SPB Git

spb/wp41_uqo

Public
152.5 KB · 614 lines
1{"event": "train", "step": 20, "loss": -6.126709461212158, "grad_norm": 0.29296189546585083, "lr": 1.5749999999999997e-05, "ex_per_s": 99.07277912707447, "data_frac": 0.941399068322179, "mem_mb": 87.387136, "nll": -6.126709461212158, "elapsed_s": 12.9}2{"event": "train", "step": 40, "loss": -5.893186569213867, "grad_norm": 1.0338581800460815, "lr": 3.0749999999999995e-05, "ex_per_s": 187.54857724172953, "data_frac": 0.9025527900646199, "mem_mb": 87.387136, "nll": -5.893186569213867, "elapsed_s": 13.7}3{"event": "train", "step": 60, "loss": -5.976109981536865, "grad_norm": 0.7631037831306458, "lr": 4.5749999999999994e-05, "ex_per_s": 267.5714321693343, "data_frac": 0.8595441506649495, "mem_mb": 87.387136, "nll": -5.976109981536865, "elapsed_s": 14.4}4{"event": "train", "step": 80, "loss": -6.07813835144043, "grad_norm": 0.19182723760604858, "lr": 6.075e-05, "ex_per_s": 342.5049518060364, "data_frac": 0.8261925092488206, "mem_mb": 87.387136, "nll": -6.07813835144043, "elapsed_s": 15.0}5{"event": "train", "step": 100, "loss": -5.948204517364502, "grad_norm": 0.22057855129241943, "lr": 7.575e-05, "ex_per_s": 412.58569993780526, "data_frac": 0.796841554126776, "mem_mb": 87.387136, "nll": -5.948204517364502, "elapsed_s": 15.5}6{"event": "train", "step": 120, "loss": -5.904655933380127, "grad_norm": 1.078939437866211, "lr": 9.074999999999998e-05, "ex_per_s": 477.8375827887266, "data_frac": 0.7695257382303832, "mem_mb": 87.387136, "nll": -5.904655933380127, "elapsed_s": 16.1}7{"event": "train", "step": 140, "loss": -5.830500602722168, "grad_norm": 0.1425323188304901, "lr": 0.00010574999999999998, "ex_per_s": 538.6897440518165, "data_frac": 0.7442727768896044, "mem_mb": 87.387136, "nll": -5.830500602722168, "elapsed_s": 16.6}8{"event": "train", "step": 160, "loss": -5.735442161560059, "grad_norm": 0.6544366478919983, "lr": 0.00012075, "ex_per_s": 595.7380114857222, "data_frac": 0.7208360408179214, "mem_mb": 87.387136, "nll": -5.735442161560059, "elapsed_s": 17.2}9{"event": "train", "step": 180, "loss": -5.972235679626465, "grad_norm": 1.2221903800964355, "lr": 0.00013575, "ex_per_s": 649.3369698137969, "data_frac": 0.6989913411133459, "mem_mb": 87.387136, "nll": -5.972235679626465, "elapsed_s": 17.8}10{"event": "train", "step": 200, "loss": -5.966368198394775, "grad_norm": 1.150103211402893, "lr": 0.00015074999999999998, "ex_per_s": 699.7021184932388, "data_frac": 0.6784860607719979, "mem_mb": 87.387136, "nll": -5.966368198394775, "elapsed_s": 18.3}11{"event": "train", "step": 220, "loss": -6.027935981750488, "grad_norm": 1.232113242149353, "lr": 0.00016575, "ex_per_s": 746.9889570964981, "data_frac": 0.6590710052364411, "mem_mb": 87.387136, "nll": -6.027935981750488, "elapsed_s": 18.9}12{"event": "train", "step": 240, "loss": -5.9130659103393555, "grad_norm": 0.4099697768688202, "lr": 0.00018075, "ex_per_s": 791.3177690066087, "data_frac": 0.641177741359346, "mem_mb": 87.387136, "nll": -5.9130659103393555, "elapsed_s": 19.4}13{"event": "train", "step": 260, "loss": -5.888187885284424, "grad_norm": 0.14474304020404816, "lr": 0.00019574999999999996, "ex_per_s": 833.5648877527472, "data_frac": 0.6239998384303513, "mem_mb": 87.387136, "nll": -5.888187885284424, "elapsed_s": 20.0}14{"event": "train", "step": 280, "loss": -5.943597316741943, "grad_norm": 0.4338245093822479, "lr": 0.00021074999999999997, "ex_per_s": 873.5533472180898, "data_frac": 0.6077477754946444, "mem_mb": 87.387136, "nll": -5.943597316741943, "elapsed_s": 20.5}15{"event": "train", "step": 300, "loss": -5.9672627449035645, "grad_norm": 0.20889081060886383, "lr": 0.00022574999999999996, "ex_per_s": 911.4090867231427, "data_frac": 0.5923149547761662, "mem_mb": 87.387136, "nll": -5.9672627449035645, "elapsed_s": 21.1}16{"event": "train", "step": 320, "loss": -5.963364601135254, "grad_norm": 0.9382969737052917, "lr": 0.00024074999999999997, "ex_per_s": 947.323012631871, "data_frac": 0.5776742189361541, "mem_mb": 87.387136, "nll": -5.963364601135254, "elapsed_s": 21.6}17{"event": "train", "step": 340, "loss": -5.834258556365967, "grad_norm": 0.5007207989692688, "lr": 0.00025575, "ex_per_s": 981.2777952025406, "data_frac": 0.5636898770958582, "mem_mb": 87.387136, "nll": -5.834258556365967, "elapsed_s": 22.2}18{"event": "train", "step": 360, "loss": -5.994662284851074, "grad_norm": 0.47530439496040344, "lr": 0.00027074999999999994, "ex_per_s": 1013.3254999307143, "data_frac": 0.5502587112708397, "mem_mb": 87.387136, "nll": -5.994662284851074, "elapsed_s": 22.7}19{"event": "train", "step": 380, "loss": -6.007313251495361, "grad_norm": 1.0708191394805908, "lr": 0.00028575, "ex_per_s": 1043.9972222157403, "data_frac": 0.5375367585003322, "mem_mb": 87.387136, "nll": -6.007313251495361, "elapsed_s": 23.3}20{"event": "train", "step": 400, "loss": -6.003469944000244, "grad_norm": 0.6356548070907593, "lr": 0.0003, "ex_per_s": 1072.997582147047, "data_frac": 0.5253570970639055, "mem_mb": 87.387136, "nll": -6.003469944000244, "elapsed_s": 23.9}21{"event": "train", "step": 420, "loss": -5.957545280456543, "grad_norm": 1.3730764389038086, "lr": 0.0002999980196294046, "ex_per_s": 1100.3901719113999, "data_frac": 0.5136847098567874, "mem_mb": 87.387136, "nll": -5.957545280456543, "elapsed_s": 24.4}22{"event": "train", "step": 440, "loss": -5.939514636993408, "grad_norm": 0.6811357140541077, "lr": 0.00029999207857572027, "ex_per_s": 1126.542491655987, "data_frac": 0.5025538575345276, "mem_mb": 87.387136, "nll": -5.939514636993408, "elapsed_s": 25.0}23{"event": "train", "step": 460, "loss": -6.097637176513672, "grad_norm": 1.1437867879867554, "lr": 0.00029998217701325043, "ex_per_s": 1151.5209995013072, "data_frac": 0.4919051671307119, "mem_mb": 87.387136, "nll": -6.097637176513672, "elapsed_s": 25.6}24{"event": "train", "step": 480, "loss": -5.87827205657959, "grad_norm": 0.1512659341096878, "lr": 0.00029996831523249534, "ex_per_s": 1175.3931113721778, "data_frac": 0.4817159800968535, "mem_mb": 87.387136, "nll": -5.87827205657959, "elapsed_s": 26.1}25{"event": "train", "step": 500, "loss": -5.923110008239746, "grad_norm": 0.5719287395477295, "lr": 0.0002999504936401433, "ex_per_s": 1198.3322668067049, "data_frac": 0.47197343817909265, "mem_mb": 87.387136, "nll": -5.923110008239746, "elapsed_s": 26.7}26{"event": "train", "step": 520, "loss": -6.040012836456299, "grad_norm": 0.29625198245048523, "lr": 0.0002999287127590589, "ex_per_s": 1220.1378380991964, "data_frac": 0.4625934904548634, "mem_mb": 87.387136, "nll": -6.040012836456299, "elapsed_s": 27.3}27{"event": "train", "step": 540, "loss": -5.880192279815674, "grad_norm": 0.231744185090065, "lr": 0.00029990297322826756, "ex_per_s": 1241.2192527532832, "data_frac": 0.4536629571526717, "mem_mb": 87.387136, "nll": -5.880192279815674, "elapsed_s": 27.9}28{"event": "train", "step": 560, "loss": -5.971814155578613, "grad_norm": 0.20757676661014557, "lr": 0.00029987327580293684, "ex_per_s": 1261.6588646190662, "data_frac": 0.44510686736912985, "mem_mb": 87.387136, "nll": -5.971814155578613, "elapsed_s": 28.4}29{"event": "train", "step": 580, "loss": -5.924593925476074, "grad_norm": 0.15755067765712738, "lr": 0.00029983962135435424, "ex_per_s": 1281.8717558533044, "data_frac": 0.437015306658457, "mem_mb": 87.387136, "nll": -5.924593925476074, "elapsed_s": 29.0}30{"event": "train", "step": 600, "loss": -5.9729533195495605, "grad_norm": 0.8026211857795715, "lr": 0.0002998020108699017, "ex_per_s": 1300.991804139551, "data_frac": 0.42915830114657455, "mem_mb": 87.387136, "nll": -5.9729533195495605, "elapsed_s": 29.5}31{"event": "train", "step": 620, "loss": -5.952296733856201, "grad_norm": 0.305502712726593, "lr": 0.0002997604454530266, "ex_per_s": 1319.2637643405146, "data_frac": 0.42156324520466043, "mem_mb": 87.387136, "nll": -5.952296733856201, "elapsed_s": 30.1}32{"event": "train", "step": 640, "loss": -6.065856456756592, "grad_norm": 0.5094771385192871, "lr": 0.00029971492632320935, "ex_per_s": 1337.2991366267204, "data_frac": 0.4143362034394023, "mem_mb": 87.387136, "nll": -6.065856456756592, "elapsed_s": 30.6}33{"event": "train", "step": 660, "loss": -6.0066046714782715, "grad_norm": 1.029942274093628, "lr": 0.0002996654548159277, "ex_per_s": 1354.7751390121616, "data_frac": 0.40737720257517956, "mem_mb": 87.387136, "nll": -6.0066046714782715, "elapsed_s": 31.2}34{"event": "train", "step": 680, "loss": -6.007057189941406, "grad_norm": 0.11710964888334274, "lr": 0.00029961203238261747, "ex_per_s": 1371.5738491602847, "data_frac": 0.4006374144721437, "mem_mb": 87.387136, "nll": -6.007057189941406, "elapsed_s": 31.7}35{"event": "train", "step": 700, "loss": -5.985086917877197, "grad_norm": 1.1003360748291016, "lr": 0.0002995546605906301, "ex_per_s": 1387.6211593344647, "data_frac": 0.394099535685363, "mem_mb": 87.387136, "nll": -5.985086917877197, "elapsed_s": 32.3}36{"event": "train", "step": 720, "loss": -6.06267786026001, "grad_norm": 0.550123393535614, "lr": 0.00029949334112318624, "ex_per_s": 1403.356652236374, "data_frac": 0.38783318892197544, "mem_mb": 87.387136, "nll": -6.06267786026001, "elapsed_s": 32.8}37{"event": "train", "step": 740, "loss": -6.034382343292236, "grad_norm": 0.1986245959997177, "lr": 0.0002994280757793271, "ex_per_s": 1418.3051541417015, "data_frac": 0.38170930606760445, "mem_mb": 87.387136, "nll": -6.034382343292236, "elapsed_s": 33.4}38{"event": "train", "step": 760, "loss": -5.959176063537598, "grad_norm": 0.9037821888923645, "lr": 0.00029935886647386114, "ex_per_s": 1432.9834722865226, "data_frac": 0.3758355715533176, "mem_mb": 87.387136, "nll": -5.959176063537598, "elapsed_s": 34.0}39{"event": "train", "step": 780, "loss": -5.8952250480651855, "grad_norm": 0.10423505306243896, "lr": 0.00029928571523730785, "ex_per_s": 1447.236620784163, "data_frac": 0.37015934047484284, "mem_mb": 87.387136, "nll": -5.8952250480651855, "elapsed_s": 34.5}40{"event": "train", "step": 800, "loss": -5.816573619842529, "grad_norm": 0.26577693223953247, "lr": 0.00029920862421583855, "ex_per_s": 1461.0652954133916, "data_frac": 0.36466288942636244, "mem_mb": 87.387136, "nll": -5.816573619842529, "elapsed_s": 35.1}41{"event": "train", "step": 820, "loss": -5.931841850280762, "grad_norm": 0.8744286298751831, "lr": 0.00029912759567121304, "ex_per_s": 1474.4408452390724, "data_frac": 0.35933203914128475, "mem_mb": 87.387136, "nll": -5.931841850280762, "elapsed_s": 35.6}42{"event": "train", "step": 840, "loss": -5.808979511260986, "grad_norm": 0.6802332401275635, "lr": 0.0002990426319807135, "ex_per_s": 1487.373886587576, "data_frac": 0.3541573243692837, "mem_mb": 87.387136, "nll": -5.808979511260986, "elapsed_s": 36.2}43{"event": "train", "step": 860, "loss": -5.915463924407959, "grad_norm": 1.4986572265625, "lr": 0.0002989537356370747, "ex_per_s": 1499.9679718074665, "data_frac": 0.3491355300827521, "mem_mb": 87.387136, "nll": -5.915463924407959, "elapsed_s": 36.7}44{"event": "train", "step": 880, "loss": -5.870353698730469, "grad_norm": 0.20672672986984253, "lr": 0.00029886090924841085, "ex_per_s": 1512.1664586052314, "data_frac": 0.34426939888606656, "mem_mb": 87.387136, "nll": -5.870353698730469, "elapsed_s": 37.3}45{"event": "train", "step": 900, "loss": -5.98939847946167, "grad_norm": 0.17975841462612152, "lr": 0.0002987641555381389, "ex_per_s": 1524.016598988959, "data_frac": 0.33953680165664635, "mem_mb": 87.387136, "nll": -5.98939847946167, "elapsed_s": 37.8}46{"event": "train", "step": 920, "loss": -6.027338027954102, "grad_norm": 0.36973750591278076, "lr": 0.0002986634773448989, "ex_per_s": 1535.527869987873, "data_frac": 0.33494396720773306, "mem_mb": 87.387136, "nll": -6.027338027954102, "elapsed_s": 38.4}47{"event": "train", "step": 940, "loss": -6.083712577819824, "grad_norm": 0.46151137351989746, "lr": 0.00029855887762247066, "ex_per_s": 1546.631860073435, "data_frac": 0.3304741581480734, "mem_mb": 87.387136, "nll": -6.083712577819824, "elapsed_s": 38.9}48{"event": "train", "step": 960, "loss": -5.966577529907227, "grad_norm": 0.5817192196846008, "lr": 0.00029845035943968696, "ex_per_s": 1557.485080137267, "data_frac": 0.3261334768591644, "mem_mb": 87.387136, "nll": -5.966577529907227, "elapsed_s": 39.5}49{"event": "train", "step": 980, "loss": -5.9267988204956055, "grad_norm": 0.1286102831363678, "lr": 0.0002983379259803436, "ex_per_s": 1568.050731774402, "data_frac": 0.32190956071076116, "mem_mb": 87.387136, "nll": -5.9267988204956055, "elapsed_s": 40.0}50{"event": "train", "step": 1000, "loss": -6.07277250289917, "grad_norm": 0.09989146143198013, "lr": 0.00029822158054310604, "ex_per_s": 1578.281899594725, "data_frac": 0.3177942919543449, "mem_mb": 87.387136, "nll": -6.07277250289917, "elapsed_s": 40.6}51{"event": "val", "step": 1000, "val_nll": -6.44646793489428, "best": -6.44646793489428, "bad": 0, "mem_mb": 87.387136, "elapsed_s": 102.9}52{"event": "train", "step": 1020, "loss": -6.027746200561523, "grad_norm": 0.6262896656990051, "lr": 0.0002981013265414125, "ex_per_s": 630.8510998602953, "data_frac": 0.12464009288383113, "mem_mb": 87.387136, "nll": -6.027746200561523, "elapsed_s": 103.5}53{"event": "train", "step": 1040, "loss": -5.9728288650512695, "grad_norm": 0.3161700963973999, "lr": 0.000297977167503374, "ex_per_s": 639.7919025899367, "data_frac": 0.12408129835447318, "mem_mb": 87.387136, "nll": -5.9728288650512695, "elapsed_s": 104.0}54{"event": "train", "step": 1060, "loss": -6.181101322174072, "grad_norm": 0.417673796415329, "lr": 0.00029784910707167064, "ex_per_s": 648.640702848875, "data_frac": 0.12353240329585514, "mem_mb": 87.387136, "nll": -6.181101322174072, "elapsed_s": 104.6}55{"event": "train", "step": 1080, "loss": -5.814739227294922, "grad_norm": 0.7007580399513245, "lr": 0.0002977171490034448, "ex_per_s": 657.4080261509125, "data_frac": 0.12298641160001743, "mem_mb": 87.387136, "nll": -5.814739227294922, "elapsed_s": 105.2}56{"event": "train", "step": 1100, "loss": -6.052592754364014, "grad_norm": 0.13131222128868103, "lr": 0.000297581297170191, "ex_per_s": 666.0778138111724, "data_frac": 0.12244561266665026, "mem_mb": 87.387136, "nll": -6.052592754364014, "elapsed_s": 105.7}57{"event": "train", "step": 1120, "loss": -6.053568363189697, "grad_norm": 0.4406443238258362, "lr": 0.0002974415555576423, "ex_per_s": 674.6439806481776, "data_frac": 0.12191394927583794, "mem_mb": 87.387136, "nll": -6.053568363189697, "elapsed_s": 106.3}58{"event": "train", "step": 1140, "loss": -6.019890785217285, "grad_norm": 0.23186486959457397, "lr": 0.000297297928265653, "ex_per_s": 683.1387369121938, "data_frac": 0.1213845388023123, "mem_mb": 87.387136, "nll": -6.019890785217285, "elapsed_s": 106.8}59{"event": "train", "step": 1160, "loss": -5.969985008239746, "grad_norm": 0.44112712144851685, "lr": 0.00029715041950807914, "ex_per_s": 691.545668233661, "data_frac": 0.12085987297111848, "mem_mb": 87.387136, "nll": -5.969985008239746, "elapsed_s": 107.4}60{"event": "train", "step": 1180, "loss": -5.956596851348877, "grad_norm": 0.22152145206928253, "lr": 0.000296999033612654, "ex_per_s": 699.8420664611023, "data_frac": 0.12034162484235793, "mem_mb": 87.387136, "nll": -5.956596851348877, "elapsed_s": 107.9}61{"event": "train", "step": 1200, "loss": -5.9472784996032715, "grad_norm": 0.2252693474292755, "lr": 0.00029684377502086165, "ex_per_s": 708.074671146022, "data_frac": 0.11982935236415727, "mem_mb": 87.387136, "nll": -5.9472784996032715, "elapsed_s": 108.5}62{"event": "train", "step": 1220, "loss": -6.033879280090332, "grad_norm": 0.5526195168495178, "lr": 0.00029668464828780665, "ex_per_s": 716.229696136611, "data_frac": 0.11932170375180338, "mem_mb": 87.387136, "nll": -6.033879280090332, "elapsed_s": 109.0}63{"event": "train", "step": 1240, "loss": -6.0588483810424805, "grad_norm": 0.06881555169820786, "lr": 0.00029652165808208004, "ex_per_s": 724.3041887290675, "data_frac": 0.11881967519895464, "mem_mb": 87.387136, "nll": -6.0588483810424805, "elapsed_s": 109.6}64{"event": "train", "step": 1260, "loss": -6.054294109344482, "grad_norm": 0.12128376215696335, "lr": 0.0002963548091856228, "ex_per_s": 732.2270805644646, "data_frac": 0.11832373355398446, "mem_mb": 87.387136, "nll": -6.054294109344482, "elapsed_s": 110.1}65{"event": "train", "step": 1280, "loss": -6.047044277191162, "grad_norm": 0.6023955941200256, "lr": 0.0002961841064935852, "ex_per_s": 740.112777675692, "data_frac": 0.11783705494424561, "mem_mb": 87.387136, "nll": -6.047044277191162, "elapsed_s": 110.7}66{"event": "train", "step": 1300, "loss": -5.863057613372803, "grad_norm": 1.1831268072128296, "lr": 0.0002960095550141835, "ex_per_s": 747.9490942765957, "data_frac": 0.11735099463168568, "mem_mb": 87.387136, "nll": -5.863057613372803, "elapsed_s": 111.2}67{"event": "train", "step": 1320, "loss": -5.827126502990723, "grad_norm": 0.4121658504009247, "lr": 0.00029583115986855273, "ex_per_s": 755.7064703605608, "data_frac": 0.11686831199079947, "mem_mb": 87.387136, "nll": -5.827126502990723, "elapsed_s": 111.8}68{"event": "train", "step": 1340, "loss": -5.953884601593018, "grad_norm": 0.11539014428853989, "lr": 0.0002956489262905966, "ex_per_s": 763.3878139667821, "data_frac": 0.11638963482950133, "mem_mb": 87.387136, "nll": -5.953884601593018, "elapsed_s": 112.4}69{"event": "train", "step": 1360, "loss": -6.037893772125244, "grad_norm": 0.8783569931983948, "lr": 0.0002954628596268339, "ex_per_s": 770.9971099874849, "data_frac": 0.11591736630492584, "mem_mb": 87.387136, "nll": -6.037893772125244, "elapsed_s": 112.9}70{"event": "train", "step": 1380, "loss": -6.07012414932251, "grad_norm": 0.09593472629785538, "lr": 0.0002952729653362417, "ex_per_s": 778.5351931134217, "data_frac": 0.11544917629127789, "mem_mb": 87.387136, "nll": -6.07012414932251, "elapsed_s": 113.5}71{"event": "train", "step": 1400, "loss": -5.797062397003174, "grad_norm": 0.15432216227054596, "lr": 0.00029507924899009505, "ex_per_s": 785.9953871285329, "data_frac": 0.11498322962479426, "mem_mb": 87.387136, "nll": -5.797062397003174, "elapsed_s": 114.0}72{"event": "train", "step": 1420, "loss": -5.990588188171387, "grad_norm": 0.4838603436946869, "lr": 0.00029488171627180373, "ex_per_s": 793.3893564780791, "data_frac": 0.11452409583646306, "mem_mb": 87.387136, "nll": -5.990588188171387, "elapsed_s": 114.6}73{"event": "train", "step": 1440, "loss": -5.745283126831055, "grad_norm": 0.13366298377513885, "lr": 0.00029468037297674526, "ex_per_s": 800.7120640051032, "data_frac": 0.1140715748737943, "mem_mb": 87.387136, "nll": -5.745283126831055, "elapsed_s": 115.1}74{"event": "train", "step": 1460, "loss": -5.800985336303711, "grad_norm": 0.3610480725765228, "lr": 0.00029447522501209515, "ex_per_s": 807.9568333142162, "data_frac": 0.1136217590486015, "mem_mb": 87.387136, "nll": -5.800985336303711, "elapsed_s": 115.7}75{"event": "train", "step": 1480, "loss": -6.043315410614014, "grad_norm": 0.76244056224823, "lr": 0.00029426627839665323, "ex_per_s": 815.1378413381527, "data_frac": 0.11317570248043561, "mem_mb": 87.387136, "nll": -6.043315410614014, "elapsed_s": 116.2}76{"event": "train", "step": 1500, "loss": -5.958849906921387, "grad_norm": 0.5434313416481018, "lr": 0.00029405353926066754, "ex_per_s": 822.2540098477745, "data_frac": 0.11273423384053402, "mem_mb": 87.387136, "nll": -5.958849906921387, "elapsed_s": 116.8}77{"event": "train", "step": 1520, "loss": -5.950752258300781, "grad_norm": 0.4597800672054291, "lr": 0.000293837013845654, "ex_per_s": 829.3021515545109, "data_frac": 0.1122948255997171, "mem_mb": 87.387136, "nll": -5.950752258300781, "elapsed_s": 117.3}78{"event": "train", "step": 1540, "loss": -5.878765106201172, "grad_norm": 0.259623259305954, "lr": 0.00029361670850421354, "ex_per_s": 836.2845971282455, "data_frac": 0.11186060444516974, "mem_mb": 87.387136, "nll": -5.878765106201172, "elapsed_s": 117.9}79{"event": "train", "step": 1560, "loss": -5.982401371002197, "grad_norm": 0.2739230692386627, "lr": 0.0002933926296998457, "ex_per_s": 843.2015329137512, "data_frac": 0.11143254995111966, "mem_mb": 87.387136, "nll": -5.982401371002197, "elapsed_s": 118.4}80{"event": "train", "step": 1580, "loss": -5.980667591094971, "grad_norm": 0.0781499445438385, "lr": 0.0002931647840067591, "ex_per_s": 850.0580501184725, "data_frac": 0.11100504983096497, "mem_mb": 87.387136, "nll": -5.980667591094971, "elapsed_s": 119.0}81{"event": "train", "step": 1600, "loss": -5.955578804016113, "grad_norm": 0.11596962809562683, "lr": 0.0002929331781096783, "ex_per_s": 856.8391205948361, "data_frac": 0.11058567797805925, "mem_mb": 87.387136, "nll": -5.955578804016113, "elapsed_s": 119.5}82{"event": "train", "step": 1620, "loss": -5.881551265716553, "grad_norm": 0.4086676239967346, "lr": 0.0002926978188036479, "ex_per_s": 863.5691800776307, "data_frac": 0.11016804425867165, "mem_mb": 87.387136, "nll": -5.881551265716553, "elapsed_s": 120.1}83{"event": "train", "step": 1640, "loss": -6.076884746551514, "grad_norm": 0.7403295636177063, "lr": 0.00029245871299383297, "ex_per_s": 870.2335450939, "data_frac": 0.10975226001118027, "mem_mb": 87.387136, "nll": -6.076884746551514, "elapsed_s": 120.6}84{"event": "train", "step": 1660, "loss": -6.037146091461182, "grad_norm": 1.1261307001113892, "lr": 0.0002922158676953168, "ex_per_s": 876.8325295415549, "data_frac": 0.10934130517491611, "mem_mb": 87.387136, "nll": -6.037146091461182, "elapsed_s": 121.2}85{"event": "train", "step": 1680, "loss": -5.867631435394287, "grad_norm": 0.16226300597190857, "lr": 0.0002919692900328947, "ex_per_s": 883.3808171253389, "data_frac": 0.10893318545350437, "mem_mb": 87.387136, "nll": -5.867631435394287, "elapsed_s": 121.7}86{"event": "train", "step": 1700, "loss": -5.9275593757629395, "grad_norm": 0.40410375595092773, "lr": 0.0002917189872408653, "ex_per_s": 889.8656550541784, "data_frac": 0.10852835271259646, "mem_mb": 87.387136, "nll": -5.9275593757629395, "elapsed_s": 122.3}87{"event": "train", "step": 1720, "loss": -5.886199474334717, "grad_norm": 1.1085450649261475, "lr": 0.00029146496666281795, "ex_per_s": 896.2895811063954, "data_frac": 0.10812809189865533, "mem_mb": 87.387136, "nll": -5.886199474334717, "elapsed_s": 122.8}88{"event": "train", "step": 1740, "loss": -6.273200035095215, "grad_norm": 0.26168251037597656, "lr": 0.00029120723575141777, "ex_per_s": 902.6659292655364, "data_frac": 0.10773200033929956, "mem_mb": 87.387136, "nll": -6.273200035095215, "elapsed_s": 123.4}89{"event": "train", "step": 1760, "loss": -5.928180694580078, "grad_norm": 0.17997372150421143, "lr": 0.00029094580206818636, "ex_per_s": 908.9800172424885, "data_frac": 0.10733860335658545, "mem_mb": 87.387136, "nll": -5.928180694580078, "elapsed_s": 123.9}90{"event": "train", "step": 1780, "loss": -5.870651721954346, "grad_norm": 0.43960997462272644, "lr": 0.00029068067328328045, "ex_per_s": 915.2387301640849, "data_frac": 0.10694862646865191, "mem_mb": 87.387136, "nll": -5.870651721954346, "elapsed_s": 124.5}91{"event": "train", "step": 1800, "loss": -6.049631118774414, "grad_norm": 0.9362449645996094, "lr": 0.0002904118571752668, "ex_per_s": 921.4503852302668, "data_frac": 0.10656358614928872, "mem_mb": 87.387136, "nll": -6.049631118774414, "elapsed_s": 125.0}92{"event": "train", "step": 1820, "loss": -5.938852787017822, "grad_norm": 0.21159812808036804, "lr": 0.00029013936163089376, "ex_per_s": 927.5982137203356, "data_frac": 0.10617976742063381, "mem_mb": 87.387136, "nll": -5.938852787017822, "elapsed_s": 125.6}93{"event": "train", "step": 1840, "loss": -6.0738935470581055, "grad_norm": 0.5510373711585999, "lr": 0.00028986319464486, "ex_per_s": 933.6920249798966, "data_frac": 0.10580112428283216, "mem_mb": 87.387136, "nll": -6.0738935470581055, "elapsed_s": 126.1}94{"event": "train", "step": 1860, "loss": -6.0498456954956055, "grad_norm": 0.23798057436943054, "lr": 0.00028958336431958003, "ex_per_s": 939.7267973437196, "data_frac": 0.10542430318048063, "mem_mb": 87.387136, "nll": -6.0498456954956055, "elapsed_s": 126.7}95{"event": "train", "step": 1880, "loss": -5.946710109710693, "grad_norm": 0.7650047540664673, "lr": 0.0002892998788649464, "ex_per_s": 945.7175590448015, "data_frac": 0.10505238535129563, "mem_mb": 87.387136, "nll": -5.946710109710693, "elapsed_s": 127.2}96{"event": "train", "step": 1900, "loss": -5.9826741218566895, "grad_norm": 0.2267639935016632, "lr": 0.0002890127465980889, "ex_per_s": 951.6533058073242, "data_frac": 0.10468395286473763, "mem_mb": 87.387136, "nll": -5.9826741218566895, "elapsed_s": 127.8}97{"event": "train", "step": 1920, "loss": -5.964985370635986, "grad_norm": 0.25300759077072144, "lr": 0.0002887219759431305, "ex_per_s": 957.5370434776405, "data_frac": 0.10431662992550104, "mem_mb": 87.387136, "nll": -5.964985370635986, "elapsed_s": 128.3}98{"event": "train", "step": 1940, "loss": -6.050823211669922, "grad_norm": 0.3881009817123413, "lr": 0.0002884275754309401, "ex_per_s": 963.3805370289896, "data_frac": 0.10395326404908316, "mem_mb": 87.387136, "nll": -6.050823211669922, "elapsed_s": 128.9}99{"event": "train", "step": 1960, "loss": -5.939585208892822, "grad_norm": 0.055183619260787964, "lr": 0.0002881295536988824, "ex_per_s": 969.1652252364594, "data_frac": 0.10359195045110157, "mem_mb": 87.387136, "nll": -5.939585208892822, "elapsed_s": 129.4}100{"event": "train", "step": 1980, "loss": -5.963694095611572, "grad_norm": 0.8392723202705383, "lr": 0.00028782791949056447, "ex_per_s": 974.8652942222055, "data_frac": 0.1032371684827485, "mem_mb": 87.387136, "nll": -5.963694095611572, "elapsed_s": 130.0}101{"event": "train", "step": 2000, "loss": -5.915077209472656, "grad_norm": 0.3372773230075836, "lr": 0.00028752268165557917, "ex_per_s": 980.5065718593831, "data_frac": 0.10288437232390438, "mem_mb": 87.387136, "nll": -5.915077209472656, "elapsed_s": 130.6}102{"event": "val", "step": 2000, "val_nll": -6.448097740310123, "best": -6.448097740310123, "bad": 0, "mem_mb": 87.387136, "elapsed_s": 192.4}103{"event": "train", "step": 2020, "loss": -5.92456579208374, "grad_norm": 0.37628281116485596, "lr": 0.00028721384914924547, "ex_per_s": 669.93426201215, "data_frac": 0.0696582525604068, "mem_mb": 87.387136, "nll": -5.92456579208374, "elapsed_s": 193.0}104{"event": "train", "step": 2040, "loss": -5.873891353607178, "grad_norm": 0.12839029729366302, "lr": 0.0002869014310323458, "ex_per_s": 674.6353526528309, "data_frac": 0.06951590713295805, "mem_mb": 87.387136, "nll": -5.873891353607178, "elapsed_s": 193.5}105{"event": "train", "step": 2060, "loss": -6.1435747146606445, "grad_norm": 0.276603102684021, "lr": 0.0002865854364708602, "ex_per_s": 679.307509004036, "data_frac": 0.06937258513965874, "mem_mb": 87.387136, "nll": -6.1435747146606445, "elapsed_s": 194.1}106{"event": "train", "step": 2080, "loss": -5.960577011108398, "grad_norm": 0.16785742342472076, "lr": 0.00028626587473569746, "ex_per_s": 683.9379256666349, "data_frac": 0.06923259109724128, "mem_mb": 87.387136, "nll": -5.960577011108398, "elapsed_s": 194.6}107{"event": "train", "step": 2100, "loss": -5.899491786956787, "grad_norm": 0.9659305810928345, "lr": 0.0002859427552024228, "ex_per_s": 688.5608881103927, "data_frac": 0.06909152764828845, "mem_mb": 87.387136, "nll": -5.899491786956787, "elapsed_s": 195.2}108{"event": "train", "step": 2120, "loss": -6.1468024253845215, "grad_norm": 1.0310970544815063, "lr": 0.0002856160873509834, "ex_per_s": 693.1558355622558, "data_frac": 0.06895227721588046, "mem_mb": 87.387136, "nll": -6.1468024253845215, "elapsed_s": 195.8}109{"event": "train", "step": 2140, "loss": -5.977762699127197, "grad_norm": 0.16244547069072723, "lr": 0.00028528588076542964, "ex_per_s": 697.7252883377536, "data_frac": 0.06881230482316222, "mem_mb": 87.387136, "nll": -5.977762699127197, "elapsed_s": 196.3}110{"event": "train", "step": 2160, "loss": -5.9270710945129395, "grad_norm": 0.32392823696136475, "lr": 0.00028495214513363427, "ex_per_s": 702.2743509682305, "data_frac": 0.06867301303847165, "mem_mb": 87.387136, "nll": -5.9270710945129395, "elapsed_s": 196.9}111{"event": "train", "step": 2180, "loss": -6.080907344818115, "grad_norm": 0.6312873959541321, "lr": 0.0002846148902470082, "ex_per_s": 706.7946271455825, "data_frac": 0.06853565644116626, "mem_mb": 87.387136, "nll": -6.080907344818115, "elapsed_s": 197.4}112{"event": "train", "step": 2200, "loss": -5.964138507843018, "grad_norm": 0.5938954949378967, "lr": 0.0002842741260002131, "ex_per_s": 711.2916790423378, "data_frac": 0.06839771320252767, "mem_mb": 87.387136, "nll": -5.964138507843018, "elapsed_s": 198.0}113{"event": "train", "step": 2220, "loss": -5.826471328735352, "grad_norm": 0.4477749764919281, "lr": 0.00028392986239087094, "ex_per_s": 715.76120553392, "data_frac": 0.06826086034085146, "mem_mb": 87.387136, "nll": -5.826471328735352, "elapsed_s": 198.5}114{"event": "train", "step": 2240, "loss": -5.961202621459961, "grad_norm": 0.385021448135376, "lr": 0.0002835821095192712, "ex_per_s": 720.2071744113925, "data_frac": 0.06812451650649798, "mem_mb": 87.387136, "nll": -5.961202621459961, "elapsed_s": 199.1}115{"event": "train", "step": 2260, "loss": -5.738562107086182, "grad_norm": 0.617154061794281, "lr": 0.0002832308775880741, "ex_per_s": 724.6232978399175, "data_frac": 0.06798993085055793, "mem_mb": 87.387136, "nll": -5.738562107086182, "elapsed_s": 199.6}116{"event": "train", "step": 2280, "loss": -6.009629726409912, "grad_norm": 0.27370303869247437, "lr": 0.0002828761769020111, "ex_per_s": 728.9897699422443, "data_frac": 0.06785913727055828, "mem_mb": 87.387136, "nll": -6.009629726409912, "elapsed_s": 200.2}117{"event": "train", "step": 2300, "loss": -5.998528957366943, "grad_norm": 0.1112547293305397, "lr": 0.0002825180178675832, "ex_per_s": 733.3414217656425, "data_frac": 0.06772850001718193, "mem_mb": 87.387136, "nll": -5.998528957366943, "elapsed_s": 200.7}118{"event": "train", "step": 2320, "loss": -6.068765163421631, "grad_norm": 0.5632175803184509, "lr": 0.00028215641099275517, "ex_per_s": 737.6918707545615, "data_frac": 0.067596342273222, "mem_mb": 87.387136, "nll": -6.068765163421631, "elapsed_s": 201.3}119{"event": "train", "step": 2340, "loss": -5.850257396697998, "grad_norm": 0.5378944277763367, "lr": 0.0002817913668866473, "ex_per_s": 742.0171290519071, "data_frac": 0.06746316387856566, "mem_mb": 87.387136, "nll": -5.850257396697998, "elapsed_s": 201.8}120{"event": "train", "step": 2360, "loss": -5.927847862243652, "grad_norm": 0.5388060808181763, "lr": 0.00028142289625922426, "ex_per_s": 746.321554231724, "data_frac": 0.06733140763264443, "mem_mb": 87.387136, "nll": -5.927847862243652, "elapsed_s": 202.4}121{"event": "train", "step": 2380, "loss": -5.93149995803833, "grad_norm": 0.2845104932785034, "lr": 0.00028105100992098075, "ex_per_s": 750.5821964960635, "data_frac": 0.0672057076769901, "mem_mb": 87.387136, "nll": -5.93149995803833, "elapsed_s": 202.9}122{"event": "train", "step": 2400, "loss": -5.957243919372559, "grad_norm": 0.9727479219436646, "lr": 0.00028067571878262454, "ex_per_s": 754.8278428856687, "data_frac": 0.06707546423074254, "mem_mb": 87.387136, "nll": -5.957243919372559, "elapsed_s": 203.5}123{"event": "train", "step": 2420, "loss": -5.995573997497559, "grad_norm": 0.18297845125198364, "lr": 0.00028029703385475607, "ex_per_s": 759.0587102201467, "data_frac": 0.06694608804836685, "mem_mb": 87.387136, "nll": -5.995573997497559, "elapsed_s": 204.1}124{"event": "train", "step": 2440, "loss": -6.097630500793457, "grad_norm": 0.5550150275230408, "lr": 0.0002799149662475457, "ex_per_s": 763.268720705212, "data_frac": 0.06681911317390023, "mem_mb": 87.387136, "nll": -6.097630500793457, "elapsed_s": 204.6}125{"event": "train", "step": 2460, "loss": -6.130674362182617, "grad_norm": 0.18153207004070282, "lr": 0.0002795295271704076, "ex_per_s": 767.4575346693026, "data_frac": 0.06669196426402169, "mem_mb": 87.387136, "nll": -6.130674362182617, "elapsed_s": 205.2}126{"event": "train", "step": 2480, "loss": -5.943713188171387, "grad_norm": 0.24111561477184296, "lr": 0.00027914072793167077, "ex_per_s": 771.6247181860418, "data_frac": 0.0665649089071843, "mem_mb": 87.387136, "nll": -5.943713188171387, "elapsed_s": 205.7}127{"event": "train", "step": 2500, "loss": -5.758356094360352, "grad_norm": 0.43684643507003784, "lr": 0.0002787485799382476, "ex_per_s": 775.7670688862555, "data_frac": 0.06643869841611623, "mem_mb": 87.387136, "nll": -5.758356094360352, "elapsed_s": 206.3}128{"event": "train", "step": 2520, "loss": -5.960775375366211, "grad_norm": 0.5992901921272278, "lr": 0.0002783530946952989, "ex_per_s": 779.8756004424941, "data_frac": 0.06631531783930669, "mem_mb": 87.387136, "nll": -5.960775375366211, "elapsed_s": 206.8}129{"event": "train", "step": 2540, "loss": -6.030770778656006, "grad_norm": 0.19597789645195007, "lr": 0.00027795428380589644, "ex_per_s": 783.9718239198831, "data_frac": 0.06619026265997377, "mem_mb": 87.387136, "nll": -6.030770778656006, "elapsed_s": 207.4}130{"event": "train", "step": 2560, "loss": -5.946702480316162, "grad_norm": 0.1972549706697464, "lr": 0.00027755215897068257, "ex_per_s": 788.054494791509, "data_frac": 0.0660677878606237, "mem_mb": 87.387136, "nll": -5.946702480316162, "elapsed_s": 207.9}131{"event": "train", "step": 2580, "loss": -5.958736419677734, "grad_norm": 0.2554219663143158, "lr": 0.0002771467319875269, "ex_per_s": 792.1082157769769, "data_frac": 0.06594421741310626, "mem_mb": 87.387136, "nll": -5.958736419677734, "elapsed_s": 208.5}132{"event": "train", "step": 2600, "loss": -5.846290588378906, "grad_norm": 0.12335797399282455, "lr": 0.0002767380147511802, "ex_per_s": 796.1458543946065, "data_frac": 0.06582083327341696, "mem_mb": 87.387136, "nll": -5.846290588378906, "elapsed_s": 209.0}133{"event": "train", "step": 2620, "loss": -5.888761520385742, "grad_norm": 1.0330591201782227, "lr": 0.0002763260192529254, "ex_per_s": 800.1597763562073, "data_frac": 0.0656993336237357, "mem_mb": 87.387136, "nll": -5.888761520385742, "elapsed_s": 209.6}134{"event": "train", "step": 2640, "loss": -5.985912799835205, "grad_norm": 0.41847720742225647, "lr": 0.0002759107575802257, "ex_per_s": 804.1498013757845, "data_frac": 0.06557772162727349, "mem_mb": 87.387136, "nll": -5.985912799835205, "elapsed_s": 210.1}135{"event": "train", "step": 2660, "loss": -5.820004940032959, "grad_norm": 0.11785310506820679, "lr": 0.00027549224191637014, "ex_per_s": 808.0902059498196, "data_frac": 0.06545997122882431, "mem_mb": 87.387136, "nll": -5.820004940032959, "elapsed_s": 210.7}136{"event": "train", "step": 2680, "loss": -5.9571757316589355, "grad_norm": 0.594344973564148, "lr": 0.000275070484540116, "ex_per_s": 812.0256264523825, "data_frac": 0.06534248981528719, "mem_mb": 87.387136, "nll": -5.9571757316589355, "elapsed_s": 211.2}137{"event": "train", "step": 2700, "loss": -6.132286071777344, "grad_norm": 0.1716897338628769, "lr": 0.00027464549782532844, "ex_per_s": 815.951604452, "data_frac": 0.06522286513360934, "mem_mb": 87.387136, "nll": -6.132286071777344, "elapsed_s": 211.8}138{"event": "train", "step": 2720, "loss": -5.884458541870117, "grad_norm": 0.22029858827590942, "lr": 0.00027421729424061787, "ex_per_s": 819.8592598991969, "data_frac": 0.06510315760764075, "mem_mb": 87.387136, "nll": -5.884458541870117, "elapsed_s": 212.3}139{"event": "train", "step": 2740, "loss": -6.126152038574219, "grad_norm": 0.14196233451366425, "lr": 0.0002737858863489736, "ex_per_s": 823.7489692038324, "data_frac": 0.06498509995231615, "mem_mb": 87.387136, "nll": -6.126152038574219, "elapsed_s": 212.9}140{"event": "train", "step": 2760, "loss": -5.9625163078308105, "grad_norm": 0.381961852312088, "lr": 0.00027335128680739583, "ex_per_s": 827.6232921444213, "data_frac": 0.06486772390987428, "mem_mb": 87.387136, "nll": -5.9625163078308105, "elapsed_s": 213.4}141{"event": "train", "step": 2780, "loss": -6.013861656188965, "grad_norm": 0.347411185503006, "lr": 0.00027291350836652374, "ex_per_s": 831.4767796093726, "data_frac": 0.06475056902216367, "mem_mb": 87.387136, "nll": -6.013861656188965, "elapsed_s": 214.0}142{"event": "train", "step": 2800, "loss": -5.981975555419922, "grad_norm": 0.24702735245227814, "lr": 0.00027247256387026185, "ex_per_s": 835.3075591956122, "data_frac": 0.06463265416565092, "mem_mb": 87.387136, "nll": -5.981975555419922, "elapsed_s": 214.5}143{"event": "train", "step": 2820, "loss": -6.006134510040283, "grad_norm": 0.5524539947509766, "lr": 0.000272028466255403, "ex_per_s": 839.1174496140079, "data_frac": 0.06451718757353306, "mem_mb": 87.387136, "nll": -6.006134510040283, "elapsed_s": 215.1}144{"event": "train", "step": 2840, "loss": -5.875261306762695, "grad_norm": 0.5831489562988281, "lr": 0.0002715812285512487, "ex_per_s": 842.9135455786738, "data_frac": 0.06440237792474818, "mem_mb": 87.387136, "nll": -5.875261306762695, "elapsed_s": 215.6}145{"event": "train", "step": 2860, "loss": -6.051813125610352, "grad_norm": 0.1396515667438507, "lr": 0.0002711308638792272, "ex_per_s": 846.6887053719466, "data_frac": 0.0642871335479779, "mem_mb": 87.387136, "nll": -6.051813125610352, "elapsed_s": 216.2}146{"event": "train", "step": 2880, "loss": -5.773238182067871, "grad_norm": 0.16466780006885529, "lr": 0.000270677385452508, "ex_per_s": 850.4422585109293, "data_frac": 0.06417228626528132, "mem_mb": 87.387136, "nll": -5.773238182067871, "elapsed_s": 216.7}147{"event": "train", "step": 2900, "loss": -5.792508602142334, "grad_norm": 0.6678301095962524, "lr": 0.00027022080657561484, "ex_per_s": 854.1798398150106, "data_frac": 0.06405893798019585, "mem_mb": 87.387136, "nll": -5.792508602142334, "elapsed_s": 217.3}148{"event": "train", "step": 2920, "loss": -5.969480037689209, "grad_norm": 0.43283340334892273, "lr": 0.0002697611406440348, "ex_per_s": 857.899888336311, "data_frac": 0.06394585326231329, "mem_mb": 87.387136, "nll": -5.969480037689209, "elapsed_s": 217.8}149{"event": "train", "step": 2940, "loss": -5.89968204498291, "grad_norm": 0.6470178365707397, "lr": 0.00026929840114382563, "ex_per_s": 861.6007385274202, "data_frac": 0.0638348274681402, "mem_mb": 87.387136, "nll": -5.89968204498291, "elapsed_s": 218.4}150{"event": "train", "step": 2960, "loss": -5.935841083526611, "grad_norm": 0.24842789769172668, "lr": 0.0002688326016512199, "ex_per_s": 865.2825940817252, "data_frac": 0.06372368621509614, "mem_mb": 87.387136, "nll": -5.935841083526611, "elapsed_s": 218.9}151{"event": "train", "step": 2980, "loss": -5.89022970199585, "grad_norm": 0.16711390018463135, "lr": 0.0002683637558322267, "ex_per_s": 868.9470727290504, "data_frac": 0.063612246668732, "mem_mb": 87.387136, "nll": -5.89022970199585, "elapsed_s": 219.5}152{"event": "train", "step": 3000, "loss": -5.994123458862305, "grad_norm": 0.21651625633239746, "lr": 0.0002678918774422309, "ex_per_s": 872.592172023114, "data_frac": 0.06350191044380722, "mem_mb": 87.387136, "nll": -5.994123458862305, "elapsed_s": 220.0}153{"event": "val", "step": 3000, "val_nll": -6.44874599689327, "best": -6.44874599689327, "bad": 0, "mem_mb": 87.387136, "elapsed_s": 282.0}154{"event": "train", "step": 3020, "loss": -5.961611747741699, "grad_norm": 0.9950322508811951, "lr": 0.0002674169803255893, "ex_per_s": 683.9693622218892, "data_frac": 0.049487368692438284, "mem_mb": 87.387136, "nll": -5.961611747741699, "elapsed_s": 282.6}155{"event": "train", "step": 3040, "loss": -5.859049320220947, "grad_norm": 1.105383038520813, "lr": 0.0002669390784152248, "ex_per_s": 687.1531840319066, "data_frac": 0.049428179361831324, "mem_mb": 87.387136, "nll": -5.859049320220947, "elapsed_s": 283.2}156{"event": "train", "step": 3060, "loss": -6.071338653564453, "grad_norm": 0.1949179321527481, "lr": 0.00026645818573221706, "ex_per_s": 690.3252472888908, "data_frac": 0.04936979453233726, "mem_mb": 87.387136, "nll": -6.071338653564453, "elapsed_s": 283.7}157{"event": "train", "step": 3080, "loss": -5.967234134674072, "grad_norm": 0.1887255758047104, "lr": 0.00026597431638539183, "ex_per_s": 693.487453346192, "data_frac": 0.04931105315956573, "mem_mb": 87.387136, "nll": -5.967234134674072, "elapsed_s": 284.3}158{"event": "train", "step": 3100, "loss": -6.012963771820068, "grad_norm": 0.30092909932136536, "lr": 0.00026548748457090647, "ex_per_s": 696.6396777357492, "data_frac": 0.049253110100718325, "mem_mb": 87.387136, "nll": -6.012963771820068, "elapsed_s": 284.8}159{"event": "train", "step": 3120, "loss": -6.087491989135742, "grad_norm": 0.9991333484649658, "lr": 0.00026499770457183376, "ex_per_s": 699.7765920118902, "data_frac": 0.04919447975926008, "mem_mb": 87.387136, "nll": -6.087491989135742, "elapsed_s": 285.4}160{"event": "train", "step": 3140, "loss": -6.009480953216553, "grad_norm": 0.37701359391212463, "lr": 0.0002645049907577427, "ex_per_s": 702.9014609272161, "data_frac": 0.04913695045551485, "mem_mb": 87.387136, "nll": -6.009480953216553, "elapsed_s": 285.9}161{"event": "train", "step": 3160, "loss": -6.0198564529418945, "grad_norm": 0.11491687595844269, "lr": 0.00026400935758427697, "ex_per_s": 706.0157578831631, "data_frac": 0.049079484691353564, "mem_mb": 87.387136, "nll": -6.0198564529418945, "elapsed_s": 286.5}162{"event": "train", "step": 3180, "loss": -6.093408107757568, "grad_norm": 0.26270055770874023, "lr": 0.00026351081959273087, "ex_per_s": 709.11696384352, "data_frac": 0.04902198842155227, "mem_mb": 87.387136, "nll": -6.093408107757568, "elapsed_s": 287.0}163{"event": "train", "step": 3200, "loss": -6.001749515533447, "grad_norm": 0.575379490852356, "lr": 0.0002630093914096226, "ex_per_s": 712.2052837272347, "data_frac": 0.04896493215761048, "mem_mb": 87.387136, "nll": -6.001749515533447, "elapsed_s": 287.6}164{"event": "train", "step": 3220, "loss": -5.9138617515563965, "grad_norm": 0.4132280945777893, "lr": 0.00026250508774626523, "ex_per_s": 715.2829957782048, "data_frac": 0.048908902933634125, "mem_mb": 87.387136, "nll": -5.9138617515563965, "elapsed_s": 288.1}165{"event": "train", "step": 3240, "loss": -5.801316261291504, "grad_norm": 0.25039154291152954, "lr": 0.00026199792339833493, "ex_per_s": 718.3489926050748, "data_frac": 0.04885352411829015, "mem_mb": 87.387136, "nll": -5.801316261291504, "elapsed_s": 288.7}166{"event": "train", "step": 3260, "loss": -5.960469722747803, "grad_norm": 0.11596778780221939, "lr": 0.00026148791324543704, "ex_per_s": 721.4051119101993, "data_frac": 0.048797613310147085, "mem_mb": 87.387136, "nll": -5.960469722747803, "elapsed_s": 289.2}167{"event": "train", "step": 3280, "loss": -6.0191473960876465, "grad_norm": 0.10604080557823181, "lr": 0.0002609750722506695, "ex_per_s": 724.4248473011909, "data_frac": 0.04874584152925874, "mem_mb": 87.387136, "nll": -6.0191473960876465, "elapsed_s": 289.8}168{"event": "train", "step": 3300, "loss": -5.801467418670654, "grad_norm": 0.6658523678779602, "lr": 0.0002604594154601839, "ex_per_s": 727.4533406627259, "data_frac": 0.04869227337581376, "mem_mb": 87.387136, "nll": -5.801467418670654, "elapsed_s": 290.3}169{"event": "train", "step": 3320, "loss": -5.988015651702881, "grad_norm": 0.5227786302566528, "lr": 0.0002599409580027438, "ex_per_s": 730.4670226539469, "data_frac": 0.04863772339150156, "mem_mb": 87.387136, "nll": -5.988015651702881, "elapsed_s": 290.9}170{"event": "train", "step": 3340, "loss": -5.928284168243408, "grad_norm": 0.5201811194419861, "lr": 0.0002594197150892811, "ex_per_s": 733.4509960816991, "data_frac": 0.0485859350196955, "mem_mb": 87.387136, "nll": -5.928284168243408, "elapsed_s": 291.5}171{"event": "train", "step": 3360, "loss": -5.909664630889893, "grad_norm": 0.5909820199012756, "lr": 0.00025889570201244967, "ex_per_s": 736.4248330997307, "data_frac": 0.048535786614155974, "mem_mb": 87.387136, "nll": -5.909664630889893, "elapsed_s": 292.0}172{"event": "train", "step": 3380, "loss": -6.120666980743408, "grad_norm": 0.2354678511619568, "lr": 0.00025836893414617684, "ex_per_s": 739.4126198617708, "data_frac": 0.04848130643620197, "mem_mb": 87.387136, "nll": -6.120666980743408, "elapsed_s": 292.6}173{"event": "train", "step": 3400, "loss": -5.881025314331055, "grad_norm": 0.28381073474884033, "lr": 0.00025783942694521214, "ex_per_s": 742.3897276984318, "data_frac": 0.048426100400222556, "mem_mb": 87.387136, "nll": -5.881025314331055, "elapsed_s": 293.1}174{"event": "train", "step": 3420, "loss": -5.836352825164795, "grad_norm": 0.29121559858322144, "lr": 0.00025730719594467387, "ex_per_s": 745.354866677292, "data_frac": 0.048371671666023144, "mem_mb": 87.387136, "nll": -5.836352825164795, "elapsed_s": 293.7}175{"event": "train", "step": 3440, "loss": -5.916317462921143, "grad_norm": 0.4118235409259796, "lr": 0.00025677225675959357, "ex_per_s": 748.3087385795653, "data_frac": 0.04831662489031198, "mem_mb": 87.387136, "nll": -5.916317462921143, "elapsed_s": 294.2}176{"event": "train", "step": 3460, "loss": -6.056397438049316, "grad_norm": 0.2838130593299866, "lr": 0.00025623462508445754, "ex_per_s": 751.2520853641126, "data_frac": 0.048262553823023704, "mem_mb": 87.387136, "nll": -6.056397438049316, "elapsed_s": 294.8}177{"event": "train", "step": 3480, "loss": -5.917623519897461, "grad_norm": 0.07283079624176025, "lr": 0.00025569431669274684, "ex_per_s": 754.184666457092, "data_frac": 0.048208354738106134, "mem_mb": 87.387136, "nll": -5.917623519897461, "elapsed_s": 295.3}178{"event": "train", "step": 3500, "loss": -6.035781383514404, "grad_norm": 0.7294445633888245, "lr": 0.00025515134743647386, "ex_per_s": 757.108950848106, "data_frac": 0.04815503070728151, "mem_mb": 87.387136, "nll": -6.035781383514404, "elapsed_s": 295.9}179{"event": "train", "step": 3520, "loss": -5.9320268630981445, "grad_norm": 0.6693244576454163, "lr": 0.0002546057332457181, "ex_per_s": 760.0202443458354, "data_frac": 0.048101394406062016, "mem_mb": 87.387136, "nll": -5.9320268630981445, "elapsed_s": 296.4}180{"event": "train", "step": 3540, "loss": -6.136391639709473, "grad_norm": 0.25614842772483826, "lr": 0.00025405749012815786, "ex_per_s": 762.9220094732851, "data_frac": 0.048047833285596926, "mem_mb": 87.387136, "nll": -6.136391639709473, "elapsed_s": 297.0}181{"event": "train", "step": 3560, "loss": -6.142132759094238, "grad_norm": 0.11818993836641312, "lr": 0.0002535066341686014, "ex_per_s": 765.8108302451595, "data_frac": 0.04799430145134938, "mem_mb": 87.387136, "nll": -6.142132759094238, "elapsed_s": 297.5}182{"event": "train", "step": 3580, "loss": -5.903182506561279, "grad_norm": 0.3154137432575226, "lr": 0.0002529531815285146, "ex_per_s": 768.6918187724492, "data_frac": 0.047940910336687875, "mem_mb": 87.387136, "nll": -5.903182506561279, "elapsed_s": 298.1}183{"event": "train", "step": 3600, "loss": -6.09623908996582, "grad_norm": 0.11119779944419861, "lr": 0.00025239714844554667, "ex_per_s": 771.5594118608379, "data_frac": 0.047888503038336076, "mem_mb": 87.387136, "nll": -6.09623908996582, "elapsed_s": 298.6}184{"event": "train", "step": 3620, "loss": -5.9068684577941895, "grad_norm": 1.3510226011276245, "lr": 0.0002518385512330544, "ex_per_s": 774.4168457990845, "data_frac": 0.04783627496074692, "mem_mb": 87.387136, "nll": -5.9068684577941895, "elapsed_s": 299.2}185{"event": "train", "step": 3640, "loss": -6.052760124206543, "grad_norm": 0.3451843559741974, "lr": 0.00025127740627962274, "ex_per_s": 777.2627479139026, "data_frac": 0.04778373679932633, "mem_mb": 87.387136, "nll": -6.052760124206543, "elapsed_s": 299.7}186{"event": "train", "step": 3660, "loss": -5.832146167755127, "grad_norm": 0.4144589900970459, "lr": 0.00025071373004858453, "ex_per_s": 780.1018043122476, "data_frac": 0.047731197020214534, "mem_mb": 87.387136, "nll": -5.832146167755127, "elapsed_s": 300.3}187{"event": "train", "step": 3680, "loss": -5.993370532989502, "grad_norm": 0.12051399052143097, "lr": 0.00025014753907753737, "ex_per_s": 782.927585651553, "data_frac": 0.04767888957433364, "mem_mb": 87.387136, "nll": -5.993370532989502, "elapsed_s": 300.8}188{"event": "train", "step": 3700, "loss": -6.068953037261963, "grad_norm": 0.5796141028404236, "lr": 0.0002495788499778582, "ex_per_s": 785.7458771934387, "data_frac": 0.04762729561256485, "mem_mb": 87.387136, "nll": -6.068953037261963, "elapsed_s": 301.4}189{"event": "train", "step": 3720, "loss": -5.946788787841797, "grad_norm": 0.27789026498794556, "lr": 0.0002490076794342163, "ex_per_s": 788.5541419325868, "data_frac": 0.047575216451295844, "mem_mb": 87.387136, "nll": -5.946788787841797, "elapsed_s": 301.9}190{"event": "train", "step": 3740, "loss": -6.187288284301758, "grad_norm": 0.7211219668388367, "lr": 0.00024843404420408344, "ex_per_s": 791.3491049501284, "data_frac": 0.047523350019591584, "mem_mb": 87.387136, "nll": -6.187288284301758, "elapsed_s": 302.5}191{"event": "train", "step": 3760, "loss": -6.146374225616455, "grad_norm": 0.19833917915821075, "lr": 0.0002478579611172425, "ex_per_s": 794.1336072371939, "data_frac": 0.047472310897729866, "mem_mb": 87.387136, "nll": -6.146374225616455, "elapsed_s": 303.0}192{"event": "train", "step": 3780, "loss": -5.961691379547119, "grad_norm": 0.18673504889011383, "lr": 0.00024727944707529366, "ex_per_s": 796.9113521633652, "data_frac": 0.04742228196213722, "mem_mb": 87.387136, "nll": -5.961691379547119, "elapsed_s": 303.6}193{"event": "train", "step": 3800, "loss": -6.0998125076293945, "grad_norm": 0.40004169940948486, "lr": 0.0002466985190511583, "ex_per_s": 799.6754871486713, "data_frac": 0.047371186494021354, "mem_mb": 87.387136, "nll": -6.0998125076293945, "elapsed_s": 304.1}194{"event": "train", "step": 3820, "loss": -5.969661712646484, "grad_norm": 0.5440827012062073, "lr": 0.0002461151940885812, "ex_per_s": 802.4304509619426, "data_frac": 0.04731990013248095, "mem_mb": 87.387136, "nll": -5.969661712646484, "elapsed_s": 304.7}195{"event": "train", "step": 3840, "loss": -5.873776435852051, "grad_norm": 0.4971930682659149, "lr": 0.0002455294893016306, "ex_per_s": 805.1743382797956, "data_frac": 0.047269436420304484, "mem_mb": 87.387136, "nll": -5.873776435852051, "elapsed_s": 305.2}196{"event": "train", "step": 3860, "loss": -6.012630939483643, "grad_norm": 0.18269334733486176, "lr": 0.000244941421874196, "ex_per_s": 807.9092006997784, "data_frac": 0.047218306667104366, "mem_mb": 87.387136, "nll": -6.012630939483643, "elapsed_s": 305.8}197{"event": "train", "step": 3880, "loss": -5.8886566162109375, "grad_norm": 0.11685912311077118, "lr": 0.00024435100905948387, "ex_per_s": 810.6342231984505, "data_frac": 0.04716776331207462, "mem_mb": 87.387136, "nll": -5.8886566162109375, "elapsed_s": 306.3}198{"event": "train", "step": 3900, "loss": -6.180679798126221, "grad_norm": 0.14571112394332886, "lr": 0.0002437582681795117, "ex_per_s": 813.3489355809127, "data_frac": 0.04711711717108108, "mem_mb": 87.387136, "nll": -6.180679798126221, "elapsed_s": 306.9}199{"event": "train", "step": 3920, "loss": -6.035394668579102, "grad_norm": 0.6779693961143494, "lr": 0.00024316321662459957, "ex_per_s": 816.0578612062883, "data_frac": 0.04706722403732482, "mem_mb": 87.387136, "nll": -6.035394668579102, "elapsed_s": 307.4}200{"event": "train", "step": 3940, "loss": -5.920629978179932, "grad_norm": 0.4710245430469513, "lr": 0.00024256587185286015, "ex_per_s": 818.7537280177114, "data_frac": 0.04701686557483968, "mem_mb": 87.387136, "nll": -5.920629978179932, "elapsed_s": 308.0}201{"event": "train", "step": 3960, "loss": -6.066068172454834, "grad_norm": 0.44925907254219055, "lr": 0.0002419662513896863, "ex_per_s": 821.4403348690654, "data_frac": 0.0469679531133506, "mem_mb": 87.387136, "nll": -6.066068172454834, "elapsed_s": 308.5}202{"event": "train", "step": 3980, "loss": -5.805137634277344, "grad_norm": 0.26259705424308777, "lr": 0.00024136437282723714, "ex_per_s": 824.1156854028688, "data_frac": 0.0469188147898776, "mem_mb": 87.387136, "nll": -5.805137634277344, "elapsed_s": 309.1}203{"event": "train", "step": 4000, "loss": -6.055163383483887, "grad_norm": 1.0784001350402832, "lr": 0.0002407602538239216, "ex_per_s": 826.7821267554624, "data_frac": 0.04686972898318268, "mem_mb": 87.387136, "nll": -6.055163383483887, "elapsed_s": 309.6}204{"event": "val", "step": 4000, "val_nll": -6.449117189626103, "best": -6.449117189626103, "bad": 0, "mem_mb": 87.387136, "elapsed_s": 371.4}205{"event": "train", "step": 4020, "loss": -5.986039161682129, "grad_norm": 0.686004638671875, "lr": 0.00024015391210388054, "ex_per_s": 691.5595306532599, "data_frac": 0.03904122289407737, "mem_mb": 87.387136, "nll": -5.986039161682129, "elapsed_s": 372.0}206{"event": "train", "step": 4040, "loss": -6.129359722137451, "grad_norm": 0.7181258201599121, "lr": 0.00023954536545646676, "ex_per_s": 693.958981954372, "data_frac": 0.039015685762636955, "mem_mb": 87.387136, "nll": -6.129359722137451, "elapsed_s": 372.6}207{"event": "train", "step": 4060, "loss": -5.867975234985352, "grad_norm": 0.13763950765132904, "lr": 0.0002389346317357231, "ex_per_s": 696.3615486104752, "data_frac": 0.03898678301379426, "mem_mb": 87.387136, "nll": -5.867975234985352, "elapsed_s": 373.2}208{"event": "train", "step": 4080, "loss": -5.931249141693115, "grad_norm": 0.10752607882022858, "lr": 0.0002383217288598584, "ex_per_s": 698.7582378248593, "data_frac": 0.0389578422121085, "mem_mb": 87.387136, "nll": -5.931249141693115, "elapsed_s": 373.7}209{"event": "train", "step": 4100, "loss": -5.942954063415527, "grad_norm": 0.24355830252170563, "lr": 0.00023770667481072205, "ex_per_s": 701.14907706092, "data_frac": 0.038929087001875806, "mem_mb": 87.387136, "nll": -5.942954063415527, "elapsed_s": 374.3}210{"event": "train", "step": 4120, "loss": -6.1049933433532715, "grad_norm": 0.1316048502922058, "lr": 0.0002370894876332764, "ex_per_s": 703.5256434796992, "data_frac": 0.03890116942687034, "mem_mb": 87.387136, "nll": -6.1049933433532715, "elapsed_s": 374.8}211{"event": "train", "step": 4140, "loss": -5.9142746925354, "grad_norm": 0.6410803198814392, "lr": 0.00023647018543506706, "ex_per_s": 705.9008075668677, "data_frac": 0.03887234477279737, "mem_mb": 87.387136, "nll": -5.9142746925354, "elapsed_s": 375.4}212{"event": "train", "step": 4160, "loss": -6.134688854217529, "grad_norm": 0.08552515506744385, "lr": 0.00023584878638569206, "ex_per_s": 708.2706466711958, "data_frac": 0.03884413881445718, "mem_mb": 87.387136, "nll": -6.134688854217529, "elapsed_s": 375.9}213{"event": "train", "step": 4180, "loss": -5.971691608428955, "grad_norm": 0.08877632021903992, "lr": 0.00023522530871626857, "ex_per_s": 710.633843808173, "data_frac": 0.038815166057861464, "mem_mb": 87.387136, "nll": -5.971691608428955, "elapsed_s": 376.5}214{"event": "train", "step": 4200, "loss": -6.0277323722839355, "grad_norm": 0.17054620385169983, "lr": 0.0002345997707188979, "ex_per_s": 712.988029615549, "data_frac": 0.03878651340488445, "mem_mb": 87.387136, "nll": -6.0277323722839355, "elapsed_s": 377.0}215{"event": "train", "step": 4220, "loss": -5.965580940246582, "grad_norm": 0.36926400661468506, "lr": 0.00023397219074612914, "ex_per_s": 715.3370791609478, "data_frac": 0.03875834191708581, "mem_mb": 87.387136, "nll": -5.965580940246582, "elapsed_s": 377.6}216{"event": "train", "step": 4240, "loss": -5.995228290557861, "grad_norm": 0.3099658191204071, "lr": 0.00023334258721042025, "ex_per_s": 717.678319168089, "data_frac": 0.038729835745044267, "mem_mb": 87.387136, "nll": -5.995228290557861, "elapsed_s": 378.1}217{"event": "train", "step": 4260, "loss": -5.887146949768066, "grad_norm": 0.4585038423538208, "lr": 0.00023271097858359853, "ex_per_s": 720.0143838787465, "data_frac": 0.03870139739698353, "mem_mb": 87.387136, "nll": -5.887146949768066, "elapsed_s": 378.7}218{"event": "train", "step": 4280, "loss": -5.848607540130615, "grad_norm": 0.4130026698112488, "lr": 0.00023207738339631794, "ex_per_s": 722.3441705822468, "data_frac": 0.038672769745984555, "mem_mb": 87.387136, "nll": -5.848607540130615, "elapsed_s": 379.2}219{"event": "train", "step": 4300, "loss": -6.256242752075195, "grad_norm": 0.1399703323841095, "lr": 0.00023144182023751594, "ex_per_s": 724.6666546616365, "data_frac": 0.03864446787788004, "mem_mb": 87.387136, "nll": -6.256242752075195, "elapsed_s": 379.8}220{"event": "train", "step": 4320, "loss": -6.134541988372803, "grad_norm": 0.5883010029792786, "lr": 0.00023080430775386808, "ex_per_s": 726.9820494763195, "data_frac": 0.038616056519064745, "mem_mb": 87.387136, "nll": -6.134541988372803, "elapsed_s": 380.3}221{"event": "train", "step": 4340, "loss": -6.083139419555664, "grad_norm": 0.6268623471260071, "lr": 0.00023016486464924057, "ex_per_s": 729.2912000243898, "data_frac": 0.03858790496751687, "mem_mb": 87.387136, "nll": -6.083139419555664, "elapsed_s": 380.9}222{"event": "train", "step": 4360, "loss": -5.850640296936035, "grad_norm": 0.16011951863765717, "lr": 0.0002295235096841419, "ex_per_s": 731.5902368404388, "data_frac": 0.03856073032313977, "mem_mb": 87.387136, "nll": -5.850640296936035, "elapsed_s": 381.4}223{"event": "train", "step": 4380, "loss": -6.0041913986206055, "grad_norm": 0.4440540075302124, "lr": 0.00022888026167517227, "ex_per_s": 733.8875070799319, "data_frac": 0.03853258888412446, "mem_mb": 87.387136, "nll": -6.0041913986206055, "elapsed_s": 382.0}224{"event": "train", "step": 4400, "loss": -6.046944618225098, "grad_norm": 0.2893553078174591, "lr": 0.00022823513949447166, "ex_per_s": 736.1764566881603, "data_frac": 0.038505176832885374, "mem_mb": 87.387136, "nll": -6.046944618225098, "elapsed_s": 382.5}225{"event": "train", "step": 4420, "loss": -5.89694356918335, "grad_norm": 0.4767192304134369, "lr": 0.00022758816206916584, "ex_per_s": 738.4556390209762, "data_frac": 0.038477906922604985, "mem_mb": 87.387136, "nll": -5.89694356918335, "elapsed_s": 383.1}226{"event": "train", "step": 4440, "loss": -6.091466426849365, "grad_norm": 0.22425322234630585, "lr": 0.00022693934838081143, "ex_per_s": 740.7318626801018, "data_frac": 0.038450921592710095, "mem_mb": 87.387136, "nll": -6.091466426849365, "elapsed_s": 383.6}227{"event": "train", "step": 4460, "loss": -5.934288501739502, "grad_norm": 0.3642868399620056, "lr": 0.0002262887174648388, "ex_per_s": 743.0007995104309, "data_frac": 0.038423274360445986, "mem_mb": 87.387136, "nll": -5.934288501739502, "elapsed_s": 384.2}228{"event": "train", "step": 4480, "loss": -6.06217098236084, "grad_norm": 0.2389584481716156, "lr": 0.00022563628840999367, "ex_per_s": 745.2635255253018, "data_frac": 0.03839584283708816, "mem_mb": 87.387136, "nll": -6.06217098236084, "elapsed_s": 384.7}229{"event": "train", "step": 4500, "loss": -6.149658679962158, "grad_norm": 1.0027055740356445, "lr": 0.000224982080357777, "ex_per_s": 747.5171238237401, "data_frac": 0.03836846927294375, "mem_mb": 87.387136, "nll": -6.149658679962158, "elapsed_s": 385.3}230{"event": "train", "step": 4520, "loss": -6.017146110534668, "grad_norm": 0.36344701051712036, "lr": 0.00022432611250188336, "ex_per_s": 749.7650376626206, "data_frac": 0.038341926560239686, "mem_mb": 87.387136, "nll": -6.017146110534668, "elapsed_s": 385.8}231{"event": "train", "step": 4540, "loss": -6.012322902679443, "grad_norm": 0.09145653992891312, "lr": 0.00022366840408763816, "ex_per_s": 752.0080301007935, "data_frac": 0.038314493425398416, "mem_mb": 87.387136, "nll": -6.012322902679443, "elapsed_s": 386.4}232{"event": "train", "step": 4560, "loss": -6.044816493988037, "grad_norm": 0.550759494304657, "lr": 0.00022300897441143247, "ex_per_s": 754.2426072182261, "data_frac": 0.03828754163017105, "mem_mb": 87.387136, "nll": -6.044816493988037, "elapsed_s": 386.9}233{"event": "train", "step": 4580, "loss": -6.019803047180176, "grad_norm": 1.1254035234451294, "lr": 0.0002223478428201573, "ex_per_s": 756.4713717174593, "data_frac": 0.038260592760330295, "mem_mb": 87.387136, "nll": -6.019803047180176, "elapsed_s": 387.5}234{"event": "train", "step": 4600, "loss": -5.8565216064453125, "grad_norm": 0.5585300922393799, "lr": 0.0002216850287106357, "ex_per_s": 758.6956795639754, "data_frac": 0.03823338608860247, "mem_mb": 87.387136, "nll": -5.8565216064453125, "elapsed_s": 388.0}235{"event": "train", "step": 4620, "loss": -6.028345584869385, "grad_norm": 0.18680547177791595, "lr": 0.00022102055152905406, "ex_per_s": 760.9143342223406, "data_frac": 0.03820623491623454, "mem_mb": 87.387136, "nll": -6.028345584869385, "elapsed_s": 388.6}236{"event": "train", "step": 4640, "loss": -5.834449291229248, "grad_norm": 0.2808162271976471, "lr": 0.00022035443077039114, "ex_per_s": 763.1262984584394, "data_frac": 0.03817988572626791, "mem_mb": 87.387136, "nll": -5.834449291229248, "elapsed_s": 389.1}237{"event": "train", "step": 4660, "loss": -5.980545520782471, "grad_norm": 0.4078221023082733, "lr": 0.00021968668597784646, "ex_per_s": 765.3271927996351, "data_frac": 0.03815343830965972, "mem_mb": 87.387136, "nll": -5.980545520782471, "elapsed_s": 389.7}238{"event": "train", "step": 4680, "loss": -6.031014442443848, "grad_norm": 0.21940258145332336, "lr": 0.00021901733674226665, "ex_per_s": 767.5259346164739, "data_frac": 0.038126793143187884, "mem_mb": 87.387136, "nll": -6.031014442443848, "elapsed_s": 390.3}239{"event": "train", "step": 4700, "loss": -5.909234523773193, "grad_norm": 0.2931768298149109, "lr": 0.00021834640270157097, "ex_per_s": 769.7184807184032, "data_frac": 0.03810032729488143, "mem_mb": 87.387136, "nll": -5.909234523773193, "elapsed_s": 390.8}240{"event": "train", "step": 4720, "loss": -6.067203521728516, "grad_norm": 0.1580803096294403, "lr": 0.00021767390354017485, "ex_per_s": 771.8977502651653, "data_frac": 0.03807439736960904, "mem_mb": 87.387136, "nll": -6.067203521728516, "elapsed_s": 391.4}241{"event": "train", "step": 4740, "loss": -6.0441083908081055, "grad_norm": 0.222369983792305, "lr": 0.0002169998589884126, "ex_per_s": 774.0638498105177, "data_frac": 0.03805030484724514, "mem_mb": 87.387136, "nll": -6.0441083908081055, "elapsed_s": 391.9}242{"event": "train", "step": 4760, "loss": -5.99399995803833, "grad_norm": 0.5899275541305542, "lr": 0.00021632428882195843, "ex_per_s": 776.226990618684, "data_frac": 0.03802568770019038, "mem_mb": 87.387136, "nll": -5.99399995803833, "elapsed_s": 392.5}243{"event": "train", "step": 4780, "loss": -6.061189651489258, "grad_norm": 0.9550632238388062, "lr": 0.00021564721286124628, "ex_per_s": 778.3955043792553, "data_frac": 0.03799971634628856, "mem_mb": 87.387136, "nll": -6.061189651489258, "elapsed_s": 393.0}244{"event": "train", "step": 4800, "loss": -6.112143516540527, "grad_norm": 0.32626667618751526, "lr": 0.00021496865097088845, "ex_per_s": 780.5605879076636, "data_frac": 0.03797380814788955, "mem_mb": 87.387136, "nll": -6.112143516540527, "elapsed_s": 393.6}245{"event": "train", "step": 4820, "loss": -6.031839847564697, "grad_norm": 0.2864052951335907, "lr": 0.00021428862305909247, "ex_per_s": 782.7193394009147, "data_frac": 0.03794739606262969, "mem_mb": 87.387136, "nll": -6.031839847564697, "elapsed_s": 394.1}246{"event": "train", "step": 4840, "loss": -5.935202598571777, "grad_norm": 0.13767454028129578, "lr": 0.00021360714907707733, "ex_per_s": 784.8723177489026, "data_frac": 0.037921390062463804, "mem_mb": 87.387136, "nll": -5.935202598571777, "elapsed_s": 394.7}247{"event": "train", "step": 4860, "loss": -5.925395965576172, "grad_norm": 0.48247814178466797, "lr": 0.00021292424901848804, "ex_per_s": 787.01767174808, "data_frac": 0.03789484803469302, "mem_mb": 87.387136, "nll": -5.925395965576172, "elapsed_s": 395.2}248{"event": "train", "step": 4880, "loss": -6.026703357696533, "grad_norm": 0.6323076486587524, "lr": 0.00021223994291880895, "ex_per_s": 789.1587120748376, "data_frac": 0.037869354496855495, "mem_mb": 87.387136, "nll": -6.026703357696533, "elapsed_s": 395.8}249{"event": "train", "step": 4900, "loss": -5.95719051361084, "grad_norm": 0.2440064549446106, "lr": 0.00021155425085477597, "ex_per_s": 791.2930870971331, "data_frac": 0.03784356671522794, "mem_mb": 87.387136, "nll": -5.95719051361084, "elapsed_s": 396.3}250{"event": "train", "step": 4920, "loss": -5.948297023773193, "grad_norm": 0.19307172298431396, "lr": 0.00021086719294378776, "ex_per_s": 793.4214559548027, "data_frac": 0.037817853384458096, "mem_mb": 87.387136, "nll": -5.948297023773193, "elapsed_s": 396.9}251{"event": "train", "step": 4940, "loss": -6.045091152191162, "grad_norm": 0.2634351849555969, "lr": 0.00021017878934331512, "ex_per_s": 795.5440753039901, "data_frac": 0.03779218865006119, "mem_mb": 87.387136, "nll": -6.045091152191162, "elapsed_s": 397.4}252{"event": "train", "step": 4960, "loss": -6.066006183624268, "grad_norm": 0.1875029057264328, "lr": 0.00020948906025030994, "ex_per_s": 797.6611584137222, "data_frac": 0.037766506324180146, "mem_mb": 87.387136, "nll": -6.066006183624268, "elapsed_s": 398.0}253{"event": "train", "step": 4980, "loss": -6.075031757354736, "grad_norm": 0.07519250363111496, "lr": 0.00020879802590061233, "ex_per_s": 799.7722356128139, "data_frac": 0.03774084290517448, "mem_mb": 87.387136, "nll": -6.075031757354736, "elapsed_s": 398.5}254{"event": "train", "step": 5000, "loss": -5.882948875427246, "grad_norm": 0.5079578161239624, "lr": 0.00020810570656835728, "ex_per_s": 801.878301534912, "data_frac": 0.03771568491959038, "mem_mb": 87.387136, "nll": -5.882948875427246, "elapsed_s": 399.1}255{"event": "val", "step": 5000, "val_nll": -6.450388795802899, "best": -6.450388795802899, "bad": 0, "mem_mb": 87.387136, "elapsed_s": 460.9}256{"event": "train", "step": 5020, "loss": -5.965717792510986, "grad_norm": 0.1806696057319641, "lr": 0.0002074121225653795, "ex_per_s": 696.2513881196535, "data_frac": 0.03264089182461191, "mem_mb": 87.387136, "nll": -5.965717792510986, "elapsed_s": 461.5}257{"event": "train", "step": 5040, "loss": -6.0916314125061035, "grad_norm": 0.4662337601184845, "lr": 0.00020671729424061788, "ex_per_s": 698.1907300048658, "data_frac": 0.03262505332529232, "mem_mb": 87.387136, "nll": -6.0916314125061035, "elapsed_s": 462.0}258{"event": "train", "step": 5060, "loss": -6.008965015411377, "grad_norm": 0.2884875237941742, "lr": 0.00020602124197951796, "ex_per_s": 700.1249486297081, "data_frac": 0.03260931496897575, "mem_mb": 87.387136, "nll": -6.008965015411377, "elapsed_s": 462.6}259{"event": "train", "step": 5080, "loss": -6.130163669586182, "grad_norm": 0.15827839076519012, "lr": 0.00020532398620343433, "ex_per_s": 702.0430929603808, "data_frac": 0.03259574406008086, "mem_mb": 87.387136, "nll": -6.130163669586182, "elapsed_s": 463.1}260{"event": "train", "step": 5100, "loss": -6.062674522399902, "grad_norm": 0.15082833170890808, "lr": 0.0002046255473690312, "ex_per_s": 703.9665165420315, "data_frac": 0.03258002247597713, "mem_mb": 87.387136, "nll": -6.062674522399902, "elapsed_s": 463.7}261{"event": "train", "step": 5120, "loss": -6.0026535987854, "grad_norm": 0.0946756899356842, "lr": 0.00020392594596768237, "ex_per_s": 705.8872439364125, "data_frac": 0.032564155708591974, "mem_mb": 87.387136, "nll": -6.0026535987854, "elapsed_s": 464.2}262{"event": "train", "step": 5140, "loss": -5.945266246795654, "grad_norm": 1.198238730430603, "lr": 0.00020322520252486997, "ex_per_s": 707.8033303174021, "data_frac": 0.03254880561180617, "mem_mb": 87.387136, "nll": -5.945266246795654, "elapsed_s": 464.8}263{"event": "train", "step": 5160, "loss": -6.165943145751953, "grad_norm": 0.39935043454170227, "lr": 0.0002025233375995822, "ex_per_s": 709.7156017406315, "data_frac": 0.03253315283840343, "mem_mb": 87.387136, "nll": -6.165943145751953, "elapsed_s": 465.3}264{"event": "train", "step": 5180, "loss": -5.938140392303467, "grad_norm": 0.32019689679145813, "lr": 0.0002018203717837105, "ex_per_s": 711.6225178005853, "data_frac": 0.03251782776060027, "mem_mb": 87.387136, "nll": -5.938140392303467, "elapsed_s": 465.9}265{"event": "train", "step": 5200, "loss": -5.826654434204102, "grad_norm": 0.08381716907024384, "lr": 0.00020111632570144484, "ex_per_s": 713.5251102714267, "data_frac": 0.03250202920751243, "mem_mb": 87.387136, "nll": -5.826654434204102, "elapsed_s": 466.4}266{"event": "train", "step": 5220, "loss": -5.924322128295898, "grad_norm": 0.2462642639875412, "lr": 0.00020041122000866903, "ex_per_s": 715.4238518432414, "data_frac": 0.03248590769672025, "mem_mb": 87.387136, "nll": -5.924322128295898, "elapsed_s": 467.0}267{"event": "train", "step": 5240, "loss": -5.778026580810547, "grad_norm": 0.08111827820539474, "lr": 0.00019970507539235469, "ex_per_s": 717.3181752415527, "data_frac": 0.03247037970674235, "mem_mb": 87.387136, "nll": -5.778026580810547, "elapsed_s": 467.5}268{"event": "train", "step": 5260, "loss": -5.945707321166992, "grad_norm": 0.14539682865142822, "lr": 0.0001989979125699542, "ex_per_s": 719.2079925074077, "data_frac": 0.03245480733887223, "mem_mb": 87.387136, "nll": -5.945707321166992, "elapsed_s": 468.1}269{"event": "train", "step": 5280, "loss": -6.000999450683594, "grad_norm": 0.9808164834976196, "lr": 0.00019828975228879288, "ex_per_s": 721.0936651503512, "data_frac": 0.03243914221133638, "mem_mb": 87.387136, "nll": -6.000999450683594, "elapsed_s": 468.6}270{"event": "train", "step": 5300, "loss": -5.89147424697876, "grad_norm": 0.2847606837749481, "lr": 0.00019758061532546025, "ex_per_s": 722.973197930945, "data_frac": 0.03242381322809763, "mem_mb": 87.387136, "nll": -5.89147424697876, "elapsed_s": 469.2}271{"event": "train", "step": 5320, "loss": -6.074265003204346, "grad_norm": 0.7212088108062744, "lr": 0.0001968705224852007, "ex_per_s": 724.8475520656987, "data_frac": 0.03240846601323826, "mem_mb": 87.387136, "nll": -6.074265003204346, "elapsed_s": 469.7}272{"event": "train", "step": 5340, "loss": -5.974186420440674, "grad_norm": 0.5298534631729126, "lr": 0.00019615949460130276, "ex_per_s": 726.7201806637977, "data_frac": 0.03239304264037756, "mem_mb": 87.387136, "nll": -5.974186420440674, "elapsed_s": 470.3}273{"event": "train", "step": 5360, "loss": -6.008090019226074, "grad_norm": 0.6200326681137085, "lr": 0.00019544755253448823, "ex_per_s": 728.58810763481, "data_frac": 0.032377598138225806, "mem_mb": 87.387136, "nll": -6.008090019226074, "elapsed_s": 470.8}274{"event": "train", "step": 5380, "loss": -5.858450412750244, "grad_norm": 0.2090979367494583, "lr": 0.0001947347171722997, "ex_per_s": 730.447036993103, "data_frac": 0.0323632177459336, "mem_mb": 87.387136, "nll": -5.858450412750244, "elapsed_s": 471.4}275{"event": "train", "step": 5400, "loss": -5.953564167022705, "grad_norm": 0.17011505365371704, "lr": 0.00019402100942848827, "ex_per_s": 732.306513244157, "data_frac": 0.03234801154808645, "mem_mb": 87.387136, "nll": -5.953564167022705, "elapsed_s": 471.9}276{"event": "train", "step": 5420, "loss": -5.838059425354004, "grad_norm": 0.2692638039588928, "lr": 0.00019330645024239946, "ex_per_s": 734.1476470099142, "data_frac": 0.03233434228768761, "mem_mb": 87.387136, "nll": -5.838059425354004, "elapsed_s": 472.5}277{"event": "train", "step": 5440, "loss": -6.019412040710449, "grad_norm": 0.7226775884628296, "lr": 0.00019259106057835926, "ex_per_s": 735.9897070045204, "data_frac": 0.032320447090656124, "mem_mb": 87.387136, "nll": -6.019412040710449, "elapsed_s": 473.1}278{"event": "train", "step": 5460, "loss": -6.100400924682617, "grad_norm": 0.15304264426231384, "lr": 0.00019187486142505883, "ex_per_s": 737.8356910662272, "data_frac": 0.032305145731437314, "mem_mb": 87.387136, "nll": -6.100400924682617, "elapsed_s": 473.6}279{"event": "train", "step": 5480, "loss": -6.100403785705566, "grad_norm": 0.30742397904396057, "lr": 0.00019115787379493887, "ex_per_s": 739.6780550343208, "data_frac": 0.03228985701403385, "mem_mb": 87.387136, "nll": -6.100403785705566, "elapsed_s": 474.2}280{"event": "train", "step": 5500, "loss": -6.080921649932861, "grad_norm": 0.6839185953140259, "lr": 0.00019044011872357296, "ex_per_s": 741.5160322028825, "data_frac": 0.03227473154203729, "mem_mb": 87.387136, "nll": -6.080921649932861, "elapsed_s": 474.7}281{"event": "train", "step": 5520, "loss": -5.926772594451904, "grad_norm": 0.2653290331363678, "lr": 0.00018972161726905055, "ex_per_s": 743.3506259695039, "data_frac": 0.032259726891941405, "mem_mb": 87.387136, "nll": -5.926772594451904, "elapsed_s": 475.3}282{"event": "train", "step": 5540, "loss": -5.919389247894287, "grad_norm": 0.7642732262611389, "lr": 0.00018900239051135913, "ex_per_s": 745.1814374788495, "data_frac": 0.03224480951093631, "mem_mb": 87.387136, "nll": -5.919389247894287, "elapsed_s": 475.8}283{"event": "train", "step": 5560, "loss": -5.976914405822754, "grad_norm": 0.7473878264427185, "lr": 0.00018828245955176568, "ex_per_s": 747.0063984650501, "data_frac": 0.032229630358982415, "mem_mb": 87.387136, "nll": -5.976914405822754, "elapsed_s": 476.4}284{"event": "train", "step": 5580, "loss": -6.031469345092773, "grad_norm": 0.32219865918159485, "lr": 0.00018756184551219765, "ex_per_s": 748.8271267030433, "data_frac": 0.032214598627991003, "mem_mb": 87.387136, "nll": -6.031469345092773, "elapsed_s": 476.9}285{"event": "train", "step": 5600, "loss": -6.083838939666748, "grad_norm": 0.547707200050354, "lr": 0.00018684056953462326, "ex_per_s": 750.6452477835737, "data_frac": 0.03219982036728943, "mem_mb": 87.387136, "nll": -6.083838939666748, "elapsed_s": 477.5}286{"event": "train", "step": 5620, "loss": -5.820559501647949, "grad_norm": 0.41368213295936584, "lr": 0.00018611865278043117, "ex_per_s": 752.4587068263926, "data_frac": 0.03218478037610986, "mem_mb": 87.387136, "nll": -5.820559501647949, "elapsed_s": 478.0}287{"event": "train", "step": 5640, "loss": -6.116333484649658, "grad_norm": 1.1059784889221191, "lr": 0.00018539611642980975, "ex_per_s": 754.2672079556728, "data_frac": 0.03216992328005768, "mem_mb": 87.387136, "nll": -6.116333484649658, "elapsed_s": 478.6}288{"event": "train", "step": 5660, "loss": -6.0049309730529785, "grad_norm": 0.40857964754104614, "lr": 0.00018467298168112556, "ex_per_s": 756.0719684841447, "data_frac": 0.03215514252804699, "mem_mb": 87.387136, "nll": -6.0049309730529785, "elapsed_s": 479.1}289{"event": "train", "step": 5680, "loss": -5.967125415802002, "grad_norm": 0.21865764260292053, "lr": 0.00018394926975030145, "ex_per_s": 757.872950040827, "data_frac": 0.03214055084989943, "mem_mb": 87.387136, "nll": -5.967125415802002, "elapsed_s": 479.7}290{"event": "train", "step": 5700, "loss": -6.032834053039551, "grad_norm": 0.4793780446052551, "lr": 0.00018322500187019415, "ex_per_s": 759.6685916554392, "data_frac": 0.03212580733698853, "mem_mb": 87.387136, "nll": -6.032834053039551, "elapsed_s": 480.2}291{"event": "train", "step": 5720, "loss": -5.833492755889893, "grad_norm": 0.4515489935874939, "lr": 0.00018250019928997124, "ex_per_s": 761.4597784414115, "data_frac": 0.0321109018865362, "mem_mb": 87.387136, "nll": -5.833492755889893, "elapsed_s": 480.8}292{"event": "train", "step": 5740, "loss": -5.998956680297852, "grad_norm": 0.3860251307487488, "lr": 0.00018177488327448776, "ex_per_s": 763.2472862851932, "data_frac": 0.03209588400045206, "mem_mb": 87.387136, "nll": -5.998956680297852, "elapsed_s": 481.3}293{"event": "train", "step": 5760, "loss": -5.996168613433838, "grad_norm": 0.2783406972885132, "lr": 0.00018104907510366235, "ex_per_s": 765.0324726695362, "data_frac": 0.032081263038943986, "mem_mb": 87.387136, "nll": -5.996168613433838, "elapsed_s": 481.9}294{"event": "train", "step": 5780, "loss": -6.009420871734619, "grad_norm": 0.4022574722766876, "lr": 0.00018032279607185304, "ex_per_s": 766.8121369730289, "data_frac": 0.032066396766903385, "mem_mb": 87.387136, "nll": -6.009420871734619, "elapsed_s": 482.4}295{"event": "train", "step": 5800, "loss": -5.982061862945557, "grad_norm": 0.9659512042999268, "lr": 0.00017959606748723213, "ex_per_s": 768.5885277064065, "data_frac": 0.032051741358750184, "mem_mb": 87.387136, "nll": -5.982061862945557, "elapsed_s": 483.0}296{"event": "train", "step": 5820, "loss": -5.906925201416016, "grad_norm": 0.4662586748600006, "lr": 0.00017886891067116147, "ex_per_s": 770.3607509673365, "data_frac": 0.03203757228492572, "mem_mb": 87.387136, "nll": -5.906925201416016, "elapsed_s": 483.5}297{"event": "train", "step": 5840, "loss": -5.905411243438721, "grad_norm": 0.8098547458648682, "lr": 0.00017814134695756664, "ex_per_s": 772.1288429436464, "data_frac": 0.0320228279190757, "mem_mb": 87.387136, "nll": -5.905411243438721, "elapsed_s": 484.1}298{"event": "train", "step": 5860, "loss": -6.00360107421875, "grad_norm": 0.13891299068927765, "lr": 0.00017741339769231104, "ex_per_s": 773.891848815916, "data_frac": 0.032008259802752285, "mem_mb": 87.387136, "nll": -6.00360107421875, "elapsed_s": 484.6}299{"event": "train", "step": 5880, "loss": -6.0144877433776855, "grad_norm": 0.23181936144828796, "lr": 0.00017668508423256986, "ex_per_s": 775.6509815861161, "data_frac": 0.03199367818696006, "mem_mb": 87.387136, "nll": -6.0144877433776855, "elapsed_s": 485.2}300{"event": "train", "step": 5900, "loss": -6.1373419761657715, "grad_norm": 0.5776628255844116, "lr": 0.0001759564279462032, "ex_per_s": 777.4059351836173, "data_frac": 0.031979076812228176, "mem_mb": 87.387136, "nll": -6.1373419761657715, "elapsed_s": 485.7}301{"event": "train", "step": 5920, "loss": -6.001518249511719, "grad_norm": 0.09498699754476547, "lr": 0.00017522745021112944, "ex_per_s": 779.1577574550221, "data_frac": 0.03196465510907732, "mem_mb": 87.387136, "nll": -6.001518249511719, "elapsed_s": 486.3}302{"event": "train", "step": 5940, "loss": -5.940080165863037, "grad_norm": 0.10419271141290665, "lr": 0.0001744981724146976, "ex_per_s": 780.9061109753519, "data_frac": 0.031950277468675625, "mem_mb": 87.387136, "nll": -5.940080165863037, "elapsed_s": 486.8}303{"event": "train", "step": 5960, "loss": -5.904377460479736, "grad_norm": 0.3416885733604431, "lr": 0.00017376861595306043, "ex_per_s": 782.6469876510582, "data_frac": 0.03193649768436896, "mem_mb": 87.387136, "nll": -5.904377460479736, "elapsed_s": 487.4}304{"event": "train", "step": 5980, "loss": -6.031105995178223, "grad_norm": 0.5392665863037109, "lr": 0.00017303880223054632, "ex_per_s": 784.3829922135309, "data_frac": 0.03192323660410068, "mem_mb": 87.387136, "nll": -6.031105995178223, "elapsed_s": 487.9}305{"event": "train", "step": 6000, "loss": -5.9906392097473145, "grad_norm": 0.16624507308006287, "lr": 0.00017230875265903138, "ex_per_s": 786.1163501031435, "data_frac": 0.03190948960442683, "mem_mb": 87.387136, "nll": -5.9906392097473145, "elapsed_s": 488.5}306{"event": "val", "step": 6000, "val_nll": -6.450991928571213, "best": -6.450991928571213, "bad": 0, "mem_mb": 87.387136, "elapsed_s": 550.4}307{"event": "train", "step": 6020, "loss": -5.8482537269592285, "grad_norm": 0.20637978613376617, "lr": 0.0001715784886573111, "ex_per_s": 699.2970335903311, "data_frac": 0.02831138148675874, "mem_mb": 87.387136, "nll": -5.8482537269592285, "elapsed_s": 551.0}308{"event": "train", "step": 6040, "loss": -5.862753391265869, "grad_norm": 0.681307852268219, "lr": 0.0001708480316504724, "ex_per_s": 700.8940625339218, "data_frac": 0.028301697090386994, "mem_mb": 87.387136, "nll": -5.862753391265869, "elapsed_s": 551.5}309{"event": "train", "step": 6060, "loss": -6.024799823760986, "grad_norm": 0.5486758351325989, "lr": 0.00017011740306926445, "ex_per_s": 702.5120060898098, "data_frac": 0.028292797300175445, "mem_mb": 87.387136, "nll": -6.024799823760986, "elapsed_s": 552.1}310{"event": "train", "step": 6080, "loss": -6.005685329437256, "grad_norm": 0.20361237227916718, "lr": 0.00016938662434947037, "ex_per_s": 704.1267144987787, "data_frac": 0.028283789148421643, "mem_mb": 87.387136, "nll": -6.005685329437256, "elapsed_s": 552.6}311{"event": "train", "step": 6100, "loss": -6.136325359344482, "grad_norm": 0.7500337958335876, "lr": 0.00016865571693127818, "ex_per_s": 705.7216436115889, "data_frac": 0.0282781829913139, "mem_mb": 87.387136, "nll": -6.136325359344482, "elapsed_s": 553.2}312{"event": "train", "step": 6120, "loss": -5.992982387542725, "grad_norm": 0.23407310247421265, "lr": 0.00016792470225865166, "ex_per_s": 707.3247740429775, "data_frac": 0.028271062148220342, "mem_mb": 87.387136, "nll": -5.992982387542725, "elapsed_s": 553.8}313{"event": "train", "step": 6140, "loss": -6.000748634338379, "grad_norm": 0.26489314436912537, "lr": 0.00016719360177870138, "ex_per_s": 708.9310635185964, "data_frac": 0.028262338669193504, "mem_mb": 87.387136, "nll": -6.000748634338379, "elapsed_s": 554.3}314{"event": "train", "step": 6160, "loss": -6.06650447845459, "grad_norm": 0.36487263441085815, "lr": 0.0001664624369410554, "ex_per_s": 710.5330955420861, "data_frac": 0.02825324757434177, "mem_mb": 87.387136, "nll": -6.06650447845459, "elapsed_s": 554.9}315{"event": "train", "step": 6180, "loss": -6.042501449584961, "grad_norm": 0.4021764397621155, "lr": 0.00016573122919722992, "ex_per_s": 712.1326782838368, "data_frac": 0.028244575805511155, "mem_mb": 87.387136, "nll": -6.042501449584961, "elapsed_s": 555.4}316{"event": "train", "step": 6200, "loss": -5.990609169006348, "grad_norm": 0.1374845653772354, "lr": 0.000165, "ex_per_s": 713.7294086133439, "data_frac": 0.028235752873578015, "mem_mb": 87.387136, "nll": -5.990609169006348, "elapsed_s": 556.0}317{"event": "train", "step": 6220, "loss": -5.93215799331665, "grad_norm": 0.10865644365549088, "lr": 0.00016426877080277007, "ex_per_s": 715.322387259641, "data_frac": 0.028226925390867078, "mem_mb": 87.387136, "nll": -5.93215799331665, "elapsed_s": 556.5}318{"event": "train", "step": 6240, "loss": -5.904057502746582, "grad_norm": 0.49983176589012146, "lr": 0.0001635375630589446, "ex_per_s": 716.912064799418, "data_frac": 0.028218148936193496, "mem_mb": 87.387136, "nll": -5.904057502746582, "elapsed_s": 557.1}319{"event": "train", "step": 6260, "loss": -5.910193920135498, "grad_norm": 0.10925904661417007, "lr": 0.00016280639822129864, "ex_per_s": 718.4994368083671, "data_frac": 0.02820931288193875, "mem_mb": 87.387136, "nll": -5.910193920135498, "elapsed_s": 557.6}320{"event": "train", "step": 6280, "loss": -5.90048885345459, "grad_norm": 0.6607616543769836, "lr": 0.00016207529774134834, "ex_per_s": 720.0842877434826, "data_frac": 0.02820043749404479, "mem_mb": 87.387136, "nll": -5.90048885345459, "elapsed_s": 558.2}321{"event": "train", "step": 6300, "loss": -5.982473373413086, "grad_norm": 0.16371466219425201, "lr": 0.00016134428306872187, "ex_per_s": 721.6648795794769, "data_frac": 0.02819183858118385, "mem_mb": 87.387136, "nll": -5.982473373413086, "elapsed_s": 558.7}322{"event": "train", "step": 6320, "loss": -5.820489406585693, "grad_norm": 0.5605255961418152, "lr": 0.00016061337565052966, "ex_per_s": 723.2423132365619, "data_frac": 0.02818281031756179, "mem_mb": 87.387136, "nll": -5.820489406585693, "elapsed_s": 559.3}323{"event": "train", "step": 6340, "loss": -5.8861894607543945, "grad_norm": 0.11822149902582169, "lr": 0.00015988259693073558, "ex_per_s": 724.8170314802378, "data_frac": 0.028174139566055037, "mem_mb": 87.387136, "nll": -5.8861894607543945, "elapsed_s": 559.8}324{"event": "train", "step": 6360, "loss": -6.164138317108154, "grad_norm": 0.2757340967655182, "lr": 0.0001591519683495276, "ex_per_s": 726.3887778393399, "data_frac": 0.02816518988413022, "mem_mb": 87.387136, "nll": -6.164138317108154, "elapsed_s": 560.4}325{"event": "train", "step": 6380, "loss": -5.897886753082275, "grad_norm": 0.22743238508701324, "lr": 0.0001584215113426889, "ex_per_s": 727.9586946737467, "data_frac": 0.02815638080270423, "mem_mb": 87.387136, "nll": -5.897886753082275, "elapsed_s": 560.9}326{"event": "train", "step": 6400, "loss": -5.887228012084961, "grad_norm": 0.37095341086387634, "lr": 0.00015769124734096862, "ex_per_s": 729.5186980641472, "data_frac": 0.028149241930626505, "mem_mb": 87.387136, "nll": -5.887228012084961, "elapsed_s": 561.5}327{"event": "train", "step": 6420, "loss": -5.984542369842529, "grad_norm": 0.4550502896308899, "lr": 0.00015696119776945368, "ex_per_s": 731.0819182791295, "data_frac": 0.028140254215317673, "mem_mb": 87.387136, "nll": -5.984542369842529, "elapsed_s": 562.0}328{"event": "train", "step": 6440, "loss": -5.976597785949707, "grad_norm": 0.42461809515953064, "lr": 0.00015623138404693952, "ex_per_s": 732.6410556572688, "data_frac": 0.02813146681346371, "mem_mb": 87.387136, "nll": -5.976597785949707, "elapsed_s": 562.6}329{"event": "train", "step": 6460, "loss": -5.928560733795166, "grad_norm": 0.5858632922172546, "lr": 0.0001555018275853024, "ex_per_s": 734.1959076908452, "data_frac": 0.028122709896877658, "mem_mb": 87.387136, "nll": -5.928560733795166, "elapsed_s": 563.1}330{"event": "train", "step": 6480, "loss": -5.9367523193359375, "grad_norm": 0.4919951260089874, "lr": 0.0001547725497888706, "ex_per_s": 735.7499776817953, "data_frac": 0.02811417885912995, "mem_mb": 87.387136, "nll": -5.9367523193359375, "elapsed_s": 563.7}331{"event": "train", "step": 6500, "loss": -5.956730842590332, "grad_norm": 0.07426013052463531, "lr": 0.0001540435720537968, "ex_per_s": 737.3008162723229, "data_frac": 0.028105332731604732, "mem_mb": 87.387136, "nll": -5.956730842590332, "elapsed_s": 564.2}332{"event": "train", "step": 6520, "loss": -6.039681911468506, "grad_norm": 0.13792191445827484, "lr": 0.0001533149157674301, "ex_per_s": 738.84794382959, "data_frac": 0.028096489502321496, "mem_mb": 87.387136, "nll": -6.039681911468506, "elapsed_s": 564.8}333{"event": "train", "step": 6540, "loss": -6.162737846374512, "grad_norm": 0.29489627480506897, "lr": 0.00015258660230768893, "ex_per_s": 740.3912023291035, "data_frac": 0.028087932257689067, "mem_mb": 87.387136, "nll": -6.162737846374512, "elapsed_s": 565.3}334{"event": "train", "step": 6560, "loss": -6.0482072830200195, "grad_norm": 0.6176924109458923, "lr": 0.0001518586530424334, "ex_per_s": 741.9326930885962, "data_frac": 0.028079379256686382, "mem_mb": 87.387136, "nll": -6.0482072830200195, "elapsed_s": 565.9}335{"event": "train", "step": 6580, "loss": -6.040205001831055, "grad_norm": 0.6630623936653137, "lr": 0.00015113108932883853, "ex_per_s": 743.4699636420033, "data_frac": 0.028071062310189774, "mem_mb": 87.387136, "nll": -6.040205001831055, "elapsed_s": 566.4}336{"event": "train", "step": 6600, "loss": -6.012368679046631, "grad_norm": 0.2605536878108978, "lr": 0.00015040393251276787, "ex_per_s": 745.004800980269, "data_frac": 0.028062412515580688, "mem_mb": 87.387136, "nll": -6.012368679046631, "elapsed_s": 567.0}337{"event": "train", "step": 6620, "loss": -6.2021613121032715, "grad_norm": 0.1239386647939682, "lr": 0.00014967720392814694, "ex_per_s": 746.536997044106, "data_frac": 0.028054640060892932, "mem_mb": 87.387136, "nll": -6.2021613121032715, "elapsed_s": 567.5}338{"event": "train", "step": 6640, "loss": -6.10203218460083, "grad_norm": 0.10553612560033798, "lr": 0.00014895092489633762, "ex_per_s": 748.0676221632069, "data_frac": 0.028046374492906975, "mem_mb": 87.387136, "nll": -6.10203218460083, "elapsed_s": 568.1}339{"event": "train", "step": 6660, "loss": -5.9736480712890625, "grad_norm": 0.11073289066553116, "lr": 0.00014822511672551224, "ex_per_s": 749.5948571344735, "data_frac": 0.02803830044268504, "mem_mb": 87.387136, "nll": -5.9736480712890625, "elapsed_s": 568.6}340{"event": "train", "step": 6680, "loss": -5.794349670410156, "grad_norm": 0.21222122013568878, "lr": 0.0001474998007100288, "ex_per_s": 751.1193720904921, "data_frac": 0.028029978823513692, "mem_mb": 87.387136, "nll": -5.794349670410156, "elapsed_s": 569.2}341{"event": "train", "step": 6700, "loss": -5.860685348510742, "grad_norm": 0.6802064776420593, "lr": 0.00014677499812980582, "ex_per_s": 752.6411996305851, "data_frac": 0.02802167569092721, "mem_mb": 87.387136, "nll": -5.860685348510742, "elapsed_s": 569.7}342{"event": "train", "step": 6720, "loss": -6.212716102600098, "grad_norm": 0.12068258225917816, "lr": 0.0001460507302496985, "ex_per_s": 754.1580250409938, "data_frac": 0.028013205154635394, "mem_mb": 87.387136, "nll": -6.212716102600098, "elapsed_s": 570.3}343{"event": "train", "step": 6740, "loss": -5.971323013305664, "grad_norm": 0.22996173799037933, "lr": 0.0001453270183188744, "ex_per_s": 755.6720278206798, "data_frac": 0.02800469232008012, "mem_mb": 87.387136, "nll": -5.971323013305664, "elapsed_s": 570.8}344{"event": "train", "step": 6760, "loss": -5.901308536529541, "grad_norm": 0.2864314317703247, "lr": 0.00014460388357019025, "ex_per_s": 757.1848853103046, "data_frac": 0.02799643350294194, "mem_mb": 87.387136, "nll": -5.901308536529541, "elapsed_s": 571.4}345{"event": "train", "step": 6780, "loss": -5.9979047775268555, "grad_norm": 0.43047767877578735, "lr": 0.0001438813472195688, "ex_per_s": 758.6933161815012, "data_frac": 0.02798799786844305, "mem_mb": 87.387136, "nll": -5.9979047775268555, "elapsed_s": 571.9}346{"event": "train", "step": 6800, "loss": -6.088345527648926, "grad_norm": 0.19597987830638885, "lr": 0.00014315943046537674, "ex_per_s": 760.2002270470028, "data_frac": 0.027979711047133406, "mem_mb": 87.387136, "nll": -6.088345527648926, "elapsed_s": 572.5}347{"event": "train", "step": 6820, "loss": -6.10868501663208, "grad_norm": 0.2608284652233124, "lr": 0.00014243815448780235, "ex_per_s": 761.6958726559168, "data_frac": 0.027973146839491196, "mem_mb": 87.387136, "nll": -6.10868501663208, "elapsed_s": 573.0}348{"event": "train", "step": 6840, "loss": -5.830193042755127, "grad_norm": 0.2952214181423187, "lr": 0.00014171754044823431, "ex_per_s": 763.1877305570182, "data_frac": 0.02796692700253854, "mem_mb": 87.387136, "nll": -5.830193042755127, "elapsed_s": 573.6}349{"event": "train", "step": 6860, "loss": -5.9489874839782715, "grad_norm": 0.12180054187774658, "lr": 0.0001409976094886409, "ex_per_s": 764.6849446942765, "data_frac": 0.02795845380425766, "mem_mb": 87.387136, "nll": -5.9489874839782715, "elapsed_s": 574.2}350{"event": "train", "step": 6880, "loss": -6.0232834815979, "grad_norm": 0.18995629251003265, "lr": 0.00014027838273094942, "ex_per_s": 766.1805710259314, "data_frac": 0.02795035974840283, "mem_mb": 87.387136, "nll": -6.0232834815979, "elapsed_s": 574.7}351{"event": "train", "step": 6900, "loss": -6.005333423614502, "grad_norm": 0.3579402267932892, "lr": 0.00013955988127642706, "ex_per_s": 767.6730087838196, "data_frac": 0.027942230634747475, "mem_mb": 87.387136, "nll": -6.005333423614502, "elapsed_s": 575.3}352{"event": "train", "step": 6920, "loss": -6.067220211029053, "grad_norm": 0.6126911640167236, "lr": 0.0001388421262050611, "ex_per_s": 769.1614431773645, "data_frac": 0.027933962602682678, "mem_mb": 87.387136, "nll": -6.067220211029053, "elapsed_s": 575.8}353{"event": "train", "step": 6940, "loss": -6.062272071838379, "grad_norm": 0.6467146277427673, "lr": 0.00013812513857494116, "ex_per_s": 770.6479703039279, "data_frac": 0.027925832913923504, "mem_mb": 87.387136, "nll": -6.062272071838379, "elapsed_s": 576.4}354{"event": "train", "step": 6960, "loss": -5.740734577178955, "grad_norm": 0.3165520131587982, "lr": 0.00013740893942164073, "ex_per_s": 772.1323285309231, "data_frac": 0.027917979656522898, "mem_mb": 87.387136, "nll": -5.740734577178955, "elapsed_s": 576.9}355{"event": "train", "step": 6980, "loss": -6.1976847648620605, "grad_norm": 0.26810377836227417, "lr": 0.0001366935497576006, "ex_per_s": 773.6131112974568, "data_frac": 0.02790953876897603, "mem_mb": 87.387136, "nll": -6.1976847648620605, "elapsed_s": 577.5}356{"event": "train", "step": 7000, "loss": -6.124111652374268, "grad_norm": 0.4738962948322296, "lr": 0.00013597899057151173, "ex_per_s": 775.0917686870716, "data_frac": 0.02790184637588485, "mem_mb": 87.387136, "nll": -6.124111652374268, "elapsed_s": 578.0}357{"event": "val", "step": 7000, "val_nll": -6.451378460830027, "best": -6.451378460830027, "bad": 0, "mem_mb": 87.387136, "elapsed_s": 639.8}358{"event": "train", "step": 7020, "loss": -6.145406246185303, "grad_norm": 0.3592066466808319, "lr": 0.0001352652828277003, "ex_per_s": 701.5966532436081, "data_frac": 0.02520130968006742, "mem_mb": 87.387136, "nll": -6.145406246185303, "elapsed_s": 640.4}359{"event": "train", "step": 7040, "loss": -6.0576348304748535, "grad_norm": 0.08366266638040543, "lr": 0.00013455244746551177, "ex_per_s": 702.9899123095878, "data_frac": 0.025196529156212295, "mem_mb": 87.387136, "nll": -6.0576348304748535, "elapsed_s": 640.9}360{"event": "train", "step": 7060, "loss": -5.989241123199463, "grad_norm": 0.8103774189949036, "lr": 0.00013384050539869724, "ex_per_s": 704.380336799942, "data_frac": 0.025191628935022244, "mem_mb": 87.387136, "nll": -5.989241123199463, "elapsed_s": 641.5}361{"event": "train", "step": 7080, "loss": -6.072149276733398, "grad_norm": 0.7666151523590088, "lr": 0.00013312947751479927, "ex_per_s": 705.7689486069011, "data_frac": 0.02518676818126654, "mem_mb": 87.387136, "nll": -6.072149276733398, "elapsed_s": 642.0}362{"event": "train", "step": 7100, "loss": -5.964911460876465, "grad_norm": 0.08737583458423615, "lr": 0.00013241938467453975, "ex_per_s": 707.1551171751165, "data_frac": 0.025181605690311443, "mem_mb": 87.387136, "nll": -5.964911460876465, "elapsed_s": 642.6}363{"event": "train", "step": 7120, "loss": -6.041417598724365, "grad_norm": 0.10068763792514801, "lr": 0.00013171024771120715, "ex_per_s": 708.5382868567999, "data_frac": 0.025176827761324787, "mem_mb": 87.387136, "nll": -6.041417598724365, "elapsed_s": 643.1}364{"event": "train", "step": 7140, "loss": -5.946241855621338, "grad_norm": 0.20865434408187866, "lr": 0.0001310020874300458, "ex_per_s": 709.9157735713792, "data_frac": 0.02517196436815736, "mem_mb": 87.387136, "nll": -5.946241855621338, "elapsed_s": 643.7}365{"event": "train", "step": 7160, "loss": -6.029646396636963, "grad_norm": 0.44283631443977356, "lr": 0.00013029492460764526, "ex_per_s": 711.2914925576032, "data_frac": 0.025168725477692007, "mem_mb": 87.387136, "nll": -6.029646396636963, "elapsed_s": 644.2}366{"event": "train", "step": 7180, "loss": -5.947843074798584, "grad_norm": 0.5925171971321106, "lr": 0.00012958877999133094, "ex_per_s": 712.6684459733574, "data_frac": 0.02516339112463613, "mem_mb": 87.387136, "nll": -5.947843074798584, "elapsed_s": 644.8}367{"event": "train", "step": 7200, "loss": -5.95745325088501, "grad_norm": 0.3017040193080902, "lr": 0.0001288836742985552, "ex_per_s": 714.0427126789662, "data_frac": 0.025158452013348098, "mem_mb": 87.387136, "nll": -5.95745325088501, "elapsed_s": 645.4}368{"event": "train", "step": 7220, "loss": -6.061054706573486, "grad_norm": 0.14479616284370422, "lr": 0.00012817962821628953, "ex_per_s": 715.4149021149428, "data_frac": 0.02515353858896623, "mem_mb": 87.387136, "nll": -6.061054706573486, "elapsed_s": 645.9}369{"event": "train", "step": 7240, "loss": -5.95499849319458, "grad_norm": 0.16110537946224213, "lr": 0.00012747666240041779, "ex_per_s": 716.7847364319447, "data_frac": 0.025148621243496816, "mem_mb": 87.387136, "nll": -5.95499849319458, "elapsed_s": 646.5}370{"event": "train", "step": 7260, "loss": -6.073244571685791, "grad_norm": 0.7027435898780823, "lr": 0.00012677479747513008, "ex_per_s": 718.1524721844238, "data_frac": 0.025143879009743013, "mem_mb": 87.387136, "nll": -6.073244571685791, "elapsed_s": 647.0}371{"event": "train", "step": 7280, "loss": -5.869675636291504, "grad_norm": 0.4892747402191162, "lr": 0.00012607405403231763, "ex_per_s": 719.5183547719735, "data_frac": 0.025139073825386854, "mem_mb": 87.387136, "nll": -5.869675636291504, "elapsed_s": 647.6}372{"event": "train", "step": 7300, "loss": -5.923182964324951, "grad_norm": 0.7605231404304504, "lr": 0.0001253744526309688, "ex_per_s": 720.8810274605654, "data_frac": 0.025134180719055862, "mem_mb": 87.387136, "nll": -5.923182964324951, "elapsed_s": 648.1}373{"event": "train", "step": 7320, "loss": -6.072534084320068, "grad_norm": 0.34328216314315796, "lr": 0.00012467601379656567, "ex_per_s": 722.2414928944547, "data_frac": 0.025129037695123646, "mem_mb": 87.387136, "nll": -6.072534084320068, "elapsed_s": 648.7}374{"event": "train", "step": 7340, "loss": -6.013071537017822, "grad_norm": 0.33346831798553467, "lr": 0.000123978758020482, "ex_per_s": 723.6000493715701, "data_frac": 0.025124203099048167, "mem_mb": 87.387136, "nll": -6.013071537017822, "elapsed_s": 649.2}375{"event": "train", "step": 7360, "loss": -5.992919445037842, "grad_norm": 0.4837609827518463, "lr": 0.00012328270575938212, "ex_per_s": 724.9557623209405, "data_frac": 0.025119324638471767, "mem_mb": 87.387136, "nll": -5.992919445037842, "elapsed_s": 649.8}376{"event": "train", "step": 7380, "loss": -6.076193809509277, "grad_norm": 0.26850518584251404, "lr": 0.00012258787743462041, "ex_per_s": 726.3096268037799, "data_frac": 0.02511445502891408, "mem_mb": 87.387136, "nll": -6.076193809509277, "elapsed_s": 650.3}377{"event": "train", "step": 7400, "loss": -6.001147270202637, "grad_norm": 0.26635468006134033, "lr": 0.00012189429343164274, "ex_per_s": 727.6608637007483, "data_frac": 0.025109383213205713, "mem_mb": 87.387136, "nll": -6.001147270202637, "elapsed_s": 650.9}378{"event": "train", "step": 7420, "loss": -6.050612926483154, "grad_norm": 0.1880367547273636, "lr": 0.00012120197409938763, "ex_per_s": 729.0067498395891, "data_frac": 0.02510528434678897, "mem_mb": 87.387136, "nll": -6.050612926483154, "elapsed_s": 651.4}379{"event": "train", "step": 7440, "loss": -5.7803874015808105, "grad_norm": 0.42726168036460876, "lr": 0.00012051093974969008, "ex_per_s": 730.3525806119513, "data_frac": 0.025100315088937886, "mem_mb": 87.387136, "nll": -5.7803874015808105, "elapsed_s": 652.0}380{"event": "train", "step": 7460, "loss": -6.214455604553223, "grad_norm": 0.1505008488893509, "lr": 0.00011982121065668482, "ex_per_s": 731.6972934656429, "data_frac": 0.025095422528736717, "mem_mb": 87.387136, "nll": -6.214455604553223, "elapsed_s": 652.5}381{"event": "train", "step": 7480, "loss": -5.96359920501709, "grad_norm": 0.32445189356803894, "lr": 0.00011913280705621224, "ex_per_s": 733.0401140376745, "data_frac": 0.025090481422016413, "mem_mb": 87.387136, "nll": -5.96359920501709, "elapsed_s": 653.1}382{"event": "train", "step": 7500, "loss": -5.777565956115723, "grad_norm": 0.11949358880519867, "lr": 0.00011844574914522403, "ex_per_s": 734.3755241834012, "data_frac": 0.025086460566782844, "mem_mb": 87.387136, "nll": -5.777565956115723, "elapsed_s": 653.6}383{"event": "train", "step": 7520, "loss": -6.06016206741333, "grad_norm": 0.6743895411491394, "lr": 0.00011776005708119105, "ex_per_s": 735.7067441148798, "data_frac": 0.02508263647976084, "mem_mb": 87.387136, "nll": -6.06016206741333, "elapsed_s": 654.2}384{"event": "train", "step": 7540, "loss": -5.842057704925537, "grad_norm": 0.3591616153717041, "lr": 0.00011707575098151199, "ex_per_s": 737.0424970927445, "data_frac": 0.025078136581162837, "mem_mb": 87.387136, "nll": -5.842057704925537, "elapsed_s": 654.7}385{"event": "train", "step": 7560, "loss": -5.902693271636963, "grad_norm": 0.3278430700302124, "lr": 0.00011639285092292267, "ex_per_s": 738.3762533904427, "data_frac": 0.025073058586643528, "mem_mb": 87.387136, "nll": -5.902693271636963, "elapsed_s": 655.3}386{"event": "train", "step": 7580, "loss": -5.869600296020508, "grad_norm": 0.240690216422081, "lr": 0.00011571137694090757, "ex_per_s": 739.7087683167987, "data_frac": 0.02506823456763216, "mem_mb": 87.387136, "nll": -5.869600296020508, "elapsed_s": 655.8}387{"event": "train", "step": 7600, "loss": -6.057699680328369, "grad_norm": 0.0535687617957592, "lr": 0.00011503134902911152, "ex_per_s": 741.0390759867863, "data_frac": 0.02506350922596906, "mem_mb": 87.387136, "nll": -6.057699680328369, "elapsed_s": 656.4}388{"event": "train", "step": 7620, "loss": -5.948822021484375, "grad_norm": 0.5082135796546936, "lr": 0.00011435278713875375, "ex_per_s": 742.3660523733686, "data_frac": 0.02505866525490062, "mem_mb": 87.387136, "nll": -5.948822021484375, "elapsed_s": 656.9}389{"event": "train", "step": 7640, "loss": -5.908698558807373, "grad_norm": 0.7556275129318237, "lr": 0.00011367571117804159, "ex_per_s": 743.6914337197013, "data_frac": 0.025053944871594916, "mem_mb": 87.387136, "nll": -5.908698558807373, "elapsed_s": 657.5}390{"event": "train", "step": 7660, "loss": -6.048027992248535, "grad_norm": 0.7106224298477173, "lr": 0.00011300014101158744, "ex_per_s": 745.0141913728521, "data_frac": 0.025049029221550972, "mem_mb": 87.387136, "nll": -6.048027992248535, "elapsed_s": 658.0}391{"event": "train", "step": 7680, "loss": -5.967001914978027, "grad_norm": 0.13099558651447296, "lr": 0.00011232609645982515, "ex_per_s": 746.3352373136054, "data_frac": 0.025043914864830406, "mem_mb": 87.387136, "nll": -5.967001914978027, "elapsed_s": 658.6}392{"event": "train", "step": 7700, "loss": -5.906396389007568, "grad_norm": 0.09452717006206512, "lr": 0.00011165359729842902, "ex_per_s": 747.654288573784, "data_frac": 0.02503936955028997, "mem_mb": 87.387136, "nll": -5.906396389007568, "elapsed_s": 659.1}393{"event": "train", "step": 7720, "loss": -6.035179615020752, "grad_norm": 0.4094908535480499, "lr": 0.00011098266325773331, "ex_per_s": 748.9702565892763, "data_frac": 0.02503472856420738, "mem_mb": 87.387136, "nll": -6.035179615020752, "elapsed_s": 659.7}394{"event": "train", "step": 7740, "loss": -6.042139530181885, "grad_norm": 0.9067054390907288, "lr": 0.00011031331402215353, "ex_per_s": 750.2846386509852, "data_frac": 0.025030130484011984, "mem_mb": 87.387136, "nll": -6.042139530181885, "elapsed_s": 660.2}395{"event": "train", "step": 7760, "loss": -5.964738845825195, "grad_norm": 0.48612549901008606, "lr": 0.00010964556922960883, "ex_per_s": 751.5960838233414, "data_frac": 0.02502512441671355, "mem_mb": 87.387136, "nll": -5.964738845825195, "elapsed_s": 660.8}396{"event": "train", "step": 7780, "loss": -6.0526323318481445, "grad_norm": 0.18920890986919403, "lr": 0.00010897944847094595, "ex_per_s": 752.9051503449501, "data_frac": 0.025020413005689104, "mem_mb": 87.387136, "nll": -6.0526323318481445, "elapsed_s": 661.3}397{"event": "train", "step": 7800, "loss": -6.0587334632873535, "grad_norm": 0.08214398473501205, "lr": 0.00010831497128936428, "ex_per_s": 754.2120444635509, "data_frac": 0.02501564486467053, "mem_mb": 87.387136, "nll": -6.0587334632873535, "elapsed_s": 661.9}398{"event": "train", "step": 7820, "loss": -5.901905059814453, "grad_norm": 0.08871162682771683, "lr": 0.00010765215717984271, "ex_per_s": 755.5179338318005, "data_frac": 0.025010934910182747, "mem_mb": 87.387136, "nll": -5.901905059814453, "elapsed_s": 662.4}399{"event": "train", "step": 7840, "loss": -5.956962585449219, "grad_norm": 0.16445249319076538, "lr": 0.00010699102558856754, "ex_per_s": 756.8221169350992, "data_frac": 0.02500584847370345, "mem_mb": 87.387136, "nll": -5.956962585449219, "elapsed_s": 663.0}400{"event": "train", "step": 7860, "loss": -6.117966175079346, "grad_norm": 0.38417962193489075, "lr": 0.0001063315959123618, "ex_per_s": 758.1238763380778, "data_frac": 0.025001111780656796, "mem_mb": 87.387136, "nll": -6.117966175079346, "elapsed_s": 663.5}401{"event": "train", "step": 7880, "loss": -6.064858436584473, "grad_norm": 0.08149275928735733, "lr": 0.0001056738874981166, "ex_per_s": 759.4233732107047, "data_frac": 0.024996511911545626, "mem_mb": 87.387136, "nll": -6.064858436584473, "elapsed_s": 664.1}402{"event": "train", "step": 7900, "loss": -6.055647850036621, "grad_norm": 0.18749792873859406, "lr": 0.00010501791964222304, "ex_per_s": 760.7207348788082, "data_frac": 0.02499199550815492, "mem_mb": 87.387136, "nll": -6.055647850036621, "elapsed_s": 664.6}403{"event": "train", "step": 7920, "loss": -6.083131313323975, "grad_norm": 0.12408920377492905, "lr": 0.00010436371159000636, "ex_per_s": 762.0147584127133, "data_frac": 0.024987152846430855, "mem_mb": 87.387136, "nll": -6.083131313323975, "elapsed_s": 665.2}404{"event": "train", "step": 7940, "loss": -6.047245025634766, "grad_norm": 0.11885043233633041, "lr": 0.00010371128253516118, "ex_per_s": 763.3077061868368, "data_frac": 0.024982218814793674, "mem_mb": 87.387136, "nll": -6.047245025634766, "elapsed_s": 665.7}405{"event": "train", "step": 7960, "loss": -5.992215156555176, "grad_norm": 0.3306664526462555, "lr": 0.00010306065161918857, "ex_per_s": 764.5975695859579, "data_frac": 0.024977524218037832, "mem_mb": 87.387136, "nll": -5.992215156555176, "elapsed_s": 666.3}406{"event": "train", "step": 7980, "loss": -5.838624000549316, "grad_norm": 0.08480846881866455, "lr": 0.00010241183793083414, "ex_per_s": 765.8852137532075, "data_frac": 0.024973044396484123, "mem_mb": 87.387136, "nll": -5.838624000549316, "elapsed_s": 666.8}407{"event": "train", "step": 8000, "loss": -6.050516605377197, "grad_norm": 0.06463274359703064, "lr": 0.00010176486050552833, "ex_per_s": 767.1715807748769, "data_frac": 0.024967950863989067, "mem_mb": 87.387136, "nll": -6.050516605377197, "elapsed_s": 667.4}408{"event": "val", "step": 8000, "val_nll": -6.451127785085121, "best": -6.451378460830027, "bad": 1, "mem_mb": 87.387136, "elapsed_s": 729.2}409{"event": "train", "step": 8020, "loss": -6.044818878173828, "grad_norm": 0.2859102189540863, "lr": 0.00010111973832482765, "ex_per_s": 703.2964327110292, "data_frac": 0.02284718021494215, "mem_mb": 87.387136, "nll": -6.044818878173828, "elapsed_s": 729.8}410{"event": "train", "step": 8040, "loss": -6.035453796386719, "grad_norm": 0.10678686946630478, "lr": 0.00010047649031585807, "ex_per_s": 704.5167650579751, "data_frac": 0.022844423210303576, "mem_mb": 87.387136, "nll": -6.035453796386719, "elapsed_s": 730.4}411{"event": "train", "step": 8060, "loss": -5.832866191864014, "grad_norm": 0.8170378804206848, "lr": 9.983513535075943e-05, "ex_per_s": 705.7354452766182, "data_frac": 0.02284153474215197, "mem_mb": 87.387136, "nll": -5.832866191864014, "elapsed_s": 730.9}412{"event": "train", "step": 8080, "loss": -6.005900859832764, "grad_norm": 0.10611089318990707, "lr": 9.919569224613192e-05, "ex_per_s": 706.9522488389171, "data_frac": 0.022838599288601425, "mem_mb": 87.387136, "nll": -6.005900859832764, "elapsed_s": 731.5}413{"event": "train", "step": 8100, "loss": -5.942192554473877, "grad_norm": 0.08372770249843597, "lr": 9.855817976248399e-05, "ex_per_s": 708.1674611987123, "data_frac": 0.02283600644191328, "mem_mb": 87.387136, "nll": -5.942192554473877, "elapsed_s": 732.0}414{"event": "train", "step": 8120, "loss": -5.902020454406738, "grad_norm": 0.6285110116004944, "lr": 9.792261660368207e-05, "ex_per_s": 709.3814639946536, "data_frac": 0.022833190855396965, "mem_mb": 87.387136, "nll": -5.902020454406738, "elapsed_s": 732.6}415{"event": "train", "step": 8140, "loss": -5.894786357879639, "grad_norm": 0.07740367203950882, "lr": 9.72890214164015e-05, "ex_per_s": 710.5929942444203, "data_frac": 0.02283058770574171, "mem_mb": 87.387136, "nll": -5.894786357879639, "elapsed_s": 733.1}416{"event": "train", "step": 8160, "loss": -6.042470455169678, "grad_norm": 0.24637894332408905, "lr": 9.66574127895797e-05, "ex_per_s": 711.8029285726047, "data_frac": 0.022827981816247817, "mem_mb": 87.387136, "nll": -6.042470455169678, "elapsed_s": 733.7}417{"event": "train", "step": 8180, "loss": -5.944397449493408, "grad_norm": 0.672427237033844, "lr": 9.602780925387089e-05, "ex_per_s": 713.0013880368803, "data_frac": 0.022827329115706615, "mem_mb": 87.387136, "nll": -5.944397449493408, "elapsed_s": 734.3}418{"event": "train", "step": 8200, "loss": -6.171375751495361, "grad_norm": 0.26455527544021606, "lr": 9.540022928110205e-05, "ex_per_s": 714.2033135381471, "data_frac": 0.02282557423202111, "mem_mb": 87.387136, "nll": -6.171375751495361, "elapsed_s": 734.8}419{"event": "train", "step": 8220, "loss": -5.905048370361328, "grad_norm": 0.10140577703714371, "lr": 9.477469128373143e-05, "ex_per_s": 715.4083822769588, "data_frac": 0.02282304656824924, "mem_mb": 87.387136, "nll": -5.905048370361328, "elapsed_s": 735.4}420{"event": "train", "step": 8240, "loss": -5.840516090393066, "grad_norm": 0.19177375733852386, "lr": 9.415121361430792e-05, "ex_per_s": 716.611648479743, "data_frac": 0.02282050887245568, "mem_mb": 87.387136, "nll": -5.840516090393066, "elapsed_s": 735.9}421{"event": "train", "step": 8260, "loss": -6.186091423034668, "grad_norm": 0.43508145213127136, "lr": 9.352981456493298e-05, "ex_per_s": 717.8133569611566, "data_frac": 0.022817716629376802, "mem_mb": 87.387136, "nll": -6.186091423034668, "elapsed_s": 736.5}422{"event": "train", "step": 8280, "loss": -6.153454303741455, "grad_norm": 0.17725025117397308, "lr": 9.291051236672361e-05, "ex_per_s": 719.0134840650497, "data_frac": 0.022815155395511426, "mem_mb": 87.387136, "nll": -6.153454303741455, "elapsed_s": 737.0}423{"event": "train", "step": 8300, "loss": -6.048291206359863, "grad_norm": 0.34130188822746277, "lr": 9.229332518927795e-05, "ex_per_s": 720.2111408905482, "data_frac": 0.022812563929743797, "mem_mb": 87.387136, "nll": -6.048291206359863, "elapsed_s": 737.6}424{"event": "train", "step": 8320, "loss": -5.882452964782715, "grad_norm": 0.3126510977745056, "lr": 9.167827114014155e-05, "ex_per_s": 721.4076934800166, "data_frac": 0.02280994182477743, "mem_mb": 87.387136, "nll": -5.882452964782715, "elapsed_s": 738.1}425{"event": "train", "step": 8340, "loss": -6.056857585906982, "grad_norm": 0.5709038972854614, "lr": 9.106536826427686e-05, "ex_per_s": 722.6024109267108, "data_frac": 0.022807240710596977, "mem_mb": 87.387136, "nll": -6.056857585906982, "elapsed_s": 738.7}426{"event": "train", "step": 8360, "loss": -6.0099053382873535, "grad_norm": 0.19273512065410614, "lr": 9.045463454353313e-05, "ex_per_s": 723.795061539513, "data_frac": 0.02280452784745561, "mem_mb": 87.387136, "nll": -6.0099053382873535, "elapsed_s": 739.2}427{"event": "train", "step": 8380, "loss": -6.0518035888671875, "grad_norm": 0.09001065045595169, "lr": 8.984608789611947e-05, "ex_per_s": 724.9859669996594, "data_frac": 0.022801766226992603, "mem_mb": 87.387136, "nll": -6.0518035888671875, "elapsed_s": 739.8}428{"event": "train", "step": 8400, "loss": -6.035555839538574, "grad_norm": 0.3559303879737854, "lr": 8.92397461760784e-05, "ex_per_s": 726.1753341258753, "data_frac": 0.02279901224124535, "mem_mb": 87.387136, "nll": -6.035555839538574, "elapsed_s": 740.3}429{"event": "train", "step": 8420, "loss": -6.0047454833984375, "grad_norm": 0.07370869070291519, "lr": 8.863562717276283e-05, "ex_per_s": 727.3641014567914, "data_frac": 0.02279636046891626, "mem_mb": 87.387136, "nll": -6.0047454833984375, "elapsed_s": 740.9}430{"event": "train", "step": 8440, "loss": -6.030483245849609, "grad_norm": 0.35605505108833313, "lr": 8.803374861031363e-05, "ex_per_s": 728.5507890709096, "data_frac": 0.022793952266636237, "mem_mb": 87.387136, "nll": -6.030483245849609, "elapsed_s": 741.4}431{"event": "train", "step": 8460, "loss": -5.909950256347656, "grad_norm": 0.40451300144195557, "lr": 8.743412814713985e-05, "ex_per_s": 729.7356575169152, "data_frac": 0.02279125561910194, "mem_mb": 87.387136, "nll": -5.909950256347656, "elapsed_s": 742.0}432{"event": "train", "step": 8480, "loss": -5.88002872467041, "grad_norm": 0.3222537338733673, "lr": 8.683678337540044e-05, "ex_per_s": 730.9179592947069, "data_frac": 0.022788795533145643, "mem_mb": 87.387136, "nll": -5.88002872467041, "elapsed_s": 742.5}433{"event": "train", "step": 8500, "loss": -5.96632719039917, "grad_norm": 0.5460098385810852, "lr": 8.62417318204883e-05, "ex_per_s": 732.0970138056751, "data_frac": 0.0227866311072631, "mem_mb": 87.387136, "nll": -5.96632719039917, "elapsed_s": 743.1}434{"event": "train", "step": 8520, "loss": -5.934908390045166, "grad_norm": 0.29210522770881653, "lr": 8.564899094051614e-05, "ex_per_s": 733.2747606530587, "data_frac": 0.022783883515947734, "mem_mb": 87.387136, "nll": -5.934908390045166, "elapsed_s": 743.6}435{"event": "train", "step": 8540, "loss": -5.905764579772949, "grad_norm": 0.29319074749946594, "lr": 8.505857812580402e-05, "ex_per_s": 734.4517460860972, "data_frac": 0.022781234666128348, "mem_mb": 87.387136, "nll": -5.905764579772949, "elapsed_s": 744.2}436{"event": "train", "step": 8560, "loss": -5.970069885253906, "grad_norm": 0.2313978672027588, "lr": 8.447051069836941e-05, "ex_per_s": 735.627931551012, "data_frac": 0.022778750165188797, "mem_mb": 87.387136, "nll": -5.970069885253906, "elapsed_s": 744.7}437{"event": "train", "step": 8580, "loss": -5.956496238708496, "grad_norm": 0.11502166092395782, "lr": 8.38848059114188e-05, "ex_per_s": 736.8013068505146, "data_frac": 0.02277575372417887, "mem_mb": 87.387136, "nll": -5.956496238708496, "elapsed_s": 745.3}438{"event": "train", "step": 8600, "loss": -6.0219244956970215, "grad_norm": 0.2989794611930847, "lr": 8.330148094884171e-05, "ex_per_s": 737.9729577931323, "data_frac": 0.022773175998459006, "mem_mb": 87.387136, "nll": -6.0219244956970215, "elapsed_s": 745.8}439{"event": "train", "step": 8620, "loss": -5.992048263549805, "grad_norm": 0.5307865738868713, "lr": 8.27205529247063e-05, "ex_per_s": 739.1432098261704, "data_frac": 0.022770494109185344, "mem_mb": 87.387136, "nll": -5.992048263549805, "elapsed_s": 746.4}440{"event": "train", "step": 8640, "loss": -5.965205669403076, "grad_norm": 0.6857157349586487, "lr": 8.214203888275743e-05, "ex_per_s": 740.3113549191095, "data_frac": 0.022767829365608516, "mem_mb": 87.387136, "nll": -5.965205669403076, "elapsed_s": 746.9}441{"event": "train", "step": 8660, "loss": -5.949089527130127, "grad_norm": 0.10125486552715302, "lr": 8.156595579591652e-05, "ex_per_s": 741.4788315135976, "data_frac": 0.02276543504981572, "mem_mb": 87.387136, "nll": -5.949089527130127, "elapsed_s": 747.5}442{"event": "train", "step": 8680, "loss": -5.899725437164307, "grad_norm": 0.3616769313812256, "lr": 8.09923205657837e-05, "ex_per_s": 742.6435789413767, "data_frac": 0.022762891493239075, "mem_mb": 87.387136, "nll": -5.899725437164307, "elapsed_s": 748.0}443{"event": "train", "step": 8700, "loss": -5.931083679199219, "grad_norm": 0.18084421753883362, "lr": 8.042115002214181e-05, "ex_per_s": 743.8070279149325, "data_frac": 0.022759939251628995, "mem_mb": 87.387136, "nll": -5.931083679199219, "elapsed_s": 748.6}444{"event": "train", "step": 8720, "loss": -6.1663079261779785, "grad_norm": 0.18849147856235504, "lr": 7.985246092246267e-05, "ex_per_s": 744.9692776168761, "data_frac": 0.02275715529021813, "mem_mb": 87.387136, "nll": -6.1663079261779785, "elapsed_s": 749.1}445{"event": "train", "step": 8740, "loss": -5.900761604309082, "grad_norm": 0.1094619482755661, "lr": 7.928626995141546e-05, "ex_per_s": 746.1290424671097, "data_frac": 0.02275440987181978, "mem_mb": 87.387136, "nll": -5.900761604309082, "elapsed_s": 749.7}446{"event": "train", "step": 8760, "loss": -6.072549819946289, "grad_norm": 0.5012131929397583, "lr": 7.872259372037726e-05, "ex_per_s": 747.2869201556146, "data_frac": 0.02275168991730239, "mem_mb": 87.387136, "nll": -6.072549819946289, "elapsed_s": 750.2}447{"event": "train", "step": 8780, "loss": -5.982771873474121, "grad_norm": 0.09783509373664856, "lr": 7.816144876694563e-05, "ex_per_s": 748.4444187094196, "data_frac": 0.022749149714096147, "mem_mb": 87.387136, "nll": -5.982771873474121, "elapsed_s": 750.8}448{"event": "train", "step": 8800, "loss": -5.953919410705566, "grad_norm": 0.10252179205417633, "lr": 7.760285155445326e-05, "ex_per_s": 749.599009170597, "data_frac": 0.02274665065720215, "mem_mb": 87.387136, "nll": -5.953919410705566, "elapsed_s": 751.3}449{"event": "train", "step": 8820, "loss": -5.986872673034668, "grad_norm": 0.18976806104183197, "lr": 7.70468184714854e-05, "ex_per_s": 750.7518312024963, "data_frac": 0.022743822241733675, "mem_mb": 87.387136, "nll": -5.986872673034668, "elapsed_s": 751.9}450{"event": "train", "step": 8840, "loss": -5.956393241882324, "grad_norm": 0.18803150951862335, "lr": 7.649336583139853e-05, "ex_per_s": 751.9028811958444, "data_frac": 0.022741239446319735, "mem_mb": 87.387136, "nll": -5.956393241882324, "elapsed_s": 752.4}451{"event": "train", "step": 8860, "loss": -5.990670680999756, "grad_norm": 0.26107850670814514, "lr": 7.594250987184208e-05, "ex_per_s": 753.0526226107928, "data_frac": 0.022738499569241368, "mem_mb": 87.387136, "nll": -5.990670680999756, "elapsed_s": 753.0}452{"event": "train", "step": 8880, "loss": -5.969250679016113, "grad_norm": 0.23125363886356354, "lr": 7.53942667542819e-05, "ex_per_s": 754.2002794915762, "data_frac": 0.022735861964481487, "mem_mb": 87.387136, "nll": -5.969250679016113, "elapsed_s": 753.6}453{"event": "train", "step": 8900, "loss": -5.9480509757995605, "grad_norm": 0.21725821495056152, "lr": 7.484865256352612e-05, "ex_per_s": 755.3432538831453, "data_frac": 0.022734157432098598, "mem_mb": 87.387136, "nll": -5.9480509757995605, "elapsed_s": 754.1}454{"event": "train", "step": 8920, "loss": -6.098246097564697, "grad_norm": 0.12495209276676178, "lr": 7.430568330725319e-05, "ex_per_s": 756.481177050616, "data_frac": 0.022732615350051662, "mem_mb": 87.387136, "nll": -6.098246097564697, "elapsed_s": 754.7}455{"event": "train", "step": 8940, "loss": -6.107264041900635, "grad_norm": 0.18885818123817444, "lr": 7.376537491554246e-05, "ex_per_s": 757.6241371690992, "data_frac": 0.022730334097374107, "mem_mb": 87.387136, "nll": -6.107264041900635, "elapsed_s": 755.2}456{"event": "train", "step": 8960, "loss": -6.056124210357666, "grad_norm": 0.2562611401081085, "lr": 7.322774324040643e-05, "ex_per_s": 758.7654344662791, "data_frac": 0.022727813688598487, "mem_mb": 87.387136, "nll": -6.056124210357666, "elapsed_s": 755.8}457{"event": "train", "step": 8980, "loss": -5.916199684143066, "grad_norm": 0.7678993344306946, "lr": 7.26928040553261e-05, "ex_per_s": 759.9058906698023, "data_frac": 0.022725386156436888, "mem_mb": 87.387136, "nll": -5.916199684143066, "elapsed_s": 756.3}458{"event": "train", "step": 9000, "loss": -6.094947338104248, "grad_norm": 0.19445332884788513, "lr": 7.216057305478784e-05, "ex_per_s": 761.0449613507149, "data_frac": 0.02272306227600385, "mem_mb": 87.387136, "nll": -6.094947338104248, "elapsed_s": 756.9}459{"event": "val", "step": 9000, "val_nll": -6.450784578466606, "best": -6.451378460830027, "bad": 2, "mem_mb": 87.387136, "elapsed_s": 818.6}460{"event": "train", "step": 9020, "loss": -6.004690170288086, "grad_norm": 0.388964980840683, "lr": 7.163106585382312e-05, "ex_per_s": 704.7304876154385, "data_frac": 0.021008559904255854, "mem_mb": 87.387136, "nll": -6.004690170288086, "elapsed_s": 819.2}461{"event": "train", "step": 9040, "loss": -6.103464603424072, "grad_norm": 0.07918325066566467, "lr": 7.11042979875503e-05, "ex_per_s": 705.8179976416443, "data_frac": 0.02100743206887752, "mem_mb": 87.387136, "nll": -6.103464603424072, "elapsed_s": 819.7}462{"event": "train", "step": 9060, "loss": -6.082679271697998, "grad_norm": 0.32133814692497253, "lr": 7.058028491071894e-05, "ex_per_s": 706.9042334340683, "data_frac": 0.021006284589278067, "mem_mb": 87.387136, "nll": -6.082679271697998, "elapsed_s": 820.3}463{"event": "train", "step": 9080, "loss": -6.094155788421631, "grad_norm": 0.3063240349292755, "lr": 7.005904199725619e-05, "ex_per_s": 707.9879581551052, "data_frac": 0.021005121372624416, "mem_mb": 87.387136, "nll": -6.094155788421631, "elapsed_s": 820.8}464{"event": "train", "step": 9100, "loss": -6.167654037475586, "grad_norm": 0.34596168994903564, "lr": 6.954058453981609e-05, "ex_per_s": 709.0702865803601, "data_frac": 0.021004124381065577, "mem_mb": 87.387136, "nll": -6.167654037475586, "elapsed_s": 821.4}465{"event": "train", "step": 9120, "loss": -5.970705032348633, "grad_norm": 0.15490353107452393, "lr": 6.902492774933048e-05, "ex_per_s": 710.1458277395308, "data_frac": 0.02101717809817138, "mem_mb": 87.387136, "nll": -5.970705032348633, "elapsed_s": 821.9}466{"event": "train", "step": 9140, "loss": -6.120774745941162, "grad_norm": 0.08929605782032013, "lr": 6.851208675456295e-05, "ex_per_s": 711.2164648774776, "data_frac": 0.021012319828086528, "mem_mb": 87.387136, "nll": -6.120774745941162, "elapsed_s": 822.5}467{"event": "train", "step": 9160, "loss": -5.980359077453613, "grad_norm": 0.23031820356845856, "lr": 6.800207660166508e-05, "ex_per_s": 712.2571462060267, "data_frac": 0.02104735730475679, "mem_mb": 87.387136, "nll": -5.980359077453613, "elapsed_s": 823.1}468{"event": "train", "step": 9180, "loss": -5.956682205200195, "grad_norm": 0.12744039297103882, "lr": 6.749491225373478e-05, "ex_per_s": 713.3310065699444, "data_frac": 0.021042157215391714, "mem_mb": 87.387136, "nll": -5.956682205200195, "elapsed_s": 823.6}469{"event": "train", "step": 9200, "loss": -6.144965171813965, "grad_norm": 0.09953243285417557, "lr": 6.699060859037737e-05, "ex_per_s": 714.4065052818397, "data_frac": 0.02103980083229723, "mem_mb": 87.387136, "nll": -6.144965171813965, "elapsed_s": 824.2}470{"event": "train", "step": 9220, "loss": -5.969365119934082, "grad_norm": 0.6003400087356567, "lr": 6.648918040726909e-05, "ex_per_s": 715.4737298957178, "data_frac": 0.02103879871835414, "mem_mb": 87.387136, "nll": -5.969365119934082, "elapsed_s": 824.8}471{"event": "train", "step": 9240, "loss": -6.115626335144043, "grad_norm": 0.36869022250175476, "lr": 6.599064241572299e-05, "ex_per_s": 716.5455865134485, "data_frac": 0.021036753875624096, "mem_mb": 87.387136, "nll": -6.115626335144043, "elapsed_s": 825.3}472{"event": "train", "step": 9260, "loss": -6.023001194000244, "grad_norm": 0.498033732175827, "lr": 6.549500924225728e-05, "ex_per_s": 717.6170935487706, "data_frac": 0.021035046703749442, "mem_mb": 87.387136, "nll": -6.023001194000244, "elapsed_s": 825.9}473{"event": "train", "step": 9280, "loss": -5.81950569152832, "grad_norm": 0.11562884598970413, "lr": 6.500229542816624e-05, "ex_per_s": 718.6867466762702, "data_frac": 0.02103308215266751, "mem_mb": 87.387136, "nll": -5.81950569152832, "elapsed_s": 826.4}474{"event": "train", "step": 9300, "loss": -6.004363059997559, "grad_norm": 0.06096707284450531, "lr": 6.45125154290935e-05, "ex_per_s": 719.7547048033463, "data_frac": 0.021031539979654325, "mem_mb": 87.387136, "nll": -6.004363059997559, "elapsed_s": 827.0}475{"event": "train", "step": 9320, "loss": -5.823749542236328, "grad_norm": 0.5874042510986328, "lr": 6.402568361460816e-05, "ex_per_s": 720.8216050189736, "data_frac": 0.021029690195051826, "mem_mb": 87.387136, "nll": -5.823749542236328, "elapsed_s": 827.5}476{"event": "train", "step": 9340, "loss": -6.052692413330078, "grad_norm": 0.35148364305496216, "lr": 6.354181426778293e-05, "ex_per_s": 721.8873799831682, "data_frac": 0.021027888658014846, "mem_mb": 87.387136, "nll": -6.052692413330078, "elapsed_s": 828.1}477{"event": "train", "step": 9360, "loss": -6.06135892868042, "grad_norm": 0.4511505663394928, "lr": 6.306092158477523e-05, "ex_per_s": 722.9518216999201, "data_frac": 0.021025901518305395, "mem_mb": 87.387136, "nll": -6.06135892868042, "elapsed_s": 828.6}478{"event": "train", "step": 9380, "loss": -5.955174446105957, "grad_norm": 0.12739114463329315, "lr": 6.258301967441065e-05, "ex_per_s": 724.0143400827327, "data_frac": 0.02102420534821116, "mem_mb": 87.387136, "nll": -5.955174446105957, "elapsed_s": 829.2}479{"event": "train", "step": 9400, "loss": -6.072584629058838, "grad_norm": 0.2889581024646759, "lr": 6.210812255776909e-05, "ex_per_s": 725.0758124410391, "data_frac": 0.02102230226577047, "mem_mb": 87.387136, "nll": -6.072584629058838, "elapsed_s": 829.7}480{"event": "train", "step": 9420, "loss": -5.9371209144592285, "grad_norm": 0.2584763467311859, "lr": 6.163624416777327e-05, "ex_per_s": 726.136061060962, "data_frac": 0.021020594912031586, "mem_mb": 87.387136, "nll": -5.9371209144592285, "elapsed_s": 830.3}481{"event": "train", "step": 9440, "loss": -5.926826000213623, "grad_norm": 0.18150700628757477, "lr": 6.116739834878009e-05, "ex_per_s": 727.1946513749307, "data_frac": 0.02101893737335877, "mem_mb": 87.387136, "nll": -5.926826000213623, "elapsed_s": 830.8}482{"event": "train", "step": 9460, "loss": -6.000977039337158, "grad_norm": 0.12678849697113037, "lr": 6.0701598856174347e-05, "ex_per_s": 728.2526158393194, "data_frac": 0.021017243712465234, "mem_mb": 87.387136, "nll": -6.000977039337158, "elapsed_s": 831.4}483{"event": "train", "step": 9480, "loss": -5.924574375152588, "grad_norm": 0.10811737179756165, "lr": 6.023885935596515e-05, "ex_per_s": 729.306022743186, "data_frac": 0.021015615796799766, "mem_mb": 87.387136, "nll": -5.924574375152588, "elapsed_s": 831.9}484{"event": "train", "step": 9500, "loss": -5.915750980377197, "grad_norm": 0.22329914569854736, "lr": 5.977919342438512e-05, "ex_per_s": 730.3601878584371, "data_frac": 0.02101348572522523, "mem_mb": 87.387136, "nll": -5.915750980377197, "elapsed_s": 832.5}485{"event": "train", "step": 9520, "loss": -5.987942218780518, "grad_norm": 0.18093310296535492, "lr": 5.9322614547491995e-05, "ex_per_s": 731.4135207262439, "data_frac": 0.021011639433657827, "mem_mb": 87.387136, "nll": -5.987942218780518, "elapsed_s": 833.0}486{"event": "train", "step": 9540, "loss": -6.003871917724609, "grad_norm": 0.17149512469768524, "lr": 5.886913612077285e-05, "ex_per_s": 732.466738577343, "data_frac": 0.02100963088064874, "mem_mb": 87.387136, "nll": -6.003871917724609, "elapsed_s": 833.6}487{"event": "train", "step": 9560, "loss": -5.96549654006958, "grad_norm": 0.12325428426265717, "lr": 5.841877144875127e-05, "ex_per_s": 733.5157750569763, "data_frac": 0.021007963439825934, "mem_mb": 87.387136, "nll": -5.96549654006958, "elapsed_s": 834.1}488{"event": "train", "step": 9580, "loss": -5.9092631340026855, "grad_norm": 0.18545927107334137, "lr": 5.7971533744597e-05, "ex_per_s": 734.5539322060309, "data_frac": 0.02100860958828001, "mem_mb": 87.387136, "nll": -5.9092631340026855, "elapsed_s": 834.7}489{"event": "train", "step": 9600, "loss": -6.155309677124023, "grad_norm": 0.5579402446746826, "lr": 5.752743612973809e-05, "ex_per_s": 735.5984743866056, "data_frac": 0.021008131352893703, "mem_mb": 87.387136, "nll": -6.155309677124023, "elapsed_s": 835.2}490{"event": "train", "step": 9620, "loss": -6.025865077972412, "grad_norm": 0.11734499782323837, "lr": 5.7086491633476214e-05, "ex_per_s": 736.6447090621042, "data_frac": 0.021006578843334314, "mem_mb": 87.387136, "nll": -6.025865077972412, "elapsed_s": 835.8}491{"event": "train", "step": 9640, "loss": -6.000345706939697, "grad_norm": 0.2466370165348053, "lr": 5.6648713192604105e-05, "ex_per_s": 737.6907297718093, "data_frac": 0.0210047622385549, "mem_mb": 87.387136, "nll": -6.000345706939697, "elapsed_s": 836.4}492{"event": "train", "step": 9660, "loss": -6.0511322021484375, "grad_norm": 0.10170773416757584, "lr": 5.621411365102632e-05, "ex_per_s": 738.7265165161223, "data_frac": 0.02100355792238687, "mem_mb": 87.387136, "nll": -6.0511322021484375, "elapsed_s": 836.9}493{"event": "train", "step": 9680, "loss": -5.931900501251221, "grad_norm": 0.14364871382713318, "lr": 5.578270575938211e-05, "ex_per_s": 739.7692568531903, "data_frac": 0.021001786277606155, "mem_mb": 87.387136, "nll": -5.931900501251221, "elapsed_s": 837.5}494{"event": "train", "step": 9700, "loss": -6.196300983428955, "grad_norm": 0.7086783647537231, "lr": 5.5354502174671515e-05, "ex_per_s": 740.809474848263, "data_frac": 0.020999810184324008, "mem_mb": 87.387136, "nll": -6.196300983428955, "elapsed_s": 838.0}495{"event": "train", "step": 9720, "loss": -6.034513473510742, "grad_norm": 0.5418042540550232, "lr": 5.492951545988398e-05, "ex_per_s": 741.849777009526, "data_frac": 0.020997866743984094, "mem_mb": 87.387136, "nll": -6.034513473510742, "elapsed_s": 838.6}496{"event": "train", "step": 9740, "loss": -5.920782566070557, "grad_norm": 0.30258262157440186, "lr": 5.4507758083629804e-05, "ex_per_s": 742.888756726648, "data_frac": 0.020995885271944386, "mem_mb": 87.387136, "nll": -5.920782566070557, "elapsed_s": 839.1}497{"event": "train", "step": 9760, "loss": -6.039750576019287, "grad_norm": 0.8714653849601746, "lr": 5.408924241977427e-05, "ex_per_s": 743.921542088196, "data_frac": 0.0209945290080635, "mem_mb": 87.387136, "nll": -6.039750576019287, "elapsed_s": 839.7}498{"event": "train", "step": 9780, "loss": -6.06117057800293, "grad_norm": 0.14143531024456024, "lr": 5.367398074707458e-05, "ex_per_s": 744.954171975676, "data_frac": 0.020993960901141407, "mem_mb": 87.387136, "nll": -6.06117057800293, "elapsed_s": 840.2}499{"event": "train", "step": 9800, "loss": -5.853957176208496, "grad_norm": 0.603736162185669, "lr": 5.326198524881979e-05, "ex_per_s": 745.9875533198127, "data_frac": 0.020992034574949497, "mem_mb": 87.387136, "nll": -5.853957176208496, "elapsed_s": 840.8}500{"event": "train", "step": 9820, "loss": -6.093336582183838, "grad_norm": 0.18343417346477509, "lr": 5.285326801247307e-05, "ex_per_s": 747.0212759508257, "data_frac": 0.020990137493443393, "mem_mb": 87.387136, "nll": -6.093336582183838, "elapsed_s": 841.3}501{"event": "train", "step": 9840, "loss": -5.9519243240356445, "grad_norm": 0.33079224824905396, "lr": 5.244784102931745e-05, "ex_per_s": 748.0526506461864, "data_frac": 0.02098855985351116, "mem_mb": 87.387136, "nll": -5.9519243240356445, "elapsed_s": 841.9}502{"event": "train", "step": 9860, "loss": -5.8804931640625, "grad_norm": 0.3901727795600891, "lr": 5.2045716194103555e-05, "ex_per_s": 749.0748090653352, "data_frac": 0.020988285361493737, "mem_mb": 87.387136, "nll": -5.8804931640625, "elapsed_s": 842.4}503{"event": "train", "step": 9880, "loss": -5.8915324211120605, "grad_norm": 0.19533923268318176, "lr": 5.164690530470111e-05, "ex_per_s": 750.0932026436237, "data_frac": 0.02098767888350054, "mem_mb": 87.387136, "nll": -5.8915324211120605, "elapsed_s": 843.0}504{"event": "train", "step": 9900, "loss": -5.938329219818115, "grad_norm": 0.40575674176216125, "lr": 5.125142006175238e-05, "ex_per_s": 751.1155748109744, "data_frac": 0.02098599823660444, "mem_mb": 87.387136, "nll": -5.938329219818115, "elapsed_s": 843.6}505{"event": "train", "step": 9920, "loss": -5.884403705596924, "grad_norm": 0.20149566233158112, "lr": 5.085927206832921e-05, "ex_per_s": 752.1365608123367, "data_frac": 0.020985227124147403, "mem_mb": 87.387136, "nll": -5.884403705596924, "elapsed_s": 844.1}506{"event": "train", "step": 9940, "loss": -5.960329532623291, "grad_norm": 0.18213415145874023, "lr": 5.047047282959241e-05, "ex_per_s": 753.1571300984514, "data_frac": 0.020984122464269783, "mem_mb": 87.387136, "nll": -5.960329532623291, "elapsed_s": 844.7}507{"event": "train", "step": 9960, "loss": -6.136186122894287, "grad_norm": 0.5019580125808716, "lr": 5.0085033752454274e-05, "ex_per_s": 754.1783586228031, "data_frac": 0.020983190025611047, "mem_mb": 87.387136, "nll": -6.136186122894287, "elapsed_s": 845.2}508{"event": "train", "step": 9980, "loss": -5.803555965423584, "grad_norm": 0.9931098222732544, "lr": 4.970296614524392e-05, "ex_per_s": 755.197400722544, "data_frac": 0.020981792335525374, "mem_mb": 87.387136, "nll": -5.803555965423584, "elapsed_s": 845.8}509{"event": "train", "step": 10000, "loss": -5.941291332244873, "grad_norm": 0.5641351342201233, "lr": 4.9324281217375474e-05, "ex_per_s": 756.2161108738793, "data_frac": 0.020980864625491903, "mem_mb": 87.387136, "nll": -5.941291332244873, "elapsed_s": 846.3}510{"event": "val", "step": 10000, "val_nll": -6.450904497069025, "best": -6.451378460830027, "bad": 3, "mem_mb": 87.387136, "elapsed_s": 908.8}511{"event": "train", "step": 10020, "loss": -6.075389385223389, "grad_norm": 0.6258154511451721, "lr": 4.8948990079019224e-05, "ex_per_s": 705.1626172108353, "data_frac": 0.01953681707811597, "mem_mb": 87.387136, "nll": -6.075389385223389, "elapsed_s": 909.4}512{"event": "train", "step": 10040, "loss": -6.080573081970215, "grad_norm": 0.3199026584625244, "lr": 4.857710374077574e-05, "ex_per_s": 706.1419390605317, "data_frac": 0.019535922534612685, "mem_mb": 87.387136, "nll": -6.080573081970215, "elapsed_s": 910.0}513{"event": "train", "step": 10060, "loss": -5.757182598114014, "grad_norm": 0.1130465641617775, "lr": 4.820863311335266e-05, "ex_per_s": 707.1205512755726, "data_frac": 0.01953519490100744, "mem_mb": 87.387136, "nll": -5.757182598114014, "elapsed_s": 910.5}514{"event": "train", "step": 10080, "loss": -6.181051254272461, "grad_norm": 0.07460504025220871, "lr": 4.78435890072448e-05, "ex_per_s": 708.0976397417427, "data_frac": 0.0195343205823378, "mem_mb": 87.387136, "nll": -6.181051254272461, "elapsed_s": 911.1}515{"event": "train", "step": 10100, "loss": -5.976009845733643, "grad_norm": 0.6353197693824768, "lr": 4.748198213241678e-05, "ex_per_s": 709.0740429258965, "data_frac": 0.019533704486429104, "mem_mb": 87.387136, "nll": -5.976009845733643, "elapsed_s": 911.6}516{"event": "train", "step": 10120, "loss": -6.063584327697754, "grad_norm": 0.4969109296798706, "lr": 4.71238230979889e-05, "ex_per_s": 710.0484000265606, "data_frac": 0.01953285277672866, "mem_mb": 87.387136, "nll": -6.063584327697754, "elapsed_s": 912.2}517{"event": "train", "step": 10140, "loss": -5.969394683837891, "grad_norm": 0.20354820787906647, "lr": 4.676912241192593e-05, "ex_per_s": 711.0215430807087, "data_frac": 0.0195321562710906, "mem_mb": 87.387136, "nll": -5.969394683837891, "elapsed_s": 912.7}518{"event": "train", "step": 10160, "loss": -5.774524688720703, "grad_norm": 0.16288608312606812, "lr": 4.641789048072876e-05, "ex_per_s": 711.9935767434569, "data_frac": 0.01953136871150558, "mem_mb": 87.387136, "nll": -5.774524688720703, "elapsed_s": 913.3}519{"event": "train", "step": 10180, "loss": -6.0105390548706055, "grad_norm": 0.17186801135540009, "lr": 4.6070137609129025e-05, "ex_per_s": 712.9649618239905, "data_frac": 0.019530746768400567, "mem_mb": 87.387136, "nll": -6.0105390548706055, "elapsed_s": 913.8}520{"event": "train", "step": 10200, "loss": -5.958827972412109, "grad_norm": 0.22463662922382355, "lr": 4.57258739997869e-05, "ex_per_s": 713.9339696443091, "data_frac": 0.019530210167560933, "mem_mb": 87.387136, "nll": -5.958827972412109, "elapsed_s": 914.4}521{"event": "train", "step": 10220, "loss": -5.782334327697754, "grad_norm": 0.12281239032745361, "lr": 4.538510975299176e-05, "ex_per_s": 714.8946760379674, "data_frac": 0.01953084358031051, "mem_mb": 87.387136, "nll": -5.782334327697754, "elapsed_s": 914.9}522{"event": "train", "step": 10240, "loss": -6.023759365081787, "grad_norm": 0.30794742703437805, "lr": 4.504785486636568e-05, "ex_per_s": 715.8511987231714, "data_frac": 0.01953187992469214, "mem_mb": 87.387136, "nll": -6.023759365081787, "elapsed_s": 915.5}523{"event": "train", "step": 10260, "loss": -6.032024383544922, "grad_norm": 0.19618935883045197, "lr": 4.471411923457034e-05, "ex_per_s": 716.8177676682618, "data_frac": 0.01953131290464718, "mem_mb": 87.387136, "nll": -6.032024383544922, "elapsed_s": 916.1}524{"event": "train", "step": 10280, "loss": -6.062406063079834, "grad_norm": 0.18295128643512726, "lr": 4.438391264901654e-05, "ex_per_s": 717.7839995387274, "data_frac": 0.019530463394185453, "mem_mb": 87.387136, "nll": -6.062406063079834, "elapsed_s": 916.6}525{"event": "train", "step": 10300, "loss": -5.992383003234863, "grad_norm": 0.12214640527963638, "lr": 4.405724479757712e-05, "ex_per_s": 718.7431233833457, "data_frac": 0.01953119807417716, "mem_mb": 87.387136, "nll": -5.992383003234863, "elapsed_s": 917.2}526{"event": "train", "step": 10320, "loss": -5.906893730163574, "grad_norm": 0.3592733144760132, "lr": 4.373412526430252e-05, "ex_per_s": 719.7057805245948, "data_frac": 0.019530469427458806, "mem_mb": 87.387136, "nll": -5.906893730163574, "elapsed_s": 917.7}527{"event": "train", "step": 10340, "loss": -5.9493231773376465, "grad_norm": 0.10114942491054535, "lr": 4.341456352913975e-05, "ex_per_s": 720.6672888731745, "data_frac": 0.019529748884732124, "mem_mb": 87.387136, "nll": -5.9493231773376465, "elapsed_s": 918.3}528{"event": "train", "step": 10360, "loss": -5.797008514404297, "grad_norm": 0.3523833155632019, "lr": 4.3098568967654146e-05, "ex_per_s": 721.6278950820939, "data_frac": 0.019529186285728948, "mem_mb": 87.387136, "nll": -5.797008514404297, "elapsed_s": 918.8}529{"event": "train", "step": 10380, "loss": -5.982280254364014, "grad_norm": 0.554698646068573, "lr": 4.27861508507545e-05, "ex_per_s": 722.5873255716822, "data_frac": 0.019528341058620294, "mem_mb": 87.387136, "nll": -5.982280254364014, "elapsed_s": 919.4}530{"event": "train", "step": 10400, "loss": -5.972064971923828, "grad_norm": 0.7184318900108337, "lr": 4.247731834442082e-05, "ex_per_s": 723.5459763527357, "data_frac": 0.019527823557169623, "mem_mb": 87.387136, "nll": -5.972064971923828, "elapsed_s": 919.9}531{"event": "train", "step": 10420, "loss": -6.0335612297058105, "grad_norm": 0.36289772391319275, "lr": 4.2172080509435485e-05, "ex_per_s": 724.5036459347192, "data_frac": 0.019527101096593953, "mem_mb": 87.387136, "nll": -6.0335612297058105, "elapsed_s": 920.5}532{"event": "train", "step": 10440, "loss": -6.123469352722168, "grad_norm": 0.5963459610939026, "lr": 4.1870446301117574e-05, "ex_per_s": 725.4592877090927, "data_frac": 0.019526503300909385, "mem_mb": 87.387136, "nll": -6.123469352722168, "elapsed_s": 921.0}533{"event": "train", "step": 10460, "loss": -5.972865581512451, "grad_norm": 0.10237566381692886, "lr": 4.157242456905989e-05, "ex_per_s": 726.4142849992342, "data_frac": 0.0195255365142016, "mem_mb": 87.387136, "nll": -5.972865581512451, "elapsed_s": 921.6}534{"event": "train", "step": 10480, "loss": -5.969245433807373, "grad_norm": 0.523353636264801, "lr": 4.1278024056869486e-05, "ex_per_s": 727.3689352539736, "data_frac": 0.019524866285716713, "mem_mb": 87.387136, "nll": -5.969245433807373, "elapsed_s": 922.1}535{"event": "train", "step": 10500, "loss": -5.994906902313232, "grad_norm": 0.4244695007801056, "lr": 4.0987253401911067e-05, "ex_per_s": 728.3211953663383, "data_frac": 0.01952431912257474, "mem_mb": 87.387136, "nll": -5.994906902313232, "elapsed_s": 922.7}536{"event": "train", "step": 10520, "loss": -6.004985809326172, "grad_norm": 0.5051804780960083, "lr": 4.070012113505359e-05, "ex_per_s": 729.2733584094173, "data_frac": 0.019523512803673752, "mem_mb": 87.387136, "nll": -6.004985809326172, "elapsed_s": 923.2}537{"event": "train", "step": 10540, "loss": -5.777639389038086, "grad_norm": 0.4886258542537689, "lr": 4.041663568041998e-05, "ex_per_s": 730.2237288604814, "data_frac": 0.019522403378133203, "mem_mb": 87.387136, "nll": -5.777639389038086, "elapsed_s": 923.8}538{"event": "train", "step": 10560, "loss": -5.870212554931641, "grad_norm": 0.13420671224594116, "lr": 4.013680535514001e-05, "ex_per_s": 731.1738872443827, "data_frac": 0.019521845305181713, "mem_mb": 87.387136, "nll": -5.870212554931641, "elapsed_s": 924.3}539{"event": "train", "step": 10580, "loss": -5.868882656097412, "grad_norm": 0.22616130113601685, "lr": 3.986063836910623e-05, "ex_per_s": 732.1221733032462, "data_frac": 0.019520799976253445, "mem_mb": 87.387136, "nll": -5.868882656097412, "elapsed_s": 924.9}540{"event": "train", "step": 10600, "loss": -5.893259525299072, "grad_norm": 0.3827493190765381, "lr": 3.958814282473316e-05, "ex_per_s": 733.0693284054435, "data_frac": 0.019519925116150742, "mem_mb": 87.387136, "nll": -5.893259525299072, "elapsed_s": 925.4}541{"event": "train", "step": 10620, "loss": -6.041603088378906, "grad_norm": 0.23788945376873016, "lr": 3.9319326716719496e-05, "ex_per_s": 734.0152237812608, "data_frac": 0.01951916956665811, "mem_mb": 87.387136, "nll": -6.041603088378906, "elapsed_s": 926.0}542{"event": "train", "step": 10640, "loss": -5.815142631530762, "grad_norm": 0.3430311679840088, "lr": 3.905419793181364e-05, "ex_per_s": 734.9601314258681, "data_frac": 0.019518381008244018, "mem_mb": 87.387136, "nll": -5.815142631530762, "elapsed_s": 926.5}543{"event": "train", "step": 10660, "loss": -5.848355770111084, "grad_norm": 0.2586807310581207, "lr": 3.87927642485822e-05, "ex_per_s": 735.9007104736237, "data_frac": 0.019517784838375084, "mem_mb": 87.387136, "nll": -5.848355770111084, "elapsed_s": 927.1}544{"event": "train", "step": 10680, "loss": -5.953320026397705, "grad_norm": 0.12650947272777557, "lr": 3.853503333718199e-05, "ex_per_s": 736.842626840963, "data_frac": 0.019517033845164622, "mem_mb": 87.387136, "nll": -5.953320026397705, "elapsed_s": 927.6}545{"event": "train", "step": 10700, "loss": -5.802043914794922, "grad_norm": 0.6718646883964539, "lr": 3.8281012759134686e-05, "ex_per_s": 737.7842155202155, "data_frac": 0.019516456008643268, "mem_mb": 87.387136, "nll": -5.802043914794922, "elapsed_s": 928.2}546{"event": "train", "step": 10720, "loss": -5.879045009613037, "grad_norm": 0.1319819986820221, "lr": 3.803070996710526e-05, "ex_per_s": 738.7254018063924, "data_frac": 0.019515768361955354, "mem_mb": 87.387136, "nll": -5.879045009613037, "elapsed_s": 928.7}547{"event": "train", "step": 10740, "loss": -6.098975658416748, "grad_norm": 0.3808552920818329, "lr": 3.778413230468319e-05, "ex_per_s": 739.6646926226439, "data_frac": 0.01951482950335317, "mem_mb": 87.387136, "nll": -6.098975658416748, "elapsed_s": 929.3}548{"event": "train", "step": 10760, "loss": -5.961232662200928, "grad_norm": 0.16332563757896423, "lr": 3.754128700616701e-05, "ex_per_s": 740.6034884774206, "data_frac": 0.019514190360759204, "mem_mb": 87.387136, "nll": -5.961232662200928, "elapsed_s": 929.8}549{"event": "train", "step": 10780, "loss": -6.0168232917785645, "grad_norm": 0.5266698598861694, "lr": 3.7302181196352115e-05, "ex_per_s": 741.5408833272627, "data_frac": 0.019513470787671225, "mem_mb": 87.387136, "nll": -6.0168232917785645, "elapsed_s": 930.4}550{"event": "train", "step": 10800, "loss": -5.930110454559326, "grad_norm": 0.22588370740413666, "lr": 3.7066821890321684e-05, "ex_per_s": 742.4714396635025, "data_frac": 0.01951311779793426, "mem_mb": 87.387136, "nll": -5.930110454559326, "elapsed_s": 931.0}551{"event": "train", "step": 10820, "loss": -6.015006065368652, "grad_norm": 0.22324518859386444, "lr": 3.683521599324089e-05, "ex_per_s": 743.403810045698, "data_frac": 0.019513455875260412, "mem_mb": 87.387136, "nll": -6.015006065368652, "elapsed_s": 931.5}552{"event": "train", "step": 10840, "loss": -5.961558818817139, "grad_norm": 0.303348571062088, "lr": 3.660737030015427e-05, "ex_per_s": 744.3385646983493, "data_frac": 0.019513075006982745, "mem_mb": 87.387136, "nll": -5.961558818817139, "elapsed_s": 932.1}553{"event": "train", "step": 10860, "loss": -5.843118190765381, "grad_norm": 0.11336714029312134, "lr": 3.638329149578648e-05, "ex_per_s": 745.2715483479587, "data_frac": 0.019512501958638975, "mem_mb": 87.387136, "nll": -5.843118190765381, "elapsed_s": 932.6}554{"event": "train", "step": 10880, "loss": -5.998651504516602, "grad_norm": 0.08524297177791595, "lr": 3.6162986154346006e-05, "ex_per_s": 746.2035187059232, "data_frac": 0.019511730296470613, "mem_mb": 87.387136, "nll": -5.998651504516602, "elapsed_s": 933.2}555{"event": "train", "step": 10900, "loss": -6.123110294342041, "grad_norm": 0.23110662400722504, "lr": 3.594646073933244e-05, "ex_per_s": 747.1350372591378, "data_frac": 0.01951084392289184, "mem_mb": 87.387136, "nll": -6.123110294342041, "elapsed_s": 933.7}556{"event": "train", "step": 10920, "loss": -5.898434162139893, "grad_norm": 0.17212854325771332, "lr": 3.5733721603346695e-05, "ex_per_s": 748.0651152836061, "data_frac": 0.019510123150043152, "mem_mb": 87.387136, "nll": -5.898434162139893, "elapsed_s": 934.3}557{"event": "train", "step": 10940, "loss": -5.89520788192749, "grad_norm": 0.34023353457450867, "lr": 3.552477498790482e-05, "ex_per_s": 748.9941945528386, "data_frac": 0.019509391445828278, "mem_mb": 87.387136, "nll": -5.89520788192749, "elapsed_s": 934.8}558{"event": "train", "step": 10960, "loss": -6.124850273132324, "grad_norm": 0.3836945593357086, "lr": 3.531962702325468e-05, "ex_per_s": 749.9167043408534, "data_frac": 0.01950985633283297, "mem_mb": 87.387136, "nll": -6.124850273132324, "elapsed_s": 935.4}559{"event": "train", "step": 10980, "loss": -5.897029876708984, "grad_norm": 0.14583748579025269, "lr": 3.511828372819624e-05, "ex_per_s": 750.8397512502445, "data_frac": 0.01950970285333484, "mem_mb": 87.387136, "nll": -5.897029876708984, "elapsed_s": 935.9}560{"event": "train", "step": 11000, "loss": -5.920523166656494, "grad_norm": 0.10986573249101639, "lr": 3.4920751009904904e-05, "ex_per_s": 751.7660560607077, "data_frac": 0.019508809852033186, "mem_mb": 87.387136, "nll": -5.920523166656494, "elapsed_s": 936.5}561{"event": "val", "step": 11000, "val_nll": -6.451677419170983, "best": -6.451677419170983, "bad": 0, "mem_mb": 87.387136, "elapsed_s": 998.3}562{"event": "train", "step": 11020, "loss": -5.804335594177246, "grad_norm": 0.3057524859905243, "lr": 3.472703466375828e-05, "ex_per_s": 706.0820195577934, "data_frac": 0.018300364862293205, "mem_mb": 87.387136, "nll": -5.804335594177246, "elapsed_s": 998.9}563{"event": "train", "step": 11040, "loss": -5.985428810119629, "grad_norm": 0.2542284429073334, "lr": 3.4537140373166085e-05, "ex_per_s": 706.9718761667016, "data_frac": 0.01830011977841537, "mem_mb": 87.387136, "nll": -5.985428810119629, "elapsed_s": 999.4}564{"event": "train", "step": 11060, "loss": -5.982811450958252, "grad_norm": 0.47602978348731995, "lr": 3.4351073709403385e-05, "ex_per_s": 707.862090312177, "data_frac": 0.018300017911420934, "mem_mb": 87.387136, "nll": -5.982811450958252, "elapsed_s": 1000.0}565{"event": "train", "step": 11080, "loss": -6.027459621429443, "grad_norm": 0.3746662139892578, "lr": 3.4168840131447255e-05, "ex_per_s": 708.7522569008842, "data_frac": 0.0182998506641848, "mem_mb": 87.387136, "nll": -6.027459621429443, "elapsed_s": 1000.5}566{"event": "train", "step": 11100, "loss": -6.139019012451172, "grad_norm": 0.5484477281570435, "lr": 3.399044498581645e-05, "ex_per_s": 709.6415650616898, "data_frac": 0.018300066324278948, "mem_mb": 87.387136, "nll": -6.139019012451172, "elapsed_s": 1001.1}567{"event": "train", "step": 11120, "loss": -5.909210205078125, "grad_norm": 0.7130416035652161, "lr": 3.3815893506414745e-05, "ex_per_s": 710.5291070796811, "data_frac": 0.018300069792495616, "mem_mb": 87.387136, "nll": -5.909210205078125, "elapsed_s": 1001.6}568{"event": "train", "step": 11140, "loss": -5.880738735198975, "grad_norm": 0.1476144790649414, "lr": 3.364519081437718e-05, "ex_per_s": 711.4164213684799, "data_frac": 0.018299938105647312, "mem_mb": 87.387136, "nll": -5.880738735198975, "elapsed_s": 1002.2}569{"event": "train", "step": 11160, "loss": -6.077532768249512, "grad_norm": 0.15938924252986908, "lr": 3.347834191791993e-05, "ex_per_s": 712.3024229264482, "data_frac": 0.01829989128780772, "mem_mb": 87.387136, "nll": -6.077532768249512, "elapsed_s": 1002.7}570{"event": "train", "step": 11180, "loss": -6.067091464996338, "grad_norm": 0.30931761860847473, "lr": 3.331535171219332e-05, "ex_per_s": 713.1815858614403, "data_frac": 0.018300863654328807, "mem_mb": 87.387136, "nll": -6.067091464996338, "elapsed_s": 1003.3}571{"event": "train", "step": 11200, "loss": -6.179079055786133, "grad_norm": 0.0774928480386734, "lr": 3.3156224979138305e-05, "ex_per_s": 714.060543208866, "data_frac": 0.018302236425698518, "mem_mb": 87.387136, "nll": -6.179079055786133, "elapsed_s": 1003.8}572{"event": "train", "step": 11220, "loss": -5.982019424438477, "grad_norm": 0.11403144896030426, "lr": 3.300096638734598e-05, "ex_per_s": 714.9377650137225, "data_frac": 0.01830305294414413, "mem_mb": 87.387136, "nll": -5.982019424438477, "elapsed_s": 1004.4}573{"event": "train", "step": 11240, "loss": -5.968616008758545, "grad_norm": 0.14043396711349487, "lr": 3.2849580491920845e-05, "ex_per_s": 715.8179189567485, "data_frac": 0.018303913904951646, "mem_mb": 87.387136, "nll": -5.968616008758545, "elapsed_s": 1005.0}574{"event": "train", "step": 11260, "loss": -6.03187894821167, "grad_norm": 0.3568243384361267, "lr": 3.2702071734346924e-05, "ex_per_s": 716.6944736409539, "data_frac": 0.018305153447512608, "mem_mb": 87.387136, "nll": -6.03187894821167, "elapsed_s": 1005.5}575{"event": "train", "step": 11280, "loss": -6.104994297027588, "grad_norm": 0.3087379038333893, "lr": 3.255844444235771e-05, "ex_per_s": 717.574436792705, "data_frac": 0.018305147547174935, "mem_mb": 87.387136, "nll": -6.104994297027588, "elapsed_s": 1006.1}576{"event": "train", "step": 11300, "loss": -5.935966491699219, "grad_norm": 0.1395108550786972, "lr": 3.2418702829808946e-05, "ex_per_s": 718.4532270672586, "data_frac": 0.018305236018521864, "mem_mb": 87.387136, "nll": -5.935966491699219, "elapsed_s": 1006.6}577{"event": "train", "step": 11320, "loss": -5.85839319229126, "grad_norm": 0.2437034696340561, "lr": 3.228285099655516e-05, "ex_per_s": 719.3309772482821, "data_frac": 0.01830524522335117, "mem_mb": 87.387136, "nll": -5.85839319229126, "elapsed_s": 1007.2}578{"event": "train", "step": 11340, "loss": -5.927273273468018, "grad_norm": 0.5453211665153503, "lr": 3.215089292832933e-05, "ex_per_s": 720.2076948894162, "data_frac": 0.018305319770334765, "mem_mb": 87.387136, "nll": -5.927273273468018, "elapsed_s": 1007.7}579{"event": "train", "step": 11360, "loss": -5.963433742523193, "grad_norm": 0.09351104497909546, "lr": 3.202283249662597e-05, "ex_per_s": 721.0839797439703, "data_frac": 0.018305194766308527, "mem_mb": 87.387136, "nll": -5.963433742523193, "elapsed_s": 1008.3}580{"event": "train", "step": 11380, "loss": -5.908868789672852, "grad_norm": 0.1917879432439804, "lr": 3.189867345858747e-05, "ex_per_s": 721.9592087533669, "data_frac": 0.01830512835134102, "mem_mb": 87.387136, "nll": -5.908868789672852, "elapsed_s": 1008.8}581{"event": "train", "step": 11400, "loss": -6.075880527496338, "grad_norm": 0.1398187130689621, "lr": 3.177841945689397e-05, "ex_per_s": 722.8331196637538, "data_frac": 0.01830511564327882, "mem_mb": 87.387136, "nll": -6.075880527496338, "elapsed_s": 1009.4}582{"event": "train", "step": 11420, "loss": -5.836301326751709, "grad_norm": 0.17609179019927979, "lr": 3.1662074019656414e-05, "ex_per_s": 723.7059593424804, "data_frac": 0.0183049255183562, "mem_mb": 87.387136, "nll": -5.836301326751709, "elapsed_s": 1009.9}583{"event": "train", "step": 11440, "loss": -5.883617877960205, "grad_norm": 0.7243898510932922, "lr": 3.154964056031306e-05, "ex_per_s": 724.5786889103151, "data_frac": 0.01830508485086036, "mem_mb": 87.387136, "nll": -5.883617877960205, "elapsed_s": 1010.5}584{"event": "train", "step": 11460, "loss": -6.137275695800781, "grad_norm": 0.6140545010566711, "lr": 3.1441122377529333e-05, "ex_per_s": 725.4507904883947, "data_frac": 0.01830497621861385, "mem_mb": 87.387136, "nll": -6.137275695800781, "elapsed_s": 1011.0}585{"event": "train", "step": 11480, "loss": -5.964150428771973, "grad_norm": 0.25568240880966187, "lr": 3.133652265510106e-05, "ex_per_s": 726.3206100064208, "data_frac": 0.018304803909857066, "mem_mb": 87.387136, "nll": -5.964150428771973, "elapsed_s": 1011.6}586{"event": "train", "step": 11500, "loss": -5.942880630493164, "grad_norm": 0.2334444373846054, "lr": 3.123584446186108e-05, "ex_per_s": 727.1882325678705, "data_frac": 0.018304849813123976, "mem_mb": 87.387136, "nll": -5.942880630493164, "elapsed_s": 1012.1}587{"event": "train", "step": 11520, "loss": -5.910140037536621, "grad_norm": 0.5161798000335693, "lr": 3.113909075158913e-05, "ex_per_s": 728.0563314723213, "data_frac": 0.01830489428038092, "mem_mb": 87.387136, "nll": -5.910140037536621, "elapsed_s": 1012.7}588{"event": "train", "step": 11540, "loss": -6.047054767608643, "grad_norm": 0.18527160584926605, "lr": 3.104626436292525e-05, "ex_per_s": 728.923492207254, "data_frac": 0.018304888271755054, "mem_mb": 87.387136, "nll": -6.047054767608643, "elapsed_s": 1013.2}589{"event": "train", "step": 11560, "loss": -6.116405487060547, "grad_norm": 0.09366846829652786, "lr": 3.095736801928647e-05, "ex_per_s": 729.7902878608222, "data_frac": 0.01830496050248617, "mem_mb": 87.387136, "nll": -6.116405487060547, "elapsed_s": 1013.8}590{"event": "train", "step": 11580, "loss": -5.957379341125488, "grad_norm": 0.09837593138217926, "lr": 3.087240432878696e-05, "ex_per_s": 730.6569240860017, "data_frac": 0.018305149778682882, "mem_mb": 87.387136, "nll": -5.957379341125488, "elapsed_s": 1014.3}591{"event": "train", "step": 11600, "loss": -6.073317050933838, "grad_norm": 0.08523277938365936, "lr": 3.0791375784161455e-05, "ex_per_s": 731.5220101909139, "data_frac": 0.018305237792894224, "mem_mb": 87.387136, "nll": -6.073317050933838, "elapsed_s": 1014.9}592{"event": "train", "step": 11620, "loss": -6.024156093597412, "grad_norm": 0.16961702704429626, "lr": 3.0714284762692116e-05, "ex_per_s": 732.3856127318041, "data_frac": 0.018305288116050047, "mem_mb": 87.387136, "nll": -6.024156093597412, "elapsed_s": 1015.4}593{"event": "train", "step": 11640, "loss": -5.9628448486328125, "grad_norm": 0.15671977400779724, "lr": 3.0641133526138836e-05, "ex_per_s": 733.242648378998, "data_frac": 0.01830665864136266, "mem_mb": 87.387136, "nll": -5.9628448486328125, "elapsed_s": 1016.0}594{"event": "train", "step": 11660, "loss": -5.921741008758545, "grad_norm": 0.3603900969028473, "lr": 3.0571924220672835e-05, "ex_per_s": 734.1033440311948, "data_frac": 0.018307190905183734, "mem_mb": 87.387136, "nll": -5.921741008758545, "elapsed_s": 1016.5}595{"event": "train", "step": 11680, "loss": -5.9648637771606445, "grad_norm": 0.31642380356788635, "lr": 3.0506658876813734e-05, "ex_per_s": 734.9635722699844, "data_frac": 0.018307054907082548, "mem_mb": 87.387136, "nll": -5.9648637771606445, "elapsed_s": 1017.1}596{"event": "train", "step": 11700, "loss": -6.047276020050049, "grad_norm": 0.13295939564704895, "lr": 3.0445339409369934e-05, "ex_per_s": 735.8227633224259, "data_frac": 0.018307030090590218, "mem_mb": 87.387136, "nll": -6.047276020050049, "elapsed_s": 1017.6}597{"event": "train", "step": 11720, "loss": -5.761789321899414, "grad_norm": 0.2536252439022064, "lr": 3.0387967617382475e-05, "ex_per_s": 736.681640254529, "data_frac": 0.018307110092381687, "mem_mb": 87.387136, "nll": -5.761789321899414, "elapsed_s": 1018.2}598{"event": "train", "step": 11740, "loss": -6.043163299560547, "grad_norm": 0.5215227603912354, "lr": 3.0334545184072252e-05, "ex_per_s": 737.5395372202314, "data_frac": 0.018307252873055582, "mem_mb": 87.387136, "nll": -6.043163299560547, "elapsed_s": 1018.8}599{"event": "train", "step": 11760, "loss": -5.901737689971924, "grad_norm": 0.20282135903835297, "lr": 3.0285073676790606e-05, "ex_per_s": 738.3966570188745, "data_frac": 0.01830734543489206, "mem_mb": 87.387136, "nll": -5.901737689971924, "elapsed_s": 1019.3}600{"event": "train", "step": 11780, "loss": -6.082774639129639, "grad_norm": 0.15525485575199127, "lr": 3.0239554546973377e-05, "ex_per_s": 739.2520505644013, "data_frac": 0.018307412002441778, "mem_mb": 87.387136, "nll": -6.082774639129639, "elapsed_s": 1019.9}601{"event": "train", "step": 11800, "loss": -5.967316627502441, "grad_norm": 0.33201226592063904, "lr": 3.0197989130098285e-05, "ex_per_s": 740.1048686334391, "data_frac": 0.018307722508421907, "mem_mb": 87.387136, "nll": -5.967316627502441, "elapsed_s": 1020.4}602{"event": "train", "step": 11820, "loss": -5.970182418823242, "grad_norm": 0.6808258891105652, "lr": 3.0160378645645744e-05, "ex_per_s": 740.9596284790159, "data_frac": 0.018307709811442786, "mem_mb": 87.387136, "nll": -5.970182418823242, "elapsed_s": 1021.0}603{"event": "train", "step": 11840, "loss": -6.031555652618408, "grad_norm": 0.11267592012882233, "lr": 3.0126724197063143e-05, "ex_per_s": 741.8130773343604, "data_frac": 0.018307682137041748, "mem_mb": 87.387136, "nll": -6.031555652618408, "elapsed_s": 1021.5}604{"event": "train", "step": 11860, "loss": -5.986674785614014, "grad_norm": 0.3639260530471802, "lr": 3.0097026771732407e-05, "ex_per_s": 742.666029881983, "data_frac": 0.018307599155000638, "mem_mb": 87.387136, "nll": -5.986674785614014, "elapsed_s": 1022.1}605{"event": "train", "step": 11880, "loss": -6.030040740966797, "grad_norm": 0.6477398872375488, "lr": 3.0071287240941065e-05, "ex_per_s": 743.5177582272897, "data_frac": 0.01830773860998495, "mem_mb": 87.387136, "nll": -6.030040740966797, "elapsed_s": 1022.6}606{"event": "train", "step": 11900, "loss": -5.886940956115723, "grad_norm": 0.15826255083084106, "lr": 3.0049506359856666e-05, "ex_per_s": 744.3679111805639, "data_frac": 0.01830781581468526, "mem_mb": 87.387136, "nll": -5.886940956115723, "elapsed_s": 1023.2}607{"event": "train", "step": 11920, "loss": -6.006839752197266, "grad_norm": 0.14892154932022095, "lr": 3.0031684767504643e-05, "ex_per_s": 745.218344564736, "data_frac": 0.018307848506737705, "mem_mb": 87.387136, "nll": -6.006839752197266, "elapsed_s": 1023.7}608{"event": "train", "step": 11940, "loss": -6.0674729347229, "grad_norm": 0.7977173328399658, "lr": 3.0017822986749544e-05, "ex_per_s": 746.0671174939163, "data_frac": 0.018307846768937243, "mem_mb": 87.387136, "nll": -6.0674729347229, "elapsed_s": 1024.3}609{"event": "train", "step": 11960, "loss": -6.163132190704346, "grad_norm": 0.14692315459251404, "lr": 3.000792142427973e-05, "ex_per_s": 746.9147340263447, "data_frac": 0.018307740651231082, "mem_mb": 87.387136, "nll": -6.163132190704346, "elapsed_s": 1024.8}610{"event": "train", "step": 11980, "loss": -6.091024875640869, "grad_norm": 0.12935681641101837, "lr": 3.000198037059536e-05, "ex_per_s": 747.7618080971732, "data_frac": 0.01830788620702322, "mem_mb": 87.387136, "nll": -6.091024875640869, "elapsed_s": 1025.4}611{"event": "train", "step": 12000, "loss": -5.93742036819458, "grad_norm": 0.11847402900457382, "lr": 2.9999999999999997e-05, "ex_per_s": 748.6084684882477, "data_frac": 0.018307979130001154, "mem_mb": 87.387136, "nll": -5.93742036819458, "elapsed_s": 1025.9}612{"event": "val", "step": 12000, "val_nll": -6.452020036474576, "best": -6.452020036474576, "bad": 0, "mem_mb": 87.387136, "elapsed_s": 1087.7}613{"event": "done", "steps": 12000, "best_val": -6.452020036474576, "wall_s": 1087.6674129962921, "ex_per_s": 706.0981976874011, "peak_mem_mb": 87.387136, "n_params": 3200259, "device": "mps", "data_time_frac": 0.01726834735532425, "elapsed_s": 1087.7}614
Files · a19bfab00d86abc6a1086d679b6b076e598d824b
Menu

Theme