1{"event": "train", "step": 20, "loss": 0.004187169950455427, "grad_norm": 0.36299845576286316, "lr": 0.0003, "ex_per_s": 92.76316243488624, "data_frac": 0.8068907394799553, "mem_mb": 63.6288, "copula_nll": 0.004187169950455427, "elapsed_s": 13.8}2{"event": "train", "step": 40, "loss": 0.0010893481085076928, "grad_norm": 0.27996984124183655, "lr": 0.00029920862421583855, "ex_per_s": 164.49601369695776, "data_frac": 0.716235007711494, "mem_mb": 63.6288, "copula_nll": 0.0010893481085076928, "elapsed_s": 15.6}3{"event": "train", "step": 60, "loss": 0.0010735804680734873, "grad_norm": 0.15422450006008148, "lr": 0.00029684377502086165, "ex_per_s": 222.0566059532456, "data_frac": 0.6456005282881486, "mem_mb": 63.6288, "copula_nll": 0.0010735804680734873, "elapsed_s": 17.3}4{"event": "train", "step": 80, "loss": 0.0018942878814414144, "grad_norm": 0.1834036409854889, "lr": 0.0002929331781096783, "ex_per_s": 269.3422742020049, "data_frac": 0.5879387147154701, "mem_mb": 63.6288, "copula_nll": 0.0018942878814414144, "elapsed_s": 19.1}5{"event": "train", "step": 100, "loss": 0.0005939342081546783, "grad_norm": 0.12928138673305511, "lr": 0.00028752268165557917, "ex_per_s": 308.9403503816227, "data_frac": 0.540070249787059, "mem_mb": 63.6288, "copula_nll": 0.0005939342081546783, "elapsed_s": 20.8}6{"event": "val", "step": 100, "val_nll": 0.0005939342081546783, "best": 0.0005939342081546783, "bad": 0, "mem_mb": 63.6288, "elapsed_s": 20.9}7{"event": "train", "step": 120, "loss": 0.0003655031614471227, "grad_norm": 0.12188612669706345, "lr": 0.00028067571878262454, "ex_per_s": 340.47762803166125, "data_frac": 0.49651395818284555, "mem_mb": 63.6288, "copula_nll": 0.0003655031614471227, "elapsed_s": 22.6}8{"event": "train", "step": 140, "loss": 0.0007432855200022459, "grad_norm": 0.14370176196098328, "lr": 0.00027247256387026185, "ex_per_s": 369.3179085444116, "data_frac": 0.4620956384634324, "mem_mb": 63.6288, "copula_nll": 0.0007432855200022459, "elapsed_s": 24.3}9{"event": "train", "step": 160, "loss": 0.00113543844781816, "grad_norm": 0.16792291402816772, "lr": 0.0002630093914096226, "ex_per_s": 394.63475457597184, "data_frac": 0.43249672425391755, "mem_mb": 63.6288, "copula_nll": 0.00113543844781816, "elapsed_s": 26.0}10{"event": "train", "step": 180, "loss": 0.00017941027181223035, "grad_norm": 0.11429471522569656, "lr": 0.00025239714844554667, "ex_per_s": 416.8555534459972, "data_frac": 0.4065079059277007, "mem_mb": 63.6288, "copula_nll": 0.00017941027181223035, "elapsed_s": 27.7}11{"event": "train", "step": 200, "loss": 0.00023519163369201124, "grad_norm": 0.13828593492507935, "lr": 0.0002407602538239216, "ex_per_s": 436.5824736249117, "data_frac": 0.38355004239324336, "mem_mb": 63.6288, "copula_nll": 0.00023519163369201124, "elapsed_s": 29.4}12{"event": "val", "step": 200, "val_nll": 0.00023519163369201124, "best": 0.00023519163369201124, "bad": 0, "mem_mb": 63.6288, "elapsed_s": 29.5}13{"event": "train", "step": 220, "loss": 0.0003238385543227196, "grad_norm": 0.07590283453464508, "lr": 0.00022823513949447166, "ex_per_s": 452.62572299474215, "data_frac": 0.3618585431765403, "mem_mb": 63.6288, "copula_nll": 0.0003238385543227196, "elapsed_s": 31.1}14{"event": "train", "step": 240, "loss": 0.0011522737331688404, "grad_norm": 0.08838767558336258, "lr": 0.00021496865097088845, "ex_per_s": 468.3770896347677, "data_frac": 0.3435866299812955, "mem_mb": 63.6288, "copula_nll": 0.0011522737331688404, "elapsed_s": 32.8}15{"event": "train", "step": 260, "loss": 0.0005164364702068269, "grad_norm": 0.10975008457899094, "lr": 0.00020111632570144484, "ex_per_s": 482.5341750183092, "data_frac": 0.32707929711833494, "mem_mb": 63.6288, "copula_nll": 0.0005164364702068269, "elapsed_s": 34.5}16{"event": "train", "step": 280, "loss": 0.00033359264489263296, "grad_norm": 0.09256277233362198, "lr": 0.00018684056953462326, "ex_per_s": 495.4477435478485, "data_frac": 0.3121542897027769, "mem_mb": 63.6288, "copula_nll": 0.00033359264489263296, "elapsed_s": 36.2}17{"event": "train", "step": 300, "loss": 6.956507422728464e-05, "grad_norm": 0.0879335030913353, "lr": 0.00017230875265903138, "ex_per_s": 507.2219431358373, "data_frac": 0.2985628291980113, "mem_mb": 63.6288, "copula_nll": 6.956507422728464e-05, "elapsed_s": 37.9}18{"event": "val", "step": 300, "val_nll": 6.956507422728464e-05, "best": 6.956507422728464e-05, "bad": 0, "mem_mb": 63.6288, "elapsed_s": 38.0}19{"event": "train", "step": 320, "loss": 0.0003083588962908834, "grad_norm": 0.07122193276882172, "lr": 0.00015769124734096862, "ex_per_s": 516.5878923469062, "data_frac": 0.28534934576351495, "mem_mb": 63.6288, "copula_nll": 0.0003083588962908834, "elapsed_s": 39.7}20{"event": "train", "step": 340, "loss": 0.00038758019218221307, "grad_norm": 0.07990402728319168, "lr": 0.00014315943046537674, "ex_per_s": 526.4486258454959, "data_frac": 0.27396525592290544, "mem_mb": 63.6288, "copula_nll": 0.00038758019218221307, "elapsed_s": 41.4}21{"event": "train", "step": 360, "loss": 8.071810952969827e-06, "grad_norm": 0.0783398300409317, "lr": 0.0001288836742985552, "ex_per_s": 535.521285306534, "data_frac": 0.2634562014205555, "mem_mb": 63.6288, "copula_nll": 8.071810952969827e-06, "elapsed_s": 43.1}22{"event": "train", "step": 380, "loss": 7.62121781008318e-05, "grad_norm": 0.0547613687813282, "lr": 0.00011503134902911152, "ex_per_s": 543.9226226055251, "data_frac": 0.2537536059068448, "mem_mb": 63.6288, "copula_nll": 7.62121781008318e-05, "elapsed_s": 44.8}23{"event": "train", "step": 400, "loss": -2.4825094442348927e-05, "grad_norm": 0.0668315589427948, "lr": 0.00010176486050552833, "ex_per_s": 551.6987283867904, "data_frac": 0.24474661534299127, "mem_mb": 63.6288, "copula_nll": -2.4825094442348927e-05, "elapsed_s": 46.4}24{"event": "val", "step": 400, "val_nll": -2.4825094442348927e-05, "best": -2.4825094442348927e-05, "bad": 0, "mem_mb": 63.6288, "elapsed_s": 46.6}25{"event": "train", "step": 420, "loss": 1.2148651876486838e-05, "grad_norm": 0.07364778220653534, "lr": 8.92397461760784e-05, "ex_per_s": 557.5524134167764, "data_frac": 0.23581477076516805, "mem_mb": 63.6288, "copula_nll": 1.2148651876486838e-05, "elapsed_s": 48.3}26{"event": "train", "step": 440, "loss": 0.00028504454530775547, "grad_norm": 0.06526606529951096, "lr": 7.760285155445326e-05, "ex_per_s": 564.2602669409349, "data_frac": 0.2280272966413325, "mem_mb": 63.6288, "copula_nll": 0.00028504454530775547, "elapsed_s": 49.9}27{"event": "train", "step": 460, "loss": 0.00025486687081865966, "grad_norm": 0.048734355717897415, "lr": 6.699060859037737e-05, "ex_per_s": 570.51972641162, "data_frac": 0.22074909893062147, "mem_mb": 63.6288, "copula_nll": 0.00025486687081865966, "elapsed_s": 51.6}28{"event": "train", "step": 480, "loss": 0.00011922304111067206, "grad_norm": 0.06257154047489166, "lr": 5.752743612973809e-05, "ex_per_s": 576.3856807797464, "data_frac": 0.21393502730729033, "mem_mb": 63.6288, "copula_nll": 0.00011922304111067206, "elapsed_s": 53.3}29{"event": "train", "step": 500, "loss": 0.00017234732513315976, "grad_norm": 0.04334617033600807, "lr": 4.9324281217375474e-05, "ex_per_s": 581.9023241382351, "data_frac": 0.2075534099192822, "mem_mb": 63.6288, "copula_nll": 0.00017234732513315976, "elapsed_s": 55.0}30{"event": "val", "step": 500, "val_nll": 0.00017234732513315976, "best": -2.4825094442348927e-05, "bad": 1, "mem_mb": 63.6288, "elapsed_s": 55.1}31{"event": "train", "step": 520, "loss": 0.00042805937118828297, "grad_norm": 0.11445201933383942, "lr": 4.247731834442082e-05, "ex_per_s": 586.4602271734927, "data_frac": 0.20133215368308796, "mem_mb": 63.6288, "copula_nll": 0.00042805937118828297, "elapsed_s": 56.8}32{"event": "train", "step": 540, "loss": 0.00013249997573439032, "grad_norm": 0.03931199014186859, "lr": 3.7066821890321684e-05, "ex_per_s": 591.3143112865489, "data_frac": 0.19567453375185886, "mem_mb": 63.6288, "copula_nll": 0.00013249997573439032, "elapsed_s": 58.5}33{"event": "train", "step": 560, "loss": -0.00027832190971821547, "grad_norm": 0.042342785745859146, "lr": 3.3156224979138305e-05, "ex_per_s": 595.881209021954, "data_frac": 0.19032979208928918, "mem_mb": 63.6288, "copula_nll": -0.00027832190971821547, "elapsed_s": 60.2}34{"event": "train", "step": 580, "loss": 0.00021181872580200434, "grad_norm": 0.06567656993865967, "lr": 3.0791375784161455e-05, "ex_per_s": 600.2082404940256, "data_frac": 0.1852814131314618, "mem_mb": 63.6288, "copula_nll": 0.00021181872580200434, "elapsed_s": 61.9}35{"event": "train", "step": 600, "loss": 0.00010817812290042639, "grad_norm": 0.05705700069665909, "lr": 2.9999999999999997e-05, "ex_per_s": 604.3051346398715, "data_frac": 0.18050654995212742, "mem_mb": 63.6288, "copula_nll": 0.00010817812290042639, "elapsed_s": 63.6}36{"event": "val", "step": 600, "val_nll": 0.00010817812290042639, "best": -2.4825094442348927e-05, "bad": 2, "mem_mb": 63.6288, "elapsed_s": 63.6}37{"event": "done", "steps": 600, "best_val": -2.4825094442348927e-05, "wall_s": 63.63427805900574, "ex_per_s": 603.4483484576203, "peak_mem_mb": 63.6288, "n_params": 3456584, "device": "mps", "data_time_frac": 0.18025062705996636, "elapsed_s": 63.6}38