1{"event": "train", "step": 20, "loss": -5.800985336303711, "grad_norm": 0.6310738325119019, "lr": 0.00012599999999999997, "ex_per_s": 561.7820117468924, "data_frac": 0.6356217764323392, "mem_mb": 86.469632, "nll": -5.800985336303711, "elapsed_s": 2.3}2{"event": "train", "step": 40, "loss": -6.0472283363342285, "grad_norm": 0.29100605845451355, "lr": 0.00024599999999999996, "ex_per_s": 904.6051716934734, "data_frac": 0.5147867318622017, "mem_mb": 86.469632, "nll": -6.0472283363342285, "elapsed_s": 2.8}3{"event": "train", "step": 60, "loss": -5.957995414733887, "grad_norm": 1.0624996423721313, "lr": 0.0002997798290300283, "ex_per_s": 1135.623380457406, "data_frac": 0.4333911669844615, "mem_mb": 86.469632, "nll": -5.957995414733887, "elapsed_s": 3.4}4{"event": "train", "step": 80, "loss": -5.887150764465332, "grad_norm": 0.9817919731140137, "lr": 0.0002980227678394385, "ex_per_s": 1292.107702735348, "data_frac": 0.3772051211777614, "mem_mb": 86.469632, "nll": -5.887150764465332, "elapsed_s": 4.0}5{"event": "train", "step": 100, "loss": -5.931876182556152, "grad_norm": 0.8043496012687683, "lr": 0.00029453155143795715, "ex_per_s": 1418.0283989768934, "data_frac": 0.33305367421318405, "mem_mb": 86.469632, "nll": -5.931876182556152, "elapsed_s": 4.5}6{"event": "val", "step": 100, "val_nll": -6.060500350732755, "best": -6.060500350732755, "bad": 0, "mem_mb": 86.469632, "elapsed_s": 29.5}7{"event": "train", "step": 120, "loss": -5.899375915527344, "grad_norm": 0.8422088027000427, "lr": 0.00028935169317050475, "ex_per_s": 255.23623122073988, "data_frac": 0.05025232190276488, "mem_mb": 86.469632, "nll": -5.899375915527344, "elapsed_s": 30.1}8{"event": "train", "step": 140, "loss": -5.784610271453857, "grad_norm": 0.13086478412151337, "lr": 0.0002825507204111699, "ex_per_s": 292.4115669882624, "data_frac": 0.04962752950526664, "mem_mb": 86.469632, "nll": -5.784610271453857, "elapsed_s": 30.7}9{"event": "train", "step": 160, "loss": -5.8238630294799805, "grad_norm": 0.24100030958652496, "lr": 0.00027421729424061787, "ex_per_s": 327.9805679464724, "data_frac": 0.04964293640588582, "mem_mb": 86.469632, "nll": -5.8238630294799805, "elapsed_s": 31.2}10{"event": "train", "step": 180, "loss": -5.77768087387085, "grad_norm": 0.5538883805274963, "lr": 0.00026446005361331463, "ex_per_s": 362.58055722423404, "data_frac": 0.04904739349203437, "mem_mb": 86.469632, "nll": -5.77768087387085, "elapsed_s": 31.8}11{"event": "train", "step": 200, "loss": -5.8120832443237305, "grad_norm": 0.9348663091659546, "lr": 0.0002534061990826135, "ex_per_s": 396.00168428958574, "data_frac": 0.04847480343525327, "mem_mb": 86.469632, "nll": -5.8120832443237305, "elapsed_s": 32.3}12{"event": "val", "step": 200, "val_nll": -6.061717885426913, "best": -6.061717885426913, "bad": 0, "mem_mb": 86.469632, "elapsed_s": 57.3}13{"event": "train", "step": 220, "loss": -5.93495512008667, "grad_norm": 0.24960874021053314, "lr": 0.00024119983454701384, "ex_per_s": 243.1372034885646, "data_frac": 0.027205146634531628, "mem_mb": 86.469632, "nll": -5.93495512008667, "elapsed_s": 57.9}14{"event": "train", "step": 240, "loss": -5.906928539276123, "grad_norm": 1.063704013824463, "lr": 0.00022800008863546595, "ex_per_s": 262.6199987477303, "data_frac": 0.027436679665487445, "mem_mb": 86.469632, "nll": -5.906928539276123, "elapsed_s": 58.5}15{"event": "train", "step": 260, "loss": -6.0408101081848145, "grad_norm": 0.6488344669342041, "lr": 0.0002139790402223365, "ex_per_s": 281.8461898603646, "data_frac": 0.027318706366990204, "mem_mb": 86.469632, "nll": -6.0408101081848145, "elapsed_s": 59.1}16{"event": "train", "step": 280, "loss": -5.8299031257629395, "grad_norm": 0.298898845911026, "lr": 0.00019931947511612076, "ex_per_s": 300.7205536582715, "data_frac": 0.02720951490637674, "mem_mb": 86.469632, "nll": -5.8299031257629395, "elapsed_s": 59.6}17{"event": "train", "step": 300, "loss": -5.930388450622559, "grad_norm": 0.30565884709358215, "lr": 0.00018421250316689352, "ex_per_s": 319.218686315573, "data_frac": 0.027092517437594114, "mem_mb": 86.469632, "nll": -5.930388450622559, "elapsed_s": 60.2}18{"event": "val", "step": 300, "val_nll": -6.063746403830828, "best": -6.063746403830828, "bad": 0, "mem_mb": 86.469632, "elapsed_s": 85.1}19{"event": "train", "step": 320, "loss": -5.870319843292236, "grad_norm": 0.7388385534286499, "lr": 0.00016885506685714895, "ex_per_s": 238.88170762243155, "data_frac": 0.019355928464945833, "mem_mb": 86.469632, "nll": -5.870319843292236, "elapsed_s": 85.7}20{"event": "train", "step": 340, "loss": -5.9455695152282715, "grad_norm": 0.46732261776924133, "lr": 0.00015344737385535666, "ex_per_s": 252.19004111335144, "data_frac": 0.019333308280503638, "mem_mb": 86.469632, "nll": -5.9455695152282715, "elapsed_s": 86.3}21{"event": "train", "step": 360, "loss": -5.928246974945068, "grad_norm": 0.5238118767738342, "lr": 0.0001381902870028244, "ex_per_s": 265.33311925856765, "data_frac": 0.019310102333383095, "mem_mb": 86.469632, "nll": -5.928246974945068, "elapsed_s": 86.8}22{"event": "train", "step": 380, "loss": -5.932854175567627, "grad_norm": 0.2111808806657791, "lr": 0.00012328270575938212, "ex_per_s": 278.28337214000436, "data_frac": 0.019276522406841842, "mem_mb": 86.469632, "nll": -5.932854175567627, "elapsed_s": 87.4}23{"event": "train", "step": 400, "loss": -5.946496486663818, "grad_norm": 0.557999312877655, "lr": 0.00010891897324474534, "ex_per_s": 291.02514102602987, "data_frac": 0.019491249270171825, "mem_mb": 86.469632, "nll": -5.946496486663818, "elapsed_s": 88.0}24{"event": "val", "step": 400, "val_nll": -6.063318440447752, "best": -6.063746403830828, "bad": 1, "mem_mb": 86.469632, "elapsed_s": 112.9}25{"event": "train", "step": 420, "loss": -5.969577312469482, "grad_norm": 0.8200492262840271, "lr": 9.528634267874014e-05, "ex_per_s": 236.78768182513096, "data_frac": 0.015179030587109682, "mem_mb": 86.469632, "nll": -5.969577312469482, "elapsed_s": 113.5}26{"event": "train", "step": 440, "loss": -5.921257495880127, "grad_norm": 0.15645407140254974, "lr": 8.256253624921584e-05, "ex_per_s": 246.86335605342077, "data_frac": 0.015181393210222735, "mem_mb": 86.469632, "nll": -5.921257495880127, "elapsed_s": 114.1}27{"event": "train", "step": 460, "loss": -5.88503885269165, "grad_norm": 0.36901313066482544, "lr": 7.09134282315354e-05, "ex_per_s": 256.82340728918615, "data_frac": 0.01517364957353039, "mem_mb": 86.469632, "nll": -5.88503885269165, "elapsed_s": 114.6}28{"event": "train", "step": 480, "loss": -5.912304878234863, "grad_norm": 0.39294031262397766, "lr": 6.049088256372386e-05, "ex_per_s": 266.6430950030507, "data_frac": 0.0153695657462188, "mem_mb": 86.469632, "nll": -5.912304878234863, "elapsed_s": 115.2}29{"event": "train", "step": 500, "loss": -5.958542346954346, "grad_norm": 0.7920178174972534, "lr": 5.143077306779053e-05, "ex_per_s": 276.4188867770969, "data_frac": 0.01537698659100627, "mem_mb": 86.469632, "nll": -5.958542346954346, "elapsed_s": 115.8}30{"event": "val", "step": 500, "val_nll": -6.064731540310882, "best": -6.064731540310882, "bad": 0, "mem_mb": 86.469632, "elapsed_s": 140.7}31{"event": "train", "step": 520, "loss": -6.000699520111084, "grad_norm": 0.12713053822517395, "lr": 4.385121212666663e-05, "ex_per_s": 235.48505522393708, "data_frac": 0.01265902055512085, "mem_mb": 86.469632, "nll": -6.000699520111084, "elapsed_s": 141.3}32{"event": "train", "step": 540, "loss": -5.948925018310547, "grad_norm": 0.4662758409976959, "lr": 3.785101090866418e-05, "ex_per_s": 243.54122819111126, "data_frac": 0.01281221301230926, "mem_mb": 86.469632, "nll": -5.948925018310547, "elapsed_s": 141.9}33{"event": "train", "step": 560, "loss": -5.8442158699035645, "grad_norm": 0.6316394209861755, "lr": 3.3508391212784984e-05, "ex_per_s": 251.58170426879605, "data_frac": 0.012824264362089636, "mem_mb": 86.469632, "nll": -5.8442158699035645, "elapsed_s": 142.5}34{"event": "train", "step": 580, "loss": -5.817366123199463, "grad_norm": 0.46420392394065857, "lr": 3.087996572794572e-05, "ex_per_s": 259.5605353250752, "data_frac": 0.012835934803148525, "mem_mb": 86.469632, "nll": -5.817366123199463, "elapsed_s": 143.0}35{"event": "train", "step": 600, "loss": -5.883575916290283, "grad_norm": 0.701759934425354, "lr": 2.9999999999999997e-05, "ex_per_s": 267.47750093326175, "data_frac": 0.012846381434265428, "mem_mb": 86.469632, "nll": -5.883575916290283, "elapsed_s": 143.6}36{"event": "val", "step": 600, "val_nll": -6.065922264944626, "best": -6.065922264944626, "bad": 0, "mem_mb": 86.469632, "elapsed_s": 168.6}37{"event": "done", "steps": 600, "best_val": -6.065922264944626, "wall_s": 168.56347799301147, "ex_per_s": 227.80735457767452, "peak_mem_mb": 86.469632, "n_params": 3200259, "device": "mps", "data_time_frac": 0.010941107794954125, "elapsed_s": 168.6}38