{"step": 1, "loss": 0.21364526450634003, "loss_ema": 0.21364526450634003, "loss_mean_50": 0.21364526450634003, "loss_std_50": 0.0, "lr": 0.00025, "grad_norm_before_clip": 0.0186767578125, "grad_clipped": false, "step_seconds": 10.10697298499872, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 2, "loss": 0.10056868940591812, "loss_ema": 0.20799143575131893, "loss_mean_50": 0.15710697695612907, "loss_std_50": 0.05653828755021095, "lr": 0.00025, "grad_norm_before_clip": 0.021484375, "grad_clipped": false, "step_seconds": 10.031426426001417, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 3, "loss": 0.17827622592449188, "loss_ema": 0.20650567525997757, "loss_mean_50": 0.16416339327891669, "loss_std_50": 0.047229630481099814, "lr": 0.00025, "grad_norm_before_clip": 0.0079345703125, "grad_clipped": false, "step_seconds": 2.111574391998147, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 4, "loss": 0.0857408195734024, "loss_ema": 0.2004674324756488, "loss_mean_50": 0.1445577498525381, "loss_std_50": 0.05316128534020625, "lr": 0.00025, "grad_norm_before_clip": 0.021728515625, "grad_clipped": false, "step_seconds": 0.4253470510011539, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 5, "loss": 0.06948555260896683, "loss_ema": 0.19391833848231468, "loss_mean_50": 0.12954331040382386, "loss_std_50": 0.056237277450590456, "lr": 0.00025, "grad_norm_before_clip": 0.0189208984375, "grad_clipped": false, "step_seconds": 0.601615749001212, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 6, "loss": 0.08113450556993484, "loss_ema": 0.18827914683669567, "loss_mean_50": 0.12147517626484235, "loss_std_50": 0.05441507225302473, "lr": 0.00025, "grad_norm_before_clip": 0.01409912109375, "grad_clipped": false, "step_seconds": 0.6685794279983384, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 7, "loss": 0.11239732801914215, "loss_ema": 0.18448505589581798, "loss_mean_50": 0.12017834080117089, "loss_std_50": 0.05047861694823225, "lr": 0.00025, "grad_norm_before_clip": 0.05517578125, "grad_clipped": false, "step_seconds": 0.35718776599969715, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 8, "loss": 0.1190834492444992, "loss_ema": 0.18121497556325203, "loss_mean_50": 0.12004147935658693, "loss_std_50": 0.047219810887652446, "lr": 0.00025, "grad_norm_before_clip": 0.0478515625, "grad_clipped": false, "step_seconds": 1.892097593001381, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 9, "loss": 0.5245077013969421, "loss_ema": 0.19837961185493655, "loss_mean_50": 0.16498217069440418, "loss_std_50": 0.1346821845235972, "lr": 0.00025, "grad_norm_before_clip": 0.0244140625, "grad_clipped": false, "step_seconds": 0.5758501069976774, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 10, "loss": 0.18972772359848022, "loss_ema": 0.1979470174421137, "loss_mean_50": 0.1674567259848118, "loss_std_50": 0.1279862201965939, "lr": 0.00025, "grad_norm_before_clip": 0.1123046875, "grad_clipped": false, "step_seconds": 1.8357110030010517, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 11, "loss": 0.5074498057365417, "loss_ema": 0.2134221568568351, "loss_mean_50": 0.19836518778042359, "loss_std_50": 0.15634790947374025, "lr": 0.00025, "grad_norm_before_clip": 0.080078125, "grad_clipped": false, "step_seconds": 2.0685308360007184, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 12, "loss": 0.09319855272769928, "loss_ema": 0.2074109766503783, "loss_mean_50": 0.18960130152602991, "loss_std_50": 0.1524876250350202, "lr": 0.00025, "grad_norm_before_clip": 0.017578125, "grad_clipped": false, "step_seconds": 2.060223196000152, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 13, "loss": 0.08769605308771133, "loss_ema": 0.20142523047224495, "loss_mean_50": 0.18176243626154387, "loss_std_50": 0.1490006627793402, "lr": 0.00025, "grad_norm_before_clip": 0.01470947265625, "grad_clipped": false, "step_seconds": 0.4812139200003003, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 14, "loss": 0.06312495470046997, "loss_ema": 0.19451021668365617, "loss_mean_50": 0.17328833043575287, "loss_std_50": 0.14679555099096517, "lr": 0.00025, "grad_norm_before_clip": 0.007293701171875, "grad_clipped": false, "step_seconds": 1.2690454509975098, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 15, "loss": 0.4654071033000946, "loss_ema": 0.20805506101447807, "loss_mean_50": 0.19276291529337566, "loss_std_50": 0.15944268744588497, "lr": 0.00025, "grad_norm_before_clip": 0.01397705078125, "grad_clipped": false, "step_seconds": 2.0587862259999383, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 16, "loss": 0.07444032281637192, "loss_ema": 0.20137432410457276, "loss_mean_50": 0.18536775326356292, "loss_std_50": 0.15701408771325595, "lr": 0.00025, "grad_norm_before_clip": 0.0081787109375, "grad_clipped": false, "step_seconds": 1.8991433980008878, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 17, "loss": 0.07906686514616013, "loss_ema": 0.19525895115665212, "loss_mean_50": 0.17911475984489217, "loss_std_50": 0.15436586789518408, "lr": 0.00025, "grad_norm_before_clip": 0.00982666015625, "grad_clipped": false, "step_seconds": 2.0679554490016017, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 18, "loss": 0.050631001591682434, "loss_ema": 0.18802755367840362, "loss_mean_50": 0.1719767732752694, "loss_std_50": 0.1528762972510802, "lr": 0.00025, "grad_norm_before_clip": 0.0125732421875, "grad_clipped": false, "step_seconds": 2.06616962499902, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 19, "loss": 0.22272807359695435, "loss_ema": 0.18976257967433116, "loss_mean_50": 0.17464789434483177, "loss_std_50": 0.149229786133289, "lr": 0.00025, "grad_norm_before_clip": 0.033935546875, "grad_clipped": false, "step_seconds": 1.8301164270014851, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 20, "loss": 0.08332343399524689, "loss_ema": 0.18444062239037695, "loss_mean_50": 0.17008167132735252, "loss_std_50": 0.14680670991538364, "lr": 0.00025, "grad_norm_before_clip": 0.028076171875, "grad_clipped": false, "step_seconds": 1.8923190649984463, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 21, "loss": 0.23371417820453644, "loss_ema": 0.18690430018108492, "loss_mean_50": 0.17311179070245653, "loss_std_50": 0.14390811939155448, "lr": 0.00025, "grad_norm_before_clip": 0.018798828125, "grad_clipped": false, "step_seconds": 0.6146930799986876, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 22, "loss": 0.05716972053050995, "loss_ema": 0.18041757119855617, "loss_mean_50": 0.1678416966037317, "loss_std_50": 0.14265852767090073, "lr": 0.00025, "grad_norm_before_clip": 0.0133056640625, "grad_clipped": false, "step_seconds": 1.5904794050002238, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 23, "loss": 0.10259830206632614, "loss_ema": 0.17652660774194467, "loss_mean_50": 0.1650050272760184, "loss_std_50": 0.14015575931234892, "lr": 0.00025, "grad_norm_before_clip": 0.01361083984375, "grad_clipped": false, "step_seconds": 2.0638304839994817, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 24, "loss": 0.05102641135454178, "loss_ema": 0.17025159792257452, "loss_mean_50": 0.1602559182792902, "loss_std_50": 0.1390823311314131, "lr": 0.00025, "grad_norm_before_clip": 0.0211181640625, "grad_clipped": false, "step_seconds": 2.07047174300169, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 25, "loss": 0.15586231648921967, "loss_ema": 0.16953213385090676, "loss_mean_50": 0.1600801742076874, "loss_std_50": 0.1362750171642587, "lr": 0.00025, "grad_norm_before_clip": 0.10302734375, "grad_clipped": false, "step_seconds": 2.061736870000459, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 26, "loss": 0.07449161261320114, "loss_ema": 0.16478010778902147, "loss_mean_50": 0.15678830645405328, "loss_std_50": 0.13463849935572428, "lr": 0.00025, "grad_norm_before_clip": 0.023681640625, "grad_clipped": false, "step_seconds": 0.5741265889992064, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 27, "loss": 0.2080274224281311, "loss_ema": 0.16694247352097694, "loss_mean_50": 0.15868605149013024, "loss_std_50": 0.13247555629954666, "lr": 0.00025, "grad_norm_before_clip": 0.00958251953125, "grad_clipped": false, "step_seconds": 1.8939291700007743, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 28, "loss": 0.039954639971256256, "loss_ema": 0.1605930818434909, "loss_mean_50": 0.15444564393588475, "loss_std_50": 0.13194121390692753, "lr": 0.00025, "grad_norm_before_clip": 0.00848388671875, "grad_clipped": false, "step_seconds": 2.0704005080006027, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 29, "loss": 0.06404057890176773, "loss_ema": 0.15576545669640474, "loss_mean_50": 0.15132822790022554, "loss_std_50": 0.13069163437160558, "lr": 0.00025, "grad_norm_before_clip": 0.0098876953125, "grad_clipped": false, "step_seconds": 2.0657387229985034, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 30, "loss": 0.04639781266450882, "loss_ema": 0.15029707449480992, "loss_mean_50": 0.1478305473923683, "loss_std_50": 0.12986816065025464, "lr": 0.00025, "grad_norm_before_clip": 0.031494140625, "grad_clipped": false, "step_seconds": 2.003965325002355, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 31, "loss": 0.2954860031604767, "loss_ema": 0.15755652092809325, "loss_mean_50": 0.1525936266106944, "loss_std_50": 0.13039283246500546, "lr": 0.00025, "grad_norm_before_clip": 0.01544189453125, "grad_clipped": false, "step_seconds": 1.1039269139982935, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 32, "loss": 0.28982383012771606, "loss_ema": 0.1641698863880744, "loss_mean_50": 0.15688207047060132, "loss_std_50": 0.13054149702289516, "lr": 0.00025, "grad_norm_before_clip": 0.0166015625, "grad_clipped": false, "step_seconds": 1.8943754909996642, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 33, "loss": 0.9423972964286804, "loss_ema": 0.20308125689010467, "loss_mean_50": 0.1806855621663007, "loss_std_50": 0.18616144860921996, "lr": 0.00025, "grad_norm_before_clip": 0.03369140625, "grad_clipped": false, "step_seconds": 2.054299928000546, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 34, "loss": 0.09143199771642685, "loss_ema": 0.19749879393142078, "loss_mean_50": 0.17806045732953968, "loss_std_50": 0.18402227343582298, "lr": 0.00025, "grad_norm_before_clip": 0.020751953125, "grad_clipped": false, "step_seconds": 0.3634469270000409, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 35, "loss": 0.05243552848696709, "loss_ema": 0.19024563065919808, "loss_mean_50": 0.17447117364832332, "loss_std_50": 0.18257784416800507, "lr": 0.00025, "grad_norm_before_clip": 0.0218505859375, "grad_clipped": false, "step_seconds": 1.5702532069990411, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 36, "loss": 0.10789015889167786, "loss_ema": 0.18612785707082208, "loss_mean_50": 0.17262170101619428, "loss_std_50": 0.18035638431678702, "lr": 0.00025, "grad_norm_before_clip": 0.0693359375, "grad_clipped": false, "step_seconds": 1.8963039320005919, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 37, "loss": 0.16299864649772644, "loss_ema": 0.18497139654216727, "loss_mean_50": 0.1723616184616411, "loss_std_50": 0.1779092855306784, "lr": 0.00025, "grad_norm_before_clip": 0.009765625, "grad_clipped": false, "step_seconds": 1.890168457000982, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 38, "loss": 0.06831441819667816, "loss_ema": 0.1791385476248928, "loss_mean_50": 0.1696235342441421, "loss_std_50": 0.1763410528801478, "lr": 0.00025, "grad_norm_before_clip": 0.01251220703125, "grad_clipped": false, "step_seconds": 1.8873140849973424, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 39, "loss": 0.09817425906658173, "loss_ema": 0.17509033319697725, "loss_mean_50": 0.16779150154728156, "loss_std_50": 0.1744315636052051, "lr": 0.00025, "grad_norm_before_clip": 0.061279296875, "grad_clipped": false, "step_seconds": 0.5220913490011299, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 40, "loss": 0.1394553929567337, "loss_ema": 0.17330858618496509, "loss_mean_50": 0.16708309883251787, "loss_std_50": 0.17229417478271836, "lr": 0.00025, "grad_norm_before_clip": 0.010498046875, "grad_clipped": false, "step_seconds": 1.2883531469997251, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 41, "loss": 0.39068299531936646, "loss_ema": 0.18417730664168513, "loss_mean_50": 0.1725367548443922, "loss_std_50": 0.17364027739957918, "lr": 0.00025, "grad_norm_before_clip": 0.016357421875, "grad_clipped": false, "step_seconds": 0.6029024920026131, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 42, "loss": 0.14437268674373627, "loss_ema": 0.18218707564678768, "loss_mean_50": 0.1718661817943766, "loss_std_50": 0.17161440110639278, "lr": 0.00025, "grad_norm_before_clip": 0.00726318359375, "grad_clipped": false, "step_seconds": 2.0666895949980244, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 43, "loss": 0.06395913660526276, "loss_ema": 0.17627567869471142, "loss_mean_50": 0.1693567156271879, "loss_std_50": 0.17038508050615261, "lr": 0.00025, "grad_norm_before_clip": 0.018798828125, "grad_clipped": false, "step_seconds": 2.0606633169991255, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 44, "loss": 0.12407145649194717, "loss_ema": 0.17366546758457319, "loss_mean_50": 0.16832750519229608, "loss_std_50": 0.16857291381520162, "lr": 0.00025, "grad_norm_before_clip": 0.00616455078125, "grad_clipped": false, "step_seconds": 1.9057077870020294, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 45, "loss": 0.05337880551815033, "loss_ema": 0.16765113448125205, "loss_mean_50": 0.1657730896439817, "loss_std_50": 0.16754833311187528, "lr": 0.00025, "grad_norm_before_clip": 0.169921875, "grad_clipped": false, "step_seconds": 1.5743348769974546, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 46, "loss": 0.05486397445201874, "loss_ema": 0.16201177647979037, "loss_mean_50": 0.1633620219224173, "loss_std_50": 0.16650456314615014, "lr": 0.00025, "grad_norm_before_clip": 0.0230712890625, "grad_clipped": false, "step_seconds": 1.887843659002101, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 47, "loss": 0.05400737375020981, "loss_ema": 0.15661155634331134, "loss_mean_50": 0.16103532728045544, "loss_std_50": 0.1654778640099858, "lr": 0.00025, "grad_norm_before_clip": 0.037353515625, "grad_clipped": false, "step_seconds": 0.6030308160006825, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 48, "loss": 0.05440378561615944, "loss_ema": 0.15150116780695375, "loss_mean_50": 0.15881383682911596, "loss_std_50": 0.16445179183170033, "lr": 0.00025, "grad_norm_before_clip": 0.007720947265625, "grad_clipped": false, "step_seconds": 2.0644823319998977, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 49, "loss": 0.09789769351482391, "loss_ema": 0.14882099409234725, "loss_mean_50": 0.15757065023086508, "loss_std_50": 0.16299279153603966, "lr": 0.00025, "grad_norm_before_clip": 0.022705078125, "grad_clipped": false, "step_seconds": 1.8939477899984922, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 50, "loss": 0.05093717575073242, "loss_ema": 0.14392680317526652, "loss_mean_50": 0.15543798074126244, "loss_std_50": 0.16204376806683704, "lr": 0.00025, "grad_norm_before_clip": 0.0240478515625, "grad_clipped": false, "step_seconds": 1.2938321949986857, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 51, "loss": 0.1337166428565979, "loss_ema": 0.14341629515933307, "loss_mean_50": 0.1538394083082676, "loss_std_50": 0.16185580591439772, "lr": 0.00025, "grad_norm_before_clip": 0.01123046875, "grad_clipped": false, "step_seconds": 1.1064193660022283, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 52, "loss": 0.4767708480358124, "loss_ema": 0.16008402280315703, "loss_mean_50": 0.16136345148086548, "loss_std_50": 0.16783810529191126, "lr": 0.00025, "grad_norm_before_clip": 0.05810546875, "grad_clipped": false, "step_seconds": 2.065631632998702, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 53, "loss": 0.07730188220739365, "loss_ema": 0.15594491577336886, "loss_mean_50": 0.1593439646065235, "loss_std_50": 0.16822947830427878, "lr": 0.00025, "grad_norm_before_clip": 0.0062255859375, "grad_clipped": false, "step_seconds": 2.0573393020022195, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 54, "loss": 0.07954277098178864, "loss_ema": 0.15212480853378985, "loss_mean_50": 0.15922000363469124, "loss_std_50": 0.1682859416443055, "lr": 0.00025, "grad_norm_before_clip": 0.015380859375, "grad_clipped": false, "step_seconds": 0.4221210299983795, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 55, "loss": 0.051134757697582245, "loss_ema": 0.14707530599197946, "loss_mean_50": 0.15885298773646356, "loss_std_50": 0.16850111699395046, "lr": 0.00025, "grad_norm_before_clip": 0.015380859375, "grad_clipped": false, "step_seconds": 1.2657016070006648, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 56, "loss": 0.13579601049423218, "loss_ema": 0.1465113412170921, "loss_mean_50": 0.1599462178349495, "loss_std_50": 0.16817033183872337, "lr": 0.00025, "grad_norm_before_clip": 0.014404296875, "grad_clipped": false, "step_seconds": 0.6673578879999695, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 57, "loss": 0.06432237476110458, "loss_ema": 0.14240189289429273, "loss_mean_50": 0.15898471876978873, "loss_std_50": 0.1685763817357697, "lr": 0.00025, "grad_norm_before_clip": 0.045654296875, "grad_clipped": false, "step_seconds": 0.5230422979984723, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 58, "loss": 0.0871826782822609, "loss_ema": 0.13964093216369114, "loss_mean_50": 0.158346703350544, "loss_std_50": 0.1687864267339575, "lr": 0.00025, "grad_norm_before_clip": 0.061279296875, "grad_clipped": false, "step_seconds": 1.8922897110023769, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 59, "loss": 0.7411826252937317, "loss_ema": 0.16971801682019316, "loss_mean_50": 0.16268020182847975, "loss_std_50": 0.1805063858950701, "lr": 0.00025, "grad_norm_before_clip": 0.037841796875, "grad_clipped": false, "step_seconds": 2.072668510998483, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 60, "loss": 0.3161301910877228, "loss_ema": 0.17703862553356964, "loss_mean_50": 0.16520825117826463, "loss_std_50": 0.18174837256707493, "lr": 0.00025, "grad_norm_before_clip": 0.0291748046875, "grad_clipped": false, "step_seconds": 2.063520443000016, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 61, "loss": 0.08561673015356064, "loss_ema": 0.17246753076456917, "loss_mean_50": 0.15677158966660498, "loss_std_50": 0.17534367466312878, "lr": 0.00025, "grad_norm_before_clip": 0.11767578125, "grad_clipped": false, "step_seconds": 0.6167511880012171, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 62, "loss": 0.2868773639202118, "loss_ema": 0.17818802242235127, "loss_mean_50": 0.16064516589045524, "loss_std_50": 0.1760344264085212, "lr": 0.00025, "grad_norm_before_clip": 0.020263671875, "grad_clipped": false, "step_seconds": 2.0620734519980033, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 63, "loss": 0.05516883730888367, "loss_ema": 0.17203706316667788, "loss_mean_50": 0.1599946215748787, "loss_std_50": 0.176362608649672, "lr": 0.00025, "grad_norm_before_clip": 0.01324462890625, "grad_clipped": false, "step_seconds": 1.8303461400028027, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 64, "loss": 0.1268303394317627, "loss_ema": 0.16977672697993212, "loss_mean_50": 0.16126872926950456, "loss_std_50": 0.17588766052311353, "lr": 0.00025, "grad_norm_before_clip": 0.02783203125, "grad_clipped": false, "step_seconds": 1.8294348629970045, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 65, "loss": 0.10386589169502258, "loss_ema": 0.16648118521568664, "loss_mean_50": 0.1540379050374031, "loss_std_50": 0.17058746439668745, "lr": 0.00025, "grad_norm_before_clip": 0.023681640625, "grad_clipped": false, "step_seconds": 0.7270793310017325, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 66, "loss": 0.05625113844871521, "loss_ema": 0.16096968287733807, "loss_mean_50": 0.15367412135004999, "loss_std_50": 0.1707761113364766, "lr": 0.00025, "grad_norm_before_clip": 0.01385498046875, "grad_clipped": false, "step_seconds": 2.0515820410000742, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 67, "loss": 0.1435173600912094, "loss_ema": 0.16009706673803162, "loss_mean_50": 0.15496313124895095, "loss_std_50": 0.17045103988861177, "lr": 0.00025, "grad_norm_before_clip": 0.0216064453125, "grad_clipped": false, "step_seconds": 0.3632198369996331, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 68, "loss": 0.05882783234119415, "loss_ema": 0.15503360501818972, "loss_mean_50": 0.1551270678639412, "loss_std_50": 0.170354530823928, "lr": 0.00025, "grad_norm_before_clip": 0.0152587890625, "grad_clipped": false, "step_seconds": 0.5718913489981787, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 69, "loss": 0.3125799596309662, "loss_ema": 0.16291092274882854, "loss_mean_50": 0.15692410558462144, "loss_std_50": 0.17152803610682646, "lr": 0.00025, "grad_norm_before_clip": 0.040771484375, "grad_clipped": false, "step_seconds": 0.6051288070011651, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 70, "loss": 0.420383095741272, "loss_ema": 0.1757845313984507, "loss_mean_50": 0.16366529881954192, "loss_std_50": 0.1750893917322044, "lr": 0.00025, "grad_norm_before_clip": 0.0322265625, "grad_clipped": false, "step_seconds": 2.068572281001252, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 71, "loss": 0.3568682372570038, "loss_ema": 0.18483871669137836, "loss_mean_50": 0.16612838000059127, "loss_std_50": 0.17691421346687894, "lr": 0.00025, "grad_norm_before_clip": 0.0108642578125, "grad_clipped": false, "step_seconds": 1.8963267039980565, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 72, "loss": 0.11224092543125153, "loss_ema": 0.18120882712837202, "loss_mean_50": 0.1672298040986061, "loss_std_50": 0.176403126835353, "lr": 0.00025, "grad_norm_before_clip": 0.00677490234375, "grad_clipped": false, "step_seconds": 2.061690491002082, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 73, "loss": 0.08280862867832184, "loss_ema": 0.1762888172058695, "loss_mean_50": 0.16683401063084602, "loss_std_50": 0.17656981786508244, "lr": 0.00025, "grad_norm_before_clip": 0.05810546875, "grad_clipped": false, "step_seconds": 0.47835121599928243, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 74, "loss": 0.172180637717247, "loss_ema": 0.17608340823143837, "loss_mean_50": 0.16925709515810014, "loss_std_50": 0.17579355212778247, "lr": 0.00025, "grad_norm_before_clip": 0.0068359375, "grad_clipped": false, "step_seconds": 1.5852467049990082, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 75, "loss": 1.055152177810669, "loss_ema": 0.2200368467103999, "loss_mean_50": 0.18724289238452912, "loss_std_50": 0.21511043120983564, "lr": 0.00025, "grad_norm_before_clip": 0.419921875, "grad_clipped": false, "step_seconds": 0.5745764999992389, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 76, "loss": 0.06651657074689865, "loss_ema": 0.21236083291222482, "loss_mean_50": 0.18708339154720308, "loss_std_50": 0.21519691458681725, "lr": 0.00025, "grad_norm_before_clip": 0.0177001953125, "grad_clipped": false, "step_seconds": 1.8890766880031151, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 77, "loss": 0.10814609378576279, "loss_ema": 0.2071500959559017, "loss_mean_50": 0.1850857649743557, "loss_std_50": 0.21545665556377716, "lr": 0.00025, "grad_norm_before_clip": 0.028076171875, "grad_clipped": false, "step_seconds": 0.3509441109999898, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 78, "loss": 0.08052494376897812, "loss_ema": 0.2008188383465555, "loss_mean_50": 0.18589717105031014, "loss_std_50": 0.2149844424246498, "lr": 0.00025, "grad_norm_before_clip": 0.0257568359375, "grad_clipped": false, "step_seconds": 1.8927308509992145, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 79, "loss": 0.055518534034490585, "loss_ema": 0.19355382313095224, "loss_mean_50": 0.1857267301529646, "loss_std_50": 0.21508433841946514, "lr": 0.00025, "grad_norm_before_clip": 0.0250244140625, "grad_clipped": false, "step_seconds": 2.066348725998978, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 80, "loss": 0.24701449275016785, "loss_ema": 0.196226856611913, "loss_mean_50": 0.18973906375467778, "loss_std_50": 0.2143176303161289, "lr": 0.00025, "grad_norm_before_clip": 0.01904296875, "grad_clipped": false, "step_seconds": 2.066611815000215, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 81, "loss": 0.12454988807439804, "loss_ema": 0.19264300818503724, "loss_mean_50": 0.1863203414529562, "loss_std_50": 0.21396659381966202, "lr": 0.00025, "grad_norm_before_clip": 0.04345703125, "grad_clipped": false, "step_seconds": 2.070608033998724, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 82, "loss": 0.15681979060173035, "loss_ema": 0.1908518473058719, "loss_mean_50": 0.1836602606624365, "loss_std_50": 0.21348951601338584, "lr": 0.00025, "grad_norm_before_clip": 0.007080078125, "grad_clipped": false, "step_seconds": 2.06779503800135, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 83, "loss": 0.06142241507768631, "loss_ema": 0.1843803756944626, "loss_mean_50": 0.1660407630354166, "loss_std_50": 0.18453327824541096, "lr": 0.00025, "grad_norm_before_clip": 0.00933837890625, "grad_clipped": false, "step_seconds": 1.2971172850011499, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 84, "loss": 0.5309082269668579, "loss_ema": 0.20170676825808237, "loss_mean_50": 0.1748302876204252, "loss_std_50": 0.19111910162261264, "lr": 0.00025, "grad_norm_before_clip": 0.05517578125, "grad_clipped": false, "step_seconds": 1.8333222740002384, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 85, "loss": 0.2983182370662689, "loss_ema": 0.2065373416984917, "loss_mean_50": 0.17974794179201126, "loss_std_50": 0.19106989196203378, "lr": 0.00025, "grad_norm_before_clip": 0.01214599609375, "grad_clipped": false, "step_seconds": 2.068814788999589, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 86, "loss": 0.46302124857902527, "loss_ema": 0.21936153704251837, "loss_mean_50": 0.18685056358575822, "loss_std_50": 0.19483033871944128, "lr": 0.00025, "grad_norm_before_clip": 0.03857421875, "grad_clipped": false, "step_seconds": 2.0615451849989768, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 87, "loss": 0.06657804548740387, "loss_ema": 0.21172236246476264, "loss_mean_50": 0.18492215156555175, "loss_std_50": 0.19553279368390214, "lr": 0.00025, "grad_norm_before_clip": 0.064453125, "grad_clipped": false, "step_seconds": 1.9050075199993444, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 88, "loss": 0.08899189531803131, "loss_ema": 0.20558583910742606, "loss_mean_50": 0.1853357011079788, "loss_std_50": 0.19530746886448855, "lr": 0.00025, "grad_norm_before_clip": 0.0137939453125, "grad_clipped": false, "step_seconds": 1.8963149359988165, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 89, "loss": 0.17458075284957886, "loss_ema": 0.2040355847945337, "loss_mean_50": 0.18686383098363876, "loss_std_50": 0.19491804239731728, "lr": 0.00025, "grad_norm_before_clip": 0.013916015625, "grad_clipped": false, "step_seconds": 1.889559070001269, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 90, "loss": 0.13865405321121216, "loss_ema": 0.20076650821536762, "loss_mean_50": 0.18684780418872834, "loss_std_50": 0.1949219727190307, "lr": 0.00025, "grad_norm_before_clip": 0.021484375, "grad_clipped": false, "step_seconds": 0.476793980000366, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 91, "loss": 0.19437916576862335, "loss_ema": 0.2004471410930304, "loss_mean_50": 0.18292172759771347, "loss_std_50": 0.19274158891261559, "lr": 0.00025, "grad_norm_before_clip": 0.01434326171875, "grad_clipped": false, "step_seconds": 0.5717134320002515, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 92, "loss": 0.16461345553398132, "loss_ema": 0.19865545681507793, "loss_mean_50": 0.18332654297351836, "loss_std_50": 0.1926814456728389, "lr": 0.00025, "grad_norm_before_clip": 0.07421875, "grad_clipped": false, "step_seconds": 0.600009110999963, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 93, "loss": 0.5372874736785889, "loss_ema": 0.21558705765825345, "loss_mean_50": 0.1927931097149849, "loss_std_50": 0.19813459787718762, "lr": 0.00025, "grad_norm_before_clip": 0.0238037109375, "grad_clipped": false, "step_seconds": 0.5752489750011591, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 94, "loss": 0.15562692284584045, "loss_ema": 0.2125890509176328, "loss_mean_50": 0.19342421904206275, "loss_std_50": 0.19796488015167188, "lr": 0.00025, "grad_norm_before_clip": 0.008544921875, "grad_clipped": false, "step_seconds": 2.068150650997268, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 95, "loss": 0.13786131143569946, "loss_ema": 0.20885266394353613, "loss_mean_50": 0.19511386916041373, "loss_std_50": 0.19712110296210084, "lr": 0.00025, "grad_norm_before_clip": 0.00604248046875, "grad_clipped": false, "step_seconds": 2.070505175001017, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 96, "loss": 0.4317935109138489, "loss_ema": 0.21999970629205176, "loss_mean_50": 0.20265245988965033, "loss_std_50": 0.19881358927374307, "lr": 0.00025, "grad_norm_before_clip": 0.031005859375, "grad_clipped": false, "step_seconds": 1.2966852210010984, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 97, "loss": 0.05225798487663269, "loss_ema": 0.21161262022128077, "loss_mean_50": 0.20261747211217881, "loss_std_50": 0.1988398973681711, "lr": 0.00025, "grad_norm_before_clip": 0.01483154296875, "grad_clipped": false, "step_seconds": 1.593328033999569, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 98, "loss": 0.3329669237136841, "loss_ema": 0.21768033539590093, "loss_mean_50": 0.20818873487412928, "loss_std_50": 0.19851130816652127, "lr": 0.00025, "grad_norm_before_clip": 0.01519775390625, "grad_clipped": false, "step_seconds": 1.89063914000144, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 99, "loss": 0.07770195603370667, "loss_ema": 0.2106814164277912, "loss_mean_50": 0.20778482012450694, "loss_std_50": 0.1987557044107645, "lr": 0.00025, "grad_norm_before_clip": 0.0179443359375, "grad_clipped": false, "step_seconds": 0.6722752689966001, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 100, "loss": 0.2521173059940338, "loss_ema": 0.21275321090610333, "loss_mean_50": 0.21180842272937297, "loss_std_50": 0.1975725800131504, "lr": 0.00025, "grad_norm_before_clip": 0.0177001953125, "grad_clipped": false, "step_seconds": 1.833373745001154, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 101, "loss": 0.13959015905857086, "loss_ema": 0.20909505831372668, "loss_mean_50": 0.21192589305341245, "loss_std_50": 0.19752785526161679, "lr": 0.00025, "grad_norm_before_clip": 0.01025390625, "grad_clipped": false, "step_seconds": 1.890494306000619, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 102, "loss": 0.06942665576934814, "loss_ema": 0.20211163818650774, "loss_mean_50": 0.20377900920808314, "loss_std_50": 0.1948182371771266, "lr": 0.00025, "grad_norm_before_clip": 0.0311279296875, "grad_clipped": false, "step_seconds": 0.7793569440000283, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 103, "loss": 0.6183346509933472, "loss_ema": 0.2229227888268497, "loss_mean_50": 0.21459966458380222, "loss_std_50": 0.20237158220381624, "lr": 0.00025, "grad_norm_before_clip": 0.0150146484375, "grad_clipped": false, "step_seconds": 2.063055352999072, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 104, "loss": 0.08742055296897888, "loss_ema": 0.21614767703395615, "loss_mean_50": 0.21475722022354604, "loss_std_50": 0.20226941365003887, "lr": 0.00025, "grad_norm_before_clip": 0.03857421875, "grad_clipped": false, "step_seconds": 0.7274722270012717, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 105, "loss": 0.17863425612449646, "loss_ema": 0.21427200598848314, "loss_mean_50": 0.21730721019208432, "loss_std_50": 0.20099020974996695, "lr": 0.00025, "grad_norm_before_clip": 0.04345703125, "grad_clipped": false, "step_seconds": 2.060281928999757, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 106, "loss": 0.09042968600988388, "loss_ema": 0.20807988998955318, "loss_mean_50": 0.21639988370239735, "loss_std_50": 0.20145798019265768, "lr": 0.00025, "grad_norm_before_clip": 0.006744384765625, "grad_clipped": false, "step_seconds": 1.102359971999249, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 107, "loss": 0.2045910507440567, "loss_ema": 0.20790544802727834, "loss_mean_50": 0.2192052572220564, "loss_std_50": 0.20029399706162296, "lr": 0.00025, "grad_norm_before_clip": 0.0634765625, "grad_clipped": false, "step_seconds": 0.4241579980007373, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 108, "loss": 0.34046733379364014, "loss_ema": 0.21453354231559643, "loss_mean_50": 0.22427095033228397, "loss_std_50": 0.20009376454433853, "lr": 0.00025, "grad_norm_before_clip": 0.029541015625, "grad_clipped": false, "step_seconds": 1.8897385339987522, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 109, "loss": 0.05774255841970444, "loss_ema": 0.20669399312080183, "loss_mean_50": 0.21060214899480342, "loss_std_50": 0.18724678633170844, "lr": 0.00025, "grad_norm_before_clip": 0.0106201171875, "grad_clipped": false, "step_seconds": 0.5988535890028288, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 110, "loss": 0.04555026814341545, "loss_ema": 0.1986368068719325, "loss_mean_50": 0.20519055053591728, "loss_std_50": 0.18802710631120337, "lr": 0.00025, "grad_norm_before_clip": 0.0201416015625, "grad_clipped": false, "step_seconds": 1.2657827059992997, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 111, "loss": 0.11464548856019974, "loss_ema": 0.19443724095634587, "loss_mean_50": 0.20577112570405007, "loss_std_50": 0.1877015339169763, "lr": 0.00025, "grad_norm_before_clip": 0.0419921875, "grad_clipped": false, "step_seconds": 0.35483165999903576, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 112, "loss": 0.051620159298181534, "loss_ema": 0.18729638687343764, "loss_mean_50": 0.20106598161160946, "loss_std_50": 0.18855612804992816, "lr": 0.00025, "grad_norm_before_clip": 0.039306640625, "grad_clipped": false, "step_seconds": 2.0698271760011266, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 113, "loss": 0.05502936244010925, "loss_ema": 0.1806830356517712, "loss_mean_50": 0.20106319211423396, "loss_std_50": 0.18855828744944264, "lr": 0.00025, "grad_norm_before_clip": 0.009521484375, "grad_clipped": false, "step_seconds": 2.062761901997874, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 114, "loss": 0.23164789378643036, "loss_ema": 0.18323127855850416, "loss_mean_50": 0.20315954320132731, "loss_std_50": 0.188303826910571, "lr": 0.00025, "grad_norm_before_clip": 0.00860595703125, "grad_clipped": false, "step_seconds": 1.897860164001031, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 115, "loss": 0.17967715859413147, "loss_ema": 0.1830535725602855, "loss_mean_50": 0.2046757685393095, "loss_std_50": 0.18780275864065055, "lr": 0.00025, "grad_norm_before_clip": 0.01116943359375, "grad_clipped": false, "step_seconds": 2.067205217001174, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 116, "loss": 0.5338679552078247, "loss_ema": 0.20059429169266246, "loss_mean_50": 0.2142281048744917, "loss_std_50": 0.19210773415586205, "lr": 0.00025, "grad_norm_before_clip": 0.0191650390625, "grad_clipped": false, "step_seconds": 1.5748346729997138, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 117, "loss": 0.06362693756818771, "loss_ema": 0.1937459239864387, "loss_mean_50": 0.21263029642403125, "loss_std_50": 0.19301928031670648, "lr": 0.00025, "grad_norm_before_clip": 0.0159912109375, "grad_clipped": false, "step_seconds": 2.06744722899748, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 118, "loss": 0.12416733056306839, "loss_ema": 0.1902669943152702, "loss_mean_50": 0.21393708638846876, "loss_std_50": 0.19219298853336836, "lr": 0.00025, "grad_norm_before_clip": 0.0093994140625, "grad_clipped": false, "step_seconds": 1.833203730002424, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 119, "loss": 0.13287734985351562, "loss_ema": 0.18739751209218247, "loss_mean_50": 0.21034303419291972, "loss_std_50": 0.19199487732518997, "lr": 0.00025, "grad_norm_before_clip": 0.006439208984375, "grad_clipped": false, "step_seconds": 2.067237164999824, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 120, "loss": 0.07676942646503448, "loss_ema": 0.18186610781082505, "loss_mean_50": 0.20347076080739498, "loss_std_50": 0.19049752334913306, "lr": 0.00025, "grad_norm_before_clip": 0.040771484375, "grad_clipped": false, "step_seconds": 1.8332056449980882, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 121, "loss": 0.12261141836643219, "loss_ema": 0.1789033733386054, "loss_mean_50": 0.19878562442958356, "loss_std_50": 0.18954552183803325, "lr": 0.00025, "grad_norm_before_clip": 0.005584716796875, "grad_clipped": false, "step_seconds": 2.0669986130014877, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 122, "loss": 0.06117578595876694, "loss_ema": 0.17301699396961348, "loss_mean_50": 0.19776432164013386, "loss_std_50": 0.19014571108455128, "lr": 0.00025, "grad_norm_before_clip": 0.040771484375, "grad_clipped": false, "step_seconds": 1.891242196001258, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 123, "loss": 0.06558235734701157, "loss_ema": 0.16764526213848338, "loss_mean_50": 0.19741979621350766, "loss_std_50": 0.19036916229495054, "lr": 0.00025, "grad_norm_before_clip": 0.01190185546875, "grad_clipped": false, "step_seconds": 0.6166232189971197, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 124, "loss": 0.04784893989562988, "loss_ema": 0.16165544602634072, "loss_mean_50": 0.1949331622570753, "loss_std_50": 0.1914913134406081, "lr": 0.00025, "grad_norm_before_clip": 0.01043701171875, "grad_clipped": false, "step_seconds": 2.0693021990009584, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 125, "loss": 0.05248887464404106, "loss_ema": 0.15619711745722573, "loss_mean_50": 0.17487989619374275, "loss_std_50": 0.1478954429672335, "lr": 0.00025, "grad_norm_before_clip": 0.006195068359375, "grad_clipped": false, "step_seconds": 2.0666344599994773, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 126, "loss": 0.25309592485427856, "loss_ema": 0.16104205782707837, "loss_mean_50": 0.17861148327589035, "loss_std_50": 0.1474674202835091, "lr": 0.00025, "grad_norm_before_clip": 0.0302734375, "grad_clipped": false, "step_seconds": 0.3662347810022766, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 127, "loss": 0.07245126366615295, "loss_ema": 0.1566125181190321, "loss_mean_50": 0.17789758667349814, "loss_std_50": 0.1478926056939496, "lr": 0.00025, "grad_norm_before_clip": 0.007293701171875, "grad_clipped": false, "step_seconds": 1.2947001609973086, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 128, "loss": 0.05840076878666878, "loss_ema": 0.15170193065241394, "loss_mean_50": 0.17745510317385196, "loss_std_50": 0.14821601865276213, "lr": 0.00025, "grad_norm_before_clip": 0.017822265625, "grad_clipped": false, "step_seconds": 2.0697202970004582, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 129, "loss": 0.3639194965362549, "loss_ema": 0.16231280894660596, "loss_mean_50": 0.18362312242388726, "loss_std_50": 0.14942540809975305, "lr": 0.00025, "grad_norm_before_clip": 0.0169677734375, "grad_clipped": false, "step_seconds": 2.057395517000259, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 130, "loss": 0.20866678655147552, "loss_ema": 0.16463050782684943, "loss_mean_50": 0.1828561682999134, "loss_std_50": 0.14919630944755738, "lr": 0.00025, "grad_norm_before_clip": 0.01171875, "grad_clipped": false, "step_seconds": 0.6150187860002916, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 131, "loss": 0.1644660234451294, "loss_ema": 0.16462228360776343, "loss_mean_50": 0.18365449100732803, "loss_std_50": 0.14898883493707507, "lr": 0.00025, "grad_norm_before_clip": 0.00604248046875, "grad_clipped": false, "step_seconds": 2.0670689799990214, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 132, "loss": 0.35527902841567993, "loss_ema": 0.17415512084815923, "loss_mean_50": 0.18762367576360703, "loss_std_50": 0.15085296192707973, "lr": 0.00025, "grad_norm_before_clip": 0.0595703125, "grad_clipped": false, "step_seconds": 1.8906894510000711, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 133, "loss": 0.09479318559169769, "loss_ema": 0.17018702408533615, "loss_mean_50": 0.18829109117388726, "loss_std_50": 0.15036617143819256, "lr": 0.00025, "grad_norm_before_clip": 0.07861328125, "grad_clipped": false, "step_seconds": 2.0703081550018396, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 134, "loss": 0.122616708278656, "loss_ema": 0.16780850829500213, "loss_mean_50": 0.18012526080012323, "loss_std_50": 0.14241431523509368, "lr": 0.00025, "grad_norm_before_clip": 0.043701171875, "grad_clipped": false, "step_seconds": 2.0654479959994205, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 135, "loss": 0.36357980966567993, "loss_ema": 0.17759707336353603, "loss_mean_50": 0.18143049225211144, "loss_std_50": 0.14378405124336177, "lr": 0.00025, "grad_norm_before_clip": 0.0172119140625, "grad_clipped": false, "step_seconds": 1.2698774259988568, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 136, "loss": 0.0410926416516304, "loss_ema": 0.17077185177794074, "loss_mean_50": 0.17299192011356354, "loss_std_50": 0.13932218436941116, "lr": 0.00025, "grad_norm_before_clip": 0.01202392578125, "grad_clipped": false, "step_seconds": 1.829554301002645, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 137, "loss": 0.06625089049339294, "loss_ema": 0.16554580371371333, "loss_mean_50": 0.17298537701368333, "loss_std_50": 0.13932718940853034, "lr": 0.00025, "grad_norm_before_clip": 0.0115966796875, "grad_clipped": false, "step_seconds": 0.5241217599977972, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 138, "loss": 0.25968438386917114, "loss_ema": 0.1702527327214862, "loss_mean_50": 0.17639922678470613, "loss_std_50": 0.13931851131371156, "lr": 0.00025, "grad_norm_before_clip": 0.01190185546875, "grad_clipped": false, "step_seconds": 1.9921720179991098, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 139, "loss": 0.1462310254573822, "loss_ema": 0.169051647358281, "loss_mean_50": 0.17583223223686217, "loss_std_50": 0.1393824321151305, "lr": 0.00025, "grad_norm_before_clip": 0.006805419921875, "grad_clipped": false, "step_seconds": 1.5884640450021834, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 140, "loss": 0.33825114369392395, "loss_ema": 0.17751162217506314, "loss_mean_50": 0.1798241740465164, "loss_std_50": 0.1411080454792779, "lr": 0.00025, "grad_norm_before_clip": 0.01519775390625, "grad_clipped": false, "step_seconds": 1.1033548589985003, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 141, "loss": 0.07515409588813782, "loss_ema": 0.17239374586071687, "loss_mean_50": 0.1774396726489067, "loss_std_50": 0.14184736233578601, "lr": 0.00025, "grad_norm_before_clip": 0.00750732421875, "grad_clipped": false, "step_seconds": 2.067458264002198, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 142, "loss": 0.10637927055358887, "loss_ema": 0.16909302209536045, "loss_mean_50": 0.17627498894929886, "loss_std_50": 0.14218656438464577, "lr": 0.00025, "grad_norm_before_clip": 0.01019287109375, "grad_clipped": false, "step_seconds": 2.0584115759993438, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 143, "loss": 0.1598220020532608, "loss_ema": 0.16862947109325546, "loss_mean_50": 0.1687256795167923, "loss_std_50": 0.13250977606029296, "lr": 0.00025, "grad_norm_before_clip": 0.027587890625, "grad_clipped": false, "step_seconds": 0.35089204299947596, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 144, "loss": 0.08613498508930206, "loss_ema": 0.1645047467930578, "loss_mean_50": 0.16733584076166153, "loss_std_50": 0.1330033909619508, "lr": 0.00025, "grad_norm_before_clip": 0.007354736328125, "grad_clipped": false, "step_seconds": 1.8925468220004404, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 145, "loss": 0.11607778072357178, "loss_ema": 0.1620833984895835, "loss_mean_50": 0.16690017014741898, "loss_std_50": 0.1331348377221043, "lr": 0.00025, "grad_norm_before_clip": 0.014892578125, "grad_clipped": false, "step_seconds": 1.295020712997939, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 146, "loss": 0.2056363821029663, "loss_ema": 0.16426104767025262, "loss_mean_50": 0.16237702757120132, "loss_std_50": 0.1277930609978944, "lr": 0.00025, "grad_norm_before_clip": 0.00775146484375, "grad_clipped": false, "step_seconds": 2.0623952970017854, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 147, "loss": 0.053704291582107544, "loss_ema": 0.15873320986584535, "loss_mean_50": 0.16240595370531083, "loss_std_50": 0.12776829341490814, "lr": 0.00025, "grad_norm_before_clip": 0.021240234375, "grad_clipped": false, "step_seconds": 1.8293670329985616, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 148, "loss": 0.14008700847625732, "loss_ema": 0.15780089979636594, "loss_mean_50": 0.15854835540056228, "loss_std_50": 0.1254511759769356, "lr": 0.00025, "grad_norm_before_clip": 0.0172119140625, "grad_clipped": false, "step_seconds": 0.6007937030008179, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 149, "loss": 0.06103575974702835, "loss_ema": 0.15296264279389907, "loss_mean_50": 0.1582150314748287, "loss_std_50": 0.1256874606695779, "lr": 0.00025, "grad_norm_before_clip": 0.0169677734375, "grad_clipped": false, "step_seconds": 2.0687861750011507, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 150, "loss": 0.043099015951156616, "loss_ema": 0.14746946145176196, "loss_mean_50": 0.15403466567397117, "loss_std_50": 0.12597040727385553, "lr": 0.00025, "grad_norm_before_clip": 0.0205078125, "grad_clipped": false, "step_seconds": 2.0690540879986656, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 151, "loss": 0.06185286492109299, "loss_ema": 0.14318863162522852, "loss_mean_50": 0.1524799197912216, "loss_std_50": 0.12661715162786885, "lr": 0.00025, "grad_norm_before_clip": 0.0211181640625, "grad_clipped": false, "step_seconds": 2.069475414999033, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 152, "loss": 0.09540572762489319, "loss_ema": 0.14079948642521176, "loss_mean_50": 0.15299950122833253, "loss_std_50": 0.1263282450460703, "lr": 0.00025, "grad_norm_before_clip": 0.0125732421875, "grad_clipped": false, "step_seconds": 0.6731133100001898, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 153, "loss": 0.1033872663974762, "loss_ema": 0.13892887542382495, "loss_mean_50": 0.1427005535364151, "loss_std_50": 0.10756973839152346, "lr": 0.00025, "grad_norm_before_clip": 0.00897216796875, "grad_clipped": false, "step_seconds": 2.0630964690026303, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 154, "loss": 0.06374206393957138, "loss_ema": 0.13516953484961228, "loss_mean_50": 0.14222698375582696, "loss_std_50": 0.10786378280784191, "lr": 0.00025, "grad_norm_before_clip": 0.033935546875, "grad_clipped": false, "step_seconds": 0.47946379099812475, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 155, "loss": 0.2989034056663513, "loss_ema": 0.14335622839044923, "loss_mean_50": 0.14463236674666405, "loss_std_50": 0.10996931393944333, "lr": 0.00025, "grad_norm_before_clip": 0.00677490234375, "grad_clipped": false, "step_seconds": 2.069311210001615, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 156, "loss": 0.061289023607969284, "loss_ema": 0.13925286815132523, "loss_mean_50": 0.14404955349862575, "loss_std_50": 0.11033165448754667, "lr": 0.00025, "grad_norm_before_clip": 0.00897216796875, "grad_clipped": false, "step_seconds": 1.8940636469997116, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 157, "loss": 0.0678068995475769, "loss_ema": 0.1356805697211378, "loss_mean_50": 0.14131387047469615, "loss_std_50": 0.11049227759534862, "lr": 0.00025, "grad_norm_before_clip": 0.0150146484375, "grad_clipped": false, "step_seconds": 1.5742507930008287, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 158, "loss": 0.053140852600336075, "loss_ema": 0.13155358386509772, "loss_mean_50": 0.13556734085083008, "loss_std_50": 0.10741400461798475, "lr": 0.00025, "grad_norm_before_clip": 0.025634765625, "grad_clipped": false, "step_seconds": 1.8899117589971866, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 159, "loss": 0.10403697192668915, "loss_ema": 0.1301777532681773, "loss_mean_50": 0.13649322912096978, "loss_std_50": 0.10693764756653514, "lr": 0.00025, "grad_norm_before_clip": 0.021484375, "grad_clipped": false, "step_seconds": 0.3647666899996693, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 160, "loss": 0.14367257058620453, "loss_ema": 0.13085249413407865, "loss_mean_50": 0.13845567516982554, "loss_std_50": 0.10614814029656854, "lr": 0.00025, "grad_norm_before_clip": 0.0556640625, "grad_clipped": false, "step_seconds": 1.893487441000616, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 161, "loss": 0.04581020399928093, "loss_ema": 0.12660037962733875, "loss_mean_50": 0.13707896947860718, "loss_std_50": 0.10689180294438065, "lr": 0.00025, "grad_norm_before_clip": 0.0120849609375, "grad_clipped": false, "step_seconds": 1.8897688050019497, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 162, "loss": 0.34621185064315796, "loss_ema": 0.1375809531781297, "loss_mean_50": 0.1429708033055067, "loss_std_50": 0.11009001274316917, "lr": 0.00025, "grad_norm_before_clip": 0.041259765625, "grad_clipped": false, "step_seconds": 0.42439055899740197, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 163, "loss": 0.05841884762048721, "loss_ema": 0.1336228479002476, "loss_mean_50": 0.14303859300911426, "loss_std_50": 0.11003687124832666, "lr": 0.00025, "grad_norm_before_clip": 0.0191650390625, "grad_clipped": false, "step_seconds": 2.060727214000508, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 164, "loss": 0.10623296350240707, "loss_ema": 0.13225335368035557, "loss_mean_50": 0.1405302944034338, "loss_std_50": 0.10941609751681132, "lr": 0.00025, "grad_norm_before_clip": 0.06298828125, "grad_clipped": false, "step_seconds": 0.7247397329992964, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 165, "loss": 0.15511594712734222, "loss_ema": 0.1333964833527049, "loss_mean_50": 0.140039070174098, "loss_std_50": 0.10929431083250961, "lr": 0.00025, "grad_norm_before_clip": 0.0791015625, "grad_clipped": false, "step_seconds": 0.575280830002157, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 166, "loss": 0.18723763525485992, "loss_ema": 0.13608854094781264, "loss_mean_50": 0.1331064637750387, "loss_std_50": 0.09401976189102183, "lr": 0.00025, "grad_norm_before_clip": 0.0184326171875, "grad_clipped": false, "step_seconds": 1.8924143789990922, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 167, "loss": 0.051684536039829254, "loss_ema": 0.13186834070241346, "loss_mean_50": 0.13286761574447156, "loss_std_50": 0.09421093934754349, "lr": 0.00025, "grad_norm_before_clip": 0.0213623046875, "grad_clipped": false, "step_seconds": 1.2946425129994168, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 168, "loss": 0.1473582535982132, "loss_ema": 0.13264283634720345, "loss_mean_50": 0.13333143420517446, "loss_std_50": 0.09422405020824269, "lr": 0.00025, "grad_norm_before_clip": 0.012451171875, "grad_clipped": false, "step_seconds": 1.8399131200021657, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 169, "loss": 0.11890513449907303, "loss_ema": 0.13195595125479692, "loss_mean_50": 0.1330519898980856, "loss_std_50": 0.09424569903787076, "lr": 0.00025, "grad_norm_before_clip": 0.03955078125, "grad_clipped": false, "step_seconds": 2.0656397649981955, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 170, "loss": 0.40698686242103577, "loss_ema": 0.14570749681310888, "loss_mean_50": 0.1396563386172056, "loss_std_50": 0.10137103192670262, "lr": 0.00025, "grad_norm_before_clip": 0.021728515625, "grad_clipped": false, "step_seconds": 0.5717807910004922, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 171, "loss": 0.06201949343085289, "loss_ema": 0.14152309664399607, "loss_mean_50": 0.13844450011849405, "loss_std_50": 0.10192819320394281, "lr": 0.00025, "grad_norm_before_clip": 0.04248046875, "grad_clipped": false, "step_seconds": 0.6167131090005569, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 172, "loss": 0.4779774248600006, "loss_ema": 0.15834581305479628, "loss_mean_50": 0.1467805328965187, "loss_std_50": 0.11183072110675345, "lr": 0.00025, "grad_norm_before_clip": 0.02978515625, "grad_clipped": false, "step_seconds": 2.062255726999865, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 173, "loss": 0.3720622956752777, "loss_ema": 0.16903163718582034, "loss_mean_50": 0.15291013166308404, "loss_std_50": 0.11554961177252011, "lr": 0.00025, "grad_norm_before_clip": 0.061767578125, "grad_clipped": false, "step_seconds": 0.35903181200046674, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 174, "loss": 0.10187978297472, "loss_ema": 0.16567404447526532, "loss_mean_50": 0.15399074852466582, "loss_std_50": 0.11481232467558415, "lr": 0.00025, "grad_norm_before_clip": 0.01153564453125, "grad_clipped": false, "step_seconds": 2.113746689999971, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 175, "loss": 0.09117960184812546, "loss_ema": 0.16194932234390833, "loss_mean_50": 0.15476456306874753, "loss_std_50": 0.11425464237121391, "lr": 0.00025, "grad_norm_before_clip": 0.0185546875, "grad_clipped": false, "step_seconds": 0.4800399749983626, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 176, "loss": 0.13634638488292694, "loss_ema": 0.16066917547085927, "loss_mean_50": 0.1524295722693205, "loss_std_50": 0.11341108664007382, "lr": 0.00025, "grad_norm_before_clip": 0.09521484375, "grad_clipped": false, "step_seconds": 2.06323558800068, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 177, "loss": 0.10767350345849991, "loss_ema": 0.15801939187024128, "loss_mean_50": 0.15313401706516744, "loss_std_50": 0.11302083811129453, "lr": 0.00025, "grad_norm_before_clip": 0.043212890625, "grad_clipped": false, "step_seconds": 0.5751924030009832, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 178, "loss": 0.04000652953982353, "loss_ema": 0.15211874875372036, "loss_mean_50": 0.15276613228023053, "loss_std_50": 0.1133580315207197, "lr": 0.00025, "grad_norm_before_clip": 0.05419921875, "grad_clipped": false, "step_seconds": 2.07067118499981, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 179, "loss": 0.0665939599275589, "loss_ema": 0.14784250931241227, "loss_mean_50": 0.1468196215480566, "loss_std_50": 0.10987028861976947, "lr": 0.00025, "grad_norm_before_clip": 0.0286865234375, "grad_clipped": false, "step_seconds": 1.8950349740007368, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 180, "loss": 0.07986277341842651, "loss_ema": 0.14444352251771297, "loss_mean_50": 0.14424354128539563, "loss_std_50": 0.10989998674122604, "lr": 0.00025, "grad_norm_before_clip": 0.005889892578125, "grad_clipped": false, "step_seconds": 2.0669310509983916, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 181, "loss": 0.0757472962141037, "loss_ema": 0.1410087112025325, "loss_mean_50": 0.14246916674077512, "loss_std_50": 0.11027472247428408, "lr": 0.00025, "grad_norm_before_clip": 0.03662109375, "grad_clipped": false, "step_seconds": 1.8292006970004877, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 182, "loss": 0.9666167497634888, "loss_ema": 0.1822891131305803, "loss_mean_50": 0.1546959211677313, "loss_std_50": 0.15712938910491905, "lr": 0.00025, "grad_norm_before_clip": 0.2021484375, "grad_clipped": false, "step_seconds": 0.521102743998199, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 183, "loss": 0.12291572242975235, "loss_ema": 0.1793204435955389, "loss_mean_50": 0.15525837190449238, "loss_std_50": 0.15696420431327784, "lr": 0.00025, "grad_norm_before_clip": 0.011474609375, "grad_clipped": false, "step_seconds": 0.6163061150000431, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 184, "loss": 0.07145357877016068, "loss_ema": 0.17392710035426998, "loss_mean_50": 0.15423510931432247, "loss_std_50": 0.1573399811628707, "lr": 0.00025, "grad_norm_before_clip": 0.01708984375, "grad_clipped": false, "step_seconds": 1.8935901290024049, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 185, "loss": 0.07913462072610855, "loss_ema": 0.16918747637286188, "loss_mean_50": 0.14854620553553105, "loss_std_50": 0.15478954629608532, "lr": 0.00025, "grad_norm_before_clip": 0.011962890625, "grad_clipped": false, "step_seconds": 1.8962409529995057, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 186, "loss": 0.3663540482521057, "loss_ema": 0.17904580496682407, "loss_mean_50": 0.15505143366754054, "loss_std_50": 0.15695657172813243, "lr": 0.00025, "grad_norm_before_clip": 0.045654296875, "grad_clipped": false, "step_seconds": 1.8894711649991223, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 187, "loss": 0.11633317917585373, "loss_ema": 0.17591017367727554, "loss_mean_50": 0.15605307944118976, "loss_std_50": 0.15654594676851175, "lr": 0.00025, "grad_norm_before_clip": 0.046630859375, "grad_clipped": false, "step_seconds": 0.3617088930004684, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 188, "loss": 0.09706868976354599, "loss_ema": 0.17196809948158906, "loss_mean_50": 0.15280076555907726, "loss_std_50": 0.15604758929065807, "lr": 0.00025, "grad_norm_before_clip": 0.05029296875, "grad_clipped": false, "step_seconds": 0.5992570279995562, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 189, "loss": 0.05217523127794266, "loss_ema": 0.16597845607140674, "loss_mean_50": 0.15091964967548846, "loss_std_50": 0.1566810717979221, "lr": 0.00025, "grad_norm_before_clip": 0.0184326171875, "grad_clipped": false, "step_seconds": 0.6003827740023553, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 190, "loss": 0.07863268256187439, "loss_ema": 0.16161116739593012, "loss_mean_50": 0.14572728045284747, "loss_std_50": 0.15467593187327983, "lr": 0.00025, "grad_norm_before_clip": 0.00799560546875, "grad_clipped": false, "step_seconds": 1.890354267998191, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 191, "loss": 0.11394251137971878, "loss_ema": 0.15922773459511955, "loss_mean_50": 0.1465030487626791, "loss_std_50": 0.15441708463672507, "lr": 0.00025, "grad_norm_before_clip": 0.0091552734375, "grad_clipped": false, "step_seconds": 0.6724709189984424, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 192, "loss": 0.04789841175079346, "loss_ema": 0.15366126845290323, "loss_mean_50": 0.14533343158662318, "loss_std_50": 0.15493717165553417, "lr": 0.00025, "grad_norm_before_clip": 0.0101318359375, "grad_clipped": false, "step_seconds": 1.8393620570022904, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 193, "loss": 0.08858944475650787, "loss_ema": 0.15040767726808346, "loss_mean_50": 0.14390878044068814, "loss_std_50": 0.15512477802425065, "lr": 0.00025, "grad_norm_before_clip": 0.00982666015625, "grad_clipped": false, "step_seconds": 2.067289165999682, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 194, "loss": 0.5142767429351807, "loss_ema": 0.16860113055143833, "loss_mean_50": 0.1524716155976057, "loss_std_50": 0.16330054174866918, "lr": 0.00025, "grad_norm_before_clip": 0.173828125, "grad_clipped": false, "step_seconds": 1.8961915289983153, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 195, "loss": 0.0973009318113327, "loss_ema": 0.16503612061443304, "loss_mean_50": 0.15209607861936092, "loss_std_50": 0.16340536028551497, "lr": 0.00025, "grad_norm_before_clip": 0.00933837890625, "grad_clipped": false, "step_seconds": 2.067746985001577, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 196, "loss": 0.29351744055747986, "loss_ema": 0.17146018661158538, "loss_mean_50": 0.1538536997884512, "loss_std_50": 0.1644411484032243, "lr": 0.00025, "grad_norm_before_clip": 0.01513671875, "grad_clipped": false, "step_seconds": 1.842230988000665, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 197, "loss": 0.780427873134613, "loss_ema": 0.20190857093773676, "loss_mean_50": 0.1683881714195013, "loss_std_50": 0.18569045812153176, "lr": 0.00025, "grad_norm_before_clip": 1.1640625, "grad_clipped": true, "step_seconds": 2.067671528999199, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 198, "loss": 0.060885943472385406, "loss_ema": 0.19485743956446916, "loss_mean_50": 0.16680415011942387, "loss_std_50": 0.18626205350982875, "lr": 0.00025, "grad_norm_before_clip": 0.036376953125, "grad_clipped": false, "step_seconds": 1.8952541520011437, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 199, "loss": 0.05609386786818504, "loss_ema": 0.18791926097965495, "loss_mean_50": 0.166705312281847, "loss_std_50": 0.18631945441274722, "lr": 0.00025, "grad_norm_before_clip": 0.03173828125, "grad_clipped": false, "step_seconds": 2.067151502997149, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 200, "loss": 0.38169461488723755, "loss_ema": 0.19760802867503405, "loss_mean_50": 0.1734772242605686, "loss_std_50": 0.18785078646537384, "lr": 0.00025, "grad_norm_before_clip": 0.04248046875, "grad_clipped": false, "step_seconds": 1.9450773640019179, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 201, "loss": 0.10959521681070328, "loss_ema": 0.1932073880818175, "loss_mean_50": 0.17443207129836083, "loss_std_50": 0.1874017729594665, "lr": 0.00025, "grad_norm_before_clip": 0.12353515625, "grad_clipped": false, "step_seconds": 0.726523298999382, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 202, "loss": 0.13622121512889862, "loss_ema": 0.19035807943417157, "loss_mean_50": 0.17524838104844093, "loss_std_50": 0.1871444796873247, "lr": 0.00025, "grad_norm_before_clip": 0.014404296875, "grad_clipped": false, "step_seconds": 1.2943238410007325, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 203, "loss": 0.4468855559825897, "loss_ema": 0.20318445326159246, "loss_mean_50": 0.1821183468401432, "loss_std_50": 0.19065233947582927, "lr": 0.00025, "grad_norm_before_clip": 0.0673828125, "grad_clipped": false, "step_seconds": 1.8922144169991952, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 204, "loss": 0.17098166048526764, "loss_ema": 0.2015743136227762, "loss_mean_50": 0.18426313877105713, "loss_std_50": 0.18991033685340075, "lr": 0.00025, "grad_norm_before_clip": 0.00958251953125, "grad_clipped": false, "step_seconds": 1.2936787559992808, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 205, "loss": 0.16961780190467834, "loss_ema": 0.1999764880368713, "loss_mean_50": 0.18167742669582368, "loss_std_50": 0.18921070804475695, "lr": 0.00025, "grad_norm_before_clip": 0.01007080078125, "grad_clipped": false, "step_seconds": 1.5944972070028598, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 206, "loss": 0.7441975474357605, "loss_ema": 0.22718754100681574, "loss_mean_50": 0.1953355971723795, "loss_std_50": 0.20409031527932966, "lr": 0.00025, "grad_norm_before_clip": 0.1875, "grad_clipped": false, "step_seconds": 2.0596610399989004, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 207, "loss": 0.06726846098899841, "loss_ema": 0.21919158700592486, "loss_mean_50": 0.19532482840120793, "loss_std_50": 0.20409705810691253, "lr": 0.00025, "grad_norm_before_clip": 0.0146484375, "grad_clipped": false, "step_seconds": 0.8066569969996635, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 208, "loss": 0.06499076634645462, "loss_ema": 0.21148154597295132, "loss_mean_50": 0.19556182667613028, "loss_std_50": 0.20393863453219718, "lr": 0.00025, "grad_norm_before_clip": 0.0218505859375, "grad_clipped": false, "step_seconds": 0.4228767050008173, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 209, "loss": 0.08190026879310608, "loss_ema": 0.20500248211395905, "loss_mean_50": 0.19511909261345864, "loss_std_50": 0.20416075444481646, "lr": 0.00025, "grad_norm_before_clip": 0.0216064453125, "grad_clipped": false, "step_seconds": 1.109200590999535, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 210, "loss": 0.05089142918586731, "loss_ema": 0.19729692946755448, "loss_mean_50": 0.1932634697854519, "loss_std_50": 0.2050396737335935, "lr": 0.00025, "grad_norm_before_clip": 0.056884765625, "grad_clipped": false, "step_seconds": 2.071660052999505, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 211, "loss": 0.10503614693880081, "loss_ema": 0.19268389034111677, "loss_mean_50": 0.1944479886442423, "loss_std_50": 0.2043543408346523, "lr": 0.00025, "grad_norm_before_clip": 0.046875, "grad_clipped": false, "step_seconds": 2.062405206001131, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 212, "loss": 0.07938400655984879, "loss_ema": 0.18701889615205336, "loss_mean_50": 0.1891114317625761, "loss_std_50": 0.20380472708428476, "lr": 0.00025, "grad_norm_before_clip": 0.00897216796875, "grad_clipped": false, "step_seconds": 2.0672169609970297, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 213, "loss": 0.09497059881687164, "loss_ema": 0.18241648128529425, "loss_mean_50": 0.1898424667865038, "loss_std_50": 0.20339978187611973, "lr": 0.00025, "grad_norm_before_clip": 0.05419921875, "grad_clipped": false, "step_seconds": 2.0609149010015244, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 214, "loss": 0.05992647632956505, "loss_ema": 0.17629198103750776, "loss_mean_50": 0.18891633704304694, "loss_std_50": 0.20388321623343902, "lr": 0.00025, "grad_norm_before_clip": 0.07666015625, "grad_clipped": false, "step_seconds": 1.2713025209995976, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 215, "loss": 0.06703150272369385, "loss_ema": 0.17082895712181706, "loss_mean_50": 0.18715464815497398, "loss_std_50": 0.20454713697220242, "lr": 0.00025, "grad_norm_before_clip": 0.01318359375, "grad_clipped": false, "step_seconds": 2.063885186002153, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 216, "loss": 0.08349248766899109, "loss_ema": 0.16646213364917575, "loss_mean_50": 0.1850797452032566, "loss_std_50": 0.20506131462043528, "lr": 0.00025, "grad_norm_before_clip": 0.1298828125, "grad_clipped": false, "step_seconds": 0.6230942929978482, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 217, "loss": 0.9552731513977051, "loss_ema": 0.2059026845366022, "loss_mean_50": 0.20315151751041413, "loss_std_50": 0.23071979937121226, "lr": 0.00025, "grad_norm_before_clip": 0.21484375, "grad_clipped": false, "step_seconds": 2.0634014579991344, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 218, "loss": 0.07104169577360153, "loss_ema": 0.19915963509845216, "loss_mean_50": 0.2016251863539219, "loss_std_50": 0.23133546738258767, "lr": 0.00025, "grad_norm_before_clip": 0.025634765625, "grad_clipped": false, "step_seconds": 2.0702511930030596, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 219, "loss": 0.183138906955719, "loss_ema": 0.19835859869131547, "loss_mean_50": 0.20290986180305481, "loss_std_50": 0.23105071042750133, "lr": 0.00025, "grad_norm_before_clip": 0.06396484375, "grad_clipped": false, "step_seconds": 2.0668789470000775, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 220, "loss": 0.07117071747779846, "loss_ema": 0.19199920463063963, "loss_mean_50": 0.19619353890419006, "loss_std_50": 0.22989884165671137, "lr": 0.00025, "grad_norm_before_clip": 0.0291748046875, "grad_clipped": false, "step_seconds": 1.8944799789969693, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 221, "loss": 0.1074662059545517, "loss_ema": 0.18777255469683524, "loss_mean_50": 0.19710247315466403, "loss_std_50": 0.22945598387555607, "lr": 0.00025, "grad_norm_before_clip": 0.0113525390625, "grad_clipped": false, "step_seconds": 2.063808760001848, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 222, "loss": 0.12671291828155518, "loss_ema": 0.18471957287607121, "loss_mean_50": 0.19007718302309512, "loss_std_50": 0.22610168697659447, "lr": 0.00025, "grad_norm_before_clip": 0.0130615234375, "grad_clipped": false, "step_seconds": 0.7233026309986599, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 223, "loss": 0.070111945271492, "loss_ema": 0.17898919149584225, "loss_mean_50": 0.18403817601501943, "loss_std_50": 0.2251909518361112, "lr": 0.00025, "grad_norm_before_clip": 0.022705078125, "grad_clipped": false, "step_seconds": 0.42358786600016174, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 224, "loss": 0.12007059156894684, "loss_ema": 0.17604326149949748, "loss_mean_50": 0.18440199218690395, "loss_std_50": 0.22507258703117278, "lr": 0.00025, "grad_norm_before_clip": 0.007659912109375, "grad_clipped": false, "step_seconds": 1.5746292410003662, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 225, "loss": 0.0582679845392704, "loss_ema": 0.1701544976514861, "loss_mean_50": 0.18374375984072686, "loss_std_50": 0.22539215523078626, "lr": 0.00025, "grad_norm_before_clip": 0.0167236328125, "grad_clipped": false, "step_seconds": 1.8397429440010455, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 226, "loss": 0.06958968192338943, "loss_ema": 0.16512625686508126, "loss_mean_50": 0.1824086257815361, "loss_std_50": 0.2258661858350386, "lr": 0.00025, "grad_norm_before_clip": 0.08642578125, "grad_clipped": false, "step_seconds": 1.8929592770000454, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 227, "loss": 0.05049067363142967, "loss_ema": 0.15939447770339868, "loss_mean_50": 0.18126496918499468, "loss_std_50": 0.2263858787669278, "lr": 0.00025, "grad_norm_before_clip": 0.0185546875, "grad_clipped": false, "step_seconds": 1.2675724399996398, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 228, "loss": 0.2625437080860138, "loss_ema": 0.16455193922252942, "loss_mean_50": 0.1857157127559185, "loss_std_50": 0.22575163946086207, "lr": 0.00025, "grad_norm_before_clip": 0.020263671875, "grad_clipped": false, "step_seconds": 0.6047987109996029, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 229, "loss": 0.07374648749828339, "loss_ema": 0.1600116666363171, "loss_mean_50": 0.185858763307333, "loss_std_50": 0.22567836529885776, "lr": 0.00025, "grad_norm_before_clip": 0.0157470703125, "grad_clipped": false, "step_seconds": 1.8927860670009977, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 230, "loss": 0.05193428695201874, "loss_ema": 0.1546077976521022, "loss_mean_50": 0.18530019357800484, "loss_std_50": 0.2259743898624631, "lr": 0.00025, "grad_norm_before_clip": 0.0235595703125, "grad_clipped": false, "step_seconds": 1.2947636850003619, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 231, "loss": 0.07097988575696945, "loss_ema": 0.15042640205734556, "loss_mean_50": 0.18520484536886214, "loss_std_50": 0.22602159563108323, "lr": 0.00025, "grad_norm_before_clip": 0.0096435546875, "grad_clipped": false, "step_seconds": 1.8345448820000456, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 232, "loss": 0.05895189568400383, "loss_ema": 0.14585267673867847, "loss_mean_50": 0.16705154828727245, "loss_std_50": 0.19713681633875055, "lr": 0.00025, "grad_norm_before_clip": 0.0172119140625, "grad_clipped": false, "step_seconds": 1.8964851269993233, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 233, "loss": 0.0705086886882782, "loss_ema": 0.14208547733615845, "loss_mean_50": 0.16600340761244298, "loss_std_50": 0.19750766266049663, "lr": 0.00025, "grad_norm_before_clip": 0.009765625, "grad_clipped": false, "step_seconds": 2.0633867100004863, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 234, "loss": 0.26960107684135437, "loss_ema": 0.14846125731141824, "loss_mean_50": 0.16996635757386686, "loss_std_50": 0.197558669705804, "lr": 0.00025, "grad_norm_before_clip": 0.03125, "grad_clipped": false, "step_seconds": 0.36502058599944576, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 235, "loss": 0.06405515223741531, "loss_ema": 0.14424095205771809, "loss_mean_50": 0.16966476820409299, "loss_std_50": 0.19770855469533882, "lr": 0.00025, "grad_norm_before_clip": 0.032470703125, "grad_clipped": false, "step_seconds": 1.8910128169991367, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 236, "loss": 0.08757022768259048, "loss_ema": 0.14140741583896171, "loss_mean_50": 0.16408909179270267, "loss_std_50": 0.19600673775722066, "lr": 0.00025, "grad_norm_before_clip": 0.0184326171875, "grad_clipped": false, "step_seconds": 0.6744617889999063, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 237, "loss": 0.25262024998664856, "loss_ema": 0.14696805754634606, "loss_mean_50": 0.16681483320891857, "loss_std_50": 0.19627112460977458, "lr": 0.00025, "grad_norm_before_clip": 0.010498046875, "grad_clipped": false, "step_seconds": 2.0634208859992214, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 238, "loss": 0.04853406548500061, "loss_ema": 0.1420463579432788, "loss_mean_50": 0.16584414072334766, "loss_std_50": 0.19673314011991286, "lr": 0.00025, "grad_norm_before_clip": 0.0101318359375, "grad_clipped": false, "step_seconds": 1.8984458039994934, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 239, "loss": 0.06517697125673294, "loss_ema": 0.1382028886089515, "loss_mean_50": 0.16610417552292348, "loss_std_50": 0.19659126624553422, "lr": 0.00025, "grad_norm_before_clip": 0.02880859375, "grad_clipped": false, "step_seconds": 2.064541707000899, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 240, "loss": 0.04634686931967735, "loss_ema": 0.13361008764448778, "loss_mean_50": 0.16545845925807953, "loss_std_50": 0.19693024150874414, "lr": 0.00025, "grad_norm_before_clip": 0.016357421875, "grad_clipped": false, "step_seconds": 0.6004311770011554, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 241, "loss": 0.10681529343128204, "loss_ema": 0.1322703479338275, "loss_mean_50": 0.1653159148991108, "loss_std_50": 0.19697005422664046, "lr": 0.00025, "grad_norm_before_clip": 0.0081787109375, "grad_clipped": false, "step_seconds": 1.8392508430006274, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 242, "loss": 0.042714182287454605, "loss_ema": 0.12779253965150883, "loss_mean_50": 0.16521223030984403, "loss_std_50": 0.1970331896088259, "lr": 0.00025, "grad_norm_before_clip": 0.0167236328125, "grad_clipped": false, "step_seconds": 2.0707309669996903, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 243, "loss": 0.04809299856424332, "loss_ema": 0.12380756259714555, "loss_mean_50": 0.16440230138599873, "loss_std_50": 0.19742932681377504, "lr": 0.00025, "grad_norm_before_clip": 0.01165771484375, "grad_clipped": false, "step_seconds": 1.6637946980008564, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 244, "loss": 0.059986140578985214, "loss_ema": 0.12061649149623753, "loss_mean_50": 0.15531648933887482, "loss_std_50": 0.19148263446420208, "lr": 0.00025, "grad_norm_before_clip": 0.0067138671875, "grad_clipped": false, "step_seconds": 2.0639549319967045, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 245, "loss": 0.0608053132891655, "loss_ema": 0.11762593258588393, "loss_mean_50": 0.15458657696843148, "loss_std_50": 0.19177173326364877, "lr": 0.00025, "grad_norm_before_clip": 0.0166015625, "grad_clipped": false, "step_seconds": 0.48323883400007617, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 246, "loss": 0.09352850914001465, "loss_ema": 0.11642106141359045, "loss_mean_50": 0.15058679834008218, "loss_std_50": 0.1909160168913417, "lr": 0.00025, "grad_norm_before_clip": 0.0087890625, "grad_clipped": false, "step_seconds": 2.061336344002484, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 247, "loss": 0.10083045810461044, "loss_ema": 0.11564153124814144, "loss_mean_50": 0.1369948500394821, "loss_std_50": 0.16846276361544957, "lr": 0.00025, "grad_norm_before_clip": 0.01470947265625, "grad_clipped": false, "step_seconds": 1.2981639059980807, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 248, "loss": 0.7799427509307861, "loss_ema": 0.14885659223227368, "loss_mean_50": 0.15137598618865014, "loss_std_50": 0.19059033133936212, "lr": 0.00025, "grad_norm_before_clip": 0.416015625, "grad_clipped": false, "step_seconds": 2.0651016110023193, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 249, "loss": 0.05407511442899704, "loss_ema": 0.14411751834210984, "loss_mean_50": 0.15133561111986638, "loss_std_50": 0.19061072457071435, "lr": 0.00025, "grad_norm_before_clip": 0.00933837890625, "grad_clipped": false, "step_seconds": 1.1068054549978115, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 250, "loss": 0.12011857330799103, "loss_ema": 0.1429175710904039, "loss_mean_50": 0.14610409028828145, "loss_std_50": 0.18778515427962475, "lr": 0.00025, "grad_norm_before_clip": 0.008056640625, "grad_clipped": false, "step_seconds": 2.063412760002393, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 251, "loss": 0.6474580764770508, "loss_ema": 0.16814459635973625, "loss_mean_50": 0.1568613474816084, "loss_std_50": 0.20036966840815607, "lr": 0.00025, "grad_norm_before_clip": 0.10546875, "grad_clipped": false, "step_seconds": 0.5242746779986192, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 252, "loss": 0.09569139033555984, "loss_ema": 0.1645219360585274, "loss_mean_50": 0.15605075098574162, "loss_std_50": 0.20053344340331253, "lr": 0.00025, "grad_norm_before_clip": 0.01953125, "grad_clipped": false, "step_seconds": 1.8933698520013422, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 253, "loss": 0.07025247812271118, "loss_ema": 0.15980846316173658, "loss_mean_50": 0.14851808942854405, "loss_std_50": 0.19650050923503673, "lr": 0.00025, "grad_norm_before_clip": 0.01483154296875, "grad_clipped": false, "step_seconds": 1.89042056299877, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 254, "loss": 0.08435840904712677, "loss_ema": 0.15603596045600607, "loss_mean_50": 0.14678562439978124, "loss_std_50": 0.19667660192761457, "lr": 0.00025, "grad_norm_before_clip": 0.027099609375, "grad_clipped": false, "step_seconds": 0.570849357998668, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 255, "loss": 0.12028802931308746, "loss_ema": 0.15424856389886013, "loss_mean_50": 0.1457990289479494, "loss_std_50": 0.1966833207456165, "lr": 0.00025, "grad_norm_before_clip": 0.0264892578125, "grad_clipped": false, "step_seconds": 0.3512261310024769, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 256, "loss": 0.4813821613788605, "loss_ema": 0.17060524377286015, "loss_mean_50": 0.1405427212268114, "loss_std_50": 0.18370466767488097, "lr": 0.00025, "grad_norm_before_clip": 0.25, "grad_clipped": false, "step_seconds": 2.0667915319972963, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 257, "loss": 0.046782560646533966, "loss_ema": 0.16441410961654382, "loss_mean_50": 0.1401330032199621, "loss_std_50": 0.18389038600934562, "lr": 0.00025, "grad_norm_before_clip": 0.046875, "grad_clipped": false, "step_seconds": 2.063560657999915, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 258, "loss": 0.10726837813854218, "loss_ema": 0.1615568230426437, "loss_mean_50": 0.14097855545580387, "loss_std_50": 0.18363995661203575, "lr": 0.00025, "grad_norm_before_clip": 0.0208740234375, "grad_clipped": false, "step_seconds": 0.5705507400016359, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 259, "loss": 0.08561442047357559, "loss_ema": 0.15775970291419028, "loss_mean_50": 0.14105283848941327, "loss_std_50": 0.18361679393848454, "lr": 0.00025, "grad_norm_before_clip": 0.0142822265625, "grad_clipped": false, "step_seconds": 0.5796445249980025, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 260, "loss": 0.05195458605885506, "loss_ema": 0.15246944707142351, "loss_mean_50": 0.141074101626873, "loss_std_50": 0.18360641312682527, "lr": 0.00025, "grad_norm_before_clip": 0.034912109375, "grad_clipped": false, "step_seconds": 2.063734965002368, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 261, "loss": 0.05493989586830139, "loss_ema": 0.14759296951126738, "loss_mean_50": 0.14007217660546303, "loss_std_50": 0.18393672398958597, "lr": 0.00025, "grad_norm_before_clip": 0.05029296875, "grad_clipped": false, "step_seconds": 1.1037029569997685, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 262, "loss": 0.12333947420120239, "loss_ema": 0.1463802947457641, "loss_mean_50": 0.1409512859582901, "loss_std_50": 0.18374951479035848, "lr": 0.00025, "grad_norm_before_clip": 0.08154296875, "grad_clipped": false, "step_seconds": 1.2914395010011503, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 263, "loss": 0.42080989480018616, "loss_ema": 0.1601017747484852, "loss_mean_50": 0.14746807187795638, "loss_std_50": 0.18773797706917986, "lr": 0.00025, "grad_norm_before_clip": 0.038330078125, "grad_clipped": false, "step_seconds": 1.889247950999561, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 264, "loss": 0.2753879725933075, "loss_ema": 0.1658660846407263, "loss_mean_50": 0.15177730180323123, "loss_std_50": 0.18815147613674563, "lr": 0.00025, "grad_norm_before_clip": 0.035400390625, "grad_clipped": false, "step_seconds": 0.6684090300004755, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 265, "loss": 0.3181735873222351, "loss_ema": 0.17348145977480173, "loss_mean_50": 0.15680014349520208, "loss_std_50": 0.18917152666622852, "lr": 0.00025, "grad_norm_before_clip": 0.051025390625, "grad_clipped": false, "step_seconds": 2.063086830999964, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 266, "loss": 0.07840418070554733, "loss_ema": 0.168727595821339, "loss_mean_50": 0.1566983773559332, "loss_std_50": 0.18921229990797153, "lr": 0.00025, "grad_norm_before_clip": 0.0224609375, "grad_clipped": false, "step_seconds": 0.5720388949994231, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 267, "loss": 0.07468393445014954, "loss_ema": 0.16402541275277951, "loss_mean_50": 0.1390865930169821, "loss_std_50": 0.15123231622589142, "lr": 0.00025, "grad_norm_before_clip": 0.00616455078125, "grad_clipped": false, "step_seconds": 2.059849156998098, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 268, "loss": 0.13049747049808502, "loss_ema": 0.16234901564004478, "loss_mean_50": 0.14027570851147175, "loss_std_50": 0.15092605068405002, "lr": 0.00025, "grad_norm_before_clip": 0.030517578125, "grad_clipped": false, "step_seconds": 0.723211471002287, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 269, "loss": 0.13490253686904907, "loss_ema": 0.160976691701495, "loss_mean_50": 0.13931098110973836, "loss_std_50": 0.15080309817764798, "lr": 0.00025, "grad_norm_before_clip": 0.0179443359375, "grad_clipped": false, "step_seconds": 0.36169307499949355, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 270, "loss": 0.9277240037918091, "loss_ema": 0.1993140573060107, "loss_mean_50": 0.15644204683601856, "loss_std_50": 0.1865131120945554, "lr": 0.00025, "grad_norm_before_clip": 0.2451171875, "grad_clipped": false, "step_seconds": 2.0616679249978915, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 271, "loss": 0.06536400318145752, "loss_ema": 0.19261655459978302, "loss_mean_50": 0.1556000027805567, "loss_std_50": 0.18682709515999749, "lr": 0.00025, "grad_norm_before_clip": 0.017333984375, "grad_clipped": false, "step_seconds": 0.5241704220024985, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 272, "loss": 0.07357928901910782, "loss_ema": 0.18666469132074925, "loss_mean_50": 0.15453733019530774, "loss_std_50": 0.18713923396658708, "lr": 0.00025, "grad_norm_before_clip": 0.0291748046875, "grad_clipped": false, "step_seconds": 2.063627739000367, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 273, "loss": 0.30766478180885315, "loss_ema": 0.19271469584515444, "loss_mean_50": 0.15928838692605496, "loss_std_50": 0.18794926912841325, "lr": 0.00025, "grad_norm_before_clip": 0.026611328125, "grad_clipped": false, "step_seconds": 1.89030588300011, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 274, "loss": 0.6961368918418884, "loss_ema": 0.21788580564499113, "loss_mean_50": 0.17080971293151379, "loss_std_50": 0.2023006486596979, "lr": 0.00025, "grad_norm_before_clip": 0.06591796875, "grad_clipped": false, "step_seconds": 0.6027730969981349, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 275, "loss": 0.08692061901092529, "loss_ema": 0.21133754631328783, "loss_mean_50": 0.1713827656209469, "loss_std_50": 0.20202143165450479, "lr": 0.00025, "grad_norm_before_clip": 0.0086669921875, "grad_clipped": false, "step_seconds": 2.058138043001236, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 276, "loss": 0.058037932962179184, "loss_ema": 0.20367256564573238, "loss_mean_50": 0.17115173064172268, "loss_std_50": 0.20214427979690602, "lr": 0.00025, "grad_norm_before_clip": 0.0126953125, "grad_clipped": false, "step_seconds": 0.47912175199962803, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 277, "loss": 0.5913259983062744, "loss_ema": 0.22305523727875948, "loss_mean_50": 0.18196843713521957, "loss_std_50": 0.20972614335322007, "lr": 0.00025, "grad_norm_before_clip": 0.029296875, "grad_clipped": false, "step_seconds": 1.894061552000494, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 278, "loss": 0.22840550541877747, "loss_ema": 0.22332275068576035, "loss_mean_50": 0.18128567308187485, "loss_std_50": 0.2095181843647525, "lr": 0.00025, "grad_norm_before_clip": 0.06591796875, "grad_clipped": false, "step_seconds": 1.892254652997508, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 279, "loss": 0.08271140605211258, "loss_ema": 0.21629218345407797, "loss_mean_50": 0.18146497145295143, "loss_std_50": 0.2094298966845135, "lr": 0.00025, "grad_norm_before_clip": 0.00732421875, "grad_clipped": false, "step_seconds": 1.3411587520022294, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 280, "loss": 0.06832720339298248, "loss_ema": 0.2088939344510232, "loss_mean_50": 0.1817928297817707, "loss_std_50": 0.20923960729481514, "lr": 0.00025, "grad_norm_before_clip": 0.02880859375, "grad_clipped": false, "step_seconds": 0.6127595269972517, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 281, "loss": 0.818049430847168, "loss_ema": 0.23935170927083044, "loss_mean_50": 0.19673422068357468, "loss_std_50": 0.22673514764042407, "lr": 0.00025, "grad_norm_before_clip": 1.9453125, "grad_clipped": true, "step_seconds": 0.42120316500222543, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 282, "loss": 0.45779240131378174, "loss_ema": 0.25027374387297796, "loss_mean_50": 0.20471103079617023, "loss_std_50": 0.22875433497047837, "lr": 0.00025, "grad_norm_before_clip": 0.032470703125, "grad_clipped": false, "step_seconds": 1.8916134130013234, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 283, "loss": 0.1867596060037613, "loss_ema": 0.24709803697951713, "loss_mean_50": 0.2070360491424799, "loss_std_50": 0.22796793581109567, "lr": 0.00025, "grad_norm_before_clip": 0.01275634765625, "grad_clipped": false, "step_seconds": 1.2692840869967768, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 284, "loss": 0.11270033568143845, "loss_ema": 0.24037815191461318, "loss_mean_50": 0.20389803431928158, "loss_std_50": 0.2281649173378898, "lr": 0.00025, "grad_norm_before_clip": 0.01055908203125, "grad_clipped": false, "step_seconds": 1.5705008910008473, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 285, "loss": 0.07143943756818771, "loss_ema": 0.23193121619729187, "loss_mean_50": 0.20404572002589702, "loss_std_50": 0.22807672537202095, "lr": 0.00025, "grad_norm_before_clip": 0.010986328125, "grad_clipped": false, "step_seconds": 2.0642389519998687, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 286, "loss": 0.14675745368003845, "loss_ema": 0.2276725280714292, "loss_mean_50": 0.20522946454584598, "loss_std_50": 0.2276222737524202, "lr": 0.00025, "grad_norm_before_clip": 0.01116943359375, "grad_clipped": false, "step_seconds": 2.0630510289993254, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 287, "loss": 0.044535208493471146, "loss_ema": 0.21851566209253132, "loss_mean_50": 0.20106776371598245, "loss_std_50": 0.2286178359229547, "lr": 0.00025, "grad_norm_before_clip": 0.0126953125, "grad_clipped": false, "step_seconds": 0.5976806780017796, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 288, "loss": 0.8414866924285889, "loss_ema": 0.2496642136093342, "loss_mean_50": 0.21692681625485422, "loss_std_50": 0.24444222805769872, "lr": 0.00025, "grad_norm_before_clip": 0.2109375, "grad_clipped": false, "step_seconds": 1.8333051090012304, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 289, "loss": 0.09008864313364029, "loss_ema": 0.24168543508554946, "loss_mean_50": 0.21742504969239235, "loss_std_50": 0.24415763915005642, "lr": 0.00025, "grad_norm_before_clip": 0.007049560546875, "grad_clipped": false, "step_seconds": 2.0637604579969775, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 290, "loss": 0.050429172813892365, "loss_ema": 0.2321226219719666, "loss_mean_50": 0.21750669576227666, "loss_std_50": 0.2441010931379539, "lr": 0.00025, "grad_norm_before_clip": 0.017333984375, "grad_clipped": false, "step_seconds": 1.5902675180004735, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 291, "loss": 0.06689687073230743, "loss_ema": 0.22386133440998363, "loss_mean_50": 0.21670832730829714, "loss_std_50": 0.24452672846923473, "lr": 0.00025, "grad_norm_before_clip": 0.0157470703125, "grad_clipped": false, "step_seconds": 1.8911972300011257, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 292, "loss": 0.058608800172805786, "loss_ema": 0.21559870769812473, "loss_mean_50": 0.2170262196660042, "loss_std_50": 0.24431056022999578, "lr": 0.00025, "grad_norm_before_clip": 0.00701904296875, "grad_clipped": false, "step_seconds": 2.0677908350007783, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 293, "loss": 0.17892633378505707, "loss_ema": 0.21376508900247135, "loss_mean_50": 0.21964288637042045, "loss_std_50": 0.24318525090452933, "lr": 0.00025, "grad_norm_before_clip": 0.01007080078125, "grad_clipped": false, "step_seconds": 2.0685030439999537, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 294, "loss": 0.10236966609954834, "loss_ema": 0.20819531785732517, "loss_mean_50": 0.22049055688083172, "loss_std_50": 0.24270064348189202, "lr": 0.00025, "grad_norm_before_clip": 0.0087890625, "grad_clipped": false, "step_seconds": 2.0647707370007993, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 295, "loss": 0.051241133362054825, "loss_ema": 0.20034760863256165, "loss_mean_50": 0.22029927328228951, "loss_std_50": 0.24283015785507278, "lr": 0.00025, "grad_norm_before_clip": 0.016845703125, "grad_clipped": false, "step_seconds": 1.8914113439968787, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 296, "loss": 0.06970933079719543, "loss_ema": 0.19381569474079333, "loss_mean_50": 0.21982288971543312, "loss_std_50": 0.2431016016099971, "lr": 0.00025, "grad_norm_before_clip": 0.01556396484375, "grad_clipped": false, "step_seconds": 2.063165313000354, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 297, "loss": 0.3755088448524475, "loss_ema": 0.202900352246376, "loss_mean_50": 0.22531645745038986, "loss_std_50": 0.24345387196718096, "lr": 0.00025, "grad_norm_before_clip": 0.0281982421875, "grad_clipped": false, "step_seconds": 1.8919100330022047, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 298, "loss": 0.23868107795715332, "loss_ema": 0.20468938853191487, "loss_mean_50": 0.21449122399091722, "loss_std_50": 0.2302259061217703, "lr": 0.00025, "grad_norm_before_clip": 0.0123291015625, "grad_clipped": false, "step_seconds": 2.065629837998131, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 299, "loss": 0.16545258462429047, "loss_ema": 0.20272754833653364, "loss_mean_50": 0.21671877339482307, "loss_std_50": 0.22919955294922004, "lr": 0.00025, "grad_norm_before_clip": 0.08056640625, "grad_clipped": false, "step_seconds": 0.3508930470015912, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 300, "loss": 0.06416043639183044, "loss_ema": 0.19579919273929847, "loss_mean_50": 0.21559961065649985, "loss_std_50": 0.22980433306953443, "lr": 0.00025, "grad_norm_before_clip": 0.033203125, "grad_clipped": false, "step_seconds": 1.8326842809983646, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 301, "loss": 0.08566455543041229, "loss_ema": 0.19029246087385415, "loss_mean_50": 0.20436374023556708, "loss_std_50": 0.22201669845967006, "lr": 0.00025, "grad_norm_before_clip": 0.0216064453125, "grad_clipped": false, "step_seconds": 1.8280077260023972, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 302, "loss": 0.06404975056648254, "loss_ema": 0.18398032535848557, "loss_mean_50": 0.20373090744018554, "loss_std_50": 0.22237036816731334, "lr": 0.00025, "grad_norm_before_clip": 0.035888671875, "grad_clipped": false, "step_seconds": 0.5290625229972648, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 303, "loss": 0.24111390113830566, "loss_ema": 0.18683700414747656, "loss_mean_50": 0.20714813590049744, "loss_std_50": 0.22160442940296116, "lr": 0.00025, "grad_norm_before_clip": 0.040283203125, "grad_clipped": false, "step_seconds": 2.0644336700024724, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 304, "loss": 0.060757409781217575, "loss_ema": 0.1805330244291636, "loss_mean_50": 0.20667611591517926, "loss_std_50": 0.22189042088744063, "lr": 0.00025, "grad_norm_before_clip": 0.0172119140625, "grad_clipped": false, "step_seconds": 1.8915459079980792, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 305, "loss": 0.07671216875314713, "loss_ema": 0.17534198164536277, "loss_mean_50": 0.20580459870398043, "loss_std_50": 0.22231318867862837, "lr": 0.00025, "grad_norm_before_clip": 0.04248046875, "grad_clipped": false, "step_seconds": 2.0636260329993092, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 306, "loss": 0.09397804737091064, "loss_ema": 0.17127378493164017, "loss_mean_50": 0.19805651642382144, "loss_std_50": 0.21930427440454442, "lr": 0.00025, "grad_norm_before_clip": 0.0712890625, "grad_clipped": false, "step_seconds": 0.597488985000382, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 307, "loss": 0.06375209242105484, "loss_ema": 0.1658977003061109, "loss_mean_50": 0.19839590705931187, "loss_std_50": 0.21908292256199335, "lr": 0.00025, "grad_norm_before_clip": 0.01031494140625, "grad_clipped": false, "step_seconds": 1.891664907001541, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 308, "loss": 0.17194153368473053, "loss_ema": 0.16619989197504187, "loss_mean_50": 0.19968937017023564, "loss_std_50": 0.21873172165544208, "lr": 0.00025, "grad_norm_before_clip": 0.01080322265625, "grad_clipped": false, "step_seconds": 2.06302324599892, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 309, "loss": 0.10063706338405609, "loss_ema": 0.16292175054549257, "loss_mean_50": 0.19998982302844526, "loss_std_50": 0.2185850889083268, "lr": 0.00025, "grad_norm_before_clip": 0.01025390625, "grad_clipped": false, "step_seconds": 0.6678184230004263, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 310, "loss": 0.07986830919981003, "loss_ema": 0.15876907847820843, "loss_mean_50": 0.20054809749126434, "loss_std_50": 0.2182416650337009, "lr": 0.00025, "grad_norm_before_clip": 0.01141357421875, "grad_clipped": false, "step_seconds": 1.1033733969998138, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 311, "loss": 0.1587352156639099, "loss_ema": 0.1587673853374935, "loss_mean_50": 0.20262400388717652, "loss_std_50": 0.21733855236348248, "lr": 0.00025, "grad_norm_before_clip": 0.006744384765625, "grad_clipped": false, "step_seconds": 2.072567152001284, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 312, "loss": 0.07471901178359985, "loss_ema": 0.1545649666597988, "loss_mean_50": 0.20165159463882446, "loss_std_50": 0.21779938882112995, "lr": 0.00025, "grad_norm_before_clip": 0.031982421875, "grad_clipped": false, "step_seconds": 2.06380855399766, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 313, "loss": 0.06717291474342346, "loss_ema": 0.15019536406398004, "loss_mean_50": 0.1945788550376892, "loss_std_50": 0.21630449191297, "lr": 0.00025, "grad_norm_before_clip": 0.033447265625, "grad_clipped": false, "step_seconds": 1.8931038980008452, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 314, "loss": 0.07868794351816177, "loss_ema": 0.14661999303668913, "loss_mean_50": 0.19064485445618629, "loss_std_50": 0.21658755425694945, "lr": 0.00025, "grad_norm_before_clip": 0.006134033203125, "grad_clipped": false, "step_seconds": 1.8931273810012499, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 315, "loss": 1.0426579713821411, "loss_ema": 0.19142189195396173, "loss_mean_50": 0.2051345421373844, "loss_std_50": 0.2467660290316671, "lr": 0.00025, "grad_norm_before_clip": 0.40234375, "grad_clipped": false, "step_seconds": 2.0807314949997817, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 316, "loss": 0.07131430506706238, "loss_ema": 0.18541651260961675, "loss_mean_50": 0.20499274462461473, "loss_std_50": 0.24684083617655572, "lr": 0.00025, "grad_norm_before_clip": 0.040283203125, "grad_clipped": false, "step_seconds": 0.4780461299997114, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 317, "loss": 0.09372551739215851, "loss_ema": 0.18083196284874384, "loss_mean_50": 0.2053735762834549, "loss_std_50": 0.2466541172746304, "lr": 0.00025, "grad_norm_before_clip": 0.0301513671875, "grad_clipped": false, "step_seconds": 1.831707761000871, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 318, "loss": 0.060947638005018234, "loss_ema": 0.17483774660655754, "loss_mean_50": 0.20398257963359356, "loss_std_50": 0.24726780442558163, "lr": 0.00025, "grad_norm_before_clip": 0.0267333984375, "grad_clipped": false, "step_seconds": 1.5720129209985316, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 319, "loss": 0.07690819352865219, "loss_ema": 0.16994126895266226, "loss_mean_50": 0.20282269276678563, "loss_std_50": 0.2477247236628105, "lr": 0.00025, "grad_norm_before_clip": 0.025146484375, "grad_clipped": false, "step_seconds": 0.573261248999188, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 320, "loss": 0.41768932342529297, "loss_ema": 0.1823286716762938, "loss_mean_50": 0.1926219991594553, "loss_std_50": 0.22732619674960478, "lr": 0.00025, "grad_norm_before_clip": 0.068359375, "grad_clipped": false, "step_seconds": 0.6156355180028186, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 321, "loss": 0.1068355068564415, "loss_ema": 0.17855401343530117, "loss_mean_50": 0.19345142923295497, "loss_std_50": 0.22693568751701346, "lr": 0.00025, "grad_norm_before_clip": 0.0234375, "grad_clipped": false, "step_seconds": 2.0641025909972086, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 322, "loss": 0.09644050151109695, "loss_ema": 0.17444833783909094, "loss_mean_50": 0.19390865348279476, "loss_std_50": 0.22671663595061808, "lr": 0.00025, "grad_norm_before_clip": 0.0281982421875, "grad_clipped": false, "step_seconds": 0.36253403600130696, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 323, "loss": 0.06566912680864334, "loss_ema": 0.16900937728756857, "loss_mean_50": 0.18906874038279056, "loss_std_50": 0.2268195464402198, "lr": 0.00025, "grad_norm_before_clip": 0.044921875, "grad_clipped": false, "step_seconds": 1.8916644169985375, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 324, "loss": 0.08628369122743607, "loss_ema": 0.16487309298456193, "loss_mean_50": 0.17687167637050152, "loss_std_50": 0.21533060952623895, "lr": 0.00025, "grad_norm_before_clip": 0.0087890625, "grad_clipped": false, "step_seconds": 2.0599955190009496, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 325, "loss": 0.06357647478580475, "loss_ema": 0.15980826207462406, "loss_mean_50": 0.1764047934859991, "loss_std_50": 0.21555033198525572, "lr": 0.00025, "grad_norm_before_clip": 0.00860595703125, "grad_clipped": false, "step_seconds": 1.2907480300018506, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 326, "loss": 0.9552216529846191, "loss_ema": 0.1995789316201238, "loss_mean_50": 0.1943484678864479, "loss_std_50": 0.24081293550668162, "lr": 0.00025, "grad_norm_before_clip": 0.6953125, "grad_clipped": false, "step_seconds": 0.7244109659986862, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 327, "loss": 0.06713671237230301, "loss_ema": 0.19295682065773276, "loss_mean_50": 0.18386468216776847, "loss_std_50": 0.2346333172548292, "lr": 0.00025, "grad_norm_before_clip": 0.01226806640625, "grad_clipped": false, "step_seconds": 1.5859979829983786, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 328, "loss": 0.0579180046916008, "loss_ema": 0.18620487985942613, "loss_mean_50": 0.18045493215322495, "loss_std_50": 0.23519936390958163, "lr": 0.00025, "grad_norm_before_clip": 0.01263427734375, "grad_clipped": false, "step_seconds": 0.5715966970019508, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 329, "loss": 0.05190513655543327, "loss_ema": 0.1794898926942265, "loss_mean_50": 0.17983880676329136, "loss_std_50": 0.23549476901631342, "lr": 0.00025, "grad_norm_before_clip": 0.025634765625, "grad_clipped": false, "step_seconds": 2.0636410669976613, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 330, "loss": 0.054949983954429626, "loss_ema": 0.17326289725723665, "loss_mean_50": 0.1795712623745203, "loss_std_50": 0.23562886551009443, "lr": 0.00025, "grad_norm_before_clip": 0.0615234375, "grad_clipped": false, "step_seconds": 1.2673111900003278, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 331, "loss": 0.10888087749481201, "loss_ema": 0.1700437962691154, "loss_mean_50": 0.1653878913074732, "loss_std_50": 0.21740894513500061, "lr": 0.00025, "grad_norm_before_clip": 0.0693359375, "grad_clipped": false, "step_seconds": 1.8282180410024012, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 332, "loss": 0.26777613162994385, "loss_ema": 0.1749304130371568, "loss_mean_50": 0.16158756591379642, "loss_std_50": 0.21389685840914527, "lr": 0.00025, "grad_norm_before_clip": 0.0673828125, "grad_clipped": false, "step_seconds": 0.34750925499974983, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 333, "loss": 0.0733431801199913, "loss_ema": 0.16985105139129852, "loss_mean_50": 0.15931923739612103, "loss_std_50": 0.2142190223193882, "lr": 0.00025, "grad_norm_before_clip": 0.0166015625, "grad_clipped": false, "step_seconds": 1.890752439998323, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 334, "loss": 0.08137176185846329, "loss_ema": 0.16542708691465674, "loss_mean_50": 0.15869266591966152, "loss_std_50": 0.21440020212343933, "lr": 0.00025, "grad_norm_before_clip": 0.0074462890625, "grad_clipped": false, "step_seconds": 1.8904056979990855, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 335, "loss": 0.3265155553817749, "loss_ema": 0.17348151033801265, "loss_mean_50": 0.16379418827593326, "loss_std_50": 0.21529618916184573, "lr": 0.00025, "grad_norm_before_clip": 0.0927734375, "grad_clipped": false, "step_seconds": 1.2911245280010917, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 336, "loss": 0.053408779203891754, "loss_ema": 0.1674778737813066, "loss_mean_50": 0.16192721478641034, "loss_std_50": 0.21583988797673356, "lr": 0.00025, "grad_norm_before_clip": 0.01336669921875, "grad_clipped": false, "step_seconds": 2.0642905429995153, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 337, "loss": 0.0678793415427208, "loss_ema": 0.16249794716937732, "loss_mean_50": 0.16239409744739533, "loss_std_50": 0.2156105787033497, "lr": 0.00025, "grad_norm_before_clip": 0.01055908203125, "grad_clipped": false, "step_seconds": 2.0632462210014637, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 338, "loss": 0.8010963201522827, "loss_ema": 0.19442786581852256, "loss_mean_50": 0.1615862900018692, "loss_std_50": 0.21312612376365728, "lr": 0.00025, "grad_norm_before_clip": 0.2294921875, "grad_clipped": false, "step_seconds": 0.5988754479985801, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 339, "loss": 0.07051308453083038, "loss_ema": 0.18823212675413795, "loss_mean_50": 0.161194778829813, "loss_std_50": 0.21327503289055058, "lr": 0.00025, "grad_norm_before_clip": 0.0277099609375, "grad_clipped": false, "step_seconds": 1.8907576830024482, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 340, "loss": 0.05319790169596672, "loss_ema": 0.18148041550122937, "loss_mean_50": 0.1612501534074545, "loss_std_50": 0.21324662414212647, "lr": 0.00025, "grad_norm_before_clip": 0.0228271484375, "grad_clipped": false, "step_seconds": 2.062573084000178, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 341, "loss": 0.5895124673843384, "loss_ema": 0.2018820180953848, "loss_mean_50": 0.1717024653404951, "loss_std_50": 0.22103165663699162, "lr": 0.00025, "grad_norm_before_clip": 0.26171875, "grad_clipped": false, "step_seconds": 0.42385139299949515, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 342, "loss": 0.40497803688049316, "loss_ema": 0.21203681903464022, "loss_mean_50": 0.17862985007464885, "loss_std_50": 0.2227993511659381, "lr": 0.00025, "grad_norm_before_clip": 0.07080078125, "grad_clipped": false, "step_seconds": 2.0636158479974256, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 343, "loss": 0.15014439821243286, "loss_ema": 0.20894219799352984, "loss_mean_50": 0.17805421136319638, "loss_std_50": 0.22283502009635056, "lr": 0.00025, "grad_norm_before_clip": 0.00738525390625, "grad_clipped": false, "step_seconds": 2.0636325020022923, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 344, "loss": 0.1773565709590912, "loss_ema": 0.2073629166418079, "loss_mean_50": 0.17955394946038722, "loss_std_50": 0.2225727829119907, "lr": 0.00025, "grad_norm_before_clip": 0.0084228515625, "grad_clipped": false, "step_seconds": 1.829331827997521, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 345, "loss": 0.10441523790359497, "loss_ema": 0.20221553270489726, "loss_mean_50": 0.18061743155121804, "loss_std_50": 0.22208364560006946, "lr": 0.00025, "grad_norm_before_clip": 0.04248046875, "grad_clipped": false, "step_seconds": 0.36882973900355864, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 346, "loss": 0.23425889015197754, "loss_ema": 0.20381770057725127, "loss_mean_50": 0.18390842273831368, "loss_std_50": 0.22163449868055912, "lr": 0.00025, "grad_norm_before_clip": 0.030517578125, "grad_clipped": false, "step_seconds": 2.062793190998491, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 347, "loss": 0.08727725595235825, "loss_ema": 0.19799067834600662, "loss_mean_50": 0.17814379096031188, "loss_std_50": 0.22032057833677005, "lr": 0.00025, "grad_norm_before_clip": 0.03955078125, "grad_clipped": false, "step_seconds": 0.5715904959979525, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 348, "loss": 0.575825035572052, "loss_ema": 0.21688239620730887, "loss_mean_50": 0.18488667011260987, "loss_std_50": 0.22712420171211703, "lr": 0.00025, "grad_norm_before_clip": 0.04150390625, "grad_clipped": false, "step_seconds": 1.586332985003537, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 349, "loss": 0.43728920817375183, "loss_ema": 0.227902736805631, "loss_mean_50": 0.1903234025835991, "loss_std_50": 0.2298313122579778, "lr": 0.00025, "grad_norm_before_clip": 0.06689453125, "grad_clipped": false, "step_seconds": 1.8296775339986198, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 350, "loss": 0.043807853013277054, "loss_ema": 0.21869799261601328, "loss_mean_50": 0.18991635091602801, "loss_std_50": 0.23007229439399213, "lr": 0.00025, "grad_norm_before_clip": 0.009033203125, "grad_clipped": false, "step_seconds": 2.0638382750003075, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 351, "loss": 0.214418426156044, "loss_ema": 0.2184840142930148, "loss_mean_50": 0.19249142833054067, "loss_std_50": 0.22961112325104574, "lr": 0.00025, "grad_norm_before_clip": 0.0625, "grad_clipped": false, "step_seconds": 1.8911051959985343, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 352, "loss": 0.06696650385856628, "loss_ema": 0.21090813877129239, "loss_mean_50": 0.19254976339638233, "loss_std_50": 0.22957885216013252, "lr": 0.00025, "grad_norm_before_clip": 0.0167236328125, "grad_clipped": false, "step_seconds": 1.8913247919990681, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 353, "loss": 0.13545972108840942, "loss_ema": 0.20713571788714824, "loss_mean_50": 0.1904366797953844, "loss_std_50": 0.22960836262205556, "lr": 0.00025, "grad_norm_before_clip": 0.01129150390625, "grad_clipped": false, "step_seconds": 1.891562247001275, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 354, "loss": 0.9011448621749878, "loss_ema": 0.2418361751015402, "loss_mean_50": 0.2072444288432598, "loss_std_50": 0.24940586983173282, "lr": 0.00025, "grad_norm_before_clip": 0.279296875, "grad_clipped": false, "step_seconds": 2.0637906650008517, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 355, "loss": 0.3670373857021332, "loss_ema": 0.24809623563156985, "loss_mean_50": 0.21305093318223953, "loss_std_50": 0.2496787431413296, "lr": 0.00025, "grad_norm_before_clip": 0.02197265625, "grad_clipped": false, "step_seconds": 1.8907494900013262, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 356, "loss": 0.054602861404418945, "loss_ema": 0.2384215669202123, "loss_mean_50": 0.2122634294629097, "loss_std_50": 0.2501147804525345, "lr": 0.00025, "grad_norm_before_clip": 0.020751953125, "grad_clipped": false, "step_seconds": 1.830252550997102, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 357, "loss": 0.04698038846254349, "loss_ema": 0.22884950799732884, "loss_mean_50": 0.21192799538373946, "loss_std_50": 0.2503248853251419, "lr": 0.00025, "grad_norm_before_clip": 0.01348876953125, "grad_clipped": false, "step_seconds": 2.0628375460000825, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 358, "loss": 0.07647807896137238, "loss_ema": 0.221230936545531, "loss_mean_50": 0.21001872628927232, "loss_std_50": 0.2509857728226099, "lr": 0.00025, "grad_norm_before_clip": 0.02001953125, "grad_clipped": false, "step_seconds": 0.5978448269997898, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 359, "loss": 0.058151159435510635, "loss_ema": 0.21307694769002997, "loss_mean_50": 0.2091690082103014, "loss_std_50": 0.25142618068171113, "lr": 0.00025, "grad_norm_before_clip": 0.0108642578125, "grad_clipped": false, "step_seconds": 0.47579869300170685, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 360, "loss": 0.08478862792253494, "loss_ema": 0.20666253170165522, "loss_mean_50": 0.2092674145847559, "loss_std_50": 0.2513765120553161, "lr": 0.00025, "grad_norm_before_clip": 0.00799560546875, "grad_clipped": false, "step_seconds": 1.2657213149977906, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 361, "loss": 0.6517403721809387, "loss_ema": 0.2289164237256194, "loss_mean_50": 0.21912751771509648, "loss_std_50": 0.25876148376618624, "lr": 0.00025, "grad_norm_before_clip": 0.0927734375, "grad_clipped": false, "step_seconds": 0.34740922300261445, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 362, "loss": 0.08128394186496735, "loss_ema": 0.2215347996325868, "loss_mean_50": 0.21925881631672384, "loss_std_50": 0.2586898315317228, "lr": 0.00025, "grad_norm_before_clip": 0.043701171875, "grad_clipped": false, "step_seconds": 1.829599379998399, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 363, "loss": 0.11221134662628174, "loss_ema": 0.21606862698227153, "loss_mean_50": 0.22015958495438098, "loss_std_50": 0.25823670989998265, "lr": 0.00025, "grad_norm_before_clip": 0.0184326171875, "grad_clipped": false, "step_seconds": 2.058613401000912, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 364, "loss": 0.09946165233850479, "loss_ema": 0.2102382782500832, "loss_mean_50": 0.22057505913078784, "loss_std_50": 0.25802538835029065, "lr": 0.00025, "grad_norm_before_clip": 0.0126953125, "grad_clipped": false, "step_seconds": 0.6129380029997265, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 365, "loss": 0.06589643657207489, "loss_ema": 0.20302118616618278, "loss_mean_50": 0.20103982843458654, "loss_std_50": 0.23055927443541366, "lr": 0.00025, "grad_norm_before_clip": 0.010986328125, "grad_clipped": false, "step_seconds": 2.059022959001595, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 366, "loss": 0.07413780689239502, "loss_ema": 0.19657701720249338, "loss_mean_50": 0.20109629847109317, "loss_std_50": 0.23052783795561374, "lr": 0.00025, "grad_norm_before_clip": 0.02587890625, "grad_clipped": false, "step_seconds": 0.6676966489976621, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 367, "loss": 0.06236957386136055, "loss_ema": 0.18986664503543674, "loss_mean_50": 0.20046917960047722, "loss_std_50": 0.23086148051210428, "lr": 0.00025, "grad_norm_before_clip": 0.0196533203125, "grad_clipped": false, "step_seconds": 0.5719256550000864, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 368, "loss": 0.06819052249193192, "loss_ema": 0.1837828389082615, "loss_mean_50": 0.20061403729021549, "loss_std_50": 0.23077614662619653, "lr": 0.00025, "grad_norm_before_clip": 0.018310546875, "grad_clipped": false, "step_seconds": 1.889278667000326, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 369, "loss": 0.05994100496172905, "loss_ema": 0.17759074721093485, "loss_mean_50": 0.20027469351887703, "loss_std_50": 0.23097019293041957, "lr": 0.00025, "grad_norm_before_clip": 0.05029296875, "grad_clipped": false, "step_seconds": 0.5209808510007861, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 370, "loss": 0.3331528604030609, "loss_ema": 0.18536885287054114, "loss_mean_50": 0.1985839642584324, "loss_std_50": 0.2296782993136841, "lr": 0.00025, "grad_norm_before_clip": 0.0164794921875, "grad_clipped": false, "step_seconds": 2.0629927549998683, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 371, "loss": 0.051075197756290436, "loss_ema": 0.17865417011482862, "loss_mean_50": 0.19746875807642936, "loss_std_50": 0.230255724415235, "lr": 0.00025, "grad_norm_before_clip": 0.01104736328125, "grad_clipped": false, "step_seconds": 1.5700533119998, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 372, "loss": 0.07702803611755371, "loss_ema": 0.17357286341496486, "loss_mean_50": 0.1970805087685585, "loss_std_50": 0.23044203840898145, "lr": 0.00025, "grad_norm_before_clip": 0.0147705078125, "grad_clipped": false, "step_seconds": 2.06285305900019, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 373, "loss": 0.2558169662952423, "loss_ema": 0.17768506855897873, "loss_mean_50": 0.20088346555829048, "loss_std_50": 0.22981011942128504, "lr": 0.00025, "grad_norm_before_clip": 0.0291748046875, "grad_clipped": false, "step_seconds": 0.7240851140013547, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 374, "loss": 0.1264430582523346, "loss_ema": 0.17512296804364652, "loss_mean_50": 0.20168665289878845, "loss_std_50": 0.22947812792897115, "lr": 0.00025, "grad_norm_before_clip": 0.02734375, "grad_clipped": false, "step_seconds": 0.4200953749968903, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 375, "loss": 0.06421492248773575, "loss_ema": 0.16957756576585098, "loss_mean_50": 0.20169942185282708, "loss_std_50": 0.2294704602819614, "lr": 0.00025, "grad_norm_before_clip": 0.0439453125, "grad_clipped": false, "step_seconds": 0.5972963900021568, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 376, "loss": 0.13587796688079834, "loss_ema": 0.16789258582159836, "loss_mean_50": 0.18531254813075065, "loss_std_50": 0.20277794932913185, "lr": 0.00025, "grad_norm_before_clip": 0.01611328125, "grad_clipped": false, "step_seconds": 2.06457570099883, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 377, "loss": 0.06175665557384491, "loss_ema": 0.16258578930921067, "loss_mean_50": 0.1852049469947815, "loss_std_50": 0.20284204634318306, "lr": 0.00025, "grad_norm_before_clip": 0.01214599609375, "grad_clipped": false, "step_seconds": 2.063203883000824, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 378, "loss": 0.08787408471107483, "loss_ema": 0.15885020407930386, "loss_mean_50": 0.18580406859517098, "loss_std_50": 0.20250916877869485, "lr": 0.00025, "grad_norm_before_clip": 0.040771484375, "grad_clipped": false, "step_seconds": 1.2914996200015594, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 379, "loss": 0.1354045867919922, "loss_ema": 0.1576779232149383, "loss_mean_50": 0.18747405759990216, "loss_std_50": 0.20174091863440824, "lr": 0.00025, "grad_norm_before_clip": 0.01495361328125, "grad_clipped": false, "step_seconds": 2.0644739189992833, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 380, "loss": 0.05541301518678665, "loss_ema": 0.1525646778135307, "loss_mean_50": 0.1874833182245493, "loss_std_50": 0.20173484563220284, "lr": 0.00025, "grad_norm_before_clip": 0.058349609375, "grad_clipped": false, "step_seconds": 2.063950402000046, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 381, "loss": 0.2296503782272339, "loss_ema": 0.15641896283421586, "loss_mean_50": 0.18989870823919774, "loss_std_50": 0.2015021294640278, "lr": 0.00025, "grad_norm_before_clip": 0.0123291015625, "grad_clipped": false, "step_seconds": 1.1031212239977322, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 382, "loss": 0.10678327083587646, "loss_ema": 0.15393717823429887, "loss_mean_50": 0.18667885102331638, "loss_std_50": 0.20151825315653774, "lr": 0.00025, "grad_norm_before_clip": 0.036865234375, "grad_clipped": false, "step_seconds": 1.8900544060015818, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 383, "loss": 0.048648037016391754, "loss_ema": 0.1486727211734035, "loss_mean_50": 0.1861849481612444, "loss_std_50": 0.2018254519171242, "lr": 0.00025, "grad_norm_before_clip": 0.0269775390625, "grad_clipped": false, "step_seconds": 1.294608332998905, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 384, "loss": 0.09086084365844727, "loss_ema": 0.1457821272976557, "loss_mean_50": 0.18637472979724407, "loss_std_50": 0.20173124359406608, "lr": 0.00025, "grad_norm_before_clip": 0.0147705078125, "grad_clipped": false, "step_seconds": 2.0634339429998363, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 385, "loss": 0.10559377819299698, "loss_ema": 0.14377270984242277, "loss_mean_50": 0.18195629425346851, "loss_std_50": 0.2010315753613979, "lr": 0.00025, "grad_norm_before_clip": 0.048095703125, "grad_clipped": false, "step_seconds": 2.0669530560007843, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 386, "loss": 0.20996332168579102, "loss_ema": 0.14708224043459117, "loss_mean_50": 0.1850873851031065, "loss_std_50": 0.20022259902813214, "lr": 0.00025, "grad_norm_before_clip": 0.00946044921875, "grad_clipped": false, "step_seconds": 1.8933944370000972, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 387, "loss": 0.11564221233129501, "loss_ema": 0.14551023902942636, "loss_mean_50": 0.18604264251887798, "loss_std_50": 0.1997745598813094, "lr": 0.00025, "grad_norm_before_clip": 0.07568359375, "grad_clipped": false, "step_seconds": 1.889024932999746, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 388, "loss": 0.16238734126091003, "loss_ema": 0.14635409414100053, "loss_mean_50": 0.17326846294105053, "loss_std_50": 0.1794214760000287, "lr": 0.00025, "grad_norm_before_clip": 0.047119140625, "grad_clipped": false, "step_seconds": 0.5872024400014197, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 389, "loss": 0.06083950027823448, "loss_ema": 0.14207836444786223, "loss_mean_50": 0.17307499125599862, "loss_std_50": 0.17953735181237396, "lr": 0.00025, "grad_norm_before_clip": 0.0089111328125, "grad_clipped": false, "step_seconds": 2.0683238679994247, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 390, "loss": 0.05437934398651123, "loss_ema": 0.1376934134247947, "loss_mean_50": 0.1730986201018095, "loss_std_50": 0.1795216503351742, "lr": 0.00025, "grad_norm_before_clip": 0.0167236328125, "grad_clipped": false, "step_seconds": 1.8932176150010491, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 391, "loss": 0.10540254414081573, "loss_ema": 0.13607886996059573, "loss_mean_50": 0.16341642163693904, "loss_std_50": 0.16958161251087633, "lr": 0.00025, "grad_norm_before_clip": 0.09375, "grad_clipped": false, "step_seconds": 0.5999426809976285, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 392, "loss": 0.050411634147167206, "loss_ema": 0.1317955081699243, "loss_mean_50": 0.15632509358227253, "loss_std_50": 0.16672131755999522, "lr": 0.00025, "grad_norm_before_clip": 0.02197265625, "grad_clipped": false, "step_seconds": 1.5897995960003755, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 393, "loss": 0.06582719087600708, "loss_ema": 0.12849709230522843, "loss_mean_50": 0.154638749435544, "loss_std_50": 0.1672010387537683, "lr": 0.00025, "grad_norm_before_clip": 0.0218505859375, "grad_clipped": false, "step_seconds": 1.891188986999623, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 394, "loss": 0.11445506662130356, "loss_ema": 0.12779499102103217, "loss_mean_50": 0.15338071934878827, "loss_std_50": 0.16726200195990346, "lr": 0.00025, "grad_norm_before_clip": 0.09521484375, "grad_clipped": false, "step_seconds": 0.574369451001985, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 395, "loss": 0.19862329959869385, "loss_ema": 0.13133640644991523, "loss_mean_50": 0.15526488058269025, "loss_std_50": 0.16723041795669108, "lr": 0.00025, "grad_norm_before_clip": 0.01251220703125, "grad_clipped": false, "step_seconds": 1.829543312000169, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 396, "loss": 0.36259278655052185, "loss_ema": 0.14289922545494554, "loss_mean_50": 0.15783155851066114, "loss_std_50": 0.16939398157929675, "lr": 0.00025, "grad_norm_before_clip": 0.01123046875, "grad_clipped": false, "step_seconds": 2.065931298999203, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 397, "loss": 0.0705532655119896, "loss_ema": 0.13928192745779774, "loss_mean_50": 0.15749707870185375, "loss_std_50": 0.16954940558707393, "lr": 0.00025, "grad_norm_before_clip": 0.0166015625, "grad_clipped": false, "step_seconds": 2.061776414000633, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 398, "loss": 0.11842869222164154, "loss_ema": 0.13823926569598993, "loss_mean_50": 0.14834915183484554, "loss_std_50": 0.1587257936046483, "lr": 0.00025, "grad_norm_before_clip": 0.00860595703125, "grad_clipped": false, "step_seconds": 1.5702721259985992, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 399, "loss": 1.0280557870864868, "loss_ema": 0.18273009176551477, "loss_mean_50": 0.16016448341310024, "loss_std_50": 0.1971350391333159, "lr": 0.00025, "grad_norm_before_clip": 0.1982421875, "grad_clipped": false, "step_seconds": 0.6684030659998825, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 400, "loss": 0.06926389783620834, "loss_ema": 0.17705678206904943, "loss_mean_50": 0.16067360430955888, "loss_std_50": 0.19686656768078017, "lr": 0.00025, "grad_norm_before_clip": 0.0186767578125, "grad_clipped": false, "step_seconds": 0.8975996389999636, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 401, "loss": 0.10054569691419601, "loss_ema": 0.17323122781130676, "loss_mean_50": 0.1583961497247219, "loss_std_50": 0.19689031428303908, "lr": 0.00025, "grad_norm_before_clip": 0.007354736328125, "grad_clipped": false, "step_seconds": 1.2989250659993559, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 402, "loss": 0.06225338950753212, "loss_ema": 0.16768233589611803, "loss_mean_50": 0.15830188743770124, "loss_std_50": 0.19693518724970036, "lr": 0.00025, "grad_norm_before_clip": 0.0093994140625, "grad_clipped": false, "step_seconds": 2.071300981002423, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 403, "loss": 0.23838520050048828, "loss_ema": 0.17121747912633653, "loss_mean_50": 0.1603603970259428, "loss_std_50": 0.19722338087737296, "lr": 0.00025, "grad_norm_before_clip": 0.0142822265625, "grad_clipped": false, "step_seconds": 2.062610572997073, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 404, "loss": 0.06447064876556396, "loss_ema": 0.1658801376082979, "loss_mean_50": 0.14362691275775433, "loss_std_50": 0.16681042194435353, "lr": 0.00025, "grad_norm_before_clip": 0.0111083984375, "grad_clipped": false, "step_seconds": 2.0683199480008625, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 405, "loss": 0.06860361248254776, "loss_ema": 0.1610163113520104, "loss_mean_50": 0.13765823729336263, "loss_std_50": 0.16402565929059115, "lr": 0.00025, "grad_norm_before_clip": 0.0076904296875, "grad_clipped": false, "step_seconds": 2.067398246999801, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 406, "loss": 0.38585028052330017, "loss_ema": 0.17225800981057487, "loss_mean_50": 0.14428318567574025, "loss_std_50": 0.16719613889625523, "lr": 0.00025, "grad_norm_before_clip": 0.0269775390625, "grad_clipped": false, "step_seconds": 2.066121029001806, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 407, "loss": 0.1229015365242958, "loss_ema": 0.1697901861462609, "loss_mean_50": 0.14580160863697528, "loss_std_50": 0.16664942263633892, "lr": 0.00025, "grad_norm_before_clip": 0.0537109375, "grad_clipped": false, "step_seconds": 1.830790533000254, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 408, "loss": 0.805134654045105, "loss_ema": 0.2015574095412031, "loss_mean_50": 0.16037474013864994, "loss_std_50": 0.19015241298303187, "lr": 0.00025, "grad_norm_before_clip": 0.3203125, "grad_clipped": false, "step_seconds": 1.265676352999435, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 409, "loss": 0.3105796277523041, "loss_ema": 0.20700852045175816, "loss_mean_50": 0.16542330950498582, "loss_std_50": 0.19072149607961864, "lr": 0.00025, "grad_norm_before_clip": 0.03564453125, "grad_clipped": false, "step_seconds": 0.34758868900098605, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 410, "loss": 0.13894227147102356, "loss_ema": 0.20360520800272142, "loss_mean_50": 0.16650638237595558, "loss_std_50": 0.19041402755583217, "lr": 0.00025, "grad_norm_before_clip": 0.018798828125, "grad_clipped": false, "step_seconds": 0.6006689179994282, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 411, "loss": 0.05597091093659401, "loss_ema": 0.19622349314941503, "loss_mean_50": 0.1545909931510687, "loss_std_50": 0.17790684838234264, "lr": 0.00025, "grad_norm_before_clip": 0.053955078125, "grad_clipped": false, "step_seconds": 1.2935387369980162, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 412, "loss": 0.42458704113960266, "loss_ema": 0.2076416705489244, "loss_mean_50": 0.1614570551365614, "loss_std_50": 0.18153286957841133, "lr": 0.00025, "grad_norm_before_clip": 0.07958984375, "grad_clipped": false, "step_seconds": 1.8323786729997664, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 413, "loss": 0.08053422719240189, "loss_ema": 0.20128629838109827, "loss_mean_50": 0.16082351274788378, "loss_std_50": 0.1817587649796572, "lr": 0.00025, "grad_norm_before_clip": 0.01153564453125, "grad_clipped": false, "step_seconds": 1.8944239640004525, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 414, "loss": 0.1232815757393837, "loss_ema": 0.1973860622490125, "loss_mean_50": 0.16129991121590137, "loss_std_50": 0.18162847814662472, "lr": 0.00025, "grad_norm_before_clip": 0.008544921875, "grad_clipped": false, "step_seconds": 2.066679332001513, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 415, "loss": 0.5953907370567322, "loss_ema": 0.2172862959893985, "loss_mean_50": 0.17188979722559453, "loss_std_50": 0.19095397044906354, "lr": 0.00025, "grad_norm_before_clip": 0.0498046875, "grad_clipped": false, "step_seconds": 1.8989160749988514, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 416, "loss": 0.5254871845245361, "loss_ema": 0.2326963404161554, "loss_mean_50": 0.18091678477823733, "loss_std_50": 0.19670141214489786, "lr": 0.00025, "grad_norm_before_clip": 0.01611328125, "grad_clipped": false, "step_seconds": 1.8937568319997808, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 417, "loss": 0.0750700831413269, "loss_ema": 0.22481502755241395, "loss_mean_50": 0.18117079496383667, "loss_std_50": 0.19655630918199352, "lr": 0.00025, "grad_norm_before_clip": 0.046875, "grad_clipped": false, "step_seconds": 1.8900063239998417, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 418, "loss": 0.09190915524959564, "loss_ema": 0.218169733937273, "loss_mean_50": 0.18164516761898994, "loss_std_50": 0.19631153714686372, "lr": 0.00025, "grad_norm_before_clip": 0.02197265625, "grad_clipped": false, "step_seconds": 0.3631880729990371, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 419, "loss": 0.07286716997623444, "loss_ema": 0.21090460573922107, "loss_mean_50": 0.18190369091928005, "loss_std_50": 0.1961595467315545, "lr": 0.00025, "grad_norm_before_clip": 0.048583984375, "grad_clipped": false, "step_seconds": 0.6164466490008635, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 420, "loss": 0.055849168449640274, "loss_ema": 0.20315183387474203, "loss_mean_50": 0.17635761708021164, "loss_std_50": 0.19572449378197615, "lr": 0.00025, "grad_norm_before_clip": 0.0113525390625, "grad_clipped": false, "step_seconds": 2.0669367500013323, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 421, "loss": 0.08902985602617264, "loss_ema": 0.19744573498231355, "loss_mean_50": 0.1771167102456093, "loss_std_50": 0.1953102923784566, "lr": 0.00025, "grad_norm_before_clip": 0.05029296875, "grad_clipped": false, "step_seconds": 1.8942814430010912, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 422, "loss": 0.06542053818702698, "loss_ema": 0.1908444751425492, "loss_mean_50": 0.17688456028699875, "loss_std_50": 0.1954359799490142, "lr": 0.00025, "grad_norm_before_clip": 0.006011962890625, "grad_clipped": false, "step_seconds": 2.067636528998264, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 423, "loss": 0.0601595900952816, "loss_ema": 0.18431023089018583, "loss_mean_50": 0.17297141276299954, "loss_std_50": 0.1957748641017203, "lr": 0.00025, "grad_norm_before_clip": 0.037109375, "grad_clipped": false, "step_seconds": 2.0656955000013113, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 424, "loss": 0.06413145363330841, "loss_ema": 0.17830129202734193, "loss_mean_50": 0.171725180670619, "loss_std_50": 0.1962647938327621, "lr": 0.00025, "grad_norm_before_clip": 0.00909423828125, "grad_clipped": false, "step_seconds": 1.103747643999668, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 425, "loss": 0.11943270266056061, "loss_ema": 0.17535786255900288, "loss_mean_50": 0.17282953627407552, "loss_std_50": 0.19581156924389168, "lr": 0.00025, "grad_norm_before_clip": 0.055419921875, "grad_clipped": false, "step_seconds": 0.7272998299995379, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 426, "loss": 0.5169261693954468, "loss_ema": 0.19243627790082507, "loss_mean_50": 0.1804505003243685, "loss_std_50": 0.20155602954413512, "lr": 0.00025, "grad_norm_before_clip": 0.0419921875, "grad_clipped": false, "step_seconds": 1.893994644000486, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 427, "loss": 0.2003888487815857, "loss_ema": 0.1928339064448631, "loss_mean_50": 0.18322314418852328, "loss_std_50": 0.20085649521673338, "lr": 0.00025, "grad_norm_before_clip": 0.0211181640625, "grad_clipped": false, "step_seconds": 2.0670457179985533, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 428, "loss": 0.054981160908937454, "loss_ema": 0.1859412691680668, "loss_mean_50": 0.18256528571248054, "loss_std_50": 0.20122124677980724, "lr": 0.00025, "grad_norm_before_clip": 0.03466796875, "grad_clipped": false, "step_seconds": 2.0634106430006796, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 429, "loss": 0.1611914038658142, "loss_ema": 0.18470377590295417, "loss_mean_50": 0.183081022053957, "loss_std_50": 0.2011327382543595, "lr": 0.00025, "grad_norm_before_clip": 0.034423828125, "grad_clipped": false, "step_seconds": 0.42044285500014666, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 430, "loss": 0.09824853390455246, "loss_ema": 0.1803810138030341, "loss_mean_50": 0.1839377324283123, "loss_std_50": 0.20067783401859515, "lr": 0.00025, "grad_norm_before_clip": 0.015380859375, "grad_clipped": false, "step_seconds": 0.474496636998083, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 431, "loss": 0.13160015642642975, "loss_ema": 0.17794197093420386, "loss_mean_50": 0.18197672799229622, "loss_std_50": 0.2007006204412758, "lr": 0.00025, "grad_norm_before_clip": 0.0157470703125, "grad_clipped": false, "step_seconds": 0.5798682900021959, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 432, "loss": 0.11051030457019806, "loss_ema": 0.17457038761600358, "loss_mean_50": 0.18205126866698265, "loss_std_50": 0.20067336983953082, "lr": 0.00025, "grad_norm_before_clip": 0.035400390625, "grad_clipped": false, "step_seconds": 1.8295626170001924, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 433, "loss": 0.04641300439834595, "loss_ema": 0.16816251845512067, "loss_mean_50": 0.18200656801462173, "loss_std_50": 0.2007033275633081, "lr": 0.00025, "grad_norm_before_clip": 0.0169677734375, "grad_clipped": false, "step_seconds": 2.06296735000069, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 434, "loss": 0.46140626072883606, "loss_ema": 0.18282470556880642, "loss_mean_50": 0.1894174763560295, "loss_std_50": 0.2040147961226036, "lr": 0.00025, "grad_norm_before_clip": 0.052001953125, "grad_clipped": false, "step_seconds": 0.7273422899997968, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 435, "loss": 0.10096569359302521, "loss_ema": 0.17873175497001736, "loss_mean_50": 0.18932491466403006, "loss_std_50": 0.20405385215497698, "lr": 0.00025, "grad_norm_before_clip": 0.0096435546875, "grad_clipped": false, "step_seconds": 1.893880719999288, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 436, "loss": 0.11524701863527298, "loss_ema": 0.17555751815328013, "loss_mean_50": 0.18743058860301973, "loss_std_50": 0.2042929708158673, "lr": 0.00025, "grad_norm_before_clip": 0.00732421875, "grad_clipped": false, "step_seconds": 2.063667539998278, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 437, "loss": 0.12916496396064758, "loss_ema": 0.17323789044364848, "loss_mean_50": 0.18770104363560677, "loss_std_50": 0.20420668701047198, "lr": 0.00025, "grad_norm_before_clip": 0.013916015625, "grad_clipped": false, "step_seconds": 0.6714573200006271, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 438, "loss": 0.7505527138710022, "loss_ema": 0.20210363161501615, "loss_mean_50": 0.1994643510878086, "loss_std_50": 0.21882691790717848, "lr": 0.00025, "grad_norm_before_clip": 0.357421875, "grad_clipped": false, "step_seconds": 1.8313753850015928, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 439, "loss": 0.04972861334681511, "loss_ema": 0.19448488070160608, "loss_mean_50": 0.19924213334918023, "loss_std_50": 0.21897317063496366, "lr": 0.00025, "grad_norm_before_clip": 0.0107421875, "grad_clipped": false, "step_seconds": 1.5901736399973743, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 440, "loss": 0.685283899307251, "loss_ema": 0.2190248316318883, "loss_mean_50": 0.21186022445559502, "loss_std_50": 0.22824342410030848, "lr": 0.00025, "grad_norm_before_clip": 0.028076171875, "grad_clipped": false, "step_seconds": 2.0657596950004518, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 441, "loss": 0.05734796077013016, "loss_ema": 0.21094098808880038, "loss_mean_50": 0.21089913278818131, "loss_std_50": 0.2287901942533031, "lr": 0.00025, "grad_norm_before_clip": 0.0087890625, "grad_clipped": false, "step_seconds": 1.111731981000048, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 442, "loss": 0.09608245640993118, "loss_ema": 0.2051980615048569, "loss_mean_50": 0.2118125492334366, "loss_std_50": 0.2282381460140957, "lr": 0.00025, "grad_norm_before_clip": 0.0291748046875, "grad_clipped": false, "step_seconds": 1.8923047429998405, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 443, "loss": 0.04736277833580971, "loss_ema": 0.19730629734640454, "loss_mean_50": 0.21144326098263264, "loss_std_50": 0.22848885086150306, "lr": 0.00025, "grad_norm_before_clip": 0.033203125, "grad_clipped": false, "step_seconds": 1.2731512840000505, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 444, "loss": 0.05921650305390358, "loss_ema": 0.1904018076317795, "loss_mean_50": 0.21033848971128463, "loss_std_50": 0.22908788699564325, "lr": 0.00025, "grad_norm_before_clip": 0.0244140625, "grad_clipped": false, "step_seconds": 2.0633298290013045, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 445, "loss": 0.25644221901893616, "loss_ema": 0.19370382820113732, "loss_mean_50": 0.21149486809968948, "loss_std_50": 0.22917174546764138, "lr": 0.00025, "grad_norm_before_clip": 0.044677734375, "grad_clipped": false, "step_seconds": 0.4794246560013562, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 446, "loss": 0.07056401669979095, "loss_ema": 0.18754683762607, "loss_mean_50": 0.20565429270267488, "loss_std_50": 0.22896767256108388, "lr": 0.00025, "grad_norm_before_clip": 0.01336669921875, "grad_clipped": false, "step_seconds": 1.8938394749966392, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 447, "loss": 0.38620665669441223, "loss_ema": 0.1974798285794871, "loss_mean_50": 0.2119673605263233, "loss_std_50": 0.22950659603722565, "lr": 0.00025, "grad_norm_before_clip": 0.054931640625, "grad_clipped": false, "step_seconds": 2.062519399001758, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 448, "loss": 0.20579929649829865, "loss_ema": 0.19789580197542767, "loss_mean_50": 0.21371477261185645, "loss_std_50": 0.22912004588536206, "lr": 0.00025, "grad_norm_before_clip": 0.028564453125, "grad_clipped": false, "step_seconds": 0.36057782900024904, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 449, "loss": 0.10917405784130096, "loss_ema": 0.1934597147687213, "loss_mean_50": 0.19533713802695274, "loss_std_50": 0.19777210985717258, "lr": 0.00025, "grad_norm_before_clip": 0.060791015625, "grad_clipped": false, "step_seconds": 0.35104415900059394, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 450, "loss": 0.1008865088224411, "loss_ema": 0.18883105447140727, "loss_mean_50": 0.1959695902466774, "loss_std_50": 0.19741817705278814, "lr": 0.00025, "grad_norm_before_clip": 0.007293701171875, "grad_clipped": false, "step_seconds": 2.0637226260005264, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 451, "loss": 0.08133473247289658, "loss_ema": 0.18345623837148173, "loss_mean_50": 0.19558537095785142, "loss_std_50": 0.19762210815175973, "lr": 0.00025, "grad_norm_before_clip": 0.10595703125, "grad_clipped": false, "step_seconds": 0.6157994639979734, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 452, "loss": 0.06616106629371643, "loss_ema": 0.17759147976759346, "loss_mean_50": 0.1956635244935751, "loss_std_50": 0.19757012980098282, "lr": 0.00025, "grad_norm_before_clip": 0.00946044921875, "grad_clipped": false, "step_seconds": 1.8936502409997047, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 453, "loss": 0.0592203252017498, "loss_ema": 0.17167292203930126, "loss_mean_50": 0.19208022698760033, "loss_std_50": 0.19838585702918407, "lr": 0.00025, "grad_norm_before_clip": 0.0147705078125, "grad_clipped": false, "step_seconds": 1.8934082810010295, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 454, "loss": 0.06374756991863251, "loss_ema": 0.1662766544332678, "loss_mean_50": 0.1920657654106617, "loss_std_50": 0.19839518489215952, "lr": 0.00025, "grad_norm_before_clip": 0.0147705078125, "grad_clipped": false, "step_seconds": 1.8294964179985982, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 455, "loss": 0.04454779252409935, "loss_ema": 0.1601902113378094, "loss_mean_50": 0.19158464901149272, "loss_std_50": 0.1987228997820405, "lr": 0.00025, "grad_norm_before_clip": 0.040283203125, "grad_clipped": false, "step_seconds": 0.6022769789997255, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 456, "loss": 0.2002137005329132, "loss_ema": 0.16219138579756456, "loss_mean_50": 0.187871917411685, "loss_std_50": 0.19678341652873158, "lr": 0.00025, "grad_norm_before_clip": 0.0130615234375, "grad_clipped": false, "step_seconds": 2.066692487002001, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 457, "loss": 0.08123035728931427, "loss_ema": 0.15814333437215203, "loss_mean_50": 0.18703849382698537, "loss_std_50": 0.19714472821991044, "lr": 0.00025, "grad_norm_before_clip": 0.026123046875, "grad_clipped": false, "step_seconds": 1.8978780270008428, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 458, "loss": 0.04616814851760864, "loss_ema": 0.15254457507942484, "loss_mean_50": 0.17185916371643542, "loss_std_50": 0.17717692744971422, "lr": 0.00025, "grad_norm_before_clip": 0.018798828125, "grad_clipped": false, "step_seconds": 1.3757660449991818, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 459, "loss": 0.0504286028444767, "loss_ema": 0.14743877646767742, "loss_mean_50": 0.16665614321827887, "loss_std_50": 0.17684635274057559, "lr": 0.00025, "grad_norm_before_clip": 0.007568359375, "grad_clipped": false, "step_seconds": 2.0671873629980837, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 460, "loss": 0.08766774088144302, "loss_ema": 0.14445022468836569, "loss_mean_50": 0.16563065260648727, "loss_std_50": 0.1771524852407645, "lr": 0.00025, "grad_norm_before_clip": 0.01806640625, "grad_clipped": false, "step_seconds": 1.8944191990012769, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 461, "loss": 0.07546475529670715, "loss_ema": 0.14100095121878276, "loss_mean_50": 0.16602052949368953, "loss_std_50": 0.17693203108788275, "lr": 0.00025, "grad_norm_before_clip": 0.01263427734375, "grad_clipped": false, "step_seconds": 2.066928240001289, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 462, "loss": 0.043453868478536606, "loss_ema": 0.13612359708177046, "loss_mean_50": 0.1583978660404682, "loss_std_50": 0.17381069453780962, "lr": 0.00025, "grad_norm_before_clip": 0.0322265625, "grad_clipped": false, "step_seconds": 2.0641126959999383, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 463, "loss": 0.12077949941158295, "loss_ema": 0.1353563921982611, "loss_mean_50": 0.15920277148485185, "loss_std_50": 0.17354122728621182, "lr": 0.00025, "grad_norm_before_clip": 0.07080078125, "grad_clipped": false, "step_seconds": 0.42422244800036424, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 464, "loss": 0.05018895864486694, "loss_ema": 0.13109802052059136, "loss_mean_50": 0.1577409191429615, "loss_std_50": 0.17414446341316075, "lr": 0.00025, "grad_norm_before_clip": 0.04833984375, "grad_clipped": false, "step_seconds": 2.0724827429985453, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 465, "loss": 0.04644867405295372, "loss_ema": 0.12686555319720946, "loss_mean_50": 0.14676207788288592, "loss_std_50": 0.16316473637397297, "lr": 0.00025, "grad_norm_before_clip": 0.02001953125, "grad_clipped": false, "step_seconds": 2.0672712349987705, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 466, "loss": 0.054421670734882355, "loss_ema": 0.1232433590740931, "loss_mean_50": 0.13734076760709285, "loss_std_50": 0.15438863622853186, "lr": 0.00025, "grad_norm_before_clip": 0.035400390625, "grad_clipped": false, "step_seconds": 2.062050773001829, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 467, "loss": 0.14989393949508667, "loss_ema": 0.12457588809514276, "loss_mean_50": 0.13883724473416806, "loss_std_50": 0.15414022976603597, "lr": 0.00025, "grad_norm_before_clip": 0.010498046875, "grad_clipped": false, "step_seconds": 1.569794789000298, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 468, "loss": 0.08919969946146011, "loss_ema": 0.12280707866345864, "loss_mean_50": 0.13878305561840534, "loss_std_50": 0.1541571934819347, "lr": 0.00025, "grad_norm_before_clip": 0.021484375, "grad_clipped": false, "step_seconds": 0.5243113619981159, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 469, "loss": 0.7832806706428528, "loss_ema": 0.15583075826242834, "loss_mean_50": 0.1529913256317377, "loss_std_50": 0.17827846421062496, "lr": 0.00025, "grad_norm_before_clip": 0.048828125, "grad_clipped": false, "step_seconds": 2.058454204001464, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 470, "loss": 0.1573411077260971, "loss_ema": 0.15590627573561178, "loss_mean_50": 0.15502116441726685, "loss_std_50": 0.1777378321963261, "lr": 0.00025, "grad_norm_before_clip": 0.0208740234375, "grad_clipped": false, "step_seconds": 0.57314045000021, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 471, "loss": 0.08233189582824707, "loss_ema": 0.15222755674024355, "loss_mean_50": 0.15488720521330834, "loss_std_50": 0.1777900351214258, "lr": 0.00025, "grad_norm_before_clip": 0.01104736328125, "grad_clipped": false, "step_seconds": 1.2910781399987172, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 472, "loss": 0.0850844755768776, "loss_ema": 0.14887040268207524, "loss_mean_50": 0.15528048396110536, "loss_std_50": 0.1776133571875225, "lr": 0.00025, "grad_norm_before_clip": 0.0233154296875, "grad_clipped": false, "step_seconds": 0.6023086400018656, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 473, "loss": 0.0792771503329277, "loss_ema": 0.14539074006461786, "loss_mean_50": 0.15566283516585827, "loss_std_50": 0.17742865863336155, "lr": 0.00025, "grad_norm_before_clip": 0.06689453125, "grad_clipped": false, "step_seconds": 1.8888993090004078, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 474, "loss": 0.05759251490235329, "loss_ema": 0.1410008288065046, "loss_mean_50": 0.15553205639123915, "loss_std_50": 0.17749847233025032, "lr": 0.00025, "grad_norm_before_clip": 0.042724609375, "grad_clipped": false, "step_seconds": 0.5790920039980847, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 475, "loss": 0.15872152149677277, "loss_ema": 0.14188686344101803, "loss_mean_50": 0.1563178327679634, "loss_std_50": 0.17742387218752764, "lr": 0.00025, "grad_norm_before_clip": 0.04150390625, "grad_clipped": false, "step_seconds": 0.7242940319993068, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 476, "loss": 0.19299983978271484, "loss_ema": 0.14444251225810287, "loss_mean_50": 0.14983930617570876, "loss_std_50": 0.16989232784053965, "lr": 0.00025, "grad_norm_before_clip": 0.01153564453125, "grad_clipped": false, "step_seconds": 0.6717754560013418, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 477, "loss": 0.06503615528345108, "loss_ema": 0.14047219440937028, "loss_mean_50": 0.1471322523057461, "loss_std_50": 0.17014347263793092, "lr": 0.00025, "grad_norm_before_clip": 0.0439453125, "grad_clipped": false, "step_seconds": 1.8893247320011142, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 478, "loss": 0.24479380249977112, "loss_ema": 0.1456882748138903, "loss_mean_50": 0.15092850513756276, "loss_std_50": 0.17016259530385877, "lr": 0.00025, "grad_norm_before_clip": 0.09814453125, "grad_clipped": false, "step_seconds": 0.42399097299858113, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 479, "loss": 0.12194106727838516, "loss_ema": 0.14450091443711502, "loss_mean_50": 0.15014349840581417, "loss_std_50": 0.1702039702658037, "lr": 0.00025, "grad_norm_before_clip": 0.01123046875, "grad_clipped": false, "step_seconds": 2.0621349460016063, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 480, "loss": 0.20280401408672333, "loss_ema": 0.14741606941959542, "loss_mean_50": 0.1522346080094576, "loss_std_50": 0.1701958272293771, "lr": 0.00025, "grad_norm_before_clip": 0.05908203125, "grad_clipped": false, "step_seconds": 0.3604790769968531, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 481, "loss": 0.08225607126951218, "loss_ema": 0.14415806951209126, "loss_mean_50": 0.15124772630631925, "loss_std_50": 0.1704554779731019, "lr": 0.00025, "grad_norm_before_clip": 0.0625, "grad_clipped": false, "step_seconds": 0.6023321700013184, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 482, "loss": 0.3068547546863556, "loss_ema": 0.15229290377080446, "loss_mean_50": 0.15517461530864238, "loss_std_50": 0.17172865139176174, "lr": 0.00025, "grad_norm_before_clip": 0.01104736328125, "grad_clipped": false, "step_seconds": 2.0643269360007253, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 483, "loss": 0.09611580520868301, "loss_ema": 0.14948404884269836, "loss_mean_50": 0.15616867132484913, "loss_std_50": 0.1712393607273815, "lr": 0.00025, "grad_norm_before_clip": 0.0296630859375, "grad_clipped": false, "step_seconds": 2.058089869999094, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 484, "loss": 0.10792915523052216, "loss_ema": 0.14740630416208955, "loss_mean_50": 0.14909912921488286, "loss_std_50": 0.16569876745237078, "lr": 0.00025, "grad_norm_before_clip": 0.03271484375, "grad_clipped": false, "step_seconds": 0.35100532000069506, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 485, "loss": 0.04715067520737648, "loss_ema": 0.14239352271435388, "loss_mean_50": 0.14802282884716988, "loss_std_50": 0.1661819975380861, "lr": 0.00025, "grad_norm_before_clip": 0.01324462890625, "grad_clipped": false, "step_seconds": 2.0625501590002386, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 486, "loss": 0.08455357700586319, "loss_ema": 0.13950152542892932, "loss_mean_50": 0.14740896001458167, "loss_std_50": 0.16635853243729284, "lr": 0.00025, "grad_norm_before_clip": 0.01519775390625, "grad_clipped": false, "step_seconds": 0.6133860369991453, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 487, "loss": 0.04932572692632675, "loss_ema": 0.1349927355037992, "loss_mean_50": 0.14581217527389526, "loss_std_50": 0.16690824167069254, "lr": 0.00025, "grad_norm_before_clip": 0.0091552734375, "grad_clipped": false, "step_seconds": 2.0666033170018636, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 488, "loss": 0.058754969388246536, "loss_ema": 0.13118084719802156, "loss_mean_50": 0.13197622038424014, "loss_std_50": 0.14319317110592836, "lr": 0.00025, "grad_norm_before_clip": 0.01458740234375, "grad_clipped": false, "step_seconds": 1.8920842649968108, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 489, "loss": 0.0683843344449997, "loss_ema": 0.12804102156037045, "loss_mean_50": 0.13234933480620384, "loss_std_50": 0.14300255319006838, "lr": 0.00025, "grad_norm_before_clip": 0.02978515625, "grad_clipped": false, "step_seconds": 1.2668510770017747, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 490, "loss": 0.057179391384124756, "loss_ema": 0.12449794005155816, "loss_mean_50": 0.11978724464774132, "loss_std_50": 0.11954163235129138, "lr": 0.00025, "grad_norm_before_clip": 0.056884765625, "grad_clipped": false, "step_seconds": 1.8902665230016282, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 491, "loss": 0.05331956595182419, "loss_ema": 0.12093902134657146, "loss_mean_50": 0.1197066767513752, "loss_std_50": 0.11958503726146881, "lr": 0.00025, "grad_norm_before_clip": 0.012451171875, "grad_clipped": false, "step_seconds": 2.0633923030000005, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 492, "loss": 0.06372296065092087, "loss_ema": 0.11807821831178893, "loss_mean_50": 0.119059486836195, "loss_std_50": 0.1197985130778929, "lr": 0.00025, "grad_norm_before_clip": 0.008544921875, "grad_clipped": false, "step_seconds": 2.0621759419991577, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 493, "loss": 0.13120713829994202, "loss_ema": 0.11873466431119657, "loss_mean_50": 0.12073637403547764, "loss_std_50": 0.11936923696464205, "lr": 0.00025, "grad_norm_before_clip": 0.0123291015625, "grad_clipped": false, "step_seconds": 2.062480571999913, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 494, "loss": 0.044942706823349, "loss_ema": 0.11504506643680419, "loss_mean_50": 0.12045089811086655, "loss_std_50": 0.11953297848834356, "lr": 0.00025, "grad_norm_before_clip": 0.011962890625, "grad_clipped": false, "step_seconds": 0.747906618998968, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 495, "loss": 0.0480748675763607, "loss_ema": 0.111696556493782, "loss_mean_50": 0.11628355108201503, "loss_std_50": 0.11834550761946196, "lr": 0.00025, "grad_norm_before_clip": 0.0252685546875, "grad_clipped": false, "step_seconds": 2.063716184999066, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 496, "loss": 0.2518722414970398, "loss_ema": 0.11870534074394488, "loss_mean_50": 0.11990971557796001, "loss_std_50": 0.11965947704901252, "lr": 0.00025, "grad_norm_before_clip": 0.017822265625, "grad_clipped": false, "step_seconds": 1.890274297002179, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 497, "loss": 0.06312262266874313, "loss_ema": 0.11592620484018479, "loss_mean_50": 0.11344803489744663, "loss_std_50": 0.11367872004860125, "lr": 0.00025, "grad_norm_before_clip": 0.040771484375, "grad_clipped": false, "step_seconds": 1.570099508997373, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 498, "loss": 0.3188587427139282, "loss_ema": 0.12607283173387196, "loss_mean_50": 0.11570922382175923, "loss_std_50": 0.11658059222642848, "lr": 0.00025, "grad_norm_before_clip": 0.022216796875, "grad_clipped": false, "step_seconds": 1.2900037070030521, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 499, "loss": 0.12333541363477707, "loss_ema": 0.1259359608289172, "loss_mean_50": 0.11599245093762875, "loss_std_50": 0.11658157347453788, "lr": 0.00025, "grad_norm_before_clip": 0.109375, "grad_clipped": false, "step_seconds": 0.4764515629976813, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 500, "loss": 0.0639076754450798, "loss_ema": 0.12283454655972532, "loss_mean_50": 0.11525287427008152, "loss_std_50": 0.11679216149712415, "lr": 0.00025, "grad_norm_before_clip": 0.00921630859375, "grad_clipped": false, "step_seconds": 2.066896120002639, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 501, "loss": 0.04379449039697647, "loss_ema": 0.11888254375158787, "loss_mean_50": 0.11450206942856311, "loss_std_50": 0.11712797481778607, "lr": 0.00025, "grad_norm_before_clip": 0.01556396484375, "grad_clipped": false, "step_seconds": 1.828813383999659, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 502, "loss": 0.04958606883883476, "loss_ema": 0.1154177200059502, "loss_mean_50": 0.11417056947946548, "loss_std_50": 0.11728766893271142, "lr": 0.00025, "grad_norm_before_clip": 0.014404296875, "grad_clipped": false, "step_seconds": 2.063779202999285, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 503, "loss": 0.14613378047943115, "loss_ema": 0.11695352302962424, "loss_mean_50": 0.11590883858501912, "loss_std_50": 0.1171043041668877, "lr": 0.00025, "grad_norm_before_clip": 0.01214599609375, "grad_clipped": false, "step_seconds": 1.828112916999089, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 504, "loss": 0.09662916511297226, "loss_ema": 0.11593730513379163, "loss_mean_50": 0.11656647048890591, "loss_std_50": 0.11690168404217272, "lr": 0.00025, "grad_norm_before_clip": 0.0146484375, "grad_clipped": false, "step_seconds": 1.29403097300019, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 505, "loss": 0.12791812419891357, "loss_ema": 0.11653634608704773, "loss_mean_50": 0.11823387712240219, "loss_std_50": 0.11645628745730142, "lr": 0.00025, "grad_norm_before_clip": 0.0086669921875, "grad_clipped": false, "step_seconds": 1.8321882009986439, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 506, "loss": 0.20649729669094086, "loss_ema": 0.12103439361724239, "loss_mean_50": 0.11835954904556274, "loss_std_50": 0.1165480411274922, "lr": 0.00025, "grad_norm_before_clip": 0.0093994140625, "grad_clipped": false, "step_seconds": 1.5950780840030347, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 507, "loss": 0.043544284999370575, "loss_ema": 0.11715988818634879, "loss_mean_50": 0.11760582759976387, "loss_std_50": 0.11690702601518158, "lr": 0.00025, "grad_norm_before_clip": 0.0159912109375, "grad_clipped": false, "step_seconds": 1.8902760589990066, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 508, "loss": 0.07475839555263519, "loss_ema": 0.1150398135546631, "loss_mean_50": 0.1181776325404644, "loss_std_50": 0.11662579894531011, "lr": 0.00025, "grad_norm_before_clip": 0.057373046875, "grad_clipped": false, "step_seconds": 1.8904152879986214, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 509, "loss": 0.07808394730091095, "loss_ema": 0.11319202024197549, "loss_mean_50": 0.11873073942959309, "loss_std_50": 0.11636847727725884, "lr": 0.00025, "grad_norm_before_clip": 0.0269775390625, "grad_clipped": false, "step_seconds": 2.062150701000064, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 510, "loss": 0.26248037815093994, "loss_ema": 0.1206564381374237, "loss_mean_50": 0.12222699217498302, "loss_std_50": 0.11799737156243173, "lr": 0.00025, "grad_norm_before_clip": 0.019775390625, "grad_clipped": false, "step_seconds": 1.8896365990003687, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 511, "loss": 0.09503113478422165, "loss_ema": 0.11937517296976359, "loss_mean_50": 0.12261831976473331, "loss_std_50": 0.11787402052864424, "lr": 0.00025, "grad_norm_before_clip": 0.0311279296875, "grad_clipped": false, "step_seconds": 1.8904616049985634, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 512, "loss": 0.0482683926820755, "loss_ema": 0.11581983395537918, "loss_mean_50": 0.12271461024880409, "loss_std_50": 0.1178112620674482, "lr": 0.00025, "grad_norm_before_clip": 0.00738525390625, "grad_clipped": false, "step_seconds": 2.064540292001766, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 513, "loss": 0.055298615247011185, "loss_ema": 0.11279377301996077, "loss_mean_50": 0.12140499256551265, "loss_std_50": 0.11818883960007036, "lr": 0.00025, "grad_norm_before_clip": 0.02587890625, "grad_clipped": false, "step_seconds": 1.1017547090013977, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 514, "loss": 0.051131051033735275, "loss_ema": 0.1097106369206495, "loss_mean_50": 0.12142383441329002, "loss_std_50": 0.11817755928439388, "lr": 0.00025, "grad_norm_before_clip": 0.024658203125, "grad_clipped": false, "step_seconds": 0.5208990379978786, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 515, "loss": 0.04589683189988136, "loss_ema": 0.10651994666961109, "loss_mean_50": 0.12141279757022858, "loss_std_50": 0.11818458641191995, "lr": 0.00025, "grad_norm_before_clip": 0.0079345703125, "grad_clipped": false, "step_seconds": 2.063082011001825, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 516, "loss": 0.6847035884857178, "loss_ema": 0.1354291287604164, "loss_mean_50": 0.1340184359252453, "loss_std_50": 0.1416505151001212, "lr": 0.00025, "grad_norm_before_clip": 0.0927734375, "grad_clipped": false, "step_seconds": 0.569788728000276, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 517, "loss": 0.27823951840400696, "loss_ema": 0.14256964824259594, "loss_mean_50": 0.1365853475034237, "loss_std_50": 0.14307072787063913, "lr": 0.00025, "grad_norm_before_clip": 0.0123291015625, "grad_clipped": false, "step_seconds": 1.8980839309988369, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 518, "loss": 0.05844378471374512, "loss_ema": 0.1383633550661534, "loss_mean_50": 0.13597022920846938, "loss_std_50": 0.14333899988075408, "lr": 0.00025, "grad_norm_before_clip": 0.0125732421875, "grad_clipped": false, "step_seconds": 1.8293202759996348, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 519, "loss": 0.2284366637468338, "loss_ema": 0.14286702050018743, "loss_mean_50": 0.124873349070549, "loss_std_50": 0.11051566774238242, "lr": 0.00025, "grad_norm_before_clip": 0.0093994140625, "grad_clipped": false, "step_seconds": 2.0637725029992, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 520, "loss": 0.05244794487953186, "loss_ema": 0.13834606671915464, "loss_mean_50": 0.1227754858136177, "loss_std_50": 0.11087442203999184, "lr": 0.00025, "grad_norm_before_clip": 0.01312255859375, "grad_clipped": false, "step_seconds": 2.057890052001312, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 521, "loss": 0.06923038512468338, "loss_ema": 0.13489028263943106, "loss_mean_50": 0.12251345559954643, "loss_std_50": 0.11098511918093709, "lr": 0.00025, "grad_norm_before_clip": 0.015625, "grad_clipped": false, "step_seconds": 0.5969838499986508, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 522, "loss": 0.08324630558490753, "loss_ema": 0.1323080837867049, "loss_mean_50": 0.12247669219970703, "loss_std_50": 0.11099781501702607, "lr": 0.00025, "grad_norm_before_clip": 0.0111083984375, "grad_clipped": false, "step_seconds": 1.8901282600018021, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 523, "loss": 0.07361245155334473, "loss_ema": 0.12937330217503687, "loss_mean_50": 0.12236339822411538, "loss_std_50": 0.11104473141467791, "lr": 0.00025, "grad_norm_before_clip": 0.050048828125, "grad_clipped": false, "step_seconds": 1.8901813329975994, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 524, "loss": 0.06427018344402313, "loss_ema": 0.12611814623848616, "loss_mean_50": 0.12249695159494876, "loss_std_50": 0.11097074219810907, "lr": 0.00025, "grad_norm_before_clip": 0.0098876953125, "grad_clipped": false, "step_seconds": 1.8904914049999206, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 525, "loss": 0.06506924331188202, "loss_ema": 0.12306570109215595, "loss_mean_50": 0.12062390603125095, "loss_std_50": 0.11113375620664027, "lr": 0.00025, "grad_norm_before_clip": 0.010009765625, "grad_clipped": false, "step_seconds": 2.063635125003202, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 526, "loss": 0.049308620393276215, "loss_ema": 0.11937784705721197, "loss_mean_50": 0.11775008164346218, "loss_std_50": 0.11108287350248776, "lr": 0.00025, "grad_norm_before_clip": 0.0286865234375, "grad_clipped": false, "step_seconds": 2.0590138449988444, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 527, "loss": 0.21532116830348969, "loss_ema": 0.12417501311952585, "loss_mean_50": 0.12075578190386295, "loss_std_50": 0.11164764990768468, "lr": 0.00025, "grad_norm_before_clip": 0.012939453125, "grad_clipped": false, "step_seconds": 1.8888084889986203, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 528, "loss": 0.06037863716483116, "loss_ema": 0.12098519432179111, "loss_mean_50": 0.11706747859716415, "loss_std_50": 0.11052960476585655, "lr": 0.00025, "grad_norm_before_clip": 0.020263671875, "grad_clipped": false, "step_seconds": 0.5202044700017723, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 529, "loss": 0.1450612097978592, "loss_ema": 0.12218899509559451, "loss_mean_50": 0.11752988144755364, "loss_std_50": 0.11059736730907509, "lr": 0.00025, "grad_norm_before_clip": 0.015869140625, "grad_clipped": false, "step_seconds": 1.1032192669990764, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 530, "loss": 0.06079232320189476, "loss_ema": 0.11911916150090952, "loss_mean_50": 0.11468964762985706, "loss_std_50": 0.11019373931365438, "lr": 0.00025, "grad_norm_before_clip": 0.007354736328125, "grad_clipped": false, "step_seconds": 2.105971477998537, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 531, "loss": 0.16860291361808777, "loss_ema": 0.12159334910676843, "loss_mean_50": 0.11641658447682858, "loss_std_50": 0.11034841159043524, "lr": 0.00025, "grad_norm_before_clip": 0.032958984375, "grad_clipped": false, "step_seconds": 0.7240619250005693, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 532, "loss": 0.07828093320131302, "loss_ema": 0.11942772831149565, "loss_mean_50": 0.11184510804712772, "loss_std_50": 0.10704964347579066, "lr": 0.00025, "grad_norm_before_clip": 0.01251220703125, "grad_clipped": false, "step_seconds": 1.8900201310025295, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 533, "loss": 0.08984101563692093, "loss_ema": 0.1179483926777669, "loss_mean_50": 0.11171961225569248, "loss_std_50": 0.1070716853393113, "lr": 0.00025, "grad_norm_before_clip": 0.007781982421875, "grad_clipped": false, "step_seconds": 2.0627251389996673, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 534, "loss": 0.24402999877929688, "loss_ema": 0.12425247298284339, "loss_mean_50": 0.11444162912666797, "loss_std_50": 0.10865896116706589, "lr": 0.00025, "grad_norm_before_clip": 0.0257568359375, "grad_clipped": false, "step_seconds": 2.0563474589980615, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 535, "loss": 0.10265029221773148, "loss_ema": 0.12317236394458779, "loss_mean_50": 0.11555162146687507, "loss_std_50": 0.10824858900625565, "lr": 0.00025, "grad_norm_before_clip": 0.05810546875, "grad_clipped": false, "step_seconds": 0.3458839289996831, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 536, "loss": 0.10768013447523117, "loss_ema": 0.12239775247111996, "loss_mean_50": 0.11601415261626244, "loss_std_50": 0.10816452615216392, "lr": 0.00025, "grad_norm_before_clip": 0.0849609375, "grad_clipped": false, "step_seconds": 0.36139521500081173, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 537, "loss": 0.04900310933589935, "loss_ema": 0.11872802031435893, "loss_mean_50": 0.1160077002644539, "loss_std_50": 0.10816851368166117, "lr": 0.00025, "grad_norm_before_clip": 0.01519775390625, "grad_clipped": false, "step_seconds": 2.0615253870018933, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 538, "loss": 0.10775349289178848, "loss_ema": 0.1181792939432304, "loss_mean_50": 0.11698767073452472, "loss_std_50": 0.10786691867717767, "lr": 0.00025, "grad_norm_before_clip": 0.0283203125, "grad_clipped": false, "step_seconds": 0.5711896439970587, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 539, "loss": 0.06570440530776978, "loss_ema": 0.11555554951145737, "loss_mean_50": 0.11693407215178013, "loss_std_50": 0.10789171910942914, "lr": 0.00025, "grad_norm_before_clip": 0.0081787109375, "grad_clipped": false, "step_seconds": 2.0619135750021087, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 540, "loss": 0.06242624670267105, "loss_ema": 0.11289908437101805, "loss_mean_50": 0.11703900925815106, "loss_std_50": 0.10783608702017422, "lr": 0.00025, "grad_norm_before_clip": 0.0159912109375, "grad_clipped": false, "step_seconds": 1.568738581998332, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 541, "loss": 0.08282651007175446, "loss_ema": 0.11139545565605487, "loss_mean_50": 0.11762914814054966, "loss_std_50": 0.10756616546492807, "lr": 0.00025, "grad_norm_before_clip": 0.0091552734375, "grad_clipped": false, "step_seconds": 1.2891556110007514, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 542, "loss": 0.06332950294017792, "loss_ema": 0.108992158020261, "loss_mean_50": 0.1176212789863348, "loss_std_50": 0.10757012307926901, "lr": 0.00025, "grad_norm_before_clip": 0.0179443359375, "grad_clipped": false, "step_seconds": 0.5700671609993151, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 543, "loss": 0.27200308442115784, "loss_ema": 0.11714270434030585, "loss_mean_50": 0.12043719790875912, "loss_std_50": 0.10971046123971928, "lr": 0.00025, "grad_norm_before_clip": 0.02587890625, "grad_clipped": false, "step_seconds": 0.475289652000356, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 544, "loss": 0.0576828308403492, "loss_ema": 0.11416971066530802, "loss_mean_50": 0.12069200038909912, "loss_std_50": 0.1095495058371611, "lr": 0.00025, "grad_norm_before_clip": 0.039306640625, "grad_clipped": false, "step_seconds": 1.585038755001733, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 545, "loss": 0.06121758371591568, "loss_ema": 0.1115221043178384, "loss_mean_50": 0.12095485471189021, "loss_std_50": 0.10939060426738959, "lr": 0.00025, "grad_norm_before_clip": 0.0162353515625, "grad_clipped": false, "step_seconds": 1.2900243080002838, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 546, "loss": 0.12191324681043625, "loss_ema": 0.11204166144246828, "loss_mean_50": 0.11835567481815815, "loss_std_50": 0.10778116576155544, "lr": 0.00025, "grad_norm_before_clip": 0.00811767578125, "grad_clipped": false, "step_seconds": 2.063505799997074, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 547, "loss": 0.05619153380393982, "loss_ema": 0.10924915506054185, "loss_mean_50": 0.11821705304086208, "loss_std_50": 0.10785654493116616, "lr": 0.00025, "grad_norm_before_clip": 0.0242919921875, "grad_clipped": false, "step_seconds": 1.8899049190004007, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 548, "loss": 0.05357206240296364, "loss_ema": 0.10646530042766293, "loss_mean_50": 0.11291131943464279, "loss_std_50": 0.10432315763521884, "lr": 0.00025, "grad_norm_before_clip": 0.055908203125, "grad_clipped": false, "step_seconds": 1.8890930750021653, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 549, "loss": 0.07697523385286331, "loss_ema": 0.10499079709892295, "loss_mean_50": 0.11198411583900451, "loss_std_50": 0.1044323528539691, "lr": 0.00025, "grad_norm_before_clip": 0.016845703125, "grad_clipped": false, "step_seconds": 0.6682527429984475, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 550, "loss": 0.06828487664461136, "loss_ema": 0.10315550107620737, "loss_mean_50": 0.11207165986299515, "loss_std_50": 0.10439384199445931, "lr": 0.00025, "grad_norm_before_clip": 0.042236328125, "grad_clipped": false, "step_seconds": 2.063366941001732, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 551, "loss": 0.044827673584222794, "loss_ema": 0.10023910970160814, "loss_mean_50": 0.11209232352674008, "loss_std_50": 0.10438042659322917, "lr": 0.00025, "grad_norm_before_clip": 0.040283203125, "grad_clipped": false, "step_seconds": 1.8279638949970831, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 552, "loss": 0.04635991156101227, "loss_ema": 0.09754514979457835, "loss_mean_50": 0.11202780038118362, "loss_std_50": 0.1044200347380839, "lr": 0.00025, "grad_norm_before_clip": 0.0162353515625, "grad_clipped": false, "step_seconds": 1.2668455190032546, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 553, "loss": 0.06318652629852295, "loss_ema": 0.09582721861977557, "loss_mean_50": 0.11036885529756546, "loss_std_50": 0.10452385660423794, "lr": 0.00025, "grad_norm_before_clip": 0.0123291015625, "grad_clipped": false, "step_seconds": 2.061876209998445, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 554, "loss": 0.1416390985250473, "loss_ema": 0.09811781261503916, "loss_mean_50": 0.11126905396580695, "loss_std_50": 0.10459544551405542, "lr": 0.00025, "grad_norm_before_clip": 0.0146484375, "grad_clipped": false, "step_seconds": 0.5977046080006403, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 555, "loss": 0.058853279799222946, "loss_ema": 0.09615458597424835, "loss_mean_50": 0.10988775707781315, "loss_std_50": 0.10482224801075421, "lr": 0.00025, "grad_norm_before_clip": 0.033447265625, "grad_clipped": false, "step_seconds": 0.6115225190005731, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 556, "loss": 0.11467458307743073, "loss_ema": 0.09708058582940746, "loss_mean_50": 0.10805130280554294, "loss_std_50": 0.10391400928128856, "lr": 0.00025, "grad_norm_before_clip": 0.1435546875, "grad_clipped": false, "step_seconds": 0.4208124709984986, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 557, "loss": 0.7199581265449524, "loss_ema": 0.1282244628651847, "loss_mean_50": 0.12157957963645458, "loss_std_50": 0.1342403931860678, "lr": 0.00025, "grad_norm_before_clip": 0.0228271484375, "grad_clipped": false, "step_seconds": 1.8282367020001402, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 558, "loss": 0.06444186717271805, "loss_ema": 0.12503533308056136, "loss_mean_50": 0.12137324906885624, "loss_std_50": 0.1343201045583963, "lr": 0.00025, "grad_norm_before_clip": 0.007049560546875, "grad_clipped": false, "step_seconds": 2.062969285998406, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 559, "loss": 0.19788624346256256, "loss_ema": 0.12867787859966143, "loss_mean_50": 0.12376929499208927, "loss_std_50": 0.1345947805294952, "lr": 0.00025, "grad_norm_before_clip": 0.01239013671875, "grad_clipped": false, "step_seconds": 2.061276913002075, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 560, "loss": 0.903060257434845, "loss_ema": 0.16739699754142062, "loss_mean_50": 0.13658089257776737, "loss_std_50": 0.17237369493099206, "lr": 0.00025, "grad_norm_before_clip": 0.1123046875, "grad_clipped": false, "step_seconds": 0.6756475119982497, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 561, "loss": 0.19412082433700562, "loss_ema": 0.16873318888119987, "loss_mean_50": 0.13856268636882305, "loss_std_50": 0.17245420384848714, "lr": 0.00025, "grad_norm_before_clip": 0.01611328125, "grad_clipped": false, "step_seconds": 1.8908594400018046, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 562, "loss": 0.06738600879907608, "loss_ema": 0.1636658298770937, "loss_mean_50": 0.13894503869116306, "loss_std_50": 0.1722746859839447, "lr": 0.00025, "grad_norm_before_clip": 0.0255126953125, "grad_clipped": false, "step_seconds": 1.8909194669977296, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 563, "loss": 0.8560429215431213, "loss_ema": 0.19828468446039507, "loss_mean_50": 0.15495992481708526, "loss_std_50": 0.19891391062958805, "lr": 0.00025, "grad_norm_before_clip": 0.10693359375, "grad_clipped": false, "step_seconds": 1.8899535170021409, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 564, "loss": 0.05204964801669121, "loss_ema": 0.19097293263820986, "loss_mean_50": 0.15497829675674438, "loss_std_50": 0.19890436220763957, "lr": 0.00025, "grad_norm_before_clip": 0.01953125, "grad_clipped": false, "step_seconds": 1.8289667860008194, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 565, "loss": 0.10510550439357758, "loss_ema": 0.18667956122597826, "loss_mean_50": 0.1561624702066183, "loss_std_50": 0.1984270990246143, "lr": 0.00025, "grad_norm_before_clip": 0.00775146484375, "grad_clipped": false, "step_seconds": 1.1036480250004388, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 566, "loss": 0.2211739420890808, "loss_ema": 0.18840428026913336, "loss_mean_50": 0.14689187727868558, "loss_std_50": 0.18380637989302961, "lr": 0.00025, "grad_norm_before_clip": 0.01806640625, "grad_clipped": false, "step_seconds": 2.1352309469984903, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 567, "loss": 0.14999432861804962, "loss_ema": 0.18648378268657914, "loss_mean_50": 0.14432697348296641, "loss_std_50": 0.18284790128793377, "lr": 0.00025, "grad_norm_before_clip": 0.01611328125, "grad_clipped": false, "step_seconds": 0.47637354800099274, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 568, "loss": 0.05750162899494171, "loss_ema": 0.18003467500199724, "loss_mean_50": 0.14430813036859036, "loss_std_50": 0.18285679920906658, "lr": 0.00025, "grad_norm_before_clip": 0.025634765625, "grad_clipped": false, "step_seconds": 2.063923227997293, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 569, "loss": 0.0868595615029335, "loss_ema": 0.17537591932704402, "loss_mean_50": 0.14147658832371235, "loss_std_50": 0.18262816306736415, "lr": 0.00025, "grad_norm_before_clip": 0.039794921875, "grad_clipped": false, "step_seconds": 2.058925427001668, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 570, "loss": 0.6211843490600586, "loss_ema": 0.19766634081369475, "loss_mean_50": 0.15285131640732288, "loss_std_50": 0.19408124647561142, "lr": 0.00025, "grad_norm_before_clip": 0.039794921875, "grad_clipped": false, "step_seconds": 1.289496462999523, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 571, "loss": 0.050461962819099426, "loss_ema": 0.19030612191396498, "loss_mean_50": 0.1524759479612112, "loss_std_50": 0.19426067981876688, "lr": 0.00025, "grad_norm_before_clip": 0.046630859375, "grad_clipped": false, "step_seconds": 0.5215789020003285, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 572, "loss": 0.2694628834724426, "loss_ema": 0.19426395999188886, "loss_mean_50": 0.1562002795189619, "loss_std_50": 0.19468231833433813, "lr": 0.00025, "grad_norm_before_clip": 0.0308837890625, "grad_clipped": false, "step_seconds": 1.2908481640006357, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 573, "loss": 0.10015866160392761, "loss_ema": 0.1895586950724908, "loss_mean_50": 0.15673120371997357, "loss_std_50": 0.19449247148343426, "lr": 0.00025, "grad_norm_before_clip": 0.1923828125, "grad_clipped": false, "step_seconds": 1.5701573400001507, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 574, "loss": 0.0756293311715126, "loss_ema": 0.1838622268774419, "loss_mean_50": 0.15695838667452336, "loss_std_50": 0.1943909445443578, "lr": 0.00025, "grad_norm_before_clip": 0.044677734375, "grad_clipped": false, "step_seconds": 1.829361162999703, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 575, "loss": 0.45572760701179504, "loss_ema": 0.19745549588415953, "loss_mean_50": 0.1647715539485216, "loss_std_50": 0.19835115927328614, "lr": 0.00025, "grad_norm_before_clip": 0.04296875, "grad_clipped": false, "step_seconds": 0.5977428719997988, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 576, "loss": 0.1783500611782074, "loss_ema": 0.19650022414886192, "loss_mean_50": 0.16735238276422024, "loss_std_50": 0.19767037067893775, "lr": 0.00025, "grad_norm_before_clip": 0.024658203125, "grad_clipped": false, "step_seconds": 0.42128581699944334, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 577, "loss": 0.06024136021733284, "loss_ema": 0.18968728095228546, "loss_mean_50": 0.1642507866024971, "loss_std_50": 0.19810954262956842, "lr": 0.00025, "grad_norm_before_clip": 0.0186767578125, "grad_clipped": false, "step_seconds": 2.0595353489989066, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 578, "loss": 0.08084806799888611, "loss_ema": 0.1842453203046155, "loss_mean_50": 0.1646601752191782, "loss_std_50": 0.19791552511789157, "lr": 0.00025, "grad_norm_before_clip": 0.009765625, "grad_clipped": false, "step_seconds": 2.0595409020024817, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 579, "loss": 0.049436166882514954, "loss_ema": 0.17750486263351045, "loss_mean_50": 0.16274767436087131, "loss_std_50": 0.19855665810087902, "lr": 0.00025, "grad_norm_before_clip": 0.0140380859375, "grad_clipped": false, "step_seconds": 2.0634793619974516, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 580, "loss": 0.057010550051927567, "loss_ema": 0.1714801470044313, "loss_mean_50": 0.16267203889787196, "loss_std_50": 0.19859619752656293, "lr": 0.00025, "grad_norm_before_clip": 0.02197265625, "grad_clipped": false, "step_seconds": 1.8276423610004713, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 581, "loss": 0.09948773682117462, "loss_ema": 0.16788052649526844, "loss_mean_50": 0.1612897353619337, "loss_std_50": 0.1987905443526517, "lr": 0.00025, "grad_norm_before_clip": 0.050048828125, "grad_clipped": false, "step_seconds": 0.361587433002569, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 582, "loss": 0.0908292680978775, "loss_ema": 0.16402796357539887, "loss_mean_50": 0.161540702059865, "loss_std_50": 0.19869348722079108, "lr": 0.00025, "grad_norm_before_clip": 0.0086669921875, "grad_clipped": false, "step_seconds": 2.0632756989980408, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 583, "loss": 0.11756972223520279, "loss_ema": 0.16170505150838907, "loss_mean_50": 0.16209527619183065, "loss_std_50": 0.1985312226128744, "lr": 0.00025, "grad_norm_before_clip": 0.0269775390625, "grad_clipped": false, "step_seconds": 2.062913640998886, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 584, "loss": 0.10477005690336227, "loss_ema": 0.15885830177813773, "loss_mean_50": 0.15931007735431194, "loss_std_50": 0.1983389690898334, "lr": 0.00025, "grad_norm_before_clip": 0.484375, "grad_clipped": false, "step_seconds": 2.062491766002495, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 585, "loss": 0.48702430725097656, "loss_ema": 0.17526660205177966, "loss_mean_50": 0.16699755765497684, "loss_std_50": 0.2033788951950167, "lr": 0.00025, "grad_norm_before_clip": 0.037353515625, "grad_clipped": false, "step_seconds": 0.5978296809989843, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 586, "loss": 0.2192089557647705, "loss_ema": 0.1774637197374292, "loss_mean_50": 0.16922813408076764, "loss_std_50": 0.20332768873803728, "lr": 0.00025, "grad_norm_before_clip": 0.01239013671875, "grad_clipped": false, "step_seconds": 1.8897459500003606, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 587, "loss": 0.13952773809432983, "loss_ema": 0.1755669206552742, "loss_mean_50": 0.17103862665593625, "loss_std_50": 0.20265101100938127, "lr": 0.00025, "grad_norm_before_clip": 0.01470947265625, "grad_clipped": false, "step_seconds": 1.5851386659996933, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 588, "loss": 0.06796740740537643, "loss_ema": 0.1701869449927793, "loss_mean_50": 0.170242904946208, "loss_std_50": 0.20297579277275915, "lr": 0.00025, "grad_norm_before_clip": 0.038818359375, "grad_clipped": false, "step_seconds": 0.6128301600001578, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 589, "loss": 0.06731129437685013, "loss_ema": 0.16504316246198283, "loss_mean_50": 0.17027504272758962, "loss_std_50": 0.20295936487348393, "lr": 0.00025, "grad_norm_before_clip": 0.1904296875, "grad_clipped": false, "step_seconds": 1.8896283419999236, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 590, "loss": 0.229600727558136, "loss_ema": 0.16827104071679047, "loss_mean_50": 0.1736185323446989, "loss_std_50": 0.20253169714531116, "lr": 0.00025, "grad_norm_before_clip": 0.0546875, "grad_clipped": false, "step_seconds": 0.7232036679997691, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 591, "loss": 0.11572926491498947, "loss_ema": 0.1656439519267004, "loss_mean_50": 0.1742765874415636, "loss_std_50": 0.20228893897480477, "lr": 0.00025, "grad_norm_before_clip": 0.043212890625, "grad_clipped": false, "step_seconds": 0.5706237269987469, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 592, "loss": 0.1410340815782547, "loss_ema": 0.1644134584092781, "loss_mean_50": 0.17583067901432514, "loss_std_50": 0.20172832176883357, "lr": 0.00025, "grad_norm_before_clip": 0.01483154296875, "grad_clipped": false, "step_seconds": 0.5707496450013423, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 593, "loss": 0.24899539351463318, "loss_ema": 0.16864255516454585, "loss_mean_50": 0.17537052519619464, "loss_std_50": 0.20153457008626888, "lr": 0.00025, "grad_norm_before_clip": 0.044189453125, "grad_clipped": false, "step_seconds": 0.3484221780017833, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 594, "loss": 0.040708888322114944, "loss_ema": 0.1622458718224243, "loss_mean_50": 0.17503104634582997, "loss_std_50": 0.20174670989838622, "lr": 0.00025, "grad_norm_before_clip": 0.006683349609375, "grad_clipped": false, "step_seconds": 2.0650931539967132, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 595, "loss": 0.2948734760284424, "loss_ema": 0.1688772520327252, "loss_mean_50": 0.1797041641920805, "loss_std_50": 0.20176241225249833, "lr": 0.00025, "grad_norm_before_clip": 0.01104736328125, "grad_clipped": false, "step_seconds": 2.0635866340016946, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 596, "loss": 0.12038826942443848, "loss_ema": 0.16645280290231085, "loss_mean_50": 0.17967366464436055, "loss_std_50": 0.20177126101734644, "lr": 0.00025, "grad_norm_before_clip": 0.21484375, "grad_clipped": false, "step_seconds": 1.889970058000472, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 597, "loss": 0.07404924929141998, "loss_ema": 0.1618326252217663, "loss_mean_50": 0.18003081895411013, "loss_std_50": 0.20156807242955013, "lr": 0.00025, "grad_norm_before_clip": 0.01190185546875, "grad_clipped": false, "step_seconds": 1.8935801629995694, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 598, "loss": 0.08974168449640274, "loss_ema": 0.1582280781854981, "loss_mean_50": 0.18075421139597891, "loss_std_50": 0.20117746081555116, "lr": 0.00025, "grad_norm_before_clip": 0.0191650390625, "grad_clipped": false, "step_seconds": 1.8936199150011817, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 599, "loss": 0.09335429221391678, "loss_ema": 0.15498438888691904, "loss_mean_50": 0.1810817925632, "loss_std_50": 0.2010214834902833, "lr": 0.00025, "grad_norm_before_clip": 0.009521484375, "grad_clipped": false, "step_seconds": 2.067751179998595, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 600, "loss": 0.07072007656097412, "loss_ema": 0.15077117327062178, "loss_mean_50": 0.18113049656152724, "loss_std_50": 0.20099444204982134, "lr": 0.00025, "grad_norm_before_clip": 0.01080322265625, "grad_clipped": false, "step_seconds": 2.4281133799995587, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 601, "loss": 0.049185823649168015, "loss_ema": 0.1456919057895491, "loss_mean_50": 0.18121765956282615, "loss_std_50": 0.20093625078789926, "lr": 0.00025, "grad_norm_before_clip": 0.0196533203125, "grad_clipped": false, "step_seconds": 1.4311984240011952, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 602, "loss": 0.059157874435186386, "loss_ema": 0.14136520422183094, "loss_mean_50": 0.18147361882030963, "loss_std_50": 0.20077238591136415, "lr": 0.00025, "grad_norm_before_clip": 0.0084228515625, "grad_clipped": false, "step_seconds": 2.06237667899768, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 603, "loss": 0.06509645283222198, "loss_ema": 0.13755176665235047, "loss_mean_50": 0.18151181735098362, "loss_std_50": 0.2007500576725626, "lr": 0.00025, "grad_norm_before_clip": 0.04638671875, "grad_clipped": false, "step_seconds": 1.5889297330031695, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 604, "loss": 0.06610170006752014, "loss_ema": 0.13397926332310894, "loss_mean_50": 0.18000106938183308, "loss_std_50": 0.2013278334592789, "lr": 0.00025, "grad_norm_before_clip": 0.01165771484375, "grad_clipped": false, "step_seconds": 1.892355106996547, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 605, "loss": 0.1349734514951706, "loss_ema": 0.13402897273171202, "loss_mean_50": 0.18152347281575204, "loss_std_50": 0.2006927821378423, "lr": 0.00025, "grad_norm_before_clip": 0.0108642578125, "grad_clipped": false, "step_seconds": 1.1061370380011795, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 606, "loss": 0.05417819321155548, "loss_ema": 0.13003643375570417, "loss_mean_50": 0.18031354501843452, "loss_std_50": 0.20127366891999857, "lr": 0.00025, "grad_norm_before_clip": 0.0247802734375, "grad_clipped": false, "step_seconds": 1.8329958159993112, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 607, "loss": 0.05681528151035309, "loss_ema": 0.1263753761434366, "loss_mean_50": 0.16705068811774254, "loss_std_50": 0.1865901855986016, "lr": 0.00025, "grad_norm_before_clip": 0.03369140625, "grad_clipped": false, "step_seconds": 2.0703712190006627, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 608, "loss": 0.1254383772611618, "loss_ema": 0.12632852619932286, "loss_mean_50": 0.1682706183195114, "loss_std_50": 0.18611413008583022, "lr": 0.00025, "grad_norm_before_clip": 0.01416015625, "grad_clipped": false, "step_seconds": 1.8375029359995096, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 609, "loss": 0.10547187924385071, "loss_ema": 0.12528569385154925, "loss_mean_50": 0.16642233103513718, "loss_std_50": 0.18626965716202235, "lr": 0.00025, "grad_norm_before_clip": 0.1875, "grad_clipped": false, "step_seconds": 2.0638563510001404, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 610, "loss": 0.07039646059274673, "loss_ema": 0.12254123218860913, "loss_mean_50": 0.14976905509829522, "loss_std_50": 0.15411282958762354, "lr": 0.00025, "grad_norm_before_clip": 0.01806640625, "grad_clipped": false, "step_seconds": 0.6678719100018498, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 611, "loss": 0.0812712237238884, "loss_ema": 0.12047773176537308, "loss_mean_50": 0.14751206308603287, "loss_std_50": 0.1542730297230476, "lr": 0.00025, "grad_norm_before_clip": 0.0311279296875, "grad_clipped": false, "step_seconds": 0.3614476420007122, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 612, "loss": 0.10460956394672394, "loss_ema": 0.11968432337444061, "loss_mean_50": 0.14825653418898582, "loss_std_50": 0.15397409606762183, "lr": 0.00025, "grad_norm_before_clip": 0.040771484375, "grad_clipped": false, "step_seconds": 0.5733939899982943, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 613, "loss": 0.06097719445824623, "loss_ema": 0.11674896692863089, "loss_mean_50": 0.13235521964728833, "loss_std_50": 0.11656883252433965, "lr": 0.00025, "grad_norm_before_clip": 0.09033203125, "grad_clipped": false, "step_seconds": 1.270985388000554, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 614, "loss": 0.0576028972864151, "loss_ema": 0.1137916634465201, "loss_mean_50": 0.1324662846326828, "loss_std_50": 0.1164948877759239, "lr": 0.00025, "grad_norm_before_clip": 0.01470947265625, "grad_clipped": false, "step_seconds": 1.889532571000018, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 615, "loss": 0.08264089375734329, "loss_ema": 0.11223412496206125, "loss_mean_50": 0.13201699241995812, "loss_std_50": 0.11664277157220007, "lr": 0.00025, "grad_norm_before_clip": 0.059814453125, "grad_clipped": false, "step_seconds": 0.52451528399979, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 616, "loss": 0.09595097601413727, "loss_ema": 0.11141996751466504, "loss_mean_50": 0.12951253309845925, "loss_std_50": 0.11604438702864113, "lr": 0.00025, "grad_norm_before_clip": 0.024169921875, "grad_clipped": false, "step_seconds": 2.059134946000995, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 617, "loss": 0.08875198662281036, "loss_ema": 0.1102865684700723, "loss_mean_50": 0.12828768625855447, "loss_std_50": 0.11614490034046941, "lr": 0.00025, "grad_norm_before_clip": 0.04052734375, "grad_clipped": false, "step_seconds": 0.6021369839982071, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 618, "loss": 0.13068000972270966, "loss_ema": 0.11130624053270416, "loss_mean_50": 0.12975125387310982, "loss_std_50": 0.11570391951488847, "lr": 0.00025, "grad_norm_before_clip": 0.0216064453125, "grad_clipped": false, "step_seconds": 2.061964516000444, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 619, "loss": 0.09997117519378662, "loss_ema": 0.11073948726575827, "loss_mean_50": 0.13001348614692687, "loss_std_50": 0.11562124087544527, "lr": 0.00025, "grad_norm_before_clip": 0.032958984375, "grad_clipped": false, "step_seconds": 0.3534416840011545, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 620, "loss": 0.06402640789747238, "loss_ema": 0.10840383329734397, "loss_mean_50": 0.11887032732367515, "loss_std_50": 0.09222912181864744, "lr": 0.00025, "grad_norm_before_clip": 0.021728515625, "grad_clipped": false, "step_seconds": 2.067974076999235, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 621, "loss": 0.061251040548086166, "loss_ema": 0.10604619365988108, "loss_mean_50": 0.1190861088782549, "loss_std_50": 0.09208132224966334, "lr": 0.00025, "grad_norm_before_clip": 0.0194091796875, "grad_clipped": false, "step_seconds": 2.0687470019984175, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 622, "loss": 0.15775448083877563, "loss_ema": 0.1086316080188258, "loss_mean_50": 0.11685194082558155, "loss_std_50": 0.08973081841493077, "lr": 0.00025, "grad_norm_before_clip": 0.0205078125, "grad_clipped": false, "step_seconds": 2.0614555280008062, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 623, "loss": 0.05026613175868988, "loss_ema": 0.105713334205819, "loss_mean_50": 0.11585409022867679, "loss_std_50": 0.09018716181119098, "lr": 0.00025, "grad_norm_before_clip": 0.03759765625, "grad_clipped": false, "step_seconds": 1.589107943000272, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 624, "loss": 0.06788181513547897, "loss_ema": 0.103821758252302, "loss_mean_50": 0.11569913990795612, "loss_std_50": 0.09026276255845632, "lr": 0.00025, "grad_norm_before_clip": 0.02001953125, "grad_clipped": false, "step_seconds": 1.8925945820010384, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 625, "loss": 0.06958779692649841, "loss_ema": 0.10211006018601182, "loss_mean_50": 0.10797634370625019, "loss_std_50": 0.07627491510137865, "lr": 0.00025, "grad_norm_before_clip": 0.05810546875, "grad_clipped": false, "step_seconds": 1.8964967319989228, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 626, "loss": 0.06958045810461044, "loss_ema": 0.10048358008194175, "loss_mean_50": 0.10580095164477825, "loss_std_50": 0.07578631816830551, "lr": 0.00025, "grad_norm_before_clip": 0.0238037109375, "grad_clipped": false, "step_seconds": 2.0643758959995466, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 627, "loss": 0.04577207192778587, "loss_ema": 0.09774800467423396, "loss_mean_50": 0.10551156587898732, "loss_std_50": 0.07598709152196194, "lr": 0.00025, "grad_norm_before_clip": 0.0172119140625, "grad_clipped": false, "step_seconds": 0.6002578799998446, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 628, "loss": 0.18156510591506958, "loss_ema": 0.10193885973627574, "loss_mean_50": 0.10752590663731099, "loss_std_50": 0.07663874715679492, "lr": 0.00025, "grad_norm_before_clip": 0.01495361328125, "grad_clipped": false, "step_seconds": 2.0668512180018297, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 629, "loss": 0.06429124623537064, "loss_ema": 0.10005647906123048, "loss_mean_50": 0.10782300822436809, "loss_std_50": 0.07644151788338542, "lr": 0.00025, "grad_norm_before_clip": 0.049560546875, "grad_clipped": false, "step_seconds": 2.064679642997362, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 630, "loss": 0.057723600417375565, "loss_ema": 0.09793983512903774, "loss_mean_50": 0.10783726923167705, "loss_std_50": 0.07643210286313801, "lr": 0.00025, "grad_norm_before_clip": 0.0140380859375, "grad_clipped": false, "step_seconds": 1.2903029740009515, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 631, "loss": 0.08749714493751526, "loss_ema": 0.09741770061946162, "loss_mean_50": 0.10759745739400386, "loss_std_50": 0.07647672167176349, "lr": 0.00025, "grad_norm_before_clip": 0.0230712890625, "grad_clipped": false, "step_seconds": 0.7269512999992003, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 632, "loss": 0.12057959288358688, "loss_ema": 0.09857579523266786, "loss_mean_50": 0.10819246388971805, "loss_std_50": 0.07645967694361794, "lr": 0.00025, "grad_norm_before_clip": 0.0155029296875, "grad_clipped": false, "step_seconds": 2.065252132000751, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 633, "loss": 0.05003894865512848, "loss_ema": 0.09614895290379089, "loss_mean_50": 0.10684184841811657, "loss_std_50": 0.07687740892727542, "lr": 0.00025, "grad_norm_before_clip": 0.064453125, "grad_clipped": false, "step_seconds": 1.8283308099998976, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 634, "loss": 0.4000418186187744, "loss_ema": 0.11134359618954007, "loss_mean_50": 0.11274728365242481, "loss_std_50": 0.08714643103522161, "lr": 0.00025, "grad_norm_before_clip": 0.022705078125, "grad_clipped": false, "step_seconds": 0.6153687140031252, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 635, "loss": 0.051342155784368515, "loss_ema": 0.10834352416928149, "loss_mean_50": 0.10403364062309266, "loss_std_50": 0.06922657621194361, "lr": 0.00025, "grad_norm_before_clip": 0.015625, "grad_clipped": false, "step_seconds": 1.8921872389983037, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 636, "loss": 0.11560574173927307, "loss_ema": 0.10870663504778107, "loss_mean_50": 0.1019615763425827, "loss_std_50": 0.06727106814798349, "lr": 0.00025, "grad_norm_before_clip": 0.00616455078125, "grad_clipped": false, "step_seconds": 2.0641431490003015, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 637, "loss": 0.06917154788970947, "loss_ema": 0.10672988068987749, "loss_mean_50": 0.1005544525384903, "loss_std_50": 0.067206368757092, "lr": 0.00025, "grad_norm_before_clip": 0.0517578125, "grad_clipped": false, "step_seconds": 1.4641054120002082, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 638, "loss": 0.04354642704129219, "loss_ema": 0.10357070800744822, "loss_mean_50": 0.10006603293120861, "loss_std_50": 0.06752938202747812, "lr": 0.00025, "grad_norm_before_clip": 0.0234375, "grad_clipped": false, "step_seconds": 1.8925308719990426, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 639, "loss": 0.09546919167041779, "loss_ema": 0.1031656321905967, "loss_mean_50": 0.10062919087707996, "loss_std_50": 0.06737110249340839, "lr": 0.00025, "grad_norm_before_clip": 0.09521484375, "grad_clipped": false, "step_seconds": 1.8891984919973765, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 640, "loss": 0.10885290801525116, "loss_ema": 0.10344999598182941, "loss_mean_50": 0.09821423448622227, "loss_std_50": 0.06482062080697397, "lr": 0.00025, "grad_norm_before_clip": 0.0194091796875, "grad_clipped": false, "step_seconds": 0.5708324410006753, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 641, "loss": 0.07789600640535355, "loss_ema": 0.10217229650300562, "loss_mean_50": 0.09745756931602954, "loss_std_50": 0.06483256451545191, "lr": 0.00025, "grad_norm_before_clip": 0.031982421875, "grad_clipped": false, "step_seconds": 0.4798199049982941, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 642, "loss": 0.3858628571033478, "loss_ema": 0.11635682453302273, "loss_mean_50": 0.1023541448265314, "loss_std_50": 0.07618962476594243, "lr": 0.00025, "grad_norm_before_clip": 0.01226806640625, "grad_clipped": false, "step_seconds": 2.0697027430069284, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 643, "loss": 0.09997518360614777, "loss_ema": 0.11553774248667897, "loss_mean_50": 0.0993737406283617, "loss_std_50": 0.07325309676446648, "lr": 0.00025, "grad_norm_before_clip": 0.026123046875, "grad_clipped": false, "step_seconds": 1.8913853819976794, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 644, "loss": 0.059340961277484894, "loss_ema": 0.11272790342621926, "loss_mean_50": 0.0997463820874691, "loss_std_50": 0.07300067461644506, "lr": 0.00025, "grad_norm_before_clip": 0.0240478515625, "grad_clipped": false, "step_seconds": 0.4242407389974687, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 645, "loss": 0.055541280657052994, "loss_ema": 0.10986857228776094, "loss_mean_50": 0.09495973818004132, "loss_std_50": 0.06770359442904085, "lr": 0.00025, "grad_norm_before_clip": 0.01055908203125, "grad_clipped": false, "step_seconds": 2.0570865490008146, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 646, "loss": 0.04278983175754547, "loss_ema": 0.10651463526125016, "loss_mean_50": 0.09340776942670345, "loss_std_50": 0.06799168971498214, "lr": 0.00025, "grad_norm_before_clip": 0.0093994140625, "grad_clipped": false, "step_seconds": 2.074367925997649, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 647, "loss": 0.06813076138496399, "loss_ema": 0.10459544156743585, "loss_mean_50": 0.09328939966857433, "loss_std_50": 0.06803042962949042, "lr": 0.00025, "grad_norm_before_clip": 0.020751953125, "grad_clipped": false, "step_seconds": 2.061938368002302, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 648, "loss": 0.051551252603530884, "loss_ema": 0.10194323211924058, "loss_mean_50": 0.09252559103071689, "loss_std_50": 0.06827990688839133, "lr": 0.00025, "grad_norm_before_clip": 0.01904296875, "grad_clipped": false, "step_seconds": 2.0656751720016473, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 649, "loss": 0.13222631812095642, "loss_ema": 0.10345738641932638, "loss_mean_50": 0.09330303154885769, "loss_std_50": 0.0685058427482284, "lr": 0.00025, "grad_norm_before_clip": 0.01312255859375, "grad_clipped": false, "step_seconds": 1.1052784909988986, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 650, "loss": 0.04673575982451439, "loss_ema": 0.10062130508958578, "loss_mean_50": 0.0928233452141285, "loss_std_50": 0.06874584213217953, "lr": 0.00025, "grad_norm_before_clip": 0.037841796875, "grad_clipped": false, "step_seconds": 2.0681610710016685, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 651, "loss": 0.23435984551906586, "loss_ema": 0.10730823211105978, "loss_mean_50": 0.09652682565152645, "loss_std_50": 0.07123792526258119, "lr": 0.00025, "grad_norm_before_clip": 0.0228271484375, "grad_clipped": false, "step_seconds": 2.0623083239988773, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 652, "loss": 0.050004467368125916, "loss_ema": 0.10444304387391308, "loss_mean_50": 0.09634375751018524, "loss_std_50": 0.07134540144870956, "lr": 0.00025, "grad_norm_before_clip": 0.02099609375, "grad_clipped": false, "step_seconds": 2.062408689998847, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 653, "loss": 0.08167456835508347, "loss_ema": 0.10330462009797159, "loss_mean_50": 0.09667531982064247, "loss_std_50": 0.07123785647763711, "lr": 0.00025, "grad_norm_before_clip": 0.0341796875, "grad_clipped": false, "step_seconds": 0.6007363009994151, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 654, "loss": 0.05284780636429787, "loss_ema": 0.1007817794112879, "loss_mean_50": 0.09641024194657803, "loss_std_50": 0.071375654276989, "lr": 0.00025, "grad_norm_before_clip": 0.0263671875, "grad_clipped": false, "step_seconds": 1.5730123390021618, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 655, "loss": 0.05315777286887169, "loss_ema": 0.09840057908416708, "loss_mean_50": 0.09477392837405205, "loss_std_50": 0.07141064066631254, "lr": 0.00025, "grad_norm_before_clip": 0.017578125, "grad_clipped": false, "step_seconds": 1.8917390869974042, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 656, "loss": 0.1070198193192482, "loss_ema": 0.09883154109592113, "loss_mean_50": 0.0958307608962059, "loss_std_50": 0.07119270796947177, "lr": 0.00025, "grad_norm_before_clip": 0.046142578125, "grad_clipped": false, "step_seconds": 0.35045669299870497, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 657, "loss": 0.07501605898141861, "loss_ema": 0.097640766990196, "loss_mean_50": 0.09619477644562721, "loss_std_50": 0.0710386517635867, "lr": 0.00025, "grad_norm_before_clip": 0.009033203125, "grad_clipped": false, "step_seconds": 1.8459905400013668, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 658, "loss": 0.05443563684821129, "loss_ema": 0.09548051048309676, "loss_mean_50": 0.0947747216373682, "loss_std_50": 0.07114946408620969, "lr": 0.00025, "grad_norm_before_clip": 0.0206298828125, "grad_clipped": false, "step_seconds": 2.0639294360007625, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 659, "loss": 0.06758779287338257, "loss_ema": 0.09408587460261104, "loss_mean_50": 0.09401703990995884, "loss_std_50": 0.07123318156905527, "lr": 0.00025, "grad_norm_before_clip": 0.0133056640625, "grad_clipped": false, "step_seconds": 0.5713126190021285, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 660, "loss": 0.07667557895183563, "loss_ema": 0.09321535982007227, "loss_mean_50": 0.09414262227714061, "loss_std_50": 0.07119695411698929, "lr": 0.00025, "grad_norm_before_clip": 0.028076171875, "grad_clipped": false, "step_seconds": 0.571725609996065, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 661, "loss": 0.09813381731510162, "loss_ema": 0.09346128269482373, "loss_mean_50": 0.09447987414896489, "loss_std_50": 0.07117511975019021, "lr": 0.00025, "grad_norm_before_clip": 0.064453125, "grad_clipped": false, "step_seconds": 0.6715269890046329, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 662, "loss": 0.1032509133219719, "loss_ema": 0.09395076422618114, "loss_mean_50": 0.09445270113646984, "loss_std_50": 0.07117150654114902, "lr": 0.00025, "grad_norm_before_clip": 0.00909423828125, "grad_clipped": false, "step_seconds": 1.8971296929958044, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 663, "loss": 0.30008989572525024, "loss_ema": 0.1042577208011346, "loss_mean_50": 0.09923495516180993, "loss_std_50": 0.07658873490790628, "lr": 0.00025, "grad_norm_before_clip": 0.0146484375, "grad_clipped": false, "step_seconds": 2.0627299890038557, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 664, "loss": 0.06861995160579681, "loss_ema": 0.1024758323413677, "loss_mean_50": 0.09945529624819756, "loss_std_50": 0.07648442173799239, "lr": 0.00025, "grad_norm_before_clip": 0.01336669921875, "grad_clipped": false, "step_seconds": 0.4767173520012875, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 665, "loss": 0.3925071656703949, "loss_ema": 0.11697739900781906, "loss_mean_50": 0.10565262168645859, "loss_std_50": 0.08673749712103622, "lr": 0.00025, "grad_norm_before_clip": 0.0235595703125, "grad_clipped": false, "step_seconds": 0.7274195219943067, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 666, "loss": 0.12845690548419952, "loss_ema": 0.11755137433163808, "loss_mean_50": 0.10630274027585983, "loss_std_50": 0.08678415190146767, "lr": 0.00025, "grad_norm_before_clip": 0.019775390625, "grad_clipped": false, "step_seconds": 1.8929181080020498, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 667, "loss": 0.06842925399541855, "loss_ema": 0.1150952683148271, "loss_mean_50": 0.10589628562331199, "loss_std_50": 0.08691289464778705, "lr": 0.00025, "grad_norm_before_clip": 0.0283203125, "grad_clipped": false, "step_seconds": 2.0665386260006926, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 668, "loss": 0.08273699134588242, "loss_ema": 0.11347735446637987, "loss_mean_50": 0.10493742525577546, "loss_std_50": 0.08689864349618336, "lr": 0.00025, "grad_norm_before_clip": 0.00909423828125, "grad_clipped": false, "step_seconds": 1.893668864002393, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 669, "loss": 0.22161585092544556, "loss_ema": 0.11888427928933315, "loss_mean_50": 0.10737031877040863, "loss_std_50": 0.08841515197162031, "lr": 0.00025, "grad_norm_before_clip": 0.0130615234375, "grad_clipped": false, "step_seconds": 2.060075869994762, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 670, "loss": 0.1800241470336914, "loss_ema": 0.12194127267655105, "loss_mean_50": 0.10969027355313302, "loss_std_50": 0.08876854505398757, "lr": 0.00025, "grad_norm_before_clip": 0.0115966796875, "grad_clipped": false, "step_seconds": 1.2698377940032515, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 671, "loss": 0.3389541506767273, "loss_ema": 0.13279191657655987, "loss_mean_50": 0.11524433575570583, "loss_std_50": 0.09409207355162431, "lr": 0.00025, "grad_norm_before_clip": 0.0296630859375, "grad_clipped": false, "step_seconds": 1.8925685310023255, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 672, "loss": 0.05770483240485191, "loss_ema": 0.12903756236797445, "loss_mean_50": 0.11324334278702736, "loss_std_50": 0.09423050446963108, "lr": 0.00025, "grad_norm_before_clip": 0.01251220703125, "grad_clipped": false, "step_seconds": 1.831824480999785, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 673, "loss": 0.24213886260986328, "loss_ema": 0.13469262738006887, "loss_mean_50": 0.11708079740405082, "loss_std_50": 0.09548623164149518, "lr": 0.00025, "grad_norm_before_clip": 0.01336669921875, "grad_clipped": false, "step_seconds": 1.6986178809966077, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 674, "loss": 0.11381013691425323, "loss_ema": 0.13364850285677807, "loss_mean_50": 0.11799936383962631, "loss_std_50": 0.0952290912277978, "lr": 0.00025, "grad_norm_before_clip": 0.031494140625, "grad_clipped": false, "step_seconds": 0.364951839997957, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 675, "loss": 0.4220737814903259, "loss_ema": 0.14806976678845546, "loss_mean_50": 0.12504908353090285, "loss_std_50": 0.10402515417685869, "lr": 0.00025, "grad_norm_before_clip": 0.0296630859375, "grad_clipped": false, "step_seconds": 1.8901606979998178, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 676, "loss": 0.08532553166151047, "loss_ema": 0.1449325550321082, "loss_mean_50": 0.12536398500204085, "loss_std_50": 0.10388049567609628, "lr": 0.00025, "grad_norm_before_clip": 0.03125, "grad_clipped": false, "step_seconds": 0.6162346710043494, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 677, "loss": 0.09099096059799194, "loss_ema": 0.14223547531040237, "loss_mean_50": 0.12626836277544498, "loss_std_50": 0.1033792635220026, "lr": 0.00025, "grad_norm_before_clip": 0.043701171875, "grad_clipped": false, "step_seconds": 2.0668459219959914, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 678, "loss": 0.1534232497215271, "loss_ema": 0.1427948640309586, "loss_mean_50": 0.12570552565157414, "loss_std_50": 0.10315303438284183, "lr": 0.00025, "grad_norm_before_clip": 0.01458740234375, "grad_clipped": false, "step_seconds": 1.8904551369996625, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 679, "loss": 0.10953821241855621, "loss_ema": 0.14113203145033845, "loss_mean_50": 0.12661046497523784, "loss_std_50": 0.1028081852443058, "lr": 0.00025, "grad_norm_before_clip": 0.01287841796875, "grad_clipped": false, "step_seconds": 0.5993544040029519, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 680, "loss": 0.08591131865978241, "loss_ema": 0.13837099581081067, "loss_mean_50": 0.127174219340086, "loss_std_50": 0.10250573419222694, "lr": 0.00025, "grad_norm_before_clip": 0.052978515625, "grad_clipped": false, "step_seconds": 1.2950503469983232, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 681, "loss": 0.3030802309513092, "loss_ema": 0.14660645756783558, "loss_mean_50": 0.13148588106036185, "loss_std_50": 0.10524356611809486, "lr": 0.00025, "grad_norm_before_clip": 0.0299072265625, "grad_clipped": false, "step_seconds": 2.069755893004185, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 682, "loss": 0.1108880415558815, "loss_ema": 0.14482053676723788, "loss_mean_50": 0.13129205003380776, "loss_std_50": 0.10527239484402404, "lr": 0.00025, "grad_norm_before_clip": 0.0279541015625, "grad_clipped": false, "step_seconds": 2.066126321995398, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 683, "loss": 0.0602445974946022, "loss_ema": 0.1405917398036061, "loss_mean_50": 0.13149616301059722, "loss_std_50": 0.10512444499098171, "lr": 0.00025, "grad_norm_before_clip": 0.0269775390625, "grad_clipped": false, "step_seconds": 0.5229592200048501, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 684, "loss": 0.06939776241779327, "loss_ema": 0.13703204093431545, "loss_mean_50": 0.1248832818865776, "loss_std_50": 0.09819474314903369, "lr": 0.00025, "grad_norm_before_clip": 0.0225830078125, "grad_clipped": false, "step_seconds": 0.43078082199644996, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 685, "loss": 0.06147629767656326, "loss_ema": 0.13325425377142783, "loss_mean_50": 0.1250859647244215, "loss_std_50": 0.09805309515837071, "lr": 0.00025, "grad_norm_before_clip": 0.009033203125, "grad_clipped": false, "step_seconds": 2.0752812499995343, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 686, "loss": 0.0900464877486229, "loss_ema": 0.13109386547028756, "loss_mean_50": 0.1245747796446085, "loss_std_50": 0.09816774402179397, "lr": 0.00025, "grad_norm_before_clip": 0.0498046875, "grad_clipped": false, "step_seconds": 1.8351533620007103, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 687, "loss": 0.08728107064962387, "loss_ema": 0.1289032257292544, "loss_mean_50": 0.12493697009980678, "loss_std_50": 0.09799592255413966, "lr": 0.00025, "grad_norm_before_clip": 0.08056640625, "grad_clipped": false, "step_seconds": 1.897238969002501, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 688, "loss": 0.08764278888702393, "loss_ema": 0.12684020388714287, "loss_mean_50": 0.1258188973367214, "loss_std_50": 0.09745640948980815, "lr": 0.00025, "grad_norm_before_clip": 0.01318359375, "grad_clipped": false, "step_seconds": 1.29389405099937, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 689, "loss": 0.18838056921958923, "loss_ema": 0.1299172221537652, "loss_mean_50": 0.12767712488770486, "loss_std_50": 0.0977453625708601, "lr": 0.00025, "grad_norm_before_clip": 0.01470947265625, "grad_clipped": false, "step_seconds": 2.0818068949956796, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 690, "loss": 0.13841797411441803, "loss_ema": 0.13034225975179783, "loss_mean_50": 0.12826842620968817, "loss_std_50": 0.09771912075372208, "lr": 0.00025, "grad_norm_before_clip": 0.042724609375, "grad_clipped": false, "step_seconds": 1.8925709250033833, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 691, "loss": 0.10969759523868561, "loss_ema": 0.1293100265261422, "loss_mean_50": 0.12890445798635483, "loss_std_50": 0.09749241983959457, "lr": 0.00025, "grad_norm_before_clip": 0.046875, "grad_clipped": false, "step_seconds": 0.7316007089975756, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 692, "loss": 0.09869489073753357, "loss_ema": 0.12777926973671178, "loss_mean_50": 0.12316109865903854, "loss_std_50": 0.09038520748080943, "lr": 0.00025, "grad_norm_before_clip": 0.059814453125, "grad_clipped": false, "step_seconds": 2.0660167340029147, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 693, "loss": 0.06864602863788605, "loss_ema": 0.12482260768177048, "loss_mean_50": 0.1225345155596733, "loss_std_50": 0.09065196770627752, "lr": 0.00025, "grad_norm_before_clip": 0.03271484375, "grad_clipped": false, "step_seconds": 0.6746362330013653, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 694, "loss": 0.18631385266780853, "loss_ema": 0.1278971699310724, "loss_mean_50": 0.12507397338747978, "loss_std_50": 0.09062459916459505, "lr": 0.00025, "grad_norm_before_clip": 0.00933837890625, "grad_clipped": false, "step_seconds": 1.8393052719984553, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 695, "loss": 0.12278114259243011, "loss_ema": 0.12764136856414027, "loss_mean_50": 0.12641877062618734, "loss_std_50": 0.09008006843039419, "lr": 0.00025, "grad_norm_before_clip": 0.00897216796875, "grad_clipped": false, "step_seconds": 1.5762375600024825, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 696, "loss": 0.1677485853433609, "loss_ema": 0.1296467294031013, "loss_mean_50": 0.12891794569790363, "loss_std_50": 0.08945646950193503, "lr": 0.00025, "grad_norm_before_clip": 0.08740234375, "grad_clipped": false, "step_seconds": 2.0677101080000284, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 697, "loss": 0.12458223104476929, "loss_ema": 0.1293935044851847, "loss_mean_50": 0.13004697509109975, "loss_std_50": 0.08903740540512785, "lr": 0.00025, "grad_norm_before_clip": 0.007171630859375, "grad_clipped": false, "step_seconds": 1.8947435809968738, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 698, "loss": 0.0610944889485836, "loss_ema": 0.12597855370835462, "loss_mean_50": 0.1302378398180008, "loss_std_50": 0.08887902153973617, "lr": 0.00025, "grad_norm_before_clip": 0.0252685546875, "grad_clipped": false, "step_seconds": 2.072572887998831, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 699, "loss": 0.05042227357625961, "loss_ema": 0.12220073970174987, "loss_mean_50": 0.12860175892710685, "loss_std_50": 0.08957753691941918, "lr": 0.00025, "grad_norm_before_clip": 0.03173828125, "grad_clipped": false, "step_seconds": 1.5907847740018042, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 700, "loss": 0.0573648065328598, "loss_ema": 0.11895894304330536, "loss_mean_50": 0.12881433986127377, "loss_std_50": 0.08939543143496444, "lr": 0.00025, "grad_norm_before_clip": 0.013427734375, "grad_clipped": false, "step_seconds": 2.0629257059990778, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 701, "loss": 0.06843703240156174, "loss_ema": 0.11643284751121817, "loss_mean_50": 0.1254958835989237, "loss_std_50": 0.08849091659209327, "lr": 0.00025, "grad_norm_before_clip": 0.068359375, "grad_clipped": false, "step_seconds": 1.8899516009987565, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 702, "loss": 0.056118112057447433, "loss_ema": 0.11341711073852963, "loss_mean_50": 0.1256181564927101, "loss_std_50": 0.08839068838875985, "lr": 0.00025, "grad_norm_before_clip": 0.0255126953125, "grad_clipped": false, "step_seconds": 0.4251640330039663, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 703, "loss": 0.3837696313858032, "loss_ema": 0.12693473677089329, "loss_mean_50": 0.1316600577533245, "loss_std_50": 0.09523986625367148, "lr": 0.00025, "grad_norm_before_clip": 0.018798828125, "grad_clipped": false, "step_seconds": 2.066101434000302, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 704, "loss": 0.10311108827590942, "loss_ema": 0.1257435543461441, "loss_mean_50": 0.13266532339155673, "loss_std_50": 0.09466622955837063, "lr": 0.00025, "grad_norm_before_clip": 0.01123046875, "grad_clipped": false, "step_seconds": 1.8332641359957051, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 705, "loss": 0.0839163139462471, "loss_ema": 0.12365219232614924, "loss_mean_50": 0.13328049421310426, "loss_std_50": 0.0942465749525895, "lr": 0.00025, "grad_norm_before_clip": 0.01422119140625, "grad_clipped": false, "step_seconds": 2.05858961299964, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 706, "loss": 0.09271567314863205, "loss_ema": 0.12210536636727337, "loss_mean_50": 0.13299441128969192, "loss_std_50": 0.09434751019553851, "lr": 0.00025, "grad_norm_before_clip": 0.04248046875, "grad_clipped": false, "step_seconds": 0.35431427800358506, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 707, "loss": 0.0849231407046318, "loss_ema": 0.12024625508414129, "loss_mean_50": 0.1331925529241562, "loss_std_50": 0.09423587732055945, "lr": 0.00025, "grad_norm_before_clip": 0.008544921875, "grad_clipped": false, "step_seconds": 2.066915600997163, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 708, "loss": 0.052661482244729996, "loss_ema": 0.11686701644217072, "loss_mean_50": 0.13315706983208656, "loss_std_50": 0.094265854611441, "lr": 0.00025, "grad_norm_before_clip": 0.01300048828125, "grad_clipped": false, "step_seconds": 2.061730859000818, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 709, "loss": 0.09462717920541763, "loss_ema": 0.11575502458033306, "loss_mean_50": 0.13369785755872726, "loss_std_50": 0.09396522411018007, "lr": 0.00025, "grad_norm_before_clip": 0.01123046875, "grad_clipped": false, "step_seconds": 1.8967976530038868, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 710, "loss": 0.0506739467382431, "loss_ema": 0.11250097068822855, "loss_mean_50": 0.1331778249144554, "loss_std_50": 0.09435052463199858, "lr": 0.00025, "grad_norm_before_clip": 0.0247802734375, "grad_clipped": false, "step_seconds": 2.067043103998003, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 711, "loss": 0.09936539828777313, "loss_ema": 0.11184419206820577, "loss_mean_50": 0.13320245653390883, "loss_std_50": 0.09434153298591198, "lr": 0.00025, "grad_norm_before_clip": 0.0140380859375, "grad_clipped": false, "step_seconds": 1.8920475260019884, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 712, "loss": 0.0767384022474289, "loss_ema": 0.11008890257716691, "loss_mean_50": 0.132672206312418, "loss_std_50": 0.09458258597938804, "lr": 0.00025, "grad_norm_before_clip": 0.01214599609375, "grad_clipped": false, "step_seconds": 1.2947055909971823, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 713, "loss": 0.0546661801636219, "loss_ema": 0.10731776645648966, "loss_mean_50": 0.1277637320011854, "loss_std_50": 0.09210264707503363, "lr": 0.00025, "grad_norm_before_clip": 0.0216064453125, "grad_clipped": false, "step_seconds": 2.0673923490030575, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 714, "loss": 0.12403815984725952, "loss_ema": 0.10815378612602815, "loss_mean_50": 0.12887209616601467, "loss_std_50": 0.09171688497585732, "lr": 0.00025, "grad_norm_before_clip": 0.0157470703125, "grad_clipped": false, "step_seconds": 1.297488854994299, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 715, "loss": 0.046232279390096664, "loss_ema": 0.10505771078923157, "loss_mean_50": 0.12194659844040871, "loss_std_50": 0.08432403133040325, "lr": 0.00025, "grad_norm_before_clip": 0.050048828125, "grad_clipped": false, "step_seconds": 1.2709218750023865, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 716, "loss": 0.2521873712539673, "loss_ema": 0.11241419381246835, "loss_mean_50": 0.12442120775580406, "loss_std_50": 0.08627180341601977, "lr": 0.00025, "grad_norm_before_clip": 0.01104736328125, "grad_clipped": false, "step_seconds": 1.889963841997087, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 717, "loss": 0.09018353372812271, "loss_ema": 0.11130266080825106, "loss_mean_50": 0.12485629335045814, "loss_std_50": 0.08604287973979077, "lr": 0.00025, "grad_norm_before_clip": 0.0194091796875, "grad_clipped": false, "step_seconds": 0.597145919004106, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 718, "loss": 0.09250666946172714, "loss_ema": 0.11036286124092486, "loss_mean_50": 0.12505168691277505, "loss_std_50": 0.08595806085853552, "lr": 0.00025, "grad_norm_before_clip": 0.050537109375, "grad_clipped": false, "step_seconds": 0.6007012749978458, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 719, "loss": 0.06263356655836105, "loss_ema": 0.10797639650679666, "loss_mean_50": 0.12187204122543335, "loss_std_50": 0.08526491517268348, "lr": 0.00025, "grad_norm_before_clip": 0.01171875, "grad_clipped": false, "step_seconds": 1.1098936999987927, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 720, "loss": 0.06612217426300049, "loss_ema": 0.10588368539460685, "loss_mean_50": 0.11959400177001953, "loss_std_50": 0.08520237027672628, "lr": 0.00025, "grad_norm_before_clip": 0.0693359375, "grad_clipped": false, "step_seconds": 1.889988250004535, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 721, "loss": 0.07283331453800201, "loss_ema": 0.10423116685177661, "loss_mean_50": 0.11427158504724502, "loss_std_50": 0.07945105109337605, "lr": 0.00025, "grad_norm_before_clip": 0.01055908203125, "grad_clipped": false, "step_seconds": 0.5207132909999928, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 722, "loss": 0.11604918539524078, "loss_ema": 0.10482206777894981, "loss_mean_50": 0.1154384721070528, "loss_std_50": 0.0790390735117874, "lr": 0.00025, "grad_norm_before_clip": 0.02685546875, "grad_clipped": false, "step_seconds": 0.5748681099939859, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 723, "loss": 0.06334170699119568, "loss_ema": 0.1027480497395621, "loss_mean_50": 0.11186252899467945, "loss_std_50": 0.07725030389647827, "lr": 0.00025, "grad_norm_before_clip": 0.046875, "grad_clipped": false, "step_seconds": 1.8953181490069255, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 724, "loss": 0.04720703139901161, "loss_ema": 0.09997099882253457, "loss_mean_50": 0.11053046688437462, "loss_std_50": 0.07777766945908855, "lr": 0.00025, "grad_norm_before_clip": 0.03515625, "grad_clipped": false, "step_seconds": 2.07096768199699, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 725, "loss": 0.035717446357011795, "loss_ema": 0.09675832119925842, "loss_mean_50": 0.10280334018170834, "loss_std_50": 0.0645012589233999, "lr": 0.00025, "grad_norm_before_clip": 0.037353515625, "grad_clipped": false, "step_seconds": 2.0671017299973755, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 726, "loss": 0.08094356954097748, "loss_ema": 0.09596758361634437, "loss_mean_50": 0.10271570093929767, "loss_std_50": 0.06452791828117965, "lr": 0.00025, "grad_norm_before_clip": 0.0498046875, "grad_clipped": false, "step_seconds": 1.8898476110043703, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 727, "loss": 0.11695440858602524, "loss_ema": 0.09701692486482841, "loss_mean_50": 0.10323496989905834, "loss_std_50": 0.06453594364877287, "lr": 0.00025, "grad_norm_before_clip": 0.02001953125, "grad_clipped": false, "step_seconds": 0.4763310489943251, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 728, "loss": 0.22457872331142426, "loss_ema": 0.10339501478715821, "loss_mean_50": 0.10465807937085629, "loss_std_50": 0.06638502574588728, "lr": 0.00025, "grad_norm_before_clip": 0.027099609375, "grad_clipped": false, "step_seconds": 0.6161981450059102, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 729, "loss": 0.06892534345388412, "loss_ema": 0.10167153122049451, "loss_mean_50": 0.10384582199156284, "loss_std_50": 0.06656855215176967, "lr": 0.00025, "grad_norm_before_clip": 0.0093994140625, "grad_clipped": false, "step_seconds": 2.063453237999056, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 730, "loss": 0.13208425045013428, "loss_ema": 0.10319216718197649, "loss_mean_50": 0.10476928062736988, "loss_std_50": 0.06663358468799255, "lr": 0.00025, "grad_norm_before_clip": 0.166015625, "grad_clipped": false, "step_seconds": 0.5709486900013871, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 731, "loss": 0.7539994120597839, "loss_ema": 0.13573252942586686, "loss_mean_50": 0.11378766424953937, "loss_std_50": 0.10955434257044572, "lr": 0.00025, "grad_norm_before_clip": 0.05419921875, "grad_clipped": false, "step_seconds": 0.3605222389960545, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 732, "loss": 0.04882906377315521, "loss_ema": 0.13138735614323127, "loss_mean_50": 0.11254648469388485, "loss_std_50": 0.10993105879884402, "lr": 0.00025, "grad_norm_before_clip": 0.0087890625, "grad_clipped": false, "step_seconds": 2.0833435890017427, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 733, "loss": 0.08567176014184952, "loss_ema": 0.1291015763431622, "loss_mean_50": 0.1130550279468298, "loss_std_50": 0.10974659147644453, "lr": 0.00025, "grad_norm_before_clip": 0.020263671875, "grad_clipped": false, "step_seconds": 1.897619613999268, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 734, "loss": 0.039259593933820724, "loss_ema": 0.12460947722269511, "loss_mean_50": 0.11245226457715035, "loss_std_50": 0.11006701232418227, "lr": 0.00025, "grad_norm_before_clip": 0.0140380859375, "grad_clipped": false, "step_seconds": 1.833943764999276, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 735, "loss": 0.051385246217250824, "loss_ema": 0.1209482656724229, "loss_mean_50": 0.1122504435479641, "loss_std_50": 0.1101695016819185, "lr": 0.00025, "grad_norm_before_clip": 0.027099609375, "grad_clipped": false, "step_seconds": 2.0636359389973222, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 736, "loss": 0.13172796368598938, "loss_ema": 0.12148725057310121, "loss_mean_50": 0.11308407306671142, "loss_std_50": 0.11015603172509202, "lr": 0.00025, "grad_norm_before_clip": 0.01397705078125, "grad_clipped": false, "step_seconds": 1.2676809720069286, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 737, "loss": 0.20397193729877472, "loss_ema": 0.1256114849093849, "loss_mean_50": 0.11541789039969444, "loss_std_50": 0.1108187745857819, "lr": 0.00025, "grad_norm_before_clip": 0.021484375, "grad_clipped": false, "step_seconds": 0.6022379089990864, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 738, "loss": 0.09817928820848465, "loss_ema": 0.12423987507433988, "loss_mean_50": 0.11562862038612366, "loss_std_50": 0.11077576746020093, "lr": 0.00025, "grad_norm_before_clip": 0.01043701171875, "grad_clipped": false, "step_seconds": 2.0670809379953425, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 739, "loss": 0.08634041994810104, "loss_ema": 0.12234490231802793, "loss_mean_50": 0.11358781740069389, "loss_std_50": 0.11035581014195757, "lr": 0.00025, "grad_norm_before_clip": 0.046875, "grad_clipped": false, "step_seconds": 1.8893617310022819, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 740, "loss": 0.09094967693090439, "loss_ema": 0.12077514104867174, "loss_mean_50": 0.11263845145702361, "loss_std_50": 0.11034229701625221, "lr": 0.00025, "grad_norm_before_clip": 0.04052734375, "grad_clipped": false, "step_seconds": 0.36436662299820455, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 741, "loss": 0.047661542892456055, "loss_ema": 0.11711946114086094, "loss_mean_50": 0.11139773041009902, "loss_std_50": 0.11071653048363087, "lr": 0.00025, "grad_norm_before_clip": 0.0224609375, "grad_clipped": false, "step_seconds": 1.5869437020010082, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 742, "loss": 0.12756842374801636, "loss_ema": 0.1176419092712187, "loss_mean_50": 0.11197520107030869, "loss_std_50": 0.11072406799659684, "lr": 0.00025, "grad_norm_before_clip": 0.02587890625, "grad_clipped": false, "step_seconds": 0.7238963440031512, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 743, "loss": 0.14868584275245667, "loss_ema": 0.1191941059452806, "loss_mean_50": 0.1135759973526001, "loss_std_50": 0.11066463647235432, "lr": 0.00025, "grad_norm_before_clip": 0.01031494140625, "grad_clipped": false, "step_seconds": 2.063271367995185, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 744, "loss": 0.084246426820755, "loss_ema": 0.11744672198905431, "loss_mean_50": 0.11153464883565903, "loss_std_50": 0.11024465167991254, "lr": 0.00025, "grad_norm_before_clip": 0.03271484375, "grad_clipped": false, "step_seconds": 0.5700435860053403, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 745, "loss": 0.07698927074670792, "loss_ema": 0.115423849426937, "loss_mean_50": 0.11061881139874458, "loss_std_50": 0.11033758406057356, "lr": 0.00025, "grad_norm_before_clip": 0.0218505859375, "grad_clipped": false, "step_seconds": 0.7106421129938099, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 746, "loss": 0.056217994540929794, "loss_ema": 0.11246355668263662, "loss_mean_50": 0.10838819958269597, "loss_std_50": 0.11028744073251011, "lr": 0.00025, "grad_norm_before_clip": 0.0162353515625, "grad_clipped": false, "step_seconds": 1.1019035999997868, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 747, "loss": 0.2463139146566391, "loss_ema": 0.11915607458133673, "loss_mean_50": 0.11082283325493336, "loss_std_50": 0.11194917269315668, "lr": 0.00025, "grad_norm_before_clip": 0.0211181640625, "grad_clipped": false, "step_seconds": 0.5750987070059637, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 748, "loss": 0.08372122794389725, "loss_ema": 0.11738433224946475, "loss_mean_50": 0.11127536803483963, "loss_std_50": 0.11179286316493947, "lr": 0.00025, "grad_norm_before_clip": 0.00946044921875, "grad_clipped": false, "step_seconds": 2.062328191997949, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 749, "loss": 0.06854420900344849, "loss_ema": 0.11494232608716393, "loss_mean_50": 0.1116378067433834, "loss_std_50": 0.11162423549452444, "lr": 0.00025, "grad_norm_before_clip": 0.01043701171875, "grad_clipped": false, "step_seconds": 1.896876312996028, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 750, "loss": 0.055231973528862, "loss_ema": 0.11195680845924882, "loss_mean_50": 0.11159515008330345, "loss_std_50": 0.11164537303230197, "lr": 0.00025, "grad_norm_before_clip": 0.058837890625, "grad_clipped": false, "step_seconds": 2.066521646003821, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 751, "loss": 0.06777498126029968, "loss_ema": 0.10974771709930135, "loss_mean_50": 0.11158190906047821, "loss_std_50": 0.11165052989428256, "lr": 0.00025, "grad_norm_before_clip": 0.01190185546875, "grad_clipped": false, "step_seconds": 1.8932036769983824, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 752, "loss": 0.08214451372623444, "loss_ema": 0.108367556930648, "loss_mean_50": 0.11210243709385395, "loss_std_50": 0.11145122891005724, "lr": 0.00025, "grad_norm_before_clip": 0.0120849609375, "grad_clipped": false, "step_seconds": 1.8936038440006087, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 753, "loss": 0.055277980864048004, "loss_ema": 0.105713078127318, "loss_mean_50": 0.10553260408341884, "loss_std_50": 0.10472216960703233, "lr": 0.00025, "grad_norm_before_clip": 0.033203125, "grad_clipped": false, "step_seconds": 1.8943279969971627, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 754, "loss": 0.05946739390492439, "loss_ema": 0.10340079391619833, "loss_mean_50": 0.10465973019599914, "loss_std_50": 0.10492041600842211, "lr": 0.00025, "grad_norm_before_clip": 0.017822265625, "grad_clipped": false, "step_seconds": 2.0639907230070094, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 755, "loss": 0.0688231885433197, "loss_ema": 0.10167191364755439, "loss_mean_50": 0.1043578676879406, "loss_std_50": 0.10500134257429694, "lr": 0.00025, "grad_norm_before_clip": 0.051513671875, "grad_clipped": false, "step_seconds": 1.8936656039950321, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 756, "loss": 0.054722972214221954, "loss_ema": 0.09932446657588777, "loss_mean_50": 0.10359801366925239, "loss_std_50": 0.10522008461969162, "lr": 0.00025, "grad_norm_before_clip": 0.00909423828125, "grad_clipped": false, "step_seconds": 2.06293278800149, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 757, "loss": 0.18193195760250092, "loss_ema": 0.10345484112721842, "loss_mean_50": 0.10553819000720978, "loss_std_50": 0.10575089144300652, "lr": 0.00025, "grad_norm_before_clip": 0.146484375, "grad_clipped": false, "step_seconds": 0.35100407199934125, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 758, "loss": 0.11001237481832504, "loss_ema": 0.10378271781177376, "loss_mean_50": 0.10668520785868169, "loss_std_50": 0.10548183176811758, "lr": 0.00025, "grad_norm_before_clip": 0.007781982421875, "grad_clipped": false, "step_seconds": 2.066780271998141, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 759, "loss": 0.09321910887956619, "loss_ema": 0.10325453736516337, "loss_mean_50": 0.10665704645216464, "loss_std_50": 0.10548523515367869, "lr": 0.00025, "grad_norm_before_clip": 0.01123046875, "grad_clipped": false, "step_seconds": 1.8938707210036227, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 760, "loss": 0.06931155920028687, "loss_ema": 0.10155738845691954, "loss_mean_50": 0.10702979870140553, "loss_std_50": 0.10531954923221806, "lr": 0.00025, "grad_norm_before_clip": 0.02001953125, "grad_clipped": false, "step_seconds": 2.0595592190002208, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 761, "loss": 0.06785222142934799, "loss_ema": 0.09987213010554097, "loss_mean_50": 0.10639953516423702, "loss_std_50": 0.10545773091099328, "lr": 0.00025, "grad_norm_before_clip": 0.0302734375, "grad_clipped": false, "step_seconds": 0.6722144870000193, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 762, "loss": 0.05633040517568588, "loss_ema": 0.0976950438590482, "loss_mean_50": 0.10599137522280216, "loss_std_50": 0.1056111221121891, "lr": 0.00025, "grad_norm_before_clip": 0.0057373046875, "grad_clipped": false, "step_seconds": 2.0665462639954058, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 763, "loss": 0.07215064018964767, "loss_ema": 0.09641782367557818, "loss_mean_50": 0.10634106442332268, "loss_std_50": 0.10546945162108874, "lr": 0.00025, "grad_norm_before_clip": 0.03662109375, "grad_clipped": false, "step_seconds": 1.8294887440060847, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 764, "loss": 0.11874455213546753, "loss_ema": 0.09753416009857263, "loss_mean_50": 0.10623519226908684, "loss_std_50": 0.10545428963659925, "lr": 0.00025, "grad_norm_before_clip": 0.030517578125, "grad_clipped": false, "step_seconds": 2.0660729499941226, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 765, "loss": 0.10877120494842529, "loss_ema": 0.09809601234106526, "loss_mean_50": 0.1074859707802534, "loss_std_50": 0.10510549173093997, "lr": 0.00025, "grad_norm_before_clip": 0.08984375, "grad_clipped": false, "step_seconds": 1.5692161460028728, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 766, "loss": 0.06058475002646446, "loss_ema": 0.09622044922533521, "loss_mean_50": 0.10365391835570335, "loss_std_50": 0.10323615815401609, "lr": 0.00025, "grad_norm_before_clip": 0.032958984375, "grad_clipped": false, "step_seconds": 0.6147222770014196, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 767, "loss": 0.35060808062553406, "loss_ema": 0.10893983079534515, "loss_mean_50": 0.10886240929365158, "loss_std_50": 0.10884242790639094, "lr": 0.00025, "grad_norm_before_clip": 0.0263671875, "grad_clipped": false, "step_seconds": 1.290566060000856, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 768, "loss": 0.9854710102081299, "loss_ema": 0.1527663897659844, "loss_mean_50": 0.12672169610857964, "loss_std_50": 0.16398543400697466, "lr": 0.00025, "grad_norm_before_clip": 0.0498046875, "grad_clipped": false, "step_seconds": 0.6003237639961299, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 769, "loss": 0.06033618748188019, "loss_ema": 0.14814487965177917, "loss_mean_50": 0.12667574852705002, "loss_std_50": 0.16400370545631945, "lr": 0.00025, "grad_norm_before_clip": 0.04638671875, "grad_clipped": false, "step_seconds": 2.0630976259999443, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 770, "loss": 0.050360020250082016, "loss_ema": 0.1432556366816943, "loss_mean_50": 0.12636050544679164, "loss_std_50": 0.1641348930780088, "lr": 0.00025, "grad_norm_before_clip": 0.022216796875, "grad_clipped": false, "step_seconds": 0.5234204060034244, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 771, "loss": 0.05391239747405052, "loss_ema": 0.13878847472131212, "loss_mean_50": 0.12598208710551262, "loss_std_50": 0.1642796131263196, "lr": 0.00025, "grad_norm_before_clip": 0.0198974609375, "grad_clipped": false, "step_seconds": 1.8327429049968487, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 772, "loss": 0.06040224805474281, "loss_ema": 0.13486916338798363, "loss_mean_50": 0.12486914835870266, "loss_std_50": 0.16453143684825058, "lr": 0.00025, "grad_norm_before_clip": 0.011962890625, "grad_clipped": false, "step_seconds": 2.062579140001617, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 773, "loss": 0.17670971155166626, "loss_ema": 0.13696119079616775, "loss_mean_50": 0.12713650844991206, "loss_std_50": 0.16444904727174706, "lr": 0.00025, "grad_norm_before_clip": 0.78125, "grad_clipped": false, "step_seconds": 0.4223862500002724, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 774, "loss": 0.13832592964172363, "loss_ema": 0.13702942773844556, "loss_mean_50": 0.1289588864147663, "loss_std_50": 0.16405760489546975, "lr": 0.00025, "grad_norm_before_clip": 0.01080322265625, "grad_clipped": false, "step_seconds": 1.288502294999489, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 775, "loss": 0.25693705677986145, "loss_ema": 0.14302480919051633, "loss_mean_50": 0.1333832786232233, "loss_std_50": 0.1644658365635334, "lr": 0.00025, "grad_norm_before_clip": 0.02490234375, "grad_clipped": false, "step_seconds": 1.8353438720005215, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 776, "loss": 0.47869303822517395, "loss_ema": 0.15980822064224923, "loss_mean_50": 0.14133826799690724, "loss_std_50": 0.17121772011203154, "lr": 0.00025, "grad_norm_before_clip": 0.04052734375, "grad_clipped": false, "step_seconds": 0.5710538439961965, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 777, "loss": 0.09128621965646744, "loss_ema": 0.1563821205929601, "loss_mean_50": 0.1408249042183161, "loss_std_50": 0.17132850561672128, "lr": 0.00025, "grad_norm_before_clip": 0.01385498046875, "grad_clipped": false, "step_seconds": 2.0623163800046314, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 778, "loss": 0.050215546041727066, "loss_ema": 0.15107379186539843, "loss_mean_50": 0.13733764067292215, "loss_std_50": 0.17136278155233622, "lr": 0.00025, "grad_norm_before_clip": 0.0289306640625, "grad_clipped": false, "step_seconds": 2.0618361449960503, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 779, "loss": 0.9608973860740662, "loss_ema": 0.1915649715758318, "loss_mean_50": 0.15517708152532578, "loss_std_50": 0.20619981932124848, "lr": 0.00025, "grad_norm_before_clip": 0.255859375, "grad_clipped": false, "step_seconds": 1.8272757710001315, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 780, "loss": 0.05009213835000992, "loss_ema": 0.18449132991454067, "loss_mean_50": 0.1535372392833233, "loss_std_50": 0.20670236509106205, "lr": 0.00025, "grad_norm_before_clip": 0.0162353515625, "grad_clipped": false, "step_seconds": 1.889027659999556, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 781, "loss": 0.4580806493759155, "loss_ema": 0.1981707958876094, "loss_mean_50": 0.1476188640296459, "loss_std_50": 0.19322183603957213, "lr": 0.00025, "grad_norm_before_clip": 0.03759765625, "grad_clipped": false, "step_seconds": 1.888684394005395, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 782, "loss": 0.052689120173454285, "loss_ema": 0.19089671210190165, "loss_mean_50": 0.1476960651576519, "loss_std_50": 0.1931831167446235, "lr": 0.00025, "grad_norm_before_clip": 0.032470703125, "grad_clipped": false, "step_seconds": 2.0628091920007137, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 783, "loss": 0.05492645129561424, "loss_ema": 0.18409819906158728, "loss_mean_50": 0.14708115898072718, "loss_std_50": 0.1934283387263324, "lr": 0.00025, "grad_norm_before_clip": 0.0380859375, "grad_clipped": false, "step_seconds": 2.062338458999875, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 784, "loss": 0.32685649394989014, "loss_ema": 0.19123611380600242, "loss_mean_50": 0.15283309698104858, "loss_std_50": 0.19441016191328953, "lr": 0.00025, "grad_norm_before_clip": 0.310546875, "grad_clipped": false, "step_seconds": 1.2903541029954795, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 785, "loss": 0.04175148531794548, "loss_ema": 0.18376188238159957, "loss_mean_50": 0.15264042176306247, "loss_std_50": 0.1945153543860057, "lr": 0.00025, "grad_norm_before_clip": 0.03759765625, "grad_clipped": false, "step_seconds": 2.063328509000712, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 786, "loss": 0.055309731513261795, "loss_ema": 0.17733927483818268, "loss_mean_50": 0.15111205711960793, "loss_std_50": 0.1949733464878304, "lr": 0.00025, "grad_norm_before_clip": 0.0113525390625, "grad_clipped": false, "step_seconds": 2.058199416998832, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 787, "loss": 0.07001455873250961, "loss_ema": 0.171973039032899, "loss_mean_50": 0.14843290954828262, "loss_std_50": 0.1951488681883562, "lr": 0.00025, "grad_norm_before_clip": 0.162109375, "grad_clipped": false, "step_seconds": 2.0575237950033625, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 788, "loss": 0.07543281465768814, "loss_ema": 0.16714602781413845, "loss_mean_50": 0.1479779800772667, "loss_std_50": 0.1952919495071373, "lr": 0.00025, "grad_norm_before_clip": 0.0732421875, "grad_clipped": false, "step_seconds": 1.5685730440018233, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 789, "loss": 0.08616408705711365, "loss_ema": 0.1630969307762872, "loss_mean_50": 0.14797445341944696, "loss_std_50": 0.19529306413924064, "lr": 0.00025, "grad_norm_before_clip": 0.01092529296875, "grad_clipped": false, "step_seconds": 1.2657319139980245, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 790, "loss": 0.20158350467681885, "loss_ema": 0.16502125947131377, "loss_mean_50": 0.15018712997436523, "loss_std_50": 0.19526117655135602, "lr": 0.00025, "grad_norm_before_clip": 0.0128173828125, "grad_clipped": false, "step_seconds": 1.8875121870005387, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 791, "loss": 0.09587609767913818, "loss_ema": 0.16156400138170499, "loss_mean_50": 0.15115142107009888, "loss_std_50": 0.1948711396162915, "lr": 0.00025, "grad_norm_before_clip": 0.1083984375, "grad_clipped": false, "step_seconds": 0.570984583995596, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 792, "loss": 0.05666455999016762, "loss_ema": 0.15631902931212813, "loss_mean_50": 0.1497333437949419, "loss_std_50": 0.1952951159022538, "lr": 0.00025, "grad_norm_before_clip": 0.03076171875, "grad_clipped": false, "step_seconds": 1.8876796559998184, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 793, "loss": 0.25637203454971313, "loss_ema": 0.1613216795740074, "loss_mean_50": 0.15188706763088702, "loss_std_50": 0.19586464211557528, "lr": 0.00025, "grad_norm_before_clip": 0.03271484375, "grad_clipped": false, "step_seconds": 0.5964312010037247, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 794, "loss": 0.1068696603178978, "loss_ema": 0.1585990786112019, "loss_mean_50": 0.1523395323008299, "loss_std_50": 0.19573395086245107, "lr": 0.00025, "grad_norm_before_clip": 0.02783203125, "grad_clipped": false, "step_seconds": 2.061432402995706, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 795, "loss": 0.0541827492415905, "loss_ema": 0.15337826214272132, "loss_mean_50": 0.15188340187072755, "loss_std_50": 0.19593548251437362, "lr": 0.00025, "grad_norm_before_clip": 0.0167236328125, "grad_clipped": false, "step_seconds": 0.4756225640012417, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 796, "loss": 1.0615922212600708, "loss_ema": 0.1987889600985888, "loss_mean_50": 0.17199088640511037, "loss_std_50": 0.23314108966920333, "lr": 0.00025, "grad_norm_before_clip": 0.8515625, "grad_clipped": false, "step_seconds": 1.8870998170023086, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 797, "loss": 0.09859149903059006, "loss_ema": 0.19377908704518887, "loss_mean_50": 0.1690364380925894, "loss_std_50": 0.23311651609888495, "lr": 0.00025, "grad_norm_before_clip": 0.07177734375, "grad_clipped": false, "step_seconds": 0.3471880709985271, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 798, "loss": 0.17699114978313446, "loss_ema": 0.19293969018208615, "loss_mean_50": 0.17090183652937413, "loss_std_50": 0.23279931746863222, "lr": 0.00025, "grad_norm_before_clip": 0.021728515625, "grad_clipped": false, "step_seconds": 1.8879404910039739, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 799, "loss": 0.14082129299640656, "loss_ema": 0.19033377032280216, "loss_mean_50": 0.1723473782092333, "loss_std_50": 0.23238327764028732, "lr": 0.00025, "grad_norm_before_clip": 0.69921875, "grad_clipped": false, "step_seconds": 0.4206555189957726, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 800, "loss": 0.6286411881446838, "loss_ema": 0.21224914121389624, "loss_mean_50": 0.18381556250154973, "loss_std_50": 0.2403335784598556, "lr": 0.00025, "grad_norm_before_clip": 0.039306640625, "grad_clipped": false, "step_seconds": 2.425478903001931, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 801, "loss": 1.0233328342437744, "loss_ema": 0.2528033258653901, "loss_mean_50": 0.20292671956121922, "loss_std_50": 0.26687350800020265, "lr": 0.00025, "grad_norm_before_clip": 0.255859375, "grad_clipped": false, "step_seconds": 0.7228092860023025, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 802, "loss": 0.1602911502122879, "loss_ema": 0.24817771708273498, "loss_mean_50": 0.2044896522909403, "loss_std_50": 0.266389968888026, "lr": 0.00025, "grad_norm_before_clip": 0.021484375, "grad_clipped": false, "step_seconds": 1.106532468998921, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 803, "loss": 0.5218762159347534, "loss_ema": 0.2618626420253359, "loss_mean_50": 0.21382161699235439, "loss_std_50": 0.26915782009224154, "lr": 0.00025, "grad_norm_before_clip": 0.041015625, "grad_clipped": false, "step_seconds": 2.0617673070009914, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 804, "loss": 0.06300776451826096, "loss_ema": 0.2519198981499821, "loss_mean_50": 0.2138924244046211, "loss_std_50": 0.269117667464217, "lr": 0.00025, "grad_norm_before_clip": 0.03564453125, "grad_clipped": false, "step_seconds": 2.056956934997288, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 805, "loss": 0.10866068303585052, "loss_ema": 0.2447569373942755, "loss_mean_50": 0.21468917429447174, "loss_std_50": 0.26874571039359185, "lr": 0.00025, "grad_norm_before_clip": 0.1015625, "grad_clipped": false, "step_seconds": 0.3667509759980021, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 806, "loss": 0.08651451021432877, "loss_ema": 0.23684481603527816, "loss_mean_50": 0.21532500505447388, "loss_std_50": 0.2684038817682832, "lr": 0.00025, "grad_norm_before_clip": 0.01007080078125, "grad_clipped": false, "step_seconds": 2.067053760998533, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 807, "loss": 0.05133669078350067, "loss_ema": 0.22756940977268927, "loss_mean_50": 0.21271309971809388, "loss_std_50": 0.2693498897916657, "lr": 0.00025, "grad_norm_before_clip": 0.01007080078125, "grad_clipped": false, "step_seconds": 2.061419654004567, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 808, "loss": 0.05718878656625748, "loss_ema": 0.21905037861236767, "loss_mean_50": 0.2116566279530525, "loss_std_50": 0.2698537649541328, "lr": 0.00025, "grad_norm_before_clip": 0.0634765625, "grad_clipped": false, "step_seconds": 1.2890381069955765, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 809, "loss": 0.07039839774370193, "loss_ema": 0.21161777956893438, "loss_mean_50": 0.21120021373033523, "loss_std_50": 0.27007290678542634, "lr": 0.00025, "grad_norm_before_clip": 0.08740234375, "grad_clipped": false, "step_seconds": 0.5197355450000032, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 810, "loss": 0.9533498883247375, "loss_ema": 0.24870438500672454, "loss_mean_50": 0.22888098031282425, "loss_std_50": 0.28851315473806344, "lr": 0.00025, "grad_norm_before_clip": 0.26171875, "grad_clipped": false, "step_seconds": 0.6121251530057634, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 811, "loss": 0.0538305900990963, "loss_ema": 0.23896069526134311, "loss_mean_50": 0.2286005476862192, "loss_std_50": 0.2886763055274544, "lr": 0.00025, "grad_norm_before_clip": 0.01220703125, "grad_clipped": false, "step_seconds": 1.8277379629944335, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 812, "loss": 0.06047850102186203, "loss_ema": 0.23003658554936907, "loss_mean_50": 0.22868350960314274, "loss_std_50": 0.28862737725617915, "lr": 0.00025, "grad_norm_before_clip": 0.0234375, "grad_clipped": false, "step_seconds": 1.586104014000739, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 813, "loss": 0.16421718895435333, "loss_ema": 0.22674561571961827, "loss_mean_50": 0.23052484057843686, "loss_std_50": 0.2879156816636489, "lr": 0.00025, "grad_norm_before_clip": 0.1083984375, "grad_clipped": false, "step_seconds": 2.0617856829994707, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 814, "loss": 0.3095627725124359, "loss_ema": 0.23088647355925915, "loss_mean_50": 0.23434120498597621, "loss_std_50": 0.2876732840333851, "lr": 0.00025, "grad_norm_before_clip": 0.029296875, "grad_clipped": false, "step_seconds": 1.889657301006082, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 815, "loss": 0.06611121445894241, "loss_ema": 0.22264771060424332, "loss_mean_50": 0.23348800517618656, "loss_std_50": 0.28810737667179426, "lr": 0.00025, "grad_norm_before_clip": 0.035400390625, "grad_clipped": false, "step_seconds": 1.8888734339998337, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 816, "loss": 0.10874288529157639, "loss_ema": 0.21695246933860995, "loss_mean_50": 0.2344511678814888, "loss_std_50": 0.2876078041294766, "lr": 0.00025, "grad_norm_before_clip": 0.1015625, "grad_clipped": false, "step_seconds": 0.7909964129939908, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 817, "loss": 0.06809654086828232, "loss_ema": 0.20950967291509356, "loss_mean_50": 0.22880093708634378, "loss_std_50": 0.2880450530023914, "lr": 0.00025, "grad_norm_before_clip": 0.0458984375, "grad_clipped": false, "step_seconds": 0.6677373690035893, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 818, "loss": 0.723369300365448, "loss_ema": 0.23520265428761128, "loss_mean_50": 0.22355890288949012, "loss_std_50": 0.2763755397364216, "lr": 0.00025, "grad_norm_before_clip": 0.080078125, "grad_clipped": false, "step_seconds": 0.36956136699882336, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 819, "loss": 0.05303197354078293, "loss_ema": 0.22609412025026987, "loss_mean_50": 0.2234128186106682, "loss_std_50": 0.2764636923723315, "lr": 0.00025, "grad_norm_before_clip": 0.0181884765625, "grad_clipped": false, "step_seconds": 1.8883288010038086, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 820, "loss": 0.4933779239654541, "loss_ema": 0.23945831043602905, "loss_mean_50": 0.23227317668497563, "loss_std_50": 0.27787109420865286, "lr": 0.00025, "grad_norm_before_clip": 0.0390625, "grad_clipped": false, "step_seconds": 0.5194498889977694, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 821, "loss": 0.09271608293056488, "loss_ema": 0.23212119906075585, "loss_mean_50": 0.2330492503941059, "loss_std_50": 0.2774256927601353, "lr": 0.00025, "grad_norm_before_clip": 0.061767578125, "grad_clipped": false, "step_seconds": 0.47599102400272386, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 822, "loss": 0.09604936093091965, "loss_ema": 0.22531760715426405, "loss_mean_50": 0.23376219265162945, "loss_std_50": 0.2770266167044271, "lr": 0.00025, "grad_norm_before_clip": 0.0074462890625, "grad_clipped": false, "step_seconds": 2.061236485998961, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 823, "loss": 0.11257657408714294, "loss_ema": 0.21968055550090798, "loss_mean_50": 0.23247952990233897, "loss_std_50": 0.2774359755149232, "lr": 0.00025, "grad_norm_before_clip": 0.060791015625, "grad_clipped": false, "step_seconds": 0.7233191880004597, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 824, "loss": 0.04715237393975258, "loss_ema": 0.21105414642285023, "loss_mean_50": 0.23065605878829956, "loss_std_50": 0.2783469427691208, "lr": 0.00025, "grad_norm_before_clip": 0.0166015625, "grad_clipped": false, "step_seconds": 1.5856503329996485, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 825, "loss": 0.07174171507358551, "loss_ema": 0.20408852485538698, "loss_mean_50": 0.22695215195417404, "loss_std_50": 0.2792034440667544, "lr": 0.00025, "grad_norm_before_clip": 0.0087890625, "grad_clipped": false, "step_seconds": 2.061662267995416, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 826, "loss": 0.21503904461860657, "loss_ema": 0.20463605084354797, "loss_mean_50": 0.22167907208204268, "loss_std_50": 0.276879263892255, "lr": 0.00025, "grad_norm_before_clip": 0.014892578125, "grad_clipped": false, "step_seconds": 2.061981304002984, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 827, "loss": 0.2081928551197052, "loss_ema": 0.20481389105735584, "loss_mean_50": 0.22401720479130746, "loss_std_50": 0.27626120167449575, "lr": 0.00025, "grad_norm_before_clip": 0.0234375, "grad_clipped": false, "step_seconds": 2.0617897310003173, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 828, "loss": 0.14167627692222595, "loss_ema": 0.20165701035059932, "loss_mean_50": 0.22584641940891742, "loss_std_50": 0.2754058201289215, "lr": 0.00025, "grad_norm_before_clip": 0.04736328125, "grad_clipped": false, "step_seconds": 1.8894606749963714, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 829, "loss": 0.17525345087051392, "loss_ema": 0.20033683237659505, "loss_mean_50": 0.21013354070484638, "loss_std_50": 0.2546500848244865, "lr": 0.00025, "grad_norm_before_clip": 0.01275634765625, "grad_clipped": false, "step_seconds": 2.0579722469992703, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 830, "loss": 0.05439326912164688, "loss_ema": 0.19303965421384764, "loss_mean_50": 0.21021956332027913, "loss_std_50": 0.2545967280538576, "lr": 0.00025, "grad_norm_before_clip": 0.0172119140625, "grad_clipped": false, "step_seconds": 1.8887726490065688, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 831, "loss": 0.07667043805122375, "loss_ema": 0.18722119340571644, "loss_mean_50": 0.20259135909378528, "loss_std_50": 0.2527633467368198, "lr": 0.00025, "grad_norm_before_clip": 0.0159912109375, "grad_clipped": false, "step_seconds": 1.1025298629974714, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 832, "loss": 0.0440380722284317, "loss_ema": 0.1800620373468522, "loss_mean_50": 0.20241833813488483, "loss_std_50": 0.25286883712350794, "lr": 0.00025, "grad_norm_before_clip": 0.01165771484375, "grad_clipped": false, "step_seconds": 2.0626100549998228, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 833, "loss": 0.05225910618901253, "loss_ema": 0.1736718907889602, "loss_mean_50": 0.2023649912327528, "loss_std_50": 0.25290022678401564, "lr": 0.00025, "grad_norm_before_clip": 0.015869140625, "grad_clipped": false, "step_seconds": 2.061534642998595, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 834, "loss": 0.13832353055477142, "loss_ema": 0.17190447277725074, "loss_mean_50": 0.19859433196485043, "loss_std_50": 0.25242101799341493, "lr": 0.00025, "grad_norm_before_clip": 0.0206298828125, "grad_clipped": false, "step_seconds": 0.6120883530020365, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 835, "loss": 0.05622747167944908, "loss_ema": 0.16612062272236067, "loss_mean_50": 0.1988838516920805, "loss_std_50": 0.2522492009708332, "lr": 0.00025, "grad_norm_before_clip": 0.0140380859375, "grad_clipped": false, "step_seconds": 1.5704208120005205, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 836, "loss": 0.06209251284599304, "loss_ema": 0.16091921722854227, "loss_mean_50": 0.19901950731873513, "loss_std_50": 0.25217376516191853, "lr": 0.00025, "grad_norm_before_clip": 0.0281982421875, "grad_clipped": false, "step_seconds": 2.061538326997834, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 837, "loss": 0.09024543315172195, "loss_ema": 0.15738552802470127, "loss_mean_50": 0.19942412480711938, "loss_std_50": 0.2519826076729298, "lr": 0.00025, "grad_norm_before_clip": 0.04248046875, "grad_clipped": false, "step_seconds": 0.3482174960008706, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 838, "loss": 0.28970906138420105, "loss_ema": 0.16400170469267625, "loss_mean_50": 0.20370964974164962, "loss_std_50": 0.25165933191775, "lr": 0.00025, "grad_norm_before_clip": 0.043212890625, "grad_clipped": false, "step_seconds": 1.889843758995994, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 839, "loss": 0.2488667517900467, "loss_ema": 0.16824495704754477, "loss_mean_50": 0.2069637030363083, "loss_std_50": 0.2511698122179803, "lr": 0.00025, "grad_norm_before_clip": 0.02978515625, "grad_clipped": false, "step_seconds": 1.829167319003318, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 840, "loss": 0.07865362614393234, "loss_ema": 0.16376539050236416, "loss_mean_50": 0.20450510546565057, "loss_std_50": 0.25181127944068715, "lr": 0.00025, "grad_norm_before_clip": 0.05419921875, "grad_clipped": false, "step_seconds": 2.059774061002827, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 841, "loss": 0.18315689265727997, "loss_ema": 0.16473496561010995, "loss_mean_50": 0.20625072136521339, "loss_std_50": 0.25135429770711226, "lr": 0.00025, "grad_norm_before_clip": 0.01190185546875, "grad_clipped": false, "step_seconds": 1.8901983209943864, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 842, "loss": 0.052242934703826904, "loss_ema": 0.15911036406479578, "loss_mean_50": 0.2061622888594866, "loss_std_50": 0.25140768231912863, "lr": 0.00025, "grad_norm_before_clip": 0.029541015625, "grad_clipped": false, "step_seconds": 2.0633244890050264, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 843, "loss": 0.10613278299570084, "loss_ema": 0.15646148501134102, "loss_mean_50": 0.20315750382840633, "loss_std_50": 0.25168728943992075, "lr": 0.00025, "grad_norm_before_clip": 0.057373046875, "grad_clipped": false, "step_seconds": 1.8898528099962277, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 844, "loss": 0.053725190460681915, "loss_ema": 0.15132467028380805, "loss_mean_50": 0.202094614431262, "loss_std_50": 0.2522033609632401, "lr": 0.00025, "grad_norm_before_clip": 0.01708984375, "grad_clipped": false, "step_seconds": 2.058850118999544, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 845, "loss": 0.04109638184309006, "loss_ema": 0.14581325586177213, "loss_mean_50": 0.201832887083292, "loss_std_50": 0.252363462095055, "lr": 0.00025, "grad_norm_before_clip": 0.04150390625, "grad_clipped": false, "step_seconds": 1.830098522004846, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 846, "loss": 0.05517267435789108, "loss_ema": 0.14128122678657806, "loss_mean_50": 0.18170449614524842, "loss_std_50": 0.22119816225434327, "lr": 0.00025, "grad_norm_before_clip": 0.040283203125, "grad_clipped": false, "step_seconds": 2.0634618969997973, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 847, "loss": 0.06418443471193314, "loss_ema": 0.1374263871828458, "loss_mean_50": 0.18101635485887527, "loss_std_50": 0.22150895543356017, "lr": 0.00025, "grad_norm_before_clip": 0.026123046875, "grad_clipped": false, "step_seconds": 0.6683003759972053, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 848, "loss": 0.05393125116825104, "loss_ema": 0.13325163038211607, "loss_mean_50": 0.1785551568865776, "loss_std_50": 0.22222251971366716, "lr": 0.00025, "grad_norm_before_clip": 0.038330078125, "grad_clipped": false, "step_seconds": 1.2910693890007678, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 849, "loss": 0.050793178379535675, "loss_ema": 0.12912870778198704, "loss_mean_50": 0.1767545945942402, "loss_std_50": 0.2228847052102985, "lr": 0.00025, "grad_norm_before_clip": 0.036376953125, "grad_clipped": false, "step_seconds": 1.8298369870026363, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 850, "loss": 0.15468955039978027, "loss_ema": 0.1304067499128767, "loss_mean_50": 0.1672755618393421, "loss_std_50": 0.21333880849392314, "lr": 0.00025, "grad_norm_before_clip": 0.01239013671875, "grad_clipped": false, "step_seconds": 2.064081347998581, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 851, "loss": 0.06744986027479172, "loss_ema": 0.12725890543097246, "loss_mean_50": 0.14815790235996246, "loss_std_50": 0.17518728448938314, "lr": 0.00025, "grad_norm_before_clip": 0.0186767578125, "grad_clipped": false, "step_seconds": 1.2902384510016418, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 852, "loss": 0.43251556158065796, "loss_ema": 0.14252173823845674, "loss_mean_50": 0.15360239058732986, "loss_std_50": 0.17965295279451063, "lr": 0.00025, "grad_norm_before_clip": 0.032958984375, "grad_clipped": false, "step_seconds": 0.7262427889945684, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 853, "loss": 0.19140416383743286, "loss_ema": 0.14496585951840554, "loss_mean_50": 0.14699294954538344, "loss_std_50": 0.1718940546074711, "lr": 0.00025, "grad_norm_before_clip": 0.0244140625, "grad_clipped": false, "step_seconds": 1.2655388839993975, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 854, "loss": 0.057099733501672745, "loss_ema": 0.1405725532175689, "loss_mean_50": 0.14687478892505168, "loss_std_50": 0.17195376595816936, "lr": 0.00025, "grad_norm_before_clip": 0.0230712890625, "grad_clipped": false, "step_seconds": 0.4211743870037026, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 855, "loss": 0.0744776576757431, "loss_ema": 0.13726780844047762, "loss_mean_50": 0.14619112841784954, "loss_std_50": 0.17217215445591702, "lr": 0.00025, "grad_norm_before_clip": 0.267578125, "grad_clipped": false, "step_seconds": 2.0642103560021496, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 856, "loss": 0.06553186476230621, "loss_ema": 0.13368101125656903, "loss_mean_50": 0.1457714755088091, "loss_std_50": 0.17234258621273207, "lr": 0.00025, "grad_norm_before_clip": 0.025146484375, "grad_clipped": false, "step_seconds": 1.8892652579961577, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 857, "loss": 0.3383134603500366, "loss_ema": 0.1439126337112424, "loss_mean_50": 0.1515110109001398, "loss_std_50": 0.17387384651865942, "lr": 0.00025, "grad_norm_before_clip": 0.033447265625, "grad_clipped": false, "step_seconds": 0.596372011998028, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 858, "loss": 0.08042947202920914, "loss_ema": 0.14073847562714076, "loss_mean_50": 0.15197582460939885, "loss_std_50": 0.17365199830324266, "lr": 0.00025, "grad_norm_before_clip": 0.0361328125, "grad_clipped": false, "step_seconds": 0.5708818540006177, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 859, "loss": 0.3812066614627838, "loss_ema": 0.15276188491892292, "loss_mean_50": 0.15819198988378047, "loss_std_50": 0.1761653024968324, "lr": 0.00025, "grad_norm_before_clip": 0.033935546875, "grad_clipped": false, "step_seconds": 0.5988996250016498, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 860, "loss": 0.06822425127029419, "loss_ema": 0.14853500323649146, "loss_mean_50": 0.1404894771426916, "loss_std_50": 0.13504516673030453, "lr": 0.00025, "grad_norm_before_clip": 0.0296630859375, "grad_clipped": false, "step_seconds": 1.8889509139989968, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 861, "loss": 0.0589367151260376, "loss_ema": 0.14405508883096876, "loss_mean_50": 0.14059159964323042, "loss_std_50": 0.13498151145245355, "lr": 0.00025, "grad_norm_before_clip": 0.01220703125, "grad_clipped": false, "step_seconds": 2.0715525409978, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 862, "loss": 0.08768118172883987, "loss_ema": 0.1412363934758623, "loss_mean_50": 0.14113565325737, "loss_std_50": 0.1347120652158408, "lr": 0.00025, "grad_norm_before_clip": 0.031494140625, "grad_clipped": false, "step_seconds": 2.0590601910007535, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 863, "loss": 0.0907590314745903, "loss_ema": 0.1387125253757987, "loss_mean_50": 0.13966649010777474, "loss_std_50": 0.13485281980366262, "lr": 0.00025, "grad_norm_before_clip": 0.0277099609375, "grad_clipped": false, "step_seconds": 1.8905353470036061, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 864, "loss": 0.059605762362480164, "loss_ema": 0.13475718722513277, "loss_mean_50": 0.1346673499047756, "loss_std_50": 0.1330833989394376, "lr": 0.00025, "grad_norm_before_clip": 0.027099609375, "grad_clipped": false, "step_seconds": 2.063883265996992, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 865, "loss": 0.06550991535186768, "loss_ema": 0.13129482363146952, "loss_mean_50": 0.13465532392263413, "loss_std_50": 0.1330896204427028, "lr": 0.00025, "grad_norm_before_clip": 0.0322265625, "grad_clipped": false, "step_seconds": 1.890551789998426, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 866, "loss": 0.035581644624471664, "loss_ema": 0.1265091646811196, "loss_mean_50": 0.13319209910929203, "loss_std_50": 0.13376691979635943, "lr": 0.00025, "grad_norm_before_clip": 0.025390625, "grad_clipped": false, "step_seconds": 2.0637757380027324, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 867, "loss": 0.07065672427415848, "loss_ema": 0.12371654266077155, "loss_mean_50": 0.13324330277740956, "loss_std_50": 0.13374248030469327, "lr": 0.00025, "grad_norm_before_clip": 0.0208740234375, "grad_clipped": false, "step_seconds": 1.8906172780043562, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 868, "loss": 0.05708640068769455, "loss_ema": 0.12038503556211769, "loss_mean_50": 0.11991764478385449, "loss_std_50": 0.1042137298594699, "lr": 0.00025, "grad_norm_before_clip": 0.02294921875, "grad_clipped": false, "step_seconds": 1.8903686839985312, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 869, "loss": 0.08687733858823776, "loss_ema": 0.11870965071342368, "loss_mean_50": 0.12059455208480357, "loss_std_50": 0.1038864893239047, "lr": 0.00025, "grad_norm_before_clip": 0.0172119140625, "grad_clipped": false, "step_seconds": 1.8291817540011834, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 870, "loss": 0.0549699068069458, "loss_ema": 0.11552266351809978, "loss_mean_50": 0.11182639174163342, "loss_std_50": 0.08956732139896206, "lr": 0.00025, "grad_norm_before_clip": 0.0169677734375, "grad_clipped": false, "step_seconds": 2.062593995993666, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 871, "loss": 0.08030235767364502, "loss_ema": 0.11376164822587705, "loss_mean_50": 0.11157811723649502, "loss_std_50": 0.08963712756875007, "lr": 0.00025, "grad_norm_before_clip": 0.038818359375, "grad_clipped": false, "step_seconds": 0.4208538930033683, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 872, "loss": 0.06123838573694229, "loss_ema": 0.1111354851014303, "loss_mean_50": 0.11088189773261548, "loss_std_50": 0.08988987063668394, "lr": 0.00025, "grad_norm_before_clip": 0.01318359375, "grad_clipped": false, "step_seconds": 2.062268779001897, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 873, "loss": 0.1730319708585739, "loss_ema": 0.11423030938928748, "loss_mean_50": 0.11209100566804409, "loss_std_50": 0.09031014391701833, "lr": 0.00025, "grad_norm_before_clip": 0.138671875, "grad_clipped": false, "step_seconds": 0.34770928799844114, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 874, "loss": 0.10453345626592636, "loss_ema": 0.11374546673311942, "loss_mean_50": 0.11323862731456756, "loss_std_50": 0.08984100885516284, "lr": 0.00025, "grad_norm_before_clip": 0.92578125, "grad_clipped": false, "step_seconds": 1.8289240160011104, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 875, "loss": 0.12905436754226685, "loss_ema": 0.11451091177357678, "loss_mean_50": 0.11438488036394119, "loss_std_50": 0.08966970421680077, "lr": 0.00025, "grad_norm_before_clip": 0.01953125, "grad_clipped": false, "step_seconds": 1.2669464529972174, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 876, "loss": 0.2047944962978363, "loss_ema": 0.11902509099978975, "loss_mean_50": 0.11417998939752579, "loss_std_50": 0.08945091744331417, "lr": 0.00025, "grad_norm_before_clip": 0.0150146484375, "grad_clipped": false, "step_seconds": 2.058233048999682, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 877, "loss": 0.09655127674341202, "loss_ema": 0.11790140028697087, "loss_mean_50": 0.11194715782999992, "loss_std_50": 0.08846427610291377, "lr": 0.00025, "grad_norm_before_clip": 0.056884765625, "grad_clipped": false, "step_seconds": 0.3608750350031187, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 878, "loss": 0.1310388743877411, "loss_ema": 0.11855827399200937, "loss_mean_50": 0.11173440977931022, "loss_std_50": 0.08840529592203125, "lr": 0.00025, "grad_norm_before_clip": 0.053466796875, "grad_clipped": false, "step_seconds": 0.6129556369996862, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 879, "loss": 0.12919586896896362, "loss_ema": 0.11909015374085709, "loss_mean_50": 0.11081325814127922, "loss_std_50": 0.08797756819916873, "lr": 0.00025, "grad_norm_before_clip": 0.0167236328125, "grad_clipped": false, "step_seconds": 0.5967627649952192, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 880, "loss": 0.16959919035434723, "loss_ema": 0.12161560557153159, "loss_mean_50": 0.11311737656593322, "loss_std_50": 0.08797837778584143, "lr": 0.00025, "grad_norm_before_clip": 0.0296630859375, "grad_clipped": false, "step_seconds": 0.4797312850059825, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 881, "loss": 0.1868332028388977, "loss_ema": 0.12487648543489989, "loss_mean_50": 0.1153206318616867, "loss_std_50": 0.08841636442565474, "lr": 0.00025, "grad_norm_before_clip": 0.0130615234375, "grad_clipped": false, "step_seconds": 2.0625854049940244, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 882, "loss": 0.06263971328735352, "loss_ema": 0.12176464682752257, "loss_mean_50": 0.11569266468286514, "loss_std_50": 0.08815439074322669, "lr": 0.00025, "grad_norm_before_clip": 0.03173828125, "grad_clipped": false, "step_seconds": 0.5733294820020092, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 883, "loss": 0.27191483974456787, "loss_ema": 0.12927215647337484, "loss_mean_50": 0.12008577935397625, "loss_std_50": 0.09033010673940492, "lr": 0.00025, "grad_norm_before_clip": 0.0263671875, "grad_clipped": false, "step_seconds": 1.292764474004798, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 884, "loss": 0.7235211730003357, "loss_ema": 0.15898460729972286, "loss_mean_50": 0.13178973220288753, "loss_std_50": 0.12368742396051524, "lr": 0.00025, "grad_norm_before_clip": 0.146484375, "grad_clipped": false, "step_seconds": 1.5711410679941764, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 885, "loss": 0.05120893195271492, "loss_ema": 0.15359582353237244, "loss_mean_50": 0.13168936140835286, "loss_std_50": 0.12375072110805162, "lr": 0.00025, "grad_norm_before_clip": 0.0194091796875, "grad_clipped": false, "step_seconds": 1.1028366070022457, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 886, "loss": 0.06616859883069992, "loss_ema": 0.1492244622972888, "loss_mean_50": 0.131770883128047, "loss_std_50": 0.12370618136920601, "lr": 0.00025, "grad_norm_before_clip": 0.01470947265625, "grad_clipped": false, "step_seconds": 0.6711414980018162, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 887, "loss": 0.054689787328243256, "loss_ema": 0.14449772854883652, "loss_mean_50": 0.13105977021157741, "loss_std_50": 0.12404457373883859, "lr": 0.00025, "grad_norm_before_clip": 0.0625, "grad_clipped": false, "step_seconds": 1.586278611997841, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 888, "loss": 0.9069497585296631, "loss_ema": 0.18262033004787784, "loss_mean_50": 0.14340458415448665, "loss_std_50": 0.16361960570304032, "lr": 0.00025, "grad_norm_before_clip": 0.181640625, "grad_clipped": false, "step_seconds": 0.7070595620025415, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 889, "loss": 0.06870382279157639, "loss_ema": 0.17692450468506277, "loss_mean_50": 0.13980132557451724, "loss_std_50": 0.1632407749214038, "lr": 0.00025, "grad_norm_before_clip": 0.0238037109375, "grad_clipped": false, "step_seconds": 2.067599374000565, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 890, "loss": 0.7617883682250977, "loss_ema": 0.2061676978620645, "loss_mean_50": 0.15346402041614055, "loss_std_50": 0.18472535875779567, "lr": 0.00025, "grad_norm_before_clip": 0.275390625, "grad_clipped": false, "step_seconds": 1.899571110996476, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 891, "loss": 0.05988208204507828, "loss_ema": 0.19885341707121518, "loss_mean_50": 0.15099852420389653, "loss_std_50": 0.1851348096581809, "lr": 0.00025, "grad_norm_before_clip": 0.0155029296875, "grad_clipped": false, "step_seconds": 1.8918947339989245, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 892, "loss": 0.07487955689430237, "loss_ema": 0.19265472406236953, "loss_mean_50": 0.15145125664770603, "loss_std_50": 0.18492031093160277, "lr": 0.00025, "grad_norm_before_clip": 0.0225830078125, "grad_clipped": false, "step_seconds": 0.5747738889986067, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 893, "loss": 0.06486056745052338, "loss_ema": 0.1862650162317772, "loss_mean_50": 0.15062581233680247, "loss_std_50": 0.18521264455959846, "lr": 0.00025, "grad_norm_before_clip": 0.01904296875, "grad_clipped": false, "step_seconds": 1.8964978860021802, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 894, "loss": 0.1736326813697815, "loss_ema": 0.18563339948867738, "loss_mean_50": 0.15302396215498448, "loss_std_50": 0.18471806704343824, "lr": 0.00025, "grad_norm_before_clip": 0.009521484375, "grad_clipped": false, "step_seconds": 2.0697578500039526, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 895, "loss": 0.06369969993829727, "loss_ema": 0.17953671451115835, "loss_mean_50": 0.15347602851688863, "loss_std_50": 0.18447108375024576, "lr": 0.00025, "grad_norm_before_clip": 0.020751953125, "grad_clipped": false, "step_seconds": 2.062978176996694, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 896, "loss": 0.03900986164808273, "loss_ema": 0.17251037186800455, "loss_mean_50": 0.15315277226269244, "loss_std_50": 0.18465712908999443, "lr": 0.00025, "grad_norm_before_clip": 0.0223388671875, "grad_clipped": false, "step_seconds": 1.8309386289984104, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 897, "loss": 0.10382568836212158, "loss_ema": 0.16907613769271038, "loss_mean_50": 0.1539455973356962, "loss_std_50": 0.18435829971702994, "lr": 0.00025, "grad_norm_before_clip": 0.0167236328125, "grad_clipped": false, "step_seconds": 1.2946188159985468, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 898, "loss": 0.05829494073987007, "loss_ema": 0.16353707784506835, "loss_mean_50": 0.1540328711271286, "loss_std_50": 0.18431196008723308, "lr": 0.00025, "grad_norm_before_clip": 0.0269775390625, "grad_clipped": false, "step_seconds": 2.0639360720015247, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 899, "loss": 0.08458691835403442, "loss_ema": 0.15958956987051665, "loss_mean_50": 0.15470874592661857, "loss_std_50": 0.18399382584318452, "lr": 0.00025, "grad_norm_before_clip": 0.1064453125, "grad_clipped": false, "step_seconds": 0.6024199349994888, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 900, "loss": 0.04603944718837738, "loss_ema": 0.15391206373640967, "loss_mean_50": 0.15253574386239052, "loss_std_50": 0.18462173851645106, "lr": 0.00025, "grad_norm_before_clip": 0.053955078125, "grad_clipped": false, "step_seconds": 2.0668059250019724, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 901, "loss": 0.13128399848937988, "loss_ema": 0.15278066047405817, "loss_mean_50": 0.1538124266266823, "loss_std_50": 0.1842492794989075, "lr": 0.00025, "grad_norm_before_clip": 0.060302734375, "grad_clipped": false, "step_seconds": 1.8941263079977944, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 902, "loss": 0.07228254526853561, "loss_ema": 0.14875575471378202, "loss_mean_50": 0.14660776630043984, "loss_std_50": 0.1802091106128433, "lr": 0.00025, "grad_norm_before_clip": 0.00927734375, "grad_clipped": false, "step_seconds": 2.063871016005578, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 903, "loss": 0.16446919739246368, "loss_ema": 0.14954142684771612, "loss_mean_50": 0.14606906697154046, "loss_std_50": 0.18011462911344067, "lr": 0.00025, "grad_norm_before_clip": 0.029296875, "grad_clipped": false, "step_seconds": 0.7224538059963379, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 904, "loss": 0.0857699066400528, "loss_ema": 0.14635285083733296, "loss_mean_50": 0.14664247043430806, "loss_std_50": 0.1798759565865137, "lr": 0.00025, "grad_norm_before_clip": 0.0269775390625, "grad_clipped": false, "step_seconds": 0.5801291140014655, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 905, "loss": 0.43010732531547546, "loss_ema": 0.1605405745612401, "loss_mean_50": 0.15375506378710269, "loss_std_50": 0.18386859995322094, "lr": 0.00025, "grad_norm_before_clip": 0.0296630859375, "grad_clipped": false, "step_seconds": 2.0668895979979425, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 906, "loss": 0.0794018805027008, "loss_ema": 0.15648363985831312, "loss_mean_50": 0.1540324641019106, "loss_std_50": 0.1837457111286956, "lr": 0.00025, "grad_norm_before_clip": 0.041748046875, "grad_clipped": false, "step_seconds": 2.065330753001035, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 907, "loss": 0.4577852487564087, "loss_ema": 0.1715487203032179, "loss_mean_50": 0.15642189987003804, "loss_std_50": 0.18687670354285268, "lr": 0.00025, "grad_norm_before_clip": 0.051025390625, "grad_clipped": false, "step_seconds": 0.6071676229985314, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 908, "loss": 0.07856129109859467, "loss_ema": 0.16689934884298674, "loss_mean_50": 0.15638453625142573, "loss_std_50": 0.18689207965343413, "lr": 0.00025, "grad_norm_before_clip": 0.011962890625, "grad_clipped": false, "step_seconds": 1.8937719489986193, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 909, "loss": 0.054560113698244095, "loss_ema": 0.1612823870857496, "loss_mean_50": 0.14985160529613495, "loss_std_50": 0.1846142864222423, "lr": 0.00025, "grad_norm_before_clip": 0.0201416015625, "grad_clipped": false, "step_seconds": 1.2950223880034173, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 910, "loss": 0.05849403887987137, "loss_ema": 0.15614296967545568, "loss_mean_50": 0.1496570010483265, "loss_std_50": 0.18470533419667473, "lr": 0.00025, "grad_norm_before_clip": 0.0223388671875, "grad_clipped": false, "step_seconds": 2.069463776999328, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 911, "loss": 0.16018840670585632, "loss_ema": 0.1563452415269757, "loss_mean_50": 0.15168203487992288, "loss_std_50": 0.184254103138118, "lr": 0.00025, "grad_norm_before_clip": 0.0225830078125, "grad_clipped": false, "step_seconds": 1.8932921099985833, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 912, "loss": 0.5866362452507019, "loss_ema": 0.17785979171316202, "loss_mean_50": 0.1616611361503601, "loss_std_50": 0.19378279897702633, "lr": 0.00025, "grad_norm_before_clip": 0.041748046875, "grad_clipped": false, "step_seconds": 1.8929562750054174, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 913, "loss": 0.047108959406614304, "loss_ema": 0.17132225009783464, "loss_mean_50": 0.16078813470900058, "loss_std_50": 0.19419812795915292, "lr": 0.00025, "grad_norm_before_clip": 0.0220947265625, "grad_clipped": false, "step_seconds": 1.5888757669963525, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 914, "loss": 0.05823059380054474, "loss_ema": 0.16566766728297014, "loss_mean_50": 0.1607606313377619, "loss_std_50": 0.19421255284062997, "lr": 0.00025, "grad_norm_before_clip": 0.0181884765625, "grad_clipped": false, "step_seconds": 1.2710695229980047, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 915, "loss": 1.0093183517456055, "loss_ema": 0.2078502015061019, "loss_mean_50": 0.17963680006563665, "loss_std_50": 0.22711617303571735, "lr": 0.00025, "grad_norm_before_clip": 0.2890625, "grad_clipped": false, "step_seconds": 2.067806108003424, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 916, "loss": 0.7634589672088623, "loss_ema": 0.23563063979123988, "loss_mean_50": 0.19419434651732445, "loss_std_50": 0.2403575723718093, "lr": 0.00025, "grad_norm_before_clip": 0.046630859375, "grad_clipped": false, "step_seconds": 2.0616804559977027, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 917, "loss": 0.1236567422747612, "loss_ema": 0.23003194491541595, "loss_mean_50": 0.1952543468773365, "loss_std_50": 0.239926903838986, "lr": 0.00025, "grad_norm_before_clip": 0.019775390625, "grad_clipped": false, "step_seconds": 1.8929940369998803, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 918, "loss": 0.070503368973732, "loss_ema": 0.22205551611833177, "loss_mean_50": 0.19552268624305724, "loss_std_50": 0.23977968154624707, "lr": 0.00025, "grad_norm_before_clip": 0.0908203125, "grad_clipped": false, "step_seconds": 1.8317274939981871, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 919, "loss": 0.10757499933242798, "loss_ema": 0.2163314902790366, "loss_mean_50": 0.19593663945794104, "loss_std_50": 0.2396095657657962, "lr": 0.00025, "grad_norm_before_clip": 0.01043701171875, "grad_clipped": false, "step_seconds": 1.1066541300024255, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 920, "loss": 0.08033687621355057, "loss_ema": 0.20953175957576226, "loss_mean_50": 0.19644397884607315, "loss_std_50": 0.23933725227004657, "lr": 0.00025, "grad_norm_before_clip": 0.01165771484375, "grad_clipped": false, "step_seconds": 2.0690152020033565, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 921, "loss": 0.6342892050743103, "loss_ema": 0.23076963185068966, "loss_mean_50": 0.20752371579408646, "loss_std_50": 0.24642230230355502, "lr": 0.00025, "grad_norm_before_clip": 0.2001953125, "grad_clipped": false, "step_seconds": 2.062320158998773, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 922, "loss": 0.3357253968715668, "loss_ema": 0.23601742010173352, "loss_mean_50": 0.21301345601677896, "loss_std_50": 0.24615958094231333, "lr": 0.00025, "grad_norm_before_clip": 0.03173828125, "grad_clipped": false, "step_seconds": 0.5210440759983612, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 923, "loss": 0.27923935651779175, "loss_ema": 0.23817851692253642, "loss_mean_50": 0.2151376037299633, "loss_std_50": 0.24626362764529838, "lr": 0.00025, "grad_norm_before_clip": 0.048583984375, "grad_clipped": false, "step_seconds": 0.5980343310002354, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 924, "loss": 0.06204424425959587, "loss_ema": 0.2293718032893894, "loss_mean_50": 0.2142878194898367, "loss_std_50": 0.24671671640430232, "lr": 0.00025, "grad_norm_before_clip": 0.045654296875, "grad_clipped": false, "step_seconds": 0.6475954950001324, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 925, "loss": 0.10065873712301254, "loss_ema": 0.22293614998107056, "loss_mean_50": 0.2137199068814516, "loss_std_50": 0.2469448362877076, "lr": 0.00025, "grad_norm_before_clip": 0.044677734375, "grad_clipped": false, "step_seconds": 0.6756347510017804, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 926, "loss": 0.9777761697769165, "loss_ema": 0.2606781509708629, "loss_mean_50": 0.22917954035103322, "loss_std_50": 0.2691036939554395, "lr": 0.00025, "grad_norm_before_clip": 0.185546875, "grad_clipped": false, "step_seconds": 1.8968383199971868, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 927, "loss": 0.06848757714033127, "loss_ema": 0.2510686222793363, "loss_mean_50": 0.2286182663589716, "loss_std_50": 0.2694088270391812, "lr": 0.00025, "grad_norm_before_clip": 0.02197265625, "grad_clipped": false, "step_seconds": 1.8980552279972471, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 928, "loss": 0.04885001480579376, "loss_ema": 0.24095769190565913, "loss_mean_50": 0.22697448916733265, "loss_std_50": 0.27024861112270665, "lr": 0.00025, "grad_norm_before_clip": 0.031494140625, "grad_clipped": false, "step_seconds": 2.064585863001412, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 929, "loss": 0.062009572982788086, "loss_ema": 0.23201028595951556, "loss_mean_50": 0.22563076324760914, "loss_std_50": 0.2708976956610782, "lr": 0.00025, "grad_norm_before_clip": 0.06787109375, "grad_clipped": false, "step_seconds": 0.6163065910004661, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 930, "loss": 0.10428240895271301, "loss_ema": 0.2256238921091754, "loss_mean_50": 0.22432442761957647, "loss_std_50": 0.2713218987112776, "lr": 0.00025, "grad_norm_before_clip": 0.00927734375, "grad_clipped": false, "step_seconds": 1.5698086940028588, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 931, "loss": 0.7743722200393677, "loss_ema": 0.253061308505685, "loss_mean_50": 0.23607520796358586, "loss_std_50": 0.2819582088214604, "lr": 0.00025, "grad_norm_before_clip": 0.0556640625, "grad_clipped": false, "step_seconds": 0.36523408599896356, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 932, "loss": 0.6021243333816528, "loss_ema": 0.27051445974948335, "loss_mean_50": 0.24686490036547185, "loss_std_50": 0.28541593488765443, "lr": 0.00025, "grad_norm_before_clip": 0.158203125, "grad_clipped": false, "step_seconds": 2.0651369449988124, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 933, "loss": 0.05799867957830429, "loss_ema": 0.25988867074092437, "loss_mean_50": 0.24258657716214657, "loss_std_50": 0.28660916041918877, "lr": 0.00025, "grad_norm_before_clip": 0.01531982421875, "grad_clipped": false, "step_seconds": 1.2939380230018287, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 934, "loss": 0.04362896829843521, "loss_ema": 0.2490756856187999, "loss_mean_50": 0.22898873306810855, "loss_std_50": 0.2795096248167511, "lr": 0.00025, "grad_norm_before_clip": 0.01470947265625, "grad_clipped": false, "step_seconds": 1.8340462920023128, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 935, "loss": 0.04630941152572632, "loss_ema": 0.23893737191414618, "loss_mean_50": 0.2288907426595688, "loss_std_50": 0.2795727853346458, "lr": 0.00025, "grad_norm_before_clip": 0.044921875, "grad_clipped": false, "step_seconds": 2.0689319819939556, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 936, "loss": 0.07014989852905273, "loss_ema": 0.2304979982448915, "loss_mean_50": 0.22897036865353584, "loss_std_50": 0.2795269918085428, "lr": 0.00025, "grad_norm_before_clip": 0.04248046875, "grad_clipped": false, "step_seconds": 2.06258181600424, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 937, "loss": 0.042941123247146606, "loss_ema": 0.22112015449500425, "loss_mean_50": 0.2287353953719139, "loss_std_50": 0.2796782921849278, "lr": 0.00025, "grad_norm_before_clip": 0.01483154296875, "grad_clipped": false, "step_seconds": 2.063125136999588, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 938, "loss": 0.0548686645925045, "loss_ema": 0.21280757999987926, "loss_mean_50": 0.21169377349317073, "loss_std_50": 0.2633146949883138, "lr": 0.00025, "grad_norm_before_clip": 0.0230712890625, "grad_clipped": false, "step_seconds": 0.4752517910019378, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 939, "loss": 0.10029657185077667, "loss_ema": 0.2071820295924241, "loss_mean_50": 0.21232562847435474, "loss_std_50": 0.26300854276375457, "lr": 0.00025, "grad_norm_before_clip": 0.036376953125, "grad_clipped": false, "step_seconds": 0.3509534429977066, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 940, "loss": 0.09666800498962402, "loss_ema": 0.2016563283622841, "loss_mean_50": 0.19902322120964527, "loss_std_50": 0.2514475833997861, "lr": 0.00025, "grad_norm_before_clip": 0.01190185546875, "grad_clipped": false, "step_seconds": 1.833451917002094, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 941, "loss": 0.5150150060653687, "loss_ema": 0.2173242622474383, "loss_mean_50": 0.20812587969005109, "loss_std_50": 0.25446579305664513, "lr": 0.00025, "grad_norm_before_clip": 0.0341796875, "grad_clipped": false, "step_seconds": 1.8907125839978107, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 942, "loss": 0.12691457569599152, "loss_ema": 0.21280377791986596, "loss_mean_50": 0.20916658006608485, "loss_std_50": 0.25402474416054666, "lr": 0.00025, "grad_norm_before_clip": 0.1435546875, "grad_clipped": false, "step_seconds": 0.7327917959992192, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 943, "loss": 0.07388008385896683, "loss_ema": 0.205857593216821, "loss_mean_50": 0.20934697039425373, "loss_std_50": 0.2539253873104078, "lr": 0.00025, "grad_norm_before_clip": 0.01373291015625, "grad_clipped": false, "step_seconds": 2.069730620001792, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 944, "loss": 0.9034655094146729, "loss_ema": 0.24073798902671362, "loss_mean_50": 0.22394362695515155, "loss_std_50": 0.2718005530778224, "lr": 0.00025, "grad_norm_before_clip": 0.06103515625, "grad_clipped": false, "step_seconds": 1.8962746559991501, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 945, "loss": 0.11851542443037033, "loss_ema": 0.23462686079689646, "loss_mean_50": 0.225039941444993, "loss_std_50": 0.271262010872857, "lr": 0.00025, "grad_norm_before_clip": 0.0103759765625, "grad_clipped": false, "step_seconds": 2.059235216998786, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 946, "loss": 0.06487910449504852, "loss_ema": 0.22613947298180406, "loss_mean_50": 0.22555732630193234, "loss_std_50": 0.2709311663831997, "lr": 0.00025, "grad_norm_before_clip": 0.0322265625, "grad_clipped": false, "step_seconds": 0.5705235970017384, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 947, "loss": 0.27288690209388733, "loss_ema": 0.2284768444374082, "loss_mean_50": 0.22893855057656765, "loss_std_50": 0.27044536265491487, "lr": 0.00025, "grad_norm_before_clip": 0.0615234375, "grad_clipped": false, "step_seconds": 0.3725726359989494, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 948, "loss": 0.03882576897740364, "loss_ema": 0.21899429066440795, "loss_mean_50": 0.22854916714131832, "loss_std_50": 0.27070466402534366, "lr": 0.00025, "grad_norm_before_clip": 0.0233154296875, "grad_clipped": false, "step_seconds": 1.8327883449965157, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 949, "loss": 0.20627541840076447, "loss_ema": 0.21835834705122575, "loss_mean_50": 0.23098293714225293, "loss_std_50": 0.2699453865225299, "lr": 0.00025, "grad_norm_before_clip": 0.01300048828125, "grad_clipped": false, "step_seconds": 2.065193292000913, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 950, "loss": 0.06603071093559265, "loss_ema": 0.2107419652454441, "loss_mean_50": 0.23138276241719724, "loss_std_50": 0.26968584444429383, "lr": 0.00025, "grad_norm_before_clip": 0.016845703125, "grad_clipped": false, "step_seconds": 0.6014779790057219, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 951, "loss": 0.06917484104633331, "loss_ema": 0.20366360903548855, "loss_mean_50": 0.2301405792683363, "loss_std_50": 0.2702864121981868, "lr": 0.00025, "grad_norm_before_clip": 0.0625, "grad_clipped": false, "step_seconds": 2.0639877909998177, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 952, "loss": 0.5057430863380432, "loss_ema": 0.21876758290061626, "loss_mean_50": 0.23880979008972644, "loss_std_50": 0.2720300370598786, "lr": 0.00025, "grad_norm_before_clip": 0.031005859375, "grad_clipped": false, "step_seconds": 0.571672073994705, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 953, "loss": 0.04228859394788742, "loss_ema": 0.2099436334529798, "loss_mean_50": 0.23636617802083493, "loss_std_50": 0.2732329609674947, "lr": 0.00025, "grad_norm_before_clip": 0.006805419921875, "grad_clipped": false, "step_seconds": 2.062583648999862, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 954, "loss": 0.09625948220491409, "loss_ema": 0.20425942589057652, "loss_mean_50": 0.23657596953213214, "loss_std_50": 0.2731212549981409, "lr": 0.00025, "grad_norm_before_clip": 0.025146484375, "grad_clipped": false, "step_seconds": 0.724245706005604, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 955, "loss": 0.23181591928005219, "loss_ema": 0.2056372505600503, "loss_mean_50": 0.2326101414114237, "loss_std_50": 0.2717183423827287, "lr": 0.00025, "grad_norm_before_clip": 0.0390625, "grad_clipped": false, "step_seconds": 0.6130531699964195, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 956, "loss": 0.07838090509176254, "loss_ema": 0.1992744332866359, "loss_mean_50": 0.23258972190320493, "loss_std_50": 0.27172989326307634, "lr": 0.00025, "grad_norm_before_clip": 0.03271484375, "grad_clipped": false, "step_seconds": 0.6007729679986369, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 957, "loss": 0.20215918123722076, "loss_ema": 0.19941867068416516, "loss_mean_50": 0.22747720055282117, "loss_std_50": 0.2698430226409706, "lr": 0.00025, "grad_norm_before_clip": 0.0177001953125, "grad_clipped": false, "step_seconds": 1.2692329409983358, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 958, "loss": 0.04988298937678337, "loss_ema": 0.19194188661879605, "loss_mean_50": 0.22690363451838494, "loss_std_50": 0.27018919853523515, "lr": 0.00025, "grad_norm_before_clip": 0.041015625, "grad_clipped": false, "step_seconds": 1.2954043660065508, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 959, "loss": 0.0902898758649826, "loss_ema": 0.18685928608110539, "loss_mean_50": 0.2276182297617197, "loss_std_50": 0.26977937830094695, "lr": 0.00025, "grad_norm_before_clip": 0.0172119140625, "grad_clipped": false, "step_seconds": 1.8926602889987407, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 960, "loss": 0.10118518769741058, "loss_ema": 0.18257558116192063, "loss_mean_50": 0.22847205273807047, "loss_std_50": 0.2693099151283785, "lr": 0.00025, "grad_norm_before_clip": 0.01239013671875, "grad_clipped": false, "step_seconds": 1.586165937995247, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 961, "loss": 0.08043725788593292, "loss_ema": 0.17746866499812125, "loss_mean_50": 0.226877029761672, "loss_std_50": 0.2699450299071802, "lr": 0.00025, "grad_norm_before_clip": 0.017333984375, "grad_clipped": false, "step_seconds": 0.524670712002262, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 962, "loss": 0.053586650639772415, "loss_ema": 0.1712745642802038, "loss_mean_50": 0.21621603786945343, "loss_std_50": 0.26602390808121185, "lr": 0.00025, "grad_norm_before_clip": 0.01611328125, "grad_clipped": false, "step_seconds": 2.0680139679971035, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 963, "loss": 0.09160993248224258, "loss_ema": 0.16729133269030574, "loss_mean_50": 0.217106057330966, "loss_std_50": 0.26553063304716806, "lr": 0.00025, "grad_norm_before_clip": 0.04443359375, "grad_clipped": false, "step_seconds": 2.066553749005834, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 964, "loss": 0.05160289630293846, "loss_ema": 0.16150691087093738, "loss_mean_50": 0.21697350338101387, "loss_std_50": 0.26561155318217133, "lr": 0.00025, "grad_norm_before_clip": 0.016357421875, "grad_clipped": false, "step_seconds": 2.07049130499945, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 965, "loss": 0.05645991861820221, "loss_ema": 0.15625456125830062, "loss_mean_50": 0.19791633471846581, "loss_std_50": 0.24113358652349248, "lr": 0.00025, "grad_norm_before_clip": 0.0267333984375, "grad_clipped": false, "step_seconds": 2.0626689400014584, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 966, "loss": 0.11925458908081055, "loss_ema": 0.15440456264942612, "loss_mean_50": 0.18503224715590477, "loss_std_50": 0.22739039306579076, "lr": 0.00025, "grad_norm_before_clip": 0.080078125, "grad_clipped": false, "step_seconds": 2.062806918998831, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 967, "loss": 0.15956230461597443, "loss_ema": 0.15466244974775353, "loss_mean_50": 0.18575035840272902, "loss_std_50": 0.2272520857601167, "lr": 0.00025, "grad_norm_before_clip": 0.02587890625, "grad_clipped": false, "step_seconds": 1.8970129919980536, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 968, "loss": 0.07591142505407333, "loss_ema": 0.1507248985130695, "loss_mean_50": 0.18585851952433585, "loss_std_50": 0.2271984886341229, "lr": 0.00025, "grad_norm_before_clip": 0.032470703125, "grad_clipped": false, "step_seconds": 1.8334162620012648, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 969, "loss": 0.05961833521723747, "loss_ema": 0.1461695703482779, "loss_mean_50": 0.18489938624203206, "loss_std_50": 0.22762776366906898, "lr": 0.00025, "grad_norm_before_clip": 0.0177001953125, "grad_clipped": false, "step_seconds": 2.0671329329998116, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 970, "loss": 0.07335907965898514, "loss_ema": 0.14252904581381326, "loss_mean_50": 0.18475983031094073, "loss_std_50": 0.2276939563249479, "lr": 0.00025, "grad_norm_before_clip": 0.038818359375, "grad_clipped": false, "step_seconds": 1.8929081019959995, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 971, "loss": 0.052192188799381256, "loss_ema": 0.13801220296309166, "loss_mean_50": 0.17311788998544217, "loss_std_50": 0.21913227372612334, "lr": 0.00025, "grad_norm_before_clip": 0.016357421875, "grad_clipped": false, "step_seconds": 1.106039280006371, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 972, "loss": 0.07017145305871964, "loss_ema": 0.13462016546787303, "loss_mean_50": 0.16780681110918522, "loss_std_50": 0.2183434935626765, "lr": 0.00025, "grad_norm_before_clip": 0.048583984375, "grad_clipped": false, "step_seconds": 1.8940183039958356, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 973, "loss": 0.13294321298599243, "loss_ema": 0.134536317843779, "loss_mean_50": 0.16488088823854924, "loss_std_50": 0.21781020481709926, "lr": 0.00025, "grad_norm_before_clip": 0.01446533203125, "grad_clipped": false, "step_seconds": 1.89325287900283, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 974, "loss": 0.13121101260185242, "loss_ema": 0.13437005258168266, "loss_mean_50": 0.16626422360539436, "loss_std_50": 0.21737188734825, "lr": 0.00025, "grad_norm_before_clip": 0.0145263671875, "grad_clipped": false, "step_seconds": 1.8314483299982385, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 975, "loss": 0.0670793354511261, "loss_ema": 0.13100551672515484, "loss_mean_50": 0.16559263557195664, "loss_std_50": 0.21762526880508465, "lr": 0.00025, "grad_norm_before_clip": 0.01025390625, "grad_clipped": false, "step_seconds": 1.294298417000391, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 976, "loss": 0.07855906337499619, "loss_ema": 0.1283831940576469, "loss_mean_50": 0.14760829344391824, "loss_std_50": 0.18437997735605968, "lr": 0.00025, "grad_norm_before_clip": 0.04736328125, "grad_clipped": false, "step_seconds": 1.8971260379985324, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 977, "loss": 0.06477489322423935, "loss_ema": 0.1252027790159765, "loss_mean_50": 0.1475340397655964, "loss_std_50": 0.18441257068075254, "lr": 0.00025, "grad_norm_before_clip": 0.01025390625, "grad_clipped": false, "step_seconds": 2.0719764930036035, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 978, "loss": 0.037161167711019516, "loss_ema": 0.12080069845072866, "loss_mean_50": 0.1473002628237009, "loss_std_50": 0.18454488412169862, "lr": 0.00025, "grad_norm_before_clip": 0.00823974609375, "grad_clipped": false, "step_seconds": 2.0670515229940065, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 979, "loss": 0.09165693074464798, "loss_ema": 0.11934351006542462, "loss_mean_50": 0.1478932099789381, "loss_std_50": 0.1843173791094825, "lr": 0.00025, "grad_norm_before_clip": 0.01190185546875, "grad_clipped": false, "step_seconds": 2.063625620001403, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 980, "loss": 0.06677556782960892, "loss_ema": 0.11671511295363383, "loss_mean_50": 0.147143073156476, "loss_std_50": 0.18456949080108184, "lr": 0.00025, "grad_norm_before_clip": 0.0255126953125, "grad_clipped": false, "step_seconds": 0.4761865559994476, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 981, "loss": 0.21444149315357208, "loss_ema": 0.12160143196363074, "loss_mean_50": 0.13594445861876012, "loss_std_50": 0.16174900760035468, "lr": 0.00025, "grad_norm_before_clip": 0.0303955078125, "grad_clipped": false, "step_seconds": 0.42204902799858246, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 982, "loss": 0.11621125042438507, "loss_ema": 0.12133192288666846, "loss_mean_50": 0.12622619695961476, "loss_std_50": 0.14740967175683523, "lr": 0.00025, "grad_norm_before_clip": 0.01446533203125, "grad_clipped": false, "step_seconds": 0.6682416380062932, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 983, "loss": 0.1309727281332016, "loss_ema": 0.1218139631489951, "loss_mean_50": 0.1276856779307127, "loss_std_50": 0.147087837537154, "lr": 0.00025, "grad_norm_before_clip": 0.060546875, "grad_clipped": false, "step_seconds": 0.35128278999764007, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 984, "loss": 0.3359029293060303, "loss_ema": 0.13251841145684684, "loss_mean_50": 0.1335311571508646, "loss_std_50": 0.14942034731175244, "lr": 0.00025, "grad_norm_before_clip": 0.024169921875, "grad_clipped": false, "step_seconds": 2.0667498230031924, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 985, "loss": 0.12344580888748169, "loss_ema": 0.1320647813283786, "loss_mean_50": 0.1350738850980997, "loss_std_50": 0.14890917301509496, "lr": 0.00025, "grad_norm_before_clip": 0.0269775390625, "grad_clipped": false, "step_seconds": 1.8901860469995881, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 986, "loss": 0.13559412956237793, "loss_ema": 0.13224124874007856, "loss_mean_50": 0.1363827697187662, "loss_std_50": 0.1486200913791099, "lr": 0.00025, "grad_norm_before_clip": 0.0152587890625, "grad_clipped": false, "step_seconds": 0.5748668140004156, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 987, "loss": 0.12946583330631256, "loss_ema": 0.13210247796839025, "loss_mean_50": 0.13811326391994952, "loss_std_50": 0.14802454865877746, "lr": 0.00025, "grad_norm_before_clip": 0.0111083984375, "grad_clipped": false, "step_seconds": 2.078059993997158, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 988, "loss": 0.06529251486063004, "loss_ema": 0.12876197981300222, "loss_mean_50": 0.13832174092531205, "loss_std_50": 0.14791446009941178, "lr": 0.00025, "grad_norm_before_clip": 0.0306396484375, "grad_clipped": false, "step_seconds": 1.8944463419975364, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 989, "loss": 0.05240008234977722, "loss_ema": 0.12494388493984096, "loss_mean_50": 0.13736381113529206, "loss_std_50": 0.1483121783118377, "lr": 0.00025, "grad_norm_before_clip": 0.04345703125, "grad_clipped": false, "step_seconds": 1.5696906279990799, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 990, "loss": 0.07558568567037582, "loss_ema": 0.1224759749763677, "loss_mean_50": 0.13694216474890708, "loss_std_50": 0.14845717299238007, "lr": 0.00025, "grad_norm_before_clip": 0.025146484375, "grad_clipped": false, "step_seconds": 1.8446458100006566, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 991, "loss": 0.07283887267112732, "loss_ema": 0.11999411986110567, "loss_mean_50": 0.12809864208102226, "loss_std_50": 0.13850894373286587, "lr": 0.00025, "grad_norm_before_clip": 0.037109375, "grad_clipped": false, "step_seconds": 2.063449679000769, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 992, "loss": 0.051452815532684326, "loss_ema": 0.1165670546446846, "loss_mean_50": 0.1265894068777561, "loss_std_50": 0.13892412791656064, "lr": 0.00025, "grad_norm_before_clip": 0.0189208984375, "grad_clipped": false, "step_seconds": 1.8293639429975883, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 993, "loss": 0.054303426295518875, "loss_ema": 0.1134538732272263, "loss_mean_50": 0.12619787372648716, "loss_std_50": 0.1390996039688147, "lr": 0.00025, "grad_norm_before_clip": 0.01080322265625, "grad_clipped": false, "step_seconds": 2.065075379003247, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 994, "loss": 0.050363939255476, "loss_ema": 0.11029937652863879, "loss_mean_50": 0.10913584232330323, "loss_std_50": 0.08420037453158319, "lr": 0.00025, "grad_norm_before_clip": 0.1455078125, "grad_clipped": false, "step_seconds": 1.290436278002744, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 995, "loss": 0.7318474650382996, "loss_ema": 0.14137678095412182, "loss_mean_50": 0.12140248313546181, "loss_std_50": 0.12121414259285582, "lr": 0.00025, "grad_norm_before_clip": 0.08056640625, "grad_clipped": false, "step_seconds": 0.5239606079994701, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 996, "loss": 0.06739123910665512, "loss_ema": 0.13767750386174846, "loss_mean_50": 0.12145272582769394, "loss_std_50": 0.12119122197240728, "lr": 0.00025, "grad_norm_before_clip": 0.0244140625, "grad_clipped": false, "step_seconds": 1.82995297099842, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 997, "loss": 0.06782995909452438, "loss_ema": 0.13418512662338725, "loss_mean_50": 0.11735158696770669, "loss_std_50": 0.1194544045677278, "lr": 0.00025, "grad_norm_before_clip": 0.011962890625, "grad_clipped": false, "step_seconds": 0.6016611270024441, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 998, "loss": 0.22096659243106842, "loss_ema": 0.1385241999137713, "loss_mean_50": 0.12099440343677997, "loss_std_50": 0.11978096661876443, "lr": 0.00025, "grad_norm_before_clip": 0.0111083984375, "grad_clipped": false, "step_seconds": 2.06871585999761, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 999, "loss": 0.050415366888046265, "loss_ema": 0.13411875826248504, "loss_mean_50": 0.1178772024065256, "loss_std_50": 0.11954887703194766, "lr": 0.00025, "grad_norm_before_clip": 0.032470703125, "grad_clipped": false, "step_seconds": 2.0637676780024776, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1000, "loss": 0.11843421310186386, "loss_ema": 0.13333453100445397, "loss_mean_50": 0.11892527244985103, "loss_std_50": 0.1193192380986629, "lr": 0.00025, "grad_norm_before_clip": 0.0267333984375, "grad_clipped": false, "step_seconds": 0.9824844949980616, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1001, "loss": 0.061032429337501526, "loss_ema": 0.12971942592110633, "loss_mean_50": 0.1187624242156744, "loss_std_50": 0.1193925608154626, "lr": 0.00025, "grad_norm_before_clip": 0.039794921875, "grad_clipped": false, "step_seconds": 0.5752152709974325, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1002, "loss": 0.25457194447517395, "loss_ema": 0.1359620518488097, "loss_mean_50": 0.11373900137841701, "loss_std_50": 0.10771793100408597, "lr": 0.00025, "grad_norm_before_clip": 0.021484375, "grad_clipped": false, "step_seconds": 2.0583055860042805, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1003, "loss": 0.08362054824829102, "loss_ema": 0.13334497666878375, "loss_mean_50": 0.11456564046442509, "loss_std_50": 0.10732431475468068, "lr": 0.00025, "grad_norm_before_clip": 0.02685546875, "grad_clipped": false, "step_seconds": 0.7271518419947824, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1004, "loss": 0.17740792036056519, "loss_ema": 0.1355481238533728, "loss_mean_50": 0.11618860922753811, "loss_std_50": 0.10764829420015061, "lr": 0.00025, "grad_norm_before_clip": 0.048095703125, "grad_clipped": false, "step_seconds": 2.065356179999071, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1005, "loss": 0.0828750804066658, "loss_ema": 0.13291447168103745, "loss_mean_50": 0.11320979245007039, "loss_std_50": 0.1064616561127002, "lr": 0.00025, "grad_norm_before_clip": 0.0673828125, "grad_clipped": false, "step_seconds": 0.5712033110030461, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1006, "loss": 0.14652486145496368, "loss_ema": 0.13359499116973375, "loss_mean_50": 0.1145726715773344, "loss_std_50": 0.10644324136039339, "lr": 0.00025, "grad_norm_before_clip": 0.0302734375, "grad_clipped": false, "step_seconds": 0.37034388299798593, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1007, "loss": 0.05529801920056343, "loss_ema": 0.12968014257127525, "loss_mean_50": 0.11163544833660126, "loss_std_50": 0.10601121688642563, "lr": 0.00025, "grad_norm_before_clip": 0.017822265625, "grad_clipped": false, "step_seconds": 2.064664878998883, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1008, "loss": 0.07049839198589325, "loss_ema": 0.12672105504200615, "loss_mean_50": 0.11204775638878345, "loss_std_50": 0.10581014097034001, "lr": 0.00025, "grad_norm_before_clip": 0.017333984375, "grad_clipped": false, "step_seconds": 1.1023232900042785, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1009, "loss": 0.1417938768863678, "loss_ema": 0.12747469613422424, "loss_mean_50": 0.11307783640921115, "loss_std_50": 0.10584400499655197, "lr": 0.00025, "grad_norm_before_clip": 0.024658203125, "grad_clipped": false, "step_seconds": 0.42415328200149816, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1010, "loss": 0.061605941504240036, "loss_ema": 0.12418125840272502, "loss_mean_50": 0.11228625148534775, "loss_std_50": 0.10607773189719391, "lr": 0.00025, "grad_norm_before_clip": 0.0380859375, "grad_clipped": false, "step_seconds": 1.8932879799976945, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1011, "loss": 0.08022338896989822, "loss_ema": 0.12198336493108368, "loss_mean_50": 0.11228197410702706, "loss_std_50": 0.106079020363806, "lr": 0.00025, "grad_norm_before_clip": 0.0203857421875, "grad_clipped": false, "step_seconds": 1.8919643710032688, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1012, "loss": 0.16860486567020416, "loss_ema": 0.12431443996803969, "loss_mean_50": 0.1145823384076357, "loss_std_50": 0.1060283425682562, "lr": 0.00025, "grad_norm_before_clip": 0.0284423828125, "grad_clipped": false, "step_seconds": 1.2940889079982298, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1013, "loss": 0.05953207239508629, "loss_ema": 0.12107532158939201, "loss_mean_50": 0.11394078120589256, "loss_std_50": 0.1062621938377399, "lr": 0.00025, "grad_norm_before_clip": 0.017333984375, "grad_clipped": false, "step_seconds": 2.0619240689993603, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1014, "loss": 0.1131167858839035, "loss_ema": 0.12067739480411759, "loss_mean_50": 0.11517105899751186, "loss_std_50": 0.10588877937367676, "lr": 0.00025, "grad_norm_before_clip": 0.0228271484375, "grad_clipped": false, "step_seconds": 1.589723862001847, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1015, "loss": 0.7262353301048279, "loss_ema": 0.1509552915691531, "loss_mean_50": 0.12856656722724438, "loss_std_50": 0.13576466721831526, "lr": 0.00025, "grad_norm_before_clip": 0.04541015625, "grad_clipped": false, "step_seconds": 2.0716905359950033, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1016, "loss": 0.04616658762097359, "loss_ema": 0.14571585637174414, "loss_mean_50": 0.12710480719804765, "loss_std_50": 0.13624965686767512, "lr": 0.00025, "grad_norm_before_clip": 0.038818359375, "grad_clipped": false, "step_seconds": 2.0722397659992566, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1017, "loss": 0.07199878245592117, "loss_ema": 0.14203000267595298, "loss_mean_50": 0.12535353675484656, "loss_std_50": 0.13638389107544513, "lr": 0.00025, "grad_norm_before_clip": 0.015625, "grad_clipped": false, "step_seconds": 2.065283359006571, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1018, "loss": 0.052177220582962036, "loss_ema": 0.13753736357130342, "loss_mean_50": 0.12487885266542435, "loss_std_50": 0.1365962863209511, "lr": 0.00025, "grad_norm_before_clip": 0.018798828125, "grad_clipped": false, "step_seconds": 1.5695121299941093, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1019, "loss": 0.08698020875453949, "loss_ema": 0.13500950583046523, "loss_mean_50": 0.1254260901361704, "loss_std_50": 0.1363883917687992, "lr": 0.00025, "grad_norm_before_clip": 0.043701171875, "grad_clipped": false, "step_seconds": 0.3519878709994373, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1020, "loss": 0.07594111561775208, "loss_ema": 0.13205608631982957, "loss_mean_50": 0.12547773085534572, "loss_std_50": 0.13636915532838323, "lr": 0.00025, "grad_norm_before_clip": 0.00970458984375, "grad_clipped": false, "step_seconds": 1.894436491005763, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1021, "loss": 0.12300892174243927, "loss_ema": 0.13160372809096005, "loss_mean_50": 0.1268940655142069, "loss_std_50": 0.13596781607421224, "lr": 0.00025, "grad_norm_before_clip": 0.051513671875, "grad_clipped": false, "step_seconds": 1.8898005209994153, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1022, "loss": 0.08189857006072998, "loss_ema": 0.12911847018944853, "loss_mean_50": 0.1271286078542471, "loss_std_50": 0.1358798542640818, "lr": 0.00025, "grad_norm_before_clip": 0.04541015625, "grad_clipped": false, "step_seconds": 0.4792235439963406, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1023, "loss": 0.599463701248169, "loss_ema": 0.15263573174238454, "loss_mean_50": 0.1364590176194906, "loss_std_50": 0.15112117940518097, "lr": 0.00025, "grad_norm_before_clip": 0.025390625, "grad_clipped": false, "step_seconds": 1.8896731019995059, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1024, "loss": 0.1975201666355133, "loss_ema": 0.15487995348704095, "loss_mean_50": 0.13778520070016384, "loss_std_50": 0.15136006920750023, "lr": 0.00025, "grad_norm_before_clip": 0.0159912109375, "grad_clipped": false, "step_seconds": 0.6715050019993214, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1025, "loss": 0.0843699723482132, "loss_ema": 0.15135445443009957, "loss_mean_50": 0.13813101343810558, "loss_std_50": 0.15121781743179294, "lr": 0.00025, "grad_norm_before_clip": 0.008544921875, "grad_clipped": false, "step_seconds": 1.8948087410026346, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1026, "loss": 0.11622091382741928, "loss_ema": 0.14959777739996555, "loss_mean_50": 0.13888425044715405, "loss_std_50": 0.15101286583419252, "lr": 0.00025, "grad_norm_before_clip": 0.047607421875, "grad_clipped": false, "step_seconds": 1.8996675849994062, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1027, "loss": 0.2537461519241333, "loss_ema": 0.15480519612617394, "loss_mean_50": 0.14266367562115193, "loss_std_50": 0.15147482557992353, "lr": 0.00025, "grad_norm_before_clip": 0.026611328125, "grad_clipped": false, "step_seconds": 1.8938944149995223, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1028, "loss": 0.05949951708316803, "loss_ema": 0.15003991217402363, "loss_mean_50": 0.1431104426085949, "loss_std_50": 0.1511956783295318, "lr": 0.00025, "grad_norm_before_clip": 0.01251220703125, "grad_clipped": false, "step_seconds": 2.071286851001787, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1029, "loss": 0.08072318881750107, "loss_ema": 0.14657407600619748, "loss_mean_50": 0.14289176777005197, "loss_std_50": 0.15127782202917736, "lr": 0.00025, "grad_norm_before_clip": 0.01007080078125, "grad_clipped": false, "step_seconds": 2.069798668002477, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1030, "loss": 0.2575918436050415, "loss_ema": 0.15212496438613968, "loss_mean_50": 0.14670809328556061, "loss_std_50": 0.15171573328931548, "lr": 0.00025, "grad_norm_before_clip": 0.00994873046875, "grad_clipped": false, "step_seconds": 2.064591706999636, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1031, "loss": 0.4728592038154602, "loss_ema": 0.1681616763576057, "loss_mean_50": 0.15187644749879836, "loss_std_50": 0.15819825065747523, "lr": 0.00025, "grad_norm_before_clip": 0.029296875, "grad_clipped": false, "step_seconds": 0.6031467449938646, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1032, "loss": 0.056820791214704514, "loss_ema": 0.16259463210046063, "loss_mean_50": 0.15068863831460477, "loss_std_50": 0.15868379580351666, "lr": 0.00025, "grad_norm_before_clip": 0.023193359375, "grad_clipped": false, "step_seconds": 1.4123510180070298, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1033, "loss": 0.07029315084218979, "loss_ema": 0.1579795580375471, "loss_mean_50": 0.14947504676878454, "loss_std_50": 0.15906152454948055, "lr": 0.00025, "grad_norm_before_clip": 0.041015625, "grad_clipped": false, "step_seconds": 1.8981530639939592, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1034, "loss": 0.41405194997787476, "loss_ema": 0.1707831776345635, "loss_mean_50": 0.15103802718222142, "loss_std_50": 0.1612545730536658, "lr": 0.00025, "grad_norm_before_clip": 0.0400390625, "grad_clipped": false, "step_seconds": 0.42080843000439927, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1035, "loss": 0.15974454581737518, "loss_ema": 0.17023124604370407, "loss_mean_50": 0.1517640019208193, "loss_std_50": 0.1612104207256012, "lr": 0.00025, "grad_norm_before_clip": 0.0186767578125, "grad_clipped": false, "step_seconds": 0.6025933279961464, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1036, "loss": 0.059833820909261703, "loss_ema": 0.16471137478698195, "loss_mean_50": 0.15024879574775696, "loss_std_50": 0.16171053697564997, "lr": 0.00025, "grad_norm_before_clip": 0.042236328125, "grad_clipped": false, "step_seconds": 2.063268802005041, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1037, "loss": 0.07579972594976425, "loss_ema": 0.16026579234512106, "loss_mean_50": 0.149175473600626, "loss_std_50": 0.16202271568468543, "lr": 0.00025, "grad_norm_before_clip": 0.053466796875, "grad_clipped": false, "step_seconds": 0.5745056519954232, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1038, "loss": 0.08267770707607269, "loss_ema": 0.15638638808166863, "loss_mean_50": 0.14952317744493485, "loss_std_50": 0.16186090183732643, "lr": 0.00025, "grad_norm_before_clip": 0.010009765625, "grad_clipped": false, "step_seconds": 1.8278586270025698, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1039, "loss": 0.08389317244291306, "loss_ema": 0.15276172729973084, "loss_mean_50": 0.15015303924679757, "loss_std_50": 0.16154269669909943, "lr": 0.00025, "grad_norm_before_clip": 0.056640625, "grad_clipped": false, "step_seconds": 0.3645173429977149, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1040, "loss": 0.3096124529838562, "loss_ema": 0.1606042635839371, "loss_mean_50": 0.15483357459306718, "loss_std_50": 0.16270057078538186, "lr": 0.00025, "grad_norm_before_clip": 0.0145263671875, "grad_clipped": false, "step_seconds": 1.8904911709978478, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1041, "loss": 0.056449100375175476, "loss_ema": 0.155396505423499, "loss_mean_50": 0.15450577914714814, "loss_std_50": 0.16288184600044278, "lr": 0.00025, "grad_norm_before_clip": 0.0196533203125, "grad_clipped": false, "step_seconds": 0.5762470230038161, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1042, "loss": 0.05523696914315224, "loss_ema": 0.15038852860948165, "loss_mean_50": 0.15458146221935748, "loss_std_50": 0.16283481720805978, "lr": 0.00025, "grad_norm_before_clip": 0.04150390625, "grad_clipped": false, "step_seconds": 1.8946355040025082, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1043, "loss": 0.0684761330485344, "loss_ema": 0.14629290883143428, "loss_mean_50": 0.1548649163544178, "loss_std_50": 0.1626722662699239, "lr": 0.00025, "grad_norm_before_clip": 0.01019287109375, "grad_clipped": false, "step_seconds": 2.064797371000168, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1044, "loss": 0.05944927781820297, "loss_ema": 0.1419507272807727, "loss_mean_50": 0.15504662312567233, "loss_std_50": 0.1625604718044799, "lr": 0.00025, "grad_norm_before_clip": 0.02587890625, "grad_clipped": false, "step_seconds": 2.06721882999409, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1045, "loss": 0.0891665518283844, "loss_ema": 0.1393115185081533, "loss_mean_50": 0.14219300486147404, "loss_std_50": 0.14033357026055376, "lr": 0.00025, "grad_norm_before_clip": 0.052001953125, "grad_clipped": false, "step_seconds": 0.4776562000042759, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1046, "loss": 0.06592302769422531, "loss_ema": 0.13564209396745688, "loss_mean_50": 0.14216364063322545, "loss_std_50": 0.1403493718723278, "lr": 0.00025, "grad_norm_before_clip": 0.05322265625, "grad_clipped": false, "step_seconds": 0.6191646449951804, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1047, "loss": 0.056991998106241226, "loss_ema": 0.1317095891743961, "loss_mean_50": 0.1419468814134598, "loss_std_50": 0.14047232271871404, "lr": 0.00025, "grad_norm_before_clip": 0.0206298828125, "grad_clipped": false, "step_seconds": 2.06327669100574, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1048, "loss": 0.09319667518138885, "loss_ema": 0.12978394347474573, "loss_mean_50": 0.1393914830684662, "loss_std_50": 0.14017343809315733, "lr": 0.00025, "grad_norm_before_clip": 0.04931640625, "grad_clipped": false, "step_seconds": 0.35057568499905756, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1049, "loss": 0.07062602788209915, "loss_ema": 0.1268260476951134, "loss_mean_50": 0.13979569628834723, "loss_std_50": 0.1399452326092526, "lr": 0.00025, "grad_norm_before_clip": 0.0126953125, "grad_clipped": false, "step_seconds": 1.1075254099996528, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1050, "loss": 0.04651198163628578, "loss_ema": 0.12281034439217202, "loss_mean_50": 0.13835725165903567, "loss_std_50": 0.14052583312762618, "lr": 0.00025, "grad_norm_before_clip": 0.020751953125, "grad_clipped": false, "step_seconds": 2.0682622009990155, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1051, "loss": 0.045084141194820404, "loss_ema": 0.11892403423230444, "loss_mean_50": 0.13803828589618206, "loss_std_50": 0.14071895016957675, "lr": 0.00025, "grad_norm_before_clip": 0.0174560546875, "grad_clipped": false, "step_seconds": 1.8939475340011995, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1052, "loss": 0.1676381379365921, "loss_ema": 0.12135973941751882, "loss_mean_50": 0.13629960976541042, "loss_std_50": 0.13980243594201172, "lr": 0.00025, "grad_norm_before_clip": 0.0228271484375, "grad_clipped": false, "step_seconds": 2.0693993199965917, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1053, "loss": 0.05088120326399803, "loss_ema": 0.11783581260984277, "loss_mean_50": 0.13564482286572457, "loss_std_50": 0.14012393358157846, "lr": 0.00025, "grad_norm_before_clip": 0.048095703125, "grad_clipped": false, "step_seconds": 2.067241158001707, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1054, "loss": 0.6240520477294922, "loss_ema": 0.14314662436582526, "loss_mean_50": 0.1445777054131031, "loss_std_50": 0.1558552841308392, "lr": 0.00025, "grad_norm_before_clip": 0.033203125, "grad_clipped": false, "step_seconds": 1.8332214519978152, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1055, "loss": 0.06988482177257538, "loss_ema": 0.13948353423616278, "loss_mean_50": 0.1443179002404213, "loss_std_50": 0.1559687095267955, "lr": 0.00025, "grad_norm_before_clip": 0.03515625, "grad_clipped": false, "step_seconds": 1.89507756200328, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1056, "loss": 0.289634644985199, "loss_ema": 0.14699108977361458, "loss_mean_50": 0.147180095911026, "loss_std_50": 0.1572904571164615, "lr": 0.00025, "grad_norm_before_clip": 0.036376953125, "grad_clipped": false, "step_seconds": 2.063567625998985, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1057, "loss": 0.03932909667491913, "loss_ema": 0.1416079901186798, "loss_mean_50": 0.1468607174605131, "loss_std_50": 0.15749278185879742, "lr": 0.00025, "grad_norm_before_clip": 0.035400390625, "grad_clipped": false, "step_seconds": 2.068000352002855, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1058, "loss": 0.0727410688996315, "loss_ema": 0.13816464405772738, "loss_mean_50": 0.14690557099878787, "loss_std_50": 0.15747134557372902, "lr": 0.00025, "grad_norm_before_clip": 0.043701171875, "grad_clipped": false, "step_seconds": 2.06756103799853, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1059, "loss": 0.1367603987455368, "loss_ema": 0.13809443179211783, "loss_mean_50": 0.14680490143597125, "loss_std_50": 0.15747619008760308, "lr": 0.00025, "grad_norm_before_clip": 0.0208740234375, "grad_clipped": false, "step_seconds": 1.892467181001848, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1060, "loss": 0.05286663770675659, "loss_ema": 0.13383304208784977, "loss_mean_50": 0.1466301153600216, "loss_std_50": 0.1575754758416257, "lr": 0.00025, "grad_norm_before_clip": 0.044189453125, "grad_clipped": false, "step_seconds": 1.2683608989973436, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1061, "loss": 0.09697014838457108, "loss_ema": 0.13198989740268582, "loss_mean_50": 0.14696505054831505, "loss_std_50": 0.15745171826733872, "lr": 0.00025, "grad_norm_before_clip": 0.0186767578125, "grad_clipped": false, "step_seconds": 0.7248167150028166, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1062, "loss": 0.0718308761715889, "loss_ema": 0.12898194634113097, "loss_mean_50": 0.14502957075834275, "loss_std_50": 0.15776829451615199, "lr": 0.00025, "grad_norm_before_clip": 0.02978515625, "grad_clipped": false, "step_seconds": 0.671846627999912, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1063, "loss": 0.2505705952644348, "loss_ema": 0.13506137878729615, "loss_mean_50": 0.1488503412157297, "loss_std_50": 0.15796461053095293, "lr": 0.00025, "grad_norm_before_clip": 0.0211181640625, "grad_clipped": false, "step_seconds": 1.590168515998812, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1064, "loss": 0.06181033328175545, "loss_ema": 0.13139882651201912, "loss_mean_50": 0.14782421216368674, "loss_std_50": 0.15835954899990737, "lr": 0.00025, "grad_norm_before_clip": 0.01556396484375, "grad_clipped": false, "step_seconds": 1.832832319996669, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1065, "loss": 0.06566748023033142, "loss_ema": 0.12811225919793473, "loss_mean_50": 0.1346128551661968, "loss_std_50": 0.13545114599227248, "lr": 0.00025, "grad_norm_before_clip": 0.048095703125, "grad_clipped": false, "step_seconds": 1.895220964004693, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1066, "loss": 0.0928841158747673, "loss_ema": 0.12635085203177637, "loss_mean_50": 0.13554720573127269, "loss_std_50": 0.13499818831541732, "lr": 0.00025, "grad_norm_before_clip": 0.01513671875, "grad_clipped": false, "step_seconds": 2.0697682719983277, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1067, "loss": 0.158051997423172, "loss_ema": 0.12793590930134613, "loss_mean_50": 0.1372682700306177, "loss_std_50": 0.13472531334938007, "lr": 0.00025, "grad_norm_before_clip": 0.0130615234375, "grad_clipped": false, "step_seconds": 2.071909689999302, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1068, "loss": 0.07281028479337692, "loss_ema": 0.12517962807594765, "loss_mean_50": 0.137680931314826, "loss_std_50": 0.13449545223184306, "lr": 0.00025, "grad_norm_before_clip": 0.02978515625, "grad_clipped": false, "step_seconds": 1.440042463997088, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1069, "loss": 0.058655206114053726, "loss_ema": 0.12185340697785295, "loss_mean_50": 0.1371144312620163, "loss_std_50": 0.13476719106757998, "lr": 0.00025, "grad_norm_before_clip": 0.06591796875, "grad_clipped": false, "step_seconds": 0.5976686530048028, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1070, "loss": 0.058942247182130814, "loss_ema": 0.11870784898806683, "loss_mean_50": 0.13677445389330387, "loss_std_50": 0.13494241186510822, "lr": 0.00025, "grad_norm_before_clip": 0.05615234375, "grad_clipped": false, "step_seconds": 0.5237208990001818, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1071, "loss": 0.10455760359764099, "loss_ema": 0.11800033671854553, "loss_mean_50": 0.1364054275304079, "loss_std_50": 0.13500476677090895, "lr": 0.00025, "grad_norm_before_clip": 0.08056640625, "grad_clipped": false, "step_seconds": 1.2971243969950592, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1072, "loss": 0.04733186215162277, "loss_ema": 0.11446691299019938, "loss_mean_50": 0.13571409337222576, "loss_std_50": 0.13537012653724836, "lr": 0.00025, "grad_norm_before_clip": 0.0225830078125, "grad_clipped": false, "step_seconds": 1.576691836999089, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1073, "loss": 0.07817821949720383, "loss_ema": 0.1126524783155496, "loss_mean_50": 0.12528838373720647, "loss_std_50": 0.11824258664654763, "lr": 0.00025, "grad_norm_before_clip": 0.0081787109375, "grad_clipped": false, "step_seconds": 2.0680546069997945, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1074, "loss": 0.06662072986364365, "loss_ema": 0.1103508908929543, "loss_mean_50": 0.12267039500176907, "loss_std_50": 0.11806330799556078, "lr": 0.00025, "grad_norm_before_clip": 0.01312255859375, "grad_clipped": false, "step_seconds": 2.063287039003626, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1075, "loss": 0.39582502841949463, "loss_ema": 0.12462459776928132, "loss_mean_50": 0.1288994961231947, "loss_std_50": 0.12394786654765913, "lr": 0.00025, "grad_norm_before_clip": 0.1455078125, "grad_clipped": false, "step_seconds": 1.8897601279968512, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1076, "loss": 0.11716390401124954, "loss_ema": 0.12425156308137973, "loss_mean_50": 0.1289183559268713, "loss_std_50": 0.12394600767869803, "lr": 0.00025, "grad_norm_before_clip": 0.0301513671875, "grad_clipped": false, "step_seconds": 1.9060177720020874, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1077, "loss": 0.08797845244407654, "loss_ema": 0.12243790754951457, "loss_mean_50": 0.12560300193727017, "loss_std_50": 0.12277419593825259, "lr": 0.00025, "grad_norm_before_clip": 0.04443359375, "grad_clipped": false, "step_seconds": 1.8321331830011331, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1078, "loss": 0.26392921805381775, "loss_ema": 0.12951247307472974, "loss_mean_50": 0.12969159595668317, "loss_std_50": 0.12390349476239224, "lr": 0.00025, "grad_norm_before_clip": 0.01416015625, "grad_clipped": false, "step_seconds": 1.1037692449972383, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1079, "loss": 0.05550440028309822, "loss_ema": 0.12581206943514817, "loss_mean_50": 0.1291872201859951, "loss_std_50": 0.12415288285617596, "lr": 0.00025, "grad_norm_before_clip": 0.06396484375, "grad_clipped": false, "step_seconds": 0.5252912069990998, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1080, "loss": 0.6752395033836365, "loss_ema": 0.15328344113257258, "loss_mean_50": 0.137540173381567, "loss_std_50": 0.14483740355243646, "lr": 0.00025, "grad_norm_before_clip": 0.037353515625, "grad_clipped": false, "step_seconds": 2.0631647490008618, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1081, "loss": 0.12614411115646362, "loss_ema": 0.15192647463376713, "loss_mean_50": 0.13060587152838707, "loss_std_50": 0.13668799979106636, "lr": 0.00025, "grad_norm_before_clip": 0.015380859375, "grad_clipped": false, "step_seconds": 1.837092098001449, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1082, "loss": 0.037937745451927185, "loss_ema": 0.1462270381746751, "loss_mean_50": 0.13022821061313153, "loss_std_50": 0.1369172360566345, "lr": 0.00025, "grad_norm_before_clip": 0.01519775390625, "grad_clipped": false, "step_seconds": 1.831534463002754, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1083, "loss": 0.06912761181592941, "loss_ema": 0.14237206685673784, "loss_mean_50": 0.1302048998326063, "loss_std_50": 0.13692753711970537, "lr": 0.00025, "grad_norm_before_clip": 0.0213623046875, "grad_clipped": false, "step_seconds": 0.573654918000102, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1084, "loss": 0.11924929916858673, "loss_ema": 0.14121592847233028, "loss_mean_50": 0.12430884681642056, "loss_std_50": 0.13078763139797814, "lr": 0.00025, "grad_norm_before_clip": 0.01361083984375, "grad_clipped": false, "step_seconds": 0.6139696499958518, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1085, "loss": 0.2717928886413574, "loss_ema": 0.14774477648078163, "loss_mean_50": 0.1265498136729002, "loss_std_50": 0.13232648892352666, "lr": 0.00025, "grad_norm_before_clip": 0.040771484375, "grad_clipped": false, "step_seconds": 0.7286845620037639, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1086, "loss": 0.10664083808660507, "loss_ema": 0.1456895795610728, "loss_mean_50": 0.12748595401644708, "loss_std_50": 0.1320164011949067, "lr": 0.00025, "grad_norm_before_clip": 0.0108642578125, "grad_clipped": false, "step_seconds": 2.070939735000138, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1087, "loss": 0.05979146435856819, "loss_ema": 0.14139467380094758, "loss_mean_50": 0.12716578878462315, "loss_std_50": 0.13216069473880837, "lr": 0.00025, "grad_norm_before_clip": 0.0289306640625, "grad_clipped": false, "step_seconds": 1.8942187919965363, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1088, "loss": 0.06667516380548477, "loss_ema": 0.13765869830117444, "loss_mean_50": 0.1268457379192114, "loss_std_50": 0.13228735890935778, "lr": 0.00025, "grad_norm_before_clip": 0.022216796875, "grad_clipped": false, "step_seconds": 1.894581428001402, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1089, "loss": 0.12290243059396744, "loss_ema": 0.13692088491581408, "loss_mean_50": 0.12762592308223247, "loss_std_50": 0.13214669566113998, "lr": 0.00025, "grad_norm_before_clip": 0.016845703125, "grad_clipped": false, "step_seconds": 2.070148042002984, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1090, "loss": 0.04720645397901535, "loss_ema": 0.13243516336897412, "loss_mean_50": 0.12237780310213565, "loss_std_50": 0.13000834700024702, "lr": 0.00025, "grad_norm_before_clip": 0.0145263671875, "grad_clipped": false, "step_seconds": 2.067558664995886, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1091, "loss": 0.05751949921250343, "loss_ema": 0.12868938016115058, "loss_mean_50": 0.12239921107888221, "loss_std_50": 0.12999757669371673, "lr": 0.00025, "grad_norm_before_clip": 0.01141357421875, "grad_clipped": false, "step_seconds": 1.8945231809993857, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1092, "loss": 0.05874582380056381, "loss_ema": 0.12519220234312123, "loss_mean_50": 0.12246938817203044, "loss_std_50": 0.12996224359719927, "lr": 0.00025, "grad_norm_before_clip": 0.0084228515625, "grad_clipped": false, "step_seconds": 2.067839613002434, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1093, "loss": 0.27718308568000793, "loss_ema": 0.13279174650996556, "loss_mean_50": 0.1266435272246599, "loss_std_50": 0.13150354541206202, "lr": 0.00025, "grad_norm_before_clip": 0.01513671875, "grad_clipped": false, "step_seconds": 2.066943206002179, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1094, "loss": 0.2509692311286926, "loss_ema": 0.1387006207409019, "loss_mean_50": 0.1304739262908697, "loss_std_50": 0.13227753690213073, "lr": 0.00025, "grad_norm_before_clip": 0.0164794921875, "grad_clipped": false, "step_seconds": 1.8904390530005912, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1095, "loss": 0.08745188266038895, "loss_ema": 0.13613818383687623, "loss_mean_50": 0.13043963290750982, "loss_std_50": 0.132288463344018, "lr": 0.00025, "grad_norm_before_clip": 0.061279296875, "grad_clipped": false, "step_seconds": 0.3645240509940777, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1096, "loss": 0.0563785694539547, "loss_ema": 0.13215020311773015, "loss_mean_50": 0.1302487437427044, "loss_std_50": 0.13238827014007265, "lr": 0.00025, "grad_norm_before_clip": 0.0289306640625, "grad_clipped": false, "step_seconds": 1.27125752500433, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1097, "loss": 0.2951076924800873, "loss_ema": 0.140298077585848, "loss_mean_50": 0.1350110576301813, "loss_std_50": 0.13394107988615364, "lr": 0.00025, "grad_norm_before_clip": 0.01348876953125, "grad_clipped": false, "step_seconds": 2.0600388339953497, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1098, "loss": 0.07326794415712357, "loss_ema": 0.13694657091441179, "loss_mean_50": 0.134612483009696, "loss_std_50": 0.1340944794127497, "lr": 0.00025, "grad_norm_before_clip": 0.02783203125, "grad_clipped": false, "step_seconds": 0.6029730240043136, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1099, "loss": 0.051267679780721664, "loss_ema": 0.13266262635772727, "loss_mean_50": 0.13422531604766846, "loss_std_50": 0.13430644551334636, "lr": 0.00025, "grad_norm_before_clip": 0.01348876953125, "grad_clipped": false, "step_seconds": 1.8933053529981407, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1100, "loss": 0.058637622743844986, "loss_ema": 0.12896137617703315, "loss_mean_50": 0.13446782886981964, "loss_std_50": 0.13415871161005224, "lr": 0.00025, "grad_norm_before_clip": 0.0203857421875, "grad_clipped": false, "step_seconds": 1.2952555410010973, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1101, "loss": 0.08163976669311523, "loss_ema": 0.12659529570283726, "loss_mean_50": 0.13519894137978553, "loss_std_50": 0.13376865322633033, "lr": 0.00025, "grad_norm_before_clip": 0.033935546875, "grad_clipped": false, "step_seconds": 2.0641693099969416, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1102, "loss": 0.05669427663087845, "loss_ema": 0.12310024474923931, "loss_mean_50": 0.13298006415367125, "loss_std_50": 0.13413181124596577, "lr": 0.00025, "grad_norm_before_clip": 0.033203125, "grad_clipped": false, "step_seconds": 0.4216164760000538, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1103, "loss": 0.05792616307735443, "loss_ema": 0.11984154066564506, "loss_mean_50": 0.1331209633499384, "loss_std_50": 0.13404917096978897, "lr": 0.00025, "grad_norm_before_clip": 0.0220947265625, "grad_clipped": false, "step_seconds": 0.5720742180055822, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1104, "loss": 0.12478457391262054, "loss_ema": 0.12008869232799382, "loss_mean_50": 0.12313561387360096, "loss_std_50": 0.11423920670723534, "lr": 0.00025, "grad_norm_before_clip": 0.1494140625, "grad_clipped": false, "step_seconds": 0.832352589997754, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1105, "loss": 0.6972883343696594, "loss_ema": 0.1489486744300771, "loss_mean_50": 0.13568368412554263, "loss_std_50": 0.1393895845245388, "lr": 0.00025, "grad_norm_before_clip": 0.03857421875, "grad_clipped": false, "step_seconds": 1.293604302998574, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1106, "loss": 0.047519974410533905, "loss_ema": 0.1438772394290999, "loss_mean_50": 0.13084139071404935, "loss_std_50": 0.13815732805632344, "lr": 0.00025, "grad_norm_before_clip": 0.04638671875, "grad_clipped": false, "step_seconds": 1.5908809969987487, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1107, "loss": 0.05852028727531433, "loss_ema": 0.13960939182141063, "loss_mean_50": 0.13122521452605723, "loss_std_50": 0.13792902818402566, "lr": 0.00025, "grad_norm_before_clip": 0.0220947265625, "grad_clipped": false, "step_seconds": 2.068602222003392, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1108, "loss": 0.0805501788854599, "loss_ema": 0.1366564311746131, "loss_mean_50": 0.1313813967257738, "loss_std_50": 0.1378671233538319, "lr": 0.00025, "grad_norm_before_clip": 0.04541015625, "grad_clipped": false, "step_seconds": 2.064647993000108, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1109, "loss": 0.08935725688934326, "loss_ema": 0.1342914724603496, "loss_mean_50": 0.13043333388864994, "loss_std_50": 0.1379898067686263, "lr": 0.00025, "grad_norm_before_clip": 0.0162353515625, "grad_clipped": false, "step_seconds": 0.6686954559991136, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1110, "loss": 0.08276772499084473, "loss_ema": 0.13171528508687436, "loss_mean_50": 0.13103135563433171, "loss_std_50": 0.1377168744211653, "lr": 0.00025, "grad_norm_before_clip": 0.064453125, "grad_clipped": false, "step_seconds": 0.47929832999943756, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1111, "loss": 0.05510077625513077, "loss_ema": 0.12788455964528717, "loss_mean_50": 0.1301939681917429, "loss_std_50": 0.13804833220114252, "lr": 0.00025, "grad_norm_before_clip": 0.01422119140625, "grad_clipped": false, "step_seconds": 1.574062617000891, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1112, "loss": 0.07851894944906235, "loss_ema": 0.12541627913547593, "loss_mean_50": 0.13032772965729236, "loss_std_50": 0.1379949465466957, "lr": 0.00025, "grad_norm_before_clip": 0.0130615234375, "grad_clipped": false, "step_seconds": 1.8942303149960935, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1113, "loss": 0.7441203594207764, "loss_ema": 0.15635148314974096, "loss_mean_50": 0.14019872494041918, "loss_std_50": 0.16183581073729328, "lr": 0.00025, "grad_norm_before_clip": 0.02294921875, "grad_clipped": false, "step_seconds": 2.0667982250015484, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1114, "loss": 0.5429820418357849, "loss_ema": 0.17568301108404316, "loss_mean_50": 0.1498221591114998, "loss_std_50": 0.17093862161664902, "lr": 0.00025, "grad_norm_before_clip": 0.05712890625, "grad_clipped": false, "step_seconds": 2.0675187570013804, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1115, "loss": 0.09840474277734756, "loss_ema": 0.17181909766870837, "loss_mean_50": 0.1504769043624401, "loss_std_50": 0.1706775277631886, "lr": 0.00025, "grad_norm_before_clip": 0.0159912109375, "grad_clipped": false, "step_seconds": 2.06321064200165, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1116, "loss": 0.08401155471801758, "loss_ema": 0.16742872052117383, "loss_mean_50": 0.15029945313930512, "loss_std_50": 0.1707419141944028, "lr": 0.00025, "grad_norm_before_clip": 0.01153564453125, "grad_clipped": false, "step_seconds": 1.8944365450006444, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1117, "loss": 0.04101628437638283, "loss_ema": 0.16110809871393428, "loss_mean_50": 0.14795873887836933, "loss_std_50": 0.17142046698150518, "lr": 0.00025, "grad_norm_before_clip": 0.032958984375, "grad_clipped": false, "step_seconds": 2.0631852869992144, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1118, "loss": 0.10976727306842804, "loss_ema": 0.15854105743165894, "loss_mean_50": 0.14869787864387035, "loss_std_50": 0.17117434425389602, "lr": 0.00025, "grad_norm_before_clip": 0.037353515625, "grad_clipped": false, "step_seconds": 0.3466109889996005, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1119, "loss": 0.0834706723690033, "loss_ema": 0.15478753817852614, "loss_mean_50": 0.14919418796896936, "loss_std_50": 0.1709483779362863, "lr": 0.00025, "grad_norm_before_clip": 0.00799560546875, "grad_clipped": false, "step_seconds": 1.8993426599990926, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1120, "loss": 0.10931798070669174, "loss_ema": 0.15251406030493442, "loss_mean_50": 0.15020170263946056, "loss_std_50": 0.17056150449334337, "lr": 0.00025, "grad_norm_before_clip": 0.0458984375, "grad_clipped": false, "step_seconds": 1.891526965002413, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1121, "loss": 0.06965146213769913, "loss_ema": 0.14837093039657265, "loss_mean_50": 0.14950357981026172, "loss_std_50": 0.17081814478824894, "lr": 0.00025, "grad_norm_before_clip": 0.01495361328125, "grad_clipped": false, "step_seconds": 2.0594018139963737, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1122, "loss": 0.047198403626680374, "loss_ema": 0.14331230405807802, "loss_mean_50": 0.14950091063976287, "loss_std_50": 0.17081974231764735, "lr": 0.00025, "grad_norm_before_clip": 0.0174560546875, "grad_clipped": false, "step_seconds": 0.5978652039993904, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1123, "loss": 0.24319124221801758, "loss_ema": 0.148306250966075, "loss_mean_50": 0.15280117109417915, "loss_std_50": 0.17100383529496932, "lr": 0.00025, "grad_norm_before_clip": 0.014404296875, "grad_clipped": false, "step_seconds": 2.0633440380042884, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1124, "loss": 0.06003126502037048, "loss_ema": 0.14389250166878978, "loss_mean_50": 0.15266938179731368, "loss_std_50": 0.17107272739281176, "lr": 0.00025, "grad_norm_before_clip": 0.0191650390625, "grad_clipped": false, "step_seconds": 0.5202204169981997, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1125, "loss": 0.09179362654685974, "loss_ema": 0.1412875579126933, "loss_mean_50": 0.146588753759861, "loss_std_50": 0.167691764264518, "lr": 0.00025, "grad_norm_before_clip": 0.0213623046875, "grad_clipped": false, "step_seconds": 0.47540779799601296, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1126, "loss": 0.0616651214659214, "loss_ema": 0.1373064360903547, "loss_mean_50": 0.14547877810895443, "loss_std_50": 0.16806611730800006, "lr": 0.00025, "grad_norm_before_clip": 0.08984375, "grad_clipped": false, "step_seconds": 0.5721051430009538, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1127, "loss": 0.12283039838075638, "loss_ema": 0.1365826342048748, "loss_mean_50": 0.14617581702768803, "loss_std_50": 0.16789838342517513, "lr": 0.00025, "grad_norm_before_clip": 0.013427734375, "grad_clipped": false, "step_seconds": 2.0644386730054975, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1128, "loss": 0.06478963047266006, "loss_ema": 0.13299298401826407, "loss_mean_50": 0.14219302527606487, "loss_std_50": 0.16741911923321454, "lr": 0.00025, "grad_norm_before_clip": 0.018798828125, "grad_clipped": false, "step_seconds": 2.06428832399979, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1129, "loss": 0.07033087313175201, "loss_ema": 0.12985987847393846, "loss_mean_50": 0.14248955473303795, "loss_std_50": 0.167278386467766, "lr": 0.00025, "grad_norm_before_clip": 0.038818359375, "grad_clipped": false, "step_seconds": 1.2923831189982593, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1130, "loss": 0.062235962599515915, "loss_ema": 0.12647868268021734, "loss_mean_50": 0.13022948391735553, "loss_std_50": 0.14927863835162786, "lr": 0.00025, "grad_norm_before_clip": 0.00848388671875, "grad_clipped": false, "step_seconds": 2.0636671129977913, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1131, "loss": 0.5557267069816589, "loss_ema": 0.1479410838952894, "loss_mean_50": 0.13882113583385944, "loss_std_50": 0.1607200028520361, "lr": 0.00025, "grad_norm_before_clip": 0.03271484375, "grad_clipped": false, "step_seconds": 0.599774725997122, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1132, "loss": 0.1266298145055771, "loss_ema": 0.1468755204258038, "loss_mean_50": 0.14059497721493244, "loss_std_50": 0.16008496556744112, "lr": 0.00025, "grad_norm_before_clip": 0.015625, "grad_clipped": false, "step_seconds": 1.8908111440032371, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1133, "loss": 0.05285258963704109, "loss_ema": 0.14217437388636567, "loss_mean_50": 0.14026947677135468, "loss_std_50": 0.16024641367408482, "lr": 0.00025, "grad_norm_before_clip": 0.037109375, "grad_clipped": false, "step_seconds": 2.064417310000863, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1134, "loss": 0.05221651494503021, "loss_ema": 0.1376764809392989, "loss_mean_50": 0.13892882108688354, "loss_std_50": 0.16069643823725635, "lr": 0.00025, "grad_norm_before_clip": 0.0184326171875, "grad_clipped": false, "step_seconds": 1.5708551299976534, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1135, "loss": 0.0672934427857399, "loss_ema": 0.13415732903162095, "loss_mean_50": 0.1348388321697712, "loss_std_50": 0.15986304323496037, "lr": 0.00025, "grad_norm_before_clip": 0.0341796875, "grad_clipped": false, "step_seconds": 2.059717581003497, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1136, "loss": 0.05420153588056564, "loss_ema": 0.13015953937406818, "loss_mean_50": 0.1337900461256504, "loss_std_50": 0.1602162213192748, "lr": 0.00025, "grad_norm_before_clip": 0.037841796875, "grad_clipped": false, "step_seconds": 1.1036149969950202, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1137, "loss": 0.07986480742692947, "loss_ema": 0.12764480277671122, "loss_mean_50": 0.13419151298701762, "loss_std_50": 0.16005536299508855, "lr": 0.00025, "grad_norm_before_clip": 0.103515625, "grad_clipped": false, "step_seconds": 1.8287451780051924, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1138, "loss": 0.15218490362167358, "loss_ema": 0.12887180781895935, "loss_mean_50": 0.1359017077833414, "loss_std_50": 0.15978141491188128, "lr": 0.00025, "grad_norm_before_clip": 0.201171875, "grad_clipped": false, "step_seconds": 0.34703286599688, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1139, "loss": 0.0864643007516861, "loss_ema": 0.1267514324655957, "loss_mean_50": 0.13517294518649578, "loss_std_50": 0.1599220778268327, "lr": 0.00025, "grad_norm_before_clip": 0.07275390625, "grad_clipped": false, "step_seconds": 0.5716733540029963, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1140, "loss": 0.05295072868466377, "loss_ema": 0.12306139727654909, "loss_mean_50": 0.13528783068060876, "loss_std_50": 0.15986089441760656, "lr": 0.00025, "grad_norm_before_clip": 0.0216064453125, "grad_clipped": false, "step_seconds": 1.3422249459981686, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1141, "loss": 0.10319263488054276, "loss_ema": 0.12206795915674877, "loss_mean_50": 0.13620129339396952, "loss_std_50": 0.15954408454765936, "lr": 0.00025, "grad_norm_before_clip": 0.049560546875, "grad_clipped": false, "step_seconds": 0.36210315600328613, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1142, "loss": 0.04476553201675415, "loss_ema": 0.11820283779974904, "loss_mean_50": 0.13592168755829334, "loss_std_50": 0.15969176468591378, "lr": 0.00025, "grad_norm_before_clip": 0.0242919921875, "grad_clipped": false, "step_seconds": 1.8303689109961851, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1143, "loss": 0.5715754628181458, "loss_ema": 0.14087146905066888, "loss_mean_50": 0.1418095351010561, "loss_std_50": 0.16989285313335475, "lr": 0.00025, "grad_norm_before_clip": 0.05712890625, "grad_clipped": false, "step_seconds": 2.0635896860039793, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1144, "loss": 0.05382949486374855, "loss_ema": 0.13651937034132286, "loss_mean_50": 0.13786674037575722, "loss_std_50": 0.16960108749772818, "lr": 0.00025, "grad_norm_before_clip": 0.09130859375, "grad_clipped": false, "step_seconds": 2.0596475709971855, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1145, "loss": 0.11477876454591751, "loss_ema": 0.13543234005155258, "loss_mean_50": 0.13841327801346778, "loss_std_50": 0.16948173385946871, "lr": 0.00025, "grad_norm_before_clip": 0.052734375, "grad_clipped": false, "step_seconds": 2.063978523001424, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1146, "loss": 0.05554242432117462, "loss_ema": 0.13143784426503366, "loss_mean_50": 0.1383965551108122, "loss_std_50": 0.1694898685232283, "lr": 0.00025, "grad_norm_before_clip": 0.020263671875, "grad_clipped": false, "step_seconds": 1.2902823269978398, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1147, "loss": 0.05824127420783043, "loss_ema": 0.12777801576217349, "loss_mean_50": 0.13365922674536704, "loss_std_50": 0.16834994214908555, "lr": 0.00025, "grad_norm_before_clip": 0.0294189453125, "grad_clipped": false, "step_seconds": 0.4215745129986317, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1148, "loss": 0.04850808158516884, "loss_ema": 0.12381451905332325, "loss_mean_50": 0.13316402949392794, "loss_std_50": 0.16856313360717823, "lr": 0.00025, "grad_norm_before_clip": 0.036376953125, "grad_clipped": false, "step_seconds": 1.8296524670004146, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1149, "loss": 0.06381651014089584, "loss_ema": 0.12081461860770187, "loss_mean_50": 0.13341500610113144, "loss_std_50": 0.16845031418520856, "lr": 0.00025, "grad_norm_before_clip": 0.0198974609375, "grad_clipped": false, "step_seconds": 1.8898743450045004, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1150, "loss": 0.06920751929283142, "loss_ema": 0.11823426364195835, "loss_mean_50": 0.13362640403211118, "loss_std_50": 0.16836294885163947, "lr": 0.00025, "grad_norm_before_clip": 0.064453125, "grad_clipped": false, "step_seconds": 0.6690565939934459, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1151, "loss": 0.3727438151836395, "loss_ema": 0.1309597412190424, "loss_mean_50": 0.13944848500192164, "loss_std_50": 0.17146917051260435, "lr": 0.00025, "grad_norm_before_clip": 0.02734375, "grad_clipped": false, "step_seconds": 1.8903422729999875, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1152, "loss": 0.13088323175907135, "loss_ema": 0.13095591574604387, "loss_mean_50": 0.1409322641044855, "loss_std_50": 0.17106717080449804, "lr": 0.00025, "grad_norm_before_clip": 0.07763671875, "grad_clipped": false, "step_seconds": 0.7241905860064435, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1153, "loss": 0.08146390318870544, "loss_ema": 0.12848131511817695, "loss_mean_50": 0.14140301890671253, "loss_std_50": 0.17087037420166537, "lr": 0.00025, "grad_norm_before_clip": 0.0228271484375, "grad_clipped": false, "step_seconds": 1.890996832997189, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1154, "loss": 0.11640938371419907, "loss_ema": 0.12787771854797805, "loss_mean_50": 0.1412355151027441, "loss_std_50": 0.17089068700199875, "lr": 0.00025, "grad_norm_before_clip": 0.12451171875, "grad_clipped": false, "step_seconds": 2.0642907520013978, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1155, "loss": 0.13247466087341309, "loss_ema": 0.1281075656642498, "loss_mean_50": 0.12993924163281917, "loss_std_50": 0.15130651474655887, "lr": 0.00025, "grad_norm_before_clip": 0.02197265625, "grad_clipped": false, "step_seconds": 1.8911293189958087, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1156, "loss": 0.09731291979551315, "loss_ema": 0.12656783337081295, "loss_mean_50": 0.13093510054051877, "loss_std_50": 0.15092415477692775, "lr": 0.00025, "grad_norm_before_clip": 0.01287841796875, "grad_clipped": false, "step_seconds": 2.064395882000099, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1157, "loss": 0.0609324686229229, "loss_ema": 0.12328606513341844, "loss_mean_50": 0.13098334416747093, "loss_std_50": 0.1509013831400681, "lr": 0.00025, "grad_norm_before_clip": 0.0206298828125, "grad_clipped": false, "step_seconds": 2.063090622003074, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1158, "loss": 0.6071805357933044, "loss_ema": 0.14748078866641273, "loss_mean_50": 0.14151595130562783, "loss_std_50": 0.1647564773946634, "lr": 0.00025, "grad_norm_before_clip": 0.06103515625, "grad_clipped": false, "step_seconds": 0.6130331870008376, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1159, "loss": 0.03556752949953079, "loss_ema": 0.1418851257080686, "loss_mean_50": 0.14044015675783159, "loss_std_50": 0.16526835838474643, "lr": 0.00025, "grad_norm_before_clip": 0.025390625, "grad_clipped": false, "step_seconds": 2.0639745270018466, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1160, "loss": 0.06781023740768433, "loss_ema": 0.13818138129304938, "loss_mean_50": 0.14014100700616836, "loss_std_50": 0.16538597494525614, "lr": 0.00025, "grad_norm_before_clip": 0.048828125, "grad_clipped": false, "step_seconds": 1.5869731999991927, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1161, "loss": 0.055490169674158096, "loss_ema": 0.13404682071210483, "loss_mean_50": 0.1401487948745489, "loss_std_50": 0.16538197941806845, "lr": 0.00025, "grad_norm_before_clip": 0.044677734375, "grad_clipped": false, "step_seconds": 1.8892821219997131, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1162, "loss": 0.06527135521173477, "loss_ema": 0.13060804743708632, "loss_mean_50": 0.13988384298980236, "loss_std_50": 0.16549107764227944, "lr": 0.00025, "grad_norm_before_clip": 0.032958984375, "grad_clipped": false, "step_seconds": 1.3004279870001483, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1163, "loss": 0.0722023993730545, "loss_ema": 0.12768776503388474, "loss_mean_50": 0.12644548378884793, "loss_std_50": 0.14140822989351523, "lr": 0.00025, "grad_norm_before_clip": 0.0146484375, "grad_clipped": false, "step_seconds": 1.8912016039976152, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1164, "loss": 0.181363046169281, "loss_ema": 0.13037152909065455, "loss_mean_50": 0.11921310387551784, "loss_std_50": 0.12858555459992735, "lr": 0.00025, "grad_norm_before_clip": 0.0150146484375, "grad_clipped": false, "step_seconds": 2.064100404000783, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1165, "loss": 0.050506483763456345, "loss_ema": 0.12637827682429464, "loss_mean_50": 0.11825513869524001, "loss_std_50": 0.12891500839544207, "lr": 0.00025, "grad_norm_before_clip": 0.01904296875, "grad_clipped": false, "step_seconds": 2.0626316359994235, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1166, "loss": 0.19895531237125397, "loss_ema": 0.1300071286016426, "loss_mean_50": 0.12055401384830475, "loss_std_50": 0.12930812991603505, "lr": 0.00025, "grad_norm_before_clip": 0.04296875, "grad_clipped": false, "step_seconds": 0.5720528579986421, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1167, "loss": 0.16172994673252106, "loss_ema": 0.1315932695081865, "loss_mean_50": 0.12296828709542751, "loss_std_50": 0.12892690909138077, "lr": 0.00025, "grad_norm_before_clip": 0.00958251953125, "grad_clipped": false, "step_seconds": 2.0592352070016204, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1168, "loss": 0.046363603323698044, "loss_ema": 0.12733178619896207, "loss_mean_50": 0.12170021370053291, "loss_std_50": 0.12936158662373787, "lr": 0.00025, "grad_norm_before_clip": 0.103515625, "grad_clipped": false, "step_seconds": 2.0587072769994847, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1169, "loss": 0.06581753492355347, "loss_ema": 0.12425607363519163, "loss_mean_50": 0.12134715095162392, "loss_std_50": 0.12948947045858059, "lr": 0.00025, "grad_norm_before_clip": 0.056884765625, "grad_clipped": false, "step_seconds": 1.8900156629970297, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1170, "loss": 0.14683657884597778, "loss_ema": 0.12538509889573093, "loss_mean_50": 0.12209752291440963, "loss_std_50": 0.12952629115941316, "lr": 0.00025, "grad_norm_before_clip": 0.037841796875, "grad_clipped": false, "step_seconds": 1.569745288004924, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1171, "loss": 0.06864460557699203, "loss_ema": 0.12254807422979398, "loss_mean_50": 0.12207738578319549, "loss_std_50": 0.12953452125833773, "lr": 0.00025, "grad_norm_before_clip": 0.03515625, "grad_clipped": false, "step_seconds": 1.8894646080007078, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1172, "loss": 0.051292743533849716, "loss_ema": 0.11898530769499677, "loss_mean_50": 0.12215927258133888, "loss_std_50": 0.1294884456790276, "lr": 0.00025, "grad_norm_before_clip": 0.0225830078125, "grad_clipped": false, "step_seconds": 0.5978015449945815, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1173, "loss": 0.08190843462944031, "loss_ema": 0.11713146404171894, "loss_mean_50": 0.11893361642956733, "loss_std_50": 0.1284378470404847, "lr": 0.00025, "grad_norm_before_clip": 0.00811767578125, "grad_clipped": false, "step_seconds": 2.0629456200040295, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1174, "loss": 0.6018933653831482, "loss_ema": 0.1413695591087904, "loss_mean_50": 0.1297708584368229, "loss_std_50": 0.14482557571170426, "lr": 0.00025, "grad_norm_before_clip": 0.0289306640625, "grad_clipped": false, "step_seconds": 2.0627187049976783, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1175, "loss": 0.11439007520675659, "loss_ema": 0.1400205849136887, "loss_mean_50": 0.13022278741002083, "loss_std_50": 0.1447415942548813, "lr": 0.00025, "grad_norm_before_clip": 0.0400390625, "grad_clipped": false, "step_seconds": 1.8887726719985949, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1176, "loss": 0.08466706424951553, "loss_ema": 0.13725290888048006, "loss_mean_50": 0.13068282626569272, "loss_std_50": 0.14455940259545708, "lr": 0.00025, "grad_norm_before_clip": 0.0419921875, "grad_clipped": false, "step_seconds": 0.7355408040020848, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1177, "loss": 0.07384856045246124, "loss_ema": 0.1340826914590791, "loss_mean_50": 0.1297031895071268, "loss_std_50": 0.1447751037074867, "lr": 0.00025, "grad_norm_before_clip": 0.0208740234375, "grad_clipped": false, "step_seconds": 1.2897568609987502, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1178, "loss": 0.09191448241472244, "loss_ema": 0.13197428100686126, "loss_mean_50": 0.13024568654596805, "loss_std_50": 0.1445815364257147, "lr": 0.00025, "grad_norm_before_clip": 0.046630859375, "grad_clipped": false, "step_seconds": 0.5993978080005036, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1179, "loss": 0.045032624155282974, "loss_ema": 0.12762719816428233, "loss_mean_50": 0.12973972156643868, "loss_std_50": 0.14483436849508657, "lr": 0.00025, "grad_norm_before_clip": 0.0189208984375, "grad_clipped": false, "step_seconds": 2.061947506997967, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1180, "loss": 0.19261740148067474, "loss_ema": 0.13087670833010195, "loss_mean_50": 0.13234735034406186, "loss_std_50": 0.14476923524237906, "lr": 0.00025, "grad_norm_before_clip": 0.026611328125, "grad_clipped": false, "step_seconds": 0.7237231649996829, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1181, "loss": 0.04587821289896965, "loss_ema": 0.12662678355854534, "loss_mean_50": 0.12215038046240806, "loss_std_50": 0.13197988452778925, "lr": 0.00025, "grad_norm_before_clip": 0.032958984375, "grad_clipped": false, "step_seconds": 1.2672638760050177, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1182, "loss": 0.048675328493118286, "loss_ema": 0.12272921080527398, "loss_mean_50": 0.1205912907421589, "loss_std_50": 0.13237760202490317, "lr": 0.00025, "grad_norm_before_clip": 0.0179443359375, "grad_clipped": false, "step_seconds": 2.06189119299961, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1183, "loss": 0.09170113503932953, "loss_ema": 0.12117780701697675, "loss_mean_50": 0.12136826165020466, "loss_std_50": 0.13209143811562582, "lr": 0.00025, "grad_norm_before_clip": 0.0830078125, "grad_clipped": false, "step_seconds": 0.47590268399653723, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1184, "loss": 0.9793730974197388, "loss_ema": 0.16408757153711484, "loss_mean_50": 0.13991139329969882, "loss_std_50": 0.17813508069640385, "lr": 0.00025, "grad_norm_before_clip": 0.0439453125, "grad_clipped": false, "step_seconds": 1.1028508700037492, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1185, "loss": 0.07548976689577103, "loss_ema": 0.15965768130504765, "loss_mean_50": 0.14007531978189947, "loss_std_50": 0.17807193963973192, "lr": 0.00025, "grad_norm_before_clip": 0.06689453125, "grad_clipped": false, "step_seconds": 1.8884916919996613, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1186, "loss": 0.12093404680490494, "loss_ema": 0.1577214995800405, "loss_mean_50": 0.14140997000038624, "loss_std_50": 0.1776729468074517, "lr": 0.00025, "grad_norm_before_clip": 0.052490234375, "grad_clipped": false, "step_seconds": 0.4211382869980298, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1187, "loss": 0.07402525842189789, "loss_ema": 0.15353668752213337, "loss_mean_50": 0.1412931790202856, "loss_std_50": 0.1777152785594896, "lr": 0.00025, "grad_norm_before_clip": 0.042724609375, "grad_clipped": false, "step_seconds": 2.0588597699970705, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1188, "loss": 0.0651393011212349, "loss_ema": 0.14911681820208844, "loss_mean_50": 0.13955226697027684, "loss_std_50": 0.1780261360806231, "lr": 0.00025, "grad_norm_before_clip": 0.054443359375, "grad_clipped": false, "step_seconds": 1.8298344840004575, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1189, "loss": 0.05174359679222107, "loss_ema": 0.14424815713159506, "loss_mean_50": 0.13885785289108754, "loss_std_50": 0.1782993649032643, "lr": 0.00025, "grad_norm_before_clip": 0.024658203125, "grad_clipped": false, "step_seconds": 1.8904481339995982, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1190, "loss": 0.0513555146753788, "loss_ema": 0.13960352500878423, "loss_mean_50": 0.13882594861090183, "loss_std_50": 0.1783148760230068, "lr": 0.00025, "grad_norm_before_clip": 0.053466796875, "grad_clipped": false, "step_seconds": 1.5861361080023926, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1191, "loss": 0.05894956365227699, "loss_ema": 0.13557082694095884, "loss_mean_50": 0.13794108718633652, "loss_std_50": 0.17859905405094575, "lr": 0.00025, "grad_norm_before_clip": 0.0286865234375, "grad_clipped": false, "step_seconds": 1.8287190219998593, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1192, "loss": 0.0559990368783474, "loss_ema": 0.13159223743782827, "loss_mean_50": 0.1381657572835684, "loss_std_50": 0.17848873337493992, "lr": 0.00025, "grad_norm_before_clip": 0.01251220703125, "grad_clipped": false, "step_seconds": 2.0636138410045533, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1193, "loss": 0.08106376975774765, "loss_ema": 0.12906581405382422, "loss_mean_50": 0.12835552342236042, "loss_std_50": 0.16754199633505445, "lr": 0.00025, "grad_norm_before_clip": 0.060546875, "grad_clipped": false, "step_seconds": 1.8281814479996683, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1194, "loss": 0.10458951443433762, "loss_ema": 0.1278419990728499, "loss_mean_50": 0.1293707238137722, "loss_std_50": 0.16724085562179677, "lr": 0.00025, "grad_norm_before_clip": 0.11962890625, "grad_clipped": false, "step_seconds": 0.3469766799971694, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1195, "loss": 0.0888407900929451, "loss_ema": 0.12589193862385467, "loss_mean_50": 0.12885196432471274, "loss_std_50": 0.16732552011796042, "lr": 0.00025, "grad_norm_before_clip": 0.034423828125, "grad_clipped": false, "step_seconds": 0.3620962269997108, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1196, "loss": 0.07681656628847122, "loss_ema": 0.1234381700070855, "loss_mean_50": 0.12927744716405867, "loss_std_50": 0.16716553630148948, "lr": 0.00025, "grad_norm_before_clip": 0.0245361328125, "grad_clipped": false, "step_seconds": 2.0631365230001393, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1197, "loss": 0.04670270159840584, "loss_ema": 0.11960139658665152, "loss_mean_50": 0.1290466757118702, "loss_std_50": 0.16727137318214602, "lr": 0.00025, "grad_norm_before_clip": 0.03271484375, "grad_clipped": false, "step_seconds": 2.0634370089974254, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1198, "loss": 0.20959794521331787, "loss_ema": 0.12410122401798483, "loss_mean_50": 0.13226847298443317, "loss_std_50": 0.1672404650378103, "lr": 0.00025, "grad_norm_before_clip": 0.0159912109375, "grad_clipped": false, "step_seconds": 2.0619252570031676, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1199, "loss": 0.07916529476642609, "loss_ema": 0.12185442755540689, "loss_mean_50": 0.1325754486769438, "loss_std_50": 0.16712858653857413, "lr": 0.00025, "grad_norm_before_clip": 0.0260009765625, "grad_clipped": false, "step_seconds": 0.5201806609984487, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1200, "loss": 0.19222193956375122, "loss_ema": 0.1253728031558241, "loss_mean_50": 0.13503573708236216, "loss_std_50": 0.16708307995532473, "lr": 0.00025, "grad_norm_before_clip": 0.025146484375, "grad_clipped": false, "step_seconds": 0.9837258490006207, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1201, "loss": 0.08642452955245972, "loss_ema": 0.12342538947565587, "loss_mean_50": 0.1293093513697386, "loss_std_50": 0.16371048376962893, "lr": 0.00025, "grad_norm_before_clip": 0.01080322265625, "grad_clipped": false, "step_seconds": 1.8925688649978838, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1202, "loss": 0.09641312807798386, "loss_ema": 0.12207477640577227, "loss_mean_50": 0.12861994929611684, "loss_std_50": 0.1637749703373945, "lr": 0.00025, "grad_norm_before_clip": 0.0218505859375, "grad_clipped": false, "step_seconds": 1.890375450006104, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1203, "loss": 0.06636839359998703, "loss_ema": 0.11928945726548301, "loss_mean_50": 0.12831803910434247, "loss_std_50": 0.16387550466525688, "lr": 0.00025, "grad_norm_before_clip": 0.048583984375, "grad_clipped": false, "step_seconds": 0.6735222269999213, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1204, "loss": 0.1372094303369522, "loss_ema": 0.12018545591905647, "loss_mean_50": 0.12873404003679753, "loss_std_50": 0.16387114690455373, "lr": 0.00025, "grad_norm_before_clip": 0.01287841796875, "grad_clipped": false, "step_seconds": 2.061012710997602, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1205, "loss": 0.23215174674987793, "loss_ema": 0.12578377046059752, "loss_mean_50": 0.13072758175432683, "loss_std_50": 0.16450958384114064, "lr": 0.00025, "grad_norm_before_clip": 0.0203857421875, "grad_clipped": false, "step_seconds": 2.076570997996896, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1206, "loss": 0.65834641456604, "loss_ema": 0.15241190266586963, "loss_mean_50": 0.14194825164973734, "loss_std_50": 0.18022985769499728, "lr": 0.00025, "grad_norm_before_clip": 0.0234375, "grad_clipped": false, "step_seconds": 1.8893266240047524, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1207, "loss": 0.05352099612355232, "loss_ema": 0.14746735733875374, "loss_mean_50": 0.14180002219974994, "loss_std_50": 0.18029946223213764, "lr": 0.00025, "grad_norm_before_clip": 0.024658203125, "grad_clipped": false, "step_seconds": 1.889869260994601, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1208, "loss": 0.10348226130008698, "loss_ema": 0.1452681025368204, "loss_mean_50": 0.1317260567098856, "loss_std_50": 0.16764306089982467, "lr": 0.00025, "grad_norm_before_clip": 0.02099609375, "grad_clipped": false, "step_seconds": 2.0590488300003926, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1209, "loss": 0.1671801507472992, "loss_ema": 0.14636370494734433, "loss_mean_50": 0.13435830913484095, "loss_std_50": 0.1671450798665896, "lr": 0.00025, "grad_norm_before_clip": 0.0194091796875, "grad_clipped": false, "step_seconds": 2.0621818720028386, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1210, "loss": 0.06515242904424667, "loss_ema": 0.14230314115218945, "loss_mean_50": 0.1343051529675722, "loss_std_50": 0.16716665653712381, "lr": 0.00025, "grad_norm_before_clip": 0.034423828125, "grad_clipped": false, "step_seconds": 0.7196714849997079, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1211, "loss": 0.051718249917030334, "loss_ema": 0.1377738965904315, "loss_mean_50": 0.13422971457242966, "loss_std_50": 0.16720305399761104, "lr": 0.00025, "grad_norm_before_clip": 0.034912109375, "grad_clipped": false, "step_seconds": 1.2908451219991548, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1212, "loss": 0.057817492634058, "loss_ema": 0.1337760763926128, "loss_mean_50": 0.13408063732087613, "loss_std_50": 0.16726778078523638, "lr": 0.00025, "grad_norm_before_clip": 0.0115966796875, "grad_clipped": false, "step_seconds": 1.8290233310035546, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1213, "loss": 0.045965805649757385, "loss_ema": 0.12938556285547004, "loss_mean_50": 0.1335559054464102, "loss_std_50": 0.16750206353909755, "lr": 0.00025, "grad_norm_before_clip": 0.400390625, "grad_clipped": false, "step_seconds": 1.2671396169971558, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1214, "loss": 0.1067342758178711, "loss_ema": 0.1282529985035901, "loss_mean_50": 0.13206333003938198, "loss_std_50": 0.1674018848482504, "lr": 0.00025, "grad_norm_before_clip": 0.06787109375, "grad_clipped": false, "step_seconds": 2.058668364996265, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1215, "loss": 0.07686663419008255, "loss_ema": 0.1256836802879147, "loss_mean_50": 0.13259053304791452, "loss_std_50": 0.16718557422486782, "lr": 0.00025, "grad_norm_before_clip": 0.09130859375, "grad_clipped": false, "step_seconds": 1.8890226540024742, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1216, "loss": 0.05244786664843559, "loss_ema": 0.12202188960594075, "loss_mean_50": 0.12966038413345815, "loss_std_50": 0.1672806077137979, "lr": 0.00025, "grad_norm_before_clip": 0.0225830078125, "grad_clipped": false, "step_seconds": 0.5719450949982274, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1217, "loss": 0.11136604100465775, "loss_ema": 0.1214890971758766, "loss_mean_50": 0.12865310601890087, "loss_std_50": 0.1672360954782246, "lr": 0.00025, "grad_norm_before_clip": 0.0084228515625, "grad_clipped": false, "step_seconds": 2.0631811800049036, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1218, "loss": 0.6326887011528015, "loss_ema": 0.14704907737472284, "loss_mean_50": 0.14037960797548293, "loss_std_50": 0.18104145149889672, "lr": 0.00025, "grad_norm_before_clip": 0.05224609375, "grad_clipped": false, "step_seconds": 2.0636202589957975, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1219, "loss": 0.07098419964313507, "loss_ema": 0.14324583348814343, "loss_mean_50": 0.14048294126987457, "loss_std_50": 0.1810003339285701, "lr": 0.00025, "grad_norm_before_clip": 0.05615234375, "grad_clipped": false, "step_seconds": 1.5847888409989537, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1220, "loss": 0.059927068650722504, "loss_ema": 0.13907989524627237, "loss_mean_50": 0.13874475106596948, "loss_std_50": 0.18134794505350374, "lr": 0.00025, "grad_norm_before_clip": 0.033203125, "grad_clipped": false, "step_seconds": 0.6680211260027136, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1221, "loss": 0.806513786315918, "loss_ema": 0.17245158979975467, "loss_mean_50": 0.153502134680748, "loss_std_50": 0.20368928760442043, "lr": 0.00025, "grad_norm_before_clip": 0.1396484375, "grad_clipped": false, "step_seconds": 1.2890241950008203, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1222, "loss": 0.05860481783747673, "loss_ema": 0.16675925120164076, "loss_mean_50": 0.15364837616682053, "loss_std_50": 0.20361846507704726, "lr": 0.00025, "grad_norm_before_clip": 0.0213623046875, "grad_clipped": false, "step_seconds": 0.6130518930003745, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1223, "loss": 0.48341289162635803, "loss_ema": 0.18259193322287662, "loss_mean_50": 0.1616784653067589, "loss_std_50": 0.20848970554352772, "lr": 0.00025, "grad_norm_before_clip": 0.04296875, "grad_clipped": false, "step_seconds": 2.0626946479969774, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1224, "loss": 0.03478921204805374, "loss_ema": 0.17520179716413545, "loss_mean_50": 0.150336382240057, "loss_std_50": 0.19946315187573033, "lr": 0.00025, "grad_norm_before_clip": 0.0172119140625, "grad_clipped": false, "step_seconds": 2.0629655640004785, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1225, "loss": 0.07341063767671585, "loss_ema": 0.17011223918976445, "loss_mean_50": 0.14951679348945618, "loss_std_50": 0.19969322946516335, "lr": 0.00025, "grad_norm_before_clip": 0.028564453125, "grad_clipped": false, "step_seconds": 2.0579418820052524, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1226, "loss": 0.37849223613739014, "loss_ema": 0.18053123903714574, "loss_mean_50": 0.15539329692721368, "loss_std_50": 0.20200826201083677, "lr": 0.00025, "grad_norm_before_clip": 0.041015625, "grad_clipped": false, "step_seconds": 0.7242091189982602, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1227, "loss": 0.09317467361688614, "loss_ema": 0.17616341076613273, "loss_mean_50": 0.15577981919050216, "loss_std_50": 0.2018703068357919, "lr": 0.00025, "grad_norm_before_clip": 0.059814453125, "grad_clipped": false, "step_seconds": 1.8893889370010584, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1228, "loss": 0.06681230664253235, "loss_ema": 0.1706958555599527, "loss_mean_50": 0.15527777567505838, "loss_std_50": 0.20205963841211624, "lr": 0.00025, "grad_norm_before_clip": 0.080078125, "grad_clipped": false, "step_seconds": 0.5978844499986735, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1229, "loss": 0.6270912885665894, "loss_ema": 0.19351562721028454, "loss_mean_50": 0.1669189489632845, "loss_std_50": 0.21190011918348992, "lr": 0.00025, "grad_norm_before_clip": 0.12890625, "grad_clipped": false, "step_seconds": 0.41992195599596016, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1230, "loss": 0.07750780880451202, "loss_ema": 0.18771523628999592, "loss_mean_50": 0.16461675710976123, "loss_std_50": 0.21223345460388865, "lr": 0.00025, "grad_norm_before_clip": 0.04296875, "grad_clipped": false, "step_seconds": 0.36122544800309697, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1231, "loss": 0.06514882296323776, "loss_ema": 0.18158691562365803, "loss_mean_50": 0.1650021693110466, "loss_std_50": 0.2120348822042674, "lr": 0.00025, "grad_norm_before_clip": 0.054931640625, "grad_clipped": false, "step_seconds": 1.102934345995891, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1232, "loss": 0.04864552244544029, "loss_ema": 0.17493984596474713, "loss_mean_50": 0.16500157319009304, "loss_std_50": 0.21203520928969496, "lr": 0.00025, "grad_norm_before_clip": 0.03125, "grad_clipped": false, "step_seconds": 2.0627170350053348, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1233, "loss": 0.06795345991849899, "loss_ema": 0.1695905266624347, "loss_mean_50": 0.16452661968767643, "loss_std_50": 0.21222538028903357, "lr": 0.00025, "grad_norm_before_clip": 0.03759765625, "grad_clipped": false, "step_seconds": 1.889899770001648, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1234, "loss": 0.12522944808006287, "loss_ema": 0.16737247273331612, "loss_mean_50": 0.14744374670088292, "loss_std_50": 0.17748007549531358, "lr": 0.00025, "grad_norm_before_clip": 0.012451171875, "grad_clipped": false, "step_seconds": 1.5701410490000853, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1235, "loss": 0.10328744351863861, "loss_ema": 0.16416822127258224, "loss_mean_50": 0.14799970023334028, "loss_std_50": 0.177297253761897, "lr": 0.00025, "grad_norm_before_clip": 0.109375, "grad_clipped": false, "step_seconds": 1.8909499109940953, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1236, "loss": 0.3297804594039917, "loss_ema": 0.1724488331791527, "loss_mean_50": 0.15217662848532199, "loss_std_50": 0.17906173057934882, "lr": 0.00025, "grad_norm_before_clip": 0.0196533203125, "grad_clipped": false, "step_seconds": 2.0617398930044146, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1237, "loss": 0.16623879969120026, "loss_ema": 0.17213833150475505, "loss_mean_50": 0.15402089931070806, "loss_std_50": 0.1787218626344259, "lr": 0.00025, "grad_norm_before_clip": 0.034912109375, "grad_clipped": false, "step_seconds": 0.5199384269944858, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1238, "loss": 0.05603407323360443, "loss_ema": 0.1663331185911975, "loss_mean_50": 0.15383879475295545, "loss_std_50": 0.17881694722035615, "lr": 0.00025, "grad_norm_before_clip": 0.035888671875, "grad_clipped": false, "step_seconds": 0.4763622330065118, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1239, "loss": 0.04390610754489899, "loss_ema": 0.16021176803888257, "loss_mean_50": 0.153682044968009, "loss_std_50": 0.178909785557178, "lr": 0.00025, "grad_norm_before_clip": 0.0322265625, "grad_clipped": false, "step_seconds": 2.0625225199983106, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1240, "loss": 0.9612155556678772, "loss_ema": 0.20026195742033231, "loss_mean_50": 0.17187924578785896, "loss_std_50": 0.2109748001361015, "lr": 0.00025, "grad_norm_before_clip": 0.1171875, "grad_clipped": false, "step_seconds": 2.061865516996477, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1241, "loss": 0.03878214210271835, "loss_ema": 0.19218796665445162, "loss_mean_50": 0.1714758973568678, "loss_std_50": 0.211209465035433, "lr": 0.00025, "grad_norm_before_clip": 0.032958984375, "grad_clipped": false, "step_seconds": 1.827417050000804, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1242, "loss": 0.042187441140413284, "loss_ema": 0.1846879403787497, "loss_mean_50": 0.1711996654421091, "loss_std_50": 0.21136928302991478, "lr": 0.00025, "grad_norm_before_clip": 0.0228271484375, "grad_clipped": false, "step_seconds": 0.598020623998309, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1243, "loss": 0.2178618460893631, "loss_ema": 0.18634663566428036, "loss_mean_50": 0.17393562696874143, "loss_std_50": 0.21107000239575427, "lr": 0.00025, "grad_norm_before_clip": 0.0235595703125, "grad_clipped": false, "step_seconds": 1.8274312390058185, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1244, "loss": 0.10457722842693329, "loss_ema": 0.18225816530241298, "loss_mean_50": 0.17393538124859334, "loss_std_50": 0.2110700831330114, "lr": 0.00025, "grad_norm_before_clip": 0.1123046875, "grad_clipped": false, "step_seconds": 0.3476294259962742, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1245, "loss": 0.04480186477303505, "loss_ema": 0.1753853502759441, "loss_mean_50": 0.17305460274219514, "loss_std_50": 0.21151475536157502, "lr": 0.00025, "grad_norm_before_clip": 0.0400390625, "grad_clipped": false, "step_seconds": 1.8894433320019743, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1246, "loss": 0.05191872641444206, "loss_ema": 0.16921201908286898, "loss_mean_50": 0.17255664594471454, "loss_std_50": 0.21176989066474786, "lr": 0.00025, "grad_norm_before_clip": 0.03515625, "grad_clipped": false, "step_seconds": 1.966668788001698, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1247, "loss": 0.11804146319627762, "loss_ema": 0.16665349128853943, "loss_mean_50": 0.17398342117667198, "loss_std_50": 0.21115658810808666, "lr": 0.00025, "grad_norm_before_clip": 0.00921630859375, "grad_clipped": false, "step_seconds": 2.061880001994723, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1248, "loss": 0.2639252543449402, "loss_ema": 0.17151707944135944, "loss_mean_50": 0.17506996735930444, "loss_std_50": 0.21147658711892292, "lr": 0.00025, "grad_norm_before_clip": 0.0211181640625, "grad_clipped": false, "step_seconds": 2.0711323310024454, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1249, "loss": 0.0810086578130722, "loss_ema": 0.16699165835994506, "loss_mean_50": 0.17510683462023735, "loss_std_50": 0.2114600246260285, "lr": 0.00025, "grad_norm_before_clip": 0.0155029296875, "grad_clipped": false, "step_seconds": 1.2892871169970022, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1250, "loss": 0.08849935233592987, "loss_ema": 0.1630670430587443, "loss_mean_50": 0.17303238287568093, "loss_std_50": 0.21179045598942384, "lr": 0.00025, "grad_norm_before_clip": 0.04736328125, "grad_clipped": false, "step_seconds": 0.34713642000133405, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1251, "loss": 0.0632392168045044, "loss_ema": 0.1580756517460323, "loss_mean_50": 0.1725686766207218, "loss_std_50": 0.21200484571323056, "lr": 0.00025, "grad_norm_before_clip": 0.0189208984375, "grad_clipped": false, "step_seconds": 0.4750046730041504, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1252, "loss": 0.050867896527051926, "loss_ema": 0.15271526398508328, "loss_mean_50": 0.17165777198970317, "loss_std_50": 0.21242752435754444, "lr": 0.00025, "grad_norm_before_clip": 0.042724609375, "grad_clipped": false, "step_seconds": 0.5199203230004059, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1253, "loss": 0.08426754921674728, "loss_ema": 0.14929287824666648, "loss_mean_50": 0.17201575510203837, "loss_std_50": 0.2122648084574047, "lr": 0.00025, "grad_norm_before_clip": 0.271484375, "grad_clipped": false, "step_seconds": 0.7238810269991518, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1254, "loss": 0.06072119250893593, "loss_ema": 0.14486429395977993, "loss_mean_50": 0.17048599034547807, "loss_std_50": 0.21278512334660768, "lr": 0.00025, "grad_norm_before_clip": 0.08349609375, "grad_clipped": false, "step_seconds": 2.0630345449972083, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1255, "loss": 0.056120891124010086, "loss_ema": 0.14042712381799144, "loss_mean_50": 0.1669653732329607, "loss_std_50": 0.21319157624286725, "lr": 0.00025, "grad_norm_before_clip": 0.032470703125, "grad_clipped": false, "step_seconds": 1.8899548199988203, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1256, "loss": 0.06305374205112457, "loss_ema": 0.1365584547296481, "loss_mean_50": 0.15505951978266239, "loss_std_50": 0.20173186367255933, "lr": 0.00025, "grad_norm_before_clip": 0.017578125, "grad_clipped": false, "step_seconds": 1.5690558310016058, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1257, "loss": 0.0962771326303482, "loss_ema": 0.13454438862468307, "loss_mean_50": 0.1559146425127983, "loss_std_50": 0.20138996879430102, "lr": 0.00025, "grad_norm_before_clip": 0.10888671875, "grad_clipped": false, "step_seconds": 0.3615171290002763, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1258, "loss": 0.05825484171509743, "loss_ema": 0.13072991127920378, "loss_mean_50": 0.1550100941210985, "loss_std_50": 0.20172473067281768, "lr": 0.00025, "grad_norm_before_clip": 0.0203857421875, "grad_clipped": false, "step_seconds": 1.1036518449982395, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1259, "loss": 0.05371613800525665, "loss_ema": 0.1268792226155064, "loss_mean_50": 0.15274081386625768, "loss_std_50": 0.2022126713079378, "lr": 0.00025, "grad_norm_before_clip": 0.06982421875, "grad_clipped": false, "step_seconds": 2.063186001003487, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1260, "loss": 0.30010277032852173, "loss_ema": 0.13554040000115716, "loss_mean_50": 0.15743982069194318, "loss_std_50": 0.20285157182895427, "lr": 0.00025, "grad_norm_before_clip": 0.0291748046875, "grad_clipped": false, "step_seconds": 2.0581459800014272, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1261, "loss": 0.17676910758018494, "loss_ema": 0.13760183538010853, "loss_mean_50": 0.15994083784520627, "loss_std_50": 0.20230283369929927, "lr": 0.00025, "grad_norm_before_clip": 0.0230712890625, "grad_clipped": false, "step_seconds": 1.5859735139965778, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1262, "loss": 0.1055913120508194, "loss_ema": 0.1360013092136441, "loss_mean_50": 0.16089631423354148, "loss_std_50": 0.2019307243932356, "lr": 0.00025, "grad_norm_before_clip": 0.0283203125, "grad_clipped": false, "step_seconds": 1.888865203000023, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1263, "loss": 0.9403309226036072, "loss_ema": 0.17621778988314224, "loss_mean_50": 0.1787836165726185, "loss_std_50": 0.22878428181901117, "lr": 0.00025, "grad_norm_before_clip": 0.1572265625, "grad_clipped": false, "step_seconds": 0.6673417549973237, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1264, "loss": 0.07888731360435486, "loss_ema": 0.17135126606920287, "loss_mean_50": 0.17822667732834815, "loss_std_50": 0.2289927961814292, "lr": 0.00025, "grad_norm_before_clip": 0.035888671875, "grad_clipped": false, "step_seconds": 0.5994029530047555, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1265, "loss": 0.0666746273636818, "loss_ema": 0.16611743413392682, "loss_mean_50": 0.17802283719182013, "loss_std_50": 0.22908744876643697, "lr": 0.00025, "grad_norm_before_clip": 0.05078125, "grad_clipped": false, "step_seconds": 2.0629646230008802, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1266, "loss": 0.04740433394908905, "loss_ema": 0.16018177912468493, "loss_mean_50": 0.1779219665378332, "loss_std_50": 0.2291438225453046, "lr": 0.00025, "grad_norm_before_clip": 0.014892578125, "grad_clipped": false, "step_seconds": 2.0588070469966624, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1267, "loss": 0.09978341311216354, "loss_ema": 0.15716186082405884, "loss_mean_50": 0.17769031397998333, "loss_std_50": 0.22921683313824065, "lr": 0.00025, "grad_norm_before_clip": 0.0225830078125, "grad_clipped": false, "step_seconds": 1.8898277890039026, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1268, "loss": 0.08230581134557724, "loss_ema": 0.15341905835013475, "loss_mean_50": 0.16668265618383885, "loss_std_50": 0.22013786926384563, "lr": 0.00025, "grad_norm_before_clip": 0.0191650390625, "grad_clipped": false, "step_seconds": 1.8900141129997792, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1269, "loss": 0.045464154332876205, "loss_ema": 0.1480213131492718, "loss_mean_50": 0.16617225527763366, "loss_std_50": 0.22038860130931295, "lr": 0.00025, "grad_norm_before_clip": 0.02880859375, "grad_clipped": false, "step_seconds": 1.2677880009941873, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1270, "loss": 0.08626589924097061, "loss_ema": 0.14493354245385673, "loss_mean_50": 0.16669903188943863, "loss_std_50": 0.22016538741262212, "lr": 0.00025, "grad_norm_before_clip": 0.0208740234375, "grad_clipped": false, "step_seconds": 1.8286378129996592, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1271, "loss": 0.06016683205962181, "loss_ema": 0.14069520693414497, "loss_mean_50": 0.1517720928043127, "loss_std_50": 0.2007229635504473, "lr": 0.00025, "grad_norm_before_clip": 0.041015625, "grad_clipped": false, "step_seconds": 0.4214081830068608, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1272, "loss": 0.11512182652950287, "loss_ema": 0.13941653791391287, "loss_mean_50": 0.15290243297815323, "loss_std_50": 0.2003539179687559, "lr": 0.00025, "grad_norm_before_clip": 0.0235595703125, "grad_clipped": false, "step_seconds": 1.8903260379956919, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1273, "loss": 0.19646026194095612, "loss_ema": 0.14226872411526503, "loss_mean_50": 0.14716338038444518, "loss_std_50": 0.19483828716276413, "lr": 0.00025, "grad_norm_before_clip": 0.0142822265625, "grad_clipped": false, "step_seconds": 1.8889274379980634, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1274, "loss": 0.14182451367378235, "loss_ema": 0.14224651359319088, "loss_mean_50": 0.14930408641695977, "loss_std_50": 0.19417874888352551, "lr": 0.00025, "grad_norm_before_clip": 0.0181884765625, "grad_clipped": false, "step_seconds": 0.5709674540048582, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1275, "loss": 0.10461441427469254, "loss_ema": 0.14036490862726594, "loss_mean_50": 0.1499281619489193, "loss_std_50": 0.1939838758051596, "lr": 0.00025, "grad_norm_before_clip": 0.03173828125, "grad_clipped": false, "step_seconds": 0.5974541830000817, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1276, "loss": 0.08643247187137604, "loss_ema": 0.13766828678947143, "loss_mean_50": 0.14408696666359902, "loss_std_50": 0.1913933850555085, "lr": 0.00025, "grad_norm_before_clip": 0.024658203125, "grad_clipped": false, "step_seconds": 1.8918025419989135, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1277, "loss": 0.12234540283679962, "loss_ema": 0.13690214259183786, "loss_mean_50": 0.14467038124799728, "loss_std_50": 0.1912817297880696, "lr": 0.00025, "grad_norm_before_clip": 0.0203857421875, "grad_clipped": false, "step_seconds": 2.0647246049993555, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1278, "loss": 0.06450894474983215, "loss_ema": 0.13328248269973755, "loss_mean_50": 0.14462431401014328, "loss_std_50": 0.19130075156916984, "lr": 0.00025, "grad_norm_before_clip": 0.0625, "grad_clipped": false, "step_seconds": 2.0644316210018587, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1279, "loss": 0.07691657543182373, "loss_ema": 0.13046418733634185, "loss_mean_50": 0.13362081974744797, "loss_std_50": 0.17863678174288428, "lr": 0.00025, "grad_norm_before_clip": 0.09033203125, "grad_clipped": false, "step_seconds": 2.059102362996782, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1280, "loss": 0.09932640939950943, "loss_ema": 0.12890729843950022, "loss_mean_50": 0.1340571917593479, "loss_std_50": 0.1785257911412318, "lr": 0.00025, "grad_norm_before_clip": 0.051025390625, "grad_clipped": false, "step_seconds": 0.6133735490002437, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1281, "loss": 0.12365760654211044, "loss_ema": 0.12864481384463072, "loss_mean_50": 0.13522736743092537, "loss_std_50": 0.17826184267293818, "lr": 0.00025, "grad_norm_before_clip": 0.0216064453125, "grad_clipped": false, "step_seconds": 2.0675959770014742, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1282, "loss": 0.0551651194691658, "loss_ema": 0.12497082912585747, "loss_mean_50": 0.13535775937139988, "loss_std_50": 0.1782008375525918, "lr": 0.00025, "grad_norm_before_clip": 0.0201416015625, "grad_clipped": false, "step_seconds": 2.1973377999966033, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1283, "loss": 0.03925454616546631, "loss_ema": 0.12068501497783791, "loss_mean_50": 0.13478378109633923, "loss_std_50": 0.17846304607245822, "lr": 0.00025, "grad_norm_before_clip": 0.0107421875, "grad_clipped": false, "step_seconds": 2.0640295070043067, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1284, "loss": 0.07307969778776169, "loss_ema": 0.11830474911833411, "loss_mean_50": 0.1337407860904932, "loss_std_50": 0.17866810892610202, "lr": 0.00025, "grad_norm_before_clip": 0.01019287109375, "grad_clipped": false, "step_seconds": 1.8909191439961432, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1285, "loss": 0.07822477072477341, "loss_ema": 0.11630075019865607, "loss_mean_50": 0.1332395326346159, "loss_std_50": 0.17878795913027307, "lr": 0.00025, "grad_norm_before_clip": 0.00872802734375, "grad_clipped": false, "step_seconds": 2.062603686004877, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1286, "loss": 0.05935756489634514, "loss_ema": 0.11345359093354052, "loss_mean_50": 0.12783107474446298, "loss_std_50": 0.17684028697796467, "lr": 0.00025, "grad_norm_before_clip": 0.076171875, "grad_clipped": false, "step_seconds": 0.5724509959982242, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1287, "loss": 0.26840001344680786, "loss_ema": 0.12120091205920387, "loss_mean_50": 0.12987429901957512, "loss_std_50": 0.1778595001591337, "lr": 0.00025, "grad_norm_before_clip": 0.038818359375, "grad_clipped": false, "step_seconds": 2.0628920390008716, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1288, "loss": 0.1741477996110916, "loss_ema": 0.12384825643679825, "loss_mean_50": 0.13223657354712487, "loss_std_50": 0.17764733810030323, "lr": 0.00025, "grad_norm_before_clip": 0.1943359375, "grad_clipped": false, "step_seconds": 1.8289219430007506, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1289, "loss": 0.0631626546382904, "loss_ema": 0.12081397634687285, "loss_mean_50": 0.1326217044889927, "loss_std_50": 0.1774762156156879, "lr": 0.00025, "grad_norm_before_clip": 0.029541015625, "grad_clipped": false, "step_seconds": 1.2916679959962494, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1290, "loss": 0.0934712216258049, "loss_ema": 0.11944683861081945, "loss_mean_50": 0.11526681780815125, "loss_std_50": 0.13227212412107323, "lr": 0.00025, "grad_norm_before_clip": 0.16796875, "grad_clipped": false, "step_seconds": 1.8281147119996604, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1291, "loss": 0.06935068964958191, "loss_ema": 0.11694203116275756, "loss_mean_50": 0.11587818875908852, "loss_std_50": 0.13198753255215864, "lr": 0.00025, "grad_norm_before_clip": 0.03759765625, "grad_clipped": false, "step_seconds": 0.6074758020040463, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1292, "loss": 0.06390590965747833, "loss_ema": 0.1142902250874936, "loss_mean_50": 0.11631255812942981, "loss_std_50": 0.13177987669847613, "lr": 0.00025, "grad_norm_before_clip": 0.0267333984375, "grad_clipped": false, "step_seconds": 1.889328311997815, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1293, "loss": 0.04832611605525017, "loss_ema": 0.11099201963588143, "loss_mean_50": 0.11292184352874755, "loss_std_50": 0.13130360591231688, "lr": 0.00025, "grad_norm_before_clip": 0.0194091796875, "grad_clipped": false, "step_seconds": 0.598036374001822, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1294, "loss": 0.10967739671468735, "loss_ema": 0.11092628848982172, "loss_mean_50": 0.11302384689450264, "loss_std_50": 0.13129906472465458, "lr": 0.00025, "grad_norm_before_clip": 0.015625, "grad_clipped": false, "step_seconds": 2.0638033249997534, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1295, "loss": 0.06523193418979645, "loss_ema": 0.10864157077482045, "loss_mean_50": 0.11343244828283787, "loss_std_50": 0.1311177867567262, "lr": 0.00025, "grad_norm_before_clip": 0.0556640625, "grad_clipped": false, "step_seconds": 2.063844110998616, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1296, "loss": 0.05859918147325516, "loss_ema": 0.10613945130974217, "loss_mean_50": 0.11356605738401412, "loss_std_50": 0.13105842642122856, "lr": 0.00025, "grad_norm_before_clip": 0.0150146484375, "grad_clipped": false, "step_seconds": 2.062919044001319, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1297, "loss": 0.10055626183748245, "loss_ema": 0.10586029183612919, "loss_mean_50": 0.11321635335683823, "loss_std_50": 0.13106934559065106, "lr": 0.00025, "grad_norm_before_clip": 0.00909423828125, "grad_clipped": false, "step_seconds": 1.103531672997633, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1298, "loss": 0.06375201046466827, "loss_ema": 0.10375487776755614, "loss_mean_50": 0.10921288847923279, "loss_std_50": 0.1294519856554914, "lr": 0.00025, "grad_norm_before_clip": 0.037841796875, "grad_clipped": false, "step_seconds": 1.8907208110031206, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1299, "loss": 0.041819922626018524, "loss_ema": 0.10065813001047925, "loss_mean_50": 0.10842911377549172, "loss_std_50": 0.1297386948899355, "lr": 0.00025, "grad_norm_before_clip": 0.01458740234375, "grad_clipped": false, "step_seconds": 2.062782030996459, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1300, "loss": 0.1315813511610031, "loss_ema": 0.10220429106800544, "loss_mean_50": 0.10929075375199318, "loss_std_50": 0.1297465342211229, "lr": 0.00025, "grad_norm_before_clip": 0.0341796875, "grad_clipped": false, "step_seconds": 1.2659519630033174, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1301, "loss": 0.6742545366287231, "loss_ema": 0.13080680334604133, "loss_mean_50": 0.12151106014847755, "loss_std_50": 0.15174351350918508, "lr": 0.00025, "grad_norm_before_clip": 0.1259765625, "grad_clipped": false, "step_seconds": 0.5212269339972408, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1302, "loss": 0.05851924791932106, "loss_ema": 0.1271924255747053, "loss_mean_50": 0.12166408717632293, "loss_std_50": 0.1516760386866376, "lr": 0.00025, "grad_norm_before_clip": 0.03271484375, "grad_clipped": false, "step_seconds": 1.569088809999812, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1303, "loss": 0.5139968395233154, "loss_ema": 0.1465326462721358, "loss_mean_50": 0.1302586729824543, "loss_std_50": 0.16119020924593255, "lr": 0.00025, "grad_norm_before_clip": 0.125, "grad_clipped": false, "step_seconds": 0.3624310029990738, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1304, "loss": 0.08700094372034073, "loss_ema": 0.14355606114454603, "loss_mean_50": 0.13078426800668239, "loss_std_50": 0.1610053499285791, "lr": 0.00025, "grad_norm_before_clip": 0.0751953125, "grad_clipped": false, "step_seconds": 2.0628463930042926, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1305, "loss": 0.676829993724823, "loss_ema": 0.17021975777355985, "loss_mean_50": 0.14319845005869866, "loss_std_50": 0.177821362845284, "lr": 0.00025, "grad_norm_before_clip": 0.052734375, "grad_clipped": false, "step_seconds": 0.7230473289964721, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1306, "loss": 0.16829736530780792, "loss_ema": 0.17012363815027226, "loss_mean_50": 0.14530332252383232, "loss_std_50": 0.1774827941592112, "lr": 0.00025, "grad_norm_before_clip": 0.146484375, "grad_clipped": false, "step_seconds": 0.3485776790039381, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1307, "loss": 0.15538695454597473, "loss_ema": 0.16938680397005737, "loss_mean_50": 0.14648551896214485, "loss_std_50": 0.17734911017799862, "lr": 0.00025, "grad_norm_before_clip": 0.240234375, "grad_clipped": false, "step_seconds": 1.8335696409994853, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1308, "loss": 0.049625497311353683, "loss_ema": 0.16339873863712218, "loss_mean_50": 0.14631293207406998, "loss_std_50": 0.17743906368787096, "lr": 0.00025, "grad_norm_before_clip": 0.054931640625, "grad_clipped": false, "step_seconds": 1.2908172789975652, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1309, "loss": 0.9701846241950989, "loss_ema": 0.20373803291502102, "loss_mean_50": 0.16464230179786682, "loss_std_50": 0.2110745448984495, "lr": 0.00025, "grad_norm_before_clip": 0.2734375, "grad_clipped": false, "step_seconds": 0.4220961660030298, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1310, "loss": 0.07332850247621536, "loss_ema": 0.19721755639308072, "loss_mean_50": 0.1601068164408207, "loss_std_50": 0.2105508639376809, "lr": 0.00025, "grad_norm_before_clip": 0.01104736328125, "grad_clipped": false, "step_seconds": 2.058635626999603, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1311, "loss": 0.07030405104160309, "loss_ema": 0.19087188112550682, "loss_mean_50": 0.15797751531004905, "loss_std_50": 0.21090962632403715, "lr": 0.00025, "grad_norm_before_clip": 0.0203857421875, "grad_clipped": false, "step_seconds": 0.477415008994285, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1312, "loss": 0.08620879054069519, "loss_ema": 0.1856387265962662, "loss_mean_50": 0.15758986487984658, "loss_std_50": 0.21102333736947437, "lr": 0.00025, "grad_norm_before_clip": 0.0120849609375, "grad_clipped": false, "step_seconds": 1.8902172689995496, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1313, "loss": 0.074375219643116, "loss_ema": 0.1800755512486087, "loss_mean_50": 0.14027075082063675, "loss_std_50": 0.17920859208604464, "lr": 0.00025, "grad_norm_before_clip": 0.008544921875, "grad_clipped": false, "step_seconds": 1.8297530280033243, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1314, "loss": 0.20494772493839264, "loss_ema": 0.18131915993309786, "loss_mean_50": 0.1427919590473175, "loss_std_50": 0.17921402478285828, "lr": 0.00025, "grad_norm_before_clip": 0.050537109375, "grad_clipped": false, "step_seconds": 2.065934322999965, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1315, "loss": 0.0514264814555645, "loss_ema": 0.17482452600922116, "loss_mean_50": 0.14248699612915516, "loss_std_50": 0.17935620904431532, "lr": 0.00025, "grad_norm_before_clip": 0.0125732421875, "grad_clipped": false, "step_seconds": 2.0649516210032743, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1316, "loss": 0.06369960308074951, "loss_ema": 0.16926827986279758, "loss_mean_50": 0.14281290151178838, "loss_std_50": 0.17919787476851395, "lr": 0.00025, "grad_norm_before_clip": 0.018310546875, "grad_clipped": false, "step_seconds": 1.8910756039986154, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1317, "loss": 0.041119132190942764, "loss_ema": 0.16286082247920483, "loss_mean_50": 0.14163961589336396, "loss_std_50": 0.17966720190214064, "lr": 0.00025, "grad_norm_before_clip": 0.0174560546875, "grad_clipped": false, "step_seconds": 2.063494369998807, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1318, "loss": 0.06649589538574219, "loss_ema": 0.15804257612453168, "loss_mean_50": 0.14132341757416725, "loss_std_50": 0.17978521915151913, "lr": 0.00025, "grad_norm_before_clip": 0.033935546875, "grad_clipped": false, "step_seconds": 2.139635332998296, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1319, "loss": 0.061880774796009064, "loss_ema": 0.15323448605810555, "loss_mean_50": 0.14165174998342991, "loss_std_50": 0.17962477536809082, "lr": 0.00025, "grad_norm_before_clip": 0.01385498046875, "grad_clipped": false, "step_seconds": 2.064124836004339, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1320, "loss": 0.05477997660636902, "loss_ema": 0.1483117605855187, "loss_mean_50": 0.1410220315307379, "loss_std_50": 0.1798728596304946, "lr": 0.00025, "grad_norm_before_clip": 0.06298828125, "grad_clipped": false, "step_seconds": 1.83000051399722, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1321, "loss": 0.08924432098865509, "loss_ema": 0.14535838860567551, "loss_mean_50": 0.14160358130931855, "loss_std_50": 0.1796573816798407, "lr": 0.00025, "grad_norm_before_clip": 0.0155029296875, "grad_clipped": false, "step_seconds": 2.0635114559991052, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1322, "loss": 0.05807248502969742, "loss_ema": 0.1409940934268766, "loss_mean_50": 0.14046259447932244, "loss_std_50": 0.18000276707678378, "lr": 0.00025, "grad_norm_before_clip": 0.052978515625, "grad_clipped": false, "step_seconds": 0.6188716809992911, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1323, "loss": 0.3287114202976227, "loss_ema": 0.1503799597704139, "loss_mean_50": 0.14310761764645577, "loss_std_50": 0.18176918782060042, "lr": 0.00025, "grad_norm_before_clip": 0.0625, "grad_clipped": false, "step_seconds": 1.2984473199976492, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1324, "loss": 0.22936603426933289, "loss_ema": 0.15432926349535983, "loss_mean_50": 0.14485844805836678, "loss_std_50": 0.18216956275810472, "lr": 0.00025, "grad_norm_before_clip": 0.0145263671875, "grad_clipped": false, "step_seconds": 2.064629235006578, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1325, "loss": 0.0708867609500885, "loss_ema": 0.15015713836809627, "loss_mean_50": 0.1441838949918747, "loss_std_50": 0.1823796565914324, "lr": 0.00025, "grad_norm_before_clip": 0.04150390625, "grad_clipped": false, "step_seconds": 0.5750800129972049, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1326, "loss": 0.17437823116779327, "loss_ema": 0.1513681930080811, "loss_mean_50": 0.14594281017780303, "loss_std_50": 0.1822382363201278, "lr": 0.00025, "grad_norm_before_clip": 0.296875, "grad_clipped": false, "step_seconds": 1.9030901000005542, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1327, "loss": 0.14856311678886414, "loss_ema": 0.15122793919712024, "loss_mean_50": 0.14646716445684432, "loss_std_50": 0.18220730061424867, "lr": 0.00025, "grad_norm_before_clip": 0.03759765625, "grad_clipped": false, "step_seconds": 2.069464809996134, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1328, "loss": 0.05683014541864395, "loss_ema": 0.1465080495081964, "loss_mean_50": 0.14631358847022058, "loss_std_50": 0.18227953730366375, "lr": 0.00025, "grad_norm_before_clip": 0.051025390625, "grad_clipped": false, "step_seconds": 0.5775312460027635, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1329, "loss": 0.1291835755109787, "loss_ema": 0.1456418258083355, "loss_mean_50": 0.14735892847180365, "loss_std_50": 0.18202825830739985, "lr": 0.00025, "grad_norm_before_clip": 0.09228515625, "grad_clipped": false, "step_seconds": 1.8976308220007923, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1330, "loss": 0.06334133446216583, "loss_ema": 0.141526801241027, "loss_mean_50": 0.14663922697305679, "loss_std_50": 0.18228769984964138, "lr": 0.00025, "grad_norm_before_clip": 0.02099609375, "grad_clipped": false, "step_seconds": 1.894327242000145, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1331, "loss": 0.05976010486483574, "loss_ema": 0.13743846642221744, "loss_mean_50": 0.14536127693951129, "loss_std_50": 0.18266791973359187, "lr": 0.00025, "grad_norm_before_clip": 0.03515625, "grad_clipped": false, "step_seconds": 1.5965161249987432, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1332, "loss": 0.06559715420007706, "loss_ema": 0.1338464008111104, "loss_mean_50": 0.14556991763412952, "loss_std_50": 0.18257071161042793, "lr": 0.00025, "grad_norm_before_clip": 0.050048828125, "grad_clipped": false, "step_seconds": 1.8912197859972366, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1333, "loss": 0.3860020637512207, "loss_ema": 0.1464541839581159, "loss_mean_50": 0.15250486798584462, "loss_std_50": 0.18497044189566153, "lr": 0.00025, "grad_norm_before_clip": 0.0220947265625, "grad_clipped": false, "step_seconds": 0.6673046320065623, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1334, "loss": 0.06550728529691696, "loss_ema": 0.14240683902505594, "loss_mean_50": 0.15235341973602773, "loss_std_50": 0.18503849835414893, "lr": 0.00025, "grad_norm_before_clip": 0.0216064453125, "grad_clipped": false, "step_seconds": 1.910237126998254, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1335, "loss": 0.3019583821296692, "loss_ema": 0.15038441618028658, "loss_mean_50": 0.15682809196412564, "loss_std_50": 0.1858950105435346, "lr": 0.00025, "grad_norm_before_clip": 0.0184326171875, "grad_clipped": false, "step_seconds": 2.0676125639947713, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1336, "loss": 0.04116705060005188, "loss_ema": 0.14492354790127482, "loss_mean_50": 0.15646428167819976, "loss_std_50": 0.1861030952023843, "lr": 0.00025, "grad_norm_before_clip": 0.0595703125, "grad_clipped": false, "step_seconds": 0.5982741700063343, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1337, "loss": 0.07088174670934677, "loss_ema": 0.14122145784167842, "loss_mean_50": 0.15251391634345055, "loss_std_50": 0.18578119361908554, "lr": 0.00025, "grad_norm_before_clip": 0.068359375, "grad_clipped": false, "step_seconds": 0.6023170459957328, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1338, "loss": 0.055628702044487, "loss_ema": 0.13694182005181882, "loss_mean_50": 0.15014353439211844, "loss_std_50": 0.18624555723245403, "lr": 0.00025, "grad_norm_before_clip": 0.01556396484375, "grad_clipped": false, "step_seconds": 1.8325053270018543, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1339, "loss": 0.18516455590724945, "loss_ema": 0.13935295684459034, "loss_mean_50": 0.15258357241749765, "loss_std_50": 0.18588886403807597, "lr": 0.00025, "grad_norm_before_clip": 0.0361328125, "grad_clipped": false, "step_seconds": 1.2947627949979506, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1340, "loss": 0.6054868698120117, "loss_ema": 0.1626596524929614, "loss_mean_50": 0.16282388538122178, "loss_std_50": 0.19616918284803386, "lr": 0.00025, "grad_norm_before_clip": 0.0947265625, "grad_clipped": false, "step_seconds": 2.063523807999445, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1341, "loss": 0.05615153908729553, "loss_ema": 0.15733424682267813, "loss_mean_50": 0.16255990236997606, "loss_std_50": 0.19630362613845528, "lr": 0.00025, "grad_norm_before_clip": 0.0301513671875, "grad_clipped": false, "step_seconds": 0.4228893430044991, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1342, "loss": 0.1308477222919464, "loss_ema": 0.15600992059614152, "loss_mean_50": 0.1638987386226654, "loss_std_50": 0.19585398181895255, "lr": 0.00025, "grad_norm_before_clip": 0.0673828125, "grad_clipped": false, "step_seconds": 0.5785277030008729, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1343, "loss": 0.04326794296503067, "loss_ema": 0.15037282171458596, "loss_mean_50": 0.16379757516086102, "loss_std_50": 0.19591494867868114, "lr": 0.00025, "grad_norm_before_clip": 0.0115966796875, "grad_clipped": false, "step_seconds": 2.068372391993762, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1344, "loss": 0.0513196662068367, "loss_ema": 0.1454201639391985, "loss_mean_50": 0.162630420550704, "loss_std_50": 0.19640710422904448, "lr": 0.00025, "grad_norm_before_clip": 0.0238037109375, "grad_clipped": false, "step_seconds": 1.5732323480042396, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1345, "loss": 0.058382049202919006, "loss_ema": 0.1410682582023845, "loss_mean_50": 0.16249342285096646, "loss_std_50": 0.19647737014459044, "lr": 0.00025, "grad_norm_before_clip": 0.04052734375, "grad_clipped": false, "step_seconds": 1.8944026749959448, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1346, "loss": 0.08827313035726547, "loss_ema": 0.13842850181012853, "loss_mean_50": 0.16308690182864666, "loss_std_50": 0.19620728205182766, "lr": 0.00025, "grad_norm_before_clip": 0.0830078125, "grad_clipped": false, "step_seconds": 2.063215754002158, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1347, "loss": 0.05104581266641617, "loss_ema": 0.13405936735294288, "loss_mean_50": 0.16209669284522532, "loss_std_50": 0.19664480545082724, "lr": 0.00025, "grad_norm_before_clip": 0.0235595703125, "grad_clipped": false, "step_seconds": 0.5262117040037992, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1348, "loss": 0.3165624737739563, "loss_ema": 0.14318452267399356, "loss_mean_50": 0.1671529021114111, "loss_std_50": 0.19730021357605781, "lr": 0.00025, "grad_norm_before_clip": 0.07470703125, "grad_clipped": false, "step_seconds": 1.8926551829936216, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1349, "loss": 0.1663597673177719, "loss_ema": 0.14434328490618248, "loss_mean_50": 0.16964369900524617, "loss_std_50": 0.1964866806537125, "lr": 0.00025, "grad_norm_before_clip": 0.040283203125, "grad_clipped": false, "step_seconds": 1.8328899460029788, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1350, "loss": 0.046561580151319504, "loss_ema": 0.13945419966843933, "loss_mean_50": 0.1679433035850525, "loss_std_50": 0.19717538802520992, "lr": 0.00025, "grad_norm_before_clip": 0.018798828125, "grad_clipped": false, "step_seconds": 2.0688010309968377, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1351, "loss": 0.11664316803216934, "loss_ema": 0.13831364808662583, "loss_mean_50": 0.1567910762131214, "loss_std_50": 0.183519411102934, "lr": 0.00025, "grad_norm_before_clip": 0.1064453125, "grad_clipped": false, "step_seconds": 1.8942023330018856, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1352, "loss": 0.2313288003206253, "loss_ema": 0.1429644056983258, "loss_mean_50": 0.1602472672611475, "loss_std_50": 0.1832631970598179, "lr": 0.00025, "grad_norm_before_clip": 0.017822265625, "grad_clipped": false, "step_seconds": 2.065920360000746, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1353, "loss": 0.07465976476669312, "loss_ema": 0.13954917365174416, "loss_mean_50": 0.15146052576601504, "loss_std_50": 0.17649907020898095, "lr": 0.00025, "grad_norm_before_clip": 0.134765625, "grad_clipped": false, "step_seconds": 2.0668932659973507, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1354, "loss": 0.08752364665269852, "loss_ema": 0.13694789730179188, "loss_mean_50": 0.1514709798246622, "loss_std_50": 0.17649526739041024, "lr": 0.00025, "grad_norm_before_clip": 0.0157470703125, "grad_clipped": false, "step_seconds": 2.022518668003613, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1355, "loss": 0.0655522495508194, "loss_ema": 0.13337811491424326, "loss_mean_50": 0.13924542494118214, "loss_std_50": 0.16008970583244203, "lr": 0.00025, "grad_norm_before_clip": 0.07568359375, "grad_clipped": false, "step_seconds": 2.0664101599977585, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1356, "loss": 0.05661556497216225, "loss_ema": 0.1295399874171392, "loss_mean_50": 0.13701178893446922, "loss_std_50": 0.1604474940723476, "lr": 0.00025, "grad_norm_before_clip": 0.014892578125, "grad_clipped": false, "step_seconds": 0.48074298500432633, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1357, "loss": 0.04465298354625702, "loss_ema": 0.1252956372235951, "loss_mean_50": 0.13479710951447488, "loss_std_50": 0.16094205042061696, "lr": 0.00025, "grad_norm_before_clip": 0.03955078125, "grad_clipped": false, "step_seconds": 1.2721824199979892, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1358, "loss": 0.06945167481899261, "loss_ema": 0.12250343910336497, "loss_mean_50": 0.13519363306462764, "loss_std_50": 0.16075603506205247, "lr": 0.00025, "grad_norm_before_clip": 0.037841796875, "grad_clipped": false, "step_seconds": 1.8294622550019994, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1359, "loss": 0.2784179449081421, "loss_ema": 0.13029916439360384, "loss_mean_50": 0.12135829947888851, "loss_std_50": 0.11007793292991648, "lr": 0.00025, "grad_norm_before_clip": 0.0218505859375, "grad_clipped": false, "step_seconds": 0.6680821389963967, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1360, "loss": 0.09166280180215836, "loss_ema": 0.12836734626403157, "loss_mean_50": 0.12172498546540737, "loss_std_50": 0.10994778798663252, "lr": 0.00025, "grad_norm_before_clip": 0.055908203125, "grad_clipped": false, "step_seconds": 0.3656114190016524, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1361, "loss": 0.1307370662689209, "loss_ema": 0.12848583226427604, "loss_mean_50": 0.12293364576995373, "loss_std_50": 0.10970778151749103, "lr": 0.00025, "grad_norm_before_clip": 0.050537109375, "grad_clipped": false, "step_seconds": 1.892138873001386, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1362, "loss": 0.13640491664409637, "loss_ema": 0.12888178648326704, "loss_mean_50": 0.12393756829202175, "loss_std_50": 0.10959673657553096, "lr": 0.00025, "grad_norm_before_clip": 0.07373046875, "grad_clipped": false, "step_seconds": 1.111675929001649, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1363, "loss": 0.10455609112977982, "loss_ema": 0.12766550171559266, "loss_mean_50": 0.12454118572175503, "loss_std_50": 0.10940504841675606, "lr": 0.00025, "grad_norm_before_clip": 0.048828125, "grad_clipped": false, "step_seconds": 0.7321239859957132, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1364, "loss": 0.13751772046089172, "loss_ema": 0.1281581126528576, "loss_mean_50": 0.123192585632205, "loss_std_50": 0.10881961898009013, "lr": 0.00025, "grad_norm_before_clip": 0.0279541015625, "grad_clipped": false, "step_seconds": 0.5777113570002257, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1365, "loss": 0.08667770773172379, "loss_ema": 0.1260840924068009, "loss_mean_50": 0.1238976101577282, "loss_std_50": 0.10846599313784158, "lr": 0.00025, "grad_norm_before_clip": 0.017333984375, "grad_clipped": false, "step_seconds": 2.070843319001142, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1366, "loss": 0.09578400105237961, "loss_ema": 0.12456908783907984, "loss_mean_50": 0.1245392981171608, "loss_std_50": 0.10820254804374425, "lr": 0.00025, "grad_norm_before_clip": 0.0283203125, "grad_clipped": false, "step_seconds": 1.8956176889987546, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1367, "loss": 0.0646832212805748, "loss_ema": 0.12157479451115459, "loss_mean_50": 0.12501057989895345, "loss_std_50": 0.10788904399272024, "lr": 0.00025, "grad_norm_before_clip": 0.0517578125, "grad_clipped": false, "step_seconds": 1.5867679119983222, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1368, "loss": 0.08760684728622437, "loss_ema": 0.11987639714990807, "loss_mean_50": 0.12543279893696307, "loss_std_50": 0.10770036658684475, "lr": 0.00025, "grad_norm_before_clip": 0.1064453125, "grad_clipped": false, "step_seconds": 0.35275159700540826, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1369, "loss": 0.10232182592153549, "loss_ema": 0.11899866858848944, "loss_mean_50": 0.1262416199594736, "loss_std_50": 0.10737141108097834, "lr": 0.00025, "grad_norm_before_clip": 0.01434326171875, "grad_clipped": false, "step_seconds": 1.2960296719975304, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1370, "loss": 0.23198796808719635, "loss_ema": 0.12464813356342479, "loss_mean_50": 0.12978577978909014, "loss_std_50": 0.10787756635211528, "lr": 0.00025, "grad_norm_before_clip": 0.01507568359375, "grad_clipped": false, "step_seconds": 2.068059983001149, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1371, "loss": 0.07149305939674377, "loss_ema": 0.12199037985509073, "loss_mean_50": 0.12943075455725192, "loss_std_50": 0.10803949234664725, "lr": 0.00025, "grad_norm_before_clip": 0.042236328125, "grad_clipped": false, "step_seconds": 2.0633848889992805, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1372, "loss": 0.03617017716169357, "loss_ema": 0.11769936972042086, "loss_mean_50": 0.12899270839989185, "loss_std_50": 0.1083718167948152, "lr": 0.00025, "grad_norm_before_clip": 0.0220947265625, "grad_clipped": false, "step_seconds": 2.072429348998412, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1373, "loss": 0.09153870493173599, "loss_ema": 0.11639133648098661, "loss_mean_50": 0.12424925409257412, "loss_std_50": 0.10465302234774422, "lr": 0.00025, "grad_norm_before_clip": 0.01202392578125, "grad_clipped": false, "step_seconds": 2.0734039430026314, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1374, "loss": 0.04982847720384598, "loss_ema": 0.11306319351712957, "loss_mean_50": 0.12065850295126439, "loss_std_50": 0.10406315334410395, "lr": 0.00025, "grad_norm_before_clip": 0.0177001953125, "grad_clipped": false, "step_seconds": 2.0657692390013835, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1375, "loss": 0.057850319892168045, "loss_ema": 0.11030254983588149, "loss_mean_50": 0.12039777413010597, "loss_std_50": 0.10420376545762336, "lr": 0.00025, "grad_norm_before_clip": 0.0322265625, "grad_clipped": false, "step_seconds": 0.6165842049958883, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1376, "loss": 0.06625685095787048, "loss_ema": 0.10810026489198092, "loss_mean_50": 0.11823534652590752, "loss_std_50": 0.1041829911204457, "lr": 0.00025, "grad_norm_before_clip": 0.024169921875, "grad_clipped": false, "step_seconds": 1.8890971970031387, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1377, "loss": 0.11469440162181854, "loss_ema": 0.1084299717284728, "loss_mean_50": 0.1175579722225666, "loss_std_50": 0.10409366981565142, "lr": 0.00025, "grad_norm_before_clip": 0.0079345703125, "grad_clipped": false, "step_seconds": 2.072511770995334, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1378, "loss": 0.07926658540964127, "loss_ema": 0.10697180241253122, "loss_mean_50": 0.11800670102238656, "loss_std_50": 0.10387905450700188, "lr": 0.00025, "grad_norm_before_clip": 0.049560546875, "grad_clipped": false, "step_seconds": 2.061814944005164, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1379, "loss": 0.3544919490814209, "loss_ema": 0.1193478097459757, "loss_mean_50": 0.12251286849379539, "loss_std_50": 0.1090254989464046, "lr": 0.00025, "grad_norm_before_clip": 0.07666015625, "grad_clipped": false, "step_seconds": 0.42105356299725827, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1380, "loss": 0.26311057806015015, "loss_ema": 0.12653594816168443, "loss_mean_50": 0.12650825336575508, "loss_std_50": 0.11043516301500335, "lr": 0.00025, "grad_norm_before_clip": 0.032958984375, "grad_clipped": false, "step_seconds": 0.602669865002099, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1381, "loss": 0.04450047016143799, "loss_ema": 0.1224341742616721, "loss_mean_50": 0.12620306067168713, "loss_std_50": 0.11064009811603367, "lr": 0.00025, "grad_norm_before_clip": 0.03173828125, "grad_clipped": false, "step_seconds": 2.0614507499994943, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1382, "loss": 0.06519895046949387, "loss_ema": 0.11957241307206318, "loss_mean_50": 0.12619509659707545, "loss_std_50": 0.11064447459744835, "lr": 0.00025, "grad_norm_before_clip": 0.0130615234375, "grad_clipped": false, "step_seconds": 1.8894450070001767, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1383, "loss": 0.41375863552093506, "loss_ema": 0.13428172419450676, "loss_mean_50": 0.12675022803246974, "loss_std_50": 0.11200783081002154, "lr": 0.00025, "grad_norm_before_clip": 0.0625, "grad_clipped": false, "step_seconds": 2.0704207949966076, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1384, "loss": 0.061501577496528625, "loss_ema": 0.13064271685960785, "loss_mean_50": 0.126670113876462, "loss_std_50": 0.11205302990872419, "lr": 0.00025, "grad_norm_before_clip": 0.019775390625, "grad_clipped": false, "step_seconds": 2.0644643920022645, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1385, "loss": 0.058484748005867004, "loss_ema": 0.1270348184169208, "loss_mean_50": 0.12180064119398594, "loss_std_50": 0.10959304307370075, "lr": 0.00025, "grad_norm_before_clip": 0.11181640625, "grad_clipped": false, "step_seconds": 1.1014452329982305, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1386, "loss": 0.11387339234352112, "loss_ema": 0.1263767471132508, "loss_mean_50": 0.12325476802885532, "loss_std_50": 0.10899422880202182, "lr": 0.00025, "grad_norm_before_clip": 0.04248046875, "grad_clipped": false, "step_seconds": 0.3536051659975783, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1387, "loss": 0.25510263442993164, "loss_ema": 0.13281304147908485, "loss_mean_50": 0.12693918578326702, "loss_std_50": 0.11026778998513621, "lr": 0.00025, "grad_norm_before_clip": 0.0194091796875, "grad_clipped": false, "step_seconds": 2.061743350001052, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1388, "loss": 0.07825328409671783, "loss_ema": 0.13008505360996647, "loss_mean_50": 0.12739167742431165, "loss_std_50": 0.11002037719513763, "lr": 0.00025, "grad_norm_before_clip": 0.045166015625, "grad_clipped": false, "step_seconds": 2.062841502003721, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1389, "loss": 0.9760569930076599, "loss_ema": 0.17238365057985114, "loss_mean_50": 0.14320952616631985, "loss_std_50": 0.16183998673706973, "lr": 0.00025, "grad_norm_before_clip": 0.10302734375, "grad_clipped": false, "step_seconds": 0.5749584599980153, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1390, "loss": 0.1163065955042839, "loss_ema": 0.16957979782607277, "loss_mean_50": 0.1334259206801653, "loss_std_50": 0.14777324140502562, "lr": 0.00025, "grad_norm_before_clip": 0.08154296875, "grad_clipped": false, "step_seconds": 2.100919791999331, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1391, "loss": 0.40213003754615784, "loss_ema": 0.18120730981207703, "loss_mean_50": 0.14034549064934254, "loss_std_50": 0.15203177915880972, "lr": 0.00025, "grad_norm_before_clip": 0.050537109375, "grad_clipped": false, "step_seconds": 2.062085843004752, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1392, "loss": 0.10197337716817856, "loss_ema": 0.1772456131798821, "loss_mean_50": 0.13976800374686718, "loss_std_50": 0.15212157181956598, "lr": 0.00025, "grad_norm_before_clip": 0.0341796875, "grad_clipped": false, "step_seconds": 0.6701195039931918, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1393, "loss": 0.11297892779111862, "loss_ema": 0.17403227891044393, "loss_mean_50": 0.14116222344338894, "loss_std_50": 0.15154912280012212, "lr": 0.00025, "grad_norm_before_clip": 0.01324462890625, "grad_clipped": false, "step_seconds": 1.2699938250007108, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1394, "loss": 0.0669977217912674, "loss_ema": 0.1686805510544851, "loss_mean_50": 0.14147578455507756, "loss_std_50": 0.15137903443081482, "lr": 0.00025, "grad_norm_before_clip": 0.0140380859375, "grad_clipped": false, "step_seconds": 2.066143006006314, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1395, "loss": 0.10968806594610214, "loss_ema": 0.16573092679906593, "loss_mean_50": 0.14250190488994122, "loss_std_50": 0.15098568457392236, "lr": 0.00025, "grad_norm_before_clip": 0.0186767578125, "grad_clipped": false, "step_seconds": 1.5726203409940354, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1396, "loss": 0.07836808264255524, "loss_ema": 0.1613627845912404, "loss_mean_50": 0.142303803935647, "loss_std_50": 0.15106318362077092, "lr": 0.00025, "grad_norm_before_clip": 0.020751953125, "grad_clipped": false, "step_seconds": 1.889805676000833, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1397, "loss": 0.05869624763727188, "loss_ema": 0.15622945774354197, "loss_mean_50": 0.14245681263506413, "loss_std_50": 0.15097452130732017, "lr": 0.00025, "grad_norm_before_clip": 0.00897216796875, "grad_clipped": false, "step_seconds": 2.0655362920006155, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1398, "loss": 0.0596671998500824, "loss_ema": 0.151401344848869, "loss_mean_50": 0.13731890715658665, "loss_std_50": 0.1493242605411399, "lr": 0.00025, "grad_norm_before_clip": 0.076171875, "grad_clipped": false, "step_seconds": 1.8929874559980817, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1399, "loss": 0.11517047882080078, "loss_ema": 0.14958980154746557, "loss_mean_50": 0.13629512138664723, "loss_std_50": 0.1492971207868135, "lr": 0.00025, "grad_norm_before_clip": 0.0184326171875, "grad_clipped": false, "step_seconds": 2.0581670300016413, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1400, "loss": 0.12168850749731064, "loss_ema": 0.14819473684495782, "loss_mean_50": 0.13779765993356705, "loss_std_50": 0.1487635625685302, "lr": 0.00025, "grad_norm_before_clip": 0.018798828125, "grad_clipped": false, "step_seconds": 0.9501642210016144, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1401, "loss": 0.0833442211151123, "loss_ema": 0.14495221105846554, "loss_mean_50": 0.1371316809952259, "loss_std_50": 0.1489312167061213, "lr": 0.00025, "grad_norm_before_clip": 0.022705078125, "grad_clipped": false, "step_seconds": 1.8956882019992918, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1402, "loss": 0.08838142454624176, "loss_ema": 0.14212367173285434, "loss_mean_50": 0.13427273347973823, "loss_std_50": 0.14846684321073886, "lr": 0.00025, "grad_norm_before_clip": 0.033935546875, "grad_clipped": false, "step_seconds": 1.8900098529993556, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1403, "loss": 0.05949658527970314, "loss_ema": 0.13799231741019677, "loss_mean_50": 0.13396946988999844, "loss_std_50": 0.14860372434826213, "lr": 0.00025, "grad_norm_before_clip": 0.021484375, "grad_clipped": false, "step_seconds": 0.5244509410040337, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1404, "loss": 0.9686012864112854, "loss_ema": 0.17952276586025118, "loss_mean_50": 0.15159102268517016, "loss_std_50": 0.18884281522030405, "lr": 0.00025, "grad_norm_before_clip": 0.181640625, "grad_clipped": false, "step_seconds": 2.06150247399637, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1405, "loss": 0.08909714967012405, "loss_ema": 0.17500148505074484, "loss_mean_50": 0.15206192068755628, "loss_std_50": 0.1886569463047687, "lr": 0.00025, "grad_norm_before_clip": 0.032958984375, "grad_clipped": false, "step_seconds": 0.36164264300168725, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1406, "loss": 0.057614516466856, "loss_ema": 0.1691321366215504, "loss_mean_50": 0.15208189971745015, "loss_std_50": 0.18864688997376108, "lr": 0.00025, "grad_norm_before_clip": 0.06591796875, "grad_clipped": false, "step_seconds": 0.6151536039978964, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1407, "loss": 0.059033673256635666, "loss_ema": 0.16362721345330467, "loss_mean_50": 0.15236951351165773, "loss_std_50": 0.18849378340792416, "lr": 0.00025, "grad_norm_before_clip": 0.051513671875, "grad_clipped": false, "step_seconds": 1.5910671570018167, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1408, "loss": 0.31009504199028015, "loss_ema": 0.17095060488015346, "loss_mean_50": 0.15718238085508346, "loss_std_50": 0.18938527445019573, "lr": 0.00025, "grad_norm_before_clip": 0.027099609375, "grad_clipped": false, "step_seconds": 1.2987174939989927, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1409, "loss": 0.0742761418223381, "loss_ema": 0.1661168817272627, "loss_mean_50": 0.15309954479336738, "loss_std_50": 0.1889275529875273, "lr": 0.00025, "grad_norm_before_clip": 0.0179443359375, "grad_clipped": false, "step_seconds": 1.8925268050006707, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1410, "loss": 0.07256371527910233, "loss_ema": 0.16143922340485467, "loss_mean_50": 0.15271756306290626, "loss_std_50": 0.18907063573367544, "lr": 0.00025, "grad_norm_before_clip": 0.0361328125, "grad_clipped": false, "step_seconds": 1.830039423002745, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1411, "loss": 0.09106283634901047, "loss_ema": 0.15792040405206245, "loss_mean_50": 0.15192407846450806, "loss_std_50": 0.18924438945624095, "lr": 0.00025, "grad_norm_before_clip": 0.01953125, "grad_clipped": false, "step_seconds": 1.2931757609985652, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1412, "loss": 0.2675926685333252, "loss_ema": 0.16340401727612558, "loss_mean_50": 0.15454783350229262, "loss_std_50": 0.18991925236022042, "lr": 0.00025, "grad_norm_before_clip": 0.052978515625, "grad_clipped": false, "step_seconds": 1.8292427459964529, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1413, "loss": 0.13053394854068756, "loss_ema": 0.16176051383935366, "loss_mean_50": 0.15506739065051078, "loss_std_50": 0.1898172866857627, "lr": 0.00025, "grad_norm_before_clip": 0.0185546875, "grad_clipped": false, "step_seconds": 0.6000427000035415, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1414, "loss": 0.13203786313533783, "loss_ema": 0.16027438130415286, "loss_mean_50": 0.1549577935039997, "loss_std_50": 0.1898289695422873, "lr": 0.00025, "grad_norm_before_clip": 0.0439453125, "grad_clipped": false, "step_seconds": 0.4778499890016974, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1415, "loss": 0.09227801114320755, "loss_ema": 0.1568745627961056, "loss_mean_50": 0.15506979957222938, "loss_std_50": 0.18979029699171102, "lr": 0.00025, "grad_norm_before_clip": 0.03271484375, "grad_clipped": false, "step_seconds": 0.7306536939940997, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1416, "loss": 0.5463553667068481, "loss_ema": 0.1763486029916427, "loss_mean_50": 0.16408122688531876, "loss_std_50": 0.1973092567337307, "lr": 0.00025, "grad_norm_before_clip": 0.1201171875, "grad_clipped": false, "step_seconds": 2.068547041999409, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1417, "loss": 0.1247062012553215, "loss_ema": 0.17376648290482663, "loss_mean_50": 0.1652816864848137, "loss_std_50": 0.1968829862432772, "lr": 0.00025, "grad_norm_before_clip": 0.03173828125, "grad_clipped": false, "step_seconds": 1.8932161040065694, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1418, "loss": 0.12426119297742844, "loss_ema": 0.1712912184084567, "loss_mean_50": 0.16601477339863777, "loss_std_50": 0.19666051678084373, "lr": 0.00025, "grad_norm_before_clip": 0.07275390625, "grad_clipped": false, "step_seconds": 1.8334237919989391, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1419, "loss": 0.09649215638637543, "loss_ema": 0.16755126530735262, "loss_mean_50": 0.16589818000793458, "loss_std_50": 0.19669996777032786, "lr": 0.00025, "grad_norm_before_clip": 0.07763671875, "grad_clipped": false, "step_seconds": 1.8887613409970072, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1420, "loss": 0.05626709759235382, "loss_ema": 0.16198705692160267, "loss_mean_50": 0.16238376259803772, "loss_std_50": 0.19705722131904232, "lr": 0.00025, "grad_norm_before_clip": 0.0634765625, "grad_clipped": false, "step_seconds": 1.9036974510017899, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1421, "loss": 0.07166178524494171, "loss_ema": 0.15747079333776962, "loss_mean_50": 0.1623871371150017, "loss_std_50": 0.19705566626596474, "lr": 0.00025, "grad_norm_before_clip": 0.12109375, "grad_clipped": false, "step_seconds": 1.8942652420009836, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1422, "loss": 0.11703769862651825, "loss_ema": 0.15544913860220702, "loss_mean_50": 0.16400448754429817, "loss_std_50": 0.19634367012190496, "lr": 0.00025, "grad_norm_before_clip": 0.0206298828125, "grad_clipped": false, "step_seconds": 2.0861436519990093, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1423, "loss": 0.09172323346138, "loss_ema": 0.15226284334516565, "loss_mean_50": 0.16400817811489105, "loss_std_50": 0.19634230971485003, "lr": 0.00025, "grad_norm_before_clip": 0.011962890625, "grad_clipped": false, "step_seconds": 2.0654036569976597, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1424, "loss": 0.0656476765871048, "loss_ema": 0.14793208500726263, "loss_mean_50": 0.16432456210255622, "loss_std_50": 0.1961707372890964, "lr": 0.00025, "grad_norm_before_clip": 0.0159912109375, "grad_clipped": false, "step_seconds": 2.063449758999923, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1425, "loss": 0.06070500612258911, "loss_ema": 0.14357073106302892, "loss_mean_50": 0.16438165582716466, "loss_std_50": 0.19614015364376178, "lr": 0.00025, "grad_norm_before_clip": 0.047119140625, "grad_clipped": false, "step_seconds": 2.057255277999502, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1426, "loss": 0.07574998587369919, "loss_ema": 0.14017969380356243, "loss_mean_50": 0.16457151852548121, "loss_std_50": 0.19604965119917026, "lr": 0.00025, "grad_norm_before_clip": 0.038330078125, "grad_clipped": false, "step_seconds": 0.6027532029984286, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1427, "loss": 0.5358275771141052, "loss_ema": 0.15996208796908956, "loss_mean_50": 0.17299418203532696, "loss_std_50": 0.2026607785430595, "lr": 0.00025, "grad_norm_before_clip": 0.17578125, "grad_clipped": false, "step_seconds": 1.829994619001809, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1428, "loss": 0.05377960205078125, "loss_ema": 0.15465296367317413, "loss_mean_50": 0.17248444236814975, "loss_std_50": 0.20292776152952557, "lr": 0.00025, "grad_norm_before_clip": 0.054931640625, "grad_clipped": false, "step_seconds": 1.2946426650014473, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1429, "loss": 0.8118005394935608, "loss_ema": 0.18751034246419346, "loss_mean_50": 0.18163061417639256, "loss_std_50": 0.22047219945270158, "lr": 0.00025, "grad_norm_before_clip": 0.12255859375, "grad_clipped": false, "step_seconds": 2.0731638299985207, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1430, "loss": 0.0782940536737442, "loss_ema": 0.18204952802467098, "loss_mean_50": 0.17793428368866443, "loss_std_50": 0.22062438006004165, "lr": 0.00025, "grad_norm_before_clip": 0.0296630859375, "grad_clipped": false, "step_seconds": 2.0615618520023418, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1431, "loss": 0.05046320706605911, "loss_ema": 0.1754702119767404, "loss_mean_50": 0.17805353842675686, "loss_std_50": 0.22055382270739107, "lr": 0.00025, "grad_norm_before_clip": 0.0322265625, "grad_clipped": false, "step_seconds": 1.5697840659995563, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1432, "loss": 0.08944259583950043, "loss_ema": 0.1711688311698784, "loss_mean_50": 0.178538411334157, "loss_std_50": 0.2203317236159399, "lr": 0.00025, "grad_norm_before_clip": 0.0311279296875, "grad_clipped": false, "step_seconds": 0.4794004689974827, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1433, "loss": 0.4539380669593811, "loss_ema": 0.18530729295935355, "loss_mean_50": 0.1793419999629259, "loss_std_50": 0.22125946534764773, "lr": 0.00025, "grad_norm_before_clip": 0.1123046875, "grad_clipped": false, "step_seconds": 1.5857855510039371, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1434, "loss": 0.104700468480587, "loss_ema": 0.18127695173541522, "loss_mean_50": 0.18020597778260708, "loss_std_50": 0.22088165249597674, "lr": 0.00025, "grad_norm_before_clip": 0.036376953125, "grad_clipped": false, "step_seconds": 0.5714191939987359, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1435, "loss": 0.052416179329156876, "loss_ema": 0.1748339131151023, "loss_mean_50": 0.18008460640907287, "loss_std_50": 0.2209501599333011, "lr": 0.00025, "grad_norm_before_clip": 0.017333984375, "grad_clipped": false, "step_seconds": 0.5282543919965974, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1436, "loss": 0.08273175358772278, "loss_ema": 0.1702288051387333, "loss_mean_50": 0.1794617736339569, "loss_std_50": 0.22117969688824063, "lr": 0.00025, "grad_norm_before_clip": 0.01470947265625, "grad_clipped": false, "step_seconds": 2.064816228004929, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1437, "loss": 0.05014774575829506, "loss_ema": 0.16422475216971139, "loss_mean_50": 0.17536267586052418, "loss_std_50": 0.2216385960808434, "lr": 0.00025, "grad_norm_before_clip": 0.050537109375, "grad_clipped": false, "step_seconds": 1.293744404996687, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1438, "loss": 0.03845827654004097, "loss_ema": 0.15793642838822786, "loss_mean_50": 0.17456677570939064, "loss_std_50": 0.22205694185800184, "lr": 0.00025, "grad_norm_before_clip": 0.0322265625, "grad_clipped": false, "step_seconds": 1.832578729998204, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1439, "loss": 0.052542686462402344, "loss_ema": 0.15266674129193658, "loss_mean_50": 0.1560964895784855, "loss_std_50": 0.19083553169209633, "lr": 0.00025, "grad_norm_before_clip": 0.039306640625, "grad_clipped": false, "step_seconds": 2.0785435380021227, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1440, "loss": 0.056343138217926025, "loss_ema": 0.14785056113823603, "loss_mean_50": 0.15489722043275833, "loss_std_50": 0.19126973576200118, "lr": 0.00025, "grad_norm_before_clip": 0.027587890625, "grad_clipped": false, "step_seconds": 2.0687830640017637, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1441, "loss": 0.06268280744552612, "loss_ema": 0.1435921734536005, "loss_mean_50": 0.1481082758307457, "loss_std_50": 0.18837624768465797, "lr": 0.00025, "grad_norm_before_clip": 0.058349609375, "grad_clipped": false, "step_seconds": 1.8972800779956742, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1442, "loss": 0.6658475399017334, "loss_ema": 0.16970494177600715, "loss_mean_50": 0.1593857590854168, "loss_std_50": 0.20168524833739196, "lr": 0.00025, "grad_norm_before_clip": 0.05322265625, "grad_clipped": false, "step_seconds": 2.066535381003632, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1443, "loss": 0.0335116907954216, "loss_ema": 0.16289527922697786, "loss_mean_50": 0.15779641434550284, "loss_std_50": 0.20235668291844258, "lr": 0.00025, "grad_norm_before_clip": 0.0191650390625, "grad_clipped": false, "step_seconds": 2.066358638003294, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1444, "loss": 0.22409191727638245, "loss_ema": 0.16595511112944808, "loss_mean_50": 0.16093829825520514, "loss_std_50": 0.20214195385925968, "lr": 0.00025, "grad_norm_before_clip": 0.0166015625, "grad_clipped": false, "step_seconds": 2.065985263994662, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1445, "loss": 0.06413187086582184, "loss_ema": 0.16086394911626675, "loss_mean_50": 0.16002717435359956, "loss_std_50": 0.20247330027615532, "lr": 0.00025, "grad_norm_before_clip": 0.0390625, "grad_clipped": false, "step_seconds": 1.890003054002591, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1446, "loss": 0.11921198666095734, "loss_ema": 0.1587813509935013, "loss_mean_50": 0.1608440524339676, "loss_std_50": 0.20222443838632914, "lr": 0.00025, "grad_norm_before_clip": 0.0177001953125, "grad_clipped": false, "step_seconds": 0.5755600600023172, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1447, "loss": 0.2050141543149948, "loss_ema": 0.16109299115957595, "loss_mean_50": 0.16377041056752206, "loss_std_50": 0.2017832906517493, "lr": 0.00025, "grad_norm_before_clip": 0.036376953125, "grad_clipped": false, "step_seconds": 2.0664130649965955, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1448, "loss": 0.06549990922212601, "loss_ema": 0.15631333706270345, "loss_mean_50": 0.16388706475496292, "loss_std_50": 0.20172475068049733, "lr": 0.00025, "grad_norm_before_clip": 0.0230712890625, "grad_clipped": false, "step_seconds": 1.8899048780003795, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1449, "loss": 0.05037263035774231, "loss_ema": 0.1510163017274554, "loss_mean_50": 0.16259110778570174, "loss_std_50": 0.20224104418755218, "lr": 0.00025, "grad_norm_before_clip": 0.0458984375, "grad_clipped": false, "step_seconds": 0.42159808699943824, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1450, "loss": 0.05998051539063454, "loss_ema": 0.14646451241061434, "loss_mean_50": 0.16135694794356822, "loss_std_50": 0.20267470247364813, "lr": 0.00025, "grad_norm_before_clip": 0.0308837890625, "grad_clipped": false, "step_seconds": 0.6693591500006733, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1451, "loss": 0.14521554112434387, "loss_ema": 0.1464020638463008, "loss_mean_50": 0.16259437434375287, "loss_std_50": 0.2023832873238889, "lr": 0.00025, "grad_norm_before_clip": 0.038330078125, "grad_clipped": false, "step_seconds": 0.7308669420017395, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1452, "loss": 0.0726357102394104, "loss_ema": 0.14271374616595628, "loss_mean_50": 0.16227946005761623, "loss_std_50": 0.20251073009500498, "lr": 0.00025, "grad_norm_before_clip": 0.055419921875, "grad_clipped": false, "step_seconds": 1.8331641600016155, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1453, "loss": 0.16572462022304535, "loss_ema": 0.14386428986881072, "loss_mean_50": 0.16440402075648308, "loss_std_50": 0.20197780333756443, "lr": 0.00025, "grad_norm_before_clip": 0.0162353515625, "grad_clipped": false, "step_seconds": 2.0662804759995197, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1454, "loss": 0.34904876351356506, "loss_ema": 0.15412351355104845, "loss_mean_50": 0.15201297029852867, "loss_std_50": 0.1684894764916778, "lr": 0.00025, "grad_norm_before_clip": 0.0240478515625, "grad_clipped": false, "step_seconds": 1.889173135998135, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1455, "loss": 1.0193675756454468, "loss_ema": 0.19738571665576835, "loss_mean_50": 0.17061837881803513, "loss_std_50": 0.2073872095402346, "lr": 0.00025, "grad_norm_before_clip": 0.11669921875, "grad_clipped": false, "step_seconds": 0.3622237390009104, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1456, "loss": 0.05107903480529785, "loss_ema": 0.19007038256324482, "loss_mean_50": 0.17048766918480396, "loss_std_50": 0.2074604377540238, "lr": 0.00025, "grad_norm_before_clip": 0.01300048828125, "grad_clipped": false, "step_seconds": 0.6010001769973314, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1457, "loss": 0.05994993820786476, "loss_ema": 0.18356436034547582, "loss_mean_50": 0.17050599448382855, "loss_std_50": 0.20745063227789423, "lr": 0.00025, "grad_norm_before_clip": 0.0244140625, "grad_clipped": false, "step_seconds": 2.06747770900256, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1458, "loss": 0.12344024330377579, "loss_ema": 0.1805581544933908, "loss_mean_50": 0.16677289851009847, "loss_std_50": 0.2065827448795196, "lr": 0.00025, "grad_norm_before_clip": 0.0223388671875, "grad_clipped": false, "step_seconds": 1.890909980997094, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1459, "loss": 0.07796362042427063, "loss_ema": 0.17542842778993478, "loss_mean_50": 0.16684664808213712, "loss_std_50": 0.2065503662559927, "lr": 0.00025, "grad_norm_before_clip": 0.056884765625, "grad_clipped": false, "step_seconds": 0.35265868499845965, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1460, "loss": 0.0826701745390892, "loss_ema": 0.1707905151273925, "loss_mean_50": 0.16704877726733686, "loss_std_50": 0.2064629290935331, "lr": 0.00025, "grad_norm_before_clip": 0.01043701171875, "grad_clipped": false, "step_seconds": 1.1045956230009324, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1461, "loss": 0.05069580674171448, "loss_ema": 0.1647857797081086, "loss_mean_50": 0.16624143667519092, "loss_std_50": 0.206837067039995, "lr": 0.00025, "grad_norm_before_clip": 0.0247802734375, "grad_clipped": false, "step_seconds": 0.8375654490009765, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1462, "loss": 0.047525808215141296, "loss_ema": 0.15892278113346023, "loss_mean_50": 0.16184009946882724, "loss_std_50": 0.2069749431630151, "lr": 0.00025, "grad_norm_before_clip": 0.03271484375, "grad_clipped": false, "step_seconds": 1.2662927319979644, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1463, "loss": 0.08168281614780426, "loss_ema": 0.15506078288417743, "loss_mean_50": 0.1608630768209696, "loss_std_50": 0.20723555395671936, "lr": 0.00025, "grad_norm_before_clip": 0.04638671875, "grad_clipped": false, "step_seconds": 2.0801032520030276, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1464, "loss": 0.23253408074378967, "loss_ema": 0.15893444777715804, "loss_mean_50": 0.16287300117313863, "loss_std_50": 0.20743348771361036, "lr": 0.00025, "grad_norm_before_clip": 0.021484375, "grad_clipped": false, "step_seconds": 1.899944949997007, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1465, "loss": 0.03966447710990906, "loss_ema": 0.1529709492437956, "loss_mean_50": 0.16182073049247264, "loss_std_50": 0.20792180815666603, "lr": 0.00025, "grad_norm_before_clip": 0.0269775390625, "grad_clipped": false, "step_seconds": 1.8336329140001908, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1466, "loss": 0.06587105244398117, "loss_ema": 0.14861595440380485, "loss_mean_50": 0.15221104420721532, "loss_std_50": 0.20091272065754845, "lr": 0.00025, "grad_norm_before_clip": 0.035400390625, "grad_clipped": false, "step_seconds": 1.8340079300032812, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1467, "loss": 0.34225210547447205, "loss_ema": 0.1582977619573382, "loss_mean_50": 0.15656196229159833, "loss_std_50": 0.2026182928698113, "lr": 0.00025, "grad_norm_before_clip": 0.042236328125, "grad_clipped": false, "step_seconds": 0.6721057269969606, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1468, "loss": 0.08752405643463135, "loss_ema": 0.15475907668120284, "loss_mean_50": 0.15582721956074239, "loss_std_50": 0.20280061788252185, "lr": 0.00025, "grad_norm_before_clip": 0.052978515625, "grad_clipped": false, "step_seconds": 1.5953598449996207, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1469, "loss": 0.09195167571306229, "loss_ema": 0.1516187066327958, "loss_mean_50": 0.1557364099472761, "loss_std_50": 0.20282818116510673, "lr": 0.00025, "grad_norm_before_clip": 0.01080322265625, "grad_clipped": false, "step_seconds": 2.073220142003265, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1470, "loss": 0.1548645943403244, "loss_ema": 0.1517810010181722, "loss_mean_50": 0.15770835988223553, "loss_std_50": 0.20233021229525067, "lr": 0.00025, "grad_norm_before_clip": 0.01312255859375, "grad_clipped": false, "step_seconds": 2.0696078879991546, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1471, "loss": 0.07543990015983582, "loss_ema": 0.1479639459752554, "loss_mean_50": 0.15778392218053341, "loss_std_50": 0.20229876625166546, "lr": 0.00025, "grad_norm_before_clip": 0.0228271484375, "grad_clipped": false, "step_seconds": 2.0618830989988055, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1472, "loss": 0.20694957673549652, "loss_ema": 0.15091322751326744, "loss_mean_50": 0.15958215974271298, "loss_std_50": 0.20232819206846592, "lr": 0.00025, "grad_norm_before_clip": 0.021240234375, "grad_clipped": false, "step_seconds": 1.894017831000383, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1473, "loss": 0.08873125910758972, "loss_ema": 0.14780412909298354, "loss_mean_50": 0.15952232025563717, "loss_std_50": 0.20234869421329327, "lr": 0.00025, "grad_norm_before_clip": 0.058837890625, "grad_clipped": false, "step_seconds": 1.5795411369981593, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1474, "loss": 0.08883871883153915, "loss_ema": 0.14485585857991132, "loss_mean_50": 0.15998614110052586, "loss_std_50": 0.20215947510759036, "lr": 0.00025, "grad_norm_before_clip": 0.012451171875, "grad_clipped": false, "step_seconds": 2.066705641002045, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1475, "loss": 0.06858848035335541, "loss_ema": 0.14104248966858351, "loss_mean_50": 0.16014381058514118, "loss_std_50": 0.20208504221901488, "lr": 0.00025, "grad_norm_before_clip": 0.01312255859375, "grad_clipped": false, "step_seconds": 2.059406824002508, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1476, "loss": 0.08110307157039642, "loss_ema": 0.13804551876367416, "loss_mean_50": 0.16025087229907511, "loss_std_50": 0.20204171658828743, "lr": 0.00025, "grad_norm_before_clip": 0.0888671875, "grad_clipped": false, "step_seconds": 0.579066572994634, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1477, "loss": 0.055864330381155014, "loss_ema": 0.1339364593445482, "loss_mean_50": 0.15065160736441613, "loss_std_50": 0.19525747907143887, "lr": 0.00025, "grad_norm_before_clip": 0.0269775390625, "grad_clipped": false, "step_seconds": 1.8907176920038182, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1478, "loss": 0.07797791063785553, "loss_ema": 0.13113853190921354, "loss_mean_50": 0.1511355735361576, "loss_std_50": 0.19504664705071553, "lr": 0.00025, "grad_norm_before_clip": 0.0224609375, "grad_clipped": false, "step_seconds": 0.5996652829999221, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1479, "loss": 0.07778408378362656, "loss_ema": 0.12847080950293419, "loss_mean_50": 0.13645524442195892, "loss_std_50": 0.170896832442657, "lr": 0.00025, "grad_norm_before_clip": 0.0498046875, "grad_clipped": false, "step_seconds": 0.5781661340006394, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1480, "loss": 0.04006216302514076, "loss_ema": 0.1240503771790445, "loss_mean_50": 0.13569060660898685, "loss_std_50": 0.1712405346539978, "lr": 0.00025, "grad_norm_before_clip": 0.0306396484375, "grad_clipped": false, "step_seconds": 2.0659643839971977, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1481, "loss": 0.07478374987840652, "loss_ema": 0.1215870458140126, "loss_mean_50": 0.1361770174652338, "loss_std_50": 0.17103216897515197, "lr": 0.00025, "grad_norm_before_clip": 0.025634765625, "grad_clipped": false, "step_seconds": 1.8925785070023267, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1482, "loss": 0.0911007896065712, "loss_ema": 0.12006273300364052, "loss_mean_50": 0.13621018134057522, "loss_std_50": 0.17102326428754597, "lr": 0.00025, "grad_norm_before_clip": 0.01080322265625, "grad_clipped": false, "step_seconds": 1.299931478999497, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1483, "loss": 0.06497256457805634, "loss_ema": 0.11730822458236131, "loss_mean_50": 0.12843087129294872, "loss_std_50": 0.16513907786346277, "lr": 0.00025, "grad_norm_before_clip": 0.01141357421875, "grad_clipped": false, "step_seconds": 2.070374394999817, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1484, "loss": 0.15596772730350494, "loss_ema": 0.11924119971841848, "loss_mean_50": 0.12945621646940708, "loss_std_50": 0.16514771155934954, "lr": 0.00025, "grad_norm_before_clip": 0.049560546875, "grad_clipped": false, "step_seconds": 2.062480457003403, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1485, "loss": 0.16699618101119995, "loss_ema": 0.12162894878305755, "loss_mean_50": 0.13174781650304795, "loss_std_50": 0.16485750463621757, "lr": 0.00025, "grad_norm_before_clip": 0.0152587890625, "grad_clipped": false, "step_seconds": 1.2924397359965951, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1486, "loss": 0.051423147320747375, "loss_ema": 0.11811865870994204, "loss_mean_50": 0.13112164437770843, "loss_std_50": 0.16510176953764147, "lr": 0.00025, "grad_norm_before_clip": 0.0242919921875, "grad_clipped": false, "step_seconds": 1.1066952429973753, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1487, "loss": 0.15453898906707764, "loss_ema": 0.1199396752277988, "loss_mean_50": 0.13320946924388408, "loss_std_50": 0.16472421584965016, "lr": 0.00025, "grad_norm_before_clip": 0.01507568359375, "grad_clipped": false, "step_seconds": 2.066914278002514, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1488, "loss": 0.39950892329216003, "loss_ema": 0.13391813763101684, "loss_mean_50": 0.14043048217892648, "loss_std_50": 0.16828748163076857, "lr": 0.00025, "grad_norm_before_clip": 0.0242919921875, "grad_clipped": false, "step_seconds": 1.89337931600312, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1489, "loss": 0.05186377465724945, "loss_ema": 0.12981541948232844, "loss_mean_50": 0.1404169039428234, "loss_std_50": 0.16829459952762651, "lr": 0.00025, "grad_norm_before_clip": 0.0177001953125, "grad_clipped": false, "step_seconds": 0.48315656499471515, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1490, "loss": 0.048664044588804245, "loss_ema": 0.12575785073765222, "loss_mean_50": 0.14026332207024098, "loss_std_50": 0.1683747380847319, "lr": 0.00025, "grad_norm_before_clip": 0.026123046875, "grad_clipped": false, "step_seconds": 1.8324973050039262, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1491, "loss": 0.04402511939406395, "loss_ema": 0.1216712141704728, "loss_mean_50": 0.13989016830921172, "loss_std_50": 0.1685668244034715, "lr": 0.00025, "grad_norm_before_clip": 0.02490234375, "grad_clipped": false, "step_seconds": 1.8908056040017982, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1492, "loss": 0.14362822473049164, "loss_ema": 0.12276906469847373, "loss_mean_50": 0.1294457820057869, "loss_std_50": 0.15090840072998432, "lr": 0.00025, "grad_norm_before_clip": 0.035400390625, "grad_clipped": false, "step_seconds": 0.5252867579984013, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1493, "loss": 0.053573280572891235, "loss_ema": 0.1193092754921946, "loss_mean_50": 0.1298470138013363, "loss_std_50": 0.150679295712033, "lr": 0.00025, "grad_norm_before_clip": 0.0081787109375, "grad_clipped": false, "step_seconds": 2.071861676995468, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1494, "loss": 0.051944393664598465, "loss_ema": 0.11594103140081478, "loss_mean_50": 0.12640406332910062, "loss_std_50": 0.1504530843026589, "lr": 0.00025, "grad_norm_before_clip": 0.0306396484375, "grad_clipped": false, "step_seconds": 2.0737299180036644, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1495, "loss": 0.08100853115320206, "loss_ema": 0.11419440638843414, "loss_mean_50": 0.1267415965348482, "loss_std_50": 0.1503318835618878, "lr": 0.00025, "grad_norm_before_clip": 0.0308837890625, "grad_clipped": false, "step_seconds": 1.8905097259994363, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1496, "loss": 0.15188845992088318, "loss_ema": 0.11607910906505658, "loss_mean_50": 0.12739512600004674, "loss_std_50": 0.15036875176967177, "lr": 0.00025, "grad_norm_before_clip": 0.037841796875, "grad_clipped": false, "step_seconds": 0.42240004800260067, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1497, "loss": 0.5848063230514526, "loss_ema": 0.13951546976437637, "loss_mean_50": 0.1349909693747759, "loss_std_50": 0.16314738877721768, "lr": 0.00025, "grad_norm_before_clip": 0.037353515625, "grad_clipped": false, "step_seconds": 0.8187880179975764, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1498, "loss": 0.06786376982927322, "loss_ema": 0.13593288476762122, "loss_mean_50": 0.13503824658691882, "loss_std_50": 0.1631275859541956, "lr": 0.00025, "grad_norm_before_clip": 0.0283203125, "grad_clipped": false, "step_seconds": 1.2729605639979127, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1499, "loss": 0.056634239852428436, "loss_ema": 0.13196795252186158, "loss_mean_50": 0.13516347877681256, "loss_std_50": 0.16306493200609753, "lr": 0.00025, "grad_norm_before_clip": 0.00677490234375, "grad_clipped": false, "step_seconds": 2.066909591005242, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1500, "loss": 0.05031907185912132, "loss_ema": 0.12788550848872457, "loss_mean_50": 0.13497024990618228, "loss_std_50": 0.16315960474575353, "lr": 0.00025, "grad_norm_before_clip": 0.0257568359375, "grad_clipped": false, "step_seconds": 2.0622815789974993, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1501, "loss": 0.0801989734172821, "loss_ema": 0.12550118173515243, "loss_mean_50": 0.13366991855204105, "loss_std_50": 0.16333176143629116, "lr": 0.00025, "grad_norm_before_clip": 0.04736328125, "grad_clipped": false, "step_seconds": 0.36318434400163824, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1502, "loss": 0.06628190726041794, "loss_ema": 0.1225402180114157, "loss_mean_50": 0.1335428424924612, "loss_std_50": 0.16338166217536904, "lr": 0.00025, "grad_norm_before_clip": 0.11865234375, "grad_clipped": false, "step_seconds": 0.6204100659961114, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1503, "loss": 0.10713892430067062, "loss_ema": 0.12177015332587844, "loss_mean_50": 0.1323711285740137, "loss_std_50": 0.16335674036671793, "lr": 0.00025, "grad_norm_before_clip": 0.08447265625, "grad_clipped": false, "step_seconds": 1.890585553002893, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1504, "loss": 0.34536999464035034, "loss_ema": 0.13295014539160202, "loss_mean_50": 0.13229755319654943, "loss_std_50": 0.1632599326202056, "lr": 0.00025, "grad_norm_before_clip": 0.0245361328125, "grad_clipped": false, "step_seconds": 0.7232550150001771, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1505, "loss": 0.1093900054693222, "loss_ema": 0.13177213839548801, "loss_mean_50": 0.11409800179302693, "loss_std_50": 0.10293304886270675, "lr": 0.00025, "grad_norm_before_clip": 0.050537109375, "grad_clipped": false, "step_seconds": 0.3471857319964329, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1506, "loss": 0.06612268090248108, "loss_ema": 0.12848966552083765, "loss_mean_50": 0.1143988747149706, "loss_std_50": 0.10277026246898577, "lr": 0.00025, "grad_norm_before_clip": 0.03515625, "grad_clipped": false, "step_seconds": 1.9017944730003364, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1507, "loss": 0.2470090240240097, "loss_ema": 0.13441563344599625, "loss_mean_50": 0.1181400564312935, "loss_std_50": 0.10411602041746454, "lr": 0.00025, "grad_norm_before_clip": 0.031494140625, "grad_clipped": false, "step_seconds": 0.6173416950041428, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1508, "loss": 0.05691147595643997, "loss_ema": 0.13054042557151843, "loss_mean_50": 0.11680948108434677, "loss_std_50": 0.10446431074073881, "lr": 0.00025, "grad_norm_before_clip": 0.0294189453125, "grad_clipped": false, "step_seconds": 2.061952043994097, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1509, "loss": 0.05057839676737785, "loss_ema": 0.1265423241313114, "loss_mean_50": 0.11626177661120891, "loss_std_50": 0.1047379747328319, "lr": 0.00025, "grad_norm_before_clip": 0.047119140625, "grad_clipped": false, "step_seconds": 0.42122994500095956, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1510, "loss": 0.0711759403347969, "loss_ema": 0.12377400494148566, "loss_mean_50": 0.11603189192712307, "loss_std_50": 0.10482402988927104, "lr": 0.00025, "grad_norm_before_clip": 0.091796875, "grad_clipped": false, "step_seconds": 0.6017626600005315, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1511, "loss": 0.05476852506399155, "loss_ema": 0.12032373094761095, "loss_mean_50": 0.1161133462935686, "loss_std_50": 0.10477479911340326, "lr": 0.00025, "grad_norm_before_clip": 0.02734375, "grad_clipped": false, "step_seconds": 1.8270648550023907, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1512, "loss": 0.0785185694694519, "loss_ema": 0.11823347287370299, "loss_mean_50": 0.11673320151865482, "loss_std_50": 0.10445839672352746, "lr": 0.00025, "grad_norm_before_clip": 0.044921875, "grad_clipped": false, "step_seconds": 0.3473679300004733, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1513, "loss": 0.1144423708319664, "loss_ema": 0.11804391777161616, "loss_mean_50": 0.11738839261233806, "loss_std_50": 0.10433916684779693, "lr": 0.00025, "grad_norm_before_clip": 0.05224609375, "grad_clipped": false, "step_seconds": 0.7274366929996177, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1514, "loss": 0.10896098613739014, "loss_ema": 0.11758977118990485, "loss_mean_50": 0.11491693072021007, "loss_std_50": 0.10303787385374727, "lr": 0.00025, "grad_norm_before_clip": 0.04638671875, "grad_clipped": false, "step_seconds": 1.8939733789957245, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1515, "loss": 0.2546762228012085, "loss_ema": 0.12444409377047003, "loss_mean_50": 0.11921716563403606, "loss_std_50": 0.10428665296651761, "lr": 0.00025, "grad_norm_before_clip": 0.055419921875, "grad_clipped": false, "step_seconds": 1.2909071480025887, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1516, "loss": 0.060544226318597794, "loss_ema": 0.12124910039787641, "loss_mean_50": 0.1191106291115284, "loss_std_50": 0.10434380076093062, "lr": 0.00025, "grad_norm_before_clip": 0.07275390625, "grad_clipped": false, "step_seconds": 0.4774541170045268, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1517, "loss": 0.04855065420269966, "loss_ema": 0.11761417808811757, "loss_mean_50": 0.11323660008609295, "loss_std_50": 0.09978404868494331, "lr": 0.00025, "grad_norm_before_clip": 0.0390625, "grad_clipped": false, "step_seconds": 0.5259484739945037, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1518, "loss": 0.043826401233673096, "loss_ema": 0.11392478924539534, "loss_mean_50": 0.11236264698207378, "loss_std_50": 0.10019593501737081, "lr": 0.00025, "grad_norm_before_clip": 0.015869140625, "grad_clipped": false, "step_seconds": 2.069966770002793, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1519, "loss": 0.05593908950686455, "loss_ema": 0.1110255042584688, "loss_mean_50": 0.11164239525794983, "loss_std_50": 0.10046913373399498, "lr": 0.00025, "grad_norm_before_clip": 0.01904296875, "grad_clipped": false, "step_seconds": 2.065531474996533, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1520, "loss": 0.04311923310160637, "loss_ema": 0.10763019070062567, "loss_mean_50": 0.10940748803317547, "loss_std_50": 0.10072535593446062, "lr": 0.00025, "grad_norm_before_clip": 0.043212890625, "grad_clipped": false, "step_seconds": 1.2706118499991135, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1521, "loss": 0.12936364114284515, "loss_ema": 0.10871686322273663, "loss_mean_50": 0.11048596285283566, "loss_std_50": 0.10064453905010053, "lr": 0.00025, "grad_norm_before_clip": 0.01263427734375, "grad_clipped": false, "step_seconds": 1.5885741500023869, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1522, "loss": 0.09835533052682877, "loss_ema": 0.10819878658794123, "loss_mean_50": 0.1083140779286623, "loss_std_50": 0.09970679327523212, "lr": 0.00025, "grad_norm_before_clip": 0.059326171875, "grad_clipped": false, "step_seconds": 0.36898607900366187, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1523, "loss": 0.18772418797016144, "loss_ema": 0.11217505665705224, "loss_mean_50": 0.11029393650591374, "loss_std_50": 0.10027948125574064, "lr": 0.00025, "grad_norm_before_clip": 0.02001953125, "grad_clipped": false, "step_seconds": 1.8908613120001974, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1524, "loss": 0.11046836525201797, "loss_ema": 0.11208972208680053, "loss_mean_50": 0.11072652943432332, "loss_std_50": 0.10023263593000725, "lr": 0.00025, "grad_norm_before_clip": 0.019775390625, "grad_clipped": false, "step_seconds": 0.6070553779936745, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1525, "loss": 0.19551482796669006, "loss_ema": 0.11626097738079501, "loss_mean_50": 0.11326505638659, "loss_std_50": 0.10073929716919446, "lr": 0.00025, "grad_norm_before_clip": 0.021728515625, "grad_clipped": false, "step_seconds": 2.0663290100055747, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1526, "loss": 0.05892443656921387, "loss_ema": 0.11339415034021594, "loss_mean_50": 0.11282148368656636, "loss_std_50": 0.10092858579857074, "lr": 0.00025, "grad_norm_before_clip": 0.03466796875, "grad_clipped": false, "step_seconds": 2.0700712139951065, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1527, "loss": 0.0645175501704216, "loss_ema": 0.11095032033172621, "loss_mean_50": 0.11299454808235168, "loss_std_50": 0.10083815018941931, "lr": 0.00025, "grad_norm_before_clip": 0.0150146484375, "grad_clipped": false, "step_seconds": 1.2953893049998442, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1528, "loss": 0.14182965457439423, "loss_ema": 0.1124942870438596, "loss_mean_50": 0.11427158296108246, "loss_std_50": 0.10079091072568078, "lr": 0.00025, "grad_norm_before_clip": 0.01287841796875, "grad_clipped": false, "step_seconds": 2.0683964580020984, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1529, "loss": 0.054556649178266525, "loss_ema": 0.10959740515057993, "loss_mean_50": 0.11380703426897526, "loss_std_50": 0.10101129933742087, "lr": 0.00025, "grad_norm_before_clip": 0.033447265625, "grad_clipped": false, "step_seconds": 1.581621682002151, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1530, "loss": 0.058227311819791794, "loss_ema": 0.10702890048404053, "loss_mean_50": 0.11417033724486828, "loss_std_50": 0.10077780804838396, "lr": 0.00025, "grad_norm_before_clip": 0.019287109375, "grad_clipped": false, "step_seconds": 1.8916387209974346, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1531, "loss": 0.11829042434692383, "loss_ema": 0.10759197667718469, "loss_mean_50": 0.11504047073423862, "loss_std_50": 0.10062168205617837, "lr": 0.00025, "grad_norm_before_clip": 0.018310546875, "grad_clipped": false, "step_seconds": 0.6786219589994289, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1532, "loss": 0.11529510468244553, "loss_ema": 0.10797713307744772, "loss_mean_50": 0.11552435703575611, "loss_std_50": 0.10056355146861568, "lr": 0.00025, "grad_norm_before_clip": 0.007232666015625, "grad_clipped": false, "step_seconds": 2.071436960999563, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1533, "loss": 0.07132203876972198, "loss_ema": 0.10614437836206142, "loss_mean_50": 0.11565134651958943, "loss_std_50": 0.10050362671088055, "lr": 0.00025, "grad_norm_before_clip": 0.03173828125, "grad_clipped": false, "step_seconds": 2.0727267270049197, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1534, "loss": 0.126761332154274, "loss_ema": 0.10717522605167204, "loss_mean_50": 0.11506721861660481, "loss_std_50": 0.1003523701815695, "lr": 0.00025, "grad_norm_before_clip": 0.0179443359375, "grad_clipped": false, "step_seconds": 2.0664486059977207, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1535, "loss": 0.05046118423342705, "loss_ema": 0.1043395239607598, "loss_mean_50": 0.11273651868104935, "loss_std_50": 0.10047244642057832, "lr": 0.00025, "grad_norm_before_clip": 0.0228271484375, "grad_clipped": false, "step_seconds": 1.111489946997608, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1536, "loss": 0.09529534727334976, "loss_ema": 0.10388731512638928, "loss_mean_50": 0.11361396268010139, "loss_std_50": 0.10012412237225674, "lr": 0.00025, "grad_norm_before_clip": 0.052490234375, "grad_clipped": false, "step_seconds": 2.0717703530026483, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1537, "loss": 0.042216233909130096, "loss_ema": 0.10080376106552631, "loss_mean_50": 0.11136750757694244, "loss_std_50": 0.10044027517914166, "lr": 0.00025, "grad_norm_before_clip": 0.01397705078125, "grad_clipped": false, "step_seconds": 2.067130345996702, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1538, "loss": 0.5376242995262146, "loss_ema": 0.12264478798856072, "loss_mean_50": 0.11412981510162354, "loss_std_50": 0.10979073767787105, "lr": 0.00025, "grad_norm_before_clip": 0.021240234375, "grad_clipped": false, "step_seconds": 2.0621477070017136, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1539, "loss": 0.04619035869836807, "loss_ema": 0.11882206652405108, "loss_mean_50": 0.1140163467824459, "loss_std_50": 0.10985794192349785, "lr": 0.00025, "grad_norm_before_clip": 0.01904296875, "grad_clipped": false, "step_seconds": 1.8941206239978783, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1540, "loss": 0.07064784318208694, "loss_ema": 0.11641335535695288, "loss_mean_50": 0.11445602275431156, "loss_std_50": 0.10963928202408522, "lr": 0.00025, "grad_norm_before_clip": 0.07958984375, "grad_clipped": false, "step_seconds": 0.5752592510034447, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1541, "loss": 0.07270359992980957, "loss_ema": 0.11422786758559571, "loss_mean_50": 0.11502959236502647, "loss_std_50": 0.10934394466055296, "lr": 0.00025, "grad_norm_before_clip": 0.0238037109375, "grad_clipped": false, "step_seconds": 1.8968839059962193, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1542, "loss": 0.0424613319337368, "loss_ema": 0.11063954080300276, "loss_mean_50": 0.11300625450909138, "loss_std_50": 0.10973135263372896, "lr": 0.00025, "grad_norm_before_clip": 0.0185546875, "grad_clipped": false, "step_seconds": 2.069447674002731, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1543, "loss": 0.1808776706457138, "loss_ema": 0.1141514472951383, "loss_mean_50": 0.11555234231054783, "loss_std_50": 0.10979969126771187, "lr": 0.00025, "grad_norm_before_clip": 0.017822265625, "grad_clipped": false, "step_seconds": 1.8991855089989258, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1544, "loss": 0.12479300796985626, "loss_ema": 0.11468352532887419, "loss_mean_50": 0.11700931459665298, "loss_std_50": 0.10942868823877948, "lr": 0.00025, "grad_norm_before_clip": 0.015869140625, "grad_clipped": false, "step_seconds": 1.8286157910042675, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1545, "loss": 0.07164128124713898, "loss_ema": 0.11253141312478743, "loss_mean_50": 0.11682196959853172, "loss_std_50": 0.10949815867100632, "lr": 0.00025, "grad_norm_before_clip": 0.026611328125, "grad_clipped": false, "step_seconds": 0.5753866300001391, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1546, "loss": 0.05583572015166283, "loss_ema": 0.1096966284761312, "loss_mean_50": 0.11490091480314732, "loss_std_50": 0.10970847541724099, "lr": 0.00025, "grad_norm_before_clip": 0.03125, "grad_clipped": false, "step_seconds": 2.06908398700034, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1547, "loss": 0.04484272375702858, "loss_ema": 0.10645393324017606, "loss_mean_50": 0.10410164281725884, "loss_std_50": 0.08718524216622407, "lr": 0.00025, "grad_norm_before_clip": 0.024658203125, "grad_clipped": false, "step_seconds": 1.8317207059953944, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1548, "loss": 0.09129004180431366, "loss_ema": 0.10569573866838293, "loss_mean_50": 0.10457016825675965, "loss_std_50": 0.08705208790794226, "lr": 0.00025, "grad_norm_before_clip": 0.01373291015625, "grad_clipped": false, "step_seconds": 1.8889148130037938, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1549, "loss": 0.2182362973690033, "loss_ema": 0.11132276660341395, "loss_mean_50": 0.10780220940709113, "loss_std_50": 0.08820466332707241, "lr": 0.00025, "grad_norm_before_clip": 0.020263671875, "grad_clipped": false, "step_seconds": 0.6111410129960859, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1550, "loss": 0.6074557900428772, "loss_ema": 0.1361294177753871, "loss_mean_50": 0.11894494377076625, "loss_std_50": 0.11217348124373616, "lr": 0.00025, "grad_norm_before_clip": 0.0302734375, "grad_clipped": false, "step_seconds": 2.0668238350044703, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1551, "loss": 0.9121841788291931, "loss_ema": 0.17493215582807736, "loss_mean_50": 0.13558464787900448, "loss_std_50": 0.1576723019996791, "lr": 0.00025, "grad_norm_before_clip": 0.0908203125, "grad_clipped": false, "step_seconds": 2.0716465699952096, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1552, "loss": 0.26011231541633606, "loss_ema": 0.17919116380749028, "loss_mean_50": 0.13946125604212284, "loss_std_50": 0.15830228087279247, "lr": 0.00025, "grad_norm_before_clip": 0.025634765625, "grad_clipped": false, "step_seconds": 1.831701556002372, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1553, "loss": 0.046933408826589584, "loss_ema": 0.17257827605844522, "loss_mean_50": 0.13825714573264122, "loss_std_50": 0.1587718351456385, "lr": 0.00025, "grad_norm_before_clip": 0.0322265625, "grad_clipped": false, "step_seconds": 1.5868707679983345, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1554, "loss": 0.07521219551563263, "loss_ema": 0.16770997203130458, "loss_mean_50": 0.13285398975014687, "loss_std_50": 0.15620781098346714, "lr": 0.00025, "grad_norm_before_clip": 0.12060546875, "grad_clipped": false, "step_seconds": 0.6161764429998584, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1555, "loss": 0.22950959205627441, "loss_ema": 0.17079995303255308, "loss_mean_50": 0.1352563814818859, "loss_std_50": 0.15675121583002302, "lr": 0.00025, "grad_norm_before_clip": 0.01446533203125, "grad_clipped": false, "step_seconds": 1.8933365410048282, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1556, "loss": 0.07572393119335175, "loss_ema": 0.166046151940593, "loss_mean_50": 0.13544840648770332, "loss_std_50": 0.15667226835728995, "lr": 0.00025, "grad_norm_before_clip": 0.01141357421875, "grad_clipped": false, "step_seconds": 2.0676952109934064, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1557, "loss": 0.08169584721326828, "loss_ema": 0.16182863670422676, "loss_mean_50": 0.1321421429514885, "loss_std_50": 0.15602608633980783, "lr": 0.00025, "grad_norm_before_clip": 0.034912109375, "grad_clipped": false, "step_seconds": 0.7305260200009798, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1558, "loss": 0.06460120528936386, "loss_ema": 0.1569672651334836, "loss_mean_50": 0.13229593753814697, "loss_std_50": 0.15595562979840602, "lr": 0.00025, "grad_norm_before_clip": 0.0213623046875, "grad_clipped": false, "step_seconds": 1.8918339410010958, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1559, "loss": 0.11355328559875488, "loss_ema": 0.15479656615674717, "loss_mean_50": 0.1335554353147745, "loss_std_50": 0.15554434332951522, "lr": 0.00025, "grad_norm_before_clip": 0.034912109375, "grad_clipped": false, "step_seconds": 0.6013162489980459, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1560, "loss": 0.05759023502469063, "loss_ema": 0.14993624960014432, "loss_mean_50": 0.13328372120857238, "loss_std_50": 0.15566489366095912, "lr": 0.00025, "grad_norm_before_clip": 0.0205078125, "grad_clipped": false, "step_seconds": 2.0683684980031103, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1561, "loss": 0.29043906927108765, "loss_ema": 0.1569613905836915, "loss_mean_50": 0.1379971320927143, "loss_std_50": 0.1567801206285272, "lr": 0.00025, "grad_norm_before_clip": 0.016357421875, "grad_clipped": false, "step_seconds": 1.8937782529974356, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1562, "loss": 0.05954664200544357, "loss_ema": 0.1520906531547791, "loss_mean_50": 0.13761769354343414, "loss_std_50": 0.15694648084728416, "lr": 0.00025, "grad_norm_before_clip": 0.0213623046875, "grad_clipped": false, "step_seconds": 2.0635305150062777, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1563, "loss": 0.059481967240571976, "loss_ema": 0.14746021885906874, "loss_mean_50": 0.13651848547160625, "loss_std_50": 0.15729701686555805, "lr": 0.00025, "grad_norm_before_clip": 0.01177978515625, "grad_clipped": false, "step_seconds": 2.065917890999117, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1564, "loss": 0.08520026504993439, "loss_ema": 0.144347221168612, "loss_mean_50": 0.13604327104985714, "loss_std_50": 0.15741540129736625, "lr": 0.00025, "grad_norm_before_clip": 0.291015625, "grad_clipped": false, "step_seconds": 0.35253727399685886, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1565, "loss": 0.09235266596078873, "loss_ema": 0.14174749340822085, "loss_mean_50": 0.13279679991304874, "loss_std_50": 0.1566070586704346, "lr": 0.00025, "grad_norm_before_clip": 0.14453125, "grad_clipped": false, "step_seconds": 1.2730947820018628, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1566, "loss": 0.05197172239422798, "loss_ema": 0.1372587048575212, "loss_mean_50": 0.13262534983456134, "loss_std_50": 0.15669073554480992, "lr": 0.00025, "grad_norm_before_clip": 0.0223388671875, "grad_clipped": false, "step_seconds": 1.8893826389976311, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1567, "loss": 0.055154718458652496, "loss_ema": 0.13315350553757777, "loss_mean_50": 0.1327574311196804, "loss_std_50": 0.15662257833949586, "lr": 0.00025, "grad_norm_before_clip": 0.0771484375, "grad_clipped": false, "step_seconds": 0.5249151219977648, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1568, "loss": 0.47110840678215027, "loss_ema": 0.1500512505998064, "loss_mean_50": 0.14130307123064995, "loss_std_50": 0.16306151346626366, "lr": 0.00025, "grad_norm_before_clip": 0.0198974609375, "grad_clipped": false, "step_seconds": 1.893079970999679, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1569, "loss": 0.04821355640888214, "loss_ema": 0.14495936589026018, "loss_mean_50": 0.1411485605686903, "loss_std_50": 0.16314596614150192, "lr": 0.00025, "grad_norm_before_clip": 0.0281982421875, "grad_clipped": false, "step_seconds": 2.0670832370014978, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1570, "loss": 0.08134165406227112, "loss_ema": 0.14177848029886073, "loss_mean_50": 0.1419130089879036, "loss_std_50": 0.16277396667558003, "lr": 0.00025, "grad_norm_before_clip": 0.01116943359375, "grad_clipped": false, "step_seconds": 1.892592747004528, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1571, "loss": 0.06812181323766708, "loss_ema": 0.13809564694580104, "loss_mean_50": 0.14068817242980003, "loss_std_50": 0.16309389038663452, "lr": 0.00025, "grad_norm_before_clip": 0.01458740234375, "grad_clipped": false, "step_seconds": 1.2973813619973953, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1572, "loss": 0.055069874972105026, "loss_ema": 0.13394435834711624, "loss_mean_50": 0.13982246331870557, "loss_std_50": 0.16343082969960707, "lr": 0.00025, "grad_norm_before_clip": 0.053466796875, "grad_clipped": false, "step_seconds": 1.2913584240013734, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1573, "loss": 0.4809075891971588, "loss_ema": 0.15129251988961837, "loss_mean_50": 0.1456861313432455, "loss_std_50": 0.1701650470896379, "lr": 0.00025, "grad_norm_before_clip": 0.06982421875, "grad_clipped": false, "step_seconds": 0.3639176060023601, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1574, "loss": 0.07154318690299988, "loss_ema": 0.14730505324028742, "loss_mean_50": 0.14490762777626515, "loss_std_50": 0.170413247391308, "lr": 0.00025, "grad_norm_before_clip": 0.010009765625, "grad_clipped": false, "step_seconds": 1.1096763579989783, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1575, "loss": 0.4973618686199188, "loss_ema": 0.16480789400926898, "loss_mean_50": 0.15094456858932973, "loss_std_50": 0.17730619925885827, "lr": 0.00025, "grad_norm_before_clip": 0.02001953125, "grad_clipped": false, "step_seconds": 1.8346264399951906, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1576, "loss": 0.908517599105835, "loss_ema": 0.20199337926409727, "loss_mean_50": 0.16793643184006213, "loss_std_50": 0.20605278138445401, "lr": 0.00025, "grad_norm_before_clip": 0.042724609375, "grad_clipped": false, "step_seconds": 1.892918498000654, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1577, "loss": 0.06589590758085251, "loss_ema": 0.195188505679935, "loss_mean_50": 0.16796399898827075, "loss_std_50": 0.20603903520068192, "lr": 0.00025, "grad_norm_before_clip": 0.04296875, "grad_clipped": false, "step_seconds": 2.0584534459994757, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1578, "loss": 0.06254914402961731, "loss_ema": 0.1885565375974191, "loss_mean_50": 0.1663783887773752, "loss_std_50": 0.2065385088075518, "lr": 0.00025, "grad_norm_before_clip": 0.02978515625, "grad_clipped": false, "step_seconds": 0.47935077900183387, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1579, "loss": 0.12735047936439514, "loss_ema": 0.18549623468576792, "loss_mean_50": 0.1678342653810978, "loss_std_50": 0.2060010135147044, "lr": 0.00025, "grad_norm_before_clip": 0.01348876953125, "grad_clipped": false, "step_seconds": 1.8313000120033394, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1580, "loss": 0.09816613793373108, "loss_ema": 0.18112972984816608, "loss_mean_50": 0.16863304190337658, "loss_std_50": 0.20565159566464233, "lr": 0.00025, "grad_norm_before_clip": 0.08056640625, "grad_clipped": false, "step_seconds": 2.0588156979938503, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1581, "loss": 0.07558382302522659, "loss_ema": 0.1758524345070191, "loss_mean_50": 0.16777890987694263, "loss_std_50": 0.2059473836146576, "lr": 0.00025, "grad_norm_before_clip": 0.0208740234375, "grad_clipped": false, "step_seconds": 0.5750767170029576, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1582, "loss": 0.11336120218038559, "loss_ema": 0.1727278728906874, "loss_mean_50": 0.16774023182690143, "loss_std_50": 0.20595741808392445, "lr": 0.00025, "grad_norm_before_clip": 0.007781982421875, "grad_clipped": false, "step_seconds": 2.0662439259976964, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1583, "loss": 0.20821379125118256, "loss_ema": 0.17450216880871214, "loss_mean_50": 0.17047806687653064, "loss_std_50": 0.20556700889062587, "lr": 0.00025, "grad_norm_before_clip": 0.00799560546875, "grad_clipped": false, "step_seconds": 2.062308757005667, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1584, "loss": 0.06666894257068634, "loss_ema": 0.16911050749681084, "loss_mean_50": 0.1692762190848589, "loss_std_50": 0.20599430655727288, "lr": 0.00025, "grad_norm_before_clip": 0.0517578125, "grad_clipped": false, "step_seconds": 0.5755249359935988, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1585, "loss": 0.05235391855239868, "loss_ema": 0.16327267804959023, "loss_mean_50": 0.16931407377123833, "loss_std_50": 0.20597264172247623, "lr": 0.00025, "grad_norm_before_clip": 0.0174560546875, "grad_clipped": false, "step_seconds": 2.0714517010055715, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1586, "loss": 0.0858546569943428, "loss_ema": 0.15940177699682787, "loss_mean_50": 0.1691252599656582, "loss_std_50": 0.20604472217217745, "lr": 0.00025, "grad_norm_before_clip": 0.01806640625, "grad_clipped": false, "step_seconds": 2.07281618900015, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1587, "loss": 0.06076328456401825, "loss_ema": 0.15446985237518737, "loss_mean_50": 0.16949620097875595, "loss_std_50": 0.20583250052929242, "lr": 0.00025, "grad_norm_before_clip": 0.0213623046875, "grad_clipped": false, "step_seconds": 2.06473712099978, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1588, "loss": 0.062057558447122574, "loss_ema": 0.14984923767878414, "loss_mean_50": 0.1599848661571741, "loss_std_50": 0.1994919749850364, "lr": 0.00025, "grad_norm_before_clip": 0.1083984375, "grad_clipped": false, "step_seconds": 0.6683392310005729, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1589, "loss": 0.35938751697540283, "loss_ema": 0.16032615164361508, "loss_mean_50": 0.1662488093227148, "loss_std_50": 0.20073379297955826, "lr": 0.00025, "grad_norm_before_clip": 0.0732421875, "grad_clipped": false, "step_seconds": 0.4237629609997384, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1590, "loss": 0.058290980756282806, "loss_ema": 0.15522439309924846, "loss_mean_50": 0.16600167207419872, "loss_std_50": 0.20085890950509439, "lr": 0.00025, "grad_norm_before_clip": 0.03662109375, "grad_clipped": false, "step_seconds": 1.5747012339998037, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1591, "loss": 0.0587632954120636, "loss_ema": 0.1504013382148892, "loss_mean_50": 0.1657228659838438, "loss_std_50": 0.20099784719743913, "lr": 0.00025, "grad_norm_before_clip": 0.03515625, "grad_clipped": false, "step_seconds": 1.888484144998074, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1592, "loss": 0.059543829411268234, "loss_ema": 0.14585846277470815, "loss_mean_50": 0.16606451593339444, "loss_std_50": 0.20080246386421394, "lr": 0.00025, "grad_norm_before_clip": 0.036865234375, "grad_clipped": false, "step_seconds": 0.4848900639990461, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1593, "loss": 0.3733825087547302, "loss_ema": 0.15723466507370926, "loss_mean_50": 0.16991461269557476, "loss_std_50": 0.20288428391760543, "lr": 0.00025, "grad_norm_before_clip": 0.0213623046875, "grad_clipped": false, "step_seconds": 0.6708612330039614, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1594, "loss": 0.04938548058271408, "loss_ema": 0.1518422058491595, "loss_mean_50": 0.16840646214783192, "loss_std_50": 0.20349345048964593, "lr": 0.00025, "grad_norm_before_clip": 0.0244140625, "grad_clipped": false, "step_seconds": 1.572403377998853, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1595, "loss": 0.08085206151008606, "loss_ema": 0.14829269863220582, "loss_mean_50": 0.16859067775309086, "loss_std_50": 0.20340992088273022, "lr": 0.00025, "grad_norm_before_clip": 0.0174560546875, "grad_clipped": false, "step_seconds": 1.8932466839978588, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1596, "loss": 0.05492929741740227, "loss_ema": 0.14362452857146563, "loss_mean_50": 0.16857254929840565, "loss_std_50": 0.20342000924982773, "lr": 0.00025, "grad_norm_before_clip": 0.0091552734375, "grad_clipped": false, "step_seconds": 2.0589685949962586, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1597, "loss": 0.07455793023109436, "loss_ema": 0.14017119865444705, "loss_mean_50": 0.16916685342788695, "loss_std_50": 0.20310081374332087, "lr": 0.00025, "grad_norm_before_clip": 0.255859375, "grad_clipped": false, "step_seconds": 0.6004479130060645, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1598, "loss": 0.06595782935619354, "loss_ema": 0.13646053018953436, "loss_mean_50": 0.16866020917892455, "loss_std_50": 0.20332592051193982, "lr": 0.00025, "grad_norm_before_clip": 0.0162353515625, "grad_clipped": false, "step_seconds": 0.5254025009999168, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1599, "loss": 0.5526460409164429, "loss_ema": 0.15726980572587979, "loss_mean_50": 0.17534840404987334, "loss_std_50": 0.21022950461458625, "lr": 0.00025, "grad_norm_before_clip": 0.0302734375, "grad_clipped": false, "step_seconds": 1.2897932279956876, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1600, "loss": 0.05569903180003166, "loss_ema": 0.15219126702958738, "loss_mean_50": 0.16431326888501643, "loss_std_50": 0.2015605467034389, "lr": 0.00025, "grad_norm_before_clip": 0.0634765625, "grad_clipped": false, "step_seconds": 0.9453228749989648, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1601, "loss": 0.11612509191036224, "loss_ema": 0.1503879582736261, "loss_mean_50": 0.14839208714663982, "loss_std_50": 0.1709777588127501, "lr": 0.00025, "grad_norm_before_clip": 0.00860595703125, "grad_clipped": false, "step_seconds": 2.065046441006416, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1602, "loss": 0.06898143887519836, "loss_ema": 0.14631763230370473, "loss_mean_50": 0.14456946961581707, "loss_std_50": 0.17057336965930275, "lr": 0.00025, "grad_norm_before_clip": 0.031494140625, "grad_clipped": false, "step_seconds": 1.8979738709967933, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1603, "loss": 0.10434065759181976, "loss_ema": 0.1442187835681105, "loss_mean_50": 0.14571761459112167, "loss_std_50": 0.17010487159118434, "lr": 0.00025, "grad_norm_before_clip": 0.10791015625, "grad_clipped": false, "step_seconds": 1.8899687480006833, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1604, "loss": 0.061768047511577606, "loss_ema": 0.14009624676528384, "loss_mean_50": 0.14544873163104058, "loss_std_50": 0.17022668813932715, "lr": 0.00025, "grad_norm_before_clip": 0.040771484375, "grad_clipped": false, "step_seconds": 0.7827935960012837, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1605, "loss": 0.08155252784490585, "loss_ema": 0.13716906081926494, "loss_mean_50": 0.1424895903468132, "loss_std_50": 0.17002558269213694, "lr": 0.00025, "grad_norm_before_clip": 0.10986328125, "grad_clipped": false, "step_seconds": 0.3480707540002186, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1606, "loss": 0.1514010727405548, "loss_ema": 0.13788066141532943, "loss_mean_50": 0.14400313317775726, "loss_std_50": 0.1697611356683995, "lr": 0.00025, "grad_norm_before_clip": 0.01611328125, "grad_clipped": false, "step_seconds": 0.6045116399982362, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1607, "loss": 0.11241414397954941, "loss_ema": 0.13660733554354043, "loss_mean_50": 0.1446174991130829, "loss_std_50": 0.16959003232798897, "lr": 0.00025, "grad_norm_before_clip": 0.01055908203125, "grad_clipped": false, "step_seconds": 2.0705861230017035, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1608, "loss": 0.0488516129553318, "loss_ema": 0.13221954941413, "loss_mean_50": 0.14430250726640226, "loss_std_50": 0.1697529080712044, "lr": 0.00025, "grad_norm_before_clip": 0.0252685546875, "grad_clipped": false, "step_seconds": 2.0635438049939694, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1609, "loss": 0.23053720593452454, "loss_ema": 0.1371354322401497, "loss_mean_50": 0.14664218567311763, "loss_std_50": 0.17011876403268467, "lr": 0.00025, "grad_norm_before_clip": 0.03466796875, "grad_clipped": false, "step_seconds": 0.6161253480022424, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1610, "loss": 0.09561924636363983, "loss_ema": 0.1350596229463242, "loss_mean_50": 0.14740276589989662, "loss_std_50": 0.1698036430768099, "lr": 0.00025, "grad_norm_before_clip": 0.00616455078125, "grad_clipped": false, "step_seconds": 2.066463241004385, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1611, "loss": 0.08528206497430801, "loss_ema": 0.1325707450477234, "loss_mean_50": 0.14329962581396102, "loss_std_50": 0.16877331954909122, "lr": 0.00025, "grad_norm_before_clip": 0.0341796875, "grad_clipped": false, "step_seconds": 1.8956658879978932, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1612, "loss": 0.1048492044210434, "loss_ema": 0.1311846680163894, "loss_mean_50": 0.14420567706227302, "loss_std_50": 0.16844254187978866, "lr": 0.00025, "grad_norm_before_clip": 0.018798828125, "grad_clipped": false, "step_seconds": 2.058488219998253, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1613, "loss": 0.07789377123117447, "loss_ema": 0.12852012317712863, "loss_mean_50": 0.14457391314208506, "loss_std_50": 0.1682769667960945, "lr": 0.00025, "grad_norm_before_clip": 0.035888671875, "grad_clipped": false, "step_seconds": 0.36496985900157597, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1614, "loss": 0.086930051445961, "loss_ema": 0.12644061959057024, "loss_mean_50": 0.14460850887000562, "loss_std_50": 0.16826493411076154, "lr": 0.00025, "grad_norm_before_clip": 0.0184326171875, "grad_clipped": false, "step_seconds": 1.832698327001708, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1615, "loss": 0.5350102186203003, "loss_ema": 0.14686909954205674, "loss_mean_50": 0.15346165992319583, "loss_std_50": 0.17671549508994489, "lr": 0.00025, "grad_norm_before_clip": 0.032470703125, "grad_clipped": false, "step_seconds": 2.072380757999781, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1616, "loss": 0.07716594636440277, "loss_ema": 0.14338394188317402, "loss_mean_50": 0.15396554440259932, "loss_std_50": 0.17646112567665428, "lr": 0.00025, "grad_norm_before_clip": 0.048583984375, "grad_clipped": false, "step_seconds": 0.7301569829942309, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1617, "loss": 0.08494974672794342, "loss_ema": 0.1404622321254125, "loss_mean_50": 0.15456144496798516, "loss_std_50": 0.17617651886016775, "lr": 0.00025, "grad_norm_before_clip": 0.00787353515625, "grad_clipped": false, "step_seconds": 2.0642844140020316, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1618, "loss": 0.05438082665205002, "loss_ema": 0.13615816185174437, "loss_mean_50": 0.14622689336538314, "loss_std_50": 0.1707787591953938, "lr": 0.00025, "grad_norm_before_clip": 0.01226806640625, "grad_clipped": false, "step_seconds": 1.106167373000062, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1619, "loss": 0.0727507621049881, "loss_ema": 0.13298779186440657, "loss_mean_50": 0.14671763747930527, "loss_std_50": 0.17053148188943257, "lr": 0.00025, "grad_norm_before_clip": 0.025634765625, "grad_clipped": false, "step_seconds": 2.065366711001843, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1620, "loss": 0.5480251908302307, "loss_ema": 0.15373966181269777, "loss_mean_50": 0.15605130821466445, "loss_std_50": 0.1792465998823548, "lr": 0.00025, "grad_norm_before_clip": 0.04150390625, "grad_clipped": false, "step_seconds": 1.8922940389966243, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1621, "loss": 0.05205420032143593, "loss_ema": 0.14865538873813466, "loss_mean_50": 0.15572995595633984, "loss_std_50": 0.17941827206105845, "lr": 0.00025, "grad_norm_before_clip": 0.01611328125, "grad_clipped": false, "step_seconds": 1.5952345420009806, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1622, "loss": 0.23706966638565063, "loss_ema": 0.15307610262051047, "loss_mean_50": 0.15936995178461075, "loss_std_50": 0.17918521342880273, "lr": 0.00025, "grad_norm_before_clip": 0.018798828125, "grad_clipped": false, "step_seconds": 1.8966783910000231, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1623, "loss": 0.0775057002902031, "loss_ema": 0.14929758250399508, "loss_mean_50": 0.15130191400647164, "loss_std_50": 0.1735181648875242, "lr": 0.00025, "grad_norm_before_clip": 0.0089111328125, "grad_clipped": false, "step_seconds": 2.062480199005222, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1624, "loss": 0.31607088446617126, "loss_ema": 0.1576362476021039, "loss_mean_50": 0.15619246795773506, "loss_std_50": 0.17464358908796934, "lr": 0.00025, "grad_norm_before_clip": 0.07958984375, "grad_clipped": false, "step_seconds": 0.42914872799883597, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1625, "loss": 0.04718692973256111, "loss_ema": 0.15211378170862674, "loss_mean_50": 0.14718896917998792, "loss_std_50": 0.16831230842572956, "lr": 0.00025, "grad_norm_before_clip": 0.0225830078125, "grad_clipped": false, "step_seconds": 2.0661927229957655, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1626, "loss": 0.055297043174505234, "loss_ema": 0.14727294478192066, "loss_mean_50": 0.1301245580613613, "loss_std_50": 0.1288964542071542, "lr": 0.00025, "grad_norm_before_clip": 0.01324462890625, "grad_clipped": false, "step_seconds": 1.832600491004996, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1627, "loss": 0.054348431527614594, "loss_ema": 0.14262671911920533, "loss_mean_50": 0.12989360854029655, "loss_std_50": 0.12902161294021225, "lr": 0.00025, "grad_norm_before_clip": 0.02490234375, "grad_clipped": false, "step_seconds": 2.0664202109983307, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1628, "loss": 0.09565256536006927, "loss_ema": 0.14027801143124852, "loss_mean_50": 0.1305556769669056, "loss_std_50": 0.12875900634071866, "lr": 0.00025, "grad_norm_before_clip": 0.01123046875, "grad_clipped": false, "step_seconds": 1.8918639549956424, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1629, "loss": 0.049300502985715866, "loss_ema": 0.13572913600897188, "loss_mean_50": 0.12899467743933202, "loss_std_50": 0.12926054152162983, "lr": 0.00025, "grad_norm_before_clip": 0.062255859375, "grad_clipped": false, "step_seconds": 1.2938572480052244, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1630, "loss": 0.0682365670800209, "loss_ema": 0.13235450756252434, "loss_mean_50": 0.1283960860222578, "loss_std_50": 0.12947104798103748, "lr": 0.00025, "grad_norm_before_clip": 0.01019287109375, "grad_clipped": false, "step_seconds": 1.8327574919967446, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1631, "loss": 0.03536081686615944, "loss_ema": 0.1275048230277061, "loss_mean_50": 0.12759162589907647, "loss_std_50": 0.12992087450790935, "lr": 0.00025, "grad_norm_before_clip": 0.00946044921875, "grad_clipped": false, "step_seconds": 2.0645583270015777, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1632, "loss": 0.050840213894844055, "loss_ema": 0.123671592571063, "loss_mean_50": 0.12634120613336564, "loss_std_50": 0.13035196802894095, "lr": 0.00025, "grad_norm_before_clip": 0.0189208984375, "grad_clipped": false, "step_seconds": 1.270521660997474, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1633, "loss": 0.06476961076259613, "loss_ema": 0.12072649348063966, "loss_mean_50": 0.1234723225235939, "loss_std_50": 0.1300967482264802, "lr": 0.00025, "grad_norm_before_clip": 0.00811767578125, "grad_clipped": false, "step_seconds": 2.069338316003268, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1634, "loss": 0.07136817276477814, "loss_ema": 0.11825857744484657, "loss_mean_50": 0.12356630712747574, "loss_std_50": 0.13005736978333332, "lr": 0.00025, "grad_norm_before_clip": 0.10791015625, "grad_clipped": false, "step_seconds": 1.8891314630018314, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1635, "loss": 0.09121686965227127, "loss_ema": 0.1169064920552178, "loss_mean_50": 0.12434356614947319, "loss_std_50": 0.1297452154791818, "lr": 0.00025, "grad_norm_before_clip": 0.10986328125, "grad_clipped": false, "step_seconds": 0.4320672069952707, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1636, "loss": 0.08798804879188538, "loss_ema": 0.11546056989205118, "loss_mean_50": 0.12438623398542405, "loss_std_50": 0.12973290126056233, "lr": 0.00025, "grad_norm_before_clip": 0.057861328125, "grad_clipped": false, "step_seconds": 0.7246624440012965, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1637, "loss": 0.5784182548522949, "loss_ema": 0.13860845414006337, "loss_mean_50": 0.13473933339118957, "loss_std_50": 0.14410198793828166, "lr": 0.00025, "grad_norm_before_clip": 0.037841796875, "grad_clipped": false, "step_seconds": 0.5276994969972293, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1638, "loss": 0.05020507425069809, "loss_ema": 0.1341882851455951, "loss_mean_50": 0.13450228370726108, "loss_std_50": 0.14423104640263917, "lr": 0.00025, "grad_norm_before_clip": 0.0228271484375, "grad_clipped": false, "step_seconds": 2.063151733003906, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1639, "loss": 0.07547003775835037, "loss_ema": 0.1312523727762329, "loss_mean_50": 0.12882393412292004, "loss_std_50": 0.14081399025236233, "lr": 0.00025, "grad_norm_before_clip": 0.024658203125, "grad_clipped": false, "step_seconds": 0.5703707939974265, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1640, "loss": 0.16233684122562408, "loss_ema": 0.13280659619870244, "loss_mean_50": 0.13090485133230687, "loss_std_50": 0.1405247806722969, "lr": 0.00025, "grad_norm_before_clip": 0.068359375, "grad_clipped": false, "step_seconds": 0.6058218550024321, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1641, "loss": 0.08403391391038895, "loss_ema": 0.13036796208428678, "loss_mean_50": 0.13141026370227338, "loss_std_50": 0.14030968660437437, "lr": 0.00025, "grad_norm_before_clip": 0.01019287109375, "grad_clipped": false, "step_seconds": 2.068445108998276, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1642, "loss": 0.09689097106456757, "loss_ema": 0.12869411253330082, "loss_mean_50": 0.13215720653533936, "loss_std_50": 0.14002423424233382, "lr": 0.00025, "grad_norm_before_clip": 0.01226806640625, "grad_clipped": false, "step_seconds": 1.8323396740015596, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1643, "loss": 0.06384003907442093, "loss_ema": 0.12545140886035683, "loss_mean_50": 0.12596635714173318, "loss_std_50": 0.1360073944229452, "lr": 0.00025, "grad_norm_before_clip": 0.0810546875, "grad_clipped": false, "step_seconds": 1.893335576998652, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1644, "loss": 0.18232454359531403, "loss_ema": 0.12829506559710468, "loss_mean_50": 0.12862513840198517, "loss_std_50": 0.1357835575850202, "lr": 0.00025, "grad_norm_before_clip": 0.010498046875, "grad_clipped": false, "step_seconds": 1.8895777450015885, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1645, "loss": 0.21690909564495087, "loss_ema": 0.132725767099497, "loss_mean_50": 0.13134627908468247, "loss_std_50": 0.13616168957693145, "lr": 0.00025, "grad_norm_before_clip": 0.00946044921875, "grad_clipped": false, "step_seconds": 2.0698263049998786, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1646, "loss": 0.06243707984685898, "loss_ema": 0.12921133273686508, "loss_mean_50": 0.1314964347332716, "loss_std_50": 0.13608145213354775, "lr": 0.00025, "grad_norm_before_clip": 0.03759765625, "grad_clipped": false, "step_seconds": 2.0630770249990746, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1647, "loss": 0.04946538433432579, "loss_ema": 0.12522403531673812, "loss_mean_50": 0.13099458381533624, "loss_std_50": 0.13633653860621303, "lr": 0.00025, "grad_norm_before_clip": 0.0269775390625, "grad_clipped": false, "step_seconds": 1.830622385001334, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1648, "loss": 0.04816297069191933, "loss_ema": 0.12137098208549717, "loss_mean_50": 0.13063868664205075, "loss_std_50": 0.1365289384356534, "lr": 0.00025, "grad_norm_before_clip": 0.044677734375, "grad_clipped": false, "step_seconds": 1.2905842309992295, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1649, "loss": 0.08713191747665405, "loss_ema": 0.11965902885505501, "loss_mean_50": 0.12132840417325497, "loss_std_50": 0.12259495221561303, "lr": 0.00025, "grad_norm_before_clip": 0.0264892578125, "grad_clipped": false, "step_seconds": 0.6026105729979463, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1650, "loss": 0.08567024022340775, "loss_ema": 0.11795958942347265, "loss_mean_50": 0.12192782834172249, "loss_std_50": 0.1223456121083961, "lr": 0.00025, "grad_norm_before_clip": 0.0096435546875, "grad_clipped": false, "step_seconds": 1.8919527570033097, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1651, "loss": 0.09901561588048935, "loss_ema": 0.11701239074632348, "loss_mean_50": 0.12158563882112502, "loss_std_50": 0.12238528368682915, "lr": 0.00025, "grad_norm_before_clip": 0.015625, "grad_clipped": false, "step_seconds": 1.5728784499951871, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1652, "loss": 0.059741195291280746, "loss_ema": 0.11414883097357134, "loss_mean_50": 0.12140083394944667, "loss_std_50": 0.12247152394387993, "lr": 0.00025, "grad_norm_before_clip": 0.2041015625, "grad_clipped": false, "step_seconds": 1.8930676780000795, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1653, "loss": 0.04916539043188095, "loss_ema": 0.11089965894648682, "loss_mean_50": 0.12029732860624791, "loss_std_50": 0.12286820030728214, "lr": 0.00025, "grad_norm_before_clip": 0.047607421875, "grad_clipped": false, "step_seconds": 2.0601286010059994, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1654, "loss": 0.05491928383708, "loss_ema": 0.10810064019101646, "loss_mean_50": 0.12016035333275794, "loss_std_50": 0.1229371714509996, "lr": 0.00025, "grad_norm_before_clip": 0.0185546875, "grad_clipped": false, "step_seconds": 1.1054104499999085, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1655, "loss": 0.052520036697387695, "loss_ema": 0.10532161001633503, "loss_mean_50": 0.11957970350980758, "loss_std_50": 0.12318646003140211, "lr": 0.00025, "grad_norm_before_clip": 0.0172119140625, "grad_clipped": false, "step_seconds": 1.5898131079957238, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1656, "loss": 0.06690645217895508, "loss_ema": 0.10340085212446602, "loss_mean_50": 0.11788981109857559, "loss_std_50": 0.12331782367536155, "lr": 0.00025, "grad_norm_before_clip": 0.050537109375, "grad_clipped": false, "step_seconds": 1.8324271520032198, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1657, "loss": 0.08704451471567154, "loss_ema": 0.10258303525402629, "loss_mean_50": 0.11738241851329803, "loss_std_50": 0.12339147935073551, "lr": 0.00025, "grad_norm_before_clip": 0.04345703125, "grad_clipped": false, "step_seconds": 2.0621172309984104, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1658, "loss": 0.1274152398109436, "loss_ema": 0.10382464548187215, "loss_mean_50": 0.11895369105041027, "loss_std_50": 0.12300842248737458, "lr": 0.00025, "grad_norm_before_clip": 0.11474609375, "grad_clipped": false, "step_seconds": 2.0657661689983797, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1659, "loss": 0.054096054285764694, "loss_ema": 0.10133821592206677, "loss_mean_50": 0.11542486801743507, "loss_std_50": 0.12228545316549475, "lr": 0.00025, "grad_norm_before_clip": 0.07275390625, "grad_clipped": false, "step_seconds": 1.891631997001241, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1660, "loss": 0.052943699061870575, "loss_ema": 0.09891849007905695, "loss_mean_50": 0.11457135707139969, "loss_std_50": 0.12256931202978426, "lr": 0.00025, "grad_norm_before_clip": 0.035888671875, "grad_clipped": false, "step_seconds": 1.2951742540026316, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1661, "loss": 0.07623306661844254, "loss_ema": 0.09778421890602622, "loss_mean_50": 0.11439037710428238, "loss_std_50": 0.12261909612850785, "lr": 0.00025, "grad_norm_before_clip": 0.0576171875, "grad_clipped": false, "step_seconds": 2.063553793996107, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1662, "loss": 0.06817576289176941, "loss_ema": 0.09630379610531338, "loss_mean_50": 0.1136569082736969, "loss_std_50": 0.1227835489414588, "lr": 0.00025, "grad_norm_before_clip": 0.021728515625, "grad_clipped": false, "step_seconds": 1.893002681004873, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1663, "loss": 0.11963729560375214, "loss_ema": 0.09747047108023532, "loss_mean_50": 0.11449177876114845, "loss_std_50": 0.12267941205906839, "lr": 0.00025, "grad_norm_before_clip": 0.039794921875, "grad_clipped": false, "step_seconds": 1.8903011259972118, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1664, "loss": 0.03576454892754555, "loss_ema": 0.09438517497260082, "loss_mean_50": 0.11346846871078015, "loss_std_50": 0.12311765729493489, "lr": 0.00025, "grad_norm_before_clip": 0.0111083984375, "grad_clipped": false, "step_seconds": 2.062826084002154, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1665, "loss": 0.4153565466403961, "loss_ema": 0.11043374355599057, "loss_mean_50": 0.11107539527118206, "loss_std_50": 0.11584908299546817, "lr": 0.00025, "grad_norm_before_clip": 0.0301513671875, "grad_clipped": false, "step_seconds": 0.5755471609954839, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1666, "loss": 0.14447811245918274, "loss_ema": 0.11213596200115017, "loss_mean_50": 0.11242163859307766, "loss_std_50": 0.11583831609453207, "lr": 0.00025, "grad_norm_before_clip": 0.00823974609375, "grad_clipped": false, "step_seconds": 2.0662425699993037, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1667, "loss": 0.05742977559566498, "loss_ema": 0.1094006526808759, "loss_mean_50": 0.1118712391704321, "loss_std_50": 0.1160327563197471, "lr": 0.00025, "grad_norm_before_clip": 0.027587890625, "grad_clipped": false, "step_seconds": 2.0664283450023504, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1668, "loss": 0.06135612353682518, "loss_ema": 0.10699842622367335, "loss_mean_50": 0.1120107451081276, "loss_std_50": 0.11596772682419026, "lr": 0.00025, "grad_norm_before_clip": 0.01470947265625, "grad_clipped": false, "step_seconds": 2.063243684002373, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1669, "loss": 0.05158058553934097, "loss_ema": 0.10422753418945672, "loss_mean_50": 0.11158734157681466, "loss_std_50": 0.11614879918717082, "lr": 0.00025, "grad_norm_before_clip": 0.0235595703125, "grad_clipped": false, "step_seconds": 0.5968291899989708, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1670, "loss": 0.04955144226551056, "loss_ema": 0.1014937295932594, "loss_mean_50": 0.10161786660552025, "loss_std_50": 0.09827798457572663, "lr": 0.00025, "grad_norm_before_clip": 0.0208740234375, "grad_clipped": false, "step_seconds": 1.2667409179994138, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1671, "loss": 0.0563013069331646, "loss_ema": 0.09923410846025467, "loss_mean_50": 0.10170280873775482, "loss_std_50": 0.09823693657887188, "lr": 0.00025, "grad_norm_before_clip": 0.0223388671875, "grad_clipped": false, "step_seconds": 0.6168061089992989, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1672, "loss": 0.0578470453619957, "loss_ema": 0.09716475530534172, "loss_mean_50": 0.09811835631728172, "loss_std_50": 0.09648642518967504, "lr": 0.00025, "grad_norm_before_clip": 0.018310546875, "grad_clipped": false, "step_seconds": 2.0581224740017205, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1673, "loss": 0.09649834781885147, "loss_ema": 0.0971314349310172, "loss_mean_50": 0.09849820926785469, "loss_std_50": 0.09644190383023131, "lr": 0.00025, "grad_norm_before_clip": 0.046630859375, "grad_clipped": false, "step_seconds": 0.36238628000137396, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1674, "loss": 0.06296762824058533, "loss_ema": 0.0954232445964956, "loss_mean_50": 0.09343614414334298, "loss_std_50": 0.09139971211649657, "lr": 0.00025, "grad_norm_before_clip": 0.030029296875, "grad_clipped": false, "step_seconds": 0.6687130539939972, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1675, "loss": 0.05778903886675835, "loss_ema": 0.09354153431000874, "loss_mean_50": 0.09364818632602692, "loss_std_50": 0.09130441905357552, "lr": 0.00025, "grad_norm_before_clip": 0.0294189453125, "grad_clipped": false, "step_seconds": 0.47929758700047387, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1676, "loss": 0.795411229133606, "loss_ema": 0.1286350190511886, "loss_mean_50": 0.10845047004520893, "loss_std_50": 0.13393058088725754, "lr": 0.00025, "grad_norm_before_clip": 0.0244140625, "grad_clipped": false, "step_seconds": 1.892553592006152, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1677, "loss": 0.3681962490081787, "loss_ema": 0.1406130805490381, "loss_mean_50": 0.11472742639482021, "loss_std_50": 0.13852370590671376, "lr": 0.00025, "grad_norm_before_clip": 0.0235595703125, "grad_clipped": false, "step_seconds": 2.0678626419976354, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1678, "loss": 0.11575144529342651, "loss_ema": 0.13936999878625753, "loss_mean_50": 0.11512940399348735, "loss_std_50": 0.13849692948929979, "lr": 0.00025, "grad_norm_before_clip": 0.00927734375, "grad_clipped": false, "step_seconds": 2.0762130099974456, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1679, "loss": 0.25522854924201965, "loss_ema": 0.14516292630904562, "loss_mean_50": 0.11924796491861343, "loss_std_50": 0.13953610091907998, "lr": 0.00025, "grad_norm_before_clip": 0.0228271484375, "grad_clipped": false, "step_seconds": 2.0603877329995157, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1680, "loss": 0.04885382950305939, "loss_ema": 0.1403474714687463, "loss_mean_50": 0.1188603101670742, "loss_std_50": 0.13970410378840847, "lr": 0.00025, "grad_norm_before_clip": 0.046875, "grad_clipped": false, "step_seconds": 2.068579121005314, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1681, "loss": 0.04213704913854599, "loss_ema": 0.13543695035223627, "loss_mean_50": 0.11899583481252193, "loss_std_50": 0.1396263016641912, "lr": 0.00025, "grad_norm_before_clip": 0.01177978515625, "grad_clipped": false, "step_seconds": 2.0644758219932555, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1682, "loss": 0.0724567323923111, "loss_ema": 0.13228793945424, "loss_mean_50": 0.11942816518247128, "loss_std_50": 0.13944795156150214, "lr": 0.00025, "grad_norm_before_clip": 0.0220947265625, "grad_clipped": false, "step_seconds": 1.9023404170002323, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1683, "loss": 0.08289527893066406, "loss_ema": 0.1298183064280612, "loss_mean_50": 0.11979067854583264, "loss_std_50": 0.13932889744375632, "lr": 0.00025, "grad_norm_before_clip": 0.061767578125, "grad_clipped": false, "step_seconds": 2.0593794460000936, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1684, "loss": 0.46149325370788574, "loss_ema": 0.1464020537920524, "loss_mean_50": 0.12759318016469479, "loss_std_50": 0.14710533928959094, "lr": 0.00025, "grad_norm_before_clip": 0.038330078125, "grad_clipped": false, "step_seconds": 0.6137048370001139, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1685, "loss": 0.05262600630521774, "loss_ema": 0.14171325141771066, "loss_mean_50": 0.12682136289775373, "loss_std_50": 0.1473951219322123, "lr": 0.00025, "grad_norm_before_clip": 0.0252685546875, "grad_clipped": false, "step_seconds": 2.0606248320036684, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1686, "loss": 0.16527123749256134, "loss_ema": 0.1428911507214532, "loss_mean_50": 0.12836702667176725, "loss_std_50": 0.14738500678273778, "lr": 0.00025, "grad_norm_before_clip": 0.0771484375, "grad_clipped": false, "step_seconds": 1.8305942009974387, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1687, "loss": 0.1082746684551239, "loss_ema": 0.14116032660813674, "loss_mean_50": 0.11896415494382381, "loss_std_50": 0.13263136177033544, "lr": 0.00025, "grad_norm_before_clip": 0.0087890625, "grad_clipped": false, "step_seconds": 2.0668124110015924, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1688, "loss": 0.21745260059833527, "loss_ema": 0.14497494030764665, "loss_mean_50": 0.12230910547077656, "loss_std_50": 0.13296365174745836, "lr": 0.00025, "grad_norm_before_clip": 0.01129150390625, "grad_clipped": false, "step_seconds": 2.0644644440035336, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1689, "loss": 0.10668612271547318, "loss_ema": 0.14306049942803797, "loss_mean_50": 0.12293342716991901, "loss_std_50": 0.13281546030095834, "lr": 0.00025, "grad_norm_before_clip": 0.044677734375, "grad_clipped": false, "step_seconds": 0.34938206099468516, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1690, "loss": 0.05859675630927086, "loss_ema": 0.1388373122720996, "loss_mean_50": 0.12085862547159194, "loss_std_50": 0.1329938842820668, "lr": 0.00025, "grad_norm_before_clip": 0.0155029296875, "grad_clipped": false, "step_seconds": 1.5742968410049798, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1691, "loss": 0.08154002577066422, "loss_ema": 0.13597244794702784, "loss_mean_50": 0.12080874770879746, "loss_std_50": 0.1330081524819365, "lr": 0.00025, "grad_norm_before_clip": 0.01226806640625, "grad_clipped": false, "step_seconds": 2.0672267909976654, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1692, "loss": 0.12760214507579803, "loss_ema": 0.13555393280346634, "loss_mean_50": 0.12142297118902207, "loss_std_50": 0.13296718828469703, "lr": 0.00025, "grad_norm_before_clip": 0.0279541015625, "grad_clipped": false, "step_seconds": 1.8934591409997665, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1693, "loss": 0.0688900500535965, "loss_ema": 0.13222073866597284, "loss_mean_50": 0.12152397140860557, "loss_std_50": 0.13292532201930465, "lr": 0.00025, "grad_norm_before_clip": 0.0184326171875, "grad_clipped": false, "step_seconds": 1.2931946070020786, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1694, "loss": 0.061198096722364426, "loss_ema": 0.12866960656879242, "loss_mean_50": 0.11910144247114658, "loss_std_50": 0.13289891880317006, "lr": 0.00025, "grad_norm_before_clip": 0.03759765625, "grad_clipped": false, "step_seconds": 1.9073564329955843, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1695, "loss": 0.5298397541046143, "loss_ema": 0.1487281139455835, "loss_mean_50": 0.12536005564033986, "loss_std_50": 0.1442419663368913, "lr": 0.00025, "grad_norm_before_clip": 0.03173828125, "grad_clipped": false, "step_seconds": 2.0658392640034435, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1696, "loss": 0.08958131074905396, "loss_ema": 0.145770773785757, "loss_mean_50": 0.12590294025838375, "loss_std_50": 0.1440550814205807, "lr": 0.00025, "grad_norm_before_clip": 0.078125, "grad_clipped": false, "step_seconds": 0.5995599140005652, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1697, "loss": 0.08413521200418472, "loss_ema": 0.1426889956966784, "loss_mean_50": 0.12659633681178092, "loss_std_50": 0.14376864243522416, "lr": 0.00025, "grad_norm_before_clip": 0.01123046875, "grad_clipped": false, "step_seconds": 1.1043066099955468, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1698, "loss": 0.07771778106689453, "loss_ema": 0.13944043496518918, "loss_mean_50": 0.12718743301928043, "loss_std_50": 0.1435054687416375, "lr": 0.00025, "grad_norm_before_clip": 0.045166015625, "grad_clipped": false, "step_seconds": 0.5713006570003927, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1699, "loss": 0.09416230767965317, "loss_ema": 0.13717652860091237, "loss_mean_50": 0.12732804082334043, "loss_std_50": 0.14346959287517613, "lr": 0.00025, "grad_norm_before_clip": 0.0244140625, "grad_clipped": false, "step_seconds": 0.7246301240011235, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1700, "loss": 0.1368025243282318, "loss_ema": 0.13715782838727833, "loss_mean_50": 0.1283506865054369, "loss_std_50": 0.14335119861834042, "lr": 0.00025, "grad_norm_before_clip": 0.015625, "grad_clipped": false, "step_seconds": 1.2679116100043757, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1701, "loss": 0.10688681900501251, "loss_ema": 0.13564427791816505, "loss_mean_50": 0.12850811056792735, "loss_std_50": 0.1433232164995244, "lr": 0.00025, "grad_norm_before_clip": 0.032958984375, "grad_clipped": false, "step_seconds": 2.0645852479938185, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1702, "loss": 0.05681370571255684, "loss_ema": 0.13170274930788464, "loss_mean_50": 0.1284495607763529, "loss_std_50": 0.14335189200254544, "lr": 0.00025, "grad_norm_before_clip": 0.054443359375, "grad_clipped": false, "step_seconds": 1.8917184150050161, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1703, "loss": 0.04981617256999016, "loss_ema": 0.12760842047098991, "loss_mean_50": 0.12846257641911507, "loss_std_50": 0.14334472216574462, "lr": 0.00025, "grad_norm_before_clip": 0.043212890625, "grad_clipped": false, "step_seconds": 2.064965161996952, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1704, "loss": 0.19992341101169586, "loss_ema": 0.1312241699980252, "loss_mean_50": 0.13136265896260738, "loss_std_50": 0.14329431110114832, "lr": 0.00025, "grad_norm_before_clip": 0.0147705078125, "grad_clipped": false, "step_seconds": 1.8287822339989361, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1705, "loss": 0.3292473554611206, "loss_ema": 0.14112532927117996, "loss_mean_50": 0.13689720533788205, "loss_std_50": 0.14546983285487636, "lr": 0.00025, "grad_norm_before_clip": 0.02099609375, "grad_clipped": false, "step_seconds": 0.6703652630021679, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1706, "loss": 0.04405840113759041, "loss_ema": 0.13627198286450048, "loss_mean_50": 0.13644024431705476, "loss_std_50": 0.14572463833615384, "lr": 0.00025, "grad_norm_before_clip": 0.045654296875, "grad_clipped": false, "step_seconds": 1.5858030349991168, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1707, "loss": 0.09543034434318542, "loss_ema": 0.13422990093843473, "loss_mean_50": 0.13660796090960503, "loss_std_50": 0.14567250792516648, "lr": 0.00025, "grad_norm_before_clip": 0.039794921875, "grad_clipped": false, "step_seconds": 0.36292968800262315, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1708, "loss": 0.13072913885116577, "loss_ema": 0.13405486283407125, "loss_mean_50": 0.13667423889040947, "loss_std_50": 0.1456690641865626, "lr": 0.00025, "grad_norm_before_clip": 0.1220703125, "grad_clipped": false, "step_seconds": 2.0640351479960373, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1709, "loss": 0.03587789461016655, "loss_ema": 0.12914601442287602, "loss_mean_50": 0.1363098756968975, "loss_std_50": 0.14589776704735874, "lr": 0.00025, "grad_norm_before_clip": 0.01708984375, "grad_clipped": false, "step_seconds": 1.828618439998536, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1710, "loss": 0.05757896602153778, "loss_ema": 0.12556766200280908, "loss_mean_50": 0.13640258103609085, "loss_std_50": 0.14584622919279328, "lr": 0.00025, "grad_norm_before_clip": 0.023681640625, "grad_clipped": false, "step_seconds": 0.4206107610007166, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1711, "loss": 0.1197429671883583, "loss_ema": 0.12527642726208654, "loss_mean_50": 0.13727277904748916, "loss_std_50": 0.14561424645460946, "lr": 0.00025, "grad_norm_before_clip": 0.0191650390625, "grad_clipped": false, "step_seconds": 0.4760130350041436, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1712, "loss": 0.08617956936359406, "loss_ema": 0.12332158436716192, "loss_mean_50": 0.13763285517692567, "loss_std_50": 0.1454651211954547, "lr": 0.00025, "grad_norm_before_clip": 0.37109375, "grad_clipped": false, "step_seconds": 0.6955637270002626, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1713, "loss": 0.09498600661754608, "loss_ema": 0.12190480547968113, "loss_mean_50": 0.13713982939720154, "loss_std_50": 0.14556701783214065, "lr": 0.00025, "grad_norm_before_clip": 0.00823974609375, "grad_clipped": false, "step_seconds": 1.8906286029960029, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1714, "loss": 0.09072673320770264, "loss_ema": 0.12034590186608221, "loss_mean_50": 0.13823907308280467, "loss_std_50": 0.14500376858555883, "lr": 0.00025, "grad_norm_before_clip": 0.0810546875, "grad_clipped": false, "step_seconds": 1.8916052280037547, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1715, "loss": 0.05297314375638962, "loss_ema": 0.11697726396059757, "loss_mean_50": 0.13099140502512455, "loss_std_50": 0.13993958661793698, "lr": 0.00025, "grad_norm_before_clip": 0.0196533203125, "grad_clipped": false, "step_seconds": 1.8902580889989622, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1716, "loss": 0.06936158984899521, "loss_ema": 0.11459648025501745, "loss_mean_50": 0.1294890745729208, "loss_std_50": 0.14018972060331705, "lr": 0.00025, "grad_norm_before_clip": 0.09375, "grad_clipped": false, "step_seconds": 0.5987646490029874, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1717, "loss": 0.06005923077464104, "loss_ema": 0.11186961778099863, "loss_mean_50": 0.12954166367650033, "loss_std_50": 0.14016316994859057, "lr": 0.00025, "grad_norm_before_clip": 0.0185546875, "grad_clipped": false, "step_seconds": 0.5718901789950905, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1718, "loss": 0.04738945886492729, "loss_ema": 0.10864560983519506, "loss_mean_50": 0.12926233038306237, "loss_std_50": 0.14031261711041504, "lr": 0.00025, "grad_norm_before_clip": 0.228515625, "grad_clipped": false, "step_seconds": 1.2916287370026112, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1719, "loss": 0.23801325261592865, "loss_ema": 0.11511399197423174, "loss_mean_50": 0.13299098372459411, "loss_std_50": 0.1406754202294178, "lr": 0.00025, "grad_norm_before_clip": 0.01177978515625, "grad_clipped": false, "step_seconds": 2.0642232800018974, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1720, "loss": 0.06827842444181442, "loss_ema": 0.11277221359761086, "loss_mean_50": 0.1333655233681202, "loss_std_50": 0.14047755956657001, "lr": 0.00025, "grad_norm_before_clip": 0.0223388671875, "grad_clipped": false, "step_seconds": 1.9034899559992482, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1721, "loss": 0.07458376884460449, "loss_ema": 0.11086279135996054, "loss_mean_50": 0.13373117260634898, "loss_std_50": 0.14030017486309984, "lr": 0.00025, "grad_norm_before_clip": 0.025634765625, "grad_clipped": false, "step_seconds": 1.8467797329940367, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1722, "loss": 0.2605687081813812, "loss_ema": 0.11834808720103157, "loss_mean_50": 0.1377856058627367, "loss_std_50": 0.14097619597239844, "lr": 0.00025, "grad_norm_before_clip": 0.0595703125, "grad_clipped": false, "step_seconds": 2.0599129420006648, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1723, "loss": 0.06767280399799347, "loss_ema": 0.11581432304087966, "loss_mean_50": 0.13720909498631953, "loss_std_50": 0.1412026161663739, "lr": 0.00025, "grad_norm_before_clip": 0.01123046875, "grad_clipped": false, "step_seconds": 1.890343863000453, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1724, "loss": 0.12752479314804077, "loss_ema": 0.11639984654623771, "loss_mean_50": 0.13850023828446864, "loss_std_50": 0.14081246950310436, "lr": 0.00025, "grad_norm_before_clip": 0.2177734375, "grad_clipped": false, "step_seconds": 0.5221884480051813, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1725, "loss": 0.2662320137023926, "loss_ema": 0.12389145490404545, "loss_mean_50": 0.14266909778118134, "loss_std_50": 0.14144537586241115, "lr": 0.00025, "grad_norm_before_clip": 0.0125732421875, "grad_clipped": false, "step_seconds": 1.8909129729945562, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1726, "loss": 0.05546098202466965, "loss_ema": 0.12046993126007664, "loss_mean_50": 0.1278700928390026, "loss_std_50": 0.1068571154102806, "lr": 0.00025, "grad_norm_before_clip": 0.056884765625, "grad_clipped": false, "step_seconds": 1.283227733001695, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1727, "loss": 0.5040169954299927, "loss_ema": 0.13964728446857244, "loss_mean_50": 0.13058650776743888, "loss_std_50": 0.11439257328838044, "lr": 0.00025, "grad_norm_before_clip": 0.036376953125, "grad_clipped": false, "step_seconds": 2.072946473999764, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1728, "loss": 0.3068977892398834, "loss_ema": 0.14800980970713798, "loss_mean_50": 0.13440943464636804, "loss_std_50": 0.11699725565812562, "lr": 0.00025, "grad_norm_before_clip": 0.0125732421875, "grad_clipped": false, "step_seconds": 1.5931943100003991, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1729, "loss": 0.4527142345905304, "loss_ema": 0.1632450309513076, "loss_mean_50": 0.13835914835333823, "loss_std_50": 0.12412562723902787, "lr": 0.00025, "grad_norm_before_clip": 0.017822265625, "grad_clipped": false, "step_seconds": 2.071788152003137, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1730, "loss": 0.04764991253614426, "loss_ema": 0.1574652750305494, "loss_mean_50": 0.13833507001399994, "loss_std_50": 0.12414310300985978, "lr": 0.00025, "grad_norm_before_clip": 0.01220703125, "grad_clipped": false, "step_seconds": 2.0686574860010296, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1731, "loss": 0.07655714452266693, "loss_ema": 0.1534198685051553, "loss_mean_50": 0.13902347192168235, "loss_std_50": 0.12370240573603783, "lr": 0.00025, "grad_norm_before_clip": 0.023193359375, "grad_clipped": false, "step_seconds": 0.6741634469944984, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1732, "loss": 0.10895650088787079, "loss_ema": 0.15119670012429107, "loss_mean_50": 0.13975346729159355, "loss_std_50": 0.12341478894506432, "lr": 0.00025, "grad_norm_before_clip": 0.010986328125, "grad_clipped": false, "step_seconds": 1.2959643130016048, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1733, "loss": 0.20158790051937103, "loss_ema": 0.15371626014404505, "loss_mean_50": 0.1421273197233677, "loss_std_50": 0.12343981458980406, "lr": 0.00025, "grad_norm_before_clip": 0.01300048828125, "grad_clipped": false, "step_seconds": 1.5730084919996443, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1734, "loss": 0.13836278021335602, "loss_ema": 0.15294858614751058, "loss_mean_50": 0.1356647102534771, "loss_std_50": 0.11469966845707646, "lr": 0.00025, "grad_norm_before_clip": 0.1025390625, "grad_clipped": false, "step_seconds": 2.06749187099922, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1735, "loss": 0.10099424421787262, "loss_ema": 0.15035086905102868, "loss_mean_50": 0.1366320750117302, "loss_std_50": 0.11419811927883111, "lr": 0.00025, "grad_norm_before_clip": 0.01318359375, "grad_clipped": false, "step_seconds": 1.2957447320004576, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1736, "loss": 0.09561148285865784, "loss_ema": 0.14761389974141015, "loss_mean_50": 0.13523887991905212, "loss_std_50": 0.1142651267942918, "lr": 0.00025, "grad_norm_before_clip": 0.0216064453125, "grad_clipped": false, "step_seconds": 1.1075002580037108, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1737, "loss": 0.12978288531303406, "loss_ema": 0.14672234901999134, "loss_mean_50": 0.13566904425621032, "loss_std_50": 0.11420327553754317, "lr": 0.00025, "grad_norm_before_clip": 0.027099609375, "grad_clipped": false, "step_seconds": 0.7293571039990638, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1738, "loss": 0.839194655418396, "loss_ema": 0.18134596433991157, "loss_mean_50": 0.14810388535261154, "loss_std_50": 0.15050899501317017, "lr": 0.00025, "grad_norm_before_clip": 0.0225830078125, "grad_clipped": false, "step_seconds": 2.0618754230017657, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1739, "loss": 0.4538446366786957, "loss_ema": 0.19497089795685077, "loss_mean_50": 0.15504705563187599, "loss_std_50": 0.1563329443384825, "lr": 0.00025, "grad_norm_before_clip": 0.031982421875, "grad_clipped": false, "step_seconds": 0.599731038993923, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1740, "loss": 0.4888156056404114, "loss_ema": 0.2096631333410288, "loss_mean_50": 0.1636514326184988, "loss_std_50": 0.16250516876985405, "lr": 0.00025, "grad_norm_before_clip": 0.03955078125, "grad_clipped": false, "step_seconds": 0.4819274149995181, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1741, "loss": 0.03688061982393265, "loss_ema": 0.20102400766517398, "loss_mean_50": 0.16275824449956416, "loss_std_50": 0.16307575910828753, "lr": 0.00025, "grad_norm_before_clip": 0.0211181640625, "grad_clipped": false, "step_seconds": 1.8326600960062933, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1742, "loss": 0.05498035252094269, "loss_ema": 0.1937218249079624, "loss_mean_50": 0.16130580864846705, "loss_std_50": 0.16370460009509782, "lr": 0.00025, "grad_norm_before_clip": 0.045166015625, "grad_clipped": false, "step_seconds": 0.5759448129974771, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1743, "loss": 0.04895396903157234, "loss_ema": 0.18648343211414287, "loss_mean_50": 0.16090708702802659, "loss_std_50": 0.16395329329396832, "lr": 0.00025, "grad_norm_before_clip": 0.02490234375, "grad_clipped": false, "step_seconds": 0.42854591499781236, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1744, "loss": 0.15566696226596832, "loss_ema": 0.18494260862173414, "loss_mean_50": 0.16279646433889866, "loss_std_50": 0.16333653650824762, "lr": 0.00025, "grad_norm_before_clip": 0.01348876953125, "grad_clipped": false, "step_seconds": 2.0686362850028672, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1745, "loss": 0.15530651807785034, "loss_ema": 0.18346080409453994, "loss_mean_50": 0.15530579961836338, "loss_std_50": 0.1546913719481915, "lr": 0.00025, "grad_norm_before_clip": 0.0218505859375, "grad_clipped": false, "step_seconds": 0.6040417499971227, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1746, "loss": 0.2062174528837204, "loss_ema": 0.18459863653399894, "loss_mean_50": 0.1576385224610567, "loss_std_50": 0.15456204165115864, "lr": 0.00025, "grad_norm_before_clip": 0.03955078125, "grad_clipped": false, "step_seconds": 1.8310390309998184, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1747, "loss": 0.11017575860023499, "loss_ema": 0.18087749263731073, "loss_mean_50": 0.15815933339297772, "loss_std_50": 0.15435722536198773, "lr": 0.00025, "grad_norm_before_clip": 0.01025390625, "grad_clipped": false, "step_seconds": 1.8913281040004222, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1748, "loss": 0.08481942117214203, "loss_ema": 0.1760745890640523, "loss_mean_50": 0.15830136619508267, "loss_std_50": 0.1542863922574288, "lr": 0.00025, "grad_norm_before_clip": 0.01507568359375, "grad_clipped": false, "step_seconds": 1.8911859080035356, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1749, "loss": 0.06260450929403305, "loss_ema": 0.17040108507555132, "loss_mean_50": 0.15767021022737027, "loss_std_50": 0.15461168736491915, "lr": 0.00025, "grad_norm_before_clip": 0.04541015625, "grad_clipped": false, "step_seconds": 1.8914948969977559, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1750, "loss": 0.17012180387973785, "loss_ema": 0.17038712101576065, "loss_mean_50": 0.15833659581840037, "loss_std_50": 0.15459211311533824, "lr": 0.00025, "grad_norm_before_clip": 0.007781982421875, "grad_clipped": false, "step_seconds": 2.064385974998004, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1751, "loss": 0.047199033200740814, "loss_ema": 0.16422771662500965, "loss_mean_50": 0.15714284010231494, "loss_std_50": 0.15521400049790865, "lr": 0.00025, "grad_norm_before_clip": 0.015869140625, "grad_clipped": false, "step_seconds": 2.0642843550012913, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1752, "loss": 0.07296357303857803, "loss_ema": 0.15966450944568805, "loss_mean_50": 0.15746583744883538, "loss_std_50": 0.15502156591648375, "lr": 0.00025, "grad_norm_before_clip": 0.0908203125, "grad_clipped": false, "step_seconds": 1.8893565219987067, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1753, "loss": 0.10510952770709991, "loss_ema": 0.15693676035875864, "loss_mean_50": 0.15857170455157757, "loss_std_50": 0.15444584038935685, "lr": 0.00025, "grad_norm_before_clip": 0.048583984375, "grad_clipped": false, "step_seconds": 0.36225408199970843, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1754, "loss": 0.04968448355793953, "loss_ema": 0.15157414651871767, "loss_mean_50": 0.15556692600250244, "loss_std_50": 0.1550722998630535, "lr": 0.00025, "grad_norm_before_clip": 0.032470703125, "grad_clipped": false, "step_seconds": 1.889553639004589, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1755, "loss": 0.06503225117921829, "loss_ema": 0.1472470517517427, "loss_mean_50": 0.1502826239168644, "loss_std_50": 0.1535582203975191, "lr": 0.00025, "grad_norm_before_clip": 0.0308837890625, "grad_clipped": false, "step_seconds": 0.571839086995169, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1756, "loss": 0.07754343003034592, "loss_ema": 0.14376187066567286, "loss_mean_50": 0.15095232449471951, "loss_std_50": 0.15316601009764885, "lr": 0.00025, "grad_norm_before_clip": 0.026123046875, "grad_clipped": false, "step_seconds": 0.6138211340003181, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1757, "loss": 0.05995757505297661, "loss_ema": 0.13957165588503806, "loss_mean_50": 0.15024286910891532, "loss_std_50": 0.15350332375948408, "lr": 0.00025, "grad_norm_before_clip": 0.01324462890625, "grad_clipped": false, "step_seconds": 2.0655540770021616, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1758, "loss": 0.05148231238126755, "loss_ema": 0.13516718870984953, "loss_mean_50": 0.14865793257951737, "loss_std_50": 0.15410456067014505, "lr": 0.00025, "grad_norm_before_clip": 0.00970458984375, "grad_clipped": false, "step_seconds": 2.060375899003702, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1759, "loss": 0.1285601705312729, "loss_ema": 0.13483683780092068, "loss_mean_50": 0.1505115780979395, "loss_std_50": 0.15329211094877013, "lr": 0.00025, "grad_norm_before_clip": 0.01043701171875, "grad_clipped": false, "step_seconds": 2.064979939998011, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1760, "loss": 0.048462629318237305, "loss_ema": 0.1305181273767865, "loss_mean_50": 0.15032925136387348, "loss_std_50": 0.1534079150086008, "lr": 0.00025, "grad_norm_before_clip": 0.01470947265625, "grad_clipped": false, "step_seconds": 2.0646684899984393, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1761, "loss": 0.050833575427532196, "loss_ema": 0.12653389977932378, "loss_mean_50": 0.14895106352865695, "loss_std_50": 0.15398495498127154, "lr": 0.00025, "grad_norm_before_clip": 0.0208740234375, "grad_clipped": false, "step_seconds": 1.8895856919989455, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1762, "loss": 0.07679305970668793, "loss_ema": 0.12404685777569198, "loss_mean_50": 0.14876333333551883, "loss_std_50": 0.1540670680537406, "lr": 0.00025, "grad_norm_before_clip": 0.138671875, "grad_clipped": false, "step_seconds": 0.34853367500181776, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1763, "loss": 0.05166282877326012, "loss_ema": 0.12042765632557038, "loss_mean_50": 0.14789686977863312, "loss_std_50": 0.15448831943524663, "lr": 0.00025, "grad_norm_before_clip": 0.0223388671875, "grad_clipped": false, "step_seconds": 2.0626370789977955, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1764, "loss": 0.08743475377559662, "loss_ema": 0.11877801119807169, "loss_mean_50": 0.147831030189991, "loss_std_50": 0.15451336953867742, "lr": 0.00025, "grad_norm_before_clip": 0.017578125, "grad_clipped": false, "step_seconds": 1.299803941998107, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1765, "loss": 0.08220057934522629, "loss_ema": 0.11694913960542941, "loss_mean_50": 0.14841557890176774, "loss_std_50": 0.15420838634195025, "lr": 0.00025, "grad_norm_before_clip": 0.0101318359375, "grad_clipped": false, "step_seconds": 1.8896630340022966, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1766, "loss": 0.08474934101104736, "loss_ema": 0.1153391496757103, "loss_mean_50": 0.14872333392500878, "loss_std_50": 0.15406559913969728, "lr": 0.00025, "grad_norm_before_clip": 0.028076171875, "grad_clipped": false, "step_seconds": 0.5996325209998759, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1767, "loss": 0.10055451095104218, "loss_ema": 0.11459991773947689, "loss_mean_50": 0.1495332395285368, "loss_std_50": 0.15370338687736207, "lr": 0.00025, "grad_norm_before_clip": 0.019287109375, "grad_clipped": false, "step_seconds": 1.8911890119998134, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1768, "loss": 0.041935283690690994, "loss_ema": 0.1109666860370376, "loss_mean_50": 0.14942415602505207, "loss_std_50": 0.15377775717510597, "lr": 0.00025, "grad_norm_before_clip": 0.0216064453125, "grad_clipped": false, "step_seconds": 0.5976388010021765, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1769, "loss": 0.056194812059402466, "loss_ema": 0.10822809233815585, "loss_mean_50": 0.14578778721392155, "loss_std_50": 0.15378962603018725, "lr": 0.00025, "grad_norm_before_clip": 0.01031494140625, "grad_clipped": false, "step_seconds": 2.0625857609993545, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1770, "loss": 0.30279895663261414, "loss_ema": 0.11795663555287876, "loss_mean_50": 0.15047819785773753, "loss_std_50": 0.154926257434287, "lr": 0.00025, "grad_norm_before_clip": 0.0301513671875, "grad_clipped": false, "step_seconds": 0.6121846059977543, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1771, "loss": 0.09262288361787796, "loss_ema": 0.11668994795612872, "loss_mean_50": 0.15083898015320302, "loss_std_50": 0.15477002467965661, "lr": 0.00025, "grad_norm_before_clip": 0.1025390625, "grad_clipped": false, "step_seconds": 0.7232511080001132, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1772, "loss": 0.20787963271141052, "loss_ema": 0.12124943219389281, "loss_mean_50": 0.1497851986438036, "loss_std_50": 0.15419763467866338, "lr": 0.00025, "grad_norm_before_clip": 0.01708984375, "grad_clipped": false, "step_seconds": 0.5212326960026985, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1773, "loss": 0.04765861853957176, "loss_ema": 0.11756989151117675, "loss_mean_50": 0.14938491493463515, "loss_std_50": 0.15443606500307186, "lr": 0.00025, "grad_norm_before_clip": 0.042236328125, "grad_clipped": false, "step_seconds": 1.8905200849985704, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1774, "loss": 0.06538736820220947, "loss_ema": 0.11496076534572837, "loss_mean_50": 0.14814216643571854, "loss_std_50": 0.15485641158328675, "lr": 0.00025, "grad_norm_before_clip": 0.06884765625, "grad_clipped": false, "step_seconds": 1.8908478560042568, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1775, "loss": 0.41949453949928284, "loss_ema": 0.1301874540534061, "loss_mean_50": 0.15120741695165635, "loss_std_50": 0.15863432924345874, "lr": 0.00025, "grad_norm_before_clip": 0.019775390625, "grad_clipped": false, "step_seconds": 2.0649619639953016, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1776, "loss": 0.1369047462940216, "loss_ema": 0.13052331866543687, "loss_mean_50": 0.15283629223704337, "loss_std_50": 0.15805992822838114, "lr": 0.00025, "grad_norm_before_clip": 0.205078125, "grad_clipped": false, "step_seconds": 2.0638128069986124, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1777, "loss": 0.056320492178201675, "loss_ema": 0.1268131773410751, "loss_mean_50": 0.14388236217200756, "loss_std_50": 0.1504078332909211, "lr": 0.00025, "grad_norm_before_clip": 0.0186767578125, "grad_clipped": false, "step_seconds": 1.5864772850036388, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1778, "loss": 0.15587948262691498, "loss_ema": 0.1282664926053671, "loss_mean_50": 0.1408619960397482, "loss_std_50": 0.1486095277310253, "lr": 0.00025, "grad_norm_before_clip": 0.0164794921875, "grad_clipped": false, "step_seconds": 1.8293216129968641, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1779, "loss": 0.22705556452274323, "loss_ema": 0.1332059462012359, "loss_mean_50": 0.13634882263839246, "loss_std_50": 0.14236563226236043, "lr": 0.00025, "grad_norm_before_clip": 0.01556396484375, "grad_clipped": false, "step_seconds": 1.8292731160036055, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1780, "loss": 0.045782096683979034, "loss_ema": 0.12883475372537306, "loss_mean_50": 0.13631146632134913, "loss_std_50": 0.14238914480360756, "lr": 0.00025, "grad_norm_before_clip": 0.0458984375, "grad_clipped": false, "step_seconds": 2.058410541001649, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1781, "loss": 0.17356151342391968, "loss_ema": 0.1310710917103004, "loss_mean_50": 0.1382515536993742, "loss_std_50": 0.14222251706244976, "lr": 0.00025, "grad_norm_before_clip": 0.031982421875, "grad_clipped": false, "step_seconds": 0.3627074179967167, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1782, "loss": 0.14145250618457794, "loss_ema": 0.13159016243401428, "loss_mean_50": 0.13890147380530835, "loss_std_50": 0.14216139730245475, "lr": 0.00025, "grad_norm_before_clip": 0.00982666015625, "grad_clipped": false, "step_seconds": 2.0643468639973435, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1783, "loss": 1.0162739753723145, "loss_ema": 0.17582435308092928, "loss_mean_50": 0.15519519530236722, "loss_std_50": 0.18778028621502854, "lr": 0.00025, "grad_norm_before_clip": 0.197265625, "grad_clipped": false, "step_seconds": 2.0638361830060603, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1784, "loss": 0.09853110462427139, "loss_ema": 0.17195969065809638, "loss_mean_50": 0.15439856179058553, "loss_std_50": 0.18793443283639877, "lr": 0.00025, "grad_norm_before_clip": 0.0849609375, "grad_clipped": false, "step_seconds": 1.889623297000071, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1785, "loss": 0.06400873512029648, "loss_ema": 0.16656214288120638, "loss_mean_50": 0.153658851608634, "loss_std_50": 0.18821575350375738, "lr": 0.00025, "grad_norm_before_clip": 0.02099609375, "grad_clipped": false, "step_seconds": 0.4216983299993444, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1786, "loss": 0.11657927930355072, "loss_ema": 0.1640629997023236, "loss_mean_50": 0.15407820753753185, "loss_std_50": 0.18810928196360407, "lr": 0.00025, "grad_norm_before_clip": 0.019775390625, "grad_clipped": false, "step_seconds": 2.0641469979964313, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1787, "loss": 0.11156114190816879, "loss_ema": 0.16143790681261586, "loss_mean_50": 0.15371377266943453, "loss_std_50": 0.18817363766437742, "lr": 0.00025, "grad_norm_before_clip": 0.169921875, "grad_clipped": false, "step_seconds": 2.0650232960033463, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1788, "loss": 0.05727905035018921, "loss_ema": 0.1562299639894945, "loss_mean_50": 0.1380754605680704, "loss_std_50": 0.16109957606261607, "lr": 0.00025, "grad_norm_before_clip": 0.018310546875, "grad_clipped": false, "step_seconds": 2.0642024439948727, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1789, "loss": 0.22894851863384247, "loss_ema": 0.1598658917217119, "loss_mean_50": 0.13357753820717336, "loss_std_50": 0.1552539754853408, "lr": 0.00025, "grad_norm_before_clip": 0.017578125, "grad_clipped": false, "step_seconds": 1.8898497670015786, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1790, "loss": 0.05752710998058319, "loss_ema": 0.15474895263465543, "loss_mean_50": 0.12495176829397678, "loss_std_50": 0.14704143857770952, "lr": 0.00025, "grad_norm_before_clip": 0.0223388671875, "grad_clipped": false, "step_seconds": 0.4769897799997125, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1791, "loss": 0.17536014318466187, "loss_ema": 0.15577951216215574, "loss_mean_50": 0.12772135876119137, "loss_std_50": 0.14666016353792863, "lr": 0.00025, "grad_norm_before_clip": 0.017333984375, "grad_clipped": false, "step_seconds": 1.890343176004535, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1792, "loss": 0.19396737217903137, "loss_ema": 0.15768890516299952, "loss_mean_50": 0.13050109915435315, "loss_std_50": 0.14657224231899046, "lr": 0.00025, "grad_norm_before_clip": 0.029296875, "grad_clipped": false, "step_seconds": 1.2668760379965534, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1793, "loss": 0.05607670545578003, "loss_ema": 0.15260829517763852, "loss_mean_50": 0.1306435538828373, "loss_std_50": 0.14649635846665254, "lr": 0.00025, "grad_norm_before_clip": 0.0113525390625, "grad_clipped": false, "step_seconds": 1.1037062579998747, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1794, "loss": 0.08089593052864075, "loss_ema": 0.14902267694518861, "loss_mean_50": 0.12914813324809074, "loss_std_50": 0.14661486917212993, "lr": 0.00025, "grad_norm_before_clip": 0.0712890625, "grad_clipped": false, "step_seconds": 1.5713264489968424, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1795, "loss": 0.06427352875471115, "loss_ema": 0.14478521953566476, "loss_mean_50": 0.12732747346162795, "loss_std_50": 0.1468437745514877, "lr": 0.00025, "grad_norm_before_clip": 0.0120849609375, "grad_clipped": false, "step_seconds": 2.0595919210027205, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1796, "loss": 0.11664299666881561, "loss_ema": 0.1433781083923223, "loss_mean_50": 0.12553598433732988, "loss_std_50": 0.14641617149436748, "lr": 0.00025, "grad_norm_before_clip": 0.02099609375, "grad_clipped": false, "step_seconds": 0.5722689449976315, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1797, "loss": 0.061673082411289215, "loss_ema": 0.13929285709327063, "loss_mean_50": 0.12456593081355095, "loss_std_50": 0.14667516812802173, "lr": 0.00025, "grad_norm_before_clip": 0.013671875, "grad_clipped": false, "step_seconds": 1.8300955750019057, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1798, "loss": 0.1867513656616211, "loss_ema": 0.14166578252168815, "loss_mean_50": 0.12660456970334052, "loss_std_50": 0.14681687151833694, "lr": 0.00025, "grad_norm_before_clip": 0.0191650390625, "grad_clipped": false, "step_seconds": 1.891169656002603, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1799, "loss": 0.043210964649915695, "loss_ema": 0.13674304162809953, "loss_mean_50": 0.1262166988104582, "loss_std_50": 0.14701092827144022, "lr": 0.00025, "grad_norm_before_clip": 0.02001953125, "grad_clipped": false, "step_seconds": 2.063756072995602, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1800, "loss": 0.06731779873371124, "loss_ema": 0.13327177948338012, "loss_mean_50": 0.12416061870753765, "loss_std_50": 0.14710137324111913, "lr": 0.00025, "grad_norm_before_clip": 0.0185546875, "grad_clipped": false, "step_seconds": 0.934116869000718, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1801, "loss": 0.049266569316387177, "loss_ema": 0.12907151897503047, "loss_mean_50": 0.12420196942985058, "loss_std_50": 0.14708002229857897, "lr": 0.00025, "grad_norm_before_clip": 0.029541015625, "grad_clipped": false, "step_seconds": 1.2914667760051088, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1802, "loss": 0.06966634839773178, "loss_ema": 0.12610126044616554, "loss_mean_50": 0.12413602493703366, "loss_std_50": 0.14710371791564802, "lr": 0.00025, "grad_norm_before_clip": 0.02392578125, "grad_clipped": false, "step_seconds": 2.059216333000222, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1803, "loss": 0.06501193344593048, "loss_ema": 0.12304679409615378, "loss_mean_50": 0.12333407305181027, "loss_std_50": 0.14731440428463366, "lr": 0.00025, "grad_norm_before_clip": 0.040283203125, "grad_clipped": false, "step_seconds": 0.6677525189952576, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1804, "loss": 0.25494128465652466, "loss_ema": 0.12964151862417234, "loss_mean_50": 0.12743920907378198, "loss_std_50": 0.14806284304393405, "lr": 0.00025, "grad_norm_before_clip": 0.09326171875, "grad_clipped": false, "step_seconds": 0.3486945350014139, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1805, "loss": 0.08789680898189545, "loss_ema": 0.12755428314205847, "loss_mean_50": 0.1278965002298355, "loss_std_50": 0.1479046173561823, "lr": 0.00025, "grad_norm_before_clip": 0.0654296875, "grad_clipped": false, "step_seconds": 2.064315517003706, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1806, "loss": 0.19584369659423828, "loss_ema": 0.13096875381466744, "loss_mean_50": 0.13026250556111335, "loss_std_50": 0.1480263680914871, "lr": 0.00025, "grad_norm_before_clip": 0.044189453125, "grad_clipped": false, "step_seconds": 2.062324925995199, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1807, "loss": 0.11730022728443146, "loss_ema": 0.13028532748815563, "loss_mean_50": 0.13140935860574246, "loss_std_50": 0.14769900177908388, "lr": 0.00025, "grad_norm_before_clip": 0.130859375, "grad_clipped": false, "step_seconds": 1.9008549909995054, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1808, "loss": 0.04284519702196121, "loss_ema": 0.12591332096484592, "loss_mean_50": 0.13123661629855632, "loss_std_50": 0.14779739796968855, "lr": 0.00025, "grad_norm_before_clip": 0.03173828125, "grad_clipped": false, "step_seconds": 1.830130296999414, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1809, "loss": 0.04916246235370636, "loss_ema": 0.12207577803428893, "loss_mean_50": 0.129648662135005, "loss_std_50": 0.14824347984266056, "lr": 0.00025, "grad_norm_before_clip": 0.0283203125, "grad_clipped": false, "step_seconds": 1.57118230600463, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1810, "loss": 0.11580730229616165, "loss_ema": 0.12176235424738256, "loss_mean_50": 0.1309955555945635, "loss_std_50": 0.1478050182511513, "lr": 0.00025, "grad_norm_before_clip": 0.056884765625, "grad_clipped": false, "step_seconds": 1.8914263639962883, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1811, "loss": 0.456447035074234, "loss_ema": 0.13849658828872513, "loss_mean_50": 0.13910782478749753, "loss_std_50": 0.15417642153516434, "lr": 0.00025, "grad_norm_before_clip": 0.0400390625, "grad_clipped": false, "step_seconds": 2.0598759220010834, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1812, "loss": 0.0713871568441391, "loss_ema": 0.13514111671649584, "loss_mean_50": 0.13899970673024653, "loss_std_50": 0.15422197134736262, "lr": 0.00025, "grad_norm_before_clip": 0.1015625, "grad_clipped": false, "step_seconds": 1.890816081999219, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1813, "loss": 0.07334116846323013, "loss_ema": 0.13205111930383254, "loss_mean_50": 0.13943327352404594, "loss_std_50": 0.15400615161097886, "lr": 0.00025, "grad_norm_before_clip": 0.0167236328125, "grad_clipped": false, "step_seconds": 1.2920195710030384, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1814, "loss": 0.7405932545661926, "loss_ema": 0.16247822606695053, "loss_mean_50": 0.15249644353985786, "loss_std_50": 0.1752741790971308, "lr": 0.00025, "grad_norm_before_clip": 0.103515625, "grad_clipped": false, "step_seconds": 2.063450510999246, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1815, "loss": 0.05101730301976204, "loss_ema": 0.1569051799145911, "loss_mean_50": 0.15187277801334859, "loss_std_50": 0.1755784128195738, "lr": 0.00025, "grad_norm_before_clip": 0.017333984375, "grad_clipped": false, "step_seconds": 1.1040340600011405, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1816, "loss": 0.40756770968437195, "loss_ema": 0.16943830640308014, "loss_mean_50": 0.15832914538681508, "loss_std_50": 0.1788954477182817, "lr": 0.00025, "grad_norm_before_clip": 0.0198974609375, "grad_clipped": false, "step_seconds": 2.06302278599469, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1817, "loss": 0.16843312978744507, "loss_ema": 0.1693880475722984, "loss_mean_50": 0.15968671776354312, "loss_std_50": 0.17870932223557345, "lr": 0.00025, "grad_norm_before_clip": 0.05419921875, "grad_clipped": false, "step_seconds": 0.420978575006302, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1818, "loss": 0.050302740186452866, "loss_ema": 0.1634337822030061, "loss_mean_50": 0.15985406689345838, "loss_std_50": 0.17860286373587123, "lr": 0.00025, "grad_norm_before_clip": 0.033447265625, "grad_clipped": false, "step_seconds": 0.5721486289985478, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1819, "loss": 0.05888395011425018, "loss_ema": 0.1582062905985683, "loss_mean_50": 0.15990784965455532, "loss_std_50": 0.17857204291115086, "lr": 0.00025, "grad_norm_before_clip": 0.029296875, "grad_clipped": false, "step_seconds": 1.8913429159947555, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1820, "loss": 0.29216277599334717, "loss_ema": 0.16490411486830722, "loss_mean_50": 0.15969512604176997, "loss_std_50": 0.1784079572212959, "lr": 0.00025, "grad_norm_before_clip": 0.0233154296875, "grad_clipped": false, "step_seconds": 2.0637580820039148, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1821, "loss": 0.059954579919576645, "loss_ema": 0.1596566381208707, "loss_mean_50": 0.15904175996780395, "loss_std_50": 0.17871195295611542, "lr": 0.00025, "grad_norm_before_clip": 0.013427734375, "grad_clipped": false, "step_seconds": 1.8302617669978645, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1822, "loss": 0.48566561937332153, "loss_ema": 0.17595708718349323, "loss_mean_50": 0.1645974797010422, "loss_std_50": 0.18437205933115575, "lr": 0.00025, "grad_norm_before_clip": 0.0299072265625, "grad_clipped": false, "step_seconds": 1.8904740479993052, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1823, "loss": 0.0732855573296547, "loss_ema": 0.17082351069080132, "loss_mean_50": 0.16511001847684384, "loss_std_50": 0.18408165841376714, "lr": 0.00025, "grad_norm_before_clip": 0.0233154296875, "grad_clipped": false, "step_seconds": 1.8289864060061518, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1824, "loss": 0.10558895021677017, "loss_ema": 0.16756178266709976, "loss_mean_50": 0.16591405011713506, "loss_std_50": 0.18373179776311105, "lr": 0.00025, "grad_norm_before_clip": 0.1787109375, "grad_clipped": false, "step_seconds": 0.5976867709978251, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1825, "loss": 0.058967530727386475, "loss_ema": 0.16213207007011407, "loss_mean_50": 0.15870350994169713, "loss_std_50": 0.18068777348766288, "lr": 0.00025, "grad_norm_before_clip": 0.036376953125, "grad_clipped": false, "step_seconds": 2.0641431389958598, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1826, "loss": 0.29103007912635803, "loss_ema": 0.16857697052292628, "loss_mean_50": 0.16178601659834385, "loss_std_50": 0.18160196126956618, "lr": 0.00025, "grad_norm_before_clip": 0.0234375, "grad_clipped": false, "step_seconds": 2.0583707460027654, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1827, "loss": 0.1582678258419037, "loss_ema": 0.16806151328887514, "loss_mean_50": 0.16382496327161788, "loss_std_50": 0.18097763135237518, "lr": 0.00025, "grad_norm_before_clip": 0.0191650390625, "grad_clipped": false, "step_seconds": 0.6155064409977058, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1828, "loss": 0.058107033371925354, "loss_ema": 0.16256378929302764, "loss_mean_50": 0.1618695142865181, "loss_std_50": 0.18158012630785042, "lr": 0.00025, "grad_norm_before_clip": 0.04638671875, "grad_clipped": false, "step_seconds": 2.0654293300030986, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1829, "loss": 0.08460148423910141, "loss_ema": 0.15866567404033133, "loss_mean_50": 0.15902043268084526, "loss_std_50": 0.1816525462260832, "lr": 0.00025, "grad_norm_before_clip": 0.0172119140625, "grad_clipped": false, "step_seconds": 0.5210711089966935, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1830, "loss": 0.12606701254844666, "loss_ema": 0.15703574096573708, "loss_mean_50": 0.1606261309981346, "loss_std_50": 0.18099814757757832, "lr": 0.00025, "grad_norm_before_clip": 0.0203857421875, "grad_clipped": false, "step_seconds": 0.6680611039992073, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1831, "loss": 0.07862314581871033, "loss_ema": 0.15311511120838572, "loss_mean_50": 0.15872736364603043, "loss_std_50": 0.18135012384405508, "lr": 0.00025, "grad_norm_before_clip": 0.076171875, "grad_clipped": false, "step_seconds": 0.34837944000173593, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1832, "loss": 0.062100473791360855, "loss_ema": 0.1485643793375345, "loss_mean_50": 0.15714032299816608, "loss_std_50": 0.1818409069024063, "lr": 0.00025, "grad_norm_before_clip": 0.018310546875, "grad_clipped": false, "step_seconds": 1.891159578000952, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1833, "loss": 0.0620773620903492, "loss_ema": 0.14424002847517522, "loss_mean_50": 0.1380563907325268, "loss_std_50": 0.13461220461112147, "lr": 0.00025, "grad_norm_before_clip": 0.07080078125, "grad_clipped": false, "step_seconds": 1.890376342998934, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1834, "loss": 0.09798362106084824, "loss_ema": 0.14192720810445886, "loss_mean_50": 0.13804544106125832, "loss_std_50": 0.13461544147279514, "lr": 0.00025, "grad_norm_before_clip": 0.049072265625, "grad_clipped": false, "step_seconds": 0.7248572630051058, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1835, "loss": 0.3421151638031006, "loss_ema": 0.15193660588939092, "loss_mean_50": 0.1436075696349144, "loss_std_50": 0.13716282355985723, "lr": 0.00025, "grad_norm_before_clip": 0.064453125, "grad_clipped": false, "step_seconds": 2.0641437619997305, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1836, "loss": 0.10750824958086014, "loss_ema": 0.14971518807396436, "loss_mean_50": 0.14342614904046058, "loss_std_50": 0.13720444563527454, "lr": 0.00025, "grad_norm_before_clip": 0.05322265625, "grad_clipped": false, "step_seconds": 2.0592483609943883, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1837, "loss": 0.9366875290870667, "loss_ema": 0.18906380512461948, "loss_mean_50": 0.15992867678403855, "loss_std_50": 0.17640207268633082, "lr": 0.00025, "grad_norm_before_clip": 0.10791015625, "grad_clipped": false, "step_seconds": 1.2909095610011718, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1838, "loss": 0.058821093291044235, "loss_ema": 0.18255166953294072, "loss_mean_50": 0.15995951764285565, "loss_std_50": 0.17638425737443422, "lr": 0.00025, "grad_norm_before_clip": 0.0191650390625, "grad_clipped": false, "step_seconds": 2.062908765998145, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1839, "loss": 0.05985274910926819, "loss_ema": 0.17641672351175708, "loss_mean_50": 0.15657760225236417, "loss_std_50": 0.17664995486779006, "lr": 0.00025, "grad_norm_before_clip": 0.0301513671875, "grad_clipped": false, "step_seconds": 0.47534546600218164, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1840, "loss": 0.9667709469795227, "loss_ema": 0.21593443468514537, "loss_mean_50": 0.17476247899234296, "loss_std_50": 0.20930017183261462, "lr": 0.00025, "grad_norm_before_clip": 0.1728515625, "grad_clipped": false, "step_seconds": 0.6134518710023258, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1841, "loss": 0.17732641100883484, "loss_ema": 0.21400403350132982, "loss_mean_50": 0.17480180434882642, "loss_std_50": 0.20930046515375048, "lr": 0.00025, "grad_norm_before_clip": 0.0164794921875, "grad_clipped": false, "step_seconds": 2.0633561229988118, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1842, "loss": 0.09767851233482361, "loss_ema": 0.2081877574430045, "loss_mean_50": 0.17287602715194225, "loss_std_50": 0.20955808206647802, "lr": 0.00025, "grad_norm_before_clip": 0.05517578125, "grad_clipped": false, "step_seconds": 0.5723088640006608, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1843, "loss": 0.056519415229558945, "loss_ema": 0.2006043403323322, "loss_mean_50": 0.17288488134741783, "loss_std_50": 0.20955315619854406, "lr": 0.00025, "grad_norm_before_clip": 0.01129150390625, "grad_clipped": false, "step_seconds": 1.5868689249982708, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1844, "loss": 0.12271522730588913, "loss_ema": 0.19670988468101003, "loss_mean_50": 0.1737212672829628, "loss_std_50": 0.2092675948961434, "lr": 0.00025, "grad_norm_before_clip": 0.10595703125, "grad_clipped": false, "step_seconds": 2.063282110997534, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1845, "loss": 0.13116152584552765, "loss_ema": 0.19343246673923592, "loss_mean_50": 0.17505902722477912, "loss_std_50": 0.20877688366866648, "lr": 0.00025, "grad_norm_before_clip": 0.01104736328125, "grad_clipped": false, "step_seconds": 1.2674238649997278, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1846, "loss": 0.05382399633526802, "loss_ema": 0.18645204321903752, "loss_mean_50": 0.17380264721810817, "loss_std_50": 0.20931296822093, "lr": 0.00025, "grad_norm_before_clip": 0.0269775390625, "grad_clipped": false, "step_seconds": 2.064192384001217, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1847, "loss": 0.07894446700811386, "loss_ema": 0.18107666440849135, "loss_mean_50": 0.17414807491004466, "loss_std_50": 0.20914181802342974, "lr": 0.00025, "grad_norm_before_clip": 0.099609375, "grad_clipped": false, "step_seconds": 2.0642356520038447, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1848, "loss": 0.49814555048942566, "loss_ema": 0.19693010871253805, "loss_mean_50": 0.18037595860660077, "loss_std_50": 0.21400426148955173, "lr": 0.00025, "grad_norm_before_clip": 0.0238037109375, "grad_clipped": false, "step_seconds": 1.889425426001253, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1849, "loss": 0.08518756926059723, "loss_ema": 0.19134298173994102, "loss_mean_50": 0.1812154906988144, "loss_std_50": 0.21354636732929352, "lr": 0.00025, "grad_norm_before_clip": 0.130859375, "grad_clipped": false, "step_seconds": 0.36054507399967406, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1850, "loss": 0.058659180998802185, "loss_ema": 0.18470879170288407, "loss_mean_50": 0.18104231834411622, "loss_std_50": 0.21364215011818075, "lr": 0.00025, "grad_norm_before_clip": 0.010986328125, "grad_clipped": false, "step_seconds": 2.0718080750011723, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1851, "loss": 0.10946652293205261, "loss_ema": 0.18094667826434246, "loss_mean_50": 0.18224631741642952, "loss_std_50": 0.21306497511907993, "lr": 0.00025, "grad_norm_before_clip": 0.0150146484375, "grad_clipped": false, "step_seconds": 0.5723847209956148, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1852, "loss": 0.05985826998949051, "loss_ema": 0.17489225785059986, "loss_mean_50": 0.1820501558482647, "loss_std_50": 0.21317302089548734, "lr": 0.00025, "grad_norm_before_clip": 0.060791015625, "grad_clipped": false, "step_seconds": 1.8896702429992729, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1853, "loss": 0.14685197174549103, "loss_ema": 0.1734902435453444, "loss_mean_50": 0.1836869566142559, "loss_std_50": 0.2125814598081898, "lr": 0.00025, "grad_norm_before_clip": 0.026611328125, "grad_clipped": false, "step_seconds": 0.42140897000354016, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1854, "loss": 0.07456573843955994, "loss_ema": 0.16854401829005516, "loss_mean_50": 0.1800794456899166, "loss_std_50": 0.2128719524477926, "lr": 0.00025, "grad_norm_before_clip": 0.00994873046875, "grad_clipped": false, "step_seconds": 2.064074627996888, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1855, "loss": 0.04096795246005058, "loss_ema": 0.1621652149985549, "loss_mean_50": 0.17914086855947972, "loss_std_50": 0.21337918031680014, "lr": 0.00025, "grad_norm_before_clip": 0.011962890625, "grad_clipped": false, "step_seconds": 2.0628265679988544, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1856, "loss": 0.10135655105113983, "loss_ema": 0.15912478180118414, "loss_mean_50": 0.17725112564861775, "loss_std_50": 0.21364112847510877, "lr": 0.00025, "grad_norm_before_clip": 0.03369140625, "grad_clipped": false, "step_seconds": 0.5994665309990523, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1857, "loss": 0.2542654573917389, "loss_ema": 0.16388181558071188, "loss_mean_50": 0.1799904302507639, "loss_std_50": 0.21373294067129175, "lr": 0.00025, "grad_norm_before_clip": 0.0101318359375, "grad_clipped": false, "step_seconds": 2.063375180005096, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1858, "loss": 0.06275974959135056, "loss_ema": 0.1588257122812438, "loss_mean_50": 0.18038872130215167, "loss_std_50": 0.21349542300022745, "lr": 0.00025, "grad_norm_before_clip": 0.025146484375, "grad_clipped": false, "step_seconds": 1.8892422659992008, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1859, "loss": 0.05001714080572128, "loss_ema": 0.15338528370746768, "loss_mean_50": 0.18040581487119198, "loss_std_50": 0.21348494960799133, "lr": 0.00025, "grad_norm_before_clip": 0.0947265625, "grad_clipped": false, "step_seconds": 0.5211729950024164, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1860, "loss": 0.06632254272699356, "loss_ema": 0.14903214665844397, "loss_mean_50": 0.1794161196798086, "loss_std_50": 0.21389643455578958, "lr": 0.00025, "grad_norm_before_clip": 0.0283203125, "grad_clipped": false, "step_seconds": 2.0584182599995984, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1861, "loss": 0.15570080280303955, "loss_ema": 0.14936557946567372, "loss_mean_50": 0.17340119503438473, "loss_std_50": 0.21021853191523074, "lr": 0.00025, "grad_norm_before_clip": 0.06787109375, "grad_clipped": false, "step_seconds": 0.6131504909935757, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1862, "loss": 0.048879656940698624, "loss_ema": 0.14434128333942495, "loss_mean_50": 0.17295104503631592, "loss_std_50": 0.21046045594077967, "lr": 0.00025, "grad_norm_before_clip": 0.01007080078125, "grad_clipped": false, "step_seconds": 2.0643577720038593, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1863, "loss": 0.057806458324193954, "loss_ema": 0.14001454208866337, "loss_mean_50": 0.17264035083353518, "loss_std_50": 0.21061868376446855, "lr": 0.00025, "grad_norm_before_clip": 0.01507568359375, "grad_clipped": false, "step_seconds": 2.0639837380003883, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1864, "loss": 0.8876442909240723, "loss_ema": 0.17739602953043382, "loss_mean_50": 0.17558137156069278, "loss_std_50": 0.21937361513521253, "lr": 0.00025, "grad_norm_before_clip": 0.166015625, "grad_clipped": false, "step_seconds": 1.8894919199956348, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1865, "loss": 0.14883936941623688, "loss_ema": 0.17596819652472398, "loss_mean_50": 0.17753781288862228, "loss_std_50": 0.2186891261699292, "lr": 0.00025, "grad_norm_before_clip": 0.0146484375, "grad_clipped": false, "step_seconds": 1.1029836170055205, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1866, "loss": 0.1031484380364418, "loss_ema": 0.17232720860030987, "loss_mean_50": 0.1714494274556637, "loss_std_50": 0.21642612130056677, "lr": 0.00025, "grad_norm_before_clip": 0.0084228515625, "grad_clipped": false, "step_seconds": 1.8893153149983846, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1867, "loss": 0.7840054631233215, "loss_ema": 0.20291112132646044, "loss_mean_50": 0.1837608741223812, "loss_std_50": 0.23279392225271966, "lr": 0.00025, "grad_norm_before_clip": 0.11279296875, "grad_clipped": false, "step_seconds": 0.5960931989975506, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1868, "loss": 0.46692848205566406, "loss_ema": 0.21611198936292061, "loss_mean_50": 0.19209338895976544, "loss_std_50": 0.2353105094288608, "lr": 0.00025, "grad_norm_before_clip": 0.08251953125, "grad_clipped": false, "step_seconds": 0.3630189870018512, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1869, "loss": 0.05320945382118225, "loss_ema": 0.2079668625858337, "loss_mean_50": 0.19197989903390408, "loss_std_50": 0.23537608804750654, "lr": 0.00025, "grad_norm_before_clip": 0.0123291015625, "grad_clipped": false, "step_seconds": 2.060737532003259, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1870, "loss": 0.07397579401731491, "loss_ema": 0.20126730915740773, "loss_mean_50": 0.18761615939438342, "loss_std_50": 0.23550080212220917, "lr": 0.00025, "grad_norm_before_clip": 0.1171875, "grad_clipped": false, "step_seconds": 2.0637665309986915, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1871, "loss": 0.12505674362182617, "loss_ema": 0.19745678088062865, "loss_mean_50": 0.18891820266842843, "loss_std_50": 0.2349707567253263, "lr": 0.00025, "grad_norm_before_clip": 0.01171875, "grad_clipped": false, "step_seconds": 1.8287794509960804, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1872, "loss": 0.09175125509500504, "loss_ema": 0.19217150459134746, "loss_mean_50": 0.1810399153828621, "loss_std_50": 0.2314667078288209, "lr": 0.00025, "grad_norm_before_clip": 0.0123291015625, "grad_clipped": false, "step_seconds": 1.2657659530013916, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1873, "loss": 0.04896846413612366, "loss_ema": 0.18501135256858628, "loss_mean_50": 0.18055357351899148, "loss_std_50": 0.23171801318802, "lr": 0.00025, "grad_norm_before_clip": 0.02197265625, "grad_clipped": false, "step_seconds": 0.5716733849985758, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1874, "loss": 0.21432526409626007, "loss_ema": 0.18647704814496996, "loss_mean_50": 0.18272829979658126, "loss_std_50": 0.2315144160698703, "lr": 0.00025, "grad_norm_before_clip": 0.04296875, "grad_clipped": false, "step_seconds": 2.0638850790055585, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1875, "loss": 0.03399894759058952, "loss_ema": 0.17885314311725092, "loss_mean_50": 0.18222892813384534, "loss_std_50": 0.2318075695939539, "lr": 0.00025, "grad_norm_before_clip": 0.01708984375, "grad_clipped": false, "step_seconds": 2.065102324995678, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1876, "loss": 0.08012725412845612, "loss_ema": 0.17391684866781115, "loss_mean_50": 0.1780108716338873, "loss_std_50": 0.23170821855939225, "lr": 0.00025, "grad_norm_before_clip": 0.19921875, "grad_clipped": false, "step_seconds": 1.8302789110020967, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1877, "loss": 0.05854600667953491, "loss_ema": 0.16814830656839735, "loss_mean_50": 0.17601643525063992, "loss_std_50": 0.2322980027755482, "lr": 0.00025, "grad_norm_before_clip": 0.068359375, "grad_clipped": false, "step_seconds": 0.6694109329982894, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1878, "loss": 0.058968447148799896, "loss_ema": 0.16268931359741748, "loss_mean_50": 0.17603366352617741, "loss_std_50": 0.23228928921937184, "lr": 0.00025, "grad_norm_before_clip": 0.043701171875, "grad_clipped": false, "step_seconds": 2.064856119999604, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1879, "loss": 0.09801878780126572, "loss_ema": 0.15945578730760987, "loss_mean_50": 0.1763020095974207, "loss_std_50": 0.2321912389064902, "lr": 0.00025, "grad_norm_before_clip": 0.0184326171875, "grad_clipped": false, "step_seconds": 2.059622186003253, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1880, "loss": 0.07632169872522354, "loss_ema": 0.15529908287849054, "loss_mean_50": 0.17530710332095623, "loss_std_50": 0.2325107131706057, "lr": 0.00025, "grad_norm_before_clip": 0.0185546875, "grad_clipped": false, "step_seconds": 1.5869831269956194, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1881, "loss": 0.06142192333936691, "loss_ema": 0.15060522490153436, "loss_mean_50": 0.17496307887136936, "loss_std_50": 0.23266618645154397, "lr": 0.00025, "grad_norm_before_clip": 0.177734375, "grad_clipped": false, "step_seconds": 1.9027363280038116, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1882, "loss": 0.059291355311870575, "loss_ema": 0.14603953142205117, "loss_mean_50": 0.17490689650177957, "loss_std_50": 0.23269377035400315, "lr": 0.00025, "grad_norm_before_clip": 0.0201416015625, "grad_clipped": false, "step_seconds": 2.0637741239988827, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1883, "loss": 0.38882899284362793, "loss_ema": 0.15817900449313, "loss_mean_50": 0.18144192911684512, "loss_std_50": 0.23401779269884032, "lr": 0.00025, "grad_norm_before_clip": 0.0634765625, "grad_clipped": false, "step_seconds": 0.7241730090026977, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1884, "loss": 0.08420119434595108, "loss_ema": 0.15448011398577102, "loss_mean_50": 0.1811662805825472, "loss_std_50": 0.2341240285486033, "lr": 0.00025, "grad_norm_before_clip": 0.07861328125, "grad_clipped": false, "step_seconds": 1.8302828609957942, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1885, "loss": 0.08802174776792526, "loss_ema": 0.15115719567487873, "loss_mean_50": 0.17608441226184368, "loss_std_50": 0.2333316578989325, "lr": 0.00025, "grad_norm_before_clip": 0.01361083984375, "grad_clipped": false, "step_seconds": 1.8895151399992756, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1886, "loss": 0.0520000234246254, "loss_ema": 0.14619933706236607, "loss_mean_50": 0.17497424773871897, "loss_std_50": 0.2337869008774812, "lr": 0.00025, "grad_norm_before_clip": 0.0380859375, "grad_clipped": false, "step_seconds": 0.47613697800261434, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1887, "loss": 0.059791598469018936, "loss_ema": 0.14187895013269872, "loss_mean_50": 0.15743632912635802, "loss_std_50": 0.20738836714447073, "lr": 0.00025, "grad_norm_before_clip": 0.0206298828125, "grad_clipped": false, "step_seconds": 1.2911843940019025, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1888, "loss": 0.09029208868741989, "loss_ema": 0.13929960706043476, "loss_mean_50": 0.15806574903428555, "loss_std_50": 0.20713571962154342, "lr": 0.00025, "grad_norm_before_clip": 0.010009765625, "grad_clipped": false, "step_seconds": 1.5697848019990488, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1889, "loss": 0.08953884243965149, "loss_ema": 0.1368115688293956, "loss_mean_50": 0.15865947090089322, "loss_std_50": 0.2068957629071773, "lr": 0.00025, "grad_norm_before_clip": 0.0615234375, "grad_clipped": false, "step_seconds": 0.3489823400013847, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1890, "loss": 0.060316454619169235, "loss_ema": 0.13298681311888427, "loss_mean_50": 0.14053038105368615, "loss_std_50": 0.1720747995259124, "lr": 0.00025, "grad_norm_before_clip": 0.02978515625, "grad_clipped": false, "step_seconds": 1.8903111469990108, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1891, "loss": 0.07251156121492386, "loss_ema": 0.12996305052368626, "loss_mean_50": 0.1384340840578079, "loss_std_50": 0.17225212454836072, "lr": 0.00025, "grad_norm_before_clip": 0.033203125, "grad_clipped": false, "step_seconds": 1.4021124969949597, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1892, "loss": 0.0637994259595871, "loss_ema": 0.12665486929548128, "loss_mean_50": 0.1377565023303032, "loss_std_50": 0.17247759740234656, "lr": 0.00025, "grad_norm_before_clip": 0.033203125, "grad_clipped": false, "step_seconds": 1.889301360999525, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1893, "loss": 0.061423853039741516, "loss_ema": 0.12339331848269429, "loss_mean_50": 0.13785459108650686, "loss_std_50": 0.1724327583866357, "lr": 0.00025, "grad_norm_before_clip": 0.017333984375, "grad_clipped": false, "step_seconds": 1.8380700670022634, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1894, "loss": 0.045828841626644135, "loss_ema": 0.11951509463989178, "loss_mean_50": 0.13631686337292195, "loss_std_50": 0.17290310096369618, "lr": 0.00025, "grad_norm_before_clip": 0.028564453125, "grad_clipped": false, "step_seconds": 2.0642743860007613, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1895, "loss": 0.0720529556274414, "loss_ema": 0.11714198768926926, "loss_mean_50": 0.1351346919685602, "loss_std_50": 0.17313621874208407, "lr": 0.00025, "grad_norm_before_clip": 0.00958251953125, "grad_clipped": false, "step_seconds": 2.0623966339990147, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1896, "loss": 0.05092976987361908, "loss_ema": 0.11383137679848675, "loss_mean_50": 0.13507680743932723, "loss_std_50": 0.1731638752260441, "lr": 0.00025, "grad_norm_before_clip": 0.06689453125, "grad_clipped": false, "step_seconds": 0.5972924540037639, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1897, "loss": 0.8006489276885986, "loss_ema": 0.14817225434299233, "loss_mean_50": 0.14951089665293693, "loss_std_50": 0.19640288415420742, "lr": 0.00025, "grad_norm_before_clip": 0.2314453125, "grad_clipped": false, "step_seconds": 2.0629729309948743, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1898, "loss": 0.06162630394101143, "loss_ema": 0.14384495682289328, "loss_mean_50": 0.14078051172196865, "loss_std_50": 0.19031927090722367, "lr": 0.00025, "grad_norm_before_clip": 0.055419921875, "grad_clipped": false, "step_seconds": 2.062973277999845, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1899, "loss": 0.05221311375498772, "loss_ema": 0.13926336466949799, "loss_mean_50": 0.14012102261185647, "loss_std_50": 0.19056773632331736, "lr": 0.00025, "grad_norm_before_clip": 0.01409912109375, "grad_clipped": false, "step_seconds": 1.1030264880027971, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1900, "loss": 0.0881686806678772, "loss_ema": 0.13670863046941695, "loss_mean_50": 0.14071121260523797, "loss_std_50": 0.1903601168497608, "lr": 0.00025, "grad_norm_before_clip": 0.072265625, "grad_clipped": false, "step_seconds": 2.0591061689992785, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1901, "loss": 0.10440252721309662, "loss_ema": 0.13509332530660093, "loss_mean_50": 0.14060993269085884, "loss_std_50": 0.19037805973766037, "lr": 0.00025, "grad_norm_before_clip": 0.0157470703125, "grad_clipped": false, "step_seconds": 1.8908526169980178, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1902, "loss": 0.0326114296913147, "loss_ema": 0.12996923052583662, "loss_mean_50": 0.14006499588489532, "loss_std_50": 0.1906472281093429, "lr": 0.00025, "grad_norm_before_clip": 0.020751953125, "grad_clipped": false, "step_seconds": 2.0627089420013363, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1903, "loss": 0.0911811888217926, "loss_ema": 0.1280298284406344, "loss_mean_50": 0.13895158022642135, "loss_std_50": 0.19076686599716497, "lr": 0.00025, "grad_norm_before_clip": 0.034912109375, "grad_clipped": false, "step_seconds": 0.6685350539992214, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1904, "loss": 0.04810663312673569, "loss_ema": 0.12403366867493947, "loss_mean_50": 0.13842239812016488, "loss_std_50": 0.19098131449044314, "lr": 0.00025, "grad_norm_before_clip": 0.0157470703125, "grad_clipped": false, "step_seconds": 1.8892882869986352, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1905, "loss": 0.09553193300962448, "loss_ema": 0.1226085818916737, "loss_mean_50": 0.13951367773115636, "loss_std_50": 0.19057679830891733, "lr": 0.00025, "grad_norm_before_clip": 0.0966796875, "grad_clipped": false, "step_seconds": 0.3478558980059461, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1906, "loss": 0.3702840805053711, "loss_ema": 0.13499235682235855, "loss_mean_50": 0.14489222832024098, "loss_std_50": 0.19320084755725161, "lr": 0.00025, "grad_norm_before_clip": 0.06494140625, "grad_clipped": false, "step_seconds": 2.0632921119977254, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1907, "loss": 0.06591664999723434, "loss_ema": 0.13153857148110232, "loss_mean_50": 0.1411252521723509, "loss_std_50": 0.19286749379358914, "lr": 0.00025, "grad_norm_before_clip": 0.0250244140625, "grad_clipped": false, "step_seconds": 2.0587694279965945, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1908, "loss": 0.08464134484529495, "loss_ema": 0.12919371014931194, "loss_mean_50": 0.14156288407742978, "loss_std_50": 0.19271394406316056, "lr": 0.00025, "grad_norm_before_clip": 0.049072265625, "grad_clipped": false, "step_seconds": 1.290042678003374, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1909, "loss": 0.41231533885002136, "loss_ema": 0.1433497915843474, "loss_mean_50": 0.14880884803831576, "loss_std_50": 0.19592010008097896, "lr": 0.00025, "grad_norm_before_clip": 0.05126953125, "grad_clipped": false, "step_seconds": 0.5983401889970992, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1910, "loss": 0.053957391530275345, "loss_ema": 0.1388801715816438, "loss_mean_50": 0.1485615450143814, "loss_std_50": 0.19603183573519595, "lr": 0.00025, "grad_norm_before_clip": 0.046875, "grad_clipped": false, "step_seconds": 0.613105930002348, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1911, "loss": 0.05136307328939438, "loss_ema": 0.13450431666703133, "loss_mean_50": 0.1464747904241085, "loss_std_50": 0.19649951031763113, "lr": 0.00025, "grad_norm_before_clip": 0.06298828125, "grad_clipped": false, "step_seconds": 1.891640132998873, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1912, "loss": 0.14957855641841888, "loss_ema": 0.1352580286546007, "loss_mean_50": 0.1484887684136629, "loss_std_50": 0.19600433136387693, "lr": 0.00025, "grad_norm_before_clip": 0.007110595703125, "grad_clipped": false, "step_seconds": 2.0636011329988833, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1913, "loss": 0.23750029504299164, "loss_ema": 0.14037014197402026, "loss_mean_50": 0.15208264514803888, "loss_std_50": 0.19595605980324518, "lr": 0.00025, "grad_norm_before_clip": 0.01171875, "grad_clipped": false, "step_seconds": 1.5692581980038085, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1914, "loss": 0.27736207842826843, "loss_ema": 0.14721973879673264, "loss_mean_50": 0.1398770008981228, "loss_std_50": 0.16656133901774103, "lr": 0.00025, "grad_norm_before_clip": 0.0245361328125, "grad_clipped": false, "step_seconds": 0.5216300839965697, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1915, "loss": 0.045756008476018906, "loss_ema": 0.14214655228069695, "loss_mean_50": 0.13781533367931842, "loss_std_50": 0.16707482732884424, "lr": 0.00025, "grad_norm_before_clip": 0.0159912109375, "grad_clipped": false, "step_seconds": 2.063025891999132, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1916, "loss": 0.3665213882923126, "loss_ema": 0.15336529408127772, "loss_mean_50": 0.14308279268443586, "loss_std_50": 0.17002454269268036, "lr": 0.00025, "grad_norm_before_clip": 0.0986328125, "grad_clipped": false, "step_seconds": 0.4207922650020919, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1917, "loss": 0.528042197227478, "loss_ema": 0.17209913923858772, "loss_mean_50": 0.13796352736651898, "loss_std_50": 0.1537217476928111, "lr": 0.00025, "grad_norm_before_clip": 0.08349609375, "grad_clipped": false, "step_seconds": 1.2910258699994301, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1918, "loss": 0.13278812170028687, "loss_ema": 0.17013358836167267, "loss_mean_50": 0.13128072015941142, "loss_std_50": 0.14636219686028387, "lr": 0.00025, "grad_norm_before_clip": 0.0177001953125, "grad_clipped": false, "step_seconds": 1.8302765250045923, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1919, "loss": 0.07197242975234985, "loss_ema": 0.16522553043120652, "loss_mean_50": 0.13165597967803477, "loss_std_50": 0.1461854947104166, "lr": 0.00025, "grad_norm_before_clip": 0.0191650390625, "grad_clipped": false, "step_seconds": 2.0592882799974177, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1920, "loss": 0.04334652051329613, "loss_ema": 0.159131579935311, "loss_mean_50": 0.1310433942079544, "loss_std_50": 0.14648977690601264, "lr": 0.00025, "grad_norm_before_clip": 0.0286865234375, "grad_clipped": false, "step_seconds": 1.2661248910007998, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1921, "loss": 0.09490552544593811, "loss_ema": 0.15592027721084234, "loss_mean_50": 0.13044036984443663, "loss_std_50": 0.14657521349484287, "lr": 0.00025, "grad_norm_before_clip": 0.02783203125, "grad_clipped": false, "step_seconds": 0.5718377430021064, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1922, "loss": 0.048584938049316406, "loss_ema": 0.15055351025276603, "loss_mean_50": 0.12957704350352287, "loss_std_50": 0.14692725125371833, "lr": 0.00025, "grad_norm_before_clip": 0.024169921875, "grad_clipped": false, "step_seconds": 2.0641802309983177, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1923, "loss": 0.94114089012146, "loss_ema": 0.19008287924620074, "loss_mean_50": 0.1474204920232296, "loss_std_50": 0.18523496210837545, "lr": 0.00025, "grad_norm_before_clip": 0.271484375, "grad_clipped": false, "step_seconds": 2.0623015919991303, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1924, "loss": 0.8082520961761475, "loss_ema": 0.22099134009269805, "loss_mean_50": 0.15929902866482734, "loss_std_50": 0.20691866702878453, "lr": 0.00025, "grad_norm_before_clip": 0.0732421875, "grad_clipped": false, "step_seconds": 0.36499189599999227, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1925, "loss": 0.6966070532798767, "loss_ema": 0.24477212575205698, "loss_mean_50": 0.17255119077861308, "loss_std_50": 0.21931646318561968, "lr": 0.00025, "grad_norm_before_clip": 0.06591796875, "grad_clipped": false, "step_seconds": 1.8896887079972657, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1926, "loss": 0.07306141406297684, "loss_ema": 0.23618659016760296, "loss_mean_50": 0.1724098739773035, "loss_std_50": 0.21937823886135843, "lr": 0.00025, "grad_norm_before_clip": 0.02099609375, "grad_clipped": false, "step_seconds": 0.5747440080012893, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1927, "loss": 0.1944829821586609, "loss_ema": 0.23410140976715585, "loss_mean_50": 0.17512861348688602, "loss_std_50": 0.21879183023587948, "lr": 0.00025, "grad_norm_before_clip": 0.015869140625, "grad_clipped": false, "step_seconds": 2.040343667998968, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1928, "loss": 0.04948178678750992, "loss_ema": 0.22487042861817352, "loss_mean_50": 0.17493888027966023, "loss_std_50": 0.21889656871589971, "lr": 0.00025, "grad_norm_before_clip": 0.03466796875, "grad_clipped": false, "step_seconds": 1.894669747001899, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1929, "loss": 0.08628997951745987, "loss_ema": 0.21794140616313784, "loss_mean_50": 0.1747043041139841, "loss_std_50": 0.2189851394711667, "lr": 0.00025, "grad_norm_before_clip": 0.016357421875, "grad_clipped": false, "step_seconds": 1.8333660970019992, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1930, "loss": 0.07908155024051666, "loss_ema": 0.21099841336700678, "loss_mean_50": 0.17475950114428998, "loss_std_50": 0.2189606808146145, "lr": 0.00025, "grad_norm_before_clip": 0.011962890625, "grad_clipped": false, "step_seconds": 1.9003024229969014, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1931, "loss": 0.07455580681562424, "loss_ema": 0.20417628303943766, "loss_mean_50": 0.17502217881381513, "loss_std_50": 0.21883239755468095, "lr": 0.00025, "grad_norm_before_clip": 0.0308837890625, "grad_clipped": false, "step_seconds": 1.8905346500032465, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1932, "loss": 0.04901126027107239, "loss_ema": 0.1964180319010194, "loss_mean_50": 0.17481657691299915, "loss_std_50": 0.2189458346465154, "lr": 0.00025, "grad_norm_before_clip": 0.03759765625, "grad_clipped": false, "step_seconds": 0.4801295500001288, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1933, "loss": 0.08015536516904831, "loss_ema": 0.19060489856442084, "loss_mean_50": 0.16864310435950755, "loss_std_50": 0.2171689557980558, "lr": 0.00025, "grad_norm_before_clip": 0.08935546875, "grad_clipped": false, "step_seconds": 2.066828147995693, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1934, "loss": 0.18132154643535614, "loss_ema": 0.1901407309579676, "loss_mean_50": 0.17058551140129566, "loss_std_50": 0.21683908446550082, "lr": 0.00025, "grad_norm_before_clip": 0.0123291015625, "grad_clipped": false, "step_seconds": 1.5880800540035125, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1935, "loss": 0.13145612180233002, "loss_ema": 0.18720650050018572, "loss_mean_50": 0.17145419888198377, "loss_std_50": 0.21659344545761097, "lr": 0.00025, "grad_norm_before_clip": 0.0224609375, "grad_clipped": false, "step_seconds": 0.7236113530016155, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1936, "loss": 0.0725291445851326, "loss_ema": 0.18147263270443306, "loss_mean_50": 0.1718647813051939, "loss_std_50": 0.21638597312196928, "lr": 0.00025, "grad_norm_before_clip": 0.01446533203125, "grad_clipped": false, "step_seconds": 1.901471720993868, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1937, "loss": 0.0874502956867218, "loss_ema": 0.1767715158535475, "loss_mean_50": 0.17241795524954795, "loss_std_50": 0.2161339665669256, "lr": 0.00025, "grad_norm_before_clip": 0.05810546875, "grad_clipped": false, "step_seconds": 0.3515278810009477, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1938, "loss": 0.08725083619356155, "loss_ema": 0.1722954818705482, "loss_mean_50": 0.1723571301996708, "loss_std_50": 0.2161574967667338, "lr": 0.00025, "grad_norm_before_clip": 0.009765625, "grad_clipped": false, "step_seconds": 2.067363617999945, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1939, "loss": 0.07078014314174652, "loss_ema": 0.1672197149341081, "loss_mean_50": 0.1719819562137127, "loss_std_50": 0.2163171351585123, "lr": 0.00025, "grad_norm_before_clip": 0.0137939453125, "grad_clipped": false, "step_seconds": 1.8908966400049394, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1940, "loss": 0.10423104465007782, "loss_ema": 0.16407028141990657, "loss_mean_50": 0.17286024801433086, "loss_std_50": 0.21595080829236526, "lr": 0.00025, "grad_norm_before_clip": 0.022216796875, "grad_clipped": false, "step_seconds": 1.894082922997768, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1941, "loss": 0.11808615177869797, "loss_ema": 0.16177107493784612, "loss_mean_50": 0.17377173982560634, "loss_std_50": 0.2156212597086038, "lr": 0.00025, "grad_norm_before_clip": 0.01214599609375, "grad_clipped": false, "step_seconds": 2.064141433998884, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1942, "loss": 0.9829082489013672, "loss_ema": 0.20282793363602217, "loss_mean_50": 0.19215391628444195, "loss_std_50": 0.24291311787757913, "lr": 0.00025, "grad_norm_before_clip": 0.3671875, "grad_clipped": false, "step_seconds": 0.670253746000526, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1943, "loss": 0.8264909386634827, "loss_ema": 0.2340110838873952, "loss_mean_50": 0.20745525799691678, "loss_std_50": 0.25783427613352655, "lr": 0.00025, "grad_norm_before_clip": 0.0859375, "grad_clipped": false, "step_seconds": 1.2703170219974709, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1944, "loss": 0.06540098041296005, "loss_ema": 0.22558057871367343, "loss_mean_50": 0.20784670077264308, "loss_std_50": 0.2576033522842376, "lr": 0.00025, "grad_norm_before_clip": 0.017333984375, "grad_clipped": false, "step_seconds": 2.0672381469994434, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1945, "loss": 0.09592084586620331, "loss_ema": 0.2190975920712999, "loss_mean_50": 0.20832405857741831, "loss_std_50": 0.2573732860129869, "lr": 0.00025, "grad_norm_before_clip": 0.01141357421875, "grad_clipped": false, "step_seconds": 2.0627453820052324, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1946, "loss": 0.03544935584068298, "loss_ema": 0.20991518025976905, "loss_mean_50": 0.2080144502967596, "loss_std_50": 0.2575716725698502, "lr": 0.00025, "grad_norm_before_clip": 0.0189208984375, "grad_clipped": false, "step_seconds": 2.0653608979991986, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1947, "loss": 0.06964078545570374, "loss_ema": 0.20290146051956576, "loss_mean_50": 0.19339428745210172, "loss_std_50": 0.2439017226166858, "lr": 0.00025, "grad_norm_before_clip": 0.036376953125, "grad_clipped": false, "step_seconds": 1.2932063760017627, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1948, "loss": 0.13368383049964905, "loss_ema": 0.1994405790185699, "loss_mean_50": 0.19483543798327446, "loss_std_50": 0.24333109978864412, "lr": 0.00025, "grad_norm_before_clip": 0.01263427734375, "grad_clipped": false, "step_seconds": 1.2946885399942403, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1949, "loss": 0.059287264943122864, "loss_ema": 0.19243291331479756, "loss_mean_50": 0.19497692100703717, "loss_std_50": 0.24325017513627806, "lr": 0.00025, "grad_norm_before_clip": 0.06884765625, "grad_clipped": false, "step_seconds": 2.062585207000666, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1950, "loss": 0.6299062371253967, "loss_ema": 0.2143065795053275, "loss_mean_50": 0.20581167213618756, "loss_std_50": 0.2502166382719104, "lr": 0.00025, "grad_norm_before_clip": 0.263671875, "grad_clipped": false, "step_seconds": 1.8960642899983213, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1951, "loss": 0.05584219843149185, "loss_ema": 0.2063833604516357, "loss_mean_50": 0.20484046556055546, "loss_std_50": 0.25070214078317743, "lr": 0.00025, "grad_norm_before_clip": 0.016845703125, "grad_clipped": false, "step_seconds": 2.0670098580012564, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1952, "loss": 0.9239352941513062, "loss_ema": 0.24226095713661921, "loss_mean_50": 0.2226669428497553, "loss_std_50": 0.26885399519984077, "lr": 0.00025, "grad_norm_before_clip": 0.265625, "grad_clipped": false, "step_seconds": 1.8926118910021614, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1953, "loss": 0.05825597792863846, "loss_ema": 0.23306070817622015, "loss_mean_50": 0.2220084386318922, "loss_std_50": 0.26921531595470355, "lr": 0.00025, "grad_norm_before_clip": 0.0140380859375, "grad_clipped": false, "step_seconds": 1.5906209390013828, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1954, "loss": 0.11154986172914505, "loss_ema": 0.22698516585386638, "loss_mean_50": 0.2232773032039404, "loss_std_50": 0.26854135903498594, "lr": 0.00025, "grad_norm_before_clip": 0.0091552734375, "grad_clipped": false, "step_seconds": 2.0666776540019782, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1955, "loss": 0.11583366245031357, "loss_ema": 0.22142759068368872, "loss_mean_50": 0.22368333779275418, "loss_std_50": 0.2683631900520564, "lr": 0.00025, "grad_norm_before_clip": 0.038818359375, "grad_clipped": false, "step_seconds": 0.6200495779994526, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1956, "loss": 0.12248149514198303, "loss_ema": 0.21648028590660343, "loss_mean_50": 0.21872728608548642, "loss_std_50": 0.267897816396908, "lr": 0.00025, "grad_norm_before_clip": 0.068359375, "grad_clipped": false, "step_seconds": 2.0642619779973757, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1957, "loss": 0.050319869071245193, "loss_ema": 0.20817226506483552, "loss_mean_50": 0.21841535046696664, "loss_std_50": 0.2680845800833677, "lr": 0.00025, "grad_norm_before_clip": 0.0238037109375, "grad_clipped": false, "step_seconds": 1.8283902170005604, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1958, "loss": 0.08900422602891922, "loss_ema": 0.2022138631130397, "loss_mean_50": 0.21850260809063912, "loss_std_50": 0.2680417309964024, "lr": 0.00025, "grad_norm_before_clip": 0.1044921875, "grad_clipped": false, "step_seconds": 0.3634072089989786, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1959, "loss": 0.09867019951343536, "loss_ema": 0.19703667993305948, "loss_mean_50": 0.2122297053039074, "loss_std_50": 0.2671010087532356, "lr": 0.00025, "grad_norm_before_clip": 0.0169677734375, "grad_clipped": false, "step_seconds": 1.107126460003201, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1960, "loss": 0.05392569303512573, "loss_ema": 0.18988113058816278, "loss_mean_50": 0.2122290713340044, "loss_std_50": 0.26710138445252746, "lr": 0.00025, "grad_norm_before_clip": 0.04052734375, "grad_clipped": false, "step_seconds": 2.067057684995234, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1961, "loss": 0.13724204897880554, "loss_ema": 0.1872491765076949, "loss_mean_50": 0.21394665084779263, "loss_std_50": 0.26633644733863765, "lr": 0.00025, "grad_norm_before_clip": 0.0106201171875, "grad_clipped": false, "step_seconds": 1.8291869509994285, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1962, "loss": 0.10344169288873672, "loss_ema": 0.18305880232674698, "loss_mean_50": 0.213023913577199, "loss_std_50": 0.26663760728741476, "lr": 0.00025, "grad_norm_before_clip": 0.02099609375, "grad_clipped": false, "step_seconds": 0.5747634020008263, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1963, "loss": 0.06050804257392883, "loss_ema": 0.17693126433910608, "loss_mean_50": 0.20948406852781773, "loss_std_50": 0.26746275080868465, "lr": 0.00025, "grad_norm_before_clip": 0.0224609375, "grad_clipped": false, "step_seconds": 2.017290317999141, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1964, "loss": 0.06981811672449112, "loss_ema": 0.17157560695837532, "loss_mean_50": 0.20533318929374217, "loss_std_50": 0.26798708242942293, "lr": 0.00025, "grad_norm_before_clip": 0.035888671875, "grad_clipped": false, "step_seconds": 0.4245040230016457, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1965, "loss": 0.7423338890075684, "loss_ema": 0.20011352106083496, "loss_mean_50": 0.21926474690437317, "loss_std_50": 0.2772743867519756, "lr": 0.00025, "grad_norm_before_clip": 0.322265625, "grad_clipped": false, "step_seconds": 2.0647273319991655, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1966, "loss": 0.7429419755935669, "loss_ema": 0.22725494378747155, "loss_mean_50": 0.22679315865039826, "loss_std_50": 0.2861389085628887, "lr": 0.00025, "grad_norm_before_clip": 0.158203125, "grad_clipped": false, "step_seconds": 0.7282211339988862, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1967, "loss": 0.14539840817451477, "loss_ema": 0.2231621170068237, "loss_mean_50": 0.21914028286933898, "loss_std_50": 0.2830801846280824, "lr": 0.00025, "grad_norm_before_clip": 0.01129150390625, "grad_clipped": false, "step_seconds": 2.0640118720039027, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1968, "loss": 0.0787310004234314, "loss_ema": 0.2159405611776541, "loss_mean_50": 0.21805914044380187, "loss_std_50": 0.2835108172138994, "lr": 0.00025, "grad_norm_before_clip": 0.07861328125, "grad_clipped": false, "step_seconds": 0.5988740719985799, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1969, "loss": 0.08896520733833313, "loss_ema": 0.20959179348568802, "loss_mean_50": 0.21839899599552154, "loss_std_50": 0.2833456304691796, "lr": 0.00025, "grad_norm_before_clip": 0.0250244140625, "grad_clipped": false, "step_seconds": 0.5260134609998204, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1970, "loss": 0.046186383813619614, "loss_ema": 0.20142152300208457, "loss_mean_50": 0.218455793261528, "loss_std_50": 0.28331081760583837, "lr": 0.00025, "grad_norm_before_clip": 0.0115966796875, "grad_clipped": false, "step_seconds": 2.0705446249994566, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1971, "loss": 0.42643624544143677, "loss_ema": 0.21267225912405216, "loss_mean_50": 0.22508640766143798, "loss_std_50": 0.28421977116555674, "lr": 0.00025, "grad_norm_before_clip": 0.1845703125, "grad_clipped": false, "step_seconds": 2.0670687329984503, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1972, "loss": 0.09869962930679321, "loss_ema": 0.20697362763318922, "loss_mean_50": 0.22608870148658752, "loss_std_50": 0.28368343396793905, "lr": 0.00025, "grad_norm_before_clip": 0.099609375, "grad_clipped": false, "step_seconds": 1.8940576169989072, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1973, "loss": 0.051787447184324265, "loss_ema": 0.19921431861074596, "loss_mean_50": 0.20830163262784482, "loss_std_50": 0.2655965677203921, "lr": 0.00025, "grad_norm_before_clip": 0.054443359375, "grad_clipped": false, "step_seconds": 1.5803140350035392, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1974, "loss": 0.11612462997436523, "loss_ema": 0.19505983417892692, "loss_mean_50": 0.19445908330380918, "loss_std_50": 0.2516367251320673, "lr": 0.00025, "grad_norm_before_clip": 0.0162353515625, "grad_clipped": false, "step_seconds": 1.8911204410032951, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1975, "loss": 0.11998701840639114, "loss_ema": 0.1913061933903001, "loss_mean_50": 0.18292668260633946, "loss_std_50": 0.24136262192668645, "lr": 0.00025, "grad_norm_before_clip": 0.0341796875, "grad_clipped": false, "step_seconds": 0.5970990339992568, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1976, "loss": 0.5437679290771484, "loss_ema": 0.20892928017464252, "loss_mean_50": 0.1923408129066229, "loss_std_50": 0.24602847390589552, "lr": 0.00025, "grad_norm_before_clip": 0.031982421875, "grad_clipped": false, "step_seconds": 0.5712374340000679, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1977, "loss": 0.07875902950763702, "loss_ema": 0.20242076764129222, "loss_mean_50": 0.1900263338536024, "loss_std_50": 0.2465412294359644, "lr": 0.00025, "grad_norm_before_clip": 0.06640625, "grad_clipped": false, "step_seconds": 0.4792066119989613, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1978, "loss": 0.08329962193965912, "loss_ema": 0.19646471035621055, "loss_mean_50": 0.1907026905566454, "loss_std_50": 0.2462008870413063, "lr": 0.00025, "grad_norm_before_clip": 0.009521484375, "grad_clipped": false, "step_seconds": 1.8278765130016836, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1979, "loss": 0.25488874316215515, "loss_ema": 0.19938591199650776, "loss_mean_50": 0.1940746658295393, "loss_std_50": 0.24590214164406038, "lr": 0.00025, "grad_norm_before_clip": 0.02587890625, "grad_clipped": false, "step_seconds": 1.5973691649996908, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1980, "loss": 0.06120336800813675, "loss_ema": 0.1924767847970892, "loss_mean_50": 0.1937171021848917, "loss_std_50": 0.24608202436669982, "lr": 0.00025, "grad_norm_before_clip": 0.01007080078125, "grad_clipped": false, "step_seconds": 1.8355511489935452, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1981, "loss": 0.08642949163913727, "loss_ema": 0.1871744201391916, "loss_mean_50": 0.19395457588136197, "loss_std_50": 0.24597262177711413, "lr": 0.00025, "grad_norm_before_clip": 0.01019287109375, "grad_clipped": false, "step_seconds": 2.064948863000609, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1982, "loss": 0.050668127834796906, "loss_ema": 0.18034910552397188, "loss_mean_50": 0.19398771323263644, "loss_std_50": 0.24595320366870022, "lr": 0.00025, "grad_norm_before_clip": 0.027099609375, "grad_clipped": false, "step_seconds": 1.1062188089999836, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1983, "loss": 0.3545920252799988, "loss_ema": 0.18906125151177322, "loss_mean_50": 0.19947644643485546, "loss_std_50": 0.24641341488316484, "lr": 0.00025, "grad_norm_before_clip": 0.05029296875, "grad_clipped": false, "step_seconds": 1.8926542940025683, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1984, "loss": 0.15964320302009583, "loss_ema": 0.18759034908718933, "loss_mean_50": 0.19904287956655026, "loss_std_50": 0.24646404360858074, "lr": 0.00025, "grad_norm_before_clip": 0.0164794921875, "grad_clipped": false, "step_seconds": 1.8359998500018264, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1985, "loss": 0.056729961186647415, "loss_ema": 0.18104732969216222, "loss_mean_50": 0.1975483563542366, "loss_std_50": 0.24709510543617563, "lr": 0.00025, "grad_norm_before_clip": 0.0130615234375, "grad_clipped": false, "step_seconds": 2.066459859001043, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1986, "loss": 0.12269512563943863, "loss_ema": 0.17812971948952605, "loss_mean_50": 0.19855167597532272, "loss_std_50": 0.24668694439915984, "lr": 0.00025, "grad_norm_before_clip": 0.014892578125, "grad_clipped": false, "step_seconds": 1.837160101000336, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1987, "loss": 0.7326471209526062, "loss_ema": 0.20585558956268007, "loss_mean_50": 0.2114556124806404, "loss_std_50": 0.257189085134906, "lr": 0.00025, "grad_norm_before_clip": 0.1513671875, "grad_clipped": false, "step_seconds": 2.062553003997891, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1988, "loss": 0.035942260175943375, "loss_ema": 0.19735992309334324, "loss_mean_50": 0.21042944096028804, "loss_std_50": 0.25778427937373316, "lr": 0.00025, "grad_norm_before_clip": 0.0185546875, "grad_clipped": false, "step_seconds": 0.6005811930008349, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1989, "loss": 0.05492820218205452, "loss_ema": 0.19023833704777882, "loss_mean_50": 0.2101124021410942, "loss_std_50": 0.2579655177659005, "lr": 0.00025, "grad_norm_before_clip": 0.0322265625, "grad_clipped": false, "step_seconds": 2.058189652998408, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1990, "loss": 0.09676479548215866, "loss_ema": 0.1855646599694978, "loss_mean_50": 0.20996307715773582, "loss_std_50": 0.25802891779740855, "lr": 0.00025, "grad_norm_before_clip": 0.033203125, "grad_clipped": false, "step_seconds": 0.47910630900150863, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1991, "loss": 0.6007049679756165, "loss_ema": 0.20632167536980373, "loss_mean_50": 0.2196154534816742, "loss_std_50": 0.26338282246088, "lr": 0.00025, "grad_norm_before_clip": 0.09619140625, "grad_clipped": false, "step_seconds": 1.8915469109997503, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1992, "loss": 0.08446479588747025, "loss_ema": 0.20022883139568703, "loss_mean_50": 0.20164658442139627, "loss_std_50": 0.24033440678838466, "lr": 0.00025, "grad_norm_before_clip": 0.06787109375, "grad_clipped": false, "step_seconds": 0.6721666289959103, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1993, "loss": 0.04141445830464363, "loss_ema": 0.19228811274113486, "loss_mean_50": 0.18594505481421947, "loss_std_50": 0.22409588661676372, "lr": 0.00025, "grad_norm_before_clip": 0.01239013671875, "grad_clipped": false, "step_seconds": 2.0663329050003085, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1994, "loss": 0.09584210067987442, "loss_ema": 0.1874658121380718, "loss_mean_50": 0.18655387721955777, "loss_std_50": 0.2238087333042251, "lr": 0.00025, "grad_norm_before_clip": 0.00970458984375, "grad_clipped": false, "step_seconds": 1.570184889002121, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1995, "loss": 0.10009483247995377, "loss_ema": 0.18309726315516592, "loss_mean_50": 0.18663735695183278, "loss_std_50": 0.22377568798972733, "lr": 0.00025, "grad_norm_before_clip": 0.0634765625, "grad_clipped": false, "step_seconds": 0.36326544400071725, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1996, "loss": 0.08682966977357864, "loss_ema": 0.17828388348608656, "loss_mean_50": 0.18766496323049067, "loss_std_50": 0.22319627643715206, "lr": 0.00025, "grad_norm_before_clip": 0.0634765625, "grad_clipped": false, "step_seconds": 0.5992705559983733, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1997, "loss": 0.06501910090446472, "loss_ema": 0.17262064435700547, "loss_mean_50": 0.1875725295394659, "loss_std_50": 0.22324608684487163, "lr": 0.00025, "grad_norm_before_clip": 0.0159912109375, "grad_clipped": false, "step_seconds": 0.5754990940040443, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1998, "loss": 0.05516907572746277, "loss_ema": 0.16674806592552832, "loss_mean_50": 0.18600223444402217, "loss_std_50": 0.22389480368870568, "lr": 0.00025, "grad_norm_before_clip": 0.0849609375, "grad_clipped": false, "step_seconds": 2.06819616699795, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 1999, "loss": 0.3098676800727844, "loss_ema": 0.17390404663289113, "loss_mean_50": 0.1910138427466154, "loss_std_50": 0.22380680553443721, "lr": 0.00025, "grad_norm_before_clip": 0.034912109375, "grad_clipped": false, "step_seconds": 2.14247853699635, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false} {"step": 2000, "loss": 0.9879177212715149, "loss_ema": 0.21460473036482233, "loss_mean_50": 0.19817407242953777, "loss_std_50": 0.24266603072887427, "lr": 0.00025, "grad_norm_before_clip": 0.1552734375, "grad_clipped": false, "step_seconds": 2.4384511540047242, "gpu_peak_gib": 36.505051136016846, "nonfinite_loss": false, "nonfinite_grad": false}