{"event": "train", "step": 20, "loss": -5.7571234703063965, "grad_norm": 1.796495795249939, "lr": 0.00012599999999999997, "ex_per_s": 501.1061165793347, "data_frac": 0.5582618058699889, "mem_mb": 35.623936, "nll": -5.7571234703063965, "elapsed_s": 2.6} {"event": "train", "step": 40, "loss": -6.029896259307861, "grad_norm": 0.7294140458106995, "lr": 0.00024599999999999996, "ex_per_s": 925.6350348653489, "data_frac": 0.5187394784990693, "mem_mb": 35.623936, "nll": -6.029896259307861, "elapsed_s": 2.8} {"event": "train", "step": 60, "loss": -5.951688766479492, "grad_norm": 0.16268973052501678, "lr": 0.0002997798290300283, "ex_per_s": 1290.0933103511748, "data_frac": 0.4849182588251465, "mem_mb": 35.623936, "nll": -5.951688766479492, "elapsed_s": 3.0} {"event": "train", "step": 80, "loss": -5.8884596824646, "grad_norm": 0.2002396434545517, "lr": 0.0002980227678394385, "ex_per_s": 1592.7776481047879, "data_frac": 0.4590140516321997, "mem_mb": 35.623936, "nll": -5.8884596824646, "elapsed_s": 3.2} {"event": "train", "step": 100, "loss": -5.931579113006592, "grad_norm": 0.1511886864900589, "lr": 0.00029453155143795715, "ex_per_s": 1867.1379616704394, "data_frac": 0.43313638928051024, "mem_mb": 35.623936, "nll": -5.931579113006592, "elapsed_s": 3.4} {"event": "val", "step": 100, "val_nll": -6.0599547057741, "best": -6.0599547057741, "bad": 0, "mem_mb": 35.623936, "elapsed_s": 28.1} {"event": "train", "step": 120, "loss": -5.895289421081543, "grad_norm": 0.4282356798648834, "lr": 0.00028935169317050475, "ex_per_s": 270.72425814390994, "data_frac": 0.05264313099861028, "mem_mb": 35.361792, "nll": -5.895289421081543, "elapsed_s": 28.4} {"event": "train", "step": 140, "loss": -5.785079479217529, "grad_norm": 0.5710427761077881, "lr": 0.0002825507204111699, "ex_per_s": 313.5239931347181, "data_frac": 0.05255672801270622, "mem_mb": 35.361792, "nll": -5.785079479217529, "elapsed_s": 28.6} {"event": "train", "step": 160, "loss": -5.828155994415283, "grad_norm": 0.5868253707885742, "lr": 0.00027421729424061787, "ex_per_s": 355.34713571687445, "data_frac": 0.05324116907036634, "mem_mb": 35.361792, "nll": -5.828155994415283, "elapsed_s": 28.8} {"event": "train", "step": 180, "loss": -5.774509906768799, "grad_norm": 0.12268310785293579, "lr": 0.00026446005361331463, "ex_per_s": 396.8578750647277, "data_frac": 0.05310988720234889, "mem_mb": 35.361792, "nll": -5.774509906768799, "elapsed_s": 29.0} {"event": "train", "step": 200, "loss": -5.815005779266357, "grad_norm": 1.4674038887023926, "lr": 0.0002534061990826135, "ex_per_s": 437.7700905058994, "data_frac": 0.053021812446637, "mem_mb": 35.361792, "nll": -5.815005779266357, "elapsed_s": 29.2} {"event": "val", "step": 200, "val_nll": -6.065618567511276, "best": -6.065618567511276, "bad": 0, "mem_mb": 35.361792, "elapsed_s": 53.8} {"event": "train", "step": 220, "loss": -5.933737277984619, "grad_norm": 0.8920566439628601, "lr": 0.00024119983454701384, "ex_per_s": 260.222978481409, "data_frac": 0.028832011235945896, "mem_mb": 35.361792, "nll": -5.933737277984619, "elapsed_s": 54.1} {"event": "train", "step": 240, "loss": -5.904402732849121, "grad_norm": 0.5953885912895203, "lr": 0.00022800008863546595, "ex_per_s": 282.6565525583383, "data_frac": 0.029272332907047005, "mem_mb": 35.361792, "nll": -5.904402732849121, "elapsed_s": 54.3} {"event": "train", "step": 260, "loss": -6.040090084075928, "grad_norm": 0.2880513072013855, "lr": 0.0002139790402223365, "ex_per_s": 305.0236705852221, "data_frac": 0.029305859457386996, "mem_mb": 35.361792, "nll": -6.040090084075928, "elapsed_s": 54.6} {"event": "train", "step": 280, "loss": -5.834377765655518, "grad_norm": 0.5397225618362427, "lr": 0.00019931947511612076, "ex_per_s": 327.21352073236517, "data_frac": 0.02937800973701689, "mem_mb": 35.361792, "nll": -5.834377765655518, "elapsed_s": 54.8} {"event": "train", "step": 300, "loss": -5.931962966918945, "grad_norm": 0.2925002872943878, "lr": 0.00018421250316689352, "ex_per_s": 349.23795021399775, "data_frac": 0.02940846077125432, "mem_mb": 35.361792, "nll": -5.931962966918945, "elapsed_s": 55.0} {"event": "val", "step": 300, "val_nll": -6.0662493399464035, "best": -6.0662493399464035, "bad": 0, "mem_mb": 35.361792, "elapsed_s": 79.6} {"event": "train", "step": 320, "loss": -5.875869274139404, "grad_norm": 0.16882193088531494, "lr": 0.00016885506685714895, "ex_per_s": 256.5139156574031, "data_frac": 0.020641870789642146, "mem_mb": 35.361792, "nll": -5.875869274139404, "elapsed_s": 79.8} {"event": "train", "step": 340, "loss": -5.946311950683594, "grad_norm": 0.4773854613304138, "lr": 0.00015344737385535666, "ex_per_s": 271.82613607198704, "data_frac": 0.02069456146333605, "mem_mb": 35.361792, "nll": -5.946311950683594, "elapsed_s": 80.1} {"event": "train", "step": 360, "loss": -5.935012340545654, "grad_norm": 0.16489626467227936, "lr": 0.0001381902870028244, "ex_per_s": 287.06155674441095, "data_frac": 0.020745273470043454, "mem_mb": 35.361792, "nll": -5.935012340545654, "elapsed_s": 80.3} {"event": "train", "step": 380, "loss": -5.9362568855285645, "grad_norm": 0.258220911026001, "lr": 0.00012328270575938212, "ex_per_s": 302.209213254781, "data_frac": 0.020784505524516106, "mem_mb": 35.361792, "nll": -5.9362568855285645, "elapsed_s": 80.5} {"event": "train", "step": 400, "loss": -5.949446678161621, "grad_norm": 0.47040072083473206, "lr": 0.00010891897324474534, "ex_per_s": 317.1807143488112, "data_frac": 0.021152733913865948, "mem_mb": 35.361792, "nll": -5.949446678161621, "elapsed_s": 80.7} {"event": "val", "step": 400, "val_nll": -6.067097822272813, "best": -6.067097822272813, "bad": 0, "mem_mb": 35.361792, "elapsed_s": 105.3} {"event": "train", "step": 420, "loss": -5.968071460723877, "grad_norm": 1.1659741401672363, "lr": 9.528634267874014e-05, "ex_per_s": 254.51243734645908, "data_frac": 0.016253218309514542, "mem_mb": 35.361792, "nll": -5.968071460723877, "elapsed_s": 105.6} {"event": "train", "step": 440, "loss": -5.919938087463379, "grad_norm": 0.23034586012363434, "lr": 8.256253624921584e-05, "ex_per_s": 266.09056794555465, "data_frac": 0.01631589334066677, "mem_mb": 35.361792, "nll": -5.919938087463379, "elapsed_s": 105.8} {"event": "train", "step": 460, "loss": -5.883735656738281, "grad_norm": 1.193274736404419, "lr": 7.09134282315354e-05, "ex_per_s": 277.6257937853613, "data_frac": 0.016353599889744976, "mem_mb": 35.361792, "nll": -5.883735656738281, "elapsed_s": 106.0} {"event": "train", "step": 480, "loss": -5.9088263511657715, "grad_norm": 0.9350354671478271, "lr": 6.049088256372386e-05, "ex_per_s": 289.05458619483943, "data_frac": 0.016631010683792354, "mem_mb": 35.361792, "nll": -5.9088263511657715, "elapsed_s": 106.3} {"event": "train", "step": 500, "loss": -5.966639995574951, "grad_norm": 0.7082943916320801, "lr": 5.143077306779053e-05, "ex_per_s": 300.50239199430524, "data_frac": 0.01667433204779628, "mem_mb": 35.361792, "nll": -5.966639995574951, "elapsed_s": 106.5} {"event": "val", "step": 500, "val_nll": -6.068014428296788, "best": -6.068014428296788, "bad": 0, "mem_mb": 35.361792, "elapsed_s": 131.0} {"event": "train", "step": 520, "loss": -6.001988887786865, "grad_norm": 0.21180765330791473, "lr": 4.385121212666663e-05, "ex_per_s": 253.53912180350645, "data_frac": 0.013616853262588512, "mem_mb": 35.361792, "nll": -6.001988887786865, "elapsed_s": 131.3} {"event": "train", "step": 540, "loss": -5.955315113067627, "grad_norm": 0.25246500968933105, "lr": 3.785101090866418e-05, "ex_per_s": 262.8181850143476, "data_frac": 0.013851756322564128, "mem_mb": 35.361792, "nll": -5.955315113067627, "elapsed_s": 131.5} {"event": "train", "step": 560, "loss": -5.850186347961426, "grad_norm": 0.9502299427986145, "lr": 3.3508391212784984e-05, "ex_per_s": 272.10265835487974, "data_frac": 0.01392078568547058, "mem_mb": 35.361792, "nll": -5.850186347961426, "elapsed_s": 131.7} {"event": "train", "step": 580, "loss": -5.817522048950195, "grad_norm": 0.6963861584663391, "lr": 3.087996572794572e-05, "ex_per_s": 281.362061042815, "data_frac": 0.013982479197707915, "mem_mb": 35.361792, "nll": -5.817522048950195, "elapsed_s": 131.9} {"event": "train", "step": 600, "loss": -5.8791584968566895, "grad_norm": 1.0272698402404785, "lr": 2.9999999999999997e-05, "ex_per_s": 290.5890644813262, "data_frac": 0.014048085798736041, "mem_mb": 35.361792, "nll": -5.8791584968566895, "elapsed_s": 132.2} {"event": "val", "step": 600, "val_nll": -6.068539780236767, "best": -6.068539780236767, "bad": 0, "mem_mb": 35.361792, "elapsed_s": 157.8} {"event": "done", "steps": 600, "best_val": -6.068539780236767, "wall_s": 157.81467294692993, "ex_per_s": 243.32338231257614, "peak_mem_mb": 35.361792, "n_params": 231459, "device": "mps", "data_time_frac": 0.011763098372841153, "elapsed_s": 157.8}