, { "epoch": 0.016551774313828916, "grad_norm": 2.0380139350891113, "learning_rate": 5.50909090909091e-06, "loss": 0.15, "step": 910 }, { "epoch": 0.016733661943651213, "grad_norm": 1.9570249319076538, "learning_rate": 5.569696969696971e-06, "loss": 0.1483, "step": 920 }, { "epoch": 0.016915549573473507, "grad_norm": 1.9929877519607544, "learning_rate": 5.630303030303031e-06, "loss": 0.1489, "step": 930 }, { "epoch": 0.017097437203295804, "grad_norm": 1.7221827507019043, "learning_rate": 5.690909090909091e-06, "loss": 0.153, "step": 940 }, { "epoch": 0.0172793248331181, "grad_norm": 1.6864351034164429, "learning_rate": 5.751515151515152e-06, "loss": 0.1524, "step": 950 }, { "epoch": 0.017461212462940395, "grad_norm": 1.8310917615890503, "learning_rate": 5.812121212121212e-06, "loss": 0.149, "step": 960 }, { "epoch": 0.01764310009276269, "grad_norm": 2.0041658878326416, "learning_rate": 5.872727272727273e-06, "loss": 0.1521, "step": 970 }, { "epoch": 0.017824987722584985, "grad_norm": 1.7045669555664062, "learning_rate": 5.933333333333335e-06, "loss": 0.155, "step": 980 }, { "epoch": 0.018006875352407282, "grad_norm": 1.9028352499008179, "learning_rate": 5.993939393939394e-06, "loss": 0.1575, "step": 990 }, { "epoch": 0.01818876298222958, "grad_norm": 1.7632054090499878, "learning_rate": 6.0545454545454555e-06, "loss": 0.1537, "step": 1000 }, { "epoch": 0.018370650612051873, "grad_norm": 1.9027011394500732, "learning_rate": 6.115151515151516e-06, "loss": 0.1512, "step": 1010 }, { "epoch": 0.01855253824187417, "grad_norm": 1.8169986009597778, "learning_rate": 6.175757575757576e-06, "loss": 0.1552, "step": 1020 }, { "epoch": 0.018734425871696467, "grad_norm": 1.7264984846115112, "learning_rate": 6.236363636363637e-06, "loss": 0.1571, "step": 1030 }, { "epoch": 0.01891631350151876, "grad_norm": 1.7387828826904297, "learning_rate": 6.296969696969697e-06, "loss": 0.1567, "step": 1040 }, { "epoch": 0.019098201131341058, "grad_norm": 1.7405774593353271, "learning_rate": 6.357575757575758e-06, "loss": 0.1562, "step": 1050 }, { "epoch": 0.019280088761163352, "grad_norm": 1.9967329502105713, "learning_rate": 6.418181818181819e-06, "loss": 0.1574, "step": 1060 }, { "epoch": 0.01946197639098565, "grad_norm": 1.5799033641815186, "learning_rate": 6.478787878787879e-06, "loss": 0.1587, "step": 1070 }, { "epoch": 0.019643864020807946, "grad_norm": 1.7459362745285034, "learning_rate": 6.5393939393939395e-06, "loss": 0.156, "step": 1080 }, { "epoch": 0.01982575165063024, "grad_norm": 1.938998818397522, "learning_rate": 6.600000000000001e-06, "loss": 0.1558, "step": 1090 }, { "epoch": 0.020007639280452537, "grad_norm": 1.787226676940918, "learning_rate": 6.660606060606061e-06, "loss": 0.1549, "step": 1100 }, { "epoch": 0.02018952691027483, "grad_norm": 1.8064942359924316, "learning_rate": 6.721212121212122e-06, "loss": 0.1587, "step": 1110 }, { "epoch": 0.020371414540097128, "grad_norm": 1.8533458709716797, "learning_rate": 6.781818181818183e-06, "loss": 0.1639, "step": 1120 }, { "epoch": 0.020553302169919425, "grad_norm": 1.5029926300048828, "learning_rate": 6.842424242424243e-06, "loss": 0.1636, "step": 1130 }, { "epoch": 0.02073518979974172, "grad_norm": 1.6515835523605347, "learning_rate": 6.903030303030304e-06, "loss": 0.1597, "step": 1140 }, { "epoch": 0.020917077429564016, "grad_norm": 1.6701991558074951, "learning_rate": 6.963636363636364e-06, "loss": 0.161, "step": 1150 }, { "epoch": 0.021098965059386313, "grad_norm": 1.6107878684997559, "learning_rate": 7.024242424242424e-06, "loss": 0.1631, "step": 1160 }, { "epoch": 0.021280852689208606, "grad_norm": 1.856719732284546, "learning_rate": 7.084848484848485e-06, "loss": 0.1606, "step": 1170 }, { "epoch": 0.021462740319030903, "grad_norm": 1.8452413082122803, "learning_rate": 7.145454545454547e-06, "loss": 0.1595, "step": 1180 }, { "epoch": 0.021644627948853197, "grad_norm": 1.7249821424484253, "learning_rate": 7.206060606060606e-06, "loss": 0.1618, "step": 1190 }, { "epoch": 0.021826515578675494, "grad_norm": 1.7419499158859253, "learning_rate": 7.266666666666668e-06, "loss": 0.1621, "step": 1200 }, { "epoch": 0.02200840320849779, "grad_norm": 2.052760362625122, "learning_rate": 7.3272727272727285e-06, "loss": 0.1679, "step": 1210 }, { "epoch": 0.022190290838320085, "grad_norm": 1.8518140316009521, "learning_rate": 7.3878787878787885e-06, "loss": 0.1603, "step": 1220 }, { "epoch": 0.022372178468142382, "grad_norm": 1.7229838371276855, "learning_rate": 7.448484848484849e-06, "loss": 0.1625, "step": 1230 }, { "epoch": 0.02255406609796468, "grad_norm": 1.7121567726135254, "learning_rate": 7.509090909090909e-06, "loss": 0.1652, "step": 1240 }, { "epoch": 0.022735953727786973, "grad_norm": 1.776513934135437, "learning_rate": 7.56969696969697e-06, "loss": 0.1672, "step": 1250 }, { "epoch": 0.02291784135760927, "grad_norm": 1.6081390380859375, "learning_rate": 7.630303030303031e-06, "loss": 0.1637, "step": 1260 }, { "epoch": 0.023099728987431564, "grad_norm": 1.5514482259750366, "learning_rate": 7.690909090909091e-06, "loss": 0.166, "step": 1270 }, { "epoch": 0.02328161661725386, "grad_norm": 1.8464757204055786, "learning_rate": 7.751515151515153e-06, "loss": 0.1679, "step": 1280 }, { "epoch": 0.023463504247076158, "grad_norm": 1.5869499444961548, "learning_rate": 7.812121212121213e-06, "loss": 0.1672, "step": 1290 }, { "epoch": 0.02364539187689845, "grad_norm": 1.6429524421691895, "learning_rate": 7.872727272727273e-06, "loss": 0.1699, "step": 1300 }, { "epoch": 0.02382727950672075, "grad_norm": 1.686038851737976, "learning_rate": 7.933333333333334e-06, "loss": 0.1677, "step": 1310 }, { "epoch": 0.024009167136543042, "grad_norm": 1.7300689220428467, "learning_rate": 7.993939393939396e-06, "loss": 0.1673, "step": 1320 }, { "epoch": 0.02419105476636534, "grad_norm": 1.577950358390808, "learning_rate": 8.054545454545454e-06, "loss": 0.1629, "step": 1330 }, { "epoch": 0.024372942396187636, "grad_norm": 1.7388083934783936, "learning_rate": 8.115151515151516e-06, "loss": 0.1664, "step": 1340 }, { "epoch": 0.02455483002600993, "grad_norm": 1.5772171020507812, "learning_rate": 8.175757575757577e-06, "loss": 0.1709, "step": 1350 }, { "epoch": 0.024736717655832227, "grad_norm": 1.7451183795928955, "learning_rate": 8.236363636363637e-06, "loss": 0.1705, "step": 1360 }, { "epoch": 0.024918605285654524, "grad_norm": 1.7142850160598755, "learning_rate": 8.296969696969697e-06, "loss": 0.1713, "step": 1370 }, { "epoch": 0.025100492915476818, "grad_norm": 1.6560472249984741, "learning_rate": 8.357575757575759e-06, "loss": 0.1693, "step": 1380 }, { "epoch": 0.025282380545299115, "grad_norm": 1.8233059644699097, "learning_rate": 8.418181818181819e-06, "loss": 0.1751, "step": 1390 }, { "epoch": 0.02546426817512141, "grad_norm": 1.787598967552185, "learning_rate": 8.478787878787879e-06, "loss": 0.1694, "step": 1400 }, { "epoch": 0.025646155804943706, "grad_norm": 1.606640100479126, "learning_rate": 8.539393939393939e-06, "loss": 0.1693, "step": 1410 }, { "epoch": 0.025828043434766003, "grad_norm": 1.634798526763916, "learning_rate": 8.6e-06, "loss": 0.1713, "step": 1420 }, { "epoch": 0.026009931064588297, "grad_norm": 1.8219115734100342, "learning_rate": 8.660606060606062e-06, "loss": 0.1703, "step": 1430 }, { "epoch": 0.026191818694410594, "grad_norm": 1.6835020780563354, "learning_rate": 8.72121212121212e-06, "loss": 0.1678, "step": 1440 }, { "epoch": 0.026373706324232887, "grad_norm": 1.5893797874450684, "learning_rate": 8.781818181818182e-06, "loss": 0.1727, "step": 1450 }, { "epoch": 0.026555593954055184, "grad_norm": 1.6859092712402344, "learning_rate": 8.842424242424244e-06, "loss": 0.172, "step": 1460 }, { "epoch": 0.02673748158387748, "grad_norm": 1.663055658340454, "learning_rate": 8.903030303030304e-06, "loss": 0.1781, "step": 1470 }, { "epoch": 0.026919369213699775, "grad_norm": 1.5171325206756592, "learning_rate": 8.963636363636364e-06, "loss": 0.1653, "step": 1480 }, { "epoch": 0.027101256843522072, "grad_norm": 1.6407140493392944, "learning_rate": 9.024242424242426e-06, "loss": 0.1724, "step": 1490 }, { "epoch": 0.02728314447334437, "grad_norm": 1.6270114183425903, "learning_rate": 9.084848484848486e-06, "loss": 0.1748, "step": 1500 }, { "epoch": 0.027465032103166663, "grad_norm": 1.9687035083770752, "learning_rate": 9.145454545454546e-06, "loss": 0.1751, "step": 1510 }, { "epoch": 0.02764691973298896, "grad_norm": 1.6873291730880737, "learning_rate": 9.206060606060607e-06, "loss": 0.1777, "step": 1520 }, { "epoch": 0.027828807362811254, "grad_norm": 1.7744474411010742, "learning_rate": 9.266666666666667e-06, "loss": 0.1765, "step": 1530 }, { "epoch": 0.02801069499263355, "grad_norm": 1.6565089225769043, "learning_rate": 9.327272727272729e-06, "loss": 0.178, "step": 1540 }, { "epoch": 0.028192582622455848, "grad_norm": 1.4741289615631104, "learning_rate": 9.387878787878789e-06, "loss": 0.1686, "step": 1550 }, { "epoch": 0.02837447025227814, "grad_norm": 1.866940975189209, "learning_rate": 9.448484848484849e-06, "loss": 0.1794, "step": 1560 }, { "epoch": 0.02855635788210044, "grad_norm": 1.5899207592010498, "learning_rate": 9.50909090909091e-06, "loss": 0.1786, "step": 1570 }, { "epoch": 0.028738245511922732, "grad_norm": 1.5176146030426025, "learning_rate": 9.56969696969697e-06, "loss": 0.1755, "step": 1580 }, { "epoch": 0.02892013314174503, "grad_norm": 1.8783880472183228, "learning_rate": 9.63030303030303e-06, "loss": 0.1743, "step": 1590 }, { "epoch": 0.029102020771567327, "grad_norm": 1.63473379611969, "learning_rate": 9.690909090909092e-06, "loss": 0.173, "step": 1600 }, { "epoch": 0.02928390840138962, "grad_norm": 1.8661726713180542, "learning_rate": 9.751515151515152e-06, "loss": 0.1782, "step": 1610 }, { "epoch": 0.029465796031211917, "grad_norm": 1.5227019786834717, "learning_rate": 9.812121212121212e-06, "loss": 0.1798, "step": 1620 }, { "epoch": 0.029647683661034215, "grad_norm": 1.8625763654708862, "learning_rate": 9.872727272727274e-06, "loss": 0.1823, "step": 1630 }, { "epoch": 0.029829571290856508, "grad_norm": 1.8941525220870972, "learning_rate": 9.933333333333334e-06, "loss": 0.1784, "step": 1640 }, { "epoch": 0.030011458920678805, "grad_norm": 1.67645263671875, "learning_rate": 9.993939393939395e-06, "loss": 0.1844, "step": 1650 }, { "epoch": 0.0301933465505011, "grad_norm": 1.7031346559524536, "learning_rate": 9.999999437803847e-06, "loss": 0.1783, "step": 1660 }, { "epoch": 0.030375234180323396, "grad_norm": 1.9023866653442383, "learning_rate": 9.99999749440994e-06, "loss": 0.1808, "step": 1670 }, { "epoch": 0.030557121810145693, "grad_norm": 1.6697500944137573, "learning_rate": 9.999994162878262e-06, "loss": 0.1812, "step": 1680 }, { "epoch": 0.030739009439967987, "grad_norm": 1.7237571477890015, "learning_rate": 9.999989443209965e-06, "loss": 0.1814, "step": 1690 }, { "epoch": 0.030920897069790284, "grad_norm": 1.6282294988632202, "learning_rate": 9.999983335406693e-06, "loss": 0.183, "step": 1700 }, { "epoch": 0.03110278469961258, "grad_norm": 1.689727783203125, "learning_rate": 9.999975839470558e-06, "loss": 0.1849, "step": 1710 }, { "epoch": 0.031284672329434875, "grad_norm": 1.548848271369934, "learning_rate": 9.999966955404166e-06, "loss": 0.1804, "step": 1720 }, { "epoch": 0.03146655995925717, "grad_norm": 2.3792967796325684, "learning_rate": 9.999956683210603e-06, "loss": 0.1853, "step": 1730 }, { "epoch": 0.03164844758907947, "grad_norm": 1.4742813110351562, "learning_rate": 9.999945022893427e-06, "loss": 0.1817, "step": 1740 }, { "epoch": 0.03183033521890176, "grad_norm": 1.469868540763855, "learning_rate": 9.99993197445669e-06, "loss": 0.1834, "step": 1750 }, { "epoch": 0.032012222848724056, "grad_norm": 1.401251196861267, "learning_rate": 9.999917537904916e-06, "loss": 0.1825, "step": 1760 }, { "epoch": 0.03219411047854635, "grad_norm": 1.7211878299713135, "learning_rate": 9.999901713243116e-06, "loss": 0.1848, "step": 1770 }, { "epoch": 0.03237599810836865, "grad_norm": 1.5446548461914062, "learning_rate": 9.999884500476786e-06, "loss": 0.1856, "step": 1780 }, { "epoch": 0.03255788573819095, "grad_norm": 1.6769757270812988, "learning_rate": 9.999865899611894e-06, "loss": 0.1818, "step": 1790 }, { "epoch": 0.032739773368013245, "grad_norm": 1.5544582605361938, "learning_rate": 9.999845910654898e-06, "loss": 0.186, "step": 1800 }, { "epoch": 0.032921660997835535, "grad_norm": 1.6814320087432861, "learning_rate": 9.999824533612735e-06, "loss": 0.184, "step": 1810 }, { "epoch": 0.03310354862765783, "grad_norm": 1.5309964418411255, "learning_rate": 9.999801768492822e-06, "loss": 0.1836, "step": 1820 }, { "epoch": 0.03328543625748013, "grad_norm": 1.5497956275939941, "learning_rate": 9.999777615303062e-06, "loss": 0.1877, "step": 1830 }, { "epoch": 0.033467323887302426, "grad_norm": 1.6039537191390991, "learning_rate": 9.999752074051832e-06, "loss": 0.1818, "step": 1840 }, { "epoch": 0.03364921151712472, "grad_norm": 1.818418264389038, "learning_rate": 9.999725144748e-06, "loss": 0.19, "step": 1850 }, { "epoch": 0.033831099146947013, "grad_norm": 1.5296515226364136, "learning_rate": 9.99969682740091e-06, "loss": 0.1892, "step": 1860 }, { "epoch": 0.03401298677676931, "grad_norm": 1.571153163909912, "learning_rate": 9.99966712202039e-06, "loss": 0.1853, "step": 1870 }, { "epoch": 0.03419487440659161, "grad_norm": 1.4336351156234741, "learning_rate": 9.999636028616749e-06, "loss": 0.1829, "step": 1880 }, { "epoch": 0.034376762036413905, "grad_norm": 1.5827103853225708, "learning_rate": 9.999603547200773e-06, "loss": 0.1873, "step": 1890 }, { "epoch": 0.0345586496662362, "grad_norm": 1.7336403131484985, "learning_rate": 9.99956967778374e-06, "loss": 0.1913, "step": 1900 }, { "epoch": 0.03474053729605849, "grad_norm": 1.6818351745605469, "learning_rate": 9.999534420377401e-06, "loss": 0.1923, "step": 1910 }, { "epoch": 0.03492242492588079, "grad_norm": 1.6289077997207642, "learning_rate": 9.999497774993994e-06, "loss": 0.186, "step": 1920 }, { "epoch": 0.035104312555703086, "grad_norm": 1.4053266048431396, "learning_rate": 9.999459741646234e-06, "loss": 0.1848, "step": 1930 }, { "epoch": 0.03528620018552538, "grad_norm": 1.6154378652572632, "learning_rate": 9.999420320347318e-06, "loss": 0.1894, "step": 1940 }, { "epoch": 0.03546808781534768, "grad_norm": 1.6070823669433594, "learning_rate": 9.99937951111093e-06, "loss": 0.1893, "step": 1950 }, { "epoch": 0.03564997544516997, "grad_norm": 1.632728099822998, "learning_rate": 9.999337313951231e-06, "loss": 0.1887, "step": 1960 }, { "epoch": 0.03583186307499227, "grad_norm": 1.3731106519699097, "learning_rate": 9.999293728882864e-06, "loss": 0.1819, "step": 1970 }, { "epoch": 0.036013750704814565, "grad_norm": 1.481343388557434, "learning_rate": 9.999248755920958e-06, "loss": 0.1912, "step": 1980 }, { "epoch": 0.03619563833463686, "grad_norm": 1.4502817392349243, "learning_rate": 9.999202395081116e-06, "loss": 0.1932, "step": 1990 }, { "epoch": 0.03637752596445916, "grad_norm": 1.4695982933044434, "learning_rate": 9.999154646379427e-06, "loss": 0.1848, "step": 2000 }, { "epoch": 0.036559413594281456, "grad_norm": 1.516126275062561, "learning_rate": 9.999105509832466e-06, "loss": 0.1888, "step": 2010 }, { "epoch": 0.036741301224103746, "grad_norm": 1.4765836000442505, "learning_rate": 9.999054985457279e-06, "loss": 0.1872, "step": 2020 }, { "epoch": 0.036923188853926044, "grad_norm": 1.583577036857605, "learning_rate": 9.999003073271405e-06, "loss": 0.1917, "step": 2030 }, { "epoch": 0.03710507648374834, "grad_norm": 1.6631110906600952, "learning_rate": 9.998949773292855e-06, "loss": 0.1855, "step": 2040 }, { "epoch": 0.03728696411357064, "grad_norm": 1.5875599384307861, "learning_rate": 9.998895085540129e-06, "loss": 0.1854, "step": 2050 }, { "epoch": 0.037468851743392935, "grad_norm": 1.3335065841674805, "learning_rate": 9.998839010032204e-06, "loss": 0.1867, "step": 2060 }, { "epoch": 0.037650739373215225, "grad_norm": 1.4972814321517944, "learning_rate": 9.998781546788542e-06, "loss": 0.1927, "step": 2070 }, { "epoch": 0.03783262700303752, "grad_norm": 1.5532270669937134, "learning_rate": 9.998722695829082e-06, "loss": 0.1863, "step": 2080 }, { "epoch": 0.03801451463285982, "grad_norm": 1.4522678852081299, "learning_rate": 9.998662457174248e-06, "loss": 0.1845, "step": 2090 }, { "epoch": 0.038196402262682116, "grad_norm": 1.5946389436721802, "learning_rate": 9.998600830844946e-06, "loss": 0.1906, "step": 2100 }, { "epoch": 0.038378289892504414, "grad_norm": 1.6205220222473145, "learning_rate": 9.998537816862564e-06, "loss": 0.1813, "step": 2110 }, { "epoch": 0.038560177522326704, "grad_norm": 1.5368601083755493, "learning_rate": 9.998473415248967e-06, "loss": 0.1909, "step": 2120 }, { "epoch": 0.038742065152149, "grad_norm": 1.548032283782959, "learning_rate": 9.998407626026505e-06, "loss": 0.1916, "step": 2130 }, { "epoch": 0.0389239527819713, "grad_norm": 1.6453593969345093, "learning_rate": 9.99834044921801e-06, "loss": 0.1871, "step": 2140 }, { "epoch": 0.039105840411793595, "grad_norm": 1.6719483137130737, "learning_rate": 9.998271884846795e-06, "loss": 0.187, "step": 2150 }, { "epoch": 0.03928772804161589, "grad_norm": 1.5126999616622925, "learning_rate": 9.998201932936653e-06, "loss": 0.1878, "step": 2160 }, { "epoch": 0.03946961567143818, "grad_norm": 1.572847604751587, "learning_rate": 9.998130593511862e-06, "loss": 0.188, "step": 2170 }, { "epoch": 0.03965150330126048, "grad_norm": 1.5013948678970337, "learning_rate": 9.998057866597176e-06, "loss": 0.1904, "step": 2180 }, { "epoch": 0.03983339093108278, "grad_norm": 1.6366902589797974, "learning_rate": 9.997983752217838e-06, "loss": 0.184, "step": 2190 }, { "epoch": 0.040015278560905074, "grad_norm": 1.5445061922073364, "learning_rate": 9.997908250399561e-06, "loss": 0.1882, "step": 2200 }, { "epoch": 0.04019716619072737, "grad_norm": 1.4088222980499268, "learning_rate": 9.997831361168554e-06, "loss": 0.1812, "step": 2210 }, { "epoch": 0.04037905382054966, "grad_norm": 1.485829472541809, "learning_rate": 9.9977530845515e-06, "loss": 0.1877, "step": 2220 }, { "epoch": 0.04056094145037196, "grad_norm": 1.3736923933029175, "learning_rate": 9.99767342057556e-06, "loss": 0.1854, "step": 2230 }, { "epoch": 0.040742829080194255, "grad_norm": 1.7130082845687866, "learning_rate": 9.997592369268381e-06, "loss": 0.1877, "step": 2240 }, { "epoch": 0.04092471671001655, "grad_norm": 1.4817094802856445, "learning_rate": 9.997509930658091e-06, "loss": 0.1891, "step": 2250 }, { "epoch": 0.04110660433983885, "grad_norm": 1.4166659116744995, "learning_rate": 9.9974261047733e-06, "loss": 0.1938, "step": 2260 }, { "epoch": 0.04128849196966115, "grad_norm": 1.4477918148040771, "learning_rate": 9.997340891643097e-06, "loss": 0.1872, "step": 2270 }, { "epoch": 0.04147037959948344, "grad_norm": 1.6571303606033325, "learning_rate": 9.997254291297056e-06, "loss": 0.1906, "step": 2280 }, { "epoch": 0.041652267229305734, "grad_norm": 1.6183624267578125, "learning_rate": 9.997166303765226e-06, "loss": 0.1924, "step": 2290 }, { "epoch": 0.04183415485912803, "grad_norm": 1.596320629119873, "learning_rate": 9.997076929078146e-06, "loss": 0.1929, "step": 2300 }, { "epoch": 0.04201604248895033, "grad_norm": 1.8356691598892212, "learning_rate": 9.996986167266833e-06, "loss": 0.189, "step": 2310 }, { "epoch": 0.042197930118772625, "grad_norm": 1.732596516609192, "learning_rate": 9.99689401836278e-06, "loss": 0.1885, "step": 2320 }, { "epoch": 0.042379817748594915, "grad_norm": 1.4517465829849243, "learning_rate": 9.996800482397968e-06, "loss": 0.1858, "step": 2330 }, { "epoch": 0.04256170537841721, "grad_norm": 1.6920828819274902, "learning_rate": 9.996705559404857e-06, "loss": 0.1883, "step": 2340 }, { "epoch": 0.04274359300823951, "grad_norm": 1.9700907468795776, "learning_rate": 9.996609249416391e-06, "loss": 0.1841, "step": 2350 }, { "epoch": 0.04292548063806181, "grad_norm": 1.3702646493911743, "learning_rate": 9.996511552465991e-06, "loss": 0.1867, "step": 2360 }, { "epoch": 0.043107368267884104, "grad_norm": 1.5118557214736938, "learning_rate": 9.99641246858756e-06, "loss": 0.1903, "step": 2370 }, { "epoch": 0.043289255897706394, "grad_norm": 1.3380440473556519, "learning_rate": 9.996311997815484e-06, "loss": 0.1914, "step": 2380 }, { "epoch": 0.04347114352752869, "grad_norm": 1.4816709756851196, "learning_rate": 9.99621014018463e-06, "loss": 0.1909, "step": 2390 }, { "epoch": 0.04365303115735099, "grad_norm": 1.5862603187561035, "learning_rate": 9.996106895730349e-06, "loss": 0.1961, "step": 2400 }, { "epoch": 0.043834918787173285, "grad_norm": 1.4945240020751953, "learning_rate": 9.996002264488466e-06, "loss": 0.1937, "step": 2410 }, { "epoch": 0.04401680641699558, "grad_norm": 1.4985971450805664, "learning_rate": 9.995896246495294e-06, "loss": 0.1932, "step": 2420 }, { "epoch": 0.04419869404681787, "grad_norm": 1.5427844524383545, "learning_rate": 9.995788841787625e-06, "loss": 0.1873, "step": 2430 }, { "epoch": 0.04438058167664017, "grad_norm": 1.3536957502365112, "learning_rate": 9.995680050402731e-06, "loss": 0.1924, "step": 2440 }, { "epoch": 0.04456246930646247, "grad_norm": 1.499593734741211, "learning_rate": 9.995569872378368e-06, "loss": 0.1917, "step": 2450 }, { "epoch": 0.044744356936284764, "grad_norm": 1.6509376764297485, "learning_rate": 9.995458307752772e-06, "loss": 0.1902, "step": 2460 }, { "epoch": 0.04492624456610706, "grad_norm": 1.5029953718185425, "learning_rate": 9.995345356564655e-06, "loss": 0.1913, "step": 2470 }, { "epoch": 0.04510813219592936, "grad_norm": 1.684455156326294, "learning_rate": 9.99523101885322e-06, "loss": 0.1964, "step": 2480 }, { "epoch": 0.04529001982575165, "grad_norm": 1.6489102840423584, "learning_rate": 9.995115294658147e-06, "loss": 0.195, "step": 2490 }, { "epoch": 0.045471907455573946, "grad_norm": 1.6504191160202026, "learning_rate": 9.994998184019591e-06, "loss": 0.1907, "step": 2500 }, { "epoch": 0.04565379508539624, "grad_norm": 1.5102119445800781, "learning_rate": 9.994879686978197e-06, "loss": 0.1862, "step": 2510 }, { "epoch": 0.04583568271521854, "grad_norm": 1.62295401096344, "learning_rate": 9.994759803575087e-06, "loss": 0.1918, "step": 2520 }, { "epoch": 0.04601757034504084, "grad_norm": 1.6990547180175781, "learning_rate": 9.994638533851864e-06, "loss": 0.1899, "step": 2530 }, { "epoch": 0.04619945797486313, "grad_norm": 1.4805915355682373, "learning_rate": 9.994515877850616e-06, "loss": 0.1939, "step": 2540 }, { "epoch": 0.046381345604685424, "grad_norm": 1.4748296737670898, "learning_rate": 9.994391835613905e-06, "loss": 0.1946, "step": 2550 }, { "epoch": 0.04656323323450772, "grad_norm": 1.3929975032806396, "learning_rate": 9.994266407184778e-06, "loss": 0.1875, "step": 2560 }, { "epoch": 0.04674512086433002, "grad_norm": 1.321103572845459, "learning_rate": 9.994139592606765e-06, "loss": 0.1928, "step": 2570 }, { "epoch": 0.046927008494152316, "grad_norm": 1.428270936012268, "learning_rate": 9.994011391923874e-06, "loss": 0.1936, "step": 2580 }, { "epoch": 0.047108896123974606, "grad_norm": 1.3334007263183594, "learning_rate": 9.993881805180596e-06, "loss": 0.1922, "step": 2590 }, { "epoch": 0.0472907837537969, "grad_norm": 1.4146026372909546, "learning_rate": 9.993750832421904e-06, "loss": 0.1894, "step": 2600 }, { "epoch": 0.0474726713836192, "grad_norm": 1.5112866163253784, "learning_rate": 9.993618473693245e-06, "loss": 0.1928, "step": 2610 }, { "epoch": 0.0476545590134415, "grad_norm": 1.5261043310165405, "learning_rate": 9.993484729040553e-06, "loss": 0.1942, "step": 2620 }, { "epoch": 0.047836446643263794, "grad_norm": 1.5727509260177612, "learning_rate": 9.993349598510247e-06, "loss": 0.1974, "step": 2630 }, { "epoch": 0.048018334273086084, "grad_norm": 1.477271556854248, "learning_rate": 9.993213082149216e-06, "loss": 0.196, "step": 2640 }, { "epoch": 0.04820022190290838, "grad_norm": 1.5518320798873901, "learning_rate": 9.993075180004841e-06, "loss": 0.1963, "step": 2650 }, { "epoch": 0.04838210953273068, "grad_norm": 1.4629360437393188, "learning_rate": 9.992935892124974e-06, "loss": 0.1958, "step": 2660 }, { "epoch": 0.048563997162552976, "grad_norm": 1.4742317199707031, "learning_rate": 9.992795218557958e-06, "loss": 0.1917, "step": 2670 }, { "epoch": 0.04874588479237527, "grad_norm": 1.7339228391647339, "learning_rate": 9.992653159352608e-06, "loss": 0.1969, "step": 2680 }, { "epoch": 0.04892777242219756, "grad_norm": 1.4217747449874878, "learning_rate": 9.992509714558223e-06, "loss": 0.1959, "step": 2690 }, { "epoch": 0.04910966005201986, "grad_norm": 1.5128493309020996, "learning_rate": 9.992364884224584e-06, "loss": 0.1966, "step": 2700 }, { "epoch": 0.04929154768184216, "grad_norm": 1.4551234245300293, "learning_rate": 9.992218668401954e-06, "loss": 0.1892, "step": 2710 }, { "epoch": 0.049473435311664454, "grad_norm": 1.3585389852523804, "learning_rate": 9.992071067141074e-06, "loss": 0.1924, "step": 2720 }, { "epoch": 0.04965532294148675, "grad_norm": 1.3882709741592407, "learning_rate": 9.991922080493167e-06, "loss": 0.19, "step": 2730 }, { "epoch": 0.04983721057130905, "grad_norm": 1.4835132360458374, "learning_rate": 9.991771708509932e-06, "loss": 0.1893, "step": 2740 }, { "epoch": 0.05001909820113134, "grad_norm": 1.5010672807693481, "learning_rate": 9.991619951243561e-06, "loss": 0.1912, "step": 2750 }, { "epoch": 0.050200985830953636, "grad_norm": 1.5885372161865234, "learning_rate": 9.991466808746714e-06, "loss": 0.1934, "step": 2760 }, { "epoch": 0.05038287346077593, "grad_norm": 1.466482162475586, "learning_rate": 9.991312281072537e-06, "loss": 0.1933, "step": 2770 }, { "epoch": 0.05056476109059823, "grad_norm": 1.4879587888717651, "learning_rate": 9.99115636827466e-06, "loss": 0.1935, "step": 2780 }, { "epoch": 0.05074664872042053, "grad_norm": 1.4405150413513184, "learning_rate": 9.990999070407186e-06, "loss": 0.1956, "step": 2790 }, { "epoch": 0.05092853635024282, "grad_norm": 1.4173868894577026, "learning_rate": 9.990840387524704e-06, "loss": 0.1937, "step": 2800 }, { "epoch": 0.051110423980065114, "grad_norm": 1.4000107049942017, "learning_rate": 9.990680319682284e-06, "loss": 0.1939, "step": 2810 }, { "epoch": 0.05129231160988741, "grad_norm": 1.6621602773666382, "learning_rate": 9.990518866935473e-06, "loss": 0.1943, "step": 2820 }, { "epoch": 0.05147419923970971, "grad_norm": 1.443499207496643, "learning_rate": 9.990356029340303e-06, "loss": 0.1942, "step": 2830 }, { "epoch": 0.051656086869532006, "grad_norm": 1.6307679414749146, "learning_rate": 9.990191806953281e-06, "loss": 0.193, "step": 2840 }, { "epoch": 0.051837974499354296, "grad_norm": 1.369185447692871, "learning_rate": 9.9900261998314e-06, "loss": 0.1911, "step": 2850 }, { "epoch": 0.05201986212917659, "grad_norm": 1.3025792837142944, "learning_rate": 9.989859208032134e-06, "loss": 0.1897, "step": 2860 }, { "epoch": 0.05220174975899889, "grad_norm": 1.4197098016738892, "learning_rate": 9.989690831613429e-06, "loss": 0.1989, "step": 2870 }, { "epoch": 0.05238363738882119, "grad_norm": 1.557252049446106, "learning_rate": 9.989521070633722e-06, "loss": 0.1984, "step": 2880 }, { "epoch": 0.052565525018643484, "grad_norm": 1.650745153427124, "learning_rate": 9.989349925151925e-06, "loss": 0.1934, "step": 2890 }, { "epoch": 0.052747412648465775, "grad_norm": 1.4879794120788574, "learning_rate": 9.98917739522743e-06, "loss": 0.1973, "step": 2900 }, { "epoch": 0.05292930027828807, "grad_norm": 1.4728347063064575, "learning_rate": 9.989003480920112e-06, "loss": 0.1929, "step": 2910 }, { "epoch": 0.05311118790811037, "grad_norm": 1.3925570249557495, "learning_rate": 9.988828182290325e-06, "loss": 0.1875, "step": 2920 }, { "epoch": 0.053293075537932666, "grad_norm": 1.4369516372680664, "learning_rate": 9.988651499398905e-06, "loss": 0.193, "step": 2930 }, { "epoch": 0.05347496316775496, "grad_norm": 1.4336378574371338, "learning_rate": 9.988473432307165e-06, "loss": 0.1953, "step": 2940 }, { "epoch": 0.05365685079757726, "grad_norm": 1.4390443563461304, "learning_rate": 9.9882939810769e-06, "loss": 0.1963, "step": 2950 }, { "epoch": 0.05383873842739955, "grad_norm": 1.422025442123413, "learning_rate": 9.98811314577039e-06, "loss": 0.1927, "step": 2960 }, { "epoch": 0.05402062605722185, "grad_norm": 1.3656927347183228, "learning_rate": 9.987930926450387e-06, "loss": 0.1941, "step": 2970 }, { "epoch": 0.054202513687044145, "grad_norm": 1.5142945051193237, "learning_rate": 9.98774732318013e-06, "loss": 0.1954, "step": 2980 }, { "epoch": 0.05438440131686644, "grad_norm": 1.3625375032424927, "learning_rate": 9.987562336023332e-06, "loss": 0.1914, "step": 2990 }, { "epoch": 0.05456628894668874, "grad_norm": 1.3754993677139282, "learning_rate": 9.987375965044193e-06, "loss": 0.193, "step": 3000 }, { "epoch": 0.05474817657651103, "grad_norm": 1.341659665107727, "learning_rate": 9.98718821030739e-06, "loss": 0.1879, "step": 3010 }, { "epoch": 0.054930064206333326, "grad_norm": 1.4178719520568848, "learning_rate": 9.986999071878081e-06, "loss": 0.1926, "step": 3020 }, { "epoch": 0.05511195183615562, "grad_norm": 1.2714202404022217, "learning_rate": 9.986808549821901e-06, "loss": 0.1989, "step": 3030 }, { "epoch": 0.05529383946597792, "grad_norm": 1.4462344646453857, "learning_rate": 9.98661664420497e-06, "loss": 0.1934, "step": 3040 }, { "epoch": 0.05547572709580022, "grad_norm": 1.5199788808822632, "learning_rate": 9.986423355093885e-06, "loss": 0.1918, "step": 3050 }, { "epoch": 0.05565761472562251, "grad_norm": 1.4134677648544312, "learning_rate": 9.986228682555723e-06, "loss": 0.1905, "step": 3060 }, { "epoch": 0.055839502355444805, "grad_norm": 1.4899158477783203, "learning_rate": 9.986032626658046e-06, "loss": 0.1946, "step": 3070 }, { "epoch": 0.0560213899852671, "grad_norm": 1.4463444948196411, "learning_rate": 9.985835187468889e-06, "loss": 0.1956, "step": 3080 }, { "epoch": 0.0562032776150894, "grad_norm": 1.3372284173965454, "learning_rate": 9.985636365056768e-06, "loss": 0.203, "step": 3090 }, { "epoch": 0.056385165244911696, "grad_norm": 1.3450403213500977, "learning_rate": 9.985436159490688e-06, "loss": 0.195, "step": 3100 }, { "epoch": 0.056567052874733986, "grad_norm": 1.5129225254058838, "learning_rate": 9.98523457084012e-06, "loss": 0.1917, "step": 3110 }, { "epoch": 0.05674894050455628, "grad_norm": 1.4894664287567139, "learning_rate": 9.985031599175027e-06, "loss": 0.202, "step": 3120 }, { "epoch": 0.05693082813437858, "grad_norm": 1.3956130743026733, "learning_rate": 9.984827244565844e-06, "loss": 0.1992, "step": 3130 }, { "epoch": 0.05711271576420088, "grad_norm": 1.3382129669189453, "learning_rate": 9.984621507083494e-06, "loss": 0.1947, "step": 3140 }, { "epoch": 0.057294603394023175, "grad_norm": 1.4366463422775269, "learning_rate": 9.984414386799371e-06, "loss": 0.196, "step": 3150 }, { "epoch": 0.057476491023845465, "grad_norm": 1.3940147161483765, "learning_rate": 9.984205883785353e-06, "loss": 0.2013, "step": 3160 }, { "epoch": 0.05765837865366776, "grad_norm": 1.4949400424957275, "learning_rate": 9.983995998113799e-06, "loss": 0.1946, "step": 3170 }, { "epoch": 0.05784026628349006, "grad_norm": 1.5186998844146729, "learning_rate": 9.983784729857545e-06, "loss": 0.1953, "step": 3180 }, { "epoch": 0.058022153913312356, "grad_norm": 1.4857509136199951, "learning_rate": 9.983572079089911e-06, "loss": 0.1898, "step": 3190 }, { "epoch": 0.05820404154313465, "grad_norm": 1.4895317554473877, "learning_rate": 9.983358045884693e-06, "loss": 0.1979, "step": 3200 }, { "epoch": 0.05838592917295695, "grad_norm": 1.5095937252044678, "learning_rate": 9.983142630316167e-06, "loss": 0.2025, "step": 3210 }, { "epoch": 0.05856781680277924, "grad_norm": 1.5274860858917236, "learning_rate": 9.982925832459092e-06, "loss": 0.1965, "step": 3220 }, { "epoch": 0.05874970443260154, "grad_norm": 1.4467787742614746, "learning_rate": 9.9827076523887e-06, "loss": 0.2005, "step": 3230 }, { "epoch": 0.058931592062423835, "grad_norm": 1.4280941486358643, "learning_rate": 9.982488090180713e-06, "loss": 0.1923, "step": 3240 }, { "epoch": 0.05911347969224613, "grad_norm": 1.4705772399902344, "learning_rate": 9.982267145911323e-06, "loss": 0.1989, "step": 3250 }, { "epoch": 0.05929536732206843, "grad_norm": 1.3121740818023682, "learning_rate": 9.982044819657204e-06, "loss": 0.2013, "step": 3260 }, { "epoch": 0.05947725495189072, "grad_norm": 1.4656754732131958, "learning_rate": 9.981821111495513e-06, "loss": 0.1916, "step": 3270 }, { "epoch": 0.059659142581713016, "grad_norm": 1.411145806312561, "learning_rate": 9.981596021503885e-06, "loss": 0.1946, "step": 3280 }, { "epoch": 0.059841030211535313, "grad_norm": 1.4107681512832642, "learning_rate": 9.981369549760433e-06, "loss": 0.1999, "step": 3290 }, { "epoch": 0.06002291784135761, "grad_norm": 1.4301096200942993, "learning_rate": 9.981141696343752e-06, "loss": 0.1981, "step": 3300 }, { "epoch": 0.06020480547117991, "grad_norm": 1.4126956462860107, "learning_rate": 9.980912461332913e-06, "loss": 0.194, "step": 3310 }, { "epoch": 0.0603866931010022, "grad_norm": 1.36647367477417, "learning_rate": 9.980681844807469e-06, "loss": 0.2005, "step": 3320 }, { "epoch": 0.060568580730824495, "grad_norm": 1.5784987211227417, "learning_rate": 9.980449846847453e-06, "loss": 0.1944, "step": 3330 }, { "epoch": 0.06075046836064679, "grad_norm": 1.3963580131530762, "learning_rate": 9.980216467533374e-06, "loss": 0.2025, "step": 3340 }, { "epoch": 0.06093235599046909, "grad_norm": 1.4201867580413818, "learning_rate": 9.979981706946226e-06, "loss": 0.1981, "step": 3350 }, { "epoch": 0.061114243620291386, "grad_norm": 1.4145359992980957, "learning_rate": 9.979745565167477e-06, "loss": 0.1928, "step": 3360 }, { "epoch": 0.061296131250113677, "grad_norm": 1.373362421989441, "learning_rate": 9.979508042279077e-06, "loss": 0.1996, "step": 3370 }, { "epoch": 0.061478018879935974, "grad_norm": 1.4794952869415283, "learning_rate": 9.979269138363455e-06, "loss": 0.1952, "step": 3380 }, { "epoch": 0.06165990650975827, "grad_norm": 1.6453673839569092, "learning_rate": 9.979028853503517e-06, "loss": 0.1989, "step": 3390 }, { "epoch": 0.06184179413958057, "grad_norm": 1.394802212715149, "learning_rate": 9.978787187782652e-06, "loss": 0.1941, "step": 3400 }, { "epoch": 0.062023681769402865, "grad_norm": 1.396864652633667, "learning_rate": 9.978544141284726e-06, "loss": 0.1911, "step": 3410 }, { "epoch": 0.06220556939922516, "grad_norm": 1.4512993097305298, "learning_rate": 9.978299714094087e-06, "loss": 0.1993, "step": 3420 }, { "epoch": 0.06238745702904745, "grad_norm": 1.4991222620010376, "learning_rate": 9.978053906295553e-06, "loss": 0.1924, "step": 3430 }, { "epoch": 0.06256934465886975, "grad_norm": 1.3346813917160034, "learning_rate": 9.977806717974434e-06, "loss": 0.1964, "step": 3440 }, { "epoch": 0.06275123228869205, "grad_norm": 1.6097724437713623, "learning_rate": 9.977558149216511e-06, "loss": 0.196, "step": 3450 }, { "epoch": 0.06293311991851434, "grad_norm": 1.5026745796203613, "learning_rate": 9.977308200108046e-06, "loss": 0.1952, "step": 3460 }, { "epoch": 0.06311500754833664, "grad_norm": 1.4450024366378784, "learning_rate": 9.977056870735781e-06, "loss": 0.1974, "step": 3470 }, { "epoch": 0.06329689517815894, "grad_norm": 1.413477897644043, "learning_rate": 9.976804161186933e-06, "loss": 0.196, "step": 3480 }, { "epoch": 0.06347878280798123, "grad_norm": 1.4108734130859375, "learning_rate": 9.976550071549204e-06, "loss": 0.1961, "step": 3490 }, { "epoch": 0.06366067043780352, "grad_norm": 1.429091215133667, "learning_rate": 9.97629460191077e-06, "loss": 0.1993, "step": 3500 }, { "epoch": 0.06384255806762582, "grad_norm": 1.470714807510376, "learning_rate": 9.976037752360288e-06, "loss": 0.1975, "step": 3510 }, { "epoch": 0.06402444569744811, "grad_norm": 1.2751047611236572, "learning_rate": 9.975779522986896e-06, "loss": 0.1996, "step": 3520 }, { "epoch": 0.06420633332727041, "grad_norm": 1.2947102785110474, "learning_rate": 9.975519913880208e-06, "loss": 0.1973, "step": 3530 }, { "epoch": 0.0643882209570927, "grad_norm": 1.384432315826416, "learning_rate": 9.975258925130313e-06, "loss": 0.1969, "step": 3540 }, { "epoch": 0.064570108586915, "grad_norm": 1.3525705337524414, "learning_rate": 9.97499655682779e-06, "loss": 0.194, "step": 3550 }, { "epoch": 0.0647519962167373, "grad_norm": 1.4675012826919556, "learning_rate": 9.974732809063683e-06, "loss": 0.1977, "step": 3560 }, { "epoch": 0.0649338838465596, "grad_norm": 1.408706545829773, "learning_rate": 9.974467681929526e-06, "loss": 0.1964, "step": 3570 }, { "epoch": 0.0651157714763819, "grad_norm": 1.398267149925232, "learning_rate": 9.974201175517327e-06, "loss": 0.1995, "step": 3580 }, { "epoch": 0.06529765910620419, "grad_norm": 1.5841115713119507, "learning_rate": 9.97393328991957e-06, "loss": 0.1949, "step": 3590 }, { "epoch": 0.06547954673602649, "grad_norm": 1.4701436758041382, "learning_rate": 9.973664025229224e-06, "loss": 0.1984, "step": 3600 }, { "epoch": 0.06566143436584877, "grad_norm": 1.4845051765441895, "learning_rate": 9.973393381539732e-06, "loss": 0.1956, "step": 3610 }, { "epoch": 0.06584332199567107, "grad_norm": 1.381178379058838, "learning_rate": 9.97312135894502e-06, "loss": 0.2001, "step": 3620 }, { "epoch": 0.06602520962549337, "grad_norm": 1.459144949913025, "learning_rate": 9.97284795753948e-06, "loss": 0.199, "step": 3630 }, { "epoch": 0.06620709725531566, "grad_norm": 1.4556596279144287, "learning_rate": 9.972573177418001e-06, "loss": 0.1957, "step": 3640 }, { "epoch": 0.06638898488513796, "grad_norm": 1.40968656539917, "learning_rate": 9.97229701867594e-06, "loss": 0.1976, "step": 3650 }, { "epoch": 0.06657087251496026, "grad_norm": 1.3727895021438599, "learning_rate": 9.972019481409128e-06, "loss": 0.1961, "step": 3660 }, { "epoch": 0.06675276014478256, "grad_norm": 1.5781313180923462, "learning_rate": 9.971740565713884e-06, "loss": 0.1972, "step": 3670 }, { "epoch": 0.06693464777460485, "grad_norm": 1.4277968406677246, "learning_rate": 9.971460271687004e-06, "loss": 0.1941, "step": 3680 }, { "epoch": 0.06711653540442715, "grad_norm": 1.3346920013427734, "learning_rate": 9.971178599425752e-06, "loss": 0.1957, "step": 3690 }, { "epoch": 0.06729842303424945, "grad_norm": 1.576047420501709, "learning_rate": 9.970895549027887e-06, "loss": 0.1994, "step": 3700 }, { "epoch": 0.06748031066407173, "grad_norm": 1.787609577178955, "learning_rate": 9.970611120591631e-06, "loss": 0.1969, "step": 3710 }, { "epoch": 0.06766219829389403, "grad_norm": 1.5408638715744019, "learning_rate": 9.970325314215693e-06, "loss": 0.1977, "step": 3720 }, { "epoch": 0.06784408592371632, "grad_norm": 1.3391757011413574, "learning_rate": 9.970038129999257e-06, "loss": 0.1987, "step": 3730 }, { "epoch": 0.06802597355353862, "grad_norm": 1.3409479856491089, "learning_rate": 9.969749568041988e-06, "loss": 0.1978, "step": 3740 }, { "epoch": 0.06820786118336092, "grad_norm": 1.480838656425476, "learning_rate": 9.969459628444025e-06, "loss": 0.1966, "step": 3750 }, { "epoch": 0.06838974881318322, "grad_norm": 1.3224890232086182, "learning_rate": 9.969168311305986e-06, "loss": 0.2017, "step": 3760 }, { "epoch": 0.06857163644300551, "grad_norm": 1.2969557046890259, "learning_rate": 9.96887561672897e-06, "loss": 0.1963, "step": 3770 }, { "epoch": 0.06875352407282781, "grad_norm": 1.5014185905456543, "learning_rate": 9.96858154481455e-06, "loss": 0.1963, "step": 3780 }, { "epoch": 0.0689354117026501, "grad_norm": 1.3806164264678955, "learning_rate": 9.968286095664783e-06, "loss": 0.2009, "step": 3790 }, { "epoch": 0.0691172993324724, "grad_norm": 1.5777373313903809, "learning_rate": 9.967989269382198e-06, "loss": 0.1975, "step": 3800 }, { "epoch": 0.0692991869622947, "grad_norm": 1.3794312477111816, "learning_rate": 9.967691066069803e-06, "loss": 0.2018, "step": 3810 }, { "epoch": 0.06948107459211698, "grad_norm": 1.3254566192626953, "learning_rate": 9.967391485831086e-06, "loss": 0.2021, "step": 3820 }, { "epoch": 0.06966296222193928, "grad_norm": 1.3208379745483398, "learning_rate": 9.96709052877001e-06, "loss": 0.1963, "step": 3830 }, { "epoch": 0.06984484985176158, "grad_norm": 1.5351831912994385, "learning_rate": 9.966788194991021e-06, "loss": 0.2028, "step": 3840 }, { "epoch": 0.07002673748158388, "grad_norm": 1.3125929832458496, "learning_rate": 9.966484484599036e-06, "loss": 0.1979, "step": 3850 }, { "epoch": 0.07020862511140617, "grad_norm": 1.495896339416504, "learning_rate": 9.966179397699457e-06, "loss": 0.1985, "step": 3860 }, { "epoch": 0.07039051274122847, "grad_norm": 1.4076049327850342, "learning_rate": 9.965872934398155e-06, "loss": 0.1952, "step": 3870 }, { "epoch": 0.07057240037105077, "grad_norm": 1.4500361680984497, "learning_rate": 9.965565094801485e-06, "loss": 0.2036, "step": 3880 }, { "epoch": 0.07075428800087306, "grad_norm": 1.3310503959655762, "learning_rate": 9.96525587901628e-06, "loss": 0.1989, "step": 3890 }, { "epoch": 0.07093617563069536, "grad_norm": 1.420925259590149, "learning_rate": 9.964945287149847e-06, "loss": 0.2004, "step": 3900 }, { "epoch": 0.07111806326051766, "grad_norm": 1.209490418434143, "learning_rate": 9.964633319309972e-06, "loss": 0.1941, "step": 3910 }, { "epoch": 0.07129995089033994, "grad_norm": 1.4402743577957153, "learning_rate": 9.964319975604918e-06, "loss": 0.1925, "step": 3920 }, { "epoch": 0.07148183852016224, "grad_norm": 1.3430827856063843, "learning_rate": 9.964005256143427e-06, "loss": 0.1958, "step": 3930 }, { "epoch": 0.07166372614998454, "grad_norm": 1.3854824304580688, "learning_rate": 9.96368916103472e-06, "loss": 0.2041, "step": 3940 }, { "epoch": 0.07184561377980683, "grad_norm": 1.4710984230041504, "learning_rate": 9.963371690388489e-06, "loss": 0.1997, "step": 3950 }, { "epoch": 0.07202750140962913, "grad_norm": 1.4233417510986328, "learning_rate": 9.963052844314908e-06, "loss": 0.2006, "step": 3960 }, { "epoch": 0.07220938903945143, "grad_norm": 1.3216158151626587, "learning_rate": 9.962732622924629e-06, "loss": 0.1978, "step": 3970 }, { "epoch": 0.07239127666927372, "grad_norm": 1.5875132083892822, "learning_rate": 9.96241102632878e-06, "loss": 0.2059, "step": 3980 }, { "epoch": 0.07257316429909602, "grad_norm": 1.3882695436477661, "learning_rate": 9.962088054638964e-06, "loss": 0.2002, "step": 3990 }, { "epoch": 0.07275505192891832, "grad_norm": 1.427248239517212, "learning_rate": 9.961763707967267e-06, "loss": 0.2008, "step": 4000 }, { "epoch": 0.07293693955874062, "grad_norm": 1.3965343236923218, "learning_rate": 9.961437986426244e-06, "loss": 0.2002, "step": 4010 }, { "epoch": 0.07311882718856291, "grad_norm": 1.5116524696350098, "learning_rate": 9.961110890128936e-06, "loss": 0.203, "step": 4020 }, { "epoch": 0.0733007148183852, "grad_norm": 1.4433568716049194, "learning_rate": 9.960782419188854e-06, "loss": 0.1977, "step": 4030 }, { "epoch": 0.07348260244820749, "grad_norm": 1.401340365409851, "learning_rate": 9.960452573719992e-06, "loss": 0.1927, "step": 4040 }, { "epoch": 0.07366449007802979, "grad_norm": 1.6165874004364014, "learning_rate": 9.960121353836813e-06, "loss": 0.1937, "step": 4050 }, { "epoch": 0.07384637770785209, "grad_norm": 1.4945926666259766, "learning_rate": 9.959788759654266e-06, "loss": 0.2003, "step": 4060 }, { "epoch": 0.07402826533767438, "grad_norm": 1.458087682723999, "learning_rate": 9.95945479128777e-06, "loss": 0.1954, "step": 4070 }, { "epoch": 0.07421015296749668, "grad_norm": 1.4807409048080444, "learning_rate": 9.959119448853224e-06, "loss": 0.2005, "step": 4080 }, { "epoch": 0.07439204059731898, "grad_norm": 1.4180232286453247, "learning_rate": 9.958782732467006e-06, "loss": 0.1985, "step": 4090 }, { "epoch": 0.07457392822714128, "grad_norm": 1.3156378269195557, "learning_rate": 9.958444642245966e-06, "loss": 0.1964, "step": 4100 }, { "epoch": 0.07475581585696357, "grad_norm": 1.363908290863037, "learning_rate": 9.958105178307431e-06, "loss": 0.2011, "step": 4110 }, { "epoch": 0.07493770348678587, "grad_norm": 1.2397117614746094, "learning_rate": 9.957764340769211e-06, "loss": 0.1958, "step": 4120 }, { "epoch": 0.07511959111660815, "grad_norm": 1.4548262357711792, "learning_rate": 9.957422129749587e-06, "loss": 0.2041, "step": 4130 }, { "epoch": 0.07530147874643045, "grad_norm": 1.475016713142395, "learning_rate": 9.957078545367316e-06, "loss": 0.2028, "step": 4140 }, { "epoch": 0.07548336637625275, "grad_norm": 1.3564308881759644, "learning_rate": 9.956733587741637e-06, "loss": 0.2031, "step": 4150 }, { "epoch": 0.07566525400607504, "grad_norm": 1.3682562112808228, "learning_rate": 9.956387256992259e-06, "loss": 0.1997, "step": 4160 }, { "epoch": 0.07584714163589734, "grad_norm": 1.3483378887176514, "learning_rate": 9.956039553239372e-06, "loss": 0.1984, "step": 4170 }, { "epoch": 0.07602902926571964, "grad_norm": 1.3411798477172852, "learning_rate": 9.955690476603643e-06, "loss": 0.1981, "step": 4180 }, { "epoch": 0.07621091689554194, "grad_norm": 1.5337358713150024, "learning_rate": 9.955340027206213e-06, "loss": 0.2047, "step": 4190 }, { "epoch": 0.07639280452536423, "grad_norm": 1.538867473602295, "learning_rate": 9.954988205168697e-06, "loss": 0.2006, "step": 4200 }, { "epoch": 0.07657469215518653, "grad_norm": 1.3451670408248901, "learning_rate": 9.954635010613194e-06, "loss": 0.2023, "step": 4210 }, { "epoch": 0.07675657978500883, "grad_norm": 1.560053825378418, "learning_rate": 9.95428044366227e-06, "loss": 0.2016, "step": 4220 }, { "epoch": 0.07693846741483111, "grad_norm": 1.4828375577926636, "learning_rate": 9.953924504438978e-06, "loss": 0.1973, "step": 4230 }, { "epoch": 0.07712035504465341, "grad_norm": 1.388662338256836, "learning_rate": 9.953567193066836e-06, "loss": 0.2038, "step": 4240 }, { "epoch": 0.0773022426744757, "grad_norm": 1.377870798110962, "learning_rate": 9.953208509669845e-06, "loss": 0.1924, "step": 4250 }, { "epoch": 0.077484130304298, "grad_norm": 1.296096920967102, "learning_rate": 9.952848454372482e-06, "loss": 0.1958, "step": 4260 }, { "epoch": 0.0776660179341203, "grad_norm": 1.2676022052764893, "learning_rate": 9.952487027299695e-06, "loss": 0.1979, "step": 4270 }, { "epoch": 0.0778479055639426, "grad_norm": 1.2610996961593628, "learning_rate": 9.952124228576916e-06, "loss": 0.1976, "step": 4280 }, { "epoch": 0.0780297931937649, "grad_norm": 1.3363850116729736, "learning_rate": 9.951760058330047e-06, "loss": 0.2045, "step": 4290 }, { "epoch": 0.07821168082358719, "grad_norm": 1.4456254243850708, "learning_rate": 9.951394516685467e-06, "loss": 0.1977, "step": 4300 }, { "epoch": 0.07839356845340949, "grad_norm": 1.4827438592910767, "learning_rate": 9.951027603770032e-06, "loss": 0.2041, "step": 4310 }, { "epoch": 0.07857545608323178, "grad_norm": 1.2539728879928589, "learning_rate": 9.950659319711075e-06, "loss": 0.2031, "step": 4320 }, { "epoch": 0.07875734371305408, "grad_norm": 1.3080464601516724, "learning_rate": 9.950289664636401e-06, "loss": 0.1928, "step": 4330 }, { "epoch": 0.07893923134287636, "grad_norm": 1.313822627067566, "learning_rate": 9.949918638674296e-06, "loss": 0.1986, "step": 4340 }, { "epoch": 0.07912111897269866, "grad_norm": 1.388775110244751, "learning_rate": 9.949546241953516e-06, "loss": 0.2006, "step": 4350 }, { "epoch": 0.07930300660252096, "grad_norm": 1.2764337062835693, "learning_rate": 9.949172474603295e-06, "loss": 0.196, "step": 4360 }, { "epoch": 0.07948489423234326, "grad_norm": 1.4924980401992798, "learning_rate": 9.948797336753348e-06, "loss": 0.1968, "step": 4370 }, { "epoch": 0.07966678186216555, "grad_norm": 1.3218250274658203, "learning_rate": 9.948420828533855e-06, "loss": 0.2028, "step": 4380 }, { "epoch": 0.07984866949198785, "grad_norm": 1.5309276580810547, "learning_rate": 9.948042950075483e-06, "loss": 0.1981, "step": 4390 }, { "epoch": 0.08003055712181015, "grad_norm": 1.3769726753234863, "learning_rate": 9.947663701509363e-06, "loss": 0.2, "step": 4400 }, { "epoch": 0.08021244475163244, "grad_norm": 1.293850302696228, "learning_rate": 9.947283082967112e-06, "loss": 0.1958, "step": 4410 }, { "epoch": 0.08039433238145474, "grad_norm": 1.5666780471801758, "learning_rate": 9.946901094580816e-06, "loss": 0.202, "step": 4420 }, { "epoch": 0.08057622001127704, "grad_norm": 1.3879072666168213, "learning_rate": 9.946517736483039e-06, "loss": 0.2009, "step": 4430 }, { "epoch": 0.08075810764109932, "grad_norm": 1.2550431489944458, "learning_rate": 9.946133008806818e-06, "loss": 0.1978, "step": 4440 }, { "epoch": 0.08093999527092162, "grad_norm": 1.1922531127929688, "learning_rate": 9.945746911685669e-06, "loss": 0.1972, "step": 4450 }, { "epoch": 0.08112188290074392, "grad_norm": 1.2960706949234009, "learning_rate": 9.945359445253578e-06, "loss": 0.1984, "step": 4460 }, { "epoch": 0.08130377053056621, "grad_norm": 1.348329782485962, "learning_rate": 9.944970609645013e-06, "loss": 0.2004, "step": 4470 }, { "epoch": 0.08148565816038851, "grad_norm": 1.370158076286316, "learning_rate": 9.944580404994912e-06, "loss": 0.2011, "step": 4480 }, { "epoch": 0.08166754579021081, "grad_norm": 1.2610018253326416, "learning_rate": 9.944188831438689e-06, "loss": 0.204, "step": 4490 }, { "epoch": 0.0818494334200331, "grad_norm": 1.3518896102905273, "learning_rate": 9.943795889112234e-06, "loss": 0.2009, "step": 4500 }, { "epoch": 0.0820313210498554, "grad_norm": 1.2875120639801025, "learning_rate": 9.94340157815191e-06, "loss": 0.2067, "step": 4510 }, { "epoch": 0.0822132086796777, "grad_norm": 1.4524809122085571, "learning_rate": 9.943005898694559e-06, "loss": 0.1964, "step": 4520 }, { "epoch": 0.0823950963095, "grad_norm": 1.4133378267288208, "learning_rate": 9.942608850877493e-06, "loss": 0.1994, "step": 4530 }, { "epoch": 0.0825769839393223, "grad_norm": 1.3483506441116333, "learning_rate": 9.942210434838506e-06, "loss": 0.202, "step": 4540 }, { "epoch": 0.08275887156914458, "grad_norm": 1.333419919013977, "learning_rate": 9.941810650715854e-06, "loss": 0.1982, "step": 4550 }, { "epoch": 0.08294075919896687, "grad_norm": 1.3400126695632935, "learning_rate": 9.941409498648287e-06, "loss": 0.2004, "step": 4560 }, { "epoch": 0.08312264682878917, "grad_norm": 1.2951198816299438, "learning_rate": 9.941006978775009e-06, "loss": 0.1999, "step": 4570 }, { "epoch": 0.08330453445861147, "grad_norm": 1.3258846998214722, "learning_rate": 9.940603091235712e-06, "loss": 0.2022, "step": 4580 }, { "epoch": 0.08348642208843376, "grad_norm": 1.708970308303833, "learning_rate": 9.940197836170558e-06, "loss": 0.1969, "step": 4590 }, { "epoch": 0.08366830971825606, "grad_norm": 1.386101484298706, "learning_rate": 9.939791213720188e-06, "loss": 0.199, "step": 4600 }, { "epoch": 0.08385019734807836, "grad_norm": 1.4247386455535889, "learning_rate": 9.939383224025712e-06, "loss": 0.1985, "step": 4610 }, { "epoch": 0.08403208497790066, "grad_norm": 1.387792944908142, "learning_rate": 9.938973867228714e-06, "loss": 0.1963, "step": 4620 }, { "epoch": 0.08421397260772295, "grad_norm": 1.3545522689819336, "learning_rate": 9.938563143471258e-06, "loss": 0.1998, "step": 4630 }, { "epoch": 0.08439586023754525, "grad_norm": 1.4259077310562134, "learning_rate": 9.938151052895878e-06, "loss": 0.2022, "step": 4640 }, { "epoch": 0.08457774786736753, "grad_norm": 1.3524292707443237, "learning_rate": 9.937737595645584e-06, "loss": 0.2018, "step": 4650 }, { "epoch": 0.08475963549718983, "grad_norm": 1.2588515281677246, "learning_rate": 9.937322771863861e-06, "loss": 0.1968, "step": 4660 }, { "epoch": 0.08494152312701213, "grad_norm": 1.308218002319336, "learning_rate": 9.936906581694667e-06, "loss": 0.1992, "step": 4670 }, { "epoch": 0.08512341075683443, "grad_norm": 1.2647380828857422, "learning_rate": 9.936489025282433e-06, "loss": 0.2014, "step": 4680 }, { "epoch": 0.08530529838665672, "grad_norm": 1.3199272155761719, "learning_rate": 9.936070102772067e-06, "loss": 0.2072, "step": 4690 }, { "epoch": 0.08548718601647902, "grad_norm": 1.2941620349884033, "learning_rate": 9.935649814308947e-06, "loss": 0.1935, "step": 4700 }, { "epoch": 0.08566907364630132, "grad_norm": 1.397042155265808, "learning_rate": 9.935228160038933e-06, "loss": 0.2039, "step": 4710 }, { "epoch": 0.08585096127612361, "grad_norm": 1.4642465114593506, "learning_rate": 9.934805140108347e-06, "loss": 0.2008, "step": 4720 }, { "epoch": 0.08603284890594591, "grad_norm": 1.3432420492172241, "learning_rate": 9.934380754663997e-06, "loss": 0.1981, "step": 4730 }, { "epoch": 0.08621473653576821, "grad_norm": 1.2501524686813354, "learning_rate": 9.933955003853157e-06, "loss": 0.2037, "step": 4740 }, { "epoch": 0.0863966241655905, "grad_norm": 1.3735986948013306, "learning_rate": 9.933527887823578e-06, "loss": 0.2033, "step": 4750 }, { "epoch": 0.08657851179541279, "grad_norm": 1.4575186967849731, "learning_rate": 9.933099406723483e-06, "loss": 0.2026, "step": 4760 }, { "epoch": 0.08676039942523509, "grad_norm": 1.364054560661316, "learning_rate": 9.932669560701571e-06, "loss": 0.2003, "step": 4770 }, { "epoch": 0.08694228705505738, "grad_norm": 1.4867210388183594, "learning_rate": 9.932238349907014e-06, "loss": 0.2012, "step": 4780 }, { "epoch": 0.08712417468487968, "grad_norm": 4.171614170074463, "learning_rate": 9.931805774489457e-06, "loss": 0.1974, "step": 4790 }, { "epoch": 0.08730606231470198, "grad_norm": 1.433622121810913, "learning_rate": 9.931371834599014e-06, "loss": 0.1987, "step": 4800 }, { "epoch": 0.08748794994452427, "grad_norm": 1.3450977802276611, "learning_rate": 9.930936530386282e-06, "loss": 0.2043, "step": 4810 }, { "epoch": 0.08766983757434657, "grad_norm": 1.4820464849472046, "learning_rate": 9.930499862002328e-06, "loss": 0.1971, "step": 4820 }, { "epoch": 0.08785172520416887, "grad_norm": 1.4885598421096802, "learning_rate": 9.930061829598687e-06, "loss": 0.1992, "step": 4830 }, { "epoch": 0.08803361283399116, "grad_norm": 1.3304864168167114, "learning_rate": 9.929622433327372e-06, "loss": 0.1983, "step": 4840 }, { "epoch": 0.08821550046381346, "grad_norm": 1.3625112771987915, "learning_rate": 9.929181673340872e-06, "loss": 0.2035, "step": 4850 }, { "epoch": 0.08839738809363575, "grad_norm": 1.622989535331726, "learning_rate": 9.928739549792142e-06, "loss": 0.2006, "step": 4860 }, { "epoch": 0.08857927572345804, "grad_norm": 1.3789777755737305, "learning_rate": 9.928296062834619e-06, "loss": 0.2021, "step": 4870 }, { "epoch": 0.08876116335328034, "grad_norm": 1.301522135734558, "learning_rate": 9.927851212622202e-06, "loss": 0.202, "step": 4880 }, { "epoch": 0.08894305098310264, "grad_norm": 1.2948260307312012, "learning_rate": 9.927404999309273e-06, "loss": 0.1962, "step": 4890 }, { "epoch": 0.08912493861292493, "grad_norm": 1.4278150796890259, "learning_rate": 9.926957423050684e-06, "loss": 0.2057, "step": 4900 }, { "epoch": 0.08930682624274723, "grad_norm": 1.352800726890564, "learning_rate": 9.92650848400176e-06, "loss": 0.1986, "step": 4910 }, { "epoch": 0.08948871387256953, "grad_norm": 1.421162486076355, "learning_rate": 9.926058182318295e-06, "loss": 0.195, "step": 4920 }, { "epoch": 0.08967060150239183, "grad_norm": 1.3156191110610962, "learning_rate": 9.925606518156564e-06, "loss": 0.2025, "step": 4930 }, { "epoch": 0.08985248913221412, "grad_norm": 1.3862154483795166, "learning_rate": 9.925153491673306e-06, "loss": 0.1922, "step": 4940 }, { "epoch": 0.09003437676203642, "grad_norm": 1.5020748376846313, "learning_rate": 9.92469910302574e-06, "loss": 0.2026, "step": 4950 }, { "epoch": 0.09021626439185872, "grad_norm": 1.3673168420791626, "learning_rate": 9.92424335237155e-06, "loss": 0.2032, "step": 4960 }, { "epoch": 0.090398152021681, "grad_norm": 1.2813433408737183, "learning_rate": 9.923786239868904e-06, "loss": 0.2043, "step": 4970 }, { "epoch": 0.0905800396515033, "grad_norm": 1.2920981645584106, "learning_rate": 9.92332776567643e-06, "loss": 0.2072, "step": 4980 }, { "epoch": 0.0907619272813256, "grad_norm": 1.3558473587036133, "learning_rate": 9.922867929953237e-06, "loss": 0.2006, "step": 4990 }, { "epoch": 0.09094381491114789, "grad_norm": 1.43364417552948, "learning_rate": 9.922406732858903e-06, "loss": 0.2042, "step": 5000 }, { "epoch": 0.09112570254097019, "grad_norm": 1.4242839813232422, "learning_rate": 9.921944174553481e-06, "loss": 0.1989, "step": 5010 }, { "epoch": 0.09130759017079249, "grad_norm": 1.263886570930481, "learning_rate": 9.921480255197495e-06, "loss": 0.1986, "step": 5020 }, { "epoch": 0.09148947780061478, "grad_norm": 1.3914079666137695, "learning_rate": 9.921014974951938e-06, "loss": 0.2034, "step": 5030 }, { "epoch": 0.09167136543043708, "grad_norm": 1.452286958694458, "learning_rate": 9.92054833397828e-06, "loss": 0.2063, "step": 5040 }, { "epoch": 0.09185325306025938, "grad_norm": 1.2708038091659546, "learning_rate": 9.920080332438464e-06, "loss": 0.1999, "step": 5050 }, { "epoch": 0.09203514069008167, "grad_norm": 1.3004201650619507, "learning_rate": 9.919610970494899e-06, "loss": 0.2013, "step": 5060 }, { "epoch": 0.09221702831990396, "grad_norm": 1.2961074113845825, "learning_rate": 9.91914024831047e-06, "loss": 0.1971, "step": 5070 }, { "epoch": 0.09239891594972625, "grad_norm": 1.4583500623703003, "learning_rate": 9.918668166048537e-06, "loss": 0.1948, "step": 5080 }, { "epoch": 0.09258080357954855, "grad_norm": 1.4541499614715576, "learning_rate": 9.918194723872927e-06, "loss": 0.1991, "step": 5090 }, { "epoch": 0.09276269120937085, "grad_norm": 1.3050438165664673, "learning_rate": 9.917719921947941e-06, "loss": 0.2044, "step": 5100 }, { "epoch": 0.09294457883919315, "grad_norm": 1.1695116758346558, "learning_rate": 9.917243760438351e-06, "loss": 0.1977, "step": 5110 }, { "epoch": 0.09312646646901544, "grad_norm": 1.3679789304733276, "learning_rate": 9.916766239509404e-06, "loss": 0.2047, "step": 5120 }, { "epoch": 0.09330835409883774, "grad_norm": 1.3517804145812988, "learning_rate": 9.916287359326816e-06, "loss": 0.2022, "step": 5130 }, { "epoch": 0.09349024172866004, "grad_norm": 1.3424577713012695, "learning_rate": 9.91580712005677e-06, "loss": 0.2038, "step": 5140 }, { "epoch": 0.09367212935848233, "grad_norm": 1.388779878616333, "learning_rate": 9.915325521865931e-06, "loss": 0.1925, "step": 5150 }, { "epoch": 0.09385401698830463, "grad_norm": 1.3131184577941895, "learning_rate": 9.91484256492143e-06, "loss": 0.1969, "step": 5160 }, { "epoch": 0.09403590461812691, "grad_norm": 1.3347338438034058, "learning_rate": 9.914358249390867e-06, "loss": 0.2021, "step": 5170 }, { "epoch": 0.09421779224794921, "grad_norm": 1.574829339981079, "learning_rate": 9.913872575442317e-06, "loss": 0.1981, "step": 5180 }, { "epoch": 0.09439967987777151, "grad_norm": 1.4480096101760864, "learning_rate": 9.913385543244328e-06, "loss": 0.202, "step": 5190 }, { "epoch": 0.0945815675075938, "grad_norm": 1.4249836206436157, "learning_rate": 9.912897152965915e-06, "loss": 0.2007, "step": 5200 }, { "epoch": 0.0947634551374161, "grad_norm": 1.331373691558838, "learning_rate": 9.912407404776567e-06, "loss": 0.1995, "step": 5210 }, { "epoch": 0.0949453427672384, "grad_norm": 1.396562933921814, "learning_rate": 9.911916298846242e-06, "loss": 0.2028, "step": 5220 }, { "epoch": 0.0951272303970607, "grad_norm": 1.3434922695159912, "learning_rate": 9.911423835345372e-06, "loss": 0.2, "step": 5230 }, { "epoch": 0.095309118026883, "grad_norm": 1.4851365089416504, "learning_rate": 9.910930014444863e-06, "loss": 0.2035, "step": 5240 }, { "epoch": 0.09549100565670529, "grad_norm": 1.426054835319519, "learning_rate": 9.91043483631608e-06, "loss": 0.1985, "step": 5250 }, { "epoch": 0.09567289328652759, "grad_norm": 1.2745842933654785, "learning_rate": 9.909938301130872e-06, "loss": 0.199, "step": 5260 }, { "epoch": 0.09585478091634989, "grad_norm": 1.399355173110962, "learning_rate": 9.909440409061555e-06, "loss": 0.1982, "step": 5270 }, { "epoch": 0.09603666854617217, "grad_norm": 1.3390458822250366, "learning_rate": 9.908941160280909e-06, "loss": 0.2007, "step": 5280 }, { "epoch": 0.09621855617599447, "grad_norm": 1.3447686433792114, "learning_rate": 9.908440554962198e-06, "loss": 0.2056, "step": 5290 }, { "epoch": 0.09640044380581676, "grad_norm": 1.2179698944091797, "learning_rate": 9.907938593279145e-06, "loss": 0.198, "step": 5300 }, { "epoch": 0.09658233143563906, "grad_norm": 1.292288064956665, "learning_rate": 9.907435275405948e-06, "loss": 0.1986, "step": 5310 }, { "epoch": 0.09676421906546136, "grad_norm": 1.295562982559204, "learning_rate": 9.906930601517277e-06, "loss": 0.1997, "step": 5320 }, { "epoch": 0.09694610669528365, "grad_norm": 1.3421874046325684, "learning_rate": 9.906424571788271e-06, "loss": 0.1989, "step": 5330 }, { "epoch": 0.09712799432510595, "grad_norm": 1.347042441368103, "learning_rate": 9.90591718639454e-06, "loss": 0.1981, "step": 5340 }, { "epoch": 0.09730988195492825, "grad_norm": 1.3517132997512817, "learning_rate": 9.905408445512163e-06, "loss": 0.197, "step": 5350 }, { "epoch": 0.09749176958475055, "grad_norm": 1.3937288522720337, "learning_rate": 9.904898349317694e-06, "loss": 0.1999, "step": 5360 }, { "epoch": 0.09767365721457284, "grad_norm": 1.375322937965393, "learning_rate": 9.90438689798815e-06, "loss": 0.199, "step": 5370 }, { "epoch": 0.09785554484439513, "grad_norm": 1.212874412536621, "learning_rate": 9.903874091701024e-06, "loss": 0.1966, "step": 5380 }, { "epoch": 0.09803743247421742, "grad_norm": 1.2691152095794678, "learning_rate": 9.90335993063428e-06, "loss": 0.1967, "step": 5390 }, { "epoch": 0.09821932010403972, "grad_norm": 1.3521469831466675, "learning_rate": 9.902844414966347e-06, "loss": 0.2046, "step": 5400 }, { "epoch": 0.09840120773386202, "grad_norm": 1.2652087211608887, "learning_rate": 9.902327544876124e-06, "loss": 0.1951, "step": 5410 }, { "epoch": 0.09858309536368431, "grad_norm": 1.2044527530670166, "learning_rate": 9.90180932054299e-06, "loss": 0.1979, "step": 5420 }, { "epoch": 0.09876498299350661, "grad_norm": 1.3442468643188477, "learning_rate": 9.901289742146783e-06, "loss": 0.2007, "step": 5430 }, { "epoch": 0.09894687062332891, "grad_norm": 1.348831295967102, "learning_rate": 9.900768809867812e-06, "loss": 0.1988, "step": 5440 }, { "epoch": 0.0991287582531512, "grad_norm": 1.388492226600647, "learning_rate": 9.900246523886862e-06, "loss": 0.2025, "step": 5450 }, { "epoch": 0.0993106458829735, "grad_norm": 1.2420367002487183, "learning_rate": 9.899722884385186e-06, "loss": 0.1972, "step": 5460 }, { "epoch": 0.0994925335127958, "grad_norm": 1.349602222442627, "learning_rate": 9.899197891544502e-06, "loss": 0.1967, "step": 5470 }, { "epoch": 0.0996744211426181, "grad_norm": 1.3191062211990356, "learning_rate": 9.898671545546999e-06, "loss": 0.2027, "step": 5480 }, { "epoch": 0.09985630877244038, "grad_norm": 1.4323886632919312, "learning_rate": 9.898143846575341e-06, "loss": 0.2045, "step": 5490 }, { "epoch": 0.10003819640226268, "grad_norm": 1.2825477123260498, "learning_rate": 9.897614794812656e-06, "loss": 0.2031, "step": 5500 }, { "epoch": 0.10022008403208497, "grad_norm": 1.3284200429916382, "learning_rate": 9.897084390442544e-06, "loss": 0.1999, "step": 5510 }, { "epoch": 0.10040197166190727, "grad_norm": 1.366426944732666, "learning_rate": 9.896552633649074e-06, "loss": 0.2043, "step": 5520 }, { "epoch": 0.10058385929172957, "grad_norm": 1.312911033630371, "learning_rate": 9.896019524616783e-06, "loss": 0.1977, "step": 5530 }, { "epoch": 0.10076574692155187, "grad_norm": 1.3504881858825684, "learning_rate": 9.89548506353068e-06, "loss": 0.1986, "step": 5540 }, { "epoch": 0.10094763455137416, "grad_norm": 1.440566062927246, "learning_rate": 9.89494925057624e-06, "loss": 0.2026, "step": 5550 }, { "epoch": 0.10112952218119646, "grad_norm": 1.1985517740249634, "learning_rate": 9.894412085939408e-06, "loss": 0.205, "step": 5560 }, { "epoch": 0.10131140981101876, "grad_norm": 1.29746675491333, "learning_rate": 9.8938735698066e-06, "loss": 0.2037, "step": 5570 }, { "epoch": 0.10149329744084105, "grad_norm": 1.5362179279327393, "learning_rate": 9.8933337023647e-06, "loss": 0.2018, "step": 5580 }, { "epoch": 0.10167518507066334, "grad_norm": 1.4425830841064453, "learning_rate": 9.89279248380106e-06, "loss": 0.1938, "step": 5590 }, { "epoch": 0.10185707270048563, "grad_norm": 1.2522568702697754, "learning_rate": 9.892249914303502e-06, "loss": 0.1985, "step": 5600 }, { "epoch": 0.10203896033030793, "grad_norm": 1.300062894821167, "learning_rate": 9.891705994060317e-06, "loss": 0.2023, "step": 5610 }, { "epoch": 0.10222084796013023, "grad_norm": 1.3818265199661255, "learning_rate": 9.891160723260263e-06, "loss": 0.2021, "step": 5620 }, { "epoch": 0.10240273558995253, "grad_norm": 1.2549062967300415, "learning_rate": 9.890614102092567e-06, "loss": 0.2051, "step": 5630 }, { "epoch": 0.10258462321977482, "grad_norm": 1.3728899955749512, "learning_rate": 9.890066130746929e-06, "loss": 0.2016, "step": 5640 }, { "epoch": 0.10276651084959712, "grad_norm": 1.5419846773147583, "learning_rate": 9.889516809413512e-06, "loss": 0.2015, "step": 5650 }, { "epoch": 0.10294839847941942, "grad_norm": 1.385050654411316, "learning_rate": 9.888966138282947e-06, "loss": 0.1964, "step": 5660 }, { "epoch": 0.10313028610924171, "grad_norm": 1.135652780532837, "learning_rate": 9.88841411754634e-06, "loss": 0.2012, "step": 5670 }, { "epoch": 0.10331217373906401, "grad_norm": 1.3606603145599365, "learning_rate": 9.887860747395258e-06, "loss": 0.205, "step": 5680 }, { "epoch": 0.10349406136888631, "grad_norm": 1.2551074028015137, "learning_rate": 9.887306028021743e-06, "loss": 0.2041, "step": 5690 }, { "epoch": 0.10367594899870859, "grad_norm": 1.1764214038848877, "learning_rate": 9.8867499596183e-06, "loss": 0.2029, "step": 5700 }, { "epoch": 0.10385783662853089, "grad_norm": 1.3003919124603271, "learning_rate": 9.886192542377905e-06, "loss": 0.1995, "step": 5710 }, { "epoch": 0.10403972425835319, "grad_norm": 1.252990961074829, "learning_rate": 9.885633776494e-06, "loss": 0.2028, "step": 5720 }, { "epoch": 0.10422161188817548, "grad_norm": 1.3583661317825317, "learning_rate": 9.885073662160494e-06, "loss": 0.1983, "step": 5730 }, { "epoch": 0.10440349951799778, "grad_norm": 1.318865418434143, "learning_rate": 9.88451219957177e-06, "loss": 0.2063, "step": 5740 }, { "epoch": 0.10458538714782008, "grad_norm": 1.421940803527832, "learning_rate": 9.883949388922673e-06, "loss": 0.2052, "step": 5750 }, { "epoch": 0.10476727477764237, "grad_norm": 1.120473027229309, "learning_rate": 9.88338523040852e-06, "loss": 0.1957, "step": 5760 }, { "epoch": 0.10494916240746467, "grad_norm": 1.2960808277130127, "learning_rate": 9.882819724225091e-06, "loss": 0.2017, "step": 5770 }, { "epoch": 0.10513105003728697, "grad_norm": 1.4100353717803955, "learning_rate": 9.882252870568638e-06, "loss": 0.2009, "step": 5780 }, { "epoch": 0.10531293766710927, "grad_norm": 1.3540055751800537, "learning_rate": 9.881684669635876e-06, "loss": 0.1999, "step": 5790 }, { "epoch": 0.10549482529693155, "grad_norm": 1.4130516052246094, "learning_rate": 9.881115121623992e-06, "loss": 0.1986, "step": 5800 }, { "epoch": 0.10567671292675385, "grad_norm": 1.1937249898910522, "learning_rate": 9.88054422673064e-06, "loss": 0.1965, "step": 5810 }, { "epoch": 0.10585860055657614, "grad_norm": 1.2592949867248535, "learning_rate": 9.87997198515394e-06, "loss": 0.2059, "step": 5820 }, { "epoch": 0.10604048818639844, "grad_norm": 1.2385478019714355, "learning_rate": 9.879398397092481e-06, "loss": 0.2029, "step": 5830 }, { "epoch": 0.10622237581622074, "grad_norm": 1.3134866952896118, "learning_rate": 9.878823462745314e-06, "loss": 0.1996, "step": 5840 }, { "epoch": 0.10640426344604303, "grad_norm": 1.3059242963790894, "learning_rate": 9.878247182311965e-06, "loss": 0.2006, "step": 5850 }, { "epoch": 0.10658615107586533, "grad_norm": 1.294008493423462, "learning_rate": 9.87766955599242e-06, "loss": 0.2023, "step": 5860 }, { "epoch": 0.10676803870568763, "grad_norm": 1.179680585861206, "learning_rate": 9.877090583987138e-06, "loss": 0.1971, "step": 5870 }, { "epoch": 0.10694992633550993, "grad_norm": 1.0662832260131836, "learning_rate": 9.87651026649704e-06, "loss": 0.1987, "step": 5880 }, { "epoch": 0.10713181396533222, "grad_norm": 1.509894609451294, "learning_rate": 9.875928603723517e-06, "loss": 0.2033, "step": 5890 }, { "epoch": 0.10731370159515452, "grad_norm": 1.3066198825836182, "learning_rate": 9.87534559586843e-06, "loss": 0.2034, "step": 5900 }, { "epoch": 0.1074955892249768, "grad_norm": 1.3719162940979004, "learning_rate": 9.874761243134098e-06, "loss": 0.2019, "step": 5910 }, { "epoch": 0.1076774768547991, "grad_norm": 1.3149073123931885, "learning_rate": 9.874175545723314e-06, "loss": 0.2064, "step": 5920 }, { "epoch": 0.1078593644846214, "grad_norm": 1.1917974948883057, "learning_rate": 9.873588503839333e-06, "loss": 0.2024, "step": 5930 }, { "epoch": 0.1080412521144437, "grad_norm": 1.1793763637542725, "learning_rate": 9.873000117685882e-06, "loss": 0.1936, "step": 5940 }, { "epoch": 0.10822313974426599, "grad_norm": 1.2577564716339111, "learning_rate": 9.872410387467149e-06, "loss": 0.1994, "step": 5950 }, { "epoch": 0.10840502737408829, "grad_norm": 1.3139779567718506, "learning_rate": 9.871819313387792e-06, "loss": 0.1954, "step": 5960 }, { "epoch": 0.10858691500391059, "grad_norm": 1.2779500484466553, "learning_rate": 9.871226895652934e-06, "loss": 0.2029, "step": 5970 }, { "epoch": 0.10876880263373288, "grad_norm": 1.2850782871246338, "learning_rate": 9.870633134468165e-06, "loss": 0.1985, "step": 5980 }, { "epoch": 0.10895069026355518, "grad_norm": 1.3416451215744019, "learning_rate": 9.87003803003954e-06, "loss": 0.2004, "step": 5990 }, { "epoch": 0.10913257789337748, "grad_norm": 1.4673434495925903, "learning_rate": 9.86944158257358e-06, "loss": 0.2045, "step": 6000 }, { "epoch": 0.10931446552319976, "grad_norm": 1.1609548330307007, "learning_rate": 9.868843792277274e-06, "loss": 0.2019, "step": 6010 }, { "epoch": 0.10949635315302206, "grad_norm": 1.4946528673171997, "learning_rate": 9.868244659358077e-06, "loss": 0.2026, "step": 6020 }, { "epoch": 0.10967824078284436, "grad_norm": 1.412505865097046, "learning_rate": 9.867644184023907e-06, "loss": 0.204, "step": 6030 }, { "epoch": 0.10986012841266665, "grad_norm": 1.3223438262939453, "learning_rate": 9.86704236648315e-06, "loss": 0.2031, "step": 6040 }, { "epoch": 0.11004201604248895, "grad_norm": 1.348317265510559, "learning_rate": 9.86643920694466e-06, "loss": 0.2005, "step": 6050 }, { "epoch": 0.11022390367231125, "grad_norm": 1.1969858407974243, "learning_rate": 9.86583470561775e-06, "loss": 0.1981, "step": 6060 }, { "epoch": 0.11040579130213354, "grad_norm": 1.295723557472229, "learning_rate": 9.865228862712206e-06, "loss": 0.2035, "step": 6070 }, { "epoch": 0.11058767893195584, "grad_norm": 1.2585691213607788, "learning_rate": 9.864621678438276e-06, "loss": 0.2047, "step": 6080 }, { "epoch": 0.11076956656177814, "grad_norm": 1.339484691619873, "learning_rate": 9.864013153006676e-06, "loss": 0.1973, "step": 6090 }, { "epoch": 0.11095145419160043, "grad_norm": 1.0862985849380493, "learning_rate": 9.86340328662858e-06, "loss": 0.2019, "step": 6100 }, { "epoch": 0.11113334182142272, "grad_norm": 1.3663614988327026, "learning_rate": 9.862792079515636e-06, "loss": 0.2, "step": 6110 }, { "epoch": 0.11131522945124502, "grad_norm": 1.1375399827957153, "learning_rate": 9.862179531879954e-06, "loss": 0.2047, "step": 6120 }, { "epoch": 0.11149711708106731, "grad_norm": 1.2515023946762085, "learning_rate": 9.861565643934113e-06, "loss": 0.1985, "step": 6130 }, { "epoch": 0.11167900471088961, "grad_norm": 1.3913748264312744, "learning_rate": 9.860950415891145e-06, "loss": 0.2023, "step": 6140 }, { "epoch": 0.1118608923407119, "grad_norm": 1.1249629259109497, "learning_rate": 9.860333847964563e-06, "loss": 0.1928, "step": 6150 }, { "epoch": 0.1120427799705342, "grad_norm": 1.190535545349121, "learning_rate": 9.859715940368334e-06, "loss": 0.203, "step": 6160 }, { "epoch": 0.1122246676003565, "grad_norm": 1.1253619194030762, "learning_rate": 9.859096693316892e-06, "loss": 0.2009, "step": 6170 }, { "epoch": 0.1124065552301788, "grad_norm": 1.2472362518310547, "learning_rate": 9.85847610702514e-06, "loss": 0.207, "step": 6180 }, { "epoch": 0.1125884428600011, "grad_norm": 1.477825403213501, "learning_rate": 9.857854181708442e-06, "loss": 0.1998, "step": 6190 }, { "epoch": 0.11277033048982339, "grad_norm": 1.199155569076538, "learning_rate": 9.857230917582628e-06, "loss": 0.2019, "step": 6200 }, { "epoch": 0.11295221811964569, "grad_norm": 1.2641313076019287, "learning_rate": 9.856606314863993e-06, "loss": 0.2066, "step": 6210 }, { "epoch": 0.11313410574946797, "grad_norm": 1.3950942754745483, "learning_rate": 9.855980373769294e-06, "loss": 0.2016, "step": 6220 }, { "epoch": 0.11331599337929027, "grad_norm": 1.2773404121398926, "learning_rate": 9.855353094515755e-06, "loss": 0.2062, "step": 6230 }, { "epoch": 0.11349788100911257, "grad_norm": 1.2882050275802612, "learning_rate": 9.854724477321063e-06, "loss": 0.2094, "step": 6240 }, { "epoch": 0.11367976863893486, "grad_norm": 1.4161255359649658, "learning_rate": 9.854094522403372e-06, "loss": 0.2077, "step": 6250 }, { "epoch": 0.11386165626875716, "grad_norm": 1.3292781114578247, "learning_rate": 9.853463229981295e-06, "loss": 0.1981, "step": 6260 }, { "epoch": 0.11404354389857946, "grad_norm": 1.279371738433838, "learning_rate": 9.852830600273913e-06, "loss": 0.1964, "step": 6270 }, { "epoch": 0.11422543152840176, "grad_norm": 1.2736986875534058, "learning_rate": 9.852196633500774e-06, "loss": 0.2042, "step": 6280 }, { "epoch": 0.11440731915822405, "grad_norm": 1.3373594284057617, "learning_rate": 9.851561329881883e-06, "loss": 0.2064, "step": 6290 }, { "epoch": 0.11458920678804635, "grad_norm": 1.2860163450241089, "learning_rate": 9.850924689637713e-06, "loss": 0.2018, "step": 6300 }, { "epoch": 0.11477109441786865, "grad_norm": 1.237523078918457, "learning_rate": 9.850286712989202e-06, "loss": 0.202, "step": 6310 }, { "epoch": 0.11495298204769093, "grad_norm": 1.200131893157959, "learning_rate": 9.84964740015775e-06, "loss": 0.1987, "step": 6320 }, { "epoch": 0.11513486967751323, "grad_norm": 1.2631765604019165, "learning_rate": 9.849006751365216e-06, "loss": 0.2044, "step": 6330 }, { "epoch": 0.11531675730733552, "grad_norm": 1.3581782579421997, "learning_rate": 9.848364766833932e-06, "loss": 0.2091, "step": 6340 }, { "epoch": 0.11549864493715782, "grad_norm": 1.256722331047058, "learning_rate": 9.847721446786686e-06, "loss": 0.2002, "step": 6350 }, { "epoch": 0.11568053256698012, "grad_norm": 1.3466540575027466, "learning_rate": 9.847076791446736e-06, "loss": 0.2023, "step": 6360 }, { "epoch": 0.11586242019680242, "grad_norm": 1.2411190271377563, "learning_rate": 9.846430801037796e-06, "loss": 0.1969, "step": 6370 }, { "epoch": 0.11604430782662471, "grad_norm": 1.4271907806396484, "learning_rate": 9.84578347578405e-06, "loss": 0.2014, "step": 6380 }, { "epoch": 0.11622619545644701, "grad_norm": 1.1762012243270874, "learning_rate": 9.845134815910141e-06, "loss": 0.1956, "step": 6390 }, { "epoch": 0.1164080830862693, "grad_norm": 1.3236180543899536, "learning_rate": 9.844484821641175e-06, "loss": 0.2023, "step": 6400 }, { "epoch": 0.1165899707160916, "grad_norm": 1.3581218719482422, "learning_rate": 9.843833493202725e-06, "loss": 0.2001, "step": 6410 }, { "epoch": 0.1167718583459139, "grad_norm": 1.3137873411178589, "learning_rate": 9.843180830820824e-06, "loss": 0.2051, "step": 6420 }, { "epoch": 0.11695374597573618, "grad_norm": 1.2004704475402832, "learning_rate": 9.842526834721968e-06, "loss": 0.2016, "step": 6430 }, { "epoch": 0.11713563360555848, "grad_norm": 1.1342967748641968, "learning_rate": 9.841871505133116e-06, "loss": 0.202, "step": 6440 }, { "epoch": 0.11731752123538078, "grad_norm": 1.2483773231506348, "learning_rate": 9.84121484228169e-06, "loss": 0.1998, "step": 6450 }, { "epoch": 0.11749940886520308, "grad_norm": 1.5301131010055542, "learning_rate": 9.840556846395575e-06, "loss": 0.1984, "step": 6460 }, { "epoch": 0.11768129649502537, "grad_norm": 1.2602758407592773, "learning_rate": 9.83989751770312e-06, "loss": 0.2007, "step": 6470 }, { "epoch": 0.11786318412484767, "grad_norm": 1.2727043628692627, "learning_rate": 9.839236856433132e-06, "loss": 0.2021, "step": 6480 }, { "epoch": 0.11804507175466997, "grad_norm": 1.3601182699203491, "learning_rate": 9.838574862814885e-06, "loss": 0.2032, "step": 6490 }, { "epoch": 0.11822695938449226, "grad_norm": 1.201776385307312, "learning_rate": 9.837911537078114e-06, "loss": 0.1976, "step": 6500 }, { "epoch": 0.11840884701431456, "grad_norm": 1.2684857845306396, "learning_rate": 9.837246879453015e-06, "loss": 0.1976, "step": 6510 }, { "epoch": 0.11859073464413686, "grad_norm": 1.1725822687149048, "learning_rate": 9.836580890170248e-06, "loss": 0.2021, "step": 6520 }, { "epoch": 0.11877262227395914, "grad_norm": 1.3211827278137207, "learning_rate": 9.835913569460936e-06, "loss": 0.2039, "step": 6530 }, { "epoch": 0.11895450990378144, "grad_norm": 1.343963623046875, "learning_rate": 9.835244917556659e-06, "loss": 0.2032, "step": 6540 }, { "epoch": 0.11913639753360374, "grad_norm": 1.1777474880218506, "learning_rate": 9.834574934689463e-06, "loss": 0.1942, "step": 6550 }, { "epoch": 0.11931828516342603, "grad_norm": 1.429918646812439, "learning_rate": 9.833903621091858e-06, "loss": 0.2122, "step": 6560 }, { "epoch": 0.11950017279324833, "grad_norm": 1.3367538452148438, "learning_rate": 9.83323097699681e-06, "loss": 0.2022, "step": 6570 }, { "epoch": 0.11968206042307063, "grad_norm": 1.1916404962539673, "learning_rate": 9.83255700263775e-06, "loss": 0.2034, "step": 6580 }, { "epoch": 0.11986394805289292, "grad_norm": 1.3650444746017456, "learning_rate": 9.831881698248576e-06, "loss": 0.2057, "step": 6590 }, { "epoch": 0.12004583568271522, "grad_norm": 1.2263579368591309, "learning_rate": 9.831205064063633e-06, "loss": 0.2011, "step": 6600 }, { "epoch": 0.12022772331253752, "grad_norm": 1.1591582298278809, "learning_rate": 9.830527100317743e-06, "loss": 0.2007, "step": 6610 }, { "epoch": 0.12040961094235982, "grad_norm": 1.2920209169387817, "learning_rate": 9.829847807246182e-06, "loss": 0.2034, "step": 6620 }, { "epoch": 0.12059149857218211, "grad_norm": 1.305131435394287, "learning_rate": 9.829167185084685e-06, "loss": 0.2024, "step": 6630 }, { "epoch": 0.1207733862020044, "grad_norm": 1.3712044954299927, "learning_rate": 9.828485234069457e-06, "loss": 0.1982, "step": 6640 }, { "epoch": 0.12095527383182669, "grad_norm": 1.3103065490722656, "learning_rate": 9.827801954437152e-06, "loss": 0.2004, "step": 6650 }, { "epoch": 0.12113716146164899, "grad_norm": 1.4186030626296997, "learning_rate": 9.8271173464249e-06, "loss": 0.2063, "step": 6660 }, { "epoch": 0.12131904909147129, "grad_norm": 1.3007049560546875, "learning_rate": 9.826431410270275e-06, "loss": 0.2041, "step": 6670 }, { "epoch": 0.12150093672129358, "grad_norm": 1.1901096105575562, "learning_rate": 9.825744146211324e-06, "loss": 0.2055, "step": 6680 }, { "epoch": 0.12168282435111588, "grad_norm": 1.327390432357788, "learning_rate": 9.825055554486555e-06, "loss": 0.2015, "step": 6690 }, { "epoch": 0.12186471198093818, "grad_norm": 1.1147143840789795, "learning_rate": 9.824365635334927e-06, "loss": 0.1986, "step": 6700 }, { "epoch": 0.12204659961076048, "grad_norm": 1.4462212324142456, "learning_rate": 9.823674388995872e-06, "loss": 0.2035, "step": 6710 }, { "epoch": 0.12222848724058277, "grad_norm": 1.3410418033599854, "learning_rate": 9.822981815709271e-06, "loss": 0.2048, "step": 6720 }, { "epoch": 0.12241037487040507, "grad_norm": 1.5063930749893188, "learning_rate": 9.822287915715475e-06, "loss": 0.1981, "step": 6730 }, { "epoch": 0.12259226250022735, "grad_norm": 1.1660511493682861, "learning_rate": 9.821592689255287e-06, "loss": 0.2019, "step": 6740 }, { "epoch": 0.12277415013004965, "grad_norm": 1.2649790048599243, "learning_rate": 9.82089613656998e-06, "loss": 0.2001, "step": 6750 }, { "epoch": 0.12295603775987195, "grad_norm": 1.2601503133773804, "learning_rate": 9.820198257901279e-06, "loss": 0.2005, "step": 6760 }, { "epoch": 0.12313792538969424, "grad_norm": 1.4218087196350098, "learning_rate": 9.819499053491371e-06, "loss": 0.2028, "step": 6770 }, { "epoch": 0.12331981301951654, "grad_norm": 1.2548253536224365, "learning_rate": 9.818798523582906e-06, "loss": 0.1999, "step": 6780 }, { "epoch": 0.12350170064933884, "grad_norm": 1.192886233329773, "learning_rate": 9.818096668418992e-06, "loss": 0.2008, "step": 6790 }, { "epoch": 0.12368358827916114, "grad_norm": 1.6069155931472778, "learning_rate": 9.817393488243199e-06, "loss": 0.2007, "step": 6800 }, { "epoch": 0.12386547590898343, "grad_norm": 1.3426796197891235, "learning_rate": 9.81668898329955e-06, "loss": 0.2093, "step": 6810 }, { "epoch": 0.12404736353880573, "grad_norm": 1.1907076835632324, "learning_rate": 9.815983153832536e-06, "loss": 0.2066, "step": 6820 }, { "epoch": 0.12422925116862803, "grad_norm": 1.4027718305587769, "learning_rate": 9.815276000087104e-06, "loss": 0.2037, "step": 6830 }, { "epoch": 0.12441113879845032, "grad_norm": 1.4948644638061523, "learning_rate": 9.81456752230866e-06, "loss": 0.2059, "step": 6840 }, { "epoch": 0.12459302642827261, "grad_norm": 1.3248716592788696, "learning_rate": 9.813857720743069e-06, "loss": 0.2049, "step": 6850 }, { "epoch": 0.1247749140580949, "grad_norm": 1.574239730834961, "learning_rate": 9.813146595636662e-06, "loss": 0.2023, "step": 6860 }, { "epoch": 0.1249568016879172, "grad_norm": 1.2213228940963745, "learning_rate": 9.81243414723622e-06, "loss": 0.208, "step": 6870 }, { "epoch": 0.1251386893177395, "grad_norm": 1.3836355209350586, "learning_rate": 9.811720375788985e-06, "loss": 0.2028, "step": 6880 }, { "epoch": 0.1253205769475618, "grad_norm": 1.372022032737732, "learning_rate": 9.811005281542667e-06, "loss": 0.2005, "step": 6890 }, { "epoch": 0.1255024645773841, "grad_norm": 1.297513723373413, "learning_rate": 9.810288864745423e-06, "loss": 0.2017, "step": 6900 }, { "epoch": 0.1256843522072064, "grad_norm": 1.3127362728118896, "learning_rate": 9.809571125645876e-06, "loss": 0.1992, "step": 6910 }, { "epoch": 0.1258662398370287, "grad_norm": 1.2699722051620483, "learning_rate": 9.808852064493108e-06, "loss": 0.2038, "step": 6920 }, { "epoch": 0.12604812746685098, "grad_norm": 1.4122538566589355, "learning_rate": 9.808131681536655e-06, "loss": 0.2054, "step": 6930 }, { "epoch": 0.12623001509667328, "grad_norm": 1.4779349565505981, "learning_rate": 9.80740997702652e-06, "loss": 0.2042, "step": 6940 }, { "epoch": 0.12641190272649558, "grad_norm": 1.1476770639419556, "learning_rate": 9.806686951213152e-06, "loss": 0.1991, "step": 6950 }, { "epoch": 0.12659379035631788, "grad_norm": 1.248610496520996, "learning_rate": 9.805962604347472e-06, "loss": 0.2072, "step": 6960 }, { "epoch": 0.12677567798614017, "grad_norm": 1.3917546272277832, "learning_rate": 9.80523693668085e-06, "loss": 0.2047, "step": 6970 }, { "epoch": 0.12695756561596247, "grad_norm": 1.124672770500183, "learning_rate": 9.804509948465121e-06, "loss": 0.1995, "step": 6980 }, { "epoch": 0.12713945324578477, "grad_norm": 1.3372440338134766, "learning_rate": 9.803781639952574e-06, "loss": 0.2031, "step": 6990 }, { "epoch": 0.12732134087560704, "grad_norm": 1.485921859741211, "learning_rate": 9.803052011395956e-06, "loss": 0.201, "step": 7000 }, { "epoch": 0.12750322850542933, "grad_norm": 1.3200222253799438, "learning_rate": 9.802321063048474e-06, "loss": 0.1999, "step": 7010 }, { "epoch": 0.12768511613525163, "grad_norm": 1.2966521978378296, "learning_rate": 9.801588795163794e-06, "loss": 0.2046, "step": 7020 }, { "epoch": 0.12786700376507393, "grad_norm": 1.34244704246521, "learning_rate": 9.800855207996037e-06, "loss": 0.2001, "step": 7030 }, { "epoch": 0.12804889139489622, "grad_norm": 1.129221796989441, "learning_rate": 9.800120301799781e-06, "loss": 0.2011, "step": 7040 }, { "epoch": 0.12823077902471852, "grad_norm": 1.301711916923523, "learning_rate": 9.799384076830067e-06, "loss": 0.2031, "step": 7050 }, { "epoch": 0.12841266665454082, "grad_norm": 1.2440203428268433, "learning_rate": 9.79864653334239e-06, "loss": 0.2044, "step": 7060 }, { "epoch": 0.12859455428436312, "grad_norm": 1.3236651420593262, "learning_rate": 9.797907671592704e-06, "loss": 0.201, "step": 7070 }, { "epoch": 0.1287764419141854, "grad_norm": 1.185644268989563, "learning_rate": 9.797167491837417e-06, "loss": 0.2041, "step": 7080 }, { "epoch": 0.1289583295440077, "grad_norm": 1.2876919507980347, "learning_rate": 9.796425994333398e-06, "loss": 0.206, "step": 7090 }, { "epoch": 0.12914021717383, "grad_norm": 1.2410295009613037, "learning_rate": 9.795683179337973e-06, "loss": 0.1994, "step": 7100 }, { "epoch": 0.1293221048036523, "grad_norm": 1.2081174850463867, "learning_rate": 9.794939047108925e-06, "loss": 0.1994, "step": 7110 }, { "epoch": 0.1295039924334746, "grad_norm": 1.2595489025115967, "learning_rate": 9.794193597904494e-06, "loss": 0.2036, "step": 7120 }, { "epoch": 0.1296858800632969, "grad_norm": 1.349601149559021, "learning_rate": 9.793446831983375e-06, "loss": 0.2033, "step": 7130 }, { "epoch": 0.1298677676931192, "grad_norm": 1.1659882068634033, "learning_rate": 9.792698749604722e-06, "loss": 0.2001, "step": 7140 }, { "epoch": 0.1300496553229415, "grad_norm": 1.2206439971923828, "learning_rate": 9.791949351028145e-06, "loss": 0.1983, "step": 7150 }, { "epoch": 0.1302315429527638, "grad_norm": 1.1265190839767456, "learning_rate": 9.791198636513715e-06, "loss": 0.2054, "step": 7160 }, { "epoch": 0.1304134305825861, "grad_norm": 1.4903628826141357, "learning_rate": 9.79044660632195e-06, "loss": 0.2059, "step": 7170 }, { "epoch": 0.13059531821240838, "grad_norm": 1.363667368888855, "learning_rate": 9.789693260713835e-06, "loss": 0.2029, "step": 7180 }, { "epoch": 0.13077720584223068, "grad_norm": 1.2748936414718628, "learning_rate": 9.788938599950805e-06, "loss": 0.1996, "step": 7190 }, { "epoch": 0.13095909347205298, "grad_norm": 1.4553683996200562, "learning_rate": 9.788182624294756e-06, "loss": 0.201, "step": 7200 }, { "epoch": 0.13114098110187525, "grad_norm": 1.3464908599853516, "learning_rate": 9.787425334008034e-06, "loss": 0.2036, "step": 7210 }, { "epoch": 0.13132286873169755, "grad_norm": 1.2837427854537964, "learning_rate": 9.78666672935345e-06, "loss": 0.2017, "step": 7220 }, { "epoch": 0.13150475636151984, "grad_norm": 1.2826616764068604, "learning_rate": 9.785906810594261e-06, "loss": 0.2044, "step": 7230 }, { "epoch": 0.13168664399134214, "grad_norm": 1.2653878927230835, "learning_rate": 9.785145577994186e-06, "loss": 0.2118, "step": 7240 }, { "epoch": 0.13186853162116444, "grad_norm": 1.2498185634613037, "learning_rate": 9.784383031817403e-06, "loss": 0.2024, "step": 7250 }, { "epoch": 0.13205041925098673, "grad_norm": 1.2798430919647217, "learning_rate": 9.783619172328537e-06, "loss": 0.2041, "step": 7260 }, { "epoch": 0.13223230688080903, "grad_norm": 1.3566452264785767, "learning_rate": 9.782853999792677e-06, "loss": 0.1995, "step": 7270 }, { "epoch": 0.13241419451063133, "grad_norm": 1.1576201915740967, "learning_rate": 9.782087514475363e-06, "loss": 0.2033, "step": 7280 }, { "epoch": 0.13259608214045362, "grad_norm": 1.350222110748291, "learning_rate": 9.78131971664259e-06, "loss": 0.2079, "step": 7290 }, { "epoch": 0.13277796977027592, "grad_norm": 1.2813305854797363, "learning_rate": 9.780550606560818e-06, "loss": 0.2032, "step": 7300 }, { "epoch": 0.13295985740009822, "grad_norm": 1.337875247001648, "learning_rate": 9.779780184496945e-06, "loss": 0.2037, "step": 7310 }, { "epoch": 0.13314174502992052, "grad_norm": 1.2992160320281982, "learning_rate": 9.779008450718341e-06, "loss": 0.1956, "step": 7320 }, { "epoch": 0.1333236326597428, "grad_norm": 1.1269599199295044, "learning_rate": 9.77823540549282e-06, "loss": 0.2046, "step": 7330 }, { "epoch": 0.1335055202895651, "grad_norm": 1.3352545499801636, "learning_rate": 9.777461049088662e-06, "loss": 0.2007, "step": 7340 }, { "epoch": 0.1336874079193874, "grad_norm": 1.1609328985214233, "learning_rate": 9.776685381774586e-06, "loss": 0.2041, "step": 7350 }, { "epoch": 0.1338692955492097, "grad_norm": 1.1933655738830566, "learning_rate": 9.775908403819784e-06, "loss": 0.2111, "step": 7360 }, { "epoch": 0.134051183179032, "grad_norm": 1.2798395156860352, "learning_rate": 9.775130115493889e-06, "loss": 0.2048, "step": 7370 }, { "epoch": 0.1342330708088543, "grad_norm": 1.380805492401123, "learning_rate": 9.774350517066998e-06, "loss": 0.2104, "step": 7380 }, { "epoch": 0.1344149584386766, "grad_norm": 1.3309656381607056, "learning_rate": 9.773569608809655e-06, "loss": 0.2029, "step": 7390 }, { "epoch": 0.1345968460684989, "grad_norm": 1.2733231782913208, "learning_rate": 9.772787390992865e-06, "loss": 0.2094, "step": 7400 }, { "epoch": 0.1347787336983212, "grad_norm": 1.3477153778076172, "learning_rate": 9.772003863888082e-06, "loss": 0.2042, "step": 7410 }, { "epoch": 0.13496062132814346, "grad_norm": 1.2158923149108887, "learning_rate": 9.77121902776722e-06, "loss": 0.2033, "step": 7420 }, { "epoch": 0.13514250895796576, "grad_norm": 1.2256052494049072, "learning_rate": 9.77043288290264e-06, "loss": 0.2025, "step": 7430 }, { "epoch": 0.13532439658778805, "grad_norm": 1.2375941276550293, "learning_rate": 9.769645429567167e-06, "loss": 0.2002, "step": 7440 }, { "epoch": 0.13550628421761035, "grad_norm": 1.3680274486541748, "learning_rate": 9.768856668034072e-06, "loss": 0.2006, "step": 7450 }, { "epoch": 0.13568817184743265, "grad_norm": 1.281759262084961, "learning_rate": 9.768066598577083e-06, "loss": 0.2026, "step": 7460 }, { "epoch": 0.13587005947725495, "grad_norm": 1.1957578659057617, "learning_rate": 9.76727522147038e-06, "loss": 0.1982, "step": 7470 }, { "epoch": 0.13605194710707724, "grad_norm": 1.219369888305664, "learning_rate": 9.7664825369886e-06, "loss": 0.2034, "step": 7480 }, { "epoch": 0.13623383473689954, "grad_norm": 1.3196187019348145, "learning_rate": 9.765688545406831e-06, "loss": 0.2036, "step": 7490 }, { "epoch": 0.13641572236672184, "grad_norm": 1.3299503326416016, "learning_rate": 9.764893247000616e-06, "loss": 0.206, "step": 7500 }, { "epoch": 0.13659760999654413, "grad_norm": 1.2020349502563477, "learning_rate": 9.76409664204595e-06, "loss": 0.2083, "step": 7510 }, { "epoch": 0.13677949762636643, "grad_norm": 1.2493973970413208, "learning_rate": 9.763298730819286e-06, "loss": 0.203, "step": 7520 }, { "epoch": 0.13696138525618873, "grad_norm": 1.2851570844650269, "learning_rate": 9.762499513597524e-06, "loss": 0.2052, "step": 7530 }, { "epoch": 0.13714327288601103, "grad_norm": 1.280176043510437, "learning_rate": 9.761698990658017e-06, "loss": 0.2016, "step": 7540 }, { "epoch": 0.13732516051583332, "grad_norm": 1.2086681127548218, "learning_rate": 9.760897162278582e-06, "loss": 0.1996, "step": 7550 }, { "epoch": 0.13750704814565562, "grad_norm": 1.2187764644622803, "learning_rate": 9.760094028737475e-06, "loss": 0.2017, "step": 7560 }, { "epoch": 0.13768893577547792, "grad_norm": 1.2019497156143188, "learning_rate": 9.759289590313414e-06, "loss": 0.2016, "step": 7570 }, { "epoch": 0.1378708234053002, "grad_norm": 1.3461536169052124, "learning_rate": 9.758483847285565e-06, "loss": 0.2022, "step": 7580 }, { "epoch": 0.1380527110351225, "grad_norm": 1.2128572463989258, "learning_rate": 9.75767679993355e-06, "loss": 0.2079, "step": 7590 }, { "epoch": 0.1382345986649448, "grad_norm": 1.3423349857330322, "learning_rate": 9.756868448537445e-06, "loss": 0.206, "step": 7600 }, { "epoch": 0.1384164862947671, "grad_norm": 1.438670039176941, "learning_rate": 9.75605879337777e-06, "loss": 0.2046, "step": 7610 }, { "epoch": 0.1385983739245894, "grad_norm": 1.2912158966064453, "learning_rate": 9.755247834735509e-06, "loss": 0.2058, "step": 7620 }, { "epoch": 0.13878026155441167, "grad_norm": 1.3144817352294922, "learning_rate": 9.75443557289209e-06, "loss": 0.2066, "step": 7630 }, { "epoch": 0.13896214918423397, "grad_norm": 1.2198230028152466, "learning_rate": 9.753622008129395e-06, "loss": 0.2007, "step": 7640 }, { "epoch": 0.13914403681405627, "grad_norm": 1.184598684310913, "learning_rate": 9.752807140729762e-06, "loss": 0.2065, "step": 7650 }, { "epoch": 0.13932592444387856, "grad_norm": 1.2358959913253784, "learning_rate": 9.751990970975977e-06, "loss": 0.2058, "step": 7660 }, { "epoch": 0.13950781207370086, "grad_norm": 1.3730189800262451, "learning_rate": 9.751173499151278e-06, "loss": 0.2006, "step": 7670 }, { "epoch": 0.13968969970352316, "grad_norm": 1.26596999168396, "learning_rate": 9.750354725539359e-06, "loss": 0.2023, "step": 7680 }, { "epoch": 0.13987158733334545, "grad_norm": 1.2239418029785156, "learning_rate": 9.749534650424359e-06, "loss": 0.2028, "step": 7690 }, { "epoch": 0.14005347496316775, "grad_norm": 1.3511844873428345, "learning_rate": 9.748713274090876e-06, "loss": 0.2018, "step": 7700 }, { "epoch": 0.14023536259299005, "grad_norm": 1.1983609199523926, "learning_rate": 9.747890596823953e-06, "loss": 0.2003, "step": 7710 }, { "epoch": 0.14041725022281235, "grad_norm": 1.3326727151870728, "learning_rate": 9.74706661890909e-06, "loss": 0.2102, "step": 7720 }, { "epoch": 0.14059913785263464, "grad_norm": 1.425252914428711, "learning_rate": 9.746241340632233e-06, "loss": 0.2051, "step": 7730 }, { "epoch": 0.14078102548245694, "grad_norm": 1.1916435956954956, "learning_rate": 9.745414762279786e-06, "loss": 0.2071, "step": 7740 }, { "epoch": 0.14096291311227924, "grad_norm": 1.248789668083191, "learning_rate": 9.744586884138598e-06, "loss": 0.2025, "step": 7750 }, { "epoch": 0.14114480074210153, "grad_norm": 1.314653992652893, "learning_rate": 9.743757706495972e-06, "loss": 0.2034, "step": 7760 }, { "epoch": 0.14132668837192383, "grad_norm": 1.271106243133545, "learning_rate": 9.74292722963966e-06, "loss": 0.1989, "step": 7770 }, { "epoch": 0.14150857600174613, "grad_norm": 1.191628098487854, "learning_rate": 9.742095453857869e-06, "loss": 0.2031, "step": 7780 }, { "epoch": 0.14169046363156843, "grad_norm": 1.3651148080825806, "learning_rate": 9.741262379439253e-06, "loss": 0.2019, "step": 7790 }, { "epoch": 0.14187235126139072, "grad_norm": 1.3211761713027954, "learning_rate": 9.740428006672914e-06, "loss": 0.2056, "step": 7800 }, { "epoch": 0.14205423889121302, "grad_norm": 1.3478931188583374, "learning_rate": 9.739592335848415e-06, "loss": 0.2028, "step": 7810 }, { "epoch": 0.14223612652103532, "grad_norm": 1.2544299364089966, "learning_rate": 9.738755367255758e-06, "loss": 0.2029, "step": 7820 }, { "epoch": 0.1424180141508576, "grad_norm": 1.3443603515625, "learning_rate": 9.7379171011854e-06, "loss": 0.2022, "step": 7830 }, { "epoch": 0.14259990178067988, "grad_norm": 1.2746943235397339, "learning_rate": 9.737077537928252e-06, "loss": 0.2006, "step": 7840 }, { "epoch": 0.14278178941050218, "grad_norm": 1.3253238201141357, "learning_rate": 9.736236677775668e-06, "loss": 0.2038, "step": 7850 }, { "epoch": 0.14296367704032448, "grad_norm": 1.3438528776168823, "learning_rate": 9.735394521019456e-06, "loss": 0.201, "step": 7860 }, { "epoch": 0.14314556467014677, "grad_norm": 1.2835006713867188, "learning_rate": 9.734551067951876e-06, "loss": 0.207, "step": 7870 }, { "epoch": 0.14332745229996907, "grad_norm": 1.3535889387130737, "learning_rate": 9.733706318865634e-06, "loss": 0.203, "step": 7880 }, { "epoch": 0.14350933992979137, "grad_norm": 1.4053682088851929, "learning_rate": 9.732860274053887e-06, "loss": 0.2046, "step": 7890 }, { "epoch": 0.14369122755961367, "grad_norm": 1.2633450031280518, "learning_rate": 9.732012933810242e-06, "loss": 0.2057, "step": 7900 }, { "epoch": 0.14387311518943596, "grad_norm": 1.280160665512085, "learning_rate": 9.731164298428753e-06, "loss": 0.198, "step": 7910 }, { "epoch": 0.14405500281925826, "grad_norm": 1.1429922580718994, "learning_rate": 9.73031436820393e-06, "loss": 0.2008, "step": 7920 }, { "epoch": 0.14423689044908056, "grad_norm": 1.2125667333602905, "learning_rate": 9.729463143430726e-06, "loss": 0.203, "step": 7930 }, { "epoch": 0.14441877807890285, "grad_norm": 1.3325953483581543, "learning_rate": 9.728610624404548e-06, "loss": 0.2051, "step": 7940 }, { "epoch": 0.14460066570872515, "grad_norm": 1.0918596982955933, "learning_rate": 9.727756811421246e-06, "loss": 0.2025, "step": 7950 }, { "epoch": 0.14478255333854745, "grad_norm": 1.2399933338165283, "learning_rate": 9.726901704777124e-06, "loss": 0.2035, "step": 7960 }, { "epoch": 0.14496444096836975, "grad_norm": 1.3124727010726929, "learning_rate": 9.726045304768935e-06, "loss": 0.2026, "step": 7970 }, { "epoch": 0.14514632859819204, "grad_norm": 1.2363300323486328, "learning_rate": 9.725187611693878e-06, "loss": 0.2021, "step": 7980 }, { "epoch": 0.14532821622801434, "grad_norm": 1.2398524284362793, "learning_rate": 9.724328625849605e-06, "loss": 0.206, "step": 7990 }, { "epoch": 0.14551010385783664, "grad_norm": 1.3217400312423706, "learning_rate": 9.723468347534208e-06, "loss": 0.2049, "step": 8000 }, { "epoch": 0.14569199148765893, "grad_norm": 1.2750279903411865, "learning_rate": 9.722606777046239e-06, "loss": 0.2029, "step": 8010 }, { "epoch": 0.14587387911748123, "grad_norm": 1.1990046501159668, "learning_rate": 9.721743914684691e-06, "loss": 0.2035, "step": 8020 }, { "epoch": 0.14605576674730353, "grad_norm": 1.2323766946792603, "learning_rate": 9.720879760749007e-06, "loss": 0.2063, "step": 8030 }, { "epoch": 0.14623765437712583, "grad_norm": 1.1522272825241089, "learning_rate": 9.720014315539077e-06, "loss": 0.2048, "step": 8040 }, { "epoch": 0.1464195420069481, "grad_norm": 1.3271764516830444, "learning_rate": 9.719147579355244e-06, "loss": 0.2058, "step": 8050 }, { "epoch": 0.1466014296367704, "grad_norm": 1.418045997619629, "learning_rate": 9.718279552498291e-06, "loss": 0.2024, "step": 8060 }, { "epoch": 0.1467833172665927, "grad_norm": 1.3318840265274048, "learning_rate": 9.717410235269458e-06, "loss": 0.2059, "step": 8070 }, { "epoch": 0.14696520489641499, "grad_norm": 1.2033460140228271, "learning_rate": 9.716539627970424e-06, "loss": 0.1986, "step": 8080 }, { "epoch": 0.14714709252623728, "grad_norm": 1.2256625890731812, "learning_rate": 9.715667730903321e-06, "loss": 0.2045, "step": 8090 }, { "epoch": 0.14732898015605958, "grad_norm": 1.2747761011123657, "learning_rate": 9.714794544370732e-06, "loss": 0.2086, "step": 8100 }, { "epoch": 0.14751086778588188, "grad_norm": 1.3151066303253174, "learning_rate": 9.713920068675676e-06, "loss": 0.2019, "step": 8110 }, { "epoch": 0.14769275541570417, "grad_norm": 1.2404409646987915, "learning_rate": 9.713044304121632e-06, "loss": 0.1994, "step": 8120 }, { "epoch": 0.14787464304552647, "grad_norm": 1.0892459154129028, "learning_rate": 9.712167251012518e-06, "loss": 0.2046, "step": 8130 }, { "epoch": 0.14805653067534877, "grad_norm": 1.214125394821167, "learning_rate": 9.7112889096527e-06, "loss": 0.2031, "step": 8140 }, { "epoch": 0.14823841830517107, "grad_norm": 1.3248231410980225, "learning_rate": 9.710409280346998e-06, "loss": 0.2059, "step": 8150 }, { "epoch": 0.14842030593499336, "grad_norm": 1.107438325881958, "learning_rate": 9.709528363400669e-06, "loss": 0.2041, "step": 8160 }, { "epoch": 0.14860219356481566, "grad_norm": 1.3557703495025635, "learning_rate": 9.708646159119423e-06, "loss": 0.2117, "step": 8170 }, { "epoch": 0.14878408119463796, "grad_norm": 1.1461529731750488, "learning_rate": 9.707762667809419e-06, "loss": 0.2001, "step": 8180 }, { "epoch": 0.14896596882446025, "grad_norm": 1.1433957815170288, "learning_rate": 9.706877889777254e-06, "loss": 0.2105, "step": 8190 }, { "epoch": 0.14914785645428255, "grad_norm": 1.3031474351882935, "learning_rate": 9.70599182532998e-06, "loss": 0.2048, "step": 8200 }, { "epoch": 0.14932974408410485, "grad_norm": 1.523716688156128, "learning_rate": 9.70510447477509e-06, "loss": 0.2024, "step": 8210 }, { "epoch": 0.14951163171392715, "grad_norm": 1.3183835744857788, "learning_rate": 9.704215838420528e-06, "loss": 0.2018, "step": 8220 }, { "epoch": 0.14969351934374944, "grad_norm": 1.272459864616394, "learning_rate": 9.703325916574677e-06, "loss": 0.2034, "step": 8230 }, { "epoch": 0.14987540697357174, "grad_norm": 1.1589417457580566, "learning_rate": 9.702434709546376e-06, "loss": 0.2039, "step": 8240 }, { "epoch": 0.15005729460339404, "grad_norm": 1.2624086141586304, "learning_rate": 9.701542217644902e-06, "loss": 0.2064, "step": 8250 }, { "epoch": 0.1502391822332163, "grad_norm": 1.206635594367981, "learning_rate": 9.700648441179982e-06, "loss": 0.2035, "step": 8260 }, { "epoch": 0.1504210698630386, "grad_norm": 1.275697112083435, "learning_rate": 9.699753380461785e-06, "loss": 0.1999, "step": 8270 }, { "epoch": 0.1506029574928609, "grad_norm": 1.3041597604751587, "learning_rate": 9.69885703580093e-06, "loss": 0.1974, "step": 8280 }, { "epoch": 0.1507848451226832, "grad_norm": 1.1456183195114136, "learning_rate": 9.69795940750848e-06, "loss": 0.2084, "step": 8290 }, { "epoch": 0.1509667327525055, "grad_norm": 1.181738018989563, "learning_rate": 9.69706049589594e-06, "loss": 0.2018, "step": 8300 }, { "epoch": 0.1511486203823278, "grad_norm": 1.2826589345932007, "learning_rate": 9.696160301275268e-06, "loss": 0.2023, "step": 8310 }, { "epoch": 0.1513305080121501, "grad_norm": 1.2810136079788208, "learning_rate": 9.69525882395886e-06, "loss": 0.2031, "step": 8320 }, { "epoch": 0.15151239564197239, "grad_norm": 1.3606480360031128, "learning_rate": 9.694356064259561e-06, "loss": 0.2013, "step": 8330 }, { "epoch": 0.15169428327179468, "grad_norm": 1.2372937202453613, "learning_rate": 9.69345202249066e-06, "loss": 0.2011, "step": 8340 }, { "epoch": 0.15187617090161698, "grad_norm": 1.1875381469726562, "learning_rate": 9.692546698965887e-06, "loss": 0.2038, "step": 8350 }, { "epoch": 0.15205805853143928, "grad_norm": 1.1035404205322266, "learning_rate": 9.691640093999427e-06, "loss": 0.2012, "step": 8360 }, { "epoch": 0.15223994616126157, "grad_norm": 1.2764623165130615, "learning_rate": 9.690732207905898e-06, "loss": 0.2061, "step": 8370 }, { "epoch": 0.15242183379108387, "grad_norm": 1.263163447380066, "learning_rate": 9.689823041000372e-06, "loss": 0.208, "step": 8380 }, { "epoch": 0.15260372142090617, "grad_norm": 1.1590152978897095, "learning_rate": 9.68891259359836e-06, "loss": 0.2031, "step": 8390 }, { "epoch": 0.15278560905072847, "grad_norm": 1.2139486074447632, "learning_rate": 9.688000866015816e-06, "loss": 0.2034, "step": 8400 }, { "epoch": 0.15296749668055076, "grad_norm": 1.325201392173767, "learning_rate": 9.687087858569145e-06, "loss": 0.2085, "step": 8410 }, { "epoch": 0.15314938431037306, "grad_norm": 1.2177867889404297, "learning_rate": 9.68617357157519e-06, "loss": 0.2022, "step": 8420 }, { "epoch": 0.15333127194019536, "grad_norm": 1.3213423490524292, "learning_rate": 9.68525800535124e-06, "loss": 0.2015, "step": 8430 }, { "epoch": 0.15351315957001765, "grad_norm": 1.2851548194885254, "learning_rate": 9.68434116021503e-06, "loss": 0.2032, "step": 8440 }, { "epoch": 0.15369504719983995, "grad_norm": 1.223699927330017, "learning_rate": 9.683423036484735e-06, "loss": 0.2043, "step": 8450 }, { "epoch": 0.15387693482966222, "grad_norm": 1.1216609477996826, "learning_rate": 9.68250363447898e-06, "loss": 0.2009, "step": 8460 }, { "epoch": 0.15405882245948452, "grad_norm": 1.1287636756896973, "learning_rate": 9.681582954516823e-06, "loss": 0.2012, "step": 8470 }, { "epoch": 0.15424071008930682, "grad_norm": 1.1828080415725708, "learning_rate": 9.680660996917778e-06, "loss": 0.2013, "step": 8480 }, { "epoch": 0.1544225977191291, "grad_norm": 1.3509445190429688, "learning_rate": 9.679737762001793e-06, "loss": 0.2059, "step": 8490 }, { "epoch": 0.1546044853489514, "grad_norm": 1.1541754007339478, "learning_rate": 9.678813250089261e-06, "loss": 0.2096, "step": 8500 }, { "epoch": 0.1547863729787737, "grad_norm": 1.2574920654296875, "learning_rate": 9.677887461501023e-06, "loss": 0.2055, "step": 8510 }, { "epoch": 0.154968260608596, "grad_norm": 1.2498879432678223, "learning_rate": 9.676960396558359e-06, "loss": 0.2116, "step": 8520 }, { "epoch": 0.1551501482384183, "grad_norm": 1.270896077156067, "learning_rate": 9.67603205558299e-06, "loss": 0.1991, "step": 8530 }, { "epoch": 0.1553320358682406, "grad_norm": 1.48837411403656, "learning_rate": 9.675102438897086e-06, "loss": 0.2024, "step": 8540 }, { "epoch": 0.1555139234980629, "grad_norm": 1.0683718919754028, "learning_rate": 9.674171546823253e-06, "loss": 0.1995, "step": 8550 }, { "epoch": 0.1556958111278852, "grad_norm": 1.3667972087860107, "learning_rate": 9.673239379684546e-06, "loss": 0.204, "step": 8560 }, { "epoch": 0.1558776987577075, "grad_norm": 1.1413387060165405, "learning_rate": 9.672305937804456e-06, "loss": 0.2069, "step": 8570 }, { "epoch": 0.1560595863875298, "grad_norm": 1.1252026557922363, "learning_rate": 9.671371221506925e-06, "loss": 0.1999, "step": 8580 }, { "epoch": 0.15624147401735208, "grad_norm": 1.208566427230835, "learning_rate": 9.670435231116325e-06, "loss": 0.209, "step": 8590 }, { "epoch": 0.15642336164717438, "grad_norm": 1.2742613554000854, "learning_rate": 9.669497966957483e-06, "loss": 0.2042, "step": 8600 }, { "epoch": 0.15660524927699668, "grad_norm": 1.2481261491775513, "learning_rate": 9.668559429355657e-06, "loss": 0.2048, "step": 8610 }, { "epoch": 0.15678713690681897, "grad_norm": 1.1632429361343384, "learning_rate": 9.667619618636556e-06, "loss": 0.2051, "step": 8620 }, { "epoch": 0.15696902453664127, "grad_norm": 1.1873955726623535, "learning_rate": 9.666678535126327e-06, "loss": 0.2095, "step": 8630 }, { "epoch": 0.15715091216646357, "grad_norm": 1.222371220588684, "learning_rate": 9.665736179151558e-06, "loss": 0.2059, "step": 8640 }, { "epoch": 0.15733279979628587, "grad_norm": 1.2221559286117554, "learning_rate": 9.664792551039275e-06, "loss": 0.2022, "step": 8650 }, { "epoch": 0.15751468742610816, "grad_norm": 1.1163411140441895, "learning_rate": 9.663847651116955e-06, "loss": 0.2049, "step": 8660 }, { "epoch": 0.15769657505593043, "grad_norm": 1.2033820152282715, "learning_rate": 9.66290147971251e-06, "loss": 0.2074, "step": 8670 }, { "epoch": 0.15787846268575273, "grad_norm": 1.1449280977249146, "learning_rate": 9.661954037154293e-06, "loss": 0.2092, "step": 8680 }, { "epoch": 0.15806035031557503, "grad_norm": 1.3219293355941772, "learning_rate": 9.661005323771098e-06, "loss": 0.2052, "step": 8690 }, { "epoch": 0.15824223794539732, "grad_norm": 1.2712159156799316, "learning_rate": 9.660055339892166e-06, "loss": 0.2053, "step": 8700 }, { "epoch": 0.15842412557521962, "grad_norm": 1.2301836013793945, "learning_rate": 9.65910408584717e-06, "loss": 0.2043, "step": 8710 }, { "epoch": 0.15860601320504192, "grad_norm": 1.2510608434677124, "learning_rate": 9.65815156196623e-06, "loss": 0.2068, "step": 8720 }, { "epoch": 0.15878790083486422, "grad_norm": 1.161624789237976, "learning_rate": 9.657197768579905e-06, "loss": 0.2056, "step": 8730 }, { "epoch": 0.1589697884646865, "grad_norm": 1.3088337182998657, "learning_rate": 9.656242706019193e-06, "loss": 0.2023, "step": 8740 }, { "epoch": 0.1591516760945088, "grad_norm": 1.261197805404663, "learning_rate": 9.655286374615534e-06, "loss": 0.2066, "step": 8750 }, { "epoch": 0.1593335637243311, "grad_norm": 1.2525779008865356, "learning_rate": 9.654328774700811e-06, "loss": 0.2046, "step": 8760 }, { "epoch": 0.1595154513541534, "grad_norm": 1.2087912559509277, "learning_rate": 9.653369906607341e-06, "loss": 0.2065, "step": 8770 }, { "epoch": 0.1596973389839757, "grad_norm": 1.0791152715682983, "learning_rate": 9.652409770667886e-06, "loss": 0.2081, "step": 8780 }, { "epoch": 0.159879226613798, "grad_norm": 1.3153890371322632, "learning_rate": 9.651448367215646e-06, "loss": 0.2082, "step": 8790 }, { "epoch": 0.1600611142436203, "grad_norm": 1.1332933902740479, "learning_rate": 9.65048569658426e-06, "loss": 0.2007, "step": 8800 }, { "epoch": 0.1602430018734426, "grad_norm": 1.1836415529251099, "learning_rate": 9.649521759107812e-06, "loss": 0.2044, "step": 8810 }, { "epoch": 0.1604248895032649, "grad_norm": 1.1890449523925781, "learning_rate": 9.648556555120817e-06, "loss": 0.209, "step": 8820 }, { "epoch": 0.1606067771330872, "grad_norm": 1.1036098003387451, "learning_rate": 9.647590084958238e-06, "loss": 0.2066, "step": 8830 }, { "epoch": 0.16078866476290948, "grad_norm": 1.1960350275039673, "learning_rate": 9.64662234895547e-06, "loss": 0.2001, "step": 8840 }, { "epoch": 0.16097055239273178, "grad_norm": 1.298035979270935, "learning_rate": 9.645653347448356e-06, "loss": 0.2054, "step": 8850 }, { "epoch": 0.16115244002255408, "grad_norm": 1.0827739238739014, "learning_rate": 9.64468308077317e-06, "loss": 0.2008, "step": 8860 }, { "epoch": 0.16133432765237637, "grad_norm": 1.3310248851776123, "learning_rate": 9.643711549266625e-06, "loss": 0.2065, "step": 8870 }, { "epoch": 0.16151621528219864, "grad_norm": 1.3814090490341187, "learning_rate": 9.642738753265882e-06, "loss": 0.2099, "step": 8880 }, { "epoch": 0.16169810291202094, "grad_norm": 1.236280918121338, "learning_rate": 9.641764693108532e-06, "loss": 0.2075, "step": 8890 }, { "epoch": 0.16187999054184324, "grad_norm": 1.329544186592102, "learning_rate": 9.640789369132609e-06, "loss": 0.2065, "step": 8900 }, { "epoch": 0.16206187817166554, "grad_norm": 1.2200772762298584, "learning_rate": 9.639812781676581e-06, "loss": 0.2059, "step": 8910 }, { "epoch": 0.16224376580148783, "grad_norm": 1.2127430438995361, "learning_rate": 9.638834931079362e-06, "loss": 0.2045, "step": 8920 }, { "epoch": 0.16242565343131013, "grad_norm": 1.2585093975067139, "learning_rate": 9.637855817680296e-06, "loss": 0.2019, "step": 8930 }, { "epoch": 0.16260754106113243, "grad_norm": 1.1843936443328857, "learning_rate": 9.636875441819173e-06, "loss": 0.202, "step": 8940 }, { "epoch": 0.16278942869095472, "grad_norm": 1.1939257383346558, "learning_rate": 9.635893803836213e-06, "loss": 0.202, "step": 8950 }, { "epoch": 0.16297131632077702, "grad_norm": 1.2697182893753052, "learning_rate": 9.634910904072084e-06, "loss": 0.2102, "step": 8960 }, { "epoch": 0.16315320395059932, "grad_norm": 1.369160532951355, "learning_rate": 9.633926742867884e-06, "loss": 0.2032, "step": 8970 }, { "epoch": 0.16333509158042162, "grad_norm": 1.1185448169708252, "learning_rate": 9.632941320565146e-06, "loss": 0.2083, "step": 8980 }, { "epoch": 0.1635169792102439, "grad_norm": 1.1596076488494873, "learning_rate": 9.631954637505854e-06, "loss": 0.2075, "step": 8990 }, { "epoch": 0.1636988668400662, "grad_norm": 1.2383697032928467, "learning_rate": 9.630966694032415e-06, "loss": 0.1999, "step": 9000 }, { "epoch": 0.1638807544698885, "grad_norm": 1.2686270475387573, "learning_rate": 9.629977490487684e-06, "loss": 0.199, "step": 9010 }, { "epoch": 0.1640626420997108, "grad_norm": 1.3145558834075928, "learning_rate": 9.628987027214945e-06, "loss": 0.2081, "step": 9020 }, { "epoch": 0.1642445297295331, "grad_norm": 1.1714657545089722, "learning_rate": 9.627995304557924e-06, "loss": 0.2066, "step": 9030 }, { "epoch": 0.1644264173593554, "grad_norm": 1.2524043321609497, "learning_rate": 9.627002322860787e-06, "loss": 0.198, "step": 9040 }, { "epoch": 0.1646083049891777, "grad_norm": 1.3364479541778564, "learning_rate": 9.62600808246813e-06, "loss": 0.2044, "step": 9050 }, { "epoch": 0.164790192619, "grad_norm": 1.1927906274795532, "learning_rate": 9.625012583724988e-06, "loss": 0.2075, "step": 9060 }, { "epoch": 0.1649720802488223, "grad_norm": 1.3231987953186035, "learning_rate": 9.624015826976832e-06, "loss": 0.2087, "step": 9070 }, { "epoch": 0.1651539678786446, "grad_norm": 1.3189643621444702, "learning_rate": 9.623017812569578e-06, "loss": 0.2078, "step": 9080 }, { "epoch": 0.16533585550846686, "grad_norm": 1.1701757907867432, "learning_rate": 9.622018540849566e-06, "loss": 0.2054, "step": 9090 }, { "epoch": 0.16551774313828915, "grad_norm": 1.2364568710327148, "learning_rate": 9.621018012163577e-06, "loss": 0.2067, "step": 9100 }, { "epoch": 0.16569963076811145, "grad_norm": 1.1360710859298706, "learning_rate": 9.620016226858835e-06, "loss": 0.209, "step": 9110 }, { "epoch": 0.16588151839793375, "grad_norm": 4.04784631729126, "learning_rate": 9.619013185282988e-06, "loss": 0.2059, "step": 9120 }, { "epoch": 0.16606340602775604, "grad_norm": 1.2707104682922363, "learning_rate": 9.61800888778413e-06, "loss": 0.2016, "step": 9130 }, { "epoch": 0.16624529365757834, "grad_norm": 1.1705626249313354, "learning_rate": 9.617003334710784e-06, "loss": 0.1991, "step": 9140 }, { "epoch": 0.16642718128740064, "grad_norm": 1.477776288986206, "learning_rate": 9.615996526411915e-06, "loss": 0.2004, "step": 9150 }, { "epoch": 0.16660906891722294, "grad_norm": 1.2454384565353394, "learning_rate": 9.614988463236919e-06, "loss": 0.2035, "step": 9160 }, { "epoch": 0.16679095654704523, "grad_norm": 1.2420705556869507, "learning_rate": 9.61397914553563e-06, "loss": 0.2075, "step": 9170 }, { "epoch": 0.16697284417686753, "grad_norm": 1.1833326816558838, "learning_rate": 9.612968573658314e-06, "loss": 0.2043, "step": 9180 }, { "epoch": 0.16715473180668983, "grad_norm": 1.251091718673706, "learning_rate": 9.611956747955675e-06, "loss": 0.2065, "step": 9190 }, { "epoch": 0.16733661943651212, "grad_norm": 1.2736601829528809, "learning_rate": 9.61094366877885e-06, "loss": 0.2021, "step": 9200 }, { "epoch": 0.16751850706633442, "grad_norm": 1.2156418561935425, "learning_rate": 9.609929336479419e-06, "loss": 0.2044, "step": 9210 }, { "epoch": 0.16770039469615672, "grad_norm": 1.2737362384796143, "learning_rate": 9.608913751409383e-06, "loss": 0.2048, "step": 9220 }, { "epoch": 0.16788228232597902, "grad_norm": 1.2745249271392822, "learning_rate": 9.607896913921192e-06, "loss": 0.2075, "step": 9230 }, { "epoch": 0.1680641699558013, "grad_norm": 1.1074798107147217, "learning_rate": 9.606878824367715e-06, "loss": 0.2027, "step": 9240 }, { "epoch": 0.1682460575856236, "grad_norm": 1.3677560091018677, "learning_rate": 9.605859483102274e-06, "loss": 0.2018, "step": 9250 }, { "epoch": 0.1684279452154459, "grad_norm": 1.2134233713150024, "learning_rate": 9.60483889047861e-06, "loss": 0.2048, "step": 9260 }, { "epoch": 0.1686098328452682, "grad_norm": 1.117648720741272, "learning_rate": 9.603817046850904e-06, "loss": 0.2027, "step": 9270 }, { "epoch": 0.1687917204750905, "grad_norm": 1.1616350412368774, "learning_rate": 9.602793952573774e-06, "loss": 0.2125, "step": 9280 }, { "epoch": 0.1689736081049128, "grad_norm": 1.2022018432617188, "learning_rate": 9.601769608002267e-06, "loss": 0.1995, "step": 9290 }, { "epoch": 0.16915549573473507, "grad_norm": 1.2443410158157349, "learning_rate": 9.600744013491868e-06, "loss": 0.2056, "step": 9300 }, { "epoch": 0.16933738336455736, "grad_norm": 1.1245598793029785, "learning_rate": 9.599717169398492e-06, "loss": 0.1994, "step": 9310 }, { "epoch": 0.16951927099437966, "grad_norm": 1.2104740142822266, "learning_rate": 9.59868907607849e-06, "loss": 0.2079, "step": 9320 }, { "epoch": 0.16970115862420196, "grad_norm": 1.2367461919784546, "learning_rate": 9.597659733888645e-06, "loss": 0.2063, "step": 9330 }, { "epoch": 0.16988304625402426, "grad_norm": 1.3144422769546509, "learning_rate": 9.596629143186178e-06, "loss": 0.208, "step": 9340 }, { "epoch": 0.17006493388384655, "grad_norm": 1.1833688020706177, "learning_rate": 9.595597304328735e-06, "loss": 0.2051, "step": 9350 }, { "epoch": 0.17024682151366885, "grad_norm": 1.2102400064468384, "learning_rate": 9.594564217674402e-06, "loss": 0.2118, "step": 9360 }, { "epoch": 0.17042870914349115, "grad_norm": 1.2092446088790894, "learning_rate": 9.593529883581699e-06, "loss": 0.2008, "step": 9370 }, { "epoch": 0.17061059677331344, "grad_norm": 1.1509653329849243, "learning_rate": 9.59249430240957e-06, "loss": 0.2054, "step": 9380 }, { "epoch": 0.17079248440313574, "grad_norm": 1.2275534868240356, "learning_rate": 9.5914574745174e-06, "loss": 0.2029, "step": 9390 }, { "epoch": 0.17097437203295804, "grad_norm": 1.1511934995651245, "learning_rate": 9.590419400265005e-06, "loss": 0.2048, "step": 9400 }, { "epoch": 0.17115625966278034, "grad_norm": 1.260895013809204, "learning_rate": 9.589380080012633e-06, "loss": 0.2032, "step": 9410 }, { "epoch": 0.17133814729260263, "grad_norm": 1.1312109231948853, "learning_rate": 9.588339514120963e-06, "loss": 0.2051, "step": 9420 }, { "epoch": 0.17152003492242493, "grad_norm": 1.1486672163009644, "learning_rate": 9.587297702951109e-06, "loss": 0.204, "step": 9430 }, { "epoch": 0.17170192255224723, "grad_norm": 1.1136952638626099, "learning_rate": 9.586254646864612e-06, "loss": 0.2011, "step": 9440 }, { "epoch": 0.17188381018206952, "grad_norm": 1.135048508644104, "learning_rate": 9.585210346223453e-06, "loss": 0.204, "step": 9450 }, { "epoch": 0.17206569781189182, "grad_norm": 1.1265591382980347, "learning_rate": 9.584164801390039e-06, "loss": 0.2037, "step": 9460 }, { "epoch": 0.17224758544171412, "grad_norm": 1.3972450494766235, "learning_rate": 9.583118012727208e-06, "loss": 0.2032, "step": 9470 }, { "epoch": 0.17242947307153642, "grad_norm": 1.0960749387741089, "learning_rate": 9.582069980598235e-06, "loss": 0.2016, "step": 9480 }, { "epoch": 0.1726113607013587, "grad_norm": 1.2884175777435303, "learning_rate": 9.58102070536682e-06, "loss": 0.2083, "step": 9490 }, { "epoch": 0.172793248331181, "grad_norm": 1.3119328022003174, "learning_rate": 9.579970187397103e-06, "loss": 0.2035, "step": 9500 }, { "epoch": 0.17297513596100328, "grad_norm": 1.2609689235687256, "learning_rate": 9.578918427053646e-06, "loss": 0.2046, "step": 9510 }, { "epoch": 0.17315702359082558, "grad_norm": 1.4489425420761108, "learning_rate": 9.577865424701448e-06, "loss": 0.2061, "step": 9520 }, { "epoch": 0.17333891122064787, "grad_norm": 1.2707335948944092, "learning_rate": 9.576811180705937e-06, "loss": 0.2079, "step": 9530 }, { "epoch": 0.17352079885047017, "grad_norm": 1.1698447465896606, "learning_rate": 9.575755695432972e-06, "loss": 0.2031, "step": 9540 }, { "epoch": 0.17370268648029247, "grad_norm": 1.050032615661621, "learning_rate": 9.574698969248844e-06, "loss": 0.204, "step": 9550 }, { "epoch": 0.17388457411011476, "grad_norm": 1.12115478515625, "learning_rate": 9.57364100252027e-06, "loss": 0.197, "step": 9560 }, { "epoch": 0.17406646173993706, "grad_norm": 1.3241275548934937, "learning_rate": 9.572581795614407e-06, "loss": 0.2057, "step": 9570 }, { "epoch": 0.17424834936975936, "grad_norm": 1.1743230819702148, "learning_rate": 9.571521348898831e-06, "loss": 0.2068, "step": 9580 }, { "epoch": 0.17443023699958166, "grad_norm": 1.065850019454956, "learning_rate": 9.570459662741558e-06, "loss": 0.2032, "step": 9590 }, { "epoch": 0.17461212462940395, "grad_norm": 1.3414216041564941, "learning_rate": 9.569396737511026e-06, "loss": 0.2051, "step": 9600 }, { "epoch": 0.17479401225922625, "grad_norm": 1.177623987197876, "learning_rate": 9.56833257357611e-06, "loss": 0.1968, "step": 9610 }, { "epoch": 0.17497589988904855, "grad_norm": 1.1938071250915527, "learning_rate": 9.56726717130611e-06, "loss": 0.1974, "step": 9620 }, { "epoch": 0.17515778751887084, "grad_norm": 1.190413236618042, "learning_rate": 9.566200531070759e-06, "loss": 0.206, "step": 9630 }, { "epoch": 0.17533967514869314, "grad_norm": 1.2499895095825195, "learning_rate": 9.565132653240218e-06, "loss": 0.1996, "step": 9640 }, { "epoch": 0.17552156277851544, "grad_norm": 1.6273319721221924, "learning_rate": 9.564063538185076e-06, "loss": 0.2022, "step": 9650 }, { "epoch": 0.17570345040833774, "grad_norm": 1.2264381647109985, "learning_rate": 9.562993186276354e-06, "loss": 0.2058, "step": 9660 }, { "epoch": 0.17588533803816003, "grad_norm": 1.1919559240341187, "learning_rate": 9.5619215978855e-06, "loss": 0.205, "step": 9670 }, { "epoch": 0.17606722566798233, "grad_norm": 1.4804728031158447, "learning_rate": 9.560848773384394e-06, "loss": 0.2036, "step": 9680 }, { "epoch": 0.17624911329780463, "grad_norm": 1.0691962242126465, "learning_rate": 9.559774713145343e-06, "loss": 0.2043, "step": 9690 }, { "epoch": 0.17643100092762692, "grad_norm": 1.1888394355773926, "learning_rate": 9.558699417541084e-06, "loss": 0.2063, "step": 9700 }, { "epoch": 0.17661288855744922, "grad_norm": 1.2647475004196167, "learning_rate": 9.557622886944779e-06, "loss": 0.2084, "step": 9710 }, { "epoch": 0.1767947761872715, "grad_norm": 1.1395543813705444, "learning_rate": 9.556545121730024e-06, "loss": 0.204, "step": 9720 }, { "epoch": 0.1769766638170938, "grad_norm": 1.2001235485076904, "learning_rate": 9.55546612227084e-06, "loss": 0.2079, "step": 9730 }, { "epoch": 0.17715855144691608, "grad_norm": 1.300815463066101, "learning_rate": 9.554385888941678e-06, "loss": 0.2077, "step": 9740 }, { "epoch": 0.17734043907673838, "grad_norm": 1.2244254350662231, "learning_rate": 9.553304422117414e-06, "loss": 0.2033, "step": 9750 }, { "epoch": 0.17752232670656068, "grad_norm": 1.165794014930725, "learning_rate": 9.552221722173356e-06, "loss": 0.2074, "step": 9760 }, { "epoch": 0.17770421433638298, "grad_norm": 1.2453926801681519, "learning_rate": 9.55113778948524e-06, "loss": 0.199, "step": 9770 }, { "epoch": 0.17788610196620527, "grad_norm": 1.2637500762939453, "learning_rate": 9.550052624429227e-06, "loss": 0.1995, "step": 9780 }, { "epoch": 0.17806798959602757, "grad_norm": 1.1128952503204346, "learning_rate": 9.548966227381903e-06, "loss": 0.2007, "step": 9790 }, { "epoch": 0.17824987722584987, "grad_norm": 1.1896483898162842, "learning_rate": 9.547878598720292e-06, "loss": 0.2057, "step": 9800 }, { "epoch": 0.17843176485567216, "grad_norm": 1.269689679145813, "learning_rate": 9.546789738821837e-06, "loss": 0.2072, "step": 9810 }, { "epoch": 0.17861365248549446, "grad_norm": 1.1925259828567505, "learning_rate": 9.545699648064408e-06, "loss": 0.2055, "step": 9820 }, { "epoch": 0.17879554011531676, "grad_norm": 1.0713609457015991, "learning_rate": 9.544608326826303e-06, "loss": 0.2019, "step": 9830 }, { "epoch": 0.17897742774513906, "grad_norm": 1.3235448598861694, "learning_rate": 9.543515775486253e-06, "loss": 0.2019, "step": 9840 }, { "epoch": 0.17915931537496135, "grad_norm": 1.2674704790115356, "learning_rate": 9.54242199442341e-06, "loss": 0.2044, "step": 9850 }, { "epoch": 0.17934120300478365, "grad_norm": 1.1459300518035889, "learning_rate": 9.541326984017349e-06, "loss": 0.2095, "step": 9860 }, { "epoch": 0.17952309063460595, "grad_norm": 1.218822956085205, "learning_rate": 9.540230744648083e-06, "loss": 0.2042, "step": 9870 }, { "epoch": 0.17970497826442824, "grad_norm": 1.3011091947555542, "learning_rate": 9.539133276696042e-06, "loss": 0.2035, "step": 9880 }, { "epoch": 0.17988686589425054, "grad_norm": 1.3885401487350464, "learning_rate": 9.538034580542083e-06, "loss": 0.2106, "step": 9890 }, { "epoch": 0.18006875352407284, "grad_norm": 1.185119867324829, "learning_rate": 9.536934656567497e-06, "loss": 0.2069, "step": 9900 }, { "epoch": 0.18025064115389514, "grad_norm": 1.2718002796173096, "learning_rate": 9.535833505153993e-06, "loss": 0.2034, "step": 9910 }, { "epoch": 0.18043252878371743, "grad_norm": 1.2904762029647827, "learning_rate": 9.534731126683708e-06, "loss": 0.2069, "step": 9920 }, { "epoch": 0.1806144164135397, "grad_norm": 1.22158944606781, "learning_rate": 9.533627521539207e-06, "loss": 0.204, "step": 9930 }, { "epoch": 0.180796304043362, "grad_norm": 1.1100610494613647, "learning_rate": 9.53252269010348e-06, "loss": 0.1993, "step": 9940 }, { "epoch": 0.1809781916731843, "grad_norm": 1.3144199848175049, "learning_rate": 9.53141663275994e-06, "loss": 0.2045, "step": 9950 }, { "epoch": 0.1811600793030066, "grad_norm": 1.1769694089889526, "learning_rate": 9.530309349892425e-06, "loss": 0.208, "step": 9960 }, { "epoch": 0.1813419669328289, "grad_norm": 1.1617391109466553, "learning_rate": 9.529200841885206e-06, "loss": 0.2054, "step": 9970 }, { "epoch": 0.1815238545626512, "grad_norm": 1.2725718021392822, "learning_rate": 9.528091109122968e-06, "loss": 0.2069, "step": 9980 }, { "epoch": 0.18170574219247349, "grad_norm": 1.164806604385376, "learning_rate": 9.52698015199083e-06, "loss": 0.2047, "step": 9990 }, { "epoch": 0.18188762982229578, "grad_norm": 1.0623137950897217, "learning_rate": 9.525867970874336e-06, "loss": 0.2073, "step": 10000 }, { "epoch": 0.18206951745211808, "grad_norm": 1.273817777633667, "learning_rate": 9.524754566159446e-06, "loss": 0.2076, "step": 10010 }, { "epoch": 0.18225140508194038, "grad_norm": 1.0059643983840942, "learning_rate": 9.523639938232549e-06, "loss": 0.2021, "step": 10020 }, { "epoch": 0.18243329271176267, "grad_norm": 1.1743406057357788, "learning_rate": 9.522524087480465e-06, "loss": 0.2027, "step": 10030 }, { "epoch": 0.18261518034158497, "grad_norm": 1.2861663103103638, "learning_rate": 9.52140701429043e-06, "loss": 0.2078, "step": 10040 }, { "epoch": 0.18279706797140727, "grad_norm": 1.1542563438415527, "learning_rate": 9.520288719050104e-06, "loss": 0.2044, "step": 10050 }, { "epoch": 0.18297895560122956, "grad_norm": 1.1543684005737305, "learning_rate": 9.519169202147581e-06, "loss": 0.2039, "step": 10060 }, { "epoch": 0.18316084323105186, "grad_norm": 1.1192841529846191, "learning_rate": 9.518048463971368e-06, "loss": 0.2079, "step": 10070 }, { "epoch": 0.18334273086087416, "grad_norm": 1.2542691230773926, "learning_rate": 9.516926504910399e-06, "loss": 0.2048, "step": 10080 }, { "epoch": 0.18352461849069646, "grad_norm": 1.1450704336166382, "learning_rate": 9.515803325354035e-06, "loss": 0.2057, "step": 10090 }, { "epoch": 0.18370650612051875, "grad_norm": 1.03681480884552, "learning_rate": 9.514678925692058e-06, "loss": 0.2052, "step": 10100 }, { "epoch": 0.18388839375034105, "grad_norm": 1.2106249332427979, "learning_rate": 9.513553306314672e-06, "loss": 0.2071, "step": 10110 }, { "epoch": 0.18407028138016335, "grad_norm": 1.1555790901184082, "learning_rate": 9.512426467612507e-06, "loss": 0.2047, "step": 10120 }, { "epoch": 0.18425216900998564, "grad_norm": 1.1019535064697266, "learning_rate": 9.511298409976617e-06, "loss": 0.2076, "step": 10130 }, { "epoch": 0.18443405663980791, "grad_norm": 1.1599723100662231, "learning_rate": 9.510169133798474e-06, "loss": 0.2042, "step": 10140 }, { "epoch": 0.1846159442696302, "grad_norm": 1.1230344772338867, "learning_rate": 9.509038639469979e-06, "loss": 0.2023, "step": 10150 }, { "epoch": 0.1847978318994525, "grad_norm": 1.1224297285079956, "learning_rate": 9.507906927383449e-06, "loss": 0.203, "step": 10160 }, { "epoch": 0.1849797195292748, "grad_norm": 1.2464066743850708, "learning_rate": 9.50677399793163e-06, "loss": 0.2048, "step": 10170 }, { "epoch": 0.1851616071590971, "grad_norm": 1.1495904922485352, "learning_rate": 9.505639851507686e-06, "loss": 0.2054, "step": 10180 }, { "epoch": 0.1853434947889194, "grad_norm": 1.2982914447784424, "learning_rate": 9.504504488505208e-06, "loss": 0.205, "step": 10190 }, { "epoch": 0.1855253824187417, "grad_norm": 1.2651275396347046, "learning_rate": 9.503367909318203e-06, "loss": 0.2045, "step": 10200 }, { "epoch": 0.185707270048564, "grad_norm": 1.1924387216567993, "learning_rate": 9.502230114341105e-06, "loss": 0.1998, "step": 10210 }, { "epoch": 0.1858891576783863, "grad_norm": 1.1721303462982178, "learning_rate": 9.501091103968768e-06, "loss": 0.202, "step": 10220 }, { "epoch": 0.1860710453082086, "grad_norm": 1.1974214315414429, "learning_rate": 9.499950878596468e-06, "loss": 0.2068, "step": 10230 }, { "epoch": 0.18625293293803089, "grad_norm": 1.3243749141693115, "learning_rate": 9.498809438619905e-06, "loss": 0.2011, "step": 10240 }, { "epoch": 0.18643482056785318, "grad_norm": 1.2544305324554443, "learning_rate": 9.497666784435194e-06, "loss": 0.2071, "step": 10250 }, { "epoch": 0.18661670819767548, "grad_norm": 1.2359765768051147, "learning_rate": 9.49652291643888e-06, "loss": 0.2094, "step": 10260 }, { "epoch": 0.18679859582749778, "grad_norm": 1.170343041419983, "learning_rate": 9.495377835027922e-06, "loss": 0.2056, "step": 10270 }, { "epoch": 0.18698048345732007, "grad_norm": 1.1282341480255127, "learning_rate": 9.494231540599703e-06, "loss": 0.2018, "step": 10280 }, { "epoch": 0.18716237108714237, "grad_norm": 1.142945408821106, "learning_rate": 9.493084033552026e-06, "loss": 0.2022, "step": 10290 }, { "epoch": 0.18734425871696467, "grad_norm": 1.2115294933319092, "learning_rate": 9.491935314283119e-06, "loss": 0.2074, "step": 10300 }, { "epoch": 0.18752614634678696, "grad_norm": 1.2698603868484497, "learning_rate": 9.490785383191625e-06, "loss": 0.2074, "step": 10310 }, { "epoch": 0.18770803397660926, "grad_norm": 1.2862892150878906, "learning_rate": 9.48963424067661e-06, "loss": 0.205, "step": 10320 }, { "epoch": 0.18788992160643156, "grad_norm": 1.1177188158035278, "learning_rate": 9.488481887137562e-06, "loss": 0.2051, "step": 10330 }, { "epoch": 0.18807180923625383, "grad_norm": 1.1950029134750366, "learning_rate": 9.487328322974388e-06, "loss": 0.2101, "step": 10340 }, { "epoch": 0.18825369686607613, "grad_norm": 1.1896470785140991, "learning_rate": 9.486173548587409e-06, "loss": 0.2043, "step": 10350 }, { "epoch": 0.18843558449589842, "grad_norm": 1.3148009777069092, "learning_rate": 9.48501756437738e-06, "loss": 0.2066, "step": 10360 }, { "epoch": 0.18861747212572072, "grad_norm": 1.1876869201660156, "learning_rate": 9.483860370745462e-06, "loss": 0.2053, "step": 10370 }, { "epoch": 0.18879935975554302, "grad_norm": 1.3242968320846558, "learning_rate": 9.482701968093242e-06, "loss": 0.204, "step": 10380 }, { "epoch": 0.18898124738536531, "grad_norm": 1.2138035297393799, "learning_rate": 9.481542356822728e-06, "loss": 0.2081, "step": 10390 }, { "epoch": 0.1891631350151876, "grad_norm": 1.1874767541885376, "learning_rate": 9.480381537336344e-06, "loss": 0.2038, "step": 10400 }, { "epoch": 0.1893450226450099, "grad_norm": 1.0907520055770874, "learning_rate": 9.479219510036935e-06, "loss": 0.2014, "step": 10410 }, { "epoch": 0.1895269102748322, "grad_norm": 1.1858952045440674, "learning_rate": 9.478056275327764e-06, "loss": 0.2093, "step": 10420 }, { "epoch": 0.1897087979046545, "grad_norm": 1.4200718402862549, "learning_rate": 9.476891833612514e-06, "loss": 0.1987, "step": 10430 }, { "epoch": 0.1898906855344768, "grad_norm": 1.1501787900924683, "learning_rate": 9.475726185295287e-06, "loss": 0.2063, "step": 10440 }, { "epoch": 0.1900725731642991, "grad_norm": 1.2744396924972534, "learning_rate": 9.474559330780603e-06, "loss": 0.2038, "step": 10450 }, { "epoch": 0.1902544607941214, "grad_norm": 1.2336671352386475, "learning_rate": 9.4733912704734e-06, "loss": 0.2033, "step": 10460 }, { "epoch": 0.1904363484239437, "grad_norm": 1.1195048093795776, "learning_rate": 9.472222004779039e-06, "loss": 0.2052, "step": 10470 }, { "epoch": 0.190618236053766, "grad_norm": 1.1645137071609497, "learning_rate": 9.471051534103294e-06, "loss": 0.2059, "step": 10480 }, { "epoch": 0.19080012368358829, "grad_norm": 1.2005435228347778, "learning_rate": 9.469879858852355e-06, "loss": 0.2037, "step": 10490 }, { "epoch": 0.19098201131341058, "grad_norm": 1.0980713367462158, "learning_rate": 9.468706979432838e-06, "loss": 0.2026, "step": 10500 }, { "epoch": 0.19116389894323288, "grad_norm": 1.1369038820266724, "learning_rate": 9.467532896251772e-06, "loss": 0.2009, "step": 10510 }, { "epoch": 0.19134578657305518, "grad_norm": 1.223918080329895, "learning_rate": 9.466357609716605e-06, "loss": 0.1988, "step": 10520 }, { "epoch": 0.19152767420287747, "grad_norm": 1.4620968103408813, "learning_rate": 9.465181120235202e-06, "loss": 0.1996, "step": 10530 }, { "epoch": 0.19170956183269977, "grad_norm": 1.0091949701309204, "learning_rate": 9.464003428215844e-06, "loss": 0.2095, "step": 10540 }, { "epoch": 0.19189144946252204, "grad_norm": 1.185293197631836, "learning_rate": 9.462824534067232e-06, "loss": 0.1964, "step": 10550 }, { "epoch": 0.19207333709234434, "grad_norm": 1.1529943943023682, "learning_rate": 9.461644438198483e-06, "loss": 0.2038, "step": 10560 }, { "epoch": 0.19225522472216663, "grad_norm": 1.0986497402191162, "learning_rate": 9.460463141019133e-06, "loss": 0.2051, "step": 10570 }, { "epoch": 0.19243711235198893, "grad_norm": 1.1696445941925049, "learning_rate": 9.459280642939132e-06, "loss": 0.2019, "step": 10580 }, { "epoch": 0.19261899998181123, "grad_norm": 1.2223596572875977, "learning_rate": 9.458096944368846e-06, "loss": 0.2073, "step": 10590 }, { "epoch": 0.19280088761163353, "grad_norm": 1.2852977514266968, "learning_rate": 9.456912045719063e-06, "loss": 0.2069, "step": 10600 }, { "epoch": 0.19298277524145582, "grad_norm": 1.2646054029464722, "learning_rate": 9.45572594740098e-06, "loss": 0.203, "step": 10610 }, { "epoch": 0.19316466287127812, "grad_norm": 1.2625905275344849, "learning_rate": 9.454538649826216e-06, "loss": 0.2053, "step": 10620 }, { "epoch": 0.19334655050110042, "grad_norm": 1.2310161590576172, "learning_rate": 9.453350153406804e-06, "loss": 0.2088, "step": 10630 }, { "epoch": 0.19352843813092271, "grad_norm": 1.2546939849853516, "learning_rate": 9.452160458555194e-06, "loss": 0.2069, "step": 10640 }, { "epoch": 0.193710325760745, "grad_norm": 1.2538132667541504, "learning_rate": 9.450969565684252e-06, "loss": 0.2039, "step": 10650 }, { "epoch": 0.1938922133905673, "grad_norm": 1.2170571088790894, "learning_rate": 9.449777475207258e-06, "loss": 0.2036, "step": 10660 }, { "epoch": 0.1940741010203896, "grad_norm": 1.243306040763855, "learning_rate": 9.448584187537908e-06, "loss": 0.1995, "step": 10670 }, { "epoch": 0.1942559886502119, "grad_norm": 1.1705322265625, "learning_rate": 9.447389703090317e-06, "loss": 0.2083, "step": 10680 }, { "epoch": 0.1944378762800342, "grad_norm": 1.1804661750793457, "learning_rate": 9.446194022279008e-06, "loss": 0.2076, "step": 10690 }, { "epoch": 0.1946197639098565, "grad_norm": 1.0582330226898193, "learning_rate": 9.444997145518927e-06, "loss": 0.2077, "step": 10700 }, { "epoch": 0.1948016515396788, "grad_norm": 1.358140230178833, "learning_rate": 9.443799073225432e-06, "loss": 0.1976, "step": 10710 }, { "epoch": 0.1949835391695011, "grad_norm": 1.2068650722503662, "learning_rate": 9.442599805814295e-06, "loss": 0.2037, "step": 10720 }, { "epoch": 0.1951654267993234, "grad_norm": 1.2527823448181152, "learning_rate": 9.441399343701702e-06, "loss": 0.2037, "step": 10730 }, { "epoch": 0.19534731442914569, "grad_norm": 1.1305378675460815, "learning_rate": 9.440197687304258e-06, "loss": 0.2085, "step": 10740 }, { "epoch": 0.19552920205896798, "grad_norm": 1.1438076496124268, "learning_rate": 9.438994837038976e-06, "loss": 0.2081, "step": 10750 }, { "epoch": 0.19571108968879025, "grad_norm": 1.1795485019683838, "learning_rate": 9.437790793323288e-06, "loss": 0.2091, "step": 10760 }, { "epoch": 0.19589297731861255, "grad_norm": 1.198729157447815, "learning_rate": 9.436585556575039e-06, "loss": 0.2069, "step": 10770 }, { "epoch": 0.19607486494843485, "grad_norm": 1.3204835653305054, "learning_rate": 9.435379127212489e-06, "loss": 0.2015, "step": 10780 }, { "epoch": 0.19625675257825714, "grad_norm": 1.0835230350494385, "learning_rate": 9.434171505654308e-06, "loss": 0.2077, "step": 10790 }, { "epoch": 0.19643864020807944, "grad_norm": 1.1877741813659668, "learning_rate": 9.432962692319585e-06, "loss": 0.2087, "step": 10800 }, { "epoch": 0.19662052783790174, "grad_norm": 1.3718692064285278, "learning_rate": 9.431752687627822e-06, "loss": 0.2013, "step": 10810 }, { "epoch": 0.19680241546772403, "grad_norm": 1.1640417575836182, "learning_rate": 9.430541491998928e-06, "loss": 0.2012, "step": 10820 }, { "epoch": 0.19698430309754633, "grad_norm": 1.2216943502426147, "learning_rate": 9.42932910585323e-06, "loss": 0.21, "step": 10830 }, { "epoch": 0.19716619072736863, "grad_norm": 1.0980168581008911, "learning_rate": 9.428115529611472e-06, "loss": 0.2062, "step": 10840 }, { "epoch": 0.19734807835719093, "grad_norm": 1.2609854936599731, "learning_rate": 9.426900763694805e-06, "loss": 0.2042, "step": 10850 }, { "epoch": 0.19752996598701322, "grad_norm": 1.150866985321045, "learning_rate": 9.425684808524793e-06, "loss": 0.2135, "step": 10860 }, { "epoch": 0.19771185361683552, "grad_norm": 1.1331489086151123, "learning_rate": 9.424467664523418e-06, "loss": 0.2048, "step": 10870 }, { "epoch": 0.19789374124665782, "grad_norm": 1.3179152011871338, "learning_rate": 9.42324933211307e-06, "loss": 0.2058, "step": 10880 }, { "epoch": 0.19807562887648011, "grad_norm": 1.3322744369506836, "learning_rate": 9.42202981171655e-06, "loss": 0.2072, "step": 10890 }, { "epoch": 0.1982575165063024, "grad_norm": 1.1612200736999512, "learning_rate": 9.420809103757078e-06, "loss": 0.1998, "step": 10900 }, { "epoch": 0.1984394041361247, "grad_norm": 1.1739659309387207, "learning_rate": 9.419587208658279e-06, "loss": 0.2036, "step": 10910 }, { "epoch": 0.198621291765947, "grad_norm": 1.2882579565048218, "learning_rate": 9.418364126844195e-06, "loss": 0.2022, "step": 10920 }, { "epoch": 0.1988031793957693, "grad_norm": 1.4153274297714233, "learning_rate": 9.417139858739275e-06, "loss": 0.2029, "step": 10930 }, { "epoch": 0.1989850670255916, "grad_norm": 1.1509428024291992, "learning_rate": 9.415914404768384e-06, "loss": 0.2046, "step": 10940 }, { "epoch": 0.1991669546554139, "grad_norm": 1.0972530841827393, "learning_rate": 9.414687765356798e-06, "loss": 0.2042, "step": 10950 }, { "epoch": 0.1993488422852362, "grad_norm": 1.2044167518615723, "learning_rate": 9.413459940930204e-06, "loss": 0.2051, "step": 10960 }, { "epoch": 0.19953072991505846, "grad_norm": 1.1149928569793701, "learning_rate": 9.412230931914696e-06, "loss": 0.2082, "step": 10970 }, { "epoch": 0.19971261754488076, "grad_norm": 1.1632695198059082, "learning_rate": 9.411000738736787e-06, "loss": 0.2078, "step": 10980 }, { "epoch": 0.19989450517470306, "grad_norm": 1.191149115562439, "learning_rate": 9.409769361823392e-06, "loss": 0.2053, "step": 10990 }, { "epoch": 0.20007639280452535, "grad_norm": 1.3285565376281738, "learning_rate": 9.408536801601845e-06, "loss": 0.2101, "step": 11000 }, { "epoch": 0.20025828043434765, "grad_norm": 1.30349600315094, "learning_rate": 9.407303058499885e-06, "loss": 0.2038, "step": 11010 }, { "epoch": 0.20044016806416995, "grad_norm": 1.095078706741333, "learning_rate": 9.406068132945667e-06, "loss": 0.2067, "step": 11020 }, { "epoch": 0.20062205569399225, "grad_norm": 1.2634528875350952, "learning_rate": 9.40483202536775e-06, "loss": 0.2053, "step": 11030 }, { "epoch": 0.20080394332381454, "grad_norm": 1.2108304500579834, "learning_rate": 9.403594736195108e-06, "loss": 0.2068, "step": 11040 }, { "epoch": 0.20098583095363684, "grad_norm": 1.184955358505249, "learning_rate": 9.40235626585712e-06, "loss": 0.2046, "step": 11050 }, { "epoch": 0.20116771858345914, "grad_norm": 1.1924490928649902, "learning_rate": 9.40111661478358e-06, "loss": 0.2045, "step": 11060 }, { "epoch": 0.20134960621328143, "grad_norm": 1.356102466583252, "learning_rate": 9.399875783404692e-06, "loss": 0.2044, "step": 11070 }, { "epoch": 0.20153149384310373, "grad_norm": 1.1390835046768188, "learning_rate": 9.398633772151064e-06, "loss": 0.2055, "step": 11080 }, { "epoch": 0.20171338147292603, "grad_norm": 1.1432411670684814, "learning_rate": 9.397390581453722e-06, "loss": 0.2099, "step": 11090 }, { "epoch": 0.20189526910274833, "grad_norm": 1.212916612625122, "learning_rate": 9.396146211744089e-06, "loss": 0.2017, "step": 11100 }, { "epoch": 0.20207715673257062, "grad_norm": 1.2518407106399536, "learning_rate": 9.39490066345401e-06, "loss": 0.2057, "step": 11110 }, { "epoch": 0.20225904436239292, "grad_norm": 1.2591809034347534, "learning_rate": 9.393653937015732e-06, "loss": 0.2013, "step": 11120 }, { "epoch": 0.20244093199221522, "grad_norm": 1.1807211637496948, "learning_rate": 9.39240603286191e-06, "loss": 0.2011, "step": 11130 }, { "epoch": 0.20262281962203751, "grad_norm": 1.1085846424102783, "learning_rate": 9.391156951425613e-06, "loss": 0.2024, "step": 11140 }, { "epoch": 0.2028047072518598, "grad_norm": 1.053900957107544, "learning_rate": 9.389906693140316e-06, "loss": 0.2067, "step": 11150 }, { "epoch": 0.2029865948816821, "grad_norm": 1.2656404972076416, "learning_rate": 9.388655258439899e-06, "loss": 0.2048, "step": 11160 }, { "epoch": 0.2031684825115044, "grad_norm": 1.133520245552063, "learning_rate": 9.387402647758656e-06, "loss": 0.2081, "step": 11170 }, { "epoch": 0.20335037014132668, "grad_norm": 1.0993170738220215, "learning_rate": 9.386148861531283e-06, "loss": 0.2027, "step": 11180 }, { "epoch": 0.20353225777114897, "grad_norm": 1.305258870124817, "learning_rate": 9.384893900192892e-06, "loss": 0.2024, "step": 11190 }, { "epoch": 0.20371414540097127, "grad_norm": 1.093003273010254, "learning_rate": 9.383637764178994e-06, "loss": 0.2081, "step": 11200 }, { "epoch": 0.20389603303079357, "grad_norm": 1.1339225769042969, "learning_rate": 9.382380453925513e-06, "loss": 0.206, "step": 11210 }, { "epoch": 0.20407792066061586, "grad_norm": 1.1990915536880493, "learning_rate": 9.381121969868779e-06, "loss": 0.2075, "step": 11220 }, { "epoch": 0.20425980829043816, "grad_norm": 1.2026305198669434, "learning_rate": 9.37986231244553e-06, "loss": 0.2046, "step": 11230 }, { "epoch": 0.20444169592026046, "grad_norm": 1.120132565498352, "learning_rate": 9.378601482092911e-06, "loss": 0.2067, "step": 11240 }, { "epoch": 0.20462358355008275, "grad_norm": 1.2466145753860474, "learning_rate": 9.377339479248474e-06, "loss": 0.2074, "step": 11250 }, { "epoch": 0.20480547117990505, "grad_norm": 1.1422334909439087, "learning_rate": 9.376076304350177e-06, "loss": 0.2045, "step": 11260 }, { "epoch": 0.20498735880972735, "grad_norm": 1.1800636053085327, "learning_rate": 9.374811957836385e-06, "loss": 0.2044, "step": 11270 }, { "epoch": 0.20516924643954965, "grad_norm": 1.2476835250854492, "learning_rate": 9.373546440145873e-06, "loss": 0.2048, "step": 11280 }, { "epoch": 0.20535113406937194, "grad_norm": 1.1608178615570068, "learning_rate": 9.372279751717815e-06, "loss": 0.2008, "step": 11290 }, { "epoch": 0.20553302169919424, "grad_norm": 1.198868751525879, "learning_rate": 9.371011892991799e-06, "loss": 0.2076, "step": 11300 }, { "epoch": 0.20571490932901654, "grad_norm": 1.094629168510437, "learning_rate": 9.369742864407817e-06, "loss": 0.1995, "step": 11310 }, { "epoch": 0.20589679695883883, "grad_norm": 1.0940017700195312, "learning_rate": 9.368472666406261e-06, "loss": 0.2061, "step": 11320 }, { "epoch": 0.20607868458866113, "grad_norm": 1.2551742792129517, "learning_rate": 9.36720129942794e-06, "loss": 0.2041, "step": 11330 }, { "epoch": 0.20626057221848343, "grad_norm": 1.1798274517059326, "learning_rate": 9.365928763914056e-06, "loss": 0.2062, "step": 11340 }, { "epoch": 0.20644245984830573, "grad_norm": 1.2825660705566406, "learning_rate": 9.364655060306227e-06, "loss": 0.2054, "step": 11350 }, { "epoch": 0.20662434747812802, "grad_norm": 1.175791621208191, "learning_rate": 9.36338018904647e-06, "loss": 0.2107, "step": 11360 }, { "epoch": 0.20680623510795032, "grad_norm": 1.1204986572265625, "learning_rate": 9.362104150577214e-06, "loss": 0.2043, "step": 11370 }, { "epoch": 0.20698812273777262, "grad_norm": 1.1782041788101196, "learning_rate": 9.360826945341283e-06, "loss": 0.2071, "step": 11380 }, { "epoch": 0.2071700103675949, "grad_norm": 1.3417091369628906, "learning_rate": 9.359548573781916e-06, "loss": 0.2012, "step": 11390 }, { "epoch": 0.20735189799741718, "grad_norm": 1.2109251022338867, "learning_rate": 9.358269036342748e-06, "loss": 0.202, "step": 11400 }, { "epoch": 0.20753378562723948, "grad_norm": 1.1761672496795654, "learning_rate": 9.356988333467826e-06, "loss": 0.2067, "step": 11410 }, { "epoch": 0.20771567325706178, "grad_norm": 1.2743104696273804, "learning_rate": 9.355706465601598e-06, "loss": 0.2097, "step": 11420 }, { "epoch": 0.20789756088688408, "grad_norm": 1.176780343055725, "learning_rate": 9.354423433188913e-06, "loss": 0.211, "step": 11430 }, { "epoch": 0.20807944851670637, "grad_norm": 1.3458741903305054, "learning_rate": 9.353139236675033e-06, "loss": 0.2063, "step": 11440 }, { "epoch": 0.20826133614652867, "grad_norm": 1.2323410511016846, "learning_rate": 9.351853876505615e-06, "loss": 0.2046, "step": 11450 }, { "epoch": 0.20844322377635097, "grad_norm": 1.2499507665634155, "learning_rate": 9.350567353126728e-06, "loss": 0.1995, "step": 11460 }, { "epoch": 0.20862511140617326, "grad_norm": 1.2195287942886353, "learning_rate": 9.349279666984833e-06, "loss": 0.209, "step": 11470 }, { "epoch": 0.20880699903599556, "grad_norm": 1.1871291399002075, "learning_rate": 9.347990818526808e-06, "loss": 0.2044, "step": 11480 }, { "epoch": 0.20898888666581786, "grad_norm": 1.2574995756149292, "learning_rate": 9.346700808199924e-06, "loss": 0.2042, "step": 11490 }, { "epoch": 0.20917077429564016, "grad_norm": 1.159767508506775, "learning_rate": 9.345409636451863e-06, "loss": 0.2029, "step": 11500 }, { "epoch": 0.20935266192546245, "grad_norm": 1.0596212148666382, "learning_rate": 9.344117303730703e-06, "loss": 0.2026, "step": 11510 }, { "epoch": 0.20953454955528475, "grad_norm": 1.1866978406906128, "learning_rate": 9.342823810484932e-06, "loss": 0.2044, "step": 11520 }, { "epoch": 0.20971643718510705, "grad_norm": 1.084310531616211, "learning_rate": 9.341529157163433e-06, "loss": 0.2061, "step": 11530 }, { "epoch": 0.20989832481492934, "grad_norm": 1.1811504364013672, "learning_rate": 9.340233344215497e-06, "loss": 0.2052, "step": 11540 }, { "epoch": 0.21008021244475164, "grad_norm": 1.19733726978302, "learning_rate": 9.338936372090816e-06, "loss": 0.2155, "step": 11550 }, { "epoch": 0.21026210007457394, "grad_norm": 1.3213340044021606, "learning_rate": 9.337638241239485e-06, "loss": 0.2085, "step": 11560 }, { "epoch": 0.21044398770439623, "grad_norm": 1.1321098804473877, "learning_rate": 9.336338952112002e-06, "loss": 0.2056, "step": 11570 }, { "epoch": 0.21062587533421853, "grad_norm": 1.2217694520950317, "learning_rate": 9.33503850515926e-06, "loss": 0.2027, "step": 11580 }, { "epoch": 0.21080776296404083, "grad_norm": 1.1952358484268188, "learning_rate": 9.333736900832565e-06, "loss": 0.2081, "step": 11590 }, { "epoch": 0.2109896505938631, "grad_norm": 1.0710875988006592, "learning_rate": 9.332434139583617e-06, "loss": 0.2041, "step": 11600 }, { "epoch": 0.2111715382236854, "grad_norm": 1.3002084493637085, "learning_rate": 9.331130221864515e-06, "loss": 0.2077, "step": 11610 }, { "epoch": 0.2113534258535077, "grad_norm": 1.15580153465271, "learning_rate": 9.32982514812777e-06, "loss": 0.2057, "step": 11620 }, { "epoch": 0.21153531348333, "grad_norm": 1.1890710592269897, "learning_rate": 9.328518918826284e-06, "loss": 0.2078, "step": 11630 }, { "epoch": 0.2117172011131523, "grad_norm": 1.0847753286361694, "learning_rate": 9.327211534413366e-06, "loss": 0.206, "step": 11640 }, { "epoch": 0.21189908874297458, "grad_norm": 1.1828001737594604, "learning_rate": 9.325902995342721e-06, "loss": 0.2051, "step": 11650 }, { "epoch": 0.21208097637279688, "grad_norm": 1.3307528495788574, "learning_rate": 9.32459330206846e-06, "loss": 0.2077, "step": 11660 }, { "epoch": 0.21226286400261918, "grad_norm": 1.059983253479004, "learning_rate": 9.323282455045092e-06, "loss": 0.2068, "step": 11670 }, { "epoch": 0.21244475163244148, "grad_norm": 1.1660367250442505, "learning_rate": 9.321970454727525e-06, "loss": 0.2043, "step": 11680 }, { "epoch": 0.21262663926226377, "grad_norm": 1.2049612998962402, "learning_rate": 9.320657301571067e-06, "loss": 0.2065, "step": 11690 }, { "epoch": 0.21280852689208607, "grad_norm": 1.2131469249725342, "learning_rate": 9.31934299603143e-06, "loss": 0.2062, "step": 11700 }, { "epoch": 0.21299041452190837, "grad_norm": 1.1160151958465576, "learning_rate": 9.318027538564725e-06, "loss": 0.2035, "step": 11710 }, { "epoch": 0.21317230215173066, "grad_norm": 1.089788556098938, "learning_rate": 9.31671092962746e-06, "loss": 0.2066, "step": 11720 }, { "epoch": 0.21335418978155296, "grad_norm": 1.1193190813064575, "learning_rate": 9.315393169676542e-06, "loss": 0.2041, "step": 11730 }, { "epoch": 0.21353607741137526, "grad_norm": 1.2167837619781494, "learning_rate": 9.31407425916928e-06, "loss": 0.199, "step": 11740 }, { "epoch": 0.21371796504119756, "grad_norm": 1.1858867406845093, "learning_rate": 9.312754198563384e-06, "loss": 0.1971, "step": 11750 }, { "epoch": 0.21389985267101985, "grad_norm": 1.0834810733795166, "learning_rate": 9.311432988316958e-06, "loss": 0.2003, "step": 11760 }, { "epoch": 0.21408174030084215, "grad_norm": 1.2119176387786865, "learning_rate": 9.31011062888851e-06, "loss": 0.202, "step": 11770 }, { "epoch": 0.21426362793066445, "grad_norm": 1.0890644788742065, "learning_rate": 9.308787120736944e-06, "loss": 0.2049, "step": 11780 }, { "epoch": 0.21444551556048674, "grad_norm": 1.2156691551208496, "learning_rate": 9.307462464321562e-06, "loss": 0.1997, "step": 11790 }, { "epoch": 0.21462740319030904, "grad_norm": 1.1456180810928345, "learning_rate": 9.306136660102067e-06, "loss": 0.2096, "step": 11800 }, { "epoch": 0.2148092908201313, "grad_norm": 1.18339204788208, "learning_rate": 9.304809708538557e-06, "loss": 0.2034, "step": 11810 }, { "epoch": 0.2149911784499536, "grad_norm": 1.0955277681350708, "learning_rate": 9.303481610091535e-06, "loss": 0.2041, "step": 11820 }, { "epoch": 0.2151730660797759, "grad_norm": 1.1013888120651245, "learning_rate": 9.30215236522189e-06, "loss": 0.2044, "step": 11830 }, { "epoch": 0.2153549537095982, "grad_norm": 1.2851464748382568, "learning_rate": 9.30082197439092e-06, "loss": 0.2074, "step": 11840 }, { "epoch": 0.2155368413394205, "grad_norm": 1.1511218547821045, "learning_rate": 9.29949043806032e-06, "loss": 0.2049, "step": 11850 }, { "epoch": 0.2157187289692428, "grad_norm": 1.21390962600708, "learning_rate": 9.298157756692173e-06, "loss": 0.2012, "step": 11860 }, { "epoch": 0.2159006165990651, "grad_norm": 1.0348773002624512, "learning_rate": 9.296823930748971e-06, "loss": 0.2033, "step": 11870 }, { "epoch": 0.2160825042288874, "grad_norm": 1.16798996925354, "learning_rate": 9.295488960693593e-06, "loss": 0.2026, "step": 11880 }, { "epoch": 0.2162643918587097, "grad_norm": 1.2871118783950806, "learning_rate": 9.294152846989322e-06, "loss": 0.2047, "step": 11890 }, { "epoch": 0.21644627948853198, "grad_norm": 1.1435632705688477, "learning_rate": 9.292815590099837e-06, "loss": 0.2057, "step": 11900 }, { "epoch": 0.21662816711835428, "grad_norm": 1.2194441556930542, "learning_rate": 9.291477190489214e-06, "loss": 0.2039, "step": 11910 }, { "epoch": 0.21681005474817658, "grad_norm": 1.1162981986999512, "learning_rate": 9.290137648621917e-06, "loss": 0.2038, "step": 11920 }, { "epoch": 0.21699194237799888, "grad_norm": 1.460074543952942, "learning_rate": 9.288796964962821e-06, "loss": 0.2328, "step": 11930 }, { "epoch": 0.21717383000782117, "grad_norm": 1.2620781660079956, "learning_rate": 9.287455139977186e-06, "loss": 0.2411, "step": 11940 }, { "epoch": 0.21735571763764347, "grad_norm": 1.2630178928375244, "learning_rate": 9.286112174130672e-06, "loss": 0.2406, "step": 11950 }, { "epoch": 0.21753760526746577, "grad_norm": 1.1954129934310913, "learning_rate": 9.284768067889335e-06, "loss": 0.2433, "step": 11960 }, { "epoch": 0.21771949289728806, "grad_norm": 1.1874427795410156, "learning_rate": 9.283422821719627e-06, "loss": 0.2414, "step": 11970 }, { "epoch": 0.21790138052711036, "grad_norm": 1.2058665752410889, "learning_rate": 9.282076436088396e-06, "loss": 0.2435, "step": 11980 }, { "epoch": 0.21808326815693266, "grad_norm": 1.2170400619506836, "learning_rate": 9.280728911462882e-06, "loss": 0.2383, "step": 11990 }, { "epoch": 0.21826515578675496, "grad_norm": 1.3015910387039185, "learning_rate": 9.279380248310724e-06, "loss": 0.2401, "step": 12000 }, { "epoch": 0.21844704341657725, "grad_norm": 1.2647545337677002, "learning_rate": 9.278030447099955e-06, "loss": 0.2411, "step": 12010 }, { "epoch": 0.21862893104639952, "grad_norm": 1.2003717422485352, "learning_rate": 9.276679508299001e-06, "loss": 0.2439, "step": 12020 }, { "epoch": 0.21881081867622182, "grad_norm": 1.101434588432312, "learning_rate": 9.275327432376689e-06, "loss": 0.2419, "step": 12030 }, { "epoch": 0.21899270630604412, "grad_norm": 1.0256600379943848, "learning_rate": 9.273974219802231e-06, "loss": 0.2389, "step": 12040 }, { "epoch": 0.2191745939358664, "grad_norm": 1.6739869117736816, "learning_rate": 9.272619871045242e-06, "loss": 0.2424, "step": 12050 }, { "epoch": 0.2193564815656887, "grad_norm": 1.0115116834640503, "learning_rate": 9.271264386575727e-06, "loss": 0.2348, "step": 12060 }, { "epoch": 0.219538369195511, "grad_norm": 1.1915202140808105, "learning_rate": 9.269907766864086e-06, "loss": 0.2405, "step": 12070 }, { "epoch": 0.2197202568253333, "grad_norm": 1.1162763833999634, "learning_rate": 9.268550012381114e-06, "loss": 0.2434, "step": 12080 }, { "epoch": 0.2199021444551556, "grad_norm": 1.1061850786209106, "learning_rate": 9.267191123597994e-06, "loss": 0.2349, "step": 12090 }, { "epoch": 0.2200840320849779, "grad_norm": 1.2275229692459106, "learning_rate": 9.265831100986317e-06, "loss": 0.2458, "step": 12100 }, { "epoch": 0.2202659197148002, "grad_norm": 1.2055981159210205, "learning_rate": 9.26446994501805e-06, "loss": 0.2367, "step": 12110 }, { "epoch": 0.2204478073446225, "grad_norm": 1.1981253623962402, "learning_rate": 9.263107656165564e-06, "loss": 0.2341, "step": 12120 }, { "epoch": 0.2206296949744448, "grad_norm": 1.0565924644470215, "learning_rate": 9.26174423490162e-06, "loss": 0.2369, "step": 12130 }, { "epoch": 0.2208115826042671, "grad_norm": 1.1049587726593018, "learning_rate": 9.260379681699373e-06, "loss": 0.2462, "step": 12140 }, { "epoch": 0.22099347023408938, "grad_norm": 1.2207977771759033, "learning_rate": 9.25901399703237e-06, "loss": 0.2449, "step": 12150 }, { "epoch": 0.22117535786391168, "grad_norm": 1.2270461320877075, "learning_rate": 9.257647181374551e-06, "loss": 0.2444, "step": 12160 }, { "epoch": 0.22135724549373398, "grad_norm": 1.2713526487350464, "learning_rate": 9.256279235200247e-06, "loss": 0.2412, "step": 12170 }, { "epoch": 0.22153913312355628, "grad_norm": 1.2892661094665527, "learning_rate": 9.254910158984185e-06, "loss": 0.246, "step": 12180 }, { "epoch": 0.22172102075337857, "grad_norm": 1.0940911769866943, "learning_rate": 9.253539953201482e-06, "loss": 0.2438, "step": 12190 }, { "epoch": 0.22190290838320087, "grad_norm": 1.1787681579589844, "learning_rate": 9.252168618327645e-06, "loss": 0.2411, "step": 12200 }, { "epoch": 0.22208479601302317, "grad_norm": 1.196402907371521, "learning_rate": 9.250796154838573e-06, "loss": 0.2434, "step": 12210 }, { "epoch": 0.22226668364284544, "grad_norm": 1.0899019241333008, "learning_rate": 9.249422563210564e-06, "loss": 0.2456, "step": 12220 }, { "epoch": 0.22244857127266773, "grad_norm": 1.2337325811386108, "learning_rate": 9.248047843920298e-06, "loss": 0.2397, "step": 12230 }, { "epoch": 0.22263045890249003, "grad_norm": 1.1983920335769653, "learning_rate": 9.246671997444851e-06, "loss": 0.247, "step": 12240 }, { "epoch": 0.22281234653231233, "grad_norm": 1.2902324199676514, "learning_rate": 9.24529502426169e-06, "loss": 0.241, "step": 12250 }, { "epoch": 0.22299423416213462, "grad_norm": 1.3313353061676025, "learning_rate": 9.243916924848672e-06, "loss": 0.2398, "step": 12260 }, { "epoch": 0.22317612179195692, "grad_norm": 1.2714611291885376, "learning_rate": 9.242537699684046e-06, "loss": 0.2452, "step": 12270 }, { "epoch": 0.22335800942177922, "grad_norm": 1.221411943435669, "learning_rate": 9.241157349246448e-06, "loss": 0.2436, "step": 12280 }, { "epoch": 0.22353989705160152, "grad_norm": 1.1067193746566772, "learning_rate": 9.239775874014912e-06, "loss": 0.2394, "step": 12290 }, { "epoch": 0.2237217846814238, "grad_norm": 1.2789084911346436, "learning_rate": 9.238393274468852e-06, "loss": 0.2444, "step": 12300 }, { "epoch": 0.2239036723112461, "grad_norm": 1.1797438859939575, "learning_rate": 9.237009551088083e-06, "loss": 0.2376, "step": 12310 }, { "epoch": 0.2240855599410684, "grad_norm": 1.1164038181304932, "learning_rate": 9.235624704352803e-06, "loss": 0.2402, "step": 12320 }, { "epoch": 0.2242674475708907, "grad_norm": 1.1792681217193604, "learning_rate": 9.234238734743603e-06, "loss": 0.2445, "step": 12330 }, { "epoch": 0.224449335200713, "grad_norm": 1.1096701622009277, "learning_rate": 9.232851642741462e-06, "loss": 0.2431, "step": 12340 }, { "epoch": 0.2246312228305353, "grad_norm": 1.192185640335083, "learning_rate": 9.231463428827746e-06, "loss": 0.2428, "step": 12350 }, { "epoch": 0.2248131104603576, "grad_norm": 1.2700989246368408, "learning_rate": 9.230074093484218e-06, "loss": 0.2494, "step": 12360 }, { "epoch": 0.2249949980901799, "grad_norm": 1.2278188467025757, "learning_rate": 9.228683637193024e-06, "loss": 0.2404, "step": 12370 }, { "epoch": 0.2251768857200022, "grad_norm": 1.1227959394454956, "learning_rate": 9.2272920604367e-06, "loss": 0.2474, "step": 12380 }, { "epoch": 0.2253587733498245, "grad_norm": 1.1174941062927246, "learning_rate": 9.22589936369817e-06, "loss": 0.2377, "step": 12390 }, { "epoch": 0.22554066097964678, "grad_norm": 1.200476884841919, "learning_rate": 9.22450554746075e-06, "loss": 0.2471, "step": 12400 }, { "epoch": 0.22572254860946908, "grad_norm": 1.201672911643982, "learning_rate": 9.223110612208142e-06, "loss": 0.245, "step": 12410 }, { "epoch": 0.22590443623929138, "grad_norm": 1.220346450805664, "learning_rate": 9.221714558424438e-06, "loss": 0.2443, "step": 12420 }, { "epoch": 0.22608632386911365, "grad_norm": 1.1108195781707764, "learning_rate": 9.220317386594113e-06, "loss": 0.2366, "step": 12430 }, { "epoch": 0.22626821149893594, "grad_norm": 1.081030011177063, "learning_rate": 9.21891909720204e-06, "loss": 0.239, "step": 12440 }, { "epoch": 0.22645009912875824, "grad_norm": 1.1370497941970825, "learning_rate": 9.217519690733469e-06, "loss": 0.2405, "step": 12450 }, { "epoch": 0.22663198675858054, "grad_norm": 1.1630958318710327, "learning_rate": 9.216119167674045e-06, "loss": 0.2457, "step": 12460 }, { "epoch": 0.22681387438840284, "grad_norm": 1.1050676107406616, "learning_rate": 9.214717528509797e-06, "loss": 0.2414, "step": 12470 }, { "epoch": 0.22699576201822513, "grad_norm": 1.2531986236572266, "learning_rate": 9.213314773727142e-06, "loss": 0.2419, "step": 12480 }, { "epoch": 0.22717764964804743, "grad_norm": 1.2503808736801147, "learning_rate": 9.211910903812887e-06, "loss": 0.2375, "step": 12490 }, { "epoch": 0.22735953727786973, "grad_norm": 1.1670805215835571, "learning_rate": 9.210505919254222e-06, "loss": 0.244, "step": 12500 }, { "epoch": 0.22754142490769202, "grad_norm": 1.1941174268722534, "learning_rate": 9.209099820538723e-06, "loss": 0.2463, "step": 12510 }, { "epoch": 0.22772331253751432, "grad_norm": 1.1826839447021484, "learning_rate": 9.20769260815436e-06, "loss": 0.2382, "step": 12520 }, { "epoch": 0.22790520016733662, "grad_norm": 1.1053701639175415, "learning_rate": 9.20628428258948e-06, "loss": 0.2445, "step": 12530 }, { "epoch": 0.22808708779715892, "grad_norm": 1.1654002666473389, "learning_rate": 9.204874844332823e-06, "loss": 0.2423, "step": 12540 }, { "epoch": 0.2282689754269812, "grad_norm": 1.0843522548675537, "learning_rate": 9.203464293873511e-06, "loss": 0.2494, "step": 12550 }, { "epoch": 0.2284508630568035, "grad_norm": 1.2886278629302979, "learning_rate": 9.202052631701055e-06, "loss": 0.2454, "step": 12560 }, { "epoch": 0.2286327506866258, "grad_norm": 1.051065444946289, "learning_rate": 9.200639858305353e-06, "loss": 0.2403, "step": 12570 }, { "epoch": 0.2288146383164481, "grad_norm": 1.1590920686721802, "learning_rate": 9.199225974176684e-06, "loss": 0.2419, "step": 12580 }, { "epoch": 0.2289965259462704, "grad_norm": 1.2408875226974487, "learning_rate": 9.197810979805713e-06, "loss": 0.2435, "step": 12590 }, { "epoch": 0.2291784135760927, "grad_norm": 1.1131696701049805, "learning_rate": 9.196394875683497e-06, "loss": 0.2413, "step": 12600 }, { "epoch": 0.229360301205915, "grad_norm": 1.0361675024032593, "learning_rate": 9.194977662301468e-06, "loss": 0.2407, "step": 12610 }, { "epoch": 0.2295421888357373, "grad_norm": 1.1329774856567383, "learning_rate": 9.193559340151451e-06, "loss": 0.2453, "step": 12620 }, { "epoch": 0.2297240764655596, "grad_norm": 1.0985132455825806, "learning_rate": 9.192139909725653e-06, "loss": 0.2422, "step": 12630 }, { "epoch": 0.22990596409538186, "grad_norm": 1.1421674489974976, "learning_rate": 9.190719371516667e-06, "loss": 0.2458, "step": 12640 }, { "epoch": 0.23008785172520416, "grad_norm": 1.2174543142318726, "learning_rate": 9.189297726017464e-06, "loss": 0.2433, "step": 12650 }, { "epoch": 0.23026973935502645, "grad_norm": 1.0511232614517212, "learning_rate": 9.18787497372141e-06, "loss": 0.2397, "step": 12660 }, { "epoch": 0.23045162698484875, "grad_norm": 1.08985435962677, "learning_rate": 9.186451115122244e-06, "loss": 0.2368, "step": 12670 }, { "epoch": 0.23063351461467105, "grad_norm": 1.075791835784912, "learning_rate": 9.185026150714097e-06, "loss": 0.237, "step": 12680 }, { "epoch": 0.23081540224449335, "grad_norm": 1.1824145317077637, "learning_rate": 9.183600080991482e-06, "loss": 0.2349, "step": 12690 }, { "epoch": 0.23099728987431564, "grad_norm": 1.2430377006530762, "learning_rate": 9.182172906449295e-06, "loss": 0.2396, "step": 12700 }, { "epoch": 0.23117917750413794, "grad_norm": 1.203497052192688, "learning_rate": 9.180744627582811e-06, "loss": 0.2449, "step": 12710 }, { "epoch": 0.23136106513396024, "grad_norm": 1.2539360523223877, "learning_rate": 9.179315244887697e-06, "loss": 0.2406, "step": 12720 }, { "epoch": 0.23154295276378253, "grad_norm": 1.1778775453567505, "learning_rate": 9.177884758859996e-06, "loss": 0.2422, "step": 12730 }, { "epoch": 0.23172484039360483, "grad_norm": 1.1435452699661255, "learning_rate": 9.176453169996135e-06, "loss": 0.2433, "step": 12740 }, { "epoch": 0.23190672802342713, "grad_norm": 1.0696369409561157, "learning_rate": 9.175020478792928e-06, "loss": 0.2436, "step": 12750 }, { "epoch": 0.23208861565324942, "grad_norm": 1.0631952285766602, "learning_rate": 9.173586685747567e-06, "loss": 0.2429, "step": 12760 }, { "epoch": 0.23227050328307172, "grad_norm": 1.2382184267044067, "learning_rate": 9.172151791357627e-06, "loss": 0.2446, "step": 12770 }, { "epoch": 0.23245239091289402, "grad_norm": 1.080805778503418, "learning_rate": 9.170715796121068e-06, "loss": 0.2383, "step": 12780 }, { "epoch": 0.23263427854271632, "grad_norm": 1.1371625661849976, "learning_rate": 9.169278700536228e-06, "loss": 0.2417, "step": 12790 }, { "epoch": 0.2328161661725386, "grad_norm": 1.3004320859909058, "learning_rate": 9.16784050510183e-06, "loss": 0.2382, "step": 12800 }, { "epoch": 0.2329980538023609, "grad_norm": 1.2150214910507202, "learning_rate": 9.166401210316978e-06, "loss": 0.2423, "step": 12810 }, { "epoch": 0.2331799414321832, "grad_norm": 1.0909066200256348, "learning_rate": 9.164960816681158e-06, "loss": 0.2457, "step": 12820 }, { "epoch": 0.2333618290620055, "grad_norm": 1.247357726097107, "learning_rate": 9.163519324694234e-06, "loss": 0.2423, "step": 12830 }, { "epoch": 0.2335437166918278, "grad_norm": 1.1862260103225708, "learning_rate": 9.162076734856455e-06, "loss": 0.2433, "step": 12840 }, { "epoch": 0.23372560432165007, "grad_norm": 1.124001383781433, "learning_rate": 9.16063304766845e-06, "loss": 0.242, "step": 12850 }, { "epoch": 0.23390749195147237, "grad_norm": 1.2859461307525635, "learning_rate": 9.159188263631228e-06, "loss": 0.2413, "step": 12860 }, { "epoch": 0.23408937958129467, "grad_norm": 1.1422553062438965, "learning_rate": 9.15774238324618e-06, "loss": 0.2449, "step": 12870 }, { "epoch": 0.23427126721111696, "grad_norm": 1.0626020431518555, "learning_rate": 9.156295407015075e-06, "loss": 0.233, "step": 12880 }, { "epoch": 0.23445315484093926, "grad_norm": 1.2899481058120728, "learning_rate": 9.154847335440066e-06, "loss": 0.2371, "step": 12890 }, { "epoch": 0.23463504247076156, "grad_norm": 1.1767812967300415, "learning_rate": 9.153398169023682e-06, "loss": 0.2401, "step": 12900 }, { "epoch": 0.23481693010058385, "grad_norm": 1.2326481342315674, "learning_rate": 9.151947908268837e-06, "loss": 0.2358, "step": 12910 }, { "epoch": 0.23499881773040615, "grad_norm": 1.1820077896118164, "learning_rate": 9.150496553678818e-06, "loss": 0.2454, "step": 12920 }, { "epoch": 0.23518070536022845, "grad_norm": 1.163394808769226, "learning_rate": 9.149044105757295e-06, "loss": 0.2395, "step": 12930 }, { "epoch": 0.23536259299005075, "grad_norm": 1.124464511871338, "learning_rate": 9.147590565008323e-06, "loss": 0.2444, "step": 12940 }, { "epoch": 0.23554448061987304, "grad_norm": 1.225743055343628, "learning_rate": 9.146135931936326e-06, "loss": 0.2452, "step": 12950 }, { "epoch": 0.23572636824969534, "grad_norm": 1.1964303255081177, "learning_rate": 9.144680207046116e-06, "loss": 0.2437, "step": 12960 }, { "epoch": 0.23590825587951764, "grad_norm": 1.122210144996643, "learning_rate": 9.143223390842874e-06, "loss": 0.2374, "step": 12970 }, { "epoch": 0.23609014350933993, "grad_norm": 1.146546483039856, "learning_rate": 9.141765483832172e-06, "loss": 0.2436, "step": 12980 }, { "epoch": 0.23627203113916223, "grad_norm": 1.16197669506073, "learning_rate": 9.140306486519952e-06, "loss": 0.2426, "step": 12990 }, { "epoch": 0.23645391876898453, "grad_norm": 1.100369930267334, "learning_rate": 9.138846399412537e-06, "loss": 0.2441, "step": 13000 }, { "epoch": 0.23663580639880682, "grad_norm": 1.124489426612854, "learning_rate": 9.137385223016627e-06, "loss": 0.2319, "step": 13010 }, { "epoch": 0.23681769402862912, "grad_norm": 1.163371205329895, "learning_rate": 9.135922957839299e-06, "loss": 0.2377, "step": 13020 }, { "epoch": 0.23699958165845142, "grad_norm": 1.1572223901748657, "learning_rate": 9.134459604388015e-06, "loss": 0.2434, "step": 13030 }, { "epoch": 0.23718146928827372, "grad_norm": 1.1032226085662842, "learning_rate": 9.132995163170604e-06, "loss": 0.2499, "step": 13040 }, { "epoch": 0.237363356918096, "grad_norm": 1.226538062095642, "learning_rate": 9.13152963469528e-06, "loss": 0.2412, "step": 13050 }, { "epoch": 0.23754524454791828, "grad_norm": 1.2178583145141602, "learning_rate": 9.130063019470634e-06, "loss": 0.2395, "step": 13060 }, { "epoch": 0.23772713217774058, "grad_norm": 1.1614960432052612, "learning_rate": 9.12859531800563e-06, "loss": 0.2416, "step": 13070 }, { "epoch": 0.23790901980756288, "grad_norm": 1.1842694282531738, "learning_rate": 9.12712653080961e-06, "loss": 0.245, "step": 13080 }, { "epoch": 0.23809090743738517, "grad_norm": 1.0508524179458618, "learning_rate": 9.125656658392299e-06, "loss": 0.2367, "step": 13090 }, { "epoch": 0.23827279506720747, "grad_norm": 1.2328228950500488, "learning_rate": 9.124185701263788e-06, "loss": 0.2414, "step": 13100 }, { "epoch": 0.23845468269702977, "grad_norm": 1.2765612602233887, "learning_rate": 9.122713659934554e-06, "loss": 0.2431, "step": 13110 }, { "epoch": 0.23863657032685207, "grad_norm": 1.0772905349731445, "learning_rate": 9.121240534915446e-06, "loss": 0.2389, "step": 13120 }, { "epoch": 0.23881845795667436, "grad_norm": 1.2631515264511108, "learning_rate": 9.119766326717688e-06, "loss": 0.2418, "step": 13130 }, { "epoch": 0.23900034558649666, "grad_norm": 1.136946439743042, "learning_rate": 9.11829103585288e-06, "loss": 0.2414, "step": 13140 }, { "epoch": 0.23918223321631896, "grad_norm": 1.1588985919952393, "learning_rate": 9.116814662833002e-06, "loss": 0.2445, "step": 13150 }, { "epoch": 0.23936412084614125, "grad_norm": 1.1583210229873657, "learning_rate": 9.115337208170405e-06, "loss": 0.2394, "step": 13160 }, { "epoch": 0.23954600847596355, "grad_norm": 1.1314738988876343, "learning_rate": 9.11385867237782e-06, "loss": 0.2405, "step": 13170 }, { "epoch": 0.23972789610578585, "grad_norm": 1.4732067584991455, "learning_rate": 9.112379055968346e-06, "loss": 0.2495, "step": 13180 }, { "epoch": 0.23990978373560815, "grad_norm": 1.0485589504241943, "learning_rate": 9.110898359455462e-06, "loss": 0.243, "step": 13190 }, { "epoch": 0.24009167136543044, "grad_norm": 1.1345514059066772, "learning_rate": 9.109416583353024e-06, "loss": 0.2395, "step": 13200 }, { "epoch": 0.24027355899525274, "grad_norm": 1.1665010452270508, "learning_rate": 9.107933728175256e-06, "loss": 0.2426, "step": 13210 }, { "epoch": 0.24045544662507504, "grad_norm": 1.1744667291641235, "learning_rate": 9.106449794436762e-06, "loss": 0.24, "step": 13220 }, { "epoch": 0.24063733425489733, "grad_norm": 1.2042723894119263, "learning_rate": 9.104964782652517e-06, "loss": 0.2433, "step": 13230 }, { "epoch": 0.24081922188471963, "grad_norm": 1.155953049659729, "learning_rate": 9.103478693337874e-06, "loss": 0.2362, "step": 13240 }, { "epoch": 0.24100110951454193, "grad_norm": 1.1755268573760986, "learning_rate": 9.101991527008557e-06, "loss": 0.2394, "step": 13250 }, { "epoch": 0.24118299714436423, "grad_norm": 1.0863829851150513, "learning_rate": 9.100503284180662e-06, "loss": 0.2389, "step": 13260 }, { "epoch": 0.2413648847741865, "grad_norm": 1.130052924156189, "learning_rate": 9.099013965370663e-06, "loss": 0.2425, "step": 13270 }, { "epoch": 0.2415467724040088, "grad_norm": 1.176639199256897, "learning_rate": 9.097523571095404e-06, "loss": 0.2482, "step": 13280 }, { "epoch": 0.2417286600338311, "grad_norm": 1.1621344089508057, "learning_rate": 9.096032101872104e-06, "loss": 0.243, "step": 13290 }, { "epoch": 0.24191054766365339, "grad_norm": 1.2151074409484863, "learning_rate": 9.094539558218355e-06, "loss": 0.2375, "step": 13300 }, { "epoch": 0.24209243529347568, "grad_norm": 1.0690991878509521, "learning_rate": 9.093045940652122e-06, "loss": 0.2361, "step": 13310 }, { "epoch": 0.24227432292329798, "grad_norm": 1.0377861261367798, "learning_rate": 9.091551249691739e-06, "loss": 0.2403, "step": 13320 }, { "epoch": 0.24245621055312028, "grad_norm": 1.1706064939498901, "learning_rate": 9.090055485855918e-06, "loss": 0.2377, "step": 13330 }, { "epoch": 0.24263809818294257, "grad_norm": 1.263335108757019, "learning_rate": 9.088558649663738e-06, "loss": 0.2366, "step": 13340 }, { "epoch": 0.24281998581276487, "grad_norm": 1.2183676958084106, "learning_rate": 9.08706074163466e-06, "loss": 0.2415, "step": 13350 }, { "epoch": 0.24300187344258717, "grad_norm": 1.1738510131835938, "learning_rate": 9.085561762288504e-06, "loss": 0.2425, "step": 13360 }, { "epoch": 0.24318376107240947, "grad_norm": 1.314589262008667, "learning_rate": 9.08406171214547e-06, "loss": 0.244, "step": 13370 }, { "epoch": 0.24336564870223176, "grad_norm": 1.300558090209961, "learning_rate": 9.082560591726123e-06, "loss": 0.241, "step": 13380 }, { "epoch": 0.24354753633205406, "grad_norm": 1.023947834968567, "learning_rate": 9.081058401551412e-06, "loss": 0.2452, "step": 13390 }, { "epoch": 0.24372942396187636, "grad_norm": 1.2469351291656494, "learning_rate": 9.079555142142642e-06, "loss": 0.2423, "step": 13400 }, { "epoch": 0.24391131159169865, "grad_norm": 1.1761417388916016, "learning_rate": 9.0780508140215e-06, "loss": 0.2384, "step": 13410 }, { "epoch": 0.24409319922152095, "grad_norm": 1.2449365854263306, "learning_rate": 9.076545417710036e-06, "loss": 0.2458, "step": 13420 }, { "epoch": 0.24427508685134325, "grad_norm": 1.179292917251587, "learning_rate": 9.07503895373068e-06, "loss": 0.2459, "step": 13430 }, { "epoch": 0.24445697448116555, "grad_norm": 1.1899425983428955, "learning_rate": 9.073531422606223e-06, "loss": 0.2485, "step": 13440 }, { "epoch": 0.24463886211098784, "grad_norm": 1.040949821472168, "learning_rate": 9.072022824859833e-06, "loss": 0.2383, "step": 13450 }, { "epoch": 0.24482074974081014, "grad_norm": 1.156585454940796, "learning_rate": 9.070513161015045e-06, "loss": 0.2365, "step": 13460 }, { "epoch": 0.24500263737063244, "grad_norm": 1.3477994203567505, "learning_rate": 9.069002431595765e-06, "loss": 0.2414, "step": 13470 }, { "epoch": 0.2451845250004547, "grad_norm": 1.1098705530166626, "learning_rate": 9.067490637126265e-06, "loss": 0.2371, "step": 13480 }, { "epoch": 0.245366412630277, "grad_norm": 1.0374101400375366, "learning_rate": 9.065977778131198e-06, "loss": 0.2403, "step": 13490 }, { "epoch": 0.2455483002600993, "grad_norm": 1.0092461109161377, "learning_rate": 9.06446385513557e-06, "loss": 0.2428, "step": 13500 }, { "epoch": 0.2457301878899216, "grad_norm": 1.1828948259353638, "learning_rate": 9.06294886866477e-06, "loss": 0.2542, "step": 13510 }, { "epoch": 0.2459120755197439, "grad_norm": 1.1725293397903442, "learning_rate": 9.061432819244548e-06, "loss": 0.2379, "step": 13520 }, { "epoch": 0.2460939631495662, "grad_norm": 1.1779288053512573, "learning_rate": 9.059915707401027e-06, "loss": 0.2362, "step": 13530 }, { "epoch": 0.2462758507793885, "grad_norm": 1.178715467453003, "learning_rate": 9.058397533660697e-06, "loss": 0.2464, "step": 13540 }, { "epoch": 0.24645773840921079, "grad_norm": 1.2208960056304932, "learning_rate": 9.056878298550418e-06, "loss": 0.2408, "step": 13550 }, { "epoch": 0.24663962603903308, "grad_norm": 1.2236849069595337, "learning_rate": 9.055358002597415e-06, "loss": 0.2473, "step": 13560 }, { "epoch": 0.24682151366885538, "grad_norm": 1.0795109272003174, "learning_rate": 9.053836646329286e-06, "loss": 0.2409, "step": 13570 }, { "epoch": 0.24700340129867768, "grad_norm": 1.2244290113449097, "learning_rate": 9.052314230273994e-06, "loss": 0.2477, "step": 13580 }, { "epoch": 0.24718528892849997, "grad_norm": 1.2313214540481567, "learning_rate": 9.050790754959867e-06, "loss": 0.2469, "step": 13590 }, { "epoch": 0.24736717655832227, "grad_norm": 0.9916638731956482, "learning_rate": 9.049266220915607e-06, "loss": 0.2473, "step": 13600 }, { "epoch": 0.24754906418814457, "grad_norm": 1.2456462383270264, "learning_rate": 9.047740628670279e-06, "loss": 0.2414, "step": 13610 }, { "epoch": 0.24773095181796687, "grad_norm": 1.037807822227478, "learning_rate": 9.046213978753317e-06, "loss": 0.2478, "step": 13620 }, { "epoch": 0.24791283944778916, "grad_norm": 1.1258243322372437, "learning_rate": 9.04468627169452e-06, "loss": 0.2413, "step": 13630 }, { "epoch": 0.24809472707761146, "grad_norm": 1.0788524150848389, "learning_rate": 9.043157508024057e-06, "loss": 0.2451, "step": 13640 }, { "epoch": 0.24827661470743376, "grad_norm": 1.1068387031555176, "learning_rate": 9.041627688272461e-06, "loss": 0.238, "step": 13650 }, { "epoch": 0.24845850233725605, "grad_norm": 1.144702672958374, "learning_rate": 9.040096812970633e-06, "loss": 0.2371, "step": 13660 }, { "epoch": 0.24864038996707835, "grad_norm": 1.0332012176513672, "learning_rate": 9.038564882649839e-06, "loss": 0.2436, "step": 13670 }, { "epoch": 0.24882227759690065, "grad_norm": 1.1377009153366089, "learning_rate": 9.037031897841714e-06, "loss": 0.2399, "step": 13680 }, { "epoch": 0.24900416522672292, "grad_norm": 1.151105284690857, "learning_rate": 9.035497859078254e-06, "loss": 0.2459, "step": 13690 }, { "epoch": 0.24918605285654521, "grad_norm": 1.0756385326385498, "learning_rate": 9.033962766891824e-06, "loss": 0.2451, "step": 13700 }, { "epoch": 0.2493679404863675, "grad_norm": 0.980371356010437, "learning_rate": 9.032426621815158e-06, "loss": 0.2409, "step": 13710 }, { "epoch": 0.2495498281161898, "grad_norm": 1.187261939048767, "learning_rate": 9.030889424381347e-06, "loss": 0.2474, "step": 13720 }, { "epoch": 0.2497317157460121, "grad_norm": 1.1111297607421875, "learning_rate": 9.029351175123853e-06, "loss": 0.2446, "step": 13730 }, { "epoch": 0.2499136033758344, "grad_norm": 1.107922077178955, "learning_rate": 9.027811874576502e-06, "loss": 0.2408, "step": 13740 }, { "epoch": 0.2500954910056567, "grad_norm": 1.0424144268035889, "learning_rate": 9.026271523273486e-06, "loss": 0.2355, "step": 13750 }, { "epoch": 0.250277378635479, "grad_norm": 1.1174860000610352, "learning_rate": 9.024730121749359e-06, "loss": 0.2402, "step": 13760 }, { "epoch": 0.2504592662653013, "grad_norm": 1.1899250745773315, "learning_rate": 9.02318767053904e-06, "loss": 0.2394, "step": 13770 }, { "epoch": 0.2506411538951236, "grad_norm": 1.1053589582443237, "learning_rate": 9.021644170177815e-06, "loss": 0.2389, "step": 13780 }, { "epoch": 0.2508230415249459, "grad_norm": 1.1784085035324097, "learning_rate": 9.020099621201332e-06, "loss": 0.2481, "step": 13790 }, { "epoch": 0.2510049291547682, "grad_norm": 1.049025297164917, "learning_rate": 9.0185540241456e-06, "loss": 0.2401, "step": 13800 }, { "epoch": 0.2511868167845905, "grad_norm": 1.0463790893554688, "learning_rate": 9.017007379547e-06, "loss": 0.2407, "step": 13810 }, { "epoch": 0.2513687044144128, "grad_norm": 0.9761174321174622, "learning_rate": 9.015459687942265e-06, "loss": 0.2419, "step": 13820 }, { "epoch": 0.2515505920442351, "grad_norm": 1.0824782848358154, "learning_rate": 9.013910949868503e-06, "loss": 0.2427, "step": 13830 }, { "epoch": 0.2517324796740574, "grad_norm": 1.1755932569503784, "learning_rate": 9.012361165863176e-06, "loss": 0.238, "step": 13840 }, { "epoch": 0.25191436730387967, "grad_norm": 1.0864869356155396, "learning_rate": 9.010810336464117e-06, "loss": 0.2368, "step": 13850 }, { "epoch": 0.25209625493370197, "grad_norm": 1.10878324508667, "learning_rate": 9.009258462209513e-06, "loss": 0.2406, "step": 13860 }, { "epoch": 0.25227814256352427, "grad_norm": 1.2308197021484375, "learning_rate": 9.00770554363792e-06, "loss": 0.2433, "step": 13870 }, { "epoch": 0.25246003019334656, "grad_norm": 1.044987678527832, "learning_rate": 9.006151581288255e-06, "loss": 0.2358, "step": 13880 }, { "epoch": 0.25264191782316886, "grad_norm": 1.2495203018188477, "learning_rate": 9.004596575699796e-06, "loss": 0.2436, "step": 13890 }, { "epoch": 0.25282380545299116, "grad_norm": 1.1755080223083496, "learning_rate": 9.003040527412186e-06, "loss": 0.2403, "step": 13900 }, { "epoch": 0.25300569308281345, "grad_norm": 1.2137928009033203, "learning_rate": 9.001483436965424e-06, "loss": 0.2319, "step": 13910 }, { "epoch": 0.25318758071263575, "grad_norm": 1.2298438549041748, "learning_rate": 8.999925304899879e-06, "loss": 0.2441, "step": 13920 }, { "epoch": 0.25336946834245805, "grad_norm": 1.1958192586898804, "learning_rate": 8.99836613175627e-06, "loss": 0.2448, "step": 13930 }, { "epoch": 0.25355135597228035, "grad_norm": 1.039143681526184, "learning_rate": 8.996805918075691e-06, "loss": 0.2357, "step": 13940 }, { "epoch": 0.25373324360210264, "grad_norm": 1.1602506637573242, "learning_rate": 8.995244664399588e-06, "loss": 0.2446, "step": 13950 }, { "epoch": 0.25391513123192494, "grad_norm": 1.1293082237243652, "learning_rate": 8.99368237126977e-06, "loss": 0.238, "step": 13960 }, { "epoch": 0.25409701886174724, "grad_norm": 1.197698712348938, "learning_rate": 8.992119039228405e-06, "loss": 0.2385, "step": 13970 }, { "epoch": 0.25427890649156953, "grad_norm": 1.1265259981155396, "learning_rate": 8.990554668818026e-06, "loss": 0.2429, "step": 13980 }, { "epoch": 0.25446079412139183, "grad_norm": 1.1568533182144165, "learning_rate": 8.98898926058152e-06, "loss": 0.2447, "step": 13990 }, { "epoch": 0.2546426817512141, "grad_norm": 1.2859561443328857, "learning_rate": 8.987422815062144e-06, "loss": 0.2424, "step": 14000 }, { "epoch": 0.25482456938103637, "grad_norm": 1.1802407503128052, "learning_rate": 8.985855332803504e-06, "loss": 0.2422, "step": 14010 }, { "epoch": 0.25500645701085867, "grad_norm": 1.01063072681427, "learning_rate": 8.984286814349573e-06, "loss": 0.2399, "step": 14020 }, { "epoch": 0.25518834464068096, "grad_norm": 1.1263175010681152, "learning_rate": 8.982717260244679e-06, "loss": 0.2436, "step": 14030 }, { "epoch": 0.25537023227050326, "grad_norm": 1.0849275588989258, "learning_rate": 8.981146671033514e-06, "loss": 0.2395, "step": 14040 }, { "epoch": 0.25555211990032556, "grad_norm": 1.1802668571472168, "learning_rate": 8.979575047261126e-06, "loss": 0.2431, "step": 14050 }, { "epoch": 0.25573400753014786, "grad_norm": 1.0504835844039917, "learning_rate": 8.978002389472923e-06, "loss": 0.2393, "step": 14060 }, { "epoch": 0.25591589515997015, "grad_norm": 1.0576434135437012, "learning_rate": 8.976428698214674e-06, "loss": 0.2423, "step": 14070 }, { "epoch": 0.25609778278979245, "grad_norm": 1.4598921537399292, "learning_rate": 8.974853974032498e-06, "loss": 0.2455, "step": 14080 }, { "epoch": 0.25627967041961475, "grad_norm": 1.0698788166046143, "learning_rate": 8.973278217472884e-06, "loss": 0.2379, "step": 14090 }, { "epoch": 0.25646155804943704, "grad_norm": 1.1171948909759521, "learning_rate": 8.971701429082675e-06, "loss": 0.2476, "step": 14100 }, { "epoch": 0.25664344567925934, "grad_norm": 1.1050981283187866, "learning_rate": 8.970123609409067e-06, "loss": 0.2397, "step": 14110 }, { "epoch": 0.25682533330908164, "grad_norm": 1.208266258239746, "learning_rate": 8.968544758999621e-06, "loss": 0.244, "step": 14120 }, { "epoch": 0.25700722093890394, "grad_norm": 1.158897042274475, "learning_rate": 8.966964878402254e-06, "loss": 0.2413, "step": 14130 }, { "epoch": 0.25718910856872623, "grad_norm": 1.0780500173568726, "learning_rate": 8.965383968165236e-06, "loss": 0.2339, "step": 14140 }, { "epoch": 0.25737099619854853, "grad_norm": 1.184244155883789, "learning_rate": 8.9638020288372e-06, "loss": 0.2441, "step": 14150 }, { "epoch": 0.2575528838283708, "grad_norm": 1.3023953437805176, "learning_rate": 8.962219060967133e-06, "loss": 0.2401, "step": 14160 }, { "epoch": 0.2577347714581931, "grad_norm": 1.0806920528411865, "learning_rate": 8.960635065104381e-06, "loss": 0.2408, "step": 14170 }, { "epoch": 0.2579166590880154, "grad_norm": 1.2880221605300903, "learning_rate": 8.959050041798642e-06, "loss": 0.2455, "step": 14180 }, { "epoch": 0.2580985467178377, "grad_norm": 1.1644631624221802, "learning_rate": 8.957463991599978e-06, "loss": 0.2406, "step": 14190 }, { "epoch": 0.25828043434766, "grad_norm": 1.102165699005127, "learning_rate": 8.955876915058801e-06, "loss": 0.2436, "step": 14200 }, { "epoch": 0.2584623219774823, "grad_norm": 1.1319584846496582, "learning_rate": 8.954288812725882e-06, "loss": 0.2425, "step": 14210 }, { "epoch": 0.2586442096073046, "grad_norm": 1.1124358177185059, "learning_rate": 8.952699685152348e-06, "loss": 0.2482, "step": 14220 }, { "epoch": 0.2588260972371269, "grad_norm": 1.0982496738433838, "learning_rate": 8.95110953288968e-06, "loss": 0.2376, "step": 14230 }, { "epoch": 0.2590079848669492, "grad_norm": 1.1643916368484497, "learning_rate": 8.949518356489719e-06, "loss": 0.2368, "step": 14240 }, { "epoch": 0.2591898724967715, "grad_norm": 1.0621930360794067, "learning_rate": 8.947926156504654e-06, "loss": 0.2484, "step": 14250 }, { "epoch": 0.2593717601265938, "grad_norm": 1.228169322013855, "learning_rate": 8.946332933487037e-06, "loss": 0.2484, "step": 14260 }, { "epoch": 0.2595536477564161, "grad_norm": 1.10387122631073, "learning_rate": 8.944738687989772e-06, "loss": 0.2389, "step": 14270 }, { "epoch": 0.2597355353862384, "grad_norm": 1.0716426372528076, "learning_rate": 8.943143420566116e-06, "loss": 0.2445, "step": 14280 }, { "epoch": 0.2599174230160607, "grad_norm": 1.1580923795700073, "learning_rate": 8.94154713176968e-06, "loss": 0.2433, "step": 14290 }, { "epoch": 0.260099310645883, "grad_norm": 1.185591459274292, "learning_rate": 8.939949822154434e-06, "loss": 0.2454, "step": 14300 }, { "epoch": 0.2602811982757053, "grad_norm": 1.1153088808059692, "learning_rate": 8.9383514922747e-06, "loss": 0.2392, "step": 14310 }, { "epoch": 0.2604630859055276, "grad_norm": 1.1453018188476562, "learning_rate": 8.936752142685156e-06, "loss": 0.2448, "step": 14320 }, { "epoch": 0.2606449735353499, "grad_norm": 1.194618582725525, "learning_rate": 8.935151773940826e-06, "loss": 0.2377, "step": 14330 }, { "epoch": 0.2608268611651722, "grad_norm": 0.9727566838264465, "learning_rate": 8.933550386597096e-06, "loss": 0.2426, "step": 14340 }, { "epoch": 0.26100874879499447, "grad_norm": 1.0964109897613525, "learning_rate": 8.931947981209705e-06, "loss": 0.2358, "step": 14350 }, { "epoch": 0.26119063642481677, "grad_norm": 1.0387821197509766, "learning_rate": 8.930344558334742e-06, "loss": 0.2357, "step": 14360 }, { "epoch": 0.26137252405463907, "grad_norm": 1.1148838996887207, "learning_rate": 8.92874011852865e-06, "loss": 0.24, "step": 14370 }, { "epoch": 0.26155441168446136, "grad_norm": 1.0496677160263062, "learning_rate": 8.927134662348224e-06, "loss": 0.2399, "step": 14380 }, { "epoch": 0.26173629931428366, "grad_norm": 1.0001507997512817, "learning_rate": 8.925528190350614e-06, "loss": 0.239, "step": 14390 }, { "epoch": 0.26191818694410596, "grad_norm": 1.0297473669052124, "learning_rate": 8.923920703093322e-06, "loss": 0.2389, "step": 14400 }, { "epoch": 0.26210007457392825, "grad_norm": 1.1497948169708252, "learning_rate": 8.922312201134203e-06, "loss": 0.2443, "step": 14410 }, { "epoch": 0.2622819622037505, "grad_norm": 1.154736876487732, "learning_rate": 8.92070268503146e-06, "loss": 0.2444, "step": 14420 }, { "epoch": 0.2624638498335728, "grad_norm": 1.1596887111663818, "learning_rate": 8.919092155343651e-06, "loss": 0.2388, "step": 14430 }, { "epoch": 0.2626457374633951, "grad_norm": 1.1537272930145264, "learning_rate": 8.917480612629687e-06, "loss": 0.2386, "step": 14440 }, { "epoch": 0.2628276250932174, "grad_norm": 1.1386953592300415, "learning_rate": 8.91586805744883e-06, "loss": 0.2443, "step": 14450 }, { "epoch": 0.2630095127230397, "grad_norm": 1.155553936958313, "learning_rate": 8.914254490360693e-06, "loss": 0.2453, "step": 14460 }, { "epoch": 0.263191400352862, "grad_norm": 1.2796547412872314, "learning_rate": 8.912639911925236e-06, "loss": 0.2477, "step": 14470 }, { "epoch": 0.2633732879826843, "grad_norm": 1.0379668474197388, "learning_rate": 8.911024322702777e-06, "loss": 0.2454, "step": 14480 }, { "epoch": 0.2635551756125066, "grad_norm": 1.142238974571228, "learning_rate": 8.90940772325398e-06, "loss": 0.244, "step": 14490 }, { "epoch": 0.2637370632423289, "grad_norm": 1.2147444486618042, "learning_rate": 8.907790114139862e-06, "loss": 0.2397, "step": 14500 }, { "epoch": 0.26391895087215117, "grad_norm": 1.0354416370391846, "learning_rate": 8.906171495921787e-06, "loss": 0.2389, "step": 14510 }, { "epoch": 0.26410083850197347, "grad_norm": 1.2040220499038696, "learning_rate": 8.904551869161473e-06, "loss": 0.2487, "step": 14520 }, { "epoch": 0.26428272613179576, "grad_norm": 1.2007484436035156, "learning_rate": 8.90293123442099e-06, "loss": 0.2422, "step": 14530 }, { "epoch": 0.26446461376161806, "grad_norm": 1.1634647846221924, "learning_rate": 8.90130959226275e-06, "loss": 0.2387, "step": 14540 }, { "epoch": 0.26464650139144036, "grad_norm": 1.310475468635559, "learning_rate": 8.899686943249521e-06, "loss": 0.2412, "step": 14550 }, { "epoch": 0.26482838902126266, "grad_norm": 1.061648964881897, "learning_rate": 8.898063287944418e-06, "loss": 0.2418, "step": 14560 }, { "epoch": 0.26501027665108495, "grad_norm": 1.1068246364593506, "learning_rate": 8.896438626910905e-06, "loss": 0.2496, "step": 14570 }, { "epoch": 0.26519216428090725, "grad_norm": 1.2746261358261108, "learning_rate": 8.894812960712798e-06, "loss": 0.2386, "step": 14580 }, { "epoch": 0.26537405191072955, "grad_norm": 1.0714538097381592, "learning_rate": 8.893186289914255e-06, "loss": 0.2428, "step": 14590 }, { "epoch": 0.26555593954055184, "grad_norm": 1.1074604988098145, "learning_rate": 8.891558615079791e-06, "loss": 0.2409, "step": 14600 }, { "epoch": 0.26573782717037414, "grad_norm": 1.1057993173599243, "learning_rate": 8.889929936774266e-06, "loss": 0.243, "step": 14610 }, { "epoch": 0.26591971480019644, "grad_norm": 1.0478131771087646, "learning_rate": 8.888300255562885e-06, "loss": 0.2435, "step": 14620 }, { "epoch": 0.26610160243001874, "grad_norm": 1.0382249355316162, "learning_rate": 8.886669572011208e-06, "loss": 0.2368, "step": 14630 }, { "epoch": 0.26628349005984103, "grad_norm": 1.1711512804031372, "learning_rate": 8.885037886685135e-06, "loss": 0.25, "step": 14640 }, { "epoch": 0.26646537768966333, "grad_norm": 1.1412230730056763, "learning_rate": 8.883405200150916e-06, "loss": 0.2407, "step": 14650 }, { "epoch": 0.2666472653194856, "grad_norm": 1.2240146398544312, "learning_rate": 8.881771512975153e-06, "loss": 0.2443, "step": 14660 }, { "epoch": 0.2668291529493079, "grad_norm": 1.1004799604415894, "learning_rate": 8.880136825724792e-06, "loss": 0.2348, "step": 14670 }, { "epoch": 0.2670110405791302, "grad_norm": 1.0732883214950562, "learning_rate": 8.878501138967126e-06, "loss": 0.2376, "step": 14680 }, { "epoch": 0.2671929282089525, "grad_norm": 1.1211254596710205, "learning_rate": 8.876864453269793e-06, "loss": 0.2452, "step": 14690 }, { "epoch": 0.2673748158387748, "grad_norm": 1.0365872383117676, "learning_rate": 8.875226769200783e-06, "loss": 0.2355, "step": 14700 }, { "epoch": 0.2675567034685971, "grad_norm": 1.1627471446990967, "learning_rate": 8.873588087328424e-06, "loss": 0.2393, "step": 14710 }, { "epoch": 0.2677385910984194, "grad_norm": 1.1162115335464478, "learning_rate": 8.871948408221402e-06, "loss": 0.2367, "step": 14720 }, { "epoch": 0.2679204787282417, "grad_norm": 1.2035032510757446, "learning_rate": 8.870307732448737e-06, "loss": 0.2384, "step": 14730 }, { "epoch": 0.268102366358064, "grad_norm": 1.1075810194015503, "learning_rate": 8.868666060579802e-06, "loss": 0.2395, "step": 14740 }, { "epoch": 0.2682842539878863, "grad_norm": 1.2566192150115967, "learning_rate": 8.867023393184315e-06, "loss": 0.243, "step": 14750 }, { "epoch": 0.2684661416177086, "grad_norm": 1.1481984853744507, "learning_rate": 8.86537973083234e-06, "loss": 0.2407, "step": 14760 }, { "epoch": 0.2686480292475309, "grad_norm": 1.1143574714660645, "learning_rate": 8.863735074094282e-06, "loss": 0.2417, "step": 14770 }, { "epoch": 0.2688299168773532, "grad_norm": 1.1777325868606567, "learning_rate": 8.862089423540893e-06, "loss": 0.2389, "step": 14780 }, { "epoch": 0.2690118045071755, "grad_norm": 1.1655361652374268, "learning_rate": 8.860442779743273e-06, "loss": 0.2416, "step": 14790 }, { "epoch": 0.2691936921369978, "grad_norm": 1.1094576120376587, "learning_rate": 8.858795143272865e-06, "loss": 0.2407, "step": 14800 }, { "epoch": 0.2693755797668201, "grad_norm": 1.1106535196304321, "learning_rate": 8.857146514701454e-06, "loss": 0.2466, "step": 14810 }, { "epoch": 0.2695574673966424, "grad_norm": 1.1875728368759155, "learning_rate": 8.855496894601173e-06, "loss": 0.2434, "step": 14820 }, { "epoch": 0.2697393550264647, "grad_norm": 1.1291401386260986, "learning_rate": 8.853846283544497e-06, "loss": 0.2384, "step": 14830 }, { "epoch": 0.2699212426562869, "grad_norm": 1.135932207107544, "learning_rate": 8.852194682104243e-06, "loss": 0.2414, "step": 14840 }, { "epoch": 0.2701031302861092, "grad_norm": 1.1127851009368896, "learning_rate": 8.850542090853576e-06, "loss": 0.2454, "step": 14850 }, { "epoch": 0.2702850179159315, "grad_norm": 1.2291078567504883, "learning_rate": 8.848888510366002e-06, "loss": 0.2425, "step": 14860 }, { "epoch": 0.2704669055457538, "grad_norm": 1.1629512310028076, "learning_rate": 8.847233941215373e-06, "loss": 0.2402, "step": 14870 }, { "epoch": 0.2706487931755761, "grad_norm": 1.2133839130401611, "learning_rate": 8.84557838397588e-06, "loss": 0.2443, "step": 14880 }, { "epoch": 0.2708306808053984, "grad_norm": 1.113277554512024, "learning_rate": 8.843921839222054e-06, "loss": 0.2404, "step": 14890 }, { "epoch": 0.2710125684352207, "grad_norm": 1.0636012554168701, "learning_rate": 8.842264307528781e-06, "loss": 0.2419, "step": 14900 }, { "epoch": 0.271194456065043, "grad_norm": 1.2140467166900635, "learning_rate": 8.84060578947128e-06, "loss": 0.239, "step": 14910 }, { "epoch": 0.2713763436948653, "grad_norm": 1.1045196056365967, "learning_rate": 8.838946285625112e-06, "loss": 0.2374, "step": 14920 }, { "epoch": 0.2715582313246876, "grad_norm": 1.076012134552002, "learning_rate": 8.837285796566184e-06, "loss": 0.2416, "step": 14930 }, { "epoch": 0.2717401189545099, "grad_norm": 1.0800269842147827, "learning_rate": 8.835624322870741e-06, "loss": 0.24, "step": 14940 }, { "epoch": 0.2719220065843322, "grad_norm": 1.1103801727294922, "learning_rate": 8.833961865115375e-06, "loss": 0.2459, "step": 14950 }, { "epoch": 0.2721038942141545, "grad_norm": 0.9823939800262451, "learning_rate": 8.832298423877016e-06, "loss": 0.2427, "step": 14960 }, { "epoch": 0.2722857818439768, "grad_norm": 1.1267204284667969, "learning_rate": 8.830633999732933e-06, "loss": 0.2459, "step": 14970 }, { "epoch": 0.2724676694737991, "grad_norm": 1.1427183151245117, "learning_rate": 8.828968593260743e-06, "loss": 0.2423, "step": 14980 }, { "epoch": 0.2726495571036214, "grad_norm": 1.0654957294464111, "learning_rate": 8.827302205038396e-06, "loss": 0.2392, "step": 14990 }, { "epoch": 0.2728314447334437, "grad_norm": 1.1139858961105347, "learning_rate": 8.825634835644188e-06, "loss": 0.2459, "step": 15000 }, { "epoch": 0.27301333236326597, "grad_norm": 1.1800897121429443, "learning_rate": 8.823966485656755e-06, "loss": 0.2421, "step": 15010 }, { "epoch": 0.27319521999308827, "grad_norm": 1.105136513710022, "learning_rate": 8.82229715565507e-06, "loss": 0.2435, "step": 15020 }, { "epoch": 0.27337710762291056, "grad_norm": 1.0124306678771973, "learning_rate": 8.820626846218452e-06, "loss": 0.2412, "step": 15030 }, { "epoch": 0.27355899525273286, "grad_norm": 1.0876699686050415, "learning_rate": 8.81895555792655e-06, "loss": 0.24, "step": 15040 }, { "epoch": 0.27374088288255516, "grad_norm": 1.1966737508773804, "learning_rate": 8.817283291359367e-06, "loss": 0.2382, "step": 15050 }, { "epoch": 0.27392277051237746, "grad_norm": 1.1016789674758911, "learning_rate": 8.815610047097232e-06, "loss": 0.2429, "step": 15060 }, { "epoch": 0.27410465814219975, "grad_norm": 1.0865273475646973, "learning_rate": 8.813935825720819e-06, "loss": 0.2454, "step": 15070 }, { "epoch": 0.27428654577202205, "grad_norm": 0.9996333122253418, "learning_rate": 8.812260627811142e-06, "loss": 0.2375, "step": 15080 }, { "epoch": 0.27446843340184435, "grad_norm": 1.0459433794021606, "learning_rate": 8.810584453949554e-06, "loss": 0.2418, "step": 15090 }, { "epoch": 0.27465032103166664, "grad_norm": 1.1739091873168945, "learning_rate": 8.808907304717743e-06, "loss": 0.2366, "step": 15100 }, { "epoch": 0.27483220866148894, "grad_norm": 1.1851654052734375, "learning_rate": 8.807229180697739e-06, "loss": 0.2378, "step": 15110 }, { "epoch": 0.27501409629131124, "grad_norm": 1.2904473543167114, "learning_rate": 8.80555008247191e-06, "loss": 0.2408, "step": 15120 }, { "epoch": 0.27519598392113354, "grad_norm": 1.1105366945266724, "learning_rate": 8.80387001062296e-06, "loss": 0.2454, "step": 15130 }, { "epoch": 0.27537787155095583, "grad_norm": 1.1040501594543457, "learning_rate": 8.802188965733933e-06, "loss": 0.2472, "step": 15140 }, { "epoch": 0.27555975918077813, "grad_norm": 1.132219672203064, "learning_rate": 8.80050694838821e-06, "loss": 0.2435, "step": 15150 }, { "epoch": 0.2757416468106004, "grad_norm": 1.2334219217300415, "learning_rate": 8.798823959169509e-06, "loss": 0.2419, "step": 15160 }, { "epoch": 0.2759235344404227, "grad_norm": 1.0602556467056274, "learning_rate": 8.797139998661885e-06, "loss": 0.2394, "step": 15170 }, { "epoch": 0.276105422070245, "grad_norm": 1.0042457580566406, "learning_rate": 8.795455067449731e-06, "loss": 0.2418, "step": 15180 }, { "epoch": 0.2762873097000673, "grad_norm": 1.3375178575515747, "learning_rate": 8.793769166117776e-06, "loss": 0.2494, "step": 15190 }, { "epoch": 0.2764691973298896, "grad_norm": 1.2540217638015747, "learning_rate": 8.792082295251088e-06, "loss": 0.2373, "step": 15200 }, { "epoch": 0.2766510849597119, "grad_norm": 1.0794984102249146, "learning_rate": 8.790394455435067e-06, "loss": 0.239, "step": 15210 }, { "epoch": 0.2768329725895342, "grad_norm": 1.1795397996902466, "learning_rate": 8.788705647255454e-06, "loss": 0.2427, "step": 15220 }, { "epoch": 0.2770148602193565, "grad_norm": 1.3286646604537964, "learning_rate": 8.787015871298324e-06, "loss": 0.2414, "step": 15230 }, { "epoch": 0.2771967478491788, "grad_norm": 1.2409565448760986, "learning_rate": 8.785325128150087e-06, "loss": 0.2425, "step": 15240 }, { "epoch": 0.27737863547900105, "grad_norm": 1.119489312171936, "learning_rate": 8.78363341839749e-06, "loss": 0.2362, "step": 15250 }, { "epoch": 0.27756052310882334, "grad_norm": 1.0998989343643188, "learning_rate": 8.781940742627613e-06, "loss": 0.2424, "step": 15260 }, { "epoch": 0.27774241073864564, "grad_norm": 1.1200876235961914, "learning_rate": 8.780247101427875e-06, "loss": 0.2455, "step": 15270 }, { "epoch": 0.27792429836846794, "grad_norm": 1.1072152853012085, "learning_rate": 8.77855249538603e-06, "loss": 0.2455, "step": 15280 }, { "epoch": 0.27810618599829023, "grad_norm": 1.1198558807373047, "learning_rate": 8.776856925090163e-06, "loss": 0.2388, "step": 15290 }, { "epoch": 0.27828807362811253, "grad_norm": 1.305768370628357, "learning_rate": 8.775160391128693e-06, "loss": 0.2448, "step": 15300 }, { "epoch": 0.27846996125793483, "grad_norm": 1.352221965789795, "learning_rate": 8.77346289409038e-06, "loss": 0.2427, "step": 15310 }, { "epoch": 0.2786518488877571, "grad_norm": 1.255691409111023, "learning_rate": 8.771764434564314e-06, "loss": 0.24, "step": 15320 }, { "epoch": 0.2788337365175794, "grad_norm": 1.087154507637024, "learning_rate": 8.770065013139917e-06, "loss": 0.2367, "step": 15330 }, { "epoch": 0.2790156241474017, "grad_norm": 1.1064964532852173, "learning_rate": 8.76836463040695e-06, "loss": 0.2459, "step": 15340 }, { "epoch": 0.279197511777224, "grad_norm": 1.1023540496826172, "learning_rate": 8.766663286955502e-06, "loss": 0.2429, "step": 15350 }, { "epoch": 0.2793793994070463, "grad_norm": 1.163469672203064, "learning_rate": 8.764960983376001e-06, "loss": 0.24, "step": 15360 }, { "epoch": 0.2795612870368686, "grad_norm": 1.1526663303375244, "learning_rate": 8.763257720259202e-06, "loss": 0.2436, "step": 15370 }, { "epoch": 0.2797431746666909, "grad_norm": 1.0679776668548584, "learning_rate": 8.761553498196198e-06, "loss": 0.2422, "step": 15380 }, { "epoch": 0.2799250622965132, "grad_norm": 1.0693891048431396, "learning_rate": 8.759848317778415e-06, "loss": 0.2358, "step": 15390 }, { "epoch": 0.2801069499263355, "grad_norm": 1.0759114027023315, "learning_rate": 8.758142179597607e-06, "loss": 0.2412, "step": 15400 }, { "epoch": 0.2802888375561578, "grad_norm": 1.0329474210739136, "learning_rate": 8.756435084245862e-06, "loss": 0.2412, "step": 15410 }, { "epoch": 0.2804707251859801, "grad_norm": 1.1175578832626343, "learning_rate": 8.754727032315606e-06, "loss": 0.241, "step": 15420 }, { "epoch": 0.2806526128158024, "grad_norm": 1.1074411869049072, "learning_rate": 8.753018024399587e-06, "loss": 0.2353, "step": 15430 }, { "epoch": 0.2808345004456247, "grad_norm": 1.1529345512390137, "learning_rate": 8.751308061090895e-06, "loss": 0.2449, "step": 15440 }, { "epoch": 0.281016388075447, "grad_norm": 0.9964255690574646, "learning_rate": 8.749597142982943e-06, "loss": 0.2366, "step": 15450 }, { "epoch": 0.2811982757052693, "grad_norm": 1.109969973564148, "learning_rate": 8.74788527066948e-06, "loss": 0.2435, "step": 15460 }, { "epoch": 0.2813801633350916, "grad_norm": 1.2078320980072021, "learning_rate": 8.746172444744586e-06, "loss": 0.2419, "step": 15470 }, { "epoch": 0.2815620509649139, "grad_norm": 0.9976600408554077, "learning_rate": 8.744458665802668e-06, "loss": 0.2401, "step": 15480 }, { "epoch": 0.2817439385947362, "grad_norm": 1.198743224143982, "learning_rate": 8.74274393443847e-06, "loss": 0.2413, "step": 15490 }, { "epoch": 0.2819258262245585, "grad_norm": 1.068706750869751, "learning_rate": 8.741028251247058e-06, "loss": 0.2413, "step": 15500 }, { "epoch": 0.28210771385438077, "grad_norm": 1.0055851936340332, "learning_rate": 8.73931161682384e-06, "loss": 0.2336, "step": 15510 }, { "epoch": 0.28228960148420307, "grad_norm": 1.1045547723770142, "learning_rate": 8.737594031764546e-06, "loss": 0.2444, "step": 15520 }, { "epoch": 0.28247148911402536, "grad_norm": 1.0840308666229248, "learning_rate": 8.735875496665237e-06, "loss": 0.2433, "step": 15530 }, { "epoch": 0.28265337674384766, "grad_norm": 1.165518879890442, "learning_rate": 8.734156012122302e-06, "loss": 0.245, "step": 15540 }, { "epoch": 0.28283526437366996, "grad_norm": 1.2632437944412231, "learning_rate": 8.732435578732465e-06, "loss": 0.2428, "step": 15550 }, { "epoch": 0.28301715200349226, "grad_norm": 1.126296043395996, "learning_rate": 8.730714197092774e-06, "loss": 0.2399, "step": 15560 }, { "epoch": 0.28319903963331455, "grad_norm": 1.1645640134811401, "learning_rate": 8.728991867800609e-06, "loss": 0.2466, "step": 15570 }, { "epoch": 0.28338092726313685, "grad_norm": 1.1628543138504028, "learning_rate": 8.727268591453675e-06, "loss": 0.24, "step": 15580 }, { "epoch": 0.28356281489295915, "grad_norm": 1.1003553867340088, "learning_rate": 8.725544368650015e-06, "loss": 0.242, "step": 15590 }, { "epoch": 0.28374470252278144, "grad_norm": 1.240368366241455, "learning_rate": 8.723819199987987e-06, "loss": 0.2381, "step": 15600 }, { "epoch": 0.28392659015260374, "grad_norm": 1.1107569932937622, "learning_rate": 8.722093086066288e-06, "loss": 0.2415, "step": 15610 }, { "epoch": 0.28410847778242604, "grad_norm": 0.9578258991241455, "learning_rate": 8.72036602748394e-06, "loss": 0.238, "step": 15620 }, { "epoch": 0.28429036541224834, "grad_norm": 1.0927367210388184, "learning_rate": 8.718638024840292e-06, "loss": 0.2373, "step": 15630 }, { "epoch": 0.28447225304207063, "grad_norm": 1.1413776874542236, "learning_rate": 8.716909078735017e-06, "loss": 0.2405, "step": 15640 }, { "epoch": 0.28465414067189293, "grad_norm": 1.2270649671554565, "learning_rate": 8.715179189768123e-06, "loss": 0.2386, "step": 15650 }, { "epoch": 0.2848360283017152, "grad_norm": 1.100249171257019, "learning_rate": 8.713448358539943e-06, "loss": 0.2435, "step": 15660 }, { "epoch": 0.28501791593153747, "grad_norm": 1.23246431350708, "learning_rate": 8.711716585651131e-06, "loss": 0.2443, "step": 15670 }, { "epoch": 0.28519980356135977, "grad_norm": 1.1247358322143555, "learning_rate": 8.709983871702675e-06, "loss": 0.2426, "step": 15680 }, { "epoch": 0.28538169119118206, "grad_norm": 1.2272391319274902, "learning_rate": 8.708250217295885e-06, "loss": 0.2395, "step": 15690 }, { "epoch": 0.28556357882100436, "grad_norm": 1.0277881622314453, "learning_rate": 8.7065156230324e-06, "loss": 0.2441, "step": 15700 }, { "epoch": 0.28574546645082666, "grad_norm": 1.1356745958328247, "learning_rate": 8.704780089514186e-06, "loss": 0.2456, "step": 15710 }, { "epoch": 0.28592735408064895, "grad_norm": 1.1764129400253296, "learning_rate": 8.703043617343533e-06, "loss": 0.2437, "step": 15720 }, { "epoch": 0.28610924171047125, "grad_norm": 1.2471063137054443, "learning_rate": 8.701306207123055e-06, "loss": 0.2411, "step": 15730 }, { "epoch": 0.28629112934029355, "grad_norm": 1.4003702402114868, "learning_rate": 8.699567859455692e-06, "loss": 0.2397, "step": 15740 }, { "epoch": 0.28647301697011585, "grad_norm": 1.0183783769607544, "learning_rate": 8.697828574944716e-06, "loss": 0.238, "step": 15750 }, { "epoch": 0.28665490459993814, "grad_norm": 1.0335110425949097, "learning_rate": 8.696088354193715e-06, "loss": 0.2353, "step": 15760 }, { "epoch": 0.28683679222976044, "grad_norm": 1.0987763404846191, "learning_rate": 8.694347197806607e-06, "loss": 0.244, "step": 15770 }, { "epoch": 0.28701867985958274, "grad_norm": 1.0315415859222412, "learning_rate": 8.692605106387634e-06, "loss": 0.2341, "step": 15780 }, { "epoch": 0.28720056748940503, "grad_norm": 1.0726174116134644, "learning_rate": 8.690862080541362e-06, "loss": 0.2367, "step": 15790 }, { "epoch": 0.28738245511922733, "grad_norm": 1.10829758644104, "learning_rate": 8.689118120872681e-06, "loss": 0.2473, "step": 15800 }, { "epoch": 0.28756434274904963, "grad_norm": 1.2717026472091675, "learning_rate": 8.687373227986807e-06, "loss": 0.2474, "step": 15810 }, { "epoch": 0.2877462303788719, "grad_norm": 1.161615252494812, "learning_rate": 8.685627402489276e-06, "loss": 0.2444, "step": 15820 }, { "epoch": 0.2879281180086942, "grad_norm": 1.1066628694534302, "learning_rate": 8.683880644985952e-06, "loss": 0.2443, "step": 15830 }, { "epoch": 0.2881100056385165, "grad_norm": 1.0989508628845215, "learning_rate": 8.682132956083019e-06, "loss": 0.243, "step": 15840 }, { "epoch": 0.2882918932683388, "grad_norm": 1.1146517992019653, "learning_rate": 8.680384336386987e-06, "loss": 0.2374, "step": 15850 }, { "epoch": 0.2884737808981611, "grad_norm": 1.1666936874389648, "learning_rate": 8.678634786504684e-06, "loss": 0.2351, "step": 15860 }, { "epoch": 0.2886556685279834, "grad_norm": 1.3485037088394165, "learning_rate": 8.676884307043271e-06, "loss": 0.2384, "step": 15870 }, { "epoch": 0.2888375561578057, "grad_norm": 1.1017310619354248, "learning_rate": 8.675132898610218e-06, "loss": 0.2392, "step": 15880 }, { "epoch": 0.289019443787628, "grad_norm": 1.1430022716522217, "learning_rate": 8.67338056181333e-06, "loss": 0.2408, "step": 15890 }, { "epoch": 0.2892013314174503, "grad_norm": 1.1213291883468628, "learning_rate": 8.671627297260725e-06, "loss": 0.2402, "step": 15900 }, { "epoch": 0.2893832190472726, "grad_norm": 1.1519625186920166, "learning_rate": 8.669873105560848e-06, "loss": 0.2444, "step": 15910 }, { "epoch": 0.2895651066770949, "grad_norm": 1.0541523694992065, "learning_rate": 8.668117987322463e-06, "loss": 0.2426, "step": 15920 }, { "epoch": 0.2897469943069172, "grad_norm": 1.0848534107208252, "learning_rate": 8.66636194315466e-06, "loss": 0.2373, "step": 15930 }, { "epoch": 0.2899288819367395, "grad_norm": 1.1315960884094238, "learning_rate": 8.664604973666845e-06, "loss": 0.2359, "step": 15940 }, { "epoch": 0.2901107695665618, "grad_norm": 1.0731089115142822, "learning_rate": 8.662847079468746e-06, "loss": 0.2461, "step": 15950 }, { "epoch": 0.2902926571963841, "grad_norm": 1.150952696800232, "learning_rate": 8.661088261170415e-06, "loss": 0.2405, "step": 15960 }, { "epoch": 0.2904745448262064, "grad_norm": 1.15884530544281, "learning_rate": 8.659328519382225e-06, "loss": 0.242, "step": 15970 }, { "epoch": 0.2906564324560287, "grad_norm": 1.1065579652786255, "learning_rate": 8.657567854714861e-06, "loss": 0.2437, "step": 15980 }, { "epoch": 0.290838320085851, "grad_norm": 1.071104645729065, "learning_rate": 8.655806267779342e-06, "loss": 0.242, "step": 15990 }, { "epoch": 0.2910202077156733, "grad_norm": 1.08843195438385, "learning_rate": 8.654043759186994e-06, "loss": 0.2413, "step": 16000 }, { "epoch": 0.29120209534549557, "grad_norm": 1.2877010107040405, "learning_rate": 8.652280329549475e-06, "loss": 0.2442, "step": 16010 }, { "epoch": 0.29138398297531787, "grad_norm": 1.1674139499664307, "learning_rate": 8.650515979478748e-06, "loss": 0.2418, "step": 16020 }, { "epoch": 0.29156587060514016, "grad_norm": 1.0175658464431763, "learning_rate": 8.648750709587109e-06, "loss": 0.2342, "step": 16030 }, { "epoch": 0.29174775823496246, "grad_norm": 1.1748112440109253, "learning_rate": 8.646984520487167e-06, "loss": 0.2454, "step": 16040 }, { "epoch": 0.29192964586478476, "grad_norm": 1.0489765405654907, "learning_rate": 8.645217412791852e-06, "loss": 0.241, "step": 16050 }, { "epoch": 0.29211153349460706, "grad_norm": 1.1443538665771484, "learning_rate": 8.643449387114408e-06, "loss": 0.2444, "step": 16060 }, { "epoch": 0.29229342112442935, "grad_norm": 1.0178362131118774, "learning_rate": 8.641680444068403e-06, "loss": 0.2454, "step": 16070 }, { "epoch": 0.29247530875425165, "grad_norm": 0.9894161224365234, "learning_rate": 8.639910584267724e-06, "loss": 0.2359, "step": 16080 }, { "epoch": 0.2926571963840739, "grad_norm": 1.0867708921432495, "learning_rate": 8.63813980832657e-06, "loss": 0.2367, "step": 16090 }, { "epoch": 0.2928390840138962, "grad_norm": 1.0345256328582764, "learning_rate": 8.636368116859462e-06, "loss": 0.2442, "step": 16100 }, { "epoch": 0.2930209716437185, "grad_norm": 1.2687456607818604, "learning_rate": 8.634595510481241e-06, "loss": 0.2452, "step": 16110 }, { "epoch": 0.2932028592735408, "grad_norm": 1.1113970279693604, "learning_rate": 8.632821989807063e-06, "loss": 0.2402, "step": 16120 }, { "epoch": 0.2933847469033631, "grad_norm": 1.1916862726211548, "learning_rate": 8.631047555452397e-06, "loss": 0.234, "step": 16130 }, { "epoch": 0.2935666345331854, "grad_norm": 1.0145052671432495, "learning_rate": 8.629272208033037e-06, "loss": 0.2386, "step": 16140 }, { "epoch": 0.2937485221630077, "grad_norm": 1.030868411064148, "learning_rate": 8.62749594816509e-06, "loss": 0.2425, "step": 16150 }, { "epoch": 0.29393040979282997, "grad_norm": 1.2650423049926758, "learning_rate": 8.625718776464976e-06, "loss": 0.2388, "step": 16160 }, { "epoch": 0.29411229742265227, "grad_norm": 1.2143903970718384, "learning_rate": 8.623940693549439e-06, "loss": 0.2478, "step": 16170 }, { "epoch": 0.29429418505247457, "grad_norm": 1.0746158361434937, "learning_rate": 8.622161700035535e-06, "loss": 0.2444, "step": 16180 }, { "epoch": 0.29447607268229686, "grad_norm": 1.0873991250991821, "learning_rate": 8.620381796540633e-06, "loss": 0.2453, "step": 16190 }, { "epoch": 0.29465796031211916, "grad_norm": 1.1526209115982056, "learning_rate": 8.618600983682425e-06, "loss": 0.2429, "step": 16200 }, { "epoch": 0.29483984794194146, "grad_norm": 1.0449796915054321, "learning_rate": 8.616819262078915e-06, "loss": 0.2385, "step": 16210 }, { "epoch": 0.29502173557176375, "grad_norm": 1.1222143173217773, "learning_rate": 8.61503663234842e-06, "loss": 0.2444, "step": 16220 }, { "epoch": 0.29520362320158605, "grad_norm": 1.1389591693878174, "learning_rate": 8.613253095109575e-06, "loss": 0.2433, "step": 16230 }, { "epoch": 0.29538551083140835, "grad_norm": 1.1018446683883667, "learning_rate": 8.611468650981327e-06, "loss": 0.2396, "step": 16240 }, { "epoch": 0.29556739846123065, "grad_norm": 1.1072074174880981, "learning_rate": 8.609683300582945e-06, "loss": 0.241, "step": 16250 }, { "epoch": 0.29574928609105294, "grad_norm": 1.1531603336334229, "learning_rate": 8.607897044534004e-06, "loss": 0.2436, "step": 16260 }, { "epoch": 0.29593117372087524, "grad_norm": 1.3103587627410889, "learning_rate": 8.606109883454397e-06, "loss": 0.2451, "step": 16270 }, { "epoch": 0.29611306135069754, "grad_norm": 1.1782392263412476, "learning_rate": 8.604321817964332e-06, "loss": 0.234, "step": 16280 }, { "epoch": 0.29629494898051983, "grad_norm": 1.0910162925720215, "learning_rate": 8.602532848684328e-06, "loss": 0.2421, "step": 16290 }, { "epoch": 0.29647683661034213, "grad_norm": 1.2660026550292969, "learning_rate": 8.600742976235222e-06, "loss": 0.2427, "step": 16300 }, { "epoch": 0.29665872424016443, "grad_norm": 1.0393041372299194, "learning_rate": 8.598952201238158e-06, "loss": 0.2455, "step": 16310 }, { "epoch": 0.2968406118699867, "grad_norm": 1.1929399967193604, "learning_rate": 8.5971605243146e-06, "loss": 0.2457, "step": 16320 }, { "epoch": 0.297022499499809, "grad_norm": 1.20790696144104, "learning_rate": 8.59536794608632e-06, "loss": 0.2403, "step": 16330 }, { "epoch": 0.2972043871296313, "grad_norm": 1.032019019126892, "learning_rate": 8.593574467175407e-06, "loss": 0.2386, "step": 16340 }, { "epoch": 0.2973862747594536, "grad_norm": 1.1964908838272095, "learning_rate": 8.59178008820426e-06, "loss": 0.2412, "step": 16350 }, { "epoch": 0.2975681623892759, "grad_norm": 1.0060573816299438, "learning_rate": 8.589984809795589e-06, "loss": 0.2426, "step": 16360 }, { "epoch": 0.2977500500190982, "grad_norm": 1.172892689704895, "learning_rate": 8.588188632572419e-06, "loss": 0.245, "step": 16370 }, { "epoch": 0.2979319376489205, "grad_norm": 1.071576714515686, "learning_rate": 8.586391557158086e-06, "loss": 0.2423, "step": 16380 }, { "epoch": 0.2981138252787428, "grad_norm": 1.2694872617721558, "learning_rate": 8.584593584176237e-06, "loss": 0.246, "step": 16390 }, { "epoch": 0.2982957129085651, "grad_norm": 0.9609643816947937, "learning_rate": 8.58279471425083e-06, "loss": 0.2381, "step": 16400 }, { "epoch": 0.2984776005383874, "grad_norm": 1.0623087882995605, "learning_rate": 8.580994948006135e-06, "loss": 0.2395, "step": 16410 }, { "epoch": 0.2986594881682097, "grad_norm": 1.0239620208740234, "learning_rate": 8.579194286066736e-06, "loss": 0.2377, "step": 16420 }, { "epoch": 0.298841375798032, "grad_norm": 1.1970970630645752, "learning_rate": 8.577392729057524e-06, "loss": 0.2402, "step": 16430 }, { "epoch": 0.2990232634278543, "grad_norm": 1.0640192031860352, "learning_rate": 8.5755902776037e-06, "loss": 0.2469, "step": 16440 }, { "epoch": 0.2992051510576766, "grad_norm": 1.0501600503921509, "learning_rate": 8.573786932330779e-06, "loss": 0.2368, "step": 16450 }, { "epoch": 0.2993870386874989, "grad_norm": 1.0291799306869507, "learning_rate": 8.571982693864582e-06, "loss": 0.2433, "step": 16460 }, { "epoch": 0.2995689263173212, "grad_norm": 1.1454262733459473, "learning_rate": 8.570177562831247e-06, "loss": 0.2415, "step": 16470 }, { "epoch": 0.2997508139471435, "grad_norm": 1.0815348625183105, "learning_rate": 8.568371539857211e-06, "loss": 0.2403, "step": 16480 }, { "epoch": 0.2999327015769658, "grad_norm": 1.0741140842437744, "learning_rate": 8.566564625569233e-06, "loss": 0.2388, "step": 16490 }, { "epoch": 0.3001145892067881, "grad_norm": 1.040873408317566, "learning_rate": 8.564756820594368e-06, "loss": 0.2416, "step": 16500 }, { "epoch": 0.3002964768366103, "grad_norm": 1.1864813566207886, "learning_rate": 8.562948125559992e-06, "loss": 0.2423, "step": 16510 }, { "epoch": 0.3004783644664326, "grad_norm": 1.1669349670410156, "learning_rate": 8.561138541093784e-06, "loss": 0.241, "step": 16520 }, { "epoch": 0.3006602520962549, "grad_norm": 1.0917984247207642, "learning_rate": 8.55932806782373e-06, "loss": 0.2451, "step": 16530 }, { "epoch": 0.3008421397260772, "grad_norm": 1.0629491806030273, "learning_rate": 8.55751670637813e-06, "loss": 0.2401, "step": 16540 }, { "epoch": 0.3010240273558995, "grad_norm": 1.1166703701019287, "learning_rate": 8.555704457385587e-06, "loss": 0.2408, "step": 16550 }, { "epoch": 0.3012059149857218, "grad_norm": 1.0851277112960815, "learning_rate": 8.553891321475016e-06, "loss": 0.2392, "step": 16560 }, { "epoch": 0.3013878026155441, "grad_norm": 1.0560604333877563, "learning_rate": 8.552077299275637e-06, "loss": 0.2414, "step": 16570 }, { "epoch": 0.3015696902453664, "grad_norm": 1.2302266359329224, "learning_rate": 8.550262391416979e-06, "loss": 0.2454, "step": 16580 }, { "epoch": 0.3017515778751887, "grad_norm": 1.4571740627288818, "learning_rate": 8.548446598528876e-06, "loss": 0.2491, "step": 16590 }, { "epoch": 0.301933465505011, "grad_norm": 1.0516165494918823, "learning_rate": 8.546629921241476e-06, "loss": 0.238, "step": 16600 }, { "epoch": 0.3021153531348333, "grad_norm": 1.2873600721359253, "learning_rate": 8.544812360185222e-06, "loss": 0.2358, "step": 16610 }, { "epoch": 0.3022972407646556, "grad_norm": 1.1078827381134033, "learning_rate": 8.542993915990879e-06, "loss": 0.2404, "step": 16620 }, { "epoch": 0.3024791283944779, "grad_norm": 1.1033910512924194, "learning_rate": 8.541174589289503e-06, "loss": 0.2475, "step": 16630 }, { "epoch": 0.3026610160243002, "grad_norm": 1.1003849506378174, "learning_rate": 8.539354380712466e-06, "loss": 0.2425, "step": 16640 }, { "epoch": 0.3028429036541225, "grad_norm": 1.1457328796386719, "learning_rate": 8.537533290891444e-06, "loss": 0.2434, "step": 16650 }, { "epoch": 0.30302479128394477, "grad_norm": 1.2128576040267944, "learning_rate": 8.535711320458417e-06, "loss": 0.2462, "step": 16660 }, { "epoch": 0.30320667891376707, "grad_norm": 1.0543389320373535, "learning_rate": 8.533888470045674e-06, "loss": 0.2363, "step": 16670 }, { "epoch": 0.30338856654358937, "grad_norm": 1.225722312927246, "learning_rate": 8.532064740285805e-06, "loss": 0.2427, "step": 16680 }, { "epoch": 0.30357045417341166, "grad_norm": 1.0980792045593262, "learning_rate": 8.53024013181171e-06, "loss": 0.2439, "step": 16690 }, { "epoch": 0.30375234180323396, "grad_norm": 1.0883729457855225, "learning_rate": 8.528414645256587e-06, "loss": 0.2368, "step": 16700 }, { "epoch": 0.30393422943305626, "grad_norm": 1.1322470903396606, "learning_rate": 8.526588281253947e-06, "loss": 0.2371, "step": 16710 }, { "epoch": 0.30411611706287855, "grad_norm": 0.9917386770248413, "learning_rate": 8.524761040437602e-06, "loss": 0.2361, "step": 16720 }, { "epoch": 0.30429800469270085, "grad_norm": 1.118303894996643, "learning_rate": 8.522932923441665e-06, "loss": 0.2413, "step": 16730 }, { "epoch": 0.30447989232252315, "grad_norm": 1.1649808883666992, "learning_rate": 8.521103930900558e-06, "loss": 0.2432, "step": 16740 }, { "epoch": 0.30466177995234545, "grad_norm": 1.051479697227478, "learning_rate": 8.519274063449002e-06, "loss": 0.2426, "step": 16750 }, { "epoch": 0.30484366758216774, "grad_norm": 0.9805173873901367, "learning_rate": 8.51744332172203e-06, "loss": 0.2373, "step": 16760 }, { "epoch": 0.30502555521199004, "grad_norm": 1.113160252571106, "learning_rate": 8.515611706354968e-06, "loss": 0.2475, "step": 16770 }, { "epoch": 0.30520744284181234, "grad_norm": 1.0846376419067383, "learning_rate": 8.513779217983452e-06, "loss": 0.2455, "step": 16780 }, { "epoch": 0.30538933047163463, "grad_norm": 1.070762276649475, "learning_rate": 8.511945857243418e-06, "loss": 0.24, "step": 16790 }, { "epoch": 0.30557121810145693, "grad_norm": 1.2653802633285522, "learning_rate": 8.510111624771104e-06, "loss": 0.2451, "step": 16800 }, { "epoch": 0.30575310573127923, "grad_norm": 1.1579831838607788, "learning_rate": 8.508276521203058e-06, "loss": 0.2403, "step": 16810 }, { "epoch": 0.3059349933611015, "grad_norm": 1.1777876615524292, "learning_rate": 8.506440547176118e-06, "loss": 0.2453, "step": 16820 }, { "epoch": 0.3061168809909238, "grad_norm": 1.2143641710281372, "learning_rate": 8.504603703327436e-06, "loss": 0.245, "step": 16830 }, { "epoch": 0.3062987686207461, "grad_norm": 1.1647313833236694, "learning_rate": 8.502765990294456e-06, "loss": 0.2408, "step": 16840 }, { "epoch": 0.3064806562505684, "grad_norm": 1.175020694732666, "learning_rate": 8.500927408714931e-06, "loss": 0.2468, "step": 16850 }, { "epoch": 0.3066625438803907, "grad_norm": 1.0475231409072876, "learning_rate": 8.499087959226912e-06, "loss": 0.239, "step": 16860 }, { "epoch": 0.306844431510213, "grad_norm": 1.0648504495620728, "learning_rate": 8.497247642468752e-06, "loss": 0.2423, "step": 16870 }, { "epoch": 0.3070263191400353, "grad_norm": 1.1327980756759644, "learning_rate": 8.495406459079103e-06, "loss": 0.2469, "step": 16880 }, { "epoch": 0.3072082067698576, "grad_norm": 1.1335347890853882, "learning_rate": 8.493564409696919e-06, "loss": 0.2397, "step": 16890 }, { "epoch": 0.3073900943996799, "grad_norm": 1.1085877418518066, "learning_rate": 8.49172149496146e-06, "loss": 0.2444, "step": 16900 }, { "epoch": 0.3075719820295022, "grad_norm": 1.1087485551834106, "learning_rate": 8.489877715512277e-06, "loss": 0.241, "step": 16910 }, { "epoch": 0.30775386965932444, "grad_norm": 1.0898114442825317, "learning_rate": 8.488033071989223e-06, "loss": 0.2446, "step": 16920 }, { "epoch": 0.30793575728914674, "grad_norm": 1.177289605140686, "learning_rate": 8.48618756503246e-06, "loss": 0.2448, "step": 16930 }, { "epoch": 0.30811764491896904, "grad_norm": 1.0080673694610596, "learning_rate": 8.484341195282438e-06, "loss": 0.2364, "step": 16940 }, { "epoch": 0.30829953254879133, "grad_norm": 1.0985051393508911, "learning_rate": 8.482493963379914e-06, "loss": 0.2459, "step": 16950 }, { "epoch": 0.30848142017861363, "grad_norm": 1.0400090217590332, "learning_rate": 8.480645869965939e-06, "loss": 0.2466, "step": 16960 }, { "epoch": 0.3086633078084359, "grad_norm": 1.0825755596160889, "learning_rate": 8.478796915681866e-06, "loss": 0.2395, "step": 16970 }, { "epoch": 0.3088451954382582, "grad_norm": 1.1496561765670776, "learning_rate": 8.476947101169348e-06, "loss": 0.2457, "step": 16980 }, { "epoch": 0.3090270830680805, "grad_norm": 1.1222411394119263, "learning_rate": 8.475096427070333e-06, "loss": 0.2398, "step": 16990 }, { "epoch": 0.3092089706979028, "grad_norm": 1.2720251083374023, "learning_rate": 8.47324489402707e-06, "loss": 0.2406, "step": 17000 }, { "epoch": 0.3093908583277251, "grad_norm": 1.120935082435608, "learning_rate": 8.471392502682104e-06, "loss": 0.2409, "step": 17010 }, { "epoch": 0.3095727459575474, "grad_norm": 1.3452249765396118, "learning_rate": 8.469539253678281e-06, "loss": 0.2388, "step": 17020 }, { "epoch": 0.3097546335873697, "grad_norm": 1.09719979763031, "learning_rate": 8.46768514765874e-06, "loss": 0.2419, "step": 17030 }, { "epoch": 0.309936521217192, "grad_norm": 1.0683366060256958, "learning_rate": 8.465830185266924e-06, "loss": 0.2414, "step": 17040 }, { "epoch": 0.3101184088470143, "grad_norm": 1.1404798030853271, "learning_rate": 8.463974367146564e-06, "loss": 0.2447, "step": 17050 }, { "epoch": 0.3103002964768366, "grad_norm": 1.0585665702819824, "learning_rate": 8.462117693941698e-06, "loss": 0.2467, "step": 17060 }, { "epoch": 0.3104821841066589, "grad_norm": 1.014467477798462, "learning_rate": 8.460260166296652e-06, "loss": 0.2337, "step": 17070 }, { "epoch": 0.3106640717364812, "grad_norm": 1.0759268999099731, "learning_rate": 8.458401784856055e-06, "loss": 0.2371, "step": 17080 }, { "epoch": 0.3108459593663035, "grad_norm": 1.0869052410125732, "learning_rate": 8.45654255026483e-06, "loss": 0.2398, "step": 17090 }, { "epoch": 0.3110278469961258, "grad_norm": 1.1731657981872559, "learning_rate": 8.454682463168193e-06, "loss": 0.2417, "step": 17100 }, { "epoch": 0.3112097346259481, "grad_norm": 1.1612756252288818, "learning_rate": 8.452821524211663e-06, "loss": 0.2425, "step": 17110 }, { "epoch": 0.3113916222557704, "grad_norm": 0.9929637908935547, "learning_rate": 8.450959734041046e-06, "loss": 0.2356, "step": 17120 }, { "epoch": 0.3115735098855927, "grad_norm": 1.1920799016952515, "learning_rate": 8.449097093302451e-06, "loss": 0.242, "step": 17130 }, { "epoch": 0.311755397515415, "grad_norm": 1.0078868865966797, "learning_rate": 8.447233602642278e-06, "loss": 0.2422, "step": 17140 }, { "epoch": 0.3119372851452373, "grad_norm": 1.1925419569015503, "learning_rate": 8.445369262707222e-06, "loss": 0.2421, "step": 17150 }, { "epoch": 0.3121191727750596, "grad_norm": 1.1631150245666504, "learning_rate": 8.443504074144272e-06, "loss": 0.2467, "step": 17160 }, { "epoch": 0.31230106040488187, "grad_norm": 1.1490345001220703, "learning_rate": 8.441638037600717e-06, "loss": 0.2377, "step": 17170 }, { "epoch": 0.31248294803470417, "grad_norm": 1.2134582996368408, "learning_rate": 8.439771153724134e-06, "loss": 0.2465, "step": 17180 }, { "epoch": 0.31266483566452646, "grad_norm": 1.1568607091903687, "learning_rate": 8.4379034231624e-06, "loss": 0.2394, "step": 17190 }, { "epoch": 0.31284672329434876, "grad_norm": 1.2982045412063599, "learning_rate": 8.436034846563676e-06, "loss": 0.244, "step": 17200 }, { "epoch": 0.31302861092417106, "grad_norm": 1.03049898147583, "learning_rate": 8.434165424576426e-06, "loss": 0.2385, "step": 17210 }, { "epoch": 0.31321049855399336, "grad_norm": 1.2870643138885498, "learning_rate": 8.432295157849404e-06, "loss": 0.2462, "step": 17220 }, { "epoch": 0.31339238618381565, "grad_norm": 1.123897910118103, "learning_rate": 8.430424047031659e-06, "loss": 0.2405, "step": 17230 }, { "epoch": 0.31357427381363795, "grad_norm": 1.1166447401046753, "learning_rate": 8.428552092772528e-06, "loss": 0.243, "step": 17240 }, { "epoch": 0.31375616144346025, "grad_norm": 1.194369912147522, "learning_rate": 8.426679295721648e-06, "loss": 0.2414, "step": 17250 }, { "epoch": 0.31393804907328254, "grad_norm": 1.0631046295166016, "learning_rate": 8.424805656528942e-06, "loss": 0.2365, "step": 17260 }, { "epoch": 0.31411993670310484, "grad_norm": 1.2828593254089355, "learning_rate": 8.422931175844626e-06, "loss": 0.2374, "step": 17270 }, { "epoch": 0.31430182433292714, "grad_norm": 1.0537234544754028, "learning_rate": 8.421055854319214e-06, "loss": 0.2386, "step": 17280 }, { "epoch": 0.31448371196274943, "grad_norm": 1.1555967330932617, "learning_rate": 8.419179692603504e-06, "loss": 0.2387, "step": 17290 }, { "epoch": 0.31466559959257173, "grad_norm": 1.0570846796035767, "learning_rate": 8.417302691348592e-06, "loss": 0.2399, "step": 17300 }, { "epoch": 0.31484748722239403, "grad_norm": 0.9920412302017212, "learning_rate": 8.415424851205862e-06, "loss": 0.2418, "step": 17310 }, { "epoch": 0.3150293748522163, "grad_norm": 1.1797183752059937, "learning_rate": 8.413546172826986e-06, "loss": 0.2374, "step": 17320 }, { "epoch": 0.3152112624820386, "grad_norm": 1.026594638824463, "learning_rate": 8.411666656863935e-06, "loss": 0.2381, "step": 17330 }, { "epoch": 0.31539315011186086, "grad_norm": 0.9913362264633179, "learning_rate": 8.409786303968965e-06, "loss": 0.2425, "step": 17340 }, { "epoch": 0.31557503774168316, "grad_norm": 1.2059279680252075, "learning_rate": 8.407905114794621e-06, "loss": 0.2425, "step": 17350 }, { "epoch": 0.31575692537150546, "grad_norm": 1.3564531803131104, "learning_rate": 8.406023089993744e-06, "loss": 0.2451, "step": 17360 }, { "epoch": 0.31593881300132776, "grad_norm": 1.1375837326049805, "learning_rate": 8.404140230219461e-06, "loss": 0.2423, "step": 17370 }, { "epoch": 0.31612070063115005, "grad_norm": 1.1811593770980835, "learning_rate": 8.402256536125188e-06, "loss": 0.241, "step": 17380 }, { "epoch": 0.31630258826097235, "grad_norm": 1.1975454092025757, "learning_rate": 8.400372008364635e-06, "loss": 0.2363, "step": 17390 }, { "epoch": 0.31648447589079465, "grad_norm": 1.1011886596679688, "learning_rate": 8.398486647591795e-06, "loss": 0.242, "step": 17400 }, { "epoch": 0.31666636352061694, "grad_norm": 1.0360519886016846, "learning_rate": 8.396600454460957e-06, "loss": 0.2419, "step": 17410 }, { "epoch": 0.31684825115043924, "grad_norm": 1.1125507354736328, "learning_rate": 8.394713429626695e-06, "loss": 0.2413, "step": 17420 }, { "epoch": 0.31703013878026154, "grad_norm": 1.1085683107376099, "learning_rate": 8.39282557374387e-06, "loss": 0.2445, "step": 17430 }, { "epoch": 0.31721202641008384, "grad_norm": 1.0825371742248535, "learning_rate": 8.390936887467633e-06, "loss": 0.2423, "step": 17440 }, { "epoch": 0.31739391403990613, "grad_norm": 1.1374180316925049, "learning_rate": 8.389047371453427e-06, "loss": 0.2456, "step": 17450 }, { "epoch": 0.31757580166972843, "grad_norm": 1.0457067489624023, "learning_rate": 8.387157026356976e-06, "loss": 0.2387, "step": 17460 }, { "epoch": 0.3177576892995507, "grad_norm": 1.145015001296997, "learning_rate": 8.385265852834296e-06, "loss": 0.2404, "step": 17470 }, { "epoch": 0.317939576929373, "grad_norm": 1.1336253881454468, "learning_rate": 8.383373851541691e-06, "loss": 0.2407, "step": 17480 }, { "epoch": 0.3181214645591953, "grad_norm": 0.9884617328643799, "learning_rate": 8.381481023135752e-06, "loss": 0.2474, "step": 17490 }, { "epoch": 0.3183033521890176, "grad_norm": 1.108948826789856, "learning_rate": 8.379587368273354e-06, "loss": 0.246, "step": 17500 }, { "epoch": 0.3184852398188399, "grad_norm": 1.136454701423645, "learning_rate": 8.37769288761166e-06, "loss": 0.2396, "step": 17510 }, { "epoch": 0.3186671274486622, "grad_norm": 1.0823819637298584, "learning_rate": 8.375797581808123e-06, "loss": 0.2414, "step": 17520 }, { "epoch": 0.3188490150784845, "grad_norm": 1.3169891834259033, "learning_rate": 8.373901451520478e-06, "loss": 0.2422, "step": 17530 }, { "epoch": 0.3190309027083068, "grad_norm": 1.014597773551941, "learning_rate": 8.372004497406749e-06, "loss": 0.237, "step": 17540 }, { "epoch": 0.3192127903381291, "grad_norm": 1.1122320890426636, "learning_rate": 8.370106720125244e-06, "loss": 0.2421, "step": 17550 }, { "epoch": 0.3193946779679514, "grad_norm": 1.1073418855667114, "learning_rate": 8.36820812033456e-06, "loss": 0.2361, "step": 17560 }, { "epoch": 0.3195765655977737, "grad_norm": 1.1521282196044922, "learning_rate": 8.366308698693573e-06, "loss": 0.2447, "step": 17570 }, { "epoch": 0.319758453227596, "grad_norm": 0.9338330030441284, "learning_rate": 8.364408455861451e-06, "loss": 0.2381, "step": 17580 }, { "epoch": 0.3199403408574183, "grad_norm": 1.101046085357666, "learning_rate": 8.362507392497643e-06, "loss": 0.2386, "step": 17590 }, { "epoch": 0.3201222284872406, "grad_norm": 1.0585565567016602, "learning_rate": 8.360605509261883e-06, "loss": 0.2376, "step": 17600 }, { "epoch": 0.3203041161170629, "grad_norm": 1.0252063274383545, "learning_rate": 8.358702806814192e-06, "loss": 0.2464, "step": 17610 }, { "epoch": 0.3204860037468852, "grad_norm": 1.1294273138046265, "learning_rate": 8.356799285814875e-06, "loss": 0.2461, "step": 17620 }, { "epoch": 0.3206678913767075, "grad_norm": 1.0226770639419556, "learning_rate": 8.354894946924514e-06, "loss": 0.2367, "step": 17630 }, { "epoch": 0.3208497790065298, "grad_norm": 1.0170129537582397, "learning_rate": 8.352989790803988e-06, "loss": 0.2382, "step": 17640 }, { "epoch": 0.3210316666363521, "grad_norm": 1.090009331703186, "learning_rate": 8.351083818114447e-06, "loss": 0.2436, "step": 17650 }, { "epoch": 0.3212135542661744, "grad_norm": 1.0107086896896362, "learning_rate": 8.349177029517333e-06, "loss": 0.2426, "step": 17660 }, { "epoch": 0.32139544189599667, "grad_norm": 1.2126145362854004, "learning_rate": 8.347269425674362e-06, "loss": 0.2392, "step": 17670 }, { "epoch": 0.32157732952581897, "grad_norm": 1.027884602546692, "learning_rate": 8.345361007247547e-06, "loss": 0.2399, "step": 17680 }, { "epoch": 0.32175921715564126, "grad_norm": 1.0584042072296143, "learning_rate": 8.343451774899169e-06, "loss": 0.2395, "step": 17690 }, { "epoch": 0.32194110478546356, "grad_norm": 1.091700792312622, "learning_rate": 8.3415417292918e-06, "loss": 0.2392, "step": 17700 }, { "epoch": 0.32212299241528586, "grad_norm": 1.108428716659546, "learning_rate": 8.339630871088293e-06, "loss": 0.2355, "step": 17710 }, { "epoch": 0.32230488004510816, "grad_norm": 1.2520198822021484, "learning_rate": 8.337719200951779e-06, "loss": 0.2447, "step": 17720 }, { "epoch": 0.32248676767493045, "grad_norm": 1.0365803241729736, "learning_rate": 8.335806719545677e-06, "loss": 0.2382, "step": 17730 }, { "epoch": 0.32266865530475275, "grad_norm": 1.0146421194076538, "learning_rate": 8.333893427533684e-06, "loss": 0.2422, "step": 17740 }, { "epoch": 0.32285054293457505, "grad_norm": 1.4381134510040283, "learning_rate": 8.331979325579775e-06, "loss": 0.2455, "step": 17750 }, { "epoch": 0.3230324305643973, "grad_norm": 1.0818530321121216, "learning_rate": 8.330064414348213e-06, "loss": 0.2453, "step": 17760 }, { "epoch": 0.3232143181942196, "grad_norm": 1.2479361295700073, "learning_rate": 8.328148694503539e-06, "loss": 0.2458, "step": 17770 }, { "epoch": 0.3233962058240419, "grad_norm": 1.076878309249878, "learning_rate": 8.326232166710574e-06, "loss": 0.2368, "step": 17780 }, { "epoch": 0.3235780934538642, "grad_norm": 1.020280122756958, "learning_rate": 8.324314831634418e-06, "loss": 0.2428, "step": 17790 }, { "epoch": 0.3237599810836865, "grad_norm": 1.1207789182662964, "learning_rate": 8.322396689940453e-06, "loss": 0.2381, "step": 17800 }, { "epoch": 0.3239418687135088, "grad_norm": 1.1924148797988892, "learning_rate": 8.32047774229434e-06, "loss": 0.237, "step": 17810 }, { "epoch": 0.32412375634333107, "grad_norm": 1.029096245765686, "learning_rate": 8.318557989362022e-06, "loss": 0.2387, "step": 17820 }, { "epoch": 0.32430564397315337, "grad_norm": 1.0677367448806763, "learning_rate": 8.31663743180972e-06, "loss": 0.2378, "step": 17830 }, { "epoch": 0.32448753160297567, "grad_norm": 1.1404725313186646, "learning_rate": 8.314716070303933e-06, "loss": 0.2461, "step": 17840 }, { "epoch": 0.32466941923279796, "grad_norm": 1.1586166620254517, "learning_rate": 8.31279390551144e-06, "loss": 0.244, "step": 17850 }, { "epoch": 0.32485130686262026, "grad_norm": 0.975590705871582, "learning_rate": 8.3108709380993e-06, "loss": 0.2424, "step": 17860 }, { "epoch": 0.32503319449244256, "grad_norm": 1.066352367401123, "learning_rate": 8.308947168734846e-06, "loss": 0.2406, "step": 17870 }, { "epoch": 0.32521508212226485, "grad_norm": 1.0884690284729004, "learning_rate": 8.307022598085699e-06, "loss": 0.2456, "step": 17880 }, { "epoch": 0.32539696975208715, "grad_norm": 1.2440153360366821, "learning_rate": 8.305097226819744e-06, "loss": 0.2461, "step": 17890 }, { "epoch": 0.32557885738190945, "grad_norm": 1.1572452783584595, "learning_rate": 8.30317105560516e-06, "loss": 0.2337, "step": 17900 }, { "epoch": 0.32576074501173174, "grad_norm": 1.1508656740188599, "learning_rate": 8.301244085110387e-06, "loss": 0.2412, "step": 17910 }, { "epoch": 0.32594263264155404, "grad_norm": 1.0944114923477173, "learning_rate": 8.299316316004154e-06, "loss": 0.2414, "step": 17920 }, { "epoch": 0.32612452027137634, "grad_norm": 1.1207197904586792, "learning_rate": 8.297387748955465e-06, "loss": 0.2414, "step": 17930 }, { "epoch": 0.32630640790119864, "grad_norm": 1.0476343631744385, "learning_rate": 8.2954583846336e-06, "loss": 0.2405, "step": 17940 }, { "epoch": 0.32648829553102093, "grad_norm": 1.2455599308013916, "learning_rate": 8.29352822370811e-06, "loss": 0.2375, "step": 17950 }, { "epoch": 0.32667018316084323, "grad_norm": 1.011339545249939, "learning_rate": 8.291597266848833e-06, "loss": 0.2367, "step": 17960 }, { "epoch": 0.3268520707906655, "grad_norm": 1.122495412826538, "learning_rate": 8.289665514725878e-06, "loss": 0.2337, "step": 17970 }, { "epoch": 0.3270339584204878, "grad_norm": 1.1096036434173584, "learning_rate": 8.287732968009628e-06, "loss": 0.2368, "step": 17980 }, { "epoch": 0.3272158460503101, "grad_norm": 1.0070850849151611, "learning_rate": 8.285799627370741e-06, "loss": 0.2415, "step": 17990 }, { "epoch": 0.3273977336801324, "grad_norm": 1.1685924530029297, "learning_rate": 8.283865493480158e-06, "loss": 0.2472, "step": 18000 }, { "epoch": 0.3275796213099547, "grad_norm": 1.0676673650741577, "learning_rate": 8.28193056700909e-06, "loss": 0.2437, "step": 18010 }, { "epoch": 0.327761508939777, "grad_norm": 1.131300449371338, "learning_rate": 8.279994848629022e-06, "loss": 0.2454, "step": 18020 }, { "epoch": 0.3279433965695993, "grad_norm": 1.1429650783538818, "learning_rate": 8.278058339011715e-06, "loss": 0.2426, "step": 18030 }, { "epoch": 0.3281252841994216, "grad_norm": 1.066686987876892, "learning_rate": 8.276121038829206e-06, "loss": 0.2391, "step": 18040 }, { "epoch": 0.3283071718292439, "grad_norm": 1.0353062152862549, "learning_rate": 8.274182948753803e-06, "loss": 0.2404, "step": 18050 }, { "epoch": 0.3284890594590662, "grad_norm": 0.9626215100288391, "learning_rate": 8.272244069458095e-06, "loss": 0.2362, "step": 18060 }, { "epoch": 0.3286709470888885, "grad_norm": 1.1340659856796265, "learning_rate": 8.270304401614936e-06, "loss": 0.2365, "step": 18070 }, { "epoch": 0.3288528347187108, "grad_norm": 1.155951976776123, "learning_rate": 8.268363945897461e-06, "loss": 0.2436, "step": 18080 }, { "epoch": 0.3290347223485331, "grad_norm": 1.2335131168365479, "learning_rate": 8.266422702979075e-06, "loss": 0.2414, "step": 18090 }, { "epoch": 0.3292166099783554, "grad_norm": 1.0515141487121582, "learning_rate": 8.264480673533454e-06, "loss": 0.2395, "step": 18100 }, { "epoch": 0.3293984976081777, "grad_norm": 1.1160974502563477, "learning_rate": 8.262537858234552e-06, "loss": 0.2423, "step": 18110 }, { "epoch": 0.329580385238, "grad_norm": 0.9704128503799438, "learning_rate": 8.260594257756592e-06, "loss": 0.241, "step": 18120 }, { "epoch": 0.3297622728678223, "grad_norm": 1.1378782987594604, "learning_rate": 8.25864987277407e-06, "loss": 0.2357, "step": 18130 }, { "epoch": 0.3299441604976446, "grad_norm": 1.0444890260696411, "learning_rate": 8.256704703961758e-06, "loss": 0.2369, "step": 18140 }, { "epoch": 0.3301260481274669, "grad_norm": 1.0622199773788452, "learning_rate": 8.254758751994693e-06, "loss": 0.242, "step": 18150 }, { "epoch": 0.3303079357572892, "grad_norm": 1.0283966064453125, "learning_rate": 8.252812017548191e-06, "loss": 0.2392, "step": 18160 }, { "epoch": 0.33048982338711147, "grad_norm": 1.020594596862793, "learning_rate": 8.250864501297838e-06, "loss": 0.2379, "step": 18170 }, { "epoch": 0.3306717110169337, "grad_norm": 1.202979326248169, "learning_rate": 8.248916203919484e-06, "loss": 0.2391, "step": 18180 }, { "epoch": 0.330853598646756, "grad_norm": 1.1230056285858154, "learning_rate": 8.246967126089257e-06, "loss": 0.2453, "step": 18190 }, { "epoch": 0.3310354862765783, "grad_norm": 1.0409367084503174, "learning_rate": 8.245017268483559e-06, "loss": 0.2484, "step": 18200 }, { "epoch": 0.3312173739064006, "grad_norm": 1.067679524421692, "learning_rate": 8.243066631779056e-06, "loss": 0.2383, "step": 18210 }, { "epoch": 0.3313992615362229, "grad_norm": 1.009343147277832, "learning_rate": 8.241115216652684e-06, "loss": 0.2332, "step": 18220 }, { "epoch": 0.3315811491660452, "grad_norm": 1.0373283624649048, "learning_rate": 8.239163023781654e-06, "loss": 0.2345, "step": 18230 }, { "epoch": 0.3317630367958675, "grad_norm": 1.1121355295181274, "learning_rate": 8.237210053843444e-06, "loss": 0.2403, "step": 18240 }, { "epoch": 0.3319449244256898, "grad_norm": 1.0550665855407715, "learning_rate": 8.235256307515801e-06, "loss": 0.2417, "step": 18250 }, { "epoch": 0.3321268120555121, "grad_norm": 1.1119602918624878, "learning_rate": 8.233301785476746e-06, "loss": 0.241, "step": 18260 }, { "epoch": 0.3323086996853344, "grad_norm": 1.1335630416870117, "learning_rate": 8.231346488404564e-06, "loss": 0.2423, "step": 18270 }, { "epoch": 0.3324905873151567, "grad_norm": 0.9856058359146118, "learning_rate": 8.22939041697781e-06, "loss": 0.2369, "step": 18280 }, { "epoch": 0.332672474944979, "grad_norm": 1.0501062870025635, "learning_rate": 8.22743357187531e-06, "loss": 0.243, "step": 18290 }, { "epoch": 0.3328543625748013, "grad_norm": 1.0610977411270142, "learning_rate": 8.225475953776156e-06, "loss": 0.241, "step": 18300 }, { "epoch": 0.3330362502046236, "grad_norm": 1.1157371997833252, "learning_rate": 8.22351756335971e-06, "loss": 0.24, "step": 18310 }, { "epoch": 0.33321813783444587, "grad_norm": 1.0618952512741089, "learning_rate": 8.221558401305602e-06, "loss": 0.2409, "step": 18320 }, { "epoch": 0.33340002546426817, "grad_norm": 1.284769058227539, "learning_rate": 8.21959846829373e-06, "loss": 0.24, "step": 18330 }, { "epoch": 0.33358191309409047, "grad_norm": 1.148939847946167, "learning_rate": 8.217637765004254e-06, "loss": 0.2398, "step": 18340 }, { "epoch": 0.33376380072391276, "grad_norm": 1.0564181804656982, "learning_rate": 8.215676292117612e-06, "loss": 0.2446, "step": 18350 }, { "epoch": 0.33394568835373506, "grad_norm": 0.9763849973678589, "learning_rate": 8.213714050314496e-06, "loss": 0.2371, "step": 18360 }, { "epoch": 0.33412757598355736, "grad_norm": 1.1640968322753906, "learning_rate": 8.211751040275879e-06, "loss": 0.2433, "step": 18370 }, { "epoch": 0.33430946361337965, "grad_norm": 1.1046663522720337, "learning_rate": 8.20978726268299e-06, "loss": 0.2491, "step": 18380 }, { "epoch": 0.33449135124320195, "grad_norm": 1.0722484588623047, "learning_rate": 8.207822718217328e-06, "loss": 0.2416, "step": 18390 }, { "epoch": 0.33467323887302425, "grad_norm": 1.1351474523544312, "learning_rate": 8.205857407560658e-06, "loss": 0.246, "step": 18400 }, { "epoch": 0.33485512650284655, "grad_norm": 1.092793345451355, "learning_rate": 8.203891331395012e-06, "loss": 0.2418, "step": 18410 }, { "epoch": 0.33503701413266884, "grad_norm": 1.1406116485595703, "learning_rate": 8.201924490402687e-06, "loss": 0.245, "step": 18420 }, { "epoch": 0.33521890176249114, "grad_norm": 1.0450466871261597, "learning_rate": 8.199956885266242e-06, "loss": 0.2395, "step": 18430 }, { "epoch": 0.33540078939231344, "grad_norm": 1.0788438320159912, "learning_rate": 8.197988516668508e-06, "loss": 0.2408, "step": 18440 }, { "epoch": 0.33558267702213573, "grad_norm": 1.0764347314834595, "learning_rate": 8.196019385292573e-06, "loss": 0.2436, "step": 18450 }, { "epoch": 0.33576456465195803, "grad_norm": 1.1865594387054443, "learning_rate": 8.194049491821799e-06, "loss": 0.2465, "step": 18460 }, { "epoch": 0.33594645228178033, "grad_norm": 1.046962857246399, "learning_rate": 8.192078836939801e-06, "loss": 0.2375, "step": 18470 }, { "epoch": 0.3361283399116026, "grad_norm": 1.2411059141159058, "learning_rate": 8.19010742133047e-06, "loss": 0.2471, "step": 18480 }, { "epoch": 0.3363102275414249, "grad_norm": 1.005616545677185, "learning_rate": 8.188135245677954e-06, "loss": 0.2381, "step": 18490 }, { "epoch": 0.3364921151712472, "grad_norm": 1.1739150285720825, "learning_rate": 8.186162310666666e-06, "loss": 0.2368, "step": 18500 }, { "epoch": 0.3366740028010695, "grad_norm": 1.223450779914856, "learning_rate": 8.184188616981281e-06, "loss": 0.2437, "step": 18510 }, { "epoch": 0.3368558904308918, "grad_norm": 1.063596248626709, "learning_rate": 8.182214165306741e-06, "loss": 0.2391, "step": 18520 }, { "epoch": 0.3370377780607141, "grad_norm": 1.5105847120285034, "learning_rate": 8.18023895632825e-06, "loss": 0.2422, "step": 18530 }, { "epoch": 0.3372196656905364, "grad_norm": 1.168526291847229, "learning_rate": 8.178262990731272e-06, "loss": 0.2444, "step": 18540 }, { "epoch": 0.3374015533203587, "grad_norm": 1.0606147050857544, "learning_rate": 8.176286269201538e-06, "loss": 0.2409, "step": 18550 }, { "epoch": 0.337583440950181, "grad_norm": 1.1473525762557983, "learning_rate": 8.174308792425034e-06, "loss": 0.2395, "step": 18560 }, { "epoch": 0.3377653285800033, "grad_norm": 1.1190978288650513, "learning_rate": 8.17233056108802e-06, "loss": 0.2443, "step": 18570 }, { "epoch": 0.3379472162098256, "grad_norm": 1.058236837387085, "learning_rate": 8.170351575877006e-06, "loss": 0.2351, "step": 18580 }, { "epoch": 0.3381291038396479, "grad_norm": 1.1129353046417236, "learning_rate": 8.168371837478768e-06, "loss": 0.2458, "step": 18590 }, { "epoch": 0.33831099146947013, "grad_norm": 1.0548298358917236, "learning_rate": 8.166391346580348e-06, "loss": 0.2375, "step": 18600 }, { "epoch": 0.33849287909929243, "grad_norm": 1.1663175821304321, "learning_rate": 8.16441010386904e-06, "loss": 0.2415, "step": 18610 }, { "epoch": 0.33867476672911473, "grad_norm": 1.2130937576293945, "learning_rate": 8.162428110032409e-06, "loss": 0.2397, "step": 18620 }, { "epoch": 0.338856654358937, "grad_norm": 1.2481780052185059, "learning_rate": 8.160445365758271e-06, "loss": 0.2434, "step": 18630 }, { "epoch": 0.3390385419887593, "grad_norm": 1.0734604597091675, "learning_rate": 8.15846187173471e-06, "loss": 0.2452, "step": 18640 }, { "epoch": 0.3392204296185816, "grad_norm": 0.9888572692871094, "learning_rate": 8.156477628650065e-06, "loss": 0.2428, "step": 18650 }, { "epoch": 0.3394023172484039, "grad_norm": 1.0340362787246704, "learning_rate": 8.15449263719294e-06, "loss": 0.2384, "step": 18660 }, { "epoch": 0.3395842048782262, "grad_norm": 1.0570247173309326, "learning_rate": 8.152506898052191e-06, "loss": 0.2433, "step": 18670 }, { "epoch": 0.3397660925080485, "grad_norm": 1.0994501113891602, "learning_rate": 8.150520411916943e-06, "loss": 0.2424, "step": 18680 }, { "epoch": 0.3399479801378708, "grad_norm": 1.1094231605529785, "learning_rate": 8.148533179476575e-06, "loss": 0.2398, "step": 18690 }, { "epoch": 0.3401298677676931, "grad_norm": 1.1259690523147583, "learning_rate": 8.146545201420722e-06, "loss": 0.24, "step": 18700 }, { "epoch": 0.3403117553975154, "grad_norm": 1.0425059795379639, "learning_rate": 8.144556478439285e-06, "loss": 0.242, "step": 18710 }, { "epoch": 0.3404936430273377, "grad_norm": 1.139539122581482, "learning_rate": 8.14256701122242e-06, "loss": 0.2393, "step": 18720 }, { "epoch": 0.34067553065716, "grad_norm": 1.1858630180358887, "learning_rate": 8.140576800460537e-06, "loss": 0.2406, "step": 18730 }, { "epoch": 0.3408574182869823, "grad_norm": 1.129947543144226, "learning_rate": 8.138585846844312e-06, "loss": 0.2368, "step": 18740 }, { "epoch": 0.3410393059168046, "grad_norm": 1.1117815971374512, "learning_rate": 8.136594151064673e-06, "loss": 0.2452, "step": 18750 }, { "epoch": 0.3412211935466269, "grad_norm": 1.0062024593353271, "learning_rate": 8.13460171381281e-06, "loss": 0.2352, "step": 18760 }, { "epoch": 0.3414030811764492, "grad_norm": 1.0165077447891235, "learning_rate": 8.132608535780163e-06, "loss": 0.2316, "step": 18770 }, { "epoch": 0.3415849688062715, "grad_norm": 1.0657894611358643, "learning_rate": 8.130614617658439e-06, "loss": 0.2369, "step": 18780 }, { "epoch": 0.3417668564360938, "grad_norm": 1.159389615058899, "learning_rate": 8.128619960139594e-06, "loss": 0.2469, "step": 18790 }, { "epoch": 0.3419487440659161, "grad_norm": 1.1313371658325195, "learning_rate": 8.126624563915842e-06, "loss": 0.2379, "step": 18800 }, { "epoch": 0.3421306316957384, "grad_norm": 1.107779860496521, "learning_rate": 8.124628429679658e-06, "loss": 0.2402, "step": 18810 }, { "epoch": 0.34231251932556067, "grad_norm": 1.0201448202133179, "learning_rate": 8.122631558123769e-06, "loss": 0.237, "step": 18820 }, { "epoch": 0.34249440695538297, "grad_norm": 1.1753418445587158, "learning_rate": 8.120633949941155e-06, "loss": 0.2356, "step": 18830 }, { "epoch": 0.34267629458520527, "grad_norm": 1.1351187229156494, "learning_rate": 8.118635605825062e-06, "loss": 0.25, "step": 18840 }, { "epoch": 0.34285818221502756, "grad_norm": 1.338121771812439, "learning_rate": 8.116636526468978e-06, "loss": 0.243, "step": 18850 }, { "epoch": 0.34304006984484986, "grad_norm": 1.0817935466766357, "learning_rate": 8.114636712566655e-06, "loss": 0.2393, "step": 18860 }, { "epoch": 0.34322195747467216, "grad_norm": 1.0730266571044922, "learning_rate": 8.112636164812099e-06, "loss": 0.2429, "step": 18870 }, { "epoch": 0.34340384510449445, "grad_norm": 1.2624022960662842, "learning_rate": 8.110634883899571e-06, "loss": 0.2392, "step": 18880 }, { "epoch": 0.34358573273431675, "grad_norm": 1.0777050256729126, "learning_rate": 8.10863287052358e-06, "loss": 0.2416, "step": 18890 }, { "epoch": 0.34376762036413905, "grad_norm": 1.1783580780029297, "learning_rate": 8.106630125378896e-06, "loss": 0.2444, "step": 18900 }, { "epoch": 0.34394950799396135, "grad_norm": 1.08743155002594, "learning_rate": 8.104626649160545e-06, "loss": 0.2394, "step": 18910 }, { "epoch": 0.34413139562378364, "grad_norm": 1.0962332487106323, "learning_rate": 8.102622442563796e-06, "loss": 0.2434, "step": 18920 }, { "epoch": 0.34431328325360594, "grad_norm": 1.0727392435073853, "learning_rate": 8.100617506284183e-06, "loss": 0.241, "step": 18930 }, { "epoch": 0.34449517088342824, "grad_norm": 1.1791802644729614, "learning_rate": 8.098611841017484e-06, "loss": 0.2418, "step": 18940 }, { "epoch": 0.34467705851325053, "grad_norm": 1.1286652088165283, "learning_rate": 8.09660544745974e-06, "loss": 0.2361, "step": 18950 }, { "epoch": 0.34485894614307283, "grad_norm": 1.301607370376587, "learning_rate": 8.094598326307235e-06, "loss": 0.2425, "step": 18960 }, { "epoch": 0.34504083377289513, "grad_norm": 0.99855637550354, "learning_rate": 8.092590478256512e-06, "loss": 0.2457, "step": 18970 }, { "epoch": 0.3452227214027174, "grad_norm": 1.179500937461853, "learning_rate": 8.090581904004363e-06, "loss": 0.2439, "step": 18980 }, { "epoch": 0.3454046090325397, "grad_norm": 1.1297231912612915, "learning_rate": 8.088572604247832e-06, "loss": 0.2375, "step": 18990 }, { "epoch": 0.345586496662362, "grad_norm": 1.1838889122009277, "learning_rate": 8.086562579684217e-06, "loss": 0.2411, "step": 19000 }, { "epoch": 0.34576838429218426, "grad_norm": 1.1058369874954224, "learning_rate": 8.084551831011065e-06, "loss": 0.2379, "step": 19010 }, { "epoch": 0.34595027192200656, "grad_norm": 1.0934828519821167, "learning_rate": 8.082540358926175e-06, "loss": 0.2452, "step": 19020 }, { "epoch": 0.34613215955182886, "grad_norm": 1.1290647983551025, "learning_rate": 8.080528164127599e-06, "loss": 0.2391, "step": 19030 }, { "epoch": 0.34631404718165115, "grad_norm": 1.1356394290924072, "learning_rate": 8.078515247313639e-06, "loss": 0.2452, "step": 19040 }, { "epoch": 0.34649593481147345, "grad_norm": 1.098906397819519, "learning_rate": 8.076501609182844e-06, "loss": 0.2419, "step": 19050 }, { "epoch": 0.34667782244129575, "grad_norm": 1.1065016984939575, "learning_rate": 8.074487250434018e-06, "loss": 0.2475, "step": 19060 }, { "epoch": 0.34685971007111804, "grad_norm": 1.182944893836975, "learning_rate": 8.072472171766214e-06, "loss": 0.2489, "step": 19070 }, { "epoch": 0.34704159770094034, "grad_norm": 1.0782215595245361, "learning_rate": 8.070456373878735e-06, "loss": 0.2408, "step": 19080 }, { "epoch": 0.34722348533076264, "grad_norm": 1.0724421739578247, "learning_rate": 8.068439857471128e-06, "loss": 0.2424, "step": 19090 }, { "epoch": 0.34740537296058493, "grad_norm": 1.0261377096176147, "learning_rate": 8.066422623243198e-06, "loss": 0.2374, "step": 19100 }, { "epoch": 0.34758726059040723, "grad_norm": 1.205142617225647, "learning_rate": 8.064404671894997e-06, "loss": 0.2414, "step": 19110 }, { "epoch": 0.34776914822022953, "grad_norm": 1.1337488889694214, "learning_rate": 8.062386004126818e-06, "loss": 0.2408, "step": 19120 }, { "epoch": 0.3479510358500518, "grad_norm": 1.043106198310852, "learning_rate": 8.060366620639215e-06, "loss": 0.2409, "step": 19130 }, { "epoch": 0.3481329234798741, "grad_norm": 1.2015527486801147, "learning_rate": 8.058346522132979e-06, "loss": 0.2409, "step": 19140 }, { "epoch": 0.3483148111096964, "grad_norm": 1.1069492101669312, "learning_rate": 8.056325709309159e-06, "loss": 0.2462, "step": 19150 }, { "epoch": 0.3484966987395187, "grad_norm": 1.0237934589385986, "learning_rate": 8.05430418286904e-06, "loss": 0.2384, "step": 19160 }, { "epoch": 0.348678586369341, "grad_norm": 1.0335527658462524, "learning_rate": 8.052281943514171e-06, "loss": 0.2353, "step": 19170 }, { "epoch": 0.3488604739991633, "grad_norm": 1.0093574523925781, "learning_rate": 8.050258991946334e-06, "loss": 0.2425, "step": 19180 }, { "epoch": 0.3490423616289856, "grad_norm": 1.1733046770095825, "learning_rate": 8.048235328867561e-06, "loss": 0.2462, "step": 19190 }, { "epoch": 0.3492242492588079, "grad_norm": 1.116045355796814, "learning_rate": 8.046210954980139e-06, "loss": 0.2468, "step": 19200 }, { "epoch": 0.3494061368886302, "grad_norm": 0.9700180292129517, "learning_rate": 8.044185870986589e-06, "loss": 0.2381, "step": 19210 }, { "epoch": 0.3495880245184525, "grad_norm": 1.0554620027542114, "learning_rate": 8.042160077589693e-06, "loss": 0.2375, "step": 19220 }, { "epoch": 0.3497699121482748, "grad_norm": 1.1598381996154785, "learning_rate": 8.040133575492466e-06, "loss": 0.244, "step": 19230 }, { "epoch": 0.3499517997780971, "grad_norm": 1.0800161361694336, "learning_rate": 8.038106365398175e-06, "loss": 0.2398, "step": 19240 }, { "epoch": 0.3501336874079194, "grad_norm": 1.078354001045227, "learning_rate": 8.036078448010334e-06, "loss": 0.2443, "step": 19250 }, { "epoch": 0.3503155750377417, "grad_norm": 1.1011981964111328, "learning_rate": 8.034049824032701e-06, "loss": 0.2395, "step": 19260 }, { "epoch": 0.350497462667564, "grad_norm": 1.1388776302337646, "learning_rate": 8.032020494169277e-06, "loss": 0.2425, "step": 19270 }, { "epoch": 0.3506793502973863, "grad_norm": 1.1568511724472046, "learning_rate": 8.029990459124313e-06, "loss": 0.2395, "step": 19280 }, { "epoch": 0.3508612379272086, "grad_norm": 1.1334553956985474, "learning_rate": 8.027959719602296e-06, "loss": 0.2377, "step": 19290 }, { "epoch": 0.3510431255570309, "grad_norm": 1.0126951932907104, "learning_rate": 8.025928276307969e-06, "loss": 0.237, "step": 19300 }, { "epoch": 0.3512250131868532, "grad_norm": 1.0850121974945068, "learning_rate": 8.023896129946306e-06, "loss": 0.2429, "step": 19310 }, { "epoch": 0.35140690081667547, "grad_norm": 1.0907610654830933, "learning_rate": 8.02186328122254e-06, "loss": 0.2329, "step": 19320 }, { "epoch": 0.35158878844649777, "grad_norm": 1.0459132194519043, "learning_rate": 8.019829730842136e-06, "loss": 0.2369, "step": 19330 }, { "epoch": 0.35177067607632007, "grad_norm": 1.0403358936309814, "learning_rate": 8.017795479510807e-06, "loss": 0.2399, "step": 19340 }, { "epoch": 0.35195256370614236, "grad_norm": 1.2093799114227295, "learning_rate": 8.01576052793451e-06, "loss": 0.2441, "step": 19350 }, { "epoch": 0.35213445133596466, "grad_norm": 1.195879578590393, "learning_rate": 8.01372487681944e-06, "loss": 0.2385, "step": 19360 }, { "epoch": 0.35231633896578696, "grad_norm": 1.1674798727035522, "learning_rate": 8.011688526872044e-06, "loss": 0.2478, "step": 19370 }, { "epoch": 0.35249822659560925, "grad_norm": 1.1101734638214111, "learning_rate": 8.009651478799e-06, "loss": 0.2349, "step": 19380 }, { "epoch": 0.35268011422543155, "grad_norm": 1.0908697843551636, "learning_rate": 8.007613733307237e-06, "loss": 0.2333, "step": 19390 }, { "epoch": 0.35286200185525385, "grad_norm": 1.0285685062408447, "learning_rate": 8.005575291103926e-06, "loss": 0.2404, "step": 19400 }, { "epoch": 0.35304388948507615, "grad_norm": 1.0605310201644897, "learning_rate": 8.003536152896475e-06, "loss": 0.2374, "step": 19410 }, { "epoch": 0.35322577711489844, "grad_norm": 1.1636619567871094, "learning_rate": 8.001496319392531e-06, "loss": 0.2355, "step": 19420 }, { "epoch": 0.3534076647447207, "grad_norm": 1.0561200380325317, "learning_rate": 7.999455791299996e-06, "loss": 0.2433, "step": 19430 }, { "epoch": 0.353589552374543, "grad_norm": 1.1859208345413208, "learning_rate": 7.997414569326996e-06, "loss": 0.2386, "step": 19440 }, { "epoch": 0.3537714400043653, "grad_norm": 1.1530942916870117, "learning_rate": 7.99537265418191e-06, "loss": 0.2401, "step": 19450 }, { "epoch": 0.3539533276341876, "grad_norm": 1.1732267141342163, "learning_rate": 7.993330046573355e-06, "loss": 0.2412, "step": 19460 }, { "epoch": 0.3541352152640099, "grad_norm": 1.1378253698349, "learning_rate": 7.991286747210181e-06, "loss": 0.2463, "step": 19470 }, { "epoch": 0.35431710289383217, "grad_norm": 1.0983445644378662, "learning_rate": 7.989242756801488e-06, "loss": 0.2423, "step": 19480 }, { "epoch": 0.35449899052365447, "grad_norm": 1.0365228652954102, "learning_rate": 7.98719807605661e-06, "loss": 0.2393, "step": 19490 }, { "epoch": 0.35468087815347676, "grad_norm": 1.1388261318206787, "learning_rate": 7.98515270568512e-06, "loss": 0.2298, "step": 19500 }, { "epoch": 0.35486276578329906, "grad_norm": 1.1338834762573242, "learning_rate": 7.983106646396837e-06, "loss": 0.2397, "step": 19510 }, { "epoch": 0.35504465341312136, "grad_norm": 1.0522189140319824, "learning_rate": 7.981059898901811e-06, "loss": 0.2395, "step": 19520 }, { "epoch": 0.35522654104294366, "grad_norm": 1.171075463294983, "learning_rate": 7.979012463910335e-06, "loss": 0.2389, "step": 19530 }, { "epoch": 0.35540842867276595, "grad_norm": 1.2701480388641357, "learning_rate": 7.976964342132943e-06, "loss": 0.2442, "step": 19540 }, { "epoch": 0.35559031630258825, "grad_norm": 1.133359670639038, "learning_rate": 7.9749155342804e-06, "loss": 0.2459, "step": 19550 }, { "epoch": 0.35577220393241055, "grad_norm": 0.969272255897522, "learning_rate": 7.972866041063715e-06, "loss": 0.2372, "step": 19560 }, { "epoch": 0.35595409156223284, "grad_norm": 1.1068692207336426, "learning_rate": 7.970815863194134e-06, "loss": 0.2437, "step": 19570 }, { "epoch": 0.35613597919205514, "grad_norm": 1.0344756841659546, "learning_rate": 7.968765001383137e-06, "loss": 0.2401, "step": 19580 }, { "epoch": 0.35631786682187744, "grad_norm": 1.1244932413101196, "learning_rate": 7.966713456342448e-06, "loss": 0.2403, "step": 19590 }, { "epoch": 0.35649975445169974, "grad_norm": 1.192271113395691, "learning_rate": 7.964661228784021e-06, "loss": 0.2415, "step": 19600 }, { "epoch": 0.35668164208152203, "grad_norm": 0.9454160332679749, "learning_rate": 7.962608319420052e-06, "loss": 0.2393, "step": 19610 }, { "epoch": 0.35686352971134433, "grad_norm": 1.1304776668548584, "learning_rate": 7.960554728962972e-06, "loss": 0.2444, "step": 19620 }, { "epoch": 0.3570454173411666, "grad_norm": 1.024312138557434, "learning_rate": 7.958500458125445e-06, "loss": 0.2391, "step": 19630 }, { "epoch": 0.3572273049709889, "grad_norm": 1.2062351703643799, "learning_rate": 7.956445507620378e-06, "loss": 0.2413, "step": 19640 }, { "epoch": 0.3574091926008112, "grad_norm": 1.1140583753585815, "learning_rate": 7.954389878160908e-06, "loss": 0.2404, "step": 19650 }, { "epoch": 0.3575910802306335, "grad_norm": 1.1345632076263428, "learning_rate": 7.95233357046041e-06, "loss": 0.247, "step": 19660 }, { "epoch": 0.3577729678604558, "grad_norm": 1.040116310119629, "learning_rate": 7.950276585232495e-06, "loss": 0.2384, "step": 19670 }, { "epoch": 0.3579548554902781, "grad_norm": 1.0723797082901, "learning_rate": 7.948218923191006e-06, "loss": 0.2368, "step": 19680 }, { "epoch": 0.3581367431201004, "grad_norm": 1.0985362529754639, "learning_rate": 7.946160585050025e-06, "loss": 0.2393, "step": 19690 }, { "epoch": 0.3583186307499227, "grad_norm": 1.0519025325775146, "learning_rate": 7.944101571523865e-06, "loss": 0.2411, "step": 19700 }, { "epoch": 0.358500518379745, "grad_norm": 0.9480572938919067, "learning_rate": 7.942041883327077e-06, "loss": 0.2467, "step": 19710 }, { "epoch": 0.3586824060095673, "grad_norm": 1.3163642883300781, "learning_rate": 7.939981521174444e-06, "loss": 0.2412, "step": 19720 }, { "epoch": 0.3588642936393896, "grad_norm": 1.1221197843551636, "learning_rate": 7.937920485780982e-06, "loss": 0.2355, "step": 19730 }, { "epoch": 0.3590461812692119, "grad_norm": 1.084874153137207, "learning_rate": 7.935858777861943e-06, "loss": 0.2339, "step": 19740 }, { "epoch": 0.3592280688990342, "grad_norm": 1.2039934396743774, "learning_rate": 7.933796398132808e-06, "loss": 0.2427, "step": 19750 }, { "epoch": 0.3594099565288565, "grad_norm": 1.134290337562561, "learning_rate": 7.931733347309298e-06, "loss": 0.2368, "step": 19760 }, { "epoch": 0.3595918441586788, "grad_norm": 1.272295355796814, "learning_rate": 7.92966962610736e-06, "loss": 0.2431, "step": 19770 }, { "epoch": 0.3597737317885011, "grad_norm": 1.1999216079711914, "learning_rate": 7.92760523524318e-06, "loss": 0.2458, "step": 19780 }, { "epoch": 0.3599556194183234, "grad_norm": 1.1256084442138672, "learning_rate": 7.92554017543317e-06, "loss": 0.2396, "step": 19790 }, { "epoch": 0.3601375070481457, "grad_norm": 1.1529816389083862, "learning_rate": 7.92347444739398e-06, "loss": 0.25, "step": 19800 }, { "epoch": 0.360319394677968, "grad_norm": 1.039035439491272, "learning_rate": 7.921408051842486e-06, "loss": 0.2461, "step": 19810 }, { "epoch": 0.36050128230779027, "grad_norm": 1.091518521308899, "learning_rate": 7.919340989495803e-06, "loss": 0.2394, "step": 19820 }, { "epoch": 0.36068316993761257, "grad_norm": 0.9092937707901001, "learning_rate": 7.917273261071271e-06, "loss": 0.2402, "step": 19830 }, { "epoch": 0.36086505756743487, "grad_norm": 1.0514779090881348, "learning_rate": 7.915204867286465e-06, "loss": 0.2448, "step": 19840 }, { "epoch": 0.3610469451972571, "grad_norm": 1.0701093673706055, "learning_rate": 7.913135808859186e-06, "loss": 0.2451, "step": 19850 }, { "epoch": 0.3612288328270794, "grad_norm": 1.1952847242355347, "learning_rate": 7.911066086507473e-06, "loss": 0.2409, "step": 19860 }, { "epoch": 0.3614107204569017, "grad_norm": 1.1036311388015747, "learning_rate": 7.908995700949588e-06, "loss": 0.2421, "step": 19870 }, { "epoch": 0.361592608086724, "grad_norm": 0.997238278388977, "learning_rate": 7.906924652904027e-06, "loss": 0.238, "step": 19880 }, { "epoch": 0.3617744957165463, "grad_norm": 1.1556787490844727, "learning_rate": 7.904852943089519e-06, "loss": 0.2383, "step": 19890 }, { "epoch": 0.3619563833463686, "grad_norm": 1.0116026401519775, "learning_rate": 7.902780572225014e-06, "loss": 0.2416, "step": 19900 }, { "epoch": 0.3621382709761909, "grad_norm": 1.0899544954299927, "learning_rate": 7.900707541029701e-06, "loss": 0.2373, "step": 19910 }, { "epoch": 0.3623201586060132, "grad_norm": 1.0376065969467163, "learning_rate": 7.898633850222991e-06, "loss": 0.2418, "step": 19920 }, { "epoch": 0.3625020462358355, "grad_norm": 0.922985315322876, "learning_rate": 7.896559500524526e-06, "loss": 0.2407, "step": 19930 }, { "epoch": 0.3626839338656578, "grad_norm": 1.1156353950500488, "learning_rate": 7.894484492654181e-06, "loss": 0.2454, "step": 19940 }, { "epoch": 0.3628658214954801, "grad_norm": 1.1574690341949463, "learning_rate": 7.892408827332051e-06, "loss": 0.2404, "step": 19950 }, { "epoch": 0.3630477091253024, "grad_norm": 1.193482518196106, "learning_rate": 7.890332505278467e-06, "loss": 0.2432, "step": 19960 }, { "epoch": 0.3632295967551247, "grad_norm": 1.0958505868911743, "learning_rate": 7.888255527213985e-06, "loss": 0.2387, "step": 19970 }, { "epoch": 0.36341148438494697, "grad_norm": 1.1883647441864014, "learning_rate": 7.886177893859385e-06, "loss": 0.2384, "step": 19980 }, { "epoch": 0.36359337201476927, "grad_norm": 1.0876684188842773, "learning_rate": 7.884099605935681e-06, "loss": 0.2374, "step": 19990 }, { "epoch": 0.36377525964459156, "grad_norm": 1.3195700645446777, "learning_rate": 7.882020664164109e-06, "loss": 0.2425, "step": 20000 }, { "epoch": 0.36395714727441386, "grad_norm": 0.9515960812568665, "learning_rate": 7.879941069266136e-06, "loss": 0.2467, "step": 20010 }, { "epoch": 0.36413903490423616, "grad_norm": 1.122153878211975, "learning_rate": 7.877860821963452e-06, "loss": 0.2382, "step": 20020 }, { "epoch": 0.36432092253405846, "grad_norm": 1.0241906642913818, "learning_rate": 7.875779922977975e-06, "loss": 0.2381, "step": 20030 }, { "epoch": 0.36450281016388075, "grad_norm": 1.0193246603012085, "learning_rate": 7.873698373031848e-06, "loss": 0.2354, "step": 20040 }, { "epoch": 0.36468469779370305, "grad_norm": 1.042682409286499, "learning_rate": 7.871616172847442e-06, "loss": 0.2377, "step": 20050 }, { "epoch": 0.36486658542352535, "grad_norm": 1.129643440246582, "learning_rate": 7.869533323147355e-06, "loss": 0.2404, "step": 20060 }, { "epoch": 0.36504847305334764, "grad_norm": 1.0649287700653076, "learning_rate": 7.867449824654404e-06, "loss": 0.2424, "step": 20070 }, { "epoch": 0.36523036068316994, "grad_norm": 1.155824899673462, "learning_rate": 7.865365678091638e-06, "loss": 0.2392, "step": 20080 }, { "epoch": 0.36541224831299224, "grad_norm": 1.0795493125915527, "learning_rate": 7.863280884182327e-06, "loss": 0.2479, "step": 20090 }, { "epoch": 0.36559413594281454, "grad_norm": 1.1557211875915527, "learning_rate": 7.861195443649966e-06, "loss": 0.2438, "step": 20100 }, { "epoch": 0.36577602357263683, "grad_norm": 1.1377887725830078, "learning_rate": 7.859109357218277e-06, "loss": 0.2427, "step": 20110 }, { "epoch": 0.36595791120245913, "grad_norm": 1.1049022674560547, "learning_rate": 7.857022625611204e-06, "loss": 0.2406, "step": 20120 }, { "epoch": 0.3661397988322814, "grad_norm": 1.0132242441177368, "learning_rate": 7.854935249552914e-06, "loss": 0.2423, "step": 20130 }, { "epoch": 0.3663216864621037, "grad_norm": 1.0356265306472778, "learning_rate": 7.852847229767798e-06, "loss": 0.238, "step": 20140 }, { "epoch": 0.366503574091926, "grad_norm": 1.0557317733764648, "learning_rate": 7.850758566980473e-06, "loss": 0.2405, "step": 20150 }, { "epoch": 0.3666854617217483, "grad_norm": 1.1048452854156494, "learning_rate": 7.848669261915779e-06, "loss": 0.2418, "step": 20160 }, { "epoch": 0.3668673493515706, "grad_norm": 1.2129817008972168, "learning_rate": 7.846579315298773e-06, "loss": 0.2412, "step": 20170 }, { "epoch": 0.3670492369813929, "grad_norm": 1.076357364654541, "learning_rate": 7.844488727854742e-06, "loss": 0.2398, "step": 20180 }, { "epoch": 0.3672311246112152, "grad_norm": 1.0466338396072388, "learning_rate": 7.842397500309192e-06, "loss": 0.2401, "step": 20190 }, { "epoch": 0.3674130122410375, "grad_norm": 1.1943999528884888, "learning_rate": 7.840305633387848e-06, "loss": 0.2391, "step": 20200 }, { "epoch": 0.3675948998708598, "grad_norm": 1.11336088180542, "learning_rate": 7.838213127816666e-06, "loss": 0.244, "step": 20210 }, { "epoch": 0.3677767875006821, "grad_norm": 1.093136191368103, "learning_rate": 7.836119984321816e-06, "loss": 0.2419, "step": 20220 }, { "epoch": 0.3679586751305044, "grad_norm": 1.0068036317825317, "learning_rate": 7.83402620362969e-06, "loss": 0.2426, "step": 20230 }, { "epoch": 0.3681405627603267, "grad_norm": 1.1445446014404297, "learning_rate": 7.831931786466902e-06, "loss": 0.2397, "step": 20240 }, { "epoch": 0.368322450390149, "grad_norm": 1.1244648694992065, "learning_rate": 7.829836733560288e-06, "loss": 0.2421, "step": 20250 }, { "epoch": 0.3685043380199713, "grad_norm": 1.2111468315124512, "learning_rate": 7.827741045636907e-06, "loss": 0.2383, "step": 20260 }, { "epoch": 0.36868622564979353, "grad_norm": 1.016542911529541, "learning_rate": 7.82564472342403e-06, "loss": 0.2439, "step": 20270 }, { "epoch": 0.36886811327961583, "grad_norm": 1.0132765769958496, "learning_rate": 7.823547767649156e-06, "loss": 0.2382, "step": 20280 }, { "epoch": 0.3690500009094381, "grad_norm": 1.0569450855255127, "learning_rate": 7.821450179040001e-06, "loss": 0.2318, "step": 20290 }, { "epoch": 0.3692318885392604, "grad_norm": 1.0725332498550415, "learning_rate": 7.819351958324504e-06, "loss": 0.24, "step": 20300 }, { "epoch": 0.3694137761690827, "grad_norm": 0.9774289131164551, "learning_rate": 7.817253106230817e-06, "loss": 0.2403, "step": 20310 }, { "epoch": 0.369595663798905, "grad_norm": 1.1501506567001343, "learning_rate": 7.815153623487314e-06, "loss": 0.2392, "step": 20320 }, { "epoch": 0.3697775514287273, "grad_norm": 1.1429778337478638, "learning_rate": 7.81305351082259e-06, "loss": 0.2444, "step": 20330 }, { "epoch": 0.3699594390585496, "grad_norm": 1.0506048202514648, "learning_rate": 7.810952768965457e-06, "loss": 0.2418, "step": 20340 }, { "epoch": 0.3701413266883719, "grad_norm": 1.0567690134048462, "learning_rate": 7.808851398644944e-06, "loss": 0.2375, "step": 20350 }, { "epoch": 0.3703232143181942, "grad_norm": 1.240209937095642, "learning_rate": 7.8067494005903e-06, "loss": 0.2406, "step": 20360 }, { "epoch": 0.3705051019480165, "grad_norm": 1.205761432647705, "learning_rate": 7.80464677553099e-06, "loss": 0.2398, "step": 20370 }, { "epoch": 0.3706869895778388, "grad_norm": 0.9883845448493958, "learning_rate": 7.8025435241967e-06, "loss": 0.2391, "step": 20380 }, { "epoch": 0.3708688772076611, "grad_norm": 1.1043930053710938, "learning_rate": 7.800439647317331e-06, "loss": 0.237, "step": 20390 }, { "epoch": 0.3710507648374834, "grad_norm": 0.9645054936408997, "learning_rate": 7.798335145622999e-06, "loss": 0.2382, "step": 20400 }, { "epoch": 0.3712326524673057, "grad_norm": 1.1453968286514282, "learning_rate": 7.796230019844041e-06, "loss": 0.2364, "step": 20410 }, { "epoch": 0.371414540097128, "grad_norm": 1.1825060844421387, "learning_rate": 7.794124270711006e-06, "loss": 0.2334, "step": 20420 }, { "epoch": 0.3715964277269503, "grad_norm": 0.9973169565200806, "learning_rate": 7.792017898954665e-06, "loss": 0.2385, "step": 20430 }, { "epoch": 0.3717783153567726, "grad_norm": 1.257887363433838, "learning_rate": 7.789910905306002e-06, "loss": 0.2402, "step": 20440 }, { "epoch": 0.3719602029865949, "grad_norm": 1.2057524919509888, "learning_rate": 7.787803290496214e-06, "loss": 0.2397, "step": 20450 }, { "epoch": 0.3721420906164172, "grad_norm": 1.0857775211334229, "learning_rate": 7.78569505525672e-06, "loss": 0.2412, "step": 20460 }, { "epoch": 0.3723239782462395, "grad_norm": 1.0445079803466797, "learning_rate": 7.783586200319144e-06, "loss": 0.2416, "step": 20470 }, { "epoch": 0.37250586587606177, "grad_norm": 1.1763966083526611, "learning_rate": 7.78147672641534e-06, "loss": 0.244, "step": 20480 }, { "epoch": 0.37268775350588407, "grad_norm": 1.0350455045700073, "learning_rate": 7.779366634277364e-06, "loss": 0.2388, "step": 20490 }, { "epoch": 0.37286964113570636, "grad_norm": 1.0422688722610474, "learning_rate": 7.777255924637494e-06, "loss": 0.2365, "step": 20500 }, { "epoch": 0.37305152876552866, "grad_norm": 1.1451988220214844, "learning_rate": 7.775144598228217e-06, "loss": 0.2435, "step": 20510 }, { "epoch": 0.37323341639535096, "grad_norm": 1.1446171998977661, "learning_rate": 7.773032655782234e-06, "loss": 0.2417, "step": 20520 }, { "epoch": 0.37341530402517326, "grad_norm": 1.0598891973495483, "learning_rate": 7.770920098032468e-06, "loss": 0.2415, "step": 20530 }, { "epoch": 0.37359719165499555, "grad_norm": 0.9509167075157166, "learning_rate": 7.768806925712046e-06, "loss": 0.237, "step": 20540 }, { "epoch": 0.37377907928481785, "grad_norm": 1.0998649597167969, "learning_rate": 7.766693139554313e-06, "loss": 0.241, "step": 20550 }, { "epoch": 0.37396096691464015, "grad_norm": 1.0373883247375488, "learning_rate": 7.764578740292827e-06, "loss": 0.2451, "step": 20560 }, { "epoch": 0.37414285454446244, "grad_norm": 1.1122106313705444, "learning_rate": 7.762463728661356e-06, "loss": 0.2398, "step": 20570 }, { "epoch": 0.37432474217428474, "grad_norm": 0.9298462867736816, "learning_rate": 7.760348105393883e-06, "loss": 0.2391, "step": 20580 }, { "epoch": 0.37450662980410704, "grad_norm": 1.0754690170288086, "learning_rate": 7.758231871224602e-06, "loss": 0.2432, "step": 20590 }, { "epoch": 0.37468851743392934, "grad_norm": 1.0495531558990479, "learning_rate": 7.756115026887922e-06, "loss": 0.2448, "step": 20600 }, { "epoch": 0.37487040506375163, "grad_norm": 1.0250312089920044, "learning_rate": 7.75399757311846e-06, "loss": 0.2389, "step": 20610 }, { "epoch": 0.37505229269357393, "grad_norm": 0.99468594789505, "learning_rate": 7.751879510651045e-06, "loss": 0.2353, "step": 20620 }, { "epoch": 0.3752341803233962, "grad_norm": 1.1579947471618652, "learning_rate": 7.749760840220717e-06, "loss": 0.2325, "step": 20630 }, { "epoch": 0.3754160679532185, "grad_norm": 1.1230144500732422, "learning_rate": 7.747641562562733e-06, "loss": 0.2349, "step": 20640 }, { "epoch": 0.3755979555830408, "grad_norm": 1.1359121799468994, "learning_rate": 7.745521678412551e-06, "loss": 0.244, "step": 20650 }, { "epoch": 0.3757798432128631, "grad_norm": 1.0038529634475708, "learning_rate": 7.743401188505846e-06, "loss": 0.237, "step": 20660 }, { "epoch": 0.3759617308426854, "grad_norm": 1.1018023490905762, "learning_rate": 7.741280093578501e-06, "loss": 0.2392, "step": 20670 }, { "epoch": 0.37614361847250766, "grad_norm": 0.9998648166656494, "learning_rate": 7.739158394366611e-06, "loss": 0.24, "step": 20680 }, { "epoch": 0.37632550610232995, "grad_norm": 1.0452847480773926, "learning_rate": 7.737036091606477e-06, "loss": 0.2397, "step": 20690 }, { "epoch": 0.37650739373215225, "grad_norm": 0.9988589882850647, "learning_rate": 7.734913186034612e-06, "loss": 0.2401, "step": 20700 }, { "epoch": 0.37668928136197455, "grad_norm": 1.102146863937378, "learning_rate": 7.73278967838774e-06, "loss": 0.2388, "step": 20710 }, { "epoch": 0.37687116899179685, "grad_norm": 1.1516517400741577, "learning_rate": 7.73066556940279e-06, "loss": 0.2409, "step": 20720 }, { "epoch": 0.37705305662161914, "grad_norm": 1.1193506717681885, "learning_rate": 7.7285408598169e-06, "loss": 0.2435, "step": 20730 }, { "epoch": 0.37723494425144144, "grad_norm": 1.0247642993927002, "learning_rate": 7.72641555036742e-06, "loss": 0.238, "step": 20740 }, { "epoch": 0.37741683188126374, "grad_norm": 1.110872507095337, "learning_rate": 7.724289641791907e-06, "loss": 0.2373, "step": 20750 }, { "epoch": 0.37759871951108603, "grad_norm": 1.133522391319275, "learning_rate": 7.722163134828122e-06, "loss": 0.2406, "step": 20760 }, { "epoch": 0.37778060714090833, "grad_norm": 1.2115174531936646, "learning_rate": 7.720036030214039e-06, "loss": 0.2374, "step": 20770 }, { "epoch": 0.37796249477073063, "grad_norm": 0.9982824325561523, "learning_rate": 7.717908328687836e-06, "loss": 0.2412, "step": 20780 }, { "epoch": 0.3781443824005529, "grad_norm": 1.0521788597106934, "learning_rate": 7.7157800309879e-06, "loss": 0.2368, "step": 20790 }, { "epoch": 0.3783262700303752, "grad_norm": 0.9637540578842163, "learning_rate": 7.713651137852821e-06, "loss": 0.243, "step": 20800 }, { "epoch": 0.3785081576601975, "grad_norm": 1.0706913471221924, "learning_rate": 7.711521650021405e-06, "loss": 0.2357, "step": 20810 }, { "epoch": 0.3786900452900198, "grad_norm": 0.9088761210441589, "learning_rate": 7.70939156823265e-06, "loss": 0.235, "step": 20820 }, { "epoch": 0.3788719329198421, "grad_norm": 1.1147007942199707, "learning_rate": 7.707260893225776e-06, "loss": 0.2361, "step": 20830 }, { "epoch": 0.3790538205496644, "grad_norm": 1.03704833984375, "learning_rate": 7.705129625740194e-06, "loss": 0.244, "step": 20840 }, { "epoch": 0.3792357081794867, "grad_norm": 1.0627440214157104, "learning_rate": 7.702997766515534e-06, "loss": 0.2394, "step": 20850 }, { "epoch": 0.379417595809309, "grad_norm": 1.1675723791122437, "learning_rate": 7.70086531629162e-06, "loss": 0.24, "step": 20860 }, { "epoch": 0.3795994834391313, "grad_norm": 1.1579525470733643, "learning_rate": 7.69873227580849e-06, "loss": 0.2441, "step": 20870 }, { "epoch": 0.3797813710689536, "grad_norm": 1.1091647148132324, "learning_rate": 7.69659864580638e-06, "loss": 0.2381, "step": 20880 }, { "epoch": 0.3799632586987759, "grad_norm": 1.0669488906860352, "learning_rate": 7.694464427025737e-06, "loss": 0.2382, "step": 20890 }, { "epoch": 0.3801451463285982, "grad_norm": 0.9688666462898254, "learning_rate": 7.692329620207206e-06, "loss": 0.2415, "step": 20900 }, { "epoch": 0.3803270339584205, "grad_norm": 1.0427874326705933, "learning_rate": 7.690194226091638e-06, "loss": 0.2422, "step": 20910 }, { "epoch": 0.3805089215882428, "grad_norm": 1.0380635261535645, "learning_rate": 7.688058245420092e-06, "loss": 0.2441, "step": 20920 }, { "epoch": 0.3806908092180651, "grad_norm": 1.1444907188415527, "learning_rate": 7.685921678933825e-06, "loss": 0.2405, "step": 20930 }, { "epoch": 0.3808726968478874, "grad_norm": 1.1620243787765503, "learning_rate": 7.683784527374302e-06, "loss": 0.2414, "step": 20940 }, { "epoch": 0.3810545844777097, "grad_norm": 1.201034665107727, "learning_rate": 7.681646791483187e-06, "loss": 0.2428, "step": 20950 }, { "epoch": 0.381236472107532, "grad_norm": 0.9535667300224304, "learning_rate": 7.679508472002346e-06, "loss": 0.2406, "step": 20960 }, { "epoch": 0.3814183597373543, "grad_norm": 1.0046578645706177, "learning_rate": 7.677369569673856e-06, "loss": 0.2405, "step": 20970 }, { "epoch": 0.38160024736717657, "grad_norm": 1.0309101343154907, "learning_rate": 7.675230085239986e-06, "loss": 0.2379, "step": 20980 }, { "epoch": 0.38178213499699887, "grad_norm": 1.0252187252044678, "learning_rate": 7.67309001944321e-06, "loss": 0.239, "step": 20990 }, { "epoch": 0.38196402262682116, "grad_norm": 1.009677767753601, "learning_rate": 7.670949373026207e-06, "loss": 0.238, "step": 21000 }, { "epoch": 0.38214591025664346, "grad_norm": 1.0273197889328003, "learning_rate": 7.668808146731858e-06, "loss": 0.2357, "step": 21010 }, { "epoch": 0.38232779788646576, "grad_norm": 1.0627318620681763, "learning_rate": 7.666666341303238e-06, "loss": 0.2443, "step": 21020 }, { "epoch": 0.38250968551628806, "grad_norm": 1.047061800956726, "learning_rate": 7.664523957483632e-06, "loss": 0.243, "step": 21030 }, { "epoch": 0.38269157314611035, "grad_norm": 1.3039597272872925, "learning_rate": 7.662380996016519e-06, "loss": 0.2391, "step": 21040 }, { "epoch": 0.38287346077593265, "grad_norm": 0.9711835384368896, "learning_rate": 7.66023745764558e-06, "loss": 0.2362, "step": 21050 }, { "epoch": 0.38305534840575495, "grad_norm": 1.2898297309875488, "learning_rate": 7.658093343114698e-06, "loss": 0.2461, "step": 21060 }, { "epoch": 0.38323723603557724, "grad_norm": 1.0630356073379517, "learning_rate": 7.655948653167958e-06, "loss": 0.2452, "step": 21070 }, { "epoch": 0.38341912366539954, "grad_norm": 0.9959006309509277, "learning_rate": 7.653803388549638e-06, "loss": 0.2389, "step": 21080 }, { "epoch": 0.38360101129522184, "grad_norm": 1.0544307231903076, "learning_rate": 7.651657550004221e-06, "loss": 0.241, "step": 21090 }, { "epoch": 0.3837828989250441, "grad_norm": 1.0523689985275269, "learning_rate": 7.649511138276387e-06, "loss": 0.238, "step": 21100 }, { "epoch": 0.3839647865548664, "grad_norm": 1.1002427339553833, "learning_rate": 7.647364154111015e-06, "loss": 0.2382, "step": 21110 }, { "epoch": 0.3841466741846887, "grad_norm": 1.0448845624923706, "learning_rate": 7.645216598253184e-06, "loss": 0.235, "step": 21120 }, { "epoch": 0.38432856181451097, "grad_norm": 1.088006615638733, "learning_rate": 7.643068471448168e-06, "loss": 0.2391, "step": 21130 }, { "epoch": 0.38451044944433327, "grad_norm": 1.0280585289001465, "learning_rate": 7.640919774441441e-06, "loss": 0.2345, "step": 21140 }, { "epoch": 0.38469233707415557, "grad_norm": 1.0401118993759155, "learning_rate": 7.638770507978679e-06, "loss": 0.2372, "step": 21150 }, { "epoch": 0.38487422470397786, "grad_norm": 1.1771221160888672, "learning_rate": 7.63662067280575e-06, "loss": 0.2412, "step": 21160 }, { "epoch": 0.38505611233380016, "grad_norm": 1.1181117296218872, "learning_rate": 7.634470269668721e-06, "loss": 0.2488, "step": 21170 }, { "epoch": 0.38523799996362246, "grad_norm": 1.2087643146514893, "learning_rate": 7.632319299313859e-06, "loss": 0.2329, "step": 21180 }, { "epoch": 0.38541988759344475, "grad_norm": 0.9765835404396057, "learning_rate": 7.630167762487619e-06, "loss": 0.2344, "step": 21190 }, { "epoch": 0.38560177522326705, "grad_norm": 0.999513566493988, "learning_rate": 7.628015659936666e-06, "loss": 0.2387, "step": 21200 }, { "epoch": 0.38578366285308935, "grad_norm": 1.0579007863998413, "learning_rate": 7.6258629924078495e-06, "loss": 0.236, "step": 21210 }, { "epoch": 0.38596555048291165, "grad_norm": 1.026792049407959, "learning_rate": 7.623709760648223e-06, "loss": 0.2377, "step": 21220 }, { "epoch": 0.38614743811273394, "grad_norm": 1.1539820432662964, "learning_rate": 7.621555965405031e-06, "loss": 0.2389, "step": 21230 }, { "epoch": 0.38632932574255624, "grad_norm": 1.0305315256118774, "learning_rate": 7.6194016074257135e-06, "loss": 0.2407, "step": 21240 }, { "epoch": 0.38651121337237854, "grad_norm": 1.0835392475128174, "learning_rate": 7.6172466874579096e-06, "loss": 0.2389, "step": 21250 }, { "epoch": 0.38669310100220083, "grad_norm": 1.1253561973571777, "learning_rate": 7.61509120624945e-06, "loss": 0.2388, "step": 21260 }, { "epoch": 0.38687498863202313, "grad_norm": 1.027140498161316, "learning_rate": 7.612935164548362e-06, "loss": 0.2401, "step": 21270 }, { "epoch": 0.38705687626184543, "grad_norm": 1.0267332792282104, "learning_rate": 7.610778563102867e-06, "loss": 0.2437, "step": 21280 }, { "epoch": 0.3872387638916677, "grad_norm": 1.1524869203567505, "learning_rate": 7.608621402661379e-06, "loss": 0.2414, "step": 21290 }, { "epoch": 0.38742065152149, "grad_norm": 0.9468303918838501, "learning_rate": 7.60646368397251e-06, "loss": 0.2409, "step": 21300 }, { "epoch": 0.3876025391513123, "grad_norm": 1.0597643852233887, "learning_rate": 7.604305407785062e-06, "loss": 0.2398, "step": 21310 }, { "epoch": 0.3877844267811346, "grad_norm": 1.0989406108856201, "learning_rate": 7.602146574848032e-06, "loss": 0.238, "step": 21320 }, { "epoch": 0.3879663144109569, "grad_norm": 1.0452088117599487, "learning_rate": 7.5999871859106075e-06, "loss": 0.2432, "step": 21330 }, { "epoch": 0.3881482020407792, "grad_norm": 1.1448529958724976, "learning_rate": 7.597827241722175e-06, "loss": 0.2434, "step": 21340 }, { "epoch": 0.3883300896706015, "grad_norm": 1.2026896476745605, "learning_rate": 7.595666743032306e-06, "loss": 0.2419, "step": 21350 }, { "epoch": 0.3885119773004238, "grad_norm": 1.1103978157043457, "learning_rate": 7.593505690590771e-06, "loss": 0.2391, "step": 21360 }, { "epoch": 0.3886938649302461, "grad_norm": 1.1986654996871948, "learning_rate": 7.5913440851475305e-06, "loss": 0.2435, "step": 21370 }, { "epoch": 0.3888757525600684, "grad_norm": 0.9713031053543091, "learning_rate": 7.589181927452734e-06, "loss": 0.2381, "step": 21380 }, { "epoch": 0.3890576401898907, "grad_norm": 1.20008385181427, "learning_rate": 7.587019218256729e-06, "loss": 0.2305, "step": 21390 }, { "epoch": 0.389239527819713, "grad_norm": 1.167276382446289, "learning_rate": 7.584855958310048e-06, "loss": 0.2462, "step": 21400 }, { "epoch": 0.3894214154495353, "grad_norm": 1.101544976234436, "learning_rate": 7.582692148363416e-06, "loss": 0.2412, "step": 21410 }, { "epoch": 0.3896033030793576, "grad_norm": 0.9762274026870728, "learning_rate": 7.580527789167753e-06, "loss": 0.2381, "step": 21420 }, { "epoch": 0.3897851907091799, "grad_norm": 1.067799687385559, "learning_rate": 7.578362881474164e-06, "loss": 0.2446, "step": 21430 }, { "epoch": 0.3899670783390022, "grad_norm": 1.1635833978652954, "learning_rate": 7.576197426033948e-06, "loss": 0.2425, "step": 21440 }, { "epoch": 0.3901489659688245, "grad_norm": 1.0069206953048706, "learning_rate": 7.574031423598595e-06, "loss": 0.2413, "step": 21450 }, { "epoch": 0.3903308535986468, "grad_norm": 1.025131106376648, "learning_rate": 7.57186487491978e-06, "loss": 0.2402, "step": 21460 }, { "epoch": 0.3905127412284691, "grad_norm": 1.1373223066329956, "learning_rate": 7.569697780749372e-06, "loss": 0.2448, "step": 21470 }, { "epoch": 0.39069462885829137, "grad_norm": 1.0666416883468628, "learning_rate": 7.567530141839427e-06, "loss": 0.2377, "step": 21480 }, { "epoch": 0.39087651648811367, "grad_norm": 0.9913526177406311, "learning_rate": 7.565361958942194e-06, "loss": 0.2398, "step": 21490 }, { "epoch": 0.39105840411793596, "grad_norm": 1.0406688451766968, "learning_rate": 7.563193232810102e-06, "loss": 0.2357, "step": 21500 }, { "epoch": 0.39124029174775826, "grad_norm": 1.0882186889648438, "learning_rate": 7.5610239641957775e-06, "loss": 0.2377, "step": 21510 }, { "epoch": 0.3914221793775805, "grad_norm": 1.1328761577606201, "learning_rate": 7.558854153852032e-06, "loss": 0.2497, "step": 21520 }, { "epoch": 0.3916040670074028, "grad_norm": 1.237585186958313, "learning_rate": 7.556683802531864e-06, "loss": 0.2372, "step": 21530 }, { "epoch": 0.3917859546372251, "grad_norm": 1.1703097820281982, "learning_rate": 7.554512910988463e-06, "loss": 0.2379, "step": 21540 }, { "epoch": 0.3919678422670474, "grad_norm": 1.0261156558990479, "learning_rate": 7.552341479975199e-06, "loss": 0.2388, "step": 21550 }, { "epoch": 0.3921497298968697, "grad_norm": 0.9599685668945312, "learning_rate": 7.550169510245638e-06, "loss": 0.2381, "step": 21560 }, { "epoch": 0.392331617526692, "grad_norm": 1.1213761568069458, "learning_rate": 7.547997002553526e-06, "loss": 0.2419, "step": 21570 }, { "epoch": 0.3925135051565143, "grad_norm": 1.173598289489746, "learning_rate": 7.5458239576528e-06, "loss": 0.2417, "step": 21580 }, { "epoch": 0.3926953927863366, "grad_norm": 1.1113269329071045, "learning_rate": 7.543650376297583e-06, "loss": 0.2353, "step": 21590 }, { "epoch": 0.3928772804161589, "grad_norm": 1.2211987972259521, "learning_rate": 7.541476259242181e-06, "loss": 0.2333, "step": 21600 }, { "epoch": 0.3930591680459812, "grad_norm": 1.0685701370239258, "learning_rate": 7.539301607241089e-06, "loss": 0.2372, "step": 21610 }, { "epoch": 0.3932410556758035, "grad_norm": 1.1269465684890747, "learning_rate": 7.537126421048985e-06, "loss": 0.2395, "step": 21620 }, { "epoch": 0.39342294330562577, "grad_norm": 1.3198050260543823, "learning_rate": 7.534950701420735e-06, "loss": 0.2402, "step": 21630 }, { "epoch": 0.39360483093544807, "grad_norm": 1.1072196960449219, "learning_rate": 7.532774449111392e-06, "loss": 0.2399, "step": 21640 }, { "epoch": 0.39378671856527037, "grad_norm": 1.108518362045288, "learning_rate": 7.530597664876186e-06, "loss": 0.2374, "step": 21650 }, { "epoch": 0.39396860619509266, "grad_norm": 1.1146671772003174, "learning_rate": 7.528420349470541e-06, "loss": 0.2417, "step": 21660 }, { "epoch": 0.39415049382491496, "grad_norm": 1.091384768486023, "learning_rate": 7.526242503650058e-06, "loss": 0.2272, "step": 21670 }, { "epoch": 0.39433238145473726, "grad_norm": 1.0724842548370361, "learning_rate": 7.524064128170524e-06, "loss": 0.2318, "step": 21680 }, { "epoch": 0.39451426908455955, "grad_norm": 1.0652724504470825, "learning_rate": 7.521885223787914e-06, "loss": 0.2374, "step": 21690 }, { "epoch": 0.39469615671438185, "grad_norm": 0.9754598140716553, "learning_rate": 7.519705791258381e-06, "loss": 0.241, "step": 21700 }, { "epoch": 0.39487804434420415, "grad_norm": 1.124104619026184, "learning_rate": 7.5175258313382675e-06, "loss": 0.2355, "step": 21710 }, { "epoch": 0.39505993197402645, "grad_norm": 1.1662229299545288, "learning_rate": 7.51534534478409e-06, "loss": 0.2381, "step": 21720 }, { "epoch": 0.39524181960384874, "grad_norm": 1.0138437747955322, "learning_rate": 7.513164332352555e-06, "loss": 0.2372, "step": 21730 }, { "epoch": 0.39542370723367104, "grad_norm": 1.092553734779358, "learning_rate": 7.510982794800551e-06, "loss": 0.2369, "step": 21740 }, { "epoch": 0.39560559486349334, "grad_norm": 1.189460277557373, "learning_rate": 7.508800732885144e-06, "loss": 0.2349, "step": 21750 }, { "epoch": 0.39578748249331563, "grad_norm": 1.0456101894378662, "learning_rate": 7.506618147363589e-06, "loss": 0.2424, "step": 21760 }, { "epoch": 0.39596937012313793, "grad_norm": 1.1122850179672241, "learning_rate": 7.5044350389933155e-06, "loss": 0.245, "step": 21770 }, { "epoch": 0.39615125775296023, "grad_norm": 1.1821974515914917, "learning_rate": 7.5022514085319385e-06, "loss": 0.2378, "step": 21780 }, { "epoch": 0.3963331453827825, "grad_norm": 1.0697275400161743, "learning_rate": 7.500067256737256e-06, "loss": 0.2479, "step": 21790 }, { "epoch": 0.3965150330126048, "grad_norm": 1.091057300567627, "learning_rate": 7.497882584367239e-06, "loss": 0.239, "step": 21800 }, { "epoch": 0.3966969206424271, "grad_norm": 1.2560312747955322, "learning_rate": 7.49569739218005e-06, "loss": 0.2391, "step": 21810 }, { "epoch": 0.3968788082722494, "grad_norm": 1.0622342824935913, "learning_rate": 7.493511680934024e-06, "loss": 0.2404, "step": 21820 }, { "epoch": 0.3970606959020717, "grad_norm": 1.1440685987472534, "learning_rate": 7.491325451387678e-06, "loss": 0.2428, "step": 21830 }, { "epoch": 0.397242583531894, "grad_norm": 1.1160210371017456, "learning_rate": 7.489138704299711e-06, "loss": 0.2391, "step": 21840 }, { "epoch": 0.3974244711617163, "grad_norm": 1.1469627618789673, "learning_rate": 7.486951440428998e-06, "loss": 0.239, "step": 21850 }, { "epoch": 0.3976063587915386, "grad_norm": 1.291122555732727, "learning_rate": 7.484763660534599e-06, "loss": 0.2405, "step": 21860 }, { "epoch": 0.3977882464213609, "grad_norm": 1.1403888463974, "learning_rate": 7.482575365375746e-06, "loss": 0.2399, "step": 21870 }, { "epoch": 0.3979701340511832, "grad_norm": 1.1028801202774048, "learning_rate": 7.4803865557118524e-06, "loss": 0.243, "step": 21880 }, { "epoch": 0.3981520216810055, "grad_norm": 1.069035530090332, "learning_rate": 7.478197232302516e-06, "loss": 0.2369, "step": 21890 }, { "epoch": 0.3983339093108278, "grad_norm": 1.2657936811447144, "learning_rate": 7.476007395907504e-06, "loss": 0.2413, "step": 21900 }, { "epoch": 0.3985157969406501, "grad_norm": 1.079011082649231, "learning_rate": 7.4738170472867656e-06, "loss": 0.2395, "step": 21910 }, { "epoch": 0.3986976845704724, "grad_norm": 1.0259153842926025, "learning_rate": 7.471626187200428e-06, "loss": 0.2399, "step": 21920 }, { "epoch": 0.3988795722002947, "grad_norm": 1.1199934482574463, "learning_rate": 7.4694348164087985e-06, "loss": 0.2408, "step": 21930 }, { "epoch": 0.3990614598301169, "grad_norm": 1.0959848165512085, "learning_rate": 7.467242935672355e-06, "loss": 0.2395, "step": 21940 }, { "epoch": 0.3992433474599392, "grad_norm": 0.9797034859657288, "learning_rate": 7.4650505457517565e-06, "loss": 0.2388, "step": 21950 }, { "epoch": 0.3994252350897615, "grad_norm": 1.1165918111801147, "learning_rate": 7.462857647407841e-06, "loss": 0.2389, "step": 21960 }, { "epoch": 0.3996071227195838, "grad_norm": 0.9834339618682861, "learning_rate": 7.460664241401617e-06, "loss": 0.2367, "step": 21970 }, { "epoch": 0.3997890103494061, "grad_norm": 1.0174416303634644, "learning_rate": 7.458470328494273e-06, "loss": 0.2366, "step": 21980 }, { "epoch": 0.3999708979792284, "grad_norm": 1.1984021663665771, "learning_rate": 7.456275909447176e-06, "loss": 0.2374, "step": 21990 }, { "epoch": 0.4001527856090507, "grad_norm": 1.0719200372695923, "learning_rate": 7.454080985021861e-06, "loss": 0.2339, "step": 22000 }, { "epoch": 0.400334673238873, "grad_norm": 1.2304654121398926, "learning_rate": 7.451885555980046e-06, "loss": 0.243, "step": 22010 }, { "epoch": 0.4005165608686953, "grad_norm": 1.0295624732971191, "learning_rate": 7.44968962308362e-06, "loss": 0.2335, "step": 22020 }, { "epoch": 0.4006984484985176, "grad_norm": 1.1502093076705933, "learning_rate": 7.447493187094648e-06, "loss": 0.2454, "step": 22030 }, { "epoch": 0.4008803361283399, "grad_norm": 1.0693467855453491, "learning_rate": 7.44529624877537e-06, "loss": 0.2359, "step": 22040 }, { "epoch": 0.4010622237581622, "grad_norm": 1.0670827627182007, "learning_rate": 7.443098808888197e-06, "loss": 0.2335, "step": 22050 }, { "epoch": 0.4012441113879845, "grad_norm": 0.9374266266822815, "learning_rate": 7.440900868195719e-06, "loss": 0.2351, "step": 22060 }, { "epoch": 0.4014259990178068, "grad_norm": 1.0353474617004395, "learning_rate": 7.438702427460698e-06, "loss": 0.2339, "step": 22070 }, { "epoch": 0.4016078866476291, "grad_norm": 1.112261176109314, "learning_rate": 7.436503487446068e-06, "loss": 0.2409, "step": 22080 }, { "epoch": 0.4017897742774514, "grad_norm": 0.9188585877418518, "learning_rate": 7.434304048914939e-06, "loss": 0.2364, "step": 22090 }, { "epoch": 0.4019716619072737, "grad_norm": 1.1033028364181519, "learning_rate": 7.432104112630591e-06, "loss": 0.2367, "step": 22100 }, { "epoch": 0.402153549537096, "grad_norm": 1.0888268947601318, "learning_rate": 7.429903679356477e-06, "loss": 0.2393, "step": 22110 }, { "epoch": 0.4023354371669183, "grad_norm": 1.0754263401031494, "learning_rate": 7.4277027498562255e-06, "loss": 0.2396, "step": 22120 }, { "epoch": 0.40251732479674057, "grad_norm": 0.97438645362854, "learning_rate": 7.4255013248936355e-06, "loss": 0.2361, "step": 22130 }, { "epoch": 0.40269921242656287, "grad_norm": 1.065803050994873, "learning_rate": 7.4232994052326766e-06, "loss": 0.2385, "step": 22140 }, { "epoch": 0.40288110005638517, "grad_norm": 0.9771719574928284, "learning_rate": 7.421096991637493e-06, "loss": 0.2376, "step": 22150 }, { "epoch": 0.40306298768620746, "grad_norm": 1.044490098953247, "learning_rate": 7.418894084872397e-06, "loss": 0.2423, "step": 22160 }, { "epoch": 0.40324487531602976, "grad_norm": 1.1814759969711304, "learning_rate": 7.416690685701873e-06, "loss": 0.2396, "step": 22170 }, { "epoch": 0.40342676294585206, "grad_norm": 1.0750343799591064, "learning_rate": 7.414486794890577e-06, "loss": 0.24, "step": 22180 }, { "epoch": 0.40360865057567435, "grad_norm": 1.0620170831680298, "learning_rate": 7.4122824132033385e-06, "loss": 0.2402, "step": 22190 }, { "epoch": 0.40379053820549665, "grad_norm": 0.9233658909797668, "learning_rate": 7.410077541405152e-06, "loss": 0.2439, "step": 22200 }, { "epoch": 0.40397242583531895, "grad_norm": 1.0570226907730103, "learning_rate": 7.407872180261183e-06, "loss": 0.2422, "step": 22210 }, { "epoch": 0.40415431346514125, "grad_norm": 1.132875680923462, "learning_rate": 7.4056663305367704e-06, "loss": 0.2479, "step": 22220 }, { "epoch": 0.40433620109496354, "grad_norm": 1.0190727710723877, "learning_rate": 7.403459992997421e-06, "loss": 0.2383, "step": 22230 }, { "epoch": 0.40451808872478584, "grad_norm": 1.0873651504516602, "learning_rate": 7.401253168408807e-06, "loss": 0.2388, "step": 22240 }, { "epoch": 0.40469997635460814, "grad_norm": 1.0912225246429443, "learning_rate": 7.399045857536779e-06, "loss": 0.2421, "step": 22250 }, { "epoch": 0.40488186398443043, "grad_norm": 1.0466713905334473, "learning_rate": 7.396838061147343e-06, "loss": 0.2348, "step": 22260 }, { "epoch": 0.40506375161425273, "grad_norm": 1.1127400398254395, "learning_rate": 7.394629780006687e-06, "loss": 0.2377, "step": 22270 }, { "epoch": 0.40524563924407503, "grad_norm": 1.1504093408584595, "learning_rate": 7.392421014881158e-06, "loss": 0.2385, "step": 22280 }, { "epoch": 0.4054275268738973, "grad_norm": 1.01735520362854, "learning_rate": 7.390211766537275e-06, "loss": 0.2415, "step": 22290 }, { "epoch": 0.4056094145037196, "grad_norm": 1.0787396430969238, "learning_rate": 7.388002035741722e-06, "loss": 0.2367, "step": 22300 }, { "epoch": 0.4057913021335419, "grad_norm": 1.0395532846450806, "learning_rate": 7.385791823261356e-06, "loss": 0.2403, "step": 22310 }, { "epoch": 0.4059731897633642, "grad_norm": 1.0740938186645508, "learning_rate": 7.383581129863195e-06, "loss": 0.2425, "step": 22320 }, { "epoch": 0.4061550773931865, "grad_norm": 1.0345404148101807, "learning_rate": 7.381369956314425e-06, "loss": 0.2384, "step": 22330 }, { "epoch": 0.4063369650230088, "grad_norm": 1.2550166845321655, "learning_rate": 7.379158303382401e-06, "loss": 0.2437, "step": 22340 }, { "epoch": 0.4065188526528311, "grad_norm": 1.0034711360931396, "learning_rate": 7.376946171834644e-06, "loss": 0.2383, "step": 22350 }, { "epoch": 0.40670074028265335, "grad_norm": 0.9295478463172913, "learning_rate": 7.37473356243884e-06, "loss": 0.2393, "step": 22360 }, { "epoch": 0.40688262791247565, "grad_norm": 0.9784128069877625, "learning_rate": 7.37252047596284e-06, "loss": 0.2381, "step": 22370 }, { "epoch": 0.40706451554229794, "grad_norm": 1.0054985284805298, "learning_rate": 7.370306913174664e-06, "loss": 0.2411, "step": 22380 }, { "epoch": 0.40724640317212024, "grad_norm": 1.0148347616195679, "learning_rate": 7.36809287484249e-06, "loss": 0.2452, "step": 22390 }, { "epoch": 0.40742829080194254, "grad_norm": 1.037340760231018, "learning_rate": 7.365878361734673e-06, "loss": 0.2424, "step": 22400 }, { "epoch": 0.40761017843176484, "grad_norm": 1.1766114234924316, "learning_rate": 7.363663374619721e-06, "loss": 0.236, "step": 22410 }, { "epoch": 0.40779206606158713, "grad_norm": 1.1524221897125244, "learning_rate": 7.361447914266314e-06, "loss": 0.2402, "step": 22420 }, { "epoch": 0.40797395369140943, "grad_norm": 1.1275984048843384, "learning_rate": 7.359231981443292e-06, "loss": 0.237, "step": 22430 }, { "epoch": 0.4081558413212317, "grad_norm": 1.068060278892517, "learning_rate": 7.357015576919659e-06, "loss": 0.2394, "step": 22440 }, { "epoch": 0.408337728951054, "grad_norm": 1.1011126041412354, "learning_rate": 7.354798701464588e-06, "loss": 0.2353, "step": 22450 }, { "epoch": 0.4085196165808763, "grad_norm": 1.0880591869354248, "learning_rate": 7.352581355847409e-06, "loss": 0.2415, "step": 22460 }, { "epoch": 0.4087015042106986, "grad_norm": 1.0561857223510742, "learning_rate": 7.350363540837618e-06, "loss": 0.2399, "step": 22470 }, { "epoch": 0.4088833918405209, "grad_norm": 1.0235008001327515, "learning_rate": 7.348145257204874e-06, "loss": 0.2452, "step": 22480 }, { "epoch": 0.4090652794703432, "grad_norm": 1.0035444498062134, "learning_rate": 7.345926505718997e-06, "loss": 0.2337, "step": 22490 }, { "epoch": 0.4092471671001655, "grad_norm": 1.1023861169815063, "learning_rate": 7.343707287149975e-06, "loss": 0.234, "step": 22500 }, { "epoch": 0.4094290547299878, "grad_norm": 0.9700878262519836, "learning_rate": 7.341487602267948e-06, "loss": 0.2385, "step": 22510 }, { "epoch": 0.4096109423598101, "grad_norm": 1.1136680841445923, "learning_rate": 7.339267451843228e-06, "loss": 0.2375, "step": 22520 }, { "epoch": 0.4097928299896324, "grad_norm": 1.0959511995315552, "learning_rate": 7.337046836646281e-06, "loss": 0.2385, "step": 22530 }, { "epoch": 0.4099747176194547, "grad_norm": 1.0515809059143066, "learning_rate": 7.33482575744774e-06, "loss": 0.2431, "step": 22540 }, { "epoch": 0.410156605249277, "grad_norm": 1.110561490058899, "learning_rate": 7.332604215018396e-06, "loss": 0.2359, "step": 22550 }, { "epoch": 0.4103384928790993, "grad_norm": 1.1301217079162598, "learning_rate": 7.3303822101292e-06, "loss": 0.2427, "step": 22560 }, { "epoch": 0.4105203805089216, "grad_norm": 1.0000431537628174, "learning_rate": 7.328159743551265e-06, "loss": 0.2355, "step": 22570 }, { "epoch": 0.4107022681387439, "grad_norm": 1.073376178741455, "learning_rate": 7.3259368160558645e-06, "loss": 0.2357, "step": 22580 }, { "epoch": 0.4108841557685662, "grad_norm": 1.1222933530807495, "learning_rate": 7.323713428414431e-06, "loss": 0.2433, "step": 22590 }, { "epoch": 0.4110660433983885, "grad_norm": 1.0707898139953613, "learning_rate": 7.321489581398557e-06, "loss": 0.2381, "step": 22600 }, { "epoch": 0.4112479310282108, "grad_norm": 1.183427095413208, "learning_rate": 7.319265275779995e-06, "loss": 0.2387, "step": 22610 }, { "epoch": 0.4114298186580331, "grad_norm": 1.0609657764434814, "learning_rate": 7.317040512330656e-06, "loss": 0.2369, "step": 22620 }, { "epoch": 0.4116117062878554, "grad_norm": 1.0563980340957642, "learning_rate": 7.31481529182261e-06, "loss": 0.2396, "step": 22630 }, { "epoch": 0.41179359391767767, "grad_norm": 0.9869409203529358, "learning_rate": 7.312589615028086e-06, "loss": 0.2308, "step": 22640 }, { "epoch": 0.41197548154749997, "grad_norm": 1.034613847732544, "learning_rate": 7.3103634827194704e-06, "loss": 0.2387, "step": 22650 }, { "epoch": 0.41215736917732226, "grad_norm": 0.9683380126953125, "learning_rate": 7.308136895669308e-06, "loss": 0.2277, "step": 22660 }, { "epoch": 0.41233925680714456, "grad_norm": 0.9941020607948303, "learning_rate": 7.3059098546503e-06, "loss": 0.2383, "step": 22670 }, { "epoch": 0.41252114443696686, "grad_norm": 0.9819587469100952, "learning_rate": 7.303682360435311e-06, "loss": 0.2354, "step": 22680 }, { "epoch": 0.41270303206678915, "grad_norm": 1.1709831953048706, "learning_rate": 7.301454413797356e-06, "loss": 0.241, "step": 22690 }, { "epoch": 0.41288491969661145, "grad_norm": 1.0668151378631592, "learning_rate": 7.299226015509612e-06, "loss": 0.2402, "step": 22700 }, { "epoch": 0.41306680732643375, "grad_norm": 1.0478585958480835, "learning_rate": 7.296997166345406e-06, "loss": 0.2355, "step": 22710 }, { "epoch": 0.41324869495625605, "grad_norm": 1.0001646280288696, "learning_rate": 7.294767867078231e-06, "loss": 0.2383, "step": 22720 }, { "epoch": 0.41343058258607834, "grad_norm": 0.9759256839752197, "learning_rate": 7.292538118481728e-06, "loss": 0.2369, "step": 22730 }, { "epoch": 0.41361247021590064, "grad_norm": 0.9180347919464111, "learning_rate": 7.290307921329698e-06, "loss": 0.2425, "step": 22740 }, { "epoch": 0.41379435784572294, "grad_norm": 1.0805007219314575, "learning_rate": 7.288077276396097e-06, "loss": 0.2371, "step": 22750 }, { "epoch": 0.41397624547554523, "grad_norm": 0.9506241679191589, "learning_rate": 7.285846184455034e-06, "loss": 0.2396, "step": 22760 }, { "epoch": 0.4141581331053675, "grad_norm": 0.8793436884880066, "learning_rate": 7.283614646280779e-06, "loss": 0.232, "step": 22770 }, { "epoch": 0.4143400207351898, "grad_norm": 1.148541808128357, "learning_rate": 7.281382662647749e-06, "loss": 0.2406, "step": 22780 }, { "epoch": 0.41452190836501207, "grad_norm": 1.1808499097824097, "learning_rate": 7.2791502343305235e-06, "loss": 0.24, "step": 22790 }, { "epoch": 0.41470379599483437, "grad_norm": 1.1933693885803223, "learning_rate": 7.276917362103829e-06, "loss": 0.2458, "step": 22800 }, { "epoch": 0.41488568362465666, "grad_norm": 1.0444930791854858, "learning_rate": 7.274684046742551e-06, "loss": 0.2401, "step": 22810 }, { "epoch": 0.41506757125447896, "grad_norm": 1.059423804283142, "learning_rate": 7.272450289021729e-06, "loss": 0.2397, "step": 22820 }, { "epoch": 0.41524945888430126, "grad_norm": 1.1424963474273682, "learning_rate": 7.270216089716552e-06, "loss": 0.2402, "step": 22830 }, { "epoch": 0.41543134651412356, "grad_norm": 0.928158164024353, "learning_rate": 7.267981449602365e-06, "loss": 0.237, "step": 22840 }, { "epoch": 0.41561323414394585, "grad_norm": 1.0149104595184326, "learning_rate": 7.265746369454668e-06, "loss": 0.2448, "step": 22850 }, { "epoch": 0.41579512177376815, "grad_norm": 0.9731175303459167, "learning_rate": 7.263510850049108e-06, "loss": 0.2352, "step": 22860 }, { "epoch": 0.41597700940359045, "grad_norm": 1.025789499282837, "learning_rate": 7.2612748921614905e-06, "loss": 0.2447, "step": 22870 }, { "epoch": 0.41615889703341274, "grad_norm": 1.1643526554107666, "learning_rate": 7.259038496567767e-06, "loss": 0.2416, "step": 22880 }, { "epoch": 0.41634078466323504, "grad_norm": 1.0501271486282349, "learning_rate": 7.256801664044047e-06, "loss": 0.2395, "step": 22890 }, { "epoch": 0.41652267229305734, "grad_norm": 1.096943736076355, "learning_rate": 7.254564395366588e-06, "loss": 0.235, "step": 22900 }, { "epoch": 0.41670455992287964, "grad_norm": 1.2223362922668457, "learning_rate": 7.252326691311801e-06, "loss": 0.2347, "step": 22910 }, { "epoch": 0.41688644755270193, "grad_norm": 0.9863176345825195, "learning_rate": 7.250088552656245e-06, "loss": 0.24, "step": 22920 }, { "epoch": 0.41706833518252423, "grad_norm": 1.037568211555481, "learning_rate": 7.247849980176631e-06, "loss": 0.2359, "step": 22930 }, { "epoch": 0.4172502228123465, "grad_norm": 1.0885649919509888, "learning_rate": 7.245610974649825e-06, "loss": 0.2407, "step": 22940 }, { "epoch": 0.4174321104421688, "grad_norm": 0.9563019871711731, "learning_rate": 7.243371536852836e-06, "loss": 0.239, "step": 22950 }, { "epoch": 0.4176139980719911, "grad_norm": 1.1007386445999146, "learning_rate": 7.241131667562828e-06, "loss": 0.243, "step": 22960 }, { "epoch": 0.4177958857018134, "grad_norm": 0.9713028073310852, "learning_rate": 7.238891367557114e-06, "loss": 0.2359, "step": 22970 }, { "epoch": 0.4179777733316357, "grad_norm": 1.1849958896636963, "learning_rate": 7.236650637613154e-06, "loss": 0.2349, "step": 22980 }, { "epoch": 0.418159660961458, "grad_norm": 1.0161100625991821, "learning_rate": 7.23440947850856e-06, "loss": 0.234, "step": 22990 }, { "epoch": 0.4183415485912803, "grad_norm": 1.023136854171753, "learning_rate": 7.232167891021092e-06, "loss": 0.234, "step": 23000 }, { "epoch": 0.4185234362211026, "grad_norm": 1.0664057731628418, "learning_rate": 7.229925875928657e-06, "loss": 0.2396, "step": 23010 }, { "epoch": 0.4187053238509249, "grad_norm": 1.2471948862075806, "learning_rate": 7.227683434009314e-06, "loss": 0.2371, "step": 23020 }, { "epoch": 0.4188872114807472, "grad_norm": 1.086255431175232, "learning_rate": 7.225440566041266e-06, "loss": 0.2439, "step": 23030 }, { "epoch": 0.4190690991105695, "grad_norm": 1.0842435359954834, "learning_rate": 7.2231972728028695e-06, "loss": 0.2379, "step": 23040 }, { "epoch": 0.4192509867403918, "grad_norm": 1.068603277206421, "learning_rate": 7.220953555072621e-06, "loss": 0.2453, "step": 23050 }, { "epoch": 0.4194328743702141, "grad_norm": 0.9778236150741577, "learning_rate": 7.218709413629171e-06, "loss": 0.238, "step": 23060 }, { "epoch": 0.4196147620000364, "grad_norm": 1.1410367488861084, "learning_rate": 7.216464849251313e-06, "loss": 0.2426, "step": 23070 }, { "epoch": 0.4197966496298587, "grad_norm": 1.1137360334396362, "learning_rate": 7.214219862717988e-06, "loss": 0.2386, "step": 23080 }, { "epoch": 0.419978537259681, "grad_norm": 1.1142606735229492, "learning_rate": 7.211974454808286e-06, "loss": 0.2365, "step": 23090 }, { "epoch": 0.4201604248895033, "grad_norm": 0.9863907098770142, "learning_rate": 7.209728626301441e-06, "loss": 0.2329, "step": 23100 }, { "epoch": 0.4203423125193256, "grad_norm": 1.2206939458847046, "learning_rate": 7.207482377976831e-06, "loss": 0.2408, "step": 23110 }, { "epoch": 0.4205242001491479, "grad_norm": 1.0902749300003052, "learning_rate": 7.205235710613986e-06, "loss": 0.2343, "step": 23120 }, { "epoch": 0.4207060877789702, "grad_norm": 1.0307663679122925, "learning_rate": 7.202988624992575e-06, "loss": 0.2366, "step": 23130 }, { "epoch": 0.42088797540879247, "grad_norm": 0.9758961796760559, "learning_rate": 7.2007411218924135e-06, "loss": 0.2314, "step": 23140 }, { "epoch": 0.42106986303861477, "grad_norm": 1.3913764953613281, "learning_rate": 7.198493202093467e-06, "loss": 0.24, "step": 23150 }, { "epoch": 0.42125175066843706, "grad_norm": 0.9324555397033691, "learning_rate": 7.196244866375836e-06, "loss": 0.2413, "step": 23160 }, { "epoch": 0.42143363829825936, "grad_norm": 1.0394598245620728, "learning_rate": 7.1939961155197755e-06, "loss": 0.244, "step": 23170 }, { "epoch": 0.42161552592808166, "grad_norm": 1.15119469165802, "learning_rate": 7.191746950305677e-06, "loss": 0.2387, "step": 23180 }, { "epoch": 0.4217974135579039, "grad_norm": 1.0587334632873535, "learning_rate": 7.189497371514082e-06, "loss": 0.241, "step": 23190 }, { "epoch": 0.4219793011877262, "grad_norm": 1.0927966833114624, "learning_rate": 7.187247379925669e-06, "loss": 0.2382, "step": 23200 }, { "epoch": 0.4221611888175485, "grad_norm": 1.063897967338562, "learning_rate": 7.184996976321266e-06, "loss": 0.2354, "step": 23210 }, { "epoch": 0.4223430764473708, "grad_norm": 1.0828526020050049, "learning_rate": 7.182746161481838e-06, "loss": 0.2383, "step": 23220 }, { "epoch": 0.4225249640771931, "grad_norm": 0.9691662192344666, "learning_rate": 7.1804949361884966e-06, "loss": 0.2378, "step": 23230 }, { "epoch": 0.4227068517070154, "grad_norm": 0.9910222887992859, "learning_rate": 7.178243301222498e-06, "loss": 0.2456, "step": 23240 }, { "epoch": 0.4228887393368377, "grad_norm": 0.952233076095581, "learning_rate": 7.175991257365236e-06, "loss": 0.2392, "step": 23250 }, { "epoch": 0.42307062696666, "grad_norm": 1.0354819297790527, "learning_rate": 7.173738805398245e-06, "loss": 0.2353, "step": 23260 }, { "epoch": 0.4232525145964823, "grad_norm": 1.0788792371749878, "learning_rate": 7.171485946103207e-06, "loss": 0.2393, "step": 23270 }, { "epoch": 0.4234344022263046, "grad_norm": 1.0296905040740967, "learning_rate": 7.1692326802619406e-06, "loss": 0.2359, "step": 23280 }, { "epoch": 0.42361628985612687, "grad_norm": 1.0408998727798462, "learning_rate": 7.166979008656409e-06, "loss": 0.2438, "step": 23290 }, { "epoch": 0.42379817748594917, "grad_norm": 1.024610996246338, "learning_rate": 7.164724932068713e-06, "loss": 0.2402, "step": 23300 }, { "epoch": 0.42398006511577147, "grad_norm": 1.118838906288147, "learning_rate": 7.162470451281095e-06, "loss": 0.2403, "step": 23310 }, { "epoch": 0.42416195274559376, "grad_norm": 1.02663254737854, "learning_rate": 7.16021556707594e-06, "loss": 0.2384, "step": 23320 }, { "epoch": 0.42434384037541606, "grad_norm": 1.062116026878357, "learning_rate": 7.157960280235768e-06, "loss": 0.2384, "step": 23330 }, { "epoch": 0.42452572800523836, "grad_norm": 1.0632929801940918, "learning_rate": 7.155704591543245e-06, "loss": 0.2382, "step": 23340 }, { "epoch": 0.42470761563506065, "grad_norm": 1.1716848611831665, "learning_rate": 7.15344850178117e-06, "loss": 0.2443, "step": 23350 }, { "epoch": 0.42488950326488295, "grad_norm": 1.1443759202957153, "learning_rate": 7.1511920117324855e-06, "loss": 0.2424, "step": 23360 }, { "epoch": 0.42507139089470525, "grad_norm": 0.9864426851272583, "learning_rate": 7.148935122180273e-06, "loss": 0.2393, "step": 23370 }, { "epoch": 0.42525327852452754, "grad_norm": 0.9501287341117859, "learning_rate": 7.146677833907749e-06, "loss": 0.2403, "step": 23380 }, { "epoch": 0.42543516615434984, "grad_norm": 1.106345295906067, "learning_rate": 7.144420147698274e-06, "loss": 0.234, "step": 23390 }, { "epoch": 0.42561705378417214, "grad_norm": 1.0831334590911865, "learning_rate": 7.14216206433534e-06, "loss": 0.2343, "step": 23400 }, { "epoch": 0.42579894141399444, "grad_norm": 1.1097761392593384, "learning_rate": 7.139903584602581e-06, "loss": 0.2386, "step": 23410 }, { "epoch": 0.42598082904381673, "grad_norm": 1.0937780141830444, "learning_rate": 7.137644709283771e-06, "loss": 0.2393, "step": 23420 }, { "epoch": 0.42616271667363903, "grad_norm": 1.0225157737731934, "learning_rate": 7.135385439162813e-06, "loss": 0.2369, "step": 23430 }, { "epoch": 0.4263446043034613, "grad_norm": 1.0780187845230103, "learning_rate": 7.133125775023757e-06, "loss": 0.2323, "step": 23440 }, { "epoch": 0.4265264919332836, "grad_norm": 1.032809853553772, "learning_rate": 7.13086571765078e-06, "loss": 0.237, "step": 23450 }, { "epoch": 0.4267083795631059, "grad_norm": 0.9844892621040344, "learning_rate": 7.128605267828205e-06, "loss": 0.2401, "step": 23460 }, { "epoch": 0.4268902671929282, "grad_norm": 1.195318579673767, "learning_rate": 7.1263444263404825e-06, "loss": 0.2368, "step": 23470 }, { "epoch": 0.4270721548227505, "grad_norm": 1.100950002670288, "learning_rate": 7.1240831939722045e-06, "loss": 0.2386, "step": 23480 }, { "epoch": 0.4272540424525728, "grad_norm": 1.076837182044983, "learning_rate": 7.1218215715080975e-06, "loss": 0.2376, "step": 23490 }, { "epoch": 0.4274359300823951, "grad_norm": 1.2090212106704712, "learning_rate": 7.1195595597330205e-06, "loss": 0.2388, "step": 23500 }, { "epoch": 0.4276178177122174, "grad_norm": 1.0686225891113281, "learning_rate": 7.117297159431973e-06, "loss": 0.2406, "step": 23510 }, { "epoch": 0.4277997053420397, "grad_norm": 1.0764039754867554, "learning_rate": 7.115034371390085e-06, "loss": 0.2346, "step": 23520 }, { "epoch": 0.427981592971862, "grad_norm": 1.2116689682006836, "learning_rate": 7.112771196392622e-06, "loss": 0.242, "step": 23530 }, { "epoch": 0.4281634806016843, "grad_norm": 1.0496398210525513, "learning_rate": 7.110507635224986e-06, "loss": 0.2404, "step": 23540 }, { "epoch": 0.4283453682315066, "grad_norm": 1.1282240152359009, "learning_rate": 7.108243688672708e-06, "loss": 0.2379, "step": 23550 }, { "epoch": 0.4285272558613289, "grad_norm": 1.0475882291793823, "learning_rate": 7.105979357521459e-06, "loss": 0.2305, "step": 23560 }, { "epoch": 0.4287091434911512, "grad_norm": 1.0305306911468506, "learning_rate": 7.103714642557038e-06, "loss": 0.2364, "step": 23570 }, { "epoch": 0.4288910311209735, "grad_norm": 1.1644500494003296, "learning_rate": 7.101449544565382e-06, "loss": 0.2406, "step": 23580 }, { "epoch": 0.4290729187507958, "grad_norm": 1.1067997217178345, "learning_rate": 7.099184064332557e-06, "loss": 0.2366, "step": 23590 }, { "epoch": 0.4292548063806181, "grad_norm": 1.1816115379333496, "learning_rate": 7.096918202644763e-06, "loss": 0.2439, "step": 23600 }, { "epoch": 0.4294366940104403, "grad_norm": 1.2005125284194946, "learning_rate": 7.094651960288332e-06, "loss": 0.237, "step": 23610 }, { "epoch": 0.4296185816402626, "grad_norm": 1.050127625465393, "learning_rate": 7.09238533804973e-06, "loss": 0.24, "step": 23620 }, { "epoch": 0.4298004692700849, "grad_norm": 1.0180304050445557, "learning_rate": 7.090118336715554e-06, "loss": 0.2389, "step": 23630 }, { "epoch": 0.4299823568999072, "grad_norm": 1.1575292348861694, "learning_rate": 7.087850957072531e-06, "loss": 0.2381, "step": 23640 }, { "epoch": 0.4301642445297295, "grad_norm": 1.1625980138778687, "learning_rate": 7.085583199907519e-06, "loss": 0.2424, "step": 23650 }, { "epoch": 0.4303461321595518, "grad_norm": 0.9848566651344299, "learning_rate": 7.083315066007511e-06, "loss": 0.234, "step": 23660 }, { "epoch": 0.4305280197893741, "grad_norm": 1.0324928760528564, "learning_rate": 7.081046556159626e-06, "loss": 0.2359, "step": 23670 }, { "epoch": 0.4307099074191964, "grad_norm": 1.059889316558838, "learning_rate": 7.0787776711511165e-06, "loss": 0.2383, "step": 23680 }, { "epoch": 0.4308917950490187, "grad_norm": 0.9881415367126465, "learning_rate": 7.076508411769365e-06, "loss": 0.2342, "step": 23690 }, { "epoch": 0.431073682678841, "grad_norm": 1.0066877603530884, "learning_rate": 7.074238778801882e-06, "loss": 0.2337, "step": 23700 }, { "epoch": 0.4312555703086633, "grad_norm": 0.9914868474006653, "learning_rate": 7.071968773036309e-06, "loss": 0.2364, "step": 23710 }, { "epoch": 0.4314374579384856, "grad_norm": 1.045775294303894, "learning_rate": 7.069698395260416e-06, "loss": 0.2407, "step": 23720 }, { "epoch": 0.4316193455683079, "grad_norm": 1.0545380115509033, "learning_rate": 7.067427646262104e-06, "loss": 0.2406, "step": 23730 }, { "epoch": 0.4318012331981302, "grad_norm": 1.0858396291732788, "learning_rate": 7.065156526829402e-06, "loss": 0.2339, "step": 23740 }, { "epoch": 0.4319831208279525, "grad_norm": 0.9773837924003601, "learning_rate": 7.062885037750466e-06, "loss": 0.2377, "step": 23750 }, { "epoch": 0.4321650084577748, "grad_norm": 1.1963391304016113, "learning_rate": 7.0606131798135805e-06, "loss": 0.2368, "step": 23760 }, { "epoch": 0.4323468960875971, "grad_norm": 1.126971960067749, "learning_rate": 7.058340953807159e-06, "loss": 0.2425, "step": 23770 }, { "epoch": 0.4325287837174194, "grad_norm": 1.1863458156585693, "learning_rate": 7.056068360519747e-06, "loss": 0.2365, "step": 23780 }, { "epoch": 0.43271067134724167, "grad_norm": 1.151723027229309, "learning_rate": 7.053795400740009e-06, "loss": 0.2333, "step": 23790 }, { "epoch": 0.43289255897706397, "grad_norm": 1.1243853569030762, "learning_rate": 7.0515220752567415e-06, "loss": 0.2382, "step": 23800 }, { "epoch": 0.43307444660688627, "grad_norm": 1.1186810731887817, "learning_rate": 7.049248384858868e-06, "loss": 0.2351, "step": 23810 }, { "epoch": 0.43325633423670856, "grad_norm": 1.0047249794006348, "learning_rate": 7.046974330335436e-06, "loss": 0.2373, "step": 23820 }, { "epoch": 0.43343822186653086, "grad_norm": 1.1954692602157593, "learning_rate": 7.0446999124756244e-06, "loss": 0.2449, "step": 23830 }, { "epoch": 0.43362010949635316, "grad_norm": 1.1137282848358154, "learning_rate": 7.042425132068733e-06, "loss": 0.2372, "step": 23840 }, { "epoch": 0.43380199712617545, "grad_norm": 1.1869487762451172, "learning_rate": 7.04014998990419e-06, "loss": 0.2396, "step": 23850 }, { "epoch": 0.43398388475599775, "grad_norm": 1.0160762071609497, "learning_rate": 7.037874486771548e-06, "loss": 0.2409, "step": 23860 }, { "epoch": 0.43416577238582005, "grad_norm": 1.0851244926452637, "learning_rate": 7.035598623460484e-06, "loss": 0.2395, "step": 23870 }, { "epoch": 0.43434766001564235, "grad_norm": 1.2595776319503784, "learning_rate": 7.033322400760806e-06, "loss": 0.2399, "step": 23880 }, { "epoch": 0.43452954764546464, "grad_norm": 1.0538264513015747, "learning_rate": 7.031045819462439e-06, "loss": 0.2446, "step": 23890 }, { "epoch": 0.43471143527528694, "grad_norm": 1.0823968648910522, "learning_rate": 7.028768880355435e-06, "loss": 0.236, "step": 23900 }, { "epoch": 0.43489332290510924, "grad_norm": 1.0586066246032715, "learning_rate": 7.0264915842299715e-06, "loss": 0.236, "step": 23910 }, { "epoch": 0.43507521053493153, "grad_norm": 1.0941617488861084, "learning_rate": 7.02421393187635e-06, "loss": 0.2341, "step": 23920 }, { "epoch": 0.43525709816475383, "grad_norm": 1.0277597904205322, "learning_rate": 7.021935924084995e-06, "loss": 0.2375, "step": 23930 }, { "epoch": 0.4354389857945761, "grad_norm": 1.0073745250701904, "learning_rate": 7.019657561646453e-06, "loss": 0.2417, "step": 23940 }, { "epoch": 0.4356208734243984, "grad_norm": 1.0006564855575562, "learning_rate": 7.017378845351396e-06, "loss": 0.2354, "step": 23950 }, { "epoch": 0.4358027610542207, "grad_norm": 1.0257742404937744, "learning_rate": 7.015099775990616e-06, "loss": 0.237, "step": 23960 }, { "epoch": 0.435984648684043, "grad_norm": 1.0938684940338135, "learning_rate": 7.012820354355029e-06, "loss": 0.2382, "step": 23970 }, { "epoch": 0.4361665363138653, "grad_norm": 1.1247373819351196, "learning_rate": 7.010540581235676e-06, "loss": 0.2363, "step": 23980 }, { "epoch": 0.4363484239436876, "grad_norm": 1.0468597412109375, "learning_rate": 7.008260457423713e-06, "loss": 0.2395, "step": 23990 }, { "epoch": 0.4365303115735099, "grad_norm": 1.0977298021316528, "learning_rate": 7.005979983710427e-06, "loss": 0.2437, "step": 24000 }, { "epoch": 0.4367121992033322, "grad_norm": 1.0002497434616089, "learning_rate": 7.003699160887218e-06, "loss": 0.2403, "step": 24010 }, { "epoch": 0.4368940868331545, "grad_norm": 1.3048286437988281, "learning_rate": 7.001417989745611e-06, "loss": 0.2355, "step": 24020 }, { "epoch": 0.43707597446297675, "grad_norm": 1.0671154260635376, "learning_rate": 6.999136471077252e-06, "loss": 0.2411, "step": 24030 }, { "epoch": 0.43725786209279904, "grad_norm": 1.0384433269500732, "learning_rate": 6.996854605673907e-06, "loss": 0.2371, "step": 24040 }, { "epoch": 0.43743974972262134, "grad_norm": 1.1062817573547363, "learning_rate": 6.994572394327462e-06, "loss": 0.237, "step": 24050 }, { "epoch": 0.43762163735244364, "grad_norm": 1.0300298929214478, "learning_rate": 6.992289837829924e-06, "loss": 0.2353, "step": 24060 }, { "epoch": 0.43780352498226593, "grad_norm": 1.021735429763794, "learning_rate": 6.990006936973419e-06, "loss": 0.2375, "step": 24070 }, { "epoch": 0.43798541261208823, "grad_norm": 1.1779485940933228, "learning_rate": 6.987723692550194e-06, "loss": 0.2393, "step": 24080 }, { "epoch": 0.43816730024191053, "grad_norm": 1.1900867223739624, "learning_rate": 6.98544010535261e-06, "loss": 0.2312, "step": 24090 }, { "epoch": 0.4383491878717328, "grad_norm": 1.0953954458236694, "learning_rate": 6.983156176173156e-06, "loss": 0.239, "step": 24100 }, { "epoch": 0.4385310755015551, "grad_norm": 0.9958499670028687, "learning_rate": 6.980871905804432e-06, "loss": 0.2314, "step": 24110 }, { "epoch": 0.4387129631313774, "grad_norm": 1.0940946340560913, "learning_rate": 6.978587295039157e-06, "loss": 0.2417, "step": 24120 }, { "epoch": 0.4388948507611997, "grad_norm": 1.0588243007659912, "learning_rate": 6.976302344670174e-06, "loss": 0.235, "step": 24130 }, { "epoch": 0.439076738391022, "grad_norm": 1.1368093490600586, "learning_rate": 6.974017055490437e-06, "loss": 0.241, "step": 24140 }, { "epoch": 0.4392586260208443, "grad_norm": 1.049422264099121, "learning_rate": 6.971731428293022e-06, "loss": 0.2368, "step": 24150 }, { "epoch": 0.4394405136506666, "grad_norm": 1.044221043586731, "learning_rate": 6.969445463871122e-06, "loss": 0.2386, "step": 24160 }, { "epoch": 0.4396224012804889, "grad_norm": 1.0923479795455933, "learning_rate": 6.9671591630180445e-06, "loss": 0.2359, "step": 24170 }, { "epoch": 0.4398042889103112, "grad_norm": 1.067741870880127, "learning_rate": 6.964872526527216e-06, "loss": 0.2455, "step": 24180 }, { "epoch": 0.4399861765401335, "grad_norm": 1.1386277675628662, "learning_rate": 6.962585555192174e-06, "loss": 0.2336, "step": 24190 }, { "epoch": 0.4401680641699558, "grad_norm": 0.9775949716567993, "learning_rate": 6.960298249806583e-06, "loss": 0.2319, "step": 24200 }, { "epoch": 0.4403499517997781, "grad_norm": 1.0393908023834229, "learning_rate": 6.958010611164216e-06, "loss": 0.24, "step": 24210 }, { "epoch": 0.4405318394296004, "grad_norm": 1.0026535987854004, "learning_rate": 6.9557226400589596e-06, "loss": 0.2364, "step": 24220 }, { "epoch": 0.4407137270594227, "grad_norm": 1.0182509422302246, "learning_rate": 6.953434337284822e-06, "loss": 0.2379, "step": 24230 }, { "epoch": 0.440895614689245, "grad_norm": 0.9284889101982117, "learning_rate": 6.951145703635923e-06, "loss": 0.238, "step": 24240 }, { "epoch": 0.4410775023190673, "grad_norm": 1.028310775756836, "learning_rate": 6.948856739906498e-06, "loss": 0.2341, "step": 24250 }, { "epoch": 0.4412593899488896, "grad_norm": 1.0044920444488525, "learning_rate": 6.946567446890895e-06, "loss": 0.2366, "step": 24260 }, { "epoch": 0.4414412775787119, "grad_norm": 1.0446250438690186, "learning_rate": 6.944277825383581e-06, "loss": 0.2379, "step": 24270 }, { "epoch": 0.4416231652085342, "grad_norm": 1.086683750152588, "learning_rate": 6.9419878761791325e-06, "loss": 0.2367, "step": 24280 }, { "epoch": 0.44180505283835647, "grad_norm": 1.1165001392364502, "learning_rate": 6.9396976000722405e-06, "loss": 0.2394, "step": 24290 }, { "epoch": 0.44198694046817877, "grad_norm": 1.1153897047042847, "learning_rate": 6.93740699785771e-06, "loss": 0.2377, "step": 24300 }, { "epoch": 0.44216882809800107, "grad_norm": 1.1123895645141602, "learning_rate": 6.935116070330461e-06, "loss": 0.237, "step": 24310 }, { "epoch": 0.44235071572782336, "grad_norm": 1.0359766483306885, "learning_rate": 6.932824818285524e-06, "loss": 0.2408, "step": 24320 }, { "epoch": 0.44253260335764566, "grad_norm": 1.1336876153945923, "learning_rate": 6.930533242518043e-06, "loss": 0.2397, "step": 24330 }, { "epoch": 0.44271449098746796, "grad_norm": 0.9595381617546082, "learning_rate": 6.928241343823273e-06, "loss": 0.2339, "step": 24340 }, { "epoch": 0.44289637861729025, "grad_norm": 1.2472938299179077, "learning_rate": 6.925949122996584e-06, "loss": 0.2381, "step": 24350 }, { "epoch": 0.44307826624711255, "grad_norm": 0.9500992894172668, "learning_rate": 6.923656580833457e-06, "loss": 0.2303, "step": 24360 }, { "epoch": 0.44326015387693485, "grad_norm": 1.0638917684555054, "learning_rate": 6.92136371812948e-06, "loss": 0.2429, "step": 24370 }, { "epoch": 0.44344204150675715, "grad_norm": 0.9212687015533447, "learning_rate": 6.91907053568036e-06, "loss": 0.2333, "step": 24380 }, { "epoch": 0.44362392913657944, "grad_norm": 0.9923521280288696, "learning_rate": 6.916777034281906e-06, "loss": 0.2371, "step": 24390 }, { "epoch": 0.44380581676640174, "grad_norm": 1.1570972204208374, "learning_rate": 6.9144832147300475e-06, "loss": 0.2311, "step": 24400 }, { "epoch": 0.44398770439622404, "grad_norm": 1.0334687232971191, "learning_rate": 6.9121890778208145e-06, "loss": 0.2385, "step": 24410 }, { "epoch": 0.44416959202604633, "grad_norm": 1.1346070766448975, "learning_rate": 6.909894624350357e-06, "loss": 0.2342, "step": 24420 }, { "epoch": 0.44435147965586863, "grad_norm": 0.9753089547157288, "learning_rate": 6.907599855114926e-06, "loss": 0.2371, "step": 24430 }, { "epoch": 0.4445333672856909, "grad_norm": 1.0082833766937256, "learning_rate": 6.905304770910888e-06, "loss": 0.235, "step": 24440 }, { "epoch": 0.44471525491551317, "grad_norm": 1.1052876710891724, "learning_rate": 6.903009372534717e-06, "loss": 0.2339, "step": 24450 }, { "epoch": 0.44489714254533547, "grad_norm": 0.9987565279006958, "learning_rate": 6.900713660782992e-06, "loss": 0.2336, "step": 24460 }, { "epoch": 0.44507903017515776, "grad_norm": 1.0185377597808838, "learning_rate": 6.89841763645241e-06, "loss": 0.2357, "step": 24470 }, { "epoch": 0.44526091780498006, "grad_norm": 1.1020069122314453, "learning_rate": 6.8961213003397675e-06, "loss": 0.2372, "step": 24480 }, { "epoch": 0.44544280543480236, "grad_norm": 0.9843516945838928, "learning_rate": 6.893824653241974e-06, "loss": 0.2311, "step": 24490 }, { "epoch": 0.44562469306462466, "grad_norm": 1.08672034740448, "learning_rate": 6.8915276959560454e-06, "loss": 0.2381, "step": 24500 }, { "epoch": 0.44580658069444695, "grad_norm": 0.9559285640716553, "learning_rate": 6.889230429279104e-06, "loss": 0.2345, "step": 24510 }, { "epoch": 0.44598846832426925, "grad_norm": 0.9677874445915222, "learning_rate": 6.886932854008386e-06, "loss": 0.2367, "step": 24520 }, { "epoch": 0.44617035595409155, "grad_norm": 1.1277265548706055, "learning_rate": 6.884634970941224e-06, "loss": 0.2391, "step": 24530 }, { "epoch": 0.44635224358391384, "grad_norm": 1.038374423980713, "learning_rate": 6.882336780875065e-06, "loss": 0.2297, "step": 24540 }, { "epoch": 0.44653413121373614, "grad_norm": 1.031153917312622, "learning_rate": 6.880038284607461e-06, "loss": 0.2379, "step": 24550 }, { "epoch": 0.44671601884355844, "grad_norm": 1.0968101024627686, "learning_rate": 6.87773948293607e-06, "loss": 0.232, "step": 24560 }, { "epoch": 0.44689790647338073, "grad_norm": 1.031129002571106, "learning_rate": 6.875440376658654e-06, "loss": 0.2426, "step": 24570 }, { "epoch": 0.44707979410320303, "grad_norm": 0.9874274134635925, "learning_rate": 6.873140966573086e-06, "loss": 0.2419, "step": 24580 }, { "epoch": 0.44726168173302533, "grad_norm": 1.1212162971496582, "learning_rate": 6.870841253477339e-06, "loss": 0.2379, "step": 24590 }, { "epoch": 0.4474435693628476, "grad_norm": 1.2875007390975952, "learning_rate": 6.8685412381694916e-06, "loss": 0.238, "step": 24600 }, { "epoch": 0.4476254569926699, "grad_norm": 1.0078275203704834, "learning_rate": 6.866240921447731e-06, "loss": 0.2374, "step": 24610 }, { "epoch": 0.4478073446224922, "grad_norm": 1.1432603597640991, "learning_rate": 6.863940304110347e-06, "loss": 0.2458, "step": 24620 }, { "epoch": 0.4479892322523145, "grad_norm": 1.0729737281799316, "learning_rate": 6.861639386955732e-06, "loss": 0.2416, "step": 24630 }, { "epoch": 0.4481711198821368, "grad_norm": 1.0763455629348755, "learning_rate": 6.859338170782386e-06, "loss": 0.2402, "step": 24640 }, { "epoch": 0.4483530075119591, "grad_norm": 1.0984035730361938, "learning_rate": 6.857036656388909e-06, "loss": 0.2311, "step": 24650 }, { "epoch": 0.4485348951417814, "grad_norm": 1.089073657989502, "learning_rate": 6.854734844574007e-06, "loss": 0.2373, "step": 24660 }, { "epoch": 0.4487167827716037, "grad_norm": 1.0589081048965454, "learning_rate": 6.8524327361364875e-06, "loss": 0.2316, "step": 24670 }, { "epoch": 0.448898670401426, "grad_norm": 1.0228137969970703, "learning_rate": 6.850130331875264e-06, "loss": 0.2406, "step": 24680 }, { "epoch": 0.4490805580312483, "grad_norm": 1.1417981386184692, "learning_rate": 6.847827632589348e-06, "loss": 0.2419, "step": 24690 }, { "epoch": 0.4492624456610706, "grad_norm": 1.0208015441894531, "learning_rate": 6.8455246390778585e-06, "loss": 0.2434, "step": 24700 }, { "epoch": 0.4494443332908929, "grad_norm": 1.0515190362930298, "learning_rate": 6.843221352140012e-06, "loss": 0.2374, "step": 24710 }, { "epoch": 0.4496262209207152, "grad_norm": 1.016745924949646, "learning_rate": 6.840917772575132e-06, "loss": 0.2346, "step": 24720 }, { "epoch": 0.4498081085505375, "grad_norm": 1.107338786125183, "learning_rate": 6.838613901182636e-06, "loss": 0.2343, "step": 24730 }, { "epoch": 0.4499899961803598, "grad_norm": 1.0406426191329956, "learning_rate": 6.836309738762049e-06, "loss": 0.2416, "step": 24740 }, { "epoch": 0.4501718838101821, "grad_norm": 0.9632081985473633, "learning_rate": 6.834005286112997e-06, "loss": 0.2351, "step": 24750 }, { "epoch": 0.4503537714400044, "grad_norm": 1.0974901914596558, "learning_rate": 6.831700544035203e-06, "loss": 0.2379, "step": 24760 }, { "epoch": 0.4505356590698267, "grad_norm": 0.9827783703804016, "learning_rate": 6.829395513328495e-06, "loss": 0.2304, "step": 24770 }, { "epoch": 0.450717546699649, "grad_norm": 1.0965068340301514, "learning_rate": 6.827090194792794e-06, "loss": 0.2368, "step": 24780 }, { "epoch": 0.45089943432947127, "grad_norm": 1.0258764028549194, "learning_rate": 6.824784589228131e-06, "loss": 0.2322, "step": 24790 }, { "epoch": 0.45108132195929357, "grad_norm": 1.0631135702133179, "learning_rate": 6.822478697434626e-06, "loss": 0.2367, "step": 24800 }, { "epoch": 0.45126320958911587, "grad_norm": 1.084489107131958, "learning_rate": 6.820172520212504e-06, "loss": 0.2386, "step": 24810 }, { "epoch": 0.45144509721893816, "grad_norm": 1.0670431852340698, "learning_rate": 6.817866058362093e-06, "loss": 0.2396, "step": 24820 }, { "epoch": 0.45162698484876046, "grad_norm": 1.0888671875, "learning_rate": 6.815559312683809e-06, "loss": 0.2315, "step": 24830 }, { "epoch": 0.45180887247858276, "grad_norm": 1.1095720529556274, "learning_rate": 6.813252283978179e-06, "loss": 0.2335, "step": 24840 }, { "epoch": 0.45199076010840505, "grad_norm": 1.06611168384552, "learning_rate": 6.810944973045817e-06, "loss": 0.2394, "step": 24850 }, { "epoch": 0.4521726477382273, "grad_norm": 0.9625853896141052, "learning_rate": 6.808637380687442e-06, "loss": 0.2384, "step": 24860 }, { "epoch": 0.4523545353680496, "grad_norm": 1.0460164546966553, "learning_rate": 6.806329507703867e-06, "loss": 0.237, "step": 24870 }, { "epoch": 0.4525364229978719, "grad_norm": 1.016377568244934, "learning_rate": 6.804021354896006e-06, "loss": 0.2376, "step": 24880 }, { "epoch": 0.4527183106276942, "grad_norm": 1.0886503458023071, "learning_rate": 6.801712923064866e-06, "loss": 0.2321, "step": 24890 }, { "epoch": 0.4529001982575165, "grad_norm": 1.0180094242095947, "learning_rate": 6.799404213011555e-06, "loss": 0.2335, "step": 24900 }, { "epoch": 0.4530820858873388, "grad_norm": 0.9098349213600159, "learning_rate": 6.7970952255372735e-06, "loss": 0.2377, "step": 24910 }, { "epoch": 0.4532639735171611, "grad_norm": 1.0139611959457397, "learning_rate": 6.794785961443322e-06, "loss": 0.2372, "step": 24920 }, { "epoch": 0.4534458611469834, "grad_norm": 0.9882906079292297, "learning_rate": 6.7924764215310936e-06, "loss": 0.235, "step": 24930 }, { "epoch": 0.4536277487768057, "grad_norm": 1.0596153736114502, "learning_rate": 6.79016660660208e-06, "loss": 0.2401, "step": 24940 }, { "epoch": 0.45380963640662797, "grad_norm": 1.179579257965088, "learning_rate": 6.787856517457865e-06, "loss": 0.2397, "step": 24950 }, { "epoch": 0.45399152403645027, "grad_norm": 1.1289035081863403, "learning_rate": 6.785546154900131e-06, "loss": 0.2328, "step": 24960 }, { "epoch": 0.45417341166627256, "grad_norm": 1.0979113578796387, "learning_rate": 6.783235519730655e-06, "loss": 0.232, "step": 24970 }, { "epoch": 0.45435529929609486, "grad_norm": 1.138047218322754, "learning_rate": 6.780924612751305e-06, "loss": 0.2377, "step": 24980 }, { "epoch": 0.45453718692591716, "grad_norm": 1.0586059093475342, "learning_rate": 6.778613434764048e-06, "loss": 0.2417, "step": 24990 }, { "epoch": 0.45471907455573946, "grad_norm": 1.1543490886688232, "learning_rate": 6.776301986570943e-06, "loss": 0.2365, "step": 25000 }, { "epoch": 0.45490096218556175, "grad_norm": 1.138046145439148, "learning_rate": 6.773990268974138e-06, "loss": 0.2375, "step": 25010 }, { "epoch": 0.45508284981538405, "grad_norm": 1.2433069944381714, "learning_rate": 6.771678282775886e-06, "loss": 0.2351, "step": 25020 }, { "epoch": 0.45526473744520635, "grad_norm": 1.194191336631775, "learning_rate": 6.769366028778522e-06, "loss": 0.238, "step": 25030 }, { "epoch": 0.45544662507502864, "grad_norm": 1.0019208192825317, "learning_rate": 6.76705350778448e-06, "loss": 0.2268, "step": 25040 }, { "epoch": 0.45562851270485094, "grad_norm": 1.0959892272949219, "learning_rate": 6.764740720596283e-06, "loss": 0.236, "step": 25050 }, { "epoch": 0.45581040033467324, "grad_norm": 1.3534208536148071, "learning_rate": 6.762427668016551e-06, "loss": 0.2369, "step": 25060 }, { "epoch": 0.45599228796449554, "grad_norm": 1.0156539678573608, "learning_rate": 6.760114350847993e-06, "loss": 0.2402, "step": 25070 }, { "epoch": 0.45617417559431783, "grad_norm": 1.0428494215011597, "learning_rate": 6.757800769893407e-06, "loss": 0.2361, "step": 25080 }, { "epoch": 0.45635606322414013, "grad_norm": 1.097024917602539, "learning_rate": 6.75548692595569e-06, "loss": 0.2307, "step": 25090 }, { "epoch": 0.4565379508539624, "grad_norm": 1.028306484222412, "learning_rate": 6.753172819837822e-06, "loss": 0.2347, "step": 25100 }, { "epoch": 0.4567198384837847, "grad_norm": 1.0607043504714966, "learning_rate": 6.750858452342882e-06, "loss": 0.2381, "step": 25110 }, { "epoch": 0.456901726113607, "grad_norm": 1.0095734596252441, "learning_rate": 6.748543824274032e-06, "loss": 0.2328, "step": 25120 }, { "epoch": 0.4570836137434293, "grad_norm": 1.0406898260116577, "learning_rate": 6.746228936434531e-06, "loss": 0.2345, "step": 25130 }, { "epoch": 0.4572655013732516, "grad_norm": 0.9867165088653564, "learning_rate": 6.743913789627727e-06, "loss": 0.2379, "step": 25140 }, { "epoch": 0.4574473890030739, "grad_norm": 1.0389689207077026, "learning_rate": 6.7415983846570486e-06, "loss": 0.236, "step": 25150 }, { "epoch": 0.4576292766328962, "grad_norm": 1.1993234157562256, "learning_rate": 6.73928272232603e-06, "loss": 0.2353, "step": 25160 }, { "epoch": 0.4578111642627185, "grad_norm": 1.121263861656189, "learning_rate": 6.736966803438282e-06, "loss": 0.2342, "step": 25170 }, { "epoch": 0.4579930518925408, "grad_norm": 1.1350700855255127, "learning_rate": 6.734650628797509e-06, "loss": 0.2343, "step": 25180 }, { "epoch": 0.4581749395223631, "grad_norm": 1.0709782838821411, "learning_rate": 6.732334199207505e-06, "loss": 0.2425, "step": 25190 }, { "epoch": 0.4583568271521854, "grad_norm": 1.0963671207427979, "learning_rate": 6.730017515472151e-06, "loss": 0.2393, "step": 25200 }, { "epoch": 0.4585387147820077, "grad_norm": 1.0402522087097168, "learning_rate": 6.7277005783954155e-06, "loss": 0.2369, "step": 25210 }, { "epoch": 0.45872060241183, "grad_norm": 1.0652168989181519, "learning_rate": 6.725383388781357e-06, "loss": 0.2383, "step": 25220 }, { "epoch": 0.4589024900416523, "grad_norm": 1.0686155557632446, "learning_rate": 6.72306594743412e-06, "loss": 0.238, "step": 25230 }, { "epoch": 0.4590843776714746, "grad_norm": 1.0270164012908936, "learning_rate": 6.7207482551579396e-06, "loss": 0.2327, "step": 25240 }, { "epoch": 0.4592662653012969, "grad_norm": 0.9497126936912537, "learning_rate": 6.71843031275713e-06, "loss": 0.2367, "step": 25250 }, { "epoch": 0.4594481529311192, "grad_norm": 1.024125337600708, "learning_rate": 6.716112121036103e-06, "loss": 0.236, "step": 25260 }, { "epoch": 0.4596300405609415, "grad_norm": 1.0048960447311401, "learning_rate": 6.7137936807993475e-06, "loss": 0.2328, "step": 25270 }, { "epoch": 0.4598119281907637, "grad_norm": 1.1482211351394653, "learning_rate": 6.711474992851446e-06, "loss": 0.2327, "step": 25280 }, { "epoch": 0.459993815820586, "grad_norm": 1.0407403707504272, "learning_rate": 6.709156057997061e-06, "loss": 0.2335, "step": 25290 }, { "epoch": 0.4601757034504083, "grad_norm": 1.0434486865997314, "learning_rate": 6.706836877040943e-06, "loss": 0.2313, "step": 25300 }, { "epoch": 0.4603575910802306, "grad_norm": 1.000118374824524, "learning_rate": 6.7045174507879316e-06, "loss": 0.2319, "step": 25310 }, { "epoch": 0.4605394787100529, "grad_norm": 1.0747923851013184, "learning_rate": 6.702197780042945e-06, "loss": 0.2368, "step": 25320 }, { "epoch": 0.4607213663398752, "grad_norm": 1.1375696659088135, "learning_rate": 6.69987786561099e-06, "loss": 0.2396, "step": 25330 }, { "epoch": 0.4609032539696975, "grad_norm": 1.1442652940750122, "learning_rate": 6.697557708297157e-06, "loss": 0.2333, "step": 25340 }, { "epoch": 0.4610851415995198, "grad_norm": 0.9597370028495789, "learning_rate": 6.695237308906623e-06, "loss": 0.2357, "step": 25350 }, { "epoch": 0.4612670292293421, "grad_norm": 1.1268872022628784, "learning_rate": 6.692916668244644e-06, "loss": 0.2396, "step": 25360 }, { "epoch": 0.4614489168591644, "grad_norm": 1.1376478672027588, "learning_rate": 6.690595787116565e-06, "loss": 0.236, "step": 25370 }, { "epoch": 0.4616308044889867, "grad_norm": 0.9720773100852966, "learning_rate": 6.688274666327811e-06, "loss": 0.238, "step": 25380 }, { "epoch": 0.461812692118809, "grad_norm": 1.1524460315704346, "learning_rate": 6.6859533066838925e-06, "loss": 0.2353, "step": 25390 }, { "epoch": 0.4619945797486313, "grad_norm": 0.9662296772003174, "learning_rate": 6.6836317089904e-06, "loss": 0.2334, "step": 25400 }, { "epoch": 0.4621764673784536, "grad_norm": 0.9634810090065002, "learning_rate": 6.681309874053011e-06, "loss": 0.2395, "step": 25410 }, { "epoch": 0.4623583550082759, "grad_norm": 1.1281145811080933, "learning_rate": 6.678987802677479e-06, "loss": 0.2411, "step": 25420 }, { "epoch": 0.4625402426380982, "grad_norm": 1.0408889055252075, "learning_rate": 6.676665495669646e-06, "loss": 0.2405, "step": 25430 }, { "epoch": 0.4627221302679205, "grad_norm": 1.0058149099349976, "learning_rate": 6.674342953835433e-06, "loss": 0.2329, "step": 25440 }, { "epoch": 0.46290401789774277, "grad_norm": 1.005355954170227, "learning_rate": 6.672020177980843e-06, "loss": 0.2364, "step": 25450 }, { "epoch": 0.46308590552756507, "grad_norm": 1.2052382230758667, "learning_rate": 6.669697168911957e-06, "loss": 0.2374, "step": 25460 }, { "epoch": 0.46326779315738736, "grad_norm": 1.0517077445983887, "learning_rate": 6.667373927434941e-06, "loss": 0.2385, "step": 25470 }, { "epoch": 0.46344968078720966, "grad_norm": 0.9892187118530273, "learning_rate": 6.665050454356043e-06, "loss": 0.2384, "step": 25480 }, { "epoch": 0.46363156841703196, "grad_norm": 1.1377670764923096, "learning_rate": 6.662726750481585e-06, "loss": 0.237, "step": 25490 }, { "epoch": 0.46381345604685426, "grad_norm": 1.071195363998413, "learning_rate": 6.660402816617974e-06, "loss": 0.2345, "step": 25500 }, { "epoch": 0.46399534367667655, "grad_norm": 0.9833534359931946, "learning_rate": 6.658078653571696e-06, "loss": 0.2367, "step": 25510 }, { "epoch": 0.46417723130649885, "grad_norm": 1.1518323421478271, "learning_rate": 6.655754262149314e-06, "loss": 0.2325, "step": 25520 }, { "epoch": 0.46435911893632115, "grad_norm": 1.0748077630996704, "learning_rate": 6.653429643157478e-06, "loss": 0.239, "step": 25530 }, { "epoch": 0.46454100656614344, "grad_norm": 1.0417370796203613, "learning_rate": 6.651104797402903e-06, "loss": 0.2373, "step": 25540 }, { "epoch": 0.46472289419596574, "grad_norm": 1.112727165222168, "learning_rate": 6.648779725692399e-06, "loss": 0.2336, "step": 25550 }, { "epoch": 0.46490478182578804, "grad_norm": 1.1188504695892334, "learning_rate": 6.646454428832843e-06, "loss": 0.2349, "step": 25560 }, { "epoch": 0.46508666945561034, "grad_norm": 1.2146960496902466, "learning_rate": 6.644128907631191e-06, "loss": 0.247, "step": 25570 }, { "epoch": 0.46526855708543263, "grad_norm": 1.0507484674453735, "learning_rate": 6.641803162894485e-06, "loss": 0.2362, "step": 25580 }, { "epoch": 0.46545044471525493, "grad_norm": 0.9122512936592102, "learning_rate": 6.6394771954298354e-06, "loss": 0.2283, "step": 25590 }, { "epoch": 0.4656323323450772, "grad_norm": 1.0337618589401245, "learning_rate": 6.637151006044435e-06, "loss": 0.2379, "step": 25600 }, { "epoch": 0.4658142199748995, "grad_norm": 1.0785592794418335, "learning_rate": 6.63482459554555e-06, "loss": 0.2362, "step": 25610 }, { "epoch": 0.4659961076047218, "grad_norm": 1.0663740634918213, "learning_rate": 6.632497964740527e-06, "loss": 0.2396, "step": 25620 }, { "epoch": 0.4661779952345441, "grad_norm": 0.9726884961128235, "learning_rate": 6.63017111443679e-06, "loss": 0.2393, "step": 25630 }, { "epoch": 0.4663598828643664, "grad_norm": 1.169529676437378, "learning_rate": 6.627844045441832e-06, "loss": 0.2316, "step": 25640 }, { "epoch": 0.4665417704941887, "grad_norm": 1.0955610275268555, "learning_rate": 6.62551675856323e-06, "loss": 0.2375, "step": 25650 }, { "epoch": 0.466723658124011, "grad_norm": 1.1322466135025024, "learning_rate": 6.623189254608632e-06, "loss": 0.236, "step": 25660 }, { "epoch": 0.4669055457538333, "grad_norm": 1.117027997970581, "learning_rate": 6.620861534385761e-06, "loss": 0.2313, "step": 25670 }, { "epoch": 0.4670874333836556, "grad_norm": 1.0804245471954346, "learning_rate": 6.61853359870242e-06, "loss": 0.237, "step": 25680 }, { "epoch": 0.4672693210134779, "grad_norm": 1.04487144947052, "learning_rate": 6.61620544836648e-06, "loss": 0.2361, "step": 25690 }, { "epoch": 0.46745120864330014, "grad_norm": 1.05718195438385, "learning_rate": 6.613877084185892e-06, "loss": 0.2411, "step": 25700 }, { "epoch": 0.46763309627312244, "grad_norm": 1.1874122619628906, "learning_rate": 6.611548506968679e-06, "loss": 0.2394, "step": 25710 }, { "epoch": 0.46781498390294474, "grad_norm": 1.136997938156128, "learning_rate": 6.6092197175229366e-06, "loss": 0.2344, "step": 25720 }, { "epoch": 0.46799687153276703, "grad_norm": 1.1017428636550903, "learning_rate": 6.606890716656838e-06, "loss": 0.2368, "step": 25730 }, { "epoch": 0.46817875916258933, "grad_norm": 1.0049110651016235, "learning_rate": 6.604561505178624e-06, "loss": 0.2367, "step": 25740 }, { "epoch": 0.46836064679241163, "grad_norm": 1.072199821472168, "learning_rate": 6.602232083896614e-06, "loss": 0.2369, "step": 25750 }, { "epoch": 0.4685425344222339, "grad_norm": 0.9965728521347046, "learning_rate": 6.599902453619196e-06, "loss": 0.2369, "step": 25760 }, { "epoch": 0.4687244220520562, "grad_norm": 0.9186218976974487, "learning_rate": 6.597572615154835e-06, "loss": 0.2249, "step": 25770 }, { "epoch": 0.4689063096818785, "grad_norm": 1.032034158706665, "learning_rate": 6.595242569312064e-06, "loss": 0.2372, "step": 25780 }, { "epoch": 0.4690881973117008, "grad_norm": 0.9614871144294739, "learning_rate": 6.592912316899489e-06, "loss": 0.2341, "step": 25790 }, { "epoch": 0.4692700849415231, "grad_norm": 0.9836626648902893, "learning_rate": 6.590581858725792e-06, "loss": 0.2348, "step": 25800 }, { "epoch": 0.4694519725713454, "grad_norm": 1.1109294891357422, "learning_rate": 6.588251195599717e-06, "loss": 0.2385, "step": 25810 }, { "epoch": 0.4696338602011677, "grad_norm": 1.1146931648254395, "learning_rate": 6.585920328330091e-06, "loss": 0.2378, "step": 25820 }, { "epoch": 0.46981574783099, "grad_norm": 1.0089303255081177, "learning_rate": 6.583589257725801e-06, "loss": 0.2389, "step": 25830 }, { "epoch": 0.4699976354608123, "grad_norm": 1.192295789718628, "learning_rate": 6.581257984595811e-06, "loss": 0.2325, "step": 25840 }, { "epoch": 0.4701795230906346, "grad_norm": 1.0589085817337036, "learning_rate": 6.5789265097491525e-06, "loss": 0.2367, "step": 25850 }, { "epoch": 0.4703614107204569, "grad_norm": 1.0319956541061401, "learning_rate": 6.576594833994932e-06, "loss": 0.2344, "step": 25860 }, { "epoch": 0.4705432983502792, "grad_norm": 1.2379382848739624, "learning_rate": 6.574262958142316e-06, "loss": 0.2377, "step": 25870 }, { "epoch": 0.4707251859801015, "grad_norm": 1.0880241394042969, "learning_rate": 6.571930883000552e-06, "loss": 0.2407, "step": 25880 }, { "epoch": 0.4709070736099238, "grad_norm": 1.0422449111938477, "learning_rate": 6.5695986093789465e-06, "loss": 0.2419, "step": 25890 }, { "epoch": 0.4710889612397461, "grad_norm": 1.009684443473816, "learning_rate": 6.567266138086883e-06, "loss": 0.2399, "step": 25900 }, { "epoch": 0.4712708488695684, "grad_norm": 1.1355949640274048, "learning_rate": 6.564933469933805e-06, "loss": 0.2377, "step": 25910 }, { "epoch": 0.4714527364993907, "grad_norm": 1.041075587272644, "learning_rate": 6.562600605729234e-06, "loss": 0.2327, "step": 25920 }, { "epoch": 0.471634624129213, "grad_norm": 1.0004427433013916, "learning_rate": 6.560267546282753e-06, "loss": 0.2375, "step": 25930 }, { "epoch": 0.4718165117590353, "grad_norm": 1.0821400880813599, "learning_rate": 6.557934292404013e-06, "loss": 0.2351, "step": 25940 }, { "epoch": 0.47199839938885757, "grad_norm": 0.9858945608139038, "learning_rate": 6.555600844902736e-06, "loss": 0.2347, "step": 25950 }, { "epoch": 0.47218028701867987, "grad_norm": 1.0183491706848145, "learning_rate": 6.553267204588707e-06, "loss": 0.2374, "step": 25960 }, { "epoch": 0.47236217464850216, "grad_norm": 1.045904517173767, "learning_rate": 6.550933372271784e-06, "loss": 0.2326, "step": 25970 }, { "epoch": 0.47254406227832446, "grad_norm": 1.0267066955566406, "learning_rate": 6.548599348761884e-06, "loss": 0.2315, "step": 25980 }, { "epoch": 0.47272594990814676, "grad_norm": 0.9404084086418152, "learning_rate": 6.546265134868994e-06, "loss": 0.2404, "step": 25990 }, { "epoch": 0.47290783753796906, "grad_norm": 1.1369411945343018, "learning_rate": 6.54393073140317e-06, "loss": 0.2345, "step": 26000 }, { "epoch": 0.47308972516779135, "grad_norm": 1.135114073753357, "learning_rate": 6.541596139174528e-06, "loss": 0.2372, "step": 26010 }, { "epoch": 0.47327161279761365, "grad_norm": 1.131955623626709, "learning_rate": 6.539261358993254e-06, "loss": 0.2407, "step": 26020 }, { "epoch": 0.47345350042743595, "grad_norm": 0.9883612990379333, "learning_rate": 6.536926391669599e-06, "loss": 0.2386, "step": 26030 }, { "epoch": 0.47363538805725824, "grad_norm": 1.0821858644485474, "learning_rate": 6.534591238013874e-06, "loss": 0.2407, "step": 26040 }, { "epoch": 0.47381727568708054, "grad_norm": 1.1800668239593506, "learning_rate": 6.532255898836461e-06, "loss": 0.233, "step": 26050 }, { "epoch": 0.47399916331690284, "grad_norm": 1.1346042156219482, "learning_rate": 6.529920374947802e-06, "loss": 0.2349, "step": 26060 }, { "epoch": 0.47418105094672514, "grad_norm": 0.9291650056838989, "learning_rate": 6.5275846671584065e-06, "loss": 0.2338, "step": 26070 }, { "epoch": 0.47436293857654743, "grad_norm": 1.0273094177246094, "learning_rate": 6.5252487762788455e-06, "loss": 0.2321, "step": 26080 }, { "epoch": 0.47454482620636973, "grad_norm": 1.0984312295913696, "learning_rate": 6.5229127031197525e-06, "loss": 0.2413, "step": 26090 }, { "epoch": 0.474726713836192, "grad_norm": 0.8594556450843811, "learning_rate": 6.520576448491829e-06, "loss": 0.2334, "step": 26100 }, { "epoch": 0.4749086014660143, "grad_norm": 1.1420360803604126, "learning_rate": 6.518240013205832e-06, "loss": 0.2408, "step": 26110 }, { "epoch": 0.47509048909583657, "grad_norm": 0.9773789644241333, "learning_rate": 6.515903398072589e-06, "loss": 0.2341, "step": 26120 }, { "epoch": 0.47527237672565886, "grad_norm": 1.0133914947509766, "learning_rate": 6.513566603902985e-06, "loss": 0.2356, "step": 26130 }, { "epoch": 0.47545426435548116, "grad_norm": 1.0969163179397583, "learning_rate": 6.511229631507969e-06, "loss": 0.2379, "step": 26140 }, { "epoch": 0.47563615198530346, "grad_norm": 1.0361448526382446, "learning_rate": 6.508892481698552e-06, "loss": 0.235, "step": 26150 }, { "epoch": 0.47581803961512575, "grad_norm": 1.1899969577789307, "learning_rate": 6.506555155285806e-06, "loss": 0.2348, "step": 26160 }, { "epoch": 0.47599992724494805, "grad_norm": 1.1260524988174438, "learning_rate": 6.504217653080862e-06, "loss": 0.2367, "step": 26170 }, { "epoch": 0.47618181487477035, "grad_norm": 0.9891395568847656, "learning_rate": 6.501879975894917e-06, "loss": 0.2285, "step": 26180 }, { "epoch": 0.47636370250459265, "grad_norm": 1.1766180992126465, "learning_rate": 6.499542124539225e-06, "loss": 0.2367, "step": 26190 }, { "epoch": 0.47654559013441494, "grad_norm": 0.9628646969795227, "learning_rate": 6.497204099825101e-06, "loss": 0.2343, "step": 26200 }, { "epoch": 0.47672747776423724, "grad_norm": 0.9696711897850037, "learning_rate": 6.49486590256392e-06, "loss": 0.2391, "step": 26210 }, { "epoch": 0.47690936539405954, "grad_norm": 1.0831176042556763, "learning_rate": 6.49252753356712e-06, "loss": 0.2409, "step": 26220 }, { "epoch": 0.47709125302388183, "grad_norm": 1.053666353225708, "learning_rate": 6.490188993646193e-06, "loss": 0.2369, "step": 26230 }, { "epoch": 0.47727314065370413, "grad_norm": 1.1069592237472534, "learning_rate": 6.487850283612697e-06, "loss": 0.2401, "step": 26240 }, { "epoch": 0.47745502828352643, "grad_norm": 1.0857397317886353, "learning_rate": 6.485511404278242e-06, "loss": 0.2396, "step": 26250 }, { "epoch": 0.4776369159133487, "grad_norm": 0.9353254437446594, "learning_rate": 6.483172356454501e-06, "loss": 0.2294, "step": 26260 }, { "epoch": 0.477818803543171, "grad_norm": 1.2247776985168457, "learning_rate": 6.480833140953208e-06, "loss": 0.2367, "step": 26270 }, { "epoch": 0.4780006911729933, "grad_norm": 1.1775330305099487, "learning_rate": 6.478493758586147e-06, "loss": 0.2416, "step": 26280 }, { "epoch": 0.4781825788028156, "grad_norm": 1.0620237588882446, "learning_rate": 6.476154210165168e-06, "loss": 0.2345, "step": 26290 }, { "epoch": 0.4783644664326379, "grad_norm": 1.0487083196640015, "learning_rate": 6.473814496502174e-06, "loss": 0.2288, "step": 26300 }, { "epoch": 0.4785463540624602, "grad_norm": 0.993137001991272, "learning_rate": 6.471474618409126e-06, "loss": 0.233, "step": 26310 }, { "epoch": 0.4787282416922825, "grad_norm": 1.1323915719985962, "learning_rate": 6.4691345766980455e-06, "loss": 0.2305, "step": 26320 }, { "epoch": 0.4789101293221048, "grad_norm": 1.1092647314071655, "learning_rate": 6.466794372181004e-06, "loss": 0.2342, "step": 26330 }, { "epoch": 0.4790920169519271, "grad_norm": 1.0640127658843994, "learning_rate": 6.464454005670139e-06, "loss": 0.2338, "step": 26340 }, { "epoch": 0.4792739045817494, "grad_norm": 0.9933304190635681, "learning_rate": 6.462113477977635e-06, "loss": 0.231, "step": 26350 }, { "epoch": 0.4794557922115717, "grad_norm": 1.1071678400039673, "learning_rate": 6.459772789915736e-06, "loss": 0.2387, "step": 26360 }, { "epoch": 0.479637679841394, "grad_norm": 1.0868339538574219, "learning_rate": 6.457431942296744e-06, "loss": 0.2329, "step": 26370 }, { "epoch": 0.4798195674712163, "grad_norm": 1.0917006731033325, "learning_rate": 6.455090935933012e-06, "loss": 0.2353, "step": 26380 }, { "epoch": 0.4800014551010386, "grad_norm": 1.0614556074142456, "learning_rate": 6.452749771636952e-06, "loss": 0.2362, "step": 26390 }, { "epoch": 0.4801833427308609, "grad_norm": 1.0863940715789795, "learning_rate": 6.45040845022103e-06, "loss": 0.2318, "step": 26400 }, { "epoch": 0.4803652303606832, "grad_norm": 1.0338623523712158, "learning_rate": 6.448066972497762e-06, "loss": 0.2402, "step": 26410 }, { "epoch": 0.4805471179905055, "grad_norm": 0.9126060009002686, "learning_rate": 6.4457253392797255e-06, "loss": 0.2353, "step": 26420 }, { "epoch": 0.4807290056203278, "grad_norm": 1.003967523574829, "learning_rate": 6.443383551379547e-06, "loss": 0.2346, "step": 26430 }, { "epoch": 0.4809108932501501, "grad_norm": 1.893058180809021, "learning_rate": 6.441041609609909e-06, "loss": 0.235, "step": 26440 }, { "epoch": 0.48109278087997237, "grad_norm": 1.0395245552062988, "learning_rate": 6.4386995147835466e-06, "loss": 0.2339, "step": 26450 }, { "epoch": 0.48127466850979467, "grad_norm": 0.946892499923706, "learning_rate": 6.436357267713245e-06, "loss": 0.2325, "step": 26460 }, { "epoch": 0.48145655613961696, "grad_norm": 1.0001158714294434, "learning_rate": 6.434014869211847e-06, "loss": 0.2388, "step": 26470 }, { "epoch": 0.48163844376943926, "grad_norm": 1.1603224277496338, "learning_rate": 6.431672320092246e-06, "loss": 0.2369, "step": 26480 }, { "epoch": 0.48182033139926156, "grad_norm": 1.0530019998550415, "learning_rate": 6.429329621167391e-06, "loss": 0.2305, "step": 26490 }, { "epoch": 0.48200221902908386, "grad_norm": 0.9836006164550781, "learning_rate": 6.4269867732502735e-06, "loss": 0.2361, "step": 26500 }, { "epoch": 0.48218410665890615, "grad_norm": 1.077296257019043, "learning_rate": 6.4246437771539476e-06, "loss": 0.2422, "step": 26510 }, { "epoch": 0.48236599428872845, "grad_norm": 1.1514861583709717, "learning_rate": 6.422300633691511e-06, "loss": 0.2385, "step": 26520 }, { "epoch": 0.4825478819185507, "grad_norm": 1.0725198984146118, "learning_rate": 6.419957343676119e-06, "loss": 0.2301, "step": 26530 }, { "epoch": 0.482729769548373, "grad_norm": 1.0643237829208374, "learning_rate": 6.417613907920972e-06, "loss": 0.2327, "step": 26540 }, { "epoch": 0.4829116571781953, "grad_norm": 1.2034404277801514, "learning_rate": 6.415270327239326e-06, "loss": 0.2415, "step": 26550 }, { "epoch": 0.4830935448080176, "grad_norm": 0.9862958788871765, "learning_rate": 6.41292660244448e-06, "loss": 0.2377, "step": 26560 }, { "epoch": 0.4832754324378399, "grad_norm": 0.9280201196670532, "learning_rate": 6.410582734349793e-06, "loss": 0.2362, "step": 26570 }, { "epoch": 0.4834573200676622, "grad_norm": 1.5178390741348267, "learning_rate": 6.408238723768665e-06, "loss": 0.2349, "step": 26580 }, { "epoch": 0.4836392076974845, "grad_norm": 1.0117470026016235, "learning_rate": 6.40589457151455e-06, "loss": 0.2371, "step": 26590 }, { "epoch": 0.48382109532730677, "grad_norm": 1.0714900493621826, "learning_rate": 6.403550278400951e-06, "loss": 0.2327, "step": 26600 }, { "epoch": 0.48400298295712907, "grad_norm": 1.0333285331726074, "learning_rate": 6.4012058452414185e-06, "loss": 0.2398, "step": 26610 }, { "epoch": 0.48418487058695137, "grad_norm": 0.8940795063972473, "learning_rate": 6.39886127284955e-06, "loss": 0.2374, "step": 26620 }, { "epoch": 0.48436675821677366, "grad_norm": 1.0531280040740967, "learning_rate": 6.396516562038997e-06, "loss": 0.2381, "step": 26630 }, { "epoch": 0.48454864584659596, "grad_norm": 1.0678032636642456, "learning_rate": 6.3941717136234526e-06, "loss": 0.2352, "step": 26640 }, { "epoch": 0.48473053347641826, "grad_norm": 0.999147355556488, "learning_rate": 6.391826728416661e-06, "loss": 0.2374, "step": 26650 }, { "epoch": 0.48491242110624055, "grad_norm": 1.0982682704925537, "learning_rate": 6.389481607232414e-06, "loss": 0.2372, "step": 26660 }, { "epoch": 0.48509430873606285, "grad_norm": 0.9629616737365723, "learning_rate": 6.38713635088455e-06, "loss": 0.2344, "step": 26670 }, { "epoch": 0.48527619636588515, "grad_norm": 1.12167227268219, "learning_rate": 6.3847909601869525e-06, "loss": 0.2355, "step": 26680 }, { "epoch": 0.48545808399570745, "grad_norm": 1.0228084325790405, "learning_rate": 6.382445435953556e-06, "loss": 0.2365, "step": 26690 }, { "epoch": 0.48563997162552974, "grad_norm": 1.0440739393234253, "learning_rate": 6.380099778998337e-06, "loss": 0.2386, "step": 26700 }, { "epoch": 0.48582185925535204, "grad_norm": 1.0537091493606567, "learning_rate": 6.3777539901353215e-06, "loss": 0.2362, "step": 26710 }, { "epoch": 0.48600374688517434, "grad_norm": 1.0171797275543213, "learning_rate": 6.375408070178577e-06, "loss": 0.2407, "step": 26720 }, { "epoch": 0.48618563451499663, "grad_norm": 1.11215341091156, "learning_rate": 6.3730620199422195e-06, "loss": 0.2308, "step": 26730 }, { "epoch": 0.48636752214481893, "grad_norm": 1.0709608793258667, "learning_rate": 6.370715840240412e-06, "loss": 0.2395, "step": 26740 }, { "epoch": 0.48654940977464123, "grad_norm": 1.0064561367034912, "learning_rate": 6.36836953188736e-06, "loss": 0.2368, "step": 26750 }, { "epoch": 0.4867312974044635, "grad_norm": 1.1293060779571533, "learning_rate": 6.36602309569731e-06, "loss": 0.2394, "step": 26760 }, { "epoch": 0.4869131850342858, "grad_norm": 1.072532057762146, "learning_rate": 6.363676532484559e-06, "loss": 0.2382, "step": 26770 }, { "epoch": 0.4870950726641081, "grad_norm": 1.241718053817749, "learning_rate": 6.361329843063448e-06, "loss": 0.2411, "step": 26780 }, { "epoch": 0.4872769602939304, "grad_norm": 1.1025112867355347, "learning_rate": 6.358983028248358e-06, "loss": 0.2374, "step": 26790 }, { "epoch": 0.4874588479237527, "grad_norm": 1.0322685241699219, "learning_rate": 6.356636088853712e-06, "loss": 0.2283, "step": 26800 }, { "epoch": 0.487640735553575, "grad_norm": 1.1131837368011475, "learning_rate": 6.354289025693982e-06, "loss": 0.2326, "step": 26810 }, { "epoch": 0.4878226231833973, "grad_norm": 0.9023326635360718, "learning_rate": 6.351941839583683e-06, "loss": 0.2315, "step": 26820 }, { "epoch": 0.4880045108132196, "grad_norm": 1.1285479068756104, "learning_rate": 6.3495945313373655e-06, "loss": 0.2326, "step": 26830 }, { "epoch": 0.4881863984430419, "grad_norm": 1.0418171882629395, "learning_rate": 6.347247101769629e-06, "loss": 0.2323, "step": 26840 }, { "epoch": 0.4883682860728642, "grad_norm": 1.1477363109588623, "learning_rate": 6.3448995516951114e-06, "loss": 0.2353, "step": 26850 }, { "epoch": 0.4885501737026865, "grad_norm": 1.0301620960235596, "learning_rate": 6.342551881928496e-06, "loss": 0.2294, "step": 26860 }, { "epoch": 0.4887320613325088, "grad_norm": 0.947162926197052, "learning_rate": 6.340204093284504e-06, "loss": 0.2352, "step": 26870 }, { "epoch": 0.4889139489623311, "grad_norm": 1.1173691749572754, "learning_rate": 6.337856186577901e-06, "loss": 0.2384, "step": 26880 }, { "epoch": 0.4890958365921534, "grad_norm": 1.065893292427063, "learning_rate": 6.335508162623492e-06, "loss": 0.2358, "step": 26890 }, { "epoch": 0.4892777242219757, "grad_norm": 1.0829477310180664, "learning_rate": 6.333160022236118e-06, "loss": 0.2361, "step": 26900 }, { "epoch": 0.489459611851798, "grad_norm": 1.0239745378494263, "learning_rate": 6.330811766230671e-06, "loss": 0.2329, "step": 26910 }, { "epoch": 0.4896414994816203, "grad_norm": 1.136513590812683, "learning_rate": 6.328463395422074e-06, "loss": 0.2323, "step": 26920 }, { "epoch": 0.4898233871114426, "grad_norm": 1.0624431371688843, "learning_rate": 6.326114910625295e-06, "loss": 0.2338, "step": 26930 }, { "epoch": 0.4900052747412649, "grad_norm": 1.112166404724121, "learning_rate": 6.3237663126553385e-06, "loss": 0.2314, "step": 26940 }, { "epoch": 0.4901871623710871, "grad_norm": 1.1055281162261963, "learning_rate": 6.321417602327249e-06, "loss": 0.2414, "step": 26950 }, { "epoch": 0.4903690500009094, "grad_norm": 0.9759424328804016, "learning_rate": 6.319068780456112e-06, "loss": 0.2367, "step": 26960 }, { "epoch": 0.4905509376307317, "grad_norm": 1.0448808670043945, "learning_rate": 6.316719847857048e-06, "loss": 0.24, "step": 26970 }, { "epoch": 0.490732825260554, "grad_norm": 1.0821455717086792, "learning_rate": 6.31437080534522e-06, "loss": 0.2337, "step": 26980 }, { "epoch": 0.4909147128903763, "grad_norm": 1.1531460285186768, "learning_rate": 6.312021653735827e-06, "loss": 0.2308, "step": 26990 }, { "epoch": 0.4910966005201986, "grad_norm": 0.9505715370178223, "learning_rate": 6.309672393844104e-06, "loss": 0.2373, "step": 27000 }, { "epoch": 0.4912784881500209, "grad_norm": 1.0728135108947754, "learning_rate": 6.307323026485328e-06, "loss": 0.2454, "step": 27010 }, { "epoch": 0.4914603757798432, "grad_norm": 1.119097113609314, "learning_rate": 6.304973552474808e-06, "loss": 0.2429, "step": 27020 }, { "epoch": 0.4916422634096655, "grad_norm": 0.9930291175842285, "learning_rate": 6.302623972627897e-06, "loss": 0.2374, "step": 27030 }, { "epoch": 0.4918241510394878, "grad_norm": 1.0245212316513062, "learning_rate": 6.3002742877599775e-06, "loss": 0.2391, "step": 27040 }, { "epoch": 0.4920060386693101, "grad_norm": 1.1050933599472046, "learning_rate": 6.297924498686472e-06, "loss": 0.2293, "step": 27050 }, { "epoch": 0.4921879262991324, "grad_norm": 1.1041287183761597, "learning_rate": 6.295574606222838e-06, "loss": 0.2379, "step": 27060 }, { "epoch": 0.4923698139289547, "grad_norm": 1.0606625080108643, "learning_rate": 6.293224611184573e-06, "loss": 0.2374, "step": 27070 }, { "epoch": 0.492551701558777, "grad_norm": 1.1404308080673218, "learning_rate": 6.290874514387203e-06, "loss": 0.2342, "step": 27080 }, { "epoch": 0.4927335891885993, "grad_norm": 1.0447267293930054, "learning_rate": 6.288524316646296e-06, "loss": 0.2324, "step": 27090 }, { "epoch": 0.49291547681842157, "grad_norm": 0.987555980682373, "learning_rate": 6.28617401877745e-06, "loss": 0.233, "step": 27100 }, { "epoch": 0.49309736444824387, "grad_norm": 1.24008047580719, "learning_rate": 6.283823621596301e-06, "loss": 0.2301, "step": 27110 }, { "epoch": 0.49327925207806617, "grad_norm": 1.033390760421753, "learning_rate": 6.281473125918518e-06, "loss": 0.2284, "step": 27120 }, { "epoch": 0.49346113970788846, "grad_norm": 0.9998188614845276, "learning_rate": 6.279122532559804e-06, "loss": 0.237, "step": 27130 }, { "epoch": 0.49364302733771076, "grad_norm": 1.0960620641708374, "learning_rate": 6.276771842335897e-06, "loss": 0.2368, "step": 27140 }, { "epoch": 0.49382491496753306, "grad_norm": 0.9821438193321228, "learning_rate": 6.2744210560625675e-06, "loss": 0.2331, "step": 27150 }, { "epoch": 0.49400680259735535, "grad_norm": 1.0667589902877808, "learning_rate": 6.27207017455562e-06, "loss": 0.2319, "step": 27160 }, { "epoch": 0.49418869022717765, "grad_norm": 1.1129165887832642, "learning_rate": 6.26971919863089e-06, "loss": 0.2333, "step": 27170 }, { "epoch": 0.49437057785699995, "grad_norm": 1.1574243307113647, "learning_rate": 6.267368129104252e-06, "loss": 0.2377, "step": 27180 }, { "epoch": 0.49455246548682225, "grad_norm": 1.0541948080062866, "learning_rate": 6.265016966791605e-06, "loss": 0.2319, "step": 27190 }, { "epoch": 0.49473435311664454, "grad_norm": 1.0495550632476807, "learning_rate": 6.262665712508884e-06, "loss": 0.2362, "step": 27200 }, { "epoch": 0.49491624074646684, "grad_norm": 0.9525271058082581, "learning_rate": 6.260314367072059e-06, "loss": 0.2362, "step": 27210 }, { "epoch": 0.49509812837628914, "grad_norm": 0.985342264175415, "learning_rate": 6.2579629312971234e-06, "loss": 0.2311, "step": 27220 }, { "epoch": 0.49528001600611143, "grad_norm": 1.0027285814285278, "learning_rate": 6.255611406000112e-06, "loss": 0.2351, "step": 27230 }, { "epoch": 0.49546190363593373, "grad_norm": 1.0518122911453247, "learning_rate": 6.253259791997082e-06, "loss": 0.2309, "step": 27240 }, { "epoch": 0.49564379126575603, "grad_norm": 1.0443973541259766, "learning_rate": 6.250908090104124e-06, "loss": 0.2325, "step": 27250 }, { "epoch": 0.4958256788955783, "grad_norm": 1.0319820642471313, "learning_rate": 6.248556301137365e-06, "loss": 0.2364, "step": 27260 }, { "epoch": 0.4960075665254006, "grad_norm": 1.074689269065857, "learning_rate": 6.246204425912953e-06, "loss": 0.2373, "step": 27270 }, { "epoch": 0.4961894541552229, "grad_norm": 0.9690266847610474, "learning_rate": 6.243852465247072e-06, "loss": 0.2349, "step": 27280 }, { "epoch": 0.4963713417850452, "grad_norm": 1.0596930980682373, "learning_rate": 6.241500419955934e-06, "loss": 0.2392, "step": 27290 }, { "epoch": 0.4965532294148675, "grad_norm": 1.0499018430709839, "learning_rate": 6.239148290855779e-06, "loss": 0.2312, "step": 27300 }, { "epoch": 0.4967351170446898, "grad_norm": 0.9947345852851868, "learning_rate": 6.236796078762879e-06, "loss": 0.2337, "step": 27310 }, { "epoch": 0.4969170046745121, "grad_norm": 1.1519755125045776, "learning_rate": 6.234443784493531e-06, "loss": 0.2351, "step": 27320 }, { "epoch": 0.4970988923043344, "grad_norm": 0.9566649794578552, "learning_rate": 6.232091408864065e-06, "loss": 0.2388, "step": 27330 }, { "epoch": 0.4972807799341567, "grad_norm": 1.0191950798034668, "learning_rate": 6.229738952690835e-06, "loss": 0.2335, "step": 27340 }, { "epoch": 0.497462667563979, "grad_norm": 1.0668261051177979, "learning_rate": 6.2273864167902256e-06, "loss": 0.2412, "step": 27350 }, { "epoch": 0.4976445551938013, "grad_norm": 1.0298593044281006, "learning_rate": 6.225033801978647e-06, "loss": 0.2334, "step": 27360 }, { "epoch": 0.49782644282362354, "grad_norm": 1.03029203414917, "learning_rate": 6.2226811090725405e-06, "loss": 0.2401, "step": 27370 }, { "epoch": 0.49800833045344584, "grad_norm": 1.0866119861602783, "learning_rate": 6.220328338888372e-06, "loss": 0.2364, "step": 27380 }, { "epoch": 0.49819021808326813, "grad_norm": 1.0183802843093872, "learning_rate": 6.217975492242632e-06, "loss": 0.2318, "step": 27390 }, { "epoch": 0.49837210571309043, "grad_norm": 1.167044758796692, "learning_rate": 6.21562256995184e-06, "loss": 0.2379, "step": 27400 }, { "epoch": 0.4985539933429127, "grad_norm": 1.1368356943130493, "learning_rate": 6.213269572832546e-06, "loss": 0.2389, "step": 27410 }, { "epoch": 0.498735880972735, "grad_norm": 1.0248609781265259, "learning_rate": 6.210916501701315e-06, "loss": 0.2298, "step": 27420 }, { "epoch": 0.4989177686025573, "grad_norm": 0.9623903036117554, "learning_rate": 6.208563357374749e-06, "loss": 0.2276, "step": 27430 }, { "epoch": 0.4990996562323796, "grad_norm": 1.2235732078552246, "learning_rate": 6.2062101406694685e-06, "loss": 0.2311, "step": 27440 }, { "epoch": 0.4992815438622019, "grad_norm": 1.042500376701355, "learning_rate": 6.203856852402121e-06, "loss": 0.2371, "step": 27450 }, { "epoch": 0.4994634314920242, "grad_norm": 1.085516095161438, "learning_rate": 6.201503493389382e-06, "loss": 0.2442, "step": 27460 }, { "epoch": 0.4996453191218465, "grad_norm": 0.9982538223266602, "learning_rate": 6.199150064447945e-06, "loss": 0.2389, "step": 27470 }, { "epoch": 0.4998272067516688, "grad_norm": 0.9843856692314148, "learning_rate": 6.196796566394534e-06, "loss": 0.2368, "step": 27480 }, { "epoch": 0.5000090943814911, "grad_norm": 1.09203040599823, "learning_rate": 6.194443000045891e-06, "loss": 0.2336, "step": 27490 }, { "epoch": 0.5001909820113134, "grad_norm": 1.0648736953735352, "learning_rate": 6.19208936621879e-06, "loss": 0.2378, "step": 27500 }, { "epoch": 0.5003728696411357, "grad_norm": 0.8945997357368469, "learning_rate": 6.189735665730021e-06, "loss": 0.234, "step": 27510 }, { "epoch": 0.500554757270958, "grad_norm": 1.0950134992599487, "learning_rate": 6.187381899396399e-06, "loss": 0.2317, "step": 27520 }, { "epoch": 0.5007366449007803, "grad_norm": 0.9360733032226562, "learning_rate": 6.185028068034762e-06, "loss": 0.2349, "step": 27530 }, { "epoch": 0.5009185325306026, "grad_norm": 1.0350356101989746, "learning_rate": 6.1826741724619735e-06, "loss": 0.2357, "step": 27540 }, { "epoch": 0.5011004201604249, "grad_norm": 0.9936109781265259, "learning_rate": 6.180320213494916e-06, "loss": 0.2366, "step": 27550 }, { "epoch": 0.5012823077902472, "grad_norm": 1.0285481214523315, "learning_rate": 6.177966191950493e-06, "loss": 0.234, "step": 27560 }, { "epoch": 0.5014641954200695, "grad_norm": 1.1668871641159058, "learning_rate": 6.175612108645634e-06, "loss": 0.2323, "step": 27570 }, { "epoch": 0.5016460830498918, "grad_norm": 1.0140082836151123, "learning_rate": 6.1732579643972875e-06, "loss": 0.2356, "step": 27580 }, { "epoch": 0.5018279706797141, "grad_norm": 1.1001384258270264, "learning_rate": 6.17090376002242e-06, "loss": 0.2359, "step": 27590 }, { "epoch": 0.5020098583095364, "grad_norm": 0.974944531917572, "learning_rate": 6.168549496338026e-06, "loss": 0.2359, "step": 27600 }, { "epoch": 0.5021917459393587, "grad_norm": 0.9938240647315979, "learning_rate": 6.1661951741611124e-06, "loss": 0.235, "step": 27610 }, { "epoch": 0.502373633569181, "grad_norm": 1.0340303182601929, "learning_rate": 6.163840794308715e-06, "loss": 0.2339, "step": 27620 }, { "epoch": 0.5025555211990033, "grad_norm": 1.0935347080230713, "learning_rate": 6.161486357597882e-06, "loss": 0.2355, "step": 27630 }, { "epoch": 0.5027374088288256, "grad_norm": 1.0289702415466309, "learning_rate": 6.159131864845686e-06, "loss": 0.2366, "step": 27640 }, { "epoch": 0.5029192964586479, "grad_norm": 1.0110191106796265, "learning_rate": 6.156777316869218e-06, "loss": 0.2288, "step": 27650 }, { "epoch": 0.5031011840884702, "grad_norm": 0.9451196789741516, "learning_rate": 6.154422714485586e-06, "loss": 0.2334, "step": 27660 }, { "epoch": 0.5032830717182925, "grad_norm": 1.103176474571228, "learning_rate": 6.152068058511921e-06, "loss": 0.2331, "step": 27670 }, { "epoch": 0.5034649593481147, "grad_norm": 1.0115736722946167, "learning_rate": 6.149713349765369e-06, "loss": 0.2338, "step": 27680 }, { "epoch": 0.503646846977937, "grad_norm": 1.0313315391540527, "learning_rate": 6.147358589063094e-06, "loss": 0.2352, "step": 27690 }, { "epoch": 0.5038287346077593, "grad_norm": 0.9711647629737854, "learning_rate": 6.145003777222283e-06, "loss": 0.2338, "step": 27700 }, { "epoch": 0.5040106222375816, "grad_norm": 1.1609171628952026, "learning_rate": 6.142648915060132e-06, "loss": 0.2375, "step": 27710 }, { "epoch": 0.5041925098674039, "grad_norm": 1.0881685018539429, "learning_rate": 6.140294003393868e-06, "loss": 0.2338, "step": 27720 }, { "epoch": 0.5043743974972262, "grad_norm": 1.1539689302444458, "learning_rate": 6.137939043040718e-06, "loss": 0.2322, "step": 27730 }, { "epoch": 0.5045562851270485, "grad_norm": 1.1117377281188965, "learning_rate": 6.13558403481794e-06, "loss": 0.2343, "step": 27740 }, { "epoch": 0.5047381727568708, "grad_norm": 1.174024224281311, "learning_rate": 6.133228979542804e-06, "loss": 0.2386, "step": 27750 }, { "epoch": 0.5049200603866931, "grad_norm": 0.9412002563476562, "learning_rate": 6.1308738780325925e-06, "loss": 0.2349, "step": 27760 }, { "epoch": 0.5051019480165154, "grad_norm": 1.002962350845337, "learning_rate": 6.12851873110461e-06, "loss": 0.2306, "step": 27770 }, { "epoch": 0.5052838356463377, "grad_norm": 1.0405523777008057, "learning_rate": 6.1261635395761746e-06, "loss": 0.2389, "step": 27780 }, { "epoch": 0.50546572327616, "grad_norm": 1.1368600130081177, "learning_rate": 6.123808304264616e-06, "loss": 0.2376, "step": 27790 }, { "epoch": 0.5056476109059823, "grad_norm": 1.0969663858413696, "learning_rate": 6.1214530259872854e-06, "loss": 0.23, "step": 27800 }, { "epoch": 0.5058294985358046, "grad_norm": 1.045153260231018, "learning_rate": 6.119097705561546e-06, "loss": 0.2346, "step": 27810 }, { "epoch": 0.5060113861656269, "grad_norm": 1.1298222541809082, "learning_rate": 6.116742343804774e-06, "loss": 0.2283, "step": 27820 }, { "epoch": 0.5061932737954492, "grad_norm": 1.012107491493225, "learning_rate": 6.114386941534361e-06, "loss": 0.2309, "step": 27830 }, { "epoch": 0.5063751614252715, "grad_norm": 0.934128999710083, "learning_rate": 6.112031499567716e-06, "loss": 0.2357, "step": 27840 }, { "epoch": 0.5065570490550938, "grad_norm": 0.9453412890434265, "learning_rate": 6.109676018722258e-06, "loss": 0.2331, "step": 27850 }, { "epoch": 0.5067389366849161, "grad_norm": 1.15156888961792, "learning_rate": 6.107320499815419e-06, "loss": 0.2364, "step": 27860 }, { "epoch": 0.5069208243147384, "grad_norm": 1.0187565088272095, "learning_rate": 6.1049649436646485e-06, "loss": 0.2317, "step": 27870 }, { "epoch": 0.5071027119445607, "grad_norm": 1.155007243156433, "learning_rate": 6.1026093510874015e-06, "loss": 0.2362, "step": 27880 }, { "epoch": 0.507284599574383, "grad_norm": 1.2311943769454956, "learning_rate": 6.100253722901156e-06, "loss": 0.2337, "step": 27890 }, { "epoch": 0.5074664872042053, "grad_norm": 1.0334389209747314, "learning_rate": 6.0978980599233925e-06, "loss": 0.2364, "step": 27900 }, { "epoch": 0.5076483748340276, "grad_norm": 0.9980051517486572, "learning_rate": 6.095542362971608e-06, "loss": 0.236, "step": 27910 }, { "epoch": 0.5078302624638499, "grad_norm": 0.9291595816612244, "learning_rate": 6.093186632863314e-06, "loss": 0.2378, "step": 27920 }, { "epoch": 0.5080121500936722, "grad_norm": 1.0597511529922485, "learning_rate": 6.090830870416028e-06, "loss": 0.2347, "step": 27930 }, { "epoch": 0.5081940377234945, "grad_norm": 1.0421253442764282, "learning_rate": 6.088475076447281e-06, "loss": 0.2314, "step": 27940 }, { "epoch": 0.5083759253533168, "grad_norm": 1.0308843851089478, "learning_rate": 6.0861192517746156e-06, "loss": 0.2368, "step": 27950 }, { "epoch": 0.5085578129831391, "grad_norm": 0.9460098743438721, "learning_rate": 6.083763397215585e-06, "loss": 0.2359, "step": 27960 }, { "epoch": 0.5087397006129614, "grad_norm": 0.9479824304580688, "learning_rate": 6.081407513587754e-06, "loss": 0.2335, "step": 27970 }, { "epoch": 0.5089215882427837, "grad_norm": 1.0459022521972656, "learning_rate": 6.079051601708693e-06, "loss": 0.2317, "step": 27980 }, { "epoch": 0.509103475872606, "grad_norm": 1.145302653312683, "learning_rate": 6.076695662395986e-06, "loss": 0.2339, "step": 27990 }, { "epoch": 0.5092853635024281, "grad_norm": 0.9731887578964233, "learning_rate": 6.074339696467226e-06, "loss": 0.2377, "step": 28000 }, { "epoch": 0.5094672511322504, "grad_norm": 1.0444252490997314, "learning_rate": 6.071983704740014e-06, "loss": 0.2368, "step": 28010 }, { "epoch": 0.5096491387620727, "grad_norm": 1.168859601020813, "learning_rate": 6.0696276880319636e-06, "loss": 0.2305, "step": 28020 }, { "epoch": 0.509831026391895, "grad_norm": 1.0365219116210938, "learning_rate": 6.0672716471606894e-06, "loss": 0.2384, "step": 28030 }, { "epoch": 0.5100129140217173, "grad_norm": 1.1377460956573486, "learning_rate": 6.0649155829438235e-06, "loss": 0.2328, "step": 28040 }, { "epoch": 0.5101948016515396, "grad_norm": 0.9943942427635193, "learning_rate": 6.062559496199001e-06, "loss": 0.2335, "step": 28050 }, { "epoch": 0.5103766892813619, "grad_norm": 1.0400941371917725, "learning_rate": 6.060203387743862e-06, "loss": 0.2324, "step": 28060 }, { "epoch": 0.5105585769111842, "grad_norm": 1.1199785470962524, "learning_rate": 6.057847258396065e-06, "loss": 0.2304, "step": 28070 }, { "epoch": 0.5107404645410065, "grad_norm": 1.1849287748336792, "learning_rate": 6.055491108973261e-06, "loss": 0.2353, "step": 28080 }, { "epoch": 0.5109223521708288, "grad_norm": 1.0458118915557861, "learning_rate": 6.053134940293121e-06, "loss": 0.2323, "step": 28090 }, { "epoch": 0.5111042398006511, "grad_norm": 1.0490484237670898, "learning_rate": 6.050778753173314e-06, "loss": 0.2351, "step": 28100 }, { "epoch": 0.5112861274304734, "grad_norm": 1.0128884315490723, "learning_rate": 6.048422548431519e-06, "loss": 0.2334, "step": 28110 }, { "epoch": 0.5114680150602957, "grad_norm": 1.090782642364502, "learning_rate": 6.046066326885423e-06, "loss": 0.2357, "step": 28120 }, { "epoch": 0.511649902690118, "grad_norm": 1.0474845170974731, "learning_rate": 6.043710089352714e-06, "loss": 0.2343, "step": 28130 }, { "epoch": 0.5118317903199403, "grad_norm": 1.012192726135254, "learning_rate": 6.041353836651088e-06, "loss": 0.2356, "step": 28140 }, { "epoch": 0.5120136779497626, "grad_norm": 1.0101693868637085, "learning_rate": 6.0389975695982484e-06, "loss": 0.2338, "step": 28150 }, { "epoch": 0.5121955655795849, "grad_norm": 1.0825836658477783, "learning_rate": 6.036641289011901e-06, "loss": 0.2307, "step": 28160 }, { "epoch": 0.5123774532094072, "grad_norm": 0.9538998007774353, "learning_rate": 6.034284995709756e-06, "loss": 0.2316, "step": 28170 }, { "epoch": 0.5125593408392295, "grad_norm": 1.00091552734375, "learning_rate": 6.031928690509528e-06, "loss": 0.2363, "step": 28180 }, { "epoch": 0.5127412284690518, "grad_norm": 0.9773130416870117, "learning_rate": 6.029572374228937e-06, "loss": 0.2283, "step": 28190 }, { "epoch": 0.5129231160988741, "grad_norm": 1.0651586055755615, "learning_rate": 6.027216047685709e-06, "loss": 0.2359, "step": 28200 }, { "epoch": 0.5131050037286964, "grad_norm": 0.9789493680000305, "learning_rate": 6.0248597116975695e-06, "loss": 0.2354, "step": 28210 }, { "epoch": 0.5132868913585187, "grad_norm": 0.9844257831573486, "learning_rate": 6.022503367082247e-06, "loss": 0.2331, "step": 28220 }, { "epoch": 0.513468778988341, "grad_norm": 1.061683177947998, "learning_rate": 6.020147014657476e-06, "loss": 0.2349, "step": 28230 }, { "epoch": 0.5136506666181633, "grad_norm": 1.1310045719146729, "learning_rate": 6.0177906552409935e-06, "loss": 0.2403, "step": 28240 }, { "epoch": 0.5138325542479856, "grad_norm": 1.1097134351730347, "learning_rate": 6.015434289650538e-06, "loss": 0.2333, "step": 28250 }, { "epoch": 0.5140144418778079, "grad_norm": 1.2118488550186157, "learning_rate": 6.013077918703848e-06, "loss": 0.2334, "step": 28260 }, { "epoch": 0.5141963295076302, "grad_norm": 1.1178908348083496, "learning_rate": 6.010721543218667e-06, "loss": 0.2355, "step": 28270 }, { "epoch": 0.5143782171374525, "grad_norm": 1.030941367149353, "learning_rate": 6.008365164012738e-06, "loss": 0.238, "step": 28280 }, { "epoch": 0.5145601047672748, "grad_norm": 1.0206964015960693, "learning_rate": 6.006008781903808e-06, "loss": 0.2322, "step": 28290 }, { "epoch": 0.5147419923970971, "grad_norm": 1.0243712663650513, "learning_rate": 6.003652397709622e-06, "loss": 0.2296, "step": 28300 }, { "epoch": 0.5149238800269194, "grad_norm": 1.0943278074264526, "learning_rate": 6.001296012247928e-06, "loss": 0.2326, "step": 28310 }, { "epoch": 0.5151057676567417, "grad_norm": 0.9322488307952881, "learning_rate": 5.998939626336473e-06, "loss": 0.2347, "step": 28320 }, { "epoch": 0.515287655286564, "grad_norm": 1.0744811296463013, "learning_rate": 5.996583240793004e-06, "loss": 0.2262, "step": 28330 }, { "epoch": 0.5154695429163862, "grad_norm": 1.0160642862319946, "learning_rate": 5.994226856435269e-06, "loss": 0.2377, "step": 28340 }, { "epoch": 0.5156514305462085, "grad_norm": 1.0481338500976562, "learning_rate": 5.991870474081014e-06, "loss": 0.2304, "step": 28350 }, { "epoch": 0.5158333181760308, "grad_norm": 1.0547266006469727, "learning_rate": 5.989514094547988e-06, "loss": 0.2374, "step": 28360 }, { "epoch": 0.5160152058058531, "grad_norm": 1.0406454801559448, "learning_rate": 5.987157718653933e-06, "loss": 0.2338, "step": 28370 }, { "epoch": 0.5161970934356754, "grad_norm": 1.1174136400222778, "learning_rate": 5.984801347216595e-06, "loss": 0.2382, "step": 28380 }, { "epoch": 0.5163789810654977, "grad_norm": 1.00761878490448, "learning_rate": 5.9824449810537156e-06, "loss": 0.233, "step": 28390 }, { "epoch": 0.51656086869532, "grad_norm": 1.0233840942382812, "learning_rate": 5.980088620983038e-06, "loss": 0.233, "step": 28400 }, { "epoch": 0.5167427563251423, "grad_norm": 1.0514453649520874, "learning_rate": 5.9777322678222985e-06, "loss": 0.2314, "step": 28410 }, { "epoch": 0.5169246439549646, "grad_norm": 1.1033023595809937, "learning_rate": 5.975375922389234e-06, "loss": 0.2407, "step": 28420 }, { "epoch": 0.5171065315847869, "grad_norm": 1.0737812519073486, "learning_rate": 5.973019585501577e-06, "loss": 0.2422, "step": 28430 }, { "epoch": 0.5172884192146092, "grad_norm": 0.9611364006996155, "learning_rate": 5.97066325797706e-06, "loss": 0.2309, "step": 28440 }, { "epoch": 0.5174703068444315, "grad_norm": 1.0496286153793335, "learning_rate": 5.968306940633408e-06, "loss": 0.2387, "step": 28450 }, { "epoch": 0.5176521944742538, "grad_norm": 0.9669097661972046, "learning_rate": 5.965950634288347e-06, "loss": 0.2331, "step": 28460 }, { "epoch": 0.5178340821040761, "grad_norm": 0.9400644898414612, "learning_rate": 5.963594339759598e-06, "loss": 0.238, "step": 28470 }, { "epoch": 0.5180159697338984, "grad_norm": 1.0730098485946655, "learning_rate": 5.961238057864872e-06, "loss": 0.2382, "step": 28480 }, { "epoch": 0.5181978573637207, "grad_norm": 0.9756695032119751, "learning_rate": 5.958881789421886e-06, "loss": 0.2311, "step": 28490 }, { "epoch": 0.518379744993543, "grad_norm": 0.9422895908355713, "learning_rate": 5.956525535248341e-06, "loss": 0.2312, "step": 28500 }, { "epoch": 0.5185616326233653, "grad_norm": 1.1865103244781494, "learning_rate": 5.954169296161944e-06, "loss": 0.2363, "step": 28510 }, { "epoch": 0.5187435202531876, "grad_norm": 0.9696475267410278, "learning_rate": 5.95181307298039e-06, "loss": 0.2358, "step": 28520 }, { "epoch": 0.5189254078830099, "grad_norm": 0.9959151744842529, "learning_rate": 5.9494568665213695e-06, "loss": 0.234, "step": 28530 }, { "epoch": 0.5191072955128322, "grad_norm": 0.9632995128631592, "learning_rate": 5.947100677602568e-06, "loss": 0.2342, "step": 28540 }, { "epoch": 0.5192891831426545, "grad_norm": 1.1029428243637085, "learning_rate": 5.944744507041665e-06, "loss": 0.2394, "step": 28550 }, { "epoch": 0.5194710707724768, "grad_norm": 1.1035962104797363, "learning_rate": 5.942388355656335e-06, "loss": 0.2358, "step": 28560 }, { "epoch": 0.5196529584022991, "grad_norm": 1.0363643169403076, "learning_rate": 5.9400322242642394e-06, "loss": 0.2361, "step": 28570 }, { "epoch": 0.5198348460321214, "grad_norm": 0.9355618953704834, "learning_rate": 5.937676113683042e-06, "loss": 0.2369, "step": 28580 }, { "epoch": 0.5200167336619437, "grad_norm": 0.9901108741760254, "learning_rate": 5.9353200247303935e-06, "loss": 0.2441, "step": 28590 }, { "epoch": 0.520198621291766, "grad_norm": 1.0443756580352783, "learning_rate": 5.932963958223937e-06, "loss": 0.2364, "step": 28600 }, { "epoch": 0.5203805089215883, "grad_norm": 1.1022025346755981, "learning_rate": 5.930607914981312e-06, "loss": 0.2405, "step": 28610 }, { "epoch": 0.5205623965514106, "grad_norm": 0.9746026992797852, "learning_rate": 5.928251895820145e-06, "loss": 0.232, "step": 28620 }, { "epoch": 0.5207442841812329, "grad_norm": 0.9750149846076965, "learning_rate": 5.925895901558057e-06, "loss": 0.2343, "step": 28630 }, { "epoch": 0.5209261718110552, "grad_norm": 1.0235331058502197, "learning_rate": 5.92353993301266e-06, "loss": 0.2392, "step": 28640 }, { "epoch": 0.5211080594408775, "grad_norm": 1.1489704847335815, "learning_rate": 5.921183991001555e-06, "loss": 0.238, "step": 28650 }, { "epoch": 0.5212899470706998, "grad_norm": 0.9754573106765747, "learning_rate": 5.91882807634234e-06, "loss": 0.235, "step": 28660 }, { "epoch": 0.521471834700522, "grad_norm": 1.0694761276245117, "learning_rate": 5.916472189852593e-06, "loss": 0.2427, "step": 28670 }, { "epoch": 0.5216537223303443, "grad_norm": 1.071991205215454, "learning_rate": 5.914116332349891e-06, "loss": 0.2304, "step": 28680 }, { "epoch": 0.5218356099601666, "grad_norm": 1.1246490478515625, "learning_rate": 5.9117605046518e-06, "loss": 0.2319, "step": 28690 }, { "epoch": 0.5220174975899889, "grad_norm": 1.0032109022140503, "learning_rate": 5.909404707575872e-06, "loss": 0.2344, "step": 28700 }, { "epoch": 0.5221993852198112, "grad_norm": 1.0275136232376099, "learning_rate": 5.90704894193965e-06, "loss": 0.2381, "step": 28710 }, { "epoch": 0.5223812728496335, "grad_norm": 1.0021271705627441, "learning_rate": 5.9046932085606675e-06, "loss": 0.2336, "step": 28720 }, { "epoch": 0.5225631604794558, "grad_norm": 0.9272720217704773, "learning_rate": 5.902337508256445e-06, "loss": 0.2348, "step": 28730 }, { "epoch": 0.5227450481092781, "grad_norm": 1.2459428310394287, "learning_rate": 5.8999818418444935e-06, "loss": 0.2351, "step": 28740 }, { "epoch": 0.5229269357391004, "grad_norm": 0.9717971086502075, "learning_rate": 5.897626210142307e-06, "loss": 0.2354, "step": 28750 }, { "epoch": 0.5231088233689227, "grad_norm": 1.0121790170669556, "learning_rate": 5.895270613967375e-06, "loss": 0.2347, "step": 28760 }, { "epoch": 0.523290710998745, "grad_norm": 1.0016226768493652, "learning_rate": 5.892915054137169e-06, "loss": 0.2378, "step": 28770 }, { "epoch": 0.5234725986285673, "grad_norm": 1.0644270181655884, "learning_rate": 5.890559531469154e-06, "loss": 0.2357, "step": 28780 }, { "epoch": 0.5236544862583896, "grad_norm": 1.0551998615264893, "learning_rate": 5.888204046780772e-06, "loss": 0.2415, "step": 28790 }, { "epoch": 0.5238363738882119, "grad_norm": 1.1688895225524902, "learning_rate": 5.885848600889461e-06, "loss": 0.235, "step": 28800 }, { "epoch": 0.5240182615180342, "grad_norm": 1.0597494840621948, "learning_rate": 5.883493194612643e-06, "loss": 0.231, "step": 28810 }, { "epoch": 0.5242001491478565, "grad_norm": 1.0534192323684692, "learning_rate": 5.8811378287677235e-06, "loss": 0.233, "step": 28820 }, { "epoch": 0.5243820367776787, "grad_norm": 1.026236653327942, "learning_rate": 5.878782504172098e-06, "loss": 0.2307, "step": 28830 }, { "epoch": 0.524563924407501, "grad_norm": 1.0780673027038574, "learning_rate": 5.876427221643145e-06, "loss": 0.2328, "step": 28840 }, { "epoch": 0.5247458120373233, "grad_norm": 1.0444656610488892, "learning_rate": 5.874071981998229e-06, "loss": 0.2279, "step": 28850 }, { "epoch": 0.5249276996671456, "grad_norm": 1.070549488067627, "learning_rate": 5.871716786054701e-06, "loss": 0.236, "step": 28860 }, { "epoch": 0.5251095872969679, "grad_norm": 1.0434720516204834, "learning_rate": 5.869361634629893e-06, "loss": 0.2385, "step": 28870 }, { "epoch": 0.5252914749267902, "grad_norm": 0.9673358201980591, "learning_rate": 5.867006528541126e-06, "loss": 0.2311, "step": 28880 }, { "epoch": 0.5254733625566125, "grad_norm": 1.0879775285720825, "learning_rate": 5.864651468605704e-06, "loss": 0.2333, "step": 28890 }, { "epoch": 0.5256552501864348, "grad_norm": 0.9107252359390259, "learning_rate": 5.8622964556409135e-06, "loss": 0.2298, "step": 28900 }, { "epoch": 0.5258371378162571, "grad_norm": 0.9896516799926758, "learning_rate": 5.859941490464025e-06, "loss": 0.2371, "step": 28910 }, { "epoch": 0.5260190254460794, "grad_norm": 0.9455205798149109, "learning_rate": 5.857586573892294e-06, "loss": 0.2298, "step": 28920 }, { "epoch": 0.5262009130759017, "grad_norm": 1.0034269094467163, "learning_rate": 5.855231706742959e-06, "loss": 0.2342, "step": 28930 }, { "epoch": 0.526382800705724, "grad_norm": 1.0246200561523438, "learning_rate": 5.852876889833238e-06, "loss": 0.2407, "step": 28940 }, { "epoch": 0.5265646883355463, "grad_norm": 1.0664772987365723, "learning_rate": 5.850522123980336e-06, "loss": 0.2339, "step": 28950 }, { "epoch": 0.5267465759653686, "grad_norm": 1.0619338750839233, "learning_rate": 5.848167410001439e-06, "loss": 0.2424, "step": 28960 }, { "epoch": 0.5269284635951909, "grad_norm": 1.0837416648864746, "learning_rate": 5.845812748713712e-06, "loss": 0.2278, "step": 28970 }, { "epoch": 0.5271103512250132, "grad_norm": 1.0308361053466797, "learning_rate": 5.843458140934306e-06, "loss": 0.238, "step": 28980 }, { "epoch": 0.5272922388548354, "grad_norm": 1.089249849319458, "learning_rate": 5.841103587480351e-06, "loss": 0.2351, "step": 28990 }, { "epoch": 0.5274741264846577, "grad_norm": 1.0258208513259888, "learning_rate": 5.838749089168959e-06, "loss": 0.2306, "step": 29000 }, { "epoch": 0.52765601411448, "grad_norm": 1.1229883432388306, "learning_rate": 5.836394646817222e-06, "loss": 0.2323, "step": 29010 }, { "epoch": 0.5278379017443023, "grad_norm": 1.0013298988342285, "learning_rate": 5.834040261242212e-06, "loss": 0.2316, "step": 29020 }, { "epoch": 0.5280197893741246, "grad_norm": 1.0482548475265503, "learning_rate": 5.831685933260985e-06, "loss": 0.2394, "step": 29030 }, { "epoch": 0.5282016770039469, "grad_norm": 1.0159963369369507, "learning_rate": 5.829331663690572e-06, "loss": 0.2304, "step": 29040 }, { "epoch": 0.5283835646337692, "grad_norm": 1.0169930458068848, "learning_rate": 5.826977453347988e-06, "loss": 0.2307, "step": 29050 }, { "epoch": 0.5285654522635915, "grad_norm": 1.0697216987609863, "learning_rate": 5.824623303050225e-06, "loss": 0.229, "step": 29060 }, { "epoch": 0.5287473398934138, "grad_norm": 0.966858446598053, "learning_rate": 5.822269213614253e-06, "loss": 0.2348, "step": 29070 }, { "epoch": 0.5289292275232361, "grad_norm": 1.0246540307998657, "learning_rate": 5.819915185857026e-06, "loss": 0.2314, "step": 29080 }, { "epoch": 0.5291111151530584, "grad_norm": 1.2161870002746582, "learning_rate": 5.8175612205954704e-06, "loss": 0.2333, "step": 29090 }, { "epoch": 0.5292930027828807, "grad_norm": 1.0519689321517944, "learning_rate": 5.815207318646494e-06, "loss": 0.227, "step": 29100 }, { "epoch": 0.529474890412703, "grad_norm": 1.0197287797927856, "learning_rate": 5.812853480826983e-06, "loss": 0.2349, "step": 29110 }, { "epoch": 0.5296567780425253, "grad_norm": 1.072658658027649, "learning_rate": 5.810499707953803e-06, "loss": 0.2309, "step": 29120 }, { "epoch": 0.5298386656723476, "grad_norm": 1.0353909730911255, "learning_rate": 5.808146000843791e-06, "loss": 0.2343, "step": 29130 }, { "epoch": 0.5300205533021699, "grad_norm": 1.0287649631500244, "learning_rate": 5.805792360313765e-06, "loss": 0.235, "step": 29140 }, { "epoch": 0.5302024409319922, "grad_norm": 1.1710437536239624, "learning_rate": 5.803438787180524e-06, "loss": 0.2351, "step": 29150 }, { "epoch": 0.5303843285618145, "grad_norm": 0.8704602718353271, "learning_rate": 5.801085282260835e-06, "loss": 0.2339, "step": 29160 }, { "epoch": 0.5305662161916368, "grad_norm": 1.110183835029602, "learning_rate": 5.798731846371448e-06, "loss": 0.2337, "step": 29170 }, { "epoch": 0.5307481038214591, "grad_norm": 1.108005404472351, "learning_rate": 5.7963784803290875e-06, "loss": 0.2421, "step": 29180 }, { "epoch": 0.5309299914512814, "grad_norm": 1.0909215211868286, "learning_rate": 5.79402518495045e-06, "loss": 0.2256, "step": 29190 }, { "epoch": 0.5311118790811037, "grad_norm": 1.0148457288742065, "learning_rate": 5.791671961052215e-06, "loss": 0.2344, "step": 29200 }, { "epoch": 0.531293766710926, "grad_norm": 1.0110915899276733, "learning_rate": 5.789318809451028e-06, "loss": 0.2375, "step": 29210 }, { "epoch": 0.5314756543407483, "grad_norm": 0.9321255087852478, "learning_rate": 5.786965730963515e-06, "loss": 0.2302, "step": 29220 }, { "epoch": 0.5316575419705706, "grad_norm": 1.1661500930786133, "learning_rate": 5.7846127264062805e-06, "loss": 0.2357, "step": 29230 }, { "epoch": 0.5318394296003929, "grad_norm": 1.1467283964157104, "learning_rate": 5.782259796595893e-06, "loss": 0.2332, "step": 29240 }, { "epoch": 0.5320213172302152, "grad_norm": 1.2630386352539062, "learning_rate": 5.779906942348905e-06, "loss": 0.2327, "step": 29250 }, { "epoch": 0.5322032048600375, "grad_norm": 0.9080108404159546, "learning_rate": 5.777554164481835e-06, "loss": 0.2373, "step": 29260 }, { "epoch": 0.5323850924898598, "grad_norm": 1.0276989936828613, "learning_rate": 5.7752014638111815e-06, "loss": 0.2277, "step": 29270 }, { "epoch": 0.5325669801196821, "grad_norm": 0.8989725112915039, "learning_rate": 5.772848841153411e-06, "loss": 0.2286, "step": 29280 }, { "epoch": 0.5327488677495044, "grad_norm": 1.0438196659088135, "learning_rate": 5.7704962973249655e-06, "loss": 0.2331, "step": 29290 }, { "epoch": 0.5329307553793267, "grad_norm": 1.071773886680603, "learning_rate": 5.768143833142261e-06, "loss": 0.2341, "step": 29300 }, { "epoch": 0.533112643009149, "grad_norm": 0.9456564784049988, "learning_rate": 5.765791449421681e-06, "loss": 0.241, "step": 29310 }, { "epoch": 0.5332945306389713, "grad_norm": 1.0218158960342407, "learning_rate": 5.763439146979587e-06, "loss": 0.2264, "step": 29320 }, { "epoch": 0.5334764182687936, "grad_norm": 1.0227123498916626, "learning_rate": 5.7610869266323076e-06, "loss": 0.2347, "step": 29330 }, { "epoch": 0.5336583058986158, "grad_norm": 1.0287529230117798, "learning_rate": 5.758734789196145e-06, "loss": 0.236, "step": 29340 }, { "epoch": 0.5338401935284381, "grad_norm": 1.089640736579895, "learning_rate": 5.7563827354873745e-06, "loss": 0.2377, "step": 29350 }, { "epoch": 0.5340220811582604, "grad_norm": 0.9276518225669861, "learning_rate": 5.754030766322237e-06, "loss": 0.2328, "step": 29360 }, { "epoch": 0.5342039687880827, "grad_norm": 1.0281058549880981, "learning_rate": 5.75167888251695e-06, "loss": 0.2365, "step": 29370 }, { "epoch": 0.534385856417905, "grad_norm": 1.0273182392120361, "learning_rate": 5.749327084887698e-06, "loss": 0.2354, "step": 29380 }, { "epoch": 0.5345677440477273, "grad_norm": 1.0579750537872314, "learning_rate": 5.746975374250635e-06, "loss": 0.2351, "step": 29390 }, { "epoch": 0.5347496316775496, "grad_norm": 0.8683820962905884, "learning_rate": 5.744623751421888e-06, "loss": 0.2384, "step": 29400 }, { "epoch": 0.5349315193073719, "grad_norm": 1.0879629850387573, "learning_rate": 5.742272217217551e-06, "loss": 0.2334, "step": 29410 }, { "epoch": 0.5351134069371942, "grad_norm": 1.0609129667282104, "learning_rate": 5.739920772453686e-06, "loss": 0.2332, "step": 29420 }, { "epoch": 0.5352952945670165, "grad_norm": 1.1081922054290771, "learning_rate": 5.737569417946328e-06, "loss": 0.2344, "step": 29430 }, { "epoch": 0.5354771821968388, "grad_norm": 0.9304471611976624, "learning_rate": 5.735218154511479e-06, "loss": 0.2396, "step": 29440 }, { "epoch": 0.5356590698266611, "grad_norm": 1.0985136032104492, "learning_rate": 5.732866982965107e-06, "loss": 0.2291, "step": 29450 }, { "epoch": 0.5358409574564834, "grad_norm": 1.0139542818069458, "learning_rate": 5.7305159041231515e-06, "loss": 0.234, "step": 29460 }, { "epoch": 0.5360228450863057, "grad_norm": 1.1079919338226318, "learning_rate": 5.728164918801517e-06, "loss": 0.2336, "step": 29470 }, { "epoch": 0.536204732716128, "grad_norm": 0.959118664264679, "learning_rate": 5.725814027816078e-06, "loss": 0.2331, "step": 29480 }, { "epoch": 0.5363866203459503, "grad_norm": 0.9697901606559753, "learning_rate": 5.723463231982673e-06, "loss": 0.2287, "step": 29490 }, { "epoch": 0.5365685079757726, "grad_norm": 1.301689863204956, "learning_rate": 5.7211125321171145e-06, "loss": 0.2407, "step": 29500 }, { "epoch": 0.5367503956055949, "grad_norm": 1.021133303642273, "learning_rate": 5.718761929035172e-06, "loss": 0.2348, "step": 29510 }, { "epoch": 0.5369322832354172, "grad_norm": 1.0549908876419067, "learning_rate": 5.716411423552588e-06, "loss": 0.2376, "step": 29520 }, { "epoch": 0.5371141708652395, "grad_norm": 0.9998169541358948, "learning_rate": 5.7140610164850705e-06, "loss": 0.2413, "step": 29530 }, { "epoch": 0.5372960584950618, "grad_norm": 1.0497316122055054, "learning_rate": 5.7117107086482896e-06, "loss": 0.2359, "step": 29540 }, { "epoch": 0.5374779461248841, "grad_norm": 1.0187480449676514, "learning_rate": 5.709360500857885e-06, "loss": 0.2324, "step": 29550 }, { "epoch": 0.5376598337547064, "grad_norm": 0.9702481627464294, "learning_rate": 5.707010393929462e-06, "loss": 0.2326, "step": 29560 }, { "epoch": 0.5378417213845287, "grad_norm": 0.9775631427764893, "learning_rate": 5.7046603886785866e-06, "loss": 0.2314, "step": 29570 }, { "epoch": 0.538023609014351, "grad_norm": 1.028932809829712, "learning_rate": 5.702310485920794e-06, "loss": 0.232, "step": 29580 }, { "epoch": 0.5382054966441733, "grad_norm": 1.1439776420593262, "learning_rate": 5.699960686471578e-06, "loss": 0.2372, "step": 29590 }, { "epoch": 0.5383873842739956, "grad_norm": 1.0690008401870728, "learning_rate": 5.697610991146406e-06, "loss": 0.2332, "step": 29600 }, { "epoch": 0.5385692719038179, "grad_norm": 1.1820428371429443, "learning_rate": 5.6952614007607e-06, "loss": 0.2356, "step": 29610 }, { "epoch": 0.5387511595336402, "grad_norm": 1.146303653717041, "learning_rate": 5.692911916129851e-06, "loss": 0.2313, "step": 29620 }, { "epoch": 0.5389330471634625, "grad_norm": 0.9535050392150879, "learning_rate": 5.69056253806921e-06, "loss": 0.2325, "step": 29630 }, { "epoch": 0.5391149347932848, "grad_norm": 1.0992079973220825, "learning_rate": 5.688213267394094e-06, "loss": 0.2376, "step": 29640 }, { "epoch": 0.5392968224231071, "grad_norm": 1.0448040962219238, "learning_rate": 5.6858641049197815e-06, "loss": 0.2319, "step": 29650 }, { "epoch": 0.5394787100529294, "grad_norm": 1.006338357925415, "learning_rate": 5.683515051461511e-06, "loss": 0.2379, "step": 29660 }, { "epoch": 0.5396605976827515, "grad_norm": 1.001916766166687, "learning_rate": 5.681166107834489e-06, "loss": 0.2305, "step": 29670 }, { "epoch": 0.5398424853125738, "grad_norm": 1.108864665031433, "learning_rate": 5.678817274853877e-06, "loss": 0.2371, "step": 29680 }, { "epoch": 0.5400243729423961, "grad_norm": 0.9151191711425781, "learning_rate": 5.676468553334805e-06, "loss": 0.2301, "step": 29690 }, { "epoch": 0.5402062605722184, "grad_norm": 1.0779138803482056, "learning_rate": 5.674119944092357e-06, "loss": 0.2249, "step": 29700 }, { "epoch": 0.5403881482020407, "grad_norm": 1.038408637046814, "learning_rate": 5.671771447941584e-06, "loss": 0.2325, "step": 29710 }, { "epoch": 0.540570035831863, "grad_norm": 1.1025434732437134, "learning_rate": 5.669423065697496e-06, "loss": 0.2347, "step": 29720 }, { "epoch": 0.5407519234616853, "grad_norm": 1.1184706687927246, "learning_rate": 5.66707479817506e-06, "loss": 0.2331, "step": 29730 }, { "epoch": 0.5409338110915076, "grad_norm": 1.0894944667816162, "learning_rate": 5.6647266461892115e-06, "loss": 0.2282, "step": 29740 }, { "epoch": 0.5411156987213299, "grad_norm": 1.0619761943817139, "learning_rate": 5.662378610554835e-06, "loss": 0.2366, "step": 29750 }, { "epoch": 0.5412975863511522, "grad_norm": 1.0463804006576538, "learning_rate": 5.6600306920867844e-06, "loss": 0.2244, "step": 29760 }, { "epoch": 0.5414794739809745, "grad_norm": 0.9921329021453857, "learning_rate": 5.657682891599865e-06, "loss": 0.2384, "step": 29770 }, { "epoch": 0.5416613616107968, "grad_norm": 1.0708236694335938, "learning_rate": 5.655335209908849e-06, "loss": 0.2328, "step": 29780 }, { "epoch": 0.5418432492406191, "grad_norm": 1.0564535856246948, "learning_rate": 5.652987647828461e-06, "loss": 0.2291, "step": 29790 }, { "epoch": 0.5420251368704414, "grad_norm": 1.0703203678131104, "learning_rate": 5.650640206173387e-06, "loss": 0.2276, "step": 29800 }, { "epoch": 0.5422070245002637, "grad_norm": 1.1085237264633179, "learning_rate": 5.648292885758268e-06, "loss": 0.2391, "step": 29810 }, { "epoch": 0.542388912130086, "grad_norm": 1.0507737398147583, "learning_rate": 5.6459456873977115e-06, "loss": 0.2328, "step": 29820 }, { "epoch": 0.5425707997599083, "grad_norm": 1.1148293018341064, "learning_rate": 5.643598611906272e-06, "loss": 0.2326, "step": 29830 }, { "epoch": 0.5427526873897306, "grad_norm": 0.8965333104133606, "learning_rate": 5.641251660098468e-06, "loss": 0.2323, "step": 29840 }, { "epoch": 0.5429345750195529, "grad_norm": 0.9898197054862976, "learning_rate": 5.6389048327887716e-06, "loss": 0.2282, "step": 29850 }, { "epoch": 0.5431164626493752, "grad_norm": 1.0468670129776, "learning_rate": 5.636558130791613e-06, "loss": 0.2302, "step": 29860 }, { "epoch": 0.5432983502791975, "grad_norm": 1.1611112356185913, "learning_rate": 5.634211554921382e-06, "loss": 0.2358, "step": 29870 }, { "epoch": 0.5434802379090198, "grad_norm": 1.0550047159194946, "learning_rate": 5.631865105992418e-06, "loss": 0.2346, "step": 29880 }, { "epoch": 0.5436621255388421, "grad_norm": 1.1149972677230835, "learning_rate": 5.629518784819024e-06, "loss": 0.2288, "step": 29890 }, { "epoch": 0.5438440131686644, "grad_norm": 1.0250184535980225, "learning_rate": 5.627172592215451e-06, "loss": 0.2312, "step": 29900 }, { "epoch": 0.5440259007984867, "grad_norm": 1.182274341583252, "learning_rate": 5.624826528995911e-06, "loss": 0.2361, "step": 29910 }, { "epoch": 0.544207788428309, "grad_norm": 0.9452041387557983, "learning_rate": 5.62248059597457e-06, "loss": 0.2298, "step": 29920 }, { "epoch": 0.5443896760581313, "grad_norm": 0.9968046545982361, "learning_rate": 5.620134793965546e-06, "loss": 0.2326, "step": 29930 }, { "epoch": 0.5445715636879536, "grad_norm": 1.0025653839111328, "learning_rate": 5.617789123782916e-06, "loss": 0.2287, "step": 29940 }, { "epoch": 0.5447534513177759, "grad_norm": 1.1165555715560913, "learning_rate": 5.615443586240707e-06, "loss": 0.2319, "step": 29950 }, { "epoch": 0.5449353389475982, "grad_norm": 1.0894076824188232, "learning_rate": 5.613098182152904e-06, "loss": 0.2305, "step": 29960 }, { "epoch": 0.5451172265774205, "grad_norm": 1.149275302886963, "learning_rate": 5.610752912333441e-06, "loss": 0.2334, "step": 29970 }, { "epoch": 0.5452991142072428, "grad_norm": 1.1058450937271118, "learning_rate": 5.608407777596208e-06, "loss": 0.2327, "step": 29980 }, { "epoch": 0.545481001837065, "grad_norm": 0.9745444655418396, "learning_rate": 5.60606277875505e-06, "loss": 0.2276, "step": 29990 }, { "epoch": 0.5456628894668873, "grad_norm": 0.9405921697616577, "learning_rate": 5.6037179166237606e-06, "loss": 0.2236, "step": 30000 }, { "epoch": 0.5458447770967096, "grad_norm": 0.9922993779182434, "learning_rate": 5.6013731920160915e-06, "loss": 0.2283, "step": 30010 }, { "epoch": 0.5460266647265319, "grad_norm": 0.9950010180473328, "learning_rate": 5.59902860574574e-06, "loss": 0.2319, "step": 30020 }, { "epoch": 0.5462085523563542, "grad_norm": 0.9998811483383179, "learning_rate": 5.596684158626361e-06, "loss": 0.2341, "step": 30030 }, { "epoch": 0.5463904399861765, "grad_norm": 1.0623749494552612, "learning_rate": 5.5943398514715566e-06, "loss": 0.2327, "step": 30040 }, { "epoch": 0.5465723276159988, "grad_norm": 0.9865375757217407, "learning_rate": 5.591995685094885e-06, "loss": 0.2359, "step": 30050 }, { "epoch": 0.5467542152458211, "grad_norm": 0.9644178152084351, "learning_rate": 5.589651660309853e-06, "loss": 0.2288, "step": 30060 }, { "epoch": 0.5469361028756434, "grad_norm": 1.0183473825454712, "learning_rate": 5.587307777929918e-06, "loss": 0.2388, "step": 30070 }, { "epoch": 0.5471179905054657, "grad_norm": 1.0867995023727417, "learning_rate": 5.584964038768486e-06, "loss": 0.2372, "step": 30080 }, { "epoch": 0.547299878135288, "grad_norm": 0.9653428792953491, "learning_rate": 5.58262044363892e-06, "loss": 0.234, "step": 30090 }, { "epoch": 0.5474817657651103, "grad_norm": 1.0408172607421875, "learning_rate": 5.580276993354525e-06, "loss": 0.2287, "step": 30100 }, { "epoch": 0.5476636533949326, "grad_norm": 0.9434533715248108, "learning_rate": 5.5779336887285625e-06, "loss": 0.2273, "step": 30110 }, { "epoch": 0.5478455410247549, "grad_norm": 1.1924347877502441, "learning_rate": 5.57559053057424e-06, "loss": 0.2336, "step": 30120 }, { "epoch": 0.5480274286545772, "grad_norm": 1.0673818588256836, "learning_rate": 5.573247519704712e-06, "loss": 0.2299, "step": 30130 }, { "epoch": 0.5482093162843995, "grad_norm": 1.07002592086792, "learning_rate": 5.5709046569330875e-06, "loss": 0.231, "step": 30140 }, { "epoch": 0.5483912039142218, "grad_norm": 1.0068707466125488, "learning_rate": 5.568561943072417e-06, "loss": 0.2334, "step": 30150 }, { "epoch": 0.5485730915440441, "grad_norm": 1.0399404764175415, "learning_rate": 5.566219378935709e-06, "loss": 0.2334, "step": 30160 }, { "epoch": 0.5487549791738664, "grad_norm": 1.1136581897735596, "learning_rate": 5.56387696533591e-06, "loss": 0.2333, "step": 30170 }, { "epoch": 0.5489368668036887, "grad_norm": 1.1521400213241577, "learning_rate": 5.56153470308592e-06, "loss": 0.2342, "step": 30180 }, { "epoch": 0.549118754433511, "grad_norm": 1.0327507257461548, "learning_rate": 5.559192592998587e-06, "loss": 0.236, "step": 30190 }, { "epoch": 0.5493006420633333, "grad_norm": 0.954231321811676, "learning_rate": 5.556850635886701e-06, "loss": 0.2289, "step": 30200 }, { "epoch": 0.5494825296931556, "grad_norm": 1.3626220226287842, "learning_rate": 5.554508832563006e-06, "loss": 0.2295, "step": 30210 }, { "epoch": 0.5496644173229779, "grad_norm": 1.0768349170684814, "learning_rate": 5.5521671838401856e-06, "loss": 0.2394, "step": 30220 }, { "epoch": 0.5498463049528002, "grad_norm": 1.0492844581604004, "learning_rate": 5.549825690530876e-06, "loss": 0.2331, "step": 30230 }, { "epoch": 0.5500281925826225, "grad_norm": 1.1057953834533691, "learning_rate": 5.5474843534476545e-06, "loss": 0.2336, "step": 30240 }, { "epoch": 0.5502100802124448, "grad_norm": 0.9441450834274292, "learning_rate": 5.545143173403046e-06, "loss": 0.2327, "step": 30250 }, { "epoch": 0.5503919678422671, "grad_norm": 1.0348663330078125, "learning_rate": 5.542802151209523e-06, "loss": 0.2293, "step": 30260 }, { "epoch": 0.5505738554720894, "grad_norm": 1.0580936670303345, "learning_rate": 5.5404612876795e-06, "loss": 0.2337, "step": 30270 }, { "epoch": 0.5507557431019117, "grad_norm": 1.1611499786376953, "learning_rate": 5.5381205836253375e-06, "loss": 0.2334, "step": 30280 }, { "epoch": 0.550937630731734, "grad_norm": 1.0718892812728882, "learning_rate": 5.535780039859341e-06, "loss": 0.2299, "step": 30290 }, { "epoch": 0.5511195183615563, "grad_norm": 1.014711618423462, "learning_rate": 5.533439657193762e-06, "loss": 0.2288, "step": 30300 }, { "epoch": 0.5513014059913786, "grad_norm": 0.9640907049179077, "learning_rate": 5.531099436440793e-06, "loss": 0.2293, "step": 30310 }, { "epoch": 0.5514832936212009, "grad_norm": 1.05460786819458, "learning_rate": 5.52875937841257e-06, "loss": 0.2386, "step": 30320 }, { "epoch": 0.5516651812510232, "grad_norm": 1.0359519720077515, "learning_rate": 5.526419483921177e-06, "loss": 0.2357, "step": 30330 }, { "epoch": 0.5518470688808454, "grad_norm": 0.9911280274391174, "learning_rate": 5.524079753778639e-06, "loss": 0.226, "step": 30340 }, { "epoch": 0.5520289565106677, "grad_norm": 0.947166919708252, "learning_rate": 5.521740188796921e-06, "loss": 0.2341, "step": 30350 }, { "epoch": 0.55221084414049, "grad_norm": 1.1330032348632812, "learning_rate": 5.519400789787936e-06, "loss": 0.2295, "step": 30360 }, { "epoch": 0.5523927317703123, "grad_norm": 1.0150935649871826, "learning_rate": 5.517061557563532e-06, "loss": 0.2352, "step": 30370 }, { "epoch": 0.5525746194001346, "grad_norm": 1.1031506061553955, "learning_rate": 5.514722492935508e-06, "loss": 0.2361, "step": 30380 }, { "epoch": 0.5527565070299569, "grad_norm": 1.0270905494689941, "learning_rate": 5.512383596715599e-06, "loss": 0.2335, "step": 30390 }, { "epoch": 0.5529383946597792, "grad_norm": 1.0546590089797974, "learning_rate": 5.510044869715483e-06, "loss": 0.2352, "step": 30400 }, { "epoch": 0.5531202822896015, "grad_norm": 1.0254411697387695, "learning_rate": 5.507706312746779e-06, "loss": 0.2376, "step": 30410 }, { "epoch": 0.5533021699194238, "grad_norm": 1.0189224481582642, "learning_rate": 5.505367926621047e-06, "loss": 0.2385, "step": 30420 }, { "epoch": 0.5534840575492461, "grad_norm": 1.0725640058517456, "learning_rate": 5.50302971214979e-06, "loss": 0.2301, "step": 30430 }, { "epoch": 0.5536659451790684, "grad_norm": 1.2169197797775269, "learning_rate": 5.500691670144448e-06, "loss": 0.2402, "step": 30440 }, { "epoch": 0.5538478328088907, "grad_norm": 0.9245725870132446, "learning_rate": 5.498353801416402e-06, "loss": 0.2321, "step": 30450 }, { "epoch": 0.554029720438713, "grad_norm": 1.0870983600616455, "learning_rate": 5.496016106776975e-06, "loss": 0.2331, "step": 30460 }, { "epoch": 0.5542116080685353, "grad_norm": 0.9866313934326172, "learning_rate": 5.493678587037425e-06, "loss": 0.2354, "step": 30470 }, { "epoch": 0.5543934956983576, "grad_norm": 0.9904917478561401, "learning_rate": 5.4913412430089574e-06, "loss": 0.2408, "step": 30480 }, { "epoch": 0.5545753833281799, "grad_norm": 1.0206598043441772, "learning_rate": 5.489004075502709e-06, "loss": 0.2322, "step": 30490 }, { "epoch": 0.5547572709580021, "grad_norm": 1.0043582916259766, "learning_rate": 5.486667085329755e-06, "loss": 0.2325, "step": 30500 }, { "epoch": 0.5549391585878244, "grad_norm": 1.1796817779541016, "learning_rate": 5.484330273301119e-06, "loss": 0.2341, "step": 30510 }, { "epoch": 0.5551210462176467, "grad_norm": 1.1425234079360962, "learning_rate": 5.481993640227748e-06, "loss": 0.2346, "step": 30520 }, { "epoch": 0.555302933847469, "grad_norm": 0.9768431782722473, "learning_rate": 5.47965718692054e-06, "loss": 0.2326, "step": 30530 }, { "epoch": 0.5554848214772913, "grad_norm": 1.0340431928634644, "learning_rate": 5.477320914190322e-06, "loss": 0.2292, "step": 30540 }, { "epoch": 0.5556667091071136, "grad_norm": 1.0538471937179565, "learning_rate": 5.474984822847864e-06, "loss": 0.238, "step": 30550 }, { "epoch": 0.5558485967369359, "grad_norm": 0.9400471448898315, "learning_rate": 5.472648913703869e-06, "loss": 0.2307, "step": 30560 }, { "epoch": 0.5560304843667582, "grad_norm": 1.0696097612380981, "learning_rate": 5.4703131875689804e-06, "loss": 0.2316, "step": 30570 }, { "epoch": 0.5562123719965805, "grad_norm": 0.9533029794692993, "learning_rate": 5.4679776452537746e-06, "loss": 0.235, "step": 30580 }, { "epoch": 0.5563942596264028, "grad_norm": 0.9796180129051208, "learning_rate": 5.465642287568764e-06, "loss": 0.2318, "step": 30590 }, { "epoch": 0.5565761472562251, "grad_norm": 1.0407708883285522, "learning_rate": 5.463307115324402e-06, "loss": 0.2368, "step": 30600 }, { "epoch": 0.5567580348860474, "grad_norm": 1.183281660079956, "learning_rate": 5.460972129331075e-06, "loss": 0.2378, "step": 30610 }, { "epoch": 0.5569399225158697, "grad_norm": 0.9969792366027832, "learning_rate": 5.4586373303991e-06, "loss": 0.2262, "step": 30620 }, { "epoch": 0.557121810145692, "grad_norm": 0.9691275954246521, "learning_rate": 5.456302719338736e-06, "loss": 0.2296, "step": 30630 }, { "epoch": 0.5573036977755143, "grad_norm": 1.0053476095199585, "learning_rate": 5.453968296960174e-06, "loss": 0.231, "step": 30640 }, { "epoch": 0.5574855854053365, "grad_norm": 1.115382194519043, "learning_rate": 5.451634064073538e-06, "loss": 0.2291, "step": 30650 }, { "epoch": 0.5576674730351588, "grad_norm": 1.0473781824111938, "learning_rate": 5.449300021488888e-06, "loss": 0.2317, "step": 30660 }, { "epoch": 0.5578493606649811, "grad_norm": 1.1361571550369263, "learning_rate": 5.446966170016218e-06, "loss": 0.2349, "step": 30670 }, { "epoch": 0.5580312482948034, "grad_norm": 1.0086969137191772, "learning_rate": 5.444632510465456e-06, "loss": 0.2327, "step": 30680 }, { "epoch": 0.5582131359246257, "grad_norm": 0.9801390171051025, "learning_rate": 5.442299043646462e-06, "loss": 0.2281, "step": 30690 }, { "epoch": 0.558395023554448, "grad_norm": 0.912772536277771, "learning_rate": 5.439965770369031e-06, "loss": 0.2342, "step": 30700 }, { "epoch": 0.5585769111842703, "grad_norm": 1.038275122642517, "learning_rate": 5.437632691442889e-06, "loss": 0.2284, "step": 30710 }, { "epoch": 0.5587587988140926, "grad_norm": 1.009528398513794, "learning_rate": 5.435299807677693e-06, "loss": 0.2272, "step": 30720 }, { "epoch": 0.5589406864439149, "grad_norm": 1.0899229049682617, "learning_rate": 5.432967119883038e-06, "loss": 0.2315, "step": 30730 }, { "epoch": 0.5591225740737372, "grad_norm": 1.0794355869293213, "learning_rate": 5.430634628868445e-06, "loss": 0.2316, "step": 30740 }, { "epoch": 0.5593044617035595, "grad_norm": 1.0033371448516846, "learning_rate": 5.428302335443372e-06, "loss": 0.2348, "step": 30750 }, { "epoch": 0.5594863493333818, "grad_norm": 1.035003423690796, "learning_rate": 5.425970240417202e-06, "loss": 0.2293, "step": 30760 }, { "epoch": 0.5596682369632041, "grad_norm": 1.102955937385559, "learning_rate": 5.423638344599256e-06, "loss": 0.2309, "step": 30770 }, { "epoch": 0.5598501245930264, "grad_norm": 0.9668028950691223, "learning_rate": 5.4213066487987805e-06, "loss": 0.2353, "step": 30780 }, { "epoch": 0.5600320122228487, "grad_norm": 1.1864781379699707, "learning_rate": 5.418975153824957e-06, "loss": 0.2374, "step": 30790 }, { "epoch": 0.560213899852671, "grad_norm": 0.9532299041748047, "learning_rate": 5.416643860486892e-06, "loss": 0.2297, "step": 30800 }, { "epoch": 0.5603957874824933, "grad_norm": 0.8968337774276733, "learning_rate": 5.414312769593628e-06, "loss": 0.2286, "step": 30810 }, { "epoch": 0.5605776751123156, "grad_norm": 1.0846155881881714, "learning_rate": 5.411981881954133e-06, "loss": 0.2314, "step": 30820 }, { "epoch": 0.5607595627421379, "grad_norm": 0.991410493850708, "learning_rate": 5.409651198377307e-06, "loss": 0.2358, "step": 30830 }, { "epoch": 0.5609414503719602, "grad_norm": 1.0356638431549072, "learning_rate": 5.407320719671976e-06, "loss": 0.2383, "step": 30840 }, { "epoch": 0.5611233380017825, "grad_norm": 0.9444685578346252, "learning_rate": 5.404990446646897e-06, "loss": 0.2246, "step": 30850 }, { "epoch": 0.5613052256316048, "grad_norm": 1.048344612121582, "learning_rate": 5.402660380110757e-06, "loss": 0.2306, "step": 30860 }, { "epoch": 0.5614871132614271, "grad_norm": 1.2397648096084595, "learning_rate": 5.400330520872168e-06, "loss": 0.2411, "step": 30870 }, { "epoch": 0.5616690008912494, "grad_norm": 0.9519614577293396, "learning_rate": 5.398000869739674e-06, "loss": 0.2313, "step": 30880 }, { "epoch": 0.5618508885210717, "grad_norm": 1.089355230331421, "learning_rate": 5.395671427521745e-06, "loss": 0.2349, "step": 30890 }, { "epoch": 0.562032776150894, "grad_norm": 1.126096487045288, "learning_rate": 5.393342195026776e-06, "loss": 0.2289, "step": 30900 }, { "epoch": 0.5622146637807163, "grad_norm": 1.0335042476654053, "learning_rate": 5.391013173063093e-06, "loss": 0.2328, "step": 30910 }, { "epoch": 0.5623965514105386, "grad_norm": 0.9649476408958435, "learning_rate": 5.3886843624389464e-06, "loss": 0.23, "step": 30920 }, { "epoch": 0.5625784390403609, "grad_norm": 1.0825494527816772, "learning_rate": 5.386355763962515e-06, "loss": 0.2369, "step": 30930 }, { "epoch": 0.5627603266701832, "grad_norm": 0.9978388547897339, "learning_rate": 5.384027378441902e-06, "loss": 0.2314, "step": 30940 }, { "epoch": 0.5629422143000055, "grad_norm": 0.9724529981613159, "learning_rate": 5.381699206685139e-06, "loss": 0.23, "step": 30950 }, { "epoch": 0.5631241019298278, "grad_norm": 1.0622512102127075, "learning_rate": 5.379371249500181e-06, "loss": 0.2293, "step": 30960 }, { "epoch": 0.56330598955965, "grad_norm": 0.9983969330787659, "learning_rate": 5.3770435076949125e-06, "loss": 0.2247, "step": 30970 }, { "epoch": 0.5634878771894724, "grad_norm": 1.1251076459884644, "learning_rate": 5.374715982077139e-06, "loss": 0.2351, "step": 30980 }, { "epoch": 0.5636697648192946, "grad_norm": 0.9814444184303284, "learning_rate": 5.37238867345459e-06, "loss": 0.2306, "step": 30990 }, { "epoch": 0.563851652449117, "grad_norm": 1.0478147268295288, "learning_rate": 5.370061582634927e-06, "loss": 0.2315, "step": 31000 }, { "epoch": 0.5640335400789392, "grad_norm": 1.1080548763275146, "learning_rate": 5.3677347104257274e-06, "loss": 0.2359, "step": 31010 }, { "epoch": 0.5642154277087615, "grad_norm": 1.0233296155929565, "learning_rate": 5.365408057634499e-06, "loss": 0.2356, "step": 31020 }, { "epoch": 0.5643973153385838, "grad_norm": 1.1115097999572754, "learning_rate": 5.363081625068669e-06, "loss": 0.233, "step": 31030 }, { "epoch": 0.5645792029684061, "grad_norm": 1.1684825420379639, "learning_rate": 5.360755413535591e-06, "loss": 0.234, "step": 31040 }, { "epoch": 0.5647610905982284, "grad_norm": 1.040555715560913, "learning_rate": 5.358429423842542e-06, "loss": 0.2337, "step": 31050 }, { "epoch": 0.5649429782280507, "grad_norm": 0.9511948823928833, "learning_rate": 5.356103656796719e-06, "loss": 0.2323, "step": 31060 }, { "epoch": 0.565124865857873, "grad_norm": 0.9752454161643982, "learning_rate": 5.353778113205245e-06, "loss": 0.2281, "step": 31070 }, { "epoch": 0.5653067534876953, "grad_norm": 1.0786898136138916, "learning_rate": 5.351452793875165e-06, "loss": 0.2384, "step": 31080 }, { "epoch": 0.5654886411175176, "grad_norm": 1.0233315229415894, "learning_rate": 5.349127699613444e-06, "loss": 0.2321, "step": 31090 }, { "epoch": 0.5656705287473399, "grad_norm": 1.0551549196243286, "learning_rate": 5.346802831226972e-06, "loss": 0.2358, "step": 31100 }, { "epoch": 0.5658524163771622, "grad_norm": 0.9914330244064331, "learning_rate": 5.3444781895225564e-06, "loss": 0.2243, "step": 31110 }, { "epoch": 0.5660343040069845, "grad_norm": 1.008002758026123, "learning_rate": 5.34215377530693e-06, "loss": 0.2371, "step": 31120 }, { "epoch": 0.5662161916368068, "grad_norm": 1.1225355863571167, "learning_rate": 5.339829589386745e-06, "loss": 0.235, "step": 31130 }, { "epoch": 0.5663980792666291, "grad_norm": 1.0346909761428833, "learning_rate": 5.337505632568574e-06, "loss": 0.2323, "step": 31140 }, { "epoch": 0.5665799668964514, "grad_norm": 1.1239229440689087, "learning_rate": 5.3351819056589096e-06, "loss": 0.2348, "step": 31150 }, { "epoch": 0.5667618545262737, "grad_norm": 1.0792022943496704, "learning_rate": 5.332858409464168e-06, "loss": 0.2247, "step": 31160 }, { "epoch": 0.566943742156096, "grad_norm": 1.1348061561584473, "learning_rate": 5.330535144790682e-06, "loss": 0.2318, "step": 31170 }, { "epoch": 0.5671256297859183, "grad_norm": 1.1325775384902954, "learning_rate": 5.328212112444705e-06, "loss": 0.2327, "step": 31180 }, { "epoch": 0.5673075174157406, "grad_norm": 1.0343570709228516, "learning_rate": 5.325889313232409e-06, "loss": 0.2306, "step": 31190 }, { "epoch": 0.5674894050455629, "grad_norm": 1.0810939073562622, "learning_rate": 5.323566747959887e-06, "loss": 0.2377, "step": 31200 }, { "epoch": 0.5676712926753852, "grad_norm": 1.023622751235962, "learning_rate": 5.321244417433149e-06, "loss": 0.2272, "step": 31210 }, { "epoch": 0.5678531803052075, "grad_norm": 1.0553888082504272, "learning_rate": 5.318922322458124e-06, "loss": 0.2372, "step": 31220 }, { "epoch": 0.5680350679350298, "grad_norm": 1.0640326738357544, "learning_rate": 5.31660046384066e-06, "loss": 0.2337, "step": 31230 }, { "epoch": 0.5682169555648521, "grad_norm": 1.105950117111206, "learning_rate": 5.314278842386521e-06, "loss": 0.2339, "step": 31240 }, { "epoch": 0.5683988431946744, "grad_norm": 0.9975911378860474, "learning_rate": 5.311957458901392e-06, "loss": 0.2307, "step": 31250 }, { "epoch": 0.5685807308244967, "grad_norm": 1.0773829221725464, "learning_rate": 5.3096363141908714e-06, "loss": 0.2317, "step": 31260 }, { "epoch": 0.568762618454319, "grad_norm": 1.081488013267517, "learning_rate": 5.307315409060481e-06, "loss": 0.2377, "step": 31270 }, { "epoch": 0.5689445060841413, "grad_norm": 0.8961349725723267, "learning_rate": 5.304994744315649e-06, "loss": 0.2355, "step": 31280 }, { "epoch": 0.5691263937139636, "grad_norm": 1.026458978652954, "learning_rate": 5.302674320761732e-06, "loss": 0.2335, "step": 31290 }, { "epoch": 0.5693082813437859, "grad_norm": 1.11603844165802, "learning_rate": 5.300354139203996e-06, "loss": 0.2288, "step": 31300 }, { "epoch": 0.5694901689736082, "grad_norm": 0.9894170165061951, "learning_rate": 5.298034200447622e-06, "loss": 0.2322, "step": 31310 }, { "epoch": 0.5696720566034305, "grad_norm": 1.1383452415466309, "learning_rate": 5.295714505297712e-06, "loss": 0.2324, "step": 31320 }, { "epoch": 0.5698539442332528, "grad_norm": 1.0508434772491455, "learning_rate": 5.293395054559278e-06, "loss": 0.2321, "step": 31330 }, { "epoch": 0.5700358318630749, "grad_norm": 1.0392154455184937, "learning_rate": 5.291075849037254e-06, "loss": 0.2314, "step": 31340 }, { "epoch": 0.5702177194928972, "grad_norm": 1.0208503007888794, "learning_rate": 5.288756889536479e-06, "loss": 0.2337, "step": 31350 }, { "epoch": 0.5703996071227195, "grad_norm": 1.040327787399292, "learning_rate": 5.286438176861719e-06, "loss": 0.2328, "step": 31360 }, { "epoch": 0.5705814947525418, "grad_norm": 1.0414096117019653, "learning_rate": 5.2841197118176426e-06, "loss": 0.2255, "step": 31370 }, { "epoch": 0.5707633823823641, "grad_norm": 1.06798255443573, "learning_rate": 5.2818014952088385e-06, "loss": 0.2352, "step": 31380 }, { "epoch": 0.5709452700121864, "grad_norm": 1.00497305393219, "learning_rate": 5.279483527839808e-06, "loss": 0.2329, "step": 31390 }, { "epoch": 0.5711271576420087, "grad_norm": 1.0053774118423462, "learning_rate": 5.277165810514968e-06, "loss": 0.2292, "step": 31400 }, { "epoch": 0.571309045271831, "grad_norm": 0.9356263875961304, "learning_rate": 5.274848344038646e-06, "loss": 0.225, "step": 31410 }, { "epoch": 0.5714909329016533, "grad_norm": 1.0934680700302124, "learning_rate": 5.272531129215079e-06, "loss": 0.2305, "step": 31420 }, { "epoch": 0.5716728205314756, "grad_norm": 0.973341166973114, "learning_rate": 5.270214166848429e-06, "loss": 0.2257, "step": 31430 }, { "epoch": 0.5718547081612979, "grad_norm": 1.0307974815368652, "learning_rate": 5.267897457742755e-06, "loss": 0.2309, "step": 31440 }, { "epoch": 0.5720365957911202, "grad_norm": 0.9903424978256226, "learning_rate": 5.265581002702041e-06, "loss": 0.2359, "step": 31450 }, { "epoch": 0.5722184834209425, "grad_norm": 0.9989016652107239, "learning_rate": 5.263264802530172e-06, "loss": 0.2346, "step": 31460 }, { "epoch": 0.5724003710507648, "grad_norm": 0.9817729592323303, "learning_rate": 5.260948858030954e-06, "loss": 0.2355, "step": 31470 }, { "epoch": 0.5725822586805871, "grad_norm": 0.9527500867843628, "learning_rate": 5.2586331700080966e-06, "loss": 0.2289, "step": 31480 }, { "epoch": 0.5727641463104094, "grad_norm": 0.9186456203460693, "learning_rate": 5.256317739265226e-06, "loss": 0.2354, "step": 31490 }, { "epoch": 0.5729460339402317, "grad_norm": 0.9378907084465027, "learning_rate": 5.254002566605878e-06, "loss": 0.2357, "step": 31500 }, { "epoch": 0.573127921570054, "grad_norm": 0.9141557812690735, "learning_rate": 5.251687652833494e-06, "loss": 0.229, "step": 31510 }, { "epoch": 0.5733098091998763, "grad_norm": 0.9638628959655762, "learning_rate": 5.249372998751432e-06, "loss": 0.23, "step": 31520 }, { "epoch": 0.5734916968296986, "grad_norm": 1.1111685037612915, "learning_rate": 5.247058605162955e-06, "loss": 0.2345, "step": 31530 }, { "epoch": 0.5736735844595209, "grad_norm": 1.0177158117294312, "learning_rate": 5.2447444728712415e-06, "loss": 0.2382, "step": 31540 }, { "epoch": 0.5738554720893432, "grad_norm": 1.0268818140029907, "learning_rate": 5.242430602679374e-06, "loss": 0.2301, "step": 31550 }, { "epoch": 0.5740373597191655, "grad_norm": 1.1126199960708618, "learning_rate": 5.240116995390343e-06, "loss": 0.2355, "step": 31560 }, { "epoch": 0.5742192473489878, "grad_norm": 1.0426604747772217, "learning_rate": 5.237803651807054e-06, "loss": 0.2361, "step": 31570 }, { "epoch": 0.5744011349788101, "grad_norm": 1.053911805152893, "learning_rate": 5.235490572732316e-06, "loss": 0.2386, "step": 31580 }, { "epoch": 0.5745830226086324, "grad_norm": 1.0753487348556519, "learning_rate": 5.2331777589688474e-06, "loss": 0.2334, "step": 31590 }, { "epoch": 0.5747649102384547, "grad_norm": 0.9375584721565247, "learning_rate": 5.230865211319274e-06, "loss": 0.2351, "step": 31600 }, { "epoch": 0.574946797868277, "grad_norm": 1.0107825994491577, "learning_rate": 5.22855293058613e-06, "loss": 0.2309, "step": 31610 }, { "epoch": 0.5751286854980993, "grad_norm": 0.9805966019630432, "learning_rate": 5.226240917571861e-06, "loss": 0.2338, "step": 31620 }, { "epoch": 0.5753105731279216, "grad_norm": 1.0483276844024658, "learning_rate": 5.2239291730788096e-06, "loss": 0.2354, "step": 31630 }, { "epoch": 0.5754924607577439, "grad_norm": 1.0051281452178955, "learning_rate": 5.221617697909236e-06, "loss": 0.2333, "step": 31640 }, { "epoch": 0.5756743483875661, "grad_norm": 0.9939745664596558, "learning_rate": 5.2193064928653e-06, "loss": 0.2304, "step": 31650 }, { "epoch": 0.5758562360173884, "grad_norm": 0.9432213306427002, "learning_rate": 5.21699555874907e-06, "loss": 0.2287, "step": 31660 }, { "epoch": 0.5760381236472107, "grad_norm": 1.0570756196975708, "learning_rate": 5.21468489636252e-06, "loss": 0.2362, "step": 31670 }, { "epoch": 0.576220011277033, "grad_norm": 0.9760013818740845, "learning_rate": 5.21237450650753e-06, "loss": 0.2307, "step": 31680 }, { "epoch": 0.5764018989068553, "grad_norm": 0.9999644756317139, "learning_rate": 5.210064389985887e-06, "loss": 0.2315, "step": 31690 }, { "epoch": 0.5765837865366776, "grad_norm": 1.062132716178894, "learning_rate": 5.20775454759928e-06, "loss": 0.2294, "step": 31700 }, { "epoch": 0.5767656741664999, "grad_norm": 1.0032590627670288, "learning_rate": 5.205444980149303e-06, "loss": 0.2302, "step": 31710 }, { "epoch": 0.5769475617963222, "grad_norm": 0.9387415647506714, "learning_rate": 5.2031356884374596e-06, "loss": 0.2293, "step": 31720 }, { "epoch": 0.5771294494261445, "grad_norm": 1.034792184829712, "learning_rate": 5.200826673265152e-06, "loss": 0.2314, "step": 31730 }, { "epoch": 0.5773113370559668, "grad_norm": 0.973124086856842, "learning_rate": 5.198517935433687e-06, "loss": 0.2374, "step": 31740 }, { "epoch": 0.5774932246857891, "grad_norm": 1.0560771226882935, "learning_rate": 5.1962094757442795e-06, "loss": 0.2292, "step": 31750 }, { "epoch": 0.5776751123156114, "grad_norm": 0.9921122193336487, "learning_rate": 5.193901294998044e-06, "loss": 0.2293, "step": 31760 }, { "epoch": 0.5778569999454337, "grad_norm": 1.1704269647598267, "learning_rate": 5.1915933939959986e-06, "loss": 0.2314, "step": 31770 }, { "epoch": 0.578038887575256, "grad_norm": 1.0494579076766968, "learning_rate": 5.189285773539065e-06, "loss": 0.2355, "step": 31780 }, { "epoch": 0.5782207752050783, "grad_norm": 1.0257377624511719, "learning_rate": 5.1869784344280675e-06, "loss": 0.2294, "step": 31790 }, { "epoch": 0.5784026628349006, "grad_norm": 1.1080572605133057, "learning_rate": 5.184671377463732e-06, "loss": 0.235, "step": 31800 }, { "epoch": 0.5785845504647229, "grad_norm": 1.0626914501190186, "learning_rate": 5.1823646034466915e-06, "loss": 0.2255, "step": 31810 }, { "epoch": 0.5787664380945452, "grad_norm": 1.1224690675735474, "learning_rate": 5.180058113177471e-06, "loss": 0.2297, "step": 31820 }, { "epoch": 0.5789483257243675, "grad_norm": 1.153935432434082, "learning_rate": 5.177751907456505e-06, "loss": 0.2361, "step": 31830 }, { "epoch": 0.5791302133541898, "grad_norm": 1.0163662433624268, "learning_rate": 5.175445987084128e-06, "loss": 0.2317, "step": 31840 }, { "epoch": 0.5793121009840121, "grad_norm": 1.0656633377075195, "learning_rate": 5.1731403528605706e-06, "loss": 0.2381, "step": 31850 }, { "epoch": 0.5794939886138344, "grad_norm": 1.0081274509429932, "learning_rate": 5.170835005585973e-06, "loss": 0.2294, "step": 31860 }, { "epoch": 0.5796758762436567, "grad_norm": 1.1081962585449219, "learning_rate": 5.168529946060365e-06, "loss": 0.2321, "step": 31870 }, { "epoch": 0.579857763873479, "grad_norm": 1.0558089017868042, "learning_rate": 5.166225175083687e-06, "loss": 0.2388, "step": 31880 }, { "epoch": 0.5800396515033013, "grad_norm": 1.0680440664291382, "learning_rate": 5.163920693455772e-06, "loss": 0.2301, "step": 31890 }, { "epoch": 0.5802215391331236, "grad_norm": 1.1843724250793457, "learning_rate": 5.161616501976354e-06, "loss": 0.2351, "step": 31900 }, { "epoch": 0.5804034267629459, "grad_norm": 1.1137070655822754, "learning_rate": 5.1593126014450686e-06, "loss": 0.2314, "step": 31910 }, { "epoch": 0.5805853143927682, "grad_norm": 0.9422705173492432, "learning_rate": 5.15700899266145e-06, "loss": 0.2312, "step": 31920 }, { "epoch": 0.5807672020225905, "grad_norm": 0.9360978603363037, "learning_rate": 5.15470567642493e-06, "loss": 0.2284, "step": 31930 }, { "epoch": 0.5809490896524128, "grad_norm": 1.06290864944458, "learning_rate": 5.152402653534838e-06, "loss": 0.2325, "step": 31940 }, { "epoch": 0.5811309772822351, "grad_norm": 1.1024699211120605, "learning_rate": 5.150099924790404e-06, "loss": 0.2371, "step": 31950 }, { "epoch": 0.5813128649120574, "grad_norm": 1.0623990297317505, "learning_rate": 5.147797490990753e-06, "loss": 0.2301, "step": 31960 }, { "epoch": 0.5814947525418797, "grad_norm": 0.9685909748077393, "learning_rate": 5.14549535293491e-06, "loss": 0.2349, "step": 31970 }, { "epoch": 0.581676640171702, "grad_norm": 1.0955525636672974, "learning_rate": 5.143193511421798e-06, "loss": 0.2322, "step": 31980 }, { "epoch": 0.5818585278015243, "grad_norm": 1.0077673196792603, "learning_rate": 5.140891967250236e-06, "loss": 0.2318, "step": 31990 }, { "epoch": 0.5820404154313465, "grad_norm": 1.0467798709869385, "learning_rate": 5.138590721218937e-06, "loss": 0.2249, "step": 32000 }, { "epoch": 0.5822223030611688, "grad_norm": 1.0142946243286133, "learning_rate": 5.136289774126515e-06, "loss": 0.2359, "step": 32010 }, { "epoch": 0.5824041906909911, "grad_norm": 1.0409880876541138, "learning_rate": 5.133989126771477e-06, "loss": 0.2356, "step": 32020 }, { "epoch": 0.5825860783208134, "grad_norm": 0.8854726552963257, "learning_rate": 5.131688779952231e-06, "loss": 0.2293, "step": 32030 }, { "epoch": 0.5827679659506357, "grad_norm": 0.9655479788780212, "learning_rate": 5.129388734467073e-06, "loss": 0.2353, "step": 32040 }, { "epoch": 0.582949853580458, "grad_norm": 1.0407323837280273, "learning_rate": 5.1270889911141995e-06, "loss": 0.2407, "step": 32050 }, { "epoch": 0.5831317412102803, "grad_norm": 1.0326703786849976, "learning_rate": 5.124789550691702e-06, "loss": 0.2254, "step": 32060 }, { "epoch": 0.5833136288401026, "grad_norm": 1.197322130203247, "learning_rate": 5.122490413997566e-06, "loss": 0.2341, "step": 32070 }, { "epoch": 0.5834955164699249, "grad_norm": 1.0346308946609497, "learning_rate": 5.1201915818296726e-06, "loss": 0.2265, "step": 32080 }, { "epoch": 0.5836774040997472, "grad_norm": 1.1197483539581299, "learning_rate": 5.117893054985793e-06, "loss": 0.233, "step": 32090 }, { "epoch": 0.5838592917295695, "grad_norm": 0.9150761365890503, "learning_rate": 5.115594834263598e-06, "loss": 0.2298, "step": 32100 }, { "epoch": 0.5840411793593918, "grad_norm": 1.0348066091537476, "learning_rate": 5.11329692046065e-06, "loss": 0.2348, "step": 32110 }, { "epoch": 0.5842230669892141, "grad_norm": 0.927821159362793, "learning_rate": 5.110999314374403e-06, "loss": 0.226, "step": 32120 }, { "epoch": 0.5844049546190364, "grad_norm": 1.0835669040679932, "learning_rate": 5.108702016802208e-06, "loss": 0.2313, "step": 32130 }, { "epoch": 0.5845868422488587, "grad_norm": 1.13704514503479, "learning_rate": 5.106405028541306e-06, "loss": 0.2355, "step": 32140 }, { "epoch": 0.584768729878681, "grad_norm": 1.1420735120773315, "learning_rate": 5.104108350388832e-06, "loss": 0.2315, "step": 32150 }, { "epoch": 0.5849506175085033, "grad_norm": 1.0489819049835205, "learning_rate": 5.101811983141812e-06, "loss": 0.2332, "step": 32160 }, { "epoch": 0.5851325051383255, "grad_norm": 1.0355250835418701, "learning_rate": 5.099515927597165e-06, "loss": 0.2324, "step": 32170 }, { "epoch": 0.5853143927681478, "grad_norm": 1.0077470541000366, "learning_rate": 5.0972201845517035e-06, "loss": 0.2327, "step": 32180 }, { "epoch": 0.5854962803979701, "grad_norm": 0.94249427318573, "learning_rate": 5.094924754802127e-06, "loss": 0.228, "step": 32190 }, { "epoch": 0.5856781680277924, "grad_norm": 0.9497967958450317, "learning_rate": 5.092629639145034e-06, "loss": 0.2322, "step": 32200 }, { "epoch": 0.5858600556576147, "grad_norm": 1.0557676553726196, "learning_rate": 5.090334838376904e-06, "loss": 0.2321, "step": 32210 }, { "epoch": 0.586041943287437, "grad_norm": 0.9220730066299438, "learning_rate": 5.088040353294115e-06, "loss": 0.2223, "step": 32220 }, { "epoch": 0.5862238309172593, "grad_norm": 1.0412043333053589, "learning_rate": 5.085746184692932e-06, "loss": 0.2342, "step": 32230 }, { "epoch": 0.5864057185470816, "grad_norm": 1.0228556394577026, "learning_rate": 5.083452333369512e-06, "loss": 0.2305, "step": 32240 }, { "epoch": 0.5865876061769039, "grad_norm": 1.0324307680130005, "learning_rate": 5.0811588001198986e-06, "loss": 0.2277, "step": 32250 }, { "epoch": 0.5867694938067262, "grad_norm": 1.1044939756393433, "learning_rate": 5.0788655857400304e-06, "loss": 0.2363, "step": 32260 }, { "epoch": 0.5869513814365485, "grad_norm": 1.0391732454299927, "learning_rate": 5.07657269102573e-06, "loss": 0.2305, "step": 32270 }, { "epoch": 0.5871332690663708, "grad_norm": 1.0322682857513428, "learning_rate": 5.074280116772715e-06, "loss": 0.2303, "step": 32280 }, { "epoch": 0.587315156696193, "grad_norm": 1.1265110969543457, "learning_rate": 5.071987863776584e-06, "loss": 0.2359, "step": 32290 }, { "epoch": 0.5874970443260153, "grad_norm": 0.9935548901557922, "learning_rate": 5.06969593283283e-06, "loss": 0.2328, "step": 32300 }, { "epoch": 0.5876789319558376, "grad_norm": 0.991986095905304, "learning_rate": 5.0674043247368306e-06, "loss": 0.2366, "step": 32310 }, { "epoch": 0.5878608195856599, "grad_norm": 0.9952406883239746, "learning_rate": 5.065113040283857e-06, "loss": 0.234, "step": 32320 }, { "epoch": 0.5880427072154822, "grad_norm": 1.0627460479736328, "learning_rate": 5.0628220802690605e-06, "loss": 0.2261, "step": 32330 }, { "epoch": 0.5882245948453045, "grad_norm": 1.0322265625, "learning_rate": 5.060531445487486e-06, "loss": 0.2377, "step": 32340 }, { "epoch": 0.5884064824751268, "grad_norm": 1.1572036743164062, "learning_rate": 5.058241136734062e-06, "loss": 0.2275, "step": 32350 }, { "epoch": 0.5885883701049491, "grad_norm": 0.9923114776611328, "learning_rate": 5.055951154803606e-06, "loss": 0.236, "step": 32360 }, { "epoch": 0.5887702577347714, "grad_norm": 1.0452744960784912, "learning_rate": 5.053661500490819e-06, "loss": 0.2356, "step": 32370 }, { "epoch": 0.5889521453645937, "grad_norm": 0.9131403565406799, "learning_rate": 5.051372174590294e-06, "loss": 0.2331, "step": 32380 }, { "epoch": 0.589134032994416, "grad_norm": 1.0960623025894165, "learning_rate": 5.049083177896503e-06, "loss": 0.2313, "step": 32390 }, { "epoch": 0.5893159206242383, "grad_norm": 1.1776304244995117, "learning_rate": 5.046794511203809e-06, "loss": 0.2304, "step": 32400 }, { "epoch": 0.5894978082540606, "grad_norm": 0.9334163069725037, "learning_rate": 5.044506175306457e-06, "loss": 0.2329, "step": 32410 }, { "epoch": 0.5896796958838829, "grad_norm": 0.9677385091781616, "learning_rate": 5.042218170998582e-06, "loss": 0.2347, "step": 32420 }, { "epoch": 0.5898615835137052, "grad_norm": 0.9925421476364136, "learning_rate": 5.0399304990742e-06, "loss": 0.2244, "step": 32430 }, { "epoch": 0.5900434711435275, "grad_norm": 0.971684992313385, "learning_rate": 5.037643160327208e-06, "loss": 0.2301, "step": 32440 }, { "epoch": 0.5902253587733498, "grad_norm": 0.9995459914207458, "learning_rate": 5.035356155551396e-06, "loss": 0.2319, "step": 32450 }, { "epoch": 0.5904072464031721, "grad_norm": 0.9708480834960938, "learning_rate": 5.033069485540434e-06, "loss": 0.2345, "step": 32460 }, { "epoch": 0.5905891340329944, "grad_norm": 0.9918359518051147, "learning_rate": 5.030783151087874e-06, "loss": 0.233, "step": 32470 }, { "epoch": 0.5907710216628167, "grad_norm": 1.0380576848983765, "learning_rate": 5.028497152987156e-06, "loss": 0.2316, "step": 32480 }, { "epoch": 0.590952909292639, "grad_norm": 0.9545225501060486, "learning_rate": 5.026211492031596e-06, "loss": 0.23, "step": 32490 }, { "epoch": 0.5911347969224613, "grad_norm": 0.9958082437515259, "learning_rate": 5.023926169014404e-06, "loss": 0.2339, "step": 32500 }, { "epoch": 0.5913166845522836, "grad_norm": 0.9508954286575317, "learning_rate": 5.021641184728658e-06, "loss": 0.2341, "step": 32510 }, { "epoch": 0.5914985721821059, "grad_norm": 1.0328869819641113, "learning_rate": 5.0193565399673326e-06, "loss": 0.2318, "step": 32520 }, { "epoch": 0.5916804598119282, "grad_norm": 0.9578921794891357, "learning_rate": 5.017072235523279e-06, "loss": 0.2256, "step": 32530 }, { "epoch": 0.5918623474417505, "grad_norm": 1.0177661180496216, "learning_rate": 5.0147882721892285e-06, "loss": 0.2262, "step": 32540 }, { "epoch": 0.5920442350715728, "grad_norm": 1.0634374618530273, "learning_rate": 5.012504650757794e-06, "loss": 0.2331, "step": 32550 }, { "epoch": 0.5922261227013951, "grad_norm": 1.168020486831665, "learning_rate": 5.010221372021476e-06, "loss": 0.2359, "step": 32560 }, { "epoch": 0.5924080103312174, "grad_norm": 1.0619227886199951, "learning_rate": 5.007938436772645e-06, "loss": 0.2334, "step": 32570 }, { "epoch": 0.5925898979610397, "grad_norm": 0.9602659344673157, "learning_rate": 5.005655845803563e-06, "loss": 0.2362, "step": 32580 }, { "epoch": 0.592771785590862, "grad_norm": 1.0997313261032104, "learning_rate": 5.003373599906366e-06, "loss": 0.2294, "step": 32590 }, { "epoch": 0.5929536732206843, "grad_norm": 1.0982638597488403, "learning_rate": 5.001091699873074e-06, "loss": 0.2333, "step": 32600 }, { "epoch": 0.5931355608505066, "grad_norm": 1.0659058094024658, "learning_rate": 4.998810146495585e-06, "loss": 0.232, "step": 32610 }, { "epoch": 0.5933174484803289, "grad_norm": 0.993105411529541, "learning_rate": 4.996528940565677e-06, "loss": 0.2295, "step": 32620 }, { "epoch": 0.5934993361101512, "grad_norm": 1.08257257938385, "learning_rate": 4.994248082875008e-06, "loss": 0.234, "step": 32630 }, { "epoch": 0.5936812237399735, "grad_norm": 0.9684922099113464, "learning_rate": 4.991967574215113e-06, "loss": 0.2358, "step": 32640 }, { "epoch": 0.5938631113697957, "grad_norm": 1.1110929250717163, "learning_rate": 4.98968741537741e-06, "loss": 0.2353, "step": 32650 }, { "epoch": 0.594044998999618, "grad_norm": 0.9887519478797913, "learning_rate": 4.987407607153191e-06, "loss": 0.2306, "step": 32660 }, { "epoch": 0.5942268866294403, "grad_norm": 1.0224299430847168, "learning_rate": 4.985128150333629e-06, "loss": 0.2316, "step": 32670 }, { "epoch": 0.5944087742592626, "grad_norm": 1.1103401184082031, "learning_rate": 4.982849045709775e-06, "loss": 0.2332, "step": 32680 }, { "epoch": 0.5945906618890849, "grad_norm": 0.968902587890625, "learning_rate": 4.980570294072558e-06, "loss": 0.2292, "step": 32690 }, { "epoch": 0.5947725495189072, "grad_norm": 1.104445457458496, "learning_rate": 4.978291896212783e-06, "loss": 0.2332, "step": 32700 }, { "epoch": 0.5949544371487295, "grad_norm": 1.0184935331344604, "learning_rate": 4.976013852921131e-06, "loss": 0.2313, "step": 32710 }, { "epoch": 0.5951363247785518, "grad_norm": 0.9942048788070679, "learning_rate": 4.973736164988166e-06, "loss": 0.237, "step": 32720 }, { "epoch": 0.5953182124083741, "grad_norm": 0.9892153739929199, "learning_rate": 4.971458833204321e-06, "loss": 0.2279, "step": 32730 }, { "epoch": 0.5955001000381964, "grad_norm": 0.9311028122901917, "learning_rate": 4.969181858359913e-06, "loss": 0.229, "step": 32740 }, { "epoch": 0.5956819876680187, "grad_norm": 1.0920376777648926, "learning_rate": 4.966905241245127e-06, "loss": 0.2329, "step": 32750 }, { "epoch": 0.595863875297841, "grad_norm": 0.9811052083969116, "learning_rate": 4.964628982650029e-06, "loss": 0.2271, "step": 32760 }, { "epoch": 0.5960457629276633, "grad_norm": 1.0900349617004395, "learning_rate": 4.9623530833645624e-06, "loss": 0.2285, "step": 32770 }, { "epoch": 0.5962276505574856, "grad_norm": 1.049673318862915, "learning_rate": 4.960077544178541e-06, "loss": 0.2307, "step": 32780 }, { "epoch": 0.5964095381873079, "grad_norm": 1.0601820945739746, "learning_rate": 4.9578023658816565e-06, "loss": 0.2291, "step": 32790 }, { "epoch": 0.5965914258171302, "grad_norm": 1.000276803970337, "learning_rate": 4.955527549263471e-06, "loss": 0.2334, "step": 32800 }, { "epoch": 0.5967733134469525, "grad_norm": 1.0226107835769653, "learning_rate": 4.953253095113431e-06, "loss": 0.2277, "step": 32810 }, { "epoch": 0.5969552010767748, "grad_norm": 0.987308919429779, "learning_rate": 4.950979004220846e-06, "loss": 0.2306, "step": 32820 }, { "epoch": 0.5971370887065971, "grad_norm": 1.1142116785049438, "learning_rate": 4.948705277374908e-06, "loss": 0.228, "step": 32830 }, { "epoch": 0.5973189763364194, "grad_norm": 1.0344361066818237, "learning_rate": 4.946431915364675e-06, "loss": 0.2268, "step": 32840 }, { "epoch": 0.5975008639662417, "grad_norm": 1.0392388105392456, "learning_rate": 4.944158918979087e-06, "loss": 0.232, "step": 32850 }, { "epoch": 0.597682751596064, "grad_norm": 1.0341469049453735, "learning_rate": 4.941886289006949e-06, "loss": 0.2301, "step": 32860 }, { "epoch": 0.5978646392258863, "grad_norm": 1.0142534971237183, "learning_rate": 4.939614026236944e-06, "loss": 0.2303, "step": 32870 }, { "epoch": 0.5980465268557086, "grad_norm": 1.0441832542419434, "learning_rate": 4.937342131457625e-06, "loss": 0.2315, "step": 32880 }, { "epoch": 0.5982284144855309, "grad_norm": 0.9953135848045349, "learning_rate": 4.9350706054574185e-06, "loss": 0.2354, "step": 32890 }, { "epoch": 0.5984103021153532, "grad_norm": 0.9956700205802917, "learning_rate": 4.932799449024624e-06, "loss": 0.2295, "step": 32900 }, { "epoch": 0.5985921897451755, "grad_norm": 1.1120680570602417, "learning_rate": 4.930528662947409e-06, "loss": 0.2338, "step": 32910 }, { "epoch": 0.5987740773749978, "grad_norm": 1.048804759979248, "learning_rate": 4.928258248013818e-06, "loss": 0.2285, "step": 32920 }, { "epoch": 0.5989559650048201, "grad_norm": 1.1142202615737915, "learning_rate": 4.925988205011759e-06, "loss": 0.2269, "step": 32930 }, { "epoch": 0.5991378526346424, "grad_norm": 0.9053483009338379, "learning_rate": 4.9237185347290214e-06, "loss": 0.227, "step": 32940 }, { "epoch": 0.5993197402644647, "grad_norm": 1.1391067504882812, "learning_rate": 4.921449237953257e-06, "loss": 0.2294, "step": 32950 }, { "epoch": 0.599501627894287, "grad_norm": 1.3665286302566528, "learning_rate": 4.919180315471989e-06, "loss": 0.2314, "step": 32960 }, { "epoch": 0.5996835155241093, "grad_norm": 1.0190109014511108, "learning_rate": 4.916911768072614e-06, "loss": 0.2303, "step": 32970 }, { "epoch": 0.5998654031539316, "grad_norm": 1.0855412483215332, "learning_rate": 4.914643596542395e-06, "loss": 0.2369, "step": 32980 }, { "epoch": 0.6000472907837539, "grad_norm": 1.0429917573928833, "learning_rate": 4.912375801668468e-06, "loss": 0.2319, "step": 32990 }, { "epoch": 0.6002291784135761, "grad_norm": 1.0076863765716553, "learning_rate": 4.910108384237834e-06, "loss": 0.2344, "step": 33000 }, { "epoch": 0.6004110660433983, "grad_norm": 0.9673401117324829, "learning_rate": 4.907841345037367e-06, "loss": 0.231, "step": 33010 }, { "epoch": 0.6005929536732206, "grad_norm": 1.015733003616333, "learning_rate": 4.90557468485381e-06, "loss": 0.2299, "step": 33020 }, { "epoch": 0.6007748413030429, "grad_norm": 0.8853901624679565, "learning_rate": 4.90330840447377e-06, "loss": 0.2296, "step": 33030 }, { "epoch": 0.6009567289328652, "grad_norm": 1.0380003452301025, "learning_rate": 4.901042504683725e-06, "loss": 0.2342, "step": 33040 }, { "epoch": 0.6011386165626875, "grad_norm": 0.9485339522361755, "learning_rate": 4.898776986270021e-06, "loss": 0.2292, "step": 33050 }, { "epoch": 0.6013205041925098, "grad_norm": 1.0085399150848389, "learning_rate": 4.896511850018873e-06, "loss": 0.2333, "step": 33060 }, { "epoch": 0.6015023918223321, "grad_norm": 0.9577423930168152, "learning_rate": 4.894247096716358e-06, "loss": 0.2259, "step": 33070 }, { "epoch": 0.6016842794521544, "grad_norm": 1.0806711912155151, "learning_rate": 4.891982727148429e-06, "loss": 0.2296, "step": 33080 }, { "epoch": 0.6018661670819767, "grad_norm": 1.01913321018219, "learning_rate": 4.889718742100896e-06, "loss": 0.2302, "step": 33090 }, { "epoch": 0.602048054711799, "grad_norm": 1.0265038013458252, "learning_rate": 4.8874551423594455e-06, "loss": 0.232, "step": 33100 }, { "epoch": 0.6022299423416213, "grad_norm": 1.0115931034088135, "learning_rate": 4.8851919287096186e-06, "loss": 0.226, "step": 33110 }, { "epoch": 0.6024118299714436, "grad_norm": 0.8991165161132812, "learning_rate": 4.882929101936835e-06, "loss": 0.2306, "step": 33120 }, { "epoch": 0.6025937176012659, "grad_norm": 1.0026841163635254, "learning_rate": 4.88066666282637e-06, "loss": 0.2308, "step": 33130 }, { "epoch": 0.6027756052310882, "grad_norm": 1.0048400163650513, "learning_rate": 4.87840461216337e-06, "loss": 0.2344, "step": 33140 }, { "epoch": 0.6029574928609105, "grad_norm": 0.9634072780609131, "learning_rate": 4.876142950732845e-06, "loss": 0.2333, "step": 33150 }, { "epoch": 0.6031393804907328, "grad_norm": 1.0011333227157593, "learning_rate": 4.873881679319669e-06, "loss": 0.2332, "step": 33160 }, { "epoch": 0.6033212681205551, "grad_norm": 1.0351907014846802, "learning_rate": 4.871620798708583e-06, "loss": 0.2308, "step": 33170 }, { "epoch": 0.6035031557503774, "grad_norm": 0.9133408069610596, "learning_rate": 4.869360309684188e-06, "loss": 0.23, "step": 33180 }, { "epoch": 0.6036850433801997, "grad_norm": 0.9372546076774597, "learning_rate": 4.867100213030957e-06, "loss": 0.2289, "step": 33190 }, { "epoch": 0.603866931010022, "grad_norm": 1.0844550132751465, "learning_rate": 4.8648405095332164e-06, "loss": 0.232, "step": 33200 }, { "epoch": 0.6040488186398443, "grad_norm": 1.1530841588974, "learning_rate": 4.862581199975165e-06, "loss": 0.2338, "step": 33210 }, { "epoch": 0.6042307062696666, "grad_norm": 1.0191621780395508, "learning_rate": 4.860322285140862e-06, "loss": 0.2295, "step": 33220 }, { "epoch": 0.6044125938994889, "grad_norm": 1.075235366821289, "learning_rate": 4.858063765814227e-06, "loss": 0.2355, "step": 33230 }, { "epoch": 0.6045944815293112, "grad_norm": 1.2071605920791626, "learning_rate": 4.855805642779046e-06, "loss": 0.2379, "step": 33240 }, { "epoch": 0.6047763691591335, "grad_norm": 1.04007887840271, "learning_rate": 4.853547916818964e-06, "loss": 0.2307, "step": 33250 }, { "epoch": 0.6049582567889558, "grad_norm": 1.0171641111373901, "learning_rate": 4.851290588717492e-06, "loss": 0.23, "step": 33260 }, { "epoch": 0.6051401444187781, "grad_norm": 0.9520862698554993, "learning_rate": 4.849033659258001e-06, "loss": 0.228, "step": 33270 }, { "epoch": 0.6053220320486004, "grad_norm": 1.0089293718338013, "learning_rate": 4.846777129223723e-06, "loss": 0.2341, "step": 33280 }, { "epoch": 0.6055039196784227, "grad_norm": 1.056749701499939, "learning_rate": 4.844520999397753e-06, "loss": 0.2321, "step": 33290 }, { "epoch": 0.605685807308245, "grad_norm": 1.0630772113800049, "learning_rate": 4.842265270563045e-06, "loss": 0.2315, "step": 33300 }, { "epoch": 0.6058676949380672, "grad_norm": 0.9077365398406982, "learning_rate": 4.840009943502416e-06, "loss": 0.2272, "step": 33310 }, { "epoch": 0.6060495825678895, "grad_norm": 1.005932331085205, "learning_rate": 4.837755018998542e-06, "loss": 0.2253, "step": 33320 }, { "epoch": 0.6062314701977118, "grad_norm": 0.9848788380622864, "learning_rate": 4.8355004978339625e-06, "loss": 0.2276, "step": 33330 }, { "epoch": 0.6064133578275341, "grad_norm": 1.023958683013916, "learning_rate": 4.833246380791071e-06, "loss": 0.2329, "step": 33340 }, { "epoch": 0.6065952454573564, "grad_norm": 1.077498197555542, "learning_rate": 4.830992668652125e-06, "loss": 0.2296, "step": 33350 }, { "epoch": 0.6067771330871787, "grad_norm": 1.1282366514205933, "learning_rate": 4.828739362199242e-06, "loss": 0.2274, "step": 33360 }, { "epoch": 0.606959020717001, "grad_norm": 0.9238568544387817, "learning_rate": 4.8264864622143956e-06, "loss": 0.2272, "step": 33370 }, { "epoch": 0.6071409083468233, "grad_norm": 1.0985161066055298, "learning_rate": 4.8242339694794225e-06, "loss": 0.2315, "step": 33380 }, { "epoch": 0.6073227959766456, "grad_norm": 1.0611600875854492, "learning_rate": 4.821981884776016e-06, "loss": 0.2378, "step": 33390 }, { "epoch": 0.6075046836064679, "grad_norm": 1.0462846755981445, "learning_rate": 4.819730208885725e-06, "loss": 0.2297, "step": 33400 }, { "epoch": 0.6076865712362902, "grad_norm": 1.0214678049087524, "learning_rate": 4.81747894258996e-06, "loss": 0.2261, "step": 33410 }, { "epoch": 0.6078684588661125, "grad_norm": 1.0545390844345093, "learning_rate": 4.81522808666999e-06, "loss": 0.2355, "step": 33420 }, { "epoch": 0.6080503464959348, "grad_norm": 0.9591225385665894, "learning_rate": 4.812977641906938e-06, "loss": 0.2282, "step": 33430 }, { "epoch": 0.6082322341257571, "grad_norm": 0.9484382271766663, "learning_rate": 4.810727609081787e-06, "loss": 0.2332, "step": 33440 }, { "epoch": 0.6084141217555794, "grad_norm": 1.0720173120498657, "learning_rate": 4.808477988975375e-06, "loss": 0.2283, "step": 33450 }, { "epoch": 0.6085960093854017, "grad_norm": 0.9639154076576233, "learning_rate": 4.806228782368402e-06, "loss": 0.2228, "step": 33460 }, { "epoch": 0.608777897015224, "grad_norm": 1.105881690979004, "learning_rate": 4.803979990041417e-06, "loss": 0.2359, "step": 33470 }, { "epoch": 0.6089597846450463, "grad_norm": 0.9850494265556335, "learning_rate": 4.80173161277483e-06, "loss": 0.2291, "step": 33480 }, { "epoch": 0.6091416722748686, "grad_norm": 1.0181496143341064, "learning_rate": 4.799483651348908e-06, "loss": 0.2308, "step": 33490 }, { "epoch": 0.6093235599046909, "grad_norm": 1.0431053638458252, "learning_rate": 4.797236106543767e-06, "loss": 0.2296, "step": 33500 }, { "epoch": 0.6095054475345132, "grad_norm": 1.075327754020691, "learning_rate": 4.794988979139387e-06, "loss": 0.2265, "step": 33510 }, { "epoch": 0.6096873351643355, "grad_norm": 1.112910270690918, "learning_rate": 4.792742269915597e-06, "loss": 0.2316, "step": 33520 }, { "epoch": 0.6098692227941578, "grad_norm": 1.0932180881500244, "learning_rate": 4.790495979652083e-06, "loss": 0.2242, "step": 33530 }, { "epoch": 0.6100511104239801, "grad_norm": 1.0411957502365112, "learning_rate": 4.788250109128387e-06, "loss": 0.2308, "step": 33540 }, { "epoch": 0.6102329980538024, "grad_norm": 1.0132696628570557, "learning_rate": 4.786004659123903e-06, "loss": 0.2252, "step": 33550 }, { "epoch": 0.6104148856836247, "grad_norm": 1.0605286359786987, "learning_rate": 4.783759630417879e-06, "loss": 0.2393, "step": 33560 }, { "epoch": 0.610596773313447, "grad_norm": 1.082736611366272, "learning_rate": 4.781515023789419e-06, "loss": 0.2356, "step": 33570 }, { "epoch": 0.6107786609432693, "grad_norm": 1.0431078672409058, "learning_rate": 4.7792708400174805e-06, "loss": 0.2286, "step": 33580 }, { "epoch": 0.6109605485730916, "grad_norm": 1.0331953763961792, "learning_rate": 4.77702707988087e-06, "loss": 0.2299, "step": 33590 }, { "epoch": 0.6111424362029139, "grad_norm": 1.0994800329208374, "learning_rate": 4.774783744158254e-06, "loss": 0.2325, "step": 33600 }, { "epoch": 0.6113243238327362, "grad_norm": 0.9995352029800415, "learning_rate": 4.772540833628143e-06, "loss": 0.2352, "step": 33610 }, { "epoch": 0.6115062114625585, "grad_norm": 1.1392416954040527, "learning_rate": 4.770298349068907e-06, "loss": 0.2281, "step": 33620 }, { "epoch": 0.6116880990923808, "grad_norm": 0.9809969663619995, "learning_rate": 4.768056291258766e-06, "loss": 0.23, "step": 33630 }, { "epoch": 0.611869986722203, "grad_norm": 1.125225305557251, "learning_rate": 4.765814660975794e-06, "loss": 0.2337, "step": 33640 }, { "epoch": 0.6120518743520253, "grad_norm": 1.0247304439544678, "learning_rate": 4.7635734589979095e-06, "loss": 0.2321, "step": 33650 }, { "epoch": 0.6122337619818476, "grad_norm": 1.0853139162063599, "learning_rate": 4.761332686102892e-06, "loss": 0.2335, "step": 33660 }, { "epoch": 0.6124156496116699, "grad_norm": 1.1975637674331665, "learning_rate": 4.759092343068365e-06, "loss": 0.2335, "step": 33670 }, { "epoch": 0.6125975372414922, "grad_norm": 0.9638726115226746, "learning_rate": 4.756852430671804e-06, "loss": 0.2251, "step": 33680 }, { "epoch": 0.6127794248713145, "grad_norm": 1.1700873374938965, "learning_rate": 4.754612949690539e-06, "loss": 0.23, "step": 33690 }, { "epoch": 0.6129613125011368, "grad_norm": 1.1248730421066284, "learning_rate": 4.7523739009017456e-06, "loss": 0.2341, "step": 33700 }, { "epoch": 0.6131432001309591, "grad_norm": 0.9498530626296997, "learning_rate": 4.750135285082452e-06, "loss": 0.2312, "step": 33710 }, { "epoch": 0.6133250877607814, "grad_norm": 1.1168748140335083, "learning_rate": 4.747897103009535e-06, "loss": 0.2355, "step": 33720 }, { "epoch": 0.6135069753906037, "grad_norm": 1.0866575241088867, "learning_rate": 4.745659355459724e-06, "loss": 0.229, "step": 33730 }, { "epoch": 0.613688863020426, "grad_norm": 1.0202277898788452, "learning_rate": 4.743422043209591e-06, "loss": 0.2263, "step": 33740 }, { "epoch": 0.6138707506502483, "grad_norm": 0.9893162250518799, "learning_rate": 4.741185167035563e-06, "loss": 0.2283, "step": 33750 }, { "epoch": 0.6140526382800706, "grad_norm": 1.193796992301941, "learning_rate": 4.738948727713915e-06, "loss": 0.2288, "step": 33760 }, { "epoch": 0.6142345259098929, "grad_norm": 1.222607135772705, "learning_rate": 4.736712726020764e-06, "loss": 0.2293, "step": 33770 }, { "epoch": 0.6144164135397152, "grad_norm": 0.9972514510154724, "learning_rate": 4.734477162732085e-06, "loss": 0.2276, "step": 33780 }, { "epoch": 0.6145983011695375, "grad_norm": 0.9688988924026489, "learning_rate": 4.732242038623694e-06, "loss": 0.2331, "step": 33790 }, { "epoch": 0.6147801887993598, "grad_norm": 0.9923522472381592, "learning_rate": 4.730007354471257e-06, "loss": 0.2314, "step": 33800 }, { "epoch": 0.6149620764291821, "grad_norm": 1.0657912492752075, "learning_rate": 4.727773111050287e-06, "loss": 0.2332, "step": 33810 }, { "epoch": 0.6151439640590044, "grad_norm": 1.0580252408981323, "learning_rate": 4.725539309136142e-06, "loss": 0.2329, "step": 33820 }, { "epoch": 0.6153258516888267, "grad_norm": 1.0309321880340576, "learning_rate": 4.723305949504031e-06, "loss": 0.2288, "step": 33830 }, { "epoch": 0.6155077393186489, "grad_norm": 1.1845779418945312, "learning_rate": 4.721073032929006e-06, "loss": 0.228, "step": 33840 }, { "epoch": 0.6156896269484712, "grad_norm": 1.0078984498977661, "learning_rate": 4.718840560185969e-06, "loss": 0.2273, "step": 33850 }, { "epoch": 0.6158715145782935, "grad_norm": 1.1703011989593506, "learning_rate": 4.716608532049661e-06, "loss": 0.2282, "step": 33860 }, { "epoch": 0.6160534022081158, "grad_norm": 0.9256883859634399, "learning_rate": 4.714376949294675e-06, "loss": 0.2321, "step": 33870 }, { "epoch": 0.6162352898379381, "grad_norm": 1.0470062494277954, "learning_rate": 4.71214581269545e-06, "loss": 0.2278, "step": 33880 }, { "epoch": 0.6164171774677604, "grad_norm": 0.9235951900482178, "learning_rate": 4.709915123026265e-06, "loss": 0.2344, "step": 33890 }, { "epoch": 0.6165990650975827, "grad_norm": 0.9984848499298096, "learning_rate": 4.707684881061245e-06, "loss": 0.2314, "step": 33900 }, { "epoch": 0.616780952727405, "grad_norm": 1.077300786972046, "learning_rate": 4.705455087574367e-06, "loss": 0.234, "step": 33910 }, { "epoch": 0.6169628403572273, "grad_norm": 0.8881286382675171, "learning_rate": 4.703225743339441e-06, "loss": 0.2243, "step": 33920 }, { "epoch": 0.6171447279870496, "grad_norm": 1.076494574546814, "learning_rate": 4.700996849130131e-06, "loss": 0.2311, "step": 33930 }, { "epoch": 0.6173266156168719, "grad_norm": 1.1733546257019043, "learning_rate": 4.698768405719937e-06, "loss": 0.2311, "step": 33940 }, { "epoch": 0.6175085032466942, "grad_norm": 1.09995436668396, "learning_rate": 4.696540413882207e-06, "loss": 0.2259, "step": 33950 }, { "epoch": 0.6176903908765164, "grad_norm": 0.9680917859077454, "learning_rate": 4.694312874390133e-06, "loss": 0.2277, "step": 33960 }, { "epoch": 0.6178722785063387, "grad_norm": 0.9790857434272766, "learning_rate": 4.6920857880167484e-06, "loss": 0.2292, "step": 33970 }, { "epoch": 0.618054166136161, "grad_norm": 1.1183606386184692, "learning_rate": 4.6898591555349275e-06, "loss": 0.2302, "step": 33980 }, { "epoch": 0.6182360537659833, "grad_norm": 1.1460540294647217, "learning_rate": 4.687632977717389e-06, "loss": 0.2267, "step": 33990 }, { "epoch": 0.6184179413958056, "grad_norm": 1.0893044471740723, "learning_rate": 4.6854072553366946e-06, "loss": 0.2345, "step": 34000 }, { "epoch": 0.6185998290256279, "grad_norm": 1.1836576461791992, "learning_rate": 4.683181989165249e-06, "loss": 0.2333, "step": 34010 }, { "epoch": 0.6187817166554502, "grad_norm": 0.9718018174171448, "learning_rate": 4.680957179975292e-06, "loss": 0.2233, "step": 34020 }, { "epoch": 0.6189636042852725, "grad_norm": 0.9945033192634583, "learning_rate": 4.678732828538914e-06, "loss": 0.2324, "step": 34030 }, { "epoch": 0.6191454919150948, "grad_norm": 1.0806177854537964, "learning_rate": 4.67650893562804e-06, "loss": 0.2223, "step": 34040 }, { "epoch": 0.6193273795449171, "grad_norm": 0.9443930387496948, "learning_rate": 4.674285502014438e-06, "loss": 0.2281, "step": 34050 }, { "epoch": 0.6195092671747394, "grad_norm": 1.1312743425369263, "learning_rate": 4.672062528469717e-06, "loss": 0.2269, "step": 34060 }, { "epoch": 0.6196911548045617, "grad_norm": 1.0608365535736084, "learning_rate": 4.669840015765327e-06, "loss": 0.2335, "step": 34070 }, { "epoch": 0.619873042434384, "grad_norm": 1.115720510482788, "learning_rate": 4.667617964672554e-06, "loss": 0.2304, "step": 34080 }, { "epoch": 0.6200549300642063, "grad_norm": 0.9386481046676636, "learning_rate": 4.665396375962529e-06, "loss": 0.2337, "step": 34090 }, { "epoch": 0.6202368176940286, "grad_norm": 1.0191057920455933, "learning_rate": 4.663175250406219e-06, "loss": 0.2274, "step": 34100 }, { "epoch": 0.6204187053238509, "grad_norm": 1.0554252862930298, "learning_rate": 4.660954588774431e-06, "loss": 0.233, "step": 34110 }, { "epoch": 0.6206005929536732, "grad_norm": 0.8729039430618286, "learning_rate": 4.6587343918378146e-06, "loss": 0.2244, "step": 34120 }, { "epoch": 0.6207824805834955, "grad_norm": 0.9442662000656128, "learning_rate": 4.656514660366853e-06, "loss": 0.2275, "step": 34130 }, { "epoch": 0.6209643682133178, "grad_norm": 0.9709875583648682, "learning_rate": 4.65429539513187e-06, "loss": 0.2273, "step": 34140 }, { "epoch": 0.6211462558431401, "grad_norm": 1.0142515897750854, "learning_rate": 4.652076596903028e-06, "loss": 0.2345, "step": 34150 }, { "epoch": 0.6213281434729624, "grad_norm": 0.9787991642951965, "learning_rate": 4.649858266450327e-06, "loss": 0.2269, "step": 34160 }, { "epoch": 0.6215100311027847, "grad_norm": 1.1541682481765747, "learning_rate": 4.647640404543603e-06, "loss": 0.2293, "step": 34170 }, { "epoch": 0.621691918732607, "grad_norm": 1.0209717750549316, "learning_rate": 4.645423011952531e-06, "loss": 0.2287, "step": 34180 }, { "epoch": 0.6218738063624293, "grad_norm": 0.999505341053009, "learning_rate": 4.6432060894466255e-06, "loss": 0.2299, "step": 34190 }, { "epoch": 0.6220556939922516, "grad_norm": 1.1190552711486816, "learning_rate": 4.640989637795235e-06, "loss": 0.2334, "step": 34200 }, { "epoch": 0.6222375816220739, "grad_norm": 1.0510551929473877, "learning_rate": 4.638773657767543e-06, "loss": 0.2278, "step": 34210 }, { "epoch": 0.6224194692518962, "grad_norm": 1.0925910472869873, "learning_rate": 4.636558150132573e-06, "loss": 0.235, "step": 34220 }, { "epoch": 0.6226013568817185, "grad_norm": 0.970253586769104, "learning_rate": 4.6343431156591835e-06, "loss": 0.2251, "step": 34230 }, { "epoch": 0.6227832445115408, "grad_norm": 0.9813908934593201, "learning_rate": 4.632128555116067e-06, "loss": 0.2291, "step": 34240 }, { "epoch": 0.6229651321413631, "grad_norm": 1.0540757179260254, "learning_rate": 4.629914469271752e-06, "loss": 0.2264, "step": 34250 }, { "epoch": 0.6231470197711854, "grad_norm": 0.9325709939002991, "learning_rate": 4.627700858894604e-06, "loss": 0.2281, "step": 34260 }, { "epoch": 0.6233289074010077, "grad_norm": 1.021346092224121, "learning_rate": 4.625487724752824e-06, "loss": 0.2294, "step": 34270 }, { "epoch": 0.62351079503083, "grad_norm": 1.1237609386444092, "learning_rate": 4.623275067614443e-06, "loss": 0.2406, "step": 34280 }, { "epoch": 0.6236926826606523, "grad_norm": 1.0674339532852173, "learning_rate": 4.621062888247333e-06, "loss": 0.2376, "step": 34290 }, { "epoch": 0.6238745702904746, "grad_norm": 0.8771055340766907, "learning_rate": 4.618851187419196e-06, "loss": 0.2305, "step": 34300 }, { "epoch": 0.6240564579202968, "grad_norm": 0.944161057472229, "learning_rate": 4.616639965897566e-06, "loss": 0.2223, "step": 34310 }, { "epoch": 0.6242383455501191, "grad_norm": 1.1120994091033936, "learning_rate": 4.614429224449818e-06, "loss": 0.2281, "step": 34320 }, { "epoch": 0.6244202331799414, "grad_norm": 1.050179123878479, "learning_rate": 4.6122189638431535e-06, "loss": 0.2363, "step": 34330 }, { "epoch": 0.6246021208097637, "grad_norm": 1.2617390155792236, "learning_rate": 4.610009184844608e-06, "loss": 0.2288, "step": 34340 }, { "epoch": 0.624784008439586, "grad_norm": 1.0629674196243286, "learning_rate": 4.607799888221055e-06, "loss": 0.2284, "step": 34350 }, { "epoch": 0.6249658960694083, "grad_norm": 0.9821167588233948, "learning_rate": 4.605591074739194e-06, "loss": 0.2252, "step": 34360 }, { "epoch": 0.6251477836992306, "grad_norm": 1.007512092590332, "learning_rate": 4.603382745165562e-06, "loss": 0.2298, "step": 34370 }, { "epoch": 0.6253296713290529, "grad_norm": 0.9306389689445496, "learning_rate": 4.601174900266524e-06, "loss": 0.2302, "step": 34380 }, { "epoch": 0.6255115589588752, "grad_norm": 1.1345140933990479, "learning_rate": 4.598967540808281e-06, "loss": 0.2363, "step": 34390 }, { "epoch": 0.6256934465886975, "grad_norm": 0.9737083315849304, "learning_rate": 4.596760667556861e-06, "loss": 0.2276, "step": 34400 }, { "epoch": 0.6258753342185198, "grad_norm": 0.9895510673522949, "learning_rate": 4.594554281278126e-06, "loss": 0.2341, "step": 34410 }, { "epoch": 0.6260572218483421, "grad_norm": 0.9495417475700378, "learning_rate": 4.59234838273777e-06, "loss": 0.2238, "step": 34420 }, { "epoch": 0.6262391094781644, "grad_norm": 1.046923279762268, "learning_rate": 4.5901429727013136e-06, "loss": 0.2321, "step": 34430 }, { "epoch": 0.6264209971079867, "grad_norm": 0.9924057722091675, "learning_rate": 4.587938051934113e-06, "loss": 0.2294, "step": 34440 }, { "epoch": 0.626602884737809, "grad_norm": 0.952174961566925, "learning_rate": 4.585733621201351e-06, "loss": 0.228, "step": 34450 }, { "epoch": 0.6267847723676313, "grad_norm": 1.0578913688659668, "learning_rate": 4.58352968126804e-06, "loss": 0.2375, "step": 34460 }, { "epoch": 0.6269666599974536, "grad_norm": 1.0101194381713867, "learning_rate": 4.581326232899028e-06, "loss": 0.2294, "step": 34470 }, { "epoch": 0.6271485476272759, "grad_norm": 1.0660455226898193, "learning_rate": 4.579123276858985e-06, "loss": 0.2261, "step": 34480 }, { "epoch": 0.6273304352570982, "grad_norm": 0.9999799132347107, "learning_rate": 4.576920813912411e-06, "loss": 0.2284, "step": 34490 }, { "epoch": 0.6275123228869205, "grad_norm": 0.9943839311599731, "learning_rate": 4.574718844823641e-06, "loss": 0.2313, "step": 34500 }, { "epoch": 0.6276942105167428, "grad_norm": 1.0759775638580322, "learning_rate": 4.572517370356833e-06, "loss": 0.2302, "step": 34510 }, { "epoch": 0.6278760981465651, "grad_norm": 0.9517611265182495, "learning_rate": 4.570316391275974e-06, "loss": 0.2265, "step": 34520 }, { "epoch": 0.6280579857763874, "grad_norm": 1.0085078477859497, "learning_rate": 4.568115908344881e-06, "loss": 0.2276, "step": 34530 }, { "epoch": 0.6282398734062097, "grad_norm": 1.0061146020889282, "learning_rate": 4.5659159223271975e-06, "loss": 0.2391, "step": 34540 }, { "epoch": 0.628421761036032, "grad_norm": 0.9676005840301514, "learning_rate": 4.563716433986396e-06, "loss": 0.2266, "step": 34550 }, { "epoch": 0.6286036486658543, "grad_norm": 1.062000036239624, "learning_rate": 4.5615174440857736e-06, "loss": 0.2304, "step": 34560 }, { "epoch": 0.6287855362956766, "grad_norm": 1.0122674703598022, "learning_rate": 4.559318953388456e-06, "loss": 0.2304, "step": 34570 }, { "epoch": 0.6289674239254989, "grad_norm": 1.189263105392456, "learning_rate": 4.557120962657398e-06, "loss": 0.2313, "step": 34580 }, { "epoch": 0.6291493115553212, "grad_norm": 0.897071123123169, "learning_rate": 4.554923472655376e-06, "loss": 0.228, "step": 34590 }, { "epoch": 0.6293311991851435, "grad_norm": 0.9573798179626465, "learning_rate": 4.5527264841449955e-06, "loss": 0.2329, "step": 34600 }, { "epoch": 0.6295130868149658, "grad_norm": 1.104583740234375, "learning_rate": 4.550529997888689e-06, "loss": 0.2227, "step": 34610 }, { "epoch": 0.6296949744447881, "grad_norm": 1.044461727142334, "learning_rate": 4.548334014648713e-06, "loss": 0.2315, "step": 34620 }, { "epoch": 0.6298768620746104, "grad_norm": 0.91045743227005, "learning_rate": 4.546138535187148e-06, "loss": 0.2275, "step": 34630 }, { "epoch": 0.6300587497044327, "grad_norm": 1.8084471225738525, "learning_rate": 4.543943560265904e-06, "loss": 0.2266, "step": 34640 }, { "epoch": 0.630240637334255, "grad_norm": 1.0223358869552612, "learning_rate": 4.541749090646712e-06, "loss": 0.2306, "step": 34650 }, { "epoch": 0.6304225249640772, "grad_norm": 0.9362687468528748, "learning_rate": 4.53955512709113e-06, "loss": 0.2273, "step": 34660 }, { "epoch": 0.6306044125938995, "grad_norm": 0.9929496645927429, "learning_rate": 4.537361670360538e-06, "loss": 0.2317, "step": 34670 }, { "epoch": 0.6307863002237217, "grad_norm": 1.0559959411621094, "learning_rate": 4.535168721216143e-06, "loss": 0.2306, "step": 34680 }, { "epoch": 0.630968187853544, "grad_norm": 1.2002332210540771, "learning_rate": 4.532976280418972e-06, "loss": 0.2298, "step": 34690 }, { "epoch": 0.6311500754833663, "grad_norm": 1.0531468391418457, "learning_rate": 4.530784348729881e-06, "loss": 0.2263, "step": 34700 }, { "epoch": 0.6313319631131886, "grad_norm": 0.9440687894821167, "learning_rate": 4.528592926909544e-06, "loss": 0.2263, "step": 34710 }, { "epoch": 0.6315138507430109, "grad_norm": 0.9987545013427734, "learning_rate": 4.526402015718462e-06, "loss": 0.2284, "step": 34720 }, { "epoch": 0.6316957383728332, "grad_norm": 0.9656444191932678, "learning_rate": 4.524211615916955e-06, "loss": 0.2275, "step": 34730 }, { "epoch": 0.6318776260026555, "grad_norm": 1.080355167388916, "learning_rate": 4.522021728265168e-06, "loss": 0.2297, "step": 34740 }, { "epoch": 0.6320595136324778, "grad_norm": 1.117940902709961, "learning_rate": 4.51983235352307e-06, "loss": 0.225, "step": 34750 }, { "epoch": 0.6322414012623001, "grad_norm": 1.073404312133789, "learning_rate": 4.51764349245045e-06, "loss": 0.2336, "step": 34760 }, { "epoch": 0.6324232888921224, "grad_norm": 1.0087950229644775, "learning_rate": 4.515455145806916e-06, "loss": 0.2316, "step": 34770 }, { "epoch": 0.6326051765219447, "grad_norm": 0.988588273525238, "learning_rate": 4.513267314351901e-06, "loss": 0.2289, "step": 34780 }, { "epoch": 0.632787064151767, "grad_norm": 1.025473713874817, "learning_rate": 4.511079998844661e-06, "loss": 0.2271, "step": 34790 }, { "epoch": 0.6329689517815893, "grad_norm": 1.0581908226013184, "learning_rate": 4.508893200044267e-06, "loss": 0.231, "step": 34800 }, { "epoch": 0.6331508394114116, "grad_norm": 0.8973144292831421, "learning_rate": 4.506706918709615e-06, "loss": 0.2264, "step": 34810 }, { "epoch": 0.6333327270412339, "grad_norm": 1.0593993663787842, "learning_rate": 4.50452115559942e-06, "loss": 0.2323, "step": 34820 }, { "epoch": 0.6335146146710562, "grad_norm": 0.9977158904075623, "learning_rate": 4.502335911472219e-06, "loss": 0.2236, "step": 34830 }, { "epoch": 0.6336965023008785, "grad_norm": 0.9671915173530579, "learning_rate": 4.500151187086366e-06, "loss": 0.2282, "step": 34840 }, { "epoch": 0.6338783899307008, "grad_norm": 0.9890559911727905, "learning_rate": 4.497966983200036e-06, "loss": 0.2327, "step": 34850 }, { "epoch": 0.6340602775605231, "grad_norm": 0.9826804399490356, "learning_rate": 4.495783300571226e-06, "loss": 0.2239, "step": 34860 }, { "epoch": 0.6342421651903454, "grad_norm": 1.0021919012069702, "learning_rate": 4.493600139957746e-06, "loss": 0.2297, "step": 34870 }, { "epoch": 0.6344240528201677, "grad_norm": 1.049775242805481, "learning_rate": 4.491417502117231e-06, "loss": 0.2258, "step": 34880 }, { "epoch": 0.63460594044999, "grad_norm": 0.9651318192481995, "learning_rate": 4.489235387807132e-06, "loss": 0.2239, "step": 34890 }, { "epoch": 0.6347878280798123, "grad_norm": 1.096611738204956, "learning_rate": 4.487053797784715e-06, "loss": 0.2296, "step": 34900 }, { "epoch": 0.6349697157096346, "grad_norm": 0.937961995601654, "learning_rate": 4.484872732807071e-06, "loss": 0.2301, "step": 34910 }, { "epoch": 0.6351516033394569, "grad_norm": 1.0612543821334839, "learning_rate": 4.482692193631104e-06, "loss": 0.2326, "step": 34920 }, { "epoch": 0.6353334909692792, "grad_norm": 1.1217859983444214, "learning_rate": 4.480512181013535e-06, "loss": 0.2265, "step": 34930 }, { "epoch": 0.6355153785991015, "grad_norm": 0.9832899570465088, "learning_rate": 4.478332695710908e-06, "loss": 0.2378, "step": 34940 }, { "epoch": 0.6356972662289238, "grad_norm": 1.0324209928512573, "learning_rate": 4.476153738479575e-06, "loss": 0.2294, "step": 34950 }, { "epoch": 0.635879153858746, "grad_norm": 0.8893582820892334, "learning_rate": 4.473975310075713e-06, "loss": 0.2328, "step": 34960 }, { "epoch": 0.6360610414885683, "grad_norm": 0.9696807861328125, "learning_rate": 4.471797411255311e-06, "loss": 0.2298, "step": 34970 }, { "epoch": 0.6362429291183906, "grad_norm": 1.0695371627807617, "learning_rate": 4.469620042774176e-06, "loss": 0.2328, "step": 34980 }, { "epoch": 0.6364248167482129, "grad_norm": 1.057108759880066, "learning_rate": 4.46744320538793e-06, "loss": 0.2323, "step": 34990 }, { "epoch": 0.6366067043780352, "grad_norm": 1.0305492877960205, "learning_rate": 4.46526689985201e-06, "loss": 0.2355, "step": 35000 }, { "epoch": 0.6367885920078575, "grad_norm": 1.0061670541763306, "learning_rate": 4.46309112692167e-06, "loss": 0.2239, "step": 35010 }, { "epoch": 0.6369704796376798, "grad_norm": 1.0510038137435913, "learning_rate": 4.460915887351982e-06, "loss": 0.2312, "step": 35020 }, { "epoch": 0.6371523672675021, "grad_norm": 1.1256122589111328, "learning_rate": 4.458741181897823e-06, "loss": 0.2287, "step": 35030 }, { "epoch": 0.6373342548973244, "grad_norm": 1.0301735401153564, "learning_rate": 4.456567011313897e-06, "loss": 0.2341, "step": 35040 }, { "epoch": 0.6375161425271467, "grad_norm": 1.0811245441436768, "learning_rate": 4.4543933763547134e-06, "loss": 0.231, "step": 35050 }, { "epoch": 0.637698030156969, "grad_norm": 1.0196235179901123, "learning_rate": 4.452220277774601e-06, "loss": 0.2336, "step": 35060 }, { "epoch": 0.6378799177867913, "grad_norm": 1.0128467082977295, "learning_rate": 4.450047716327698e-06, "loss": 0.222, "step": 35070 }, { "epoch": 0.6380618054166136, "grad_norm": 1.0267530679702759, "learning_rate": 4.44787569276796e-06, "loss": 0.2252, "step": 35080 }, { "epoch": 0.6382436930464359, "grad_norm": 0.9775059819221497, "learning_rate": 4.445704207849154e-06, "loss": 0.2331, "step": 35090 }, { "epoch": 0.6384255806762582, "grad_norm": 1.0627954006195068, "learning_rate": 4.4435332623248605e-06, "loss": 0.2288, "step": 35100 }, { "epoch": 0.6386074683060805, "grad_norm": 1.2532583475112915, "learning_rate": 4.441362856948473e-06, "loss": 0.2279, "step": 35110 }, { "epoch": 0.6387893559359028, "grad_norm": 0.9441149234771729, "learning_rate": 4.439192992473197e-06, "loss": 0.2278, "step": 35120 }, { "epoch": 0.6389712435657251, "grad_norm": 1.1137416362762451, "learning_rate": 4.437023669652051e-06, "loss": 0.2304, "step": 35130 }, { "epoch": 0.6391531311955474, "grad_norm": 1.0415589809417725, "learning_rate": 4.434854889237866e-06, "loss": 0.2235, "step": 35140 }, { "epoch": 0.6393350188253697, "grad_norm": 1.213550329208374, "learning_rate": 4.432686651983282e-06, "loss": 0.2326, "step": 35150 }, { "epoch": 0.639516906455192, "grad_norm": 1.066896677017212, "learning_rate": 4.430518958640755e-06, "loss": 0.2314, "step": 35160 }, { "epoch": 0.6396987940850143, "grad_norm": 0.9255173802375793, "learning_rate": 4.428351809962547e-06, "loss": 0.2254, "step": 35170 }, { "epoch": 0.6398806817148366, "grad_norm": 1.0145187377929688, "learning_rate": 4.426185206700737e-06, "loss": 0.23, "step": 35180 }, { "epoch": 0.6400625693446589, "grad_norm": 1.0036258697509766, "learning_rate": 4.4240191496072085e-06, "loss": 0.2297, "step": 35190 }, { "epoch": 0.6402444569744812, "grad_norm": 0.9670072793960571, "learning_rate": 4.421853639433659e-06, "loss": 0.2255, "step": 35200 }, { "epoch": 0.6404263446043035, "grad_norm": 1.0939384698867798, "learning_rate": 4.419688676931597e-06, "loss": 0.2282, "step": 35210 }, { "epoch": 0.6406082322341258, "grad_norm": 1.0673348903656006, "learning_rate": 4.417524262852337e-06, "loss": 0.2287, "step": 35220 }, { "epoch": 0.6407901198639481, "grad_norm": 1.0531182289123535, "learning_rate": 4.415360397947009e-06, "loss": 0.2287, "step": 35230 }, { "epoch": 0.6409720074937704, "grad_norm": 1.1888744831085205, "learning_rate": 4.413197082966545e-06, "loss": 0.2288, "step": 35240 }, { "epoch": 0.6411538951235927, "grad_norm": 0.9702010750770569, "learning_rate": 4.411034318661695e-06, "loss": 0.229, "step": 35250 }, { "epoch": 0.641335782753415, "grad_norm": 1.1089341640472412, "learning_rate": 4.40887210578301e-06, "loss": 0.2211, "step": 35260 }, { "epoch": 0.6415176703832373, "grad_norm": 1.0559593439102173, "learning_rate": 4.406710445080854e-06, "loss": 0.227, "step": 35270 }, { "epoch": 0.6416995580130596, "grad_norm": 1.0184268951416016, "learning_rate": 4.404549337305396e-06, "loss": 0.2287, "step": 35280 }, { "epoch": 0.6418814456428819, "grad_norm": 1.0204039812088013, "learning_rate": 4.4023887832066185e-06, "loss": 0.2284, "step": 35290 }, { "epoch": 0.6420633332727042, "grad_norm": 0.9995204210281372, "learning_rate": 4.400228783534307e-06, "loss": 0.2243, "step": 35300 }, { "epoch": 0.6422452209025264, "grad_norm": 1.2048383951187134, "learning_rate": 4.3980693390380565e-06, "loss": 0.2321, "step": 35310 }, { "epoch": 0.6424271085323487, "grad_norm": 1.0676233768463135, "learning_rate": 4.395910450467271e-06, "loss": 0.232, "step": 35320 }, { "epoch": 0.642608996162171, "grad_norm": 0.9660900235176086, "learning_rate": 4.3937521185711555e-06, "loss": 0.2322, "step": 35330 }, { "epoch": 0.6427908837919933, "grad_norm": 0.9588336944580078, "learning_rate": 4.391594344098729e-06, "loss": 0.2262, "step": 35340 }, { "epoch": 0.6429727714218156, "grad_norm": 1.0967961549758911, "learning_rate": 4.389437127798813e-06, "loss": 0.234, "step": 35350 }, { "epoch": 0.6431546590516379, "grad_norm": 1.0954645872116089, "learning_rate": 4.387280470420037e-06, "loss": 0.2306, "step": 35360 }, { "epoch": 0.6433365466814602, "grad_norm": 1.0551491975784302, "learning_rate": 4.385124372710835e-06, "loss": 0.2263, "step": 35370 }, { "epoch": 0.6435184343112825, "grad_norm": 1.0946751832962036, "learning_rate": 4.38296883541945e-06, "loss": 0.2334, "step": 35380 }, { "epoch": 0.6437003219411048, "grad_norm": 1.1892290115356445, "learning_rate": 4.380813859293924e-06, "loss": 0.229, "step": 35390 }, { "epoch": 0.6438822095709271, "grad_norm": 1.0772526264190674, "learning_rate": 4.378659445082112e-06, "loss": 0.2323, "step": 35400 }, { "epoch": 0.6440640972007494, "grad_norm": 1.0363010168075562, "learning_rate": 4.376505593531668e-06, "loss": 0.2316, "step": 35410 }, { "epoch": 0.6442459848305717, "grad_norm": 0.9913173913955688, "learning_rate": 4.3743523053900545e-06, "loss": 0.2265, "step": 35420 }, { "epoch": 0.644427872460394, "grad_norm": 1.051899790763855, "learning_rate": 4.372199581404538e-06, "loss": 0.2303, "step": 35430 }, { "epoch": 0.6446097600902163, "grad_norm": 1.1021772623062134, "learning_rate": 4.370047422322186e-06, "loss": 0.2256, "step": 35440 }, { "epoch": 0.6447916477200386, "grad_norm": 0.8999760150909424, "learning_rate": 4.367895828889874e-06, "loss": 0.2198, "step": 35450 }, { "epoch": 0.6449735353498609, "grad_norm": 1.0771315097808838, "learning_rate": 4.365744801854278e-06, "loss": 0.2281, "step": 35460 }, { "epoch": 0.6451554229796832, "grad_norm": 0.9928827881813049, "learning_rate": 4.363594341961881e-06, "loss": 0.2331, "step": 35470 }, { "epoch": 0.6453373106095055, "grad_norm": 0.9627540111541748, "learning_rate": 4.361444449958966e-06, "loss": 0.2286, "step": 35480 }, { "epoch": 0.6455191982393278, "grad_norm": 1.1254987716674805, "learning_rate": 4.35929512659162e-06, "loss": 0.2288, "step": 35490 }, { "epoch": 0.6457010858691501, "grad_norm": 1.0977801084518433, "learning_rate": 4.357146372605733e-06, "loss": 0.23, "step": 35500 }, { "epoch": 0.6458829734989724, "grad_norm": 1.055215835571289, "learning_rate": 4.354998188746995e-06, "loss": 0.2309, "step": 35510 }, { "epoch": 0.6460648611287946, "grad_norm": 1.0650640726089478, "learning_rate": 4.352850575760904e-06, "loss": 0.2266, "step": 35520 }, { "epoch": 0.6462467487586169, "grad_norm": 1.1715863943099976, "learning_rate": 4.350703534392755e-06, "loss": 0.2339, "step": 35530 }, { "epoch": 0.6464286363884392, "grad_norm": 1.133113145828247, "learning_rate": 4.3485570653876435e-06, "loss": 0.2397, "step": 35540 }, { "epoch": 0.6466105240182615, "grad_norm": 1.025874137878418, "learning_rate": 4.346411169490471e-06, "loss": 0.2206, "step": 35550 }, { "epoch": 0.6467924116480838, "grad_norm": 1.0263713598251343, "learning_rate": 4.344265847445935e-06, "loss": 0.2279, "step": 35560 }, { "epoch": 0.6469742992779061, "grad_norm": 1.0440007448196411, "learning_rate": 4.34212109999854e-06, "loss": 0.227, "step": 35570 }, { "epoch": 0.6471561869077284, "grad_norm": 1.0281603336334229, "learning_rate": 4.339976927892587e-06, "loss": 0.2275, "step": 35580 }, { "epoch": 0.6473380745375507, "grad_norm": 1.0489246845245361, "learning_rate": 4.337833331872176e-06, "loss": 0.23, "step": 35590 }, { "epoch": 0.647519962167373, "grad_norm": 1.0286476612091064, "learning_rate": 4.335690312681211e-06, "loss": 0.2257, "step": 35600 }, { "epoch": 0.6477018497971953, "grad_norm": 1.0878384113311768, "learning_rate": 4.3335478710633934e-06, "loss": 0.2306, "step": 35610 }, { "epoch": 0.6478837374270175, "grad_norm": 0.9689228534698486, "learning_rate": 4.3314060077622255e-06, "loss": 0.2311, "step": 35620 }, { "epoch": 0.6480656250568398, "grad_norm": 1.0111113786697388, "learning_rate": 4.329264723521007e-06, "loss": 0.2265, "step": 35630 }, { "epoch": 0.6482475126866621, "grad_norm": 1.1684194803237915, "learning_rate": 4.327124019082838e-06, "loss": 0.2309, "step": 35640 }, { "epoch": 0.6484294003164844, "grad_norm": 1.0521570444107056, "learning_rate": 4.324983895190616e-06, "loss": 0.2297, "step": 35650 }, { "epoch": 0.6486112879463067, "grad_norm": 1.1856184005737305, "learning_rate": 4.32284435258704e-06, "loss": 0.2312, "step": 35660 }, { "epoch": 0.648793175576129, "grad_norm": 0.8585806488990784, "learning_rate": 4.320705392014605e-06, "loss": 0.2304, "step": 35670 }, { "epoch": 0.6489750632059513, "grad_norm": 1.0937283039093018, "learning_rate": 4.318567014215604e-06, "loss": 0.2246, "step": 35680 }, { "epoch": 0.6491569508357736, "grad_norm": 0.9863806962966919, "learning_rate": 4.316429219932129e-06, "loss": 0.2263, "step": 35690 }, { "epoch": 0.6493388384655959, "grad_norm": 1.1214354038238525, "learning_rate": 4.3142920099060665e-06, "loss": 0.2269, "step": 35700 }, { "epoch": 0.6495207260954182, "grad_norm": 0.9759724140167236, "learning_rate": 4.312155384879105e-06, "loss": 0.2278, "step": 35710 }, { "epoch": 0.6497026137252405, "grad_norm": 0.970343291759491, "learning_rate": 4.310019345592728e-06, "loss": 0.2316, "step": 35720 }, { "epoch": 0.6498845013550628, "grad_norm": 1.0756537914276123, "learning_rate": 4.307883892788212e-06, "loss": 0.2297, "step": 35730 }, { "epoch": 0.6500663889848851, "grad_norm": 1.020764946937561, "learning_rate": 4.3057490272066364e-06, "loss": 0.2295, "step": 35740 }, { "epoch": 0.6502482766147074, "grad_norm": 1.022598385810852, "learning_rate": 4.303614749588869e-06, "loss": 0.2298, "step": 35750 }, { "epoch": 0.6504301642445297, "grad_norm": 1.0732240676879883, "learning_rate": 4.301481060675584e-06, "loss": 0.2346, "step": 35760 }, { "epoch": 0.650612051874352, "grad_norm": 1.0587788820266724, "learning_rate": 4.2993479612072405e-06, "loss": 0.2279, "step": 35770 }, { "epoch": 0.6507939395041743, "grad_norm": 1.1023931503295898, "learning_rate": 4.297215451924102e-06, "loss": 0.2254, "step": 35780 }, { "epoch": 0.6509758271339966, "grad_norm": 0.9371799230575562, "learning_rate": 4.295083533566221e-06, "loss": 0.2274, "step": 35790 }, { "epoch": 0.6511577147638189, "grad_norm": 1.0311172008514404, "learning_rate": 4.292952206873444e-06, "loss": 0.2367, "step": 35800 }, { "epoch": 0.6513396023936412, "grad_norm": 1.0948482751846313, "learning_rate": 4.290821472585422e-06, "loss": 0.2286, "step": 35810 }, { "epoch": 0.6515214900234635, "grad_norm": 1.00542151927948, "learning_rate": 4.288691331441589e-06, "loss": 0.2292, "step": 35820 }, { "epoch": 0.6517033776532858, "grad_norm": 0.944648802280426, "learning_rate": 4.286561784181177e-06, "loss": 0.2325, "step": 35830 }, { "epoch": 0.6518852652831081, "grad_norm": 1.0490294694900513, "learning_rate": 4.284432831543217e-06, "loss": 0.2288, "step": 35840 }, { "epoch": 0.6520671529129304, "grad_norm": 1.0123852491378784, "learning_rate": 4.282304474266525e-06, "loss": 0.2312, "step": 35850 }, { "epoch": 0.6522490405427527, "grad_norm": 0.9818184971809387, "learning_rate": 4.2801767130897175e-06, "loss": 0.2299, "step": 35860 }, { "epoch": 0.652430928172575, "grad_norm": 1.0150995254516602, "learning_rate": 4.2780495487512015e-06, "loss": 0.2301, "step": 35870 }, { "epoch": 0.6526128158023973, "grad_norm": 1.141837477684021, "learning_rate": 4.2759229819891755e-06, "loss": 0.229, "step": 35880 }, { "epoch": 0.6527947034322196, "grad_norm": 0.9928088188171387, "learning_rate": 4.2737970135416315e-06, "loss": 0.2255, "step": 35890 }, { "epoch": 0.6529765910620419, "grad_norm": 1.0572943687438965, "learning_rate": 4.271671644146354e-06, "loss": 0.2376, "step": 35900 }, { "epoch": 0.6531584786918642, "grad_norm": 0.9939349293708801, "learning_rate": 4.269546874540923e-06, "loss": 0.2329, "step": 35910 }, { "epoch": 0.6533403663216865, "grad_norm": 1.029391884803772, "learning_rate": 4.267422705462704e-06, "loss": 0.2327, "step": 35920 }, { "epoch": 0.6535222539515088, "grad_norm": 0.9821317195892334, "learning_rate": 4.265299137648859e-06, "loss": 0.2263, "step": 35930 }, { "epoch": 0.653704141581331, "grad_norm": 1.0932074785232544, "learning_rate": 4.263176171836338e-06, "loss": 0.2224, "step": 35940 }, { "epoch": 0.6538860292111534, "grad_norm": 1.0160936117172241, "learning_rate": 4.261053808761883e-06, "loss": 0.2327, "step": 35950 }, { "epoch": 0.6540679168409756, "grad_norm": 1.0567348003387451, "learning_rate": 4.258932049162031e-06, "loss": 0.2308, "step": 35960 }, { "epoch": 0.654249804470798, "grad_norm": 0.9402511715888977, "learning_rate": 4.256810893773105e-06, "loss": 0.2304, "step": 35970 }, { "epoch": 0.6544316921006202, "grad_norm": 0.9902397394180298, "learning_rate": 4.254690343331218e-06, "loss": 0.23, "step": 35980 }, { "epoch": 0.6546135797304425, "grad_norm": 1.1015371084213257, "learning_rate": 4.252570398572275e-06, "loss": 0.2317, "step": 35990 }, { "epoch": 0.6547954673602648, "grad_norm": 1.0096180438995361, "learning_rate": 4.2504510602319694e-06, "loss": 0.2273, "step": 36000 }, { "epoch": 0.6549773549900871, "grad_norm": 1.0461565256118774, "learning_rate": 4.2483323290457875e-06, "loss": 0.2369, "step": 36010 }, { "epoch": 0.6551592426199094, "grad_norm": 0.9099552631378174, "learning_rate": 4.246214205749002e-06, "loss": 0.2347, "step": 36020 }, { "epoch": 0.6553411302497317, "grad_norm": 1.1422394514083862, "learning_rate": 4.244096691076673e-06, "loss": 0.2309, "step": 36030 }, { "epoch": 0.655523017879554, "grad_norm": 0.9433681964874268, "learning_rate": 4.2419797857636545e-06, "loss": 0.2299, "step": 36040 }, { "epoch": 0.6557049055093763, "grad_norm": 0.9482448697090149, "learning_rate": 4.239863490544581e-06, "loss": 0.2251, "step": 36050 }, { "epoch": 0.6558867931391986, "grad_norm": 0.976412296295166, "learning_rate": 4.237747806153886e-06, "loss": 0.227, "step": 36060 }, { "epoch": 0.6560686807690209, "grad_norm": 1.0689549446105957, "learning_rate": 4.235632733325782e-06, "loss": 0.231, "step": 36070 }, { "epoch": 0.6562505683988432, "grad_norm": 1.081779956817627, "learning_rate": 4.233518272794274e-06, "loss": 0.2307, "step": 36080 }, { "epoch": 0.6564324560286655, "grad_norm": 1.0844422578811646, "learning_rate": 4.231404425293151e-06, "loss": 0.2289, "step": 36090 }, { "epoch": 0.6566143436584878, "grad_norm": 1.1093719005584717, "learning_rate": 4.229291191555994e-06, "loss": 0.2333, "step": 36100 }, { "epoch": 0.6567962312883101, "grad_norm": 1.0415148735046387, "learning_rate": 4.227178572316167e-06, "loss": 0.2289, "step": 36110 }, { "epoch": 0.6569781189181324, "grad_norm": 0.8484458923339844, "learning_rate": 4.225066568306823e-06, "loss": 0.2235, "step": 36120 }, { "epoch": 0.6571600065479547, "grad_norm": 0.9933307766914368, "learning_rate": 4.222955180260899e-06, "loss": 0.2267, "step": 36130 }, { "epoch": 0.657341894177777, "grad_norm": 1.053653597831726, "learning_rate": 4.2208444089111225e-06, "loss": 0.2285, "step": 36140 }, { "epoch": 0.6575237818075993, "grad_norm": 0.929388701915741, "learning_rate": 4.21873425499e-06, "loss": 0.2299, "step": 36150 }, { "epoch": 0.6577056694374216, "grad_norm": 1.1021196842193604, "learning_rate": 4.216624719229832e-06, "loss": 0.2231, "step": 36160 }, { "epoch": 0.6578875570672439, "grad_norm": 1.0605757236480713, "learning_rate": 4.2145158023627e-06, "loss": 0.2289, "step": 36170 }, { "epoch": 0.6580694446970662, "grad_norm": 1.0673308372497559, "learning_rate": 4.21240750512047e-06, "loss": 0.2287, "step": 36180 }, { "epoch": 0.6582513323268885, "grad_norm": 1.1138449907302856, "learning_rate": 4.210299828234795e-06, "loss": 0.2301, "step": 36190 }, { "epoch": 0.6584332199567108, "grad_norm": 0.9576336145401001, "learning_rate": 4.20819277243711e-06, "loss": 0.2268, "step": 36200 }, { "epoch": 0.6586151075865331, "grad_norm": 0.9425252079963684, "learning_rate": 4.206086338458638e-06, "loss": 0.2254, "step": 36210 }, { "epoch": 0.6587969952163554, "grad_norm": 1.0988132953643799, "learning_rate": 4.2039805270303866e-06, "loss": 0.2306, "step": 36220 }, { "epoch": 0.6589788828461777, "grad_norm": 1.058688998222351, "learning_rate": 4.2018753388831425e-06, "loss": 0.2275, "step": 36230 }, { "epoch": 0.659160770476, "grad_norm": 1.089938998222351, "learning_rate": 4.199770774747478e-06, "loss": 0.2295, "step": 36240 }, { "epoch": 0.6593426581058223, "grad_norm": 1.0905506610870361, "learning_rate": 4.197666835353753e-06, "loss": 0.2251, "step": 36250 }, { "epoch": 0.6595245457356446, "grad_norm": 1.021070957183838, "learning_rate": 4.1955635214321065e-06, "loss": 0.2292, "step": 36260 }, { "epoch": 0.6597064333654669, "grad_norm": 1.1195157766342163, "learning_rate": 4.193460833712461e-06, "loss": 0.2297, "step": 36270 }, { "epoch": 0.6598883209952892, "grad_norm": 1.0755164623260498, "learning_rate": 4.1913587729245205e-06, "loss": 0.2241, "step": 36280 }, { "epoch": 0.6600702086251115, "grad_norm": 1.1276775598526, "learning_rate": 4.189257339797774e-06, "loss": 0.2253, "step": 36290 }, { "epoch": 0.6602520962549338, "grad_norm": 0.9993970394134521, "learning_rate": 4.187156535061493e-06, "loss": 0.2286, "step": 36300 }, { "epoch": 0.660433983884756, "grad_norm": 1.0909414291381836, "learning_rate": 4.185056359444728e-06, "loss": 0.2299, "step": 36310 }, { "epoch": 0.6606158715145783, "grad_norm": 1.0830891132354736, "learning_rate": 4.182956813676315e-06, "loss": 0.2243, "step": 36320 }, { "epoch": 0.6607977591444006, "grad_norm": 0.9704794883728027, "learning_rate": 4.180857898484865e-06, "loss": 0.2286, "step": 36330 }, { "epoch": 0.6609796467742229, "grad_norm": 1.072483777999878, "learning_rate": 4.178759614598776e-06, "loss": 0.2294, "step": 36340 }, { "epoch": 0.6611615344040451, "grad_norm": 1.1173909902572632, "learning_rate": 4.176661962746227e-06, "loss": 0.2311, "step": 36350 }, { "epoch": 0.6613434220338674, "grad_norm": 0.9965020418167114, "learning_rate": 4.174564943655176e-06, "loss": 0.2265, "step": 36360 }, { "epoch": 0.6615253096636897, "grad_norm": 0.9390847086906433, "learning_rate": 4.172468558053359e-06, "loss": 0.2243, "step": 36370 }, { "epoch": 0.661707197293512, "grad_norm": 1.0194498300552368, "learning_rate": 4.170372806668294e-06, "loss": 0.2294, "step": 36380 }, { "epoch": 0.6618890849233343, "grad_norm": 0.95277339220047, "learning_rate": 4.168277690227284e-06, "loss": 0.2225, "step": 36390 }, { "epoch": 0.6620709725531566, "grad_norm": 1.0470409393310547, "learning_rate": 4.166183209457402e-06, "loss": 0.2345, "step": 36400 }, { "epoch": 0.6622528601829789, "grad_norm": 1.0388418436050415, "learning_rate": 4.164089365085507e-06, "loss": 0.2322, "step": 36410 }, { "epoch": 0.6624347478128012, "grad_norm": 0.9784204363822937, "learning_rate": 4.161996157838237e-06, "loss": 0.2293, "step": 36420 }, { "epoch": 0.6626166354426235, "grad_norm": 1.0451868772506714, "learning_rate": 4.159903588442007e-06, "loss": 0.2248, "step": 36430 }, { "epoch": 0.6627985230724458, "grad_norm": 1.0621014833450317, "learning_rate": 4.1578116576230085e-06, "loss": 0.232, "step": 36440 }, { "epoch": 0.6629804107022681, "grad_norm": 0.920569658279419, "learning_rate": 4.1557203661072185e-06, "loss": 0.2268, "step": 36450 }, { "epoch": 0.6631622983320904, "grad_norm": 1.010171890258789, "learning_rate": 4.153629714620384e-06, "loss": 0.2298, "step": 36460 }, { "epoch": 0.6633441859619127, "grad_norm": 0.9757713675498962, "learning_rate": 4.151539703888034e-06, "loss": 0.2314, "step": 36470 }, { "epoch": 0.663526073591735, "grad_norm": 0.8729656338691711, "learning_rate": 4.1494503346354774e-06, "loss": 0.2269, "step": 36480 }, { "epoch": 0.6637079612215573, "grad_norm": 0.9141420722007751, "learning_rate": 4.147361607587793e-06, "loss": 0.2264, "step": 36490 }, { "epoch": 0.6638898488513796, "grad_norm": 1.039692759513855, "learning_rate": 4.145273523469846e-06, "loss": 0.2284, "step": 36500 }, { "epoch": 0.6640717364812019, "grad_norm": 1.0489275455474854, "learning_rate": 4.143186083006272e-06, "loss": 0.2284, "step": 36510 }, { "epoch": 0.6642536241110242, "grad_norm": 1.0469385385513306, "learning_rate": 4.1410992869214865e-06, "loss": 0.2267, "step": 36520 }, { "epoch": 0.6644355117408465, "grad_norm": 1.0679316520690918, "learning_rate": 4.139013135939678e-06, "loss": 0.2295, "step": 36530 }, { "epoch": 0.6646173993706688, "grad_norm": 1.0776679515838623, "learning_rate": 4.136927630784812e-06, "loss": 0.2271, "step": 36540 }, { "epoch": 0.6647992870004911, "grad_norm": 0.93632572889328, "learning_rate": 4.134842772180635e-06, "loss": 0.2234, "step": 36550 }, { "epoch": 0.6649811746303134, "grad_norm": 0.9700039029121399, "learning_rate": 4.132758560850663e-06, "loss": 0.2309, "step": 36560 }, { "epoch": 0.6651630622601357, "grad_norm": 1.0095453262329102, "learning_rate": 4.130674997518192e-06, "loss": 0.2248, "step": 36570 }, { "epoch": 0.665344949889958, "grad_norm": 1.0951637029647827, "learning_rate": 4.128592082906287e-06, "loss": 0.2299, "step": 36580 }, { "epoch": 0.6655268375197803, "grad_norm": 1.1420496702194214, "learning_rate": 4.126509817737792e-06, "loss": 0.2299, "step": 36590 }, { "epoch": 0.6657087251496026, "grad_norm": 1.0388160943984985, "learning_rate": 4.124428202735327e-06, "loss": 0.224, "step": 36600 }, { "epoch": 0.6658906127794249, "grad_norm": 0.9397167563438416, "learning_rate": 4.122347238621286e-06, "loss": 0.2318, "step": 36610 }, { "epoch": 0.6660725004092471, "grad_norm": 1.0498284101486206, "learning_rate": 4.120266926117832e-06, "loss": 0.2339, "step": 36620 }, { "epoch": 0.6662543880390694, "grad_norm": 1.0729303359985352, "learning_rate": 4.118187265946906e-06, "loss": 0.2326, "step": 36630 }, { "epoch": 0.6664362756688917, "grad_norm": 1.0018390417099, "learning_rate": 4.116108258830226e-06, "loss": 0.226, "step": 36640 }, { "epoch": 0.666618163298714, "grad_norm": 1.0734063386917114, "learning_rate": 4.114029905489275e-06, "loss": 0.2236, "step": 36650 }, { "epoch": 0.6668000509285363, "grad_norm": 1.0029221773147583, "learning_rate": 4.111952206645316e-06, "loss": 0.231, "step": 36660 }, { "epoch": 0.6669819385583586, "grad_norm": 1.0424234867095947, "learning_rate": 4.109875163019383e-06, "loss": 0.2211, "step": 36670 }, { "epoch": 0.6671638261881809, "grad_norm": 1.0025221109390259, "learning_rate": 4.107798775332279e-06, "loss": 0.2238, "step": 36680 }, { "epoch": 0.6673457138180032, "grad_norm": 1.0510585308074951, "learning_rate": 4.105723044304585e-06, "loss": 0.2294, "step": 36690 }, { "epoch": 0.6675276014478255, "grad_norm": 1.0408614873886108, "learning_rate": 4.103647970656653e-06, "loss": 0.2264, "step": 36700 }, { "epoch": 0.6677094890776478, "grad_norm": 1.0289076566696167, "learning_rate": 4.101573555108604e-06, "loss": 0.2327, "step": 36710 }, { "epoch": 0.6678913767074701, "grad_norm": 1.0130047798156738, "learning_rate": 4.099499798380331e-06, "loss": 0.229, "step": 36720 }, { "epoch": 0.6680732643372924, "grad_norm": 1.109006643295288, "learning_rate": 4.097426701191501e-06, "loss": 0.227, "step": 36730 }, { "epoch": 0.6682551519671147, "grad_norm": 1.0521125793457031, "learning_rate": 4.095354264261548e-06, "loss": 0.2266, "step": 36740 }, { "epoch": 0.668437039596937, "grad_norm": 0.9443548917770386, "learning_rate": 4.093282488309682e-06, "loss": 0.2271, "step": 36750 }, { "epoch": 0.6686189272267593, "grad_norm": 0.9788421988487244, "learning_rate": 4.091211374054881e-06, "loss": 0.2251, "step": 36760 }, { "epoch": 0.6688008148565816, "grad_norm": 1.0574913024902344, "learning_rate": 4.089140922215893e-06, "loss": 0.2335, "step": 36770 }, { "epoch": 0.6689827024864039, "grad_norm": 1.0531178712844849, "learning_rate": 4.0870711335112365e-06, "loss": 0.232, "step": 36780 }, { "epoch": 0.6691645901162262, "grad_norm": 0.9943070411682129, "learning_rate": 4.085002008659198e-06, "loss": 0.2274, "step": 36790 }, { "epoch": 0.6693464777460485, "grad_norm": 1.081281304359436, "learning_rate": 4.082933548377839e-06, "loss": 0.2242, "step": 36800 }, { "epoch": 0.6695283653758708, "grad_norm": 1.0827614068984985, "learning_rate": 4.080865753384985e-06, "loss": 0.2299, "step": 36810 }, { "epoch": 0.6697102530056931, "grad_norm": 1.0653057098388672, "learning_rate": 4.078798624398233e-06, "loss": 0.2269, "step": 36820 }, { "epoch": 0.6698921406355154, "grad_norm": 1.0181727409362793, "learning_rate": 4.0767321621349474e-06, "loss": 0.2315, "step": 36830 }, { "epoch": 0.6700740282653377, "grad_norm": 1.0611629486083984, "learning_rate": 4.0746663673122636e-06, "loss": 0.2211, "step": 36840 }, { "epoch": 0.67025591589516, "grad_norm": 1.0703856945037842, "learning_rate": 4.072601240647084e-06, "loss": 0.2268, "step": 36850 }, { "epoch": 0.6704378035249823, "grad_norm": 0.9948020577430725, "learning_rate": 4.0705367828560785e-06, "loss": 0.2262, "step": 36860 }, { "epoch": 0.6706196911548046, "grad_norm": 1.0253857374191284, "learning_rate": 4.068472994655685e-06, "loss": 0.2226, "step": 36870 }, { "epoch": 0.6708015787846269, "grad_norm": 1.0178947448730469, "learning_rate": 4.066409876762108e-06, "loss": 0.2275, "step": 36880 }, { "epoch": 0.6709834664144492, "grad_norm": 0.9733319282531738, "learning_rate": 4.064347429891326e-06, "loss": 0.226, "step": 36890 }, { "epoch": 0.6711653540442715, "grad_norm": 1.0567940473556519, "learning_rate": 4.062285654759077e-06, "loss": 0.2346, "step": 36900 }, { "epoch": 0.6713472416740938, "grad_norm": 1.1771031618118286, "learning_rate": 4.060224552080866e-06, "loss": 0.2364, "step": 36910 }, { "epoch": 0.6715291293039161, "grad_norm": 0.9492975473403931, "learning_rate": 4.0581641225719705e-06, "loss": 0.2268, "step": 36920 }, { "epoch": 0.6717110169337384, "grad_norm": 1.148786187171936, "learning_rate": 4.0561043669474275e-06, "loss": 0.2253, "step": 36930 }, { "epoch": 0.6718929045635607, "grad_norm": 1.0617055892944336, "learning_rate": 4.054045285922046e-06, "loss": 0.2319, "step": 36940 }, { "epoch": 0.672074792193383, "grad_norm": 0.9874722361564636, "learning_rate": 4.051986880210399e-06, "loss": 0.2315, "step": 36950 }, { "epoch": 0.6722566798232052, "grad_norm": 0.9659709334373474, "learning_rate": 4.049929150526825e-06, "loss": 0.2233, "step": 36960 }, { "epoch": 0.6724385674530275, "grad_norm": 0.9923990964889526, "learning_rate": 4.047872097585425e-06, "loss": 0.2281, "step": 36970 }, { "epoch": 0.6726204550828498, "grad_norm": 1.0497393608093262, "learning_rate": 4.04581572210007e-06, "loss": 0.2263, "step": 36980 }, { "epoch": 0.6728023427126721, "grad_norm": 1.1274715662002563, "learning_rate": 4.04376002478439e-06, "loss": 0.2304, "step": 36990 }, { "epoch": 0.6729842303424944, "grad_norm": 0.9225826263427734, "learning_rate": 4.0417050063517875e-06, "loss": 0.2267, "step": 37000 }, { "epoch": 0.6731661179723167, "grad_norm": 0.9764724969863892, "learning_rate": 4.039650667515423e-06, "loss": 0.232, "step": 37010 }, { "epoch": 0.673348005602139, "grad_norm": 1.0971262454986572, "learning_rate": 4.037597008988224e-06, "loss": 0.2288, "step": 37020 }, { "epoch": 0.6735298932319613, "grad_norm": 1.0430476665496826, "learning_rate": 4.03554403148288e-06, "loss": 0.2261, "step": 37030 }, { "epoch": 0.6737117808617836, "grad_norm": 1.0410863161087036, "learning_rate": 4.033491735711847e-06, "loss": 0.228, "step": 37040 }, { "epoch": 0.6738936684916059, "grad_norm": 0.951379120349884, "learning_rate": 4.031440122387342e-06, "loss": 0.2239, "step": 37050 }, { "epoch": 0.6740755561214282, "grad_norm": 0.8687132000923157, "learning_rate": 4.0293891922213455e-06, "loss": 0.2253, "step": 37060 }, { "epoch": 0.6742574437512505, "grad_norm": 0.8784888386726379, "learning_rate": 4.0273389459256026e-06, "loss": 0.2276, "step": 37070 }, { "epoch": 0.6744393313810728, "grad_norm": 0.9675646424293518, "learning_rate": 4.025289384211617e-06, "loss": 0.2251, "step": 37080 }, { "epoch": 0.6746212190108951, "grad_norm": 0.982462465763092, "learning_rate": 4.023240507790661e-06, "loss": 0.2226, "step": 37090 }, { "epoch": 0.6748031066407174, "grad_norm": 1.0837334394454956, "learning_rate": 4.021192317373764e-06, "loss": 0.2283, "step": 37100 }, { "epoch": 0.6749849942705397, "grad_norm": 1.0249576568603516, "learning_rate": 4.019144813671719e-06, "loss": 0.2314, "step": 37110 }, { "epoch": 0.675166881900362, "grad_norm": 0.9781671166419983, "learning_rate": 4.017097997395083e-06, "loss": 0.2227, "step": 37120 }, { "epoch": 0.6753487695301843, "grad_norm": 1.0584161281585693, "learning_rate": 4.015051869254167e-06, "loss": 0.2251, "step": 37130 }, { "epoch": 0.6755306571600066, "grad_norm": 1.1518276929855347, "learning_rate": 4.013006429959055e-06, "loss": 0.2297, "step": 37140 }, { "epoch": 0.6757125447898289, "grad_norm": 0.989780604839325, "learning_rate": 4.010961680219581e-06, "loss": 0.2222, "step": 37150 }, { "epoch": 0.6758944324196512, "grad_norm": 1.060133934020996, "learning_rate": 4.008917620745345e-06, "loss": 0.235, "step": 37160 }, { "epoch": 0.6760763200494735, "grad_norm": 1.0646179914474487, "learning_rate": 4.006874252245706e-06, "loss": 0.2244, "step": 37170 }, { "epoch": 0.6762582076792958, "grad_norm": 0.9715378284454346, "learning_rate": 4.004831575429784e-06, "loss": 0.2281, "step": 37180 }, { "epoch": 0.676440095309118, "grad_norm": 1.1419854164123535, "learning_rate": 4.0027895910064595e-06, "loss": 0.2361, "step": 37190 }, { "epoch": 0.6766219829389403, "grad_norm": 0.9770339727401733, "learning_rate": 4.000748299684372e-06, "loss": 0.2271, "step": 37200 }, { "epoch": 0.6768038705687626, "grad_norm": 1.0742179155349731, "learning_rate": 3.998707702171917e-06, "loss": 0.2305, "step": 37210 }, { "epoch": 0.6769857581985849, "grad_norm": 1.0290157794952393, "learning_rate": 3.996667799177258e-06, "loss": 0.2266, "step": 37220 }, { "epoch": 0.6771676458284072, "grad_norm": 1.0276693105697632, "learning_rate": 3.994628591408307e-06, "loss": 0.2204, "step": 37230 }, { "epoch": 0.6773495334582295, "grad_norm": 1.1054023504257202, "learning_rate": 3.992590079572744e-06, "loss": 0.2308, "step": 37240 }, { "epoch": 0.6775314210880518, "grad_norm": 1.1471400260925293, "learning_rate": 3.990552264378001e-06, "loss": 0.2298, "step": 37250 }, { "epoch": 0.677713308717874, "grad_norm": 0.9988109469413757, "learning_rate": 3.988515146531271e-06, "loss": 0.2237, "step": 37260 }, { "epoch": 0.6778951963476963, "grad_norm": 0.9288300275802612, "learning_rate": 3.986478726739504e-06, "loss": 0.2231, "step": 37270 }, { "epoch": 0.6780770839775186, "grad_norm": 0.9838886260986328, "learning_rate": 3.984443005709407e-06, "loss": 0.2293, "step": 37280 }, { "epoch": 0.6782589716073409, "grad_norm": 0.9939622282981873, "learning_rate": 3.982407984147449e-06, "loss": 0.2293, "step": 37290 }, { "epoch": 0.6784408592371632, "grad_norm": 0.9832051396369934, "learning_rate": 3.980373662759851e-06, "loss": 0.2267, "step": 37300 }, { "epoch": 0.6786227468669855, "grad_norm": 1.1446959972381592, "learning_rate": 3.978340042252594e-06, "loss": 0.2243, "step": 37310 }, { "epoch": 0.6788046344968078, "grad_norm": 1.0628173351287842, "learning_rate": 3.976307123331414e-06, "loss": 0.232, "step": 37320 }, { "epoch": 0.6789865221266301, "grad_norm": 1.2291074991226196, "learning_rate": 3.974274906701803e-06, "loss": 0.2303, "step": 37330 }, { "epoch": 0.6791684097564524, "grad_norm": 0.8970046639442444, "learning_rate": 3.972243393069013e-06, "loss": 0.2245, "step": 37340 }, { "epoch": 0.6793502973862747, "grad_norm": 1.0530948638916016, "learning_rate": 3.970212583138049e-06, "loss": 0.2257, "step": 37350 }, { "epoch": 0.679532185016097, "grad_norm": 0.9542867541313171, "learning_rate": 3.968182477613672e-06, "loss": 0.2214, "step": 37360 }, { "epoch": 0.6797140726459193, "grad_norm": 0.97577965259552, "learning_rate": 3.9661530772004e-06, "loss": 0.2263, "step": 37370 }, { "epoch": 0.6798959602757416, "grad_norm": 0.9261910915374756, "learning_rate": 3.964124382602503e-06, "loss": 0.2234, "step": 37380 }, { "epoch": 0.6800778479055639, "grad_norm": 1.019255518913269, "learning_rate": 3.96209639452401e-06, "loss": 0.2295, "step": 37390 }, { "epoch": 0.6802597355353862, "grad_norm": 0.9418621063232422, "learning_rate": 3.960069113668703e-06, "loss": 0.2274, "step": 37400 }, { "epoch": 0.6804416231652085, "grad_norm": 1.0295920372009277, "learning_rate": 3.95804254074012e-06, "loss": 0.227, "step": 37410 }, { "epoch": 0.6806235107950308, "grad_norm": 1.1472117900848389, "learning_rate": 3.956016676441548e-06, "loss": 0.2329, "step": 37420 }, { "epoch": 0.6808053984248531, "grad_norm": 0.908531665802002, "learning_rate": 3.9539915214760366e-06, "loss": 0.2243, "step": 37430 }, { "epoch": 0.6809872860546754, "grad_norm": 1.101921796798706, "learning_rate": 3.951967076546383e-06, "loss": 0.2321, "step": 37440 }, { "epoch": 0.6811691736844977, "grad_norm": 1.0857964754104614, "learning_rate": 3.9499433423551406e-06, "loss": 0.2318, "step": 37450 }, { "epoch": 0.68135106131432, "grad_norm": 1.106821894645691, "learning_rate": 3.9479203196046134e-06, "loss": 0.2322, "step": 37460 }, { "epoch": 0.6815329489441423, "grad_norm": 0.9697529673576355, "learning_rate": 3.9458980089968605e-06, "loss": 0.228, "step": 37470 }, { "epoch": 0.6817148365739646, "grad_norm": 1.0069853067398071, "learning_rate": 3.9438764112336936e-06, "loss": 0.2311, "step": 37480 }, { "epoch": 0.6818967242037869, "grad_norm": 0.9977846741676331, "learning_rate": 3.941855527016681e-06, "loss": 0.2198, "step": 37490 }, { "epoch": 0.6820786118336092, "grad_norm": 1.086702823638916, "learning_rate": 3.939835357047137e-06, "loss": 0.2277, "step": 37500 }, { "epoch": 0.6822604994634315, "grad_norm": 1.1763324737548828, "learning_rate": 3.937815902026129e-06, "loss": 0.2252, "step": 37510 }, { "epoch": 0.6824423870932538, "grad_norm": 1.0435668230056763, "learning_rate": 3.935797162654481e-06, "loss": 0.2215, "step": 37520 }, { "epoch": 0.6826242747230761, "grad_norm": 0.9678309559822083, "learning_rate": 3.933779139632763e-06, "loss": 0.2317, "step": 37530 }, { "epoch": 0.6828061623528984, "grad_norm": 0.9414541125297546, "learning_rate": 3.9317618336613e-06, "loss": 0.2226, "step": 37540 }, { "epoch": 0.6829880499827207, "grad_norm": 1.132460594177246, "learning_rate": 3.929745245440168e-06, "loss": 0.2261, "step": 37550 }, { "epoch": 0.683169937612543, "grad_norm": 1.0514429807662964, "learning_rate": 3.92772937566919e-06, "loss": 0.234, "step": 37560 }, { "epoch": 0.6833518252423653, "grad_norm": 1.0821878910064697, "learning_rate": 3.925714225047946e-06, "loss": 0.2339, "step": 37570 }, { "epoch": 0.6835337128721876, "grad_norm": 0.9907398819923401, "learning_rate": 3.923699794275761e-06, "loss": 0.2244, "step": 37580 }, { "epoch": 0.6837156005020099, "grad_norm": 1.0719683170318604, "learning_rate": 3.921686084051713e-06, "loss": 0.2237, "step": 37590 }, { "epoch": 0.6838974881318322, "grad_norm": 1.0958715677261353, "learning_rate": 3.919673095074631e-06, "loss": 0.2306, "step": 37600 }, { "epoch": 0.6840793757616545, "grad_norm": 0.9991689324378967, "learning_rate": 3.917660828043088e-06, "loss": 0.2254, "step": 37610 }, { "epoch": 0.6842612633914767, "grad_norm": 1.0699479579925537, "learning_rate": 3.915649283655413e-06, "loss": 0.2236, "step": 37620 }, { "epoch": 0.684443151021299, "grad_norm": 0.963096022605896, "learning_rate": 3.9136384626096825e-06, "loss": 0.2284, "step": 37630 }, { "epoch": 0.6846250386511213, "grad_norm": 0.8829547166824341, "learning_rate": 3.9116283656037204e-06, "loss": 0.229, "step": 37640 }, { "epoch": 0.6848069262809436, "grad_norm": 0.9818018674850464, "learning_rate": 3.9096189933350985e-06, "loss": 0.2253, "step": 37650 }, { "epoch": 0.6849888139107659, "grad_norm": 1.0232049226760864, "learning_rate": 3.907610346501139e-06, "loss": 0.2284, "step": 37660 }, { "epoch": 0.6851707015405882, "grad_norm": 1.0689998865127563, "learning_rate": 3.905602425798912e-06, "loss": 0.2296, "step": 37670 }, { "epoch": 0.6853525891704105, "grad_norm": 0.9759604930877686, "learning_rate": 3.903595231925238e-06, "loss": 0.2248, "step": 37680 }, { "epoch": 0.6855344768002328, "grad_norm": 0.9803677201271057, "learning_rate": 3.90158876557668e-06, "loss": 0.2275, "step": 37690 }, { "epoch": 0.6857163644300551, "grad_norm": 0.9475717544555664, "learning_rate": 3.899583027449554e-06, "loss": 0.2298, "step": 37700 }, { "epoch": 0.6858982520598774, "grad_norm": 0.9996252059936523, "learning_rate": 3.897578018239918e-06, "loss": 0.2238, "step": 37710 }, { "epoch": 0.6860801396896997, "grad_norm": 1.040241003036499, "learning_rate": 3.895573738643579e-06, "loss": 0.2271, "step": 37720 }, { "epoch": 0.686262027319522, "grad_norm": 1.0512491464614868, "learning_rate": 3.893570189356095e-06, "loss": 0.2262, "step": 37730 }, { "epoch": 0.6864439149493443, "grad_norm": 1.127382516860962, "learning_rate": 3.891567371072766e-06, "loss": 0.2267, "step": 37740 }, { "epoch": 0.6866258025791666, "grad_norm": 1.1362277269363403, "learning_rate": 3.889565284488638e-06, "loss": 0.2351, "step": 37750 }, { "epoch": 0.6868076902089889, "grad_norm": 1.0016266107559204, "learning_rate": 3.887563930298503e-06, "loss": 0.2271, "step": 37760 }, { "epoch": 0.6869895778388112, "grad_norm": 0.9184446334838867, "learning_rate": 3.885563309196905e-06, "loss": 0.2337, "step": 37770 }, { "epoch": 0.6871714654686335, "grad_norm": 0.963651180267334, "learning_rate": 3.883563421878124e-06, "loss": 0.2272, "step": 37780 }, { "epoch": 0.6873533530984558, "grad_norm": 0.9826354384422302, "learning_rate": 3.881564269036193e-06, "loss": 0.2211, "step": 37790 }, { "epoch": 0.6875352407282781, "grad_norm": 1.0465344190597534, "learning_rate": 3.879565851364888e-06, "loss": 0.2325, "step": 37800 }, { "epoch": 0.6877171283581004, "grad_norm": 1.0330665111541748, "learning_rate": 3.877568169557728e-06, "loss": 0.224, "step": 37810 }, { "epoch": 0.6878990159879227, "grad_norm": 1.0436604022979736, "learning_rate": 3.8755712243079765e-06, "loss": 0.22, "step": 37820 }, { "epoch": 0.688080903617745, "grad_norm": 1.1160664558410645, "learning_rate": 3.873575016308645e-06, "loss": 0.223, "step": 37830 }, { "epoch": 0.6882627912475673, "grad_norm": 0.991527795791626, "learning_rate": 3.871579546252486e-06, "loss": 0.2256, "step": 37840 }, { "epoch": 0.6884446788773896, "grad_norm": 0.9470672011375427, "learning_rate": 3.869584814831996e-06, "loss": 0.2298, "step": 37850 }, { "epoch": 0.6886265665072119, "grad_norm": 1.0025269985198975, "learning_rate": 3.867590822739417e-06, "loss": 0.2224, "step": 37860 }, { "epoch": 0.6888084541370342, "grad_norm": 1.0016402006149292, "learning_rate": 3.865597570666732e-06, "loss": 0.2274, "step": 37870 }, { "epoch": 0.6889903417668565, "grad_norm": 1.0634032487869263, "learning_rate": 3.863605059305671e-06, "loss": 0.2301, "step": 37880 }, { "epoch": 0.6891722293966788, "grad_norm": 1.0739128589630127, "learning_rate": 3.861613289347702e-06, "loss": 0.2263, "step": 37890 }, { "epoch": 0.6893541170265011, "grad_norm": 1.0474637746810913, "learning_rate": 3.859622261484039e-06, "loss": 0.2248, "step": 37900 }, { "epoch": 0.6895360046563234, "grad_norm": 0.972844660282135, "learning_rate": 3.857631976405639e-06, "loss": 0.2254, "step": 37910 }, { "epoch": 0.6897178922861457, "grad_norm": 1.1449986696243286, "learning_rate": 3.855642434803198e-06, "loss": 0.2289, "step": 37920 }, { "epoch": 0.689899779915968, "grad_norm": 1.0049103498458862, "learning_rate": 3.853653637367156e-06, "loss": 0.2242, "step": 37930 }, { "epoch": 0.6900816675457903, "grad_norm": 0.990366518497467, "learning_rate": 3.851665584787697e-06, "loss": 0.2228, "step": 37940 }, { "epoch": 0.6902635551756126, "grad_norm": 1.007523536682129, "learning_rate": 3.849678277754742e-06, "loss": 0.2269, "step": 37950 }, { "epoch": 0.6904454428054349, "grad_norm": 0.9775503277778625, "learning_rate": 3.847691716957957e-06, "loss": 0.2246, "step": 37960 }, { "epoch": 0.6906273304352571, "grad_norm": 1.0246235132217407, "learning_rate": 3.845705903086742e-06, "loss": 0.2257, "step": 37970 }, { "epoch": 0.6908092180650794, "grad_norm": 1.0333411693572998, "learning_rate": 3.8437208368302515e-06, "loss": 0.2215, "step": 37980 }, { "epoch": 0.6909911056949017, "grad_norm": 1.0212043523788452, "learning_rate": 3.841736518877368e-06, "loss": 0.2306, "step": 37990 }, { "epoch": 0.691172993324724, "grad_norm": 1.100812315940857, "learning_rate": 3.839752949916719e-06, "loss": 0.229, "step": 38000 }, { "epoch": 0.6913548809545463, "grad_norm": 1.0717827081680298, "learning_rate": 3.8377701306366695e-06, "loss": 0.2237, "step": 38010 }, { "epoch": 0.6915367685843685, "grad_norm": 0.965070366859436, "learning_rate": 3.835788061725332e-06, "loss": 0.2266, "step": 38020 }, { "epoch": 0.6917186562141908, "grad_norm": 1.082040786743164, "learning_rate": 3.8338067438705475e-06, "loss": 0.2337, "step": 38030 }, { "epoch": 0.6919005438440131, "grad_norm": 1.150926947593689, "learning_rate": 3.8318261777599045e-06, "loss": 0.2327, "step": 38040 }, { "epoch": 0.6920824314738354, "grad_norm": 1.080736517906189, "learning_rate": 3.829846364080728e-06, "loss": 0.227, "step": 38050 }, { "epoch": 0.6922643191036577, "grad_norm": 1.0034098625183105, "learning_rate": 3.827867303520083e-06, "loss": 0.2313, "step": 38060 }, { "epoch": 0.69244620673348, "grad_norm": 1.0631098747253418, "learning_rate": 3.8258889967647695e-06, "loss": 0.2255, "step": 38070 }, { "epoch": 0.6926280943633023, "grad_norm": 0.92012619972229, "learning_rate": 3.82391144450133e-06, "loss": 0.2203, "step": 38080 }, { "epoch": 0.6928099819931246, "grad_norm": 1.101719617843628, "learning_rate": 3.8219346474160454e-06, "loss": 0.2282, "step": 38090 }, { "epoch": 0.6929918696229469, "grad_norm": 1.009193778038025, "learning_rate": 3.81995860619493e-06, "loss": 0.2354, "step": 38100 }, { "epoch": 0.6931737572527692, "grad_norm": 0.9369474053382874, "learning_rate": 3.81798332152374e-06, "loss": 0.2207, "step": 38110 }, { "epoch": 0.6933556448825915, "grad_norm": 1.0513299703598022, "learning_rate": 3.816008794087965e-06, "loss": 0.2256, "step": 38120 }, { "epoch": 0.6935375325124138, "grad_norm": 1.0351046323776245, "learning_rate": 3.8140350245728397e-06, "loss": 0.2362, "step": 38130 }, { "epoch": 0.6937194201422361, "grad_norm": 0.8524655699729919, "learning_rate": 3.812062013663327e-06, "loss": 0.2282, "step": 38140 }, { "epoch": 0.6939013077720584, "grad_norm": 0.9966171979904175, "learning_rate": 3.81008976204413e-06, "loss": 0.2261, "step": 38150 }, { "epoch": 0.6940831954018807, "grad_norm": 0.9128750562667847, "learning_rate": 3.80811827039969e-06, "loss": 0.2281, "step": 38160 }, { "epoch": 0.694265083031703, "grad_norm": 1.0195536613464355, "learning_rate": 3.80614753941418e-06, "loss": 0.2219, "step": 38170 }, { "epoch": 0.6944469706615253, "grad_norm": 1.1172174215316772, "learning_rate": 3.8041775697715155e-06, "loss": 0.2278, "step": 38180 }, { "epoch": 0.6946288582913476, "grad_norm": 0.9880733489990234, "learning_rate": 3.802208362155342e-06, "loss": 0.225, "step": 38190 }, { "epoch": 0.6948107459211699, "grad_norm": 1.044958233833313, "learning_rate": 3.8002399172490445e-06, "loss": 0.2269, "step": 38200 }, { "epoch": 0.6949926335509922, "grad_norm": 0.9557856917381287, "learning_rate": 3.798272235735738e-06, "loss": 0.2265, "step": 38210 }, { "epoch": 0.6951745211808145, "grad_norm": 1.08904230594635, "learning_rate": 3.7963053182982814e-06, "loss": 0.2302, "step": 38220 }, { "epoch": 0.6953564088106368, "grad_norm": 0.9989499449729919, "learning_rate": 3.7943391656192598e-06, "loss": 0.2239, "step": 38230 }, { "epoch": 0.6955382964404591, "grad_norm": 1.088232398033142, "learning_rate": 3.792373778380998e-06, "loss": 0.2308, "step": 38240 }, { "epoch": 0.6957201840702814, "grad_norm": 0.9884992241859436, "learning_rate": 3.790409157265551e-06, "loss": 0.2223, "step": 38250 }, { "epoch": 0.6959020717001037, "grad_norm": 0.9894301891326904, "learning_rate": 3.788445302954712e-06, "loss": 0.2212, "step": 38260 }, { "epoch": 0.696083959329926, "grad_norm": 0.9807180762290955, "learning_rate": 3.7864822161300076e-06, "loss": 0.2295, "step": 38270 }, { "epoch": 0.6962658469597482, "grad_norm": 1.004472017288208, "learning_rate": 3.7845198974726956e-06, "loss": 0.2341, "step": 38280 }, { "epoch": 0.6964477345895705, "grad_norm": 0.9250956773757935, "learning_rate": 3.782558347663769e-06, "loss": 0.223, "step": 38290 }, { "epoch": 0.6966296222193928, "grad_norm": 1.1322519779205322, "learning_rate": 3.7805975673839533e-06, "loss": 0.2274, "step": 38300 }, { "epoch": 0.6968115098492151, "grad_norm": 0.9694535732269287, "learning_rate": 3.778637557313707e-06, "loss": 0.2241, "step": 38310 }, { "epoch": 0.6969933974790374, "grad_norm": 0.9457369446754456, "learning_rate": 3.776678318133221e-06, "loss": 0.2266, "step": 38320 }, { "epoch": 0.6971752851088597, "grad_norm": 1.0462794303894043, "learning_rate": 3.7747198505224214e-06, "loss": 0.2258, "step": 38330 }, { "epoch": 0.697357172738682, "grad_norm": 0.9890527725219727, "learning_rate": 3.772762155160964e-06, "loss": 0.2305, "step": 38340 }, { "epoch": 0.6975390603685043, "grad_norm": 1.0786572694778442, "learning_rate": 3.770805232728235e-06, "loss": 0.227, "step": 38350 }, { "epoch": 0.6977209479983266, "grad_norm": 0.9103183150291443, "learning_rate": 3.7688490839033552e-06, "loss": 0.2273, "step": 38360 }, { "epoch": 0.6979028356281489, "grad_norm": 1.0190715789794922, "learning_rate": 3.7668937093651756e-06, "loss": 0.2241, "step": 38370 }, { "epoch": 0.6980847232579712, "grad_norm": 0.9721336364746094, "learning_rate": 3.7649391097922804e-06, "loss": 0.2248, "step": 38380 }, { "epoch": 0.6982666108877935, "grad_norm": 0.9843193292617798, "learning_rate": 3.7629852858629827e-06, "loss": 0.2298, "step": 38390 }, { "epoch": 0.6984484985176158, "grad_norm": 1.0629949569702148, "learning_rate": 3.7610322382553266e-06, "loss": 0.2229, "step": 38400 }, { "epoch": 0.6986303861474381, "grad_norm": 1.054898977279663, "learning_rate": 3.759079967647087e-06, "loss": 0.2264, "step": 38410 }, { "epoch": 0.6988122737772604, "grad_norm": 0.9919925928115845, "learning_rate": 3.757128474715771e-06, "loss": 0.2252, "step": 38420 }, { "epoch": 0.6989941614070827, "grad_norm": 1.0753018856048584, "learning_rate": 3.755177760138613e-06, "loss": 0.2313, "step": 38430 }, { "epoch": 0.699176049036905, "grad_norm": 1.0736924409866333, "learning_rate": 3.75322782459258e-06, "loss": 0.2252, "step": 38440 }, { "epoch": 0.6993579366667273, "grad_norm": 1.013752818107605, "learning_rate": 3.7512786687543655e-06, "loss": 0.2266, "step": 38450 }, { "epoch": 0.6995398242965496, "grad_norm": 0.9932569861412048, "learning_rate": 3.749330293300394e-06, "loss": 0.2277, "step": 38460 }, { "epoch": 0.6997217119263719, "grad_norm": 1.0183956623077393, "learning_rate": 3.747382698906821e-06, "loss": 0.2277, "step": 38470 }, { "epoch": 0.6999035995561942, "grad_norm": 0.9572027325630188, "learning_rate": 3.7454358862495284e-06, "loss": 0.2244, "step": 38480 }, { "epoch": 0.7000854871860165, "grad_norm": 1.047674298286438, "learning_rate": 3.743489856004127e-06, "loss": 0.2259, "step": 38490 }, { "epoch": 0.7002673748158388, "grad_norm": 1.1051393747329712, "learning_rate": 3.741544608845958e-06, "loss": 0.2303, "step": 38500 }, { "epoch": 0.7004492624456611, "grad_norm": 0.8885807991027832, "learning_rate": 3.739600145450088e-06, "loss": 0.2266, "step": 38510 }, { "epoch": 0.7006311500754834, "grad_norm": 1.172953486442566, "learning_rate": 3.737656466491315e-06, "loss": 0.2277, "step": 38520 }, { "epoch": 0.7008130377053057, "grad_norm": 1.119729995727539, "learning_rate": 3.735713572644163e-06, "loss": 0.2308, "step": 38530 }, { "epoch": 0.700994925335128, "grad_norm": 0.9139930605888367, "learning_rate": 3.733771464582882e-06, "loss": 0.2177, "step": 38540 }, { "epoch": 0.7011768129649503, "grad_norm": 1.0375818014144897, "learning_rate": 3.731830142981452e-06, "loss": 0.2286, "step": 38550 }, { "epoch": 0.7013587005947726, "grad_norm": 1.0134018659591675, "learning_rate": 3.729889608513576e-06, "loss": 0.2212, "step": 38560 }, { "epoch": 0.7015405882245949, "grad_norm": 0.967987596988678, "learning_rate": 3.727949861852691e-06, "loss": 0.2258, "step": 38570 }, { "epoch": 0.7017224758544172, "grad_norm": 1.0550811290740967, "learning_rate": 3.726010903671955e-06, "loss": 0.2252, "step": 38580 }, { "epoch": 0.7019043634842395, "grad_norm": 1.014256238937378, "learning_rate": 3.7240727346442508e-06, "loss": 0.2268, "step": 38590 }, { "epoch": 0.7020862511140618, "grad_norm": 1.0545783042907715, "learning_rate": 3.7221353554421948e-06, "loss": 0.2249, "step": 38600 }, { "epoch": 0.702268138743884, "grad_norm": 0.9440546035766602, "learning_rate": 3.72019876673812e-06, "loss": 0.2243, "step": 38610 }, { "epoch": 0.7024500263737063, "grad_norm": 1.0820131301879883, "learning_rate": 3.7182629692040937e-06, "loss": 0.2267, "step": 38620 }, { "epoch": 0.7026319140035286, "grad_norm": 1.0345181226730347, "learning_rate": 3.716327963511904e-06, "loss": 0.2247, "step": 38630 }, { "epoch": 0.7028138016333509, "grad_norm": 0.8967955112457275, "learning_rate": 3.7143937503330647e-06, "loss": 0.2267, "step": 38640 }, { "epoch": 0.7029956892631732, "grad_norm": 1.0786789655685425, "learning_rate": 3.712460330338814e-06, "loss": 0.2304, "step": 38650 }, { "epoch": 0.7031775768929955, "grad_norm": 1.0823158025741577, "learning_rate": 3.710527704200114e-06, "loss": 0.2253, "step": 38660 }, { "epoch": 0.7033594645228178, "grad_norm": 1.044409155845642, "learning_rate": 3.708595872587657e-06, "loss": 0.2239, "step": 38670 }, { "epoch": 0.7035413521526401, "grad_norm": 0.928075909614563, "learning_rate": 3.706664836171853e-06, "loss": 0.2157, "step": 38680 }, { "epoch": 0.7037232397824624, "grad_norm": 1.131911039352417, "learning_rate": 3.70473459562284e-06, "loss": 0.2243, "step": 38690 }, { "epoch": 0.7039051274122847, "grad_norm": 1.1055309772491455, "learning_rate": 3.702805151610476e-06, "loss": 0.2273, "step": 38700 }, { "epoch": 0.704087015042107, "grad_norm": 1.1578195095062256, "learning_rate": 3.700876504804346e-06, "loss": 0.2317, "step": 38710 }, { "epoch": 0.7042689026719293, "grad_norm": 1.0823307037353516, "learning_rate": 3.698948655873759e-06, "loss": 0.2263, "step": 38720 }, { "epoch": 0.7044507903017516, "grad_norm": 1.0200296640396118, "learning_rate": 3.697021605487744e-06, "loss": 0.2323, "step": 38730 }, { "epoch": 0.7046326779315739, "grad_norm": 0.9948307871818542, "learning_rate": 3.6950953543150538e-06, "loss": 0.2287, "step": 38740 }, { "epoch": 0.7048145655613962, "grad_norm": 0.9733783006668091, "learning_rate": 3.6931699030241653e-06, "loss": 0.233, "step": 38750 }, { "epoch": 0.7049964531912185, "grad_norm": 1.1030446290969849, "learning_rate": 3.6912452522832747e-06, "loss": 0.2271, "step": 38760 }, { "epoch": 0.7051783408210408, "grad_norm": 1.0550490617752075, "learning_rate": 3.6893214027603064e-06, "loss": 0.2259, "step": 38770 }, { "epoch": 0.7053602284508631, "grad_norm": 0.9828023910522461, "learning_rate": 3.6873983551229e-06, "loss": 0.2351, "step": 38780 }, { "epoch": 0.7055421160806854, "grad_norm": 1.1142655611038208, "learning_rate": 3.6854761100384206e-06, "loss": 0.2282, "step": 38790 }, { "epoch": 0.7057240037105077, "grad_norm": 1.0222651958465576, "learning_rate": 3.683554668173952e-06, "loss": 0.2228, "step": 38800 }, { "epoch": 0.70590589134033, "grad_norm": 1.0303343534469604, "learning_rate": 3.681634030196305e-06, "loss": 0.2231, "step": 38810 }, { "epoch": 0.7060877789701523, "grad_norm": 1.0257666110992432, "learning_rate": 3.6797141967720052e-06, "loss": 0.2258, "step": 38820 }, { "epoch": 0.7062696665999746, "grad_norm": 1.016453742980957, "learning_rate": 3.6777951685673008e-06, "loss": 0.2322, "step": 38830 }, { "epoch": 0.7064515542297969, "grad_norm": 1.066719651222229, "learning_rate": 3.6758769462481614e-06, "loss": 0.2304, "step": 38840 }, { "epoch": 0.7066334418596192, "grad_norm": 0.9133002758026123, "learning_rate": 3.673959530480277e-06, "loss": 0.2253, "step": 38850 }, { "epoch": 0.7068153294894414, "grad_norm": 0.9237119555473328, "learning_rate": 3.6720429219290553e-06, "loss": 0.2278, "step": 38860 }, { "epoch": 0.7069972171192637, "grad_norm": 0.916232705116272, "learning_rate": 3.670127121259631e-06, "loss": 0.2219, "step": 38870 }, { "epoch": 0.707179104749086, "grad_norm": 1.2842185497283936, "learning_rate": 3.66821212913685e-06, "loss": 0.2223, "step": 38880 }, { "epoch": 0.7073609923789083, "grad_norm": 1.0425487756729126, "learning_rate": 3.666297946225281e-06, "loss": 0.231, "step": 38890 }, { "epoch": 0.7075428800087306, "grad_norm": 0.9884723424911499, "learning_rate": 3.6643845731892118e-06, "loss": 0.2218, "step": 38900 }, { "epoch": 0.7077247676385529, "grad_norm": 1.0641332864761353, "learning_rate": 3.662472010692648e-06, "loss": 0.2226, "step": 38910 }, { "epoch": 0.7079066552683752, "grad_norm": 0.9510948657989502, "learning_rate": 3.6605602593993185e-06, "loss": 0.2247, "step": 38920 }, { "epoch": 0.7080885428981974, "grad_norm": 0.9665051698684692, "learning_rate": 3.6586493199726647e-06, "loss": 0.2213, "step": 38930 }, { "epoch": 0.7082704305280197, "grad_norm": 0.9703660607337952, "learning_rate": 3.656739193075849e-06, "loss": 0.2241, "step": 38940 }, { "epoch": 0.708452318157842, "grad_norm": 1.1974529027938843, "learning_rate": 3.654829879371753e-06, "loss": 0.229, "step": 38950 }, { "epoch": 0.7086342057876643, "grad_norm": 1.0499862432479858, "learning_rate": 3.6529213795229712e-06, "loss": 0.2274, "step": 38960 }, { "epoch": 0.7088160934174866, "grad_norm": 0.9518017768859863, "learning_rate": 3.651013694191824e-06, "loss": 0.2279, "step": 38970 }, { "epoch": 0.7089979810473089, "grad_norm": 1.0371450185775757, "learning_rate": 3.6491068240403418e-06, "loss": 0.2261, "step": 38980 }, { "epoch": 0.7091798686771312, "grad_norm": 1.0285859107971191, "learning_rate": 3.6472007697302743e-06, "loss": 0.2228, "step": 38990 }, { "epoch": 0.7093617563069535, "grad_norm": 1.0165804624557495, "learning_rate": 3.6452955319230877e-06, "loss": 0.2263, "step": 39000 }, { "epoch": 0.7095436439367758, "grad_norm": 0.9665330052375793, "learning_rate": 3.643391111279968e-06, "loss": 0.227, "step": 39010 }, { "epoch": 0.7097255315665981, "grad_norm": 0.9141767621040344, "learning_rate": 3.6414875084618153e-06, "loss": 0.228, "step": 39020 }, { "epoch": 0.7099074191964204, "grad_norm": 1.084652304649353, "learning_rate": 3.6395847241292436e-06, "loss": 0.2267, "step": 39030 }, { "epoch": 0.7100893068262427, "grad_norm": 0.8555041551589966, "learning_rate": 3.6376827589425857e-06, "loss": 0.2256, "step": 39040 }, { "epoch": 0.710271194456065, "grad_norm": 0.9540870785713196, "learning_rate": 3.6357816135618883e-06, "loss": 0.2309, "step": 39050 }, { "epoch": 0.7104530820858873, "grad_norm": 1.0102461576461792, "learning_rate": 3.6338812886469175e-06, "loss": 0.2226, "step": 39060 }, { "epoch": 0.7106349697157096, "grad_norm": 1.084857702255249, "learning_rate": 3.6319817848571504e-06, "loss": 0.2314, "step": 39070 }, { "epoch": 0.7108168573455319, "grad_norm": 1.0021203756332397, "learning_rate": 3.6300831028517815e-06, "loss": 0.227, "step": 39080 }, { "epoch": 0.7109987449753542, "grad_norm": 1.0530072450637817, "learning_rate": 3.628185243289718e-06, "loss": 0.2265, "step": 39090 }, { "epoch": 0.7111806326051765, "grad_norm": 1.0326077938079834, "learning_rate": 3.6262882068295816e-06, "loss": 0.2224, "step": 39100 }, { "epoch": 0.7113625202349988, "grad_norm": 0.9897919297218323, "learning_rate": 3.6243919941297133e-06, "loss": 0.2214, "step": 39110 }, { "epoch": 0.7115444078648211, "grad_norm": 0.9566752910614014, "learning_rate": 3.6224966058481624e-06, "loss": 0.2242, "step": 39120 }, { "epoch": 0.7117262954946434, "grad_norm": 0.8931500315666199, "learning_rate": 3.6206020426426957e-06, "loss": 0.2283, "step": 39130 }, { "epoch": 0.7119081831244657, "grad_norm": 1.0459926128387451, "learning_rate": 3.6187083051707895e-06, "loss": 0.2245, "step": 39140 }, { "epoch": 0.712090070754288, "grad_norm": 1.0193703174591064, "learning_rate": 3.61681539408964e-06, "loss": 0.2301, "step": 39150 }, { "epoch": 0.7122719583841103, "grad_norm": 0.9379159212112427, "learning_rate": 3.6149233100561498e-06, "loss": 0.2266, "step": 39160 }, { "epoch": 0.7124538460139326, "grad_norm": 0.9580371379852295, "learning_rate": 3.613032053726939e-06, "loss": 0.2286, "step": 39170 }, { "epoch": 0.7126357336437549, "grad_norm": 1.1168186664581299, "learning_rate": 3.611141625758341e-06, "loss": 0.2274, "step": 39180 }, { "epoch": 0.7128176212735772, "grad_norm": 1.0658667087554932, "learning_rate": 3.6092520268063975e-06, "loss": 0.2234, "step": 39190 }, { "epoch": 0.7129995089033995, "grad_norm": 1.099603295326233, "learning_rate": 3.6073632575268634e-06, "loss": 0.2278, "step": 39200 }, { "epoch": 0.7131813965332218, "grad_norm": 0.9173485040664673, "learning_rate": 3.6054753185752104e-06, "loss": 0.2286, "step": 39210 }, { "epoch": 0.7133632841630441, "grad_norm": 0.9214757680892944, "learning_rate": 3.603588210606618e-06, "loss": 0.2273, "step": 39220 }, { "epoch": 0.7135451717928664, "grad_norm": 1.0672017335891724, "learning_rate": 3.6017019342759775e-06, "loss": 0.2263, "step": 39230 }, { "epoch": 0.7137270594226887, "grad_norm": 1.0424164533615112, "learning_rate": 3.5998164902378906e-06, "loss": 0.2246, "step": 39240 }, { "epoch": 0.713908947052511, "grad_norm": 1.0937645435333252, "learning_rate": 3.597931879146673e-06, "loss": 0.225, "step": 39250 }, { "epoch": 0.7140908346823333, "grad_norm": 1.0006009340286255, "learning_rate": 3.596048101656352e-06, "loss": 0.2285, "step": 39260 }, { "epoch": 0.7142727223121556, "grad_norm": 1.0704447031021118, "learning_rate": 3.5941651584206614e-06, "loss": 0.2211, "step": 39270 }, { "epoch": 0.7144546099419778, "grad_norm": 1.1988385915756226, "learning_rate": 3.5922830500930483e-06, "loss": 0.2235, "step": 39280 }, { "epoch": 0.7146364975718001, "grad_norm": 1.082528829574585, "learning_rate": 3.5904017773266696e-06, "loss": 0.2218, "step": 39290 }, { "epoch": 0.7148183852016224, "grad_norm": 0.9418493509292603, "learning_rate": 3.5885213407743913e-06, "loss": 0.2229, "step": 39300 }, { "epoch": 0.7150002728314447, "grad_norm": 1.0742307901382446, "learning_rate": 3.5866417410887917e-06, "loss": 0.2266, "step": 39310 }, { "epoch": 0.715182160461267, "grad_norm": 0.8867419958114624, "learning_rate": 3.584762978922156e-06, "loss": 0.2261, "step": 39320 }, { "epoch": 0.7153640480910893, "grad_norm": 1.017149806022644, "learning_rate": 3.582885054926481e-06, "loss": 0.2261, "step": 39330 }, { "epoch": 0.7155459357209116, "grad_norm": 0.9892767667770386, "learning_rate": 3.5810079697534694e-06, "loss": 0.2245, "step": 39340 }, { "epoch": 0.7157278233507339, "grad_norm": 1.048552393913269, "learning_rate": 3.5791317240545354e-06, "loss": 0.226, "step": 39350 }, { "epoch": 0.7159097109805562, "grad_norm": 1.0442287921905518, "learning_rate": 3.5772563184808028e-06, "loss": 0.2228, "step": 39360 }, { "epoch": 0.7160915986103785, "grad_norm": 0.975202739238739, "learning_rate": 3.5753817536831e-06, "loss": 0.2234, "step": 39370 }, { "epoch": 0.7162734862402008, "grad_norm": 1.0365620851516724, "learning_rate": 3.573508030311969e-06, "loss": 0.2322, "step": 39380 }, { "epoch": 0.7164553738700231, "grad_norm": 1.0585733652114868, "learning_rate": 3.5716351490176528e-06, "loss": 0.232, "step": 39390 }, { "epoch": 0.7166372614998454, "grad_norm": 0.9456053376197815, "learning_rate": 3.5697631104501103e-06, "loss": 0.2233, "step": 39400 }, { "epoch": 0.7168191491296677, "grad_norm": 1.0316987037658691, "learning_rate": 3.5678919152589996e-06, "loss": 0.2212, "step": 39410 }, { "epoch": 0.71700103675949, "grad_norm": 0.9606568813323975, "learning_rate": 3.5660215640936947e-06, "loss": 0.2304, "step": 39420 }, { "epoch": 0.7171829243893123, "grad_norm": 0.9590197801589966, "learning_rate": 3.564152057603269e-06, "loss": 0.2257, "step": 39430 }, { "epoch": 0.7173648120191346, "grad_norm": 0.921314001083374, "learning_rate": 3.562283396436508e-06, "loss": 0.2262, "step": 39440 }, { "epoch": 0.7175466996489569, "grad_norm": 1.0513544082641602, "learning_rate": 3.5604155812418985e-06, "loss": 0.2239, "step": 39450 }, { "epoch": 0.7177285872787792, "grad_norm": 0.9950818419456482, "learning_rate": 3.558548612667642e-06, "loss": 0.2307, "step": 39460 }, { "epoch": 0.7179104749086015, "grad_norm": 0.9861069917678833, "learning_rate": 3.556682491361638e-06, "loss": 0.2271, "step": 39470 }, { "epoch": 0.7180923625384238, "grad_norm": 0.9529068470001221, "learning_rate": 3.554817217971497e-06, "loss": 0.2217, "step": 39480 }, { "epoch": 0.7182742501682461, "grad_norm": 0.9429428577423096, "learning_rate": 3.552952793144533e-06, "loss": 0.2233, "step": 39490 }, { "epoch": 0.7184561377980684, "grad_norm": 1.051757574081421, "learning_rate": 3.5510892175277633e-06, "loss": 0.2236, "step": 39500 }, { "epoch": 0.7186380254278907, "grad_norm": 0.9484663605690002, "learning_rate": 3.5492264917679183e-06, "loss": 0.2281, "step": 39510 }, { "epoch": 0.718819913057713, "grad_norm": 1.085644245147705, "learning_rate": 3.547364616511425e-06, "loss": 0.2271, "step": 39520 }, { "epoch": 0.7190018006875353, "grad_norm": 0.9762524962425232, "learning_rate": 3.5455035924044194e-06, "loss": 0.2244, "step": 39530 }, { "epoch": 0.7191836883173576, "grad_norm": 0.9102628231048584, "learning_rate": 3.543643420092742e-06, "loss": 0.2251, "step": 39540 }, { "epoch": 0.7193655759471799, "grad_norm": 0.9606788158416748, "learning_rate": 3.5417841002219337e-06, "loss": 0.2222, "step": 39550 }, { "epoch": 0.7195474635770022, "grad_norm": 0.9825555682182312, "learning_rate": 3.5399256334372485e-06, "loss": 0.2223, "step": 39560 }, { "epoch": 0.7197293512068245, "grad_norm": 1.0059466361999512, "learning_rate": 3.5380680203836353e-06, "loss": 0.2274, "step": 39570 }, { "epoch": 0.7199112388366468, "grad_norm": 1.0548558235168457, "learning_rate": 3.536211261705751e-06, "loss": 0.2287, "step": 39580 }, { "epoch": 0.7200931264664691, "grad_norm": 0.9666028022766113, "learning_rate": 3.534355358047954e-06, "loss": 0.2225, "step": 39590 }, { "epoch": 0.7202750140962914, "grad_norm": 1.0764973163604736, "learning_rate": 3.5325003100543093e-06, "loss": 0.2286, "step": 39600 }, { "epoch": 0.7204569017261137, "grad_norm": 1.023422360420227, "learning_rate": 3.5306461183685812e-06, "loss": 0.2295, "step": 39610 }, { "epoch": 0.720638789355936, "grad_norm": 1.0408382415771484, "learning_rate": 3.5287927836342407e-06, "loss": 0.2275, "step": 39620 }, { "epoch": 0.7208206769857582, "grad_norm": 1.0338811874389648, "learning_rate": 3.5269403064944567e-06, "loss": 0.2265, "step": 39630 }, { "epoch": 0.7210025646155805, "grad_norm": 1.0825419425964355, "learning_rate": 3.525088687592103e-06, "loss": 0.2249, "step": 39640 }, { "epoch": 0.7211844522454028, "grad_norm": 1.0145151615142822, "learning_rate": 3.5232379275697576e-06, "loss": 0.2329, "step": 39650 }, { "epoch": 0.7213663398752251, "grad_norm": 1.055267572402954, "learning_rate": 3.5213880270696974e-06, "loss": 0.233, "step": 39660 }, { "epoch": 0.7215482275050474, "grad_norm": 0.9095268249511719, "learning_rate": 3.5195389867339012e-06, "loss": 0.2261, "step": 39670 }, { "epoch": 0.7217301151348697, "grad_norm": 1.0775773525238037, "learning_rate": 3.5176908072040515e-06, "loss": 0.221, "step": 39680 }, { "epoch": 0.7219120027646919, "grad_norm": 1.0039910078048706, "learning_rate": 3.5158434891215276e-06, "loss": 0.2277, "step": 39690 }, { "epoch": 0.7220938903945142, "grad_norm": 0.9566814303398132, "learning_rate": 3.513997033127416e-06, "loss": 0.2208, "step": 39700 }, { "epoch": 0.7222757780243365, "grad_norm": 1.0139429569244385, "learning_rate": 3.512151439862501e-06, "loss": 0.2236, "step": 39710 }, { "epoch": 0.7224576656541588, "grad_norm": 1.0819032192230225, "learning_rate": 3.5103067099672673e-06, "loss": 0.2264, "step": 39720 }, { "epoch": 0.7226395532839811, "grad_norm": 1.0994322299957275, "learning_rate": 3.5084628440818978e-06, "loss": 0.2383, "step": 39730 }, { "epoch": 0.7228214409138034, "grad_norm": 1.090857744216919, "learning_rate": 3.5066198428462793e-06, "loss": 0.2259, "step": 39740 }, { "epoch": 0.7230033285436257, "grad_norm": 0.9654910564422607, "learning_rate": 3.5047777068999976e-06, "loss": 0.2261, "step": 39750 }, { "epoch": 0.723185216173448, "grad_norm": 1.0838216543197632, "learning_rate": 3.502936436882338e-06, "loss": 0.2263, "step": 39760 }, { "epoch": 0.7233671038032703, "grad_norm": 0.9965794682502747, "learning_rate": 3.5010960334322847e-06, "loss": 0.2246, "step": 39770 }, { "epoch": 0.7235489914330926, "grad_norm": 1.0013644695281982, "learning_rate": 3.4992564971885205e-06, "loss": 0.2276, "step": 39780 }, { "epoch": 0.7237308790629149, "grad_norm": 1.034096598625183, "learning_rate": 3.497417828789428e-06, "loss": 0.2285, "step": 39790 }, { "epoch": 0.7239127666927372, "grad_norm": 1.1122640371322632, "learning_rate": 3.495580028873091e-06, "loss": 0.2267, "step": 39800 }, { "epoch": 0.7240946543225595, "grad_norm": 0.9045590162277222, "learning_rate": 3.4937430980772883e-06, "loss": 0.2269, "step": 39810 }, { "epoch": 0.7242765419523818, "grad_norm": 1.0341230630874634, "learning_rate": 3.4919070370394992e-06, "loss": 0.2261, "step": 39820 }, { "epoch": 0.7244584295822041, "grad_norm": 1.1251468658447266, "learning_rate": 3.4900718463968997e-06, "loss": 0.2255, "step": 39830 }, { "epoch": 0.7246403172120264, "grad_norm": 1.099722981452942, "learning_rate": 3.4882375267863632e-06, "loss": 0.23, "step": 39840 }, { "epoch": 0.7248222048418487, "grad_norm": 1.0051381587982178, "learning_rate": 3.4864040788444664e-06, "loss": 0.2252, "step": 39850 }, { "epoch": 0.725004092471671, "grad_norm": 1.0344635248184204, "learning_rate": 3.4845715032074758e-06, "loss": 0.2238, "step": 39860 }, { "epoch": 0.7251859801014933, "grad_norm": 1.0432615280151367, "learning_rate": 3.48273980051136e-06, "loss": 0.2235, "step": 39870 }, { "epoch": 0.7253678677313156, "grad_norm": 1.0335874557495117, "learning_rate": 3.4809089713917833e-06, "loss": 0.2235, "step": 39880 }, { "epoch": 0.7255497553611379, "grad_norm": 1.1280962228775024, "learning_rate": 3.4790790164841047e-06, "loss": 0.2219, "step": 39890 }, { "epoch": 0.7257316429909602, "grad_norm": 0.976334810256958, "learning_rate": 3.4772499364233854e-06, "loss": 0.2217, "step": 39900 }, { "epoch": 0.7259135306207825, "grad_norm": 1.048184871673584, "learning_rate": 3.4754217318443783e-06, "loss": 0.226, "step": 39910 }, { "epoch": 0.7260954182506048, "grad_norm": 0.9700600504875183, "learning_rate": 3.4735944033815327e-06, "loss": 0.2242, "step": 39920 }, { "epoch": 0.726277305880427, "grad_norm": 0.9178264141082764, "learning_rate": 3.471767951668995e-06, "loss": 0.2196, "step": 39930 }, { "epoch": 0.7264591935102493, "grad_norm": 1.1073514223098755, "learning_rate": 3.469942377340609e-06, "loss": 0.225, "step": 39940 }, { "epoch": 0.7266410811400716, "grad_norm": 1.0976300239562988, "learning_rate": 3.4681176810299112e-06, "loss": 0.2266, "step": 39950 }, { "epoch": 0.7268229687698939, "grad_norm": 1.0103930234909058, "learning_rate": 3.4662938633701335e-06, "loss": 0.2247, "step": 39960 }, { "epoch": 0.7270048563997162, "grad_norm": 1.1469089984893799, "learning_rate": 3.4644709249942054e-06, "loss": 0.2209, "step": 39970 }, { "epoch": 0.7271867440295385, "grad_norm": 0.9643105864524841, "learning_rate": 3.4626488665347502e-06, "loss": 0.2264, "step": 39980 }, { "epoch": 0.7273686316593608, "grad_norm": 1.0131168365478516, "learning_rate": 3.460827688624082e-06, "loss": 0.2249, "step": 39990 }, { "epoch": 0.7275505192891831, "grad_norm": 1.0567377805709839, "learning_rate": 3.459007391894217e-06, "loss": 0.2245, "step": 40000 }, { "epoch": 0.7277324069190054, "grad_norm": 1.0117990970611572, "learning_rate": 3.457187976976859e-06, "loss": 0.2233, "step": 40010 }, { "epoch": 0.7279142945488277, "grad_norm": 0.9737592339515686, "learning_rate": 3.455369444503408e-06, "loss": 0.2271, "step": 40020 }, { "epoch": 0.72809618217865, "grad_norm": 1.1766866445541382, "learning_rate": 3.453551795104958e-06, "loss": 0.228, "step": 40030 }, { "epoch": 0.7282780698084723, "grad_norm": 0.9524601101875305, "learning_rate": 3.451735029412295e-06, "loss": 0.2284, "step": 40040 }, { "epoch": 0.7284599574382946, "grad_norm": 1.0175305604934692, "learning_rate": 3.4499191480559023e-06, "loss": 0.2247, "step": 40050 }, { "epoch": 0.7286418450681169, "grad_norm": 1.0697991847991943, "learning_rate": 3.448104151665952e-06, "loss": 0.2267, "step": 40060 }, { "epoch": 0.7288237326979392, "grad_norm": 0.9677073359489441, "learning_rate": 3.446290040872312e-06, "loss": 0.2242, "step": 40070 }, { "epoch": 0.7290056203277615, "grad_norm": 1.0248783826828003, "learning_rate": 3.4444768163045405e-06, "loss": 0.2234, "step": 40080 }, { "epoch": 0.7291875079575838, "grad_norm": 0.9701741933822632, "learning_rate": 3.4426644785918878e-06, "loss": 0.2296, "step": 40090 }, { "epoch": 0.7293693955874061, "grad_norm": 0.9851295948028564, "learning_rate": 3.440853028363301e-06, "loss": 0.2262, "step": 40100 }, { "epoch": 0.7295512832172284, "grad_norm": 1.1325019598007202, "learning_rate": 3.4390424662474155e-06, "loss": 0.2275, "step": 40110 }, { "epoch": 0.7297331708470507, "grad_norm": 0.9892047047615051, "learning_rate": 3.4372327928725573e-06, "loss": 0.2202, "step": 40120 }, { "epoch": 0.729915058476873, "grad_norm": 0.9734076261520386, "learning_rate": 3.435424008866747e-06, "loss": 0.2229, "step": 40130 }, { "epoch": 0.7300969461066953, "grad_norm": 1.0852165222167969, "learning_rate": 3.4336161148576952e-06, "loss": 0.2289, "step": 40140 }, { "epoch": 0.7302788337365176, "grad_norm": 0.8698779344558716, "learning_rate": 3.4318091114728037e-06, "loss": 0.2238, "step": 40150 }, { "epoch": 0.7304607213663399, "grad_norm": 0.9165798425674438, "learning_rate": 3.4300029993391657e-06, "loss": 0.2272, "step": 40160 }, { "epoch": 0.7306426089961622, "grad_norm": 1.0310250520706177, "learning_rate": 3.428197779083565e-06, "loss": 0.2306, "step": 40170 }, { "epoch": 0.7308244966259845, "grad_norm": 1.0101385116577148, "learning_rate": 3.4263934513324736e-06, "loss": 0.2272, "step": 40180 }, { "epoch": 0.7310063842558068, "grad_norm": 0.9616357684135437, "learning_rate": 3.4245900167120582e-06, "loss": 0.226, "step": 40190 }, { "epoch": 0.7311882718856291, "grad_norm": 1.0379709005355835, "learning_rate": 3.4227874758481714e-06, "loss": 0.2293, "step": 40200 }, { "epoch": 0.7313701595154514, "grad_norm": 0.9670861959457397, "learning_rate": 3.420985829366359e-06, "loss": 0.2243, "step": 40210 }, { "epoch": 0.7315520471452737, "grad_norm": 1.060729742050171, "learning_rate": 3.4191850778918535e-06, "loss": 0.229, "step": 40220 }, { "epoch": 0.731733934775096, "grad_norm": 0.9955342411994934, "learning_rate": 3.417385222049576e-06, "loss": 0.2272, "step": 40230 }, { "epoch": 0.7319158224049183, "grad_norm": 1.0512956380844116, "learning_rate": 3.415586262464143e-06, "loss": 0.2235, "step": 40240 }, { "epoch": 0.7320977100347406, "grad_norm": 1.181087851524353, "learning_rate": 3.4137881997598545e-06, "loss": 0.2261, "step": 40250 }, { "epoch": 0.7322795976645629, "grad_norm": 0.97022545337677, "learning_rate": 3.4119910345607e-06, "loss": 0.2221, "step": 40260 }, { "epoch": 0.7324614852943852, "grad_norm": 1.0055204629898071, "learning_rate": 3.410194767490359e-06, "loss": 0.2226, "step": 40270 }, { "epoch": 0.7326433729242074, "grad_norm": 1.1231684684753418, "learning_rate": 3.4083993991721974e-06, "loss": 0.2251, "step": 40280 }, { "epoch": 0.7328252605540297, "grad_norm": 0.9133597016334534, "learning_rate": 3.4066049302292694e-06, "loss": 0.2291, "step": 40290 }, { "epoch": 0.733007148183852, "grad_norm": 1.1526724100112915, "learning_rate": 3.40481136128432e-06, "loss": 0.2225, "step": 40300 }, { "epoch": 0.7331890358136743, "grad_norm": 1.1000378131866455, "learning_rate": 3.4030186929597784e-06, "loss": 0.2269, "step": 40310 }, { "epoch": 0.7333709234434966, "grad_norm": 1.0105363130569458, "learning_rate": 3.401226925877764e-06, "loss": 0.2186, "step": 40320 }, { "epoch": 0.7335528110733189, "grad_norm": 0.8574592471122742, "learning_rate": 3.3994360606600796e-06, "loss": 0.2244, "step": 40330 }, { "epoch": 0.7337346987031412, "grad_norm": 1.203989028930664, "learning_rate": 3.39764609792822e-06, "loss": 0.2242, "step": 40340 }, { "epoch": 0.7339165863329635, "grad_norm": 0.991537868976593, "learning_rate": 3.395857038303364e-06, "loss": 0.2185, "step": 40350 }, { "epoch": 0.7340984739627858, "grad_norm": 0.9802651405334473, "learning_rate": 3.394068882406376e-06, "loss": 0.2246, "step": 40360 }, { "epoch": 0.7342803615926081, "grad_norm": 0.948989748954773, "learning_rate": 3.39228163085781e-06, "loss": 0.2247, "step": 40370 }, { "epoch": 0.7344622492224304, "grad_norm": 0.9466442465782166, "learning_rate": 3.3904952842779003e-06, "loss": 0.2245, "step": 40380 }, { "epoch": 0.7346441368522527, "grad_norm": 0.8759536147117615, "learning_rate": 3.3887098432865763e-06, "loss": 0.2223, "step": 40390 }, { "epoch": 0.734826024482075, "grad_norm": 1.1500260829925537, "learning_rate": 3.3869253085034447e-06, "loss": 0.2233, "step": 40400 }, { "epoch": 0.7350079121118973, "grad_norm": 1.0405325889587402, "learning_rate": 3.385141680547801e-06, "loss": 0.2245, "step": 40410 }, { "epoch": 0.7351897997417196, "grad_norm": 1.11989164352417, "learning_rate": 3.3833589600386275e-06, "loss": 0.2246, "step": 40420 }, { "epoch": 0.7353716873715419, "grad_norm": 0.9173451066017151, "learning_rate": 3.3815771475945876e-06, "loss": 0.2236, "step": 40430 }, { "epoch": 0.7355535750013642, "grad_norm": 0.9932881593704224, "learning_rate": 3.3797962438340346e-06, "loss": 0.2247, "step": 40440 }, { "epoch": 0.7357354626311865, "grad_norm": 1.122541069984436, "learning_rate": 3.3780162493750023e-06, "loss": 0.2292, "step": 40450 }, { "epoch": 0.7359173502610088, "grad_norm": 0.9191031455993652, "learning_rate": 3.3762371648352114e-06, "loss": 0.2187, "step": 40460 }, { "epoch": 0.7360992378908311, "grad_norm": 1.136909008026123, "learning_rate": 3.374458990832065e-06, "loss": 0.2293, "step": 40470 }, { "epoch": 0.7362811255206534, "grad_norm": 0.9493772983551025, "learning_rate": 3.3726817279826508e-06, "loss": 0.2231, "step": 40480 }, { "epoch": 0.7364630131504757, "grad_norm": 1.1747050285339355, "learning_rate": 3.3709053769037416e-06, "loss": 0.224, "step": 40490 }, { "epoch": 0.736644900780298, "grad_norm": 1.1267133951187134, "learning_rate": 3.3691299382117926e-06, "loss": 0.2276, "step": 40500 }, { "epoch": 0.7368267884101203, "grad_norm": 1.0414090156555176, "learning_rate": 3.3673554125229422e-06, "loss": 0.2306, "step": 40510 }, { "epoch": 0.7370086760399426, "grad_norm": 1.0547276735305786, "learning_rate": 3.3655818004530115e-06, "loss": 0.2264, "step": 40520 }, { "epoch": 0.7371905636697648, "grad_norm": 0.8940314054489136, "learning_rate": 3.363809102617507e-06, "loss": 0.2315, "step": 40530 }, { "epoch": 0.7373724512995871, "grad_norm": 0.953575074672699, "learning_rate": 3.3620373196316157e-06, "loss": 0.2285, "step": 40540 }, { "epoch": 0.7375543389294094, "grad_norm": 0.9605063796043396, "learning_rate": 3.3602664521102093e-06, "loss": 0.2232, "step": 40550 }, { "epoch": 0.7377362265592317, "grad_norm": 1.0013294219970703, "learning_rate": 3.3584965006678373e-06, "loss": 0.227, "step": 40560 }, { "epoch": 0.737918114189054, "grad_norm": 1.0066992044448853, "learning_rate": 3.356727465918736e-06, "loss": 0.2253, "step": 40570 }, { "epoch": 0.7381000018188763, "grad_norm": 1.2448369264602661, "learning_rate": 3.3549593484768194e-06, "loss": 0.2268, "step": 40580 }, { "epoch": 0.7382818894486985, "grad_norm": 1.0688987970352173, "learning_rate": 3.3531921489556897e-06, "loss": 0.2281, "step": 40590 }, { "epoch": 0.7384637770785208, "grad_norm": 1.131683111190796, "learning_rate": 3.351425867968623e-06, "loss": 0.2327, "step": 40600 }, { "epoch": 0.7386456647083431, "grad_norm": 1.138440728187561, "learning_rate": 3.349660506128582e-06, "loss": 0.2247, "step": 40610 }, { "epoch": 0.7388275523381654, "grad_norm": 1.0320110321044922, "learning_rate": 3.347896064048207e-06, "loss": 0.2216, "step": 40620 }, { "epoch": 0.7390094399679877, "grad_norm": 1.1043646335601807, "learning_rate": 3.3461325423398196e-06, "loss": 0.2209, "step": 40630 }, { "epoch": 0.73919132759781, "grad_norm": 1.0351148843765259, "learning_rate": 3.3443699416154264e-06, "loss": 0.2275, "step": 40640 }, { "epoch": 0.7393732152276323, "grad_norm": 0.983749508857727, "learning_rate": 3.3426082624867086e-06, "loss": 0.2249, "step": 40650 }, { "epoch": 0.7395551028574546, "grad_norm": 0.8857402801513672, "learning_rate": 3.3408475055650304e-06, "loss": 0.2257, "step": 40660 }, { "epoch": 0.7397369904872769, "grad_norm": 1.0518083572387695, "learning_rate": 3.339087671461436e-06, "loss": 0.227, "step": 40670 }, { "epoch": 0.7399188781170992, "grad_norm": 1.095323920249939, "learning_rate": 3.337328760786646e-06, "loss": 0.2325, "step": 40680 }, { "epoch": 0.7401007657469215, "grad_norm": 1.0118931531906128, "learning_rate": 3.335570774151068e-06, "loss": 0.221, "step": 40690 }, { "epoch": 0.7402826533767438, "grad_norm": 1.0446853637695312, "learning_rate": 3.333813712164781e-06, "loss": 0.227, "step": 40700 }, { "epoch": 0.7404645410065661, "grad_norm": 1.0248676538467407, "learning_rate": 3.3320575754375473e-06, "loss": 0.2243, "step": 40710 }, { "epoch": 0.7406464286363884, "grad_norm": 1.075701117515564, "learning_rate": 3.330302364578807e-06, "loss": 0.2296, "step": 40720 }, { "epoch": 0.7408283162662107, "grad_norm": 0.9572486281394958, "learning_rate": 3.328548080197679e-06, "loss": 0.2229, "step": 40730 }, { "epoch": 0.741010203896033, "grad_norm": 1.025866150856018, "learning_rate": 3.326794722902962e-06, "loss": 0.2249, "step": 40740 }, { "epoch": 0.7411920915258553, "grad_norm": 0.99025559425354, "learning_rate": 3.3250422933031314e-06, "loss": 0.2298, "step": 40750 }, { "epoch": 0.7413739791556776, "grad_norm": 1.0298738479614258, "learning_rate": 3.3232907920063397e-06, "loss": 0.2241, "step": 40760 }, { "epoch": 0.7415558667854999, "grad_norm": 1.0335428714752197, "learning_rate": 3.3215402196204184e-06, "loss": 0.2206, "step": 40770 }, { "epoch": 0.7417377544153222, "grad_norm": 1.0743790864944458, "learning_rate": 3.319790576752878e-06, "loss": 0.226, "step": 40780 }, { "epoch": 0.7419196420451445, "grad_norm": 0.9552496075630188, "learning_rate": 3.318041864010905e-06, "loss": 0.2281, "step": 40790 }, { "epoch": 0.7421015296749668, "grad_norm": 1.0064516067504883, "learning_rate": 3.3162940820013644e-06, "loss": 0.2283, "step": 40800 }, { "epoch": 0.7422834173047891, "grad_norm": 1.0188826322555542, "learning_rate": 3.3145472313307957e-06, "loss": 0.2284, "step": 40810 }, { "epoch": 0.7424653049346114, "grad_norm": 1.0296093225479126, "learning_rate": 3.312801312605417e-06, "loss": 0.2253, "step": 40820 }, { "epoch": 0.7426471925644337, "grad_norm": 1.0803273916244507, "learning_rate": 3.3110563264311213e-06, "loss": 0.2226, "step": 40830 }, { "epoch": 0.742829080194256, "grad_norm": 1.1112000942230225, "learning_rate": 3.309312273413482e-06, "loss": 0.223, "step": 40840 }, { "epoch": 0.7430109678240783, "grad_norm": 1.0394047498703003, "learning_rate": 3.307569154157744e-06, "loss": 0.2252, "step": 40850 }, { "epoch": 0.7431928554539006, "grad_norm": 1.0285221338272095, "learning_rate": 3.3058269692688305e-06, "loss": 0.2212, "step": 40860 }, { "epoch": 0.7433747430837229, "grad_norm": 0.8995693325996399, "learning_rate": 3.304085719351341e-06, "loss": 0.2199, "step": 40870 }, { "epoch": 0.7435566307135452, "grad_norm": 1.0177698135375977, "learning_rate": 3.3023454050095473e-06, "loss": 0.2262, "step": 40880 }, { "epoch": 0.7437385183433675, "grad_norm": 1.0085077285766602, "learning_rate": 3.3006060268474004e-06, "loss": 0.2232, "step": 40890 }, { "epoch": 0.7439204059731898, "grad_norm": 1.0370198488235474, "learning_rate": 3.298867585468525e-06, "loss": 0.2257, "step": 40900 }, { "epoch": 0.744102293603012, "grad_norm": 0.9766439199447632, "learning_rate": 3.297130081476221e-06, "loss": 0.2216, "step": 40910 }, { "epoch": 0.7442841812328344, "grad_norm": 0.9190120100975037, "learning_rate": 3.2953935154734594e-06, "loss": 0.2193, "step": 40920 }, { "epoch": 0.7444660688626566, "grad_norm": 0.946776807308197, "learning_rate": 3.2936578880628935e-06, "loss": 0.2244, "step": 40930 }, { "epoch": 0.744647956492479, "grad_norm": 0.9635629057884216, "learning_rate": 3.2919231998468425e-06, "loss": 0.2241, "step": 40940 }, { "epoch": 0.7448298441223012, "grad_norm": 0.9305209517478943, "learning_rate": 3.2901894514273055e-06, "loss": 0.2271, "step": 40950 }, { "epoch": 0.7450117317521235, "grad_norm": 1.0482350587844849, "learning_rate": 3.288456643405952e-06, "loss": 0.2231, "step": 40960 }, { "epoch": 0.7451936193819458, "grad_norm": 1.0545371770858765, "learning_rate": 3.2867247763841247e-06, "loss": 0.2222, "step": 40970 }, { "epoch": 0.7453755070117681, "grad_norm": 0.9879759550094604, "learning_rate": 3.2849938509628455e-06, "loss": 0.2252, "step": 40980 }, { "epoch": 0.7455573946415904, "grad_norm": 0.9411875605583191, "learning_rate": 3.283263867742804e-06, "loss": 0.2261, "step": 40990 }, { "epoch": 0.7457392822714127, "grad_norm": 0.9325774908065796, "learning_rate": 3.281534827324363e-06, "loss": 0.2189, "step": 41000 }, { "epoch": 0.745921169901235, "grad_norm": 1.0183863639831543, "learning_rate": 3.279806730307561e-06, "loss": 0.2221, "step": 41010 }, { "epoch": 0.7461030575310573, "grad_norm": 0.9900813102722168, "learning_rate": 3.278079577292105e-06, "loss": 0.2253, "step": 41020 }, { "epoch": 0.7462849451608796, "grad_norm": 1.0002174377441406, "learning_rate": 3.2763533688773808e-06, "loss": 0.2238, "step": 41030 }, { "epoch": 0.7464668327907019, "grad_norm": 0.9560858607292175, "learning_rate": 3.274628105662441e-06, "loss": 0.2247, "step": 41040 }, { "epoch": 0.7466487204205242, "grad_norm": 1.0141634941101074, "learning_rate": 3.2729037882460108e-06, "loss": 0.2279, "step": 41050 }, { "epoch": 0.7468306080503465, "grad_norm": 1.1285710334777832, "learning_rate": 3.2711804172264887e-06, "loss": 0.22, "step": 41060 }, { "epoch": 0.7470124956801688, "grad_norm": 0.9310078620910645, "learning_rate": 3.2694579932019434e-06, "loss": 0.2211, "step": 41070 }, { "epoch": 0.7471943833099911, "grad_norm": 1.0062452554702759, "learning_rate": 3.267736516770116e-06, "loss": 0.2285, "step": 41080 }, { "epoch": 0.7473762709398134, "grad_norm": 0.9772156476974487, "learning_rate": 3.2660159885284204e-06, "loss": 0.2244, "step": 41090 }, { "epoch": 0.7475581585696357, "grad_norm": 1.1134989261627197, "learning_rate": 3.264296409073937e-06, "loss": 0.2256, "step": 41100 }, { "epoch": 0.747740046199458, "grad_norm": 1.104334831237793, "learning_rate": 3.262577779003422e-06, "loss": 0.2276, "step": 41110 }, { "epoch": 0.7479219338292803, "grad_norm": 1.0890153646469116, "learning_rate": 3.2608600989132966e-06, "loss": 0.2279, "step": 41120 }, { "epoch": 0.7481038214591026, "grad_norm": 0.978481650352478, "learning_rate": 3.259143369399659e-06, "loss": 0.2259, "step": 41130 }, { "epoch": 0.7482857090889249, "grad_norm": 1.0581084489822388, "learning_rate": 3.2574275910582705e-06, "loss": 0.2247, "step": 41140 }, { "epoch": 0.7484675967187472, "grad_norm": 0.9410079121589661, "learning_rate": 3.255712764484568e-06, "loss": 0.2232, "step": 41150 }, { "epoch": 0.7486494843485695, "grad_norm": 1.0087394714355469, "learning_rate": 3.253998890273653e-06, "loss": 0.2243, "step": 41160 }, { "epoch": 0.7488313719783918, "grad_norm": 1.0442885160446167, "learning_rate": 3.252285969020301e-06, "loss": 0.2228, "step": 41170 }, { "epoch": 0.7490132596082141, "grad_norm": 1.0271538496017456, "learning_rate": 3.2505740013189556e-06, "loss": 0.2215, "step": 41180 }, { "epoch": 0.7491951472380364, "grad_norm": 0.9658655524253845, "learning_rate": 3.248862987763729e-06, "loss": 0.2202, "step": 41190 }, { "epoch": 0.7493770348678587, "grad_norm": 1.7764461040496826, "learning_rate": 3.2471529289484007e-06, "loss": 0.2242, "step": 41200 }, { "epoch": 0.749558922497681, "grad_norm": 1.0163350105285645, "learning_rate": 3.2454438254664224e-06, "loss": 0.2235, "step": 41210 }, { "epoch": 0.7497408101275033, "grad_norm": 1.135709524154663, "learning_rate": 3.2437356779109087e-06, "loss": 0.2226, "step": 41220 }, { "epoch": 0.7499226977573256, "grad_norm": 0.9674757122993469, "learning_rate": 3.24202848687465e-06, "loss": 0.227, "step": 41230 }, { "epoch": 0.7501045853871479, "grad_norm": 0.992882251739502, "learning_rate": 3.240322252950099e-06, "loss": 0.2255, "step": 41240 }, { "epoch": 0.7502864730169702, "grad_norm": 0.9489679932594299, "learning_rate": 3.2386169767293783e-06, "loss": 0.2218, "step": 41250 }, { "epoch": 0.7504683606467925, "grad_norm": 1.0670738220214844, "learning_rate": 3.2369126588042775e-06, "loss": 0.2246, "step": 41260 }, { "epoch": 0.7506502482766148, "grad_norm": 1.015752911567688, "learning_rate": 3.235209299766253e-06, "loss": 0.2268, "step": 41270 }, { "epoch": 0.750832135906437, "grad_norm": 0.9713588953018188, "learning_rate": 3.233506900206432e-06, "loss": 0.2251, "step": 41280 }, { "epoch": 0.7510140235362593, "grad_norm": 1.0735505819320679, "learning_rate": 3.231805460715605e-06, "loss": 0.2232, "step": 41290 }, { "epoch": 0.7511959111660816, "grad_norm": 1.0616105794906616, "learning_rate": 3.2301049818842295e-06, "loss": 0.2215, "step": 41300 }, { "epoch": 0.7513777987959039, "grad_norm": 1.0079889297485352, "learning_rate": 3.2284054643024304e-06, "loss": 0.2255, "step": 41310 }, { "epoch": 0.7515596864257262, "grad_norm": 1.0575647354125977, "learning_rate": 3.226706908560001e-06, "loss": 0.2281, "step": 41320 }, { "epoch": 0.7517415740555485, "grad_norm": 1.0231540203094482, "learning_rate": 3.225009315246399e-06, "loss": 0.2255, "step": 41330 }, { "epoch": 0.7519234616853708, "grad_norm": 1.0239845514297485, "learning_rate": 3.2233126849507458e-06, "loss": 0.2336, "step": 41340 }, { "epoch": 0.7521053493151931, "grad_norm": 1.1640222072601318, "learning_rate": 3.221617018261833e-06, "loss": 0.2284, "step": 41350 }, { "epoch": 0.7522872369450153, "grad_norm": 0.9731427431106567, "learning_rate": 3.2199223157681153e-06, "loss": 0.2205, "step": 41360 }, { "epoch": 0.7524691245748376, "grad_norm": 1.0512293577194214, "learning_rate": 3.2182285780577127e-06, "loss": 0.2264, "step": 41370 }, { "epoch": 0.7526510122046599, "grad_norm": 0.9214732646942139, "learning_rate": 3.2165358057184116e-06, "loss": 0.2179, "step": 41380 }, { "epoch": 0.7528328998344822, "grad_norm": 1.0533369779586792, "learning_rate": 3.214843999337662e-06, "loss": 0.2216, "step": 41390 }, { "epoch": 0.7530147874643045, "grad_norm": 1.023456335067749, "learning_rate": 3.21315315950258e-06, "loss": 0.2263, "step": 41400 }, { "epoch": 0.7531966750941268, "grad_norm": 1.0152350664138794, "learning_rate": 3.2114632867999452e-06, "loss": 0.2238, "step": 41410 }, { "epoch": 0.7533785627239491, "grad_norm": 1.012320876121521, "learning_rate": 3.209774381816201e-06, "loss": 0.2257, "step": 41420 }, { "epoch": 0.7535604503537714, "grad_norm": 0.9701526165008545, "learning_rate": 3.2080864451374583e-06, "loss": 0.2266, "step": 41430 }, { "epoch": 0.7537423379835937, "grad_norm": 1.0155868530273438, "learning_rate": 3.206399477349488e-06, "loss": 0.223, "step": 41440 }, { "epoch": 0.753924225613416, "grad_norm": 1.0016207695007324, "learning_rate": 3.2047134790377267e-06, "loss": 0.229, "step": 41450 }, { "epoch": 0.7541061132432383, "grad_norm": 0.9542732238769531, "learning_rate": 3.2030284507872745e-06, "loss": 0.2224, "step": 41460 }, { "epoch": 0.7542880008730606, "grad_norm": 1.007574200630188, "learning_rate": 3.2013443931828924e-06, "loss": 0.2203, "step": 41470 }, { "epoch": 0.7544698885028829, "grad_norm": 0.9397510290145874, "learning_rate": 3.1996613068090114e-06, "loss": 0.2258, "step": 41480 }, { "epoch": 0.7546517761327052, "grad_norm": 0.9571300745010376, "learning_rate": 3.197979192249717e-06, "loss": 0.2234, "step": 41490 }, { "epoch": 0.7548336637625275, "grad_norm": 1.1046873331069946, "learning_rate": 3.1962980500887637e-06, "loss": 0.2242, "step": 41500 }, { "epoch": 0.7550155513923498, "grad_norm": 1.0027753114700317, "learning_rate": 3.1946178809095625e-06, "loss": 0.228, "step": 41510 }, { "epoch": 0.7551974390221721, "grad_norm": 1.0115450620651245, "learning_rate": 3.1929386852951944e-06, "loss": 0.2274, "step": 41520 }, { "epoch": 0.7553793266519944, "grad_norm": 1.0472209453582764, "learning_rate": 3.191260463828396e-06, "loss": 0.2226, "step": 41530 }, { "epoch": 0.7555612142818167, "grad_norm": 1.006067156791687, "learning_rate": 3.1895832170915697e-06, "loss": 0.2225, "step": 41540 }, { "epoch": 0.755743101911639, "grad_norm": 0.9458088874816895, "learning_rate": 3.1879069456667777e-06, "loss": 0.2243, "step": 41550 }, { "epoch": 0.7559249895414613, "grad_norm": 1.011466145515442, "learning_rate": 3.1862316501357424e-06, "loss": 0.2216, "step": 41560 }, { "epoch": 0.7561068771712836, "grad_norm": 1.0207699537277222, "learning_rate": 3.1845573310798535e-06, "loss": 0.2267, "step": 41570 }, { "epoch": 0.7562887648011059, "grad_norm": 1.0730522871017456, "learning_rate": 3.182883989080154e-06, "loss": 0.233, "step": 41580 }, { "epoch": 0.7564706524309281, "grad_norm": 1.0227510929107666, "learning_rate": 3.1812116247173547e-06, "loss": 0.2267, "step": 41590 }, { "epoch": 0.7566525400607504, "grad_norm": 1.0439783334732056, "learning_rate": 3.1795402385718212e-06, "loss": 0.2232, "step": 41600 }, { "epoch": 0.7568344276905727, "grad_norm": 1.0156528949737549, "learning_rate": 3.1778698312235824e-06, "loss": 0.225, "step": 41610 }, { "epoch": 0.757016315320395, "grad_norm": 1.0003060102462769, "learning_rate": 3.176200403252329e-06, "loss": 0.2266, "step": 41620 }, { "epoch": 0.7571982029502173, "grad_norm": 1.0823966264724731, "learning_rate": 3.174531955237411e-06, "loss": 0.2271, "step": 41630 }, { "epoch": 0.7573800905800396, "grad_norm": 1.1248583793640137, "learning_rate": 3.1728644877578376e-06, "loss": 0.232, "step": 41640 }, { "epoch": 0.7575619782098619, "grad_norm": 1.0047192573547363, "learning_rate": 3.1711980013922763e-06, "loss": 0.2259, "step": 41650 }, { "epoch": 0.7577438658396842, "grad_norm": 1.0110406875610352, "learning_rate": 3.169532496719056e-06, "loss": 0.2251, "step": 41660 }, { "epoch": 0.7579257534695065, "grad_norm": 1.09419584274292, "learning_rate": 3.1678679743161634e-06, "loss": 0.2243, "step": 41670 }, { "epoch": 0.7581076410993288, "grad_norm": 1.0086225271224976, "learning_rate": 3.1662044347612485e-06, "loss": 0.218, "step": 41680 }, { "epoch": 0.7582895287291511, "grad_norm": 0.9790023565292358, "learning_rate": 3.1645418786316164e-06, "loss": 0.2283, "step": 41690 }, { "epoch": 0.7584714163589734, "grad_norm": 1.0873905420303345, "learning_rate": 3.16288030650423e-06, "loss": 0.227, "step": 41700 }, { "epoch": 0.7586533039887957, "grad_norm": 0.9719481468200684, "learning_rate": 3.1612197189557124e-06, "loss": 0.2255, "step": 41710 }, { "epoch": 0.758835191618618, "grad_norm": 0.8739980459213257, "learning_rate": 3.1595601165623472e-06, "loss": 0.2178, "step": 41720 }, { "epoch": 0.7590170792484403, "grad_norm": 1.0571880340576172, "learning_rate": 3.1579014999000728e-06, "loss": 0.2221, "step": 41730 }, { "epoch": 0.7591989668782626, "grad_norm": 1.0174086093902588, "learning_rate": 3.1562438695444865e-06, "loss": 0.2254, "step": 41740 }, { "epoch": 0.7593808545080849, "grad_norm": 0.9019321203231812, "learning_rate": 3.1545872260708436e-06, "loss": 0.2134, "step": 41750 }, { "epoch": 0.7595627421379072, "grad_norm": 1.01734459400177, "learning_rate": 3.1529315700540554e-06, "loss": 0.2198, "step": 41760 }, { "epoch": 0.7597446297677295, "grad_norm": 1.0227842330932617, "learning_rate": 3.151276902068695e-06, "loss": 0.2232, "step": 41770 }, { "epoch": 0.7599265173975518, "grad_norm": 0.9270710945129395, "learning_rate": 3.149623222688987e-06, "loss": 0.2237, "step": 41780 }, { "epoch": 0.7601084050273741, "grad_norm": 0.980209231376648, "learning_rate": 3.1479705324888172e-06, "loss": 0.2238, "step": 41790 }, { "epoch": 0.7602902926571964, "grad_norm": 1.1279540061950684, "learning_rate": 3.146318832041725e-06, "loss": 0.2208, "step": 41800 }, { "epoch": 0.7604721802870187, "grad_norm": 1.0877469778060913, "learning_rate": 3.1446681219209065e-06, "loss": 0.2288, "step": 41810 }, { "epoch": 0.760654067916841, "grad_norm": 1.052452564239502, "learning_rate": 3.1430184026992196e-06, "loss": 0.2192, "step": 41820 }, { "epoch": 0.7608359555466633, "grad_norm": 0.9585065841674805, "learning_rate": 3.1413696749491703e-06, "loss": 0.2269, "step": 41830 }, { "epoch": 0.7610178431764856, "grad_norm": 0.9691552519798279, "learning_rate": 3.1397219392429255e-06, "loss": 0.2218, "step": 41840 }, { "epoch": 0.7611997308063079, "grad_norm": 1.0943222045898438, "learning_rate": 3.138075196152307e-06, "loss": 0.2293, "step": 41850 }, { "epoch": 0.7613816184361302, "grad_norm": 1.0376389026641846, "learning_rate": 3.1364294462487897e-06, "loss": 0.2284, "step": 41860 }, { "epoch": 0.7615635060659525, "grad_norm": 0.9097104668617249, "learning_rate": 3.134784690103509e-06, "loss": 0.225, "step": 41870 }, { "epoch": 0.7617453936957748, "grad_norm": 0.9906215667724609, "learning_rate": 3.1331409282872504e-06, "loss": 0.219, "step": 41880 }, { "epoch": 0.7619272813255971, "grad_norm": 0.9990463256835938, "learning_rate": 3.1314981613704554e-06, "loss": 0.2222, "step": 41890 }, { "epoch": 0.7621091689554194, "grad_norm": 0.9972063899040222, "learning_rate": 3.1298563899232235e-06, "loss": 0.223, "step": 41900 }, { "epoch": 0.7622910565852417, "grad_norm": 0.9345612525939941, "learning_rate": 3.1282156145153036e-06, "loss": 0.2192, "step": 41910 }, { "epoch": 0.762472944215064, "grad_norm": 1.1232956647872925, "learning_rate": 3.1265758357161045e-06, "loss": 0.2246, "step": 41920 }, { "epoch": 0.7626548318448862, "grad_norm": 1.0516451597213745, "learning_rate": 3.1249370540946846e-06, "loss": 0.2235, "step": 41930 }, { "epoch": 0.7628367194747085, "grad_norm": 0.9987391233444214, "learning_rate": 3.1232992702197584e-06, "loss": 0.2238, "step": 41940 }, { "epoch": 0.7630186071045308, "grad_norm": 0.9442387819290161, "learning_rate": 3.1216624846596934e-06, "loss": 0.2236, "step": 41950 }, { "epoch": 0.7632004947343531, "grad_norm": 0.9251434803009033, "learning_rate": 3.1200266979825095e-06, "loss": 0.2252, "step": 41960 }, { "epoch": 0.7633823823641754, "grad_norm": 0.8395904302597046, "learning_rate": 3.118391910755884e-06, "loss": 0.2215, "step": 41970 }, { "epoch": 0.7635642699939977, "grad_norm": 1.0057668685913086, "learning_rate": 3.1167581235471432e-06, "loss": 0.2238, "step": 41980 }, { "epoch": 0.76374615762382, "grad_norm": 0.9621633887290955, "learning_rate": 3.1151253369232693e-06, "loss": 0.2253, "step": 41990 }, { "epoch": 0.7639280452536423, "grad_norm": 1.0548299551010132, "learning_rate": 3.113493551450894e-06, "loss": 0.2228, "step": 42000 }, { "epoch": 0.7641099328834646, "grad_norm": 0.9513607621192932, "learning_rate": 3.111862767696303e-06, "loss": 0.2228, "step": 42010 }, { "epoch": 0.7642918205132869, "grad_norm": 1.1862115859985352, "learning_rate": 3.1102329862254377e-06, "loss": 0.2291, "step": 42020 }, { "epoch": 0.7644737081431092, "grad_norm": 1.0723832845687866, "learning_rate": 3.108604207603887e-06, "loss": 0.2277, "step": 42030 }, { "epoch": 0.7646555957729315, "grad_norm": 0.9593611359596252, "learning_rate": 3.106976432396893e-06, "loss": 0.2268, "step": 42040 }, { "epoch": 0.7648374834027538, "grad_norm": 1.023695468902588, "learning_rate": 3.1053496611693513e-06, "loss": 0.2244, "step": 42050 }, { "epoch": 0.7650193710325761, "grad_norm": 1.0112428665161133, "learning_rate": 3.1037238944858057e-06, "loss": 0.2205, "step": 42060 }, { "epoch": 0.7652012586623984, "grad_norm": 0.9875343441963196, "learning_rate": 3.102099132910457e-06, "loss": 0.2201, "step": 42070 }, { "epoch": 0.7653831462922207, "grad_norm": 1.2074856758117676, "learning_rate": 3.1004753770071504e-06, "loss": 0.2229, "step": 42080 }, { "epoch": 0.765565033922043, "grad_norm": 0.9050807356834412, "learning_rate": 3.098852627339388e-06, "loss": 0.2204, "step": 42090 }, { "epoch": 0.7657469215518653, "grad_norm": 0.9774457812309265, "learning_rate": 3.0972308844703184e-06, "loss": 0.2279, "step": 42100 }, { "epoch": 0.7659288091816876, "grad_norm": 0.9778152704238892, "learning_rate": 3.0956101489627444e-06, "loss": 0.2243, "step": 42110 }, { "epoch": 0.7661106968115099, "grad_norm": 0.8497558832168579, "learning_rate": 3.093990421379116e-06, "loss": 0.2249, "step": 42120 }, { "epoch": 0.7662925844413322, "grad_norm": 1.0962015390396118, "learning_rate": 3.092371702281536e-06, "loss": 0.2226, "step": 42130 }, { "epoch": 0.7664744720711545, "grad_norm": 0.9380297064781189, "learning_rate": 3.0907539922317555e-06, "loss": 0.2224, "step": 42140 }, { "epoch": 0.7666563597009768, "grad_norm": 1.094885230064392, "learning_rate": 3.0891372917911733e-06, "loss": 0.2261, "step": 42150 }, { "epoch": 0.7668382473307991, "grad_norm": 1.0139778852462769, "learning_rate": 3.0875216015208444e-06, "loss": 0.2217, "step": 42160 }, { "epoch": 0.7670201349606214, "grad_norm": 1.0423494577407837, "learning_rate": 3.085906921981467e-06, "loss": 0.2216, "step": 42170 }, { "epoch": 0.7672020225904437, "grad_norm": 0.9817461371421814, "learning_rate": 3.0842932537333923e-06, "loss": 0.2227, "step": 42180 }, { "epoch": 0.767383910220266, "grad_norm": 1.0193676948547363, "learning_rate": 3.0826805973366193e-06, "loss": 0.2239, "step": 42190 }, { "epoch": 0.7675657978500882, "grad_norm": 1.0500943660736084, "learning_rate": 3.0810689533507944e-06, "loss": 0.2257, "step": 42200 }, { "epoch": 0.7677476854799105, "grad_norm": 1.0478789806365967, "learning_rate": 3.079458322335214e-06, "loss": 0.222, "step": 42210 }, { "epoch": 0.7679295731097328, "grad_norm": 0.9437698125839233, "learning_rate": 3.0778487048488236e-06, "loss": 0.2204, "step": 42220 }, { "epoch": 0.768111460739555, "grad_norm": 1.058676838874817, "learning_rate": 3.0762401014502158e-06, "loss": 0.2219, "step": 42230 }, { "epoch": 0.7682933483693773, "grad_norm": 0.9487638473510742, "learning_rate": 3.0746325126976324e-06, "loss": 0.2187, "step": 42240 }, { "epoch": 0.7684752359991996, "grad_norm": 0.9931320548057556, "learning_rate": 3.0730259391489616e-06, "loss": 0.2269, "step": 42250 }, { "epoch": 0.7686571236290219, "grad_norm": 1.0806828737258911, "learning_rate": 3.0714203813617395e-06, "loss": 0.2311, "step": 42260 }, { "epoch": 0.7688390112588442, "grad_norm": 1.006449580192566, "learning_rate": 3.069815839893152e-06, "loss": 0.2258, "step": 42270 }, { "epoch": 0.7690208988886665, "grad_norm": 1.054297685623169, "learning_rate": 3.0682123153000297e-06, "loss": 0.2198, "step": 42280 }, { "epoch": 0.7692027865184888, "grad_norm": 0.9744758009910583, "learning_rate": 3.066609808138852e-06, "loss": 0.2195, "step": 42290 }, { "epoch": 0.7693846741483111, "grad_norm": 1.0437334775924683, "learning_rate": 3.065008318965742e-06, "loss": 0.2305, "step": 42300 }, { "epoch": 0.7695665617781334, "grad_norm": 0.9644580483436584, "learning_rate": 3.063407848336475e-06, "loss": 0.2265, "step": 42310 }, { "epoch": 0.7697484494079557, "grad_norm": 1.015281319618225, "learning_rate": 3.061808396806468e-06, "loss": 0.2308, "step": 42320 }, { "epoch": 0.769930337037778, "grad_norm": 1.0188612937927246, "learning_rate": 3.0602099649307872e-06, "loss": 0.2252, "step": 42330 }, { "epoch": 0.7701122246676003, "grad_norm": 1.0173659324645996, "learning_rate": 3.058612553264143e-06, "loss": 0.2205, "step": 42340 }, { "epoch": 0.7702941122974226, "grad_norm": 0.977533757686615, "learning_rate": 3.0570161623608925e-06, "loss": 0.2256, "step": 42350 }, { "epoch": 0.7704759999272449, "grad_norm": 1.0004795789718628, "learning_rate": 3.055420792775039e-06, "loss": 0.2293, "step": 42360 }, { "epoch": 0.7706578875570672, "grad_norm": 1.1197783946990967, "learning_rate": 3.0538264450602317e-06, "loss": 0.2219, "step": 42370 }, { "epoch": 0.7708397751868895, "grad_norm": 0.988921582698822, "learning_rate": 3.0522331197697642e-06, "loss": 0.2222, "step": 42380 }, { "epoch": 0.7710216628167118, "grad_norm": 1.0667216777801514, "learning_rate": 3.050640817456575e-06, "loss": 0.224, "step": 42390 }, { "epoch": 0.7712035504465341, "grad_norm": 1.0161266326904297, "learning_rate": 3.049049538673247e-06, "loss": 0.2226, "step": 42400 }, { "epoch": 0.7713854380763564, "grad_norm": 1.0432935953140259, "learning_rate": 3.0474592839720116e-06, "loss": 0.2187, "step": 42410 }, { "epoch": 0.7715673257061787, "grad_norm": 0.9216803312301636, "learning_rate": 3.0458700539047426e-06, "loss": 0.2212, "step": 42420 }, { "epoch": 0.771749213336001, "grad_norm": 1.1149756908416748, "learning_rate": 3.0442818490229554e-06, "loss": 0.2302, "step": 42430 }, { "epoch": 0.7719311009658233, "grad_norm": 1.0001018047332764, "learning_rate": 3.042694669877814e-06, "loss": 0.2223, "step": 42440 }, { "epoch": 0.7721129885956456, "grad_norm": 1.008204460144043, "learning_rate": 3.041108517020123e-06, "loss": 0.2236, "step": 42450 }, { "epoch": 0.7722948762254679, "grad_norm": 0.9972517490386963, "learning_rate": 3.0395233910003317e-06, "loss": 0.2299, "step": 42460 }, { "epoch": 0.7724767638552902, "grad_norm": 1.0746959447860718, "learning_rate": 3.037939292368538e-06, "loss": 0.2275, "step": 42470 }, { "epoch": 0.7726586514851125, "grad_norm": 1.0281002521514893, "learning_rate": 3.036356221674477e-06, "loss": 0.2255, "step": 42480 }, { "epoch": 0.7728405391149348, "grad_norm": 0.933452308177948, "learning_rate": 3.0347741794675272e-06, "loss": 0.2226, "step": 42490 }, { "epoch": 0.7730224267447571, "grad_norm": 1.0657762289047241, "learning_rate": 3.0331931662967127e-06, "loss": 0.2225, "step": 42500 }, { "epoch": 0.7732043143745794, "grad_norm": 1.0098363161087036, "learning_rate": 3.0316131827107022e-06, "loss": 0.222, "step": 42510 }, { "epoch": 0.7733862020044017, "grad_norm": 0.9840245246887207, "learning_rate": 3.0300342292578027e-06, "loss": 0.2208, "step": 42520 }, { "epoch": 0.773568089634224, "grad_norm": 0.869941234588623, "learning_rate": 3.028456306485966e-06, "loss": 0.224, "step": 42530 }, { "epoch": 0.7737499772640463, "grad_norm": 0.9715585708618164, "learning_rate": 3.0268794149427873e-06, "loss": 0.224, "step": 42540 }, { "epoch": 0.7739318648938686, "grad_norm": 1.0423235893249512, "learning_rate": 3.0253035551754992e-06, "loss": 0.2282, "step": 42550 }, { "epoch": 0.7741137525236909, "grad_norm": 0.9641357660293579, "learning_rate": 3.0237287277309836e-06, "loss": 0.2214, "step": 42560 }, { "epoch": 0.7742956401535132, "grad_norm": 1.055687665939331, "learning_rate": 3.0221549331557587e-06, "loss": 0.2204, "step": 42570 }, { "epoch": 0.7744775277833355, "grad_norm": 1.0532023906707764, "learning_rate": 3.020582171995986e-06, "loss": 0.228, "step": 42580 }, { "epoch": 0.7746594154131577, "grad_norm": 0.9986780881881714, "learning_rate": 3.0190104447974675e-06, "loss": 0.2277, "step": 42590 }, { "epoch": 0.77484130304298, "grad_norm": 0.9063992500305176, "learning_rate": 3.0174397521056465e-06, "loss": 0.2215, "step": 42600 }, { "epoch": 0.7750231906728023, "grad_norm": 1.0739573240280151, "learning_rate": 3.0158700944656093e-06, "loss": 0.2263, "step": 42610 }, { "epoch": 0.7752050783026246, "grad_norm": 1.0597399473190308, "learning_rate": 3.014301472422081e-06, "loss": 0.2202, "step": 42620 }, { "epoch": 0.7753869659324469, "grad_norm": 0.9686228632926941, "learning_rate": 3.0127338865194276e-06, "loss": 0.2229, "step": 42630 }, { "epoch": 0.7755688535622692, "grad_norm": 1.0322505235671997, "learning_rate": 3.011167337301656e-06, "loss": 0.224, "step": 42640 }, { "epoch": 0.7757507411920915, "grad_norm": 1.0502586364746094, "learning_rate": 3.0096018253124117e-06, "loss": 0.2263, "step": 42650 }, { "epoch": 0.7759326288219138, "grad_norm": 1.0710793733596802, "learning_rate": 3.0080373510949845e-06, "loss": 0.2312, "step": 42660 }, { "epoch": 0.7761145164517361, "grad_norm": 1.0213178396224976, "learning_rate": 3.0064739151922984e-06, "loss": 0.2243, "step": 42670 }, { "epoch": 0.7762964040815584, "grad_norm": 0.9121502637863159, "learning_rate": 3.004911518146922e-06, "loss": 0.2216, "step": 42680 }, { "epoch": 0.7764782917113807, "grad_norm": 1.0305994749069214, "learning_rate": 3.0033501605010585e-06, "loss": 0.2328, "step": 42690 }, { "epoch": 0.776660179341203, "grad_norm": 1.1562353372573853, "learning_rate": 3.0017898427965553e-06, "loss": 0.2221, "step": 42700 }, { "epoch": 0.7768420669710253, "grad_norm": 1.0079669952392578, "learning_rate": 3.000230565574897e-06, "loss": 0.2268, "step": 42710 }, { "epoch": 0.7770239546008476, "grad_norm": 1.0499202013015747, "learning_rate": 2.9986723293772046e-06, "loss": 0.2183, "step": 42720 }, { "epoch": 0.7772058422306699, "grad_norm": 1.015758752822876, "learning_rate": 2.997115134744243e-06, "loss": 0.2287, "step": 42730 }, { "epoch": 0.7773877298604922, "grad_norm": 0.9919911623001099, "learning_rate": 2.9955589822164112e-06, "loss": 0.2221, "step": 42740 }, { "epoch": 0.7775696174903145, "grad_norm": 1.0310574769973755, "learning_rate": 2.994003872333747e-06, "loss": 0.2253, "step": 42750 }, { "epoch": 0.7777515051201368, "grad_norm": 1.0149425268173218, "learning_rate": 2.9924498056359295e-06, "loss": 0.2183, "step": 42760 }, { "epoch": 0.7779333927499591, "grad_norm": 1.1043009757995605, "learning_rate": 2.9908967826622726e-06, "loss": 0.2251, "step": 42770 }, { "epoch": 0.7781152803797814, "grad_norm": 1.0646169185638428, "learning_rate": 2.989344803951729e-06, "loss": 0.2228, "step": 42780 }, { "epoch": 0.7782971680096037, "grad_norm": 1.0975682735443115, "learning_rate": 2.98779387004289e-06, "loss": 0.2245, "step": 42790 }, { "epoch": 0.778479055639426, "grad_norm": 1.0264967679977417, "learning_rate": 2.9862439814739815e-06, "loss": 0.2249, "step": 42800 }, { "epoch": 0.7786609432692483, "grad_norm": 1.079259991645813, "learning_rate": 2.9846951387828705e-06, "loss": 0.2284, "step": 42810 }, { "epoch": 0.7788428308990706, "grad_norm": 1.1294838190078735, "learning_rate": 2.9831473425070588e-06, "loss": 0.2276, "step": 42820 }, { "epoch": 0.7790247185288929, "grad_norm": 0.9243553280830383, "learning_rate": 2.9816005931836845e-06, "loss": 0.224, "step": 42830 }, { "epoch": 0.7792066061587152, "grad_norm": 1.080892562866211, "learning_rate": 2.9800548913495243e-06, "loss": 0.2202, "step": 42840 }, { "epoch": 0.7793884937885375, "grad_norm": 1.0095688104629517, "learning_rate": 2.978510237540988e-06, "loss": 0.2237, "step": 42850 }, { "epoch": 0.7795703814183598, "grad_norm": 0.9821520447731018, "learning_rate": 2.9769666322941272e-06, "loss": 0.2262, "step": 42860 }, { "epoch": 0.7797522690481821, "grad_norm": 1.0111538171768188, "learning_rate": 2.975424076144625e-06, "loss": 0.2244, "step": 42870 }, { "epoch": 0.7799341566780044, "grad_norm": 0.9374005794525146, "learning_rate": 2.9738825696278016e-06, "loss": 0.2248, "step": 42880 }, { "epoch": 0.7801160443078267, "grad_norm": 1.0423915386199951, "learning_rate": 2.9723421132786125e-06, "loss": 0.229, "step": 42890 }, { "epoch": 0.780297931937649, "grad_norm": 0.91768878698349, "learning_rate": 2.9708027076316513e-06, "loss": 0.2193, "step": 42900 }, { "epoch": 0.7804798195674713, "grad_norm": 1.0346136093139648, "learning_rate": 2.969264353221144e-06, "loss": 0.2243, "step": 42910 }, { "epoch": 0.7806617071972936, "grad_norm": 1.0155085325241089, "learning_rate": 2.967727050580953e-06, "loss": 0.2199, "step": 42920 }, { "epoch": 0.7808435948271158, "grad_norm": 1.0129344463348389, "learning_rate": 2.9661908002445763e-06, "loss": 0.2226, "step": 42930 }, { "epoch": 0.7810254824569381, "grad_norm": 0.9926137328147888, "learning_rate": 2.964655602745143e-06, "loss": 0.2286, "step": 42940 }, { "epoch": 0.7812073700867604, "grad_norm": 0.9695350527763367, "learning_rate": 2.963121458615425e-06, "loss": 0.2231, "step": 42950 }, { "epoch": 0.7813892577165827, "grad_norm": 0.9467591643333435, "learning_rate": 2.96158836838782e-06, "loss": 0.2248, "step": 42960 }, { "epoch": 0.781571145346405, "grad_norm": 1.0627413988113403, "learning_rate": 2.9600563325943633e-06, "loss": 0.2221, "step": 42970 }, { "epoch": 0.7817530329762273, "grad_norm": 0.9747713208198547, "learning_rate": 2.958525351766727e-06, "loss": 0.2234, "step": 42980 }, { "epoch": 0.7819349206060496, "grad_norm": 1.0177266597747803, "learning_rate": 2.95699542643621e-06, "loss": 0.2205, "step": 42990 }, { "epoch": 0.7821168082358719, "grad_norm": 0.9680036306381226, "learning_rate": 2.9554665571337525e-06, "loss": 0.2163, "step": 43000 }, { "epoch": 0.7822986958656942, "grad_norm": 1.0278232097625732, "learning_rate": 2.953938744389926e-06, "loss": 0.222, "step": 43010 }, { "epoch": 0.7824805834955165, "grad_norm": 0.9525260329246521, "learning_rate": 2.9524119887349334e-06, "loss": 0.2248, "step": 43020 }, { "epoch": 0.7826624711253388, "grad_norm": 1.0004374980926514, "learning_rate": 2.9508862906986125e-06, "loss": 0.2228, "step": 43030 }, { "epoch": 0.782844358755161, "grad_norm": 0.9735926389694214, "learning_rate": 2.9493616508104315e-06, "loss": 0.2286, "step": 43040 }, { "epoch": 0.7830262463849833, "grad_norm": 0.9513868093490601, "learning_rate": 2.9478380695994934e-06, "loss": 0.2216, "step": 43050 }, { "epoch": 0.7832081340148056, "grad_norm": 1.07111656665802, "learning_rate": 2.946315547594536e-06, "loss": 0.2219, "step": 43060 }, { "epoch": 0.7833900216446279, "grad_norm": 0.9514635801315308, "learning_rate": 2.9447940853239264e-06, "loss": 0.2249, "step": 43070 }, { "epoch": 0.7835719092744502, "grad_norm": 1.0454280376434326, "learning_rate": 2.943273683315664e-06, "loss": 0.2216, "step": 43080 }, { "epoch": 0.7837537969042725, "grad_norm": 1.142444372177124, "learning_rate": 2.9417543420973805e-06, "loss": 0.2226, "step": 43090 }, { "epoch": 0.7839356845340948, "grad_norm": 1.043087363243103, "learning_rate": 2.940236062196341e-06, "loss": 0.2218, "step": 43100 }, { "epoch": 0.7841175721639171, "grad_norm": 0.9096753001213074, "learning_rate": 2.9387188441394408e-06, "loss": 0.2182, "step": 43110 }, { "epoch": 0.7842994597937394, "grad_norm": 0.9688875675201416, "learning_rate": 2.9372026884532084e-06, "loss": 0.2239, "step": 43120 }, { "epoch": 0.7844813474235617, "grad_norm": 0.9774050116539001, "learning_rate": 2.9356875956638008e-06, "loss": 0.2277, "step": 43130 }, { "epoch": 0.784663235053384, "grad_norm": 1.1142789125442505, "learning_rate": 2.9341735662970073e-06, "loss": 0.2251, "step": 43140 }, { "epoch": 0.7848451226832063, "grad_norm": 1.0595271587371826, "learning_rate": 2.93266060087825e-06, "loss": 0.2261, "step": 43150 }, { "epoch": 0.7850270103130286, "grad_norm": 1.1221669912338257, "learning_rate": 2.93114869993258e-06, "loss": 0.2221, "step": 43160 }, { "epoch": 0.7852088979428509, "grad_norm": 0.9699030518531799, "learning_rate": 2.9296378639846797e-06, "loss": 0.2189, "step": 43170 }, { "epoch": 0.7853907855726732, "grad_norm": 0.9606396555900574, "learning_rate": 2.9281280935588606e-06, "loss": 0.2235, "step": 43180 }, { "epoch": 0.7855726732024955, "grad_norm": 0.9698804020881653, "learning_rate": 2.926619389179065e-06, "loss": 0.222, "step": 43190 }, { "epoch": 0.7857545608323178, "grad_norm": 1.0965737104415894, "learning_rate": 2.9251117513688665e-06, "loss": 0.2188, "step": 43200 }, { "epoch": 0.7859364484621401, "grad_norm": 0.9659901857376099, "learning_rate": 2.9236051806514687e-06, "loss": 0.2252, "step": 43210 }, { "epoch": 0.7861183360919624, "grad_norm": 1.027941107749939, "learning_rate": 2.9220996775497023e-06, "loss": 0.2283, "step": 43220 }, { "epoch": 0.7863002237217847, "grad_norm": 1.0863381624221802, "learning_rate": 2.9205952425860277e-06, "loss": 0.2215, "step": 43230 }, { "epoch": 0.786482111351607, "grad_norm": 0.9744716286659241, "learning_rate": 2.919091876282539e-06, "loss": 0.2207, "step": 43240 }, { "epoch": 0.7866639989814292, "grad_norm": 1.0929560661315918, "learning_rate": 2.9175895791609556e-06, "loss": 0.2309, "step": 43250 }, { "epoch": 0.7868458866112515, "grad_norm": 0.97994065284729, "learning_rate": 2.9160883517426265e-06, "loss": 0.2265, "step": 43260 }, { "epoch": 0.7870277742410738, "grad_norm": 1.0885083675384521, "learning_rate": 2.9145881945485277e-06, "loss": 0.2256, "step": 43270 }, { "epoch": 0.7872096618708961, "grad_norm": 1.0820889472961426, "learning_rate": 2.9130891080992684e-06, "loss": 0.221, "step": 43280 }, { "epoch": 0.7873915495007184, "grad_norm": 1.0074574947357178, "learning_rate": 2.911591092915082e-06, "loss": 0.2232, "step": 43290 }, { "epoch": 0.7875734371305407, "grad_norm": 0.9417107701301575, "learning_rate": 2.910094149515833e-06, "loss": 0.2176, "step": 43300 }, { "epoch": 0.787755324760363, "grad_norm": 1.079524040222168, "learning_rate": 2.908598278421012e-06, "loss": 0.2225, "step": 43310 }, { "epoch": 0.7879372123901853, "grad_norm": 1.077722430229187, "learning_rate": 2.9071034801497376e-06, "loss": 0.2217, "step": 43320 }, { "epoch": 0.7881191000200076, "grad_norm": 1.107809066772461, "learning_rate": 2.9056097552207572e-06, "loss": 0.2247, "step": 43330 }, { "epoch": 0.7883009876498299, "grad_norm": 0.9996061325073242, "learning_rate": 2.9041171041524452e-06, "loss": 0.224, "step": 43340 }, { "epoch": 0.7884828752796522, "grad_norm": 1.0058493614196777, "learning_rate": 2.902625527462802e-06, "loss": 0.2187, "step": 43350 }, { "epoch": 0.7886647629094745, "grad_norm": 0.9888401627540588, "learning_rate": 2.9011350256694592e-06, "loss": 0.2291, "step": 43360 }, { "epoch": 0.7888466505392968, "grad_norm": 0.995897650718689, "learning_rate": 2.89964559928967e-06, "loss": 0.2191, "step": 43370 }, { "epoch": 0.7890285381691191, "grad_norm": 0.8860456943511963, "learning_rate": 2.8981572488403176e-06, "loss": 0.2196, "step": 43380 }, { "epoch": 0.7892104257989414, "grad_norm": 0.9556050300598145, "learning_rate": 2.8966699748379107e-06, "loss": 0.2165, "step": 43390 }, { "epoch": 0.7893923134287637, "grad_norm": 0.962722659111023, "learning_rate": 2.8951837777985862e-06, "loss": 0.2291, "step": 43400 }, { "epoch": 0.789574201058586, "grad_norm": 1.1376715898513794, "learning_rate": 2.893698658238106e-06, "loss": 0.2239, "step": 43410 }, { "epoch": 0.7897560886884083, "grad_norm": 1.0388892889022827, "learning_rate": 2.892214616671857e-06, "loss": 0.2177, "step": 43420 }, { "epoch": 0.7899379763182306, "grad_norm": 0.9921071529388428, "learning_rate": 2.8907316536148524e-06, "loss": 0.2207, "step": 43430 }, { "epoch": 0.7901198639480529, "grad_norm": 1.0287737846374512, "learning_rate": 2.8892497695817344e-06, "loss": 0.2187, "step": 43440 }, { "epoch": 0.7903017515778752, "grad_norm": 1.1230801343917847, "learning_rate": 2.887768965086766e-06, "loss": 0.2177, "step": 43450 }, { "epoch": 0.7904836392076975, "grad_norm": 1.0371381044387817, "learning_rate": 2.886289240643839e-06, "loss": 0.2298, "step": 43460 }, { "epoch": 0.7906655268375198, "grad_norm": 1.0685036182403564, "learning_rate": 2.8848105967664676e-06, "loss": 0.2256, "step": 43470 }, { "epoch": 0.7908474144673421, "grad_norm": 1.1212258338928223, "learning_rate": 2.883333033967792e-06, "loss": 0.2273, "step": 43480 }, { "epoch": 0.7910293020971644, "grad_norm": 1.0240336656570435, "learning_rate": 2.881856552760581e-06, "loss": 0.2269, "step": 43490 }, { "epoch": 0.7912111897269867, "grad_norm": 0.9120620489120483, "learning_rate": 2.8803811536572227e-06, "loss": 0.2217, "step": 43500 }, { "epoch": 0.791393077356809, "grad_norm": 0.9415653944015503, "learning_rate": 2.8789068371697316e-06, "loss": 0.2216, "step": 43510 }, { "epoch": 0.7915749649866313, "grad_norm": 1.0183041095733643, "learning_rate": 2.877433603809746e-06, "loss": 0.2296, "step": 43520 }, { "epoch": 0.7917568526164536, "grad_norm": 0.9307426810264587, "learning_rate": 2.8759614540885296e-06, "loss": 0.2232, "step": 43530 }, { "epoch": 0.7919387402462759, "grad_norm": 0.9421246647834778, "learning_rate": 2.87449038851697e-06, "loss": 0.2242, "step": 43540 }, { "epoch": 0.7921206278760982, "grad_norm": 0.9676687121391296, "learning_rate": 2.8730204076055755e-06, "loss": 0.2203, "step": 43550 }, { "epoch": 0.7923025155059205, "grad_norm": 1.0517562627792358, "learning_rate": 2.871551511864485e-06, "loss": 0.2293, "step": 43560 }, { "epoch": 0.7924844031357428, "grad_norm": 0.9216522574424744, "learning_rate": 2.8700837018034534e-06, "loss": 0.2228, "step": 43570 }, { "epoch": 0.792666290765565, "grad_norm": 1.0129079818725586, "learning_rate": 2.8686169779318608e-06, "loss": 0.2257, "step": 43580 }, { "epoch": 0.7928481783953873, "grad_norm": 1.03448486328125, "learning_rate": 2.8671513407587115e-06, "loss": 0.2258, "step": 43590 }, { "epoch": 0.7930300660252096, "grad_norm": 0.9085607528686523, "learning_rate": 2.8656867907926345e-06, "loss": 0.2242, "step": 43600 }, { "epoch": 0.7932119536550319, "grad_norm": 1.0849491357803345, "learning_rate": 2.8642233285418787e-06, "loss": 0.2243, "step": 43610 }, { "epoch": 0.7933938412848542, "grad_norm": 0.9323189854621887, "learning_rate": 2.862760954514314e-06, "loss": 0.2241, "step": 43620 }, { "epoch": 0.7935757289146765, "grad_norm": 0.9408084154129028, "learning_rate": 2.8612996692174367e-06, "loss": 0.2205, "step": 43630 }, { "epoch": 0.7937576165444988, "grad_norm": 1.0064846277236938, "learning_rate": 2.8598394731583635e-06, "loss": 0.2252, "step": 43640 }, { "epoch": 0.7939395041743211, "grad_norm": 0.9861797094345093, "learning_rate": 2.858380366843834e-06, "loss": 0.2276, "step": 43650 }, { "epoch": 0.7941213918041434, "grad_norm": 0.9897487163543701, "learning_rate": 2.856922350780206e-06, "loss": 0.2235, "step": 43660 }, { "epoch": 0.7943032794339657, "grad_norm": 0.947838306427002, "learning_rate": 2.8554654254734637e-06, "loss": 0.2199, "step": 43670 }, { "epoch": 0.794485167063788, "grad_norm": 0.9508879780769348, "learning_rate": 2.854009591429209e-06, "loss": 0.2198, "step": 43680 }, { "epoch": 0.7946670546936103, "grad_norm": 1.0373586416244507, "learning_rate": 2.8525548491526684e-06, "loss": 0.2249, "step": 43690 }, { "epoch": 0.7948489423234326, "grad_norm": 0.9972711801528931, "learning_rate": 2.851101199148687e-06, "loss": 0.222, "step": 43700 }, { "epoch": 0.7950308299532549, "grad_norm": 0.97149658203125, "learning_rate": 2.8496486419217327e-06, "loss": 0.2166, "step": 43710 }, { "epoch": 0.7952127175830772, "grad_norm": 1.0728939771652222, "learning_rate": 2.848197177975893e-06, "loss": 0.2216, "step": 43720 }, { "epoch": 0.7953946052128995, "grad_norm": 1.009264349937439, "learning_rate": 2.846746807814873e-06, "loss": 0.2248, "step": 43730 }, { "epoch": 0.7955764928427218, "grad_norm": 1.057274341583252, "learning_rate": 2.8452975319420062e-06, "loss": 0.2242, "step": 43740 }, { "epoch": 0.7957583804725441, "grad_norm": 1.045536994934082, "learning_rate": 2.8438493508602393e-06, "loss": 0.2231, "step": 43750 }, { "epoch": 0.7959402681023664, "grad_norm": 0.9564407467842102, "learning_rate": 2.8424022650721415e-06, "loss": 0.2223, "step": 43760 }, { "epoch": 0.7961221557321887, "grad_norm": 0.9497394561767578, "learning_rate": 2.8409562750799026e-06, "loss": 0.2204, "step": 43770 }, { "epoch": 0.796304043362011, "grad_norm": 0.8792627453804016, "learning_rate": 2.8395113813853286e-06, "loss": 0.22, "step": 43780 }, { "epoch": 0.7964859309918333, "grad_norm": 1.034838080406189, "learning_rate": 2.8380675844898507e-06, "loss": 0.2239, "step": 43790 }, { "epoch": 0.7966678186216556, "grad_norm": 1.069647192955017, "learning_rate": 2.8366248848945144e-06, "loss": 0.2201, "step": 43800 }, { "epoch": 0.7968497062514779, "grad_norm": 0.9620888829231262, "learning_rate": 2.835183283099989e-06, "loss": 0.2185, "step": 43810 }, { "epoch": 0.7970315938813002, "grad_norm": 1.0037379264831543, "learning_rate": 2.833742779606556e-06, "loss": 0.2197, "step": 43820 }, { "epoch": 0.7972134815111225, "grad_norm": 0.9222750067710876, "learning_rate": 2.8323033749141228e-06, "loss": 0.2246, "step": 43830 }, { "epoch": 0.7973953691409448, "grad_norm": 0.9481472969055176, "learning_rate": 2.8308650695222145e-06, "loss": 0.2238, "step": 43840 }, { "epoch": 0.7975772567707671, "grad_norm": 0.9374997019767761, "learning_rate": 2.82942786392997e-06, "loss": 0.2246, "step": 43850 }, { "epoch": 0.7977591444005894, "grad_norm": 0.9594138860702515, "learning_rate": 2.8279917586361516e-06, "loss": 0.2195, "step": 43860 }, { "epoch": 0.7979410320304116, "grad_norm": 0.9104349613189697, "learning_rate": 2.826556754139135e-06, "loss": 0.2243, "step": 43870 }, { "epoch": 0.7981229196602339, "grad_norm": 0.989269495010376, "learning_rate": 2.8251228509369183e-06, "loss": 0.2262, "step": 43880 }, { "epoch": 0.7983048072900562, "grad_norm": 1.006292462348938, "learning_rate": 2.8236900495271158e-06, "loss": 0.2227, "step": 43890 }, { "epoch": 0.7984866949198784, "grad_norm": 0.9648720026016235, "learning_rate": 2.8222583504069582e-06, "loss": 0.223, "step": 43900 }, { "epoch": 0.7986685825497007, "grad_norm": 1.1538965702056885, "learning_rate": 2.8208277540732953e-06, "loss": 0.2254, "step": 43910 }, { "epoch": 0.798850470179523, "grad_norm": 1.1156715154647827, "learning_rate": 2.8193982610225946e-06, "loss": 0.2272, "step": 43920 }, { "epoch": 0.7990323578093453, "grad_norm": 1.0434162616729736, "learning_rate": 2.8179698717509362e-06, "loss": 0.2208, "step": 43930 }, { "epoch": 0.7992142454391676, "grad_norm": 0.9618845582008362, "learning_rate": 2.816542586754025e-06, "loss": 0.2194, "step": 43940 }, { "epoch": 0.7993961330689899, "grad_norm": 1.0847927331924438, "learning_rate": 2.8151164065271763e-06, "loss": 0.221, "step": 43950 }, { "epoch": 0.7995780206988122, "grad_norm": 0.9194183945655823, "learning_rate": 2.813691331565325e-06, "loss": 0.2177, "step": 43960 }, { "epoch": 0.7997599083286345, "grad_norm": 1.091462254524231, "learning_rate": 2.8122673623630203e-06, "loss": 0.23, "step": 43970 }, { "epoch": 0.7999417959584568, "grad_norm": 1.048332691192627, "learning_rate": 2.810844499414428e-06, "loss": 0.2194, "step": 43980 }, { "epoch": 0.8001236835882791, "grad_norm": 1.0637898445129395, "learning_rate": 2.8094227432133335e-06, "loss": 0.2235, "step": 43990 }, { "epoch": 0.8003055712181014, "grad_norm": 1.0327093601226807, "learning_rate": 2.808002094253134e-06, "loss": 0.2218, "step": 44000 }, { "epoch": 0.8004874588479237, "grad_norm": 1.0266443490982056, "learning_rate": 2.8065825530268442e-06, "loss": 0.2253, "step": 44010 }, { "epoch": 0.800669346477746, "grad_norm": 1.0395704507827759, "learning_rate": 2.805164120027093e-06, "loss": 0.2229, "step": 44020 }, { "epoch": 0.8008512341075683, "grad_norm": 0.9720150232315063, "learning_rate": 2.8037467957461284e-06, "loss": 0.221, "step": 44030 }, { "epoch": 0.8010331217373906, "grad_norm": 0.9836148619651794, "learning_rate": 2.8023305806758093e-06, "loss": 0.2239, "step": 44040 }, { "epoch": 0.8012150093672129, "grad_norm": 0.9741009473800659, "learning_rate": 2.800915475307612e-06, "loss": 0.2266, "step": 44050 }, { "epoch": 0.8013968969970352, "grad_norm": 0.9533312916755676, "learning_rate": 2.799501480132626e-06, "loss": 0.2164, "step": 44060 }, { "epoch": 0.8015787846268575, "grad_norm": 0.9519167542457581, "learning_rate": 2.798088595641556e-06, "loss": 0.2217, "step": 44070 }, { "epoch": 0.8017606722566798, "grad_norm": 1.0319180488586426, "learning_rate": 2.7966768223247256e-06, "loss": 0.2192, "step": 44080 }, { "epoch": 0.8019425598865021, "grad_norm": 1.066082239151001, "learning_rate": 2.7952661606720654e-06, "loss": 0.2228, "step": 44090 }, { "epoch": 0.8021244475163244, "grad_norm": 1.0164188146591187, "learning_rate": 2.7938566111731245e-06, "loss": 0.2236, "step": 44100 }, { "epoch": 0.8023063351461467, "grad_norm": 1.0140495300292969, "learning_rate": 2.792448174317068e-06, "loss": 0.2257, "step": 44110 }, { "epoch": 0.802488222775969, "grad_norm": 1.2520900964736938, "learning_rate": 2.79104085059267e-06, "loss": 0.2283, "step": 44120 }, { "epoch": 0.8026701104057913, "grad_norm": 1.0057297945022583, "learning_rate": 2.7896346404883202e-06, "loss": 0.2245, "step": 44130 }, { "epoch": 0.8028519980356136, "grad_norm": 0.9313770532608032, "learning_rate": 2.788229544492024e-06, "loss": 0.2239, "step": 44140 }, { "epoch": 0.8030338856654359, "grad_norm": 0.9771432876586914, "learning_rate": 2.7868255630913976e-06, "loss": 0.2213, "step": 44150 }, { "epoch": 0.8032157732952582, "grad_norm": 0.9113761186599731, "learning_rate": 2.785422696773671e-06, "loss": 0.2214, "step": 44160 }, { "epoch": 0.8033976609250805, "grad_norm": 1.0070172548294067, "learning_rate": 2.7840209460256874e-06, "loss": 0.217, "step": 44170 }, { "epoch": 0.8035795485549028, "grad_norm": 0.9579640030860901, "learning_rate": 2.7826203113339024e-06, "loss": 0.2232, "step": 44180 }, { "epoch": 0.8037614361847251, "grad_norm": 0.9842847585678101, "learning_rate": 2.7812207931843865e-06, "loss": 0.2205, "step": 44190 }, { "epoch": 0.8039433238145474, "grad_norm": 0.9915623664855957, "learning_rate": 2.7798223920628197e-06, "loss": 0.2255, "step": 44200 }, { "epoch": 0.8041252114443697, "grad_norm": 1.041277527809143, "learning_rate": 2.7784251084544962e-06, "loss": 0.2274, "step": 44210 }, { "epoch": 0.804307099074192, "grad_norm": 0.9947366118431091, "learning_rate": 2.7770289428443208e-06, "loss": 0.2261, "step": 44220 }, { "epoch": 0.8044889867040143, "grad_norm": 0.991459310054779, "learning_rate": 2.7756338957168126e-06, "loss": 0.2213, "step": 44230 }, { "epoch": 0.8046708743338365, "grad_norm": 0.9859628081321716, "learning_rate": 2.7742399675561003e-06, "loss": 0.2191, "step": 44240 }, { "epoch": 0.8048527619636588, "grad_norm": 1.1321364641189575, "learning_rate": 2.7728471588459273e-06, "loss": 0.2274, "step": 44250 }, { "epoch": 0.8050346495934811, "grad_norm": 1.074383020401001, "learning_rate": 2.771455470069644e-06, "loss": 0.2223, "step": 44260 }, { "epoch": 0.8052165372233034, "grad_norm": 0.9538015723228455, "learning_rate": 2.7700649017102155e-06, "loss": 0.2237, "step": 44270 }, { "epoch": 0.8053984248531257, "grad_norm": 0.9737296104431152, "learning_rate": 2.768675454250217e-06, "loss": 0.2281, "step": 44280 }, { "epoch": 0.805580312482948, "grad_norm": 1.0212516784667969, "learning_rate": 2.767287128171836e-06, "loss": 0.2228, "step": 44290 }, { "epoch": 0.8057622001127703, "grad_norm": 1.0518425703048706, "learning_rate": 2.7658999239568697e-06, "loss": 0.2269, "step": 44300 }, { "epoch": 0.8059440877425926, "grad_norm": 1.0400155782699585, "learning_rate": 2.7645138420867247e-06, "loss": 0.2226, "step": 44310 }, { "epoch": 0.8061259753724149, "grad_norm": 1.1135287284851074, "learning_rate": 2.7631288830424195e-06, "loss": 0.2204, "step": 44320 }, { "epoch": 0.8063078630022372, "grad_norm": 0.9952766299247742, "learning_rate": 2.7617450473045844e-06, "loss": 0.2277, "step": 44330 }, { "epoch": 0.8064897506320595, "grad_norm": 1.0173605680465698, "learning_rate": 2.760362335353457e-06, "loss": 0.218, "step": 44340 }, { "epoch": 0.8066716382618818, "grad_norm": 1.1673275232315063, "learning_rate": 2.758980747668888e-06, "loss": 0.2248, "step": 44350 }, { "epoch": 0.8068535258917041, "grad_norm": 0.9442208409309387, "learning_rate": 2.757600284730334e-06, "loss": 0.2212, "step": 44360 }, { "epoch": 0.8070354135215264, "grad_norm": 0.9935293793678284, "learning_rate": 2.7562209470168643e-06, "loss": 0.2217, "step": 44370 }, { "epoch": 0.8072173011513487, "grad_norm": 1.0524063110351562, "learning_rate": 2.754842735007156e-06, "loss": 0.2304, "step": 44380 }, { "epoch": 0.807399188781171, "grad_norm": 1.0023491382598877, "learning_rate": 2.7534656491794996e-06, "loss": 0.2234, "step": 44390 }, { "epoch": 0.8075810764109933, "grad_norm": 1.3313415050506592, "learning_rate": 2.752089690011789e-06, "loss": 0.2262, "step": 44400 }, { "epoch": 0.8077629640408156, "grad_norm": 0.8770501017570496, "learning_rate": 2.75071485798153e-06, "loss": 0.223, "step": 44410 }, { "epoch": 0.8079448516706379, "grad_norm": 1.0588210821151733, "learning_rate": 2.7493411535658364e-06, "loss": 0.2272, "step": 44420 }, { "epoch": 0.8081267393004602, "grad_norm": 1.1327773332595825, "learning_rate": 2.7479685772414336e-06, "loss": 0.2223, "step": 44430 }, { "epoch": 0.8083086269302825, "grad_norm": 0.9941517114639282, "learning_rate": 2.7465971294846517e-06, "loss": 0.2192, "step": 44440 }, { "epoch": 0.8084905145601048, "grad_norm": 0.905812680721283, "learning_rate": 2.7452268107714302e-06, "loss": 0.2229, "step": 44450 }, { "epoch": 0.8086724021899271, "grad_norm": 1.024550437927246, "learning_rate": 2.7438576215773173e-06, "loss": 0.22, "step": 44460 }, { "epoch": 0.8088542898197494, "grad_norm": 1.0377891063690186, "learning_rate": 2.7424895623774684e-06, "loss": 0.2205, "step": 44470 }, { "epoch": 0.8090361774495717, "grad_norm": 0.9489783048629761, "learning_rate": 2.7411226336466497e-06, "loss": 0.2211, "step": 44480 }, { "epoch": 0.809218065079394, "grad_norm": 1.0196104049682617, "learning_rate": 2.7397568358592313e-06, "loss": 0.2196, "step": 44490 }, { "epoch": 0.8093999527092163, "grad_norm": 1.0149619579315186, "learning_rate": 2.738392169489193e-06, "loss": 0.2183, "step": 44500 }, { "epoch": 0.8095818403390386, "grad_norm": 0.8915206789970398, "learning_rate": 2.7370286350101204e-06, "loss": 0.2219, "step": 44510 }, { "epoch": 0.8097637279688609, "grad_norm": 0.9819897413253784, "learning_rate": 2.7356662328952073e-06, "loss": 0.2245, "step": 44520 }, { "epoch": 0.8099456155986832, "grad_norm": 1.0852417945861816, "learning_rate": 2.734304963617256e-06, "loss": 0.2235, "step": 44530 }, { "epoch": 0.8101275032285055, "grad_norm": 1.1220831871032715, "learning_rate": 2.7329448276486724e-06, "loss": 0.224, "step": 44540 }, { "epoch": 0.8103093908583278, "grad_norm": 1.0146201848983765, "learning_rate": 2.731585825461472e-06, "loss": 0.2274, "step": 44550 }, { "epoch": 0.8104912784881501, "grad_norm": 0.9456096887588501, "learning_rate": 2.730227957527275e-06, "loss": 0.2239, "step": 44560 }, { "epoch": 0.8106731661179724, "grad_norm": 0.9560818672180176, "learning_rate": 2.7288712243173075e-06, "loss": 0.2262, "step": 44570 }, { "epoch": 0.8108550537477947, "grad_norm": 0.9176381826400757, "learning_rate": 2.727515626302406e-06, "loss": 0.2254, "step": 44580 }, { "epoch": 0.811036941377617, "grad_norm": 1.0253549814224243, "learning_rate": 2.726161163953007e-06, "loss": 0.2189, "step": 44590 }, { "epoch": 0.8112188290074392, "grad_norm": 1.0547857284545898, "learning_rate": 2.7248078377391583e-06, "loss": 0.2229, "step": 44600 }, { "epoch": 0.8114007166372615, "grad_norm": 1.1865084171295166, "learning_rate": 2.7234556481305086e-06, "loss": 0.222, "step": 44610 }, { "epoch": 0.8115826042670838, "grad_norm": 0.9389124512672424, "learning_rate": 2.722104595596316e-06, "loss": 0.2221, "step": 44620 }, { "epoch": 0.8117644918969061, "grad_norm": 1.0560612678527832, "learning_rate": 2.720754680605443e-06, "loss": 0.2224, "step": 44630 }, { "epoch": 0.8119463795267284, "grad_norm": 1.1060501337051392, "learning_rate": 2.719405903626356e-06, "loss": 0.2216, "step": 44640 }, { "epoch": 0.8121282671565507, "grad_norm": 1.0781084299087524, "learning_rate": 2.7180582651271265e-06, "loss": 0.2234, "step": 44650 }, { "epoch": 0.812310154786373, "grad_norm": 0.9307900071144104, "learning_rate": 2.7167117655754343e-06, "loss": 0.2204, "step": 44660 }, { "epoch": 0.8124920424161953, "grad_norm": 0.9623140692710876, "learning_rate": 2.715366405438558e-06, "loss": 0.2158, "step": 44670 }, { "epoch": 0.8126739300460176, "grad_norm": 0.9945633411407471, "learning_rate": 2.7140221851833882e-06, "loss": 0.2196, "step": 44680 }, { "epoch": 0.8128558176758399, "grad_norm": 0.9559438824653625, "learning_rate": 2.7126791052764133e-06, "loss": 0.2189, "step": 44690 }, { "epoch": 0.8130377053056622, "grad_norm": 1.0379374027252197, "learning_rate": 2.7113371661837296e-06, "loss": 0.22, "step": 44700 }, { "epoch": 0.8132195929354844, "grad_norm": 0.9728280901908875, "learning_rate": 2.7099963683710354e-06, "loss": 0.2271, "step": 44710 }, { "epoch": 0.8134014805653067, "grad_norm": 1.0203497409820557, "learning_rate": 2.7086567123036343e-06, "loss": 0.2224, "step": 44720 }, { "epoch": 0.813583368195129, "grad_norm": 1.1351407766342163, "learning_rate": 2.7073181984464347e-06, "loss": 0.2228, "step": 44730 }, { "epoch": 0.8137652558249513, "grad_norm": 1.062680721282959, "learning_rate": 2.705980827263946e-06, "loss": 0.2191, "step": 44740 }, { "epoch": 0.8139471434547736, "grad_norm": 0.9409022927284241, "learning_rate": 2.7046445992202828e-06, "loss": 0.2171, "step": 44750 }, { "epoch": 0.8141290310845959, "grad_norm": 1.0605251789093018, "learning_rate": 2.7033095147791623e-06, "loss": 0.221, "step": 44760 }, { "epoch": 0.8143109187144182, "grad_norm": 1.0575889348983765, "learning_rate": 2.7019755744039045e-06, "loss": 0.2226, "step": 44770 }, { "epoch": 0.8144928063442405, "grad_norm": 0.933271050453186, "learning_rate": 2.7006427785574345e-06, "loss": 0.2204, "step": 44780 }, { "epoch": 0.8146746939740628, "grad_norm": 1.0247336626052856, "learning_rate": 2.6993111277022775e-06, "loss": 0.2215, "step": 44790 }, { "epoch": 0.8148565816038851, "grad_norm": 1.119522213935852, "learning_rate": 2.6979806223005638e-06, "loss": 0.225, "step": 44800 }, { "epoch": 0.8150384692337074, "grad_norm": 1.0373032093048096, "learning_rate": 2.6966512628140223e-06, "loss": 0.2265, "step": 44810 }, { "epoch": 0.8152203568635297, "grad_norm": 1.0728118419647217, "learning_rate": 2.6953230497039893e-06, "loss": 0.2246, "step": 44820 }, { "epoch": 0.815402244493352, "grad_norm": 0.9771208763122559, "learning_rate": 2.6939959834314e-06, "loss": 0.222, "step": 44830 }, { "epoch": 0.8155841321231743, "grad_norm": 0.9419203400611877, "learning_rate": 2.6926700644567925e-06, "loss": 0.2174, "step": 44840 }, { "epoch": 0.8157660197529966, "grad_norm": 0.9388554096221924, "learning_rate": 2.691345293240306e-06, "loss": 0.2159, "step": 44850 }, { "epoch": 0.8159479073828189, "grad_norm": 1.1119645833969116, "learning_rate": 2.6900216702416828e-06, "loss": 0.2184, "step": 44860 }, { "epoch": 0.8161297950126412, "grad_norm": 1.0003283023834229, "learning_rate": 2.6886991959202652e-06, "loss": 0.2219, "step": 44870 }, { "epoch": 0.8163116826424635, "grad_norm": 1.1034090518951416, "learning_rate": 2.6873778707349983e-06, "loss": 0.2301, "step": 44880 }, { "epoch": 0.8164935702722858, "grad_norm": 1.0011792182922363, "learning_rate": 2.686057695144428e-06, "loss": 0.2229, "step": 44890 }, { "epoch": 0.816675457902108, "grad_norm": 0.9099850654602051, "learning_rate": 2.6847386696066995e-06, "loss": 0.2195, "step": 44900 }, { "epoch": 0.8168573455319303, "grad_norm": 1.0517704486846924, "learning_rate": 2.6834207945795614e-06, "loss": 0.2225, "step": 44910 }, { "epoch": 0.8170392331617526, "grad_norm": 1.0462573766708374, "learning_rate": 2.6821040705203627e-06, "loss": 0.2266, "step": 44920 }, { "epoch": 0.8172211207915749, "grad_norm": 1.0631307363510132, "learning_rate": 2.680788497886051e-06, "loss": 0.2265, "step": 44930 }, { "epoch": 0.8174030084213972, "grad_norm": 1.1005617380142212, "learning_rate": 2.6794740771331763e-06, "loss": 0.2226, "step": 44940 }, { "epoch": 0.8175848960512195, "grad_norm": 0.9842298030853271, "learning_rate": 2.6781608087178885e-06, "loss": 0.224, "step": 44950 }, { "epoch": 0.8177667836810418, "grad_norm": 1.1224092245101929, "learning_rate": 2.676848693095937e-06, "loss": 0.2166, "step": 44960 }, { "epoch": 0.8179486713108641, "grad_norm": 1.0010684728622437, "learning_rate": 2.6755377307226694e-06, "loss": 0.2288, "step": 44970 }, { "epoch": 0.8181305589406864, "grad_norm": 1.0084420442581177, "learning_rate": 2.6742279220530387e-06, "loss": 0.2294, "step": 44980 }, { "epoch": 0.8183124465705087, "grad_norm": 0.9362150430679321, "learning_rate": 2.6729192675415916e-06, "loss": 0.2228, "step": 44990 }, { "epoch": 0.818494334200331, "grad_norm": 1.17454993724823, "learning_rate": 2.671611767642477e-06, "loss": 0.222, "step": 45000 }, { "epoch": 0.8186762218301533, "grad_norm": 0.9910193085670471, "learning_rate": 2.670305422809442e-06, "loss": 0.2225, "step": 45010 }, { "epoch": 0.8188581094599756, "grad_norm": 1.0515499114990234, "learning_rate": 2.6690002334958354e-06, "loss": 0.2239, "step": 45020 }, { "epoch": 0.8190399970897979, "grad_norm": 1.025694489479065, "learning_rate": 2.6676962001546014e-06, "loss": 0.2255, "step": 45030 }, { "epoch": 0.8192218847196202, "grad_norm": 0.9849122166633606, "learning_rate": 2.666393323238286e-06, "loss": 0.2158, "step": 45040 }, { "epoch": 0.8194037723494425, "grad_norm": 1.0386641025543213, "learning_rate": 2.6650916031990322e-06, "loss": 0.2229, "step": 45050 }, { "epoch": 0.8195856599792648, "grad_norm": 0.8869991302490234, "learning_rate": 2.6637910404885804e-06, "loss": 0.2194, "step": 45060 }, { "epoch": 0.8197675476090871, "grad_norm": 0.9872651100158691, "learning_rate": 2.6624916355582732e-06, "loss": 0.2208, "step": 45070 }, { "epoch": 0.8199494352389094, "grad_norm": 0.9550368785858154, "learning_rate": 2.6611933888590487e-06, "loss": 0.2228, "step": 45080 }, { "epoch": 0.8201313228687317, "grad_norm": 1.0715761184692383, "learning_rate": 2.6598963008414424e-06, "loss": 0.2184, "step": 45090 }, { "epoch": 0.820313210498554, "grad_norm": 0.9814006686210632, "learning_rate": 2.6586003719555896e-06, "loss": 0.2206, "step": 45100 }, { "epoch": 0.8204950981283763, "grad_norm": 1.1378893852233887, "learning_rate": 2.6573056026512215e-06, "loss": 0.2205, "step": 45110 }, { "epoch": 0.8206769857581986, "grad_norm": 1.0848628282546997, "learning_rate": 2.6560119933776707e-06, "loss": 0.219, "step": 45120 }, { "epoch": 0.8208588733880209, "grad_norm": 1.0404765605926514, "learning_rate": 2.6547195445838614e-06, "loss": 0.2246, "step": 45130 }, { "epoch": 0.8210407610178432, "grad_norm": 1.0723414421081543, "learning_rate": 2.6534282567183202e-06, "loss": 0.2187, "step": 45140 }, { "epoch": 0.8212226486476655, "grad_norm": 1.100123643875122, "learning_rate": 2.6521381302291678e-06, "loss": 0.2211, "step": 45150 }, { "epoch": 0.8214045362774878, "grad_norm": 1.0197398662567139, "learning_rate": 2.6508491655641214e-06, "loss": 0.2234, "step": 45160 }, { "epoch": 0.8215864239073101, "grad_norm": 1.0212470293045044, "learning_rate": 2.649561363170499e-06, "loss": 0.2185, "step": 45170 }, { "epoch": 0.8217683115371324, "grad_norm": 1.000978708267212, "learning_rate": 2.6482747234952123e-06, "loss": 0.22, "step": 45180 }, { "epoch": 0.8219501991669547, "grad_norm": 1.0122380256652832, "learning_rate": 2.646989246984768e-06, "loss": 0.2167, "step": 45190 }, { "epoch": 0.822132086796777, "grad_norm": 1.0904041528701782, "learning_rate": 2.645704934085272e-06, "loss": 0.2241, "step": 45200 }, { "epoch": 0.8223139744265993, "grad_norm": 1.0525559186935425, "learning_rate": 2.6444217852424253e-06, "loss": 0.2178, "step": 45210 }, { "epoch": 0.8224958620564216, "grad_norm": 1.0217771530151367, "learning_rate": 2.6431398009015265e-06, "loss": 0.2234, "step": 45220 }, { "epoch": 0.8226777496862439, "grad_norm": 1.0103594064712524, "learning_rate": 2.641858981507466e-06, "loss": 0.2208, "step": 45230 }, { "epoch": 0.8228596373160662, "grad_norm": 1.2114310264587402, "learning_rate": 2.6405793275047353e-06, "loss": 0.2286, "step": 45240 }, { "epoch": 0.8230415249458884, "grad_norm": 0.9984057545661926, "learning_rate": 2.639300839337417e-06, "loss": 0.2235, "step": 45250 }, { "epoch": 0.8232234125757107, "grad_norm": 1.0261670351028442, "learning_rate": 2.6380235174491897e-06, "loss": 0.2244, "step": 45260 }, { "epoch": 0.823405300205533, "grad_norm": 0.9633432030677795, "learning_rate": 2.636747362283331e-06, "loss": 0.2265, "step": 45270 }, { "epoch": 0.8235871878353553, "grad_norm": 0.9061949253082275, "learning_rate": 2.63547237428271e-06, "loss": 0.2168, "step": 45280 }, { "epoch": 0.8237690754651776, "grad_norm": 0.9818053245544434, "learning_rate": 2.6341985538897906e-06, "loss": 0.2216, "step": 45290 }, { "epoch": 0.8239509630949999, "grad_norm": 1.0422415733337402, "learning_rate": 2.6329259015466345e-06, "loss": 0.2245, "step": 45300 }, { "epoch": 0.8241328507248222, "grad_norm": 1.1089386940002441, "learning_rate": 2.6316544176948934e-06, "loss": 0.2248, "step": 45310 }, { "epoch": 0.8243147383546445, "grad_norm": 1.065927267074585, "learning_rate": 2.630384102775819e-06, "loss": 0.2217, "step": 45320 }, { "epoch": 0.8244966259844668, "grad_norm": 0.9989216327667236, "learning_rate": 2.6291149572302544e-06, "loss": 0.2178, "step": 45330 }, { "epoch": 0.8246785136142891, "grad_norm": 1.1084251403808594, "learning_rate": 2.6278469814986357e-06, "loss": 0.2242, "step": 45340 }, { "epoch": 0.8248604012441114, "grad_norm": 1.0397945642471313, "learning_rate": 2.6265801760209946e-06, "loss": 0.2243, "step": 45350 }, { "epoch": 0.8250422888739337, "grad_norm": 1.094642162322998, "learning_rate": 2.625314541236957e-06, "loss": 0.2273, "step": 45360 }, { "epoch": 0.825224176503756, "grad_norm": 1.1421188116073608, "learning_rate": 2.6240500775857424e-06, "loss": 0.2161, "step": 45370 }, { "epoch": 0.8254060641335783, "grad_norm": 1.186858892440796, "learning_rate": 2.6227867855061627e-06, "loss": 0.2212, "step": 45380 }, { "epoch": 0.8255879517634006, "grad_norm": 1.0875937938690186, "learning_rate": 2.6215246654366254e-06, "loss": 0.2281, "step": 45390 }, { "epoch": 0.8257698393932229, "grad_norm": 1.1750167608261108, "learning_rate": 2.6202637178151275e-06, "loss": 0.2229, "step": 45400 }, { "epoch": 0.8259517270230452, "grad_norm": 0.9085155129432678, "learning_rate": 2.6190039430792635e-06, "loss": 0.2158, "step": 45410 }, { "epoch": 0.8261336146528675, "grad_norm": 1.0824395418167114, "learning_rate": 2.617745341666219e-06, "loss": 0.2216, "step": 45420 }, { "epoch": 0.8263155022826898, "grad_norm": 0.9006544947624207, "learning_rate": 2.6164879140127714e-06, "loss": 0.2212, "step": 45430 }, { "epoch": 0.8264973899125121, "grad_norm": 0.9351937174797058, "learning_rate": 2.6152316605552924e-06, "loss": 0.2182, "step": 45440 }, { "epoch": 0.8266792775423344, "grad_norm": 1.0441341400146484, "learning_rate": 2.6139765817297434e-06, "loss": 0.2224, "step": 45450 }, { "epoch": 0.8268611651721567, "grad_norm": 1.1005812883377075, "learning_rate": 2.6127226779716836e-06, "loss": 0.2186, "step": 45460 }, { "epoch": 0.827043052801979, "grad_norm": 0.9451075196266174, "learning_rate": 2.6114699497162587e-06, "loss": 0.2195, "step": 45470 }, { "epoch": 0.8272249404318013, "grad_norm": 1.1063755750656128, "learning_rate": 2.6102183973982086e-06, "loss": 0.2231, "step": 45480 }, { "epoch": 0.8274068280616236, "grad_norm": 0.9973794221878052, "learning_rate": 2.6089680214518674e-06, "loss": 0.2218, "step": 45490 }, { "epoch": 0.8275887156914459, "grad_norm": 1.010552167892456, "learning_rate": 2.6077188223111583e-06, "loss": 0.2224, "step": 45500 }, { "epoch": 0.8277706033212682, "grad_norm": 1.0175256729125977, "learning_rate": 2.6064708004095945e-06, "loss": 0.2229, "step": 45510 }, { "epoch": 0.8279524909510905, "grad_norm": 1.0669503211975098, "learning_rate": 2.6052239561802856e-06, "loss": 0.2181, "step": 45520 }, { "epoch": 0.8281343785809128, "grad_norm": 0.9606948494911194, "learning_rate": 2.6039782900559285e-06, "loss": 0.2186, "step": 45530 }, { "epoch": 0.828316266210735, "grad_norm": 0.925467848777771, "learning_rate": 2.602733802468813e-06, "loss": 0.2201, "step": 45540 }, { "epoch": 0.8284981538405572, "grad_norm": 0.992935061454773, "learning_rate": 2.601490493850819e-06, "loss": 0.2245, "step": 45550 }, { "epoch": 0.8286800414703795, "grad_norm": 1.0629651546478271, "learning_rate": 2.6002483646334165e-06, "loss": 0.2257, "step": 45560 }, { "epoch": 0.8288619291002018, "grad_norm": 1.0241538286209106, "learning_rate": 2.59900741524767e-06, "loss": 0.2172, "step": 45570 }, { "epoch": 0.8290438167300241, "grad_norm": 0.9978601336479187, "learning_rate": 2.59776764612423e-06, "loss": 0.2237, "step": 45580 }, { "epoch": 0.8292257043598464, "grad_norm": 0.9450433850288391, "learning_rate": 2.5965290576933412e-06, "loss": 0.2137, "step": 45590 }, { "epoch": 0.8294075919896687, "grad_norm": 0.9804511070251465, "learning_rate": 2.5952916503848337e-06, "loss": 0.2222, "step": 45600 }, { "epoch": 0.829589479619491, "grad_norm": 0.9059440493583679, "learning_rate": 2.5940554246281333e-06, "loss": 0.2289, "step": 45610 }, { "epoch": 0.8297713672493133, "grad_norm": 1.1093337535858154, "learning_rate": 2.592820380852253e-06, "loss": 0.2311, "step": 45620 }, { "epoch": 0.8299532548791356, "grad_norm": 0.9798998832702637, "learning_rate": 2.591586519485795e-06, "loss": 0.2156, "step": 45630 }, { "epoch": 0.8301351425089579, "grad_norm": 1.1356996297836304, "learning_rate": 2.5903538409569516e-06, "loss": 0.2235, "step": 45640 }, { "epoch": 0.8303170301387802, "grad_norm": 0.959176242351532, "learning_rate": 2.5891223456935057e-06, "loss": 0.2205, "step": 45650 }, { "epoch": 0.8304989177686025, "grad_norm": 1.0896254777908325, "learning_rate": 2.5878920341228286e-06, "loss": 0.2208, "step": 45660 }, { "epoch": 0.8306808053984248, "grad_norm": 0.9666463136672974, "learning_rate": 2.586662906671882e-06, "loss": 0.2222, "step": 45670 }, { "epoch": 0.8308626930282471, "grad_norm": 1.0942425727844238, "learning_rate": 2.585434963767214e-06, "loss": 0.223, "step": 45680 }, { "epoch": 0.8310445806580694, "grad_norm": 1.0032063722610474, "learning_rate": 2.584208205834964e-06, "loss": 0.2166, "step": 45690 }, { "epoch": 0.8312264682878917, "grad_norm": 0.9055570960044861, "learning_rate": 2.5829826333008603e-06, "loss": 0.219, "step": 45700 }, { "epoch": 0.831408355917714, "grad_norm": 1.0769062042236328, "learning_rate": 2.581758246590218e-06, "loss": 0.2202, "step": 45710 }, { "epoch": 0.8315902435475363, "grad_norm": 0.9754552245140076, "learning_rate": 2.5805350461279435e-06, "loss": 0.2194, "step": 45720 }, { "epoch": 0.8317721311773586, "grad_norm": 1.1223217248916626, "learning_rate": 2.579313032338527e-06, "loss": 0.2244, "step": 45730 }, { "epoch": 0.8319540188071809, "grad_norm": 1.0862840414047241, "learning_rate": 2.5780922056460523e-06, "loss": 0.2202, "step": 45740 }, { "epoch": 0.8321359064370032, "grad_norm": 0.8867900371551514, "learning_rate": 2.5768725664741856e-06, "loss": 0.2214, "step": 45750 }, { "epoch": 0.8323177940668255, "grad_norm": 1.0075997114181519, "learning_rate": 2.5756541152461866e-06, "loss": 0.2232, "step": 45760 }, { "epoch": 0.8324996816966478, "grad_norm": 1.0649938583374023, "learning_rate": 2.5744368523848995e-06, "loss": 0.2222, "step": 45770 }, { "epoch": 0.8326815693264701, "grad_norm": 1.0437426567077637, "learning_rate": 2.5732207783127567e-06, "loss": 0.2236, "step": 45780 }, { "epoch": 0.8328634569562924, "grad_norm": 0.9160332083702087, "learning_rate": 2.572005893451778e-06, "loss": 0.2204, "step": 45790 }, { "epoch": 0.8330453445861147, "grad_norm": 0.9410653710365295, "learning_rate": 2.570792198223569e-06, "loss": 0.2174, "step": 45800 }, { "epoch": 0.833227232215937, "grad_norm": 1.0612871646881104, "learning_rate": 2.569579693049327e-06, "loss": 0.222, "step": 45810 }, { "epoch": 0.8334091198457593, "grad_norm": 1.0225571393966675, "learning_rate": 2.568368378349831e-06, "loss": 0.2274, "step": 45820 }, { "epoch": 0.8335910074755816, "grad_norm": 1.1378533840179443, "learning_rate": 2.56715825454545e-06, "loss": 0.2217, "step": 45830 }, { "epoch": 0.8337728951054039, "grad_norm": 1.010451078414917, "learning_rate": 2.5659493220561395e-06, "loss": 0.2271, "step": 45840 }, { "epoch": 0.8339547827352262, "grad_norm": 0.9431526064872742, "learning_rate": 2.564741581301438e-06, "loss": 0.2216, "step": 45850 }, { "epoch": 0.8341366703650485, "grad_norm": 0.9566556215286255, "learning_rate": 2.5635350327004777e-06, "loss": 0.2222, "step": 45860 }, { "epoch": 0.8343185579948708, "grad_norm": 0.9817479848861694, "learning_rate": 2.562329676671971e-06, "loss": 0.2252, "step": 45870 }, { "epoch": 0.834500445624693, "grad_norm": 0.9123132228851318, "learning_rate": 2.5611255136342172e-06, "loss": 0.2209, "step": 45880 }, { "epoch": 0.8346823332545154, "grad_norm": 1.1162023544311523, "learning_rate": 2.5599225440051032e-06, "loss": 0.2209, "step": 45890 }, { "epoch": 0.8348642208843376, "grad_norm": 1.034985899925232, "learning_rate": 2.558720768202101e-06, "loss": 0.2217, "step": 45900 }, { "epoch": 0.83504610851416, "grad_norm": 0.9749163389205933, "learning_rate": 2.55752018664227e-06, "loss": 0.2259, "step": 45910 }, { "epoch": 0.8352279961439822, "grad_norm": 1.4150331020355225, "learning_rate": 2.5563207997422517e-06, "loss": 0.2221, "step": 45920 }, { "epoch": 0.8354098837738045, "grad_norm": 0.9566100835800171, "learning_rate": 2.5551226079182763e-06, "loss": 0.2229, "step": 45930 }, { "epoch": 0.8355917714036268, "grad_norm": 0.9062387943267822, "learning_rate": 2.5539256115861566e-06, "loss": 0.2245, "step": 45940 }, { "epoch": 0.8357736590334491, "grad_norm": 1.0234832763671875, "learning_rate": 2.552729811161292e-06, "loss": 0.2248, "step": 45950 }, { "epoch": 0.8359555466632714, "grad_norm": 0.9478579163551331, "learning_rate": 2.551535207058668e-06, "loss": 0.2209, "step": 45960 }, { "epoch": 0.8361374342930937, "grad_norm": 1.041063904762268, "learning_rate": 2.5503417996928527e-06, "loss": 0.2195, "step": 45970 }, { "epoch": 0.836319321922916, "grad_norm": 1.033801794052124, "learning_rate": 2.5491495894779994e-06, "loss": 0.2184, "step": 45980 }, { "epoch": 0.8365012095527383, "grad_norm": 1.0131111145019531, "learning_rate": 2.547958576827847e-06, "loss": 0.2192, "step": 45990 }, { "epoch": 0.8366830971825606, "grad_norm": 1.1148781776428223, "learning_rate": 2.546768762155718e-06, "loss": 0.219, "step": 46000 }, { "epoch": 0.8368649848123829, "grad_norm": 1.1122589111328125, "learning_rate": 2.54558014587452e-06, "loss": 0.2254, "step": 46010 }, { "epoch": 0.8370468724422052, "grad_norm": 0.988564133644104, "learning_rate": 2.544392728396743e-06, "loss": 0.2247, "step": 46020 }, { "epoch": 0.8372287600720275, "grad_norm": 1.0775463581085205, "learning_rate": 2.5432065101344616e-06, "loss": 0.2194, "step": 46030 }, { "epoch": 0.8374106477018498, "grad_norm": 1.1823444366455078, "learning_rate": 2.542021491499337e-06, "loss": 0.2193, "step": 46040 }, { "epoch": 0.8375925353316721, "grad_norm": 0.9716647863388062, "learning_rate": 2.5408376729026086e-06, "loss": 0.217, "step": 46050 }, { "epoch": 0.8377744229614944, "grad_norm": 1.0075008869171143, "learning_rate": 2.539655054755105e-06, "loss": 0.2187, "step": 46060 }, { "epoch": 0.8379563105913167, "grad_norm": 0.9386841654777527, "learning_rate": 2.538473637467235e-06, "loss": 0.2199, "step": 46070 }, { "epoch": 0.838138198221139, "grad_norm": 1.035090684890747, "learning_rate": 2.5372934214489914e-06, "loss": 0.2176, "step": 46080 }, { "epoch": 0.8383200858509613, "grad_norm": 1.0509997606277466, "learning_rate": 2.5361144071099496e-06, "loss": 0.2209, "step": 46090 }, { "epoch": 0.8385019734807836, "grad_norm": 1.0427205562591553, "learning_rate": 2.5349365948592675e-06, "loss": 0.2297, "step": 46100 }, { "epoch": 0.8386838611106059, "grad_norm": 1.0533828735351562, "learning_rate": 2.533759985105689e-06, "loss": 0.2249, "step": 46110 }, { "epoch": 0.8388657487404282, "grad_norm": 1.1118217706680298, "learning_rate": 2.5325845782575377e-06, "loss": 0.224, "step": 46120 }, { "epoch": 0.8390476363702505, "grad_norm": 1.0336945056915283, "learning_rate": 2.5314103747227204e-06, "loss": 0.2191, "step": 46130 }, { "epoch": 0.8392295240000728, "grad_norm": 0.9509017467498779, "learning_rate": 2.5302373749087255e-06, "loss": 0.2268, "step": 46140 }, { "epoch": 0.8394114116298951, "grad_norm": 1.0178346633911133, "learning_rate": 2.529065579222625e-06, "loss": 0.2176, "step": 46150 }, { "epoch": 0.8395932992597174, "grad_norm": 0.9817126989364624, "learning_rate": 2.5278949880710733e-06, "loss": 0.2247, "step": 46160 }, { "epoch": 0.8397751868895397, "grad_norm": 0.9314928650856018, "learning_rate": 2.526725601860306e-06, "loss": 0.2233, "step": 46170 }, { "epoch": 0.839957074519362, "grad_norm": 0.9503644704818726, "learning_rate": 2.5255574209961405e-06, "loss": 0.2207, "step": 46180 }, { "epoch": 0.8401389621491843, "grad_norm": 0.9812507033348083, "learning_rate": 2.5243904458839745e-06, "loss": 0.2256, "step": 46190 }, { "epoch": 0.8403208497790066, "grad_norm": 0.9053927063941956, "learning_rate": 2.523224676928792e-06, "loss": 0.2195, "step": 46200 }, { "epoch": 0.8405027374088289, "grad_norm": 1.0274786949157715, "learning_rate": 2.522060114535153e-06, "loss": 0.219, "step": 46210 }, { "epoch": 0.8406846250386512, "grad_norm": 1.0019451379776, "learning_rate": 2.5208967591072014e-06, "loss": 0.2201, "step": 46220 }, { "epoch": 0.8408665126684735, "grad_norm": 1.1457656621932983, "learning_rate": 2.5197346110486625e-06, "loss": 0.2212, "step": 46230 }, { "epoch": 0.8410484002982958, "grad_norm": 1.1107145547866821, "learning_rate": 2.51857367076284e-06, "loss": 0.2213, "step": 46240 }, { "epoch": 0.841230287928118, "grad_norm": 1.080235481262207, "learning_rate": 2.517413938652623e-06, "loss": 0.2199, "step": 46250 }, { "epoch": 0.8414121755579403, "grad_norm": 1.0944706201553345, "learning_rate": 2.516255415120478e-06, "loss": 0.2199, "step": 46260 }, { "epoch": 0.8415940631877626, "grad_norm": 0.9848400354385376, "learning_rate": 2.5150981005684516e-06, "loss": 0.2243, "step": 46270 }, { "epoch": 0.8417759508175849, "grad_norm": 1.0978325605392456, "learning_rate": 2.5139419953981726e-06, "loss": 0.2251, "step": 46280 }, { "epoch": 0.8419578384474072, "grad_norm": 1.2674081325531006, "learning_rate": 2.5127871000108493e-06, "loss": 0.2212, "step": 46290 }, { "epoch": 0.8421397260772295, "grad_norm": 1.0315848588943481, "learning_rate": 2.511633414807272e-06, "loss": 0.2238, "step": 46300 }, { "epoch": 0.8423216137070518, "grad_norm": 1.1228336095809937, "learning_rate": 2.510480940187808e-06, "loss": 0.2198, "step": 46310 }, { "epoch": 0.8425035013368741, "grad_norm": 0.9244106411933899, "learning_rate": 2.5093296765524073e-06, "loss": 0.2251, "step": 46320 }, { "epoch": 0.8426853889666964, "grad_norm": 0.9634957313537598, "learning_rate": 2.508179624300596e-06, "loss": 0.2231, "step": 46330 }, { "epoch": 0.8428672765965187, "grad_norm": 1.0214321613311768, "learning_rate": 2.5070307838314847e-06, "loss": 0.2228, "step": 46340 }, { "epoch": 0.843049164226341, "grad_norm": 1.1112011671066284, "learning_rate": 2.505883155543758e-06, "loss": 0.2221, "step": 46350 }, { "epoch": 0.8432310518561633, "grad_norm": 0.8715990781784058, "learning_rate": 2.504736739835685e-06, "loss": 0.2215, "step": 46360 }, { "epoch": 0.8434129394859856, "grad_norm": 1.0619350671768188, "learning_rate": 2.5035915371051123e-06, "loss": 0.2183, "step": 46370 }, { "epoch": 0.8435948271158078, "grad_norm": 0.9898329973220825, "learning_rate": 2.5024475477494625e-06, "loss": 0.2214, "step": 46380 }, { "epoch": 0.8437767147456301, "grad_norm": 0.9968706965446472, "learning_rate": 2.5013047721657402e-06, "loss": 0.2182, "step": 46390 }, { "epoch": 0.8439586023754524, "grad_norm": 1.057762861251831, "learning_rate": 2.5001632107505287e-06, "loss": 0.2192, "step": 46400 }, { "epoch": 0.8441404900052747, "grad_norm": 1.0654566287994385, "learning_rate": 2.49902286389999e-06, "loss": 0.2205, "step": 46410 }, { "epoch": 0.844322377635097, "grad_norm": 0.9706518054008484, "learning_rate": 2.497883732009863e-06, "loss": 0.2174, "step": 46420 }, { "epoch": 0.8445042652649193, "grad_norm": 0.9498987793922424, "learning_rate": 2.4967458154754663e-06, "loss": 0.2177, "step": 46430 }, { "epoch": 0.8446861528947416, "grad_norm": 1.028548240661621, "learning_rate": 2.4956091146916953e-06, "loss": 0.2194, "step": 46440 }, { "epoch": 0.8448680405245639, "grad_norm": 1.0364956855773926, "learning_rate": 2.4944736300530267e-06, "loss": 0.2232, "step": 46450 }, { "epoch": 0.8450499281543862, "grad_norm": 1.0209287405014038, "learning_rate": 2.4933393619535116e-06, "loss": 0.2168, "step": 46460 }, { "epoch": 0.8452318157842085, "grad_norm": 0.9926218390464783, "learning_rate": 2.492206310786781e-06, "loss": 0.2206, "step": 46470 }, { "epoch": 0.8454137034140308, "grad_norm": 0.9134261012077332, "learning_rate": 2.491074476946042e-06, "loss": 0.2248, "step": 46480 }, { "epoch": 0.8455955910438531, "grad_norm": 0.9745477437973022, "learning_rate": 2.4899438608240804e-06, "loss": 0.2197, "step": 46490 }, { "epoch": 0.8457774786736754, "grad_norm": 1.042567253112793, "learning_rate": 2.48881446281326e-06, "loss": 0.2169, "step": 46500 }, { "epoch": 0.8459593663034977, "grad_norm": 1.0442872047424316, "learning_rate": 2.4876862833055204e-06, "loss": 0.216, "step": 46510 }, { "epoch": 0.84614125393332, "grad_norm": 0.8939103484153748, "learning_rate": 2.4865593226923796e-06, "loss": 0.2175, "step": 46520 }, { "epoch": 0.8463231415631423, "grad_norm": 0.9720823764801025, "learning_rate": 2.4854335813649307e-06, "loss": 0.2228, "step": 46530 }, { "epoch": 0.8465050291929646, "grad_norm": 1.0844451189041138, "learning_rate": 2.4843090597138464e-06, "loss": 0.2256, "step": 46540 }, { "epoch": 0.8466869168227869, "grad_norm": 0.9710667729377747, "learning_rate": 2.483185758129374e-06, "loss": 0.2237, "step": 46550 }, { "epoch": 0.8468688044526091, "grad_norm": 1.01625657081604, "learning_rate": 2.4820636770013385e-06, "loss": 0.2182, "step": 46560 }, { "epoch": 0.8470506920824314, "grad_norm": 1.0612510442733765, "learning_rate": 2.480942816719141e-06, "loss": 0.2237, "step": 46570 }, { "epoch": 0.8472325797122537, "grad_norm": 1.0833090543746948, "learning_rate": 2.4798231776717565e-06, "loss": 0.212, "step": 46580 }, { "epoch": 0.847414467342076, "grad_norm": 1.002211570739746, "learning_rate": 2.4787047602477415e-06, "loss": 0.2237, "step": 46590 }, { "epoch": 0.8475963549718983, "grad_norm": 1.0371882915496826, "learning_rate": 2.4775875648352253e-06, "loss": 0.2242, "step": 46600 }, { "epoch": 0.8477782426017206, "grad_norm": 0.8940765857696533, "learning_rate": 2.476471591821913e-06, "loss": 0.219, "step": 46610 }, { "epoch": 0.8479601302315429, "grad_norm": 1.1127417087554932, "learning_rate": 2.4753568415950848e-06, "loss": 0.2208, "step": 46620 }, { "epoch": 0.8481420178613652, "grad_norm": 1.0407006740570068, "learning_rate": 2.4742433145415994e-06, "loss": 0.2207, "step": 46630 }, { "epoch": 0.8483239054911875, "grad_norm": 0.9077007174491882, "learning_rate": 2.4731310110478875e-06, "loss": 0.2191, "step": 46640 }, { "epoch": 0.8485057931210098, "grad_norm": 0.9852501153945923, "learning_rate": 2.4720199314999584e-06, "loss": 0.2213, "step": 46650 }, { "epoch": 0.8486876807508321, "grad_norm": 0.9704689383506775, "learning_rate": 2.4709100762833953e-06, "loss": 0.2221, "step": 46660 }, { "epoch": 0.8488695683806544, "grad_norm": 0.9281551837921143, "learning_rate": 2.4698014457833557e-06, "loss": 0.2221, "step": 46670 }, { "epoch": 0.8490514560104767, "grad_norm": 1.178658366203308, "learning_rate": 2.4686940403845726e-06, "loss": 0.2229, "step": 46680 }, { "epoch": 0.849233343640299, "grad_norm": 0.9627045392990112, "learning_rate": 2.467587860471354e-06, "loss": 0.219, "step": 46690 }, { "epoch": 0.8494152312701213, "grad_norm": 0.9633352160453796, "learning_rate": 2.466482906427584e-06, "loss": 0.2146, "step": 46700 }, { "epoch": 0.8495971188999436, "grad_norm": 0.9678441882133484, "learning_rate": 2.4653791786367183e-06, "loss": 0.2237, "step": 46710 }, { "epoch": 0.8497790065297659, "grad_norm": 1.0637812614440918, "learning_rate": 2.4642766774817905e-06, "loss": 0.2219, "step": 46720 }, { "epoch": 0.8499608941595882, "grad_norm": 0.9873220324516296, "learning_rate": 2.4631754033454037e-06, "loss": 0.2162, "step": 46730 }, { "epoch": 0.8501427817894105, "grad_norm": 1.1517233848571777, "learning_rate": 2.462075356609742e-06, "loss": 0.2279, "step": 46740 }, { "epoch": 0.8503246694192328, "grad_norm": 0.902126669883728, "learning_rate": 2.4609765376565574e-06, "loss": 0.2178, "step": 46750 }, { "epoch": 0.8505065570490551, "grad_norm": 0.9616915583610535, "learning_rate": 2.459878946867179e-06, "loss": 0.218, "step": 46760 }, { "epoch": 0.8506884446788774, "grad_norm": 0.9963560104370117, "learning_rate": 2.458782584622508e-06, "loss": 0.2229, "step": 46770 }, { "epoch": 0.8508703323086997, "grad_norm": 0.9801801443099976, "learning_rate": 2.45768745130302e-06, "loss": 0.2253, "step": 46780 }, { "epoch": 0.851052219938522, "grad_norm": 0.9284557104110718, "learning_rate": 2.456593547288766e-06, "loss": 0.2178, "step": 46790 }, { "epoch": 0.8512341075683443, "grad_norm": 1.1070151329040527, "learning_rate": 2.4555008729593675e-06, "loss": 0.2265, "step": 46800 }, { "epoch": 0.8514159951981666, "grad_norm": 1.132606029510498, "learning_rate": 2.45440942869402e-06, "loss": 0.2193, "step": 46810 }, { "epoch": 0.8515978828279889, "grad_norm": 0.9556341171264648, "learning_rate": 2.4533192148714926e-06, "loss": 0.2189, "step": 46820 }, { "epoch": 0.8517797704578112, "grad_norm": 1.0153870582580566, "learning_rate": 2.452230231870127e-06, "loss": 0.226, "step": 46830 }, { "epoch": 0.8519616580876335, "grad_norm": 1.1174598932266235, "learning_rate": 2.4511424800678394e-06, "loss": 0.2219, "step": 46840 }, { "epoch": 0.8521435457174558, "grad_norm": 0.8994749188423157, "learning_rate": 2.4500559598421156e-06, "loss": 0.2199, "step": 46850 }, { "epoch": 0.8523254333472781, "grad_norm": 1.1184794902801514, "learning_rate": 2.448970671570017e-06, "loss": 0.2233, "step": 46860 }, { "epoch": 0.8525073209771004, "grad_norm": 1.1004436016082764, "learning_rate": 2.4478866156281767e-06, "loss": 0.2245, "step": 46870 }, { "epoch": 0.8526892086069227, "grad_norm": 1.033408522605896, "learning_rate": 2.4468037923927986e-06, "loss": 0.2189, "step": 46880 }, { "epoch": 0.852871096236745, "grad_norm": 1.0278949737548828, "learning_rate": 2.445722202239659e-06, "loss": 0.223, "step": 46890 }, { "epoch": 0.8530529838665672, "grad_norm": 0.932388424873352, "learning_rate": 2.4446418455441097e-06, "loss": 0.2204, "step": 46900 }, { "epoch": 0.8532348714963895, "grad_norm": 1.0796031951904297, "learning_rate": 2.4435627226810707e-06, "loss": 0.2177, "step": 46910 }, { "epoch": 0.8534167591262118, "grad_norm": 0.9813504815101624, "learning_rate": 2.4424848340250346e-06, "loss": 0.2202, "step": 46920 }, { "epoch": 0.8535986467560341, "grad_norm": 0.9732363224029541, "learning_rate": 2.441408179950066e-06, "loss": 0.2203, "step": 46930 }, { "epoch": 0.8537805343858564, "grad_norm": 1.018663763999939, "learning_rate": 2.440332760829802e-06, "loss": 0.2282, "step": 46940 }, { "epoch": 0.8539624220156787, "grad_norm": 1.1063417196273804, "learning_rate": 2.4392585770374504e-06, "loss": 0.225, "step": 46950 }, { "epoch": 0.854144309645501, "grad_norm": 0.9420072436332703, "learning_rate": 2.4381856289457895e-06, "loss": 0.2201, "step": 46960 }, { "epoch": 0.8543261972753233, "grad_norm": 1.0746166706085205, "learning_rate": 2.43711391692717e-06, "loss": 0.2218, "step": 46970 }, { "epoch": 0.8545080849051456, "grad_norm": 1.065211534500122, "learning_rate": 2.4360434413535117e-06, "loss": 0.2193, "step": 46980 }, { "epoch": 0.8546899725349679, "grad_norm": 1.04986572265625, "learning_rate": 2.4349742025963093e-06, "loss": 0.2197, "step": 46990 }, { "epoch": 0.8548718601647902, "grad_norm": 0.9414594173431396, "learning_rate": 2.4339062010266236e-06, "loss": 0.2195, "step": 47000 }, { "epoch": 0.8550537477946125, "grad_norm": 0.9357734322547913, "learning_rate": 2.432839437015089e-06, "loss": 0.2211, "step": 47010 }, { "epoch": 0.8552356354244348, "grad_norm": 1.0674961805343628, "learning_rate": 2.4317739109319092e-06, "loss": 0.2198, "step": 47020 }, { "epoch": 0.8554175230542571, "grad_norm": 0.9659944176673889, "learning_rate": 2.4307096231468587e-06, "loss": 0.2175, "step": 47030 }, { "epoch": 0.8555994106840794, "grad_norm": 1.116426706314087, "learning_rate": 2.4296465740292834e-06, "loss": 0.2219, "step": 47040 }, { "epoch": 0.8557812983139017, "grad_norm": 0.9752377271652222, "learning_rate": 2.4285847639480965e-06, "loss": 0.2147, "step": 47050 }, { "epoch": 0.855963185943724, "grad_norm": 1.0829821825027466, "learning_rate": 2.427524193271784e-06, "loss": 0.2207, "step": 47060 }, { "epoch": 0.8561450735735463, "grad_norm": 0.9269568920135498, "learning_rate": 2.4264648623684007e-06, "loss": 0.216, "step": 47070 }, { "epoch": 0.8563269612033686, "grad_norm": 1.2749253511428833, "learning_rate": 2.4254067716055695e-06, "loss": 0.2242, "step": 47080 }, { "epoch": 0.8565088488331909, "grad_norm": 1.0229692459106445, "learning_rate": 2.424349921350487e-06, "loss": 0.2212, "step": 47090 }, { "epoch": 0.8566907364630132, "grad_norm": 1.0770928859710693, "learning_rate": 2.4232943119699158e-06, "loss": 0.219, "step": 47100 }, { "epoch": 0.8568726240928355, "grad_norm": 1.0631471872329712, "learning_rate": 2.422239943830188e-06, "loss": 0.2228, "step": 47110 }, { "epoch": 0.8570545117226578, "grad_norm": 1.1609134674072266, "learning_rate": 2.421186817297207e-06, "loss": 0.2195, "step": 47120 }, { "epoch": 0.8572363993524801, "grad_norm": 0.9934793710708618, "learning_rate": 2.420134932736444e-06, "loss": 0.2218, "step": 47130 }, { "epoch": 0.8574182869823024, "grad_norm": 1.0360925197601318, "learning_rate": 2.4190842905129393e-06, "loss": 0.2213, "step": 47140 }, { "epoch": 0.8576001746121247, "grad_norm": 0.9008083343505859, "learning_rate": 2.4180348909913027e-06, "loss": 0.2193, "step": 47150 }, { "epoch": 0.857782062241947, "grad_norm": 1.0633419752120972, "learning_rate": 2.4169867345357113e-06, "loss": 0.2195, "step": 47160 }, { "epoch": 0.8579639498717693, "grad_norm": 1.0587279796600342, "learning_rate": 2.415939821509913e-06, "loss": 0.2209, "step": 47170 }, { "epoch": 0.8581458375015916, "grad_norm": 0.978306770324707, "learning_rate": 2.414894152277221e-06, "loss": 0.2203, "step": 47180 }, { "epoch": 0.8583277251314139, "grad_norm": 0.9527724981307983, "learning_rate": 2.41384972720052e-06, "loss": 0.2252, "step": 47190 }, { "epoch": 0.8585096127612362, "grad_norm": 1.0055979490280151, "learning_rate": 2.412806546642262e-06, "loss": 0.2213, "step": 47200 }, { "epoch": 0.8586915003910583, "grad_norm": 1.1580499410629272, "learning_rate": 2.411764610964467e-06, "loss": 0.2223, "step": 47210 }, { "epoch": 0.8588733880208806, "grad_norm": 0.9637002348899841, "learning_rate": 2.4107239205287214e-06, "loss": 0.2197, "step": 47220 }, { "epoch": 0.8590552756507029, "grad_norm": 1.044795274734497, "learning_rate": 2.409684475696181e-06, "loss": 0.2202, "step": 47230 }, { "epoch": 0.8592371632805252, "grad_norm": 0.8444093465805054, "learning_rate": 2.4086462768275705e-06, "loss": 0.2173, "step": 47240 }, { "epoch": 0.8594190509103475, "grad_norm": 1.0278189182281494, "learning_rate": 2.40760932428318e-06, "loss": 0.2165, "step": 47250 }, { "epoch": 0.8596009385401698, "grad_norm": 1.047140121459961, "learning_rate": 2.4065736184228683e-06, "loss": 0.2181, "step": 47260 }, { "epoch": 0.8597828261699921, "grad_norm": 0.9679595232009888, "learning_rate": 2.4055391596060613e-06, "loss": 0.2225, "step": 47270 }, { "epoch": 0.8599647137998144, "grad_norm": 1.1101776361465454, "learning_rate": 2.404505948191751e-06, "loss": 0.2225, "step": 47280 }, { "epoch": 0.8601466014296367, "grad_norm": 1.0164434909820557, "learning_rate": 2.403473984538499e-06, "loss": 0.2285, "step": 47290 }, { "epoch": 0.860328489059459, "grad_norm": 0.9565553665161133, "learning_rate": 2.402443269004432e-06, "loss": 0.2173, "step": 47300 }, { "epoch": 0.8605103766892813, "grad_norm": 0.9449731707572937, "learning_rate": 2.401413801947244e-06, "loss": 0.2219, "step": 47310 }, { "epoch": 0.8606922643191036, "grad_norm": 0.9532617330551147, "learning_rate": 2.4003855837241944e-06, "loss": 0.2213, "step": 47320 }, { "epoch": 0.8608741519489259, "grad_norm": 0.8956452012062073, "learning_rate": 2.3993586146921134e-06, "loss": 0.2162, "step": 47330 }, { "epoch": 0.8610560395787482, "grad_norm": 1.003064513206482, "learning_rate": 2.398332895207392e-06, "loss": 0.2152, "step": 47340 }, { "epoch": 0.8612379272085705, "grad_norm": 1.0600762367248535, "learning_rate": 2.3973084256259928e-06, "loss": 0.2202, "step": 47350 }, { "epoch": 0.8614198148383928, "grad_norm": 1.0959163904190063, "learning_rate": 2.396285206303441e-06, "loss": 0.2229, "step": 47360 }, { "epoch": 0.8616017024682151, "grad_norm": 1.0543880462646484, "learning_rate": 2.3952632375948277e-06, "loss": 0.2252, "step": 47370 }, { "epoch": 0.8617835900980374, "grad_norm": 1.1258389949798584, "learning_rate": 2.3942425198548147e-06, "loss": 0.2206, "step": 47380 }, { "epoch": 0.8619654777278597, "grad_norm": 1.0473010540008545, "learning_rate": 2.3932230534376245e-06, "loss": 0.2211, "step": 47390 }, { "epoch": 0.862147365357682, "grad_norm": 0.9824640154838562, "learning_rate": 2.3922048386970477e-06, "loss": 0.2182, "step": 47400 }, { "epoch": 0.8623292529875043, "grad_norm": 1.0761005878448486, "learning_rate": 2.3911878759864383e-06, "loss": 0.2224, "step": 47410 }, { "epoch": 0.8625111406173266, "grad_norm": 0.9958451390266418, "learning_rate": 2.390172165658721e-06, "loss": 0.2182, "step": 47420 }, { "epoch": 0.8626930282471489, "grad_norm": 0.9025494456291199, "learning_rate": 2.3891577080663797e-06, "loss": 0.2138, "step": 47430 }, { "epoch": 0.8628749158769712, "grad_norm": 1.0303434133529663, "learning_rate": 2.3881445035614683e-06, "loss": 0.2237, "step": 47440 }, { "epoch": 0.8630568035067935, "grad_norm": 1.003402590751648, "learning_rate": 2.387132552495603e-06, "loss": 0.2257, "step": 47450 }, { "epoch": 0.8632386911366158, "grad_norm": 0.9320489764213562, "learning_rate": 2.3861218552199657e-06, "loss": 0.225, "step": 47460 }, { "epoch": 0.8634205787664381, "grad_norm": 0.9709509015083313, "learning_rate": 2.3851124120853042e-06, "loss": 0.2224, "step": 47470 }, { "epoch": 0.8636024663962604, "grad_norm": 1.0121980905532837, "learning_rate": 2.3841042234419284e-06, "loss": 0.221, "step": 47480 }, { "epoch": 0.8637843540260827, "grad_norm": 1.079336166381836, "learning_rate": 2.3830972896397165e-06, "loss": 0.2234, "step": 47490 }, { "epoch": 0.863966241655905, "grad_norm": 0.9938313364982605, "learning_rate": 2.382091611028109e-06, "loss": 0.2199, "step": 47500 }, { "epoch": 0.8641481292857273, "grad_norm": 1.021723747253418, "learning_rate": 2.381087187956111e-06, "loss": 0.2206, "step": 47510 }, { "epoch": 0.8643300169155496, "grad_norm": 0.9989994168281555, "learning_rate": 2.380084020772292e-06, "loss": 0.22, "step": 47520 }, { "epoch": 0.8645119045453719, "grad_norm": 0.9052411913871765, "learning_rate": 2.379082109824786e-06, "loss": 0.2186, "step": 47530 }, { "epoch": 0.8646937921751942, "grad_norm": 1.1513855457305908, "learning_rate": 2.3780814554612915e-06, "loss": 0.2226, "step": 47540 }, { "epoch": 0.8648756798050165, "grad_norm": 1.1340218782424927, "learning_rate": 2.3770820580290693e-06, "loss": 0.2219, "step": 47550 }, { "epoch": 0.8650575674348387, "grad_norm": 1.036847472190857, "learning_rate": 2.3760839178749452e-06, "loss": 0.2246, "step": 47560 }, { "epoch": 0.865239455064661, "grad_norm": 0.9943318367004395, "learning_rate": 2.375087035345308e-06, "loss": 0.2231, "step": 47570 }, { "epoch": 0.8654213426944833, "grad_norm": 1.0322109460830688, "learning_rate": 2.374091410786111e-06, "loss": 0.2171, "step": 47580 }, { "epoch": 0.8656032303243056, "grad_norm": 0.9904865026473999, "learning_rate": 2.3730970445428715e-06, "loss": 0.2198, "step": 47590 }, { "epoch": 0.8657851179541279, "grad_norm": 0.9989230632781982, "learning_rate": 2.372103936960667e-06, "loss": 0.222, "step": 47600 }, { "epoch": 0.8659670055839502, "grad_norm": 1.001457929611206, "learning_rate": 2.3711120883841416e-06, "loss": 0.2163, "step": 47610 }, { "epoch": 0.8661488932137725, "grad_norm": 0.9901447296142578, "learning_rate": 2.3701214991574996e-06, "loss": 0.2184, "step": 47620 }, { "epoch": 0.8663307808435948, "grad_norm": 0.9393513798713684, "learning_rate": 2.369132169624512e-06, "loss": 0.2245, "step": 47630 }, { "epoch": 0.8665126684734171, "grad_norm": 0.9990282654762268, "learning_rate": 2.368144100128509e-06, "loss": 0.2136, "step": 47640 }, { "epoch": 0.8666945561032394, "grad_norm": 0.9857451319694519, "learning_rate": 2.3671572910123855e-06, "loss": 0.2191, "step": 47650 }, { "epoch": 0.8668764437330617, "grad_norm": 1.039778470993042, "learning_rate": 2.366171742618599e-06, "loss": 0.2177, "step": 47660 }, { "epoch": 0.867058331362884, "grad_norm": 1.0830647945404053, "learning_rate": 2.3651874552891666e-06, "loss": 0.2209, "step": 47670 }, { "epoch": 0.8672402189927063, "grad_norm": 0.9862698316574097, "learning_rate": 2.3642044293656718e-06, "loss": 0.2213, "step": 47680 }, { "epoch": 0.8674221066225286, "grad_norm": 0.9928057789802551, "learning_rate": 2.36322266518926e-06, "loss": 0.2148, "step": 47690 }, { "epoch": 0.8676039942523509, "grad_norm": 1.098871111869812, "learning_rate": 2.362242163100635e-06, "loss": 0.2171, "step": 47700 }, { "epoch": 0.8677858818821732, "grad_norm": 1.205004334449768, "learning_rate": 2.3612629234400667e-06, "loss": 0.2232, "step": 47710 }, { "epoch": 0.8679677695119955, "grad_norm": 1.0479062795639038, "learning_rate": 2.3602849465473835e-06, "loss": 0.2214, "step": 47720 }, { "epoch": 0.8681496571418178, "grad_norm": 1.1000932455062866, "learning_rate": 2.3593082327619796e-06, "loss": 0.2222, "step": 47730 }, { "epoch": 0.8683315447716401, "grad_norm": 0.9922612905502319, "learning_rate": 2.358332782422807e-06, "loss": 0.2238, "step": 47740 }, { "epoch": 0.8685134324014624, "grad_norm": 1.0232930183410645, "learning_rate": 2.3573585958683806e-06, "loss": 0.2141, "step": 47750 }, { "epoch": 0.8686953200312847, "grad_norm": 1.0170302391052246, "learning_rate": 2.3563856734367772e-06, "loss": 0.2238, "step": 47760 }, { "epoch": 0.868877207661107, "grad_norm": 0.9266634583473206, "learning_rate": 2.355414015465635e-06, "loss": 0.2233, "step": 47770 }, { "epoch": 0.8690590952909293, "grad_norm": 1.0667692422866821, "learning_rate": 2.354443622292153e-06, "loss": 0.219, "step": 47780 }, { "epoch": 0.8692409829207516, "grad_norm": 0.9775679707527161, "learning_rate": 2.353474494253091e-06, "loss": 0.2191, "step": 47790 }, { "epoch": 0.8694228705505739, "grad_norm": 0.9972792267799377, "learning_rate": 2.3525066316847693e-06, "loss": 0.221, "step": 47800 }, { "epoch": 0.8696047581803962, "grad_norm": 0.9871259927749634, "learning_rate": 2.351540034923071e-06, "loss": 0.2192, "step": 47810 }, { "epoch": 0.8697866458102185, "grad_norm": 1.07982337474823, "learning_rate": 2.350574704303438e-06, "loss": 0.2266, "step": 47820 }, { "epoch": 0.8699685334400408, "grad_norm": 0.9890723824501038, "learning_rate": 2.3496106401608736e-06, "loss": 0.2218, "step": 47830 }, { "epoch": 0.8701504210698631, "grad_norm": 1.1246246099472046, "learning_rate": 2.3486478428299415e-06, "loss": 0.2226, "step": 47840 }, { "epoch": 0.8703323086996854, "grad_norm": 1.0826724767684937, "learning_rate": 2.347686312644766e-06, "loss": 0.2207, "step": 47850 }, { "epoch": 0.8705141963295077, "grad_norm": 1.0690481662750244, "learning_rate": 2.3467260499390316e-06, "loss": 0.2188, "step": 47860 }, { "epoch": 0.87069608395933, "grad_norm": 1.1464859247207642, "learning_rate": 2.345767055045981e-06, "loss": 0.2237, "step": 47870 }, { "epoch": 0.8708779715891523, "grad_norm": 0.9516382217407227, "learning_rate": 2.3448093282984207e-06, "loss": 0.2239, "step": 47880 }, { "epoch": 0.8710598592189746, "grad_norm": 0.8873171806335449, "learning_rate": 2.3438528700287145e-06, "loss": 0.2212, "step": 47890 }, { "epoch": 0.8712417468487968, "grad_norm": 1.0362290143966675, "learning_rate": 2.3428976805687865e-06, "loss": 0.2221, "step": 47900 }, { "epoch": 0.8714236344786191, "grad_norm": 1.024158239364624, "learning_rate": 2.3419437602501194e-06, "loss": 0.2206, "step": 47910 }, { "epoch": 0.8716055221084414, "grad_norm": 1.099452018737793, "learning_rate": 2.3409911094037585e-06, "loss": 0.2211, "step": 47920 }, { "epoch": 0.8717874097382637, "grad_norm": 1.0557656288146973, "learning_rate": 2.340039728360306e-06, "loss": 0.2211, "step": 47930 }, { "epoch": 0.871969297368086, "grad_norm": 0.9552907943725586, "learning_rate": 2.3390896174499237e-06, "loss": 0.2181, "step": 47940 }, { "epoch": 0.8721511849979083, "grad_norm": 1.047949194908142, "learning_rate": 2.3381407770023333e-06, "loss": 0.2168, "step": 47950 }, { "epoch": 0.8723330726277306, "grad_norm": 0.9165541529655457, "learning_rate": 2.337193207346815e-06, "loss": 0.2149, "step": 47960 }, { "epoch": 0.8725149602575529, "grad_norm": 1.0529606342315674, "learning_rate": 2.3362469088122077e-06, "loss": 0.2222, "step": 47970 }, { "epoch": 0.8726968478873752, "grad_norm": 1.063109040260315, "learning_rate": 2.3353018817269125e-06, "loss": 0.229, "step": 47980 }, { "epoch": 0.8728787355171975, "grad_norm": 1.0255229473114014, "learning_rate": 2.334358126418884e-06, "loss": 0.2179, "step": 47990 }, { "epoch": 0.8730606231470198, "grad_norm": 1.0680323839187622, "learning_rate": 2.3334156432156386e-06, "loss": 0.2219, "step": 48000 }, { "epoch": 0.8732425107768421, "grad_norm": 0.9979690313339233, "learning_rate": 2.332474432444251e-06, "loss": 0.2203, "step": 48010 }, { "epoch": 0.8734243984066644, "grad_norm": 1.035853624343872, "learning_rate": 2.3315344944313532e-06, "loss": 0.2198, "step": 48020 }, { "epoch": 0.8736062860364867, "grad_norm": 0.9094853401184082, "learning_rate": 2.3305958295031373e-06, "loss": 0.2211, "step": 48030 }, { "epoch": 0.873788173666309, "grad_norm": 1.092643141746521, "learning_rate": 2.3296584379853524e-06, "loss": 0.2232, "step": 48040 }, { "epoch": 0.8739700612961312, "grad_norm": 0.9724991321563721, "learning_rate": 2.328722320203305e-06, "loss": 0.2233, "step": 48050 }, { "epoch": 0.8741519489259535, "grad_norm": 0.9713260531425476, "learning_rate": 2.3277874764818612e-06, "loss": 0.2192, "step": 48060 }, { "epoch": 0.8743338365557758, "grad_norm": 0.9748836755752563, "learning_rate": 2.326853907145443e-06, "loss": 0.2207, "step": 48070 }, { "epoch": 0.8745157241855981, "grad_norm": 0.965522050857544, "learning_rate": 2.3259216125180336e-06, "loss": 0.2206, "step": 48080 }, { "epoch": 0.8746976118154204, "grad_norm": 1.0690813064575195, "learning_rate": 2.3249905929231693e-06, "loss": 0.222, "step": 48090 }, { "epoch": 0.8748794994452427, "grad_norm": 0.9459896683692932, "learning_rate": 2.3240608486839475e-06, "loss": 0.219, "step": 48100 }, { "epoch": 0.875061387075065, "grad_norm": 1.1509844064712524, "learning_rate": 2.3231323801230205e-06, "loss": 0.2215, "step": 48110 }, { "epoch": 0.8752432747048873, "grad_norm": 0.9814848303794861, "learning_rate": 2.3222051875626004e-06, "loss": 0.2193, "step": 48120 }, { "epoch": 0.8754251623347096, "grad_norm": 0.9716874361038208, "learning_rate": 2.3212792713244547e-06, "loss": 0.217, "step": 48130 }, { "epoch": 0.8756070499645319, "grad_norm": 1.0602940320968628, "learning_rate": 2.320354631729908e-06, "loss": 0.2163, "step": 48140 }, { "epoch": 0.8757889375943542, "grad_norm": 1.0917819738388062, "learning_rate": 2.3194312690998427e-06, "loss": 0.2265, "step": 48150 }, { "epoch": 0.8759708252241765, "grad_norm": 1.1961299180984497, "learning_rate": 2.318509183754697e-06, "loss": 0.2201, "step": 48160 }, { "epoch": 0.8761527128539988, "grad_norm": 1.068651556968689, "learning_rate": 2.3175883760144676e-06, "loss": 0.2177, "step": 48170 }, { "epoch": 0.8763346004838211, "grad_norm": 1.045971155166626, "learning_rate": 2.316668846198706e-06, "loss": 0.2201, "step": 48180 }, { "epoch": 0.8765164881136434, "grad_norm": 1.0282126665115356, "learning_rate": 2.3157505946265203e-06, "loss": 0.2176, "step": 48190 }, { "epoch": 0.8766983757434657, "grad_norm": 0.9962639808654785, "learning_rate": 2.314833621616577e-06, "loss": 0.2189, "step": 48200 }, { "epoch": 0.876880263373288, "grad_norm": 1.0432841777801514, "learning_rate": 2.313917927487096e-06, "loss": 0.2216, "step": 48210 }, { "epoch": 0.8770621510031102, "grad_norm": 1.0859683752059937, "learning_rate": 2.3130035125558566e-06, "loss": 0.2161, "step": 48220 }, { "epoch": 0.8772440386329325, "grad_norm": 1.0325125455856323, "learning_rate": 2.312090377140191e-06, "loss": 0.2178, "step": 48230 }, { "epoch": 0.8774259262627548, "grad_norm": 0.889004647731781, "learning_rate": 2.31117852155699e-06, "loss": 0.221, "step": 48240 }, { "epoch": 0.8776078138925771, "grad_norm": 0.904416561126709, "learning_rate": 2.3102679461226983e-06, "loss": 0.2183, "step": 48250 }, { "epoch": 0.8777897015223994, "grad_norm": 1.0203497409820557, "learning_rate": 2.3093586511533175e-06, "loss": 0.2205, "step": 48260 }, { "epoch": 0.8779715891522217, "grad_norm": 0.9571070075035095, "learning_rate": 2.3084506369644034e-06, "loss": 0.2203, "step": 48270 }, { "epoch": 0.878153476782044, "grad_norm": 0.8580355644226074, "learning_rate": 2.307543903871071e-06, "loss": 0.2221, "step": 48280 }, { "epoch": 0.8783353644118663, "grad_norm": 1.0300512313842773, "learning_rate": 2.306638452187986e-06, "loss": 0.2217, "step": 48290 }, { "epoch": 0.8785172520416886, "grad_norm": 0.9437963962554932, "learning_rate": 2.3057342822293713e-06, "loss": 0.2147, "step": 48300 }, { "epoch": 0.8786991396715109, "grad_norm": 1.0007628202438354, "learning_rate": 2.3048313943090058e-06, "loss": 0.2166, "step": 48310 }, { "epoch": 0.8788810273013332, "grad_norm": 1.042641520500183, "learning_rate": 2.3039297887402235e-06, "loss": 0.2213, "step": 48320 }, { "epoch": 0.8790629149311555, "grad_norm": 1.0247929096221924, "learning_rate": 2.303029465835912e-06, "loss": 0.2134, "step": 48330 }, { "epoch": 0.8792448025609778, "grad_norm": 1.0082882642745972, "learning_rate": 2.3021304259085154e-06, "loss": 0.2157, "step": 48340 }, { "epoch": 0.8794266901908001, "grad_norm": 0.8195844888687134, "learning_rate": 2.3012326692700306e-06, "loss": 0.22, "step": 48350 }, { "epoch": 0.8796085778206224, "grad_norm": 0.9533538222312927, "learning_rate": 2.30033619623201e-06, "loss": 0.2157, "step": 48360 }, { "epoch": 0.8797904654504447, "grad_norm": 1.075189232826233, "learning_rate": 2.299441007105563e-06, "loss": 0.2232, "step": 48370 }, { "epoch": 0.879972353080267, "grad_norm": 1.083745002746582, "learning_rate": 2.2985471022013495e-06, "loss": 0.2216, "step": 48380 }, { "epoch": 0.8801542407100893, "grad_norm": 0.9505921602249146, "learning_rate": 2.297654481829586e-06, "loss": 0.2213, "step": 48390 }, { "epoch": 0.8803361283399116, "grad_norm": 1.1097731590270996, "learning_rate": 2.296763146300043e-06, "loss": 0.2175, "step": 48400 }, { "epoch": 0.8805180159697339, "grad_norm": 1.0091371536254883, "learning_rate": 2.295873095922043e-06, "loss": 0.2205, "step": 48410 }, { "epoch": 0.8806999035995562, "grad_norm": 0.9489006400108337, "learning_rate": 2.2949843310044665e-06, "loss": 0.2174, "step": 48420 }, { "epoch": 0.8808817912293785, "grad_norm": 0.9508060216903687, "learning_rate": 2.2940968518557454e-06, "loss": 0.2179, "step": 48430 }, { "epoch": 0.8810636788592008, "grad_norm": 0.9710690379142761, "learning_rate": 2.2932106587838644e-06, "loss": 0.2166, "step": 48440 }, { "epoch": 0.8812455664890231, "grad_norm": 1.1104813814163208, "learning_rate": 2.292325752096365e-06, "loss": 0.2196, "step": 48450 }, { "epoch": 0.8814274541188454, "grad_norm": 0.9923499822616577, "learning_rate": 2.291442132100339e-06, "loss": 0.2178, "step": 48460 }, { "epoch": 0.8816093417486677, "grad_norm": 1.0243771076202393, "learning_rate": 2.2905597991024335e-06, "loss": 0.2229, "step": 48470 }, { "epoch": 0.88179122937849, "grad_norm": 1.0304192304611206, "learning_rate": 2.2896787534088494e-06, "loss": 0.2177, "step": 48480 }, { "epoch": 0.8819731170083123, "grad_norm": 0.9893465638160706, "learning_rate": 2.2887989953253397e-06, "loss": 0.2223, "step": 48490 }, { "epoch": 0.8821550046381346, "grad_norm": 1.0505845546722412, "learning_rate": 2.28792052515721e-06, "loss": 0.2205, "step": 48500 }, { "epoch": 0.8823368922679569, "grad_norm": 1.0188542604446411, "learning_rate": 2.287043343209321e-06, "loss": 0.2182, "step": 48510 }, { "epoch": 0.8825187798977792, "grad_norm": 1.025320291519165, "learning_rate": 2.2861674497860853e-06, "loss": 0.2184, "step": 48520 }, { "epoch": 0.8827006675276015, "grad_norm": 1.036966323852539, "learning_rate": 2.2852928451914677e-06, "loss": 0.2205, "step": 48530 }, { "epoch": 0.8828825551574238, "grad_norm": 0.9489051103591919, "learning_rate": 2.284419529728987e-06, "loss": 0.213, "step": 48540 }, { "epoch": 0.883064442787246, "grad_norm": 1.0678906440734863, "learning_rate": 2.2835475037017133e-06, "loss": 0.2193, "step": 48550 }, { "epoch": 0.8832463304170683, "grad_norm": 1.0754244327545166, "learning_rate": 2.2826767674122694e-06, "loss": 0.2242, "step": 48560 }, { "epoch": 0.8834282180468906, "grad_norm": 1.03080415725708, "learning_rate": 2.2818073211628324e-06, "loss": 0.2205, "step": 48570 }, { "epoch": 0.8836101056767129, "grad_norm": 1.104439377784729, "learning_rate": 2.2809391652551294e-06, "loss": 0.2181, "step": 48580 }, { "epoch": 0.8837919933065352, "grad_norm": 1.1010830402374268, "learning_rate": 2.2800722999904407e-06, "loss": 0.2238, "step": 48590 }, { "epoch": 0.8839738809363575, "grad_norm": 0.982247531414032, "learning_rate": 2.2792067256695985e-06, "loss": 0.2209, "step": 48600 }, { "epoch": 0.8841557685661798, "grad_norm": 1.0359705686569214, "learning_rate": 2.278342442592987e-06, "loss": 0.2215, "step": 48610 }, { "epoch": 0.8843376561960021, "grad_norm": 1.0257881879806519, "learning_rate": 2.2774794510605424e-06, "loss": 0.2161, "step": 48620 }, { "epoch": 0.8845195438258244, "grad_norm": 1.013580560684204, "learning_rate": 2.276617751371753e-06, "loss": 0.2197, "step": 48630 }, { "epoch": 0.8847014314556467, "grad_norm": 0.992938756942749, "learning_rate": 2.2757573438256587e-06, "loss": 0.2266, "step": 48640 }, { "epoch": 0.884883319085469, "grad_norm": 0.9786552786827087, "learning_rate": 2.27489822872085e-06, "loss": 0.2158, "step": 48650 }, { "epoch": 0.8850652067152913, "grad_norm": 1.0968017578125, "learning_rate": 2.2740404063554696e-06, "loss": 0.2161, "step": 48660 }, { "epoch": 0.8852470943451136, "grad_norm": 0.947352409362793, "learning_rate": 2.273183877027212e-06, "loss": 0.2218, "step": 48670 }, { "epoch": 0.8854289819749359, "grad_norm": 1.1105660200119019, "learning_rate": 2.2723286410333234e-06, "loss": 0.2197, "step": 48680 }, { "epoch": 0.8856108696047582, "grad_norm": 0.9952221512794495, "learning_rate": 2.2714746986705986e-06, "loss": 0.2215, "step": 48690 }, { "epoch": 0.8857927572345805, "grad_norm": 0.9688575267791748, "learning_rate": 2.270622050235386e-06, "loss": 0.2168, "step": 48700 }, { "epoch": 0.8859746448644028, "grad_norm": 0.9946843981742859, "learning_rate": 2.2697706960235842e-06, "loss": 0.2163, "step": 48710 }, { "epoch": 0.8861565324942251, "grad_norm": 1.082174301147461, "learning_rate": 2.2689206363306434e-06, "loss": 0.2241, "step": 48720 }, { "epoch": 0.8863384201240474, "grad_norm": 1.1445916891098022, "learning_rate": 2.2680718714515627e-06, "loss": 0.2225, "step": 48730 }, { "epoch": 0.8865203077538697, "grad_norm": 1.0169639587402344, "learning_rate": 2.267224401680893e-06, "loss": 0.2216, "step": 48740 }, { "epoch": 0.886702195383692, "grad_norm": 0.9263100028038025, "learning_rate": 2.2663782273127355e-06, "loss": 0.2135, "step": 48750 }, { "epoch": 0.8868840830135143, "grad_norm": 1.051608681678772, "learning_rate": 2.2655333486407427e-06, "loss": 0.2145, "step": 48760 }, { "epoch": 0.8870659706433366, "grad_norm": 1.0448174476623535, "learning_rate": 2.264689765958117e-06, "loss": 0.2162, "step": 48770 }, { "epoch": 0.8872478582731589, "grad_norm": 1.133402943611145, "learning_rate": 2.263847479557609e-06, "loss": 0.2199, "step": 48780 }, { "epoch": 0.8874297459029812, "grad_norm": 1.0377596616744995, "learning_rate": 2.2630064897315242e-06, "loss": 0.2181, "step": 48790 }, { "epoch": 0.8876116335328035, "grad_norm": 0.9135207533836365, "learning_rate": 2.262166796771713e-06, "loss": 0.2217, "step": 48800 }, { "epoch": 0.8877935211626258, "grad_norm": 0.9771300554275513, "learning_rate": 2.2613284009695782e-06, "loss": 0.2279, "step": 48810 }, { "epoch": 0.8879754087924481, "grad_norm": 1.0481973886489868, "learning_rate": 2.260491302616073e-06, "loss": 0.218, "step": 48820 }, { "epoch": 0.8881572964222704, "grad_norm": 0.9855868816375732, "learning_rate": 2.2596555020016995e-06, "loss": 0.2206, "step": 48830 }, { "epoch": 0.8883391840520927, "grad_norm": 0.9249936938285828, "learning_rate": 2.2588209994165093e-06, "loss": 0.2243, "step": 48840 }, { "epoch": 0.888521071681915, "grad_norm": 0.9320968985557556, "learning_rate": 2.2579877951501032e-06, "loss": 0.2197, "step": 48850 }, { "epoch": 0.8887029593117373, "grad_norm": 0.9917884469032288, "learning_rate": 2.2571558894916316e-06, "loss": 0.2237, "step": 48860 }, { "epoch": 0.8888848469415596, "grad_norm": 0.9531525373458862, "learning_rate": 2.256325282729796e-06, "loss": 0.2216, "step": 48870 }, { "epoch": 0.8890667345713817, "grad_norm": 1.130022644996643, "learning_rate": 2.2554959751528456e-06, "loss": 0.2195, "step": 48880 }, { "epoch": 0.889248622201204, "grad_norm": 1.0246096849441528, "learning_rate": 2.254667967048578e-06, "loss": 0.219, "step": 48890 }, { "epoch": 0.8894305098310263, "grad_norm": 1.0111467838287354, "learning_rate": 2.25384125870434e-06, "loss": 0.2201, "step": 48900 }, { "epoch": 0.8896123974608486, "grad_norm": 0.9242514967918396, "learning_rate": 2.25301585040703e-06, "loss": 0.2185, "step": 48910 }, { "epoch": 0.8897942850906709, "grad_norm": 0.9885618686676025, "learning_rate": 2.2521917424430924e-06, "loss": 0.2205, "step": 48920 }, { "epoch": 0.8899761727204932, "grad_norm": 1.1256972551345825, "learning_rate": 2.251368935098521e-06, "loss": 0.2158, "step": 48930 }, { "epoch": 0.8901580603503155, "grad_norm": 1.090020775794983, "learning_rate": 2.250547428658859e-06, "loss": 0.2208, "step": 48940 }, { "epoch": 0.8903399479801378, "grad_norm": 1.0668623447418213, "learning_rate": 2.2497272234091965e-06, "loss": 0.2207, "step": 48950 }, { "epoch": 0.8905218356099601, "grad_norm": 1.008733868598938, "learning_rate": 2.2489083196341744e-06, "loss": 0.2161, "step": 48960 }, { "epoch": 0.8907037232397824, "grad_norm": 1.0210314989089966, "learning_rate": 2.2480907176179806e-06, "loss": 0.2178, "step": 48970 }, { "epoch": 0.8908856108696047, "grad_norm": 0.9738550186157227, "learning_rate": 2.2472744176443506e-06, "loss": 0.2169, "step": 48980 }, { "epoch": 0.891067498499427, "grad_norm": 1.0206574201583862, "learning_rate": 2.2464594199965693e-06, "loss": 0.2169, "step": 48990 }, { "epoch": 0.8912493861292493, "grad_norm": 0.9760768413543701, "learning_rate": 2.2456457249574683e-06, "loss": 0.223, "step": 49000 }, { "epoch": 0.8914312737590716, "grad_norm": 0.8910099864006042, "learning_rate": 2.2448333328094296e-06, "loss": 0.223, "step": 49010 }, { "epoch": 0.8916131613888939, "grad_norm": 0.9755825400352478, "learning_rate": 2.24402224383438e-06, "loss": 0.2183, "step": 49020 }, { "epoch": 0.8917950490187162, "grad_norm": 1.0790802240371704, "learning_rate": 2.2432124583137965e-06, "loss": 0.2238, "step": 49030 }, { "epoch": 0.8919769366485385, "grad_norm": 1.0324113368988037, "learning_rate": 2.242403976528702e-06, "loss": 0.223, "step": 49040 }, { "epoch": 0.8921588242783608, "grad_norm": 1.131880760192871, "learning_rate": 2.2415967987596673e-06, "loss": 0.2174, "step": 49050 }, { "epoch": 0.8923407119081831, "grad_norm": 0.8835222721099854, "learning_rate": 2.240790925286811e-06, "loss": 0.2201, "step": 49060 }, { "epoch": 0.8925225995380054, "grad_norm": 1.0716150999069214, "learning_rate": 2.2399863563898014e-06, "loss": 0.2196, "step": 49070 }, { "epoch": 0.8927044871678277, "grad_norm": 1.8619354963302612, "learning_rate": 2.23918309234785e-06, "loss": 0.2203, "step": 49080 }, { "epoch": 0.89288637479765, "grad_norm": 1.000885248184204, "learning_rate": 2.2383811334397178e-06, "loss": 0.222, "step": 49090 }, { "epoch": 0.8930682624274723, "grad_norm": 0.9939632415771484, "learning_rate": 2.2375804799437107e-06, "loss": 0.2171, "step": 49100 }, { "epoch": 0.8932501500572946, "grad_norm": 0.9855164289474487, "learning_rate": 2.236781132137685e-06, "loss": 0.2141, "step": 49110 }, { "epoch": 0.8934320376871169, "grad_norm": 1.0360188484191895, "learning_rate": 2.235983090299042e-06, "loss": 0.2194, "step": 49120 }, { "epoch": 0.8936139253169392, "grad_norm": 1.1273777484893799, "learning_rate": 2.2351863547047298e-06, "loss": 0.2195, "step": 49130 }, { "epoch": 0.8937958129467615, "grad_norm": 0.9928491115570068, "learning_rate": 2.2343909256312426e-06, "loss": 0.2207, "step": 49140 }, { "epoch": 0.8939777005765838, "grad_norm": 1.0315052270889282, "learning_rate": 2.2335968033546214e-06, "loss": 0.2163, "step": 49150 }, { "epoch": 0.8941595882064061, "grad_norm": 1.1127065420150757, "learning_rate": 2.232803988150456e-06, "loss": 0.222, "step": 49160 }, { "epoch": 0.8943414758362284, "grad_norm": 0.9085960388183594, "learning_rate": 2.2320124802938793e-06, "loss": 0.2181, "step": 49170 }, { "epoch": 0.8945233634660507, "grad_norm": 0.9887523055076599, "learning_rate": 2.231222280059573e-06, "loss": 0.2196, "step": 49180 }, { "epoch": 0.894705251095873, "grad_norm": 0.9343292117118835, "learning_rate": 2.230433387721763e-06, "loss": 0.2084, "step": 49190 }, { "epoch": 0.8948871387256953, "grad_norm": 0.985170304775238, "learning_rate": 2.229645803554222e-06, "loss": 0.2112, "step": 49200 }, { "epoch": 0.8950690263555175, "grad_norm": 1.0086250305175781, "learning_rate": 2.2288595278302704e-06, "loss": 0.2225, "step": 49210 }, { "epoch": 0.8952509139853398, "grad_norm": 1.1657524108886719, "learning_rate": 2.228074560822772e-06, "loss": 0.2257, "step": 49220 }, { "epoch": 0.8954328016151621, "grad_norm": 0.9726791977882385, "learning_rate": 2.227290902804139e-06, "loss": 0.2194, "step": 49230 }, { "epoch": 0.8956146892449844, "grad_norm": 0.9862969517707825, "learning_rate": 2.2265085540463256e-06, "loss": 0.2165, "step": 49240 }, { "epoch": 0.8957965768748067, "grad_norm": 1.0852818489074707, "learning_rate": 2.2257275148208353e-06, "loss": 0.218, "step": 49250 }, { "epoch": 0.895978464504629, "grad_norm": 0.9395306706428528, "learning_rate": 2.2249477853987154e-06, "loss": 0.2163, "step": 49260 }, { "epoch": 0.8961603521344513, "grad_norm": 1.0232399702072144, "learning_rate": 2.2241693660505605e-06, "loss": 0.2202, "step": 49270 }, { "epoch": 0.8963422397642736, "grad_norm": 1.065872073173523, "learning_rate": 2.223392257046507e-06, "loss": 0.2197, "step": 49280 }, { "epoch": 0.8965241273940959, "grad_norm": 1.067436695098877, "learning_rate": 2.2226164586562395e-06, "loss": 0.2174, "step": 49290 }, { "epoch": 0.8967060150239182, "grad_norm": 1.070875883102417, "learning_rate": 2.221841971148987e-06, "loss": 0.22, "step": 49300 }, { "epoch": 0.8968879026537405, "grad_norm": 1.076474905014038, "learning_rate": 2.221068794793524e-06, "loss": 0.2204, "step": 49310 }, { "epoch": 0.8970697902835628, "grad_norm": 1.0977987051010132, "learning_rate": 2.220296929858169e-06, "loss": 0.2243, "step": 49320 }, { "epoch": 0.8972516779133851, "grad_norm": 1.0264873504638672, "learning_rate": 2.2195263766107854e-06, "loss": 0.2246, "step": 49330 }, { "epoch": 0.8974335655432074, "grad_norm": 1.0219733715057373, "learning_rate": 2.218757135318782e-06, "loss": 0.2175, "step": 49340 }, { "epoch": 0.8976154531730297, "grad_norm": 1.1031183004379272, "learning_rate": 2.217989206249112e-06, "loss": 0.2233, "step": 49350 }, { "epoch": 0.897797340802852, "grad_norm": 1.0345066785812378, "learning_rate": 2.2172225896682758e-06, "loss": 0.2183, "step": 49360 }, { "epoch": 0.8979792284326743, "grad_norm": 1.0805751085281372, "learning_rate": 2.216457285842313e-06, "loss": 0.2147, "step": 49370 }, { "epoch": 0.8981611160624966, "grad_norm": 0.9697491526603699, "learning_rate": 2.2156932950368118e-06, "loss": 0.2184, "step": 49380 }, { "epoch": 0.8983430036923189, "grad_norm": 1.0121058225631714, "learning_rate": 2.2149306175169037e-06, "loss": 0.2166, "step": 49390 }, { "epoch": 0.8985248913221412, "grad_norm": 1.1341594457626343, "learning_rate": 2.214169253547263e-06, "loss": 0.2182, "step": 49400 }, { "epoch": 0.8987067789519635, "grad_norm": 1.0302540063858032, "learning_rate": 2.2134092033921105e-06, "loss": 0.2181, "step": 49410 }, { "epoch": 0.8988886665817858, "grad_norm": 1.0734676122665405, "learning_rate": 2.21265046731521e-06, "loss": 0.2261, "step": 49420 }, { "epoch": 0.8990705542116081, "grad_norm": 0.901306688785553, "learning_rate": 2.2118930455798695e-06, "loss": 0.2168, "step": 49430 }, { "epoch": 0.8992524418414304, "grad_norm": 0.9411107301712036, "learning_rate": 2.211136938448939e-06, "loss": 0.2203, "step": 49440 }, { "epoch": 0.8994343294712527, "grad_norm": 0.9342565536499023, "learning_rate": 2.2103821461848144e-06, "loss": 0.2182, "step": 49450 }, { "epoch": 0.899616217101075, "grad_norm": 0.9786953926086426, "learning_rate": 2.2096286690494363e-06, "loss": 0.2186, "step": 49460 }, { "epoch": 0.8997981047308973, "grad_norm": 0.9577221870422363, "learning_rate": 2.2088765073042857e-06, "loss": 0.2214, "step": 49470 }, { "epoch": 0.8999799923607196, "grad_norm": 1.0659631490707397, "learning_rate": 2.2081256612103897e-06, "loss": 0.2218, "step": 49480 }, { "epoch": 0.9001618799905419, "grad_norm": 1.0675607919692993, "learning_rate": 2.2073761310283177e-06, "loss": 0.223, "step": 49490 }, { "epoch": 0.9003437676203642, "grad_norm": 1.0640678405761719, "learning_rate": 2.2066279170181825e-06, "loss": 0.2202, "step": 49500 }, { "epoch": 0.9005256552501865, "grad_norm": 1.0510852336883545, "learning_rate": 2.2058810194396404e-06, "loss": 0.2156, "step": 49510 }, { "epoch": 0.9007075428800088, "grad_norm": 1.0717664957046509, "learning_rate": 2.205135438551891e-06, "loss": 0.2165, "step": 49520 }, { "epoch": 0.9008894305098311, "grad_norm": 1.0283339023590088, "learning_rate": 2.204391174613676e-06, "loss": 0.2256, "step": 49530 }, { "epoch": 0.9010713181396534, "grad_norm": 1.0281919240951538, "learning_rate": 2.2036482278832814e-06, "loss": 0.226, "step": 49540 }, { "epoch": 0.9012532057694757, "grad_norm": 1.0302907228469849, "learning_rate": 2.2029065986185358e-06, "loss": 0.2188, "step": 49550 }, { "epoch": 0.901435093399298, "grad_norm": 1.012940764427185, "learning_rate": 2.202166287076809e-06, "loss": 0.2195, "step": 49560 }, { "epoch": 0.9016169810291202, "grad_norm": 0.9691174030303955, "learning_rate": 2.201427293515016e-06, "loss": 0.2155, "step": 49570 }, { "epoch": 0.9017988686589425, "grad_norm": 1.0270189046859741, "learning_rate": 2.200689618189613e-06, "loss": 0.2208, "step": 49580 }, { "epoch": 0.9019807562887648, "grad_norm": 1.0972816944122314, "learning_rate": 2.1999532613565978e-06, "loss": 0.2165, "step": 49590 }, { "epoch": 0.9021626439185871, "grad_norm": 1.0964717864990234, "learning_rate": 2.1992182232715124e-06, "loss": 0.2226, "step": 49600 }, { "epoch": 0.9023445315484094, "grad_norm": 0.9129855632781982, "learning_rate": 2.198484504189441e-06, "loss": 0.22, "step": 49610 }, { "epoch": 0.9025264191782317, "grad_norm": 0.96333909034729, "learning_rate": 2.1977521043650096e-06, "loss": 0.2246, "step": 49620 }, { "epoch": 0.902708306808054, "grad_norm": 1.220102310180664, "learning_rate": 2.197021024052386e-06, "loss": 0.2179, "step": 49630 }, { "epoch": 0.9028901944378763, "grad_norm": 1.134250521659851, "learning_rate": 2.19629126350528e-06, "loss": 0.2221, "step": 49640 }, { "epoch": 0.9030720820676986, "grad_norm": 1.1843374967575073, "learning_rate": 2.195562822976944e-06, "loss": 0.2189, "step": 49650 }, { "epoch": 0.9032539696975209, "grad_norm": 0.9659371376037598, "learning_rate": 2.1948357027201722e-06, "loss": 0.2259, "step": 49660 }, { "epoch": 0.9034358573273432, "grad_norm": 1.020161747932434, "learning_rate": 2.1941099029873008e-06, "loss": 0.227, "step": 49670 }, { "epoch": 0.9036177449571655, "grad_norm": 0.9759588837623596, "learning_rate": 2.193385424030207e-06, "loss": 0.2178, "step": 49680 }, { "epoch": 0.9037996325869878, "grad_norm": 1.035779356956482, "learning_rate": 2.1926622661003106e-06, "loss": 0.2172, "step": 49690 }, { "epoch": 0.9039815202168101, "grad_norm": 0.9959318041801453, "learning_rate": 2.1919404294485723e-06, "loss": 0.2195, "step": 49700 }, { "epoch": 0.9041634078466324, "grad_norm": 0.9299396872520447, "learning_rate": 2.1912199143254943e-06, "loss": 0.2177, "step": 49710 }, { "epoch": 0.9043452954764546, "grad_norm": 1.0576361417770386, "learning_rate": 2.19050072098112e-06, "loss": 0.2196, "step": 49720 }, { "epoch": 0.9045271831062769, "grad_norm": 0.9745627045631409, "learning_rate": 2.1897828496650353e-06, "loss": 0.2144, "step": 49730 }, { "epoch": 0.9047090707360992, "grad_norm": 0.9010319113731384, "learning_rate": 2.1890663006263657e-06, "loss": 0.2156, "step": 49740 }, { "epoch": 0.9048909583659215, "grad_norm": 0.9444818496704102, "learning_rate": 2.188351074113779e-06, "loss": 0.2186, "step": 49750 }, { "epoch": 0.9050728459957438, "grad_norm": 0.9957970380783081, "learning_rate": 2.1876371703754833e-06, "loss": 0.2181, "step": 49760 }, { "epoch": 0.9052547336255661, "grad_norm": 0.9100670218467712, "learning_rate": 2.1869245896592274e-06, "loss": 0.2228, "step": 49770 }, { "epoch": 0.9054366212553884, "grad_norm": 0.9854258298873901, "learning_rate": 2.186213332212303e-06, "loss": 0.2255, "step": 49780 }, { "epoch": 0.9056185088852107, "grad_norm": 0.973223865032196, "learning_rate": 2.1855033982815397e-06, "loss": 0.2173, "step": 49790 }, { "epoch": 0.905800396515033, "grad_norm": 1.0142550468444824, "learning_rate": 2.1847947881133093e-06, "loss": 0.2202, "step": 49800 }, { "epoch": 0.9059822841448553, "grad_norm": 0.9444631934165955, "learning_rate": 2.1840875019535256e-06, "loss": 0.2162, "step": 49810 }, { "epoch": 0.9061641717746776, "grad_norm": 0.958588182926178, "learning_rate": 2.1833815400476393e-06, "loss": 0.2197, "step": 49820 }, { "epoch": 0.9063460594044999, "grad_norm": 1.0321245193481445, "learning_rate": 2.1826769026406446e-06, "loss": 0.2215, "step": 49830 }, { "epoch": 0.9065279470343222, "grad_norm": 1.0929768085479736, "learning_rate": 2.181973589977075e-06, "loss": 0.2177, "step": 49840 }, { "epoch": 0.9067098346641445, "grad_norm": 0.9492863416671753, "learning_rate": 2.1812716023010057e-06, "loss": 0.2241, "step": 49850 }, { "epoch": 0.9068917222939668, "grad_norm": 0.984443187713623, "learning_rate": 2.180570939856049e-06, "loss": 0.2169, "step": 49860 }, { "epoch": 0.907073609923789, "grad_norm": 0.9171675443649292, "learning_rate": 2.1798716028853593e-06, "loss": 0.2183, "step": 49870 }, { "epoch": 0.9072554975536113, "grad_norm": 1.1326794624328613, "learning_rate": 2.179173591631631e-06, "loss": 0.224, "step": 49880 }, { "epoch": 0.9074373851834336, "grad_norm": 0.9495893120765686, "learning_rate": 2.178476906337098e-06, "loss": 0.2231, "step": 49890 }, { "epoch": 0.9076192728132559, "grad_norm": 0.9850584864616394, "learning_rate": 2.1777815472435353e-06, "loss": 0.2211, "step": 49900 }, { "epoch": 0.9078011604430782, "grad_norm": 0.9779699444770813, "learning_rate": 2.1770875145922557e-06, "loss": 0.2226, "step": 49910 }, { "epoch": 0.9079830480729005, "grad_norm": 0.9441952705383301, "learning_rate": 2.1763948086241128e-06, "loss": 0.2175, "step": 49920 }, { "epoch": 0.9081649357027228, "grad_norm": 0.930846631526947, "learning_rate": 2.1757034295794995e-06, "loss": 0.2173, "step": 49930 }, { "epoch": 0.9083468233325451, "grad_norm": 1.0873545408248901, "learning_rate": 2.1750133776983474e-06, "loss": 0.2227, "step": 49940 }, { "epoch": 0.9085287109623674, "grad_norm": 1.1264839172363281, "learning_rate": 2.1743246532201304e-06, "loss": 0.2223, "step": 49950 }, { "epoch": 0.9087105985921897, "grad_norm": 0.9264421463012695, "learning_rate": 2.1736372563838583e-06, "loss": 0.2206, "step": 49960 }, { "epoch": 0.908892486222012, "grad_norm": 0.935222864151001, "learning_rate": 2.1729511874280822e-06, "loss": 0.2174, "step": 49970 }, { "epoch": 0.9090743738518343, "grad_norm": 1.0096547603607178, "learning_rate": 2.1722664465908914e-06, "loss": 0.2233, "step": 49980 }, { "epoch": 0.9092562614816566, "grad_norm": 1.0503716468811035, "learning_rate": 2.1715830341099146e-06, "loss": 0.2174, "step": 49990 }, { "epoch": 0.9094381491114789, "grad_norm": 0.9120515584945679, "learning_rate": 2.1709009502223204e-06, "loss": 0.2243, "step": 50000 }, { "epoch": 0.9096200367413012, "grad_norm": 0.9363812208175659, "learning_rate": 2.170220195164815e-06, "loss": 0.2175, "step": 50010 }, { "epoch": 0.9098019243711235, "grad_norm": 1.025675654411316, "learning_rate": 2.1695407691736444e-06, "loss": 0.2206, "step": 50020 }, { "epoch": 0.9099838120009458, "grad_norm": 1.1202521324157715, "learning_rate": 2.168862672484592e-06, "loss": 0.2157, "step": 50030 }, { "epoch": 0.9101656996307681, "grad_norm": 0.9873199462890625, "learning_rate": 2.1681859053329816e-06, "loss": 0.2204, "step": 50040 }, { "epoch": 0.9103475872605904, "grad_norm": 1.0131508111953735, "learning_rate": 2.1675104679536748e-06, "loss": 0.2161, "step": 50050 }, { "epoch": 0.9105294748904127, "grad_norm": 1.0567446947097778, "learning_rate": 2.1668363605810726e-06, "loss": 0.2201, "step": 50060 }, { "epoch": 0.910711362520235, "grad_norm": 0.9780047535896301, "learning_rate": 2.166163583449112e-06, "loss": 0.2184, "step": 50070 }, { "epoch": 0.9108932501500573, "grad_norm": 1.0703679323196411, "learning_rate": 2.1654921367912703e-06, "loss": 0.217, "step": 50080 }, { "epoch": 0.9110751377798796, "grad_norm": 0.9783108830451965, "learning_rate": 2.164822020840564e-06, "loss": 0.2242, "step": 50090 }, { "epoch": 0.9112570254097019, "grad_norm": 1.1611164808273315, "learning_rate": 2.1641532358295454e-06, "loss": 0.2169, "step": 50100 }, { "epoch": 0.9114389130395242, "grad_norm": 0.9453316330909729, "learning_rate": 2.163485781990307e-06, "loss": 0.2185, "step": 50110 }, { "epoch": 0.9116208006693465, "grad_norm": 1.0769485235214233, "learning_rate": 2.1628196595544776e-06, "loss": 0.2207, "step": 50120 }, { "epoch": 0.9118026882991688, "grad_norm": 1.1406688690185547, "learning_rate": 2.1621548687532245e-06, "loss": 0.2202, "step": 50130 }, { "epoch": 0.9119845759289911, "grad_norm": 1.1388379335403442, "learning_rate": 2.1614914098172543e-06, "loss": 0.2196, "step": 50140 }, { "epoch": 0.9121664635588134, "grad_norm": 1.005816102027893, "learning_rate": 2.1608292829768097e-06, "loss": 0.2266, "step": 50150 }, { "epoch": 0.9123483511886357, "grad_norm": 1.0045641660690308, "learning_rate": 2.1601684884616708e-06, "loss": 0.2198, "step": 50160 }, { "epoch": 0.912530238818458, "grad_norm": 1.1141029596328735, "learning_rate": 2.1595090265011574e-06, "loss": 0.2185, "step": 50170 }, { "epoch": 0.9127121264482803, "grad_norm": 1.0433286428451538, "learning_rate": 2.158850897324126e-06, "loss": 0.2234, "step": 50180 }, { "epoch": 0.9128940140781026, "grad_norm": 1.0251669883728027, "learning_rate": 2.158194101158968e-06, "loss": 0.2222, "step": 50190 }, { "epoch": 0.9130759017079249, "grad_norm": 0.9853386282920837, "learning_rate": 2.157538638233617e-06, "loss": 0.2172, "step": 50200 }, { "epoch": 0.9132577893377472, "grad_norm": 1.104098916053772, "learning_rate": 2.1568845087755406e-06, "loss": 0.2185, "step": 50210 }, { "epoch": 0.9134396769675694, "grad_norm": 1.1173125505447388, "learning_rate": 2.1562317130117436e-06, "loss": 0.2196, "step": 50220 }, { "epoch": 0.9136215645973917, "grad_norm": 1.02001953125, "learning_rate": 2.1555802511687694e-06, "loss": 0.2211, "step": 50230 }, { "epoch": 0.913803452227214, "grad_norm": 1.0972050428390503, "learning_rate": 2.1549301234726978e-06, "loss": 0.2237, "step": 50240 }, { "epoch": 0.9139853398570363, "grad_norm": 0.9400871396064758, "learning_rate": 2.154281330149145e-06, "loss": 0.2205, "step": 50250 }, { "epoch": 0.9141672274868586, "grad_norm": 0.9508055448532104, "learning_rate": 2.153633871423266e-06, "loss": 0.2148, "step": 50260 }, { "epoch": 0.9143491151166809, "grad_norm": 1.01235830783844, "learning_rate": 2.1529877475197505e-06, "loss": 0.226, "step": 50270 }, { "epoch": 0.9145310027465032, "grad_norm": 1.056320309638977, "learning_rate": 2.1523429586628256e-06, "loss": 0.2201, "step": 50280 }, { "epoch": 0.9147128903763255, "grad_norm": 0.8765395879745483, "learning_rate": 2.151699505076256e-06, "loss": 0.2102, "step": 50290 }, { "epoch": 0.9148947780061478, "grad_norm": 0.981096625328064, "learning_rate": 2.151057386983343e-06, "loss": 0.2171, "step": 50300 }, { "epoch": 0.9150766656359701, "grad_norm": 0.9598449468612671, "learning_rate": 2.1504166046069227e-06, "loss": 0.2174, "step": 50310 }, { "epoch": 0.9152585532657924, "grad_norm": 0.9128077030181885, "learning_rate": 2.149777158169369e-06, "loss": 0.2149, "step": 50320 }, { "epoch": 0.9154404408956147, "grad_norm": 0.9727381467819214, "learning_rate": 2.149139047892592e-06, "loss": 0.2161, "step": 50330 }, { "epoch": 0.915622328525437, "grad_norm": 0.9955872893333435, "learning_rate": 2.1485022739980385e-06, "loss": 0.2219, "step": 50340 }, { "epoch": 0.9158042161552593, "grad_norm": 0.9832108616828918, "learning_rate": 2.147866836706691e-06, "loss": 0.216, "step": 50350 }, { "epoch": 0.9159861037850816, "grad_norm": 0.9656097888946533, "learning_rate": 2.1472327362390684e-06, "loss": 0.225, "step": 50360 }, { "epoch": 0.9161679914149039, "grad_norm": 0.9750476479530334, "learning_rate": 2.1465999728152254e-06, "loss": 0.2202, "step": 50370 }, { "epoch": 0.9163498790447262, "grad_norm": 0.9611464738845825, "learning_rate": 2.145968546654752e-06, "loss": 0.2171, "step": 50380 }, { "epoch": 0.9165317666745485, "grad_norm": 0.9842739701271057, "learning_rate": 2.1453384579767756e-06, "loss": 0.2208, "step": 50390 }, { "epoch": 0.9167136543043708, "grad_norm": 1.0615019798278809, "learning_rate": 2.14470970699996e-06, "loss": 0.2174, "step": 50400 }, { "epoch": 0.9168955419341931, "grad_norm": 1.0029759407043457, "learning_rate": 2.144082293942502e-06, "loss": 0.2233, "step": 50410 }, { "epoch": 0.9170774295640154, "grad_norm": 1.0402920246124268, "learning_rate": 2.1434562190221355e-06, "loss": 0.2254, "step": 50420 }, { "epoch": 0.9172593171938377, "grad_norm": 1.0092333555221558, "learning_rate": 2.142831482456131e-06, "loss": 0.2174, "step": 50430 }, { "epoch": 0.91744120482366, "grad_norm": 1.0712521076202393, "learning_rate": 2.1422080844612943e-06, "loss": 0.2239, "step": 50440 }, { "epoch": 0.9176230924534823, "grad_norm": 1.0410335063934326, "learning_rate": 2.141586025253965e-06, "loss": 0.2165, "step": 50450 }, { "epoch": 0.9178049800833046, "grad_norm": 1.034846305847168, "learning_rate": 2.1409653050500196e-06, "loss": 0.2218, "step": 50460 }, { "epoch": 0.9179868677131269, "grad_norm": 1.1084116697311401, "learning_rate": 2.14034592406487e-06, "loss": 0.2233, "step": 50470 }, { "epoch": 0.9181687553429492, "grad_norm": 1.0870945453643799, "learning_rate": 2.1397278825134614e-06, "loss": 0.2201, "step": 50480 }, { "epoch": 0.9183506429727715, "grad_norm": 1.0680650472640991, "learning_rate": 2.139111180610278e-06, "loss": 0.2176, "step": 50490 }, { "epoch": 0.9185325306025938, "grad_norm": 0.9367012977600098, "learning_rate": 2.1384958185693343e-06, "loss": 0.2167, "step": 50500 }, { "epoch": 0.9187144182324161, "grad_norm": 1.085263967514038, "learning_rate": 2.137881796604184e-06, "loss": 0.2173, "step": 50510 }, { "epoch": 0.9188963058622384, "grad_norm": 1.0263234376907349, "learning_rate": 2.137269114927913e-06, "loss": 0.2183, "step": 50520 }, { "epoch": 0.9190781934920607, "grad_norm": 0.9786432981491089, "learning_rate": 2.1366577737531434e-06, "loss": 0.212, "step": 50530 }, { "epoch": 0.919260081121883, "grad_norm": 0.9785593152046204, "learning_rate": 2.1360477732920323e-06, "loss": 0.2217, "step": 50540 }, { "epoch": 0.9194419687517053, "grad_norm": 0.9515863656997681, "learning_rate": 2.13543911375627e-06, "loss": 0.2148, "step": 50550 }, { "epoch": 0.9196238563815274, "grad_norm": 1.0963867902755737, "learning_rate": 2.134831795357083e-06, "loss": 0.2251, "step": 50560 }, { "epoch": 0.9198057440113497, "grad_norm": 1.0958924293518066, "learning_rate": 2.1342258183052322e-06, "loss": 0.2228, "step": 50570 }, { "epoch": 0.919987631641172, "grad_norm": 1.0395830869674683, "learning_rate": 2.1336211828110114e-06, "loss": 0.2248, "step": 50580 }, { "epoch": 0.9201695192709943, "grad_norm": 1.159134030342102, "learning_rate": 2.133017889084252e-06, "loss": 0.2184, "step": 50590 }, { "epoch": 0.9203514069008166, "grad_norm": 0.9786425828933716, "learning_rate": 2.1324159373343165e-06, "loss": 0.2207, "step": 50600 }, { "epoch": 0.9205332945306389, "grad_norm": 0.9749329686164856, "learning_rate": 2.1318153277701027e-06, "loss": 0.222, "step": 50610 }, { "epoch": 0.9207151821604612, "grad_norm": 1.027547836303711, "learning_rate": 2.1312160606000435e-06, "loss": 0.2192, "step": 50620 }, { "epoch": 0.9208970697902835, "grad_norm": 1.0701862573623657, "learning_rate": 2.1306181360321056e-06, "loss": 0.2218, "step": 50630 }, { "epoch": 0.9210789574201058, "grad_norm": 1.0034618377685547, "learning_rate": 2.13002155427379e-06, "loss": 0.2181, "step": 50640 }, { "epoch": 0.9212608450499281, "grad_norm": 1.0503402948379517, "learning_rate": 2.12942631553213e-06, "loss": 0.2245, "step": 50650 }, { "epoch": 0.9214427326797504, "grad_norm": 0.9860204458236694, "learning_rate": 2.1288324200136945e-06, "loss": 0.223, "step": 50660 }, { "epoch": 0.9216246203095727, "grad_norm": 1.1278706789016724, "learning_rate": 2.1282398679245857e-06, "loss": 0.2179, "step": 50670 }, { "epoch": 0.921806507939395, "grad_norm": 0.988832414150238, "learning_rate": 2.1276486594704406e-06, "loss": 0.2157, "step": 50680 }, { "epoch": 0.9219883955692173, "grad_norm": 1.0015689134597778, "learning_rate": 2.127058794856429e-06, "loss": 0.2153, "step": 50690 }, { "epoch": 0.9221702831990396, "grad_norm": 1.002034306526184, "learning_rate": 2.126470274287253e-06, "loss": 0.22, "step": 50700 }, { "epoch": 0.9223521708288619, "grad_norm": 1.0272670984268188, "learning_rate": 2.12588309796715e-06, "loss": 0.2133, "step": 50710 }, { "epoch": 0.9225340584586842, "grad_norm": 1.0159516334533691, "learning_rate": 2.1252972660998913e-06, "loss": 0.2212, "step": 50720 }, { "epoch": 0.9227159460885065, "grad_norm": 1.0691665410995483, "learning_rate": 2.1247127788887804e-06, "loss": 0.2163, "step": 50730 }, { "epoch": 0.9228978337183288, "grad_norm": 1.058031678199768, "learning_rate": 2.1241296365366547e-06, "loss": 0.2149, "step": 50740 }, { "epoch": 0.9230797213481511, "grad_norm": 0.9425684809684753, "learning_rate": 2.1235478392458857e-06, "loss": 0.2227, "step": 50750 }, { "epoch": 0.9232616089779734, "grad_norm": 0.9352062344551086, "learning_rate": 2.1229673872183763e-06, "loss": 0.2205, "step": 50760 }, { "epoch": 0.9234434966077957, "grad_norm": 1.0651869773864746, "learning_rate": 2.1223882806555625e-06, "loss": 0.2168, "step": 50770 }, { "epoch": 0.923625384237618, "grad_norm": 0.9277414083480835, "learning_rate": 2.121810519758416e-06, "loss": 0.2165, "step": 50780 }, { "epoch": 0.9238072718674403, "grad_norm": 1.0709866285324097, "learning_rate": 2.1212341047274397e-06, "loss": 0.2216, "step": 50790 }, { "epoch": 0.9239891594972626, "grad_norm": 0.9877578020095825, "learning_rate": 2.1206590357626685e-06, "loss": 0.2181, "step": 50800 }, { "epoch": 0.9241710471270849, "grad_norm": 1.0435523986816406, "learning_rate": 2.120085313063672e-06, "loss": 0.2166, "step": 50810 }, { "epoch": 0.9243529347569072, "grad_norm": 0.8744575381278992, "learning_rate": 2.1195129368295516e-06, "loss": 0.2177, "step": 50820 }, { "epoch": 0.9245348223867295, "grad_norm": 0.9327054023742676, "learning_rate": 2.118941907258942e-06, "loss": 0.2132, "step": 50830 }, { "epoch": 0.9247167100165518, "grad_norm": 0.9790618419647217, "learning_rate": 2.1183722245500106e-06, "loss": 0.2209, "step": 50840 }, { "epoch": 0.924898597646374, "grad_norm": 1.0119765996932983, "learning_rate": 2.117803888900457e-06, "loss": 0.2232, "step": 50850 }, { "epoch": 0.9250804852761964, "grad_norm": 0.9614843130111694, "learning_rate": 2.117236900507512e-06, "loss": 0.2212, "step": 50860 }, { "epoch": 0.9252623729060186, "grad_norm": 1.0565814971923828, "learning_rate": 2.1166712595679407e-06, "loss": 0.2163, "step": 50870 }, { "epoch": 0.925444260535841, "grad_norm": 0.9433305859565735, "learning_rate": 2.1161069662780415e-06, "loss": 0.2188, "step": 50880 }, { "epoch": 0.9256261481656632, "grad_norm": 0.9603264331817627, "learning_rate": 2.115544020833643e-06, "loss": 0.2117, "step": 50890 }, { "epoch": 0.9258080357954855, "grad_norm": 0.9880537986755371, "learning_rate": 2.1149824234301055e-06, "loss": 0.2243, "step": 50900 }, { "epoch": 0.9259899234253078, "grad_norm": 1.0101677179336548, "learning_rate": 2.1144221742623246e-06, "loss": 0.2182, "step": 50910 }, { "epoch": 0.9261718110551301, "grad_norm": 1.2035956382751465, "learning_rate": 2.1138632735247253e-06, "loss": 0.2188, "step": 50920 }, { "epoch": 0.9263536986849524, "grad_norm": 1.0194079875946045, "learning_rate": 2.1133057214112655e-06, "loss": 0.2122, "step": 50930 }, { "epoch": 0.9265355863147747, "grad_norm": 1.177993655204773, "learning_rate": 2.1127495181154353e-06, "loss": 0.218, "step": 50940 }, { "epoch": 0.926717473944597, "grad_norm": 0.9876782298088074, "learning_rate": 2.112194663830257e-06, "loss": 0.2153, "step": 50950 }, { "epoch": 0.9268993615744193, "grad_norm": 0.9176049828529358, "learning_rate": 2.1116411587482836e-06, "loss": 0.2156, "step": 50960 }, { "epoch": 0.9270812492042416, "grad_norm": 1.0580894947052002, "learning_rate": 2.1110890030616e-06, "loss": 0.2205, "step": 50970 }, { "epoch": 0.9272631368340639, "grad_norm": 1.042250633239746, "learning_rate": 2.1105381969618246e-06, "loss": 0.2179, "step": 50980 }, { "epoch": 0.9274450244638862, "grad_norm": 1.0456092357635498, "learning_rate": 2.1099887406401053e-06, "loss": 0.2195, "step": 50990 }, { "epoch": 0.9276269120937085, "grad_norm": 0.9836710691452026, "learning_rate": 2.109440634287123e-06, "loss": 0.2207, "step": 51000 }, { "epoch": 0.9278087997235308, "grad_norm": 1.1064010858535767, "learning_rate": 2.1088938780930896e-06, "loss": 0.2179, "step": 51010 }, { "epoch": 0.9279906873533531, "grad_norm": 1.0799050331115723, "learning_rate": 2.1083484722477485e-06, "loss": 0.216, "step": 51020 }, { "epoch": 0.9281725749831754, "grad_norm": 0.9644990563392639, "learning_rate": 2.107804416940374e-06, "loss": 0.2213, "step": 51030 }, { "epoch": 0.9283544626129977, "grad_norm": 1.0484766960144043, "learning_rate": 2.1072617123597723e-06, "loss": 0.2207, "step": 51040 }, { "epoch": 0.92853635024282, "grad_norm": 0.9928408265113831, "learning_rate": 2.106720358694281e-06, "loss": 0.2216, "step": 51050 }, { "epoch": 0.9287182378726423, "grad_norm": 1.019343376159668, "learning_rate": 2.106180356131769e-06, "loss": 0.221, "step": 51060 }, { "epoch": 0.9289001255024646, "grad_norm": 1.0188078880310059, "learning_rate": 2.105641704859634e-06, "loss": 0.2149, "step": 51070 }, { "epoch": 0.9290820131322869, "grad_norm": 0.9403066039085388, "learning_rate": 2.1051044050648088e-06, "loss": 0.2166, "step": 51080 }, { "epoch": 0.9292639007621092, "grad_norm": 0.9873102903366089, "learning_rate": 2.104568456933754e-06, "loss": 0.2201, "step": 51090 }, { "epoch": 0.9294457883919315, "grad_norm": 0.9997217059135437, "learning_rate": 2.1040338606524624e-06, "loss": 0.218, "step": 51100 }, { "epoch": 0.9296276760217538, "grad_norm": 1.1595381498336792, "learning_rate": 2.103500616406457e-06, "loss": 0.2211, "step": 51110 }, { "epoch": 0.9298095636515761, "grad_norm": 0.9895651936531067, "learning_rate": 2.1029687243807926e-06, "loss": 0.2164, "step": 51120 }, { "epoch": 0.9299914512813984, "grad_norm": 1.0492875576019287, "learning_rate": 2.1024381847600545e-06, "loss": 0.2212, "step": 51130 }, { "epoch": 0.9301733389112207, "grad_norm": 0.9541158676147461, "learning_rate": 2.101908997728358e-06, "loss": 0.223, "step": 51140 }, { "epoch": 0.930355226541043, "grad_norm": 0.9395267963409424, "learning_rate": 2.101381163469349e-06, "loss": 0.2155, "step": 51150 }, { "epoch": 0.9305371141708653, "grad_norm": 0.987350344657898, "learning_rate": 2.100854682166204e-06, "loss": 0.216, "step": 51160 }, { "epoch": 0.9307190018006876, "grad_norm": 1.005720853805542, "learning_rate": 2.100329554001631e-06, "loss": 0.2181, "step": 51170 }, { "epoch": 0.9309008894305099, "grad_norm": 1.0411951541900635, "learning_rate": 2.0998057791578676e-06, "loss": 0.2191, "step": 51180 }, { "epoch": 0.9310827770603322, "grad_norm": 1.036994457244873, "learning_rate": 2.0992833578166813e-06, "loss": 0.2205, "step": 51190 }, { "epoch": 0.9312646646901545, "grad_norm": 1.0198848247528076, "learning_rate": 2.098762290159371e-06, "loss": 0.2124, "step": 51200 }, { "epoch": 0.9314465523199768, "grad_norm": 0.934201180934906, "learning_rate": 2.0982425763667644e-06, "loss": 0.2212, "step": 51210 }, { "epoch": 0.931628439949799, "grad_norm": 1.1349257230758667, "learning_rate": 2.0977242166192218e-06, "loss": 0.2243, "step": 51220 }, { "epoch": 0.9318103275796213, "grad_norm": 0.9570160508155823, "learning_rate": 2.0972072110966304e-06, "loss": 0.2183, "step": 51230 }, { "epoch": 0.9319922152094436, "grad_norm": 1.034578800201416, "learning_rate": 2.0966915599784096e-06, "loss": 0.2182, "step": 51240 }, { "epoch": 0.9321741028392659, "grad_norm": 1.006939172744751, "learning_rate": 2.0961772634435087e-06, "loss": 0.2148, "step": 51250 }, { "epoch": 0.9323559904690882, "grad_norm": 0.9934409856796265, "learning_rate": 2.095664321670405e-06, "loss": 0.2179, "step": 51260 }, { "epoch": 0.9325378780989105, "grad_norm": 1.0886871814727783, "learning_rate": 2.095152734837109e-06, "loss": 0.223, "step": 51270 }, { "epoch": 0.9327197657287328, "grad_norm": 1.0539145469665527, "learning_rate": 2.0946425031211574e-06, "loss": 0.2157, "step": 51280 }, { "epoch": 0.9329016533585551, "grad_norm": 1.0415384769439697, "learning_rate": 2.0941336266996194e-06, "loss": 0.2185, "step": 51290 }, { "epoch": 0.9330835409883774, "grad_norm": 1.14793062210083, "learning_rate": 2.0936261057490925e-06, "loss": 0.2254, "step": 51300 }, { "epoch": 0.9332654286181997, "grad_norm": 1.013717532157898, "learning_rate": 2.0931199404457035e-06, "loss": 0.2161, "step": 51310 }, { "epoch": 0.933447316248022, "grad_norm": 1.0698362588882446, "learning_rate": 2.0926151309651095e-06, "loss": 0.2154, "step": 51320 }, { "epoch": 0.9336292038778443, "grad_norm": 0.9392372965812683, "learning_rate": 2.092111677482498e-06, "loss": 0.2209, "step": 51330 }, { "epoch": 0.9338110915076666, "grad_norm": 0.9648811221122742, "learning_rate": 2.0916095801725834e-06, "loss": 0.218, "step": 51340 }, { "epoch": 0.9339929791374889, "grad_norm": 1.093062162399292, "learning_rate": 2.091108839209611e-06, "loss": 0.2199, "step": 51350 }, { "epoch": 0.9341748667673112, "grad_norm": 1.0139596462249756, "learning_rate": 2.0906094547673565e-06, "loss": 0.214, "step": 51360 }, { "epoch": 0.9343567543971335, "grad_norm": 0.9550477266311646, "learning_rate": 2.0901114270191224e-06, "loss": 0.2187, "step": 51370 }, { "epoch": 0.9345386420269558, "grad_norm": 1.0060465335845947, "learning_rate": 2.0896147561377416e-06, "loss": 0.2171, "step": 51380 }, { "epoch": 0.934720529656778, "grad_norm": 1.0337440967559814, "learning_rate": 2.089119442295577e-06, "loss": 0.2164, "step": 51390 }, { "epoch": 0.9349024172866003, "grad_norm": 0.9330530166625977, "learning_rate": 2.0886254856645188e-06, "loss": 0.218, "step": 51400 }, { "epoch": 0.9350843049164226, "grad_norm": 0.9793212413787842, "learning_rate": 2.0881328864159872e-06, "loss": 0.2148, "step": 51410 }, { "epoch": 0.9352661925462449, "grad_norm": 1.024221420288086, "learning_rate": 2.0876416447209313e-06, "loss": 0.2172, "step": 51420 }, { "epoch": 0.9354480801760672, "grad_norm": 1.0043543577194214, "learning_rate": 2.08715176074983e-06, "loss": 0.2196, "step": 51430 }, { "epoch": 0.9356299678058895, "grad_norm": 1.0451810359954834, "learning_rate": 2.0866632346726883e-06, "loss": 0.2117, "step": 51440 }, { "epoch": 0.9358118554357118, "grad_norm": 1.0153279304504395, "learning_rate": 2.0861760666590427e-06, "loss": 0.2174, "step": 51450 }, { "epoch": 0.9359937430655341, "grad_norm": 0.9480035305023193, "learning_rate": 2.0856902568779573e-06, "loss": 0.2183, "step": 51460 }, { "epoch": 0.9361756306953564, "grad_norm": 0.9770836234092712, "learning_rate": 2.0852058054980246e-06, "loss": 0.22, "step": 51470 }, { "epoch": 0.9363575183251787, "grad_norm": 0.8515194654464722, "learning_rate": 2.0847227126873667e-06, "loss": 0.2164, "step": 51480 }, { "epoch": 0.936539405955001, "grad_norm": 1.0212386846542358, "learning_rate": 2.0842409786136327e-06, "loss": 0.22, "step": 51490 }, { "epoch": 0.9367212935848233, "grad_norm": 0.9332212209701538, "learning_rate": 2.0837606034440017e-06, "loss": 0.2243, "step": 51500 }, { "epoch": 0.9369031812146456, "grad_norm": 0.9747300744056702, "learning_rate": 2.0832815873451793e-06, "loss": 0.2198, "step": 51510 }, { "epoch": 0.9370850688444679, "grad_norm": 0.9465929269790649, "learning_rate": 2.082803930483403e-06, "loss": 0.2192, "step": 51520 }, { "epoch": 0.9372669564742901, "grad_norm": 1.1347110271453857, "learning_rate": 2.0823276330244345e-06, "loss": 0.2213, "step": 51530 }, { "epoch": 0.9374488441041124, "grad_norm": 1.0529600381851196, "learning_rate": 2.0818526951335656e-06, "loss": 0.2163, "step": 51540 }, { "epoch": 0.9376307317339347, "grad_norm": 0.9221251010894775, "learning_rate": 2.0813791169756165e-06, "loss": 0.2175, "step": 51550 }, { "epoch": 0.937812619363757, "grad_norm": 0.9673653244972229, "learning_rate": 2.0809068987149357e-06, "loss": 0.2167, "step": 51560 }, { "epoch": 0.9379945069935793, "grad_norm": 0.983202338218689, "learning_rate": 2.0804360405153983e-06, "loss": 0.218, "step": 51570 }, { "epoch": 0.9381763946234016, "grad_norm": 1.125699520111084, "learning_rate": 2.07996654254041e-06, "loss": 0.2237, "step": 51580 }, { "epoch": 0.9383582822532239, "grad_norm": 0.9538814425468445, "learning_rate": 2.079498404952901e-06, "loss": 0.2183, "step": 51590 }, { "epoch": 0.9385401698830462, "grad_norm": 1.2127059698104858, "learning_rate": 2.0790316279153326e-06, "loss": 0.2231, "step": 51600 }, { "epoch": 0.9387220575128685, "grad_norm": 1.0795354843139648, "learning_rate": 2.078566211589692e-06, "loss": 0.2217, "step": 51610 }, { "epoch": 0.9389039451426908, "grad_norm": 0.8942591547966003, "learning_rate": 2.078102156137495e-06, "loss": 0.2168, "step": 51620 }, { "epoch": 0.9390858327725131, "grad_norm": 1.139607310295105, "learning_rate": 2.077639461719785e-06, "loss": 0.2253, "step": 51630 }, { "epoch": 0.9392677204023354, "grad_norm": 1.063031792640686, "learning_rate": 2.077178128497133e-06, "loss": 0.2215, "step": 51640 }, { "epoch": 0.9394496080321577, "grad_norm": 0.9405277371406555, "learning_rate": 2.0767181566296375e-06, "loss": 0.2204, "step": 51650 }, { "epoch": 0.93963149566198, "grad_norm": 1.063250184059143, "learning_rate": 2.076259546276924e-06, "loss": 0.2196, "step": 51660 }, { "epoch": 0.9398133832918023, "grad_norm": 0.8859453797340393, "learning_rate": 2.0758022975981474e-06, "loss": 0.2205, "step": 51670 }, { "epoch": 0.9399952709216246, "grad_norm": 0.9899684190750122, "learning_rate": 2.0753464107519885e-06, "loss": 0.2192, "step": 51680 }, { "epoch": 0.9401771585514469, "grad_norm": 1.0534582138061523, "learning_rate": 2.074891885896656e-06, "loss": 0.2262, "step": 51690 }, { "epoch": 0.9403590461812692, "grad_norm": 0.9609195590019226, "learning_rate": 2.074438723189884e-06, "loss": 0.222, "step": 51700 }, { "epoch": 0.9405409338110915, "grad_norm": 1.032922625541687, "learning_rate": 2.0739869227889385e-06, "loss": 0.2187, "step": 51710 }, { "epoch": 0.9407228214409138, "grad_norm": 1.1559892892837524, "learning_rate": 2.0735364848506078e-06, "loss": 0.2194, "step": 51720 }, { "epoch": 0.9409047090707361, "grad_norm": 1.0096423625946045, "learning_rate": 2.0730874095312104e-06, "loss": 0.2176, "step": 51730 }, { "epoch": 0.9410865967005584, "grad_norm": 1.0829038619995117, "learning_rate": 2.072639696986591e-06, "loss": 0.2201, "step": 51740 }, { "epoch": 0.9412684843303807, "grad_norm": 1.0790120363235474, "learning_rate": 2.072193347372121e-06, "loss": 0.2202, "step": 51750 }, { "epoch": 0.941450371960203, "grad_norm": 1.0045809745788574, "learning_rate": 2.0717483608426988e-06, "loss": 0.2241, "step": 51760 }, { "epoch": 0.9416322595900253, "grad_norm": 1.014689564704895, "learning_rate": 2.071304737552751e-06, "loss": 0.2238, "step": 51770 }, { "epoch": 0.9418141472198476, "grad_norm": 0.9740641117095947, "learning_rate": 2.07086247765623e-06, "loss": 0.2269, "step": 51780 }, { "epoch": 0.9419960348496699, "grad_norm": 1.0788333415985107, "learning_rate": 2.0704215813066153e-06, "loss": 0.2216, "step": 51790 }, { "epoch": 0.9421779224794922, "grad_norm": 1.116563320159912, "learning_rate": 2.069982048656913e-06, "loss": 0.2201, "step": 51800 }, { "epoch": 0.9423598101093145, "grad_norm": 0.9668092131614685, "learning_rate": 2.0695438798596566e-06, "loss": 0.2136, "step": 51810 }, { "epoch": 0.9425416977391368, "grad_norm": 1.0589921474456787, "learning_rate": 2.0691070750669048e-06, "loss": 0.2196, "step": 51820 }, { "epoch": 0.9427235853689591, "grad_norm": 0.9874095320701599, "learning_rate": 2.0686716344302457e-06, "loss": 0.2158, "step": 51830 }, { "epoch": 0.9429054729987814, "grad_norm": 1.0002796649932861, "learning_rate": 2.0682375581007914e-06, "loss": 0.2136, "step": 51840 }, { "epoch": 0.9430873606286037, "grad_norm": 1.0213795900344849, "learning_rate": 2.067804846229181e-06, "loss": 0.2118, "step": 51850 }, { "epoch": 0.943269248258426, "grad_norm": 1.0628570318222046, "learning_rate": 2.0673734989655812e-06, "loss": 0.2177, "step": 51860 }, { "epoch": 0.9434511358882482, "grad_norm": 1.0730541944503784, "learning_rate": 2.066943516459684e-06, "loss": 0.2071, "step": 51870 }, { "epoch": 0.9436330235180705, "grad_norm": 1.0267469882965088, "learning_rate": 2.066514898860709e-06, "loss": 0.2254, "step": 51880 }, { "epoch": 0.9438149111478928, "grad_norm": 0.9479590058326721, "learning_rate": 2.0660876463174e-06, "loss": 0.224, "step": 51890 }, { "epoch": 0.9439967987777151, "grad_norm": 1.1217676401138306, "learning_rate": 2.06566175897803e-06, "loss": 0.2205, "step": 51900 }, { "epoch": 0.9441786864075374, "grad_norm": 0.9698897004127502, "learning_rate": 2.0652372369903947e-06, "loss": 0.2221, "step": 51910 }, { "epoch": 0.9443605740373597, "grad_norm": 1.068892240524292, "learning_rate": 2.06481408050182e-06, "loss": 0.2253, "step": 51920 }, { "epoch": 0.944542461667182, "grad_norm": 0.979950487613678, "learning_rate": 2.064392289659155e-06, "loss": 0.2151, "step": 51930 }, { "epoch": 0.9447243492970043, "grad_norm": 0.9268290996551514, "learning_rate": 2.0639718646087758e-06, "loss": 0.2174, "step": 51940 }, { "epoch": 0.9449062369268266, "grad_norm": 1.0287888050079346, "learning_rate": 2.0635528054965843e-06, "loss": 0.2187, "step": 51950 }, { "epoch": 0.9450881245566489, "grad_norm": 1.1152727603912354, "learning_rate": 2.0631351124680082e-06, "loss": 0.221, "step": 51960 }, { "epoch": 0.9452700121864712, "grad_norm": 0.9351426959037781, "learning_rate": 2.0627187856680024e-06, "loss": 0.2196, "step": 51970 }, { "epoch": 0.9454518998162935, "grad_norm": 0.8360930681228638, "learning_rate": 2.0623038252410465e-06, "loss": 0.2168, "step": 51980 }, { "epoch": 0.9456337874461158, "grad_norm": 0.9770666360855103, "learning_rate": 2.0618902313311453e-06, "loss": 0.2167, "step": 51990 }, { "epoch": 0.9458156750759381, "grad_norm": 0.9127971529960632, "learning_rate": 2.0614780040818305e-06, "loss": 0.2151, "step": 52000 }, { "epoch": 0.9459975627057604, "grad_norm": 1.1427135467529297, "learning_rate": 2.0610671436361603e-06, "loss": 0.2163, "step": 52010 }, { "epoch": 0.9461794503355827, "grad_norm": 0.9507139325141907, "learning_rate": 2.0606576501367165e-06, "loss": 0.2218, "step": 52020 }, { "epoch": 0.946361337965405, "grad_norm": 1.1042839288711548, "learning_rate": 2.0602495237256078e-06, "loss": 0.2197, "step": 52030 }, { "epoch": 0.9465432255952273, "grad_norm": 1.04312002658844, "learning_rate": 2.0598427645444684e-06, "loss": 0.2163, "step": 52040 }, { "epoch": 0.9467251132250496, "grad_norm": 1.0626236200332642, "learning_rate": 2.0594373727344574e-06, "loss": 0.2258, "step": 52050 }, { "epoch": 0.9469070008548719, "grad_norm": 1.0234336853027344, "learning_rate": 2.05903334843626e-06, "loss": 0.2199, "step": 52060 }, { "epoch": 0.9470888884846942, "grad_norm": 0.9290069341659546, "learning_rate": 2.0586306917900875e-06, "loss": 0.2186, "step": 52070 }, { "epoch": 0.9472707761145165, "grad_norm": 0.9418688416481018, "learning_rate": 2.058229402935674e-06, "loss": 0.2126, "step": 52080 }, { "epoch": 0.9474526637443388, "grad_norm": 1.1064765453338623, "learning_rate": 2.0578294820122816e-06, "loss": 0.2163, "step": 52090 }, { "epoch": 0.9476345513741611, "grad_norm": 0.9492932558059692, "learning_rate": 2.0574309291586974e-06, "loss": 0.2209, "step": 52100 }, { "epoch": 0.9478164390039834, "grad_norm": 0.889927327632904, "learning_rate": 2.057033744513231e-06, "loss": 0.2142, "step": 52110 }, { "epoch": 0.9479983266338057, "grad_norm": 1.1919007301330566, "learning_rate": 2.0566379282137215e-06, "loss": 0.2202, "step": 52120 }, { "epoch": 0.948180214263628, "grad_norm": 0.9640122652053833, "learning_rate": 2.05624348039753e-06, "loss": 0.2164, "step": 52130 }, { "epoch": 0.9483621018934503, "grad_norm": 0.9227055311203003, "learning_rate": 2.0558504012015435e-06, "loss": 0.2209, "step": 52140 }, { "epoch": 0.9485439895232726, "grad_norm": 1.1444154977798462, "learning_rate": 2.055458690762174e-06, "loss": 0.2128, "step": 52150 }, { "epoch": 0.9487258771530949, "grad_norm": 1.0124015808105469, "learning_rate": 2.0550683492153584e-06, "loss": 0.2209, "step": 52160 }, { "epoch": 0.9489077647829172, "grad_norm": 0.9836930632591248, "learning_rate": 2.0546793766965594e-06, "loss": 0.2182, "step": 52170 }, { "epoch": 0.9490896524127395, "grad_norm": 1.1620041131973267, "learning_rate": 2.0542917733407636e-06, "loss": 0.2164, "step": 52180 }, { "epoch": 0.9492715400425618, "grad_norm": 1.0114210844039917, "learning_rate": 2.053905539282483e-06, "loss": 0.2188, "step": 52190 }, { "epoch": 0.949453427672384, "grad_norm": 0.9145107865333557, "learning_rate": 2.053520674655753e-06, "loss": 0.2143, "step": 52200 }, { "epoch": 0.9496353153022064, "grad_norm": 0.9818302989006042, "learning_rate": 2.053137179594137e-06, "loss": 0.216, "step": 52210 }, { "epoch": 0.9498172029320286, "grad_norm": 1.061311960220337, "learning_rate": 2.0527550542307198e-06, "loss": 0.2157, "step": 52220 }, { "epoch": 0.9499990905618508, "grad_norm": 0.9926944375038147, "learning_rate": 2.0523742986981125e-06, "loss": 0.2162, "step": 52230 }, { "epoch": 0.9501809781916731, "grad_norm": 1.0315511226654053, "learning_rate": 2.05199491312845e-06, "loss": 0.2201, "step": 52240 }, { "epoch": 0.9503628658214954, "grad_norm": 1.0509774684906006, "learning_rate": 2.0516168976533925e-06, "loss": 0.218, "step": 52250 }, { "epoch": 0.9505447534513177, "grad_norm": 1.0633182525634766, "learning_rate": 2.051240252404125e-06, "loss": 0.226, "step": 52260 }, { "epoch": 0.95072664108114, "grad_norm": 1.149706244468689, "learning_rate": 2.0508649775113564e-06, "loss": 0.2171, "step": 52270 }, { "epoch": 0.9509085287109623, "grad_norm": 1.0855449438095093, "learning_rate": 2.0504910731053195e-06, "loss": 0.2232, "step": 52280 }, { "epoch": 0.9510904163407846, "grad_norm": 1.0344364643096924, "learning_rate": 2.0501185393157714e-06, "loss": 0.2217, "step": 52290 }, { "epoch": 0.9512723039706069, "grad_norm": 1.0421452522277832, "learning_rate": 2.0497473762719958e-06, "loss": 0.2159, "step": 52300 }, { "epoch": 0.9514541916004292, "grad_norm": 0.9661158323287964, "learning_rate": 2.0493775841027987e-06, "loss": 0.2195, "step": 52310 }, { "epoch": 0.9516360792302515, "grad_norm": 0.9808648824691772, "learning_rate": 2.0490091629365103e-06, "loss": 0.2186, "step": 52320 }, { "epoch": 0.9518179668600738, "grad_norm": 1.0297309160232544, "learning_rate": 2.048642112900986e-06, "loss": 0.2213, "step": 52330 }, { "epoch": 0.9519998544898961, "grad_norm": 0.9023817777633667, "learning_rate": 2.0482764341236048e-06, "loss": 0.2172, "step": 52340 }, { "epoch": 0.9521817421197184, "grad_norm": 0.9629757404327393, "learning_rate": 2.0479121267312694e-06, "loss": 0.2155, "step": 52350 }, { "epoch": 0.9523636297495407, "grad_norm": 0.9992483258247375, "learning_rate": 2.0475491908504073e-06, "loss": 0.2173, "step": 52360 }, { "epoch": 0.952545517379363, "grad_norm": 1.0041064023971558, "learning_rate": 2.0471876266069705e-06, "loss": 0.2221, "step": 52370 }, { "epoch": 0.9527274050091853, "grad_norm": 1.0142927169799805, "learning_rate": 2.046827434126434e-06, "loss": 0.2165, "step": 52380 }, { "epoch": 0.9529092926390076, "grad_norm": 0.9843560457229614, "learning_rate": 2.0464686135337965e-06, "loss": 0.2164, "step": 52390 }, { "epoch": 0.9530911802688299, "grad_norm": 0.9394570589065552, "learning_rate": 2.0461111649535817e-06, "loss": 0.2152, "step": 52400 }, { "epoch": 0.9532730678986522, "grad_norm": 1.0174832344055176, "learning_rate": 2.0457550885098364e-06, "loss": 0.2145, "step": 52410 }, { "epoch": 0.9534549555284745, "grad_norm": 1.116393804550171, "learning_rate": 2.0454003843261323e-06, "loss": 0.2212, "step": 52420 }, { "epoch": 0.9536368431582968, "grad_norm": 1.1700621843338013, "learning_rate": 2.0450470525255627e-06, "loss": 0.2231, "step": 52430 }, { "epoch": 0.9538187307881191, "grad_norm": 1.023358941078186, "learning_rate": 2.044695093230747e-06, "loss": 0.2158, "step": 52440 }, { "epoch": 0.9540006184179414, "grad_norm": 1.03300142288208, "learning_rate": 2.0443445065638266e-06, "loss": 0.2164, "step": 52450 }, { "epoch": 0.9541825060477637, "grad_norm": 0.9958537817001343, "learning_rate": 2.0439952926464684e-06, "loss": 0.2139, "step": 52460 }, { "epoch": 0.954364393677586, "grad_norm": 1.083014726638794, "learning_rate": 2.04364745159986e-06, "loss": 0.2154, "step": 52470 }, { "epoch": 0.9545462813074083, "grad_norm": 0.9867051243782043, "learning_rate": 2.0433009835447153e-06, "loss": 0.2141, "step": 52480 }, { "epoch": 0.9547281689372306, "grad_norm": 1.1113662719726562, "learning_rate": 2.042955888601271e-06, "loss": 0.2243, "step": 52490 }, { "epoch": 0.9549100565670529, "grad_norm": 0.994591474533081, "learning_rate": 2.042612166889286e-06, "loss": 0.219, "step": 52500 }, { "epoch": 0.9550919441968752, "grad_norm": 0.8671774864196777, "learning_rate": 2.0422698185280445e-06, "loss": 0.2186, "step": 52510 }, { "epoch": 0.9552738318266975, "grad_norm": 0.9693871140480042, "learning_rate": 2.0419288436363534e-06, "loss": 0.2219, "step": 52520 }, { "epoch": 0.9554557194565197, "grad_norm": 0.9873690009117126, "learning_rate": 2.041589242332542e-06, "loss": 0.2157, "step": 52530 }, { "epoch": 0.955637607086342, "grad_norm": 1.0852140188217163, "learning_rate": 2.041251014734464e-06, "loss": 0.221, "step": 52540 }, { "epoch": 0.9558194947161643, "grad_norm": 0.9870039820671082, "learning_rate": 2.040914160959497e-06, "loss": 0.2246, "step": 52550 }, { "epoch": 0.9560013823459866, "grad_norm": 0.9315917491912842, "learning_rate": 2.0405786811245397e-06, "loss": 0.2196, "step": 52560 }, { "epoch": 0.9561832699758089, "grad_norm": 0.9728351831436157, "learning_rate": 2.0402445753460155e-06, "loss": 0.2168, "step": 52570 }, { "epoch": 0.9563651576056312, "grad_norm": 1.0221359729766846, "learning_rate": 2.039911843739871e-06, "loss": 0.2153, "step": 52580 }, { "epoch": 0.9565470452354535, "grad_norm": 0.9398824572563171, "learning_rate": 2.0395804864215756e-06, "loss": 0.2163, "step": 52590 }, { "epoch": 0.9567289328652758, "grad_norm": 1.0585192441940308, "learning_rate": 2.0392505035061215e-06, "loss": 0.2172, "step": 52600 }, { "epoch": 0.9569108204950981, "grad_norm": 1.0444186925888062, "learning_rate": 2.0389218951080246e-06, "loss": 0.2146, "step": 52610 }, { "epoch": 0.9570927081249204, "grad_norm": 1.063467264175415, "learning_rate": 2.038594661341323e-06, "loss": 0.2176, "step": 52620 }, { "epoch": 0.9572745957547427, "grad_norm": 0.9168427586555481, "learning_rate": 2.038268802319578e-06, "loss": 0.2232, "step": 52630 }, { "epoch": 0.957456483384565, "grad_norm": 1.0196940898895264, "learning_rate": 2.037944318155875e-06, "loss": 0.2187, "step": 52640 }, { "epoch": 0.9576383710143873, "grad_norm": 0.9977497458457947, "learning_rate": 2.03762120896282e-06, "loss": 0.2208, "step": 52650 }, { "epoch": 0.9578202586442096, "grad_norm": 0.9954604506492615, "learning_rate": 2.037299474852544e-06, "loss": 0.2158, "step": 52660 }, { "epoch": 0.9580021462740319, "grad_norm": 1.0454648733139038, "learning_rate": 2.0369791159366993e-06, "loss": 0.2134, "step": 52670 }, { "epoch": 0.9581840339038542, "grad_norm": 1.0757291316986084, "learning_rate": 2.036660132326462e-06, "loss": 0.2183, "step": 52680 }, { "epoch": 0.9583659215336765, "grad_norm": 1.0640476942062378, "learning_rate": 2.0363425241325306e-06, "loss": 0.2165, "step": 52690 }, { "epoch": 0.9585478091634988, "grad_norm": 0.997562050819397, "learning_rate": 2.036026291465125e-06, "loss": 0.2164, "step": 52700 }, { "epoch": 0.9587296967933211, "grad_norm": 0.9817377924919128, "learning_rate": 2.03571143443399e-06, "loss": 0.2168, "step": 52710 }, { "epoch": 0.9589115844231434, "grad_norm": 0.8699998259544373, "learning_rate": 2.035397953148392e-06, "loss": 0.218, "step": 52720 }, { "epoch": 0.9590934720529657, "grad_norm": 0.8602861762046814, "learning_rate": 2.035085847717119e-06, "loss": 0.2186, "step": 52730 }, { "epoch": 0.959275359682788, "grad_norm": 1.0599945783615112, "learning_rate": 2.034775118248483e-06, "loss": 0.223, "step": 52740 }, { "epoch": 0.9594572473126103, "grad_norm": 1.0542784929275513, "learning_rate": 2.034465764850318e-06, "loss": 0.2173, "step": 52750 }, { "epoch": 0.9596391349424326, "grad_norm": 0.9162759184837341, "learning_rate": 2.0341577876299797e-06, "loss": 0.2164, "step": 52760 }, { "epoch": 0.9598210225722549, "grad_norm": 0.9896430969238281, "learning_rate": 2.033851186694348e-06, "loss": 0.2135, "step": 52770 }, { "epoch": 0.9600029102020772, "grad_norm": 1.0408730506896973, "learning_rate": 2.0335459621498225e-06, "loss": 0.2216, "step": 52780 }, { "epoch": 0.9601847978318995, "grad_norm": 1.0322086811065674, "learning_rate": 2.0332421141023283e-06, "loss": 0.2183, "step": 52790 }, { "epoch": 0.9603666854617218, "grad_norm": 1.035201072692871, "learning_rate": 2.03293964265731e-06, "loss": 0.2138, "step": 52800 }, { "epoch": 0.9605485730915441, "grad_norm": 0.916583240032196, "learning_rate": 2.032638547919736e-06, "loss": 0.2114, "step": 52810 }, { "epoch": 0.9607304607213664, "grad_norm": 0.965078592300415, "learning_rate": 2.032338829994097e-06, "loss": 0.2191, "step": 52820 }, { "epoch": 0.9609123483511887, "grad_norm": 1.2432674169540405, "learning_rate": 2.0320404889844053e-06, "loss": 0.2239, "step": 52830 }, { "epoch": 0.961094235981011, "grad_norm": 1.0000189542770386, "learning_rate": 2.031743524994195e-06, "loss": 0.2159, "step": 52840 }, { "epoch": 0.9612761236108333, "grad_norm": 0.962149441242218, "learning_rate": 2.031447938126524e-06, "loss": 0.2241, "step": 52850 }, { "epoch": 0.9614580112406556, "grad_norm": 1.036426305770874, "learning_rate": 2.031153728483971e-06, "loss": 0.2183, "step": 52860 }, { "epoch": 0.9616398988704778, "grad_norm": 1.0210624933242798, "learning_rate": 2.0308608961686353e-06, "loss": 0.2222, "step": 52870 }, { "epoch": 0.9618217865003001, "grad_norm": 1.1824133396148682, "learning_rate": 2.030569441282142e-06, "loss": 0.2226, "step": 52880 }, { "epoch": 0.9620036741301224, "grad_norm": 1.1512562036514282, "learning_rate": 2.0302793639256347e-06, "loss": 0.213, "step": 52890 }, { "epoch": 0.9621855617599447, "grad_norm": 0.9639707803726196, "learning_rate": 2.029990664199781e-06, "loss": 0.2191, "step": 52900 }, { "epoch": 0.962367449389767, "grad_norm": 1.1163064241409302, "learning_rate": 2.0297033422047692e-06, "loss": 0.2178, "step": 52910 }, { "epoch": 0.9625493370195893, "grad_norm": 0.8838971853256226, "learning_rate": 2.0294173980403105e-06, "loss": 0.2134, "step": 52920 }, { "epoch": 0.9627312246494116, "grad_norm": 1.1131560802459717, "learning_rate": 2.0291328318056376e-06, "loss": 0.2168, "step": 52930 }, { "epoch": 0.9629131122792339, "grad_norm": 0.9595938920974731, "learning_rate": 2.028849643599504e-06, "loss": 0.2185, "step": 52940 }, { "epoch": 0.9630949999090562, "grad_norm": 0.8955169320106506, "learning_rate": 2.0285678335201863e-06, "loss": 0.2209, "step": 52950 }, { "epoch": 0.9632768875388785, "grad_norm": 0.9718515276908875, "learning_rate": 2.028287401665482e-06, "loss": 0.2166, "step": 52960 }, { "epoch": 0.9634587751687008, "grad_norm": 1.0113071203231812, "learning_rate": 2.0280083481327115e-06, "loss": 0.2183, "step": 52970 }, { "epoch": 0.9636406627985231, "grad_norm": 0.9132651686668396, "learning_rate": 2.027730673018715e-06, "loss": 0.2191, "step": 52980 }, { "epoch": 0.9638225504283454, "grad_norm": 1.0265698432922363, "learning_rate": 2.027454376419856e-06, "loss": 0.216, "step": 52990 }, { "epoch": 0.9640044380581677, "grad_norm": 1.2740031480789185, "learning_rate": 2.027179458432019e-06, "loss": 0.2156, "step": 53000 }, { "epoch": 0.96418632568799, "grad_norm": 1.0505378246307373, "learning_rate": 2.0269059191506105e-06, "loss": 0.2205, "step": 53010 }, { "epoch": 0.9643682133178123, "grad_norm": 1.1327850818634033, "learning_rate": 2.026633758670557e-06, "loss": 0.2186, "step": 53020 }, { "epoch": 0.9645501009476346, "grad_norm": 0.9723937511444092, "learning_rate": 2.026362977086309e-06, "loss": 0.2219, "step": 53030 }, { "epoch": 0.9647319885774569, "grad_norm": 0.9673178791999817, "learning_rate": 2.0260935744918358e-06, "loss": 0.2212, "step": 53040 }, { "epoch": 0.9649138762072792, "grad_norm": 0.9914476275444031, "learning_rate": 2.02582555098063e-06, "loss": 0.2235, "step": 53050 }, { "epoch": 0.9650957638371014, "grad_norm": 0.957523763179779, "learning_rate": 2.025558906645705e-06, "loss": 0.2224, "step": 53060 }, { "epoch": 0.9652776514669237, "grad_norm": 1.0088144540786743, "learning_rate": 2.025293641579596e-06, "loss": 0.2191, "step": 53070 }, { "epoch": 0.965459539096746, "grad_norm": 0.9872651696205139, "learning_rate": 2.025029755874358e-06, "loss": 0.2199, "step": 53080 }, { "epoch": 0.9656414267265683, "grad_norm": 1.03352952003479, "learning_rate": 2.02476724962157e-06, "loss": 0.2194, "step": 53090 }, { "epoch": 0.9658233143563906, "grad_norm": 1.006712555885315, "learning_rate": 2.0245061229123297e-06, "loss": 0.2142, "step": 53100 }, { "epoch": 0.9660052019862129, "grad_norm": 1.0149028301239014, "learning_rate": 2.0242463758372576e-06, "loss": 0.2281, "step": 53110 }, { "epoch": 0.9661870896160352, "grad_norm": 1.0771595239639282, "learning_rate": 2.0239880084864948e-06, "loss": 0.2215, "step": 53120 }, { "epoch": 0.9663689772458575, "grad_norm": 1.046921968460083, "learning_rate": 2.023731020949704e-06, "loss": 0.2208, "step": 53130 }, { "epoch": 0.9665508648756798, "grad_norm": 0.9715319275856018, "learning_rate": 2.023475413316068e-06, "loss": 0.2177, "step": 53140 }, { "epoch": 0.9667327525055021, "grad_norm": 1.0507829189300537, "learning_rate": 2.023221185674293e-06, "loss": 0.2165, "step": 53150 }, { "epoch": 0.9669146401353244, "grad_norm": 1.0635485649108887, "learning_rate": 2.022968338112603e-06, "loss": 0.2154, "step": 53160 }, { "epoch": 0.9670965277651467, "grad_norm": 1.1850779056549072, "learning_rate": 2.0227168707187457e-06, "loss": 0.2212, "step": 53170 }, { "epoch": 0.967278415394969, "grad_norm": 1.0439897775650024, "learning_rate": 2.0224667835799893e-06, "loss": 0.2165, "step": 53180 }, { "epoch": 0.9674603030247912, "grad_norm": 1.043090581893921, "learning_rate": 2.022218076783122e-06, "loss": 0.2188, "step": 53190 }, { "epoch": 0.9676421906546135, "grad_norm": 0.9684791564941406, "learning_rate": 2.0219707504144543e-06, "loss": 0.218, "step": 53200 }, { "epoch": 0.9678240782844358, "grad_norm": 0.9676417708396912, "learning_rate": 2.0217248045598175e-06, "loss": 0.2138, "step": 53210 }, { "epoch": 0.9680059659142581, "grad_norm": 1.0374157428741455, "learning_rate": 2.0214802393045617e-06, "loss": 0.2198, "step": 53220 }, { "epoch": 0.9681878535440804, "grad_norm": 1.0933938026428223, "learning_rate": 2.021237054733561e-06, "loss": 0.2222, "step": 53230 }, { "epoch": 0.9683697411739027, "grad_norm": 1.0850520133972168, "learning_rate": 2.0209952509312083e-06, "loss": 0.2146, "step": 53240 }, { "epoch": 0.968551628803725, "grad_norm": 1.0094233751296997, "learning_rate": 2.020754827981418e-06, "loss": 0.2153, "step": 53250 }, { "epoch": 0.9687335164335473, "grad_norm": 1.0251014232635498, "learning_rate": 2.0205157859676246e-06, "loss": 0.2158, "step": 53260 }, { "epoch": 0.9689154040633696, "grad_norm": 0.954368531703949, "learning_rate": 2.020278124972785e-06, "loss": 0.2146, "step": 53270 }, { "epoch": 0.9690972916931919, "grad_norm": 1.0515717267990112, "learning_rate": 2.0200418450793757e-06, "loss": 0.2215, "step": 53280 }, { "epoch": 0.9692791793230142, "grad_norm": 0.9248121380805969, "learning_rate": 2.0198069463693933e-06, "loss": 0.2167, "step": 53290 }, { "epoch": 0.9694610669528365, "grad_norm": 1.0220775604248047, "learning_rate": 2.019573428924356e-06, "loss": 0.2175, "step": 53300 }, { "epoch": 0.9696429545826588, "grad_norm": 0.9673506021499634, "learning_rate": 2.0193412928253025e-06, "loss": 0.2167, "step": 53310 }, { "epoch": 0.9698248422124811, "grad_norm": 0.9527473449707031, "learning_rate": 2.019110538152792e-06, "loss": 0.2187, "step": 53320 }, { "epoch": 0.9700067298423034, "grad_norm": 1.0482168197631836, "learning_rate": 2.018881164986905e-06, "loss": 0.2171, "step": 53330 }, { "epoch": 0.9701886174721257, "grad_norm": 0.8957347869873047, "learning_rate": 2.018653173407241e-06, "loss": 0.2133, "step": 53340 }, { "epoch": 0.970370505101948, "grad_norm": 0.9840524196624756, "learning_rate": 2.018426563492922e-06, "loss": 0.2226, "step": 53350 }, { "epoch": 0.9705523927317703, "grad_norm": 1.0724962949752808, "learning_rate": 2.018201335322588e-06, "loss": 0.2276, "step": 53360 }, { "epoch": 0.9707342803615926, "grad_norm": 0.9534935355186462, "learning_rate": 2.0179774889744026e-06, "loss": 0.2154, "step": 53370 }, { "epoch": 0.9709161679914149, "grad_norm": 1.1135156154632568, "learning_rate": 2.0177550245260476e-06, "loss": 0.218, "step": 53380 }, { "epoch": 0.9710980556212372, "grad_norm": 1.1064800024032593, "learning_rate": 2.0175339420547254e-06, "loss": 0.2196, "step": 53390 }, { "epoch": 0.9712799432510595, "grad_norm": 0.9510077238082886, "learning_rate": 2.0173142416371602e-06, "loss": 0.2155, "step": 53400 }, { "epoch": 0.9714618308808818, "grad_norm": 1.0378668308258057, "learning_rate": 2.017095923349595e-06, "loss": 0.223, "step": 53410 }, { "epoch": 0.9716437185107041, "grad_norm": 0.9340543746948242, "learning_rate": 2.016878987267794e-06, "loss": 0.2144, "step": 53420 }, { "epoch": 0.9718256061405264, "grad_norm": 1.2110066413879395, "learning_rate": 2.0166634334670408e-06, "loss": 0.217, "step": 53430 }, { "epoch": 0.9720074937703487, "grad_norm": 1.01598060131073, "learning_rate": 2.0164492620221415e-06, "loss": 0.219, "step": 53440 }, { "epoch": 0.972189381400171, "grad_norm": 0.9808478951454163, "learning_rate": 2.01623647300742e-06, "loss": 0.2174, "step": 53450 }, { "epoch": 0.9723712690299933, "grad_norm": 1.0272045135498047, "learning_rate": 2.0160250664967216e-06, "loss": 0.2179, "step": 53460 }, { "epoch": 0.9725531566598156, "grad_norm": 1.1524288654327393, "learning_rate": 2.0158150425634124e-06, "loss": 0.2175, "step": 53470 }, { "epoch": 0.9727350442896379, "grad_norm": 1.0134477615356445, "learning_rate": 2.015606401280377e-06, "loss": 0.2181, "step": 53480 }, { "epoch": 0.9729169319194602, "grad_norm": 0.9886091947555542, "learning_rate": 2.0153991427200215e-06, "loss": 0.2151, "step": 53490 }, { "epoch": 0.9730988195492825, "grad_norm": 1.1076860427856445, "learning_rate": 2.015193266954272e-06, "loss": 0.212, "step": 53500 }, { "epoch": 0.9732807071791048, "grad_norm": 0.911866307258606, "learning_rate": 2.0149887740545747e-06, "loss": 0.2195, "step": 53510 }, { "epoch": 0.973462594808927, "grad_norm": 0.9286797642707825, "learning_rate": 2.014785664091895e-06, "loss": 0.2229, "step": 53520 }, { "epoch": 0.9736444824387493, "grad_norm": 1.083666205406189, "learning_rate": 2.0145839371367194e-06, "loss": 0.2179, "step": 53530 }, { "epoch": 0.9738263700685716, "grad_norm": 1.0180327892303467, "learning_rate": 2.0143835932590543e-06, "loss": 0.2139, "step": 53540 }, { "epoch": 0.9740082576983939, "grad_norm": 1.0086389780044556, "learning_rate": 2.014184632528426e-06, "loss": 0.218, "step": 53550 }, { "epoch": 0.9741901453282162, "grad_norm": 1.0166265964508057, "learning_rate": 2.0139870550138814e-06, "loss": 0.2202, "step": 53560 }, { "epoch": 0.9743720329580385, "grad_norm": 0.9812893867492676, "learning_rate": 2.0137908607839853e-06, "loss": 0.2182, "step": 53570 }, { "epoch": 0.9745539205878608, "grad_norm": 0.9694347381591797, "learning_rate": 2.0135960499068246e-06, "loss": 0.2151, "step": 53580 }, { "epoch": 0.9747358082176831, "grad_norm": 1.0540918111801147, "learning_rate": 2.0134026224500054e-06, "loss": 0.214, "step": 53590 }, { "epoch": 0.9749176958475054, "grad_norm": 0.9678305983543396, "learning_rate": 2.013210578480654e-06, "loss": 0.2173, "step": 53600 }, { "epoch": 0.9750995834773277, "grad_norm": 1.2849807739257812, "learning_rate": 2.0130199180654164e-06, "loss": 0.2177, "step": 53610 }, { "epoch": 0.97528147110715, "grad_norm": 0.9601233601570129, "learning_rate": 2.0128306412704577e-06, "loss": 0.2132, "step": 53620 }, { "epoch": 0.9754633587369723, "grad_norm": 0.9379364252090454, "learning_rate": 2.012642748161464e-06, "loss": 0.2199, "step": 53630 }, { "epoch": 0.9756452463667946, "grad_norm": 1.0932776927947998, "learning_rate": 2.012456238803641e-06, "loss": 0.2212, "step": 53640 }, { "epoch": 0.9758271339966169, "grad_norm": 0.963916003704071, "learning_rate": 2.012271113261713e-06, "loss": 0.2149, "step": 53650 }, { "epoch": 0.9760090216264392, "grad_norm": 1.0278346538543701, "learning_rate": 2.0120873715999257e-06, "loss": 0.2175, "step": 53660 }, { "epoch": 0.9761909092562615, "grad_norm": 1.002237319946289, "learning_rate": 2.011905013882044e-06, "loss": 0.2189, "step": 53670 }, { "epoch": 0.9763727968860838, "grad_norm": 1.2026863098144531, "learning_rate": 2.011724040171351e-06, "loss": 0.22, "step": 53680 }, { "epoch": 0.9765546845159061, "grad_norm": 0.8771064281463623, "learning_rate": 2.0115444505306524e-06, "loss": 0.2158, "step": 53690 }, { "epoch": 0.9767365721457284, "grad_norm": 0.9433237314224243, "learning_rate": 2.0113662450222716e-06, "loss": 0.2156, "step": 53700 }, { "epoch": 0.9769184597755507, "grad_norm": 1.0158779621124268, "learning_rate": 2.011189423708052e-06, "loss": 0.2199, "step": 53710 }, { "epoch": 0.977100347405373, "grad_norm": 1.2047250270843506, "learning_rate": 2.0110139866493562e-06, "loss": 0.216, "step": 53720 }, { "epoch": 0.9772822350351953, "grad_norm": 1.0479954481124878, "learning_rate": 2.0108399339070675e-06, "loss": 0.2216, "step": 53730 }, { "epoch": 0.9774641226650176, "grad_norm": 1.0611143112182617, "learning_rate": 2.0106672655415884e-06, "loss": 0.2196, "step": 53740 }, { "epoch": 0.9776460102948399, "grad_norm": 0.9943484663963318, "learning_rate": 2.0104959816128404e-06, "loss": 0.2183, "step": 53750 }, { "epoch": 0.9778278979246622, "grad_norm": 0.9850167036056519, "learning_rate": 2.010326082180265e-06, "loss": 0.2181, "step": 53760 }, { "epoch": 0.9780097855544845, "grad_norm": 1.0748499631881714, "learning_rate": 2.0101575673028238e-06, "loss": 0.2227, "step": 53770 }, { "epoch": 0.9781916731843068, "grad_norm": 0.9561355710029602, "learning_rate": 2.0099904370389963e-06, "loss": 0.213, "step": 53780 }, { "epoch": 0.9783735608141291, "grad_norm": 1.1529735326766968, "learning_rate": 2.0098246914467833e-06, "loss": 0.2186, "step": 53790 }, { "epoch": 0.9785554484439514, "grad_norm": 0.9216739535331726, "learning_rate": 2.0096603305837043e-06, "loss": 0.2226, "step": 53800 }, { "epoch": 0.9787373360737737, "grad_norm": 0.9457110166549683, "learning_rate": 2.009497354506797e-06, "loss": 0.212, "step": 53810 }, { "epoch": 0.978919223703596, "grad_norm": 1.0063366889953613, "learning_rate": 2.009335763272621e-06, "loss": 0.2211, "step": 53820 }, { "epoch": 0.9791011113334183, "grad_norm": 1.017280101776123, "learning_rate": 2.009175556937254e-06, "loss": 0.217, "step": 53830 }, { "epoch": 0.9792829989632406, "grad_norm": 0.9983012080192566, "learning_rate": 2.0090167355562927e-06, "loss": 0.2191, "step": 53840 }, { "epoch": 0.9794648865930629, "grad_norm": 0.9945818781852722, "learning_rate": 2.008859299184854e-06, "loss": 0.2134, "step": 53850 }, { "epoch": 0.9796467742228852, "grad_norm": 0.8834373354911804, "learning_rate": 2.0087032478775735e-06, "loss": 0.2185, "step": 53860 }, { "epoch": 0.9798286618527074, "grad_norm": 0.9494906067848206, "learning_rate": 2.008548581688606e-06, "loss": 0.216, "step": 53870 }, { "epoch": 0.9800105494825297, "grad_norm": 1.0739188194274902, "learning_rate": 2.0083953006716264e-06, "loss": 0.2185, "step": 53880 }, { "epoch": 0.980192437112352, "grad_norm": 1.047847032546997, "learning_rate": 2.0082434048798292e-06, "loss": 0.2209, "step": 53890 }, { "epoch": 0.9803743247421742, "grad_norm": 1.0239135026931763, "learning_rate": 2.008092894365927e-06, "loss": 0.226, "step": 53900 }, { "epoch": 0.9805562123719965, "grad_norm": 1.0239651203155518, "learning_rate": 2.007943769182151e-06, "loss": 0.2196, "step": 53910 }, { "epoch": 0.9807381000018188, "grad_norm": 1.1065601110458374, "learning_rate": 2.007796029380255e-06, "loss": 0.218, "step": 53920 }, { "epoch": 0.9809199876316411, "grad_norm": 0.9932688474655151, "learning_rate": 2.007649675011508e-06, "loss": 0.2206, "step": 53930 }, { "epoch": 0.9811018752614634, "grad_norm": 1.0820831060409546, "learning_rate": 2.0075047061267015e-06, "loss": 0.2206, "step": 53940 }, { "epoch": 0.9812837628912857, "grad_norm": 1.0632272958755493, "learning_rate": 2.0073611227761433e-06, "loss": 0.2174, "step": 53950 }, { "epoch": 0.981465650521108, "grad_norm": 0.9828553199768066, "learning_rate": 2.007218925009663e-06, "loss": 0.2185, "step": 53960 }, { "epoch": 0.9816475381509303, "grad_norm": 1.0186635255813599, "learning_rate": 2.007078112876607e-06, "loss": 0.2226, "step": 53970 }, { "epoch": 0.9818294257807526, "grad_norm": 1.0764888525009155, "learning_rate": 2.006938686425843e-06, "loss": 0.2191, "step": 53980 }, { "epoch": 0.9820113134105749, "grad_norm": 0.9762018322944641, "learning_rate": 2.006800645705757e-06, "loss": 0.2208, "step": 53990 }, { "epoch": 0.9821932010403972, "grad_norm": 1.0159308910369873, "learning_rate": 2.0066639907642522e-06, "loss": 0.2181, "step": 54000 }, { "epoch": 0.9823750886702195, "grad_norm": 0.9960979223251343, "learning_rate": 2.006528721648755e-06, "loss": 0.2184, "step": 54010 }, { "epoch": 0.9825569763000418, "grad_norm": 0.9580124020576477, "learning_rate": 2.0063948384062066e-06, "loss": 0.2172, "step": 54020 }, { "epoch": 0.9827388639298641, "grad_norm": 1.0285533666610718, "learning_rate": 2.006262341083069e-06, "loss": 0.2193, "step": 54030 }, { "epoch": 0.9829207515596864, "grad_norm": 1.0161300897598267, "learning_rate": 2.006131229725325e-06, "loss": 0.2149, "step": 54040 }, { "epoch": 0.9831026391895087, "grad_norm": 1.0348050594329834, "learning_rate": 2.0060015043784745e-06, "loss": 0.2177, "step": 54050 }, { "epoch": 0.983284526819331, "grad_norm": 0.937641978263855, "learning_rate": 2.0058731650875354e-06, "loss": 0.2187, "step": 54060 }, { "epoch": 0.9834664144491533, "grad_norm": 1.0032374858856201, "learning_rate": 2.0057462118970466e-06, "loss": 0.2173, "step": 54070 }, { "epoch": 0.9836483020789756, "grad_norm": 1.0276998281478882, "learning_rate": 2.0056206448510647e-06, "loss": 0.2176, "step": 54080 }, { "epoch": 0.9838301897087979, "grad_norm": 1.012143611907959, "learning_rate": 2.0054964639931674e-06, "loss": 0.2173, "step": 54090 }, { "epoch": 0.9840120773386202, "grad_norm": 0.9660089015960693, "learning_rate": 2.005373669366448e-06, "loss": 0.2136, "step": 54100 }, { "epoch": 0.9841939649684425, "grad_norm": 0.9792198538780212, "learning_rate": 2.0052522610135212e-06, "loss": 0.2173, "step": 54110 }, { "epoch": 0.9843758525982648, "grad_norm": 0.95306795835495, "learning_rate": 2.00513223897652e-06, "loss": 0.2119, "step": 54120 }, { "epoch": 0.9845577402280871, "grad_norm": 1.0557807683944702, "learning_rate": 2.0050136032970966e-06, "loss": 0.2178, "step": 54130 }, { "epoch": 0.9847396278579094, "grad_norm": 0.9953875541687012, "learning_rate": 2.0048963540164208e-06, "loss": 0.2188, "step": 54140 }, { "epoch": 0.9849215154877317, "grad_norm": 1.0351041555404663, "learning_rate": 2.0047804911751824e-06, "loss": 0.214, "step": 54150 }, { "epoch": 0.985103403117554, "grad_norm": 0.998131275177002, "learning_rate": 2.00466601481359e-06, "loss": 0.2149, "step": 54160 }, { "epoch": 0.9852852907473763, "grad_norm": 1.0050764083862305, "learning_rate": 2.004552924971371e-06, "loss": 0.2147, "step": 54170 }, { "epoch": 0.9854671783771985, "grad_norm": 0.9494879841804504, "learning_rate": 2.004441221687772e-06, "loss": 0.2142, "step": 54180 }, { "epoch": 0.9856490660070208, "grad_norm": 1.0352630615234375, "learning_rate": 2.0043309050015566e-06, "loss": 0.2166, "step": 54190 }, { "epoch": 0.9858309536368431, "grad_norm": 1.0083630084991455, "learning_rate": 2.0042219749510093e-06, "loss": 0.2148, "step": 54200 }, { "epoch": 0.9860128412666654, "grad_norm": 0.9694600105285645, "learning_rate": 2.0041144315739325e-06, "loss": 0.2154, "step": 54210 }, { "epoch": 0.9861947288964877, "grad_norm": 1.036368489265442, "learning_rate": 2.0040082749076474e-06, "loss": 0.2165, "step": 54220 }, { "epoch": 0.98637661652631, "grad_norm": 0.9371940493583679, "learning_rate": 2.003903504988994e-06, "loss": 0.2123, "step": 54230 }, { "epoch": 0.9865585041561323, "grad_norm": 1.0884873867034912, "learning_rate": 2.0038001218543315e-06, "loss": 0.2187, "step": 54240 }, { "epoch": 0.9867403917859546, "grad_norm": 1.01853609085083, "learning_rate": 2.003698125539537e-06, "loss": 0.2196, "step": 54250 }, { "epoch": 0.9869222794157769, "grad_norm": 1.07811439037323, "learning_rate": 2.0035975160800066e-06, "loss": 0.2207, "step": 54260 }, { "epoch": 0.9871041670455992, "grad_norm": 1.0956552028656006, "learning_rate": 2.0034982935106563e-06, "loss": 0.2228, "step": 54270 }, { "epoch": 0.9872860546754215, "grad_norm": 0.9440680146217346, "learning_rate": 2.0034004578659184e-06, "loss": 0.2127, "step": 54280 }, { "epoch": 0.9874679423052438, "grad_norm": 0.9131145477294922, "learning_rate": 2.0033040091797463e-06, "loss": 0.2127, "step": 54290 }, { "epoch": 0.9876498299350661, "grad_norm": 1.068057656288147, "learning_rate": 2.0032089474856104e-06, "loss": 0.2211, "step": 54300 }, { "epoch": 0.9878317175648884, "grad_norm": 0.8812510967254639, "learning_rate": 2.0031152728165002e-06, "loss": 0.2127, "step": 54310 }, { "epoch": 0.9880136051947107, "grad_norm": 0.9299179315567017, "learning_rate": 2.0030229852049248e-06, "loss": 0.2143, "step": 54320 }, { "epoch": 0.988195492824533, "grad_norm": 1.0018349885940552, "learning_rate": 2.0029320846829113e-06, "loss": 0.2139, "step": 54330 }, { "epoch": 0.9883773804543553, "grad_norm": 1.131984829902649, "learning_rate": 2.0028425712820044e-06, "loss": 0.2181, "step": 54340 }, { "epoch": 0.9885592680841776, "grad_norm": 0.8518328666687012, "learning_rate": 2.0027544450332693e-06, "loss": 0.2139, "step": 54350 }, { "epoch": 0.9887411557139999, "grad_norm": 1.1199736595153809, "learning_rate": 2.002667705967288e-06, "loss": 0.2161, "step": 54360 }, { "epoch": 0.9889230433438222, "grad_norm": 1.0543639659881592, "learning_rate": 2.002582354114163e-06, "loss": 0.2191, "step": 54370 }, { "epoch": 0.9891049309736445, "grad_norm": 0.9061278104782104, "learning_rate": 2.002498389503513e-06, "loss": 0.2116, "step": 54380 }, { "epoch": 0.9892868186034668, "grad_norm": 0.9008918404579163, "learning_rate": 2.002415812164478e-06, "loss": 0.2118, "step": 54390 }, { "epoch": 0.9894687062332891, "grad_norm": 1.02567458152771, "learning_rate": 2.002334622125714e-06, "loss": 0.2158, "step": 54400 }, { "epoch": 0.9896505938631114, "grad_norm": 0.9914021492004395, "learning_rate": 2.002254819415398e-06, "loss": 0.2143, "step": 54410 }, { "epoch": 0.9898324814929337, "grad_norm": 0.9069268107414246, "learning_rate": 2.002176404061223e-06, "loss": 0.2133, "step": 54420 }, { "epoch": 0.990014369122756, "grad_norm": 0.9920116066932678, "learning_rate": 2.0020993760904027e-06, "loss": 0.2127, "step": 54430 }, { "epoch": 0.9901962567525783, "grad_norm": 0.9922382235527039, "learning_rate": 2.0020237355296686e-06, "loss": 0.2129, "step": 54440 }, { "epoch": 0.9903781443824006, "grad_norm": 0.9771965146064758, "learning_rate": 2.0019494824052697e-06, "loss": 0.2144, "step": 54450 }, { "epoch": 0.9905600320122229, "grad_norm": 1.0094854831695557, "learning_rate": 2.0018766167429755e-06, "loss": 0.2196, "step": 54460 }, { "epoch": 0.9907419196420452, "grad_norm": 0.9582799673080444, "learning_rate": 2.001805138568072e-06, "loss": 0.2207, "step": 54470 }, { "epoch": 0.9909238072718675, "grad_norm": 1.0055713653564453, "learning_rate": 2.001735047905365e-06, "loss": 0.2198, "step": 54480 }, { "epoch": 0.9911056949016898, "grad_norm": 1.0134953260421753, "learning_rate": 2.0016663447791785e-06, "loss": 0.2163, "step": 54490 }, { "epoch": 0.9912875825315121, "grad_norm": 1.0230180025100708, "learning_rate": 2.001599029213355e-06, "loss": 0.219, "step": 54500 }, { "epoch": 0.9914694701613344, "grad_norm": 0.9659442901611328, "learning_rate": 2.0015331012312547e-06, "loss": 0.2167, "step": 54510 }, { "epoch": 0.9916513577911567, "grad_norm": 1.1874651908874512, "learning_rate": 2.0014685608557575e-06, "loss": 0.2176, "step": 54520 }, { "epoch": 0.991833245420979, "grad_norm": 0.9718601107597351, "learning_rate": 2.00140540810926e-06, "loss": 0.2127, "step": 54530 }, { "epoch": 0.9920151330508012, "grad_norm": 1.021325945854187, "learning_rate": 2.0013436430136802e-06, "loss": 0.2146, "step": 54540 }, { "epoch": 0.9921970206806235, "grad_norm": 1.0644222497940063, "learning_rate": 2.0012832655904522e-06, "loss": 0.2235, "step": 54550 }, { "epoch": 0.9923789083104458, "grad_norm": 0.9746378660202026, "learning_rate": 2.001224275860528e-06, "loss": 0.2179, "step": 54560 }, { "epoch": 0.9925607959402681, "grad_norm": 0.9416080117225647, "learning_rate": 2.00116667384438e-06, "loss": 0.2204, "step": 54570 }, { "epoch": 0.9927426835700904, "grad_norm": 0.8720496296882629, "learning_rate": 2.001110459561998e-06, "loss": 0.2169, "step": 54580 }, { "epoch": 0.9929245711999127, "grad_norm": 0.9769662618637085, "learning_rate": 2.00105563303289e-06, "loss": 0.2166, "step": 54590 }, { "epoch": 0.993106458829735, "grad_norm": 0.9483774304389954, "learning_rate": 2.001002194276083e-06, "loss": 0.2186, "step": 54600 }, { "epoch": 0.9932883464595573, "grad_norm": 1.0602209568023682, "learning_rate": 2.0009501433101217e-06, "loss": 0.2199, "step": 54610 }, { "epoch": 0.9934702340893796, "grad_norm": 0.908362090587616, "learning_rate": 2.00089948015307e-06, "loss": 0.2199, "step": 54620 }, { "epoch": 0.9936521217192019, "grad_norm": 1.1202993392944336, "learning_rate": 2.0008502048225096e-06, "loss": 0.2192, "step": 54630 }, { "epoch": 0.9938340093490242, "grad_norm": 0.9329814314842224, "learning_rate": 2.0008023173355405e-06, "loss": 0.213, "step": 54640 }, { "epoch": 0.9940158969788465, "grad_norm": 1.0682350397109985, "learning_rate": 2.0007558177087825e-06, "loss": 0.2226, "step": 54650 }, { "epoch": 0.9941977846086688, "grad_norm": 1.0304162502288818, "learning_rate": 2.000710705958371e-06, "loss": 0.2157, "step": 54660 }, { "epoch": 0.9943796722384911, "grad_norm": 0.9822694659233093, "learning_rate": 2.0006669820999622e-06, "loss": 0.2179, "step": 54670 }, { "epoch": 0.9945615598683134, "grad_norm": 0.9626803994178772, "learning_rate": 2.00062464614873e-06, "loss": 0.212, "step": 54680 }, { "epoch": 0.9947434474981357, "grad_norm": 1.0234310626983643, "learning_rate": 2.000583698119365e-06, "loss": 0.2144, "step": 54690 }, { "epoch": 0.994925335127958, "grad_norm": 1.0375109910964966, "learning_rate": 2.0005441380260797e-06, "loss": 0.2117, "step": 54700 }, { "epoch": 0.9951072227577803, "grad_norm": 0.9325792789459229, "learning_rate": 2.0005059658826016e-06, "loss": 0.219, "step": 54710 }, { "epoch": 0.9952891103876026, "grad_norm": 1.0958082675933838, "learning_rate": 2.0004691817021777e-06, "loss": 0.2139, "step": 54720 }, { "epoch": 0.9954709980174248, "grad_norm": 0.9188241958618164, "learning_rate": 2.0004337854975734e-06, "loss": 0.2136, "step": 54730 }, { "epoch": 0.9956528856472471, "grad_norm": 0.9681688547134399, "learning_rate": 2.0003997772810734e-06, "loss": 0.2193, "step": 54740 }, { "epoch": 0.9958347732770694, "grad_norm": 1.086104393005371, "learning_rate": 2.000367157064478e-06, "loss": 0.2201, "step": 54750 }, { "epoch": 0.9960166609068917, "grad_norm": 1.0971709489822388, "learning_rate": 2.000335924859109e-06, "loss": 0.2188, "step": 54760 }, { "epoch": 0.996198548536714, "grad_norm": 0.8964868783950806, "learning_rate": 2.000306080675804e-06, "loss": 0.2133, "step": 54770 }, { "epoch": 0.9963804361665363, "grad_norm": 1.0347673892974854, "learning_rate": 2.000277624524922e-06, "loss": 0.2195, "step": 54780 }, { "epoch": 0.9965623237963586, "grad_norm": 0.928534984588623, "learning_rate": 2.0002505564163357e-06, "loss": 0.214, "step": 54790 }, { "epoch": 0.9967442114261809, "grad_norm": 1.0491340160369873, "learning_rate": 2.0002248763594397e-06, "loss": 0.2148, "step": 54800 }, { "epoch": 0.9969260990560032, "grad_norm": 1.040202260017395, "learning_rate": 2.000200584363146e-06, "loss": 0.2165, "step": 54810 }, { "epoch": 0.9971079866858255, "grad_norm": 1.1188472509384155, "learning_rate": 2.000177680435885e-06, "loss": 0.218, "step": 54820 }, { "epoch": 0.9972898743156478, "grad_norm": 0.9703745245933533, "learning_rate": 2.0001561645856044e-06, "loss": 0.2163, "step": 54830 }, { "epoch": 0.99747176194547, "grad_norm": 0.9982457160949707, "learning_rate": 2.000136036819772e-06, "loss": 0.2198, "step": 54840 }, { "epoch": 0.9976536495752923, "grad_norm": 1.065281867980957, "learning_rate": 2.0001172971453717e-06, "loss": 0.2153, "step": 54850 }, { "epoch": 0.9978355372051146, "grad_norm": 1.0456359386444092, "learning_rate": 2.000099945568908e-06, "loss": 0.2248, "step": 54860 }, { "epoch": 0.9980174248349369, "grad_norm": 1.0893419981002808, "learning_rate": 2.000083982096401e-06, "loss": 0.2172, "step": 54870 }, { "epoch": 0.9981993124647592, "grad_norm": 0.9799977540969849, "learning_rate": 2.0000694067333924e-06, "loss": 0.216, "step": 54880 }, { "epoch": 0.9983812000945815, "grad_norm": 0.9896212816238403, "learning_rate": 2.0000562194849385e-06, "loss": 0.2152, "step": 54890 }, { "epoch": 0.9985630877244038, "grad_norm": 1.068423867225647, "learning_rate": 2.0000444203556176e-06, "loss": 0.2161, "step": 54900 }, { "epoch": 0.9987449753542261, "grad_norm": 1.0909819602966309, "learning_rate": 2.0000340093495226e-06, "loss": 0.2168, "step": 54910 }, { "epoch": 0.9989268629840484, "grad_norm": 1.0839061737060547, "learning_rate": 2.0000249864702677e-06, "loss": 0.2194, "step": 54920 }, { "epoch": 0.9991087506138707, "grad_norm": 0.9454774260520935, "learning_rate": 2.0000173517209833e-06, "loss": 0.2151, "step": 54930 }, { "epoch": 0.999290638243693, "grad_norm": 0.9733779430389404, "learning_rate": 2.0000111051043196e-06, "loss": 0.2157, "step": 54940 }, { "epoch": 0.9994725258735153, "grad_norm": 1.121142864227295, "learning_rate": 2.000006246622444e-06, "loss": 0.22, "step": 54950 }, { "epoch": 0.9996544135033376, "grad_norm": 1.0394141674041748, "learning_rate": 2.000002776277043e-06, "loss": 0.2188, "step": 54960 }, { "epoch": 0.9998363011331599, "grad_norm": 1.0501214265823364, "learning_rate": 2.0000006940693213e-06, "loss": 0.2134, "step": 54970 }, { "epoch": 1.0, "step": 54979, "total_flos": 2980695134699520.0, "train_loss": 0.22231142784002364, "train_runtime": 336346.2144, "train_samples_per_second": 20.923, "train_steps_per_second": 0.163 } ], "logging_steps": 10, "max_steps": 54979, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 50000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 2980695134699520.0, "train_batch_size": 8, "trial_name": null, "trial_params": null }