{ "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": 25.52570152282715, "learning_rate": 3.1080186174614134e-06, "loss": 2.9286, "step": 10 }, { "epoch": 0.00036377525964459157, "grad_norm": 4.41333532333374, "learning_rate": 4.043625448399395e-06, "loss": 1.7375, "step": 20 }, { "epoch": 0.0005456628894668873, "grad_norm": 4.662846565246582, "learning_rate": 4.590920359916674e-06, "loss": 1.5079, "step": 30 }, { "epoch": 0.0007275505192891831, "grad_norm": 11.236649513244629, "learning_rate": 4.979232279337378e-06, "loss": 1.5, "step": 40 }, { "epoch": 0.000909438149111479, "grad_norm": 10.691333770751953, "learning_rate": 5.280430403984844e-06, "loss": 1.4118, "step": 50 }, { "epoch": 0.0010913257789337747, "grad_norm": 10.45896053314209, "learning_rate": 5.5265271908546555e-06, "loss": 1.4797, "step": 60 }, { "epoch": 0.0012732134087560705, "grad_norm": 9.242533683776855, "learning_rate": 5.734599059405875e-06, "loss": 1.3055, "step": 70 }, { "epoch": 0.0014551010385783663, "grad_norm": 8.152229309082031, "learning_rate": 5.914839110275359e-06, "loss": 1.3248, "step": 80 }, { "epoch": 0.001636988668400662, "grad_norm": 6.054175853729248, "learning_rate": 6.073822102371933e-06, "loss": 1.2695, "step": 90 }, { "epoch": 0.001818876298222958, "grad_norm": 6.100965976715088, "learning_rate": 6.216037234922827e-06, "loss": 1.2747, "step": 100 }, { "epoch": 0.0020007639280452535, "grad_norm": 4.275759220123291, "learning_rate": 6.34468647102131e-06, "loss": 1.2406, "step": 110 }, { "epoch": 0.0021826515578675493, "grad_norm": 3.17997145652771, "learning_rate": 6.462134021792637e-06, "loss": 1.2036, "step": 120 }, { "epoch": 0.002364539187689845, "grad_norm": 2.4806201457977295, "learning_rate": 6.570175295228441e-06, "loss": 1.1418, "step": 130 }, { "epoch": 0.002546426817512141, "grad_norm": 2.5229175090789795, "learning_rate": 6.670205890343856e-06, "loss": 1.1462, "step": 140 }, { "epoch": 0.0027283144473344368, "grad_norm": 2.527791976928711, "learning_rate": 6.763332146440104e-06, "loss": 1.1609, "step": 150 }, { "epoch": 0.0029102020771567326, "grad_norm": 3.125991106033325, "learning_rate": 6.8504459412133415e-06, "loss": 1.1488, "step": 160 }, { "epoch": 0.0030920897069790284, "grad_norm": 2.314793586730957, "learning_rate": 6.932276772940404e-06, "loss": 1.1911, "step": 170 }, { "epoch": 0.003273977336801324, "grad_norm": 2.5680654048919678, "learning_rate": 7.009428933309917e-06, "loss": 1.2174, "step": 180 }, { "epoch": 0.00345586496662362, "grad_norm": 2.252018690109253, "learning_rate": 7.082408616368629e-06, "loss": 1.1741, "step": 190 }, { "epoch": 0.003637752596445916, "grad_norm": 2.6119725704193115, "learning_rate": 7.151644065860809e-06, "loss": 1.1344, "step": 200 }, { "epoch": 0.0038196402262682116, "grad_norm": 2.5633702278137207, "learning_rate": 7.217500801861133e-06, "loss": 1.1387, "step": 210 }, { "epoch": 0.004001527856090507, "grad_norm": 2.0281286239624023, "learning_rate": 7.280293301959292e-06, "loss": 1.125, "step": 220 }, { "epoch": 0.004183415485912803, "grad_norm": 2.2588398456573486, "learning_rate": 7.340294083719535e-06, "loss": 1.0991, "step": 230 }, { "epoch": 0.004365303115735099, "grad_norm": 2.5346293449401855, "learning_rate": 7.397740852730619e-06, "loss": 1.1328, "step": 240 }, { "epoch": 0.004547190745557395, "grad_norm": 2.4390437602996826, "learning_rate": 7.452842190508276e-06, "loss": 1.1699, "step": 250 }, { "epoch": 0.00472907837537969, "grad_norm": 2.3022284507751465, "learning_rate": 7.5057821261664224e-06, "loss": 1.0707, "step": 260 }, { "epoch": 0.0049109660052019865, "grad_norm": 2.235152006149292, "learning_rate": 7.556723844827194e-06, "loss": 1.2179, "step": 270 }, { "epoch": 0.005092853635024282, "grad_norm": 2.500591516494751, "learning_rate": 7.6058127212818375e-06, "loss": 1.1368, "step": 280 }, { "epoch": 0.005274741264846578, "grad_norm": 2.3847954273223877, "learning_rate": 7.653178821069666e-06, "loss": 1.1978, "step": 290 }, { "epoch": 0.0054566288946688735, "grad_norm": 2.2241158485412598, "learning_rate": 7.698938977378086e-06, "loss": 1.0558, "step": 300 }, { "epoch": 0.00563851652449117, "grad_norm": 2.370058536529541, "learning_rate": 7.743198527267122e-06, "loss": 1.2129, "step": 310 }, { "epoch": 0.005820404154313465, "grad_norm": 2.3130123615264893, "learning_rate": 7.786052772151323e-06, "loss": 1.1383, "step": 320 }, { "epoch": 0.0060022917841357605, "grad_norm": 2.157639265060425, "learning_rate": 7.827588213476568e-06, "loss": 1.1908, "step": 330 }, { "epoch": 0.006184179413958057, "grad_norm": 2.4065439701080322, "learning_rate": 7.867883603878387e-06, "loss": 1.1235, "step": 340 }, { "epoch": 0.006366067043780352, "grad_norm": 2.1777374744415283, "learning_rate": 7.907010845929305e-06, "loss": 1.0693, "step": 350 }, { "epoch": 0.006547954673602648, "grad_norm": 2.665274143218994, "learning_rate": 7.945035764247898e-06, "loss": 1.1398, "step": 360 }, { "epoch": 0.006729842303424944, "grad_norm": 2.160505771636963, "learning_rate": 7.98201877179672e-06, "loss": 1.1131, "step": 370 }, { "epoch": 0.00691172993324724, "grad_norm": 2.5119383335113525, "learning_rate": 8.018015447306611e-06, "loss": 1.1596, "step": 380 }, { "epoch": 0.007093617563069535, "grad_norm": 2.418747901916504, "learning_rate": 8.053077037683701e-06, "loss": 1.1558, "step": 390 }, { "epoch": 0.007275505192891832, "grad_norm": 2.1949877738952637, "learning_rate": 8.08725089679879e-06, "loss": 1.1648, "step": 400 }, { "epoch": 0.007457392822714127, "grad_norm": 2.3044469356536865, "learning_rate": 8.120580870087571e-06, "loss": 1.11, "step": 410 }, { "epoch": 0.007639280452536423, "grad_norm": 2.097435235977173, "learning_rate": 8.153107632799117e-06, "loss": 1.1227, "step": 420 }, { "epoch": 0.007821168082358719, "grad_norm": 2.27219820022583, "learning_rate": 8.18486898843871e-06, "loss": 1.139, "step": 430 }, { "epoch": 0.008003055712181014, "grad_norm": 2.229113817214966, "learning_rate": 8.215900132897273e-06, "loss": 1.1217, "step": 440 }, { "epoch": 0.008184943342003311, "grad_norm": 2.1199800968170166, "learning_rate": 8.246233888895365e-06, "loss": 1.1604, "step": 450 }, { "epoch": 0.008366830971825607, "grad_norm": 2.1087863445281982, "learning_rate": 8.275900914657518e-06, "loss": 1.0971, "step": 460 }, { "epoch": 0.008548718601647902, "grad_norm": 2.3510258197784424, "learning_rate": 8.304929890142974e-06, "loss": 1.0879, "step": 470 }, { "epoch": 0.008730606231470197, "grad_norm": 2.3183298110961914, "learning_rate": 8.333347683668602e-06, "loss": 1.0899, "step": 480 }, { "epoch": 0.008912493861292493, "grad_norm": 2.7054505348205566, "learning_rate": 8.361179501350334e-06, "loss": 1.1005, "step": 490 }, { "epoch": 0.00909438149111479, "grad_norm": 2.290515899658203, "learning_rate": 8.388449021446257e-06, "loss": 1.1283, "step": 500 }, { "epoch": 0.009276269120937085, "grad_norm": 2.184718370437622, "learning_rate": 8.415178515395664e-06, "loss": 1.1275, "step": 510 }, { "epoch": 0.00945815675075938, "grad_norm": 2.231933116912842, "learning_rate": 8.441388957104405e-06, "loss": 1.1478, "step": 520 }, { "epoch": 0.009640044380581676, "grad_norm": 2.043821096420288, "learning_rate": 8.467100121820135e-06, "loss": 1.107, "step": 530 }, { "epoch": 0.009821932010403973, "grad_norm": 2.126964569091797, "learning_rate": 8.492330675765176e-06, "loss": 1.0849, "step": 540 }, { "epoch": 0.010003819640226268, "grad_norm": 2.070014476776123, "learning_rate": 8.51709825754474e-06, "loss": 1.1021, "step": 550 }, { "epoch": 0.010185707270048564, "grad_norm": 2.4281294345855713, "learning_rate": 8.541419552219821e-06, "loss": 1.1638, "step": 560 }, { "epoch": 0.01036759489987086, "grad_norm": 2.1897330284118652, "learning_rate": 8.56531035882389e-06, "loss": 1.1641, "step": 570 }, { "epoch": 0.010549482529693156, "grad_norm": 2.1623220443725586, "learning_rate": 8.588785652007649e-06, "loss": 1.1121, "step": 580 }, { "epoch": 0.010731370159515452, "grad_norm": 1.859534740447998, "learning_rate": 8.611859638414193e-06, "loss": 1.0443, "step": 590 }, { "epoch": 0.010913257789337747, "grad_norm": 2.305708885192871, "learning_rate": 8.634545808316069e-06, "loss": 1.0983, "step": 600 }, { "epoch": 0.011095145419160042, "grad_norm": 2.2217178344726562, "learning_rate": 8.656856982984191e-06, "loss": 1.1227, "step": 610 }, { "epoch": 0.01127703304898234, "grad_norm": 2.3903679847717285, "learning_rate": 8.678805358205103e-06, "loss": 1.1389, "step": 620 }, { "epoch": 0.011458920678804635, "grad_norm": 1.8387632369995117, "learning_rate": 8.700402544316394e-06, "loss": 1.0905, "step": 630 }, { "epoch": 0.01164080830862693, "grad_norm": 2.007169723510742, "learning_rate": 8.721659603089306e-06, "loss": 1.0935, "step": 640 }, { "epoch": 0.011822695938449226, "grad_norm": 2.7365691661834717, "learning_rate": 8.742587081751872e-06, "loss": 1.1236, "step": 650 }, { "epoch": 0.012004583568271521, "grad_norm": 2.1970369815826416, "learning_rate": 8.763195044414551e-06, "loss": 1.1369, "step": 660 }, { "epoch": 0.012186471198093818, "grad_norm": 2.2955799102783203, "learning_rate": 8.78349310113276e-06, "loss": 1.1189, "step": 670 }, { "epoch": 0.012368358827916114, "grad_norm": 2.0956482887268066, "learning_rate": 8.803490434816368e-06, "loss": 1.0898, "step": 680 }, { "epoch": 0.012550246457738409, "grad_norm": 2.1907670497894287, "learning_rate": 8.823195826174793e-06, "loss": 1.0911, "step": 690 }, { "epoch": 0.012732134087560704, "grad_norm": 2.2326161861419678, "learning_rate": 8.842617676867288e-06, "loss": 1.1408, "step": 700 }, { "epoch": 0.012914021717383001, "grad_norm": 2.154390811920166, "learning_rate": 8.861764031011172e-06, "loss": 1.1061, "step": 710 }, { "epoch": 0.013095909347205297, "grad_norm": 2.4368627071380615, "learning_rate": 8.88064259518588e-06, "loss": 1.1197, "step": 720 }, { "epoch": 0.013277796977027592, "grad_norm": 2.1372616291046143, "learning_rate": 8.89926075705724e-06, "loss": 1.1126, "step": 730 }, { "epoch": 0.013459684606849888, "grad_norm": 2.3460845947265625, "learning_rate": 8.917625602734702e-06, "loss": 1.1191, "step": 740 }, { "epoch": 0.013641572236672185, "grad_norm": 2.5043632984161377, "learning_rate": 8.935743932963536e-06, "loss": 1.1313, "step": 750 }, { "epoch": 0.01382345986649448, "grad_norm": 2.0339291095733643, "learning_rate": 8.953622278244594e-06, "loss": 1.1145, "step": 760 }, { "epoch": 0.014005347496316775, "grad_norm": 2.1438963413238525, "learning_rate": 8.971266912965769e-06, "loss": 1.1042, "step": 770 }, { "epoch": 0.01418723512613907, "grad_norm": 1.861353874206543, "learning_rate": 8.988683868621682e-06, "loss": 1.0491, "step": 780 }, { "epoch": 0.014369122755961366, "grad_norm": 2.371600866317749, "learning_rate": 9.005878946191255e-06, "loss": 1.125, "step": 790 }, { "epoch": 0.014551010385783663, "grad_norm": 1.6678138971328735, "learning_rate": 9.022857727736773e-06, "loss": 1.0407, "step": 800 }, { "epoch": 0.014732898015605959, "grad_norm": 2.4710280895233154, "learning_rate": 9.039625587282453e-06, "loss": 1.1209, "step": 810 }, { "epoch": 0.014914785645428254, "grad_norm": 1.9889713525772095, "learning_rate": 9.056187701025554e-06, "loss": 1.0456, "step": 820 }, { "epoch": 0.01509667327525055, "grad_norm": 2.253756046295166, "learning_rate": 9.072549056928585e-06, "loss": 1.135, "step": 830 }, { "epoch": 0.015278560905072847, "grad_norm": 2.099663496017456, "learning_rate": 9.088714463737098e-06, "loss": 1.094, "step": 840 }, { "epoch": 0.015460448534895142, "grad_norm": 2.239149332046509, "learning_rate": 9.104688559463835e-06, "loss": 1.1276, "step": 850 }, { "epoch": 0.015642336164717437, "grad_norm": 2.0930051803588867, "learning_rate": 9.120475819376691e-06, "loss": 1.0715, "step": 860 }, { "epoch": 0.015824223794539734, "grad_norm": 2.291969060897827, "learning_rate": 9.136080563524925e-06, "loss": 1.1242, "step": 870 }, { "epoch": 0.016006111424362028, "grad_norm": 1.9396501779556274, "learning_rate": 9.151506963835255e-06, "loss": 1.0648, "step": 880 }, { "epoch": 0.016187999054184325, "grad_norm": 2.4726831912994385, "learning_rate": 9.166759050807031e-06, "loss": 1.1543, "step": 890 }, { "epoch": 0.016369886684006622, "grad_norm": 2.2952706813812256, "learning_rate": 9.181840719833347e-06, "loss": 1.1238, "step": 900 }, { "epoch": 0.016551774313828916, "grad_norm": 2.1331019401550293, "learning_rate": 9.196755737172901e-06, "loss": 1.0868, "step": 910 }, { "epoch": 0.016733661943651213, "grad_norm": 2.1127774715423584, "learning_rate": 9.2115077455955e-06, "loss": 1.1265, "step": 920 }, { "epoch": 0.016915549573473507, "grad_norm": 2.196153402328491, "learning_rate": 9.226100269722382e-06, "loss": 1.056, "step": 930 }, { "epoch": 0.017097437203295804, "grad_norm": 2.084660053253174, "learning_rate": 9.240536721080955e-06, "loss": 1.0759, "step": 940 }, { "epoch": 0.0172793248331181, "grad_norm": 2.327160120010376, "learning_rate": 9.254820402892061e-06, "loss": 1.0906, "step": 950 }, { "epoch": 0.017461212462940395, "grad_norm": 2.200631856918335, "learning_rate": 9.268954514606585e-06, "loss": 1.1172, "step": 960 }, { "epoch": 0.01764310009276269, "grad_norm": 2.194944381713867, "learning_rate": 9.282942156207003e-06, "loss": 1.1024, "step": 970 }, { "epoch": 0.017824987722584985, "grad_norm": 2.226675510406494, "learning_rate": 9.296786332288317e-06, "loss": 1.079, "step": 980 }, { "epoch": 0.018006875352407282, "grad_norm": 2.1736607551574707, "learning_rate": 9.31048995593183e-06, "loss": 1.1187, "step": 990 }, { "epoch": 0.01818876298222958, "grad_norm": 2.387242555618286, "learning_rate": 9.32405585238424e-06, "loss": 1.1077, "step": 1000 }, { "epoch": 0.018370650612051873, "grad_norm": 1.8114150762557983, "learning_rate": 9.337486762553703e-06, "loss": 1.1239, "step": 1010 }, { "epoch": 0.01855253824187417, "grad_norm": 1.954827070236206, "learning_rate": 9.350785346333647e-06, "loss": 1.0637, "step": 1020 }, { "epoch": 0.018734425871696467, "grad_norm": 2.103395462036133, "learning_rate": 9.36395418576445e-06, "loss": 1.1761, "step": 1030 }, { "epoch": 0.01891631350151876, "grad_norm": 2.291724681854248, "learning_rate": 9.376995788042386e-06, "loss": 1.0797, "step": 1040 }, { "epoch": 0.019098201131341058, "grad_norm": 2.218689441680908, "learning_rate": 9.389912588384565e-06, "loss": 1.0911, "step": 1050 }, { "epoch": 0.019280088761163352, "grad_norm": 2.457857370376587, "learning_rate": 9.402706952758116e-06, "loss": 1.1306, "step": 1060 }, { "epoch": 0.01946197639098565, "grad_norm": 2.220818519592285, "learning_rate": 9.41538118048122e-06, "loss": 1.1598, "step": 1070 }, { "epoch": 0.019643864020807946, "grad_norm": 2.352590560913086, "learning_rate": 9.427937506703159e-06, "loss": 1.049, "step": 1080 }, { "epoch": 0.01982575165063024, "grad_norm": 1.9947540760040283, "learning_rate": 9.44037810477008e-06, "loss": 1.1102, "step": 1090 }, { "epoch": 0.020007639280452537, "grad_norm": 2.2105612754821777, "learning_rate": 9.452705088482722e-06, "loss": 1.1041, "step": 1100 }, { "epoch": 0.02018952691027483, "grad_norm": 2.3346028327941895, "learning_rate": 9.46492051425198e-06, "loss": 1.1259, "step": 1110 }, { "epoch": 0.020371414540097128, "grad_norm": 2.1265478134155273, "learning_rate": 9.477026383157802e-06, "loss": 1.1122, "step": 1120 }, { "epoch": 0.020553302169919425, "grad_norm": 1.8557378053665161, "learning_rate": 9.489024642916581e-06, "loss": 1.1446, "step": 1130 }, { "epoch": 0.02073518979974172, "grad_norm": 2.2170047760009766, "learning_rate": 9.500917189761871e-06, "loss": 1.0924, "step": 1140 }, { "epoch": 0.020917077429564016, "grad_norm": 2.569016456604004, "learning_rate": 9.512705870242966e-06, "loss": 1.0599, "step": 1150 }, { "epoch": 0.021098965059386313, "grad_norm": 2.2807047367095947, "learning_rate": 9.52439248294563e-06, "loss": 1.1035, "step": 1160 }, { "epoch": 0.021280852689208606, "grad_norm": 2.267941951751709, "learning_rate": 9.535978780138962e-06, "loss": 0.9948, "step": 1170 }, { "epoch": 0.021462740319030903, "grad_norm": 2.207578659057617, "learning_rate": 9.547466469352174e-06, "loss": 1.0577, "step": 1180 }, { "epoch": 0.021644627948853197, "grad_norm": 2.4154961109161377, "learning_rate": 9.558857214884864e-06, "loss": 1.1145, "step": 1190 }, { "epoch": 0.021826515578675494, "grad_norm": 2.261624813079834, "learning_rate": 9.570152639254052e-06, "loss": 1.101, "step": 1200 }, { "epoch": 0.02200840320849779, "grad_norm": 2.2443344593048096, "learning_rate": 9.581354324581205e-06, "loss": 1.109, "step": 1210 }, { "epoch": 0.022190290838320085, "grad_norm": 2.3841586112976074, "learning_rate": 9.592463813922174e-06, "loss": 1.1066, "step": 1220 }, { "epoch": 0.022372178468142382, "grad_norm": 2.0505576133728027, "learning_rate": 9.603482612542831e-06, "loss": 1.1092, "step": 1230 }, { "epoch": 0.02255406609796468, "grad_norm": 2.4716007709503174, "learning_rate": 9.614412189143086e-06, "loss": 1.1073, "step": 1240 }, { "epoch": 0.022735953727786973, "grad_norm": 2.4736626148223877, "learning_rate": 9.625253977031707e-06, "loss": 1.1065, "step": 1250 }, { "epoch": 0.02291784135760927, "grad_norm": 2.082811117172241, "learning_rate": 9.636009375254376e-06, "loss": 1.0555, "step": 1260 }, { "epoch": 0.023099728987431564, "grad_norm": 1.8840736150741577, "learning_rate": 9.646679749677123e-06, "loss": 1.0615, "step": 1270 }, { "epoch": 0.02328161661725386, "grad_norm": 1.953975796699524, "learning_rate": 9.657266434027289e-06, "loss": 1.0571, "step": 1280 }, { "epoch": 0.023463504247076158, "grad_norm": 1.9492075443267822, "learning_rate": 9.66777073089397e-06, "loss": 1.104, "step": 1290 }, { "epoch": 0.02364539187689845, "grad_norm": 2.0191433429718018, "learning_rate": 9.678193912689853e-06, "loss": 1.1043, "step": 1300 }, { "epoch": 0.02382727950672075, "grad_norm": 2.2549564838409424, "learning_rate": 9.688537222576177e-06, "loss": 1.1231, "step": 1310 }, { "epoch": 0.024009167136543042, "grad_norm": 2.2347633838653564, "learning_rate": 9.698801875352534e-06, "loss": 1.1141, "step": 1320 }, { "epoch": 0.02419105476636534, "grad_norm": 2.223905324935913, "learning_rate": 9.70898905831309e-06, "loss": 1.0639, "step": 1330 }, { "epoch": 0.024372942396187636, "grad_norm": 2.27856707572937, "learning_rate": 9.71909993207074e-06, "loss": 1.0891, "step": 1340 }, { "epoch": 0.02455483002600993, "grad_norm": 1.8044217824935913, "learning_rate": 9.729135631350626e-06, "loss": 1.0856, "step": 1350 }, { "epoch": 0.024736717655832227, "grad_norm": 1.888077735900879, "learning_rate": 9.739097265754349e-06, "loss": 1.083, "step": 1360 }, { "epoch": 0.024918605285654524, "grad_norm": 2.092176675796509, "learning_rate": 9.748985920496237e-06, "loss": 1.0807, "step": 1370 }, { "epoch": 0.025100492915476818, "grad_norm": 2.375063419342041, "learning_rate": 9.758802657112778e-06, "loss": 1.1016, "step": 1380 }, { "epoch": 0.025282380545299115, "grad_norm": 2.418849229812622, "learning_rate": 9.768548514146493e-06, "loss": 1.1417, "step": 1390 }, { "epoch": 0.02546426817512141, "grad_norm": 2.329854965209961, "learning_rate": 9.778224507805269e-06, "loss": 1.1169, "step": 1400 }, { "epoch": 0.025646155804943706, "grad_norm": 1.759024739265442, "learning_rate": 9.787831632598233e-06, "loss": 1.0783, "step": 1410 }, { "epoch": 0.025828043434766003, "grad_norm": 1.8483268022537231, "learning_rate": 9.797370861949155e-06, "loss": 1.1116, "step": 1420 }, { "epoch": 0.026009931064588297, "grad_norm": 2.314424514770508, "learning_rate": 9.806843148788337e-06, "loss": 1.0831, "step": 1430 }, { "epoch": 0.026191818694410594, "grad_norm": 2.067441940307617, "learning_rate": 9.816249426123863e-06, "loss": 1.0352, "step": 1440 }, { "epoch": 0.026373706324232887, "grad_norm": 2.0936992168426514, "learning_rate": 9.825590607593097e-06, "loss": 1.0742, "step": 1450 }, { "epoch": 0.026555593954055184, "grad_norm": 2.2518866062164307, "learning_rate": 9.834867587995221e-06, "loss": 1.0817, "step": 1460 }, { "epoch": 0.02673748158387748, "grad_norm": 2.0199484825134277, "learning_rate": 9.844081243805595e-06, "loss": 1.1271, "step": 1470 }, { "epoch": 0.026919369213699775, "grad_norm": 2.1645147800445557, "learning_rate": 9.853232433672685e-06, "loss": 1.059, "step": 1480 }, { "epoch": 0.027101256843522072, "grad_norm": 2.0219526290893555, "learning_rate": 9.862321998898258e-06, "loss": 1.0667, "step": 1490 }, { "epoch": 0.02728314447334437, "grad_norm": 2.223036050796509, "learning_rate": 9.871350763901519e-06, "loss": 1.0595, "step": 1500 }, { "epoch": 0.027465032103166663, "grad_norm": 2.440768003463745, "learning_rate": 9.880319536667822e-06, "loss": 1.0905, "step": 1510 }, { "epoch": 0.02764691973298896, "grad_norm": 2.1718924045562744, "learning_rate": 9.889229109182575e-06, "loss": 1.156, "step": 1520 }, { "epoch": 0.027828807362811254, "grad_norm": 1.8884831666946411, "learning_rate": 9.898080257850923e-06, "loss": 1.0563, "step": 1530 }, { "epoch": 0.02801069499263355, "grad_norm": 1.9383727312088013, "learning_rate": 9.906873743903753e-06, "loss": 1.1086, "step": 1540 }, { "epoch": 0.028192582622455848, "grad_norm": 2.0660831928253174, "learning_rate": 9.915610313790553e-06, "loss": 1.0101, "step": 1550 }, { "epoch": 0.02837447025227814, "grad_norm": 1.8989975452423096, "learning_rate": 9.924290699559666e-06, "loss": 1.0818, "step": 1560 }, { "epoch": 0.02855635788210044, "grad_norm": 2.1190500259399414, "learning_rate": 9.932915619226359e-06, "loss": 1.0777, "step": 1570 }, { "epoch": 0.028738245511922732, "grad_norm": 1.95400869846344, "learning_rate": 9.941485777129238e-06, "loss": 1.0513, "step": 1580 }, { "epoch": 0.02892013314174503, "grad_norm": 2.332594871520996, "learning_rate": 9.950001864275395e-06, "loss": 1.0353, "step": 1590 }, { "epoch": 0.029102020771567327, "grad_norm": 2.164888381958008, "learning_rate": 9.958464558674756e-06, "loss": 1.048, "step": 1600 }, { "epoch": 0.02928390840138962, "grad_norm": 1.941528558731079, "learning_rate": 9.966874525663997e-06, "loss": 1.0839, "step": 1610 }, { "epoch": 0.029465796031211917, "grad_norm": 1.946907639503479, "learning_rate": 9.975232418220436e-06, "loss": 1.0767, "step": 1620 }, { "epoch": 0.029647683661034215, "grad_norm": 2.4455347061157227, "learning_rate": 9.98353887726628e-06, "loss": 1.1282, "step": 1630 }, { "epoch": 0.029829571290856508, "grad_norm": 2.301375150680542, "learning_rate": 9.991794531963535e-06, "loss": 1.0592, "step": 1640 }, { "epoch": 0.030011458920678805, "grad_norm": 2.0533745288848877, "learning_rate": 1e-05, "loss": 1.0922, "step": 1650 }, { "epoch": 0.0301933465505011, "grad_norm": 2.078913450241089, "learning_rate": 9.998312362879484e-06, "loss": 1.0714, "step": 1660 }, { "epoch": 0.030375234180323396, "grad_norm": 2.4876811504364014, "learning_rate": 9.996437210523355e-06, "loss": 1.0715, "step": 1670 }, { "epoch": 0.030557121810145693, "grad_norm": 2.1432297229766846, "learning_rate": 9.994562058167227e-06, "loss": 1.0948, "step": 1680 }, { "epoch": 0.030739009439967987, "grad_norm": 2.4991774559020996, "learning_rate": 9.992686905811098e-06, "loss": 1.1053, "step": 1690 }, { "epoch": 0.030920897069790284, "grad_norm": 2.112098455429077, "learning_rate": 9.99081175345497e-06, "loss": 1.1215, "step": 1700 }, { "epoch": 0.03110278469961258, "grad_norm": 2.095259666442871, "learning_rate": 9.98893660109884e-06, "loss": 1.0678, "step": 1710 }, { "epoch": 0.031284672329434875, "grad_norm": 2.0996689796447754, "learning_rate": 9.98706144874271e-06, "loss": 1.1001, "step": 1720 }, { "epoch": 0.03146655995925717, "grad_norm": 1.864997386932373, "learning_rate": 9.985186296386582e-06, "loss": 1.0655, "step": 1730 }, { "epoch": 0.03164844758907947, "grad_norm": 2.0542449951171875, "learning_rate": 9.983311144030453e-06, "loss": 1.0926, "step": 1740 }, { "epoch": 0.03183033521890176, "grad_norm": 2.0685179233551025, "learning_rate": 9.981435991674324e-06, "loss": 1.1118, "step": 1750 }, { "epoch": 0.032012222848724056, "grad_norm": 1.861316204071045, "learning_rate": 9.979560839318196e-06, "loss": 1.0472, "step": 1760 }, { "epoch": 0.03219411047854635, "grad_norm": 2.070981979370117, "learning_rate": 9.977685686962067e-06, "loss": 1.0368, "step": 1770 }, { "epoch": 0.03237599810836865, "grad_norm": 2.3046417236328125, "learning_rate": 9.975810534605938e-06, "loss": 1.1441, "step": 1780 }, { "epoch": 0.03255788573819095, "grad_norm": 2.5218398571014404, "learning_rate": 9.973935382249808e-06, "loss": 1.0674, "step": 1790 }, { "epoch": 0.032739773368013245, "grad_norm": 2.218770742416382, "learning_rate": 9.97206022989368e-06, "loss": 1.1103, "step": 1800 }, { "epoch": 0.032921660997835535, "grad_norm": 2.19288969039917, "learning_rate": 9.97018507753755e-06, "loss": 1.0968, "step": 1810 }, { "epoch": 0.03310354862765783, "grad_norm": 2.0518405437469482, "learning_rate": 9.968309925181422e-06, "loss": 1.1132, "step": 1820 }, { "epoch": 0.03328543625748013, "grad_norm": 2.027134418487549, "learning_rate": 9.966434772825293e-06, "loss": 1.0865, "step": 1830 }, { "epoch": 0.033467323887302426, "grad_norm": 2.2364461421966553, "learning_rate": 9.964559620469165e-06, "loss": 1.076, "step": 1840 }, { "epoch": 0.03364921151712472, "grad_norm": 2.2125043869018555, "learning_rate": 9.962684468113036e-06, "loss": 1.0956, "step": 1850 }, { "epoch": 0.033831099146947013, "grad_norm": 2.2807345390319824, "learning_rate": 9.960809315756906e-06, "loss": 1.1085, "step": 1860 }, { "epoch": 0.03401298677676931, "grad_norm": 2.1539998054504395, "learning_rate": 9.958934163400777e-06, "loss": 1.0706, "step": 1870 }, { "epoch": 0.03419487440659161, "grad_norm": 1.906841516494751, "learning_rate": 9.957059011044648e-06, "loss": 1.0821, "step": 1880 }, { "epoch": 0.034376762036413905, "grad_norm": 2.2091033458709717, "learning_rate": 9.95518385868852e-06, "loss": 1.1329, "step": 1890 }, { "epoch": 0.0345586496662362, "grad_norm": 2.0988004207611084, "learning_rate": 9.953308706332391e-06, "loss": 1.0506, "step": 1900 }, { "epoch": 0.03474053729605849, "grad_norm": 2.256591558456421, "learning_rate": 9.951433553976262e-06, "loss": 1.1358, "step": 1910 }, { "epoch": 0.03492242492588079, "grad_norm": 2.1920981407165527, "learning_rate": 9.949558401620134e-06, "loss": 1.0602, "step": 1920 }, { "epoch": 0.035104312555703086, "grad_norm": 2.1666300296783447, "learning_rate": 9.947683249264003e-06, "loss": 1.0628, "step": 1930 }, { "epoch": 0.03528620018552538, "grad_norm": 2.159074306488037, "learning_rate": 9.945808096907875e-06, "loss": 1.1096, "step": 1940 }, { "epoch": 0.03546808781534768, "grad_norm": 2.303570508956909, "learning_rate": 9.943932944551744e-06, "loss": 1.0579, "step": 1950 }, { "epoch": 0.03564997544516997, "grad_norm": 2.1957900524139404, "learning_rate": 9.942057792195616e-06, "loss": 1.0694, "step": 1960 }, { "epoch": 0.03583186307499227, "grad_norm": 1.8738420009613037, "learning_rate": 9.940182639839487e-06, "loss": 1.1123, "step": 1970 }, { "epoch": 0.036013750704814565, "grad_norm": 1.8341156244277954, "learning_rate": 9.938307487483358e-06, "loss": 1.0623, "step": 1980 }, { "epoch": 0.03619563833463686, "grad_norm": 2.119471788406372, "learning_rate": 9.93643233512723e-06, "loss": 1.0728, "step": 1990 }, { "epoch": 0.03637752596445916, "grad_norm": 2.261554002761841, "learning_rate": 9.934557182771101e-06, "loss": 1.0776, "step": 2000 }, { "epoch": 0.036559413594281456, "grad_norm": 2.4454526901245117, "learning_rate": 9.932682030414972e-06, "loss": 1.0907, "step": 2010 }, { "epoch": 0.036741301224103746, "grad_norm": 2.012326717376709, "learning_rate": 9.930806878058842e-06, "loss": 1.09, "step": 2020 }, { "epoch": 0.036923188853926044, "grad_norm": 1.8640422821044922, "learning_rate": 9.928931725702713e-06, "loss": 1.0694, "step": 2030 }, { "epoch": 0.03710507648374834, "grad_norm": 2.2236013412475586, "learning_rate": 9.927056573346585e-06, "loss": 1.0468, "step": 2040 }, { "epoch": 0.03728696411357064, "grad_norm": 2.0866546630859375, "learning_rate": 9.925181420990456e-06, "loss": 1.0321, "step": 2050 }, { "epoch": 0.037468851743392935, "grad_norm": 1.930566668510437, "learning_rate": 9.923306268634328e-06, "loss": 1.021, "step": 2060 }, { "epoch": 0.037650739373215225, "grad_norm": 2.0315401554107666, "learning_rate": 9.921431116278199e-06, "loss": 1.1325, "step": 2070 }, { "epoch": 0.03783262700303752, "grad_norm": 1.808922290802002, "learning_rate": 9.91955596392207e-06, "loss": 1.0451, "step": 2080 }, { "epoch": 0.03801451463285982, "grad_norm": 2.1626944541931152, "learning_rate": 9.91768081156594e-06, "loss": 1.1076, "step": 2090 }, { "epoch": 0.038196402262682116, "grad_norm": 2.702723979949951, "learning_rate": 9.915805659209811e-06, "loss": 1.1283, "step": 2100 }, { "epoch": 0.038378289892504414, "grad_norm": 2.102219581604004, "learning_rate": 9.913930506853683e-06, "loss": 1.0416, "step": 2110 }, { "epoch": 0.038560177522326704, "grad_norm": 2.005981922149658, "learning_rate": 9.912055354497554e-06, "loss": 1.0958, "step": 2120 }, { "epoch": 0.038742065152149, "grad_norm": 2.101555109024048, "learning_rate": 9.910180202141425e-06, "loss": 1.1123, "step": 2130 }, { "epoch": 0.0389239527819713, "grad_norm": 2.2241809368133545, "learning_rate": 9.908305049785297e-06, "loss": 1.0976, "step": 2140 }, { "epoch": 0.039105840411793595, "grad_norm": 2.0759167671203613, "learning_rate": 9.906429897429168e-06, "loss": 1.0907, "step": 2150 }, { "epoch": 0.03928772804161589, "grad_norm": 1.9940487146377563, "learning_rate": 9.904554745073038e-06, "loss": 1.0718, "step": 2160 }, { "epoch": 0.03946961567143818, "grad_norm": 2.1747734546661377, "learning_rate": 9.902679592716909e-06, "loss": 1.0737, "step": 2170 }, { "epoch": 0.03965150330126048, "grad_norm": 1.978407621383667, "learning_rate": 9.90080444036078e-06, "loss": 1.0666, "step": 2180 }, { "epoch": 0.03983339093108278, "grad_norm": 2.268183469772339, "learning_rate": 9.898929288004652e-06, "loss": 1.0369, "step": 2190 }, { "epoch": 0.040015278560905074, "grad_norm": 2.271346092224121, "learning_rate": 9.897054135648523e-06, "loss": 1.0403, "step": 2200 }, { "epoch": 0.04019716619072737, "grad_norm": 1.8638702630996704, "learning_rate": 9.895178983292393e-06, "loss": 1.0375, "step": 2210 }, { "epoch": 0.04037905382054966, "grad_norm": 2.0516111850738525, "learning_rate": 9.893303830936264e-06, "loss": 1.1368, "step": 2220 }, { "epoch": 0.04056094145037196, "grad_norm": 1.8387516736984253, "learning_rate": 9.891428678580135e-06, "loss": 1.0197, "step": 2230 }, { "epoch": 0.040742829080194255, "grad_norm": 2.0782952308654785, "learning_rate": 9.889553526224007e-06, "loss": 1.0209, "step": 2240 }, { "epoch": 0.04092471671001655, "grad_norm": 2.322962522506714, "learning_rate": 9.887678373867878e-06, "loss": 1.094, "step": 2250 }, { "epoch": 0.04110660433983885, "grad_norm": 2.2086918354034424, "learning_rate": 9.885803221511748e-06, "loss": 1.0873, "step": 2260 }, { "epoch": 0.04128849196966115, "grad_norm": 2.08990216255188, "learning_rate": 9.883928069155619e-06, "loss": 1.059, "step": 2270 }, { "epoch": 0.04147037959948344, "grad_norm": 2.1425018310546875, "learning_rate": 9.88205291679949e-06, "loss": 1.0827, "step": 2280 }, { "epoch": 0.041652267229305734, "grad_norm": 2.236010789871216, "learning_rate": 9.880177764443362e-06, "loss": 1.0617, "step": 2290 }, { "epoch": 0.04183415485912803, "grad_norm": 2.0994324684143066, "learning_rate": 9.878302612087233e-06, "loss": 1.0604, "step": 2300 }, { "epoch": 0.04201604248895033, "grad_norm": 2.211843729019165, "learning_rate": 9.876427459731104e-06, "loss": 1.0471, "step": 2310 }, { "epoch": 0.042197930118772625, "grad_norm": 2.2948849201202393, "learning_rate": 9.874552307374976e-06, "loss": 1.1088, "step": 2320 }, { "epoch": 0.042379817748594915, "grad_norm": 2.068437337875366, "learning_rate": 9.872677155018845e-06, "loss": 1.034, "step": 2330 }, { "epoch": 0.04256170537841721, "grad_norm": 2.4303810596466064, "learning_rate": 9.870802002662717e-06, "loss": 1.0891, "step": 2340 }, { "epoch": 0.04274359300823951, "grad_norm": 2.3631670475006104, "learning_rate": 9.868926850306588e-06, "loss": 0.9964, "step": 2350 }, { "epoch": 0.04292548063806181, "grad_norm": 2.0966973304748535, "learning_rate": 9.86705169795046e-06, "loss": 1.0992, "step": 2360 }, { "epoch": 0.043107368267884104, "grad_norm": 1.9453641176223755, "learning_rate": 9.86517654559433e-06, "loss": 1.0674, "step": 2370 }, { "epoch": 0.043289255897706394, "grad_norm": 1.9895405769348145, "learning_rate": 9.863301393238202e-06, "loss": 1.0961, "step": 2380 }, { "epoch": 0.04347114352752869, "grad_norm": 2.2851083278656006, "learning_rate": 9.861426240882073e-06, "loss": 1.0348, "step": 2390 }, { "epoch": 0.04365303115735099, "grad_norm": 1.8202131986618042, "learning_rate": 9.859551088525943e-06, "loss": 1.0666, "step": 2400 }, { "epoch": 0.043834918787173285, "grad_norm": 2.2696967124938965, "learning_rate": 9.857675936169814e-06, "loss": 1.0427, "step": 2410 }, { "epoch": 0.04401680641699558, "grad_norm": 1.8943482637405396, "learning_rate": 9.855800783813686e-06, "loss": 1.0625, "step": 2420 }, { "epoch": 0.04419869404681787, "grad_norm": 2.0973594188690186, "learning_rate": 9.853925631457557e-06, "loss": 1.0684, "step": 2430 }, { "epoch": 0.04438058167664017, "grad_norm": 2.0698273181915283, "learning_rate": 9.852050479101429e-06, "loss": 1.0635, "step": 2440 }, { "epoch": 0.04456246930646247, "grad_norm": 1.8531159162521362, "learning_rate": 9.8501753267453e-06, "loss": 1.0336, "step": 2450 }, { "epoch": 0.044744356936284764, "grad_norm": 2.1260156631469727, "learning_rate": 9.848300174389171e-06, "loss": 1.0623, "step": 2460 }, { "epoch": 0.04492624456610706, "grad_norm": 2.116804838180542, "learning_rate": 9.846425022033041e-06, "loss": 1.0796, "step": 2470 }, { "epoch": 0.04510813219592936, "grad_norm": 2.451788902282715, "learning_rate": 9.844549869676912e-06, "loss": 1.1251, "step": 2480 }, { "epoch": 0.04529001982575165, "grad_norm": 2.1061205863952637, "learning_rate": 9.842674717320782e-06, "loss": 1.0933, "step": 2490 }, { "epoch": 0.045471907455573946, "grad_norm": 2.273237705230713, "learning_rate": 9.840799564964653e-06, "loss": 1.0548, "step": 2500 }, { "epoch": 0.04565379508539624, "grad_norm": 2.126223087310791, "learning_rate": 9.838924412608525e-06, "loss": 1.0339, "step": 2510 }, { "epoch": 0.04583568271521854, "grad_norm": 1.8449867963790894, "learning_rate": 9.837049260252396e-06, "loss": 1.0736, "step": 2520 }, { "epoch": 0.04601757034504084, "grad_norm": 2.1887238025665283, "learning_rate": 9.835174107896267e-06, "loss": 1.0222, "step": 2530 }, { "epoch": 0.04619945797486313, "grad_norm": 2.2490627765655518, "learning_rate": 9.833298955540139e-06, "loss": 1.071, "step": 2540 }, { "epoch": 0.046381345604685424, "grad_norm": 2.093247413635254, "learning_rate": 9.83142380318401e-06, "loss": 1.0648, "step": 2550 }, { "epoch": 0.04656323323450772, "grad_norm": 1.952309489250183, "learning_rate": 9.82954865082788e-06, "loss": 1.0413, "step": 2560 }, { "epoch": 0.04674512086433002, "grad_norm": 1.791938066482544, "learning_rate": 9.827673498471751e-06, "loss": 1.0994, "step": 2570 }, { "epoch": 0.046927008494152316, "grad_norm": 2.0889620780944824, "learning_rate": 9.825798346115622e-06, "loss": 1.0682, "step": 2580 }, { "epoch": 0.047108896123974606, "grad_norm": 1.7925217151641846, "learning_rate": 9.823923193759494e-06, "loss": 1.0172, "step": 2590 }, { "epoch": 0.0472907837537969, "grad_norm": 1.8989485502243042, "learning_rate": 9.822048041403365e-06, "loss": 1.0729, "step": 2600 }, { "epoch": 0.0474726713836192, "grad_norm": 1.85089111328125, "learning_rate": 9.820172889047236e-06, "loss": 1.0487, "step": 2610 }, { "epoch": 0.0476545590134415, "grad_norm": 2.102358102798462, "learning_rate": 9.818297736691108e-06, "loss": 1.0481, "step": 2620 }, { "epoch": 0.047836446643263794, "grad_norm": 1.9082953929901123, "learning_rate": 9.816422584334977e-06, "loss": 1.0918, "step": 2630 }, { "epoch": 0.048018334273086084, "grad_norm": 2.3200759887695312, "learning_rate": 9.814547431978849e-06, "loss": 1.0302, "step": 2640 }, { "epoch": 0.04820022190290838, "grad_norm": 1.9435070753097534, "learning_rate": 9.81267227962272e-06, "loss": 1.0849, "step": 2650 }, { "epoch": 0.04838210953273068, "grad_norm": 2.0293588638305664, "learning_rate": 9.810797127266591e-06, "loss": 1.1034, "step": 2660 }, { "epoch": 0.048563997162552976, "grad_norm": 1.9240785837173462, "learning_rate": 9.808921974910463e-06, "loss": 1.0387, "step": 2670 }, { "epoch": 0.04874588479237527, "grad_norm": 1.8327438831329346, "learning_rate": 9.807046822554334e-06, "loss": 1.0738, "step": 2680 }, { "epoch": 0.04892777242219756, "grad_norm": 2.0993192195892334, "learning_rate": 9.805171670198205e-06, "loss": 1.0941, "step": 2690 }, { "epoch": 0.04910966005201986, "grad_norm": 1.7831382751464844, "learning_rate": 9.803296517842075e-06, "loss": 1.0471, "step": 2700 }, { "epoch": 0.04929154768184216, "grad_norm": 2.063632011413574, "learning_rate": 9.801421365485946e-06, "loss": 1.0436, "step": 2710 }, { "epoch": 0.049473435311664454, "grad_norm": 1.7250831127166748, "learning_rate": 9.799546213129818e-06, "loss": 1.0632, "step": 2720 }, { "epoch": 0.04965532294148675, "grad_norm": 2.0084586143493652, "learning_rate": 9.797671060773689e-06, "loss": 1.0458, "step": 2730 }, { "epoch": 0.04983721057130905, "grad_norm": 2.218557357788086, "learning_rate": 9.79579590841756e-06, "loss": 1.0464, "step": 2740 }, { "epoch": 0.05001909820113134, "grad_norm": 2.0410192012786865, "learning_rate": 9.793920756061432e-06, "loss": 1.0952, "step": 2750 }, { "epoch": 0.050200985830953636, "grad_norm": 2.141200065612793, "learning_rate": 9.792045603705301e-06, "loss": 1.0414, "step": 2760 }, { "epoch": 0.05038287346077593, "grad_norm": 1.9114173650741577, "learning_rate": 9.790170451349173e-06, "loss": 1.0833, "step": 2770 }, { "epoch": 0.05056476109059823, "grad_norm": 1.8680387735366821, "learning_rate": 9.788295298993044e-06, "loss": 1.0807, "step": 2780 }, { "epoch": 0.05074664872042053, "grad_norm": 2.077197313308716, "learning_rate": 9.786420146636915e-06, "loss": 1.0633, "step": 2790 }, { "epoch": 0.05092853635024282, "grad_norm": 2.0504746437072754, "learning_rate": 9.784544994280785e-06, "loss": 1.0808, "step": 2800 }, { "epoch": 0.051110423980065114, "grad_norm": 1.9054299592971802, "learning_rate": 9.782669841924656e-06, "loss": 1.0973, "step": 2810 }, { "epoch": 0.05129231160988741, "grad_norm": 2.4688613414764404, "learning_rate": 9.780794689568528e-06, "loss": 1.0901, "step": 2820 }, { "epoch": 0.05147419923970971, "grad_norm": 1.8642398118972778, "learning_rate": 9.778919537212399e-06, "loss": 1.0878, "step": 2830 }, { "epoch": 0.051656086869532006, "grad_norm": 2.250351667404175, "learning_rate": 9.77704438485627e-06, "loss": 1.0934, "step": 2840 }, { "epoch": 0.051837974499354296, "grad_norm": 2.0136661529541016, "learning_rate": 9.775169232500142e-06, "loss": 1.0596, "step": 2850 }, { "epoch": 0.05201986212917659, "grad_norm": 1.8060574531555176, "learning_rate": 9.773294080144013e-06, "loss": 0.9879, "step": 2860 }, { "epoch": 0.05220174975899889, "grad_norm": 1.9669241905212402, "learning_rate": 9.771418927787883e-06, "loss": 1.0545, "step": 2870 }, { "epoch": 0.05238363738882119, "grad_norm": 1.992021918296814, "learning_rate": 9.769543775431754e-06, "loss": 1.0571, "step": 2880 }, { "epoch": 0.052565525018643484, "grad_norm": 2.03399658203125, "learning_rate": 9.767668623075626e-06, "loss": 1.0298, "step": 2890 }, { "epoch": 0.052747412648465775, "grad_norm": 2.177325963973999, "learning_rate": 9.765793470719497e-06, "loss": 1.0684, "step": 2900 }, { "epoch": 0.05292930027828807, "grad_norm": 2.0124611854553223, "learning_rate": 9.763918318363368e-06, "loss": 1.0659, "step": 2910 }, { "epoch": 0.05311118790811037, "grad_norm": 1.6754887104034424, "learning_rate": 9.76204316600724e-06, "loss": 0.9972, "step": 2920 }, { "epoch": 0.053293075537932666, "grad_norm": 2.0895204544067383, "learning_rate": 9.760168013651111e-06, "loss": 1.0678, "step": 2930 }, { "epoch": 0.05347496316775496, "grad_norm": 1.8113572597503662, "learning_rate": 9.75829286129498e-06, "loss": 1.0403, "step": 2940 }, { "epoch": 0.05365685079757726, "grad_norm": 1.8781177997589111, "learning_rate": 9.756417708938852e-06, "loss": 1.0844, "step": 2950 }, { "epoch": 0.05383873842739955, "grad_norm": 1.8169450759887695, "learning_rate": 9.754542556582723e-06, "loss": 1.0245, "step": 2960 }, { "epoch": 0.05402062605722185, "grad_norm": 2.1107654571533203, "learning_rate": 9.752667404226595e-06, "loss": 1.0685, "step": 2970 }, { "epoch": 0.054202513687044145, "grad_norm": 1.9126217365264893, "learning_rate": 9.750792251870466e-06, "loss": 1.0481, "step": 2980 }, { "epoch": 0.05438440131686644, "grad_norm": 2.0116934776306152, "learning_rate": 9.748917099514337e-06, "loss": 1.065, "step": 2990 }, { "epoch": 0.05456628894668874, "grad_norm": 1.9778796434402466, "learning_rate": 9.747041947158209e-06, "loss": 1.0879, "step": 3000 }, { "epoch": 0.05474817657651103, "grad_norm": 2.0754051208496094, "learning_rate": 9.745166794802078e-06, "loss": 1.0535, "step": 3010 }, { "epoch": 0.054930064206333326, "grad_norm": 1.944700837135315, "learning_rate": 9.74329164244595e-06, "loss": 1.0443, "step": 3020 }, { "epoch": 0.05511195183615562, "grad_norm": 2.2578976154327393, "learning_rate": 9.74141649008982e-06, "loss": 1.0476, "step": 3030 }, { "epoch": 0.05529383946597792, "grad_norm": 2.187514066696167, "learning_rate": 9.73954133773369e-06, "loss": 1.0217, "step": 3040 }, { "epoch": 0.05547572709580022, "grad_norm": 2.170315742492676, "learning_rate": 9.737666185377562e-06, "loss": 1.0328, "step": 3050 }, { "epoch": 0.05565761472562251, "grad_norm": 1.9472947120666504, "learning_rate": 9.735791033021433e-06, "loss": 0.9958, "step": 3060 }, { "epoch": 0.055839502355444805, "grad_norm": 2.156726121902466, "learning_rate": 9.733915880665305e-06, "loss": 1.0736, "step": 3070 }, { "epoch": 0.0560213899852671, "grad_norm": 1.6079156398773193, "learning_rate": 9.732040728309176e-06, "loss": 1.0274, "step": 3080 }, { "epoch": 0.0562032776150894, "grad_norm": 1.728289246559143, "learning_rate": 9.730165575953047e-06, "loss": 1.0879, "step": 3090 }, { "epoch": 0.056385165244911696, "grad_norm": 1.872959017753601, "learning_rate": 9.728290423596917e-06, "loss": 1.041, "step": 3100 }, { "epoch": 0.056567052874733986, "grad_norm": 2.1834208965301514, "learning_rate": 9.726415271240788e-06, "loss": 1.0053, "step": 3110 }, { "epoch": 0.05674894050455628, "grad_norm": 1.9437049627304077, "learning_rate": 9.72454011888466e-06, "loss": 1.073, "step": 3120 }, { "epoch": 0.05693082813437858, "grad_norm": 2.2302844524383545, "learning_rate": 9.722664966528531e-06, "loss": 1.0919, "step": 3130 }, { "epoch": 0.05711271576420088, "grad_norm": 2.0193328857421875, "learning_rate": 9.720789814172402e-06, "loss": 1.0451, "step": 3140 }, { "epoch": 0.057294603394023175, "grad_norm": 1.9554122686386108, "learning_rate": 9.718914661816274e-06, "loss": 1.0729, "step": 3150 }, { "epoch": 0.057476491023845465, "grad_norm": 2.064631462097168, "learning_rate": 9.717039509460145e-06, "loss": 1.0698, "step": 3160 }, { "epoch": 0.05765837865366776, "grad_norm": 1.9653854370117188, "learning_rate": 9.715164357104015e-06, "loss": 1.0644, "step": 3170 }, { "epoch": 0.05784026628349006, "grad_norm": 1.8224579095840454, "learning_rate": 9.713289204747886e-06, "loss": 1.0707, "step": 3180 }, { "epoch": 0.058022153913312356, "grad_norm": 2.089186429977417, "learning_rate": 9.711414052391757e-06, "loss": 0.9834, "step": 3190 }, { "epoch": 0.05820404154313465, "grad_norm": 2.1762919425964355, "learning_rate": 9.709538900035629e-06, "loss": 1.07, "step": 3200 }, { "epoch": 0.05838592917295695, "grad_norm": 2.1300547122955322, "learning_rate": 9.7076637476795e-06, "loss": 1.1359, "step": 3210 }, { "epoch": 0.05856781680277924, "grad_norm": 2.81547474861145, "learning_rate": 9.705788595323371e-06, "loss": 1.0573, "step": 3220 }, { "epoch": 0.05874970443260154, "grad_norm": 2.074758768081665, "learning_rate": 9.703913442967243e-06, "loss": 1.0838, "step": 3230 }, { "epoch": 0.058931592062423835, "grad_norm": 2.108267307281494, "learning_rate": 9.702038290611112e-06, "loss": 1.0146, "step": 3240 }, { "epoch": 0.05911347969224613, "grad_norm": 1.914854884147644, "learning_rate": 9.700163138254984e-06, "loss": 1.0881, "step": 3250 }, { "epoch": 0.05929536732206843, "grad_norm": 1.8367441892623901, "learning_rate": 9.698287985898855e-06, "loss": 1.0528, "step": 3260 }, { "epoch": 0.05947725495189072, "grad_norm": 2.411186933517456, "learning_rate": 9.696412833542727e-06, "loss": 1.0627, "step": 3270 }, { "epoch": 0.059659142581713016, "grad_norm": 1.851426362991333, "learning_rate": 9.694537681186598e-06, "loss": 1.0374, "step": 3280 }, { "epoch": 0.059841030211535313, "grad_norm": 2.051708221435547, "learning_rate": 9.69266252883047e-06, "loss": 1.0518, "step": 3290 }, { "epoch": 0.06002291784135761, "grad_norm": 1.6065384149551392, "learning_rate": 9.69078737647434e-06, "loss": 1.0317, "step": 3300 }, { "epoch": 0.06020480547117991, "grad_norm": 1.9180585145950317, "learning_rate": 9.68891222411821e-06, "loss": 1.0545, "step": 3310 }, { "epoch": 0.0603866931010022, "grad_norm": 2.064558744430542, "learning_rate": 9.687037071762082e-06, "loss": 1.0981, "step": 3320 }, { "epoch": 0.060568580730824495, "grad_norm": 2.3542895317077637, "learning_rate": 9.685161919405953e-06, "loss": 1.0284, "step": 3330 }, { "epoch": 0.06075046836064679, "grad_norm": 1.910372257232666, "learning_rate": 9.683286767049823e-06, "loss": 1.0738, "step": 3340 }, { "epoch": 0.06093235599046909, "grad_norm": 2.22318434715271, "learning_rate": 9.681411614693694e-06, "loss": 1.0419, "step": 3350 }, { "epoch": 0.061114243620291386, "grad_norm": 2.02980375289917, "learning_rate": 9.679536462337565e-06, "loss": 1.0319, "step": 3360 }, { "epoch": 0.061296131250113677, "grad_norm": 1.83599853515625, "learning_rate": 9.677661309981437e-06, "loss": 1.0873, "step": 3370 }, { "epoch": 0.061478018879935974, "grad_norm": 1.9949945211410522, "learning_rate": 9.675786157625308e-06, "loss": 1.0588, "step": 3380 }, { "epoch": 0.06165990650975827, "grad_norm": 2.0783708095550537, "learning_rate": 9.67391100526918e-06, "loss": 1.0598, "step": 3390 }, { "epoch": 0.06184179413958057, "grad_norm": 1.9110045433044434, "learning_rate": 9.67203585291305e-06, "loss": 1.0447, "step": 3400 }, { "epoch": 0.062023681769402865, "grad_norm": 1.5796464681625366, "learning_rate": 9.67016070055692e-06, "loss": 0.9729, "step": 3410 }, { "epoch": 0.06220556939922516, "grad_norm": 2.0420589447021484, "learning_rate": 9.668285548200792e-06, "loss": 1.0633, "step": 3420 }, { "epoch": 0.06238745702904745, "grad_norm": 1.870244026184082, "learning_rate": 9.666410395844663e-06, "loss": 1.0075, "step": 3430 }, { "epoch": 0.06256934465886975, "grad_norm": 1.68318510055542, "learning_rate": 9.664535243488534e-06, "loss": 1.0464, "step": 3440 }, { "epoch": 0.06275123228869205, "grad_norm": 1.8888028860092163, "learning_rate": 9.662660091132406e-06, "loss": 1.0524, "step": 3450 }, { "epoch": 0.06293311991851434, "grad_norm": 2.0802059173583984, "learning_rate": 9.660784938776277e-06, "loss": 1.0667, "step": 3460 }, { "epoch": 0.06311500754833664, "grad_norm": 2.195378065109253, "learning_rate": 9.658909786420148e-06, "loss": 1.0847, "step": 3470 }, { "epoch": 0.06329689517815894, "grad_norm": 2.083913803100586, "learning_rate": 9.657034634064018e-06, "loss": 1.0647, "step": 3480 }, { "epoch": 0.06347878280798123, "grad_norm": 1.9541796445846558, "learning_rate": 9.65515948170789e-06, "loss": 1.0105, "step": 3490 }, { "epoch": 0.06366067043780352, "grad_norm": 1.9778636693954468, "learning_rate": 9.65328432935176e-06, "loss": 1.0297, "step": 3500 }, { "epoch": 0.06384255806762582, "grad_norm": 1.7345699071884155, "learning_rate": 9.651409176995632e-06, "loss": 1.0563, "step": 3510 }, { "epoch": 0.06402444569744811, "grad_norm": 1.9002227783203125, "learning_rate": 9.649534024639503e-06, "loss": 1.08, "step": 3520 }, { "epoch": 0.06420633332727041, "grad_norm": 1.6361489295959473, "learning_rate": 9.647658872283375e-06, "loss": 1.0636, "step": 3530 }, { "epoch": 0.0643882209570927, "grad_norm": 1.794331669807434, "learning_rate": 9.645783719927246e-06, "loss": 1.0257, "step": 3540 }, { "epoch": 0.064570108586915, "grad_norm": 1.98244309425354, "learning_rate": 9.643908567571116e-06, "loss": 0.996, "step": 3550 }, { "epoch": 0.0647519962167373, "grad_norm": 2.2512729167938232, "learning_rate": 9.642033415214987e-06, "loss": 1.0442, "step": 3560 }, { "epoch": 0.0649338838465596, "grad_norm": 1.970611572265625, "learning_rate": 9.640158262858858e-06, "loss": 1.0451, "step": 3570 }, { "epoch": 0.0651157714763819, "grad_norm": 1.9105221033096313, "learning_rate": 9.63828311050273e-06, "loss": 1.0365, "step": 3580 }, { "epoch": 0.06529765910620419, "grad_norm": 2.2752978801727295, "learning_rate": 9.6364079581466e-06, "loss": 1.0527, "step": 3590 }, { "epoch": 0.06547954673602649, "grad_norm": 1.9438990354537964, "learning_rate": 9.63453280579047e-06, "loss": 1.036, "step": 3600 }, { "epoch": 0.06566143436584877, "grad_norm": 1.8714165687561035, "learning_rate": 9.632657653434342e-06, "loss": 1.0342, "step": 3610 }, { "epoch": 0.06584332199567107, "grad_norm": 2.1653854846954346, "learning_rate": 9.630782501078213e-06, "loss": 1.025, "step": 3620 }, { "epoch": 0.06602520962549337, "grad_norm": 2.180068016052246, "learning_rate": 9.628907348722085e-06, "loss": 1.0717, "step": 3630 }, { "epoch": 0.06620709725531566, "grad_norm": 1.9516308307647705, "learning_rate": 9.627032196365954e-06, "loss": 1.051, "step": 3640 }, { "epoch": 0.06638898488513796, "grad_norm": 2.0084168910980225, "learning_rate": 9.625157044009826e-06, "loss": 1.0359, "step": 3650 }, { "epoch": 0.06657087251496026, "grad_norm": 2.114259719848633, "learning_rate": 9.623281891653697e-06, "loss": 1.1023, "step": 3660 }, { "epoch": 0.06675276014478256, "grad_norm": 2.0614378452301025, "learning_rate": 9.621406739297568e-06, "loss": 1.0323, "step": 3670 }, { "epoch": 0.06693464777460485, "grad_norm": 3.006847620010376, "learning_rate": 9.61953158694144e-06, "loss": 1.0493, "step": 3680 }, { "epoch": 0.06711653540442715, "grad_norm": 1.9178725481033325, "learning_rate": 9.617656434585311e-06, "loss": 1.0462, "step": 3690 }, { "epoch": 0.06729842303424945, "grad_norm": 1.9819490909576416, "learning_rate": 9.615781282229183e-06, "loss": 1.0408, "step": 3700 }, { "epoch": 0.06748031066407173, "grad_norm": 2.2754716873168945, "learning_rate": 9.613906129873052e-06, "loss": 1.0372, "step": 3710 }, { "epoch": 0.06766219829389403, "grad_norm": 2.0944437980651855, "learning_rate": 9.612030977516924e-06, "loss": 1.0644, "step": 3720 }, { "epoch": 0.06784408592371632, "grad_norm": 2.650284767150879, "learning_rate": 9.610155825160795e-06, "loss": 1.05, "step": 3730 }, { "epoch": 0.06802597355353862, "grad_norm": 3.915095329284668, "learning_rate": 9.608280672804666e-06, "loss": 1.0725, "step": 3740 }, { "epoch": 0.06820786118336092, "grad_norm": 2.4063615798950195, "learning_rate": 9.606405520448538e-06, "loss": 1.0837, "step": 3750 }, { "epoch": 0.06838974881318322, "grad_norm": 1.962277889251709, "learning_rate": 9.604530368092409e-06, "loss": 1.108, "step": 3760 }, { "epoch": 0.06857163644300551, "grad_norm": 1.8112766742706299, "learning_rate": 9.60265521573628e-06, "loss": 1.0446, "step": 3770 }, { "epoch": 0.06875352407282781, "grad_norm": 2.072768449783325, "learning_rate": 9.60078006338015e-06, "loss": 1.0269, "step": 3780 }, { "epoch": 0.0689354117026501, "grad_norm": 2.203702926635742, "learning_rate": 9.598904911024021e-06, "loss": 1.059, "step": 3790 }, { "epoch": 0.0691172993324724, "grad_norm": 2.0095455646514893, "learning_rate": 9.597029758667893e-06, "loss": 1.0299, "step": 3800 }, { "epoch": 0.0692991869622947, "grad_norm": 2.243889808654785, "learning_rate": 9.595154606311764e-06, "loss": 1.0727, "step": 3810 }, { "epoch": 0.06948107459211698, "grad_norm": 1.7506276369094849, "learning_rate": 9.593279453955635e-06, "loss": 1.0628, "step": 3820 }, { "epoch": 0.06966296222193928, "grad_norm": 1.9709569215774536, "learning_rate": 9.591404301599507e-06, "loss": 1.0415, "step": 3830 }, { "epoch": 0.06984484985176158, "grad_norm": 1.9766770601272583, "learning_rate": 9.589529149243378e-06, "loss": 1.0633, "step": 3840 }, { "epoch": 0.07002673748158388, "grad_norm": 1.777722716331482, "learning_rate": 9.587653996887248e-06, "loss": 1.0134, "step": 3850 }, { "epoch": 0.07020862511140617, "grad_norm": 2.0629265308380127, "learning_rate": 9.585778844531119e-06, "loss": 1.0743, "step": 3860 }, { "epoch": 0.07039051274122847, "grad_norm": 2.193633556365967, "learning_rate": 9.58390369217499e-06, "loss": 0.9857, "step": 3870 }, { "epoch": 0.07057240037105077, "grad_norm": 1.9299228191375732, "learning_rate": 9.58202853981886e-06, "loss": 1.0823, "step": 3880 }, { "epoch": 0.07075428800087306, "grad_norm": 2.1046764850616455, "learning_rate": 9.580153387462731e-06, "loss": 1.0945, "step": 3890 }, { "epoch": 0.07093617563069536, "grad_norm": 2.1045219898223877, "learning_rate": 9.578278235106603e-06, "loss": 1.0542, "step": 3900 }, { "epoch": 0.07111806326051766, "grad_norm": 1.8538627624511719, "learning_rate": 9.576403082750474e-06, "loss": 1.0214, "step": 3910 }, { "epoch": 0.07129995089033994, "grad_norm": 1.9710133075714111, "learning_rate": 9.574527930394345e-06, "loss": 0.9567, "step": 3920 }, { "epoch": 0.07148183852016224, "grad_norm": 1.9644601345062256, "learning_rate": 9.572652778038217e-06, "loss": 0.975, "step": 3930 }, { "epoch": 0.07166372614998454, "grad_norm": 2.3125038146972656, "learning_rate": 9.570777625682088e-06, "loss": 1.0786, "step": 3940 }, { "epoch": 0.07184561377980683, "grad_norm": 1.726649522781372, "learning_rate": 9.568902473325958e-06, "loss": 1.0694, "step": 3950 }, { "epoch": 0.07202750140962913, "grad_norm": 1.9652243852615356, "learning_rate": 9.567027320969829e-06, "loss": 1.0287, "step": 3960 }, { "epoch": 0.07220938903945143, "grad_norm": 2.018324613571167, "learning_rate": 9.5651521686137e-06, "loss": 1.0653, "step": 3970 }, { "epoch": 0.07239127666927372, "grad_norm": 1.8757928609848022, "learning_rate": 9.563277016257572e-06, "loss": 1.0983, "step": 3980 }, { "epoch": 0.07257316429909602, "grad_norm": 2.0476598739624023, "learning_rate": 9.561401863901443e-06, "loss": 1.0542, "step": 3990 }, { "epoch": 0.07275505192891832, "grad_norm": 2.1414577960968018, "learning_rate": 9.559526711545314e-06, "loss": 1.0645, "step": 4000 }, { "epoch": 0.07293693955874062, "grad_norm": 2.1879947185516357, "learning_rate": 9.557651559189186e-06, "loss": 1.0398, "step": 4010 }, { "epoch": 0.07311882718856291, "grad_norm": 2.161365270614624, "learning_rate": 9.555776406833055e-06, "loss": 1.0976, "step": 4020 }, { "epoch": 0.0733007148183852, "grad_norm": 2.020900011062622, "learning_rate": 9.553901254476927e-06, "loss": 1.0584, "step": 4030 }, { "epoch": 0.07348260244820749, "grad_norm": 2.0236494541168213, "learning_rate": 9.552026102120798e-06, "loss": 0.9835, "step": 4040 }, { "epoch": 0.07366449007802979, "grad_norm": 2.110966205596924, "learning_rate": 9.55015094976467e-06, "loss": 1.032, "step": 4050 }, { "epoch": 0.07384637770785209, "grad_norm": 2.1377997398376465, "learning_rate": 9.54827579740854e-06, "loss": 1.0905, "step": 4060 }, { "epoch": 0.07402826533767438, "grad_norm": 2.1384310722351074, "learning_rate": 9.546400645052412e-06, "loss": 1.0453, "step": 4070 }, { "epoch": 0.07421015296749668, "grad_norm": 2.242419958114624, "learning_rate": 9.544525492696284e-06, "loss": 1.0803, "step": 4080 }, { "epoch": 0.07439204059731898, "grad_norm": 2.0814545154571533, "learning_rate": 9.542650340340153e-06, "loss": 1.0593, "step": 4090 }, { "epoch": 0.07457392822714128, "grad_norm": 2.0891764163970947, "learning_rate": 9.540775187984025e-06, "loss": 1.0706, "step": 4100 }, { "epoch": 0.07475581585696357, "grad_norm": 1.974657654762268, "learning_rate": 9.538900035627896e-06, "loss": 0.9992, "step": 4110 }, { "epoch": 0.07493770348678587, "grad_norm": 1.6557669639587402, "learning_rate": 9.537024883271767e-06, "loss": 1.024, "step": 4120 }, { "epoch": 0.07511959111660815, "grad_norm": 2.0316240787506104, "learning_rate": 9.535149730915639e-06, "loss": 1.0229, "step": 4130 }, { "epoch": 0.07530147874643045, "grad_norm": 1.706313967704773, "learning_rate": 9.533274578559508e-06, "loss": 1.0628, "step": 4140 }, { "epoch": 0.07548336637625275, "grad_norm": 2.0057051181793213, "learning_rate": 9.53139942620338e-06, "loss": 1.0441, "step": 4150 }, { "epoch": 0.07566525400607504, "grad_norm": 2.0549216270446777, "learning_rate": 9.529524273847251e-06, "loss": 1.0396, "step": 4160 }, { "epoch": 0.07584714163589734, "grad_norm": 2.589655637741089, "learning_rate": 9.527649121491122e-06, "loss": 0.9863, "step": 4170 }, { "epoch": 0.07602902926571964, "grad_norm": 2.003836154937744, "learning_rate": 9.525773969134992e-06, "loss": 1.0434, "step": 4180 }, { "epoch": 0.07621091689554194, "grad_norm": 2.1027398109436035, "learning_rate": 9.523898816778863e-06, "loss": 1.074, "step": 4190 }, { "epoch": 0.07639280452536423, "grad_norm": 2.0363476276397705, "learning_rate": 9.522023664422735e-06, "loss": 1.0803, "step": 4200 }, { "epoch": 0.07657469215518653, "grad_norm": 1.9485821723937988, "learning_rate": 9.520148512066606e-06, "loss": 1.0664, "step": 4210 }, { "epoch": 0.07675657978500883, "grad_norm": 2.194520950317383, "learning_rate": 9.518273359710477e-06, "loss": 0.994, "step": 4220 }, { "epoch": 0.07693846741483111, "grad_norm": 1.7853076457977295, "learning_rate": 9.516398207354349e-06, "loss": 1.0069, "step": 4230 }, { "epoch": 0.07712035504465341, "grad_norm": 2.006049156188965, "learning_rate": 9.51452305499822e-06, "loss": 1.0827, "step": 4240 }, { "epoch": 0.0773022426744757, "grad_norm": 1.867207646369934, "learning_rate": 9.51264790264209e-06, "loss": 0.982, "step": 4250 }, { "epoch": 0.077484130304298, "grad_norm": 1.5956778526306152, "learning_rate": 9.510772750285961e-06, "loss": 1.0043, "step": 4260 }, { "epoch": 0.0776660179341203, "grad_norm": 1.8642373085021973, "learning_rate": 9.508897597929832e-06, "loss": 1.0728, "step": 4270 }, { "epoch": 0.0778479055639426, "grad_norm": 1.7842274904251099, "learning_rate": 9.507022445573704e-06, "loss": 1.0246, "step": 4280 }, { "epoch": 0.0780297931937649, "grad_norm": 1.9453967809677124, "learning_rate": 9.505147293217575e-06, "loss": 1.0835, "step": 4290 }, { "epoch": 0.07821168082358719, "grad_norm": 1.915901780128479, "learning_rate": 9.503272140861446e-06, "loss": 1.0492, "step": 4300 }, { "epoch": 0.07839356845340949, "grad_norm": 2.0741169452667236, "learning_rate": 9.501396988505318e-06, "loss": 1.0344, "step": 4310 }, { "epoch": 0.07857545608323178, "grad_norm": 1.8070313930511475, "learning_rate": 9.499521836149187e-06, "loss": 1.0836, "step": 4320 }, { "epoch": 0.07875734371305408, "grad_norm": 1.938774585723877, "learning_rate": 9.497646683793059e-06, "loss": 0.9954, "step": 4330 }, { "epoch": 0.07893923134287636, "grad_norm": 1.7961421012878418, "learning_rate": 9.49577153143693e-06, "loss": 1.0404, "step": 4340 }, { "epoch": 0.07912111897269866, "grad_norm": 1.9945311546325684, "learning_rate": 9.493896379080801e-06, "loss": 1.0216, "step": 4350 }, { "epoch": 0.07930300660252096, "grad_norm": 1.5361127853393555, "learning_rate": 9.492021226724673e-06, "loss": 1.0231, "step": 4360 }, { "epoch": 0.07948489423234326, "grad_norm": 2.064429998397827, "learning_rate": 9.490146074368544e-06, "loss": 1.0328, "step": 4370 }, { "epoch": 0.07966678186216555, "grad_norm": 1.8996516466140747, "learning_rate": 9.488270922012415e-06, "loss": 1.0248, "step": 4380 }, { "epoch": 0.07984866949198785, "grad_norm": 1.8914896249771118, "learning_rate": 9.486395769656285e-06, "loss": 1.0447, "step": 4390 }, { "epoch": 0.08003055712181015, "grad_norm": 1.6722475290298462, "learning_rate": 9.484520617300156e-06, "loss": 1.0215, "step": 4400 }, { "epoch": 0.08021244475163244, "grad_norm": 1.766554355621338, "learning_rate": 9.482645464944028e-06, "loss": 1.0194, "step": 4410 }, { "epoch": 0.08039433238145474, "grad_norm": 2.324265480041504, "learning_rate": 9.480770312587897e-06, "loss": 1.0536, "step": 4420 }, { "epoch": 0.08057622001127704, "grad_norm": 1.943267583847046, "learning_rate": 9.478895160231769e-06, "loss": 1.073, "step": 4430 }, { "epoch": 0.08075810764109932, "grad_norm": 2.042475461959839, "learning_rate": 9.47702000787564e-06, "loss": 1.049, "step": 4440 }, { "epoch": 0.08093999527092162, "grad_norm": 1.7257215976715088, "learning_rate": 9.475144855519511e-06, "loss": 1.0211, "step": 4450 }, { "epoch": 0.08112188290074392, "grad_norm": 2.067899703979492, "learning_rate": 9.473269703163383e-06, "loss": 1.0348, "step": 4460 }, { "epoch": 0.08130377053056621, "grad_norm": 2.0670583248138428, "learning_rate": 9.471394550807254e-06, "loss": 1.0397, "step": 4470 }, { "epoch": 0.08148565816038851, "grad_norm": 1.8541423082351685, "learning_rate": 9.469519398451126e-06, "loss": 1.0939, "step": 4480 }, { "epoch": 0.08166754579021081, "grad_norm": 1.8613542318344116, "learning_rate": 9.467644246094995e-06, "loss": 1.0324, "step": 4490 }, { "epoch": 0.0818494334200331, "grad_norm": 1.7599711418151855, "learning_rate": 9.465769093738866e-06, "loss": 1.0413, "step": 4500 }, { "epoch": 0.0820313210498554, "grad_norm": 1.908406376838684, "learning_rate": 9.463893941382738e-06, "loss": 1.0578, "step": 4510 }, { "epoch": 0.0822132086796777, "grad_norm": 2.2544240951538086, "learning_rate": 9.46201878902661e-06, "loss": 1.0689, "step": 4520 }, { "epoch": 0.0823950963095, "grad_norm": 2.085304021835327, "learning_rate": 9.46014363667048e-06, "loss": 1.0344, "step": 4530 }, { "epoch": 0.0825769839393223, "grad_norm": 2.0172603130340576, "learning_rate": 9.458268484314352e-06, "loss": 1.1153, "step": 4540 }, { "epoch": 0.08275887156914458, "grad_norm": 1.9172337055206299, "learning_rate": 9.456393331958223e-06, "loss": 1.0544, "step": 4550 }, { "epoch": 0.08294075919896687, "grad_norm": 2.1904711723327637, "learning_rate": 9.454518179602093e-06, "loss": 1.0221, "step": 4560 }, { "epoch": 0.08312264682878917, "grad_norm": 1.9674263000488281, "learning_rate": 9.452643027245964e-06, "loss": 1.0214, "step": 4570 }, { "epoch": 0.08330453445861147, "grad_norm": 2.1888327598571777, "learning_rate": 9.450767874889836e-06, "loss": 1.1126, "step": 4580 }, { "epoch": 0.08348642208843376, "grad_norm": 2.1656699180603027, "learning_rate": 9.448892722533707e-06, "loss": 1.0239, "step": 4590 }, { "epoch": 0.08366830971825606, "grad_norm": 2.030341148376465, "learning_rate": 9.447017570177578e-06, "loss": 1.0392, "step": 4600 }, { "epoch": 0.08385019734807836, "grad_norm": 2.0368993282318115, "learning_rate": 9.44514241782145e-06, "loss": 1.0324, "step": 4610 }, { "epoch": 0.08403208497790066, "grad_norm": 1.9260778427124023, "learning_rate": 9.443267265465321e-06, "loss": 1.0194, "step": 4620 }, { "epoch": 0.08421397260772295, "grad_norm": 1.9122116565704346, "learning_rate": 9.44139211310919e-06, "loss": 1.0269, "step": 4630 }, { "epoch": 0.08439586023754525, "grad_norm": 1.8831114768981934, "learning_rate": 9.439516960753062e-06, "loss": 1.0249, "step": 4640 }, { "epoch": 0.08457774786736753, "grad_norm": 1.9987093210220337, "learning_rate": 9.437641808396933e-06, "loss": 1.0654, "step": 4650 }, { "epoch": 0.08475963549718983, "grad_norm": 1.3792049884796143, "learning_rate": 9.435766656040805e-06, "loss": 0.9917, "step": 4660 }, { "epoch": 0.08494152312701213, "grad_norm": 1.9566466808319092, "learning_rate": 9.433891503684676e-06, "loss": 1.0666, "step": 4670 }, { "epoch": 0.08512341075683443, "grad_norm": 1.811153531074524, "learning_rate": 9.432016351328547e-06, "loss": 1.0303, "step": 4680 }, { "epoch": 0.08530529838665672, "grad_norm": 2.100069999694824, "learning_rate": 9.430141198972417e-06, "loss": 1.124, "step": 4690 }, { "epoch": 0.08548718601647902, "grad_norm": 1.9234416484832764, "learning_rate": 9.428266046616288e-06, "loss": 0.9807, "step": 4700 }, { "epoch": 0.08566907364630132, "grad_norm": 2.102048635482788, "learning_rate": 9.42639089426016e-06, "loss": 1.0293, "step": 4710 }, { "epoch": 0.08585096127612361, "grad_norm": 2.3038628101348877, "learning_rate": 9.42451574190403e-06, "loss": 1.0337, "step": 4720 }, { "epoch": 0.08603284890594591, "grad_norm": 1.978508472442627, "learning_rate": 9.4226405895479e-06, "loss": 1.0226, "step": 4730 }, { "epoch": 0.08621473653576821, "grad_norm": 1.786424160003662, "learning_rate": 9.420765437191772e-06, "loss": 1.0737, "step": 4740 }, { "epoch": 0.0863966241655905, "grad_norm": 1.9865491390228271, "learning_rate": 9.418890284835643e-06, "loss": 1.0795, "step": 4750 }, { "epoch": 0.08657851179541279, "grad_norm": 2.008565902709961, "learning_rate": 9.417015132479515e-06, "loss": 1.0593, "step": 4760 }, { "epoch": 0.08676039942523509, "grad_norm": 2.0048575401306152, "learning_rate": 9.415139980123386e-06, "loss": 1.02, "step": 4770 }, { "epoch": 0.08694228705505738, "grad_norm": 2.0685856342315674, "learning_rate": 9.413264827767257e-06, "loss": 1.039, "step": 4780 }, { "epoch": 0.08712417468487968, "grad_norm": 2.117199659347534, "learning_rate": 9.411389675411127e-06, "loss": 0.9881, "step": 4790 }, { "epoch": 0.08730606231470198, "grad_norm": 2.102531671524048, "learning_rate": 9.409514523054998e-06, "loss": 1.0469, "step": 4800 }, { "epoch": 0.08748794994452427, "grad_norm": 2.0938868522644043, "learning_rate": 9.40763937069887e-06, "loss": 1.0642, "step": 4810 }, { "epoch": 0.08766983757434657, "grad_norm": 2.089996099472046, "learning_rate": 9.405764218342741e-06, "loss": 1.0272, "step": 4820 }, { "epoch": 0.08785172520416887, "grad_norm": 1.9860011339187622, "learning_rate": 9.403889065986612e-06, "loss": 0.998, "step": 4830 }, { "epoch": 0.08803361283399116, "grad_norm": 1.6000299453735352, "learning_rate": 9.402013913630484e-06, "loss": 1.0205, "step": 4840 }, { "epoch": 0.08821550046381346, "grad_norm": 2.323111057281494, "learning_rate": 9.400138761274355e-06, "loss": 1.0215, "step": 4850 }, { "epoch": 0.08839738809363575, "grad_norm": 2.23721981048584, "learning_rate": 9.398263608918225e-06, "loss": 1.0622, "step": 4860 }, { "epoch": 0.08857927572345804, "grad_norm": 1.9715780019760132, "learning_rate": 9.396388456562096e-06, "loss": 1.02, "step": 4870 }, { "epoch": 0.08876116335328034, "grad_norm": 1.9683715105056763, "learning_rate": 9.394513304205967e-06, "loss": 1.0182, "step": 4880 }, { "epoch": 0.08894305098310264, "grad_norm": 1.6821422576904297, "learning_rate": 9.392638151849839e-06, "loss": 0.9696, "step": 4890 }, { "epoch": 0.08912493861292493, "grad_norm": 1.990465521812439, "learning_rate": 9.39076299949371e-06, "loss": 1.0728, "step": 4900 }, { "epoch": 0.08930682624274723, "grad_norm": 1.7876290082931519, "learning_rate": 9.388887847137582e-06, "loss": 1.0152, "step": 4910 }, { "epoch": 0.08948871387256953, "grad_norm": 2.0891644954681396, "learning_rate": 9.387012694781453e-06, "loss": 0.9841, "step": 4920 }, { "epoch": 0.08967060150239183, "grad_norm": 2.068150520324707, "learning_rate": 9.385137542425323e-06, "loss": 1.0547, "step": 4930 }, { "epoch": 0.08985248913221412, "grad_norm": 1.8828814029693604, "learning_rate": 9.383262390069194e-06, "loss": 0.9552, "step": 4940 }, { "epoch": 0.09003437676203642, "grad_norm": 2.041520595550537, "learning_rate": 9.381387237713065e-06, "loss": 1.0375, "step": 4950 }, { "epoch": 0.09021626439185872, "grad_norm": 2.0959219932556152, "learning_rate": 9.379512085356935e-06, "loss": 1.0768, "step": 4960 }, { "epoch": 0.090398152021681, "grad_norm": 2.1579341888427734, "learning_rate": 9.377636933000806e-06, "loss": 1.0722, "step": 4970 }, { "epoch": 0.0905800396515033, "grad_norm": 1.864747166633606, "learning_rate": 9.375761780644678e-06, "loss": 1.0729, "step": 4980 }, { "epoch": 0.0907619272813256, "grad_norm": 2.0034849643707275, "learning_rate": 9.373886628288549e-06, "loss": 0.9843, "step": 4990 }, { "epoch": 0.09094381491114789, "grad_norm": 1.8855655193328857, "learning_rate": 9.37201147593242e-06, "loss": 1.0152, "step": 5000 }, { "epoch": 0.09112570254097019, "grad_norm": 1.9928489923477173, "learning_rate": 9.370136323576292e-06, "loss": 1.0043, "step": 5010 }, { "epoch": 0.09130759017079249, "grad_norm": 1.572189211845398, "learning_rate": 9.368261171220163e-06, "loss": 0.9994, "step": 5020 }, { "epoch": 0.09148947780061478, "grad_norm": 1.808457612991333, "learning_rate": 9.366386018864033e-06, "loss": 1.0145, "step": 5030 }, { "epoch": 0.09167136543043708, "grad_norm": 2.1113319396972656, "learning_rate": 9.364510866507904e-06, "loss": 1.0645, "step": 5040 }, { "epoch": 0.09185325306025938, "grad_norm": 1.7390172481536865, "learning_rate": 9.362635714151775e-06, "loss": 0.9303, "step": 5050 }, { "epoch": 0.09203514069008167, "grad_norm": 1.99275803565979, "learning_rate": 9.360760561795647e-06, "loss": 1.0049, "step": 5060 }, { "epoch": 0.09221702831990396, "grad_norm": 2.0742483139038086, "learning_rate": 9.358885409439518e-06, "loss": 1.0311, "step": 5070 }, { "epoch": 0.09239891594972625, "grad_norm": 1.9582628011703491, "learning_rate": 9.35701025708339e-06, "loss": 1.0192, "step": 5080 }, { "epoch": 0.09258080357954855, "grad_norm": 1.8456971645355225, "learning_rate": 9.35513510472726e-06, "loss": 1.0577, "step": 5090 }, { "epoch": 0.09276269120937085, "grad_norm": 2.0255110263824463, "learning_rate": 9.35325995237113e-06, "loss": 1.0726, "step": 5100 }, { "epoch": 0.09294457883919315, "grad_norm": 1.8210622072219849, "learning_rate": 9.351384800015002e-06, "loss": 1.0334, "step": 5110 }, { "epoch": 0.09312646646901544, "grad_norm": 1.7578660249710083, "learning_rate": 9.349509647658873e-06, "loss": 1.0523, "step": 5120 }, { "epoch": 0.09330835409883774, "grad_norm": 1.8944443464279175, "learning_rate": 9.347634495302744e-06, "loss": 1.0041, "step": 5130 }, { "epoch": 0.09349024172866004, "grad_norm": 1.9928462505340576, "learning_rate": 9.345759342946616e-06, "loss": 1.0684, "step": 5140 }, { "epoch": 0.09367212935848233, "grad_norm": 2.2086362838745117, "learning_rate": 9.343884190590487e-06, "loss": 1.0026, "step": 5150 }, { "epoch": 0.09385401698830463, "grad_norm": 2.05908203125, "learning_rate": 9.342009038234358e-06, "loss": 1.0124, "step": 5160 }, { "epoch": 0.09403590461812691, "grad_norm": 1.9666191339492798, "learning_rate": 9.340133885878228e-06, "loss": 1.002, "step": 5170 }, { "epoch": 0.09421779224794921, "grad_norm": 2.197920799255371, "learning_rate": 9.3382587335221e-06, "loss": 1.0106, "step": 5180 }, { "epoch": 0.09439967987777151, "grad_norm": 1.7947094440460205, "learning_rate": 9.33638358116597e-06, "loss": 1.0332, "step": 5190 }, { "epoch": 0.0945815675075938, "grad_norm": 2.1186506748199463, "learning_rate": 9.334508428809842e-06, "loss": 1.0058, "step": 5200 }, { "epoch": 0.0947634551374161, "grad_norm": 2.0192208290100098, "learning_rate": 9.332633276453713e-06, "loss": 1.0087, "step": 5210 }, { "epoch": 0.0949453427672384, "grad_norm": 1.9937084913253784, "learning_rate": 9.330758124097585e-06, "loss": 1.082, "step": 5220 }, { "epoch": 0.0951272303970607, "grad_norm": 2.007432222366333, "learning_rate": 9.328882971741456e-06, "loss": 1.0082, "step": 5230 }, { "epoch": 0.095309118026883, "grad_norm": 2.060500144958496, "learning_rate": 9.327007819385326e-06, "loss": 1.0405, "step": 5240 }, { "epoch": 0.09549100565670529, "grad_norm": 2.117885112762451, "learning_rate": 9.325132667029197e-06, "loss": 1.0441, "step": 5250 }, { "epoch": 0.09567289328652759, "grad_norm": 2.03344988822937, "learning_rate": 9.323257514673067e-06, "loss": 0.9958, "step": 5260 }, { "epoch": 0.09585478091634989, "grad_norm": 1.92274010181427, "learning_rate": 9.321382362316938e-06, "loss": 0.978, "step": 5270 }, { "epoch": 0.09603666854617217, "grad_norm": 2.2551653385162354, "learning_rate": 9.31950720996081e-06, "loss": 1.0233, "step": 5280 }, { "epoch": 0.09621855617599447, "grad_norm": 2.3190758228302, "learning_rate": 9.31763205760468e-06, "loss": 1.0359, "step": 5290 }, { "epoch": 0.09640044380581676, "grad_norm": 1.8211904764175415, "learning_rate": 9.315756905248552e-06, "loss": 0.9773, "step": 5300 }, { "epoch": 0.09658233143563906, "grad_norm": 1.9140715599060059, "learning_rate": 9.313881752892424e-06, "loss": 1.0518, "step": 5310 }, { "epoch": 0.09676421906546136, "grad_norm": 1.8997334241867065, "learning_rate": 9.312006600536295e-06, "loss": 1.0445, "step": 5320 }, { "epoch": 0.09694610669528365, "grad_norm": 2.084719181060791, "learning_rate": 9.310131448180165e-06, "loss": 1.0312, "step": 5330 }, { "epoch": 0.09712799432510595, "grad_norm": 1.8229697942733765, "learning_rate": 9.308256295824036e-06, "loss": 1.0311, "step": 5340 }, { "epoch": 0.09730988195492825, "grad_norm": 2.027582883834839, "learning_rate": 9.306381143467907e-06, "loss": 1.0168, "step": 5350 }, { "epoch": 0.09749176958475055, "grad_norm": 2.087104320526123, "learning_rate": 9.304505991111779e-06, "loss": 1.0382, "step": 5360 }, { "epoch": 0.09767365721457284, "grad_norm": 1.950005292892456, "learning_rate": 9.30263083875565e-06, "loss": 1.0425, "step": 5370 }, { "epoch": 0.09785554484439513, "grad_norm": 2.183122158050537, "learning_rate": 9.300755686399521e-06, "loss": 1.0677, "step": 5380 }, { "epoch": 0.09803743247421742, "grad_norm": 1.8889926671981812, "learning_rate": 9.298880534043393e-06, "loss": 1.0014, "step": 5390 }, { "epoch": 0.09821932010403972, "grad_norm": 2.19388747215271, "learning_rate": 9.297005381687262e-06, "loss": 1.0312, "step": 5400 }, { "epoch": 0.09840120773386202, "grad_norm": 1.8085167407989502, "learning_rate": 9.295130229331134e-06, "loss": 0.9943, "step": 5410 }, { "epoch": 0.09858309536368431, "grad_norm": 1.7752214670181274, "learning_rate": 9.293255076975005e-06, "loss": 0.9966, "step": 5420 }, { "epoch": 0.09876498299350661, "grad_norm": 2.031956672668457, "learning_rate": 9.291379924618876e-06, "loss": 1.0907, "step": 5430 }, { "epoch": 0.09894687062332891, "grad_norm": 2.046603202819824, "learning_rate": 9.289504772262748e-06, "loss": 0.9971, "step": 5440 }, { "epoch": 0.0991287582531512, "grad_norm": 1.9869693517684937, "learning_rate": 9.287629619906619e-06, "loss": 1.0066, "step": 5450 }, { "epoch": 0.0993106458829735, "grad_norm": 1.8923308849334717, "learning_rate": 9.28575446755049e-06, "loss": 1.005, "step": 5460 }, { "epoch": 0.0994925335127958, "grad_norm": 1.922049880027771, "learning_rate": 9.28387931519436e-06, "loss": 1.0302, "step": 5470 }, { "epoch": 0.0996744211426181, "grad_norm": 2.079739809036255, "learning_rate": 9.282004162838231e-06, "loss": 1.0375, "step": 5480 }, { "epoch": 0.09985630877244038, "grad_norm": 2.048841953277588, "learning_rate": 9.280129010482103e-06, "loss": 1.0511, "step": 5490 }, { "epoch": 0.10003819640226268, "grad_norm": 2.0528860092163086, "learning_rate": 9.278253858125974e-06, "loss": 1.0567, "step": 5500 }, { "epoch": 0.10022008403208497, "grad_norm": 1.9672099351882935, "learning_rate": 9.276378705769844e-06, "loss": 1.0384, "step": 5510 }, { "epoch": 0.10040197166190727, "grad_norm": 1.9451953172683716, "learning_rate": 9.274503553413715e-06, "loss": 1.0372, "step": 5520 }, { "epoch": 0.10058385929172957, "grad_norm": 1.6539404392242432, "learning_rate": 9.272628401057586e-06, "loss": 0.9863, "step": 5530 }, { "epoch": 0.10076574692155187, "grad_norm": 1.9275966882705688, "learning_rate": 9.270753248701458e-06, "loss": 1.0068, "step": 5540 }, { "epoch": 0.10094763455137416, "grad_norm": 2.21311092376709, "learning_rate": 9.268878096345329e-06, "loss": 1.0259, "step": 5550 }, { "epoch": 0.10112952218119646, "grad_norm": 2.095170021057129, "learning_rate": 9.267002943989199e-06, "loss": 1.0594, "step": 5560 }, { "epoch": 0.10131140981101876, "grad_norm": 1.963150978088379, "learning_rate": 9.26512779163307e-06, "loss": 1.0749, "step": 5570 }, { "epoch": 0.10149329744084105, "grad_norm": 1.7124818563461304, "learning_rate": 9.263252639276941e-06, "loss": 1.0174, "step": 5580 }, { "epoch": 0.10167518507066334, "grad_norm": 2.0797741413116455, "learning_rate": 9.261377486920813e-06, "loss": 1.0014, "step": 5590 }, { "epoch": 0.10185707270048563, "grad_norm": 1.8911874294281006, "learning_rate": 9.259502334564684e-06, "loss": 0.9917, "step": 5600 }, { "epoch": 0.10203896033030793, "grad_norm": 1.871750831604004, "learning_rate": 9.257627182208555e-06, "loss": 1.0404, "step": 5610 }, { "epoch": 0.10222084796013023, "grad_norm": 1.8750195503234863, "learning_rate": 9.255752029852427e-06, "loss": 1.0171, "step": 5620 }, { "epoch": 0.10240273558995253, "grad_norm": 1.9541122913360596, "learning_rate": 9.253876877496298e-06, "loss": 1.0833, "step": 5630 }, { "epoch": 0.10258462321977482, "grad_norm": 1.532046914100647, "learning_rate": 9.252001725140168e-06, "loss": 0.9713, "step": 5640 }, { "epoch": 0.10276651084959712, "grad_norm": 2.205430030822754, "learning_rate": 9.250126572784039e-06, "loss": 1.0293, "step": 5650 }, { "epoch": 0.10294839847941942, "grad_norm": 1.9942052364349365, "learning_rate": 9.24825142042791e-06, "loss": 1.0049, "step": 5660 }, { "epoch": 0.10313028610924171, "grad_norm": 1.7362048625946045, "learning_rate": 9.246376268071782e-06, "loss": 1.0183, "step": 5670 }, { "epoch": 0.10331217373906401, "grad_norm": 2.1242446899414062, "learning_rate": 9.244501115715653e-06, "loss": 1.0636, "step": 5680 }, { "epoch": 0.10349406136888631, "grad_norm": 2.039032220840454, "learning_rate": 9.242625963359524e-06, "loss": 1.0018, "step": 5690 }, { "epoch": 0.10367594899870859, "grad_norm": 1.829018235206604, "learning_rate": 9.240750811003396e-06, "loss": 1.0359, "step": 5700 }, { "epoch": 0.10385783662853089, "grad_norm": 1.7071303129196167, "learning_rate": 9.238875658647265e-06, "loss": 1.0256, "step": 5710 }, { "epoch": 0.10403972425835319, "grad_norm": 1.8934966325759888, "learning_rate": 9.237000506291137e-06, "loss": 1.0281, "step": 5720 }, { "epoch": 0.10422161188817548, "grad_norm": 2.2118289470672607, "learning_rate": 9.235125353935008e-06, "loss": 1.0369, "step": 5730 }, { "epoch": 0.10440349951799778, "grad_norm": 1.9886342287063599, "learning_rate": 9.23325020157888e-06, "loss": 1.0379, "step": 5740 }, { "epoch": 0.10458538714782008, "grad_norm": 2.0176427364349365, "learning_rate": 9.231375049222751e-06, "loss": 1.031, "step": 5750 }, { "epoch": 0.10476727477764237, "grad_norm": 1.631627082824707, "learning_rate": 9.229499896866622e-06, "loss": 0.9671, "step": 5760 }, { "epoch": 0.10494916240746467, "grad_norm": 1.8476777076721191, "learning_rate": 9.227624744510494e-06, "loss": 1.0421, "step": 5770 }, { "epoch": 0.10513105003728697, "grad_norm": 2.1775872707366943, "learning_rate": 9.225749592154363e-06, "loss": 1.0718, "step": 5780 }, { "epoch": 0.10531293766710927, "grad_norm": 1.6606237888336182, "learning_rate": 9.223874439798235e-06, "loss": 1.0605, "step": 5790 }, { "epoch": 0.10549482529693155, "grad_norm": 2.2568509578704834, "learning_rate": 9.221999287442104e-06, "loss": 0.9948, "step": 5800 }, { "epoch": 0.10567671292675385, "grad_norm": 1.909332036972046, "learning_rate": 9.220124135085976e-06, "loss": 0.9495, "step": 5810 }, { "epoch": 0.10585860055657614, "grad_norm": 1.8219736814498901, "learning_rate": 9.218248982729847e-06, "loss": 1.0559, "step": 5820 }, { "epoch": 0.10604048818639844, "grad_norm": 1.758324384689331, "learning_rate": 9.216373830373718e-06, "loss": 1.0622, "step": 5830 }, { "epoch": 0.10622237581622074, "grad_norm": 1.9071513414382935, "learning_rate": 9.21449867801759e-06, "loss": 1.0103, "step": 5840 }, { "epoch": 0.10640426344604303, "grad_norm": 1.7847944498062134, "learning_rate": 9.212623525661461e-06, "loss": 1.0489, "step": 5850 }, { "epoch": 0.10658615107586533, "grad_norm": 2.000011682510376, "learning_rate": 9.210748373305332e-06, "loss": 1.0618, "step": 5860 }, { "epoch": 0.10676803870568763, "grad_norm": 1.758354663848877, "learning_rate": 9.208873220949202e-06, "loss": 0.9801, "step": 5870 }, { "epoch": 0.10694992633550993, "grad_norm": 1.5367717742919922, "learning_rate": 9.206998068593073e-06, "loss": 0.9596, "step": 5880 }, { "epoch": 0.10713181396533222, "grad_norm": 2.2171790599823, "learning_rate": 9.205122916236945e-06, "loss": 1.0392, "step": 5890 }, { "epoch": 0.10731370159515452, "grad_norm": 1.5651408433914185, "learning_rate": 9.203247763880816e-06, "loss": 1.0162, "step": 5900 }, { "epoch": 0.1074955892249768, "grad_norm": 1.8953105211257935, "learning_rate": 9.201372611524687e-06, "loss": 1.0601, "step": 5910 }, { "epoch": 0.1076774768547991, "grad_norm": 1.9191170930862427, "learning_rate": 9.199497459168559e-06, "loss": 1.0543, "step": 5920 }, { "epoch": 0.1078593644846214, "grad_norm": 1.719555139541626, "learning_rate": 9.19762230681243e-06, "loss": 1.0268, "step": 5930 }, { "epoch": 0.1080412521144437, "grad_norm": 1.564473271369934, "learning_rate": 9.1957471544563e-06, "loss": 0.9658, "step": 5940 }, { "epoch": 0.10822313974426599, "grad_norm": 1.8770138025283813, "learning_rate": 9.193872002100171e-06, "loss": 1.0426, "step": 5950 }, { "epoch": 0.10840502737408829, "grad_norm": 1.9052034616470337, "learning_rate": 9.191996849744042e-06, "loss": 0.962, "step": 5960 }, { "epoch": 0.10858691500391059, "grad_norm": 1.8631007671356201, "learning_rate": 9.190121697387914e-06, "loss": 1.0029, "step": 5970 }, { "epoch": 0.10876880263373288, "grad_norm": 1.7974355220794678, "learning_rate": 9.188246545031785e-06, "loss": 0.9768, "step": 5980 }, { "epoch": 0.10895069026355518, "grad_norm": 1.9188841581344604, "learning_rate": 9.186371392675656e-06, "loss": 1.0498, "step": 5990 }, { "epoch": 0.10913257789337748, "grad_norm": 1.880696177482605, "learning_rate": 9.184496240319528e-06, "loss": 1.0334, "step": 6000 }, { "epoch": 0.10931446552319976, "grad_norm": 1.9165536165237427, "learning_rate": 9.182621087963397e-06, "loss": 1.0544, "step": 6010 }, { "epoch": 0.10949635315302206, "grad_norm": 2.101367712020874, "learning_rate": 9.180745935607269e-06, "loss": 1.0384, "step": 6020 }, { "epoch": 0.10967824078284436, "grad_norm": 2.1352245807647705, "learning_rate": 9.17887078325114e-06, "loss": 1.0354, "step": 6030 }, { "epoch": 0.10986012841266665, "grad_norm": 1.973879337310791, "learning_rate": 9.176995630895011e-06, "loss": 1.0252, "step": 6040 }, { "epoch": 0.11004201604248895, "grad_norm": 1.8103615045547485, "learning_rate": 9.175120478538883e-06, "loss": 1.0186, "step": 6050 }, { "epoch": 0.11022390367231125, "grad_norm": 1.7248214483261108, "learning_rate": 9.173245326182754e-06, "loss": 1.0059, "step": 6060 }, { "epoch": 0.11040579130213354, "grad_norm": 1.9427331686019897, "learning_rate": 9.171370173826624e-06, "loss": 1.0403, "step": 6070 }, { "epoch": 0.11058767893195584, "grad_norm": 1.6445019245147705, "learning_rate": 9.169495021470495e-06, "loss": 1.0314, "step": 6080 }, { "epoch": 0.11076956656177814, "grad_norm": 1.6631091833114624, "learning_rate": 9.167619869114366e-06, "loss": 0.9426, "step": 6090 }, { "epoch": 0.11095145419160043, "grad_norm": 1.7824392318725586, "learning_rate": 9.165744716758236e-06, "loss": 1.0101, "step": 6100 }, { "epoch": 0.11113334182142272, "grad_norm": 1.726915955543518, "learning_rate": 9.163869564402107e-06, "loss": 0.9914, "step": 6110 }, { "epoch": 0.11131522945124502, "grad_norm": 1.5673582553863525, "learning_rate": 9.161994412045979e-06, "loss": 1.0192, "step": 6120 }, { "epoch": 0.11149711708106731, "grad_norm": 1.7408254146575928, "learning_rate": 9.16011925968985e-06, "loss": 0.9645, "step": 6130 }, { "epoch": 0.11167900471088961, "grad_norm": 2.4571914672851562, "learning_rate": 9.158244107333722e-06, "loss": 1.063, "step": 6140 }, { "epoch": 0.1118608923407119, "grad_norm": 1.7167598009109497, "learning_rate": 9.156368954977593e-06, "loss": 0.977, "step": 6150 }, { "epoch": 0.1120427799705342, "grad_norm": 1.9731882810592651, "learning_rate": 9.154493802621464e-06, "loss": 1.0442, "step": 6160 }, { "epoch": 0.1122246676003565, "grad_norm": 1.7314952611923218, "learning_rate": 9.152618650265334e-06, "loss": 1.0259, "step": 6170 }, { "epoch": 0.1124065552301788, "grad_norm": 2.0027003288269043, "learning_rate": 9.150743497909205e-06, "loss": 1.0512, "step": 6180 }, { "epoch": 0.1125884428600011, "grad_norm": 2.214618682861328, "learning_rate": 9.148868345553077e-06, "loss": 1.0419, "step": 6190 }, { "epoch": 0.11277033048982339, "grad_norm": 2.0078604221343994, "learning_rate": 9.146993193196948e-06, "loss": 1.0516, "step": 6200 }, { "epoch": 0.11295221811964569, "grad_norm": 1.9138555526733398, "learning_rate": 9.14511804084082e-06, "loss": 1.0386, "step": 6210 }, { "epoch": 0.11313410574946797, "grad_norm": 2.108290433883667, "learning_rate": 9.14324288848469e-06, "loss": 1.0227, "step": 6220 }, { "epoch": 0.11331599337929027, "grad_norm": 2.0754683017730713, "learning_rate": 9.141367736128562e-06, "loss": 1.0355, "step": 6230 }, { "epoch": 0.11349788100911257, "grad_norm": 1.7072536945343018, "learning_rate": 9.139492583772432e-06, "loss": 1.0895, "step": 6240 }, { "epoch": 0.11367976863893486, "grad_norm": 2.033083438873291, "learning_rate": 9.137617431416303e-06, "loss": 1.0315, "step": 6250 }, { "epoch": 0.11386165626875716, "grad_norm": 1.8223384618759155, "learning_rate": 9.135742279060174e-06, "loss": 0.9786, "step": 6260 }, { "epoch": 0.11404354389857946, "grad_norm": 1.6871641874313354, "learning_rate": 9.133867126704046e-06, "loss": 0.9602, "step": 6270 }, { "epoch": 0.11422543152840176, "grad_norm": 1.9660054445266724, "learning_rate": 9.131991974347917e-06, "loss": 1.0524, "step": 6280 }, { "epoch": 0.11440731915822405, "grad_norm": 2.29740047454834, "learning_rate": 9.130116821991788e-06, "loss": 1.0807, "step": 6290 }, { "epoch": 0.11458920678804635, "grad_norm": 1.8881101608276367, "learning_rate": 9.12824166963566e-06, "loss": 0.9766, "step": 6300 }, { "epoch": 0.11477109441786865, "grad_norm": 1.8412954807281494, "learning_rate": 9.126366517279531e-06, "loss": 0.9808, "step": 6310 }, { "epoch": 0.11495298204769093, "grad_norm": 1.4697201251983643, "learning_rate": 9.1244913649234e-06, "loss": 0.9826, "step": 6320 }, { "epoch": 0.11513486967751323, "grad_norm": 1.6962069272994995, "learning_rate": 9.122616212567272e-06, "loss": 1.0632, "step": 6330 }, { "epoch": 0.11531675730733552, "grad_norm": 2.1623051166534424, "learning_rate": 9.120741060211142e-06, "loss": 1.0758, "step": 6340 }, { "epoch": 0.11549864493715782, "grad_norm": 1.9333816766738892, "learning_rate": 9.118865907855013e-06, "loss": 0.9939, "step": 6350 }, { "epoch": 0.11568053256698012, "grad_norm": 1.6929733753204346, "learning_rate": 9.116990755498884e-06, "loss": 0.9922, "step": 6360 }, { "epoch": 0.11586242019680242, "grad_norm": 1.7999440431594849, "learning_rate": 9.115115603142756e-06, "loss": 0.994, "step": 6370 }, { "epoch": 0.11604430782662471, "grad_norm": 2.0934410095214844, "learning_rate": 9.113240450786627e-06, "loss": 1.0591, "step": 6380 }, { "epoch": 0.11622619545644701, "grad_norm": 1.5314563512802124, "learning_rate": 9.111365298430498e-06, "loss": 0.9685, "step": 6390 }, { "epoch": 0.1164080830862693, "grad_norm": 2.105696439743042, "learning_rate": 9.10949014607437e-06, "loss": 1.0515, "step": 6400 }, { "epoch": 0.1165899707160916, "grad_norm": 2.225750684738159, "learning_rate": 9.10761499371824e-06, "loss": 1.0296, "step": 6410 }, { "epoch": 0.1167718583459139, "grad_norm": 1.9414364099502563, "learning_rate": 9.10573984136211e-06, "loss": 1.0067, "step": 6420 }, { "epoch": 0.11695374597573618, "grad_norm": 1.8306856155395508, "learning_rate": 9.103864689005982e-06, "loss": 1.0285, "step": 6430 }, { "epoch": 0.11713563360555848, "grad_norm": 1.4717962741851807, "learning_rate": 9.101989536649853e-06, "loss": 0.9989, "step": 6440 }, { "epoch": 0.11731752123538078, "grad_norm": 1.961300015449524, "learning_rate": 9.100114384293725e-06, "loss": 0.9956, "step": 6450 }, { "epoch": 0.11749940886520308, "grad_norm": 2.0891945362091064, "learning_rate": 9.098239231937596e-06, "loss": 0.9935, "step": 6460 }, { "epoch": 0.11768129649502537, "grad_norm": 1.846771001815796, "learning_rate": 9.096364079581467e-06, "loss": 0.982, "step": 6470 }, { "epoch": 0.11786318412484767, "grad_norm": 1.5889613628387451, "learning_rate": 9.094488927225337e-06, "loss": 1.0266, "step": 6480 }, { "epoch": 0.11804507175466997, "grad_norm": 1.94053053855896, "learning_rate": 9.092613774869208e-06, "loss": 1.0798, "step": 6490 }, { "epoch": 0.11822695938449226, "grad_norm": 1.9005244970321655, "learning_rate": 9.09073862251308e-06, "loss": 0.9885, "step": 6500 }, { "epoch": 0.11840884701431456, "grad_norm": 1.8771830797195435, "learning_rate": 9.088863470156951e-06, "loss": 0.9956, "step": 6510 }, { "epoch": 0.11859073464413686, "grad_norm": 2.129850387573242, "learning_rate": 9.086988317800822e-06, "loss": 1.003, "step": 6520 }, { "epoch": 0.11877262227395914, "grad_norm": 1.6929882764816284, "learning_rate": 9.085113165444694e-06, "loss": 1.0399, "step": 6530 }, { "epoch": 0.11895450990378144, "grad_norm": 2.094172239303589, "learning_rate": 9.083238013088565e-06, "loss": 1.0143, "step": 6540 }, { "epoch": 0.11913639753360374, "grad_norm": 1.7076716423034668, "learning_rate": 9.081362860732435e-06, "loss": 0.9344, "step": 6550 }, { "epoch": 0.11931828516342603, "grad_norm": 1.8881752490997314, "learning_rate": 9.079487708376306e-06, "loss": 1.0941, "step": 6560 }, { "epoch": 0.11950017279324833, "grad_norm": 1.9843730926513672, "learning_rate": 9.077612556020178e-06, "loss": 1.0388, "step": 6570 }, { "epoch": 0.11968206042307063, "grad_norm": 1.5764013528823853, "learning_rate": 9.075737403664049e-06, "loss": 1.0132, "step": 6580 }, { "epoch": 0.11986394805289292, "grad_norm": 1.9505935907363892, "learning_rate": 9.07386225130792e-06, "loss": 1.0367, "step": 6590 }, { "epoch": 0.12004583568271522, "grad_norm": 1.9338325262069702, "learning_rate": 9.071987098951792e-06, "loss": 0.9799, "step": 6600 }, { "epoch": 0.12022772331253752, "grad_norm": 1.6739287376403809, "learning_rate": 9.070111946595663e-06, "loss": 0.9801, "step": 6610 }, { "epoch": 0.12040961094235982, "grad_norm": 1.9977126121520996, "learning_rate": 9.068236794239533e-06, "loss": 1.0408, "step": 6620 }, { "epoch": 0.12059149857218211, "grad_norm": 2.1551601886749268, "learning_rate": 9.066361641883404e-06, "loss": 1.0079, "step": 6630 }, { "epoch": 0.1207733862020044, "grad_norm": 2.1326162815093994, "learning_rate": 9.064486489527274e-06, "loss": 1.0316, "step": 6640 }, { "epoch": 0.12095527383182669, "grad_norm": 1.8646138906478882, "learning_rate": 9.062611337171145e-06, "loss": 1.0002, "step": 6650 }, { "epoch": 0.12113716146164899, "grad_norm": 1.9916980266571045, "learning_rate": 9.060736184815016e-06, "loss": 0.9875, "step": 6660 }, { "epoch": 0.12131904909147129, "grad_norm": 1.9356499910354614, "learning_rate": 9.058861032458888e-06, "loss": 1.006, "step": 6670 }, { "epoch": 0.12150093672129358, "grad_norm": 1.6276991367340088, "learning_rate": 9.056985880102759e-06, "loss": 1.0645, "step": 6680 }, { "epoch": 0.12168282435111588, "grad_norm": 1.6595115661621094, "learning_rate": 9.05511072774663e-06, "loss": 0.97, "step": 6690 }, { "epoch": 0.12186471198093818, "grad_norm": 1.5573678016662598, "learning_rate": 9.053235575390502e-06, "loss": 0.9987, "step": 6700 }, { "epoch": 0.12204659961076048, "grad_norm": 2.3925092220306396, "learning_rate": 9.051360423034371e-06, "loss": 1.069, "step": 6710 }, { "epoch": 0.12222848724058277, "grad_norm": 2.164938449859619, "learning_rate": 9.049485270678243e-06, "loss": 1.0038, "step": 6720 }, { "epoch": 0.12241037487040507, "grad_norm": 2.1018762588500977, "learning_rate": 9.047610118322114e-06, "loss": 1.0369, "step": 6730 }, { "epoch": 0.12259226250022735, "grad_norm": 1.5960421562194824, "learning_rate": 9.045734965965985e-06, "loss": 1.0044, "step": 6740 }, { "epoch": 0.12277415013004965, "grad_norm": 1.7212382555007935, "learning_rate": 9.043859813609857e-06, "loss": 0.9844, "step": 6750 }, { "epoch": 0.12295603775987195, "grad_norm": 1.7172038555145264, "learning_rate": 9.041984661253728e-06, "loss": 1.006, "step": 6760 }, { "epoch": 0.12313792538969424, "grad_norm": 1.9231442213058472, "learning_rate": 9.0401095088976e-06, "loss": 1.0047, "step": 6770 }, { "epoch": 0.12331981301951654, "grad_norm": 2.178776979446411, "learning_rate": 9.038234356541469e-06, "loss": 1.0322, "step": 6780 }, { "epoch": 0.12350170064933884, "grad_norm": 1.754514455795288, "learning_rate": 9.03635920418534e-06, "loss": 0.9822, "step": 6790 }, { "epoch": 0.12368358827916114, "grad_norm": 2.019707441329956, "learning_rate": 9.034484051829212e-06, "loss": 1.0063, "step": 6800 }, { "epoch": 0.12386547590898343, "grad_norm": 2.1651487350463867, "learning_rate": 9.032608899473083e-06, "loss": 1.0617, "step": 6810 }, { "epoch": 0.12404736353880573, "grad_norm": 1.8383890390396118, "learning_rate": 9.030733747116954e-06, "loss": 1.0499, "step": 6820 }, { "epoch": 0.12422925116862803, "grad_norm": 2.209059715270996, "learning_rate": 9.028858594760826e-06, "loss": 1.044, "step": 6830 }, { "epoch": 0.12441113879845032, "grad_norm": 2.404381513595581, "learning_rate": 9.026983442404697e-06, "loss": 1.0296, "step": 6840 }, { "epoch": 0.12459302642827261, "grad_norm": 1.9811248779296875, "learning_rate": 9.025108290048567e-06, "loss": 1.0334, "step": 6850 }, { "epoch": 0.1247749140580949, "grad_norm": 1.7408416271209717, "learning_rate": 9.023233137692438e-06, "loss": 0.9952, "step": 6860 }, { "epoch": 0.1249568016879172, "grad_norm": 2.002206802368164, "learning_rate": 9.02135798533631e-06, "loss": 1.0494, "step": 6870 }, { "epoch": 0.1251386893177395, "grad_norm": 2.048963785171509, "learning_rate": 9.01948283298018e-06, "loss": 1.0266, "step": 6880 }, { "epoch": 0.1253205769475618, "grad_norm": 1.9269241094589233, "learning_rate": 9.01760768062405e-06, "loss": 0.9714, "step": 6890 }, { "epoch": 0.1255024645773841, "grad_norm": 1.878318190574646, "learning_rate": 9.015732528267922e-06, "loss": 1.018, "step": 6900 }, { "epoch": 0.1256843522072064, "grad_norm": 1.963857650756836, "learning_rate": 9.013857375911793e-06, "loss": 0.998, "step": 6910 }, { "epoch": 0.1258662398370287, "grad_norm": 2.116774082183838, "learning_rate": 9.011982223555664e-06, "loss": 1.0398, "step": 6920 }, { "epoch": 0.12604812746685098, "grad_norm": 2.058079719543457, "learning_rate": 9.010107071199536e-06, "loss": 1.0125, "step": 6930 }, { "epoch": 0.12623001509667328, "grad_norm": 2.3310933113098145, "learning_rate": 9.008231918843407e-06, "loss": 1.0, "step": 6940 }, { "epoch": 0.12641190272649558, "grad_norm": 1.8501083850860596, "learning_rate": 9.006356766487277e-06, "loss": 1.0221, "step": 6950 }, { "epoch": 0.12659379035631788, "grad_norm": 1.6123971939086914, "learning_rate": 9.004481614131148e-06, "loss": 1.0182, "step": 6960 }, { "epoch": 0.12677567798614017, "grad_norm": 1.9837661981582642, "learning_rate": 9.00260646177502e-06, "loss": 1.0502, "step": 6970 }, { "epoch": 0.12695756561596247, "grad_norm": 1.6761548519134521, "learning_rate": 9.000731309418891e-06, "loss": 0.9868, "step": 6980 }, { "epoch": 0.12713945324578477, "grad_norm": 1.9050461053848267, "learning_rate": 8.998856157062762e-06, "loss": 1.0497, "step": 6990 }, { "epoch": 0.12732134087560704, "grad_norm": 1.9055516719818115, "learning_rate": 8.996981004706634e-06, "loss": 1.004, "step": 7000 }, { "epoch": 0.12750322850542933, "grad_norm": 1.906802773475647, "learning_rate": 8.995105852350505e-06, "loss": 0.9695, "step": 7010 }, { "epoch": 0.12768511613525163, "grad_norm": 2.1746010780334473, "learning_rate": 8.993230699994375e-06, "loss": 1.014, "step": 7020 }, { "epoch": 0.12786700376507393, "grad_norm": 1.9818288087844849, "learning_rate": 8.991355547638246e-06, "loss": 0.9747, "step": 7030 }, { "epoch": 0.12804889139489622, "grad_norm": 1.7605993747711182, "learning_rate": 8.989480395282117e-06, "loss": 0.9885, "step": 7040 }, { "epoch": 0.12823077902471852, "grad_norm": 2.1218433380126953, "learning_rate": 8.987605242925989e-06, "loss": 1.0498, "step": 7050 }, { "epoch": 0.12841266665454082, "grad_norm": 2.037904739379883, "learning_rate": 8.98573009056986e-06, "loss": 1.0824, "step": 7060 }, { "epoch": 0.12859455428436312, "grad_norm": 1.9100219011306763, "learning_rate": 8.983854938213731e-06, "loss": 0.9961, "step": 7070 }, { "epoch": 0.1287764419141854, "grad_norm": 1.7678216695785522, "learning_rate": 8.981979785857603e-06, "loss": 0.9855, "step": 7080 }, { "epoch": 0.1289583295440077, "grad_norm": 2.0349838733673096, "learning_rate": 8.980104633501472e-06, "loss": 1.0636, "step": 7090 }, { "epoch": 0.12914021717383, "grad_norm": 1.7976396083831787, "learning_rate": 8.978229481145344e-06, "loss": 0.9919, "step": 7100 }, { "epoch": 0.1293221048036523, "grad_norm": 1.9442434310913086, "learning_rate": 8.976354328789215e-06, "loss": 1.0486, "step": 7110 }, { "epoch": 0.1295039924334746, "grad_norm": 1.8663753271102905, "learning_rate": 8.974479176433086e-06, "loss": 1.0214, "step": 7120 }, { "epoch": 0.1296858800632969, "grad_norm": 2.0721962451934814, "learning_rate": 8.972604024076958e-06, "loss": 0.9913, "step": 7130 }, { "epoch": 0.1298677676931192, "grad_norm": 1.781520128250122, "learning_rate": 8.970728871720829e-06, "loss": 0.966, "step": 7140 }, { "epoch": 0.1300496553229415, "grad_norm": 1.567594289779663, "learning_rate": 8.9688537193647e-06, "loss": 1.0437, "step": 7150 }, { "epoch": 0.1302315429527638, "grad_norm": 1.8216781616210938, "learning_rate": 8.96697856700857e-06, "loss": 1.0339, "step": 7160 }, { "epoch": 0.1304134305825861, "grad_norm": 2.1657092571258545, "learning_rate": 8.965103414652441e-06, "loss": 1.0726, "step": 7170 }, { "epoch": 0.13059531821240838, "grad_norm": 2.1490163803100586, "learning_rate": 8.963228262296311e-06, "loss": 0.993, "step": 7180 }, { "epoch": 0.13077720584223068, "grad_norm": 1.916015386581421, "learning_rate": 8.961353109940182e-06, "loss": 1.0231, "step": 7190 }, { "epoch": 0.13095909347205298, "grad_norm": 2.0953304767608643, "learning_rate": 8.959477957584054e-06, "loss": 0.9761, "step": 7200 }, { "epoch": 0.13114098110187525, "grad_norm": 2.03989839553833, "learning_rate": 8.957602805227925e-06, "loss": 0.9952, "step": 7210 }, { "epoch": 0.13132286873169755, "grad_norm": 1.8135217428207397, "learning_rate": 8.955727652871796e-06, "loss": 0.9984, "step": 7220 }, { "epoch": 0.13150475636151984, "grad_norm": 1.6570298671722412, "learning_rate": 8.953852500515668e-06, "loss": 1.0285, "step": 7230 }, { "epoch": 0.13168664399134214, "grad_norm": 1.751917839050293, "learning_rate": 8.951977348159539e-06, "loss": 1.0529, "step": 7240 }, { "epoch": 0.13186853162116444, "grad_norm": 3.489983320236206, "learning_rate": 8.950102195803409e-06, "loss": 1.0006, "step": 7250 }, { "epoch": 0.13205041925098673, "grad_norm": 1.745742678642273, "learning_rate": 8.94822704344728e-06, "loss": 0.9957, "step": 7260 }, { "epoch": 0.13223230688080903, "grad_norm": 1.7659106254577637, "learning_rate": 8.946351891091151e-06, "loss": 0.9831, "step": 7270 }, { "epoch": 0.13241419451063133, "grad_norm": 1.6759850978851318, "learning_rate": 8.944476738735023e-06, "loss": 0.9888, "step": 7280 }, { "epoch": 0.13259608214045362, "grad_norm": 1.9594956636428833, "learning_rate": 8.942601586378894e-06, "loss": 1.0053, "step": 7290 }, { "epoch": 0.13277796977027592, "grad_norm": 2.049525022506714, "learning_rate": 8.940726434022765e-06, "loss": 1.0189, "step": 7300 }, { "epoch": 0.13295985740009822, "grad_norm": 1.91941237449646, "learning_rate": 8.938851281666637e-06, "loss": 1.0237, "step": 7310 }, { "epoch": 0.13314174502992052, "grad_norm": 1.6621403694152832, "learning_rate": 8.936976129310506e-06, "loss": 0.9662, "step": 7320 }, { "epoch": 0.1333236326597428, "grad_norm": 1.55476713180542, "learning_rate": 8.935100976954378e-06, "loss": 1.0159, "step": 7330 }, { "epoch": 0.1335055202895651, "grad_norm": 1.9162460565567017, "learning_rate": 8.93322582459825e-06, "loss": 0.9704, "step": 7340 }, { "epoch": 0.1336874079193874, "grad_norm": 1.7184839248657227, "learning_rate": 8.93135067224212e-06, "loss": 0.9897, "step": 7350 }, { "epoch": 0.1338692955492097, "grad_norm": 1.962281584739685, "learning_rate": 8.929475519885992e-06, "loss": 1.0423, "step": 7360 }, { "epoch": 0.134051183179032, "grad_norm": 1.8257099390029907, "learning_rate": 8.927600367529863e-06, "loss": 1.0324, "step": 7370 }, { "epoch": 0.1342330708088543, "grad_norm": 2.1000680923461914, "learning_rate": 8.925725215173735e-06, "loss": 1.0486, "step": 7380 }, { "epoch": 0.1344149584386766, "grad_norm": 1.854967713356018, "learning_rate": 8.923850062817604e-06, "loss": 0.9881, "step": 7390 }, { "epoch": 0.1345968460684989, "grad_norm": 1.9156250953674316, "learning_rate": 8.921974910461476e-06, "loss": 0.9847, "step": 7400 }, { "epoch": 0.1347787336983212, "grad_norm": 2.0147359371185303, "learning_rate": 8.920099758105347e-06, "loss": 1.0426, "step": 7410 }, { "epoch": 0.13496062132814346, "grad_norm": 1.8597078323364258, "learning_rate": 8.918224605749218e-06, "loss": 1.0227, "step": 7420 }, { "epoch": 0.13514250895796576, "grad_norm": 1.586326241493225, "learning_rate": 8.91634945339309e-06, "loss": 1.0294, "step": 7430 }, { "epoch": 0.13532439658778805, "grad_norm": 2.0155322551727295, "learning_rate": 8.91447430103696e-06, "loss": 0.9588, "step": 7440 }, { "epoch": 0.13550628421761035, "grad_norm": 1.8451392650604248, "learning_rate": 8.91259914868083e-06, "loss": 0.9646, "step": 7450 }, { "epoch": 0.13568817184743265, "grad_norm": 1.9579941034317017, "learning_rate": 8.910723996324702e-06, "loss": 1.0131, "step": 7460 }, { "epoch": 0.13587005947725495, "grad_norm": 2.080066204071045, "learning_rate": 8.908848843968573e-06, "loss": 1.0164, "step": 7470 }, { "epoch": 0.13605194710707724, "grad_norm": 1.8979116678237915, "learning_rate": 8.906973691612445e-06, "loss": 0.9967, "step": 7480 }, { "epoch": 0.13623383473689954, "grad_norm": 1.9077279567718506, "learning_rate": 8.905098539256314e-06, "loss": 0.9954, "step": 7490 }, { "epoch": 0.13641572236672184, "grad_norm": 2.124659538269043, "learning_rate": 8.903223386900186e-06, "loss": 1.0232, "step": 7500 }, { "epoch": 0.13659760999654413, "grad_norm": 1.8870562314987183, "learning_rate": 8.901348234544057e-06, "loss": 1.0225, "step": 7510 }, { "epoch": 0.13677949762636643, "grad_norm": 1.936928153038025, "learning_rate": 8.899473082187928e-06, "loss": 1.0008, "step": 7520 }, { "epoch": 0.13696138525618873, "grad_norm": 1.8801982402801514, "learning_rate": 8.8975979298318e-06, "loss": 1.0099, "step": 7530 }, { "epoch": 0.13714327288601103, "grad_norm": 2.1173958778381348, "learning_rate": 8.895722777475671e-06, "loss": 1.0092, "step": 7540 }, { "epoch": 0.13732516051583332, "grad_norm": 1.8792171478271484, "learning_rate": 8.893847625119542e-06, "loss": 0.9627, "step": 7550 }, { "epoch": 0.13750704814565562, "grad_norm": 1.7120834589004517, "learning_rate": 8.891972472763412e-06, "loss": 1.0037, "step": 7560 }, { "epoch": 0.13768893577547792, "grad_norm": 2.6391725540161133, "learning_rate": 8.890097320407283e-06, "loss": 1.0506, "step": 7570 }, { "epoch": 0.1378708234053002, "grad_norm": 2.0720865726470947, "learning_rate": 8.888222168051155e-06, "loss": 1.0074, "step": 7580 }, { "epoch": 0.1380527110351225, "grad_norm": 1.9443409442901611, "learning_rate": 8.886347015695026e-06, "loss": 1.0408, "step": 7590 }, { "epoch": 0.1382345986649448, "grad_norm": 1.8200019598007202, "learning_rate": 8.884471863338897e-06, "loss": 0.9927, "step": 7600 }, { "epoch": 0.1384164862947671, "grad_norm": 2.059541940689087, "learning_rate": 8.882596710982769e-06, "loss": 1.0067, "step": 7610 }, { "epoch": 0.1385983739245894, "grad_norm": 2.076218366622925, "learning_rate": 8.88072155862664e-06, "loss": 1.0204, "step": 7620 }, { "epoch": 0.13878026155441167, "grad_norm": 2.270941734313965, "learning_rate": 8.87884640627051e-06, "loss": 1.0297, "step": 7630 }, { "epoch": 0.13896214918423397, "grad_norm": 1.76186203956604, "learning_rate": 8.876971253914381e-06, "loss": 0.9884, "step": 7640 }, { "epoch": 0.13914403681405627, "grad_norm": 1.8881423473358154, "learning_rate": 8.875096101558252e-06, "loss": 0.9602, "step": 7650 }, { "epoch": 0.13932592444387856, "grad_norm": 1.5613855123519897, "learning_rate": 8.873220949202124e-06, "loss": 1.02, "step": 7660 }, { "epoch": 0.13950781207370086, "grad_norm": 2.10691237449646, "learning_rate": 8.871345796845995e-06, "loss": 1.0181, "step": 7670 }, { "epoch": 0.13968969970352316, "grad_norm": 1.7975183725357056, "learning_rate": 8.869470644489866e-06, "loss": 1.0258, "step": 7680 }, { "epoch": 0.13987158733334545, "grad_norm": 1.7161943912506104, "learning_rate": 8.867595492133738e-06, "loss": 1.0149, "step": 7690 }, { "epoch": 0.14005347496316775, "grad_norm": 2.2564918994903564, "learning_rate": 8.865720339777607e-06, "loss": 1.0358, "step": 7700 }, { "epoch": 0.14023536259299005, "grad_norm": 1.8036478757858276, "learning_rate": 8.863845187421479e-06, "loss": 0.9986, "step": 7710 }, { "epoch": 0.14041725022281235, "grad_norm": 2.00091814994812, "learning_rate": 8.861970035065348e-06, "loss": 1.0422, "step": 7720 }, { "epoch": 0.14059913785263464, "grad_norm": 2.2291436195373535, "learning_rate": 8.86009488270922e-06, "loss": 1.0366, "step": 7730 }, { "epoch": 0.14078102548245694, "grad_norm": 1.8454982042312622, "learning_rate": 8.858219730353091e-06, "loss": 0.9721, "step": 7740 }, { "epoch": 0.14096291311227924, "grad_norm": 1.9020845890045166, "learning_rate": 8.856344577996962e-06, "loss": 1.002, "step": 7750 }, { "epoch": 0.14114480074210153, "grad_norm": 1.965834617614746, "learning_rate": 8.854469425640834e-06, "loss": 1.03, "step": 7760 }, { "epoch": 0.14132668837192383, "grad_norm": 1.97315514087677, "learning_rate": 8.852594273284705e-06, "loss": 0.9801, "step": 7770 }, { "epoch": 0.14150857600174613, "grad_norm": 1.6884329319000244, "learning_rate": 8.850719120928577e-06, "loss": 1.0064, "step": 7780 }, { "epoch": 0.14169046363156843, "grad_norm": 2.116067409515381, "learning_rate": 8.848843968572446e-06, "loss": 1.0158, "step": 7790 }, { "epoch": 0.14187235126139072, "grad_norm": 1.5746667385101318, "learning_rate": 8.846968816216318e-06, "loss": 1.0099, "step": 7800 }, { "epoch": 0.14205423889121302, "grad_norm": 1.8346058130264282, "learning_rate": 8.845093663860189e-06, "loss": 1.0229, "step": 7810 }, { "epoch": 0.14223612652103532, "grad_norm": 1.9840772151947021, "learning_rate": 8.84321851150406e-06, "loss": 0.9771, "step": 7820 }, { "epoch": 0.1424180141508576, "grad_norm": 2.308147668838501, "learning_rate": 8.841343359147932e-06, "loss": 1.0386, "step": 7830 }, { "epoch": 0.14259990178067988, "grad_norm": 1.8353471755981445, "learning_rate": 8.839468206791803e-06, "loss": 0.9621, "step": 7840 }, { "epoch": 0.14278178941050218, "grad_norm": 1.9703474044799805, "learning_rate": 8.837593054435674e-06, "loss": 0.9959, "step": 7850 }, { "epoch": 0.14296367704032448, "grad_norm": 1.8827356100082397, "learning_rate": 8.835717902079544e-06, "loss": 0.9934, "step": 7860 }, { "epoch": 0.14314556467014677, "grad_norm": 1.8526721000671387, "learning_rate": 8.833842749723415e-06, "loss": 1.0452, "step": 7870 }, { "epoch": 0.14332745229996907, "grad_norm": 2.299969434738159, "learning_rate": 8.831967597367287e-06, "loss": 1.0218, "step": 7880 }, { "epoch": 0.14350933992979137, "grad_norm": 2.011289119720459, "learning_rate": 8.830092445011158e-06, "loss": 1.0294, "step": 7890 }, { "epoch": 0.14369122755961367, "grad_norm": 1.8865152597427368, "learning_rate": 8.82821729265503e-06, "loss": 1.0173, "step": 7900 }, { "epoch": 0.14387311518943596, "grad_norm": 2.0604190826416016, "learning_rate": 8.8263421402989e-06, "loss": 0.9885, "step": 7910 }, { "epoch": 0.14405500281925826, "grad_norm": 2.0624663829803467, "learning_rate": 8.824466987942772e-06, "loss": 1.0164, "step": 7920 }, { "epoch": 0.14423689044908056, "grad_norm": 2.0279347896575928, "learning_rate": 8.822591835586642e-06, "loss": 1.016, "step": 7930 }, { "epoch": 0.14441877807890285, "grad_norm": 2.024566888809204, "learning_rate": 8.820716683230513e-06, "loss": 1.0264, "step": 7940 }, { "epoch": 0.14460066570872515, "grad_norm": 1.8158067464828491, "learning_rate": 8.818841530874384e-06, "loss": 0.9506, "step": 7950 }, { "epoch": 0.14478255333854745, "grad_norm": 1.6889722347259521, "learning_rate": 8.816966378518256e-06, "loss": 1.021, "step": 7960 }, { "epoch": 0.14496444096836975, "grad_norm": 2.0333611965179443, "learning_rate": 8.815091226162127e-06, "loss": 0.9832, "step": 7970 }, { "epoch": 0.14514632859819204, "grad_norm": 2.0525667667388916, "learning_rate": 8.813216073805998e-06, "loss": 1.039, "step": 7980 }, { "epoch": 0.14532821622801434, "grad_norm": 2.073119640350342, "learning_rate": 8.811340921449868e-06, "loss": 1.0089, "step": 7990 }, { "epoch": 0.14551010385783664, "grad_norm": 2.016915798187256, "learning_rate": 8.80946576909374e-06, "loss": 0.9786, "step": 8000 }, { "epoch": 0.14569199148765893, "grad_norm": 1.6649905443191528, "learning_rate": 8.80759061673761e-06, "loss": 1.002, "step": 8010 }, { "epoch": 0.14587387911748123, "grad_norm": 1.9489091634750366, "learning_rate": 8.805715464381482e-06, "loss": 0.9948, "step": 8020 }, { "epoch": 0.14605576674730353, "grad_norm": 1.8132866621017456, "learning_rate": 8.803840312025352e-06, "loss": 1.0448, "step": 8030 }, { "epoch": 0.14623765437712583, "grad_norm": 2.077514410018921, "learning_rate": 8.801965159669223e-06, "loss": 1.0106, "step": 8040 }, { "epoch": 0.1464195420069481, "grad_norm": 1.9303265810012817, "learning_rate": 8.800090007313094e-06, "loss": 1.0233, "step": 8050 }, { "epoch": 0.1466014296367704, "grad_norm": 2.12864089012146, "learning_rate": 8.798214854956966e-06, "loss": 1.0168, "step": 8060 }, { "epoch": 0.1467833172665927, "grad_norm": 1.9853078126907349, "learning_rate": 8.796339702600837e-06, "loss": 0.9921, "step": 8070 }, { "epoch": 0.14696520489641499, "grad_norm": 1.7428709268569946, "learning_rate": 8.794464550244708e-06, "loss": 0.9489, "step": 8080 }, { "epoch": 0.14714709252623728, "grad_norm": 2.0870895385742188, "learning_rate": 8.79258939788858e-06, "loss": 1.007, "step": 8090 }, { "epoch": 0.14732898015605958, "grad_norm": 2.0257561206817627, "learning_rate": 8.79071424553245e-06, "loss": 1.0683, "step": 8100 }, { "epoch": 0.14751086778588188, "grad_norm": 1.749794840812683, "learning_rate": 8.78883909317632e-06, "loss": 0.9352, "step": 8110 }, { "epoch": 0.14769275541570417, "grad_norm": 1.8992363214492798, "learning_rate": 8.786963940820192e-06, "loss": 1.001, "step": 8120 }, { "epoch": 0.14787464304552647, "grad_norm": 1.5930064916610718, "learning_rate": 8.785088788464063e-06, "loss": 0.9871, "step": 8130 }, { "epoch": 0.14805653067534877, "grad_norm": 1.907882809638977, "learning_rate": 8.783213636107935e-06, "loss": 1.0102, "step": 8140 }, { "epoch": 0.14823841830517107, "grad_norm": 1.9129756689071655, "learning_rate": 8.781338483751806e-06, "loss": 1.0017, "step": 8150 }, { "epoch": 0.14842030593499336, "grad_norm": 1.71849524974823, "learning_rate": 8.779463331395678e-06, "loss": 1.0046, "step": 8160 }, { "epoch": 0.14860219356481566, "grad_norm": 2.0375993251800537, "learning_rate": 8.777588179039547e-06, "loss": 1.0145, "step": 8170 }, { "epoch": 0.14878408119463796, "grad_norm": 1.857675313949585, "learning_rate": 8.775713026683419e-06, "loss": 0.9717, "step": 8180 }, { "epoch": 0.14896596882446025, "grad_norm": 1.8244816064834595, "learning_rate": 8.77383787432729e-06, "loss": 1.0667, "step": 8190 }, { "epoch": 0.14914785645428255, "grad_norm": 2.0875372886657715, "learning_rate": 8.771962721971161e-06, "loss": 0.9892, "step": 8200 }, { "epoch": 0.14932974408410485, "grad_norm": 2.1089751720428467, "learning_rate": 8.770087569615033e-06, "loss": 1.0347, "step": 8210 }, { "epoch": 0.14951163171392715, "grad_norm": 2.282026529312134, "learning_rate": 8.768212417258904e-06, "loss": 0.9797, "step": 8220 }, { "epoch": 0.14969351934374944, "grad_norm": 1.9321129322052002, "learning_rate": 8.766337264902775e-06, "loss": 1.0035, "step": 8230 }, { "epoch": 0.14987540697357174, "grad_norm": 1.809800624847412, "learning_rate": 8.764462112546645e-06, "loss": 0.9518, "step": 8240 }, { "epoch": 0.15005729460339404, "grad_norm": 2.2866039276123047, "learning_rate": 8.762586960190516e-06, "loss": 1.0453, "step": 8250 }, { "epoch": 0.1502391822332163, "grad_norm": 1.9624278545379639, "learning_rate": 8.760711807834388e-06, "loss": 0.9807, "step": 8260 }, { "epoch": 0.1504210698630386, "grad_norm": 1.98380446434021, "learning_rate": 8.758836655478257e-06, "loss": 1.0119, "step": 8270 }, { "epoch": 0.1506029574928609, "grad_norm": 1.8660717010498047, "learning_rate": 8.756961503122129e-06, "loss": 0.9823, "step": 8280 }, { "epoch": 0.1507848451226832, "grad_norm": 1.8808702230453491, "learning_rate": 8.755086350766e-06, "loss": 0.9973, "step": 8290 }, { "epoch": 0.1509667327525055, "grad_norm": 1.707129955291748, "learning_rate": 8.753211198409871e-06, "loss": 0.9717, "step": 8300 }, { "epoch": 0.1511486203823278, "grad_norm": 1.975527048110962, "learning_rate": 8.751336046053743e-06, "loss": 0.975, "step": 8310 }, { "epoch": 0.1513305080121501, "grad_norm": 1.6645090579986572, "learning_rate": 8.749460893697614e-06, "loss": 0.9768, "step": 8320 }, { "epoch": 0.15151239564197239, "grad_norm": 2.120394706726074, "learning_rate": 8.747585741341484e-06, "loss": 1.0357, "step": 8330 }, { "epoch": 0.15169428327179468, "grad_norm": 1.6850720643997192, "learning_rate": 8.745710588985355e-06, "loss": 0.9342, "step": 8340 }, { "epoch": 0.15187617090161698, "grad_norm": 1.6781442165374756, "learning_rate": 8.743835436629226e-06, "loss": 1.0104, "step": 8350 }, { "epoch": 0.15205805853143928, "grad_norm": 1.7403934001922607, "learning_rate": 8.741960284273098e-06, "loss": 1.0197, "step": 8360 }, { "epoch": 0.15223994616126157, "grad_norm": 2.053558349609375, "learning_rate": 8.740085131916969e-06, "loss": 1.0266, "step": 8370 }, { "epoch": 0.15242183379108387, "grad_norm": 1.7690942287445068, "learning_rate": 8.73820997956084e-06, "loss": 1.0006, "step": 8380 }, { "epoch": 0.15260372142090617, "grad_norm": 1.8674019575119019, "learning_rate": 8.736334827204712e-06, "loss": 1.0298, "step": 8390 }, { "epoch": 0.15278560905072847, "grad_norm": 1.860883355140686, "learning_rate": 8.734459674848581e-06, "loss": 1.049, "step": 8400 }, { "epoch": 0.15296749668055076, "grad_norm": 1.9633513689041138, "learning_rate": 8.732584522492453e-06, "loss": 1.0514, "step": 8410 }, { "epoch": 0.15314938431037306, "grad_norm": 1.8330522775650024, "learning_rate": 8.730709370136324e-06, "loss": 1.0062, "step": 8420 }, { "epoch": 0.15333127194019536, "grad_norm": 1.8137083053588867, "learning_rate": 8.728834217780195e-06, "loss": 0.9455, "step": 8430 }, { "epoch": 0.15351315957001765, "grad_norm": 1.7965846061706543, "learning_rate": 8.726959065424067e-06, "loss": 0.9433, "step": 8440 }, { "epoch": 0.15369504719983995, "grad_norm": 1.6799397468566895, "learning_rate": 8.725083913067938e-06, "loss": 0.9653, "step": 8450 }, { "epoch": 0.15387693482966222, "grad_norm": 1.786130428314209, "learning_rate": 8.72320876071181e-06, "loss": 1.0111, "step": 8460 }, { "epoch": 0.15405882245948452, "grad_norm": 1.745202898979187, "learning_rate": 8.721333608355679e-06, "loss": 1.0076, "step": 8470 }, { "epoch": 0.15424071008930682, "grad_norm": 1.4879215955734253, "learning_rate": 8.71945845599955e-06, "loss": 0.9721, "step": 8480 }, { "epoch": 0.1544225977191291, "grad_norm": 2.06437087059021, "learning_rate": 8.717583303643422e-06, "loss": 0.9899, "step": 8490 }, { "epoch": 0.1546044853489514, "grad_norm": 2.025437593460083, "learning_rate": 8.715708151287293e-06, "loss": 1.0457, "step": 8500 }, { "epoch": 0.1547863729787737, "grad_norm": 1.755220651626587, "learning_rate": 8.713832998931164e-06, "loss": 0.9713, "step": 8510 }, { "epoch": 0.154968260608596, "grad_norm": 1.7720798254013062, "learning_rate": 8.711957846575036e-06, "loss": 1.0669, "step": 8520 }, { "epoch": 0.1551501482384183, "grad_norm": 1.9302481412887573, "learning_rate": 8.710082694218907e-06, "loss": 0.9652, "step": 8530 }, { "epoch": 0.1553320358682406, "grad_norm": 1.7999954223632812, "learning_rate": 8.708207541862777e-06, "loss": 0.9918, "step": 8540 }, { "epoch": 0.1555139234980629, "grad_norm": 1.638383150100708, "learning_rate": 8.706332389506648e-06, "loss": 0.9535, "step": 8550 }, { "epoch": 0.1556958111278852, "grad_norm": 1.9563740491867065, "learning_rate": 8.70445723715052e-06, "loss": 1.0048, "step": 8560 }, { "epoch": 0.1558776987577075, "grad_norm": 1.8302547931671143, "learning_rate": 8.702582084794389e-06, "loss": 0.9954, "step": 8570 }, { "epoch": 0.1560595863875298, "grad_norm": 1.5323671102523804, "learning_rate": 8.70070693243826e-06, "loss": 0.9425, "step": 8580 }, { "epoch": 0.15624147401735208, "grad_norm": 1.8860400915145874, "learning_rate": 8.698831780082132e-06, "loss": 1.0565, "step": 8590 }, { "epoch": 0.15642336164717438, "grad_norm": 1.6690120697021484, "learning_rate": 8.696956627726003e-06, "loss": 1.033, "step": 8600 }, { "epoch": 0.15660524927699668, "grad_norm": 1.8538683652877808, "learning_rate": 8.695081475369875e-06, "loss": 0.9713, "step": 8610 }, { "epoch": 0.15678713690681897, "grad_norm": 1.8018652200698853, "learning_rate": 8.693206323013746e-06, "loss": 0.9768, "step": 8620 }, { "epoch": 0.15696902453664127, "grad_norm": 1.9127740859985352, "learning_rate": 8.691331170657617e-06, "loss": 1.0038, "step": 8630 }, { "epoch": 0.15715091216646357, "grad_norm": 1.982724666595459, "learning_rate": 8.689456018301487e-06, "loss": 0.998, "step": 8640 }, { "epoch": 0.15733279979628587, "grad_norm": 1.8344866037368774, "learning_rate": 8.687580865945358e-06, "loss": 0.9956, "step": 8650 }, { "epoch": 0.15751468742610816, "grad_norm": 1.8258346319198608, "learning_rate": 8.68570571358923e-06, "loss": 1.0099, "step": 8660 }, { "epoch": 0.15769657505593043, "grad_norm": 1.7866837978363037, "learning_rate": 8.683830561233101e-06, "loss": 1.0176, "step": 8670 }, { "epoch": 0.15787846268575273, "grad_norm": 1.8536311388015747, "learning_rate": 8.681955408876972e-06, "loss": 1.0355, "step": 8680 }, { "epoch": 0.15806035031557503, "grad_norm": 1.8701895475387573, "learning_rate": 8.680080256520844e-06, "loss": 1.0002, "step": 8690 }, { "epoch": 0.15824223794539732, "grad_norm": 2.0518224239349365, "learning_rate": 8.678205104164715e-06, "loss": 0.9419, "step": 8700 }, { "epoch": 0.15842412557521962, "grad_norm": 1.6014643907546997, "learning_rate": 8.676329951808585e-06, "loss": 1.0484, "step": 8710 }, { "epoch": 0.15860601320504192, "grad_norm": 1.7277088165283203, "learning_rate": 8.674454799452456e-06, "loss": 1.052, "step": 8720 }, { "epoch": 0.15878790083486422, "grad_norm": 1.9712945222854614, "learning_rate": 8.672579647096327e-06, "loss": 0.9677, "step": 8730 }, { "epoch": 0.1589697884646865, "grad_norm": 2.0613019466400146, "learning_rate": 8.670704494740199e-06, "loss": 1.0026, "step": 8740 }, { "epoch": 0.1591516760945088, "grad_norm": 1.8830273151397705, "learning_rate": 8.66882934238407e-06, "loss": 1.0437, "step": 8750 }, { "epoch": 0.1593335637243311, "grad_norm": 1.7918322086334229, "learning_rate": 8.666954190027941e-06, "loss": 1.0003, "step": 8760 }, { "epoch": 0.1595154513541534, "grad_norm": 1.8037041425704956, "learning_rate": 8.665079037671813e-06, "loss": 1.0145, "step": 8770 }, { "epoch": 0.1596973389839757, "grad_norm": 1.585322618484497, "learning_rate": 8.663203885315682e-06, "loss": 1.035, "step": 8780 }, { "epoch": 0.159879226613798, "grad_norm": 2.101459264755249, "learning_rate": 8.661328732959554e-06, "loss": 0.9958, "step": 8790 }, { "epoch": 0.1600611142436203, "grad_norm": 1.6609373092651367, "learning_rate": 8.659453580603425e-06, "loss": 0.9755, "step": 8800 }, { "epoch": 0.1602430018734426, "grad_norm": 1.730338454246521, "learning_rate": 8.657578428247296e-06, "loss": 0.983, "step": 8810 }, { "epoch": 0.1604248895032649, "grad_norm": 1.832344889640808, "learning_rate": 8.655703275891166e-06, "loss": 1.0666, "step": 8820 }, { "epoch": 0.1606067771330872, "grad_norm": 1.766767144203186, "learning_rate": 8.653828123535037e-06, "loss": 1.0116, "step": 8830 }, { "epoch": 0.16078866476290948, "grad_norm": 1.7701267004013062, "learning_rate": 8.651952971178909e-06, "loss": 0.9389, "step": 8840 }, { "epoch": 0.16097055239273178, "grad_norm": 2.0997347831726074, "learning_rate": 8.65007781882278e-06, "loss": 1.0254, "step": 8850 }, { "epoch": 0.16115244002255408, "grad_norm": 1.7084864377975464, "learning_rate": 8.648202666466651e-06, "loss": 0.9732, "step": 8860 }, { "epoch": 0.16133432765237637, "grad_norm": 2.0284552574157715, "learning_rate": 8.646327514110521e-06, "loss": 1.0583, "step": 8870 }, { "epoch": 0.16151621528219864, "grad_norm": 2.074686050415039, "learning_rate": 8.644452361754392e-06, "loss": 1.017, "step": 8880 }, { "epoch": 0.16169810291202094, "grad_norm": 1.8955098390579224, "learning_rate": 8.642577209398264e-06, "loss": 1.0342, "step": 8890 }, { "epoch": 0.16187999054184324, "grad_norm": 2.102757453918457, "learning_rate": 8.640702057042135e-06, "loss": 1.0242, "step": 8900 }, { "epoch": 0.16206187817166554, "grad_norm": 1.7492499351501465, "learning_rate": 8.638826904686006e-06, "loss": 0.9939, "step": 8910 }, { "epoch": 0.16224376580148783, "grad_norm": 1.6390219926834106, "learning_rate": 8.636951752329878e-06, "loss": 0.9915, "step": 8920 }, { "epoch": 0.16242565343131013, "grad_norm": 1.9659079313278198, "learning_rate": 8.635076599973749e-06, "loss": 0.9785, "step": 8930 }, { "epoch": 0.16260754106113243, "grad_norm": 1.8005989789962769, "learning_rate": 8.633201447617619e-06, "loss": 0.9924, "step": 8940 }, { "epoch": 0.16278942869095472, "grad_norm": 1.8950822353363037, "learning_rate": 8.63132629526149e-06, "loss": 0.9673, "step": 8950 }, { "epoch": 0.16297131632077702, "grad_norm": 1.9198206663131714, "learning_rate": 8.629451142905361e-06, "loss": 1.0405, "step": 8960 }, { "epoch": 0.16315320395059932, "grad_norm": 1.5994789600372314, "learning_rate": 8.627575990549233e-06, "loss": 0.9783, "step": 8970 }, { "epoch": 0.16333509158042162, "grad_norm": 1.845847249031067, "learning_rate": 8.625700838193104e-06, "loss": 1.0318, "step": 8980 }, { "epoch": 0.1635169792102439, "grad_norm": 1.768350601196289, "learning_rate": 8.623825685836976e-06, "loss": 1.0347, "step": 8990 }, { "epoch": 0.1636988668400662, "grad_norm": 1.9537410736083984, "learning_rate": 8.621950533480847e-06, "loss": 0.9809, "step": 9000 }, { "epoch": 0.1638807544698885, "grad_norm": 1.8067792654037476, "learning_rate": 8.620075381124717e-06, "loss": 0.93, "step": 9010 }, { "epoch": 0.1640626420997108, "grad_norm": 2.0004212856292725, "learning_rate": 8.618200228768588e-06, "loss": 0.9956, "step": 9020 }, { "epoch": 0.1642445297295331, "grad_norm": 1.8644200563430786, "learning_rate": 8.61632507641246e-06, "loss": 1.0111, "step": 9030 }, { "epoch": 0.1644264173593554, "grad_norm": 1.8750951290130615, "learning_rate": 8.61444992405633e-06, "loss": 0.9648, "step": 9040 }, { "epoch": 0.1646083049891777, "grad_norm": 1.9473575353622437, "learning_rate": 8.612574771700202e-06, "loss": 1.0158, "step": 9050 }, { "epoch": 0.164790192619, "grad_norm": 2.0611822605133057, "learning_rate": 8.610699619344073e-06, "loss": 1.0034, "step": 9060 }, { "epoch": 0.1649720802488223, "grad_norm": 1.9632188081741333, "learning_rate": 8.608824466987945e-06, "loss": 1.0311, "step": 9070 }, { "epoch": 0.1651539678786446, "grad_norm": 2.119523763656616, "learning_rate": 8.606949314631814e-06, "loss": 1.0269, "step": 9080 }, { "epoch": 0.16533585550846686, "grad_norm": 2.0425703525543213, "learning_rate": 8.605074162275686e-06, "loss": 1.0154, "step": 9090 }, { "epoch": 0.16551774313828915, "grad_norm": 1.7819550037384033, "learning_rate": 8.603199009919557e-06, "loss": 0.9666, "step": 9100 }, { "epoch": 0.16569963076811145, "grad_norm": 1.6669549942016602, "learning_rate": 8.601323857563427e-06, "loss": 1.0024, "step": 9110 }, { "epoch": 0.16588151839793375, "grad_norm": 1.7890113592147827, "learning_rate": 8.599448705207298e-06, "loss": 1.0276, "step": 9120 }, { "epoch": 0.16606340602775604, "grad_norm": 2.1156680583953857, "learning_rate": 8.59757355285117e-06, "loss": 0.9562, "step": 9130 }, { "epoch": 0.16624529365757834, "grad_norm": 1.6905243396759033, "learning_rate": 8.59569840049504e-06, "loss": 0.995, "step": 9140 }, { "epoch": 0.16642718128740064, "grad_norm": 1.7820976972579956, "learning_rate": 8.593823248138912e-06, "loss": 0.9241, "step": 9150 }, { "epoch": 0.16660906891722294, "grad_norm": 1.8553690910339355, "learning_rate": 8.591948095782783e-06, "loss": 0.9606, "step": 9160 }, { "epoch": 0.16679095654704523, "grad_norm": 2.028427839279175, "learning_rate": 8.590072943426655e-06, "loss": 1.0117, "step": 9170 }, { "epoch": 0.16697284417686753, "grad_norm": 1.8538340330123901, "learning_rate": 8.588197791070524e-06, "loss": 0.9966, "step": 9180 }, { "epoch": 0.16715473180668983, "grad_norm": 2.0141379833221436, "learning_rate": 8.586322638714396e-06, "loss": 1.0044, "step": 9190 }, { "epoch": 0.16733661943651212, "grad_norm": 2.0313010215759277, "learning_rate": 8.584447486358267e-06, "loss": 0.9715, "step": 9200 }, { "epoch": 0.16751850706633442, "grad_norm": 1.8195600509643555, "learning_rate": 8.582572334002138e-06, "loss": 1.0064, "step": 9210 }, { "epoch": 0.16770039469615672, "grad_norm": 2.041232109069824, "learning_rate": 8.58069718164601e-06, "loss": 0.9759, "step": 9220 }, { "epoch": 0.16788228232597902, "grad_norm": 1.768041729927063, "learning_rate": 8.578822029289881e-06, "loss": 0.9993, "step": 9230 }, { "epoch": 0.1680641699558013, "grad_norm": 1.6297677755355835, "learning_rate": 8.576946876933752e-06, "loss": 1.0012, "step": 9240 }, { "epoch": 0.1682460575856236, "grad_norm": 2.0828018188476562, "learning_rate": 8.575071724577622e-06, "loss": 0.9812, "step": 9250 }, { "epoch": 0.1684279452154459, "grad_norm": 2.0518126487731934, "learning_rate": 8.573196572221493e-06, "loss": 0.9987, "step": 9260 }, { "epoch": 0.1686098328452682, "grad_norm": 1.9933569431304932, "learning_rate": 8.571321419865365e-06, "loss": 0.9435, "step": 9270 }, { "epoch": 0.1687917204750905, "grad_norm": 1.7144036293029785, "learning_rate": 8.569446267509236e-06, "loss": 1.0098, "step": 9280 }, { "epoch": 0.1689736081049128, "grad_norm": 1.6733895540237427, "learning_rate": 8.567571115153107e-06, "loss": 0.9703, "step": 9290 }, { "epoch": 0.16915549573473507, "grad_norm": 2.3429300785064697, "learning_rate": 8.565695962796979e-06, "loss": 1.0126, "step": 9300 }, { "epoch": 0.16933738336455736, "grad_norm": 1.669849157333374, "learning_rate": 8.56382081044085e-06, "loss": 0.9645, "step": 9310 }, { "epoch": 0.16951927099437966, "grad_norm": 1.8966795206069946, "learning_rate": 8.56194565808472e-06, "loss": 1.0173, "step": 9320 }, { "epoch": 0.16970115862420196, "grad_norm": 1.7924338579177856, "learning_rate": 8.560070505728591e-06, "loss": 1.026, "step": 9330 }, { "epoch": 0.16988304625402426, "grad_norm": 2.233508825302124, "learning_rate": 8.558195353372462e-06, "loss": 1.0208, "step": 9340 }, { "epoch": 0.17006493388384655, "grad_norm": 1.6625133752822876, "learning_rate": 8.556320201016334e-06, "loss": 0.9707, "step": 9350 }, { "epoch": 0.17024682151366885, "grad_norm": 2.0872902870178223, "learning_rate": 8.554445048660205e-06, "loss": 1.0204, "step": 9360 }, { "epoch": 0.17042870914349115, "grad_norm": 1.6963367462158203, "learning_rate": 8.552569896304075e-06, "loss": 0.9313, "step": 9370 }, { "epoch": 0.17061059677331344, "grad_norm": 1.8073759078979492, "learning_rate": 8.550694743947946e-06, "loss": 1.0063, "step": 9380 }, { "epoch": 0.17079248440313574, "grad_norm": 2.0576305389404297, "learning_rate": 8.548819591591817e-06, "loss": 1.0013, "step": 9390 }, { "epoch": 0.17097437203295804, "grad_norm": 2.037397623062134, "learning_rate": 8.546944439235689e-06, "loss": 1.0169, "step": 9400 }, { "epoch": 0.17115625966278034, "grad_norm": 1.9468375444412231, "learning_rate": 8.545069286879558e-06, "loss": 1.0003, "step": 9410 }, { "epoch": 0.17133814729260263, "grad_norm": 1.8905898332595825, "learning_rate": 8.54319413452343e-06, "loss": 0.994, "step": 9420 }, { "epoch": 0.17152003492242493, "grad_norm": 2.1806724071502686, "learning_rate": 8.541318982167301e-06, "loss": 1.0467, "step": 9430 }, { "epoch": 0.17170192255224723, "grad_norm": 1.76972496509552, "learning_rate": 8.539443829811173e-06, "loss": 0.9833, "step": 9440 }, { "epoch": 0.17188381018206952, "grad_norm": 1.5887510776519775, "learning_rate": 8.537568677455044e-06, "loss": 0.9894, "step": 9450 }, { "epoch": 0.17206569781189182, "grad_norm": 1.7436683177947998, "learning_rate": 8.535693525098915e-06, "loss": 1.0068, "step": 9460 }, { "epoch": 0.17224758544171412, "grad_norm": 2.060023069381714, "learning_rate": 8.533818372742787e-06, "loss": 0.9948, "step": 9470 }, { "epoch": 0.17242947307153642, "grad_norm": 1.6808582544326782, "learning_rate": 8.531943220386656e-06, "loss": 0.9363, "step": 9480 }, { "epoch": 0.1726113607013587, "grad_norm": 1.8470091819763184, "learning_rate": 8.530068068030528e-06, "loss": 1.0256, "step": 9490 }, { "epoch": 0.172793248331181, "grad_norm": 2.1948657035827637, "learning_rate": 8.528192915674399e-06, "loss": 0.998, "step": 9500 }, { "epoch": 0.17297513596100328, "grad_norm": 1.9377003908157349, "learning_rate": 8.52631776331827e-06, "loss": 0.9894, "step": 9510 }, { "epoch": 0.17315702359082558, "grad_norm": 1.965267300605774, "learning_rate": 8.524442610962142e-06, "loss": 1.0175, "step": 9520 }, { "epoch": 0.17333891122064787, "grad_norm": 2.137708902359009, "learning_rate": 8.522567458606013e-06, "loss": 1.0042, "step": 9530 }, { "epoch": 0.17352079885047017, "grad_norm": 1.9071402549743652, "learning_rate": 8.520692306249884e-06, "loss": 0.9868, "step": 9540 }, { "epoch": 0.17370268648029247, "grad_norm": 1.5734281539916992, "learning_rate": 8.518817153893754e-06, "loss": 0.9808, "step": 9550 }, { "epoch": 0.17388457411011476, "grad_norm": 1.4781156778335571, "learning_rate": 8.516942001537625e-06, "loss": 0.9326, "step": 9560 }, { "epoch": 0.17406646173993706, "grad_norm": 1.9321955442428589, "learning_rate": 8.515066849181497e-06, "loss": 0.977, "step": 9570 }, { "epoch": 0.17424834936975936, "grad_norm": 1.7477235794067383, "learning_rate": 8.513191696825368e-06, "loss": 0.9908, "step": 9580 }, { "epoch": 0.17443023699958166, "grad_norm": 1.616288423538208, "learning_rate": 8.51131654446924e-06, "loss": 0.9846, "step": 9590 }, { "epoch": 0.17461212462940395, "grad_norm": 2.1463751792907715, "learning_rate": 8.50944139211311e-06, "loss": 1.0027, "step": 9600 }, { "epoch": 0.17479401225922625, "grad_norm": 1.9579887390136719, "learning_rate": 8.507566239756982e-06, "loss": 0.9568, "step": 9610 }, { "epoch": 0.17497589988904855, "grad_norm": 1.7791880369186401, "learning_rate": 8.505691087400852e-06, "loss": 0.9594, "step": 9620 }, { "epoch": 0.17515778751887084, "grad_norm": 1.9479594230651855, "learning_rate": 8.503815935044723e-06, "loss": 0.9757, "step": 9630 }, { "epoch": 0.17533967514869314, "grad_norm": 2.1653013229370117, "learning_rate": 8.501940782688594e-06, "loss": 0.9942, "step": 9640 }, { "epoch": 0.17552156277851544, "grad_norm": 2.0509161949157715, "learning_rate": 8.500065630332464e-06, "loss": 0.9941, "step": 9650 }, { "epoch": 0.17570345040833774, "grad_norm": 1.9873849153518677, "learning_rate": 8.498190477976335e-06, "loss": 1.0089, "step": 9660 }, { "epoch": 0.17588533803816003, "grad_norm": 1.889290690422058, "learning_rate": 8.496315325620207e-06, "loss": 1.026, "step": 9670 }, { "epoch": 0.17606722566798233, "grad_norm": 1.8124101161956787, "learning_rate": 8.494440173264078e-06, "loss": 1.0029, "step": 9680 }, { "epoch": 0.17624911329780463, "grad_norm": 1.9056084156036377, "learning_rate": 8.49256502090795e-06, "loss": 1.0137, "step": 9690 }, { "epoch": 0.17643100092762692, "grad_norm": 1.981221318244934, "learning_rate": 8.49068986855182e-06, "loss": 1.0187, "step": 9700 }, { "epoch": 0.17661288855744922, "grad_norm": 1.7072160243988037, "learning_rate": 8.488814716195692e-06, "loss": 1.0471, "step": 9710 }, { "epoch": 0.1767947761872715, "grad_norm": 1.8729819059371948, "learning_rate": 8.486939563839562e-06, "loss": 0.9758, "step": 9720 }, { "epoch": 0.1769766638170938, "grad_norm": 1.8641771078109741, "learning_rate": 8.485064411483433e-06, "loss": 1.0249, "step": 9730 }, { "epoch": 0.17715855144691608, "grad_norm": 1.9458508491516113, "learning_rate": 8.483189259127304e-06, "loss": 1.0501, "step": 9740 }, { "epoch": 0.17734043907673838, "grad_norm": 1.7997204065322876, "learning_rate": 8.481314106771176e-06, "loss": 0.9872, "step": 9750 }, { "epoch": 0.17752232670656068, "grad_norm": 1.829720377922058, "learning_rate": 8.479438954415047e-06, "loss": 1.0305, "step": 9760 }, { "epoch": 0.17770421433638298, "grad_norm": 1.957440972328186, "learning_rate": 8.477563802058918e-06, "loss": 1.0025, "step": 9770 }, { "epoch": 0.17788610196620527, "grad_norm": 2.047772169113159, "learning_rate": 8.47568864970279e-06, "loss": 0.9468, "step": 9780 }, { "epoch": 0.17806798959602757, "grad_norm": 1.669567584991455, "learning_rate": 8.47381349734666e-06, "loss": 0.9358, "step": 9790 }, { "epoch": 0.17824987722584987, "grad_norm": 1.5368571281433105, "learning_rate": 8.47193834499053e-06, "loss": 0.9757, "step": 9800 }, { "epoch": 0.17843176485567216, "grad_norm": 1.9632482528686523, "learning_rate": 8.470063192634402e-06, "loss": 0.9919, "step": 9810 }, { "epoch": 0.17861365248549446, "grad_norm": 1.9858685731887817, "learning_rate": 8.468188040278274e-06, "loss": 1.0357, "step": 9820 }, { "epoch": 0.17879554011531676, "grad_norm": 1.7345324754714966, "learning_rate": 8.466312887922145e-06, "loss": 1.0658, "step": 9830 }, { "epoch": 0.17897742774513906, "grad_norm": 1.885174036026001, "learning_rate": 8.464437735566016e-06, "loss": 0.9674, "step": 9840 }, { "epoch": 0.17915931537496135, "grad_norm": 1.975999355316162, "learning_rate": 8.462562583209888e-06, "loss": 0.9367, "step": 9850 }, { "epoch": 0.17934120300478365, "grad_norm": 1.809954047203064, "learning_rate": 8.460687430853757e-06, "loss": 1.0279, "step": 9860 }, { "epoch": 0.17952309063460595, "grad_norm": 1.924049735069275, "learning_rate": 8.458812278497629e-06, "loss": 0.9925, "step": 9870 }, { "epoch": 0.17970497826442824, "grad_norm": 2.012298583984375, "learning_rate": 8.4569371261415e-06, "loss": 1.0173, "step": 9880 }, { "epoch": 0.17988686589425054, "grad_norm": 2.084808349609375, "learning_rate": 8.455061973785371e-06, "loss": 1.0512, "step": 9890 }, { "epoch": 0.18006875352407284, "grad_norm": 2.021421194076538, "learning_rate": 8.453186821429243e-06, "loss": 1.0075, "step": 9900 }, { "epoch": 0.18025064115389514, "grad_norm": 1.997519612312317, "learning_rate": 8.451311669073114e-06, "loss": 0.9348, "step": 9910 }, { "epoch": 0.18043252878371743, "grad_norm": 2.177257537841797, "learning_rate": 8.449436516716984e-06, "loss": 1.0284, "step": 9920 }, { "epoch": 0.1806144164135397, "grad_norm": 1.9272831678390503, "learning_rate": 8.447561364360855e-06, "loss": 1.0022, "step": 9930 }, { "epoch": 0.180796304043362, "grad_norm": 1.7378673553466797, "learning_rate": 8.445686212004726e-06, "loss": 0.9788, "step": 9940 }, { "epoch": 0.1809781916731843, "grad_norm": 1.8969510793685913, "learning_rate": 8.443811059648596e-06, "loss": 1.0251, "step": 9950 }, { "epoch": 0.1811600793030066, "grad_norm": 2.0862691402435303, "learning_rate": 8.441935907292467e-06, "loss": 1.019, "step": 9960 }, { "epoch": 0.1813419669328289, "grad_norm": 1.8014501333236694, "learning_rate": 8.440060754936339e-06, "loss": 0.9999, "step": 9970 }, { "epoch": 0.1815238545626512, "grad_norm": 1.7559843063354492, "learning_rate": 8.43818560258021e-06, "loss": 0.9672, "step": 9980 }, { "epoch": 0.18170574219247349, "grad_norm": 1.954372763633728, "learning_rate": 8.436310450224081e-06, "loss": 1.0019, "step": 9990 }, { "epoch": 0.18188762982229578, "grad_norm": 1.6844967603683472, "learning_rate": 8.434435297867953e-06, "loss": 1.0004, "step": 10000 }, { "epoch": 0.18206951745211808, "grad_norm": 2.08164381980896, "learning_rate": 8.432560145511824e-06, "loss": 1.0493, "step": 10010 }, { "epoch": 0.18225140508194038, "grad_norm": 1.7799490690231323, "learning_rate": 8.430684993155694e-06, "loss": 0.9585, "step": 10020 }, { "epoch": 0.18243329271176267, "grad_norm": 1.6546871662139893, "learning_rate": 8.428809840799565e-06, "loss": 0.9937, "step": 10030 }, { "epoch": 0.18261518034158497, "grad_norm": 2.026031255722046, "learning_rate": 8.426934688443436e-06, "loss": 1.024, "step": 10040 }, { "epoch": 0.18279706797140727, "grad_norm": 1.878083348274231, "learning_rate": 8.425059536087308e-06, "loss": 0.9927, "step": 10050 }, { "epoch": 0.18297895560122956, "grad_norm": 1.8991910219192505, "learning_rate": 8.423184383731179e-06, "loss": 0.9459, "step": 10060 }, { "epoch": 0.18316084323105186, "grad_norm": 1.8727904558181763, "learning_rate": 8.42130923137505e-06, "loss": 1.0278, "step": 10070 }, { "epoch": 0.18334273086087416, "grad_norm": 1.9268906116485596, "learning_rate": 8.419434079018922e-06, "loss": 0.9972, "step": 10080 }, { "epoch": 0.18352461849069646, "grad_norm": 1.521740198135376, "learning_rate": 8.417558926662791e-06, "loss": 0.933, "step": 10090 }, { "epoch": 0.18370650612051875, "grad_norm": 1.721168875694275, "learning_rate": 8.415683774306663e-06, "loss": 1.0042, "step": 10100 }, { "epoch": 0.18388839375034105, "grad_norm": 1.9219352006912231, "learning_rate": 8.413808621950534e-06, "loss": 1.0661, "step": 10110 }, { "epoch": 0.18407028138016335, "grad_norm": 1.7432963848114014, "learning_rate": 8.411933469594405e-06, "loss": 1.0086, "step": 10120 }, { "epoch": 0.18425216900998564, "grad_norm": 1.6798317432403564, "learning_rate": 8.410058317238277e-06, "loss": 1.0332, "step": 10130 }, { "epoch": 0.18443405663980791, "grad_norm": 1.93979012966156, "learning_rate": 8.408183164882148e-06, "loss": 1.0286, "step": 10140 }, { "epoch": 0.1846159442696302, "grad_norm": 1.811020016670227, "learning_rate": 8.40630801252602e-06, "loss": 0.9664, "step": 10150 }, { "epoch": 0.1847978318994525, "grad_norm": 1.7855523824691772, "learning_rate": 8.404432860169889e-06, "loss": 0.9597, "step": 10160 }, { "epoch": 0.1849797195292748, "grad_norm": 2.1909375190734863, "learning_rate": 8.40255770781376e-06, "loss": 1.0563, "step": 10170 }, { "epoch": 0.1851616071590971, "grad_norm": 1.7991702556610107, "learning_rate": 8.400682555457632e-06, "loss": 0.9911, "step": 10180 }, { "epoch": 0.1853434947889194, "grad_norm": 2.0792362689971924, "learning_rate": 8.398807403101503e-06, "loss": 1.0399, "step": 10190 }, { "epoch": 0.1855253824187417, "grad_norm": 1.9448586702346802, "learning_rate": 8.396932250745373e-06, "loss": 0.9906, "step": 10200 }, { "epoch": 0.185707270048564, "grad_norm": 1.737175703048706, "learning_rate": 8.395057098389244e-06, "loss": 0.9748, "step": 10210 }, { "epoch": 0.1858891576783863, "grad_norm": 1.900800108909607, "learning_rate": 8.393181946033115e-06, "loss": 0.9713, "step": 10220 }, { "epoch": 0.1860710453082086, "grad_norm": 1.9840604066848755, "learning_rate": 8.391306793676987e-06, "loss": 1.0309, "step": 10230 }, { "epoch": 0.18625293293803089, "grad_norm": 2.1194610595703125, "learning_rate": 8.389431641320858e-06, "loss": 0.9391, "step": 10240 }, { "epoch": 0.18643482056785318, "grad_norm": 1.7872644662857056, "learning_rate": 8.38755648896473e-06, "loss": 0.9896, "step": 10250 }, { "epoch": 0.18661670819767548, "grad_norm": 1.8703609704971313, "learning_rate": 8.3856813366086e-06, "loss": 1.0059, "step": 10260 }, { "epoch": 0.18679859582749778, "grad_norm": 1.8459398746490479, "learning_rate": 8.38380618425247e-06, "loss": 1.0329, "step": 10270 }, { "epoch": 0.18698048345732007, "grad_norm": 1.8908511400222778, "learning_rate": 8.381931031896342e-06, "loss": 1.0428, "step": 10280 }, { "epoch": 0.18716237108714237, "grad_norm": 1.6407402753829956, "learning_rate": 8.380055879540213e-06, "loss": 0.9386, "step": 10290 }, { "epoch": 0.18734425871696467, "grad_norm": 2.0332183837890625, "learning_rate": 8.378180727184085e-06, "loss": 0.969, "step": 10300 }, { "epoch": 0.18752614634678696, "grad_norm": 2.0101680755615234, "learning_rate": 8.376305574827956e-06, "loss": 0.9968, "step": 10310 }, { "epoch": 0.18770803397660926, "grad_norm": 2.0028209686279297, "learning_rate": 8.374430422471827e-06, "loss": 0.9692, "step": 10320 }, { "epoch": 0.18788992160643156, "grad_norm": 1.6362500190734863, "learning_rate": 8.372555270115697e-06, "loss": 0.9819, "step": 10330 }, { "epoch": 0.18807180923625383, "grad_norm": 2.009300708770752, "learning_rate": 8.370680117759568e-06, "loss": 1.0136, "step": 10340 }, { "epoch": 0.18825369686607613, "grad_norm": 1.5453765392303467, "learning_rate": 8.36880496540344e-06, "loss": 0.9342, "step": 10350 }, { "epoch": 0.18843558449589842, "grad_norm": 2.118241548538208, "learning_rate": 8.366929813047311e-06, "loss": 0.9872, "step": 10360 }, { "epoch": 0.18861747212572072, "grad_norm": 2.1445538997650146, "learning_rate": 8.365054660691182e-06, "loss": 0.9593, "step": 10370 }, { "epoch": 0.18879935975554302, "grad_norm": 2.257965087890625, "learning_rate": 8.363179508335054e-06, "loss": 0.9857, "step": 10380 }, { "epoch": 0.18898124738536531, "grad_norm": 2.02932071685791, "learning_rate": 8.361304355978925e-06, "loss": 0.989, "step": 10390 }, { "epoch": 0.1891631350151876, "grad_norm": 1.8212867975234985, "learning_rate": 8.359429203622795e-06, "loss": 0.9619, "step": 10400 }, { "epoch": 0.1893450226450099, "grad_norm": 1.7713617086410522, "learning_rate": 8.357554051266666e-06, "loss": 0.9509, "step": 10410 }, { "epoch": 0.1895269102748322, "grad_norm": 1.873590111732483, "learning_rate": 8.355678898910537e-06, "loss": 1.0381, "step": 10420 }, { "epoch": 0.1897087979046545, "grad_norm": 1.75608491897583, "learning_rate": 8.353803746554409e-06, "loss": 0.987, "step": 10430 }, { "epoch": 0.1898906855344768, "grad_norm": 1.8407073020935059, "learning_rate": 8.35192859419828e-06, "loss": 1.032, "step": 10440 }, { "epoch": 0.1900725731642991, "grad_norm": 1.8600488901138306, "learning_rate": 8.350053441842151e-06, "loss": 0.9886, "step": 10450 }, { "epoch": 0.1902544607941214, "grad_norm": 1.9448249340057373, "learning_rate": 8.348178289486023e-06, "loss": 1.0013, "step": 10460 }, { "epoch": 0.1904363484239437, "grad_norm": 1.8869528770446777, "learning_rate": 8.346303137129892e-06, "loss": 0.9809, "step": 10470 }, { "epoch": 0.190618236053766, "grad_norm": 1.6909080743789673, "learning_rate": 8.344427984773764e-06, "loss": 0.9675, "step": 10480 }, { "epoch": 0.19080012368358829, "grad_norm": 1.6651352643966675, "learning_rate": 8.342552832417633e-06, "loss": 0.952, "step": 10490 }, { "epoch": 0.19098201131341058, "grad_norm": 2.0679590702056885, "learning_rate": 8.340677680061505e-06, "loss": 0.9344, "step": 10500 }, { "epoch": 0.19116389894323288, "grad_norm": 2.042505979537964, "learning_rate": 8.338802527705376e-06, "loss": 1.0025, "step": 10510 }, { "epoch": 0.19134578657305518, "grad_norm": 1.9665930271148682, "learning_rate": 8.336927375349247e-06, "loss": 0.9486, "step": 10520 }, { "epoch": 0.19152767420287747, "grad_norm": 2.1786909103393555, "learning_rate": 8.335052222993119e-06, "loss": 0.9696, "step": 10530 }, { "epoch": 0.19170956183269977, "grad_norm": 1.9954205751419067, "learning_rate": 8.33317707063699e-06, "loss": 1.0502, "step": 10540 }, { "epoch": 0.19189144946252204, "grad_norm": 1.5932093858718872, "learning_rate": 8.331301918280861e-06, "loss": 0.9362, "step": 10550 }, { "epoch": 0.19207333709234434, "grad_norm": 1.6336421966552734, "learning_rate": 8.329426765924731e-06, "loss": 1.0048, "step": 10560 }, { "epoch": 0.19225522472216663, "grad_norm": 1.7362372875213623, "learning_rate": 8.327551613568602e-06, "loss": 0.9917, "step": 10570 }, { "epoch": 0.19243711235198893, "grad_norm": 1.7530592679977417, "learning_rate": 8.325676461212474e-06, "loss": 0.9773, "step": 10580 }, { "epoch": 0.19261899998181123, "grad_norm": 2.042093515396118, "learning_rate": 8.323801308856345e-06, "loss": 0.9897, "step": 10590 }, { "epoch": 0.19280088761163353, "grad_norm": 2.2356042861938477, "learning_rate": 8.321926156500216e-06, "loss": 0.9834, "step": 10600 }, { "epoch": 0.19298277524145582, "grad_norm": 1.9126919507980347, "learning_rate": 8.320051004144088e-06, "loss": 0.9641, "step": 10610 }, { "epoch": 0.19316466287127812, "grad_norm": 1.8801255226135254, "learning_rate": 8.31817585178796e-06, "loss": 0.9942, "step": 10620 }, { "epoch": 0.19334655050110042, "grad_norm": 1.8805824518203735, "learning_rate": 8.316300699431829e-06, "loss": 0.9942, "step": 10630 }, { "epoch": 0.19352843813092271, "grad_norm": 1.7055130004882812, "learning_rate": 8.3144255470757e-06, "loss": 0.9783, "step": 10640 }, { "epoch": 0.193710325760745, "grad_norm": 3.394209861755371, "learning_rate": 8.312550394719572e-06, "loss": 0.9732, "step": 10650 }, { "epoch": 0.1938922133905673, "grad_norm": 1.902644395828247, "learning_rate": 8.310675242363443e-06, "loss": 0.9822, "step": 10660 }, { "epoch": 0.1940741010203896, "grad_norm": 1.6994283199310303, "learning_rate": 8.308800090007314e-06, "loss": 0.9305, "step": 10670 }, { "epoch": 0.1942559886502119, "grad_norm": 2.0488579273223877, "learning_rate": 8.306924937651186e-06, "loss": 0.9807, "step": 10680 }, { "epoch": 0.1944378762800342, "grad_norm": 1.5741785764694214, "learning_rate": 8.305049785295057e-06, "loss": 0.9985, "step": 10690 }, { "epoch": 0.1946197639098565, "grad_norm": 1.8194721937179565, "learning_rate": 8.303174632938927e-06, "loss": 0.9875, "step": 10700 }, { "epoch": 0.1948016515396788, "grad_norm": 1.9223861694335938, "learning_rate": 8.301299480582798e-06, "loss": 0.9517, "step": 10710 }, { "epoch": 0.1949835391695011, "grad_norm": 1.967619776725769, "learning_rate": 8.29942432822667e-06, "loss": 0.9798, "step": 10720 }, { "epoch": 0.1951654267993234, "grad_norm": 1.9623382091522217, "learning_rate": 8.29754917587054e-06, "loss": 0.9794, "step": 10730 }, { "epoch": 0.19534731442914569, "grad_norm": 1.5949537754058838, "learning_rate": 8.295674023514412e-06, "loss": 1.0328, "step": 10740 }, { "epoch": 0.19552920205896798, "grad_norm": 1.805527925491333, "learning_rate": 8.293798871158282e-06, "loss": 0.9606, "step": 10750 }, { "epoch": 0.19571108968879025, "grad_norm": 1.7760205268859863, "learning_rate": 8.291923718802153e-06, "loss": 0.9832, "step": 10760 }, { "epoch": 0.19589297731861255, "grad_norm": 1.76691734790802, "learning_rate": 8.290048566446024e-06, "loss": 1.0458, "step": 10770 }, { "epoch": 0.19607486494843485, "grad_norm": 2.026409864425659, "learning_rate": 8.288173414089896e-06, "loss": 0.973, "step": 10780 }, { "epoch": 0.19625675257825714, "grad_norm": 1.7038363218307495, "learning_rate": 8.286298261733767e-06, "loss": 0.9532, "step": 10790 }, { "epoch": 0.19643864020807944, "grad_norm": 1.8819013833999634, "learning_rate": 8.284423109377637e-06, "loss": 1.0334, "step": 10800 }, { "epoch": 0.19662052783790174, "grad_norm": 1.8777210712432861, "learning_rate": 8.282547957021508e-06, "loss": 1.0085, "step": 10810 }, { "epoch": 0.19680241546772403, "grad_norm": 1.8163093328475952, "learning_rate": 8.28067280466538e-06, "loss": 1.0241, "step": 10820 }, { "epoch": 0.19698430309754633, "grad_norm": 1.732369065284729, "learning_rate": 8.27879765230925e-06, "loss": 1.045, "step": 10830 }, { "epoch": 0.19716619072736863, "grad_norm": 1.7144161462783813, "learning_rate": 8.276922499953122e-06, "loss": 0.9715, "step": 10840 }, { "epoch": 0.19734807835719093, "grad_norm": 1.760724663734436, "learning_rate": 8.275047347596993e-06, "loss": 1.0052, "step": 10850 }, { "epoch": 0.19752996598701322, "grad_norm": 1.7421636581420898, "learning_rate": 8.273172195240865e-06, "loss": 1.0288, "step": 10860 }, { "epoch": 0.19771185361683552, "grad_norm": 1.7709577083587646, "learning_rate": 8.271297042884734e-06, "loss": 0.9413, "step": 10870 }, { "epoch": 0.19789374124665782, "grad_norm": 1.999746322631836, "learning_rate": 8.269421890528606e-06, "loss": 1.0305, "step": 10880 }, { "epoch": 0.19807562887648011, "grad_norm": 1.791949987411499, "learning_rate": 8.267546738172477e-06, "loss": 1.0504, "step": 10890 }, { "epoch": 0.1982575165063024, "grad_norm": 1.7948578596115112, "learning_rate": 8.265671585816348e-06, "loss": 0.952, "step": 10900 }, { "epoch": 0.1984394041361247, "grad_norm": 1.9292140007019043, "learning_rate": 8.26379643346022e-06, "loss": 0.9635, "step": 10910 }, { "epoch": 0.198621291765947, "grad_norm": 2.0732581615448, "learning_rate": 8.261921281104091e-06, "loss": 0.9672, "step": 10920 }, { "epoch": 0.1988031793957693, "grad_norm": 1.9955917596817017, "learning_rate": 8.260046128747962e-06, "loss": 0.9559, "step": 10930 }, { "epoch": 0.1989850670255916, "grad_norm": 1.9403129816055298, "learning_rate": 8.258170976391832e-06, "loss": 1.002, "step": 10940 }, { "epoch": 0.1991669546554139, "grad_norm": 1.8930565118789673, "learning_rate": 8.256295824035703e-06, "loss": 0.994, "step": 10950 }, { "epoch": 0.1993488422852362, "grad_norm": 2.046185255050659, "learning_rate": 8.254420671679575e-06, "loss": 1.0279, "step": 10960 }, { "epoch": 0.19953072991505846, "grad_norm": 1.9677140712738037, "learning_rate": 8.252545519323446e-06, "loss": 0.9992, "step": 10970 }, { "epoch": 0.19971261754488076, "grad_norm": 1.7812222242355347, "learning_rate": 8.250670366967317e-06, "loss": 0.983, "step": 10980 }, { "epoch": 0.19989450517470306, "grad_norm": 1.8911069631576538, "learning_rate": 8.248795214611189e-06, "loss": 1.0304, "step": 10990 }, { "epoch": 0.20007639280452535, "grad_norm": 1.9693381786346436, "learning_rate": 8.24692006225506e-06, "loss": 1.0514, "step": 11000 }, { "epoch": 0.20025828043434765, "grad_norm": 2.162670373916626, "learning_rate": 8.24504490989893e-06, "loss": 0.9919, "step": 11010 }, { "epoch": 0.20044016806416995, "grad_norm": 1.6367141008377075, "learning_rate": 8.243169757542801e-06, "loss": 0.9598, "step": 11020 }, { "epoch": 0.20062205569399225, "grad_norm": 1.9117881059646606, "learning_rate": 8.24129460518667e-06, "loss": 0.9927, "step": 11030 }, { "epoch": 0.20080394332381454, "grad_norm": 1.720945119857788, "learning_rate": 8.239419452830542e-06, "loss": 1.0044, "step": 11040 }, { "epoch": 0.20098583095363684, "grad_norm": 1.870945692062378, "learning_rate": 8.237544300474414e-06, "loss": 0.9753, "step": 11050 }, { "epoch": 0.20116771858345914, "grad_norm": 1.8370575904846191, "learning_rate": 8.235669148118285e-06, "loss": 1.0248, "step": 11060 }, { "epoch": 0.20134960621328143, "grad_norm": 2.118319272994995, "learning_rate": 8.233793995762156e-06, "loss": 1.0266, "step": 11070 }, { "epoch": 0.20153149384310373, "grad_norm": 1.9159519672393799, "learning_rate": 8.231918843406028e-06, "loss": 0.9947, "step": 11080 }, { "epoch": 0.20171338147292603, "grad_norm": 1.8845148086547852, "learning_rate": 8.230043691049899e-06, "loss": 1.0161, "step": 11090 }, { "epoch": 0.20189526910274833, "grad_norm": 1.9567432403564453, "learning_rate": 8.228168538693769e-06, "loss": 0.9669, "step": 11100 }, { "epoch": 0.20207715673257062, "grad_norm": 2.033036708831787, "learning_rate": 8.22629338633764e-06, "loss": 0.9536, "step": 11110 }, { "epoch": 0.20225904436239292, "grad_norm": 1.9330066442489624, "learning_rate": 8.224418233981511e-06, "loss": 0.9697, "step": 11120 }, { "epoch": 0.20244093199221522, "grad_norm": 1.6622217893600464, "learning_rate": 8.222543081625383e-06, "loss": 0.96, "step": 11130 }, { "epoch": 0.20262281962203751, "grad_norm": 1.8531430959701538, "learning_rate": 8.220667929269254e-06, "loss": 0.9743, "step": 11140 }, { "epoch": 0.2028047072518598, "grad_norm": 1.3430469036102295, "learning_rate": 8.218792776913125e-06, "loss": 0.941, "step": 11150 }, { "epoch": 0.2029865948816821, "grad_norm": 1.8915189504623413, "learning_rate": 8.216917624556997e-06, "loss": 0.9764, "step": 11160 }, { "epoch": 0.2031684825115044, "grad_norm": 1.8248958587646484, "learning_rate": 8.215042472200866e-06, "loss": 1.0178, "step": 11170 }, { "epoch": 0.20335037014132668, "grad_norm": 1.856082797050476, "learning_rate": 8.213167319844738e-06, "loss": 0.9268, "step": 11180 }, { "epoch": 0.20353225777114897, "grad_norm": 2.1717777252197266, "learning_rate": 8.211292167488609e-06, "loss": 0.9989, "step": 11190 }, { "epoch": 0.20371414540097127, "grad_norm": 1.8282170295715332, "learning_rate": 8.20941701513248e-06, "loss": 0.9856, "step": 11200 }, { "epoch": 0.20389603303079357, "grad_norm": 1.9032844305038452, "learning_rate": 8.207541862776352e-06, "loss": 0.9714, "step": 11210 }, { "epoch": 0.20407792066061586, "grad_norm": 2.054607629776001, "learning_rate": 8.205666710420223e-06, "loss": 1.0015, "step": 11220 }, { "epoch": 0.20425980829043816, "grad_norm": 1.9749099016189575, "learning_rate": 8.203791558064094e-06, "loss": 1.0006, "step": 11230 }, { "epoch": 0.20444169592026046, "grad_norm": 1.956068515777588, "learning_rate": 8.201916405707964e-06, "loss": 1.011, "step": 11240 }, { "epoch": 0.20462358355008275, "grad_norm": 1.9118454456329346, "learning_rate": 8.200041253351835e-06, "loss": 0.9923, "step": 11250 }, { "epoch": 0.20480547117990505, "grad_norm": 1.6765189170837402, "learning_rate": 8.198166100995707e-06, "loss": 0.991, "step": 11260 }, { "epoch": 0.20498735880972735, "grad_norm": 2.0210773944854736, "learning_rate": 8.196290948639578e-06, "loss": 0.9965, "step": 11270 }, { "epoch": 0.20516924643954965, "grad_norm": 1.7786681652069092, "learning_rate": 8.19441579628345e-06, "loss": 0.9793, "step": 11280 }, { "epoch": 0.20535113406937194, "grad_norm": 2.0084500312805176, "learning_rate": 8.19254064392732e-06, "loss": 1.0006, "step": 11290 }, { "epoch": 0.20553302169919424, "grad_norm": 1.7777109146118164, "learning_rate": 8.19066549157119e-06, "loss": 1.0246, "step": 11300 }, { "epoch": 0.20571490932901654, "grad_norm": 1.8567805290222168, "learning_rate": 8.188790339215062e-06, "loss": 0.924, "step": 11310 }, { "epoch": 0.20589679695883883, "grad_norm": 1.7877449989318848, "learning_rate": 8.186915186858933e-06, "loss": 0.9478, "step": 11320 }, { "epoch": 0.20607868458866113, "grad_norm": 2.029236078262329, "learning_rate": 8.185040034502804e-06, "loss": 0.9835, "step": 11330 }, { "epoch": 0.20626057221848343, "grad_norm": 1.9978158473968506, "learning_rate": 8.183164882146674e-06, "loss": 0.9715, "step": 11340 }, { "epoch": 0.20644245984830573, "grad_norm": 1.7695239782333374, "learning_rate": 8.181289729790545e-06, "loss": 0.9469, "step": 11350 }, { "epoch": 0.20662434747812802, "grad_norm": 2.0341503620147705, "learning_rate": 8.179414577434417e-06, "loss": 0.9961, "step": 11360 }, { "epoch": 0.20680623510795032, "grad_norm": 1.7673083543777466, "learning_rate": 8.177539425078288e-06, "loss": 1.0507, "step": 11370 }, { "epoch": 0.20698812273777262, "grad_norm": 1.630971908569336, "learning_rate": 8.17566427272216e-06, "loss": 1.0043, "step": 11380 }, { "epoch": 0.2071700103675949, "grad_norm": 2.2276577949523926, "learning_rate": 8.17378912036603e-06, "loss": 1.0213, "step": 11390 }, { "epoch": 0.20735189799741718, "grad_norm": 2.200019121170044, "learning_rate": 8.171913968009902e-06, "loss": 0.973, "step": 11400 }, { "epoch": 0.20753378562723948, "grad_norm": 2.006477117538452, "learning_rate": 8.170038815653772e-06, "loss": 0.9734, "step": 11410 }, { "epoch": 0.20771567325706178, "grad_norm": 1.8218709230422974, "learning_rate": 8.168163663297643e-06, "loss": 0.9869, "step": 11420 }, { "epoch": 0.20789756088688408, "grad_norm": 1.9025583267211914, "learning_rate": 8.166288510941514e-06, "loss": 1.0049, "step": 11430 }, { "epoch": 0.20807944851670637, "grad_norm": 2.041063070297241, "learning_rate": 8.164413358585386e-06, "loss": 1.0146, "step": 11440 }, { "epoch": 0.20826133614652867, "grad_norm": 1.7757340669631958, "learning_rate": 8.162538206229257e-06, "loss": 0.9985, "step": 11450 }, { "epoch": 0.20844322377635097, "grad_norm": 1.8128620386123657, "learning_rate": 8.160663053873129e-06, "loss": 0.9494, "step": 11460 }, { "epoch": 0.20862511140617326, "grad_norm": 1.8823188543319702, "learning_rate": 8.158787901517e-06, "loss": 0.942, "step": 11470 }, { "epoch": 0.20880699903599556, "grad_norm": 1.961885690689087, "learning_rate": 8.15691274916087e-06, "loss": 0.997, "step": 11480 }, { "epoch": 0.20898888666581786, "grad_norm": 1.9007952213287354, "learning_rate": 8.155037596804741e-06, "loss": 0.9836, "step": 11490 }, { "epoch": 0.20917077429564016, "grad_norm": 1.8076527118682861, "learning_rate": 8.153162444448612e-06, "loss": 0.9707, "step": 11500 }, { "epoch": 0.20935266192546245, "grad_norm": 1.6519578695297241, "learning_rate": 8.151287292092484e-06, "loss": 0.9999, "step": 11510 }, { "epoch": 0.20953454955528475, "grad_norm": 1.8570841550827026, "learning_rate": 8.149412139736355e-06, "loss": 0.9501, "step": 11520 }, { "epoch": 0.20971643718510705, "grad_norm": 2.070585012435913, "learning_rate": 8.147536987380226e-06, "loss": 1.0006, "step": 11530 }, { "epoch": 0.20989832481492934, "grad_norm": 1.8866047859191895, "learning_rate": 8.145661835024098e-06, "loss": 0.9632, "step": 11540 }, { "epoch": 0.21008021244475164, "grad_norm": 1.8845032453536987, "learning_rate": 8.143786682667967e-06, "loss": 1.0292, "step": 11550 }, { "epoch": 0.21026210007457394, "grad_norm": 1.9951459169387817, "learning_rate": 8.141911530311839e-06, "loss": 1.0135, "step": 11560 }, { "epoch": 0.21044398770439623, "grad_norm": 1.811766505241394, "learning_rate": 8.14003637795571e-06, "loss": 1.03, "step": 11570 }, { "epoch": 0.21062587533421853, "grad_norm": 2.9458611011505127, "learning_rate": 8.13816122559958e-06, "loss": 0.968, "step": 11580 }, { "epoch": 0.21080776296404083, "grad_norm": 1.929782509803772, "learning_rate": 8.136286073243451e-06, "loss": 0.9962, "step": 11590 }, { "epoch": 0.2109896505938631, "grad_norm": 1.7960035800933838, "learning_rate": 8.134410920887322e-06, "loss": 1.0159, "step": 11600 }, { "epoch": 0.2111715382236854, "grad_norm": 2.015784502029419, "learning_rate": 8.132535768531194e-06, "loss": 0.9929, "step": 11610 }, { "epoch": 0.2113534258535077, "grad_norm": 1.8309593200683594, "learning_rate": 8.130660616175065e-06, "loss": 0.9568, "step": 11620 }, { "epoch": 0.21153531348333, "grad_norm": 2.0688846111297607, "learning_rate": 8.128785463818936e-06, "loss": 0.9935, "step": 11630 }, { "epoch": 0.2117172011131523, "grad_norm": 2.307091236114502, "learning_rate": 8.126910311462806e-06, "loss": 0.9811, "step": 11640 }, { "epoch": 0.21189908874297458, "grad_norm": 1.8823400735855103, "learning_rate": 8.125035159106677e-06, "loss": 1.0247, "step": 11650 }, { "epoch": 0.21208097637279688, "grad_norm": 2.063138723373413, "learning_rate": 8.123160006750549e-06, "loss": 1.0213, "step": 11660 }, { "epoch": 0.21226286400261918, "grad_norm": 1.8086344003677368, "learning_rate": 8.12128485439442e-06, "loss": 1.0236, "step": 11670 }, { "epoch": 0.21244475163244148, "grad_norm": 1.7887110710144043, "learning_rate": 8.119409702038291e-06, "loss": 0.9555, "step": 11680 }, { "epoch": 0.21262663926226377, "grad_norm": 1.8482232093811035, "learning_rate": 8.117534549682163e-06, "loss": 0.9935, "step": 11690 }, { "epoch": 0.21280852689208607, "grad_norm": 1.8163851499557495, "learning_rate": 8.115659397326034e-06, "loss": 1.0119, "step": 11700 }, { "epoch": 0.21299041452190837, "grad_norm": 1.9031025171279907, "learning_rate": 8.113784244969904e-06, "loss": 0.9904, "step": 11710 }, { "epoch": 0.21317230215173066, "grad_norm": 1.9545713663101196, "learning_rate": 8.111909092613775e-06, "loss": 0.9784, "step": 11720 }, { "epoch": 0.21335418978155296, "grad_norm": 1.8925461769104004, "learning_rate": 8.110033940257646e-06, "loss": 0.9227, "step": 11730 }, { "epoch": 0.21353607741137526, "grad_norm": 1.957401990890503, "learning_rate": 8.108158787901518e-06, "loss": 0.9666, "step": 11740 }, { "epoch": 0.21371796504119756, "grad_norm": 1.8446511030197144, "learning_rate": 8.106283635545389e-06, "loss": 0.955, "step": 11750 }, { "epoch": 0.21389985267101985, "grad_norm": 1.9065465927124023, "learning_rate": 8.10440848318926e-06, "loss": 0.9433, "step": 11760 }, { "epoch": 0.21408174030084215, "grad_norm": 1.533137559890747, "learning_rate": 8.102533330833132e-06, "loss": 0.9727, "step": 11770 }, { "epoch": 0.21426362793066445, "grad_norm": 1.7309361696243286, "learning_rate": 8.100658178477001e-06, "loss": 0.9759, "step": 11780 }, { "epoch": 0.21444551556048674, "grad_norm": 1.7198470830917358, "learning_rate": 8.098783026120873e-06, "loss": 0.9571, "step": 11790 }, { "epoch": 0.21462740319030904, "grad_norm": 1.7714506387710571, "learning_rate": 8.096907873764744e-06, "loss": 0.9574, "step": 11800 }, { "epoch": 0.2148092908201313, "grad_norm": 1.8110978603363037, "learning_rate": 8.095032721408615e-06, "loss": 0.9688, "step": 11810 }, { "epoch": 0.2149911784499536, "grad_norm": 1.6015188694000244, "learning_rate": 8.093157569052487e-06, "loss": 0.9442, "step": 11820 }, { "epoch": 0.2151730660797759, "grad_norm": 1.8035808801651, "learning_rate": 8.091282416696358e-06, "loss": 1.0031, "step": 11830 }, { "epoch": 0.2153549537095982, "grad_norm": 1.9116731882095337, "learning_rate": 8.08940726434023e-06, "loss": 1.0088, "step": 11840 }, { "epoch": 0.2155368413394205, "grad_norm": 1.7762588262557983, "learning_rate": 8.0875321119841e-06, "loss": 0.9855, "step": 11850 }, { "epoch": 0.2157187289692428, "grad_norm": 1.9459120035171509, "learning_rate": 8.08565695962797e-06, "loss": 0.9326, "step": 11860 }, { "epoch": 0.2159006165990651, "grad_norm": 1.9648765325546265, "learning_rate": 8.083781807271842e-06, "loss": 0.9502, "step": 11870 }, { "epoch": 0.2160825042288874, "grad_norm": 1.4599266052246094, "learning_rate": 8.081906654915712e-06, "loss": 0.9584, "step": 11880 }, { "epoch": 0.2162643918587097, "grad_norm": 1.8366084098815918, "learning_rate": 8.080031502559583e-06, "loss": 0.9675, "step": 11890 }, { "epoch": 0.21644627948853198, "grad_norm": 1.8567276000976562, "learning_rate": 8.078156350203454e-06, "loss": 0.9978, "step": 11900 }, { "epoch": 0.21662816711835428, "grad_norm": 2.2244911193847656, "learning_rate": 8.076281197847326e-06, "loss": 0.9946, "step": 11910 }, { "epoch": 0.21681005474817658, "grad_norm": 1.7405041456222534, "learning_rate": 8.074406045491197e-06, "loss": 0.974, "step": 11920 }, { "epoch": 0.21699194237799888, "grad_norm": 1.6735631227493286, "learning_rate": 8.072530893135068e-06, "loss": 0.9454, "step": 11930 }, { "epoch": 0.21717383000782117, "grad_norm": 1.8188669681549072, "learning_rate": 8.07065574077894e-06, "loss": 0.9576, "step": 11940 }, { "epoch": 0.21735571763764347, "grad_norm": 1.6393500566482544, "learning_rate": 8.06878058842281e-06, "loss": 0.9324, "step": 11950 }, { "epoch": 0.21753760526746577, "grad_norm": 1.9130679368972778, "learning_rate": 8.06690543606668e-06, "loss": 0.9743, "step": 11960 }, { "epoch": 0.21771949289728806, "grad_norm": 1.8405766487121582, "learning_rate": 8.065030283710552e-06, "loss": 0.9923, "step": 11970 }, { "epoch": 0.21790138052711036, "grad_norm": 1.77022123336792, "learning_rate": 8.063155131354423e-06, "loss": 0.9691, "step": 11980 }, { "epoch": 0.21808326815693266, "grad_norm": 1.9239343404769897, "learning_rate": 8.061279978998295e-06, "loss": 0.9736, "step": 11990 }, { "epoch": 0.21826515578675496, "grad_norm": 2.168621063232422, "learning_rate": 8.059404826642166e-06, "loss": 1.0086, "step": 12000 }, { "epoch": 0.21844704341657725, "grad_norm": 2.1215877532958984, "learning_rate": 8.057529674286037e-06, "loss": 0.9864, "step": 12010 }, { "epoch": 0.21862893104639952, "grad_norm": 2.244178533554077, "learning_rate": 8.055654521929907e-06, "loss": 0.9807, "step": 12020 }, { "epoch": 0.21881081867622182, "grad_norm": 1.7695603370666504, "learning_rate": 8.053779369573778e-06, "loss": 0.9679, "step": 12030 }, { "epoch": 0.21899270630604412, "grad_norm": 1.5863369703292847, "learning_rate": 8.05190421721765e-06, "loss": 0.9377, "step": 12040 }, { "epoch": 0.2191745939358664, "grad_norm": 1.8923125267028809, "learning_rate": 8.050029064861521e-06, "loss": 1.0003, "step": 12050 }, { "epoch": 0.2193564815656887, "grad_norm": 1.6434249877929688, "learning_rate": 8.048153912505392e-06, "loss": 0.9814, "step": 12060 }, { "epoch": 0.219538369195511, "grad_norm": 1.5944768190383911, "learning_rate": 8.046278760149264e-06, "loss": 0.9629, "step": 12070 }, { "epoch": 0.2197202568253333, "grad_norm": 1.7468922138214111, "learning_rate": 8.044403607793135e-06, "loss": 0.966, "step": 12080 }, { "epoch": 0.2199021444551556, "grad_norm": 1.510898470878601, "learning_rate": 8.042528455437005e-06, "loss": 0.9685, "step": 12090 }, { "epoch": 0.2200840320849779, "grad_norm": 2.0824756622314453, "learning_rate": 8.040653303080876e-06, "loss": 1.0105, "step": 12100 }, { "epoch": 0.2202659197148002, "grad_norm": 1.8687022924423218, "learning_rate": 8.038778150724747e-06, "loss": 0.9891, "step": 12110 }, { "epoch": 0.2204478073446225, "grad_norm": 1.935361623764038, "learning_rate": 8.036902998368619e-06, "loss": 0.9334, "step": 12120 }, { "epoch": 0.2206296949744448, "grad_norm": 1.8009439706802368, "learning_rate": 8.035027846012488e-06, "loss": 0.9135, "step": 12130 }, { "epoch": 0.2208115826042671, "grad_norm": 1.8109952211380005, "learning_rate": 8.03315269365636e-06, "loss": 0.9866, "step": 12140 }, { "epoch": 0.22099347023408938, "grad_norm": 2.015763998031616, "learning_rate": 8.031277541300231e-06, "loss": 1.0104, "step": 12150 }, { "epoch": 0.22117535786391168, "grad_norm": 1.8925752639770508, "learning_rate": 8.029402388944102e-06, "loss": 0.9968, "step": 12160 }, { "epoch": 0.22135724549373398, "grad_norm": 1.9365222454071045, "learning_rate": 8.027527236587974e-06, "loss": 0.9357, "step": 12170 }, { "epoch": 0.22153913312355628, "grad_norm": 2.1653497219085693, "learning_rate": 8.025652084231843e-06, "loss": 1.0105, "step": 12180 }, { "epoch": 0.22172102075337857, "grad_norm": 1.7420566082000732, "learning_rate": 8.023776931875715e-06, "loss": 0.9531, "step": 12190 }, { "epoch": 0.22190290838320087, "grad_norm": 1.850256085395813, "learning_rate": 8.021901779519586e-06, "loss": 1.0038, "step": 12200 }, { "epoch": 0.22208479601302317, "grad_norm": 2.018156051635742, "learning_rate": 8.020026627163457e-06, "loss": 0.9771, "step": 12210 }, { "epoch": 0.22226668364284544, "grad_norm": 1.6256285905838013, "learning_rate": 8.018151474807329e-06, "loss": 1.0064, "step": 12220 }, { "epoch": 0.22244857127266773, "grad_norm": 2.0152883529663086, "learning_rate": 8.0162763224512e-06, "loss": 1.0018, "step": 12230 }, { "epoch": 0.22263045890249003, "grad_norm": 1.9446229934692383, "learning_rate": 8.014401170095071e-06, "loss": 1.0287, "step": 12240 }, { "epoch": 0.22281234653231233, "grad_norm": 2.000818967819214, "learning_rate": 8.012526017738941e-06, "loss": 1.0203, "step": 12250 }, { "epoch": 0.22299423416213462, "grad_norm": 2.011997938156128, "learning_rate": 8.010650865382812e-06, "loss": 0.9352, "step": 12260 }, { "epoch": 0.22317612179195692, "grad_norm": 2.086158037185669, "learning_rate": 8.008775713026684e-06, "loss": 0.9917, "step": 12270 }, { "epoch": 0.22335800942177922, "grad_norm": 2.115729570388794, "learning_rate": 8.006900560670555e-06, "loss": 1.0187, "step": 12280 }, { "epoch": 0.22353989705160152, "grad_norm": 1.5539520978927612, "learning_rate": 8.005025408314427e-06, "loss": 0.9945, "step": 12290 }, { "epoch": 0.2237217846814238, "grad_norm": 1.9139909744262695, "learning_rate": 8.003150255958298e-06, "loss": 1.0173, "step": 12300 }, { "epoch": 0.2239036723112461, "grad_norm": 1.7895619869232178, "learning_rate": 8.00127510360217e-06, "loss": 0.9546, "step": 12310 }, { "epoch": 0.2240855599410684, "grad_norm": 1.6601608991622925, "learning_rate": 7.999399951246039e-06, "loss": 0.9771, "step": 12320 }, { "epoch": 0.2242674475708907, "grad_norm": 1.9098025560379028, "learning_rate": 7.99752479888991e-06, "loss": 0.9953, "step": 12330 }, { "epoch": 0.224449335200713, "grad_norm": 1.817420482635498, "learning_rate": 7.995649646533782e-06, "loss": 0.971, "step": 12340 }, { "epoch": 0.2246312228305353, "grad_norm": 1.8927239179611206, "learning_rate": 7.993774494177653e-06, "loss": 0.9838, "step": 12350 }, { "epoch": 0.2248131104603576, "grad_norm": 1.871390461921692, "learning_rate": 7.991899341821524e-06, "loss": 0.9977, "step": 12360 }, { "epoch": 0.2249949980901799, "grad_norm": 1.9121569395065308, "learning_rate": 7.990024189465396e-06, "loss": 0.9553, "step": 12370 }, { "epoch": 0.2251768857200022, "grad_norm": 1.8793610334396362, "learning_rate": 7.988149037109267e-06, "loss": 1.0076, "step": 12380 }, { "epoch": 0.2253587733498245, "grad_norm": 1.7325600385665894, "learning_rate": 7.986273884753137e-06, "loss": 0.9736, "step": 12390 }, { "epoch": 0.22554066097964678, "grad_norm": 1.9257549047470093, "learning_rate": 7.984398732397008e-06, "loss": 1.0194, "step": 12400 }, { "epoch": 0.22572254860946908, "grad_norm": 1.8143707513809204, "learning_rate": 7.98252358004088e-06, "loss": 1.0367, "step": 12410 }, { "epoch": 0.22590443623929138, "grad_norm": 2.1535956859588623, "learning_rate": 7.980648427684749e-06, "loss": 1.0314, "step": 12420 }, { "epoch": 0.22608632386911365, "grad_norm": 1.720963716506958, "learning_rate": 7.97877327532862e-06, "loss": 0.9661, "step": 12430 }, { "epoch": 0.22626821149893594, "grad_norm": 1.6227248907089233, "learning_rate": 7.976898122972492e-06, "loss": 0.9881, "step": 12440 }, { "epoch": 0.22645009912875824, "grad_norm": 1.8718997240066528, "learning_rate": 7.975022970616363e-06, "loss": 0.9953, "step": 12450 }, { "epoch": 0.22663198675858054, "grad_norm": 2.049736261367798, "learning_rate": 7.973147818260234e-06, "loss": 1.032, "step": 12460 }, { "epoch": 0.22681387438840284, "grad_norm": 2.0638585090637207, "learning_rate": 7.971272665904106e-06, "loss": 1.0031, "step": 12470 }, { "epoch": 0.22699576201822513, "grad_norm": 1.996248483657837, "learning_rate": 7.969397513547977e-06, "loss": 0.9748, "step": 12480 }, { "epoch": 0.22717764964804743, "grad_norm": 1.5725581645965576, "learning_rate": 7.967522361191847e-06, "loss": 0.9354, "step": 12490 }, { "epoch": 0.22735953727786973, "grad_norm": 1.909485936164856, "learning_rate": 7.965647208835718e-06, "loss": 1.0284, "step": 12500 }, { "epoch": 0.22754142490769202, "grad_norm": 2.011371374130249, "learning_rate": 7.96377205647959e-06, "loss": 0.9995, "step": 12510 }, { "epoch": 0.22772331253751432, "grad_norm": 1.9009560346603394, "learning_rate": 7.96189690412346e-06, "loss": 0.9718, "step": 12520 }, { "epoch": 0.22790520016733662, "grad_norm": 1.7891513109207153, "learning_rate": 7.960021751767332e-06, "loss": 0.9895, "step": 12530 }, { "epoch": 0.22808708779715892, "grad_norm": 1.908170223236084, "learning_rate": 7.958146599411203e-06, "loss": 0.9717, "step": 12540 }, { "epoch": 0.2282689754269812, "grad_norm": 1.6964102983474731, "learning_rate": 7.956271447055075e-06, "loss": 1.0298, "step": 12550 }, { "epoch": 0.2284508630568035, "grad_norm": 1.9340261220932007, "learning_rate": 7.954396294698944e-06, "loss": 1.0038, "step": 12560 }, { "epoch": 0.2286327506866258, "grad_norm": 1.6998636722564697, "learning_rate": 7.952521142342816e-06, "loss": 0.9722, "step": 12570 }, { "epoch": 0.2288146383164481, "grad_norm": 2.0776619911193848, "learning_rate": 7.950645989986687e-06, "loss": 1.0009, "step": 12580 }, { "epoch": 0.2289965259462704, "grad_norm": 1.6786268949508667, "learning_rate": 7.948770837630558e-06, "loss": 0.9725, "step": 12590 }, { "epoch": 0.2291784135760927, "grad_norm": 1.8172553777694702, "learning_rate": 7.94689568527443e-06, "loss": 0.971, "step": 12600 }, { "epoch": 0.229360301205915, "grad_norm": 1.9086748361587524, "learning_rate": 7.945020532918301e-06, "loss": 0.9637, "step": 12610 }, { "epoch": 0.2295421888357373, "grad_norm": 1.876025676727295, "learning_rate": 7.943145380562172e-06, "loss": 1.0055, "step": 12620 }, { "epoch": 0.2297240764655596, "grad_norm": 1.7510485649108887, "learning_rate": 7.941270228206042e-06, "loss": 1.0209, "step": 12630 }, { "epoch": 0.22990596409538186, "grad_norm": 2.0021560192108154, "learning_rate": 7.939395075849913e-06, "loss": 1.0122, "step": 12640 }, { "epoch": 0.23008785172520416, "grad_norm": 1.8680665493011475, "learning_rate": 7.937519923493785e-06, "loss": 0.9616, "step": 12650 }, { "epoch": 0.23026973935502645, "grad_norm": 1.8153934478759766, "learning_rate": 7.935644771137656e-06, "loss": 0.959, "step": 12660 }, { "epoch": 0.23045162698484875, "grad_norm": 1.6201245784759521, "learning_rate": 7.933769618781528e-06, "loss": 0.9439, "step": 12670 }, { "epoch": 0.23063351461467105, "grad_norm": 1.670819640159607, "learning_rate": 7.931894466425397e-06, "loss": 0.9701, "step": 12680 }, { "epoch": 0.23081540224449335, "grad_norm": 1.9798805713653564, "learning_rate": 7.930019314069269e-06, "loss": 0.9486, "step": 12690 }, { "epoch": 0.23099728987431564, "grad_norm": 1.9202210903167725, "learning_rate": 7.92814416171314e-06, "loss": 0.9486, "step": 12700 }, { "epoch": 0.23117917750413794, "grad_norm": 1.9742478132247925, "learning_rate": 7.926269009357011e-06, "loss": 1.0165, "step": 12710 }, { "epoch": 0.23136106513396024, "grad_norm": 1.9634573459625244, "learning_rate": 7.924393857000881e-06, "loss": 0.9938, "step": 12720 }, { "epoch": 0.23154295276378253, "grad_norm": 2.073284864425659, "learning_rate": 7.922518704644752e-06, "loss": 0.9964, "step": 12730 }, { "epoch": 0.23172484039360483, "grad_norm": 1.8703467845916748, "learning_rate": 7.920643552288624e-06, "loss": 0.9877, "step": 12740 }, { "epoch": 0.23190672802342713, "grad_norm": 1.9589588642120361, "learning_rate": 7.918768399932495e-06, "loss": 1.0136, "step": 12750 }, { "epoch": 0.23208861565324942, "grad_norm": 1.6830650568008423, "learning_rate": 7.916893247576366e-06, "loss": 0.9724, "step": 12760 }, { "epoch": 0.23227050328307172, "grad_norm": 2.015071153640747, "learning_rate": 7.915018095220238e-06, "loss": 0.977, "step": 12770 }, { "epoch": 0.23245239091289402, "grad_norm": 1.76607346534729, "learning_rate": 7.913142942864109e-06, "loss": 1.0174, "step": 12780 }, { "epoch": 0.23263427854271632, "grad_norm": 1.9155534505844116, "learning_rate": 7.911267790507979e-06, "loss": 0.9801, "step": 12790 }, { "epoch": 0.2328161661725386, "grad_norm": 1.9722321033477783, "learning_rate": 7.90939263815185e-06, "loss": 0.9641, "step": 12800 }, { "epoch": 0.2329980538023609, "grad_norm": 1.9565088748931885, "learning_rate": 7.907517485795721e-06, "loss": 1.001, "step": 12810 }, { "epoch": 0.2331799414321832, "grad_norm": 1.6000837087631226, "learning_rate": 7.905642333439593e-06, "loss": 0.9824, "step": 12820 }, { "epoch": 0.2333618290620055, "grad_norm": 1.9354525804519653, "learning_rate": 7.903767181083464e-06, "loss": 0.9996, "step": 12830 }, { "epoch": 0.2335437166918278, "grad_norm": 1.871201753616333, "learning_rate": 7.901892028727335e-06, "loss": 1.0001, "step": 12840 }, { "epoch": 0.23372560432165007, "grad_norm": 1.7988711595535278, "learning_rate": 7.900016876371207e-06, "loss": 0.9694, "step": 12850 }, { "epoch": 0.23390749195147237, "grad_norm": 1.7857673168182373, "learning_rate": 7.898141724015076e-06, "loss": 0.9754, "step": 12860 }, { "epoch": 0.23408937958129467, "grad_norm": 2.0402657985687256, "learning_rate": 7.896266571658948e-06, "loss": 1.0082, "step": 12870 }, { "epoch": 0.23427126721111696, "grad_norm": 1.5837410688400269, "learning_rate": 7.894391419302819e-06, "loss": 0.9401, "step": 12880 }, { "epoch": 0.23445315484093926, "grad_norm": 2.0028114318847656, "learning_rate": 7.89251626694669e-06, "loss": 0.9377, "step": 12890 }, { "epoch": 0.23463504247076156, "grad_norm": 2.018592357635498, "learning_rate": 7.890641114590562e-06, "loss": 0.9659, "step": 12900 }, { "epoch": 0.23481693010058385, "grad_norm": 1.6541537046432495, "learning_rate": 7.888765962234433e-06, "loss": 0.9311, "step": 12910 }, { "epoch": 0.23499881773040615, "grad_norm": 1.9219393730163574, "learning_rate": 7.886890809878304e-06, "loss": 1.0272, "step": 12920 }, { "epoch": 0.23518070536022845, "grad_norm": 1.9779529571533203, "learning_rate": 7.885015657522174e-06, "loss": 0.9265, "step": 12930 }, { "epoch": 0.23536259299005075, "grad_norm": 1.8813639879226685, "learning_rate": 7.883140505166045e-06, "loss": 1.023, "step": 12940 }, { "epoch": 0.23554448061987304, "grad_norm": 2.011969566345215, "learning_rate": 7.881265352809917e-06, "loss": 0.9948, "step": 12950 }, { "epoch": 0.23572636824969534, "grad_norm": 1.809483528137207, "learning_rate": 7.879390200453786e-06, "loss": 0.9652, "step": 12960 }, { "epoch": 0.23590825587951764, "grad_norm": 1.3860406875610352, "learning_rate": 7.877515048097658e-06, "loss": 0.9313, "step": 12970 }, { "epoch": 0.23609014350933993, "grad_norm": 1.8466994762420654, "learning_rate": 7.875639895741529e-06, "loss": 0.9778, "step": 12980 }, { "epoch": 0.23627203113916223, "grad_norm": 1.9828274250030518, "learning_rate": 7.8737647433854e-06, "loss": 0.9891, "step": 12990 }, { "epoch": 0.23645391876898453, "grad_norm": 2.1969563961029053, "learning_rate": 7.871889591029272e-06, "loss": 0.9933, "step": 13000 }, { "epoch": 0.23663580639880682, "grad_norm": 1.8729944229125977, "learning_rate": 7.870014438673143e-06, "loss": 0.9191, "step": 13010 }, { "epoch": 0.23681769402862912, "grad_norm": 1.8716546297073364, "learning_rate": 7.868139286317014e-06, "loss": 0.9488, "step": 13020 }, { "epoch": 0.23699958165845142, "grad_norm": 1.8012906312942505, "learning_rate": 7.866264133960884e-06, "loss": 0.995, "step": 13030 }, { "epoch": 0.23718146928827372, "grad_norm": 1.5061815977096558, "learning_rate": 7.864388981604755e-06, "loss": 0.9814, "step": 13040 }, { "epoch": 0.237363356918096, "grad_norm": 2.001615285873413, "learning_rate": 7.862513829248627e-06, "loss": 1.0141, "step": 13050 }, { "epoch": 0.23754524454791828, "grad_norm": 1.6915583610534668, "learning_rate": 7.860638676892498e-06, "loss": 0.9485, "step": 13060 }, { "epoch": 0.23772713217774058, "grad_norm": 1.8033761978149414, "learning_rate": 7.85876352453637e-06, "loss": 0.9256, "step": 13070 }, { "epoch": 0.23790901980756288, "grad_norm": 2.2278897762298584, "learning_rate": 7.85688837218024e-06, "loss": 1.0047, "step": 13080 }, { "epoch": 0.23809090743738517, "grad_norm": 1.8029603958129883, "learning_rate": 7.855013219824112e-06, "loss": 0.994, "step": 13090 }, { "epoch": 0.23827279506720747, "grad_norm": 2.0576272010803223, "learning_rate": 7.853138067467982e-06, "loss": 0.9624, "step": 13100 }, { "epoch": 0.23845468269702977, "grad_norm": 2.079256057739258, "learning_rate": 7.851262915111853e-06, "loss": 0.9764, "step": 13110 }, { "epoch": 0.23863657032685207, "grad_norm": 1.8635168075561523, "learning_rate": 7.849387762755725e-06, "loss": 0.9752, "step": 13120 }, { "epoch": 0.23881845795667436, "grad_norm": 1.9097139835357666, "learning_rate": 7.847512610399596e-06, "loss": 0.9802, "step": 13130 }, { "epoch": 0.23900034558649666, "grad_norm": 2.3948822021484375, "learning_rate": 7.845637458043467e-06, "loss": 0.9588, "step": 13140 }, { "epoch": 0.23918223321631896, "grad_norm": 2.248418092727661, "learning_rate": 7.843762305687339e-06, "loss": 0.9926, "step": 13150 }, { "epoch": 0.23936412084614125, "grad_norm": 1.8968299627304077, "learning_rate": 7.84188715333121e-06, "loss": 0.9695, "step": 13160 }, { "epoch": 0.23954600847596355, "grad_norm": 2.1176114082336426, "learning_rate": 7.84001200097508e-06, "loss": 0.9706, "step": 13170 }, { "epoch": 0.23972789610578585, "grad_norm": 1.993798851966858, "learning_rate": 7.838136848618951e-06, "loss": 1.0407, "step": 13180 }, { "epoch": 0.23990978373560815, "grad_norm": 1.7589892148971558, "learning_rate": 7.836261696262822e-06, "loss": 0.9604, "step": 13190 }, { "epoch": 0.24009167136543044, "grad_norm": 1.7941608428955078, "learning_rate": 7.834386543906694e-06, "loss": 0.9928, "step": 13200 }, { "epoch": 0.24027355899525274, "grad_norm": 1.8732818365097046, "learning_rate": 7.832511391550565e-06, "loss": 0.9436, "step": 13210 }, { "epoch": 0.24045544662507504, "grad_norm": 2.005150556564331, "learning_rate": 7.830636239194436e-06, "loss": 0.9928, "step": 13220 }, { "epoch": 0.24063733425489733, "grad_norm": 2.196274518966675, "learning_rate": 7.828761086838306e-06, "loss": 0.9562, "step": 13230 }, { "epoch": 0.24081922188471963, "grad_norm": 1.8464711904525757, "learning_rate": 7.826885934482177e-06, "loss": 0.9421, "step": 13240 }, { "epoch": 0.24100110951454193, "grad_norm": 2.0278499126434326, "learning_rate": 7.825010782126049e-06, "loss": 0.9692, "step": 13250 }, { "epoch": 0.24118299714436423, "grad_norm": 1.7955461740493774, "learning_rate": 7.823135629769918e-06, "loss": 0.9577, "step": 13260 }, { "epoch": 0.2413648847741865, "grad_norm": 1.6109719276428223, "learning_rate": 7.82126047741379e-06, "loss": 0.9192, "step": 13270 }, { "epoch": 0.2415467724040088, "grad_norm": 1.8968068361282349, "learning_rate": 7.819385325057661e-06, "loss": 1.014, "step": 13280 }, { "epoch": 0.2417286600338311, "grad_norm": 1.7367167472839355, "learning_rate": 7.817510172701532e-06, "loss": 0.9586, "step": 13290 }, { "epoch": 0.24191054766365339, "grad_norm": 2.258734941482544, "learning_rate": 7.815635020345404e-06, "loss": 0.9467, "step": 13300 }, { "epoch": 0.24209243529347568, "grad_norm": 1.9671674966812134, "learning_rate": 7.813759867989275e-06, "loss": 0.963, "step": 13310 }, { "epoch": 0.24227432292329798, "grad_norm": 1.7689012289047241, "learning_rate": 7.811884715633146e-06, "loss": 0.9972, "step": 13320 }, { "epoch": 0.24245621055312028, "grad_norm": 2.033461809158325, "learning_rate": 7.810009563277016e-06, "loss": 0.9756, "step": 13330 }, { "epoch": 0.24263809818294257, "grad_norm": 1.7762607336044312, "learning_rate": 7.808134410920887e-06, "loss": 0.9407, "step": 13340 }, { "epoch": 0.24281998581276487, "grad_norm": 1.889312505722046, "learning_rate": 7.806259258564759e-06, "loss": 1.01, "step": 13350 }, { "epoch": 0.24300187344258717, "grad_norm": 1.8854376077651978, "learning_rate": 7.80438410620863e-06, "loss": 0.9847, "step": 13360 }, { "epoch": 0.24318376107240947, "grad_norm": 2.122243642807007, "learning_rate": 7.802508953852501e-06, "loss": 1.015, "step": 13370 }, { "epoch": 0.24336564870223176, "grad_norm": 1.8480827808380127, "learning_rate": 7.800633801496373e-06, "loss": 0.9365, "step": 13380 }, { "epoch": 0.24354753633205406, "grad_norm": 1.733091950416565, "learning_rate": 7.798758649140244e-06, "loss": 1.0239, "step": 13390 }, { "epoch": 0.24372942396187636, "grad_norm": 1.927022933959961, "learning_rate": 7.796883496784114e-06, "loss": 0.9833, "step": 13400 }, { "epoch": 0.24391131159169865, "grad_norm": 1.9980850219726562, "learning_rate": 7.795008344427985e-06, "loss": 0.9602, "step": 13410 }, { "epoch": 0.24409319922152095, "grad_norm": 2.017420768737793, "learning_rate": 7.793133192071856e-06, "loss": 0.9656, "step": 13420 }, { "epoch": 0.24427508685134325, "grad_norm": 1.8918589353561401, "learning_rate": 7.791258039715728e-06, "loss": 0.9793, "step": 13430 }, { "epoch": 0.24445697448116555, "grad_norm": 1.9195411205291748, "learning_rate": 7.789382887359599e-06, "loss": 1.023, "step": 13440 }, { "epoch": 0.24463886211098784, "grad_norm": 1.6275864839553833, "learning_rate": 7.78750773500347e-06, "loss": 0.9586, "step": 13450 }, { "epoch": 0.24482074974081014, "grad_norm": 1.9340627193450928, "learning_rate": 7.785632582647342e-06, "loss": 0.9578, "step": 13460 }, { "epoch": 0.24500263737063244, "grad_norm": 1.877102255821228, "learning_rate": 7.783757430291211e-06, "loss": 0.9353, "step": 13470 }, { "epoch": 0.2451845250004547, "grad_norm": 1.868805170059204, "learning_rate": 7.781882277935083e-06, "loss": 0.9431, "step": 13480 }, { "epoch": 0.245366412630277, "grad_norm": 1.4692630767822266, "learning_rate": 7.780007125578954e-06, "loss": 0.9765, "step": 13490 }, { "epoch": 0.2455483002600993, "grad_norm": 1.5086939334869385, "learning_rate": 7.778131973222826e-06, "loss": 0.9174, "step": 13500 }, { "epoch": 0.2457301878899216, "grad_norm": 1.9399229288101196, "learning_rate": 7.776256820866695e-06, "loss": 1.0391, "step": 13510 }, { "epoch": 0.2459120755197439, "grad_norm": 1.9816889762878418, "learning_rate": 7.774381668510567e-06, "loss": 0.9518, "step": 13520 }, { "epoch": 0.2460939631495662, "grad_norm": 1.9061853885650635, "learning_rate": 7.772506516154438e-06, "loss": 0.9529, "step": 13530 }, { "epoch": 0.2462758507793885, "grad_norm": 1.8229578733444214, "learning_rate": 7.77063136379831e-06, "loss": 0.9953, "step": 13540 }, { "epoch": 0.24645773840921079, "grad_norm": 1.903462529182434, "learning_rate": 7.76875621144218e-06, "loss": 0.975, "step": 13550 }, { "epoch": 0.24663962603903308, "grad_norm": 1.8648415803909302, "learning_rate": 7.76688105908605e-06, "loss": 1.0395, "step": 13560 }, { "epoch": 0.24682151366885538, "grad_norm": 1.798852562904358, "learning_rate": 7.765005906729922e-06, "loss": 0.9824, "step": 13570 }, { "epoch": 0.24700340129867768, "grad_norm": 2.2597126960754395, "learning_rate": 7.763130754373793e-06, "loss": 1.0031, "step": 13580 }, { "epoch": 0.24718528892849997, "grad_norm": 2.145716428756714, "learning_rate": 7.761255602017664e-06, "loss": 1.0353, "step": 13590 }, { "epoch": 0.24736717655832227, "grad_norm": 1.630487084388733, "learning_rate": 7.759380449661536e-06, "loss": 1.009, "step": 13600 }, { "epoch": 0.24754906418814457, "grad_norm": 1.8363187313079834, "learning_rate": 7.757505297305407e-06, "loss": 0.9758, "step": 13610 }, { "epoch": 0.24773095181796687, "grad_norm": 1.6943910121917725, "learning_rate": 7.755630144949278e-06, "loss": 1.058, "step": 13620 }, { "epoch": 0.24791283944778916, "grad_norm": 1.7142800092697144, "learning_rate": 7.75375499259315e-06, "loss": 0.9589, "step": 13630 }, { "epoch": 0.24809472707761146, "grad_norm": 1.9603590965270996, "learning_rate": 7.75187984023702e-06, "loss": 0.9749, "step": 13640 }, { "epoch": 0.24827661470743376, "grad_norm": 1.7630757093429565, "learning_rate": 7.75000468788089e-06, "loss": 0.934, "step": 13650 }, { "epoch": 0.24845850233725605, "grad_norm": 1.6553025245666504, "learning_rate": 7.748129535524762e-06, "loss": 0.9016, "step": 13660 }, { "epoch": 0.24864038996707835, "grad_norm": 1.743787407875061, "learning_rate": 7.746254383168633e-06, "loss": 0.9439, "step": 13670 }, { "epoch": 0.24882227759690065, "grad_norm": 1.9637888669967651, "learning_rate": 7.744379230812505e-06, "loss": 0.9532, "step": 13680 }, { "epoch": 0.24900416522672292, "grad_norm": 1.8140965700149536, "learning_rate": 7.742504078456376e-06, "loss": 0.9962, "step": 13690 }, { "epoch": 0.24918605285654521, "grad_norm": 1.7508203983306885, "learning_rate": 7.740628926100247e-06, "loss": 1.001, "step": 13700 }, { "epoch": 0.2493679404863675, "grad_norm": 1.5710957050323486, "learning_rate": 7.738753773744117e-06, "loss": 0.9853, "step": 13710 }, { "epoch": 0.2495498281161898, "grad_norm": 3.076228618621826, "learning_rate": 7.736878621387988e-06, "loss": 1.0112, "step": 13720 }, { "epoch": 0.2497317157460121, "grad_norm": 1.8057913780212402, "learning_rate": 7.73500346903186e-06, "loss": 0.982, "step": 13730 }, { "epoch": 0.2499136033758344, "grad_norm": 1.646047830581665, "learning_rate": 7.733128316675731e-06, "loss": 0.9658, "step": 13740 }, { "epoch": 0.2500954910056567, "grad_norm": 1.7876461744308472, "learning_rate": 7.731253164319602e-06, "loss": 0.9508, "step": 13750 }, { "epoch": 0.250277378635479, "grad_norm": 1.916103720664978, "learning_rate": 7.729378011963474e-06, "loss": 0.9921, "step": 13760 }, { "epoch": 0.2504592662653013, "grad_norm": 1.6938910484313965, "learning_rate": 7.727502859607345e-06, "loss": 0.9752, "step": 13770 }, { "epoch": 0.2506411538951236, "grad_norm": 1.9247857332229614, "learning_rate": 7.725627707251215e-06, "loss": 0.9823, "step": 13780 }, { "epoch": 0.2508230415249459, "grad_norm": 1.9572477340698242, "learning_rate": 7.723752554895086e-06, "loss": 0.9849, "step": 13790 }, { "epoch": 0.2510049291547682, "grad_norm": 1.4677276611328125, "learning_rate": 7.721877402538956e-06, "loss": 0.9263, "step": 13800 }, { "epoch": 0.2511868167845905, "grad_norm": 1.5601004362106323, "learning_rate": 7.720002250182827e-06, "loss": 0.9655, "step": 13810 }, { "epoch": 0.2513687044144128, "grad_norm": 1.3916957378387451, "learning_rate": 7.718127097826698e-06, "loss": 0.9503, "step": 13820 }, { "epoch": 0.2515505920442351, "grad_norm": 1.526648998260498, "learning_rate": 7.71625194547057e-06, "loss": 0.9676, "step": 13830 }, { "epoch": 0.2517324796740574, "grad_norm": 1.7039613723754883, "learning_rate": 7.714376793114441e-06, "loss": 0.9197, "step": 13840 }, { "epoch": 0.25191436730387967, "grad_norm": 2.674018383026123, "learning_rate": 7.712501640758312e-06, "loss": 0.9377, "step": 13850 }, { "epoch": 0.25209625493370197, "grad_norm": 1.9297279119491577, "learning_rate": 7.710626488402184e-06, "loss": 0.9737, "step": 13860 }, { "epoch": 0.25227814256352427, "grad_norm": 1.8367328643798828, "learning_rate": 7.708751336046053e-06, "loss": 1.0243, "step": 13870 }, { "epoch": 0.25246003019334656, "grad_norm": 1.7327096462249756, "learning_rate": 7.706876183689925e-06, "loss": 0.9762, "step": 13880 }, { "epoch": 0.25264191782316886, "grad_norm": 1.9901331663131714, "learning_rate": 7.705001031333796e-06, "loss": 1.0009, "step": 13890 }, { "epoch": 0.25282380545299116, "grad_norm": 1.9485225677490234, "learning_rate": 7.703125878977668e-06, "loss": 0.9676, "step": 13900 }, { "epoch": 0.25300569308281345, "grad_norm": 1.802854299545288, "learning_rate": 7.701250726621539e-06, "loss": 0.9325, "step": 13910 }, { "epoch": 0.25318758071263575, "grad_norm": 2.04207706451416, "learning_rate": 7.69937557426541e-06, "loss": 1.0061, "step": 13920 }, { "epoch": 0.25336946834245805, "grad_norm": 2.1753017902374268, "learning_rate": 7.697500421909282e-06, "loss": 1.0192, "step": 13930 }, { "epoch": 0.25355135597228035, "grad_norm": 1.717956304550171, "learning_rate": 7.695625269553151e-06, "loss": 0.9693, "step": 13940 }, { "epoch": 0.25373324360210264, "grad_norm": 1.9084745645523071, "learning_rate": 7.693750117197023e-06, "loss": 0.9668, "step": 13950 }, { "epoch": 0.25391513123192494, "grad_norm": 1.9097872972488403, "learning_rate": 7.691874964840894e-06, "loss": 0.9306, "step": 13960 }, { "epoch": 0.25409701886174724, "grad_norm": 1.8033891916275024, "learning_rate": 7.689999812484765e-06, "loss": 0.9776, "step": 13970 }, { "epoch": 0.25427890649156953, "grad_norm": 1.8245972394943237, "learning_rate": 7.688124660128637e-06, "loss": 0.9796, "step": 13980 }, { "epoch": 0.25446079412139183, "grad_norm": 1.9585059881210327, "learning_rate": 7.686249507772508e-06, "loss": 1.0142, "step": 13990 }, { "epoch": 0.2546426817512141, "grad_norm": 2.0619916915893555, "learning_rate": 7.68437435541638e-06, "loss": 0.9861, "step": 14000 }, { "epoch": 0.25482456938103637, "grad_norm": 1.934654712677002, "learning_rate": 7.682499203060249e-06, "loss": 1.0008, "step": 14010 }, { "epoch": 0.25500645701085867, "grad_norm": 1.706204891204834, "learning_rate": 7.68062405070412e-06, "loss": 0.9777, "step": 14020 }, { "epoch": 0.25518834464068096, "grad_norm": 1.8890771865844727, "learning_rate": 7.678748898347992e-06, "loss": 0.9609, "step": 14030 }, { "epoch": 0.25537023227050326, "grad_norm": 1.6950863599777222, "learning_rate": 7.676873745991863e-06, "loss": 0.9758, "step": 14040 }, { "epoch": 0.25555211990032556, "grad_norm": 1.8255680799484253, "learning_rate": 7.674998593635734e-06, "loss": 0.9953, "step": 14050 }, { "epoch": 0.25573400753014786, "grad_norm": 1.4436343908309937, "learning_rate": 7.673123441279604e-06, "loss": 0.9516, "step": 14060 }, { "epoch": 0.25591589515997015, "grad_norm": 1.5000497102737427, "learning_rate": 7.671248288923475e-06, "loss": 0.9587, "step": 14070 }, { "epoch": 0.25609778278979245, "grad_norm": 1.7461103200912476, "learning_rate": 7.669373136567347e-06, "loss": 0.9908, "step": 14080 }, { "epoch": 0.25627967041961475, "grad_norm": 1.8010388612747192, "learning_rate": 7.667497984211218e-06, "loss": 0.9958, "step": 14090 }, { "epoch": 0.25646155804943704, "grad_norm": 1.9153543710708618, "learning_rate": 7.665622831855088e-06, "loss": 1.0091, "step": 14100 }, { "epoch": 0.25664344567925934, "grad_norm": 1.7615113258361816, "learning_rate": 7.663747679498959e-06, "loss": 1.0052, "step": 14110 }, { "epoch": 0.25682533330908164, "grad_norm": 1.9884371757507324, "learning_rate": 7.66187252714283e-06, "loss": 0.9513, "step": 14120 }, { "epoch": 0.25700722093890394, "grad_norm": 1.8211458921432495, "learning_rate": 7.659997374786702e-06, "loss": 0.9929, "step": 14130 }, { "epoch": 0.25718910856872623, "grad_norm": 1.8082714080810547, "learning_rate": 7.658122222430573e-06, "loss": 0.9105, "step": 14140 }, { "epoch": 0.25737099619854853, "grad_norm": 2.0523805618286133, "learning_rate": 7.656247070074444e-06, "loss": 0.9808, "step": 14150 }, { "epoch": 0.2575528838283708, "grad_norm": 1.8339991569519043, "learning_rate": 7.654371917718316e-06, "loss": 0.9322, "step": 14160 }, { "epoch": 0.2577347714581931, "grad_norm": 1.8894540071487427, "learning_rate": 7.652496765362185e-06, "loss": 1.04, "step": 14170 }, { "epoch": 0.2579166590880154, "grad_norm": 1.9678466320037842, "learning_rate": 7.650621613006057e-06, "loss": 0.9951, "step": 14180 }, { "epoch": 0.2580985467178377, "grad_norm": 2.076616048812866, "learning_rate": 7.648746460649928e-06, "loss": 1.0056, "step": 14190 }, { "epoch": 0.25828043434766, "grad_norm": 1.8141287565231323, "learning_rate": 7.6468713082938e-06, "loss": 0.9881, "step": 14200 }, { "epoch": 0.2584623219774823, "grad_norm": 1.667710304260254, "learning_rate": 7.64499615593767e-06, "loss": 0.9944, "step": 14210 }, { "epoch": 0.2586442096073046, "grad_norm": 1.9056310653686523, "learning_rate": 7.643121003581542e-06, "loss": 1.0054, "step": 14220 }, { "epoch": 0.2588260972371269, "grad_norm": 1.819300889968872, "learning_rate": 7.641245851225413e-06, "loss": 0.9279, "step": 14230 }, { "epoch": 0.2590079848669492, "grad_norm": 1.996396780014038, "learning_rate": 7.639370698869283e-06, "loss": 0.9421, "step": 14240 }, { "epoch": 0.2591898724967715, "grad_norm": 1.787083387374878, "learning_rate": 7.637495546513154e-06, "loss": 1.0017, "step": 14250 }, { "epoch": 0.2593717601265938, "grad_norm": 2.2667624950408936, "learning_rate": 7.635620394157026e-06, "loss": 1.0188, "step": 14260 }, { "epoch": 0.2595536477564161, "grad_norm": 1.7121225595474243, "learning_rate": 7.633745241800897e-06, "loss": 0.9781, "step": 14270 }, { "epoch": 0.2597355353862384, "grad_norm": 1.8308659791946411, "learning_rate": 7.631870089444768e-06, "loss": 1.0336, "step": 14280 }, { "epoch": 0.2599174230160607, "grad_norm": 1.9134451150894165, "learning_rate": 7.62999493708864e-06, "loss": 1.0007, "step": 14290 }, { "epoch": 0.260099310645883, "grad_norm": 1.882537603378296, "learning_rate": 7.62811978473251e-06, "loss": 0.9867, "step": 14300 }, { "epoch": 0.2602811982757053, "grad_norm": 1.7899043560028076, "learning_rate": 7.626244632376382e-06, "loss": 0.9372, "step": 14310 }, { "epoch": 0.2604630859055276, "grad_norm": 1.9811822175979614, "learning_rate": 7.624369480020253e-06, "loss": 1.027, "step": 14320 }, { "epoch": 0.2606449735353499, "grad_norm": 1.8965691328048706, "learning_rate": 7.622494327664123e-06, "loss": 0.9668, "step": 14330 }, { "epoch": 0.2608268611651722, "grad_norm": 1.5549839735031128, "learning_rate": 7.620619175307994e-06, "loss": 0.9943, "step": 14340 }, { "epoch": 0.26100874879499447, "grad_norm": 1.8324050903320312, "learning_rate": 7.618744022951865e-06, "loss": 0.9361, "step": 14350 }, { "epoch": 0.26119063642481677, "grad_norm": 1.6670117378234863, "learning_rate": 7.616868870595736e-06, "loss": 0.944, "step": 14360 }, { "epoch": 0.26137252405463907, "grad_norm": 1.7637135982513428, "learning_rate": 7.614993718239607e-06, "loss": 0.984, "step": 14370 }, { "epoch": 0.26155441168446136, "grad_norm": 1.6880879402160645, "learning_rate": 7.6131185658834786e-06, "loss": 0.9585, "step": 14380 }, { "epoch": 0.26173629931428366, "grad_norm": 1.7147941589355469, "learning_rate": 7.611243413527349e-06, "loss": 0.9647, "step": 14390 }, { "epoch": 0.26191818694410596, "grad_norm": 1.7164943218231201, "learning_rate": 7.60936826117122e-06, "loss": 0.9541, "step": 14400 }, { "epoch": 0.26210007457392825, "grad_norm": 1.8968594074249268, "learning_rate": 7.607493108815092e-06, "loss": 1.0111, "step": 14410 }, { "epoch": 0.2622819622037505, "grad_norm": 1.7109686136245728, "learning_rate": 7.605617956458963e-06, "loss": 0.9436, "step": 14420 }, { "epoch": 0.2624638498335728, "grad_norm": 1.6927847862243652, "learning_rate": 7.603742804102834e-06, "loss": 0.9473, "step": 14430 }, { "epoch": 0.2626457374633951, "grad_norm": 2.112290382385254, "learning_rate": 7.601867651746705e-06, "loss": 0.9969, "step": 14440 }, { "epoch": 0.2628276250932174, "grad_norm": 1.7495760917663574, "learning_rate": 7.599992499390576e-06, "loss": 1.044, "step": 14450 }, { "epoch": 0.2630095127230397, "grad_norm": 1.8300254344940186, "learning_rate": 7.598117347034447e-06, "loss": 0.9874, "step": 14460 }, { "epoch": 0.263191400352862, "grad_norm": 2.054642677307129, "learning_rate": 7.596242194678318e-06, "loss": 0.9955, "step": 14470 }, { "epoch": 0.2633732879826843, "grad_norm": 1.5287891626358032, "learning_rate": 7.5943670423221895e-06, "loss": 0.967, "step": 14480 }, { "epoch": 0.2635551756125066, "grad_norm": 1.7404342889785767, "learning_rate": 7.592491889966061e-06, "loss": 0.9852, "step": 14490 }, { "epoch": 0.2637370632423289, "grad_norm": 2.0065712928771973, "learning_rate": 7.590616737609931e-06, "loss": 0.9254, "step": 14500 }, { "epoch": 0.26391895087215117, "grad_norm": 1.9136399030685425, "learning_rate": 7.588741585253803e-06, "loss": 0.9616, "step": 14510 }, { "epoch": 0.26410083850197347, "grad_norm": 1.9669866561889648, "learning_rate": 7.586866432897674e-06, "loss": 1.0085, "step": 14520 }, { "epoch": 0.26428272613179576, "grad_norm": 2.050832509994507, "learning_rate": 7.5849912805415445e-06, "loss": 0.9836, "step": 14530 }, { "epoch": 0.26446461376161806, "grad_norm": 1.8162351846694946, "learning_rate": 7.583116128185416e-06, "loss": 0.9754, "step": 14540 }, { "epoch": 0.26464650139144036, "grad_norm": 2.409945011138916, "learning_rate": 7.581240975829287e-06, "loss": 0.9636, "step": 14550 }, { "epoch": 0.26482838902126266, "grad_norm": 1.9509608745574951, "learning_rate": 7.5793658234731586e-06, "loss": 0.9876, "step": 14560 }, { "epoch": 0.26501027665108495, "grad_norm": 1.8925732374191284, "learning_rate": 7.577490671117029e-06, "loss": 1.0204, "step": 14570 }, { "epoch": 0.26519216428090725, "grad_norm": 1.6911696195602417, "learning_rate": 7.5756155187609e-06, "loss": 0.9316, "step": 14580 }, { "epoch": 0.26537405191072955, "grad_norm": 1.7440366744995117, "learning_rate": 7.573740366404772e-06, "loss": 0.9776, "step": 14590 }, { "epoch": 0.26555593954055184, "grad_norm": 2.005509853363037, "learning_rate": 7.571865214048642e-06, "loss": 0.9756, "step": 14600 }, { "epoch": 0.26573782717037414, "grad_norm": 1.7425146102905273, "learning_rate": 7.569990061692513e-06, "loss": 0.9594, "step": 14610 }, { "epoch": 0.26591971480019644, "grad_norm": 1.6862713098526, "learning_rate": 7.568114909336384e-06, "loss": 0.992, "step": 14620 }, { "epoch": 0.26610160243001874, "grad_norm": 1.500666856765747, "learning_rate": 7.566239756980255e-06, "loss": 0.9737, "step": 14630 }, { "epoch": 0.26628349005984103, "grad_norm": 2.1204779148101807, "learning_rate": 7.564364604624126e-06, "loss": 1.0061, "step": 14640 }, { "epoch": 0.26646537768966333, "grad_norm": 1.924917459487915, "learning_rate": 7.562489452267997e-06, "loss": 1.0022, "step": 14650 }, { "epoch": 0.2666472653194856, "grad_norm": 1.8320016860961914, "learning_rate": 7.560614299911868e-06, "loss": 1.0449, "step": 14660 }, { "epoch": 0.2668291529493079, "grad_norm": 1.7578608989715576, "learning_rate": 7.558739147555739e-06, "loss": 0.9219, "step": 14670 }, { "epoch": 0.2670110405791302, "grad_norm": 1.6839576959609985, "learning_rate": 7.5568639951996105e-06, "loss": 0.9763, "step": 14680 }, { "epoch": 0.2671929282089525, "grad_norm": 2.0553314685821533, "learning_rate": 7.554988842843482e-06, "loss": 0.9815, "step": 14690 }, { "epoch": 0.2673748158387748, "grad_norm": 1.2649132013320923, "learning_rate": 7.553113690487352e-06, "loss": 0.9388, "step": 14700 }, { "epoch": 0.2675567034685971, "grad_norm": 1.9819074869155884, "learning_rate": 7.551238538131224e-06, "loss": 0.9567, "step": 14710 }, { "epoch": 0.2677385910984194, "grad_norm": 2.1308209896087646, "learning_rate": 7.549363385775095e-06, "loss": 0.9468, "step": 14720 }, { "epoch": 0.2679204787282417, "grad_norm": 2.0336458683013916, "learning_rate": 7.5474882334189655e-06, "loss": 1.0138, "step": 14730 }, { "epoch": 0.268102366358064, "grad_norm": 1.8846548795700073, "learning_rate": 7.545613081062837e-06, "loss": 0.9729, "step": 14740 }, { "epoch": 0.2682842539878863, "grad_norm": 2.151585578918457, "learning_rate": 7.543737928706708e-06, "loss": 1.01, "step": 14750 }, { "epoch": 0.2684661416177086, "grad_norm": 1.935870885848999, "learning_rate": 7.5418627763505795e-06, "loss": 0.992, "step": 14760 }, { "epoch": 0.2686480292475309, "grad_norm": 1.6901012659072876, "learning_rate": 7.53998762399445e-06, "loss": 1.0018, "step": 14770 }, { "epoch": 0.2688299168773532, "grad_norm": 1.827883005142212, "learning_rate": 7.538112471638321e-06, "loss": 0.9227, "step": 14780 }, { "epoch": 0.2690118045071755, "grad_norm": 1.9543110132217407, "learning_rate": 7.536237319282193e-06, "loss": 0.9702, "step": 14790 }, { "epoch": 0.2691936921369978, "grad_norm": 1.8198139667510986, "learning_rate": 7.534362166926063e-06, "loss": 0.9418, "step": 14800 }, { "epoch": 0.2693755797668201, "grad_norm": 1.8370306491851807, "learning_rate": 7.532487014569935e-06, "loss": 0.9849, "step": 14810 }, { "epoch": 0.2695574673966424, "grad_norm": 1.8377439975738525, "learning_rate": 7.530611862213806e-06, "loss": 0.9813, "step": 14820 }, { "epoch": 0.2697393550264647, "grad_norm": 1.8755351305007935, "learning_rate": 7.528736709857677e-06, "loss": 0.9886, "step": 14830 }, { "epoch": 0.2699212426562869, "grad_norm": 1.6415935754776, "learning_rate": 7.526861557501548e-06, "loss": 0.9653, "step": 14840 }, { "epoch": 0.2701031302861092, "grad_norm": 1.8105888366699219, "learning_rate": 7.524986405145419e-06, "loss": 1.0047, "step": 14850 }, { "epoch": 0.2702850179159315, "grad_norm": 1.6765233278274536, "learning_rate": 7.5231112527892905e-06, "loss": 0.9971, "step": 14860 }, { "epoch": 0.2704669055457538, "grad_norm": 1.7640339136123657, "learning_rate": 7.521236100433161e-06, "loss": 0.9505, "step": 14870 }, { "epoch": 0.2706487931755761, "grad_norm": 1.9182533025741577, "learning_rate": 7.5193609480770315e-06, "loss": 0.9728, "step": 14880 }, { "epoch": 0.2708306808053984, "grad_norm": 1.4863581657409668, "learning_rate": 7.517485795720903e-06, "loss": 0.9375, "step": 14890 }, { "epoch": 0.2710125684352207, "grad_norm": 1.8148013353347778, "learning_rate": 7.515610643364773e-06, "loss": 0.9626, "step": 14900 }, { "epoch": 0.271194456065043, "grad_norm": 1.6603494882583618, "learning_rate": 7.513735491008645e-06, "loss": 0.9715, "step": 14910 }, { "epoch": 0.2713763436948653, "grad_norm": 1.7779719829559326, "learning_rate": 7.511860338652516e-06, "loss": 0.9601, "step": 14920 }, { "epoch": 0.2715582313246876, "grad_norm": 1.7564115524291992, "learning_rate": 7.5099851862963865e-06, "loss": 0.9646, "step": 14930 }, { "epoch": 0.2717401189545099, "grad_norm": 1.3310190439224243, "learning_rate": 7.508110033940258e-06, "loss": 0.9039, "step": 14940 }, { "epoch": 0.2719220065843322, "grad_norm": 1.9702794551849365, "learning_rate": 7.506234881584129e-06, "loss": 1.0121, "step": 14950 }, { "epoch": 0.2721038942141545, "grad_norm": 1.6510754823684692, "learning_rate": 7.5043597292280005e-06, "loss": 0.946, "step": 14960 }, { "epoch": 0.2722857818439768, "grad_norm": 1.9840177297592163, "learning_rate": 7.502484576871871e-06, "loss": 0.9934, "step": 14970 }, { "epoch": 0.2724676694737991, "grad_norm": 2.0782999992370605, "learning_rate": 7.500609424515742e-06, "loss": 0.9676, "step": 14980 }, { "epoch": 0.2726495571036214, "grad_norm": 1.4374816417694092, "learning_rate": 7.498734272159614e-06, "loss": 0.9601, "step": 14990 }, { "epoch": 0.2728314447334437, "grad_norm": 1.8038861751556396, "learning_rate": 7.496859119803484e-06, "loss": 1.0033, "step": 15000 }, { "epoch": 0.27301333236326597, "grad_norm": 1.4830902814865112, "learning_rate": 7.4949839674473556e-06, "loss": 0.9342, "step": 15010 }, { "epoch": 0.27319521999308827, "grad_norm": 1.7690520286560059, "learning_rate": 7.493108815091227e-06, "loss": 0.9492, "step": 15020 }, { "epoch": 0.27337710762291056, "grad_norm": 1.6788280010223389, "learning_rate": 7.491233662735098e-06, "loss": 0.9494, "step": 15030 }, { "epoch": 0.27355899525273286, "grad_norm": 1.7570347785949707, "learning_rate": 7.489358510378969e-06, "loss": 0.9675, "step": 15040 }, { "epoch": 0.27374088288255516, "grad_norm": 1.7836564779281616, "learning_rate": 7.48748335802284e-06, "loss": 0.9552, "step": 15050 }, { "epoch": 0.27392277051237746, "grad_norm": 1.861364722251892, "learning_rate": 7.4856082056667115e-06, "loss": 0.9996, "step": 15060 }, { "epoch": 0.27410465814219975, "grad_norm": 1.8897819519042969, "learning_rate": 7.483733053310582e-06, "loss": 1.0047, "step": 15070 }, { "epoch": 0.27428654577202205, "grad_norm": 1.7170814275741577, "learning_rate": 7.481857900954453e-06, "loss": 0.9739, "step": 15080 }, { "epoch": 0.27446843340184435, "grad_norm": 1.7191603183746338, "learning_rate": 7.479982748598325e-06, "loss": 0.9967, "step": 15090 }, { "epoch": 0.27465032103166664, "grad_norm": 1.7801636457443237, "learning_rate": 7.478107596242196e-06, "loss": 0.9753, "step": 15100 }, { "epoch": 0.27483220866148894, "grad_norm": 1.8506035804748535, "learning_rate": 7.4762324438860665e-06, "loss": 0.9439, "step": 15110 }, { "epoch": 0.27501409629131124, "grad_norm": 1.922729253768921, "learning_rate": 7.474357291529938e-06, "loss": 1.0047, "step": 15120 }, { "epoch": 0.27519598392113354, "grad_norm": 1.852864384651184, "learning_rate": 7.472482139173809e-06, "loss": 0.9714, "step": 15130 }, { "epoch": 0.27537787155095583, "grad_norm": 1.845689296722412, "learning_rate": 7.47060698681768e-06, "loss": 1.023, "step": 15140 }, { "epoch": 0.27555975918077813, "grad_norm": 2.032780647277832, "learning_rate": 7.468731834461551e-06, "loss": 0.9543, "step": 15150 }, { "epoch": 0.2757416468106004, "grad_norm": 1.5877729654312134, "learning_rate": 7.4668566821054215e-06, "loss": 0.971, "step": 15160 }, { "epoch": 0.2759235344404227, "grad_norm": 1.5429811477661133, "learning_rate": 7.464981529749292e-06, "loss": 0.9662, "step": 15170 }, { "epoch": 0.276105422070245, "grad_norm": 1.7685881853103638, "learning_rate": 7.463106377393163e-06, "loss": 0.9658, "step": 15180 }, { "epoch": 0.2762873097000673, "grad_norm": 2.0490689277648926, "learning_rate": 7.461231225037035e-06, "loss": 0.9909, "step": 15190 }, { "epoch": 0.2764691973298896, "grad_norm": 1.987021565437317, "learning_rate": 7.459356072680905e-06, "loss": 0.9182, "step": 15200 }, { "epoch": 0.2766510849597119, "grad_norm": 1.617103099822998, "learning_rate": 7.4574809203247766e-06, "loss": 0.9774, "step": 15210 }, { "epoch": 0.2768329725895342, "grad_norm": 1.7891179323196411, "learning_rate": 7.455605767968648e-06, "loss": 0.964, "step": 15220 }, { "epoch": 0.2770148602193565, "grad_norm": 1.8492369651794434, "learning_rate": 7.453730615612519e-06, "loss": 1.0068, "step": 15230 }, { "epoch": 0.2771967478491788, "grad_norm": 1.8903502225875854, "learning_rate": 7.45185546325639e-06, "loss": 0.9832, "step": 15240 }, { "epoch": 0.27737863547900105, "grad_norm": 1.7082234621047974, "learning_rate": 7.449980310900261e-06, "loss": 0.9505, "step": 15250 }, { "epoch": 0.27756052310882334, "grad_norm": 1.8289300203323364, "learning_rate": 7.4481051585441324e-06, "loss": 0.9482, "step": 15260 }, { "epoch": 0.27774241073864564, "grad_norm": 1.7719106674194336, "learning_rate": 7.446230006188003e-06, "loss": 0.9676, "step": 15270 }, { "epoch": 0.27792429836846794, "grad_norm": 1.9199049472808838, "learning_rate": 7.444354853831874e-06, "loss": 1.0154, "step": 15280 }, { "epoch": 0.27810618599829023, "grad_norm": 1.3415733575820923, "learning_rate": 7.442479701475746e-06, "loss": 0.9631, "step": 15290 }, { "epoch": 0.27828807362811253, "grad_norm": 2.1288068294525146, "learning_rate": 7.440604549119617e-06, "loss": 0.968, "step": 15300 }, { "epoch": 0.27846996125793483, "grad_norm": 2.0070481300354004, "learning_rate": 7.4387293967634875e-06, "loss": 0.9763, "step": 15310 }, { "epoch": 0.2786518488877571, "grad_norm": 2.0138168334960938, "learning_rate": 7.436854244407359e-06, "loss": 0.9732, "step": 15320 }, { "epoch": 0.2788337365175794, "grad_norm": 1.7294713258743286, "learning_rate": 7.43497909205123e-06, "loss": 0.9328, "step": 15330 }, { "epoch": 0.2790156241474017, "grad_norm": 1.9535155296325684, "learning_rate": 7.433103939695101e-06, "loss": 0.9765, "step": 15340 }, { "epoch": 0.279197511777224, "grad_norm": 1.7219698429107666, "learning_rate": 7.431228787338972e-06, "loss": 0.9815, "step": 15350 }, { "epoch": 0.2793793994070463, "grad_norm": 1.6790530681610107, "learning_rate": 7.429353634982843e-06, "loss": 0.9523, "step": 15360 }, { "epoch": 0.2795612870368686, "grad_norm": 1.802009105682373, "learning_rate": 7.427478482626715e-06, "loss": 0.9673, "step": 15370 }, { "epoch": 0.2797431746666909, "grad_norm": 1.9603451490402222, "learning_rate": 7.425603330270585e-06, "loss": 1.004, "step": 15380 }, { "epoch": 0.2799250622965132, "grad_norm": 1.8540652990341187, "learning_rate": 7.4237281779144566e-06, "loss": 0.9356, "step": 15390 }, { "epoch": 0.2801069499263355, "grad_norm": 2.233053207397461, "learning_rate": 7.421853025558328e-06, "loss": 0.9745, "step": 15400 }, { "epoch": 0.2802888375561578, "grad_norm": 1.692794680595398, "learning_rate": 7.419977873202198e-06, "loss": 0.9609, "step": 15410 }, { "epoch": 0.2804707251859801, "grad_norm": 1.9265594482421875, "learning_rate": 7.41810272084607e-06, "loss": 0.9233, "step": 15420 }, { "epoch": 0.2806526128158024, "grad_norm": 1.741700530052185, "learning_rate": 7.416227568489941e-06, "loss": 0.9573, "step": 15430 }, { "epoch": 0.2808345004456247, "grad_norm": 2.4362001419067383, "learning_rate": 7.414352416133811e-06, "loss": 0.9996, "step": 15440 }, { "epoch": 0.281016388075447, "grad_norm": 1.4864639043807983, "learning_rate": 7.412477263777682e-06, "loss": 0.9341, "step": 15450 }, { "epoch": 0.2811982757052693, "grad_norm": 1.8208688497543335, "learning_rate": 7.4106021114215534e-06, "loss": 0.9298, "step": 15460 }, { "epoch": 0.2813801633350916, "grad_norm": 2.6285853385925293, "learning_rate": 7.408726959065424e-06, "loss": 0.9535, "step": 15470 }, { "epoch": 0.2815620509649139, "grad_norm": 1.6491061449050903, "learning_rate": 7.406851806709295e-06, "loss": 0.9807, "step": 15480 }, { "epoch": 0.2817439385947362, "grad_norm": 1.9990066289901733, "learning_rate": 7.404976654353167e-06, "loss": 0.9685, "step": 15490 }, { "epoch": 0.2819258262245585, "grad_norm": 1.6802910566329956, "learning_rate": 7.403101501997038e-06, "loss": 0.9337, "step": 15500 }, { "epoch": 0.28210771385438077, "grad_norm": 1.5782105922698975, "learning_rate": 7.4012263496409085e-06, "loss": 0.9192, "step": 15510 }, { "epoch": 0.28228960148420307, "grad_norm": 1.9223039150238037, "learning_rate": 7.39935119728478e-06, "loss": 1.0411, "step": 15520 }, { "epoch": 0.28247148911402536, "grad_norm": 1.7367079257965088, "learning_rate": 7.397476044928651e-06, "loss": 0.9946, "step": 15530 }, { "epoch": 0.28265337674384766, "grad_norm": 1.8746544122695923, "learning_rate": 7.395600892572522e-06, "loss": 0.969, "step": 15540 }, { "epoch": 0.28283526437366996, "grad_norm": 1.9507691860198975, "learning_rate": 7.393725740216393e-06, "loss": 0.9377, "step": 15550 }, { "epoch": 0.28301715200349226, "grad_norm": 1.9424742460250854, "learning_rate": 7.391850587860264e-06, "loss": 0.9603, "step": 15560 }, { "epoch": 0.28319903963331455, "grad_norm": 1.938085913658142, "learning_rate": 7.389975435504136e-06, "loss": 0.9948, "step": 15570 }, { "epoch": 0.28338092726313685, "grad_norm": 1.9010441303253174, "learning_rate": 7.388100283148006e-06, "loss": 0.9874, "step": 15580 }, { "epoch": 0.28356281489295915, "grad_norm": 1.7152457237243652, "learning_rate": 7.3862251307918776e-06, "loss": 0.9202, "step": 15590 }, { "epoch": 0.28374470252278144, "grad_norm": 1.934708833694458, "learning_rate": 7.384349978435749e-06, "loss": 0.9469, "step": 15600 }, { "epoch": 0.28392659015260374, "grad_norm": 1.815783143043518, "learning_rate": 7.382474826079619e-06, "loss": 0.9837, "step": 15610 }, { "epoch": 0.28410847778242604, "grad_norm": 1.5728338956832886, "learning_rate": 7.380599673723491e-06, "loss": 0.9218, "step": 15620 }, { "epoch": 0.28429036541224834, "grad_norm": 1.7345935106277466, "learning_rate": 7.378724521367362e-06, "loss": 0.9007, "step": 15630 }, { "epoch": 0.28447225304207063, "grad_norm": 1.8177725076675415, "learning_rate": 7.3768493690112334e-06, "loss": 0.9373, "step": 15640 }, { "epoch": 0.28465414067189293, "grad_norm": 1.869949460029602, "learning_rate": 7.374974216655104e-06, "loss": 0.9755, "step": 15650 }, { "epoch": 0.2848360283017152, "grad_norm": 1.5880954265594482, "learning_rate": 7.373099064298975e-06, "loss": 0.9764, "step": 15660 }, { "epoch": 0.28501791593153747, "grad_norm": 2.083223819732666, "learning_rate": 7.371223911942847e-06, "loss": 0.9785, "step": 15670 }, { "epoch": 0.28519980356135977, "grad_norm": 1.7031662464141846, "learning_rate": 7.369348759586717e-06, "loss": 0.9157, "step": 15680 }, { "epoch": 0.28538169119118206, "grad_norm": 2.011521577835083, "learning_rate": 7.3674736072305885e-06, "loss": 1.0207, "step": 15690 }, { "epoch": 0.28556357882100436, "grad_norm": 1.7368990182876587, "learning_rate": 7.36559845487446e-06, "loss": 1.0107, "step": 15700 }, { "epoch": 0.28574546645082666, "grad_norm": 1.8622167110443115, "learning_rate": 7.3637233025183295e-06, "loss": 0.978, "step": 15710 }, { "epoch": 0.28592735408064895, "grad_norm": 2.2967188358306885, "learning_rate": 7.361848150162201e-06, "loss": 1.0154, "step": 15720 }, { "epoch": 0.28610924171047125, "grad_norm": 2.104937791824341, "learning_rate": 7.359972997806072e-06, "loss": 0.9879, "step": 15730 }, { "epoch": 0.28629112934029355, "grad_norm": 1.853859543800354, "learning_rate": 7.358097845449943e-06, "loss": 0.92, "step": 15740 }, { "epoch": 0.28647301697011585, "grad_norm": 1.7890266180038452, "learning_rate": 7.356222693093814e-06, "loss": 0.9551, "step": 15750 }, { "epoch": 0.28665490459993814, "grad_norm": 1.7031465768814087, "learning_rate": 7.354347540737685e-06, "loss": 0.9332, "step": 15760 }, { "epoch": 0.28683679222976044, "grad_norm": 1.8133784532546997, "learning_rate": 7.352472388381557e-06, "loss": 0.9778, "step": 15770 }, { "epoch": 0.28701867985958274, "grad_norm": 1.892357587814331, "learning_rate": 7.350597236025427e-06, "loss": 0.9393, "step": 15780 }, { "epoch": 0.28720056748940503, "grad_norm": 1.6976877450942993, "learning_rate": 7.3487220836692985e-06, "loss": 0.9569, "step": 15790 }, { "epoch": 0.28738245511922733, "grad_norm": 1.8674887418746948, "learning_rate": 7.34684693131317e-06, "loss": 0.9626, "step": 15800 }, { "epoch": 0.28756434274904963, "grad_norm": 1.7864760160446167, "learning_rate": 7.34497177895704e-06, "loss": 0.9804, "step": 15810 }, { "epoch": 0.2877462303788719, "grad_norm": 1.8293278217315674, "learning_rate": 7.343096626600912e-06, "loss": 0.9654, "step": 15820 }, { "epoch": 0.2879281180086942, "grad_norm": 1.704266905784607, "learning_rate": 7.341221474244783e-06, "loss": 0.943, "step": 15830 }, { "epoch": 0.2881100056385165, "grad_norm": 1.6457997560501099, "learning_rate": 7.3393463218886544e-06, "loss": 0.999, "step": 15840 }, { "epoch": 0.2882918932683388, "grad_norm": 1.886089563369751, "learning_rate": 7.337471169532525e-06, "loss": 0.957, "step": 15850 }, { "epoch": 0.2884737808981611, "grad_norm": 1.8241212368011475, "learning_rate": 7.335596017176396e-06, "loss": 0.9138, "step": 15860 }, { "epoch": 0.2886556685279834, "grad_norm": 1.7088687419891357, "learning_rate": 7.333720864820268e-06, "loss": 0.9781, "step": 15870 }, { "epoch": 0.2888375561578057, "grad_norm": 1.8628582954406738, "learning_rate": 7.331845712464138e-06, "loss": 0.9739, "step": 15880 }, { "epoch": 0.289019443787628, "grad_norm": 2.0835862159729004, "learning_rate": 7.3299705601080095e-06, "loss": 0.9389, "step": 15890 }, { "epoch": 0.2892013314174503, "grad_norm": 1.6511784791946411, "learning_rate": 7.328095407751881e-06, "loss": 0.9464, "step": 15900 }, { "epoch": 0.2893832190472726, "grad_norm": 1.8004348278045654, "learning_rate": 7.326220255395752e-06, "loss": 0.9691, "step": 15910 }, { "epoch": 0.2895651066770949, "grad_norm": 1.7398090362548828, "learning_rate": 7.324345103039623e-06, "loss": 0.9685, "step": 15920 }, { "epoch": 0.2897469943069172, "grad_norm": 1.8331530094146729, "learning_rate": 7.322469950683494e-06, "loss": 0.9154, "step": 15930 }, { "epoch": 0.2899288819367395, "grad_norm": 1.9380791187286377, "learning_rate": 7.320594798327365e-06, "loss": 0.8892, "step": 15940 }, { "epoch": 0.2901107695665618, "grad_norm": 2.0279407501220703, "learning_rate": 7.318719645971236e-06, "loss": 1.0314, "step": 15950 }, { "epoch": 0.2902926571963841, "grad_norm": 1.9823241233825684, "learning_rate": 7.316844493615107e-06, "loss": 0.9689, "step": 15960 }, { "epoch": 0.2904745448262064, "grad_norm": 1.7714208364486694, "learning_rate": 7.3149693412589785e-06, "loss": 0.9642, "step": 15970 }, { "epoch": 0.2906564324560287, "grad_norm": 1.8479591608047485, "learning_rate": 7.31309418890285e-06, "loss": 1.0045, "step": 15980 }, { "epoch": 0.290838320085851, "grad_norm": 1.6393401622772217, "learning_rate": 7.3112190365467195e-06, "loss": 0.9934, "step": 15990 }, { "epoch": 0.2910202077156733, "grad_norm": 1.6877065896987915, "learning_rate": 7.309343884190591e-06, "loss": 0.958, "step": 16000 }, { "epoch": 0.29120209534549557, "grad_norm": 2.3334555625915527, "learning_rate": 7.307468731834461e-06, "loss": 1.0032, "step": 16010 }, { "epoch": 0.29138398297531787, "grad_norm": 1.9124318361282349, "learning_rate": 7.305593579478333e-06, "loss": 0.9192, "step": 16020 }, { "epoch": 0.29156587060514016, "grad_norm": 1.6890385150909424, "learning_rate": 7.303718427122204e-06, "loss": 0.9296, "step": 16030 }, { "epoch": 0.29174775823496246, "grad_norm": 1.9010239839553833, "learning_rate": 7.301843274766075e-06, "loss": 0.9579, "step": 16040 }, { "epoch": 0.29192964586478476, "grad_norm": 1.673276424407959, "learning_rate": 7.299968122409946e-06, "loss": 0.9646, "step": 16050 }, { "epoch": 0.29211153349460706, "grad_norm": 1.9287877082824707, "learning_rate": 7.298092970053817e-06, "loss": 0.9881, "step": 16060 }, { "epoch": 0.29229342112442935, "grad_norm": 1.8797739744186401, "learning_rate": 7.296217817697689e-06, "loss": 1.0082, "step": 16070 }, { "epoch": 0.29247530875425165, "grad_norm": 1.3847905397415161, "learning_rate": 7.294342665341559e-06, "loss": 0.9211, "step": 16080 }, { "epoch": 0.2926571963840739, "grad_norm": 1.6554455757141113, "learning_rate": 7.2924675129854305e-06, "loss": 0.9407, "step": 16090 }, { "epoch": 0.2928390840138962, "grad_norm": 1.4946279525756836, "learning_rate": 7.290592360629302e-06, "loss": 0.918, "step": 16100 }, { "epoch": 0.2930209716437185, "grad_norm": 1.8290605545043945, "learning_rate": 7.288717208273173e-06, "loss": 1.0208, "step": 16110 }, { "epoch": 0.2932028592735408, "grad_norm": 1.7519444227218628, "learning_rate": 7.286842055917044e-06, "loss": 0.9356, "step": 16120 }, { "epoch": 0.2933847469033631, "grad_norm": 1.5251578092575073, "learning_rate": 7.284966903560915e-06, "loss": 0.9284, "step": 16130 }, { "epoch": 0.2935666345331854, "grad_norm": 1.4722367525100708, "learning_rate": 7.283091751204786e-06, "loss": 0.9726, "step": 16140 }, { "epoch": 0.2937485221630077, "grad_norm": 1.9106196165084839, "learning_rate": 7.281216598848657e-06, "loss": 0.9911, "step": 16150 }, { "epoch": 0.29393040979282997, "grad_norm": 1.824966549873352, "learning_rate": 7.279341446492528e-06, "loss": 0.9472, "step": 16160 }, { "epoch": 0.29411229742265227, "grad_norm": 2.0068728923797607, "learning_rate": 7.2774662941363995e-06, "loss": 1.0127, "step": 16170 }, { "epoch": 0.29429418505247457, "grad_norm": 1.7025574445724487, "learning_rate": 7.275591141780271e-06, "loss": 0.9911, "step": 16180 }, { "epoch": 0.29447607268229686, "grad_norm": 1.8473609685897827, "learning_rate": 7.273715989424141e-06, "loss": 0.9175, "step": 16190 }, { "epoch": 0.29465796031211916, "grad_norm": 1.6580541133880615, "learning_rate": 7.271840837068013e-06, "loss": 0.9885, "step": 16200 }, { "epoch": 0.29483984794194146, "grad_norm": 1.5334750413894653, "learning_rate": 7.269965684711884e-06, "loss": 0.9577, "step": 16210 }, { "epoch": 0.29502173557176375, "grad_norm": 1.9514312744140625, "learning_rate": 7.2680905323557546e-06, "loss": 0.9247, "step": 16220 }, { "epoch": 0.29520362320158605, "grad_norm": 1.9120029211044312, "learning_rate": 7.266215379999626e-06, "loss": 0.9722, "step": 16230 }, { "epoch": 0.29538551083140835, "grad_norm": 1.762048602104187, "learning_rate": 7.264340227643497e-06, "loss": 0.9485, "step": 16240 }, { "epoch": 0.29556739846123065, "grad_norm": 1.8751541376113892, "learning_rate": 7.262465075287369e-06, "loss": 0.9561, "step": 16250 }, { "epoch": 0.29574928609105294, "grad_norm": 2.013787269592285, "learning_rate": 7.260589922931238e-06, "loss": 0.9782, "step": 16260 }, { "epoch": 0.29593117372087524, "grad_norm": 2.2378485202789307, "learning_rate": 7.25871477057511e-06, "loss": 0.9894, "step": 16270 }, { "epoch": 0.29611306135069754, "grad_norm": 1.6353799104690552, "learning_rate": 7.25683961821898e-06, "loss": 0.949, "step": 16280 }, { "epoch": 0.29629494898051983, "grad_norm": 1.9163002967834473, "learning_rate": 7.2549644658628514e-06, "loss": 0.9718, "step": 16290 }, { "epoch": 0.29647683661034213, "grad_norm": 1.7738646268844604, "learning_rate": 7.253089313506723e-06, "loss": 0.9672, "step": 16300 }, { "epoch": 0.29665872424016443, "grad_norm": 1.8424556255340576, "learning_rate": 7.251214161150594e-06, "loss": 0.9776, "step": 16310 }, { "epoch": 0.2968406118699867, "grad_norm": 1.9713268280029297, "learning_rate": 7.249339008794465e-06, "loss": 1.0156, "step": 16320 }, { "epoch": 0.297022499499809, "grad_norm": 1.932427167892456, "learning_rate": 7.247463856438336e-06, "loss": 0.9706, "step": 16330 }, { "epoch": 0.2972043871296313, "grad_norm": 1.7492613792419434, "learning_rate": 7.245588704082207e-06, "loss": 0.9562, "step": 16340 }, { "epoch": 0.2973862747594536, "grad_norm": 1.9578547477722168, "learning_rate": 7.243713551726078e-06, "loss": 0.9813, "step": 16350 }, { "epoch": 0.2975681623892759, "grad_norm": 1.8969460725784302, "learning_rate": 7.241838399369949e-06, "loss": 0.9866, "step": 16360 }, { "epoch": 0.2977500500190982, "grad_norm": 1.6135566234588623, "learning_rate": 7.2399632470138205e-06, "loss": 0.9935, "step": 16370 }, { "epoch": 0.2979319376489205, "grad_norm": 1.7874118089675903, "learning_rate": 7.238088094657692e-06, "loss": 0.9346, "step": 16380 }, { "epoch": 0.2981138252787428, "grad_norm": 2.1884615421295166, "learning_rate": 7.236212942301562e-06, "loss": 0.9602, "step": 16390 }, { "epoch": 0.2982957129085651, "grad_norm": 1.611219048500061, "learning_rate": 7.234337789945434e-06, "loss": 0.9374, "step": 16400 }, { "epoch": 0.2984776005383874, "grad_norm": 1.8737378120422363, "learning_rate": 7.232462637589305e-06, "loss": 0.9712, "step": 16410 }, { "epoch": 0.2986594881682097, "grad_norm": 1.7604877948760986, "learning_rate": 7.2305874852331756e-06, "loss": 0.9436, "step": 16420 }, { "epoch": 0.298841375798032, "grad_norm": 2.0272700786590576, "learning_rate": 7.228712332877047e-06, "loss": 0.963, "step": 16430 }, { "epoch": 0.2990232634278543, "grad_norm": 1.7695460319519043, "learning_rate": 7.226837180520918e-06, "loss": 1.0239, "step": 16440 }, { "epoch": 0.2992051510576766, "grad_norm": 1.8327637910842896, "learning_rate": 7.22496202816479e-06, "loss": 0.9032, "step": 16450 }, { "epoch": 0.2993870386874989, "grad_norm": 1.5888820886611938, "learning_rate": 7.22308687580866e-06, "loss": 1.006, "step": 16460 }, { "epoch": 0.2995689263173212, "grad_norm": 1.8765348196029663, "learning_rate": 7.2212117234525314e-06, "loss": 0.9778, "step": 16470 }, { "epoch": 0.2997508139471435, "grad_norm": 1.7241919040679932, "learning_rate": 7.219336571096403e-06, "loss": 0.9525, "step": 16480 }, { "epoch": 0.2999327015769658, "grad_norm": 2.0809221267700195, "learning_rate": 7.217461418740273e-06, "loss": 0.9113, "step": 16490 }, { "epoch": 0.3001145892067881, "grad_norm": 1.7928916215896606, "learning_rate": 7.215586266384145e-06, "loss": 0.9437, "step": 16500 }, { "epoch": 0.3002964768366103, "grad_norm": 1.9963165521621704, "learning_rate": 7.213711114028016e-06, "loss": 0.9711, "step": 16510 }, { "epoch": 0.3004783644664326, "grad_norm": 1.7790125608444214, "learning_rate": 7.211835961671887e-06, "loss": 0.9683, "step": 16520 }, { "epoch": 0.3006602520962549, "grad_norm": 1.8868930339813232, "learning_rate": 7.209960809315758e-06, "loss": 0.9759, "step": 16530 }, { "epoch": 0.3008421397260772, "grad_norm": 1.6990758180618286, "learning_rate": 7.208085656959628e-06, "loss": 0.9628, "step": 16540 }, { "epoch": 0.3010240273558995, "grad_norm": 1.5543116331100464, "learning_rate": 7.206210504603499e-06, "loss": 0.9897, "step": 16550 }, { "epoch": 0.3012059149857218, "grad_norm": 1.9792580604553223, "learning_rate": 7.20433535224737e-06, "loss": 0.9534, "step": 16560 }, { "epoch": 0.3013878026155441, "grad_norm": 1.9167406558990479, "learning_rate": 7.2024601998912415e-06, "loss": 0.9781, "step": 16570 }, { "epoch": 0.3015696902453664, "grad_norm": 1.9952348470687866, "learning_rate": 7.200585047535113e-06, "loss": 0.9733, "step": 16580 }, { "epoch": 0.3017515778751887, "grad_norm": 2.0631120204925537, "learning_rate": 7.198709895178983e-06, "loss": 0.9703, "step": 16590 }, { "epoch": 0.301933465505011, "grad_norm": 1.812882423400879, "learning_rate": 7.196834742822855e-06, "loss": 0.9768, "step": 16600 }, { "epoch": 0.3021153531348333, "grad_norm": 1.8575290441513062, "learning_rate": 7.194959590466726e-06, "loss": 0.9544, "step": 16610 }, { "epoch": 0.3022972407646556, "grad_norm": 1.7783591747283936, "learning_rate": 7.1930844381105965e-06, "loss": 0.9669, "step": 16620 }, { "epoch": 0.3024791283944779, "grad_norm": 1.7712740898132324, "learning_rate": 7.191209285754468e-06, "loss": 1.0228, "step": 16630 }, { "epoch": 0.3026610160243002, "grad_norm": 1.8130881786346436, "learning_rate": 7.189334133398339e-06, "loss": 1.0014, "step": 16640 }, { "epoch": 0.3028429036541225, "grad_norm": 1.8524647951126099, "learning_rate": 7.187458981042211e-06, "loss": 0.9975, "step": 16650 }, { "epoch": 0.30302479128394477, "grad_norm": 2.3552539348602295, "learning_rate": 7.185583828686081e-06, "loss": 0.9542, "step": 16660 }, { "epoch": 0.30320667891376707, "grad_norm": 1.790747880935669, "learning_rate": 7.1837086763299524e-06, "loss": 0.9481, "step": 16670 }, { "epoch": 0.30338856654358937, "grad_norm": 2.172921657562256, "learning_rate": 7.181833523973824e-06, "loss": 0.9577, "step": 16680 }, { "epoch": 0.30357045417341166, "grad_norm": 1.98452889919281, "learning_rate": 7.179958371617694e-06, "loss": 0.9945, "step": 16690 }, { "epoch": 0.30375234180323396, "grad_norm": 1.7650880813598633, "learning_rate": 7.178083219261566e-06, "loss": 0.9714, "step": 16700 }, { "epoch": 0.30393422943305626, "grad_norm": 1.7654763460159302, "learning_rate": 7.176208066905437e-06, "loss": 0.9438, "step": 16710 }, { "epoch": 0.30411611706287855, "grad_norm": 1.9195278882980347, "learning_rate": 7.174332914549308e-06, "loss": 0.9646, "step": 16720 }, { "epoch": 0.30429800469270085, "grad_norm": 1.7103880643844604, "learning_rate": 7.172457762193179e-06, "loss": 0.9462, "step": 16730 }, { "epoch": 0.30447989232252315, "grad_norm": 1.9434927701950073, "learning_rate": 7.17058260983705e-06, "loss": 0.9748, "step": 16740 }, { "epoch": 0.30466177995234545, "grad_norm": 1.7774547338485718, "learning_rate": 7.1687074574809215e-06, "loss": 0.9953, "step": 16750 }, { "epoch": 0.30484366758216774, "grad_norm": 1.4004487991333008, "learning_rate": 7.166832305124792e-06, "loss": 0.9515, "step": 16760 }, { "epoch": 0.30502555521199004, "grad_norm": 1.745778203010559, "learning_rate": 7.164957152768663e-06, "loss": 0.9826, "step": 16770 }, { "epoch": 0.30520744284181234, "grad_norm": 1.8207790851593018, "learning_rate": 7.163082000412535e-06, "loss": 0.9778, "step": 16780 }, { "epoch": 0.30538933047163463, "grad_norm": 1.9178853034973145, "learning_rate": 7.161206848056406e-06, "loss": 0.9026, "step": 16790 }, { "epoch": 0.30557121810145693, "grad_norm": 1.9085743427276611, "learning_rate": 7.1593316957002765e-06, "loss": 0.9737, "step": 16800 }, { "epoch": 0.30575310573127923, "grad_norm": 1.979950189590454, "learning_rate": 7.157456543344147e-06, "loss": 0.9952, "step": 16810 }, { "epoch": 0.3059349933611015, "grad_norm": 2.031468629837036, "learning_rate": 7.1555813909880175e-06, "loss": 0.985, "step": 16820 }, { "epoch": 0.3061168809909238, "grad_norm": 1.9560374021530151, "learning_rate": 7.153706238631889e-06, "loss": 0.9552, "step": 16830 }, { "epoch": 0.3062987686207461, "grad_norm": 1.9647176265716553, "learning_rate": 7.15183108627576e-06, "loss": 0.952, "step": 16840 }, { "epoch": 0.3064806562505684, "grad_norm": 2.3069026470184326, "learning_rate": 7.149955933919632e-06, "loss": 0.9915, "step": 16850 }, { "epoch": 0.3066625438803907, "grad_norm": 1.8992050886154175, "learning_rate": 7.148080781563502e-06, "loss": 0.9603, "step": 16860 }, { "epoch": 0.306844431510213, "grad_norm": 1.4912513494491577, "learning_rate": 7.146205629207373e-06, "loss": 0.9393, "step": 16870 }, { "epoch": 0.3070263191400353, "grad_norm": 2.0222220420837402, "learning_rate": 7.144330476851245e-06, "loss": 1.0115, "step": 16880 }, { "epoch": 0.3072082067698576, "grad_norm": 1.64396333694458, "learning_rate": 7.142455324495115e-06, "loss": 0.9398, "step": 16890 }, { "epoch": 0.3073900943996799, "grad_norm": 1.5727607011795044, "learning_rate": 7.140580172138987e-06, "loss": 1.0116, "step": 16900 }, { "epoch": 0.3075719820295022, "grad_norm": 1.8319810628890991, "learning_rate": 7.138705019782858e-06, "loss": 0.9487, "step": 16910 }, { "epoch": 0.30775386965932444, "grad_norm": 1.7792812585830688, "learning_rate": 7.136829867426729e-06, "loss": 1.0021, "step": 16920 }, { "epoch": 0.30793575728914674, "grad_norm": 1.6446012258529663, "learning_rate": 7.1349547150706e-06, "loss": 1.0132, "step": 16930 }, { "epoch": 0.30811764491896904, "grad_norm": 1.8370671272277832, "learning_rate": 7.133079562714471e-06, "loss": 0.9468, "step": 16940 }, { "epoch": 0.30829953254879133, "grad_norm": 1.8273836374282837, "learning_rate": 7.1312044103583425e-06, "loss": 0.9913, "step": 16950 }, { "epoch": 0.30848142017861363, "grad_norm": 1.5634357929229736, "learning_rate": 7.129329258002213e-06, "loss": 0.9999, "step": 16960 }, { "epoch": 0.3086633078084359, "grad_norm": 1.8204610347747803, "learning_rate": 7.127454105646084e-06, "loss": 0.9943, "step": 16970 }, { "epoch": 0.3088451954382582, "grad_norm": 2.0715255737304688, "learning_rate": 7.125578953289956e-06, "loss": 0.9649, "step": 16980 }, { "epoch": 0.3090270830680805, "grad_norm": 1.9158834218978882, "learning_rate": 7.123703800933827e-06, "loss": 0.9529, "step": 16990 }, { "epoch": 0.3092089706979028, "grad_norm": 2.175045967102051, "learning_rate": 7.1218286485776975e-06, "loss": 0.9656, "step": 17000 }, { "epoch": 0.3093908583277251, "grad_norm": 1.906081199645996, "learning_rate": 7.119953496221569e-06, "loss": 1.0125, "step": 17010 }, { "epoch": 0.3095727459575474, "grad_norm": 1.989466905593872, "learning_rate": 7.11807834386544e-06, "loss": 0.9579, "step": 17020 }, { "epoch": 0.3097546335873697, "grad_norm": 1.9105993509292603, "learning_rate": 7.116203191509311e-06, "loss": 1.0182, "step": 17030 }, { "epoch": 0.309936521217192, "grad_norm": 1.934565544128418, "learning_rate": 7.114328039153182e-06, "loss": 1.0116, "step": 17040 }, { "epoch": 0.3101184088470143, "grad_norm": 1.8717756271362305, "learning_rate": 7.112452886797053e-06, "loss": 0.9674, "step": 17050 }, { "epoch": 0.3103002964768366, "grad_norm": 1.8642945289611816, "learning_rate": 7.110577734440925e-06, "loss": 1.0047, "step": 17060 }, { "epoch": 0.3104821841066589, "grad_norm": 1.6120551824569702, "learning_rate": 7.108702582084795e-06, "loss": 0.8852, "step": 17070 }, { "epoch": 0.3106640717364812, "grad_norm": 1.8083833456039429, "learning_rate": 7.106827429728667e-06, "loss": 0.9396, "step": 17080 }, { "epoch": 0.3108459593663035, "grad_norm": 1.637912392616272, "learning_rate": 7.104952277372536e-06, "loss": 0.9174, "step": 17090 }, { "epoch": 0.3110278469961258, "grad_norm": 1.7549060583114624, "learning_rate": 7.103077125016408e-06, "loss": 0.9804, "step": 17100 }, { "epoch": 0.3112097346259481, "grad_norm": 1.9938567876815796, "learning_rate": 7.101201972660279e-06, "loss": 0.975, "step": 17110 }, { "epoch": 0.3113916222557704, "grad_norm": 1.6747959852218628, "learning_rate": 7.09932682030415e-06, "loss": 0.9598, "step": 17120 }, { "epoch": 0.3115735098855927, "grad_norm": 1.81370210647583, "learning_rate": 7.097451667948021e-06, "loss": 0.9928, "step": 17130 }, { "epoch": 0.311755397515415, "grad_norm": 1.625963807106018, "learning_rate": 7.095576515591892e-06, "loss": 0.9987, "step": 17140 }, { "epoch": 0.3119372851452373, "grad_norm": 2.256035089492798, "learning_rate": 7.0937013632357635e-06, "loss": 0.9708, "step": 17150 }, { "epoch": 0.3121191727750596, "grad_norm": 2.0494306087493896, "learning_rate": 7.091826210879634e-06, "loss": 0.9937, "step": 17160 }, { "epoch": 0.31230106040488187, "grad_norm": 1.8634881973266602, "learning_rate": 7.089951058523505e-06, "loss": 0.9653, "step": 17170 }, { "epoch": 0.31248294803470417, "grad_norm": 1.842549443244934, "learning_rate": 7.088075906167377e-06, "loss": 0.9954, "step": 17180 }, { "epoch": 0.31266483566452646, "grad_norm": 1.740484356880188, "learning_rate": 7.086200753811248e-06, "loss": 0.9122, "step": 17190 }, { "epoch": 0.31284672329434876, "grad_norm": 2.12878155708313, "learning_rate": 7.0843256014551185e-06, "loss": 0.9726, "step": 17200 }, { "epoch": 0.31302861092417106, "grad_norm": 1.7786214351654053, "learning_rate": 7.08245044909899e-06, "loss": 0.9791, "step": 17210 }, { "epoch": 0.31321049855399336, "grad_norm": 2.275665044784546, "learning_rate": 7.080575296742861e-06, "loss": 0.9805, "step": 17220 }, { "epoch": 0.31339238618381565, "grad_norm": 1.9142580032348633, "learning_rate": 7.078700144386732e-06, "loss": 0.9689, "step": 17230 }, { "epoch": 0.31357427381363795, "grad_norm": 1.923582911491394, "learning_rate": 7.076824992030603e-06, "loss": 0.9778, "step": 17240 }, { "epoch": 0.31375616144346025, "grad_norm": 1.5965503454208374, "learning_rate": 7.074949839674474e-06, "loss": 0.9325, "step": 17250 }, { "epoch": 0.31393804907328254, "grad_norm": 1.6698459386825562, "learning_rate": 7.073074687318346e-06, "loss": 0.9436, "step": 17260 }, { "epoch": 0.31411993670310484, "grad_norm": 1.6553102731704712, "learning_rate": 7.071199534962216e-06, "loss": 0.9149, "step": 17270 }, { "epoch": 0.31430182433292714, "grad_norm": 2.039456844329834, "learning_rate": 7.069324382606088e-06, "loss": 0.9464, "step": 17280 }, { "epoch": 0.31448371196274943, "grad_norm": 1.7551900148391724, "learning_rate": 7.067449230249959e-06, "loss": 0.9534, "step": 17290 }, { "epoch": 0.31466559959257173, "grad_norm": 1.9351842403411865, "learning_rate": 7.0655740778938294e-06, "loss": 0.9645, "step": 17300 }, { "epoch": 0.31484748722239403, "grad_norm": 1.6156986951828003, "learning_rate": 7.063698925537701e-06, "loss": 0.9343, "step": 17310 }, { "epoch": 0.3150293748522163, "grad_norm": 1.811111569404602, "learning_rate": 7.061823773181572e-06, "loss": 0.9427, "step": 17320 }, { "epoch": 0.3152112624820386, "grad_norm": 1.5763550996780396, "learning_rate": 7.0599486208254435e-06, "loss": 0.9362, "step": 17330 }, { "epoch": 0.31539315011186086, "grad_norm": 1.5860577821731567, "learning_rate": 7.058073468469314e-06, "loss": 0.9213, "step": 17340 }, { "epoch": 0.31557503774168316, "grad_norm": 1.913583517074585, "learning_rate": 7.056198316113185e-06, "loss": 0.9719, "step": 17350 }, { "epoch": 0.31575692537150546, "grad_norm": 2.0779712200164795, "learning_rate": 7.054323163757057e-06, "loss": 1.0207, "step": 17360 }, { "epoch": 0.31593881300132776, "grad_norm": 1.583220362663269, "learning_rate": 7.052448011400926e-06, "loss": 0.9318, "step": 17370 }, { "epoch": 0.31612070063115005, "grad_norm": 1.8386383056640625, "learning_rate": 7.050572859044798e-06, "loss": 0.9694, "step": 17380 }, { "epoch": 0.31630258826097235, "grad_norm": 2.196873664855957, "learning_rate": 7.048697706688669e-06, "loss": 0.9567, "step": 17390 }, { "epoch": 0.31648447589079465, "grad_norm": 1.889480710029602, "learning_rate": 7.0468225543325395e-06, "loss": 0.9447, "step": 17400 }, { "epoch": 0.31666636352061694, "grad_norm": 1.868050456047058, "learning_rate": 7.044947401976411e-06, "loss": 0.9611, "step": 17410 }, { "epoch": 0.31684825115043924, "grad_norm": 1.9232454299926758, "learning_rate": 7.043072249620282e-06, "loss": 0.9762, "step": 17420 }, { "epoch": 0.31703013878026154, "grad_norm": 1.9982192516326904, "learning_rate": 7.041197097264153e-06, "loss": 0.9799, "step": 17430 }, { "epoch": 0.31721202641008384, "grad_norm": 1.8170651197433472, "learning_rate": 7.039321944908024e-06, "loss": 0.9273, "step": 17440 }, { "epoch": 0.31739391403990613, "grad_norm": 1.9631552696228027, "learning_rate": 7.037446792551895e-06, "loss": 0.9786, "step": 17450 }, { "epoch": 0.31757580166972843, "grad_norm": 1.762860655784607, "learning_rate": 7.035571640195767e-06, "loss": 0.9634, "step": 17460 }, { "epoch": 0.3177576892995507, "grad_norm": 1.659618854522705, "learning_rate": 7.033696487839637e-06, "loss": 0.9651, "step": 17470 }, { "epoch": 0.317939576929373, "grad_norm": 2.0607964992523193, "learning_rate": 7.031821335483509e-06, "loss": 0.9386, "step": 17480 }, { "epoch": 0.3181214645591953, "grad_norm": 1.5752860307693481, "learning_rate": 7.02994618312738e-06, "loss": 0.9651, "step": 17490 }, { "epoch": 0.3183033521890176, "grad_norm": 1.7745592594146729, "learning_rate": 7.0280710307712504e-06, "loss": 0.9699, "step": 17500 }, { "epoch": 0.3184852398188399, "grad_norm": 2.013772964477539, "learning_rate": 7.026195878415122e-06, "loss": 0.9863, "step": 17510 }, { "epoch": 0.3186671274486622, "grad_norm": 2.00417160987854, "learning_rate": 7.024320726058993e-06, "loss": 0.9831, "step": 17520 }, { "epoch": 0.3188490150784845, "grad_norm": 2.0747883319854736, "learning_rate": 7.0224455737028645e-06, "loss": 0.9513, "step": 17530 }, { "epoch": 0.3190309027083068, "grad_norm": 1.6616145372390747, "learning_rate": 7.020570421346735e-06, "loss": 0.9043, "step": 17540 }, { "epoch": 0.3192127903381291, "grad_norm": 1.8961272239685059, "learning_rate": 7.018695268990606e-06, "loss": 0.9521, "step": 17550 }, { "epoch": 0.3193946779679514, "grad_norm": 1.6588767766952515, "learning_rate": 7.016820116634478e-06, "loss": 0.9203, "step": 17560 }, { "epoch": 0.3195765655977737, "grad_norm": 1.851897954940796, "learning_rate": 7.014944964278348e-06, "loss": 0.9714, "step": 17570 }, { "epoch": 0.319758453227596, "grad_norm": 1.3482609987258911, "learning_rate": 7.0130698119222195e-06, "loss": 0.9349, "step": 17580 }, { "epoch": 0.3199403408574183, "grad_norm": 1.9141026735305786, "learning_rate": 7.011194659566091e-06, "loss": 0.9513, "step": 17590 }, { "epoch": 0.3201222284872406, "grad_norm": 1.7604691982269287, "learning_rate": 7.009319507209962e-06, "loss": 0.9461, "step": 17600 }, { "epoch": 0.3203041161170629, "grad_norm": 1.807477355003357, "learning_rate": 7.007444354853833e-06, "loss": 0.9807, "step": 17610 }, { "epoch": 0.3204860037468852, "grad_norm": 1.8562551736831665, "learning_rate": 7.005569202497704e-06, "loss": 0.9554, "step": 17620 }, { "epoch": 0.3206678913767075, "grad_norm": 1.6959906816482544, "learning_rate": 7.003694050141575e-06, "loss": 0.9501, "step": 17630 }, { "epoch": 0.3208497790065298, "grad_norm": 1.3080979585647583, "learning_rate": 7.001818897785445e-06, "loss": 0.9702, "step": 17640 }, { "epoch": 0.3210316666363521, "grad_norm": 1.8562761545181274, "learning_rate": 6.999943745429316e-06, "loss": 0.96, "step": 17650 }, { "epoch": 0.3212135542661744, "grad_norm": 1.727596640586853, "learning_rate": 6.998068593073187e-06, "loss": 1.0123, "step": 17660 }, { "epoch": 0.32139544189599667, "grad_norm": 1.8950073719024658, "learning_rate": 6.996193440717058e-06, "loss": 0.9692, "step": 17670 }, { "epoch": 0.32157732952581897, "grad_norm": 1.7806305885314941, "learning_rate": 6.99431828836093e-06, "loss": 0.9452, "step": 17680 }, { "epoch": 0.32175921715564126, "grad_norm": 1.8167861700057983, "learning_rate": 6.992443136004801e-06, "loss": 0.9797, "step": 17690 }, { "epoch": 0.32194110478546356, "grad_norm": 1.8731948137283325, "learning_rate": 6.9905679836486714e-06, "loss": 0.9511, "step": 17700 }, { "epoch": 0.32212299241528586, "grad_norm": 1.9191720485687256, "learning_rate": 6.988692831292543e-06, "loss": 1.0076, "step": 17710 }, { "epoch": 0.32230488004510816, "grad_norm": 2.084716558456421, "learning_rate": 6.986817678936414e-06, "loss": 1.0114, "step": 17720 }, { "epoch": 0.32248676767493045, "grad_norm": 1.7297780513763428, "learning_rate": 6.9849425265802855e-06, "loss": 0.9695, "step": 17730 }, { "epoch": 0.32266865530475275, "grad_norm": 1.8233392238616943, "learning_rate": 6.983067374224156e-06, "loss": 0.9732, "step": 17740 }, { "epoch": 0.32285054293457505, "grad_norm": 1.9199872016906738, "learning_rate": 6.981192221868027e-06, "loss": 0.9691, "step": 17750 }, { "epoch": 0.3230324305643973, "grad_norm": 1.9149428606033325, "learning_rate": 6.979317069511899e-06, "loss": 1.0056, "step": 17760 }, { "epoch": 0.3232143181942196, "grad_norm": 1.949476957321167, "learning_rate": 6.977441917155769e-06, "loss": 0.9538, "step": 17770 }, { "epoch": 0.3233962058240419, "grad_norm": 1.8582606315612793, "learning_rate": 6.9755667647996405e-06, "loss": 0.9455, "step": 17780 }, { "epoch": 0.3235780934538642, "grad_norm": 1.6459109783172607, "learning_rate": 6.973691612443512e-06, "loss": 0.9654, "step": 17790 }, { "epoch": 0.3237599810836865, "grad_norm": 1.8793400526046753, "learning_rate": 6.971816460087383e-06, "loss": 0.9297, "step": 17800 }, { "epoch": 0.3239418687135088, "grad_norm": 1.8913977146148682, "learning_rate": 6.969941307731254e-06, "loss": 0.9503, "step": 17810 }, { "epoch": 0.32412375634333107, "grad_norm": 1.9153498411178589, "learning_rate": 6.968066155375125e-06, "loss": 0.9478, "step": 17820 }, { "epoch": 0.32430564397315337, "grad_norm": 1.8137028217315674, "learning_rate": 6.966191003018996e-06, "loss": 0.976, "step": 17830 }, { "epoch": 0.32448753160297567, "grad_norm": 1.6685136556625366, "learning_rate": 6.964315850662867e-06, "loss": 1.0035, "step": 17840 }, { "epoch": 0.32466941923279796, "grad_norm": 1.9644677639007568, "learning_rate": 6.962440698306738e-06, "loss": 0.99, "step": 17850 }, { "epoch": 0.32485130686262026, "grad_norm": 1.6844993829727173, "learning_rate": 6.9605655459506096e-06, "loss": 0.9759, "step": 17860 }, { "epoch": 0.32503319449244256, "grad_norm": 1.7223246097564697, "learning_rate": 6.958690393594481e-06, "loss": 0.953, "step": 17870 }, { "epoch": 0.32521508212226485, "grad_norm": 1.9517260789871216, "learning_rate": 6.956815241238351e-06, "loss": 0.9843, "step": 17880 }, { "epoch": 0.32539696975208715, "grad_norm": 1.9121862649917603, "learning_rate": 6.954940088882223e-06, "loss": 0.9818, "step": 17890 }, { "epoch": 0.32557885738190945, "grad_norm": 1.9232604503631592, "learning_rate": 6.953064936526094e-06, "loss": 0.9713, "step": 17900 }, { "epoch": 0.32576074501173174, "grad_norm": 1.7684513330459595, "learning_rate": 6.951189784169965e-06, "loss": 0.91, "step": 17910 }, { "epoch": 0.32594263264155404, "grad_norm": 1.6830466985702515, "learning_rate": 6.949314631813835e-06, "loss": 0.9556, "step": 17920 }, { "epoch": 0.32612452027137634, "grad_norm": 1.8105405569076538, "learning_rate": 6.947439479457706e-06, "loss": 0.9427, "step": 17930 }, { "epoch": 0.32630640790119864, "grad_norm": 1.8196887969970703, "learning_rate": 6.945564327101577e-06, "loss": 0.9769, "step": 17940 }, { "epoch": 0.32648829553102093, "grad_norm": 2.058438301086426, "learning_rate": 6.943689174745448e-06, "loss": 0.9244, "step": 17950 }, { "epoch": 0.32667018316084323, "grad_norm": 1.5371114015579224, "learning_rate": 6.94181402238932e-06, "loss": 0.9431, "step": 17960 }, { "epoch": 0.3268520707906655, "grad_norm": 1.8851964473724365, "learning_rate": 6.93993887003319e-06, "loss": 0.9517, "step": 17970 }, { "epoch": 0.3270339584204878, "grad_norm": 2.2233192920684814, "learning_rate": 6.9380637176770615e-06, "loss": 0.9975, "step": 17980 }, { "epoch": 0.3272158460503101, "grad_norm": 1.696704387664795, "learning_rate": 6.936188565320933e-06, "loss": 0.9676, "step": 17990 }, { "epoch": 0.3273977336801324, "grad_norm": 1.9966048002243042, "learning_rate": 6.934313412964803e-06, "loss": 0.9873, "step": 18000 }, { "epoch": 0.3275796213099547, "grad_norm": 1.7879928350448608, "learning_rate": 6.932438260608675e-06, "loss": 0.9825, "step": 18010 }, { "epoch": 0.327761508939777, "grad_norm": 1.9097920656204224, "learning_rate": 6.930563108252546e-06, "loss": 0.9194, "step": 18020 }, { "epoch": 0.3279433965695993, "grad_norm": 1.806315302848816, "learning_rate": 6.928687955896417e-06, "loss": 0.9697, "step": 18030 }, { "epoch": 0.3281252841994216, "grad_norm": 1.9251766204833984, "learning_rate": 6.926812803540288e-06, "loss": 0.9759, "step": 18040 }, { "epoch": 0.3283071718292439, "grad_norm": 2.0474250316619873, "learning_rate": 6.924937651184159e-06, "loss": 0.9039, "step": 18050 }, { "epoch": 0.3284890594590662, "grad_norm": 1.4173970222473145, "learning_rate": 6.9230624988280306e-06, "loss": 0.9736, "step": 18060 }, { "epoch": 0.3286709470888885, "grad_norm": 1.7997825145721436, "learning_rate": 6.921187346471902e-06, "loss": 0.9542, "step": 18070 }, { "epoch": 0.3288528347187108, "grad_norm": 1.8603745698928833, "learning_rate": 6.919312194115772e-06, "loss": 1.0007, "step": 18080 }, { "epoch": 0.3290347223485331, "grad_norm": 1.6843491792678833, "learning_rate": 6.917437041759644e-06, "loss": 0.9749, "step": 18090 }, { "epoch": 0.3292166099783554, "grad_norm": 1.840662956237793, "learning_rate": 6.915561889403515e-06, "loss": 0.989, "step": 18100 }, { "epoch": 0.3293984976081777, "grad_norm": 1.8024054765701294, "learning_rate": 6.913686737047386e-06, "loss": 0.9931, "step": 18110 }, { "epoch": 0.329580385238, "grad_norm": 1.4752856492996216, "learning_rate": 6.911811584691257e-06, "loss": 0.932, "step": 18120 }, { "epoch": 0.3297622728678223, "grad_norm": 1.6828447580337524, "learning_rate": 6.909936432335128e-06, "loss": 0.9281, "step": 18130 }, { "epoch": 0.3299441604976446, "grad_norm": 1.7964787483215332, "learning_rate": 6.908061279979e-06, "loss": 0.9662, "step": 18140 }, { "epoch": 0.3301260481274669, "grad_norm": 1.7154407501220703, "learning_rate": 6.90618612762287e-06, "loss": 0.9919, "step": 18150 }, { "epoch": 0.3303079357572892, "grad_norm": 1.5258060693740845, "learning_rate": 6.9043109752667415e-06, "loss": 0.9648, "step": 18160 }, { "epoch": 0.33048982338711147, "grad_norm": 1.5440808534622192, "learning_rate": 6.902435822910613e-06, "loss": 0.9249, "step": 18170 }, { "epoch": 0.3306717110169337, "grad_norm": 2.0045745372772217, "learning_rate": 6.900560670554483e-06, "loss": 0.9843, "step": 18180 }, { "epoch": 0.330853598646756, "grad_norm": 1.8393396139144897, "learning_rate": 6.898685518198354e-06, "loss": 0.942, "step": 18190 }, { "epoch": 0.3310354862765783, "grad_norm": 1.8636430501937866, "learning_rate": 6.896810365842224e-06, "loss": 0.9711, "step": 18200 }, { "epoch": 0.3312173739064006, "grad_norm": 1.7236974239349365, "learning_rate": 6.894935213486096e-06, "loss": 0.976, "step": 18210 }, { "epoch": 0.3313992615362229, "grad_norm": 1.560594081878662, "learning_rate": 6.893060061129967e-06, "loss": 0.896, "step": 18220 }, { "epoch": 0.3315811491660452, "grad_norm": 1.6876016855239868, "learning_rate": 6.891184908773838e-06, "loss": 0.9377, "step": 18230 }, { "epoch": 0.3317630367958675, "grad_norm": 1.9315391778945923, "learning_rate": 6.889309756417709e-06, "loss": 0.9542, "step": 18240 }, { "epoch": 0.3319449244256898, "grad_norm": 1.7926255464553833, "learning_rate": 6.88743460406158e-06, "loss": 1.0271, "step": 18250 }, { "epoch": 0.3321268120555121, "grad_norm": 1.836022973060608, "learning_rate": 6.8855594517054516e-06, "loss": 0.9803, "step": 18260 }, { "epoch": 0.3323086996853344, "grad_norm": 1.6379709243774414, "learning_rate": 6.883684299349322e-06, "loss": 0.994, "step": 18270 }, { "epoch": 0.3324905873151567, "grad_norm": 1.405582070350647, "learning_rate": 6.881809146993193e-06, "loss": 0.9285, "step": 18280 }, { "epoch": 0.332672474944979, "grad_norm": 1.710940957069397, "learning_rate": 6.879933994637065e-06, "loss": 0.9423, "step": 18290 }, { "epoch": 0.3328543625748013, "grad_norm": 1.7459790706634521, "learning_rate": 6.878058842280936e-06, "loss": 0.9567, "step": 18300 }, { "epoch": 0.3330362502046236, "grad_norm": 1.7908908128738403, "learning_rate": 6.876183689924807e-06, "loss": 0.9474, "step": 18310 }, { "epoch": 0.33321813783444587, "grad_norm": 1.6309661865234375, "learning_rate": 6.874308537568678e-06, "loss": 0.9753, "step": 18320 }, { "epoch": 0.33340002546426817, "grad_norm": 1.975085973739624, "learning_rate": 6.872433385212549e-06, "loss": 0.947, "step": 18330 }, { "epoch": 0.33358191309409047, "grad_norm": 1.8583658933639526, "learning_rate": 6.87055823285642e-06, "loss": 0.9861, "step": 18340 }, { "epoch": 0.33376380072391276, "grad_norm": 1.7887905836105347, "learning_rate": 6.868683080500291e-06, "loss": 0.9816, "step": 18350 }, { "epoch": 0.33394568835373506, "grad_norm": 1.6085526943206787, "learning_rate": 6.8668079281441625e-06, "loss": 0.9364, "step": 18360 }, { "epoch": 0.33412757598355736, "grad_norm": 1.9245717525482178, "learning_rate": 6.864932775788034e-06, "loss": 0.9527, "step": 18370 }, { "epoch": 0.33430946361337965, "grad_norm": 1.7649201154708862, "learning_rate": 6.863057623431904e-06, "loss": 1.0457, "step": 18380 }, { "epoch": 0.33449135124320195, "grad_norm": 1.813568353652954, "learning_rate": 6.861182471075776e-06, "loss": 0.9096, "step": 18390 }, { "epoch": 0.33467323887302425, "grad_norm": 1.9370174407958984, "learning_rate": 6.859307318719647e-06, "loss": 0.9987, "step": 18400 }, { "epoch": 0.33485512650284655, "grad_norm": 1.8475254774093628, "learning_rate": 6.857432166363518e-06, "loss": 0.961, "step": 18410 }, { "epoch": 0.33503701413266884, "grad_norm": 1.912887692451477, "learning_rate": 6.855557014007389e-06, "loss": 0.9856, "step": 18420 }, { "epoch": 0.33521890176249114, "grad_norm": 1.4538737535476685, "learning_rate": 6.85368186165126e-06, "loss": 0.9487, "step": 18430 }, { "epoch": 0.33540078939231344, "grad_norm": 1.8133844137191772, "learning_rate": 6.8518067092951316e-06, "loss": 0.9523, "step": 18440 }, { "epoch": 0.33558267702213573, "grad_norm": 2.0079169273376465, "learning_rate": 6.849931556939002e-06, "loss": 0.996, "step": 18450 }, { "epoch": 0.33576456465195803, "grad_norm": 1.6972055435180664, "learning_rate": 6.848056404582873e-06, "loss": 0.9768, "step": 18460 }, { "epoch": 0.33594645228178033, "grad_norm": 1.8006527423858643, "learning_rate": 6.846181252226743e-06, "loss": 0.951, "step": 18470 }, { "epoch": 0.3361283399116026, "grad_norm": 2.082578182220459, "learning_rate": 6.844306099870614e-06, "loss": 1.0473, "step": 18480 }, { "epoch": 0.3363102275414249, "grad_norm": 1.6940659284591675, "learning_rate": 6.842430947514486e-06, "loss": 0.912, "step": 18490 }, { "epoch": 0.3364921151712472, "grad_norm": 1.7546778917312622, "learning_rate": 6.840555795158357e-06, "loss": 0.9014, "step": 18500 }, { "epoch": 0.3366740028010695, "grad_norm": 2.1499483585357666, "learning_rate": 6.838680642802228e-06, "loss": 1.0025, "step": 18510 }, { "epoch": 0.3368558904308918, "grad_norm": 1.8628674745559692, "learning_rate": 6.836805490446099e-06, "loss": 0.9625, "step": 18520 }, { "epoch": 0.3370377780607141, "grad_norm": 1.8907338380813599, "learning_rate": 6.83493033808997e-06, "loss": 0.9897, "step": 18530 }, { "epoch": 0.3372196656905364, "grad_norm": 1.9896732568740845, "learning_rate": 6.833055185733841e-06, "loss": 0.9433, "step": 18540 }, { "epoch": 0.3374015533203587, "grad_norm": 1.83692467212677, "learning_rate": 6.831180033377712e-06, "loss": 0.9662, "step": 18550 }, { "epoch": 0.337583440950181, "grad_norm": 1.4928460121154785, "learning_rate": 6.8293048810215835e-06, "loss": 0.9393, "step": 18560 }, { "epoch": 0.3377653285800033, "grad_norm": 2.0950260162353516, "learning_rate": 6.827429728665455e-06, "loss": 0.9537, "step": 18570 }, { "epoch": 0.3379472162098256, "grad_norm": 1.855189561843872, "learning_rate": 6.825554576309325e-06, "loss": 0.9349, "step": 18580 }, { "epoch": 0.3381291038396479, "grad_norm": 1.9965275526046753, "learning_rate": 6.823679423953197e-06, "loss": 0.9631, "step": 18590 }, { "epoch": 0.33831099146947013, "grad_norm": 1.808714509010315, "learning_rate": 6.821804271597068e-06, "loss": 0.9381, "step": 18600 }, { "epoch": 0.33849287909929243, "grad_norm": 2.0526275634765625, "learning_rate": 6.8199291192409385e-06, "loss": 0.9154, "step": 18610 }, { "epoch": 0.33867476672911473, "grad_norm": 2.1310393810272217, "learning_rate": 6.81805396688481e-06, "loss": 0.9556, "step": 18620 }, { "epoch": 0.338856654358937, "grad_norm": 2.1013660430908203, "learning_rate": 6.816178814528681e-06, "loss": 0.928, "step": 18630 }, { "epoch": 0.3390385419887593, "grad_norm": 1.7940229177474976, "learning_rate": 6.8143036621725525e-06, "loss": 0.9935, "step": 18640 }, { "epoch": 0.3392204296185816, "grad_norm": 1.7725869417190552, "learning_rate": 6.812428509816423e-06, "loss": 0.9728, "step": 18650 }, { "epoch": 0.3394023172484039, "grad_norm": 1.6583549976348877, "learning_rate": 6.810553357460294e-06, "loss": 0.9437, "step": 18660 }, { "epoch": 0.3395842048782262, "grad_norm": 1.8341736793518066, "learning_rate": 6.808678205104166e-06, "loss": 0.9978, "step": 18670 }, { "epoch": 0.3397660925080485, "grad_norm": 2.0058207511901855, "learning_rate": 6.806803052748036e-06, "loss": 0.9837, "step": 18680 }, { "epoch": 0.3399479801378708, "grad_norm": 1.7527529001235962, "learning_rate": 6.804927900391908e-06, "loss": 0.9413, "step": 18690 }, { "epoch": 0.3401298677676931, "grad_norm": 1.4731731414794922, "learning_rate": 6.803052748035779e-06, "loss": 0.9447, "step": 18700 }, { "epoch": 0.3403117553975154, "grad_norm": 1.7836195230484009, "learning_rate": 6.80117759567965e-06, "loss": 0.9462, "step": 18710 }, { "epoch": 0.3404936430273377, "grad_norm": 1.7428367137908936, "learning_rate": 6.799302443323521e-06, "loss": 0.924, "step": 18720 }, { "epoch": 0.34067553065716, "grad_norm": 1.8812856674194336, "learning_rate": 6.797427290967392e-06, "loss": 0.9921, "step": 18730 }, { "epoch": 0.3408574182869823, "grad_norm": 1.7676701545715332, "learning_rate": 6.795552138611262e-06, "loss": 0.9469, "step": 18740 }, { "epoch": 0.3410393059168046, "grad_norm": 1.8761498928070068, "learning_rate": 6.793676986255133e-06, "loss": 1.007, "step": 18750 }, { "epoch": 0.3412211935466269, "grad_norm": 1.5796834230422974, "learning_rate": 6.7918018338990045e-06, "loss": 0.9644, "step": 18760 }, { "epoch": 0.3414030811764492, "grad_norm": 1.6537748575210571, "learning_rate": 6.789926681542876e-06, "loss": 0.9148, "step": 18770 }, { "epoch": 0.3415849688062715, "grad_norm": 1.6503853797912598, "learning_rate": 6.788051529186746e-06, "loss": 0.9433, "step": 18780 }, { "epoch": 0.3417668564360938, "grad_norm": 1.8981844186782837, "learning_rate": 6.786176376830618e-06, "loss": 1.0025, "step": 18790 }, { "epoch": 0.3419487440659161, "grad_norm": 68.24463653564453, "learning_rate": 6.784301224474489e-06, "loss": 0.9301, "step": 18800 }, { "epoch": 0.3421306316957384, "grad_norm": 1.7434653043746948, "learning_rate": 6.7824260721183595e-06, "loss": 0.9542, "step": 18810 }, { "epoch": 0.34231251932556067, "grad_norm": 1.8632564544677734, "learning_rate": 6.780550919762231e-06, "loss": 0.9501, "step": 18820 }, { "epoch": 0.34249440695538297, "grad_norm": 1.5848464965820312, "learning_rate": 6.778675767406102e-06, "loss": 0.8921, "step": 18830 }, { "epoch": 0.34267629458520527, "grad_norm": 1.8699554204940796, "learning_rate": 6.7768006150499735e-06, "loss": 1.0087, "step": 18840 }, { "epoch": 0.34285818221502756, "grad_norm": 1.9188249111175537, "learning_rate": 6.774925462693844e-06, "loss": 0.9688, "step": 18850 }, { "epoch": 0.34304006984484986, "grad_norm": 1.5972836017608643, "learning_rate": 6.773050310337715e-06, "loss": 0.9323, "step": 18860 }, { "epoch": 0.34322195747467216, "grad_norm": 1.7466974258422852, "learning_rate": 6.771175157981587e-06, "loss": 0.9369, "step": 18870 }, { "epoch": 0.34340384510449445, "grad_norm": 2.0047826766967773, "learning_rate": 6.769300005625457e-06, "loss": 0.9414, "step": 18880 }, { "epoch": 0.34358573273431675, "grad_norm": 1.6249563694000244, "learning_rate": 6.7674248532693286e-06, "loss": 0.9434, "step": 18890 }, { "epoch": 0.34376762036413905, "grad_norm": 2.1605453491210938, "learning_rate": 6.7655497009132e-06, "loss": 0.9958, "step": 18900 }, { "epoch": 0.34394950799396135, "grad_norm": 1.695791244506836, "learning_rate": 6.763674548557071e-06, "loss": 0.9478, "step": 18910 }, { "epoch": 0.34413139562378364, "grad_norm": 2.5433456897735596, "learning_rate": 6.761799396200942e-06, "loss": 0.9998, "step": 18920 }, { "epoch": 0.34431328325360594, "grad_norm": 1.7417856454849243, "learning_rate": 6.759924243844813e-06, "loss": 0.9752, "step": 18930 }, { "epoch": 0.34449517088342824, "grad_norm": 1.7776873111724854, "learning_rate": 6.7580490914886845e-06, "loss": 0.9398, "step": 18940 }, { "epoch": 0.34467705851325053, "grad_norm": 1.9862675666809082, "learning_rate": 6.756173939132555e-06, "loss": 0.9203, "step": 18950 }, { "epoch": 0.34485894614307283, "grad_norm": 1.8355683088302612, "learning_rate": 6.754298786776426e-06, "loss": 1.0044, "step": 18960 }, { "epoch": 0.34504083377289513, "grad_norm": 1.8910837173461914, "learning_rate": 6.752423634420298e-06, "loss": 1.0117, "step": 18970 }, { "epoch": 0.3452227214027174, "grad_norm": 1.7873389720916748, "learning_rate": 6.750548482064169e-06, "loss": 0.9935, "step": 18980 }, { "epoch": 0.3454046090325397, "grad_norm": 1.94937002658844, "learning_rate": 6.7486733297080395e-06, "loss": 1.003, "step": 18990 }, { "epoch": 0.345586496662362, "grad_norm": 1.9022846221923828, "learning_rate": 6.746798177351911e-06, "loss": 0.984, "step": 19000 }, { "epoch": 0.34576838429218426, "grad_norm": 1.7488044500350952, "learning_rate": 6.744923024995782e-06, "loss": 0.9374, "step": 19010 }, { "epoch": 0.34595027192200656, "grad_norm": 1.9037729501724243, "learning_rate": 6.743047872639652e-06, "loss": 0.9471, "step": 19020 }, { "epoch": 0.34613215955182886, "grad_norm": 1.6361767053604126, "learning_rate": 6.741172720283523e-06, "loss": 0.9445, "step": 19030 }, { "epoch": 0.34631404718165115, "grad_norm": 1.8356149196624756, "learning_rate": 6.7392975679273945e-06, "loss": 0.9567, "step": 19040 }, { "epoch": 0.34649593481147345, "grad_norm": 1.8782070875167847, "learning_rate": 6.737422415571265e-06, "loss": 0.9741, "step": 19050 }, { "epoch": 0.34667782244129575, "grad_norm": 1.9188792705535889, "learning_rate": 6.735547263215136e-06, "loss": 0.9653, "step": 19060 }, { "epoch": 0.34685971007111804, "grad_norm": 2.1017820835113525, "learning_rate": 6.733672110859008e-06, "loss": 1.0116, "step": 19070 }, { "epoch": 0.34704159770094034, "grad_norm": 1.9035903215408325, "learning_rate": 6.731796958502878e-06, "loss": 0.94, "step": 19080 }, { "epoch": 0.34722348533076264, "grad_norm": 1.831088662147522, "learning_rate": 6.7299218061467496e-06, "loss": 0.9941, "step": 19090 }, { "epoch": 0.34740537296058493, "grad_norm": 1.4933007955551147, "learning_rate": 6.728046653790621e-06, "loss": 0.937, "step": 19100 }, { "epoch": 0.34758726059040723, "grad_norm": 1.9686388969421387, "learning_rate": 6.726171501434492e-06, "loss": 0.9052, "step": 19110 }, { "epoch": 0.34776914822022953, "grad_norm": 1.6532413959503174, "learning_rate": 6.724296349078363e-06, "loss": 0.9183, "step": 19120 }, { "epoch": 0.3479510358500518, "grad_norm": 1.8595399856567383, "learning_rate": 6.722421196722234e-06, "loss": 0.9623, "step": 19130 }, { "epoch": 0.3481329234798741, "grad_norm": 2.2733583450317383, "learning_rate": 6.7205460443661054e-06, "loss": 0.976, "step": 19140 }, { "epoch": 0.3483148111096964, "grad_norm": 1.7943824529647827, "learning_rate": 6.718670892009976e-06, "loss": 1.0011, "step": 19150 }, { "epoch": 0.3484966987395187, "grad_norm": 1.8029237985610962, "learning_rate": 6.716795739653847e-06, "loss": 0.9373, "step": 19160 }, { "epoch": 0.348678586369341, "grad_norm": 1.7467204332351685, "learning_rate": 6.714920587297719e-06, "loss": 0.9034, "step": 19170 }, { "epoch": 0.3488604739991633, "grad_norm": 1.824460506439209, "learning_rate": 6.71304543494159e-06, "loss": 0.975, "step": 19180 }, { "epoch": 0.3490423616289856, "grad_norm": 1.8969218730926514, "learning_rate": 6.7111702825854605e-06, "loss": 0.9915, "step": 19190 }, { "epoch": 0.3492242492588079, "grad_norm": 1.8673970699310303, "learning_rate": 6.709295130229332e-06, "loss": 0.9872, "step": 19200 }, { "epoch": 0.3494061368886302, "grad_norm": 1.3765614032745361, "learning_rate": 6.707419977873203e-06, "loss": 0.9458, "step": 19210 }, { "epoch": 0.3495880245184525, "grad_norm": 1.7563319206237793, "learning_rate": 6.705544825517074e-06, "loss": 0.9306, "step": 19220 }, { "epoch": 0.3497699121482748, "grad_norm": 1.7018917798995972, "learning_rate": 6.703669673160945e-06, "loss": 0.9735, "step": 19230 }, { "epoch": 0.3499517997780971, "grad_norm": 1.743767499923706, "learning_rate": 6.701794520804816e-06, "loss": 0.956, "step": 19240 }, { "epoch": 0.3501336874079194, "grad_norm": 1.8742784261703491, "learning_rate": 6.699919368448688e-06, "loss": 0.9992, "step": 19250 }, { "epoch": 0.3503155750377417, "grad_norm": 1.787313461303711, "learning_rate": 6.698044216092558e-06, "loss": 0.9643, "step": 19260 }, { "epoch": 0.350497462667564, "grad_norm": 1.914232850074768, "learning_rate": 6.6961690637364296e-06, "loss": 0.9655, "step": 19270 }, { "epoch": 0.3506793502973863, "grad_norm": 1.8536165952682495, "learning_rate": 6.694293911380301e-06, "loss": 0.9808, "step": 19280 }, { "epoch": 0.3508612379272086, "grad_norm": 1.9159001111984253, "learning_rate": 6.6924187590241706e-06, "loss": 0.9594, "step": 19290 }, { "epoch": 0.3510431255570309, "grad_norm": 1.7481118440628052, "learning_rate": 6.690543606668042e-06, "loss": 0.9564, "step": 19300 }, { "epoch": 0.3512250131868532, "grad_norm": 1.991297721862793, "learning_rate": 6.688668454311913e-06, "loss": 0.9455, "step": 19310 }, { "epoch": 0.35140690081667547, "grad_norm": 1.7424366474151611, "learning_rate": 6.686793301955784e-06, "loss": 0.9464, "step": 19320 }, { "epoch": 0.35158878844649777, "grad_norm": 1.814192295074463, "learning_rate": 6.684918149599655e-06, "loss": 0.9513, "step": 19330 }, { "epoch": 0.35177067607632007, "grad_norm": 1.5638848543167114, "learning_rate": 6.6830429972435264e-06, "loss": 0.9885, "step": 19340 }, { "epoch": 0.35195256370614236, "grad_norm": 2.099931001663208, "learning_rate": 6.681167844887397e-06, "loss": 1.0088, "step": 19350 }, { "epoch": 0.35213445133596466, "grad_norm": 1.8483749628067017, "learning_rate": 6.679292692531268e-06, "loss": 0.9194, "step": 19360 }, { "epoch": 0.35231633896578696, "grad_norm": 1.7257847785949707, "learning_rate": 6.67741754017514e-06, "loss": 0.981, "step": 19370 }, { "epoch": 0.35249822659560925, "grad_norm": 1.852994680404663, "learning_rate": 6.675542387819011e-06, "loss": 0.9343, "step": 19380 }, { "epoch": 0.35268011422543155, "grad_norm": 1.9208077192306519, "learning_rate": 6.6736672354628815e-06, "loss": 0.9661, "step": 19390 }, { "epoch": 0.35286200185525385, "grad_norm": 1.694713830947876, "learning_rate": 6.671792083106753e-06, "loss": 0.9408, "step": 19400 }, { "epoch": 0.35304388948507615, "grad_norm": 2.1170647144317627, "learning_rate": 6.669916930750624e-06, "loss": 0.938, "step": 19410 }, { "epoch": 0.35322577711489844, "grad_norm": 1.6976631879806519, "learning_rate": 6.668041778394495e-06, "loss": 0.9181, "step": 19420 }, { "epoch": 0.3534076647447207, "grad_norm": 1.5510655641555786, "learning_rate": 6.666166626038366e-06, "loss": 0.9841, "step": 19430 }, { "epoch": 0.353589552374543, "grad_norm": 1.8570460081100464, "learning_rate": 6.664291473682237e-06, "loss": 0.9714, "step": 19440 }, { "epoch": 0.3537714400043653, "grad_norm": 1.9883953332901, "learning_rate": 6.662416321326109e-06, "loss": 0.9362, "step": 19450 }, { "epoch": 0.3539533276341876, "grad_norm": 1.8509176969528198, "learning_rate": 6.660541168969979e-06, "loss": 0.9305, "step": 19460 }, { "epoch": 0.3541352152640099, "grad_norm": 1.8314805030822754, "learning_rate": 6.6586660166138505e-06, "loss": 1.0106, "step": 19470 }, { "epoch": 0.35431710289383217, "grad_norm": 1.97035551071167, "learning_rate": 6.656790864257722e-06, "loss": 0.973, "step": 19480 }, { "epoch": 0.35449899052365447, "grad_norm": 1.3411452770233154, "learning_rate": 6.654915711901592e-06, "loss": 0.9337, "step": 19490 }, { "epoch": 0.35468087815347676, "grad_norm": 1.977065920829773, "learning_rate": 6.653040559545464e-06, "loss": 0.8917, "step": 19500 }, { "epoch": 0.35486276578329906, "grad_norm": 1.7520662546157837, "learning_rate": 6.651165407189335e-06, "loss": 0.932, "step": 19510 }, { "epoch": 0.35504465341312136, "grad_norm": 1.6292095184326172, "learning_rate": 6.6492902548332064e-06, "loss": 0.924, "step": 19520 }, { "epoch": 0.35522654104294366, "grad_norm": 1.827060341835022, "learning_rate": 6.647415102477077e-06, "loss": 1.0192, "step": 19530 }, { "epoch": 0.35540842867276595, "grad_norm": 1.8754503726959229, "learning_rate": 6.645539950120948e-06, "loss": 0.9454, "step": 19540 }, { "epoch": 0.35559031630258825, "grad_norm": 1.8662265539169312, "learning_rate": 6.64366479776482e-06, "loss": 0.968, "step": 19550 }, { "epoch": 0.35577220393241055, "grad_norm": 1.783555269241333, "learning_rate": 6.64178964540869e-06, "loss": 0.9491, "step": 19560 }, { "epoch": 0.35595409156223284, "grad_norm": 1.966017484664917, "learning_rate": 6.639914493052561e-06, "loss": 0.9859, "step": 19570 }, { "epoch": 0.35613597919205514, "grad_norm": 1.6040916442871094, "learning_rate": 6.638039340696432e-06, "loss": 0.9802, "step": 19580 }, { "epoch": 0.35631786682187744, "grad_norm": 1.340786337852478, "learning_rate": 6.6361641883403025e-06, "loss": 0.9336, "step": 19590 }, { "epoch": 0.35649975445169974, "grad_norm": 1.9224059581756592, "learning_rate": 6.634289035984174e-06, "loss": 0.9528, "step": 19600 }, { "epoch": 0.35668164208152203, "grad_norm": 1.6971544027328491, "learning_rate": 6.632413883628045e-06, "loss": 0.9914, "step": 19610 }, { "epoch": 0.35686352971134433, "grad_norm": 1.9191011190414429, "learning_rate": 6.630538731271916e-06, "loss": 0.9742, "step": 19620 }, { "epoch": 0.3570454173411666, "grad_norm": 1.6802769899368286, "learning_rate": 6.628663578915787e-06, "loss": 0.9271, "step": 19630 }, { "epoch": 0.3572273049709889, "grad_norm": 2.0676968097686768, "learning_rate": 6.626788426559658e-06, "loss": 0.967, "step": 19640 }, { "epoch": 0.3574091926008112, "grad_norm": 1.9657933712005615, "learning_rate": 6.62491327420353e-06, "loss": 0.9832, "step": 19650 }, { "epoch": 0.3575910802306335, "grad_norm": 1.784475326538086, "learning_rate": 6.6230381218474e-06, "loss": 0.9617, "step": 19660 }, { "epoch": 0.3577729678604558, "grad_norm": 1.7431219816207886, "learning_rate": 6.6211629694912715e-06, "loss": 0.9353, "step": 19670 }, { "epoch": 0.3579548554902781, "grad_norm": 1.7144967317581177, "learning_rate": 6.619287817135143e-06, "loss": 0.8983, "step": 19680 }, { "epoch": 0.3581367431201004, "grad_norm": 1.7171357870101929, "learning_rate": 6.617412664779013e-06, "loss": 0.9446, "step": 19690 }, { "epoch": 0.3583186307499227, "grad_norm": 2.0818846225738525, "learning_rate": 6.615537512422885e-06, "loss": 0.9838, "step": 19700 }, { "epoch": 0.358500518379745, "grad_norm": 1.8617734909057617, "learning_rate": 6.613662360066756e-06, "loss": 0.9781, "step": 19710 }, { "epoch": 0.3586824060095673, "grad_norm": 1.9534735679626465, "learning_rate": 6.6117872077106274e-06, "loss": 0.9619, "step": 19720 }, { "epoch": 0.3588642936393896, "grad_norm": 1.8849050998687744, "learning_rate": 6.609912055354498e-06, "loss": 0.9112, "step": 19730 }, { "epoch": 0.3590461812692119, "grad_norm": 1.6902272701263428, "learning_rate": 6.608036902998369e-06, "loss": 0.9207, "step": 19740 }, { "epoch": 0.3592280688990342, "grad_norm": 1.8684391975402832, "learning_rate": 6.606161750642241e-06, "loss": 0.9288, "step": 19750 }, { "epoch": 0.3594099565288565, "grad_norm": 2.2166008949279785, "learning_rate": 6.604286598286111e-06, "loss": 0.9345, "step": 19760 }, { "epoch": 0.3595918441586788, "grad_norm": 1.9590837955474854, "learning_rate": 6.6024114459299825e-06, "loss": 0.9986, "step": 19770 }, { "epoch": 0.3597737317885011, "grad_norm": 1.9226467609405518, "learning_rate": 6.600536293573854e-06, "loss": 1.0006, "step": 19780 }, { "epoch": 0.3599556194183234, "grad_norm": 1.8462949991226196, "learning_rate": 6.598661141217725e-06, "loss": 0.9498, "step": 19790 }, { "epoch": 0.3601375070481457, "grad_norm": 2.0686347484588623, "learning_rate": 6.596785988861596e-06, "loss": 1.001, "step": 19800 }, { "epoch": 0.360319394677968, "grad_norm": 1.9141149520874023, "learning_rate": 6.594910836505467e-06, "loss": 1.0154, "step": 19810 }, { "epoch": 0.36050128230779027, "grad_norm": 1.8426600694656372, "learning_rate": 6.593035684149338e-06, "loss": 0.9578, "step": 19820 }, { "epoch": 0.36068316993761257, "grad_norm": 1.3228956460952759, "learning_rate": 6.591160531793209e-06, "loss": 0.9828, "step": 19830 }, { "epoch": 0.36086505756743487, "grad_norm": 1.8849674463272095, "learning_rate": 6.58928537943708e-06, "loss": 0.9503, "step": 19840 }, { "epoch": 0.3610469451972571, "grad_norm": 1.6956053972244263, "learning_rate": 6.587410227080951e-06, "loss": 0.9375, "step": 19850 }, { "epoch": 0.3612288328270794, "grad_norm": 1.9367643594741821, "learning_rate": 6.585535074724821e-06, "loss": 0.9734, "step": 19860 }, { "epoch": 0.3614107204569017, "grad_norm": 1.8484889268875122, "learning_rate": 6.5836599223686925e-06, "loss": 1.0258, "step": 19870 }, { "epoch": 0.361592608086724, "grad_norm": 1.8099972009658813, "learning_rate": 6.581784770012564e-06, "loss": 0.9411, "step": 19880 }, { "epoch": 0.3617744957165463, "grad_norm": 1.999625563621521, "learning_rate": 6.579909617656434e-06, "loss": 0.949, "step": 19890 }, { "epoch": 0.3619563833463686, "grad_norm": 1.880810022354126, "learning_rate": 6.578034465300306e-06, "loss": 0.9739, "step": 19900 }, { "epoch": 0.3621382709761909, "grad_norm": 1.7183822393417358, "learning_rate": 6.576159312944177e-06, "loss": 0.9371, "step": 19910 }, { "epoch": 0.3623201586060132, "grad_norm": 1.9615190029144287, "learning_rate": 6.574284160588048e-06, "loss": 0.971, "step": 19920 }, { "epoch": 0.3625020462358355, "grad_norm": 1.533467411994934, "learning_rate": 6.572409008231919e-06, "loss": 0.9537, "step": 19930 }, { "epoch": 0.3626839338656578, "grad_norm": 2.102810859680176, "learning_rate": 6.57053385587579e-06, "loss": 0.9717, "step": 19940 }, { "epoch": 0.3628658214954801, "grad_norm": 1.8652774095535278, "learning_rate": 6.568658703519662e-06, "loss": 0.9512, "step": 19950 }, { "epoch": 0.3630477091253024, "grad_norm": 1.7110936641693115, "learning_rate": 6.566783551163532e-06, "loss": 0.9793, "step": 19960 }, { "epoch": 0.3632295967551247, "grad_norm": 1.8577954769134521, "learning_rate": 6.5649083988074035e-06, "loss": 0.9232, "step": 19970 }, { "epoch": 0.36341148438494697, "grad_norm": 1.8973751068115234, "learning_rate": 6.563033246451275e-06, "loss": 0.9199, "step": 19980 }, { "epoch": 0.36359337201476927, "grad_norm": 2.0244901180267334, "learning_rate": 6.561158094095146e-06, "loss": 0.9056, "step": 19990 }, { "epoch": 0.36377525964459156, "grad_norm": 1.6958764791488647, "learning_rate": 6.559282941739017e-06, "loss": 0.9487, "step": 20000 }, { "epoch": 0.36395714727441386, "grad_norm": 1.6772500276565552, "learning_rate": 6.557407789382888e-06, "loss": 0.9429, "step": 20010 }, { "epoch": 0.36413903490423616, "grad_norm": 1.855457067489624, "learning_rate": 6.555532637026759e-06, "loss": 0.9795, "step": 20020 }, { "epoch": 0.36432092253405846, "grad_norm": 1.5845664739608765, "learning_rate": 6.55365748467063e-06, "loss": 0.927, "step": 20030 }, { "epoch": 0.36450281016388075, "grad_norm": 1.846027135848999, "learning_rate": 6.551782332314501e-06, "loss": 0.9541, "step": 20040 }, { "epoch": 0.36468469779370305, "grad_norm": 1.695282220840454, "learning_rate": 6.5499071799583725e-06, "loss": 0.9042, "step": 20050 }, { "epoch": 0.36486658542352535, "grad_norm": 1.8382288217544556, "learning_rate": 6.548032027602244e-06, "loss": 0.9752, "step": 20060 }, { "epoch": 0.36504847305334764, "grad_norm": 1.868682622909546, "learning_rate": 6.546156875246114e-06, "loss": 0.9941, "step": 20070 }, { "epoch": 0.36523036068316994, "grad_norm": 1.9499953985214233, "learning_rate": 6.544281722889986e-06, "loss": 0.937, "step": 20080 }, { "epoch": 0.36541224831299224, "grad_norm": 1.8389796018600464, "learning_rate": 6.542406570533857e-06, "loss": 0.9592, "step": 20090 }, { "epoch": 0.36559413594281454, "grad_norm": 2.0468904972076416, "learning_rate": 6.5405314181777276e-06, "loss": 0.9128, "step": 20100 }, { "epoch": 0.36577602357263683, "grad_norm": 2.0885043144226074, "learning_rate": 6.538656265821599e-06, "loss": 0.9659, "step": 20110 }, { "epoch": 0.36595791120245913, "grad_norm": 2.0680675506591797, "learning_rate": 6.536781113465469e-06, "loss": 0.9801, "step": 20120 }, { "epoch": 0.3661397988322814, "grad_norm": 1.9860502481460571, "learning_rate": 6.53490596110934e-06, "loss": 0.9728, "step": 20130 }, { "epoch": 0.3663216864621037, "grad_norm": 1.5683821439743042, "learning_rate": 6.533030808753211e-06, "loss": 0.937, "step": 20140 }, { "epoch": 0.366503574091926, "grad_norm": 1.711961269378662, "learning_rate": 6.531155656397083e-06, "loss": 0.9328, "step": 20150 }, { "epoch": 0.3666854617217483, "grad_norm": 1.8659642934799194, "learning_rate": 6.529280504040953e-06, "loss": 0.9524, "step": 20160 }, { "epoch": 0.3668673493515706, "grad_norm": 2.628143787384033, "learning_rate": 6.5274053516848244e-06, "loss": 0.9715, "step": 20170 }, { "epoch": 0.3670492369813929, "grad_norm": 2.067209482192993, "learning_rate": 6.525530199328696e-06, "loss": 0.995, "step": 20180 }, { "epoch": 0.3672311246112152, "grad_norm": 1.7550941705703735, "learning_rate": 6.523655046972567e-06, "loss": 0.9433, "step": 20190 }, { "epoch": 0.3674130122410375, "grad_norm": 1.7698055505752563, "learning_rate": 6.521779894616438e-06, "loss": 0.9845, "step": 20200 }, { "epoch": 0.3675948998708598, "grad_norm": 1.850812554359436, "learning_rate": 6.519904742260309e-06, "loss": 0.9546, "step": 20210 }, { "epoch": 0.3677767875006821, "grad_norm": 1.7769960165023804, "learning_rate": 6.51802958990418e-06, "loss": 0.9749, "step": 20220 }, { "epoch": 0.3679586751305044, "grad_norm": 1.4312729835510254, "learning_rate": 6.516154437548051e-06, "loss": 0.9469, "step": 20230 }, { "epoch": 0.3681405627603267, "grad_norm": 1.7375304698944092, "learning_rate": 6.514279285191922e-06, "loss": 0.9209, "step": 20240 }, { "epoch": 0.368322450390149, "grad_norm": 1.9318410158157349, "learning_rate": 6.5124041328357935e-06, "loss": 0.9681, "step": 20250 }, { "epoch": 0.3685043380199713, "grad_norm": 2.0149803161621094, "learning_rate": 6.510528980479665e-06, "loss": 0.9195, "step": 20260 }, { "epoch": 0.36868622564979353, "grad_norm": 1.648154616355896, "learning_rate": 6.508653828123535e-06, "loss": 0.9797, "step": 20270 }, { "epoch": 0.36886811327961583, "grad_norm": 1.5204428434371948, "learning_rate": 6.506778675767407e-06, "loss": 0.9497, "step": 20280 }, { "epoch": 0.3690500009094381, "grad_norm": 1.9163322448730469, "learning_rate": 6.504903523411278e-06, "loss": 0.9302, "step": 20290 }, { "epoch": 0.3692318885392604, "grad_norm": 1.7874022722244263, "learning_rate": 6.5030283710551486e-06, "loss": 0.9064, "step": 20300 }, { "epoch": 0.3694137761690827, "grad_norm": 1.6145976781845093, "learning_rate": 6.50115321869902e-06, "loss": 0.9593, "step": 20310 }, { "epoch": 0.369595663798905, "grad_norm": 2.081843614578247, "learning_rate": 6.499278066342891e-06, "loss": 0.8867, "step": 20320 }, { "epoch": 0.3697775514287273, "grad_norm": 1.8081263303756714, "learning_rate": 6.497402913986763e-06, "loss": 0.9521, "step": 20330 }, { "epoch": 0.3699594390585496, "grad_norm": 1.7748967409133911, "learning_rate": 6.495527761630633e-06, "loss": 0.9172, "step": 20340 }, { "epoch": 0.3701413266883719, "grad_norm": 1.72898268699646, "learning_rate": 6.4936526092745044e-06, "loss": 0.9247, "step": 20350 }, { "epoch": 0.3703232143181942, "grad_norm": 1.9666167497634888, "learning_rate": 6.491777456918376e-06, "loss": 0.9187, "step": 20360 }, { "epoch": 0.3705051019480165, "grad_norm": 1.9864271879196167, "learning_rate": 6.489902304562246e-06, "loss": 0.9586, "step": 20370 }, { "epoch": 0.3706869895778388, "grad_norm": 1.7449681758880615, "learning_rate": 6.488027152206118e-06, "loss": 0.9461, "step": 20380 }, { "epoch": 0.3708688772076611, "grad_norm": 1.869722604751587, "learning_rate": 6.486151999849989e-06, "loss": 0.9284, "step": 20390 }, { "epoch": 0.3710507648374834, "grad_norm": 1.5209869146347046, "learning_rate": 6.484276847493859e-06, "loss": 0.9524, "step": 20400 }, { "epoch": 0.3712326524673057, "grad_norm": 2.012038230895996, "learning_rate": 6.48240169513773e-06, "loss": 0.9918, "step": 20410 }, { "epoch": 0.371414540097128, "grad_norm": 2.0340659618377686, "learning_rate": 6.480526542781601e-06, "loss": 0.945, "step": 20420 }, { "epoch": 0.3715964277269503, "grad_norm": 1.7765617370605469, "learning_rate": 6.478651390425472e-06, "loss": 0.9673, "step": 20430 }, { "epoch": 0.3717783153567726, "grad_norm": 1.700187087059021, "learning_rate": 6.476776238069343e-06, "loss": 0.9574, "step": 20440 }, { "epoch": 0.3719602029865949, "grad_norm": 1.8520267009735107, "learning_rate": 6.4749010857132145e-06, "loss": 0.9523, "step": 20450 }, { "epoch": 0.3721420906164172, "grad_norm": 1.9015381336212158, "learning_rate": 6.473025933357086e-06, "loss": 0.9604, "step": 20460 }, { "epoch": 0.3723239782462395, "grad_norm": 1.843801736831665, "learning_rate": 6.471150781000956e-06, "loss": 1.0019, "step": 20470 }, { "epoch": 0.37250586587606177, "grad_norm": 1.923731803894043, "learning_rate": 6.469275628644828e-06, "loss": 0.9899, "step": 20480 }, { "epoch": 0.37268775350588407, "grad_norm": 1.5584584474563599, "learning_rate": 6.467400476288699e-06, "loss": 0.9379, "step": 20490 }, { "epoch": 0.37286964113570636, "grad_norm": 1.7982107400894165, "learning_rate": 6.4655253239325695e-06, "loss": 0.9003, "step": 20500 }, { "epoch": 0.37305152876552866, "grad_norm": 1.9607254266738892, "learning_rate": 6.463650171576441e-06, "loss": 0.9717, "step": 20510 }, { "epoch": 0.37323341639535096, "grad_norm": 1.7398099899291992, "learning_rate": 6.461775019220312e-06, "loss": 0.9587, "step": 20520 }, { "epoch": 0.37341530402517326, "grad_norm": 1.6344388723373413, "learning_rate": 6.459899866864184e-06, "loss": 0.9402, "step": 20530 }, { "epoch": 0.37359719165499555, "grad_norm": 1.5545814037322998, "learning_rate": 6.458024714508054e-06, "loss": 0.9241, "step": 20540 }, { "epoch": 0.37377907928481785, "grad_norm": 1.94661283493042, "learning_rate": 6.4561495621519254e-06, "loss": 0.9272, "step": 20550 }, { "epoch": 0.37396096691464015, "grad_norm": 2.1296989917755127, "learning_rate": 6.454274409795797e-06, "loss": 0.9714, "step": 20560 }, { "epoch": 0.37414285454446244, "grad_norm": 1.8342506885528564, "learning_rate": 6.452399257439667e-06, "loss": 0.9542, "step": 20570 }, { "epoch": 0.37432474217428474, "grad_norm": 1.6379458904266357, "learning_rate": 6.450524105083539e-06, "loss": 0.9628, "step": 20580 }, { "epoch": 0.37450662980410704, "grad_norm": 1.6199287176132202, "learning_rate": 6.44864895272741e-06, "loss": 0.9614, "step": 20590 }, { "epoch": 0.37468851743392934, "grad_norm": 1.7503761053085327, "learning_rate": 6.446773800371281e-06, "loss": 1.0194, "step": 20600 }, { "epoch": 0.37487040506375163, "grad_norm": 1.7898787260055542, "learning_rate": 6.444898648015152e-06, "loss": 0.9172, "step": 20610 }, { "epoch": 0.37505229269357393, "grad_norm": 1.6331182718276978, "learning_rate": 6.443023495659023e-06, "loss": 0.9141, "step": 20620 }, { "epoch": 0.3752341803233962, "grad_norm": 1.9052802324295044, "learning_rate": 6.4411483433028945e-06, "loss": 0.9119, "step": 20630 }, { "epoch": 0.3754160679532185, "grad_norm": 1.553583025932312, "learning_rate": 6.439273190946765e-06, "loss": 0.9322, "step": 20640 }, { "epoch": 0.3755979555830408, "grad_norm": 1.7648203372955322, "learning_rate": 6.437398038590636e-06, "loss": 1.0251, "step": 20650 }, { "epoch": 0.3757798432128631, "grad_norm": 1.720937728881836, "learning_rate": 6.435522886234508e-06, "loss": 0.905, "step": 20660 }, { "epoch": 0.3759617308426854, "grad_norm": 2.0783395767211914, "learning_rate": 6.433647733878377e-06, "loss": 0.8988, "step": 20670 }, { "epoch": 0.37614361847250766, "grad_norm": 1.6438990831375122, "learning_rate": 6.431772581522249e-06, "loss": 0.9297, "step": 20680 }, { "epoch": 0.37632550610232995, "grad_norm": 1.8916932344436646, "learning_rate": 6.42989742916612e-06, "loss": 0.9452, "step": 20690 }, { "epoch": 0.37650739373215225, "grad_norm": 1.8049689531326294, "learning_rate": 6.4280222768099905e-06, "loss": 0.948, "step": 20700 }, { "epoch": 0.37668928136197455, "grad_norm": 1.9805068969726562, "learning_rate": 6.426147124453862e-06, "loss": 0.9279, "step": 20710 }, { "epoch": 0.37687116899179685, "grad_norm": 1.9190044403076172, "learning_rate": 6.424271972097733e-06, "loss": 0.9858, "step": 20720 }, { "epoch": 0.37705305662161914, "grad_norm": 1.9311233758926392, "learning_rate": 6.4223968197416046e-06, "loss": 0.9706, "step": 20730 }, { "epoch": 0.37723494425144144, "grad_norm": 1.7969591617584229, "learning_rate": 6.420521667385475e-06, "loss": 0.96, "step": 20740 }, { "epoch": 0.37741683188126374, "grad_norm": 1.6175260543823242, "learning_rate": 6.418646515029346e-06, "loss": 0.9248, "step": 20750 }, { "epoch": 0.37759871951108603, "grad_norm": 1.9255784749984741, "learning_rate": 6.416771362673218e-06, "loss": 0.9414, "step": 20760 }, { "epoch": 0.37778060714090833, "grad_norm": 2.089294672012329, "learning_rate": 6.414896210317088e-06, "loss": 0.9433, "step": 20770 }, { "epoch": 0.37796249477073063, "grad_norm": 2.3150439262390137, "learning_rate": 6.41302105796096e-06, "loss": 1.0043, "step": 20780 }, { "epoch": 0.3781443824005529, "grad_norm": 1.6228495836257935, "learning_rate": 6.411145905604831e-06, "loss": 0.9375, "step": 20790 }, { "epoch": 0.3783262700303752, "grad_norm": 1.681469202041626, "learning_rate": 6.409270753248702e-06, "loss": 0.9534, "step": 20800 }, { "epoch": 0.3785081576601975, "grad_norm": 1.9003984928131104, "learning_rate": 6.407395600892573e-06, "loss": 0.8746, "step": 20810 }, { "epoch": 0.3786900452900198, "grad_norm": 1.7853254079818726, "learning_rate": 6.405520448536444e-06, "loss": 0.9269, "step": 20820 }, { "epoch": 0.3788719329198421, "grad_norm": 1.99488365650177, "learning_rate": 6.4036452961803155e-06, "loss": 0.9555, "step": 20830 }, { "epoch": 0.3790538205496644, "grad_norm": 2.258359432220459, "learning_rate": 6.401770143824186e-06, "loss": 0.9849, "step": 20840 }, { "epoch": 0.3792357081794867, "grad_norm": 1.7384345531463623, "learning_rate": 6.399894991468057e-06, "loss": 0.9455, "step": 20850 }, { "epoch": 0.379417595809309, "grad_norm": 1.8588263988494873, "learning_rate": 6.398019839111929e-06, "loss": 0.9549, "step": 20860 }, { "epoch": 0.3795994834391313, "grad_norm": 1.8968509435653687, "learning_rate": 6.3961446867558e-06, "loss": 0.9546, "step": 20870 }, { "epoch": 0.3797813710689536, "grad_norm": 1.7750802040100098, "learning_rate": 6.3942695343996705e-06, "loss": 0.9099, "step": 20880 }, { "epoch": 0.3799632586987759, "grad_norm": 1.615633487701416, "learning_rate": 6.392394382043542e-06, "loss": 0.9497, "step": 20890 }, { "epoch": 0.3801451463285982, "grad_norm": 1.7936372756958008, "learning_rate": 6.390519229687413e-06, "loss": 1.0196, "step": 20900 }, { "epoch": 0.3803270339584205, "grad_norm": 1.7461737394332886, "learning_rate": 6.388644077331284e-06, "loss": 0.922, "step": 20910 }, { "epoch": 0.3805089215882428, "grad_norm": 1.773910641670227, "learning_rate": 6.386768924975155e-06, "loss": 0.9878, "step": 20920 }, { "epoch": 0.3806908092180651, "grad_norm": 1.8795124292373657, "learning_rate": 6.384893772619026e-06, "loss": 0.974, "step": 20930 }, { "epoch": 0.3808726968478874, "grad_norm": 1.9447557926177979, "learning_rate": 6.383018620262898e-06, "loss": 0.9517, "step": 20940 }, { "epoch": 0.3810545844777097, "grad_norm": 1.9501694440841675, "learning_rate": 6.381143467906767e-06, "loss": 0.931, "step": 20950 }, { "epoch": 0.381236472107532, "grad_norm": 1.5729373693466187, "learning_rate": 6.379268315550639e-06, "loss": 0.9618, "step": 20960 }, { "epoch": 0.3814183597373543, "grad_norm": 1.7744735479354858, "learning_rate": 6.377393163194509e-06, "loss": 0.9682, "step": 20970 }, { "epoch": 0.38160024736717657, "grad_norm": 2.0297963619232178, "learning_rate": 6.375518010838381e-06, "loss": 0.9522, "step": 20980 }, { "epoch": 0.38178213499699887, "grad_norm": 1.9708538055419922, "learning_rate": 6.373642858482252e-06, "loss": 0.9603, "step": 20990 }, { "epoch": 0.38196402262682116, "grad_norm": 1.6064218282699585, "learning_rate": 6.371767706126123e-06, "loss": 0.9297, "step": 21000 }, { "epoch": 0.38214591025664346, "grad_norm": 1.9688817262649536, "learning_rate": 6.369892553769994e-06, "loss": 0.9717, "step": 21010 }, { "epoch": 0.38232779788646576, "grad_norm": 1.6251115798950195, "learning_rate": 6.368017401413865e-06, "loss": 0.9706, "step": 21020 }, { "epoch": 0.38250968551628806, "grad_norm": 1.7978028059005737, "learning_rate": 6.3661422490577365e-06, "loss": 0.9229, "step": 21030 }, { "epoch": 0.38269157314611035, "grad_norm": 1.9247300624847412, "learning_rate": 6.364267096701607e-06, "loss": 0.9564, "step": 21040 }, { "epoch": 0.38287346077593265, "grad_norm": 1.5639925003051758, "learning_rate": 6.362391944345478e-06, "loss": 0.9524, "step": 21050 }, { "epoch": 0.38305534840575495, "grad_norm": 1.824642539024353, "learning_rate": 6.36051679198935e-06, "loss": 0.9651, "step": 21060 }, { "epoch": 0.38323723603557724, "grad_norm": 1.714865803718567, "learning_rate": 6.358641639633221e-06, "loss": 0.9569, "step": 21070 }, { "epoch": 0.38341912366539954, "grad_norm": 1.5053967237472534, "learning_rate": 6.3567664872770915e-06, "loss": 0.941, "step": 21080 }, { "epoch": 0.38360101129522184, "grad_norm": 1.7000781297683716, "learning_rate": 6.354891334920963e-06, "loss": 0.9477, "step": 21090 }, { "epoch": 0.3837828989250441, "grad_norm": 1.6597474813461304, "learning_rate": 6.353016182564834e-06, "loss": 0.9506, "step": 21100 }, { "epoch": 0.3839647865548664, "grad_norm": 2.0158236026763916, "learning_rate": 6.351141030208705e-06, "loss": 0.9319, "step": 21110 }, { "epoch": 0.3841466741846887, "grad_norm": 1.638370394706726, "learning_rate": 6.349265877852576e-06, "loss": 0.9511, "step": 21120 }, { "epoch": 0.38432856181451097, "grad_norm": 1.608657717704773, "learning_rate": 6.347390725496447e-06, "loss": 0.9663, "step": 21130 }, { "epoch": 0.38451044944433327, "grad_norm": 2.0680153369903564, "learning_rate": 6.345515573140319e-06, "loss": 0.9421, "step": 21140 }, { "epoch": 0.38469233707415557, "grad_norm": 1.6429840326309204, "learning_rate": 6.343640420784189e-06, "loss": 0.9154, "step": 21150 }, { "epoch": 0.38487422470397786, "grad_norm": 1.7993927001953125, "learning_rate": 6.341765268428061e-06, "loss": 0.9632, "step": 21160 }, { "epoch": 0.38505611233380016, "grad_norm": 1.8652188777923584, "learning_rate": 6.339890116071932e-06, "loss": 0.932, "step": 21170 }, { "epoch": 0.38523799996362246, "grad_norm": 1.9601812362670898, "learning_rate": 6.3380149637158024e-06, "loss": 0.9283, "step": 21180 }, { "epoch": 0.38541988759344475, "grad_norm": 1.5970784425735474, "learning_rate": 6.336139811359674e-06, "loss": 0.9256, "step": 21190 }, { "epoch": 0.38560177522326705, "grad_norm": 1.6464331150054932, "learning_rate": 6.334264659003545e-06, "loss": 0.9433, "step": 21200 }, { "epoch": 0.38578366285308935, "grad_norm": 1.8009922504425049, "learning_rate": 6.3323895066474165e-06, "loss": 0.9451, "step": 21210 }, { "epoch": 0.38596555048291165, "grad_norm": 1.901947259902954, "learning_rate": 6.330514354291286e-06, "loss": 0.9488, "step": 21220 }, { "epoch": 0.38614743811273394, "grad_norm": 2.981391429901123, "learning_rate": 6.3286392019351575e-06, "loss": 0.9589, "step": 21230 }, { "epoch": 0.38632932574255624, "grad_norm": 1.7724177837371826, "learning_rate": 6.326764049579028e-06, "loss": 0.9738, "step": 21240 }, { "epoch": 0.38651121337237854, "grad_norm": 1.9357391595840454, "learning_rate": 6.324888897222899e-06, "loss": 0.9385, "step": 21250 }, { "epoch": 0.38669310100220083, "grad_norm": 1.7862352132797241, "learning_rate": 6.323013744866771e-06, "loss": 0.9332, "step": 21260 }, { "epoch": 0.38687498863202313, "grad_norm": 1.6897590160369873, "learning_rate": 6.321138592510642e-06, "loss": 0.9442, "step": 21270 }, { "epoch": 0.38705687626184543, "grad_norm": 1.494464635848999, "learning_rate": 6.3192634401545125e-06, "loss": 0.9755, "step": 21280 }, { "epoch": 0.3872387638916677, "grad_norm": 2.153212785720825, "learning_rate": 6.317388287798384e-06, "loss": 0.9467, "step": 21290 }, { "epoch": 0.38742065152149, "grad_norm": 1.5123788118362427, "learning_rate": 6.315513135442255e-06, "loss": 0.9152, "step": 21300 }, { "epoch": 0.3876025391513123, "grad_norm": 1.880175232887268, "learning_rate": 6.313637983086126e-06, "loss": 0.9675, "step": 21310 }, { "epoch": 0.3877844267811346, "grad_norm": 1.7143162488937378, "learning_rate": 6.311762830729997e-06, "loss": 0.9665, "step": 21320 }, { "epoch": 0.3879663144109569, "grad_norm": 1.93434739112854, "learning_rate": 6.309887678373868e-06, "loss": 0.9925, "step": 21330 }, { "epoch": 0.3881482020407792, "grad_norm": 1.8013455867767334, "learning_rate": 6.30801252601774e-06, "loss": 0.9351, "step": 21340 }, { "epoch": 0.3883300896706015, "grad_norm": 1.9994934797286987, "learning_rate": 6.30613737366161e-06, "loss": 0.9781, "step": 21350 }, { "epoch": 0.3885119773004238, "grad_norm": 2.039247989654541, "learning_rate": 6.304262221305482e-06, "loss": 0.9969, "step": 21360 }, { "epoch": 0.3886938649302461, "grad_norm": 1.6346608400344849, "learning_rate": 6.302387068949353e-06, "loss": 0.9866, "step": 21370 }, { "epoch": 0.3888757525600684, "grad_norm": 1.6564929485321045, "learning_rate": 6.3005119165932234e-06, "loss": 0.939, "step": 21380 }, { "epoch": 0.3890576401898907, "grad_norm": 1.8050354719161987, "learning_rate": 6.298636764237095e-06, "loss": 0.8928, "step": 21390 }, { "epoch": 0.389239527819713, "grad_norm": 2.037611722946167, "learning_rate": 6.296761611880966e-06, "loss": 1.0127, "step": 21400 }, { "epoch": 0.3894214154495353, "grad_norm": 2.0243113040924072, "learning_rate": 6.2948864595248375e-06, "loss": 0.9337, "step": 21410 }, { "epoch": 0.3896033030793576, "grad_norm": 1.744860291481018, "learning_rate": 6.293011307168708e-06, "loss": 0.9128, "step": 21420 }, { "epoch": 0.3897851907091799, "grad_norm": 1.8823028802871704, "learning_rate": 6.291136154812579e-06, "loss": 0.9948, "step": 21430 }, { "epoch": 0.3899670783390022, "grad_norm": 2.086362838745117, "learning_rate": 6.289261002456451e-06, "loss": 0.9425, "step": 21440 }, { "epoch": 0.3901489659688245, "grad_norm": 1.596874713897705, "learning_rate": 6.287385850100321e-06, "loss": 0.949, "step": 21450 }, { "epoch": 0.3903308535986468, "grad_norm": 1.595170497894287, "learning_rate": 6.2855106977441925e-06, "loss": 0.9874, "step": 21460 }, { "epoch": 0.3905127412284691, "grad_norm": 1.8800153732299805, "learning_rate": 6.283635545388064e-06, "loss": 0.9796, "step": 21470 }, { "epoch": 0.39069462885829137, "grad_norm": 1.9131873846054077, "learning_rate": 6.281760393031935e-06, "loss": 0.9553, "step": 21480 }, { "epoch": 0.39087651648811367, "grad_norm": 1.7636244297027588, "learning_rate": 6.279885240675806e-06, "loss": 0.9447, "step": 21490 }, { "epoch": 0.39105840411793596, "grad_norm": 1.4394025802612305, "learning_rate": 6.278010088319676e-06, "loss": 0.8917, "step": 21500 }, { "epoch": 0.39124029174775826, "grad_norm": 1.888911247253418, "learning_rate": 6.276134935963547e-06, "loss": 0.9007, "step": 21510 }, { "epoch": 0.3914221793775805, "grad_norm": 1.9244552850723267, "learning_rate": 6.274259783607418e-06, "loss": 0.9663, "step": 21520 }, { "epoch": 0.3916040670074028, "grad_norm": 1.7919135093688965, "learning_rate": 6.272384631251289e-06, "loss": 0.9313, "step": 21530 }, { "epoch": 0.3917859546372251, "grad_norm": 2.0028162002563477, "learning_rate": 6.270509478895161e-06, "loss": 0.9299, "step": 21540 }, { "epoch": 0.3919678422670474, "grad_norm": 1.9816657304763794, "learning_rate": 6.268634326539031e-06, "loss": 0.9387, "step": 21550 }, { "epoch": 0.3921497298968697, "grad_norm": 1.8701618909835815, "learning_rate": 6.266759174182903e-06, "loss": 0.9535, "step": 21560 }, { "epoch": 0.392331617526692, "grad_norm": 1.8784172534942627, "learning_rate": 6.264884021826774e-06, "loss": 0.9625, "step": 21570 }, { "epoch": 0.3925135051565143, "grad_norm": 2.0701005458831787, "learning_rate": 6.2630088694706444e-06, "loss": 0.993, "step": 21580 }, { "epoch": 0.3926953927863366, "grad_norm": 1.8376721143722534, "learning_rate": 6.261133717114516e-06, "loss": 0.9596, "step": 21590 }, { "epoch": 0.3928772804161589, "grad_norm": 2.068044662475586, "learning_rate": 6.259258564758387e-06, "loss": 0.925, "step": 21600 }, { "epoch": 0.3930591680459812, "grad_norm": 3.8939874172210693, "learning_rate": 6.2573834124022585e-06, "loss": 0.9691, "step": 21610 }, { "epoch": 0.3932410556758035, "grad_norm": 1.5158485174179077, "learning_rate": 6.255508260046129e-06, "loss": 0.9634, "step": 21620 }, { "epoch": 0.39342294330562577, "grad_norm": 2.1105868816375732, "learning_rate": 6.25363310769e-06, "loss": 0.9577, "step": 21630 }, { "epoch": 0.39360483093544807, "grad_norm": 1.6287899017333984, "learning_rate": 6.251757955333872e-06, "loss": 0.9487, "step": 21640 }, { "epoch": 0.39378671856527037, "grad_norm": 1.5715581178665161, "learning_rate": 6.249882802977742e-06, "loss": 0.9239, "step": 21650 }, { "epoch": 0.39396860619509266, "grad_norm": 1.8160897493362427, "learning_rate": 6.2480076506216135e-06, "loss": 0.9435, "step": 21660 }, { "epoch": 0.39415049382491496, "grad_norm": 1.930176019668579, "learning_rate": 6.246132498265485e-06, "loss": 0.8901, "step": 21670 }, { "epoch": 0.39433238145473726, "grad_norm": 1.8919724225997925, "learning_rate": 6.244257345909356e-06, "loss": 0.9267, "step": 21680 }, { "epoch": 0.39451426908455955, "grad_norm": 1.6844450235366821, "learning_rate": 6.242382193553227e-06, "loss": 0.9528, "step": 21690 }, { "epoch": 0.39469615671438185, "grad_norm": 1.6391801834106445, "learning_rate": 6.240507041197098e-06, "loss": 0.9365, "step": 21700 }, { "epoch": 0.39487804434420415, "grad_norm": 1.908337950706482, "learning_rate": 6.238631888840969e-06, "loss": 0.8926, "step": 21710 }, { "epoch": 0.39505993197402645, "grad_norm": 1.8427716493606567, "learning_rate": 6.23675673648484e-06, "loss": 0.9286, "step": 21720 }, { "epoch": 0.39524181960384874, "grad_norm": 1.879414439201355, "learning_rate": 6.234881584128711e-06, "loss": 0.9697, "step": 21730 }, { "epoch": 0.39542370723367104, "grad_norm": 1.7199918031692505, "learning_rate": 6.2330064317725826e-06, "loss": 0.9653, "step": 21740 }, { "epoch": 0.39560559486349334, "grad_norm": 2.013948678970337, "learning_rate": 6.231131279416454e-06, "loss": 0.9454, "step": 21750 }, { "epoch": 0.39578748249331563, "grad_norm": 1.7800954580307007, "learning_rate": 6.229256127060324e-06, "loss": 1.0005, "step": 21760 }, { "epoch": 0.39596937012313793, "grad_norm": 1.8713427782058716, "learning_rate": 6.227380974704195e-06, "loss": 0.9953, "step": 21770 }, { "epoch": 0.39615125775296023, "grad_norm": 1.8825263977050781, "learning_rate": 6.225505822348065e-06, "loss": 0.9523, "step": 21780 }, { "epoch": 0.3963331453827825, "grad_norm": 1.8878095149993896, "learning_rate": 6.223630669991937e-06, "loss": 0.9467, "step": 21790 }, { "epoch": 0.3965150330126048, "grad_norm": 1.9152753353118896, "learning_rate": 6.221755517635808e-06, "loss": 0.9719, "step": 21800 }, { "epoch": 0.3966969206424271, "grad_norm": 2.027570962905884, "learning_rate": 6.2198803652796795e-06, "loss": 0.9331, "step": 21810 }, { "epoch": 0.3968788082722494, "grad_norm": 1.869139313697815, "learning_rate": 6.21800521292355e-06, "loss": 0.9681, "step": 21820 }, { "epoch": 0.3970606959020717, "grad_norm": 1.4616950750350952, "learning_rate": 6.216130060567421e-06, "loss": 0.9123, "step": 21830 }, { "epoch": 0.397242583531894, "grad_norm": 1.9072171449661255, "learning_rate": 6.214254908211293e-06, "loss": 0.9477, "step": 21840 }, { "epoch": 0.3974244711617163, "grad_norm": 1.9351584911346436, "learning_rate": 6.212379755855163e-06, "loss": 0.9573, "step": 21850 }, { "epoch": 0.3976063587915386, "grad_norm": 2.007991313934326, "learning_rate": 6.2105046034990345e-06, "loss": 0.9706, "step": 21860 }, { "epoch": 0.3977882464213609, "grad_norm": 2.098093032836914, "learning_rate": 6.208629451142906e-06, "loss": 0.946, "step": 21870 }, { "epoch": 0.3979701340511832, "grad_norm": 1.8840709924697876, "learning_rate": 6.206754298786777e-06, "loss": 0.9617, "step": 21880 }, { "epoch": 0.3981520216810055, "grad_norm": 1.8902676105499268, "learning_rate": 6.204879146430648e-06, "loss": 0.911, "step": 21890 }, { "epoch": 0.3983339093108278, "grad_norm": 1.8380446434020996, "learning_rate": 6.203003994074519e-06, "loss": 0.9642, "step": 21900 }, { "epoch": 0.3985157969406501, "grad_norm": 1.7922409772872925, "learning_rate": 6.20112884171839e-06, "loss": 0.9629, "step": 21910 }, { "epoch": 0.3986976845704724, "grad_norm": 1.6750752925872803, "learning_rate": 6.199253689362261e-06, "loss": 0.9599, "step": 21920 }, { "epoch": 0.3988795722002947, "grad_norm": 2.0001938343048096, "learning_rate": 6.197378537006132e-06, "loss": 1.0037, "step": 21930 }, { "epoch": 0.3990614598301169, "grad_norm": 1.9192121028900146, "learning_rate": 6.1955033846500036e-06, "loss": 0.9575, "step": 21940 }, { "epoch": 0.3992433474599392, "grad_norm": 1.5950777530670166, "learning_rate": 6.193628232293875e-06, "loss": 0.9266, "step": 21950 }, { "epoch": 0.3994252350897615, "grad_norm": 1.994168758392334, "learning_rate": 6.191753079937745e-06, "loss": 0.9295, "step": 21960 }, { "epoch": 0.3996071227195838, "grad_norm": 1.8170902729034424, "learning_rate": 6.189877927581617e-06, "loss": 0.9457, "step": 21970 }, { "epoch": 0.3997890103494061, "grad_norm": 1.921403169631958, "learning_rate": 6.188002775225488e-06, "loss": 0.9603, "step": 21980 }, { "epoch": 0.3999708979792284, "grad_norm": 1.7057158946990967, "learning_rate": 6.186127622869359e-06, "loss": 0.949, "step": 21990 }, { "epoch": 0.4001527856090507, "grad_norm": 1.7338143587112427, "learning_rate": 6.18425247051323e-06, "loss": 0.9758, "step": 22000 }, { "epoch": 0.400334673238873, "grad_norm": 1.886673927307129, "learning_rate": 6.182377318157101e-06, "loss": 0.9895, "step": 22010 }, { "epoch": 0.4005165608686953, "grad_norm": 1.7016479969024658, "learning_rate": 6.180502165800973e-06, "loss": 0.9237, "step": 22020 }, { "epoch": 0.4006984484985176, "grad_norm": 1.962141752243042, "learning_rate": 6.178627013444843e-06, "loss": 0.9302, "step": 22030 }, { "epoch": 0.4008803361283399, "grad_norm": 1.741377353668213, "learning_rate": 6.1767518610887145e-06, "loss": 0.9179, "step": 22040 }, { "epoch": 0.4010622237581622, "grad_norm": 1.598309874534607, "learning_rate": 6.174876708732584e-06, "loss": 0.8707, "step": 22050 }, { "epoch": 0.4012441113879845, "grad_norm": 1.5971617698669434, "learning_rate": 6.1730015563764555e-06, "loss": 0.9013, "step": 22060 }, { "epoch": 0.4014259990178068, "grad_norm": 1.6641191244125366, "learning_rate": 6.171126404020327e-06, "loss": 0.8852, "step": 22070 }, { "epoch": 0.4016078866476291, "grad_norm": 1.9258111715316772, "learning_rate": 6.169251251664198e-06, "loss": 0.939, "step": 22080 }, { "epoch": 0.4017897742774514, "grad_norm": 1.677895188331604, "learning_rate": 6.167376099308069e-06, "loss": 0.9253, "step": 22090 }, { "epoch": 0.4019716619072737, "grad_norm": 1.8439784049987793, "learning_rate": 6.16550094695194e-06, "loss": 0.923, "step": 22100 }, { "epoch": 0.402153549537096, "grad_norm": 1.782586932182312, "learning_rate": 6.163625794595811e-06, "loss": 0.9175, "step": 22110 }, { "epoch": 0.4023354371669183, "grad_norm": 1.8405555486679077, "learning_rate": 6.161750642239682e-06, "loss": 0.992, "step": 22120 }, { "epoch": 0.40251732479674057, "grad_norm": 1.5856215953826904, "learning_rate": 6.159875489883553e-06, "loss": 0.9579, "step": 22130 }, { "epoch": 0.40269921242656287, "grad_norm": 1.6120870113372803, "learning_rate": 6.1580003375274246e-06, "loss": 0.9252, "step": 22140 }, { "epoch": 0.40288110005638517, "grad_norm": 1.5404319763183594, "learning_rate": 6.156125185171296e-06, "loss": 0.9287, "step": 22150 }, { "epoch": 0.40306298768620746, "grad_norm": 1.9407260417938232, "learning_rate": 6.154250032815166e-06, "loss": 0.951, "step": 22160 }, { "epoch": 0.40324487531602976, "grad_norm": 1.862470269203186, "learning_rate": 6.152374880459038e-06, "loss": 0.9546, "step": 22170 }, { "epoch": 0.40342676294585206, "grad_norm": 1.8645063638687134, "learning_rate": 6.150499728102909e-06, "loss": 0.9491, "step": 22180 }, { "epoch": 0.40360865057567435, "grad_norm": 1.5913957357406616, "learning_rate": 6.14862457574678e-06, "loss": 0.9468, "step": 22190 }, { "epoch": 0.40379053820549665, "grad_norm": 1.5572479963302612, "learning_rate": 6.146749423390651e-06, "loss": 0.9786, "step": 22200 }, { "epoch": 0.40397242583531895, "grad_norm": 1.6258306503295898, "learning_rate": 6.144874271034522e-06, "loss": 0.9473, "step": 22210 }, { "epoch": 0.40415431346514125, "grad_norm": 2.0182578563690186, "learning_rate": 6.142999118678394e-06, "loss": 1.0221, "step": 22220 }, { "epoch": 0.40433620109496354, "grad_norm": 1.4751932621002197, "learning_rate": 6.141123966322264e-06, "loss": 0.9649, "step": 22230 }, { "epoch": 0.40451808872478584, "grad_norm": 1.7424744367599487, "learning_rate": 6.1392488139661355e-06, "loss": 0.9715, "step": 22240 }, { "epoch": 0.40469997635460814, "grad_norm": 1.902964472770691, "learning_rate": 6.137373661610007e-06, "loss": 0.9432, "step": 22250 }, { "epoch": 0.40488186398443043, "grad_norm": 1.9100854396820068, "learning_rate": 6.135498509253877e-06, "loss": 0.9532, "step": 22260 }, { "epoch": 0.40506375161425273, "grad_norm": 1.797090768814087, "learning_rate": 6.133623356897749e-06, "loss": 0.9731, "step": 22270 }, { "epoch": 0.40524563924407503, "grad_norm": 1.9406412839889526, "learning_rate": 6.13174820454162e-06, "loss": 0.9287, "step": 22280 }, { "epoch": 0.4054275268738973, "grad_norm": 1.755383849143982, "learning_rate": 6.129873052185491e-06, "loss": 0.9723, "step": 22290 }, { "epoch": 0.4056094145037196, "grad_norm": 1.6863954067230225, "learning_rate": 6.127997899829362e-06, "loss": 0.9216, "step": 22300 }, { "epoch": 0.4057913021335419, "grad_norm": 1.7671152353286743, "learning_rate": 6.126122747473233e-06, "loss": 0.9417, "step": 22310 }, { "epoch": 0.4059731897633642, "grad_norm": 1.8690663576126099, "learning_rate": 6.1242475951171045e-06, "loss": 0.9652, "step": 22320 }, { "epoch": 0.4061550773931865, "grad_norm": 1.750935435295105, "learning_rate": 6.122372442760974e-06, "loss": 0.9454, "step": 22330 }, { "epoch": 0.4063369650230088, "grad_norm": 1.981649398803711, "learning_rate": 6.1204972904048455e-06, "loss": 0.9861, "step": 22340 }, { "epoch": 0.4065188526528311, "grad_norm": 1.5552715063095093, "learning_rate": 6.118622138048717e-06, "loss": 0.9388, "step": 22350 }, { "epoch": 0.40670074028265335, "grad_norm": 1.664083480834961, "learning_rate": 6.116746985692587e-06, "loss": 0.9379, "step": 22360 }, { "epoch": 0.40688262791247565, "grad_norm": 1.737690806388855, "learning_rate": 6.114871833336459e-06, "loss": 0.9522, "step": 22370 }, { "epoch": 0.40706451554229794, "grad_norm": 1.8287957906723022, "learning_rate": 6.11299668098033e-06, "loss": 0.936, "step": 22380 }, { "epoch": 0.40724640317212024, "grad_norm": 1.6600207090377808, "learning_rate": 6.111121528624201e-06, "loss": 0.9813, "step": 22390 }, { "epoch": 0.40742829080194254, "grad_norm": 1.6726586818695068, "learning_rate": 6.109246376268072e-06, "loss": 0.9688, "step": 22400 }, { "epoch": 0.40761017843176484, "grad_norm": 1.8603947162628174, "learning_rate": 6.107371223911943e-06, "loss": 0.952, "step": 22410 }, { "epoch": 0.40779206606158713, "grad_norm": 2.5591893196105957, "learning_rate": 6.105496071555815e-06, "loss": 1.0006, "step": 22420 }, { "epoch": 0.40797395369140943, "grad_norm": 1.7697805166244507, "learning_rate": 6.103620919199685e-06, "loss": 0.9419, "step": 22430 }, { "epoch": 0.4081558413212317, "grad_norm": 1.96444833278656, "learning_rate": 6.1017457668435565e-06, "loss": 0.9486, "step": 22440 }, { "epoch": 0.408337728951054, "grad_norm": 1.9395101070404053, "learning_rate": 6.099870614487428e-06, "loss": 0.8948, "step": 22450 }, { "epoch": 0.4085196165808763, "grad_norm": 1.9680722951889038, "learning_rate": 6.097995462131298e-06, "loss": 0.9875, "step": 22460 }, { "epoch": 0.4087015042106986, "grad_norm": 1.8145864009857178, "learning_rate": 6.09612030977517e-06, "loss": 0.9639, "step": 22470 }, { "epoch": 0.4088833918405209, "grad_norm": 1.803026795387268, "learning_rate": 6.094245157419041e-06, "loss": 0.9936, "step": 22480 }, { "epoch": 0.4090652794703432, "grad_norm": 1.5634328126907349, "learning_rate": 6.092370005062912e-06, "loss": 0.9322, "step": 22490 }, { "epoch": 0.4092471671001655, "grad_norm": 2.0123960971832275, "learning_rate": 6.090494852706783e-06, "loss": 0.9099, "step": 22500 }, { "epoch": 0.4094290547299878, "grad_norm": 1.7367254495620728, "learning_rate": 6.088619700350654e-06, "loss": 0.9677, "step": 22510 }, { "epoch": 0.4096109423598101, "grad_norm": 1.9406116008758545, "learning_rate": 6.0867445479945255e-06, "loss": 0.9286, "step": 22520 }, { "epoch": 0.4097928299896324, "grad_norm": 1.7692252397537231, "learning_rate": 6.084869395638396e-06, "loss": 0.9239, "step": 22530 }, { "epoch": 0.4099747176194547, "grad_norm": 1.799065351486206, "learning_rate": 6.082994243282267e-06, "loss": 1.0147, "step": 22540 }, { "epoch": 0.410156605249277, "grad_norm": 2.069366455078125, "learning_rate": 6.081119090926139e-06, "loss": 0.9433, "step": 22550 }, { "epoch": 0.4103384928790993, "grad_norm": 1.463561773300171, "learning_rate": 6.07924393857001e-06, "loss": 0.9436, "step": 22560 }, { "epoch": 0.4105203805089216, "grad_norm": 1.874988079071045, "learning_rate": 6.077368786213881e-06, "loss": 0.9512, "step": 22570 }, { "epoch": 0.4107022681387439, "grad_norm": 1.8613309860229492, "learning_rate": 6.075493633857752e-06, "loss": 0.9248, "step": 22580 }, { "epoch": 0.4108841557685662, "grad_norm": 1.9254941940307617, "learning_rate": 6.073618481501623e-06, "loss": 0.9831, "step": 22590 }, { "epoch": 0.4110660433983885, "grad_norm": 1.6314630508422852, "learning_rate": 6.071743329145493e-06, "loss": 0.9226, "step": 22600 }, { "epoch": 0.4112479310282108, "grad_norm": 1.7021420001983643, "learning_rate": 6.069868176789364e-06, "loss": 0.9512, "step": 22610 }, { "epoch": 0.4114298186580331, "grad_norm": 1.9547184705734253, "learning_rate": 6.067993024433236e-06, "loss": 0.9668, "step": 22620 }, { "epoch": 0.4116117062878554, "grad_norm": 1.6963647603988647, "learning_rate": 6.066117872077106e-06, "loss": 0.9275, "step": 22630 }, { "epoch": 0.41179359391767767, "grad_norm": 1.7449941635131836, "learning_rate": 6.0642427197209775e-06, "loss": 0.9104, "step": 22640 }, { "epoch": 0.41197548154749997, "grad_norm": 1.7072960138320923, "learning_rate": 6.062367567364849e-06, "loss": 0.9231, "step": 22650 }, { "epoch": 0.41215736917732226, "grad_norm": 1.5960581302642822, "learning_rate": 6.060492415008719e-06, "loss": 0.9072, "step": 22660 }, { "epoch": 0.41233925680714456, "grad_norm": 1.7913954257965088, "learning_rate": 6.058617262652591e-06, "loss": 0.9064, "step": 22670 }, { "epoch": 0.41252114443696686, "grad_norm": 1.7762765884399414, "learning_rate": 6.056742110296462e-06, "loss": 0.9186, "step": 22680 }, { "epoch": 0.41270303206678915, "grad_norm": 1.88654625415802, "learning_rate": 6.054866957940333e-06, "loss": 0.9615, "step": 22690 }, { "epoch": 0.41288491969661145, "grad_norm": 1.7871001958847046, "learning_rate": 6.052991805584204e-06, "loss": 0.9638, "step": 22700 }, { "epoch": 0.41306680732643375, "grad_norm": 1.6648297309875488, "learning_rate": 6.051116653228075e-06, "loss": 0.9447, "step": 22710 }, { "epoch": 0.41324869495625605, "grad_norm": 1.5894651412963867, "learning_rate": 6.0492415008719465e-06, "loss": 0.9419, "step": 22720 }, { "epoch": 0.41343058258607834, "grad_norm": 1.7180280685424805, "learning_rate": 6.047366348515817e-06, "loss": 0.9541, "step": 22730 }, { "epoch": 0.41361247021590064, "grad_norm": 1.506583333015442, "learning_rate": 6.045491196159688e-06, "loss": 0.9106, "step": 22740 }, { "epoch": 0.41379435784572294, "grad_norm": 1.9351799488067627, "learning_rate": 6.04361604380356e-06, "loss": 0.9414, "step": 22750 }, { "epoch": 0.41397624547554523, "grad_norm": 1.4377590417861938, "learning_rate": 6.041740891447431e-06, "loss": 0.9389, "step": 22760 }, { "epoch": 0.4141581331053675, "grad_norm": 1.5525861978530884, "learning_rate": 6.0398657390913016e-06, "loss": 0.9219, "step": 22770 }, { "epoch": 0.4143400207351898, "grad_norm": 1.5868505239486694, "learning_rate": 6.037990586735173e-06, "loss": 0.9829, "step": 22780 }, { "epoch": 0.41452190836501207, "grad_norm": 1.905914306640625, "learning_rate": 6.036115434379044e-06, "loss": 1.0056, "step": 22790 }, { "epoch": 0.41470379599483437, "grad_norm": 2.017577886581421, "learning_rate": 6.034240282022915e-06, "loss": 0.9603, "step": 22800 }, { "epoch": 0.41488568362465666, "grad_norm": 1.707759976387024, "learning_rate": 6.032365129666786e-06, "loss": 0.9578, "step": 22810 }, { "epoch": 0.41506757125447896, "grad_norm": 1.9023910760879517, "learning_rate": 6.0304899773106575e-06, "loss": 0.9564, "step": 22820 }, { "epoch": 0.41524945888430126, "grad_norm": 1.861293911933899, "learning_rate": 6.028614824954529e-06, "loss": 0.9529, "step": 22830 }, { "epoch": 0.41543134651412356, "grad_norm": 1.570849061012268, "learning_rate": 6.026739672598399e-06, "loss": 0.9186, "step": 22840 }, { "epoch": 0.41561323414394585, "grad_norm": 1.6846044063568115, "learning_rate": 6.024864520242271e-06, "loss": 0.9987, "step": 22850 }, { "epoch": 0.41579512177376815, "grad_norm": 1.5576562881469727, "learning_rate": 6.022989367886142e-06, "loss": 0.9031, "step": 22860 }, { "epoch": 0.41597700940359045, "grad_norm": 1.668095588684082, "learning_rate": 6.0211142155300125e-06, "loss": 0.9606, "step": 22870 }, { "epoch": 0.41615889703341274, "grad_norm": 1.7058451175689697, "learning_rate": 6.019239063173883e-06, "loss": 0.9486, "step": 22880 }, { "epoch": 0.41634078466323504, "grad_norm": 1.8578392267227173, "learning_rate": 6.017363910817754e-06, "loss": 0.9344, "step": 22890 }, { "epoch": 0.41652267229305734, "grad_norm": 1.9656736850738525, "learning_rate": 6.015488758461625e-06, "loss": 0.9032, "step": 22900 }, { "epoch": 0.41670455992287964, "grad_norm": 1.9480291604995728, "learning_rate": 6.013613606105496e-06, "loss": 1.017, "step": 22910 }, { "epoch": 0.41688644755270193, "grad_norm": 1.719743013381958, "learning_rate": 6.0117384537493675e-06, "loss": 0.9488, "step": 22920 }, { "epoch": 0.41706833518252423, "grad_norm": 1.814772129058838, "learning_rate": 6.009863301393238e-06, "loss": 0.9722, "step": 22930 }, { "epoch": 0.4172502228123465, "grad_norm": 1.9117565155029297, "learning_rate": 6.007988149037109e-06, "loss": 0.9557, "step": 22940 }, { "epoch": 0.4174321104421688, "grad_norm": 1.6588844060897827, "learning_rate": 6.006112996680981e-06, "loss": 0.9621, "step": 22950 }, { "epoch": 0.4176139980719911, "grad_norm": 1.6730375289916992, "learning_rate": 6.004237844324852e-06, "loss": 0.9607, "step": 22960 }, { "epoch": 0.4177958857018134, "grad_norm": 1.5419020652770996, "learning_rate": 6.0023626919687226e-06, "loss": 0.9137, "step": 22970 }, { "epoch": 0.4179777733316357, "grad_norm": 1.9511784315109253, "learning_rate": 6.000487539612594e-06, "loss": 1.0098, "step": 22980 }, { "epoch": 0.418159660961458, "grad_norm": 1.8249188661575317, "learning_rate": 5.998612387256465e-06, "loss": 0.9164, "step": 22990 }, { "epoch": 0.4183415485912803, "grad_norm": 3.311690092086792, "learning_rate": 5.996737234900336e-06, "loss": 0.9074, "step": 23000 }, { "epoch": 0.4185234362211026, "grad_norm": 1.9374357461929321, "learning_rate": 5.994862082544207e-06, "loss": 0.9974, "step": 23010 }, { "epoch": 0.4187053238509249, "grad_norm": 1.4140807390213013, "learning_rate": 5.9929869301880784e-06, "loss": 0.8839, "step": 23020 }, { "epoch": 0.4188872114807472, "grad_norm": 1.6931018829345703, "learning_rate": 5.99111177783195e-06, "loss": 0.9413, "step": 23030 }, { "epoch": 0.4190690991105695, "grad_norm": 1.8674830198287964, "learning_rate": 5.98923662547582e-06, "loss": 0.933, "step": 23040 }, { "epoch": 0.4192509867403918, "grad_norm": 1.8617889881134033, "learning_rate": 5.987361473119692e-06, "loss": 0.9733, "step": 23050 }, { "epoch": 0.4194328743702141, "grad_norm": 1.6805315017700195, "learning_rate": 5.985486320763563e-06, "loss": 0.9191, "step": 23060 }, { "epoch": 0.4196147620000364, "grad_norm": 1.6826789379119873, "learning_rate": 5.9836111684074335e-06, "loss": 0.934, "step": 23070 }, { "epoch": 0.4197966496298587, "grad_norm": 1.6881111860275269, "learning_rate": 5.981736016051305e-06, "loss": 0.9486, "step": 23080 }, { "epoch": 0.419978537259681, "grad_norm": 1.7724345922470093, "learning_rate": 5.979860863695176e-06, "loss": 0.9147, "step": 23090 }, { "epoch": 0.4201604248895033, "grad_norm": 1.7864292860031128, "learning_rate": 5.9779857113390475e-06, "loss": 0.8997, "step": 23100 }, { "epoch": 0.4203423125193256, "grad_norm": 1.6389120817184448, "learning_rate": 5.976110558982918e-06, "loss": 0.9156, "step": 23110 }, { "epoch": 0.4205242001491479, "grad_norm": 1.8215222358703613, "learning_rate": 5.974235406626789e-06, "loss": 0.9356, "step": 23120 }, { "epoch": 0.4207060877789702, "grad_norm": 1.7632427215576172, "learning_rate": 5.972360254270661e-06, "loss": 0.9116, "step": 23130 }, { "epoch": 0.42088797540879247, "grad_norm": 1.6765843629837036, "learning_rate": 5.970485101914531e-06, "loss": 0.9202, "step": 23140 }, { "epoch": 0.42106986303861477, "grad_norm": 1.7467560768127441, "learning_rate": 5.968609949558402e-06, "loss": 0.9665, "step": 23150 }, { "epoch": 0.42125175066843706, "grad_norm": 1.5864354372024536, "learning_rate": 5.966734797202273e-06, "loss": 0.9602, "step": 23160 }, { "epoch": 0.42143363829825936, "grad_norm": 2.029357433319092, "learning_rate": 5.9648596448461436e-06, "loss": 0.9333, "step": 23170 }, { "epoch": 0.42161552592808166, "grad_norm": 1.805400013923645, "learning_rate": 5.962984492490015e-06, "loss": 0.9633, "step": 23180 }, { "epoch": 0.4217974135579039, "grad_norm": 1.4319820404052734, "learning_rate": 5.961109340133886e-06, "loss": 0.906, "step": 23190 }, { "epoch": 0.4219793011877262, "grad_norm": 1.9155900478363037, "learning_rate": 5.959234187777757e-06, "loss": 0.972, "step": 23200 }, { "epoch": 0.4221611888175485, "grad_norm": 1.978089451789856, "learning_rate": 5.957359035421628e-06, "loss": 0.8894, "step": 23210 }, { "epoch": 0.4223430764473708, "grad_norm": 1.9235475063323975, "learning_rate": 5.9554838830654994e-06, "loss": 0.9187, "step": 23220 }, { "epoch": 0.4225249640771931, "grad_norm": 1.7764322757720947, "learning_rate": 5.953608730709371e-06, "loss": 0.9665, "step": 23230 }, { "epoch": 0.4227068517070154, "grad_norm": 1.7964248657226562, "learning_rate": 5.951733578353241e-06, "loss": 1.0108, "step": 23240 }, { "epoch": 0.4228887393368377, "grad_norm": 1.6739617586135864, "learning_rate": 5.949858425997113e-06, "loss": 0.9464, "step": 23250 }, { "epoch": 0.42307062696666, "grad_norm": 1.7193934917449951, "learning_rate": 5.947983273640984e-06, "loss": 0.9127, "step": 23260 }, { "epoch": 0.4232525145964823, "grad_norm": 1.9022221565246582, "learning_rate": 5.9461081212848545e-06, "loss": 0.9644, "step": 23270 }, { "epoch": 0.4234344022263046, "grad_norm": 1.8553431034088135, "learning_rate": 5.944232968928726e-06, "loss": 0.9237, "step": 23280 }, { "epoch": 0.42361628985612687, "grad_norm": 1.9243379831314087, "learning_rate": 5.942357816572597e-06, "loss": 0.9442, "step": 23290 }, { "epoch": 0.42379817748594917, "grad_norm": 1.7660346031188965, "learning_rate": 5.9404826642164685e-06, "loss": 0.9495, "step": 23300 }, { "epoch": 0.42398006511577147, "grad_norm": 1.7359954118728638, "learning_rate": 5.938607511860339e-06, "loss": 0.9508, "step": 23310 }, { "epoch": 0.42416195274559376, "grad_norm": 1.6419157981872559, "learning_rate": 5.93673235950421e-06, "loss": 0.9787, "step": 23320 }, { "epoch": 0.42434384037541606, "grad_norm": 1.863930106163025, "learning_rate": 5.934857207148082e-06, "loss": 0.9553, "step": 23330 }, { "epoch": 0.42452572800523836, "grad_norm": 1.6468404531478882, "learning_rate": 5.932982054791952e-06, "loss": 0.8919, "step": 23340 }, { "epoch": 0.42470761563506065, "grad_norm": 2.275084972381592, "learning_rate": 5.9311069024358235e-06, "loss": 0.987, "step": 23350 }, { "epoch": 0.42488950326488295, "grad_norm": 1.934873342514038, "learning_rate": 5.929231750079695e-06, "loss": 0.983, "step": 23360 }, { "epoch": 0.42507139089470525, "grad_norm": 1.5639519691467285, "learning_rate": 5.927356597723566e-06, "loss": 0.9366, "step": 23370 }, { "epoch": 0.42525327852452754, "grad_norm": 1.7250670194625854, "learning_rate": 5.925481445367437e-06, "loss": 0.9538, "step": 23380 }, { "epoch": 0.42543516615434984, "grad_norm": 1.8523356914520264, "learning_rate": 5.923606293011308e-06, "loss": 0.9238, "step": 23390 }, { "epoch": 0.42561705378417214, "grad_norm": 1.9187796115875244, "learning_rate": 5.9217311406551794e-06, "loss": 0.9565, "step": 23400 }, { "epoch": 0.42579894141399444, "grad_norm": 1.7960031032562256, "learning_rate": 5.91985598829905e-06, "loss": 0.9366, "step": 23410 }, { "epoch": 0.42598082904381673, "grad_norm": 2.0311131477355957, "learning_rate": 5.917980835942921e-06, "loss": 0.962, "step": 23420 }, { "epoch": 0.42616271667363903, "grad_norm": 1.4386647939682007, "learning_rate": 5.916105683586792e-06, "loss": 0.9166, "step": 23430 }, { "epoch": 0.4263446043034613, "grad_norm": 1.9514163732528687, "learning_rate": 5.914230531230662e-06, "loss": 0.91, "step": 23440 }, { "epoch": 0.4265264919332836, "grad_norm": 1.4519870281219482, "learning_rate": 5.912355378874534e-06, "loss": 0.9378, "step": 23450 }, { "epoch": 0.4267083795631059, "grad_norm": 1.6975163221359253, "learning_rate": 5.910480226518405e-06, "loss": 0.9722, "step": 23460 }, { "epoch": 0.4268902671929282, "grad_norm": 2.0366060733795166, "learning_rate": 5.9086050741622755e-06, "loss": 0.954, "step": 23470 }, { "epoch": 0.4270721548227505, "grad_norm": 1.8747756481170654, "learning_rate": 5.906729921806147e-06, "loss": 1.0096, "step": 23480 }, { "epoch": 0.4272540424525728, "grad_norm": 1.9512778520584106, "learning_rate": 5.904854769450018e-06, "loss": 0.9274, "step": 23490 }, { "epoch": 0.4274359300823951, "grad_norm": 1.7077785730361938, "learning_rate": 5.9029796170938895e-06, "loss": 0.9575, "step": 23500 }, { "epoch": 0.4276178177122174, "grad_norm": 1.7355226278305054, "learning_rate": 5.90110446473776e-06, "loss": 0.9691, "step": 23510 }, { "epoch": 0.4277997053420397, "grad_norm": 1.8634899854660034, "learning_rate": 5.899229312381631e-06, "loss": 0.8993, "step": 23520 }, { "epoch": 0.427981592971862, "grad_norm": 1.7795253992080688, "learning_rate": 5.897354160025503e-06, "loss": 0.9328, "step": 23530 }, { "epoch": 0.4281634806016843, "grad_norm": 1.9061211347579956, "learning_rate": 5.895479007669373e-06, "loss": 0.9663, "step": 23540 }, { "epoch": 0.4283453682315066, "grad_norm": 1.6887754201889038, "learning_rate": 5.8936038553132445e-06, "loss": 0.9411, "step": 23550 }, { "epoch": 0.4285272558613289, "grad_norm": 1.8195197582244873, "learning_rate": 5.891728702957116e-06, "loss": 0.9392, "step": 23560 }, { "epoch": 0.4287091434911512, "grad_norm": 1.8348307609558105, "learning_rate": 5.889853550600987e-06, "loss": 0.9229, "step": 23570 }, { "epoch": 0.4288910311209735, "grad_norm": 1.5944244861602783, "learning_rate": 5.887978398244858e-06, "loss": 0.9554, "step": 23580 }, { "epoch": 0.4290729187507958, "grad_norm": 1.8346295356750488, "learning_rate": 5.886103245888729e-06, "loss": 0.9559, "step": 23590 }, { "epoch": 0.4292548063806181, "grad_norm": 1.9847493171691895, "learning_rate": 5.8842280935326e-06, "loss": 0.981, "step": 23600 }, { "epoch": 0.4294366940104403, "grad_norm": 2.0419065952301025, "learning_rate": 5.882352941176471e-06, "loss": 0.9433, "step": 23610 }, { "epoch": 0.4296185816402626, "grad_norm": 1.7267695665359497, "learning_rate": 5.880477788820342e-06, "loss": 0.9425, "step": 23620 }, { "epoch": 0.4298004692700849, "grad_norm": 1.790099024772644, "learning_rate": 5.878602636464214e-06, "loss": 0.9595, "step": 23630 }, { "epoch": 0.4299823568999072, "grad_norm": 1.7923002243041992, "learning_rate": 5.876727484108085e-06, "loss": 0.9559, "step": 23640 }, { "epoch": 0.4301642445297295, "grad_norm": 2.2292892932891846, "learning_rate": 5.8748523317519555e-06, "loss": 1.0014, "step": 23650 }, { "epoch": 0.4303461321595518, "grad_norm": 1.7498859167099, "learning_rate": 5.872977179395827e-06, "loss": 0.8888, "step": 23660 }, { "epoch": 0.4305280197893741, "grad_norm": 1.8529729843139648, "learning_rate": 5.871102027039698e-06, "loss": 0.9451, "step": 23670 }, { "epoch": 0.4307099074191964, "grad_norm": 2.024423360824585, "learning_rate": 5.869226874683569e-06, "loss": 0.9644, "step": 23680 }, { "epoch": 0.4308917950490187, "grad_norm": 1.67050302028656, "learning_rate": 5.86735172232744e-06, "loss": 0.8918, "step": 23690 }, { "epoch": 0.431073682678841, "grad_norm": 1.7207499742507935, "learning_rate": 5.8654765699713105e-06, "loss": 0.9025, "step": 23700 }, { "epoch": 0.4312555703086633, "grad_norm": 1.6824506521224976, "learning_rate": 5.863601417615181e-06, "loss": 0.8958, "step": 23710 }, { "epoch": 0.4314374579384856, "grad_norm": 1.6344856023788452, "learning_rate": 5.861726265259052e-06, "loss": 0.9508, "step": 23720 }, { "epoch": 0.4316193455683079, "grad_norm": 1.9322179555892944, "learning_rate": 5.859851112902924e-06, "loss": 1.0383, "step": 23730 }, { "epoch": 0.4318012331981302, "grad_norm": 1.8529424667358398, "learning_rate": 5.857975960546794e-06, "loss": 0.9424, "step": 23740 }, { "epoch": 0.4319831208279525, "grad_norm": 1.8732798099517822, "learning_rate": 5.8561008081906655e-06, "loss": 0.9742, "step": 23750 }, { "epoch": 0.4321650084577748, "grad_norm": 1.6418994665145874, "learning_rate": 5.854225655834537e-06, "loss": 0.8892, "step": 23760 }, { "epoch": 0.4323468960875971, "grad_norm": 1.917565107345581, "learning_rate": 5.852350503478408e-06, "loss": 0.9769, "step": 23770 }, { "epoch": 0.4325287837174194, "grad_norm": 1.7532726526260376, "learning_rate": 5.850475351122279e-06, "loss": 0.8989, "step": 23780 }, { "epoch": 0.43271067134724167, "grad_norm": 2.052868366241455, "learning_rate": 5.84860019876615e-06, "loss": 0.9616, "step": 23790 }, { "epoch": 0.43289255897706397, "grad_norm": 1.9024713039398193, "learning_rate": 5.846725046410021e-06, "loss": 0.9291, "step": 23800 }, { "epoch": 0.43307444660688627, "grad_norm": 1.593246579170227, "learning_rate": 5.844849894053892e-06, "loss": 0.8837, "step": 23810 }, { "epoch": 0.43325633423670856, "grad_norm": 1.8394228219985962, "learning_rate": 5.842974741697763e-06, "loss": 0.9913, "step": 23820 }, { "epoch": 0.43343822186653086, "grad_norm": 2.193136692047119, "learning_rate": 5.841099589341635e-06, "loss": 1.0084, "step": 23830 }, { "epoch": 0.43362010949635316, "grad_norm": 1.8788574934005737, "learning_rate": 5.839224436985506e-06, "loss": 0.9426, "step": 23840 }, { "epoch": 0.43380199712617545, "grad_norm": 1.8859025239944458, "learning_rate": 5.8373492846293764e-06, "loss": 0.9618, "step": 23850 }, { "epoch": 0.43398388475599775, "grad_norm": 1.6324436664581299, "learning_rate": 5.835474132273248e-06, "loss": 0.9893, "step": 23860 }, { "epoch": 0.43416577238582005, "grad_norm": 1.645495891571045, "learning_rate": 5.833598979917119e-06, "loss": 0.9884, "step": 23870 }, { "epoch": 0.43434766001564235, "grad_norm": 1.864006519317627, "learning_rate": 5.83172382756099e-06, "loss": 0.9714, "step": 23880 }, { "epoch": 0.43452954764546464, "grad_norm": 2.106480598449707, "learning_rate": 5.829848675204861e-06, "loss": 1.039, "step": 23890 }, { "epoch": 0.43471143527528694, "grad_norm": 1.8152413368225098, "learning_rate": 5.827973522848732e-06, "loss": 0.9717, "step": 23900 }, { "epoch": 0.43489332290510924, "grad_norm": 1.7892295122146606, "learning_rate": 5.826098370492604e-06, "loss": 0.9592, "step": 23910 }, { "epoch": 0.43507521053493153, "grad_norm": 1.7493058443069458, "learning_rate": 5.824223218136474e-06, "loss": 0.865, "step": 23920 }, { "epoch": 0.43525709816475383, "grad_norm": 1.8927497863769531, "learning_rate": 5.8223480657803455e-06, "loss": 0.9419, "step": 23930 }, { "epoch": 0.4354389857945761, "grad_norm": 1.6567891836166382, "learning_rate": 5.820472913424217e-06, "loss": 0.9289, "step": 23940 }, { "epoch": 0.4356208734243984, "grad_norm": 1.8641098737716675, "learning_rate": 5.818597761068087e-06, "loss": 0.935, "step": 23950 }, { "epoch": 0.4358027610542207, "grad_norm": 1.81911039352417, "learning_rate": 5.816722608711959e-06, "loss": 0.9022, "step": 23960 }, { "epoch": 0.435984648684043, "grad_norm": 1.7088016271591187, "learning_rate": 5.81484745635583e-06, "loss": 0.9586, "step": 23970 }, { "epoch": 0.4361665363138653, "grad_norm": 1.745242714881897, "learning_rate": 5.8129723039997e-06, "loss": 0.9484, "step": 23980 }, { "epoch": 0.4363484239436876, "grad_norm": 1.885690450668335, "learning_rate": 5.811097151643571e-06, "loss": 0.9551, "step": 23990 }, { "epoch": 0.4365303115735099, "grad_norm": 1.7696515321731567, "learning_rate": 5.809221999287442e-06, "loss": 0.9637, "step": 24000 }, { "epoch": 0.4367121992033322, "grad_norm": 1.8968489170074463, "learning_rate": 5.807346846931313e-06, "loss": 0.9737, "step": 24010 }, { "epoch": 0.4368940868331545, "grad_norm": 2.0199108123779297, "learning_rate": 5.805471694575184e-06, "loss": 0.9364, "step": 24020 }, { "epoch": 0.43707597446297675, "grad_norm": 1.8231583833694458, "learning_rate": 5.803596542219056e-06, "loss": 0.9677, "step": 24030 }, { "epoch": 0.43725786209279904, "grad_norm": 1.8977023363113403, "learning_rate": 5.801721389862927e-06, "loss": 0.9228, "step": 24040 }, { "epoch": 0.43743974972262134, "grad_norm": 1.7747828960418701, "learning_rate": 5.7998462375067974e-06, "loss": 0.9355, "step": 24050 }, { "epoch": 0.43762163735244364, "grad_norm": 1.898495078086853, "learning_rate": 5.797971085150669e-06, "loss": 0.952, "step": 24060 }, { "epoch": 0.43780352498226593, "grad_norm": 1.8068677186965942, "learning_rate": 5.79609593279454e-06, "loss": 0.9735, "step": 24070 }, { "epoch": 0.43798541261208823, "grad_norm": 2.0763535499572754, "learning_rate": 5.794220780438411e-06, "loss": 0.9733, "step": 24080 }, { "epoch": 0.43816730024191053, "grad_norm": 2.0287771224975586, "learning_rate": 5.792345628082282e-06, "loss": 0.9098, "step": 24090 }, { "epoch": 0.4383491878717328, "grad_norm": 1.6352459192276, "learning_rate": 5.790470475726153e-06, "loss": 0.947, "step": 24100 }, { "epoch": 0.4385310755015551, "grad_norm": 1.720505952835083, "learning_rate": 5.788595323370025e-06, "loss": 0.8982, "step": 24110 }, { "epoch": 0.4387129631313774, "grad_norm": 4.568820476531982, "learning_rate": 5.786720171013895e-06, "loss": 0.9502, "step": 24120 }, { "epoch": 0.4388948507611997, "grad_norm": 1.6555933952331543, "learning_rate": 5.7848450186577665e-06, "loss": 0.9084, "step": 24130 }, { "epoch": 0.439076738391022, "grad_norm": 1.6641035079956055, "learning_rate": 5.782969866301638e-06, "loss": 0.925, "step": 24140 }, { "epoch": 0.4392586260208443, "grad_norm": 1.8679566383361816, "learning_rate": 5.781094713945508e-06, "loss": 0.9336, "step": 24150 }, { "epoch": 0.4394405136506666, "grad_norm": 1.6929429769515991, "learning_rate": 5.77921956158938e-06, "loss": 0.9645, "step": 24160 }, { "epoch": 0.4396224012804889, "grad_norm": 1.7856169939041138, "learning_rate": 5.777344409233251e-06, "loss": 0.9575, "step": 24170 }, { "epoch": 0.4398042889103112, "grad_norm": 1.7718777656555176, "learning_rate": 5.775469256877122e-06, "loss": 1.0009, "step": 24180 }, { "epoch": 0.4399861765401335, "grad_norm": 1.8451727628707886, "learning_rate": 5.773594104520993e-06, "loss": 0.926, "step": 24190 }, { "epoch": 0.4401680641699558, "grad_norm": 1.5885201692581177, "learning_rate": 5.771718952164864e-06, "loss": 0.9187, "step": 24200 }, { "epoch": 0.4403499517997781, "grad_norm": 1.7429572343826294, "learning_rate": 5.769843799808736e-06, "loss": 0.9499, "step": 24210 }, { "epoch": 0.4405318394296004, "grad_norm": 1.8374218940734863, "learning_rate": 5.767968647452606e-06, "loss": 0.924, "step": 24220 }, { "epoch": 0.4407137270594227, "grad_norm": 1.906712532043457, "learning_rate": 5.7660934950964774e-06, "loss": 0.9416, "step": 24230 }, { "epoch": 0.440895614689245, "grad_norm": 1.5675550699234009, "learning_rate": 5.764218342740349e-06, "loss": 0.9132, "step": 24240 }, { "epoch": 0.4410775023190673, "grad_norm": 1.5951001644134521, "learning_rate": 5.76234319038422e-06, "loss": 0.941, "step": 24250 }, { "epoch": 0.4412593899488896, "grad_norm": 1.5989505052566528, "learning_rate": 5.76046803802809e-06, "loss": 0.9449, "step": 24260 }, { "epoch": 0.4414412775787119, "grad_norm": 1.4987530708312988, "learning_rate": 5.758592885671961e-06, "loss": 0.9095, "step": 24270 }, { "epoch": 0.4416231652085342, "grad_norm": 1.9366507530212402, "learning_rate": 5.756717733315832e-06, "loss": 0.96, "step": 24280 }, { "epoch": 0.44180505283835647, "grad_norm": 1.5156174898147583, "learning_rate": 5.754842580959703e-06, "loss": 0.9693, "step": 24290 }, { "epoch": 0.44198694046817877, "grad_norm": 2.0374417304992676, "learning_rate": 5.752967428603574e-06, "loss": 0.9128, "step": 24300 }, { "epoch": 0.44216882809800107, "grad_norm": 1.8761966228485107, "learning_rate": 5.751092276247446e-06, "loss": 0.9476, "step": 24310 }, { "epoch": 0.44235071572782336, "grad_norm": 1.8373512029647827, "learning_rate": 5.749217123891316e-06, "loss": 0.9603, "step": 24320 }, { "epoch": 0.44253260335764566, "grad_norm": 1.831644058227539, "learning_rate": 5.7473419715351875e-06, "loss": 1.01, "step": 24330 }, { "epoch": 0.44271449098746796, "grad_norm": 1.5575881004333496, "learning_rate": 5.745466819179059e-06, "loss": 0.9249, "step": 24340 }, { "epoch": 0.44289637861729025, "grad_norm": 1.8411716222763062, "learning_rate": 5.743591666822929e-06, "loss": 0.9176, "step": 24350 }, { "epoch": 0.44307826624711255, "grad_norm": 1.5131239891052246, "learning_rate": 5.741716514466801e-06, "loss": 0.9029, "step": 24360 }, { "epoch": 0.44326015387693485, "grad_norm": 1.855910301208496, "learning_rate": 5.739841362110672e-06, "loss": 0.9678, "step": 24370 }, { "epoch": 0.44344204150675715, "grad_norm": 1.3760943412780762, "learning_rate": 5.737966209754543e-06, "loss": 0.9216, "step": 24380 }, { "epoch": 0.44362392913657944, "grad_norm": 1.5986158847808838, "learning_rate": 5.736091057398414e-06, "loss": 0.9324, "step": 24390 }, { "epoch": 0.44380581676640174, "grad_norm": 1.9240859746932983, "learning_rate": 5.734215905042285e-06, "loss": 0.8943, "step": 24400 }, { "epoch": 0.44398770439622404, "grad_norm": 1.975664734840393, "learning_rate": 5.732340752686157e-06, "loss": 0.9634, "step": 24410 }, { "epoch": 0.44416959202604633, "grad_norm": 1.9624677896499634, "learning_rate": 5.730465600330027e-06, "loss": 0.9599, "step": 24420 }, { "epoch": 0.44435147965586863, "grad_norm": 1.5714447498321533, "learning_rate": 5.7285904479738984e-06, "loss": 0.9501, "step": 24430 }, { "epoch": 0.4445333672856909, "grad_norm": 1.731459379196167, "learning_rate": 5.72671529561777e-06, "loss": 0.9421, "step": 24440 }, { "epoch": 0.44471525491551317, "grad_norm": 1.8668715953826904, "learning_rate": 5.724840143261641e-06, "loss": 0.9214, "step": 24450 }, { "epoch": 0.44489714254533547, "grad_norm": 1.668696403503418, "learning_rate": 5.722964990905512e-06, "loss": 0.9073, "step": 24460 }, { "epoch": 0.44507903017515776, "grad_norm": 1.8570729494094849, "learning_rate": 5.721089838549383e-06, "loss": 0.9584, "step": 24470 }, { "epoch": 0.44526091780498006, "grad_norm": 2.1509225368499756, "learning_rate": 5.719214686193254e-06, "loss": 0.9507, "step": 24480 }, { "epoch": 0.44544280543480236, "grad_norm": 1.4645363092422485, "learning_rate": 5.717339533837125e-06, "loss": 0.8864, "step": 24490 }, { "epoch": 0.44562469306462466, "grad_norm": 1.7392913103103638, "learning_rate": 5.715464381480996e-06, "loss": 0.9243, "step": 24500 }, { "epoch": 0.44580658069444695, "grad_norm": 1.7126799821853638, "learning_rate": 5.7135892291248675e-06, "loss": 0.9566, "step": 24510 }, { "epoch": 0.44598846832426925, "grad_norm": 1.7097699642181396, "learning_rate": 5.711714076768739e-06, "loss": 0.9363, "step": 24520 }, { "epoch": 0.44617035595409155, "grad_norm": 1.7786710262298584, "learning_rate": 5.7098389244126085e-06, "loss": 0.9238, "step": 24530 }, { "epoch": 0.44635224358391384, "grad_norm": 1.8594014644622803, "learning_rate": 5.70796377205648e-06, "loss": 0.9323, "step": 24540 }, { "epoch": 0.44653413121373614, "grad_norm": 1.5484310388565063, "learning_rate": 5.70608861970035e-06, "loss": 0.9355, "step": 24550 }, { "epoch": 0.44671601884355844, "grad_norm": 1.7573254108428955, "learning_rate": 5.704213467344222e-06, "loss": 0.9238, "step": 24560 }, { "epoch": 0.44689790647338073, "grad_norm": 1.8144683837890625, "learning_rate": 5.702338314988093e-06, "loss": 0.982, "step": 24570 }, { "epoch": 0.44707979410320303, "grad_norm": 1.8707637786865234, "learning_rate": 5.700463162631964e-06, "loss": 0.9573, "step": 24580 }, { "epoch": 0.44726168173302533, "grad_norm": 1.9777191877365112, "learning_rate": 5.698588010275835e-06, "loss": 0.9921, "step": 24590 }, { "epoch": 0.4474435693628476, "grad_norm": 1.9554080963134766, "learning_rate": 5.696712857919706e-06, "loss": 0.9282, "step": 24600 }, { "epoch": 0.4476254569926699, "grad_norm": 1.8734979629516602, "learning_rate": 5.6948377055635776e-06, "loss": 0.9934, "step": 24610 }, { "epoch": 0.4478073446224922, "grad_norm": 1.8916754722595215, "learning_rate": 5.692962553207448e-06, "loss": 0.9648, "step": 24620 }, { "epoch": 0.4479892322523145, "grad_norm": 1.7224067449569702, "learning_rate": 5.691087400851319e-06, "loss": 0.9533, "step": 24630 }, { "epoch": 0.4481711198821368, "grad_norm": 1.2640917301177979, "learning_rate": 5.689212248495191e-06, "loss": 0.9448, "step": 24640 }, { "epoch": 0.4483530075119591, "grad_norm": 1.6344876289367676, "learning_rate": 5.687337096139062e-06, "loss": 0.9174, "step": 24650 }, { "epoch": 0.4485348951417814, "grad_norm": 1.8791377544403076, "learning_rate": 5.685461943782933e-06, "loss": 0.925, "step": 24660 }, { "epoch": 0.4487167827716037, "grad_norm": 1.682818055152893, "learning_rate": 5.683586791426804e-06, "loss": 0.8917, "step": 24670 }, { "epoch": 0.448898670401426, "grad_norm": 1.6948832273483276, "learning_rate": 5.681711639070675e-06, "loss": 0.949, "step": 24680 }, { "epoch": 0.4490805580312483, "grad_norm": 1.8898659944534302, "learning_rate": 5.679836486714546e-06, "loss": 0.9833, "step": 24690 }, { "epoch": 0.4492624456610706, "grad_norm": 1.7765005826950073, "learning_rate": 5.677961334358417e-06, "loss": 0.9494, "step": 24700 }, { "epoch": 0.4494443332908929, "grad_norm": 1.9873895645141602, "learning_rate": 5.6760861820022885e-06, "loss": 0.8998, "step": 24710 }, { "epoch": 0.4496262209207152, "grad_norm": 1.59677255153656, "learning_rate": 5.67421102964616e-06, "loss": 0.9034, "step": 24720 }, { "epoch": 0.4498081085505375, "grad_norm": 1.5882203578948975, "learning_rate": 5.67233587729003e-06, "loss": 0.9543, "step": 24730 }, { "epoch": 0.4499899961803598, "grad_norm": 1.7846215963363647, "learning_rate": 5.670460724933902e-06, "loss": 0.9304, "step": 24740 }, { "epoch": 0.4501718838101821, "grad_norm": 1.630859613418579, "learning_rate": 5.668585572577773e-06, "loss": 0.9, "step": 24750 }, { "epoch": 0.4503537714400044, "grad_norm": 1.8120801448822021, "learning_rate": 5.6667104202216435e-06, "loss": 0.8987, "step": 24760 }, { "epoch": 0.4505356590698267, "grad_norm": 1.7587453126907349, "learning_rate": 5.664835267865515e-06, "loss": 0.9429, "step": 24770 }, { "epoch": 0.450717546699649, "grad_norm": 1.9521197080612183, "learning_rate": 5.662960115509386e-06, "loss": 0.9733, "step": 24780 }, { "epoch": 0.45089943432947127, "grad_norm": 1.718642234802246, "learning_rate": 5.6610849631532576e-06, "loss": 0.9261, "step": 24790 }, { "epoch": 0.45108132195929357, "grad_norm": 1.7561429738998413, "learning_rate": 5.659209810797128e-06, "loss": 0.9364, "step": 24800 }, { "epoch": 0.45126320958911587, "grad_norm": 1.87411367893219, "learning_rate": 5.6573346584409986e-06, "loss": 0.9641, "step": 24810 }, { "epoch": 0.45144509721893816, "grad_norm": 1.7848961353302002, "learning_rate": 5.655459506084869e-06, "loss": 0.9438, "step": 24820 }, { "epoch": 0.45162698484876046, "grad_norm": 2.5736875534057617, "learning_rate": 5.65358435372874e-06, "loss": 0.9243, "step": 24830 }, { "epoch": 0.45180887247858276, "grad_norm": 1.8918070793151855, "learning_rate": 5.651709201372612e-06, "loss": 0.9208, "step": 24840 }, { "epoch": 0.45199076010840505, "grad_norm": 1.6735316514968872, "learning_rate": 5.649834049016483e-06, "loss": 0.9571, "step": 24850 }, { "epoch": 0.4521726477382273, "grad_norm": 1.6286576986312866, "learning_rate": 5.647958896660354e-06, "loss": 0.9579, "step": 24860 }, { "epoch": 0.4523545353680496, "grad_norm": 1.8146171569824219, "learning_rate": 5.646083744304225e-06, "loss": 0.9382, "step": 24870 }, { "epoch": 0.4525364229978719, "grad_norm": 1.663692831993103, "learning_rate": 5.644208591948096e-06, "loss": 0.97, "step": 24880 }, { "epoch": 0.4527183106276942, "grad_norm": 1.5580780506134033, "learning_rate": 5.642333439591967e-06, "loss": 0.9141, "step": 24890 }, { "epoch": 0.4529001982575165, "grad_norm": 1.7900398969650269, "learning_rate": 5.640458287235838e-06, "loss": 0.9221, "step": 24900 }, { "epoch": 0.4530820858873388, "grad_norm": 1.5524049997329712, "learning_rate": 5.6385831348797095e-06, "loss": 0.9731, "step": 24910 }, { "epoch": 0.4532639735171611, "grad_norm": 1.9595086574554443, "learning_rate": 5.636707982523581e-06, "loss": 0.9729, "step": 24920 }, { "epoch": 0.4534458611469834, "grad_norm": 1.6639009714126587, "learning_rate": 5.634832830167451e-06, "loss": 0.902, "step": 24930 }, { "epoch": 0.4536277487768057, "grad_norm": 1.6980737447738647, "learning_rate": 5.632957677811323e-06, "loss": 0.9284, "step": 24940 }, { "epoch": 0.45380963640662797, "grad_norm": 1.807883620262146, "learning_rate": 5.631082525455194e-06, "loss": 0.9738, "step": 24950 }, { "epoch": 0.45399152403645027, "grad_norm": 1.869215488433838, "learning_rate": 5.6292073730990645e-06, "loss": 0.9053, "step": 24960 }, { "epoch": 0.45417341166627256, "grad_norm": 3.0838093757629395, "learning_rate": 5.627332220742936e-06, "loss": 0.9015, "step": 24970 }, { "epoch": 0.45435529929609486, "grad_norm": 1.6845874786376953, "learning_rate": 5.625457068386807e-06, "loss": 0.9322, "step": 24980 }, { "epoch": 0.45453718692591716, "grad_norm": 1.8890109062194824, "learning_rate": 5.6235819160306786e-06, "loss": 0.9961, "step": 24990 }, { "epoch": 0.45471907455573946, "grad_norm": 1.8497660160064697, "learning_rate": 5.621706763674549e-06, "loss": 0.946, "step": 25000 }, { "epoch": 0.45490096218556175, "grad_norm": 1.826682448387146, "learning_rate": 5.61983161131842e-06, "loss": 0.967, "step": 25010 }, { "epoch": 0.45508284981538405, "grad_norm": 2.0935964584350586, "learning_rate": 5.617956458962292e-06, "loss": 0.8951, "step": 25020 }, { "epoch": 0.45526473744520635, "grad_norm": 1.5101534128189087, "learning_rate": 5.616081306606162e-06, "loss": 0.9103, "step": 25030 }, { "epoch": 0.45544662507502864, "grad_norm": 1.4822194576263428, "learning_rate": 5.614206154250034e-06, "loss": 0.8677, "step": 25040 }, { "epoch": 0.45562851270485094, "grad_norm": 1.7467336654663086, "learning_rate": 5.612331001893905e-06, "loss": 0.9345, "step": 25050 }, { "epoch": 0.45581040033467324, "grad_norm": 2.061743974685669, "learning_rate": 5.610455849537776e-06, "loss": 0.9114, "step": 25060 }, { "epoch": 0.45599228796449554, "grad_norm": 1.4925720691680908, "learning_rate": 5.608580697181647e-06, "loss": 0.9563, "step": 25070 }, { "epoch": 0.45617417559431783, "grad_norm": 1.8822121620178223, "learning_rate": 5.606705544825517e-06, "loss": 0.9228, "step": 25080 }, { "epoch": 0.45635606322414013, "grad_norm": 1.8797225952148438, "learning_rate": 5.604830392469388e-06, "loss": 0.9518, "step": 25090 }, { "epoch": 0.4565379508539624, "grad_norm": 1.7793980836868286, "learning_rate": 5.602955240113259e-06, "loss": 0.9318, "step": 25100 }, { "epoch": 0.4567198384837847, "grad_norm": 1.8565032482147217, "learning_rate": 5.6010800877571305e-06, "loss": 0.9525, "step": 25110 }, { "epoch": 0.456901726113607, "grad_norm": 1.6069177389144897, "learning_rate": 5.599204935401002e-06, "loss": 0.8834, "step": 25120 }, { "epoch": 0.4570836137434293, "grad_norm": 1.7974411249160767, "learning_rate": 5.597329783044872e-06, "loss": 0.9049, "step": 25130 }, { "epoch": 0.4572655013732516, "grad_norm": 1.512403130531311, "learning_rate": 5.595454630688744e-06, "loss": 0.8933, "step": 25140 }, { "epoch": 0.4574473890030739, "grad_norm": 1.78144109249115, "learning_rate": 5.593579478332615e-06, "loss": 0.9037, "step": 25150 }, { "epoch": 0.4576292766328962, "grad_norm": 2.052403211593628, "learning_rate": 5.5917043259764855e-06, "loss": 0.9363, "step": 25160 }, { "epoch": 0.4578111642627185, "grad_norm": 1.7542611360549927, "learning_rate": 5.589829173620357e-06, "loss": 0.9336, "step": 25170 }, { "epoch": 0.4579930518925408, "grad_norm": 2.008397102355957, "learning_rate": 5.587954021264228e-06, "loss": 0.9687, "step": 25180 }, { "epoch": 0.4581749395223631, "grad_norm": 1.7846548557281494, "learning_rate": 5.5860788689080995e-06, "loss": 0.968, "step": 25190 }, { "epoch": 0.4583568271521854, "grad_norm": 1.8089210987091064, "learning_rate": 5.58420371655197e-06, "loss": 0.9621, "step": 25200 }, { "epoch": 0.4585387147820077, "grad_norm": 1.9440875053405762, "learning_rate": 5.582328564195841e-06, "loss": 0.9216, "step": 25210 }, { "epoch": 0.45872060241183, "grad_norm": 2.089184284210205, "learning_rate": 5.580453411839713e-06, "loss": 0.9442, "step": 25220 }, { "epoch": 0.4589024900416523, "grad_norm": 1.804408073425293, "learning_rate": 5.578578259483583e-06, "loss": 0.9012, "step": 25230 }, { "epoch": 0.4590843776714746, "grad_norm": 1.587821125984192, "learning_rate": 5.576703107127455e-06, "loss": 0.9034, "step": 25240 }, { "epoch": 0.4592662653012969, "grad_norm": 1.6983145475387573, "learning_rate": 5.574827954771326e-06, "loss": 0.9522, "step": 25250 }, { "epoch": 0.4594481529311192, "grad_norm": 1.5479928255081177, "learning_rate": 5.572952802415197e-06, "loss": 0.9158, "step": 25260 }, { "epoch": 0.4596300405609415, "grad_norm": 1.2768175601959229, "learning_rate": 5.571077650059068e-06, "loss": 0.9188, "step": 25270 }, { "epoch": 0.4598119281907637, "grad_norm": 1.8626618385314941, "learning_rate": 5.569202497702939e-06, "loss": 0.9037, "step": 25280 }, { "epoch": 0.459993815820586, "grad_norm": 1.7442610263824463, "learning_rate": 5.5673273453468105e-06, "loss": 0.909, "step": 25290 }, { "epoch": 0.4601757034504083, "grad_norm": 1.350962519645691, "learning_rate": 5.565452192990681e-06, "loss": 0.8698, "step": 25300 }, { "epoch": 0.4603575910802306, "grad_norm": 1.6970657110214233, "learning_rate": 5.563577040634552e-06, "loss": 0.9463, "step": 25310 }, { "epoch": 0.4605394787100529, "grad_norm": 1.8357535600662231, "learning_rate": 5.561701888278424e-06, "loss": 0.9761, "step": 25320 }, { "epoch": 0.4607213663398752, "grad_norm": 2.0518150329589844, "learning_rate": 5.559826735922295e-06, "loss": 0.9441, "step": 25330 }, { "epoch": 0.4609032539696975, "grad_norm": 2.3117856979370117, "learning_rate": 5.5579515835661655e-06, "loss": 0.9605, "step": 25340 }, { "epoch": 0.4610851415995198, "grad_norm": 1.6193277835845947, "learning_rate": 5.556076431210037e-06, "loss": 0.982, "step": 25350 }, { "epoch": 0.4612670292293421, "grad_norm": 1.7919234037399292, "learning_rate": 5.5542012788539065e-06, "loss": 0.9001, "step": 25360 }, { "epoch": 0.4614489168591644, "grad_norm": 1.7843897342681885, "learning_rate": 5.552326126497778e-06, "loss": 0.953, "step": 25370 }, { "epoch": 0.4616308044889867, "grad_norm": 1.5884428024291992, "learning_rate": 5.550450974141649e-06, "loss": 0.9152, "step": 25380 }, { "epoch": 0.461812692118809, "grad_norm": 1.7955889701843262, "learning_rate": 5.5485758217855205e-06, "loss": 0.9358, "step": 25390 }, { "epoch": 0.4619945797486313, "grad_norm": 1.9264470338821411, "learning_rate": 5.546700669429391e-06, "loss": 0.9183, "step": 25400 }, { "epoch": 0.4621764673784536, "grad_norm": 1.656365156173706, "learning_rate": 5.544825517073262e-06, "loss": 0.9668, "step": 25410 }, { "epoch": 0.4623583550082759, "grad_norm": 1.7884774208068848, "learning_rate": 5.542950364717134e-06, "loss": 0.9638, "step": 25420 }, { "epoch": 0.4625402426380982, "grad_norm": 1.7192962169647217, "learning_rate": 5.541075212361004e-06, "loss": 0.9647, "step": 25430 }, { "epoch": 0.4627221302679205, "grad_norm": 2.031825065612793, "learning_rate": 5.539200060004876e-06, "loss": 0.9502, "step": 25440 }, { "epoch": 0.46290401789774277, "grad_norm": 1.677132248878479, "learning_rate": 5.537324907648747e-06, "loss": 0.9656, "step": 25450 }, { "epoch": 0.46308590552756507, "grad_norm": 2.0672240257263184, "learning_rate": 5.535449755292618e-06, "loss": 0.9249, "step": 25460 }, { "epoch": 0.46326779315738736, "grad_norm": 1.8657370805740356, "learning_rate": 5.533574602936489e-06, "loss": 0.9713, "step": 25470 }, { "epoch": 0.46344968078720966, "grad_norm": 1.72834050655365, "learning_rate": 5.53169945058036e-06, "loss": 0.9498, "step": 25480 }, { "epoch": 0.46363156841703196, "grad_norm": 1.956283688545227, "learning_rate": 5.5298242982242315e-06, "loss": 0.943, "step": 25490 }, { "epoch": 0.46381345604685426, "grad_norm": 1.7781356573104858, "learning_rate": 5.527949145868102e-06, "loss": 0.9472, "step": 25500 }, { "epoch": 0.46399534367667655, "grad_norm": 1.7738869190216064, "learning_rate": 5.526073993511973e-06, "loss": 0.9359, "step": 25510 }, { "epoch": 0.46417723130649885, "grad_norm": 1.8052818775177002, "learning_rate": 5.524198841155845e-06, "loss": 0.9468, "step": 25520 }, { "epoch": 0.46435911893632115, "grad_norm": 1.6228158473968506, "learning_rate": 5.522323688799716e-06, "loss": 0.9589, "step": 25530 }, { "epoch": 0.46454100656614344, "grad_norm": 1.5916061401367188, "learning_rate": 5.5204485364435865e-06, "loss": 0.9243, "step": 25540 }, { "epoch": 0.46472289419596574, "grad_norm": 1.7903155088424683, "learning_rate": 5.518573384087458e-06, "loss": 0.9213, "step": 25550 }, { "epoch": 0.46490478182578804, "grad_norm": 1.6573275327682495, "learning_rate": 5.516698231731329e-06, "loss": 0.9261, "step": 25560 }, { "epoch": 0.46508666945561034, "grad_norm": 1.9876381158828735, "learning_rate": 5.5148230793752e-06, "loss": 0.9893, "step": 25570 }, { "epoch": 0.46526855708543263, "grad_norm": 4.872640609741211, "learning_rate": 5.512947927019071e-06, "loss": 0.8895, "step": 25580 }, { "epoch": 0.46545044471525493, "grad_norm": 1.5064125061035156, "learning_rate": 5.511072774662942e-06, "loss": 0.8874, "step": 25590 }, { "epoch": 0.4656323323450772, "grad_norm": 1.8287690877914429, "learning_rate": 5.509197622306814e-06, "loss": 0.9556, "step": 25600 }, { "epoch": 0.4658142199748995, "grad_norm": 1.6951038837432861, "learning_rate": 5.507322469950684e-06, "loss": 0.9535, "step": 25610 }, { "epoch": 0.4659961076047218, "grad_norm": 1.646454095840454, "learning_rate": 5.5054473175945556e-06, "loss": 0.9306, "step": 25620 }, { "epoch": 0.4661779952345441, "grad_norm": 1.7710955142974854, "learning_rate": 5.503572165238425e-06, "loss": 0.9711, "step": 25630 }, { "epoch": 0.4663598828643664, "grad_norm": 1.702586054801941, "learning_rate": 5.5016970128822966e-06, "loss": 0.8898, "step": 25640 }, { "epoch": 0.4665417704941887, "grad_norm": 1.9196209907531738, "learning_rate": 5.499821860526168e-06, "loss": 0.9663, "step": 25650 }, { "epoch": 0.466723658124011, "grad_norm": 1.8831323385238647, "learning_rate": 5.497946708170039e-06, "loss": 0.9291, "step": 25660 }, { "epoch": 0.4669055457538333, "grad_norm": 1.8339132070541382, "learning_rate": 5.49607155581391e-06, "loss": 0.9298, "step": 25670 }, { "epoch": 0.4670874333836556, "grad_norm": 1.9934489727020264, "learning_rate": 5.494196403457781e-06, "loss": 0.9197, "step": 25680 }, { "epoch": 0.4672693210134779, "grad_norm": 2.016810655593872, "learning_rate": 5.4923212511016525e-06, "loss": 0.9332, "step": 25690 }, { "epoch": 0.46745120864330014, "grad_norm": 1.74915611743927, "learning_rate": 5.490446098745523e-06, "loss": 0.9013, "step": 25700 }, { "epoch": 0.46763309627312244, "grad_norm": 1.7488962411880493, "learning_rate": 5.488570946389394e-06, "loss": 0.934, "step": 25710 }, { "epoch": 0.46781498390294474, "grad_norm": 1.7157257795333862, "learning_rate": 5.486695794033266e-06, "loss": 0.9073, "step": 25720 }, { "epoch": 0.46799687153276703, "grad_norm": 1.9035598039627075, "learning_rate": 5.484820641677137e-06, "loss": 0.9525, "step": 25730 }, { "epoch": 0.46817875916258933, "grad_norm": 1.7672237157821655, "learning_rate": 5.4829454893210075e-06, "loss": 0.9504, "step": 25740 }, { "epoch": 0.46836064679241163, "grad_norm": 1.5747566223144531, "learning_rate": 5.481070336964879e-06, "loss": 0.9278, "step": 25750 }, { "epoch": 0.4685425344222339, "grad_norm": 1.8140782117843628, "learning_rate": 5.47919518460875e-06, "loss": 0.9562, "step": 25760 }, { "epoch": 0.4687244220520562, "grad_norm": 1.4695918560028076, "learning_rate": 5.477320032252621e-06, "loss": 0.8502, "step": 25770 }, { "epoch": 0.4689063096818785, "grad_norm": 1.6936376094818115, "learning_rate": 5.475444879896492e-06, "loss": 0.9023, "step": 25780 }, { "epoch": 0.4690881973117008, "grad_norm": 1.7057700157165527, "learning_rate": 5.473569727540363e-06, "loss": 0.9331, "step": 25790 }, { "epoch": 0.4692700849415231, "grad_norm": 1.7180604934692383, "learning_rate": 5.471694575184235e-06, "loss": 0.9036, "step": 25800 }, { "epoch": 0.4694519725713454, "grad_norm": 1.8229283094406128, "learning_rate": 5.469819422828105e-06, "loss": 0.9456, "step": 25810 }, { "epoch": 0.4696338602011677, "grad_norm": 1.9319720268249512, "learning_rate": 5.4679442704719766e-06, "loss": 0.9625, "step": 25820 }, { "epoch": 0.46981574783099, "grad_norm": 1.7636637687683105, "learning_rate": 5.466069118115848e-06, "loss": 0.9897, "step": 25830 }, { "epoch": 0.4699976354608123, "grad_norm": 1.8712573051452637, "learning_rate": 5.464193965759718e-06, "loss": 0.923, "step": 25840 }, { "epoch": 0.4701795230906346, "grad_norm": 1.8372734785079956, "learning_rate": 5.46231881340359e-06, "loss": 0.9414, "step": 25850 }, { "epoch": 0.4703614107204569, "grad_norm": 1.8823912143707275, "learning_rate": 5.460443661047461e-06, "loss": 0.9079, "step": 25860 }, { "epoch": 0.4705432983502792, "grad_norm": 1.7433570623397827, "learning_rate": 5.4585685086913324e-06, "loss": 0.9545, "step": 25870 }, { "epoch": 0.4707251859801015, "grad_norm": 1.7278735637664795, "learning_rate": 5.456693356335203e-06, "loss": 0.9686, "step": 25880 }, { "epoch": 0.4709070736099238, "grad_norm": 1.881070613861084, "learning_rate": 5.454818203979074e-06, "loss": 0.9303, "step": 25890 }, { "epoch": 0.4710889612397461, "grad_norm": 1.6203715801239014, "learning_rate": 5.452943051622946e-06, "loss": 0.9137, "step": 25900 }, { "epoch": 0.4712708488695684, "grad_norm": 2.0224497318267822, "learning_rate": 5.451067899266815e-06, "loss": 1.0028, "step": 25910 }, { "epoch": 0.4714527364993907, "grad_norm": 1.7020373344421387, "learning_rate": 5.449192746910687e-06, "loss": 0.9336, "step": 25920 }, { "epoch": 0.471634624129213, "grad_norm": 1.7350817918777466, "learning_rate": 5.447317594554557e-06, "loss": 0.9466, "step": 25930 }, { "epoch": 0.4718165117590353, "grad_norm": 1.9133930206298828, "learning_rate": 5.4454424421984285e-06, "loss": 0.9481, "step": 25940 }, { "epoch": 0.47199839938885757, "grad_norm": 2.032752275466919, "learning_rate": 5.4435672898423e-06, "loss": 0.9418, "step": 25950 }, { "epoch": 0.47218028701867987, "grad_norm": 1.6111680269241333, "learning_rate": 5.441692137486171e-06, "loss": 0.9672, "step": 25960 }, { "epoch": 0.47236217464850216, "grad_norm": 1.8706839084625244, "learning_rate": 5.439816985130042e-06, "loss": 0.9241, "step": 25970 }, { "epoch": 0.47254406227832446, "grad_norm": 1.743367075920105, "learning_rate": 5.437941832773913e-06, "loss": 0.9215, "step": 25980 }, { "epoch": 0.47272594990814676, "grad_norm": 1.571146845817566, "learning_rate": 5.436066680417784e-06, "loss": 0.953, "step": 25990 }, { "epoch": 0.47290783753796906, "grad_norm": 1.8292210102081299, "learning_rate": 5.434191528061656e-06, "loss": 0.9502, "step": 26000 }, { "epoch": 0.47308972516779135, "grad_norm": 2.0385780334472656, "learning_rate": 5.432316375705526e-06, "loss": 0.9984, "step": 26010 }, { "epoch": 0.47327161279761365, "grad_norm": 1.8323901891708374, "learning_rate": 5.4304412233493976e-06, "loss": 0.9975, "step": 26020 }, { "epoch": 0.47345350042743595, "grad_norm": 27.173025131225586, "learning_rate": 5.428566070993269e-06, "loss": 0.9438, "step": 26030 }, { "epoch": 0.47363538805725824, "grad_norm": 1.937531590461731, "learning_rate": 5.426690918637139e-06, "loss": 0.9584, "step": 26040 }, { "epoch": 0.47381727568708054, "grad_norm": 1.9865853786468506, "learning_rate": 5.424815766281011e-06, "loss": 0.9465, "step": 26050 }, { "epoch": 0.47399916331690284, "grad_norm": 1.9251933097839355, "learning_rate": 5.422940613924882e-06, "loss": 0.9324, "step": 26060 }, { "epoch": 0.47418105094672514, "grad_norm": 1.527958869934082, "learning_rate": 5.4210654615687534e-06, "loss": 0.9245, "step": 26070 }, { "epoch": 0.47436293857654743, "grad_norm": 1.6725372076034546, "learning_rate": 5.419190309212624e-06, "loss": 0.9273, "step": 26080 }, { "epoch": 0.47454482620636973, "grad_norm": 1.7708693742752075, "learning_rate": 5.417315156856495e-06, "loss": 0.9818, "step": 26090 }, { "epoch": 0.474726713836192, "grad_norm": 1.5165467262268066, "learning_rate": 5.415440004500367e-06, "loss": 0.9144, "step": 26100 }, { "epoch": 0.4749086014660143, "grad_norm": 1.8606237173080444, "learning_rate": 5.413564852144237e-06, "loss": 0.9431, "step": 26110 }, { "epoch": 0.47509048909583657, "grad_norm": 1.573801040649414, "learning_rate": 5.4116896997881085e-06, "loss": 0.9137, "step": 26120 }, { "epoch": 0.47527237672565886, "grad_norm": 1.8134078979492188, "learning_rate": 5.40981454743198e-06, "loss": 0.9399, "step": 26130 }, { "epoch": 0.47545426435548116, "grad_norm": 1.905177116394043, "learning_rate": 5.407939395075851e-06, "loss": 0.9368, "step": 26140 }, { "epoch": 0.47563615198530346, "grad_norm": 1.6472920179367065, "learning_rate": 5.406064242719722e-06, "loss": 0.9123, "step": 26150 }, { "epoch": 0.47581803961512575, "grad_norm": 1.7117635011672974, "learning_rate": 5.404189090363593e-06, "loss": 0.9115, "step": 26160 }, { "epoch": 0.47599992724494805, "grad_norm": 1.7651386260986328, "learning_rate": 5.402313938007464e-06, "loss": 0.9404, "step": 26170 }, { "epoch": 0.47618181487477035, "grad_norm": 1.57883620262146, "learning_rate": 5.400438785651334e-06, "loss": 0.9515, "step": 26180 }, { "epoch": 0.47636370250459265, "grad_norm": 1.9204127788543701, "learning_rate": 5.398563633295205e-06, "loss": 0.9295, "step": 26190 }, { "epoch": 0.47654559013441494, "grad_norm": 1.7813009023666382, "learning_rate": 5.396688480939076e-06, "loss": 0.9507, "step": 26200 }, { "epoch": 0.47672747776423724, "grad_norm": 1.7327282428741455, "learning_rate": 5.394813328582947e-06, "loss": 0.9462, "step": 26210 }, { "epoch": 0.47690936539405954, "grad_norm": 1.990752935409546, "learning_rate": 5.3929381762268185e-06, "loss": 0.9911, "step": 26220 }, { "epoch": 0.47709125302388183, "grad_norm": 1.728952407836914, "learning_rate": 5.39106302387069e-06, "loss": 0.9545, "step": 26230 }, { "epoch": 0.47727314065370413, "grad_norm": 2.0776143074035645, "learning_rate": 5.38918787151456e-06, "loss": 1.0009, "step": 26240 }, { "epoch": 0.47745502828352643, "grad_norm": 1.7980643510818481, "learning_rate": 5.387312719158432e-06, "loss": 0.9613, "step": 26250 }, { "epoch": 0.4776369159133487, "grad_norm": 1.3760435581207275, "learning_rate": 5.385437566802303e-06, "loss": 0.8707, "step": 26260 }, { "epoch": 0.477818803543171, "grad_norm": 1.9111219644546509, "learning_rate": 5.383562414446174e-06, "loss": 0.9791, "step": 26270 }, { "epoch": 0.4780006911729933, "grad_norm": 1.930521845817566, "learning_rate": 5.381687262090045e-06, "loss": 0.9798, "step": 26280 }, { "epoch": 0.4781825788028156, "grad_norm": 1.6972194910049438, "learning_rate": 5.379812109733916e-06, "loss": 0.9352, "step": 26290 }, { "epoch": 0.4783644664326379, "grad_norm": 1.4849518537521362, "learning_rate": 5.377936957377788e-06, "loss": 0.9142, "step": 26300 }, { "epoch": 0.4785463540624602, "grad_norm": 1.6689987182617188, "learning_rate": 5.376061805021658e-06, "loss": 0.8718, "step": 26310 }, { "epoch": 0.4787282416922825, "grad_norm": 1.8172708749771118, "learning_rate": 5.3741866526655295e-06, "loss": 0.9018, "step": 26320 }, { "epoch": 0.4789101293221048, "grad_norm": 1.7628045082092285, "learning_rate": 5.372311500309401e-06, "loss": 0.9216, "step": 26330 }, { "epoch": 0.4790920169519271, "grad_norm": 1.835808277130127, "learning_rate": 5.370436347953272e-06, "loss": 0.8854, "step": 26340 }, { "epoch": 0.4792739045817494, "grad_norm": 1.767014503479004, "learning_rate": 5.368561195597143e-06, "loss": 0.891, "step": 26350 }, { "epoch": 0.4794557922115717, "grad_norm": 1.8796334266662598, "learning_rate": 5.366686043241014e-06, "loss": 0.9415, "step": 26360 }, { "epoch": 0.479637679841394, "grad_norm": 1.7756295204162598, "learning_rate": 5.364810890884885e-06, "loss": 0.9018, "step": 26370 }, { "epoch": 0.4798195674712163, "grad_norm": 1.8216146230697632, "learning_rate": 5.362935738528756e-06, "loss": 0.9463, "step": 26380 }, { "epoch": 0.4800014551010386, "grad_norm": 2.0933403968811035, "learning_rate": 5.361060586172627e-06, "loss": 0.8995, "step": 26390 }, { "epoch": 0.4801833427308609, "grad_norm": 1.5769665241241455, "learning_rate": 5.3591854338164985e-06, "loss": 0.9017, "step": 26400 }, { "epoch": 0.4803652303606832, "grad_norm": 1.7467364072799683, "learning_rate": 5.35731028146037e-06, "loss": 0.9519, "step": 26410 }, { "epoch": 0.4805471179905055, "grad_norm": 1.481995940208435, "learning_rate": 5.35543512910424e-06, "loss": 0.9, "step": 26420 }, { "epoch": 0.4807290056203278, "grad_norm": 1.4602344036102295, "learning_rate": 5.353559976748112e-06, "loss": 0.9218, "step": 26430 }, { "epoch": 0.4809108932501501, "grad_norm": 1.8295810222625732, "learning_rate": 5.351684824391983e-06, "loss": 0.9213, "step": 26440 }, { "epoch": 0.48109278087997237, "grad_norm": 1.892196536064148, "learning_rate": 5.3498096720358536e-06, "loss": 0.9426, "step": 26450 }, { "epoch": 0.48127466850979467, "grad_norm": 1.5648273229599, "learning_rate": 5.347934519679724e-06, "loss": 0.9236, "step": 26460 }, { "epoch": 0.48145655613961696, "grad_norm": 1.8066247701644897, "learning_rate": 5.3460593673235946e-06, "loss": 0.9639, "step": 26470 }, { "epoch": 0.48163844376943926, "grad_norm": 2.039532423019409, "learning_rate": 5.344184214967466e-06, "loss": 0.9652, "step": 26480 }, { "epoch": 0.48182033139926156, "grad_norm": 1.9863473176956177, "learning_rate": 5.342309062611337e-06, "loss": 0.905, "step": 26490 }, { "epoch": 0.48200221902908386, "grad_norm": 1.823171854019165, "learning_rate": 5.340433910255209e-06, "loss": 0.9206, "step": 26500 }, { "epoch": 0.48218410665890615, "grad_norm": 1.9079914093017578, "learning_rate": 5.338558757899079e-06, "loss": 0.9767, "step": 26510 }, { "epoch": 0.48236599428872845, "grad_norm": 1.96554434299469, "learning_rate": 5.3366836055429505e-06, "loss": 0.8972, "step": 26520 }, { "epoch": 0.4825478819185507, "grad_norm": 1.6178168058395386, "learning_rate": 5.334808453186822e-06, "loss": 0.8748, "step": 26530 }, { "epoch": 0.482729769548373, "grad_norm": 1.8007620573043823, "learning_rate": 5.332933300830692e-06, "loss": 0.9107, "step": 26540 }, { "epoch": 0.4829116571781953, "grad_norm": 2.015913724899292, "learning_rate": 5.331058148474564e-06, "loss": 1.0105, "step": 26550 }, { "epoch": 0.4830935448080176, "grad_norm": 1.6925578117370605, "learning_rate": 5.329182996118435e-06, "loss": 0.9408, "step": 26560 }, { "epoch": 0.4832754324378399, "grad_norm": 1.8472177982330322, "learning_rate": 5.327307843762306e-06, "loss": 0.9327, "step": 26570 }, { "epoch": 0.4834573200676622, "grad_norm": 2.0298852920532227, "learning_rate": 5.325432691406177e-06, "loss": 0.9333, "step": 26580 }, { "epoch": 0.4836392076974845, "grad_norm": 1.6047093868255615, "learning_rate": 5.323557539050048e-06, "loss": 0.8972, "step": 26590 }, { "epoch": 0.48382109532730677, "grad_norm": 1.6971629858016968, "learning_rate": 5.3216823866939195e-06, "loss": 0.9127, "step": 26600 }, { "epoch": 0.48400298295712907, "grad_norm": 1.9122039079666138, "learning_rate": 5.31980723433779e-06, "loss": 0.9714, "step": 26610 }, { "epoch": 0.48418487058695137, "grad_norm": 1.3488558530807495, "learning_rate": 5.317932081981661e-06, "loss": 0.9239, "step": 26620 }, { "epoch": 0.48436675821677366, "grad_norm": 1.8838000297546387, "learning_rate": 5.316056929625533e-06, "loss": 0.9707, "step": 26630 }, { "epoch": 0.48454864584659596, "grad_norm": 1.7706851959228516, "learning_rate": 5.314181777269404e-06, "loss": 0.9282, "step": 26640 }, { "epoch": 0.48473053347641826, "grad_norm": 1.7789227962493896, "learning_rate": 5.3123066249132746e-06, "loss": 0.9621, "step": 26650 }, { "epoch": 0.48491242110624055, "grad_norm": 1.7115477323532104, "learning_rate": 5.310431472557146e-06, "loss": 0.917, "step": 26660 }, { "epoch": 0.48509430873606285, "grad_norm": 1.5613596439361572, "learning_rate": 5.308556320201017e-06, "loss": 0.905, "step": 26670 }, { "epoch": 0.48527619636588515, "grad_norm": 1.7320533990859985, "learning_rate": 5.306681167844889e-06, "loss": 0.9288, "step": 26680 }, { "epoch": 0.48545808399570745, "grad_norm": 1.777734398841858, "learning_rate": 5.304806015488759e-06, "loss": 0.909, "step": 26690 }, { "epoch": 0.48563997162552974, "grad_norm": 1.6741708517074585, "learning_rate": 5.3029308631326304e-06, "loss": 0.9501, "step": 26700 }, { "epoch": 0.48582185925535204, "grad_norm": 2.0318093299865723, "learning_rate": 5.301055710776502e-06, "loss": 0.975, "step": 26710 }, { "epoch": 0.48600374688517434, "grad_norm": 1.8086800575256348, "learning_rate": 5.299180558420372e-06, "loss": 0.9591, "step": 26720 }, { "epoch": 0.48618563451499663, "grad_norm": 1.8341994285583496, "learning_rate": 5.297305406064244e-06, "loss": 0.9025, "step": 26730 }, { "epoch": 0.48636752214481893, "grad_norm": 1.4962538480758667, "learning_rate": 5.295430253708113e-06, "loss": 0.9445, "step": 26740 }, { "epoch": 0.48654940977464123, "grad_norm": 1.7776670455932617, "learning_rate": 5.293555101351985e-06, "loss": 0.9775, "step": 26750 }, { "epoch": 0.4867312974044635, "grad_norm": 2.1104695796966553, "learning_rate": 5.291679948995856e-06, "loss": 0.9393, "step": 26760 }, { "epoch": 0.4869131850342858, "grad_norm": 1.9966048002243042, "learning_rate": 5.289804796639727e-06, "loss": 0.9474, "step": 26770 }, { "epoch": 0.4870950726641081, "grad_norm": 2.0334503650665283, "learning_rate": 5.287929644283598e-06, "loss": 1.0127, "step": 26780 }, { "epoch": 0.4872769602939304, "grad_norm": 1.999276041984558, "learning_rate": 5.286054491927469e-06, "loss": 0.9537, "step": 26790 }, { "epoch": 0.4874588479237527, "grad_norm": 1.681791067123413, "learning_rate": 5.2841793395713405e-06, "loss": 0.9053, "step": 26800 }, { "epoch": 0.487640735553575, "grad_norm": 1.6309877634048462, "learning_rate": 5.282304187215211e-06, "loss": 0.9173, "step": 26810 }, { "epoch": 0.4878226231833973, "grad_norm": 1.3635234832763672, "learning_rate": 5.280429034859082e-06, "loss": 0.9002, "step": 26820 }, { "epoch": 0.4880045108132196, "grad_norm": 1.9246203899383545, "learning_rate": 5.278553882502954e-06, "loss": 0.8952, "step": 26830 }, { "epoch": 0.4881863984430419, "grad_norm": 1.638728141784668, "learning_rate": 5.276678730146825e-06, "loss": 0.9151, "step": 26840 }, { "epoch": 0.4883682860728642, "grad_norm": 1.8514955043792725, "learning_rate": 5.2748035777906956e-06, "loss": 0.9559, "step": 26850 }, { "epoch": 0.4885501737026865, "grad_norm": 1.6871751546859741, "learning_rate": 5.272928425434567e-06, "loss": 0.9223, "step": 26860 }, { "epoch": 0.4887320613325088, "grad_norm": 1.4504338502883911, "learning_rate": 5.271053273078438e-06, "loss": 0.9244, "step": 26870 }, { "epoch": 0.4889139489623311, "grad_norm": 1.8438400030136108, "learning_rate": 5.269178120722309e-06, "loss": 0.968, "step": 26880 }, { "epoch": 0.4890958365921534, "grad_norm": 1.8410753011703491, "learning_rate": 5.26730296836618e-06, "loss": 0.911, "step": 26890 }, { "epoch": 0.4892777242219757, "grad_norm": 1.870131492614746, "learning_rate": 5.2654278160100514e-06, "loss": 0.9549, "step": 26900 }, { "epoch": 0.489459611851798, "grad_norm": 1.6683014631271362, "learning_rate": 5.263552663653923e-06, "loss": 0.9328, "step": 26910 }, { "epoch": 0.4896414994816203, "grad_norm": 1.8470538854599, "learning_rate": 5.261677511297793e-06, "loss": 0.9093, "step": 26920 }, { "epoch": 0.4898233871114426, "grad_norm": 1.7992953062057495, "learning_rate": 5.259802358941665e-06, "loss": 0.949, "step": 26930 }, { "epoch": 0.4900052747412649, "grad_norm": 1.9223506450653076, "learning_rate": 5.257927206585536e-06, "loss": 0.9582, "step": 26940 }, { "epoch": 0.4901871623710871, "grad_norm": 1.7717034816741943, "learning_rate": 5.2560520542294065e-06, "loss": 0.9932, "step": 26950 }, { "epoch": 0.4903690500009094, "grad_norm": 1.6849770545959473, "learning_rate": 5.254176901873278e-06, "loss": 0.9178, "step": 26960 }, { "epoch": 0.4905509376307317, "grad_norm": 1.9492264986038208, "learning_rate": 5.252301749517149e-06, "loss": 0.9551, "step": 26970 }, { "epoch": 0.490732825260554, "grad_norm": 1.7765681743621826, "learning_rate": 5.2504265971610205e-06, "loss": 0.924, "step": 26980 }, { "epoch": 0.4909147128903763, "grad_norm": 1.5905905961990356, "learning_rate": 5.248551444804891e-06, "loss": 0.8829, "step": 26990 }, { "epoch": 0.4910966005201986, "grad_norm": 1.7968873977661133, "learning_rate": 5.246676292448762e-06, "loss": 0.9815, "step": 27000 }, { "epoch": 0.4912784881500209, "grad_norm": 1.9420381784439087, "learning_rate": 5.244801140092632e-06, "loss": 0.9592, "step": 27010 }, { "epoch": 0.4914603757798432, "grad_norm": 1.8319895267486572, "learning_rate": 5.242925987736503e-06, "loss": 0.9263, "step": 27020 }, { "epoch": 0.4916422634096655, "grad_norm": 2.2768213748931885, "learning_rate": 5.241050835380375e-06, "loss": 0.9788, "step": 27030 }, { "epoch": 0.4918241510394878, "grad_norm": 1.768019199371338, "learning_rate": 5.239175683024246e-06, "loss": 0.9447, "step": 27040 }, { "epoch": 0.4920060386693101, "grad_norm": 1.959582805633545, "learning_rate": 5.2373005306681165e-06, "loss": 0.9061, "step": 27050 }, { "epoch": 0.4921879262991324, "grad_norm": 1.936706304550171, "learning_rate": 5.235425378311988e-06, "loss": 0.9314, "step": 27060 }, { "epoch": 0.4923698139289547, "grad_norm": 2.0715396404266357, "learning_rate": 5.233550225955859e-06, "loss": 0.9348, "step": 27070 }, { "epoch": 0.492551701558777, "grad_norm": 2.01753568649292, "learning_rate": 5.23167507359973e-06, "loss": 0.9596, "step": 27080 }, { "epoch": 0.4927335891885993, "grad_norm": 1.7768604755401611, "learning_rate": 5.229799921243601e-06, "loss": 0.9023, "step": 27090 }, { "epoch": 0.49291547681842157, "grad_norm": 1.9886380434036255, "learning_rate": 5.2279247688874724e-06, "loss": 0.9456, "step": 27100 }, { "epoch": 0.49309736444824387, "grad_norm": 1.7425211668014526, "learning_rate": 5.226049616531344e-06, "loss": 0.9045, "step": 27110 }, { "epoch": 0.49327925207806617, "grad_norm": 1.731164574623108, "learning_rate": 5.224174464175214e-06, "loss": 0.8562, "step": 27120 }, { "epoch": 0.49346113970788846, "grad_norm": 1.636728048324585, "learning_rate": 5.222299311819086e-06, "loss": 0.9546, "step": 27130 }, { "epoch": 0.49364302733771076, "grad_norm": 1.8449267148971558, "learning_rate": 5.220424159462957e-06, "loss": 0.9444, "step": 27140 }, { "epoch": 0.49382491496753306, "grad_norm": 1.7578495740890503, "learning_rate": 5.2185490071068275e-06, "loss": 0.8954, "step": 27150 }, { "epoch": 0.49400680259735535, "grad_norm": 1.7692879438400269, "learning_rate": 5.216673854750699e-06, "loss": 0.9535, "step": 27160 }, { "epoch": 0.49418869022717765, "grad_norm": 1.8460845947265625, "learning_rate": 5.21479870239457e-06, "loss": 0.9392, "step": 27170 }, { "epoch": 0.49437057785699995, "grad_norm": 1.9563292264938354, "learning_rate": 5.2129235500384415e-06, "loss": 0.9473, "step": 27180 }, { "epoch": 0.49455246548682225, "grad_norm": 1.6125167608261108, "learning_rate": 5.211048397682312e-06, "loss": 0.9148, "step": 27190 }, { "epoch": 0.49473435311664454, "grad_norm": 1.7871448993682861, "learning_rate": 5.209173245326183e-06, "loss": 0.9407, "step": 27200 }, { "epoch": 0.49491624074646684, "grad_norm": 1.5834555625915527, "learning_rate": 5.207298092970055e-06, "loss": 0.9489, "step": 27210 }, { "epoch": 0.49509812837628914, "grad_norm": 1.6739935874938965, "learning_rate": 5.205422940613925e-06, "loss": 0.9264, "step": 27220 }, { "epoch": 0.49528001600611143, "grad_norm": 1.95569908618927, "learning_rate": 5.2035477882577965e-06, "loss": 0.9284, "step": 27230 }, { "epoch": 0.49546190363593373, "grad_norm": 1.9844462871551514, "learning_rate": 5.201672635901668e-06, "loss": 0.9209, "step": 27240 }, { "epoch": 0.49564379126575603, "grad_norm": 1.7542649507522583, "learning_rate": 5.199797483545539e-06, "loss": 0.9009, "step": 27250 }, { "epoch": 0.4958256788955783, "grad_norm": 1.904186487197876, "learning_rate": 5.19792233118941e-06, "loss": 0.9577, "step": 27260 }, { "epoch": 0.4960075665254006, "grad_norm": 1.9254201650619507, "learning_rate": 5.196047178833281e-06, "loss": 0.9762, "step": 27270 }, { "epoch": 0.4961894541552229, "grad_norm": 1.6309130191802979, "learning_rate": 5.1941720264771524e-06, "loss": 0.9425, "step": 27280 }, { "epoch": 0.4963713417850452, "grad_norm": 1.8007079362869263, "learning_rate": 5.192296874121022e-06, "loss": 0.9509, "step": 27290 }, { "epoch": 0.4965532294148675, "grad_norm": 2.3031022548675537, "learning_rate": 5.1904217217648934e-06, "loss": 0.8894, "step": 27300 }, { "epoch": 0.4967351170446898, "grad_norm": 1.7578288316726685, "learning_rate": 5.188546569408765e-06, "loss": 0.9159, "step": 27310 }, { "epoch": 0.4969170046745121, "grad_norm": 1.8686363697052002, "learning_rate": 5.186671417052635e-06, "loss": 0.9671, "step": 27320 }, { "epoch": 0.4970988923043344, "grad_norm": 1.660099983215332, "learning_rate": 5.184796264696507e-06, "loss": 0.9839, "step": 27330 }, { "epoch": 0.4972807799341567, "grad_norm": 1.5739390850067139, "learning_rate": 5.182921112340378e-06, "loss": 0.9091, "step": 27340 }, { "epoch": 0.497462667563979, "grad_norm": 1.9636890888214111, "learning_rate": 5.1810459599842485e-06, "loss": 0.9671, "step": 27350 }, { "epoch": 0.4976445551938013, "grad_norm": 1.6441775560379028, "learning_rate": 5.17917080762812e-06, "loss": 0.9447, "step": 27360 }, { "epoch": 0.49782644282362354, "grad_norm": 1.7975026369094849, "learning_rate": 5.177295655271991e-06, "loss": 0.9543, "step": 27370 }, { "epoch": 0.49800833045344584, "grad_norm": 2.0956289768218994, "learning_rate": 5.1754205029158625e-06, "loss": 0.9415, "step": 27380 }, { "epoch": 0.49819021808326813, "grad_norm": 1.7382558584213257, "learning_rate": 5.173545350559733e-06, "loss": 0.9431, "step": 27390 }, { "epoch": 0.49837210571309043, "grad_norm": 1.9321858882904053, "learning_rate": 5.171670198203604e-06, "loss": 0.9144, "step": 27400 }, { "epoch": 0.4985539933429127, "grad_norm": 2.007011890411377, "learning_rate": 5.169795045847476e-06, "loss": 0.9245, "step": 27410 }, { "epoch": 0.498735880972735, "grad_norm": 1.946610927581787, "learning_rate": 5.167919893491346e-06, "loss": 0.937, "step": 27420 }, { "epoch": 0.4989177686025573, "grad_norm": 1.5769853591918945, "learning_rate": 5.1660447411352175e-06, "loss": 0.8864, "step": 27430 }, { "epoch": 0.4990996562323796, "grad_norm": 2.004315137863159, "learning_rate": 5.164169588779089e-06, "loss": 0.8975, "step": 27440 }, { "epoch": 0.4992815438622019, "grad_norm": 1.9189525842666626, "learning_rate": 5.16229443642296e-06, "loss": 0.9627, "step": 27450 }, { "epoch": 0.4994634314920242, "grad_norm": 1.91278076171875, "learning_rate": 5.160419284066831e-06, "loss": 0.991, "step": 27460 }, { "epoch": 0.4996453191218465, "grad_norm": 1.7369033098220825, "learning_rate": 5.158544131710702e-06, "loss": 0.9226, "step": 27470 }, { "epoch": 0.4998272067516688, "grad_norm": 1.7931630611419678, "learning_rate": 5.156668979354573e-06, "loss": 0.934, "step": 27480 }, { "epoch": 0.5000090943814911, "grad_norm": 1.8640613555908203, "learning_rate": 5.154793826998444e-06, "loss": 0.9325, "step": 27490 }, { "epoch": 0.5001909820113134, "grad_norm": 1.865349531173706, "learning_rate": 5.152918674642315e-06, "loss": 0.9444, "step": 27500 }, { "epoch": 0.5003728696411357, "grad_norm": 1.7172973155975342, "learning_rate": 5.151043522286187e-06, "loss": 0.9344, "step": 27510 }, { "epoch": 0.500554757270958, "grad_norm": 1.7100263833999634, "learning_rate": 5.149168369930058e-06, "loss": 0.9011, "step": 27520 }, { "epoch": 0.5007366449007803, "grad_norm": 1.5951790809631348, "learning_rate": 5.1472932175739285e-06, "loss": 0.9084, "step": 27530 }, { "epoch": 0.5009185325306026, "grad_norm": 1.6689066886901855, "learning_rate": 5.1454180652178e-06, "loss": 0.9085, "step": 27540 }, { "epoch": 0.5011004201604249, "grad_norm": 1.867164134979248, "learning_rate": 5.143542912861671e-06, "loss": 0.9503, "step": 27550 }, { "epoch": 0.5012823077902472, "grad_norm": 1.6398952007293701, "learning_rate": 5.141667760505541e-06, "loss": 0.9185, "step": 27560 }, { "epoch": 0.5014641954200695, "grad_norm": 1.9351627826690674, "learning_rate": 5.139792608149412e-06, "loss": 0.901, "step": 27570 }, { "epoch": 0.5016460830498918, "grad_norm": 1.7612413167953491, "learning_rate": 5.1379174557932835e-06, "loss": 0.9301, "step": 27580 }, { "epoch": 0.5018279706797141, "grad_norm": 1.7336935997009277, "learning_rate": 5.136042303437154e-06, "loss": 0.938, "step": 27590 }, { "epoch": 0.5020098583095364, "grad_norm": 1.6922791004180908, "learning_rate": 5.134167151081025e-06, "loss": 0.9599, "step": 27600 }, { "epoch": 0.5021917459393587, "grad_norm": 1.7060588598251343, "learning_rate": 5.132291998724897e-06, "loss": 0.975, "step": 27610 }, { "epoch": 0.502373633569181, "grad_norm": 1.765971064567566, "learning_rate": 5.130416846368767e-06, "loss": 0.9502, "step": 27620 }, { "epoch": 0.5025555211990033, "grad_norm": 1.700237512588501, "learning_rate": 5.1285416940126385e-06, "loss": 0.9401, "step": 27630 }, { "epoch": 0.5027374088288256, "grad_norm": 1.764011025428772, "learning_rate": 5.12666654165651e-06, "loss": 0.9438, "step": 27640 }, { "epoch": 0.5029192964586479, "grad_norm": 1.7321616411209106, "learning_rate": 5.124791389300381e-06, "loss": 0.9492, "step": 27650 }, { "epoch": 0.5031011840884702, "grad_norm": 1.7623158693313599, "learning_rate": 5.122916236944252e-06, "loss": 0.8928, "step": 27660 }, { "epoch": 0.5032830717182925, "grad_norm": 1.7516065835952759, "learning_rate": 5.121041084588123e-06, "loss": 0.9181, "step": 27670 }, { "epoch": 0.5034649593481147, "grad_norm": 1.918930172920227, "learning_rate": 5.119165932231994e-06, "loss": 0.9543, "step": 27680 }, { "epoch": 0.503646846977937, "grad_norm": 1.832951307296753, "learning_rate": 5.117290779875865e-06, "loss": 0.9677, "step": 27690 }, { "epoch": 0.5038287346077593, "grad_norm": 1.82257878780365, "learning_rate": 5.115415627519736e-06, "loss": 0.9443, "step": 27700 }, { "epoch": 0.5040106222375816, "grad_norm": 1.817644476890564, "learning_rate": 5.113540475163608e-06, "loss": 0.9841, "step": 27710 }, { "epoch": 0.5041925098674039, "grad_norm": 1.62765634059906, "learning_rate": 5.111665322807479e-06, "loss": 0.9122, "step": 27720 }, { "epoch": 0.5043743974972262, "grad_norm": 1.5139933824539185, "learning_rate": 5.1097901704513494e-06, "loss": 0.9603, "step": 27730 }, { "epoch": 0.5045562851270485, "grad_norm": 1.7089998722076416, "learning_rate": 5.107915018095221e-06, "loss": 0.9455, "step": 27740 }, { "epoch": 0.5047381727568708, "grad_norm": 1.7287033796310425, "learning_rate": 5.106039865739092e-06, "loss": 0.9129, "step": 27750 }, { "epoch": 0.5049200603866931, "grad_norm": 1.7069302797317505, "learning_rate": 5.104164713382963e-06, "loss": 0.9268, "step": 27760 }, { "epoch": 0.5051019480165154, "grad_norm": 1.5218079090118408, "learning_rate": 5.102289561026834e-06, "loss": 0.9148, "step": 27770 }, { "epoch": 0.5052838356463377, "grad_norm": 1.715531587600708, "learning_rate": 5.100414408670705e-06, "loss": 0.9548, "step": 27780 }, { "epoch": 0.50546572327616, "grad_norm": 2.0890305042266846, "learning_rate": 5.098539256314577e-06, "loss": 0.9657, "step": 27790 }, { "epoch": 0.5056476109059823, "grad_norm": 1.872141718864441, "learning_rate": 5.096664103958447e-06, "loss": 0.9244, "step": 27800 }, { "epoch": 0.5058294985358046, "grad_norm": 1.8111239671707153, "learning_rate": 5.0947889516023185e-06, "loss": 0.8968, "step": 27810 }, { "epoch": 0.5060113861656269, "grad_norm": 1.7465794086456299, "learning_rate": 5.09291379924619e-06, "loss": 0.9015, "step": 27820 }, { "epoch": 0.5061932737954492, "grad_norm": 1.7096943855285645, "learning_rate": 5.09103864689006e-06, "loss": 0.9342, "step": 27830 }, { "epoch": 0.5063751614252715, "grad_norm": 1.6699851751327515, "learning_rate": 5.089163494533931e-06, "loss": 0.9367, "step": 27840 }, { "epoch": 0.5065570490550938, "grad_norm": 1.6692363023757935, "learning_rate": 5.087288342177802e-06, "loss": 0.9067, "step": 27850 }, { "epoch": 0.5067389366849161, "grad_norm": 1.8607301712036133, "learning_rate": 5.085413189821673e-06, "loss": 0.9805, "step": 27860 }, { "epoch": 0.5069208243147384, "grad_norm": 2.0388152599334717, "learning_rate": 5.083538037465544e-06, "loss": 0.9326, "step": 27870 }, { "epoch": 0.5071027119445607, "grad_norm": 1.650810956954956, "learning_rate": 5.081662885109415e-06, "loss": 0.8882, "step": 27880 }, { "epoch": 0.507284599574383, "grad_norm": 2.055002450942993, "learning_rate": 5.079787732753286e-06, "loss": 0.9463, "step": 27890 }, { "epoch": 0.5074664872042053, "grad_norm": 1.9631452560424805, "learning_rate": 5.077912580397157e-06, "loss": 0.9533, "step": 27900 }, { "epoch": 0.5076483748340276, "grad_norm": 1.5816434621810913, "learning_rate": 5.076037428041029e-06, "loss": 0.9337, "step": 27910 }, { "epoch": 0.5078302624638499, "grad_norm": 1.4585213661193848, "learning_rate": 5.0741622756849e-06, "loss": 0.9481, "step": 27920 }, { "epoch": 0.5080121500936722, "grad_norm": 1.8132458925247192, "learning_rate": 5.0722871233287704e-06, "loss": 0.9446, "step": 27930 }, { "epoch": 0.5081940377234945, "grad_norm": 1.667555570602417, "learning_rate": 5.070411970972642e-06, "loss": 0.8997, "step": 27940 }, { "epoch": 0.5083759253533168, "grad_norm": 1.915305256843567, "learning_rate": 5.068536818616513e-06, "loss": 0.9278, "step": 27950 }, { "epoch": 0.5085578129831391, "grad_norm": 1.588841199874878, "learning_rate": 5.066661666260384e-06, "loss": 0.9166, "step": 27960 }, { "epoch": 0.5087397006129614, "grad_norm": 1.5478357076644897, "learning_rate": 5.064786513904255e-06, "loss": 0.9168, "step": 27970 }, { "epoch": 0.5089215882427837, "grad_norm": 1.732718825340271, "learning_rate": 5.062911361548126e-06, "loss": 0.9263, "step": 27980 }, { "epoch": 0.509103475872606, "grad_norm": 1.934731364250183, "learning_rate": 5.061036209191998e-06, "loss": 0.8988, "step": 27990 }, { "epoch": 0.5092853635024281, "grad_norm": 1.408164381980896, "learning_rate": 5.059161056835868e-06, "loss": 0.9461, "step": 28000 }, { "epoch": 0.5094672511322504, "grad_norm": 1.6941912174224854, "learning_rate": 5.0572859044797395e-06, "loss": 0.941, "step": 28010 }, { "epoch": 0.5096491387620727, "grad_norm": 1.9431991577148438, "learning_rate": 5.055410752123611e-06, "loss": 0.9035, "step": 28020 }, { "epoch": 0.509831026391895, "grad_norm": 1.7643742561340332, "learning_rate": 5.053535599767481e-06, "loss": 0.967, "step": 28030 }, { "epoch": 0.5100129140217173, "grad_norm": 1.6412506103515625, "learning_rate": 5.051660447411353e-06, "loss": 0.9071, "step": 28040 }, { "epoch": 0.5101948016515396, "grad_norm": 1.698790431022644, "learning_rate": 5.049785295055224e-06, "loss": 0.8922, "step": 28050 }, { "epoch": 0.5103766892813619, "grad_norm": 1.8880112171173096, "learning_rate": 5.047910142699095e-06, "loss": 0.9471, "step": 28060 }, { "epoch": 0.5105585769111842, "grad_norm": 1.7147902250289917, "learning_rate": 5.046034990342966e-06, "loss": 0.9482, "step": 28070 }, { "epoch": 0.5107404645410065, "grad_norm": 2.1546804904937744, "learning_rate": 5.044159837986837e-06, "loss": 0.9323, "step": 28080 }, { "epoch": 0.5109223521708288, "grad_norm": 1.5421981811523438, "learning_rate": 5.042284685630709e-06, "loss": 0.895, "step": 28090 }, { "epoch": 0.5111042398006511, "grad_norm": 1.9185025691986084, "learning_rate": 5.040409533274579e-06, "loss": 0.9307, "step": 28100 }, { "epoch": 0.5112861274304734, "grad_norm": 1.9070810079574585, "learning_rate": 5.03853438091845e-06, "loss": 0.9189, "step": 28110 }, { "epoch": 0.5114680150602957, "grad_norm": 2.1129114627838135, "learning_rate": 5.036659228562321e-06, "loss": 0.9656, "step": 28120 }, { "epoch": 0.511649902690118, "grad_norm": 1.7082821130752563, "learning_rate": 5.0347840762061914e-06, "loss": 0.9475, "step": 28130 }, { "epoch": 0.5118317903199403, "grad_norm": 1.9067262411117554, "learning_rate": 5.032908923850063e-06, "loss": 0.9739, "step": 28140 }, { "epoch": 0.5120136779497626, "grad_norm": 1.7255147695541382, "learning_rate": 5.031033771493934e-06, "loss": 0.9225, "step": 28150 }, { "epoch": 0.5121955655795849, "grad_norm": 2.1110501289367676, "learning_rate": 5.029158619137805e-06, "loss": 0.9147, "step": 28160 }, { "epoch": 0.5123774532094072, "grad_norm": 1.7028230428695679, "learning_rate": 5.027283466781676e-06, "loss": 0.9006, "step": 28170 }, { "epoch": 0.5125593408392295, "grad_norm": 1.8196064233779907, "learning_rate": 5.025408314425547e-06, "loss": 0.9515, "step": 28180 }, { "epoch": 0.5127412284690518, "grad_norm": 1.6594399213790894, "learning_rate": 5.023533162069419e-06, "loss": 0.9086, "step": 28190 }, { "epoch": 0.5129231160988741, "grad_norm": 1.8304585218429565, "learning_rate": 5.021658009713289e-06, "loss": 0.9297, "step": 28200 }, { "epoch": 0.5131050037286964, "grad_norm": 1.6912245750427246, "learning_rate": 5.0197828573571605e-06, "loss": 0.9092, "step": 28210 }, { "epoch": 0.5132868913585187, "grad_norm": 1.5861530303955078, "learning_rate": 5.017907705001032e-06, "loss": 0.9328, "step": 28220 }, { "epoch": 0.513468778988341, "grad_norm": 2.1063311100006104, "learning_rate": 5.016032552644902e-06, "loss": 0.9394, "step": 28230 }, { "epoch": 0.5136506666181633, "grad_norm": 1.5556379556655884, "learning_rate": 5.014157400288774e-06, "loss": 0.9796, "step": 28240 }, { "epoch": 0.5138325542479856, "grad_norm": 1.6517800092697144, "learning_rate": 5.012282247932645e-06, "loss": 0.9379, "step": 28250 }, { "epoch": 0.5140144418778079, "grad_norm": 1.659123182296753, "learning_rate": 5.010407095576516e-06, "loss": 0.9206, "step": 28260 }, { "epoch": 0.5141963295076302, "grad_norm": 1.7413959503173828, "learning_rate": 5.008531943220387e-06, "loss": 0.8984, "step": 28270 }, { "epoch": 0.5143782171374525, "grad_norm": 1.710426926612854, "learning_rate": 5.006656790864258e-06, "loss": 0.9274, "step": 28280 }, { "epoch": 0.5145601047672748, "grad_norm": 1.7196457386016846, "learning_rate": 5.00478163850813e-06, "loss": 0.9352, "step": 28290 }, { "epoch": 0.5147419923970971, "grad_norm": 1.6752634048461914, "learning_rate": 5.002906486152e-06, "loss": 0.8981, "step": 28300 }, { "epoch": 0.5149238800269194, "grad_norm": 1.5469954013824463, "learning_rate": 5.001031333795871e-06, "loss": 0.8669, "step": 28310 }, { "epoch": 0.5151057676567417, "grad_norm": 1.7417079210281372, "learning_rate": 4.999156181439742e-06, "loss": 0.9374, "step": 28320 }, { "epoch": 0.515287655286564, "grad_norm": 1.862420678138733, "learning_rate": 4.997281029083613e-06, "loss": 0.9026, "step": 28330 }, { "epoch": 0.5154695429163862, "grad_norm": 1.7872474193572998, "learning_rate": 4.995405876727485e-06, "loss": 0.9559, "step": 28340 }, { "epoch": 0.5156514305462085, "grad_norm": 1.556990146636963, "learning_rate": 4.993530724371355e-06, "loss": 0.8655, "step": 28350 }, { "epoch": 0.5158333181760308, "grad_norm": 2.1369731426239014, "learning_rate": 4.9916555720152265e-06, "loss": 0.9265, "step": 28360 }, { "epoch": 0.5160152058058531, "grad_norm": 1.9577378034591675, "learning_rate": 4.989780419659098e-06, "loss": 0.9026, "step": 28370 }, { "epoch": 0.5161970934356754, "grad_norm": 1.8826055526733398, "learning_rate": 4.987905267302969e-06, "loss": 0.9655, "step": 28380 }, { "epoch": 0.5163789810654977, "grad_norm": 1.9350978136062622, "learning_rate": 4.98603011494684e-06, "loss": 0.9394, "step": 28390 }, { "epoch": 0.51656086869532, "grad_norm": 1.6946358680725098, "learning_rate": 4.984154962590711e-06, "loss": 0.9476, "step": 28400 }, { "epoch": 0.5167427563251423, "grad_norm": 1.5382535457611084, "learning_rate": 4.982279810234582e-06, "loss": 0.9502, "step": 28410 }, { "epoch": 0.5169246439549646, "grad_norm": 2.8472537994384766, "learning_rate": 4.980404657878453e-06, "loss": 0.9557, "step": 28420 }, { "epoch": 0.5171065315847869, "grad_norm": 1.8121683597564697, "learning_rate": 4.978529505522324e-06, "loss": 0.9659, "step": 28430 }, { "epoch": 0.5172884192146092, "grad_norm": 1.8850538730621338, "learning_rate": 4.9766543531661955e-06, "loss": 0.9372, "step": 28440 }, { "epoch": 0.5174703068444315, "grad_norm": 1.7608709335327148, "learning_rate": 4.974779200810067e-06, "loss": 0.9723, "step": 28450 }, { "epoch": 0.5176521944742538, "grad_norm": 1.6121933460235596, "learning_rate": 4.972904048453937e-06, "loss": 0.8833, "step": 28460 }, { "epoch": 0.5178340821040761, "grad_norm": 1.660927653312683, "learning_rate": 4.971028896097808e-06, "loss": 0.982, "step": 28470 }, { "epoch": 0.5180159697338984, "grad_norm": 1.903847098350525, "learning_rate": 4.969153743741679e-06, "loss": 0.9583, "step": 28480 }, { "epoch": 0.5181978573637207, "grad_norm": 1.4804279804229736, "learning_rate": 4.9672785913855506e-06, "loss": 0.9581, "step": 28490 }, { "epoch": 0.518379744993543, "grad_norm": 1.6075106859207153, "learning_rate": 4.965403439029421e-06, "loss": 0.941, "step": 28500 }, { "epoch": 0.5185616326233653, "grad_norm": 1.9201306104660034, "learning_rate": 4.963528286673292e-06, "loss": 0.9239, "step": 28510 }, { "epoch": 0.5187435202531876, "grad_norm": 1.5300731658935547, "learning_rate": 4.961653134317164e-06, "loss": 0.887, "step": 28520 }, { "epoch": 0.5189254078830099, "grad_norm": 1.8794856071472168, "learning_rate": 4.959777981961035e-06, "loss": 0.9442, "step": 28530 }, { "epoch": 0.5191072955128322, "grad_norm": 1.7621675729751587, "learning_rate": 4.957902829604906e-06, "loss": 0.9311, "step": 28540 }, { "epoch": 0.5192891831426545, "grad_norm": 1.7185416221618652, "learning_rate": 4.956027677248777e-06, "loss": 0.9438, "step": 28550 }, { "epoch": 0.5194710707724768, "grad_norm": 1.759386658668518, "learning_rate": 4.954152524892648e-06, "loss": 0.9464, "step": 28560 }, { "epoch": 0.5196529584022991, "grad_norm": 1.8944493532180786, "learning_rate": 4.952277372536519e-06, "loss": 0.9422, "step": 28570 }, { "epoch": 0.5198348460321214, "grad_norm": 1.6695692539215088, "learning_rate": 4.95040222018039e-06, "loss": 0.9711, "step": 28580 }, { "epoch": 0.5200167336619437, "grad_norm": 1.8397133350372314, "learning_rate": 4.9485270678242615e-06, "loss": 0.9763, "step": 28590 }, { "epoch": 0.520198621291766, "grad_norm": 1.6710008382797241, "learning_rate": 4.946651915468132e-06, "loss": 0.9497, "step": 28600 }, { "epoch": 0.5203805089215883, "grad_norm": 2.1245839595794678, "learning_rate": 4.944776763112003e-06, "loss": 0.9584, "step": 28610 }, { "epoch": 0.5205623965514106, "grad_norm": 1.7579052448272705, "learning_rate": 4.942901610755874e-06, "loss": 0.9338, "step": 28620 }, { "epoch": 0.5207442841812329, "grad_norm": 1.6415201425552368, "learning_rate": 4.941026458399745e-06, "loss": 0.9359, "step": 28630 }, { "epoch": 0.5209261718110552, "grad_norm": 1.7992873191833496, "learning_rate": 4.9391513060436165e-06, "loss": 0.9735, "step": 28640 }, { "epoch": 0.5211080594408775, "grad_norm": 1.8018790483474731, "learning_rate": 4.937276153687488e-06, "loss": 0.9659, "step": 28650 }, { "epoch": 0.5212899470706998, "grad_norm": 1.7031961679458618, "learning_rate": 4.935401001331358e-06, "loss": 0.9445, "step": 28660 }, { "epoch": 0.521471834700522, "grad_norm": 1.8574272394180298, "learning_rate": 4.93352584897523e-06, "loss": 0.9872, "step": 28670 }, { "epoch": 0.5216537223303443, "grad_norm": 1.9039596319198608, "learning_rate": 4.931650696619101e-06, "loss": 0.944, "step": 28680 }, { "epoch": 0.5218356099601666, "grad_norm": 1.8077912330627441, "learning_rate": 4.9297755442629716e-06, "loss": 0.9336, "step": 28690 }, { "epoch": 0.5220174975899889, "grad_norm": 1.6581701040267944, "learning_rate": 4.927900391906843e-06, "loss": 0.9287, "step": 28700 }, { "epoch": 0.5221993852198112, "grad_norm": 1.734019160270691, "learning_rate": 4.926025239550714e-06, "loss": 0.9618, "step": 28710 }, { "epoch": 0.5223812728496335, "grad_norm": 1.778205156326294, "learning_rate": 4.924150087194586e-06, "loss": 0.9313, "step": 28720 }, { "epoch": 0.5225631604794558, "grad_norm": 1.4659720659255981, "learning_rate": 4.922274934838456e-06, "loss": 0.9554, "step": 28730 }, { "epoch": 0.5227450481092781, "grad_norm": 1.7408289909362793, "learning_rate": 4.920399782482327e-06, "loss": 0.9622, "step": 28740 }, { "epoch": 0.5229269357391004, "grad_norm": 1.8289926052093506, "learning_rate": 4.918524630126198e-06, "loss": 0.9637, "step": 28750 }, { "epoch": 0.5231088233689227, "grad_norm": 1.5667299032211304, "learning_rate": 4.916649477770069e-06, "loss": 0.9106, "step": 28760 }, { "epoch": 0.523290710998745, "grad_norm": 1.7546194791793823, "learning_rate": 4.91477432541394e-06, "loss": 0.9315, "step": 28770 }, { "epoch": 0.5234725986285673, "grad_norm": 1.8818204402923584, "learning_rate": 4.912899173057811e-06, "loss": 0.9353, "step": 28780 }, { "epoch": 0.5236544862583896, "grad_norm": 1.7691248655319214, "learning_rate": 4.9110240207016825e-06, "loss": 0.9278, "step": 28790 }, { "epoch": 0.5238363738882119, "grad_norm": 1.982174277305603, "learning_rate": 4.909148868345554e-06, "loss": 0.9083, "step": 28800 }, { "epoch": 0.5240182615180342, "grad_norm": 1.8157793283462524, "learning_rate": 4.907273715989424e-06, "loss": 0.9533, "step": 28810 }, { "epoch": 0.5242001491478565, "grad_norm": 1.907267451286316, "learning_rate": 4.905398563633296e-06, "loss": 0.9204, "step": 28820 }, { "epoch": 0.5243820367776787, "grad_norm": 1.7415707111358643, "learning_rate": 4.903523411277167e-06, "loss": 0.9133, "step": 28830 }, { "epoch": 0.524563924407501, "grad_norm": 1.8638232946395874, "learning_rate": 4.9016482589210375e-06, "loss": 0.8909, "step": 28840 }, { "epoch": 0.5247458120373233, "grad_norm": 1.6917760372161865, "learning_rate": 4.899773106564909e-06, "loss": 0.9063, "step": 28850 }, { "epoch": 0.5249276996671456, "grad_norm": 1.8226263523101807, "learning_rate": 4.89789795420878e-06, "loss": 0.9116, "step": 28860 }, { "epoch": 0.5251095872969679, "grad_norm": 2.0195095539093018, "learning_rate": 4.896022801852651e-06, "loss": 0.9451, "step": 28870 }, { "epoch": 0.5252914749267902, "grad_norm": 1.8048875331878662, "learning_rate": 4.894147649496522e-06, "loss": 0.9143, "step": 28880 }, { "epoch": 0.5254733625566125, "grad_norm": 2.0613269805908203, "learning_rate": 4.8922724971403926e-06, "loss": 0.9495, "step": 28890 }, { "epoch": 0.5256552501864348, "grad_norm": 1.6987861394882202, "learning_rate": 4.890397344784264e-06, "loss": 0.9138, "step": 28900 }, { "epoch": 0.5258371378162571, "grad_norm": 1.6912016868591309, "learning_rate": 4.888522192428135e-06, "loss": 0.9401, "step": 28910 }, { "epoch": 0.5260190254460794, "grad_norm": 1.6561483144760132, "learning_rate": 4.886647040072007e-06, "loss": 0.9268, "step": 28920 }, { "epoch": 0.5262009130759017, "grad_norm": 1.8610457181930542, "learning_rate": 4.884771887715877e-06, "loss": 0.9325, "step": 28930 }, { "epoch": 0.526382800705724, "grad_norm": 1.88031804561615, "learning_rate": 4.8828967353597484e-06, "loss": 0.9453, "step": 28940 }, { "epoch": 0.5265646883355463, "grad_norm": 1.9766699075698853, "learning_rate": 4.88102158300362e-06, "loss": 0.9402, "step": 28950 }, { "epoch": 0.5267465759653686, "grad_norm": 1.9315474033355713, "learning_rate": 4.87914643064749e-06, "loss": 0.9484, "step": 28960 }, { "epoch": 0.5269284635951909, "grad_norm": 1.897608995437622, "learning_rate": 4.877271278291362e-06, "loss": 0.8919, "step": 28970 }, { "epoch": 0.5271103512250132, "grad_norm": 1.8112660646438599, "learning_rate": 4.875396125935233e-06, "loss": 0.9496, "step": 28980 }, { "epoch": 0.5272922388548354, "grad_norm": 1.6700665950775146, "learning_rate": 4.873520973579104e-06, "loss": 0.9261, "step": 28990 }, { "epoch": 0.5274741264846577, "grad_norm": 1.944116234779358, "learning_rate": 4.871645821222975e-06, "loss": 0.9007, "step": 29000 }, { "epoch": 0.52765601411448, "grad_norm": 1.869511604309082, "learning_rate": 4.869770668866845e-06, "loss": 0.9158, "step": 29010 }, { "epoch": 0.5278379017443023, "grad_norm": 1.532313585281372, "learning_rate": 4.867895516510717e-06, "loss": 0.9169, "step": 29020 }, { "epoch": 0.5280197893741246, "grad_norm": 2.1330137252807617, "learning_rate": 4.866020364154588e-06, "loss": 0.9876, "step": 29030 }, { "epoch": 0.5282016770039469, "grad_norm": 1.4437569379806519, "learning_rate": 4.8641452117984585e-06, "loss": 0.9063, "step": 29040 }, { "epoch": 0.5283835646337692, "grad_norm": 1.6962612867355347, "learning_rate": 4.86227005944233e-06, "loss": 0.9274, "step": 29050 }, { "epoch": 0.5285654522635915, "grad_norm": 1.628200888633728, "learning_rate": 4.860394907086201e-06, "loss": 0.8904, "step": 29060 }, { "epoch": 0.5287473398934138, "grad_norm": 1.7013524770736694, "learning_rate": 4.8585197547300725e-06, "loss": 0.9552, "step": 29070 }, { "epoch": 0.5289292275232361, "grad_norm": 1.8084032535552979, "learning_rate": 4.856644602373943e-06, "loss": 0.9629, "step": 29080 }, { "epoch": 0.5291111151530584, "grad_norm": 1.9745742082595825, "learning_rate": 4.854769450017814e-06, "loss": 0.959, "step": 29090 }, { "epoch": 0.5292930027828807, "grad_norm": 1.6550484895706177, "learning_rate": 4.852894297661686e-06, "loss": 0.8903, "step": 29100 }, { "epoch": 0.529474890412703, "grad_norm": 1.7503348588943481, "learning_rate": 4.851019145305556e-06, "loss": 0.9302, "step": 29110 }, { "epoch": 0.5296567780425253, "grad_norm": 1.8597761392593384, "learning_rate": 4.849143992949428e-06, "loss": 0.9265, "step": 29120 }, { "epoch": 0.5298386656723476, "grad_norm": 1.6740473508834839, "learning_rate": 4.847268840593299e-06, "loss": 0.9676, "step": 29130 }, { "epoch": 0.5300205533021699, "grad_norm": 1.8406171798706055, "learning_rate": 4.84539368823717e-06, "loss": 0.9776, "step": 29140 }, { "epoch": 0.5302024409319922, "grad_norm": 2.006584644317627, "learning_rate": 4.843518535881041e-06, "loss": 0.9671, "step": 29150 }, { "epoch": 0.5303843285618145, "grad_norm": 1.362302541732788, "learning_rate": 4.841643383524911e-06, "loss": 0.8999, "step": 29160 }, { "epoch": 0.5305662161916368, "grad_norm": 1.7892321348190308, "learning_rate": 4.839768231168783e-06, "loss": 0.9498, "step": 29170 }, { "epoch": 0.5307481038214591, "grad_norm": 1.8787633180618286, "learning_rate": 4.837893078812654e-06, "loss": 0.9572, "step": 29180 }, { "epoch": 0.5309299914512814, "grad_norm": 1.9824796915054321, "learning_rate": 4.836017926456525e-06, "loss": 0.8553, "step": 29190 }, { "epoch": 0.5311118790811037, "grad_norm": 1.6781057119369507, "learning_rate": 4.834142774100396e-06, "loss": 0.9454, "step": 29200 }, { "epoch": 0.531293766710926, "grad_norm": 2.0033795833587646, "learning_rate": 4.832267621744267e-06, "loss": 0.9391, "step": 29210 }, { "epoch": 0.5314756543407483, "grad_norm": 1.5616178512573242, "learning_rate": 4.8303924693881385e-06, "loss": 0.905, "step": 29220 }, { "epoch": 0.5316575419705706, "grad_norm": 1.8889305591583252, "learning_rate": 4.828517317032009e-06, "loss": 0.9435, "step": 29230 }, { "epoch": 0.5318394296003929, "grad_norm": 1.572322130203247, "learning_rate": 4.82664216467588e-06, "loss": 0.9032, "step": 29240 }, { "epoch": 0.5320213172302152, "grad_norm": 1.9575142860412598, "learning_rate": 4.824767012319752e-06, "loss": 0.9207, "step": 29250 }, { "epoch": 0.5322032048600375, "grad_norm": 1.4209097623825073, "learning_rate": 4.822891859963623e-06, "loss": 0.9404, "step": 29260 }, { "epoch": 0.5323850924898598, "grad_norm": 1.593551516532898, "learning_rate": 4.8210167076074935e-06, "loss": 0.8788, "step": 29270 }, { "epoch": 0.5325669801196821, "grad_norm": 1.6503963470458984, "learning_rate": 4.819141555251365e-06, "loss": 0.8742, "step": 29280 }, { "epoch": 0.5327488677495044, "grad_norm": 1.9016822576522827, "learning_rate": 4.817266402895235e-06, "loss": 0.9035, "step": 29290 }, { "epoch": 0.5329307553793267, "grad_norm": 1.9850623607635498, "learning_rate": 4.815391250539107e-06, "loss": 0.9298, "step": 29300 }, { "epoch": 0.533112643009149, "grad_norm": 1.5893802642822266, "learning_rate": 4.813516098182977e-06, "loss": 0.9888, "step": 29310 }, { "epoch": 0.5332945306389713, "grad_norm": 1.7152743339538574, "learning_rate": 4.8116409458268486e-06, "loss": 0.8404, "step": 29320 }, { "epoch": 0.5334764182687936, "grad_norm": 1.7532532215118408, "learning_rate": 4.80976579347072e-06, "loss": 0.9348, "step": 29330 }, { "epoch": 0.5336583058986158, "grad_norm": 1.7479454278945923, "learning_rate": 4.807890641114591e-06, "loss": 0.9324, "step": 29340 }, { "epoch": 0.5338401935284381, "grad_norm": 1.8707996606826782, "learning_rate": 4.806015488758462e-06, "loss": 0.9279, "step": 29350 }, { "epoch": 0.5340220811582604, "grad_norm": 1.8385226726531982, "learning_rate": 4.804140336402333e-06, "loss": 0.9306, "step": 29360 }, { "epoch": 0.5342039687880827, "grad_norm": 1.5597219467163086, "learning_rate": 4.8022651840462045e-06, "loss": 0.9111, "step": 29370 }, { "epoch": 0.534385856417905, "grad_norm": 1.6995303630828857, "learning_rate": 4.800390031690075e-06, "loss": 0.9496, "step": 29380 }, { "epoch": 0.5345677440477273, "grad_norm": 1.8509466648101807, "learning_rate": 4.798514879333946e-06, "loss": 0.9439, "step": 29390 }, { "epoch": 0.5347496316775496, "grad_norm": 1.7241860628128052, "learning_rate": 4.796639726977818e-06, "loss": 0.9657, "step": 29400 }, { "epoch": 0.5349315193073719, "grad_norm": 2.0764145851135254, "learning_rate": 4.794764574621689e-06, "loss": 0.9118, "step": 29410 }, { "epoch": 0.5351134069371942, "grad_norm": 1.7709485292434692, "learning_rate": 4.7928894222655595e-06, "loss": 0.9242, "step": 29420 }, { "epoch": 0.5352952945670165, "grad_norm": 1.7392661571502686, "learning_rate": 4.79101426990943e-06, "loss": 0.9553, "step": 29430 }, { "epoch": 0.5354771821968388, "grad_norm": 1.6478362083435059, "learning_rate": 4.789139117553301e-06, "loss": 0.9509, "step": 29440 }, { "epoch": 0.5356590698266611, "grad_norm": 1.8979146480560303, "learning_rate": 4.787263965197173e-06, "loss": 0.8933, "step": 29450 }, { "epoch": 0.5358409574564834, "grad_norm": 1.6993882656097412, "learning_rate": 4.785388812841044e-06, "loss": 0.9345, "step": 29460 }, { "epoch": 0.5360228450863057, "grad_norm": 1.7007336616516113, "learning_rate": 4.7835136604849145e-06, "loss": 0.9318, "step": 29470 }, { "epoch": 0.536204732716128, "grad_norm": 1.3423607349395752, "learning_rate": 4.781638508128786e-06, "loss": 0.8831, "step": 29480 }, { "epoch": 0.5363866203459503, "grad_norm": 1.3738720417022705, "learning_rate": 4.779763355772657e-06, "loss": 0.903, "step": 29490 }, { "epoch": 0.5365685079757726, "grad_norm": 1.9113595485687256, "learning_rate": 4.777888203416528e-06, "loss": 0.9977, "step": 29500 }, { "epoch": 0.5367503956055949, "grad_norm": 1.8054567575454712, "learning_rate": 4.776013051060399e-06, "loss": 0.9309, "step": 29510 }, { "epoch": 0.5369322832354172, "grad_norm": 1.8454992771148682, "learning_rate": 4.77413789870427e-06, "loss": 0.9219, "step": 29520 }, { "epoch": 0.5371141708652395, "grad_norm": 1.7343976497650146, "learning_rate": 4.772262746348142e-06, "loss": 0.9641, "step": 29530 }, { "epoch": 0.5372960584950618, "grad_norm": 1.8239952325820923, "learning_rate": 4.770387593992012e-06, "loss": 0.9254, "step": 29540 }, { "epoch": 0.5374779461248841, "grad_norm": 1.6698511838912964, "learning_rate": 4.768512441635884e-06, "loss": 0.9562, "step": 29550 }, { "epoch": 0.5376598337547064, "grad_norm": 1.4010450839996338, "learning_rate": 4.766637289279754e-06, "loss": 0.9566, "step": 29560 }, { "epoch": 0.5378417213845287, "grad_norm": 2.1057088375091553, "learning_rate": 4.7647621369236254e-06, "loss": 0.9243, "step": 29570 }, { "epoch": 0.538023609014351, "grad_norm": 1.7416343688964844, "learning_rate": 4.762886984567496e-06, "loss": 0.9301, "step": 29580 }, { "epoch": 0.5382054966441733, "grad_norm": 1.822641134262085, "learning_rate": 4.761011832211367e-06, "loss": 0.9426, "step": 29590 }, { "epoch": 0.5383873842739956, "grad_norm": 1.9623198509216309, "learning_rate": 4.759136679855239e-06, "loss": 0.9128, "step": 29600 }, { "epoch": 0.5385692719038179, "grad_norm": 2.0367817878723145, "learning_rate": 4.75726152749911e-06, "loss": 0.9295, "step": 29610 }, { "epoch": 0.5387511595336402, "grad_norm": 1.9567114114761353, "learning_rate": 4.7553863751429805e-06, "loss": 0.9026, "step": 29620 }, { "epoch": 0.5389330471634625, "grad_norm": 1.6800367832183838, "learning_rate": 4.753511222786852e-06, "loss": 0.9469, "step": 29630 }, { "epoch": 0.5391149347932848, "grad_norm": 1.6842610836029053, "learning_rate": 4.751636070430723e-06, "loss": 1.0201, "step": 29640 }, { "epoch": 0.5392968224231071, "grad_norm": 1.7676087617874146, "learning_rate": 4.749760918074594e-06, "loss": 0.9155, "step": 29650 }, { "epoch": 0.5394787100529294, "grad_norm": 1.9534484148025513, "learning_rate": 4.747885765718465e-06, "loss": 0.9419, "step": 29660 }, { "epoch": 0.5396605976827515, "grad_norm": 1.6623222827911377, "learning_rate": 4.746010613362336e-06, "loss": 0.9265, "step": 29670 }, { "epoch": 0.5398424853125738, "grad_norm": 2.0141758918762207, "learning_rate": 4.744135461006208e-06, "loss": 0.9347, "step": 29680 }, { "epoch": 0.5400243729423961, "grad_norm": 1.4771662950515747, "learning_rate": 4.742260308650078e-06, "loss": 0.9045, "step": 29690 }, { "epoch": 0.5402062605722184, "grad_norm": 1.9478840827941895, "learning_rate": 4.740385156293949e-06, "loss": 0.8804, "step": 29700 }, { "epoch": 0.5403881482020407, "grad_norm": 2.0552260875701904, "learning_rate": 4.73851000393782e-06, "loss": 0.952, "step": 29710 }, { "epoch": 0.540570035831863, "grad_norm": 1.8416754007339478, "learning_rate": 4.736634851581691e-06, "loss": 0.9077, "step": 29720 }, { "epoch": 0.5407519234616853, "grad_norm": 2.022157669067383, "learning_rate": 4.734759699225563e-06, "loss": 0.9121, "step": 29730 }, { "epoch": 0.5409338110915076, "grad_norm": 2.017209053039551, "learning_rate": 4.732884546869433e-06, "loss": 0.8809, "step": 29740 }, { "epoch": 0.5411156987213299, "grad_norm": 1.88423490524292, "learning_rate": 4.731009394513305e-06, "loss": 1.0161, "step": 29750 }, { "epoch": 0.5412975863511522, "grad_norm": 1.9593629837036133, "learning_rate": 4.729134242157176e-06, "loss": 0.937, "step": 29760 }, { "epoch": 0.5414794739809745, "grad_norm": 1.5457786321640015, "learning_rate": 4.7272590898010464e-06, "loss": 0.9517, "step": 29770 }, { "epoch": 0.5416613616107968, "grad_norm": 1.8909605741500854, "learning_rate": 4.725383937444918e-06, "loss": 0.9701, "step": 29780 }, { "epoch": 0.5418432492406191, "grad_norm": 1.8771177530288696, "learning_rate": 4.723508785088789e-06, "loss": 0.8876, "step": 29790 }, { "epoch": 0.5420251368704414, "grad_norm": 1.8317878246307373, "learning_rate": 4.7216336327326605e-06, "loss": 0.879, "step": 29800 }, { "epoch": 0.5422070245002637, "grad_norm": 1.4886330366134644, "learning_rate": 4.719758480376531e-06, "loss": 0.9784, "step": 29810 }, { "epoch": 0.542388912130086, "grad_norm": 1.799000859260559, "learning_rate": 4.717883328020402e-06, "loss": 0.9249, "step": 29820 }, { "epoch": 0.5425707997599083, "grad_norm": 1.8290956020355225, "learning_rate": 4.716008175664274e-06, "loss": 0.9102, "step": 29830 }, { "epoch": 0.5427526873897306, "grad_norm": 1.5032947063446045, "learning_rate": 4.714133023308144e-06, "loss": 0.9375, "step": 29840 }, { "epoch": 0.5429345750195529, "grad_norm": 1.9619691371917725, "learning_rate": 4.712257870952015e-06, "loss": 0.9363, "step": 29850 }, { "epoch": 0.5431164626493752, "grad_norm": 1.9494624137878418, "learning_rate": 4.710382718595886e-06, "loss": 0.8863, "step": 29860 }, { "epoch": 0.5432983502791975, "grad_norm": 1.8041497468948364, "learning_rate": 4.708507566239757e-06, "loss": 0.9264, "step": 29870 }, { "epoch": 0.5434802379090198, "grad_norm": 1.6270350217819214, "learning_rate": 4.706632413883629e-06, "loss": 0.9172, "step": 29880 }, { "epoch": 0.5436621255388421, "grad_norm": 1.6229300498962402, "learning_rate": 4.704757261527499e-06, "loss": 0.8881, "step": 29890 }, { "epoch": 0.5438440131686644, "grad_norm": 1.6086441278457642, "learning_rate": 4.7028821091713706e-06, "loss": 0.9164, "step": 29900 }, { "epoch": 0.5440259007984867, "grad_norm": 1.9682828187942505, "learning_rate": 4.701006956815242e-06, "loss": 0.9701, "step": 29910 }, { "epoch": 0.544207788428309, "grad_norm": 1.6798253059387207, "learning_rate": 4.699131804459112e-06, "loss": 0.9161, "step": 29920 }, { "epoch": 0.5443896760581313, "grad_norm": 2.0817670822143555, "learning_rate": 4.697256652102984e-06, "loss": 0.9901, "step": 29930 }, { "epoch": 0.5445715636879536, "grad_norm": 1.5867139101028442, "learning_rate": 4.695381499746855e-06, "loss": 0.8794, "step": 29940 }, { "epoch": 0.5447534513177759, "grad_norm": 1.6647294759750366, "learning_rate": 4.6935063473907264e-06, "loss": 0.8879, "step": 29950 }, { "epoch": 0.5449353389475982, "grad_norm": 1.9279026985168457, "learning_rate": 4.691631195034597e-06, "loss": 0.9216, "step": 29960 }, { "epoch": 0.5451172265774205, "grad_norm": 2.089212656021118, "learning_rate": 4.6897560426784674e-06, "loss": 0.9806, "step": 29970 }, { "epoch": 0.5452991142072428, "grad_norm": 1.7073360681533813, "learning_rate": 4.687880890322339e-06, "loss": 0.9092, "step": 29980 }, { "epoch": 0.545481001837065, "grad_norm": 1.8282935619354248, "learning_rate": 4.68600573796621e-06, "loss": 0.9304, "step": 29990 }, { "epoch": 0.5456628894668873, "grad_norm": 1.7891631126403809, "learning_rate": 4.6841305856100815e-06, "loss": 0.8425, "step": 30000 }, { "epoch": 0.5458447770967096, "grad_norm": 1.7716052532196045, "learning_rate": 4.682255433253952e-06, "loss": 0.8744, "step": 30010 }, { "epoch": 0.5460266647265319, "grad_norm": 1.4290790557861328, "learning_rate": 4.680380280897823e-06, "loss": 0.9235, "step": 30020 }, { "epoch": 0.5462085523563542, "grad_norm": 1.7622125148773193, "learning_rate": 4.678505128541695e-06, "loss": 0.8872, "step": 30030 }, { "epoch": 0.5463904399861765, "grad_norm": 1.7551732063293457, "learning_rate": 4.676629976185565e-06, "loss": 0.8504, "step": 30040 }, { "epoch": 0.5465723276159988, "grad_norm": 1.6092852354049683, "learning_rate": 4.6747548238294365e-06, "loss": 0.9278, "step": 30050 }, { "epoch": 0.5467542152458211, "grad_norm": 1.907472848892212, "learning_rate": 4.672879671473308e-06, "loss": 0.8809, "step": 30060 }, { "epoch": 0.5469361028756434, "grad_norm": 1.7210683822631836, "learning_rate": 4.671004519117179e-06, "loss": 0.9356, "step": 30070 }, { "epoch": 0.5471179905054657, "grad_norm": 1.824859619140625, "learning_rate": 4.66912936676105e-06, "loss": 0.9546, "step": 30080 }, { "epoch": 0.547299878135288, "grad_norm": 1.5705376863479614, "learning_rate": 4.667254214404921e-06, "loss": 0.937, "step": 30090 }, { "epoch": 0.5474817657651103, "grad_norm": 1.9508434534072876, "learning_rate": 4.665379062048792e-06, "loss": 0.8994, "step": 30100 }, { "epoch": 0.5476636533949326, "grad_norm": 1.3149646520614624, "learning_rate": 4.663503909692663e-06, "loss": 0.8673, "step": 30110 }, { "epoch": 0.5478455410247549, "grad_norm": 1.8131937980651855, "learning_rate": 4.661628757336533e-06, "loss": 0.8833, "step": 30120 }, { "epoch": 0.5480274286545772, "grad_norm": 1.8727097511291504, "learning_rate": 4.659753604980405e-06, "loss": 0.9303, "step": 30130 }, { "epoch": 0.5482093162843995, "grad_norm": 1.7763073444366455, "learning_rate": 4.657878452624276e-06, "loss": 0.9095, "step": 30140 }, { "epoch": 0.5483912039142218, "grad_norm": 1.8258219957351685, "learning_rate": 4.6560033002681474e-06, "loss": 0.9607, "step": 30150 }, { "epoch": 0.5485730915440441, "grad_norm": 1.7472764253616333, "learning_rate": 4.654128147912018e-06, "loss": 0.9391, "step": 30160 }, { "epoch": 0.5487549791738664, "grad_norm": 2.0682904720306396, "learning_rate": 4.652252995555889e-06, "loss": 0.8985, "step": 30170 }, { "epoch": 0.5489368668036887, "grad_norm": 1.97250497341156, "learning_rate": 4.650377843199761e-06, "loss": 0.9635, "step": 30180 }, { "epoch": 0.549118754433511, "grad_norm": 1.9141507148742676, "learning_rate": 4.648502690843631e-06, "loss": 0.9347, "step": 30190 }, { "epoch": 0.5493006420633333, "grad_norm": 1.7201939821243286, "learning_rate": 4.6466275384875025e-06, "loss": 0.8892, "step": 30200 }, { "epoch": 0.5494825296931556, "grad_norm": 1.5154184103012085, "learning_rate": 4.644752386131374e-06, "loss": 0.9237, "step": 30210 }, { "epoch": 0.5496644173229779, "grad_norm": 1.5725326538085938, "learning_rate": 4.642877233775245e-06, "loss": 0.9012, "step": 30220 }, { "epoch": 0.5498463049528002, "grad_norm": 1.9159680604934692, "learning_rate": 4.641002081419116e-06, "loss": 0.9208, "step": 30230 }, { "epoch": 0.5500281925826225, "grad_norm": 1.5341054201126099, "learning_rate": 4.639126929062987e-06, "loss": 0.9135, "step": 30240 }, { "epoch": 0.5502100802124448, "grad_norm": 1.5628615617752075, "learning_rate": 4.6372517767068575e-06, "loss": 0.9099, "step": 30250 }, { "epoch": 0.5503919678422671, "grad_norm": 1.8907455205917358, "learning_rate": 4.635376624350729e-06, "loss": 0.9225, "step": 30260 }, { "epoch": 0.5505738554720894, "grad_norm": 1.7685147523880005, "learning_rate": 4.633501471994599e-06, "loss": 0.9338, "step": 30270 }, { "epoch": 0.5507557431019117, "grad_norm": 1.627261757850647, "learning_rate": 4.631626319638471e-06, "loss": 0.9214, "step": 30280 }, { "epoch": 0.550937630731734, "grad_norm": 1.8197919130325317, "learning_rate": 4.629751167282342e-06, "loss": 0.8928, "step": 30290 }, { "epoch": 0.5511195183615563, "grad_norm": 1.5422248840332031, "learning_rate": 4.627876014926213e-06, "loss": 0.8953, "step": 30300 }, { "epoch": 0.5513014059913786, "grad_norm": 1.7484406232833862, "learning_rate": 4.626000862570084e-06, "loss": 0.9117, "step": 30310 }, { "epoch": 0.5514832936212009, "grad_norm": 1.88404381275177, "learning_rate": 4.624125710213955e-06, "loss": 0.9612, "step": 30320 }, { "epoch": 0.5516651812510232, "grad_norm": 1.6889588832855225, "learning_rate": 4.6222505578578266e-06, "loss": 0.8943, "step": 30330 }, { "epoch": 0.5518470688808454, "grad_norm": 1.9049310684204102, "learning_rate": 4.620375405501698e-06, "loss": 0.891, "step": 30340 }, { "epoch": 0.5520289565106677, "grad_norm": 1.672253966331482, "learning_rate": 4.618500253145568e-06, "loss": 0.9141, "step": 30350 }, { "epoch": 0.55221084414049, "grad_norm": 1.973272681236267, "learning_rate": 4.61662510078944e-06, "loss": 0.884, "step": 30360 }, { "epoch": 0.5523927317703123, "grad_norm": 1.984662652015686, "learning_rate": 4.614749948433311e-06, "loss": 0.9356, "step": 30370 }, { "epoch": 0.5525746194001346, "grad_norm": 2.11368727684021, "learning_rate": 4.612874796077182e-06, "loss": 0.9393, "step": 30380 }, { "epoch": 0.5527565070299569, "grad_norm": 1.3701233863830566, "learning_rate": 4.610999643721052e-06, "loss": 0.9083, "step": 30390 }, { "epoch": 0.5529383946597792, "grad_norm": 1.6295621395111084, "learning_rate": 4.6091244913649235e-06, "loss": 0.8948, "step": 30400 }, { "epoch": 0.5531202822896015, "grad_norm": 1.8967338800430298, "learning_rate": 4.607249339008795e-06, "loss": 0.9814, "step": 30410 }, { "epoch": 0.5533021699194238, "grad_norm": 1.9587637186050415, "learning_rate": 4.605374186652666e-06, "loss": 0.9811, "step": 30420 }, { "epoch": 0.5534840575492461, "grad_norm": 1.9537910223007202, "learning_rate": 4.603499034296537e-06, "loss": 0.9445, "step": 30430 }, { "epoch": 0.5536659451790684, "grad_norm": 2.0022172927856445, "learning_rate": 4.601623881940408e-06, "loss": 0.9302, "step": 30440 }, { "epoch": 0.5538478328088907, "grad_norm": 1.5055766105651855, "learning_rate": 4.599748729584279e-06, "loss": 0.9321, "step": 30450 }, { "epoch": 0.554029720438713, "grad_norm": 1.9123190641403198, "learning_rate": 4.59787357722815e-06, "loss": 0.9067, "step": 30460 }, { "epoch": 0.5542116080685353, "grad_norm": 1.813563585281372, "learning_rate": 4.595998424872021e-06, "loss": 0.945, "step": 30470 }, { "epoch": 0.5543934956983576, "grad_norm": 1.7191497087478638, "learning_rate": 4.5941232725158925e-06, "loss": 0.9725, "step": 30480 }, { "epoch": 0.5545753833281799, "grad_norm": 1.8983311653137207, "learning_rate": 4.592248120159764e-06, "loss": 0.9242, "step": 30490 }, { "epoch": 0.5547572709580021, "grad_norm": 1.7595092058181763, "learning_rate": 4.590372967803634e-06, "loss": 0.971, "step": 30500 }, { "epoch": 0.5549391585878244, "grad_norm": 1.859959363937378, "learning_rate": 4.588497815447506e-06, "loss": 0.9367, "step": 30510 }, { "epoch": 0.5551210462176467, "grad_norm": 2.15088152885437, "learning_rate": 4.586622663091377e-06, "loss": 0.9389, "step": 30520 }, { "epoch": 0.555302933847469, "grad_norm": 1.603073239326477, "learning_rate": 4.5847475107352476e-06, "loss": 0.9571, "step": 30530 }, { "epoch": 0.5554848214772913, "grad_norm": 1.6694890260696411, "learning_rate": 4.582872358379118e-06, "loss": 0.9345, "step": 30540 }, { "epoch": 0.5556667091071136, "grad_norm": 2.0611371994018555, "learning_rate": 4.580997206022989e-06, "loss": 0.9695, "step": 30550 }, { "epoch": 0.5558485967369359, "grad_norm": 1.7429736852645874, "learning_rate": 4.579122053666861e-06, "loss": 0.8794, "step": 30560 }, { "epoch": 0.5560304843667582, "grad_norm": 2.035836696624756, "learning_rate": 4.577246901310732e-06, "loss": 0.9432, "step": 30570 }, { "epoch": 0.5562123719965805, "grad_norm": 1.6249909400939941, "learning_rate": 4.575371748954603e-06, "loss": 0.9817, "step": 30580 }, { "epoch": 0.5563942596264028, "grad_norm": 1.5527379512786865, "learning_rate": 4.573496596598474e-06, "loss": 0.8979, "step": 30590 }, { "epoch": 0.5565761472562251, "grad_norm": 1.7922154664993286, "learning_rate": 4.571621444242345e-06, "loss": 0.9391, "step": 30600 }, { "epoch": 0.5567580348860474, "grad_norm": 1.676177978515625, "learning_rate": 4.569746291886216e-06, "loss": 0.8832, "step": 30610 }, { "epoch": 0.5569399225158697, "grad_norm": 1.8242237567901611, "learning_rate": 4.567871139530087e-06, "loss": 0.86, "step": 30620 }, { "epoch": 0.557121810145692, "grad_norm": 1.3871036767959595, "learning_rate": 4.5659959871739585e-06, "loss": 0.913, "step": 30630 }, { "epoch": 0.5573036977755143, "grad_norm": 1.685050368309021, "learning_rate": 4.56412083481783e-06, "loss": 0.931, "step": 30640 }, { "epoch": 0.5574855854053365, "grad_norm": 1.8825631141662598, "learning_rate": 4.5622456824617e-06, "loss": 0.9164, "step": 30650 }, { "epoch": 0.5576674730351588, "grad_norm": 1.9449098110198975, "learning_rate": 4.560370530105571e-06, "loss": 0.8962, "step": 30660 }, { "epoch": 0.5578493606649811, "grad_norm": 1.978756308555603, "learning_rate": 4.558495377749442e-06, "loss": 0.9303, "step": 30670 }, { "epoch": 0.5580312482948034, "grad_norm": 1.659315586090088, "learning_rate": 4.5566202253933135e-06, "loss": 0.9357, "step": 30680 }, { "epoch": 0.5582131359246257, "grad_norm": 1.5804767608642578, "learning_rate": 4.554745073037185e-06, "loss": 0.9096, "step": 30690 }, { "epoch": 0.558395023554448, "grad_norm": 1.3145716190338135, "learning_rate": 4.552869920681055e-06, "loss": 0.9217, "step": 30700 }, { "epoch": 0.5585769111842703, "grad_norm": 1.9159820079803467, "learning_rate": 4.550994768324927e-06, "loss": 0.8969, "step": 30710 }, { "epoch": 0.5587587988140926, "grad_norm": 1.6911141872406006, "learning_rate": 4.549119615968798e-06, "loss": 0.898, "step": 30720 }, { "epoch": 0.5589406864439149, "grad_norm": 1.9616305828094482, "learning_rate": 4.5472444636126686e-06, "loss": 0.9316, "step": 30730 }, { "epoch": 0.5591225740737372, "grad_norm": 1.8773064613342285, "learning_rate": 4.54536931125654e-06, "loss": 0.9259, "step": 30740 }, { "epoch": 0.5593044617035595, "grad_norm": 1.9241373538970947, "learning_rate": 4.543494158900411e-06, "loss": 0.9334, "step": 30750 }, { "epoch": 0.5594863493333818, "grad_norm": 1.9375174045562744, "learning_rate": 4.541619006544283e-06, "loss": 0.9207, "step": 30760 }, { "epoch": 0.5596682369632041, "grad_norm": 1.903158187866211, "learning_rate": 4.539743854188153e-06, "loss": 0.9282, "step": 30770 }, { "epoch": 0.5598501245930264, "grad_norm": 1.6123579740524292, "learning_rate": 4.5378687018320244e-06, "loss": 0.952, "step": 30780 }, { "epoch": 0.5600320122228487, "grad_norm": 1.759907841682434, "learning_rate": 4.535993549475896e-06, "loss": 0.9521, "step": 30790 }, { "epoch": 0.560213899852671, "grad_norm": 1.2514804601669312, "learning_rate": 4.534118397119766e-06, "loss": 0.9, "step": 30800 }, { "epoch": 0.5603957874824933, "grad_norm": 1.7788814306259155, "learning_rate": 4.532243244763637e-06, "loss": 0.9568, "step": 30810 }, { "epoch": 0.5605776751123156, "grad_norm": 1.6959012746810913, "learning_rate": 4.530368092407508e-06, "loss": 0.9116, "step": 30820 }, { "epoch": 0.5607595627421379, "grad_norm": 1.6073685884475708, "learning_rate": 4.5284929400513795e-06, "loss": 0.9321, "step": 30830 }, { "epoch": 0.5609414503719602, "grad_norm": 1.869703769683838, "learning_rate": 4.526617787695251e-06, "loss": 0.9602, "step": 30840 }, { "epoch": 0.5611233380017825, "grad_norm": 1.6927474737167358, "learning_rate": 4.524742635339121e-06, "loss": 0.914, "step": 30850 }, { "epoch": 0.5613052256316048, "grad_norm": 1.4882655143737793, "learning_rate": 4.522867482982993e-06, "loss": 0.9185, "step": 30860 }, { "epoch": 0.5614871132614271, "grad_norm": 2.19917631149292, "learning_rate": 4.520992330626864e-06, "loss": 0.9475, "step": 30870 }, { "epoch": 0.5616690008912494, "grad_norm": 1.5046412944793701, "learning_rate": 4.5191171782707345e-06, "loss": 0.8479, "step": 30880 }, { "epoch": 0.5618508885210717, "grad_norm": 1.8980952501296997, "learning_rate": 4.517242025914606e-06, "loss": 0.9642, "step": 30890 }, { "epoch": 0.562032776150894, "grad_norm": 1.54112708568573, "learning_rate": 4.515366873558477e-06, "loss": 0.8485, "step": 30900 }, { "epoch": 0.5622146637807163, "grad_norm": 1.8419818878173828, "learning_rate": 4.5134917212023485e-06, "loss": 0.9299, "step": 30910 }, { "epoch": 0.5623965514105386, "grad_norm": 1.7992204427719116, "learning_rate": 4.511616568846219e-06, "loss": 0.8812, "step": 30920 }, { "epoch": 0.5625784390403609, "grad_norm": 2.0396971702575684, "learning_rate": 4.50974141649009e-06, "loss": 0.9516, "step": 30930 }, { "epoch": 0.5627603266701832, "grad_norm": 1.7274892330169678, "learning_rate": 4.507866264133961e-06, "loss": 0.9096, "step": 30940 }, { "epoch": 0.5629422143000055, "grad_norm": 1.757537603378296, "learning_rate": 4.505991111777832e-06, "loss": 0.9386, "step": 30950 }, { "epoch": 0.5631241019298278, "grad_norm": 1.7126554250717163, "learning_rate": 4.504115959421704e-06, "loss": 0.9435, "step": 30960 }, { "epoch": 0.56330598955965, "grad_norm": 1.579878807067871, "learning_rate": 4.502240807065574e-06, "loss": 0.9014, "step": 30970 }, { "epoch": 0.5634878771894724, "grad_norm": 1.9369961023330688, "learning_rate": 4.5003656547094454e-06, "loss": 0.948, "step": 30980 }, { "epoch": 0.5636697648192946, "grad_norm": 1.4946503639221191, "learning_rate": 4.498490502353317e-06, "loss": 0.893, "step": 30990 }, { "epoch": 0.563851652449117, "grad_norm": 1.8246204853057861, "learning_rate": 4.496615349997187e-06, "loss": 0.9295, "step": 31000 }, { "epoch": 0.5640335400789392, "grad_norm": 1.9484518766403198, "learning_rate": 4.494740197641059e-06, "loss": 0.957, "step": 31010 }, { "epoch": 0.5642154277087615, "grad_norm": 1.8797485828399658, "learning_rate": 4.49286504528493e-06, "loss": 0.9632, "step": 31020 }, { "epoch": 0.5643973153385838, "grad_norm": 1.7848237752914429, "learning_rate": 4.490989892928801e-06, "loss": 0.9363, "step": 31030 }, { "epoch": 0.5645792029684061, "grad_norm": 1.9805647134780884, "learning_rate": 4.489114740572672e-06, "loss": 0.9472, "step": 31040 }, { "epoch": 0.5647610905982284, "grad_norm": 1.9926567077636719, "learning_rate": 4.487239588216543e-06, "loss": 0.8983, "step": 31050 }, { "epoch": 0.5649429782280507, "grad_norm": 1.443089485168457, "learning_rate": 4.4853644358604145e-06, "loss": 0.9067, "step": 31060 }, { "epoch": 0.565124865857873, "grad_norm": 1.6422239542007446, "learning_rate": 4.483489283504285e-06, "loss": 0.8882, "step": 31070 }, { "epoch": 0.5653067534876953, "grad_norm": 1.8046696186065674, "learning_rate": 4.4816141311481555e-06, "loss": 0.9612, "step": 31080 }, { "epoch": 0.5654886411175176, "grad_norm": 1.7526752948760986, "learning_rate": 4.479738978792027e-06, "loss": 0.9308, "step": 31090 }, { "epoch": 0.5656705287473399, "grad_norm": 1.7753565311431885, "learning_rate": 4.477863826435898e-06, "loss": 0.946, "step": 31100 }, { "epoch": 0.5658524163771622, "grad_norm": 1.4720994234085083, "learning_rate": 4.4759886740797695e-06, "loss": 0.8746, "step": 31110 }, { "epoch": 0.5660343040069845, "grad_norm": 1.7843143939971924, "learning_rate": 4.47411352172364e-06, "loss": 0.9379, "step": 31120 }, { "epoch": 0.5662161916368068, "grad_norm": 1.9427281618118286, "learning_rate": 4.472238369367511e-06, "loss": 0.916, "step": 31130 }, { "epoch": 0.5663980792666291, "grad_norm": 1.8097238540649414, "learning_rate": 4.470363217011383e-06, "loss": 0.9415, "step": 31140 }, { "epoch": 0.5665799668964514, "grad_norm": 2.001749038696289, "learning_rate": 4.468488064655253e-06, "loss": 0.9107, "step": 31150 }, { "epoch": 0.5667618545262737, "grad_norm": 1.5637723207473755, "learning_rate": 4.466612912299125e-06, "loss": 0.8666, "step": 31160 }, { "epoch": 0.566943742156096, "grad_norm": 1.7936627864837646, "learning_rate": 4.464737759942996e-06, "loss": 0.9443, "step": 31170 }, { "epoch": 0.5671256297859183, "grad_norm": 1.8238015174865723, "learning_rate": 4.462862607586867e-06, "loss": 0.9339, "step": 31180 }, { "epoch": 0.5673075174157406, "grad_norm": 1.4359797239303589, "learning_rate": 4.460987455230738e-06, "loss": 0.8707, "step": 31190 }, { "epoch": 0.5674894050455629, "grad_norm": 1.515069842338562, "learning_rate": 4.459112302874609e-06, "loss": 0.9748, "step": 31200 }, { "epoch": 0.5676712926753852, "grad_norm": 1.820872187614441, "learning_rate": 4.45723715051848e-06, "loss": 0.9102, "step": 31210 }, { "epoch": 0.5678531803052075, "grad_norm": 2.0222394466400146, "learning_rate": 4.455361998162351e-06, "loss": 1.0162, "step": 31220 }, { "epoch": 0.5680350679350298, "grad_norm": 1.9202109575271606, "learning_rate": 4.453486845806222e-06, "loss": 0.9451, "step": 31230 }, { "epoch": 0.5682169555648521, "grad_norm": 1.848773717880249, "learning_rate": 4.451611693450093e-06, "loss": 0.932, "step": 31240 }, { "epoch": 0.5683988431946744, "grad_norm": 1.5891112089157104, "learning_rate": 4.449736541093964e-06, "loss": 0.9456, "step": 31250 }, { "epoch": 0.5685807308244967, "grad_norm": 1.7102514505386353, "learning_rate": 4.4478613887378355e-06, "loss": 0.9193, "step": 31260 }, { "epoch": 0.568762618454319, "grad_norm": 1.7325708866119385, "learning_rate": 4.445986236381706e-06, "loss": 0.9526, "step": 31270 }, { "epoch": 0.5689445060841413, "grad_norm": 1.5652101039886475, "learning_rate": 4.444111084025577e-06, "loss": 0.9548, "step": 31280 }, { "epoch": 0.5691263937139636, "grad_norm": 1.7290343046188354, "learning_rate": 4.442235931669449e-06, "loss": 0.9459, "step": 31290 }, { "epoch": 0.5693082813437859, "grad_norm": 1.991104006767273, "learning_rate": 4.44036077931332e-06, "loss": 0.9002, "step": 31300 }, { "epoch": 0.5694901689736082, "grad_norm": 1.8551522493362427, "learning_rate": 4.4384856269571905e-06, "loss": 0.9605, "step": 31310 }, { "epoch": 0.5696720566034305, "grad_norm": 1.8179130554199219, "learning_rate": 4.436610474601062e-06, "loss": 0.8975, "step": 31320 }, { "epoch": 0.5698539442332528, "grad_norm": 1.7932413816452026, "learning_rate": 4.434735322244933e-06, "loss": 0.9671, "step": 31330 }, { "epoch": 0.5700358318630749, "grad_norm": 2.1218950748443604, "learning_rate": 4.432860169888804e-06, "loss": 0.9344, "step": 31340 }, { "epoch": 0.5702177194928972, "grad_norm": 1.7737703323364258, "learning_rate": 4.430985017532674e-06, "loss": 0.9758, "step": 31350 }, { "epoch": 0.5703996071227195, "grad_norm": 1.7583048343658447, "learning_rate": 4.4291098651765456e-06, "loss": 0.9219, "step": 31360 }, { "epoch": 0.5705814947525418, "grad_norm": 1.5862075090408325, "learning_rate": 4.427234712820417e-06, "loss": 0.879, "step": 31370 }, { "epoch": 0.5707633823823641, "grad_norm": 2.1365504264831543, "learning_rate": 4.425359560464288e-06, "loss": 0.9634, "step": 31380 }, { "epoch": 0.5709452700121864, "grad_norm": 1.8059425354003906, "learning_rate": 4.423484408108159e-06, "loss": 0.9145, "step": 31390 }, { "epoch": 0.5711271576420087, "grad_norm": 2.255171537399292, "learning_rate": 4.42160925575203e-06, "loss": 0.8902, "step": 31400 }, { "epoch": 0.571309045271831, "grad_norm": 1.4382256269454956, "learning_rate": 4.4197341033959015e-06, "loss": 0.8819, "step": 31410 }, { "epoch": 0.5714909329016533, "grad_norm": 1.8230832815170288, "learning_rate": 4.417858951039772e-06, "loss": 0.9151, "step": 31420 }, { "epoch": 0.5716728205314756, "grad_norm": 1.7198976278305054, "learning_rate": 4.415983798683643e-06, "loss": 0.8615, "step": 31430 }, { "epoch": 0.5718547081612979, "grad_norm": 1.7542462348937988, "learning_rate": 4.414108646327515e-06, "loss": 0.9084, "step": 31440 }, { "epoch": 0.5720365957911202, "grad_norm": 1.6572433710098267, "learning_rate": 4.412233493971386e-06, "loss": 0.9765, "step": 31450 }, { "epoch": 0.5722184834209425, "grad_norm": 1.685448169708252, "learning_rate": 4.4103583416152565e-06, "loss": 0.904, "step": 31460 }, { "epoch": 0.5724003710507648, "grad_norm": 1.9538639783859253, "learning_rate": 4.408483189259128e-06, "loss": 0.967, "step": 31470 }, { "epoch": 0.5725822586805871, "grad_norm": 1.6254172325134277, "learning_rate": 4.406608036902999e-06, "loss": 0.8964, "step": 31480 }, { "epoch": 0.5727641463104094, "grad_norm": 1.362546443939209, "learning_rate": 4.40473288454687e-06, "loss": 0.9496, "step": 31490 }, { "epoch": 0.5729460339402317, "grad_norm": 1.7110003232955933, "learning_rate": 4.402857732190741e-06, "loss": 0.9567, "step": 31500 }, { "epoch": 0.573127921570054, "grad_norm": 1.5600433349609375, "learning_rate": 4.4009825798346115e-06, "loss": 0.8707, "step": 31510 }, { "epoch": 0.5733098091998763, "grad_norm": 1.6994414329528809, "learning_rate": 4.399107427478483e-06, "loss": 0.9317, "step": 31520 }, { "epoch": 0.5734916968296986, "grad_norm": 1.9185165166854858, "learning_rate": 4.397232275122354e-06, "loss": 0.9102, "step": 31530 }, { "epoch": 0.5736735844595209, "grad_norm": 1.722122073173523, "learning_rate": 4.395357122766225e-06, "loss": 0.9617, "step": 31540 }, { "epoch": 0.5738554720893432, "grad_norm": 1.7180355787277222, "learning_rate": 4.393481970410096e-06, "loss": 0.8582, "step": 31550 }, { "epoch": 0.5740373597191655, "grad_norm": 1.6962560415267944, "learning_rate": 4.391606818053967e-06, "loss": 0.9596, "step": 31560 }, { "epoch": 0.5742192473489878, "grad_norm": 1.8084471225738525, "learning_rate": 4.389731665697839e-06, "loss": 0.9352, "step": 31570 }, { "epoch": 0.5744011349788101, "grad_norm": 1.7610275745391846, "learning_rate": 4.387856513341709e-06, "loss": 0.9628, "step": 31580 }, { "epoch": 0.5745830226086324, "grad_norm": 1.8994617462158203, "learning_rate": 4.385981360985581e-06, "loss": 0.9617, "step": 31590 }, { "epoch": 0.5747649102384547, "grad_norm": 1.7015775442123413, "learning_rate": 4.384106208629452e-06, "loss": 0.9435, "step": 31600 }, { "epoch": 0.574946797868277, "grad_norm": 1.754062533378601, "learning_rate": 4.3822310562733224e-06, "loss": 0.938, "step": 31610 }, { "epoch": 0.5751286854980993, "grad_norm": 1.7728357315063477, "learning_rate": 4.380355903917194e-06, "loss": 0.9467, "step": 31620 }, { "epoch": 0.5753105731279216, "grad_norm": 1.876059651374817, "learning_rate": 4.378480751561064e-06, "loss": 0.978, "step": 31630 }, { "epoch": 0.5754924607577439, "grad_norm": 1.727156162261963, "learning_rate": 4.376605599204936e-06, "loss": 0.9712, "step": 31640 }, { "epoch": 0.5756743483875661, "grad_norm": 1.8606857061386108, "learning_rate": 4.374730446848807e-06, "loss": 0.9266, "step": 31650 }, { "epoch": 0.5758562360173884, "grad_norm": 1.5963188409805298, "learning_rate": 4.3728552944926775e-06, "loss": 0.9485, "step": 31660 }, { "epoch": 0.5760381236472107, "grad_norm": 1.9372062683105469, "learning_rate": 4.370980142136549e-06, "loss": 0.927, "step": 31670 }, { "epoch": 0.576220011277033, "grad_norm": 1.7517775297164917, "learning_rate": 4.36910498978042e-06, "loss": 0.9077, "step": 31680 }, { "epoch": 0.5764018989068553, "grad_norm": 1.6144378185272217, "learning_rate": 4.367229837424291e-06, "loss": 0.9481, "step": 31690 }, { "epoch": 0.5765837865366776, "grad_norm": 1.6868209838867188, "learning_rate": 4.365354685068162e-06, "loss": 0.9069, "step": 31700 }, { "epoch": 0.5767656741664999, "grad_norm": 1.5446077585220337, "learning_rate": 4.363479532712033e-06, "loss": 0.9404, "step": 31710 }, { "epoch": 0.5769475617963222, "grad_norm": 1.6294294595718384, "learning_rate": 4.361604380355905e-06, "loss": 0.8993, "step": 31720 }, { "epoch": 0.5771294494261445, "grad_norm": 1.837283730506897, "learning_rate": 4.359729227999775e-06, "loss": 0.9148, "step": 31730 }, { "epoch": 0.5773113370559668, "grad_norm": 1.625562071800232, "learning_rate": 4.3578540756436466e-06, "loss": 0.9409, "step": 31740 }, { "epoch": 0.5774932246857891, "grad_norm": 1.8818601369857788, "learning_rate": 4.355978923287518e-06, "loss": 0.9573, "step": 31750 }, { "epoch": 0.5776751123156114, "grad_norm": 1.5567842721939087, "learning_rate": 4.354103770931388e-06, "loss": 0.9068, "step": 31760 }, { "epoch": 0.5778569999454337, "grad_norm": 2.000521659851074, "learning_rate": 4.35222861857526e-06, "loss": 0.9347, "step": 31770 }, { "epoch": 0.578038887575256, "grad_norm": 1.775222659111023, "learning_rate": 4.35035346621913e-06, "loss": 0.9136, "step": 31780 }, { "epoch": 0.5782207752050783, "grad_norm": 1.846109390258789, "learning_rate": 4.348478313863002e-06, "loss": 0.9141, "step": 31790 }, { "epoch": 0.5784026628349006, "grad_norm": 1.9380476474761963, "learning_rate": 4.346603161506873e-06, "loss": 0.9416, "step": 31800 }, { "epoch": 0.5785845504647229, "grad_norm": 2.134873390197754, "learning_rate": 4.3447280091507434e-06, "loss": 0.9165, "step": 31810 }, { "epoch": 0.5787664380945452, "grad_norm": 1.9128013849258423, "learning_rate": 4.342852856794615e-06, "loss": 0.9429, "step": 31820 }, { "epoch": 0.5789483257243675, "grad_norm": 2.237440347671509, "learning_rate": 4.340977704438486e-06, "loss": 0.9671, "step": 31830 }, { "epoch": 0.5791302133541898, "grad_norm": 1.7541285753250122, "learning_rate": 4.3391025520823575e-06, "loss": 0.902, "step": 31840 }, { "epoch": 0.5793121009840121, "grad_norm": 1.9243026971817017, "learning_rate": 4.337227399726228e-06, "loss": 0.9686, "step": 31850 }, { "epoch": 0.5794939886138344, "grad_norm": 1.5740324258804321, "learning_rate": 4.335352247370099e-06, "loss": 0.8912, "step": 31860 }, { "epoch": 0.5796758762436567, "grad_norm": 1.7372163534164429, "learning_rate": 4.333477095013971e-06, "loss": 0.8537, "step": 31870 }, { "epoch": 0.579857763873479, "grad_norm": 1.7979557514190674, "learning_rate": 4.331601942657841e-06, "loss": 0.9447, "step": 31880 }, { "epoch": 0.5800396515033013, "grad_norm": 1.8965040445327759, "learning_rate": 4.3297267903017125e-06, "loss": 0.9268, "step": 31890 }, { "epoch": 0.5802215391331236, "grad_norm": 2.126237392425537, "learning_rate": 4.327851637945583e-06, "loss": 0.9321, "step": 31900 }, { "epoch": 0.5804034267629459, "grad_norm": 1.8655073642730713, "learning_rate": 4.325976485589454e-06, "loss": 0.9172, "step": 31910 }, { "epoch": 0.5805853143927682, "grad_norm": 1.6200999021530151, "learning_rate": 4.324101333233326e-06, "loss": 0.9616, "step": 31920 }, { "epoch": 0.5807672020225905, "grad_norm": 1.5737221240997314, "learning_rate": 4.322226180877196e-06, "loss": 0.904, "step": 31930 }, { "epoch": 0.5809490896524128, "grad_norm": 1.8981579542160034, "learning_rate": 4.3203510285210675e-06, "loss": 0.9617, "step": 31940 }, { "epoch": 0.5811309772822351, "grad_norm": 1.7035857439041138, "learning_rate": 4.318475876164939e-06, "loss": 0.9997, "step": 31950 }, { "epoch": 0.5813128649120574, "grad_norm": 1.767116904258728, "learning_rate": 4.316600723808809e-06, "loss": 0.8808, "step": 31960 }, { "epoch": 0.5814947525418797, "grad_norm": 1.4543989896774292, "learning_rate": 4.314725571452681e-06, "loss": 0.9175, "step": 31970 }, { "epoch": 0.581676640171702, "grad_norm": 1.859422206878662, "learning_rate": 4.312850419096552e-06, "loss": 0.9311, "step": 31980 }, { "epoch": 0.5818585278015243, "grad_norm": 1.9386622905731201, "learning_rate": 4.3109752667404234e-06, "loss": 0.9125, "step": 31990 }, { "epoch": 0.5820404154313465, "grad_norm": 1.8602176904678345, "learning_rate": 4.309100114384294e-06, "loss": 0.8924, "step": 32000 }, { "epoch": 0.5822223030611688, "grad_norm": 1.6976393461227417, "learning_rate": 4.307224962028165e-06, "loss": 0.9395, "step": 32010 }, { "epoch": 0.5824041906909911, "grad_norm": 3.1305367946624756, "learning_rate": 4.305349809672037e-06, "loss": 0.932, "step": 32020 }, { "epoch": 0.5825860783208134, "grad_norm": 1.3420015573501587, "learning_rate": 4.303474657315907e-06, "loss": 0.8864, "step": 32030 }, { "epoch": 0.5827679659506357, "grad_norm": 1.7448850870132446, "learning_rate": 4.3015995049597785e-06, "loss": 0.9141, "step": 32040 }, { "epoch": 0.582949853580458, "grad_norm": 1.9569635391235352, "learning_rate": 4.299724352603649e-06, "loss": 0.999, "step": 32050 }, { "epoch": 0.5831317412102803, "grad_norm": 1.8964941501617432, "learning_rate": 4.29784920024752e-06, "loss": 0.8681, "step": 32060 }, { "epoch": 0.5833136288401026, "grad_norm": 1.9122952222824097, "learning_rate": 4.295974047891392e-06, "loss": 0.9335, "step": 32070 }, { "epoch": 0.5834955164699249, "grad_norm": 1.8947848081588745, "learning_rate": 4.294098895535262e-06, "loss": 0.8933, "step": 32080 }, { "epoch": 0.5836774040997472, "grad_norm": 2.161971092224121, "learning_rate": 4.2922237431791335e-06, "loss": 0.9307, "step": 32090 }, { "epoch": 0.5838592917295695, "grad_norm": 1.7527996301651, "learning_rate": 4.290348590823005e-06, "loss": 0.8911, "step": 32100 }, { "epoch": 0.5840411793593918, "grad_norm": 1.7986754179000854, "learning_rate": 4.288473438466876e-06, "loss": 0.9219, "step": 32110 }, { "epoch": 0.5842230669892141, "grad_norm": 1.4105918407440186, "learning_rate": 4.286598286110747e-06, "loss": 0.8914, "step": 32120 }, { "epoch": 0.5844049546190364, "grad_norm": 1.7740788459777832, "learning_rate": 4.284723133754618e-06, "loss": 0.9161, "step": 32130 }, { "epoch": 0.5845868422488587, "grad_norm": 2.212419271469116, "learning_rate": 4.282847981398489e-06, "loss": 0.9574, "step": 32140 }, { "epoch": 0.584768729878681, "grad_norm": 2.0142276287078857, "learning_rate": 4.28097282904236e-06, "loss": 0.9367, "step": 32150 }, { "epoch": 0.5849506175085033, "grad_norm": 1.7376099824905396, "learning_rate": 4.279097676686231e-06, "loss": 0.9357, "step": 32160 }, { "epoch": 0.5851325051383255, "grad_norm": 1.7898401021957397, "learning_rate": 4.2772225243301026e-06, "loss": 0.932, "step": 32170 }, { "epoch": 0.5853143927681478, "grad_norm": 1.9171831607818604, "learning_rate": 4.275347371973973e-06, "loss": 0.9262, "step": 32180 }, { "epoch": 0.5854962803979701, "grad_norm": 1.7057424783706665, "learning_rate": 4.273472219617844e-06, "loss": 0.9259, "step": 32190 }, { "epoch": 0.5856781680277924, "grad_norm": 1.7356023788452148, "learning_rate": 4.271597067261715e-06, "loss": 0.9352, "step": 32200 }, { "epoch": 0.5858600556576147, "grad_norm": 1.9151338338851929, "learning_rate": 4.269721914905586e-06, "loss": 0.9344, "step": 32210 }, { "epoch": 0.586041943287437, "grad_norm": 1.5106738805770874, "learning_rate": 4.267846762549458e-06, "loss": 0.8774, "step": 32220 }, { "epoch": 0.5862238309172593, "grad_norm": 1.891909122467041, "learning_rate": 4.265971610193328e-06, "loss": 0.9148, "step": 32230 }, { "epoch": 0.5864057185470816, "grad_norm": 1.8983814716339111, "learning_rate": 4.2640964578371995e-06, "loss": 0.9264, "step": 32240 }, { "epoch": 0.5865876061769039, "grad_norm": 1.7322580814361572, "learning_rate": 4.262221305481071e-06, "loss": 0.929, "step": 32250 }, { "epoch": 0.5867694938067262, "grad_norm": 1.9573256969451904, "learning_rate": 4.260346153124942e-06, "loss": 0.9694, "step": 32260 }, { "epoch": 0.5869513814365485, "grad_norm": 1.7728848457336426, "learning_rate": 4.258471000768813e-06, "loss": 0.8989, "step": 32270 }, { "epoch": 0.5871332690663708, "grad_norm": 1.8944664001464844, "learning_rate": 4.256595848412684e-06, "loss": 0.9562, "step": 32280 }, { "epoch": 0.587315156696193, "grad_norm": 1.789947509765625, "learning_rate": 4.254720696056555e-06, "loss": 0.9544, "step": 32290 }, { "epoch": 0.5874970443260153, "grad_norm": 1.9856559038162231, "learning_rate": 4.252845543700426e-06, "loss": 0.9583, "step": 32300 }, { "epoch": 0.5876789319558376, "grad_norm": 1.784870982170105, "learning_rate": 4.250970391344297e-06, "loss": 0.9397, "step": 32310 }, { "epoch": 0.5878608195856599, "grad_norm": 1.7352540493011475, "learning_rate": 4.249095238988168e-06, "loss": 0.9677, "step": 32320 }, { "epoch": 0.5880427072154822, "grad_norm": 1.7433189153671265, "learning_rate": 4.247220086632039e-06, "loss": 0.8849, "step": 32330 }, { "epoch": 0.5882245948453045, "grad_norm": 1.9549473524093628, "learning_rate": 4.24534493427591e-06, "loss": 0.9242, "step": 32340 }, { "epoch": 0.5884064824751268, "grad_norm": 1.8258134126663208, "learning_rate": 4.243469781919781e-06, "loss": 0.8984, "step": 32350 }, { "epoch": 0.5885883701049491, "grad_norm": 1.5680292844772339, "learning_rate": 4.241594629563652e-06, "loss": 0.9302, "step": 32360 }, { "epoch": 0.5887702577347714, "grad_norm": 2.288522958755493, "learning_rate": 4.2397194772075236e-06, "loss": 0.9923, "step": 32370 }, { "epoch": 0.5889521453645937, "grad_norm": 1.4036264419555664, "learning_rate": 4.237844324851395e-06, "loss": 0.8868, "step": 32380 }, { "epoch": 0.589134032994416, "grad_norm": 1.8607784509658813, "learning_rate": 4.235969172495265e-06, "loss": 0.9595, "step": 32390 }, { "epoch": 0.5893159206242383, "grad_norm": 1.6006252765655518, "learning_rate": 4.234094020139137e-06, "loss": 0.8987, "step": 32400 }, { "epoch": 0.5894978082540606, "grad_norm": 1.590961217880249, "learning_rate": 4.232218867783008e-06, "loss": 0.9218, "step": 32410 }, { "epoch": 0.5896796958838829, "grad_norm": 1.7900917530059814, "learning_rate": 4.230343715426879e-06, "loss": 0.9005, "step": 32420 }, { "epoch": 0.5898615835137052, "grad_norm": 3.171160936355591, "learning_rate": 4.22846856307075e-06, "loss": 0.8727, "step": 32430 }, { "epoch": 0.5900434711435275, "grad_norm": 1.7931150197982788, "learning_rate": 4.226593410714621e-06, "loss": 0.9281, "step": 32440 }, { "epoch": 0.5902253587733498, "grad_norm": 2.0892159938812256, "learning_rate": 4.224718258358492e-06, "loss": 0.8764, "step": 32450 }, { "epoch": 0.5904072464031721, "grad_norm": 1.7574422359466553, "learning_rate": 4.222843106002363e-06, "loss": 0.9667, "step": 32460 }, { "epoch": 0.5905891340329944, "grad_norm": 1.6149563789367676, "learning_rate": 4.220967953646234e-06, "loss": 0.9338, "step": 32470 }, { "epoch": 0.5907710216628167, "grad_norm": 2.0389606952667236, "learning_rate": 4.219092801290105e-06, "loss": 0.9061, "step": 32480 }, { "epoch": 0.590952909292639, "grad_norm": 1.61555814743042, "learning_rate": 4.217217648933976e-06, "loss": 0.8959, "step": 32490 }, { "epoch": 0.5911347969224613, "grad_norm": 1.886434555053711, "learning_rate": 4.215342496577847e-06, "loss": 0.9317, "step": 32500 }, { "epoch": 0.5913166845522836, "grad_norm": 2.0425941944122314, "learning_rate": 4.213467344221718e-06, "loss": 0.9987, "step": 32510 }, { "epoch": 0.5914985721821059, "grad_norm": 1.8700565099716187, "learning_rate": 4.2115921918655895e-06, "loss": 0.942, "step": 32520 }, { "epoch": 0.5916804598119282, "grad_norm": 1.7127858400344849, "learning_rate": 4.209717039509461e-06, "loss": 0.9127, "step": 32530 }, { "epoch": 0.5918623474417505, "grad_norm": 1.6361597776412964, "learning_rate": 4.207841887153331e-06, "loss": 0.9197, "step": 32540 }, { "epoch": 0.5920442350715728, "grad_norm": 1.6431283950805664, "learning_rate": 4.205966734797203e-06, "loss": 0.95, "step": 32550 }, { "epoch": 0.5922261227013951, "grad_norm": 1.9848246574401855, "learning_rate": 4.204091582441074e-06, "loss": 0.9746, "step": 32560 }, { "epoch": 0.5924080103312174, "grad_norm": 1.8002452850341797, "learning_rate": 4.2022164300849446e-06, "loss": 0.8775, "step": 32570 }, { "epoch": 0.5925898979610397, "grad_norm": 1.6153194904327393, "learning_rate": 4.200341277728816e-06, "loss": 0.9404, "step": 32580 }, { "epoch": 0.592771785590862, "grad_norm": 1.9195470809936523, "learning_rate": 4.198466125372686e-06, "loss": 0.9268, "step": 32590 }, { "epoch": 0.5929536732206843, "grad_norm": 1.8937604427337646, "learning_rate": 4.196590973016558e-06, "loss": 0.9357, "step": 32600 }, { "epoch": 0.5931355608505066, "grad_norm": 1.7308577299118042, "learning_rate": 4.194715820660429e-06, "loss": 0.9139, "step": 32610 }, { "epoch": 0.5933174484803289, "grad_norm": 1.7044434547424316, "learning_rate": 4.1928406683043e-06, "loss": 0.8785, "step": 32620 }, { "epoch": 0.5934993361101512, "grad_norm": 1.763940453529358, "learning_rate": 4.190965515948171e-06, "loss": 0.9027, "step": 32630 }, { "epoch": 0.5936812237399735, "grad_norm": 1.7217826843261719, "learning_rate": 4.189090363592042e-06, "loss": 0.9472, "step": 32640 }, { "epoch": 0.5938631113697957, "grad_norm": 1.8365895748138428, "learning_rate": 4.187215211235914e-06, "loss": 0.9606, "step": 32650 }, { "epoch": 0.594044998999618, "grad_norm": 2.218208074569702, "learning_rate": 4.185340058879784e-06, "loss": 0.9534, "step": 32660 }, { "epoch": 0.5942268866294403, "grad_norm": 1.9316097497940063, "learning_rate": 4.1834649065236555e-06, "loss": 0.9067, "step": 32670 }, { "epoch": 0.5944087742592626, "grad_norm": 2.1024065017700195, "learning_rate": 4.181589754167527e-06, "loss": 0.9554, "step": 32680 }, { "epoch": 0.5945906618890849, "grad_norm": 1.7858511209487915, "learning_rate": 4.179714601811397e-06, "loss": 0.911, "step": 32690 }, { "epoch": 0.5947725495189072, "grad_norm": 1.9187378883361816, "learning_rate": 4.177839449455269e-06, "loss": 0.9575, "step": 32700 }, { "epoch": 0.5949544371487295, "grad_norm": 1.863858938217163, "learning_rate": 4.17596429709914e-06, "loss": 0.9132, "step": 32710 }, { "epoch": 0.5951363247785518, "grad_norm": 1.867635726928711, "learning_rate": 4.174089144743011e-06, "loss": 0.9502, "step": 32720 }, { "epoch": 0.5953182124083741, "grad_norm": 1.7699298858642578, "learning_rate": 4.172213992386882e-06, "loss": 0.8841, "step": 32730 }, { "epoch": 0.5955001000381964, "grad_norm": 1.7802077531814575, "learning_rate": 4.170338840030752e-06, "loss": 0.9407, "step": 32740 }, { "epoch": 0.5956819876680187, "grad_norm": 1.4759904146194458, "learning_rate": 4.168463687674624e-06, "loss": 0.8836, "step": 32750 }, { "epoch": 0.595863875297841, "grad_norm": 1.6442184448242188, "learning_rate": 4.166588535318495e-06, "loss": 0.9099, "step": 32760 }, { "epoch": 0.5960457629276633, "grad_norm": 1.9345614910125732, "learning_rate": 4.1647133829623655e-06, "loss": 0.9325, "step": 32770 }, { "epoch": 0.5962276505574856, "grad_norm": 1.8315421342849731, "learning_rate": 4.162838230606237e-06, "loss": 0.9282, "step": 32780 }, { "epoch": 0.5964095381873079, "grad_norm": 1.6720154285430908, "learning_rate": 4.160963078250108e-06, "loss": 0.9328, "step": 32790 }, { "epoch": 0.5965914258171302, "grad_norm": 1.8685444593429565, "learning_rate": 4.15908792589398e-06, "loss": 0.941, "step": 32800 }, { "epoch": 0.5967733134469525, "grad_norm": 1.429648995399475, "learning_rate": 4.15721277353785e-06, "loss": 0.9009, "step": 32810 }, { "epoch": 0.5969552010767748, "grad_norm": 1.734580397605896, "learning_rate": 4.1553376211817214e-06, "loss": 0.9249, "step": 32820 }, { "epoch": 0.5971370887065971, "grad_norm": 1.834291934967041, "learning_rate": 4.153462468825593e-06, "loss": 0.8918, "step": 32830 }, { "epoch": 0.5973189763364194, "grad_norm": 1.697350025177002, "learning_rate": 4.151587316469463e-06, "loss": 0.9155, "step": 32840 }, { "epoch": 0.5975008639662417, "grad_norm": 1.6191293001174927, "learning_rate": 4.149712164113335e-06, "loss": 0.944, "step": 32850 }, { "epoch": 0.597682751596064, "grad_norm": 1.8364585638046265, "learning_rate": 4.147837011757206e-06, "loss": 0.9107, "step": 32860 }, { "epoch": 0.5978646392258863, "grad_norm": 1.7132437229156494, "learning_rate": 4.1459618594010765e-06, "loss": 0.9046, "step": 32870 }, { "epoch": 0.5980465268557086, "grad_norm": 1.6637591123580933, "learning_rate": 4.144086707044948e-06, "loss": 0.8733, "step": 32880 }, { "epoch": 0.5982284144855309, "grad_norm": 2.3139266967773438, "learning_rate": 4.142211554688818e-06, "loss": 0.9604, "step": 32890 }, { "epoch": 0.5984103021153532, "grad_norm": 1.8437250852584839, "learning_rate": 4.14033640233269e-06, "loss": 0.9278, "step": 32900 }, { "epoch": 0.5985921897451755, "grad_norm": 1.8004294633865356, "learning_rate": 4.138461249976561e-06, "loss": 0.9573, "step": 32910 }, { "epoch": 0.5987740773749978, "grad_norm": 1.5493444204330444, "learning_rate": 4.136586097620432e-06, "loss": 0.8853, "step": 32920 }, { "epoch": 0.5989559650048201, "grad_norm": 1.7530975341796875, "learning_rate": 4.134710945264303e-06, "loss": 0.9233, "step": 32930 }, { "epoch": 0.5991378526346424, "grad_norm": 1.4887069463729858, "learning_rate": 4.132835792908174e-06, "loss": 0.825, "step": 32940 }, { "epoch": 0.5993197402644647, "grad_norm": 1.8427399396896362, "learning_rate": 4.1309606405520455e-06, "loss": 0.8931, "step": 32950 }, { "epoch": 0.599501627894287, "grad_norm": 1.8213533163070679, "learning_rate": 4.129085488195916e-06, "loss": 0.9404, "step": 32960 }, { "epoch": 0.5996835155241093, "grad_norm": 1.7114263772964478, "learning_rate": 4.127210335839787e-06, "loss": 0.9063, "step": 32970 }, { "epoch": 0.5998654031539316, "grad_norm": 1.8428988456726074, "learning_rate": 4.125335183483659e-06, "loss": 0.922, "step": 32980 }, { "epoch": 0.6000472907837539, "grad_norm": 1.9516984224319458, "learning_rate": 4.12346003112753e-06, "loss": 0.9435, "step": 32990 }, { "epoch": 0.6002291784135761, "grad_norm": 1.708216667175293, "learning_rate": 4.121584878771401e-06, "loss": 0.934, "step": 33000 }, { "epoch": 0.6004110660433983, "grad_norm": 1.7633997201919556, "learning_rate": 4.119709726415271e-06, "loss": 0.9086, "step": 33010 }, { "epoch": 0.6005929536732206, "grad_norm": 1.746015191078186, "learning_rate": 4.1178345740591424e-06, "loss": 0.9211, "step": 33020 }, { "epoch": 0.6007748413030429, "grad_norm": 1.8567267656326294, "learning_rate": 4.115959421703014e-06, "loss": 0.9087, "step": 33030 }, { "epoch": 0.6009567289328652, "grad_norm": 1.8014817237854004, "learning_rate": 4.114084269346884e-06, "loss": 0.9221, "step": 33040 }, { "epoch": 0.6011386165626875, "grad_norm": 1.5375348329544067, "learning_rate": 4.112209116990756e-06, "loss": 0.935, "step": 33050 }, { "epoch": 0.6013205041925098, "grad_norm": 1.814067006111145, "learning_rate": 4.110333964634627e-06, "loss": 0.9493, "step": 33060 }, { "epoch": 0.6015023918223321, "grad_norm": 1.4392801523208618, "learning_rate": 4.108458812278498e-06, "loss": 0.8952, "step": 33070 }, { "epoch": 0.6016842794521544, "grad_norm": 1.931409478187561, "learning_rate": 4.106583659922369e-06, "loss": 0.9151, "step": 33080 }, { "epoch": 0.6018661670819767, "grad_norm": 1.4638195037841797, "learning_rate": 4.10470850756624e-06, "loss": 0.8548, "step": 33090 }, { "epoch": 0.602048054711799, "grad_norm": 1.8952786922454834, "learning_rate": 4.1028333552101115e-06, "loss": 0.9218, "step": 33100 }, { "epoch": 0.6022299423416213, "grad_norm": 1.7525330781936646, "learning_rate": 4.100958202853982e-06, "loss": 0.922, "step": 33110 }, { "epoch": 0.6024118299714436, "grad_norm": 1.7436785697937012, "learning_rate": 4.099083050497853e-06, "loss": 0.9121, "step": 33120 }, { "epoch": 0.6025937176012659, "grad_norm": 2.1517248153686523, "learning_rate": 4.097207898141725e-06, "loss": 0.9349, "step": 33130 }, { "epoch": 0.6027756052310882, "grad_norm": 1.6758590936660767, "learning_rate": 4.095332745785595e-06, "loss": 0.944, "step": 33140 }, { "epoch": 0.6029574928609105, "grad_norm": 1.3388575315475464, "learning_rate": 4.0934575934294665e-06, "loss": 0.9304, "step": 33150 }, { "epoch": 0.6031393804907328, "grad_norm": 2.0148470401763916, "learning_rate": 4.091582441073337e-06, "loss": 0.908, "step": 33160 }, { "epoch": 0.6033212681205551, "grad_norm": 1.9134258031845093, "learning_rate": 4.089707288717208e-06, "loss": 0.9152, "step": 33170 }, { "epoch": 0.6035031557503774, "grad_norm": 1.5400763750076294, "learning_rate": 4.08783213636108e-06, "loss": 0.8737, "step": 33180 }, { "epoch": 0.6036850433801997, "grad_norm": 1.9717614650726318, "learning_rate": 4.085956984004951e-06, "loss": 0.8817, "step": 33190 }, { "epoch": 0.603866931010022, "grad_norm": 1.9369003772735596, "learning_rate": 4.0840818316488216e-06, "loss": 0.9149, "step": 33200 }, { "epoch": 0.6040488186398443, "grad_norm": 2.0410382747650146, "learning_rate": 4.082206679292693e-06, "loss": 0.9415, "step": 33210 }, { "epoch": 0.6042307062696666, "grad_norm": 1.7528213262557983, "learning_rate": 4.080331526936564e-06, "loss": 0.8959, "step": 33220 }, { "epoch": 0.6044125938994889, "grad_norm": 2.0048575401306152, "learning_rate": 4.078456374580435e-06, "loss": 0.9652, "step": 33230 }, { "epoch": 0.6045944815293112, "grad_norm": 2.0665643215179443, "learning_rate": 4.076581222224306e-06, "loss": 0.9804, "step": 33240 }, { "epoch": 0.6047763691591335, "grad_norm": 1.756378173828125, "learning_rate": 4.0747060698681775e-06, "loss": 0.901, "step": 33250 }, { "epoch": 0.6049582567889558, "grad_norm": 1.8725515604019165, "learning_rate": 4.072830917512049e-06, "loss": 0.9172, "step": 33260 }, { "epoch": 0.6051401444187781, "grad_norm": 1.6469242572784424, "learning_rate": 4.070955765155919e-06, "loss": 0.9217, "step": 33270 }, { "epoch": 0.6053220320486004, "grad_norm": 1.8482000827789307, "learning_rate": 4.06908061279979e-06, "loss": 0.9697, "step": 33280 }, { "epoch": 0.6055039196784227, "grad_norm": 1.8334046602249146, "learning_rate": 4.067205460443661e-06, "loss": 0.9129, "step": 33290 }, { "epoch": 0.605685807308245, "grad_norm": 1.7526013851165771, "learning_rate": 4.0653303080875325e-06, "loss": 0.9313, "step": 33300 }, { "epoch": 0.6058676949380672, "grad_norm": 1.6693048477172852, "learning_rate": 4.063455155731403e-06, "loss": 0.8882, "step": 33310 }, { "epoch": 0.6060495825678895, "grad_norm": 1.6963971853256226, "learning_rate": 4.061580003375274e-06, "loss": 0.9251, "step": 33320 }, { "epoch": 0.6062314701977118, "grad_norm": 1.7954249382019043, "learning_rate": 4.059704851019146e-06, "loss": 0.9206, "step": 33330 }, { "epoch": 0.6064133578275341, "grad_norm": 1.9595900774002075, "learning_rate": 4.057829698663017e-06, "loss": 0.964, "step": 33340 }, { "epoch": 0.6065952454573564, "grad_norm": 1.744148850440979, "learning_rate": 4.0559545463068875e-06, "loss": 0.9033, "step": 33350 }, { "epoch": 0.6067771330871787, "grad_norm": 2.0368125438690186, "learning_rate": 4.054079393950759e-06, "loss": 0.9015, "step": 33360 }, { "epoch": 0.606959020717001, "grad_norm": 1.4904348850250244, "learning_rate": 4.05220424159463e-06, "loss": 0.9004, "step": 33370 }, { "epoch": 0.6071409083468233, "grad_norm": 1.9952279329299927, "learning_rate": 4.050329089238501e-06, "loss": 0.9317, "step": 33380 }, { "epoch": 0.6073227959766456, "grad_norm": 2.0680718421936035, "learning_rate": 4.048453936882372e-06, "loss": 0.9746, "step": 33390 }, { "epoch": 0.6075046836064679, "grad_norm": 1.764790654182434, "learning_rate": 4.046578784526243e-06, "loss": 0.9012, "step": 33400 }, { "epoch": 0.6076865712362902, "grad_norm": 1.8025588989257812, "learning_rate": 4.044703632170115e-06, "loss": 0.8638, "step": 33410 }, { "epoch": 0.6078684588661125, "grad_norm": 1.6312255859375, "learning_rate": 4.042828479813985e-06, "loss": 0.9431, "step": 33420 }, { "epoch": 0.6080503464959348, "grad_norm": 1.7149649858474731, "learning_rate": 4.040953327457856e-06, "loss": 0.9064, "step": 33430 }, { "epoch": 0.6082322341257571, "grad_norm": 1.735232949256897, "learning_rate": 4.039078175101727e-06, "loss": 0.9244, "step": 33440 }, { "epoch": 0.6084141217555794, "grad_norm": 1.404343605041504, "learning_rate": 4.0372030227455984e-06, "loss": 0.8917, "step": 33450 }, { "epoch": 0.6085960093854017, "grad_norm": 1.7129203081130981, "learning_rate": 4.03532787038947e-06, "loss": 0.9128, "step": 33460 }, { "epoch": 0.608777897015224, "grad_norm": 1.76793372631073, "learning_rate": 4.03345271803334e-06, "loss": 0.9582, "step": 33470 }, { "epoch": 0.6089597846450463, "grad_norm": 1.890964388847351, "learning_rate": 4.031577565677212e-06, "loss": 0.9291, "step": 33480 }, { "epoch": 0.6091416722748686, "grad_norm": 1.7620793581008911, "learning_rate": 4.029702413321083e-06, "loss": 0.9023, "step": 33490 }, { "epoch": 0.6093235599046909, "grad_norm": 2.0071473121643066, "learning_rate": 4.0278272609649535e-06, "loss": 0.8982, "step": 33500 }, { "epoch": 0.6095054475345132, "grad_norm": 2.2239506244659424, "learning_rate": 4.025952108608825e-06, "loss": 0.9396, "step": 33510 }, { "epoch": 0.6096873351643355, "grad_norm": 1.7975202798843384, "learning_rate": 4.024076956252696e-06, "loss": 0.8795, "step": 33520 }, { "epoch": 0.6098692227941578, "grad_norm": 1.7224763631820679, "learning_rate": 4.0222018038965675e-06, "loss": 0.8331, "step": 33530 }, { "epoch": 0.6100511104239801, "grad_norm": 1.7153115272521973, "learning_rate": 4.020326651540438e-06, "loss": 0.9213, "step": 33540 }, { "epoch": 0.6102329980538024, "grad_norm": 1.6516104936599731, "learning_rate": 4.018451499184309e-06, "loss": 0.8884, "step": 33550 }, { "epoch": 0.6104148856836247, "grad_norm": 1.8389549255371094, "learning_rate": 4.01657634682818e-06, "loss": 0.9351, "step": 33560 }, { "epoch": 0.610596773313447, "grad_norm": 1.9694089889526367, "learning_rate": 4.014701194472051e-06, "loss": 0.9951, "step": 33570 }, { "epoch": 0.6107786609432693, "grad_norm": 1.68708336353302, "learning_rate": 4.012826042115922e-06, "loss": 0.8487, "step": 33580 }, { "epoch": 0.6109605485730916, "grad_norm": 1.6834707260131836, "learning_rate": 4.010950889759793e-06, "loss": 0.9152, "step": 33590 }, { "epoch": 0.6111424362029139, "grad_norm": 1.8037166595458984, "learning_rate": 4.009075737403664e-06, "loss": 0.9211, "step": 33600 }, { "epoch": 0.6113243238327362, "grad_norm": 1.924572229385376, "learning_rate": 4.007200585047536e-06, "loss": 0.9081, "step": 33610 }, { "epoch": 0.6115062114625585, "grad_norm": 2.10718035697937, "learning_rate": 4.005325432691406e-06, "loss": 0.9266, "step": 33620 }, { "epoch": 0.6116880990923808, "grad_norm": 1.7143288850784302, "learning_rate": 4.003450280335278e-06, "loss": 0.9186, "step": 33630 }, { "epoch": 0.611869986722203, "grad_norm": 1.8098031282424927, "learning_rate": 4.001575127979149e-06, "loss": 0.9296, "step": 33640 }, { "epoch": 0.6120518743520253, "grad_norm": 1.8423343896865845, "learning_rate": 3.9996999756230194e-06, "loss": 0.9305, "step": 33650 }, { "epoch": 0.6122337619818476, "grad_norm": 2.0067684650421143, "learning_rate": 3.997824823266891e-06, "loss": 0.9957, "step": 33660 }, { "epoch": 0.6124156496116699, "grad_norm": 1.748978853225708, "learning_rate": 3.995949670910762e-06, "loss": 0.9273, "step": 33670 }, { "epoch": 0.6125975372414922, "grad_norm": 1.8037883043289185, "learning_rate": 3.9940745185546335e-06, "loss": 0.9083, "step": 33680 }, { "epoch": 0.6127794248713145, "grad_norm": 2.0527584552764893, "learning_rate": 3.992199366198504e-06, "loss": 0.9313, "step": 33690 }, { "epoch": 0.6129613125011368, "grad_norm": 1.796504020690918, "learning_rate": 3.9903242138423745e-06, "loss": 0.9445, "step": 33700 }, { "epoch": 0.6131432001309591, "grad_norm": 1.710259199142456, "learning_rate": 3.988449061486246e-06, "loss": 0.9115, "step": 33710 }, { "epoch": 0.6133250877607814, "grad_norm": 1.8895467519760132, "learning_rate": 3.986573909130117e-06, "loss": 0.9248, "step": 33720 }, { "epoch": 0.6135069753906037, "grad_norm": 1.95646333694458, "learning_rate": 3.9846987567739885e-06, "loss": 0.8936, "step": 33730 }, { "epoch": 0.613688863020426, "grad_norm": 1.9211786985397339, "learning_rate": 3.982823604417859e-06, "loss": 0.8684, "step": 33740 }, { "epoch": 0.6138707506502483, "grad_norm": 2.034348249435425, "learning_rate": 3.98094845206173e-06, "loss": 0.8885, "step": 33750 }, { "epoch": 0.6140526382800706, "grad_norm": 2.1393723487854004, "learning_rate": 3.979073299705602e-06, "loss": 0.9396, "step": 33760 }, { "epoch": 0.6142345259098929, "grad_norm": 1.5773297548294067, "learning_rate": 3.977198147349472e-06, "loss": 0.9185, "step": 33770 }, { "epoch": 0.6144164135397152, "grad_norm": 1.8818999528884888, "learning_rate": 3.9753229949933435e-06, "loss": 0.874, "step": 33780 }, { "epoch": 0.6145983011695375, "grad_norm": 1.770891547203064, "learning_rate": 3.973447842637215e-06, "loss": 0.9264, "step": 33790 }, { "epoch": 0.6147801887993598, "grad_norm": 1.8127703666687012, "learning_rate": 3.971572690281086e-06, "loss": 0.9097, "step": 33800 }, { "epoch": 0.6149620764291821, "grad_norm": 1.842868447303772, "learning_rate": 3.969697537924957e-06, "loss": 0.9342, "step": 33810 }, { "epoch": 0.6151439640590044, "grad_norm": 1.7400097846984863, "learning_rate": 3.967822385568828e-06, "loss": 0.925, "step": 33820 }, { "epoch": 0.6153258516888267, "grad_norm": 1.7091853618621826, "learning_rate": 3.965947233212699e-06, "loss": 0.9222, "step": 33830 }, { "epoch": 0.6155077393186489, "grad_norm": 1.771470546722412, "learning_rate": 3.96407208085657e-06, "loss": 0.9173, "step": 33840 }, { "epoch": 0.6156896269484712, "grad_norm": 1.5563055276870728, "learning_rate": 3.9621969285004404e-06, "loss": 0.8792, "step": 33850 }, { "epoch": 0.6158715145782935, "grad_norm": 1.8599662780761719, "learning_rate": 3.960321776144312e-06, "loss": 0.9096, "step": 33860 }, { "epoch": 0.6160534022081158, "grad_norm": 1.5384700298309326, "learning_rate": 3.958446623788183e-06, "loss": 0.9182, "step": 33870 }, { "epoch": 0.6162352898379381, "grad_norm": 2.026608467102051, "learning_rate": 3.9565714714320545e-06, "loss": 0.9365, "step": 33880 }, { "epoch": 0.6164171774677604, "grad_norm": 1.700036883354187, "learning_rate": 3.954696319075925e-06, "loss": 0.9545, "step": 33890 }, { "epoch": 0.6165990650975827, "grad_norm": 1.8236943483352661, "learning_rate": 3.952821166719796e-06, "loss": 0.943, "step": 33900 }, { "epoch": 0.616780952727405, "grad_norm": 1.8015422821044922, "learning_rate": 3.950946014363668e-06, "loss": 0.9206, "step": 33910 }, { "epoch": 0.6169628403572273, "grad_norm": 1.465366005897522, "learning_rate": 3.949070862007538e-06, "loss": 0.8643, "step": 33920 }, { "epoch": 0.6171447279870496, "grad_norm": 1.8882282972335815, "learning_rate": 3.9471957096514095e-06, "loss": 0.9284, "step": 33930 }, { "epoch": 0.6173266156168719, "grad_norm": 1.7146151065826416, "learning_rate": 3.945320557295281e-06, "loss": 0.941, "step": 33940 }, { "epoch": 0.6175085032466942, "grad_norm": 2.0660033226013184, "learning_rate": 3.943445404939152e-06, "loss": 0.89, "step": 33950 }, { "epoch": 0.6176903908765164, "grad_norm": 1.6402182579040527, "learning_rate": 3.941570252583023e-06, "loss": 0.914, "step": 33960 }, { "epoch": 0.6178722785063387, "grad_norm": 1.8043001890182495, "learning_rate": 3.939695100226893e-06, "loss": 0.9095, "step": 33970 }, { "epoch": 0.618054166136161, "grad_norm": 2.0465705394744873, "learning_rate": 3.9378199478707645e-06, "loss": 0.9249, "step": 33980 }, { "epoch": 0.6182360537659833, "grad_norm": 1.8645507097244263, "learning_rate": 3.935944795514636e-06, "loss": 0.898, "step": 33990 }, { "epoch": 0.6184179413958056, "grad_norm": 2.0160515308380127, "learning_rate": 3.934069643158507e-06, "loss": 0.9276, "step": 34000 }, { "epoch": 0.6185998290256279, "grad_norm": 1.997955560684204, "learning_rate": 3.932194490802378e-06, "loss": 0.943, "step": 34010 }, { "epoch": 0.6187817166554502, "grad_norm": 1.796348214149475, "learning_rate": 3.930319338446249e-06, "loss": 0.8916, "step": 34020 }, { "epoch": 0.6189636042852725, "grad_norm": 1.4995973110198975, "learning_rate": 3.92844418609012e-06, "loss": 0.9415, "step": 34030 }, { "epoch": 0.6191454919150948, "grad_norm": 1.629045009613037, "learning_rate": 3.926569033733991e-06, "loss": 0.8847, "step": 34040 }, { "epoch": 0.6193273795449171, "grad_norm": 1.6081360578536987, "learning_rate": 3.924693881377862e-06, "loss": 0.8844, "step": 34050 }, { "epoch": 0.6195092671747394, "grad_norm": 1.846148133277893, "learning_rate": 3.922818729021734e-06, "loss": 0.9141, "step": 34060 }, { "epoch": 0.6196911548045617, "grad_norm": 1.918429970741272, "learning_rate": 3.920943576665605e-06, "loss": 0.9427, "step": 34070 }, { "epoch": 0.619873042434384, "grad_norm": 2.0083189010620117, "learning_rate": 3.9190684243094755e-06, "loss": 0.8847, "step": 34080 }, { "epoch": 0.6200549300642063, "grad_norm": 1.564800500869751, "learning_rate": 3.917193271953347e-06, "loss": 0.9217, "step": 34090 }, { "epoch": 0.6202368176940286, "grad_norm": 1.8400012254714966, "learning_rate": 3.915318119597218e-06, "loss": 0.8892, "step": 34100 }, { "epoch": 0.6204187053238509, "grad_norm": 1.6783236265182495, "learning_rate": 3.913442967241089e-06, "loss": 0.939, "step": 34110 }, { "epoch": 0.6206005929536732, "grad_norm": 1.341929316520691, "learning_rate": 3.911567814884959e-06, "loss": 0.8874, "step": 34120 }, { "epoch": 0.6207824805834955, "grad_norm": 1.6890172958374023, "learning_rate": 3.9096926625288305e-06, "loss": 0.8936, "step": 34130 }, { "epoch": 0.6209643682133178, "grad_norm": 1.6114534139633179, "learning_rate": 3.907817510172702e-06, "loss": 0.8885, "step": 34140 }, { "epoch": 0.6211462558431401, "grad_norm": 2.056154251098633, "learning_rate": 3.905942357816573e-06, "loss": 0.908, "step": 34150 }, { "epoch": 0.6213281434729624, "grad_norm": 1.8053280115127563, "learning_rate": 3.904067205460444e-06, "loss": 0.8862, "step": 34160 }, { "epoch": 0.6215100311027847, "grad_norm": 2.0730466842651367, "learning_rate": 3.902192053104315e-06, "loss": 0.9088, "step": 34170 }, { "epoch": 0.621691918732607, "grad_norm": 1.7616539001464844, "learning_rate": 3.900316900748186e-06, "loss": 0.8975, "step": 34180 }, { "epoch": 0.6218738063624293, "grad_norm": 2.073641538619995, "learning_rate": 3.898441748392057e-06, "loss": 0.8683, "step": 34190 }, { "epoch": 0.6220556939922516, "grad_norm": 1.9246840476989746, "learning_rate": 3.896566596035928e-06, "loss": 0.9167, "step": 34200 }, { "epoch": 0.6222375816220739, "grad_norm": 1.7265483140945435, "learning_rate": 3.8946914436797996e-06, "loss": 0.9034, "step": 34210 }, { "epoch": 0.6224194692518962, "grad_norm": 2.0743906497955322, "learning_rate": 3.892816291323671e-06, "loss": 0.9863, "step": 34220 }, { "epoch": 0.6226013568817185, "grad_norm": 1.663174033164978, "learning_rate": 3.890941138967541e-06, "loss": 0.864, "step": 34230 }, { "epoch": 0.6227832445115408, "grad_norm": 1.7176766395568848, "learning_rate": 3.889065986611413e-06, "loss": 0.9206, "step": 34240 }, { "epoch": 0.6229651321413631, "grad_norm": 1.842840313911438, "learning_rate": 3.887190834255283e-06, "loss": 0.8715, "step": 34250 }, { "epoch": 0.6231470197711854, "grad_norm": 1.6389003992080688, "learning_rate": 3.885315681899155e-06, "loss": 0.9317, "step": 34260 }, { "epoch": 0.6233289074010077, "grad_norm": 1.7731000185012817, "learning_rate": 3.883440529543025e-06, "loss": 0.9268, "step": 34270 }, { "epoch": 0.62351079503083, "grad_norm": 1.5640262365341187, "learning_rate": 3.8815653771868965e-06, "loss": 0.9754, "step": 34280 }, { "epoch": 0.6236926826606523, "grad_norm": 1.7656984329223633, "learning_rate": 3.879690224830768e-06, "loss": 0.9778, "step": 34290 }, { "epoch": 0.6238745702904746, "grad_norm": 1.4673734903335571, "learning_rate": 3.877815072474639e-06, "loss": 0.877, "step": 34300 }, { "epoch": 0.6240564579202968, "grad_norm": 1.6534192562103271, "learning_rate": 3.87593992011851e-06, "loss": 0.8306, "step": 34310 }, { "epoch": 0.6242383455501191, "grad_norm": 1.7187418937683105, "learning_rate": 3.874064767762381e-06, "loss": 0.9125, "step": 34320 }, { "epoch": 0.6244202331799414, "grad_norm": 1.8531609773635864, "learning_rate": 3.872189615406252e-06, "loss": 0.9666, "step": 34330 }, { "epoch": 0.6246021208097637, "grad_norm": 1.9790321588516235, "learning_rate": 3.870314463050124e-06, "loss": 0.9199, "step": 34340 }, { "epoch": 0.624784008439586, "grad_norm": 1.9025084972381592, "learning_rate": 3.868439310693994e-06, "loss": 0.9186, "step": 34350 }, { "epoch": 0.6249658960694083, "grad_norm": 1.7078937292099, "learning_rate": 3.8665641583378655e-06, "loss": 0.8903, "step": 34360 }, { "epoch": 0.6251477836992306, "grad_norm": 2.122840404510498, "learning_rate": 3.864689005981737e-06, "loss": 0.8899, "step": 34370 }, { "epoch": 0.6253296713290529, "grad_norm": 1.859971284866333, "learning_rate": 3.862813853625607e-06, "loss": 0.9576, "step": 34380 }, { "epoch": 0.6255115589588752, "grad_norm": 2.153463363647461, "learning_rate": 3.860938701269478e-06, "loss": 0.9693, "step": 34390 }, { "epoch": 0.6256934465886975, "grad_norm": 1.718993902206421, "learning_rate": 3.859063548913349e-06, "loss": 0.9277, "step": 34400 }, { "epoch": 0.6258753342185198, "grad_norm": 1.8264517784118652, "learning_rate": 3.8571883965572206e-06, "loss": 0.9705, "step": 34410 }, { "epoch": 0.6260572218483421, "grad_norm": 1.6097959280014038, "learning_rate": 3.855313244201092e-06, "loss": 0.919, "step": 34420 }, { "epoch": 0.6262391094781644, "grad_norm": 1.8570421934127808, "learning_rate": 3.853438091844962e-06, "loss": 0.9198, "step": 34430 }, { "epoch": 0.6264209971079867, "grad_norm": 1.816131591796875, "learning_rate": 3.851562939488834e-06, "loss": 0.9384, "step": 34440 }, { "epoch": 0.626602884737809, "grad_norm": 1.76336669921875, "learning_rate": 3.849687787132705e-06, "loss": 0.883, "step": 34450 }, { "epoch": 0.6267847723676313, "grad_norm": 1.887971043586731, "learning_rate": 3.847812634776576e-06, "loss": 1.0007, "step": 34460 }, { "epoch": 0.6269666599974536, "grad_norm": 1.8503254652023315, "learning_rate": 3.845937482420447e-06, "loss": 0.8967, "step": 34470 }, { "epoch": 0.6271485476272759, "grad_norm": 1.7494583129882812, "learning_rate": 3.844062330064318e-06, "loss": 0.8956, "step": 34480 }, { "epoch": 0.6273304352570982, "grad_norm": 1.8156715631484985, "learning_rate": 3.84218717770819e-06, "loss": 0.8871, "step": 34490 }, { "epoch": 0.6275123228869205, "grad_norm": 1.5177156925201416, "learning_rate": 3.84031202535206e-06, "loss": 0.9089, "step": 34500 }, { "epoch": 0.6276942105167428, "grad_norm": 1.8937116861343384, "learning_rate": 3.8384368729959315e-06, "loss": 0.9415, "step": 34510 }, { "epoch": 0.6278760981465651, "grad_norm": 1.779102087020874, "learning_rate": 3.836561720639802e-06, "loss": 0.9258, "step": 34520 }, { "epoch": 0.6280579857763874, "grad_norm": 1.5168299674987793, "learning_rate": 3.834686568283673e-06, "loss": 0.9153, "step": 34530 }, { "epoch": 0.6282398734062097, "grad_norm": 1.9190194606781006, "learning_rate": 3.832811415927544e-06, "loss": 0.9599, "step": 34540 }, { "epoch": 0.628421761036032, "grad_norm": 1.6612945795059204, "learning_rate": 3.830936263571415e-06, "loss": 0.8928, "step": 34550 }, { "epoch": 0.6286036486658543, "grad_norm": 1.972699761390686, "learning_rate": 3.8290611112152865e-06, "loss": 0.9411, "step": 34560 }, { "epoch": 0.6287855362956766, "grad_norm": 1.789770245552063, "learning_rate": 3.827185958859158e-06, "loss": 0.9573, "step": 34570 }, { "epoch": 0.6289674239254989, "grad_norm": 2.1483521461486816, "learning_rate": 3.825310806503028e-06, "loss": 0.9132, "step": 34580 }, { "epoch": 0.6291493115553212, "grad_norm": 1.9080004692077637, "learning_rate": 3.8234356541469e-06, "loss": 0.9038, "step": 34590 }, { "epoch": 0.6293311991851435, "grad_norm": 1.380690574645996, "learning_rate": 3.821560501790771e-06, "loss": 0.9047, "step": 34600 }, { "epoch": 0.6295130868149658, "grad_norm": 1.9013036489486694, "learning_rate": 3.8196853494346416e-06, "loss": 0.9016, "step": 34610 }, { "epoch": 0.6296949744447881, "grad_norm": 1.7418197393417358, "learning_rate": 3.817810197078513e-06, "loss": 0.9289, "step": 34620 }, { "epoch": 0.6298768620746104, "grad_norm": 1.5931085348129272, "learning_rate": 3.815935044722384e-06, "loss": 0.8675, "step": 34630 }, { "epoch": 0.6300587497044327, "grad_norm": 1.8142422437667847, "learning_rate": 3.814059892366255e-06, "loss": 0.8719, "step": 34640 }, { "epoch": 0.630240637334255, "grad_norm": 1.9265135526657104, "learning_rate": 3.8121847400101265e-06, "loss": 0.9295, "step": 34650 }, { "epoch": 0.6304225249640772, "grad_norm": 1.7070248126983643, "learning_rate": 3.810309587653997e-06, "loss": 0.9281, "step": 34660 }, { "epoch": 0.6306044125938995, "grad_norm": 1.7525990009307861, "learning_rate": 3.808434435297868e-06, "loss": 0.9067, "step": 34670 }, { "epoch": 0.6307863002237217, "grad_norm": 1.7604687213897705, "learning_rate": 3.8065592829417393e-06, "loss": 0.9638, "step": 34680 }, { "epoch": 0.630968187853544, "grad_norm": 1.9701699018478394, "learning_rate": 3.80468413058561e-06, "loss": 0.8845, "step": 34690 }, { "epoch": 0.6311500754833663, "grad_norm": 1.7774897813796997, "learning_rate": 3.8028089782294815e-06, "loss": 0.9153, "step": 34700 }, { "epoch": 0.6313319631131886, "grad_norm": 1.6965547800064087, "learning_rate": 3.8009338258733525e-06, "loss": 0.8861, "step": 34710 }, { "epoch": 0.6315138507430109, "grad_norm": 1.6440776586532593, "learning_rate": 3.7990586735172234e-06, "loss": 0.9408, "step": 34720 }, { "epoch": 0.6316957383728332, "grad_norm": 1.4108350276947021, "learning_rate": 3.7971835211610947e-06, "loss": 0.8614, "step": 34730 }, { "epoch": 0.6318776260026555, "grad_norm": 1.9351489543914795, "learning_rate": 3.7953083688049657e-06, "loss": 0.945, "step": 34740 }, { "epoch": 0.6320595136324778, "grad_norm": 1.7050602436065674, "learning_rate": 3.793433216448837e-06, "loss": 0.9111, "step": 34750 }, { "epoch": 0.6322414012623001, "grad_norm": 1.7805261611938477, "learning_rate": 3.791558064092708e-06, "loss": 0.9619, "step": 34760 }, { "epoch": 0.6324232888921224, "grad_norm": 1.7948118448257446, "learning_rate": 3.7896829117365793e-06, "loss": 0.9252, "step": 34770 }, { "epoch": 0.6326051765219447, "grad_norm": 1.7354973554611206, "learning_rate": 3.78780775938045e-06, "loss": 0.9078, "step": 34780 }, { "epoch": 0.632787064151767, "grad_norm": 1.9071037769317627, "learning_rate": 3.785932607024321e-06, "loss": 0.8712, "step": 34790 }, { "epoch": 0.6329689517815893, "grad_norm": 2.3774046897888184, "learning_rate": 3.784057454668192e-06, "loss": 0.891, "step": 34800 }, { "epoch": 0.6331508394114116, "grad_norm": 1.6190460920333862, "learning_rate": 3.782182302312063e-06, "loss": 0.9002, "step": 34810 }, { "epoch": 0.6333327270412339, "grad_norm": 1.6844547986984253, "learning_rate": 3.780307149955934e-06, "loss": 0.8995, "step": 34820 }, { "epoch": 0.6335146146710562, "grad_norm": 1.7245121002197266, "learning_rate": 3.7784319975998052e-06, "loss": 0.8586, "step": 34830 }, { "epoch": 0.6336965023008785, "grad_norm": 1.8026998043060303, "learning_rate": 3.776556845243676e-06, "loss": 0.9317, "step": 34840 }, { "epoch": 0.6338783899307008, "grad_norm": 1.5943574905395508, "learning_rate": 3.7746816928875475e-06, "loss": 0.9382, "step": 34850 }, { "epoch": 0.6340602775605231, "grad_norm": 1.493780493736267, "learning_rate": 3.7728065405314184e-06, "loss": 0.8298, "step": 34860 }, { "epoch": 0.6342421651903454, "grad_norm": 1.8566195964813232, "learning_rate": 3.7709313881752898e-06, "loss": 0.8947, "step": 34870 }, { "epoch": 0.6344240528201677, "grad_norm": 1.6876417398452759, "learning_rate": 3.7690562358191607e-06, "loss": 0.857, "step": 34880 }, { "epoch": 0.63460594044999, "grad_norm": 1.7192668914794922, "learning_rate": 3.7671810834630316e-06, "loss": 0.8945, "step": 34890 }, { "epoch": 0.6347878280798123, "grad_norm": 2.164149045944214, "learning_rate": 3.765305931106903e-06, "loss": 0.9565, "step": 34900 }, { "epoch": 0.6349697157096346, "grad_norm": 1.6938353776931763, "learning_rate": 3.763430778750774e-06, "loss": 0.942, "step": 34910 }, { "epoch": 0.6351516033394569, "grad_norm": 1.8224318027496338, "learning_rate": 3.7615556263946452e-06, "loss": 0.9605, "step": 34920 }, { "epoch": 0.6353334909692792, "grad_norm": 2.060216188430786, "learning_rate": 3.7596804740385157e-06, "loss": 0.8905, "step": 34930 }, { "epoch": 0.6355153785991015, "grad_norm": 1.7174073457717896, "learning_rate": 3.7578053216823867e-06, "loss": 0.922, "step": 34940 }, { "epoch": 0.6356972662289238, "grad_norm": 1.6929856538772583, "learning_rate": 3.755930169326258e-06, "loss": 0.892, "step": 34950 }, { "epoch": 0.635879153858746, "grad_norm": 1.5236437320709229, "learning_rate": 3.754055016970129e-06, "loss": 0.9058, "step": 34960 }, { "epoch": 0.6360610414885683, "grad_norm": 1.7446197271347046, "learning_rate": 3.7521798646140003e-06, "loss": 0.9249, "step": 34970 }, { "epoch": 0.6362429291183906, "grad_norm": 2.330638885498047, "learning_rate": 3.750304712257871e-06, "loss": 0.9822, "step": 34980 }, { "epoch": 0.6364248167482129, "grad_norm": 1.7049263715744019, "learning_rate": 3.748429559901742e-06, "loss": 0.8866, "step": 34990 }, { "epoch": 0.6366067043780352, "grad_norm": 1.840646743774414, "learning_rate": 3.7465544075456135e-06, "loss": 0.959, "step": 35000 }, { "epoch": 0.6367885920078575, "grad_norm": 1.8237131834030151, "learning_rate": 3.7446792551894844e-06, "loss": 0.8532, "step": 35010 }, { "epoch": 0.6369704796376798, "grad_norm": 1.6714988946914673, "learning_rate": 3.7428041028333557e-06, "loss": 0.9319, "step": 35020 }, { "epoch": 0.6371523672675021, "grad_norm": 2.006824493408203, "learning_rate": 3.7409289504772267e-06, "loss": 0.9241, "step": 35030 }, { "epoch": 0.6373342548973244, "grad_norm": 1.6985468864440918, "learning_rate": 3.739053798121098e-06, "loss": 0.9173, "step": 35040 }, { "epoch": 0.6375161425271467, "grad_norm": 2.0127639770507812, "learning_rate": 3.737178645764969e-06, "loss": 0.9359, "step": 35050 }, { "epoch": 0.637698030156969, "grad_norm": 1.9136996269226074, "learning_rate": 3.73530349340884e-06, "loss": 0.9647, "step": 35060 }, { "epoch": 0.6378799177867913, "grad_norm": 1.7063285112380981, "learning_rate": 3.7334283410527108e-06, "loss": 0.891, "step": 35070 }, { "epoch": 0.6380618054166136, "grad_norm": 1.9084405899047852, "learning_rate": 3.7315531886965817e-06, "loss": 0.9037, "step": 35080 }, { "epoch": 0.6382436930464359, "grad_norm": 1.6186113357543945, "learning_rate": 3.7296780363404526e-06, "loss": 0.9414, "step": 35090 }, { "epoch": 0.6384255806762582, "grad_norm": 1.9261200428009033, "learning_rate": 3.727802883984324e-06, "loss": 0.9343, "step": 35100 }, { "epoch": 0.6386074683060805, "grad_norm": 2.2500412464141846, "learning_rate": 3.725927731628195e-06, "loss": 0.925, "step": 35110 }, { "epoch": 0.6387893559359028, "grad_norm": 1.7531427145004272, "learning_rate": 3.7240525792720662e-06, "loss": 0.9438, "step": 35120 }, { "epoch": 0.6389712435657251, "grad_norm": 1.883966326713562, "learning_rate": 3.722177426915937e-06, "loss": 0.949, "step": 35130 }, { "epoch": 0.6391531311955474, "grad_norm": 1.9611972570419312, "learning_rate": 3.7203022745598085e-06, "loss": 0.9166, "step": 35140 }, { "epoch": 0.6393350188253697, "grad_norm": 1.8424853086471558, "learning_rate": 3.7184271222036794e-06, "loss": 0.9653, "step": 35150 }, { "epoch": 0.639516906455192, "grad_norm": 1.9691050052642822, "learning_rate": 3.7165519698475503e-06, "loss": 0.9101, "step": 35160 }, { "epoch": 0.6396987940850143, "grad_norm": 1.673384428024292, "learning_rate": 3.7146768174914217e-06, "loss": 0.8755, "step": 35170 }, { "epoch": 0.6398806817148366, "grad_norm": 2.0371744632720947, "learning_rate": 3.7128016651352926e-06, "loss": 0.8843, "step": 35180 }, { "epoch": 0.6400625693446589, "grad_norm": 1.9245977401733398, "learning_rate": 3.710926512779164e-06, "loss": 0.9506, "step": 35190 }, { "epoch": 0.6402444569744812, "grad_norm": 1.75563383102417, "learning_rate": 3.709051360423035e-06, "loss": 0.8976, "step": 35200 }, { "epoch": 0.6404263446043035, "grad_norm": 2.0239827632904053, "learning_rate": 3.7071762080669054e-06, "loss": 0.916, "step": 35210 }, { "epoch": 0.6406082322341258, "grad_norm": 1.7960189580917358, "learning_rate": 3.7053010557107767e-06, "loss": 0.8689, "step": 35220 }, { "epoch": 0.6407901198639481, "grad_norm": 2.0056376457214355, "learning_rate": 3.7034259033546476e-06, "loss": 0.9146, "step": 35230 }, { "epoch": 0.6409720074937704, "grad_norm": 1.9857748746871948, "learning_rate": 3.701550750998519e-06, "loss": 0.9104, "step": 35240 }, { "epoch": 0.6411538951235927, "grad_norm": 1.8049699068069458, "learning_rate": 3.69967559864239e-06, "loss": 0.8429, "step": 35250 }, { "epoch": 0.641335782753415, "grad_norm": 2.1852478981018066, "learning_rate": 3.697800446286261e-06, "loss": 0.8605, "step": 35260 }, { "epoch": 0.6415176703832373, "grad_norm": 1.7811357975006104, "learning_rate": 3.695925293930132e-06, "loss": 0.8988, "step": 35270 }, { "epoch": 0.6416995580130596, "grad_norm": 1.864500880241394, "learning_rate": 3.694050141574003e-06, "loss": 0.8863, "step": 35280 }, { "epoch": 0.6418814456428819, "grad_norm": 2.1878674030303955, "learning_rate": 3.6921749892178744e-06, "loss": 0.9667, "step": 35290 }, { "epoch": 0.6420633332727042, "grad_norm": 1.854137659072876, "learning_rate": 3.6902998368617454e-06, "loss": 0.9042, "step": 35300 }, { "epoch": 0.6422452209025264, "grad_norm": 1.7948673963546753, "learning_rate": 3.6884246845056167e-06, "loss": 0.9057, "step": 35310 }, { "epoch": 0.6424271085323487, "grad_norm": 1.8818168640136719, "learning_rate": 3.6865495321494876e-06, "loss": 0.9194, "step": 35320 }, { "epoch": 0.642608996162171, "grad_norm": 1.765102505683899, "learning_rate": 3.6846743797933586e-06, "loss": 0.9221, "step": 35330 }, { "epoch": 0.6427908837919933, "grad_norm": 1.458070158958435, "learning_rate": 3.68279922743723e-06, "loss": 0.8615, "step": 35340 }, { "epoch": 0.6429727714218156, "grad_norm": 1.8439390659332275, "learning_rate": 3.6809240750811004e-06, "loss": 0.9589, "step": 35350 }, { "epoch": 0.6431546590516379, "grad_norm": 1.7078166007995605, "learning_rate": 3.6790489227249713e-06, "loss": 0.9262, "step": 35360 }, { "epoch": 0.6433365466814602, "grad_norm": 1.8527930974960327, "learning_rate": 3.6771737703688427e-06, "loss": 0.879, "step": 35370 }, { "epoch": 0.6435184343112825, "grad_norm": 1.8724119663238525, "learning_rate": 3.6752986180127136e-06, "loss": 0.9355, "step": 35380 }, { "epoch": 0.6437003219411048, "grad_norm": 2.0734996795654297, "learning_rate": 3.673423465656585e-06, "loss": 0.8984, "step": 35390 }, { "epoch": 0.6438822095709271, "grad_norm": 1.8043962717056274, "learning_rate": 3.671548313300456e-06, "loss": 0.9344, "step": 35400 }, { "epoch": 0.6440640972007494, "grad_norm": 1.852675199508667, "learning_rate": 3.6696731609443272e-06, "loss": 0.9416, "step": 35410 }, { "epoch": 0.6442459848305717, "grad_norm": 1.7103484869003296, "learning_rate": 3.667798008588198e-06, "loss": 0.8702, "step": 35420 }, { "epoch": 0.644427872460394, "grad_norm": 1.6592025756835938, "learning_rate": 3.665922856232069e-06, "loss": 0.9126, "step": 35430 }, { "epoch": 0.6446097600902163, "grad_norm": 2.191230297088623, "learning_rate": 3.6640477038759404e-06, "loss": 0.9099, "step": 35440 }, { "epoch": 0.6447916477200386, "grad_norm": 1.4987645149230957, "learning_rate": 3.6621725515198113e-06, "loss": 0.8721, "step": 35450 }, { "epoch": 0.6449735353498609, "grad_norm": 1.694178581237793, "learning_rate": 3.6602973991636827e-06, "loss": 0.9185, "step": 35460 }, { "epoch": 0.6451554229796832, "grad_norm": 1.8120923042297363, "learning_rate": 3.6584222468075536e-06, "loss": 0.9027, "step": 35470 }, { "epoch": 0.6453373106095055, "grad_norm": 1.7366586923599243, "learning_rate": 3.656547094451425e-06, "loss": 0.9188, "step": 35480 }, { "epoch": 0.6455191982393278, "grad_norm": 1.853173851966858, "learning_rate": 3.6546719420952954e-06, "loss": 0.9421, "step": 35490 }, { "epoch": 0.6457010858691501, "grad_norm": 1.881579041481018, "learning_rate": 3.6527967897391664e-06, "loss": 0.935, "step": 35500 }, { "epoch": 0.6458829734989724, "grad_norm": 2.019984483718872, "learning_rate": 3.6509216373830377e-06, "loss": 0.9245, "step": 35510 }, { "epoch": 0.6460648611287946, "grad_norm": 1.8136979341506958, "learning_rate": 3.6490464850269086e-06, "loss": 0.8619, "step": 35520 }, { "epoch": 0.6462467487586169, "grad_norm": 3.4603359699249268, "learning_rate": 3.6471713326707796e-06, "loss": 0.9606, "step": 35530 }, { "epoch": 0.6464286363884392, "grad_norm": 2.0132787227630615, "learning_rate": 3.645296180314651e-06, "loss": 0.9539, "step": 35540 }, { "epoch": 0.6466105240182615, "grad_norm": 1.6312474012374878, "learning_rate": 3.643421027958522e-06, "loss": 0.8786, "step": 35550 }, { "epoch": 0.6467924116480838, "grad_norm": 1.9252760410308838, "learning_rate": 3.641545875602393e-06, "loss": 0.9419, "step": 35560 }, { "epoch": 0.6469742992779061, "grad_norm": 1.6863114833831787, "learning_rate": 3.639670723246264e-06, "loss": 0.901, "step": 35570 }, { "epoch": 0.6471561869077284, "grad_norm": 1.980211853981018, "learning_rate": 3.6377955708901354e-06, "loss": 0.942, "step": 35580 }, { "epoch": 0.6473380745375507, "grad_norm": 1.6630967855453491, "learning_rate": 3.6359204185340064e-06, "loss": 0.923, "step": 35590 }, { "epoch": 0.647519962167373, "grad_norm": 1.4877395629882812, "learning_rate": 3.6340452661778773e-06, "loss": 0.9171, "step": 35600 }, { "epoch": 0.6477018497971953, "grad_norm": 1.9042991399765015, "learning_rate": 3.6321701138217486e-06, "loss": 0.9139, "step": 35610 }, { "epoch": 0.6478837374270175, "grad_norm": 1.6390870809555054, "learning_rate": 3.630294961465619e-06, "loss": 0.8901, "step": 35620 }, { "epoch": 0.6480656250568398, "grad_norm": 1.5580377578735352, "learning_rate": 3.62841980910949e-06, "loss": 0.9139, "step": 35630 }, { "epoch": 0.6482475126866621, "grad_norm": 1.7888315916061401, "learning_rate": 3.6265446567533614e-06, "loss": 0.8956, "step": 35640 }, { "epoch": 0.6484294003164844, "grad_norm": 1.7918812036514282, "learning_rate": 3.6246695043972323e-06, "loss": 0.9238, "step": 35650 }, { "epoch": 0.6486112879463067, "grad_norm": 2.0927155017852783, "learning_rate": 3.6227943520411037e-06, "loss": 0.883, "step": 35660 }, { "epoch": 0.648793175576129, "grad_norm": 1.5681899785995483, "learning_rate": 3.6209191996849746e-06, "loss": 0.9657, "step": 35670 }, { "epoch": 0.6489750632059513, "grad_norm": 2.1521875858306885, "learning_rate": 3.619044047328846e-06, "loss": 0.8893, "step": 35680 }, { "epoch": 0.6491569508357736, "grad_norm": 1.562870740890503, "learning_rate": 3.617168894972717e-06, "loss": 0.9054, "step": 35690 }, { "epoch": 0.6493388384655959, "grad_norm": 1.9648867845535278, "learning_rate": 3.6152937426165878e-06, "loss": 0.9082, "step": 35700 }, { "epoch": 0.6495207260954182, "grad_norm": 1.5617092847824097, "learning_rate": 3.613418590260459e-06, "loss": 0.9174, "step": 35710 }, { "epoch": 0.6497026137252405, "grad_norm": 1.7817894220352173, "learning_rate": 3.61154343790433e-06, "loss": 0.896, "step": 35720 }, { "epoch": 0.6498845013550628, "grad_norm": 1.81378173828125, "learning_rate": 3.6096682855482014e-06, "loss": 0.9252, "step": 35730 }, { "epoch": 0.6500663889848851, "grad_norm": 1.989039659500122, "learning_rate": 3.6077931331920723e-06, "loss": 0.9312, "step": 35740 }, { "epoch": 0.6502482766147074, "grad_norm": 1.8471498489379883, "learning_rate": 3.6059179808359437e-06, "loss": 0.9359, "step": 35750 }, { "epoch": 0.6504301642445297, "grad_norm": 1.7470799684524536, "learning_rate": 3.604042828479814e-06, "loss": 0.9495, "step": 35760 }, { "epoch": 0.650612051874352, "grad_norm": 1.5103219747543335, "learning_rate": 3.602167676123685e-06, "loss": 0.867, "step": 35770 }, { "epoch": 0.6507939395041743, "grad_norm": 1.7779618501663208, "learning_rate": 3.6002925237675564e-06, "loss": 0.9057, "step": 35780 }, { "epoch": 0.6509758271339966, "grad_norm": 1.5336410999298096, "learning_rate": 3.5984173714114274e-06, "loss": 0.877, "step": 35790 }, { "epoch": 0.6511577147638189, "grad_norm": 2.0906546115875244, "learning_rate": 3.5965422190552983e-06, "loss": 0.9533, "step": 35800 }, { "epoch": 0.6513396023936412, "grad_norm": 1.9192923307418823, "learning_rate": 3.5946670666991696e-06, "loss": 0.9325, "step": 35810 }, { "epoch": 0.6515214900234635, "grad_norm": 1.7867056131362915, "learning_rate": 3.5927919143430405e-06, "loss": 0.935, "step": 35820 }, { "epoch": 0.6517033776532858, "grad_norm": 2.7808799743652344, "learning_rate": 3.590916761986912e-06, "loss": 0.9281, "step": 35830 }, { "epoch": 0.6518852652831081, "grad_norm": 1.9201149940490723, "learning_rate": 3.589041609630783e-06, "loss": 0.8967, "step": 35840 }, { "epoch": 0.6520671529129304, "grad_norm": 1.9394557476043701, "learning_rate": 3.587166457274654e-06, "loss": 0.9322, "step": 35850 }, { "epoch": 0.6522490405427527, "grad_norm": 1.6852123737335205, "learning_rate": 3.585291304918525e-06, "loss": 0.876, "step": 35860 }, { "epoch": 0.652430928172575, "grad_norm": 1.707686185836792, "learning_rate": 3.583416152562396e-06, "loss": 0.9491, "step": 35870 }, { "epoch": 0.6526128158023973, "grad_norm": 1.8361778259277344, "learning_rate": 3.5815410002062673e-06, "loss": 0.9007, "step": 35880 }, { "epoch": 0.6527947034322196, "grad_norm": 1.5855261087417603, "learning_rate": 3.5796658478501383e-06, "loss": 0.8849, "step": 35890 }, { "epoch": 0.6529765910620419, "grad_norm": 1.9242916107177734, "learning_rate": 3.5777906954940088e-06, "loss": 0.988, "step": 35900 }, { "epoch": 0.6531584786918642, "grad_norm": 1.8328014612197876, "learning_rate": 3.57591554313788e-06, "loss": 0.9023, "step": 35910 }, { "epoch": 0.6533403663216865, "grad_norm": 1.9542458057403564, "learning_rate": 3.574040390781751e-06, "loss": 0.9281, "step": 35920 }, { "epoch": 0.6535222539515088, "grad_norm": 1.926924705505371, "learning_rate": 3.5721652384256224e-06, "loss": 0.8696, "step": 35930 }, { "epoch": 0.653704141581331, "grad_norm": 1.8536232709884644, "learning_rate": 3.5702900860694933e-06, "loss": 0.9047, "step": 35940 }, { "epoch": 0.6538860292111534, "grad_norm": 1.7543623447418213, "learning_rate": 3.5684149337133647e-06, "loss": 0.934, "step": 35950 }, { "epoch": 0.6540679168409756, "grad_norm": 1.8524726629257202, "learning_rate": 3.5665397813572356e-06, "loss": 0.8785, "step": 35960 }, { "epoch": 0.654249804470798, "grad_norm": 1.5600829124450684, "learning_rate": 3.5646646290011065e-06, "loss": 0.9482, "step": 35970 }, { "epoch": 0.6544316921006202, "grad_norm": 1.652187466621399, "learning_rate": 3.562789476644978e-06, "loss": 0.9279, "step": 35980 }, { "epoch": 0.6546135797304425, "grad_norm": 2.007418155670166, "learning_rate": 3.5609143242888488e-06, "loss": 0.9426, "step": 35990 }, { "epoch": 0.6547954673602648, "grad_norm": 1.8493926525115967, "learning_rate": 3.55903917193272e-06, "loss": 0.9006, "step": 36000 }, { "epoch": 0.6549773549900871, "grad_norm": 2.3347792625427246, "learning_rate": 3.557164019576591e-06, "loss": 0.9454, "step": 36010 }, { "epoch": 0.6551592426199094, "grad_norm": 1.7334908246994019, "learning_rate": 3.5552888672204624e-06, "loss": 0.9494, "step": 36020 }, { "epoch": 0.6553411302497317, "grad_norm": 1.8617274761199951, "learning_rate": 3.5534137148643333e-06, "loss": 0.936, "step": 36030 }, { "epoch": 0.655523017879554, "grad_norm": 1.4432669878005981, "learning_rate": 3.551538562508204e-06, "loss": 0.9152, "step": 36040 }, { "epoch": 0.6557049055093763, "grad_norm": 1.712950587272644, "learning_rate": 3.549663410152075e-06, "loss": 0.8923, "step": 36050 }, { "epoch": 0.6558867931391986, "grad_norm": 1.5483108758926392, "learning_rate": 3.547788257795946e-06, "loss": 0.8993, "step": 36060 }, { "epoch": 0.6560686807690209, "grad_norm": 1.777249813079834, "learning_rate": 3.545913105439817e-06, "loss": 0.9153, "step": 36070 }, { "epoch": 0.6562505683988432, "grad_norm": 2.0317466259002686, "learning_rate": 3.5440379530836883e-06, "loss": 0.9364, "step": 36080 }, { "epoch": 0.6564324560286655, "grad_norm": 1.9500826597213745, "learning_rate": 3.5421628007275593e-06, "loss": 0.952, "step": 36090 }, { "epoch": 0.6566143436584878, "grad_norm": 1.9679385423660278, "learning_rate": 3.5402876483714306e-06, "loss": 0.9517, "step": 36100 }, { "epoch": 0.6567962312883101, "grad_norm": 1.9924405813217163, "learning_rate": 3.5384124960153015e-06, "loss": 0.9184, "step": 36110 }, { "epoch": 0.6569781189181324, "grad_norm": 1.4569213390350342, "learning_rate": 3.536537343659173e-06, "loss": 0.9239, "step": 36120 }, { "epoch": 0.6571600065479547, "grad_norm": 1.9975745677947998, "learning_rate": 3.534662191303044e-06, "loss": 0.9333, "step": 36130 }, { "epoch": 0.657341894177777, "grad_norm": 2.0379834175109863, "learning_rate": 3.5327870389469147e-06, "loss": 0.9151, "step": 36140 }, { "epoch": 0.6575237818075993, "grad_norm": 1.747615098953247, "learning_rate": 3.530911886590786e-06, "loss": 0.9581, "step": 36150 }, { "epoch": 0.6577056694374216, "grad_norm": 1.8305943012237549, "learning_rate": 3.529036734234657e-06, "loss": 0.9107, "step": 36160 }, { "epoch": 0.6578875570672439, "grad_norm": 2.0407192707061768, "learning_rate": 3.5271615818785283e-06, "loss": 0.9639, "step": 36170 }, { "epoch": 0.6580694446970662, "grad_norm": 1.8832200765609741, "learning_rate": 3.525286429522399e-06, "loss": 0.9219, "step": 36180 }, { "epoch": 0.6582513323268885, "grad_norm": 1.6436736583709717, "learning_rate": 3.5234112771662698e-06, "loss": 0.8852, "step": 36190 }, { "epoch": 0.6584332199567108, "grad_norm": 1.6881084442138672, "learning_rate": 3.521536124810141e-06, "loss": 0.9114, "step": 36200 }, { "epoch": 0.6586151075865331, "grad_norm": 1.647737741470337, "learning_rate": 3.519660972454012e-06, "loss": 0.9, "step": 36210 }, { "epoch": 0.6587969952163554, "grad_norm": 2.001925230026245, "learning_rate": 3.5177858200978834e-06, "loss": 0.8938, "step": 36220 }, { "epoch": 0.6589788828461777, "grad_norm": 2.0436782836914062, "learning_rate": 3.5159106677417543e-06, "loss": 0.9152, "step": 36230 }, { "epoch": 0.659160770476, "grad_norm": 2.097855567932129, "learning_rate": 3.5140355153856252e-06, "loss": 0.9373, "step": 36240 }, { "epoch": 0.6593426581058223, "grad_norm": 1.6855969429016113, "learning_rate": 3.5121603630294966e-06, "loss": 0.8881, "step": 36250 }, { "epoch": 0.6595245457356446, "grad_norm": 2.083529472351074, "learning_rate": 3.5102852106733675e-06, "loss": 0.9607, "step": 36260 }, { "epoch": 0.6597064333654669, "grad_norm": 1.7592012882232666, "learning_rate": 3.508410058317239e-06, "loss": 0.9502, "step": 36270 }, { "epoch": 0.6598883209952892, "grad_norm": 1.63860285282135, "learning_rate": 3.5065349059611098e-06, "loss": 0.8953, "step": 36280 }, { "epoch": 0.6600702086251115, "grad_norm": 2.0438530445098877, "learning_rate": 3.504659753604981e-06, "loss": 0.9282, "step": 36290 }, { "epoch": 0.6602520962549338, "grad_norm": 1.580081820487976, "learning_rate": 3.502784601248852e-06, "loss": 0.8776, "step": 36300 }, { "epoch": 0.660433983884756, "grad_norm": 2.026538372039795, "learning_rate": 3.5009094488927225e-06, "loss": 0.9501, "step": 36310 }, { "epoch": 0.6606158715145783, "grad_norm": 1.954871416091919, "learning_rate": 3.4990342965365934e-06, "loss": 0.8651, "step": 36320 }, { "epoch": 0.6607977591444006, "grad_norm": 2.6242668628692627, "learning_rate": 3.497159144180465e-06, "loss": 0.9642, "step": 36330 }, { "epoch": 0.6609796467742229, "grad_norm": 1.8665703535079956, "learning_rate": 3.4952839918243357e-06, "loss": 0.934, "step": 36340 }, { "epoch": 0.6611615344040451, "grad_norm": 2.0320990085601807, "learning_rate": 3.493408839468207e-06, "loss": 0.8859, "step": 36350 }, { "epoch": 0.6613434220338674, "grad_norm": 1.791208267211914, "learning_rate": 3.491533687112078e-06, "loss": 0.9377, "step": 36360 }, { "epoch": 0.6615253096636897, "grad_norm": 1.842911720275879, "learning_rate": 3.4896585347559493e-06, "loss": 0.913, "step": 36370 }, { "epoch": 0.661707197293512, "grad_norm": 1.826214075088501, "learning_rate": 3.4877833823998203e-06, "loss": 0.9143, "step": 36380 }, { "epoch": 0.6618890849233343, "grad_norm": 1.5603677034378052, "learning_rate": 3.4859082300436916e-06, "loss": 0.8646, "step": 36390 }, { "epoch": 0.6620709725531566, "grad_norm": 1.9510743618011475, "learning_rate": 3.4840330776875625e-06, "loss": 0.9487, "step": 36400 }, { "epoch": 0.6622528601829789, "grad_norm": 1.6193963289260864, "learning_rate": 3.4821579253314334e-06, "loss": 0.9569, "step": 36410 }, { "epoch": 0.6624347478128012, "grad_norm": 1.8703267574310303, "learning_rate": 3.4802827729753048e-06, "loss": 0.9261, "step": 36420 }, { "epoch": 0.6626166354426235, "grad_norm": 1.9272736310958862, "learning_rate": 3.4784076206191757e-06, "loss": 0.9331, "step": 36430 }, { "epoch": 0.6627985230724458, "grad_norm": 1.8767096996307373, "learning_rate": 3.476532468263047e-06, "loss": 0.955, "step": 36440 }, { "epoch": 0.6629804107022681, "grad_norm": 1.4386097192764282, "learning_rate": 3.4746573159069176e-06, "loss": 0.9143, "step": 36450 }, { "epoch": 0.6631622983320904, "grad_norm": 1.7262775897979736, "learning_rate": 3.4727821635507885e-06, "loss": 0.8707, "step": 36460 }, { "epoch": 0.6633441859619127, "grad_norm": 1.854810118675232, "learning_rate": 3.47090701119466e-06, "loss": 0.9223, "step": 36470 }, { "epoch": 0.663526073591735, "grad_norm": 1.4906370639801025, "learning_rate": 3.4690318588385307e-06, "loss": 0.9241, "step": 36480 }, { "epoch": 0.6637079612215573, "grad_norm": 1.6353435516357422, "learning_rate": 3.4671567064824017e-06, "loss": 0.9396, "step": 36490 }, { "epoch": 0.6638898488513796, "grad_norm": 1.8414195775985718, "learning_rate": 3.465281554126273e-06, "loss": 0.9095, "step": 36500 }, { "epoch": 0.6640717364812019, "grad_norm": 1.879328966140747, "learning_rate": 3.463406401770144e-06, "loss": 0.9349, "step": 36510 }, { "epoch": 0.6642536241110242, "grad_norm": 1.8018476963043213, "learning_rate": 3.4615312494140153e-06, "loss": 0.8732, "step": 36520 }, { "epoch": 0.6644355117408465, "grad_norm": 1.759614109992981, "learning_rate": 3.459656097057886e-06, "loss": 0.8976, "step": 36530 }, { "epoch": 0.6646173993706688, "grad_norm": 1.7969610691070557, "learning_rate": 3.4577809447017576e-06, "loss": 0.9503, "step": 36540 }, { "epoch": 0.6647992870004911, "grad_norm": 1.7359538078308105, "learning_rate": 3.4559057923456285e-06, "loss": 0.9107, "step": 36550 }, { "epoch": 0.6649811746303134, "grad_norm": 1.8232042789459229, "learning_rate": 3.4540306399895e-06, "loss": 0.9276, "step": 36560 }, { "epoch": 0.6651630622601357, "grad_norm": 1.7117555141448975, "learning_rate": 3.4521554876333707e-06, "loss": 0.9325, "step": 36570 }, { "epoch": 0.665344949889958, "grad_norm": 1.855176329612732, "learning_rate": 3.4502803352772417e-06, "loss": 0.9007, "step": 36580 }, { "epoch": 0.6655268375197803, "grad_norm": 2.021364688873291, "learning_rate": 3.448405182921112e-06, "loss": 0.9938, "step": 36590 }, { "epoch": 0.6657087251496026, "grad_norm": 1.6558122634887695, "learning_rate": 3.4465300305649835e-06, "loss": 0.9153, "step": 36600 }, { "epoch": 0.6658906127794249, "grad_norm": 1.714416742324829, "learning_rate": 3.4446548782088544e-06, "loss": 0.9524, "step": 36610 }, { "epoch": 0.6660725004092471, "grad_norm": 1.526942253112793, "learning_rate": 3.4427797258527258e-06, "loss": 0.9145, "step": 36620 }, { "epoch": 0.6662543880390694, "grad_norm": 1.750524640083313, "learning_rate": 3.4409045734965967e-06, "loss": 0.9305, "step": 36630 }, { "epoch": 0.6664362756688917, "grad_norm": 1.5641123056411743, "learning_rate": 3.439029421140468e-06, "loss": 0.8997, "step": 36640 }, { "epoch": 0.666618163298714, "grad_norm": 1.8152329921722412, "learning_rate": 3.437154268784339e-06, "loss": 0.9025, "step": 36650 }, { "epoch": 0.6668000509285363, "grad_norm": 1.7433604001998901, "learning_rate": 3.43527911642821e-06, "loss": 0.9487, "step": 36660 }, { "epoch": 0.6669819385583586, "grad_norm": 1.7978482246398926, "learning_rate": 3.4334039640720812e-06, "loss": 0.8798, "step": 36670 }, { "epoch": 0.6671638261881809, "grad_norm": 1.9091829061508179, "learning_rate": 3.431528811715952e-06, "loss": 0.8582, "step": 36680 }, { "epoch": 0.6673457138180032, "grad_norm": 2.053354263305664, "learning_rate": 3.4296536593598235e-06, "loss": 0.9681, "step": 36690 }, { "epoch": 0.6675276014478255, "grad_norm": 1.756234884262085, "learning_rate": 3.4277785070036944e-06, "loss": 0.9249, "step": 36700 }, { "epoch": 0.6677094890776478, "grad_norm": 1.9640408754348755, "learning_rate": 3.4259033546475658e-06, "loss": 0.9225, "step": 36710 }, { "epoch": 0.6678913767074701, "grad_norm": 1.660300612449646, "learning_rate": 3.4240282022914367e-06, "loss": 0.9257, "step": 36720 }, { "epoch": 0.6680732643372924, "grad_norm": 1.8015929460525513, "learning_rate": 3.422153049935307e-06, "loss": 0.9531, "step": 36730 }, { "epoch": 0.6682551519671147, "grad_norm": 2.047502279281616, "learning_rate": 3.4202778975791785e-06, "loss": 0.8984, "step": 36740 }, { "epoch": 0.668437039596937, "grad_norm": 1.7929562330245972, "learning_rate": 3.4184027452230495e-06, "loss": 0.94, "step": 36750 }, { "epoch": 0.6686189272267593, "grad_norm": 1.7851004600524902, "learning_rate": 3.4165275928669204e-06, "loss": 0.9228, "step": 36760 }, { "epoch": 0.6688008148565816, "grad_norm": 1.8746616840362549, "learning_rate": 3.4146524405107917e-06, "loss": 0.9281, "step": 36770 }, { "epoch": 0.6689827024864039, "grad_norm": 1.8719826936721802, "learning_rate": 3.4127772881546627e-06, "loss": 0.9735, "step": 36780 }, { "epoch": 0.6691645901162262, "grad_norm": 1.6705026626586914, "learning_rate": 3.410902135798534e-06, "loss": 0.9544, "step": 36790 }, { "epoch": 0.6693464777460485, "grad_norm": 2.014188528060913, "learning_rate": 3.409026983442405e-06, "loss": 0.8979, "step": 36800 }, { "epoch": 0.6695283653758708, "grad_norm": 1.9278243780136108, "learning_rate": 3.4071518310862763e-06, "loss": 0.9784, "step": 36810 }, { "epoch": 0.6697102530056931, "grad_norm": 1.8418865203857422, "learning_rate": 3.405276678730147e-06, "loss": 0.9556, "step": 36820 }, { "epoch": 0.6698921406355154, "grad_norm": 1.8624428510665894, "learning_rate": 3.403401526374018e-06, "loss": 0.9212, "step": 36830 }, { "epoch": 0.6700740282653377, "grad_norm": 1.6920396089553833, "learning_rate": 3.4015263740178895e-06, "loss": 0.8805, "step": 36840 }, { "epoch": 0.67025591589516, "grad_norm": 1.5917963981628418, "learning_rate": 3.3996512216617604e-06, "loss": 0.914, "step": 36850 }, { "epoch": 0.6704378035249823, "grad_norm": 2.188643455505371, "learning_rate": 3.397776069305631e-06, "loss": 0.9116, "step": 36860 }, { "epoch": 0.6706196911548046, "grad_norm": 1.994233250617981, "learning_rate": 3.3959009169495022e-06, "loss": 0.8526, "step": 36870 }, { "epoch": 0.6708015787846269, "grad_norm": 1.9204285144805908, "learning_rate": 3.394025764593373e-06, "loss": 0.8952, "step": 36880 }, { "epoch": 0.6709834664144492, "grad_norm": 1.7116070985794067, "learning_rate": 3.3921506122372445e-06, "loss": 0.8831, "step": 36890 }, { "epoch": 0.6711653540442715, "grad_norm": 1.903015375137329, "learning_rate": 3.3902754598811154e-06, "loss": 0.977, "step": 36900 }, { "epoch": 0.6713472416740938, "grad_norm": 2.2148380279541016, "learning_rate": 3.3884003075249868e-06, "loss": 0.9703, "step": 36910 }, { "epoch": 0.6715291293039161, "grad_norm": 1.7400041818618774, "learning_rate": 3.3865251551688577e-06, "loss": 0.9096, "step": 36920 }, { "epoch": 0.6717110169337384, "grad_norm": 1.9059343338012695, "learning_rate": 3.3846500028127286e-06, "loss": 0.9385, "step": 36930 }, { "epoch": 0.6718929045635607, "grad_norm": 1.7508853673934937, "learning_rate": 3.3827748504566e-06, "loss": 0.9913, "step": 36940 }, { "epoch": 0.672074792193383, "grad_norm": 1.611193299293518, "learning_rate": 3.380899698100471e-06, "loss": 0.9333, "step": 36950 }, { "epoch": 0.6722566798232052, "grad_norm": 1.7654218673706055, "learning_rate": 3.3790245457443422e-06, "loss": 0.8774, "step": 36960 }, { "epoch": 0.6724385674530275, "grad_norm": 1.8349578380584717, "learning_rate": 3.377149393388213e-06, "loss": 0.9097, "step": 36970 }, { "epoch": 0.6726204550828498, "grad_norm": 1.8652702569961548, "learning_rate": 3.3752742410320845e-06, "loss": 0.8667, "step": 36980 }, { "epoch": 0.6728023427126721, "grad_norm": 1.8159314393997192, "learning_rate": 3.3733990886759554e-06, "loss": 0.9101, "step": 36990 }, { "epoch": 0.6729842303424944, "grad_norm": 1.8206707239151, "learning_rate": 3.371523936319826e-06, "loss": 0.8891, "step": 37000 }, { "epoch": 0.6731661179723167, "grad_norm": 1.5865856409072876, "learning_rate": 3.3696487839636973e-06, "loss": 0.9165, "step": 37010 }, { "epoch": 0.673348005602139, "grad_norm": 1.8478156328201294, "learning_rate": 3.367773631607568e-06, "loss": 0.9387, "step": 37020 }, { "epoch": 0.6735298932319613, "grad_norm": 1.7724337577819824, "learning_rate": 3.365898479251439e-06, "loss": 0.8895, "step": 37030 }, { "epoch": 0.6737117808617836, "grad_norm": 1.8671132326126099, "learning_rate": 3.3640233268953105e-06, "loss": 0.922, "step": 37040 }, { "epoch": 0.6738936684916059, "grad_norm": 1.9989399909973145, "learning_rate": 3.3621481745391814e-06, "loss": 0.8899, "step": 37050 }, { "epoch": 0.6740755561214282, "grad_norm": 1.5742394924163818, "learning_rate": 3.3602730221830527e-06, "loss": 0.8392, "step": 37060 }, { "epoch": 0.6742574437512505, "grad_norm": 1.6119225025177002, "learning_rate": 3.3583978698269236e-06, "loss": 0.9342, "step": 37070 }, { "epoch": 0.6744393313810728, "grad_norm": 2.5035603046417236, "learning_rate": 3.356522717470795e-06, "loss": 0.9169, "step": 37080 }, { "epoch": 0.6746212190108951, "grad_norm": 1.6229885816574097, "learning_rate": 3.354647565114666e-06, "loss": 0.8819, "step": 37090 }, { "epoch": 0.6748031066407174, "grad_norm": 2.1613261699676514, "learning_rate": 3.352772412758537e-06, "loss": 0.9399, "step": 37100 }, { "epoch": 0.6749849942705397, "grad_norm": 1.8503670692443848, "learning_rate": 3.350897260402408e-06, "loss": 0.9694, "step": 37110 }, { "epoch": 0.675166881900362, "grad_norm": 2.771941900253296, "learning_rate": 3.349022108046279e-06, "loss": 0.9166, "step": 37120 }, { "epoch": 0.6753487695301843, "grad_norm": 1.994605302810669, "learning_rate": 3.3471469556901505e-06, "loss": 0.9141, "step": 37130 }, { "epoch": 0.6755306571600066, "grad_norm": 1.7458540201187134, "learning_rate": 3.345271803334021e-06, "loss": 0.9565, "step": 37140 }, { "epoch": 0.6757125447898289, "grad_norm": 1.6830155849456787, "learning_rate": 3.343396650977892e-06, "loss": 0.9323, "step": 37150 }, { "epoch": 0.6758944324196512, "grad_norm": 1.9454553127288818, "learning_rate": 3.3415214986217632e-06, "loss": 0.9261, "step": 37160 }, { "epoch": 0.6760763200494735, "grad_norm": 2.0181732177734375, "learning_rate": 3.339646346265634e-06, "loss": 0.8796, "step": 37170 }, { "epoch": 0.6762582076792958, "grad_norm": 1.8411269187927246, "learning_rate": 3.3377711939095055e-06, "loss": 0.9144, "step": 37180 }, { "epoch": 0.676440095309118, "grad_norm": 2.0486574172973633, "learning_rate": 3.3358960415533764e-06, "loss": 0.9739, "step": 37190 }, { "epoch": 0.6766219829389403, "grad_norm": 1.9298819303512573, "learning_rate": 3.3340208891972473e-06, "loss": 0.98, "step": 37200 }, { "epoch": 0.6768038705687626, "grad_norm": 1.9493865966796875, "learning_rate": 3.3321457368411187e-06, "loss": 0.9349, "step": 37210 }, { "epoch": 0.6769857581985849, "grad_norm": 1.8308111429214478, "learning_rate": 3.3302705844849896e-06, "loss": 0.9635, "step": 37220 }, { "epoch": 0.6771676458284072, "grad_norm": 1.83572256565094, "learning_rate": 3.328395432128861e-06, "loss": 0.8399, "step": 37230 }, { "epoch": 0.6773495334582295, "grad_norm": 1.884605050086975, "learning_rate": 3.326520279772732e-06, "loss": 0.9032, "step": 37240 }, { "epoch": 0.6775314210880518, "grad_norm": 1.9072190523147583, "learning_rate": 3.3246451274166032e-06, "loss": 0.8926, "step": 37250 }, { "epoch": 0.677713308717874, "grad_norm": 1.7049870491027832, "learning_rate": 3.322769975060474e-06, "loss": 0.9009, "step": 37260 }, { "epoch": 0.6778951963476963, "grad_norm": 1.6335763931274414, "learning_rate": 3.320894822704345e-06, "loss": 0.854, "step": 37270 }, { "epoch": 0.6780770839775186, "grad_norm": 1.7307498455047607, "learning_rate": 3.319019670348216e-06, "loss": 0.8942, "step": 37280 }, { "epoch": 0.6782589716073409, "grad_norm": 1.8881864547729492, "learning_rate": 3.317144517992087e-06, "loss": 0.949, "step": 37290 }, { "epoch": 0.6784408592371632, "grad_norm": 1.6831691265106201, "learning_rate": 3.315269365635958e-06, "loss": 0.8999, "step": 37300 }, { "epoch": 0.6786227468669855, "grad_norm": 1.492645502090454, "learning_rate": 3.313394213279829e-06, "loss": 0.8633, "step": 37310 }, { "epoch": 0.6788046344968078, "grad_norm": 1.7764889001846313, "learning_rate": 3.3115190609237e-06, "loss": 0.9353, "step": 37320 }, { "epoch": 0.6789865221266301, "grad_norm": 2.246408462524414, "learning_rate": 3.3096439085675714e-06, "loss": 0.9767, "step": 37330 }, { "epoch": 0.6791684097564524, "grad_norm": 1.6311293840408325, "learning_rate": 3.3077687562114424e-06, "loss": 0.9355, "step": 37340 }, { "epoch": 0.6793502973862747, "grad_norm": 1.8130003213882446, "learning_rate": 3.3058936038553137e-06, "loss": 0.8973, "step": 37350 }, { "epoch": 0.679532185016097, "grad_norm": 1.5171222686767578, "learning_rate": 3.3040184514991846e-06, "loss": 0.8765, "step": 37360 }, { "epoch": 0.6797140726459193, "grad_norm": 2.6933183670043945, "learning_rate": 3.3021432991430556e-06, "loss": 0.9355, "step": 37370 }, { "epoch": 0.6798959602757416, "grad_norm": 1.6485912799835205, "learning_rate": 3.300268146786927e-06, "loss": 0.8557, "step": 37380 }, { "epoch": 0.6800778479055639, "grad_norm": 1.651051640510559, "learning_rate": 3.298392994430798e-06, "loss": 0.9244, "step": 37390 }, { "epoch": 0.6802597355353862, "grad_norm": 1.7844021320343018, "learning_rate": 3.296517842074669e-06, "loss": 0.913, "step": 37400 }, { "epoch": 0.6804416231652085, "grad_norm": 1.7459601163864136, "learning_rate": 3.29464268971854e-06, "loss": 0.849, "step": 37410 }, { "epoch": 0.6806235107950308, "grad_norm": 1.8495736122131348, "learning_rate": 3.2927675373624106e-06, "loss": 0.9721, "step": 37420 }, { "epoch": 0.6808053984248531, "grad_norm": 1.520182490348816, "learning_rate": 3.290892385006282e-06, "loss": 0.9055, "step": 37430 }, { "epoch": 0.6809872860546754, "grad_norm": 2.0290565490722656, "learning_rate": 3.289017232650153e-06, "loss": 0.9197, "step": 37440 }, { "epoch": 0.6811691736844977, "grad_norm": 1.9555387496948242, "learning_rate": 3.287142080294024e-06, "loss": 0.9193, "step": 37450 }, { "epoch": 0.68135106131432, "grad_norm": 1.693869709968567, "learning_rate": 3.285266927937895e-06, "loss": 0.9513, "step": 37460 }, { "epoch": 0.6815329489441423, "grad_norm": 1.6342384815216064, "learning_rate": 3.283391775581766e-06, "loss": 0.9143, "step": 37470 }, { "epoch": 0.6817148365739646, "grad_norm": 1.8228451013565063, "learning_rate": 3.2815166232256374e-06, "loss": 0.9035, "step": 37480 }, { "epoch": 0.6818967242037869, "grad_norm": 1.7356908321380615, "learning_rate": 3.2796414708695083e-06, "loss": 0.8496, "step": 37490 }, { "epoch": 0.6820786118336092, "grad_norm": 1.6314960718154907, "learning_rate": 3.2777663185133797e-06, "loss": 0.87, "step": 37500 }, { "epoch": 0.6822604994634315, "grad_norm": 1.7308720350265503, "learning_rate": 3.2758911661572506e-06, "loss": 0.8836, "step": 37510 }, { "epoch": 0.6824423870932538, "grad_norm": 1.8439710140228271, "learning_rate": 3.274016013801122e-06, "loss": 0.863, "step": 37520 }, { "epoch": 0.6826242747230761, "grad_norm": 1.8766230344772339, "learning_rate": 3.272140861444993e-06, "loss": 0.9635, "step": 37530 }, { "epoch": 0.6828061623528984, "grad_norm": 1.7240102291107178, "learning_rate": 3.2702657090888638e-06, "loss": 0.887, "step": 37540 }, { "epoch": 0.6829880499827207, "grad_norm": 2.0339934825897217, "learning_rate": 3.2683905567327347e-06, "loss": 0.9355, "step": 37550 }, { "epoch": 0.683169937612543, "grad_norm": 1.8354023694992065, "learning_rate": 3.2665154043766056e-06, "loss": 0.9235, "step": 37560 }, { "epoch": 0.6833518252423653, "grad_norm": 1.7509652376174927, "learning_rate": 3.2646402520204765e-06, "loss": 0.9029, "step": 37570 }, { "epoch": 0.6835337128721876, "grad_norm": 1.9638590812683105, "learning_rate": 3.262765099664348e-06, "loss": 0.8913, "step": 37580 }, { "epoch": 0.6837156005020099, "grad_norm": 1.8342894315719604, "learning_rate": 3.260889947308219e-06, "loss": 0.9008, "step": 37590 }, { "epoch": 0.6838974881318322, "grad_norm": 1.6409437656402588, "learning_rate": 3.25901479495209e-06, "loss": 0.9386, "step": 37600 }, { "epoch": 0.6840793757616545, "grad_norm": 1.9657869338989258, "learning_rate": 3.257139642595961e-06, "loss": 0.9037, "step": 37610 }, { "epoch": 0.6842612633914767, "grad_norm": 1.893599271774292, "learning_rate": 3.2552644902398324e-06, "loss": 0.9146, "step": 37620 }, { "epoch": 0.684443151021299, "grad_norm": 1.7069648504257202, "learning_rate": 3.2533893378837034e-06, "loss": 0.9169, "step": 37630 }, { "epoch": 0.6846250386511213, "grad_norm": 1.689510703086853, "learning_rate": 3.2515141855275743e-06, "loss": 0.9258, "step": 37640 }, { "epoch": 0.6848069262809436, "grad_norm": 1.5168652534484863, "learning_rate": 3.2496390331714456e-06, "loss": 0.885, "step": 37650 }, { "epoch": 0.6849888139107659, "grad_norm": 1.8248481750488281, "learning_rate": 3.2477638808153165e-06, "loss": 0.9172, "step": 37660 }, { "epoch": 0.6851707015405882, "grad_norm": 1.791353464126587, "learning_rate": 3.245888728459188e-06, "loss": 0.9201, "step": 37670 }, { "epoch": 0.6853525891704105, "grad_norm": 1.7076419591903687, "learning_rate": 3.244013576103059e-06, "loss": 0.886, "step": 37680 }, { "epoch": 0.6855344768002328, "grad_norm": 1.6321697235107422, "learning_rate": 3.2421384237469293e-06, "loss": 0.859, "step": 37690 }, { "epoch": 0.6857163644300551, "grad_norm": 1.8626185655593872, "learning_rate": 3.2402632713908007e-06, "loss": 0.9174, "step": 37700 }, { "epoch": 0.6858982520598774, "grad_norm": 1.8376775979995728, "learning_rate": 3.2383881190346716e-06, "loss": 0.8682, "step": 37710 }, { "epoch": 0.6860801396896997, "grad_norm": 1.9244205951690674, "learning_rate": 3.236512966678543e-06, "loss": 0.9088, "step": 37720 }, { "epoch": 0.686262027319522, "grad_norm": 1.6889758110046387, "learning_rate": 3.234637814322414e-06, "loss": 0.8943, "step": 37730 }, { "epoch": 0.6864439149493443, "grad_norm": 2.0755650997161865, "learning_rate": 3.2327626619662848e-06, "loss": 0.9136, "step": 37740 }, { "epoch": 0.6866258025791666, "grad_norm": 1.8653230667114258, "learning_rate": 3.230887509610156e-06, "loss": 0.9658, "step": 37750 }, { "epoch": 0.6868076902089889, "grad_norm": 1.6779088973999023, "learning_rate": 3.229012357254027e-06, "loss": 0.8748, "step": 37760 }, { "epoch": 0.6869895778388112, "grad_norm": 1.697314977645874, "learning_rate": 3.2271372048978984e-06, "loss": 0.9563, "step": 37770 }, { "epoch": 0.6871714654686335, "grad_norm": 1.619655966758728, "learning_rate": 3.2252620525417693e-06, "loss": 0.8989, "step": 37780 }, { "epoch": 0.6873533530984558, "grad_norm": 1.6589019298553467, "learning_rate": 3.2233869001856407e-06, "loss": 0.8738, "step": 37790 }, { "epoch": 0.6875352407282781, "grad_norm": 1.804831862449646, "learning_rate": 3.2215117478295116e-06, "loss": 0.958, "step": 37800 }, { "epoch": 0.6877171283581004, "grad_norm": 1.795035719871521, "learning_rate": 3.2196365954733825e-06, "loss": 0.8831, "step": 37810 }, { "epoch": 0.6878990159879227, "grad_norm": 1.8527395725250244, "learning_rate": 3.217761443117254e-06, "loss": 0.9139, "step": 37820 }, { "epoch": 0.688080903617745, "grad_norm": 1.5000611543655396, "learning_rate": 3.2158862907611243e-06, "loss": 0.9079, "step": 37830 }, { "epoch": 0.6882627912475673, "grad_norm": 1.8019914627075195, "learning_rate": 3.2140111384049953e-06, "loss": 0.9067, "step": 37840 }, { "epoch": 0.6884446788773896, "grad_norm": 1.7130310535430908, "learning_rate": 3.2121359860488666e-06, "loss": 0.9041, "step": 37850 }, { "epoch": 0.6886265665072119, "grad_norm": 1.8222442865371704, "learning_rate": 3.2102608336927375e-06, "loss": 0.8654, "step": 37860 }, { "epoch": 0.6888084541370342, "grad_norm": 1.6548694372177124, "learning_rate": 3.208385681336609e-06, "loss": 0.9154, "step": 37870 }, { "epoch": 0.6889903417668565, "grad_norm": 1.8606284856796265, "learning_rate": 3.20651052898048e-06, "loss": 0.9107, "step": 37880 }, { "epoch": 0.6891722293966788, "grad_norm": 1.5015639066696167, "learning_rate": 3.204635376624351e-06, "loss": 0.902, "step": 37890 }, { "epoch": 0.6893541170265011, "grad_norm": 1.7111890316009521, "learning_rate": 3.202760224268222e-06, "loss": 0.9028, "step": 37900 }, { "epoch": 0.6895360046563234, "grad_norm": 1.9795054197311401, "learning_rate": 3.200885071912093e-06, "loss": 0.8877, "step": 37910 }, { "epoch": 0.6897178922861457, "grad_norm": 1.976670503616333, "learning_rate": 3.1990099195559643e-06, "loss": 0.9263, "step": 37920 }, { "epoch": 0.689899779915968, "grad_norm": 1.9477120637893677, "learning_rate": 3.1971347671998353e-06, "loss": 0.8955, "step": 37930 }, { "epoch": 0.6900816675457903, "grad_norm": 1.9500881433486938, "learning_rate": 3.1952596148437066e-06, "loss": 0.872, "step": 37940 }, { "epoch": 0.6902635551756126, "grad_norm": 1.748915195465088, "learning_rate": 3.1933844624875775e-06, "loss": 0.9283, "step": 37950 }, { "epoch": 0.6904454428054349, "grad_norm": 1.866405725479126, "learning_rate": 3.191509310131449e-06, "loss": 0.9154, "step": 37960 }, { "epoch": 0.6906273304352571, "grad_norm": 1.8558822870254517, "learning_rate": 3.1896341577753194e-06, "loss": 0.8763, "step": 37970 }, { "epoch": 0.6908092180650794, "grad_norm": 1.702698826789856, "learning_rate": 3.1877590054191903e-06, "loss": 0.8291, "step": 37980 }, { "epoch": 0.6909911056949017, "grad_norm": 1.715094804763794, "learning_rate": 3.1858838530630616e-06, "loss": 0.9323, "step": 37990 }, { "epoch": 0.691172993324724, "grad_norm": 1.8437352180480957, "learning_rate": 3.1840087007069326e-06, "loss": 0.8791, "step": 38000 }, { "epoch": 0.6913548809545463, "grad_norm": 1.7660740613937378, "learning_rate": 3.1821335483508035e-06, "loss": 0.9208, "step": 38010 }, { "epoch": 0.6915367685843685, "grad_norm": 1.8270787000656128, "learning_rate": 3.180258395994675e-06, "loss": 0.9272, "step": 38020 }, { "epoch": 0.6917186562141908, "grad_norm": 1.637895941734314, "learning_rate": 3.1783832436385458e-06, "loss": 0.9704, "step": 38030 }, { "epoch": 0.6919005438440131, "grad_norm": 1.9364575147628784, "learning_rate": 3.176508091282417e-06, "loss": 0.9348, "step": 38040 }, { "epoch": 0.6920824314738354, "grad_norm": 2.0131678581237793, "learning_rate": 3.174632938926288e-06, "loss": 0.9133, "step": 38050 }, { "epoch": 0.6922643191036577, "grad_norm": 1.6272459030151367, "learning_rate": 3.1727577865701594e-06, "loss": 0.973, "step": 38060 }, { "epoch": 0.69244620673348, "grad_norm": 1.944472074508667, "learning_rate": 3.1708826342140303e-06, "loss": 0.8946, "step": 38070 }, { "epoch": 0.6926280943633023, "grad_norm": 1.763022780418396, "learning_rate": 3.1690074818579012e-06, "loss": 0.9133, "step": 38080 }, { "epoch": 0.6928099819931246, "grad_norm": 2.026297092437744, "learning_rate": 3.1671323295017726e-06, "loss": 0.8948, "step": 38090 }, { "epoch": 0.6929918696229469, "grad_norm": 1.8063064813613892, "learning_rate": 3.165257177145643e-06, "loss": 0.9562, "step": 38100 }, { "epoch": 0.6931737572527692, "grad_norm": 1.7922275066375732, "learning_rate": 3.163382024789514e-06, "loss": 0.9031, "step": 38110 }, { "epoch": 0.6933556448825915, "grad_norm": 1.7323517799377441, "learning_rate": 3.1615068724333853e-06, "loss": 0.8953, "step": 38120 }, { "epoch": 0.6935375325124138, "grad_norm": 1.9057224988937378, "learning_rate": 3.1596317200772563e-06, "loss": 0.9408, "step": 38130 }, { "epoch": 0.6937194201422361, "grad_norm": 1.6619354486465454, "learning_rate": 3.1577565677211276e-06, "loss": 0.8999, "step": 38140 }, { "epoch": 0.6939013077720584, "grad_norm": 1.7458879947662354, "learning_rate": 3.1558814153649985e-06, "loss": 0.9277, "step": 38150 }, { "epoch": 0.6940831954018807, "grad_norm": 1.4708800315856934, "learning_rate": 3.15400626300887e-06, "loss": 0.9182, "step": 38160 }, { "epoch": 0.694265083031703, "grad_norm": 1.7970385551452637, "learning_rate": 3.152131110652741e-06, "loss": 0.8558, "step": 38170 }, { "epoch": 0.6944469706615253, "grad_norm": 1.7528120279312134, "learning_rate": 3.1502559582966117e-06, "loss": 0.9138, "step": 38180 }, { "epoch": 0.6946288582913476, "grad_norm": 1.5162310600280762, "learning_rate": 3.148380805940483e-06, "loss": 0.8592, "step": 38190 }, { "epoch": 0.6948107459211699, "grad_norm": 1.7899062633514404, "learning_rate": 3.146505653584354e-06, "loss": 0.8887, "step": 38200 }, { "epoch": 0.6949926335509922, "grad_norm": 1.7564643621444702, "learning_rate": 3.1446305012282253e-06, "loss": 0.9028, "step": 38210 }, { "epoch": 0.6951745211808145, "grad_norm": 1.969649314880371, "learning_rate": 3.1427553488720963e-06, "loss": 0.9232, "step": 38220 }, { "epoch": 0.6953564088106368, "grad_norm": 1.8757492303848267, "learning_rate": 3.1408801965159676e-06, "loss": 0.886, "step": 38230 }, { "epoch": 0.6955382964404591, "grad_norm": 2.396071434020996, "learning_rate": 3.139005044159838e-06, "loss": 0.9392, "step": 38240 }, { "epoch": 0.6957201840702814, "grad_norm": 1.4768074750900269, "learning_rate": 3.137129891803709e-06, "loss": 0.8774, "step": 38250 }, { "epoch": 0.6959020717001037, "grad_norm": 1.6370301246643066, "learning_rate": 3.1352547394475804e-06, "loss": 0.8842, "step": 38260 }, { "epoch": 0.696083959329926, "grad_norm": 1.6612927913665771, "learning_rate": 3.1333795870914513e-06, "loss": 0.9002, "step": 38270 }, { "epoch": 0.6962658469597482, "grad_norm": 1.905998706817627, "learning_rate": 3.1315044347353222e-06, "loss": 0.9614, "step": 38280 }, { "epoch": 0.6964477345895705, "grad_norm": 1.683234691619873, "learning_rate": 3.1296292823791936e-06, "loss": 0.8763, "step": 38290 }, { "epoch": 0.6966296222193928, "grad_norm": 2.014326810836792, "learning_rate": 3.1277541300230645e-06, "loss": 0.9011, "step": 38300 }, { "epoch": 0.6968115098492151, "grad_norm": 1.8310861587524414, "learning_rate": 3.125878977666936e-06, "loss": 0.9185, "step": 38310 }, { "epoch": 0.6969933974790374, "grad_norm": 1.703579306602478, "learning_rate": 3.1240038253108067e-06, "loss": 0.9551, "step": 38320 }, { "epoch": 0.6971752851088597, "grad_norm": 1.8551876544952393, "learning_rate": 3.122128672954678e-06, "loss": 0.8872, "step": 38330 }, { "epoch": 0.697357172738682, "grad_norm": 1.5861525535583496, "learning_rate": 3.120253520598549e-06, "loss": 0.9208, "step": 38340 }, { "epoch": 0.6975390603685043, "grad_norm": 1.9386872053146362, "learning_rate": 3.11837836824242e-06, "loss": 0.8911, "step": 38350 }, { "epoch": 0.6977209479983266, "grad_norm": 1.6845636367797852, "learning_rate": 3.1165032158862913e-06, "loss": 0.9247, "step": 38360 }, { "epoch": 0.6979028356281489, "grad_norm": 1.788857340812683, "learning_rate": 3.114628063530162e-06, "loss": 0.8642, "step": 38370 }, { "epoch": 0.6980847232579712, "grad_norm": 1.6848896741867065, "learning_rate": 3.1127529111740327e-06, "loss": 0.8619, "step": 38380 }, { "epoch": 0.6982666108877935, "grad_norm": 1.7279715538024902, "learning_rate": 3.110877758817904e-06, "loss": 0.8906, "step": 38390 }, { "epoch": 0.6984484985176158, "grad_norm": 1.5337413549423218, "learning_rate": 3.109002606461775e-06, "loss": 0.8673, "step": 38400 }, { "epoch": 0.6986303861474381, "grad_norm": 1.871241569519043, "learning_rate": 3.1071274541056463e-06, "loss": 0.8829, "step": 38410 }, { "epoch": 0.6988122737772604, "grad_norm": 1.6116584539413452, "learning_rate": 3.1052523017495172e-06, "loss": 0.8752, "step": 38420 }, { "epoch": 0.6989941614070827, "grad_norm": 1.628462553024292, "learning_rate": 3.1033771493933886e-06, "loss": 0.9199, "step": 38430 }, { "epoch": 0.699176049036905, "grad_norm": 1.7513644695281982, "learning_rate": 3.1015019970372595e-06, "loss": 0.8841, "step": 38440 }, { "epoch": 0.6993579366667273, "grad_norm": 1.955033779144287, "learning_rate": 3.0996268446811304e-06, "loss": 0.889, "step": 38450 }, { "epoch": 0.6995398242965496, "grad_norm": 1.723875641822815, "learning_rate": 3.0977516923250018e-06, "loss": 0.8684, "step": 38460 }, { "epoch": 0.6997217119263719, "grad_norm": 1.872239351272583, "learning_rate": 3.0958765399688727e-06, "loss": 0.9173, "step": 38470 }, { "epoch": 0.6999035995561942, "grad_norm": 1.7132093906402588, "learning_rate": 3.094001387612744e-06, "loss": 0.8843, "step": 38480 }, { "epoch": 0.7000854871860165, "grad_norm": 1.7412558794021606, "learning_rate": 3.092126235256615e-06, "loss": 0.9076, "step": 38490 }, { "epoch": 0.7002673748158388, "grad_norm": 1.909279227256775, "learning_rate": 3.0902510829004863e-06, "loss": 0.9603, "step": 38500 }, { "epoch": 0.7004492624456611, "grad_norm": 1.3535352945327759, "learning_rate": 3.0883759305443572e-06, "loss": 0.8901, "step": 38510 }, { "epoch": 0.7006311500754834, "grad_norm": 1.9524568319320679, "learning_rate": 3.0865007781882277e-06, "loss": 0.8755, "step": 38520 }, { "epoch": 0.7008130377053057, "grad_norm": 1.6803064346313477, "learning_rate": 3.084625625832099e-06, "loss": 0.9228, "step": 38530 }, { "epoch": 0.700994925335128, "grad_norm": 1.593557357788086, "learning_rate": 3.08275047347597e-06, "loss": 0.8728, "step": 38540 }, { "epoch": 0.7011768129649503, "grad_norm": 1.968599557876587, "learning_rate": 3.080875321119841e-06, "loss": 0.9623, "step": 38550 }, { "epoch": 0.7013587005947726, "grad_norm": 1.7453272342681885, "learning_rate": 3.0790001687637123e-06, "loss": 0.9001, "step": 38560 }, { "epoch": 0.7015405882245949, "grad_norm": 1.591606855392456, "learning_rate": 3.077125016407583e-06, "loss": 0.9208, "step": 38570 }, { "epoch": 0.7017224758544172, "grad_norm": 1.8722937107086182, "learning_rate": 3.0752498640514545e-06, "loss": 0.9128, "step": 38580 }, { "epoch": 0.7019043634842395, "grad_norm": 1.835373044013977, "learning_rate": 3.0733747116953255e-06, "loss": 0.8971, "step": 38590 }, { "epoch": 0.7020862511140618, "grad_norm": 1.6987261772155762, "learning_rate": 3.071499559339197e-06, "loss": 0.9239, "step": 38600 }, { "epoch": 0.702268138743884, "grad_norm": 1.6778908967971802, "learning_rate": 3.0696244069830677e-06, "loss": 0.9517, "step": 38610 }, { "epoch": 0.7024500263737063, "grad_norm": 1.9615843296051025, "learning_rate": 3.0677492546269387e-06, "loss": 0.9382, "step": 38620 }, { "epoch": 0.7026319140035286, "grad_norm": 1.7224866151809692, "learning_rate": 3.06587410227081e-06, "loss": 0.8874, "step": 38630 }, { "epoch": 0.7028138016333509, "grad_norm": 1.4005235433578491, "learning_rate": 3.063998949914681e-06, "loss": 0.9126, "step": 38640 }, { "epoch": 0.7029956892631732, "grad_norm": 1.9338330030441284, "learning_rate": 3.0621237975585523e-06, "loss": 0.9874, "step": 38650 }, { "epoch": 0.7031775768929955, "grad_norm": 2.1316325664520264, "learning_rate": 3.0602486452024228e-06, "loss": 0.9238, "step": 38660 }, { "epoch": 0.7033594645228178, "grad_norm": 1.809402585029602, "learning_rate": 3.0583734928462937e-06, "loss": 0.8958, "step": 38670 }, { "epoch": 0.7035413521526401, "grad_norm": 1.7938590049743652, "learning_rate": 3.056498340490165e-06, "loss": 0.8574, "step": 38680 }, { "epoch": 0.7037232397824624, "grad_norm": 1.8578448295593262, "learning_rate": 3.054623188134036e-06, "loss": 0.8991, "step": 38690 }, { "epoch": 0.7039051274122847, "grad_norm": 2.0084869861602783, "learning_rate": 3.0527480357779073e-06, "loss": 0.9218, "step": 38700 }, { "epoch": 0.704087015042107, "grad_norm": 2.100285291671753, "learning_rate": 3.0508728834217782e-06, "loss": 0.9776, "step": 38710 }, { "epoch": 0.7042689026719293, "grad_norm": 1.9038145542144775, "learning_rate": 3.048997731065649e-06, "loss": 0.9565, "step": 38720 }, { "epoch": 0.7044507903017516, "grad_norm": 1.915717601776123, "learning_rate": 3.0471225787095205e-06, "loss": 0.9412, "step": 38730 }, { "epoch": 0.7046326779315739, "grad_norm": 1.7140101194381714, "learning_rate": 3.0452474263533914e-06, "loss": 0.918, "step": 38740 }, { "epoch": 0.7048145655613962, "grad_norm": 1.7961276769638062, "learning_rate": 3.0433722739972628e-06, "loss": 0.9603, "step": 38750 }, { "epoch": 0.7049964531912185, "grad_norm": 2.013453245162964, "learning_rate": 3.0414971216411337e-06, "loss": 0.947, "step": 38760 }, { "epoch": 0.7051783408210408, "grad_norm": 1.8846160173416138, "learning_rate": 3.039621969285005e-06, "loss": 0.868, "step": 38770 }, { "epoch": 0.7053602284508631, "grad_norm": 1.7218619585037231, "learning_rate": 3.037746816928876e-06, "loss": 0.9549, "step": 38780 }, { "epoch": 0.7055421160806854, "grad_norm": 2.0393569469451904, "learning_rate": 3.0358716645727465e-06, "loss": 0.9386, "step": 38790 }, { "epoch": 0.7057240037105077, "grad_norm": 1.7464286088943481, "learning_rate": 3.033996512216618e-06, "loss": 0.9048, "step": 38800 }, { "epoch": 0.70590589134033, "grad_norm": 1.4702826738357544, "learning_rate": 3.0321213598604887e-06, "loss": 0.8605, "step": 38810 }, { "epoch": 0.7060877789701523, "grad_norm": 1.7789815664291382, "learning_rate": 3.0302462075043597e-06, "loss": 0.9089, "step": 38820 }, { "epoch": 0.7062696665999746, "grad_norm": 1.7770309448242188, "learning_rate": 3.028371055148231e-06, "loss": 0.9569, "step": 38830 }, { "epoch": 0.7064515542297969, "grad_norm": 1.8225973844528198, "learning_rate": 3.026495902792102e-06, "loss": 0.9344, "step": 38840 }, { "epoch": 0.7066334418596192, "grad_norm": 1.4480127096176147, "learning_rate": 3.0246207504359733e-06, "loss": 0.8863, "step": 38850 }, { "epoch": 0.7068153294894414, "grad_norm": 1.8408502340316772, "learning_rate": 3.022745598079844e-06, "loss": 0.9621, "step": 38860 }, { "epoch": 0.7069972171192637, "grad_norm": 1.827304482460022, "learning_rate": 3.0208704457237155e-06, "loss": 0.8994, "step": 38870 }, { "epoch": 0.707179104749086, "grad_norm": 1.346479058265686, "learning_rate": 3.0189952933675865e-06, "loss": 0.8821, "step": 38880 }, { "epoch": 0.7073609923789083, "grad_norm": 1.873375415802002, "learning_rate": 3.0171201410114574e-06, "loss": 0.9455, "step": 38890 }, { "epoch": 0.7075428800087306, "grad_norm": 1.7255735397338867, "learning_rate": 3.0152449886553287e-06, "loss": 0.8746, "step": 38900 }, { "epoch": 0.7077247676385529, "grad_norm": 1.7998709678649902, "learning_rate": 3.0133698362991996e-06, "loss": 0.8684, "step": 38910 }, { "epoch": 0.7079066552683752, "grad_norm": 1.7421581745147705, "learning_rate": 3.011494683943071e-06, "loss": 0.8965, "step": 38920 }, { "epoch": 0.7080885428981974, "grad_norm": 1.829933762550354, "learning_rate": 3.0096195315869415e-06, "loss": 0.9061, "step": 38930 }, { "epoch": 0.7082704305280197, "grad_norm": 1.4395661354064941, "learning_rate": 3.0077443792308124e-06, "loss": 0.9326, "step": 38940 }, { "epoch": 0.708452318157842, "grad_norm": 1.8470537662506104, "learning_rate": 3.0058692268746838e-06, "loss": 0.984, "step": 38950 }, { "epoch": 0.7086342057876643, "grad_norm": 1.832635521888733, "learning_rate": 3.0039940745185547e-06, "loss": 0.9317, "step": 38960 }, { "epoch": 0.7088160934174866, "grad_norm": 1.680385708808899, "learning_rate": 3.002118922162426e-06, "loss": 0.9328, "step": 38970 }, { "epoch": 0.7089979810473089, "grad_norm": 1.9391248226165771, "learning_rate": 3.000243769806297e-06, "loss": 0.8929, "step": 38980 }, { "epoch": 0.7091798686771312, "grad_norm": 1.5152174234390259, "learning_rate": 2.998368617450168e-06, "loss": 0.8667, "step": 38990 }, { "epoch": 0.7093617563069535, "grad_norm": 1.6641680002212524, "learning_rate": 2.9964934650940392e-06, "loss": 0.8917, "step": 39000 }, { "epoch": 0.7095436439367758, "grad_norm": 1.7776756286621094, "learning_rate": 2.99461831273791e-06, "loss": 0.8779, "step": 39010 }, { "epoch": 0.7097255315665981, "grad_norm": 1.768468976020813, "learning_rate": 2.9927431603817815e-06, "loss": 0.9602, "step": 39020 }, { "epoch": 0.7099074191964204, "grad_norm": 1.7321580648422241, "learning_rate": 2.9908680080256524e-06, "loss": 0.9105, "step": 39030 }, { "epoch": 0.7100893068262427, "grad_norm": 1.3769457340240479, "learning_rate": 2.9889928556695238e-06, "loss": 0.942, "step": 39040 }, { "epoch": 0.710271194456065, "grad_norm": 1.7495852708816528, "learning_rate": 2.9871177033133947e-06, "loss": 0.9224, "step": 39050 }, { "epoch": 0.7104530820858873, "grad_norm": 6.844880104064941, "learning_rate": 2.9852425509572656e-06, "loss": 0.8657, "step": 39060 }, { "epoch": 0.7106349697157096, "grad_norm": 1.9384958744049072, "learning_rate": 2.9833673986011365e-06, "loss": 0.9174, "step": 39070 }, { "epoch": 0.7108168573455319, "grad_norm": 2.0016157627105713, "learning_rate": 2.9814922462450074e-06, "loss": 0.9547, "step": 39080 }, { "epoch": 0.7109987449753542, "grad_norm": 1.8984565734863281, "learning_rate": 2.9796170938888784e-06, "loss": 0.9001, "step": 39090 }, { "epoch": 0.7111806326051765, "grad_norm": 1.7250792980194092, "learning_rate": 2.9777419415327497e-06, "loss": 0.9069, "step": 39100 }, { "epoch": 0.7113625202349988, "grad_norm": 2.013827085494995, "learning_rate": 2.9758667891766206e-06, "loss": 0.9005, "step": 39110 }, { "epoch": 0.7115444078648211, "grad_norm": 1.6951851844787598, "learning_rate": 2.973991636820492e-06, "loss": 0.9335, "step": 39120 }, { "epoch": 0.7117262954946434, "grad_norm": 1.5578699111938477, "learning_rate": 2.972116484464363e-06, "loss": 0.9325, "step": 39130 }, { "epoch": 0.7119081831244657, "grad_norm": 1.796589732170105, "learning_rate": 2.9702413321082343e-06, "loss": 0.8803, "step": 39140 }, { "epoch": 0.712090070754288, "grad_norm": 1.968949317932129, "learning_rate": 2.968366179752105e-06, "loss": 0.9159, "step": 39150 }, { "epoch": 0.7122719583841103, "grad_norm": 1.9323244094848633, "learning_rate": 2.966491027395976e-06, "loss": 0.9034, "step": 39160 }, { "epoch": 0.7124538460139326, "grad_norm": 1.820145845413208, "learning_rate": 2.9646158750398474e-06, "loss": 0.958, "step": 39170 }, { "epoch": 0.7126357336437549, "grad_norm": 1.7012977600097656, "learning_rate": 2.9627407226837184e-06, "loss": 0.9327, "step": 39180 }, { "epoch": 0.7128176212735772, "grad_norm": 2.155755043029785, "learning_rate": 2.9608655703275897e-06, "loss": 0.9096, "step": 39190 }, { "epoch": 0.7129995089033995, "grad_norm": 1.8512717485427856, "learning_rate": 2.9589904179714606e-06, "loss": 0.9056, "step": 39200 }, { "epoch": 0.7131813965332218, "grad_norm": 1.7218918800354004, "learning_rate": 2.957115265615331e-06, "loss": 0.9321, "step": 39210 }, { "epoch": 0.7133632841630441, "grad_norm": 1.857317328453064, "learning_rate": 2.9552401132592025e-06, "loss": 0.9207, "step": 39220 }, { "epoch": 0.7135451717928664, "grad_norm": 1.94147527217865, "learning_rate": 2.9533649609030734e-06, "loss": 0.8887, "step": 39230 }, { "epoch": 0.7137270594226887, "grad_norm": 1.5865204334259033, "learning_rate": 2.9514898085469448e-06, "loss": 0.8963, "step": 39240 }, { "epoch": 0.713908947052511, "grad_norm": 1.8394529819488525, "learning_rate": 2.9496146561908157e-06, "loss": 0.8976, "step": 39250 }, { "epoch": 0.7140908346823333, "grad_norm": 1.8867342472076416, "learning_rate": 2.9477395038346866e-06, "loss": 0.9323, "step": 39260 }, { "epoch": 0.7142727223121556, "grad_norm": 1.8464454412460327, "learning_rate": 2.945864351478558e-06, "loss": 0.8801, "step": 39270 }, { "epoch": 0.7144546099419778, "grad_norm": 2.017324447631836, "learning_rate": 2.943989199122429e-06, "loss": 0.8537, "step": 39280 }, { "epoch": 0.7146364975718001, "grad_norm": 1.9291441440582275, "learning_rate": 2.9421140467663e-06, "loss": 0.9151, "step": 39290 }, { "epoch": 0.7148183852016224, "grad_norm": 1.6303460597991943, "learning_rate": 2.940238894410171e-06, "loss": 0.8785, "step": 39300 }, { "epoch": 0.7150002728314447, "grad_norm": 1.6513794660568237, "learning_rate": 2.9383637420540425e-06, "loss": 0.9317, "step": 39310 }, { "epoch": 0.715182160461267, "grad_norm": 1.6161624193191528, "learning_rate": 2.9364885896979134e-06, "loss": 0.9419, "step": 39320 }, { "epoch": 0.7153640480910893, "grad_norm": 1.6376582384109497, "learning_rate": 2.9346134373417843e-06, "loss": 0.8719, "step": 39330 }, { "epoch": 0.7155459357209116, "grad_norm": 1.731296420097351, "learning_rate": 2.9327382849856552e-06, "loss": 0.8913, "step": 39340 }, { "epoch": 0.7157278233507339, "grad_norm": 1.6924587488174438, "learning_rate": 2.930863132629526e-06, "loss": 0.9403, "step": 39350 }, { "epoch": 0.7159097109805562, "grad_norm": 1.5644476413726807, "learning_rate": 2.928987980273397e-06, "loss": 0.9073, "step": 39360 }, { "epoch": 0.7160915986103785, "grad_norm": 1.8443293571472168, "learning_rate": 2.9271128279172684e-06, "loss": 0.9171, "step": 39370 }, { "epoch": 0.7162734862402008, "grad_norm": 1.9595967531204224, "learning_rate": 2.9252376755611394e-06, "loss": 0.9229, "step": 39380 }, { "epoch": 0.7164553738700231, "grad_norm": 1.682437777519226, "learning_rate": 2.9233625232050107e-06, "loss": 0.8587, "step": 39390 }, { "epoch": 0.7166372614998454, "grad_norm": 1.793497920036316, "learning_rate": 2.9214873708488816e-06, "loss": 0.9012, "step": 39400 }, { "epoch": 0.7168191491296677, "grad_norm": 1.8954229354858398, "learning_rate": 2.919612218492753e-06, "loss": 0.8818, "step": 39410 }, { "epoch": 0.71700103675949, "grad_norm": 1.785857081413269, "learning_rate": 2.917737066136624e-06, "loss": 0.8971, "step": 39420 }, { "epoch": 0.7171829243893123, "grad_norm": 1.708052158355713, "learning_rate": 2.915861913780495e-06, "loss": 0.8647, "step": 39430 }, { "epoch": 0.7173648120191346, "grad_norm": 1.353613257408142, "learning_rate": 2.913986761424366e-06, "loss": 0.9049, "step": 39440 }, { "epoch": 0.7175466996489569, "grad_norm": 1.7237468957901, "learning_rate": 2.912111609068237e-06, "loss": 0.9415, "step": 39450 }, { "epoch": 0.7177285872787792, "grad_norm": 1.6817879676818848, "learning_rate": 2.9102364567121084e-06, "loss": 0.9333, "step": 39460 }, { "epoch": 0.7179104749086015, "grad_norm": 1.6707212924957275, "learning_rate": 2.9083613043559794e-06, "loss": 0.9328, "step": 39470 }, { "epoch": 0.7180923625384238, "grad_norm": 1.645775318145752, "learning_rate": 2.90648615199985e-06, "loss": 0.8886, "step": 39480 }, { "epoch": 0.7182742501682461, "grad_norm": 1.4391015768051147, "learning_rate": 2.904610999643721e-06, "loss": 0.8632, "step": 39490 }, { "epoch": 0.7184561377980684, "grad_norm": 4.017463207244873, "learning_rate": 2.902735847287592e-06, "loss": 0.9116, "step": 39500 }, { "epoch": 0.7186380254278907, "grad_norm": 1.8768538236618042, "learning_rate": 2.9008606949314635e-06, "loss": 0.9386, "step": 39510 }, { "epoch": 0.718819913057713, "grad_norm": 1.7421550750732422, "learning_rate": 2.8989855425753344e-06, "loss": 0.9187, "step": 39520 }, { "epoch": 0.7190018006875353, "grad_norm": 1.761550784111023, "learning_rate": 2.8971103902192053e-06, "loss": 0.8974, "step": 39530 }, { "epoch": 0.7191836883173576, "grad_norm": 1.5770529508590698, "learning_rate": 2.8952352378630767e-06, "loss": 0.9303, "step": 39540 }, { "epoch": 0.7193655759471799, "grad_norm": 1.6169497966766357, "learning_rate": 2.8933600855069476e-06, "loss": 0.872, "step": 39550 }, { "epoch": 0.7195474635770022, "grad_norm": 1.6402531862258911, "learning_rate": 2.891484933150819e-06, "loss": 0.8822, "step": 39560 }, { "epoch": 0.7197293512068245, "grad_norm": 1.9320067167282104, "learning_rate": 2.88960978079469e-06, "loss": 0.9388, "step": 39570 }, { "epoch": 0.7199112388366468, "grad_norm": 1.5705825090408325, "learning_rate": 2.887734628438561e-06, "loss": 0.8883, "step": 39580 }, { "epoch": 0.7200931264664691, "grad_norm": 1.9071835279464722, "learning_rate": 2.885859476082432e-06, "loss": 0.897, "step": 39590 }, { "epoch": 0.7202750140962914, "grad_norm": 1.78468918800354, "learning_rate": 2.883984323726303e-06, "loss": 0.906, "step": 39600 }, { "epoch": 0.7204569017261137, "grad_norm": 1.7948132753372192, "learning_rate": 2.8821091713701744e-06, "loss": 0.9092, "step": 39610 }, { "epoch": 0.720638789355936, "grad_norm": 1.6585773229599, "learning_rate": 2.880234019014045e-06, "loss": 0.9306, "step": 39620 }, { "epoch": 0.7208206769857582, "grad_norm": 1.781528353691101, "learning_rate": 2.878358866657916e-06, "loss": 0.8766, "step": 39630 }, { "epoch": 0.7210025646155805, "grad_norm": 1.631090521812439, "learning_rate": 2.876483714301787e-06, "loss": 0.8799, "step": 39640 }, { "epoch": 0.7211844522454028, "grad_norm": 1.7867375612258911, "learning_rate": 2.874608561945658e-06, "loss": 0.9594, "step": 39650 }, { "epoch": 0.7213663398752251, "grad_norm": 1.9951767921447754, "learning_rate": 2.8727334095895294e-06, "loss": 0.9424, "step": 39660 }, { "epoch": 0.7215482275050474, "grad_norm": 1.754673957824707, "learning_rate": 2.8708582572334003e-06, "loss": 0.8837, "step": 39670 }, { "epoch": 0.7217301151348697, "grad_norm": 1.8818414211273193, "learning_rate": 2.8689831048772717e-06, "loss": 0.8868, "step": 39680 }, { "epoch": 0.7219120027646919, "grad_norm": 1.726767659187317, "learning_rate": 2.8671079525211426e-06, "loss": 0.9368, "step": 39690 }, { "epoch": 0.7220938903945142, "grad_norm": 1.6542056798934937, "learning_rate": 2.8652328001650135e-06, "loss": 0.873, "step": 39700 }, { "epoch": 0.7222757780243365, "grad_norm": 1.6072378158569336, "learning_rate": 2.863357647808885e-06, "loss": 0.8842, "step": 39710 }, { "epoch": 0.7224576656541588, "grad_norm": 1.7366503477096558, "learning_rate": 2.861482495452756e-06, "loss": 0.9239, "step": 39720 }, { "epoch": 0.7226395532839811, "grad_norm": 1.810336709022522, "learning_rate": 2.859607343096627e-06, "loss": 1.0057, "step": 39730 }, { "epoch": 0.7228214409138034, "grad_norm": 1.7609331607818604, "learning_rate": 2.857732190740498e-06, "loss": 0.9095, "step": 39740 }, { "epoch": 0.7230033285436257, "grad_norm": 1.4538705348968506, "learning_rate": 2.8558570383843694e-06, "loss": 0.8653, "step": 39750 }, { "epoch": 0.723185216173448, "grad_norm": 1.8064906597137451, "learning_rate": 2.85398188602824e-06, "loss": 0.9202, "step": 39760 }, { "epoch": 0.7233671038032703, "grad_norm": 1.5089035034179688, "learning_rate": 2.852106733672111e-06, "loss": 0.8956, "step": 39770 }, { "epoch": 0.7235489914330926, "grad_norm": 1.7800097465515137, "learning_rate": 2.850231581315982e-06, "loss": 0.9631, "step": 39780 }, { "epoch": 0.7237308790629149, "grad_norm": 1.783300757408142, "learning_rate": 2.848356428959853e-06, "loss": 0.9019, "step": 39790 }, { "epoch": 0.7239127666927372, "grad_norm": 1.7428406476974487, "learning_rate": 2.846481276603724e-06, "loss": 0.8807, "step": 39800 }, { "epoch": 0.7240946543225595, "grad_norm": 1.5030767917633057, "learning_rate": 2.8446061242475954e-06, "loss": 0.905, "step": 39810 }, { "epoch": 0.7242765419523818, "grad_norm": 1.920192003250122, "learning_rate": 2.8427309718914663e-06, "loss": 0.9172, "step": 39820 }, { "epoch": 0.7244584295822041, "grad_norm": 1.9215575456619263, "learning_rate": 2.8408558195353377e-06, "loss": 0.9419, "step": 39830 }, { "epoch": 0.7246403172120264, "grad_norm": 1.5755795240402222, "learning_rate": 2.8389806671792086e-06, "loss": 0.8972, "step": 39840 }, { "epoch": 0.7248222048418487, "grad_norm": 1.7251877784729004, "learning_rate": 2.83710551482308e-06, "loss": 0.905, "step": 39850 }, { "epoch": 0.725004092471671, "grad_norm": 1.6891934871673584, "learning_rate": 2.835230362466951e-06, "loss": 0.9127, "step": 39860 }, { "epoch": 0.7251859801014933, "grad_norm": 1.8949593305587769, "learning_rate": 2.8333552101108218e-06, "loss": 0.8796, "step": 39870 }, { "epoch": 0.7253678677313156, "grad_norm": 1.7994686365127563, "learning_rate": 2.831480057754693e-06, "loss": 0.8801, "step": 39880 }, { "epoch": 0.7255497553611379, "grad_norm": 2.127346992492676, "learning_rate": 2.829604905398564e-06, "loss": 0.8761, "step": 39890 }, { "epoch": 0.7257316429909602, "grad_norm": 2.007687568664551, "learning_rate": 2.8277297530424345e-06, "loss": 0.8877, "step": 39900 }, { "epoch": 0.7259135306207825, "grad_norm": 1.9667831659317017, "learning_rate": 2.825854600686306e-06, "loss": 0.9539, "step": 39910 }, { "epoch": 0.7260954182506048, "grad_norm": 1.777165174484253, "learning_rate": 2.823979448330177e-06, "loss": 0.8385, "step": 39920 }, { "epoch": 0.726277305880427, "grad_norm": 1.37406325340271, "learning_rate": 2.822104295974048e-06, "loss": 0.871, "step": 39930 }, { "epoch": 0.7264591935102493, "grad_norm": 1.9779056310653687, "learning_rate": 2.820229143617919e-06, "loss": 0.8948, "step": 39940 }, { "epoch": 0.7266410811400716, "grad_norm": 1.9791241884231567, "learning_rate": 2.8183539912617904e-06, "loss": 0.9204, "step": 39950 }, { "epoch": 0.7268229687698939, "grad_norm": 1.8525265455245972, "learning_rate": 2.8164788389056613e-06, "loss": 0.87, "step": 39960 }, { "epoch": 0.7270048563997162, "grad_norm": 2.0277316570281982, "learning_rate": 2.8146036865495323e-06, "loss": 0.8946, "step": 39970 }, { "epoch": 0.7271867440295385, "grad_norm": 1.7675131559371948, "learning_rate": 2.8127285341934036e-06, "loss": 0.9078, "step": 39980 }, { "epoch": 0.7273686316593608, "grad_norm": 1.9046142101287842, "learning_rate": 2.8108533818372745e-06, "loss": 0.9544, "step": 39990 }, { "epoch": 0.7275505192891831, "grad_norm": 1.9923175573349, "learning_rate": 2.808978229481146e-06, "loss": 0.903, "step": 40000 }, { "epoch": 0.7277324069190054, "grad_norm": 1.9380340576171875, "learning_rate": 2.807103077125017e-06, "loss": 0.9251, "step": 40010 }, { "epoch": 0.7279142945488277, "grad_norm": 1.711450219154358, "learning_rate": 2.805227924768888e-06, "loss": 0.9378, "step": 40020 }, { "epoch": 0.72809618217865, "grad_norm": 1.8682087659835815, "learning_rate": 2.8033527724127586e-06, "loss": 0.9152, "step": 40030 }, { "epoch": 0.7282780698084723, "grad_norm": 1.8580515384674072, "learning_rate": 2.8014776200566296e-06, "loss": 0.9323, "step": 40040 }, { "epoch": 0.7284599574382946, "grad_norm": 1.677230715751648, "learning_rate": 2.799602467700501e-06, "loss": 0.8807, "step": 40050 }, { "epoch": 0.7286418450681169, "grad_norm": 1.9025514125823975, "learning_rate": 2.797727315344372e-06, "loss": 0.9577, "step": 40060 }, { "epoch": 0.7288237326979392, "grad_norm": 1.7173291444778442, "learning_rate": 2.7958521629882428e-06, "loss": 0.8754, "step": 40070 }, { "epoch": 0.7290056203277615, "grad_norm": 1.9744226932525635, "learning_rate": 2.793977010632114e-06, "loss": 0.9214, "step": 40080 }, { "epoch": 0.7291875079575838, "grad_norm": 1.7740237712860107, "learning_rate": 2.792101858275985e-06, "loss": 0.8953, "step": 40090 }, { "epoch": 0.7293693955874061, "grad_norm": 1.5919040441513062, "learning_rate": 2.7902267059198564e-06, "loss": 0.9064, "step": 40100 }, { "epoch": 0.7295512832172284, "grad_norm": 2.0218799114227295, "learning_rate": 2.7883515535637273e-06, "loss": 0.9447, "step": 40110 }, { "epoch": 0.7297331708470507, "grad_norm": 1.6308109760284424, "learning_rate": 2.7864764012075986e-06, "loss": 0.8546, "step": 40120 }, { "epoch": 0.729915058476873, "grad_norm": 1.8326921463012695, "learning_rate": 2.7846012488514696e-06, "loss": 0.8899, "step": 40130 }, { "epoch": 0.7300969461066953, "grad_norm": 1.8055962324142456, "learning_rate": 2.7827260964953405e-06, "loss": 0.9408, "step": 40140 }, { "epoch": 0.7302788337365176, "grad_norm": 1.6050397157669067, "learning_rate": 2.780850944139212e-06, "loss": 0.9027, "step": 40150 }, { "epoch": 0.7304607213663399, "grad_norm": 1.6172078847885132, "learning_rate": 2.7789757917830828e-06, "loss": 0.8889, "step": 40160 }, { "epoch": 0.7306426089961622, "grad_norm": 1.881604552268982, "learning_rate": 2.7771006394269533e-06, "loss": 0.9624, "step": 40170 }, { "epoch": 0.7308244966259845, "grad_norm": 1.5730267763137817, "learning_rate": 2.7752254870708246e-06, "loss": 0.9103, "step": 40180 }, { "epoch": 0.7310063842558068, "grad_norm": 1.6301779747009277, "learning_rate": 2.7733503347146955e-06, "loss": 0.9253, "step": 40190 }, { "epoch": 0.7311882718856291, "grad_norm": 1.9564887285232544, "learning_rate": 2.771475182358567e-06, "loss": 0.9039, "step": 40200 }, { "epoch": 0.7313701595154514, "grad_norm": 1.5323816537857056, "learning_rate": 2.769600030002438e-06, "loss": 0.8738, "step": 40210 }, { "epoch": 0.7315520471452737, "grad_norm": 1.8194167613983154, "learning_rate": 2.767724877646309e-06, "loss": 0.8993, "step": 40220 }, { "epoch": 0.731733934775096, "grad_norm": 1.9931113719940186, "learning_rate": 2.76584972529018e-06, "loss": 0.9386, "step": 40230 }, { "epoch": 0.7319158224049183, "grad_norm": 1.5983598232269287, "learning_rate": 2.763974572934051e-06, "loss": 0.9137, "step": 40240 }, { "epoch": 0.7320977100347406, "grad_norm": 1.727988839149475, "learning_rate": 2.7620994205779223e-06, "loss": 0.9076, "step": 40250 }, { "epoch": 0.7322795976645629, "grad_norm": 1.59010648727417, "learning_rate": 2.7602242682217932e-06, "loss": 0.8753, "step": 40260 }, { "epoch": 0.7324614852943852, "grad_norm": 1.5284749269485474, "learning_rate": 2.7583491158656646e-06, "loss": 0.8684, "step": 40270 }, { "epoch": 0.7326433729242074, "grad_norm": 1.9912151098251343, "learning_rate": 2.7564739635095355e-06, "loss": 0.9436, "step": 40280 }, { "epoch": 0.7328252605540297, "grad_norm": 1.5803437232971191, "learning_rate": 2.754598811153407e-06, "loss": 0.8986, "step": 40290 }, { "epoch": 0.733007148183852, "grad_norm": 1.970643401145935, "learning_rate": 2.7527236587972778e-06, "loss": 0.9301, "step": 40300 }, { "epoch": 0.7331890358136743, "grad_norm": 1.9903005361557007, "learning_rate": 2.7508485064411483e-06, "loss": 0.9167, "step": 40310 }, { "epoch": 0.7333709234434966, "grad_norm": 1.756358027458191, "learning_rate": 2.7489733540850196e-06, "loss": 0.87, "step": 40320 }, { "epoch": 0.7335528110733189, "grad_norm": 1.7039042711257935, "learning_rate": 2.7470982017288906e-06, "loss": 0.9381, "step": 40330 }, { "epoch": 0.7337346987031412, "grad_norm": 2.0224292278289795, "learning_rate": 2.7452230493727615e-06, "loss": 0.9081, "step": 40340 }, { "epoch": 0.7339165863329635, "grad_norm": 1.7248506546020508, "learning_rate": 2.743347897016633e-06, "loss": 0.8585, "step": 40350 }, { "epoch": 0.7340984739627858, "grad_norm": 1.725488543510437, "learning_rate": 2.7414727446605037e-06, "loss": 0.8835, "step": 40360 }, { "epoch": 0.7342803615926081, "grad_norm": 1.6959264278411865, "learning_rate": 2.739597592304375e-06, "loss": 0.8887, "step": 40370 }, { "epoch": 0.7344622492224304, "grad_norm": 1.6208345890045166, "learning_rate": 2.737722439948246e-06, "loss": 0.8965, "step": 40380 }, { "epoch": 0.7346441368522527, "grad_norm": 1.6114270687103271, "learning_rate": 2.7358472875921174e-06, "loss": 0.8747, "step": 40390 }, { "epoch": 0.734826024482075, "grad_norm": 1.483841061592102, "learning_rate": 2.7339721352359883e-06, "loss": 0.9243, "step": 40400 }, { "epoch": 0.7350079121118973, "grad_norm": 1.7323224544525146, "learning_rate": 2.732096982879859e-06, "loss": 0.9208, "step": 40410 }, { "epoch": 0.7351897997417196, "grad_norm": 1.8800326585769653, "learning_rate": 2.7302218305237305e-06, "loss": 0.8965, "step": 40420 }, { "epoch": 0.7353716873715419, "grad_norm": 1.7635490894317627, "learning_rate": 2.7283466781676015e-06, "loss": 0.8817, "step": 40430 }, { "epoch": 0.7355535750013642, "grad_norm": 2.0497331619262695, "learning_rate": 2.726471525811473e-06, "loss": 0.9103, "step": 40440 }, { "epoch": 0.7357354626311865, "grad_norm": 2.1919009685516357, "learning_rate": 2.7245963734553433e-06, "loss": 0.9348, "step": 40450 }, { "epoch": 0.7359173502610088, "grad_norm": 1.6016602516174316, "learning_rate": 2.7227212210992142e-06, "loss": 0.8656, "step": 40460 }, { "epoch": 0.7360992378908311, "grad_norm": 1.7455639839172363, "learning_rate": 2.7208460687430856e-06, "loss": 0.9106, "step": 40470 }, { "epoch": 0.7362811255206534, "grad_norm": 1.5791932344436646, "learning_rate": 2.7189709163869565e-06, "loss": 0.888, "step": 40480 }, { "epoch": 0.7364630131504757, "grad_norm": 1.9649479389190674, "learning_rate": 2.717095764030828e-06, "loss": 0.9363, "step": 40490 }, { "epoch": 0.736644900780298, "grad_norm": 1.9192004203796387, "learning_rate": 2.7152206116746988e-06, "loss": 0.9117, "step": 40500 }, { "epoch": 0.7368267884101203, "grad_norm": 1.86349356174469, "learning_rate": 2.7133454593185697e-06, "loss": 0.9539, "step": 40510 }, { "epoch": 0.7370086760399426, "grad_norm": 1.8399274349212646, "learning_rate": 2.711470306962441e-06, "loss": 0.9447, "step": 40520 }, { "epoch": 0.7371905636697648, "grad_norm": 1.7847198247909546, "learning_rate": 2.709595154606312e-06, "loss": 0.9325, "step": 40530 }, { "epoch": 0.7373724512995871, "grad_norm": 1.7283507585525513, "learning_rate": 2.7077200022501833e-06, "loss": 0.9713, "step": 40540 }, { "epoch": 0.7375543389294094, "grad_norm": 1.38370680809021, "learning_rate": 2.7058448498940542e-06, "loss": 0.8867, "step": 40550 }, { "epoch": 0.7377362265592317, "grad_norm": 1.532311201095581, "learning_rate": 2.7039696975379256e-06, "loss": 0.8955, "step": 40560 }, { "epoch": 0.737918114189054, "grad_norm": 1.795778512954712, "learning_rate": 2.7020945451817965e-06, "loss": 0.9562, "step": 40570 }, { "epoch": 0.7381000018188763, "grad_norm": 1.8748584985733032, "learning_rate": 2.700219392825667e-06, "loss": 0.8994, "step": 40580 }, { "epoch": 0.7382818894486985, "grad_norm": 1.947282314300537, "learning_rate": 2.698344240469538e-06, "loss": 0.9064, "step": 40590 }, { "epoch": 0.7384637770785208, "grad_norm": 1.8637973070144653, "learning_rate": 2.6964690881134093e-06, "loss": 0.9006, "step": 40600 }, { "epoch": 0.7386456647083431, "grad_norm": 1.9060519933700562, "learning_rate": 2.69459393575728e-06, "loss": 0.9281, "step": 40610 }, { "epoch": 0.7388275523381654, "grad_norm": 1.5365848541259766, "learning_rate": 2.6927187834011515e-06, "loss": 0.8343, "step": 40620 }, { "epoch": 0.7390094399679877, "grad_norm": 2.012528657913208, "learning_rate": 2.6908436310450225e-06, "loss": 0.8929, "step": 40630 }, { "epoch": 0.73919132759781, "grad_norm": 1.8086872100830078, "learning_rate": 2.688968478688894e-06, "loss": 0.9004, "step": 40640 }, { "epoch": 0.7393732152276323, "grad_norm": 1.77633798122406, "learning_rate": 2.6870933263327647e-06, "loss": 0.9155, "step": 40650 }, { "epoch": 0.7395551028574546, "grad_norm": 1.6024633646011353, "learning_rate": 2.685218173976636e-06, "loss": 0.887, "step": 40660 }, { "epoch": 0.7397369904872769, "grad_norm": 1.8593266010284424, "learning_rate": 2.683343021620507e-06, "loss": 0.916, "step": 40670 }, { "epoch": 0.7399188781170992, "grad_norm": 1.7034687995910645, "learning_rate": 2.681467869264378e-06, "loss": 0.9274, "step": 40680 }, { "epoch": 0.7401007657469215, "grad_norm": 1.5720094442367554, "learning_rate": 2.6795927169082493e-06, "loss": 0.8399, "step": 40690 }, { "epoch": 0.7402826533767438, "grad_norm": 1.847389817237854, "learning_rate": 2.67771756455212e-06, "loss": 0.912, "step": 40700 }, { "epoch": 0.7404645410065661, "grad_norm": 2.0109097957611084, "learning_rate": 2.6758424121959915e-06, "loss": 0.9045, "step": 40710 }, { "epoch": 0.7406464286363884, "grad_norm": 1.889481782913208, "learning_rate": 2.673967259839862e-06, "loss": 0.9518, "step": 40720 }, { "epoch": 0.7408283162662107, "grad_norm": 1.7611159086227417, "learning_rate": 2.672092107483733e-06, "loss": 0.9261, "step": 40730 }, { "epoch": 0.741010203896033, "grad_norm": 1.6470115184783936, "learning_rate": 2.6702169551276043e-06, "loss": 0.9234, "step": 40740 }, { "epoch": 0.7411920915258553, "grad_norm": 1.7276546955108643, "learning_rate": 2.6683418027714752e-06, "loss": 0.9342, "step": 40750 }, { "epoch": 0.7413739791556776, "grad_norm": 1.6235677003860474, "learning_rate": 2.666466650415346e-06, "loss": 0.9233, "step": 40760 }, { "epoch": 0.7415558667854999, "grad_norm": 1.6541471481323242, "learning_rate": 2.6645914980592175e-06, "loss": 0.8966, "step": 40770 }, { "epoch": 0.7417377544153222, "grad_norm": 2.1836297512054443, "learning_rate": 2.6627163457030884e-06, "loss": 0.8952, "step": 40780 }, { "epoch": 0.7419196420451445, "grad_norm": 1.4739614725112915, "learning_rate": 2.6608411933469598e-06, "loss": 0.9226, "step": 40790 }, { "epoch": 0.7421015296749668, "grad_norm": 1.9336802959442139, "learning_rate": 2.6589660409908307e-06, "loss": 0.9547, "step": 40800 }, { "epoch": 0.7422834173047891, "grad_norm": 2.0044543743133545, "learning_rate": 2.657090888634702e-06, "loss": 0.9482, "step": 40810 }, { "epoch": 0.7424653049346114, "grad_norm": 1.8528282642364502, "learning_rate": 2.655215736278573e-06, "loss": 0.8727, "step": 40820 }, { "epoch": 0.7426471925644337, "grad_norm": 1.72135591506958, "learning_rate": 2.6533405839224443e-06, "loss": 0.8645, "step": 40830 }, { "epoch": 0.742829080194256, "grad_norm": 1.8336480855941772, "learning_rate": 2.6514654315663152e-06, "loss": 0.9453, "step": 40840 }, { "epoch": 0.7430109678240783, "grad_norm": 1.752600073814392, "learning_rate": 2.649590279210186e-06, "loss": 0.9429, "step": 40850 }, { "epoch": 0.7431928554539006, "grad_norm": 1.8927382230758667, "learning_rate": 2.6477151268540566e-06, "loss": 0.9094, "step": 40860 }, { "epoch": 0.7433747430837229, "grad_norm": 1.615052580833435, "learning_rate": 2.645839974497928e-06, "loss": 0.8541, "step": 40870 }, { "epoch": 0.7435566307135452, "grad_norm": 1.819195032119751, "learning_rate": 2.643964822141799e-06, "loss": 0.8975, "step": 40880 }, { "epoch": 0.7437385183433675, "grad_norm": 1.8146382570266724, "learning_rate": 2.6420896697856703e-06, "loss": 0.8708, "step": 40890 }, { "epoch": 0.7439204059731898, "grad_norm": 1.7582571506500244, "learning_rate": 2.640214517429541e-06, "loss": 0.9707, "step": 40900 }, { "epoch": 0.744102293603012, "grad_norm": 1.5470499992370605, "learning_rate": 2.6383393650734125e-06, "loss": 0.8531, "step": 40910 }, { "epoch": 0.7442841812328344, "grad_norm": 1.769972562789917, "learning_rate": 2.6364642127172835e-06, "loss": 0.897, "step": 40920 }, { "epoch": 0.7444660688626566, "grad_norm": 1.6212055683135986, "learning_rate": 2.6345890603611544e-06, "loss": 0.9174, "step": 40930 }, { "epoch": 0.744647956492479, "grad_norm": 1.7788008451461792, "learning_rate": 2.6327139080050257e-06, "loss": 0.9485, "step": 40940 }, { "epoch": 0.7448298441223012, "grad_norm": 1.7009748220443726, "learning_rate": 2.6308387556488966e-06, "loss": 0.9102, "step": 40950 }, { "epoch": 0.7450117317521235, "grad_norm": 1.977396011352539, "learning_rate": 2.628963603292768e-06, "loss": 0.8992, "step": 40960 }, { "epoch": 0.7451936193819458, "grad_norm": 1.94304621219635, "learning_rate": 2.627088450936639e-06, "loss": 0.8446, "step": 40970 }, { "epoch": 0.7453755070117681, "grad_norm": 1.8245964050292969, "learning_rate": 2.6252132985805103e-06, "loss": 0.9026, "step": 40980 }, { "epoch": 0.7455573946415904, "grad_norm": 1.489411473274231, "learning_rate": 2.623338146224381e-06, "loss": 0.9102, "step": 40990 }, { "epoch": 0.7457392822714127, "grad_norm": 1.6625012159347534, "learning_rate": 2.6214629938682517e-06, "loss": 0.861, "step": 41000 }, { "epoch": 0.745921169901235, "grad_norm": 1.704991102218628, "learning_rate": 2.619587841512123e-06, "loss": 0.8857, "step": 41010 }, { "epoch": 0.7461030575310573, "grad_norm": 1.6752591133117676, "learning_rate": 2.617712689155994e-06, "loss": 0.9109, "step": 41020 }, { "epoch": 0.7462849451608796, "grad_norm": 2.1574864387512207, "learning_rate": 2.615837536799865e-06, "loss": 0.9392, "step": 41030 }, { "epoch": 0.7464668327907019, "grad_norm": 1.777748465538025, "learning_rate": 2.6139623844437362e-06, "loss": 0.9079, "step": 41040 }, { "epoch": 0.7466487204205242, "grad_norm": 1.719100832939148, "learning_rate": 2.612087232087607e-06, "loss": 0.957, "step": 41050 }, { "epoch": 0.7468306080503465, "grad_norm": 1.9337327480316162, "learning_rate": 2.6102120797314785e-06, "loss": 0.9047, "step": 41060 }, { "epoch": 0.7470124956801688, "grad_norm": 1.4162306785583496, "learning_rate": 2.6083369273753494e-06, "loss": 0.8617, "step": 41070 }, { "epoch": 0.7471943833099911, "grad_norm": 1.7374448776245117, "learning_rate": 2.6064617750192208e-06, "loss": 0.9193, "step": 41080 }, { "epoch": 0.7473762709398134, "grad_norm": 1.7036640644073486, "learning_rate": 2.6045866226630917e-06, "loss": 0.886, "step": 41090 }, { "epoch": 0.7475581585696357, "grad_norm": 1.9265984296798706, "learning_rate": 2.6027114703069626e-06, "loss": 0.9217, "step": 41100 }, { "epoch": 0.747740046199458, "grad_norm": 1.95360267162323, "learning_rate": 2.600836317950834e-06, "loss": 0.8978, "step": 41110 }, { "epoch": 0.7479219338292803, "grad_norm": 1.9509252309799194, "learning_rate": 2.598961165594705e-06, "loss": 0.9462, "step": 41120 }, { "epoch": 0.7481038214591026, "grad_norm": 1.658699870109558, "learning_rate": 2.5970860132385762e-06, "loss": 0.8809, "step": 41130 }, { "epoch": 0.7482857090889249, "grad_norm": 2.0344479084014893, "learning_rate": 2.5952108608824467e-06, "loss": 0.9189, "step": 41140 }, { "epoch": 0.7484675967187472, "grad_norm": 1.5683519840240479, "learning_rate": 2.5933357085263176e-06, "loss": 0.8967, "step": 41150 }, { "epoch": 0.7486494843485695, "grad_norm": 1.9873541593551636, "learning_rate": 2.591460556170189e-06, "loss": 0.8879, "step": 41160 }, { "epoch": 0.7488313719783918, "grad_norm": 1.7596672773361206, "learning_rate": 2.58958540381406e-06, "loss": 0.8881, "step": 41170 }, { "epoch": 0.7490132596082141, "grad_norm": 1.8116083145141602, "learning_rate": 2.5877102514579312e-06, "loss": 0.8831, "step": 41180 }, { "epoch": 0.7491951472380364, "grad_norm": 1.447912573814392, "learning_rate": 2.585835099101802e-06, "loss": 0.8457, "step": 41190 }, { "epoch": 0.7493770348678587, "grad_norm": 1.8823761940002441, "learning_rate": 2.583959946745673e-06, "loss": 0.8895, "step": 41200 }, { "epoch": 0.749558922497681, "grad_norm": 2.0123910903930664, "learning_rate": 2.5820847943895444e-06, "loss": 0.8966, "step": 41210 }, { "epoch": 0.7497408101275033, "grad_norm": 1.6193405389785767, "learning_rate": 2.5802096420334154e-06, "loss": 0.8694, "step": 41220 }, { "epoch": 0.7499226977573256, "grad_norm": 1.6276943683624268, "learning_rate": 2.5783344896772867e-06, "loss": 0.9123, "step": 41230 }, { "epoch": 0.7501045853871479, "grad_norm": 1.7284924983978271, "learning_rate": 2.5764593373211576e-06, "loss": 0.9323, "step": 41240 }, { "epoch": 0.7502864730169702, "grad_norm": 1.5596427917480469, "learning_rate": 2.574584184965029e-06, "loss": 0.894, "step": 41250 }, { "epoch": 0.7504683606467925, "grad_norm": 1.882275938987732, "learning_rate": 2.5727090326089e-06, "loss": 0.9333, "step": 41260 }, { "epoch": 0.7506502482766148, "grad_norm": 1.8861663341522217, "learning_rate": 2.5708338802527704e-06, "loss": 0.9142, "step": 41270 }, { "epoch": 0.750832135906437, "grad_norm": 1.8780133724212646, "learning_rate": 2.5689587278966417e-06, "loss": 0.8865, "step": 41280 }, { "epoch": 0.7510140235362593, "grad_norm": 2.0838255882263184, "learning_rate": 2.5670835755405127e-06, "loss": 0.9031, "step": 41290 }, { "epoch": 0.7511959111660816, "grad_norm": 1.9446743726730347, "learning_rate": 2.5652084231843836e-06, "loss": 0.8621, "step": 41300 }, { "epoch": 0.7513777987959039, "grad_norm": 1.870200753211975, "learning_rate": 2.563333270828255e-06, "loss": 0.8999, "step": 41310 }, { "epoch": 0.7515596864257262, "grad_norm": 1.862046241760254, "learning_rate": 2.561458118472126e-06, "loss": 1.0117, "step": 41320 }, { "epoch": 0.7517415740555485, "grad_norm": 1.6889514923095703, "learning_rate": 2.559582966115997e-06, "loss": 0.8802, "step": 41330 }, { "epoch": 0.7519234616853708, "grad_norm": 1.7318564653396606, "learning_rate": 2.557707813759868e-06, "loss": 0.9397, "step": 41340 }, { "epoch": 0.7521053493151931, "grad_norm": 2.5712671279907227, "learning_rate": 2.5558326614037395e-06, "loss": 0.9453, "step": 41350 }, { "epoch": 0.7522872369450153, "grad_norm": 1.5138585567474365, "learning_rate": 2.5539575090476104e-06, "loss": 0.8751, "step": 41360 }, { "epoch": 0.7524691245748376, "grad_norm": 1.674451231956482, "learning_rate": 2.5520823566914813e-06, "loss": 0.9433, "step": 41370 }, { "epoch": 0.7526510122046599, "grad_norm": 1.5435336828231812, "learning_rate": 2.5502072043353527e-06, "loss": 0.8433, "step": 41380 }, { "epoch": 0.7528328998344822, "grad_norm": 1.8863016366958618, "learning_rate": 2.5483320519792236e-06, "loss": 0.9165, "step": 41390 }, { "epoch": 0.7530147874643045, "grad_norm": 1.6064555644989014, "learning_rate": 2.546456899623095e-06, "loss": 0.8868, "step": 41400 }, { "epoch": 0.7531966750941268, "grad_norm": 1.7135251760482788, "learning_rate": 2.5445817472669654e-06, "loss": 0.8968, "step": 41410 }, { "epoch": 0.7533785627239491, "grad_norm": 1.8201032876968384, "learning_rate": 2.5427065949108364e-06, "loss": 0.9203, "step": 41420 }, { "epoch": 0.7535604503537714, "grad_norm": 1.8795796632766724, "learning_rate": 2.5408314425547077e-06, "loss": 0.9124, "step": 41430 }, { "epoch": 0.7537423379835937, "grad_norm": 1.7568587064743042, "learning_rate": 2.5389562901985786e-06, "loss": 0.9065, "step": 41440 }, { "epoch": 0.753924225613416, "grad_norm": 1.8279571533203125, "learning_rate": 2.53708113784245e-06, "loss": 0.9667, "step": 41450 }, { "epoch": 0.7541061132432383, "grad_norm": 1.6854184865951538, "learning_rate": 2.535205985486321e-06, "loss": 0.8698, "step": 41460 }, { "epoch": 0.7542880008730606, "grad_norm": 1.6303843259811401, "learning_rate": 2.533330833130192e-06, "loss": 0.8856, "step": 41470 }, { "epoch": 0.7544698885028829, "grad_norm": 1.7443337440490723, "learning_rate": 2.531455680774063e-06, "loss": 0.8949, "step": 41480 }, { "epoch": 0.7546517761327052, "grad_norm": 1.617971420288086, "learning_rate": 2.529580528417934e-06, "loss": 0.8974, "step": 41490 }, { "epoch": 0.7548336637625275, "grad_norm": 1.8271180391311646, "learning_rate": 2.5277053760618054e-06, "loss": 0.9291, "step": 41500 }, { "epoch": 0.7550155513923498, "grad_norm": 1.8845607042312622, "learning_rate": 2.5258302237056764e-06, "loss": 0.9053, "step": 41510 }, { "epoch": 0.7551974390221721, "grad_norm": 2.139348268508911, "learning_rate": 2.5239550713495477e-06, "loss": 0.9005, "step": 41520 }, { "epoch": 0.7553793266519944, "grad_norm": 1.995959758758545, "learning_rate": 2.5220799189934186e-06, "loss": 0.9168, "step": 41530 }, { "epoch": 0.7555612142818167, "grad_norm": 1.6316792964935303, "learning_rate": 2.5202047666372895e-06, "loss": 0.8627, "step": 41540 }, { "epoch": 0.755743101911639, "grad_norm": 1.8154624700546265, "learning_rate": 2.5183296142811605e-06, "loss": 0.926, "step": 41550 }, { "epoch": 0.7559249895414613, "grad_norm": 1.8335298299789429, "learning_rate": 2.5164544619250314e-06, "loss": 0.8983, "step": 41560 }, { "epoch": 0.7561068771712836, "grad_norm": 1.7737282514572144, "learning_rate": 2.5145793095689023e-06, "loss": 0.9261, "step": 41570 }, { "epoch": 0.7562887648011059, "grad_norm": 1.8745111227035522, "learning_rate": 2.5127041572127737e-06, "loss": 0.9838, "step": 41580 }, { "epoch": 0.7564706524309281, "grad_norm": 1.862898588180542, "learning_rate": 2.5108290048566446e-06, "loss": 0.9046, "step": 41590 }, { "epoch": 0.7566525400607504, "grad_norm": 1.6752500534057617, "learning_rate": 2.508953852500516e-06, "loss": 0.8718, "step": 41600 }, { "epoch": 0.7568344276905727, "grad_norm": 1.8706533908843994, "learning_rate": 2.507078700144387e-06, "loss": 0.899, "step": 41610 }, { "epoch": 0.757016315320395, "grad_norm": 2.1989855766296387, "learning_rate": 2.505203547788258e-06, "loss": 0.9323, "step": 41620 }, { "epoch": 0.7571982029502173, "grad_norm": 1.7956820726394653, "learning_rate": 2.503328395432129e-06, "loss": 0.9304, "step": 41630 }, { "epoch": 0.7573800905800396, "grad_norm": 1.9564679861068726, "learning_rate": 2.501453243076e-06, "loss": 0.9248, "step": 41640 }, { "epoch": 0.7575619782098619, "grad_norm": 2.0875728130340576, "learning_rate": 2.499578090719871e-06, "loss": 0.9218, "step": 41650 }, { "epoch": 0.7577438658396842, "grad_norm": 1.801546573638916, "learning_rate": 2.4977029383637423e-06, "loss": 0.9206, "step": 41660 }, { "epoch": 0.7579257534695065, "grad_norm": 1.867814302444458, "learning_rate": 2.4958277860076132e-06, "loss": 0.8669, "step": 41670 }, { "epoch": 0.7581076410993288, "grad_norm": 1.785407543182373, "learning_rate": 2.4939526336514846e-06, "loss": 0.8686, "step": 41680 }, { "epoch": 0.7582895287291511, "grad_norm": 1.8583086729049683, "learning_rate": 2.4920774812953555e-06, "loss": 0.9489, "step": 41690 }, { "epoch": 0.7584714163589734, "grad_norm": 1.7577475309371948, "learning_rate": 2.4902023289392264e-06, "loss": 0.9178, "step": 41700 }, { "epoch": 0.7586533039887957, "grad_norm": 1.597063422203064, "learning_rate": 2.4883271765830978e-06, "loss": 0.8923, "step": 41710 }, { "epoch": 0.758835191618618, "grad_norm": 1.4178770780563354, "learning_rate": 2.4864520242269687e-06, "loss": 0.8763, "step": 41720 }, { "epoch": 0.7590170792484403, "grad_norm": 1.8417973518371582, "learning_rate": 2.4845768718708396e-06, "loss": 0.8878, "step": 41730 }, { "epoch": 0.7591989668782626, "grad_norm": 1.5246634483337402, "learning_rate": 2.4827017195147105e-06, "loss": 0.9065, "step": 41740 }, { "epoch": 0.7593808545080849, "grad_norm": 1.6137406826019287, "learning_rate": 2.480826567158582e-06, "loss": 0.8554, "step": 41750 }, { "epoch": 0.7595627421379072, "grad_norm": 1.6949782371520996, "learning_rate": 2.478951414802453e-06, "loss": 0.9109, "step": 41760 }, { "epoch": 0.7597446297677295, "grad_norm": 1.6687166690826416, "learning_rate": 2.477076262446324e-06, "loss": 0.8881, "step": 41770 }, { "epoch": 0.7599265173975518, "grad_norm": 1.6663562059402466, "learning_rate": 2.475201110090195e-06, "loss": 0.9253, "step": 41780 }, { "epoch": 0.7601084050273741, "grad_norm": 1.557867169380188, "learning_rate": 2.473325957734066e-06, "loss": 0.933, "step": 41790 }, { "epoch": 0.7602902926571964, "grad_norm": 1.9198566675186157, "learning_rate": 2.471450805377937e-06, "loss": 0.8923, "step": 41800 }, { "epoch": 0.7604721802870187, "grad_norm": 1.8614556789398193, "learning_rate": 2.4695756530218083e-06, "loss": 0.9467, "step": 41810 }, { "epoch": 0.760654067916841, "grad_norm": 1.6808192729949951, "learning_rate": 2.467700500665679e-06, "loss": 0.895, "step": 41820 }, { "epoch": 0.7608359555466633, "grad_norm": 1.8271970748901367, "learning_rate": 2.4658253483095505e-06, "loss": 0.9107, "step": 41830 }, { "epoch": 0.7610178431764856, "grad_norm": 1.7041412591934204, "learning_rate": 2.4639501959534215e-06, "loss": 0.9241, "step": 41840 }, { "epoch": 0.7611997308063079, "grad_norm": 2.068920373916626, "learning_rate": 2.462075043597293e-06, "loss": 0.9008, "step": 41850 }, { "epoch": 0.7613816184361302, "grad_norm": 1.7394276857376099, "learning_rate": 2.4601998912411633e-06, "loss": 0.8925, "step": 41860 }, { "epoch": 0.7615635060659525, "grad_norm": 1.380502462387085, "learning_rate": 2.4583247388850346e-06, "loss": 0.8883, "step": 41870 }, { "epoch": 0.7617453936957748, "grad_norm": 1.7763675451278687, "learning_rate": 2.4564495865289056e-06, "loss": 0.8668, "step": 41880 }, { "epoch": 0.7619272813255971, "grad_norm": 1.5892289876937866, "learning_rate": 2.454574434172777e-06, "loss": 0.8604, "step": 41890 }, { "epoch": 0.7621091689554194, "grad_norm": 1.8746265172958374, "learning_rate": 2.452699281816648e-06, "loss": 0.8981, "step": 41900 }, { "epoch": 0.7622910565852417, "grad_norm": 1.6167066097259521, "learning_rate": 2.4508241294605188e-06, "loss": 0.8977, "step": 41910 }, { "epoch": 0.762472944215064, "grad_norm": 2.0337586402893066, "learning_rate": 2.44894897710439e-06, "loss": 0.8889, "step": 41920 }, { "epoch": 0.7626548318448862, "grad_norm": 1.8232733011245728, "learning_rate": 2.447073824748261e-06, "loss": 0.9215, "step": 41930 }, { "epoch": 0.7628367194747085, "grad_norm": 1.8637588024139404, "learning_rate": 2.445198672392132e-06, "loss": 0.9234, "step": 41940 }, { "epoch": 0.7630186071045308, "grad_norm": 1.7160189151763916, "learning_rate": 2.4433235200360033e-06, "loss": 0.9221, "step": 41950 }, { "epoch": 0.7632004947343531, "grad_norm": 1.7039849758148193, "learning_rate": 2.4414483676798742e-06, "loss": 0.9097, "step": 41960 }, { "epoch": 0.7633823823641754, "grad_norm": 1.5583056211471558, "learning_rate": 2.439573215323745e-06, "loss": 0.9086, "step": 41970 }, { "epoch": 0.7635642699939977, "grad_norm": 1.687617540359497, "learning_rate": 2.4376980629676165e-06, "loss": 0.9147, "step": 41980 }, { "epoch": 0.76374615762382, "grad_norm": 1.7017635107040405, "learning_rate": 2.4358229106114874e-06, "loss": 0.9206, "step": 41990 }, { "epoch": 0.7639280452536423, "grad_norm": 2.1266071796417236, "learning_rate": 2.4339477582553583e-06, "loss": 0.9041, "step": 42000 }, { "epoch": 0.7641099328834646, "grad_norm": 1.6725590229034424, "learning_rate": 2.4320726058992293e-06, "loss": 0.8927, "step": 42010 }, { "epoch": 0.7642918205132869, "grad_norm": 1.6485366821289062, "learning_rate": 2.4301974535431006e-06, "loss": 0.9143, "step": 42020 }, { "epoch": 0.7644737081431092, "grad_norm": 1.7721554040908813, "learning_rate": 2.4283223011869715e-06, "loss": 0.921, "step": 42030 }, { "epoch": 0.7646555957729315, "grad_norm": 1.5313493013381958, "learning_rate": 2.426447148830843e-06, "loss": 0.9077, "step": 42040 }, { "epoch": 0.7648374834027538, "grad_norm": 1.6091798543930054, "learning_rate": 2.424571996474714e-06, "loss": 0.9207, "step": 42050 }, { "epoch": 0.7650193710325761, "grad_norm": 1.6185857057571411, "learning_rate": 2.422696844118585e-06, "loss": 0.8639, "step": 42060 }, { "epoch": 0.7652012586623984, "grad_norm": 1.851375937461853, "learning_rate": 2.4208216917624556e-06, "loss": 0.8728, "step": 42070 }, { "epoch": 0.7653831462922207, "grad_norm": 1.2374882698059082, "learning_rate": 2.418946539406327e-06, "loss": 0.8953, "step": 42080 }, { "epoch": 0.765565033922043, "grad_norm": 1.3764516115188599, "learning_rate": 2.417071387050198e-06, "loss": 0.8881, "step": 42090 }, { "epoch": 0.7657469215518653, "grad_norm": 1.6998172998428345, "learning_rate": 2.4151962346940693e-06, "loss": 0.9295, "step": 42100 }, { "epoch": 0.7659288091816876, "grad_norm": 1.8076428174972534, "learning_rate": 2.41332108233794e-06, "loss": 0.8954, "step": 42110 }, { "epoch": 0.7661106968115099, "grad_norm": 1.6420116424560547, "learning_rate": 2.4114459299818115e-06, "loss": 0.9265, "step": 42120 }, { "epoch": 0.7662925844413322, "grad_norm": 1.8474366664886475, "learning_rate": 2.4095707776256824e-06, "loss": 0.9278, "step": 42130 }, { "epoch": 0.7664744720711545, "grad_norm": 1.726036548614502, "learning_rate": 2.4076956252695534e-06, "loss": 0.9023, "step": 42140 }, { "epoch": 0.7666563597009768, "grad_norm": 1.5912531614303589, "learning_rate": 2.4058204729134243e-06, "loss": 0.8913, "step": 42150 }, { "epoch": 0.7668382473307991, "grad_norm": 1.5874515771865845, "learning_rate": 2.4039453205572956e-06, "loss": 0.87, "step": 42160 }, { "epoch": 0.7670201349606214, "grad_norm": 1.8068519830703735, "learning_rate": 2.4020701682011666e-06, "loss": 0.8868, "step": 42170 }, { "epoch": 0.7672020225904437, "grad_norm": 1.613627314567566, "learning_rate": 2.4001950158450375e-06, "loss": 0.9522, "step": 42180 }, { "epoch": 0.767383910220266, "grad_norm": 1.60687255859375, "learning_rate": 2.398319863488909e-06, "loss": 0.9041, "step": 42190 }, { "epoch": 0.7675657978500882, "grad_norm": 1.5712878704071045, "learning_rate": 2.3964447111327797e-06, "loss": 0.9457, "step": 42200 }, { "epoch": 0.7677476854799105, "grad_norm": 1.4283004999160767, "learning_rate": 2.3945695587766507e-06, "loss": 0.8585, "step": 42210 }, { "epoch": 0.7679295731097328, "grad_norm": 1.661405324935913, "learning_rate": 2.392694406420522e-06, "loss": 0.8865, "step": 42220 }, { "epoch": 0.768111460739555, "grad_norm": 1.7527471780776978, "learning_rate": 2.390819254064393e-06, "loss": 0.9091, "step": 42230 }, { "epoch": 0.7682933483693773, "grad_norm": 1.4924428462982178, "learning_rate": 2.388944101708264e-06, "loss": 0.8916, "step": 42240 }, { "epoch": 0.7684752359991996, "grad_norm": 1.8193546533584595, "learning_rate": 2.387068949352135e-06, "loss": 0.9331, "step": 42250 }, { "epoch": 0.7686571236290219, "grad_norm": 1.7097461223602295, "learning_rate": 2.385193796996006e-06, "loss": 0.971, "step": 42260 }, { "epoch": 0.7688390112588442, "grad_norm": 1.7990764379501343, "learning_rate": 2.383318644639877e-06, "loss": 0.8876, "step": 42270 }, { "epoch": 0.7690208988886665, "grad_norm": 1.6781110763549805, "learning_rate": 2.381443492283748e-06, "loss": 0.844, "step": 42280 }, { "epoch": 0.7692027865184888, "grad_norm": 1.763645887374878, "learning_rate": 2.3795683399276193e-06, "loss": 0.9001, "step": 42290 }, { "epoch": 0.7693846741483111, "grad_norm": 1.9159892797470093, "learning_rate": 2.3776931875714902e-06, "loss": 0.9753, "step": 42300 }, { "epoch": 0.7695665617781334, "grad_norm": 1.6262919902801514, "learning_rate": 2.3758180352153616e-06, "loss": 0.9634, "step": 42310 }, { "epoch": 0.7697484494079557, "grad_norm": 1.7619863748550415, "learning_rate": 2.3739428828592325e-06, "loss": 0.9202, "step": 42320 }, { "epoch": 0.769930337037778, "grad_norm": 1.82402503490448, "learning_rate": 2.372067730503104e-06, "loss": 0.8989, "step": 42330 }, { "epoch": 0.7701122246676003, "grad_norm": 2.1735100746154785, "learning_rate": 2.3701925781469744e-06, "loss": 0.8952, "step": 42340 }, { "epoch": 0.7702941122974226, "grad_norm": 1.8289909362792969, "learning_rate": 2.3683174257908457e-06, "loss": 0.9365, "step": 42350 }, { "epoch": 0.7704759999272449, "grad_norm": 1.8887302875518799, "learning_rate": 2.3664422734347166e-06, "loss": 0.955, "step": 42360 }, { "epoch": 0.7706578875570672, "grad_norm": 1.7182507514953613, "learning_rate": 2.364567121078588e-06, "loss": 0.9072, "step": 42370 }, { "epoch": 0.7708397751868895, "grad_norm": 1.6853119134902954, "learning_rate": 2.362691968722459e-06, "loss": 0.88, "step": 42380 }, { "epoch": 0.7710216628167118, "grad_norm": 1.7241839170455933, "learning_rate": 2.3608168163663302e-06, "loss": 0.9031, "step": 42390 }, { "epoch": 0.7712035504465341, "grad_norm": 1.8425084352493286, "learning_rate": 2.358941664010201e-06, "loss": 0.9154, "step": 42400 }, { "epoch": 0.7713854380763564, "grad_norm": 1.8742754459381104, "learning_rate": 2.357066511654072e-06, "loss": 0.8777, "step": 42410 }, { "epoch": 0.7715673257061787, "grad_norm": 1.6006019115447998, "learning_rate": 2.355191359297943e-06, "loss": 0.8947, "step": 42420 }, { "epoch": 0.771749213336001, "grad_norm": 2.060551881790161, "learning_rate": 2.3533162069418144e-06, "loss": 0.956, "step": 42430 }, { "epoch": 0.7719311009658233, "grad_norm": 2.299767017364502, "learning_rate": 2.3514410545856853e-06, "loss": 0.9184, "step": 42440 }, { "epoch": 0.7721129885956456, "grad_norm": 1.6904571056365967, "learning_rate": 2.349565902229556e-06, "loss": 0.8799, "step": 42450 }, { "epoch": 0.7722948762254679, "grad_norm": 1.8340848684310913, "learning_rate": 2.3476907498734275e-06, "loss": 0.9508, "step": 42460 }, { "epoch": 0.7724767638552902, "grad_norm": 1.6514467000961304, "learning_rate": 2.3458155975172985e-06, "loss": 0.9307, "step": 42470 }, { "epoch": 0.7726586514851125, "grad_norm": 1.5540434122085571, "learning_rate": 2.3439404451611694e-06, "loss": 0.9406, "step": 42480 }, { "epoch": 0.7728405391149348, "grad_norm": 1.6356875896453857, "learning_rate": 2.3420652928050407e-06, "loss": 0.8882, "step": 42490 }, { "epoch": 0.7730224267447571, "grad_norm": 1.8356835842132568, "learning_rate": 2.3401901404489117e-06, "loss": 0.9032, "step": 42500 }, { "epoch": 0.7732043143745794, "grad_norm": 1.502287745475769, "learning_rate": 2.3383149880927826e-06, "loss": 0.8866, "step": 42510 }, { "epoch": 0.7733862020044017, "grad_norm": 1.763626217842102, "learning_rate": 2.336439835736654e-06, "loss": 0.8481, "step": 42520 }, { "epoch": 0.773568089634224, "grad_norm": 1.4792379140853882, "learning_rate": 2.334564683380525e-06, "loss": 0.9065, "step": 42530 }, { "epoch": 0.7737499772640463, "grad_norm": 1.8244445323944092, "learning_rate": 2.332689531024396e-06, "loss": 0.8481, "step": 42540 }, { "epoch": 0.7739318648938686, "grad_norm": 1.701676607131958, "learning_rate": 2.3308143786682667e-06, "loss": 0.9471, "step": 42550 }, { "epoch": 0.7741137525236909, "grad_norm": 1.527466893196106, "learning_rate": 2.328939226312138e-06, "loss": 0.8499, "step": 42560 }, { "epoch": 0.7742956401535132, "grad_norm": 1.9091274738311768, "learning_rate": 2.327064073956009e-06, "loss": 0.8756, "step": 42570 }, { "epoch": 0.7744775277833355, "grad_norm": 2.066077470779419, "learning_rate": 2.3251889215998803e-06, "loss": 0.9404, "step": 42580 }, { "epoch": 0.7746594154131577, "grad_norm": 1.6858938932418823, "learning_rate": 2.3233137692437512e-06, "loss": 0.9469, "step": 42590 }, { "epoch": 0.77484130304298, "grad_norm": 1.316709041595459, "learning_rate": 2.3214386168876226e-06, "loss": 0.8573, "step": 42600 }, { "epoch": 0.7750231906728023, "grad_norm": 1.8855088949203491, "learning_rate": 2.3195634645314935e-06, "loss": 0.8755, "step": 42610 }, { "epoch": 0.7752050783026246, "grad_norm": 1.6909294128417969, "learning_rate": 2.3176883121753644e-06, "loss": 0.8828, "step": 42620 }, { "epoch": 0.7753869659324469, "grad_norm": 1.7338963747024536, "learning_rate": 2.3158131598192353e-06, "loss": 0.8943, "step": 42630 }, { "epoch": 0.7755688535622692, "grad_norm": 1.8949676752090454, "learning_rate": 2.3139380074631067e-06, "loss": 0.8736, "step": 42640 }, { "epoch": 0.7757507411920915, "grad_norm": 2.0595011711120605, "learning_rate": 2.3120628551069776e-06, "loss": 0.901, "step": 42650 }, { "epoch": 0.7759326288219138, "grad_norm": 1.957818865776062, "learning_rate": 2.310187702750849e-06, "loss": 0.9169, "step": 42660 }, { "epoch": 0.7761145164517361, "grad_norm": 1.6250289678573608, "learning_rate": 2.30831255039472e-06, "loss": 0.9123, "step": 42670 }, { "epoch": 0.7762964040815584, "grad_norm": 1.6227766275405884, "learning_rate": 2.306437398038591e-06, "loss": 0.8745, "step": 42680 }, { "epoch": 0.7764782917113807, "grad_norm": 1.809598684310913, "learning_rate": 2.3045622456824617e-06, "loss": 0.9414, "step": 42690 }, { "epoch": 0.776660179341203, "grad_norm": 2.1282601356506348, "learning_rate": 2.302687093326333e-06, "loss": 0.9335, "step": 42700 }, { "epoch": 0.7768420669710253, "grad_norm": 1.5858864784240723, "learning_rate": 2.300811940970204e-06, "loss": 0.9004, "step": 42710 }, { "epoch": 0.7770239546008476, "grad_norm": 2.0132319927215576, "learning_rate": 2.298936788614075e-06, "loss": 0.8865, "step": 42720 }, { "epoch": 0.7772058422306699, "grad_norm": 1.9529809951782227, "learning_rate": 2.2970616362579463e-06, "loss": 0.9097, "step": 42730 }, { "epoch": 0.7773877298604922, "grad_norm": 1.7068140506744385, "learning_rate": 2.295186483901817e-06, "loss": 0.9053, "step": 42740 }, { "epoch": 0.7775696174903145, "grad_norm": 1.705329418182373, "learning_rate": 2.2933113315456885e-06, "loss": 0.9196, "step": 42750 }, { "epoch": 0.7777515051201368, "grad_norm": 1.8144283294677734, "learning_rate": 2.291436179189559e-06, "loss": 0.8781, "step": 42760 }, { "epoch": 0.7779333927499591, "grad_norm": 1.9679603576660156, "learning_rate": 2.2895610268334304e-06, "loss": 0.8963, "step": 42770 }, { "epoch": 0.7781152803797814, "grad_norm": 1.9503834247589111, "learning_rate": 2.2876858744773013e-06, "loss": 0.8876, "step": 42780 }, { "epoch": 0.7782971680096037, "grad_norm": 1.792384147644043, "learning_rate": 2.2858107221211726e-06, "loss": 0.9028, "step": 42790 }, { "epoch": 0.778479055639426, "grad_norm": 1.7463992834091187, "learning_rate": 2.2839355697650436e-06, "loss": 0.8863, "step": 42800 }, { "epoch": 0.7786609432692483, "grad_norm": 1.7808376550674438, "learning_rate": 2.282060417408915e-06, "loss": 0.9174, "step": 42810 }, { "epoch": 0.7788428308990706, "grad_norm": 1.7383902072906494, "learning_rate": 2.2801852650527854e-06, "loss": 0.9135, "step": 42820 }, { "epoch": 0.7790247185288929, "grad_norm": 1.4799249172210693, "learning_rate": 2.2783101126966568e-06, "loss": 0.8796, "step": 42830 }, { "epoch": 0.7792066061587152, "grad_norm": 1.8857707977294922, "learning_rate": 2.2764349603405277e-06, "loss": 0.8797, "step": 42840 }, { "epoch": 0.7793884937885375, "grad_norm": 1.7146985530853271, "learning_rate": 2.274559807984399e-06, "loss": 0.902, "step": 42850 }, { "epoch": 0.7795703814183598, "grad_norm": 1.8550995588302612, "learning_rate": 2.27268465562827e-06, "loss": 0.9167, "step": 42860 }, { "epoch": 0.7797522690481821, "grad_norm": 2.076760768890381, "learning_rate": 2.2708095032721413e-06, "loss": 0.9099, "step": 42870 }, { "epoch": 0.7799341566780044, "grad_norm": 1.6754891872406006, "learning_rate": 2.2689343509160122e-06, "loss": 0.9383, "step": 42880 }, { "epoch": 0.7801160443078267, "grad_norm": 1.8648638725280762, "learning_rate": 2.267059198559883e-06, "loss": 0.9724, "step": 42890 }, { "epoch": 0.780297931937649, "grad_norm": 1.386359691619873, "learning_rate": 2.265184046203754e-06, "loss": 0.847, "step": 42900 }, { "epoch": 0.7804798195674713, "grad_norm": 1.674035668373108, "learning_rate": 2.2633088938476254e-06, "loss": 0.9051, "step": 42910 }, { "epoch": 0.7806617071972936, "grad_norm": 1.6052314043045044, "learning_rate": 2.2614337414914963e-06, "loss": 0.8636, "step": 42920 }, { "epoch": 0.7808435948271158, "grad_norm": 1.7114990949630737, "learning_rate": 2.2595585891353673e-06, "loss": 0.9099, "step": 42930 }, { "epoch": 0.7810254824569381, "grad_norm": 1.8969143629074097, "learning_rate": 2.2576834367792386e-06, "loss": 0.9217, "step": 42940 }, { "epoch": 0.7812073700867604, "grad_norm": 1.3768519163131714, "learning_rate": 2.2558082844231095e-06, "loss": 0.909, "step": 42950 }, { "epoch": 0.7813892577165827, "grad_norm": 1.7688244581222534, "learning_rate": 2.2539331320669804e-06, "loss": 0.9019, "step": 42960 }, { "epoch": 0.781571145346405, "grad_norm": 1.90581214427948, "learning_rate": 2.252057979710852e-06, "loss": 0.9166, "step": 42970 }, { "epoch": 0.7817530329762273, "grad_norm": 1.758557677268982, "learning_rate": 2.2501828273547227e-06, "loss": 0.9213, "step": 42980 }, { "epoch": 0.7819349206060496, "grad_norm": 1.5379631519317627, "learning_rate": 2.2483076749985936e-06, "loss": 0.8509, "step": 42990 }, { "epoch": 0.7821168082358719, "grad_norm": 1.7358235120773315, "learning_rate": 2.246432522642465e-06, "loss": 0.902, "step": 43000 }, { "epoch": 0.7822986958656942, "grad_norm": 1.6083182096481323, "learning_rate": 2.244557370286336e-06, "loss": 0.8704, "step": 43010 }, { "epoch": 0.7824805834955165, "grad_norm": 1.8944133520126343, "learning_rate": 2.2426822179302073e-06, "loss": 0.8854, "step": 43020 }, { "epoch": 0.7826624711253388, "grad_norm": 1.5608144998550415, "learning_rate": 2.2408070655740778e-06, "loss": 0.8785, "step": 43030 }, { "epoch": 0.782844358755161, "grad_norm": 1.922378420829773, "learning_rate": 2.238931913217949e-06, "loss": 0.9259, "step": 43040 }, { "epoch": 0.7830262463849833, "grad_norm": 1.541318416595459, "learning_rate": 2.23705676086182e-06, "loss": 0.8903, "step": 43050 }, { "epoch": 0.7832081340148056, "grad_norm": 1.792204737663269, "learning_rate": 2.2351816085056914e-06, "loss": 0.9087, "step": 43060 }, { "epoch": 0.7833900216446279, "grad_norm": 1.765285849571228, "learning_rate": 2.2333064561495623e-06, "loss": 0.9046, "step": 43070 }, { "epoch": 0.7835719092744502, "grad_norm": 2.0389106273651123, "learning_rate": 2.2314313037934336e-06, "loss": 0.8935, "step": 43080 }, { "epoch": 0.7837537969042725, "grad_norm": 1.9015138149261475, "learning_rate": 2.2295561514373046e-06, "loss": 0.86, "step": 43090 }, { "epoch": 0.7839356845340948, "grad_norm": 1.7733100652694702, "learning_rate": 2.2276809990811755e-06, "loss": 0.8777, "step": 43100 }, { "epoch": 0.7841175721639171, "grad_norm": 1.3989719152450562, "learning_rate": 2.2258058467250464e-06, "loss": 0.8214, "step": 43110 }, { "epoch": 0.7842994597937394, "grad_norm": 1.6865805387496948, "learning_rate": 2.2239306943689177e-06, "loss": 0.879, "step": 43120 }, { "epoch": 0.7844813474235617, "grad_norm": 1.671446442604065, "learning_rate": 2.2220555420127887e-06, "loss": 0.9195, "step": 43130 }, { "epoch": 0.784663235053384, "grad_norm": 1.4694942235946655, "learning_rate": 2.22018038965666e-06, "loss": 0.8887, "step": 43140 }, { "epoch": 0.7848451226832063, "grad_norm": 1.85934317111969, "learning_rate": 2.218305237300531e-06, "loss": 0.9267, "step": 43150 }, { "epoch": 0.7850270103130286, "grad_norm": 1.959727168083191, "learning_rate": 2.216430084944402e-06, "loss": 0.9055, "step": 43160 }, { "epoch": 0.7852088979428509, "grad_norm": 2.0022506713867188, "learning_rate": 2.2145549325882728e-06, "loss": 0.8292, "step": 43170 }, { "epoch": 0.7853907855726732, "grad_norm": 1.6167529821395874, "learning_rate": 2.212679780232144e-06, "loss": 0.9049, "step": 43180 }, { "epoch": 0.7855726732024955, "grad_norm": 1.917586088180542, "learning_rate": 2.210804627876015e-06, "loss": 0.9349, "step": 43190 }, { "epoch": 0.7857545608323178, "grad_norm": 2.025250196456909, "learning_rate": 2.208929475519886e-06, "loss": 0.886, "step": 43200 }, { "epoch": 0.7859364484621401, "grad_norm": 1.7453548908233643, "learning_rate": 2.2070543231637573e-06, "loss": 0.9379, "step": 43210 }, { "epoch": 0.7861183360919624, "grad_norm": 1.6541645526885986, "learning_rate": 2.2051791708076282e-06, "loss": 0.891, "step": 43220 }, { "epoch": 0.7863002237217847, "grad_norm": 2.0104525089263916, "learning_rate": 2.2033040184514996e-06, "loss": 0.8741, "step": 43230 }, { "epoch": 0.786482111351607, "grad_norm": 1.68494713306427, "learning_rate": 2.2014288660953705e-06, "loss": 0.8914, "step": 43240 }, { "epoch": 0.7866639989814292, "grad_norm": 1.912609577178955, "learning_rate": 2.1995537137392414e-06, "loss": 0.9695, "step": 43250 }, { "epoch": 0.7868458866112515, "grad_norm": 1.5977191925048828, "learning_rate": 2.1976785613831124e-06, "loss": 0.9063, "step": 43260 }, { "epoch": 0.7870277742410738, "grad_norm": 1.7781933546066284, "learning_rate": 2.1958034090269837e-06, "loss": 0.9185, "step": 43270 }, { "epoch": 0.7872096618708961, "grad_norm": 1.8781943321228027, "learning_rate": 2.1939282566708546e-06, "loss": 0.9266, "step": 43280 }, { "epoch": 0.7873915495007184, "grad_norm": 1.8348217010498047, "learning_rate": 2.192053104314726e-06, "loss": 0.9565, "step": 43290 }, { "epoch": 0.7875734371305407, "grad_norm": 1.6926681995391846, "learning_rate": 2.190177951958597e-06, "loss": 0.8811, "step": 43300 }, { "epoch": 0.787755324760363, "grad_norm": 2.1221425533294678, "learning_rate": 2.188302799602468e-06, "loss": 0.9025, "step": 43310 }, { "epoch": 0.7879372123901853, "grad_norm": 1.9504355192184448, "learning_rate": 2.1864276472463387e-06, "loss": 0.9244, "step": 43320 }, { "epoch": 0.7881191000200076, "grad_norm": 2.2104175090789795, "learning_rate": 2.18455249489021e-06, "loss": 0.8892, "step": 43330 }, { "epoch": 0.7883009876498299, "grad_norm": 1.722556710243225, "learning_rate": 2.182677342534081e-06, "loss": 0.906, "step": 43340 }, { "epoch": 0.7884828752796522, "grad_norm": 1.8197990655899048, "learning_rate": 2.1808021901779524e-06, "loss": 0.8749, "step": 43350 }, { "epoch": 0.7886647629094745, "grad_norm": 1.7930173873901367, "learning_rate": 2.1789270378218233e-06, "loss": 0.9272, "step": 43360 }, { "epoch": 0.7888466505392968, "grad_norm": 1.7470405101776123, "learning_rate": 2.177051885465694e-06, "loss": 0.8804, "step": 43370 }, { "epoch": 0.7890285381691191, "grad_norm": 1.900444746017456, "learning_rate": 2.175176733109565e-06, "loss": 0.9153, "step": 43380 }, { "epoch": 0.7892104257989414, "grad_norm": 1.8325953483581543, "learning_rate": 2.1733015807534365e-06, "loss": 0.8825, "step": 43390 }, { "epoch": 0.7893923134287637, "grad_norm": 1.683533787727356, "learning_rate": 2.1714264283973074e-06, "loss": 0.95, "step": 43400 }, { "epoch": 0.789574201058586, "grad_norm": 1.8311876058578491, "learning_rate": 2.1695512760411787e-06, "loss": 0.9595, "step": 43410 }, { "epoch": 0.7897560886884083, "grad_norm": 1.7295600175857544, "learning_rate": 2.1676761236850497e-06, "loss": 0.8543, "step": 43420 }, { "epoch": 0.7899379763182306, "grad_norm": 1.8142833709716797, "learning_rate": 2.1658009713289206e-06, "loss": 0.911, "step": 43430 }, { "epoch": 0.7901198639480529, "grad_norm": 1.7737582921981812, "learning_rate": 2.1639258189727915e-06, "loss": 0.9242, "step": 43440 }, { "epoch": 0.7903017515778752, "grad_norm": 1.9711682796478271, "learning_rate": 2.162050666616663e-06, "loss": 0.8729, "step": 43450 }, { "epoch": 0.7904836392076975, "grad_norm": 1.7207939624786377, "learning_rate": 2.1601755142605338e-06, "loss": 0.9389, "step": 43460 }, { "epoch": 0.7906655268375198, "grad_norm": 1.78118896484375, "learning_rate": 2.1583003619044047e-06, "loss": 0.9566, "step": 43470 }, { "epoch": 0.7908474144673421, "grad_norm": 1.901597261428833, "learning_rate": 2.156425209548276e-06, "loss": 0.9468, "step": 43480 }, { "epoch": 0.7910293020971644, "grad_norm": 1.899863839149475, "learning_rate": 2.154550057192147e-06, "loss": 0.9347, "step": 43490 }, { "epoch": 0.7912111897269867, "grad_norm": 1.512224793434143, "learning_rate": 2.1526749048360183e-06, "loss": 0.8485, "step": 43500 }, { "epoch": 0.791393077356809, "grad_norm": 1.4757670164108276, "learning_rate": 2.1507997524798892e-06, "loss": 0.8697, "step": 43510 }, { "epoch": 0.7915749649866313, "grad_norm": 1.8467522859573364, "learning_rate": 2.14892460012376e-06, "loss": 0.9861, "step": 43520 }, { "epoch": 0.7917568526164536, "grad_norm": 1.285565733909607, "learning_rate": 2.147049447767631e-06, "loss": 0.8934, "step": 43530 }, { "epoch": 0.7919387402462759, "grad_norm": 1.7831802368164062, "learning_rate": 2.1451742954115024e-06, "loss": 0.936, "step": 43540 }, { "epoch": 0.7921206278760982, "grad_norm": 1.636218547821045, "learning_rate": 2.1432991430553733e-06, "loss": 0.87, "step": 43550 }, { "epoch": 0.7923025155059205, "grad_norm": 1.837083101272583, "learning_rate": 2.1414239906992447e-06, "loss": 0.9259, "step": 43560 }, { "epoch": 0.7924844031357428, "grad_norm": 1.8146953582763672, "learning_rate": 2.1395488383431156e-06, "loss": 0.9263, "step": 43570 }, { "epoch": 0.792666290765565, "grad_norm": 1.794218897819519, "learning_rate": 2.1376736859869865e-06, "loss": 0.9166, "step": 43580 }, { "epoch": 0.7928481783953873, "grad_norm": 1.9038695096969604, "learning_rate": 2.1357985336308575e-06, "loss": 0.9401, "step": 43590 }, { "epoch": 0.7930300660252096, "grad_norm": 1.6904195547103882, "learning_rate": 2.133923381274729e-06, "loss": 0.8891, "step": 43600 }, { "epoch": 0.7932119536550319, "grad_norm": 2.0176522731781006, "learning_rate": 2.1320482289185997e-06, "loss": 0.881, "step": 43610 }, { "epoch": 0.7933938412848542, "grad_norm": 1.7209473848342896, "learning_rate": 2.130173076562471e-06, "loss": 0.8974, "step": 43620 }, { "epoch": 0.7935757289146765, "grad_norm": 1.5715478658676147, "learning_rate": 2.128297924206342e-06, "loss": 0.8287, "step": 43630 }, { "epoch": 0.7937576165444988, "grad_norm": 1.8117355108261108, "learning_rate": 2.126422771850213e-06, "loss": 0.8986, "step": 43640 }, { "epoch": 0.7939395041743211, "grad_norm": 1.7448678016662598, "learning_rate": 2.124547619494084e-06, "loss": 0.9137, "step": 43650 }, { "epoch": 0.7941213918041434, "grad_norm": 1.3828229904174805, "learning_rate": 2.122672467137955e-06, "loss": 0.9037, "step": 43660 }, { "epoch": 0.7943032794339657, "grad_norm": 1.5643621683120728, "learning_rate": 2.120797314781826e-06, "loss": 0.8866, "step": 43670 }, { "epoch": 0.794485167063788, "grad_norm": 1.8307050466537476, "learning_rate": 2.1189221624256975e-06, "loss": 0.9131, "step": 43680 }, { "epoch": 0.7946670546936103, "grad_norm": 1.8553756475448608, "learning_rate": 2.1170470100695684e-06, "loss": 0.9398, "step": 43690 }, { "epoch": 0.7948489423234326, "grad_norm": 1.6899996995925903, "learning_rate": 2.1151718577134393e-06, "loss": 0.9432, "step": 43700 }, { "epoch": 0.7950308299532549, "grad_norm": 1.7770848274230957, "learning_rate": 2.1132967053573106e-06, "loss": 0.8741, "step": 43710 }, { "epoch": 0.7952127175830772, "grad_norm": 1.751620888710022, "learning_rate": 2.1114215530011816e-06, "loss": 0.8945, "step": 43720 }, { "epoch": 0.7953946052128995, "grad_norm": 1.8687301874160767, "learning_rate": 2.1095464006450525e-06, "loss": 0.9024, "step": 43730 }, { "epoch": 0.7955764928427218, "grad_norm": 1.881684422492981, "learning_rate": 2.1076712482889234e-06, "loss": 0.9507, "step": 43740 }, { "epoch": 0.7957583804725441, "grad_norm": 1.9262299537658691, "learning_rate": 2.1057960959327948e-06, "loss": 0.8982, "step": 43750 }, { "epoch": 0.7959402681023664, "grad_norm": 1.684360146522522, "learning_rate": 2.1039209435766657e-06, "loss": 0.9338, "step": 43760 }, { "epoch": 0.7961221557321887, "grad_norm": 1.7470130920410156, "learning_rate": 2.102045791220537e-06, "loss": 0.8692, "step": 43770 }, { "epoch": 0.796304043362011, "grad_norm": 1.2289574146270752, "learning_rate": 2.100170638864408e-06, "loss": 0.8613, "step": 43780 }, { "epoch": 0.7964859309918333, "grad_norm": 1.8759946823120117, "learning_rate": 2.098295486508279e-06, "loss": 0.9271, "step": 43790 }, { "epoch": 0.7966678186216556, "grad_norm": 1.841951847076416, "learning_rate": 2.09642033415215e-06, "loss": 0.8778, "step": 43800 }, { "epoch": 0.7968497062514779, "grad_norm": 1.5339889526367188, "learning_rate": 2.094545181796021e-06, "loss": 0.894, "step": 43810 }, { "epoch": 0.7970315938813002, "grad_norm": 1.7188634872436523, "learning_rate": 2.092670029439892e-06, "loss": 0.879, "step": 43820 }, { "epoch": 0.7972134815111225, "grad_norm": 1.3540223836898804, "learning_rate": 2.0907948770837634e-06, "loss": 0.9123, "step": 43830 }, { "epoch": 0.7973953691409448, "grad_norm": 2.1570932865142822, "learning_rate": 2.0889197247276343e-06, "loss": 0.8973, "step": 43840 }, { "epoch": 0.7975772567707671, "grad_norm": 1.876043438911438, "learning_rate": 2.0870445723715057e-06, "loss": 0.9745, "step": 43850 }, { "epoch": 0.7977591444005894, "grad_norm": 1.9116426706314087, "learning_rate": 2.085169420015376e-06, "loss": 0.9401, "step": 43860 }, { "epoch": 0.7979410320304116, "grad_norm": 1.690272331237793, "learning_rate": 2.0832942676592475e-06, "loss": 0.8736, "step": 43870 }, { "epoch": 0.7981229196602339, "grad_norm": 2.1505374908447266, "learning_rate": 2.0814191153031184e-06, "loss": 0.9132, "step": 43880 }, { "epoch": 0.7983048072900562, "grad_norm": 1.7553631067276, "learning_rate": 2.07954396294699e-06, "loss": 0.9118, "step": 43890 }, { "epoch": 0.7984866949198784, "grad_norm": 1.7497611045837402, "learning_rate": 2.0776688105908607e-06, "loss": 0.9103, "step": 43900 }, { "epoch": 0.7986685825497007, "grad_norm": 1.9496064186096191, "learning_rate": 2.0757936582347316e-06, "loss": 0.8947, "step": 43910 }, { "epoch": 0.798850470179523, "grad_norm": 1.748233437538147, "learning_rate": 2.073918505878603e-06, "loss": 0.9216, "step": 43920 }, { "epoch": 0.7990323578093453, "grad_norm": 3.071859359741211, "learning_rate": 2.072043353522474e-06, "loss": 0.9271, "step": 43930 }, { "epoch": 0.7992142454391676, "grad_norm": 1.6989850997924805, "learning_rate": 2.070168201166345e-06, "loss": 0.8804, "step": 43940 }, { "epoch": 0.7993961330689899, "grad_norm": 2.064868927001953, "learning_rate": 2.068293048810216e-06, "loss": 0.8758, "step": 43950 }, { "epoch": 0.7995780206988122, "grad_norm": 1.5635185241699219, "learning_rate": 2.066417896454087e-06, "loss": 0.8795, "step": 43960 }, { "epoch": 0.7997599083286345, "grad_norm": 1.5630178451538086, "learning_rate": 2.064542744097958e-06, "loss": 0.9514, "step": 43970 }, { "epoch": 0.7999417959584568, "grad_norm": 1.6353487968444824, "learning_rate": 2.0626675917418294e-06, "loss": 0.8631, "step": 43980 }, { "epoch": 0.8001236835882791, "grad_norm": 1.7716807126998901, "learning_rate": 2.0607924393857003e-06, "loss": 0.9069, "step": 43990 }, { "epoch": 0.8003055712181014, "grad_norm": 1.7452951669692993, "learning_rate": 2.0589172870295712e-06, "loss": 0.8919, "step": 44000 }, { "epoch": 0.8004874588479237, "grad_norm": 1.6576308012008667, "learning_rate": 2.057042134673442e-06, "loss": 0.958, "step": 44010 }, { "epoch": 0.800669346477746, "grad_norm": 1.8645013570785522, "learning_rate": 2.0551669823173135e-06, "loss": 0.8899, "step": 44020 }, { "epoch": 0.8008512341075683, "grad_norm": 2.34377384185791, "learning_rate": 2.0532918299611844e-06, "loss": 0.9146, "step": 44030 }, { "epoch": 0.8010331217373906, "grad_norm": 1.6898078918457031, "learning_rate": 2.0514166776050557e-06, "loss": 0.9275, "step": 44040 }, { "epoch": 0.8012150093672129, "grad_norm": 1.6284592151641846, "learning_rate": 2.0495415252489267e-06, "loss": 0.9096, "step": 44050 }, { "epoch": 0.8013968969970352, "grad_norm": 1.7936326265335083, "learning_rate": 2.0476663728927976e-06, "loss": 0.8501, "step": 44060 }, { "epoch": 0.8015787846268575, "grad_norm": 1.7518349885940552, "learning_rate": 2.0457912205366685e-06, "loss": 0.887, "step": 44070 }, { "epoch": 0.8017606722566798, "grad_norm": 1.8617668151855469, "learning_rate": 2.04391606818054e-06, "loss": 0.8837, "step": 44080 }, { "epoch": 0.8019425598865021, "grad_norm": 1.596958875656128, "learning_rate": 2.0420409158244108e-06, "loss": 0.8794, "step": 44090 }, { "epoch": 0.8021244475163244, "grad_norm": 1.8482812643051147, "learning_rate": 2.040165763468282e-06, "loss": 0.9212, "step": 44100 }, { "epoch": 0.8023063351461467, "grad_norm": 1.5570991039276123, "learning_rate": 2.038290611112153e-06, "loss": 0.9247, "step": 44110 }, { "epoch": 0.802488222775969, "grad_norm": 2.004068374633789, "learning_rate": 2.0364154587560244e-06, "loss": 0.9222, "step": 44120 }, { "epoch": 0.8026701104057913, "grad_norm": 1.6813607215881348, "learning_rate": 2.034540306399895e-06, "loss": 0.9254, "step": 44130 }, { "epoch": 0.8028519980356136, "grad_norm": 1.6122387647628784, "learning_rate": 2.0326651540437662e-06, "loss": 0.9043, "step": 44140 }, { "epoch": 0.8030338856654359, "grad_norm": 1.596990942955017, "learning_rate": 2.030790001687637e-06, "loss": 0.8642, "step": 44150 }, { "epoch": 0.8032157732952582, "grad_norm": 1.5284372568130493, "learning_rate": 2.0289148493315085e-06, "loss": 0.8915, "step": 44160 }, { "epoch": 0.8033976609250805, "grad_norm": 1.6142971515655518, "learning_rate": 2.0270396969753794e-06, "loss": 0.8231, "step": 44170 }, { "epoch": 0.8035795485549028, "grad_norm": 1.8757975101470947, "learning_rate": 2.0251645446192504e-06, "loss": 0.9239, "step": 44180 }, { "epoch": 0.8037614361847251, "grad_norm": 1.5529043674468994, "learning_rate": 2.0232893922631217e-06, "loss": 0.8587, "step": 44190 }, { "epoch": 0.8039433238145474, "grad_norm": 2.154627799987793, "learning_rate": 2.0214142399069926e-06, "loss": 0.8868, "step": 44200 }, { "epoch": 0.8041252114443697, "grad_norm": 1.6290653944015503, "learning_rate": 2.0195390875508635e-06, "loss": 0.9324, "step": 44210 }, { "epoch": 0.804307099074192, "grad_norm": 1.7672550678253174, "learning_rate": 2.017663935194735e-06, "loss": 0.8857, "step": 44220 }, { "epoch": 0.8044889867040143, "grad_norm": 1.6443912982940674, "learning_rate": 2.015788782838606e-06, "loss": 0.9036, "step": 44230 }, { "epoch": 0.8046708743338365, "grad_norm": 1.744558572769165, "learning_rate": 2.0139136304824767e-06, "loss": 0.8736, "step": 44240 }, { "epoch": 0.8048527619636588, "grad_norm": 1.8745841979980469, "learning_rate": 2.012038478126348e-06, "loss": 0.9599, "step": 44250 }, { "epoch": 0.8050346495934811, "grad_norm": 1.9033523797988892, "learning_rate": 2.010163325770219e-06, "loss": 0.9015, "step": 44260 }, { "epoch": 0.8052165372233034, "grad_norm": 1.6970300674438477, "learning_rate": 2.00828817341409e-06, "loss": 0.9684, "step": 44270 }, { "epoch": 0.8053984248531257, "grad_norm": 1.711282730102539, "learning_rate": 2.006413021057961e-06, "loss": 0.9226, "step": 44280 }, { "epoch": 0.805580312482948, "grad_norm": 1.7947242259979248, "learning_rate": 2.004537868701832e-06, "loss": 0.9388, "step": 44290 }, { "epoch": 0.8057622001127703, "grad_norm": 2.022362470626831, "learning_rate": 2.002662716345703e-06, "loss": 0.9, "step": 44300 }, { "epoch": 0.8059440877425926, "grad_norm": 1.7889494895935059, "learning_rate": 2.0007875639895745e-06, "loss": 0.8923, "step": 44310 }, { "epoch": 0.8061259753724149, "grad_norm": 1.9051053524017334, "learning_rate": 1.9989124116334454e-06, "loss": 0.898, "step": 44320 }, { "epoch": 0.8063078630022372, "grad_norm": 1.742843747138977, "learning_rate": 1.9970372592773167e-06, "loss": 0.9115, "step": 44330 }, { "epoch": 0.8064897506320595, "grad_norm": 1.5725526809692383, "learning_rate": 1.9951621069211872e-06, "loss": 0.8514, "step": 44340 }, { "epoch": 0.8066716382618818, "grad_norm": 1.8262654542922974, "learning_rate": 1.9932869545650586e-06, "loss": 0.9188, "step": 44350 }, { "epoch": 0.8068535258917041, "grad_norm": 1.680727243423462, "learning_rate": 1.9914118022089295e-06, "loss": 0.9278, "step": 44360 }, { "epoch": 0.8070354135215264, "grad_norm": 1.5791391134262085, "learning_rate": 1.989536649852801e-06, "loss": 0.8448, "step": 44370 }, { "epoch": 0.8072173011513487, "grad_norm": 2.051184892654419, "learning_rate": 1.9876614974966718e-06, "loss": 0.9509, "step": 44380 }, { "epoch": 0.807399188781171, "grad_norm": 1.8539022207260132, "learning_rate": 1.985786345140543e-06, "loss": 0.9415, "step": 44390 }, { "epoch": 0.8075810764109933, "grad_norm": 1.9750038385391235, "learning_rate": 1.983911192784414e-06, "loss": 0.9428, "step": 44400 }, { "epoch": 0.8077629640408156, "grad_norm": 1.4476054906845093, "learning_rate": 1.982036040428285e-06, "loss": 0.9281, "step": 44410 }, { "epoch": 0.8079448516706379, "grad_norm": 1.7063692808151245, "learning_rate": 1.980160888072156e-06, "loss": 0.8751, "step": 44420 }, { "epoch": 0.8081267393004602, "grad_norm": 2.038158655166626, "learning_rate": 1.9782857357160272e-06, "loss": 0.9091, "step": 44430 }, { "epoch": 0.8083086269302825, "grad_norm": 1.8662910461425781, "learning_rate": 1.976410583359898e-06, "loss": 0.8875, "step": 44440 }, { "epoch": 0.8084905145601048, "grad_norm": 1.707326054573059, "learning_rate": 1.974535431003769e-06, "loss": 0.9194, "step": 44450 }, { "epoch": 0.8086724021899271, "grad_norm": 1.7738233804702759, "learning_rate": 1.9726602786476404e-06, "loss": 0.8822, "step": 44460 }, { "epoch": 0.8088542898197494, "grad_norm": 2.1574432849884033, "learning_rate": 1.9707851262915113e-06, "loss": 0.8994, "step": 44470 }, { "epoch": 0.8090361774495717, "grad_norm": 1.7068837881088257, "learning_rate": 1.9689099739353823e-06, "loss": 0.9063, "step": 44480 }, { "epoch": 0.809218065079394, "grad_norm": 1.8208048343658447, "learning_rate": 1.9670348215792536e-06, "loss": 0.897, "step": 44490 }, { "epoch": 0.8093999527092163, "grad_norm": 1.5451215505599976, "learning_rate": 1.9651596692231245e-06, "loss": 0.8671, "step": 44500 }, { "epoch": 0.8095818403390386, "grad_norm": 1.665170669555664, "learning_rate": 1.9632845168669955e-06, "loss": 0.8671, "step": 44510 }, { "epoch": 0.8097637279688609, "grad_norm": 1.5790987014770508, "learning_rate": 1.961409364510867e-06, "loss": 0.9003, "step": 44520 }, { "epoch": 0.8099456155986832, "grad_norm": 1.7878040075302124, "learning_rate": 1.9595342121547377e-06, "loss": 0.9247, "step": 44530 }, { "epoch": 0.8101275032285055, "grad_norm": 2.0879805088043213, "learning_rate": 1.957659059798609e-06, "loss": 0.915, "step": 44540 }, { "epoch": 0.8103093908583278, "grad_norm": 1.8220006227493286, "learning_rate": 1.9557839074424796e-06, "loss": 0.9155, "step": 44550 }, { "epoch": 0.8104912784881501, "grad_norm": 1.9212394952774048, "learning_rate": 1.953908755086351e-06, "loss": 0.9111, "step": 44560 }, { "epoch": 0.8106731661179724, "grad_norm": 1.9031013250350952, "learning_rate": 1.952033602730222e-06, "loss": 0.8956, "step": 44570 }, { "epoch": 0.8108550537477947, "grad_norm": 1.3798933029174805, "learning_rate": 1.950158450374093e-06, "loss": 0.887, "step": 44580 }, { "epoch": 0.811036941377617, "grad_norm": 1.8057076930999756, "learning_rate": 1.948283298017964e-06, "loss": 0.9109, "step": 44590 }, { "epoch": 0.8112188290074392, "grad_norm": 2.0451467037200928, "learning_rate": 1.9464081456618355e-06, "loss": 0.8789, "step": 44600 }, { "epoch": 0.8114007166372615, "grad_norm": 1.882991075515747, "learning_rate": 1.9445329933057064e-06, "loss": 0.852, "step": 44610 }, { "epoch": 0.8115826042670838, "grad_norm": 1.8698128461837769, "learning_rate": 1.9426578409495773e-06, "loss": 0.8878, "step": 44620 }, { "epoch": 0.8117644918969061, "grad_norm": 1.9194303750991821, "learning_rate": 1.9407826885934482e-06, "loss": 0.9612, "step": 44630 }, { "epoch": 0.8119463795267284, "grad_norm": 1.9082269668579102, "learning_rate": 1.9389075362373196e-06, "loss": 0.9153, "step": 44640 }, { "epoch": 0.8121282671565507, "grad_norm": 1.7781450748443604, "learning_rate": 1.9370323838811905e-06, "loss": 0.9345, "step": 44650 }, { "epoch": 0.812310154786373, "grad_norm": 1.6955065727233887, "learning_rate": 1.935157231525062e-06, "loss": 0.8839, "step": 44660 }, { "epoch": 0.8124920424161953, "grad_norm": 1.574458360671997, "learning_rate": 1.9332820791689328e-06, "loss": 0.8685, "step": 44670 }, { "epoch": 0.8126739300460176, "grad_norm": 1.5905708074569702, "learning_rate": 1.9314069268128037e-06, "loss": 0.877, "step": 44680 }, { "epoch": 0.8128558176758399, "grad_norm": 1.709013819694519, "learning_rate": 1.9295317744566746e-06, "loss": 0.9072, "step": 44690 }, { "epoch": 0.8130377053056622, "grad_norm": 1.941942811012268, "learning_rate": 1.927656622100546e-06, "loss": 0.9515, "step": 44700 }, { "epoch": 0.8132195929354844, "grad_norm": 1.7764863967895508, "learning_rate": 1.925781469744417e-06, "loss": 0.9457, "step": 44710 }, { "epoch": 0.8134014805653067, "grad_norm": 1.845354437828064, "learning_rate": 1.923906317388288e-06, "loss": 0.9117, "step": 44720 }, { "epoch": 0.813583368195129, "grad_norm": 2.0215368270874023, "learning_rate": 1.922031165032159e-06, "loss": 0.9, "step": 44730 }, { "epoch": 0.8137652558249513, "grad_norm": 1.6849457025527954, "learning_rate": 1.92015601267603e-06, "loss": 0.8835, "step": 44740 }, { "epoch": 0.8139471434547736, "grad_norm": 1.7370853424072266, "learning_rate": 1.918280860319901e-06, "loss": 0.8695, "step": 44750 }, { "epoch": 0.8141290310845959, "grad_norm": 1.8464813232421875, "learning_rate": 1.916405707963772e-06, "loss": 0.8889, "step": 44760 }, { "epoch": 0.8143109187144182, "grad_norm": 2.107661247253418, "learning_rate": 1.9145305556076433e-06, "loss": 0.9062, "step": 44770 }, { "epoch": 0.8144928063442405, "grad_norm": 1.423943281173706, "learning_rate": 1.912655403251514e-06, "loss": 0.8506, "step": 44780 }, { "epoch": 0.8146746939740628, "grad_norm": 1.7184113264083862, "learning_rate": 1.9107802508953855e-06, "loss": 0.8669, "step": 44790 }, { "epoch": 0.8148565816038851, "grad_norm": 1.8541791439056396, "learning_rate": 1.9089050985392564e-06, "loss": 0.9056, "step": 44800 }, { "epoch": 0.8150384692337074, "grad_norm": 1.9066154956817627, "learning_rate": 1.9070299461831276e-06, "loss": 0.9648, "step": 44810 }, { "epoch": 0.8152203568635297, "grad_norm": 1.7331770658493042, "learning_rate": 1.9051547938269985e-06, "loss": 0.9218, "step": 44820 }, { "epoch": 0.815402244493352, "grad_norm": 1.760361671447754, "learning_rate": 1.9032796414708696e-06, "loss": 0.9109, "step": 44830 }, { "epoch": 0.8155841321231743, "grad_norm": 1.5598036050796509, "learning_rate": 1.9014044891147408e-06, "loss": 0.9221, "step": 44840 }, { "epoch": 0.8157660197529966, "grad_norm": 1.6882387399673462, "learning_rate": 1.8995293367586117e-06, "loss": 0.8673, "step": 44850 }, { "epoch": 0.8159479073828189, "grad_norm": 1.8201713562011719, "learning_rate": 1.8976541844024828e-06, "loss": 0.8472, "step": 44860 }, { "epoch": 0.8161297950126412, "grad_norm": 1.7626882791519165, "learning_rate": 1.895779032046354e-06, "loss": 0.8814, "step": 44870 }, { "epoch": 0.8163116826424635, "grad_norm": 1.6628409624099731, "learning_rate": 1.893903879690225e-06, "loss": 0.9195, "step": 44880 }, { "epoch": 0.8164935702722858, "grad_norm": 1.6599780321121216, "learning_rate": 1.892028727334096e-06, "loss": 0.9138, "step": 44890 }, { "epoch": 0.816675457902108, "grad_norm": 1.6723989248275757, "learning_rate": 1.890153574977967e-06, "loss": 0.8757, "step": 44900 }, { "epoch": 0.8168573455319303, "grad_norm": 1.617401361465454, "learning_rate": 1.888278422621838e-06, "loss": 0.8657, "step": 44910 }, { "epoch": 0.8170392331617526, "grad_norm": 1.5623418092727661, "learning_rate": 1.8864032702657092e-06, "loss": 0.9629, "step": 44920 }, { "epoch": 0.8172211207915749, "grad_norm": 1.811767816543579, "learning_rate": 1.8845281179095803e-06, "loss": 0.9413, "step": 44930 }, { "epoch": 0.8174030084213972, "grad_norm": 1.77046537399292, "learning_rate": 1.8826529655534515e-06, "loss": 0.9105, "step": 44940 }, { "epoch": 0.8175848960512195, "grad_norm": 1.7027602195739746, "learning_rate": 1.8807778131973226e-06, "loss": 0.9403, "step": 44950 }, { "epoch": 0.8177667836810418, "grad_norm": 1.9617947340011597, "learning_rate": 1.8789026608411933e-06, "loss": 0.8719, "step": 44960 }, { "epoch": 0.8179486713108641, "grad_norm": 1.5565698146820068, "learning_rate": 1.8770275084850645e-06, "loss": 0.9453, "step": 44970 }, { "epoch": 0.8181305589406864, "grad_norm": 1.7119390964508057, "learning_rate": 1.8751523561289356e-06, "loss": 0.8941, "step": 44980 }, { "epoch": 0.8183124465705087, "grad_norm": 1.588463306427002, "learning_rate": 1.8732772037728067e-06, "loss": 0.8986, "step": 44990 }, { "epoch": 0.818494334200331, "grad_norm": 1.9405667781829834, "learning_rate": 1.8714020514166779e-06, "loss": 0.9208, "step": 45000 }, { "epoch": 0.8186762218301533, "grad_norm": 1.627561330795288, "learning_rate": 1.869526899060549e-06, "loss": 0.8806, "step": 45010 }, { "epoch": 0.8188581094599756, "grad_norm": 1.495926022529602, "learning_rate": 1.86765174670442e-06, "loss": 0.9075, "step": 45020 }, { "epoch": 0.8190399970897979, "grad_norm": 1.8598580360412598, "learning_rate": 1.8657765943482908e-06, "loss": 0.9188, "step": 45030 }, { "epoch": 0.8192218847196202, "grad_norm": 1.9177981615066528, "learning_rate": 1.863901441992162e-06, "loss": 0.8609, "step": 45040 }, { "epoch": 0.8194037723494425, "grad_norm": 1.5915836095809937, "learning_rate": 1.8620262896360331e-06, "loss": 0.9074, "step": 45050 }, { "epoch": 0.8195856599792648, "grad_norm": 1.677458643913269, "learning_rate": 1.8601511372799042e-06, "loss": 0.9113, "step": 45060 }, { "epoch": 0.8197675476090871, "grad_norm": 1.69912850856781, "learning_rate": 1.8582759849237752e-06, "loss": 0.8836, "step": 45070 }, { "epoch": 0.8199494352389094, "grad_norm": 1.781753659248352, "learning_rate": 1.8564008325676463e-06, "loss": 0.9368, "step": 45080 }, { "epoch": 0.8201313228687317, "grad_norm": 2.2370402812957764, "learning_rate": 1.8545256802115174e-06, "loss": 0.8649, "step": 45090 }, { "epoch": 0.820313210498554, "grad_norm": 1.6365618705749512, "learning_rate": 1.8526505278553884e-06, "loss": 0.8956, "step": 45100 }, { "epoch": 0.8204950981283763, "grad_norm": 1.7817528247833252, "learning_rate": 1.8507753754992595e-06, "loss": 0.9129, "step": 45110 }, { "epoch": 0.8206769857581986, "grad_norm": 1.8317053318023682, "learning_rate": 1.8489002231431304e-06, "loss": 0.8486, "step": 45120 }, { "epoch": 0.8208588733880209, "grad_norm": 1.9757976531982422, "learning_rate": 1.8470250707870016e-06, "loss": 0.9407, "step": 45130 }, { "epoch": 0.8210407610178432, "grad_norm": 2.0251362323760986, "learning_rate": 1.8451499184308727e-06, "loss": 0.8876, "step": 45140 }, { "epoch": 0.8212226486476655, "grad_norm": 1.971279501914978, "learning_rate": 1.8432747660747438e-06, "loss": 0.8778, "step": 45150 }, { "epoch": 0.8214045362774878, "grad_norm": 1.5502376556396484, "learning_rate": 1.841399613718615e-06, "loss": 0.8983, "step": 45160 }, { "epoch": 0.8215864239073101, "grad_norm": 1.8610812425613403, "learning_rate": 1.8395244613624857e-06, "loss": 0.8647, "step": 45170 }, { "epoch": 0.8217683115371324, "grad_norm": 1.9750860929489136, "learning_rate": 1.8376493090063568e-06, "loss": 0.9034, "step": 45180 }, { "epoch": 0.8219501991669547, "grad_norm": 2.406996011734009, "learning_rate": 1.835774156650228e-06, "loss": 0.8473, "step": 45190 }, { "epoch": 0.822132086796777, "grad_norm": 1.9336426258087158, "learning_rate": 1.833899004294099e-06, "loss": 0.914, "step": 45200 }, { "epoch": 0.8223139744265993, "grad_norm": 1.9570890665054321, "learning_rate": 1.8320238519379702e-06, "loss": 0.9094, "step": 45210 }, { "epoch": 0.8224958620564216, "grad_norm": 1.6490387916564941, "learning_rate": 1.8301486995818413e-06, "loss": 0.8498, "step": 45220 }, { "epoch": 0.8226777496862439, "grad_norm": 1.8650176525115967, "learning_rate": 1.8282735472257125e-06, "loss": 0.8967, "step": 45230 }, { "epoch": 0.8228596373160662, "grad_norm": 2.0445950031280518, "learning_rate": 1.8263983948695832e-06, "loss": 0.9267, "step": 45240 }, { "epoch": 0.8230415249458884, "grad_norm": 1.785946249961853, "learning_rate": 1.8245232425134543e-06, "loss": 0.9368, "step": 45250 }, { "epoch": 0.8232234125757107, "grad_norm": 1.9058325290679932, "learning_rate": 1.8226480901573255e-06, "loss": 0.9431, "step": 45260 }, { "epoch": 0.823405300205533, "grad_norm": 1.6929415464401245, "learning_rate": 1.8207729378011966e-06, "loss": 0.9344, "step": 45270 }, { "epoch": 0.8235871878353553, "grad_norm": 1.673387885093689, "learning_rate": 1.8188977854450677e-06, "loss": 0.8709, "step": 45280 }, { "epoch": 0.8237690754651776, "grad_norm": 1.8241955041885376, "learning_rate": 1.8170226330889386e-06, "loss": 0.9216, "step": 45290 }, { "epoch": 0.8239509630949999, "grad_norm": 1.7530510425567627, "learning_rate": 1.8151474807328096e-06, "loss": 0.9234, "step": 45300 }, { "epoch": 0.8241328507248222, "grad_norm": 1.726881980895996, "learning_rate": 1.8132723283766807e-06, "loss": 0.8702, "step": 45310 }, { "epoch": 0.8243147383546445, "grad_norm": 1.8624993562698364, "learning_rate": 1.8113971760205518e-06, "loss": 0.9375, "step": 45320 }, { "epoch": 0.8244966259844668, "grad_norm": 1.7984529733657837, "learning_rate": 1.809522023664423e-06, "loss": 0.8909, "step": 45330 }, { "epoch": 0.8246785136142891, "grad_norm": 1.7877568006515503, "learning_rate": 1.8076468713082939e-06, "loss": 0.891, "step": 45340 }, { "epoch": 0.8248604012441114, "grad_norm": 1.8455404043197632, "learning_rate": 1.805771718952165e-06, "loss": 0.9137, "step": 45350 }, { "epoch": 0.8250422888739337, "grad_norm": 1.9671860933303833, "learning_rate": 1.8038965665960362e-06, "loss": 0.9561, "step": 45360 }, { "epoch": 0.825224176503756, "grad_norm": 2.027639865875244, "learning_rate": 1.802021414239907e-06, "loss": 0.8834, "step": 45370 }, { "epoch": 0.8254060641335783, "grad_norm": 1.9367787837982178, "learning_rate": 1.8001462618837782e-06, "loss": 0.8977, "step": 45380 }, { "epoch": 0.8255879517634006, "grad_norm": 1.8950835466384888, "learning_rate": 1.7982711095276491e-06, "loss": 0.9072, "step": 45390 }, { "epoch": 0.8257698393932229, "grad_norm": 1.689517617225647, "learning_rate": 1.7963959571715203e-06, "loss": 0.853, "step": 45400 }, { "epoch": 0.8259517270230452, "grad_norm": 1.634333848953247, "learning_rate": 1.7945208048153914e-06, "loss": 0.843, "step": 45410 }, { "epoch": 0.8261336146528675, "grad_norm": 1.766903281211853, "learning_rate": 1.7926456524592625e-06, "loss": 0.9023, "step": 45420 }, { "epoch": 0.8263155022826898, "grad_norm": 1.4411345720291138, "learning_rate": 1.7907705001031337e-06, "loss": 0.9101, "step": 45430 }, { "epoch": 0.8264973899125121, "grad_norm": 1.6243600845336914, "learning_rate": 1.7888953477470044e-06, "loss": 0.8474, "step": 45440 }, { "epoch": 0.8266792775423344, "grad_norm": 1.7562636137008667, "learning_rate": 1.7870201953908755e-06, "loss": 0.9018, "step": 45450 }, { "epoch": 0.8268611651721567, "grad_norm": 1.7605478763580322, "learning_rate": 1.7851450430347467e-06, "loss": 0.868, "step": 45460 }, { "epoch": 0.827043052801979, "grad_norm": 1.7102333307266235, "learning_rate": 1.7832698906786178e-06, "loss": 0.8767, "step": 45470 }, { "epoch": 0.8272249404318013, "grad_norm": 2.041778087615967, "learning_rate": 1.781394738322489e-06, "loss": 0.9157, "step": 45480 }, { "epoch": 0.8274068280616236, "grad_norm": 1.905777931213379, "learning_rate": 1.77951958596636e-06, "loss": 0.8921, "step": 45490 }, { "epoch": 0.8275887156914459, "grad_norm": 1.7578328847885132, "learning_rate": 1.7776444336102312e-06, "loss": 0.9143, "step": 45500 }, { "epoch": 0.8277706033212682, "grad_norm": 1.600845456123352, "learning_rate": 1.775769281254102e-06, "loss": 0.9079, "step": 45510 }, { "epoch": 0.8279524909510905, "grad_norm": 1.5442759990692139, "learning_rate": 1.773894128897973e-06, "loss": 0.8659, "step": 45520 }, { "epoch": 0.8281343785809128, "grad_norm": 1.5706710815429688, "learning_rate": 1.7720189765418442e-06, "loss": 0.8681, "step": 45530 }, { "epoch": 0.828316266210735, "grad_norm": 1.7032403945922852, "learning_rate": 1.7701438241857153e-06, "loss": 0.821, "step": 45540 }, { "epoch": 0.8284981538405572, "grad_norm": 1.5703773498535156, "learning_rate": 1.7682686718295864e-06, "loss": 0.9218, "step": 45550 }, { "epoch": 0.8286800414703795, "grad_norm": 1.8635770082473755, "learning_rate": 1.7663935194734574e-06, "loss": 0.9118, "step": 45560 }, { "epoch": 0.8288619291002018, "grad_norm": 1.7056772708892822, "learning_rate": 1.7645183671173285e-06, "loss": 0.8666, "step": 45570 }, { "epoch": 0.8290438167300241, "grad_norm": 1.7045425176620483, "learning_rate": 1.7626432147611994e-06, "loss": 0.9131, "step": 45580 }, { "epoch": 0.8292257043598464, "grad_norm": 1.858498454093933, "learning_rate": 1.7607680624050706e-06, "loss": 0.8649, "step": 45590 }, { "epoch": 0.8294075919896687, "grad_norm": 1.8064758777618408, "learning_rate": 1.7588929100489417e-06, "loss": 0.8999, "step": 45600 }, { "epoch": 0.829589479619491, "grad_norm": 1.4411253929138184, "learning_rate": 1.7570177576928126e-06, "loss": 0.9408, "step": 45610 }, { "epoch": 0.8297713672493133, "grad_norm": 2.161160945892334, "learning_rate": 1.7551426053366837e-06, "loss": 0.9687, "step": 45620 }, { "epoch": 0.8299532548791356, "grad_norm": 1.9632954597473145, "learning_rate": 1.7532674529805549e-06, "loss": 0.8809, "step": 45630 }, { "epoch": 0.8301351425089579, "grad_norm": 1.9131149053573608, "learning_rate": 1.751392300624426e-06, "loss": 0.8875, "step": 45640 }, { "epoch": 0.8303170301387802, "grad_norm": 1.7691329717636108, "learning_rate": 1.7495171482682967e-06, "loss": 0.9372, "step": 45650 }, { "epoch": 0.8304989177686025, "grad_norm": 1.9863189458847046, "learning_rate": 1.7476419959121679e-06, "loss": 0.9136, "step": 45660 }, { "epoch": 0.8306808053984248, "grad_norm": 1.686842441558838, "learning_rate": 1.745766843556039e-06, "loss": 0.9056, "step": 45670 }, { "epoch": 0.8308626930282471, "grad_norm": 3.608361005783081, "learning_rate": 1.7438916911999101e-06, "loss": 0.9091, "step": 45680 }, { "epoch": 0.8310445806580694, "grad_norm": 1.8972378969192505, "learning_rate": 1.7420165388437813e-06, "loss": 0.8904, "step": 45690 }, { "epoch": 0.8312264682878917, "grad_norm": 1.4691672325134277, "learning_rate": 1.7401413864876524e-06, "loss": 0.8918, "step": 45700 }, { "epoch": 0.831408355917714, "grad_norm": 1.9478739500045776, "learning_rate": 1.7382662341315235e-06, "loss": 0.9113, "step": 45710 }, { "epoch": 0.8315902435475363, "grad_norm": 1.8693246841430664, "learning_rate": 1.7363910817753942e-06, "loss": 0.9095, "step": 45720 }, { "epoch": 0.8317721311773586, "grad_norm": 1.919787049293518, "learning_rate": 1.7345159294192654e-06, "loss": 0.9141, "step": 45730 }, { "epoch": 0.8319540188071809, "grad_norm": 1.9091148376464844, "learning_rate": 1.7326407770631365e-06, "loss": 0.8999, "step": 45740 }, { "epoch": 0.8321359064370032, "grad_norm": 1.4702906608581543, "learning_rate": 1.7307656247070076e-06, "loss": 0.9096, "step": 45750 }, { "epoch": 0.8323177940668255, "grad_norm": 1.9031301736831665, "learning_rate": 1.7288904723508788e-06, "loss": 0.955, "step": 45760 }, { "epoch": 0.8324996816966478, "grad_norm": 2.009009838104248, "learning_rate": 1.72701531999475e-06, "loss": 0.9046, "step": 45770 }, { "epoch": 0.8326815693264701, "grad_norm": 2.1217665672302246, "learning_rate": 1.7251401676386208e-06, "loss": 0.896, "step": 45780 }, { "epoch": 0.8328634569562924, "grad_norm": 1.6917264461517334, "learning_rate": 1.7232650152824918e-06, "loss": 0.8979, "step": 45790 }, { "epoch": 0.8330453445861147, "grad_norm": 1.6899542808532715, "learning_rate": 1.7213898629263629e-06, "loss": 0.8416, "step": 45800 }, { "epoch": 0.833227232215937, "grad_norm": 2.0701563358306885, "learning_rate": 1.719514710570234e-06, "loss": 0.9355, "step": 45810 }, { "epoch": 0.8334091198457593, "grad_norm": 1.6863945722579956, "learning_rate": 1.717639558214105e-06, "loss": 0.9461, "step": 45820 }, { "epoch": 0.8335910074755816, "grad_norm": 1.750415563583374, "learning_rate": 1.715764405857976e-06, "loss": 0.8824, "step": 45830 }, { "epoch": 0.8337728951054039, "grad_norm": 1.5878591537475586, "learning_rate": 1.7138892535018472e-06, "loss": 0.8971, "step": 45840 }, { "epoch": 0.8339547827352262, "grad_norm": 1.8241804838180542, "learning_rate": 1.7120141011457183e-06, "loss": 0.9022, "step": 45850 }, { "epoch": 0.8341366703650485, "grad_norm": 1.6398735046386719, "learning_rate": 1.7101389487895893e-06, "loss": 0.9208, "step": 45860 }, { "epoch": 0.8343185579948708, "grad_norm": 1.7644445896148682, "learning_rate": 1.7082637964334602e-06, "loss": 0.8908, "step": 45870 }, { "epoch": 0.834500445624693, "grad_norm": 1.7306506633758545, "learning_rate": 1.7063886440773313e-06, "loss": 0.8788, "step": 45880 }, { "epoch": 0.8346823332545154, "grad_norm": 1.7694214582443237, "learning_rate": 1.7045134917212025e-06, "loss": 0.8888, "step": 45890 }, { "epoch": 0.8348642208843376, "grad_norm": 2.0004711151123047, "learning_rate": 1.7026383393650736e-06, "loss": 0.915, "step": 45900 }, { "epoch": 0.83504610851416, "grad_norm": 1.729305386543274, "learning_rate": 1.7007631870089447e-06, "loss": 0.9047, "step": 45910 }, { "epoch": 0.8352279961439822, "grad_norm": 1.8754594326019287, "learning_rate": 1.6988880346528154e-06, "loss": 0.929, "step": 45920 }, { "epoch": 0.8354098837738045, "grad_norm": 1.6761198043823242, "learning_rate": 1.6970128822966866e-06, "loss": 0.9007, "step": 45930 }, { "epoch": 0.8355917714036268, "grad_norm": 1.587982177734375, "learning_rate": 1.6951377299405577e-06, "loss": 0.9505, "step": 45940 }, { "epoch": 0.8357736590334491, "grad_norm": 1.7194969654083252, "learning_rate": 1.6932625775844288e-06, "loss": 0.893, "step": 45950 }, { "epoch": 0.8359555466632714, "grad_norm": 1.5771125555038452, "learning_rate": 1.6913874252283e-06, "loss": 0.9358, "step": 45960 }, { "epoch": 0.8361374342930937, "grad_norm": 2.2855162620544434, "learning_rate": 1.6895122728721711e-06, "loss": 0.9182, "step": 45970 }, { "epoch": 0.836319321922916, "grad_norm": 2.0989303588867188, "learning_rate": 1.6876371205160422e-06, "loss": 0.855, "step": 45980 }, { "epoch": 0.8365012095527383, "grad_norm": 1.7607722282409668, "learning_rate": 1.685761968159913e-06, "loss": 0.896, "step": 45990 }, { "epoch": 0.8366830971825606, "grad_norm": 1.6783257722854614, "learning_rate": 1.683886815803784e-06, "loss": 0.9061, "step": 46000 }, { "epoch": 0.8368649848123829, "grad_norm": 1.8708394765853882, "learning_rate": 1.6820116634476552e-06, "loss": 0.9201, "step": 46010 }, { "epoch": 0.8370468724422052, "grad_norm": 1.7289897203445435, "learning_rate": 1.6801365110915264e-06, "loss": 0.9371, "step": 46020 }, { "epoch": 0.8372287600720275, "grad_norm": 2.0834784507751465, "learning_rate": 1.6782613587353975e-06, "loss": 0.9071, "step": 46030 }, { "epoch": 0.8374106477018498, "grad_norm": 1.8862941265106201, "learning_rate": 1.6763862063792684e-06, "loss": 0.8893, "step": 46040 }, { "epoch": 0.8375925353316721, "grad_norm": 1.7346763610839844, "learning_rate": 1.6745110540231396e-06, "loss": 0.8419, "step": 46050 }, { "epoch": 0.8377744229614944, "grad_norm": 1.7481769323349, "learning_rate": 1.6726359016670105e-06, "loss": 0.8731, "step": 46060 }, { "epoch": 0.8379563105913167, "grad_norm": 1.7097737789154053, "learning_rate": 1.6707607493108816e-06, "loss": 0.8813, "step": 46070 }, { "epoch": 0.838138198221139, "grad_norm": 1.637918472290039, "learning_rate": 1.6688855969547527e-06, "loss": 0.9097, "step": 46080 }, { "epoch": 0.8383200858509613, "grad_norm": 2.037473440170288, "learning_rate": 1.6670104445986237e-06, "loss": 0.8705, "step": 46090 }, { "epoch": 0.8385019734807836, "grad_norm": 1.948991060256958, "learning_rate": 1.6651352922424948e-06, "loss": 0.9158, "step": 46100 }, { "epoch": 0.8386838611106059, "grad_norm": 1.6718732118606567, "learning_rate": 1.663260139886366e-06, "loss": 0.9575, "step": 46110 }, { "epoch": 0.8388657487404282, "grad_norm": 1.8737668991088867, "learning_rate": 1.661384987530237e-06, "loss": 0.9229, "step": 46120 }, { "epoch": 0.8390476363702505, "grad_norm": 2.021787405014038, "learning_rate": 1.659509835174108e-06, "loss": 0.9044, "step": 46130 }, { "epoch": 0.8392295240000728, "grad_norm": 1.9651635885238647, "learning_rate": 1.657634682817979e-06, "loss": 0.9241, "step": 46140 }, { "epoch": 0.8394114116298951, "grad_norm": 1.5412344932556152, "learning_rate": 1.65575953046185e-06, "loss": 0.8882, "step": 46150 }, { "epoch": 0.8395932992597174, "grad_norm": 1.7628792524337769, "learning_rate": 1.6538843781057212e-06, "loss": 0.9319, "step": 46160 }, { "epoch": 0.8397751868895397, "grad_norm": 1.565423846244812, "learning_rate": 1.6520092257495923e-06, "loss": 0.9119, "step": 46170 }, { "epoch": 0.839957074519362, "grad_norm": 1.5338667631149292, "learning_rate": 1.6501340733934635e-06, "loss": 0.9078, "step": 46180 }, { "epoch": 0.8401389621491843, "grad_norm": 1.772363543510437, "learning_rate": 1.6482589210373346e-06, "loss": 0.892, "step": 46190 }, { "epoch": 0.8403208497790066, "grad_norm": 1.5643211603164673, "learning_rate": 1.6463837686812053e-06, "loss": 0.9164, "step": 46200 }, { "epoch": 0.8405027374088289, "grad_norm": 1.8098127841949463, "learning_rate": 1.6445086163250764e-06, "loss": 0.9027, "step": 46210 }, { "epoch": 0.8406846250386512, "grad_norm": 1.5178920030593872, "learning_rate": 1.6426334639689476e-06, "loss": 0.8685, "step": 46220 }, { "epoch": 0.8408665126684735, "grad_norm": 1.7321058511734009, "learning_rate": 1.6407583116128187e-06, "loss": 0.9131, "step": 46230 }, { "epoch": 0.8410484002982958, "grad_norm": 1.861142873764038, "learning_rate": 1.6388831592566898e-06, "loss": 0.8996, "step": 46240 }, { "epoch": 0.841230287928118, "grad_norm": 1.6858949661254883, "learning_rate": 1.637008006900561e-06, "loss": 0.8744, "step": 46250 }, { "epoch": 0.8414121755579403, "grad_norm": 2.0355873107910156, "learning_rate": 1.6351328545444319e-06, "loss": 0.9161, "step": 46260 }, { "epoch": 0.8415940631877626, "grad_norm": 1.7796436548233032, "learning_rate": 1.6332577021883028e-06, "loss": 0.9029, "step": 46270 }, { "epoch": 0.8417759508175849, "grad_norm": 1.8385387659072876, "learning_rate": 1.631382549832174e-06, "loss": 0.9058, "step": 46280 }, { "epoch": 0.8419578384474072, "grad_norm": 2.0174827575683594, "learning_rate": 1.629507397476045e-06, "loss": 0.9364, "step": 46290 }, { "epoch": 0.8421397260772295, "grad_norm": 1.8372151851654053, "learning_rate": 1.6276322451199162e-06, "loss": 0.9397, "step": 46300 }, { "epoch": 0.8423216137070518, "grad_norm": 2.044064521789551, "learning_rate": 1.6257570927637871e-06, "loss": 0.9349, "step": 46310 }, { "epoch": 0.8425035013368741, "grad_norm": 1.6412426233291626, "learning_rate": 1.6238819404076583e-06, "loss": 0.923, "step": 46320 }, { "epoch": 0.8426853889666964, "grad_norm": 1.5479168891906738, "learning_rate": 1.6220067880515294e-06, "loss": 0.9246, "step": 46330 }, { "epoch": 0.8428672765965187, "grad_norm": 1.7140381336212158, "learning_rate": 1.6201316356954003e-06, "loss": 0.8887, "step": 46340 }, { "epoch": 0.843049164226341, "grad_norm": 1.827712893486023, "learning_rate": 1.6182564833392715e-06, "loss": 0.8984, "step": 46350 }, { "epoch": 0.8432310518561633, "grad_norm": 1.6561492681503296, "learning_rate": 1.6163813309831424e-06, "loss": 0.8861, "step": 46360 }, { "epoch": 0.8434129394859856, "grad_norm": 1.9708144664764404, "learning_rate": 1.6145061786270135e-06, "loss": 0.8702, "step": 46370 }, { "epoch": 0.8435948271158078, "grad_norm": 1.9172801971435547, "learning_rate": 1.6126310262708847e-06, "loss": 0.9335, "step": 46380 }, { "epoch": 0.8437767147456301, "grad_norm": 1.5856781005859375, "learning_rate": 1.6107558739147558e-06, "loss": 0.9083, "step": 46390 }, { "epoch": 0.8439586023754524, "grad_norm": 1.853096604347229, "learning_rate": 1.608880721558627e-06, "loss": 0.896, "step": 46400 }, { "epoch": 0.8441404900052747, "grad_norm": 2.00207257270813, "learning_rate": 1.6070055692024976e-06, "loss": 0.9285, "step": 46410 }, { "epoch": 0.844322377635097, "grad_norm": 1.5327212810516357, "learning_rate": 1.6051304168463688e-06, "loss": 0.8925, "step": 46420 }, { "epoch": 0.8445042652649193, "grad_norm": 1.5268219709396362, "learning_rate": 1.60325526449024e-06, "loss": 0.9074, "step": 46430 }, { "epoch": 0.8446861528947416, "grad_norm": 1.7001073360443115, "learning_rate": 1.601380112134111e-06, "loss": 0.8747, "step": 46440 }, { "epoch": 0.8448680405245639, "grad_norm": 1.8167352676391602, "learning_rate": 1.5995049597779822e-06, "loss": 0.9101, "step": 46450 }, { "epoch": 0.8450499281543862, "grad_norm": 1.9709932804107666, "learning_rate": 1.5976298074218533e-06, "loss": 0.8494, "step": 46460 }, { "epoch": 0.8452318157842085, "grad_norm": 1.7178077697753906, "learning_rate": 1.5957546550657244e-06, "loss": 0.9157, "step": 46470 }, { "epoch": 0.8454137034140308, "grad_norm": 1.7152385711669922, "learning_rate": 1.5938795027095952e-06, "loss": 0.9134, "step": 46480 }, { "epoch": 0.8455955910438531, "grad_norm": 1.824422836303711, "learning_rate": 1.5920043503534663e-06, "loss": 0.8811, "step": 46490 }, { "epoch": 0.8457774786736754, "grad_norm": 1.8230528831481934, "learning_rate": 1.5901291979973374e-06, "loss": 0.8759, "step": 46500 }, { "epoch": 0.8459593663034977, "grad_norm": 1.5380420684814453, "learning_rate": 1.5882540456412086e-06, "loss": 0.8951, "step": 46510 }, { "epoch": 0.84614125393332, "grad_norm": 1.7538009881973267, "learning_rate": 1.5863788932850797e-06, "loss": 0.9044, "step": 46520 }, { "epoch": 0.8463231415631423, "grad_norm": 1.7480829954147339, "learning_rate": 1.5845037409289506e-06, "loss": 0.8728, "step": 46530 }, { "epoch": 0.8465050291929646, "grad_norm": 1.915043592453003, "learning_rate": 1.5826285885728215e-06, "loss": 0.9203, "step": 46540 }, { "epoch": 0.8466869168227869, "grad_norm": 1.6775882244110107, "learning_rate": 1.5807534362166927e-06, "loss": 0.9394, "step": 46550 }, { "epoch": 0.8468688044526091, "grad_norm": 1.5777791738510132, "learning_rate": 1.5788782838605638e-06, "loss": 0.863, "step": 46560 }, { "epoch": 0.8470506920824314, "grad_norm": 2.0322792530059814, "learning_rate": 1.577003131504435e-06, "loss": 0.9454, "step": 46570 }, { "epoch": 0.8472325797122537, "grad_norm": 1.7240757942199707, "learning_rate": 1.5751279791483059e-06, "loss": 0.8313, "step": 46580 }, { "epoch": 0.847414467342076, "grad_norm": 1.9246574640274048, "learning_rate": 1.573252826792177e-06, "loss": 0.9197, "step": 46590 }, { "epoch": 0.8475963549718983, "grad_norm": 1.802577018737793, "learning_rate": 1.5713776744360481e-06, "loss": 0.9391, "step": 46600 }, { "epoch": 0.8477782426017206, "grad_norm": 1.7070770263671875, "learning_rate": 1.569502522079919e-06, "loss": 0.875, "step": 46610 }, { "epoch": 0.8479601302315429, "grad_norm": 1.948799967765808, "learning_rate": 1.5676273697237902e-06, "loss": 0.9111, "step": 46620 }, { "epoch": 0.8481420178613652, "grad_norm": 1.6612634658813477, "learning_rate": 1.5657522173676611e-06, "loss": 0.8401, "step": 46630 }, { "epoch": 0.8483239054911875, "grad_norm": 1.6038167476654053, "learning_rate": 1.5638770650115322e-06, "loss": 0.9012, "step": 46640 }, { "epoch": 0.8485057931210098, "grad_norm": 1.8618981838226318, "learning_rate": 1.5620019126554034e-06, "loss": 0.8672, "step": 46650 }, { "epoch": 0.8486876807508321, "grad_norm": 1.795745611190796, "learning_rate": 1.5601267602992745e-06, "loss": 0.9092, "step": 46660 }, { "epoch": 0.8488695683806544, "grad_norm": 1.58577299118042, "learning_rate": 1.5582516079431456e-06, "loss": 0.9214, "step": 46670 }, { "epoch": 0.8490514560104767, "grad_norm": 1.9421213865280151, "learning_rate": 1.5563764555870164e-06, "loss": 0.8468, "step": 46680 }, { "epoch": 0.849233343640299, "grad_norm": 1.6970014572143555, "learning_rate": 1.5545013032308875e-06, "loss": 0.8926, "step": 46690 }, { "epoch": 0.8494152312701213, "grad_norm": 1.4808770418167114, "learning_rate": 1.5526261508747586e-06, "loss": 0.844, "step": 46700 }, { "epoch": 0.8495971188999436, "grad_norm": 1.687977910041809, "learning_rate": 1.5507509985186298e-06, "loss": 0.8924, "step": 46710 }, { "epoch": 0.8497790065297659, "grad_norm": 1.685655117034912, "learning_rate": 1.5488758461625009e-06, "loss": 0.8868, "step": 46720 }, { "epoch": 0.8499608941595882, "grad_norm": 1.4672225713729858, "learning_rate": 1.547000693806372e-06, "loss": 0.8246, "step": 46730 }, { "epoch": 0.8501427817894105, "grad_norm": 1.9655958414077759, "learning_rate": 1.5451255414502432e-06, "loss": 0.934, "step": 46740 }, { "epoch": 0.8503246694192328, "grad_norm": 1.7282836437225342, "learning_rate": 1.5432503890941139e-06, "loss": 0.8837, "step": 46750 }, { "epoch": 0.8505065570490551, "grad_norm": 1.7622408866882324, "learning_rate": 1.541375236737985e-06, "loss": 0.8907, "step": 46760 }, { "epoch": 0.8506884446788774, "grad_norm": 1.8384699821472168, "learning_rate": 1.5395000843818561e-06, "loss": 0.9517, "step": 46770 }, { "epoch": 0.8508703323086997, "grad_norm": 1.8208422660827637, "learning_rate": 1.5376249320257273e-06, "loss": 0.9378, "step": 46780 }, { "epoch": 0.851052219938522, "grad_norm": 1.6693705320358276, "learning_rate": 1.5357497796695984e-06, "loss": 0.9137, "step": 46790 }, { "epoch": 0.8512341075683443, "grad_norm": 1.9139161109924316, "learning_rate": 1.5338746273134693e-06, "loss": 0.9327, "step": 46800 }, { "epoch": 0.8514159951981666, "grad_norm": 1.9386764764785767, "learning_rate": 1.5319994749573405e-06, "loss": 0.8753, "step": 46810 }, { "epoch": 0.8515978828279889, "grad_norm": 1.486682415008545, "learning_rate": 1.5301243226012114e-06, "loss": 0.8662, "step": 46820 }, { "epoch": 0.8517797704578112, "grad_norm": 2.004983425140381, "learning_rate": 1.5282491702450825e-06, "loss": 0.9593, "step": 46830 }, { "epoch": 0.8519616580876335, "grad_norm": 2.051278829574585, "learning_rate": 1.5263740178889537e-06, "loss": 0.9337, "step": 46840 }, { "epoch": 0.8521435457174558, "grad_norm": 1.8978350162506104, "learning_rate": 1.5244988655328246e-06, "loss": 0.8577, "step": 46850 }, { "epoch": 0.8523254333472781, "grad_norm": 2.168947458267212, "learning_rate": 1.5226237131766957e-06, "loss": 0.895, "step": 46860 }, { "epoch": 0.8525073209771004, "grad_norm": 2.1260039806365967, "learning_rate": 1.5207485608205668e-06, "loss": 0.9075, "step": 46870 }, { "epoch": 0.8526892086069227, "grad_norm": 1.896706461906433, "learning_rate": 1.518873408464438e-06, "loss": 0.8859, "step": 46880 }, { "epoch": 0.852871096236745, "grad_norm": 1.9242801666259766, "learning_rate": 1.516998256108309e-06, "loss": 0.9218, "step": 46890 }, { "epoch": 0.8530529838665672, "grad_norm": 1.5217615365982056, "learning_rate": 1.5151231037521798e-06, "loss": 0.8668, "step": 46900 }, { "epoch": 0.8532348714963895, "grad_norm": 1.9427827596664429, "learning_rate": 1.513247951396051e-06, "loss": 0.8824, "step": 46910 }, { "epoch": 0.8534167591262118, "grad_norm": 1.8184987306594849, "learning_rate": 1.511372799039922e-06, "loss": 0.8858, "step": 46920 }, { "epoch": 0.8535986467560341, "grad_norm": 1.8070068359375, "learning_rate": 1.5094976466837932e-06, "loss": 0.9266, "step": 46930 }, { "epoch": 0.8537805343858564, "grad_norm": 2.0484883785247803, "learning_rate": 1.5076224943276644e-06, "loss": 0.9543, "step": 46940 }, { "epoch": 0.8539624220156787, "grad_norm": 1.6942461729049683, "learning_rate": 1.5057473419715355e-06, "loss": 0.9101, "step": 46950 }, { "epoch": 0.854144309645501, "grad_norm": 1.7476123571395874, "learning_rate": 1.5038721896154062e-06, "loss": 0.867, "step": 46960 }, { "epoch": 0.8543261972753233, "grad_norm": 2.1844003200531006, "learning_rate": 1.5019970372592773e-06, "loss": 0.9106, "step": 46970 }, { "epoch": 0.8545080849051456, "grad_norm": 1.7911380529403687, "learning_rate": 1.5001218849031485e-06, "loss": 0.8729, "step": 46980 }, { "epoch": 0.8546899725349679, "grad_norm": 1.9447240829467773, "learning_rate": 1.4982467325470196e-06, "loss": 0.8964, "step": 46990 }, { "epoch": 0.8548718601647902, "grad_norm": 1.6648777723312378, "learning_rate": 1.4963715801908907e-06, "loss": 0.8847, "step": 47000 }, { "epoch": 0.8550537477946125, "grad_norm": 1.8111579418182373, "learning_rate": 1.4944964278347619e-06, "loss": 0.8886, "step": 47010 }, { "epoch": 0.8552356354244348, "grad_norm": 1.9380193948745728, "learning_rate": 1.4926212754786328e-06, "loss": 0.8693, "step": 47020 }, { "epoch": 0.8554175230542571, "grad_norm": 1.6359202861785889, "learning_rate": 1.4907461231225037e-06, "loss": 0.9185, "step": 47030 }, { "epoch": 0.8555994106840794, "grad_norm": 1.9616470336914062, "learning_rate": 1.4888709707663749e-06, "loss": 0.9102, "step": 47040 }, { "epoch": 0.8557812983139017, "grad_norm": 1.759590744972229, "learning_rate": 1.486995818410246e-06, "loss": 0.8851, "step": 47050 }, { "epoch": 0.855963185943724, "grad_norm": 2.102386236190796, "learning_rate": 1.4851206660541171e-06, "loss": 0.856, "step": 47060 }, { "epoch": 0.8561450735735463, "grad_norm": 1.6090333461761475, "learning_rate": 1.483245513697988e-06, "loss": 0.8972, "step": 47070 }, { "epoch": 0.8563269612033686, "grad_norm": 1.8703964948654175, "learning_rate": 1.4813703613418592e-06, "loss": 0.9289, "step": 47080 }, { "epoch": 0.8565088488331909, "grad_norm": 1.8084806203842163, "learning_rate": 1.4794952089857303e-06, "loss": 0.8876, "step": 47090 }, { "epoch": 0.8566907364630132, "grad_norm": 1.9502842426300049, "learning_rate": 1.4776200566296012e-06, "loss": 0.8847, "step": 47100 }, { "epoch": 0.8568726240928355, "grad_norm": 1.938835859298706, "learning_rate": 1.4757449042734724e-06, "loss": 0.897, "step": 47110 }, { "epoch": 0.8570545117226578, "grad_norm": 2.05036997795105, "learning_rate": 1.4738697519173433e-06, "loss": 0.8905, "step": 47120 }, { "epoch": 0.8572363993524801, "grad_norm": 1.6561862230300903, "learning_rate": 1.4719945995612144e-06, "loss": 0.8881, "step": 47130 }, { "epoch": 0.8574182869823024, "grad_norm": 1.542962670326233, "learning_rate": 1.4701194472050856e-06, "loss": 0.9201, "step": 47140 }, { "epoch": 0.8576001746121247, "grad_norm": 1.6383963823318481, "learning_rate": 1.4682442948489567e-06, "loss": 0.8819, "step": 47150 }, { "epoch": 0.857782062241947, "grad_norm": 2.351818323135376, "learning_rate": 1.4663691424928276e-06, "loss": 0.9113, "step": 47160 }, { "epoch": 0.8579639498717693, "grad_norm": 1.492540955543518, "learning_rate": 1.4644939901366985e-06, "loss": 0.9267, "step": 47170 }, { "epoch": 0.8581458375015916, "grad_norm": 1.704263687133789, "learning_rate": 1.4626188377805697e-06, "loss": 0.8998, "step": 47180 }, { "epoch": 0.8583277251314139, "grad_norm": 1.7587172985076904, "learning_rate": 1.4607436854244408e-06, "loss": 0.9325, "step": 47190 }, { "epoch": 0.8585096127612362, "grad_norm": 1.7278181314468384, "learning_rate": 1.458868533068312e-06, "loss": 0.8872, "step": 47200 }, { "epoch": 0.8586915003910583, "grad_norm": 1.9696022272109985, "learning_rate": 1.456993380712183e-06, "loss": 0.88, "step": 47210 }, { "epoch": 0.8588733880208806, "grad_norm": 1.534305214881897, "learning_rate": 1.4551182283560542e-06, "loss": 0.9003, "step": 47220 }, { "epoch": 0.8590552756507029, "grad_norm": 1.686722755432129, "learning_rate": 1.453243075999925e-06, "loss": 0.8947, "step": 47230 }, { "epoch": 0.8592371632805252, "grad_norm": 1.3928815126419067, "learning_rate": 1.451367923643796e-06, "loss": 0.9001, "step": 47240 }, { "epoch": 0.8594190509103475, "grad_norm": 1.8042067289352417, "learning_rate": 1.4494927712876672e-06, "loss": 0.9011, "step": 47250 }, { "epoch": 0.8596009385401698, "grad_norm": 1.8555521965026855, "learning_rate": 1.4476176189315383e-06, "loss": 0.9098, "step": 47260 }, { "epoch": 0.8597828261699921, "grad_norm": 8.642380714416504, "learning_rate": 1.4457424665754095e-06, "loss": 0.9172, "step": 47270 }, { "epoch": 0.8599647137998144, "grad_norm": 1.8814316987991333, "learning_rate": 1.4438673142192806e-06, "loss": 0.9177, "step": 47280 }, { "epoch": 0.8601466014296367, "grad_norm": 1.8904975652694702, "learning_rate": 1.4419921618631515e-06, "loss": 0.939, "step": 47290 }, { "epoch": 0.860328489059459, "grad_norm": 1.6905385255813599, "learning_rate": 1.4401170095070224e-06, "loss": 0.8359, "step": 47300 }, { "epoch": 0.8605103766892813, "grad_norm": 1.8164564371109009, "learning_rate": 1.4382418571508936e-06, "loss": 0.9061, "step": 47310 }, { "epoch": 0.8606922643191036, "grad_norm": 1.8230478763580322, "learning_rate": 1.4363667047947647e-06, "loss": 0.9022, "step": 47320 }, { "epoch": 0.8608741519489259, "grad_norm": 1.630476474761963, "learning_rate": 1.4344915524386358e-06, "loss": 0.8861, "step": 47330 }, { "epoch": 0.8610560395787482, "grad_norm": 1.9157793521881104, "learning_rate": 1.4326164000825068e-06, "loss": 0.9037, "step": 47340 }, { "epoch": 0.8612379272085705, "grad_norm": 1.8326528072357178, "learning_rate": 1.430741247726378e-06, "loss": 0.9295, "step": 47350 }, { "epoch": 0.8614198148383928, "grad_norm": 2.0506675243377686, "learning_rate": 1.428866095370249e-06, "loss": 0.9004, "step": 47360 }, { "epoch": 0.8616017024682151, "grad_norm": 1.9227102994918823, "learning_rate": 1.42699094301412e-06, "loss": 0.9447, "step": 47370 }, { "epoch": 0.8617835900980374, "grad_norm": 2.0652763843536377, "learning_rate": 1.425115790657991e-06, "loss": 0.8965, "step": 47380 }, { "epoch": 0.8619654777278597, "grad_norm": 1.9539414644241333, "learning_rate": 1.423240638301862e-06, "loss": 0.8959, "step": 47390 }, { "epoch": 0.862147365357682, "grad_norm": 1.737230658531189, "learning_rate": 1.4213654859457332e-06, "loss": 0.9052, "step": 47400 }, { "epoch": 0.8623292529875043, "grad_norm": 1.8643970489501953, "learning_rate": 1.4194903335896043e-06, "loss": 0.9229, "step": 47410 }, { "epoch": 0.8625111406173266, "grad_norm": 1.8586868047714233, "learning_rate": 1.4176151812334754e-06, "loss": 0.9021, "step": 47420 }, { "epoch": 0.8626930282471489, "grad_norm": 1.6353585720062256, "learning_rate": 1.4157400288773466e-06, "loss": 0.8811, "step": 47430 }, { "epoch": 0.8628749158769712, "grad_norm": 1.7509915828704834, "learning_rate": 1.4138648765212173e-06, "loss": 0.9377, "step": 47440 }, { "epoch": 0.8630568035067935, "grad_norm": 1.8273578882217407, "learning_rate": 1.4119897241650884e-06, "loss": 0.9429, "step": 47450 }, { "epoch": 0.8632386911366158, "grad_norm": 1.5357245206832886, "learning_rate": 1.4101145718089595e-06, "loss": 0.8842, "step": 47460 }, { "epoch": 0.8634205787664381, "grad_norm": 1.456519603729248, "learning_rate": 1.4082394194528307e-06, "loss": 0.8713, "step": 47470 }, { "epoch": 0.8636024663962604, "grad_norm": 1.7181285619735718, "learning_rate": 1.4063642670967018e-06, "loss": 0.8862, "step": 47480 }, { "epoch": 0.8637843540260827, "grad_norm": 2.110457181930542, "learning_rate": 1.404489114740573e-06, "loss": 0.9014, "step": 47490 }, { "epoch": 0.863966241655905, "grad_norm": 1.8342326879501343, "learning_rate": 1.402613962384444e-06, "loss": 0.8692, "step": 47500 }, { "epoch": 0.8641481292857273, "grad_norm": 1.7618730068206787, "learning_rate": 1.4007388100283148e-06, "loss": 0.8833, "step": 47510 }, { "epoch": 0.8643300169155496, "grad_norm": 1.6208598613739014, "learning_rate": 1.398863657672186e-06, "loss": 0.9046, "step": 47520 }, { "epoch": 0.8645119045453719, "grad_norm": 1.596264123916626, "learning_rate": 1.396988505316057e-06, "loss": 0.8883, "step": 47530 }, { "epoch": 0.8646937921751942, "grad_norm": 2.324251413345337, "learning_rate": 1.3951133529599282e-06, "loss": 0.9114, "step": 47540 }, { "epoch": 0.8648756798050165, "grad_norm": 1.9712543487548828, "learning_rate": 1.3932382006037993e-06, "loss": 0.9212, "step": 47550 }, { "epoch": 0.8650575674348387, "grad_norm": 1.9031940698623657, "learning_rate": 1.3913630482476702e-06, "loss": 0.9844, "step": 47560 }, { "epoch": 0.865239455064661, "grad_norm": 1.8241022825241089, "learning_rate": 1.3894878958915414e-06, "loss": 0.9142, "step": 47570 }, { "epoch": 0.8654213426944833, "grad_norm": 1.6783117055892944, "learning_rate": 1.3876127435354123e-06, "loss": 0.879, "step": 47580 }, { "epoch": 0.8656032303243056, "grad_norm": 1.6201047897338867, "learning_rate": 1.3857375911792834e-06, "loss": 0.9154, "step": 47590 }, { "epoch": 0.8657851179541279, "grad_norm": 1.5437127351760864, "learning_rate": 1.3838624388231546e-06, "loss": 0.896, "step": 47600 }, { "epoch": 0.8659670055839502, "grad_norm": 1.8303335905075073, "learning_rate": 1.3819872864670255e-06, "loss": 0.8657, "step": 47610 }, { "epoch": 0.8661488932137725, "grad_norm": 1.6360700130462646, "learning_rate": 1.3801121341108966e-06, "loss": 0.8819, "step": 47620 }, { "epoch": 0.8663307808435948, "grad_norm": 1.8034600019454956, "learning_rate": 1.3782369817547678e-06, "loss": 0.9115, "step": 47630 }, { "epoch": 0.8665126684734171, "grad_norm": 1.757656455039978, "learning_rate": 1.3763618293986389e-06, "loss": 0.8456, "step": 47640 }, { "epoch": 0.8666945561032394, "grad_norm": 1.7960538864135742, "learning_rate": 1.3744866770425098e-06, "loss": 0.884, "step": 47650 }, { "epoch": 0.8668764437330617, "grad_norm": 1.978799819946289, "learning_rate": 1.3726115246863807e-06, "loss": 0.9116, "step": 47660 }, { "epoch": 0.867058331362884, "grad_norm": 1.6369656324386597, "learning_rate": 1.3707363723302519e-06, "loss": 0.8988, "step": 47670 }, { "epoch": 0.8672402189927063, "grad_norm": 1.7737622261047363, "learning_rate": 1.368861219974123e-06, "loss": 0.8913, "step": 47680 }, { "epoch": 0.8674221066225286, "grad_norm": 2.060060501098633, "learning_rate": 1.3669860676179941e-06, "loss": 0.8956, "step": 47690 }, { "epoch": 0.8676039942523509, "grad_norm": 1.9154603481292725, "learning_rate": 1.3651109152618653e-06, "loss": 0.8947, "step": 47700 }, { "epoch": 0.8677858818821732, "grad_norm": 1.9789036512374878, "learning_rate": 1.3632357629057364e-06, "loss": 0.9016, "step": 47710 }, { "epoch": 0.8679677695119955, "grad_norm": 1.8004735708236694, "learning_rate": 1.3613606105496071e-06, "loss": 0.8961, "step": 47720 }, { "epoch": 0.8681496571418178, "grad_norm": 1.6868882179260254, "learning_rate": 1.3594854581934783e-06, "loss": 0.8729, "step": 47730 }, { "epoch": 0.8683315447716401, "grad_norm": 1.7516523599624634, "learning_rate": 1.3576103058373494e-06, "loss": 0.9085, "step": 47740 }, { "epoch": 0.8685134324014624, "grad_norm": 1.9671599864959717, "learning_rate": 1.3557351534812205e-06, "loss": 0.8794, "step": 47750 }, { "epoch": 0.8686953200312847, "grad_norm": 1.8379542827606201, "learning_rate": 1.3538600011250917e-06, "loss": 0.9064, "step": 47760 }, { "epoch": 0.868877207661107, "grad_norm": 1.9927996397018433, "learning_rate": 1.3519848487689628e-06, "loss": 0.898, "step": 47770 }, { "epoch": 0.8690590952909293, "grad_norm": 1.8128291368484497, "learning_rate": 1.3501096964128335e-06, "loss": 0.9153, "step": 47780 }, { "epoch": 0.8692409829207516, "grad_norm": 1.751623511314392, "learning_rate": 1.3482345440567046e-06, "loss": 0.8691, "step": 47790 }, { "epoch": 0.8694228705505739, "grad_norm": 1.642625093460083, "learning_rate": 1.3463593917005758e-06, "loss": 0.8673, "step": 47800 }, { "epoch": 0.8696047581803962, "grad_norm": 1.6479588747024536, "learning_rate": 1.344484239344447e-06, "loss": 0.9098, "step": 47810 }, { "epoch": 0.8697866458102185, "grad_norm": 1.7840758562088013, "learning_rate": 1.342609086988318e-06, "loss": 0.9189, "step": 47820 }, { "epoch": 0.8699685334400408, "grad_norm": 1.8536478281021118, "learning_rate": 1.340733934632189e-06, "loss": 0.9237, "step": 47830 }, { "epoch": 0.8701504210698631, "grad_norm": 1.9401061534881592, "learning_rate": 1.33885878227606e-06, "loss": 0.8942, "step": 47840 }, { "epoch": 0.8703323086996854, "grad_norm": 2.060764789581299, "learning_rate": 1.336983629919931e-06, "loss": 0.9384, "step": 47850 }, { "epoch": 0.8705141963295077, "grad_norm": 1.8219040632247925, "learning_rate": 1.3351084775638022e-06, "loss": 0.8573, "step": 47860 }, { "epoch": 0.87069608395933, "grad_norm": 2.0371274948120117, "learning_rate": 1.333233325207673e-06, "loss": 0.8971, "step": 47870 }, { "epoch": 0.8708779715891523, "grad_norm": 1.547310709953308, "learning_rate": 1.3313581728515442e-06, "loss": 0.8917, "step": 47880 }, { "epoch": 0.8710598592189746, "grad_norm": 1.4884638786315918, "learning_rate": 1.3294830204954153e-06, "loss": 0.8754, "step": 47890 }, { "epoch": 0.8712417468487968, "grad_norm": 1.7032403945922852, "learning_rate": 1.3276078681392865e-06, "loss": 0.8839, "step": 47900 }, { "epoch": 0.8714236344786191, "grad_norm": 1.8696184158325195, "learning_rate": 1.3257327157831576e-06, "loss": 0.899, "step": 47910 }, { "epoch": 0.8716055221084414, "grad_norm": 2.6163299083709717, "learning_rate": 1.3238575634270283e-06, "loss": 0.9422, "step": 47920 }, { "epoch": 0.8717874097382637, "grad_norm": 2.2697017192840576, "learning_rate": 1.3219824110708995e-06, "loss": 0.9406, "step": 47930 }, { "epoch": 0.871969297368086, "grad_norm": 1.9394186735153198, "learning_rate": 1.3201072587147706e-06, "loss": 0.8796, "step": 47940 }, { "epoch": 0.8721511849979083, "grad_norm": 1.4450395107269287, "learning_rate": 1.3182321063586417e-06, "loss": 0.8289, "step": 47950 }, { "epoch": 0.8723330726277306, "grad_norm": 1.8871665000915527, "learning_rate": 1.3163569540025129e-06, "loss": 0.9036, "step": 47960 }, { "epoch": 0.8725149602575529, "grad_norm": 1.703977346420288, "learning_rate": 1.314481801646384e-06, "loss": 0.88, "step": 47970 }, { "epoch": 0.8726968478873752, "grad_norm": 1.5225331783294678, "learning_rate": 1.3126066492902551e-06, "loss": 0.9248, "step": 47980 }, { "epoch": 0.8728787355171975, "grad_norm": 1.9322733879089355, "learning_rate": 1.3107314969341258e-06, "loss": 0.8872, "step": 47990 }, { "epoch": 0.8730606231470198, "grad_norm": 1.673452377319336, "learning_rate": 1.308856344577997e-06, "loss": 0.8849, "step": 48000 }, { "epoch": 0.8732425107768421, "grad_norm": 1.6404955387115479, "learning_rate": 1.3069811922218681e-06, "loss": 0.9167, "step": 48010 }, { "epoch": 0.8734243984066644, "grad_norm": 1.7973906993865967, "learning_rate": 1.3051060398657392e-06, "loss": 0.9219, "step": 48020 }, { "epoch": 0.8736062860364867, "grad_norm": 1.5009113550186157, "learning_rate": 1.3032308875096104e-06, "loss": 0.8567, "step": 48030 }, { "epoch": 0.873788173666309, "grad_norm": 1.6717288494110107, "learning_rate": 1.3013557351534813e-06, "loss": 0.9141, "step": 48040 }, { "epoch": 0.8739700612961312, "grad_norm": 1.780364751815796, "learning_rate": 1.2994805827973524e-06, "loss": 0.9262, "step": 48050 }, { "epoch": 0.8741519489259535, "grad_norm": 1.7084497213363647, "learning_rate": 1.2976054304412234e-06, "loss": 0.8518, "step": 48060 }, { "epoch": 0.8743338365557758, "grad_norm": 1.8010286092758179, "learning_rate": 1.2957302780850945e-06, "loss": 0.8513, "step": 48070 }, { "epoch": 0.8745157241855981, "grad_norm": 1.9118068218231201, "learning_rate": 1.2938551257289656e-06, "loss": 0.8526, "step": 48080 }, { "epoch": 0.8746976118154204, "grad_norm": 1.9658195972442627, "learning_rate": 1.2919799733728365e-06, "loss": 0.89, "step": 48090 }, { "epoch": 0.8748794994452427, "grad_norm": 1.8146311044692993, "learning_rate": 1.2901048210167077e-06, "loss": 0.8963, "step": 48100 }, { "epoch": 0.875061387075065, "grad_norm": 2.1317546367645264, "learning_rate": 1.2882296686605788e-06, "loss": 0.8779, "step": 48110 }, { "epoch": 0.8752432747048873, "grad_norm": 1.6187872886657715, "learning_rate": 1.28635451630445e-06, "loss": 0.8524, "step": 48120 }, { "epoch": 0.8754251623347096, "grad_norm": 1.9870805740356445, "learning_rate": 1.2844793639483209e-06, "loss": 0.8657, "step": 48130 }, { "epoch": 0.8756070499645319, "grad_norm": 1.9682689905166626, "learning_rate": 1.2826042115921918e-06, "loss": 0.8558, "step": 48140 }, { "epoch": 0.8757889375943542, "grad_norm": 1.8981071710586548, "learning_rate": 1.280729059236063e-06, "loss": 0.9535, "step": 48150 }, { "epoch": 0.8759708252241765, "grad_norm": 2.1204748153686523, "learning_rate": 1.278853906879934e-06, "loss": 0.89, "step": 48160 }, { "epoch": 0.8761527128539988, "grad_norm": 1.7126439809799194, "learning_rate": 1.2769787545238052e-06, "loss": 0.8912, "step": 48170 }, { "epoch": 0.8763346004838211, "grad_norm": 1.9081432819366455, "learning_rate": 1.2751036021676763e-06, "loss": 0.9423, "step": 48180 }, { "epoch": 0.8765164881136434, "grad_norm": 1.8506437540054321, "learning_rate": 1.2732284498115475e-06, "loss": 0.8482, "step": 48190 }, { "epoch": 0.8766983757434657, "grad_norm": 1.9329702854156494, "learning_rate": 1.2713532974554182e-06, "loss": 0.8945, "step": 48200 }, { "epoch": 0.876880263373288, "grad_norm": 1.8249307870864868, "learning_rate": 1.2694781450992893e-06, "loss": 0.8953, "step": 48210 }, { "epoch": 0.8770621510031102, "grad_norm": 1.9135252237319946, "learning_rate": 1.2676029927431604e-06, "loss": 0.8809, "step": 48220 }, { "epoch": 0.8772440386329325, "grad_norm": 1.9186420440673828, "learning_rate": 1.2657278403870316e-06, "loss": 0.9, "step": 48230 }, { "epoch": 0.8774259262627548, "grad_norm": 1.6891684532165527, "learning_rate": 1.2638526880309027e-06, "loss": 0.8933, "step": 48240 }, { "epoch": 0.8776078138925771, "grad_norm": 1.415194034576416, "learning_rate": 1.2619775356747738e-06, "loss": 0.9017, "step": 48250 }, { "epoch": 0.8777897015223994, "grad_norm": 1.9045689105987549, "learning_rate": 1.2601023833186448e-06, "loss": 0.8607, "step": 48260 }, { "epoch": 0.8779715891522217, "grad_norm": 1.7670533657073975, "learning_rate": 1.2582272309625157e-06, "loss": 0.8807, "step": 48270 }, { "epoch": 0.878153476782044, "grad_norm": 1.6032848358154297, "learning_rate": 1.2563520786063868e-06, "loss": 0.8933, "step": 48280 }, { "epoch": 0.8783353644118663, "grad_norm": 1.8763773441314697, "learning_rate": 1.254476926250258e-06, "loss": 0.8732, "step": 48290 }, { "epoch": 0.8785172520416886, "grad_norm": 1.6578282117843628, "learning_rate": 1.252601773894129e-06, "loss": 0.8173, "step": 48300 }, { "epoch": 0.8786991396715109, "grad_norm": 1.7172664403915405, "learning_rate": 1.250726621538e-06, "loss": 0.8977, "step": 48310 }, { "epoch": 0.8788810273013332, "grad_norm": 1.866369366645813, "learning_rate": 1.2488514691818712e-06, "loss": 0.9297, "step": 48320 }, { "epoch": 0.8790629149311555, "grad_norm": 1.7037276029586792, "learning_rate": 1.2469763168257423e-06, "loss": 0.8415, "step": 48330 }, { "epoch": 0.8792448025609778, "grad_norm": 1.6417561769485474, "learning_rate": 1.2451011644696132e-06, "loss": 0.8403, "step": 48340 }, { "epoch": 0.8794266901908001, "grad_norm": 1.6400291919708252, "learning_rate": 1.2432260121134843e-06, "loss": 0.9082, "step": 48350 }, { "epoch": 0.8796085778206224, "grad_norm": 1.7235053777694702, "learning_rate": 1.2413508597573553e-06, "loss": 0.8755, "step": 48360 }, { "epoch": 0.8797904654504447, "grad_norm": 1.7136924266815186, "learning_rate": 1.2394757074012264e-06, "loss": 0.8986, "step": 48370 }, { "epoch": 0.879972353080267, "grad_norm": 1.9081246852874756, "learning_rate": 1.2376005550450975e-06, "loss": 0.9413, "step": 48380 }, { "epoch": 0.8801542407100893, "grad_norm": 1.8677886724472046, "learning_rate": 1.2357254026889685e-06, "loss": 0.9323, "step": 48390 }, { "epoch": 0.8803361283399116, "grad_norm": 1.8068042993545532, "learning_rate": 1.2338502503328396e-06, "loss": 0.8623, "step": 48400 }, { "epoch": 0.8805180159697339, "grad_norm": 2.158266067504883, "learning_rate": 1.2319750979767107e-06, "loss": 0.9046, "step": 48410 }, { "epoch": 0.8806999035995562, "grad_norm": 1.7304216623306274, "learning_rate": 1.2300999456205816e-06, "loss": 0.9106, "step": 48420 }, { "epoch": 0.8808817912293785, "grad_norm": 1.7798995971679688, "learning_rate": 1.2282247932644528e-06, "loss": 0.8731, "step": 48430 }, { "epoch": 0.8810636788592008, "grad_norm": 1.561847448348999, "learning_rate": 1.226349640908324e-06, "loss": 0.9105, "step": 48440 }, { "epoch": 0.8812455664890231, "grad_norm": 2.0570967197418213, "learning_rate": 1.224474488552195e-06, "loss": 0.9239, "step": 48450 }, { "epoch": 0.8814274541188454, "grad_norm": 1.6215455532073975, "learning_rate": 1.222599336196066e-06, "loss": 0.8925, "step": 48460 }, { "epoch": 0.8816093417486677, "grad_norm": 1.956658124923706, "learning_rate": 1.2207241838399371e-06, "loss": 0.8849, "step": 48470 }, { "epoch": 0.88179122937849, "grad_norm": 1.6236462593078613, "learning_rate": 1.2188490314838082e-06, "loss": 0.8825, "step": 48480 }, { "epoch": 0.8819731170083123, "grad_norm": 1.854007601737976, "learning_rate": 1.2169738791276792e-06, "loss": 0.8862, "step": 48490 }, { "epoch": 0.8821550046381346, "grad_norm": 2.001497268676758, "learning_rate": 1.2150987267715503e-06, "loss": 0.9095, "step": 48500 }, { "epoch": 0.8823368922679569, "grad_norm": 3.110832452774048, "learning_rate": 1.2132235744154214e-06, "loss": 0.8385, "step": 48510 }, { "epoch": 0.8825187798977792, "grad_norm": 1.468031644821167, "learning_rate": 1.2113484220592926e-06, "loss": 0.8709, "step": 48520 }, { "epoch": 0.8827006675276015, "grad_norm": 1.9637237787246704, "learning_rate": 1.2094732697031635e-06, "loss": 0.9309, "step": 48530 }, { "epoch": 0.8828825551574238, "grad_norm": 1.5584791898727417, "learning_rate": 1.2075981173470346e-06, "loss": 0.8136, "step": 48540 }, { "epoch": 0.883064442787246, "grad_norm": 2.0505027770996094, "learning_rate": 1.2057229649909058e-06, "loss": 0.9003, "step": 48550 }, { "epoch": 0.8832463304170683, "grad_norm": 1.8504356145858765, "learning_rate": 1.2038478126347767e-06, "loss": 0.9436, "step": 48560 }, { "epoch": 0.8834282180468906, "grad_norm": 2.0573360919952393, "learning_rate": 1.2019726602786478e-06, "loss": 0.9185, "step": 48570 }, { "epoch": 0.8836101056767129, "grad_norm": 1.989993929862976, "learning_rate": 1.2000975079225187e-06, "loss": 0.8564, "step": 48580 }, { "epoch": 0.8837919933065352, "grad_norm": 1.785801887512207, "learning_rate": 1.1982223555663899e-06, "loss": 0.9219, "step": 48590 }, { "epoch": 0.8839738809363575, "grad_norm": 1.8146960735321045, "learning_rate": 1.196347203210261e-06, "loss": 0.892, "step": 48600 }, { "epoch": 0.8841557685661798, "grad_norm": 1.7214933633804321, "learning_rate": 1.194472050854132e-06, "loss": 0.8644, "step": 48610 }, { "epoch": 0.8843376561960021, "grad_norm": 1.794289231300354, "learning_rate": 1.192596898498003e-06, "loss": 0.8567, "step": 48620 }, { "epoch": 0.8845195438258244, "grad_norm": 1.830572247505188, "learning_rate": 1.190721746141874e-06, "loss": 0.8741, "step": 48630 }, { "epoch": 0.8847014314556467, "grad_norm": 1.7420719861984253, "learning_rate": 1.1888465937857451e-06, "loss": 0.9411, "step": 48640 }, { "epoch": 0.884883319085469, "grad_norm": 1.5929256677627563, "learning_rate": 1.1869714414296163e-06, "loss": 0.9122, "step": 48650 }, { "epoch": 0.8850652067152913, "grad_norm": 1.70958411693573, "learning_rate": 1.1850962890734872e-06, "loss": 0.8952, "step": 48660 }, { "epoch": 0.8852470943451136, "grad_norm": 1.6430943012237549, "learning_rate": 1.1832211367173583e-06, "loss": 0.9348, "step": 48670 }, { "epoch": 0.8854289819749359, "grad_norm": 3.798797845840454, "learning_rate": 1.1813459843612294e-06, "loss": 0.9381, "step": 48680 }, { "epoch": 0.8856108696047582, "grad_norm": 1.5168273448944092, "learning_rate": 1.1794708320051006e-06, "loss": 0.8871, "step": 48690 }, { "epoch": 0.8857927572345805, "grad_norm": 1.6840554475784302, "learning_rate": 1.1775956796489715e-06, "loss": 0.8922, "step": 48700 }, { "epoch": 0.8859746448644028, "grad_norm": 1.5215253829956055, "learning_rate": 1.1757205272928426e-06, "loss": 0.8752, "step": 48710 }, { "epoch": 0.8861565324942251, "grad_norm": 1.925917387008667, "learning_rate": 1.1738453749367138e-06, "loss": 0.9293, "step": 48720 }, { "epoch": 0.8863384201240474, "grad_norm": 1.9157216548919678, "learning_rate": 1.1719702225805847e-06, "loss": 0.9038, "step": 48730 }, { "epoch": 0.8865203077538697, "grad_norm": 1.8140559196472168, "learning_rate": 1.1700950702244558e-06, "loss": 0.8909, "step": 48740 }, { "epoch": 0.886702195383692, "grad_norm": 1.651581883430481, "learning_rate": 1.168219917868327e-06, "loss": 0.8567, "step": 48750 }, { "epoch": 0.8868840830135143, "grad_norm": 1.714636206626892, "learning_rate": 1.166344765512198e-06, "loss": 0.8799, "step": 48760 }, { "epoch": 0.8870659706433366, "grad_norm": 1.508941650390625, "learning_rate": 1.164469613156069e-06, "loss": 0.8653, "step": 48770 }, { "epoch": 0.8872478582731589, "grad_norm": 1.9882034063339233, "learning_rate": 1.1625944607999402e-06, "loss": 0.9347, "step": 48780 }, { "epoch": 0.8874297459029812, "grad_norm": 1.7992470264434814, "learning_rate": 1.1607193084438113e-06, "loss": 0.8694, "step": 48790 }, { "epoch": 0.8876116335328035, "grad_norm": 1.3543908596038818, "learning_rate": 1.1588441560876822e-06, "loss": 0.8967, "step": 48800 }, { "epoch": 0.8877935211626258, "grad_norm": 1.8564542531967163, "learning_rate": 1.1569690037315533e-06, "loss": 0.9359, "step": 48810 }, { "epoch": 0.8879754087924481, "grad_norm": 1.7832627296447754, "learning_rate": 1.1550938513754245e-06, "loss": 0.8808, "step": 48820 }, { "epoch": 0.8881572964222704, "grad_norm": 1.6531888246536255, "learning_rate": 1.1532186990192954e-06, "loss": 0.9042, "step": 48830 }, { "epoch": 0.8883391840520927, "grad_norm": 1.7851561307907104, "learning_rate": 1.1513435466631665e-06, "loss": 0.8953, "step": 48840 }, { "epoch": 0.888521071681915, "grad_norm": 1.7705506086349487, "learning_rate": 1.1494683943070375e-06, "loss": 0.9069, "step": 48850 }, { "epoch": 0.8887029593117373, "grad_norm": 1.6958900690078735, "learning_rate": 1.1475932419509086e-06, "loss": 0.9178, "step": 48860 }, { "epoch": 0.8888848469415596, "grad_norm": 1.8572609424591064, "learning_rate": 1.1457180895947795e-06, "loss": 0.8895, "step": 48870 }, { "epoch": 0.8890667345713817, "grad_norm": 1.7569680213928223, "learning_rate": 1.1438429372386506e-06, "loss": 0.8986, "step": 48880 }, { "epoch": 0.889248622201204, "grad_norm": 1.6796817779541016, "learning_rate": 1.1419677848825218e-06, "loss": 0.8702, "step": 48890 }, { "epoch": 0.8894305098310263, "grad_norm": 2.656599521636963, "learning_rate": 1.1400926325263927e-06, "loss": 0.868, "step": 48900 }, { "epoch": 0.8896123974608486, "grad_norm": 1.7070457935333252, "learning_rate": 1.1382174801702638e-06, "loss": 0.8697, "step": 48910 }, { "epoch": 0.8897942850906709, "grad_norm": 1.8265495300292969, "learning_rate": 1.136342327814135e-06, "loss": 0.8685, "step": 48920 }, { "epoch": 0.8899761727204932, "grad_norm": 2.045860767364502, "learning_rate": 1.1344671754580061e-06, "loss": 0.8916, "step": 48930 }, { "epoch": 0.8901580603503155, "grad_norm": 1.7349786758422852, "learning_rate": 1.132592023101877e-06, "loss": 0.8823, "step": 48940 }, { "epoch": 0.8903399479801378, "grad_norm": 1.8291077613830566, "learning_rate": 1.1307168707457482e-06, "loss": 0.8535, "step": 48950 }, { "epoch": 0.8905218356099601, "grad_norm": 1.8757717609405518, "learning_rate": 1.1288417183896193e-06, "loss": 0.8691, "step": 48960 }, { "epoch": 0.8907037232397824, "grad_norm": 1.745502233505249, "learning_rate": 1.1269665660334902e-06, "loss": 0.8674, "step": 48970 }, { "epoch": 0.8908856108696047, "grad_norm": 1.8926265239715576, "learning_rate": 1.1250914136773614e-06, "loss": 0.9018, "step": 48980 }, { "epoch": 0.891067498499427, "grad_norm": 1.7576011419296265, "learning_rate": 1.1232162613212325e-06, "loss": 0.8706, "step": 48990 }, { "epoch": 0.8912493861292493, "grad_norm": 2.15202260017395, "learning_rate": 1.1213411089651036e-06, "loss": 0.9133, "step": 49000 }, { "epoch": 0.8914312737590716, "grad_norm": 1.5561023950576782, "learning_rate": 1.1194659566089745e-06, "loss": 0.8457, "step": 49010 }, { "epoch": 0.8916131613888939, "grad_norm": 1.6043211221694946, "learning_rate": 1.1175908042528457e-06, "loss": 0.8885, "step": 49020 }, { "epoch": 0.8917950490187162, "grad_norm": 1.9906657934188843, "learning_rate": 1.1157156518967168e-06, "loss": 0.9319, "step": 49030 }, { "epoch": 0.8919769366485385, "grad_norm": 1.910563349723816, "learning_rate": 1.1138404995405877e-06, "loss": 0.9075, "step": 49040 }, { "epoch": 0.8921588242783608, "grad_norm": 1.8960026502609253, "learning_rate": 1.1119653471844589e-06, "loss": 0.8595, "step": 49050 }, { "epoch": 0.8923407119081831, "grad_norm": 1.6238704919815063, "learning_rate": 1.11009019482833e-06, "loss": 0.856, "step": 49060 }, { "epoch": 0.8925225995380054, "grad_norm": 1.6732537746429443, "learning_rate": 1.108215042472201e-06, "loss": 0.8595, "step": 49070 }, { "epoch": 0.8927044871678277, "grad_norm": 1.842214822769165, "learning_rate": 1.106339890116072e-06, "loss": 0.9086, "step": 49080 }, { "epoch": 0.89288637479765, "grad_norm": 1.6111053228378296, "learning_rate": 1.104464737759943e-06, "loss": 0.9083, "step": 49090 }, { "epoch": 0.8930682624274723, "grad_norm": 1.7506638765335083, "learning_rate": 1.1025895854038141e-06, "loss": 0.8776, "step": 49100 }, { "epoch": 0.8932501500572946, "grad_norm": 1.73333740234375, "learning_rate": 1.1007144330476853e-06, "loss": 0.8639, "step": 49110 }, { "epoch": 0.8934320376871169, "grad_norm": 1.8790631294250488, "learning_rate": 1.0988392806915562e-06, "loss": 0.9102, "step": 49120 }, { "epoch": 0.8936139253169392, "grad_norm": 1.8280110359191895, "learning_rate": 1.0969641283354273e-06, "loss": 0.9151, "step": 49130 }, { "epoch": 0.8937958129467615, "grad_norm": 1.7765936851501465, "learning_rate": 1.0950889759792984e-06, "loss": 0.9231, "step": 49140 }, { "epoch": 0.8939777005765838, "grad_norm": 1.877920150756836, "learning_rate": 1.0932138236231694e-06, "loss": 0.8943, "step": 49150 }, { "epoch": 0.8941595882064061, "grad_norm": 1.7343271970748901, "learning_rate": 1.0913386712670405e-06, "loss": 0.8861, "step": 49160 }, { "epoch": 0.8943414758362284, "grad_norm": 1.6693294048309326, "learning_rate": 1.0894635189109116e-06, "loss": 0.8481, "step": 49170 }, { "epoch": 0.8945233634660507, "grad_norm": 2.1241793632507324, "learning_rate": 1.0875883665547826e-06, "loss": 0.8501, "step": 49180 }, { "epoch": 0.894705251095873, "grad_norm": 1.7148388624191284, "learning_rate": 1.0857132141986537e-06, "loss": 0.8398, "step": 49190 }, { "epoch": 0.8948871387256953, "grad_norm": 1.7002122402191162, "learning_rate": 1.0838380618425248e-06, "loss": 0.8555, "step": 49200 }, { "epoch": 0.8950690263555175, "grad_norm": 1.770699143409729, "learning_rate": 1.0819629094863958e-06, "loss": 0.9073, "step": 49210 }, { "epoch": 0.8952509139853398, "grad_norm": 1.9964693784713745, "learning_rate": 1.0800877571302669e-06, "loss": 0.9165, "step": 49220 }, { "epoch": 0.8954328016151621, "grad_norm": 1.6941115856170654, "learning_rate": 1.078212604774138e-06, "loss": 0.9112, "step": 49230 }, { "epoch": 0.8956146892449844, "grad_norm": 1.672477126121521, "learning_rate": 1.0763374524180092e-06, "loss": 0.8498, "step": 49240 }, { "epoch": 0.8957965768748067, "grad_norm": 1.8207308053970337, "learning_rate": 1.07446230006188e-06, "loss": 0.9371, "step": 49250 }, { "epoch": 0.895978464504629, "grad_norm": 1.8124430179595947, "learning_rate": 1.0725871477057512e-06, "loss": 0.8955, "step": 49260 }, { "epoch": 0.8961603521344513, "grad_norm": 1.9175975322723389, "learning_rate": 1.0707119953496223e-06, "loss": 0.8755, "step": 49270 }, { "epoch": 0.8963422397642736, "grad_norm": 1.9916229248046875, "learning_rate": 1.0688368429934933e-06, "loss": 0.9186, "step": 49280 }, { "epoch": 0.8965241273940959, "grad_norm": 1.7918182611465454, "learning_rate": 1.0669616906373644e-06, "loss": 0.9135, "step": 49290 }, { "epoch": 0.8967060150239182, "grad_norm": 1.9802017211914062, "learning_rate": 1.0650865382812355e-06, "loss": 0.8723, "step": 49300 }, { "epoch": 0.8968879026537405, "grad_norm": 1.893986463546753, "learning_rate": 1.0632113859251065e-06, "loss": 0.8686, "step": 49310 }, { "epoch": 0.8970697902835628, "grad_norm": 1.8826664686203003, "learning_rate": 1.0613362335689776e-06, "loss": 0.9186, "step": 49320 }, { "epoch": 0.8972516779133851, "grad_norm": 1.7353806495666504, "learning_rate": 1.0594610812128487e-06, "loss": 0.9456, "step": 49330 }, { "epoch": 0.8974335655432074, "grad_norm": 1.9892303943634033, "learning_rate": 1.0575859288567197e-06, "loss": 0.8729, "step": 49340 }, { "epoch": 0.8976154531730297, "grad_norm": 2.161444664001465, "learning_rate": 1.0557107765005908e-06, "loss": 0.9048, "step": 49350 }, { "epoch": 0.897797340802852, "grad_norm": 1.8103123903274536, "learning_rate": 1.0538356241444617e-06, "loss": 0.9368, "step": 49360 }, { "epoch": 0.8979792284326743, "grad_norm": 1.489651083946228, "learning_rate": 1.0519604717883328e-06, "loss": 0.8585, "step": 49370 }, { "epoch": 0.8981611160624966, "grad_norm": 1.7312076091766357, "learning_rate": 1.050085319432204e-06, "loss": 0.8849, "step": 49380 }, { "epoch": 0.8983430036923189, "grad_norm": 1.6740593910217285, "learning_rate": 1.048210167076075e-06, "loss": 0.8743, "step": 49390 }, { "epoch": 0.8985248913221412, "grad_norm": 2.0570967197418213, "learning_rate": 1.046335014719946e-06, "loss": 0.8934, "step": 49400 }, { "epoch": 0.8987067789519635, "grad_norm": 1.9919805526733398, "learning_rate": 1.0444598623638172e-06, "loss": 0.8867, "step": 49410 }, { "epoch": 0.8988886665817858, "grad_norm": 1.6576260328292847, "learning_rate": 1.042584710007688e-06, "loss": 0.8577, "step": 49420 }, { "epoch": 0.8990705542116081, "grad_norm": 1.7258964776992798, "learning_rate": 1.0407095576515592e-06, "loss": 0.8903, "step": 49430 }, { "epoch": 0.8992524418414304, "grad_norm": 1.6305196285247803, "learning_rate": 1.0388344052954304e-06, "loss": 0.9143, "step": 49440 }, { "epoch": 0.8994343294712527, "grad_norm": 1.7588136196136475, "learning_rate": 1.0369592529393015e-06, "loss": 0.8738, "step": 49450 }, { "epoch": 0.899616217101075, "grad_norm": 1.7689805030822754, "learning_rate": 1.0350841005831724e-06, "loss": 0.8458, "step": 49460 }, { "epoch": 0.8997981047308973, "grad_norm": 1.7410783767700195, "learning_rate": 1.0332089482270435e-06, "loss": 0.9325, "step": 49470 }, { "epoch": 0.8999799923607196, "grad_norm": 2.0345141887664795, "learning_rate": 1.0313337958709147e-06, "loss": 0.909, "step": 49480 }, { "epoch": 0.9001618799905419, "grad_norm": 1.5492429733276367, "learning_rate": 1.0294586435147856e-06, "loss": 0.9198, "step": 49490 }, { "epoch": 0.9003437676203642, "grad_norm": 1.90488862991333, "learning_rate": 1.0275834911586567e-06, "loss": 0.8923, "step": 49500 }, { "epoch": 0.9005256552501865, "grad_norm": 1.7980070114135742, "learning_rate": 1.0257083388025279e-06, "loss": 0.8574, "step": 49510 }, { "epoch": 0.9007075428800088, "grad_norm": 1.9675266742706299, "learning_rate": 1.0238331864463988e-06, "loss": 0.8545, "step": 49520 }, { "epoch": 0.9008894305098311, "grad_norm": 1.6712146997451782, "learning_rate": 1.02195803409027e-06, "loss": 0.9128, "step": 49530 }, { "epoch": 0.9010713181396534, "grad_norm": 1.9118119478225708, "learning_rate": 1.020082881734141e-06, "loss": 0.9505, "step": 49540 }, { "epoch": 0.9012532057694757, "grad_norm": 2.0669400691986084, "learning_rate": 1.0182077293780122e-06, "loss": 0.9119, "step": 49550 }, { "epoch": 0.901435093399298, "grad_norm": 1.7780697345733643, "learning_rate": 1.0163325770218831e-06, "loss": 0.8806, "step": 49560 }, { "epoch": 0.9016169810291202, "grad_norm": 1.6896228790283203, "learning_rate": 1.0144574246657543e-06, "loss": 0.8888, "step": 49570 }, { "epoch": 0.9017988686589425, "grad_norm": 1.8604635000228882, "learning_rate": 1.0125822723096252e-06, "loss": 0.9255, "step": 49580 }, { "epoch": 0.9019807562887648, "grad_norm": 1.709917664527893, "learning_rate": 1.0107071199534963e-06, "loss": 0.8818, "step": 49590 }, { "epoch": 0.9021626439185871, "grad_norm": 1.9300017356872559, "learning_rate": 1.0088319675973674e-06, "loss": 0.8983, "step": 49600 }, { "epoch": 0.9023445315484094, "grad_norm": 1.8096506595611572, "learning_rate": 1.0069568152412384e-06, "loss": 0.9106, "step": 49610 }, { "epoch": 0.9025264191782317, "grad_norm": 1.763456106185913, "learning_rate": 1.0050816628851095e-06, "loss": 0.9664, "step": 49620 }, { "epoch": 0.902708306808054, "grad_norm": 1.793454647064209, "learning_rate": 1.0032065105289804e-06, "loss": 0.8854, "step": 49630 }, { "epoch": 0.9028901944378763, "grad_norm": 1.8329869508743286, "learning_rate": 1.0013313581728516e-06, "loss": 0.9254, "step": 49640 }, { "epoch": 0.9030720820676986, "grad_norm": 2.0848400592803955, "learning_rate": 9.994562058167227e-07, "loss": 0.889, "step": 49650 }, { "epoch": 0.9032539696975209, "grad_norm": 1.630940556526184, "learning_rate": 9.975810534605936e-07, "loss": 0.9163, "step": 49660 }, { "epoch": 0.9034358573273432, "grad_norm": 2.2208728790283203, "learning_rate": 9.957059011044648e-07, "loss": 0.9444, "step": 49670 }, { "epoch": 0.9036177449571655, "grad_norm": 1.765824794769287, "learning_rate": 9.938307487483359e-07, "loss": 0.8808, "step": 49680 }, { "epoch": 0.9037996325869878, "grad_norm": 1.7552082538604736, "learning_rate": 9.91955596392207e-07, "loss": 0.889, "step": 49690 }, { "epoch": 0.9039815202168101, "grad_norm": 1.7811224460601807, "learning_rate": 9.90080444036078e-07, "loss": 0.8806, "step": 49700 }, { "epoch": 0.9041634078466324, "grad_norm": 1.5960954427719116, "learning_rate": 9.88205291679949e-07, "loss": 0.8988, "step": 49710 }, { "epoch": 0.9043452954764546, "grad_norm": 2.250556230545044, "learning_rate": 9.863301393238202e-07, "loss": 0.9023, "step": 49720 }, { "epoch": 0.9045271831062769, "grad_norm": 1.4743168354034424, "learning_rate": 9.844549869676911e-07, "loss": 0.8497, "step": 49730 }, { "epoch": 0.9047090707360992, "grad_norm": 1.6820226907730103, "learning_rate": 9.825798346115623e-07, "loss": 0.8371, "step": 49740 }, { "epoch": 0.9048909583659215, "grad_norm": 1.776291847229004, "learning_rate": 9.807046822554334e-07, "loss": 0.9075, "step": 49750 }, { "epoch": 0.9050728459957438, "grad_norm": 1.7363306283950806, "learning_rate": 9.788295298993045e-07, "loss": 0.8681, "step": 49760 }, { "epoch": 0.9052547336255661, "grad_norm": 1.6264342069625854, "learning_rate": 9.769543775431755e-07, "loss": 0.9116, "step": 49770 }, { "epoch": 0.9054366212553884, "grad_norm": 2.0522043704986572, "learning_rate": 9.750792251870466e-07, "loss": 0.9085, "step": 49780 }, { "epoch": 0.9056185088852107, "grad_norm": 1.779392957687378, "learning_rate": 9.732040728309177e-07, "loss": 0.9034, "step": 49790 }, { "epoch": 0.905800396515033, "grad_norm": 1.8619095087051392, "learning_rate": 9.713289204747887e-07, "loss": 0.9274, "step": 49800 }, { "epoch": 0.9059822841448553, "grad_norm": 1.705400824546814, "learning_rate": 9.694537681186598e-07, "loss": 0.8525, "step": 49810 }, { "epoch": 0.9061641717746776, "grad_norm": 1.809912919998169, "learning_rate": 9.67578615762531e-07, "loss": 0.9024, "step": 49820 }, { "epoch": 0.9063460594044999, "grad_norm": 1.6143596172332764, "learning_rate": 9.657034634064018e-07, "loss": 0.9096, "step": 49830 }, { "epoch": 0.9065279470343222, "grad_norm": 1.7694703340530396, "learning_rate": 9.63828311050273e-07, "loss": 0.8998, "step": 49840 }, { "epoch": 0.9067098346641445, "grad_norm": 1.5529597997665405, "learning_rate": 9.61953158694144e-07, "loss": 0.8947, "step": 49850 }, { "epoch": 0.9068917222939668, "grad_norm": 1.7930577993392944, "learning_rate": 9.60078006338015e-07, "loss": 0.8892, "step": 49860 }, { "epoch": 0.907073609923789, "grad_norm": 1.504612922668457, "learning_rate": 9.58202853981886e-07, "loss": 0.9015, "step": 49870 }, { "epoch": 0.9072554975536113, "grad_norm": 1.8212376832962036, "learning_rate": 9.56327701625757e-07, "loss": 0.9032, "step": 49880 }, { "epoch": 0.9074373851834336, "grad_norm": 1.609329104423523, "learning_rate": 9.544525492696282e-07, "loss": 0.8865, "step": 49890 }, { "epoch": 0.9076192728132559, "grad_norm": 1.8133670091629028, "learning_rate": 9.525773969134993e-07, "loss": 0.9196, "step": 49900 }, { "epoch": 0.9078011604430782, "grad_norm": 1.7704354524612427, "learning_rate": 9.507022445573704e-07, "loss": 0.9286, "step": 49910 }, { "epoch": 0.9079830480729005, "grad_norm": 1.7733662128448486, "learning_rate": 9.488270922012414e-07, "loss": 0.9021, "step": 49920 }, { "epoch": 0.9081649357027228, "grad_norm": 1.8267284631729126, "learning_rate": 9.469519398451126e-07, "loss": 0.8855, "step": 49930 }, { "epoch": 0.9083468233325451, "grad_norm": 1.8780417442321777, "learning_rate": 9.450767874889835e-07, "loss": 0.9243, "step": 49940 }, { "epoch": 0.9085287109623674, "grad_norm": 1.857926845550537, "learning_rate": 9.432016351328546e-07, "loss": 0.9446, "step": 49950 }, { "epoch": 0.9087105985921897, "grad_norm": 1.6661603450775146, "learning_rate": 9.413264827767257e-07, "loss": 0.9021, "step": 49960 }, { "epoch": 0.908892486222012, "grad_norm": 1.8867639303207397, "learning_rate": 9.394513304205967e-07, "loss": 0.8475, "step": 49970 }, { "epoch": 0.9090743738518343, "grad_norm": 1.7363148927688599, "learning_rate": 9.375761780644678e-07, "loss": 0.9581, "step": 49980 }, { "epoch": 0.9092562614816566, "grad_norm": 1.9234106540679932, "learning_rate": 9.357010257083389e-07, "loss": 0.8719, "step": 49990 }, { "epoch": 0.9094381491114789, "grad_norm": 1.7329853773117065, "learning_rate": 9.3382587335221e-07, "loss": 0.9397, "step": 50000 }, { "epoch": 0.9096200367413012, "grad_norm": 1.7642637491226196, "learning_rate": 9.31950720996081e-07, "loss": 0.9087, "step": 50010 }, { "epoch": 0.9098019243711235, "grad_norm": 1.9270793199539185, "learning_rate": 9.300755686399521e-07, "loss": 0.8839, "step": 50020 }, { "epoch": 0.9099838120009458, "grad_norm": 1.8870320320129395, "learning_rate": 9.282004162838232e-07, "loss": 0.8818, "step": 50030 }, { "epoch": 0.9101656996307681, "grad_norm": 1.685808777809143, "learning_rate": 9.263252639276942e-07, "loss": 0.9188, "step": 50040 }, { "epoch": 0.9103475872605904, "grad_norm": 1.5664621591567993, "learning_rate": 9.244501115715652e-07, "loss": 0.8736, "step": 50050 }, { "epoch": 0.9105294748904127, "grad_norm": 1.816100835800171, "learning_rate": 9.225749592154363e-07, "loss": 0.9342, "step": 50060 }, { "epoch": 0.910711362520235, "grad_norm": 1.8616538047790527, "learning_rate": 9.206998068593075e-07, "loss": 0.8993, "step": 50070 }, { "epoch": 0.9108932501500573, "grad_norm": 2.1562466621398926, "learning_rate": 9.188246545031784e-07, "loss": 0.8842, "step": 50080 }, { "epoch": 0.9110751377798796, "grad_norm": 1.6774230003356934, "learning_rate": 9.169495021470495e-07, "loss": 0.8542, "step": 50090 }, { "epoch": 0.9112570254097019, "grad_norm": 1.4940860271453857, "learning_rate": 9.150743497909207e-07, "loss": 0.8743, "step": 50100 }, { "epoch": 0.9114389130395242, "grad_norm": 1.6367684602737427, "learning_rate": 9.131991974347916e-07, "loss": 0.8851, "step": 50110 }, { "epoch": 0.9116208006693465, "grad_norm": 2.0696732997894287, "learning_rate": 9.113240450786627e-07, "loss": 0.8581, "step": 50120 }, { "epoch": 0.9118026882991688, "grad_norm": 1.9739935398101807, "learning_rate": 9.094488927225339e-07, "loss": 0.899, "step": 50130 }, { "epoch": 0.9119845759289911, "grad_norm": 2.1304879188537598, "learning_rate": 9.075737403664048e-07, "loss": 0.8479, "step": 50140 }, { "epoch": 0.9121664635588134, "grad_norm": 1.8487704992294312, "learning_rate": 9.056985880102759e-07, "loss": 0.9473, "step": 50150 }, { "epoch": 0.9123483511886357, "grad_norm": 1.8136298656463623, "learning_rate": 9.038234356541469e-07, "loss": 0.874, "step": 50160 }, { "epoch": 0.912530238818458, "grad_norm": 2.120307207107544, "learning_rate": 9.019482832980181e-07, "loss": 0.9075, "step": 50170 }, { "epoch": 0.9127121264482803, "grad_norm": 1.875233769416809, "learning_rate": 9.000731309418891e-07, "loss": 0.9059, "step": 50180 }, { "epoch": 0.9128940140781026, "grad_norm": 1.6870744228363037, "learning_rate": 8.981979785857601e-07, "loss": 0.887, "step": 50190 }, { "epoch": 0.9130759017079249, "grad_norm": 1.993452548980713, "learning_rate": 8.963228262296313e-07, "loss": 0.9134, "step": 50200 }, { "epoch": 0.9132577893377472, "grad_norm": 2.1090731620788574, "learning_rate": 8.944476738735022e-07, "loss": 0.8782, "step": 50210 }, { "epoch": 0.9134396769675694, "grad_norm": 1.8874701261520386, "learning_rate": 8.925725215173733e-07, "loss": 0.9049, "step": 50220 }, { "epoch": 0.9136215645973917, "grad_norm": 1.88060462474823, "learning_rate": 8.906973691612445e-07, "loss": 0.8659, "step": 50230 }, { "epoch": 0.913803452227214, "grad_norm": 1.9726207256317139, "learning_rate": 8.888222168051156e-07, "loss": 0.9216, "step": 50240 }, { "epoch": 0.9139853398570363, "grad_norm": 1.7966594696044922, "learning_rate": 8.869470644489865e-07, "loss": 0.9385, "step": 50250 }, { "epoch": 0.9141672274868586, "grad_norm": 1.5711458921432495, "learning_rate": 8.850719120928577e-07, "loss": 0.866, "step": 50260 }, { "epoch": 0.9143491151166809, "grad_norm": 1.7015725374221802, "learning_rate": 8.831967597367287e-07, "loss": 0.9234, "step": 50270 }, { "epoch": 0.9145310027465032, "grad_norm": 1.9586646556854248, "learning_rate": 8.813216073805997e-07, "loss": 0.9412, "step": 50280 }, { "epoch": 0.9147128903763255, "grad_norm": 1.5071704387664795, "learning_rate": 8.794464550244708e-07, "loss": 0.8501, "step": 50290 }, { "epoch": 0.9148947780061478, "grad_norm": 1.738677978515625, "learning_rate": 8.775713026683419e-07, "loss": 0.9159, "step": 50300 }, { "epoch": 0.9150766656359701, "grad_norm": 1.7049206495285034, "learning_rate": 8.75696150312213e-07, "loss": 0.8636, "step": 50310 }, { "epoch": 0.9152585532657924, "grad_norm": 1.637962818145752, "learning_rate": 8.738209979560839e-07, "loss": 0.8407, "step": 50320 }, { "epoch": 0.9154404408956147, "grad_norm": 1.60097336769104, "learning_rate": 8.719458455999551e-07, "loss": 0.8645, "step": 50330 }, { "epoch": 0.915622328525437, "grad_norm": 1.66801917552948, "learning_rate": 8.700706932438262e-07, "loss": 0.8842, "step": 50340 }, { "epoch": 0.9158042161552593, "grad_norm": 1.6513950824737549, "learning_rate": 8.681955408876971e-07, "loss": 0.8403, "step": 50350 }, { "epoch": 0.9159861037850816, "grad_norm": 1.849062442779541, "learning_rate": 8.663203885315683e-07, "loss": 0.9399, "step": 50360 }, { "epoch": 0.9161679914149039, "grad_norm": 1.5929611921310425, "learning_rate": 8.644452361754394e-07, "loss": 0.8866, "step": 50370 }, { "epoch": 0.9163498790447262, "grad_norm": 1.6532337665557861, "learning_rate": 8.625700838193104e-07, "loss": 0.8442, "step": 50380 }, { "epoch": 0.9165317666745485, "grad_norm": 1.6718623638153076, "learning_rate": 8.606949314631814e-07, "loss": 0.9242, "step": 50390 }, { "epoch": 0.9167136543043708, "grad_norm": 1.9363311529159546, "learning_rate": 8.588197791070525e-07, "loss": 0.9607, "step": 50400 }, { "epoch": 0.9168955419341931, "grad_norm": 1.920437216758728, "learning_rate": 8.569446267509236e-07, "loss": 0.929, "step": 50410 }, { "epoch": 0.9170774295640154, "grad_norm": 1.9925472736358643, "learning_rate": 8.550694743947946e-07, "loss": 0.9477, "step": 50420 }, { "epoch": 0.9172593171938377, "grad_norm": 1.7847900390625, "learning_rate": 8.531943220386657e-07, "loss": 0.8873, "step": 50430 }, { "epoch": 0.91744120482366, "grad_norm": 1.7530759572982788, "learning_rate": 8.513191696825368e-07, "loss": 0.9122, "step": 50440 }, { "epoch": 0.9176230924534823, "grad_norm": 1.9129037857055664, "learning_rate": 8.494440173264077e-07, "loss": 0.8793, "step": 50450 }, { "epoch": 0.9178049800833046, "grad_norm": 1.858827829360962, "learning_rate": 8.475688649702789e-07, "loss": 0.8824, "step": 50460 }, { "epoch": 0.9179868677131269, "grad_norm": 1.827688455581665, "learning_rate": 8.4569371261415e-07, "loss": 0.9272, "step": 50470 }, { "epoch": 0.9181687553429492, "grad_norm": 1.8716037273406982, "learning_rate": 8.438185602580211e-07, "loss": 0.9004, "step": 50480 }, { "epoch": 0.9183506429727715, "grad_norm": 1.8300449848175049, "learning_rate": 8.41943407901892e-07, "loss": 0.8736, "step": 50490 }, { "epoch": 0.9185325306025938, "grad_norm": 1.5421918630599976, "learning_rate": 8.400682555457632e-07, "loss": 0.8601, "step": 50500 }, { "epoch": 0.9187144182324161, "grad_norm": 1.580862283706665, "learning_rate": 8.381931031896342e-07, "loss": 0.8772, "step": 50510 }, { "epoch": 0.9188963058622384, "grad_norm": 1.5880405902862549, "learning_rate": 8.363179508335052e-07, "loss": 0.8753, "step": 50520 }, { "epoch": 0.9190781934920607, "grad_norm": 1.6648014783859253, "learning_rate": 8.344427984773764e-07, "loss": 0.857, "step": 50530 }, { "epoch": 0.919260081121883, "grad_norm": 1.9576537609100342, "learning_rate": 8.325676461212474e-07, "loss": 0.947, "step": 50540 }, { "epoch": 0.9194419687517053, "grad_norm": 1.689488172531128, "learning_rate": 8.306924937651185e-07, "loss": 0.8721, "step": 50550 }, { "epoch": 0.9196238563815274, "grad_norm": 1.9104033708572388, "learning_rate": 8.288173414089895e-07, "loss": 0.9719, "step": 50560 }, { "epoch": 0.9198057440113497, "grad_norm": 1.9841681718826294, "learning_rate": 8.269421890528606e-07, "loss": 0.8554, "step": 50570 }, { "epoch": 0.919987631641172, "grad_norm": 1.8018689155578613, "learning_rate": 8.250670366967317e-07, "loss": 0.9214, "step": 50580 }, { "epoch": 0.9201695192709943, "grad_norm": 2.05644154548645, "learning_rate": 8.231918843406026e-07, "loss": 0.8954, "step": 50590 }, { "epoch": 0.9203514069008166, "grad_norm": 1.4941238164901733, "learning_rate": 8.213167319844738e-07, "loss": 0.8971, "step": 50600 }, { "epoch": 0.9205332945306389, "grad_norm": 1.4249035120010376, "learning_rate": 8.194415796283449e-07, "loss": 0.8638, "step": 50610 }, { "epoch": 0.9207151821604612, "grad_norm": 1.846476674079895, "learning_rate": 8.175664272722159e-07, "loss": 0.9007, "step": 50620 }, { "epoch": 0.9208970697902835, "grad_norm": 2.1954925060272217, "learning_rate": 8.15691274916087e-07, "loss": 0.9316, "step": 50630 }, { "epoch": 0.9210789574201058, "grad_norm": 1.7516717910766602, "learning_rate": 8.138161225599581e-07, "loss": 0.8428, "step": 50640 }, { "epoch": 0.9212608450499281, "grad_norm": 1.858127474784851, "learning_rate": 8.119409702038291e-07, "loss": 0.9377, "step": 50650 }, { "epoch": 0.9214427326797504, "grad_norm": 1.5954208374023438, "learning_rate": 8.100658178477002e-07, "loss": 0.8922, "step": 50660 }, { "epoch": 0.9216246203095727, "grad_norm": 1.986305832862854, "learning_rate": 8.081906654915712e-07, "loss": 0.8917, "step": 50670 }, { "epoch": 0.921806507939395, "grad_norm": 1.6177059412002563, "learning_rate": 8.063155131354423e-07, "loss": 0.8852, "step": 50680 }, { "epoch": 0.9219883955692173, "grad_norm": 1.8063507080078125, "learning_rate": 8.044403607793135e-07, "loss": 0.874, "step": 50690 }, { "epoch": 0.9221702831990396, "grad_norm": 1.8663599491119385, "learning_rate": 8.025652084231844e-07, "loss": 0.9308, "step": 50700 }, { "epoch": 0.9223521708288619, "grad_norm": 1.8332772254943848, "learning_rate": 8.006900560670555e-07, "loss": 0.8623, "step": 50710 }, { "epoch": 0.9225340584586842, "grad_norm": 1.6091424226760864, "learning_rate": 7.988149037109267e-07, "loss": 0.9183, "step": 50720 }, { "epoch": 0.9227159460885065, "grad_norm": 1.8558006286621094, "learning_rate": 7.969397513547976e-07, "loss": 0.8839, "step": 50730 }, { "epoch": 0.9228978337183288, "grad_norm": 1.7615864276885986, "learning_rate": 7.950645989986687e-07, "loss": 0.8166, "step": 50740 }, { "epoch": 0.9230797213481511, "grad_norm": 1.8519591093063354, "learning_rate": 7.931894466425398e-07, "loss": 0.9092, "step": 50750 }, { "epoch": 0.9232616089779734, "grad_norm": 1.8583025932312012, "learning_rate": 7.913142942864108e-07, "loss": 0.851, "step": 50760 }, { "epoch": 0.9234434966077957, "grad_norm": 6.129265308380127, "learning_rate": 7.894391419302819e-07, "loss": 0.8885, "step": 50770 }, { "epoch": 0.923625384237618, "grad_norm": 1.853402018547058, "learning_rate": 7.875639895741529e-07, "loss": 0.9056, "step": 50780 }, { "epoch": 0.9238072718674403, "grad_norm": 2.0528194904327393, "learning_rate": 7.856888372180241e-07, "loss": 0.9504, "step": 50790 }, { "epoch": 0.9239891594972626, "grad_norm": 1.6408730745315552, "learning_rate": 7.838136848618951e-07, "loss": 0.8708, "step": 50800 }, { "epoch": 0.9241710471270849, "grad_norm": 1.883144497871399, "learning_rate": 7.819385325057661e-07, "loss": 0.914, "step": 50810 }, { "epoch": 0.9243529347569072, "grad_norm": 1.385356068611145, "learning_rate": 7.800633801496373e-07, "loss": 0.9134, "step": 50820 }, { "epoch": 0.9245348223867295, "grad_norm": 1.627859354019165, "learning_rate": 7.781882277935082e-07, "loss": 0.9133, "step": 50830 }, { "epoch": 0.9247167100165518, "grad_norm": 1.6206778287887573, "learning_rate": 7.763130754373793e-07, "loss": 0.8883, "step": 50840 }, { "epoch": 0.924898597646374, "grad_norm": 1.9484621286392212, "learning_rate": 7.744379230812504e-07, "loss": 0.9106, "step": 50850 }, { "epoch": 0.9250804852761964, "grad_norm": 1.6105276346206665, "learning_rate": 7.725627707251216e-07, "loss": 0.9475, "step": 50860 }, { "epoch": 0.9252623729060186, "grad_norm": 1.7613332271575928, "learning_rate": 7.706876183689925e-07, "loss": 0.8728, "step": 50870 }, { "epoch": 0.925444260535841, "grad_norm": 1.9078750610351562, "learning_rate": 7.688124660128636e-07, "loss": 0.8894, "step": 50880 }, { "epoch": 0.9256261481656632, "grad_norm": 1.753584623336792, "learning_rate": 7.669373136567347e-07, "loss": 0.8438, "step": 50890 }, { "epoch": 0.9258080357954855, "grad_norm": 1.9063968658447266, "learning_rate": 7.650621613006057e-07, "loss": 0.8992, "step": 50900 }, { "epoch": 0.9259899234253078, "grad_norm": 1.8514769077301025, "learning_rate": 7.631870089444768e-07, "loss": 0.9207, "step": 50910 }, { "epoch": 0.9261718110551301, "grad_norm": 2.065622329711914, "learning_rate": 7.613118565883479e-07, "loss": 0.9266, "step": 50920 }, { "epoch": 0.9263536986849524, "grad_norm": 1.7618910074234009, "learning_rate": 7.59436704232219e-07, "loss": 0.8956, "step": 50930 }, { "epoch": 0.9265355863147747, "grad_norm": 2.1542959213256836, "learning_rate": 7.575615518760899e-07, "loss": 0.8703, "step": 50940 }, { "epoch": 0.926717473944597, "grad_norm": 1.7254914045333862, "learning_rate": 7.55686399519961e-07, "loss": 0.8824, "step": 50950 }, { "epoch": 0.9268993615744193, "grad_norm": 1.725892424583435, "learning_rate": 7.538112471638322e-07, "loss": 0.9084, "step": 50960 }, { "epoch": 0.9270812492042416, "grad_norm": 2.0286083221435547, "learning_rate": 7.519360948077031e-07, "loss": 0.9088, "step": 50970 }, { "epoch": 0.9272631368340639, "grad_norm": 1.9914634227752686, "learning_rate": 7.500609424515742e-07, "loss": 0.8947, "step": 50980 }, { "epoch": 0.9274450244638862, "grad_norm": 1.6489503383636475, "learning_rate": 7.481857900954454e-07, "loss": 0.9037, "step": 50990 }, { "epoch": 0.9276269120937085, "grad_norm": 1.9179999828338623, "learning_rate": 7.463106377393164e-07, "loss": 0.9135, "step": 51000 }, { "epoch": 0.9278087997235308, "grad_norm": 2.0015499591827393, "learning_rate": 7.444354853831874e-07, "loss": 0.8846, "step": 51010 }, { "epoch": 0.9279906873533531, "grad_norm": 1.8603886365890503, "learning_rate": 7.425603330270586e-07, "loss": 0.9021, "step": 51020 }, { "epoch": 0.9281725749831754, "grad_norm": 1.7738934755325317, "learning_rate": 7.406851806709296e-07, "loss": 0.897, "step": 51030 }, { "epoch": 0.9283544626129977, "grad_norm": 1.6922544240951538, "learning_rate": 7.388100283148006e-07, "loss": 0.9193, "step": 51040 }, { "epoch": 0.92853635024282, "grad_norm": 1.7083415985107422, "learning_rate": 7.369348759586716e-07, "loss": 0.8975, "step": 51050 }, { "epoch": 0.9287182378726423, "grad_norm": 1.7263835668563843, "learning_rate": 7.350597236025428e-07, "loss": 0.8902, "step": 51060 }, { "epoch": 0.9289001255024646, "grad_norm": 1.7597914934158325, "learning_rate": 7.331845712464138e-07, "loss": 0.9027, "step": 51070 }, { "epoch": 0.9290820131322869, "grad_norm": 1.7099032402038574, "learning_rate": 7.313094188902848e-07, "loss": 0.8684, "step": 51080 }, { "epoch": 0.9292639007621092, "grad_norm": 1.8699753284454346, "learning_rate": 7.29434266534156e-07, "loss": 0.8756, "step": 51090 }, { "epoch": 0.9294457883919315, "grad_norm": 1.8212120532989502, "learning_rate": 7.275591141780271e-07, "loss": 0.9304, "step": 51100 }, { "epoch": 0.9296276760217538, "grad_norm": 1.9474639892578125, "learning_rate": 7.25683961821898e-07, "loss": 0.9224, "step": 51110 }, { "epoch": 0.9298095636515761, "grad_norm": 1.7813037633895874, "learning_rate": 7.238088094657692e-07, "loss": 0.8816, "step": 51120 }, { "epoch": 0.9299914512813984, "grad_norm": 1.8521595001220703, "learning_rate": 7.219336571096403e-07, "loss": 0.9188, "step": 51130 }, { "epoch": 0.9301733389112207, "grad_norm": 1.8585125207901, "learning_rate": 7.200585047535112e-07, "loss": 0.8474, "step": 51140 }, { "epoch": 0.930355226541043, "grad_norm": 1.1800601482391357, "learning_rate": 7.181833523973824e-07, "loss": 0.8448, "step": 51150 }, { "epoch": 0.9305371141708653, "grad_norm": 1.4210795164108276, "learning_rate": 7.163082000412534e-07, "loss": 0.8283, "step": 51160 }, { "epoch": 0.9307190018006876, "grad_norm": 1.986207127571106, "learning_rate": 7.144330476851245e-07, "loss": 0.9083, "step": 51170 }, { "epoch": 0.9309008894305099, "grad_norm": 1.721281886100769, "learning_rate": 7.125578953289955e-07, "loss": 0.8909, "step": 51180 }, { "epoch": 0.9310827770603322, "grad_norm": 2.147038221359253, "learning_rate": 7.106827429728666e-07, "loss": 0.9096, "step": 51190 }, { "epoch": 0.9312646646901545, "grad_norm": 1.78541100025177, "learning_rate": 7.088075906167377e-07, "loss": 0.8551, "step": 51200 }, { "epoch": 0.9314465523199768, "grad_norm": 1.8562521934509277, "learning_rate": 7.069324382606086e-07, "loss": 0.8956, "step": 51210 }, { "epoch": 0.931628439949799, "grad_norm": 1.6017941236495972, "learning_rate": 7.050572859044798e-07, "loss": 0.8832, "step": 51220 }, { "epoch": 0.9318103275796213, "grad_norm": 1.854802131652832, "learning_rate": 7.031821335483509e-07, "loss": 0.9084, "step": 51230 }, { "epoch": 0.9319922152094436, "grad_norm": 1.8054172992706299, "learning_rate": 7.01306981192222e-07, "loss": 0.8563, "step": 51240 }, { "epoch": 0.9321741028392659, "grad_norm": 1.7013791799545288, "learning_rate": 6.99431828836093e-07, "loss": 0.8644, "step": 51250 }, { "epoch": 0.9323559904690882, "grad_norm": 1.9377349615097046, "learning_rate": 6.975566764799641e-07, "loss": 0.9188, "step": 51260 }, { "epoch": 0.9325378780989105, "grad_norm": 1.7423549890518188, "learning_rate": 6.956815241238351e-07, "loss": 0.9276, "step": 51270 }, { "epoch": 0.9327197657287328, "grad_norm": 1.9830927848815918, "learning_rate": 6.938063717677061e-07, "loss": 0.9097, "step": 51280 }, { "epoch": 0.9329016533585551, "grad_norm": 1.8194503784179688, "learning_rate": 6.919312194115773e-07, "loss": 0.8756, "step": 51290 }, { "epoch": 0.9330835409883774, "grad_norm": 1.9150867462158203, "learning_rate": 6.900560670554483e-07, "loss": 0.9497, "step": 51300 }, { "epoch": 0.9332654286181997, "grad_norm": 2.039517402648926, "learning_rate": 6.881809146993194e-07, "loss": 0.912, "step": 51310 }, { "epoch": 0.933447316248022, "grad_norm": 2.052469253540039, "learning_rate": 6.863057623431904e-07, "loss": 0.837, "step": 51320 }, { "epoch": 0.9336292038778443, "grad_norm": 1.5961371660232544, "learning_rate": 6.844306099870615e-07, "loss": 0.9321, "step": 51330 }, { "epoch": 0.9338110915076666, "grad_norm": 1.615105390548706, "learning_rate": 6.825554576309326e-07, "loss": 0.9078, "step": 51340 }, { "epoch": 0.9339929791374889, "grad_norm": 1.8198567628860474, "learning_rate": 6.806803052748036e-07, "loss": 0.9049, "step": 51350 }, { "epoch": 0.9341748667673112, "grad_norm": 1.869363784790039, "learning_rate": 6.788051529186747e-07, "loss": 0.8703, "step": 51360 }, { "epoch": 0.9343567543971335, "grad_norm": 1.502649188041687, "learning_rate": 6.769300005625458e-07, "loss": 0.8812, "step": 51370 }, { "epoch": 0.9345386420269558, "grad_norm": 1.8592524528503418, "learning_rate": 6.750548482064168e-07, "loss": 0.918, "step": 51380 }, { "epoch": 0.934720529656778, "grad_norm": 1.6851568222045898, "learning_rate": 6.731796958502879e-07, "loss": 0.8814, "step": 51390 }, { "epoch": 0.9349024172866003, "grad_norm": 1.6589332818984985, "learning_rate": 6.71304543494159e-07, "loss": 0.8613, "step": 51400 }, { "epoch": 0.9350843049164226, "grad_norm": 1.7079718112945557, "learning_rate": 6.6942939113803e-07, "loss": 0.8871, "step": 51410 }, { "epoch": 0.9352661925462449, "grad_norm": 1.7061790227890015, "learning_rate": 6.675542387819011e-07, "loss": 0.9166, "step": 51420 }, { "epoch": 0.9354480801760672, "grad_norm": 1.6939040422439575, "learning_rate": 6.656790864257721e-07, "loss": 0.8794, "step": 51430 }, { "epoch": 0.9356299678058895, "grad_norm": 1.6002079248428345, "learning_rate": 6.638039340696432e-07, "loss": 0.8306, "step": 51440 }, { "epoch": 0.9358118554357118, "grad_norm": 2.048654317855835, "learning_rate": 6.619287817135142e-07, "loss": 0.8809, "step": 51450 }, { "epoch": 0.9359937430655341, "grad_norm": 1.9224117994308472, "learning_rate": 6.600536293573853e-07, "loss": 0.894, "step": 51460 }, { "epoch": 0.9361756306953564, "grad_norm": 1.8000904321670532, "learning_rate": 6.581784770012564e-07, "loss": 0.913, "step": 51470 }, { "epoch": 0.9363575183251787, "grad_norm": 1.4006829261779785, "learning_rate": 6.563033246451276e-07, "loss": 0.8444, "step": 51480 }, { "epoch": 0.936539405955001, "grad_norm": 1.7436860799789429, "learning_rate": 6.544281722889985e-07, "loss": 0.8878, "step": 51490 }, { "epoch": 0.9367212935848233, "grad_norm": 1.7650325298309326, "learning_rate": 6.525530199328696e-07, "loss": 0.9006, "step": 51500 }, { "epoch": 0.9369031812146456, "grad_norm": 1.4958856105804443, "learning_rate": 6.506778675767406e-07, "loss": 0.9085, "step": 51510 }, { "epoch": 0.9370850688444679, "grad_norm": 1.737479329109192, "learning_rate": 6.488027152206117e-07, "loss": 0.8911, "step": 51520 }, { "epoch": 0.9372669564742901, "grad_norm": 1.9269897937774658, "learning_rate": 6.469275628644828e-07, "loss": 0.8797, "step": 51530 }, { "epoch": 0.9374488441041124, "grad_norm": 1.6477235555648804, "learning_rate": 6.450524105083538e-07, "loss": 0.8711, "step": 51540 }, { "epoch": 0.9376307317339347, "grad_norm": 1.706518292427063, "learning_rate": 6.43177258152225e-07, "loss": 0.8476, "step": 51550 }, { "epoch": 0.937812619363757, "grad_norm": 1.8781756162643433, "learning_rate": 6.413021057960959e-07, "loss": 0.8853, "step": 51560 }, { "epoch": 0.9379945069935793, "grad_norm": 1.8773550987243652, "learning_rate": 6.39426953439967e-07, "loss": 0.9048, "step": 51570 }, { "epoch": 0.9381763946234016, "grad_norm": 1.8638713359832764, "learning_rate": 6.375518010838382e-07, "loss": 0.928, "step": 51580 }, { "epoch": 0.9383582822532239, "grad_norm": 2.1017277240753174, "learning_rate": 6.356766487277091e-07, "loss": 0.8577, "step": 51590 }, { "epoch": 0.9385401698830462, "grad_norm": 2.077603340148926, "learning_rate": 6.338014963715802e-07, "loss": 0.9466, "step": 51600 }, { "epoch": 0.9387220575128685, "grad_norm": 1.7290314435958862, "learning_rate": 6.319263440154514e-07, "loss": 0.9106, "step": 51610 }, { "epoch": 0.9389039451426908, "grad_norm": 1.6290613412857056, "learning_rate": 6.300511916593224e-07, "loss": 0.8923, "step": 51620 }, { "epoch": 0.9390858327725131, "grad_norm": 1.7675915956497192, "learning_rate": 6.281760393031934e-07, "loss": 0.9622, "step": 51630 }, { "epoch": 0.9392677204023354, "grad_norm": 2.3633041381835938, "learning_rate": 6.263008869470645e-07, "loss": 0.9292, "step": 51640 }, { "epoch": 0.9394496080321577, "grad_norm": 1.7897794246673584, "learning_rate": 6.244257345909356e-07, "loss": 0.8772, "step": 51650 }, { "epoch": 0.93963149566198, "grad_norm": 1.6002141237258911, "learning_rate": 6.225505822348066e-07, "loss": 0.9321, "step": 51660 }, { "epoch": 0.9398133832918023, "grad_norm": 1.682660698890686, "learning_rate": 6.206754298786776e-07, "loss": 0.919, "step": 51670 }, { "epoch": 0.9399952709216246, "grad_norm": 1.4977120161056519, "learning_rate": 6.188002775225488e-07, "loss": 0.8582, "step": 51680 }, { "epoch": 0.9401771585514469, "grad_norm": 1.925803303718567, "learning_rate": 6.169251251664198e-07, "loss": 0.8956, "step": 51690 }, { "epoch": 0.9403590461812692, "grad_norm": 1.5783852338790894, "learning_rate": 6.150499728102908e-07, "loss": 0.9226, "step": 51700 }, { "epoch": 0.9405409338110915, "grad_norm": 1.6097571849822998, "learning_rate": 6.13174820454162e-07, "loss": 0.9281, "step": 51710 }, { "epoch": 0.9407228214409138, "grad_norm": 2.0341601371765137, "learning_rate": 6.11299668098033e-07, "loss": 0.8796, "step": 51720 }, { "epoch": 0.9409047090707361, "grad_norm": 1.655587911605835, "learning_rate": 6.094245157419041e-07, "loss": 0.9001, "step": 51730 }, { "epoch": 0.9410865967005584, "grad_norm": 2.0425949096679688, "learning_rate": 6.075493633857752e-07, "loss": 0.9327, "step": 51740 }, { "epoch": 0.9412684843303807, "grad_norm": 1.9119153022766113, "learning_rate": 6.056742110296463e-07, "loss": 0.8743, "step": 51750 }, { "epoch": 0.941450371960203, "grad_norm": 1.4835243225097656, "learning_rate": 6.037990586735173e-07, "loss": 0.9121, "step": 51760 }, { "epoch": 0.9416322595900253, "grad_norm": 1.8352574110031128, "learning_rate": 6.019239063173883e-07, "loss": 0.8895, "step": 51770 }, { "epoch": 0.9418141472198476, "grad_norm": 1.5766887664794922, "learning_rate": 6.000487539612594e-07, "loss": 0.953, "step": 51780 }, { "epoch": 0.9419960348496699, "grad_norm": 1.6938852071762085, "learning_rate": 5.981736016051305e-07, "loss": 0.917, "step": 51790 }, { "epoch": 0.9421779224794922, "grad_norm": 1.962127685546875, "learning_rate": 5.962984492490015e-07, "loss": 0.918, "step": 51800 }, { "epoch": 0.9423598101093145, "grad_norm": 1.7468737363815308, "learning_rate": 5.944232968928726e-07, "loss": 0.8497, "step": 51810 }, { "epoch": 0.9425416977391368, "grad_norm": 2.0887844562530518, "learning_rate": 5.925481445367436e-07, "loss": 0.9155, "step": 51820 }, { "epoch": 0.9427235853689591, "grad_norm": 1.7625057697296143, "learning_rate": 5.906729921806147e-07, "loss": 0.9184, "step": 51830 }, { "epoch": 0.9429054729987814, "grad_norm": 1.8219964504241943, "learning_rate": 5.887978398244858e-07, "loss": 0.8759, "step": 51840 }, { "epoch": 0.9430873606286037, "grad_norm": 2.0816304683685303, "learning_rate": 5.869226874683569e-07, "loss": 0.8244, "step": 51850 }, { "epoch": 0.943269248258426, "grad_norm": 1.9049627780914307, "learning_rate": 5.850475351122279e-07, "loss": 0.8676, "step": 51860 }, { "epoch": 0.9434511358882482, "grad_norm": 2.1388561725616455, "learning_rate": 5.83172382756099e-07, "loss": 0.8803, "step": 51870 }, { "epoch": 0.9436330235180705, "grad_norm": 1.7594056129455566, "learning_rate": 5.812972303999701e-07, "loss": 0.9521, "step": 51880 }, { "epoch": 0.9438149111478928, "grad_norm": 1.559926986694336, "learning_rate": 5.794220780438411e-07, "loss": 0.9453, "step": 51890 }, { "epoch": 0.9439967987777151, "grad_norm": 1.896848440170288, "learning_rate": 5.775469256877122e-07, "loss": 0.9089, "step": 51900 }, { "epoch": 0.9441786864075374, "grad_norm": 1.8226450681686401, "learning_rate": 5.756717733315833e-07, "loss": 0.9129, "step": 51910 }, { "epoch": 0.9443605740373597, "grad_norm": 1.8784912824630737, "learning_rate": 5.737966209754543e-07, "loss": 0.9184, "step": 51920 }, { "epoch": 0.944542461667182, "grad_norm": 1.7365614175796509, "learning_rate": 5.719214686193253e-07, "loss": 0.8657, "step": 51930 }, { "epoch": 0.9447243492970043, "grad_norm": 1.5169492959976196, "learning_rate": 5.700463162631964e-07, "loss": 0.8733, "step": 51940 }, { "epoch": 0.9449062369268266, "grad_norm": 2.062864065170288, "learning_rate": 5.681711639070675e-07, "loss": 0.9066, "step": 51950 }, { "epoch": 0.9450881245566489, "grad_norm": 1.990669846534729, "learning_rate": 5.662960115509385e-07, "loss": 0.9337, "step": 51960 }, { "epoch": 0.9452700121864712, "grad_norm": 1.63023042678833, "learning_rate": 5.644208591948097e-07, "loss": 0.9269, "step": 51970 }, { "epoch": 0.9454518998162935, "grad_norm": 1.378880500793457, "learning_rate": 5.625457068386807e-07, "loss": 0.8676, "step": 51980 }, { "epoch": 0.9456337874461158, "grad_norm": 1.7651969194412231, "learning_rate": 5.606705544825518e-07, "loss": 0.8973, "step": 51990 }, { "epoch": 0.9458156750759381, "grad_norm": 1.568773865699768, "learning_rate": 5.587954021264228e-07, "loss": 0.8702, "step": 52000 }, { "epoch": 0.9459975627057604, "grad_norm": 1.9192440509796143, "learning_rate": 5.569202497702939e-07, "loss": 0.8941, "step": 52010 }, { "epoch": 0.9461794503355827, "grad_norm": 1.6015069484710693, "learning_rate": 5.55045097414165e-07, "loss": 0.9183, "step": 52020 }, { "epoch": 0.946361337965405, "grad_norm": 1.7115941047668457, "learning_rate": 5.53169945058036e-07, "loss": 0.9108, "step": 52030 }, { "epoch": 0.9465432255952273, "grad_norm": 1.5182929039001465, "learning_rate": 5.512947927019071e-07, "loss": 0.8854, "step": 52040 }, { "epoch": 0.9467251132250496, "grad_norm": 1.957244873046875, "learning_rate": 5.494196403457781e-07, "loss": 0.9601, "step": 52050 }, { "epoch": 0.9469070008548719, "grad_norm": 1.6242049932479858, "learning_rate": 5.475444879896492e-07, "loss": 0.9225, "step": 52060 }, { "epoch": 0.9470888884846942, "grad_norm": 1.789332389831543, "learning_rate": 5.456693356335203e-07, "loss": 0.8817, "step": 52070 }, { "epoch": 0.9472707761145165, "grad_norm": 1.53301203250885, "learning_rate": 5.437941832773913e-07, "loss": 0.8909, "step": 52080 }, { "epoch": 0.9474526637443388, "grad_norm": 2.0192480087280273, "learning_rate": 5.419190309212624e-07, "loss": 0.8829, "step": 52090 }, { "epoch": 0.9476345513741611, "grad_norm": 1.6977570056915283, "learning_rate": 5.400438785651334e-07, "loss": 0.9133, "step": 52100 }, { "epoch": 0.9478164390039834, "grad_norm": 1.6228829622268677, "learning_rate": 5.381687262090046e-07, "loss": 0.8444, "step": 52110 }, { "epoch": 0.9479983266338057, "grad_norm": 1.8420363664627075, "learning_rate": 5.362935738528756e-07, "loss": 0.8797, "step": 52120 }, { "epoch": 0.948180214263628, "grad_norm": 1.5505741834640503, "learning_rate": 5.344184214967466e-07, "loss": 0.8562, "step": 52130 }, { "epoch": 0.9483621018934503, "grad_norm": 1.649553894996643, "learning_rate": 5.325432691406178e-07, "loss": 0.875, "step": 52140 }, { "epoch": 0.9485439895232726, "grad_norm": 1.7693121433258057, "learning_rate": 5.306681167844888e-07, "loss": 0.8376, "step": 52150 }, { "epoch": 0.9487258771530949, "grad_norm": 2.0773918628692627, "learning_rate": 5.287929644283598e-07, "loss": 0.9334, "step": 52160 }, { "epoch": 0.9489077647829172, "grad_norm": 1.91910719871521, "learning_rate": 5.269178120722309e-07, "loss": 0.8782, "step": 52170 }, { "epoch": 0.9490896524127395, "grad_norm": 1.7494957447052002, "learning_rate": 5.25042659716102e-07, "loss": 0.9001, "step": 52180 }, { "epoch": 0.9492715400425618, "grad_norm": 1.7433851957321167, "learning_rate": 5.23167507359973e-07, "loss": 0.9118, "step": 52190 }, { "epoch": 0.949453427672384, "grad_norm": 1.4176981449127197, "learning_rate": 5.21292355003844e-07, "loss": 0.8582, "step": 52200 }, { "epoch": 0.9496353153022064, "grad_norm": 1.6157079935073853, "learning_rate": 5.194172026477152e-07, "loss": 0.8324, "step": 52210 }, { "epoch": 0.9498172029320286, "grad_norm": 1.9110054969787598, "learning_rate": 5.175420502915862e-07, "loss": 0.8268, "step": 52220 }, { "epoch": 0.9499990905618508, "grad_norm": 1.6925430297851562, "learning_rate": 5.156668979354573e-07, "loss": 0.897, "step": 52230 }, { "epoch": 0.9501809781916731, "grad_norm": 1.8037737607955933, "learning_rate": 5.137917455793284e-07, "loss": 0.8895, "step": 52240 }, { "epoch": 0.9503628658214954, "grad_norm": 1.7020078897476196, "learning_rate": 5.119165932231994e-07, "loss": 0.8878, "step": 52250 }, { "epoch": 0.9505447534513177, "grad_norm": 1.950561285018921, "learning_rate": 5.100414408670705e-07, "loss": 0.9302, "step": 52260 }, { "epoch": 0.95072664108114, "grad_norm": 2.098365068435669, "learning_rate": 5.081662885109416e-07, "loss": 0.8681, "step": 52270 }, { "epoch": 0.9509085287109623, "grad_norm": 1.7038792371749878, "learning_rate": 5.062911361548126e-07, "loss": 0.923, "step": 52280 }, { "epoch": 0.9510904163407846, "grad_norm": 1.7639466524124146, "learning_rate": 5.044159837986837e-07, "loss": 0.8978, "step": 52290 }, { "epoch": 0.9512723039706069, "grad_norm": 1.9037680625915527, "learning_rate": 5.025408314425548e-07, "loss": 0.9166, "step": 52300 }, { "epoch": 0.9514541916004292, "grad_norm": 1.610097050666809, "learning_rate": 5.006656790864258e-07, "loss": 0.8883, "step": 52310 }, { "epoch": 0.9516360792302515, "grad_norm": 1.8120169639587402, "learning_rate": 4.987905267302968e-07, "loss": 0.8968, "step": 52320 }, { "epoch": 0.9518179668600738, "grad_norm": 1.7286356687545776, "learning_rate": 4.969153743741679e-07, "loss": 0.9109, "step": 52330 }, { "epoch": 0.9519998544898961, "grad_norm": 1.432952880859375, "learning_rate": 4.95040222018039e-07, "loss": 0.8809, "step": 52340 }, { "epoch": 0.9521817421197184, "grad_norm": 1.9174970388412476, "learning_rate": 4.931650696619101e-07, "loss": 0.914, "step": 52350 }, { "epoch": 0.9523636297495407, "grad_norm": 1.3783937692642212, "learning_rate": 4.912899173057811e-07, "loss": 0.8672, "step": 52360 }, { "epoch": 0.952545517379363, "grad_norm": 1.7206238508224487, "learning_rate": 4.894147649496523e-07, "loss": 0.9231, "step": 52370 }, { "epoch": 0.9527274050091853, "grad_norm": 1.9233392477035522, "learning_rate": 4.875396125935233e-07, "loss": 0.8124, "step": 52380 }, { "epoch": 0.9529092926390076, "grad_norm": 1.7230353355407715, "learning_rate": 4.856644602373943e-07, "loss": 0.8904, "step": 52390 }, { "epoch": 0.9530911802688299, "grad_norm": 1.9402155876159668, "learning_rate": 4.837893078812655e-07, "loss": 0.8974, "step": 52400 }, { "epoch": 0.9532730678986522, "grad_norm": 1.7505662441253662, "learning_rate": 4.819141555251365e-07, "loss": 0.8259, "step": 52410 }, { "epoch": 0.9534549555284745, "grad_norm": 2.093898296356201, "learning_rate": 4.800390031690075e-07, "loss": 0.9501, "step": 52420 }, { "epoch": 0.9536368431582968, "grad_norm": 1.8039040565490723, "learning_rate": 4.781638508128785e-07, "loss": 0.948, "step": 52430 }, { "epoch": 0.9538187307881191, "grad_norm": 2.015381097793579, "learning_rate": 4.7628869845674963e-07, "loss": 0.8941, "step": 52440 }, { "epoch": 0.9540006184179414, "grad_norm": 2.380089044570923, "learning_rate": 4.744135461006207e-07, "loss": 0.8902, "step": 52450 }, { "epoch": 0.9541825060477637, "grad_norm": 1.7351382970809937, "learning_rate": 4.7253839374449174e-07, "loss": 0.8738, "step": 52460 }, { "epoch": 0.954364393677586, "grad_norm": 1.847638487815857, "learning_rate": 4.7066324138836287e-07, "loss": 0.8923, "step": 52470 }, { "epoch": 0.9545462813074083, "grad_norm": 1.8015400171279907, "learning_rate": 4.687880890322339e-07, "loss": 0.9048, "step": 52480 }, { "epoch": 0.9547281689372306, "grad_norm": 1.9891849756240845, "learning_rate": 4.66912936676105e-07, "loss": 0.9025, "step": 52490 }, { "epoch": 0.9549100565670529, "grad_norm": 1.670513391494751, "learning_rate": 4.6503778431997606e-07, "loss": 0.8899, "step": 52500 }, { "epoch": 0.9550919441968752, "grad_norm": 1.5729060173034668, "learning_rate": 4.631626319638471e-07, "loss": 0.9209, "step": 52510 }, { "epoch": 0.9552738318266975, "grad_norm": 1.6298617124557495, "learning_rate": 4.6128747960771817e-07, "loss": 0.9194, "step": 52520 }, { "epoch": 0.9554557194565197, "grad_norm": 1.7886607646942139, "learning_rate": 4.594123272515892e-07, "loss": 0.8727, "step": 52530 }, { "epoch": 0.955637607086342, "grad_norm": 1.8196879625320435, "learning_rate": 4.5753717489546033e-07, "loss": 0.9165, "step": 52540 }, { "epoch": 0.9558194947161643, "grad_norm": 1.8953953981399536, "learning_rate": 4.5566202253933136e-07, "loss": 0.8814, "step": 52550 }, { "epoch": 0.9560013823459866, "grad_norm": 1.669822096824646, "learning_rate": 4.537868701832024e-07, "loss": 0.8527, "step": 52560 }, { "epoch": 0.9561832699758089, "grad_norm": 1.8620238304138184, "learning_rate": 4.5191171782707347e-07, "loss": 0.8881, "step": 52570 }, { "epoch": 0.9563651576056312, "grad_norm": 1.7563786506652832, "learning_rate": 4.5003656547094455e-07, "loss": 0.8822, "step": 52580 }, { "epoch": 0.9565470452354535, "grad_norm": 1.6187026500701904, "learning_rate": 4.4816141311481563e-07, "loss": 0.8921, "step": 52590 }, { "epoch": 0.9567289328652758, "grad_norm": 2.428680896759033, "learning_rate": 4.4628626075868666e-07, "loss": 0.896, "step": 52600 }, { "epoch": 0.9569108204950981, "grad_norm": 1.7326472997665405, "learning_rate": 4.444111084025578e-07, "loss": 0.8725, "step": 52610 }, { "epoch": 0.9570927081249204, "grad_norm": 1.652511715888977, "learning_rate": 4.425359560464288e-07, "loss": 0.8815, "step": 52620 }, { "epoch": 0.9572745957547427, "grad_norm": 1.7680652141571045, "learning_rate": 4.4066080369029985e-07, "loss": 0.9376, "step": 52630 }, { "epoch": 0.957456483384565, "grad_norm": 1.742314338684082, "learning_rate": 4.3878565133417094e-07, "loss": 0.8975, "step": 52640 }, { "epoch": 0.9576383710143873, "grad_norm": 1.8613542318344116, "learning_rate": 4.3691049897804196e-07, "loss": 0.947, "step": 52650 }, { "epoch": 0.9578202586442096, "grad_norm": 1.6704055070877075, "learning_rate": 4.350353466219131e-07, "loss": 0.8506, "step": 52660 }, { "epoch": 0.9580021462740319, "grad_norm": 1.8931320905685425, "learning_rate": 4.3316019426578413e-07, "loss": 0.8931, "step": 52670 }, { "epoch": 0.9581840339038542, "grad_norm": 1.8270903825759888, "learning_rate": 4.312850419096552e-07, "loss": 0.8674, "step": 52680 }, { "epoch": 0.9583659215336765, "grad_norm": 1.9683502912521362, "learning_rate": 4.2940988955352624e-07, "loss": 0.9078, "step": 52690 }, { "epoch": 0.9585478091634988, "grad_norm": 1.814916968345642, "learning_rate": 4.275347371973973e-07, "loss": 0.8928, "step": 52700 }, { "epoch": 0.9587296967933211, "grad_norm": 1.8994702100753784, "learning_rate": 4.256595848412684e-07, "loss": 0.8721, "step": 52710 }, { "epoch": 0.9589115844231434, "grad_norm": 1.5650765895843506, "learning_rate": 4.2378443248513943e-07, "loss": 0.8912, "step": 52720 }, { "epoch": 0.9590934720529657, "grad_norm": 2.037667989730835, "learning_rate": 4.2190928012901056e-07, "loss": 0.8517, "step": 52730 }, { "epoch": 0.959275359682788, "grad_norm": 1.6376808881759644, "learning_rate": 4.200341277728816e-07, "loss": 0.9108, "step": 52740 }, { "epoch": 0.9594572473126103, "grad_norm": 1.8358657360076904, "learning_rate": 4.181589754167526e-07, "loss": 0.8922, "step": 52750 }, { "epoch": 0.9596391349424326, "grad_norm": 1.7188764810562134, "learning_rate": 4.162838230606237e-07, "loss": 0.9185, "step": 52760 }, { "epoch": 0.9598210225722549, "grad_norm": 1.687708854675293, "learning_rate": 4.1440867070449473e-07, "loss": 0.8877, "step": 52770 }, { "epoch": 0.9600029102020772, "grad_norm": 1.7727680206298828, "learning_rate": 4.1253351834836586e-07, "loss": 0.9252, "step": 52780 }, { "epoch": 0.9601847978318995, "grad_norm": 1.9233661890029907, "learning_rate": 4.106583659922369e-07, "loss": 0.8963, "step": 52790 }, { "epoch": 0.9603666854617218, "grad_norm": 1.5684216022491455, "learning_rate": 4.0878321363610797e-07, "loss": 0.8713, "step": 52800 }, { "epoch": 0.9605485730915441, "grad_norm": 1.6073098182678223, "learning_rate": 4.0690806127997905e-07, "loss": 0.9224, "step": 52810 }, { "epoch": 0.9607304607213664, "grad_norm": 1.569822907447815, "learning_rate": 4.050329089238501e-07, "loss": 0.8787, "step": 52820 }, { "epoch": 0.9609123483511887, "grad_norm": 2.2744388580322266, "learning_rate": 4.0315775656772116e-07, "loss": 0.9431, "step": 52830 }, { "epoch": 0.961094235981011, "grad_norm": 1.7103184461593628, "learning_rate": 4.012826042115922e-07, "loss": 0.8421, "step": 52840 }, { "epoch": 0.9612761236108333, "grad_norm": 1.9208362102508545, "learning_rate": 3.994074518554633e-07, "loss": 0.926, "step": 52850 }, { "epoch": 0.9614580112406556, "grad_norm": 1.7388149499893188, "learning_rate": 3.9753229949933435e-07, "loss": 0.8826, "step": 52860 }, { "epoch": 0.9616398988704778, "grad_norm": 1.7541353702545166, "learning_rate": 3.956571471432054e-07, "loss": 0.9389, "step": 52870 }, { "epoch": 0.9618217865003001, "grad_norm": 2.083998680114746, "learning_rate": 3.9378199478707646e-07, "loss": 0.8952, "step": 52880 }, { "epoch": 0.9620036741301224, "grad_norm": 2.6282780170440674, "learning_rate": 3.9190684243094755e-07, "loss": 0.8697, "step": 52890 }, { "epoch": 0.9621855617599447, "grad_norm": 1.7573750019073486, "learning_rate": 3.9003169007481863e-07, "loss": 0.8998, "step": 52900 }, { "epoch": 0.962367449389767, "grad_norm": 1.97195303440094, "learning_rate": 3.8815653771868966e-07, "loss": 0.9134, "step": 52910 }, { "epoch": 0.9625493370195893, "grad_norm": 1.4472529888153076, "learning_rate": 3.862813853625608e-07, "loss": 0.8425, "step": 52920 }, { "epoch": 0.9627312246494116, "grad_norm": 1.8536056280136108, "learning_rate": 3.844062330064318e-07, "loss": 0.9322, "step": 52930 }, { "epoch": 0.9629131122792339, "grad_norm": 1.7456477880477905, "learning_rate": 3.8253108065030285e-07, "loss": 0.918, "step": 52940 }, { "epoch": 0.9630949999090562, "grad_norm": 1.5978821516036987, "learning_rate": 3.8065592829417393e-07, "loss": 0.9096, "step": 52950 }, { "epoch": 0.9632768875388785, "grad_norm": 1.4595468044281006, "learning_rate": 3.7878077593804496e-07, "loss": 0.8926, "step": 52960 }, { "epoch": 0.9634587751687008, "grad_norm": 1.572182059288025, "learning_rate": 3.769056235819161e-07, "loss": 0.8693, "step": 52970 }, { "epoch": 0.9636406627985231, "grad_norm": 1.4221917390823364, "learning_rate": 3.750304712257871e-07, "loss": 0.872, "step": 52980 }, { "epoch": 0.9638225504283454, "grad_norm": 1.8172396421432495, "learning_rate": 3.731553188696582e-07, "loss": 0.9109, "step": 52990 }, { "epoch": 0.9640044380581677, "grad_norm": 1.954426884651184, "learning_rate": 3.712801665135293e-07, "loss": 0.8644, "step": 53000 }, { "epoch": 0.96418632568799, "grad_norm": 2.0348682403564453, "learning_rate": 3.694050141574003e-07, "loss": 0.9303, "step": 53010 }, { "epoch": 0.9643682133178123, "grad_norm": 1.9775394201278687, "learning_rate": 3.675298618012714e-07, "loss": 0.9167, "step": 53020 }, { "epoch": 0.9645501009476346, "grad_norm": 1.85639226436615, "learning_rate": 3.656547094451424e-07, "loss": 0.8898, "step": 53030 }, { "epoch": 0.9647319885774569, "grad_norm": 1.6937601566314697, "learning_rate": 3.6377955708901355e-07, "loss": 0.9153, "step": 53040 }, { "epoch": 0.9649138762072792, "grad_norm": 1.564937949180603, "learning_rate": 3.619044047328846e-07, "loss": 0.9227, "step": 53050 }, { "epoch": 0.9650957638371014, "grad_norm": 1.6579904556274414, "learning_rate": 3.600292523767556e-07, "loss": 0.8659, "step": 53060 }, { "epoch": 0.9652776514669237, "grad_norm": 1.7725434303283691, "learning_rate": 3.581541000206267e-07, "loss": 0.8865, "step": 53070 }, { "epoch": 0.965459539096746, "grad_norm": 1.7490605115890503, "learning_rate": 3.562789476644978e-07, "loss": 0.9291, "step": 53080 }, { "epoch": 0.9656414267265683, "grad_norm": 1.6928088665008545, "learning_rate": 3.5440379530836886e-07, "loss": 0.9085, "step": 53090 }, { "epoch": 0.9658233143563906, "grad_norm": 1.9419893026351929, "learning_rate": 3.525286429522399e-07, "loss": 0.8871, "step": 53100 }, { "epoch": 0.9660052019862129, "grad_norm": 1.8326853513717651, "learning_rate": 3.50653490596111e-07, "loss": 0.9914, "step": 53110 }, { "epoch": 0.9661870896160352, "grad_norm": 2.1202774047851562, "learning_rate": 3.4877833823998205e-07, "loss": 0.923, "step": 53120 }, { "epoch": 0.9663689772458575, "grad_norm": 1.695317029953003, "learning_rate": 3.469031858838531e-07, "loss": 0.877, "step": 53130 }, { "epoch": 0.9665508648756798, "grad_norm": 1.865455985069275, "learning_rate": 3.4502803352772416e-07, "loss": 0.8965, "step": 53140 }, { "epoch": 0.9667327525055021, "grad_norm": 2.123208522796631, "learning_rate": 3.431528811715952e-07, "loss": 0.9032, "step": 53150 }, { "epoch": 0.9669146401353244, "grad_norm": 1.907958984375, "learning_rate": 3.412777288154663e-07, "loss": 0.8931, "step": 53160 }, { "epoch": 0.9670965277651467, "grad_norm": 1.9719887971878052, "learning_rate": 3.3940257645933735e-07, "loss": 0.9038, "step": 53170 }, { "epoch": 0.967278415394969, "grad_norm": 1.6569899320602417, "learning_rate": 3.375274241032084e-07, "loss": 0.8691, "step": 53180 }, { "epoch": 0.9674603030247912, "grad_norm": 1.565130591392517, "learning_rate": 3.356522717470795e-07, "loss": 0.8986, "step": 53190 }, { "epoch": 0.9676421906546135, "grad_norm": 1.6619609594345093, "learning_rate": 3.3377711939095054e-07, "loss": 0.874, "step": 53200 }, { "epoch": 0.9678240782844358, "grad_norm": 1.7984410524368286, "learning_rate": 3.319019670348216e-07, "loss": 0.8931, "step": 53210 }, { "epoch": 0.9680059659142581, "grad_norm": 1.9547232389450073, "learning_rate": 3.3002681467869265e-07, "loss": 0.8809, "step": 53220 }, { "epoch": 0.9681878535440804, "grad_norm": 1.9077638387680054, "learning_rate": 3.281516623225638e-07, "loss": 0.9414, "step": 53230 }, { "epoch": 0.9683697411739027, "grad_norm": 1.9439486265182495, "learning_rate": 3.262765099664348e-07, "loss": 0.8854, "step": 53240 }, { "epoch": 0.968551628803725, "grad_norm": 1.7924941778182983, "learning_rate": 3.2440135761030584e-07, "loss": 0.9053, "step": 53250 }, { "epoch": 0.9687335164335473, "grad_norm": 1.8472057580947876, "learning_rate": 3.225262052541769e-07, "loss": 0.8687, "step": 53260 }, { "epoch": 0.9689154040633696, "grad_norm": 1.743963360786438, "learning_rate": 3.2065105289804795e-07, "loss": 0.9261, "step": 53270 }, { "epoch": 0.9690972916931919, "grad_norm": 1.9565738439559937, "learning_rate": 3.187759005419191e-07, "loss": 0.9159, "step": 53280 }, { "epoch": 0.9692791793230142, "grad_norm": 1.5245027542114258, "learning_rate": 3.169007481857901e-07, "loss": 0.8644, "step": 53290 }, { "epoch": 0.9694610669528365, "grad_norm": 1.7999204397201538, "learning_rate": 3.150255958296612e-07, "loss": 0.8587, "step": 53300 }, { "epoch": 0.9696429545826588, "grad_norm": 1.5883069038391113, "learning_rate": 3.131504434735323e-07, "loss": 0.9148, "step": 53310 }, { "epoch": 0.9698248422124811, "grad_norm": 1.798343300819397, "learning_rate": 3.112752911174033e-07, "loss": 0.9222, "step": 53320 }, { "epoch": 0.9700067298423034, "grad_norm": 1.9169659614562988, "learning_rate": 3.094001387612744e-07, "loss": 0.8916, "step": 53330 }, { "epoch": 0.9701886174721257, "grad_norm": 1.589854121208191, "learning_rate": 3.075249864051454e-07, "loss": 0.8491, "step": 53340 }, { "epoch": 0.970370505101948, "grad_norm": 1.8152964115142822, "learning_rate": 3.056498340490165e-07, "loss": 0.9286, "step": 53350 }, { "epoch": 0.9705523927317703, "grad_norm": 1.7528034448623657, "learning_rate": 3.037746816928876e-07, "loss": 0.9374, "step": 53360 }, { "epoch": 0.9707342803615926, "grad_norm": 1.7482283115386963, "learning_rate": 3.0189952933675866e-07, "loss": 0.8791, "step": 53370 }, { "epoch": 0.9709161679914149, "grad_norm": 2.149507761001587, "learning_rate": 3.000243769806297e-07, "loss": 0.9215, "step": 53380 }, { "epoch": 0.9710980556212372, "grad_norm": 1.823594093322754, "learning_rate": 2.9814922462450077e-07, "loss": 0.8948, "step": 53390 }, { "epoch": 0.9712799432510595, "grad_norm": 1.6475759744644165, "learning_rate": 2.962740722683718e-07, "loss": 0.9014, "step": 53400 }, { "epoch": 0.9714618308808818, "grad_norm": 2.014751434326172, "learning_rate": 2.943989199122429e-07, "loss": 0.8713, "step": 53410 }, { "epoch": 0.9716437185107041, "grad_norm": 1.465847134590149, "learning_rate": 2.9252376755611396e-07, "loss": 0.8725, "step": 53420 }, { "epoch": 0.9718256061405264, "grad_norm": 2.1486079692840576, "learning_rate": 2.9064861519998504e-07, "loss": 0.8509, "step": 53430 }, { "epoch": 0.9720074937703487, "grad_norm": 1.795898675918579, "learning_rate": 2.887734628438561e-07, "loss": 0.9795, "step": 53440 }, { "epoch": 0.972189381400171, "grad_norm": 1.4920406341552734, "learning_rate": 2.8689831048772715e-07, "loss": 0.8703, "step": 53450 }, { "epoch": 0.9723712690299933, "grad_norm": 1.735447883605957, "learning_rate": 2.850231581315982e-07, "loss": 0.8702, "step": 53460 }, { "epoch": 0.9725531566598156, "grad_norm": 1.8761446475982666, "learning_rate": 2.8314800577546926e-07, "loss": 0.9115, "step": 53470 }, { "epoch": 0.9727350442896379, "grad_norm": 1.8616496324539185, "learning_rate": 2.8127285341934034e-07, "loss": 0.8654, "step": 53480 }, { "epoch": 0.9729169319194602, "grad_norm": 1.6233702898025513, "learning_rate": 2.793977010632114e-07, "loss": 0.8601, "step": 53490 }, { "epoch": 0.9730988195492825, "grad_norm": 1.7886806726455688, "learning_rate": 2.775225487070825e-07, "loss": 0.8382, "step": 53500 }, { "epoch": 0.9732807071791048, "grad_norm": 1.4740511178970337, "learning_rate": 2.7564739635095353e-07, "loss": 0.8942, "step": 53510 }, { "epoch": 0.973462594808927, "grad_norm": 1.8439741134643555, "learning_rate": 2.737722439948246e-07, "loss": 0.9292, "step": 53520 }, { "epoch": 0.9736444824387493, "grad_norm": 1.7670116424560547, "learning_rate": 2.7189709163869564e-07, "loss": 0.8707, "step": 53530 }, { "epoch": 0.9738263700685716, "grad_norm": 1.6349842548370361, "learning_rate": 2.700219392825667e-07, "loss": 0.8973, "step": 53540 }, { "epoch": 0.9740082576983939, "grad_norm": 1.641207218170166, "learning_rate": 2.681467869264378e-07, "loss": 0.8853, "step": 53550 }, { "epoch": 0.9741901453282162, "grad_norm": 1.3784902095794678, "learning_rate": 2.662716345703089e-07, "loss": 0.8932, "step": 53560 }, { "epoch": 0.9743720329580385, "grad_norm": 1.720582127571106, "learning_rate": 2.643964822141799e-07, "loss": 0.8808, "step": 53570 }, { "epoch": 0.9745539205878608, "grad_norm": 1.766666293144226, "learning_rate": 2.62521329858051e-07, "loss": 0.8913, "step": 53580 }, { "epoch": 0.9747358082176831, "grad_norm": 1.7842615842819214, "learning_rate": 2.60646177501922e-07, "loss": 0.9047, "step": 53590 }, { "epoch": 0.9749176958475054, "grad_norm": 1.5344343185424805, "learning_rate": 2.587710251457931e-07, "loss": 0.8399, "step": 53600 }, { "epoch": 0.9750995834773277, "grad_norm": 1.9425544738769531, "learning_rate": 2.568958727896642e-07, "loss": 0.8855, "step": 53610 }, { "epoch": 0.97528147110715, "grad_norm": 1.4499845504760742, "learning_rate": 2.5502072043353527e-07, "loss": 0.8795, "step": 53620 }, { "epoch": 0.9754633587369723, "grad_norm": 1.695581078529358, "learning_rate": 2.531455680774063e-07, "loss": 0.8983, "step": 53630 }, { "epoch": 0.9756452463667946, "grad_norm": 2.0921430587768555, "learning_rate": 2.512704157212774e-07, "loss": 0.9343, "step": 53640 }, { "epoch": 0.9758271339966169, "grad_norm": 1.6871775388717651, "learning_rate": 2.493952633651484e-07, "loss": 0.8724, "step": 53650 }, { "epoch": 0.9760090216264392, "grad_norm": 1.5626379251480103, "learning_rate": 2.475201110090195e-07, "loss": 0.8396, "step": 53660 }, { "epoch": 0.9761909092562615, "grad_norm": 1.982351303100586, "learning_rate": 2.4564495865289057e-07, "loss": 0.9085, "step": 53670 }, { "epoch": 0.9763727968860838, "grad_norm": 1.8668577671051025, "learning_rate": 2.4376980629676165e-07, "loss": 0.9371, "step": 53680 }, { "epoch": 0.9765546845159061, "grad_norm": 1.549087643623352, "learning_rate": 2.4189465394063273e-07, "loss": 0.8914, "step": 53690 }, { "epoch": 0.9767365721457284, "grad_norm": 1.7944905757904053, "learning_rate": 2.4001950158450376e-07, "loss": 0.8603, "step": 53700 }, { "epoch": 0.9769184597755507, "grad_norm": 1.9123302698135376, "learning_rate": 2.3814434922837481e-07, "loss": 0.9059, "step": 53710 }, { "epoch": 0.977100347405373, "grad_norm": 2.768615484237671, "learning_rate": 2.3626919687224587e-07, "loss": 0.9108, "step": 53720 }, { "epoch": 0.9772822350351953, "grad_norm": 2.0607664585113525, "learning_rate": 2.3439404451611695e-07, "loss": 0.9546, "step": 53730 }, { "epoch": 0.9774641226650176, "grad_norm": 1.8563134670257568, "learning_rate": 2.3251889215998803e-07, "loss": 0.9011, "step": 53740 }, { "epoch": 0.9776460102948399, "grad_norm": 1.4857020378112793, "learning_rate": 2.3064373980385909e-07, "loss": 0.8862, "step": 53750 }, { "epoch": 0.9778278979246622, "grad_norm": 1.8280316591262817, "learning_rate": 2.2876858744773017e-07, "loss": 0.872, "step": 53760 }, { "epoch": 0.9780097855544845, "grad_norm": 1.934570550918579, "learning_rate": 2.268934350916012e-07, "loss": 0.8954, "step": 53770 }, { "epoch": 0.9781916731843068, "grad_norm": 1.7383111715316772, "learning_rate": 2.2501828273547228e-07, "loss": 0.8521, "step": 53780 }, { "epoch": 0.9783735608141291, "grad_norm": 2.0376129150390625, "learning_rate": 2.2314313037934333e-07, "loss": 0.9012, "step": 53790 }, { "epoch": 0.9785554484439514, "grad_norm": 1.8012861013412476, "learning_rate": 2.212679780232144e-07, "loss": 0.8918, "step": 53800 }, { "epoch": 0.9787373360737737, "grad_norm": 1.5509483814239502, "learning_rate": 2.1939282566708547e-07, "loss": 0.8454, "step": 53810 }, { "epoch": 0.978919223703596, "grad_norm": 1.7901334762573242, "learning_rate": 2.1751767331095655e-07, "loss": 0.887, "step": 53820 }, { "epoch": 0.9791011113334183, "grad_norm": 1.6910285949707031, "learning_rate": 2.156425209548276e-07, "loss": 0.8841, "step": 53830 }, { "epoch": 0.9792829989632406, "grad_norm": 1.9000298976898193, "learning_rate": 2.1376736859869866e-07, "loss": 0.9097, "step": 53840 }, { "epoch": 0.9794648865930629, "grad_norm": 1.815666675567627, "learning_rate": 2.1189221624256971e-07, "loss": 0.8448, "step": 53850 }, { "epoch": 0.9796467742228852, "grad_norm": 1.4845683574676514, "learning_rate": 2.100170638864408e-07, "loss": 0.9011, "step": 53860 }, { "epoch": 0.9798286618527074, "grad_norm": 1.6411792039871216, "learning_rate": 2.0814191153031185e-07, "loss": 0.9205, "step": 53870 }, { "epoch": 0.9800105494825297, "grad_norm": 2.0277342796325684, "learning_rate": 2.0626675917418293e-07, "loss": 0.9151, "step": 53880 }, { "epoch": 0.980192437112352, "grad_norm": 1.980673909187317, "learning_rate": 2.0439160681805399e-07, "loss": 0.8991, "step": 53890 }, { "epoch": 0.9803743247421742, "grad_norm": 1.8375296592712402, "learning_rate": 2.0251645446192504e-07, "loss": 0.9435, "step": 53900 }, { "epoch": 0.9805562123719965, "grad_norm": 1.601364254951477, "learning_rate": 2.006413021057961e-07, "loss": 0.9325, "step": 53910 }, { "epoch": 0.9807381000018188, "grad_norm": 1.9345512390136719, "learning_rate": 1.9876614974966718e-07, "loss": 0.872, "step": 53920 }, { "epoch": 0.9809199876316411, "grad_norm": 1.6383129358291626, "learning_rate": 1.9689099739353823e-07, "loss": 0.8769, "step": 53930 }, { "epoch": 0.9811018752614634, "grad_norm": 1.8747522830963135, "learning_rate": 1.9501584503740931e-07, "loss": 0.9094, "step": 53940 }, { "epoch": 0.9812837628912857, "grad_norm": 1.8687227964401245, "learning_rate": 1.931406926812804e-07, "loss": 0.8862, "step": 53950 }, { "epoch": 0.981465650521108, "grad_norm": 1.969436526298523, "learning_rate": 1.9126554032515142e-07, "loss": 0.901, "step": 53960 }, { "epoch": 0.9816475381509303, "grad_norm": 1.7446562051773071, "learning_rate": 1.8939038796902248e-07, "loss": 0.9282, "step": 53970 }, { "epoch": 0.9818294257807526, "grad_norm": 1.793647289276123, "learning_rate": 1.8751523561289356e-07, "loss": 0.9146, "step": 53980 }, { "epoch": 0.9820113134105749, "grad_norm": 1.9479408264160156, "learning_rate": 1.8564008325676464e-07, "loss": 0.9307, "step": 53990 }, { "epoch": 0.9821932010403972, "grad_norm": 1.8225338459014893, "learning_rate": 1.837649309006357e-07, "loss": 0.8904, "step": 54000 }, { "epoch": 0.9823750886702195, "grad_norm": 1.8953123092651367, "learning_rate": 1.8188977854450678e-07, "loss": 0.8885, "step": 54010 }, { "epoch": 0.9825569763000418, "grad_norm": 1.7914330959320068, "learning_rate": 1.800146261883778e-07, "loss": 0.9116, "step": 54020 }, { "epoch": 0.9827388639298641, "grad_norm": 1.8612347841262817, "learning_rate": 1.781394738322489e-07, "loss": 0.9354, "step": 54030 }, { "epoch": 0.9829207515596864, "grad_norm": 1.9441192150115967, "learning_rate": 1.7626432147611994e-07, "loss": 0.8784, "step": 54040 }, { "epoch": 0.9831026391895087, "grad_norm": 1.8987913131713867, "learning_rate": 1.7438916911999102e-07, "loss": 0.8467, "step": 54050 }, { "epoch": 0.983284526819331, "grad_norm": 1.622682809829712, "learning_rate": 1.7251401676386208e-07, "loss": 0.8451, "step": 54060 }, { "epoch": 0.9834664144491533, "grad_norm": 1.9034028053283691, "learning_rate": 1.7063886440773316e-07, "loss": 0.9107, "step": 54070 }, { "epoch": 0.9836483020789756, "grad_norm": 1.8997676372528076, "learning_rate": 1.687637120516042e-07, "loss": 0.8817, "step": 54080 }, { "epoch": 0.9838301897087979, "grad_norm": 1.7178410291671753, "learning_rate": 1.6688855969547527e-07, "loss": 0.9336, "step": 54090 }, { "epoch": 0.9840120773386202, "grad_norm": 1.8917608261108398, "learning_rate": 1.6501340733934632e-07, "loss": 0.887, "step": 54100 }, { "epoch": 0.9841939649684425, "grad_norm": 1.838948130607605, "learning_rate": 1.631382549832174e-07, "loss": 0.912, "step": 54110 }, { "epoch": 0.9843758525982648, "grad_norm": 1.633209228515625, "learning_rate": 1.6126310262708846e-07, "loss": 0.8805, "step": 54120 }, { "epoch": 0.9845577402280871, "grad_norm": 1.7487061023712158, "learning_rate": 1.5938795027095954e-07, "loss": 0.9159, "step": 54130 }, { "epoch": 0.9847396278579094, "grad_norm": 1.6534188985824585, "learning_rate": 1.575127979148306e-07, "loss": 0.8991, "step": 54140 }, { "epoch": 0.9849215154877317, "grad_norm": 1.9699021577835083, "learning_rate": 1.5563764555870165e-07, "loss": 0.9108, "step": 54150 }, { "epoch": 0.985103403117554, "grad_norm": 1.6002508401870728, "learning_rate": 1.537624932025727e-07, "loss": 0.9114, "step": 54160 }, { "epoch": 0.9852852907473763, "grad_norm": 1.7416579723358154, "learning_rate": 1.518873408464438e-07, "loss": 0.8563, "step": 54170 }, { "epoch": 0.9854671783771985, "grad_norm": 1.7942638397216797, "learning_rate": 1.5001218849031484e-07, "loss": 0.8828, "step": 54180 }, { "epoch": 0.9856490660070208, "grad_norm": 1.877023458480835, "learning_rate": 1.481370361341859e-07, "loss": 0.9082, "step": 54190 }, { "epoch": 0.9858309536368431, "grad_norm": 1.7696330547332764, "learning_rate": 1.4626188377805698e-07, "loss": 0.8544, "step": 54200 }, { "epoch": 0.9860128412666654, "grad_norm": 1.7806761264801025, "learning_rate": 1.4438673142192806e-07, "loss": 0.8888, "step": 54210 }, { "epoch": 0.9861947288964877, "grad_norm": 2.0327887535095215, "learning_rate": 1.425115790657991e-07, "loss": 0.8706, "step": 54220 }, { "epoch": 0.98637661652631, "grad_norm": 1.6407501697540283, "learning_rate": 1.4063642670967017e-07, "loss": 0.8738, "step": 54230 }, { "epoch": 0.9865585041561323, "grad_norm": 1.8894076347351074, "learning_rate": 1.3876127435354125e-07, "loss": 0.9314, "step": 54240 }, { "epoch": 0.9867403917859546, "grad_norm": 1.8624190092086792, "learning_rate": 1.368861219974123e-07, "loss": 0.896, "step": 54250 }, { "epoch": 0.9869222794157769, "grad_norm": 1.767255187034607, "learning_rate": 1.3501096964128336e-07, "loss": 0.8535, "step": 54260 }, { "epoch": 0.9871041670455992, "grad_norm": 1.6854933500289917, "learning_rate": 1.3313581728515444e-07, "loss": 0.9384, "step": 54270 }, { "epoch": 0.9872860546754215, "grad_norm": 1.6461013555526733, "learning_rate": 1.312606649290255e-07, "loss": 0.8606, "step": 54280 }, { "epoch": 0.9874679423052438, "grad_norm": 1.7928451299667358, "learning_rate": 1.2938551257289655e-07, "loss": 0.8751, "step": 54290 }, { "epoch": 0.9876498299350661, "grad_norm": 2.050334930419922, "learning_rate": 1.2751036021676763e-07, "loss": 0.9696, "step": 54300 }, { "epoch": 0.9878317175648884, "grad_norm": 1.5446444749832153, "learning_rate": 1.256352078606387e-07, "loss": 0.8433, "step": 54310 }, { "epoch": 0.9880136051947107, "grad_norm": 1.8227653503417969, "learning_rate": 1.2376005550450974e-07, "loss": 0.9081, "step": 54320 }, { "epoch": 0.988195492824533, "grad_norm": 1.7241973876953125, "learning_rate": 1.2188490314838082e-07, "loss": 0.877, "step": 54330 }, { "epoch": 0.9883773804543553, "grad_norm": 1.9050542116165161, "learning_rate": 1.2000975079225188e-07, "loss": 0.875, "step": 54340 }, { "epoch": 0.9885592680841776, "grad_norm": 1.3942615985870361, "learning_rate": 1.1813459843612293e-07, "loss": 0.8483, "step": 54350 }, { "epoch": 0.9887411557139999, "grad_norm": 2.131425619125366, "learning_rate": 1.1625944607999402e-07, "loss": 0.9181, "step": 54360 }, { "epoch": 0.9889230433438222, "grad_norm": 1.8306736946105957, "learning_rate": 1.1438429372386508e-07, "loss": 0.9186, "step": 54370 }, { "epoch": 0.9891049309736445, "grad_norm": 1.8415859937667847, "learning_rate": 1.1250914136773614e-07, "loss": 0.8545, "step": 54380 }, { "epoch": 0.9892868186034668, "grad_norm": 1.8132848739624023, "learning_rate": 1.106339890116072e-07, "loss": 0.8315, "step": 54390 }, { "epoch": 0.9894687062332891, "grad_norm": 1.588100790977478, "learning_rate": 1.0875883665547827e-07, "loss": 0.8651, "step": 54400 }, { "epoch": 0.9896505938631114, "grad_norm": 1.5314209461212158, "learning_rate": 1.0688368429934933e-07, "loss": 0.8609, "step": 54410 }, { "epoch": 0.9898324814929337, "grad_norm": 1.859622836112976, "learning_rate": 1.050085319432204e-07, "loss": 0.8614, "step": 54420 }, { "epoch": 0.990014369122756, "grad_norm": 1.6222505569458008, "learning_rate": 1.0313337958709147e-07, "loss": 0.8486, "step": 54430 }, { "epoch": 0.9901962567525783, "grad_norm": 1.7431626319885254, "learning_rate": 1.0125822723096252e-07, "loss": 0.8872, "step": 54440 }, { "epoch": 0.9903781443824006, "grad_norm": 1.75123929977417, "learning_rate": 9.938307487483359e-08, "loss": 0.8597, "step": 54450 }, { "epoch": 0.9905600320122229, "grad_norm": 1.8754692077636719, "learning_rate": 9.750792251870466e-08, "loss": 0.9286, "step": 54460 }, { "epoch": 0.9907419196420452, "grad_norm": 1.7613874673843384, "learning_rate": 9.563277016257571e-08, "loss": 0.9274, "step": 54470 }, { "epoch": 0.9909238072718675, "grad_norm": 2.216367483139038, "learning_rate": 9.375761780644678e-08, "loss": 0.9231, "step": 54480 }, { "epoch": 0.9911056949016898, "grad_norm": 1.733625054359436, "learning_rate": 9.188246545031785e-08, "loss": 0.8639, "step": 54490 }, { "epoch": 0.9912875825315121, "grad_norm": 2.0046017169952393, "learning_rate": 9.00073130941889e-08, "loss": 0.9279, "step": 54500 }, { "epoch": 0.9914694701613344, "grad_norm": 1.6595464944839478, "learning_rate": 8.813216073805997e-08, "loss": 0.8956, "step": 54510 }, { "epoch": 0.9916513577911567, "grad_norm": 2.118455410003662, "learning_rate": 8.625700838193104e-08, "loss": 0.8661, "step": 54520 }, { "epoch": 0.991833245420979, "grad_norm": 1.5925854444503784, "learning_rate": 8.43818560258021e-08, "loss": 0.894, "step": 54530 }, { "epoch": 0.9920151330508012, "grad_norm": 1.5309463739395142, "learning_rate": 8.250670366967316e-08, "loss": 0.8322, "step": 54540 }, { "epoch": 0.9921970206806235, "grad_norm": 2.10439133644104, "learning_rate": 8.063155131354423e-08, "loss": 0.9242, "step": 54550 }, { "epoch": 0.9923789083104458, "grad_norm": 1.8576916456222534, "learning_rate": 7.87563989574153e-08, "loss": 0.9263, "step": 54560 }, { "epoch": 0.9925607959402681, "grad_norm": 1.6757699251174927, "learning_rate": 7.688124660128635e-08, "loss": 0.9149, "step": 54570 }, { "epoch": 0.9927426835700904, "grad_norm": 1.266911506652832, "learning_rate": 7.500609424515742e-08, "loss": 0.8856, "step": 54580 }, { "epoch": 0.9929245711999127, "grad_norm": 1.7421590089797974, "learning_rate": 7.313094188902849e-08, "loss": 0.9208, "step": 54590 }, { "epoch": 0.993106458829735, "grad_norm": 1.4577891826629639, "learning_rate": 7.125578953289954e-08, "loss": 0.9026, "step": 54600 }, { "epoch": 0.9932883464595573, "grad_norm": 1.7901201248168945, "learning_rate": 6.938063717677063e-08, "loss": 0.8716, "step": 54610 }, { "epoch": 0.9934702340893796, "grad_norm": 1.5472559928894043, "learning_rate": 6.750548482064168e-08, "loss": 0.9001, "step": 54620 }, { "epoch": 0.9936521217192019, "grad_norm": 2.012564182281494, "learning_rate": 6.563033246451275e-08, "loss": 0.8702, "step": 54630 }, { "epoch": 0.9938340093490242, "grad_norm": 1.5856733322143555, "learning_rate": 6.375518010838382e-08, "loss": 0.9076, "step": 54640 }, { "epoch": 0.9940158969788465, "grad_norm": 2.6662232875823975, "learning_rate": 6.188002775225487e-08, "loss": 0.9497, "step": 54650 }, { "epoch": 0.9941977846086688, "grad_norm": 1.9168440103530884, "learning_rate": 6.000487539612594e-08, "loss": 0.9113, "step": 54660 }, { "epoch": 0.9943796722384911, "grad_norm": 1.7326258420944214, "learning_rate": 5.812972303999701e-08, "loss": 0.9053, "step": 54670 }, { "epoch": 0.9945615598683134, "grad_norm": 1.6351253986358643, "learning_rate": 5.625457068386807e-08, "loss": 0.8334, "step": 54680 }, { "epoch": 0.9947434474981357, "grad_norm": 1.8571479320526123, "learning_rate": 5.437941832773914e-08, "loss": 0.8642, "step": 54690 }, { "epoch": 0.994925335127958, "grad_norm": 1.833138346672058, "learning_rate": 5.25042659716102e-08, "loss": 0.8897, "step": 54700 }, { "epoch": 0.9951072227577803, "grad_norm": 1.679102897644043, "learning_rate": 5.062911361548126e-08, "loss": 0.9059, "step": 54710 }, { "epoch": 0.9952891103876026, "grad_norm": 2.013524293899536, "learning_rate": 4.875396125935233e-08, "loss": 0.8678, "step": 54720 }, { "epoch": 0.9954709980174248, "grad_norm": 1.6181925535202026, "learning_rate": 4.687880890322339e-08, "loss": 0.8784, "step": 54730 }, { "epoch": 0.9956528856472471, "grad_norm": 1.91368567943573, "learning_rate": 4.500365654709445e-08, "loss": 0.9072, "step": 54740 }, { "epoch": 0.9958347732770694, "grad_norm": 1.6468901634216309, "learning_rate": 4.312850419096552e-08, "loss": 0.8944, "step": 54750 }, { "epoch": 0.9960166609068917, "grad_norm": 1.7494819164276123, "learning_rate": 4.125335183483658e-08, "loss": 0.8983, "step": 54760 }, { "epoch": 0.996198548536714, "grad_norm": 1.6869902610778809, "learning_rate": 3.937819947870765e-08, "loss": 0.8231, "step": 54770 }, { "epoch": 0.9963804361665363, "grad_norm": 1.9149837493896484, "learning_rate": 3.750304712257871e-08, "loss": 0.8976, "step": 54780 }, { "epoch": 0.9965623237963586, "grad_norm": 1.7124042510986328, "learning_rate": 3.562789476644977e-08, "loss": 0.8979, "step": 54790 }, { "epoch": 0.9967442114261809, "grad_norm": 1.940405011177063, "learning_rate": 3.375274241032084e-08, "loss": 0.9139, "step": 54800 }, { "epoch": 0.9969260990560032, "grad_norm": 1.7076953649520874, "learning_rate": 3.187759005419191e-08, "loss": 0.9063, "step": 54810 }, { "epoch": 0.9971079866858255, "grad_norm": 1.9651620388031006, "learning_rate": 3.000243769806297e-08, "loss": 0.8718, "step": 54820 }, { "epoch": 0.9972898743156478, "grad_norm": 1.6900427341461182, "learning_rate": 2.8127285341934035e-08, "loss": 0.8737, "step": 54830 }, { "epoch": 0.99747176194547, "grad_norm": 1.7732889652252197, "learning_rate": 2.62521329858051e-08, "loss": 0.9103, "step": 54840 }, { "epoch": 0.9976536495752923, "grad_norm": 1.831407070159912, "learning_rate": 2.4376980629676164e-08, "loss": 0.8824, "step": 54850 }, { "epoch": 0.9978355372051146, "grad_norm": 2.0630836486816406, "learning_rate": 2.2501828273547226e-08, "loss": 0.9519, "step": 54860 }, { "epoch": 0.9980174248349369, "grad_norm": 1.9086077213287354, "learning_rate": 2.062667591741829e-08, "loss": 0.9044, "step": 54870 }, { "epoch": 0.9981993124647592, "grad_norm": 1.63712739944458, "learning_rate": 1.8751523561289355e-08, "loss": 0.8753, "step": 54880 }, { "epoch": 0.9983812000945815, "grad_norm": 1.6966229677200317, "learning_rate": 1.687637120516042e-08, "loss": 0.8882, "step": 54890 }, { "epoch": 0.9985630877244038, "grad_norm": 1.8537898063659668, "learning_rate": 1.5001218849031485e-08, "loss": 0.8466, "step": 54900 }, { "epoch": 0.9987449753542261, "grad_norm": 1.6853166818618774, "learning_rate": 1.312606649290255e-08, "loss": 0.9461, "step": 54910 }, { "epoch": 0.9989268629840484, "grad_norm": 2.0704269409179688, "learning_rate": 1.1250914136773613e-08, "loss": 0.9135, "step": 54920 }, { "epoch": 0.9991087506138707, "grad_norm": 1.6965166330337524, "learning_rate": 9.375761780644678e-09, "loss": 0.8522, "step": 54930 }, { "epoch": 0.999290638243693, "grad_norm": 1.7757551670074463, "learning_rate": 7.500609424515742e-09, "loss": 0.88, "step": 54940 }, { "epoch": 0.9994725258735153, "grad_norm": 2.757058620452881, "learning_rate": 5.6254570683868064e-09, "loss": 0.9368, "step": 54950 }, { "epoch": 0.9996544135033376, "grad_norm": 1.6327440738677979, "learning_rate": 3.750304712257871e-09, "loss": 0.9245, "step": 54960 }, { "epoch": 0.9998363011331599, "grad_norm": 1.6856091022491455, "learning_rate": 1.8751523561289356e-09, "loss": 0.8651, "step": 54970 }, { "epoch": 1.0, "step": 54979, "total_flos": 2422898127863808.0, "train_loss": 0.9497830198162343, "train_runtime": 125931.082, "train_samples_per_second": 55.882, "train_steps_per_second": 0.437 } ], "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": 2422898127863808.0, "train_batch_size": 4, "trial_name": null, "trial_params": null }