1{"event": "train", "step": 20, "loss": -5.731091499328613, "grad_norm": 0.8772358298301697, "lr": 0.00012599999999999997, "ex_per_s": 783.3876515424337, "data_frac": 0.8889249628932253, "mem_mb": 14.961152, "nll": -5.731091499328613, "elapsed_s": 1.6}2{"event": "train", "step": 40, "loss": -6.012518882751465, "grad_norm": 0.44088655710220337, "lr": 0.00024599999999999996, "ex_per_s": 1499.7956826227985, "data_frac": 0.8618830956409704, "mem_mb": 14.961152, "nll": -6.012518882751465, "elapsed_s": 1.7}3{"event": "train", "step": 60, "loss": -5.9410223960876465, "grad_norm": 0.2476016730070114, "lr": 0.0002997798290300283, "ex_per_s": 2128.6935283126822, "data_frac": 0.8384986498511869, "mem_mb": 14.961152, "nll": -5.9410223960876465, "elapsed_s": 1.8}4{"event": "train", "step": 80, "loss": -5.889024257659912, "grad_norm": 0.12048814445734024, "lr": 0.0002980227678394385, "ex_per_s": 2674.9421074593774, "data_frac": 0.8193758213286997, "mem_mb": 14.961152, "nll": -5.889024257659912, "elapsed_s": 1.9}5{"event": "train", "step": 100, "loss": -5.937258243560791, "grad_norm": 0.0761365219950676, "lr": 0.00029453155143795715, "ex_per_s": 3184.0682866341003, "data_frac": 0.8007222752500688, "mem_mb": 14.961152, "nll": -5.937258243560791, "elapsed_s": 2.0}6{"event": "val", "step": 100, "val_nll": -6.055456592192292, "best": -6.055456592192292, "bad": 0, "mem_mb": 14.961152, "elapsed_s": 26.6}7{"event": "train", "step": 120, "loss": -5.898717880249023, "grad_norm": 0.10158388316631317, "lr": 0.00028935169317050475, "ex_per_s": 288.31675080582517, "data_frac": 0.06082074856002979, "mem_mb": 14.961152, "nll": -5.898717880249023, "elapsed_s": 26.6}8{"event": "train", "step": 140, "loss": -5.784311294555664, "grad_norm": 0.3645849823951721, "lr": 0.0002825507204111699, "ex_per_s": 335.35760067303244, "data_frac": 0.06194142316779178, "mem_mb": 14.961152, "nll": -5.784311294555664, "elapsed_s": 26.7}9{"event": "train", "step": 160, "loss": -5.826564788818359, "grad_norm": 0.3471275269985199, "lr": 0.00027421729424061787, "ex_per_s": 381.6315638629743, "data_frac": 0.06412483596715171, "mem_mb": 14.961152, "nll": -5.826564788818359, "elapsed_s": 26.8}10{"event": "train", "step": 180, "loss": -5.774465560913086, "grad_norm": 0.21533596515655518, "lr": 0.00026446005361331463, "ex_per_s": 427.8159680555978, "data_frac": 0.06542159138188999, "mem_mb": 14.961152, "nll": -5.774465560913086, "elapsed_s": 26.9}11{"event": "train", "step": 200, "loss": -5.813832759857178, "grad_norm": 0.8862797021865845, "lr": 0.0002534061990826135, "ex_per_s": 473.66117826924165, "data_frac": 0.06672882736718183, "mem_mb": 14.961152, "nll": -5.813832759857178, "elapsed_s": 27.0}12{"event": "val", "step": 200, "val_nll": -6.063561781314903, "best": -6.063561781314903, "bad": 0, "mem_mb": 14.961152, "elapsed_s": 51.6}13{"event": "train", "step": 220, "loss": -5.934475898742676, "grad_norm": 0.6205616593360901, "lr": 0.00024119983454701384, "ex_per_s": 272.6577307521014, "data_frac": 0.035112188652049, "mem_mb": 14.961152, "nll": -5.934475898742676, "elapsed_s": 51.6}14{"event": "train", "step": 240, "loss": -5.9126458168029785, "grad_norm": 0.4978834092617035, "lr": 0.00022800008863546595, "ex_per_s": 296.8725333270123, "data_frac": 0.03594098275634613, "mem_mb": 14.961152, "nll": -5.9126458168029785, "elapsed_s": 51.7}15{"event": "train", "step": 260, "loss": -6.046988010406494, "grad_norm": 0.3671414852142334, "lr": 0.0002139790402223365, "ex_per_s": 321.0091949600026, "data_frac": 0.036711529239587926, "mem_mb": 14.961152, "nll": -6.046988010406494, "elapsed_s": 51.8}16{"event": "train", "step": 280, "loss": -5.838945388793945, "grad_norm": 0.1125865951180458, "lr": 0.00019931947511612076, "ex_per_s": 345.0525923146492, "data_frac": 0.037650901433205014, "mem_mb": 14.961152, "nll": -5.838945388793945, "elapsed_s": 51.9}17{"event": "train", "step": 300, "loss": -5.932683944702148, "grad_norm": 0.2066040188074112, "lr": 0.00018421250316689352, "ex_per_s": 369.00885852074424, "data_frac": 0.03851757598022707, "mem_mb": 14.961152, "nll": -5.932683944702148, "elapsed_s": 52.0}18{"event": "val", "step": 300, "val_nll": -6.0651017761790404, "best": -6.0651017761790404, "bad": 0, "mem_mb": 14.961152, "elapsed_s": 76.6}19{"event": "train", "step": 320, "loss": -5.878216743469238, "grad_norm": 0.1710980236530304, "lr": 0.00016885506685714895, "ex_per_s": 266.89619723837575, "data_frac": 0.026951181256453835, "mem_mb": 14.961152, "nll": -5.878216743469238, "elapsed_s": 76.7}20{"event": "train", "step": 340, "loss": -5.946746826171875, "grad_norm": 0.36603567004203796, "lr": 0.00015344737385535666, "ex_per_s": 283.21451482884555, "data_frac": 0.027476455340682344, "mem_mb": 14.961152, "nll": -5.946746826171875, "elapsed_s": 76.8}21{"event": "train", "step": 360, "loss": -5.933969974517822, "grad_norm": 0.14381878077983856, "lr": 0.0001381902870028244, "ex_per_s": 299.4951439903658, "data_frac": 0.027987531088925124, "mem_mb": 14.961152, "nll": -5.933969974517822, "elapsed_s": 76.9}22{"event": "train", "step": 380, "loss": -5.939368724822998, "grad_norm": 0.24535012245178223, "lr": 0.00012328270575938212, "ex_per_s": 315.730270156329, "data_frac": 0.02852378309674674, "mem_mb": 14.961152, "nll": -5.939368724822998, "elapsed_s": 77.0}23{"event": "train", "step": 400, "loss": -5.9516377449035645, "grad_norm": 0.1344529688358307, "lr": 0.00010891897324474534, "ex_per_s": 331.83802216740605, "data_frac": 0.029297609466424533, "mem_mb": 14.961152, "nll": -5.9516377449035645, "elapsed_s": 77.1}24{"event": "val", "step": 400, "val_nll": -6.0654186238547725, "best": -6.0654186238547725, "bad": 0, "mem_mb": 14.961152, "elapsed_s": 101.7}25{"event": "train", "step": 420, "loss": -5.9801716804504395, "grad_norm": 0.2856941521167755, "lr": 9.528634267874014e-05, "ex_per_s": 264.17471783728587, "data_frac": 0.022307234749681215, "mem_mb": 14.961152, "nll": -5.9801716804504395, "elapsed_s": 101.8}26{"event": "train", "step": 440, "loss": -5.919494152069092, "grad_norm": 0.1811201125383377, "lr": 8.256253624921584e-05, "ex_per_s": 276.529741082351, "data_frac": 0.022648413035444234, "mem_mb": 14.961152, "nll": -5.919494152069092, "elapsed_s": 101.8}27{"event": "train", "step": 460, "loss": -5.88907527923584, "grad_norm": 0.39054781198501587, "lr": 7.09134282315354e-05, "ex_per_s": 288.82315676841654, "data_frac": 0.02310671849231397, "mem_mb": 14.961152, "nll": -5.88907527923584, "elapsed_s": 101.9}28{"event": "train", "step": 480, "loss": -5.921980857849121, "grad_norm": 0.3483763039112091, "lr": 6.049088256372386e-05, "ex_per_s": 301.0318951503883, "data_frac": 0.02369368246991016, "mem_mb": 14.961152, "nll": -5.921980857849121, "elapsed_s": 102.1}29{"event": "train", "step": 500, "loss": -5.969280242919922, "grad_norm": 0.3606049120426178, "lr": 5.143077306779053e-05, "ex_per_s": 313.2776563241587, "data_frac": 0.02407478009622936, "mem_mb": 14.961152, "nll": -5.969280242919922, "elapsed_s": 102.1}30{"event": "val", "step": 500, "val_nll": -6.0658178044129025, "best": -6.0658178044129025, "bad": 0, "mem_mb": 14.961152, "elapsed_s": 126.7}31{"event": "train", "step": 520, "loss": -6.010374546051025, "grad_norm": 0.1669560968875885, "lr": 4.385121212666663e-05, "ex_per_s": 262.43418770704704, "data_frac": 0.019464527415162688, "mem_mb": 14.961152, "nll": -6.010374546051025, "elapsed_s": 126.8}32{"event": "train", "step": 540, "loss": -5.9557647705078125, "grad_norm": 0.1022234559059143, "lr": 3.785101090866418e-05, "ex_per_s": 272.3113466024373, "data_frac": 0.019819342911721008, "mem_mb": 14.961152, "nll": -5.9557647705078125, "elapsed_s": 126.9}33{"event": "train", "step": 560, "loss": -5.852337837219238, "grad_norm": 0.41757574677467346, "lr": 3.3508391212784984e-05, "ex_per_s": 282.1838401930878, "data_frac": 0.020203229063328113, "mem_mb": 14.961152, "nll": -5.852337837219238, "elapsed_s": 127.0}34{"event": "train", "step": 580, "loss": -5.82141637802124, "grad_norm": 0.3278888165950775, "lr": 3.087996572794572e-05, "ex_per_s": 292.0349766794339, "data_frac": 0.020588826376247166, "mem_mb": 14.961152, "nll": -5.82141637802124, "elapsed_s": 127.1}35{"event": "train", "step": 600, "loss": -5.887289524078369, "grad_norm": 0.3881126344203949, "lr": 2.9999999999999997e-05, "ex_per_s": 301.8779065792281, "data_frac": 0.020898655156931196, "mem_mb": 14.961152, "nll": -5.887289524078369, "elapsed_s": 127.2}36{"event": "val", "step": 600, "val_nll": -6.066584629400399, "best": -6.066584629400399, "bad": 0, "mem_mb": 14.961152, "elapsed_s": 151.7}37{"event": "done", "steps": 600, "best_val": -6.066584629400399, "wall_s": 151.7184603214264, "ex_per_s": 253.10038026122106, "peak_mem_mb": 14.961152, "n_params": 301827, "device": "mps", "data_time_frac": 0.017521843937191844, "elapsed_s": 151.7}38