SPB Git

spb/wp41_uqo

Public
154.2 KB · 614 lines
1{"event": "train", "step": 20, "loss": -6.107192516326904, "grad_norm": 2.3131539821624756, "lr": 1.5749999999999997e-05, "ex_per_s": 67.3234347265316, "data_frac": 0.6420170877865525, "mem_mb": 402.962432, "nll": -6.107192516326904, "elapsed_s": 19.2}2{"event": "train", "step": 40, "loss": -5.9028520584106445, "grad_norm": 1.0449564456939697, "lr": 3.0749999999999995e-05, "ex_per_s": 101.45678060363271, "data_frac": 0.4846856969256571, "mem_mb": 402.962432, "nll": -5.9028520584106445, "elapsed_s": 25.4}3{"event": "train", "step": 60, "loss": -5.981936454772949, "grad_norm": 0.6540200710296631, "lr": 4.5749999999999994e-05, "ex_per_s": 122.14391498668732, "data_frac": 0.3896226376675068, "mem_mb": 402.962432, "nll": -5.981936454772949, "elapsed_s": 31.6}4{"event": "train", "step": 80, "loss": -6.079314708709717, "grad_norm": 1.2294350862503052, "lr": 6.075e-05, "ex_per_s": 136.02272576695043, "data_frac": 0.3257566018491327, "mem_mb": 402.962432, "nll": -6.079314708709717, "elapsed_s": 37.8}5{"event": "train", "step": 100, "loss": -5.950352191925049, "grad_norm": 0.2080681025981903, "lr": 7.575e-05, "ex_per_s": 145.9532072860374, "data_frac": 0.2798813489734415, "mem_mb": 402.962432, "nll": -5.950352191925049, "elapsed_s": 44.0}6{"event": "train", "step": 120, "loss": -5.911990642547607, "grad_norm": 1.2063695192337036, "lr": 9.074999999999998e-05, "ex_per_s": 153.42747573626912, "data_frac": 0.24538629767444137, "mem_mb": 402.962432, "nll": -5.911990642547607, "elapsed_s": 50.2}7{"event": "train", "step": 140, "loss": -5.828505992889404, "grad_norm": 1.4288463592529297, "lr": 0.00010574999999999998, "ex_per_s": 159.2513564536868, "data_frac": 0.21851446245757, "mem_mb": 402.962432, "nll": -5.828505992889404, "elapsed_s": 56.5}8{"event": "train", "step": 160, "loss": -5.7371649742126465, "grad_norm": 0.3293968737125397, "lr": 0.00012075, "ex_per_s": 163.92916349734537, "data_frac": 0.19699243915536702, "mem_mb": 402.962432, "nll": -5.7371649742126465, "elapsed_s": 62.7}9{"event": "train", "step": 180, "loss": -5.970113277435303, "grad_norm": 2.049943208694458, "lr": 0.00013575, "ex_per_s": 167.74683488756628, "data_frac": 0.17933192238893933, "mem_mb": 402.962432, "nll": -5.970113277435303, "elapsed_s": 68.9}10{"event": "train", "step": 200, "loss": -5.967708110809326, "grad_norm": 1.3611959218978882, "lr": 0.00015074999999999998, "ex_per_s": 170.9279079542092, "data_frac": 0.1646034039910139, "mem_mb": 402.962432, "nll": -5.967708110809326, "elapsed_s": 75.1}11{"event": "train", "step": 220, "loss": -6.0331549644470215, "grad_norm": 1.2237341403961182, "lr": 0.00016575, "ex_per_s": 173.62026719601505, "data_frac": 0.1521259127740247, "mem_mb": 402.962432, "nll": -6.0331549644470215, "elapsed_s": 81.3}12{"event": "train", "step": 240, "loss": -5.9127960205078125, "grad_norm": 1.1356967687606812, "lr": 0.00018075, "ex_per_s": 175.9465184056412, "data_frac": 0.14142902679204533, "mem_mb": 402.962432, "nll": -5.9127960205078125, "elapsed_s": 87.5}13{"event": "train", "step": 260, "loss": -5.887232780456543, "grad_norm": 0.186460480093956, "lr": 0.00019574999999999996, "ex_per_s": 177.96250478209146, "data_frac": 0.13215149053808534, "mem_mb": 402.962432, "nll": -5.887232780456543, "elapsed_s": 93.7}14{"event": "train", "step": 280, "loss": -5.939558982849121, "grad_norm": 1.395185947418213, "lr": 0.00021074999999999997, "ex_per_s": 179.7303138330984, "data_frac": 0.12403157323418927, "mem_mb": 402.962432, "nll": -5.939558982849121, "elapsed_s": 99.9}15{"event": "train", "step": 300, "loss": -5.962705135345459, "grad_norm": 1.3127481937408447, "lr": 0.00022574999999999996, "ex_per_s": 181.28408383397593, "data_frac": 0.1168558814365634, "mem_mb": 402.962432, "nll": -5.962705135345459, "elapsed_s": 106.1}16{"event": "train", "step": 320, "loss": -5.966651439666748, "grad_norm": 0.15619665384292603, "lr": 0.00024074999999999997, "ex_per_s": 182.66482107871119, "data_frac": 0.11048050368326003, "mem_mb": 402.962432, "nll": -5.966651439666748, "elapsed_s": 112.3}17{"event": "train", "step": 340, "loss": -5.83209228515625, "grad_norm": 1.3084068298339844, "lr": 0.00025575, "ex_per_s": 183.90620394312813, "data_frac": 0.10477537837790023, "mem_mb": 402.962432, "nll": -5.83209228515625, "elapsed_s": 118.5}18{"event": "train", "step": 360, "loss": -5.995278835296631, "grad_norm": 0.5597890019416809, "lr": 0.00027074999999999994, "ex_per_s": 185.0077445391544, "data_frac": 0.09965265834504829, "mem_mb": 402.962432, "nll": -5.995278835296631, "elapsed_s": 124.7}19{"event": "train", "step": 380, "loss": -6.012885570526123, "grad_norm": 0.5047875046730042, "lr": 0.00028575, "ex_per_s": 186.00454483553332, "data_frac": 0.09499274313942957, "mem_mb": 402.962432, "nll": -6.012885570526123, "elapsed_s": 130.9}20{"event": "train", "step": 400, "loss": -6.003971099853516, "grad_norm": 0.6321337819099426, "lr": 0.0003, "ex_per_s": 186.91429344372466, "data_frac": 0.09076017235455153, "mem_mb": 402.962432, "nll": -6.003971099853516, "elapsed_s": 137.2}21{"event": "train", "step": 420, "loss": -5.955572605133057, "grad_norm": 2.2136952877044678, "lr": 0.0002999980196294046, "ex_per_s": 187.74260364526805, "data_frac": 0.08689522310888519, "mem_mb": 402.962432, "nll": -5.955572605133057, "elapsed_s": 143.4}22{"event": "train", "step": 440, "loss": -5.928787708282471, "grad_norm": 2.5115468502044678, "lr": 0.00029999207857572027, "ex_per_s": 188.51173057440553, "data_frac": 0.08335001761637481, "mem_mb": 402.962432, "nll": -5.928787708282471, "elapsed_s": 149.6}23{"event": "train", "step": 460, "loss": -6.096061706542969, "grad_norm": 1.9107006788253784, "lr": 0.00029998217701325043, "ex_per_s": 189.22222712592816, "data_frac": 0.08009098305328269, "mem_mb": 402.962432, "nll": -6.096061706542969, "elapsed_s": 155.8}24{"event": "train", "step": 480, "loss": -5.874442100524902, "grad_norm": 1.4291235208511353, "lr": 0.00029996831523249534, "ex_per_s": 189.87244004154402, "data_frac": 0.0770772431149375, "mem_mb": 402.962432, "nll": -5.874442100524902, "elapsed_s": 162.0}25{"event": "train", "step": 500, "loss": -5.924901008605957, "grad_norm": 0.9261434674263, "lr": 0.0002999504936401433, "ex_per_s": 190.4703412911506, "data_frac": 0.07428709737105027, "mem_mb": 402.962432, "nll": -5.924901008605957, "elapsed_s": 168.2}26{"event": "train", "step": 520, "loss": -6.03727388381958, "grad_norm": 1.3809996843338013, "lr": 0.0002999287127590589, "ex_per_s": 191.0288960305573, "data_frac": 0.07169644989552229, "mem_mb": 402.962432, "nll": -6.03727388381958, "elapsed_s": 174.4}27{"event": "train", "step": 540, "loss": -5.879265785217285, "grad_norm": 0.787273108959198, "lr": 0.00029990297322826756, "ex_per_s": 191.54505485979357, "data_frac": 0.0692822290294702, "mem_mb": 402.962432, "nll": -5.879265785217285, "elapsed_s": 180.6}28{"event": "train", "step": 560, "loss": -5.971715450286865, "grad_norm": 0.4000416696071625, "lr": 0.00029987327580293684, "ex_per_s": 192.03323637049073, "data_frac": 0.0670308098477601, "mem_mb": 402.962432, "nll": -5.971715450286865, "elapsed_s": 186.8}29{"event": "train", "step": 580, "loss": -5.922009468078613, "grad_norm": 1.1068432331085205, "lr": 0.00029983962135435424, "ex_per_s": 192.4861776022718, "data_frac": 0.06498531566932515, "mem_mb": 402.962432, "nll": -5.922009468078613, "elapsed_s": 193.0}30{"event": "train", "step": 600, "loss": -5.974067687988281, "grad_norm": 0.9065124988555908, "lr": 0.0002998020108699017, "ex_per_s": 192.90448953131568, "data_frac": 0.06300715403147421, "mem_mb": 402.962432, "nll": -5.974067687988281, "elapsed_s": 199.3}31{"event": "train", "step": 620, "loss": -5.952701091766357, "grad_norm": 0.12640467286109924, "lr": 0.0002997604454530266, "ex_per_s": 193.29913324696818, "data_frac": 0.06114937704443646, "mem_mb": 402.962432, "nll": -5.952701091766357, "elapsed_s": 205.5}32{"event": "train", "step": 640, "loss": -6.062942028045654, "grad_norm": 1.208238959312439, "lr": 0.00029971492632320935, "ex_per_s": 193.67454201450357, "data_frac": 0.05940174317163187, "mem_mb": 402.962432, "nll": -6.062942028045654, "elapsed_s": 211.7}33{"event": "train", "step": 660, "loss": -6.009639263153076, "grad_norm": 0.686811625957489, "lr": 0.0002996654548159277, "ex_per_s": 194.0336364636171, "data_frac": 0.05775532986934239, "mem_mb": 402.962432, "nll": -6.009639263153076, "elapsed_s": 217.9}34{"event": "train", "step": 680, "loss": -6.006066799163818, "grad_norm": 0.45405441522598267, "lr": 0.00029961203238261747, "ex_per_s": 194.3728782001613, "data_frac": 0.05620007541261892, "mem_mb": 402.962432, "nll": -6.006066799163818, "elapsed_s": 224.1}35{"event": "train", "step": 700, "loss": -5.983856201171875, "grad_norm": 1.774038314819336, "lr": 0.0002995546605906301, "ex_per_s": 194.69453307058018, "data_frac": 0.05472896721576836, "mem_mb": 402.962432, "nll": -5.983856201171875, "elapsed_s": 230.3}36{"event": "train", "step": 720, "loss": -6.062272548675537, "grad_norm": 0.6640989184379578, "lr": 0.00029949334112318624, "ex_per_s": 194.99336000558122, "data_frac": 0.053331184379669466, "mem_mb": 402.962432, "nll": -6.062272548675537, "elapsed_s": 236.5}37{"event": "train", "step": 740, "loss": -6.033543586730957, "grad_norm": 0.5723575353622437, "lr": 0.0002994280757793271, "ex_per_s": 195.27617511170166, "data_frac": 0.05200715060189315, "mem_mb": 402.962432, "nll": -6.033543586730957, "elapsed_s": 242.7}38{"event": "train", "step": 760, "loss": -5.955488204956055, "grad_norm": 1.7677544355392456, "lr": 0.00029935886647386114, "ex_per_s": 195.54409660474212, "data_frac": 0.05120116250018694, "mem_mb": 402.962432, "nll": -5.955488204956055, "elapsed_s": 248.9}39{"event": "train", "step": 780, "loss": -5.8937578201293945, "grad_norm": 0.46684330701828003, "lr": 0.00029928571523730785, "ex_per_s": 195.80105552503352, "data_frac": 0.049993327575705515, "mem_mb": 402.962432, "nll": -5.8937578201293945, "elapsed_s": 255.1}40{"event": "train", "step": 800, "loss": -5.8148603439331055, "grad_norm": 0.5022826194763184, "lr": 0.00029920862421583855, "ex_per_s": 196.04050539340474, "data_frac": 0.04884154465352752, "mem_mb": 402.962432, "nll": -5.8148603439331055, "elapsed_s": 261.4}41{"event": "train", "step": 820, "loss": -5.9306135177612305, "grad_norm": 1.1579512357711792, "lr": 0.00029912759567121304, "ex_per_s": 196.26773406917843, "data_frac": 0.04774256212393715, "mem_mb": 402.962432, "nll": -5.9306135177612305, "elapsed_s": 267.6}42{"event": "train", "step": 840, "loss": -5.806775093078613, "grad_norm": 0.35485219955444336, "lr": 0.0002990426319807135, "ex_per_s": 196.48564567743912, "data_frac": 0.04669617973610877, "mem_mb": 402.962432, "nll": -5.806775093078613, "elapsed_s": 273.8}43{"event": "train", "step": 860, "loss": -5.918132781982422, "grad_norm": 1.7352442741394043, "lr": 0.0002989537356370747, "ex_per_s": 196.69809931730686, "data_frac": 0.04569708935591142, "mem_mb": 402.962432, "nll": -5.918132781982422, "elapsed_s": 280.0}44{"event": "train", "step": 880, "loss": -5.869797706604004, "grad_norm": 0.293961763381958, "lr": 0.00029886090924841085, "ex_per_s": 196.90484439943916, "data_frac": 0.0447414541102662, "mem_mb": 402.962432, "nll": -5.869797706604004, "elapsed_s": 286.2}45{"event": "train", "step": 900, "loss": -5.987619876861572, "grad_norm": 0.2516532242298126, "lr": 0.0002987641555381389, "ex_per_s": 197.10143497325325, "data_frac": 0.043825706169394685, "mem_mb": 402.962432, "nll": -5.987619876861572, "elapsed_s": 292.4}46{"event": "train", "step": 920, "loss": -6.0222625732421875, "grad_norm": 0.6545571684837341, "lr": 0.0002986634773448989, "ex_per_s": 197.29207032759862, "data_frac": 0.04294759266534887, "mem_mb": 402.962432, "nll": -6.0222625732421875, "elapsed_s": 298.6}47{"event": "train", "step": 940, "loss": -6.081490516662598, "grad_norm": 1.1060117483139038, "lr": 0.00029855887762247066, "ex_per_s": 197.47045383809086, "data_frac": 0.04210602717185179, "mem_mb": 402.962432, "nll": -6.081490516662598, "elapsed_s": 304.8}48{"event": "train", "step": 960, "loss": -5.9663286209106445, "grad_norm": 0.9056603908538818, "lr": 0.00029845035943968696, "ex_per_s": 197.64379435064743, "data_frac": 0.04129761715015369, "mem_mb": 402.962432, "nll": -5.9663286209106445, "elapsed_s": 311.1}49{"event": "train", "step": 980, "loss": -5.9246506690979, "grad_norm": 0.10377504676580429, "lr": 0.0002983379259803436, "ex_per_s": 197.80603756983515, "data_frac": 0.04052074192036315, "mem_mb": 402.962432, "nll": -5.9246506690979, "elapsed_s": 317.3}50{"event": "train", "step": 1000, "loss": -6.070138931274414, "grad_norm": 0.08210939913988113, "lr": 0.00029822158054310604, "ex_per_s": 197.9605442986181, "data_frac": 0.03977228948517049, "mem_mb": 402.962432, "nll": -6.070138931274414, "elapsed_s": 323.5}51{"event": "val", "step": 1000, "val_nll": -6.4440107478472815, "best": -6.4440107478472815, "bad": 0, "mem_mb": 402.962432, "elapsed_s": 393.5}52{"event": "train", "step": 1020, "loss": -6.028340816497803, "grad_norm": 0.9137111306190491, "lr": 0.0002981013265414125, "ex_per_s": 163.37293713725006, "data_frac": 0.03220906396678216, "mem_mb": 402.962432, "nll": -6.028340816497803, "elapsed_s": 399.8}53{"event": "train", "step": 1040, "loss": -5.970975875854492, "grad_norm": 0.16366882622241974, "lr": 0.000297977167503374, "ex_per_s": 164.01327951156415, "data_frac": 0.03174015263366205, "mem_mb": 402.962432, "nll": -5.970975875854492, "elapsed_s": 406.0}54{"event": "train", "step": 1060, "loss": -6.179327011108398, "grad_norm": 0.5901069045066833, "lr": 0.00029784910707167064, "ex_per_s": 164.632987696976, "data_frac": 0.03128548895440014, "mem_mb": 402.962432, "nll": -6.179327011108398, "elapsed_s": 412.3}55{"event": "train", "step": 1080, "loss": -5.811544418334961, "grad_norm": 1.0459524393081665, "lr": 0.0002977171490034448, "ex_per_s": 165.23504351172454, "data_frac": 0.03084273891400805, "mem_mb": 402.962432, "nll": -5.811544418334961, "elapsed_s": 418.5}56{"event": "train", "step": 1100, "loss": -6.052161693572998, "grad_norm": 0.09682709723711014, "lr": 0.000297581297170191, "ex_per_s": 165.8206518905721, "data_frac": 0.030414061649607697, "mem_mb": 402.962432, "nll": -6.052161693572998, "elapsed_s": 424.7}57{"event": "train", "step": 1120, "loss": -6.053364276885986, "grad_norm": 0.4704352915287018, "lr": 0.0002974415555576423, "ex_per_s": 166.38899336790283, "data_frac": 0.02999734916766622, "mem_mb": 402.962432, "nll": -6.053364276885986, "elapsed_s": 431.0}58{"event": "train", "step": 1140, "loss": -6.019702911376953, "grad_norm": 0.6478318572044373, "lr": 0.000297297928265653, "ex_per_s": 166.9405790036385, "data_frac": 0.0295925462608847, "mem_mb": 402.962432, "nll": -6.019702911376953, "elapsed_s": 437.2}59{"event": "train", "step": 1160, "loss": -5.968024730682373, "grad_norm": 0.4760126769542694, "lr": 0.00029715041950807914, "ex_per_s": 167.47695065301667, "data_frac": 0.029198737062800184, "mem_mb": 402.962432, "nll": -5.968024730682373, "elapsed_s": 443.5}60{"event": "train", "step": 1180, "loss": -5.958501815795898, "grad_norm": 0.2333880215883255, "lr": 0.000296999033612654, "ex_per_s": 167.9998383105347, "data_frac": 0.02881650212178679, "mem_mb": 402.962432, "nll": -5.958501815795898, "elapsed_s": 449.7}61{"event": "train", "step": 1200, "loss": -5.944843292236328, "grad_norm": 0.28152742981910706, "lr": 0.00029684377502086165, "ex_per_s": 168.50982985991226, "data_frac": 0.02844446243616811, "mem_mb": 402.962432, "nll": -5.944843292236328, "elapsed_s": 456.0}62{"event": "train", "step": 1220, "loss": -6.033777713775635, "grad_norm": 0.7517232298851013, "lr": 0.00029668464828780665, "ex_per_s": 169.00850736246593, "data_frac": 0.028082924678043138, "mem_mb": 402.962432, "nll": -6.033777713775635, "elapsed_s": 462.2}63{"event": "train", "step": 1240, "loss": -6.05835485458374, "grad_norm": 0.0666666328907013, "lr": 0.00029652165808208004, "ex_per_s": 169.4949971957145, "data_frac": 0.027730568329168547, "mem_mb": 402.962432, "nll": -6.05835485458374, "elapsed_s": 468.4}64{"event": "train", "step": 1260, "loss": -6.054354667663574, "grad_norm": 0.15130338072776794, "lr": 0.0002963548091856228, "ex_per_s": 169.969422510214, "data_frac": 0.027388669411069256, "mem_mb": 402.962432, "nll": -6.054354667663574, "elapsed_s": 474.6}65{"event": "train", "step": 1280, "loss": -6.047410011291504, "grad_norm": 0.7766024470329285, "lr": 0.0002961841064935852, "ex_per_s": 170.43381219567536, "data_frac": 0.027055105464670652, "mem_mb": 402.962432, "nll": -6.047410011291504, "elapsed_s": 480.8}66{"event": "train", "step": 1300, "loss": -5.863131999969482, "grad_norm": 1.2461575269699097, "lr": 0.0002960095550141835, "ex_per_s": 170.8903762376587, "data_frac": 0.02673155449400741, "mem_mb": 402.962432, "nll": -5.863131999969482, "elapsed_s": 487.1}67{"event": "train", "step": 1320, "loss": -5.825352191925049, "grad_norm": 0.47604134678840637, "lr": 0.00029583115986855273, "ex_per_s": 171.3347576060307, "data_frac": 0.02641555023495374, "mem_mb": 402.962432, "nll": -5.825352191925049, "elapsed_s": 493.3}68{"event": "train", "step": 1340, "loss": -5.9531755447387695, "grad_norm": 0.8112426400184631, "lr": 0.0002956489262905966, "ex_per_s": 171.76253511879838, "data_frac": 0.026108280815755665, "mem_mb": 402.962432, "nll": -5.9531755447387695, "elapsed_s": 499.5}69{"event": "train", "step": 1360, "loss": -6.034916400909424, "grad_norm": 1.0386714935302734, "lr": 0.0002954628596268339, "ex_per_s": 172.1803615762114, "data_frac": 0.025806615978768706, "mem_mb": 402.962432, "nll": -6.034916400909424, "elapsed_s": 505.7}70{"event": "train", "step": 1380, "loss": -6.064829349517822, "grad_norm": 0.27478161454200745, "lr": 0.0002952729653362417, "ex_per_s": 172.58800431818352, "data_frac": 0.02551237570036973, "mem_mb": 402.962432, "nll": -6.064829349517822, "elapsed_s": 511.9}71{"event": "train", "step": 1400, "loss": -5.792366027832031, "grad_norm": 0.8609079122543335, "lr": 0.00029507924899009505, "ex_per_s": 172.98871923784495, "data_frac": 0.02522645765768454, "mem_mb": 402.962432, "nll": -5.792366027832031, "elapsed_s": 518.1}72{"event": "train", "step": 1420, "loss": -5.990638732910156, "grad_norm": 0.7710092067718506, "lr": 0.00029488171627180373, "ex_per_s": 173.37970769531915, "data_frac": 0.024946453352330377, "mem_mb": 402.962432, "nll": -5.990638732910156, "elapsed_s": 524.4}73{"event": "train", "step": 1440, "loss": -5.74508810043335, "grad_norm": 0.2322263866662979, "lr": 0.00029468037297674526, "ex_per_s": 173.76644140307647, "data_frac": 0.024673252296667034, "mem_mb": 402.962432, "nll": -5.74508810043335, "elapsed_s": 530.6}74{"event": "train", "step": 1460, "loss": -5.796363353729248, "grad_norm": 0.6004461050033569, "lr": 0.00029447522501209515, "ex_per_s": 174.1424604246741, "data_frac": 0.024406589891696768, "mem_mb": 402.962432, "nll": -5.796363353729248, "elapsed_s": 536.8}75{"event": "train", "step": 1480, "loss": -6.046204566955566, "grad_norm": 0.298090398311615, "lr": 0.00029426627839665323, "ex_per_s": 174.50932714503733, "data_frac": 0.0241454632502495, "mem_mb": 402.962432, "nll": -6.046204566955566, "elapsed_s": 543.0}76{"event": "train", "step": 1500, "loss": -5.954391002655029, "grad_norm": 0.8765631914138794, "lr": 0.00029405353926066754, "ex_per_s": 174.86694193408164, "data_frac": 0.02388984041815105, "mem_mb": 402.962432, "nll": -5.954391002655029, "elapsed_s": 549.2}77{"event": "train", "step": 1520, "loss": -5.947937488555908, "grad_norm": 0.7409071922302246, "lr": 0.000293837013845654, "ex_per_s": 175.21427283948825, "data_frac": 0.023641466859259535, "mem_mb": 402.962432, "nll": -5.947937488555908, "elapsed_s": 555.4}78{"event": "train", "step": 1540, "loss": -5.877068042755127, "grad_norm": 0.3400191068649292, "lr": 0.00029361670850421354, "ex_per_s": 175.5512144847955, "data_frac": 0.023396747565234455, "mem_mb": 402.962432, "nll": -5.877068042755127, "elapsed_s": 561.6}79{"event": "train", "step": 1560, "loss": -5.977288246154785, "grad_norm": 0.9983322620391846, "lr": 0.0002933926296998457, "ex_per_s": 175.8819621332905, "data_frac": 0.0231572689136058, "mem_mb": 402.962432, "nll": -5.977288246154785, "elapsed_s": 567.8}80{"event": "train", "step": 1580, "loss": -5.978350639343262, "grad_norm": 0.2510492503643036, "lr": 0.0002931647840067591, "ex_per_s": 176.20399850096655, "data_frac": 0.022924376403087438, "mem_mb": 402.962432, "nll": -5.978350639343262, "elapsed_s": 574.1}81{"event": "train", "step": 1600, "loss": -5.95406436920166, "grad_norm": 0.5631495714187622, "lr": 0.0002929331781096783, "ex_per_s": 176.52199043776344, "data_frac": 0.022695706641131876, "mem_mb": 402.962432, "nll": -5.95406436920166, "elapsed_s": 580.3}82{"event": "train", "step": 1620, "loss": -5.878049373626709, "grad_norm": 0.9028067588806152, "lr": 0.0002926978188036479, "ex_per_s": 176.8324882698205, "data_frac": 0.022473066922246532, "mem_mb": 402.962432, "nll": -5.878049373626709, "elapsed_s": 586.5}83{"event": "train", "step": 1640, "loss": -6.075830936431885, "grad_norm": 0.6609979867935181, "lr": 0.00029245871299383297, "ex_per_s": 177.140074018753, "data_frac": 0.022254293399887812, "mem_mb": 402.962432, "nll": -6.075830936431885, "elapsed_s": 592.7}84{"event": "train", "step": 1660, "loss": -6.03749418258667, "grad_norm": 1.123270869255066, "lr": 0.0002922158676953168, "ex_per_s": 177.4413292572932, "data_frac": 0.022040069591852007, "mem_mb": 402.962432, "nll": -6.03749418258667, "elapsed_s": 598.9}85{"event": "train", "step": 1680, "loss": -5.8611836433410645, "grad_norm": 0.5286575555801392, "lr": 0.0002919692900328947, "ex_per_s": 177.7365441672799, "data_frac": 0.021830049780023186, "mem_mb": 402.962432, "nll": -5.8611836433410645, "elapsed_s": 605.1}86{"event": "train", "step": 1700, "loss": -5.93002462387085, "grad_norm": 0.1151401698589325, "lr": 0.0002917189872408653, "ex_per_s": 178.0250687733279, "data_frac": 0.021624087853215738, "mem_mb": 402.962432, "nll": -5.93002462387085, "elapsed_s": 611.3}87{"event": "train", "step": 1720, "loss": -5.882476806640625, "grad_norm": 1.4967271089553833, "lr": 0.00029146496666281795, "ex_per_s": 178.3050102835655, "data_frac": 0.021422353924365818, "mem_mb": 402.962432, "nll": -5.882476806640625, "elapsed_s": 617.6}88{"event": "train", "step": 1740, "loss": -6.271738529205322, "grad_norm": 0.8109913468360901, "lr": 0.00029120723575141777, "ex_per_s": 178.57789679675378, "data_frac": 0.021225372855967452, "mem_mb": 402.962432, "nll": -6.271738529205322, "elapsed_s": 623.8}89{"event": "train", "step": 1760, "loss": -5.926327705383301, "grad_norm": 0.3191085755825043, "lr": 0.00029094580206818636, "ex_per_s": 178.84970054555316, "data_frac": 0.02103196579201417, "mem_mb": 402.962432, "nll": -5.926327705383301, "elapsed_s": 630.0}90{"event": "train", "step": 1780, "loss": -5.869332790374756, "grad_norm": 0.5815069079399109, "lr": 0.00029068067328328045, "ex_per_s": 179.1146046672617, "data_frac": 0.02084293020000459, "mem_mb": 402.962432, "nll": -5.869332790374756, "elapsed_s": 636.2}91{"event": "train", "step": 1800, "loss": -6.044037342071533, "grad_norm": 1.790061354637146, "lr": 0.0002904118571752668, "ex_per_s": 179.37552005746778, "data_frac": 0.020658305685307923, "mem_mb": 402.962432, "nll": -6.044037342071533, "elapsed_s": 642.4}92{"event": "train", "step": 1820, "loss": -5.931137561798096, "grad_norm": 1.6130210161209106, "lr": 0.00029013936163089376, "ex_per_s": 179.63180498330675, "data_frac": 0.02047614455765362, "mem_mb": 402.962432, "nll": -5.931137561798096, "elapsed_s": 648.6}93{"event": "train", "step": 1840, "loss": -6.071759223937988, "grad_norm": 0.40874549746513367, "lr": 0.00028986319464486, "ex_per_s": 179.88391502261615, "data_frac": 0.020296958266374662, "mem_mb": 402.962432, "nll": -6.071759223937988, "elapsed_s": 654.8}94{"event": "train", "step": 1860, "loss": -6.048898220062256, "grad_norm": 0.5703567266464233, "lr": 0.00028958336431958003, "ex_per_s": 180.1313773395533, "data_frac": 0.020121459098307384, "mem_mb": 402.962432, "nll": -6.048898220062256, "elapsed_s": 661.0}95{"event": "train", "step": 1880, "loss": -5.944430351257324, "grad_norm": 1.0417710542678833, "lr": 0.0002892998788649464, "ex_per_s": 180.3741227529576, "data_frac": 0.019949369936297603, "mem_mb": 402.962432, "nll": -5.944430351257324, "elapsed_s": 667.2}96{"event": "train", "step": 1900, "loss": -5.980104446411133, "grad_norm": 0.14741484820842743, "lr": 0.0002890127465980889, "ex_per_s": 180.61172751937727, "data_frac": 0.019780494351352068, "mem_mb": 402.962432, "nll": -5.980104446411133, "elapsed_s": 673.5}97{"event": "train", "step": 1920, "loss": -5.961438179016113, "grad_norm": 0.3133113980293274, "lr": 0.0002887219759431305, "ex_per_s": 180.84314863195138, "data_frac": 0.019614490587764554, "mem_mb": 402.962432, "nll": -5.961438179016113, "elapsed_s": 679.7}98{"event": "train", "step": 1940, "loss": -6.047362804412842, "grad_norm": 0.4715988039970398, "lr": 0.0002884275754309401, "ex_per_s": 181.06952102645332, "data_frac": 0.019451283079100625, "mem_mb": 402.962432, "nll": -6.047362804412842, "elapsed_s": 685.9}99{"event": "train", "step": 1960, "loss": -5.935074806213379, "grad_norm": 0.12194885313510895, "lr": 0.0002881295536988824, "ex_per_s": 181.29341733151907, "data_frac": 0.019291054512986634, "mem_mb": 402.962432, "nll": -5.935074806213379, "elapsed_s": 692.1}100{"event": "train", "step": 1980, "loss": -5.960320472717285, "grad_norm": 1.2137799263000488, "lr": 0.00028782791949056447, "ex_per_s": 181.51456522279335, "data_frac": 0.019133714921995892, "mem_mb": 402.962432, "nll": -5.960320472717285, "elapsed_s": 698.3}101{"event": "train", "step": 2000, "loss": -5.915528774261475, "grad_norm": 0.10422791540622711, "lr": 0.00028752268165557917, "ex_per_s": 181.73071481496797, "data_frac": 0.018979396975367362, "mem_mb": 402.962432, "nll": -5.915528774261475, "elapsed_s": 704.5}102{"event": "val", "step": 2000, "val_nll": -6.443407567357992, "best": -6.4440107478472815, "bad": 1, "mem_mb": 402.962432, "elapsed_s": 773.3}103{"event": "train", "step": 2020, "loss": -5.923832416534424, "grad_norm": 0.7397171258926392, "lr": 0.00028721384914924547, "ex_per_s": 165.86447619331383, "data_frac": 0.017164679193111168, "mem_mb": 402.962432, "nll": -5.923832416534424, "elapsed_s": 779.6}104{"event": "train", "step": 2040, "loss": -5.874258995056152, "grad_norm": 0.2966825067996979, "lr": 0.0002869014310323458, "ex_per_s": 166.1745076773503, "data_frac": 0.01704083457006182, "mem_mb": 402.962432, "nll": -5.874258995056152, "elapsed_s": 785.9}105{"event": "train", "step": 2060, "loss": -6.142366409301758, "grad_norm": 0.47748425602912903, "lr": 0.0002865854364708602, "ex_per_s": 166.480426551088, "data_frac": 0.01691895340152989, "mem_mb": 402.962432, "nll": -6.142366409301758, "elapsed_s": 792.1}106{"event": "train", "step": 2080, "loss": -5.956662178039551, "grad_norm": 0.23323269188404083, "lr": 0.00028626587473569746, "ex_per_s": 166.78110622440093, "data_frac": 0.016799132213568557, "mem_mb": 402.962432, "nll": -5.956662178039551, "elapsed_s": 798.4}107{"event": "train", "step": 2100, "loss": -5.902310848236084, "grad_norm": 0.6785365343093872, "lr": 0.0002859427552024228, "ex_per_s": 167.07801147781444, "data_frac": 0.01668118916330314, "mem_mb": 402.962432, "nll": -5.902310848236084, "elapsed_s": 804.6}108{"event": "train", "step": 2120, "loss": -6.147507667541504, "grad_norm": 0.7863346934318542, "lr": 0.0002856160873509834, "ex_per_s": 167.37015292901467, "data_frac": 0.01656500078075001, "mem_mb": 402.962432, "nll": -6.147507667541504, "elapsed_s": 810.9}109{"event": "train", "step": 2140, "loss": -5.9748077392578125, "grad_norm": 0.7714583873748779, "lr": 0.00028528588076542964, "ex_per_s": 167.65768583913984, "data_frac": 0.016451999297526367, "mem_mb": 402.962432, "nll": -5.9748077392578125, "elapsed_s": 817.1}110{"event": "train", "step": 2160, "loss": -5.9232587814331055, "grad_norm": 1.5048294067382812, "lr": 0.00028495214513363427, "ex_per_s": 167.94166956249316, "data_frac": 0.01634006755953865, "mem_mb": 402.962432, "nll": -5.9232587814331055, "elapsed_s": 823.3}111{"event": "train", "step": 2180, "loss": -6.079946041107178, "grad_norm": 1.0343011617660522, "lr": 0.0002846148902470082, "ex_per_s": 168.2235517775022, "data_frac": 0.0162290928706063, "mem_mb": 402.962432, "nll": -6.079946041107178, "elapsed_s": 829.6}112{"event": "train", "step": 2200, "loss": -5.961045742034912, "grad_norm": 1.0314335823059082, "lr": 0.0002842741260002131, "ex_per_s": 168.50055299374446, "data_frac": 0.01611970601430847, "mem_mb": 402.962432, "nll": -5.961045742034912, "elapsed_s": 835.8}113{"event": "train", "step": 2220, "loss": -5.825003623962402, "grad_norm": 0.7837941646575928, "lr": 0.00028392986239087094, "ex_per_s": 168.7756521317824, "data_frac": 0.016012521788840803, "mem_mb": 402.962432, "nll": -5.825003623962402, "elapsed_s": 842.0}114{"event": "train", "step": 2240, "loss": -5.959245204925537, "grad_norm": 0.14749103784561157, "lr": 0.0002835821095192712, "ex_per_s": 169.0473131265897, "data_frac": 0.015906663893801055, "mem_mb": 402.962432, "nll": -5.959245204925537, "elapsed_s": 848.2}115{"event": "train", "step": 2260, "loss": -5.737170696258545, "grad_norm": 1.25873863697052, "lr": 0.0002832308775880741, "ex_per_s": 169.31441480317065, "data_frac": 0.015802710665582108, "mem_mb": 402.962432, "nll": -5.737170696258545, "elapsed_s": 854.5}116{"event": "train", "step": 2280, "loss": -6.008480548858643, "grad_norm": 0.68392413854599, "lr": 0.0002828761769020111, "ex_per_s": 169.57615882438378, "data_frac": 0.015700225161033577, "mem_mb": 402.962432, "nll": -6.008480548858643, "elapsed_s": 860.7}117{"event": "train", "step": 2300, "loss": -5.998634338378906, "grad_norm": 0.11143399029970169, "lr": 0.0002825180178675832, "ex_per_s": 169.83426507893748, "data_frac": 0.015598550417734317, "mem_mb": 402.962432, "nll": -5.998634338378906, "elapsed_s": 866.9}118{"event": "train", "step": 2320, "loss": -6.068146705627441, "grad_norm": 0.3305736780166626, "lr": 0.00028215641099275517, "ex_per_s": 170.08907618885922, "data_frac": 0.015498480597754813, "mem_mb": 402.962432, "nll": -6.068146705627441, "elapsed_s": 873.1}119{"event": "train", "step": 2340, "loss": -5.852524280548096, "grad_norm": 0.3394244313240051, "lr": 0.0002817913668866473, "ex_per_s": 170.33958764746683, "data_frac": 0.015399983141250627, "mem_mb": 402.962432, "nll": -5.852524280548096, "elapsed_s": 879.4}120{"event": "train", "step": 2360, "loss": -5.926583290100098, "grad_norm": 0.7364758253097534, "lr": 0.00028142289625922426, "ex_per_s": 170.58760749149275, "data_frac": 0.015302963871526814, "mem_mb": 402.962432, "nll": -5.926583290100098, "elapsed_s": 885.6}121{"event": "train", "step": 2380, "loss": -5.932802677154541, "grad_norm": 0.14720652997493744, "lr": 0.00028105100992098075, "ex_per_s": 170.83020902259352, "data_frac": 0.015207282527666654, "mem_mb": 402.962432, "nll": -5.932802677154541, "elapsed_s": 891.8}122{"event": "train", "step": 2400, "loss": -5.9576897621154785, "grad_norm": 0.7608413100242615, "lr": 0.00028067571878262454, "ex_per_s": 171.07086945732684, "data_frac": 0.015112962579800538, "mem_mb": 402.962432, "nll": -5.9576897621154785, "elapsed_s": 898.1}123{"event": "train", "step": 2420, "loss": -5.996270179748535, "grad_norm": 0.30033889412879944, "lr": 0.00028029703385475607, "ex_per_s": 171.30984347511696, "data_frac": 0.015020287717713571, "mem_mb": 402.962432, "nll": -5.996270179748535, "elapsed_s": 904.3}124{"event": "train", "step": 2440, "loss": -6.09568977355957, "grad_norm": 1.213372826576233, "lr": 0.0002799149662475457, "ex_per_s": 171.5456195079122, "data_frac": 0.014928734185242492, "mem_mb": 402.962432, "nll": -6.09568977355957, "elapsed_s": 910.5}125{"event": "train", "step": 2460, "loss": -6.129843235015869, "grad_norm": 0.46670204401016235, "lr": 0.0002795295271704076, "ex_per_s": 171.77738326619567, "data_frac": 0.014838940273057894, "mem_mb": 402.962432, "nll": -6.129843235015869, "elapsed_s": 916.7}126{"event": "train", "step": 2480, "loss": -5.940186500549316, "grad_norm": 0.12344925105571747, "lr": 0.00027914072793167077, "ex_per_s": 172.0040298069021, "data_frac": 0.014749439068102884, "mem_mb": 402.962432, "nll": -5.940186500549316, "elapsed_s": 923.0}127{"event": "train", "step": 2500, "loss": -5.754850387573242, "grad_norm": 0.653318464756012, "lr": 0.0002787485799382476, "ex_per_s": 172.22838438660438, "data_frac": 0.014661555751221194, "mem_mb": 402.962432, "nll": -5.754850387573242, "elapsed_s": 929.2}128{"event": "train", "step": 2520, "loss": -5.961744785308838, "grad_norm": 0.6117812395095825, "lr": 0.0002783530946952989, "ex_per_s": 172.45065205231086, "data_frac": 0.01457463196797852, "mem_mb": 402.962432, "nll": -5.961744785308838, "elapsed_s": 935.4}129{"event": "train", "step": 2540, "loss": -6.030368804931641, "grad_norm": 0.1386711299419403, "lr": 0.00027795428380589644, "ex_per_s": 172.66977855774275, "data_frac": 0.014489082109189652, "mem_mb": 402.962432, "nll": -6.030368804931641, "elapsed_s": 941.6}130{"event": "train", "step": 2560, "loss": -5.945569038391113, "grad_norm": 1.0471605062484741, "lr": 0.00027755215897068257, "ex_per_s": 172.8862563915837, "data_frac": 0.014405314594178795, "mem_mb": 402.962432, "nll": -5.945569038391113, "elapsed_s": 947.9}131{"event": "train", "step": 2580, "loss": -5.95817232131958, "grad_norm": 0.9344031810760498, "lr": 0.0002771467319875269, "ex_per_s": 173.10051043306663, "data_frac": 0.01432238646053255, "mem_mb": 402.962432, "nll": -5.95817232131958, "elapsed_s": 954.1}132{"event": "train", "step": 2600, "loss": -5.84984016418457, "grad_norm": 0.2737126648426056, "lr": 0.0002767380147511802, "ex_per_s": 173.313522460924, "data_frac": 0.01424009126172363, "mem_mb": 402.962432, "nll": -5.84984016418457, "elapsed_s": 960.3}133{"event": "train", "step": 2620, "loss": -5.8896918296813965, "grad_norm": 1.4560209512710571, "lr": 0.0002763260192529254, "ex_per_s": 173.52417948840824, "data_frac": 0.014158882204973647, "mem_mb": 402.962432, "nll": -5.8896918296813965, "elapsed_s": 966.5}134{"event": "train", "step": 2640, "loss": -5.985092639923096, "grad_norm": 0.4049466848373413, "lr": 0.0002759107575802257, "ex_per_s": 173.7323381546281, "data_frac": 0.014078630871187775, "mem_mb": 402.962432, "nll": -5.985092639923096, "elapsed_s": 972.7}135{"event": "train", "step": 2660, "loss": -5.819899559020996, "grad_norm": 0.29040083289146423, "lr": 0.00027549224191637014, "ex_per_s": 173.93556791152105, "data_frac": 0.01399923448194578, "mem_mb": 402.962432, "nll": -5.819899559020996, "elapsed_s": 978.9}136{"event": "train", "step": 2680, "loss": -5.957686901092529, "grad_norm": 0.6378191113471985, "lr": 0.000275070484540116, "ex_per_s": 174.13610782003641, "data_frac": 0.013920779494741535, "mem_mb": 402.962432, "nll": -5.957686901092529, "elapsed_s": 985.2}137{"event": "train", "step": 2700, "loss": -6.1328277587890625, "grad_norm": 0.12337521463632584, "lr": 0.00027464549782532844, "ex_per_s": 174.33322598812805, "data_frac": 0.013843337049106204, "mem_mb": 402.962432, "nll": -6.1328277587890625, "elapsed_s": 991.4}138{"event": "train", "step": 2720, "loss": -5.881932258605957, "grad_norm": 0.20433562994003296, "lr": 0.00027421729424061787, "ex_per_s": 174.5291140012918, "data_frac": 0.01376738036610418, "mem_mb": 402.962432, "nll": -5.881932258605957, "elapsed_s": 997.6}139{"event": "train", "step": 2740, "loss": -6.122591972351074, "grad_norm": 0.329846054315567, "lr": 0.0002737858863489736, "ex_per_s": 174.72302792407623, "data_frac": 0.013691990741575408, "mem_mb": 402.962432, "nll": -6.122591972351074, "elapsed_s": 1003.8}140{"event": "train", "step": 2760, "loss": -5.961249828338623, "grad_norm": 0.5588884353637695, "lr": 0.00027335128680739583, "ex_per_s": 174.9146663630665, "data_frac": 0.013617259836018091, "mem_mb": 402.962432, "nll": -5.961249828338623, "elapsed_s": 1010.1}141{"event": "train", "step": 2780, "loss": -6.010849952697754, "grad_norm": 0.6349753141403198, "lr": 0.00027291350836652374, "ex_per_s": 175.10374223147912, "data_frac": 0.013543750873058676, "mem_mb": 402.962432, "nll": -6.010849952697754, "elapsed_s": 1016.3}142{"event": "train", "step": 2800, "loss": -5.979385852813721, "grad_norm": 0.7746085524559021, "lr": 0.00027247256387026185, "ex_per_s": 175.2906016658483, "data_frac": 0.013471923854305063, "mem_mb": 402.962432, "nll": -5.979385852813721, "elapsed_s": 1022.5}143{"event": "train", "step": 2820, "loss": -6.008054256439209, "grad_norm": 0.25368770956993103, "lr": 0.000272028466255403, "ex_per_s": 175.47674100167075, "data_frac": 0.01340032199871768, "mem_mb": 402.962432, "nll": -6.008054256439209, "elapsed_s": 1028.7}144{"event": "train", "step": 2840, "loss": -5.874866962432861, "grad_norm": 0.14622652530670166, "lr": 0.0002715812285512487, "ex_per_s": 175.6599355670821, "data_frac": 0.013329064155039377, "mem_mb": 402.962432, "nll": -5.874866962432861, "elapsed_s": 1034.9}145{"event": "train", "step": 2860, "loss": -6.050624847412109, "grad_norm": 0.3414463698863983, "lr": 0.0002711308638792272, "ex_per_s": 175.83935770905495, "data_frac": 0.013259520152768173, "mem_mb": 402.962432, "nll": -6.050624847412109, "elapsed_s": 1041.1}146{"event": "train", "step": 2880, "loss": -5.7729902267456055, "grad_norm": 0.11631906777620316, "lr": 0.000270677385452508, "ex_per_s": 176.01631570697262, "data_frac": 0.013190077373960064, "mem_mb": 402.962432, "nll": -5.7729902267456055, "elapsed_s": 1047.4}147{"event": "train", "step": 2900, "loss": -5.7899298667907715, "grad_norm": 0.713985800743103, "lr": 0.00027022080657561484, "ex_per_s": 176.1916094147032, "data_frac": 0.01312158567922688, "mem_mb": 402.962432, "nll": -5.7899298667907715, "elapsed_s": 1053.6}148{"event": "train", "step": 2920, "loss": -5.971068382263184, "grad_norm": 0.5636841058731079, "lr": 0.0002697611406440348, "ex_per_s": 176.3646810461843, "data_frac": 0.013053845439538563, "mem_mb": 402.962432, "nll": -5.971068382263184, "elapsed_s": 1059.8}149{"event": "train", "step": 2940, "loss": -5.896920680999756, "grad_norm": 1.1226177215576172, "lr": 0.00026929840114382563, "ex_per_s": 176.53580598595394, "data_frac": 0.012987016124476081, "mem_mb": 402.962432, "nll": -5.896920680999756, "elapsed_s": 1066.0}150{"event": "train", "step": 2960, "loss": -5.933407306671143, "grad_norm": 0.6426848769187927, "lr": 0.0002688326016512199, "ex_per_s": 176.70598515729588, "data_frac": 0.012920720817351159, "mem_mb": 402.962432, "nll": -5.933407306671143, "elapsed_s": 1072.3}151{"event": "train", "step": 2980, "loss": -5.890815258026123, "grad_norm": 0.09740150719881058, "lr": 0.0002683637558322267, "ex_per_s": 176.8736820070671, "data_frac": 0.012855469072128659, "mem_mb": 402.962432, "nll": -5.890815258026123, "elapsed_s": 1078.5}152{"event": "train", "step": 3000, "loss": -5.993322849273682, "grad_norm": 0.12731651961803436, "lr": 0.0002678918774422309, "ex_per_s": 177.04035157040028, "data_frac": 0.01279125375819366, "mem_mb": 402.962432, "nll": -5.993322849273682, "elapsed_s": 1084.7}153{"event": "val", "step": 3000, "val_nll": -6.445353803539011, "best": -6.445353803539011, "bad": 0, "mem_mb": 402.962432, "elapsed_s": 1153.1}154{"event": "train", "step": 3020, "loss": -5.961369514465332, "grad_norm": 1.3589694499969482, "lr": 0.0002674169803255893, "ex_per_s": 166.74211397270662, "data_frac": 0.01197656303798006, "mem_mb": 402.962432, "nll": -5.961369514465332, "elapsed_s": 1159.3}155{"event": "train", "step": 3040, "loss": -5.852590084075928, "grad_norm": 2.000166416168213, "lr": 0.0002669390784152248, "ex_per_s": 166.9477087507199, "data_frac": 0.011921107657122669, "mem_mb": 402.962432, "nll": -5.852590084075928, "elapsed_s": 1165.6}156{"event": "train", "step": 3060, "loss": -6.072350025177002, "grad_norm": 0.800068736076355, "lr": 0.00026645818573221706, "ex_per_s": 167.15122539203247, "data_frac": 0.011866132837294431, "mem_mb": 402.962432, "nll": -6.072350025177002, "elapsed_s": 1171.8}157{"event": "train", "step": 3080, "loss": -5.9670491218566895, "grad_norm": 0.2104136049747467, "lr": 0.00026597431638539183, "ex_per_s": 167.35246179386644, "data_frac": 0.011812586608246748, "mem_mb": 402.962432, "nll": -5.9670491218566895, "elapsed_s": 1178.1}158{"event": "train", "step": 3100, "loss": -6.012156009674072, "grad_norm": 0.1734142303466797, "lr": 0.00026548748457090647, "ex_per_s": 167.5523748950482, "data_frac": 0.011758842596228626, "mem_mb": 402.962432, "nll": -6.012156009674072, "elapsed_s": 1184.3}159{"event": "train", "step": 3120, "loss": -6.087346076965332, "grad_norm": 1.214215636253357, "lr": 0.00026499770457183376, "ex_per_s": 167.7507436537073, "data_frac": 0.011705749144045026, "mem_mb": 402.962432, "nll": -6.087346076965332, "elapsed_s": 1190.5}160{"event": "train", "step": 3140, "loss": -6.0089898109436035, "grad_norm": 0.6937364935874939, "lr": 0.0002645049907577427, "ex_per_s": 167.9471588334792, "data_frac": 0.011653245441671706, "mem_mb": 402.962432, "nll": -6.0089898109436035, "elapsed_s": 1196.8}161{"event": "train", "step": 3160, "loss": -6.020596027374268, "grad_norm": 0.6121866703033447, "lr": 0.00026400935758427697, "ex_per_s": 168.14093663904686, "data_frac": 0.011601487988385837, "mem_mb": 402.962432, "nll": -6.020596027374268, "elapsed_s": 1203.0}162{"event": "train", "step": 3180, "loss": -6.090184211730957, "grad_norm": 0.353160560131073, "lr": 0.00026351081959273087, "ex_per_s": 168.3347623684438, "data_frac": 0.011550460253158952, "mem_mb": 402.962432, "nll": -6.090184211730957, "elapsed_s": 1209.2}163{"event": "train", "step": 3200, "loss": -5.996584415435791, "grad_norm": 1.1635832786560059, "lr": 0.0002630093914096226, "ex_per_s": 168.52644245522623, "data_frac": 0.011499596875500125, "mem_mb": 402.962432, "nll": -5.996584415435791, "elapsed_s": 1215.4}164{"event": "train", "step": 3220, "loss": -5.913236141204834, "grad_norm": 0.769187867641449, "lr": 0.00026250508774626523, "ex_per_s": 168.71551867572188, "data_frac": 0.011449418502331252, "mem_mb": 402.962432, "nll": -5.913236141204834, "elapsed_s": 1221.7}165{"event": "train", "step": 3240, "loss": -5.799216270446777, "grad_norm": 0.7228909134864807, "lr": 0.00026199792339833493, "ex_per_s": 168.90339239286996, "data_frac": 0.011399467274775076, "mem_mb": 402.962432, "nll": -5.799216270446777, "elapsed_s": 1227.9}166{"event": "train", "step": 3260, "loss": -5.958597183227539, "grad_norm": 0.28226834535598755, "lr": 0.00026148791324543704, "ex_per_s": 169.08919571299108, "data_frac": 0.011350018597897749, "mem_mb": 402.962432, "nll": -5.958597183227539, "elapsed_s": 1234.1}167{"event": "train", "step": 3280, "loss": -6.016233921051025, "grad_norm": 0.17407867312431335, "lr": 0.0002609750722506695, "ex_per_s": 169.27312974212006, "data_frac": 0.011301105670069029, "mem_mb": 402.962432, "nll": -6.016233921051025, "elapsed_s": 1240.3}168{"event": "train", "step": 3300, "loss": -5.800965785980225, "grad_norm": 0.9536173939704895, "lr": 0.0002604594154601839, "ex_per_s": 169.45518909859618, "data_frac": 0.011252705441709086, "mem_mb": 402.962432, "nll": -5.800965785980225, "elapsed_s": 1246.5}169{"event": "train", "step": 3320, "loss": -5.9851555824279785, "grad_norm": 0.9405432343482971, "lr": 0.0002599409580027438, "ex_per_s": 169.63542035218723, "data_frac": 0.01120535040847046, "mem_mb": 402.962432, "nll": -5.9851555824279785, "elapsed_s": 1252.8}170{"event": "train", "step": 3340, "loss": -5.928009510040283, "grad_norm": 0.6089335680007935, "lr": 0.0002594197150892811, "ex_per_s": 169.81374443968866, "data_frac": 0.011157900640110033, "mem_mb": 402.962432, "nll": -5.928009510040283, "elapsed_s": 1259.0}171{"event": "train", "step": 3360, "loss": -5.9058732986450195, "grad_norm": 0.3709668219089508, "lr": 0.00025889570201244967, "ex_per_s": 169.9909465613543, "data_frac": 0.011111047949985162, "mem_mb": 402.962432, "nll": -5.9058732986450195, "elapsed_s": 1265.2}172{"event": "train", "step": 3380, "loss": -6.122546672821045, "grad_norm": 0.08987385034561157, "lr": 0.00025836893414617684, "ex_per_s": 170.16687210498912, "data_frac": 0.011064678404815613, "mem_mb": 402.962432, "nll": -6.122546672821045, "elapsed_s": 1271.4}173{"event": "train", "step": 3400, "loss": -5.878023147583008, "grad_norm": 0.48551082611083984, "lr": 0.00025783942694521214, "ex_per_s": 170.33985820039172, "data_frac": 0.011018612637324003, "mem_mb": 402.962432, "nll": -5.878023147583008, "elapsed_s": 1277.6}174{"event": "train", "step": 3420, "loss": -5.835014820098877, "grad_norm": 0.29327115416526794, "lr": 0.00025730719594467387, "ex_per_s": 170.51078003628996, "data_frac": 0.010973026584587, "mem_mb": 402.962432, "nll": -5.835014820098877, "elapsed_s": 1283.9}175{"event": "train", "step": 3440, "loss": -5.9144978523254395, "grad_norm": 0.8563370704650879, "lr": 0.00025677225675959357, "ex_per_s": 170.68034607273523, "data_frac": 0.010928129395356201, "mem_mb": 402.962432, "nll": -5.9144978523254395, "elapsed_s": 1290.1}176{"event": "train", "step": 3460, "loss": -6.0572590827941895, "grad_norm": 0.48602941632270813, "lr": 0.00025623462508445754, "ex_per_s": 170.8493486349561, "data_frac": 0.010883402356445445, "mem_mb": 402.962432, "nll": -6.0572590827941895, "elapsed_s": 1296.3}177{"event": "train", "step": 3480, "loss": -5.916203498840332, "grad_norm": 0.15965838730335236, "lr": 0.00025569431669274684, "ex_per_s": 171.0159746989147, "data_frac": 0.010839147459868772, "mem_mb": 402.962432, "nll": -5.916203498840332, "elapsed_s": 1302.5}178{"event": "train", "step": 3500, "loss": -6.029778003692627, "grad_norm": 1.1786530017852783, "lr": 0.00025515134743647386, "ex_per_s": 171.1809130012439, "data_frac": 0.010795497474267998, "mem_mb": 402.962432, "nll": -6.029778003692627, "elapsed_s": 1308.7}179{"event": "train", "step": 3520, "loss": -5.9321370124816895, "grad_norm": 0.9172441959381104, "lr": 0.0002546057332457181, "ex_per_s": 171.34481136651968, "data_frac": 0.010752013573864868, "mem_mb": 402.962432, "nll": -5.9321370124816895, "elapsed_s": 1315.0}180{"event": "train", "step": 3540, "loss": -6.1367034912109375, "grad_norm": 0.48071035742759705, "lr": 0.00025405749012815786, "ex_per_s": 171.5066567697813, "data_frac": 0.010708975104369014, "mem_mb": 402.962432, "nll": -6.1367034912109375, "elapsed_s": 1321.2}181{"event": "train", "step": 3560, "loss": -6.139217376708984, "grad_norm": 0.19888757169246674, "lr": 0.0002535066341686014, "ex_per_s": 171.66760916945836, "data_frac": 0.010666207273092781, "mem_mb": 402.962432, "nll": -6.139217376708984, "elapsed_s": 1327.4}182{"event": "train", "step": 3580, "loss": -5.900086402893066, "grad_norm": 0.980117678642273, "lr": 0.0002529531815285146, "ex_per_s": 171.82751515116144, "data_frac": 0.010624294804499848, "mem_mb": 402.962432, "nll": -5.900086402893066, "elapsed_s": 1333.6}183{"event": "train", "step": 3600, "loss": -6.093390941619873, "grad_norm": 0.2152511328458786, "lr": 0.00025239714844554667, "ex_per_s": 171.98465947908326, "data_frac": 0.010582583608445672, "mem_mb": 402.962432, "nll": -6.093390941619873, "elapsed_s": 1339.8}184{"event": "train", "step": 3620, "loss": -5.903722286224365, "grad_norm": 2.0552937984466553, "lr": 0.0002518385512330544, "ex_per_s": 172.1401521042927, "data_frac": 0.010541135882774048, "mem_mb": 402.962432, "nll": -5.903722286224365, "elapsed_s": 1346.1}185{"event": "train", "step": 3640, "loss": -6.05393648147583, "grad_norm": 0.35602784156799316, "lr": 0.00025127740627962274, "ex_per_s": 172.29420886383335, "data_frac": 0.010500079478954583, "mem_mb": 402.962432, "nll": -6.05393648147583, "elapsed_s": 1352.3}186{"event": "train", "step": 3660, "loss": -5.831101894378662, "grad_norm": 0.8548459410667419, "lr": 0.00025071373004858453, "ex_per_s": 172.4475308349999, "data_frac": 0.010459337833875005, "mem_mb": 402.962432, "nll": -5.831101894378662, "elapsed_s": 1358.5}187{"event": "train", "step": 3680, "loss": -5.986873149871826, "grad_norm": 0.21263039112091064, "lr": 0.00025014753907753737, "ex_per_s": 172.5992895321291, "data_frac": 0.010419063883927904, "mem_mb": 402.962432, "nll": -5.986873149871826, "elapsed_s": 1364.7}188{"event": "train", "step": 3700, "loss": -6.067651271820068, "grad_norm": 0.858372151851654, "lr": 0.0002495788499778582, "ex_per_s": 172.75019360144157, "data_frac": 0.01037910502898768, "mem_mb": 402.962432, "nll": -6.067651271820068, "elapsed_s": 1371.0}189{"event": "train", "step": 3720, "loss": -5.943192005157471, "grad_norm": 0.9326543807983398, "lr": 0.0002490076794342163, "ex_per_s": 172.89977492766258, "data_frac": 0.01033946527169308, "mem_mb": 402.962432, "nll": -5.943192005157471, "elapsed_s": 1377.2}190{"event": "train", "step": 3740, "loss": -6.188230514526367, "grad_norm": 0.42178696393966675, "lr": 0.00024843404420408344, "ex_per_s": 173.0481453488196, "data_frac": 0.010301038562224139, "mem_mb": 402.962432, "nll": -6.188230514526367, "elapsed_s": 1383.4}191{"event": "train", "step": 3760, "loss": -6.143298625946045, "grad_norm": 0.7471750974655151, "lr": 0.0002478579611172425, "ex_per_s": 173.19568416017995, "data_frac": 0.010262151807341515, "mem_mb": 402.962432, "nll": -6.143298625946045, "elapsed_s": 1389.6}192{"event": "train", "step": 3780, "loss": -5.962978839874268, "grad_norm": 0.37575408816337585, "lr": 0.00024727944707529366, "ex_per_s": 173.3404817675404, "data_frac": 0.010223993570377975, "mem_mb": 402.962432, "nll": -5.962978839874268, "elapsed_s": 1395.8}193{"event": "train", "step": 3800, "loss": -6.096671104431152, "grad_norm": 0.09653101116418839, "lr": 0.0002466985190511583, "ex_per_s": 173.4839477203616, "data_frac": 0.010185873990680033, "mem_mb": 402.962432, "nll": -6.096671104431152, "elapsed_s": 1402.1}194{"event": "train", "step": 3820, "loss": -5.9663262367248535, "grad_norm": 1.047383427619934, "lr": 0.0002461151940885812, "ex_per_s": 173.62622644826314, "data_frac": 0.010148459170280365, "mem_mb": 402.962432, "nll": -5.9663262367248535, "elapsed_s": 1408.3}195{"event": "train", "step": 3840, "loss": -5.871755599975586, "grad_norm": 0.24734215438365936, "lr": 0.0002455294893016306, "ex_per_s": 173.76732099184434, "data_frac": 0.010110840494188058, "mem_mb": 402.962432, "nll": -5.871755599975586, "elapsed_s": 1414.5}196{"event": "train", "step": 3860, "loss": -6.014182090759277, "grad_norm": 0.21969468891620636, "lr": 0.000244941421874196, "ex_per_s": 173.90685408863015, "data_frac": 0.010073900969504227, "mem_mb": 402.962432, "nll": -6.014182090759277, "elapsed_s": 1420.7}197{"event": "train", "step": 3880, "loss": -5.887627124786377, "grad_norm": 0.10920683294534683, "lr": 0.00024435100905948387, "ex_per_s": 174.0453749285378, "data_frac": 0.010037330580166407, "mem_mb": 402.962432, "nll": -5.887627124786377, "elapsed_s": 1426.9}198{"event": "train", "step": 3900, "loss": -6.1780009269714355, "grad_norm": 0.4355701804161072, "lr": 0.0002437582681795117, "ex_per_s": 174.18376594356096, "data_frac": 0.010000599257030429, "mem_mb": 402.962432, "nll": -6.1780009269714355, "elapsed_s": 1433.2}199{"event": "train", "step": 3920, "loss": -6.030921459197998, "grad_norm": 1.3763583898544312, "lr": 0.00024316321662459957, "ex_per_s": 174.3207802863324, "data_frac": 0.009964163706731451, "mem_mb": 402.962432, "nll": -6.030921459197998, "elapsed_s": 1439.4}200{"event": "train", "step": 3940, "loss": -5.920808792114258, "grad_norm": 0.30705738067626953, "lr": 0.00024256587185286015, "ex_per_s": 174.45716253616655, "data_frac": 0.00992834846074627, "mem_mb": 402.962432, "nll": -5.920808792114258, "elapsed_s": 1445.6}201{"event": "train", "step": 3960, "loss": -6.06489896774292, "grad_norm": 0.5782903432846069, "lr": 0.0002419662513896863, "ex_per_s": 174.5926621080692, "data_frac": 0.009892524755277219, "mem_mb": 402.962432, "nll": -6.06489896774292, "elapsed_s": 1451.8}202{"event": "train", "step": 3980, "loss": -5.801294803619385, "grad_norm": 0.42012736201286316, "lr": 0.00024136437282723714, "ex_per_s": 174.7254575650527, "data_frac": 0.00985707695003143, "mem_mb": 402.962432, "nll": -5.801294803619385, "elapsed_s": 1458.0}203{"event": "train", "step": 4000, "loss": -6.053316593170166, "grad_norm": 1.3878573179244995, "lr": 0.0002407602538239216, "ex_per_s": 174.85687274212134, "data_frac": 0.00982226434711655, "mem_mb": 402.962432, "nll": -6.053316593170166, "elapsed_s": 1464.2}204{"event": "val", "step": 4000, "val_nll": -6.445725998428882, "best": -6.445725998428882, "bad": 0, "mem_mb": 402.962432, "elapsed_s": 1532.5}205{"event": "train", "step": 4020, "loss": -5.988306045532227, "grad_norm": 0.5980711579322815, "lr": 0.00024015391210388054, "ex_per_s": 167.21446982028635, "data_frac": 0.009352925412093772, "mem_mb": 402.962432, "nll": -5.988306045532227, "elapsed_s": 1538.8}206{"event": "train", "step": 4040, "loss": -6.129611968994141, "grad_norm": 0.6950089335441589, "lr": 0.00023954536545646676, "ex_per_s": 167.36613444429705, "data_frac": 0.00932197950755352, "mem_mb": 402.962432, "nll": -6.129611968994141, "elapsed_s": 1545.1}207{"event": "train", "step": 4060, "loss": -5.867908000946045, "grad_norm": 0.43384379148483276, "lr": 0.0002389346317357231, "ex_per_s": 167.51640376322678, "data_frac": 0.009290918353273567, "mem_mb": 402.962432, "nll": -5.867908000946045, "elapsed_s": 1551.3}208{"event": "train", "step": 4080, "loss": -5.930375099182129, "grad_norm": 0.7557711005210876, "lr": 0.0002383217288598584, "ex_per_s": 167.66586605647467, "data_frac": 0.009260159120538918, "mem_mb": 402.962432, "nll": -5.930375099182129, "elapsed_s": 1557.6}209{"event": "train", "step": 4100, "loss": -5.938459396362305, "grad_norm": 0.2282855212688446, "lr": 0.00023770667481072205, "ex_per_s": 167.81401063983424, "data_frac": 0.00922955462767534, "mem_mb": 402.962432, "nll": -5.938459396362305, "elapsed_s": 1563.8}210{"event": "train", "step": 4120, "loss": -6.103120803833008, "grad_norm": 0.5063855051994324, "lr": 0.0002370894876332764, "ex_per_s": 167.9612712949834, "data_frac": 0.009199165643953936, "mem_mb": 402.962432, "nll": -6.103120803833008, "elapsed_s": 1570.1}211{"event": "train", "step": 4140, "loss": -5.914392471313477, "grad_norm": 0.6012131571769714, "lr": 0.00023647018543506706, "ex_per_s": 168.10784403177965, "data_frac": 0.009169168941003968, "mem_mb": 402.962432, "nll": -5.914392471313477, "elapsed_s": 1576.3}212{"event": "train", "step": 4160, "loss": -6.135086536407471, "grad_norm": 0.1510469913482666, "lr": 0.00023584878638569206, "ex_per_s": 168.25383186755093, "data_frac": 0.009139473396563536, "mem_mb": 402.962432, "nll": -6.135086536407471, "elapsed_s": 1582.6}213{"event": "train", "step": 4180, "loss": -5.972435474395752, "grad_norm": 0.22425273060798645, "lr": 0.00023522530871626857, "ex_per_s": 168.39910126788195, "data_frac": 0.00910981324703235, "mem_mb": 402.962432, "nll": -5.972435474395752, "elapsed_s": 1588.8}214{"event": "train", "step": 4200, "loss": -6.0263142585754395, "grad_norm": 0.6705049276351929, "lr": 0.0002345997707188979, "ex_per_s": 168.5433612496609, "data_frac": 0.009080489185979765, "mem_mb": 402.962432, "nll": -6.0263142585754395, "elapsed_s": 1595.0}215{"event": "train", "step": 4220, "loss": -5.963652610778809, "grad_norm": 0.71949702501297, "lr": 0.00023397219074612914, "ex_per_s": 168.68671315317894, "data_frac": 0.009051400600124432, "mem_mb": 402.962432, "nll": -5.963652610778809, "elapsed_s": 1601.3}216{"event": "train", "step": 4240, "loss": -5.992450714111328, "grad_norm": 0.9753960371017456, "lr": 0.00023334258721042025, "ex_per_s": 168.82911925530314, "data_frac": 0.009022555972741604, "mem_mb": 402.962432, "nll": -5.992450714111328, "elapsed_s": 1607.5}217{"event": "train", "step": 4260, "loss": -5.887804985046387, "grad_norm": 0.18018020689487457, "lr": 0.00023271097858359853, "ex_per_s": 168.97074258376637, "data_frac": 0.008993936379696348, "mem_mb": 402.962432, "nll": -5.887804985046387, "elapsed_s": 1613.7}218{"event": "train", "step": 4280, "loss": -5.849370956420898, "grad_norm": 0.2865234613418579, "lr": 0.00023207738339631794, "ex_per_s": 169.1112519400371, "data_frac": 0.008965299311646002, "mem_mb": 402.962432, "nll": -5.849370956420898, "elapsed_s": 1620.0}219{"event": "train", "step": 4300, "loss": -6.25685453414917, "grad_norm": 0.2018587440252304, "lr": 0.00023144182023751594, "ex_per_s": 169.2502500847578, "data_frac": 0.008937082284544486, "mem_mb": 402.962432, "nll": -6.25685453414917, "elapsed_s": 1626.2}220{"event": "train", "step": 4320, "loss": -6.132263660430908, "grad_norm": 1.214367151260376, "lr": 0.00023080430775386808, "ex_per_s": 169.38931439306833, "data_frac": 0.008909069212475902, "mem_mb": 402.962432, "nll": -6.132263660430908, "elapsed_s": 1632.4}221{"event": "train", "step": 4340, "loss": -6.081571102142334, "grad_norm": 0.6601248979568481, "lr": 0.00023016486464924057, "ex_per_s": 169.52627788863626, "data_frac": 0.00888134672437666, "mem_mb": 402.962432, "nll": -6.081571102142334, "elapsed_s": 1638.6}222{"event": "train", "step": 4360, "loss": -5.848581314086914, "grad_norm": 0.6448945999145508, "lr": 0.0002295235096841419, "ex_per_s": 169.66266875086552, "data_frac": 0.008853782494307535, "mem_mb": 402.962432, "nll": -5.848581314086914, "elapsed_s": 1644.9}223{"event": "train", "step": 4380, "loss": -6.004859924316406, "grad_norm": 0.8814634084701538, "lr": 0.00022888026167517227, "ex_per_s": 169.79865161586483, "data_frac": 0.008826374003499825, "mem_mb": 402.962432, "nll": -6.004859924316406, "elapsed_s": 1651.1}224{"event": "train", "step": 4400, "loss": -6.041928291320801, "grad_norm": 0.6646705865859985, "lr": 0.00022823513949447166, "ex_per_s": 169.93303513694883, "data_frac": 0.008799140355775798, "mem_mb": 402.962432, "nll": -6.041928291320801, "elapsed_s": 1657.3}225{"event": "train", "step": 4420, "loss": -5.896877765655518, "grad_norm": 0.2720322012901306, "lr": 0.00022758816206916584, "ex_per_s": 170.067255439617, "data_frac": 0.00877223091667288, "mem_mb": 402.962432, "nll": -5.896877765655518, "elapsed_s": 1663.5}226{"event": "train", "step": 4440, "loss": -6.089483737945557, "grad_norm": 0.2887682020664215, "lr": 0.00022693934838081143, "ex_per_s": 170.20005455425516, "data_frac": 0.00874555683688181, "mem_mb": 402.962432, "nll": -6.089483737945557, "elapsed_s": 1669.8}227{"event": "train", "step": 4460, "loss": -5.932926177978516, "grad_norm": 0.5415876507759094, "lr": 0.0002262887174648388, "ex_per_s": 170.33209642520382, "data_frac": 0.0087188938234123, "mem_mb": 402.962432, "nll": -5.932926177978516, "elapsed_s": 1676.0}228{"event": "train", "step": 4480, "loss": -6.060652732849121, "grad_norm": 0.33356985449790955, "lr": 0.00022563628840999367, "ex_per_s": 170.46294638380297, "data_frac": 0.008692465098108629, "mem_mb": 402.962432, "nll": -6.060652732849121, "elapsed_s": 1682.2}229{"event": "train", "step": 4500, "loss": -6.14852237701416, "grad_norm": 1.7588077783584595, "lr": 0.000224982080357777, "ex_per_s": 170.59345307905951, "data_frac": 0.008666215876166575, "mem_mb": 402.962432, "nll": -6.14852237701416, "elapsed_s": 1688.4}230{"event": "train", "step": 4520, "loss": -6.014455795288086, "grad_norm": 1.101172685623169, "lr": 0.00022432611250188336, "ex_per_s": 170.7228823162355, "data_frac": 0.008640156708366957, "mem_mb": 402.962432, "nll": -6.014455795288086, "elapsed_s": 1694.6}231{"event": "train", "step": 4540, "loss": -6.008378028869629, "grad_norm": 0.8486756086349487, "lr": 0.00022366840408763816, "ex_per_s": 170.8503004793215, "data_frac": 0.008614492496783593, "mem_mb": 402.962432, "nll": -6.008378028869629, "elapsed_s": 1700.9}232{"event": "train", "step": 4560, "loss": -6.044539928436279, "grad_norm": 0.4328679144382477, "lr": 0.00022300897441143247, "ex_per_s": 170.97718713041132, "data_frac": 0.008588872580672887, "mem_mb": 402.962432, "nll": -6.044539928436279, "elapsed_s": 1707.1}233{"event": "train", "step": 4580, "loss": -6.019132137298584, "grad_norm": 1.3317209482192993, "lr": 0.0002223478428201573, "ex_per_s": 171.1034876045394, "data_frac": 0.00856382305858301, "mem_mb": 402.962432, "nll": -6.019132137298584, "elapsed_s": 1713.3}234{"event": "train", "step": 4600, "loss": -5.853925704956055, "grad_norm": 1.3748066425323486, "lr": 0.0002216850287106357, "ex_per_s": 171.2288370292752, "data_frac": 0.008538668268037777, "mem_mb": 402.962432, "nll": -5.853925704956055, "elapsed_s": 1719.5}235{"event": "train", "step": 4620, "loss": -6.030454635620117, "grad_norm": 0.15149271488189697, "lr": 0.00022102055152905406, "ex_per_s": 171.35254701410526, "data_frac": 0.00851393354645868, "mem_mb": 402.962432, "nll": -6.030454635620117, "elapsed_s": 1725.8}236{"event": "train", "step": 4640, "loss": -5.835590839385986, "grad_norm": 0.37376368045806885, "lr": 0.00022035443077039114, "ex_per_s": 171.47676210004437, "data_frac": 0.008489033483456785, "mem_mb": 402.962432, "nll": -5.835590839385986, "elapsed_s": 1732.0}237{"event": "train", "step": 4660, "loss": -5.976449966430664, "grad_norm": 0.7215741872787476, "lr": 0.00021968668597784646, "ex_per_s": 171.59960071650488, "data_frac": 0.008464335015173524, "mem_mb": 402.962432, "nll": -5.976449966430664, "elapsed_s": 1738.2}238{"event": "train", "step": 4680, "loss": -6.030572414398193, "grad_norm": 0.46611031889915466, "lr": 0.00021901733674226665, "ex_per_s": 171.72170011927474, "data_frac": 0.008439755439880747, "mem_mb": 402.962432, "nll": -6.030572414398193, "elapsed_s": 1744.4}239{"event": "train", "step": 4700, "loss": -5.908234596252441, "grad_norm": 0.5777314901351929, "lr": 0.00021834640270157097, "ex_per_s": 171.84360543461594, "data_frac": 0.008415327490964639, "mem_mb": 402.962432, "nll": -5.908234596252441, "elapsed_s": 1750.6}240{"event": "train", "step": 4720, "loss": -6.065445899963379, "grad_norm": 0.19900265336036682, "lr": 0.00021767390354017485, "ex_per_s": 171.96330535207161, "data_frac": 0.008391339537160638, "mem_mb": 402.962432, "nll": -6.065445899963379, "elapsed_s": 1756.8}241{"event": "train", "step": 4740, "loss": -6.044275283813477, "grad_norm": 0.0984557643532753, "lr": 0.0002169998589884126, "ex_per_s": 172.0819309030275, "data_frac": 0.008367569365867253, "mem_mb": 402.962432, "nll": -6.044275283813477, "elapsed_s": 1763.1}242{"event": "train", "step": 4760, "loss": -5.993269920349121, "grad_norm": 0.8964200019836426, "lr": 0.00021632428882195843, "ex_per_s": 172.19956859230427, "data_frac": 0.00834415401270904, "mem_mb": 402.962432, "nll": -5.993269920349121, "elapsed_s": 1769.3}243{"event": "train", "step": 4780, "loss": -6.062222957611084, "grad_norm": 1.254046082496643, "lr": 0.00021564721286124628, "ex_per_s": 172.31769765636702, "data_frac": 0.008321022425225292, "mem_mb": 402.962432, "nll": -6.062222957611084, "elapsed_s": 1775.5}244{"event": "train", "step": 4800, "loss": -6.117073059082031, "grad_norm": 0.3326781392097473, "lr": 0.00021496865097088845, "ex_per_s": 172.43499748900345, "data_frac": 0.008298028177553991, "mem_mb": 402.962432, "nll": -6.117073059082031, "elapsed_s": 1781.7}245{"event": "train", "step": 4820, "loss": -6.032899379730225, "grad_norm": 0.22458259761333466, "lr": 0.00021428862305909247, "ex_per_s": 172.55180163303802, "data_frac": 0.00827509393444901, "mem_mb": 402.962432, "nll": -6.032899379730225, "elapsed_s": 1787.9}246{"event": "train", "step": 4840, "loss": -5.936406135559082, "grad_norm": 0.2257961630821228, "lr": 0.00021360714907707733, "ex_per_s": 172.6681052605357, "data_frac": 0.008252110123676324, "mem_mb": 402.962432, "nll": -5.936406135559082, "elapsed_s": 1794.2}247{"event": "train", "step": 4860, "loss": -5.924894332885742, "grad_norm": 0.4360259175300598, "lr": 0.00021292424901848804, "ex_per_s": 172.78388700829615, "data_frac": 0.008229255036288359, "mem_mb": 402.962432, "nll": -5.924894332885742, "elapsed_s": 1800.4}248{"event": "train", "step": 4880, "loss": -6.025900840759277, "grad_norm": 0.9050400257110596, "lr": 0.00021223994291880895, "ex_per_s": 172.8980976241909, "data_frac": 0.00820679391907628, "mem_mb": 402.962432, "nll": -6.025900840759277, "elapsed_s": 1806.6}249{"event": "train", "step": 4900, "loss": -5.9571990966796875, "grad_norm": 0.46810781955718994, "lr": 0.00021155425085477597, "ex_per_s": 173.01220005521637, "data_frac": 0.008184291705549558, "mem_mb": 402.962432, "nll": -5.9571990966796875, "elapsed_s": 1812.8}250{"event": "train", "step": 4920, "loss": -5.946750640869141, "grad_norm": 0.3089849650859833, "lr": 0.00021086719294378776, "ex_per_s": 173.12396065697044, "data_frac": 0.008161899539375631, "mem_mb": 402.962432, "nll": -5.946750640869141, "elapsed_s": 1819.0}251{"event": "train", "step": 4940, "loss": -6.04337215423584, "grad_norm": 0.1655796766281128, "lr": 0.00021017878934331512, "ex_per_s": 173.23465975014, "data_frac": 0.008139588937733169, "mem_mb": 402.962432, "nll": -6.04337215423584, "elapsed_s": 1825.2}252{"event": "train", "step": 4960, "loss": -6.065276622772217, "grad_norm": 0.38418054580688477, "lr": 0.00020948906025030994, "ex_per_s": 173.3450465005296, "data_frac": 0.00811729340618195, "mem_mb": 402.962432, "nll": -6.065276622772217, "elapsed_s": 1831.5}253{"event": "train", "step": 4980, "loss": -6.072689056396484, "grad_norm": 0.36151036620140076, "lr": 0.00020879802590061233, "ex_per_s": 173.45493699180494, "data_frac": 0.00809525090155133, "mem_mb": 402.962432, "nll": -6.072689056396484, "elapsed_s": 1837.7}254{"event": "train", "step": 5000, "loss": -5.879858016967773, "grad_norm": 0.8225620985031128, "lr": 0.00020810570656835728, "ex_per_s": 173.56386229019915, "data_frac": 0.008073546229211954, "mem_mb": 402.962432, "nll": -5.879858016967773, "elapsed_s": 1843.9}255{"event": "val", "step": 5000, "val_nll": -6.4489264462635445, "best": -6.4489264462635445, "bad": 0, "mem_mb": 402.962432, "elapsed_s": 1912.6}256{"event": "train", "step": 5020, "loss": -5.963318347930908, "grad_norm": 0.13394220173358917, "lr": 0.0002074121225653795, "ex_per_s": 167.44646388996145, "data_frac": 0.007763466076474562, "mem_mb": 402.962432, "nll": -5.963318347930908, "elapsed_s": 1918.9}257{"event": "train", "step": 5040, "loss": -6.088360786437988, "grad_norm": 0.6430414319038391, "lr": 0.00020671729424061788, "ex_per_s": 167.56717476756788, "data_frac": 0.00774398795084293, "mem_mb": 402.962432, "nll": -6.088360786437988, "elapsed_s": 1925.2}258{"event": "train", "step": 5060, "loss": -6.006345748901367, "grad_norm": 0.2772347033023834, "lr": 0.00020602124197951796, "ex_per_s": 167.68715973912308, "data_frac": 0.00772450479614808, "mem_mb": 402.962432, "nll": -6.006345748901367, "elapsed_s": 1931.4}259{"event": "train", "step": 5080, "loss": -6.12503719329834, "grad_norm": 1.0095676183700562, "lr": 0.00020532398620343433, "ex_per_s": 167.80669764500698, "data_frac": 0.007704881079085077, "mem_mb": 402.962432, "nll": -6.12503719329834, "elapsed_s": 1937.7}260{"event": "train", "step": 5100, "loss": -6.062652587890625, "grad_norm": 0.28471216559410095, "lr": 0.0002046255473690312, "ex_per_s": 167.92630237229844, "data_frac": 0.007685404929864102, "mem_mb": 402.962432, "nll": -6.062652587890625, "elapsed_s": 1943.9}261{"event": "train", "step": 5120, "loss": -6.002140522003174, "grad_norm": 0.4218657314777374, "lr": 0.00020392594596768237, "ex_per_s": 168.04555095190844, "data_frac": 0.007665976919445198, "mem_mb": 402.962432, "nll": -6.002140522003174, "elapsed_s": 1950.1}262{"event": "train", "step": 5140, "loss": -5.946596622467041, "grad_norm": 1.1655378341674805, "lr": 0.00020322520252486997, "ex_per_s": 168.16385525088432, "data_frac": 0.007646660064553952, "mem_mb": 402.962432, "nll": -5.946596622467041, "elapsed_s": 1956.4}263{"event": "train", "step": 5160, "loss": -6.16330623626709, "grad_norm": 0.36529597640037537, "lr": 0.0002025233375995822, "ex_per_s": 168.28108778943204, "data_frac": 0.007628060973999831, "mem_mb": 402.962432, "nll": -6.16330623626709, "elapsed_s": 1962.6}264{"event": "train", "step": 5180, "loss": -5.939869403839111, "grad_norm": 0.18888939917087555, "lr": 0.0002018203717837105, "ex_per_s": 168.3977236525145, "data_frac": 0.00760943595944707, "mem_mb": 402.962432, "nll": -5.939869403839111, "elapsed_s": 1968.9}265{"event": "train", "step": 5200, "loss": -5.826303482055664, "grad_norm": 0.1316034495830536, "lr": 0.00020111632570144484, "ex_per_s": 168.51579415256265, "data_frac": 0.007590686912689042, "mem_mb": 402.962432, "nll": -5.826303482055664, "elapsed_s": 1975.1}266{"event": "train", "step": 5220, "loss": -5.922908782958984, "grad_norm": 0.2656936049461365, "lr": 0.00020041122000866903, "ex_per_s": 168.6330713830728, "data_frac": 0.007571895165786695, "mem_mb": 402.962432, "nll": -5.922908782958984, "elapsed_s": 1981.3}267{"event": "train", "step": 5240, "loss": -5.7782673835754395, "grad_norm": 0.08759492635726929, "lr": 0.00019970507539235469, "ex_per_s": 168.7497737327068, "data_frac": 0.007553517950898343, "mem_mb": 402.962432, "nll": -5.7782673835754395, "elapsed_s": 1987.5}268{"event": "train", "step": 5260, "loss": -5.946219444274902, "grad_norm": 0.21369138360023499, "lr": 0.0001989979125699542, "ex_per_s": 168.86567023412408, "data_frac": 0.0075354779900296195, "mem_mb": 402.962432, "nll": -5.946219444274902, "elapsed_s": 1993.7}269{"event": "train", "step": 5280, "loss": -6.003143310546875, "grad_norm": 1.0139466524124146, "lr": 0.00019828975228879288, "ex_per_s": 168.97863021563782, "data_frac": 0.007518223560362105, "mem_mb": 402.962432, "nll": -6.003143310546875, "elapsed_s": 2000.0}270{"event": "train", "step": 5300, "loss": -5.894676685333252, "grad_norm": 0.21179679036140442, "lr": 0.00019758061532546025, "ex_per_s": 169.09265953341074, "data_frac": 0.007499994184537997, "mem_mb": 402.962432, "nll": -5.894676685333252, "elapsed_s": 2006.2}271{"event": "train", "step": 5320, "loss": -6.0689897537231445, "grad_norm": 1.588341474533081, "lr": 0.0001968705224852007, "ex_per_s": 169.20524930568604, "data_frac": 0.007482068278848647, "mem_mb": 402.962432, "nll": -6.0689897537231445, "elapsed_s": 2012.4}272{"event": "train", "step": 5340, "loss": -5.972391605377197, "grad_norm": 1.545356035232544, "lr": 0.00019615949460130276, "ex_per_s": 169.31893402224514, "data_frac": 0.0074640658142335415, "mem_mb": 402.962432, "nll": -5.972391605377197, "elapsed_s": 2018.6}273{"event": "train", "step": 5360, "loss": -6.007456302642822, "grad_norm": 0.8819142580032349, "lr": 0.00019544755253448823, "ex_per_s": 169.43059351138928, "data_frac": 0.007446364689075864, "mem_mb": 402.962432, "nll": -6.007456302642822, "elapsed_s": 2024.9}274{"event": "train", "step": 5380, "loss": -5.85875940322876, "grad_norm": 0.16367821395397186, "lr": 0.0001947347171722997, "ex_per_s": 169.5430771479248, "data_frac": 0.007428729696024308, "mem_mb": 402.962432, "nll": -5.85875940322876, "elapsed_s": 2031.1}275{"event": "train", "step": 5400, "loss": -5.9532246589660645, "grad_norm": 0.10501307994127274, "lr": 0.00019402100942848827, "ex_per_s": 169.65385158198248, "data_frac": 0.007411152522638257, "mem_mb": 402.962432, "nll": -5.9532246589660645, "elapsed_s": 2037.3}276{"event": "train", "step": 5420, "loss": -5.839375972747803, "grad_norm": 0.36099880933761597, "lr": 0.00019330645024239946, "ex_per_s": 169.7638674948831, "data_frac": 0.007393588475406213, "mem_mb": 402.962432, "nll": -5.839375972747803, "elapsed_s": 2043.5}277{"event": "train", "step": 5440, "loss": -6.023930072784424, "grad_norm": 0.8194254636764526, "lr": 0.00019259106057835926, "ex_per_s": 169.8742324269518, "data_frac": 0.00737612132053819, "mem_mb": 402.962432, "nll": -6.023930072784424, "elapsed_s": 2049.7}278{"event": "train", "step": 5460, "loss": -6.099534511566162, "grad_norm": 0.4170929193496704, "lr": 0.00019187486142505883, "ex_per_s": 169.98348504147535, "data_frac": 0.007358615910452437, "mem_mb": 402.962432, "nll": -6.099534511566162, "elapsed_s": 2055.9}279{"event": "train", "step": 5480, "loss": -6.101171970367432, "grad_norm": 0.3425576686859131, "lr": 0.00019115787379493887, "ex_per_s": 170.09132393043956, "data_frac": 0.007341369320423095, "mem_mb": 402.962432, "nll": -6.101171970367432, "elapsed_s": 2062.1}280{"event": "train", "step": 5500, "loss": -6.079975128173828, "grad_norm": 1.1244423389434814, "lr": 0.00019044011872357296, "ex_per_s": 170.19753325762, "data_frac": 0.007324595960652918, "mem_mb": 402.962432, "nll": -6.079975128173828, "elapsed_s": 2068.4}281{"event": "train", "step": 5520, "loss": -5.928621292114258, "grad_norm": 0.5367259979248047, "lr": 0.00018972161726905055, "ex_per_s": 170.30407363221315, "data_frac": 0.0073076765291994385, "mem_mb": 402.962432, "nll": -5.928621292114258, "elapsed_s": 2074.6}282{"event": "train", "step": 5540, "loss": -5.918962478637695, "grad_norm": 0.7860292196273804, "lr": 0.00018900239051135913, "ex_per_s": 170.41018000760175, "data_frac": 0.007291118397652003, "mem_mb": 402.962432, "nll": -5.918962478637695, "elapsed_s": 2080.8}283{"event": "train", "step": 5560, "loss": -5.976602554321289, "grad_norm": 0.6437584161758423, "lr": 0.00018828245955176568, "ex_per_s": 170.51612308702775, "data_frac": 0.007274264717419442, "mem_mb": 402.962432, "nll": -5.976602554321289, "elapsed_s": 2087.0}284{"event": "train", "step": 5580, "loss": -6.030399322509766, "grad_norm": 0.34126681089401245, "lr": 0.00018756184551219765, "ex_per_s": 170.62174986852358, "data_frac": 0.007257653129507772, "mem_mb": 402.962432, "nll": -6.030399322509766, "elapsed_s": 2093.2}285{"event": "train", "step": 5600, "loss": -6.083240509033203, "grad_norm": 0.7856508493423462, "lr": 0.00018684056953462326, "ex_per_s": 170.72671196639772, "data_frac": 0.00724107260573521, "mem_mb": 402.962432, "nll": -6.083240509033203, "elapsed_s": 2099.5}286{"event": "train", "step": 5620, "loss": -5.819654941558838, "grad_norm": 0.32190558314323425, "lr": 0.00018611865278043117, "ex_per_s": 170.82985727491297, "data_frac": 0.007224626104271378, "mem_mb": 402.962432, "nll": -5.819654941558838, "elapsed_s": 2105.7}287{"event": "train", "step": 5640, "loss": -6.117583274841309, "grad_norm": 1.196703314781189, "lr": 0.00018539611642980975, "ex_per_s": 170.93256781948205, "data_frac": 0.007208463060029928, "mem_mb": 402.962432, "nll": -6.117583274841309, "elapsed_s": 2111.9}288{"event": "train", "step": 5660, "loss": -6.009285926818848, "grad_norm": 0.7244869470596313, "lr": 0.00018467298168112556, "ex_per_s": 171.03359692118843, "data_frac": 0.007192060480493254, "mem_mb": 402.962432, "nll": -6.009285926818848, "elapsed_s": 2118.1}289{"event": "train", "step": 5680, "loss": -5.965942859649658, "grad_norm": 0.4052329361438751, "lr": 0.00018394926975030145, "ex_per_s": 171.13489444774865, "data_frac": 0.007175894280622576, "mem_mb": 402.962432, "nll": -5.965942859649658, "elapsed_s": 2124.4}290{"event": "train", "step": 5700, "loss": -6.030463695526123, "grad_norm": 1.2005029916763306, "lr": 0.00018322500187019415, "ex_per_s": 171.23530211528904, "data_frac": 0.007159734911324642, "mem_mb": 402.962432, "nll": -6.030463695526123, "elapsed_s": 2130.6}291{"event": "train", "step": 5720, "loss": -5.832675933837891, "grad_norm": 0.19435912370681763, "lr": 0.00018250019928997124, "ex_per_s": 171.33563751013224, "data_frac": 0.007143599890740043, "mem_mb": 402.962432, "nll": -5.832675933837891, "elapsed_s": 2136.8}292{"event": "train", "step": 5740, "loss": -5.998289108276367, "grad_norm": 0.32657572627067566, "lr": 0.00018177488327448776, "ex_per_s": 171.43578966000783, "data_frac": 0.0071275369234387, "mem_mb": 402.962432, "nll": -5.998289108276367, "elapsed_s": 2143.0}293{"event": "train", "step": 5760, "loss": -5.9973907470703125, "grad_norm": 0.2571961581707001, "lr": 0.00018104907510366235, "ex_per_s": 171.53541547243478, "data_frac": 0.007111723990203747, "mem_mb": 402.962432, "nll": -5.9973907470703125, "elapsed_s": 2149.3}294{"event": "train", "step": 5780, "loss": -6.008396625518799, "grad_norm": 0.16593782603740692, "lr": 0.00018032279607185304, "ex_per_s": 171.63513796374542, "data_frac": 0.007096014453857635, "mem_mb": 402.962432, "nll": -6.008396625518799, "elapsed_s": 2155.5}295{"event": "train", "step": 5800, "loss": -5.9827494621276855, "grad_norm": 1.2352627515792847, "lr": 0.00017959606748723213, "ex_per_s": 171.73449906657507, "data_frac": 0.007080391759625103, "mem_mb": 402.962432, "nll": -5.9827494621276855, "elapsed_s": 2161.7}296{"event": "train", "step": 5820, "loss": -5.904667377471924, "grad_norm": 1.2810490131378174, "lr": 0.00017886891067116147, "ex_per_s": 171.83267738168675, "data_frac": 0.007064760586837069, "mem_mb": 402.962432, "nll": -5.904667377471924, "elapsed_s": 2167.9}297{"event": "train", "step": 5840, "loss": -5.912580490112305, "grad_norm": 0.7933849692344666, "lr": 0.00017814134695756664, "ex_per_s": 171.93053952040944, "data_frac": 0.007049171468364267, "mem_mb": 402.962432, "nll": -5.912580490112305, "elapsed_s": 2174.1}298{"event": "train", "step": 5860, "loss": -6.002255916595459, "grad_norm": 0.18979892134666443, "lr": 0.00017741339769231104, "ex_per_s": 172.02661829302085, "data_frac": 0.007033931202455845, "mem_mb": 402.962432, "nll": -6.002255916595459, "elapsed_s": 2180.3}299{"event": "train", "step": 5880, "loss": -6.016279220581055, "grad_norm": 0.24562188982963562, "lr": 0.00017668508423256986, "ex_per_s": 172.12267799615418, "data_frac": 0.00701858456361713, "mem_mb": 402.962432, "nll": -6.016279220581055, "elapsed_s": 2186.5}300{"event": "train", "step": 5900, "loss": -6.140002727508545, "grad_norm": 0.6389336585998535, "lr": 0.0001759564279462032, "ex_per_s": 172.21850671042762, "data_frac": 0.007003202818724838, "mem_mb": 402.962432, "nll": -6.140002727508545, "elapsed_s": 2192.8}301{"event": "train", "step": 5920, "loss": -6.002124786376953, "grad_norm": 0.2604208290576935, "lr": 0.00017522745021112944, "ex_per_s": 172.31376370374593, "data_frac": 0.0069884120135421625, "mem_mb": 402.962432, "nll": -6.002124786376953, "elapsed_s": 2199.0}302{"event": "train", "step": 5940, "loss": -5.942373275756836, "grad_norm": 0.1681799441576004, "lr": 0.0001744981724146976, "ex_per_s": 172.40888813857927, "data_frac": 0.006973346090813007, "mem_mb": 402.962432, "nll": -5.942373275756836, "elapsed_s": 2205.2}303{"event": "train", "step": 5960, "loss": -5.9048261642456055, "grad_norm": 0.3486751317977905, "lr": 0.00017376861595306043, "ex_per_s": 172.5031656118506, "data_frac": 0.006958316076694589, "mem_mb": 402.962432, "nll": -5.9048261642456055, "elapsed_s": 2211.4}304{"event": "train", "step": 5980, "loss": -6.03150749206543, "grad_norm": 0.5117249488830566, "lr": 0.00017303880223054632, "ex_per_s": 172.59706682562103, "data_frac": 0.006943662552689964, "mem_mb": 402.962432, "nll": -6.03150749206543, "elapsed_s": 2217.6}305{"event": "train", "step": 6000, "loss": -5.9896769523620605, "grad_norm": 0.14993490278720856, "lr": 0.00017230875265903138, "ex_per_s": 172.69031351728145, "data_frac": 0.006928855936063942, "mem_mb": 402.962432, "nll": -5.9896769523620605, "elapsed_s": 2223.8}306{"event": "val", "step": 6000, "val_nll": -6.4506186911372145, "best": -6.4506186911372145, "bad": 0, "mem_mb": 402.962432, "elapsed_s": 2292.3}307{"event": "train", "step": 6020, "loss": -5.848527431488037, "grad_norm": 0.2666621208190918, "lr": 0.0001715784886573111, "ex_per_s": 167.62796549127097, "data_frac": 0.006708013056252328, "mem_mb": 402.962432, "nll": -5.848527431488037, "elapsed_s": 2298.6}308{"event": "train", "step": 6040, "loss": -5.862646102905273, "grad_norm": 1.1279398202896118, "lr": 0.0001708480316504724, "ex_per_s": 167.72932211510644, "data_frac": 0.006694271250730104, "mem_mb": 402.962432, "nll": -5.862646102905273, "elapsed_s": 2304.9}309{"event": "train", "step": 6060, "loss": -6.024358749389648, "grad_norm": 1.198371171951294, "lr": 0.00017011740306926445, "ex_per_s": 167.83035500250475, "data_frac": 0.006680690903868884, "mem_mb": 402.962432, "nll": -6.024358749389648, "elapsed_s": 2311.1}310{"event": "train", "step": 6080, "loss": -6.0070600509643555, "grad_norm": 0.12183865904808044, "lr": 0.00016938662434947037, "ex_per_s": 167.9310532276669, "data_frac": 0.0066672852289124005, "mem_mb": 402.962432, "nll": -6.0070600509643555, "elapsed_s": 2317.3}311{"event": "train", "step": 6100, "loss": -6.136606693267822, "grad_norm": 1.0656052827835083, "lr": 0.00016865571693127818, "ex_per_s": 168.0315204566369, "data_frac": 0.006653761933027376, "mem_mb": 402.962432, "nll": -6.136606693267822, "elapsed_s": 2323.6}312{"event": "train", "step": 6120, "loss": -5.995362758636475, "grad_norm": 0.3057081699371338, "lr": 0.00016792470225865166, "ex_per_s": 168.13117627677894, "data_frac": 0.006640432300573587, "mem_mb": 402.962432, "nll": -5.995362758636475, "elapsed_s": 2329.8}313{"event": "train", "step": 6140, "loss": -6.002242088317871, "grad_norm": 0.09592124074697495, "lr": 0.00016719360177870138, "ex_per_s": 168.23131621096556, "data_frac": 0.006627120580126769, "mem_mb": 402.962432, "nll": -6.002242088317871, "elapsed_s": 2336.0}314{"event": "train", "step": 6160, "loss": -6.064213752746582, "grad_norm": 0.47827446460723877, "lr": 0.0001664624369410554, "ex_per_s": 168.33191057025508, "data_frac": 0.006613732359340341, "mem_mb": 402.962432, "nll": -6.064213752746582, "elapsed_s": 2342.2}315{"event": "train", "step": 6180, "loss": -6.0418806076049805, "grad_norm": 0.4654652178287506, "lr": 0.00016573122919722992, "ex_per_s": 168.4316132193384, "data_frac": 0.006600516267741302, "mem_mb": 402.962432, "nll": -6.0418806076049805, "elapsed_s": 2348.4}316{"event": "train", "step": 6200, "loss": -5.991701126098633, "grad_norm": 0.316052109003067, "lr": 0.000165, "ex_per_s": 168.53122676899693, "data_frac": 0.006587344988867979, "mem_mb": 402.962432, "nll": -5.991701126098633, "elapsed_s": 2354.7}317{"event": "train", "step": 6220, "loss": -5.932340621948242, "grad_norm": 0.594242513179779, "lr": 0.00016426877080277007, "ex_per_s": 168.62913395346558, "data_frac": 0.006574076283399035, "mem_mb": 402.962432, "nll": -5.932340621948242, "elapsed_s": 2360.9}318{"event": "train", "step": 6240, "loss": -5.905035495758057, "grad_norm": 0.3939434289932251, "lr": 0.0001635375630589446, "ex_per_s": 168.72652319927656, "data_frac": 0.006561251952780359, "mem_mb": 402.962432, "nll": -5.905035495758057, "elapsed_s": 2367.1}319{"event": "train", "step": 6260, "loss": -5.906564235687256, "grad_norm": 0.1427047699689865, "lr": 0.00016280639822129864, "ex_per_s": 168.82354932547156, "data_frac": 0.006548207556892099, "mem_mb": 402.962432, "nll": -5.906564235687256, "elapsed_s": 2373.3}320{"event": "train", "step": 6280, "loss": -5.900151252746582, "grad_norm": 0.8367927074432373, "lr": 0.00016207529774134834, "ex_per_s": 168.9200319529077, "data_frac": 0.006535194710840197, "mem_mb": 402.962432, "nll": -5.900151252746582, "elapsed_s": 2379.5}321{"event": "train", "step": 6300, "loss": -5.985235214233398, "grad_norm": 0.25111791491508484, "lr": 0.00016134428306872187, "ex_per_s": 169.0160773158177, "data_frac": 0.006522282275973542, "mem_mb": 402.962432, "nll": -5.985235214233398, "elapsed_s": 2385.8}322{"event": "train", "step": 6320, "loss": -5.824914932250977, "grad_norm": 0.30666986107826233, "lr": 0.00016061337565052966, "ex_per_s": 169.11141262950102, "data_frac": 0.00650957155809313, "mem_mb": 402.962432, "nll": -5.824914932250977, "elapsed_s": 2392.0}323{"event": "train", "step": 6340, "loss": -5.881357192993164, "grad_norm": 0.11378569900989532, "lr": 0.00015988259693073558, "ex_per_s": 169.2063627216808, "data_frac": 0.006496801484777407, "mem_mb": 402.962432, "nll": -5.881357192993164, "elapsed_s": 2398.2}324{"event": "train", "step": 6360, "loss": -6.164787292480469, "grad_norm": 0.30635586380958557, "lr": 0.0001591519683495276, "ex_per_s": 169.30089032315433, "data_frac": 0.006484116122006258, "mem_mb": 402.962432, "nll": -6.164787292480469, "elapsed_s": 2404.4}325{"event": "train", "step": 6380, "loss": -5.896584987640381, "grad_norm": 1.0949552059173584, "lr": 0.0001584215113426889, "ex_per_s": 169.39532503825853, "data_frac": 0.0064718255741072015, "mem_mb": 402.962432, "nll": -5.896584987640381, "elapsed_s": 2410.6}326{"event": "train", "step": 6400, "loss": -5.888556957244873, "grad_norm": 0.514153778553009, "lr": 0.00015769124734096862, "ex_per_s": 169.48895983675573, "data_frac": 0.006459299488166643, "mem_mb": 402.962432, "nll": -5.888556957244873, "elapsed_s": 2416.9}327{"event": "train", "step": 6420, "loss": -5.9845871925354, "grad_norm": 0.24108901619911194, "lr": 0.00015696119776945368, "ex_per_s": 169.58151961532533, "data_frac": 0.006446805584866279, "mem_mb": 402.962432, "nll": -5.9845871925354, "elapsed_s": 2423.1}328{"event": "train", "step": 6440, "loss": -5.974432945251465, "grad_norm": 0.9779754877090454, "lr": 0.00015623138404693952, "ex_per_s": 169.67364305148783, "data_frac": 0.006434349434941761, "mem_mb": 402.962432, "nll": -5.974432945251465, "elapsed_s": 2429.3}329{"event": "train", "step": 6460, "loss": -5.927667617797852, "grad_norm": 0.3492763936519623, "lr": 0.0001555018275853024, "ex_per_s": 169.7656596363418, "data_frac": 0.00642206730899038, "mem_mb": 402.962432, "nll": -5.927667617797852, "elapsed_s": 2435.6}330{"event": "train", "step": 6480, "loss": -5.932928085327148, "grad_norm": 0.8329316973686218, "lr": 0.0001547725497888706, "ex_per_s": 169.8573403916162, "data_frac": 0.00640962217323191, "mem_mb": 402.962432, "nll": -5.932928085327148, "elapsed_s": 2441.8}331{"event": "train", "step": 6500, "loss": -5.956485271453857, "grad_norm": 0.24409914016723633, "lr": 0.0001540435720537968, "ex_per_s": 169.94841590900901, "data_frac": 0.0063974093081004425, "mem_mb": 402.962432, "nll": -5.956485271453857, "elapsed_s": 2448.0}332{"event": "train", "step": 6520, "loss": -6.039414405822754, "grad_norm": 0.32345643639564514, "lr": 0.0001533149157674301, "ex_per_s": 170.03916635450582, "data_frac": 0.006385255266000318, "mem_mb": 402.962432, "nll": -6.039414405822754, "elapsed_s": 2454.2}333{"event": "train", "step": 6540, "loss": -6.1584978103637695, "grad_norm": 0.4873272478580475, "lr": 0.00015258660230768893, "ex_per_s": 170.1294027843288, "data_frac": 0.006373145124764855, "mem_mb": 402.962432, "nll": -6.1584978103637695, "elapsed_s": 2460.4}334{"event": "train", "step": 6560, "loss": -6.048852443695068, "grad_norm": 1.0007928609848022, "lr": 0.0001518586530424334, "ex_per_s": 170.21928029636575, "data_frac": 0.006361003599944653, "mem_mb": 402.962432, "nll": -6.048852443695068, "elapsed_s": 2466.7}335{"event": "train", "step": 6580, "loss": -6.042402744293213, "grad_norm": 0.7366060614585876, "lr": 0.00015113108932883853, "ex_per_s": 170.30894883772254, "data_frac": 0.006349068958443762, "mem_mb": 402.962432, "nll": -6.042402744293213, "elapsed_s": 2472.9}336{"event": "train", "step": 6600, "loss": -6.010514736175537, "grad_norm": 0.9423399567604065, "lr": 0.00015040393251276787, "ex_per_s": 170.39737023847542, "data_frac": 0.006336992752340792, "mem_mb": 402.962432, "nll": -6.010514736175537, "elapsed_s": 2479.1}337{"event": "train", "step": 6620, "loss": -6.20301628112793, "grad_norm": 0.13075625896453857, "lr": 0.00014967720392814694, "ex_per_s": 170.48522167516467, "data_frac": 0.00632515888294331, "mem_mb": 402.962432, "nll": -6.20301628112793, "elapsed_s": 2485.3}338{"event": "train", "step": 6640, "loss": -6.099913597106934, "grad_norm": 0.5732356905937195, "lr": 0.00014895092489633762, "ex_per_s": 170.5731650960092, "data_frac": 0.006313435260776171, "mem_mb": 402.962432, "nll": -6.099913597106934, "elapsed_s": 2491.6}339{"event": "train", "step": 6660, "loss": -5.970026016235352, "grad_norm": 0.253370463848114, "lr": 0.00014822511672551224, "ex_per_s": 170.6608365726498, "data_frac": 0.006301638832038421, "mem_mb": 402.962432, "nll": -5.970026016235352, "elapsed_s": 2497.8}340{"event": "train", "step": 6680, "loss": -5.799136638641357, "grad_norm": 0.2497052103281021, "lr": 0.0001474998007100288, "ex_per_s": 170.7480789348525, "data_frac": 0.006289974196856321, "mem_mb": 402.962432, "nll": -5.799136638641357, "elapsed_s": 2504.0}341{"event": "train", "step": 6700, "loss": -5.8587517738342285, "grad_norm": 1.2802735567092896, "lr": 0.00014677499812980582, "ex_per_s": 170.83490779649674, "data_frac": 0.006278395901675819, "mem_mb": 402.962432, "nll": -5.8587517738342285, "elapsed_s": 2510.2}342{"event": "train", "step": 6720, "loss": -6.209876537322998, "grad_norm": 0.9127509593963623, "lr": 0.0001460507302496985, "ex_per_s": 170.92140288798944, "data_frac": 0.006266800085018975, "mem_mb": 402.962432, "nll": -6.209876537322998, "elapsed_s": 2516.4}343{"event": "train", "step": 6740, "loss": -5.971711158752441, "grad_norm": 0.12166061997413635, "lr": 0.0001453270183188744, "ex_per_s": 171.00741558233037, "data_frac": 0.006255284444248027, "mem_mb": 402.962432, "nll": -5.971711158752441, "elapsed_s": 2522.7}344{"event": "train", "step": 6760, "loss": -5.903552532196045, "grad_norm": 0.8457618951797485, "lr": 0.00014460388357019025, "ex_per_s": 171.09308326337117, "data_frac": 0.006243880534457735, "mem_mb": 402.962432, "nll": -5.903552532196045, "elapsed_s": 2528.9}345{"event": "train", "step": 6780, "loss": -5.996397495269775, "grad_norm": 0.26182684302330017, "lr": 0.0001438813472195688, "ex_per_s": 171.17848789625415, "data_frac": 0.006232430324682615, "mem_mb": 402.962432, "nll": -5.996397495269775, "elapsed_s": 2535.1}346{"event": "train", "step": 6800, "loss": -6.0887579917907715, "grad_norm": 0.13881723582744598, "lr": 0.00014315943046537674, "ex_per_s": 171.26305039400697, "data_frac": 0.006221138938649973, "mem_mb": 402.962432, "nll": -6.0887579917907715, "elapsed_s": 2541.3}347{"event": "train", "step": 6820, "loss": -6.107300758361816, "grad_norm": 0.5961145758628845, "lr": 0.00014243815448780235, "ex_per_s": 171.3470079418878, "data_frac": 0.006209884062481218, "mem_mb": 402.962432, "nll": -6.107300758361816, "elapsed_s": 2547.5}348{"event": "train", "step": 6840, "loss": -5.8319902420043945, "grad_norm": 0.281216561794281, "lr": 0.00014171754044823431, "ex_per_s": 171.4305569456672, "data_frac": 0.006198620880363398, "mem_mb": 402.962432, "nll": -5.8319902420043945, "elapsed_s": 2553.8}349{"event": "train", "step": 6860, "loss": -5.948894500732422, "grad_norm": 0.20376679301261902, "lr": 0.0001409976094886409, "ex_per_s": 171.51389628548293, "data_frac": 0.006187351117653615, "mem_mb": 402.962432, "nll": -5.948894500732422, "elapsed_s": 2560.0}350{"event": "train", "step": 6880, "loss": -6.026114463806152, "grad_norm": 0.3162730932235718, "lr": 0.00014027838273094942, "ex_per_s": 171.59675143547707, "data_frac": 0.00617624063081258, "mem_mb": 402.962432, "nll": -6.026114463806152, "elapsed_s": 2566.2}351{"event": "train", "step": 6900, "loss": -6.00633430480957, "grad_norm": 0.35987409949302673, "lr": 0.00013955988127642706, "ex_per_s": 171.6790281854791, "data_frac": 0.0061654796989765075, "mem_mb": 402.962432, "nll": -6.00633430480957, "elapsed_s": 2572.4}352{"event": "train", "step": 6920, "loss": -6.064913272857666, "grad_norm": 1.0396634340286255, "lr": 0.0001388421262050611, "ex_per_s": 171.76133220538617, "data_frac": 0.006154410909406029, "mem_mb": 402.962432, "nll": -6.064913272857666, "elapsed_s": 2578.7}353{"event": "train", "step": 6940, "loss": -6.063670635223389, "grad_norm": 0.4161675274372101, "lr": 0.00013812513857494116, "ex_per_s": 171.84343883347844, "data_frac": 0.0061433123730768695, "mem_mb": 402.962432, "nll": -6.063670635223389, "elapsed_s": 2584.9}354{"event": "train", "step": 6960, "loss": -5.742068767547607, "grad_norm": 0.17276278138160706, "lr": 0.00013740893942164073, "ex_per_s": 171.92517194824882, "data_frac": 0.006132365739370617, "mem_mb": 402.962432, "nll": -5.742068767547607, "elapsed_s": 2591.1}355{"event": "train", "step": 6980, "loss": -6.196084022521973, "grad_norm": 0.8734515905380249, "lr": 0.0001366935497576006, "ex_per_s": 172.0064688253392, "data_frac": 0.006121490104403234, "mem_mb": 402.962432, "nll": -6.196084022521973, "elapsed_s": 2597.3}356{"event": "train", "step": 7000, "loss": -6.123588562011719, "grad_norm": 0.7763359546661377, "lr": 0.00013597899057151173, "ex_per_s": 172.08696864153217, "data_frac": 0.006110613434288598, "mem_mb": 402.962432, "nll": -6.123588562011719, "elapsed_s": 2603.5}357{"event": "val", "step": 7000, "val_nll": -6.4508794186491585, "best": -6.4508794186491585, "bad": 0, "mem_mb": 402.962432, "elapsed_s": 2672.2}358{"event": "train", "step": 7020, "loss": -6.14436674118042, "grad_norm": 0.29105594754219055, "lr": 0.0001352652828277003, "ex_per_s": 167.74966257779369, "data_frac": 0.005943710082333315, "mem_mb": 402.962432, "nll": -6.14436674118042, "elapsed_s": 2678.5}359{"event": "train", "step": 7040, "loss": -6.057741165161133, "grad_norm": 0.08071581274271011, "lr": 0.00013455244746551177, "ex_per_s": 167.83609844610217, "data_frac": 0.005933767350018824, "mem_mb": 402.962432, "nll": -6.057741165161133, "elapsed_s": 2684.7}360{"event": "train", "step": 7060, "loss": -5.989640712738037, "grad_norm": 1.0020086765289307, "lr": 0.00013384050539869724, "ex_per_s": 167.92182076872814, "data_frac": 0.005923760622628078, "mem_mb": 402.962432, "nll": -5.989640712738037, "elapsed_s": 2691.0}361{"event": "train", "step": 7080, "loss": -6.067507743835449, "grad_norm": 1.6379846334457397, "lr": 0.00013312947751479927, "ex_per_s": 168.0073969777036, "data_frac": 0.005913769002522967, "mem_mb": 402.962432, "nll": -6.067507743835449, "elapsed_s": 2697.2}362{"event": "train", "step": 7100, "loss": -5.964295387268066, "grad_norm": 0.2642773687839508, "lr": 0.00013241938467453975, "ex_per_s": 168.0932001672138, "data_frac": 0.005903836263881396, "mem_mb": 402.962432, "nll": -5.964295387268066, "elapsed_s": 2703.5}363{"event": "train", "step": 7120, "loss": -6.043099403381348, "grad_norm": 0.6463243961334229, "lr": 0.00013171024771120715, "ex_per_s": 168.1784916365957, "data_frac": 0.005894417963599133, "mem_mb": 402.962432, "nll": -6.043099403381348, "elapsed_s": 2709.7}364{"event": "train", "step": 7140, "loss": -5.944856643676758, "grad_norm": 0.18349876999855042, "lr": 0.0001310020874300458, "ex_per_s": 168.26382294327465, "data_frac": 0.005884856389204909, "mem_mb": 402.962432, "nll": -5.944856643676758, "elapsed_s": 2715.9}365{"event": "train", "step": 7160, "loss": -6.032393932342529, "grad_norm": 0.7376070022583008, "lr": 0.00013029492460764526, "ex_per_s": 168.34855408958205, "data_frac": 0.00587517915378286, "mem_mb": 402.962432, "nll": -6.032393932342529, "elapsed_s": 2722.2}366{"event": "train", "step": 7180, "loss": -5.947120189666748, "grad_norm": 0.2666208744049072, "lr": 0.00012958877999133094, "ex_per_s": 168.4327565554145, "data_frac": 0.0058658180917892195, "mem_mb": 402.962432, "nll": -5.947120189666748, "elapsed_s": 2728.4}367{"event": "train", "step": 7200, "loss": -5.958819389343262, "grad_norm": 0.3135946989059448, "lr": 0.0001288836742985552, "ex_per_s": 168.51699182143506, "data_frac": 0.005856219183456937, "mem_mb": 402.962432, "nll": -5.958819389343262, "elapsed_s": 2734.6}368{"event": "train", "step": 7220, "loss": -6.062134742736816, "grad_norm": 0.3694804906845093, "lr": 0.00012817962821628953, "ex_per_s": 168.6005249114643, "data_frac": 0.005846586313796748, "mem_mb": 402.962432, "nll": -6.062134742736816, "elapsed_s": 2740.9}369{"event": "train", "step": 7240, "loss": -5.955502986907959, "grad_norm": 0.1534363478422165, "lr": 0.00012747666240041779, "ex_per_s": 168.6841461179564, "data_frac": 0.00583707776632926, "mem_mb": 402.962432, "nll": -5.955502986907959, "elapsed_s": 2747.1}370{"event": "train", "step": 7260, "loss": -6.074743270874023, "grad_norm": 1.0458329916000366, "lr": 0.00012677479747513008, "ex_per_s": 168.76814852659874, "data_frac": 0.005828114514759784, "mem_mb": 402.962432, "nll": -6.074743270874023, "elapsed_s": 2753.3}371{"event": "train", "step": 7280, "loss": -5.864043235778809, "grad_norm": 1.3810899257659912, "lr": 0.00012607405403231763, "ex_per_s": 168.85227321526392, "data_frac": 0.005818676921821266, "mem_mb": 402.962432, "nll": -5.864043235778809, "elapsed_s": 2759.5}372{"event": "train", "step": 7300, "loss": -5.921907901763916, "grad_norm": 1.4686088562011719, "lr": 0.0001253744526309688, "ex_per_s": 168.93552487371028, "data_frac": 0.005809518217812306, "mem_mb": 402.962432, "nll": -5.921907901763916, "elapsed_s": 2765.7}373{"event": "train", "step": 7320, "loss": -6.073697090148926, "grad_norm": 0.09646118432283401, "lr": 0.00012467601379656567, "ex_per_s": 169.01856477778526, "data_frac": 0.005800266887938112, "mem_mb": 402.962432, "nll": -6.073697090148926, "elapsed_s": 2772.0}374{"event": "train", "step": 7340, "loss": -6.017061710357666, "grad_norm": 0.42236918210983276, "lr": 0.000123978758020482, "ex_per_s": 169.10095327498559, "data_frac": 0.005791001619649187, "mem_mb": 402.962432, "nll": -6.017061710357666, "elapsed_s": 2778.2}375{"event": "train", "step": 7360, "loss": -5.992798805236816, "grad_norm": 0.5445763468742371, "lr": 0.00012328270575938212, "ex_per_s": 169.18260913561497, "data_frac": 0.005781711077807518, "mem_mb": 402.962432, "nll": -5.992798805236816, "elapsed_s": 2784.4}376{"event": "train", "step": 7380, "loss": -6.077399730682373, "grad_norm": 0.5555641055107117, "lr": 0.00012258787743462041, "ex_per_s": 169.26383059151533, "data_frac": 0.005772410041225629, "mem_mb": 402.962432, "nll": -6.077399730682373, "elapsed_s": 2790.6}377{"event": "train", "step": 7400, "loss": -5.995854377746582, "grad_norm": 0.6650177836418152, "lr": 0.00012189429343164274, "ex_per_s": 169.3449795019785, "data_frac": 0.005763673895397549, "mem_mb": 402.962432, "nll": -5.995854377746582, "elapsed_s": 2796.8}378{"event": "train", "step": 7420, "loss": -6.050312519073486, "grad_norm": 0.11941340565681458, "lr": 0.00012120197409938763, "ex_per_s": 169.42609855883492, "data_frac": 0.005754522804333396, "mem_mb": 402.962432, "nll": -6.050312519073486, "elapsed_s": 2803.1}379{"event": "train", "step": 7440, "loss": -5.775550365447998, "grad_norm": 0.8809807896614075, "lr": 0.00012051093974969008, "ex_per_s": 169.50709781153134, "data_frac": 0.0057454355968442235, "mem_mb": 402.962432, "nll": -5.775550365447998, "elapsed_s": 2809.3}380{"event": "train", "step": 7460, "loss": -6.210602283477783, "grad_norm": 0.6610187888145447, "lr": 0.00011982121065668482, "ex_per_s": 169.58783227756987, "data_frac": 0.005736335986258545, "mem_mb": 402.962432, "nll": -6.210602283477783, "elapsed_s": 2815.5}381{"event": "train", "step": 7480, "loss": -5.965475559234619, "grad_norm": 0.19537125527858734, "lr": 0.00011913280705621224, "ex_per_s": 169.6682913785633, "data_frac": 0.005727294768448791, "mem_mb": 402.962432, "nll": -5.965475559234619, "elapsed_s": 2821.7}382{"event": "train", "step": 7500, "loss": -5.77569055557251, "grad_norm": 0.23681271076202393, "lr": 0.00011844574914522403, "ex_per_s": 169.7483231559356, "data_frac": 0.005718158895266531, "mem_mb": 402.962432, "nll": -5.77569055557251, "elapsed_s": 2827.9}383{"event": "train", "step": 7520, "loss": -6.058894634246826, "grad_norm": 1.2348499298095703, "lr": 0.00011776005708119105, "ex_per_s": 169.8272981174431, "data_frac": 0.005709295671443629, "mem_mb": 402.962432, "nll": -6.058894634246826, "elapsed_s": 2834.1}384{"event": "train", "step": 7540, "loss": -5.839022636413574, "grad_norm": 1.1065566539764404, "lr": 0.00011707575098151199, "ex_per_s": 169.90592940281178, "data_frac": 0.005700450878176298, "mem_mb": 402.962432, "nll": -5.839022636413574, "elapsed_s": 2840.4}385{"event": "train", "step": 7560, "loss": -5.901429653167725, "grad_norm": 0.37142717838287354, "lr": 0.00011639285092292267, "ex_per_s": 169.98463815838744, "data_frac": 0.005691796395491098, "mem_mb": 402.962432, "nll": -5.901429653167725, "elapsed_s": 2846.6}386{"event": "train", "step": 7580, "loss": -5.87166690826416, "grad_norm": 0.5758063197135925, "lr": 0.00011571137694090757, "ex_per_s": 170.0635430197688, "data_frac": 0.005683161532534106, "mem_mb": 402.962432, "nll": -5.87166690826416, "elapsed_s": 2852.8}387{"event": "train", "step": 7600, "loss": -6.058672904968262, "grad_norm": 0.3588241934776306, "lr": 0.00011503134902911152, "ex_per_s": 170.14249582913737, "data_frac": 0.005674332075340728, "mem_mb": 402.962432, "nll": -6.058672904968262, "elapsed_s": 2859.0}388{"event": "train", "step": 7620, "loss": -5.948451995849609, "grad_norm": 0.2862606644630432, "lr": 0.00011435278713875375, "ex_per_s": 170.22084650735903, "data_frac": 0.00566574428388539, "mem_mb": 402.962432, "nll": -5.948451995849609, "elapsed_s": 2865.2}389{"event": "train", "step": 7640, "loss": -5.912473201751709, "grad_norm": 0.8391689658164978, "lr": 0.00011367571117804159, "ex_per_s": 170.29897269304774, "data_frac": 0.005657283670656039, "mem_mb": 402.962432, "nll": -5.912473201751709, "elapsed_s": 2871.4}390{"event": "train", "step": 7660, "loss": -6.051687240600586, "grad_norm": 0.5329217910766602, "lr": 0.00011300014101158744, "ex_per_s": 170.37637532475, "data_frac": 0.005648663503927868, "mem_mb": 402.962432, "nll": -6.051687240600586, "elapsed_s": 2877.6}391{"event": "train", "step": 7680, "loss": -5.961897373199463, "grad_norm": 0.21684826910495758, "lr": 0.00011232609645982515, "ex_per_s": 170.45357613699045, "data_frac": 0.005640130032383704, "mem_mb": 402.962432, "nll": -5.961897373199463, "elapsed_s": 2883.8}392{"event": "train", "step": 7700, "loss": -5.911105155944824, "grad_norm": 0.47113513946533203, "lr": 0.00011165359729842902, "ex_per_s": 170.53062673740945, "data_frac": 0.005631594817068838, "mem_mb": 402.962432, "nll": -5.911105155944824, "elapsed_s": 2890.0}393{"event": "train", "step": 7720, "loss": -6.0396037101745605, "grad_norm": 0.33693069219589233, "lr": 0.00011098266325773331, "ex_per_s": 170.60684146909605, "data_frac": 0.0056231266681886615, "mem_mb": 402.962432, "nll": -6.0396037101745605, "elapsed_s": 2896.2}394{"event": "train", "step": 7740, "loss": -6.050535678863525, "grad_norm": 0.6541227102279663, "lr": 0.00011031331402215353, "ex_per_s": 170.68238195193277, "data_frac": 0.005614766127852376, "mem_mb": 402.962432, "nll": -6.050535678863525, "elapsed_s": 2902.4}395{"event": "train", "step": 7760, "loss": -5.963062763214111, "grad_norm": 1.1617698669433594, "lr": 0.00010964556922960883, "ex_per_s": 170.75852435153163, "data_frac": 0.0056065134253030805, "mem_mb": 402.962432, "nll": -5.963062763214111, "elapsed_s": 2908.6}396{"event": "train", "step": 7780, "loss": -6.049520015716553, "grad_norm": 0.07860706746578217, "lr": 0.00010897944847094595, "ex_per_s": 170.83321008990785, "data_frac": 0.005598140935014811, "mem_mb": 402.962432, "nll": -6.049520015716553, "elapsed_s": 2914.8}397{"event": "train", "step": 7800, "loss": -6.059013366699219, "grad_norm": 0.4568462669849396, "lr": 0.00010831497128936428, "ex_per_s": 170.90810237449062, "data_frac": 0.0055902191154854316, "mem_mb": 402.962432, "nll": -6.059013366699219, "elapsed_s": 2921.1}398{"event": "train", "step": 7820, "loss": -5.903046131134033, "grad_norm": 0.32464057207107544, "lr": 0.00010765215717984271, "ex_per_s": 170.98250489554687, "data_frac": 0.005582182849182597, "mem_mb": 402.962432, "nll": -5.903046131134033, "elapsed_s": 2927.3}399{"event": "train", "step": 7840, "loss": -5.959072589874268, "grad_norm": 0.11727619916200638, "lr": 0.00010699102558856754, "ex_per_s": 171.05569852493372, "data_frac": 0.0055741268904700905, "mem_mb": 402.962432, "nll": -5.959072589874268, "elapsed_s": 2933.5}400{"event": "train", "step": 7860, "loss": -6.115908622741699, "grad_norm": 0.43551990389823914, "lr": 0.0001063315959123618, "ex_per_s": 171.12859005677646, "data_frac": 0.005566132923910586, "mem_mb": 402.962432, "nll": -6.115908622741699, "elapsed_s": 2939.7}401{"event": "train", "step": 7880, "loss": -6.064240455627441, "grad_norm": 0.6292474865913391, "lr": 0.0001056738874981166, "ex_per_s": 171.20020151992256, "data_frac": 0.0055583846772133605, "mem_mb": 402.962432, "nll": -6.064240455627441, "elapsed_s": 2946.0}402{"event": "train", "step": 7900, "loss": -6.05820894241333, "grad_norm": 0.7682384848594666, "lr": 0.00010501791964222304, "ex_per_s": 171.2714976281273, "data_frac": 0.00555087420923911, "mem_mb": 402.962432, "nll": -6.05820894241333, "elapsed_s": 2952.2}403{"event": "train", "step": 7920, "loss": -6.085340976715088, "grad_norm": 0.14386491477489471, "lr": 0.00010436371159000636, "ex_per_s": 171.34259137532408, "data_frac": 0.005543125606511474, "mem_mb": 402.962432, "nll": -6.085340976715088, "elapsed_s": 2958.5}404{"event": "train", "step": 7940, "loss": -6.048066139221191, "grad_norm": 0.12522479891777039, "lr": 0.00010371128253516118, "ex_per_s": 171.4143064855591, "data_frac": 0.00553524114600819, "mem_mb": 402.962432, "nll": -6.048066139221191, "elapsed_s": 2964.7}405{"event": "train", "step": 7960, "loss": -5.992875576019287, "grad_norm": 0.15464483201503754, "lr": 0.00010306065161918857, "ex_per_s": 171.4855517632896, "data_frac": 0.0055276107042259515, "mem_mb": 402.962432, "nll": -5.992875576019287, "elapsed_s": 2970.9}406{"event": "train", "step": 7980, "loss": -5.840022563934326, "grad_norm": 0.4414929449558258, "lr": 0.00010241183793083414, "ex_per_s": 171.55700549521865, "data_frac": 0.005519943945643031, "mem_mb": 402.962432, "nll": -5.840022563934326, "elapsed_s": 2977.2}407{"event": "train", "step": 8000, "loss": -6.049948215484619, "grad_norm": 0.08454224467277527, "lr": 0.00010176486050552833, "ex_per_s": 171.6275670044648, "data_frac": 0.005512465761353355, "mem_mb": 402.962432, "nll": -6.049948215484619, "elapsed_s": 2983.4}408{"event": "val", "step": 8000, "val_nll": -6.451578834152486, "best": -6.451578834152486, "bad": 0, "mem_mb": 402.962432, "elapsed_s": 3054.3}409{"event": "train", "step": 8020, "loss": -6.041386127471924, "grad_norm": 0.5554257035255432, "lr": 0.00010111973832482765, "ex_per_s": 167.7145336809419, "data_frac": 0.005377268187545515, "mem_mb": 402.962432, "nll": -6.041386127471924, "elapsed_s": 3060.6}410{"event": "train", "step": 8040, "loss": -6.033954620361328, "grad_norm": 0.3682277798652649, "lr": 0.00010047649031585807, "ex_per_s": 167.78935157383324, "data_frac": 0.005370381168821069, "mem_mb": 402.962432, "nll": -6.033954620361328, "elapsed_s": 3066.9}411{"event": "train", "step": 8060, "loss": -5.8316121101379395, "grad_norm": 1.0635325908660889, "lr": 9.983513535075943e-05, "ex_per_s": 167.8633860444277, "data_frac": 0.005363222202285278, "mem_mb": 402.962432, "nll": -5.8316121101379395, "elapsed_s": 3073.2}412{"event": "train", "step": 8080, "loss": -6.000182151794434, "grad_norm": 0.2083154171705246, "lr": 9.919569224613192e-05, "ex_per_s": 167.93857751749903, "data_frac": 0.0053563170083859276, "mem_mb": 402.962432, "nll": -6.000182151794434, "elapsed_s": 3079.4}413{"event": "train", "step": 8100, "loss": -5.9430365562438965, "grad_norm": 0.28857824206352234, "lr": 9.855817976248399e-05, "ex_per_s": 168.01457002150562, "data_frac": 0.005349392596096413, "mem_mb": 402.962432, "nll": -5.9430365562438965, "elapsed_s": 3085.6}414{"event": "train", "step": 8120, "loss": -5.899012565612793, "grad_norm": 0.9594932198524475, "lr": 9.792261660368207e-05, "ex_per_s": 168.09044183922174, "data_frac": 0.005342720819067962, "mem_mb": 402.962432, "nll": -5.899012565612793, "elapsed_s": 3091.9}415{"event": "train", "step": 8140, "loss": -5.894298076629639, "grad_norm": 0.39783674478530884, "lr": 9.72890214164015e-05, "ex_per_s": 168.16610982313603, "data_frac": 0.005335812990105307, "mem_mb": 402.962432, "nll": -5.894298076629639, "elapsed_s": 3098.1}416{"event": "train", "step": 8160, "loss": -6.043814182281494, "grad_norm": 0.15043452382087708, "lr": 9.66574127895797e-05, "ex_per_s": 168.2415981164981, "data_frac": 0.005328925623721548, "mem_mb": 402.962432, "nll": -6.043814182281494, "elapsed_s": 3104.3}417{"event": "train", "step": 8180, "loss": -5.94650936126709, "grad_norm": 0.7168315649032593, "lr": 9.602780925387089e-05, "ex_per_s": 168.31528993202645, "data_frac": 0.005322232321617455, "mem_mb": 402.962432, "nll": -5.94650936126709, "elapsed_s": 3110.5}418{"event": "train", "step": 8200, "loss": -6.169884204864502, "grad_norm": 0.14468476176261902, "lr": 9.540022928110205e-05, "ex_per_s": 168.38857977880286, "data_frac": 0.005315474047747818, "mem_mb": 402.962432, "nll": -6.169884204864502, "elapsed_s": 3116.8}419{"event": "train", "step": 8220, "loss": -5.903765678405762, "grad_norm": 0.2792345881462097, "lr": 9.477469128373143e-05, "ex_per_s": 168.4612456530618, "data_frac": 0.0053086550548532, "mem_mb": 402.962432, "nll": -5.903765678405762, "elapsed_s": 3123.0}420{"event": "train", "step": 8240, "loss": -5.839463233947754, "grad_norm": 0.3618135452270508, "lr": 9.415121361430792e-05, "ex_per_s": 168.53453231018656, "data_frac": 0.005301931301644983, "mem_mb": 402.962432, "nll": -5.839463233947754, "elapsed_s": 3129.3}421{"event": "train", "step": 8260, "loss": -6.1885457038879395, "grad_norm": 0.4989303946495056, "lr": 9.352981456493298e-05, "ex_per_s": 168.60654781340602, "data_frac": 0.005295342843491599, "mem_mb": 402.962432, "nll": -6.1885457038879395, "elapsed_s": 3135.5}422{"event": "train", "step": 8280, "loss": -6.153131008148193, "grad_norm": 0.13894334435462952, "lr": 9.291051236672361e-05, "ex_per_s": 168.6776066356762, "data_frac": 0.005288867381514196, "mem_mb": 402.962432, "nll": -6.153131008148193, "elapsed_s": 3141.8}423{"event": "train", "step": 8300, "loss": -6.044859886169434, "grad_norm": 0.4222394824028015, "lr": 9.229332518927795e-05, "ex_per_s": 168.74956147549236, "data_frac": 0.005282405380915063, "mem_mb": 402.962432, "nll": -6.044859886169434, "elapsed_s": 3148.0}424{"event": "train", "step": 8320, "loss": -5.892317771911621, "grad_norm": 0.23210157454013824, "lr": 9.167827114014155e-05, "ex_per_s": 168.82128664597082, "data_frac": 0.005275710065780618, "mem_mb": 402.962432, "nll": -5.892317771911621, "elapsed_s": 3154.3}425{"event": "train", "step": 8340, "loss": -6.060439109802246, "grad_norm": 0.9909804463386536, "lr": 9.106536826427686e-05, "ex_per_s": 168.89275231154636, "data_frac": 0.005269057949723949, "mem_mb": 402.962432, "nll": -6.060439109802246, "elapsed_s": 3160.5}426{"event": "train", "step": 8360, "loss": -6.0111775398254395, "grad_norm": 0.2852710485458374, "lr": 9.045463454353313e-05, "ex_per_s": 168.9644071652211, "data_frac": 0.005262408941816324, "mem_mb": 402.962432, "nll": -6.0111775398254395, "elapsed_s": 3166.8}427{"event": "train", "step": 8380, "loss": -6.051549911499023, "grad_norm": 0.16504880785942078, "lr": 8.984608789611947e-05, "ex_per_s": 169.03479379585227, "data_frac": 0.0052558845637516215, "mem_mb": 402.962432, "nll": -6.051549911499023, "elapsed_s": 3173.0}428{"event": "train", "step": 8400, "loss": -6.03855037689209, "grad_norm": 0.19673381745815277, "lr": 8.92397461760784e-05, "ex_per_s": 169.10469703552852, "data_frac": 0.005249479865014122, "mem_mb": 402.962432, "nll": -6.03855037689209, "elapsed_s": 3179.3}429{"event": "train", "step": 8420, "loss": -6.001935005187988, "grad_norm": 0.4497053623199463, "lr": 8.863562717276283e-05, "ex_per_s": 169.174303323134, "data_frac": 0.005243032350898503, "mem_mb": 402.962432, "nll": -6.001935005187988, "elapsed_s": 3185.5}430{"event": "train", "step": 8440, "loss": -6.029311180114746, "grad_norm": 0.48249101638793945, "lr": 8.803374861031363e-05, "ex_per_s": 169.24460441386634, "data_frac": 0.005236619821250333, "mem_mb": 402.962432, "nll": -6.029311180114746, "elapsed_s": 3191.8}431{"event": "train", "step": 8460, "loss": -5.9100236892700195, "grad_norm": 0.17144620418548584, "lr": 8.743412814713985e-05, "ex_per_s": 169.3143525755685, "data_frac": 0.005230271054765502, "mem_mb": 402.962432, "nll": -5.9100236892700195, "elapsed_s": 3198.0}432{"event": "train", "step": 8480, "loss": -5.879218578338623, "grad_norm": 0.539370596408844, "lr": 8.683678337540044e-05, "ex_per_s": 169.38399501348474, "data_frac": 0.005224027614823115, "mem_mb": 402.962432, "nll": -5.879218578338623, "elapsed_s": 3204.3}433{"event": "train", "step": 8500, "loss": -5.966991901397705, "grad_norm": 1.1326541900634766, "lr": 8.62417318204883e-05, "ex_per_s": 169.45376930493913, "data_frac": 0.005217735520761659, "mem_mb": 402.962432, "nll": -5.966991901397705, "elapsed_s": 3210.5}434{"event": "train", "step": 8520, "loss": -5.933323383331299, "grad_norm": 0.6090326309204102, "lr": 8.564899094051614e-05, "ex_per_s": 169.52293989760105, "data_frac": 0.005211542740209451, "mem_mb": 402.962432, "nll": -5.933323383331299, "elapsed_s": 3216.7}435{"event": "train", "step": 8540, "loss": -5.906728744506836, "grad_norm": 0.6669726371765137, "lr": 8.505857812580402e-05, "ex_per_s": 169.59225183630411, "data_frac": 0.005205140228153734, "mem_mb": 402.962432, "nll": -5.906728744506836, "elapsed_s": 3223.0}436{"event": "train", "step": 8560, "loss": -5.967252731323242, "grad_norm": 0.3507549464702606, "lr": 8.447051069836941e-05, "ex_per_s": 169.6609175852049, "data_frac": 0.005198870077247928, "mem_mb": 402.962432, "nll": -5.967252731323242, "elapsed_s": 3229.2}437{"event": "train", "step": 8580, "loss": -5.9598212242126465, "grad_norm": 0.47758615016937256, "lr": 8.38848059114188e-05, "ex_per_s": 169.72808815258819, "data_frac": 0.005192900333304698, "mem_mb": 402.962432, "nll": -5.9598212242126465, "elapsed_s": 3235.5}438{"event": "train", "step": 8600, "loss": -6.019575119018555, "grad_norm": 0.20863677561283112, "lr": 8.330148094884171e-05, "ex_per_s": 169.7972145685344, "data_frac": 0.005186720879118573, "mem_mb": 402.962432, "nll": -6.019575119018555, "elapsed_s": 3241.7}439{"event": "train", "step": 8620, "loss": -5.9936604499816895, "grad_norm": 0.5983206033706665, "lr": 8.27205529247063e-05, "ex_per_s": 169.8648856366211, "data_frac": 0.0051805087501273425, "mem_mb": 402.962432, "nll": -5.9936604499816895, "elapsed_s": 3247.9}440{"event": "train", "step": 8640, "loss": -5.966482162475586, "grad_norm": 0.8744270205497742, "lr": 8.214203888275743e-05, "ex_per_s": 169.93272324202238, "data_frac": 0.005174233867309572, "mem_mb": 402.962432, "nll": -5.966482162475586, "elapsed_s": 3254.2}441{"event": "train", "step": 8660, "loss": -5.950627326965332, "grad_norm": 0.42221418023109436, "lr": 8.156595579591652e-05, "ex_per_s": 170.00038382634457, "data_frac": 0.005167876606443823, "mem_mb": 402.962432, "nll": -5.950627326965332, "elapsed_s": 3260.4}442{"event": "train", "step": 8680, "loss": -5.902470111846924, "grad_norm": 0.18559107184410095, "lr": 8.09923205657837e-05, "ex_per_s": 170.067485876809, "data_frac": 0.005161741571107382, "mem_mb": 402.962432, "nll": -5.902470111846924, "elapsed_s": 3266.7}443{"event": "train", "step": 8700, "loss": -5.931341171264648, "grad_norm": 0.1590994894504547, "lr": 8.042115002214181e-05, "ex_per_s": 170.13409942325887, "data_frac": 0.0051556905600787075, "mem_mb": 402.962432, "nll": -5.931341171264648, "elapsed_s": 3272.9}444{"event": "train", "step": 8720, "loss": -6.164746284484863, "grad_norm": 0.4628412425518036, "lr": 7.985246092246267e-05, "ex_per_s": 170.2010626953996, "data_frac": 0.005149550554807322, "mem_mb": 402.962432, "nll": -6.164746284484863, "elapsed_s": 3279.1}445{"event": "train", "step": 8740, "loss": -5.898258209228516, "grad_norm": 0.17290282249450684, "lr": 7.928626995141546e-05, "ex_per_s": 170.26781040015996, "data_frac": 0.005143445172442289, "mem_mb": 402.962432, "nll": -5.898258209228516, "elapsed_s": 3285.4}446{"event": "train", "step": 8760, "loss": -6.0724921226501465, "grad_norm": 0.5969040393829346, "lr": 7.872259372037726e-05, "ex_per_s": 170.33383273804316, "data_frac": 0.005137479404682678, "mem_mb": 402.962432, "nll": -6.0724921226501465, "elapsed_s": 3291.6}447{"event": "train", "step": 8780, "loss": -5.981385707855225, "grad_norm": 0.11361300945281982, "lr": 7.816144876694563e-05, "ex_per_s": 170.3996401996028, "data_frac": 0.005131558024141794, "mem_mb": 402.962432, "nll": -5.981385707855225, "elapsed_s": 3297.8}448{"event": "train", "step": 8800, "loss": -5.9542765617370605, "grad_norm": 0.1451764702796936, "lr": 7.760285155445326e-05, "ex_per_s": 170.46549553095156, "data_frac": 0.005125361235227835, "mem_mb": 402.962432, "nll": -5.9542765617370605, "elapsed_s": 3304.1}449{"event": "train", "step": 8820, "loss": -5.985783576965332, "grad_norm": 0.08317501097917557, "lr": 7.70468184714854e-05, "ex_per_s": 170.53129383511296, "data_frac": 0.0051191462762462244, "mem_mb": 402.962432, "nll": -5.985783576965332, "elapsed_s": 3310.3}450{"event": "train", "step": 8840, "loss": -5.959237575531006, "grad_norm": 0.12760227918624878, "lr": 7.649336583139853e-05, "ex_per_s": 170.59647457852466, "data_frac": 0.005113072187796582, "mem_mb": 402.962432, "nll": -5.959237575531006, "elapsed_s": 3316.6}451{"event": "train", "step": 8860, "loss": -5.990834712982178, "grad_norm": 0.14821936190128326, "lr": 7.594250987184208e-05, "ex_per_s": 170.66172704936292, "data_frac": 0.00510704329529317, "mem_mb": 402.962432, "nll": -5.990834712982178, "elapsed_s": 3322.8}452{"event": "train", "step": 8880, "loss": -5.971665382385254, "grad_norm": 0.5600646138191223, "lr": 7.53942667542819e-05, "ex_per_s": 170.7268738783102, "data_frac": 0.005100960903943636, "mem_mb": 402.962432, "nll": -5.971665382385254, "elapsed_s": 3329.0}453{"event": "train", "step": 8900, "loss": -5.946408748626709, "grad_norm": 0.5900327563285828, "lr": 7.484865256352612e-05, "ex_per_s": 170.79193422803453, "data_frac": 0.005095052395717595, "mem_mb": 402.962432, "nll": -5.946408748626709, "elapsed_s": 3335.2}454{"event": "train", "step": 8920, "loss": -6.099116802215576, "grad_norm": 0.5251909494400024, "lr": 7.430568330725319e-05, "ex_per_s": 170.85632739476887, "data_frac": 0.005089340137641455, "mem_mb": 402.962432, "nll": -6.099116802215576, "elapsed_s": 3341.5}455{"event": "train", "step": 8940, "loss": -6.1055145263671875, "grad_norm": 0.14071038365364075, "lr": 7.376537491554246e-05, "ex_per_s": 170.92029683116033, "data_frac": 0.005083543479145903, "mem_mb": 402.962432, "nll": -6.1055145263671875, "elapsed_s": 3347.7}456{"event": "train", "step": 8960, "loss": -6.059932708740234, "grad_norm": 0.3594752550125122, "lr": 7.322774324040643e-05, "ex_per_s": 170.98441570048416, "data_frac": 0.0050777912649267315, "mem_mb": 402.962432, "nll": -6.059932708740234, "elapsed_s": 3353.9}457{"event": "train", "step": 8980, "loss": -5.913944244384766, "grad_norm": 1.402725100517273, "lr": 7.26928040553261e-05, "ex_per_s": 171.04853277594512, "data_frac": 0.005071771983462447, "mem_mb": 402.962432, "nll": -5.913944244384766, "elapsed_s": 3360.2}458{"event": "train", "step": 9000, "loss": -6.093773365020752, "grad_norm": 0.19493430852890015, "lr": 7.216057305478784e-05, "ex_per_s": 171.11219094671955, "data_frac": 0.005065971465173078, "mem_mb": 402.962432, "nll": -6.093773365020752, "elapsed_s": 3366.4}459{"event": "val", "step": 9000, "val_nll": -6.451599169007232, "best": -6.451599169007232, "bad": 0, "mem_mb": 402.962432, "elapsed_s": 3436.2}460{"event": "train", "step": 9020, "loss": -6.004270553588867, "grad_norm": 0.20608972012996674, "lr": 7.163106585382312e-05, "ex_per_s": 167.6975449573921, "data_frac": 0.004957627033783389, "mem_mb": 402.962432, "nll": -6.004270553588867, "elapsed_s": 3442.6}461{"event": "train", "step": 9040, "loss": -6.105991363525391, "grad_norm": 0.12940235435962677, "lr": 7.11042979875503e-05, "ex_per_s": 167.76458353381472, "data_frac": 0.004952263835514603, "mem_mb": 402.962432, "nll": -6.105991363525391, "elapsed_s": 3448.8}462{"event": "train", "step": 9060, "loss": -6.0831379890441895, "grad_norm": 0.0770779699087143, "lr": 7.058028491071894e-05, "ex_per_s": 167.83141904418514, "data_frac": 0.004947073180689778, "mem_mb": 402.962432, "nll": -6.0831379890441895, "elapsed_s": 3455.1}463{"event": "train", "step": 9080, "loss": -6.093325614929199, "grad_norm": 0.41548293828964233, "lr": 7.005904199725619e-05, "ex_per_s": 167.89880029354015, "data_frac": 0.004941847785291688, "mem_mb": 402.962432, "nll": -6.093325614929199, "elapsed_s": 3461.3}464{"event": "train", "step": 9100, "loss": -6.168005466461182, "grad_norm": 0.6455527544021606, "lr": 6.954058453981609e-05, "ex_per_s": 167.96557510532492, "data_frac": 0.0049363922347009336, "mem_mb": 402.962432, "nll": -6.168005466461182, "elapsed_s": 3467.6}465{"event": "train", "step": 9120, "loss": -5.972580432891846, "grad_norm": 0.25296080112457275, "lr": 6.902492774933048e-05, "ex_per_s": 168.0319807284051, "data_frac": 0.004935574652306731, "mem_mb": 402.962432, "nll": -5.972580432891846, "elapsed_s": 3473.8}466{"event": "train", "step": 9140, "loss": -6.125594615936279, "grad_norm": 0.2673346698284149, "lr": 6.851208675456295e-05, "ex_per_s": 168.0983695763788, "data_frac": 0.0049292695212756225, "mem_mb": 402.962432, "nll": -6.125594615936279, "elapsed_s": 3480.1}467{"event": "train", "step": 9160, "loss": -5.9819254875183105, "grad_norm": 0.5124772191047668, "lr": 6.800207660166508e-05, "ex_per_s": 168.16439306275942, "data_frac": 0.004933511067346551, "mem_mb": 402.962432, "nll": -5.9819254875183105, "elapsed_s": 3486.3}468{"event": "train", "step": 9180, "loss": -5.956428050994873, "grad_norm": 0.3169077932834625, "lr": 6.749491225373478e-05, "ex_per_s": 168.23059541647888, "data_frac": 0.004928002201336425, "mem_mb": 402.962432, "nll": -5.956428050994873, "elapsed_s": 3492.5}469{"event": "train", "step": 9200, "loss": -6.1427435874938965, "grad_norm": 0.07727958261966705, "lr": 6.699060859037737e-05, "ex_per_s": 168.29539195134603, "data_frac": 0.004923849121885375, "mem_mb": 402.962432, "nll": -6.1427435874938965, "elapsed_s": 3498.8}470{"event": "train", "step": 9220, "loss": -5.9665021896362305, "grad_norm": 0.35729289054870605, "lr": 6.648918040726909e-05, "ex_per_s": 168.3603951171964, "data_frac": 0.0049185489580479925, "mem_mb": 402.962432, "nll": -5.9665021896362305, "elapsed_s": 3505.1}471{"event": "train", "step": 9240, "loss": -6.115077495574951, "grad_norm": 1.100300669670105, "lr": 6.599064241572299e-05, "ex_per_s": 168.42641318811027, "data_frac": 0.00491336018594193, "mem_mb": 402.962432, "nll": -6.115077495574951, "elapsed_s": 3511.3}472{"event": "train", "step": 9260, "loss": -6.0255231857299805, "grad_norm": 0.4490249454975128, "lr": 6.549500924225728e-05, "ex_per_s": 168.49179542949094, "data_frac": 0.004908205454183394, "mem_mb": 402.962432, "nll": -6.0255231857299805, "elapsed_s": 3517.5}473{"event": "train", "step": 9280, "loss": -5.818054676055908, "grad_norm": 0.38110947608947754, "lr": 6.500229542816624e-05, "ex_per_s": 168.55719434682558, "data_frac": 0.004903123881805761, "mem_mb": 402.962432, "nll": -5.818054676055908, "elapsed_s": 3523.7}474{"event": "train", "step": 9300, "loss": -6.002753734588623, "grad_norm": 0.453721821308136, "lr": 6.45125154290935e-05, "ex_per_s": 168.62244934169797, "data_frac": 0.004897943552349888, "mem_mb": 402.962432, "nll": -6.002753734588623, "elapsed_s": 3530.0}475{"event": "train", "step": 9320, "loss": -5.82620096206665, "grad_norm": 0.6800503730773926, "lr": 6.402568361460816e-05, "ex_per_s": 168.68732758689575, "data_frac": 0.004892789547252166, "mem_mb": 402.962432, "nll": -5.82620096206665, "elapsed_s": 3536.2}476{"event": "train", "step": 9340, "loss": -6.051785945892334, "grad_norm": 0.3203694522380829, "lr": 6.354181426778293e-05, "ex_per_s": 168.75208409899676, "data_frac": 0.0048876821321934385, "mem_mb": 402.962432, "nll": -6.051785945892334, "elapsed_s": 3542.4}477{"event": "train", "step": 9360, "loss": -6.063374996185303, "grad_norm": 0.5574522018432617, "lr": 6.306092158477523e-05, "ex_per_s": 168.81666156693902, "data_frac": 0.004882463963875923, "mem_mb": 402.962432, "nll": -6.063374996185303, "elapsed_s": 3548.7}478{"event": "train", "step": 9380, "loss": -5.953561305999756, "grad_norm": 0.17294421792030334, "lr": 6.258301967441065e-05, "ex_per_s": 168.88154148442626, "data_frac": 0.004877127161347117, "mem_mb": 402.962432, "nll": -5.953561305999756, "elapsed_s": 3554.9}479{"event": "train", "step": 9400, "loss": -6.072327613830566, "grad_norm": 0.5813221335411072, "lr": 6.210812255776909e-05, "ex_per_s": 168.94535181732104, "data_frac": 0.004872028393616953, "mem_mb": 402.962432, "nll": -6.072327613830566, "elapsed_s": 3561.1}480{"event": "train", "step": 9420, "loss": -5.938012599945068, "grad_norm": 0.22019505500793457, "lr": 6.163624416777327e-05, "ex_per_s": 169.00899149375334, "data_frac": 0.004867035375443161, "mem_mb": 402.962432, "nll": -5.938012599945068, "elapsed_s": 3567.3}481{"event": "train", "step": 9440, "loss": -5.925532341003418, "grad_norm": 0.32375189661979675, "lr": 6.116739834878009e-05, "ex_per_s": 169.0726134274448, "data_frac": 0.004862007001297891, "mem_mb": 402.962432, "nll": -5.925532341003418, "elapsed_s": 3573.6}482{"event": "train", "step": 9460, "loss": -6.002009868621826, "grad_norm": 0.3272174596786499, "lr": 6.0701598856174347e-05, "ex_per_s": 169.13623631488076, "data_frac": 0.004857062188717283, "mem_mb": 402.962432, "nll": -6.002009868621826, "elapsed_s": 3579.8}483{"event": "train", "step": 9480, "loss": -5.927430629730225, "grad_norm": 0.19090132415294647, "lr": 6.023885935596515e-05, "ex_per_s": 169.19849043500378, "data_frac": 0.004852076555950782, "mem_mb": 402.962432, "nll": -5.927430629730225, "elapsed_s": 3586.0}484{"event": "train", "step": 9500, "loss": -5.918243408203125, "grad_norm": 0.08601050823926926, "lr": 5.977919342438512e-05, "ex_per_s": 169.2613758485766, "data_frac": 0.0048471012188872065, "mem_mb": 402.962432, "nll": -5.918243408203125, "elapsed_s": 3592.3}485{"event": "train", "step": 9520, "loss": -5.985201835632324, "grad_norm": 0.10161086171865463, "lr": 5.9322614547491995e-05, "ex_per_s": 169.3239867253664, "data_frac": 0.004842130620052128, "mem_mb": 402.962432, "nll": -5.985201835632324, "elapsed_s": 3598.5}486{"event": "train", "step": 9540, "loss": -6.005059719085693, "grad_norm": 0.14960062503814697, "lr": 5.886913612077285e-05, "ex_per_s": 169.38677080122991, "data_frac": 0.004836989373118055, "mem_mb": 402.962432, "nll": -6.005059719085693, "elapsed_s": 3604.7}487{"event": "train", "step": 9560, "loss": -5.965365409851074, "grad_norm": 0.2301606833934784, "lr": 5.841877144875127e-05, "ex_per_s": 169.44940621520772, "data_frac": 0.004831936186576662, "mem_mb": 402.962432, "nll": -5.965365409851074, "elapsed_s": 3610.9}488{"event": "train", "step": 9580, "loss": -5.911915302276611, "grad_norm": 0.557964563369751, "lr": 5.7971533744597e-05, "ex_per_s": 169.51133060360226, "data_frac": 0.004826916396697318, "mem_mb": 402.962432, "nll": -5.911915302276611, "elapsed_s": 3617.2}489{"event": "train", "step": 9600, "loss": -6.152686595916748, "grad_norm": 1.430990219116211, "lr": 5.752743612973809e-05, "ex_per_s": 169.57278459903253, "data_frac": 0.004821923270775791, "mem_mb": 402.962432, "nll": -6.152686595916748, "elapsed_s": 3623.4}490{"event": "train", "step": 9620, "loss": -6.026523590087891, "grad_norm": 0.8097477555274963, "lr": 5.7086491633476214e-05, "ex_per_s": 169.63441236493935, "data_frac": 0.0048167692496314364, "mem_mb": 402.962432, "nll": -6.026523590087891, "elapsed_s": 3629.6}491{"event": "train", "step": 9640, "loss": -6.002678394317627, "grad_norm": 0.15721718966960907, "lr": 5.6648713192604105e-05, "ex_per_s": 169.69604769394044, "data_frac": 0.004811841661502587, "mem_mb": 402.962432, "nll": -6.002678394317627, "elapsed_s": 3635.9}492{"event": "train", "step": 9660, "loss": -6.050014019012451, "grad_norm": 0.467062771320343, "lr": 5.621411365102632e-05, "ex_per_s": 169.75765268891797, "data_frac": 0.004806890264288509, "mem_mb": 402.962432, "nll": -6.050014019012451, "elapsed_s": 3642.1}493{"event": "train", "step": 9680, "loss": -5.925485610961914, "grad_norm": 0.4714173972606659, "lr": 5.578270575938211e-05, "ex_per_s": 169.81840430724307, "data_frac": 0.004802203748613479, "mem_mb": 402.962432, "nll": -5.925485610961914, "elapsed_s": 3648.3}494{"event": "train", "step": 9700, "loss": -6.194398880004883, "grad_norm": 0.6225925087928772, "lr": 5.5354502174671515e-05, "ex_per_s": 169.87906342232566, "data_frac": 0.004797296113641643, "mem_mb": 402.962432, "nll": -6.194398880004883, "elapsed_s": 3654.6}495{"event": "train", "step": 9720, "loss": -6.0365095138549805, "grad_norm": 0.2612350285053253, "lr": 5.492951545988398e-05, "ex_per_s": 169.9400024606095, "data_frac": 0.004792319712811704, "mem_mb": 402.962432, "nll": -6.0365095138549805, "elapsed_s": 3660.8}496{"event": "train", "step": 9740, "loss": -5.922812461853027, "grad_norm": 0.33104637265205383, "lr": 5.4507758083629804e-05, "ex_per_s": 170.00078082917565, "data_frac": 0.004787195045172955, "mem_mb": 402.962432, "nll": -5.922812461853027, "elapsed_s": 3667.0}497{"event": "train", "step": 9760, "loss": -6.039021968841553, "grad_norm": 1.1227748394012451, "lr": 5.408924241977427e-05, "ex_per_s": 170.06165388258452, "data_frac": 0.0047821224762049615, "mem_mb": 402.962432, "nll": -6.039021968841553, "elapsed_s": 3673.2}498{"event": "train", "step": 9780, "loss": -6.0624799728393555, "grad_norm": 0.15512743592262268, "lr": 5.367398074707458e-05, "ex_per_s": 170.12189950530558, "data_frac": 0.004777306884814662, "mem_mb": 402.962432, "nll": -6.0624799728393555, "elapsed_s": 3679.4}499{"event": "train", "step": 9800, "loss": -5.8508806228637695, "grad_norm": 1.320143461227417, "lr": 5.326198524881979e-05, "ex_per_s": 170.18153993684953, "data_frac": 0.004772453465014706, "mem_mb": 402.962432, "nll": -5.8508806228637695, "elapsed_s": 3685.7}500{"event": "train", "step": 9820, "loss": -6.096225738525391, "grad_norm": 0.8134450316429138, "lr": 5.285326801247307e-05, "ex_per_s": 170.24123674432963, "data_frac": 0.0047676228742436115, "mem_mb": 402.962432, "nll": -6.096225738525391, "elapsed_s": 3691.9}501{"event": "train", "step": 9840, "loss": -5.952772617340088, "grad_norm": 0.754314124584198, "lr": 5.244784102931745e-05, "ex_per_s": 170.30065810096934, "data_frac": 0.004763052822447652, "mem_mb": 402.962432, "nll": -5.952772617340088, "elapsed_s": 3698.1}502{"event": "train", "step": 9860, "loss": -5.883011341094971, "grad_norm": 0.36068496108055115, "lr": 5.2045716194103555e-05, "ex_per_s": 170.36054085777357, "data_frac": 0.004758164271565623, "mem_mb": 402.962432, "nll": -5.883011341094971, "elapsed_s": 3704.3}503{"event": "train", "step": 9880, "loss": -5.8909149169921875, "grad_norm": 0.39724090695381165, "lr": 5.164690530470111e-05, "ex_per_s": 170.42011149101037, "data_frac": 0.0047533657704046765, "mem_mb": 402.962432, "nll": -5.8909149169921875, "elapsed_s": 3710.6}504{"event": "train", "step": 9900, "loss": -5.937314987182617, "grad_norm": 0.4609275758266449, "lr": 5.125142006175238e-05, "ex_per_s": 170.47873686526694, "data_frac": 0.004748696742704713, "mem_mb": 402.962432, "nll": -5.937314987182617, "elapsed_s": 3716.8}505{"event": "train", "step": 9920, "loss": -5.8813066482543945, "grad_norm": 0.08582615107297897, "lr": 5.085927206832921e-05, "ex_per_s": 170.53750077619281, "data_frac": 0.004744021024923876, "mem_mb": 402.962432, "nll": -5.8813066482543945, "elapsed_s": 3723.0}506{"event": "train", "step": 9940, "loss": -5.958974361419678, "grad_norm": 0.34765559434890747, "lr": 5.047047282959241e-05, "ex_per_s": 170.5957882458454, "data_frac": 0.004739311062623027, "mem_mb": 402.962432, "nll": -5.958974361419678, "elapsed_s": 3729.2}507{"event": "train", "step": 9960, "loss": -6.137790203094482, "grad_norm": 0.6326130628585815, "lr": 5.0085033752454274e-05, "ex_per_s": 170.6532110144797, "data_frac": 0.004734689047680732, "mem_mb": 402.962432, "nll": -6.137790203094482, "elapsed_s": 3735.5}508{"event": "train", "step": 9980, "loss": -5.80530309677124, "grad_norm": 1.3279533386230469, "lr": 4.970296614524392e-05, "ex_per_s": 170.71091835413412, "data_frac": 0.004730089288712689, "mem_mb": 402.962432, "nll": -5.80530309677124, "elapsed_s": 3741.7}509{"event": "train", "step": 10000, "loss": -5.939015865325928, "grad_norm": 0.5378219485282898, "lr": 4.9324281217375474e-05, "ex_per_s": 170.76875558371165, "data_frac": 0.004725554267054376, "mem_mb": 402.962432, "nll": -5.939015865325928, "elapsed_s": 3747.9}510{"event": "val", "step": 10000, "val_nll": -6.4520093189281065, "best": -6.4520093189281065, "bad": 0, "mem_mb": 402.962432, "elapsed_s": 3817.2}511{"event": "train", "step": 10020, "loss": -6.075103282928467, "grad_norm": 1.2599050998687744, "lr": 4.8948990079019224e-05, "ex_per_s": 167.72788133540325, "data_frac": 0.004635400048228473, "mem_mb": 402.962432, "nll": -6.075103282928467, "elapsed_s": 3823.5}512{"event": "train", "step": 10040, "loss": -6.080260753631592, "grad_norm": 0.08734667301177979, "lr": 4.857710374077574e-05, "ex_per_s": 167.78741885776802, "data_frac": 0.0046308270113514256, "mem_mb": 402.962432, "nll": -6.080260753631592, "elapsed_s": 3829.8}513{"event": "train", "step": 10060, "loss": -5.756725311279297, "grad_norm": 0.2772590219974518, "lr": 4.820863311335266e-05, "ex_per_s": 167.84672562630746, "data_frac": 0.004626528053470193, "mem_mb": 402.962432, "nll": -5.756725311279297, "elapsed_s": 3836.1}514{"event": "train", "step": 10080, "loss": -6.178429126739502, "grad_norm": 0.26500359177589417, "lr": 4.78435890072448e-05, "ex_per_s": 167.90589184785185, "data_frac": 0.004622204694516356, "mem_mb": 402.962432, "nll": -6.178429126739502, "elapsed_s": 3842.3}515{"event": "train", "step": 10100, "loss": -5.977549076080322, "grad_norm": 1.071378469467163, "lr": 4.748198213241678e-05, "ex_per_s": 167.9649676931983, "data_frac": 0.004617755010209347, "mem_mb": 402.962432, "nll": -5.977549076080322, "elapsed_s": 3848.6}516{"event": "train", "step": 10120, "loss": -6.067094802856445, "grad_norm": 0.5620936155319214, "lr": 4.71238230979889e-05, "ex_per_s": 168.0237359876701, "data_frac": 0.00461328725228907, "mem_mb": 402.962432, "nll": -6.067094802856445, "elapsed_s": 3854.9}517{"event": "train", "step": 10140, "loss": -5.972516059875488, "grad_norm": 0.18475045263767242, "lr": 4.676912241192593e-05, "ex_per_s": 168.08275728229012, "data_frac": 0.004608848281864951, "mem_mb": 402.962432, "nll": -5.972516059875488, "elapsed_s": 3861.1}518{"event": "train", "step": 10160, "loss": -5.773355484008789, "grad_norm": 0.1312653124332428, "lr": 4.641789048072876e-05, "ex_per_s": 168.142670887938, "data_frac": 0.004604539791270631, "mem_mb": 402.962432, "nll": -5.773355484008789, "elapsed_s": 3867.4}519{"event": "train", "step": 10180, "loss": -6.010976314544678, "grad_norm": 0.20595954358577728, "lr": 4.6070137609129025e-05, "ex_per_s": 168.20279371568614, "data_frac": 0.004600196714594914, "mem_mb": 402.962432, "nll": -6.010976314544678, "elapsed_s": 3873.6}520{"event": "train", "step": 10200, "loss": -5.95725679397583, "grad_norm": 0.5351605415344238, "lr": 4.57258739997869e-05, "ex_per_s": 168.26246087141038, "data_frac": 0.00459600629755784, "mem_mb": 402.962432, "nll": -5.95725679397583, "elapsed_s": 3879.8}521{"event": "train", "step": 10220, "loss": -5.782290935516357, "grad_norm": 0.252296507358551, "lr": 4.538510975299176e-05, "ex_per_s": 168.32223061610438, "data_frac": 0.004591829492255898, "mem_mb": 402.962432, "nll": -5.782290935516357, "elapsed_s": 3886.1}522{"event": "train", "step": 10240, "loss": -6.021128177642822, "grad_norm": 0.21625137329101562, "lr": 4.504785486636568e-05, "ex_per_s": 168.381915777079, "data_frac": 0.00458763551135489, "mem_mb": 402.962432, "nll": -6.021128177642822, "elapsed_s": 3892.3}523{"event": "train", "step": 10260, "loss": -6.033493518829346, "grad_norm": 0.23460876941680908, "lr": 4.471411923457034e-05, "ex_per_s": 168.4411357397169, "data_frac": 0.004583517906356263, "mem_mb": 402.962432, "nll": -6.033493518829346, "elapsed_s": 3898.5}524{"event": "train", "step": 10280, "loss": -6.066337585449219, "grad_norm": 0.2597792446613312, "lr": 4.438391264901654e-05, "ex_per_s": 168.5001641910562, "data_frac": 0.004579181890411106, "mem_mb": 402.962432, "nll": -6.066337585449219, "elapsed_s": 3904.8}525{"event": "train", "step": 10300, "loss": -5.991567611694336, "grad_norm": 0.3955135941505432, "lr": 4.405724479757712e-05, "ex_per_s": 168.559052838424, "data_frac": 0.004574940437898904, "mem_mb": 402.962432, "nll": -5.991567611694336, "elapsed_s": 3911.0}526{"event": "train", "step": 10320, "loss": -5.905091762542725, "grad_norm": 0.22609388828277588, "lr": 4.373412526430252e-05, "ex_per_s": 168.61670720912446, "data_frac": 0.004570770710243772, "mem_mb": 402.962432, "nll": -5.905091762542725, "elapsed_s": 3917.2}527{"event": "train", "step": 10340, "loss": -5.947495937347412, "grad_norm": 0.10460443794727325, "lr": 4.341456352913975e-05, "ex_per_s": 168.6748621864437, "data_frac": 0.004566703967548137, "mem_mb": 402.962432, "nll": -5.947495937347412, "elapsed_s": 3923.5}528{"event": "train", "step": 10360, "loss": -5.798453330993652, "grad_norm": 0.4653777480125427, "lr": 4.3098568967654146e-05, "ex_per_s": 168.7329941861367, "data_frac": 0.004562674651764631, "mem_mb": 402.962432, "nll": -5.798453330993652, "elapsed_s": 3929.7}529{"event": "train", "step": 10380, "loss": -5.982715129852295, "grad_norm": 0.5036903023719788, "lr": 4.27861508507545e-05, "ex_per_s": 168.79129284057606, "data_frac": 0.004558643223546631, "mem_mb": 402.962432, "nll": -5.982715129852295, "elapsed_s": 3935.9}530{"event": "train", "step": 10400, "loss": -5.9735541343688965, "grad_norm": 1.2531126737594604, "lr": 4.247731834442082e-05, "ex_per_s": 168.849326546541, "data_frac": 0.004554413446056095, "mem_mb": 402.962432, "nll": -5.9735541343688965, "elapsed_s": 3942.2}531{"event": "train", "step": 10420, "loss": -6.034498691558838, "grad_norm": 0.5183435082435608, "lr": 4.2172080509435485e-05, "ex_per_s": 168.90729850074024, "data_frac": 0.004550117310334998, "mem_mb": 402.962432, "nll": -6.034498691558838, "elapsed_s": 3948.4}532{"event": "train", "step": 10440, "loss": -6.1286797523498535, "grad_norm": 1.0192668437957764, "lr": 4.1870446301117574e-05, "ex_per_s": 168.96498191505867, "data_frac": 0.00454593850422228, "mem_mb": 402.962432, "nll": -6.1286797523498535, "elapsed_s": 3954.6}533{"event": "train", "step": 10460, "loss": -5.97251558303833, "grad_norm": 0.310337632894516, "lr": 4.157242456905989e-05, "ex_per_s": 169.02227238930547, "data_frac": 0.004541696424108991, "mem_mb": 402.962432, "nll": -5.97251558303833, "elapsed_s": 3960.9}534{"event": "train", "step": 10480, "loss": -5.9728522300720215, "grad_norm": 0.5854835510253906, "lr": 4.1278024056869486e-05, "ex_per_s": 169.0797875426726, "data_frac": 0.004537589483377337, "mem_mb": 402.962432, "nll": -5.9728522300720215, "elapsed_s": 3967.1}535{"event": "train", "step": 10500, "loss": -5.994482517242432, "grad_norm": 0.49066510796546936, "lr": 4.0987253401911067e-05, "ex_per_s": 169.13695058923, "data_frac": 0.004533559988900779, "mem_mb": 402.962432, "nll": -5.994482517242432, "elapsed_s": 3973.3}536{"event": "train", "step": 10520, "loss": -6.004948139190674, "grad_norm": 0.7500579953193665, "lr": 4.070012113505359e-05, "ex_per_s": 169.1933027652548, "data_frac": 0.004529718810332438, "mem_mb": 402.962432, "nll": -6.004948139190674, "elapsed_s": 3979.5}537{"event": "train", "step": 10540, "loss": -5.780322551727295, "grad_norm": 0.3492540419101715, "lr": 4.041663568041998e-05, "ex_per_s": 169.25000587852074, "data_frac": 0.00452566711899585, "mem_mb": 402.962432, "nll": -5.780322551727295, "elapsed_s": 3985.8}538{"event": "train", "step": 10560, "loss": -5.8715033531188965, "grad_norm": 0.19713634252548218, "lr": 4.013680535514001e-05, "ex_per_s": 169.3067187826264, "data_frac": 0.004521520894570214, "mem_mb": 402.962432, "nll": -5.8715033531188965, "elapsed_s": 3992.0}539{"event": "train", "step": 10580, "loss": -5.86972713470459, "grad_norm": 0.09528922289609909, "lr": 3.986063836910623e-05, "ex_per_s": 169.36329824225825, "data_frac": 0.004517506093037105, "mem_mb": 402.962432, "nll": -5.86972713470459, "elapsed_s": 3998.2}540{"event": "train", "step": 10600, "loss": -5.89226770401001, "grad_norm": 0.40070974826812744, "lr": 3.958814282473316e-05, "ex_per_s": 169.41975236743872, "data_frac": 0.0045134634406367785, "mem_mb": 402.962432, "nll": -5.89226770401001, "elapsed_s": 4004.4}541{"event": "train", "step": 10620, "loss": -6.041473865509033, "grad_norm": 0.3789278566837311, "lr": 3.9319326716719496e-05, "ex_per_s": 169.47626283301173, "data_frac": 0.0045093882497751, "mem_mb": 402.962432, "nll": -6.041473865509033, "elapsed_s": 4010.7}542{"event": "train", "step": 10640, "loss": -5.817171096801758, "grad_norm": 0.21709351241588593, "lr": 3.905419793181364e-05, "ex_per_s": 169.53279175172253, "data_frac": 0.004505367675325975, "mem_mb": 402.962432, "nll": -5.817171096801758, "elapsed_s": 4016.9}543{"event": "train", "step": 10660, "loss": -5.85053825378418, "grad_norm": 0.3903542757034302, "lr": 3.87927642485822e-05, "ex_per_s": 169.58870742110142, "data_frac": 0.0045012492593879, "mem_mb": 402.962432, "nll": -5.85053825378418, "elapsed_s": 4023.1}544{"event": "train", "step": 10680, "loss": -5.950218200683594, "grad_norm": 0.14481545984745026, "lr": 3.853503333718199e-05, "ex_per_s": 169.64463276755882, "data_frac": 0.004497272015365124, "mem_mb": 402.962432, "nll": -5.950218200683594, "elapsed_s": 4029.3}545{"event": "train", "step": 10700, "loss": -5.803576469421387, "grad_norm": 0.8008978962898254, "lr": 3.8281012759134686e-05, "ex_per_s": 169.69979673308626, "data_frac": 0.004493349630725713, "mem_mb": 402.962432, "nll": -5.803576469421387, "elapsed_s": 4035.6}546{"event": "train", "step": 10720, "loss": -5.880484580993652, "grad_norm": 0.13614435493946075, "lr": 3.803070996710526e-05, "ex_per_s": 169.7545765398474, "data_frac": 0.004489475804217849, "mem_mb": 402.962432, "nll": -5.880484580993652, "elapsed_s": 4041.8}547{"event": "train", "step": 10740, "loss": -6.0990400314331055, "grad_norm": 0.921842098236084, "lr": 3.778413230468319e-05, "ex_per_s": 169.80915916016804, "data_frac": 0.004485552207063057, "mem_mb": 402.962432, "nll": -6.0990400314331055, "elapsed_s": 4048.0}548{"event": "train", "step": 10760, "loss": -5.962587356567383, "grad_norm": 0.2259231060743332, "lr": 3.754128700616701e-05, "ex_per_s": 169.8641923862593, "data_frac": 0.004481625955232518, "mem_mb": 402.962432, "nll": -5.962587356567383, "elapsed_s": 4054.3}549{"event": "train", "step": 10780, "loss": -6.01778507232666, "grad_norm": 0.44043025374412537, "lr": 3.7302181196352115e-05, "ex_per_s": 169.91881350894306, "data_frac": 0.004477785302889641, "mem_mb": 402.962432, "nll": -6.01778507232666, "elapsed_s": 4060.5}550{"event": "train", "step": 10800, "loss": -5.928957462310791, "grad_norm": 0.5930541157722473, "lr": 3.7066821890321684e-05, "ex_per_s": 169.9739101198748, "data_frac": 0.004473780915677937, "mem_mb": 402.962432, "nll": -5.928957462310791, "elapsed_s": 4066.7}551{"event": "train", "step": 10820, "loss": -6.01577091217041, "grad_norm": 0.8187345266342163, "lr": 3.683521599324089e-05, "ex_per_s": 170.02822890784287, "data_frac": 0.0044699080660528905, "mem_mb": 402.962432, "nll": -6.01577091217041, "elapsed_s": 4072.9}552{"event": "train", "step": 10840, "loss": -5.962541103363037, "grad_norm": 0.16645225882530212, "lr": 3.660737030015427e-05, "ex_per_s": 170.08202006536843, "data_frac": 0.004465976641098133, "mem_mb": 402.962432, "nll": -5.962541103363037, "elapsed_s": 4079.2}553{"event": "train", "step": 10860, "loss": -5.841565132141113, "grad_norm": 0.3517197370529175, "lr": 3.638329149578648e-05, "ex_per_s": 170.1355408106245, "data_frac": 0.0044621959060532745, "mem_mb": 402.962432, "nll": -5.841565132141113, "elapsed_s": 4085.4}554{"event": "train", "step": 10880, "loss": -6.000791549682617, "grad_norm": 0.25617340207099915, "lr": 3.6162986154346006e-05, "ex_per_s": 170.18984177495773, "data_frac": 0.004458290004675812, "mem_mb": 402.962432, "nll": -6.000791549682617, "elapsed_s": 4091.6}555{"event": "train", "step": 10900, "loss": -6.122554779052734, "grad_norm": 0.6251921057701111, "lr": 3.594646073933244e-05, "ex_per_s": 170.243252098443, "data_frac": 0.0044544510911778915, "mem_mb": 402.962432, "nll": -6.122554779052734, "elapsed_s": 4097.9}556{"event": "train", "step": 10920, "loss": -5.898369312286377, "grad_norm": 0.765903651714325, "lr": 3.5733721603346695e-05, "ex_per_s": 170.2969161709693, "data_frac": 0.004450587716200504, "mem_mb": 402.962432, "nll": -5.898369312286377, "elapsed_s": 4104.1}557{"event": "train", "step": 10940, "loss": -5.894509792327881, "grad_norm": 0.40478989481925964, "lr": 3.552477498790482e-05, "ex_per_s": 170.3501856182919, "data_frac": 0.0044465913477023155, "mem_mb": 402.962432, "nll": -5.894509792327881, "elapsed_s": 4110.3}558{"event": "train", "step": 10960, "loss": -6.125576019287109, "grad_norm": 0.23683302104473114, "lr": 3.531962702325468e-05, "ex_per_s": 170.40362329731676, "data_frac": 0.00444276434804322, "mem_mb": 402.962432, "nll": -6.125576019287109, "elapsed_s": 4116.5}559{"event": "train", "step": 10980, "loss": -5.8948588371276855, "grad_norm": 0.08881129324436188, "lr": 3.511828372819624e-05, "ex_per_s": 170.45676936453836, "data_frac": 0.004438880736026431, "mem_mb": 402.962432, "nll": -5.8948588371276855, "elapsed_s": 4122.8}560{"event": "train", "step": 11000, "loss": -5.920395851135254, "grad_norm": 0.08673154562711716, "lr": 3.4920751009904904e-05, "ex_per_s": 170.50975965342008, "data_frac": 0.004435045044068853, "mem_mb": 402.962432, "nll": -5.920395851135254, "elapsed_s": 4129.0}561{"event": "val", "step": 11000, "val_nll": -6.452306132872168, "best": -6.452306132872168, "bad": 0, "mem_mb": 402.962432, "elapsed_s": 4198.6}562{"event": "train", "step": 11020, "loss": -5.803661346435547, "grad_norm": 0.1265897899866104, "lr": 3.472703466375828e-05, "ex_per_s": 167.7361593984731, "data_frac": 0.004357892834676093, "mem_mb": 402.962432, "nll": -5.803661346435547, "elapsed_s": 4204.9}563{"event": "train", "step": 11040, "loss": -5.9849724769592285, "grad_norm": 0.48222312331199646, "lr": 3.4537140373166085e-05, "ex_per_s": 167.79086477732136, "data_frac": 0.004354234376798416, "mem_mb": 402.962432, "nll": -5.9849724769592285, "elapsed_s": 4211.1}564{"event": "train", "step": 11060, "loss": -5.982487201690674, "grad_norm": 0.9215158224105835, "lr": 3.4351073709403385e-05, "ex_per_s": 167.84541251463475, "data_frac": 0.004350626528820405, "mem_mb": 402.962432, "nll": -5.982487201690674, "elapsed_s": 4217.4}565{"event": "train", "step": 11080, "loss": -6.029166221618652, "grad_norm": 0.38918977975845337, "lr": 3.4168840131447255e-05, "ex_per_s": 167.9005839441206, "data_frac": 0.004347250501022211, "mem_mb": 402.962432, "nll": -6.029166221618652, "elapsed_s": 4223.6}566{"event": "train", "step": 11100, "loss": -6.1360907554626465, "grad_norm": 1.087996244430542, "lr": 3.399044498581645e-05, "ex_per_s": 167.9553942662204, "data_frac": 0.0043436362422648025, "mem_mb": 402.962432, "nll": -6.1360907554626465, "elapsed_s": 4229.9}567{"event": "train", "step": 11120, "loss": -5.907586574554443, "grad_norm": 0.4136078357696533, "lr": 3.3815893506414745e-05, "ex_per_s": 168.01023955586533, "data_frac": 0.004340092534649512, "mem_mb": 402.962432, "nll": -5.907586574554443, "elapsed_s": 4236.1}568{"event": "train", "step": 11140, "loss": -5.882693767547607, "grad_norm": 0.2868080139160156, "lr": 3.364519081437718e-05, "ex_per_s": 168.06559473047264, "data_frac": 0.004336463431811769, "mem_mb": 402.962432, "nll": -5.882693767547607, "elapsed_s": 4242.3}569{"event": "train", "step": 11160, "loss": -6.077145099639893, "grad_norm": 0.2666875422000885, "lr": 3.347834191791993e-05, "ex_per_s": 168.1211116841624, "data_frac": 0.004332823112002387, "mem_mb": 402.962432, "nll": -6.077145099639893, "elapsed_s": 4248.6}570{"event": "train", "step": 11180, "loss": -6.070017337799072, "grad_norm": 0.1357773244380951, "lr": 3.331535171219332e-05, "ex_per_s": 168.17667589065306, "data_frac": 0.004329087858168498, "mem_mb": 402.962432, "nll": -6.070017337799072, "elapsed_s": 4254.8}571{"event": "train", "step": 11200, "loss": -6.178225040435791, "grad_norm": 0.08946181833744049, "lr": 3.3156224979138305e-05, "ex_per_s": 168.23180319555308, "data_frac": 0.004325419032711069, "mem_mb": 402.962432, "nll": -6.178225040435791, "elapsed_s": 4261.0}572{"event": "train", "step": 11220, "loss": -5.9817280769348145, "grad_norm": 0.20637904107570648, "lr": 3.300096638734598e-05, "ex_per_s": 168.286664501121, "data_frac": 0.004321864907330657, "mem_mb": 402.962432, "nll": -5.9817280769348145, "elapsed_s": 4267.2}573{"event": "train", "step": 11240, "loss": -5.972356796264648, "grad_norm": 0.16738617420196533, "lr": 3.2849580491920845e-05, "ex_per_s": 168.34075272119904, "data_frac": 0.004318470926405481, "mem_mb": 402.962432, "nll": -5.972356796264648, "elapsed_s": 4273.4}574{"event": "train", "step": 11260, "loss": -6.038340091705322, "grad_norm": 0.49446606636047363, "lr": 3.2702071734346924e-05, "ex_per_s": 168.39409950080113, "data_frac": 0.00431495994396903, "mem_mb": 402.962432, "nll": -6.038340091705322, "elapsed_s": 4279.7}575{"event": "train", "step": 11280, "loss": -6.102568626403809, "grad_norm": 0.619586169719696, "lr": 3.255844444235771e-05, "ex_per_s": 168.44769718292557, "data_frac": 0.0043114895163571705, "mem_mb": 402.962432, "nll": -6.102568626403809, "elapsed_s": 4285.9}576{"event": "train", "step": 11300, "loss": -5.936993598937988, "grad_norm": 0.09018643200397491, "lr": 3.2418702829808946e-05, "ex_per_s": 168.50164369995596, "data_frac": 0.0043080490203661805, "mem_mb": 402.962432, "nll": -5.936993598937988, "elapsed_s": 4292.1}577{"event": "train", "step": 11320, "loss": -5.861301422119141, "grad_norm": 0.6009054183959961, "lr": 3.228285099655516e-05, "ex_per_s": 168.5558984053379, "data_frac": 0.004304546423642266, "mem_mb": 402.962432, "nll": -5.861301422119141, "elapsed_s": 4298.3}578{"event": "train", "step": 11340, "loss": -5.925404071807861, "grad_norm": 1.0661998987197876, "lr": 3.215089292832933e-05, "ex_per_s": 168.60981469312546, "data_frac": 0.004301027697661361, "mem_mb": 402.962432, "nll": -5.925404071807861, "elapsed_s": 4304.6}579{"event": "train", "step": 11360, "loss": -5.9605793952941895, "grad_norm": 0.3157835304737091, "lr": 3.202283249662597e-05, "ex_per_s": 168.66354025896695, "data_frac": 0.004297435038700376, "mem_mb": 402.962432, "nll": -5.9605793952941895, "elapsed_s": 4310.8}580{"event": "train", "step": 11380, "loss": -5.910501480102539, "grad_norm": 0.2130451500415802, "lr": 3.189867345858747e-05, "ex_per_s": 168.71730884152942, "data_frac": 0.004293933207900066, "mem_mb": 402.962432, "nll": -5.910501480102539, "elapsed_s": 4317.0}581{"event": "train", "step": 11400, "loss": -6.081836700439453, "grad_norm": 0.2678438425064087, "lr": 3.177841945689397e-05, "ex_per_s": 168.77016110337064, "data_frac": 0.004290499453441901, "mem_mb": 402.962432, "nll": -6.081836700439453, "elapsed_s": 4323.2}582{"event": "train", "step": 11420, "loss": -5.832274436950684, "grad_norm": 0.23890140652656555, "lr": 3.1662074019656414e-05, "ex_per_s": 168.82272333273994, "data_frac": 0.004287099124627312, "mem_mb": 402.962432, "nll": -5.832274436950684, "elapsed_s": 4329.5}583{"event": "train", "step": 11440, "loss": -5.879212856292725, "grad_norm": 1.588288426399231, "lr": 3.154964056031306e-05, "ex_per_s": 168.87476335058008, "data_frac": 0.00428372539764945, "mem_mb": 402.962432, "nll": -5.879212856292725, "elapsed_s": 4335.7}584{"event": "train", "step": 11460, "loss": -6.138811111450195, "grad_norm": 0.9203355312347412, "lr": 3.1441122377529333e-05, "ex_per_s": 168.92680596878168, "data_frac": 0.004280315430159619, "mem_mb": 402.962432, "nll": -6.138811111450195, "elapsed_s": 4342.0}585{"event": "train", "step": 11480, "loss": -5.963490962982178, "grad_norm": 0.2971441149711609, "lr": 3.133652265510106e-05, "ex_per_s": 168.978858988803, "data_frac": 0.004277007087085256, "mem_mb": 402.962432, "nll": -5.963490962982178, "elapsed_s": 4348.2}586{"event": "train", "step": 11500, "loss": -5.94342041015625, "grad_norm": 0.44156673550605774, "lr": 3.123584446186108e-05, "ex_per_s": 169.0310647688476, "data_frac": 0.004273564839318929, "mem_mb": 402.962432, "nll": -5.94342041015625, "elapsed_s": 4354.4}587{"event": "train", "step": 11520, "loss": -5.911505222320557, "grad_norm": 0.4921186864376068, "lr": 3.113909075158913e-05, "ex_per_s": 169.08319308314972, "data_frac": 0.004270185088138561, "mem_mb": 402.962432, "nll": -5.911505222320557, "elapsed_s": 4360.6}588{"event": "train", "step": 11540, "loss": -6.046837329864502, "grad_norm": 0.4137609302997589, "lr": 3.104626436292525e-05, "ex_per_s": 169.13508876185864, "data_frac": 0.004266872263970791, "mem_mb": 402.962432, "nll": -6.046837329864502, "elapsed_s": 4366.9}589{"event": "train", "step": 11560, "loss": -6.117463111877441, "grad_norm": 0.11132170259952545, "lr": 3.095736801928647e-05, "ex_per_s": 169.1868189343993, "data_frac": 0.004263661721263715, "mem_mb": 402.962432, "nll": -6.117463111877441, "elapsed_s": 4373.1}590{"event": "train", "step": 11580, "loss": -5.959076881408691, "grad_norm": 0.08436115086078644, "lr": 3.087240432878696e-05, "ex_per_s": 169.23834305829678, "data_frac": 0.004260356423443369, "mem_mb": 402.962432, "nll": -5.959076881408691, "elapsed_s": 4379.3}591{"event": "train", "step": 11600, "loss": -6.075795650482178, "grad_norm": 0.23954693973064423, "lr": 3.0791375784161455e-05, "ex_per_s": 169.29012703234682, "data_frac": 0.004257014249509953, "mem_mb": 402.962432, "nll": -6.075795650482178, "elapsed_s": 4385.6}592{"event": "train", "step": 11620, "loss": -6.024567604064941, "grad_norm": 0.19943343102931976, "lr": 3.0714284762692116e-05, "ex_per_s": 169.34150089032696, "data_frac": 0.004253629759911706, "mem_mb": 402.962432, "nll": -6.024567604064941, "elapsed_s": 4391.8}593{"event": "train", "step": 11640, "loss": -5.960623264312744, "grad_norm": 0.17396120727062225, "lr": 3.0641133526138836e-05, "ex_per_s": 169.3926556864967, "data_frac": 0.004250324572119709, "mem_mb": 402.962432, "nll": -5.960623264312744, "elapsed_s": 4398.0}594{"event": "train", "step": 11660, "loss": -5.9212446212768555, "grad_norm": 0.4753583073616028, "lr": 3.0571924220672835e-05, "ex_per_s": 169.443593357245, "data_frac": 0.004247059077995052, "mem_mb": 402.962432, "nll": -5.9212446212768555, "elapsed_s": 4404.3}595{"event": "train", "step": 11680, "loss": -5.965421199798584, "grad_norm": 0.7613980770111084, "lr": 3.0506658876813734e-05, "ex_per_s": 169.49432464446556, "data_frac": 0.004243892988676388, "mem_mb": 402.962432, "nll": -5.965421199798584, "elapsed_s": 4410.5}596{"event": "train", "step": 11700, "loss": -6.048730850219727, "grad_norm": 0.10865963250398636, "lr": 3.0445339409369934e-05, "ex_per_s": 169.54532003195246, "data_frac": 0.004240647073168917, "mem_mb": 402.962432, "nll": -6.048730850219727, "elapsed_s": 4416.7}597{"event": "train", "step": 11720, "loss": -5.76594352722168, "grad_norm": 0.4873335659503937, "lr": 3.0387967617382475e-05, "ex_per_s": 169.59628388904994, "data_frac": 0.004237512818674314, "mem_mb": 402.962432, "nll": -5.76594352722168, "elapsed_s": 4422.9}598{"event": "train", "step": 11740, "loss": -6.044073581695557, "grad_norm": 0.7387941479682922, "lr": 3.0334545184072252e-05, "ex_per_s": 169.64706858665568, "data_frac": 0.004234259587682094, "mem_mb": 402.962432, "nll": -6.044073581695557, "elapsed_s": 4429.1}599{"event": "train", "step": 11760, "loss": -5.903907775878906, "grad_norm": 0.22823630273342133, "lr": 3.0285073676790606e-05, "ex_per_s": 169.69762706376332, "data_frac": 0.0042309252885562405, "mem_mb": 402.962432, "nll": -5.903907775878906, "elapsed_s": 4435.4}600{"event": "train", "step": 11780, "loss": -6.083099365234375, "grad_norm": 0.42420437932014465, "lr": 3.0239554546973377e-05, "ex_per_s": 169.7477026675299, "data_frac": 0.004227665593498769, "mem_mb": 402.962432, "nll": -6.083099365234375, "elapsed_s": 4441.6}601{"event": "train", "step": 11800, "loss": -5.96582555770874, "grad_norm": 0.45792558789253235, "lr": 3.0197989130098285e-05, "ex_per_s": 169.79781774747164, "data_frac": 0.004256408653245235, "mem_mb": 402.962432, "nll": -5.96582555770874, "elapsed_s": 4447.8}602{"event": "train", "step": 11820, "loss": -5.970834732055664, "grad_norm": 0.6530676484107971, "lr": 3.0160378645645744e-05, "ex_per_s": 169.84775581477518, "data_frac": 0.004253155167465038, "mem_mb": 402.962432, "nll": -5.970834732055664, "elapsed_s": 4454.1}603{"event": "train", "step": 11840, "loss": -6.031537055969238, "grad_norm": 0.24615062773227692, "lr": 3.0126724197063143e-05, "ex_per_s": 169.89717709428243, "data_frac": 0.0042498610781183666, "mem_mb": 402.962432, "nll": -6.031537055969238, "elapsed_s": 4460.3}604{"event": "train", "step": 11860, "loss": -5.988275527954102, "grad_norm": 0.748943567276001, "lr": 3.0097026771732407e-05, "ex_per_s": 169.94692247678972, "data_frac": 0.0042466606579877894, "mem_mb": 402.962432, "nll": -5.988275527954102, "elapsed_s": 4466.5}605{"event": "train", "step": 11880, "loss": -6.027023792266846, "grad_norm": 1.1444087028503418, "lr": 3.0071287240941065e-05, "ex_per_s": 169.9965128113243, "data_frac": 0.004243475345453158, "mem_mb": 402.962432, "nll": -6.027023792266846, "elapsed_s": 4472.8}606{"event": "train", "step": 11900, "loss": -5.888978958129883, "grad_norm": 0.24101412296295166, "lr": 3.0049506359856666e-05, "ex_per_s": 170.04571093373426, "data_frac": 0.004240248524349701, "mem_mb": 402.962432, "nll": -5.888978958129883, "elapsed_s": 4479.0}607{"event": "train", "step": 11920, "loss": -6.010696887969971, "grad_norm": 0.0788135677576065, "lr": 3.0031684767504643e-05, "ex_per_s": 170.0946929685911, "data_frac": 0.004237107158710484, "mem_mb": 402.962432, "nll": -6.010696887969971, "elapsed_s": 4485.2}608{"event": "train", "step": 11940, "loss": -6.070146083831787, "grad_norm": 1.0397427082061768, "lr": 3.0017822986749544e-05, "ex_per_s": 170.14365827215113, "data_frac": 0.004233891433174438, "mem_mb": 402.962432, "nll": -6.070146083831787, "elapsed_s": 4491.5}609{"event": "train", "step": 11960, "loss": -6.163085460662842, "grad_norm": 0.11632367968559265, "lr": 3.000792142427973e-05, "ex_per_s": 170.1928741001049, "data_frac": 0.0042306651175574635, "mem_mb": 402.962432, "nll": -6.163085460662842, "elapsed_s": 4497.7}610{"event": "train", "step": 11980, "loss": -6.0911054611206055, "grad_norm": 0.3330000936985016, "lr": 3.000198037059536e-05, "ex_per_s": 170.2419882133843, "data_frac": 0.0042273510553315925, "mem_mb": 402.962432, "nll": -6.0911054611206055, "elapsed_s": 4503.9}611{"event": "train", "step": 12000, "loss": -5.938515663146973, "grad_norm": 0.11778673529624939, "lr": 2.9999999999999997e-05, "ex_per_s": 170.29098163518753, "data_frac": 0.004224088055693102, "mem_mb": 402.962432, "nll": -5.938515663146973, "elapsed_s": 4510.1}612{"event": "val", "step": 12000, "val_nll": -6.452572455371413, "best": -6.452572455371413, "bad": 0, "mem_mb": 402.962432, "elapsed_s": 4580.0}613{"event": "done", "steps": 12000, "best_val": -6.452572455371413, "wall_s": 4579.964827299118, "ex_per_s": 167.6868772926587, "peak_mem_mb": 402.962432, "n_params": 20718339, "device": "mps", "data_time_frac": 0.004159492937716627, "elapsed_s": 4580.0}614
Files · 963d827ea83682ea0d9c4f03450180447979764d
Menu

Theme