{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 37963, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00026341524449873725, "grad_norm": 1289.6014404296875, "learning_rate": 1.5803336259877089e-07, "loss": 2.413991165161133, "step": 10 }, { "epoch": 0.0005268304889974745, "grad_norm": 60.15559768676758, "learning_rate": 3.336259877085163e-07, "loss": 0.7373998641967774, "step": 20 }, { "epoch": 0.0007902457334962118, "grad_norm": 31.268722534179688, "learning_rate": 5.092186128182617e-07, "loss": 0.31741483211517335, "step": 30 }, { "epoch": 0.001053660977994949, "grad_norm": 15.827521324157715, "learning_rate": 6.848112379280071e-07, "loss": 0.2593031167984009, "step": 40 }, { "epoch": 0.0013170762224936862, "grad_norm": 20.322460174560547, "learning_rate": 8.604038630377525e-07, "loss": 0.2671781539916992, "step": 50 }, { "epoch": 0.0015804914669924235, "grad_norm": 13.949934005737305, "learning_rate": 1.035996488147498e-06, "loss": 0.25182194709777833, "step": 60 }, { "epoch": 0.0018439067114911607, "grad_norm": 14.03654956817627, "learning_rate": 1.2115891132572432e-06, "loss": 0.22838876247406006, "step": 70 }, { "epoch": 0.002107321955989898, "grad_norm": 9.701815605163574, "learning_rate": 1.3871817383669885e-06, "loss": 0.24621906280517578, "step": 80 }, { "epoch": 0.002370737200488635, "grad_norm": 14.71701717376709, "learning_rate": 1.5627743634767342e-06, "loss": 0.22284703254699706, "step": 90 }, { "epoch": 0.0026341524449873723, "grad_norm": 12.438298225402832, "learning_rate": 1.7383669885864795e-06, "loss": 0.22346246242523193, "step": 100 }, { "epoch": 0.00289756768948611, "grad_norm": 10.914676666259766, "learning_rate": 1.913959613696225e-06, "loss": 0.2267392635345459, "step": 110 }, { "epoch": 0.003160982933984847, "grad_norm": 18.423648834228516, "learning_rate": 2.08955223880597e-06, "loss": 0.25090048313140867, "step": 120 }, { "epoch": 0.003424398178483584, "grad_norm": 13.623218536376953, "learning_rate": 2.265144863915716e-06, "loss": 0.2193898916244507, "step": 130 }, { "epoch": 0.0036878134229823213, "grad_norm": 16.107337951660156, "learning_rate": 2.440737489025461e-06, "loss": 0.22636075019836427, "step": 140 }, { "epoch": 0.0039512286674810585, "grad_norm": 7.983529090881348, "learning_rate": 2.6163301141352064e-06, "loss": 0.22446055412292482, "step": 150 }, { "epoch": 0.004214643911979796, "grad_norm": 15.91808795928955, "learning_rate": 2.791922739244952e-06, "loss": 0.25168163776397706, "step": 160 }, { "epoch": 0.004478059156478534, "grad_norm": 9.312753677368164, "learning_rate": 2.9675153643546974e-06, "loss": 0.24585163593292236, "step": 170 }, { "epoch": 0.00474147440097727, "grad_norm": 6.510652542114258, "learning_rate": 3.1431079894644427e-06, "loss": 0.24500660896301268, "step": 180 }, { "epoch": 0.005004889645476008, "grad_norm": 6.967971324920654, "learning_rate": 3.318700614574188e-06, "loss": 0.22339446544647218, "step": 190 }, { "epoch": 0.005268304889974745, "grad_norm": 12.927844047546387, "learning_rate": 3.4942932396839337e-06, "loss": 0.2363447666168213, "step": 200 }, { "epoch": 0.005531720134473482, "grad_norm": 17.01195526123047, "learning_rate": 3.669885864793679e-06, "loss": 0.23123292922973632, "step": 210 }, { "epoch": 0.00579513537897222, "grad_norm": 23.698759078979492, "learning_rate": 3.845478489903424e-06, "loss": 0.2144315242767334, "step": 220 }, { "epoch": 0.0060585506234709565, "grad_norm": 15.127054214477539, "learning_rate": 4.0210711150131695e-06, "loss": 0.2204211473464966, "step": 230 }, { "epoch": 0.006321965867969694, "grad_norm": 12.879171371459961, "learning_rate": 4.196663740122916e-06, "loss": 0.24736757278442384, "step": 240 }, { "epoch": 0.006585381112468432, "grad_norm": 11.151238441467285, "learning_rate": 4.37225636523266e-06, "loss": 0.19195048809051513, "step": 250 }, { "epoch": 0.006848796356967168, "grad_norm": 9.844902992248535, "learning_rate": 4.547848990342406e-06, "loss": 0.22496633529663085, "step": 260 }, { "epoch": 0.007112211601465906, "grad_norm": 7.0281219482421875, "learning_rate": 4.7234416154521515e-06, "loss": 0.2010673999786377, "step": 270 }, { "epoch": 0.007375626845964643, "grad_norm": 12.537249565124512, "learning_rate": 4.899034240561897e-06, "loss": 0.20958292484283447, "step": 280 }, { "epoch": 0.00763904209046338, "grad_norm": 9.31061840057373, "learning_rate": 5.074626865671642e-06, "loss": 0.1994800567626953, "step": 290 }, { "epoch": 0.007902457334962117, "grad_norm": 7.942080020904541, "learning_rate": 5.250219490781387e-06, "loss": 0.2052018165588379, "step": 300 }, { "epoch": 0.008165872579460855, "grad_norm": 8.613945007324219, "learning_rate": 5.425812115891133e-06, "loss": 0.2028278112411499, "step": 310 }, { "epoch": 0.008429287823959592, "grad_norm": 11.384459495544434, "learning_rate": 5.601404741000879e-06, "loss": 0.20828416347503662, "step": 320 }, { "epoch": 0.008692703068458329, "grad_norm": 6.942869663238525, "learning_rate": 5.776997366110624e-06, "loss": 0.2186208963394165, "step": 330 }, { "epoch": 0.008956118312957067, "grad_norm": 12.008971214294434, "learning_rate": 5.9525899912203685e-06, "loss": 0.20838432312011718, "step": 340 }, { "epoch": 0.009219533557455804, "grad_norm": 10.14953899383545, "learning_rate": 6.128182616330115e-06, "loss": 0.22505829334259034, "step": 350 }, { "epoch": 0.00948294880195454, "grad_norm": 9.246990203857422, "learning_rate": 6.30377524143986e-06, "loss": 0.20929298400878907, "step": 360 }, { "epoch": 0.00974636404645328, "grad_norm": 7.53009557723999, "learning_rate": 6.479367866549605e-06, "loss": 0.21752710342407228, "step": 370 }, { "epoch": 0.010009779290952016, "grad_norm": 8.697466850280762, "learning_rate": 6.654960491659351e-06, "loss": 0.21396634578704835, "step": 380 }, { "epoch": 0.010273194535450753, "grad_norm": 13.373923301696777, "learning_rate": 6.830553116769097e-06, "loss": 0.21112780570983886, "step": 390 }, { "epoch": 0.01053660977994949, "grad_norm": 6.869314670562744, "learning_rate": 7.006145741878841e-06, "loss": 0.20314462184906007, "step": 400 }, { "epoch": 0.010800025024448228, "grad_norm": 4.798375129699707, "learning_rate": 7.181738366988587e-06, "loss": 0.191727876663208, "step": 410 }, { "epoch": 0.011063440268946964, "grad_norm": 5.486581325531006, "learning_rate": 7.3573309920983325e-06, "loss": 0.20453305244445802, "step": 420 }, { "epoch": 0.011326855513445701, "grad_norm": 14.050649642944336, "learning_rate": 7.532923617208078e-06, "loss": 0.1888379454612732, "step": 430 }, { "epoch": 0.01159027075794444, "grad_norm": 4.085829734802246, "learning_rate": 7.708516242317823e-06, "loss": 0.1985569953918457, "step": 440 }, { "epoch": 0.011853686002443176, "grad_norm": 8.606118202209473, "learning_rate": 7.88410886742757e-06, "loss": 0.2077577829360962, "step": 450 }, { "epoch": 0.012117101246941913, "grad_norm": 7.416240692138672, "learning_rate": 8.059701492537314e-06, "loss": 0.1977151870727539, "step": 460 }, { "epoch": 0.012380516491440651, "grad_norm": 4.727428913116455, "learning_rate": 8.23529411764706e-06, "loss": 0.1897322416305542, "step": 470 }, { "epoch": 0.012643931735939388, "grad_norm": 5.476516246795654, "learning_rate": 8.410886742756806e-06, "loss": 0.2021705150604248, "step": 480 }, { "epoch": 0.012907346980438125, "grad_norm": 9.653079986572266, "learning_rate": 8.58647936786655e-06, "loss": 0.20684437751770018, "step": 490 }, { "epoch": 0.013170762224936863, "grad_norm": 12.809711456298828, "learning_rate": 8.762071992976295e-06, "loss": 0.225067400932312, "step": 500 }, { "epoch": 0.0134341774694356, "grad_norm": 8.395285606384277, "learning_rate": 8.937664618086041e-06, "loss": 0.22475228309631348, "step": 510 }, { "epoch": 0.013697592713934337, "grad_norm": 8.014936447143555, "learning_rate": 9.113257243195787e-06, "loss": 0.20959186553955078, "step": 520 }, { "epoch": 0.013961007958433075, "grad_norm": 5.109429836273193, "learning_rate": 9.288849868305532e-06, "loss": 0.20533463954925538, "step": 530 }, { "epoch": 0.014224423202931812, "grad_norm": 9.933526992797852, "learning_rate": 9.464442493415278e-06, "loss": 0.22217044830322266, "step": 540 }, { "epoch": 0.014487838447430549, "grad_norm": 2.8154866695404053, "learning_rate": 9.640035118525024e-06, "loss": 0.20622563362121582, "step": 550 }, { "epoch": 0.014751253691929285, "grad_norm": 6.356856822967529, "learning_rate": 9.815627743634768e-06, "loss": 0.1982288360595703, "step": 560 }, { "epoch": 0.015014668936428024, "grad_norm": 4.393381595611572, "learning_rate": 9.991220368744513e-06, "loss": 0.20429975986480714, "step": 570 }, { "epoch": 0.01527808418092676, "grad_norm": 5.146602153778076, "learning_rate": 1.0166812993854259e-05, "loss": 0.20192668437957764, "step": 580 }, { "epoch": 0.015541499425425497, "grad_norm": 8.42093276977539, "learning_rate": 1.0342405618964005e-05, "loss": 0.19306442737579346, "step": 590 }, { "epoch": 0.015804914669924234, "grad_norm": 9.291626930236816, "learning_rate": 1.0517998244073751e-05, "loss": 0.20412278175354004, "step": 600 }, { "epoch": 0.016068329914422974, "grad_norm": 5.013097763061523, "learning_rate": 1.0693590869183494e-05, "loss": 0.20329804420471193, "step": 610 }, { "epoch": 0.01633174515892171, "grad_norm": 5.3797101974487305, "learning_rate": 1.086918349429324e-05, "loss": 0.19328720569610597, "step": 620 }, { "epoch": 0.016595160403420448, "grad_norm": 10.365560531616211, "learning_rate": 1.1044776119402986e-05, "loss": 0.22095119953155518, "step": 630 }, { "epoch": 0.016858575647919184, "grad_norm": 8.092788696289062, "learning_rate": 1.122036874451273e-05, "loss": 0.18830541372299195, "step": 640 }, { "epoch": 0.01712199089241792, "grad_norm": 7.106931209564209, "learning_rate": 1.1395961369622477e-05, "loss": 0.19556691646575927, "step": 650 }, { "epoch": 0.017385406136916658, "grad_norm": 4.923224449157715, "learning_rate": 1.1571553994732223e-05, "loss": 0.2208932876586914, "step": 660 }, { "epoch": 0.017648821381415394, "grad_norm": 5.950439453125, "learning_rate": 1.1747146619841967e-05, "loss": 0.20590548515319823, "step": 670 }, { "epoch": 0.017912236625914135, "grad_norm": 6.09124231338501, "learning_rate": 1.1922739244951713e-05, "loss": 0.20061566829681396, "step": 680 }, { "epoch": 0.01817565187041287, "grad_norm": 8.927456855773926, "learning_rate": 1.209833187006146e-05, "loss": 0.2188793420791626, "step": 690 }, { "epoch": 0.018439067114911608, "grad_norm": 16.856815338134766, "learning_rate": 1.2273924495171204e-05, "loss": 0.20925378799438477, "step": 700 }, { "epoch": 0.018702482359410345, "grad_norm": 7.197027683258057, "learning_rate": 1.244951712028095e-05, "loss": 0.18801695108413696, "step": 710 }, { "epoch": 0.01896589760390908, "grad_norm": 7.076358795166016, "learning_rate": 1.2625109745390696e-05, "loss": 0.19598019123077393, "step": 720 }, { "epoch": 0.019229312848407818, "grad_norm": 5.115788459777832, "learning_rate": 1.2800702370500439e-05, "loss": 0.19689607620239258, "step": 730 }, { "epoch": 0.01949272809290656, "grad_norm": 8.356791496276855, "learning_rate": 1.2976294995610185e-05, "loss": 0.20569064617156982, "step": 740 }, { "epoch": 0.019756143337405295, "grad_norm": 6.786954879760742, "learning_rate": 1.315188762071993e-05, "loss": 0.22605838775634765, "step": 750 }, { "epoch": 0.02001955858190403, "grad_norm": 5.884575843811035, "learning_rate": 1.3327480245829676e-05, "loss": 0.20726189613342286, "step": 760 }, { "epoch": 0.02028297382640277, "grad_norm": 8.398327827453613, "learning_rate": 1.3503072870939422e-05, "loss": 0.21570098400115967, "step": 770 }, { "epoch": 0.020546389070901505, "grad_norm": 3.800367832183838, "learning_rate": 1.3678665496049166e-05, "loss": 0.20306749343872071, "step": 780 }, { "epoch": 0.020809804315400242, "grad_norm": 6.291841506958008, "learning_rate": 1.3854258121158912e-05, "loss": 0.19225581884384155, "step": 790 }, { "epoch": 0.02107321955989898, "grad_norm": 15.661744117736816, "learning_rate": 1.4029850746268658e-05, "loss": 0.21528444290161133, "step": 800 }, { "epoch": 0.02133663480439772, "grad_norm": 4.332355499267578, "learning_rate": 1.4205443371378403e-05, "loss": 0.195716392993927, "step": 810 }, { "epoch": 0.021600050048896455, "grad_norm": 4.03119421005249, "learning_rate": 1.4381035996488149e-05, "loss": 0.19138283729553224, "step": 820 }, { "epoch": 0.021863465293395192, "grad_norm": 3.378347396850586, "learning_rate": 1.4556628621597895e-05, "loss": 0.19553327560424805, "step": 830 }, { "epoch": 0.02212688053789393, "grad_norm": 3.450166940689087, "learning_rate": 1.4732221246707638e-05, "loss": 0.19388415813446044, "step": 840 }, { "epoch": 0.022390295782392666, "grad_norm": 5.600718021392822, "learning_rate": 1.4907813871817384e-05, "loss": 0.1953861951828003, "step": 850 }, { "epoch": 0.022653711026891402, "grad_norm": 3.8659346103668213, "learning_rate": 1.508340649692713e-05, "loss": 0.19567711353302003, "step": 860 }, { "epoch": 0.022917126271390142, "grad_norm": 7.669979572296143, "learning_rate": 1.5258999122036876e-05, "loss": 0.20021870136260986, "step": 870 }, { "epoch": 0.02318054151588888, "grad_norm": 4.061251640319824, "learning_rate": 1.5434591747146622e-05, "loss": 0.21754918098449708, "step": 880 }, { "epoch": 0.023443956760387616, "grad_norm": 6.225703716278076, "learning_rate": 1.561018437225637e-05, "loss": 0.20131199359893798, "step": 890 }, { "epoch": 0.023707372004886353, "grad_norm": 4.976027011871338, "learning_rate": 1.578577699736611e-05, "loss": 0.18431692123413085, "step": 900 }, { "epoch": 0.02397078724938509, "grad_norm": 4.326873302459717, "learning_rate": 1.5961369622475857e-05, "loss": 0.19792138338088988, "step": 910 }, { "epoch": 0.024234202493883826, "grad_norm": 11.18858528137207, "learning_rate": 1.6136962247585604e-05, "loss": 0.20222058296203613, "step": 920 }, { "epoch": 0.024497617738382566, "grad_norm": 2.854438304901123, "learning_rate": 1.6312554872695346e-05, "loss": 0.20661487579345703, "step": 930 }, { "epoch": 0.024761032982881303, "grad_norm": 3.32027268409729, "learning_rate": 1.6488147497805093e-05, "loss": 0.18844704627990722, "step": 940 }, { "epoch": 0.02502444822738004, "grad_norm": 3.5729265213012695, "learning_rate": 1.666374012291484e-05, "loss": 0.18531883955001832, "step": 950 }, { "epoch": 0.025287863471878776, "grad_norm": 4.550631999969482, "learning_rate": 1.6839332748024585e-05, "loss": 0.18942469358444214, "step": 960 }, { "epoch": 0.025551278716377513, "grad_norm": 2.2843403816223145, "learning_rate": 1.701492537313433e-05, "loss": 0.18100420236587525, "step": 970 }, { "epoch": 0.02581469396087625, "grad_norm": 8.400833129882812, "learning_rate": 1.7190517998244077e-05, "loss": 0.2040170907974243, "step": 980 }, { "epoch": 0.026078109205374986, "grad_norm": 4.813568592071533, "learning_rate": 1.736611062335382e-05, "loss": 0.18595497608184813, "step": 990 }, { "epoch": 0.026341524449873727, "grad_norm": 6.092161655426025, "learning_rate": 1.7541703248463566e-05, "loss": 0.19469231367111206, "step": 1000 }, { "epoch": 0.026604939694372463, "grad_norm": 4.305901527404785, "learning_rate": 1.771729587357331e-05, "loss": 0.2061032772064209, "step": 1010 }, { "epoch": 0.0268683549388712, "grad_norm": 6.567442417144775, "learning_rate": 1.7892888498683055e-05, "loss": 0.20711908340454102, "step": 1020 }, { "epoch": 0.027131770183369937, "grad_norm": 4.914463996887207, "learning_rate": 1.80684811237928e-05, "loss": 0.18982532024383544, "step": 1030 }, { "epoch": 0.027395185427868674, "grad_norm": 3.1977298259735107, "learning_rate": 1.8244073748902547e-05, "loss": 0.19963892698287963, "step": 1040 }, { "epoch": 0.02765860067236741, "grad_norm": 2.9278969764709473, "learning_rate": 1.8419666374012293e-05, "loss": 0.18439323902130128, "step": 1050 }, { "epoch": 0.02792201591686615, "grad_norm": 3.881558895111084, "learning_rate": 1.859525899912204e-05, "loss": 0.20480668544769287, "step": 1060 }, { "epoch": 0.028185431161364887, "grad_norm": 4.101785659790039, "learning_rate": 1.8770851624231782e-05, "loss": 0.193247652053833, "step": 1070 }, { "epoch": 0.028448846405863624, "grad_norm": 3.6158151626586914, "learning_rate": 1.8946444249341528e-05, "loss": 0.21265344619750975, "step": 1080 }, { "epoch": 0.02871226165036236, "grad_norm": 7.155221462249756, "learning_rate": 1.9122036874451274e-05, "loss": 0.20929193496704102, "step": 1090 }, { "epoch": 0.028975676894861097, "grad_norm": 3.99430775642395, "learning_rate": 1.929762949956102e-05, "loss": 0.19105074405670167, "step": 1100 }, { "epoch": 0.029239092139359834, "grad_norm": 3.5792417526245117, "learning_rate": 1.9473222124670767e-05, "loss": 0.20809152126312255, "step": 1110 }, { "epoch": 0.02950250738385857, "grad_norm": 8.660551071166992, "learning_rate": 1.9648814749780513e-05, "loss": 0.1879955768585205, "step": 1120 }, { "epoch": 0.02976592262835731, "grad_norm": 5.079763412475586, "learning_rate": 1.9824407374890256e-05, "loss": 0.19388556480407715, "step": 1130 }, { "epoch": 0.030029337872856048, "grad_norm": 2.9841113090515137, "learning_rate": 2e-05, "loss": 0.19943675994873047, "step": 1140 }, { "epoch": 0.030292753117354784, "grad_norm": 8.4368896484375, "learning_rate": 1.9999996360784282e-05, "loss": 0.1987345576286316, "step": 1150 }, { "epoch": 0.03055616836185352, "grad_norm": 5.8243279457092285, "learning_rate": 1.9999985443139778e-05, "loss": 0.1869898796081543, "step": 1160 }, { "epoch": 0.030819583606352258, "grad_norm": 3.484631061553955, "learning_rate": 1.999996724707443e-05, "loss": 0.19906144142150878, "step": 1170 }, { "epoch": 0.031082998850850994, "grad_norm": 5.49529504776001, "learning_rate": 1.9999941772601482e-05, "loss": 0.1874009132385254, "step": 1180 }, { "epoch": 0.03134641409534973, "grad_norm": 8.123685836791992, "learning_rate": 1.9999909019739482e-05, "loss": 0.19805971384048462, "step": 1190 }, { "epoch": 0.03160982933984847, "grad_norm": 5.616997718811035, "learning_rate": 1.9999868988512265e-05, "loss": 0.1849887490272522, "step": 1200 }, { "epoch": 0.031873244584347205, "grad_norm": 12.930646896362305, "learning_rate": 1.9999821678948964e-05, "loss": 0.19004735946655274, "step": 1210 }, { "epoch": 0.03213665982884595, "grad_norm": 7.840391159057617, "learning_rate": 1.999976709108402e-05, "loss": 0.20391764640808105, "step": 1220 }, { "epoch": 0.032400075073344685, "grad_norm": 8.024591445922852, "learning_rate": 1.9999705224957156e-05, "loss": 0.21714406013488768, "step": 1230 }, { "epoch": 0.03266349031784342, "grad_norm": 12.377995491027832, "learning_rate": 1.9999636080613406e-05, "loss": 0.21191973686218263, "step": 1240 }, { "epoch": 0.03292690556234216, "grad_norm": 4.593820095062256, "learning_rate": 1.9999559658103097e-05, "loss": 0.19745036363601684, "step": 1250 }, { "epoch": 0.033190320806840895, "grad_norm": 4.265445709228516, "learning_rate": 1.9999475957481848e-05, "loss": 0.20346131324768066, "step": 1260 }, { "epoch": 0.03345373605133963, "grad_norm": 1.9713696241378784, "learning_rate": 1.999938497881059e-05, "loss": 0.2019831657409668, "step": 1270 }, { "epoch": 0.03371715129583837, "grad_norm": 2.55804443359375, "learning_rate": 1.999928672215553e-05, "loss": 0.187864089012146, "step": 1280 }, { "epoch": 0.033980566540337105, "grad_norm": 3.2363650798797607, "learning_rate": 1.9999181187588187e-05, "loss": 0.2095876693725586, "step": 1290 }, { "epoch": 0.03424398178483584, "grad_norm": 4.583410739898682, "learning_rate": 1.9999068375185372e-05, "loss": 0.19353764057159423, "step": 1300 }, { "epoch": 0.03450739702933458, "grad_norm": 8.890861511230469, "learning_rate": 1.99989482850292e-05, "loss": 0.2021193027496338, "step": 1310 }, { "epoch": 0.034770812273833315, "grad_norm": 2.3753862380981445, "learning_rate": 1.9998820917207076e-05, "loss": 0.19254012107849122, "step": 1320 }, { "epoch": 0.03503422751833205, "grad_norm": 2.3384273052215576, "learning_rate": 1.99986862718117e-05, "loss": 0.19241068363189698, "step": 1330 }, { "epoch": 0.03529764276283079, "grad_norm": 4.0200700759887695, "learning_rate": 1.999854434894108e-05, "loss": 0.1987602472305298, "step": 1340 }, { "epoch": 0.03556105800732953, "grad_norm": 4.7808356285095215, "learning_rate": 1.9998395148698505e-05, "loss": 0.2087228775024414, "step": 1350 }, { "epoch": 0.03582447325182827, "grad_norm": 10.173592567443848, "learning_rate": 1.9998238671192575e-05, "loss": 0.1962893486022949, "step": 1360 }, { "epoch": 0.036087888496327006, "grad_norm": 5.859401702880859, "learning_rate": 1.999807491653718e-05, "loss": 0.19477157592773436, "step": 1370 }, { "epoch": 0.03635130374082574, "grad_norm": 4.282824516296387, "learning_rate": 1.9997903884851503e-05, "loss": 0.19119102954864503, "step": 1380 }, { "epoch": 0.03661471898532448, "grad_norm": 3.0151710510253906, "learning_rate": 1.999772557626004e-05, "loss": 0.186972975730896, "step": 1390 }, { "epoch": 0.036878134229823216, "grad_norm": 3.7359418869018555, "learning_rate": 1.9997539990892555e-05, "loss": 0.1804179072380066, "step": 1400 }, { "epoch": 0.03714154947432195, "grad_norm": 4.767220973968506, "learning_rate": 1.9997347128884138e-05, "loss": 0.18955593109130858, "step": 1410 }, { "epoch": 0.03740496471882069, "grad_norm": 2.4493298530578613, "learning_rate": 1.9997146990375162e-05, "loss": 0.2060218334197998, "step": 1420 }, { "epoch": 0.037668379963319426, "grad_norm": 3.9790430068969727, "learning_rate": 1.999693957551129e-05, "loss": 0.18828206062316893, "step": 1430 }, { "epoch": 0.03793179520781816, "grad_norm": 6.32949686050415, "learning_rate": 1.999672488444349e-05, "loss": 0.18192107677459718, "step": 1440 }, { "epoch": 0.0381952104523169, "grad_norm": 3.3326783180236816, "learning_rate": 1.999650291732802e-05, "loss": 0.20310354232788086, "step": 1450 }, { "epoch": 0.038458625696815636, "grad_norm": 4.431863307952881, "learning_rate": 1.9996273674326446e-05, "loss": 0.19434974193572999, "step": 1460 }, { "epoch": 0.03872204094131437, "grad_norm": 3.508920431137085, "learning_rate": 1.9996037155605615e-05, "loss": 0.18634659051895142, "step": 1470 }, { "epoch": 0.03898545618581312, "grad_norm": 4.161068439483643, "learning_rate": 1.9995793361337676e-05, "loss": 0.19613630771636964, "step": 1480 }, { "epoch": 0.03924887143031185, "grad_norm": 5.102060317993164, "learning_rate": 1.9995542291700075e-05, "loss": 0.19299402236938476, "step": 1490 }, { "epoch": 0.03951228667481059, "grad_norm": 4.437551021575928, "learning_rate": 1.9995283946875548e-05, "loss": 0.18666152954101561, "step": 1500 }, { "epoch": 0.03977570191930933, "grad_norm": 5.511523246765137, "learning_rate": 1.999501832705213e-05, "loss": 0.1759363055229187, "step": 1510 }, { "epoch": 0.04003911716380806, "grad_norm": 2.2573740482330322, "learning_rate": 1.9994745432423154e-05, "loss": 0.17456045150756835, "step": 1520 }, { "epoch": 0.0403025324083068, "grad_norm": 5.215588569641113, "learning_rate": 1.9994465263187242e-05, "loss": 0.18085215091705323, "step": 1530 }, { "epoch": 0.04056594765280554, "grad_norm": 6.665383338928223, "learning_rate": 1.9994177819548313e-05, "loss": 0.20858979225158691, "step": 1540 }, { "epoch": 0.040829362897304274, "grad_norm": 3.927070140838623, "learning_rate": 1.999388310171558e-05, "loss": 0.19421851634979248, "step": 1550 }, { "epoch": 0.04109277814180301, "grad_norm": 6.070815086364746, "learning_rate": 1.9993581109903557e-05, "loss": 0.1759726285934448, "step": 1560 }, { "epoch": 0.04135619338630175, "grad_norm": 6.443704605102539, "learning_rate": 1.999327184433204e-05, "loss": 0.18363442420959472, "step": 1570 }, { "epoch": 0.041619608630800484, "grad_norm": 3.78230881690979, "learning_rate": 1.9992955305226127e-05, "loss": 0.18802294731140137, "step": 1580 }, { "epoch": 0.04188302387529922, "grad_norm": 3.0033769607543945, "learning_rate": 1.9992631492816214e-05, "loss": 0.1903481364250183, "step": 1590 }, { "epoch": 0.04214643911979796, "grad_norm": 2.7623050212860107, "learning_rate": 1.9992300407337977e-05, "loss": 0.2034848928451538, "step": 1600 }, { "epoch": 0.0424098543642967, "grad_norm": 8.225008010864258, "learning_rate": 1.9991962049032403e-05, "loss": 0.18417246341705323, "step": 1610 }, { "epoch": 0.04267326960879544, "grad_norm": 7.186054706573486, "learning_rate": 1.9991616418145757e-05, "loss": 0.19282492399215698, "step": 1620 }, { "epoch": 0.042936684853294174, "grad_norm": 3.883014678955078, "learning_rate": 1.9991263514929606e-05, "loss": 0.1795957088470459, "step": 1630 }, { "epoch": 0.04320010009779291, "grad_norm": 7.728805065155029, "learning_rate": 1.999090333964081e-05, "loss": 0.19285954236984254, "step": 1640 }, { "epoch": 0.04346351534229165, "grad_norm": 1.9994605779647827, "learning_rate": 1.9990535892541524e-05, "loss": 0.1797998309135437, "step": 1650 }, { "epoch": 0.043726930586790384, "grad_norm": 5.271703243255615, "learning_rate": 1.999016117389918e-05, "loss": 0.20926718711853026, "step": 1660 }, { "epoch": 0.04399034583128912, "grad_norm": 5.857334136962891, "learning_rate": 1.998977918398653e-05, "loss": 0.2037198781967163, "step": 1670 }, { "epoch": 0.04425376107578786, "grad_norm": 2.4153635501861572, "learning_rate": 1.9989389923081587e-05, "loss": 0.17998857498168946, "step": 1680 }, { "epoch": 0.044517176320286594, "grad_norm": 3.9861385822296143, "learning_rate": 1.998899339146768e-05, "loss": 0.18005175590515138, "step": 1690 }, { "epoch": 0.04478059156478533, "grad_norm": 2.412702798843384, "learning_rate": 1.9988589589433423e-05, "loss": 0.1967910885810852, "step": 1700 }, { "epoch": 0.04504400680928407, "grad_norm": 5.418268203735352, "learning_rate": 1.9988178517272717e-05, "loss": 0.1961337447166443, "step": 1710 }, { "epoch": 0.045307422053782805, "grad_norm": 4.243969440460205, "learning_rate": 1.998776017528476e-05, "loss": 0.19343302249908448, "step": 1720 }, { "epoch": 0.04557083729828155, "grad_norm": 4.495187282562256, "learning_rate": 1.9987334563774038e-05, "loss": 0.1940288782119751, "step": 1730 }, { "epoch": 0.045834252542780285, "grad_norm": 6.383896827697754, "learning_rate": 1.998690168305033e-05, "loss": 0.18307141065597535, "step": 1740 }, { "epoch": 0.04609766778727902, "grad_norm": 2.4367008209228516, "learning_rate": 1.9986461533428708e-05, "loss": 0.2010852575302124, "step": 1750 }, { "epoch": 0.04636108303177776, "grad_norm": 3.989253520965576, "learning_rate": 1.9986014115229525e-05, "loss": 0.19172250032424926, "step": 1760 }, { "epoch": 0.046624498276276495, "grad_norm": 1.7378616333007812, "learning_rate": 1.998555942877844e-05, "loss": 0.2011963129043579, "step": 1770 }, { "epoch": 0.04688791352077523, "grad_norm": 2.9213643074035645, "learning_rate": 1.9985097474406387e-05, "loss": 0.19361844062805175, "step": 1780 }, { "epoch": 0.04715132876527397, "grad_norm": 2.718158721923828, "learning_rate": 1.99846282524496e-05, "loss": 0.17704784870147705, "step": 1790 }, { "epoch": 0.047414744009772705, "grad_norm": 4.917280673980713, "learning_rate": 1.9984151763249598e-05, "loss": 0.19352484941482545, "step": 1800 }, { "epoch": 0.04767815925427144, "grad_norm": 2.854273557662964, "learning_rate": 1.9983668007153185e-05, "loss": 0.19346017837524415, "step": 1810 }, { "epoch": 0.04794157449877018, "grad_norm": 4.789058685302734, "learning_rate": 1.998317698451247e-05, "loss": 0.18207008838653566, "step": 1820 }, { "epoch": 0.048204989743268915, "grad_norm": 3.5670313835144043, "learning_rate": 1.9982678695684833e-05, "loss": 0.1915727138519287, "step": 1830 }, { "epoch": 0.04846840498776765, "grad_norm": 1.954155683517456, "learning_rate": 1.998217314103295e-05, "loss": 0.1965188980102539, "step": 1840 }, { "epoch": 0.04873182023226639, "grad_norm": 2.256221055984497, "learning_rate": 1.998166032092479e-05, "loss": 0.190895414352417, "step": 1850 }, { "epoch": 0.04899523547676513, "grad_norm": 2.570856809616089, "learning_rate": 1.99811402357336e-05, "loss": 0.1738641381263733, "step": 1860 }, { "epoch": 0.04925865072126387, "grad_norm": 4.29306697845459, "learning_rate": 1.9980612885837925e-05, "loss": 0.203872013092041, "step": 1870 }, { "epoch": 0.049522065965762606, "grad_norm": 2.5009875297546387, "learning_rate": 1.998007827162159e-05, "loss": 0.1840360641479492, "step": 1880 }, { "epoch": 0.04978548121026134, "grad_norm": 5.466083526611328, "learning_rate": 1.9979536393473715e-05, "loss": 0.19663636684417723, "step": 1890 }, { "epoch": 0.05004889645476008, "grad_norm": 3.240764856338501, "learning_rate": 1.9978987251788696e-05, "loss": 0.18610994815826415, "step": 1900 }, { "epoch": 0.050312311699258816, "grad_norm": 5.236108303070068, "learning_rate": 1.997843084696623e-05, "loss": 0.19160077571868897, "step": 1910 }, { "epoch": 0.05057572694375755, "grad_norm": 5.601428985595703, "learning_rate": 1.9977867179411282e-05, "loss": 0.16876891851425171, "step": 1920 }, { "epoch": 0.05083914218825629, "grad_norm": 2.9731242656707764, "learning_rate": 1.997729624953412e-05, "loss": 0.20094051361083984, "step": 1930 }, { "epoch": 0.051102557432755026, "grad_norm": 1.6045525074005127, "learning_rate": 1.997671805775029e-05, "loss": 0.19954390525817872, "step": 1940 }, { "epoch": 0.05136597267725376, "grad_norm": 5.317564010620117, "learning_rate": 1.9976132604480625e-05, "loss": 0.1913461923599243, "step": 1950 }, { "epoch": 0.0516293879217525, "grad_norm": 3.703770875930786, "learning_rate": 1.997553989015125e-05, "loss": 0.19999196529388427, "step": 1960 }, { "epoch": 0.051892803166251236, "grad_norm": 5.3549652099609375, "learning_rate": 1.9974939915193557e-05, "loss": 0.1827751636505127, "step": 1970 }, { "epoch": 0.05215621841074997, "grad_norm": 2.7885472774505615, "learning_rate": 1.9974332680044237e-05, "loss": 0.19259028434753417, "step": 1980 }, { "epoch": 0.05241963365524872, "grad_norm": 2.5733935832977295, "learning_rate": 1.9973718185145262e-05, "loss": 0.17758619785308838, "step": 1990 }, { "epoch": 0.05268304889974745, "grad_norm": 5.633772850036621, "learning_rate": 1.997309643094389e-05, "loss": 0.18264939785003662, "step": 2000 }, { "epoch": 0.05294646414424619, "grad_norm": 1.8114683628082275, "learning_rate": 1.997246741789266e-05, "loss": 0.1809229373931885, "step": 2010 }, { "epoch": 0.05320987938874493, "grad_norm": 4.842754364013672, "learning_rate": 1.99718311464494e-05, "loss": 0.19791162014007568, "step": 2020 }, { "epoch": 0.05347329463324366, "grad_norm": 3.468501567840576, "learning_rate": 1.9971187617077204e-05, "loss": 0.182889187335968, "step": 2030 }, { "epoch": 0.0537367098777424, "grad_norm": 3.392738103866577, "learning_rate": 1.9970536830244466e-05, "loss": 0.18658167123794556, "step": 2040 }, { "epoch": 0.05400012512224114, "grad_norm": 2.5641894340515137, "learning_rate": 1.9969878786424863e-05, "loss": 0.18103630542755128, "step": 2050 }, { "epoch": 0.054263540366739874, "grad_norm": 3.3263938426971436, "learning_rate": 1.9969213486097335e-05, "loss": 0.2000105381011963, "step": 2060 }, { "epoch": 0.05452695561123861, "grad_norm": 6.9016828536987305, "learning_rate": 1.996854092974613e-05, "loss": 0.21134014129638673, "step": 2070 }, { "epoch": 0.05479037085573735, "grad_norm": 5.948678970336914, "learning_rate": 1.9967861117860754e-05, "loss": 0.18224475383758545, "step": 2080 }, { "epoch": 0.055053786100236084, "grad_norm": 3.1152126789093018, "learning_rate": 1.9967174050936008e-05, "loss": 0.1916734218597412, "step": 2090 }, { "epoch": 0.05531720134473482, "grad_norm": 3.684882879257202, "learning_rate": 1.9966479729471966e-05, "loss": 0.19970328807830812, "step": 2100 }, { "epoch": 0.05558061658923356, "grad_norm": 5.683203220367432, "learning_rate": 1.9965778153973986e-05, "loss": 0.1761358380317688, "step": 2110 }, { "epoch": 0.0558440318337323, "grad_norm": 4.704127788543701, "learning_rate": 1.9965069324952703e-05, "loss": 0.20139927864074708, "step": 2120 }, { "epoch": 0.05610744707823104, "grad_norm": 7.783297061920166, "learning_rate": 1.9964353242924038e-05, "loss": 0.20428757667541503, "step": 2130 }, { "epoch": 0.056370862322729774, "grad_norm": 3.179382085800171, "learning_rate": 1.9963629908409184e-05, "loss": 0.18872315883636476, "step": 2140 }, { "epoch": 0.05663427756722851, "grad_norm": 2.664132595062256, "learning_rate": 1.9962899321934616e-05, "loss": 0.1772153377532959, "step": 2150 }, { "epoch": 0.05689769281172725, "grad_norm": 3.0712883472442627, "learning_rate": 1.9962161484032085e-05, "loss": 0.16853346824645996, "step": 2160 }, { "epoch": 0.057161108056225984, "grad_norm": 3.729604721069336, "learning_rate": 1.9961416395238623e-05, "loss": 0.17138276100158692, "step": 2170 }, { "epoch": 0.05742452330072472, "grad_norm": 3.4849793910980225, "learning_rate": 1.9960664056096534e-05, "loss": 0.21526758670806884, "step": 2180 }, { "epoch": 0.05768793854522346, "grad_norm": 8.013435363769531, "learning_rate": 1.9959904467153408e-05, "loss": 0.1991405725479126, "step": 2190 }, { "epoch": 0.057951353789722194, "grad_norm": 1.8459198474884033, "learning_rate": 1.99591376289621e-05, "loss": 0.17872191667556764, "step": 2200 }, { "epoch": 0.05821476903422093, "grad_norm": 5.307249069213867, "learning_rate": 1.9958363542080758e-05, "loss": 0.19273022413253785, "step": 2210 }, { "epoch": 0.05847818427871967, "grad_norm": 2.01235294342041, "learning_rate": 1.9957582207072784e-05, "loss": 0.1797632694244385, "step": 2220 }, { "epoch": 0.058741599523218405, "grad_norm": 3.9431867599487305, "learning_rate": 1.9956793624506875e-05, "loss": 0.19611239433288574, "step": 2230 }, { "epoch": 0.05900501476771714, "grad_norm": 5.2949957847595215, "learning_rate": 1.9955997794956994e-05, "loss": 0.17748968601226806, "step": 2240 }, { "epoch": 0.059268430012215885, "grad_norm": 4.127165794372559, "learning_rate": 1.995519471900238e-05, "loss": 0.18551356792449952, "step": 2250 }, { "epoch": 0.05953184525671462, "grad_norm": 2.153325080871582, "learning_rate": 1.9954384397227545e-05, "loss": 0.17884693145751954, "step": 2260 }, { "epoch": 0.05979526050121336, "grad_norm": 4.7212724685668945, "learning_rate": 1.9953566830222276e-05, "loss": 0.1770597815513611, "step": 2270 }, { "epoch": 0.060058675745712095, "grad_norm": 4.126374244689941, "learning_rate": 1.9952742018581634e-05, "loss": 0.19542912244796753, "step": 2280 }, { "epoch": 0.06032209099021083, "grad_norm": 4.265449047088623, "learning_rate": 1.9951909962905956e-05, "loss": 0.1781148076057434, "step": 2290 }, { "epoch": 0.06058550623470957, "grad_norm": 5.5661540031433105, "learning_rate": 1.9951070663800843e-05, "loss": 0.18060775995254516, "step": 2300 }, { "epoch": 0.060848921479208305, "grad_norm": 2.0413224697113037, "learning_rate": 1.995022412187718e-05, "loss": 0.1768651247024536, "step": 2310 }, { "epoch": 0.06111233672370704, "grad_norm": 4.222019672393799, "learning_rate": 1.9949370337751105e-05, "loss": 0.1885414481163025, "step": 2320 }, { "epoch": 0.06137575196820578, "grad_norm": 3.455770492553711, "learning_rate": 1.994850931204405e-05, "loss": 0.18390562534332275, "step": 2330 }, { "epoch": 0.061639167212704515, "grad_norm": 7.060954570770264, "learning_rate": 1.99476410453827e-05, "loss": 0.1971544861793518, "step": 2340 }, { "epoch": 0.06190258245720325, "grad_norm": 4.8201518058776855, "learning_rate": 1.994676553839902e-05, "loss": 0.20229337215423585, "step": 2350 }, { "epoch": 0.06216599770170199, "grad_norm": 6.087865352630615, "learning_rate": 1.9945882791730246e-05, "loss": 0.2004711389541626, "step": 2360 }, { "epoch": 0.06242941294620073, "grad_norm": 5.38054084777832, "learning_rate": 1.994499280601887e-05, "loss": 0.18110883235931396, "step": 2370 }, { "epoch": 0.06269282819069946, "grad_norm": 3.8969900608062744, "learning_rate": 1.994409558191267e-05, "loss": 0.17410547733306886, "step": 2380 }, { "epoch": 0.0629562434351982, "grad_norm": 6.07520866394043, "learning_rate": 1.9943191120064672e-05, "loss": 0.1887937903404236, "step": 2390 }, { "epoch": 0.06321965867969694, "grad_norm": 4.087823867797852, "learning_rate": 1.99422794211332e-05, "loss": 0.1824866533279419, "step": 2400 }, { "epoch": 0.06348307392419568, "grad_norm": 2.113745927810669, "learning_rate": 1.9941360485781814e-05, "loss": 0.18805363178253173, "step": 2410 }, { "epoch": 0.06374648916869441, "grad_norm": 2.2344987392425537, "learning_rate": 1.9940434314679363e-05, "loss": 0.18308205604553224, "step": 2420 }, { "epoch": 0.06400990441319315, "grad_norm": 3.702775239944458, "learning_rate": 1.9939500908499947e-05, "loss": 0.18080848455429077, "step": 2430 }, { "epoch": 0.0642733196576919, "grad_norm": 8.351118087768555, "learning_rate": 1.9938560267922948e-05, "loss": 0.19009337425231934, "step": 2440 }, { "epoch": 0.06453673490219063, "grad_norm": 2.0780444145202637, "learning_rate": 1.9937612393632995e-05, "loss": 0.16345133781433105, "step": 2450 }, { "epoch": 0.06480015014668937, "grad_norm": 2.673079252243042, "learning_rate": 1.9936657286320003e-05, "loss": 0.1774242639541626, "step": 2460 }, { "epoch": 0.0650635653911881, "grad_norm": 2.974534749984741, "learning_rate": 1.993569494667913e-05, "loss": 0.1810241937637329, "step": 2470 }, { "epoch": 0.06532698063568684, "grad_norm": 4.120419025421143, "learning_rate": 1.9934725375410812e-05, "loss": 0.1956166982650757, "step": 2480 }, { "epoch": 0.06559039588018557, "grad_norm": 3.244030475616455, "learning_rate": 1.9933748573220745e-05, "loss": 0.2022094249725342, "step": 2490 }, { "epoch": 0.06585381112468432, "grad_norm": 6.8200907707214355, "learning_rate": 1.993276454081989e-05, "loss": 0.18424432277679442, "step": 2500 }, { "epoch": 0.06611722636918305, "grad_norm": 3.571593999862671, "learning_rate": 1.9931773278924467e-05, "loss": 0.19122068881988524, "step": 2510 }, { "epoch": 0.06638064161368179, "grad_norm": 4.449934005737305, "learning_rate": 1.9930774788255957e-05, "loss": 0.1949880003929138, "step": 2520 }, { "epoch": 0.06664405685818052, "grad_norm": 2.6547319889068604, "learning_rate": 1.992976906954111e-05, "loss": 0.1954352378845215, "step": 2530 }, { "epoch": 0.06690747210267926, "grad_norm": 2.850346326828003, "learning_rate": 1.992875612351192e-05, "loss": 0.18321626186370848, "step": 2540 }, { "epoch": 0.067170887347178, "grad_norm": 1.535256266593933, "learning_rate": 1.9927735950905663e-05, "loss": 0.1647108316421509, "step": 2550 }, { "epoch": 0.06743430259167674, "grad_norm": 1.7613862752914429, "learning_rate": 1.9926708552464865e-05, "loss": 0.20731356143951415, "step": 2560 }, { "epoch": 0.06769771783617548, "grad_norm": 5.982910633087158, "learning_rate": 1.9925673928937307e-05, "loss": 0.19659221172332764, "step": 2570 }, { "epoch": 0.06796113308067421, "grad_norm": 3.0849854946136475, "learning_rate": 1.9924632081076032e-05, "loss": 0.18935306072235109, "step": 2580 }, { "epoch": 0.06822454832517295, "grad_norm": 2.256213665008545, "learning_rate": 1.9923583009639344e-05, "loss": 0.20212216377258302, "step": 2590 }, { "epoch": 0.06848796356967168, "grad_norm": 3.7723281383514404, "learning_rate": 1.9922526715390802e-05, "loss": 0.18694641590118408, "step": 2600 }, { "epoch": 0.06875137881417043, "grad_norm": 5.261135101318359, "learning_rate": 1.9921463199099222e-05, "loss": 0.18219997882843017, "step": 2610 }, { "epoch": 0.06901479405866916, "grad_norm": 2.2253568172454834, "learning_rate": 1.992039246153868e-05, "loss": 0.18656585216522217, "step": 2620 }, { "epoch": 0.0692782093031679, "grad_norm": 4.933704853057861, "learning_rate": 1.99193145034885e-05, "loss": 0.17344341278076172, "step": 2630 }, { "epoch": 0.06954162454766663, "grad_norm": 2.3711061477661133, "learning_rate": 1.9918229325733267e-05, "loss": 0.1775611400604248, "step": 2640 }, { "epoch": 0.06980503979216537, "grad_norm": 1.6992485523223877, "learning_rate": 1.9917136929062824e-05, "loss": 0.178259539604187, "step": 2650 }, { "epoch": 0.0700684550366641, "grad_norm": 4.324896335601807, "learning_rate": 1.991603731427226e-05, "loss": 0.18100647926330565, "step": 2660 }, { "epoch": 0.07033187028116285, "grad_norm": 5.618589401245117, "learning_rate": 1.9914930482161928e-05, "loss": 0.17496981620788574, "step": 2670 }, { "epoch": 0.07059528552566158, "grad_norm": 2.260545015335083, "learning_rate": 1.9913816433537425e-05, "loss": 0.1692511558532715, "step": 2680 }, { "epoch": 0.07085870077016032, "grad_norm": 3.6312239170074463, "learning_rate": 1.99126951692096e-05, "loss": 0.19629414081573487, "step": 2690 }, { "epoch": 0.07112211601465906, "grad_norm": 3.235715389251709, "learning_rate": 1.991156668999456e-05, "loss": 0.19286153316497803, "step": 2700 }, { "epoch": 0.0713855312591578, "grad_norm": 2.4739370346069336, "learning_rate": 1.9910430996713668e-05, "loss": 0.17220664024353027, "step": 2710 }, { "epoch": 0.07164894650365654, "grad_norm": 3.8688344955444336, "learning_rate": 1.9909288090193518e-05, "loss": 0.1894251823425293, "step": 2720 }, { "epoch": 0.07191236174815527, "grad_norm": 2.425143241882324, "learning_rate": 1.9908137971265977e-05, "loss": 0.19652724266052246, "step": 2730 }, { "epoch": 0.07217577699265401, "grad_norm": 2.979468822479248, "learning_rate": 1.9906980640768146e-05, "loss": 0.19069619178771974, "step": 2740 }, { "epoch": 0.07243919223715274, "grad_norm": 5.627507209777832, "learning_rate": 1.9905816099542382e-05, "loss": 0.1848670721054077, "step": 2750 }, { "epoch": 0.07270260748165149, "grad_norm": 3.759413719177246, "learning_rate": 1.9904644348436282e-05, "loss": 0.18021647930145263, "step": 2760 }, { "epoch": 0.07296602272615021, "grad_norm": 2.2049646377563477, "learning_rate": 1.990346538830271e-05, "loss": 0.178458571434021, "step": 2770 }, { "epoch": 0.07322943797064896, "grad_norm": 3.609325647354126, "learning_rate": 1.9902279219999752e-05, "loss": 0.18197252750396728, "step": 2780 }, { "epoch": 0.07349285321514769, "grad_norm": 5.374975204467773, "learning_rate": 1.9901085844390757e-05, "loss": 0.18253933191299437, "step": 2790 }, { "epoch": 0.07375626845964643, "grad_norm": 7.501921653747559, "learning_rate": 1.9899885262344316e-05, "loss": 0.19106523990631102, "step": 2800 }, { "epoch": 0.07401968370414516, "grad_norm": 2.9862067699432373, "learning_rate": 1.9898677474734266e-05, "loss": 0.17870492935180665, "step": 2810 }, { "epoch": 0.0742830989486439, "grad_norm": 3.5537593364715576, "learning_rate": 1.9897462482439684e-05, "loss": 0.17488914728164673, "step": 2820 }, { "epoch": 0.07454651419314265, "grad_norm": 4.422292709350586, "learning_rate": 1.9896240286344892e-05, "loss": 0.19606459140777588, "step": 2830 }, { "epoch": 0.07480992943764138, "grad_norm": 4.46948766708374, "learning_rate": 1.989501088733946e-05, "loss": 0.18325624465942383, "step": 2840 }, { "epoch": 0.07507334468214012, "grad_norm": 1.8666903972625732, "learning_rate": 1.9893774286318198e-05, "loss": 0.18278217315673828, "step": 2850 }, { "epoch": 0.07533675992663885, "grad_norm": 8.038261413574219, "learning_rate": 1.9892530484181155e-05, "loss": 0.18845727443695068, "step": 2860 }, { "epoch": 0.0756001751711376, "grad_norm": 3.7114603519439697, "learning_rate": 1.989127948183363e-05, "loss": 0.1833769679069519, "step": 2870 }, { "epoch": 0.07586359041563633, "grad_norm": 4.456262111663818, "learning_rate": 1.9890021280186148e-05, "loss": 0.20225114822387696, "step": 2880 }, { "epoch": 0.07612700566013507, "grad_norm": 3.581117868423462, "learning_rate": 1.9888755880154485e-05, "loss": 0.17572877407073975, "step": 2890 }, { "epoch": 0.0763904209046338, "grad_norm": 5.044270992279053, "learning_rate": 1.988748328265966e-05, "loss": 0.16917843818664552, "step": 2900 }, { "epoch": 0.07665383614913254, "grad_norm": 2.739537477493286, "learning_rate": 1.9886203488627918e-05, "loss": 0.1939738869667053, "step": 2910 }, { "epoch": 0.07691725139363127, "grad_norm": 3.971508502960205, "learning_rate": 1.988491649899075e-05, "loss": 0.18552069664001464, "step": 2920 }, { "epoch": 0.07718066663813002, "grad_norm": 5.139636993408203, "learning_rate": 1.988362231468488e-05, "loss": 0.18283467292785643, "step": 2930 }, { "epoch": 0.07744408188262875, "grad_norm": 2.1441259384155273, "learning_rate": 1.9882320936652278e-05, "loss": 0.19348957538604736, "step": 2940 }, { "epoch": 0.07770749712712749, "grad_norm": 2.6506121158599854, "learning_rate": 1.9881012365840138e-05, "loss": 0.17645202875137328, "step": 2950 }, { "epoch": 0.07797091237162623, "grad_norm": 2.769500970840454, "learning_rate": 1.987969660320089e-05, "loss": 0.1893176555633545, "step": 2960 }, { "epoch": 0.07823432761612496, "grad_norm": 2.2236979007720947, "learning_rate": 1.9878373649692213e-05, "loss": 0.19788388013839722, "step": 2970 }, { "epoch": 0.0784977428606237, "grad_norm": 1.887575387954712, "learning_rate": 1.9877043506277004e-05, "loss": 0.19241900444030763, "step": 2980 }, { "epoch": 0.07876115810512244, "grad_norm": 4.567235469818115, "learning_rate": 1.9875706173923396e-05, "loss": 0.1803753614425659, "step": 2990 }, { "epoch": 0.07902457334962118, "grad_norm": 3.3310654163360596, "learning_rate": 1.9874361653604765e-05, "loss": 0.1731792211532593, "step": 3000 }, { "epoch": 0.07928798859411991, "grad_norm": 3.720592975616455, "learning_rate": 1.9873009946299702e-05, "loss": 0.19136548042297363, "step": 3010 }, { "epoch": 0.07955140383861865, "grad_norm": 3.811061143875122, "learning_rate": 1.9871651052992046e-05, "loss": 0.17310857772827148, "step": 3020 }, { "epoch": 0.07981481908311738, "grad_norm": 5.058263778686523, "learning_rate": 1.987028497467085e-05, "loss": 0.16523392200469972, "step": 3030 }, { "epoch": 0.08007823432761613, "grad_norm": 2.781978130340576, "learning_rate": 1.986891171233041e-05, "loss": 0.171396005153656, "step": 3040 }, { "epoch": 0.08034164957211486, "grad_norm": 4.159412384033203, "learning_rate": 1.9867531266970247e-05, "loss": 0.18382549285888672, "step": 3050 }, { "epoch": 0.0806050648166136, "grad_norm": 1.7240400314331055, "learning_rate": 1.9866143639595107e-05, "loss": 0.1920453429222107, "step": 3060 }, { "epoch": 0.08086848006111233, "grad_norm": 3.9383609294891357, "learning_rate": 1.9864748831214964e-05, "loss": 0.17406328916549682, "step": 3070 }, { "epoch": 0.08113189530561107, "grad_norm": 3.711118459701538, "learning_rate": 1.986334684284502e-05, "loss": 0.1935066819190979, "step": 3080 }, { "epoch": 0.08139531055010982, "grad_norm": 5.658510684967041, "learning_rate": 1.98619376755057e-05, "loss": 0.18799118995666503, "step": 3090 }, { "epoch": 0.08165872579460855, "grad_norm": 1.8765764236450195, "learning_rate": 1.986052133022266e-05, "loss": 0.2039560317993164, "step": 3100 }, { "epoch": 0.08192214103910729, "grad_norm": 4.4753570556640625, "learning_rate": 1.985909780802678e-05, "loss": 0.176315975189209, "step": 3110 }, { "epoch": 0.08218555628360602, "grad_norm": 2.6213200092315674, "learning_rate": 1.985766710995416e-05, "loss": 0.17661077976226808, "step": 3120 }, { "epoch": 0.08244897152810476, "grad_norm": 7.07561731338501, "learning_rate": 1.9856229237046115e-05, "loss": 0.196761155128479, "step": 3130 }, { "epoch": 0.0827123867726035, "grad_norm": 3.318992853164673, "learning_rate": 1.9854784190349203e-05, "loss": 0.18834874629974366, "step": 3140 }, { "epoch": 0.08297580201710224, "grad_norm": 5.9729533195495605, "learning_rate": 1.985333197091518e-05, "loss": 0.1899608850479126, "step": 3150 }, { "epoch": 0.08323921726160097, "grad_norm": 5.7527079582214355, "learning_rate": 1.9851872579801044e-05, "loss": 0.17790971994400023, "step": 3160 }, { "epoch": 0.08350263250609971, "grad_norm": 1.212623953819275, "learning_rate": 1.9850406018068997e-05, "loss": 0.17326457500457765, "step": 3170 }, { "epoch": 0.08376604775059844, "grad_norm": 4.852723598480225, "learning_rate": 1.984893228678647e-05, "loss": 0.16704214811325074, "step": 3180 }, { "epoch": 0.08402946299509718, "grad_norm": 3.620713472366333, "learning_rate": 1.9847451387026103e-05, "loss": 0.17509396076202394, "step": 3190 }, { "epoch": 0.08429287823959591, "grad_norm": 5.292346954345703, "learning_rate": 1.9845963319865758e-05, "loss": 0.1908737063407898, "step": 3200 }, { "epoch": 0.08455629348409466, "grad_norm": 2.765699863433838, "learning_rate": 1.9844468086388523e-05, "loss": 0.17498217821121215, "step": 3210 }, { "epoch": 0.0848197087285934, "grad_norm": 6.491759777069092, "learning_rate": 1.9842965687682686e-05, "loss": 0.17661381959915162, "step": 3220 }, { "epoch": 0.08508312397309213, "grad_norm": 2.6886861324310303, "learning_rate": 1.9841456124841763e-05, "loss": 0.1853492498397827, "step": 3230 }, { "epoch": 0.08534653921759087, "grad_norm": 4.231110095977783, "learning_rate": 1.983993939896447e-05, "loss": 0.1698911190032959, "step": 3240 }, { "epoch": 0.0856099544620896, "grad_norm": 2.411189079284668, "learning_rate": 1.9838415511154755e-05, "loss": 0.19206788539886474, "step": 3250 }, { "epoch": 0.08587336970658835, "grad_norm": 4.202157497406006, "learning_rate": 1.9836884462521768e-05, "loss": 0.17713676691055297, "step": 3260 }, { "epoch": 0.08613678495108708, "grad_norm": 2.8477396965026855, "learning_rate": 1.9835346254179865e-05, "loss": 0.1824232220649719, "step": 3270 }, { "epoch": 0.08640020019558582, "grad_norm": 2.88810396194458, "learning_rate": 1.9833800887248628e-05, "loss": 0.18650965690612792, "step": 3280 }, { "epoch": 0.08666361544008455, "grad_norm": 5.302834510803223, "learning_rate": 1.983224836285284e-05, "loss": 0.18692878484725953, "step": 3290 }, { "epoch": 0.0869270306845833, "grad_norm": 6.293217182159424, "learning_rate": 1.9830688682122494e-05, "loss": 0.20510215759277345, "step": 3300 }, { "epoch": 0.08719044592908202, "grad_norm": 4.565415382385254, "learning_rate": 1.982912184619279e-05, "loss": 0.1802081823348999, "step": 3310 }, { "epoch": 0.08745386117358077, "grad_norm": 7.400785446166992, "learning_rate": 1.9827547856204147e-05, "loss": 0.16985639333724975, "step": 3320 }, { "epoch": 0.0877172764180795, "grad_norm": 2.4583890438079834, "learning_rate": 1.9825966713302172e-05, "loss": 0.18036284446716308, "step": 3330 }, { "epoch": 0.08798069166257824, "grad_norm": 4.334658145904541, "learning_rate": 1.98243784186377e-05, "loss": 0.17204302549362183, "step": 3340 }, { "epoch": 0.08824410690707699, "grad_norm": 3.7295656204223633, "learning_rate": 1.982278297336675e-05, "loss": 0.1842737913131714, "step": 3350 }, { "epoch": 0.08850752215157572, "grad_norm": 5.631339073181152, "learning_rate": 1.9821180378650563e-05, "loss": 0.18675651550292968, "step": 3360 }, { "epoch": 0.08877093739607446, "grad_norm": 4.010130882263184, "learning_rate": 1.9819570635655577e-05, "loss": 0.1886468291282654, "step": 3370 }, { "epoch": 0.08903435264057319, "grad_norm": 5.325526714324951, "learning_rate": 1.981795374555343e-05, "loss": 0.17968275547027587, "step": 3380 }, { "epoch": 0.08929776788507193, "grad_norm": 2.3901684284210205, "learning_rate": 1.981632970952096e-05, "loss": 0.17193782329559326, "step": 3390 }, { "epoch": 0.08956118312957066, "grad_norm": 2.116180181503296, "learning_rate": 1.9814698528740222e-05, "loss": 0.1863250970840454, "step": 3400 }, { "epoch": 0.0898245983740694, "grad_norm": 3.55928111076355, "learning_rate": 1.9813060204398448e-05, "loss": 0.17451035976409912, "step": 3410 }, { "epoch": 0.09008801361856814, "grad_norm": 3.283046245574951, "learning_rate": 1.9811414737688088e-05, "loss": 0.17659912109375, "step": 3420 }, { "epoch": 0.09035142886306688, "grad_norm": 2.042447566986084, "learning_rate": 1.9809762129806777e-05, "loss": 0.18425227403640748, "step": 3430 }, { "epoch": 0.09061484410756561, "grad_norm": 3.5048859119415283, "learning_rate": 1.9808102381957364e-05, "loss": 0.1910179615020752, "step": 3440 }, { "epoch": 0.09087825935206435, "grad_norm": 2.8837807178497314, "learning_rate": 1.980643549534788e-05, "loss": 0.17469851970672606, "step": 3450 }, { "epoch": 0.0911416745965631, "grad_norm": 1.7181791067123413, "learning_rate": 1.980476147119155e-05, "loss": 0.17176910638809204, "step": 3460 }, { "epoch": 0.09140508984106183, "grad_norm": 1.8453935384750366, "learning_rate": 1.980308031070682e-05, "loss": 0.17428576946258545, "step": 3470 }, { "epoch": 0.09166850508556057, "grad_norm": 3.24802827835083, "learning_rate": 1.9801392015117286e-05, "loss": 0.1999279260635376, "step": 3480 }, { "epoch": 0.0919319203300593, "grad_norm": 3.699427366256714, "learning_rate": 1.9799696585651782e-05, "loss": 0.16575183868408203, "step": 3490 }, { "epoch": 0.09219533557455804, "grad_norm": 1.8403193950653076, "learning_rate": 1.9797994023544307e-05, "loss": 0.19016122817993164, "step": 3500 }, { "epoch": 0.09245875081905677, "grad_norm": 2.37239933013916, "learning_rate": 1.979628433003406e-05, "loss": 0.17708756923675537, "step": 3510 }, { "epoch": 0.09272216606355552, "grad_norm": 2.9692983627319336, "learning_rate": 1.9794567506365432e-05, "loss": 0.17922472953796387, "step": 3520 }, { "epoch": 0.09298558130805425, "grad_norm": 3.0422770977020264, "learning_rate": 1.9792843553788e-05, "loss": 0.17406315803527833, "step": 3530 }, { "epoch": 0.09324899655255299, "grad_norm": 2.6803009510040283, "learning_rate": 1.9791112473556528e-05, "loss": 0.18567763566970824, "step": 3540 }, { "epoch": 0.09351241179705172, "grad_norm": 2.0695884227752686, "learning_rate": 1.978937426693097e-05, "loss": 0.18569003343582152, "step": 3550 }, { "epoch": 0.09377582704155046, "grad_norm": 3.5505635738372803, "learning_rate": 1.9787628935176476e-05, "loss": 0.18931155204772948, "step": 3560 }, { "epoch": 0.0940392422860492, "grad_norm": 3.178205728530884, "learning_rate": 1.9785876479563367e-05, "loss": 0.19232065677642823, "step": 3570 }, { "epoch": 0.09430265753054794, "grad_norm": 1.9780986309051514, "learning_rate": 1.978411690136716e-05, "loss": 0.18871079683303832, "step": 3580 }, { "epoch": 0.09456607277504668, "grad_norm": 3.0686018466949463, "learning_rate": 1.9782350201868547e-05, "loss": 0.1748607873916626, "step": 3590 }, { "epoch": 0.09482948801954541, "grad_norm": 4.023895740509033, "learning_rate": 1.978057638235341e-05, "loss": 0.17725872993469238, "step": 3600 }, { "epoch": 0.09509290326404415, "grad_norm": 4.106515884399414, "learning_rate": 1.9778795444112812e-05, "loss": 0.17951292991638185, "step": 3610 }, { "epoch": 0.09535631850854288, "grad_norm": 7.878188610076904, "learning_rate": 1.9777007388443e-05, "loss": 0.18134677410125732, "step": 3620 }, { "epoch": 0.09561973375304163, "grad_norm": 2.343435525894165, "learning_rate": 1.9775212216645393e-05, "loss": 0.1640856981277466, "step": 3630 }, { "epoch": 0.09588314899754036, "grad_norm": 1.7705345153808594, "learning_rate": 1.977340993002659e-05, "loss": 0.19374022483825684, "step": 3640 }, { "epoch": 0.0961465642420391, "grad_norm": 4.034841537475586, "learning_rate": 1.977160052989839e-05, "loss": 0.1891244411468506, "step": 3650 }, { "epoch": 0.09640997948653783, "grad_norm": 2.636733293533325, "learning_rate": 1.9769784017577734e-05, "loss": 0.169287371635437, "step": 3660 }, { "epoch": 0.09667339473103657, "grad_norm": 9.41122817993164, "learning_rate": 1.976796039438677e-05, "loss": 0.18921362161636351, "step": 3670 }, { "epoch": 0.0969368099755353, "grad_norm": 6.148897171020508, "learning_rate": 1.9766129661652803e-05, "loss": 0.20015082359313965, "step": 3680 }, { "epoch": 0.09720022522003405, "grad_norm": 5.165173530578613, "learning_rate": 1.9764291820708324e-05, "loss": 0.19519431591033937, "step": 3690 }, { "epoch": 0.09746364046453278, "grad_norm": 3.4715499877929688, "learning_rate": 1.9762446872890984e-05, "loss": 0.173844051361084, "step": 3700 }, { "epoch": 0.09772705570903152, "grad_norm": 3.480919361114502, "learning_rate": 1.976059481954363e-05, "loss": 0.19756587743759155, "step": 3710 }, { "epoch": 0.09799047095353026, "grad_norm": 1.643642544746399, "learning_rate": 1.9758735662014253e-05, "loss": 0.16814329624176025, "step": 3720 }, { "epoch": 0.098253886198029, "grad_norm": 3.230102777481079, "learning_rate": 1.975686940165604e-05, "loss": 0.17327139377593995, "step": 3730 }, { "epoch": 0.09851730144252774, "grad_norm": 2.7431771755218506, "learning_rate": 1.9754996039827322e-05, "loss": 0.17094879150390624, "step": 3740 }, { "epoch": 0.09878071668702647, "grad_norm": 2.4860153198242188, "learning_rate": 1.9753115577891623e-05, "loss": 0.17388330698013305, "step": 3750 }, { "epoch": 0.09904413193152521, "grad_norm": 1.9694406986236572, "learning_rate": 1.9751228017217623e-05, "loss": 0.170121431350708, "step": 3760 }, { "epoch": 0.09930754717602394, "grad_norm": 2.1289803981781006, "learning_rate": 1.9749333359179166e-05, "loss": 0.17444690465927123, "step": 3770 }, { "epoch": 0.09957096242052269, "grad_norm": 2.7408668994903564, "learning_rate": 1.974743160515527e-05, "loss": 0.17443399429321288, "step": 3780 }, { "epoch": 0.09983437766502141, "grad_norm": 3.383720636367798, "learning_rate": 1.974552275653011e-05, "loss": 0.18171422481536864, "step": 3790 }, { "epoch": 0.10009779290952016, "grad_norm": 3.2132740020751953, "learning_rate": 1.974360681469303e-05, "loss": 0.1838989734649658, "step": 3800 }, { "epoch": 0.10036120815401889, "grad_norm": 2.3213913440704346, "learning_rate": 1.9741683781038538e-05, "loss": 0.17851035594940184, "step": 3810 }, { "epoch": 0.10062462339851763, "grad_norm": 3.8238346576690674, "learning_rate": 1.9739753656966294e-05, "loss": 0.17728607654571532, "step": 3820 }, { "epoch": 0.10088803864301636, "grad_norm": 4.809046745300293, "learning_rate": 1.9737816443881134e-05, "loss": 0.18565654754638672, "step": 3830 }, { "epoch": 0.1011514538875151, "grad_norm": 3.2826762199401855, "learning_rate": 1.973587214319304e-05, "loss": 0.16768890619277954, "step": 3840 }, { "epoch": 0.10141486913201385, "grad_norm": 2.66808819770813, "learning_rate": 1.9733920756317157e-05, "loss": 0.17981516122817992, "step": 3850 }, { "epoch": 0.10167828437651258, "grad_norm": 2.2997124195098877, "learning_rate": 1.973196228467379e-05, "loss": 0.17636209726333618, "step": 3860 }, { "epoch": 0.10194169962101132, "grad_norm": 3.488969564437866, "learning_rate": 1.9729996729688403e-05, "loss": 0.18993115425109863, "step": 3870 }, { "epoch": 0.10220511486551005, "grad_norm": 2.886735200881958, "learning_rate": 1.9728024092791606e-05, "loss": 0.18792369365692138, "step": 3880 }, { "epoch": 0.1024685301100088, "grad_norm": 1.7869794368743896, "learning_rate": 1.9726044375419172e-05, "loss": 0.18258368968963623, "step": 3890 }, { "epoch": 0.10273194535450753, "grad_norm": 6.162009239196777, "learning_rate": 1.9724057579012026e-05, "loss": 0.18940904140472412, "step": 3900 }, { "epoch": 0.10299536059900627, "grad_norm": 2.3549232482910156, "learning_rate": 1.972206370501624e-05, "loss": 0.18727580308914185, "step": 3910 }, { "epoch": 0.103258775843505, "grad_norm": 3.22198748588562, "learning_rate": 1.9720062754883048e-05, "loss": 0.1852555274963379, "step": 3920 }, { "epoch": 0.10352219108800374, "grad_norm": 2.042529344558716, "learning_rate": 1.971805473006882e-05, "loss": 0.17797679901123048, "step": 3930 }, { "epoch": 0.10378560633250247, "grad_norm": 2.0486814975738525, "learning_rate": 1.9716039632035086e-05, "loss": 0.1756739139556885, "step": 3940 }, { "epoch": 0.10404902157700122, "grad_norm": 2.3551793098449707, "learning_rate": 1.9714017462248523e-05, "loss": 0.20177278518676758, "step": 3950 }, { "epoch": 0.10431243682149995, "grad_norm": 2.2095141410827637, "learning_rate": 1.9711988222180958e-05, "loss": 0.19150362014770508, "step": 3960 }, { "epoch": 0.10457585206599869, "grad_norm": 5.258943557739258, "learning_rate": 1.9709951913309347e-05, "loss": 0.1879992127418518, "step": 3970 }, { "epoch": 0.10483926731049743, "grad_norm": 2.521838665008545, "learning_rate": 1.9707908537115816e-05, "loss": 0.18166037797927856, "step": 3980 }, { "epoch": 0.10510268255499616, "grad_norm": 3.1548871994018555, "learning_rate": 1.9705858095087613e-05, "loss": 0.18472139835357665, "step": 3990 }, { "epoch": 0.1053660977994949, "grad_norm": 3.5417940616607666, "learning_rate": 1.9703800588717144e-05, "loss": 0.17024569511413573, "step": 4000 }, { "epoch": 0.10562951304399364, "grad_norm": 2.482041358947754, "learning_rate": 1.9701736019501948e-05, "loss": 0.19342424869537353, "step": 4010 }, { "epoch": 0.10589292828849238, "grad_norm": 4.9656982421875, "learning_rate": 1.969966438894471e-05, "loss": 0.17637221813201903, "step": 4020 }, { "epoch": 0.10615634353299111, "grad_norm": 1.6734966039657593, "learning_rate": 1.969758569855325e-05, "loss": 0.18694820404052734, "step": 4030 }, { "epoch": 0.10641975877748985, "grad_norm": 1.2512799501419067, "learning_rate": 1.969549994984053e-05, "loss": 0.1920862913131714, "step": 4040 }, { "epoch": 0.10668317402198858, "grad_norm": 3.693570613861084, "learning_rate": 1.9693407144324642e-05, "loss": 0.15829806327819823, "step": 4050 }, { "epoch": 0.10694658926648733, "grad_norm": 3.7436840534210205, "learning_rate": 1.9691307283528834e-05, "loss": 0.1939062714576721, "step": 4060 }, { "epoch": 0.10721000451098606, "grad_norm": 3.2761824131011963, "learning_rate": 1.968920036898146e-05, "loss": 0.18506211042404175, "step": 4070 }, { "epoch": 0.1074734197554848, "grad_norm": 2.7804830074310303, "learning_rate": 1.9687086402216033e-05, "loss": 0.17584538459777832, "step": 4080 }, { "epoch": 0.10773683499998353, "grad_norm": 1.6076340675354004, "learning_rate": 1.968496538477118e-05, "loss": 0.16897599697113036, "step": 4090 }, { "epoch": 0.10800025024448227, "grad_norm": 3.753634214401245, "learning_rate": 1.9682837318190683e-05, "loss": 0.20372614860534669, "step": 4100 }, { "epoch": 0.10826366548898102, "grad_norm": 5.629850387573242, "learning_rate": 1.968070220402343e-05, "loss": 0.1626866102218628, "step": 4110 }, { "epoch": 0.10852708073347975, "grad_norm": 1.2817552089691162, "learning_rate": 1.967856004382345e-05, "loss": 0.17655810117721557, "step": 4120 }, { "epoch": 0.10879049597797849, "grad_norm": 5.423095226287842, "learning_rate": 1.96764108391499e-05, "loss": 0.1749451994895935, "step": 4130 }, { "epoch": 0.10905391122247722, "grad_norm": 6.1288743019104, "learning_rate": 1.967425459156707e-05, "loss": 0.18257157802581786, "step": 4140 }, { "epoch": 0.10931732646697596, "grad_norm": 2.0104916095733643, "learning_rate": 1.9672091302644363e-05, "loss": 0.18118659257888795, "step": 4150 }, { "epoch": 0.1095807417114747, "grad_norm": 5.6273088455200195, "learning_rate": 1.966992097395632e-05, "loss": 0.18505115509033204, "step": 4160 }, { "epoch": 0.10984415695597344, "grad_norm": 1.6928863525390625, "learning_rate": 1.9667743607082593e-05, "loss": 0.18137009143829347, "step": 4170 }, { "epoch": 0.11010757220047217, "grad_norm": 2.9265847206115723, "learning_rate": 1.9665559203607965e-05, "loss": 0.16977577209472655, "step": 4180 }, { "epoch": 0.11037098744497091, "grad_norm": 1.433243751525879, "learning_rate": 1.9663367765122346e-05, "loss": 0.17668259143829346, "step": 4190 }, { "epoch": 0.11063440268946964, "grad_norm": 1.8410611152648926, "learning_rate": 1.966116929322075e-05, "loss": 0.1839062213897705, "step": 4200 }, { "epoch": 0.11089781793396838, "grad_norm": 2.1936771869659424, "learning_rate": 1.9658963789503328e-05, "loss": 0.17293670177459716, "step": 4210 }, { "epoch": 0.11116123317846711, "grad_norm": 2.056149959564209, "learning_rate": 1.9656751255575336e-05, "loss": 0.16012840270996093, "step": 4220 }, { "epoch": 0.11142464842296586, "grad_norm": 3.5573697090148926, "learning_rate": 1.965453169304715e-05, "loss": 0.18307063579559327, "step": 4230 }, { "epoch": 0.1116880636674646, "grad_norm": 4.347428321838379, "learning_rate": 1.9652305103534265e-05, "loss": 0.18872573375701904, "step": 4240 }, { "epoch": 0.11195147891196333, "grad_norm": 2.1119728088378906, "learning_rate": 1.9650071488657293e-05, "loss": 0.17237843275070192, "step": 4250 }, { "epoch": 0.11221489415646207, "grad_norm": 1.5902894735336304, "learning_rate": 1.964783085004195e-05, "loss": 0.18079754114151, "step": 4260 }, { "epoch": 0.1124783094009608, "grad_norm": 2.277003049850464, "learning_rate": 1.9645583189319074e-05, "loss": 0.1892380714416504, "step": 4270 }, { "epoch": 0.11274172464545955, "grad_norm": 6.822234630584717, "learning_rate": 1.9643328508124605e-05, "loss": 0.19064587354660034, "step": 4280 }, { "epoch": 0.11300513988995828, "grad_norm": 3.5220093727111816, "learning_rate": 1.9641066808099597e-05, "loss": 0.18300204277038573, "step": 4290 }, { "epoch": 0.11326855513445702, "grad_norm": 2.33072566986084, "learning_rate": 1.963879809089022e-05, "loss": 0.15880531072616577, "step": 4300 }, { "epoch": 0.11353197037895575, "grad_norm": 4.983034610748291, "learning_rate": 1.9636522358147737e-05, "loss": 0.19845268726348878, "step": 4310 }, { "epoch": 0.1137953856234545, "grad_norm": 3.5299675464630127, "learning_rate": 1.9634239611528527e-05, "loss": 0.19030361175537108, "step": 4320 }, { "epoch": 0.11405880086795323, "grad_norm": 2.0722413063049316, "learning_rate": 1.9631949852694073e-05, "loss": 0.18256970643997192, "step": 4330 }, { "epoch": 0.11432221611245197, "grad_norm": 3.865605354309082, "learning_rate": 1.9629653083310956e-05, "loss": 0.1917954683303833, "step": 4340 }, { "epoch": 0.1145856313569507, "grad_norm": 3.961416006088257, "learning_rate": 1.9627349305050867e-05, "loss": 0.19051201343536378, "step": 4350 }, { "epoch": 0.11484904660144944, "grad_norm": 3.524592638015747, "learning_rate": 1.9625038519590596e-05, "loss": 0.18392462730407716, "step": 4360 }, { "epoch": 0.11511246184594819, "grad_norm": 6.3851494789123535, "learning_rate": 1.962272072861203e-05, "loss": 0.19349901676177977, "step": 4370 }, { "epoch": 0.11537587709044692, "grad_norm": 3.854297399520874, "learning_rate": 1.9620395933802158e-05, "loss": 0.19726747274398804, "step": 4380 }, { "epoch": 0.11563929233494566, "grad_norm": 2.3233978748321533, "learning_rate": 1.9618064136853066e-05, "loss": 0.16752591133117675, "step": 4390 }, { "epoch": 0.11590270757944439, "grad_norm": 2.932774782180786, "learning_rate": 1.9615725339461936e-05, "loss": 0.1914124608039856, "step": 4400 }, { "epoch": 0.11616612282394313, "grad_norm": 3.380866765975952, "learning_rate": 1.961337954333105e-05, "loss": 0.17598845958709716, "step": 4410 }, { "epoch": 0.11642953806844186, "grad_norm": 2.3669707775115967, "learning_rate": 1.961102675016777e-05, "loss": 0.18732092380523682, "step": 4420 }, { "epoch": 0.1166929533129406, "grad_norm": 3.6404170989990234, "learning_rate": 1.960866696168457e-05, "loss": 0.1672316312789917, "step": 4430 }, { "epoch": 0.11695636855743934, "grad_norm": 3.955735445022583, "learning_rate": 1.9606300179599004e-05, "loss": 0.16967992782592772, "step": 4440 }, { "epoch": 0.11721978380193808, "grad_norm": 2.1858203411102295, "learning_rate": 1.9603926405633712e-05, "loss": 0.19269531965255737, "step": 4450 }, { "epoch": 0.11748319904643681, "grad_norm": 1.8912041187286377, "learning_rate": 1.9601545641516434e-05, "loss": 0.19165244102478027, "step": 4460 }, { "epoch": 0.11774661429093555, "grad_norm": 3.4244039058685303, "learning_rate": 1.9599157888979997e-05, "loss": 0.17427042722702027, "step": 4470 }, { "epoch": 0.11801002953543428, "grad_norm": 2.078967571258545, "learning_rate": 1.95967631497623e-05, "loss": 0.17824192047119142, "step": 4480 }, { "epoch": 0.11827344477993303, "grad_norm": 3.52314829826355, "learning_rate": 1.9594361425606348e-05, "loss": 0.17914429903030396, "step": 4490 }, { "epoch": 0.11853686002443177, "grad_norm": 2.438929319381714, "learning_rate": 1.959195271826021e-05, "loss": 0.1743268847465515, "step": 4500 }, { "epoch": 0.1188002752689305, "grad_norm": 3.4397900104522705, "learning_rate": 1.9589537029477052e-05, "loss": 0.16791825294494628, "step": 4510 }, { "epoch": 0.11906369051342924, "grad_norm": 2.0406908988952637, "learning_rate": 1.9587114361015118e-05, "loss": 0.18456268310546875, "step": 4520 }, { "epoch": 0.11932710575792797, "grad_norm": 3.681727886199951, "learning_rate": 1.958468471463773e-05, "loss": 0.1866883873939514, "step": 4530 }, { "epoch": 0.11959052100242672, "grad_norm": 2.5507326126098633, "learning_rate": 1.9582248092113286e-05, "loss": 0.18842267990112305, "step": 4540 }, { "epoch": 0.11985393624692545, "grad_norm": 1.9068635702133179, "learning_rate": 1.9579804495215268e-05, "loss": 0.16454286575317384, "step": 4550 }, { "epoch": 0.12011735149142419, "grad_norm": 2.465121030807495, "learning_rate": 1.9577353925722228e-05, "loss": 0.19311864376068116, "step": 4560 }, { "epoch": 0.12038076673592292, "grad_norm": 1.9529057741165161, "learning_rate": 1.9574896385417803e-05, "loss": 0.19175084829330444, "step": 4570 }, { "epoch": 0.12064418198042166, "grad_norm": 2.128451347351074, "learning_rate": 1.957243187609069e-05, "loss": 0.1859065532684326, "step": 4580 }, { "epoch": 0.1209075972249204, "grad_norm": 2.172701120376587, "learning_rate": 1.9569960399534668e-05, "loss": 0.17087680101394653, "step": 4590 }, { "epoch": 0.12117101246941914, "grad_norm": 3.3099329471588135, "learning_rate": 1.9567481957548585e-05, "loss": 0.18908867835998536, "step": 4600 }, { "epoch": 0.12143442771391787, "grad_norm": 4.594443321228027, "learning_rate": 1.9564996551936354e-05, "loss": 0.18507471084594726, "step": 4610 }, { "epoch": 0.12169784295841661, "grad_norm": 2.4332892894744873, "learning_rate": 1.9562504184506966e-05, "loss": 0.17927322387695313, "step": 4620 }, { "epoch": 0.12196125820291535, "grad_norm": 4.164041996002197, "learning_rate": 1.9560004857074472e-05, "loss": 0.18097437620162965, "step": 4630 }, { "epoch": 0.12222467344741408, "grad_norm": 3.2002112865448, "learning_rate": 1.955749857145799e-05, "loss": 0.1758970022201538, "step": 4640 }, { "epoch": 0.12248808869191283, "grad_norm": 4.122158527374268, "learning_rate": 1.95549853294817e-05, "loss": 0.17485592365264893, "step": 4650 }, { "epoch": 0.12275150393641156, "grad_norm": 1.2336034774780273, "learning_rate": 1.955246513297485e-05, "loss": 0.1980224609375, "step": 4660 }, { "epoch": 0.1230149191809103, "grad_norm": 1.6910487413406372, "learning_rate": 1.9549937983771748e-05, "loss": 0.18650232553482055, "step": 4670 }, { "epoch": 0.12327833442540903, "grad_norm": 7.091366291046143, "learning_rate": 1.954740388371176e-05, "loss": 0.17352170944213868, "step": 4680 }, { "epoch": 0.12354174966990777, "grad_norm": 1.8483797311782837, "learning_rate": 1.9544862834639316e-05, "loss": 0.1729900598526001, "step": 4690 }, { "epoch": 0.1238051649144065, "grad_norm": 2.0726969242095947, "learning_rate": 1.9542314838403907e-05, "loss": 0.164430832862854, "step": 4700 }, { "epoch": 0.12406858015890525, "grad_norm": 2.20929217338562, "learning_rate": 1.9539759896860064e-05, "loss": 0.18524510860443116, "step": 4710 }, { "epoch": 0.12433199540340398, "grad_norm": 3.572299003601074, "learning_rate": 1.9537198011867383e-05, "loss": 0.1891021251678467, "step": 4720 }, { "epoch": 0.12459541064790272, "grad_norm": 4.74025297164917, "learning_rate": 1.9534629185290524e-05, "loss": 0.1961202621459961, "step": 4730 }, { "epoch": 0.12485882589240146, "grad_norm": 1.8080745935440063, "learning_rate": 1.9532053418999186e-05, "loss": 0.1926150441169739, "step": 4740 }, { "epoch": 0.1251222411369002, "grad_norm": 3.039684295654297, "learning_rate": 1.952947071486812e-05, "loss": 0.19513792991638185, "step": 4750 }, { "epoch": 0.12538565638139892, "grad_norm": 4.393320560455322, "learning_rate": 1.952688107477713e-05, "loss": 0.1682377576828003, "step": 4760 }, { "epoch": 0.12564907162589767, "grad_norm": 2.145610809326172, "learning_rate": 1.9524284500611076e-05, "loss": 0.19049279689788817, "step": 4770 }, { "epoch": 0.1259124868703964, "grad_norm": 2.932485580444336, "learning_rate": 1.9521680994259845e-05, "loss": 0.1876606225967407, "step": 4780 }, { "epoch": 0.12617590211489516, "grad_norm": 2.0945847034454346, "learning_rate": 1.9519070557618386e-05, "loss": 0.1894790291786194, "step": 4790 }, { "epoch": 0.12643931735939387, "grad_norm": 1.5317434072494507, "learning_rate": 1.9516453192586686e-05, "loss": 0.1768333911895752, "step": 4800 }, { "epoch": 0.12670273260389261, "grad_norm": 1.1293529272079468, "learning_rate": 1.951382890106978e-05, "loss": 0.17407338619232177, "step": 4810 }, { "epoch": 0.12696614784839136, "grad_norm": 2.3742048740386963, "learning_rate": 1.951119768497774e-05, "loss": 0.20017702579498292, "step": 4820 }, { "epoch": 0.1272295630928901, "grad_norm": 2.4977173805236816, "learning_rate": 1.9508559546225672e-05, "loss": 0.17015371322631836, "step": 4830 }, { "epoch": 0.12749297833738882, "grad_norm": 1.1653995513916016, "learning_rate": 1.9505914486733735e-05, "loss": 0.17741246223449708, "step": 4840 }, { "epoch": 0.12775639358188756, "grad_norm": 1.6170341968536377, "learning_rate": 1.9503262508427115e-05, "loss": 0.1854646921157837, "step": 4850 }, { "epoch": 0.1280198088263863, "grad_norm": 3.2640836238861084, "learning_rate": 1.9500603613236034e-05, "loss": 0.17293426990509034, "step": 4860 }, { "epoch": 0.12828322407088505, "grad_norm": 1.4440158605575562, "learning_rate": 1.9497937803095756e-05, "loss": 0.18607325553894044, "step": 4870 }, { "epoch": 0.1285466393153838, "grad_norm": 5.041088104248047, "learning_rate": 1.9495265079946565e-05, "loss": 0.17196440696716309, "step": 4880 }, { "epoch": 0.1288100545598825, "grad_norm": 2.0358850955963135, "learning_rate": 1.9492585445733788e-05, "loss": 0.19668554067611693, "step": 4890 }, { "epoch": 0.12907346980438125, "grad_norm": 3.9136962890625, "learning_rate": 1.9489898902407783e-05, "loss": 0.18490116596221923, "step": 4900 }, { "epoch": 0.12933688504888, "grad_norm": 6.424747467041016, "learning_rate": 1.9487205451923925e-05, "loss": 0.17804936170578003, "step": 4910 }, { "epoch": 0.12960030029337874, "grad_norm": 5.476686000823975, "learning_rate": 1.9484505096242628e-05, "loss": 0.18520655632019042, "step": 4920 }, { "epoch": 0.12986371553787746, "grad_norm": 6.762352466583252, "learning_rate": 1.9481797837329323e-05, "loss": 0.18097749948501587, "step": 4930 }, { "epoch": 0.1301271307823762, "grad_norm": 2.412205457687378, "learning_rate": 1.9479083677154474e-05, "loss": 0.17379655838012695, "step": 4940 }, { "epoch": 0.13039054602687494, "grad_norm": 1.1345587968826294, "learning_rate": 1.9476362617693564e-05, "loss": 0.1814725160598755, "step": 4950 }, { "epoch": 0.1306539612713737, "grad_norm": 1.7278225421905518, "learning_rate": 1.9473634660927092e-05, "loss": 0.19065215587615966, "step": 4960 }, { "epoch": 0.1309173765158724, "grad_norm": 5.192452430725098, "learning_rate": 1.9470899808840585e-05, "loss": 0.18629525899887084, "step": 4970 }, { "epoch": 0.13118079176037115, "grad_norm": 2.7008216381073, "learning_rate": 1.9468158063424593e-05, "loss": 0.19236698150634765, "step": 4980 }, { "epoch": 0.1314442070048699, "grad_norm": 4.050326347351074, "learning_rate": 1.946540942667467e-05, "loss": 0.17149667739868163, "step": 4990 }, { "epoch": 0.13170762224936863, "grad_norm": 1.4168819189071655, "learning_rate": 1.9462653900591393e-05, "loss": 0.18426789045333863, "step": 5000 }, { "epoch": 0.13197103749386738, "grad_norm": 3.234848737716675, "learning_rate": 1.9459891487180353e-05, "loss": 0.17668113708496094, "step": 5010 }, { "epoch": 0.1322344527383661, "grad_norm": 2.032395362854004, "learning_rate": 1.9457122188452156e-05, "loss": 0.16568760871887206, "step": 5020 }, { "epoch": 0.13249786798286484, "grad_norm": 4.816248416900635, "learning_rate": 1.9454346006422414e-05, "loss": 0.18161842823028565, "step": 5030 }, { "epoch": 0.13276128322736358, "grad_norm": 1.7118101119995117, "learning_rate": 1.9451562943111753e-05, "loss": 0.19084627628326417, "step": 5040 }, { "epoch": 0.13302469847186232, "grad_norm": 3.3775229454040527, "learning_rate": 1.944877300054581e-05, "loss": 0.19304308891296387, "step": 5050 }, { "epoch": 0.13328811371636104, "grad_norm": 3.3297362327575684, "learning_rate": 1.944597618075522e-05, "loss": 0.17240257263183595, "step": 5060 }, { "epoch": 0.13355152896085978, "grad_norm": 2.4651007652282715, "learning_rate": 1.944317248577563e-05, "loss": 0.16218013763427735, "step": 5070 }, { "epoch": 0.13381494420535853, "grad_norm": 3.6757659912109375, "learning_rate": 1.9440361917647693e-05, "loss": 0.18480314016342164, "step": 5080 }, { "epoch": 0.13407835944985727, "grad_norm": 1.6834139823913574, "learning_rate": 1.943754447841706e-05, "loss": 0.17092785835266114, "step": 5090 }, { "epoch": 0.134341774694356, "grad_norm": 4.072108745574951, "learning_rate": 1.9434720170134387e-05, "loss": 0.18227226734161378, "step": 5100 }, { "epoch": 0.13460518993885473, "grad_norm": 2.910949468612671, "learning_rate": 1.9431888994855324e-05, "loss": 0.18998732566833496, "step": 5110 }, { "epoch": 0.13486860518335347, "grad_norm": 5.122820854187012, "learning_rate": 1.9429050954640528e-05, "loss": 0.17575745582580565, "step": 5120 }, { "epoch": 0.13513202042785222, "grad_norm": 5.55932092666626, "learning_rate": 1.942620605155564e-05, "loss": 0.18213706016540526, "step": 5130 }, { "epoch": 0.13539543567235096, "grad_norm": 1.321701169013977, "learning_rate": 1.9423354287671307e-05, "loss": 0.18606085777282716, "step": 5140 }, { "epoch": 0.13565885091684968, "grad_norm": 2.6069798469543457, "learning_rate": 1.9420495665063166e-05, "loss": 0.17525628805160523, "step": 5150 }, { "epoch": 0.13592226616134842, "grad_norm": 1.6079288721084595, "learning_rate": 1.9417630185811846e-05, "loss": 0.1817202687263489, "step": 5160 }, { "epoch": 0.13618568140584716, "grad_norm": 3.506030559539795, "learning_rate": 1.9414757852002962e-05, "loss": 0.1995662808418274, "step": 5170 }, { "epoch": 0.1364490966503459, "grad_norm": 3.812627077102661, "learning_rate": 1.941187866572713e-05, "loss": 0.17895548343658446, "step": 5180 }, { "epoch": 0.13671251189484462, "grad_norm": 3.1897499561309814, "learning_rate": 1.940899262907994e-05, "loss": 0.179298996925354, "step": 5190 }, { "epoch": 0.13697592713934337, "grad_norm": 1.6837998628616333, "learning_rate": 1.940609974416198e-05, "loss": 0.1811548113822937, "step": 5200 }, { "epoch": 0.1372393423838421, "grad_norm": 7.751375198364258, "learning_rate": 1.940320001307881e-05, "loss": 0.18375797271728517, "step": 5210 }, { "epoch": 0.13750275762834085, "grad_norm": 2.632138252258301, "learning_rate": 1.940029343794098e-05, "loss": 0.1856711983680725, "step": 5220 }, { "epoch": 0.13776617287283957, "grad_norm": 1.3336983919143677, "learning_rate": 1.9397380020864024e-05, "loss": 0.1822662830352783, "step": 5230 }, { "epoch": 0.13802958811733831, "grad_norm": 4.267247676849365, "learning_rate": 1.939445976396845e-05, "loss": 0.17783191204071044, "step": 5240 }, { "epoch": 0.13829300336183706, "grad_norm": 1.9450397491455078, "learning_rate": 1.939153266937975e-05, "loss": 0.18340067863464354, "step": 5250 }, { "epoch": 0.1385564186063358, "grad_norm": 2.556339979171753, "learning_rate": 1.938859873922839e-05, "loss": 0.1900884747505188, "step": 5260 }, { "epoch": 0.13881983385083455, "grad_norm": 3.035917282104492, "learning_rate": 1.9385657975649807e-05, "loss": 0.17934370040893555, "step": 5270 }, { "epoch": 0.13908324909533326, "grad_norm": 2.1907777786254883, "learning_rate": 1.9382710380784418e-05, "loss": 0.17593536376953126, "step": 5280 }, { "epoch": 0.139346664339832, "grad_norm": 2.9567837715148926, "learning_rate": 1.937975595677761e-05, "loss": 0.17602074146270752, "step": 5290 }, { "epoch": 0.13961007958433075, "grad_norm": 2.291205406188965, "learning_rate": 1.9376794705779735e-05, "loss": 0.18233461380004884, "step": 5300 }, { "epoch": 0.1398734948288295, "grad_norm": 1.8750004768371582, "learning_rate": 1.937382662994613e-05, "loss": 0.17878128290176393, "step": 5310 }, { "epoch": 0.1401369100733282, "grad_norm": 2.627429246902466, "learning_rate": 1.9370851731437077e-05, "loss": 0.1809088706970215, "step": 5320 }, { "epoch": 0.14040032531782695, "grad_norm": 5.168630599975586, "learning_rate": 1.9367870012417844e-05, "loss": 0.19205701351165771, "step": 5330 }, { "epoch": 0.1406637405623257, "grad_norm": 1.431382417678833, "learning_rate": 1.936488147505865e-05, "loss": 0.18022897243499755, "step": 5340 }, { "epoch": 0.14092715580682444, "grad_norm": 3.9417848587036133, "learning_rate": 1.9361886121534686e-05, "loss": 0.175744366645813, "step": 5350 }, { "epoch": 0.14119057105132315, "grad_norm": 1.6710342168807983, "learning_rate": 1.93588839540261e-05, "loss": 0.18413205146789552, "step": 5360 }, { "epoch": 0.1414539862958219, "grad_norm": 2.43135142326355, "learning_rate": 1.9355874974717988e-05, "loss": 0.17488625049591064, "step": 5370 }, { "epoch": 0.14171740154032064, "grad_norm": 2.362250566482544, "learning_rate": 1.9352859185800424e-05, "loss": 0.1628824234008789, "step": 5380 }, { "epoch": 0.14198081678481939, "grad_norm": 2.062446117401123, "learning_rate": 1.934983658946843e-05, "loss": 0.17712619304656982, "step": 5390 }, { "epoch": 0.14224423202931813, "grad_norm": 3.3558642864227295, "learning_rate": 1.9346807187921983e-05, "loss": 0.17560830116271972, "step": 5400 }, { "epoch": 0.14250764727381685, "grad_norm": 7.118057727813721, "learning_rate": 1.9343770983366004e-05, "loss": 0.17566461563110353, "step": 5410 }, { "epoch": 0.1427710625183156, "grad_norm": 1.6665467023849487, "learning_rate": 1.9340727978010377e-05, "loss": 0.19399229288101197, "step": 5420 }, { "epoch": 0.14303447776281433, "grad_norm": 2.087799549102783, "learning_rate": 1.933767817406994e-05, "loss": 0.18292211294174193, "step": 5430 }, { "epoch": 0.14329789300731308, "grad_norm": 3.065047264099121, "learning_rate": 1.9334621573764462e-05, "loss": 0.18212318420410156, "step": 5440 }, { "epoch": 0.1435613082518118, "grad_norm": 1.452666163444519, "learning_rate": 1.9331558179318676e-05, "loss": 0.18235863447189332, "step": 5450 }, { "epoch": 0.14382472349631054, "grad_norm": 2.658135175704956, "learning_rate": 1.9328487992962247e-05, "loss": 0.18585920333862305, "step": 5460 }, { "epoch": 0.14408813874080928, "grad_norm": 3.9550881385803223, "learning_rate": 1.9325411016929793e-05, "loss": 0.1859891414642334, "step": 5470 }, { "epoch": 0.14435155398530802, "grad_norm": 1.5532444715499878, "learning_rate": 1.9322327253460867e-05, "loss": 0.18108553886413575, "step": 5480 }, { "epoch": 0.14461496922980674, "grad_norm": 2.0241003036499023, "learning_rate": 1.931923670479997e-05, "loss": 0.17458044290542601, "step": 5490 }, { "epoch": 0.14487838447430548, "grad_norm": 0.8830637335777283, "learning_rate": 1.9316139373196533e-05, "loss": 0.18031331300735473, "step": 5500 }, { "epoch": 0.14514179971880423, "grad_norm": 3.1175448894500732, "learning_rate": 1.9313035260904923e-05, "loss": 0.1884773015975952, "step": 5510 }, { "epoch": 0.14540521496330297, "grad_norm": 1.5407496690750122, "learning_rate": 1.9309924370184455e-05, "loss": 0.17755422592163086, "step": 5520 }, { "epoch": 0.1456686302078017, "grad_norm": 3.5790276527404785, "learning_rate": 1.9306806703299367e-05, "loss": 0.18368289470672608, "step": 5530 }, { "epoch": 0.14593204545230043, "grad_norm": 2.780658721923828, "learning_rate": 1.9303682262518828e-05, "loss": 0.19319040775299073, "step": 5540 }, { "epoch": 0.14619546069679917, "grad_norm": 1.7344589233398438, "learning_rate": 1.9300551050116946e-05, "loss": 0.1779391884803772, "step": 5550 }, { "epoch": 0.14645887594129792, "grad_norm": 5.131676197052002, "learning_rate": 1.9297413068372745e-05, "loss": 0.16601235866546632, "step": 5560 }, { "epoch": 0.14672229118579666, "grad_norm": 4.0762939453125, "learning_rate": 1.9294268319570195e-05, "loss": 0.1735318660736084, "step": 5570 }, { "epoch": 0.14698570643029538, "grad_norm": 2.973914623260498, "learning_rate": 1.9291116805998168e-05, "loss": 0.19191927909851075, "step": 5580 }, { "epoch": 0.14724912167479412, "grad_norm": 1.1818104982376099, "learning_rate": 1.9287958529950477e-05, "loss": 0.17865958213806152, "step": 5590 }, { "epoch": 0.14751253691929286, "grad_norm": 3.285039186477661, "learning_rate": 1.9284793493725852e-05, "loss": 0.18028868436813356, "step": 5600 }, { "epoch": 0.1477759521637916, "grad_norm": 3.821606397628784, "learning_rate": 1.9281621699627943e-05, "loss": 0.17893388271331787, "step": 5610 }, { "epoch": 0.14803936740829032, "grad_norm": 1.3179876804351807, "learning_rate": 1.927844314996532e-05, "loss": 0.16318044662475586, "step": 5620 }, { "epoch": 0.14830278265278907, "grad_norm": 3.0638813972473145, "learning_rate": 1.9275257847051464e-05, "loss": 0.1754059910774231, "step": 5630 }, { "epoch": 0.1485661978972878, "grad_norm": 3.619817018508911, "learning_rate": 1.9272065793204778e-05, "loss": 0.1745691180229187, "step": 5640 }, { "epoch": 0.14882961314178655, "grad_norm": 5.259445667266846, "learning_rate": 1.9268866990748578e-05, "loss": 0.17275922298431395, "step": 5650 }, { "epoch": 0.1490930283862853, "grad_norm": 2.3624908924102783, "learning_rate": 1.9265661442011087e-05, "loss": 0.19458112716674805, "step": 5660 }, { "epoch": 0.149356443630784, "grad_norm": 6.161642551422119, "learning_rate": 1.9262449149325445e-05, "loss": 0.19058482646942138, "step": 5670 }, { "epoch": 0.14961985887528276, "grad_norm": 1.5710898637771606, "learning_rate": 1.92592301150297e-05, "loss": 0.17940921783447267, "step": 5680 }, { "epoch": 0.1498832741197815, "grad_norm": 1.6259984970092773, "learning_rate": 1.9256004341466795e-05, "loss": 0.17520549297332763, "step": 5690 }, { "epoch": 0.15014668936428024, "grad_norm": 3.525204658508301, "learning_rate": 1.9252771830984594e-05, "loss": 0.15869619846343994, "step": 5700 }, { "epoch": 0.15041010460877896, "grad_norm": 6.939025402069092, "learning_rate": 1.9249532585935855e-05, "loss": 0.175435209274292, "step": 5710 }, { "epoch": 0.1506735198532777, "grad_norm": 5.428125858306885, "learning_rate": 1.9246286608678243e-05, "loss": 0.17616983652114868, "step": 5720 }, { "epoch": 0.15093693509777645, "grad_norm": 3.779758930206299, "learning_rate": 1.9243033901574316e-05, "loss": 0.18240280151367189, "step": 5730 }, { "epoch": 0.1512003503422752, "grad_norm": 4.405834197998047, "learning_rate": 1.923977446699154e-05, "loss": 0.1812157154083252, "step": 5740 }, { "epoch": 0.1514637655867739, "grad_norm": 5.304318428039551, "learning_rate": 1.923650830730227e-05, "loss": 0.17232847213745117, "step": 5750 }, { "epoch": 0.15172718083127265, "grad_norm": 3.5318102836608887, "learning_rate": 1.923323542488375e-05, "loss": 0.17210733890533447, "step": 5760 }, { "epoch": 0.1519905960757714, "grad_norm": 1.6846632957458496, "learning_rate": 1.9229955822118138e-05, "loss": 0.17217986583709716, "step": 5770 }, { "epoch": 0.15225401132027014, "grad_norm": 2.3396048545837402, "learning_rate": 1.9226669501392465e-05, "loss": 0.17506622076034545, "step": 5780 }, { "epoch": 0.15251742656476888, "grad_norm": 2.4203145503997803, "learning_rate": 1.9223376465098657e-05, "loss": 0.1688312768936157, "step": 5790 }, { "epoch": 0.1527808418092676, "grad_norm": 8.27571964263916, "learning_rate": 1.9220076715633526e-05, "loss": 0.18482381105422974, "step": 5800 }, { "epoch": 0.15304425705376634, "grad_norm": 4.3753509521484375, "learning_rate": 1.921677025539877e-05, "loss": 0.179252552986145, "step": 5810 }, { "epoch": 0.15330767229826509, "grad_norm": 1.0418715476989746, "learning_rate": 1.921345708680098e-05, "loss": 0.16909897327423096, "step": 5820 }, { "epoch": 0.15357108754276383, "grad_norm": 2.559551477432251, "learning_rate": 1.921013721225162e-05, "loss": 0.1811893105506897, "step": 5830 }, { "epoch": 0.15383450278726254, "grad_norm": 3.5429282188415527, "learning_rate": 1.9206810634167036e-05, "loss": 0.16368815898895264, "step": 5840 }, { "epoch": 0.1540979180317613, "grad_norm": 1.4047377109527588, "learning_rate": 1.920347735496846e-05, "loss": 0.17367401123046874, "step": 5850 }, { "epoch": 0.15436133327626003, "grad_norm": 1.4870657920837402, "learning_rate": 1.9200137377081993e-05, "loss": 0.17575229406356813, "step": 5860 }, { "epoch": 0.15462474852075878, "grad_norm": 1.5693936347961426, "learning_rate": 1.919679070293861e-05, "loss": 0.16983580589294434, "step": 5870 }, { "epoch": 0.1548881637652575, "grad_norm": 5.10973596572876, "learning_rate": 1.9193437334974178e-05, "loss": 0.16701154708862304, "step": 5880 }, { "epoch": 0.15515157900975624, "grad_norm": 4.643651962280273, "learning_rate": 1.9190077275629408e-05, "loss": 0.17767454385757447, "step": 5890 }, { "epoch": 0.15541499425425498, "grad_norm": 1.653040885925293, "learning_rate": 1.9186710527349907e-05, "loss": 0.17311737537384034, "step": 5900 }, { "epoch": 0.15567840949875372, "grad_norm": 4.181748390197754, "learning_rate": 1.9183337092586137e-05, "loss": 0.17828924655914308, "step": 5910 }, { "epoch": 0.15594182474325247, "grad_norm": 1.6596457958221436, "learning_rate": 1.9179956973793422e-05, "loss": 0.17891929149627686, "step": 5920 }, { "epoch": 0.15620523998775118, "grad_norm": 2.0447330474853516, "learning_rate": 1.917657017343197e-05, "loss": 0.18013291358947753, "step": 5930 }, { "epoch": 0.15646865523224993, "grad_norm": 2.9528064727783203, "learning_rate": 1.9173176693966835e-05, "loss": 0.18151146173477173, "step": 5940 }, { "epoch": 0.15673207047674867, "grad_norm": 1.9334349632263184, "learning_rate": 1.9169776537867936e-05, "loss": 0.1765996813774109, "step": 5950 }, { "epoch": 0.1569954857212474, "grad_norm": 2.4763479232788086, "learning_rate": 1.9166369707610055e-05, "loss": 0.16755632162094117, "step": 5960 }, { "epoch": 0.15725890096574613, "grad_norm": 2.0141963958740234, "learning_rate": 1.9162956205672833e-05, "loss": 0.18354787826538085, "step": 5970 }, { "epoch": 0.15752231621024487, "grad_norm": 1.4612727165222168, "learning_rate": 1.9159536034540757e-05, "loss": 0.18183515071868897, "step": 5980 }, { "epoch": 0.15778573145474362, "grad_norm": 4.397353649139404, "learning_rate": 1.9156109196703177e-05, "loss": 0.18541947603225709, "step": 5990 }, { "epoch": 0.15804914669924236, "grad_norm": 1.5073760747909546, "learning_rate": 1.91526756946543e-05, "loss": 0.1833973526954651, "step": 6000 }, { "epoch": 0.15831256194374108, "grad_norm": 15.85710334777832, "learning_rate": 1.9149235530893172e-05, "loss": 2.73309326171875, "step": 6010 }, { "epoch": 0.15857597718823982, "grad_norm": 7.43599271774292, "learning_rate": 1.9145788707923692e-05, "loss": 0.3128080129623413, "step": 6020 }, { "epoch": 0.15883939243273856, "grad_norm": 4.0734357833862305, "learning_rate": 1.914233522825461e-05, "loss": 0.23382091522216797, "step": 6030 }, { "epoch": 0.1591028076772373, "grad_norm": 5.530435085296631, "learning_rate": 1.913887509439951e-05, "loss": 0.20445771217346193, "step": 6040 }, { "epoch": 0.15936622292173605, "grad_norm": 3.4625518321990967, "learning_rate": 1.9135408308876834e-05, "loss": 0.21750569343566895, "step": 6050 }, { "epoch": 0.15962963816623477, "grad_norm": 1.7544667720794678, "learning_rate": 1.913193487420986e-05, "loss": 0.19774242639541625, "step": 6060 }, { "epoch": 0.1598930534107335, "grad_norm": 1.4012534618377686, "learning_rate": 1.9128454792926694e-05, "loss": 0.20007283687591554, "step": 6070 }, { "epoch": 0.16015646865523225, "grad_norm": 4.442547798156738, "learning_rate": 1.9124968067560296e-05, "loss": 0.1902616024017334, "step": 6080 }, { "epoch": 0.160419883899731, "grad_norm": 3.077335834503174, "learning_rate": 1.9121474700648454e-05, "loss": 0.1850942611694336, "step": 6090 }, { "epoch": 0.1606832991442297, "grad_norm": 2.4999032020568848, "learning_rate": 1.9117974694733795e-05, "loss": 0.18695333003997802, "step": 6100 }, { "epoch": 0.16094671438872846, "grad_norm": 6.032928466796875, "learning_rate": 1.9114468052363764e-05, "loss": 0.1999738931655884, "step": 6110 }, { "epoch": 0.1612101296332272, "grad_norm": 2.2615878582000732, "learning_rate": 1.9110954776090656e-05, "loss": 0.2021209716796875, "step": 6120 }, { "epoch": 0.16147354487772594, "grad_norm": 2.814140558242798, "learning_rate": 1.910743486847158e-05, "loss": 0.18465156555175782, "step": 6130 }, { "epoch": 0.16173696012222466, "grad_norm": 2.4647674560546875, "learning_rate": 1.9103908332068483e-05, "loss": 0.1883750557899475, "step": 6140 }, { "epoch": 0.1620003753667234, "grad_norm": 3.700788974761963, "learning_rate": 1.910037516944812e-05, "loss": 0.16426308155059816, "step": 6150 }, { "epoch": 0.16226379061122215, "grad_norm": 3.9518325328826904, "learning_rate": 1.909683538318209e-05, "loss": 0.18233124017715455, "step": 6160 }, { "epoch": 0.1625272058557209, "grad_norm": 4.152988910675049, "learning_rate": 1.9093288975846797e-05, "loss": 0.18055331707000732, "step": 6170 }, { "epoch": 0.16279062110021963, "grad_norm": 2.189345598220825, "learning_rate": 1.9089735950023463e-05, "loss": 0.1929121732711792, "step": 6180 }, { "epoch": 0.16305403634471835, "grad_norm": 2.52712082862854, "learning_rate": 1.9086176308298147e-05, "loss": 0.17818033695220947, "step": 6190 }, { "epoch": 0.1633174515892171, "grad_norm": 9.565000534057617, "learning_rate": 1.90826100532617e-05, "loss": 0.1871973156929016, "step": 6200 }, { "epoch": 0.16358086683371584, "grad_norm": 1.7438315153121948, "learning_rate": 1.90790371875098e-05, "loss": 0.1840827465057373, "step": 6210 }, { "epoch": 0.16384428207821458, "grad_norm": 1.9141180515289307, "learning_rate": 1.9075457713642934e-05, "loss": 0.1893906831741333, "step": 6220 }, { "epoch": 0.1641076973227133, "grad_norm": 3.170661687850952, "learning_rate": 1.9071871634266392e-05, "loss": 0.17287994623184205, "step": 6230 }, { "epoch": 0.16437111256721204, "grad_norm": 1.3510822057724, "learning_rate": 1.9068278951990282e-05, "loss": 0.16872681379318238, "step": 6240 }, { "epoch": 0.16463452781171078, "grad_norm": 5.984193801879883, "learning_rate": 1.9064679669429515e-05, "loss": 0.18211286067962645, "step": 6250 }, { "epoch": 0.16489794305620953, "grad_norm": 3.0595340728759766, "learning_rate": 1.9061073789203798e-05, "loss": 0.1829797148704529, "step": 6260 }, { "epoch": 0.16516135830070824, "grad_norm": 1.4860490560531616, "learning_rate": 1.9057461313937646e-05, "loss": 0.19249308109283447, "step": 6270 }, { "epoch": 0.165424773545207, "grad_norm": 2.7337605953216553, "learning_rate": 1.905384224626038e-05, "loss": 0.1908900260925293, "step": 6280 }, { "epoch": 0.16568818878970573, "grad_norm": 5.0986714363098145, "learning_rate": 1.9050216588806113e-05, "loss": 0.17525864839553834, "step": 6290 }, { "epoch": 0.16595160403420448, "grad_norm": 5.265958786010742, "learning_rate": 1.9046584344213754e-05, "loss": 0.1703040599822998, "step": 6300 }, { "epoch": 0.16621501927870322, "grad_norm": 4.103235244750977, "learning_rate": 1.9042945515127004e-05, "loss": 0.17146971225738525, "step": 6310 }, { "epoch": 0.16647843452320193, "grad_norm": 2.9859871864318848, "learning_rate": 1.9039300104194366e-05, "loss": 0.16919984817504882, "step": 6320 }, { "epoch": 0.16674184976770068, "grad_norm": 3.3729538917541504, "learning_rate": 1.903564811406912e-05, "loss": 0.1813592791557312, "step": 6330 }, { "epoch": 0.16700526501219942, "grad_norm": 3.340883731842041, "learning_rate": 1.9031989547409347e-05, "loss": 0.16642916202545166, "step": 6340 }, { "epoch": 0.16726868025669817, "grad_norm": 1.5311027765274048, "learning_rate": 1.9028324406877905e-05, "loss": 0.161936616897583, "step": 6350 }, { "epoch": 0.16753209550119688, "grad_norm": 2.640218496322632, "learning_rate": 1.9024652695142447e-05, "loss": 0.15541021823883056, "step": 6360 }, { "epoch": 0.16779551074569563, "grad_norm": 1.8640979528427124, "learning_rate": 1.90209744148754e-05, "loss": 0.1692471981048584, "step": 6370 }, { "epoch": 0.16805892599019437, "grad_norm": 4.572014808654785, "learning_rate": 1.901728956875398e-05, "loss": 0.16244359016418458, "step": 6380 }, { "epoch": 0.1683223412346931, "grad_norm": 2.798328399658203, "learning_rate": 1.9013598159460168e-05, "loss": 0.16133053302764894, "step": 6390 }, { "epoch": 0.16858575647919183, "grad_norm": 5.52630615234375, "learning_rate": 1.900990018968074e-05, "loss": 0.1708275318145752, "step": 6400 }, { "epoch": 0.16884917172369057, "grad_norm": 2.3344638347625732, "learning_rate": 1.900619566210723e-05, "loss": 0.14985733032226561, "step": 6410 }, { "epoch": 0.16911258696818932, "grad_norm": 1.5691567659378052, "learning_rate": 1.900248457943596e-05, "loss": 0.16859116554260253, "step": 6420 }, { "epoch": 0.16937600221268806, "grad_norm": 4.644334316253662, "learning_rate": 1.899876694436801e-05, "loss": 0.17591350078582763, "step": 6430 }, { "epoch": 0.1696394174571868, "grad_norm": 2.7646586894989014, "learning_rate": 1.899504275960924e-05, "loss": 0.17829575538635253, "step": 6440 }, { "epoch": 0.16990283270168552, "grad_norm": 4.271925926208496, "learning_rate": 1.899131202787027e-05, "loss": 0.1762990713119507, "step": 6450 }, { "epoch": 0.17016624794618426, "grad_norm": 1.7724915742874146, "learning_rate": 1.8987574751866488e-05, "loss": 0.19789012670516967, "step": 6460 }, { "epoch": 0.170429663190683, "grad_norm": 4.9075798988342285, "learning_rate": 1.8983830934318044e-05, "loss": 0.18895361423492432, "step": 6470 }, { "epoch": 0.17069307843518175, "grad_norm": 2.552563190460205, "learning_rate": 1.8980080577949848e-05, "loss": 0.17543749809265136, "step": 6480 }, { "epoch": 0.17095649367968047, "grad_norm": 4.4371771812438965, "learning_rate": 1.8976323685491577e-05, "loss": 0.1848665475845337, "step": 6490 }, { "epoch": 0.1712199089241792, "grad_norm": 2.770991086959839, "learning_rate": 1.8972560259677658e-05, "loss": 0.17656024694442748, "step": 6500 }, { "epoch": 0.17148332416867795, "grad_norm": 6.787798881530762, "learning_rate": 1.896879030324727e-05, "loss": 0.17333624362945557, "step": 6510 }, { "epoch": 0.1717467394131767, "grad_norm": 2.429497003555298, "learning_rate": 1.8965013818944357e-05, "loss": 0.16790871620178222, "step": 6520 }, { "epoch": 0.1720101546576754, "grad_norm": 2.3371469974517822, "learning_rate": 1.89612308095176e-05, "loss": 0.1837393045425415, "step": 6530 }, { "epoch": 0.17227356990217416, "grad_norm": 2.5578720569610596, "learning_rate": 1.8957441277720438e-05, "loss": 0.18616821765899658, "step": 6540 }, { "epoch": 0.1725369851466729, "grad_norm": 1.411241054534912, "learning_rate": 1.8953645226311056e-05, "loss": 0.16700761318206786, "step": 6550 }, { "epoch": 0.17280040039117164, "grad_norm": 1.9491820335388184, "learning_rate": 1.894984265805239e-05, "loss": 0.1916356325149536, "step": 6560 }, { "epoch": 0.1730638156356704, "grad_norm": 2.1756982803344727, "learning_rate": 1.8946033575712104e-05, "loss": 0.16772878170013428, "step": 6570 }, { "epoch": 0.1733272308801691, "grad_norm": 2.4929039478302, "learning_rate": 1.894221798206262e-05, "loss": 0.18074071407318115, "step": 6580 }, { "epoch": 0.17359064612466785, "grad_norm": 3.8554160594940186, "learning_rate": 1.8938395879881086e-05, "loss": 0.17960437536239623, "step": 6590 }, { "epoch": 0.1738540613691666, "grad_norm": 1.8957128524780273, "learning_rate": 1.8934567271949397e-05, "loss": 0.19486334323883056, "step": 6600 }, { "epoch": 0.17411747661366533, "grad_norm": 2.746741533279419, "learning_rate": 1.8930732161054176e-05, "loss": 0.17009119987487792, "step": 6610 }, { "epoch": 0.17438089185816405, "grad_norm": 3.6750528812408447, "learning_rate": 1.8926890549986786e-05, "loss": 0.17198578119277955, "step": 6620 }, { "epoch": 0.1746443071026628, "grad_norm": 3.3586056232452393, "learning_rate": 1.8923042441543313e-05, "loss": 0.17743918895721436, "step": 6630 }, { "epoch": 0.17490772234716154, "grad_norm": 7.173677444458008, "learning_rate": 1.891918783852458e-05, "loss": 0.18204948902130128, "step": 6640 }, { "epoch": 0.17517113759166028, "grad_norm": 1.8086907863616943, "learning_rate": 1.8915326743736134e-05, "loss": 0.1589393734931946, "step": 6650 }, { "epoch": 0.175434552836159, "grad_norm": 4.871527194976807, "learning_rate": 1.8911459159988238e-05, "loss": 0.18315073251724243, "step": 6660 }, { "epoch": 0.17569796808065774, "grad_norm": 4.00826358795166, "learning_rate": 1.8907585090095897e-05, "loss": 0.17458012104034423, "step": 6670 }, { "epoch": 0.17596138332515648, "grad_norm": 2.548219680786133, "learning_rate": 1.890370453687882e-05, "loss": 0.17490650415420533, "step": 6680 }, { "epoch": 0.17622479856965523, "grad_norm": 4.547149181365967, "learning_rate": 1.889981750316144e-05, "loss": 0.19159034490585328, "step": 6690 }, { "epoch": 0.17648821381415397, "grad_norm": 4.143864154815674, "learning_rate": 1.8895923991772913e-05, "loss": 0.18115596771240233, "step": 6700 }, { "epoch": 0.1767516290586527, "grad_norm": 3.5309274196624756, "learning_rate": 1.8892024005547104e-05, "loss": 0.1863016128540039, "step": 6710 }, { "epoch": 0.17701504430315143, "grad_norm": 5.8086652755737305, "learning_rate": 1.8888117547322586e-05, "loss": 0.16894066333770752, "step": 6720 }, { "epoch": 0.17727845954765017, "grad_norm": 3.8529000282287598, "learning_rate": 1.888420461994265e-05, "loss": 0.17706985473632814, "step": 6730 }, { "epoch": 0.17754187479214892, "grad_norm": 2.7246103286743164, "learning_rate": 1.88802852262553e-05, "loss": 0.1672576665878296, "step": 6740 }, { "epoch": 0.17780529003664763, "grad_norm": 5.377591133117676, "learning_rate": 1.887635936911323e-05, "loss": 0.18482811450958253, "step": 6750 }, { "epoch": 0.17806870528114638, "grad_norm": 3.7044596672058105, "learning_rate": 1.8872427051373852e-05, "loss": 0.17983131408691405, "step": 6760 }, { "epoch": 0.17833212052564512, "grad_norm": 1.600089192390442, "learning_rate": 1.886848827589928e-05, "loss": 0.16688690185546876, "step": 6770 }, { "epoch": 0.17859553577014387, "grad_norm": 2.5465011596679688, "learning_rate": 1.8864543045556318e-05, "loss": 0.17496142387390137, "step": 6780 }, { "epoch": 0.1788589510146426, "grad_norm": 6.103038311004639, "learning_rate": 1.8860591363216482e-05, "loss": 0.16993095874786376, "step": 6790 }, { "epoch": 0.17912236625914132, "grad_norm": 2.633439302444458, "learning_rate": 1.8856633231755972e-05, "loss": 0.17429254055023194, "step": 6800 }, { "epoch": 0.17938578150364007, "grad_norm": 1.4998492002487183, "learning_rate": 1.8852668654055686e-05, "loss": 0.174491810798645, "step": 6810 }, { "epoch": 0.1796491967481388, "grad_norm": 2.408454179763794, "learning_rate": 1.884869763300122e-05, "loss": 0.17880375385284425, "step": 6820 }, { "epoch": 0.17991261199263756, "grad_norm": 3.154846429824829, "learning_rate": 1.8844720171482854e-05, "loss": 0.16761109828948975, "step": 6830 }, { "epoch": 0.18017602723713627, "grad_norm": 2.986057758331299, "learning_rate": 1.884073627239555e-05, "loss": 0.1699138641357422, "step": 6840 }, { "epoch": 0.18043944248163502, "grad_norm": 4.8546881675720215, "learning_rate": 1.8836745938638963e-05, "loss": 0.17545138597488402, "step": 6850 }, { "epoch": 0.18070285772613376, "grad_norm": 2.1798319816589355, "learning_rate": 1.8832749173117435e-05, "loss": 0.17779030799865722, "step": 6860 }, { "epoch": 0.1809662729706325, "grad_norm": 2.269099235534668, "learning_rate": 1.8828745978739982e-05, "loss": 0.175394868850708, "step": 6870 }, { "epoch": 0.18122968821513122, "grad_norm": 1.4294084310531616, "learning_rate": 1.88247363584203e-05, "loss": 0.1728590488433838, "step": 6880 }, { "epoch": 0.18149310345962996, "grad_norm": 2.9401683807373047, "learning_rate": 1.8820720315076767e-05, "loss": 0.1677929401397705, "step": 6890 }, { "epoch": 0.1817565187041287, "grad_norm": 3.772951364517212, "learning_rate": 1.8816697851632425e-05, "loss": 0.18112549781799317, "step": 6900 }, { "epoch": 0.18201993394862745, "grad_norm": 2.5563395023345947, "learning_rate": 1.8812668971015006e-05, "loss": 0.1713331937789917, "step": 6910 }, { "epoch": 0.1822833491931262, "grad_norm": 1.5148910284042358, "learning_rate": 1.88086336761569e-05, "loss": 0.17558641433715821, "step": 6920 }, { "epoch": 0.1825467644376249, "grad_norm": 1.4694788455963135, "learning_rate": 1.8804591969995167e-05, "loss": 0.17224881649017335, "step": 6930 }, { "epoch": 0.18281017968212365, "grad_norm": 3.0944385528564453, "learning_rate": 1.8800543855471535e-05, "loss": 0.1688995838165283, "step": 6940 }, { "epoch": 0.1830735949266224, "grad_norm": 3.5399391651153564, "learning_rate": 1.87964893355324e-05, "loss": 0.16903927326202392, "step": 6950 }, { "epoch": 0.18333701017112114, "grad_norm": 3.38930344581604, "learning_rate": 1.879242841312881e-05, "loss": 0.17393819093704224, "step": 6960 }, { "epoch": 0.18360042541561986, "grad_norm": 5.651376247406006, "learning_rate": 1.878836109121648e-05, "loss": 0.1719711422920227, "step": 6970 }, { "epoch": 0.1838638406601186, "grad_norm": 2.3674938678741455, "learning_rate": 1.8784287372755795e-05, "loss": 0.192412805557251, "step": 6980 }, { "epoch": 0.18412725590461734, "grad_norm": 4.673818111419678, "learning_rate": 1.8780207260711767e-05, "loss": 0.1824009656906128, "step": 6990 }, { "epoch": 0.1843906711491161, "grad_norm": 2.471508264541626, "learning_rate": 1.8776120758054087e-05, "loss": 0.17918541431427001, "step": 7000 }, { "epoch": 0.1846540863936148, "grad_norm": 4.438976764678955, "learning_rate": 1.8772027867757082e-05, "loss": 0.16786428689956664, "step": 7010 }, { "epoch": 0.18491750163811355, "grad_norm": 4.130398750305176, "learning_rate": 1.8767928592799742e-05, "loss": 0.1730055332183838, "step": 7020 }, { "epoch": 0.1851809168826123, "grad_norm": 3.1959588527679443, "learning_rate": 1.876382293616569e-05, "loss": 0.18155776262283324, "step": 7030 }, { "epoch": 0.18544433212711103, "grad_norm": 2.622697591781616, "learning_rate": 1.8759710900843202e-05, "loss": 0.17819459438323976, "step": 7040 }, { "epoch": 0.18570774737160978, "grad_norm": 2.2626750469207764, "learning_rate": 1.8755592489825193e-05, "loss": 0.1621330738067627, "step": 7050 }, { "epoch": 0.1859711626161085, "grad_norm": 1.7582013607025146, "learning_rate": 1.8751467706109224e-05, "loss": 0.18982861042022706, "step": 7060 }, { "epoch": 0.18623457786060724, "grad_norm": 4.6863555908203125, "learning_rate": 1.8747336552697483e-05, "loss": 0.1751790761947632, "step": 7070 }, { "epoch": 0.18649799310510598, "grad_norm": 1.3928864002227783, "learning_rate": 1.874319903259681e-05, "loss": 0.1633940100669861, "step": 7080 }, { "epoch": 0.18676140834960472, "grad_norm": 1.746917724609375, "learning_rate": 1.873905514881867e-05, "loss": 0.18057374954223632, "step": 7090 }, { "epoch": 0.18702482359410344, "grad_norm": 5.285078048706055, "learning_rate": 1.873490490437915e-05, "loss": 0.17007777690887452, "step": 7100 }, { "epoch": 0.18728823883860218, "grad_norm": 4.987746238708496, "learning_rate": 1.8730748302298993e-05, "loss": 0.17654755115509033, "step": 7110 }, { "epoch": 0.18755165408310093, "grad_norm": 4.0062055587768555, "learning_rate": 1.8726585345603544e-05, "loss": 0.20309696197509766, "step": 7120 }, { "epoch": 0.18781506932759967, "grad_norm": 2.2442538738250732, "learning_rate": 1.8722416037322778e-05, "loss": 0.18031530380249022, "step": 7130 }, { "epoch": 0.1880784845720984, "grad_norm": 1.814377784729004, "learning_rate": 1.871824038049131e-05, "loss": 0.18988701105117797, "step": 7140 }, { "epoch": 0.18834189981659713, "grad_norm": 3.286808967590332, "learning_rate": 1.871405837814835e-05, "loss": 0.16753276586532592, "step": 7150 }, { "epoch": 0.18860531506109587, "grad_norm": 1.957589864730835, "learning_rate": 1.8709870033337753e-05, "loss": 0.18962515592575074, "step": 7160 }, { "epoch": 0.18886873030559462, "grad_norm": 5.090994834899902, "learning_rate": 1.8705675349107964e-05, "loss": 0.17549383640289307, "step": 7170 }, { "epoch": 0.18913214555009336, "grad_norm": 2.5738751888275146, "learning_rate": 1.870147432851206e-05, "loss": 0.17000937461853027, "step": 7180 }, { "epoch": 0.18939556079459208, "grad_norm": 2.5103890895843506, "learning_rate": 1.8697266974607732e-05, "loss": 0.1742405414581299, "step": 7190 }, { "epoch": 0.18965897603909082, "grad_norm": 3.9350650310516357, "learning_rate": 1.8693053290457264e-05, "loss": 0.16105775833129882, "step": 7200 }, { "epoch": 0.18992239128358956, "grad_norm": 1.8911657333374023, "learning_rate": 1.868883327912756e-05, "loss": 0.17262306213378906, "step": 7210 }, { "epoch": 0.1901858065280883, "grad_norm": 1.4389026165008545, "learning_rate": 1.868460694369013e-05, "loss": 0.1833512306213379, "step": 7220 }, { "epoch": 0.19044922177258702, "grad_norm": 2.501546859741211, "learning_rate": 1.868037428722108e-05, "loss": 0.17171536684036254, "step": 7230 }, { "epoch": 0.19071263701708577, "grad_norm": 2.831418037414551, "learning_rate": 1.867613531280112e-05, "loss": 0.1714720606803894, "step": 7240 }, { "epoch": 0.1909760522615845, "grad_norm": 2.1661858558654785, "learning_rate": 1.867189002351556e-05, "loss": 0.16609654426574708, "step": 7250 }, { "epoch": 0.19123946750608325, "grad_norm": 4.480917930603027, "learning_rate": 1.8667638422454306e-05, "loss": 0.17936902046203612, "step": 7260 }, { "epoch": 0.19150288275058197, "grad_norm": 2.491471529006958, "learning_rate": 1.8663380512711852e-05, "loss": 0.1787546157836914, "step": 7270 }, { "epoch": 0.19176629799508071, "grad_norm": 1.5245507955551147, "learning_rate": 1.865911629738729e-05, "loss": 0.16974191665649413, "step": 7280 }, { "epoch": 0.19202971323957946, "grad_norm": 2.386057138442993, "learning_rate": 1.86548457795843e-05, "loss": 0.18117134571075438, "step": 7290 }, { "epoch": 0.1922931284840782, "grad_norm": 2.3506927490234375, "learning_rate": 1.8650568962411154e-05, "loss": 0.17339119911193848, "step": 7300 }, { "epoch": 0.19255654372857695, "grad_norm": 1.438976526260376, "learning_rate": 1.8646285848980693e-05, "loss": 0.16731410026550292, "step": 7310 }, { "epoch": 0.19281995897307566, "grad_norm": 6.182242393493652, "learning_rate": 1.8641996442410365e-05, "loss": 0.15706915855407716, "step": 7320 }, { "epoch": 0.1930833742175744, "grad_norm": 1.6648197174072266, "learning_rate": 1.8637700745822176e-05, "loss": 0.1810060977935791, "step": 7330 }, { "epoch": 0.19334678946207315, "grad_norm": 1.2559313774108887, "learning_rate": 1.8633398762342722e-05, "loss": 0.17461568117141724, "step": 7340 }, { "epoch": 0.1936102047065719, "grad_norm": 3.20802903175354, "learning_rate": 1.8629090495103175e-05, "loss": 0.17328842878341674, "step": 7350 }, { "epoch": 0.1938736199510706, "grad_norm": 3.3574395179748535, "learning_rate": 1.8624775947239272e-05, "loss": 0.15186117887496947, "step": 7360 }, { "epoch": 0.19413703519556935, "grad_norm": 3.83345890045166, "learning_rate": 1.862045512189133e-05, "loss": 0.17403628826141357, "step": 7370 }, { "epoch": 0.1944004504400681, "grad_norm": 2.723224639892578, "learning_rate": 1.8616128022204233e-05, "loss": 0.1778666138648987, "step": 7380 }, { "epoch": 0.19466386568456684, "grad_norm": 1.6794661283493042, "learning_rate": 1.8611794651327432e-05, "loss": 0.17614469528198243, "step": 7390 }, { "epoch": 0.19492728092906556, "grad_norm": 1.4335960149765015, "learning_rate": 1.8607455012414936e-05, "loss": 0.16793832778930665, "step": 7400 }, { "epoch": 0.1951906961735643, "grad_norm": 1.5740610361099243, "learning_rate": 1.8603109108625327e-05, "loss": 0.1767776846885681, "step": 7410 }, { "epoch": 0.19545411141806304, "grad_norm": 3.663703441619873, "learning_rate": 1.859875694312174e-05, "loss": 0.1629639148712158, "step": 7420 }, { "epoch": 0.19571752666256179, "grad_norm": 1.4307992458343506, "learning_rate": 1.8594398519071867e-05, "loss": 0.17678825855255126, "step": 7430 }, { "epoch": 0.19598094190706053, "grad_norm": 1.7223546504974365, "learning_rate": 1.8590033839647955e-05, "loss": 0.18179872035980224, "step": 7440 }, { "epoch": 0.19624435715155925, "grad_norm": 1.0513262748718262, "learning_rate": 1.8585662908026814e-05, "loss": 0.1646634817123413, "step": 7450 }, { "epoch": 0.196507772396058, "grad_norm": 1.8444241285324097, "learning_rate": 1.8581285727389787e-05, "loss": 0.17335801124572753, "step": 7460 }, { "epoch": 0.19677118764055673, "grad_norm": 5.035979270935059, "learning_rate": 1.857690230092278e-05, "loss": 0.18263535499572753, "step": 7470 }, { "epoch": 0.19703460288505548, "grad_norm": 3.11230731010437, "learning_rate": 1.8572512631816243e-05, "loss": 0.17051244974136354, "step": 7480 }, { "epoch": 0.1972980181295542, "grad_norm": 1.2927052974700928, "learning_rate": 1.856811672326516e-05, "loss": 0.16075198650360106, "step": 7490 }, { "epoch": 0.19756143337405294, "grad_norm": 2.848365068435669, "learning_rate": 1.8563714578469064e-05, "loss": 0.18449366092681885, "step": 7500 }, { "epoch": 0.19782484861855168, "grad_norm": 5.049249172210693, "learning_rate": 1.8559306200632036e-05, "loss": 0.1840359926223755, "step": 7510 }, { "epoch": 0.19808826386305042, "grad_norm": 2.898898124694824, "learning_rate": 1.8554891592962666e-05, "loss": 0.16484248638153076, "step": 7520 }, { "epoch": 0.19835167910754914, "grad_norm": 1.533366322517395, "learning_rate": 1.8550470758674112e-05, "loss": 0.1708270788192749, "step": 7530 }, { "epoch": 0.19861509435204788, "grad_norm": 2.408667802810669, "learning_rate": 1.8546043700984036e-05, "loss": 0.1763487696647644, "step": 7540 }, { "epoch": 0.19887850959654663, "grad_norm": 2.029545307159424, "learning_rate": 1.8541610423114647e-05, "loss": 0.18353618383407594, "step": 7550 }, { "epoch": 0.19914192484104537, "grad_norm": 2.1936285495758057, "learning_rate": 1.853717092829268e-05, "loss": 0.18035943508148194, "step": 7560 }, { "epoch": 0.1994053400855441, "grad_norm": 2.3960301876068115, "learning_rate": 1.853272521974938e-05, "loss": 0.17906506061553956, "step": 7570 }, { "epoch": 0.19966875533004283, "grad_norm": 1.7327014207839966, "learning_rate": 1.8528273300720538e-05, "loss": 0.1650938868522644, "step": 7580 }, { "epoch": 0.19993217057454157, "grad_norm": 1.8246501684188843, "learning_rate": 1.8523815174446444e-05, "loss": 0.16900116205215454, "step": 7590 }, { "epoch": 0.20019558581904032, "grad_norm": 1.2938809394836426, "learning_rate": 1.851935084417192e-05, "loss": 0.16768531799316405, "step": 7600 }, { "epoch": 0.20045900106353906, "grad_norm": 1.4242558479309082, "learning_rate": 1.8514880313146295e-05, "loss": 0.18860580921173095, "step": 7610 }, { "epoch": 0.20072241630803778, "grad_norm": 2.571415424346924, "learning_rate": 1.8510403584623414e-05, "loss": 0.17398475408554076, "step": 7620 }, { "epoch": 0.20098583155253652, "grad_norm": 5.352071285247803, "learning_rate": 1.850592066186163e-05, "loss": 0.16258074045181276, "step": 7630 }, { "epoch": 0.20124924679703526, "grad_norm": 2.9876816272735596, "learning_rate": 1.850143154812382e-05, "loss": 0.16750212907791137, "step": 7640 }, { "epoch": 0.201512662041534, "grad_norm": 3.1431078910827637, "learning_rate": 1.849693624667734e-05, "loss": 0.16658368110656738, "step": 7650 }, { "epoch": 0.20177607728603272, "grad_norm": 1.900048017501831, "learning_rate": 1.8492434760794072e-05, "loss": 0.1760498285293579, "step": 7660 }, { "epoch": 0.20203949253053147, "grad_norm": 1.627996563911438, "learning_rate": 1.848792709375039e-05, "loss": 0.16794265508651735, "step": 7670 }, { "epoch": 0.2023029077750302, "grad_norm": 1.3785876035690308, "learning_rate": 1.848341324882717e-05, "loss": 0.1574561834335327, "step": 7680 }, { "epoch": 0.20256632301952895, "grad_norm": 1.131449818611145, "learning_rate": 1.8478893229309784e-05, "loss": 0.18249088525772095, "step": 7690 }, { "epoch": 0.2028297382640277, "grad_norm": 1.5123895406723022, "learning_rate": 1.847436703848809e-05, "loss": 0.17462401390075682, "step": 7700 }, { "epoch": 0.20309315350852641, "grad_norm": 4.292091369628906, "learning_rate": 1.8469834679656452e-05, "loss": 0.1828974962234497, "step": 7710 }, { "epoch": 0.20335656875302516, "grad_norm": 2.9433798789978027, "learning_rate": 1.8465296156113715e-05, "loss": 0.17362685203552247, "step": 7720 }, { "epoch": 0.2036199839975239, "grad_norm": 2.4305474758148193, "learning_rate": 1.8460751471163212e-05, "loss": 0.18315958976745605, "step": 7730 }, { "epoch": 0.20388339924202264, "grad_norm": 2.923457384109497, "learning_rate": 1.845620062811276e-05, "loss": 0.1834866762161255, "step": 7740 }, { "epoch": 0.20414681448652136, "grad_norm": 3.3196065425872803, "learning_rate": 1.8451643630274664e-05, "loss": 0.1665391206741333, "step": 7750 }, { "epoch": 0.2044102297310201, "grad_norm": 2.0016956329345703, "learning_rate": 1.8447080480965694e-05, "loss": 0.17829010486602784, "step": 7760 }, { "epoch": 0.20467364497551885, "grad_norm": 2.993433713912964, "learning_rate": 1.844251118350711e-05, "loss": 0.1676165819168091, "step": 7770 }, { "epoch": 0.2049370602200176, "grad_norm": 2.1966440677642822, "learning_rate": 1.8437935741224652e-05, "loss": 0.1835112452507019, "step": 7780 }, { "epoch": 0.2052004754645163, "grad_norm": 1.7451486587524414, "learning_rate": 1.8433354157448513e-05, "loss": 0.17223832607269288, "step": 7790 }, { "epoch": 0.20546389070901505, "grad_norm": 5.763281345367432, "learning_rate": 1.8428766435513373e-05, "loss": 0.17887030839920043, "step": 7800 }, { "epoch": 0.2057273059535138, "grad_norm": 5.842153072357178, "learning_rate": 1.8424172578758374e-05, "loss": 0.17423752546310425, "step": 7810 }, { "epoch": 0.20599072119801254, "grad_norm": 3.5426957607269287, "learning_rate": 1.841957259052712e-05, "loss": 0.16894991397857667, "step": 7820 }, { "epoch": 0.20625413644251128, "grad_norm": 1.6789965629577637, "learning_rate": 1.8414966474167688e-05, "loss": 0.18286067247390747, "step": 7830 }, { "epoch": 0.20651755168701, "grad_norm": 6.365306377410889, "learning_rate": 1.8410354233032604e-05, "loss": 0.16613813638687133, "step": 7840 }, { "epoch": 0.20678096693150874, "grad_norm": 1.5800989866256714, "learning_rate": 1.840573587047885e-05, "loss": 0.17427504062652588, "step": 7850 }, { "epoch": 0.20704438217600749, "grad_norm": 1.9295448064804077, "learning_rate": 1.8401111389867882e-05, "loss": 0.18362643718719482, "step": 7860 }, { "epoch": 0.20730779742050623, "grad_norm": 3.9158241748809814, "learning_rate": 1.8396480794565583e-05, "loss": 0.17236547470092772, "step": 7870 }, { "epoch": 0.20757121266500494, "grad_norm": 2.6407511234283447, "learning_rate": 1.839184408794231e-05, "loss": 0.17401416301727296, "step": 7880 }, { "epoch": 0.2078346279095037, "grad_norm": 1.447827935218811, "learning_rate": 1.8387201273372855e-05, "loss": 0.1701011538505554, "step": 7890 }, { "epoch": 0.20809804315400243, "grad_norm": 5.494963645935059, "learning_rate": 1.838255235423646e-05, "loss": 0.1613600492477417, "step": 7900 }, { "epoch": 0.20836145839850118, "grad_norm": 2.0205307006835938, "learning_rate": 1.8377897333916808e-05, "loss": 0.17053998708724977, "step": 7910 }, { "epoch": 0.2086248736429999, "grad_norm": 5.545360088348389, "learning_rate": 1.837323621580202e-05, "loss": 0.16154714822769164, "step": 7920 }, { "epoch": 0.20888828888749864, "grad_norm": 4.0520734786987305, "learning_rate": 1.8368569003284662e-05, "loss": 0.17192496061325074, "step": 7930 }, { "epoch": 0.20915170413199738, "grad_norm": 2.412184238433838, "learning_rate": 1.8363895699761736e-05, "loss": 0.17406964302062988, "step": 7940 }, { "epoch": 0.20941511937649612, "grad_norm": 6.206525802612305, "learning_rate": 1.8359216308634668e-05, "loss": 0.16669867038726807, "step": 7950 }, { "epoch": 0.20967853462099487, "grad_norm": 1.248530387878418, "learning_rate": 1.8354530833309323e-05, "loss": 0.17784507274627687, "step": 7960 }, { "epoch": 0.20994194986549358, "grad_norm": 3.614555597305298, "learning_rate": 1.8349839277195993e-05, "loss": 0.1587599515914917, "step": 7970 }, { "epoch": 0.21020536510999233, "grad_norm": 1.478338599205017, "learning_rate": 1.8345141643709394e-05, "loss": 0.17357242107391357, "step": 7980 }, { "epoch": 0.21046878035449107, "grad_norm": 3.6460981369018555, "learning_rate": 1.8340437936268667e-05, "loss": 0.1725849151611328, "step": 7990 }, { "epoch": 0.2107321955989898, "grad_norm": 3.333696126937866, "learning_rate": 1.8335728158297374e-05, "loss": 0.17631611824035645, "step": 8000 }, { "epoch": 0.21099561084348853, "grad_norm": 1.2146270275115967, "learning_rate": 1.833101231322349e-05, "loss": 0.17607378959655762, "step": 8010 }, { "epoch": 0.21125902608798727, "grad_norm": 4.999386310577393, "learning_rate": 1.832629040447942e-05, "loss": 0.1798863887786865, "step": 8020 }, { "epoch": 0.21152244133248602, "grad_norm": 2.1861648559570312, "learning_rate": 1.8321562435501963e-05, "loss": 0.18826262950897216, "step": 8030 }, { "epoch": 0.21178585657698476, "grad_norm": 2.71037220954895, "learning_rate": 1.831682840973234e-05, "loss": 0.17238330841064453, "step": 8040 }, { "epoch": 0.21204927182148348, "grad_norm": 3.219214916229248, "learning_rate": 1.8312088330616188e-05, "loss": 0.1789308547973633, "step": 8050 }, { "epoch": 0.21231268706598222, "grad_norm": 1.6196449995040894, "learning_rate": 1.8307342201603533e-05, "loss": 0.1595104455947876, "step": 8060 }, { "epoch": 0.21257610231048096, "grad_norm": 2.8373658657073975, "learning_rate": 1.830259002614881e-05, "loss": 0.15425562858581543, "step": 8070 }, { "epoch": 0.2128395175549797, "grad_norm": 1.432456374168396, "learning_rate": 1.829783180771087e-05, "loss": 0.17822372913360596, "step": 8080 }, { "epoch": 0.21310293279947845, "grad_norm": 1.1693257093429565, "learning_rate": 1.8293067549752937e-05, "loss": 0.17488741874694824, "step": 8090 }, { "epoch": 0.21336634804397717, "grad_norm": 1.554022192955017, "learning_rate": 1.8288297255742646e-05, "loss": 0.18957319259643554, "step": 8100 }, { "epoch": 0.2136297632884759, "grad_norm": 2.518721103668213, "learning_rate": 1.828352092915203e-05, "loss": 0.1735781192779541, "step": 8110 }, { "epoch": 0.21389317853297465, "grad_norm": 1.450464129447937, "learning_rate": 1.8278738573457496e-05, "loss": 0.15429649353027344, "step": 8120 }, { "epoch": 0.2141565937774734, "grad_norm": 1.526811957359314, "learning_rate": 1.8273950192139854e-05, "loss": 0.173258638381958, "step": 8130 }, { "epoch": 0.2144200090219721, "grad_norm": 3.974033832550049, "learning_rate": 1.826915578868429e-05, "loss": 0.18754026889801026, "step": 8140 }, { "epoch": 0.21468342426647086, "grad_norm": 3.6525332927703857, "learning_rate": 1.8264355366580387e-05, "loss": 0.16819357872009277, "step": 8150 }, { "epoch": 0.2149468395109696, "grad_norm": 1.3311904668807983, "learning_rate": 1.825954892932209e-05, "loss": 0.16043691635131835, "step": 8160 }, { "epoch": 0.21521025475546834, "grad_norm": 3.295257806777954, "learning_rate": 1.8254736480407738e-05, "loss": 0.16700499057769774, "step": 8170 }, { "epoch": 0.21547366999996706, "grad_norm": 2.3726046085357666, "learning_rate": 1.8249918023340036e-05, "loss": 0.16461658477783203, "step": 8180 }, { "epoch": 0.2157370852444658, "grad_norm": 3.6377429962158203, "learning_rate": 1.8245093561626063e-05, "loss": 0.1750023365020752, "step": 8190 }, { "epoch": 0.21600050048896455, "grad_norm": 5.535947799682617, "learning_rate": 1.824026309877727e-05, "loss": 0.1887821912765503, "step": 8200 }, { "epoch": 0.2162639157334633, "grad_norm": 2.885030508041382, "learning_rate": 1.8235426638309483e-05, "loss": 0.17903099060058594, "step": 8210 }, { "epoch": 0.21652733097796203, "grad_norm": 2.346409559249878, "learning_rate": 1.823058418374288e-05, "loss": 0.17671188116073608, "step": 8220 }, { "epoch": 0.21679074622246075, "grad_norm": 2.338432788848877, "learning_rate": 1.822573573860201e-05, "loss": 0.16042206287384034, "step": 8230 }, { "epoch": 0.2170541614669595, "grad_norm": 1.2075117826461792, "learning_rate": 1.8220881306415783e-05, "loss": 0.16313039064407348, "step": 8240 }, { "epoch": 0.21731757671145824, "grad_norm": 2.001793622970581, "learning_rate": 1.821602089071746e-05, "loss": 0.1662074565887451, "step": 8250 }, { "epoch": 0.21758099195595698, "grad_norm": 1.694093942642212, "learning_rate": 1.8211154495044666e-05, "loss": 0.18088335990905763, "step": 8260 }, { "epoch": 0.2178444072004557, "grad_norm": 2.4892454147338867, "learning_rate": 1.8206282122939368e-05, "loss": 0.1635227918624878, "step": 8270 }, { "epoch": 0.21810782244495444, "grad_norm": 1.3374106884002686, "learning_rate": 1.8201403777947896e-05, "loss": 0.16733468770980836, "step": 8280 }, { "epoch": 0.21837123768945318, "grad_norm": 1.7170699834823608, "learning_rate": 1.8196519463620912e-05, "loss": 0.17629702091217042, "step": 8290 }, { "epoch": 0.21863465293395193, "grad_norm": 1.847398042678833, "learning_rate": 1.8191629183513435e-05, "loss": 0.1610762357711792, "step": 8300 }, { "epoch": 0.21889806817845064, "grad_norm": 1.6706644296646118, "learning_rate": 1.8186732941184824e-05, "loss": 0.1740634799003601, "step": 8310 }, { "epoch": 0.2191614834229494, "grad_norm": 0.9241598844528198, "learning_rate": 1.818183074019877e-05, "loss": 0.16702831983566285, "step": 8320 }, { "epoch": 0.21942489866744813, "grad_norm": 3.33933424949646, "learning_rate": 1.8176922584123308e-05, "loss": 0.16513785123825073, "step": 8330 }, { "epoch": 0.21968831391194688, "grad_norm": 4.204290866851807, "learning_rate": 1.817200847653081e-05, "loss": 0.19111297130584717, "step": 8340 }, { "epoch": 0.21995172915644562, "grad_norm": 1.2349704504013062, "learning_rate": 1.8167088420997975e-05, "loss": 0.1879722833633423, "step": 8350 }, { "epoch": 0.22021514440094433, "grad_norm": 2.2318332195281982, "learning_rate": 1.816216242110582e-05, "loss": 0.1755699872970581, "step": 8360 }, { "epoch": 0.22047855964544308, "grad_norm": 2.313042163848877, "learning_rate": 1.815723048043972e-05, "loss": 0.15785808563232423, "step": 8370 }, { "epoch": 0.22074197488994182, "grad_norm": 2.0751304626464844, "learning_rate": 1.8152292602589336e-05, "loss": 0.17781322002410888, "step": 8380 }, { "epoch": 0.22100539013444057, "grad_norm": 1.304282784461975, "learning_rate": 1.8147348791148678e-05, "loss": 0.18703110218048097, "step": 8390 }, { "epoch": 0.22126880537893928, "grad_norm": 1.4592136144638062, "learning_rate": 1.8142399049716065e-05, "loss": 0.1662621021270752, "step": 8400 }, { "epoch": 0.22153222062343803, "grad_norm": 2.1981868743896484, "learning_rate": 1.8137443381894126e-05, "loss": 0.19285778999328612, "step": 8410 }, { "epoch": 0.22179563586793677, "grad_norm": 1.6157572269439697, "learning_rate": 1.813248179128982e-05, "loss": 0.16745699644088746, "step": 8420 }, { "epoch": 0.2220590511124355, "grad_norm": 2.206062078475952, "learning_rate": 1.81275142815144e-05, "loss": 0.18395360708236694, "step": 8430 }, { "epoch": 0.22232246635693423, "grad_norm": 2.3403255939483643, "learning_rate": 1.812254085618343e-05, "loss": 0.18276693820953369, "step": 8440 }, { "epoch": 0.22258588160143297, "grad_norm": 3.7015748023986816, "learning_rate": 1.8117561518916792e-05, "loss": 0.1732175827026367, "step": 8450 }, { "epoch": 0.22284929684593172, "grad_norm": 1.7182708978652954, "learning_rate": 1.811257627333866e-05, "loss": 0.16934211254119874, "step": 8460 }, { "epoch": 0.22311271209043046, "grad_norm": 0.8623030185699463, "learning_rate": 1.810758512307751e-05, "loss": 0.17307037115097046, "step": 8470 }, { "epoch": 0.2233761273349292, "grad_norm": 3.8190724849700928, "learning_rate": 1.8102588071766116e-05, "loss": 0.17566931247711182, "step": 8480 }, { "epoch": 0.22363954257942792, "grad_norm": 2.8250210285186768, "learning_rate": 1.8097585123041547e-05, "loss": 0.17042367458343505, "step": 8490 }, { "epoch": 0.22390295782392666, "grad_norm": 4.785216331481934, "learning_rate": 1.809257628054517e-05, "loss": 0.16935018301010132, "step": 8500 }, { "epoch": 0.2241663730684254, "grad_norm": 2.6920666694641113, "learning_rate": 1.808756154792263e-05, "loss": 0.17528994083404542, "step": 8510 }, { "epoch": 0.22442978831292415, "grad_norm": 3.5736098289489746, "learning_rate": 1.8082540928823866e-05, "loss": 0.18016383647918702, "step": 8520 }, { "epoch": 0.22469320355742287, "grad_norm": 1.243861436843872, "learning_rate": 1.8077514426903107e-05, "loss": 0.17992398738861085, "step": 8530 }, { "epoch": 0.2249566188019216, "grad_norm": 3.111149549484253, "learning_rate": 1.8072482045818852e-05, "loss": 0.16529126167297364, "step": 8540 }, { "epoch": 0.22522003404642035, "grad_norm": 2.487227439880371, "learning_rate": 1.806744378923389e-05, "loss": 0.17820796966552735, "step": 8550 }, { "epoch": 0.2254834492909191, "grad_norm": 2.515232563018799, "learning_rate": 1.8062399660815277e-05, "loss": 0.17742340564727782, "step": 8560 }, { "epoch": 0.2257468645354178, "grad_norm": 1.5604991912841797, "learning_rate": 1.8057349664234348e-05, "loss": 0.15590879917144776, "step": 8570 }, { "epoch": 0.22601027977991656, "grad_norm": 1.6328636407852173, "learning_rate": 1.8052293803166714e-05, "loss": 0.1809088706970215, "step": 8580 }, { "epoch": 0.2262736950244153, "grad_norm": 2.2534232139587402, "learning_rate": 1.8047232081292242e-05, "loss": 0.16983637809753419, "step": 8590 }, { "epoch": 0.22653711026891404, "grad_norm": 2.6252787113189697, "learning_rate": 1.8042164502295074e-05, "loss": 0.18439631462097167, "step": 8600 }, { "epoch": 0.2268005255134128, "grad_norm": 0.9206722974777222, "learning_rate": 1.8037091069863615e-05, "loss": 0.16512007713317872, "step": 8610 }, { "epoch": 0.2270639407579115, "grad_norm": 1.97080659866333, "learning_rate": 1.8032011787690523e-05, "loss": 0.16615110635757446, "step": 8620 }, { "epoch": 0.22732735600241025, "grad_norm": 3.32802152633667, "learning_rate": 1.8026926659472727e-05, "loss": 0.16637130975723266, "step": 8630 }, { "epoch": 0.227590771246909, "grad_norm": 2.158510208129883, "learning_rate": 1.802183568891139e-05, "loss": 0.16700482368469238, "step": 8640 }, { "epoch": 0.22785418649140773, "grad_norm": 3.2146835327148438, "learning_rate": 1.801673887971195e-05, "loss": 0.1850825309753418, "step": 8650 }, { "epoch": 0.22811760173590645, "grad_norm": 1.9397488832473755, "learning_rate": 1.801163623558408e-05, "loss": 0.17983596324920653, "step": 8660 }, { "epoch": 0.2283810169804052, "grad_norm": 2.0263288021087646, "learning_rate": 1.800652776024171e-05, "loss": 0.1671680450439453, "step": 8670 }, { "epoch": 0.22864443222490394, "grad_norm": 2.680807113647461, "learning_rate": 1.8001413457403005e-05, "loss": 0.17099359035491943, "step": 8680 }, { "epoch": 0.22890784746940268, "grad_norm": 2.997537851333618, "learning_rate": 1.7996293330790374e-05, "loss": 0.1773221492767334, "step": 8690 }, { "epoch": 0.2291712627139014, "grad_norm": 1.8608683347702026, "learning_rate": 1.7991167384130466e-05, "loss": 0.1697613000869751, "step": 8700 }, { "epoch": 0.22943467795840014, "grad_norm": 2.115640878677368, "learning_rate": 1.798603562115417e-05, "loss": 0.1674631118774414, "step": 8710 }, { "epoch": 0.22969809320289888, "grad_norm": 1.5536547899246216, "learning_rate": 1.79808980455966e-05, "loss": 0.15869386196136476, "step": 8720 }, { "epoch": 0.22996150844739763, "grad_norm": 1.1747506856918335, "learning_rate": 1.7975754661197106e-05, "loss": 0.17284801006317138, "step": 8730 }, { "epoch": 0.23022492369189637, "grad_norm": 2.5048134326934814, "learning_rate": 1.797060547169927e-05, "loss": 0.17022570371627807, "step": 8740 }, { "epoch": 0.2304883389363951, "grad_norm": 2.45426082611084, "learning_rate": 1.796545048085089e-05, "loss": 0.16912513971328735, "step": 8750 }, { "epoch": 0.23075175418089383, "grad_norm": 0.9700652956962585, "learning_rate": 1.796028969240399e-05, "loss": 0.17089654207229615, "step": 8760 }, { "epoch": 0.23101516942539257, "grad_norm": 1.7735426425933838, "learning_rate": 1.7955123110114815e-05, "loss": 0.1677473545074463, "step": 8770 }, { "epoch": 0.23127858466989132, "grad_norm": 1.522149682044983, "learning_rate": 1.794995073774383e-05, "loss": 0.158442485332489, "step": 8780 }, { "epoch": 0.23154199991439003, "grad_norm": 1.6023308038711548, "learning_rate": 1.7944772579055705e-05, "loss": 0.1794361352920532, "step": 8790 }, { "epoch": 0.23180541515888878, "grad_norm": 3.1382369995117188, "learning_rate": 1.793958863781933e-05, "loss": 0.1687293529510498, "step": 8800 }, { "epoch": 0.23206883040338752, "grad_norm": 4.938356876373291, "learning_rate": 1.7934398917807802e-05, "loss": 0.2004534959793091, "step": 8810 }, { "epoch": 0.23233224564788627, "grad_norm": 1.1474403142929077, "learning_rate": 1.7929203422798426e-05, "loss": 0.17093443870544434, "step": 8820 }, { "epoch": 0.23259566089238498, "grad_norm": 2.3286845684051514, "learning_rate": 1.79240021565727e-05, "loss": 0.16432745456695558, "step": 8830 }, { "epoch": 0.23285907613688372, "grad_norm": 7.317227363586426, "learning_rate": 1.7918795122916337e-05, "loss": 0.17037522792816162, "step": 8840 }, { "epoch": 0.23312249138138247, "grad_norm": 1.221976637840271, "learning_rate": 1.7913582325619236e-05, "loss": 0.17334136962890626, "step": 8850 }, { "epoch": 0.2333859066258812, "grad_norm": 4.4593000411987305, "learning_rate": 1.7908363768475498e-05, "loss": 0.1788012742996216, "step": 8860 }, { "epoch": 0.23364932187037996, "grad_norm": 2.1842379570007324, "learning_rate": 1.7903139455283412e-05, "loss": 0.1715841293334961, "step": 8870 }, { "epoch": 0.23391273711487867, "grad_norm": 4.039061069488525, "learning_rate": 1.7897909389845462e-05, "loss": 0.17425801753997802, "step": 8880 }, { "epoch": 0.23417615235937742, "grad_norm": 5.854161739349365, "learning_rate": 1.789267357596831e-05, "loss": 0.1589876413345337, "step": 8890 }, { "epoch": 0.23443956760387616, "grad_norm": 2.721358060836792, "learning_rate": 1.7887432017462812e-05, "loss": 0.18348448276519774, "step": 8900 }, { "epoch": 0.2347029828483749, "grad_norm": 4.438284873962402, "learning_rate": 1.7882184718144002e-05, "loss": 0.17681790590286256, "step": 8910 }, { "epoch": 0.23496639809287362, "grad_norm": 2.8331925868988037, "learning_rate": 1.7876931681831087e-05, "loss": 0.1674894332885742, "step": 8920 }, { "epoch": 0.23522981333737236, "grad_norm": 1.98861825466156, "learning_rate": 1.787167291234745e-05, "loss": 0.16834622621536255, "step": 8930 }, { "epoch": 0.2354932285818711, "grad_norm": 6.180789470672607, "learning_rate": 1.7866408413520662e-05, "loss": 0.16492016315460206, "step": 8940 }, { "epoch": 0.23575664382636985, "grad_norm": 2.0994229316711426, "learning_rate": 1.786113818918244e-05, "loss": 0.18637005090713502, "step": 8950 }, { "epoch": 0.23602005907086857, "grad_norm": 2.335238218307495, "learning_rate": 1.7855862243168684e-05, "loss": 0.161917781829834, "step": 8960 }, { "epoch": 0.2362834743153673, "grad_norm": 2.878754138946533, "learning_rate": 1.785058057931946e-05, "loss": 0.17837060689926149, "step": 8970 }, { "epoch": 0.23654688955986605, "grad_norm": 4.384566783905029, "learning_rate": 1.7845293201478985e-05, "loss": 0.18997454643249512, "step": 8980 }, { "epoch": 0.2368103048043648, "grad_norm": 2.505840539932251, "learning_rate": 1.7840000113495647e-05, "loss": 0.16963764429092407, "step": 8990 }, { "epoch": 0.23707372004886354, "grad_norm": 5.674057483673096, "learning_rate": 1.7834701319221978e-05, "loss": 0.16956237554550171, "step": 9000 }, { "epoch": 0.23733713529336226, "grad_norm": 0.8848870992660522, "learning_rate": 1.782939682251467e-05, "loss": 0.18124821186065673, "step": 9010 }, { "epoch": 0.237600550537861, "grad_norm": 2.5023555755615234, "learning_rate": 1.7824086627234565e-05, "loss": 0.17937697172164918, "step": 9020 }, { "epoch": 0.23786396578235974, "grad_norm": 4.978768825531006, "learning_rate": 1.7818770737246653e-05, "loss": 0.1825796365737915, "step": 9030 }, { "epoch": 0.2381273810268585, "grad_norm": 1.752318024635315, "learning_rate": 1.7813449156420072e-05, "loss": 0.1467057228088379, "step": 9040 }, { "epoch": 0.2383907962713572, "grad_norm": 2.3018860816955566, "learning_rate": 1.7808121888628093e-05, "loss": 0.17485209703445434, "step": 9050 }, { "epoch": 0.23865421151585595, "grad_norm": 1.5911225080490112, "learning_rate": 1.780278893774813e-05, "loss": 0.168038272857666, "step": 9060 }, { "epoch": 0.2389176267603547, "grad_norm": 2.9547786712646484, "learning_rate": 1.7797450307661733e-05, "loss": 0.1780907392501831, "step": 9070 }, { "epoch": 0.23918104200485343, "grad_norm": 2.8957359790802, "learning_rate": 1.77921060022546e-05, "loss": 0.17150572538375855, "step": 9080 }, { "epoch": 0.23944445724935215, "grad_norm": 1.5714499950408936, "learning_rate": 1.7786756025416535e-05, "loss": 0.17696740627288818, "step": 9090 }, { "epoch": 0.2397078724938509, "grad_norm": 2.7014729976654053, "learning_rate": 1.7781400381041484e-05, "loss": 0.17434186935424806, "step": 9100 }, { "epoch": 0.23997128773834964, "grad_norm": 1.0096958875656128, "learning_rate": 1.777603907302752e-05, "loss": 0.16798844337463378, "step": 9110 }, { "epoch": 0.24023470298284838, "grad_norm": 2.0623669624328613, "learning_rate": 1.7770672105276833e-05, "loss": 0.1486300826072693, "step": 9120 }, { "epoch": 0.24049811822734712, "grad_norm": 5.355993747711182, "learning_rate": 1.776529948169573e-05, "loss": 0.17146706581115723, "step": 9130 }, { "epoch": 0.24076153347184584, "grad_norm": 3.308626890182495, "learning_rate": 1.775992120619464e-05, "loss": 0.18470346927642822, "step": 9140 }, { "epoch": 0.24102494871634458, "grad_norm": 1.724395751953125, "learning_rate": 1.7754537282688104e-05, "loss": 0.1703742504119873, "step": 9150 }, { "epoch": 0.24128836396084333, "grad_norm": 2.659172773361206, "learning_rate": 1.7749147715094777e-05, "loss": 0.1627340078353882, "step": 9160 }, { "epoch": 0.24155177920534207, "grad_norm": 0.8493542075157166, "learning_rate": 1.7743752507337414e-05, "loss": 0.17803412675857544, "step": 9170 }, { "epoch": 0.2418151944498408, "grad_norm": 3.234072208404541, "learning_rate": 1.7738351663342887e-05, "loss": 0.17585341930389403, "step": 9180 }, { "epoch": 0.24207860969433953, "grad_norm": 2.1761882305145264, "learning_rate": 1.7732945187042155e-05, "loss": 0.16437897682189942, "step": 9190 }, { "epoch": 0.24234202493883827, "grad_norm": 1.7425565719604492, "learning_rate": 1.7727533082370288e-05, "loss": 0.15520083904266357, "step": 9200 }, { "epoch": 0.24260544018333702, "grad_norm": 1.4637290239334106, "learning_rate": 1.772211535326645e-05, "loss": 0.18810080289840697, "step": 9210 }, { "epoch": 0.24286885542783573, "grad_norm": 2.3215205669403076, "learning_rate": 1.7716692003673898e-05, "loss": 0.16909829378128052, "step": 9220 }, { "epoch": 0.24313227067233448, "grad_norm": 2.6744112968444824, "learning_rate": 1.7711263037539978e-05, "loss": 0.1943747043609619, "step": 9230 }, { "epoch": 0.24339568591683322, "grad_norm": 1.9995051622390747, "learning_rate": 1.770582845881613e-05, "loss": 0.176696515083313, "step": 9240 }, { "epoch": 0.24365910116133196, "grad_norm": 1.8143203258514404, "learning_rate": 1.7700388271457872e-05, "loss": 0.17390177249908448, "step": 9250 }, { "epoch": 0.2439225164058307, "grad_norm": 1.2784675359725952, "learning_rate": 1.7694942479424803e-05, "loss": 0.16808344125747682, "step": 9260 }, { "epoch": 0.24418593165032942, "grad_norm": 1.828185796737671, "learning_rate": 1.768949108668061e-05, "loss": 0.15756406784057617, "step": 9270 }, { "epoch": 0.24444934689482817, "grad_norm": 1.4914684295654297, "learning_rate": 1.7684034097193055e-05, "loss": 0.16339138746261597, "step": 9280 }, { "epoch": 0.2447127621393269, "grad_norm": 5.248774528503418, "learning_rate": 1.7678571514933966e-05, "loss": 0.17548015117645263, "step": 9290 }, { "epoch": 0.24497617738382566, "grad_norm": 4.055760383605957, "learning_rate": 1.7673103343879242e-05, "loss": 0.16767470836639403, "step": 9300 }, { "epoch": 0.24523959262832437, "grad_norm": 1.294032096862793, "learning_rate": 1.7667629588008862e-05, "loss": 0.16103932857513428, "step": 9310 }, { "epoch": 0.24550300787282311, "grad_norm": 1.6302331686019897, "learning_rate": 1.766215025130686e-05, "loss": 0.16695818901062012, "step": 9320 }, { "epoch": 0.24576642311732186, "grad_norm": 1.0765403509140015, "learning_rate": 1.7656665337761327e-05, "loss": 0.1740670919418335, "step": 9330 }, { "epoch": 0.2460298383618206, "grad_norm": 1.7826926708221436, "learning_rate": 1.7651174851364427e-05, "loss": 0.16798934936523438, "step": 9340 }, { "epoch": 0.24629325360631932, "grad_norm": 6.334932804107666, "learning_rate": 1.764567879611237e-05, "loss": 0.16571921110153198, "step": 9350 }, { "epoch": 0.24655666885081806, "grad_norm": 3.123012065887451, "learning_rate": 1.7640177176005425e-05, "loss": 0.1683189630508423, "step": 9360 }, { "epoch": 0.2468200840953168, "grad_norm": 1.2713251113891602, "learning_rate": 1.7634669995047904e-05, "loss": 0.19032095670700072, "step": 9370 }, { "epoch": 0.24708349933981555, "grad_norm": 2.042850971221924, "learning_rate": 1.7629157257248175e-05, "loss": 0.16526777744293214, "step": 9380 }, { "epoch": 0.2473469145843143, "grad_norm": 1.6381175518035889, "learning_rate": 1.762363896661864e-05, "loss": 0.16868135929107667, "step": 9390 }, { "epoch": 0.247610329828813, "grad_norm": 4.822269916534424, "learning_rate": 1.761811512717576e-05, "loss": 0.17145494222640992, "step": 9400 }, { "epoch": 0.24787374507331175, "grad_norm": 1.1917823553085327, "learning_rate": 1.761258574294001e-05, "loss": 0.15338144302368165, "step": 9410 }, { "epoch": 0.2481371603178105, "grad_norm": 2.2591683864593506, "learning_rate": 1.760705081793593e-05, "loss": 0.17086596488952638, "step": 9420 }, { "epoch": 0.24840057556230924, "grad_norm": 4.829523086547852, "learning_rate": 1.760151035619206e-05, "loss": 0.1860605239868164, "step": 9430 }, { "epoch": 0.24866399080680796, "grad_norm": 1.2262787818908691, "learning_rate": 1.7595964361741e-05, "loss": 0.1689380884170532, "step": 9440 }, { "epoch": 0.2489274060513067, "grad_norm": 3.705484390258789, "learning_rate": 1.759041283861936e-05, "loss": 0.1812572956085205, "step": 9450 }, { "epoch": 0.24919082129580544, "grad_norm": 2.1851301193237305, "learning_rate": 1.7584855790867773e-05, "loss": 0.1700753688812256, "step": 9460 }, { "epoch": 0.2494542365403042, "grad_norm": 2.0020666122436523, "learning_rate": 1.7579293222530907e-05, "loss": 0.17313158512115479, "step": 9470 }, { "epoch": 0.24971765178480293, "grad_norm": 1.3905415534973145, "learning_rate": 1.7573725137657432e-05, "loss": 0.16992504596710206, "step": 9480 }, { "epoch": 0.24998106702930165, "grad_norm": 2.93188214302063, "learning_rate": 1.756815154030005e-05, "loss": 0.16876245737075807, "step": 9490 }, { "epoch": 0.2502444822738004, "grad_norm": 1.0209119319915771, "learning_rate": 1.756257243451545e-05, "loss": 0.15713037252426149, "step": 9500 }, { "epoch": 0.2505078975182991, "grad_norm": 1.7380954027175903, "learning_rate": 1.7556987824364363e-05, "loss": 0.17576712369918823, "step": 9510 }, { "epoch": 0.25077131276279785, "grad_norm": 1.3447625637054443, "learning_rate": 1.75513977139115e-05, "loss": 0.17139730453491211, "step": 9520 }, { "epoch": 0.2510347280072966, "grad_norm": 4.612795829772949, "learning_rate": 1.7545802107225585e-05, "loss": 0.17533681392669678, "step": 9530 }, { "epoch": 0.25129814325179534, "grad_norm": 1.2729588747024536, "learning_rate": 1.754020100837934e-05, "loss": 0.17575066089630126, "step": 9540 }, { "epoch": 0.2515615584962941, "grad_norm": 1.7297163009643555, "learning_rate": 1.753459442144949e-05, "loss": 0.17944008111953735, "step": 9550 }, { "epoch": 0.2518249737407928, "grad_norm": 5.010579586029053, "learning_rate": 1.7528982350516754e-05, "loss": 0.17031596899032592, "step": 9560 }, { "epoch": 0.25208838898529157, "grad_norm": 3.540562152862549, "learning_rate": 1.7523364799665832e-05, "loss": 0.16586103439331054, "step": 9570 }, { "epoch": 0.2523518042297903, "grad_norm": 2.5059635639190674, "learning_rate": 1.7517741772985424e-05, "loss": 0.1739649772644043, "step": 9580 }, { "epoch": 0.252615219474289, "grad_norm": 1.0392271280288696, "learning_rate": 1.751211327456821e-05, "loss": 0.17046668529510497, "step": 9590 }, { "epoch": 0.25287863471878774, "grad_norm": 4.536862373352051, "learning_rate": 1.7506479308510855e-05, "loss": 0.16681783199310302, "step": 9600 }, { "epoch": 0.2531420499632865, "grad_norm": 4.6880388259887695, "learning_rate": 1.7500839878914004e-05, "loss": 0.16446478366851808, "step": 9610 }, { "epoch": 0.25340546520778523, "grad_norm": 2.770036220550537, "learning_rate": 1.749519498988227e-05, "loss": 0.1886721134185791, "step": 9620 }, { "epoch": 0.253668880452284, "grad_norm": 4.637833118438721, "learning_rate": 1.748954464552426e-05, "loss": 0.1709519624710083, "step": 9630 }, { "epoch": 0.2539322956967827, "grad_norm": 1.40414559841156, "learning_rate": 1.7483888849952525e-05, "loss": 0.17696449756622315, "step": 9640 }, { "epoch": 0.25419571094128146, "grad_norm": 3.125488042831421, "learning_rate": 1.7478227607283604e-05, "loss": 0.16023489236831664, "step": 9650 }, { "epoch": 0.2544591261857802, "grad_norm": 1.4052754640579224, "learning_rate": 1.7472560921637993e-05, "loss": 0.1691415786743164, "step": 9660 }, { "epoch": 0.25472254143027895, "grad_norm": 4.789284706115723, "learning_rate": 1.746688879714015e-05, "loss": 0.17095000743865968, "step": 9670 }, { "epoch": 0.25498595667477764, "grad_norm": 1.2956119775772095, "learning_rate": 1.746121123791849e-05, "loss": 0.17258856296539307, "step": 9680 }, { "epoch": 0.2552493719192764, "grad_norm": 5.387714385986328, "learning_rate": 1.7455528248105385e-05, "loss": 0.18593918085098265, "step": 9690 }, { "epoch": 0.2555127871637751, "grad_norm": 1.1660876274108887, "learning_rate": 1.7449839831837166e-05, "loss": 0.17938224077224732, "step": 9700 }, { "epoch": 0.25577620240827387, "grad_norm": 3.1837246417999268, "learning_rate": 1.7444145993254105e-05, "loss": 0.15535261631011962, "step": 9710 }, { "epoch": 0.2560396176527726, "grad_norm": 1.1233632564544678, "learning_rate": 1.7438446736500418e-05, "loss": 0.18584858179092406, "step": 9720 }, { "epoch": 0.25630303289727135, "grad_norm": 3.697352170944214, "learning_rate": 1.7432742065724276e-05, "loss": 0.180510413646698, "step": 9730 }, { "epoch": 0.2565664481417701, "grad_norm": 2.7849578857421875, "learning_rate": 1.7427031985077785e-05, "loss": 0.17489216327667237, "step": 9740 }, { "epoch": 0.25682986338626884, "grad_norm": 2.4175074100494385, "learning_rate": 1.7421316498716986e-05, "loss": 0.15456712245941162, "step": 9750 }, { "epoch": 0.2570932786307676, "grad_norm": 3.688279390335083, "learning_rate": 1.741559561080185e-05, "loss": 0.1601771354675293, "step": 9760 }, { "epoch": 0.2573566938752663, "grad_norm": 1.734889030456543, "learning_rate": 1.74098693254963e-05, "loss": 0.14326508045196534, "step": 9770 }, { "epoch": 0.257620109119765, "grad_norm": 1.860126256942749, "learning_rate": 1.7404137646968162e-05, "loss": 0.1721107006072998, "step": 9780 }, { "epoch": 0.25788352436426376, "grad_norm": 1.8858153820037842, "learning_rate": 1.7398400579389204e-05, "loss": 0.16574654579162598, "step": 9790 }, { "epoch": 0.2581469396087625, "grad_norm": 2.389451265335083, "learning_rate": 1.739265812693511e-05, "loss": 0.1689637780189514, "step": 9800 }, { "epoch": 0.25841035485326125, "grad_norm": 3.761762857437134, "learning_rate": 1.7386910293785485e-05, "loss": 0.16992031335830687, "step": 9810 }, { "epoch": 0.25867377009776, "grad_norm": 2.6216018199920654, "learning_rate": 1.738115708412385e-05, "loss": 0.15462884902954102, "step": 9820 }, { "epoch": 0.25893718534225874, "grad_norm": 2.538630247116089, "learning_rate": 1.737539850213764e-05, "loss": 0.1772835969924927, "step": 9830 }, { "epoch": 0.2592006005867575, "grad_norm": 2.8525545597076416, "learning_rate": 1.7369634552018196e-05, "loss": 0.1680350661277771, "step": 9840 }, { "epoch": 0.25946401583125617, "grad_norm": 1.2727465629577637, "learning_rate": 1.7363865237960773e-05, "loss": 0.16386003494262696, "step": 9850 }, { "epoch": 0.2597274310757549, "grad_norm": 4.096906661987305, "learning_rate": 1.7358090564164524e-05, "loss": 0.1631781816482544, "step": 9860 }, { "epoch": 0.25999084632025365, "grad_norm": 1.98016357421875, "learning_rate": 1.735231053483251e-05, "loss": 0.1669553756713867, "step": 9870 }, { "epoch": 0.2602542615647524, "grad_norm": 1.3753807544708252, "learning_rate": 1.734652515417168e-05, "loss": 0.18242881298065186, "step": 9880 }, { "epoch": 0.26051767680925114, "grad_norm": 1.9208524227142334, "learning_rate": 1.7340734426392894e-05, "loss": 0.17158615589141846, "step": 9890 }, { "epoch": 0.2607810920537499, "grad_norm": 2.9068875312805176, "learning_rate": 1.7334938355710876e-05, "loss": 0.1805248737335205, "step": 9900 }, { "epoch": 0.26104450729824863, "grad_norm": 1.7249900102615356, "learning_rate": 1.7329136946344272e-05, "loss": 0.17583423852920532, "step": 9910 }, { "epoch": 0.2613079225427474, "grad_norm": 1.1726938486099243, "learning_rate": 1.7323330202515592e-05, "loss": 0.16635890007019044, "step": 9920 }, { "epoch": 0.2615713377872461, "grad_norm": 4.860715866088867, "learning_rate": 1.731751812845124e-05, "loss": 0.18519020080566406, "step": 9930 }, { "epoch": 0.2618347530317448, "grad_norm": 1.8250468969345093, "learning_rate": 1.7311700728381486e-05, "loss": 0.1645756721496582, "step": 9940 }, { "epoch": 0.26209816827624355, "grad_norm": 2.8991730213165283, "learning_rate": 1.7305878006540483e-05, "loss": 0.16487739086151124, "step": 9950 }, { "epoch": 0.2623615835207423, "grad_norm": 1.8146789073944092, "learning_rate": 1.7300049967166273e-05, "loss": 0.16720428466796874, "step": 9960 }, { "epoch": 0.26262499876524104, "grad_norm": 1.5719938278198242, "learning_rate": 1.729421661450074e-05, "loss": 0.16030718088150026, "step": 9970 }, { "epoch": 0.2628884140097398, "grad_norm": 3.219130754470825, "learning_rate": 1.728837795278966e-05, "loss": 0.16419185400009156, "step": 9980 }, { "epoch": 0.2631518292542385, "grad_norm": 2.787770986557007, "learning_rate": 1.7282533986282656e-05, "loss": 0.1808975338935852, "step": 9990 }, { "epoch": 0.26341524449873727, "grad_norm": 10.209324836730957, "learning_rate": 1.7276684719233222e-05, "loss": 0.1865769624710083, "step": 10000 }, { "epoch": 0.263678659743236, "grad_norm": 2.0888376235961914, "learning_rate": 1.7270830155898704e-05, "loss": 0.18856287002563477, "step": 10010 }, { "epoch": 0.26394207498773475, "grad_norm": 2.9740633964538574, "learning_rate": 1.726497030054031e-05, "loss": 0.17927258014678954, "step": 10020 }, { "epoch": 0.26420549023223344, "grad_norm": 1.1273671388626099, "learning_rate": 1.72591051574231e-05, "loss": 0.16494956016540527, "step": 10030 }, { "epoch": 0.2644689054767322, "grad_norm": 2.920545816421509, "learning_rate": 1.7253234730815964e-05, "loss": 0.18573532104492188, "step": 10040 }, { "epoch": 0.26473232072123093, "grad_norm": 1.6102410554885864, "learning_rate": 1.724735902499166e-05, "loss": 0.17615766525268556, "step": 10050 }, { "epoch": 0.2649957359657297, "grad_norm": 1.0876981019973755, "learning_rate": 1.7241478044226782e-05, "loss": 0.18118655681610107, "step": 10060 }, { "epoch": 0.2652591512102284, "grad_norm": 1.785025715827942, "learning_rate": 1.723559179280176e-05, "loss": 0.1745419979095459, "step": 10070 }, { "epoch": 0.26552256645472716, "grad_norm": 2.031411647796631, "learning_rate": 1.722970027500086e-05, "loss": 0.19144388437271118, "step": 10080 }, { "epoch": 0.2657859816992259, "grad_norm": 3.6531872749328613, "learning_rate": 1.7223803495112185e-05, "loss": 0.17464807033538818, "step": 10090 }, { "epoch": 0.26604939694372465, "grad_norm": 1.5366889238357544, "learning_rate": 1.7217901457427667e-05, "loss": 0.1676336407661438, "step": 10100 }, { "epoch": 0.26631281218822334, "grad_norm": 2.3555822372436523, "learning_rate": 1.7211994166243058e-05, "loss": 0.17107443809509276, "step": 10110 }, { "epoch": 0.2665762274327221, "grad_norm": 1.4524599313735962, "learning_rate": 1.7206081625857943e-05, "loss": 0.18081462383270264, "step": 10120 }, { "epoch": 0.2668396426772208, "grad_norm": 3.52183198928833, "learning_rate": 1.7200163840575728e-05, "loss": 0.17457337379455568, "step": 10130 }, { "epoch": 0.26710305792171957, "grad_norm": 2.989924192428589, "learning_rate": 1.7194240814703626e-05, "loss": 0.17100927829742432, "step": 10140 }, { "epoch": 0.2673664731662183, "grad_norm": 1.430578589439392, "learning_rate": 1.7188312552552672e-05, "loss": 0.18143026828765868, "step": 10150 }, { "epoch": 0.26762988841071705, "grad_norm": 2.7429685592651367, "learning_rate": 1.718237905843771e-05, "loss": 0.17431403398513795, "step": 10160 }, { "epoch": 0.2678933036552158, "grad_norm": 2.2560367584228516, "learning_rate": 1.71764403366774e-05, "loss": 0.1720438241958618, "step": 10170 }, { "epoch": 0.26815671889971454, "grad_norm": 2.4131298065185547, "learning_rate": 1.7170496391594194e-05, "loss": 0.16546313762664794, "step": 10180 }, { "epoch": 0.2684201341442133, "grad_norm": 2.150101661682129, "learning_rate": 1.716454722751435e-05, "loss": 0.1707775115966797, "step": 10190 }, { "epoch": 0.268683549388712, "grad_norm": 1.9172316789627075, "learning_rate": 1.715859284876793e-05, "loss": 0.17281148433685303, "step": 10200 }, { "epoch": 0.2689469646332107, "grad_norm": 0.7473113536834717, "learning_rate": 1.7152633259688788e-05, "loss": 0.17431385517120362, "step": 10210 }, { "epoch": 0.26921037987770946, "grad_norm": 4.507720470428467, "learning_rate": 1.714666846461457e-05, "loss": 0.16435869932174682, "step": 10220 }, { "epoch": 0.2694737951222082, "grad_norm": 1.0565967559814453, "learning_rate": 1.7140698467886707e-05, "loss": 0.16374456882476807, "step": 10230 }, { "epoch": 0.26973721036670695, "grad_norm": 1.214439868927002, "learning_rate": 1.713472327385042e-05, "loss": 0.16906923055648804, "step": 10240 }, { "epoch": 0.2700006256112057, "grad_norm": 3.024686813354492, "learning_rate": 1.7128742886854723e-05, "loss": 0.16165820360183716, "step": 10250 }, { "epoch": 0.27026404085570443, "grad_norm": 2.0641636848449707, "learning_rate": 1.712275731125239e-05, "loss": 0.1718827962875366, "step": 10260 }, { "epoch": 0.2705274561002032, "grad_norm": 6.185838222503662, "learning_rate": 1.7116766551399986e-05, "loss": 0.1606245756149292, "step": 10270 }, { "epoch": 0.2707908713447019, "grad_norm": 3.708057403564453, "learning_rate": 1.711077061165784e-05, "loss": 0.1680593729019165, "step": 10280 }, { "epoch": 0.2710542865892006, "grad_norm": 3.7919576168060303, "learning_rate": 1.710476949639006e-05, "loss": 0.18077954053878784, "step": 10290 }, { "epoch": 0.27131770183369935, "grad_norm": 2.959454298019409, "learning_rate": 1.7098763209964514e-05, "loss": 0.1658491611480713, "step": 10300 }, { "epoch": 0.2715811170781981, "grad_norm": 1.167809247970581, "learning_rate": 1.7092751756752838e-05, "loss": 0.15471465587615968, "step": 10310 }, { "epoch": 0.27184453232269684, "grad_norm": 2.891160726547241, "learning_rate": 1.7086735141130423e-05, "loss": 0.17677171230316163, "step": 10320 }, { "epoch": 0.2721079475671956, "grad_norm": 3.456111431121826, "learning_rate": 1.7080713367476428e-05, "loss": 0.1697321653366089, "step": 10330 }, { "epoch": 0.27237136281169433, "grad_norm": 0.9731795191764832, "learning_rate": 1.7074686440173752e-05, "loss": 0.17252004146575928, "step": 10340 }, { "epoch": 0.2726347780561931, "grad_norm": 1.8922146558761597, "learning_rate": 1.7068654363609058e-05, "loss": 0.15962450504302977, "step": 10350 }, { "epoch": 0.2728981933006918, "grad_norm": 2.673231840133667, "learning_rate": 1.7062617142172754e-05, "loss": 0.18620500564575196, "step": 10360 }, { "epoch": 0.2731616085451905, "grad_norm": 2.8000898361206055, "learning_rate": 1.7056574780258986e-05, "loss": 0.17930977344512938, "step": 10370 }, { "epoch": 0.27342502378968925, "grad_norm": 1.974880337715149, "learning_rate": 1.705052728226564e-05, "loss": 0.16646065711975097, "step": 10380 }, { "epoch": 0.273688439034188, "grad_norm": 1.236221432685852, "learning_rate": 1.704447465259436e-05, "loss": 0.16874320507049562, "step": 10390 }, { "epoch": 0.27395185427868674, "grad_norm": 3.3085696697235107, "learning_rate": 1.7038416895650504e-05, "loss": 0.17372771501541137, "step": 10400 }, { "epoch": 0.2742152695231855, "grad_norm": 2.1481246948242188, "learning_rate": 1.7032354015843162e-05, "loss": 0.1740834355354309, "step": 10410 }, { "epoch": 0.2744786847676842, "grad_norm": 5.534914970397949, "learning_rate": 1.7026286017585168e-05, "loss": 0.1607227325439453, "step": 10420 }, { "epoch": 0.27474210001218297, "grad_norm": 2.0112550258636475, "learning_rate": 1.7020212905293072e-05, "loss": 0.17409423589706421, "step": 10430 }, { "epoch": 0.2750055152566817, "grad_norm": 2.2311933040618896, "learning_rate": 1.7014134683387147e-05, "loss": 0.1786928653717041, "step": 10440 }, { "epoch": 0.27526893050118045, "grad_norm": 1.5783532857894897, "learning_rate": 1.7008051356291382e-05, "loss": 0.16195348501205445, "step": 10450 }, { "epoch": 0.27553234574567914, "grad_norm": 1.5903146266937256, "learning_rate": 1.7001962928433488e-05, "loss": 0.17699579000473023, "step": 10460 }, { "epoch": 0.2757957609901779, "grad_norm": 2.41642689704895, "learning_rate": 1.699586940424488e-05, "loss": 0.18720369338989257, "step": 10470 }, { "epoch": 0.27605917623467663, "grad_norm": 2.029627561569214, "learning_rate": 1.6989770788160697e-05, "loss": 0.1786409378051758, "step": 10480 }, { "epoch": 0.2763225914791754, "grad_norm": 4.369278430938721, "learning_rate": 1.6983667084619768e-05, "loss": 0.1774250388145447, "step": 10490 }, { "epoch": 0.2765860067236741, "grad_norm": 2.3421478271484375, "learning_rate": 1.6977558298064636e-05, "loss": 0.16955904960632323, "step": 10500 }, { "epoch": 0.27684942196817286, "grad_norm": 2.494091033935547, "learning_rate": 1.6971444432941533e-05, "loss": 0.1665808916091919, "step": 10510 }, { "epoch": 0.2771128372126716, "grad_norm": 1.4831452369689941, "learning_rate": 1.69653254937004e-05, "loss": 0.16984509229660033, "step": 10520 }, { "epoch": 0.27737625245717035, "grad_norm": 1.9178398847579956, "learning_rate": 1.6959201484794867e-05, "loss": 0.1642884612083435, "step": 10530 }, { "epoch": 0.2776396677016691, "grad_norm": 1.6281826496124268, "learning_rate": 1.6953072410682243e-05, "loss": 0.18041410446166992, "step": 10540 }, { "epoch": 0.2779030829461678, "grad_norm": 3.7830657958984375, "learning_rate": 1.694693827582354e-05, "loss": 0.17654249668121338, "step": 10550 }, { "epoch": 0.2781664981906665, "grad_norm": 2.0940942764282227, "learning_rate": 1.694079908468344e-05, "loss": 0.16939419507980347, "step": 10560 }, { "epoch": 0.27842991343516527, "grad_norm": 2.842390298843384, "learning_rate": 1.6934654841730318e-05, "loss": 0.17385404109954833, "step": 10570 }, { "epoch": 0.278693328679664, "grad_norm": 1.8808211088180542, "learning_rate": 1.6928505551436215e-05, "loss": 0.1783228635787964, "step": 10580 }, { "epoch": 0.27895674392416275, "grad_norm": 2.8746538162231445, "learning_rate": 1.6922351218276852e-05, "loss": 0.17705997228622436, "step": 10590 }, { "epoch": 0.2792201591686615, "grad_norm": 4.109329700469971, "learning_rate": 1.6916191846731615e-05, "loss": 0.16037931442260742, "step": 10600 }, { "epoch": 0.27948357441316024, "grad_norm": 4.155433654785156, "learning_rate": 1.6910027441283562e-05, "loss": 0.17279670238494874, "step": 10610 }, { "epoch": 0.279746989657659, "grad_norm": 3.4361977577209473, "learning_rate": 1.6903858006419413e-05, "loss": 0.188608717918396, "step": 10620 }, { "epoch": 0.2800104049021577, "grad_norm": 1.4942295551300049, "learning_rate": 1.689768354662955e-05, "loss": 0.1678479552268982, "step": 10630 }, { "epoch": 0.2802738201466564, "grad_norm": 2.254150867462158, "learning_rate": 1.689150406640801e-05, "loss": 0.1775215744972229, "step": 10640 }, { "epoch": 0.28053723539115516, "grad_norm": 1.717659592628479, "learning_rate": 1.6885319570252485e-05, "loss": 0.17733983993530272, "step": 10650 }, { "epoch": 0.2808006506356539, "grad_norm": 4.543431758880615, "learning_rate": 1.687913006266432e-05, "loss": 0.1607531189918518, "step": 10660 }, { "epoch": 0.28106406588015265, "grad_norm": 3.4744760990142822, "learning_rate": 1.6872935548148503e-05, "loss": 0.1759956121444702, "step": 10670 }, { "epoch": 0.2813274811246514, "grad_norm": 3.7958908081054688, "learning_rate": 1.686673603121367e-05, "loss": 0.1668034315109253, "step": 10680 }, { "epoch": 0.28159089636915013, "grad_norm": 4.411660194396973, "learning_rate": 1.6860531516372102e-05, "loss": 0.17831530570983886, "step": 10690 }, { "epoch": 0.2818543116136489, "grad_norm": 2.313586711883545, "learning_rate": 1.6854322008139704e-05, "loss": 0.16945666074752808, "step": 10700 }, { "epoch": 0.2821177268581476, "grad_norm": 1.1074116230010986, "learning_rate": 1.6848107511036026e-05, "loss": 0.17578552961349486, "step": 10710 }, { "epoch": 0.2823811421026463, "grad_norm": 2.2493722438812256, "learning_rate": 1.6841888029584255e-05, "loss": 0.15166597366333007, "step": 10720 }, { "epoch": 0.28264455734714505, "grad_norm": 2.375441789627075, "learning_rate": 1.6835663568311187e-05, "loss": 0.15140150785446166, "step": 10730 }, { "epoch": 0.2829079725916438, "grad_norm": 1.8186980485916138, "learning_rate": 1.682943413174726e-05, "loss": 0.17163283824920655, "step": 10740 }, { "epoch": 0.28317138783614254, "grad_norm": 2.471677780151367, "learning_rate": 1.6823199724426523e-05, "loss": 0.18757717609405516, "step": 10750 }, { "epoch": 0.2834348030806413, "grad_norm": 2.4040393829345703, "learning_rate": 1.6816960350886653e-05, "loss": 0.1849602222442627, "step": 10760 }, { "epoch": 0.28369821832514003, "grad_norm": 3.6748008728027344, "learning_rate": 1.681071601566893e-05, "loss": 0.1777123212814331, "step": 10770 }, { "epoch": 0.28396163356963877, "grad_norm": 1.480320692062378, "learning_rate": 1.6804466723318248e-05, "loss": 0.17503823041915895, "step": 10780 }, { "epoch": 0.2842250488141375, "grad_norm": 3.835320234298706, "learning_rate": 1.6798212478383118e-05, "loss": 0.1691078543663025, "step": 10790 }, { "epoch": 0.28448846405863626, "grad_norm": 3.2488515377044678, "learning_rate": 1.6791953285415645e-05, "loss": 0.16605117321014404, "step": 10800 }, { "epoch": 0.28475187930313495, "grad_norm": 2.8188626766204834, "learning_rate": 1.678568914897154e-05, "loss": 0.17151789665222167, "step": 10810 }, { "epoch": 0.2850152945476337, "grad_norm": 5.404301643371582, "learning_rate": 1.6779420073610116e-05, "loss": 0.1654367208480835, "step": 10820 }, { "epoch": 0.28527870979213243, "grad_norm": 6.018418788909912, "learning_rate": 1.6773146063894273e-05, "loss": 0.19121536016464233, "step": 10830 }, { "epoch": 0.2855421250366312, "grad_norm": 6.746064186096191, "learning_rate": 1.6766867124390505e-05, "loss": 0.1681220054626465, "step": 10840 }, { "epoch": 0.2858055402811299, "grad_norm": 2.697192430496216, "learning_rate": 1.6760583259668897e-05, "loss": 0.18066556453704835, "step": 10850 }, { "epoch": 0.28606895552562867, "grad_norm": 1.1370174884796143, "learning_rate": 1.6754294474303116e-05, "loss": 0.17344357967376708, "step": 10860 }, { "epoch": 0.2863323707701274, "grad_norm": 1.3687772750854492, "learning_rate": 1.674800077287041e-05, "loss": 0.17156126499176025, "step": 10870 }, { "epoch": 0.28659578601462615, "grad_norm": 3.298612356185913, "learning_rate": 1.6741702159951613e-05, "loss": 0.18230879306793213, "step": 10880 }, { "epoch": 0.2868592012591249, "grad_norm": 1.6946183443069458, "learning_rate": 1.6735398640131117e-05, "loss": 0.17144508361816407, "step": 10890 }, { "epoch": 0.2871226165036236, "grad_norm": 1.6103929281234741, "learning_rate": 1.6729090217996904e-05, "loss": 0.1714092254638672, "step": 10900 }, { "epoch": 0.28738603174812233, "grad_norm": 2.1315696239471436, "learning_rate": 1.6722776898140513e-05, "loss": 0.1714531421661377, "step": 10910 }, { "epoch": 0.28764944699262107, "grad_norm": 5.364084243774414, "learning_rate": 1.6716458685157047e-05, "loss": 0.16786682605743408, "step": 10920 }, { "epoch": 0.2879128622371198, "grad_norm": 2.103130340576172, "learning_rate": 1.671013558364518e-05, "loss": 0.1628779172897339, "step": 10930 }, { "epoch": 0.28817627748161856, "grad_norm": 6.633499622344971, "learning_rate": 1.6703807598207132e-05, "loss": 0.16901111602783203, "step": 10940 }, { "epoch": 0.2884396927261173, "grad_norm": 1.4333893060684204, "learning_rate": 1.669747473344869e-05, "loss": 0.16038328409194946, "step": 10950 }, { "epoch": 0.28870310797061605, "grad_norm": 3.430236577987671, "learning_rate": 1.6691136993979183e-05, "loss": 0.15643708705902098, "step": 10960 }, { "epoch": 0.2889665232151148, "grad_norm": 1.6689856052398682, "learning_rate": 1.6684794384411494e-05, "loss": 0.16259673833847046, "step": 10970 }, { "epoch": 0.2892299384596135, "grad_norm": 1.5429521799087524, "learning_rate": 1.667844690936204e-05, "loss": 0.17135932445526122, "step": 10980 }, { "epoch": 0.2894933537041122, "grad_norm": 1.997921109199524, "learning_rate": 1.6672094573450798e-05, "loss": 0.18754165172576903, "step": 10990 }, { "epoch": 0.28975676894861097, "grad_norm": 1.991801142692566, "learning_rate": 1.6665737381301263e-05, "loss": 0.1735477328300476, "step": 11000 }, { "epoch": 0.2900201841931097, "grad_norm": 1.9267622232437134, "learning_rate": 1.665937533754048e-05, "loss": 0.16066467761993408, "step": 11010 }, { "epoch": 0.29028359943760845, "grad_norm": 2.112894058227539, "learning_rate": 1.6653008446799014e-05, "loss": 0.17180747985839845, "step": 11020 }, { "epoch": 0.2905470146821072, "grad_norm": 1.5868146419525146, "learning_rate": 1.6646636713710967e-05, "loss": 0.16032593250274657, "step": 11030 }, { "epoch": 0.29081042992660594, "grad_norm": 3.518279552459717, "learning_rate": 1.664026014291396e-05, "loss": 0.18648384809494017, "step": 11040 }, { "epoch": 0.2910738451711047, "grad_norm": 5.187108039855957, "learning_rate": 1.6633878739049135e-05, "loss": 0.1732252359390259, "step": 11050 }, { "epoch": 0.2913372604156034, "grad_norm": 2.319106340408325, "learning_rate": 1.6627492506761155e-05, "loss": 0.18140662908554078, "step": 11060 }, { "epoch": 0.2916006756601021, "grad_norm": 2.147510528564453, "learning_rate": 1.6621101450698193e-05, "loss": 0.18545727729797362, "step": 11070 }, { "epoch": 0.29186409090460086, "grad_norm": 1.9996960163116455, "learning_rate": 1.6614705575511937e-05, "loss": 0.1806283950805664, "step": 11080 }, { "epoch": 0.2921275061490996, "grad_norm": 4.427528381347656, "learning_rate": 1.660830488585758e-05, "loss": 0.16274924278259278, "step": 11090 }, { "epoch": 0.29239092139359835, "grad_norm": 2.4766368865966797, "learning_rate": 1.660189938639382e-05, "loss": 0.1641000270843506, "step": 11100 }, { "epoch": 0.2926543366380971, "grad_norm": 1.605986475944519, "learning_rate": 1.6595489081782858e-05, "loss": 0.17620511054992677, "step": 11110 }, { "epoch": 0.29291775188259583, "grad_norm": 1.4490472078323364, "learning_rate": 1.6589073976690387e-05, "loss": 0.1850878953933716, "step": 11120 }, { "epoch": 0.2931811671270946, "grad_norm": 2.833563804626465, "learning_rate": 1.65826540757856e-05, "loss": 0.16977790594100953, "step": 11130 }, { "epoch": 0.2934445823715933, "grad_norm": 1.6285055875778198, "learning_rate": 1.6576229383741177e-05, "loss": 0.17304465770721436, "step": 11140 }, { "epoch": 0.29370799761609206, "grad_norm": 2.9233787059783936, "learning_rate": 1.6569799905233285e-05, "loss": 0.16548423767089843, "step": 11150 }, { "epoch": 0.29397141286059075, "grad_norm": 2.106975555419922, "learning_rate": 1.6563365644941578e-05, "loss": 0.174916410446167, "step": 11160 }, { "epoch": 0.2942348281050895, "grad_norm": 3.436124324798584, "learning_rate": 1.6556926607549186e-05, "loss": 0.16773226261138915, "step": 11170 }, { "epoch": 0.29449824334958824, "grad_norm": 2.7380642890930176, "learning_rate": 1.655048279774272e-05, "loss": 0.1744521975517273, "step": 11180 }, { "epoch": 0.294761658594087, "grad_norm": 3.499770402908325, "learning_rate": 1.6544034220212262e-05, "loss": 0.16936330795288085, "step": 11190 }, { "epoch": 0.2950250738385857, "grad_norm": 1.1885102987289429, "learning_rate": 1.6537580879651364e-05, "loss": 0.16240496635437013, "step": 11200 }, { "epoch": 0.29528848908308447, "grad_norm": 1.3388488292694092, "learning_rate": 1.6531122780757046e-05, "loss": 0.17792179584503173, "step": 11210 }, { "epoch": 0.2955519043275832, "grad_norm": 3.9184062480926514, "learning_rate": 1.6524659928229792e-05, "loss": 0.1840757369995117, "step": 11220 }, { "epoch": 0.29581531957208196, "grad_norm": 4.946585655212402, "learning_rate": 1.6518192326773546e-05, "loss": 0.1812190294265747, "step": 11230 }, { "epoch": 0.29607873481658065, "grad_norm": 2.966855764389038, "learning_rate": 1.6511719981095703e-05, "loss": 0.18624267578125, "step": 11240 }, { "epoch": 0.2963421500610794, "grad_norm": 3.9261057376861572, "learning_rate": 1.6505242895907122e-05, "loss": 0.17331089973449706, "step": 11250 }, { "epoch": 0.29660556530557813, "grad_norm": 3.9662585258483887, "learning_rate": 1.64987610759221e-05, "loss": 0.17197532653808595, "step": 11260 }, { "epoch": 0.2968689805500769, "grad_norm": 1.8090510368347168, "learning_rate": 1.6492274525858386e-05, "loss": 0.1777012586593628, "step": 11270 }, { "epoch": 0.2971323957945756, "grad_norm": 2.8187780380249023, "learning_rate": 1.648578325043717e-05, "loss": 0.17142391204833984, "step": 11280 }, { "epoch": 0.29739581103907436, "grad_norm": 0.9598867297172546, "learning_rate": 1.6479287254383084e-05, "loss": 0.18273191452026366, "step": 11290 }, { "epoch": 0.2976592262835731, "grad_norm": 1.6398122310638428, "learning_rate": 1.6472786542424196e-05, "loss": 0.15789536237716675, "step": 11300 }, { "epoch": 0.29792264152807185, "grad_norm": 3.1827757358551025, "learning_rate": 1.6466281119292004e-05, "loss": 0.18213844299316406, "step": 11310 }, { "epoch": 0.2981860567725706, "grad_norm": 2.140352249145508, "learning_rate": 1.6459770989721433e-05, "loss": 0.18527143001556395, "step": 11320 }, { "epoch": 0.2984494720170693, "grad_norm": 3.304299831390381, "learning_rate": 1.6453256158450835e-05, "loss": 0.17215118408203126, "step": 11330 }, { "epoch": 0.298712887261568, "grad_norm": 0.8981837034225464, "learning_rate": 1.644673663022199e-05, "loss": 0.1757228970527649, "step": 11340 }, { "epoch": 0.29897630250606677, "grad_norm": 3.39699125289917, "learning_rate": 1.6440212409780088e-05, "loss": 0.1738342523574829, "step": 11350 }, { "epoch": 0.2992397177505655, "grad_norm": 1.4501633644104004, "learning_rate": 1.643368350187374e-05, "loss": 0.18781490325927735, "step": 11360 }, { "epoch": 0.29950313299506426, "grad_norm": 2.818020820617676, "learning_rate": 1.6427149911254973e-05, "loss": 0.1667075276374817, "step": 11370 }, { "epoch": 0.299766548239563, "grad_norm": 1.5989407300949097, "learning_rate": 1.64206116426792e-05, "loss": 0.16073310375213623, "step": 11380 }, { "epoch": 0.30002996348406175, "grad_norm": 1.7347092628479004, "learning_rate": 1.641406870090527e-05, "loss": 0.16213557720184327, "step": 11390 }, { "epoch": 0.3002933787285605, "grad_norm": 2.2834274768829346, "learning_rate": 1.640752109069541e-05, "loss": 0.17495269775390626, "step": 11400 }, { "epoch": 0.30055679397305923, "grad_norm": 0.921018123626709, "learning_rate": 1.640096881681526e-05, "loss": 0.1677407741546631, "step": 11410 }, { "epoch": 0.3008202092175579, "grad_norm": 3.1042966842651367, "learning_rate": 1.6394411884033844e-05, "loss": 0.17492382526397704, "step": 11420 }, { "epoch": 0.30108362446205666, "grad_norm": 1.14279043674469, "learning_rate": 1.638785029712358e-05, "loss": 0.1726557970046997, "step": 11430 }, { "epoch": 0.3013470397065554, "grad_norm": 3.7108266353607178, "learning_rate": 1.638128406086027e-05, "loss": 0.17790144681930542, "step": 11440 }, { "epoch": 0.30161045495105415, "grad_norm": 2.2659263610839844, "learning_rate": 1.6374713180023114e-05, "loss": 0.16549460887908934, "step": 11450 }, { "epoch": 0.3018738701955529, "grad_norm": 1.2753863334655762, "learning_rate": 1.6368137659394673e-05, "loss": 0.17325005531311036, "step": 11460 }, { "epoch": 0.30213728544005164, "grad_norm": 3.214409828186035, "learning_rate": 1.63615575037609e-05, "loss": 0.19662760496139525, "step": 11470 }, { "epoch": 0.3024007006845504, "grad_norm": 5.250167369842529, "learning_rate": 1.6354972717911115e-05, "loss": 0.15144660472869872, "step": 11480 }, { "epoch": 0.3026641159290491, "grad_norm": 4.225930690765381, "learning_rate": 1.634838330663801e-05, "loss": 0.17326028347015382, "step": 11490 }, { "epoch": 0.3029275311735478, "grad_norm": 2.6131210327148438, "learning_rate": 1.6341789274737644e-05, "loss": 0.17267507314682007, "step": 11500 }, { "epoch": 0.30319094641804656, "grad_norm": 2.3443148136138916, "learning_rate": 1.6335190627009432e-05, "loss": 0.18530236482620238, "step": 11510 }, { "epoch": 0.3034543616625453, "grad_norm": 4.956183910369873, "learning_rate": 1.632858736825616e-05, "loss": 0.17068881988525392, "step": 11520 }, { "epoch": 0.30371777690704405, "grad_norm": 1.7407631874084473, "learning_rate": 1.632197950328396e-05, "loss": 0.16635396480560302, "step": 11530 }, { "epoch": 0.3039811921515428, "grad_norm": 2.964989423751831, "learning_rate": 1.631536703690233e-05, "loss": 0.1691913366317749, "step": 11540 }, { "epoch": 0.30424460739604153, "grad_norm": 2.6277005672454834, "learning_rate": 1.6308749973924096e-05, "loss": 0.17363357543945312, "step": 11550 }, { "epoch": 0.3045080226405403, "grad_norm": 1.2841287851333618, "learning_rate": 1.6302128319165454e-05, "loss": 0.18526341915130615, "step": 11560 }, { "epoch": 0.304771437885039, "grad_norm": 2.7807295322418213, "learning_rate": 1.629550207744592e-05, "loss": 0.17908432483673095, "step": 11570 }, { "epoch": 0.30503485312953776, "grad_norm": 1.670256495475769, "learning_rate": 1.6288871253588366e-05, "loss": 0.1812579870223999, "step": 11580 }, { "epoch": 0.30529826837403645, "grad_norm": 2.528287172317505, "learning_rate": 1.6282235852418983e-05, "loss": 0.1762699604034424, "step": 11590 }, { "epoch": 0.3055616836185352, "grad_norm": 1.9335589408874512, "learning_rate": 1.6275595878767315e-05, "loss": 0.17405856847763063, "step": 11600 }, { "epoch": 0.30582509886303394, "grad_norm": 2.077650785446167, "learning_rate": 1.6268951337466212e-05, "loss": 0.17014691829681397, "step": 11610 }, { "epoch": 0.3060885141075327, "grad_norm": 1.3672311305999756, "learning_rate": 1.626230223335186e-05, "loss": 0.19484846591949462, "step": 11620 }, { "epoch": 0.3063519293520314, "grad_norm": 1.5202871561050415, "learning_rate": 1.6255648571263762e-05, "loss": 0.18641340732574463, "step": 11630 }, { "epoch": 0.30661534459653017, "grad_norm": 2.732577323913574, "learning_rate": 1.6248990356044742e-05, "loss": 0.18058688640594484, "step": 11640 }, { "epoch": 0.3068787598410289, "grad_norm": 2.4948654174804688, "learning_rate": 1.6242327592540942e-05, "loss": 0.15918664932250975, "step": 11650 }, { "epoch": 0.30714217508552766, "grad_norm": 1.4330512285232544, "learning_rate": 1.62356602856018e-05, "loss": 0.17521049976348876, "step": 11660 }, { "epoch": 0.3074055903300264, "grad_norm": 2.05549693107605, "learning_rate": 1.6228988440080068e-05, "loss": 0.16280291080474854, "step": 11670 }, { "epoch": 0.3076690055745251, "grad_norm": 3.7981553077697754, "learning_rate": 1.6222312060831814e-05, "loss": 0.1589975118637085, "step": 11680 }, { "epoch": 0.30793242081902383, "grad_norm": 4.670704364776611, "learning_rate": 1.6215631152716387e-05, "loss": 0.18129923343658447, "step": 11690 }, { "epoch": 0.3081958360635226, "grad_norm": 2.5332674980163574, "learning_rate": 1.620894572059644e-05, "loss": 0.17721703052520751, "step": 11700 }, { "epoch": 0.3084592513080213, "grad_norm": 1.820286512374878, "learning_rate": 1.6202255769337924e-05, "loss": 0.1575068712234497, "step": 11710 }, { "epoch": 0.30872266655252006, "grad_norm": 1.3636670112609863, "learning_rate": 1.619556130381007e-05, "loss": 0.1742151975631714, "step": 11720 }, { "epoch": 0.3089860817970188, "grad_norm": 2.2883670330047607, "learning_rate": 1.61888623288854e-05, "loss": 0.18428945541381836, "step": 11730 }, { "epoch": 0.30924949704151755, "grad_norm": 3.6709206104278564, "learning_rate": 1.6182158849439718e-05, "loss": 0.16043424606323242, "step": 11740 }, { "epoch": 0.3095129122860163, "grad_norm": 3.0373895168304443, "learning_rate": 1.61754508703521e-05, "loss": 0.17408145666122438, "step": 11750 }, { "epoch": 0.309776327530515, "grad_norm": 1.952756404876709, "learning_rate": 1.6168738396504912e-05, "loss": 0.18194248676300048, "step": 11760 }, { "epoch": 0.3100397427750137, "grad_norm": 4.568816184997559, "learning_rate": 1.6162021432783773e-05, "loss": 0.17580976486206054, "step": 11770 }, { "epoch": 0.31030315801951247, "grad_norm": 2.670384407043457, "learning_rate": 1.6155299984077584e-05, "loss": 0.16558566093444824, "step": 11780 }, { "epoch": 0.3105665732640112, "grad_norm": 1.4052619934082031, "learning_rate": 1.6148574055278507e-05, "loss": 0.14677757024765015, "step": 11790 }, { "epoch": 0.31082998850850996, "grad_norm": 2.622354030609131, "learning_rate": 1.6141843651281955e-05, "loss": 0.16540083885192872, "step": 11800 }, { "epoch": 0.3110934037530087, "grad_norm": 2.8344273567199707, "learning_rate": 1.6135108776986615e-05, "loss": 0.1806156277656555, "step": 11810 }, { "epoch": 0.31135681899750745, "grad_norm": 1.3378770351409912, "learning_rate": 1.612836943729442e-05, "loss": 0.1546389102935791, "step": 11820 }, { "epoch": 0.3116202342420062, "grad_norm": 2.8203697204589844, "learning_rate": 1.6121625637110546e-05, "loss": 0.17861319780349733, "step": 11830 }, { "epoch": 0.31188364948650493, "grad_norm": 2.407794952392578, "learning_rate": 1.6114877381343422e-05, "loss": 0.17292120456695556, "step": 11840 }, { "epoch": 0.3121470647310036, "grad_norm": 1.7539100646972656, "learning_rate": 1.610812467490472e-05, "loss": 0.18991432189941407, "step": 11850 }, { "epoch": 0.31241047997550236, "grad_norm": 4.004004001617432, "learning_rate": 1.6101367522709356e-05, "loss": 0.17057435512542723, "step": 11860 }, { "epoch": 0.3126738952200011, "grad_norm": 2.6717467308044434, "learning_rate": 1.6094605929675476e-05, "loss": 0.16109434366226197, "step": 11870 }, { "epoch": 0.31293731046449985, "grad_norm": 1.6745964288711548, "learning_rate": 1.608783990072445e-05, "loss": 0.1729148507118225, "step": 11880 }, { "epoch": 0.3132007257089986, "grad_norm": 4.8190178871154785, "learning_rate": 1.6081069440780895e-05, "loss": 0.1738576889038086, "step": 11890 }, { "epoch": 0.31346414095349734, "grad_norm": 2.7170135974884033, "learning_rate": 1.607429455477264e-05, "loss": 0.15835213661193848, "step": 11900 }, { "epoch": 0.3137275561979961, "grad_norm": 0.9887146949768066, "learning_rate": 1.606751524763075e-05, "loss": 0.15864717960357666, "step": 11910 }, { "epoch": 0.3139909714424948, "grad_norm": 1.088428020477295, "learning_rate": 1.606073152428948e-05, "loss": 0.17476146221160888, "step": 11920 }, { "epoch": 0.31425438668699357, "grad_norm": 2.6736528873443604, "learning_rate": 1.6053943389686323e-05, "loss": 0.17149133682250978, "step": 11930 }, { "epoch": 0.31451780193149226, "grad_norm": 2.1328094005584717, "learning_rate": 1.604715084876198e-05, "loss": 0.1655435562133789, "step": 11940 }, { "epoch": 0.314781217175991, "grad_norm": 1.3722493648529053, "learning_rate": 1.6040353906460353e-05, "loss": 0.17612483501434326, "step": 11950 }, { "epoch": 0.31504463242048975, "grad_norm": 1.546751618385315, "learning_rate": 1.6033552567728553e-05, "loss": 0.16072731018066405, "step": 11960 }, { "epoch": 0.3153080476649885, "grad_norm": 1.8716782331466675, "learning_rate": 1.602674683751688e-05, "loss": 0.17773960828781127, "step": 11970 }, { "epoch": 0.31557146290948723, "grad_norm": 2.770393133163452, "learning_rate": 1.6019936720778845e-05, "loss": 0.1795259714126587, "step": 11980 }, { "epoch": 0.315834878153986, "grad_norm": 1.652958869934082, "learning_rate": 1.601312222247114e-05, "loss": 0.16624579429626465, "step": 11990 }, { "epoch": 0.3160982933984847, "grad_norm": 2.1126325130462646, "learning_rate": 1.6006303347553655e-05, "loss": 0.1809604287147522, "step": 12000 }, { "epoch": 0.31636170864298346, "grad_norm": 1.6994988918304443, "learning_rate": 1.5999480100989458e-05, "loss": 0.17713605165481566, "step": 12010 }, { "epoch": 0.31662512388748215, "grad_norm": 2.4320952892303467, "learning_rate": 1.5992652487744806e-05, "loss": 0.17048470973968505, "step": 12020 }, { "epoch": 0.3168885391319809, "grad_norm": 1.537379264831543, "learning_rate": 1.5985820512789128e-05, "loss": 0.15424277782440185, "step": 12030 }, { "epoch": 0.31715195437647964, "grad_norm": 3.024289608001709, "learning_rate": 1.5978984181095034e-05, "loss": 0.18160529136657716, "step": 12040 }, { "epoch": 0.3174153696209784, "grad_norm": 1.6183611154556274, "learning_rate": 1.5972143497638294e-05, "loss": 0.17687267065048218, "step": 12050 }, { "epoch": 0.3176787848654771, "grad_norm": 1.8800886869430542, "learning_rate": 1.5965298467397855e-05, "loss": 0.1628877639770508, "step": 12060 }, { "epoch": 0.31794220010997587, "grad_norm": 1.5302654504776, "learning_rate": 1.595844909535583e-05, "loss": 0.17538225650787354, "step": 12070 }, { "epoch": 0.3182056153544746, "grad_norm": 1.4013060331344604, "learning_rate": 1.595159538649748e-05, "loss": 0.16928647756576537, "step": 12080 }, { "epoch": 0.31846903059897336, "grad_norm": 4.349745750427246, "learning_rate": 1.5944737345811238e-05, "loss": 0.17584481239318847, "step": 12090 }, { "epoch": 0.3187324458434721, "grad_norm": 0.6865906119346619, "learning_rate": 1.5937874978288676e-05, "loss": 0.16750085353851318, "step": 12100 }, { "epoch": 0.3189958610879708, "grad_norm": 1.7541159391403198, "learning_rate": 1.5931008288924523e-05, "loss": 0.17192480564117432, "step": 12110 }, { "epoch": 0.31925927633246953, "grad_norm": 1.5393681526184082, "learning_rate": 1.5924137282716653e-05, "loss": 0.1640103578567505, "step": 12120 }, { "epoch": 0.3195226915769683, "grad_norm": 2.270939588546753, "learning_rate": 1.591726196466608e-05, "loss": 0.183471155166626, "step": 12130 }, { "epoch": 0.319786106821467, "grad_norm": 1.4934097528457642, "learning_rate": 1.5910382339776953e-05, "loss": 0.16969950199127198, "step": 12140 }, { "epoch": 0.32004952206596576, "grad_norm": 3.272989511489868, "learning_rate": 1.5903498413056568e-05, "loss": 0.16956121921539308, "step": 12150 }, { "epoch": 0.3203129373104645, "grad_norm": 2.303516387939453, "learning_rate": 1.5896610189515335e-05, "loss": 0.1679680347442627, "step": 12160 }, { "epoch": 0.32057635255496325, "grad_norm": 2.2209789752960205, "learning_rate": 1.588971767416681e-05, "loss": 0.17508589029312133, "step": 12170 }, { "epoch": 0.320839767799462, "grad_norm": 4.899314880371094, "learning_rate": 1.5882820872027654e-05, "loss": 0.17356717586517334, "step": 12180 }, { "epoch": 0.32110318304396074, "grad_norm": 1.2358345985412598, "learning_rate": 1.5875919788117662e-05, "loss": 0.1759329319000244, "step": 12190 }, { "epoch": 0.3213665982884594, "grad_norm": 1.0962650775909424, "learning_rate": 1.586901442745974e-05, "loss": 0.17887670993804933, "step": 12200 }, { "epoch": 0.32163001353295817, "grad_norm": 3.0361549854278564, "learning_rate": 1.5862104795079903e-05, "loss": 0.1631428122520447, "step": 12210 }, { "epoch": 0.3218934287774569, "grad_norm": 2.3433544635772705, "learning_rate": 1.5855190896007287e-05, "loss": 0.1943231463432312, "step": 12220 }, { "epoch": 0.32215684402195566, "grad_norm": 1.5203372240066528, "learning_rate": 1.584827273527412e-05, "loss": 0.15167717933654784, "step": 12230 }, { "epoch": 0.3224202592664544, "grad_norm": 4.899445533752441, "learning_rate": 1.5841350317915745e-05, "loss": 0.1698099732398987, "step": 12240 }, { "epoch": 0.32268367451095314, "grad_norm": 1.4070968627929688, "learning_rate": 1.5834423648970588e-05, "loss": 0.17722467184066773, "step": 12250 }, { "epoch": 0.3229470897554519, "grad_norm": 2.4146080017089844, "learning_rate": 1.5827492733480183e-05, "loss": 0.1815449833869934, "step": 12260 }, { "epoch": 0.32321050499995063, "grad_norm": 2.230311155319214, "learning_rate": 1.5820557576489145e-05, "loss": 0.15341194868087768, "step": 12270 }, { "epoch": 0.3234739202444493, "grad_norm": 1.1937671899795532, "learning_rate": 1.581361818304518e-05, "loss": 0.1622283339500427, "step": 12280 }, { "epoch": 0.32373733548894806, "grad_norm": 1.4783891439437866, "learning_rate": 1.5806674558199088e-05, "loss": 0.18211731910705567, "step": 12290 }, { "epoch": 0.3240007507334468, "grad_norm": 1.2803457975387573, "learning_rate": 1.5799726707004728e-05, "loss": 0.17660973072052003, "step": 12300 }, { "epoch": 0.32426416597794555, "grad_norm": 1.6764092445373535, "learning_rate": 1.5792774634519046e-05, "loss": 0.18259804248809813, "step": 12310 }, { "epoch": 0.3245275812224443, "grad_norm": 3.859341859817505, "learning_rate": 1.5785818345802068e-05, "loss": 0.17063016891479493, "step": 12320 }, { "epoch": 0.32479099646694304, "grad_norm": 1.5385537147521973, "learning_rate": 1.5778857845916874e-05, "loss": 0.17476805448532104, "step": 12330 }, { "epoch": 0.3250544117114418, "grad_norm": 1.928330421447754, "learning_rate": 1.577189313992962e-05, "loss": 0.1610261917114258, "step": 12340 }, { "epoch": 0.3253178269559405, "grad_norm": 1.696282148361206, "learning_rate": 1.5764924232909518e-05, "loss": 0.18324992656707764, "step": 12350 }, { "epoch": 0.32558124220043927, "grad_norm": 1.2840908765792847, "learning_rate": 1.5757951129928837e-05, "loss": 0.16594611406326293, "step": 12360 }, { "epoch": 0.32584465744493796, "grad_norm": 1.1543594598770142, "learning_rate": 1.5750973836062907e-05, "loss": 0.16473464965820311, "step": 12370 }, { "epoch": 0.3261080726894367, "grad_norm": 3.0727837085723877, "learning_rate": 1.57439923563901e-05, "loss": 0.1795044422149658, "step": 12380 }, { "epoch": 0.32637148793393544, "grad_norm": 2.0482370853424072, "learning_rate": 1.573700669599184e-05, "loss": 0.16641101837158204, "step": 12390 }, { "epoch": 0.3266349031784342, "grad_norm": 1.7943121194839478, "learning_rate": 1.573001685995259e-05, "loss": 0.16267645359039307, "step": 12400 }, { "epoch": 0.32689831842293293, "grad_norm": 26.108943939208984, "learning_rate": 1.572302285335985e-05, "loss": 0.4353433132171631, "step": 12410 }, { "epoch": 0.3271617336674317, "grad_norm": 2.6373348236083984, "learning_rate": 1.5716024681304177e-05, "loss": 0.42107448577880857, "step": 12420 }, { "epoch": 0.3274251489119304, "grad_norm": 4.880199432373047, "learning_rate": 1.5709022348879124e-05, "loss": 0.20741209983825684, "step": 12430 }, { "epoch": 0.32768856415642916, "grad_norm": 1.224829077720642, "learning_rate": 1.5702015861181295e-05, "loss": 0.18882279396057128, "step": 12440 }, { "epoch": 0.3279519794009279, "grad_norm": 3.8538095951080322, "learning_rate": 1.569500522331032e-05, "loss": 0.18720386028289795, "step": 12450 }, { "epoch": 0.3282153946454266, "grad_norm": 4.249204635620117, "learning_rate": 1.568799044036884e-05, "loss": 0.17898087501525878, "step": 12460 }, { "epoch": 0.32847880988992534, "grad_norm": 2.615129232406616, "learning_rate": 1.5680971517462516e-05, "loss": 0.1571852445602417, "step": 12470 }, { "epoch": 0.3287422251344241, "grad_norm": 4.423629283905029, "learning_rate": 1.5673948459700024e-05, "loss": 0.16121962070465087, "step": 12480 }, { "epoch": 0.3290056403789228, "grad_norm": 3.443063974380493, "learning_rate": 1.5666921272193046e-05, "loss": 0.16702237129211425, "step": 12490 }, { "epoch": 0.32926905562342157, "grad_norm": 3.9424638748168945, "learning_rate": 1.565988996005628e-05, "loss": 0.17729473114013672, "step": 12500 }, { "epoch": 0.3295324708679203, "grad_norm": 2.4912285804748535, "learning_rate": 1.5652854528407407e-05, "loss": 0.16977760791778565, "step": 12510 }, { "epoch": 0.32979588611241906, "grad_norm": 2.3079800605773926, "learning_rate": 1.5645814982367125e-05, "loss": 0.16804964542388917, "step": 12520 }, { "epoch": 0.3300593013569178, "grad_norm": 1.0278490781784058, "learning_rate": 1.5638771327059117e-05, "loss": 0.17996437549591066, "step": 12530 }, { "epoch": 0.3303227166014165, "grad_norm": 3.498908281326294, "learning_rate": 1.563172356761006e-05, "loss": 0.17160542011260987, "step": 12540 }, { "epoch": 0.33058613184591523, "grad_norm": 1.3162167072296143, "learning_rate": 1.5624671709149618e-05, "loss": 0.15848937034606933, "step": 12550 }, { "epoch": 0.330849547090414, "grad_norm": 2.65483021736145, "learning_rate": 1.561761575681044e-05, "loss": 0.16250481605529785, "step": 12560 }, { "epoch": 0.3311129623349127, "grad_norm": 2.1822383403778076, "learning_rate": 1.5610555715728146e-05, "loss": 0.1594555139541626, "step": 12570 }, { "epoch": 0.33137637757941146, "grad_norm": 2.1958744525909424, "learning_rate": 1.560349159104134e-05, "loss": 0.17031978368759154, "step": 12580 }, { "epoch": 0.3316397928239102, "grad_norm": 2.955902099609375, "learning_rate": 1.55964233878916e-05, "loss": 0.1657567501068115, "step": 12590 }, { "epoch": 0.33190320806840895, "grad_norm": 1.063578724861145, "learning_rate": 1.5589351111423474e-05, "loss": 0.16193270683288574, "step": 12600 }, { "epoch": 0.3321666233129077, "grad_norm": 2.678223133087158, "learning_rate": 1.5582274766784457e-05, "loss": 0.18764374256134034, "step": 12610 }, { "epoch": 0.33243003855740644, "grad_norm": 2.0107901096343994, "learning_rate": 1.557519435912503e-05, "loss": 0.1520442008972168, "step": 12620 }, { "epoch": 0.3326934538019051, "grad_norm": 1.6881669759750366, "learning_rate": 1.556810989359861e-05, "loss": 0.17687602043151857, "step": 12630 }, { "epoch": 0.33295686904640387, "grad_norm": 3.4163098335266113, "learning_rate": 1.5561021375361583e-05, "loss": 0.17832667827606202, "step": 12640 }, { "epoch": 0.3332202842909026, "grad_norm": 1.9847726821899414, "learning_rate": 1.5553928809573275e-05, "loss": 0.17117931842803955, "step": 12650 }, { "epoch": 0.33348369953540136, "grad_norm": 2.1208977699279785, "learning_rate": 1.5546832201395964e-05, "loss": 0.1707533597946167, "step": 12660 }, { "epoch": 0.3337471147799001, "grad_norm": 2.4550750255584717, "learning_rate": 1.5539731555994864e-05, "loss": 0.17072292566299438, "step": 12670 }, { "epoch": 0.33401053002439884, "grad_norm": 1.8355180025100708, "learning_rate": 1.5532626878538137e-05, "loss": 0.16214442253112793, "step": 12680 }, { "epoch": 0.3342739452688976, "grad_norm": 4.215703010559082, "learning_rate": 1.5525518174196867e-05, "loss": 0.16860904693603515, "step": 12690 }, { "epoch": 0.33453736051339633, "grad_norm": 2.9794743061065674, "learning_rate": 1.551840544814508e-05, "loss": 0.16263968944549562, "step": 12700 }, { "epoch": 0.3348007757578951, "grad_norm": 1.4102451801300049, "learning_rate": 1.551128870555972e-05, "loss": 0.16734684705734254, "step": 12710 }, { "epoch": 0.33506419100239376, "grad_norm": 3.4653806686401367, "learning_rate": 1.550416795162067e-05, "loss": 0.16929252147674562, "step": 12720 }, { "epoch": 0.3353276062468925, "grad_norm": 2.6115453243255615, "learning_rate": 1.549704319151071e-05, "loss": 0.1703181505203247, "step": 12730 }, { "epoch": 0.33559102149139125, "grad_norm": 1.4676663875579834, "learning_rate": 1.548991443041555e-05, "loss": 0.17308815717697143, "step": 12740 }, { "epoch": 0.33585443673589, "grad_norm": 3.2148799896240234, "learning_rate": 1.5482781673523816e-05, "loss": 0.16745203733444214, "step": 12750 }, { "epoch": 0.33611785198038874, "grad_norm": 1.129506230354309, "learning_rate": 1.5475644926027035e-05, "loss": 0.1539173722267151, "step": 12760 }, { "epoch": 0.3363812672248875, "grad_norm": 3.6446378231048584, "learning_rate": 1.5468504193119636e-05, "loss": 0.18486095666885377, "step": 12770 }, { "epoch": 0.3366446824693862, "grad_norm": 3.175168514251709, "learning_rate": 1.5461359479998953e-05, "loss": 0.15932785272598265, "step": 12780 }, { "epoch": 0.33690809771388497, "grad_norm": 2.618865966796875, "learning_rate": 1.5454210791865213e-05, "loss": 0.17195465564727783, "step": 12790 }, { "epoch": 0.33717151295838366, "grad_norm": 2.303323745727539, "learning_rate": 1.5447058133921552e-05, "loss": 0.17787363529205322, "step": 12800 }, { "epoch": 0.3374349282028824, "grad_norm": 3.608412027359009, "learning_rate": 1.543990151137397e-05, "loss": 0.16813769340515136, "step": 12810 }, { "epoch": 0.33769834344738114, "grad_norm": 1.5908324718475342, "learning_rate": 1.5432740929431372e-05, "loss": 0.15287187099456787, "step": 12820 }, { "epoch": 0.3379617586918799, "grad_norm": 3.7586588859558105, "learning_rate": 1.5425576393305537e-05, "loss": 0.1617181420326233, "step": 12830 }, { "epoch": 0.33822517393637863, "grad_norm": 1.7033060789108276, "learning_rate": 1.5418407908211125e-05, "loss": 0.16495177745819092, "step": 12840 }, { "epoch": 0.3384885891808774, "grad_norm": 2.3892300128936768, "learning_rate": 1.5411235479365664e-05, "loss": 0.1602534055709839, "step": 12850 }, { "epoch": 0.3387520044253761, "grad_norm": 2.3795371055603027, "learning_rate": 1.5404059111989565e-05, "loss": 0.1731044292449951, "step": 12860 }, { "epoch": 0.33901541966987486, "grad_norm": 2.9756791591644287, "learning_rate": 1.5396878811306088e-05, "loss": 0.1695300579071045, "step": 12870 }, { "epoch": 0.3392788349143736, "grad_norm": 1.2918059825897217, "learning_rate": 1.5389694582541376e-05, "loss": 0.1666462540626526, "step": 12880 }, { "epoch": 0.3395422501588723, "grad_norm": 1.074819564819336, "learning_rate": 1.5382506430924414e-05, "loss": 0.16790990829467772, "step": 12890 }, { "epoch": 0.33980566540337104, "grad_norm": 3.790964126586914, "learning_rate": 1.5375314361687053e-05, "loss": 0.17001084089279175, "step": 12900 }, { "epoch": 0.3400690806478698, "grad_norm": 3.6068553924560547, "learning_rate": 1.5368118380063984e-05, "loss": 0.16823533773422242, "step": 12910 }, { "epoch": 0.3403324958923685, "grad_norm": 2.531392812728882, "learning_rate": 1.536091849129276e-05, "loss": 0.17001526355743407, "step": 12920 }, { "epoch": 0.34059591113686727, "grad_norm": 1.2058568000793457, "learning_rate": 1.5353714700613768e-05, "loss": 0.17036560773849488, "step": 12930 }, { "epoch": 0.340859326381366, "grad_norm": 4.65857458114624, "learning_rate": 1.5346507013270237e-05, "loss": 0.16232266426086425, "step": 12940 }, { "epoch": 0.34112274162586476, "grad_norm": 3.9076433181762695, "learning_rate": 1.5339295434508238e-05, "loss": 0.1585991859436035, "step": 12950 }, { "epoch": 0.3413861568703635, "grad_norm": 1.2662116289138794, "learning_rate": 1.533207996957666e-05, "loss": 0.15807099342346193, "step": 12960 }, { "epoch": 0.34164957211486224, "grad_norm": 1.992757797241211, "learning_rate": 1.5324860623727236e-05, "loss": 0.1677384614944458, "step": 12970 }, { "epoch": 0.34191298735936093, "grad_norm": 2.9901909828186035, "learning_rate": 1.5317637402214515e-05, "loss": 0.15979373455047607, "step": 12980 }, { "epoch": 0.3421764026038597, "grad_norm": 1.2495907545089722, "learning_rate": 1.5310410310295876e-05, "loss": 0.14801959991455077, "step": 12990 }, { "epoch": 0.3424398178483584, "grad_norm": 2.6909360885620117, "learning_rate": 1.5303179353231494e-05, "loss": 0.16927385330200195, "step": 13000 }, { "epoch": 0.34270323309285716, "grad_norm": 3.613199472427368, "learning_rate": 1.529594453628439e-05, "loss": 0.15986435413360595, "step": 13010 }, { "epoch": 0.3429666483373559, "grad_norm": 1.9954915046691895, "learning_rate": 1.5288705864720356e-05, "loss": 0.1704043745994568, "step": 13020 }, { "epoch": 0.34323006358185465, "grad_norm": 2.5342087745666504, "learning_rate": 1.528146334380803e-05, "loss": 0.18544801473617553, "step": 13030 }, { "epoch": 0.3434934788263534, "grad_norm": 2.4785356521606445, "learning_rate": 1.5274216978818813e-05, "loss": 0.15762743949890137, "step": 13040 }, { "epoch": 0.34375689407085214, "grad_norm": 5.331356525421143, "learning_rate": 1.526696677502693e-05, "loss": 0.16476960182189943, "step": 13050 }, { "epoch": 0.3440203093153508, "grad_norm": 3.0559747219085693, "learning_rate": 1.525971273770939e-05, "loss": 0.16054863929748536, "step": 13060 }, { "epoch": 0.34428372455984957, "grad_norm": 2.700594663619995, "learning_rate": 1.5252454872146003e-05, "loss": 0.16602880954742433, "step": 13070 }, { "epoch": 0.3445471398043483, "grad_norm": 2.945889472961426, "learning_rate": 1.5245193183619347e-05, "loss": 0.16745076179504395, "step": 13080 }, { "epoch": 0.34481055504884706, "grad_norm": 1.0306801795959473, "learning_rate": 1.5237927677414797e-05, "loss": 0.17200818061828613, "step": 13090 }, { "epoch": 0.3450739702933458, "grad_norm": 1.0870540142059326, "learning_rate": 1.5230658358820493e-05, "loss": 0.16000906229019166, "step": 13100 }, { "epoch": 0.34533738553784454, "grad_norm": 3.0273118019104004, "learning_rate": 1.5223385233127372e-05, "loss": 0.162811017036438, "step": 13110 }, { "epoch": 0.3456008007823433, "grad_norm": 3.225558042526245, "learning_rate": 1.5216108305629123e-05, "loss": 0.15966157913208007, "step": 13120 }, { "epoch": 0.34586421602684203, "grad_norm": 2.732994556427002, "learning_rate": 1.5208827581622205e-05, "loss": 0.18311525583267213, "step": 13130 }, { "epoch": 0.3461276312713408, "grad_norm": 1.5970290899276733, "learning_rate": 1.5201543066405844e-05, "loss": 0.15258222818374634, "step": 13140 }, { "epoch": 0.34639104651583946, "grad_norm": 3.1545090675354004, "learning_rate": 1.5194254765282025e-05, "loss": 0.17747365236282348, "step": 13150 }, { "epoch": 0.3466544617603382, "grad_norm": 1.334242343902588, "learning_rate": 1.5186962683555493e-05, "loss": 0.16341812610626222, "step": 13160 }, { "epoch": 0.34691787700483695, "grad_norm": 1.4305905103683472, "learning_rate": 1.517966682653373e-05, "loss": 0.15201990604400634, "step": 13170 }, { "epoch": 0.3471812922493357, "grad_norm": 3.074871301651001, "learning_rate": 1.5172367199526985e-05, "loss": 0.161749005317688, "step": 13180 }, { "epoch": 0.34744470749383444, "grad_norm": 3.98740553855896, "learning_rate": 1.5165063807848234e-05, "loss": 0.16811532974243165, "step": 13190 }, { "epoch": 0.3477081227383332, "grad_norm": 3.0699303150177, "learning_rate": 1.5157756656813207e-05, "loss": 0.17700860500335694, "step": 13200 }, { "epoch": 0.3479715379828319, "grad_norm": 2.2511894702911377, "learning_rate": 1.5150445751740356e-05, "loss": 0.1583308458328247, "step": 13210 }, { "epoch": 0.34823495322733067, "grad_norm": 1.9364821910858154, "learning_rate": 1.5143131097950882e-05, "loss": 0.15955959558486937, "step": 13220 }, { "epoch": 0.3484983684718294, "grad_norm": 2.5469417572021484, "learning_rate": 1.5135812700768702e-05, "loss": 0.17207159996032714, "step": 13230 }, { "epoch": 0.3487617837163281, "grad_norm": 1.3210556507110596, "learning_rate": 1.512849056552046e-05, "loss": 0.1515653610229492, "step": 13240 }, { "epoch": 0.34902519896082684, "grad_norm": 1.88347589969635, "learning_rate": 1.512116469753552e-05, "loss": 0.16708308458328247, "step": 13250 }, { "epoch": 0.3492886142053256, "grad_norm": 3.035707712173462, "learning_rate": 1.5113835102145968e-05, "loss": 0.16994246244430541, "step": 13260 }, { "epoch": 0.34955202944982433, "grad_norm": 1.7004952430725098, "learning_rate": 1.51065017846866e-05, "loss": 0.16543760299682617, "step": 13270 }, { "epoch": 0.3498154446943231, "grad_norm": 1.5904133319854736, "learning_rate": 1.509916475049492e-05, "loss": 0.15441532135009767, "step": 13280 }, { "epoch": 0.3500788599388218, "grad_norm": 4.090301036834717, "learning_rate": 1.5091824004911137e-05, "loss": 0.19567564725875855, "step": 13290 }, { "epoch": 0.35034227518332056, "grad_norm": 4.454212665557861, "learning_rate": 1.5084479553278164e-05, "loss": 0.158274507522583, "step": 13300 }, { "epoch": 0.3506056904278193, "grad_norm": 1.928676724433899, "learning_rate": 1.5077131400941609e-05, "loss": 0.1536906599998474, "step": 13310 }, { "epoch": 0.350869105672318, "grad_norm": 0.9499024748802185, "learning_rate": 1.5069779553249775e-05, "loss": 0.15964841842651367, "step": 13320 }, { "epoch": 0.35113252091681674, "grad_norm": 2.5423011779785156, "learning_rate": 1.5062424015553653e-05, "loss": 0.18677899837493897, "step": 13330 }, { "epoch": 0.3513959361613155, "grad_norm": 2.6154909133911133, "learning_rate": 1.505506479320692e-05, "loss": 0.15652141571044922, "step": 13340 }, { "epoch": 0.3516593514058142, "grad_norm": 3.0865089893341064, "learning_rate": 1.5047701891565938e-05, "loss": 0.16635675430297853, "step": 13350 }, { "epoch": 0.35192276665031297, "grad_norm": 1.8586972951889038, "learning_rate": 1.504033531598974e-05, "loss": 0.16585408449172973, "step": 13360 }, { "epoch": 0.3521861818948117, "grad_norm": 3.1771247386932373, "learning_rate": 1.5032965071840045e-05, "loss": 0.1660893440246582, "step": 13370 }, { "epoch": 0.35244959713931046, "grad_norm": 1.440751552581787, "learning_rate": 1.5025591164481226e-05, "loss": 0.18377569913864136, "step": 13380 }, { "epoch": 0.3527130123838092, "grad_norm": 3.5726706981658936, "learning_rate": 1.5018213599280339e-05, "loss": 0.1612151861190796, "step": 13390 }, { "epoch": 0.35297642762830794, "grad_norm": 2.3722283840179443, "learning_rate": 1.5010832381607085e-05, "loss": 0.16999001502990724, "step": 13400 }, { "epoch": 0.35323984287280663, "grad_norm": 5.286029815673828, "learning_rate": 1.500344751683384e-05, "loss": 0.16898765563964843, "step": 13410 }, { "epoch": 0.3535032581173054, "grad_norm": 4.881797790527344, "learning_rate": 1.4996059010335626e-05, "loss": 0.17388522624969482, "step": 13420 }, { "epoch": 0.3537666733618041, "grad_norm": 1.1206573247909546, "learning_rate": 1.4988666867490117e-05, "loss": 0.15946290493011475, "step": 13430 }, { "epoch": 0.35403008860630286, "grad_norm": 4.606796741485596, "learning_rate": 1.4981271093677632e-05, "loss": 0.15743587017059327, "step": 13440 }, { "epoch": 0.3542935038508016, "grad_norm": 3.7578399181365967, "learning_rate": 1.4973871694281133e-05, "loss": 0.16415679454803467, "step": 13450 }, { "epoch": 0.35455691909530035, "grad_norm": 3.264176607131958, "learning_rate": 1.4966468674686226e-05, "loss": 0.15727078914642334, "step": 13460 }, { "epoch": 0.3548203343397991, "grad_norm": 4.871591091156006, "learning_rate": 1.4959062040281143e-05, "loss": 0.1678762912750244, "step": 13470 }, { "epoch": 0.35508374958429784, "grad_norm": 1.98879075050354, "learning_rate": 1.4951651796456758e-05, "loss": 0.17919983863830566, "step": 13480 }, { "epoch": 0.3553471648287966, "grad_norm": 1.0140492916107178, "learning_rate": 1.4944237948606561e-05, "loss": 0.15424734354019165, "step": 13490 }, { "epoch": 0.35561058007329527, "grad_norm": 1.123770833015442, "learning_rate": 1.4936820502126676e-05, "loss": 0.15314486026763915, "step": 13500 }, { "epoch": 0.355873995317794, "grad_norm": 3.041717052459717, "learning_rate": 1.4929399462415832e-05, "loss": 0.16122959852218627, "step": 13510 }, { "epoch": 0.35613741056229276, "grad_norm": 1.2382557392120361, "learning_rate": 1.492197483487539e-05, "loss": 0.16041765213012696, "step": 13520 }, { "epoch": 0.3564008258067915, "grad_norm": 2.5167272090911865, "learning_rate": 1.491454662490931e-05, "loss": 0.16682634353637696, "step": 13530 }, { "epoch": 0.35666424105129024, "grad_norm": 1.4623053073883057, "learning_rate": 1.4907114837924168e-05, "loss": 0.16269075870513916, "step": 13540 }, { "epoch": 0.356927656295789, "grad_norm": 1.1592049598693848, "learning_rate": 1.4899679479329133e-05, "loss": 0.1558336138725281, "step": 13550 }, { "epoch": 0.35719107154028773, "grad_norm": 2.5673017501831055, "learning_rate": 1.4892240554535981e-05, "loss": 0.15633227825164794, "step": 13560 }, { "epoch": 0.3574544867847865, "grad_norm": 1.31145441532135, "learning_rate": 1.4884798068959085e-05, "loss": 0.15439966917037964, "step": 13570 }, { "epoch": 0.3577179020292852, "grad_norm": 1.4015440940856934, "learning_rate": 1.4877352028015408e-05, "loss": 0.16521252393722535, "step": 13580 }, { "epoch": 0.3579813172737839, "grad_norm": 1.841369867324829, "learning_rate": 1.4869902437124499e-05, "loss": 0.17203593254089355, "step": 13590 }, { "epoch": 0.35824473251828265, "grad_norm": 1.9540482759475708, "learning_rate": 1.486244930170849e-05, "loss": 0.16097251176834107, "step": 13600 }, { "epoch": 0.3585081477627814, "grad_norm": 2.3398020267486572, "learning_rate": 1.4854992627192096e-05, "loss": 0.17609153985977172, "step": 13610 }, { "epoch": 0.35877156300728014, "grad_norm": 4.124735355377197, "learning_rate": 1.4847532419002602e-05, "loss": 0.16160476207733154, "step": 13620 }, { "epoch": 0.3590349782517789, "grad_norm": 0.8726623058319092, "learning_rate": 1.4840068682569877e-05, "loss": 0.16750503778457643, "step": 13630 }, { "epoch": 0.3592983934962776, "grad_norm": 3.986280679702759, "learning_rate": 1.4832601423326346e-05, "loss": 0.14513174295425416, "step": 13640 }, { "epoch": 0.35956180874077637, "grad_norm": 2.0235490798950195, "learning_rate": 1.4825130646707002e-05, "loss": 0.1424412727355957, "step": 13650 }, { "epoch": 0.3598252239852751, "grad_norm": 2.027879476547241, "learning_rate": 1.4817656358149398e-05, "loss": 0.17259159088134765, "step": 13660 }, { "epoch": 0.3600886392297738, "grad_norm": 0.9624882340431213, "learning_rate": 1.4810178563093648e-05, "loss": 0.17081626653671264, "step": 13670 }, { "epoch": 0.36035205447427254, "grad_norm": 2.1374564170837402, "learning_rate": 1.4802697266982411e-05, "loss": 0.1683843493461609, "step": 13680 }, { "epoch": 0.3606154697187713, "grad_norm": 2.8017427921295166, "learning_rate": 1.4795212475260896e-05, "loss": 0.16582261323928832, "step": 13690 }, { "epoch": 0.36087888496327003, "grad_norm": 1.4509450197219849, "learning_rate": 1.4787724193376858e-05, "loss": 0.15269869565963745, "step": 13700 }, { "epoch": 0.3611423002077688, "grad_norm": 3.819132089614868, "learning_rate": 1.4780232426780592e-05, "loss": 0.16530723571777345, "step": 13710 }, { "epoch": 0.3614057154522675, "grad_norm": 1.4769039154052734, "learning_rate": 1.477273718092493e-05, "loss": 0.16488196849822997, "step": 13720 }, { "epoch": 0.36166913069676626, "grad_norm": 1.7215454578399658, "learning_rate": 1.4765238461265235e-05, "loss": 0.16765968799591063, "step": 13730 }, { "epoch": 0.361932545941265, "grad_norm": 3.1199803352355957, "learning_rate": 1.4757736273259396e-05, "loss": 0.1768568992614746, "step": 13740 }, { "epoch": 0.36219596118576375, "grad_norm": 1.9606221914291382, "learning_rate": 1.4750230622367831e-05, "loss": 0.16254266500473022, "step": 13750 }, { "epoch": 0.36245937643026244, "grad_norm": 3.9938268661499023, "learning_rate": 1.474272151405348e-05, "loss": 0.16599817276000978, "step": 13760 }, { "epoch": 0.3627227916747612, "grad_norm": 2.2239859104156494, "learning_rate": 1.473520895378179e-05, "loss": 0.1690974712371826, "step": 13770 }, { "epoch": 0.3629862069192599, "grad_norm": 2.741199254989624, "learning_rate": 1.472769294702073e-05, "loss": 0.174873685836792, "step": 13780 }, { "epoch": 0.36324962216375867, "grad_norm": 5.916696071624756, "learning_rate": 1.4720173499240771e-05, "loss": 0.16812121868133545, "step": 13790 }, { "epoch": 0.3635130374082574, "grad_norm": 2.330273389816284, "learning_rate": 1.4712650615914899e-05, "loss": 0.16963601112365723, "step": 13800 }, { "epoch": 0.36377645265275615, "grad_norm": 3.470391035079956, "learning_rate": 1.4705124302518581e-05, "loss": 0.18198015689849853, "step": 13810 }, { "epoch": 0.3640398678972549, "grad_norm": 2.7970104217529297, "learning_rate": 1.4697594564529805e-05, "loss": 0.1764854907989502, "step": 13820 }, { "epoch": 0.36430328314175364, "grad_norm": 0.9528728127479553, "learning_rate": 1.4690061407429028e-05, "loss": 0.1647954225540161, "step": 13830 }, { "epoch": 0.3645666983862524, "grad_norm": 2.35282301902771, "learning_rate": 1.468252483669922e-05, "loss": 0.17824809551239013, "step": 13840 }, { "epoch": 0.3648301136307511, "grad_norm": 1.256014347076416, "learning_rate": 1.4674984857825808e-05, "loss": 0.17479848861694336, "step": 13850 }, { "epoch": 0.3650935288752498, "grad_norm": 3.942803382873535, "learning_rate": 1.4667441476296723e-05, "loss": 0.1705479860305786, "step": 13860 }, { "epoch": 0.36535694411974856, "grad_norm": 1.1489735841751099, "learning_rate": 1.4659894697602358e-05, "loss": 0.1649183750152588, "step": 13870 }, { "epoch": 0.3656203593642473, "grad_norm": 3.187908411026001, "learning_rate": 1.465234452723559e-05, "loss": 0.1600733518600464, "step": 13880 }, { "epoch": 0.36588377460874605, "grad_norm": 2.4962048530578613, "learning_rate": 1.4644790970691755e-05, "loss": 0.15309834480285645, "step": 13890 }, { "epoch": 0.3661471898532448, "grad_norm": 2.705753803253174, "learning_rate": 1.4637234033468658e-05, "loss": 0.16826821565628053, "step": 13900 }, { "epoch": 0.36641060509774354, "grad_norm": 1.0322003364562988, "learning_rate": 1.4629673721066566e-05, "loss": 0.1648475408554077, "step": 13910 }, { "epoch": 0.3666740203422423, "grad_norm": 2.6366512775421143, "learning_rate": 1.4622110038988196e-05, "loss": 0.17266156673431396, "step": 13920 }, { "epoch": 0.36693743558674097, "grad_norm": 1.3335579633712769, "learning_rate": 1.4614542992738727e-05, "loss": 0.17629343271255493, "step": 13930 }, { "epoch": 0.3672008508312397, "grad_norm": 1.0502954721450806, "learning_rate": 1.4606972587825779e-05, "loss": 0.16633857488632203, "step": 13940 }, { "epoch": 0.36746426607573845, "grad_norm": 2.8611292839050293, "learning_rate": 1.4599398829759419e-05, "loss": 0.16196240186691285, "step": 13950 }, { "epoch": 0.3677276813202372, "grad_norm": 1.053652286529541, "learning_rate": 1.4591821724052155e-05, "loss": 0.15830824375152588, "step": 13960 }, { "epoch": 0.36799109656473594, "grad_norm": 1.2588666677474976, "learning_rate": 1.4584241276218935e-05, "loss": 0.17136363983154296, "step": 13970 }, { "epoch": 0.3682545118092347, "grad_norm": 1.1531388759613037, "learning_rate": 1.4576657491777134e-05, "loss": 0.17173981666564941, "step": 13980 }, { "epoch": 0.36851792705373343, "grad_norm": 2.125673294067383, "learning_rate": 1.4569070376246552e-05, "loss": 0.16820569038391114, "step": 13990 }, { "epoch": 0.3687813422982322, "grad_norm": 1.5968555212020874, "learning_rate": 1.4561479935149426e-05, "loss": 0.16500096321105956, "step": 14000 }, { "epoch": 0.3690447575427309, "grad_norm": 3.06333327293396, "learning_rate": 1.4553886174010405e-05, "loss": 0.1798495054244995, "step": 14010 }, { "epoch": 0.3693081727872296, "grad_norm": 0.6711460947990417, "learning_rate": 1.4546289098356556e-05, "loss": 0.15132505893707277, "step": 14020 }, { "epoch": 0.36957158803172835, "grad_norm": 1.0251543521881104, "learning_rate": 1.4538688713717357e-05, "loss": 0.15507135391235352, "step": 14030 }, { "epoch": 0.3698350032762271, "grad_norm": 4.651401996612549, "learning_rate": 1.4531085025624694e-05, "loss": 0.17015016078948975, "step": 14040 }, { "epoch": 0.37009841852072584, "grad_norm": 1.3403618335723877, "learning_rate": 1.4523478039612863e-05, "loss": 0.18107739686965943, "step": 14050 }, { "epoch": 0.3703618337652246, "grad_norm": 1.8202483654022217, "learning_rate": 1.4515867761218558e-05, "loss": 0.1644890785217285, "step": 14060 }, { "epoch": 0.3706252490097233, "grad_norm": 3.001830577850342, "learning_rate": 1.4508254195980861e-05, "loss": 0.1566838264465332, "step": 14070 }, { "epoch": 0.37088866425422207, "grad_norm": 1.7589973211288452, "learning_rate": 1.450063734944126e-05, "loss": 0.17618714570999144, "step": 14080 }, { "epoch": 0.3711520794987208, "grad_norm": 1.6754298210144043, "learning_rate": 1.449301722714362e-05, "loss": 0.16729092597961426, "step": 14090 }, { "epoch": 0.37141549474321955, "grad_norm": 2.4304423332214355, "learning_rate": 1.4485393834634202e-05, "loss": 0.16778470277786256, "step": 14100 }, { "epoch": 0.37167890998771824, "grad_norm": 2.76680850982666, "learning_rate": 1.447776717746163e-05, "loss": 0.16145991086959838, "step": 14110 }, { "epoch": 0.371942325232217, "grad_norm": 1.4182759523391724, "learning_rate": 1.4470137261176922e-05, "loss": 0.16668018102645873, "step": 14120 }, { "epoch": 0.37220574047671573, "grad_norm": 4.476673126220703, "learning_rate": 1.4462504091333454e-05, "loss": 0.1647454857826233, "step": 14130 }, { "epoch": 0.3724691557212145, "grad_norm": 2.1962392330169678, "learning_rate": 1.4454867673486983e-05, "loss": 0.16595137119293213, "step": 14140 }, { "epoch": 0.3727325709657132, "grad_norm": 1.4183558225631714, "learning_rate": 1.4447228013195616e-05, "loss": 0.1671093463897705, "step": 14150 }, { "epoch": 0.37299598621021196, "grad_norm": 0.9042845368385315, "learning_rate": 1.4439585116019835e-05, "loss": 0.16464745998382568, "step": 14160 }, { "epoch": 0.3732594014547107, "grad_norm": 2.5350747108459473, "learning_rate": 1.4431938987522463e-05, "loss": 0.1694819450378418, "step": 14170 }, { "epoch": 0.37352281669920945, "grad_norm": 2.308405876159668, "learning_rate": 1.4424289633268689e-05, "loss": 0.1744166612625122, "step": 14180 }, { "epoch": 0.37378623194370814, "grad_norm": 1.574723720550537, "learning_rate": 1.4416637058826038e-05, "loss": 0.15501210689544678, "step": 14190 }, { "epoch": 0.3740496471882069, "grad_norm": 2.8200619220733643, "learning_rate": 1.4408981269764386e-05, "loss": 0.1668046236038208, "step": 14200 }, { "epoch": 0.3743130624327056, "grad_norm": 2.9689128398895264, "learning_rate": 1.4401322271655947e-05, "loss": 0.16351659297943116, "step": 14210 }, { "epoch": 0.37457647767720437, "grad_norm": 1.7798763513565063, "learning_rate": 1.4393660070075267e-05, "loss": 0.1776472210884094, "step": 14220 }, { "epoch": 0.3748398929217031, "grad_norm": 3.416825532913208, "learning_rate": 1.4385994670599233e-05, "loss": 0.16080756187438966, "step": 14230 }, { "epoch": 0.37510330816620185, "grad_norm": 1.8663899898529053, "learning_rate": 1.4378326078807048e-05, "loss": 0.16305481195449828, "step": 14240 }, { "epoch": 0.3753667234107006, "grad_norm": 1.2076812982559204, "learning_rate": 1.4370654300280246e-05, "loss": 0.17604517936706543, "step": 14250 }, { "epoch": 0.37563013865519934, "grad_norm": 2.0026915073394775, "learning_rate": 1.4362979340602677e-05, "loss": 0.1712841272354126, "step": 14260 }, { "epoch": 0.3758935538996981, "grad_norm": 0.8673966526985168, "learning_rate": 1.4355301205360513e-05, "loss": 0.16524934768676758, "step": 14270 }, { "epoch": 0.3761569691441968, "grad_norm": 0.8770243525505066, "learning_rate": 1.4347619900142223e-05, "loss": 0.19150779247283936, "step": 14280 }, { "epoch": 0.3764203843886955, "grad_norm": 2.3936960697174072, "learning_rate": 1.43399354305386e-05, "loss": 0.15141639709472657, "step": 14290 }, { "epoch": 0.37668379963319426, "grad_norm": 4.905852317810059, "learning_rate": 1.4332247802142729e-05, "loss": 0.18192846775054933, "step": 14300 }, { "epoch": 0.376947214877693, "grad_norm": 1.2279235124588013, "learning_rate": 1.4324557020550001e-05, "loss": 0.16421202421188355, "step": 14310 }, { "epoch": 0.37721063012219175, "grad_norm": 1.375461459159851, "learning_rate": 1.4316863091358093e-05, "loss": 0.15609469413757324, "step": 14320 }, { "epoch": 0.3774740453666905, "grad_norm": 1.6511584520339966, "learning_rate": 1.4309166020166984e-05, "loss": 0.14448850154876708, "step": 14330 }, { "epoch": 0.37773746061118924, "grad_norm": 1.5993614196777344, "learning_rate": 1.430146581257893e-05, "loss": 0.16564590930938722, "step": 14340 }, { "epoch": 0.378000875855688, "grad_norm": 3.025562047958374, "learning_rate": 1.4293762474198478e-05, "loss": 0.16440365314483643, "step": 14350 }, { "epoch": 0.3782642911001867, "grad_norm": 2.618290901184082, "learning_rate": 1.428605601063245e-05, "loss": 0.1612714171409607, "step": 14360 }, { "epoch": 0.3785277063446854, "grad_norm": 1.3728407621383667, "learning_rate": 1.427834642748994e-05, "loss": 0.15818169116973876, "step": 14370 }, { "epoch": 0.37879112158918415, "grad_norm": 2.248598575592041, "learning_rate": 1.4270633730382315e-05, "loss": 0.17638970613479615, "step": 14380 }, { "epoch": 0.3790545368336829, "grad_norm": 3.1680471897125244, "learning_rate": 1.4262917924923212e-05, "loss": 0.15515754222869874, "step": 14390 }, { "epoch": 0.37931795207818164, "grad_norm": 2.981949806213379, "learning_rate": 1.4255199016728527e-05, "loss": 0.16106157302856444, "step": 14400 }, { "epoch": 0.3795813673226804, "grad_norm": 2.2901713848114014, "learning_rate": 1.4247477011416409e-05, "loss": 0.16530077457427977, "step": 14410 }, { "epoch": 0.37984478256717913, "grad_norm": 0.760411262512207, "learning_rate": 1.4239751914607274e-05, "loss": 0.1669844627380371, "step": 14420 }, { "epoch": 0.3801081978116779, "grad_norm": 3.668520450592041, "learning_rate": 1.4232023731923772e-05, "loss": 0.14987165927886964, "step": 14430 }, { "epoch": 0.3803716130561766, "grad_norm": 1.5040769577026367, "learning_rate": 1.422429246899082e-05, "loss": 0.17430546283721923, "step": 14440 }, { "epoch": 0.3806350283006753, "grad_norm": 3.5190494060516357, "learning_rate": 1.4216558131435551e-05, "loss": 0.17292602062225343, "step": 14450 }, { "epoch": 0.38089844354517405, "grad_norm": 2.506058931350708, "learning_rate": 1.420882072488736e-05, "loss": 0.16889843940734864, "step": 14460 }, { "epoch": 0.3811618587896728, "grad_norm": 2.9722089767456055, "learning_rate": 1.420108025497786e-05, "loss": 0.17683342695236207, "step": 14470 }, { "epoch": 0.38142527403417154, "grad_norm": 1.7310631275177002, "learning_rate": 1.41933367273409e-05, "loss": 0.16167910099029542, "step": 14480 }, { "epoch": 0.3816886892786703, "grad_norm": 1.3145440816879272, "learning_rate": 1.4185590147612555e-05, "loss": 0.16318910121917723, "step": 14490 }, { "epoch": 0.381952104523169, "grad_norm": 5.19720983505249, "learning_rate": 1.4177840521431117e-05, "loss": 0.16141180992126464, "step": 14500 }, { "epoch": 0.38221551976766777, "grad_norm": 5.091682434082031, "learning_rate": 1.4170087854437102e-05, "loss": 0.15133132934570312, "step": 14510 }, { "epoch": 0.3824789350121665, "grad_norm": 3.4646193981170654, "learning_rate": 1.4162332152273233e-05, "loss": 0.17873475551605225, "step": 14520 }, { "epoch": 0.38274235025666525, "grad_norm": 3.3313496112823486, "learning_rate": 1.4154573420584448e-05, "loss": 0.15732795000076294, "step": 14530 }, { "epoch": 0.38300576550116394, "grad_norm": 1.4589215517044067, "learning_rate": 1.414681166501788e-05, "loss": 0.16770553588867188, "step": 14540 }, { "epoch": 0.3832691807456627, "grad_norm": 1.7002211809158325, "learning_rate": 1.4139046891222877e-05, "loss": 0.1772599697113037, "step": 14550 }, { "epoch": 0.38353259599016143, "grad_norm": 1.591063141822815, "learning_rate": 1.4131279104850968e-05, "loss": 0.19260685443878173, "step": 14560 }, { "epoch": 0.3837960112346602, "grad_norm": 0.9814028143882751, "learning_rate": 1.4123508311555894e-05, "loss": 0.1612797975540161, "step": 14570 }, { "epoch": 0.3840594264791589, "grad_norm": 1.9824777841567993, "learning_rate": 1.4115734516993562e-05, "loss": 0.17878903150558473, "step": 14580 }, { "epoch": 0.38432284172365766, "grad_norm": 5.497615337371826, "learning_rate": 1.4107957726822084e-05, "loss": 0.1732774019241333, "step": 14590 }, { "epoch": 0.3845862569681564, "grad_norm": 4.366610050201416, "learning_rate": 1.4100177946701738e-05, "loss": 0.17304242849349977, "step": 14600 }, { "epoch": 0.38484967221265515, "grad_norm": 1.3526524305343628, "learning_rate": 1.4092395182294989e-05, "loss": 0.17076704502105713, "step": 14610 }, { "epoch": 0.3851130874571539, "grad_norm": 2.159681797027588, "learning_rate": 1.4084609439266464e-05, "loss": 0.17609138488769532, "step": 14620 }, { "epoch": 0.3853765027016526, "grad_norm": 2.2721500396728516, "learning_rate": 1.4076820723282962e-05, "loss": 0.16387999057769775, "step": 14630 }, { "epoch": 0.3856399179461513, "grad_norm": 1.195629358291626, "learning_rate": 1.4069029040013448e-05, "loss": 0.18718419075012208, "step": 14640 }, { "epoch": 0.38590333319065007, "grad_norm": 2.0035507678985596, "learning_rate": 1.4061234395129049e-05, "loss": 0.17593241930007936, "step": 14650 }, { "epoch": 0.3861667484351488, "grad_norm": 2.093315839767456, "learning_rate": 1.405343679430304e-05, "loss": 0.1682530641555786, "step": 14660 }, { "epoch": 0.38643016367964755, "grad_norm": 1.5969064235687256, "learning_rate": 1.4045636243210852e-05, "loss": 0.17491557598114013, "step": 14670 }, { "epoch": 0.3866935789241463, "grad_norm": 1.0703272819519043, "learning_rate": 1.4037832747530063e-05, "loss": 0.15064680576324463, "step": 14680 }, { "epoch": 0.38695699416864504, "grad_norm": 1.4022960662841797, "learning_rate": 1.403002631294039e-05, "loss": 0.1619959831237793, "step": 14690 }, { "epoch": 0.3872204094131438, "grad_norm": 2.0651583671569824, "learning_rate": 1.40222169451237e-05, "loss": 0.15016932487487794, "step": 14700 }, { "epoch": 0.3874838246576425, "grad_norm": 1.478929877281189, "learning_rate": 1.4014404649763983e-05, "loss": 0.17101845741271973, "step": 14710 }, { "epoch": 0.3877472399021412, "grad_norm": 3.122354507446289, "learning_rate": 1.4006589432547367e-05, "loss": 0.16244759559631347, "step": 14720 }, { "epoch": 0.38801065514663996, "grad_norm": 3.3151748180389404, "learning_rate": 1.3998771299162099e-05, "loss": 0.1681267499923706, "step": 14730 }, { "epoch": 0.3882740703911387, "grad_norm": 2.602635145187378, "learning_rate": 1.3990950255298565e-05, "loss": 0.16605570316314697, "step": 14740 }, { "epoch": 0.38853748563563745, "grad_norm": 1.1204966306686401, "learning_rate": 1.3983126306649243e-05, "loss": 0.17280709743499756, "step": 14750 }, { "epoch": 0.3888009008801362, "grad_norm": 4.019490718841553, "learning_rate": 1.3975299458908753e-05, "loss": 0.15635888576507567, "step": 14760 }, { "epoch": 0.38906431612463493, "grad_norm": 2.305058002471924, "learning_rate": 1.3967469717773807e-05, "loss": 0.16733275651931762, "step": 14770 }, { "epoch": 0.3893277313691337, "grad_norm": 1.7258644104003906, "learning_rate": 1.395963708894323e-05, "loss": 0.1608535408973694, "step": 14780 }, { "epoch": 0.3895911466136324, "grad_norm": 3.325407028198242, "learning_rate": 1.3951801578117946e-05, "loss": 0.1735924243927002, "step": 14790 }, { "epoch": 0.3898545618581311, "grad_norm": 1.1585524082183838, "learning_rate": 1.3943963191000976e-05, "loss": 0.1626773953437805, "step": 14800 }, { "epoch": 0.39011797710262985, "grad_norm": 3.2905163764953613, "learning_rate": 1.393612193329744e-05, "loss": 0.18178069591522217, "step": 14810 }, { "epoch": 0.3903813923471286, "grad_norm": 4.4113450050354, "learning_rate": 1.3928277810714542e-05, "loss": 0.15989940166473388, "step": 14820 }, { "epoch": 0.39064480759162734, "grad_norm": 2.849717855453491, "learning_rate": 1.3920430828961576e-05, "loss": 0.1597743511199951, "step": 14830 }, { "epoch": 0.3909082228361261, "grad_norm": 2.1004929542541504, "learning_rate": 1.3912580993749907e-05, "loss": 0.16907172203063964, "step": 14840 }, { "epoch": 0.39117163808062483, "grad_norm": 2.5157997608184814, "learning_rate": 1.3904728310792988e-05, "loss": 0.17417107820510863, "step": 14850 }, { "epoch": 0.39143505332512357, "grad_norm": 1.859971284866333, "learning_rate": 1.3896872785806342e-05, "loss": 0.1581346035003662, "step": 14860 }, { "epoch": 0.3916984685696223, "grad_norm": 3.9845004081726074, "learning_rate": 1.3889014424507557e-05, "loss": 0.16312024593353272, "step": 14870 }, { "epoch": 0.39196188381412106, "grad_norm": 1.5912668704986572, "learning_rate": 1.3881153232616284e-05, "loss": 0.16571998596191406, "step": 14880 }, { "epoch": 0.39222529905861975, "grad_norm": 1.6515403985977173, "learning_rate": 1.3873289215854247e-05, "loss": 0.16461939811706544, "step": 14890 }, { "epoch": 0.3924887143031185, "grad_norm": 2.989976406097412, "learning_rate": 1.3865422379945205e-05, "loss": 0.15174756050109864, "step": 14900 }, { "epoch": 0.39275212954761723, "grad_norm": 1.4490909576416016, "learning_rate": 1.3857552730614989e-05, "loss": 0.18400250673294066, "step": 14910 }, { "epoch": 0.393015544792116, "grad_norm": 1.3577688932418823, "learning_rate": 1.3849680273591468e-05, "loss": 0.16901054382324218, "step": 14920 }, { "epoch": 0.3932789600366147, "grad_norm": 1.6493719816207886, "learning_rate": 1.3841805014604553e-05, "loss": 0.18169076442718507, "step": 14930 }, { "epoch": 0.39354237528111347, "grad_norm": 4.364186763763428, "learning_rate": 1.3833926959386199e-05, "loss": 0.17136571407318116, "step": 14940 }, { "epoch": 0.3938057905256122, "grad_norm": 3.0021462440490723, "learning_rate": 1.3826046113670395e-05, "loss": 0.17324841022491455, "step": 14950 }, { "epoch": 0.39406920577011095, "grad_norm": 1.2278276681900024, "learning_rate": 1.3818162483193159e-05, "loss": 0.16994601488113403, "step": 14960 }, { "epoch": 0.39433262101460964, "grad_norm": 1.654066801071167, "learning_rate": 1.3810276073692537e-05, "loss": 0.15472745895385742, "step": 14970 }, { "epoch": 0.3945960362591084, "grad_norm": 1.3889390230178833, "learning_rate": 1.3802386890908602e-05, "loss": 0.16282424926757813, "step": 14980 }, { "epoch": 0.39485945150360713, "grad_norm": 2.19498348236084, "learning_rate": 1.3794494940583437e-05, "loss": 0.16643836498260497, "step": 14990 }, { "epoch": 0.39512286674810587, "grad_norm": 1.8637348413467407, "learning_rate": 1.3786600228461147e-05, "loss": 0.17420953512191772, "step": 15000 }, { "epoch": 0.3953862819926046, "grad_norm": 3.4262003898620605, "learning_rate": 1.377870276028784e-05, "loss": 0.16000287532806395, "step": 15010 }, { "epoch": 0.39564969723710336, "grad_norm": 2.0908265113830566, "learning_rate": 1.377080254181164e-05, "loss": 0.17581336498260497, "step": 15020 }, { "epoch": 0.3959131124816021, "grad_norm": 1.0547244548797607, "learning_rate": 1.3762899578782662e-05, "loss": 0.1752563714981079, "step": 15030 }, { "epoch": 0.39617652772610085, "grad_norm": 3.8478336334228516, "learning_rate": 1.375499387695303e-05, "loss": 0.1681584119796753, "step": 15040 }, { "epoch": 0.3964399429705996, "grad_norm": 1.8298532962799072, "learning_rate": 1.3747085442076844e-05, "loss": 0.16229665279388428, "step": 15050 }, { "epoch": 0.3967033582150983, "grad_norm": 1.8491822481155396, "learning_rate": 1.3739174279910214e-05, "loss": 0.1653910756111145, "step": 15060 }, { "epoch": 0.396966773459597, "grad_norm": 1.7427021265029907, "learning_rate": 1.3731260396211219e-05, "loss": 0.15263838768005372, "step": 15070 }, { "epoch": 0.39723018870409577, "grad_norm": 3.9203860759735107, "learning_rate": 1.372334379673993e-05, "loss": 0.16617746353149415, "step": 15080 }, { "epoch": 0.3974936039485945, "grad_norm": 1.2799631357192993, "learning_rate": 1.3715424487258386e-05, "loss": 0.16147689819335936, "step": 15090 }, { "epoch": 0.39775701919309325, "grad_norm": 2.919062852859497, "learning_rate": 1.37075024735306e-05, "loss": 0.1741289496421814, "step": 15100 }, { "epoch": 0.398020434437592, "grad_norm": 1.630544662475586, "learning_rate": 1.3699577761322563e-05, "loss": 0.16854658126831054, "step": 15110 }, { "epoch": 0.39828384968209074, "grad_norm": 1.699919581413269, "learning_rate": 1.3691650356402212e-05, "loss": 0.17027909755706788, "step": 15120 }, { "epoch": 0.3985472649265895, "grad_norm": 1.1397665739059448, "learning_rate": 1.3683720264539469e-05, "loss": 0.150761616230011, "step": 15130 }, { "epoch": 0.3988106801710882, "grad_norm": 2.1830196380615234, "learning_rate": 1.3675787491506184e-05, "loss": 0.1621282458305359, "step": 15140 }, { "epoch": 0.3990740954155869, "grad_norm": 2.8384883403778076, "learning_rate": 1.3667852043076173e-05, "loss": 0.16317250728607177, "step": 15150 }, { "epoch": 0.39933751066008566, "grad_norm": 3.3046116828918457, "learning_rate": 1.3659913925025203e-05, "loss": 0.17266368865966797, "step": 15160 }, { "epoch": 0.3996009259045844, "grad_norm": 3.8348195552825928, "learning_rate": 1.3651973143130979e-05, "loss": 0.16038193702697753, "step": 15170 }, { "epoch": 0.39986434114908315, "grad_norm": 2.353288173675537, "learning_rate": 1.3644029703173139e-05, "loss": 0.151317036151886, "step": 15180 }, { "epoch": 0.4001277563935819, "grad_norm": 1.271840214729309, "learning_rate": 1.3636083610933262e-05, "loss": 0.16768852472305298, "step": 15190 }, { "epoch": 0.40039117163808063, "grad_norm": 2.9056429862976074, "learning_rate": 1.3628134872194861e-05, "loss": 0.1600516676902771, "step": 15200 }, { "epoch": 0.4006545868825794, "grad_norm": 1.8164604902267456, "learning_rate": 1.362018349274337e-05, "loss": 0.1596779227256775, "step": 15210 }, { "epoch": 0.4009180021270781, "grad_norm": 3.1642425060272217, "learning_rate": 1.3612229478366144e-05, "loss": 0.15835285186767578, "step": 15220 }, { "epoch": 0.4011814173715768, "grad_norm": 3.2881405353546143, "learning_rate": 1.360427283485246e-05, "loss": 0.17056984901428224, "step": 15230 }, { "epoch": 0.40144483261607555, "grad_norm": 0.8802847862243652, "learning_rate": 1.3596313567993504e-05, "loss": 0.17495627403259278, "step": 15240 }, { "epoch": 0.4017082478605743, "grad_norm": 1.6504604816436768, "learning_rate": 1.3588351683582374e-05, "loss": 0.16008588075637817, "step": 15250 }, { "epoch": 0.40197166310507304, "grad_norm": 1.8120076656341553, "learning_rate": 1.3580387187414077e-05, "loss": 0.15944665670394897, "step": 15260 }, { "epoch": 0.4022350783495718, "grad_norm": 1.9405163526535034, "learning_rate": 1.3572420085285512e-05, "loss": 0.15447018146514893, "step": 15270 }, { "epoch": 0.4024984935940705, "grad_norm": 2.906134605407715, "learning_rate": 1.3564450382995482e-05, "loss": 0.16640176773071289, "step": 15280 }, { "epoch": 0.40276190883856927, "grad_norm": 1.4905495643615723, "learning_rate": 1.3556478086344678e-05, "loss": 0.15556095838546752, "step": 15290 }, { "epoch": 0.403025324083068, "grad_norm": 3.2023470401763916, "learning_rate": 1.3548503201135687e-05, "loss": 0.14305920600891114, "step": 15300 }, { "epoch": 0.40328873932756676, "grad_norm": 1.9550646543502808, "learning_rate": 1.3540525733172966e-05, "loss": 0.1758166790008545, "step": 15310 }, { "epoch": 0.40355215457206545, "grad_norm": 1.7308063507080078, "learning_rate": 1.353254568826287e-05, "loss": 0.1679958701133728, "step": 15320 }, { "epoch": 0.4038155698165642, "grad_norm": 2.2797317504882812, "learning_rate": 1.3524563072213612e-05, "loss": 0.1702889919281006, "step": 15330 }, { "epoch": 0.40407898506106293, "grad_norm": 1.95553457736969, "learning_rate": 1.3516577890835292e-05, "loss": 0.16023459434509277, "step": 15340 }, { "epoch": 0.4043424003055617, "grad_norm": 1.3306602239608765, "learning_rate": 1.350859014993986e-05, "loss": 0.17195043563842774, "step": 15350 }, { "epoch": 0.4046058155500604, "grad_norm": 1.5247278213500977, "learning_rate": 1.3500599855341146e-05, "loss": 0.164687180519104, "step": 15360 }, { "epoch": 0.40486923079455917, "grad_norm": 3.732239246368408, "learning_rate": 1.3492607012854824e-05, "loss": 0.16813154220581056, "step": 15370 }, { "epoch": 0.4051326460390579, "grad_norm": 1.111462950706482, "learning_rate": 1.3484611628298438e-05, "loss": 0.14615514278411865, "step": 15380 }, { "epoch": 0.40539606128355665, "grad_norm": 5.22926664352417, "learning_rate": 1.3476613707491372e-05, "loss": 0.1616245150566101, "step": 15390 }, { "epoch": 0.4056594765280554, "grad_norm": 1.068763256072998, "learning_rate": 1.3468613256254852e-05, "loss": 0.15503045320510864, "step": 15400 }, { "epoch": 0.4059228917725541, "grad_norm": 1.6430211067199707, "learning_rate": 1.3460610280411957e-05, "loss": 0.17036550045013427, "step": 15410 }, { "epoch": 0.40618630701705283, "grad_norm": 2.660526990890503, "learning_rate": 1.3452604785787595e-05, "loss": 0.16559982299804688, "step": 15420 }, { "epoch": 0.40644972226155157, "grad_norm": 2.981189489364624, "learning_rate": 1.3444596778208516e-05, "loss": 0.16689577102661132, "step": 15430 }, { "epoch": 0.4067131375060503, "grad_norm": 2.5699052810668945, "learning_rate": 1.3436586263503286e-05, "loss": 0.16025055646896363, "step": 15440 }, { "epoch": 0.40697655275054906, "grad_norm": 1.784273624420166, "learning_rate": 1.3428573247502307e-05, "loss": 0.17597622871398927, "step": 15450 }, { "epoch": 0.4072399679950478, "grad_norm": 4.353977680206299, "learning_rate": 1.3420557736037796e-05, "loss": 0.1674705982208252, "step": 15460 }, { "epoch": 0.40750338323954655, "grad_norm": 1.382250428199768, "learning_rate": 1.3412539734943792e-05, "loss": 0.14594167470932007, "step": 15470 }, { "epoch": 0.4077667984840453, "grad_norm": 1.1939588785171509, "learning_rate": 1.3404519250056139e-05, "loss": 0.15441718101501464, "step": 15480 }, { "epoch": 0.408030213728544, "grad_norm": 1.8973113298416138, "learning_rate": 1.3396496287212491e-05, "loss": 0.15494420528411865, "step": 15490 }, { "epoch": 0.4082936289730427, "grad_norm": 4.011678218841553, "learning_rate": 1.3388470852252307e-05, "loss": 0.1696608543395996, "step": 15500 }, { "epoch": 0.40855704421754147, "grad_norm": 1.4695079326629639, "learning_rate": 1.3380442951016846e-05, "loss": 0.15011262893676758, "step": 15510 }, { "epoch": 0.4088204594620402, "grad_norm": 3.2266299724578857, "learning_rate": 1.3372412589349164e-05, "loss": 0.16586943864822387, "step": 15520 }, { "epoch": 0.40908387470653895, "grad_norm": 2.3294894695281982, "learning_rate": 1.3364379773094099e-05, "loss": 0.16731526851654052, "step": 15530 }, { "epoch": 0.4093472899510377, "grad_norm": 2.086106061935425, "learning_rate": 1.335634450809828e-05, "loss": 0.17382440567016602, "step": 15540 }, { "epoch": 0.40961070519553644, "grad_norm": 1.949090600013733, "learning_rate": 1.3348306800210127e-05, "loss": 0.1698546051979065, "step": 15550 }, { "epoch": 0.4098741204400352, "grad_norm": 2.2009551525115967, "learning_rate": 1.3340266655279828e-05, "loss": 0.15916380882263184, "step": 15560 }, { "epoch": 0.4101375356845339, "grad_norm": 0.9065308570861816, "learning_rate": 1.3332224079159342e-05, "loss": 0.15913442373275757, "step": 15570 }, { "epoch": 0.4104009509290326, "grad_norm": 0.9145050644874573, "learning_rate": 1.3324179077702408e-05, "loss": 0.16140646934509278, "step": 15580 }, { "epoch": 0.41066436617353136, "grad_norm": 4.454693794250488, "learning_rate": 1.3316131656764523e-05, "loss": 0.14530831575393677, "step": 15590 }, { "epoch": 0.4109277814180301, "grad_norm": 1.473672866821289, "learning_rate": 1.3308081822202953e-05, "loss": 0.16854504346847535, "step": 15600 }, { "epoch": 0.41119119666252885, "grad_norm": 3.2901129722595215, "learning_rate": 1.3300029579876705e-05, "loss": 0.1797114133834839, "step": 15610 }, { "epoch": 0.4114546119070276, "grad_norm": 2.246598720550537, "learning_rate": 1.3291974935646554e-05, "loss": 0.16479771137237548, "step": 15620 }, { "epoch": 0.41171802715152633, "grad_norm": 1.519062876701355, "learning_rate": 1.3283917895375018e-05, "loss": 0.15549973249435425, "step": 15630 }, { "epoch": 0.4119814423960251, "grad_norm": 1.681406021118164, "learning_rate": 1.327585846492636e-05, "loss": 0.15100445747375488, "step": 15640 }, { "epoch": 0.4122448576405238, "grad_norm": 4.648960113525391, "learning_rate": 1.3267796650166576e-05, "loss": 0.16888148784637452, "step": 15650 }, { "epoch": 0.41250827288502256, "grad_norm": 1.1525909900665283, "learning_rate": 1.3259732456963406e-05, "loss": 0.17142386436462403, "step": 15660 }, { "epoch": 0.41277168812952125, "grad_norm": 2.574064254760742, "learning_rate": 1.3251665891186313e-05, "loss": 0.16495436429977417, "step": 15670 }, { "epoch": 0.41303510337402, "grad_norm": 2.3627421855926514, "learning_rate": 1.32435969587065e-05, "loss": 0.16349687576293945, "step": 15680 }, { "epoch": 0.41329851861851874, "grad_norm": 1.589638113975525, "learning_rate": 1.323552566539688e-05, "loss": 0.16826446056365968, "step": 15690 }, { "epoch": 0.4135619338630175, "grad_norm": 3.3698439598083496, "learning_rate": 1.3227452017132085e-05, "loss": 0.1671851396560669, "step": 15700 }, { "epoch": 0.4138253491075162, "grad_norm": 4.436885356903076, "learning_rate": 1.321937601978847e-05, "loss": 0.1621719479560852, "step": 15710 }, { "epoch": 0.41408876435201497, "grad_norm": 2.196563959121704, "learning_rate": 1.3211297679244093e-05, "loss": 0.17033989429473878, "step": 15720 }, { "epoch": 0.4143521795965137, "grad_norm": 2.4124600887298584, "learning_rate": 1.3203217001378716e-05, "loss": 0.17522871494293213, "step": 15730 }, { "epoch": 0.41461559484101246, "grad_norm": 3.067631721496582, "learning_rate": 1.3195133992073808e-05, "loss": 0.16057913303375243, "step": 15740 }, { "epoch": 0.41487901008551115, "grad_norm": 3.4214606285095215, "learning_rate": 1.318704865721253e-05, "loss": 0.1619301915168762, "step": 15750 }, { "epoch": 0.4151424253300099, "grad_norm": 2.3831305503845215, "learning_rate": 1.3178961002679738e-05, "loss": 0.17017183303833008, "step": 15760 }, { "epoch": 0.41540584057450863, "grad_norm": 2.878424644470215, "learning_rate": 1.317087103436198e-05, "loss": 0.17917828559875487, "step": 15770 }, { "epoch": 0.4156692558190074, "grad_norm": 1.3427948951721191, "learning_rate": 1.3162778758147477e-05, "loss": 0.1616436719894409, "step": 15780 }, { "epoch": 0.4159326710635061, "grad_norm": 1.5541446208953857, "learning_rate": 1.315468417992614e-05, "loss": 0.17421810626983641, "step": 15790 }, { "epoch": 0.41619608630800486, "grad_norm": 1.3653860092163086, "learning_rate": 1.3146587305589552e-05, "loss": 0.16640043258666992, "step": 15800 }, { "epoch": 0.4164595015525036, "grad_norm": 1.8544694185256958, "learning_rate": 1.3138488141030967e-05, "loss": 0.15464634895324708, "step": 15810 }, { "epoch": 0.41672291679700235, "grad_norm": 3.9855027198791504, "learning_rate": 1.313038669214531e-05, "loss": 0.16999213695526122, "step": 15820 }, { "epoch": 0.4169863320415011, "grad_norm": 1.372261881828308, "learning_rate": 1.312228296482916e-05, "loss": 0.161803138256073, "step": 15830 }, { "epoch": 0.4172497472859998, "grad_norm": 1.307998538017273, "learning_rate": 1.311417696498076e-05, "loss": 0.16529455184936523, "step": 15840 }, { "epoch": 0.4175131625304985, "grad_norm": 1.633057951927185, "learning_rate": 1.310606869850001e-05, "loss": 0.148525869846344, "step": 15850 }, { "epoch": 0.41777657777499727, "grad_norm": 1.3422342538833618, "learning_rate": 1.3097958171288453e-05, "loss": 0.16022958755493164, "step": 15860 }, { "epoch": 0.418039993019496, "grad_norm": 4.8603291511535645, "learning_rate": 1.3089845389249282e-05, "loss": 0.15656884908676147, "step": 15870 }, { "epoch": 0.41830340826399476, "grad_norm": 1.8034406900405884, "learning_rate": 1.308173035828733e-05, "loss": 0.16262803077697754, "step": 15880 }, { "epoch": 0.4185668235084935, "grad_norm": 1.68562650680542, "learning_rate": 1.3073613084309065e-05, "loss": 0.16424031257629396, "step": 15890 }, { "epoch": 0.41883023875299225, "grad_norm": 3.4653966426849365, "learning_rate": 1.3065493573222594e-05, "loss": 0.17510900497436524, "step": 15900 }, { "epoch": 0.419093653997491, "grad_norm": 1.3093316555023193, "learning_rate": 1.305737183093764e-05, "loss": 0.15756523609161377, "step": 15910 }, { "epoch": 0.41935706924198973, "grad_norm": 2.150019407272339, "learning_rate": 1.3049247863365562e-05, "loss": 0.16162872314453125, "step": 15920 }, { "epoch": 0.4196204844864884, "grad_norm": 2.015519142150879, "learning_rate": 1.3041121676419333e-05, "loss": 0.16224609613418578, "step": 15930 }, { "epoch": 0.41988389973098716, "grad_norm": 1.4861056804656982, "learning_rate": 1.3032993276013546e-05, "loss": 0.15915272235870362, "step": 15940 }, { "epoch": 0.4201473149754859, "grad_norm": 0.978771984577179, "learning_rate": 1.3024862668064393e-05, "loss": 0.15683655738830565, "step": 15950 }, { "epoch": 0.42041073021998465, "grad_norm": 1.4222267866134644, "learning_rate": 1.3016729858489689e-05, "loss": 0.17029733657836915, "step": 15960 }, { "epoch": 0.4206741454644834, "grad_norm": 1.658936619758606, "learning_rate": 1.300859485320884e-05, "loss": 0.16732747554779054, "step": 15970 }, { "epoch": 0.42093756070898214, "grad_norm": 3.5594820976257324, "learning_rate": 1.3000457658142858e-05, "loss": 0.16622709035873412, "step": 15980 }, { "epoch": 0.4212009759534809, "grad_norm": 2.96176815032959, "learning_rate": 1.299231827921434e-05, "loss": 0.1698773145675659, "step": 15990 }, { "epoch": 0.4214643911979796, "grad_norm": 1.750180721282959, "learning_rate": 1.2984176722347478e-05, "loss": 0.1647020697593689, "step": 16000 }, { "epoch": 0.4217278064424783, "grad_norm": 1.3388428688049316, "learning_rate": 1.2976032993468049e-05, "loss": 0.163219678401947, "step": 16010 }, { "epoch": 0.42199122168697706, "grad_norm": 3.9195828437805176, "learning_rate": 1.296788709850341e-05, "loss": 0.1542607545852661, "step": 16020 }, { "epoch": 0.4222546369314758, "grad_norm": 2.6036977767944336, "learning_rate": 1.29597390433825e-05, "loss": 0.14489393234252929, "step": 16030 }, { "epoch": 0.42251805217597455, "grad_norm": 1.9037563800811768, "learning_rate": 1.2951588834035813e-05, "loss": 0.1587005615234375, "step": 16040 }, { "epoch": 0.4227814674204733, "grad_norm": 1.323176622390747, "learning_rate": 1.2943436476395435e-05, "loss": 0.1667156219482422, "step": 16050 }, { "epoch": 0.42304488266497203, "grad_norm": 1.1117252111434937, "learning_rate": 1.2935281976394998e-05, "loss": 0.1782158374786377, "step": 16060 }, { "epoch": 0.4233082979094708, "grad_norm": 3.0106799602508545, "learning_rate": 1.2927125339969701e-05, "loss": 0.16703643798828124, "step": 16070 }, { "epoch": 0.4235717131539695, "grad_norm": 3.236419916152954, "learning_rate": 1.2918966573056291e-05, "loss": 0.15371503829956054, "step": 16080 }, { "epoch": 0.42383512839846826, "grad_norm": 1.9663243293762207, "learning_rate": 1.2910805681593078e-05, "loss": 0.17175720930099486, "step": 16090 }, { "epoch": 0.42409854364296695, "grad_norm": 1.1184742450714111, "learning_rate": 1.2902642671519903e-05, "loss": 0.16069447994232178, "step": 16100 }, { "epoch": 0.4243619588874657, "grad_norm": 0.7529984712600708, "learning_rate": 1.2894477548778164e-05, "loss": 0.16816496849060059, "step": 16110 }, { "epoch": 0.42462537413196444, "grad_norm": 1.0020078420639038, "learning_rate": 1.2886310319310787e-05, "loss": 0.15931947231292726, "step": 16120 }, { "epoch": 0.4248887893764632, "grad_norm": 2.08776593208313, "learning_rate": 1.2878140989062233e-05, "loss": 0.172555410861969, "step": 16130 }, { "epoch": 0.4251522046209619, "grad_norm": 2.3193109035491943, "learning_rate": 1.2869969563978493e-05, "loss": 0.17338956594467164, "step": 16140 }, { "epoch": 0.42541561986546067, "grad_norm": 2.499490976333618, "learning_rate": 1.2861796050007082e-05, "loss": 0.16009494066238403, "step": 16150 }, { "epoch": 0.4256790351099594, "grad_norm": 2.0348308086395264, "learning_rate": 1.2853620453097039e-05, "loss": 0.1818298578262329, "step": 16160 }, { "epoch": 0.42594245035445816, "grad_norm": 2.1569488048553467, "learning_rate": 1.2845442779198916e-05, "loss": 0.15935273170471193, "step": 16170 }, { "epoch": 0.4262058655989569, "grad_norm": 1.6808607578277588, "learning_rate": 1.2837263034264773e-05, "loss": 0.1629301905632019, "step": 16180 }, { "epoch": 0.4264692808434556, "grad_norm": 0.920377254486084, "learning_rate": 1.2829081224248183e-05, "loss": 0.17481424808502197, "step": 16190 }, { "epoch": 0.42673269608795433, "grad_norm": 2.333655834197998, "learning_rate": 1.2820897355104224e-05, "loss": 0.17733159065246581, "step": 16200 }, { "epoch": 0.4269961113324531, "grad_norm": 1.866317629814148, "learning_rate": 1.2812711432789464e-05, "loss": 0.16241743564605712, "step": 16210 }, { "epoch": 0.4272595265769518, "grad_norm": 1.317667841911316, "learning_rate": 1.2804523463261975e-05, "loss": 0.14455286264419556, "step": 16220 }, { "epoch": 0.42752294182145056, "grad_norm": 2.096668004989624, "learning_rate": 1.2796333452481309e-05, "loss": 0.1658354878425598, "step": 16230 }, { "epoch": 0.4277863570659493, "grad_norm": 3.289599657058716, "learning_rate": 1.2788141406408516e-05, "loss": 0.16937354803085328, "step": 16240 }, { "epoch": 0.42804977231044805, "grad_norm": 1.2676527500152588, "learning_rate": 1.2779947331006113e-05, "loss": 0.15978046655654907, "step": 16250 }, { "epoch": 0.4283131875549468, "grad_norm": 1.6446702480316162, "learning_rate": 1.2771751232238107e-05, "loss": 0.15553294420242308, "step": 16260 }, { "epoch": 0.42857660279944554, "grad_norm": 5.549127578735352, "learning_rate": 1.2763553116069969e-05, "loss": 0.16435370445251465, "step": 16270 }, { "epoch": 0.4288400180439442, "grad_norm": 1.0072972774505615, "learning_rate": 1.2755352988468646e-05, "loss": 0.14891555309295654, "step": 16280 }, { "epoch": 0.42910343328844297, "grad_norm": 2.6050212383270264, "learning_rate": 1.2747150855402541e-05, "loss": 0.17341674566268922, "step": 16290 }, { "epoch": 0.4293668485329417, "grad_norm": 1.1411511898040771, "learning_rate": 1.2738946722841519e-05, "loss": 0.15573568344116212, "step": 16300 }, { "epoch": 0.42963026377744046, "grad_norm": 3.943164110183716, "learning_rate": 1.2730740596756908e-05, "loss": 0.1488109350204468, "step": 16310 }, { "epoch": 0.4298936790219392, "grad_norm": 4.784982204437256, "learning_rate": 1.2722532483121475e-05, "loss": 0.17180193662643434, "step": 16320 }, { "epoch": 0.43015709426643794, "grad_norm": 1.4563041925430298, "learning_rate": 1.271432238790944e-05, "loss": 0.1579676389694214, "step": 16330 }, { "epoch": 0.4304205095109367, "grad_norm": 2.7879443168640137, "learning_rate": 1.2706110317096463e-05, "loss": 0.1558327317237854, "step": 16340 }, { "epoch": 0.43068392475543543, "grad_norm": 3.55940842628479, "learning_rate": 1.269789627665965e-05, "loss": 0.1673831820487976, "step": 16350 }, { "epoch": 0.4309473399999341, "grad_norm": 2.3111114501953125, "learning_rate": 1.2689680272577527e-05, "loss": 0.17837551832199097, "step": 16360 }, { "epoch": 0.43121075524443286, "grad_norm": 2.2037792205810547, "learning_rate": 1.2681462310830061e-05, "loss": 0.16136665344238282, "step": 16370 }, { "epoch": 0.4314741704889316, "grad_norm": 2.9981019496917725, "learning_rate": 1.2673242397398635e-05, "loss": 0.16922414302825928, "step": 16380 }, { "epoch": 0.43173758573343035, "grad_norm": 1.0197727680206299, "learning_rate": 1.266502053826606e-05, "loss": 0.1677658200263977, "step": 16390 }, { "epoch": 0.4320010009779291, "grad_norm": 3.4116363525390625, "learning_rate": 1.2656796739416557e-05, "loss": 0.1675863027572632, "step": 16400 }, { "epoch": 0.43226441622242784, "grad_norm": 1.4036113023757935, "learning_rate": 1.2648571006835764e-05, "loss": 0.1535239577293396, "step": 16410 }, { "epoch": 0.4325278314669266, "grad_norm": 1.3738130331039429, "learning_rate": 1.2640343346510724e-05, "loss": 0.1632307291030884, "step": 16420 }, { "epoch": 0.4327912467114253, "grad_norm": 0.9281796216964722, "learning_rate": 1.2632113764429884e-05, "loss": 0.17912954092025757, "step": 16430 }, { "epoch": 0.43305466195592407, "grad_norm": 1.252395510673523, "learning_rate": 1.2623882266583083e-05, "loss": 0.16457724571228027, "step": 16440 }, { "epoch": 0.43331807720042276, "grad_norm": 2.024332046508789, "learning_rate": 1.2615648858961564e-05, "loss": 0.15653350353240966, "step": 16450 }, { "epoch": 0.4335814924449215, "grad_norm": 2.6753294467926025, "learning_rate": 1.260741354755796e-05, "loss": 0.15935895442962647, "step": 16460 }, { "epoch": 0.43384490768942024, "grad_norm": 1.0169576406478882, "learning_rate": 1.2599176338366282e-05, "loss": 0.1735864520072937, "step": 16470 }, { "epoch": 0.434108322933919, "grad_norm": 3.326626777648926, "learning_rate": 1.2590937237381927e-05, "loss": 0.14588464498519899, "step": 16480 }, { "epoch": 0.43437173817841773, "grad_norm": 1.669320821762085, "learning_rate": 1.2582696250601667e-05, "loss": 0.16426663398742675, "step": 16490 }, { "epoch": 0.4346351534229165, "grad_norm": 3.0486559867858887, "learning_rate": 1.257445338402365e-05, "loss": 0.16554529666900636, "step": 16500 }, { "epoch": 0.4348985686674152, "grad_norm": 1.6733574867248535, "learning_rate": 1.256620864364739e-05, "loss": 0.16886008977890016, "step": 16510 }, { "epoch": 0.43516198391191396, "grad_norm": 2.0443811416625977, "learning_rate": 1.255796203547376e-05, "loss": 0.15588550567626952, "step": 16520 }, { "epoch": 0.4354253991564127, "grad_norm": 2.158287525177002, "learning_rate": 1.2549713565505001e-05, "loss": 0.15578796863555908, "step": 16530 }, { "epoch": 0.4356888144009114, "grad_norm": 1.3083877563476562, "learning_rate": 1.254146323974471e-05, "loss": 0.15930577516555786, "step": 16540 }, { "epoch": 0.43595222964541014, "grad_norm": 2.3547494411468506, "learning_rate": 1.253321106419782e-05, "loss": 0.15193334817886353, "step": 16550 }, { "epoch": 0.4362156448899089, "grad_norm": 1.043045997619629, "learning_rate": 1.2524957044870627e-05, "loss": 0.15750924348831177, "step": 16560 }, { "epoch": 0.4364790601344076, "grad_norm": 1.03643000125885, "learning_rate": 1.2516701187770761e-05, "loss": 0.1600201368331909, "step": 16570 }, { "epoch": 0.43674247537890637, "grad_norm": 1.912192702293396, "learning_rate": 1.250844349890719e-05, "loss": 0.15904653072357178, "step": 16580 }, { "epoch": 0.4370058906234051, "grad_norm": 1.6667695045471191, "learning_rate": 1.250018398429022e-05, "loss": 0.17199625968933105, "step": 16590 }, { "epoch": 0.43726930586790386, "grad_norm": 4.5193257331848145, "learning_rate": 1.2491922649931475e-05, "loss": 0.17228596210479735, "step": 16600 }, { "epoch": 0.4375327211124026, "grad_norm": 3.454594850540161, "learning_rate": 1.2483659501843916e-05, "loss": 0.1706434965133667, "step": 16610 }, { "epoch": 0.4377961363569013, "grad_norm": 1.2611337900161743, "learning_rate": 1.2475394546041822e-05, "loss": 0.1528936505317688, "step": 16620 }, { "epoch": 0.43805955160140003, "grad_norm": 2.4401400089263916, "learning_rate": 1.2467127788540774e-05, "loss": 0.1492776870727539, "step": 16630 }, { "epoch": 0.4383229668458988, "grad_norm": 1.9390034675598145, "learning_rate": 1.245885923535768e-05, "loss": 0.16300721168518068, "step": 16640 }, { "epoch": 0.4385863820903975, "grad_norm": 0.87894207239151, "learning_rate": 1.2450588892510756e-05, "loss": 0.16711750030517578, "step": 16650 }, { "epoch": 0.43884979733489626, "grad_norm": 1.1897140741348267, "learning_rate": 1.2442316766019502e-05, "loss": 0.1760650873184204, "step": 16660 }, { "epoch": 0.439113212579395, "grad_norm": 2.9461121559143066, "learning_rate": 1.2434042861904743e-05, "loss": 0.15859558582305908, "step": 16670 }, { "epoch": 0.43937662782389375, "grad_norm": 1.5665311813354492, "learning_rate": 1.2425767186188569e-05, "loss": 0.1619487762451172, "step": 16680 }, { "epoch": 0.4396400430683925, "grad_norm": 2.356192111968994, "learning_rate": 1.2417489744894381e-05, "loss": 0.15897903442382813, "step": 16690 }, { "epoch": 0.43990345831289124, "grad_norm": 2.0059990882873535, "learning_rate": 1.2409210544046859e-05, "loss": 0.14641600847244263, "step": 16700 }, { "epoch": 0.4401668735573899, "grad_norm": 2.2083380222320557, "learning_rate": 1.2400929589671962e-05, "loss": 0.16532454490661622, "step": 16710 }, { "epoch": 0.44043028880188867, "grad_norm": 3.439023017883301, "learning_rate": 1.2392646887796924e-05, "loss": 0.1765817880630493, "step": 16720 }, { "epoch": 0.4406937040463874, "grad_norm": 1.6195241212844849, "learning_rate": 1.2384362444450253e-05, "loss": 0.16858491897583008, "step": 16730 }, { "epoch": 0.44095711929088616, "grad_norm": 1.3542710542678833, "learning_rate": 1.2376076265661722e-05, "loss": 0.17207159996032714, "step": 16740 }, { "epoch": 0.4412205345353849, "grad_norm": 2.5262858867645264, "learning_rate": 1.2367788357462375e-05, "loss": 0.1616477608680725, "step": 16750 }, { "epoch": 0.44148394977988364, "grad_norm": 3.1763851642608643, "learning_rate": 1.2359498725884509e-05, "loss": 0.15781328678131104, "step": 16760 }, { "epoch": 0.4417473650243824, "grad_norm": 1.5171833038330078, "learning_rate": 1.2351207376961666e-05, "loss": 0.17039743661880494, "step": 16770 }, { "epoch": 0.44201078026888113, "grad_norm": 2.899630069732666, "learning_rate": 1.234291431672866e-05, "loss": 0.17259519100189208, "step": 16780 }, { "epoch": 0.4422741955133799, "grad_norm": 1.6766637563705444, "learning_rate": 1.2334619551221531e-05, "loss": 0.15695855617523194, "step": 16790 }, { "epoch": 0.44253761075787856, "grad_norm": 3.455327033996582, "learning_rate": 1.232632308647757e-05, "loss": 0.17006523609161378, "step": 16800 }, { "epoch": 0.4428010260023773, "grad_norm": 3.539679527282715, "learning_rate": 1.2318024928535296e-05, "loss": 0.16052764654159546, "step": 16810 }, { "epoch": 0.44306444124687605, "grad_norm": 2.18220591545105, "learning_rate": 1.2309725083434477e-05, "loss": 0.15761148929595947, "step": 16820 }, { "epoch": 0.4433278564913748, "grad_norm": 0.9301260113716125, "learning_rate": 1.2301423557216088e-05, "loss": 0.16994829177856446, "step": 16830 }, { "epoch": 0.44359127173587354, "grad_norm": 2.795553684234619, "learning_rate": 1.2293120355922346e-05, "loss": 0.15537469387054442, "step": 16840 }, { "epoch": 0.4438546869803723, "grad_norm": 1.085246205329895, "learning_rate": 1.2284815485596671e-05, "loss": 0.1465196967124939, "step": 16850 }, { "epoch": 0.444118102224871, "grad_norm": 1.7054948806762695, "learning_rate": 1.2276508952283712e-05, "loss": 0.16167153120040895, "step": 16860 }, { "epoch": 0.44438151746936977, "grad_norm": 4.113103866577148, "learning_rate": 1.2268200762029319e-05, "loss": 0.16112957000732422, "step": 16870 }, { "epoch": 0.44464493271386846, "grad_norm": 1.7430205345153809, "learning_rate": 1.2259890920880555e-05, "loss": 0.1638568162918091, "step": 16880 }, { "epoch": 0.4449083479583672, "grad_norm": 3.03969144821167, "learning_rate": 1.2251579434885679e-05, "loss": 0.16101272106170655, "step": 16890 }, { "epoch": 0.44517176320286594, "grad_norm": 1.5307598114013672, "learning_rate": 1.2243266310094146e-05, "loss": 0.15664498805999755, "step": 16900 }, { "epoch": 0.4454351784473647, "grad_norm": 2.053541660308838, "learning_rate": 1.2234951552556613e-05, "loss": 0.16278046369552612, "step": 16910 }, { "epoch": 0.44569859369186343, "grad_norm": 1.4705405235290527, "learning_rate": 1.2226635168324916e-05, "loss": 0.15733711719512938, "step": 16920 }, { "epoch": 0.4459620089363622, "grad_norm": 2.6857690811157227, "learning_rate": 1.2218317163452078e-05, "loss": 0.1525759220123291, "step": 16930 }, { "epoch": 0.4462254241808609, "grad_norm": 3.8095650672912598, "learning_rate": 1.2209997543992301e-05, "loss": 0.15789613723754883, "step": 16940 }, { "epoch": 0.44648883942535966, "grad_norm": 4.578228950500488, "learning_rate": 1.2201676316000964e-05, "loss": 0.1727142572402954, "step": 16950 }, { "epoch": 0.4467522546698584, "grad_norm": 1.2320659160614014, "learning_rate": 1.2193353485534614e-05, "loss": 0.16519172191619874, "step": 16960 }, { "epoch": 0.4470156699143571, "grad_norm": 0.9870443344116211, "learning_rate": 1.2185029058650972e-05, "loss": 0.15995147228240966, "step": 16970 }, { "epoch": 0.44727908515885584, "grad_norm": 2.4086170196533203, "learning_rate": 1.2176703041408908e-05, "loss": 0.15714911222457886, "step": 16980 }, { "epoch": 0.4475425004033546, "grad_norm": 3.9423513412475586, "learning_rate": 1.2168375439868464e-05, "loss": 0.16401267051696777, "step": 16990 }, { "epoch": 0.4478059156478533, "grad_norm": 0.8298709392547607, "learning_rate": 1.2160046260090816e-05, "loss": 0.14493305683135987, "step": 17000 }, { "epoch": 0.44806933089235207, "grad_norm": 2.5729033946990967, "learning_rate": 1.2151715508138313e-05, "loss": 0.15746308565139772, "step": 17010 }, { "epoch": 0.4483327461368508, "grad_norm": 3.338959217071533, "learning_rate": 1.2143383190074429e-05, "loss": 0.15852725505828857, "step": 17020 }, { "epoch": 0.44859616138134956, "grad_norm": 3.6629714965820312, "learning_rate": 1.2135049311963787e-05, "loss": 0.16790661811828614, "step": 17030 }, { "epoch": 0.4488595766258483, "grad_norm": 1.1411627531051636, "learning_rate": 1.2126713879872138e-05, "loss": 0.16349164247512818, "step": 17040 }, { "epoch": 0.44912299187034704, "grad_norm": 1.3860563039779663, "learning_rate": 1.2118376899866375e-05, "loss": 0.16812214851379395, "step": 17050 }, { "epoch": 0.44938640711484573, "grad_norm": 5.227680206298828, "learning_rate": 1.2110038378014512e-05, "loss": 0.16549073457717894, "step": 17060 }, { "epoch": 0.4496498223593445, "grad_norm": 1.35225510597229, "learning_rate": 1.210169832038568e-05, "loss": 0.1568026065826416, "step": 17070 }, { "epoch": 0.4499132376038432, "grad_norm": 1.7641832828521729, "learning_rate": 1.2093356733050137e-05, "loss": 0.14945378303527831, "step": 17080 }, { "epoch": 0.45017665284834196, "grad_norm": 3.2126147747039795, "learning_rate": 1.2085013622079248e-05, "loss": 0.16679577827453612, "step": 17090 }, { "epoch": 0.4504400680928407, "grad_norm": 2.140160083770752, "learning_rate": 1.2076668993545491e-05, "loss": 0.15582554340362548, "step": 17100 }, { "epoch": 0.45070348333733945, "grad_norm": 4.454182147979736, "learning_rate": 1.206832285352244e-05, "loss": 0.17105987071990966, "step": 17110 }, { "epoch": 0.4509668985818382, "grad_norm": 2.019986391067505, "learning_rate": 1.2059975208084787e-05, "loss": 0.16447772979736328, "step": 17120 }, { "epoch": 0.45123031382633694, "grad_norm": 4.518513202667236, "learning_rate": 1.2051626063308299e-05, "loss": 0.1584460496902466, "step": 17130 }, { "epoch": 0.4514937290708356, "grad_norm": 1.098647952079773, "learning_rate": 1.2043275425269854e-05, "loss": 0.1649040699005127, "step": 17140 }, { "epoch": 0.45175714431533437, "grad_norm": 1.6341429948806763, "learning_rate": 1.2034923300047391e-05, "loss": 0.16294312477111816, "step": 17150 }, { "epoch": 0.4520205595598331, "grad_norm": 2.990582227706909, "learning_rate": 1.2026569693719963e-05, "loss": 0.15227413177490234, "step": 17160 }, { "epoch": 0.45228397480433186, "grad_norm": 1.3255850076675415, "learning_rate": 1.2018214612367673e-05, "loss": 0.15714179277420043, "step": 17170 }, { "epoch": 0.4525473900488306, "grad_norm": 1.1995974779129028, "learning_rate": 1.200985806207172e-05, "loss": 0.1490180015563965, "step": 17180 }, { "epoch": 0.45281080529332934, "grad_norm": 3.7351272106170654, "learning_rate": 1.2001500048914354e-05, "loss": 0.1470033884048462, "step": 17190 }, { "epoch": 0.4530742205378281, "grad_norm": 1.5902684926986694, "learning_rate": 1.1993140578978898e-05, "loss": 0.16541614532470703, "step": 17200 }, { "epoch": 0.45333763578232683, "grad_norm": 1.2647154331207275, "learning_rate": 1.1984779658349743e-05, "loss": 0.16150095462799072, "step": 17210 }, { "epoch": 0.4536010510268256, "grad_norm": 2.563264846801758, "learning_rate": 1.1976417293112318e-05, "loss": 0.1537063479423523, "step": 17220 }, { "epoch": 0.45386446627132426, "grad_norm": 1.830583095550537, "learning_rate": 1.1968053489353122e-05, "loss": 0.15241432189941406, "step": 17230 }, { "epoch": 0.454127881515823, "grad_norm": 2.7018556594848633, "learning_rate": 1.195968825315968e-05, "loss": 0.16067178249359132, "step": 17240 }, { "epoch": 0.45439129676032175, "grad_norm": 2.667846441268921, "learning_rate": 1.1951321590620583e-05, "loss": 0.16531755924224853, "step": 17250 }, { "epoch": 0.4546547120048205, "grad_norm": 3.670248031616211, "learning_rate": 1.1942953507825446e-05, "loss": 0.1741105318069458, "step": 17260 }, { "epoch": 0.45491812724931924, "grad_norm": 2.049520969390869, "learning_rate": 1.1934584010864921e-05, "loss": 0.16132049560546874, "step": 17270 }, { "epoch": 0.455181542493818, "grad_norm": 1.2342816591262817, "learning_rate": 1.1926213105830683e-05, "loss": 0.16140425205230713, "step": 17280 }, { "epoch": 0.4554449577383167, "grad_norm": 2.380225896835327, "learning_rate": 1.1917840798815448e-05, "loss": 0.17134900093078614, "step": 17290 }, { "epoch": 0.45570837298281547, "grad_norm": 1.386798620223999, "learning_rate": 1.1909467095912932e-05, "loss": 0.17553403377532958, "step": 17300 }, { "epoch": 0.4559717882273142, "grad_norm": 2.63531231880188, "learning_rate": 1.1901092003217885e-05, "loss": 0.16458430290222167, "step": 17310 }, { "epoch": 0.4562352034718129, "grad_norm": 2.6070666313171387, "learning_rate": 1.1892715526826058e-05, "loss": 0.17310547828674316, "step": 17320 }, { "epoch": 0.45649861871631164, "grad_norm": 0.8819878697395325, "learning_rate": 1.1884337672834212e-05, "loss": 0.15642634630203248, "step": 17330 }, { "epoch": 0.4567620339608104, "grad_norm": 1.807442545890808, "learning_rate": 1.1875958447340108e-05, "loss": 0.14970002174377442, "step": 17340 }, { "epoch": 0.45702544920530913, "grad_norm": 4.426146030426025, "learning_rate": 1.1867577856442514e-05, "loss": 0.16150664091110228, "step": 17350 }, { "epoch": 0.4572888644498079, "grad_norm": 2.905287981033325, "learning_rate": 1.1859195906241182e-05, "loss": 0.16452075242996217, "step": 17360 }, { "epoch": 0.4575522796943066, "grad_norm": 6.034582614898682, "learning_rate": 1.1850812602836851e-05, "loss": 0.16993050575256347, "step": 17370 }, { "epoch": 0.45781569493880536, "grad_norm": 1.7400367259979248, "learning_rate": 1.1842427952331263e-05, "loss": 0.15781996250152588, "step": 17380 }, { "epoch": 0.4580791101833041, "grad_norm": 2.970752716064453, "learning_rate": 1.183404196082712e-05, "loss": 0.17087574005126954, "step": 17390 }, { "epoch": 0.4583425254278028, "grad_norm": 2.2272698879241943, "learning_rate": 1.1825654634428111e-05, "loss": 0.15886332988739013, "step": 17400 }, { "epoch": 0.45860594067230154, "grad_norm": 2.83119797706604, "learning_rate": 1.1817265979238892e-05, "loss": 0.16951702833175658, "step": 17410 }, { "epoch": 0.4588693559168003, "grad_norm": 0.9239487648010254, "learning_rate": 1.1808876001365093e-05, "loss": 0.16689863204956054, "step": 17420 }, { "epoch": 0.459132771161299, "grad_norm": 3.8171260356903076, "learning_rate": 1.1800484706913294e-05, "loss": 0.15868542194366456, "step": 17430 }, { "epoch": 0.45939618640579777, "grad_norm": 2.8218719959259033, "learning_rate": 1.1792092101991052e-05, "loss": 0.15398695468902587, "step": 17440 }, { "epoch": 0.4596596016502965, "grad_norm": 5.654313564300537, "learning_rate": 1.1783698192706856e-05, "loss": 0.16085968017578126, "step": 17450 }, { "epoch": 0.45992301689479526, "grad_norm": 1.6495472192764282, "learning_rate": 1.1775302985170161e-05, "loss": 0.157107412815094, "step": 17460 }, { "epoch": 0.460186432139294, "grad_norm": 2.001307964324951, "learning_rate": 1.1766906485491361e-05, "loss": 0.17138925790786744, "step": 17470 }, { "epoch": 0.46044984738379274, "grad_norm": 0.9514077305793762, "learning_rate": 1.175850869978179e-05, "loss": 0.16766948699951173, "step": 17480 }, { "epoch": 0.46071326262829143, "grad_norm": 1.4404332637786865, "learning_rate": 1.1750109634153719e-05, "loss": 0.14930291175842286, "step": 17490 }, { "epoch": 0.4609766778727902, "grad_norm": 2.910261869430542, "learning_rate": 1.1741709294720346e-05, "loss": 0.14801719188690185, "step": 17500 }, { "epoch": 0.4612400931172889, "grad_norm": 3.0564184188842773, "learning_rate": 1.173330768759581e-05, "loss": 0.15724716186523438, "step": 17510 }, { "epoch": 0.46150350836178766, "grad_norm": 1.8757438659667969, "learning_rate": 1.1724904818895158e-05, "loss": 0.1507468581199646, "step": 17520 }, { "epoch": 0.4617669236062864, "grad_norm": 1.2458558082580566, "learning_rate": 1.1716500694734359e-05, "loss": 0.1728585958480835, "step": 17530 }, { "epoch": 0.46203033885078515, "grad_norm": 2.4242918491363525, "learning_rate": 1.1708095321230295e-05, "loss": 0.15739630460739135, "step": 17540 }, { "epoch": 0.4622937540952839, "grad_norm": 2.029374361038208, "learning_rate": 1.1699688704500765e-05, "loss": 0.15042076110839844, "step": 17550 }, { "epoch": 0.46255716933978264, "grad_norm": 3.0315053462982178, "learning_rate": 1.1691280850664463e-05, "loss": 0.16180498600006105, "step": 17560 }, { "epoch": 0.4628205845842814, "grad_norm": 2.337989330291748, "learning_rate": 1.1682871765840996e-05, "loss": 0.16689202785491944, "step": 17570 }, { "epoch": 0.46308399982878007, "grad_norm": 3.0139472484588623, "learning_rate": 1.1674461456150845e-05, "loss": 0.16289983987808226, "step": 17580 }, { "epoch": 0.4633474150732788, "grad_norm": 4.060619354248047, "learning_rate": 1.1666049927715408e-05, "loss": 0.16790425777435303, "step": 17590 }, { "epoch": 0.46361083031777756, "grad_norm": 1.242868185043335, "learning_rate": 1.1657637186656952e-05, "loss": 0.14586982727050782, "step": 17600 }, { "epoch": 0.4638742455622763, "grad_norm": 1.3561797142028809, "learning_rate": 1.1649223239098639e-05, "loss": 0.16947829723358154, "step": 17610 }, { "epoch": 0.46413766080677504, "grad_norm": 3.3659634590148926, "learning_rate": 1.1640808091164495e-05, "loss": 0.16189126968383788, "step": 17620 }, { "epoch": 0.4644010760512738, "grad_norm": 1.9483041763305664, "learning_rate": 1.1632391748979435e-05, "loss": 0.15316735506057738, "step": 17630 }, { "epoch": 0.46466449129577253, "grad_norm": 1.845205307006836, "learning_rate": 1.162397421866923e-05, "loss": 0.15993640422821045, "step": 17640 }, { "epoch": 0.4649279065402713, "grad_norm": 3.0508649349212646, "learning_rate": 1.1615555506360526e-05, "loss": 0.1618328332901001, "step": 17650 }, { "epoch": 0.46519132178476996, "grad_norm": 1.8535231351852417, "learning_rate": 1.1607135618180823e-05, "loss": 0.16833899021148682, "step": 17660 }, { "epoch": 0.4654547370292687, "grad_norm": 1.7698609828948975, "learning_rate": 1.1598714560258477e-05, "loss": 0.16336469650268554, "step": 17670 }, { "epoch": 0.46571815227376745, "grad_norm": 2.560708522796631, "learning_rate": 1.1590292338722702e-05, "loss": 0.17221155166625976, "step": 17680 }, { "epoch": 0.4659815675182662, "grad_norm": 2.0904951095581055, "learning_rate": 1.1581868959703553e-05, "loss": 0.15183128118515016, "step": 17690 }, { "epoch": 0.46624498276276494, "grad_norm": 4.153950214385986, "learning_rate": 1.1573444429331926e-05, "loss": 0.15240474939346313, "step": 17700 }, { "epoch": 0.4665083980072637, "grad_norm": 1.8710368871688843, "learning_rate": 1.1565018753739559e-05, "loss": 0.16465506553649903, "step": 17710 }, { "epoch": 0.4667718132517624, "grad_norm": 1.2529762983322144, "learning_rate": 1.1556591939059024e-05, "loss": 0.15789659023284913, "step": 17720 }, { "epoch": 0.46703522849626117, "grad_norm": 4.088364124298096, "learning_rate": 1.1548163991423716e-05, "loss": 0.15517597198486327, "step": 17730 }, { "epoch": 0.4672986437407599, "grad_norm": 1.5545498132705688, "learning_rate": 1.1539734916967864e-05, "loss": 0.16203732490539552, "step": 17740 }, { "epoch": 0.4675620589852586, "grad_norm": 1.5346015691757202, "learning_rate": 1.1531304721826506e-05, "loss": 0.14837806224822997, "step": 17750 }, { "epoch": 0.46782547422975734, "grad_norm": 3.4369773864746094, "learning_rate": 1.1522873412135509e-05, "loss": 0.1578960657119751, "step": 17760 }, { "epoch": 0.4680888894742561, "grad_norm": 1.5254985094070435, "learning_rate": 1.1514440994031536e-05, "loss": 0.15184051990509034, "step": 17770 }, { "epoch": 0.46835230471875483, "grad_norm": 1.2992305755615234, "learning_rate": 1.1506007473652075e-05, "loss": 0.16279627084732057, "step": 17780 }, { "epoch": 0.4686157199632536, "grad_norm": 4.610172271728516, "learning_rate": 1.1497572857135397e-05, "loss": 0.1577073335647583, "step": 17790 }, { "epoch": 0.4688791352077523, "grad_norm": 3.430170774459839, "learning_rate": 1.1489137150620582e-05, "loss": 0.16766208410263062, "step": 17800 }, { "epoch": 0.46914255045225106, "grad_norm": 0.8536844849586487, "learning_rate": 1.1480700360247504e-05, "loss": 0.16670070886611937, "step": 17810 }, { "epoch": 0.4694059656967498, "grad_norm": 2.361471652984619, "learning_rate": 1.147226249215682e-05, "loss": 0.1675114631652832, "step": 17820 }, { "epoch": 0.46966938094124855, "grad_norm": 1.5547020435333252, "learning_rate": 1.1463823552489977e-05, "loss": 0.16182464361190796, "step": 17830 }, { "epoch": 0.46993279618574724, "grad_norm": 2.1922311782836914, "learning_rate": 1.1455383547389193e-05, "loss": 0.15407836437225342, "step": 17840 }, { "epoch": 0.470196211430246, "grad_norm": 2.771745443344116, "learning_rate": 1.1446942482997479e-05, "loss": 0.1489264965057373, "step": 17850 }, { "epoch": 0.4704596266747447, "grad_norm": 2.260547637939453, "learning_rate": 1.1438500365458593e-05, "loss": 0.15409795045852662, "step": 17860 }, { "epoch": 0.47072304191924347, "grad_norm": 1.9662903547286987, "learning_rate": 1.1430057200917084e-05, "loss": 0.15779590606689453, "step": 17870 }, { "epoch": 0.4709864571637422, "grad_norm": 1.7129346132278442, "learning_rate": 1.1421612995518243e-05, "loss": 0.15012021064758302, "step": 17880 }, { "epoch": 0.47124987240824096, "grad_norm": 2.1971826553344727, "learning_rate": 1.1413167755408132e-05, "loss": 0.16841399669647217, "step": 17890 }, { "epoch": 0.4715132876527397, "grad_norm": 1.9748622179031372, "learning_rate": 1.1404721486733558e-05, "loss": 0.16436694860458373, "step": 17900 }, { "epoch": 0.47177670289723844, "grad_norm": 1.901901364326477, "learning_rate": 1.1396274195642085e-05, "loss": 0.15046207904815673, "step": 17910 }, { "epoch": 0.47204011814173713, "grad_norm": 1.9627782106399536, "learning_rate": 1.1387825888282006e-05, "loss": 0.16290569305419922, "step": 17920 }, { "epoch": 0.4723035333862359, "grad_norm": 1.452849268913269, "learning_rate": 1.1379376570802373e-05, "loss": 0.14839630126953124, "step": 17930 }, { "epoch": 0.4725669486307346, "grad_norm": 1.117431640625, "learning_rate": 1.137092624935296e-05, "loss": 0.1530773401260376, "step": 17940 }, { "epoch": 0.47283036387523336, "grad_norm": 4.2336249351501465, "learning_rate": 1.1362474930084278e-05, "loss": 0.16545209884643555, "step": 17950 }, { "epoch": 0.4730937791197321, "grad_norm": 1.3908365964889526, "learning_rate": 1.1354022619147557e-05, "loss": 0.15424270629882814, "step": 17960 }, { "epoch": 0.47335719436423085, "grad_norm": 1.4467229843139648, "learning_rate": 1.1345569322694756e-05, "loss": 0.16371290683746337, "step": 17970 }, { "epoch": 0.4736206096087296, "grad_norm": 3.4270780086517334, "learning_rate": 1.133711504687855e-05, "loss": 0.16420586109161378, "step": 17980 }, { "epoch": 0.47388402485322834, "grad_norm": 1.597594141960144, "learning_rate": 1.1328659797852325e-05, "loss": 0.1536606192588806, "step": 17990 }, { "epoch": 0.4741474400977271, "grad_norm": 2.134446620941162, "learning_rate": 1.1320203581770177e-05, "loss": 0.16456315517425538, "step": 18000 }, { "epoch": 0.47441085534222577, "grad_norm": 4.6324896812438965, "learning_rate": 1.1311746404786903e-05, "loss": 0.1707651734352112, "step": 18010 }, { "epoch": 0.4746742705867245, "grad_norm": 2.852555513381958, "learning_rate": 1.1303288273058002e-05, "loss": 0.15520042181015015, "step": 18020 }, { "epoch": 0.47493768583122326, "grad_norm": 1.6306298971176147, "learning_rate": 1.1294829192739665e-05, "loss": 0.15359745025634766, "step": 18030 }, { "epoch": 0.475201101075722, "grad_norm": 1.5242273807525635, "learning_rate": 1.1286369169988782e-05, "loss": 0.16946446895599365, "step": 18040 }, { "epoch": 0.47546451632022074, "grad_norm": 2.050584316253662, "learning_rate": 1.1277908210962916e-05, "loss": 0.16047724485397338, "step": 18050 }, { "epoch": 0.4757279315647195, "grad_norm": 2.363957166671753, "learning_rate": 1.126944632182032e-05, "loss": 0.15926183462142945, "step": 18060 }, { "epoch": 0.47599134680921823, "grad_norm": 1.6202011108398438, "learning_rate": 1.1260983508719922e-05, "loss": 0.14951848983764648, "step": 18070 }, { "epoch": 0.476254762053717, "grad_norm": 2.896294116973877, "learning_rate": 1.1252519777821324e-05, "loss": 0.16268365383148192, "step": 18080 }, { "epoch": 0.4765181772982157, "grad_norm": 3.675147771835327, "learning_rate": 1.1244055135284795e-05, "loss": 0.15135730504989625, "step": 18090 }, { "epoch": 0.4767815925427144, "grad_norm": 2.5802671909332275, "learning_rate": 1.123558958727126e-05, "loss": 0.16273877620697022, "step": 18100 }, { "epoch": 0.47704500778721315, "grad_norm": 2.2331926822662354, "learning_rate": 1.1227123139942322e-05, "loss": 0.14999455213546753, "step": 18110 }, { "epoch": 0.4773084230317119, "grad_norm": 1.2662789821624756, "learning_rate": 1.1218655799460216e-05, "loss": 0.16341941356658934, "step": 18120 }, { "epoch": 0.47757183827621064, "grad_norm": 2.253513813018799, "learning_rate": 1.1210187571987845e-05, "loss": 0.1620806097984314, "step": 18130 }, { "epoch": 0.4778352535207094, "grad_norm": 5.047486782073975, "learning_rate": 1.1201718463688745e-05, "loss": 0.1705404043197632, "step": 18140 }, { "epoch": 0.4780986687652081, "grad_norm": 1.7908602952957153, "learning_rate": 1.1193248480727101e-05, "loss": 0.16067860126495362, "step": 18150 }, { "epoch": 0.47836208400970687, "grad_norm": 2.2775461673736572, "learning_rate": 1.118477762926773e-05, "loss": 0.15348243713378906, "step": 18160 }, { "epoch": 0.4786254992542056, "grad_norm": 2.2345807552337646, "learning_rate": 1.117630591547609e-05, "loss": 0.1704486846923828, "step": 18170 }, { "epoch": 0.4788889144987043, "grad_norm": 1.4964834451675415, "learning_rate": 1.1167833345518248e-05, "loss": 0.14685357809066774, "step": 18180 }, { "epoch": 0.47915232974320304, "grad_norm": 2.919999122619629, "learning_rate": 1.1159359925560918e-05, "loss": 0.1626855731010437, "step": 18190 }, { "epoch": 0.4794157449877018, "grad_norm": 3.616553783416748, "learning_rate": 1.115088566177141e-05, "loss": 0.16559758186340331, "step": 18200 }, { "epoch": 0.47967916023220053, "grad_norm": 1.9140464067459106, "learning_rate": 1.1142410560317669e-05, "loss": 0.16047315597534179, "step": 18210 }, { "epoch": 0.4799425754766993, "grad_norm": 1.4006867408752441, "learning_rate": 1.113393462736823e-05, "loss": 0.17965371608734132, "step": 18220 }, { "epoch": 0.480205990721198, "grad_norm": 1.1743228435516357, "learning_rate": 1.1125457869092251e-05, "loss": 0.17116411924362182, "step": 18230 }, { "epoch": 0.48046940596569676, "grad_norm": 1.3340212106704712, "learning_rate": 1.1116980291659474e-05, "loss": 0.16163933277130127, "step": 18240 }, { "epoch": 0.4807328212101955, "grad_norm": 2.278843879699707, "learning_rate": 1.1108501901240254e-05, "loss": 0.15230828523635864, "step": 18250 }, { "epoch": 0.48099623645469425, "grad_norm": 2.1647024154663086, "learning_rate": 1.1100022704005524e-05, "loss": 0.15776861906051637, "step": 18260 }, { "epoch": 0.48125965169919294, "grad_norm": 3.2165520191192627, "learning_rate": 1.109154270612681e-05, "loss": 0.15442652702331544, "step": 18270 }, { "epoch": 0.4815230669436917, "grad_norm": 1.504679799079895, "learning_rate": 1.1083061913776224e-05, "loss": 0.14990427494049072, "step": 18280 }, { "epoch": 0.4817864821881904, "grad_norm": 2.2533888816833496, "learning_rate": 1.1074580333126448e-05, "loss": 0.15904197692871094, "step": 18290 }, { "epoch": 0.48204989743268917, "grad_norm": 1.34426748752594, "learning_rate": 1.1066097970350747e-05, "loss": 0.16354832649230958, "step": 18300 }, { "epoch": 0.4823133126771879, "grad_norm": 1.1781983375549316, "learning_rate": 1.1057614831622942e-05, "loss": 0.1478770136833191, "step": 18310 }, { "epoch": 0.48257672792168665, "grad_norm": 0.890922486782074, "learning_rate": 1.1049130923117437e-05, "loss": 0.1567893147468567, "step": 18320 }, { "epoch": 0.4828401431661854, "grad_norm": 1.201900601387024, "learning_rate": 1.1040646251009181e-05, "loss": 0.17284270524978637, "step": 18330 }, { "epoch": 0.48310355841068414, "grad_norm": 2.0787353515625, "learning_rate": 1.1032160821473692e-05, "loss": 0.1614445447921753, "step": 18340 }, { "epoch": 0.4833669736551829, "grad_norm": 3.2369842529296875, "learning_rate": 1.1023674640687019e-05, "loss": 0.16488101482391357, "step": 18350 }, { "epoch": 0.4836303888996816, "grad_norm": 1.195396065711975, "learning_rate": 1.1015187714825782e-05, "loss": 0.15748071670532227, "step": 18360 }, { "epoch": 0.4838938041441803, "grad_norm": 1.2416008710861206, "learning_rate": 1.1006700050067122e-05, "loss": 0.1431671380996704, "step": 18370 }, { "epoch": 0.48415721938867906, "grad_norm": 2.768425703048706, "learning_rate": 1.0998211652588737e-05, "loss": 0.16150543689727784, "step": 18380 }, { "epoch": 0.4844206346331778, "grad_norm": 2.3195102214813232, "learning_rate": 1.0989722528568845e-05, "loss": 0.16678128242492676, "step": 18390 }, { "epoch": 0.48468404987767655, "grad_norm": 2.0543510913848877, "learning_rate": 1.0981232684186191e-05, "loss": 0.1670436143875122, "step": 18400 }, { "epoch": 0.4849474651221753, "grad_norm": 2.6411712169647217, "learning_rate": 1.0972742125620062e-05, "loss": 0.15532324314117432, "step": 18410 }, { "epoch": 0.48521088036667404, "grad_norm": 2.1973202228546143, "learning_rate": 1.0964250859050245e-05, "loss": 0.1565201163291931, "step": 18420 }, { "epoch": 0.4854742956111728, "grad_norm": 1.4138691425323486, "learning_rate": 1.0955758890657051e-05, "loss": 0.1519381284713745, "step": 18430 }, { "epoch": 0.48573771085567147, "grad_norm": 3.5480620861053467, "learning_rate": 1.0947266226621298e-05, "loss": 0.13841559886932372, "step": 18440 }, { "epoch": 0.4860011261001702, "grad_norm": 1.080889105796814, "learning_rate": 1.0938772873124321e-05, "loss": 0.16399086713790895, "step": 18450 }, { "epoch": 0.48626454134466895, "grad_norm": 1.410297155380249, "learning_rate": 1.0930278836347943e-05, "loss": 0.15316035747528076, "step": 18460 }, { "epoch": 0.4865279565891677, "grad_norm": 1.2855443954467773, "learning_rate": 1.0921784122474495e-05, "loss": 0.1494288682937622, "step": 18470 }, { "epoch": 0.48679137183366644, "grad_norm": 2.380707025527954, "learning_rate": 1.0913288737686789e-05, "loss": 0.16172568798065184, "step": 18480 }, { "epoch": 0.4870547870781652, "grad_norm": 2.0207769870758057, "learning_rate": 1.090479268816814e-05, "loss": 0.162811541557312, "step": 18490 }, { "epoch": 0.48731820232266393, "grad_norm": 2.164940118789673, "learning_rate": 1.0896295980102334e-05, "loss": 0.1715444564819336, "step": 18500 }, { "epoch": 0.4875816175671627, "grad_norm": 1.3578565120697021, "learning_rate": 1.0887798619673648e-05, "loss": 0.14958250522613525, "step": 18510 }, { "epoch": 0.4878450328116614, "grad_norm": 1.2484729290008545, "learning_rate": 1.087930061306682e-05, "loss": 0.16092066764831542, "step": 18520 }, { "epoch": 0.4881084480561601, "grad_norm": 1.596464991569519, "learning_rate": 1.0870801966467073e-05, "loss": 0.15613336563110353, "step": 18530 }, { "epoch": 0.48837186330065885, "grad_norm": 1.5419564247131348, "learning_rate": 1.0862302686060082e-05, "loss": 0.14779927730560302, "step": 18540 }, { "epoch": 0.4886352785451576, "grad_norm": 3.6230366230010986, "learning_rate": 1.0853802778031995e-05, "loss": 0.15992087125778198, "step": 18550 }, { "epoch": 0.48889869378965634, "grad_norm": 3.897662878036499, "learning_rate": 1.0845302248569411e-05, "loss": 0.16268694400787354, "step": 18560 }, { "epoch": 0.4891621090341551, "grad_norm": 2.2916314601898193, "learning_rate": 1.0836801103859377e-05, "loss": 0.15720276832580565, "step": 18570 }, { "epoch": 0.4894255242786538, "grad_norm": 1.6976141929626465, "learning_rate": 1.0828299350089401e-05, "loss": 0.1526874542236328, "step": 18580 }, { "epoch": 0.48968893952315257, "grad_norm": 1.8197667598724365, "learning_rate": 1.0819796993447422e-05, "loss": 0.16386427879333496, "step": 18590 }, { "epoch": 0.4899523547676513, "grad_norm": 1.5571391582489014, "learning_rate": 1.081129404012182e-05, "loss": 0.15391082763671876, "step": 18600 }, { "epoch": 0.49021577001215005, "grad_norm": 1.7813259363174438, "learning_rate": 1.0802790496301414e-05, "loss": 0.16532355546951294, "step": 18610 }, { "epoch": 0.49047918525664874, "grad_norm": 2.7002644538879395, "learning_rate": 1.079428636817545e-05, "loss": 0.16333986520767213, "step": 18620 }, { "epoch": 0.4907426005011475, "grad_norm": 0.9184061884880066, "learning_rate": 1.0785781661933596e-05, "loss": 0.151596736907959, "step": 18630 }, { "epoch": 0.49100601574564623, "grad_norm": 1.1625558137893677, "learning_rate": 1.0777276383765953e-05, "loss": 0.15542771816253662, "step": 18640 }, { "epoch": 0.491269430990145, "grad_norm": 2.222142457962036, "learning_rate": 1.0768770539863017e-05, "loss": 0.16766867637634278, "step": 18650 }, { "epoch": 0.4915328462346437, "grad_norm": 1.179381012916565, "learning_rate": 1.076026413641572e-05, "loss": 0.1572582244873047, "step": 18660 }, { "epoch": 0.49179626147914246, "grad_norm": 2.43353009223938, "learning_rate": 1.0751757179615378e-05, "loss": 0.17592885494232177, "step": 18670 }, { "epoch": 0.4920596767236412, "grad_norm": 3.8554303646087646, "learning_rate": 1.0743249675653733e-05, "loss": 0.158453106880188, "step": 18680 }, { "epoch": 0.49232309196813995, "grad_norm": 1.462989091873169, "learning_rate": 1.0734741630722904e-05, "loss": 0.16088238954544068, "step": 18690 }, { "epoch": 0.49258650721263864, "grad_norm": 1.4464510679244995, "learning_rate": 1.0726233051015415e-05, "loss": 0.15942426919937133, "step": 18700 }, { "epoch": 0.4928499224571374, "grad_norm": 2.195713996887207, "learning_rate": 1.0717723942724182e-05, "loss": 0.14852602481842042, "step": 18710 }, { "epoch": 0.4931133377016361, "grad_norm": 2.7798850536346436, "learning_rate": 1.07092143120425e-05, "loss": 0.1490856647491455, "step": 18720 }, { "epoch": 0.49337675294613487, "grad_norm": 4.439784526824951, "learning_rate": 1.0700704165164042e-05, "loss": 0.15099480152130126, "step": 18730 }, { "epoch": 0.4936401681906336, "grad_norm": 2.4314680099487305, "learning_rate": 1.069219350828286e-05, "loss": 0.16324299573898315, "step": 18740 }, { "epoch": 0.49390358343513235, "grad_norm": 0.9601092338562012, "learning_rate": 1.068368234759338e-05, "loss": 0.14920501708984374, "step": 18750 }, { "epoch": 0.4941669986796311, "grad_norm": 1.7846492528915405, "learning_rate": 1.0675170689290393e-05, "loss": 0.16314772367477418, "step": 18760 }, { "epoch": 0.49443041392412984, "grad_norm": 2.987828016281128, "learning_rate": 1.0666658539569048e-05, "loss": 0.15782172679901124, "step": 18770 }, { "epoch": 0.4946938291686286, "grad_norm": 3.7724950313568115, "learning_rate": 1.0658145904624852e-05, "loss": 0.16672720909118652, "step": 18780 }, { "epoch": 0.4949572444131273, "grad_norm": 1.957360029220581, "learning_rate": 1.0649632790653677e-05, "loss": 0.16248611211776734, "step": 18790 }, { "epoch": 0.495220659657626, "grad_norm": 1.2883286476135254, "learning_rate": 1.0641119203851724e-05, "loss": 0.16260408163070678, "step": 18800 }, { "epoch": 0.49548407490212476, "grad_norm": 0.8817960619926453, "learning_rate": 1.0632605150415561e-05, "loss": 0.1743881344795227, "step": 18810 }, { "epoch": 0.4957474901466235, "grad_norm": 2.713728427886963, "learning_rate": 1.062409063654207e-05, "loss": 0.17767078876495362, "step": 18820 }, { "epoch": 0.49601090539112225, "grad_norm": 5.9982218742370605, "learning_rate": 1.0615575668428494e-05, "loss": 0.1593568205833435, "step": 18830 }, { "epoch": 0.496274320635621, "grad_norm": 1.9735069274902344, "learning_rate": 1.0607060252272383e-05, "loss": 0.1538442850112915, "step": 18840 }, { "epoch": 0.49653773588011973, "grad_norm": 2.3023507595062256, "learning_rate": 1.0598544394271633e-05, "loss": 0.14283354282379152, "step": 18850 }, { "epoch": 0.4968011511246185, "grad_norm": 1.5171362161636353, "learning_rate": 1.0590028100624448e-05, "loss": 0.15250186920166015, "step": 18860 }, { "epoch": 0.4970645663691172, "grad_norm": 4.112742900848389, "learning_rate": 1.0581511377529354e-05, "loss": 0.16422786712646484, "step": 18870 }, { "epoch": 0.4973279816136159, "grad_norm": 1.6772422790527344, "learning_rate": 1.0572994231185194e-05, "loss": 0.17124562263488768, "step": 18880 }, { "epoch": 0.49759139685811465, "grad_norm": 0.8000403642654419, "learning_rate": 1.056447666779111e-05, "loss": 0.15877749919891357, "step": 18890 }, { "epoch": 0.4978548121026134, "grad_norm": 1.1480902433395386, "learning_rate": 1.0555958693546552e-05, "loss": 0.15187355279922485, "step": 18900 }, { "epoch": 0.49811822734711214, "grad_norm": 1.8673231601715088, "learning_rate": 1.0547440314651274e-05, "loss": 0.1381618022918701, "step": 18910 }, { "epoch": 0.4983816425916109, "grad_norm": 2.6932506561279297, "learning_rate": 1.0538921537305315e-05, "loss": 0.1706507086753845, "step": 18920 }, { "epoch": 0.49864505783610963, "grad_norm": 3.3750767707824707, "learning_rate": 1.0530402367709007e-05, "loss": 0.15527933835983276, "step": 18930 }, { "epoch": 0.4989084730806084, "grad_norm": 2.8509957790374756, "learning_rate": 1.0521882812062978e-05, "loss": 0.1531803011894226, "step": 18940 }, { "epoch": 0.4991718883251071, "grad_norm": 1.555138111114502, "learning_rate": 1.0513362876568118e-05, "loss": 0.15717835426330568, "step": 18950 }, { "epoch": 0.49943530356960586, "grad_norm": 3.9117257595062256, "learning_rate": 1.0504842567425611e-05, "loss": 0.16523246765136718, "step": 18960 }, { "epoch": 0.49969871881410455, "grad_norm": 1.695755958557129, "learning_rate": 1.0496321890836901e-05, "loss": 0.15761640071868896, "step": 18970 }, { "epoch": 0.4999621340586033, "grad_norm": 2.2149081230163574, "learning_rate": 1.0487800853003706e-05, "loss": 0.1655901312828064, "step": 18980 }, { "epoch": 0.500225549303102, "grad_norm": 3.292912483215332, "learning_rate": 1.0479279460128005e-05, "loss": 0.1573721170425415, "step": 18990 }, { "epoch": 0.5004889645476008, "grad_norm": 2.6144659519195557, "learning_rate": 1.0470757718412035e-05, "loss": 0.14543402194976807, "step": 19000 }, { "epoch": 0.5007523797920995, "grad_norm": 2.3796679973602295, "learning_rate": 1.0462235634058287e-05, "loss": 0.1509539246559143, "step": 19010 }, { "epoch": 0.5010157950365982, "grad_norm": 1.2347304821014404, "learning_rate": 1.0453713213269503e-05, "loss": 0.16018129587173463, "step": 19020 }, { "epoch": 0.501279210281097, "grad_norm": 1.7158881425857544, "learning_rate": 1.0445190462248667e-05, "loss": 0.1536634683609009, "step": 19030 }, { "epoch": 0.5015426255255957, "grad_norm": 2.937049627304077, "learning_rate": 1.0436667387199004e-05, "loss": 0.1573346495628357, "step": 19040 }, { "epoch": 0.5018060407700945, "grad_norm": 1.7396029233932495, "learning_rate": 1.0428143994323978e-05, "loss": 0.16195236444473265, "step": 19050 }, { "epoch": 0.5020694560145932, "grad_norm": 1.9451385736465454, "learning_rate": 1.041962028982728e-05, "loss": 0.15992172956466674, "step": 19060 }, { "epoch": 0.502332871259092, "grad_norm": 3.140862226486206, "learning_rate": 1.0411096279912835e-05, "loss": 0.17484234571456908, "step": 19070 }, { "epoch": 0.5025962865035907, "grad_norm": 1.5655015707015991, "learning_rate": 1.0402571970784778e-05, "loss": 0.16122565269470215, "step": 19080 }, { "epoch": 0.5028597017480895, "grad_norm": 1.571516752243042, "learning_rate": 1.0394047368647473e-05, "loss": 0.1525239586830139, "step": 19090 }, { "epoch": 0.5031231169925882, "grad_norm": 2.0518460273742676, "learning_rate": 1.0385522479705493e-05, "loss": 0.14480807781219482, "step": 19100 }, { "epoch": 0.5033865322370868, "grad_norm": 1.0620979070663452, "learning_rate": 1.0376997310163622e-05, "loss": 0.16836537122726442, "step": 19110 }, { "epoch": 0.5036499474815856, "grad_norm": 1.7377628087997437, "learning_rate": 1.0368471866226837e-05, "loss": 0.15135738849639893, "step": 19120 }, { "epoch": 0.5039133627260843, "grad_norm": 5.387238502502441, "learning_rate": 1.0359946154100336e-05, "loss": 0.15443273782730102, "step": 19130 }, { "epoch": 0.5041767779705831, "grad_norm": 1.894454836845398, "learning_rate": 1.035142017998949e-05, "loss": 0.16029045581817628, "step": 19140 }, { "epoch": 0.5044401932150818, "grad_norm": 1.7274363040924072, "learning_rate": 1.0342893950099879e-05, "loss": 0.16461924314498902, "step": 19150 }, { "epoch": 0.5047036084595806, "grad_norm": 1.7422114610671997, "learning_rate": 1.0334367470637257e-05, "loss": 0.1542970895767212, "step": 19160 }, { "epoch": 0.5049670237040793, "grad_norm": 1.4486957788467407, "learning_rate": 1.0325840747807562e-05, "loss": 0.15284123420715331, "step": 19170 }, { "epoch": 0.505230438948578, "grad_norm": 1.8484388589859009, "learning_rate": 1.0317313787816915e-05, "loss": 0.15952398777008056, "step": 19180 }, { "epoch": 0.5054938541930768, "grad_norm": 2.615324020385742, "learning_rate": 1.0308786596871606e-05, "loss": 0.16887452602386474, "step": 19190 }, { "epoch": 0.5057572694375755, "grad_norm": 1.9221680164337158, "learning_rate": 1.0300259181178086e-05, "loss": 0.15857605934143065, "step": 19200 }, { "epoch": 0.5060206846820743, "grad_norm": 1.1506390571594238, "learning_rate": 1.0291731546942977e-05, "loss": 0.14363713264465333, "step": 19210 }, { "epoch": 0.506284099926573, "grad_norm": 1.4499422311782837, "learning_rate": 1.0283203700373065e-05, "loss": 0.14765558242797852, "step": 19220 }, { "epoch": 0.5065475151710718, "grad_norm": 1.023091435432434, "learning_rate": 1.0274675647675281e-05, "loss": 0.16054351329803468, "step": 19230 }, { "epoch": 0.5068109304155705, "grad_norm": 3.022087574005127, "learning_rate": 1.0266147395056712e-05, "loss": 0.1555554151535034, "step": 19240 }, { "epoch": 0.5070743456600693, "grad_norm": 2.5118515491485596, "learning_rate": 1.0257618948724583e-05, "loss": 0.16713078022003175, "step": 19250 }, { "epoch": 0.507337760904568, "grad_norm": 1.5194969177246094, "learning_rate": 1.024909031488627e-05, "loss": 0.16857630014419556, "step": 19260 }, { "epoch": 0.5076011761490666, "grad_norm": 2.30271053314209, "learning_rate": 1.0240561499749278e-05, "loss": 0.146337890625, "step": 19270 }, { "epoch": 0.5078645913935654, "grad_norm": 1.2149070501327515, "learning_rate": 1.023203250952125e-05, "loss": 0.15593966245651245, "step": 19280 }, { "epoch": 0.5081280066380641, "grad_norm": 1.371152400970459, "learning_rate": 1.0223503350409949e-05, "loss": 0.15275187492370607, "step": 19290 }, { "epoch": 0.5083914218825629, "grad_norm": 1.6411782503128052, "learning_rate": 1.0214974028623268e-05, "loss": 0.1636442184448242, "step": 19300 }, { "epoch": 0.5086548371270616, "grad_norm": 1.8914084434509277, "learning_rate": 1.0206444550369212e-05, "loss": 0.16912438869476318, "step": 19310 }, { "epoch": 0.5089182523715604, "grad_norm": 1.7983207702636719, "learning_rate": 1.0197914921855909e-05, "loss": 0.15526256561279297, "step": 19320 }, { "epoch": 0.5091816676160591, "grad_norm": 1.7750922441482544, "learning_rate": 1.0189385149291586e-05, "loss": 0.16371619701385498, "step": 19330 }, { "epoch": 0.5094450828605579, "grad_norm": 1.1984350681304932, "learning_rate": 1.018085523888458e-05, "loss": 0.15677036046981813, "step": 19340 }, { "epoch": 0.5097084981050566, "grad_norm": 3.5126760005950928, "learning_rate": 1.0172325196843328e-05, "loss": 0.1574757695198059, "step": 19350 }, { "epoch": 0.5099719133495553, "grad_norm": 1.6525864601135254, "learning_rate": 1.0163795029376362e-05, "loss": 0.1501251220703125, "step": 19360 }, { "epoch": 0.5102353285940541, "grad_norm": 1.9471900463104248, "learning_rate": 1.015526474269231e-05, "loss": 0.15542320013046265, "step": 19370 }, { "epoch": 0.5104987438385528, "grad_norm": 3.057561159133911, "learning_rate": 1.0146734342999875e-05, "loss": 0.14965965747833251, "step": 19380 }, { "epoch": 0.5107621590830516, "grad_norm": 2.0350611209869385, "learning_rate": 1.0138203836507859e-05, "loss": 0.1680455446243286, "step": 19390 }, { "epoch": 0.5110255743275502, "grad_norm": 2.0766584873199463, "learning_rate": 1.0129673229425125e-05, "loss": 0.16256932020187378, "step": 19400 }, { "epoch": 0.511288989572049, "grad_norm": 3.1251132488250732, "learning_rate": 1.0121142527960624e-05, "loss": 0.1496471643447876, "step": 19410 }, { "epoch": 0.5115524048165477, "grad_norm": 2.5526156425476074, "learning_rate": 1.0112611738323362e-05, "loss": 0.1491311550140381, "step": 19420 }, { "epoch": 0.5118158200610465, "grad_norm": 2.351992130279541, "learning_rate": 1.0104080866722419e-05, "loss": 0.14468069076538087, "step": 19430 }, { "epoch": 0.5120792353055452, "grad_norm": 3.294828414916992, "learning_rate": 1.0095549919366928e-05, "loss": 0.1490342140197754, "step": 19440 }, { "epoch": 0.5123426505500439, "grad_norm": 2.137716770172119, "learning_rate": 1.0087018902466086e-05, "loss": 0.1661937117576599, "step": 19450 }, { "epoch": 0.5126060657945427, "grad_norm": 4.223395824432373, "learning_rate": 1.0078487822229132e-05, "loss": 0.1683075785636902, "step": 19460 }, { "epoch": 0.5128694810390414, "grad_norm": 1.5501258373260498, "learning_rate": 1.0069956684865354e-05, "loss": 0.1619966745376587, "step": 19470 }, { "epoch": 0.5131328962835402, "grad_norm": 2.01901912689209, "learning_rate": 1.0061425496584084e-05, "loss": 0.16237061023712157, "step": 19480 }, { "epoch": 0.5133963115280389, "grad_norm": 3.0577597618103027, "learning_rate": 1.0052894263594687e-05, "loss": 0.13875463008880615, "step": 19490 }, { "epoch": 0.5136597267725377, "grad_norm": 2.2666473388671875, "learning_rate": 1.0044362992106563e-05, "loss": 0.15395309925079345, "step": 19500 }, { "epoch": 0.5139231420170364, "grad_norm": 1.6495226621627808, "learning_rate": 1.0035831688329139e-05, "loss": 0.16008172035217286, "step": 19510 }, { "epoch": 0.5141865572615352, "grad_norm": 1.5471975803375244, "learning_rate": 1.0027300358471865e-05, "loss": 0.16285134553909303, "step": 19520 }, { "epoch": 0.5144499725060339, "grad_norm": 2.909989595413208, "learning_rate": 1.0018769008744214e-05, "loss": 0.16586616039276122, "step": 19530 }, { "epoch": 0.5147133877505325, "grad_norm": 2.210388660430908, "learning_rate": 1.0010237645355669e-05, "loss": 0.1581364870071411, "step": 19540 }, { "epoch": 0.5149768029950313, "grad_norm": 1.8451000452041626, "learning_rate": 1.0001706274515724e-05, "loss": 0.15912315845489503, "step": 19550 }, { "epoch": 0.51524021823953, "grad_norm": 3.953000545501709, "learning_rate": 9.993174902433877e-06, "loss": 0.14735848903656007, "step": 19560 }, { "epoch": 0.5155036334840288, "grad_norm": 1.2780835628509521, "learning_rate": 9.984643535319632e-06, "loss": 0.15493234395980834, "step": 19570 }, { "epoch": 0.5157670487285275, "grad_norm": 3.827397346496582, "learning_rate": 9.976112179382483e-06, "loss": 0.16872299909591676, "step": 19580 }, { "epoch": 0.5160304639730263, "grad_norm": 1.712674856185913, "learning_rate": 9.967580840831922e-06, "loss": 0.1730565071105957, "step": 19590 }, { "epoch": 0.516293879217525, "grad_norm": 2.1661882400512695, "learning_rate": 9.959049525877425e-06, "loss": 0.1628730058670044, "step": 19600 }, { "epoch": 0.5165572944620238, "grad_norm": 2.961266279220581, "learning_rate": 9.950518240728446e-06, "loss": 0.16858322620391847, "step": 19610 }, { "epoch": 0.5168207097065225, "grad_norm": 1.1655664443969727, "learning_rate": 9.94198699159443e-06, "loss": 0.15004472732543944, "step": 19620 }, { "epoch": 0.5170841249510212, "grad_norm": 1.1639149188995361, "learning_rate": 9.93345578468478e-06, "loss": 0.1602458953857422, "step": 19630 }, { "epoch": 0.51734754019552, "grad_norm": 1.6253156661987305, "learning_rate": 9.924924626208884e-06, "loss": 0.14787919521331788, "step": 19640 }, { "epoch": 0.5176109554400187, "grad_norm": 1.6254117488861084, "learning_rate": 9.916393522376087e-06, "loss": 0.15716748237609862, "step": 19650 }, { "epoch": 0.5178743706845175, "grad_norm": 3.3979828357696533, "learning_rate": 9.907862479395684e-06, "loss": 0.15606313943862915, "step": 19660 }, { "epoch": 0.5181377859290162, "grad_norm": 1.5340206623077393, "learning_rate": 9.899331503476951e-06, "loss": 0.15981748104095458, "step": 19670 }, { "epoch": 0.518401201173515, "grad_norm": 1.4853127002716064, "learning_rate": 9.890800600829086e-06, "loss": 0.1472477912902832, "step": 19680 }, { "epoch": 0.5186646164180136, "grad_norm": 3.6219208240509033, "learning_rate": 9.882269777661261e-06, "loss": 0.14220106601715088, "step": 19690 }, { "epoch": 0.5189280316625123, "grad_norm": 1.795997977256775, "learning_rate": 9.873739040182569e-06, "loss": 0.14819468259811402, "step": 19700 }, { "epoch": 0.5191914469070111, "grad_norm": 1.69784677028656, "learning_rate": 9.86520839460205e-06, "loss": 0.17189364433288573, "step": 19710 }, { "epoch": 0.5194548621515098, "grad_norm": 3.260606050491333, "learning_rate": 9.856677847128678e-06, "loss": 0.15009126663208008, "step": 19720 }, { "epoch": 0.5197182773960086, "grad_norm": 1.1614912748336792, "learning_rate": 9.848147403971353e-06, "loss": 0.15468780994415282, "step": 19730 }, { "epoch": 0.5199816926405073, "grad_norm": 3.5047354698181152, "learning_rate": 9.839617071338897e-06, "loss": 0.16821537017822266, "step": 19740 }, { "epoch": 0.5202451078850061, "grad_norm": 2.144775390625, "learning_rate": 9.831086855440056e-06, "loss": 0.16761116981506347, "step": 19750 }, { "epoch": 0.5205085231295048, "grad_norm": 1.5799715518951416, "learning_rate": 9.82255676248349e-06, "loss": 0.1466035008430481, "step": 19760 }, { "epoch": 0.5207719383740036, "grad_norm": 3.6713335514068604, "learning_rate": 9.814026798677767e-06, "loss": 0.1456329345703125, "step": 19770 }, { "epoch": 0.5210353536185023, "grad_norm": 1.1480560302734375, "learning_rate": 9.805496970231367e-06, "loss": 0.14946129322052001, "step": 19780 }, { "epoch": 0.521298768863001, "grad_norm": 1.071624517440796, "learning_rate": 9.796967283352657e-06, "loss": 0.14545783996582032, "step": 19790 }, { "epoch": 0.5215621841074998, "grad_norm": 2.0692641735076904, "learning_rate": 9.788437744249925e-06, "loss": 0.15603485107421874, "step": 19800 }, { "epoch": 0.5218255993519985, "grad_norm": 3.037128210067749, "learning_rate": 9.779908359131327e-06, "loss": 0.15036942958831787, "step": 19810 }, { "epoch": 0.5220890145964973, "grad_norm": 1.475661039352417, "learning_rate": 9.77137913420492e-06, "loss": 0.14207813739776612, "step": 19820 }, { "epoch": 0.522352429840996, "grad_norm": 1.4564274549484253, "learning_rate": 9.762850075678644e-06, "loss": 0.1633122444152832, "step": 19830 }, { "epoch": 0.5226158450854947, "grad_norm": 3.119992971420288, "learning_rate": 9.754321189760312e-06, "loss": 0.15362017154693602, "step": 19840 }, { "epoch": 0.5228792603299934, "grad_norm": 0.9828754663467407, "learning_rate": 9.74579248265762e-06, "loss": 0.14542964696884156, "step": 19850 }, { "epoch": 0.5231426755744922, "grad_norm": 1.153295636177063, "learning_rate": 9.737263960578127e-06, "loss": 0.1517823576927185, "step": 19860 }, { "epoch": 0.5234060908189909, "grad_norm": 1.3068430423736572, "learning_rate": 9.728735629729256e-06, "loss": 0.1642720580101013, "step": 19870 }, { "epoch": 0.5236695060634896, "grad_norm": 2.313164472579956, "learning_rate": 9.7202074963183e-06, "loss": 0.15084776878356934, "step": 19880 }, { "epoch": 0.5239329213079884, "grad_norm": 1.5911214351654053, "learning_rate": 9.711679566552396e-06, "loss": 0.15739707946777343, "step": 19890 }, { "epoch": 0.5241963365524871, "grad_norm": 2.2101666927337646, "learning_rate": 9.703151846638545e-06, "loss": 0.15900707244873047, "step": 19900 }, { "epoch": 0.5244597517969859, "grad_norm": 1.86976158618927, "learning_rate": 9.694624342783591e-06, "loss": 0.15887658596038817, "step": 19910 }, { "epoch": 0.5247231670414846, "grad_norm": 1.666396141052246, "learning_rate": 9.686097061194209e-06, "loss": 0.17313835620880128, "step": 19920 }, { "epoch": 0.5249865822859834, "grad_norm": 2.3264641761779785, "learning_rate": 9.677570008076933e-06, "loss": 0.15282702445983887, "step": 19930 }, { "epoch": 0.5252499975304821, "grad_norm": 1.8412483930587769, "learning_rate": 9.669043189638115e-06, "loss": 0.15437095165252684, "step": 19940 }, { "epoch": 0.5255134127749809, "grad_norm": 1.697350025177002, "learning_rate": 9.66051661208394e-06, "loss": 0.1703418493270874, "step": 19950 }, { "epoch": 0.5257768280194796, "grad_norm": 1.5954238176345825, "learning_rate": 9.651990281620423e-06, "loss": 0.1438295364379883, "step": 19960 }, { "epoch": 0.5260402432639782, "grad_norm": 4.510148048400879, "learning_rate": 9.643464204453392e-06, "loss": 0.14812800884246827, "step": 19970 }, { "epoch": 0.526303658508477, "grad_norm": 3.3616676330566406, "learning_rate": 9.634938386788497e-06, "loss": 0.1731477737426758, "step": 19980 }, { "epoch": 0.5265670737529757, "grad_norm": 2.45900297164917, "learning_rate": 9.626412834831193e-06, "loss": 0.17140510082244872, "step": 19990 }, { "epoch": 0.5268304889974745, "grad_norm": 3.433534622192383, "learning_rate": 9.617887554786746e-06, "loss": 0.14622676372528076, "step": 20000 }, { "epoch": 0.5270939042419732, "grad_norm": 2.3644187450408936, "learning_rate": 9.609362552860225e-06, "loss": 0.14655935764312744, "step": 20010 }, { "epoch": 0.527357319486472, "grad_norm": 2.8691766262054443, "learning_rate": 9.600837835256486e-06, "loss": 0.16045408248901366, "step": 20020 }, { "epoch": 0.5276207347309707, "grad_norm": 2.584439277648926, "learning_rate": 9.592313408180199e-06, "loss": 0.1377565383911133, "step": 20030 }, { "epoch": 0.5278841499754695, "grad_norm": 5.7555928230285645, "learning_rate": 9.583789277835798e-06, "loss": 0.17043137550354004, "step": 20040 }, { "epoch": 0.5281475652199682, "grad_norm": 1.701331377029419, "learning_rate": 9.575265450427517e-06, "loss": 0.1621212363243103, "step": 20050 }, { "epoch": 0.5284109804644669, "grad_norm": 1.2800464630126953, "learning_rate": 9.566741932159369e-06, "loss": 0.16303017139434814, "step": 20060 }, { "epoch": 0.5286743957089657, "grad_norm": 1.2150346040725708, "learning_rate": 9.558218729235132e-06, "loss": 0.16407653093338012, "step": 20070 }, { "epoch": 0.5289378109534644, "grad_norm": 5.032706260681152, "learning_rate": 9.549695847858363e-06, "loss": 0.16129859685897827, "step": 20080 }, { "epoch": 0.5292012261979632, "grad_norm": 2.233376979827881, "learning_rate": 9.541173294232387e-06, "loss": 0.16194778680801392, "step": 20090 }, { "epoch": 0.5294646414424619, "grad_norm": 1.877744197845459, "learning_rate": 9.53265107456028e-06, "loss": 0.1578907251358032, "step": 20100 }, { "epoch": 0.5297280566869607, "grad_norm": 1.1419438123703003, "learning_rate": 9.524129195044886e-06, "loss": 0.14385558366775514, "step": 20110 }, { "epoch": 0.5299914719314593, "grad_norm": 3.026160717010498, "learning_rate": 9.515607661888799e-06, "loss": 0.16765084266662597, "step": 20120 }, { "epoch": 0.5302548871759581, "grad_norm": 3.0028793811798096, "learning_rate": 9.507086481294345e-06, "loss": 0.1785462975502014, "step": 20130 }, { "epoch": 0.5305183024204568, "grad_norm": 1.4511961936950684, "learning_rate": 9.498565659463624e-06, "loss": 0.15418097972869874, "step": 20140 }, { "epoch": 0.5307817176649555, "grad_norm": 1.2663971185684204, "learning_rate": 9.490045202598442e-06, "loss": 0.1501399278640747, "step": 20150 }, { "epoch": 0.5310451329094543, "grad_norm": 1.6932134628295898, "learning_rate": 9.48152511690037e-06, "loss": 0.16105144023895263, "step": 20160 }, { "epoch": 0.531308548153953, "grad_norm": 3.0733590126037598, "learning_rate": 9.473005408570685e-06, "loss": 0.15466649532318116, "step": 20170 }, { "epoch": 0.5315719633984518, "grad_norm": 1.8717421293258667, "learning_rate": 9.464486083810397e-06, "loss": 0.1482807517051697, "step": 20180 }, { "epoch": 0.5318353786429505, "grad_norm": 2.2918312549591064, "learning_rate": 9.455967148820243e-06, "loss": 0.15587683916091918, "step": 20190 }, { "epoch": 0.5320987938874493, "grad_norm": 3.8501720428466797, "learning_rate": 9.44744860980067e-06, "loss": 0.15944314002990723, "step": 20200 }, { "epoch": 0.532362209131948, "grad_norm": 0.8495195508003235, "learning_rate": 9.438930472951838e-06, "loss": 0.1381974458694458, "step": 20210 }, { "epoch": 0.5326256243764467, "grad_norm": 1.3953897953033447, "learning_rate": 9.430412744473615e-06, "loss": 0.15007736682891845, "step": 20220 }, { "epoch": 0.5328890396209455, "grad_norm": 3.0569510459899902, "learning_rate": 9.421895430565568e-06, "loss": 0.15149104595184326, "step": 20230 }, { "epoch": 0.5331524548654442, "grad_norm": 2.5338199138641357, "learning_rate": 9.413378537426972e-06, "loss": 0.1484626054763794, "step": 20240 }, { "epoch": 0.533415870109943, "grad_norm": 1.373223900794983, "learning_rate": 9.404862071256785e-06, "loss": 0.14908428192138673, "step": 20250 }, { "epoch": 0.5336792853544416, "grad_norm": 2.7286694049835205, "learning_rate": 9.396346038253654e-06, "loss": 0.1734738230705261, "step": 20260 }, { "epoch": 0.5339427005989404, "grad_norm": 4.595427989959717, "learning_rate": 9.387830444615926e-06, "loss": 0.14774326086044312, "step": 20270 }, { "epoch": 0.5342061158434391, "grad_norm": 5.876338481903076, "learning_rate": 9.379315296541611e-06, "loss": 0.14545888900756837, "step": 20280 }, { "epoch": 0.5344695310879379, "grad_norm": 1.717044711112976, "learning_rate": 9.370800600228399e-06, "loss": 0.16347600221633912, "step": 20290 }, { "epoch": 0.5347329463324366, "grad_norm": 0.8655709028244019, "learning_rate": 9.362286361873658e-06, "loss": 0.15550763607025148, "step": 20300 }, { "epoch": 0.5349963615769353, "grad_norm": 2.6758337020874023, "learning_rate": 9.353772587674413e-06, "loss": 0.158047878742218, "step": 20310 }, { "epoch": 0.5352597768214341, "grad_norm": 1.5839687585830688, "learning_rate": 9.345259283827362e-06, "loss": 0.15303782224655152, "step": 20320 }, { "epoch": 0.5355231920659328, "grad_norm": 2.9795854091644287, "learning_rate": 9.336746456528853e-06, "loss": 0.14995687007904052, "step": 20330 }, { "epoch": 0.5357866073104316, "grad_norm": 0.9015664458274841, "learning_rate": 9.328234111974885e-06, "loss": 0.14864944219589232, "step": 20340 }, { "epoch": 0.5360500225549303, "grad_norm": 1.785194754600525, "learning_rate": 9.319722256361117e-06, "loss": 0.16822893619537355, "step": 20350 }, { "epoch": 0.5363134377994291, "grad_norm": 5.40424108505249, "learning_rate": 9.311210895882835e-06, "loss": 0.1520548939704895, "step": 20360 }, { "epoch": 0.5365768530439278, "grad_norm": 2.0298173427581787, "learning_rate": 9.302700036734985e-06, "loss": 0.14977588653564453, "step": 20370 }, { "epoch": 0.5368402682884266, "grad_norm": 2.5361990928649902, "learning_rate": 9.294189685112133e-06, "loss": 0.162720251083374, "step": 20380 }, { "epoch": 0.5371036835329253, "grad_norm": 1.1583213806152344, "learning_rate": 9.285679847208474e-06, "loss": 0.1382606267929077, "step": 20390 }, { "epoch": 0.537367098777424, "grad_norm": 2.008822202682495, "learning_rate": 9.277170529217847e-06, "loss": 0.15886919498443602, "step": 20400 }, { "epoch": 0.5376305140219227, "grad_norm": 1.1866884231567383, "learning_rate": 9.268661737333691e-06, "loss": 0.15071730613708495, "step": 20410 }, { "epoch": 0.5378939292664214, "grad_norm": 2.2162058353424072, "learning_rate": 9.260153477749072e-06, "loss": 0.1532289743423462, "step": 20420 }, { "epoch": 0.5381573445109202, "grad_norm": 2.8463635444641113, "learning_rate": 9.251645756656676e-06, "loss": 0.15837483406066893, "step": 20430 }, { "epoch": 0.5384207597554189, "grad_norm": 1.661497712135315, "learning_rate": 9.243138580248782e-06, "loss": 0.1469140648841858, "step": 20440 }, { "epoch": 0.5386841749999177, "grad_norm": 4.910411834716797, "learning_rate": 9.234631954717285e-06, "loss": 0.1687198519706726, "step": 20450 }, { "epoch": 0.5389475902444164, "grad_norm": 2.9947774410247803, "learning_rate": 9.226125886253673e-06, "loss": 0.16655175685882567, "step": 20460 }, { "epoch": 0.5392110054889152, "grad_norm": 1.2091035842895508, "learning_rate": 9.217620381049026e-06, "loss": 0.15386669635772704, "step": 20470 }, { "epoch": 0.5394744207334139, "grad_norm": 1.6893515586853027, "learning_rate": 9.209115445294022e-06, "loss": 0.14903850555419923, "step": 20480 }, { "epoch": 0.5397378359779126, "grad_norm": 2.275068759918213, "learning_rate": 9.200611085178917e-06, "loss": 0.15284998416900636, "step": 20490 }, { "epoch": 0.5400012512224114, "grad_norm": 1.8460414409637451, "learning_rate": 9.192107306893556e-06, "loss": 0.15073950290679933, "step": 20500 }, { "epoch": 0.5402646664669101, "grad_norm": 4.1188201904296875, "learning_rate": 9.183604116627354e-06, "loss": 0.167917799949646, "step": 20510 }, { "epoch": 0.5405280817114089, "grad_norm": 1.560364007949829, "learning_rate": 9.175101520569293e-06, "loss": 0.16584885120391846, "step": 20520 }, { "epoch": 0.5407914969559076, "grad_norm": 1.9335297346115112, "learning_rate": 9.166599524907937e-06, "loss": 0.15956406593322753, "step": 20530 }, { "epoch": 0.5410549122004064, "grad_norm": 2.049795627593994, "learning_rate": 9.158098135831404e-06, "loss": 0.16120803356170654, "step": 20540 }, { "epoch": 0.541318327444905, "grad_norm": 1.9140596389770508, "learning_rate": 9.149597359527368e-06, "loss": 0.15464072227478026, "step": 20550 }, { "epoch": 0.5415817426894038, "grad_norm": 1.3707996606826782, "learning_rate": 9.141097202183063e-06, "loss": 0.15705151557922364, "step": 20560 }, { "epoch": 0.5418451579339025, "grad_norm": 1.8554515838623047, "learning_rate": 9.13259766998527e-06, "loss": 0.14657289981842042, "step": 20570 }, { "epoch": 0.5421085731784012, "grad_norm": 2.8093652725219727, "learning_rate": 9.124098769120318e-06, "loss": 0.15248311758041383, "step": 20580 }, { "epoch": 0.5423719884229, "grad_norm": 1.0202397108078003, "learning_rate": 9.115600505774072e-06, "loss": 0.1624840021133423, "step": 20590 }, { "epoch": 0.5426354036673987, "grad_norm": 1.4390980005264282, "learning_rate": 9.10710288613193e-06, "loss": 0.1547492265701294, "step": 20600 }, { "epoch": 0.5428988189118975, "grad_norm": 1.796067714691162, "learning_rate": 9.098605916378837e-06, "loss": 0.15064367055892944, "step": 20610 }, { "epoch": 0.5431622341563962, "grad_norm": 2.9771876335144043, "learning_rate": 9.090109602699239e-06, "loss": 0.15061794519424437, "step": 20620 }, { "epoch": 0.543425649400895, "grad_norm": 2.9544529914855957, "learning_rate": 9.081613951277135e-06, "loss": 0.1566182017326355, "step": 20630 }, { "epoch": 0.5436890646453937, "grad_norm": 1.814065933227539, "learning_rate": 9.07311896829602e-06, "loss": 0.15215754508972168, "step": 20640 }, { "epoch": 0.5439524798898925, "grad_norm": 1.4152992963790894, "learning_rate": 9.064624659938902e-06, "loss": 0.1370638370513916, "step": 20650 }, { "epoch": 0.5442158951343912, "grad_norm": 1.7820321321487427, "learning_rate": 9.056131032388318e-06, "loss": 0.1583823800086975, "step": 20660 }, { "epoch": 0.5444793103788899, "grad_norm": 3.417924642562866, "learning_rate": 9.047638091826291e-06, "loss": 0.1554866313934326, "step": 20670 }, { "epoch": 0.5447427256233887, "grad_norm": 1.0550373792648315, "learning_rate": 9.039145844434344e-06, "loss": 0.1617178201675415, "step": 20680 }, { "epoch": 0.5450061408678873, "grad_norm": 2.5362119674682617, "learning_rate": 9.030654296393512e-06, "loss": 0.1452026128768921, "step": 20690 }, { "epoch": 0.5452695561123861, "grad_norm": 1.5982290506362915, "learning_rate": 9.022163453884299e-06, "loss": 0.14522453546524047, "step": 20700 }, { "epoch": 0.5455329713568848, "grad_norm": 1.0945608615875244, "learning_rate": 9.013673323086716e-06, "loss": 0.15095056295394899, "step": 20710 }, { "epoch": 0.5457963866013836, "grad_norm": 2.723308563232422, "learning_rate": 9.005183910180243e-06, "loss": 0.15140933990478517, "step": 20720 }, { "epoch": 0.5460598018458823, "grad_norm": 1.352957844734192, "learning_rate": 8.996695221343834e-06, "loss": 0.1521397352218628, "step": 20730 }, { "epoch": 0.546323217090381, "grad_norm": 2.6924705505371094, "learning_rate": 8.988207262755936e-06, "loss": 0.14818162918090821, "step": 20740 }, { "epoch": 0.5465866323348798, "grad_norm": 1.5719616413116455, "learning_rate": 8.979720040594438e-06, "loss": 0.16209136247634887, "step": 20750 }, { "epoch": 0.5468500475793785, "grad_norm": 2.220827341079712, "learning_rate": 8.971233561036718e-06, "loss": 0.14809532165527345, "step": 20760 }, { "epoch": 0.5471134628238773, "grad_norm": 1.7069342136383057, "learning_rate": 8.962747830259598e-06, "loss": 0.14867268800735473, "step": 20770 }, { "epoch": 0.547376878068376, "grad_norm": 1.8203846216201782, "learning_rate": 8.954262854439354e-06, "loss": 0.16189303398132324, "step": 20780 }, { "epoch": 0.5476402933128748, "grad_norm": 1.4964431524276733, "learning_rate": 8.945778639751723e-06, "loss": 0.15910892486572265, "step": 20790 }, { "epoch": 0.5479037085573735, "grad_norm": 4.091558456420898, "learning_rate": 8.937295192371882e-06, "loss": 0.15478501319885254, "step": 20800 }, { "epoch": 0.5481671238018723, "grad_norm": 1.377855896949768, "learning_rate": 8.928812518474445e-06, "loss": 0.13054977655410765, "step": 20810 }, { "epoch": 0.548430539046371, "grad_norm": 1.4105064868927002, "learning_rate": 8.920330624233477e-06, "loss": 0.13957880735397338, "step": 20820 }, { "epoch": 0.5486939542908696, "grad_norm": 1.5862802267074585, "learning_rate": 8.911849515822457e-06, "loss": 0.15461745262145996, "step": 20830 }, { "epoch": 0.5489573695353684, "grad_norm": 1.1540553569793701, "learning_rate": 8.903369199414311e-06, "loss": 0.1571873664855957, "step": 20840 }, { "epoch": 0.5492207847798671, "grad_norm": 1.5341836214065552, "learning_rate": 8.894889681181374e-06, "loss": 0.15518150329589844, "step": 20850 }, { "epoch": 0.5494842000243659, "grad_norm": 2.940678358078003, "learning_rate": 8.886410967295401e-06, "loss": 0.16143124103546141, "step": 20860 }, { "epoch": 0.5497476152688646, "grad_norm": 2.5645227432250977, "learning_rate": 8.877933063927576e-06, "loss": 0.14090176820755004, "step": 20870 }, { "epoch": 0.5500110305133634, "grad_norm": 1.852903127670288, "learning_rate": 8.869455977248473e-06, "loss": 0.14861774444580078, "step": 20880 }, { "epoch": 0.5502744457578621, "grad_norm": 2.8419744968414307, "learning_rate": 8.860979713428085e-06, "loss": 0.13398683071136475, "step": 20890 }, { "epoch": 0.5505378610023609, "grad_norm": 2.281986951828003, "learning_rate": 8.852504278635806e-06, "loss": 0.14829174280166627, "step": 20900 }, { "epoch": 0.5508012762468596, "grad_norm": 3.0567266941070557, "learning_rate": 8.844029679040418e-06, "loss": 0.14746508598327637, "step": 20910 }, { "epoch": 0.5510646914913583, "grad_norm": 2.038241386413574, "learning_rate": 8.835555920810104e-06, "loss": 0.14576928615570067, "step": 20920 }, { "epoch": 0.5513281067358571, "grad_norm": 1.266901969909668, "learning_rate": 8.827083010112429e-06, "loss": 0.1409524440765381, "step": 20930 }, { "epoch": 0.5515915219803558, "grad_norm": 1.1499358415603638, "learning_rate": 8.818610953114342e-06, "loss": 0.13817958831787108, "step": 20940 }, { "epoch": 0.5518549372248546, "grad_norm": 2.351799488067627, "learning_rate": 8.810139755982173e-06, "loss": 0.14823726415634156, "step": 20950 }, { "epoch": 0.5521183524693533, "grad_norm": 2.6761157512664795, "learning_rate": 8.801669424881624e-06, "loss": 0.15958502292633056, "step": 20960 }, { "epoch": 0.5523817677138521, "grad_norm": 1.0058557987213135, "learning_rate": 8.79319996597777e-06, "loss": 0.1437225580215454, "step": 20970 }, { "epoch": 0.5526451829583507, "grad_norm": 3.710235357284546, "learning_rate": 8.784731385435049e-06, "loss": 0.15269854068756103, "step": 20980 }, { "epoch": 0.5529085982028495, "grad_norm": 1.5595353841781616, "learning_rate": 8.77626368941725e-06, "loss": 0.1407456398010254, "step": 20990 }, { "epoch": 0.5531720134473482, "grad_norm": 2.1029608249664307, "learning_rate": 8.767796884087539e-06, "loss": 0.14953968524932862, "step": 21000 }, { "epoch": 0.5534354286918469, "grad_norm": 2.7203619480133057, "learning_rate": 8.759330975608418e-06, "loss": 0.15344691276550293, "step": 21010 }, { "epoch": 0.5536988439363457, "grad_norm": 1.47426438331604, "learning_rate": 8.750865970141736e-06, "loss": 0.1533078670501709, "step": 21020 }, { "epoch": 0.5539622591808444, "grad_norm": 2.3285772800445557, "learning_rate": 8.742401873848696e-06, "loss": 0.15346482992172242, "step": 21030 }, { "epoch": 0.5542256744253432, "grad_norm": 1.5418825149536133, "learning_rate": 8.733938692889827e-06, "loss": 0.15955232381820678, "step": 21040 }, { "epoch": 0.5544890896698419, "grad_norm": 1.9579263925552368, "learning_rate": 8.725476433425003e-06, "loss": 0.1570706844329834, "step": 21050 }, { "epoch": 0.5547525049143407, "grad_norm": 3.180915594100952, "learning_rate": 8.717015101613417e-06, "loss": 0.15493900775909425, "step": 21060 }, { "epoch": 0.5550159201588394, "grad_norm": 1.1183475255966187, "learning_rate": 8.708554703613592e-06, "loss": 0.14308042526245118, "step": 21070 }, { "epoch": 0.5552793354033382, "grad_norm": 1.3879623413085938, "learning_rate": 8.700095245583374e-06, "loss": 0.13010308742523194, "step": 21080 }, { "epoch": 0.5555427506478369, "grad_norm": 1.297886610031128, "learning_rate": 8.691636733679915e-06, "loss": 0.1478807210922241, "step": 21090 }, { "epoch": 0.5558061658923356, "grad_norm": 1.4021432399749756, "learning_rate": 8.683179174059691e-06, "loss": 0.15958867073059083, "step": 21100 }, { "epoch": 0.5560695811368344, "grad_norm": 1.6735790967941284, "learning_rate": 8.67472257287848e-06, "loss": 0.1480097532272339, "step": 21110 }, { "epoch": 0.556332996381333, "grad_norm": 3.487928628921509, "learning_rate": 8.666266936291354e-06, "loss": 0.15235320329666138, "step": 21120 }, { "epoch": 0.5565964116258318, "grad_norm": 2.8474669456481934, "learning_rate": 8.657812270452695e-06, "loss": 0.16285637617111207, "step": 21130 }, { "epoch": 0.5568598268703305, "grad_norm": 2.7220983505249023, "learning_rate": 8.649358581516171e-06, "loss": 0.157392680644989, "step": 21140 }, { "epoch": 0.5571232421148293, "grad_norm": 1.2702735662460327, "learning_rate": 8.640905875634744e-06, "loss": 0.14674720764160157, "step": 21150 }, { "epoch": 0.557386657359328, "grad_norm": 2.303466796875, "learning_rate": 8.632454158960656e-06, "loss": 0.14456517696380616, "step": 21160 }, { "epoch": 0.5576500726038268, "grad_norm": 2.2393035888671875, "learning_rate": 8.624003437645431e-06, "loss": 0.1340503692626953, "step": 21170 }, { "epoch": 0.5579134878483255, "grad_norm": 4.425219535827637, "learning_rate": 8.615553717839873e-06, "loss": 0.1290259838104248, "step": 21180 }, { "epoch": 0.5581769030928242, "grad_norm": 2.04120135307312, "learning_rate": 8.607105005694048e-06, "loss": 0.16320137977600097, "step": 21190 }, { "epoch": 0.558440318337323, "grad_norm": 1.5081661939620972, "learning_rate": 8.598657307357293e-06, "loss": 0.15151288509368896, "step": 21200 }, { "epoch": 0.5587037335818217, "grad_norm": 1.3132966756820679, "learning_rate": 8.590210628978213e-06, "loss": 0.14079298973083496, "step": 21210 }, { "epoch": 0.5589671488263205, "grad_norm": 3.512824296951294, "learning_rate": 8.581764976704653e-06, "loss": 0.15577102899551393, "step": 21220 }, { "epoch": 0.5592305640708192, "grad_norm": 1.7644121646881104, "learning_rate": 8.573320356683739e-06, "loss": 0.1571846127510071, "step": 21230 }, { "epoch": 0.559493979315318, "grad_norm": 1.4654150009155273, "learning_rate": 8.564876775061819e-06, "loss": 0.1391467809677124, "step": 21240 }, { "epoch": 0.5597573945598167, "grad_norm": 2.932528495788574, "learning_rate": 8.556434237984496e-06, "loss": 0.1303308367729187, "step": 21250 }, { "epoch": 0.5600208098043153, "grad_norm": 3.408749580383301, "learning_rate": 8.547992751596617e-06, "loss": 0.15472233295440674, "step": 21260 }, { "epoch": 0.5602842250488141, "grad_norm": 1.299394965171814, "learning_rate": 8.53955232204226e-06, "loss": 0.15391652584075927, "step": 21270 }, { "epoch": 0.5605476402933128, "grad_norm": 2.265997886657715, "learning_rate": 8.531112955464728e-06, "loss": 0.15530860424041748, "step": 21280 }, { "epoch": 0.5608110555378116, "grad_norm": 1.207743763923645, "learning_rate": 8.522674658006564e-06, "loss": 0.14503846168518067, "step": 21290 }, { "epoch": 0.5610744707823103, "grad_norm": 2.017824649810791, "learning_rate": 8.514237435809517e-06, "loss": 0.14512238502502442, "step": 21300 }, { "epoch": 0.5613378860268091, "grad_norm": 0.9884617328643799, "learning_rate": 8.505801295014569e-06, "loss": 0.14308631420135498, "step": 21310 }, { "epoch": 0.5616013012713078, "grad_norm": 2.8635237216949463, "learning_rate": 8.497366241761907e-06, "loss": 0.15134611129760742, "step": 21320 }, { "epoch": 0.5618647165158066, "grad_norm": 0.9962643980979919, "learning_rate": 8.488932282190914e-06, "loss": 0.14448215961456298, "step": 21330 }, { "epoch": 0.5621281317603053, "grad_norm": 1.3781794309616089, "learning_rate": 8.480499422440207e-06, "loss": 0.1481541633605957, "step": 21340 }, { "epoch": 0.562391547004804, "grad_norm": 2.697906970977783, "learning_rate": 8.472067668647571e-06, "loss": 0.15261608362197876, "step": 21350 }, { "epoch": 0.5626549622493028, "grad_norm": 1.6755796670913696, "learning_rate": 8.463637026950013e-06, "loss": 0.1537643313407898, "step": 21360 }, { "epoch": 0.5629183774938015, "grad_norm": 1.3861384391784668, "learning_rate": 8.455207503483708e-06, "loss": 0.14919593334197997, "step": 21370 }, { "epoch": 0.5631817927383003, "grad_norm": 4.602996349334717, "learning_rate": 8.446779104384028e-06, "loss": 0.15156329870224, "step": 21380 }, { "epoch": 0.563445207982799, "grad_norm": 1.9435186386108398, "learning_rate": 8.43835183578553e-06, "loss": 0.1408689022064209, "step": 21390 }, { "epoch": 0.5637086232272978, "grad_norm": 1.5923078060150146, "learning_rate": 8.429925703821941e-06, "loss": 0.16210002899169923, "step": 21400 }, { "epoch": 0.5639720384717964, "grad_norm": 2.894946336746216, "learning_rate": 8.421500714626164e-06, "loss": 0.1658656120300293, "step": 21410 }, { "epoch": 0.5642354537162952, "grad_norm": 2.1379106044769287, "learning_rate": 8.413076874330266e-06, "loss": 0.1526726722717285, "step": 21420 }, { "epoch": 0.5644988689607939, "grad_norm": 1.1701951026916504, "learning_rate": 8.404654189065486e-06, "loss": 0.15173811912536622, "step": 21430 }, { "epoch": 0.5647622842052926, "grad_norm": 1.2878739833831787, "learning_rate": 8.396232664962218e-06, "loss": 0.15814650058746338, "step": 21440 }, { "epoch": 0.5650256994497914, "grad_norm": 0.9315606951713562, "learning_rate": 8.387812308150011e-06, "loss": 0.12377278804779053, "step": 21450 }, { "epoch": 0.5652891146942901, "grad_norm": 1.4814682006835938, "learning_rate": 8.379393124757554e-06, "loss": 0.17328524589538574, "step": 21460 }, { "epoch": 0.5655525299387889, "grad_norm": 1.082111120223999, "learning_rate": 8.370975120912708e-06, "loss": 0.15778810977935792, "step": 21470 }, { "epoch": 0.5658159451832876, "grad_norm": 4.145496368408203, "learning_rate": 8.362558302742447e-06, "loss": 0.14600108861923217, "step": 21480 }, { "epoch": 0.5660793604277864, "grad_norm": 3.5621421337127686, "learning_rate": 8.354142676372898e-06, "loss": 0.1531627893447876, "step": 21490 }, { "epoch": 0.5663427756722851, "grad_norm": 0.9966408610343933, "learning_rate": 8.345728247929319e-06, "loss": 0.14134550094604492, "step": 21500 }, { "epoch": 0.5666061909167839, "grad_norm": 2.519148588180542, "learning_rate": 8.33731502353609e-06, "loss": 0.14773252010345458, "step": 21510 }, { "epoch": 0.5668696061612826, "grad_norm": 2.901920795440674, "learning_rate": 8.328903009316724e-06, "loss": 0.1700361490249634, "step": 21520 }, { "epoch": 0.5671330214057813, "grad_norm": 2.468611478805542, "learning_rate": 8.320492211393844e-06, "loss": 0.15331441164016724, "step": 21530 }, { "epoch": 0.5673964366502801, "grad_norm": 4.126303195953369, "learning_rate": 8.312082635889193e-06, "loss": 0.1564542293548584, "step": 21540 }, { "epoch": 0.5676598518947787, "grad_norm": 1.1334903240203857, "learning_rate": 8.303674288923621e-06, "loss": 0.1498885154724121, "step": 21550 }, { "epoch": 0.5679232671392775, "grad_norm": 2.221412181854248, "learning_rate": 8.29526717661709e-06, "loss": 0.15272374153137208, "step": 21560 }, { "epoch": 0.5681866823837762, "grad_norm": 2.0534751415252686, "learning_rate": 8.286861305088655e-06, "loss": 0.1403123617172241, "step": 21570 }, { "epoch": 0.568450097628275, "grad_norm": 1.445168375968933, "learning_rate": 8.278456680456476e-06, "loss": 0.14494216442108154, "step": 21580 }, { "epoch": 0.5687135128727737, "grad_norm": 1.9889863729476929, "learning_rate": 8.270053308837791e-06, "loss": 0.1439493179321289, "step": 21590 }, { "epoch": 0.5689769281172725, "grad_norm": 2.009777784347534, "learning_rate": 8.261651196348953e-06, "loss": 0.14841580390930176, "step": 21600 }, { "epoch": 0.5692403433617712, "grad_norm": 2.4991977214813232, "learning_rate": 8.253250349105369e-06, "loss": 0.14973641633987428, "step": 21610 }, { "epoch": 0.5695037586062699, "grad_norm": 1.4002765417099, "learning_rate": 8.24485077322154e-06, "loss": 0.14732744693756103, "step": 21620 }, { "epoch": 0.5697671738507687, "grad_norm": 1.7210713624954224, "learning_rate": 8.236452474811046e-06, "loss": 0.1476450204849243, "step": 21630 }, { "epoch": 0.5700305890952674, "grad_norm": 1.7765886783599854, "learning_rate": 8.228055459986521e-06, "loss": 0.1518510937690735, "step": 21640 }, { "epoch": 0.5702940043397662, "grad_norm": 2.5670037269592285, "learning_rate": 8.219659734859686e-06, "loss": 0.13456717729568482, "step": 21650 }, { "epoch": 0.5705574195842649, "grad_norm": 2.2721190452575684, "learning_rate": 8.211265305541304e-06, "loss": 0.1520301342010498, "step": 21660 }, { "epoch": 0.5708208348287637, "grad_norm": 2.8183276653289795, "learning_rate": 8.202872178141204e-06, "loss": 0.15167454481124878, "step": 21670 }, { "epoch": 0.5710842500732624, "grad_norm": 6.4548749923706055, "learning_rate": 8.19448035876827e-06, "loss": 0.1387078046798706, "step": 21680 }, { "epoch": 0.5713476653177612, "grad_norm": 1.7755076885223389, "learning_rate": 8.186089853530426e-06, "loss": 0.15101919174194336, "step": 21690 }, { "epoch": 0.5716110805622598, "grad_norm": 4.630657196044922, "learning_rate": 8.177700668534646e-06, "loss": 0.15347015857696533, "step": 21700 }, { "epoch": 0.5718744958067585, "grad_norm": 1.2079713344573975, "learning_rate": 8.169312809886944e-06, "loss": 0.13818483352661132, "step": 21710 }, { "epoch": 0.5721379110512573, "grad_norm": 1.9203137159347534, "learning_rate": 8.160926283692355e-06, "loss": 0.14755859375, "step": 21720 }, { "epoch": 0.572401326295756, "grad_norm": 2.2704005241394043, "learning_rate": 8.152541096054965e-06, "loss": 0.1381232738494873, "step": 21730 }, { "epoch": 0.5726647415402548, "grad_norm": 1.5123305320739746, "learning_rate": 8.144157253077871e-06, "loss": 0.1511567234992981, "step": 21740 }, { "epoch": 0.5729281567847535, "grad_norm": 1.9170429706573486, "learning_rate": 8.135774760863193e-06, "loss": 0.15164774656295776, "step": 21750 }, { "epoch": 0.5731915720292523, "grad_norm": 1.375971794128418, "learning_rate": 8.127393625512077e-06, "loss": 0.14528651237487794, "step": 21760 }, { "epoch": 0.573454987273751, "grad_norm": 2.1837260723114014, "learning_rate": 8.11901385312467e-06, "loss": 0.1514669179916382, "step": 21770 }, { "epoch": 0.5737184025182498, "grad_norm": 1.0490000247955322, "learning_rate": 8.110635449800132e-06, "loss": 0.1559117078781128, "step": 21780 }, { "epoch": 0.5739818177627485, "grad_norm": 1.7101370096206665, "learning_rate": 8.10225842163663e-06, "loss": 0.15693798065185546, "step": 21790 }, { "epoch": 0.5742452330072472, "grad_norm": 2.257594108581543, "learning_rate": 8.093882774731322e-06, "loss": 0.1618669033050537, "step": 21800 }, { "epoch": 0.574508648251746, "grad_norm": 3.0679738521575928, "learning_rate": 8.085508515180373e-06, "loss": 0.15004403591156007, "step": 21810 }, { "epoch": 0.5747720634962447, "grad_norm": 2.156895399093628, "learning_rate": 8.077135649078917e-06, "loss": 0.14363982677459716, "step": 21820 }, { "epoch": 0.5750354787407435, "grad_norm": 2.7678658962249756, "learning_rate": 8.068764182521102e-06, "loss": 0.15894632339477538, "step": 21830 }, { "epoch": 0.5752988939852421, "grad_norm": 2.5457229614257812, "learning_rate": 8.060394121600033e-06, "loss": 0.15251970291137695, "step": 21840 }, { "epoch": 0.5755623092297409, "grad_norm": 0.973848283290863, "learning_rate": 8.052025472407802e-06, "loss": 0.15320416688919067, "step": 21850 }, { "epoch": 0.5758257244742396, "grad_norm": 1.452889323234558, "learning_rate": 8.043658241035477e-06, "loss": 0.1571507692337036, "step": 21860 }, { "epoch": 0.5760891397187383, "grad_norm": 1.3033028841018677, "learning_rate": 8.035292433573088e-06, "loss": 0.14051241874694825, "step": 21870 }, { "epoch": 0.5763525549632371, "grad_norm": 2.092637538909912, "learning_rate": 8.026928056109628e-06, "loss": 0.14146326780319213, "step": 21880 }, { "epoch": 0.5766159702077358, "grad_norm": 3.0527215003967285, "learning_rate": 8.018565114733056e-06, "loss": 0.16850028038024903, "step": 21890 }, { "epoch": 0.5768793854522346, "grad_norm": 1.1087770462036133, "learning_rate": 8.010203615530279e-06, "loss": 0.1451714515686035, "step": 21900 }, { "epoch": 0.5771428006967333, "grad_norm": 2.4026777744293213, "learning_rate": 8.00184356458716e-06, "loss": 0.15095179080963134, "step": 21910 }, { "epoch": 0.5774062159412321, "grad_norm": 2.5038814544677734, "learning_rate": 7.993484967988504e-06, "loss": 0.14310754537582399, "step": 21920 }, { "epoch": 0.5776696311857308, "grad_norm": 1.315515398979187, "learning_rate": 7.98512783181805e-06, "loss": 0.14041528701782227, "step": 21930 }, { "epoch": 0.5779330464302296, "grad_norm": 0.9767900705337524, "learning_rate": 7.976772162158498e-06, "loss": 0.15893347263336183, "step": 21940 }, { "epoch": 0.5781964616747283, "grad_norm": 1.9000744819641113, "learning_rate": 7.968417965091449e-06, "loss": 0.13721256256103515, "step": 21950 }, { "epoch": 0.578459876919227, "grad_norm": 3.3908188343048096, "learning_rate": 7.960065246697456e-06, "loss": 0.14710378646850586, "step": 21960 }, { "epoch": 0.5787232921637258, "grad_norm": 1.091012716293335, "learning_rate": 7.951714013055988e-06, "loss": 0.15285124778747558, "step": 21970 }, { "epoch": 0.5789867074082244, "grad_norm": 2.5444278717041016, "learning_rate": 7.94336427024543e-06, "loss": 0.14912824630737304, "step": 21980 }, { "epoch": 0.5792501226527232, "grad_norm": 2.319856643676758, "learning_rate": 7.93501602434309e-06, "loss": 0.1481982111930847, "step": 21990 }, { "epoch": 0.5795135378972219, "grad_norm": 1.7409952878952026, "learning_rate": 7.926669281425177e-06, "loss": 0.15709508657455445, "step": 22000 }, { "epoch": 0.5797769531417207, "grad_norm": 2.6837894916534424, "learning_rate": 7.91832404756681e-06, "loss": 0.16535511016845703, "step": 22010 }, { "epoch": 0.5800403683862194, "grad_norm": 3.0951321125030518, "learning_rate": 7.909980328842015e-06, "loss": 0.15162594318389894, "step": 22020 }, { "epoch": 0.5803037836307182, "grad_norm": 1.483004093170166, "learning_rate": 7.901638131323705e-06, "loss": 0.1454930305480957, "step": 22030 }, { "epoch": 0.5805671988752169, "grad_norm": 1.8210543394088745, "learning_rate": 7.893297461083697e-06, "loss": 0.1514162540435791, "step": 22040 }, { "epoch": 0.5808306141197156, "grad_norm": 2.857923984527588, "learning_rate": 7.884958324192689e-06, "loss": 0.15727224349975585, "step": 22050 }, { "epoch": 0.5810940293642144, "grad_norm": 2.150503396987915, "learning_rate": 7.876620726720257e-06, "loss": 0.16279019117355348, "step": 22060 }, { "epoch": 0.5813574446087131, "grad_norm": 1.2481147050857544, "learning_rate": 7.868284674734877e-06, "loss": 0.15616388320922853, "step": 22070 }, { "epoch": 0.5816208598532119, "grad_norm": 1.1596511602401733, "learning_rate": 7.859950174303879e-06, "loss": 0.15304977893829347, "step": 22080 }, { "epoch": 0.5818842750977106, "grad_norm": 2.127354145050049, "learning_rate": 7.851617231493472e-06, "loss": 0.14736783504486084, "step": 22090 }, { "epoch": 0.5821476903422094, "grad_norm": 3.0403342247009277, "learning_rate": 7.843285852368734e-06, "loss": 0.15434261560440063, "step": 22100 }, { "epoch": 0.5824111055867081, "grad_norm": 1.2195839881896973, "learning_rate": 7.8349560429936e-06, "loss": 0.1397296667098999, "step": 22110 }, { "epoch": 0.5826745208312069, "grad_norm": 3.459505319595337, "learning_rate": 7.826627809430867e-06, "loss": 0.16076960563659667, "step": 22120 }, { "epoch": 0.5829379360757055, "grad_norm": 1.397902011871338, "learning_rate": 7.818301157742182e-06, "loss": 0.1449904203414917, "step": 22130 }, { "epoch": 0.5832013513202042, "grad_norm": 2.325383424758911, "learning_rate": 7.809976093988039e-06, "loss": 0.14542801380157472, "step": 22140 }, { "epoch": 0.583464766564703, "grad_norm": 2.8130452632904053, "learning_rate": 7.801652624227782e-06, "loss": 0.15702426433563232, "step": 22150 }, { "epoch": 0.5837281818092017, "grad_norm": 2.5039663314819336, "learning_rate": 7.79333075451959e-06, "loss": 0.1441938042640686, "step": 22160 }, { "epoch": 0.5839915970537005, "grad_norm": 1.5281000137329102, "learning_rate": 7.785010490920477e-06, "loss": 0.14781497716903685, "step": 22170 }, { "epoch": 0.5842550122981992, "grad_norm": 3.237339735031128, "learning_rate": 7.776691839486297e-06, "loss": 0.14616014957427978, "step": 22180 }, { "epoch": 0.584518427542698, "grad_norm": 1.6914552450180054, "learning_rate": 7.76837480627171e-06, "loss": 0.1515043020248413, "step": 22190 }, { "epoch": 0.5847818427871967, "grad_norm": 3.3709263801574707, "learning_rate": 7.760059397330225e-06, "loss": 0.13043227195739746, "step": 22200 }, { "epoch": 0.5850452580316955, "grad_norm": 2.3561058044433594, "learning_rate": 7.75174561871415e-06, "loss": 0.15133163928985596, "step": 22210 }, { "epoch": 0.5853086732761942, "grad_norm": 1.774337887763977, "learning_rate": 7.743433476474608e-06, "loss": 0.14595566987991332, "step": 22220 }, { "epoch": 0.5855720885206929, "grad_norm": 3.8062052726745605, "learning_rate": 7.73512297666154e-06, "loss": 0.14536504745483397, "step": 22230 }, { "epoch": 0.5858355037651917, "grad_norm": 1.8209583759307861, "learning_rate": 7.726814125323683e-06, "loss": 0.14177652597427368, "step": 22240 }, { "epoch": 0.5860989190096904, "grad_norm": 4.58866024017334, "learning_rate": 7.718506928508583e-06, "loss": 0.14348435401916504, "step": 22250 }, { "epoch": 0.5863623342541892, "grad_norm": 1.9337043762207031, "learning_rate": 7.71020139226257e-06, "loss": 0.14208612442016602, "step": 22260 }, { "epoch": 0.5866257494986878, "grad_norm": 1.4246512651443481, "learning_rate": 7.701897522630772e-06, "loss": 0.14236719608306886, "step": 22270 }, { "epoch": 0.5868891647431866, "grad_norm": 1.5509198904037476, "learning_rate": 7.693595325657109e-06, "loss": 0.15730087757110595, "step": 22280 }, { "epoch": 0.5871525799876853, "grad_norm": 1.3911042213439941, "learning_rate": 7.685294807384273e-06, "loss": 0.14318013191223145, "step": 22290 }, { "epoch": 0.5874159952321841, "grad_norm": 2.1066770553588867, "learning_rate": 7.676995973853743e-06, "loss": 0.14343042373657228, "step": 22300 }, { "epoch": 0.5876794104766828, "grad_norm": 2.8381683826446533, "learning_rate": 7.668698831105768e-06, "loss": 0.1340481758117676, "step": 22310 }, { "epoch": 0.5879428257211815, "grad_norm": 1.5812296867370605, "learning_rate": 7.66040338517936e-06, "loss": 0.15555635690689087, "step": 22320 }, { "epoch": 0.5882062409656803, "grad_norm": 1.455777883529663, "learning_rate": 7.652109642112308e-06, "loss": 0.14766513109207152, "step": 22330 }, { "epoch": 0.588469656210179, "grad_norm": 1.2300065755844116, "learning_rate": 7.643817607941157e-06, "loss": 0.13448946475982665, "step": 22340 }, { "epoch": 0.5887330714546778, "grad_norm": 2.4300692081451416, "learning_rate": 7.635527288701201e-06, "loss": 0.13493034839630128, "step": 22350 }, { "epoch": 0.5889964866991765, "grad_norm": 2.1854538917541504, "learning_rate": 7.6272386904265005e-06, "loss": 0.14899163246154784, "step": 22360 }, { "epoch": 0.5892599019436753, "grad_norm": 1.184741497039795, "learning_rate": 7.618951819149848e-06, "loss": 0.13965182304382323, "step": 22370 }, { "epoch": 0.589523317188174, "grad_norm": 1.4863011837005615, "learning_rate": 7.610666680902791e-06, "loss": 0.12393391132354736, "step": 22380 }, { "epoch": 0.5897867324326727, "grad_norm": 1.2632588148117065, "learning_rate": 7.602383281715608e-06, "loss": 0.13394389152526856, "step": 22390 }, { "epoch": 0.5900501476771715, "grad_norm": 3.718641757965088, "learning_rate": 7.594101627617312e-06, "loss": 0.13468509912490845, "step": 22400 }, { "epoch": 0.5903135629216701, "grad_norm": 3.2452502250671387, "learning_rate": 7.585821724635656e-06, "loss": 0.14766185283660888, "step": 22410 }, { "epoch": 0.5905769781661689, "grad_norm": 2.3887319564819336, "learning_rate": 7.577543578797097e-06, "loss": 0.1434611439704895, "step": 22420 }, { "epoch": 0.5908403934106676, "grad_norm": 4.553908824920654, "learning_rate": 7.569267196126844e-06, "loss": 0.14465751647949218, "step": 22430 }, { "epoch": 0.5911038086551664, "grad_norm": 1.9369608163833618, "learning_rate": 7.560992582648793e-06, "loss": 0.16043039560317993, "step": 22440 }, { "epoch": 0.5913672238996651, "grad_norm": 0.8241461515426636, "learning_rate": 7.552719744385566e-06, "loss": 0.141318678855896, "step": 22450 }, { "epoch": 0.5916306391441639, "grad_norm": 1.6627668142318726, "learning_rate": 7.544448687358495e-06, "loss": 0.1510116457939148, "step": 22460 }, { "epoch": 0.5918940543886626, "grad_norm": 1.6888306140899658, "learning_rate": 7.536179417587612e-06, "loss": 0.1490260362625122, "step": 22470 }, { "epoch": 0.5921574696331613, "grad_norm": 1.988524317741394, "learning_rate": 7.527911941091644e-06, "loss": 0.138970685005188, "step": 22480 }, { "epoch": 0.5924208848776601, "grad_norm": 1.8566113710403442, "learning_rate": 7.519646263888023e-06, "loss": 0.15685584545135497, "step": 22490 }, { "epoch": 0.5926843001221588, "grad_norm": 2.0054051876068115, "learning_rate": 7.511382391992861e-06, "loss": 0.13944950103759765, "step": 22500 }, { "epoch": 0.5929477153666576, "grad_norm": 2.328723907470703, "learning_rate": 7.503120331420963e-06, "loss": 0.14911603927612305, "step": 22510 }, { "epoch": 0.5932111306111563, "grad_norm": 1.5910354852676392, "learning_rate": 7.494860088185815e-06, "loss": 0.14473676681518555, "step": 22520 }, { "epoch": 0.5934745458556551, "grad_norm": 1.2476476430892944, "learning_rate": 7.48660166829957e-06, "loss": 0.14211198091506957, "step": 22530 }, { "epoch": 0.5937379611001538, "grad_norm": 2.3416574001312256, "learning_rate": 7.478345077773075e-06, "loss": 0.1385371446609497, "step": 22540 }, { "epoch": 0.5940013763446526, "grad_norm": 2.116255760192871, "learning_rate": 7.470090322615822e-06, "loss": 0.14965008497238158, "step": 22550 }, { "epoch": 0.5942647915891512, "grad_norm": 3.205706834793091, "learning_rate": 7.4618374088359805e-06, "loss": 0.14894652366638184, "step": 22560 }, { "epoch": 0.5945282068336499, "grad_norm": 1.7929936647415161, "learning_rate": 7.45358634244038e-06, "loss": 0.138082754611969, "step": 22570 }, { "epoch": 0.5947916220781487, "grad_norm": 2.7964565753936768, "learning_rate": 7.4453371294345e-06, "loss": 0.143700110912323, "step": 22580 }, { "epoch": 0.5950550373226474, "grad_norm": 1.4694583415985107, "learning_rate": 7.437089775822477e-06, "loss": 0.1593989133834839, "step": 22590 }, { "epoch": 0.5953184525671462, "grad_norm": 3.3918967247009277, "learning_rate": 7.4288442876070865e-06, "loss": 0.14521771669387817, "step": 22600 }, { "epoch": 0.5955818678116449, "grad_norm": 1.9721447229385376, "learning_rate": 7.420600670789753e-06, "loss": 0.15154783725738524, "step": 22610 }, { "epoch": 0.5958452830561437, "grad_norm": 1.5324522256851196, "learning_rate": 7.412358931370536e-06, "loss": 0.14409937858581542, "step": 22620 }, { "epoch": 0.5961086983006424, "grad_norm": 1.1540203094482422, "learning_rate": 7.404119075348129e-06, "loss": 0.135159695148468, "step": 22630 }, { "epoch": 0.5963721135451412, "grad_norm": 1.3409498929977417, "learning_rate": 7.3958811087198555e-06, "loss": 0.14414625167846679, "step": 22640 }, { "epoch": 0.5966355287896399, "grad_norm": 2.053375244140625, "learning_rate": 7.3876450374816644e-06, "loss": 0.1285862445831299, "step": 22650 }, { "epoch": 0.5968989440341386, "grad_norm": 2.684173583984375, "learning_rate": 7.379410867628115e-06, "loss": 0.14595745801925658, "step": 22660 }, { "epoch": 0.5971623592786374, "grad_norm": 1.9410406351089478, "learning_rate": 7.371178605152405e-06, "loss": 0.147645902633667, "step": 22670 }, { "epoch": 0.597425774523136, "grad_norm": 2.034729480743408, "learning_rate": 7.362948256046321e-06, "loss": 0.13098952770233155, "step": 22680 }, { "epoch": 0.5976891897676349, "grad_norm": 2.2822771072387695, "learning_rate": 7.354719826300266e-06, "loss": 0.1466310977935791, "step": 22690 }, { "epoch": 0.5979526050121335, "grad_norm": 4.305606365203857, "learning_rate": 7.346493321903251e-06, "loss": 0.14816040992736818, "step": 22700 }, { "epoch": 0.5982160202566323, "grad_norm": 2.3036773204803467, "learning_rate": 7.338268748842876e-06, "loss": 0.14286338090896605, "step": 22710 }, { "epoch": 0.598479435501131, "grad_norm": 3.4856984615325928, "learning_rate": 7.330046113105345e-06, "loss": 0.15753631591796874, "step": 22720 }, { "epoch": 0.5987428507456298, "grad_norm": 2.0250704288482666, "learning_rate": 7.321825420675444e-06, "loss": 0.13230926990509034, "step": 22730 }, { "epoch": 0.5990062659901285, "grad_norm": 1.7638185024261475, "learning_rate": 7.3136066775365455e-06, "loss": 0.13256580829620362, "step": 22740 }, { "epoch": 0.5992696812346272, "grad_norm": 1.670737385749817, "learning_rate": 7.305389889670611e-06, "loss": 0.13362975120544435, "step": 22750 }, { "epoch": 0.599533096479126, "grad_norm": 2.7952139377593994, "learning_rate": 7.2971750630581675e-06, "loss": 0.15373237133026124, "step": 22760 }, { "epoch": 0.5997965117236247, "grad_norm": 1.9694678783416748, "learning_rate": 7.2889622036783245e-06, "loss": 0.13975625038146972, "step": 22770 }, { "epoch": 0.6000599269681235, "grad_norm": 1.3929451704025269, "learning_rate": 7.280751317508756e-06, "loss": 0.14234366416931152, "step": 22780 }, { "epoch": 0.6003233422126222, "grad_norm": 2.7170510292053223, "learning_rate": 7.272542410525691e-06, "loss": 0.14348337650299073, "step": 22790 }, { "epoch": 0.600586757457121, "grad_norm": 4.237412929534912, "learning_rate": 7.2643354887039395e-06, "loss": 0.15357513427734376, "step": 22800 }, { "epoch": 0.6008501727016197, "grad_norm": 2.204831600189209, "learning_rate": 7.256130558016843e-06, "loss": 0.1461409091949463, "step": 22810 }, { "epoch": 0.6011135879461185, "grad_norm": 1.6913361549377441, "learning_rate": 7.247927624436304e-06, "loss": 0.14630048274993895, "step": 22820 }, { "epoch": 0.6013770031906172, "grad_norm": 3.1044535636901855, "learning_rate": 7.239726693932779e-06, "loss": 0.15100336074829102, "step": 22830 }, { "epoch": 0.6016404184351158, "grad_norm": 1.8617202043533325, "learning_rate": 7.2315277724752495e-06, "loss": 0.13495848178863526, "step": 22840 }, { "epoch": 0.6019038336796146, "grad_norm": 1.5215630531311035, "learning_rate": 7.223330866031253e-06, "loss": 0.1442808151245117, "step": 22850 }, { "epoch": 0.6021672489241133, "grad_norm": 1.9715726375579834, "learning_rate": 7.215135980566847e-06, "loss": 0.14068679809570311, "step": 22860 }, { "epoch": 0.6024306641686121, "grad_norm": 2.365339756011963, "learning_rate": 7.2069431220466215e-06, "loss": 0.14341208934783936, "step": 22870 }, { "epoch": 0.6026940794131108, "grad_norm": 2.5653836727142334, "learning_rate": 7.198752296433697e-06, "loss": 0.1483376979827881, "step": 22880 }, { "epoch": 0.6029574946576096, "grad_norm": 1.2213667631149292, "learning_rate": 7.190563509689705e-06, "loss": 0.14487230777740479, "step": 22890 }, { "epoch": 0.6032209099021083, "grad_norm": 1.8696280717849731, "learning_rate": 7.182376767774803e-06, "loss": 0.1382954478263855, "step": 22900 }, { "epoch": 0.603484325146607, "grad_norm": 1.84858238697052, "learning_rate": 7.174192076647653e-06, "loss": 0.14678280353546141, "step": 22910 }, { "epoch": 0.6037477403911058, "grad_norm": 0.9320951700210571, "learning_rate": 7.166009442265423e-06, "loss": 0.13281883001327516, "step": 22920 }, { "epoch": 0.6040111556356045, "grad_norm": 1.936766266822815, "learning_rate": 7.15782887058379e-06, "loss": 0.1299246907234192, "step": 22930 }, { "epoch": 0.6042745708801033, "grad_norm": 2.4732720851898193, "learning_rate": 7.149650367556926e-06, "loss": 0.13457320928573607, "step": 22940 }, { "epoch": 0.604537986124602, "grad_norm": 1.5709097385406494, "learning_rate": 7.141473939137498e-06, "loss": 0.14223105907440187, "step": 22950 }, { "epoch": 0.6048014013691008, "grad_norm": 1.5986446142196655, "learning_rate": 7.133299591276665e-06, "loss": 0.14995031356811522, "step": 22960 }, { "epoch": 0.6050648166135995, "grad_norm": 4.234095573425293, "learning_rate": 7.12512732992407e-06, "loss": 0.13652594089508058, "step": 22970 }, { "epoch": 0.6053282318580983, "grad_norm": 4.0054521560668945, "learning_rate": 7.1169571610278375e-06, "loss": 0.15890403985977172, "step": 22980 }, { "epoch": 0.6055916471025969, "grad_norm": 1.6259632110595703, "learning_rate": 7.108789090534569e-06, "loss": 0.1283024787902832, "step": 22990 }, { "epoch": 0.6058550623470956, "grad_norm": 1.241329550743103, "learning_rate": 7.100623124389334e-06, "loss": 0.1365225076675415, "step": 23000 }, { "epoch": 0.6061184775915944, "grad_norm": 1.8893052339553833, "learning_rate": 7.092459268535685e-06, "loss": 0.15215952396392823, "step": 23010 }, { "epoch": 0.6063818928360931, "grad_norm": 1.96189546585083, "learning_rate": 7.084297528915616e-06, "loss": 0.14554303884506226, "step": 23020 }, { "epoch": 0.6066453080805919, "grad_norm": 2.0850231647491455, "learning_rate": 7.076137911469603e-06, "loss": 0.13679624795913697, "step": 23030 }, { "epoch": 0.6069087233250906, "grad_norm": 1.818786859512329, "learning_rate": 7.067980422136565e-06, "loss": 0.14406239986419678, "step": 23040 }, { "epoch": 0.6071721385695894, "grad_norm": 3.93894100189209, "learning_rate": 7.059825066853872e-06, "loss": 0.14530675411224364, "step": 23050 }, { "epoch": 0.6074355538140881, "grad_norm": 1.8208248615264893, "learning_rate": 7.0516718515573445e-06, "loss": 0.14363372325897217, "step": 23060 }, { "epoch": 0.6076989690585869, "grad_norm": 1.6967777013778687, "learning_rate": 7.043520782181248e-06, "loss": 0.13623671531677245, "step": 23070 }, { "epoch": 0.6079623843030856, "grad_norm": 1.4872030019760132, "learning_rate": 7.035371864658276e-06, "loss": 0.15852892398834229, "step": 23080 }, { "epoch": 0.6082257995475843, "grad_norm": 2.995548725128174, "learning_rate": 7.0272251049195684e-06, "loss": 0.1336752653121948, "step": 23090 }, { "epoch": 0.6084892147920831, "grad_norm": 1.7018704414367676, "learning_rate": 7.019080508894683e-06, "loss": 0.14861745834350587, "step": 23100 }, { "epoch": 0.6087526300365818, "grad_norm": 1.4979008436203003, "learning_rate": 7.010938082511614e-06, "loss": 0.1443453311920166, "step": 23110 }, { "epoch": 0.6090160452810806, "grad_norm": 1.282771348953247, "learning_rate": 7.0027978316967705e-06, "loss": 0.13439433574676513, "step": 23120 }, { "epoch": 0.6092794605255792, "grad_norm": 2.6663129329681396, "learning_rate": 6.994659762374967e-06, "loss": 0.14060474634170533, "step": 23130 }, { "epoch": 0.609542875770078, "grad_norm": 2.2609763145446777, "learning_rate": 6.986523880469458e-06, "loss": 0.14457812309265136, "step": 23140 }, { "epoch": 0.6098062910145767, "grad_norm": 2.042227029800415, "learning_rate": 6.978390191901878e-06, "loss": 0.16356531381607056, "step": 23150 }, { "epoch": 0.6100697062590755, "grad_norm": 2.6301188468933105, "learning_rate": 6.970258702592277e-06, "loss": 0.14999274015426636, "step": 23160 }, { "epoch": 0.6103331215035742, "grad_norm": 2.023432970046997, "learning_rate": 6.962129418459109e-06, "loss": 0.1418467402458191, "step": 23170 }, { "epoch": 0.6105965367480729, "grad_norm": 1.841757893562317, "learning_rate": 6.954002345419214e-06, "loss": 0.1386662721633911, "step": 23180 }, { "epoch": 0.6108599519925717, "grad_norm": 2.175351619720459, "learning_rate": 6.945877489387829e-06, "loss": 0.15313611030578614, "step": 23190 }, { "epoch": 0.6111233672370704, "grad_norm": 1.3695399761199951, "learning_rate": 6.9377548562785735e-06, "loss": 0.15199632644653321, "step": 23200 }, { "epoch": 0.6113867824815692, "grad_norm": 2.601106643676758, "learning_rate": 6.929634452003447e-06, "loss": 0.14113422632217407, "step": 23210 }, { "epoch": 0.6116501977260679, "grad_norm": 2.807974100112915, "learning_rate": 6.921516282472837e-06, "loss": 0.15714645385742188, "step": 23220 }, { "epoch": 0.6119136129705667, "grad_norm": 2.5840556621551514, "learning_rate": 6.91340035359549e-06, "loss": 0.15036648511886597, "step": 23230 }, { "epoch": 0.6121770282150654, "grad_norm": 2.2286429405212402, "learning_rate": 6.905286671278536e-06, "loss": 0.14945926666259765, "step": 23240 }, { "epoch": 0.6124404434595642, "grad_norm": 3.274634599685669, "learning_rate": 6.897175241427462e-06, "loss": 0.141487979888916, "step": 23250 }, { "epoch": 0.6127038587040629, "grad_norm": 1.392308235168457, "learning_rate": 6.889066069946107e-06, "loss": 0.14425967931747435, "step": 23260 }, { "epoch": 0.6129672739485615, "grad_norm": 0.9942537546157837, "learning_rate": 6.880959162736691e-06, "loss": 0.14732975959777833, "step": 23270 }, { "epoch": 0.6132306891930603, "grad_norm": 1.9396220445632935, "learning_rate": 6.872854525699761e-06, "loss": 0.1408051371574402, "step": 23280 }, { "epoch": 0.613494104437559, "grad_norm": 2.2801594734191895, "learning_rate": 6.8647521647342205e-06, "loss": 0.1406729817390442, "step": 23290 }, { "epoch": 0.6137575196820578, "grad_norm": 1.797753930091858, "learning_rate": 6.856652085737323e-06, "loss": 0.14693881273269654, "step": 23300 }, { "epoch": 0.6140209349265565, "grad_norm": 1.7028121948242188, "learning_rate": 6.848554294604652e-06, "loss": 0.1388055920600891, "step": 23310 }, { "epoch": 0.6142843501710553, "grad_norm": 1.131317138671875, "learning_rate": 6.8404587972301316e-06, "loss": 0.13730791807174683, "step": 23320 }, { "epoch": 0.614547765415554, "grad_norm": 1.8143483400344849, "learning_rate": 6.8323655995060135e-06, "loss": 0.154953670501709, "step": 23330 }, { "epoch": 0.6148111806600528, "grad_norm": 1.9665580987930298, "learning_rate": 6.824274707322874e-06, "loss": 0.1536205768585205, "step": 23340 }, { "epoch": 0.6150745959045515, "grad_norm": 2.641129970550537, "learning_rate": 6.816186126569615e-06, "loss": 0.15119999647140503, "step": 23350 }, { "epoch": 0.6153380111490502, "grad_norm": 4.644169330596924, "learning_rate": 6.808099863133453e-06, "loss": 0.1507734775543213, "step": 23360 }, { "epoch": 0.615601426393549, "grad_norm": 1.7185797691345215, "learning_rate": 6.800015922899923e-06, "loss": 0.15338102579116822, "step": 23370 }, { "epoch": 0.6158648416380477, "grad_norm": 1.231269359588623, "learning_rate": 6.7919343117528655e-06, "loss": 0.14025022983551025, "step": 23380 }, { "epoch": 0.6161282568825465, "grad_norm": 2.6468043327331543, "learning_rate": 6.783855035574418e-06, "loss": 0.1500670313835144, "step": 23390 }, { "epoch": 0.6163916721270452, "grad_norm": 2.0409257411956787, "learning_rate": 6.775778100245038e-06, "loss": 0.1482539176940918, "step": 23400 }, { "epoch": 0.616655087371544, "grad_norm": 2.384662389755249, "learning_rate": 6.76770351164346e-06, "loss": 0.13786070346832274, "step": 23410 }, { "epoch": 0.6169185026160426, "grad_norm": 4.181931495666504, "learning_rate": 6.759631275646715e-06, "loss": 0.14515602588653564, "step": 23420 }, { "epoch": 0.6171819178605413, "grad_norm": 2.5952630043029785, "learning_rate": 6.751561398130133e-06, "loss": 0.14480243921279906, "step": 23430 }, { "epoch": 0.6174453331050401, "grad_norm": 2.4398136138916016, "learning_rate": 6.743493884967312e-06, "loss": 0.14892103672027587, "step": 23440 }, { "epoch": 0.6177087483495388, "grad_norm": 2.649855852127075, "learning_rate": 6.735428742030141e-06, "loss": 0.13542081117630006, "step": 23450 }, { "epoch": 0.6179721635940376, "grad_norm": 2.019101619720459, "learning_rate": 6.7273659751887775e-06, "loss": 0.1472460389137268, "step": 23460 }, { "epoch": 0.6182355788385363, "grad_norm": 1.593988299369812, "learning_rate": 6.719305590311649e-06, "loss": 0.1406148910522461, "step": 23470 }, { "epoch": 0.6184989940830351, "grad_norm": 1.822778582572937, "learning_rate": 6.7112475932654554e-06, "loss": 0.1389913558959961, "step": 23480 }, { "epoch": 0.6187624093275338, "grad_norm": 1.985480546951294, "learning_rate": 6.70319198991515e-06, "loss": 0.1378423571586609, "step": 23490 }, { "epoch": 0.6190258245720326, "grad_norm": 1.858218789100647, "learning_rate": 6.695138786123952e-06, "loss": 0.13166096210479736, "step": 23500 }, { "epoch": 0.6192892398165313, "grad_norm": 1.6147857904434204, "learning_rate": 6.687087987753331e-06, "loss": 0.1412494659423828, "step": 23510 }, { "epoch": 0.61955265506103, "grad_norm": 2.318169593811035, "learning_rate": 6.679039600663001e-06, "loss": 0.14975005388259888, "step": 23520 }, { "epoch": 0.6198160703055288, "grad_norm": 1.2934218645095825, "learning_rate": 6.670993630710929e-06, "loss": 0.13770445585250854, "step": 23530 }, { "epoch": 0.6200794855500275, "grad_norm": 2.9619529247283936, "learning_rate": 6.662950083753319e-06, "loss": 0.1296025037765503, "step": 23540 }, { "epoch": 0.6203429007945263, "grad_norm": 1.5472904443740845, "learning_rate": 6.654908965644609e-06, "loss": 0.1476206064224243, "step": 23550 }, { "epoch": 0.6206063160390249, "grad_norm": 2.527945041656494, "learning_rate": 6.646870282237474e-06, "loss": 0.13451204299926758, "step": 23560 }, { "epoch": 0.6208697312835237, "grad_norm": 1.4486083984375, "learning_rate": 6.6388340393828125e-06, "loss": 0.14741653203964233, "step": 23570 }, { "epoch": 0.6211331465280224, "grad_norm": 2.033540725708008, "learning_rate": 6.6308002429297516e-06, "loss": 0.13671888113021852, "step": 23580 }, { "epoch": 0.6213965617725212, "grad_norm": 1.3214659690856934, "learning_rate": 6.622768898725634e-06, "loss": 0.12484915256500244, "step": 23590 }, { "epoch": 0.6216599770170199, "grad_norm": 2.931459665298462, "learning_rate": 6.614740012616017e-06, "loss": 0.14784932136535645, "step": 23600 }, { "epoch": 0.6219233922615186, "grad_norm": 4.474693775177002, "learning_rate": 6.606713590444673e-06, "loss": 0.14010016918182372, "step": 23610 }, { "epoch": 0.6221868075060174, "grad_norm": 2.364163637161255, "learning_rate": 6.598689638053577e-06, "loss": 0.1408665657043457, "step": 23620 }, { "epoch": 0.6224502227505161, "grad_norm": 2.2308523654937744, "learning_rate": 6.590668161282903e-06, "loss": 0.14291830062866212, "step": 23630 }, { "epoch": 0.6227136379950149, "grad_norm": 2.0526113510131836, "learning_rate": 6.582649165971037e-06, "loss": 0.13811343908309937, "step": 23640 }, { "epoch": 0.6229770532395136, "grad_norm": 1.0455418825149536, "learning_rate": 6.574632657954542e-06, "loss": 0.13383941650390624, "step": 23650 }, { "epoch": 0.6232404684840124, "grad_norm": 1.5266530513763428, "learning_rate": 6.5666186430681835e-06, "loss": 0.14243090152740479, "step": 23660 }, { "epoch": 0.6235038837285111, "grad_norm": 2.5418221950531006, "learning_rate": 6.558607127144905e-06, "loss": 0.13394877910614014, "step": 23670 }, { "epoch": 0.6237672989730099, "grad_norm": 1.9422070980072021, "learning_rate": 6.550598116015832e-06, "loss": 0.13483588695526122, "step": 23680 }, { "epoch": 0.6240307142175086, "grad_norm": 5.6151347160339355, "learning_rate": 6.54259161551027e-06, "loss": 0.15506430864334106, "step": 23690 }, { "epoch": 0.6242941294620072, "grad_norm": 2.182663679122925, "learning_rate": 6.534587631455696e-06, "loss": 0.13694182634353638, "step": 23700 }, { "epoch": 0.624557544706506, "grad_norm": 1.372154712677002, "learning_rate": 6.5265861696777556e-06, "loss": 0.1494031310081482, "step": 23710 }, { "epoch": 0.6248209599510047, "grad_norm": 3.571348190307617, "learning_rate": 6.518587236000259e-06, "loss": 0.14168102741241456, "step": 23720 }, { "epoch": 0.6250843751955035, "grad_norm": 2.499030828475952, "learning_rate": 6.510590836245166e-06, "loss": 0.13537235260009767, "step": 23730 }, { "epoch": 0.6253477904400022, "grad_norm": 2.0244529247283936, "learning_rate": 6.502596976232616e-06, "loss": 0.13965990543365478, "step": 23740 }, { "epoch": 0.625611205684501, "grad_norm": 4.270667552947998, "learning_rate": 6.494605661780874e-06, "loss": 0.14281845092773438, "step": 23750 }, { "epoch": 0.6258746209289997, "grad_norm": 2.0598065853118896, "learning_rate": 6.486616898706365e-06, "loss": 0.1438465714454651, "step": 23760 }, { "epoch": 0.6261380361734985, "grad_norm": 1.2261303663253784, "learning_rate": 6.47863069282366e-06, "loss": 0.1351830244064331, "step": 23770 }, { "epoch": 0.6264014514179972, "grad_norm": 1.3791226148605347, "learning_rate": 6.47064704994546e-06, "loss": 0.14115020036697387, "step": 23780 }, { "epoch": 0.6266648666624959, "grad_norm": 1.9163416624069214, "learning_rate": 6.462665975882607e-06, "loss": 0.13598206043243408, "step": 23790 }, { "epoch": 0.6269282819069947, "grad_norm": 3.239778995513916, "learning_rate": 6.454687476444071e-06, "loss": 0.13889918327331544, "step": 23800 }, { "epoch": 0.6271916971514934, "grad_norm": 1.7103655338287354, "learning_rate": 6.446711557436944e-06, "loss": 0.12743937969207764, "step": 23810 }, { "epoch": 0.6274551123959922, "grad_norm": 1.7402116060256958, "learning_rate": 6.438738224666452e-06, "loss": 0.14568272829055787, "step": 23820 }, { "epoch": 0.6277185276404909, "grad_norm": 2.184877395629883, "learning_rate": 6.4307674839359225e-06, "loss": 0.14200191497802733, "step": 23830 }, { "epoch": 0.6279819428849897, "grad_norm": 2.421497344970703, "learning_rate": 6.42279934104681e-06, "loss": 0.1328498363494873, "step": 23840 }, { "epoch": 0.6282453581294883, "grad_norm": 1.8640345335006714, "learning_rate": 6.4148338017986725e-06, "loss": 0.13810030221939087, "step": 23850 }, { "epoch": 0.6285087733739871, "grad_norm": 1.2449342012405396, "learning_rate": 6.406870871989166e-06, "loss": 0.1303342580795288, "step": 23860 }, { "epoch": 0.6287721886184858, "grad_norm": 2.023069143295288, "learning_rate": 6.398910557414064e-06, "loss": 0.14443204402923585, "step": 23870 }, { "epoch": 0.6290356038629845, "grad_norm": 2.503727436065674, "learning_rate": 6.390952863867224e-06, "loss": 0.13106073141098024, "step": 23880 }, { "epoch": 0.6292990191074833, "grad_norm": 2.995204210281372, "learning_rate": 6.382997797140591e-06, "loss": 0.14545685052871704, "step": 23890 }, { "epoch": 0.629562434351982, "grad_norm": 1.5962474346160889, "learning_rate": 6.375045363024217e-06, "loss": 0.13217689990997314, "step": 23900 }, { "epoch": 0.6298258495964808, "grad_norm": 3.4191198348999023, "learning_rate": 6.367095567306219e-06, "loss": 0.1593393325805664, "step": 23910 }, { "epoch": 0.6300892648409795, "grad_norm": 1.583286166191101, "learning_rate": 6.359148415772805e-06, "loss": 0.12039204835891723, "step": 23920 }, { "epoch": 0.6303526800854783, "grad_norm": 2.918832540512085, "learning_rate": 6.351203914208254e-06, "loss": 0.14583313465118408, "step": 23930 }, { "epoch": 0.630616095329977, "grad_norm": 1.4273062944412231, "learning_rate": 6.343262068394914e-06, "loss": 0.14617397785186767, "step": 23940 }, { "epoch": 0.6308795105744757, "grad_norm": 2.5822243690490723, "learning_rate": 6.335322884113208e-06, "loss": 0.15378227233886718, "step": 23950 }, { "epoch": 0.6311429258189745, "grad_norm": 3.160316228866577, "learning_rate": 6.3273863671416146e-06, "loss": 0.14725239276885987, "step": 23960 }, { "epoch": 0.6314063410634732, "grad_norm": 3.8029897212982178, "learning_rate": 6.319452523256673e-06, "loss": 0.1364882469177246, "step": 23970 }, { "epoch": 0.631669756307972, "grad_norm": 1.7435119152069092, "learning_rate": 6.31152135823298e-06, "loss": 0.144120991230011, "step": 23980 }, { "epoch": 0.6319331715524706, "grad_norm": 1.623769760131836, "learning_rate": 6.303592877843171e-06, "loss": 0.1386682152748108, "step": 23990 }, { "epoch": 0.6321965867969694, "grad_norm": 1.7071197032928467, "learning_rate": 6.295667087857948e-06, "loss": 0.13643925189971923, "step": 24000 }, { "epoch": 0.6324600020414681, "grad_norm": 3.2100648880004883, "learning_rate": 6.287743994046035e-06, "loss": 0.13293192386627198, "step": 24010 }, { "epoch": 0.6327234172859669, "grad_norm": 5.304239273071289, "learning_rate": 6.279823602174202e-06, "loss": 0.13834480047225953, "step": 24020 }, { "epoch": 0.6329868325304656, "grad_norm": 3.8826122283935547, "learning_rate": 6.271905918007255e-06, "loss": 0.15050172805786133, "step": 24030 }, { "epoch": 0.6332502477749643, "grad_norm": 2.140209436416626, "learning_rate": 6.263990947308022e-06, "loss": 0.1444446086883545, "step": 24040 }, { "epoch": 0.6335136630194631, "grad_norm": 1.5051078796386719, "learning_rate": 6.2560786958373645e-06, "loss": 0.14263710975646973, "step": 24050 }, { "epoch": 0.6337770782639618, "grad_norm": 1.29814875125885, "learning_rate": 6.248169169354159e-06, "loss": 0.1496143341064453, "step": 24060 }, { "epoch": 0.6340404935084606, "grad_norm": 1.57966148853302, "learning_rate": 6.240262373615298e-06, "loss": 0.13770569562911988, "step": 24070 }, { "epoch": 0.6343039087529593, "grad_norm": 1.2000198364257812, "learning_rate": 6.232358314375693e-06, "loss": 0.12784290313720703, "step": 24080 }, { "epoch": 0.6345673239974581, "grad_norm": 2.0438733100891113, "learning_rate": 6.224456997388254e-06, "loss": 0.133545184135437, "step": 24090 }, { "epoch": 0.6348307392419568, "grad_norm": 1.9427968263626099, "learning_rate": 6.2165584284039046e-06, "loss": 0.13822567462921143, "step": 24100 }, { "epoch": 0.6350941544864556, "grad_norm": 3.7758116722106934, "learning_rate": 6.208662613171566e-06, "loss": 0.14005563259124756, "step": 24110 }, { "epoch": 0.6353575697309543, "grad_norm": 3.6557090282440186, "learning_rate": 6.200769557438144e-06, "loss": 0.1433924674987793, "step": 24120 }, { "epoch": 0.6356209849754529, "grad_norm": 1.2158029079437256, "learning_rate": 6.192879266948554e-06, "loss": 0.13291711807250978, "step": 24130 }, { "epoch": 0.6358844002199517, "grad_norm": 1.9816151857376099, "learning_rate": 6.184991747445686e-06, "loss": 0.13894343376159668, "step": 24140 }, { "epoch": 0.6361478154644504, "grad_norm": 2.6586549282073975, "learning_rate": 6.1771070046704165e-06, "loss": 0.12944302558898926, "step": 24150 }, { "epoch": 0.6364112307089492, "grad_norm": 1.5191223621368408, "learning_rate": 6.169225044361604e-06, "loss": 0.14281437397003174, "step": 24160 }, { "epoch": 0.6366746459534479, "grad_norm": 4.165299415588379, "learning_rate": 6.161345872256075e-06, "loss": 0.14837509393692017, "step": 24170 }, { "epoch": 0.6369380611979467, "grad_norm": 2.219874858856201, "learning_rate": 6.153469494088638e-06, "loss": 0.13465571403503418, "step": 24180 }, { "epoch": 0.6372014764424454, "grad_norm": 2.891469717025757, "learning_rate": 6.1455959155920555e-06, "loss": 0.13796782493591309, "step": 24190 }, { "epoch": 0.6374648916869442, "grad_norm": 2.7262728214263916, "learning_rate": 6.1377251424970555e-06, "loss": 0.1318334698677063, "step": 24200 }, { "epoch": 0.6377283069314429, "grad_norm": 2.9026594161987305, "learning_rate": 6.1298571805323344e-06, "loss": 0.14014526605606079, "step": 24210 }, { "epoch": 0.6379917221759416, "grad_norm": 3.238727331161499, "learning_rate": 6.121992035424529e-06, "loss": 0.13226425647735596, "step": 24220 }, { "epoch": 0.6382551374204404, "grad_norm": 2.507070779800415, "learning_rate": 6.114129712898227e-06, "loss": 0.14738850593566893, "step": 24230 }, { "epoch": 0.6385185526649391, "grad_norm": 1.3882051706314087, "learning_rate": 6.106270218675975e-06, "loss": 0.13259215354919435, "step": 24240 }, { "epoch": 0.6387819679094379, "grad_norm": 1.859290361404419, "learning_rate": 6.098413558478244e-06, "loss": 0.14588794708251954, "step": 24250 }, { "epoch": 0.6390453831539366, "grad_norm": 1.455867886543274, "learning_rate": 6.090559738023456e-06, "loss": 0.1381139039993286, "step": 24260 }, { "epoch": 0.6393087983984354, "grad_norm": 2.547788143157959, "learning_rate": 6.0827087630279605e-06, "loss": 0.14690812826156616, "step": 24270 }, { "epoch": 0.639572213642934, "grad_norm": 1.3920692205429077, "learning_rate": 6.074860639206028e-06, "loss": 0.13803181648254395, "step": 24280 }, { "epoch": 0.6398356288874328, "grad_norm": 1.1971780061721802, "learning_rate": 6.067015372269872e-06, "loss": 0.1392286539077759, "step": 24290 }, { "epoch": 0.6400990441319315, "grad_norm": 1.7806363105773926, "learning_rate": 6.059172967929609e-06, "loss": 0.1335673451423645, "step": 24300 }, { "epoch": 0.6403624593764302, "grad_norm": 1.330469012260437, "learning_rate": 6.0513334318932825e-06, "loss": 0.1274864912033081, "step": 24310 }, { "epoch": 0.640625874620929, "grad_norm": 1.5502625703811646, "learning_rate": 6.043496769866848e-06, "loss": 0.14508620500564576, "step": 24320 }, { "epoch": 0.6408892898654277, "grad_norm": 1.8129595518112183, "learning_rate": 6.035662987554154e-06, "loss": 0.1340814471244812, "step": 24330 }, { "epoch": 0.6411527051099265, "grad_norm": 1.6284931898117065, "learning_rate": 6.027832090656982e-06, "loss": 0.14275140762329103, "step": 24340 }, { "epoch": 0.6414161203544252, "grad_norm": 1.9515572786331177, "learning_rate": 6.020004084874984e-06, "loss": 0.13780471086502075, "step": 24350 }, { "epoch": 0.641679535598924, "grad_norm": 1.728320837020874, "learning_rate": 6.012178975905725e-06, "loss": 0.1405393362045288, "step": 24360 }, { "epoch": 0.6419429508434227, "grad_norm": 2.415755271911621, "learning_rate": 6.0043567694446566e-06, "loss": 0.12265567779541016, "step": 24370 }, { "epoch": 0.6422063660879215, "grad_norm": 2.9262795448303223, "learning_rate": 5.9965374711851175e-06, "loss": 0.13656229972839357, "step": 24380 }, { "epoch": 0.6424697813324202, "grad_norm": 1.3684114217758179, "learning_rate": 5.9887210868183325e-06, "loss": 0.14409276247024536, "step": 24390 }, { "epoch": 0.6427331965769189, "grad_norm": 1.5813868045806885, "learning_rate": 5.9809076220334024e-06, "loss": 0.1393463969230652, "step": 24400 }, { "epoch": 0.6429966118214177, "grad_norm": 2.1806154251098633, "learning_rate": 5.973097082517301e-06, "loss": 0.1483757257461548, "step": 24410 }, { "epoch": 0.6432600270659163, "grad_norm": 3.973217487335205, "learning_rate": 5.965289473954879e-06, "loss": 0.12524446249008178, "step": 24420 }, { "epoch": 0.6435234423104151, "grad_norm": 1.9851083755493164, "learning_rate": 5.95748480202885e-06, "loss": 0.14493331909179688, "step": 24430 }, { "epoch": 0.6437868575549138, "grad_norm": 2.9258298873901367, "learning_rate": 5.949683072419792e-06, "loss": 0.14058104753494263, "step": 24440 }, { "epoch": 0.6440502727994126, "grad_norm": 2.16603422164917, "learning_rate": 5.941884290806142e-06, "loss": 0.1372858166694641, "step": 24450 }, { "epoch": 0.6443136880439113, "grad_norm": 1.4942771196365356, "learning_rate": 5.934088462864181e-06, "loss": 0.1450076460838318, "step": 24460 }, { "epoch": 0.6445771032884101, "grad_norm": 2.263479709625244, "learning_rate": 5.926295594268059e-06, "loss": 0.14720609188079833, "step": 24470 }, { "epoch": 0.6448405185329088, "grad_norm": 2.7532949447631836, "learning_rate": 5.918505690689759e-06, "loss": 0.1354233980178833, "step": 24480 }, { "epoch": 0.6451039337774075, "grad_norm": 2.926424980163574, "learning_rate": 5.910718757799105e-06, "loss": 0.14601573944091797, "step": 24490 }, { "epoch": 0.6453673490219063, "grad_norm": 3.1190297603607178, "learning_rate": 5.902934801263766e-06, "loss": 0.14451277256011963, "step": 24500 }, { "epoch": 0.645630764266405, "grad_norm": 1.068003535270691, "learning_rate": 5.89515382674924e-06, "loss": 0.1269064426422119, "step": 24510 }, { "epoch": 0.6458941795109038, "grad_norm": 1.9763262271881104, "learning_rate": 5.88737583991886e-06, "loss": 0.1625727415084839, "step": 24520 }, { "epoch": 0.6461575947554025, "grad_norm": 2.501380681991577, "learning_rate": 5.879600846433777e-06, "loss": 0.13451437950134276, "step": 24530 }, { "epoch": 0.6464210099999013, "grad_norm": 2.7632617950439453, "learning_rate": 5.871828851952965e-06, "loss": 0.14542877674102783, "step": 24540 }, { "epoch": 0.6466844252444, "grad_norm": 2.3201088905334473, "learning_rate": 5.864059862133221e-06, "loss": 0.1337432622909546, "step": 24550 }, { "epoch": 0.6469478404888986, "grad_norm": 1.1483792066574097, "learning_rate": 5.856293882629147e-06, "loss": 0.12900588512420655, "step": 24560 }, { "epoch": 0.6472112557333974, "grad_norm": 2.8969950675964355, "learning_rate": 5.848530919093162e-06, "loss": 0.14127684831619264, "step": 24570 }, { "epoch": 0.6474746709778961, "grad_norm": 2.3614752292633057, "learning_rate": 5.840770977175485e-06, "loss": 0.13639228343963622, "step": 24580 }, { "epoch": 0.6477380862223949, "grad_norm": 2.5407259464263916, "learning_rate": 5.833014062524129e-06, "loss": 0.13084619045257567, "step": 24590 }, { "epoch": 0.6480015014668936, "grad_norm": 1.9204645156860352, "learning_rate": 5.825260180784923e-06, "loss": 0.13130083084106445, "step": 24600 }, { "epoch": 0.6482649167113924, "grad_norm": 3.185931921005249, "learning_rate": 5.817509337601475e-06, "loss": 0.14075706005096436, "step": 24610 }, { "epoch": 0.6485283319558911, "grad_norm": 2.052656650543213, "learning_rate": 5.8097615386151686e-06, "loss": 0.13644609451293946, "step": 24620 }, { "epoch": 0.6487917472003899, "grad_norm": 2.918344020843506, "learning_rate": 5.8020167894652035e-06, "loss": 0.13392226696014403, "step": 24630 }, { "epoch": 0.6490551624448886, "grad_norm": 2.311426877975464, "learning_rate": 5.79427509578853e-06, "loss": 0.14268937110900878, "step": 24640 }, { "epoch": 0.6493185776893873, "grad_norm": 3.1074697971343994, "learning_rate": 5.786536463219899e-06, "loss": 0.14428868293762206, "step": 24650 }, { "epoch": 0.6495819929338861, "grad_norm": 1.8420931100845337, "learning_rate": 5.778800897391812e-06, "loss": 0.13111039400100707, "step": 24660 }, { "epoch": 0.6498454081783848, "grad_norm": 1.0847142934799194, "learning_rate": 5.771068403934545e-06, "loss": 0.14102423191070557, "step": 24670 }, { "epoch": 0.6501088234228836, "grad_norm": 2.9007747173309326, "learning_rate": 5.76333898847615e-06, "loss": 0.1290513038635254, "step": 24680 }, { "epoch": 0.6503722386673823, "grad_norm": 1.8083775043487549, "learning_rate": 5.755612656642424e-06, "loss": 0.1308225631713867, "step": 24690 }, { "epoch": 0.650635653911881, "grad_norm": 3.2962305545806885, "learning_rate": 5.747889414056925e-06, "loss": 0.1445026993751526, "step": 24700 }, { "epoch": 0.6508990691563797, "grad_norm": 2.5385589599609375, "learning_rate": 5.740169266340963e-06, "loss": 0.13474504947662352, "step": 24710 }, { "epoch": 0.6511624844008785, "grad_norm": 1.3232536315917969, "learning_rate": 5.732452219113589e-06, "loss": 0.14361793994903566, "step": 24720 }, { "epoch": 0.6514258996453772, "grad_norm": 3.021866798400879, "learning_rate": 5.724738277991611e-06, "loss": 0.14013113975524902, "step": 24730 }, { "epoch": 0.6516893148898759, "grad_norm": 2.1057982444763184, "learning_rate": 5.717027448589568e-06, "loss": 0.148160719871521, "step": 24740 }, { "epoch": 0.6519527301343747, "grad_norm": 1.1774977445602417, "learning_rate": 5.709319736519725e-06, "loss": 0.12676191329956055, "step": 24750 }, { "epoch": 0.6522161453788734, "grad_norm": 1.7519723176956177, "learning_rate": 5.701615147392096e-06, "loss": 0.1447360873222351, "step": 24760 }, { "epoch": 0.6524795606233722, "grad_norm": 1.5336990356445312, "learning_rate": 5.6939136868144075e-06, "loss": 0.13689138889312744, "step": 24770 }, { "epoch": 0.6527429758678709, "grad_norm": 1.6814764738082886, "learning_rate": 5.686215360392127e-06, "loss": 0.1281390905380249, "step": 24780 }, { "epoch": 0.6530063911123697, "grad_norm": 6.732689380645752, "learning_rate": 5.678520173728414e-06, "loss": 0.1344269633293152, "step": 24790 }, { "epoch": 0.6532698063568684, "grad_norm": 2.9469027519226074, "learning_rate": 5.67082813242416e-06, "loss": 0.14146735668182372, "step": 24800 }, { "epoch": 0.6535332216013672, "grad_norm": 1.103434443473816, "learning_rate": 5.663139242077971e-06, "loss": 0.14332296848297119, "step": 24810 }, { "epoch": 0.6537966368458659, "grad_norm": 1.2305998802185059, "learning_rate": 5.6554535082861496e-06, "loss": 0.15609861612319947, "step": 24820 }, { "epoch": 0.6540600520903646, "grad_norm": 2.317927122116089, "learning_rate": 5.647770936642705e-06, "loss": 0.127937650680542, "step": 24830 }, { "epoch": 0.6543234673348634, "grad_norm": 2.1750247478485107, "learning_rate": 5.6400915327393435e-06, "loss": 0.13392212390899658, "step": 24840 }, { "epoch": 0.654586882579362, "grad_norm": 1.6312428712844849, "learning_rate": 5.632415302165464e-06, "loss": 0.13735724687576295, "step": 24850 }, { "epoch": 0.6548502978238608, "grad_norm": 2.4964914321899414, "learning_rate": 5.624742250508164e-06, "loss": 0.12270524501800537, "step": 24860 }, { "epoch": 0.6551137130683595, "grad_norm": 2.1008424758911133, "learning_rate": 5.617072383352222e-06, "loss": 0.14595283269882203, "step": 24870 }, { "epoch": 0.6553771283128583, "grad_norm": 1.6361243724822998, "learning_rate": 5.6094057062800886e-06, "loss": 0.13130266666412355, "step": 24880 }, { "epoch": 0.655640543557357, "grad_norm": 1.9565337896347046, "learning_rate": 5.601742224871911e-06, "loss": 0.15120890140533447, "step": 24890 }, { "epoch": 0.6559039588018558, "grad_norm": 3.500741720199585, "learning_rate": 5.594081944705496e-06, "loss": 0.1421817421913147, "step": 24900 }, { "epoch": 0.6561673740463545, "grad_norm": 1.974251627922058, "learning_rate": 5.586424871356337e-06, "loss": 0.13171682357788086, "step": 24910 }, { "epoch": 0.6564307892908532, "grad_norm": 1.79145348072052, "learning_rate": 5.578771010397572e-06, "loss": 0.1361074924468994, "step": 24920 }, { "epoch": 0.656694204535352, "grad_norm": 1.7009466886520386, "learning_rate": 5.571120367400008e-06, "loss": 0.1402850031852722, "step": 24930 }, { "epoch": 0.6569576197798507, "grad_norm": 3.9037070274353027, "learning_rate": 5.563472947932123e-06, "loss": 0.13419132232666015, "step": 24940 }, { "epoch": 0.6572210350243495, "grad_norm": 2.175704002380371, "learning_rate": 5.555828757560035e-06, "loss": 0.13881711959838866, "step": 24950 }, { "epoch": 0.6574844502688482, "grad_norm": 1.451477289199829, "learning_rate": 5.548187801847513e-06, "loss": 0.13673672676086426, "step": 24960 }, { "epoch": 0.657747865513347, "grad_norm": 1.3721615076065063, "learning_rate": 5.5405500863559795e-06, "loss": 0.14293191432952881, "step": 24970 }, { "epoch": 0.6580112807578457, "grad_norm": 2.0115325450897217, "learning_rate": 5.532915616644484e-06, "loss": 0.13326632976531982, "step": 24980 }, { "epoch": 0.6582746960023445, "grad_norm": 1.6762009859085083, "learning_rate": 5.525284398269733e-06, "loss": 0.12303886413574219, "step": 24990 }, { "epoch": 0.6585381112468431, "grad_norm": 1.704554557800293, "learning_rate": 5.517656436786059e-06, "loss": 0.13142681121826172, "step": 25000 }, { "epoch": 0.6588015264913418, "grad_norm": 3.1544461250305176, "learning_rate": 5.510031737745405e-06, "loss": 0.13742295503616334, "step": 25010 }, { "epoch": 0.6590649417358406, "grad_norm": 1.792137861251831, "learning_rate": 5.50241030669737e-06, "loss": 0.1409471035003662, "step": 25020 }, { "epoch": 0.6593283569803393, "grad_norm": 2.0136210918426514, "learning_rate": 5.4947921491891565e-06, "loss": 0.1410142421722412, "step": 25030 }, { "epoch": 0.6595917722248381, "grad_norm": 1.4579253196716309, "learning_rate": 5.487177270765588e-06, "loss": 0.1338104248046875, "step": 25040 }, { "epoch": 0.6598551874693368, "grad_norm": 2.4094057083129883, "learning_rate": 5.479565676969101e-06, "loss": 0.13392772674560546, "step": 25050 }, { "epoch": 0.6601186027138356, "grad_norm": 1.184865951538086, "learning_rate": 5.47195737333974e-06, "loss": 0.1346058130264282, "step": 25060 }, { "epoch": 0.6603820179583343, "grad_norm": 0.8957417011260986, "learning_rate": 5.464352365415162e-06, "loss": 0.13239672183990478, "step": 25070 }, { "epoch": 0.660645433202833, "grad_norm": 1.6183232069015503, "learning_rate": 5.456750658730616e-06, "loss": 0.14466533660888672, "step": 25080 }, { "epoch": 0.6609088484473318, "grad_norm": 1.4534506797790527, "learning_rate": 5.449152258818955e-06, "loss": 0.13347634077072143, "step": 25090 }, { "epoch": 0.6611722636918305, "grad_norm": 3.081951379776001, "learning_rate": 5.4415571712106184e-06, "loss": 0.14378509521484376, "step": 25100 }, { "epoch": 0.6614356789363293, "grad_norm": 1.9530768394470215, "learning_rate": 5.433965401433639e-06, "loss": 0.12522320747375487, "step": 25110 }, { "epoch": 0.661699094180828, "grad_norm": 1.5312949419021606, "learning_rate": 5.426376955013641e-06, "loss": 0.13333799839019775, "step": 25120 }, { "epoch": 0.6619625094253268, "grad_norm": 2.7465338706970215, "learning_rate": 5.4187918374738154e-06, "loss": 0.14498212337493896, "step": 25130 }, { "epoch": 0.6622259246698254, "grad_norm": 1.09040105342865, "learning_rate": 5.4112100543349365e-06, "loss": 0.12879202365875245, "step": 25140 }, { "epoch": 0.6624893399143242, "grad_norm": 2.502664089202881, "learning_rate": 5.403631611115362e-06, "loss": 0.13873090744018554, "step": 25150 }, { "epoch": 0.6627527551588229, "grad_norm": 2.4419476985931396, "learning_rate": 5.396056513331005e-06, "loss": 0.13295091390609742, "step": 25160 }, { "epoch": 0.6630161704033216, "grad_norm": 2.1607024669647217, "learning_rate": 5.388484766495348e-06, "loss": 0.14452189207077026, "step": 25170 }, { "epoch": 0.6632795856478204, "grad_norm": 2.453864812850952, "learning_rate": 5.3809163761194364e-06, "loss": 0.1417858600616455, "step": 25180 }, { "epoch": 0.6635430008923191, "grad_norm": 1.116143822669983, "learning_rate": 5.373351347711865e-06, "loss": 0.11600279808044434, "step": 25190 }, { "epoch": 0.6638064161368179, "grad_norm": 2.438565731048584, "learning_rate": 5.365789686778797e-06, "loss": 0.14100605249404907, "step": 25200 }, { "epoch": 0.6640698313813166, "grad_norm": 2.1910176277160645, "learning_rate": 5.358231398823937e-06, "loss": 0.1262168765068054, "step": 25210 }, { "epoch": 0.6643332466258154, "grad_norm": 2.5838382244110107, "learning_rate": 5.350676489348521e-06, "loss": 0.13618384599685668, "step": 25220 }, { "epoch": 0.6645966618703141, "grad_norm": 1.7495651245117188, "learning_rate": 5.3431249638513475e-06, "loss": 0.15561046600341796, "step": 25230 }, { "epoch": 0.6648600771148129, "grad_norm": 1.7453478574752808, "learning_rate": 5.335576827828737e-06, "loss": 0.1363005518913269, "step": 25240 }, { "epoch": 0.6651234923593116, "grad_norm": 3.0727219581604004, "learning_rate": 5.328032086774557e-06, "loss": 0.12736417055130006, "step": 25250 }, { "epoch": 0.6653869076038103, "grad_norm": 2.8325092792510986, "learning_rate": 5.320490746180189e-06, "loss": 0.12670162916183472, "step": 25260 }, { "epoch": 0.665650322848309, "grad_norm": 2.667487144470215, "learning_rate": 5.312952811534542e-06, "loss": 0.1376153349876404, "step": 25270 }, { "epoch": 0.6659137380928077, "grad_norm": 2.004223346710205, "learning_rate": 5.305418288324058e-06, "loss": 0.13342679738998414, "step": 25280 }, { "epoch": 0.6661771533373065, "grad_norm": 2.227752923965454, "learning_rate": 5.297887182032686e-06, "loss": 0.1347031831741333, "step": 25290 }, { "epoch": 0.6664405685818052, "grad_norm": 2.7379419803619385, "learning_rate": 5.29035949814189e-06, "loss": 0.13059114217758178, "step": 25300 }, { "epoch": 0.666703983826304, "grad_norm": 2.2985384464263916, "learning_rate": 5.2828352421306415e-06, "loss": 0.13991379737854004, "step": 25310 }, { "epoch": 0.6669673990708027, "grad_norm": 2.145608425140381, "learning_rate": 5.275314419475417e-06, "loss": 0.12297461032867432, "step": 25320 }, { "epoch": 0.6672308143153015, "grad_norm": 3.146267890930176, "learning_rate": 5.267797035650202e-06, "loss": 0.12664068937301637, "step": 25330 }, { "epoch": 0.6674942295598002, "grad_norm": 1.5577589273452759, "learning_rate": 5.2602830961264704e-06, "loss": 0.13807107210159303, "step": 25340 }, { "epoch": 0.6677576448042989, "grad_norm": 2.5126445293426514, "learning_rate": 5.252772606373186e-06, "loss": 0.13416097164154053, "step": 25350 }, { "epoch": 0.6680210600487977, "grad_norm": 1.5852323770523071, "learning_rate": 5.245265571856812e-06, "loss": 0.11855721473693848, "step": 25360 }, { "epoch": 0.6682844752932964, "grad_norm": 2.426969051361084, "learning_rate": 5.237761998041291e-06, "loss": 0.1337523102760315, "step": 25370 }, { "epoch": 0.6685478905377952, "grad_norm": 1.912156105041504, "learning_rate": 5.230261890388053e-06, "loss": 0.1433022141456604, "step": 25380 }, { "epoch": 0.6688113057822939, "grad_norm": 1.4306714534759521, "learning_rate": 5.2227652543559924e-06, "loss": 0.13389546871185304, "step": 25390 }, { "epoch": 0.6690747210267927, "grad_norm": 2.015132188796997, "learning_rate": 5.215272095401483e-06, "loss": 0.13553969860076903, "step": 25400 }, { "epoch": 0.6693381362712914, "grad_norm": 1.7847398519515991, "learning_rate": 5.207782418978378e-06, "loss": 0.134661865234375, "step": 25410 }, { "epoch": 0.6696015515157901, "grad_norm": 1.7296595573425293, "learning_rate": 5.200296230537982e-06, "loss": 0.12925866842269898, "step": 25420 }, { "epoch": 0.6698649667602888, "grad_norm": 1.6886136531829834, "learning_rate": 5.192813535529066e-06, "loss": 0.1432201862335205, "step": 25430 }, { "epoch": 0.6701283820047875, "grad_norm": 1.340245008468628, "learning_rate": 5.1853343393978585e-06, "loss": 0.12907793521881103, "step": 25440 }, { "epoch": 0.6703917972492863, "grad_norm": 1.7430837154388428, "learning_rate": 5.177858647588037e-06, "loss": 0.13902363777160645, "step": 25450 }, { "epoch": 0.670655212493785, "grad_norm": 2.020220994949341, "learning_rate": 5.170386465540741e-06, "loss": 0.12684981822967528, "step": 25460 }, { "epoch": 0.6709186277382838, "grad_norm": 1.8540809154510498, "learning_rate": 5.162917798694546e-06, "loss": 0.14070849418640136, "step": 25470 }, { "epoch": 0.6711820429827825, "grad_norm": 1.92711341381073, "learning_rate": 5.15545265248546e-06, "loss": 0.13963496685028076, "step": 25480 }, { "epoch": 0.6714454582272813, "grad_norm": 2.232292413711548, "learning_rate": 5.147991032346948e-06, "loss": 0.12764986753463745, "step": 25490 }, { "epoch": 0.67170887347178, "grad_norm": 1.7543108463287354, "learning_rate": 5.140532943709893e-06, "loss": 0.1372525930404663, "step": 25500 }, { "epoch": 0.6719722887162788, "grad_norm": 2.007835865020752, "learning_rate": 5.133078392002623e-06, "loss": 0.13369332551956176, "step": 25510 }, { "epoch": 0.6722357039607775, "grad_norm": 1.379243016242981, "learning_rate": 5.125627382650873e-06, "loss": 0.13418697118759154, "step": 25520 }, { "epoch": 0.6724991192052762, "grad_norm": 3.8190205097198486, "learning_rate": 5.118179921077808e-06, "loss": 0.13990364074707032, "step": 25530 }, { "epoch": 0.672762534449775, "grad_norm": 1.8681241273880005, "learning_rate": 5.1107360127040206e-06, "loss": 0.13747386932373046, "step": 25540 }, { "epoch": 0.6730259496942737, "grad_norm": 2.1761868000030518, "learning_rate": 5.103295662947504e-06, "loss": 0.1334511637687683, "step": 25550 }, { "epoch": 0.6732893649387724, "grad_norm": 3.8584671020507812, "learning_rate": 5.095858877223665e-06, "loss": 0.12498798370361328, "step": 25560 }, { "epoch": 0.6735527801832711, "grad_norm": 1.528371810913086, "learning_rate": 5.088425660945319e-06, "loss": 0.13356505632400512, "step": 25570 }, { "epoch": 0.6738161954277699, "grad_norm": 1.4811004400253296, "learning_rate": 5.080996019522675e-06, "loss": 0.12033462524414062, "step": 25580 }, { "epoch": 0.6740796106722686, "grad_norm": 3.427171230316162, "learning_rate": 5.073569958363356e-06, "loss": 0.13599884510040283, "step": 25590 }, { "epoch": 0.6743430259167673, "grad_norm": 2.2240891456604004, "learning_rate": 5.06614748287237e-06, "loss": 0.12339534759521484, "step": 25600 }, { "epoch": 0.6746064411612661, "grad_norm": 2.0682716369628906, "learning_rate": 5.058728598452103e-06, "loss": 0.14314072132110595, "step": 25610 }, { "epoch": 0.6748698564057648, "grad_norm": 1.8825700283050537, "learning_rate": 5.05131331050235e-06, "loss": 0.12358329296112061, "step": 25620 }, { "epoch": 0.6751332716502636, "grad_norm": 1.6397619247436523, "learning_rate": 5.043901624420273e-06, "loss": 0.1457810640335083, "step": 25630 }, { "epoch": 0.6753966868947623, "grad_norm": 3.435025691986084, "learning_rate": 5.036493545600421e-06, "loss": 0.1242777943611145, "step": 25640 }, { "epoch": 0.6756601021392611, "grad_norm": 1.4267864227294922, "learning_rate": 5.029089079434708e-06, "loss": 0.12854623794555664, "step": 25650 }, { "epoch": 0.6759235173837598, "grad_norm": 2.3331551551818848, "learning_rate": 5.021688231312423e-06, "loss": 0.11921353340148926, "step": 25660 }, { "epoch": 0.6761869326282586, "grad_norm": 1.4622584581375122, "learning_rate": 5.014291006620229e-06, "loss": 0.13585360050201417, "step": 25670 }, { "epoch": 0.6764503478727573, "grad_norm": 1.292392611503601, "learning_rate": 5.006897410742142e-06, "loss": 0.12939805984497071, "step": 25680 }, { "epoch": 0.676713763117256, "grad_norm": 1.644866943359375, "learning_rate": 4.999507449059542e-06, "loss": 0.12658994197845458, "step": 25690 }, { "epoch": 0.6769771783617548, "grad_norm": 1.5335235595703125, "learning_rate": 4.992121126951159e-06, "loss": 0.14284777641296387, "step": 25700 }, { "epoch": 0.6772405936062534, "grad_norm": 1.9015681743621826, "learning_rate": 4.984738449793074e-06, "loss": 0.13484224081039428, "step": 25710 }, { "epoch": 0.6775040088507522, "grad_norm": 2.0798442363739014, "learning_rate": 4.977359422958731e-06, "loss": 0.13627951145172118, "step": 25720 }, { "epoch": 0.6777674240952509, "grad_norm": 1.5397039651870728, "learning_rate": 4.96998405181889e-06, "loss": 0.14501953125, "step": 25730 }, { "epoch": 0.6780308393397497, "grad_norm": 1.410347580909729, "learning_rate": 4.962612341741666e-06, "loss": 0.12313547134399414, "step": 25740 }, { "epoch": 0.6782942545842484, "grad_norm": 1.648043155670166, "learning_rate": 4.955244298092514e-06, "loss": 0.13202767372131347, "step": 25750 }, { "epoch": 0.6785576698287472, "grad_norm": 1.4670875072479248, "learning_rate": 4.947879926234212e-06, "loss": 0.13951926231384276, "step": 25760 }, { "epoch": 0.6788210850732459, "grad_norm": 2.6973114013671875, "learning_rate": 4.940519231526867e-06, "loss": 0.13046882152557374, "step": 25770 }, { "epoch": 0.6790845003177446, "grad_norm": 1.5323296785354614, "learning_rate": 4.933162219327909e-06, "loss": 0.11979756355285645, "step": 25780 }, { "epoch": 0.6793479155622434, "grad_norm": 1.831184983253479, "learning_rate": 4.925808894992087e-06, "loss": 0.12588067054748536, "step": 25790 }, { "epoch": 0.6796113308067421, "grad_norm": 1.7759428024291992, "learning_rate": 4.918459263871474e-06, "loss": 0.1304501175880432, "step": 25800 }, { "epoch": 0.6798747460512409, "grad_norm": 2.2679636478424072, "learning_rate": 4.911113331315448e-06, "loss": 0.11866483688354493, "step": 25810 }, { "epoch": 0.6801381612957396, "grad_norm": 3.208303689956665, "learning_rate": 4.903771102670686e-06, "loss": 0.1281005859375, "step": 25820 }, { "epoch": 0.6804015765402384, "grad_norm": 1.5508756637573242, "learning_rate": 4.896432583281189e-06, "loss": 0.12656838893890382, "step": 25830 }, { "epoch": 0.680664991784737, "grad_norm": 1.9717658758163452, "learning_rate": 4.88909777848824e-06, "loss": 0.1316227912902832, "step": 25840 }, { "epoch": 0.6809284070292358, "grad_norm": 2.5472824573516846, "learning_rate": 4.881766693630439e-06, "loss": 0.14081323146820068, "step": 25850 }, { "epoch": 0.6811918222737345, "grad_norm": 2.7943930625915527, "learning_rate": 4.874439334043654e-06, "loss": 0.12331626415252686, "step": 25860 }, { "epoch": 0.6814552375182332, "grad_norm": 1.5475043058395386, "learning_rate": 4.8671157050610525e-06, "loss": 0.13238818645477296, "step": 25870 }, { "epoch": 0.681718652762732, "grad_norm": 2.6085665225982666, "learning_rate": 4.859795812013099e-06, "loss": 0.1448438882827759, "step": 25880 }, { "epoch": 0.6819820680072307, "grad_norm": 2.530513048171997, "learning_rate": 4.852479660227518e-06, "loss": 0.12993661165237427, "step": 25890 }, { "epoch": 0.6822454832517295, "grad_norm": 2.8745739459991455, "learning_rate": 4.8451672550293236e-06, "loss": 0.11822075843811035, "step": 25900 }, { "epoch": 0.6825088984962282, "grad_norm": 2.5601987838745117, "learning_rate": 4.837858601740799e-06, "loss": 0.13011293411254882, "step": 25910 }, { "epoch": 0.682772313740727, "grad_norm": 2.4086649417877197, "learning_rate": 4.830553705681496e-06, "loss": 0.13313424587249756, "step": 25920 }, { "epoch": 0.6830357289852257, "grad_norm": 2.0409653186798096, "learning_rate": 4.823252572168236e-06, "loss": 0.13079447746276857, "step": 25930 }, { "epoch": 0.6832991442297245, "grad_norm": 1.1552385091781616, "learning_rate": 4.815955206515104e-06, "loss": 0.12838344573974608, "step": 25940 }, { "epoch": 0.6835625594742232, "grad_norm": 3.5262248516082764, "learning_rate": 4.808661614033422e-06, "loss": 0.131368088722229, "step": 25950 }, { "epoch": 0.6838259747187219, "grad_norm": 1.4061897993087769, "learning_rate": 4.801371800031796e-06, "loss": 0.12986191511154174, "step": 25960 }, { "epoch": 0.6840893899632207, "grad_norm": 1.8097083568572998, "learning_rate": 4.794085769816057e-06, "loss": 0.11714441776275634, "step": 25970 }, { "epoch": 0.6843528052077194, "grad_norm": 1.7405327558517456, "learning_rate": 4.786803528689304e-06, "loss": 0.14184169769287108, "step": 25980 }, { "epoch": 0.6846162204522181, "grad_norm": 4.278316974639893, "learning_rate": 4.779525081951857e-06, "loss": 0.14256048202514648, "step": 25990 }, { "epoch": 0.6848796356967168, "grad_norm": 1.9799456596374512, "learning_rate": 4.772250434901281e-06, "loss": 0.12441730499267578, "step": 26000 }, { "epoch": 0.6851430509412156, "grad_norm": 1.8735640048980713, "learning_rate": 4.7649795928323875e-06, "loss": 0.11471614837646485, "step": 26010 }, { "epoch": 0.6854064661857143, "grad_norm": 1.6730196475982666, "learning_rate": 4.757712561037203e-06, "loss": 0.13143041133880615, "step": 26020 }, { "epoch": 0.6856698814302131, "grad_norm": 2.918426036834717, "learning_rate": 4.750449344804989e-06, "loss": 0.12211046218872071, "step": 26030 }, { "epoch": 0.6859332966747118, "grad_norm": 4.7307939529418945, "learning_rate": 4.743189949422227e-06, "loss": 0.14706308841705323, "step": 26040 }, { "epoch": 0.6861967119192105, "grad_norm": 1.9632047414779663, "learning_rate": 4.735934380172613e-06, "loss": 0.12199954986572266, "step": 26050 }, { "epoch": 0.6864601271637093, "grad_norm": 3.5249509811401367, "learning_rate": 4.728682642337072e-06, "loss": 0.13122849464416503, "step": 26060 }, { "epoch": 0.686723542408208, "grad_norm": 1.7517142295837402, "learning_rate": 4.7214347411937325e-06, "loss": 0.1373657464981079, "step": 26070 }, { "epoch": 0.6869869576527068, "grad_norm": 1.4454935789108276, "learning_rate": 4.714190682017918e-06, "loss": 0.13230059146881104, "step": 26080 }, { "epoch": 0.6872503728972055, "grad_norm": 1.6555095911026, "learning_rate": 4.706950470082177e-06, "loss": 0.12991571426391602, "step": 26090 }, { "epoch": 0.6875137881417043, "grad_norm": 2.4000182151794434, "learning_rate": 4.699714110656242e-06, "loss": 0.12058683633804321, "step": 26100 }, { "epoch": 0.687777203386203, "grad_norm": 1.6102405786514282, "learning_rate": 4.6924816090070575e-06, "loss": 0.13224514722824096, "step": 26110 }, { "epoch": 0.6880406186307017, "grad_norm": 2.508474111557007, "learning_rate": 4.685252970398742e-06, "loss": 0.13696095943450928, "step": 26120 }, { "epoch": 0.6883040338752004, "grad_norm": 1.6621441841125488, "learning_rate": 4.678028200092607e-06, "loss": 0.13980429172515868, "step": 26130 }, { "epoch": 0.6885674491196991, "grad_norm": 1.292726993560791, "learning_rate": 4.670807303347161e-06, "loss": 0.125760018825531, "step": 26140 }, { "epoch": 0.6888308643641979, "grad_norm": 3.6824381351470947, "learning_rate": 4.663590285418082e-06, "loss": 0.1288179874420166, "step": 26150 }, { "epoch": 0.6890942796086966, "grad_norm": 5.061931133270264, "learning_rate": 4.656377151558224e-06, "loss": 0.11919209957122803, "step": 26160 }, { "epoch": 0.6893576948531954, "grad_norm": 1.7242554426193237, "learning_rate": 4.64916790701762e-06, "loss": 0.14516544342041016, "step": 26170 }, { "epoch": 0.6896211100976941, "grad_norm": 3.059019088745117, "learning_rate": 4.641962557043462e-06, "loss": 0.12060695886611938, "step": 26180 }, { "epoch": 0.6898845253421929, "grad_norm": 2.5267655849456787, "learning_rate": 4.634761106880127e-06, "loss": 0.12690563201904298, "step": 26190 }, { "epoch": 0.6901479405866916, "grad_norm": 1.9166629314422607, "learning_rate": 4.627563561769136e-06, "loss": 0.13219319581985473, "step": 26200 }, { "epoch": 0.6904113558311903, "grad_norm": 1.7112702131271362, "learning_rate": 4.620369926949164e-06, "loss": 0.13577332496643066, "step": 26210 }, { "epoch": 0.6906747710756891, "grad_norm": 1.663558006286621, "learning_rate": 4.613180207656061e-06, "loss": 0.11519391536712646, "step": 26220 }, { "epoch": 0.6909381863201878, "grad_norm": 1.5115883350372314, "learning_rate": 4.60599440912281e-06, "loss": 0.1301860809326172, "step": 26230 }, { "epoch": 0.6912016015646866, "grad_norm": 3.0607972145080566, "learning_rate": 4.598812536579545e-06, "loss": 0.12881579399108886, "step": 26240 }, { "epoch": 0.6914650168091853, "grad_norm": 1.6752123832702637, "learning_rate": 4.591634595253541e-06, "loss": 0.13154296875, "step": 26250 }, { "epoch": 0.6917284320536841, "grad_norm": 1.8187639713287354, "learning_rate": 4.584460590369214e-06, "loss": 0.13328657150268555, "step": 26260 }, { "epoch": 0.6919918472981827, "grad_norm": 2.1449263095855713, "learning_rate": 4.577290527148118e-06, "loss": 0.12391645908355713, "step": 26270 }, { "epoch": 0.6922552625426815, "grad_norm": 2.3018100261688232, "learning_rate": 4.5701244108089325e-06, "loss": 0.13458837270736695, "step": 26280 }, { "epoch": 0.6925186777871802, "grad_norm": 1.756182074546814, "learning_rate": 4.562962246567466e-06, "loss": 0.13622363805770873, "step": 26290 }, { "epoch": 0.6927820930316789, "grad_norm": 4.184998035430908, "learning_rate": 4.555804039636649e-06, "loss": 0.13083289861679076, "step": 26300 }, { "epoch": 0.6930455082761777, "grad_norm": 5.8816070556640625, "learning_rate": 4.548649795226532e-06, "loss": 0.12540223598480224, "step": 26310 }, { "epoch": 0.6933089235206764, "grad_norm": 1.621597409248352, "learning_rate": 4.541499518544292e-06, "loss": 0.12802958488464355, "step": 26320 }, { "epoch": 0.6935723387651752, "grad_norm": 3.4191293716430664, "learning_rate": 4.5343532147941985e-06, "loss": 0.12953453063964843, "step": 26330 }, { "epoch": 0.6938357540096739, "grad_norm": 2.2042460441589355, "learning_rate": 4.527210889177639e-06, "loss": 0.14789261817932128, "step": 26340 }, { "epoch": 0.6940991692541727, "grad_norm": 4.207907676696777, "learning_rate": 4.520072546893115e-06, "loss": 0.13552845716476442, "step": 26350 }, { "epoch": 0.6943625844986714, "grad_norm": 2.1865248680114746, "learning_rate": 4.512938193136216e-06, "loss": 0.12706663608551025, "step": 26360 }, { "epoch": 0.6946259997431702, "grad_norm": 1.1873949766159058, "learning_rate": 4.5058078330996315e-06, "loss": 0.12218828201293945, "step": 26370 }, { "epoch": 0.6948894149876689, "grad_norm": 2.375988006591797, "learning_rate": 4.498681471973145e-06, "loss": 0.12523798942565917, "step": 26380 }, { "epoch": 0.6951528302321676, "grad_norm": 2.409468650817871, "learning_rate": 4.491559114943628e-06, "loss": 0.12572288513183594, "step": 26390 }, { "epoch": 0.6954162454766664, "grad_norm": 2.330598831176758, "learning_rate": 4.484440767195044e-06, "loss": 0.1447676420211792, "step": 26400 }, { "epoch": 0.695679660721165, "grad_norm": 2.071511745452881, "learning_rate": 4.477326433908435e-06, "loss": 0.12920405864715576, "step": 26410 }, { "epoch": 0.6959430759656638, "grad_norm": 1.1866559982299805, "learning_rate": 4.4702161202619085e-06, "loss": 0.12888896465301514, "step": 26420 }, { "epoch": 0.6962064912101625, "grad_norm": 2.0528931617736816, "learning_rate": 4.463109831430668e-06, "loss": 0.12598778009414674, "step": 26430 }, { "epoch": 0.6964699064546613, "grad_norm": 1.8965119123458862, "learning_rate": 4.456007572586971e-06, "loss": 0.12817705869674684, "step": 26440 }, { "epoch": 0.69673332169916, "grad_norm": 2.534158706665039, "learning_rate": 4.448909348900159e-06, "loss": 0.12854275703430176, "step": 26450 }, { "epoch": 0.6969967369436588, "grad_norm": 2.0147171020507812, "learning_rate": 4.441815165536612e-06, "loss": 0.14320147037506104, "step": 26460 }, { "epoch": 0.6972601521881575, "grad_norm": 1.479232668876648, "learning_rate": 4.434725027659786e-06, "loss": 0.12688775062561036, "step": 26470 }, { "epoch": 0.6975235674326562, "grad_norm": 1.8705233335494995, "learning_rate": 4.427638940430192e-06, "loss": 0.12779202461242675, "step": 26480 }, { "epoch": 0.697786982677155, "grad_norm": 2.571664571762085, "learning_rate": 4.420556909005392e-06, "loss": 0.1310532808303833, "step": 26490 }, { "epoch": 0.6980503979216537, "grad_norm": 3.106891393661499, "learning_rate": 4.41347893853999e-06, "loss": 0.13648160696029663, "step": 26500 }, { "epoch": 0.6983138131661525, "grad_norm": 3.716116428375244, "learning_rate": 4.40640503418564e-06, "loss": 0.12361481189727783, "step": 26510 }, { "epoch": 0.6985772284106512, "grad_norm": 1.7285288572311401, "learning_rate": 4.39933520109103e-06, "loss": 0.12104291915893554, "step": 26520 }, { "epoch": 0.69884064365515, "grad_norm": 3.4231648445129395, "learning_rate": 4.3922694444018985e-06, "loss": 0.12028640508651733, "step": 26530 }, { "epoch": 0.6991040588996487, "grad_norm": 2.6737804412841797, "learning_rate": 4.385207769261007e-06, "loss": 0.11552554368972778, "step": 26540 }, { "epoch": 0.6993674741441475, "grad_norm": 3.9655823707580566, "learning_rate": 4.378150180808136e-06, "loss": 0.13276180028915405, "step": 26550 }, { "epoch": 0.6996308893886461, "grad_norm": 2.1666948795318604, "learning_rate": 4.371096684180114e-06, "loss": 0.12559547424316406, "step": 26560 }, { "epoch": 0.6998943046331448, "grad_norm": 1.6365089416503906, "learning_rate": 4.364047284510774e-06, "loss": 0.10413579940795899, "step": 26570 }, { "epoch": 0.7001577198776436, "grad_norm": 2.508159875869751, "learning_rate": 4.357001986930982e-06, "loss": 0.12345263957977295, "step": 26580 }, { "epoch": 0.7004211351221423, "grad_norm": 3.7619099617004395, "learning_rate": 4.349960796568601e-06, "loss": 0.13068017959594727, "step": 26590 }, { "epoch": 0.7006845503666411, "grad_norm": 1.6679869890213013, "learning_rate": 4.342923718548512e-06, "loss": 0.12330443859100342, "step": 26600 }, { "epoch": 0.7009479656111398, "grad_norm": 1.250872254371643, "learning_rate": 4.335890757992611e-06, "loss": 0.12203066349029541, "step": 26610 }, { "epoch": 0.7012113808556386, "grad_norm": 1.796110987663269, "learning_rate": 4.328861920019787e-06, "loss": 0.12370085716247559, "step": 26620 }, { "epoch": 0.7014747961001373, "grad_norm": 3.433562994003296, "learning_rate": 4.321837209745931e-06, "loss": 0.12490694522857666, "step": 26630 }, { "epoch": 0.701738211344636, "grad_norm": 1.5566234588623047, "learning_rate": 4.3148166322839305e-06, "loss": 0.12991682291030884, "step": 26640 }, { "epoch": 0.7020016265891348, "grad_norm": 2.932202100753784, "learning_rate": 4.3078001927436605e-06, "loss": 0.1216015338897705, "step": 26650 }, { "epoch": 0.7022650418336335, "grad_norm": 2.0464072227478027, "learning_rate": 4.300787896231997e-06, "loss": 0.12126473188400269, "step": 26660 }, { "epoch": 0.7025284570781323, "grad_norm": 3.5245115756988525, "learning_rate": 4.293779747852791e-06, "loss": 0.1292950391769409, "step": 26670 }, { "epoch": 0.702791872322631, "grad_norm": 3.561640739440918, "learning_rate": 4.286775752706865e-06, "loss": 0.12414448261260987, "step": 26680 }, { "epoch": 0.7030552875671298, "grad_norm": 3.0330655574798584, "learning_rate": 4.279775915892038e-06, "loss": 0.1349464774131775, "step": 26690 }, { "epoch": 0.7033187028116284, "grad_norm": 1.5112367868423462, "learning_rate": 4.2727802425030885e-06, "loss": 0.11555520296096802, "step": 26700 }, { "epoch": 0.7035821180561272, "grad_norm": 1.7241195440292358, "learning_rate": 4.265788737631779e-06, "loss": 0.1163262128829956, "step": 26710 }, { "epoch": 0.7038455333006259, "grad_norm": 1.7383612394332886, "learning_rate": 4.258801406366817e-06, "loss": 0.1218579888343811, "step": 26720 }, { "epoch": 0.7041089485451246, "grad_norm": 3.2516472339630127, "learning_rate": 4.251818253793885e-06, "loss": 0.1362799048423767, "step": 26730 }, { "epoch": 0.7043723637896234, "grad_norm": 2.462310314178467, "learning_rate": 4.244839284995627e-06, "loss": 0.13013435602188111, "step": 26740 }, { "epoch": 0.7046357790341221, "grad_norm": 2.6032447814941406, "learning_rate": 4.237864505051638e-06, "loss": 0.13634960651397704, "step": 26750 }, { "epoch": 0.7048991942786209, "grad_norm": 2.114231586456299, "learning_rate": 4.230893919038461e-06, "loss": 0.13489736318588258, "step": 26760 }, { "epoch": 0.7051626095231196, "grad_norm": 2.3653533458709717, "learning_rate": 4.223927532029592e-06, "loss": 0.12054059505462647, "step": 26770 }, { "epoch": 0.7054260247676184, "grad_norm": 2.0806503295898438, "learning_rate": 4.21696534909546e-06, "loss": 0.1372678518295288, "step": 26780 }, { "epoch": 0.7056894400121171, "grad_norm": 2.0510175228118896, "learning_rate": 4.2100073753034535e-06, "loss": 0.12735724449157715, "step": 26790 }, { "epoch": 0.7059528552566159, "grad_norm": 2.488318681716919, "learning_rate": 4.203053615717884e-06, "loss": 0.12763352394104005, "step": 26800 }, { "epoch": 0.7062162705011146, "grad_norm": 1.9284024238586426, "learning_rate": 4.196104075399986e-06, "loss": 0.14094668626785278, "step": 26810 }, { "epoch": 0.7064796857456133, "grad_norm": 1.5897138118743896, "learning_rate": 4.1891587594079485e-06, "loss": 0.13808525800704957, "step": 26820 }, { "epoch": 0.7067431009901121, "grad_norm": 1.997230887413025, "learning_rate": 4.182217672796867e-06, "loss": 0.1261540174484253, "step": 26830 }, { "epoch": 0.7070065162346107, "grad_norm": 1.7482988834381104, "learning_rate": 4.175280820618763e-06, "loss": 0.12237517833709717, "step": 26840 }, { "epoch": 0.7072699314791095, "grad_norm": 1.899963617324829, "learning_rate": 4.168348207922577e-06, "loss": 0.11527303457260132, "step": 26850 }, { "epoch": 0.7075333467236082, "grad_norm": 2.610476493835449, "learning_rate": 4.16141983975416e-06, "loss": 0.12420682907104492, "step": 26860 }, { "epoch": 0.707796761968107, "grad_norm": 3.9612720012664795, "learning_rate": 4.154495721156287e-06, "loss": 0.13474855422973633, "step": 26870 }, { "epoch": 0.7080601772126057, "grad_norm": 2.754361152648926, "learning_rate": 4.147575857168623e-06, "loss": 0.12857463359832763, "step": 26880 }, { "epoch": 0.7083235924571045, "grad_norm": 1.4808933734893799, "learning_rate": 4.140660252827746e-06, "loss": 0.1154819369316101, "step": 26890 }, { "epoch": 0.7085870077016032, "grad_norm": 2.6180977821350098, "learning_rate": 4.13374891316713e-06, "loss": 0.123085618019104, "step": 26900 }, { "epoch": 0.7088504229461019, "grad_norm": 1.828384518623352, "learning_rate": 4.126841843217144e-06, "loss": 0.12290180921554565, "step": 26910 }, { "epoch": 0.7091138381906007, "grad_norm": 2.3448164463043213, "learning_rate": 4.1199390480050605e-06, "loss": 0.12349241971969604, "step": 26920 }, { "epoch": 0.7093772534350994, "grad_norm": 1.768783688545227, "learning_rate": 4.113040532555021e-06, "loss": 0.12463610172271729, "step": 26930 }, { "epoch": 0.7096406686795982, "grad_norm": 2.6648662090301514, "learning_rate": 4.106146301888065e-06, "loss": 0.13218207359313966, "step": 26940 }, { "epoch": 0.7099040839240969, "grad_norm": 1.8446124792099, "learning_rate": 4.0992563610221155e-06, "loss": 0.1171257734298706, "step": 26950 }, { "epoch": 0.7101674991685957, "grad_norm": 1.908250331878662, "learning_rate": 4.092370714971966e-06, "loss": 0.12040934562683106, "step": 26960 }, { "epoch": 0.7104309144130944, "grad_norm": 1.9877536296844482, "learning_rate": 4.085489368749287e-06, "loss": 0.12984558343887329, "step": 26970 }, { "epoch": 0.7106943296575932, "grad_norm": 2.1418726444244385, "learning_rate": 4.07861232736262e-06, "loss": 0.1308051824569702, "step": 26980 }, { "epoch": 0.7109577449020918, "grad_norm": 3.3621511459350586, "learning_rate": 4.0717395958173665e-06, "loss": 0.12623634338378906, "step": 26990 }, { "epoch": 0.7112211601465905, "grad_norm": 4.0524773597717285, "learning_rate": 4.064871179115807e-06, "loss": 0.1298891305923462, "step": 27000 }, { "epoch": 0.7114845753910893, "grad_norm": 2.3543038368225098, "learning_rate": 4.0580070822570694e-06, "loss": 0.12242857217788697, "step": 27010 }, { "epoch": 0.711747990635588, "grad_norm": 1.5567960739135742, "learning_rate": 4.051147310237129e-06, "loss": 0.12595291137695314, "step": 27020 }, { "epoch": 0.7120114058800868, "grad_norm": 1.9177881479263306, "learning_rate": 4.044291868048837e-06, "loss": 0.12079306840896606, "step": 27030 }, { "epoch": 0.7122748211245855, "grad_norm": 2.6661500930786133, "learning_rate": 4.0374407606818684e-06, "loss": 0.1314191460609436, "step": 27040 }, { "epoch": 0.7125382363690843, "grad_norm": 3.113430976867676, "learning_rate": 4.03059399312277e-06, "loss": 0.1481895327568054, "step": 27050 }, { "epoch": 0.712801651613583, "grad_norm": 2.6026899814605713, "learning_rate": 4.0237515703549015e-06, "loss": 0.12350316047668457, "step": 27060 }, { "epoch": 0.7130650668580818, "grad_norm": 1.4333323240280151, "learning_rate": 4.0169134973584726e-06, "loss": 0.10749036073684692, "step": 27070 }, { "epoch": 0.7133284821025805, "grad_norm": 2.3970301151275635, "learning_rate": 4.010079779110537e-06, "loss": 0.1289878249168396, "step": 27080 }, { "epoch": 0.7135918973470792, "grad_norm": 1.7220412492752075, "learning_rate": 4.003250420584966e-06, "loss": 0.13287477493286132, "step": 27090 }, { "epoch": 0.713855312591578, "grad_norm": 2.897998809814453, "learning_rate": 3.996425426752459e-06, "loss": 0.13208507299423217, "step": 27100 }, { "epoch": 0.7141187278360767, "grad_norm": 2.5319886207580566, "learning_rate": 3.989604802580545e-06, "loss": 0.13421577215194702, "step": 27110 }, { "epoch": 0.7143821430805755, "grad_norm": 1.6109377145767212, "learning_rate": 3.982788553033561e-06, "loss": 0.12183690071105957, "step": 27120 }, { "epoch": 0.7146455583250741, "grad_norm": 1.6199915409088135, "learning_rate": 3.975976683072676e-06, "loss": 0.12956141233444213, "step": 27130 }, { "epoch": 0.714908973569573, "grad_norm": 1.7350647449493408, "learning_rate": 3.9691691976558636e-06, "loss": 0.13756474256515502, "step": 27140 }, { "epoch": 0.7151723888140716, "grad_norm": 1.8590775728225708, "learning_rate": 3.9623661017378946e-06, "loss": 0.1252762794494629, "step": 27150 }, { "epoch": 0.7154358040585704, "grad_norm": 2.0839319229125977, "learning_rate": 3.955567400270365e-06, "loss": 0.12641396522521972, "step": 27160 }, { "epoch": 0.7156992193030691, "grad_norm": 2.0178394317626953, "learning_rate": 3.948773098201657e-06, "loss": 0.1191185712814331, "step": 27170 }, { "epoch": 0.7159626345475678, "grad_norm": 3.9553658962249756, "learning_rate": 3.941983200476969e-06, "loss": 0.12997546195983886, "step": 27180 }, { "epoch": 0.7162260497920666, "grad_norm": 2.48163104057312, "learning_rate": 3.935197712038267e-06, "loss": 0.11872448921203613, "step": 27190 }, { "epoch": 0.7164894650365653, "grad_norm": 1.5602996349334717, "learning_rate": 3.928416637824326e-06, "loss": 0.12093286514282227, "step": 27200 }, { "epoch": 0.7167528802810641, "grad_norm": 2.0945427417755127, "learning_rate": 3.921639982770709e-06, "loss": 0.1272021770477295, "step": 27210 }, { "epoch": 0.7170162955255628, "grad_norm": 1.823068618774414, "learning_rate": 3.914867751809756e-06, "loss": 0.13217601776123047, "step": 27220 }, { "epoch": 0.7172797107700616, "grad_norm": 2.6504032611846924, "learning_rate": 3.908099949870588e-06, "loss": 0.13289804458618165, "step": 27230 }, { "epoch": 0.7175431260145603, "grad_norm": 2.7193310260772705, "learning_rate": 3.901336581879102e-06, "loss": 0.11899769306182861, "step": 27240 }, { "epoch": 0.717806541259059, "grad_norm": 2.184269666671753, "learning_rate": 3.894577652757968e-06, "loss": 0.13173773288726806, "step": 27250 }, { "epoch": 0.7180699565035578, "grad_norm": 1.871490478515625, "learning_rate": 3.887823167426633e-06, "loss": 0.12887003421783447, "step": 27260 }, { "epoch": 0.7183333717480564, "grad_norm": 1.4235807657241821, "learning_rate": 3.8810731308013005e-06, "loss": 0.121368145942688, "step": 27270 }, { "epoch": 0.7185967869925552, "grad_norm": 1.8367677927017212, "learning_rate": 3.874327547794931e-06, "loss": 0.12137656211853028, "step": 27280 }, { "epoch": 0.7188602022370539, "grad_norm": 2.378333568572998, "learning_rate": 3.8675864233172575e-06, "loss": 0.12463327646255493, "step": 27290 }, { "epoch": 0.7191236174815527, "grad_norm": 1.5338904857635498, "learning_rate": 3.860849762274759e-06, "loss": 0.11712589263916015, "step": 27300 }, { "epoch": 0.7193870327260514, "grad_norm": 2.4341135025024414, "learning_rate": 3.854117569570674e-06, "loss": 0.1278071880340576, "step": 27310 }, { "epoch": 0.7196504479705502, "grad_norm": 1.7199453115463257, "learning_rate": 3.8473898501049745e-06, "loss": 0.1279874324798584, "step": 27320 }, { "epoch": 0.7199138632150489, "grad_norm": 1.9621164798736572, "learning_rate": 3.840666608774385e-06, "loss": 0.1159110426902771, "step": 27330 }, { "epoch": 0.7201772784595476, "grad_norm": 2.5985753536224365, "learning_rate": 3.833947850472378e-06, "loss": 0.12104796171188355, "step": 27340 }, { "epoch": 0.7204406937040464, "grad_norm": 3.165576219558716, "learning_rate": 3.827233580089152e-06, "loss": 0.139307701587677, "step": 27350 }, { "epoch": 0.7207041089485451, "grad_norm": 2.917698860168457, "learning_rate": 3.82052380251164e-06, "loss": 0.1260754346847534, "step": 27360 }, { "epoch": 0.7209675241930439, "grad_norm": 2.098769187927246, "learning_rate": 3.8138185226235114e-06, "loss": 0.11844885349273682, "step": 27370 }, { "epoch": 0.7212309394375426, "grad_norm": 2.7083306312561035, "learning_rate": 3.8071177453051523e-06, "loss": 0.13595292568206788, "step": 27380 }, { "epoch": 0.7214943546820414, "grad_norm": 3.3041718006134033, "learning_rate": 3.8004214754336853e-06, "loss": 0.12756446599960328, "step": 27390 }, { "epoch": 0.7217577699265401, "grad_norm": 1.9615823030471802, "learning_rate": 3.7937297178829458e-06, "loss": 0.11078341007232666, "step": 27400 }, { "epoch": 0.7220211851710389, "grad_norm": 2.251208782196045, "learning_rate": 3.7870424775234703e-06, "loss": 0.1274811863899231, "step": 27410 }, { "epoch": 0.7222846004155375, "grad_norm": 1.3139832019805908, "learning_rate": 3.780359759222534e-06, "loss": 0.13380268812179566, "step": 27420 }, { "epoch": 0.7225480156600362, "grad_norm": 1.739539384841919, "learning_rate": 3.773681567844103e-06, "loss": 0.12030198574066162, "step": 27430 }, { "epoch": 0.722811430904535, "grad_norm": 2.033121109008789, "learning_rate": 3.767007908248852e-06, "loss": 0.12691547870635986, "step": 27440 }, { "epoch": 0.7230748461490337, "grad_norm": 2.9237091541290283, "learning_rate": 3.7603387852941608e-06, "loss": 0.12521060705184936, "step": 27450 }, { "epoch": 0.7233382613935325, "grad_norm": 2.905122756958008, "learning_rate": 3.753674203834099e-06, "loss": 0.11795032024383545, "step": 27460 }, { "epoch": 0.7236016766380312, "grad_norm": 2.7583088874816895, "learning_rate": 3.747014168719445e-06, "loss": 0.11972950696945191, "step": 27470 }, { "epoch": 0.72386509188253, "grad_norm": 3.4112935066223145, "learning_rate": 3.740358684797656e-06, "loss": 0.12818758487701415, "step": 27480 }, { "epoch": 0.7241285071270287, "grad_norm": 2.128758430480957, "learning_rate": 3.7337077569128797e-06, "loss": 0.12172324657440185, "step": 27490 }, { "epoch": 0.7243919223715275, "grad_norm": 3.595384120941162, "learning_rate": 3.7270613899059495e-06, "loss": 0.10602176189422607, "step": 27500 }, { "epoch": 0.7246553376160262, "grad_norm": 2.068352222442627, "learning_rate": 3.7204195886143736e-06, "loss": 0.10839182138442993, "step": 27510 }, { "epoch": 0.7249187528605249, "grad_norm": 2.2644553184509277, "learning_rate": 3.713782357872352e-06, "loss": 0.1175281286239624, "step": 27520 }, { "epoch": 0.7251821681050237, "grad_norm": 2.3819308280944824, "learning_rate": 3.7071497025107373e-06, "loss": 0.12796974182128906, "step": 27530 }, { "epoch": 0.7254455833495224, "grad_norm": 1.8756500482559204, "learning_rate": 3.7005216273570623e-06, "loss": 0.1114802360534668, "step": 27540 }, { "epoch": 0.7257089985940212, "grad_norm": 2.4373981952667236, "learning_rate": 3.693898137235532e-06, "loss": 0.11612585783004761, "step": 27550 }, { "epoch": 0.7259724138385198, "grad_norm": 2.2018604278564453, "learning_rate": 3.6872792369670074e-06, "loss": 0.11932381391525268, "step": 27560 }, { "epoch": 0.7262358290830186, "grad_norm": 1.5882081985473633, "learning_rate": 3.6806649313690066e-06, "loss": 0.12196342945098877, "step": 27570 }, { "epoch": 0.7264992443275173, "grad_norm": 1.6762802600860596, "learning_rate": 3.6740552252557093e-06, "loss": 0.13056737184524536, "step": 27580 }, { "epoch": 0.7267626595720161, "grad_norm": 1.835031270980835, "learning_rate": 3.667450123437939e-06, "loss": 0.1252953052520752, "step": 27590 }, { "epoch": 0.7270260748165148, "grad_norm": 1.9258373975753784, "learning_rate": 3.6608496307231823e-06, "loss": 0.12069153785705566, "step": 27600 }, { "epoch": 0.7272894900610135, "grad_norm": 1.7411946058273315, "learning_rate": 3.6542537519155617e-06, "loss": 0.10250592231750488, "step": 27610 }, { "epoch": 0.7275529053055123, "grad_norm": 2.378275156021118, "learning_rate": 3.647662491815833e-06, "loss": 0.12335205078125, "step": 27620 }, { "epoch": 0.727816320550011, "grad_norm": 1.9625660181045532, "learning_rate": 3.641075855221409e-06, "loss": 0.13631432056427, "step": 27630 }, { "epoch": 0.7280797357945098, "grad_norm": 1.7217278480529785, "learning_rate": 3.634493846926321e-06, "loss": 0.12540068626403808, "step": 27640 }, { "epoch": 0.7283431510390085, "grad_norm": 2.8468289375305176, "learning_rate": 3.62791647172125e-06, "loss": 0.13165031671524047, "step": 27650 }, { "epoch": 0.7286065662835073, "grad_norm": 3.202653646469116, "learning_rate": 3.621343734393482e-06, "loss": 0.12333580255508422, "step": 27660 }, { "epoch": 0.728869981528006, "grad_norm": 1.6785054206848145, "learning_rate": 3.6147756397269397e-06, "loss": 0.12770543098449708, "step": 27670 }, { "epoch": 0.7291333967725048, "grad_norm": 2.613853693008423, "learning_rate": 3.6082121925021717e-06, "loss": 0.1221621036529541, "step": 27680 }, { "epoch": 0.7293968120170035, "grad_norm": 1.9359465837478638, "learning_rate": 3.601653397496335e-06, "loss": 0.11619105339050292, "step": 27690 }, { "epoch": 0.7296602272615021, "grad_norm": 2.1502315998077393, "learning_rate": 3.5950992594832044e-06, "loss": 0.1334481954574585, "step": 27700 }, { "epoch": 0.729923642506001, "grad_norm": 2.1016392707824707, "learning_rate": 3.588549783233164e-06, "loss": 0.11094794273376465, "step": 27710 }, { "epoch": 0.7301870577504996, "grad_norm": 2.482919216156006, "learning_rate": 3.582004973513201e-06, "loss": 0.1267782211303711, "step": 27720 }, { "epoch": 0.7304504729949984, "grad_norm": 4.837303161621094, "learning_rate": 3.575464835086917e-06, "loss": 0.13400440216064452, "step": 27730 }, { "epoch": 0.7307138882394971, "grad_norm": 3.417283296585083, "learning_rate": 3.568929372714508e-06, "loss": 0.1429229497909546, "step": 27740 }, { "epoch": 0.7309773034839959, "grad_norm": 2.9903159141540527, "learning_rate": 3.562398591152755e-06, "loss": 0.12521674633026122, "step": 27750 }, { "epoch": 0.7312407187284946, "grad_norm": 2.2149131298065186, "learning_rate": 3.5558724951550515e-06, "loss": 0.12159026861190796, "step": 27760 }, { "epoch": 0.7315041339729933, "grad_norm": 1.5602940320968628, "learning_rate": 3.549351089471367e-06, "loss": 0.11654517650604249, "step": 27770 }, { "epoch": 0.7317675492174921, "grad_norm": 2.84265398979187, "learning_rate": 3.54283437884827e-06, "loss": 0.122379469871521, "step": 27780 }, { "epoch": 0.7320309644619908, "grad_norm": 1.8162760734558105, "learning_rate": 3.536322368028895e-06, "loss": 0.1254256248474121, "step": 27790 }, { "epoch": 0.7322943797064896, "grad_norm": 1.5879454612731934, "learning_rate": 3.5298150617529635e-06, "loss": 0.12882823944091798, "step": 27800 }, { "epoch": 0.7325577949509883, "grad_norm": 2.379849910736084, "learning_rate": 3.5233124647567817e-06, "loss": 0.142625093460083, "step": 27810 }, { "epoch": 0.7328212101954871, "grad_norm": 1.8069883584976196, "learning_rate": 3.516814581773216e-06, "loss": 0.11262744665145874, "step": 27820 }, { "epoch": 0.7330846254399858, "grad_norm": 2.8740899562835693, "learning_rate": 3.5103214175317056e-06, "loss": 0.12245486974716187, "step": 27830 }, { "epoch": 0.7333480406844846, "grad_norm": 1.9654358625411987, "learning_rate": 3.5038329767582567e-06, "loss": 0.12913093566894532, "step": 27840 }, { "epoch": 0.7336114559289832, "grad_norm": 3.1827900409698486, "learning_rate": 3.4973492641754315e-06, "loss": 0.1217995047569275, "step": 27850 }, { "epoch": 0.7338748711734819, "grad_norm": 1.8757539987564087, "learning_rate": 3.4908702845023643e-06, "loss": 0.11858587265014649, "step": 27860 }, { "epoch": 0.7341382864179807, "grad_norm": 2.0816283226013184, "learning_rate": 3.484396042454734e-06, "loss": 0.12572598457336426, "step": 27870 }, { "epoch": 0.7344017016624794, "grad_norm": 2.2465474605560303, "learning_rate": 3.477926542744766e-06, "loss": 0.1218374490737915, "step": 27880 }, { "epoch": 0.7346651169069782, "grad_norm": 4.081144332885742, "learning_rate": 3.4714617900812485e-06, "loss": 0.13501815795898436, "step": 27890 }, { "epoch": 0.7349285321514769, "grad_norm": 3.665175437927246, "learning_rate": 3.465001789169502e-06, "loss": 0.12465204000473022, "step": 27900 }, { "epoch": 0.7351919473959757, "grad_norm": 1.835732340812683, "learning_rate": 3.458546544711404e-06, "loss": 0.12090495824813843, "step": 27910 }, { "epoch": 0.7354553626404744, "grad_norm": 2.055253505706787, "learning_rate": 3.45209606140535e-06, "loss": 0.11719338893890381, "step": 27920 }, { "epoch": 0.7357187778849732, "grad_norm": 2.246192693710327, "learning_rate": 3.4456503439462797e-06, "loss": 0.1198773980140686, "step": 27930 }, { "epoch": 0.7359821931294719, "grad_norm": 2.5386788845062256, "learning_rate": 3.4392093970256703e-06, "loss": 0.13448913097381593, "step": 27940 }, { "epoch": 0.7362456083739706, "grad_norm": 2.1159284114837646, "learning_rate": 3.4327732253315195e-06, "loss": 0.13451638221740722, "step": 27950 }, { "epoch": 0.7365090236184694, "grad_norm": 1.797441005706787, "learning_rate": 3.4263418335483477e-06, "loss": 0.12088861465454101, "step": 27960 }, { "epoch": 0.7367724388629681, "grad_norm": 2.2967541217803955, "learning_rate": 3.419915226357203e-06, "loss": 0.11033720970153808, "step": 27970 }, { "epoch": 0.7370358541074669, "grad_norm": 2.6956703662872314, "learning_rate": 3.413493408435642e-06, "loss": 0.13396589756011962, "step": 27980 }, { "epoch": 0.7372992693519655, "grad_norm": 2.272451400756836, "learning_rate": 3.407076384457747e-06, "loss": 0.11289000511169434, "step": 27990 }, { "epoch": 0.7375626845964643, "grad_norm": 1.7562942504882812, "learning_rate": 3.400664159094106e-06, "loss": 0.11186416149139404, "step": 28000 }, { "epoch": 0.737826099840963, "grad_norm": 2.3577520847320557, "learning_rate": 3.394256737011803e-06, "loss": 0.11410292387008666, "step": 28010 }, { "epoch": 0.7380895150854618, "grad_norm": 1.965087652206421, "learning_rate": 3.3878541228744455e-06, "loss": 0.11230289936065674, "step": 28020 }, { "epoch": 0.7383529303299605, "grad_norm": 1.6735106706619263, "learning_rate": 3.38145632134213e-06, "loss": 0.11329684257507325, "step": 28030 }, { "epoch": 0.7386163455744592, "grad_norm": 2.23020601272583, "learning_rate": 3.3750633370714527e-06, "loss": 0.11820834875106812, "step": 28040 }, { "epoch": 0.738879760818958, "grad_norm": 1.899933099746704, "learning_rate": 3.368675174715501e-06, "loss": 0.12387804985046387, "step": 28050 }, { "epoch": 0.7391431760634567, "grad_norm": 3.510939598083496, "learning_rate": 3.3622918389238544e-06, "loss": 0.11805315017700195, "step": 28060 }, { "epoch": 0.7394065913079555, "grad_norm": 1.9937975406646729, "learning_rate": 3.3559133343425865e-06, "loss": 0.1333825945854187, "step": 28070 }, { "epoch": 0.7396700065524542, "grad_norm": 2.471860647201538, "learning_rate": 3.3495396656142442e-06, "loss": 0.12684314250946044, "step": 28080 }, { "epoch": 0.739933421796953, "grad_norm": 2.0876216888427734, "learning_rate": 3.343170837377858e-06, "loss": 0.12169758081436158, "step": 28090 }, { "epoch": 0.7401968370414517, "grad_norm": 2.1361289024353027, "learning_rate": 3.336806854268936e-06, "loss": 0.11502563953399658, "step": 28100 }, { "epoch": 0.7404602522859505, "grad_norm": 3.192746877670288, "learning_rate": 3.3304477209194574e-06, "loss": 0.1196405291557312, "step": 28110 }, { "epoch": 0.7407236675304492, "grad_norm": 1.896697998046875, "learning_rate": 3.3240934419578828e-06, "loss": 0.11688466072082519, "step": 28120 }, { "epoch": 0.7409870827749478, "grad_norm": 1.6780589818954468, "learning_rate": 3.3177440220091207e-06, "loss": 0.11305469274520874, "step": 28130 }, { "epoch": 0.7412504980194466, "grad_norm": 2.458468437194824, "learning_rate": 3.3113994656945537e-06, "loss": 0.1114769458770752, "step": 28140 }, { "epoch": 0.7415139132639453, "grad_norm": 4.0025954246521, "learning_rate": 3.3050597776320293e-06, "loss": 0.10587105751037598, "step": 28150 }, { "epoch": 0.7417773285084441, "grad_norm": 2.9751927852630615, "learning_rate": 3.298724962435842e-06, "loss": 0.11328818798065185, "step": 28160 }, { "epoch": 0.7420407437529428, "grad_norm": 4.339273452758789, "learning_rate": 3.292395024716746e-06, "loss": 0.11864991188049316, "step": 28170 }, { "epoch": 0.7423041589974416, "grad_norm": 1.7529709339141846, "learning_rate": 3.2860699690819407e-06, "loss": 0.11984333992004395, "step": 28180 }, { "epoch": 0.7425675742419403, "grad_norm": 3.852163553237915, "learning_rate": 3.2797498001350737e-06, "loss": 0.12677309513092042, "step": 28190 }, { "epoch": 0.7428309894864391, "grad_norm": 2.647850275039673, "learning_rate": 3.27343452247624e-06, "loss": 0.11199663877487183, "step": 28200 }, { "epoch": 0.7430944047309378, "grad_norm": 2.194591522216797, "learning_rate": 3.267124140701974e-06, "loss": 0.12527174949645997, "step": 28210 }, { "epoch": 0.7433578199754365, "grad_norm": 1.8245127201080322, "learning_rate": 3.2608186594052314e-06, "loss": 0.11133168935775757, "step": 28220 }, { "epoch": 0.7436212352199353, "grad_norm": 2.672757148742676, "learning_rate": 3.2545180831754264e-06, "loss": 0.1271038293838501, "step": 28230 }, { "epoch": 0.743884650464434, "grad_norm": 2.0978808403015137, "learning_rate": 3.248222416598381e-06, "loss": 0.10357989072799682, "step": 28240 }, { "epoch": 0.7441480657089328, "grad_norm": 1.8448913097381592, "learning_rate": 3.2419316642563646e-06, "loss": 0.11404430866241455, "step": 28250 }, { "epoch": 0.7444114809534315, "grad_norm": 2.0631260871887207, "learning_rate": 3.235645830728047e-06, "loss": 0.12097327709197998, "step": 28260 }, { "epoch": 0.7446748961979303, "grad_norm": 2.0825648307800293, "learning_rate": 3.2293649205885293e-06, "loss": 0.106662917137146, "step": 28270 }, { "epoch": 0.744938311442429, "grad_norm": 1.7878354787826538, "learning_rate": 3.2230889384093355e-06, "loss": 0.1100920557975769, "step": 28280 }, { "epoch": 0.7452017266869276, "grad_norm": 3.306718587875366, "learning_rate": 3.2168178887583947e-06, "loss": 0.11626062393188477, "step": 28290 }, { "epoch": 0.7454651419314264, "grad_norm": 4.599343776702881, "learning_rate": 3.210551776200044e-06, "loss": 0.1171034574508667, "step": 28300 }, { "epoch": 0.7457285571759251, "grad_norm": 2.559324264526367, "learning_rate": 3.204290605295033e-06, "loss": 0.12510756254196168, "step": 28310 }, { "epoch": 0.7459919724204239, "grad_norm": 2.3412513732910156, "learning_rate": 3.1980343806005077e-06, "loss": 0.12118568420410156, "step": 28320 }, { "epoch": 0.7462553876649226, "grad_norm": 2.0650124549865723, "learning_rate": 3.1917831066700255e-06, "loss": 0.12568293809890746, "step": 28330 }, { "epoch": 0.7465188029094214, "grad_norm": 2.000950336456299, "learning_rate": 3.1855367880535336e-06, "loss": 0.12882006168365479, "step": 28340 }, { "epoch": 0.7467822181539201, "grad_norm": 2.3690743446350098, "learning_rate": 3.179295429297361e-06, "loss": 0.13339346647262573, "step": 28350 }, { "epoch": 0.7470456333984189, "grad_norm": 2.061357021331787, "learning_rate": 3.173059034944248e-06, "loss": 0.11411210298538207, "step": 28360 }, { "epoch": 0.7473090486429176, "grad_norm": 2.5839695930480957, "learning_rate": 3.1668276095333063e-06, "loss": 0.11598038673400879, "step": 28370 }, { "epoch": 0.7475724638874163, "grad_norm": 1.8226842880249023, "learning_rate": 3.160601157600045e-06, "loss": 0.12673490047454833, "step": 28380 }, { "epoch": 0.7478358791319151, "grad_norm": 2.987696647644043, "learning_rate": 3.154379683676334e-06, "loss": 0.10721387863159179, "step": 28390 }, { "epoch": 0.7480992943764138, "grad_norm": 3.0277318954467773, "learning_rate": 3.148163192290432e-06, "loss": 0.11178420782089234, "step": 28400 }, { "epoch": 0.7483627096209126, "grad_norm": 2.995807647705078, "learning_rate": 3.1419516879669744e-06, "loss": 0.11688053607940674, "step": 28410 }, { "epoch": 0.7486261248654112, "grad_norm": 2.1594133377075195, "learning_rate": 3.1357451752269595e-06, "loss": 0.11442089080810547, "step": 28420 }, { "epoch": 0.74888954010991, "grad_norm": 2.1829943656921387, "learning_rate": 3.1295436585877558e-06, "loss": 0.1325732707977295, "step": 28430 }, { "epoch": 0.7491529553544087, "grad_norm": 6.310051918029785, "learning_rate": 3.1233471425630947e-06, "loss": 0.13088138103485109, "step": 28440 }, { "epoch": 0.7494163705989075, "grad_norm": 2.902792453765869, "learning_rate": 3.1171556316630626e-06, "loss": 0.11159462928771972, "step": 28450 }, { "epoch": 0.7496797858434062, "grad_norm": 2.317699909210205, "learning_rate": 3.110969130394117e-06, "loss": 0.09889677166938782, "step": 28460 }, { "epoch": 0.7499432010879049, "grad_norm": 1.7918360233306885, "learning_rate": 3.1047876432590596e-06, "loss": 0.12483820915222169, "step": 28470 }, { "epoch": 0.7502066163324037, "grad_norm": 2.716852903366089, "learning_rate": 3.0986111747570335e-06, "loss": 0.12196767330169678, "step": 28480 }, { "epoch": 0.7504700315769024, "grad_norm": 2.4043145179748535, "learning_rate": 3.092439729383548e-06, "loss": 0.12616250514984131, "step": 28490 }, { "epoch": 0.7507334468214012, "grad_norm": 2.2133755683898926, "learning_rate": 3.0862733116304435e-06, "loss": 0.11538310050964355, "step": 28500 }, { "epoch": 0.7509968620658999, "grad_norm": 2.6891629695892334, "learning_rate": 3.0801119259859115e-06, "loss": 0.12831244468688965, "step": 28510 }, { "epoch": 0.7512602773103987, "grad_norm": 2.5105440616607666, "learning_rate": 3.0739555769344666e-06, "loss": 0.10287344455718994, "step": 28520 }, { "epoch": 0.7515236925548974, "grad_norm": 2.393852710723877, "learning_rate": 3.067804268956964e-06, "loss": 0.13213424682617186, "step": 28530 }, { "epoch": 0.7517871077993962, "grad_norm": 2.923264503479004, "learning_rate": 3.0616580065305977e-06, "loss": 0.118871009349823, "step": 28540 }, { "epoch": 0.7520505230438949, "grad_norm": 1.8745907545089722, "learning_rate": 3.055516794128881e-06, "loss": 0.11890735626220703, "step": 28550 }, { "epoch": 0.7523139382883935, "grad_norm": 2.7566685676574707, "learning_rate": 3.0493806362216516e-06, "loss": 0.11749769449234009, "step": 28560 }, { "epoch": 0.7525773535328923, "grad_norm": 2.645024538040161, "learning_rate": 3.0432495372750692e-06, "loss": 0.11714366674423218, "step": 28570 }, { "epoch": 0.752840768777391, "grad_norm": 2.9116933345794678, "learning_rate": 3.037123501751611e-06, "loss": 0.11801784038543701, "step": 28580 }, { "epoch": 0.7531041840218898, "grad_norm": 3.029991388320923, "learning_rate": 3.0310025341100757e-06, "loss": 0.12302699089050292, "step": 28590 }, { "epoch": 0.7533675992663885, "grad_norm": 1.9763587713241577, "learning_rate": 3.0248866388055674e-06, "loss": 0.11933374404907227, "step": 28600 }, { "epoch": 0.7536310145108873, "grad_norm": 2.0800869464874268, "learning_rate": 3.018775820289488e-06, "loss": 0.10295345783233642, "step": 28610 }, { "epoch": 0.753894429755386, "grad_norm": 3.23380708694458, "learning_rate": 3.0126700830095668e-06, "loss": 0.11900286674499512, "step": 28620 }, { "epoch": 0.7541578449998848, "grad_norm": 3.6248223781585693, "learning_rate": 3.006569431409817e-06, "loss": 0.11644473075866699, "step": 28630 }, { "epoch": 0.7544212602443835, "grad_norm": 3.3569495677948, "learning_rate": 3.000473869930558e-06, "loss": 0.11170973777770996, "step": 28640 }, { "epoch": 0.7546846754888822, "grad_norm": 2.0438992977142334, "learning_rate": 2.9943834030084006e-06, "loss": 0.11815745830535888, "step": 28650 }, { "epoch": 0.754948090733381, "grad_norm": 1.9687085151672363, "learning_rate": 2.988298035076249e-06, "loss": 0.10643864870071411, "step": 28660 }, { "epoch": 0.7552115059778797, "grad_norm": 2.218473434448242, "learning_rate": 2.9822177705633005e-06, "loss": 0.11687982082366943, "step": 28670 }, { "epoch": 0.7554749212223785, "grad_norm": 3.054957628250122, "learning_rate": 2.976142613895031e-06, "loss": 0.11273727416992188, "step": 28680 }, { "epoch": 0.7557383364668772, "grad_norm": 2.461019277572632, "learning_rate": 2.970072569493202e-06, "loss": 0.09957402348518371, "step": 28690 }, { "epoch": 0.756001751711376, "grad_norm": 3.5358943939208984, "learning_rate": 2.9640076417758546e-06, "loss": 0.1221457839012146, "step": 28700 }, { "epoch": 0.7562651669558746, "grad_norm": 1.859423041343689, "learning_rate": 2.9579478351573e-06, "loss": 0.11317722797393799, "step": 28710 }, { "epoch": 0.7565285822003734, "grad_norm": 3.611952304840088, "learning_rate": 2.9518931540481377e-06, "loss": 0.1188472032546997, "step": 28720 }, { "epoch": 0.7567919974448721, "grad_norm": 1.548575758934021, "learning_rate": 2.945843602855215e-06, "loss": 0.1219332456588745, "step": 28730 }, { "epoch": 0.7570554126893708, "grad_norm": 2.0780982971191406, "learning_rate": 2.9397991859816588e-06, "loss": 0.12235510349273682, "step": 28740 }, { "epoch": 0.7573188279338696, "grad_norm": 2.810734748840332, "learning_rate": 2.9337599078268597e-06, "loss": 0.1128653883934021, "step": 28750 }, { "epoch": 0.7575822431783683, "grad_norm": 2.7552995681762695, "learning_rate": 2.927725772786464e-06, "loss": 0.1284663200378418, "step": 28760 }, { "epoch": 0.7578456584228671, "grad_norm": 1.3084403276443481, "learning_rate": 2.9216967852523747e-06, "loss": 0.10647549629211425, "step": 28770 }, { "epoch": 0.7581090736673658, "grad_norm": 2.6496591567993164, "learning_rate": 2.915672949612749e-06, "loss": 0.11731593608856201, "step": 28780 }, { "epoch": 0.7583724889118646, "grad_norm": 1.536673665046692, "learning_rate": 2.9096542702519904e-06, "loss": 0.11057268381118775, "step": 28790 }, { "epoch": 0.7586359041563633, "grad_norm": 1.9953309297561646, "learning_rate": 2.9036407515507613e-06, "loss": 0.11291948556900025, "step": 28800 }, { "epoch": 0.758899319400862, "grad_norm": 1.8951972723007202, "learning_rate": 2.89763239788596e-06, "loss": 0.1164654016494751, "step": 28810 }, { "epoch": 0.7591627346453608, "grad_norm": 2.81252384185791, "learning_rate": 2.8916292136307147e-06, "loss": 0.12181081771850585, "step": 28820 }, { "epoch": 0.7594261498898595, "grad_norm": 2.1134631633758545, "learning_rate": 2.885631203154412e-06, "loss": 0.1274409055709839, "step": 28830 }, { "epoch": 0.7596895651343583, "grad_norm": 2.770573377609253, "learning_rate": 2.879638370822656e-06, "loss": 0.12655628919601442, "step": 28840 }, { "epoch": 0.759952980378857, "grad_norm": 2.2271296977996826, "learning_rate": 2.8736507209972996e-06, "loss": 0.11736359596252441, "step": 28850 }, { "epoch": 0.7602163956233557, "grad_norm": 2.5906193256378174, "learning_rate": 2.867668258036401e-06, "loss": 0.11780822277069092, "step": 28860 }, { "epoch": 0.7604798108678544, "grad_norm": 2.1039836406707764, "learning_rate": 2.8616909862942565e-06, "loss": 0.10961490869522095, "step": 28870 }, { "epoch": 0.7607432261123532, "grad_norm": 2.1933295726776123, "learning_rate": 2.855718910121389e-06, "loss": 0.11098673343658447, "step": 28880 }, { "epoch": 0.7610066413568519, "grad_norm": 1.9009956121444702, "learning_rate": 2.8497520338645278e-06, "loss": 0.09906232357025146, "step": 28890 }, { "epoch": 0.7612700566013506, "grad_norm": 2.4102237224578857, "learning_rate": 2.843790361866625e-06, "loss": 0.10889662504196167, "step": 28900 }, { "epoch": 0.7615334718458494, "grad_norm": 1.5940042734146118, "learning_rate": 2.837833898466842e-06, "loss": 0.10789523124694825, "step": 28910 }, { "epoch": 0.7617968870903481, "grad_norm": 2.145087718963623, "learning_rate": 2.831882648000547e-06, "loss": 0.10669511556625366, "step": 28920 }, { "epoch": 0.7620603023348469, "grad_norm": 2.416409492492676, "learning_rate": 2.8259366147993217e-06, "loss": 0.11529054641723632, "step": 28930 }, { "epoch": 0.7623237175793456, "grad_norm": 1.6907166242599487, "learning_rate": 2.819995803190948e-06, "loss": 0.1151268720626831, "step": 28940 }, { "epoch": 0.7625871328238444, "grad_norm": 2.7375285625457764, "learning_rate": 2.8140602174993927e-06, "loss": 0.10674018859863281, "step": 28950 }, { "epoch": 0.7628505480683431, "grad_norm": 2.5392818450927734, "learning_rate": 2.8081298620448427e-06, "loss": 0.12054781913757324, "step": 28960 }, { "epoch": 0.7631139633128419, "grad_norm": 2.6401875019073486, "learning_rate": 2.8022047411436604e-06, "loss": 0.09953733682632446, "step": 28970 }, { "epoch": 0.7633773785573406, "grad_norm": 3.005009174346924, "learning_rate": 2.7962848591084122e-06, "loss": 0.130575430393219, "step": 28980 }, { "epoch": 0.7636407938018392, "grad_norm": 2.1627349853515625, "learning_rate": 2.7903702202478355e-06, "loss": 0.11203433275222778, "step": 28990 }, { "epoch": 0.763904209046338, "grad_norm": 3.923448085784912, "learning_rate": 2.7844608288668586e-06, "loss": 0.10739712715148926, "step": 29000 }, { "epoch": 0.7641676242908367, "grad_norm": 2.438535690307617, "learning_rate": 2.7785566892665993e-06, "loss": 0.11827080249786377, "step": 29010 }, { "epoch": 0.7644310395353355, "grad_norm": 2.8706812858581543, "learning_rate": 2.77265780574434e-06, "loss": 0.11493736505508423, "step": 29020 }, { "epoch": 0.7646944547798342, "grad_norm": 2.424362897872925, "learning_rate": 2.766764182593544e-06, "loss": 0.11397206783294678, "step": 29030 }, { "epoch": 0.764957870024333, "grad_norm": 2.1609432697296143, "learning_rate": 2.760875824103846e-06, "loss": 0.12205902338027955, "step": 29040 }, { "epoch": 0.7652212852688317, "grad_norm": 2.705913782119751, "learning_rate": 2.7549927345610407e-06, "loss": 0.1027780294418335, "step": 29050 }, { "epoch": 0.7654847005133305, "grad_norm": 2.4224650859832764, "learning_rate": 2.749114918247102e-06, "loss": 0.10807712078094482, "step": 29060 }, { "epoch": 0.7657481157578292, "grad_norm": 2.0975418090820312, "learning_rate": 2.7432423794401598e-06, "loss": 0.10942602157592773, "step": 29070 }, { "epoch": 0.7660115310023279, "grad_norm": 2.188889503479004, "learning_rate": 2.7373751224144905e-06, "loss": 0.12077691555023193, "step": 29080 }, { "epoch": 0.7662749462468267, "grad_norm": 2.6809520721435547, "learning_rate": 2.731513151440547e-06, "loss": 0.12171189785003662, "step": 29090 }, { "epoch": 0.7665383614913254, "grad_norm": 1.9311028718948364, "learning_rate": 2.7256564707849165e-06, "loss": 0.1107512354850769, "step": 29100 }, { "epoch": 0.7668017767358242, "grad_norm": 2.673798084259033, "learning_rate": 2.719805084710356e-06, "loss": 0.12725434303283692, "step": 29110 }, { "epoch": 0.7670651919803229, "grad_norm": 2.0340874195098877, "learning_rate": 2.7139589974757454e-06, "loss": 0.09881865978240967, "step": 29120 }, { "epoch": 0.7673286072248217, "grad_norm": 2.0231549739837646, "learning_rate": 2.7081182133361207e-06, "loss": 0.11652929782867431, "step": 29130 }, { "epoch": 0.7675920224693203, "grad_norm": 2.271456480026245, "learning_rate": 2.7022827365426606e-06, "loss": 0.11468322277069092, "step": 29140 }, { "epoch": 0.7678554377138191, "grad_norm": 1.960116982460022, "learning_rate": 2.696452571342676e-06, "loss": 0.11467940807342529, "step": 29150 }, { "epoch": 0.7681188529583178, "grad_norm": 2.1705329418182373, "learning_rate": 2.6906277219796116e-06, "loss": 0.1085382342338562, "step": 29160 }, { "epoch": 0.7683822682028165, "grad_norm": 2.266528367996216, "learning_rate": 2.6848081926930447e-06, "loss": 0.09873645901679992, "step": 29170 }, { "epoch": 0.7686456834473153, "grad_norm": 3.2041549682617188, "learning_rate": 2.6789939877186753e-06, "loss": 0.11889538764953614, "step": 29180 }, { "epoch": 0.768909098691814, "grad_norm": 2.24985933303833, "learning_rate": 2.67318511128834e-06, "loss": 0.10896039009094238, "step": 29190 }, { "epoch": 0.7691725139363128, "grad_norm": 3.131006956100464, "learning_rate": 2.66738156762999e-06, "loss": 0.10100078582763672, "step": 29200 }, { "epoch": 0.7694359291808115, "grad_norm": 2.583136558532715, "learning_rate": 2.6615833609676846e-06, "loss": 0.11479448080062866, "step": 29210 }, { "epoch": 0.7696993444253103, "grad_norm": 2.264907121658325, "learning_rate": 2.6557904955216185e-06, "loss": 0.11842861175537109, "step": 29220 }, { "epoch": 0.769962759669809, "grad_norm": 1.9222725629806519, "learning_rate": 2.650002975508086e-06, "loss": 0.10224096775054932, "step": 29230 }, { "epoch": 0.7702261749143078, "grad_norm": 2.7816455364227295, "learning_rate": 2.6442208051394936e-06, "loss": 0.11137328147888184, "step": 29240 }, { "epoch": 0.7704895901588065, "grad_norm": 2.574155569076538, "learning_rate": 2.638443988624355e-06, "loss": 0.10529007911682128, "step": 29250 }, { "epoch": 0.7707530054033052, "grad_norm": 2.289538860321045, "learning_rate": 2.632672530167283e-06, "loss": 0.10358741283416747, "step": 29260 }, { "epoch": 0.771016420647804, "grad_norm": 2.251479387283325, "learning_rate": 2.6269064339689986e-06, "loss": 0.11280766725540162, "step": 29270 }, { "epoch": 0.7712798358923026, "grad_norm": 1.912541151046753, "learning_rate": 2.6211457042263157e-06, "loss": 0.10693224668502807, "step": 29280 }, { "epoch": 0.7715432511368014, "grad_norm": 2.3536839485168457, "learning_rate": 2.6153903451321384e-06, "loss": 0.11044926643371582, "step": 29290 }, { "epoch": 0.7718066663813001, "grad_norm": 2.3933510780334473, "learning_rate": 2.6096403608754685e-06, "loss": 0.10358012914657592, "step": 29300 }, { "epoch": 0.7720700816257989, "grad_norm": 2.658748149871826, "learning_rate": 2.603895755641387e-06, "loss": 0.12783713340759278, "step": 29310 }, { "epoch": 0.7723334968702976, "grad_norm": 1.7865756750106812, "learning_rate": 2.598156533611077e-06, "loss": 0.11146925687789917, "step": 29320 }, { "epoch": 0.7725969121147963, "grad_norm": 3.595104455947876, "learning_rate": 2.592422698961781e-06, "loss": 0.10829260349273681, "step": 29330 }, { "epoch": 0.7728603273592951, "grad_norm": 2.3512842655181885, "learning_rate": 2.586694255866831e-06, "loss": 0.12005445957183838, "step": 29340 }, { "epoch": 0.7731237426037938, "grad_norm": 3.3601531982421875, "learning_rate": 2.580971208495642e-06, "loss": 0.10594971179962158, "step": 29350 }, { "epoch": 0.7733871578482926, "grad_norm": 3.2162082195281982, "learning_rate": 2.575253561013692e-06, "loss": 0.10743436813354493, "step": 29360 }, { "epoch": 0.7736505730927913, "grad_norm": 2.5249886512756348, "learning_rate": 2.56954131758253e-06, "loss": 0.08974694013595581, "step": 29370 }, { "epoch": 0.7739139883372901, "grad_norm": 3.384404420852661, "learning_rate": 2.5638344823597762e-06, "loss": 0.10821253061294556, "step": 29380 }, { "epoch": 0.7741774035817888, "grad_norm": 2.4968183040618896, "learning_rate": 2.558133059499105e-06, "loss": 0.11371484994888306, "step": 29390 }, { "epoch": 0.7744408188262876, "grad_norm": 3.043464183807373, "learning_rate": 2.552437053150266e-06, "loss": 0.10206865072250366, "step": 29400 }, { "epoch": 0.7747042340707863, "grad_norm": 1.6694424152374268, "learning_rate": 2.546746467459058e-06, "loss": 0.10691351890563965, "step": 29410 }, { "epoch": 0.774967649315285, "grad_norm": 3.490496873855591, "learning_rate": 2.541061306567325e-06, "loss": 0.12137477397918701, "step": 29420 }, { "epoch": 0.7752310645597837, "grad_norm": 2.4776833057403564, "learning_rate": 2.5353815746129827e-06, "loss": 0.10609071254730225, "step": 29430 }, { "epoch": 0.7754944798042824, "grad_norm": 2.66365385055542, "learning_rate": 2.529707275729979e-06, "loss": 0.11133675575256348, "step": 29440 }, { "epoch": 0.7757578950487812, "grad_norm": 3.805272102355957, "learning_rate": 2.524038414048321e-06, "loss": 0.12484568357467651, "step": 29450 }, { "epoch": 0.7760213102932799, "grad_norm": 2.60904860496521, "learning_rate": 2.518374993694043e-06, "loss": 0.11197469234466553, "step": 29460 }, { "epoch": 0.7762847255377787, "grad_norm": 3.124025583267212, "learning_rate": 2.5127170187892257e-06, "loss": 0.11041439771652221, "step": 29470 }, { "epoch": 0.7765481407822774, "grad_norm": 3.656919479370117, "learning_rate": 2.507064493451995e-06, "loss": 0.10264760255813599, "step": 29480 }, { "epoch": 0.7768115560267762, "grad_norm": 2.117657423019409, "learning_rate": 2.501417421796498e-06, "loss": 0.1161865234375, "step": 29490 }, { "epoch": 0.7770749712712749, "grad_norm": 2.577728033065796, "learning_rate": 2.4957758079329173e-06, "loss": 0.11339452266693115, "step": 29500 }, { "epoch": 0.7773383865157736, "grad_norm": 2.5522615909576416, "learning_rate": 2.490139655967465e-06, "loss": 0.11866680383682252, "step": 29510 }, { "epoch": 0.7776018017602724, "grad_norm": 1.842865228652954, "learning_rate": 2.4845089700023694e-06, "loss": 0.11606624126434326, "step": 29520 }, { "epoch": 0.7778652170047711, "grad_norm": 2.435579776763916, "learning_rate": 2.4788837541358957e-06, "loss": 0.10506585836410523, "step": 29530 }, { "epoch": 0.7781286322492699, "grad_norm": 3.322579860687256, "learning_rate": 2.4732640124623152e-06, "loss": 0.11336712837219239, "step": 29540 }, { "epoch": 0.7783920474937686, "grad_norm": 2.54613995552063, "learning_rate": 2.467649749071913e-06, "loss": 0.11100215911865234, "step": 29550 }, { "epoch": 0.7786554627382674, "grad_norm": 1.911004900932312, "learning_rate": 2.4620409680509993e-06, "loss": 0.11286530494689942, "step": 29560 }, { "epoch": 0.778918877982766, "grad_norm": 2.8584821224212646, "learning_rate": 2.4564376734818818e-06, "loss": 0.12212491035461426, "step": 29570 }, { "epoch": 0.7791822932272648, "grad_norm": 2.2221524715423584, "learning_rate": 2.450839869442887e-06, "loss": 0.10473718643188476, "step": 29580 }, { "epoch": 0.7794457084717635, "grad_norm": 3.483386993408203, "learning_rate": 2.4452475600083323e-06, "loss": 0.10374007225036622, "step": 29590 }, { "epoch": 0.7797091237162622, "grad_norm": 2.402254104614258, "learning_rate": 2.4396607492485393e-06, "loss": 0.10778200626373291, "step": 29600 }, { "epoch": 0.779972538960761, "grad_norm": 2.626793622970581, "learning_rate": 2.4340794412298364e-06, "loss": 0.12552030086517335, "step": 29610 }, { "epoch": 0.7802359542052597, "grad_norm": 2.353288173675537, "learning_rate": 2.428503640014539e-06, "loss": 0.10887128114700317, "step": 29620 }, { "epoch": 0.7804993694497585, "grad_norm": 1.8163460493087769, "learning_rate": 2.4229333496609532e-06, "loss": 0.11660180091857911, "step": 29630 }, { "epoch": 0.7807627846942572, "grad_norm": 2.965050220489502, "learning_rate": 2.4173685742233787e-06, "loss": 0.10872695446014405, "step": 29640 }, { "epoch": 0.781026199938756, "grad_norm": 2.799130439758301, "learning_rate": 2.411809317752095e-06, "loss": 0.10723826885223389, "step": 29650 }, { "epoch": 0.7812896151832547, "grad_norm": 2.2457075119018555, "learning_rate": 2.4062555842933753e-06, "loss": 0.10560915470123292, "step": 29660 }, { "epoch": 0.7815530304277535, "grad_norm": 2.179614305496216, "learning_rate": 2.4007073778894665e-06, "loss": 0.10557563304901123, "step": 29670 }, { "epoch": 0.7818164456722522, "grad_norm": 2.1456947326660156, "learning_rate": 2.395164702578583e-06, "loss": 0.11028563976287842, "step": 29680 }, { "epoch": 0.7820798609167509, "grad_norm": 3.4363436698913574, "learning_rate": 2.3896275623949327e-06, "loss": 0.10988218784332275, "step": 29690 }, { "epoch": 0.7823432761612497, "grad_norm": 2.636457681655884, "learning_rate": 2.38409596136868e-06, "loss": 0.11228814125061035, "step": 29700 }, { "epoch": 0.7826066914057483, "grad_norm": 2.134840250015259, "learning_rate": 2.378569903525969e-06, "loss": 0.11996212005615234, "step": 29710 }, { "epoch": 0.7828701066502471, "grad_norm": 3.7303524017333984, "learning_rate": 2.3730493928888963e-06, "loss": 0.12836227416992188, "step": 29720 }, { "epoch": 0.7831335218947458, "grad_norm": 2.353520393371582, "learning_rate": 2.367534433475527e-06, "loss": 0.11127676963806152, "step": 29730 }, { "epoch": 0.7833969371392446, "grad_norm": 2.450260639190674, "learning_rate": 2.3620250292998925e-06, "loss": 0.11827961206436158, "step": 29740 }, { "epoch": 0.7836603523837433, "grad_norm": 2.2529661655426025, "learning_rate": 2.356521184371973e-06, "loss": 0.11196300983428956, "step": 29750 }, { "epoch": 0.7839237676282421, "grad_norm": 2.2451536655426025, "learning_rate": 2.3510229026977018e-06, "loss": 0.10846577882766724, "step": 29760 }, { "epoch": 0.7841871828727408, "grad_norm": 2.8861472606658936, "learning_rate": 2.345530188278967e-06, "loss": 0.1121056079864502, "step": 29770 }, { "epoch": 0.7844505981172395, "grad_norm": 2.1899003982543945, "learning_rate": 2.3400430451136e-06, "loss": 0.10628859996795655, "step": 29780 }, { "epoch": 0.7847140133617383, "grad_norm": 3.7910988330841064, "learning_rate": 2.334561477195387e-06, "loss": 0.0969285249710083, "step": 29790 }, { "epoch": 0.784977428606237, "grad_norm": 1.7524710893630981, "learning_rate": 2.329085488514049e-06, "loss": 0.11059653759002686, "step": 29800 }, { "epoch": 0.7852408438507358, "grad_norm": 2.2332656383514404, "learning_rate": 2.3236150830552374e-06, "loss": 0.10003794431686401, "step": 29810 }, { "epoch": 0.7855042590952345, "grad_norm": 2.453404426574707, "learning_rate": 2.3181502648005593e-06, "loss": 0.10588675737380981, "step": 29820 }, { "epoch": 0.7857676743397333, "grad_norm": 3.064716339111328, "learning_rate": 2.3126910377275415e-06, "loss": 0.10239554643630981, "step": 29830 }, { "epoch": 0.786031089584232, "grad_norm": 2.5366268157958984, "learning_rate": 2.3072374058096447e-06, "loss": 0.12278655767440796, "step": 29840 }, { "epoch": 0.7862945048287308, "grad_norm": 2.159226179122925, "learning_rate": 2.301789373016259e-06, "loss": 0.10622470378875733, "step": 29850 }, { "epoch": 0.7865579200732294, "grad_norm": 2.437521457672119, "learning_rate": 2.296346943312693e-06, "loss": 0.10723904371261597, "step": 29860 }, { "epoch": 0.7868213353177281, "grad_norm": 3.1769561767578125, "learning_rate": 2.290910120660188e-06, "loss": 0.11450079679489136, "step": 29870 }, { "epoch": 0.7870847505622269, "grad_norm": 2.4654738903045654, "learning_rate": 2.285478909015896e-06, "loss": 0.09952437877655029, "step": 29880 }, { "epoch": 0.7873481658067256, "grad_norm": 2.7623672485351562, "learning_rate": 2.2800533123328874e-06, "loss": 0.12513447999954225, "step": 29890 }, { "epoch": 0.7876115810512244, "grad_norm": 3.341433048248291, "learning_rate": 2.2746333345601435e-06, "loss": 0.10991413593292236, "step": 29900 }, { "epoch": 0.7878749962957231, "grad_norm": 2.8977160453796387, "learning_rate": 2.269218979642558e-06, "loss": 0.10130587816238404, "step": 29910 }, { "epoch": 0.7881384115402219, "grad_norm": 2.327092170715332, "learning_rate": 2.2638102515209382e-06, "loss": 0.11158301830291747, "step": 29920 }, { "epoch": 0.7884018267847206, "grad_norm": 2.9963276386260986, "learning_rate": 2.258407154131982e-06, "loss": 0.11254801750183105, "step": 29930 }, { "epoch": 0.7886652420292193, "grad_norm": 2.891008138656616, "learning_rate": 2.2530096914082965e-06, "loss": 0.11863704919815063, "step": 29940 }, { "epoch": 0.7889286572737181, "grad_norm": 2.3726484775543213, "learning_rate": 2.2476178672783943e-06, "loss": 0.10747698545455933, "step": 29950 }, { "epoch": 0.7891920725182168, "grad_norm": 3.178818702697754, "learning_rate": 2.242231685666675e-06, "loss": 0.10248062610626221, "step": 29960 }, { "epoch": 0.7894554877627156, "grad_norm": 2.6864562034606934, "learning_rate": 2.2368511504934333e-06, "loss": 0.10796010494232178, "step": 29970 }, { "epoch": 0.7897189030072143, "grad_norm": 2.0640218257904053, "learning_rate": 2.2314762656748544e-06, "loss": 0.09941590428352357, "step": 29980 }, { "epoch": 0.7899823182517131, "grad_norm": 1.9324783086776733, "learning_rate": 2.2261070351230107e-06, "loss": 0.11207756996154786, "step": 29990 }, { "epoch": 0.7902457334962117, "grad_norm": 2.7422685623168945, "learning_rate": 2.2207434627458626e-06, "loss": 0.11095852851867676, "step": 30000 }, { "epoch": 0.7905091487407105, "grad_norm": 2.9182963371276855, "learning_rate": 2.2153855524472513e-06, "loss": 0.11992000341415406, "step": 30010 }, { "epoch": 0.7907725639852092, "grad_norm": 2.040980815887451, "learning_rate": 2.2100333081268853e-06, "loss": 0.10605534315109252, "step": 30020 }, { "epoch": 0.7910359792297079, "grad_norm": 3.162644386291504, "learning_rate": 2.204686733680369e-06, "loss": 0.10425312519073486, "step": 30030 }, { "epoch": 0.7912993944742067, "grad_norm": 2.9849767684936523, "learning_rate": 2.1993458329991635e-06, "loss": 0.10443499088287353, "step": 30040 }, { "epoch": 0.7915628097187054, "grad_norm": 2.2556686401367188, "learning_rate": 2.1940106099706137e-06, "loss": 0.11454148292541504, "step": 30050 }, { "epoch": 0.7918262249632042, "grad_norm": 1.9302639961242676, "learning_rate": 2.1886810684779204e-06, "loss": 0.10940110683441162, "step": 30060 }, { "epoch": 0.7920896402077029, "grad_norm": 2.675387144088745, "learning_rate": 2.18335721240015e-06, "loss": 0.10774338245391846, "step": 30070 }, { "epoch": 0.7923530554522017, "grad_norm": 2.858506202697754, "learning_rate": 2.1780390456122424e-06, "loss": 0.09790219068527221, "step": 30080 }, { "epoch": 0.7926164706967004, "grad_norm": 2.3895041942596436, "learning_rate": 2.1727265719849853e-06, "loss": 0.1141669511795044, "step": 30090 }, { "epoch": 0.7928798859411992, "grad_norm": 2.294370412826538, "learning_rate": 2.1674197953850274e-06, "loss": 0.1063225507736206, "step": 30100 }, { "epoch": 0.7931433011856979, "grad_norm": 2.9662277698516846, "learning_rate": 2.162118719674867e-06, "loss": 0.11137359142303467, "step": 30110 }, { "epoch": 0.7934067164301966, "grad_norm": 2.6668930053710938, "learning_rate": 2.1568233487128556e-06, "loss": 0.09982303380966187, "step": 30120 }, { "epoch": 0.7936701316746954, "grad_norm": 4.511026382446289, "learning_rate": 2.1515336863531976e-06, "loss": 0.11463935375213623, "step": 30130 }, { "epoch": 0.793933546919194, "grad_norm": 3.045987606048584, "learning_rate": 2.1462497364459356e-06, "loss": 0.1089849591255188, "step": 30140 }, { "epoch": 0.7941969621636928, "grad_norm": 3.009650945663452, "learning_rate": 2.1409715028369504e-06, "loss": 0.10296461582183838, "step": 30150 }, { "epoch": 0.7944603774081915, "grad_norm": 2.3335297107696533, "learning_rate": 2.135698989367976e-06, "loss": 0.11586787700653076, "step": 30160 }, { "epoch": 0.7947237926526903, "grad_norm": 3.102854013442993, "learning_rate": 2.1304321998765685e-06, "loss": 0.10197765827178955, "step": 30170 }, { "epoch": 0.794987207897189, "grad_norm": 1.9542112350463867, "learning_rate": 2.1251711381961336e-06, "loss": 0.10587058067321778, "step": 30180 }, { "epoch": 0.7952506231416878, "grad_norm": 3.9170644283294678, "learning_rate": 2.11991580815589e-06, "loss": 0.10531576871871948, "step": 30190 }, { "epoch": 0.7955140383861865, "grad_norm": 2.7885563373565674, "learning_rate": 2.1146662135808948e-06, "loss": 0.10875024795532226, "step": 30200 }, { "epoch": 0.7957774536306852, "grad_norm": 2.7890737056732178, "learning_rate": 2.1094223582920346e-06, "loss": 0.11051335334777831, "step": 30210 }, { "epoch": 0.796040868875184, "grad_norm": 3.0502092838287354, "learning_rate": 2.1041842461060103e-06, "loss": 0.10606826543807983, "step": 30220 }, { "epoch": 0.7963042841196827, "grad_norm": 2.1548945903778076, "learning_rate": 2.098951880835346e-06, "loss": 0.1070556879043579, "step": 30230 }, { "epoch": 0.7965676993641815, "grad_norm": 2.3260509967803955, "learning_rate": 2.093725266288382e-06, "loss": 0.10880136489868164, "step": 30240 }, { "epoch": 0.7968311146086802, "grad_norm": 4.0262346267700195, "learning_rate": 2.088504406269274e-06, "loss": 0.11035466194152832, "step": 30250 }, { "epoch": 0.797094529853179, "grad_norm": 3.3541433811187744, "learning_rate": 2.083289304577992e-06, "loss": 0.10738885402679443, "step": 30260 }, { "epoch": 0.7973579450976777, "grad_norm": 3.4272220134735107, "learning_rate": 2.0780799650103123e-06, "loss": 0.10007245540618896, "step": 30270 }, { "epoch": 0.7976213603421765, "grad_norm": 3.6515345573425293, "learning_rate": 2.0728763913578097e-06, "loss": 0.10623466968536377, "step": 30280 }, { "epoch": 0.7978847755866751, "grad_norm": 2.4740447998046875, "learning_rate": 2.0676785874078773e-06, "loss": 0.09987521171569824, "step": 30290 }, { "epoch": 0.7981481908311738, "grad_norm": 2.4538040161132812, "learning_rate": 2.062486556943699e-06, "loss": 0.10614453554153443, "step": 30300 }, { "epoch": 0.7984116060756726, "grad_norm": 2.367638349533081, "learning_rate": 2.0573003037442575e-06, "loss": 0.11405912637710572, "step": 30310 }, { "epoch": 0.7986750213201713, "grad_norm": 2.665231466293335, "learning_rate": 2.0521198315843315e-06, "loss": 0.10186423063278198, "step": 30320 }, { "epoch": 0.7989384365646701, "grad_norm": 2.2501401901245117, "learning_rate": 2.046945144234491e-06, "loss": 0.10970971584320069, "step": 30330 }, { "epoch": 0.7992018518091688, "grad_norm": 1.9746222496032715, "learning_rate": 2.0417762454611012e-06, "loss": 0.11975278854370117, "step": 30340 }, { "epoch": 0.7994652670536676, "grad_norm": 2.647014856338501, "learning_rate": 2.0366131390263055e-06, "loss": 0.10874041318893432, "step": 30350 }, { "epoch": 0.7997286822981663, "grad_norm": 2.515493631362915, "learning_rate": 2.0314558286880383e-06, "loss": 0.10149402618408203, "step": 30360 }, { "epoch": 0.7999920975426651, "grad_norm": 3.012254476547241, "learning_rate": 2.026304318200011e-06, "loss": 0.12133454084396363, "step": 30370 }, { "epoch": 0.8002555127871638, "grad_norm": 2.955758571624756, "learning_rate": 2.021158611311712e-06, "loss": 0.10652695894241333, "step": 30380 }, { "epoch": 0.8005189280316625, "grad_norm": 2.251390218734741, "learning_rate": 2.016018711768416e-06, "loss": 0.11363308429718018, "step": 30390 }, { "epoch": 0.8007823432761613, "grad_norm": 2.4574763774871826, "learning_rate": 2.010884623311162e-06, "loss": 0.10959346294403076, "step": 30400 }, { "epoch": 0.80104575852066, "grad_norm": 2.6107170581817627, "learning_rate": 2.005756349676753e-06, "loss": 0.1065071702003479, "step": 30410 }, { "epoch": 0.8013091737651588, "grad_norm": 3.4752278327941895, "learning_rate": 2.0006338945977777e-06, "loss": 0.09422449469566345, "step": 30420 }, { "epoch": 0.8015725890096574, "grad_norm": 2.9676578044891357, "learning_rate": 1.9955172618025743e-06, "loss": 0.10078740119934082, "step": 30430 }, { "epoch": 0.8018360042541562, "grad_norm": 2.7762274742126465, "learning_rate": 1.9904064550152515e-06, "loss": 0.09523159265518188, "step": 30440 }, { "epoch": 0.8020994194986549, "grad_norm": 3.171487331390381, "learning_rate": 1.9853014779556734e-06, "loss": 0.09946731328964234, "step": 30450 }, { "epoch": 0.8023628347431536, "grad_norm": 2.7213995456695557, "learning_rate": 1.9802023343394606e-06, "loss": 0.09615564346313477, "step": 30460 }, { "epoch": 0.8026262499876524, "grad_norm": 2.531419515609741, "learning_rate": 1.975109027877995e-06, "loss": 0.11699367761611938, "step": 30470 }, { "epoch": 0.8028896652321511, "grad_norm": 3.641385555267334, "learning_rate": 1.9700215622784026e-06, "loss": 0.09523487091064453, "step": 30480 }, { "epoch": 0.8031530804766499, "grad_norm": 4.488335132598877, "learning_rate": 1.96493994124356e-06, "loss": 0.11141613721847535, "step": 30490 }, { "epoch": 0.8034164957211486, "grad_norm": 2.0835556983947754, "learning_rate": 1.9598641684720897e-06, "loss": 0.09263993501663208, "step": 30500 }, { "epoch": 0.8036799109656474, "grad_norm": 2.5380449295043945, "learning_rate": 1.954794247658356e-06, "loss": 0.11805497407913208, "step": 30510 }, { "epoch": 0.8039433262101461, "grad_norm": 2.7328152656555176, "learning_rate": 1.949730182492473e-06, "loss": 0.09879330396652222, "step": 30520 }, { "epoch": 0.8042067414546449, "grad_norm": 2.681596279144287, "learning_rate": 1.9446719766602793e-06, "loss": 0.10307911634445191, "step": 30530 }, { "epoch": 0.8044701566991436, "grad_norm": 4.028637886047363, "learning_rate": 1.9396196338433547e-06, "loss": 0.10225174427032471, "step": 30540 }, { "epoch": 0.8047335719436423, "grad_norm": 2.716273784637451, "learning_rate": 1.9345731577190187e-06, "loss": 0.11104879379272461, "step": 30550 }, { "epoch": 0.804996987188141, "grad_norm": 2.782177448272705, "learning_rate": 1.929532551960309e-06, "loss": 0.10883151292800904, "step": 30560 }, { "epoch": 0.8052604024326397, "grad_norm": 2.572028398513794, "learning_rate": 1.9244978202359974e-06, "loss": 0.08394054174423218, "step": 30570 }, { "epoch": 0.8055238176771385, "grad_norm": 1.9692386388778687, "learning_rate": 1.9194689662105803e-06, "loss": 0.0955109417438507, "step": 30580 }, { "epoch": 0.8057872329216372, "grad_norm": 3.1990203857421875, "learning_rate": 1.9144459935442705e-06, "loss": 0.0997592568397522, "step": 30590 }, { "epoch": 0.806050648166136, "grad_norm": 2.3791825771331787, "learning_rate": 1.9094289058930095e-06, "loss": 0.09540140628814697, "step": 30600 }, { "epoch": 0.8063140634106347, "grad_norm": 2.2074570655822754, "learning_rate": 1.9044177069084502e-06, "loss": 0.10997228622436524, "step": 30610 }, { "epoch": 0.8065774786551335, "grad_norm": 2.3145792484283447, "learning_rate": 1.899412400237951e-06, "loss": 0.10221612453460693, "step": 30620 }, { "epoch": 0.8068408938996322, "grad_norm": 2.834839105606079, "learning_rate": 1.8944129895246e-06, "loss": 0.10689527988433838, "step": 30630 }, { "epoch": 0.8071043091441309, "grad_norm": 2.8995046615600586, "learning_rate": 1.8894194784071761e-06, "loss": 0.1073765754699707, "step": 30640 }, { "epoch": 0.8073677243886297, "grad_norm": 3.4976580142974854, "learning_rate": 1.8844318705201815e-06, "loss": 0.10817155838012696, "step": 30650 }, { "epoch": 0.8076311396331284, "grad_norm": 2.818986415863037, "learning_rate": 1.879450169493805e-06, "loss": 0.11415486335754395, "step": 30660 }, { "epoch": 0.8078945548776272, "grad_norm": 3.05401873588562, "learning_rate": 1.874474378953942e-06, "loss": 0.11806070804595947, "step": 30670 }, { "epoch": 0.8081579701221259, "grad_norm": 4.321625232696533, "learning_rate": 1.8695045025221936e-06, "loss": 0.09705685377120972, "step": 30680 }, { "epoch": 0.8084213853666247, "grad_norm": 2.6905932426452637, "learning_rate": 1.864540543815848e-06, "loss": 0.09171236753463745, "step": 30690 }, { "epoch": 0.8086848006111234, "grad_norm": 3.6291770935058594, "learning_rate": 1.859582506447889e-06, "loss": 0.10777089595794678, "step": 30700 }, { "epoch": 0.8089482158556222, "grad_norm": 2.3738183975219727, "learning_rate": 1.8546303940269883e-06, "loss": 0.10259437561035156, "step": 30710 }, { "epoch": 0.8092116311001208, "grad_norm": 2.7342607975006104, "learning_rate": 1.8496842101575053e-06, "loss": 0.100028395652771, "step": 30720 }, { "epoch": 0.8094750463446195, "grad_norm": 3.1948366165161133, "learning_rate": 1.844743958439491e-06, "loss": 0.10885860919952392, "step": 30730 }, { "epoch": 0.8097384615891183, "grad_norm": 3.2972469329833984, "learning_rate": 1.839809642468674e-06, "loss": 0.10226060152053833, "step": 30740 }, { "epoch": 0.810001876833617, "grad_norm": 1.6437890529632568, "learning_rate": 1.8348812658364557e-06, "loss": 0.11404252052307129, "step": 30750 }, { "epoch": 0.8102652920781158, "grad_norm": 2.03418231010437, "learning_rate": 1.8299588321299266e-06, "loss": 0.10449473857879639, "step": 30760 }, { "epoch": 0.8105287073226145, "grad_norm": 2.3469181060791016, "learning_rate": 1.8250423449318434e-06, "loss": 0.10139027833938599, "step": 30770 }, { "epoch": 0.8107921225671133, "grad_norm": 2.332538604736328, "learning_rate": 1.8201318078206443e-06, "loss": 0.11305674314498901, "step": 30780 }, { "epoch": 0.811055537811612, "grad_norm": 2.4009010791778564, "learning_rate": 1.8152272243704216e-06, "loss": 0.12614088058471679, "step": 30790 }, { "epoch": 0.8113189530561108, "grad_norm": 2.7223496437072754, "learning_rate": 1.8103285981509432e-06, "loss": 0.1107494592666626, "step": 30800 }, { "epoch": 0.8115823683006095, "grad_norm": 2.459045171737671, "learning_rate": 1.8054359327276473e-06, "loss": 0.10410108566284179, "step": 30810 }, { "epoch": 0.8118457835451082, "grad_norm": 2.584019422531128, "learning_rate": 1.8005492316616224e-06, "loss": 0.10946646928787232, "step": 30820 }, { "epoch": 0.812109198789607, "grad_norm": 2.474379777908325, "learning_rate": 1.7956684985096217e-06, "loss": 0.08909925818443298, "step": 30830 }, { "epoch": 0.8123726140341057, "grad_norm": 2.2501296997070312, "learning_rate": 1.7907937368240525e-06, "loss": 0.09653893113136292, "step": 30840 }, { "epoch": 0.8126360292786045, "grad_norm": 3.7433550357818604, "learning_rate": 1.7859249501529742e-06, "loss": 0.1152353286743164, "step": 30850 }, { "epoch": 0.8128994445231031, "grad_norm": 3.109346866607666, "learning_rate": 1.7810621420401054e-06, "loss": 0.10734360218048096, "step": 30860 }, { "epoch": 0.8131628597676019, "grad_norm": 1.8859317302703857, "learning_rate": 1.776205316024806e-06, "loss": 0.09422690868377685, "step": 30870 }, { "epoch": 0.8134262750121006, "grad_norm": 2.3608527183532715, "learning_rate": 1.7713544756420785e-06, "loss": 0.0962480366230011, "step": 30880 }, { "epoch": 0.8136896902565994, "grad_norm": 2.1877031326293945, "learning_rate": 1.76650962442258e-06, "loss": 0.09843153953552246, "step": 30890 }, { "epoch": 0.8139531055010981, "grad_norm": 4.585300445556641, "learning_rate": 1.761670765892599e-06, "loss": 0.1119913101196289, "step": 30900 }, { "epoch": 0.8142165207455968, "grad_norm": 2.5552656650543213, "learning_rate": 1.7568379035740669e-06, "loss": 0.10763862133026122, "step": 30910 }, { "epoch": 0.8144799359900956, "grad_norm": 2.5389935970306396, "learning_rate": 1.7520110409845493e-06, "loss": 0.1003667950630188, "step": 30920 }, { "epoch": 0.8147433512345943, "grad_norm": 3.862097978591919, "learning_rate": 1.7471901816372416e-06, "loss": 0.10624641180038452, "step": 30930 }, { "epoch": 0.8150067664790931, "grad_norm": 3.4884121417999268, "learning_rate": 1.7423753290409784e-06, "loss": 0.1018014669418335, "step": 30940 }, { "epoch": 0.8152701817235918, "grad_norm": 4.052643775939941, "learning_rate": 1.7375664867002151e-06, "loss": 0.10298426151275634, "step": 30950 }, { "epoch": 0.8155335969680906, "grad_norm": 1.6789445877075195, "learning_rate": 1.7327636581150343e-06, "loss": 0.10444581508636475, "step": 30960 }, { "epoch": 0.8157970122125893, "grad_norm": 2.963805913925171, "learning_rate": 1.7279668467811427e-06, "loss": 0.09996403455734253, "step": 30970 }, { "epoch": 0.816060427457088, "grad_norm": 3.9829533100128174, "learning_rate": 1.7231760561898625e-06, "loss": 0.10953686237335206, "step": 30980 }, { "epoch": 0.8163238427015868, "grad_norm": 2.9973809719085693, "learning_rate": 1.718391289828144e-06, "loss": 0.1005103349685669, "step": 30990 }, { "epoch": 0.8165872579460854, "grad_norm": 2.1631224155426025, "learning_rate": 1.7136125511785462e-06, "loss": 0.09374960660934448, "step": 31000 }, { "epoch": 0.8168506731905842, "grad_norm": 2.3718512058258057, "learning_rate": 1.7088398437192344e-06, "loss": 0.10499078035354614, "step": 31010 }, { "epoch": 0.8171140884350829, "grad_norm": 2.3474104404449463, "learning_rate": 1.7040731709239967e-06, "loss": 0.1049435019493103, "step": 31020 }, { "epoch": 0.8173775036795817, "grad_norm": 2.8945295810699463, "learning_rate": 1.6993125362622231e-06, "loss": 0.09264531135559081, "step": 31030 }, { "epoch": 0.8176409189240804, "grad_norm": 2.551286220550537, "learning_rate": 1.6945579431989067e-06, "loss": 0.1019892692565918, "step": 31040 }, { "epoch": 0.8179043341685792, "grad_norm": 2.0242884159088135, "learning_rate": 1.689809395194647e-06, "loss": 0.1009521722793579, "step": 31050 }, { "epoch": 0.8181677494130779, "grad_norm": 3.6750895977020264, "learning_rate": 1.6850668957056393e-06, "loss": 0.1092991590499878, "step": 31060 }, { "epoch": 0.8184311646575766, "grad_norm": 3.2364299297332764, "learning_rate": 1.6803304481836835e-06, "loss": 0.10060317516326904, "step": 31070 }, { "epoch": 0.8186945799020754, "grad_norm": 3.0883960723876953, "learning_rate": 1.6756000560761687e-06, "loss": 0.0921593189239502, "step": 31080 }, { "epoch": 0.8189579951465741, "grad_norm": 2.490187406539917, "learning_rate": 1.6708757228260786e-06, "loss": 0.10444489717483521, "step": 31090 }, { "epoch": 0.8192214103910729, "grad_norm": 2.6539480686187744, "learning_rate": 1.6661574518719859e-06, "loss": 0.09342672228813172, "step": 31100 }, { "epoch": 0.8194848256355716, "grad_norm": 2.8205771446228027, "learning_rate": 1.6614452466480502e-06, "loss": 0.09440644979476928, "step": 31110 }, { "epoch": 0.8197482408800704, "grad_norm": 2.75632643699646, "learning_rate": 1.6567391105840247e-06, "loss": 0.09497983455657959, "step": 31120 }, { "epoch": 0.820011656124569, "grad_norm": 3.1307032108306885, "learning_rate": 1.6520390471052306e-06, "loss": 0.08957209587097167, "step": 31130 }, { "epoch": 0.8202750713690679, "grad_norm": 2.0633785724639893, "learning_rate": 1.647345059632578e-06, "loss": 0.08848430514335633, "step": 31140 }, { "epoch": 0.8205384866135665, "grad_norm": 3.5871522426605225, "learning_rate": 1.6426571515825563e-06, "loss": 0.10404777526855469, "step": 31150 }, { "epoch": 0.8208019018580652, "grad_norm": 3.130476236343384, "learning_rate": 1.6379753263672272e-06, "loss": 0.10503010749816895, "step": 31160 }, { "epoch": 0.821065317102564, "grad_norm": 3.0473546981811523, "learning_rate": 1.6332995873942237e-06, "loss": 0.0991775393486023, "step": 31170 }, { "epoch": 0.8213287323470627, "grad_norm": 3.369915008544922, "learning_rate": 1.6286299380667504e-06, "loss": 0.08790453672409057, "step": 31180 }, { "epoch": 0.8215921475915615, "grad_norm": 3.5597012042999268, "learning_rate": 1.6239663817835783e-06, "loss": 0.10164046287536621, "step": 31190 }, { "epoch": 0.8218555628360602, "grad_norm": 3.1603634357452393, "learning_rate": 1.6193089219390478e-06, "loss": 0.11233987808227539, "step": 31200 }, { "epoch": 0.822118978080559, "grad_norm": 2.66412615776062, "learning_rate": 1.6146575619230597e-06, "loss": 0.10114245414733887, "step": 31210 }, { "epoch": 0.8223823933250577, "grad_norm": 2.824716329574585, "learning_rate": 1.610012305121068e-06, "loss": 0.11164848804473877, "step": 31220 }, { "epoch": 0.8226458085695565, "grad_norm": 2.4790732860565186, "learning_rate": 1.6053731549140982e-06, "loss": 0.11426855325698852, "step": 31230 }, { "epoch": 0.8229092238140552, "grad_norm": 3.1557395458221436, "learning_rate": 1.6007401146787195e-06, "loss": 0.10166184902191162, "step": 31240 }, { "epoch": 0.8231726390585539, "grad_norm": 2.838250160217285, "learning_rate": 1.596113187787064e-06, "loss": 0.09579348564147949, "step": 31250 }, { "epoch": 0.8234360543030527, "grad_norm": 3.055900812149048, "learning_rate": 1.5914923776068025e-06, "loss": 0.09884945154190064, "step": 31260 }, { "epoch": 0.8236994695475514, "grad_norm": 2.6006739139556885, "learning_rate": 1.5868776875011604e-06, "loss": 0.09698261022567749, "step": 31270 }, { "epoch": 0.8239628847920502, "grad_norm": 3.1110403537750244, "learning_rate": 1.582269120828912e-06, "loss": 0.11328785419464112, "step": 31280 }, { "epoch": 0.8242263000365488, "grad_norm": 2.0877957344055176, "learning_rate": 1.5776666809443697e-06, "loss": 0.08865755796432495, "step": 31290 }, { "epoch": 0.8244897152810476, "grad_norm": 2.48775577545166, "learning_rate": 1.5730703711973883e-06, "loss": 0.09703344702720643, "step": 31300 }, { "epoch": 0.8247531305255463, "grad_norm": 2.1470913887023926, "learning_rate": 1.5684801949333584e-06, "loss": 0.10360528230667114, "step": 31310 }, { "epoch": 0.8250165457700451, "grad_norm": 3.287182092666626, "learning_rate": 1.5638961554932064e-06, "loss": 0.11953060626983643, "step": 31320 }, { "epoch": 0.8252799610145438, "grad_norm": 3.4582486152648926, "learning_rate": 1.5593182562134002e-06, "loss": 0.09870579242706298, "step": 31330 }, { "epoch": 0.8255433762590425, "grad_norm": 2.9928882122039795, "learning_rate": 1.554746500425931e-06, "loss": 0.09011585116386414, "step": 31340 }, { "epoch": 0.8258067915035413, "grad_norm": 2.06645131111145, "learning_rate": 1.5501808914583139e-06, "loss": 0.10362417697906494, "step": 31350 }, { "epoch": 0.82607020674804, "grad_norm": 2.4803879261016846, "learning_rate": 1.5456214326336017e-06, "loss": 0.1041306734085083, "step": 31360 }, { "epoch": 0.8263336219925388, "grad_norm": 2.6155788898468018, "learning_rate": 1.5410681272703643e-06, "loss": 0.0959934949874878, "step": 31370 }, { "epoch": 0.8265970372370375, "grad_norm": 2.1442482471466064, "learning_rate": 1.536520978682693e-06, "loss": 0.09762048721313477, "step": 31380 }, { "epoch": 0.8268604524815363, "grad_norm": 2.07627534866333, "learning_rate": 1.531979990180199e-06, "loss": 0.08684511184692383, "step": 31390 }, { "epoch": 0.827123867726035, "grad_norm": 2.548261880874634, "learning_rate": 1.5274451650680066e-06, "loss": 0.11228606700897217, "step": 31400 }, { "epoch": 0.8273872829705338, "grad_norm": 2.72992205619812, "learning_rate": 1.5229165066467632e-06, "loss": 0.10080177783966064, "step": 31410 }, { "epoch": 0.8276506982150325, "grad_norm": 2.3708994388580322, "learning_rate": 1.5183940182126189e-06, "loss": 0.09733443856239318, "step": 31420 }, { "epoch": 0.8279141134595311, "grad_norm": 2.7469894886016846, "learning_rate": 1.513877703057235e-06, "loss": 0.10048518180847169, "step": 31430 }, { "epoch": 0.8281775287040299, "grad_norm": 2.181967258453369, "learning_rate": 1.509367564467782e-06, "loss": 0.10275371074676513, "step": 31440 }, { "epoch": 0.8284409439485286, "grad_norm": 2.8121204376220703, "learning_rate": 1.5048636057269294e-06, "loss": 0.11017169952392578, "step": 31450 }, { "epoch": 0.8287043591930274, "grad_norm": 2.1310009956359863, "learning_rate": 1.50036583011286e-06, "loss": 0.1039198875427246, "step": 31460 }, { "epoch": 0.8289677744375261, "grad_norm": 2.496077299118042, "learning_rate": 1.4958742408992454e-06, "loss": 0.10383485555648804, "step": 31470 }, { "epoch": 0.8292311896820249, "grad_norm": 2.38124418258667, "learning_rate": 1.4913888413552536e-06, "loss": 0.09052261114120483, "step": 31480 }, { "epoch": 0.8294946049265236, "grad_norm": 3.4932007789611816, "learning_rate": 1.486909634745557e-06, "loss": 0.09234562516212463, "step": 31490 }, { "epoch": 0.8297580201710223, "grad_norm": 2.4057157039642334, "learning_rate": 1.4824366243303156e-06, "loss": 0.0981772780418396, "step": 31500 }, { "epoch": 0.8300214354155211, "grad_norm": 2.1400814056396484, "learning_rate": 1.4779698133651766e-06, "loss": 0.10066416263580322, "step": 31510 }, { "epoch": 0.8302848506600198, "grad_norm": 3.012725353240967, "learning_rate": 1.47350920510128e-06, "loss": 0.09747887849807739, "step": 31520 }, { "epoch": 0.8305482659045186, "grad_norm": 2.7900829315185547, "learning_rate": 1.469054802785246e-06, "loss": 0.10526881217956544, "step": 31530 }, { "epoch": 0.8308116811490173, "grad_norm": 3.6693363189697266, "learning_rate": 1.4646066096591838e-06, "loss": 0.10853846073150634, "step": 31540 }, { "epoch": 0.8310750963935161, "grad_norm": 3.0055880546569824, "learning_rate": 1.4601646289606809e-06, "loss": 0.10415226221084595, "step": 31550 }, { "epoch": 0.8313385116380148, "grad_norm": 2.837575674057007, "learning_rate": 1.4557288639227995e-06, "loss": 0.08840492367744446, "step": 31560 }, { "epoch": 0.8316019268825136, "grad_norm": 2.754716157913208, "learning_rate": 1.451299317774083e-06, "loss": 0.10289350748062134, "step": 31570 }, { "epoch": 0.8318653421270122, "grad_norm": 3.357877254486084, "learning_rate": 1.4468759937385423e-06, "loss": 0.09659180641174317, "step": 31580 }, { "epoch": 0.8321287573715109, "grad_norm": 3.250744104385376, "learning_rate": 1.4424588950356688e-06, "loss": 0.09317456483840943, "step": 31590 }, { "epoch": 0.8323921726160097, "grad_norm": 2.594719648361206, "learning_rate": 1.4380480248804185e-06, "loss": 0.1243171215057373, "step": 31600 }, { "epoch": 0.8326555878605084, "grad_norm": 3.116384267807007, "learning_rate": 1.4336433864832033e-06, "loss": 0.10725511312484741, "step": 31610 }, { "epoch": 0.8329190031050072, "grad_norm": 3.4483346939086914, "learning_rate": 1.4292449830499189e-06, "loss": 0.10217146873474121, "step": 31620 }, { "epoch": 0.8331824183495059, "grad_norm": 2.31672739982605, "learning_rate": 1.4248528177819098e-06, "loss": 0.10112659931182862, "step": 31630 }, { "epoch": 0.8334458335940047, "grad_norm": 4.164510726928711, "learning_rate": 1.4204668938759835e-06, "loss": 0.09262876510620117, "step": 31640 }, { "epoch": 0.8337092488385034, "grad_norm": 3.5817055702209473, "learning_rate": 1.416087214524403e-06, "loss": 0.10295970439910888, "step": 31650 }, { "epoch": 0.8339726640830022, "grad_norm": 2.7480204105377197, "learning_rate": 1.4117137829148885e-06, "loss": 0.09637471437454223, "step": 31660 }, { "epoch": 0.8342360793275009, "grad_norm": 2.1775028705596924, "learning_rate": 1.4073466022306148e-06, "loss": 0.09313619136810303, "step": 31670 }, { "epoch": 0.8344994945719996, "grad_norm": 2.6742095947265625, "learning_rate": 1.4029856756502026e-06, "loss": 0.10195503234863282, "step": 31680 }, { "epoch": 0.8347629098164984, "grad_norm": 2.7346208095550537, "learning_rate": 1.3986310063477215e-06, "loss": 0.1003653883934021, "step": 31690 }, { "epoch": 0.835026325060997, "grad_norm": 2.8415145874023438, "learning_rate": 1.3942825974926888e-06, "loss": 0.10103813409805298, "step": 31700 }, { "epoch": 0.8352897403054959, "grad_norm": 2.7581284046173096, "learning_rate": 1.3899404522500626e-06, "loss": 0.09744948744773865, "step": 31710 }, { "epoch": 0.8355531555499945, "grad_norm": 3.2209293842315674, "learning_rate": 1.385604573780248e-06, "loss": 0.08730711340904236, "step": 31720 }, { "epoch": 0.8358165707944933, "grad_norm": 2.6793642044067383, "learning_rate": 1.38127496523908e-06, "loss": 0.10303528308868408, "step": 31730 }, { "epoch": 0.836079986038992, "grad_norm": 2.9000492095947266, "learning_rate": 1.376951629777833e-06, "loss": 0.09354376196861267, "step": 31740 }, { "epoch": 0.8363434012834908, "grad_norm": 2.8113338947296143, "learning_rate": 1.372634570543222e-06, "loss": 0.10353108644485473, "step": 31750 }, { "epoch": 0.8366068165279895, "grad_norm": 2.3229594230651855, "learning_rate": 1.3683237906773873e-06, "loss": 0.10174623727798462, "step": 31760 }, { "epoch": 0.8368702317724882, "grad_norm": 2.3273282051086426, "learning_rate": 1.3640192933178997e-06, "loss": 0.09373795986175537, "step": 31770 }, { "epoch": 0.837133647016987, "grad_norm": 2.523293972015381, "learning_rate": 1.359721081597758e-06, "loss": 0.10426838397979736, "step": 31780 }, { "epoch": 0.8373970622614857, "grad_norm": 2.929839611053467, "learning_rate": 1.355429158645384e-06, "loss": 0.09247338771820068, "step": 31790 }, { "epoch": 0.8376604775059845, "grad_norm": 2.9739394187927246, "learning_rate": 1.3511435275846297e-06, "loss": 0.11776881217956543, "step": 31800 }, { "epoch": 0.8379238927504832, "grad_norm": 4.622543811798096, "learning_rate": 1.3468641915347613e-06, "loss": 0.09650975465774536, "step": 31810 }, { "epoch": 0.838187307994982, "grad_norm": 3.7516400814056396, "learning_rate": 1.3425911536104573e-06, "loss": 0.09294931888580323, "step": 31820 }, { "epoch": 0.8384507232394807, "grad_norm": 2.379699945449829, "learning_rate": 1.3383244169218268e-06, "loss": 0.09486672282218933, "step": 31830 }, { "epoch": 0.8387141384839795, "grad_norm": 3.552319049835205, "learning_rate": 1.3340639845743797e-06, "loss": 0.10537576675415039, "step": 31840 }, { "epoch": 0.8389775537284782, "grad_norm": 3.4632225036621094, "learning_rate": 1.32980985966905e-06, "loss": 0.09679782986640931, "step": 31850 }, { "epoch": 0.8392409689729768, "grad_norm": 2.930457830429077, "learning_rate": 1.3255620453021656e-06, "loss": 0.10558923482894897, "step": 31860 }, { "epoch": 0.8395043842174756, "grad_norm": 3.947465658187866, "learning_rate": 1.3213205445654688e-06, "loss": 0.1034471869468689, "step": 31870 }, { "epoch": 0.8397677994619743, "grad_norm": 2.5698606967926025, "learning_rate": 1.3170853605461132e-06, "loss": 0.10485668182373047, "step": 31880 }, { "epoch": 0.8400312147064731, "grad_norm": 3.6404857635498047, "learning_rate": 1.3128564963266444e-06, "loss": 0.10190834999084472, "step": 31890 }, { "epoch": 0.8402946299509718, "grad_norm": 4.308140754699707, "learning_rate": 1.3086339549850125e-06, "loss": 0.1023023247718811, "step": 31900 }, { "epoch": 0.8405580451954706, "grad_norm": 2.7963407039642334, "learning_rate": 1.3044177395945657e-06, "loss": 0.09297573566436768, "step": 31910 }, { "epoch": 0.8408214604399693, "grad_norm": 2.307206630706787, "learning_rate": 1.3002078532240447e-06, "loss": 0.09723693132400513, "step": 31920 }, { "epoch": 0.8410848756844681, "grad_norm": 2.293464422225952, "learning_rate": 1.296004298937592e-06, "loss": 0.10110684633255004, "step": 31930 }, { "epoch": 0.8413482909289668, "grad_norm": 2.347639322280884, "learning_rate": 1.2918070797947346e-06, "loss": 0.08641494512557983, "step": 31940 }, { "epoch": 0.8416117061734655, "grad_norm": 2.5382087230682373, "learning_rate": 1.2876161988503843e-06, "loss": 0.10684244632720948, "step": 31950 }, { "epoch": 0.8418751214179643, "grad_norm": 3.074571371078491, "learning_rate": 1.2834316591548502e-06, "loss": 0.10367840528488159, "step": 31960 }, { "epoch": 0.842138536662463, "grad_norm": 3.5813732147216797, "learning_rate": 1.279253463753819e-06, "loss": 0.10456660985946656, "step": 31970 }, { "epoch": 0.8424019519069618, "grad_norm": 2.5340733528137207, "learning_rate": 1.2750816156883627e-06, "loss": 0.09715067148208618, "step": 31980 }, { "epoch": 0.8426653671514605, "grad_norm": 3.1938464641571045, "learning_rate": 1.2709161179949303e-06, "loss": 0.09390976428985595, "step": 31990 }, { "epoch": 0.8429287823959593, "grad_norm": 3.123161792755127, "learning_rate": 1.266756973705351e-06, "loss": 0.0966241180896759, "step": 32000 }, { "epoch": 0.8431921976404579, "grad_norm": 2.3354361057281494, "learning_rate": 1.2626041858468318e-06, "loss": 0.097409588098526, "step": 32010 }, { "epoch": 0.8434556128849566, "grad_norm": 3.052126884460449, "learning_rate": 1.2584577574419499e-06, "loss": 0.09553653001785278, "step": 32020 }, { "epoch": 0.8437190281294554, "grad_norm": 2.473496913909912, "learning_rate": 1.2543176915086542e-06, "loss": 0.10930265188217163, "step": 32030 }, { "epoch": 0.8439824433739541, "grad_norm": 3.4798264503479004, "learning_rate": 1.2501839910602632e-06, "loss": 0.09148523807525635, "step": 32040 }, { "epoch": 0.8442458586184529, "grad_norm": 3.407564640045166, "learning_rate": 1.24605665910546e-06, "loss": 0.10897283554077149, "step": 32050 }, { "epoch": 0.8445092738629516, "grad_norm": 3.075438976287842, "learning_rate": 1.2419356986483e-06, "loss": 0.09628790616989136, "step": 32060 }, { "epoch": 0.8447726891074504, "grad_norm": 2.884411096572876, "learning_rate": 1.2378211126881956e-06, "loss": 0.10152907371520996, "step": 32070 }, { "epoch": 0.8450361043519491, "grad_norm": 2.965564727783203, "learning_rate": 1.2337129042199147e-06, "loss": 0.09770654439926148, "step": 32080 }, { "epoch": 0.8452995195964479, "grad_norm": 2.6177046298980713, "learning_rate": 1.2296110762335933e-06, "loss": 0.08548307418823242, "step": 32090 }, { "epoch": 0.8455629348409466, "grad_norm": 3.460092067718506, "learning_rate": 1.2255156317147165e-06, "loss": 0.08670414090156556, "step": 32100 }, { "epoch": 0.8458263500854453, "grad_norm": 2.8399932384490967, "learning_rate": 1.2214265736441277e-06, "loss": 0.08992614150047303, "step": 32110 }, { "epoch": 0.8460897653299441, "grad_norm": 3.0949032306671143, "learning_rate": 1.2173439049980174e-06, "loss": 0.09587855339050293, "step": 32120 }, { "epoch": 0.8463531805744428, "grad_norm": 3.495741605758667, "learning_rate": 1.2132676287479273e-06, "loss": 0.09526111483573914, "step": 32130 }, { "epoch": 0.8466165958189416, "grad_norm": 3.6453373432159424, "learning_rate": 1.209197747860752e-06, "loss": 0.0948829174041748, "step": 32140 }, { "epoch": 0.8468800110634402, "grad_norm": 3.276627540588379, "learning_rate": 1.2051342652987218e-06, "loss": 0.09603761434555054, "step": 32150 }, { "epoch": 0.847143426307939, "grad_norm": 3.1106505393981934, "learning_rate": 1.2010771840194168e-06, "loss": 0.09287630915641784, "step": 32160 }, { "epoch": 0.8474068415524377, "grad_norm": 2.6881425380706787, "learning_rate": 1.1970265069757548e-06, "loss": 0.1033096194267273, "step": 32170 }, { "epoch": 0.8476702567969365, "grad_norm": 3.161653995513916, "learning_rate": 1.1929822371159915e-06, "loss": 0.09839882850646972, "step": 32180 }, { "epoch": 0.8479336720414352, "grad_norm": 2.2989633083343506, "learning_rate": 1.188944377383725e-06, "loss": 0.1026153564453125, "step": 32190 }, { "epoch": 0.8481970872859339, "grad_norm": 2.164799690246582, "learning_rate": 1.1849129307178841e-06, "loss": 0.08761952519416809, "step": 32200 }, { "epoch": 0.8484605025304327, "grad_norm": 2.3179595470428467, "learning_rate": 1.180887900052724e-06, "loss": 0.09566687345504761, "step": 32210 }, { "epoch": 0.8487239177749314, "grad_norm": 2.5638089179992676, "learning_rate": 1.17686928831784e-06, "loss": 0.10208464860916137, "step": 32220 }, { "epoch": 0.8489873330194302, "grad_norm": 2.7768349647521973, "learning_rate": 1.1728570984381505e-06, "loss": 0.09999068975448608, "step": 32230 }, { "epoch": 0.8492507482639289, "grad_norm": 2.7774455547332764, "learning_rate": 1.1688513333339003e-06, "loss": 0.10092257261276245, "step": 32240 }, { "epoch": 0.8495141635084277, "grad_norm": 3.0750656127929688, "learning_rate": 1.1648519959206584e-06, "loss": 0.08827012777328491, "step": 32250 }, { "epoch": 0.8497775787529264, "grad_norm": 3.7494544982910156, "learning_rate": 1.1608590891093118e-06, "loss": 0.09237549304962159, "step": 32260 }, { "epoch": 0.8500409939974252, "grad_norm": 2.796520948410034, "learning_rate": 1.156872615806075e-06, "loss": 0.09784605503082275, "step": 32270 }, { "epoch": 0.8503044092419239, "grad_norm": 2.744584798812866, "learning_rate": 1.1528925789124734e-06, "loss": 0.09736354351043701, "step": 32280 }, { "epoch": 0.8505678244864225, "grad_norm": 2.2028770446777344, "learning_rate": 1.1489189813253499e-06, "loss": 0.09503034353256226, "step": 32290 }, { "epoch": 0.8508312397309213, "grad_norm": 2.4877216815948486, "learning_rate": 1.1449518259368596e-06, "loss": 0.08983701467514038, "step": 32300 }, { "epoch": 0.85109465497542, "grad_norm": 2.952904462814331, "learning_rate": 1.1409911156344677e-06, "loss": 0.09558749794960023, "step": 32310 }, { "epoch": 0.8513580702199188, "grad_norm": 4.250616073608398, "learning_rate": 1.1370368533009557e-06, "loss": 0.09060054421424865, "step": 32320 }, { "epoch": 0.8516214854644175, "grad_norm": 2.5171709060668945, "learning_rate": 1.1330890418144014e-06, "loss": 0.09205657243728638, "step": 32330 }, { "epoch": 0.8518849007089163, "grad_norm": 4.067124366760254, "learning_rate": 1.1291476840481908e-06, "loss": 0.10266202688217163, "step": 32340 }, { "epoch": 0.852148315953415, "grad_norm": 3.9704246520996094, "learning_rate": 1.125212782871019e-06, "loss": 0.09467578530311585, "step": 32350 }, { "epoch": 0.8524117311979138, "grad_norm": 3.64198899269104, "learning_rate": 1.1212843411468755e-06, "loss": 0.09568746089935302, "step": 32360 }, { "epoch": 0.8526751464424125, "grad_norm": 3.045559883117676, "learning_rate": 1.1173623617350472e-06, "loss": 0.09182089567184448, "step": 32370 }, { "epoch": 0.8529385616869112, "grad_norm": 2.6892495155334473, "learning_rate": 1.1134468474901227e-06, "loss": 0.09767194986343383, "step": 32380 }, { "epoch": 0.85320197693141, "grad_norm": 3.637403726577759, "learning_rate": 1.1095378012619796e-06, "loss": 0.09306480288505554, "step": 32390 }, { "epoch": 0.8534653921759087, "grad_norm": 2.5590133666992188, "learning_rate": 1.1056352258957924e-06, "loss": 0.08390724658966064, "step": 32400 }, { "epoch": 0.8537288074204075, "grad_norm": 3.050424814224243, "learning_rate": 1.1017391242320263e-06, "loss": 0.0932813048362732, "step": 32410 }, { "epoch": 0.8539922226649062, "grad_norm": 3.449057102203369, "learning_rate": 1.0978494991064248e-06, "loss": 0.1005234956741333, "step": 32420 }, { "epoch": 0.854255637909405, "grad_norm": 2.747753381729126, "learning_rate": 1.0939663533500311e-06, "loss": 0.09148759841918945, "step": 32430 }, { "epoch": 0.8545190531539036, "grad_norm": 2.2850589752197266, "learning_rate": 1.0900896897891634e-06, "loss": 0.09877142906188965, "step": 32440 }, { "epoch": 0.8547824683984024, "grad_norm": 4.0545196533203125, "learning_rate": 1.0862195112454289e-06, "loss": 0.093322491645813, "step": 32450 }, { "epoch": 0.8550458836429011, "grad_norm": 2.6945109367370605, "learning_rate": 1.0823558205357054e-06, "loss": 0.0885776162147522, "step": 32460 }, { "epoch": 0.8553092988873998, "grad_norm": 3.889582872390747, "learning_rate": 1.0784986204721537e-06, "loss": 0.10043749809265137, "step": 32470 }, { "epoch": 0.8555727141318986, "grad_norm": 4.7545881271362305, "learning_rate": 1.0746479138622134e-06, "loss": 0.10133559703826904, "step": 32480 }, { "epoch": 0.8558361293763973, "grad_norm": 2.8051345348358154, "learning_rate": 1.070803703508595e-06, "loss": 0.10720604658126831, "step": 32490 }, { "epoch": 0.8560995446208961, "grad_norm": 2.9621422290802, "learning_rate": 1.066965992209279e-06, "loss": 0.09037641286849976, "step": 32500 }, { "epoch": 0.8563629598653948, "grad_norm": 3.2146759033203125, "learning_rate": 1.0631347827575179e-06, "loss": 0.09018529653549194, "step": 32510 }, { "epoch": 0.8566263751098936, "grad_norm": 4.278082847595215, "learning_rate": 1.0593100779418286e-06, "loss": 0.10064477920532226, "step": 32520 }, { "epoch": 0.8568897903543923, "grad_norm": 3.4601659774780273, "learning_rate": 1.0554918805460013e-06, "loss": 0.10758099555969239, "step": 32530 }, { "epoch": 0.8571532055988911, "grad_norm": 2.9993104934692383, "learning_rate": 1.051680193349084e-06, "loss": 0.08933628797531128, "step": 32540 }, { "epoch": 0.8574166208433898, "grad_norm": 2.8683714866638184, "learning_rate": 1.0478750191253817e-06, "loss": 0.09918599724769592, "step": 32550 }, { "epoch": 0.8576800360878885, "grad_norm": 3.360795497894287, "learning_rate": 1.0440763606444694e-06, "loss": 0.09154472351074219, "step": 32560 }, { "epoch": 0.8579434513323873, "grad_norm": 2.6390256881713867, "learning_rate": 1.0402842206711728e-06, "loss": 0.10961036682128907, "step": 32570 }, { "epoch": 0.8582068665768859, "grad_norm": 5.154048919677734, "learning_rate": 1.0364986019655765e-06, "loss": 0.0985318124294281, "step": 32580 }, { "epoch": 0.8584702818213847, "grad_norm": 2.765957832336426, "learning_rate": 1.0327195072830153e-06, "loss": 0.10430132150650025, "step": 32590 }, { "epoch": 0.8587336970658834, "grad_norm": 3.4682202339172363, "learning_rate": 1.028946939374076e-06, "loss": 0.09897711277008056, "step": 32600 }, { "epoch": 0.8589971123103822, "grad_norm": 3.124267578125, "learning_rate": 1.0251809009846003e-06, "loss": 0.1007282018661499, "step": 32610 }, { "epoch": 0.8592605275548809, "grad_norm": 2.4952609539031982, "learning_rate": 1.0214213948556718e-06, "loss": 0.0782744824886322, "step": 32620 }, { "epoch": 0.8595239427993796, "grad_norm": 2.784400701522827, "learning_rate": 1.0176684237236212e-06, "loss": 0.10003726482391358, "step": 32630 }, { "epoch": 0.8597873580438784, "grad_norm": 3.235421895980835, "learning_rate": 1.0139219903200227e-06, "loss": 0.09830948114395141, "step": 32640 }, { "epoch": 0.8600507732883771, "grad_norm": 3.3138840198516846, "learning_rate": 1.0101820973716891e-06, "loss": 0.11063623428344727, "step": 32650 }, { "epoch": 0.8603141885328759, "grad_norm": 4.109111309051514, "learning_rate": 1.0064487476006812e-06, "loss": 0.09431079626083375, "step": 32660 }, { "epoch": 0.8605776037773746, "grad_norm": 2.5596723556518555, "learning_rate": 1.0027219437242907e-06, "loss": 0.09403108358383179, "step": 32670 }, { "epoch": 0.8608410190218734, "grad_norm": 3.5394387245178223, "learning_rate": 9.990016884550425e-07, "loss": 0.09399282932281494, "step": 32680 }, { "epoch": 0.8611044342663721, "grad_norm": 3.1608877182006836, "learning_rate": 9.952879845007024e-07, "loss": 0.09063210487365722, "step": 32690 }, { "epoch": 0.8613678495108709, "grad_norm": 4.103646278381348, "learning_rate": 9.915808345642642e-07, "loss": 0.08589286804199218, "step": 32700 }, { "epoch": 0.8616312647553696, "grad_norm": 3.354959011077881, "learning_rate": 9.878802413439514e-07, "loss": 0.09053229093551636, "step": 32710 }, { "epoch": 0.8618946799998682, "grad_norm": 3.588238477706909, "learning_rate": 9.841862075332143e-07, "loss": 0.09863103628158569, "step": 32720 }, { "epoch": 0.862158095244367, "grad_norm": 2.5255398750305176, "learning_rate": 9.804987358207274e-07, "loss": 0.08136323690414429, "step": 32730 }, { "epoch": 0.8624215104888657, "grad_norm": 3.6075046062469482, "learning_rate": 9.768178288903985e-07, "loss": 0.09846740961074829, "step": 32740 }, { "epoch": 0.8626849257333645, "grad_norm": 3.258514642715454, "learning_rate": 9.731434894213465e-07, "loss": 0.09619115591049195, "step": 32750 }, { "epoch": 0.8629483409778632, "grad_norm": 3.1102330684661865, "learning_rate": 9.694757200879146e-07, "loss": 0.10208790302276612, "step": 32760 }, { "epoch": 0.863211756222362, "grad_norm": 2.589855670928955, "learning_rate": 9.658145235596628e-07, "loss": 0.09742194414138794, "step": 32770 }, { "epoch": 0.8634751714668607, "grad_norm": 4.626916408538818, "learning_rate": 9.621599025013672e-07, "loss": 0.09794178009033203, "step": 32780 }, { "epoch": 0.8637385867113595, "grad_norm": 2.8832106590270996, "learning_rate": 9.585118595730226e-07, "loss": 0.09953277707099914, "step": 32790 }, { "epoch": 0.8640020019558582, "grad_norm": 3.419935464859009, "learning_rate": 9.548703974298311e-07, "loss": 0.09181653261184693, "step": 32800 }, { "epoch": 0.8642654172003569, "grad_norm": 3.1691689491271973, "learning_rate": 9.512355187222e-07, "loss": 0.09524131417274476, "step": 32810 }, { "epoch": 0.8645288324448557, "grad_norm": 2.405031681060791, "learning_rate": 9.47607226095758e-07, "loss": 0.0942453920841217, "step": 32820 }, { "epoch": 0.8647922476893544, "grad_norm": 4.489805698394775, "learning_rate": 9.439855221913297e-07, "loss": 0.1004828929901123, "step": 32830 }, { "epoch": 0.8650556629338532, "grad_norm": 3.552455425262451, "learning_rate": 9.403704096449484e-07, "loss": 0.08998320102691651, "step": 32840 }, { "epoch": 0.8653190781783519, "grad_norm": 3.3587889671325684, "learning_rate": 9.367618910878484e-07, "loss": 0.09884260892868042, "step": 32850 }, { "epoch": 0.8655824934228507, "grad_norm": 3.2108242511749268, "learning_rate": 9.331599691464644e-07, "loss": 0.10096892118453979, "step": 32860 }, { "epoch": 0.8658459086673493, "grad_norm": 2.907284736633301, "learning_rate": 9.295646464424323e-07, "loss": 0.08884146809577942, "step": 32870 }, { "epoch": 0.8661093239118481, "grad_norm": 4.2629804611206055, "learning_rate": 9.259759255925827e-07, "loss": 0.0881452739238739, "step": 32880 }, { "epoch": 0.8663727391563468, "grad_norm": 4.2284746170043945, "learning_rate": 9.223938092089413e-07, "loss": 0.09875713586807251, "step": 32890 }, { "epoch": 0.8666361544008455, "grad_norm": 3.3294177055358887, "learning_rate": 9.188182998987261e-07, "loss": 0.09521440267562867, "step": 32900 }, { "epoch": 0.8668995696453443, "grad_norm": 3.1175029277801514, "learning_rate": 9.152494002643464e-07, "loss": 0.09293731451034545, "step": 32910 }, { "epoch": 0.867162984889843, "grad_norm": 3.127610445022583, "learning_rate": 9.116871129034066e-07, "loss": 0.10393292903900146, "step": 32920 }, { "epoch": 0.8674264001343418, "grad_norm": 2.529930353164673, "learning_rate": 9.081314404086872e-07, "loss": 0.09163670539855957, "step": 32930 }, { "epoch": 0.8676898153788405, "grad_norm": 3.4980156421661377, "learning_rate": 9.045823853681602e-07, "loss": 0.09750587940216064, "step": 32940 }, { "epoch": 0.8679532306233393, "grad_norm": 2.854724407196045, "learning_rate": 9.010399503649836e-07, "loss": 0.09310706853866577, "step": 32950 }, { "epoch": 0.868216645867838, "grad_norm": 3.842050552368164, "learning_rate": 8.975041379774951e-07, "loss": 0.10486527681350707, "step": 32960 }, { "epoch": 0.8684800611123368, "grad_norm": 3.6087090969085693, "learning_rate": 8.939749507792094e-07, "loss": 0.09951534271240234, "step": 32970 }, { "epoch": 0.8687434763568355, "grad_norm": 2.890326499938965, "learning_rate": 8.904523913388218e-07, "loss": 0.09398296475410461, "step": 32980 }, { "epoch": 0.8690068916013342, "grad_norm": 2.5978307723999023, "learning_rate": 8.869364622202015e-07, "loss": 0.09099581241607665, "step": 32990 }, { "epoch": 0.869270306845833, "grad_norm": 3.134243965148926, "learning_rate": 8.834271659823967e-07, "loss": 0.0951198697090149, "step": 33000 }, { "epoch": 0.8695337220903316, "grad_norm": 2.99690842628479, "learning_rate": 8.799245051796246e-07, "loss": 0.09096696376800537, "step": 33010 }, { "epoch": 0.8697971373348304, "grad_norm": 2.2944250106811523, "learning_rate": 8.764284823612667e-07, "loss": 0.09029741287231445, "step": 33020 }, { "epoch": 0.8700605525793291, "grad_norm": 3.7101006507873535, "learning_rate": 8.729391000718867e-07, "loss": 0.09935340285301208, "step": 33030 }, { "epoch": 0.8703239678238279, "grad_norm": 2.8055312633514404, "learning_rate": 8.694563608512041e-07, "loss": 0.09654632806777955, "step": 33040 }, { "epoch": 0.8705873830683266, "grad_norm": 2.96016526222229, "learning_rate": 8.659802672341078e-07, "loss": 0.0953908085823059, "step": 33050 }, { "epoch": 0.8708507983128254, "grad_norm": 2.147493362426758, "learning_rate": 8.625108217506472e-07, "loss": 0.09060289263725281, "step": 33060 }, { "epoch": 0.8711142135573241, "grad_norm": 3.3414535522460938, "learning_rate": 8.590480269260337e-07, "loss": 0.09266586303710937, "step": 33070 }, { "epoch": 0.8713776288018228, "grad_norm": 2.9966635704040527, "learning_rate": 8.555918852806421e-07, "loss": 0.09379749298095703, "step": 33080 }, { "epoch": 0.8716410440463216, "grad_norm": 3.2365458011627197, "learning_rate": 8.521423993300005e-07, "loss": 0.09489472508430481, "step": 33090 }, { "epoch": 0.8719044592908203, "grad_norm": 3.2773678302764893, "learning_rate": 8.486995715847923e-07, "loss": 0.09378722906112671, "step": 33100 }, { "epoch": 0.8721678745353191, "grad_norm": 3.706392765045166, "learning_rate": 8.452634045508557e-07, "loss": 0.08658404350280761, "step": 33110 }, { "epoch": 0.8724312897798178, "grad_norm": 2.850878953933716, "learning_rate": 8.41833900729182e-07, "loss": 0.09218142032623292, "step": 33120 }, { "epoch": 0.8726947050243166, "grad_norm": 3.1603126525878906, "learning_rate": 8.384110626159126e-07, "loss": 0.09671308994293212, "step": 33130 }, { "epoch": 0.8729581202688153, "grad_norm": 2.4211442470550537, "learning_rate": 8.349948927023388e-07, "loss": 0.08493578433990479, "step": 33140 }, { "epoch": 0.8732215355133139, "grad_norm": 2.888091564178467, "learning_rate": 8.315853934748908e-07, "loss": 0.10119633674621582, "step": 33150 }, { "epoch": 0.8734849507578127, "grad_norm": 3.0804309844970703, "learning_rate": 8.281825674151533e-07, "loss": 0.09750462174415589, "step": 33160 }, { "epoch": 0.8737483660023114, "grad_norm": 2.4247679710388184, "learning_rate": 8.247864169998499e-07, "loss": 0.08794012069702148, "step": 33170 }, { "epoch": 0.8740117812468102, "grad_norm": 2.3123862743377686, "learning_rate": 8.213969447008441e-07, "loss": 0.0880415141582489, "step": 33180 }, { "epoch": 0.8742751964913089, "grad_norm": 2.9148871898651123, "learning_rate": 8.180141529851404e-07, "loss": 0.09066598415374756, "step": 33190 }, { "epoch": 0.8745386117358077, "grad_norm": 4.550212383270264, "learning_rate": 8.146380443148794e-07, "loss": 0.09450763463973999, "step": 33200 }, { "epoch": 0.8748020269803064, "grad_norm": 5.148479461669922, "learning_rate": 8.11268621147342e-07, "loss": 0.09260139465332032, "step": 33210 }, { "epoch": 0.8750654422248052, "grad_norm": 3.4639859199523926, "learning_rate": 8.079058859349364e-07, "loss": 0.09737783074378967, "step": 33220 }, { "epoch": 0.8753288574693039, "grad_norm": 2.521502733230591, "learning_rate": 8.045498411252095e-07, "loss": 0.09138240814208984, "step": 33230 }, { "epoch": 0.8755922727138026, "grad_norm": 3.7582485675811768, "learning_rate": 8.012004891608327e-07, "loss": 0.10589154958724975, "step": 33240 }, { "epoch": 0.8758556879583014, "grad_norm": 3.4743521213531494, "learning_rate": 7.978578324796072e-07, "loss": 0.09324346780776978, "step": 33250 }, { "epoch": 0.8761191032028001, "grad_norm": 3.1209452152252197, "learning_rate": 7.945218735144689e-07, "loss": 0.09343381524085999, "step": 33260 }, { "epoch": 0.8763825184472989, "grad_norm": 3.251784563064575, "learning_rate": 7.91192614693469e-07, "loss": 0.09007018804550171, "step": 33270 }, { "epoch": 0.8766459336917976, "grad_norm": 3.119563341140747, "learning_rate": 7.878700584397836e-07, "loss": 0.08518782854080201, "step": 33280 }, { "epoch": 0.8769093489362964, "grad_norm": 3.917464017868042, "learning_rate": 7.84554207171716e-07, "loss": 0.08925743103027343, "step": 33290 }, { "epoch": 0.877172764180795, "grad_norm": 4.797036647796631, "learning_rate": 7.812450633026847e-07, "loss": 0.07941200733184814, "step": 33300 }, { "epoch": 0.8774361794252938, "grad_norm": 3.4335336685180664, "learning_rate": 7.779426292412273e-07, "loss": 0.08665021657943725, "step": 33310 }, { "epoch": 0.8776995946697925, "grad_norm": 2.922316312789917, "learning_rate": 7.746469073909979e-07, "loss": 0.09500949382781983, "step": 33320 }, { "epoch": 0.8779630099142912, "grad_norm": 3.1882729530334473, "learning_rate": 7.713579001507632e-07, "loss": 0.09935475587844848, "step": 33330 }, { "epoch": 0.87822642515879, "grad_norm": 2.3582522869110107, "learning_rate": 7.680756099144072e-07, "loss": 0.09477354288101196, "step": 33340 }, { "epoch": 0.8784898404032887, "grad_norm": 3.4036035537719727, "learning_rate": 7.648000390709221e-07, "loss": 0.09919623136520386, "step": 33350 }, { "epoch": 0.8787532556477875, "grad_norm": 2.686612367630005, "learning_rate": 7.615311900044087e-07, "loss": 0.07210625410079956, "step": 33360 }, { "epoch": 0.8790166708922862, "grad_norm": 3.601705551147461, "learning_rate": 7.582690650940771e-07, "loss": 0.09329230785369873, "step": 33370 }, { "epoch": 0.879280086136785, "grad_norm": 3.1223747730255127, "learning_rate": 7.550136667142393e-07, "loss": 0.09315357208251954, "step": 33380 }, { "epoch": 0.8795435013812837, "grad_norm": 3.924896717071533, "learning_rate": 7.517649972343199e-07, "loss": 0.08391173481941223, "step": 33390 }, { "epoch": 0.8798069166257825, "grad_norm": 2.5680642127990723, "learning_rate": 7.485230590188397e-07, "loss": 0.08640161752700806, "step": 33400 }, { "epoch": 0.8800703318702812, "grad_norm": 3.867201805114746, "learning_rate": 7.452878544274167e-07, "loss": 0.09501200914382935, "step": 33410 }, { "epoch": 0.8803337471147799, "grad_norm": 3.571079730987549, "learning_rate": 7.420593858147773e-07, "loss": 0.0985738217830658, "step": 33420 }, { "epoch": 0.8805971623592787, "grad_norm": 3.0092992782592773, "learning_rate": 7.388376555307386e-07, "loss": 0.10275831222534179, "step": 33430 }, { "epoch": 0.8808605776037773, "grad_norm": 3.037632942199707, "learning_rate": 7.356226659202137e-07, "loss": 0.08866103887557983, "step": 33440 }, { "epoch": 0.8811239928482761, "grad_norm": 2.9582574367523193, "learning_rate": 7.324144193232119e-07, "loss": 0.0929018497467041, "step": 33450 }, { "epoch": 0.8813874080927748, "grad_norm": 3.143890619277954, "learning_rate": 7.29212918074833e-07, "loss": 0.08487708568572998, "step": 33460 }, { "epoch": 0.8816508233372736, "grad_norm": 2.827176332473755, "learning_rate": 7.260181645052689e-07, "loss": 0.09305484890937805, "step": 33470 }, { "epoch": 0.8819142385817723, "grad_norm": 3.064573049545288, "learning_rate": 7.228301609397981e-07, "loss": 0.10136544704437256, "step": 33480 }, { "epoch": 0.8821776538262711, "grad_norm": 3.7227039337158203, "learning_rate": 7.196489096987869e-07, "loss": 0.10507712364196778, "step": 33490 }, { "epoch": 0.8824410690707698, "grad_norm": 3.137948989868164, "learning_rate": 7.164744130976886e-07, "loss": 0.0746012032032013, "step": 33500 }, { "epoch": 0.8827044843152685, "grad_norm": 3.250241279602051, "learning_rate": 7.133066734470351e-07, "loss": 0.09656537175178528, "step": 33510 }, { "epoch": 0.8829678995597673, "grad_norm": 3.6164119243621826, "learning_rate": 7.101456930524509e-07, "loss": 0.09905675649642945, "step": 33520 }, { "epoch": 0.883231314804266, "grad_norm": 3.2489030361175537, "learning_rate": 7.069914742146267e-07, "loss": 0.09201960563659668, "step": 33530 }, { "epoch": 0.8834947300487648, "grad_norm": 3.450777053833008, "learning_rate": 7.038440192293416e-07, "loss": 0.08826680779457093, "step": 33540 }, { "epoch": 0.8837581452932635, "grad_norm": 3.1368155479431152, "learning_rate": 7.007033303874511e-07, "loss": 0.09453619122505189, "step": 33550 }, { "epoch": 0.8840215605377623, "grad_norm": 2.248361349105835, "learning_rate": 6.97569409974882e-07, "loss": 0.09741613864898682, "step": 33560 }, { "epoch": 0.884284975782261, "grad_norm": 3.096714496612549, "learning_rate": 6.944422602726375e-07, "loss": 0.08358001112937927, "step": 33570 }, { "epoch": 0.8845483910267598, "grad_norm": 3.4126904010772705, "learning_rate": 6.913218835567925e-07, "loss": 0.09701161980628967, "step": 33580 }, { "epoch": 0.8848118062712584, "grad_norm": 3.582350969314575, "learning_rate": 6.882082820984892e-07, "loss": 0.09142806529998779, "step": 33590 }, { "epoch": 0.8850752215157571, "grad_norm": 3.0215229988098145, "learning_rate": 6.851014581639437e-07, "loss": 0.08295121788978577, "step": 33600 }, { "epoch": 0.8853386367602559, "grad_norm": 3.177119731903076, "learning_rate": 6.820014140144382e-07, "loss": 0.087735116481781, "step": 33610 }, { "epoch": 0.8856020520047546, "grad_norm": 3.2377405166625977, "learning_rate": 6.789081519063134e-07, "loss": 0.09260669946670533, "step": 33620 }, { "epoch": 0.8858654672492534, "grad_norm": 3.1456665992736816, "learning_rate": 6.758216740909829e-07, "loss": 0.1016180157661438, "step": 33630 }, { "epoch": 0.8861288824937521, "grad_norm": 3.712839365005493, "learning_rate": 6.727419828149173e-07, "loss": 0.10379312038421631, "step": 33640 }, { "epoch": 0.8863922977382509, "grad_norm": 2.558928966522217, "learning_rate": 6.696690803196482e-07, "loss": 0.09313035607337952, "step": 33650 }, { "epoch": 0.8866557129827496, "grad_norm": 3.1525793075561523, "learning_rate": 6.666029688417674e-07, "loss": 0.09393897056579589, "step": 33660 }, { "epoch": 0.8869191282272483, "grad_norm": 2.714195966720581, "learning_rate": 6.635436506129223e-07, "loss": 0.09124189615249634, "step": 33670 }, { "epoch": 0.8871825434717471, "grad_norm": 2.9455885887145996, "learning_rate": 6.60491127859818e-07, "loss": 0.08968055248260498, "step": 33680 }, { "epoch": 0.8874459587162458, "grad_norm": 2.9816956520080566, "learning_rate": 6.57445402804211e-07, "loss": 0.0822935163974762, "step": 33690 }, { "epoch": 0.8877093739607446, "grad_norm": 4.465208530426025, "learning_rate": 6.544064776629123e-07, "loss": 0.08924562931060791, "step": 33700 }, { "epoch": 0.8879727892052433, "grad_norm": 3.5152060985565186, "learning_rate": 6.513743546477824e-07, "loss": 0.08055285215377808, "step": 33710 }, { "epoch": 0.888236204449742, "grad_norm": 4.438351631164551, "learning_rate": 6.483490359657296e-07, "loss": 0.09286034107208252, "step": 33720 }, { "epoch": 0.8884996196942407, "grad_norm": 2.86714243888855, "learning_rate": 6.453305238187147e-07, "loss": 0.07532873153686523, "step": 33730 }, { "epoch": 0.8887630349387395, "grad_norm": 1.8379672765731812, "learning_rate": 6.423188204037412e-07, "loss": 0.08671334385871887, "step": 33740 }, { "epoch": 0.8890264501832382, "grad_norm": 3.1369266510009766, "learning_rate": 6.393139279128513e-07, "loss": 0.09184643030166625, "step": 33750 }, { "epoch": 0.8892898654277369, "grad_norm": 3.7496235370635986, "learning_rate": 6.3631584853314e-07, "loss": 0.09355286359786988, "step": 33760 }, { "epoch": 0.8895532806722357, "grad_norm": 4.482152938842773, "learning_rate": 6.333245844467384e-07, "loss": 0.08624956607818604, "step": 33770 }, { "epoch": 0.8898166959167344, "grad_norm": 2.3341617584228516, "learning_rate": 6.303401378308172e-07, "loss": 0.08655081987380982, "step": 33780 }, { "epoch": 0.8900801111612332, "grad_norm": 2.957627058029175, "learning_rate": 6.273625108575843e-07, "loss": 0.08603551983833313, "step": 33790 }, { "epoch": 0.8903435264057319, "grad_norm": 2.560284376144409, "learning_rate": 6.243917056942849e-07, "loss": 0.08966533541679382, "step": 33800 }, { "epoch": 0.8906069416502307, "grad_norm": 3.2787351608276367, "learning_rate": 6.214277245032019e-07, "loss": 0.07748063802719116, "step": 33810 }, { "epoch": 0.8908703568947294, "grad_norm": 3.1458675861358643, "learning_rate": 6.184705694416471e-07, "loss": 0.09301666021347046, "step": 33820 }, { "epoch": 0.8911337721392282, "grad_norm": 4.075384140014648, "learning_rate": 6.155202426619655e-07, "loss": 0.09430253505706787, "step": 33830 }, { "epoch": 0.8913971873837269, "grad_norm": 2.676725149154663, "learning_rate": 6.125767463115317e-07, "loss": 0.0902071475982666, "step": 33840 }, { "epoch": 0.8916606026282256, "grad_norm": 4.281543731689453, "learning_rate": 6.096400825327487e-07, "loss": 0.10286161899566651, "step": 33850 }, { "epoch": 0.8919240178727244, "grad_norm": 3.380580186843872, "learning_rate": 6.067102534630498e-07, "loss": 0.08890211582183838, "step": 33860 }, { "epoch": 0.892187433117223, "grad_norm": 3.6860034465789795, "learning_rate": 6.037872612348905e-07, "loss": 0.08869246244430543, "step": 33870 }, { "epoch": 0.8924508483617218, "grad_norm": 3.3586788177490234, "learning_rate": 6.008711079757467e-07, "loss": 0.09289317131042481, "step": 33880 }, { "epoch": 0.8927142636062205, "grad_norm": 3.1002888679504395, "learning_rate": 5.979617958081241e-07, "loss": 0.08716063499450684, "step": 33890 }, { "epoch": 0.8929776788507193, "grad_norm": 2.019514322280884, "learning_rate": 5.950593268495464e-07, "loss": 0.0818358063697815, "step": 33900 }, { "epoch": 0.893241094095218, "grad_norm": 2.809530258178711, "learning_rate": 5.921637032125538e-07, "loss": 0.0873609721660614, "step": 33910 }, { "epoch": 0.8935045093397168, "grad_norm": 3.483830213546753, "learning_rate": 5.89274927004706e-07, "loss": 0.09064873456954955, "step": 33920 }, { "epoch": 0.8937679245842155, "grad_norm": 2.911546230316162, "learning_rate": 5.863930003285779e-07, "loss": 0.08068569302558899, "step": 33930 }, { "epoch": 0.8940313398287142, "grad_norm": 3.141630172729492, "learning_rate": 5.835179252817635e-07, "loss": 0.08355578184127807, "step": 33940 }, { "epoch": 0.894294755073213, "grad_norm": 2.904794692993164, "learning_rate": 5.806497039568648e-07, "loss": 0.08437957763671874, "step": 33950 }, { "epoch": 0.8945581703177117, "grad_norm": 3.1209805011749268, "learning_rate": 5.777883384414973e-07, "loss": 0.08031675219535828, "step": 33960 }, { "epoch": 0.8948215855622105, "grad_norm": 3.126784086227417, "learning_rate": 5.749338308182851e-07, "loss": 0.08566299080848694, "step": 33970 }, { "epoch": 0.8950850008067092, "grad_norm": 4.4366655349731445, "learning_rate": 5.720861831648617e-07, "loss": 0.09027067422866822, "step": 33980 }, { "epoch": 0.895348416051208, "grad_norm": 3.327463388442993, "learning_rate": 5.692453975538703e-07, "loss": 0.08990063071250916, "step": 33990 }, { "epoch": 0.8956118312957067, "grad_norm": 2.3177220821380615, "learning_rate": 5.664114760529571e-07, "loss": 0.07992850542068482, "step": 34000 }, { "epoch": 0.8958752465402054, "grad_norm": 2.821127414703369, "learning_rate": 5.635844207247687e-07, "loss": 0.09459837675094604, "step": 34010 }, { "epoch": 0.8961386617847041, "grad_norm": 3.0097031593322754, "learning_rate": 5.607642336269615e-07, "loss": 0.08804070949554443, "step": 34020 }, { "epoch": 0.8964020770292028, "grad_norm": 3.522275447845459, "learning_rate": 5.579509168121888e-07, "loss": 0.08316448330879211, "step": 34030 }, { "epoch": 0.8966654922737016, "grad_norm": 2.7551212310791016, "learning_rate": 5.551444723281029e-07, "loss": 0.08426660299301147, "step": 34040 }, { "epoch": 0.8969289075182003, "grad_norm": 4.999213695526123, "learning_rate": 5.523449022173566e-07, "loss": 0.1041040301322937, "step": 34050 }, { "epoch": 0.8971923227626991, "grad_norm": 2.178546905517578, "learning_rate": 5.495522085175942e-07, "loss": 0.08285924196243286, "step": 34060 }, { "epoch": 0.8974557380071978, "grad_norm": 2.5019991397857666, "learning_rate": 5.467663932614642e-07, "loss": 0.08257007598876953, "step": 34070 }, { "epoch": 0.8977191532516966, "grad_norm": 3.6004087924957275, "learning_rate": 5.439874584766014e-07, "loss": 0.09957419037818908, "step": 34080 }, { "epoch": 0.8979825684961953, "grad_norm": 3.5056817531585693, "learning_rate": 5.412154061856301e-07, "loss": 0.07874701023101807, "step": 34090 }, { "epoch": 0.8982459837406941, "grad_norm": 3.1284756660461426, "learning_rate": 5.38450238406174e-07, "loss": 0.09986464977264405, "step": 34100 }, { "epoch": 0.8985093989851928, "grad_norm": 3.944567918777466, "learning_rate": 5.356919571508401e-07, "loss": 0.09943461418151855, "step": 34110 }, { "epoch": 0.8987728142296915, "grad_norm": 3.649904727935791, "learning_rate": 5.329405644272268e-07, "loss": 0.09751099348068237, "step": 34120 }, { "epoch": 0.8990362294741903, "grad_norm": 4.335876941680908, "learning_rate": 5.301960622379132e-07, "loss": 0.08551092147827148, "step": 34130 }, { "epoch": 0.899299644718689, "grad_norm": 3.6163437366485596, "learning_rate": 5.27458452580466e-07, "loss": 0.09651188254356384, "step": 34140 }, { "epoch": 0.8995630599631877, "grad_norm": 2.4474363327026367, "learning_rate": 5.24727737447438e-07, "loss": 0.08607638478279114, "step": 34150 }, { "epoch": 0.8998264752076864, "grad_norm": 3.027268886566162, "learning_rate": 5.220039188263615e-07, "loss": 0.09180448055267335, "step": 34160 }, { "epoch": 0.9000898904521852, "grad_norm": 2.7910265922546387, "learning_rate": 5.192869986997484e-07, "loss": 0.08466476202011108, "step": 34170 }, { "epoch": 0.9003533056966839, "grad_norm": 3.6583406925201416, "learning_rate": 5.165769790450903e-07, "loss": 0.09360975623130799, "step": 34180 }, { "epoch": 0.9006167209411826, "grad_norm": 3.685453414916992, "learning_rate": 5.138738618348548e-07, "loss": 0.08972256183624268, "step": 34190 }, { "epoch": 0.9008801361856814, "grad_norm": 4.044149875640869, "learning_rate": 5.111776490364906e-07, "loss": 0.10873928070068359, "step": 34200 }, { "epoch": 0.9011435514301801, "grad_norm": 3.4308018684387207, "learning_rate": 5.084883426124165e-07, "loss": 0.08426411151885986, "step": 34210 }, { "epoch": 0.9014069666746789, "grad_norm": 2.6933844089508057, "learning_rate": 5.058059445200237e-07, "loss": 0.0856201171875, "step": 34220 }, { "epoch": 0.9016703819191776, "grad_norm": 3.133965492248535, "learning_rate": 5.031304567116802e-07, "loss": 0.09320997595787048, "step": 34230 }, { "epoch": 0.9019337971636764, "grad_norm": 4.144275188446045, "learning_rate": 5.004618811347195e-07, "loss": 0.10303610563278198, "step": 34240 }, { "epoch": 0.9021972124081751, "grad_norm": 3.065229892730713, "learning_rate": 4.978002197314469e-07, "loss": 0.09607212543487549, "step": 34250 }, { "epoch": 0.9024606276526739, "grad_norm": 3.0982580184936523, "learning_rate": 4.951454744391338e-07, "loss": 0.09232580661773682, "step": 34260 }, { "epoch": 0.9027240428971726, "grad_norm": 3.6009817123413086, "learning_rate": 4.924976471900178e-07, "loss": 0.0853078842163086, "step": 34270 }, { "epoch": 0.9029874581416713, "grad_norm": 4.033894062042236, "learning_rate": 4.89856739911303e-07, "loss": 0.08223153352737426, "step": 34280 }, { "epoch": 0.90325087338617, "grad_norm": 4.006823539733887, "learning_rate": 4.872227545251562e-07, "loss": 0.08862242698669434, "step": 34290 }, { "epoch": 0.9035142886306687, "grad_norm": 3.1725809574127197, "learning_rate": 4.84595692948705e-07, "loss": 0.0778022825717926, "step": 34300 }, { "epoch": 0.9037777038751675, "grad_norm": 3.6119778156280518, "learning_rate": 4.819755570940377e-07, "loss": 0.09101446866989135, "step": 34310 }, { "epoch": 0.9040411191196662, "grad_norm": 2.768740653991699, "learning_rate": 4.793623488682008e-07, "loss": 0.08097474575042725, "step": 34320 }, { "epoch": 0.904304534364165, "grad_norm": 4.539244651794434, "learning_rate": 4.7675607017320323e-07, "loss": 0.09078863859176636, "step": 34330 }, { "epoch": 0.9045679496086637, "grad_norm": 4.481202602386475, "learning_rate": 4.741567229060062e-07, "loss": 0.08245304822921753, "step": 34340 }, { "epoch": 0.9048313648531625, "grad_norm": 3.208521842956543, "learning_rate": 4.715643089585242e-07, "loss": 0.09515695571899414, "step": 34350 }, { "epoch": 0.9050947800976612, "grad_norm": 2.8009157180786133, "learning_rate": 4.6897883021763013e-07, "loss": 0.08357650637626649, "step": 34360 }, { "epoch": 0.9053581953421599, "grad_norm": 4.166088104248047, "learning_rate": 4.664002885651475e-07, "loss": 0.09209330081939697, "step": 34370 }, { "epoch": 0.9056216105866587, "grad_norm": 6.172910690307617, "learning_rate": 4.638286858778496e-07, "loss": 0.09699270725250245, "step": 34380 }, { "epoch": 0.9058850258311574, "grad_norm": 3.151268243789673, "learning_rate": 4.612640240274591e-07, "loss": 0.0796167016029358, "step": 34390 }, { "epoch": 0.9061484410756562, "grad_norm": 3.2701032161712646, "learning_rate": 4.587063048806462e-07, "loss": 0.09580548405647278, "step": 34400 }, { "epoch": 0.9064118563201549, "grad_norm": 2.5519790649414062, "learning_rate": 4.561555302990328e-07, "loss": 0.10598872900009156, "step": 34410 }, { "epoch": 0.9066752715646537, "grad_norm": 2.922117233276367, "learning_rate": 4.536117021391806e-07, "loss": 0.09263491630554199, "step": 34420 }, { "epoch": 0.9069386868091524, "grad_norm": 3.366900682449341, "learning_rate": 4.5107482225259735e-07, "loss": 0.09019941091537476, "step": 34430 }, { "epoch": 0.9072021020536511, "grad_norm": 3.3690853118896484, "learning_rate": 4.4854489248573494e-07, "loss": 0.08484718799591065, "step": 34440 }, { "epoch": 0.9074655172981498, "grad_norm": 3.087433338165283, "learning_rate": 4.460219146799838e-07, "loss": 0.08383032083511352, "step": 34450 }, { "epoch": 0.9077289325426485, "grad_norm": 3.207143545150757, "learning_rate": 4.4350589067167717e-07, "loss": 0.08902240991592407, "step": 34460 }, { "epoch": 0.9079923477871473, "grad_norm": 4.185367107391357, "learning_rate": 4.409968222920868e-07, "loss": 0.10086619853973389, "step": 34470 }, { "epoch": 0.908255763031646, "grad_norm": 3.956326723098755, "learning_rate": 4.384947113674165e-07, "loss": 0.09269272089004517, "step": 34480 }, { "epoch": 0.9085191782761448, "grad_norm": 3.4396326541900635, "learning_rate": 4.35999559718816e-07, "loss": 0.08737400770187378, "step": 34490 }, { "epoch": 0.9087825935206435, "grad_norm": 2.9092631340026855, "learning_rate": 4.3351136916236156e-07, "loss": 0.10193984508514405, "step": 34500 }, { "epoch": 0.9090460087651423, "grad_norm": 4.45781946182251, "learning_rate": 4.3103014150906675e-07, "loss": 0.08284825086593628, "step": 34510 }, { "epoch": 0.909309424009641, "grad_norm": 3.3677327632904053, "learning_rate": 4.285558785648747e-07, "loss": 0.09505324959754943, "step": 34520 }, { "epoch": 0.9095728392541398, "grad_norm": 3.688452959060669, "learning_rate": 4.260885821306604e-07, "loss": 0.08238040804862976, "step": 34530 }, { "epoch": 0.9098362544986385, "grad_norm": 3.0169668197631836, "learning_rate": 4.236282540022307e-07, "loss": 0.0830733597278595, "step": 34540 }, { "epoch": 0.9100996697431372, "grad_norm": 2.5692360401153564, "learning_rate": 4.2117489597031767e-07, "loss": 0.09210373163223266, "step": 34550 }, { "epoch": 0.910363084987636, "grad_norm": 4.466651439666748, "learning_rate": 4.1872850982058067e-07, "loss": 0.09334124326705932, "step": 34560 }, { "epoch": 0.9106265002321347, "grad_norm": 3.0223512649536133, "learning_rate": 4.1628909733360534e-07, "loss": 0.0919814646244049, "step": 34570 }, { "epoch": 0.9108899154766334, "grad_norm": 3.4052681922912598, "learning_rate": 4.1385666028490036e-07, "loss": 0.09342589974403381, "step": 34580 }, { "epoch": 0.9111533307211321, "grad_norm": 4.161135673522949, "learning_rate": 4.114312004449006e-07, "loss": 0.07698467373847961, "step": 34590 }, { "epoch": 0.9114167459656309, "grad_norm": 3.9734840393066406, "learning_rate": 4.090127195789606e-07, "loss": 0.08851102590560914, "step": 34600 }, { "epoch": 0.9116801612101296, "grad_norm": 3.1197097301483154, "learning_rate": 4.066012194473512e-07, "loss": 0.08989826440811158, "step": 34610 }, { "epoch": 0.9119435764546284, "grad_norm": 3.4292261600494385, "learning_rate": 4.041967018052706e-07, "loss": 0.09169580340385437, "step": 34620 }, { "epoch": 0.9122069916991271, "grad_norm": 3.0280518531799316, "learning_rate": 4.0179916840282776e-07, "loss": 0.08320804238319397, "step": 34630 }, { "epoch": 0.9124704069436258, "grad_norm": 3.550360679626465, "learning_rate": 3.9940862098505117e-07, "loss": 0.09336429238319396, "step": 34640 }, { "epoch": 0.9127338221881246, "grad_norm": 4.780481815338135, "learning_rate": 3.970250612918858e-07, "loss": 0.1015979528427124, "step": 34650 }, { "epoch": 0.9129972374326233, "grad_norm": 2.794116973876953, "learning_rate": 3.946484910581871e-07, "loss": 0.08243285417556763, "step": 34660 }, { "epoch": 0.9132606526771221, "grad_norm": 2.762831211090088, "learning_rate": 3.9227891201372827e-07, "loss": 0.080685293674469, "step": 34670 }, { "epoch": 0.9135240679216208, "grad_norm": 3.943877935409546, "learning_rate": 3.899163258831895e-07, "loss": 0.09383389353752136, "step": 34680 }, { "epoch": 0.9137874831661196, "grad_norm": 2.683112382888794, "learning_rate": 3.8756073438616203e-07, "loss": 0.08591522574424744, "step": 34690 }, { "epoch": 0.9140508984106183, "grad_norm": 2.8172786235809326, "learning_rate": 3.8521213923714775e-07, "loss": 0.09910482168197632, "step": 34700 }, { "epoch": 0.914314313655117, "grad_norm": 3.1517558097839355, "learning_rate": 3.8287054214555694e-07, "loss": 0.08033649921417237, "step": 34710 }, { "epoch": 0.9145777288996157, "grad_norm": 2.924345016479492, "learning_rate": 3.805359448157031e-07, "loss": 0.08253288865089417, "step": 34720 }, { "epoch": 0.9148411441441144, "grad_norm": 2.885835647583008, "learning_rate": 3.78208348946808e-07, "loss": 0.082448410987854, "step": 34730 }, { "epoch": 0.9151045593886132, "grad_norm": 2.8155643939971924, "learning_rate": 3.7588775623299433e-07, "loss": 0.08016645312309265, "step": 34740 }, { "epoch": 0.9153679746331119, "grad_norm": 4.197465896606445, "learning_rate": 3.735741683632932e-07, "loss": 0.0941754400730133, "step": 34750 }, { "epoch": 0.9156313898776107, "grad_norm": 2.4367313385009766, "learning_rate": 3.712675870216309e-07, "loss": 0.08752024173736572, "step": 34760 }, { "epoch": 0.9158948051221094, "grad_norm": 3.431107997894287, "learning_rate": 3.689680138868379e-07, "loss": 0.08618452548980712, "step": 34770 }, { "epoch": 0.9161582203666082, "grad_norm": 3.0793747901916504, "learning_rate": 3.66675450632642e-07, "loss": 0.07918673753738403, "step": 34780 }, { "epoch": 0.9164216356111069, "grad_norm": 3.3247764110565186, "learning_rate": 3.643898989276684e-07, "loss": 0.0951232671737671, "step": 34790 }, { "epoch": 0.9166850508556056, "grad_norm": 3.9379019737243652, "learning_rate": 3.6211136043544425e-07, "loss": 0.08597785234451294, "step": 34800 }, { "epoch": 0.9169484661001044, "grad_norm": 2.8252718448638916, "learning_rate": 3.598398368143863e-07, "loss": 0.09560022950172424, "step": 34810 }, { "epoch": 0.9172118813446031, "grad_norm": 3.0144307613372803, "learning_rate": 3.575753297178053e-07, "loss": 0.08024955987930298, "step": 34820 }, { "epoch": 0.9174752965891019, "grad_norm": 3.11038875579834, "learning_rate": 3.553178407939095e-07, "loss": 0.08413999080657959, "step": 34830 }, { "epoch": 0.9177387118336006, "grad_norm": 2.642026662826538, "learning_rate": 3.5306737168579685e-07, "loss": 0.08070448637008668, "step": 34840 }, { "epoch": 0.9180021270780994, "grad_norm": 2.8060691356658936, "learning_rate": 3.5082392403145373e-07, "loss": 0.09195908308029174, "step": 34850 }, { "epoch": 0.918265542322598, "grad_norm": 2.8963239192962646, "learning_rate": 3.485874994637606e-07, "loss": 0.0834100604057312, "step": 34860 }, { "epoch": 0.9185289575670968, "grad_norm": 5.191790580749512, "learning_rate": 3.463580996104821e-07, "loss": 0.09674763679504395, "step": 34870 }, { "epoch": 0.9187923728115955, "grad_norm": 3.6697263717651367, "learning_rate": 3.4413572609427235e-07, "loss": 0.09424487352371216, "step": 34880 }, { "epoch": 0.9190557880560942, "grad_norm": 4.342655181884766, "learning_rate": 3.4192038053267095e-07, "loss": 0.09001199007034302, "step": 34890 }, { "epoch": 0.919319203300593, "grad_norm": 2.801103115081787, "learning_rate": 3.3971206453810133e-07, "loss": 0.08310393691062927, "step": 34900 }, { "epoch": 0.9195826185450917, "grad_norm": 2.6554250717163086, "learning_rate": 3.3751077971787115e-07, "loss": 0.08447160720825195, "step": 34910 }, { "epoch": 0.9198460337895905, "grad_norm": 3.4725122451782227, "learning_rate": 3.353165276741699e-07, "loss": 0.08659831285476685, "step": 34920 }, { "epoch": 0.9201094490340892, "grad_norm": 4.1573591232299805, "learning_rate": 3.331293100040711e-07, "loss": 0.08777194619178771, "step": 34930 }, { "epoch": 0.920372864278588, "grad_norm": 4.207382678985596, "learning_rate": 3.3094912829952477e-07, "loss": 0.09274623990058899, "step": 34940 }, { "epoch": 0.9206362795230867, "grad_norm": 4.022140979766846, "learning_rate": 3.2877598414736033e-07, "loss": 0.07913624048233033, "step": 34950 }, { "epoch": 0.9208996947675855, "grad_norm": 3.9144253730773926, "learning_rate": 3.266098791292871e-07, "loss": 0.09982212781906127, "step": 34960 }, { "epoch": 0.9211631100120842, "grad_norm": 2.894251823425293, "learning_rate": 3.244508148218895e-07, "loss": 0.07378747463226318, "step": 34970 }, { "epoch": 0.9214265252565829, "grad_norm": 3.4446849822998047, "learning_rate": 3.22298792796627e-07, "loss": 0.09052566289901734, "step": 34980 }, { "epoch": 0.9216899405010817, "grad_norm": 3.320096254348755, "learning_rate": 3.2015381461983577e-07, "loss": 0.09383386373519897, "step": 34990 }, { "epoch": 0.9219533557455803, "grad_norm": 3.6354925632476807, "learning_rate": 3.1801588185272016e-07, "loss": 0.07943417429924012, "step": 35000 }, { "epoch": 0.9222167709900791, "grad_norm": 2.906405210494995, "learning_rate": 3.1588499605136435e-07, "loss": 0.0927205204963684, "step": 35010 }, { "epoch": 0.9224801862345778, "grad_norm": 3.2836525440216064, "learning_rate": 3.1376115876671553e-07, "loss": 0.09153398275375366, "step": 35020 }, { "epoch": 0.9227436014790766, "grad_norm": 2.4693377017974854, "learning_rate": 3.1164437154459494e-07, "loss": 0.07831282615661621, "step": 35030 }, { "epoch": 0.9230070167235753, "grad_norm": 4.4522929191589355, "learning_rate": 3.0953463592569343e-07, "loss": 0.07599896192550659, "step": 35040 }, { "epoch": 0.9232704319680741, "grad_norm": 3.037710666656494, "learning_rate": 3.074319534455639e-07, "loss": 0.09107817411422729, "step": 35050 }, { "epoch": 0.9235338472125728, "grad_norm": 3.788597345352173, "learning_rate": 3.053363256346331e-07, "loss": 0.09831574559211731, "step": 35060 }, { "epoch": 0.9237972624570715, "grad_norm": 5.649660110473633, "learning_rate": 3.0324775401818884e-07, "loss": 0.08625438213348388, "step": 35070 }, { "epoch": 0.9240606777015703, "grad_norm": 3.810404062271118, "learning_rate": 3.0116624011637953e-07, "loss": 0.09638125896453857, "step": 35080 }, { "epoch": 0.924324092946069, "grad_norm": 2.884836435317993, "learning_rate": 2.9909178544422656e-07, "loss": 0.0820243239402771, "step": 35090 }, { "epoch": 0.9245875081905678, "grad_norm": 3.886169672012329, "learning_rate": 2.9702439151160334e-07, "loss": 0.0970758855342865, "step": 35100 }, { "epoch": 0.9248509234350665, "grad_norm": 4.603621959686279, "learning_rate": 2.949640598232517e-07, "loss": 0.09020019769668579, "step": 35110 }, { "epoch": 0.9251143386795653, "grad_norm": 3.6102075576782227, "learning_rate": 2.9291079187876656e-07, "loss": 0.09280143976211548, "step": 35120 }, { "epoch": 0.925377753924064, "grad_norm": 4.62855863571167, "learning_rate": 2.9086458917260585e-07, "loss": 0.09362294673919677, "step": 35130 }, { "epoch": 0.9256411691685628, "grad_norm": 3.2662384510040283, "learning_rate": 2.8882545319408706e-07, "loss": 0.07871664762496948, "step": 35140 }, { "epoch": 0.9259045844130614, "grad_norm": 3.08040452003479, "learning_rate": 2.8679338542737745e-07, "loss": 0.09655821919441224, "step": 35150 }, { "epoch": 0.9261679996575601, "grad_norm": 5.07653284072876, "learning_rate": 2.8476838735150613e-07, "loss": 0.08652898669242859, "step": 35160 }, { "epoch": 0.9264314149020589, "grad_norm": 3.701508045196533, "learning_rate": 2.8275046044035304e-07, "loss": 0.08410835266113281, "step": 35170 }, { "epoch": 0.9266948301465576, "grad_norm": 3.5966343879699707, "learning_rate": 2.8073960616265215e-07, "loss": 0.10220491886138916, "step": 35180 }, { "epoch": 0.9269582453910564, "grad_norm": 3.610872268676758, "learning_rate": 2.7873582598199056e-07, "loss": 0.08869006633758544, "step": 35190 }, { "epoch": 0.9272216606355551, "grad_norm": 2.5466601848602295, "learning_rate": 2.7673912135680714e-07, "loss": 0.07696690559387206, "step": 35200 }, { "epoch": 0.9274850758800539, "grad_norm": 3.2234947681427, "learning_rate": 2.7474949374038717e-07, "loss": 0.08002400398254395, "step": 35210 }, { "epoch": 0.9277484911245526, "grad_norm": 2.9634053707122803, "learning_rate": 2.7276694458086894e-07, "loss": 0.09134860634803772, "step": 35220 }, { "epoch": 0.9280119063690514, "grad_norm": 2.8772363662719727, "learning_rate": 2.7079147532123704e-07, "loss": 0.09678640365600585, "step": 35230 }, { "epoch": 0.9282753216135501, "grad_norm": 2.442417860031128, "learning_rate": 2.688230873993225e-07, "loss": 0.0677062451839447, "step": 35240 }, { "epoch": 0.9285387368580488, "grad_norm": 2.895190715789795, "learning_rate": 2.6686178224780367e-07, "loss": 0.08104737997055053, "step": 35250 }, { "epoch": 0.9288021521025476, "grad_norm": 3.4340639114379883, "learning_rate": 2.6490756129420313e-07, "loss": 0.07870402336120605, "step": 35260 }, { "epoch": 0.9290655673470463, "grad_norm": 2.708466053009033, "learning_rate": 2.6296042596088643e-07, "loss": 0.08725212216377258, "step": 35270 }, { "epoch": 0.9293289825915451, "grad_norm": 3.4588022232055664, "learning_rate": 2.610203776650655e-07, "loss": 0.0913013219833374, "step": 35280 }, { "epoch": 0.9295923978360437, "grad_norm": 3.766982316970825, "learning_rate": 2.5908741781878744e-07, "loss": 0.08135780692100525, "step": 35290 }, { "epoch": 0.9298558130805425, "grad_norm": 2.849013328552246, "learning_rate": 2.5716154782894573e-07, "loss": 0.08325653076171875, "step": 35300 }, { "epoch": 0.9301192283250412, "grad_norm": 5.144394874572754, "learning_rate": 2.552427690972703e-07, "loss": 0.08073135018348694, "step": 35310 }, { "epoch": 0.9303826435695399, "grad_norm": 3.655245542526245, "learning_rate": 2.5333108302033394e-07, "loss": 0.08385292887687683, "step": 35320 }, { "epoch": 0.9306460588140387, "grad_norm": 4.0114216804504395, "learning_rate": 2.514264909895414e-07, "loss": 0.08643591403961182, "step": 35330 }, { "epoch": 0.9309094740585374, "grad_norm": 2.764897108078003, "learning_rate": 2.495289943911361e-07, "loss": 0.08982962369918823, "step": 35340 }, { "epoch": 0.9311728893030362, "grad_norm": 3.212315797805786, "learning_rate": 2.476385946062021e-07, "loss": 0.08351254463195801, "step": 35350 }, { "epoch": 0.9314363045475349, "grad_norm": 3.3719260692596436, "learning_rate": 2.4575529301065104e-07, "loss": 0.08810253143310547, "step": 35360 }, { "epoch": 0.9316997197920337, "grad_norm": 3.7494077682495117, "learning_rate": 2.43879090975232e-07, "loss": 0.0906517505645752, "step": 35370 }, { "epoch": 0.9319631350365324, "grad_norm": 4.802675247192383, "learning_rate": 2.4200998986552484e-07, "loss": 0.090618097782135, "step": 35380 }, { "epoch": 0.9322265502810312, "grad_norm": 4.087437629699707, "learning_rate": 2.401479910419413e-07, "loss": 0.09374334812164306, "step": 35390 }, { "epoch": 0.9324899655255299, "grad_norm": 2.8628053665161133, "learning_rate": 2.3829309585972738e-07, "loss": 0.08706703782081604, "step": 35400 }, { "epoch": 0.9327533807700286, "grad_norm": 4.084856033325195, "learning_rate": 2.364453056689553e-07, "loss": 0.09569993019104003, "step": 35410 }, { "epoch": 0.9330167960145274, "grad_norm": 3.726652145385742, "learning_rate": 2.3460462181452258e-07, "loss": 0.09575282335281372, "step": 35420 }, { "epoch": 0.933280211259026, "grad_norm": 2.470247268676758, "learning_rate": 2.3277104563616205e-07, "loss": 0.09072656631469726, "step": 35430 }, { "epoch": 0.9335436265035248, "grad_norm": 2.9172914028167725, "learning_rate": 2.3094457846842943e-07, "loss": 0.08340265154838562, "step": 35440 }, { "epoch": 0.9338070417480235, "grad_norm": 5.065999984741211, "learning_rate": 2.291252216407047e-07, "loss": 0.07446525096893311, "step": 35450 }, { "epoch": 0.9340704569925223, "grad_norm": 3.481147527694702, "learning_rate": 2.273129764771953e-07, "loss": 0.07956342697143555, "step": 35460 }, { "epoch": 0.934333872237021, "grad_norm": 4.346490383148193, "learning_rate": 2.2550784429693052e-07, "loss": 0.09397023916244507, "step": 35470 }, { "epoch": 0.9345972874815198, "grad_norm": 3.992570400238037, "learning_rate": 2.2370982641376494e-07, "loss": 0.0782361626625061, "step": 35480 }, { "epoch": 0.9348607027260185, "grad_norm": 5.265741348266602, "learning_rate": 2.2191892413637173e-07, "loss": 0.08230496644973755, "step": 35490 }, { "epoch": 0.9351241179705172, "grad_norm": 2.6835684776306152, "learning_rate": 2.2013513876824933e-07, "loss": 0.08089259862899781, "step": 35500 }, { "epoch": 0.935387533215016, "grad_norm": 3.226628303527832, "learning_rate": 2.183584716077114e-07, "loss": 0.08644272089004516, "step": 35510 }, { "epoch": 0.9356509484595147, "grad_norm": 2.7770867347717285, "learning_rate": 2.1658892394789244e-07, "loss": 0.08217641711235046, "step": 35520 }, { "epoch": 0.9359143637040135, "grad_norm": 3.6899688243865967, "learning_rate": 2.1482649707674775e-07, "loss": 0.07848262190818786, "step": 35530 }, { "epoch": 0.9361777789485122, "grad_norm": 2.97949481010437, "learning_rate": 2.13071192277049e-07, "loss": 0.08125452399253845, "step": 35540 }, { "epoch": 0.936441194193011, "grad_norm": 2.552117347717285, "learning_rate": 2.1132301082637752e-07, "loss": 0.08689943552017212, "step": 35550 }, { "epoch": 0.9367046094375097, "grad_norm": 4.18700647354126, "learning_rate": 2.0958195399713888e-07, "loss": 0.08684868812561035, "step": 35560 }, { "epoch": 0.9369680246820085, "grad_norm": 3.380070924758911, "learning_rate": 2.0784802305654938e-07, "loss": 0.07831243872642517, "step": 35570 }, { "epoch": 0.9372314399265071, "grad_norm": 3.736722230911255, "learning_rate": 2.061212192666373e-07, "loss": 0.07715579867362976, "step": 35580 }, { "epoch": 0.9374948551710058, "grad_norm": 2.8802330493927, "learning_rate": 2.0440154388424616e-07, "loss": 0.08545064926147461, "step": 35590 }, { "epoch": 0.9377582704155046, "grad_norm": 3.3513002395629883, "learning_rate": 2.0268899816102916e-07, "loss": 0.08320066332817078, "step": 35600 }, { "epoch": 0.9380216856600033, "grad_norm": 3.302381992340088, "learning_rate": 2.0098358334345147e-07, "loss": 0.08506070375442505, "step": 35610 }, { "epoch": 0.9382851009045021, "grad_norm": 3.967123031616211, "learning_rate": 1.9928530067278684e-07, "loss": 0.08740626573562622, "step": 35620 }, { "epoch": 0.9385485161490008, "grad_norm": 2.9335482120513916, "learning_rate": 1.9759415138511984e-07, "loss": 0.08078117370605468, "step": 35630 }, { "epoch": 0.9388119313934996, "grad_norm": 3.6432719230651855, "learning_rate": 1.959101367113414e-07, "loss": 0.08096858263015747, "step": 35640 }, { "epoch": 0.9390753466379983, "grad_norm": 3.1051268577575684, "learning_rate": 1.9423325787714774e-07, "loss": 0.088394033908844, "step": 35650 }, { "epoch": 0.9393387618824971, "grad_norm": 3.329352378845215, "learning_rate": 1.9256351610304814e-07, "loss": 0.08853623867034913, "step": 35660 }, { "epoch": 0.9396021771269958, "grad_norm": 3.577147960662842, "learning_rate": 1.909009126043504e-07, "loss": 0.08185936808586121, "step": 35670 }, { "epoch": 0.9398655923714945, "grad_norm": 3.1278069019317627, "learning_rate": 1.8924544859116656e-07, "loss": 0.08816752433776856, "step": 35680 }, { "epoch": 0.9401290076159933, "grad_norm": 3.798156976699829, "learning_rate": 1.8759712526841943e-07, "loss": 0.09113901853561401, "step": 35690 }, { "epoch": 0.940392422860492, "grad_norm": 3.9660720825195312, "learning_rate": 1.8595594383582605e-07, "loss": 0.09360637664794921, "step": 35700 }, { "epoch": 0.9406558381049908, "grad_norm": 3.68886137008667, "learning_rate": 1.8432190548791197e-07, "loss": 0.0937181293964386, "step": 35710 }, { "epoch": 0.9409192533494894, "grad_norm": 2.7293756008148193, "learning_rate": 1.8269501141399915e-07, "loss": 0.07304525375366211, "step": 35720 }, { "epoch": 0.9411826685939882, "grad_norm": 3.365615129470825, "learning_rate": 1.8107526279821041e-07, "loss": 0.08371279239654542, "step": 35730 }, { "epoch": 0.9414460838384869, "grad_norm": 3.3052690029144287, "learning_rate": 1.794626608194705e-07, "loss": 0.08127114772796631, "step": 35740 }, { "epoch": 0.9417094990829857, "grad_norm": 2.955592393875122, "learning_rate": 1.7785720665150054e-07, "loss": 0.09623112678527831, "step": 35750 }, { "epoch": 0.9419729143274844, "grad_norm": 5.20686674118042, "learning_rate": 1.7625890146281798e-07, "loss": 0.08827178478240967, "step": 35760 }, { "epoch": 0.9422363295719831, "grad_norm": 2.878920316696167, "learning_rate": 1.7466774641674012e-07, "loss": 0.06959090232849122, "step": 35770 }, { "epoch": 0.9424997448164819, "grad_norm": 3.8736343383789062, "learning_rate": 1.7308374267137606e-07, "loss": 0.08041296005249024, "step": 35780 }, { "epoch": 0.9427631600609806, "grad_norm": 3.63692045211792, "learning_rate": 1.715068913796336e-07, "loss": 0.09394584894180298, "step": 35790 }, { "epoch": 0.9430265753054794, "grad_norm": 3.9377553462982178, "learning_rate": 1.6993719368921468e-07, "loss": 0.09127916097640991, "step": 35800 }, { "epoch": 0.9432899905499781, "grad_norm": 3.467766761779785, "learning_rate": 1.683746507426087e-07, "loss": 0.07548606395721436, "step": 35810 }, { "epoch": 0.9435534057944769, "grad_norm": 3.7000772953033447, "learning_rate": 1.66819263677106e-07, "loss": 0.0997403621673584, "step": 35820 }, { "epoch": 0.9438168210389756, "grad_norm": 3.8340518474578857, "learning_rate": 1.6527103362478202e-07, "loss": 0.09227544069290161, "step": 35830 }, { "epoch": 0.9440802362834743, "grad_norm": 4.467344760894775, "learning_rate": 1.6372996171250544e-07, "loss": 0.09280847311019898, "step": 35840 }, { "epoch": 0.9443436515279731, "grad_norm": 3.6634621620178223, "learning_rate": 1.6219604906193677e-07, "loss": 0.08204265832901, "step": 35850 }, { "epoch": 0.9446070667724717, "grad_norm": 3.5889415740966797, "learning_rate": 1.606692967895207e-07, "loss": 0.08826712369918824, "step": 35860 }, { "epoch": 0.9448704820169705, "grad_norm": 4.193356037139893, "learning_rate": 1.5914970600649615e-07, "loss": 0.09566786289215087, "step": 35870 }, { "epoch": 0.9451338972614692, "grad_norm": 4.451437473297119, "learning_rate": 1.5763727781888726e-07, "loss": 0.06987161040306092, "step": 35880 }, { "epoch": 0.945397312505968, "grad_norm": 3.350801706314087, "learning_rate": 1.5613201332750127e-07, "loss": 0.08200619220733643, "step": 35890 }, { "epoch": 0.9456607277504667, "grad_norm": 3.2984931468963623, "learning_rate": 1.5463391362793735e-07, "loss": 0.07682634592056274, "step": 35900 }, { "epoch": 0.9459241429949655, "grad_norm": 2.538004159927368, "learning_rate": 1.5314297981057658e-07, "loss": 0.08133072853088379, "step": 35910 }, { "epoch": 0.9461875582394642, "grad_norm": 2.6968374252319336, "learning_rate": 1.5165921296058427e-07, "loss": 0.08223453164100647, "step": 35920 }, { "epoch": 0.9464509734839629, "grad_norm": 2.633880853652954, "learning_rate": 1.501826141579099e-07, "loss": 0.08298977017402649, "step": 35930 }, { "epoch": 0.9467143887284617, "grad_norm": 2.6313204765319824, "learning_rate": 1.48713184477286e-07, "loss": 0.08283977508544922, "step": 35940 }, { "epoch": 0.9469778039729604, "grad_norm": 4.78745698928833, "learning_rate": 1.4725092498822814e-07, "loss": 0.08558396100997925, "step": 35950 }, { "epoch": 0.9472412192174592, "grad_norm": 3.494431734085083, "learning_rate": 1.457958367550305e-07, "loss": 0.08929576873779296, "step": 35960 }, { "epoch": 0.9475046344619579, "grad_norm": 2.711953639984131, "learning_rate": 1.443479208367693e-07, "loss": 0.07609738111495971, "step": 35970 }, { "epoch": 0.9477680497064567, "grad_norm": 3.515850067138672, "learning_rate": 1.4290717828730037e-07, "loss": 0.0819555938243866, "step": 35980 }, { "epoch": 0.9480314649509554, "grad_norm": 3.8100647926330566, "learning_rate": 1.4147361015525717e-07, "loss": 0.08971592187881469, "step": 35990 }, { "epoch": 0.9482948801954542, "grad_norm": 3.0770342350006104, "learning_rate": 1.4004721748405392e-07, "loss": 0.09595319032669067, "step": 36000 }, { "epoch": 0.9485582954399528, "grad_norm": 3.5885469913482666, "learning_rate": 1.3862800131188237e-07, "loss": 0.09062127470970154, "step": 36010 }, { "epoch": 0.9488217106844515, "grad_norm": 3.683445692062378, "learning_rate": 1.37215962671704e-07, "loss": 0.08800913691520691, "step": 36020 }, { "epoch": 0.9490851259289503, "grad_norm": 3.2093610763549805, "learning_rate": 1.3581110259126674e-07, "loss": 0.08520069718360901, "step": 36030 }, { "epoch": 0.949348541173449, "grad_norm": 3.3426856994628906, "learning_rate": 1.3441342209308595e-07, "loss": 0.08956202268600463, "step": 36040 }, { "epoch": 0.9496119564179478, "grad_norm": 4.8597493171691895, "learning_rate": 1.3302292219445457e-07, "loss": 0.08989509344100952, "step": 36050 }, { "epoch": 0.9498753716624465, "grad_norm": 3.3539135456085205, "learning_rate": 1.3163960390743746e-07, "loss": 0.08164318799972534, "step": 36060 }, { "epoch": 0.9501387869069453, "grad_norm": 4.800418376922607, "learning_rate": 1.302634682388726e-07, "loss": 0.083348947763443, "step": 36070 }, { "epoch": 0.950402202151444, "grad_norm": 2.9217710494995117, "learning_rate": 1.2889451619037318e-07, "loss": 0.08314765691757202, "step": 36080 }, { "epoch": 0.9506656173959428, "grad_norm": 5.455418109893799, "learning_rate": 1.2753274875832112e-07, "loss": 0.08682215809822083, "step": 36090 }, { "epoch": 0.9509290326404415, "grad_norm": 2.96856427192688, "learning_rate": 1.2617816693386907e-07, "loss": 0.0873587965965271, "step": 36100 }, { "epoch": 0.9511924478849402, "grad_norm": 4.21444845199585, "learning_rate": 1.2483077170293954e-07, "loss": 0.09286571741104126, "step": 36110 }, { "epoch": 0.951455863129439, "grad_norm": 3.181164026260376, "learning_rate": 1.2349056404622472e-07, "loss": 0.06337594985961914, "step": 36120 }, { "epoch": 0.9517192783739377, "grad_norm": 3.138577461242676, "learning_rate": 1.2215754493918763e-07, "loss": 0.09095133543014526, "step": 36130 }, { "epoch": 0.9519826936184365, "grad_norm": 4.22736930847168, "learning_rate": 1.208317153520555e-07, "loss": 0.09109320640563964, "step": 36140 }, { "epoch": 0.9522461088629351, "grad_norm": 3.2362430095672607, "learning_rate": 1.195130762498231e-07, "loss": 0.08646397590637207, "step": 36150 }, { "epoch": 0.952509524107434, "grad_norm": 3.2926886081695557, "learning_rate": 1.1820162859225493e-07, "loss": 0.0890853762626648, "step": 36160 }, { "epoch": 0.9527729393519326, "grad_norm": 2.523433208465576, "learning_rate": 1.1689737333387852e-07, "loss": 0.06886869072914123, "step": 36170 }, { "epoch": 0.9530363545964314, "grad_norm": 3.4738330841064453, "learning_rate": 1.1560031142398787e-07, "loss": 0.08684483170509338, "step": 36180 }, { "epoch": 0.9532997698409301, "grad_norm": 3.911371946334839, "learning_rate": 1.1431044380663891e-07, "loss": 0.09843932390213013, "step": 36190 }, { "epoch": 0.9535631850854288, "grad_norm": 3.2071728706359863, "learning_rate": 1.1302777142065402e-07, "loss": 0.08496583104133607, "step": 36200 }, { "epoch": 0.9538266003299276, "grad_norm": 3.3202414512634277, "learning_rate": 1.117522951996175e-07, "loss": 0.08740538358688354, "step": 36210 }, { "epoch": 0.9540900155744263, "grad_norm": 4.190752983093262, "learning_rate": 1.1048401607187565e-07, "loss": 0.08525798320770264, "step": 36220 }, { "epoch": 0.9543534308189251, "grad_norm": 2.8901407718658447, "learning_rate": 1.0922293496053671e-07, "loss": 0.09251134991645812, "step": 36230 }, { "epoch": 0.9546168460634238, "grad_norm": 3.332538604736328, "learning_rate": 1.0796905278347092e-07, "loss": 0.08066227436065673, "step": 36240 }, { "epoch": 0.9548802613079226, "grad_norm": 3.408548355102539, "learning_rate": 1.0672237045330491e-07, "loss": 0.08460988998413085, "step": 36250 }, { "epoch": 0.9551436765524213, "grad_norm": 3.3742728233337402, "learning_rate": 1.0548288887743174e-07, "loss": 0.0769084632396698, "step": 36260 }, { "epoch": 0.9554070917969201, "grad_norm": 3.824032783508301, "learning_rate": 1.0425060895799865e-07, "loss": 0.07697364091873168, "step": 36270 }, { "epoch": 0.9556705070414188, "grad_norm": 4.217365264892578, "learning_rate": 1.0302553159190932e-07, "loss": 0.08638783693313598, "step": 36280 }, { "epoch": 0.9559339222859174, "grad_norm": 3.1134331226348877, "learning_rate": 1.0180765767083045e-07, "loss": 0.08496474027633667, "step": 36290 }, { "epoch": 0.9561973375304162, "grad_norm": 3.290674924850464, "learning_rate": 1.00596988081183e-07, "loss": 0.07196952104568481, "step": 36300 }, { "epoch": 0.9564607527749149, "grad_norm": 3.379718780517578, "learning_rate": 9.939352370414435e-08, "loss": 0.07535833716392518, "step": 36310 }, { "epoch": 0.9567241680194137, "grad_norm": 3.1187427043914795, "learning_rate": 9.819726541564711e-08, "loss": 0.08550009727478028, "step": 36320 }, { "epoch": 0.9569875832639124, "grad_norm": 2.814427614212036, "learning_rate": 9.700821408638039e-08, "loss": 0.08921580314636231, "step": 36330 }, { "epoch": 0.9572509985084112, "grad_norm": 2.6075828075408936, "learning_rate": 9.582637058178635e-08, "loss": 0.08373249769210815, "step": 36340 }, { "epoch": 0.9575144137529099, "grad_norm": 3.903782606124878, "learning_rate": 9.465173576206355e-08, "loss": 0.07348794341087342, "step": 36350 }, { "epoch": 0.9577778289974086, "grad_norm": 3.249020576477051, "learning_rate": 9.348431048215923e-08, "loss": 0.0837772011756897, "step": 36360 }, { "epoch": 0.9580412442419074, "grad_norm": 3.403526782989502, "learning_rate": 9.232409559177701e-08, "loss": 0.08314073085784912, "step": 36370 }, { "epoch": 0.9583046594864061, "grad_norm": 3.9898242950439453, "learning_rate": 9.117109193537254e-08, "loss": 0.08913297057151795, "step": 36380 }, { "epoch": 0.9585680747309049, "grad_norm": 3.2273380756378174, "learning_rate": 9.002530035215007e-08, "loss": 0.08609622716903687, "step": 36390 }, { "epoch": 0.9588314899754036, "grad_norm": 4.269287109375, "learning_rate": 8.888672167606693e-08, "loss": 0.08323277235031128, "step": 36400 }, { "epoch": 0.9590949052199024, "grad_norm": 3.5089168548583984, "learning_rate": 8.775535673582913e-08, "loss": 0.07605867385864258, "step": 36410 }, { "epoch": 0.9593583204644011, "grad_norm": 4.35137939453125, "learning_rate": 8.663120635489241e-08, "loss": 0.07801430225372315, "step": 36420 }, { "epoch": 0.9596217357088999, "grad_norm": 2.956132650375366, "learning_rate": 8.55142713514645e-08, "loss": 0.08809442520141601, "step": 36430 }, { "epoch": 0.9598851509533985, "grad_norm": 4.4333648681640625, "learning_rate": 8.440455253849622e-08, "loss": 0.09253816604614258, "step": 36440 }, { "epoch": 0.9601485661978972, "grad_norm": 2.550842046737671, "learning_rate": 8.330205072369035e-08, "loss": 0.09530552625656127, "step": 36450 }, { "epoch": 0.960411981442396, "grad_norm": 3.742039442062378, "learning_rate": 8.22067667094939e-08, "loss": 0.09099365472793579, "step": 36460 }, { "epoch": 0.9606753966868947, "grad_norm": 2.824193239212036, "learning_rate": 8.11187012931025e-08, "loss": 0.08593811988830566, "step": 36470 }, { "epoch": 0.9609388119313935, "grad_norm": 3.3027782440185547, "learning_rate": 8.003785526645824e-08, "loss": 0.08317024111747742, "step": 36480 }, { "epoch": 0.9612022271758922, "grad_norm": 5.35274076461792, "learning_rate": 7.896422941624626e-08, "loss": 0.08354168534278869, "step": 36490 }, { "epoch": 0.961465642420391, "grad_norm": 3.721208095550537, "learning_rate": 7.789782452389705e-08, "loss": 0.0831300973892212, "step": 36500 }, { "epoch": 0.9617290576648897, "grad_norm": 3.9565184116363525, "learning_rate": 7.683864136558749e-08, "loss": 0.0841926634311676, "step": 36510 }, { "epoch": 0.9619924729093885, "grad_norm": 4.403808116912842, "learning_rate": 7.578668071223649e-08, "loss": 0.08312062621116638, "step": 36520 }, { "epoch": 0.9622558881538872, "grad_norm": 4.4483232498168945, "learning_rate": 7.474194332950602e-08, "loss": 0.0911523461341858, "step": 36530 }, { "epoch": 0.9625193033983859, "grad_norm": 3.3329203128814697, "learning_rate": 7.370442997780114e-08, "loss": 0.07601130604743958, "step": 36540 }, { "epoch": 0.9627827186428847, "grad_norm": 2.787531614303589, "learning_rate": 7.267414141227002e-08, "loss": 0.07714304924011231, "step": 36550 }, { "epoch": 0.9630461338873834, "grad_norm": 2.8296806812286377, "learning_rate": 7.165107838279838e-08, "loss": 0.09041921496391296, "step": 36560 }, { "epoch": 0.9633095491318822, "grad_norm": 2.8307437896728516, "learning_rate": 7.063524163401836e-08, "loss": 0.09038071632385254, "step": 36570 }, { "epoch": 0.9635729643763808, "grad_norm": 4.040412425994873, "learning_rate": 6.962663190529851e-08, "loss": 0.0888642430305481, "step": 36580 }, { "epoch": 0.9638363796208796, "grad_norm": 3.608409881591797, "learning_rate": 6.862524993074825e-08, "loss": 0.07175285816192627, "step": 36590 }, { "epoch": 0.9640997948653783, "grad_norm": 3.9100334644317627, "learning_rate": 6.763109643921683e-08, "loss": 0.08014992475509644, "step": 36600 }, { "epoch": 0.9643632101098771, "grad_norm": 3.636239528656006, "learning_rate": 6.664417215429319e-08, "loss": 0.10354714393615723, "step": 36610 }, { "epoch": 0.9646266253543758, "grad_norm": 3.478410243988037, "learning_rate": 6.566447779430163e-08, "loss": 0.08182239532470703, "step": 36620 }, { "epoch": 0.9648900405988745, "grad_norm": 2.789543390274048, "learning_rate": 6.469201407230729e-08, "loss": 0.0902144432067871, "step": 36630 }, { "epoch": 0.9651534558433733, "grad_norm": 3.5071120262145996, "learning_rate": 6.372678169611068e-08, "loss": 0.07805896401405335, "step": 36640 }, { "epoch": 0.965416871087872, "grad_norm": 4.441227912902832, "learning_rate": 6.276878136824982e-08, "loss": 0.07816243171691895, "step": 36650 }, { "epoch": 0.9656802863323708, "grad_norm": 2.3020682334899902, "learning_rate": 6.181801378599805e-08, "loss": 0.07663674354553222, "step": 36660 }, { "epoch": 0.9659437015768695, "grad_norm": 3.179515838623047, "learning_rate": 6.087447964136517e-08, "loss": 0.08028925657272339, "step": 36670 }, { "epoch": 0.9662071168213683, "grad_norm": 3.873351573944092, "learning_rate": 5.993817962109627e-08, "loss": 0.0960257112979889, "step": 36680 }, { "epoch": 0.966470532065867, "grad_norm": 3.7292115688323975, "learning_rate": 5.9009114406671787e-08, "loss": 0.08256404399871826, "step": 36690 }, { "epoch": 0.9667339473103658, "grad_norm": 4.834585666656494, "learning_rate": 5.8087284674304136e-08, "loss": 0.0871350884437561, "step": 36700 }, { "epoch": 0.9669973625548645, "grad_norm": 3.9148576259613037, "learning_rate": 5.7172691094942165e-08, "loss": 0.07400652170181274, "step": 36710 }, { "epoch": 0.9672607777993631, "grad_norm": 3.513040065765381, "learning_rate": 5.626533433426451e-08, "loss": 0.09040920734405518, "step": 36720 }, { "epoch": 0.967524193043862, "grad_norm": 3.987504005432129, "learning_rate": 5.5365215052686216e-08, "loss": 0.0909152090549469, "step": 36730 }, { "epoch": 0.9677876082883606, "grad_norm": 2.5424656867980957, "learning_rate": 5.4472333905352115e-08, "loss": 0.08081549406051636, "step": 36740 }, { "epoch": 0.9680510235328594, "grad_norm": 3.0678207874298096, "learning_rate": 5.3586691542140135e-08, "loss": 0.07455427646636963, "step": 36750 }, { "epoch": 0.9683144387773581, "grad_norm": 3.4890859127044678, "learning_rate": 5.2708288607659084e-08, "loss": 0.07640300989151001, "step": 36760 }, { "epoch": 0.9685778540218569, "grad_norm": 3.436936378479004, "learning_rate": 5.183712574124755e-08, "loss": 0.07987082004547119, "step": 36770 }, { "epoch": 0.9688412692663556, "grad_norm": 3.7764840126037598, "learning_rate": 5.0973203576976104e-08, "loss": 0.07973310947418213, "step": 36780 }, { "epoch": 0.9691046845108544, "grad_norm": 3.245797872543335, "learning_rate": 5.011652274364509e-08, "loss": 0.08673604726791381, "step": 36790 }, { "epoch": 0.9693680997553531, "grad_norm": 3.592090129852295, "learning_rate": 4.926708386478352e-08, "loss": 0.07665194272994995, "step": 36800 }, { "epoch": 0.9696315149998518, "grad_norm": 4.21805477142334, "learning_rate": 4.8424887558649045e-08, "loss": 0.09435230493545532, "step": 36810 }, { "epoch": 0.9698949302443506, "grad_norm": 2.651928663253784, "learning_rate": 4.7589934438228017e-08, "loss": 0.09383895993232727, "step": 36820 }, { "epoch": 0.9701583454888493, "grad_norm": 3.6219372749328613, "learning_rate": 4.676222511123762e-08, "loss": 0.09558022618293763, "step": 36830 }, { "epoch": 0.9704217607333481, "grad_norm": 5.036308288574219, "learning_rate": 4.59417601801182e-08, "loss": 0.09406188726425171, "step": 36840 }, { "epoch": 0.9706851759778468, "grad_norm": 3.5273091793060303, "learning_rate": 4.512854024203983e-08, "loss": 0.10500516891479492, "step": 36850 }, { "epoch": 0.9709485912223456, "grad_norm": 2.6987054347991943, "learning_rate": 4.432256588890016e-08, "loss": 0.07790957689285279, "step": 36860 }, { "epoch": 0.9712120064668442, "grad_norm": 3.5836498737335205, "learning_rate": 4.352383770732216e-08, "loss": 0.07565362453460693, "step": 36870 }, { "epoch": 0.9714754217113429, "grad_norm": 5.095977306365967, "learning_rate": 4.273235627865302e-08, "loss": 0.08167170286178589, "step": 36880 }, { "epoch": 0.9717388369558417, "grad_norm": 3.4837374687194824, "learning_rate": 4.19481221789686e-08, "loss": 0.08992707133293151, "step": 36890 }, { "epoch": 0.9720022522003404, "grad_norm": 3.561575174331665, "learning_rate": 4.117113597906675e-08, "loss": 0.08245635032653809, "step": 36900 }, { "epoch": 0.9722656674448392, "grad_norm": 3.0392439365386963, "learning_rate": 4.040139824447398e-08, "loss": 0.08046376705169678, "step": 36910 }, { "epoch": 0.9725290826893379, "grad_norm": 3.796013116836548, "learning_rate": 3.963890953543659e-08, "loss": 0.08400149345397949, "step": 36920 }, { "epoch": 0.9727924979338367, "grad_norm": 3.7544546127319336, "learning_rate": 3.88836704069262e-08, "loss": 0.07627551555633545, "step": 36930 }, { "epoch": 0.9730559131783354, "grad_norm": 3.7214512825012207, "learning_rate": 3.813568140864088e-08, "loss": 0.09899497032165527, "step": 36940 }, { "epoch": 0.9733193284228342, "grad_norm": 2.9571545124053955, "learning_rate": 3.739494308499736e-08, "loss": 0.09374057054519654, "step": 36950 }, { "epoch": 0.9735827436673329, "grad_norm": 3.2792117595672607, "learning_rate": 3.66614559751377e-08, "loss": 0.0801843285560608, "step": 36960 }, { "epoch": 0.9738461589118316, "grad_norm": 3.530320882797241, "learning_rate": 3.593522061292598e-08, "loss": 0.06336323022842408, "step": 36970 }, { "epoch": 0.9741095741563304, "grad_norm": 3.8483784198760986, "learning_rate": 3.521623752694603e-08, "loss": 0.08603591322898865, "step": 36980 }, { "epoch": 0.9743729894008291, "grad_norm": 4.537707805633545, "learning_rate": 3.4504507240507027e-08, "loss": 0.08166532516479492, "step": 36990 }, { "epoch": 0.9746364046453279, "grad_norm": 4.257575035095215, "learning_rate": 3.380003027163459e-08, "loss": 0.07923673391342163, "step": 37000 }, { "epoch": 0.9748998198898265, "grad_norm": 3.3928463459014893, "learning_rate": 3.310280713307856e-08, "loss": 0.07542917728424073, "step": 37010 }, { "epoch": 0.9751632351343253, "grad_norm": 4.730133533477783, "learning_rate": 3.241283833230746e-08, "loss": 0.07578556537628174, "step": 37020 }, { "epoch": 0.975426650378824, "grad_norm": 3.541816473007202, "learning_rate": 3.1730124371511794e-08, "loss": 0.07391899824142456, "step": 37030 }, { "epoch": 0.9756900656233228, "grad_norm": 3.063847541809082, "learning_rate": 3.1054665747599635e-08, "loss": 0.07374464869499206, "step": 37040 }, { "epoch": 0.9759534808678215, "grad_norm": 4.168177604675293, "learning_rate": 3.038646295219771e-08, "loss": 0.10145843029022217, "step": 37050 }, { "epoch": 0.9762168961123202, "grad_norm": 3.681044578552246, "learning_rate": 2.972551647165367e-08, "loss": 0.092701256275177, "step": 37060 }, { "epoch": 0.976480311356819, "grad_norm": 4.0198893547058105, "learning_rate": 2.9071826787033797e-08, "loss": 0.09479172229766845, "step": 37070 }, { "epoch": 0.9767437266013177, "grad_norm": 2.8580422401428223, "learning_rate": 2.8425394374120842e-08, "loss": 0.07860974073410035, "step": 37080 }, { "epoch": 0.9770071418458165, "grad_norm": 3.515366554260254, "learning_rate": 2.7786219703415108e-08, "loss": 0.07782836556434632, "step": 37090 }, { "epoch": 0.9772705570903152, "grad_norm": 3.0366175174713135, "learning_rate": 2.715430324013779e-08, "loss": 0.07275362014770508, "step": 37100 }, { "epoch": 0.977533972334814, "grad_norm": 3.9504001140594482, "learning_rate": 2.6529645444223206e-08, "loss": 0.08102408647537232, "step": 37110 }, { "epoch": 0.9777973875793127, "grad_norm": 3.0301830768585205, "learning_rate": 2.5912246770324333e-08, "loss": 0.08298348188400269, "step": 37120 }, { "epoch": 0.9780608028238115, "grad_norm": 3.8687050342559814, "learning_rate": 2.5302107667811716e-08, "loss": 0.08792224526405334, "step": 37130 }, { "epoch": 0.9783242180683102, "grad_norm": 4.277554512023926, "learning_rate": 2.469922858076901e-08, "loss": 0.0909547746181488, "step": 37140 }, { "epoch": 0.9785876333128088, "grad_norm": 6.5058064460754395, "learning_rate": 2.410360994799965e-08, "loss": 0.09240667819976807, "step": 37150 }, { "epoch": 0.9788510485573076, "grad_norm": 4.101076602935791, "learning_rate": 2.351525220301909e-08, "loss": 0.08693333268165589, "step": 37160 }, { "epoch": 0.9791144638018063, "grad_norm": 3.027273178100586, "learning_rate": 2.293415577406144e-08, "loss": 0.09951878786087036, "step": 37170 }, { "epoch": 0.9793778790463051, "grad_norm": 4.475859642028809, "learning_rate": 2.2360321084071712e-08, "loss": 0.07666131258010864, "step": 37180 }, { "epoch": 0.9796412942908038, "grad_norm": 3.071180820465088, "learning_rate": 2.179374855071137e-08, "loss": 0.08867835998535156, "step": 37190 }, { "epoch": 0.9799047095353026, "grad_norm": 3.3043200969696045, "learning_rate": 2.123443858635832e-08, "loss": 0.07895694375038147, "step": 37200 }, { "epoch": 0.9801681247798013, "grad_norm": 3.0482540130615234, "learning_rate": 2.068239159810137e-08, "loss": 0.08902370929718018, "step": 37210 }, { "epoch": 0.9804315400243001, "grad_norm": 4.75997257232666, "learning_rate": 2.0137607987743558e-08, "loss": 0.083711177110672, "step": 37220 }, { "epoch": 0.9806949552687988, "grad_norm": 4.430932998657227, "learning_rate": 1.9600088151803255e-08, "loss": 0.07371033430099487, "step": 37230 }, { "epoch": 0.9809583705132975, "grad_norm": 3.801527500152588, "learning_rate": 1.906983248150973e-08, "loss": 0.07636693716049195, "step": 37240 }, { "epoch": 0.9812217857577963, "grad_norm": 2.796309471130371, "learning_rate": 1.854684136280538e-08, "loss": 0.08025997877120972, "step": 37250 }, { "epoch": 0.981485201002295, "grad_norm": 4.757577419281006, "learning_rate": 1.8031115176346815e-08, "loss": 0.09313318729400635, "step": 37260 }, { "epoch": 0.9817486162467938, "grad_norm": 2.9475083351135254, "learning_rate": 1.752265429750044e-08, "loss": 0.08498886227607727, "step": 37270 }, { "epoch": 0.9820120314912925, "grad_norm": 4.222609043121338, "learning_rate": 1.7021459096348004e-08, "loss": 0.08530120849609375, "step": 37280 }, { "epoch": 0.9822754467357913, "grad_norm": 3.6331121921539307, "learning_rate": 1.652752993767992e-08, "loss": 0.08369654417037964, "step": 37290 }, { "epoch": 0.98253886198029, "grad_norm": 3.30229115486145, "learning_rate": 1.604086718099973e-08, "loss": 0.08782702684402466, "step": 37300 }, { "epoch": 0.9828022772247887, "grad_norm": 4.863348484039307, "learning_rate": 1.556147118051965e-08, "loss": 0.08793606758117675, "step": 37310 }, { "epoch": 0.9830656924692874, "grad_norm": 3.288719892501831, "learning_rate": 1.508934228516723e-08, "loss": 0.07925713658332825, "step": 37320 }, { "epoch": 0.9833291077137861, "grad_norm": 4.669800281524658, "learning_rate": 1.4624480838576483e-08, "loss": 0.09114296436309814, "step": 37330 }, { "epoch": 0.9835925229582849, "grad_norm": 2.759136199951172, "learning_rate": 1.4166887179094535e-08, "loss": 0.06705195903778076, "step": 37340 }, { "epoch": 0.9838559382027836, "grad_norm": 3.277724027633667, "learning_rate": 1.3716561639777192e-08, "loss": 0.08704003691673279, "step": 37350 }, { "epoch": 0.9841193534472824, "grad_norm": 4.529204845428467, "learning_rate": 1.3273504548392269e-08, "loss": 0.08393763899803161, "step": 37360 }, { "epoch": 0.9843827686917811, "grad_norm": 3.505577325820923, "learning_rate": 1.2837716227414032e-08, "loss": 0.07413549423217773, "step": 37370 }, { "epoch": 0.9846461839362799, "grad_norm": 2.6630425453186035, "learning_rate": 1.240919699402765e-08, "loss": 0.08374332189559937, "step": 37380 }, { "epoch": 0.9849095991807786, "grad_norm": 3.3014328479766846, "learning_rate": 1.1987947160130297e-08, "loss": 0.08510485291481018, "step": 37390 }, { "epoch": 0.9851730144252773, "grad_norm": 4.4060845375061035, "learning_rate": 1.1573967032324495e-08, "loss": 0.09638545513153077, "step": 37400 }, { "epoch": 0.9854364296697761, "grad_norm": 3.8934273719787598, "learning_rate": 1.1167256911922552e-08, "loss": 0.07618547677993774, "step": 37410 }, { "epoch": 0.9856998449142748, "grad_norm": 3.626394748687744, "learning_rate": 1.0767817094945454e-08, "loss": 0.08563627004623413, "step": 37420 }, { "epoch": 0.9859632601587736, "grad_norm": 3.511981964111328, "learning_rate": 1.0375647872123974e-08, "loss": 0.08079684972763061, "step": 37430 }, { "epoch": 0.9862266754032722, "grad_norm": 5.940303325653076, "learning_rate": 9.990749528894227e-09, "loss": 0.08409411907196045, "step": 37440 }, { "epoch": 0.986490090647771, "grad_norm": 2.9966113567352295, "learning_rate": 9.613122345403236e-09, "loss": 0.08685506582260132, "step": 37450 }, { "epoch": 0.9867535058922697, "grad_norm": 3.435263156890869, "learning_rate": 9.24276659650336e-09, "loss": 0.08394573330879211, "step": 37460 }, { "epoch": 0.9870169211367685, "grad_norm": 3.76529598236084, "learning_rate": 8.879682551755643e-09, "loss": 0.08725910782814025, "step": 37470 }, { "epoch": 0.9872803363812672, "grad_norm": 2.798969268798828, "learning_rate": 8.523870475429797e-09, "loss": 0.08028888702392578, "step": 37480 }, { "epoch": 0.9875437516257659, "grad_norm": 3.2598459720611572, "learning_rate": 8.175330626498668e-09, "loss": 0.07529643774032593, "step": 37490 }, { "epoch": 0.9878071668702647, "grad_norm": 5.039193630218506, "learning_rate": 7.834063258645997e-09, "loss": 0.09003254175186157, "step": 37500 }, { "epoch": 0.9880705821147634, "grad_norm": 3.328373432159424, "learning_rate": 7.500068620260869e-09, "loss": 0.0951402723789215, "step": 37510 }, { "epoch": 0.9883339973592622, "grad_norm": 3.1802775859832764, "learning_rate": 7.173346954438831e-09, "loss": 0.0817640781402588, "step": 37520 }, { "epoch": 0.9885974126037609, "grad_norm": 3.714486598968506, "learning_rate": 6.853898498982992e-09, "loss": 0.08045803308486939, "step": 37530 }, { "epoch": 0.9888608278482597, "grad_norm": 2.737558603286743, "learning_rate": 6.541723486399587e-09, "loss": 0.07878758311271668, "step": 37540 }, { "epoch": 0.9891242430927584, "grad_norm": 3.9417641162872314, "learning_rate": 6.236822143904642e-09, "loss": 0.07443664669990539, "step": 37550 }, { "epoch": 0.9893876583372572, "grad_norm": 2.6929421424865723, "learning_rate": 5.939194693418415e-09, "loss": 0.07679800987243653, "step": 37560 }, { "epoch": 0.9896510735817559, "grad_norm": 4.58318567276001, "learning_rate": 5.648841351566514e-09, "loss": 0.08762773275375366, "step": 37570 }, { "epoch": 0.9899144888262545, "grad_norm": 5.092545509338379, "learning_rate": 5.365762329681002e-09, "loss": 0.07866828441619873, "step": 37580 }, { "epoch": 0.9901779040707533, "grad_norm": 2.7590205669403076, "learning_rate": 5.089957833798176e-09, "loss": 0.08652526736259461, "step": 37590 }, { "epoch": 0.990441319315252, "grad_norm": 3.362258195877075, "learning_rate": 4.8214280646619035e-09, "loss": 0.08739187717437744, "step": 37600 }, { "epoch": 0.9907047345597508, "grad_norm": 2.909883975982666, "learning_rate": 4.560173217718067e-09, "loss": 0.07971587181091308, "step": 37610 }, { "epoch": 0.9909681498042495, "grad_norm": 4.205813407897949, "learning_rate": 4.3061934831212235e-09, "loss": 0.07072982788085938, "step": 37620 }, { "epoch": 0.9912315650487483, "grad_norm": 2.935624361038208, "learning_rate": 4.0594890457268384e-09, "loss": 0.07116802334785462, "step": 37630 }, { "epoch": 0.991494980293247, "grad_norm": 5.943000793457031, "learning_rate": 3.820060085097943e-09, "loss": 0.08554408550262452, "step": 37640 }, { "epoch": 0.9917583955377458, "grad_norm": 5.335630893707275, "learning_rate": 3.587906775501804e-09, "loss": 0.07806146144866943, "step": 37650 }, { "epoch": 0.9920218107822445, "grad_norm": 2.513282299041748, "learning_rate": 3.363029285907704e-09, "loss": 0.07814821004867553, "step": 37660 }, { "epoch": 0.9922852260267432, "grad_norm": 3.9476637840270996, "learning_rate": 3.145427779993604e-09, "loss": 0.09550424814224243, "step": 37670 }, { "epoch": 0.992548641271242, "grad_norm": 3.2351129055023193, "learning_rate": 2.935102416138369e-09, "loss": 0.08475596904754638, "step": 37680 }, { "epoch": 0.9928120565157407, "grad_norm": 2.9832234382629395, "learning_rate": 2.732053347425101e-09, "loss": 0.07765300869941712, "step": 37690 }, { "epoch": 0.9930754717602395, "grad_norm": 3.62060284614563, "learning_rate": 2.536280721642248e-09, "loss": 0.08731594085693359, "step": 37700 }, { "epoch": 0.9933388870047382, "grad_norm": 4.05382776260376, "learning_rate": 2.3477846812813843e-09, "loss": 0.07800358533859253, "step": 37710 }, { "epoch": 0.993602302249237, "grad_norm": 3.262683629989624, "learning_rate": 2.16656536353832e-09, "loss": 0.08350295424461365, "step": 37720 }, { "epoch": 0.9938657174937356, "grad_norm": 3.6598987579345703, "learning_rate": 1.9926229003119912e-09, "loss": 0.07640161514282226, "step": 37730 }, { "epoch": 0.9941291327382344, "grad_norm": 4.363567352294922, "learning_rate": 1.8259574182055707e-09, "loss": 0.08886727094650268, "step": 37740 }, { "epoch": 0.9943925479827331, "grad_norm": 3.3178560733795166, "learning_rate": 1.6665690385253564e-09, "loss": 0.08084766268730163, "step": 37750 }, { "epoch": 0.9946559632272318, "grad_norm": 4.141059398651123, "learning_rate": 1.5144578772807728e-09, "loss": 0.08237512111663818, "step": 37760 }, { "epoch": 0.9949193784717306, "grad_norm": 2.715752124786377, "learning_rate": 1.36962404518437e-09, "loss": 0.09294235110282897, "step": 37770 }, { "epoch": 0.9951827937162293, "grad_norm": 4.137963771820068, "learning_rate": 1.232067647654045e-09, "loss": 0.0935674786567688, "step": 37780 }, { "epoch": 0.9954462089607281, "grad_norm": 3.001352310180664, "learning_rate": 1.1017887848085995e-09, "loss": 0.0820215106010437, "step": 37790 }, { "epoch": 0.9957096242052268, "grad_norm": 3.5653367042541504, "learning_rate": 9.78787551469962e-10, "loss": 0.07771285772323608, "step": 37800 }, { "epoch": 0.9959730394497256, "grad_norm": 3.7755491733551025, "learning_rate": 8.630640371642963e-10, "loss": 0.08483846783638001, "step": 37810 }, { "epoch": 0.9962364546942243, "grad_norm": 3.881047487258911, "learning_rate": 7.546183261197826e-10, "loss": 0.08562873005867004, "step": 37820 }, { "epoch": 0.9964998699387231, "grad_norm": 3.326624631881714, "learning_rate": 6.534504972688372e-10, "loss": 0.08777405023574829, "step": 37830 }, { "epoch": 0.9967632851832218, "grad_norm": 2.958233594894409, "learning_rate": 5.595606242447815e-10, "loss": 0.09797561764717103, "step": 37840 }, { "epoch": 0.9970267004277205, "grad_norm": 3.9857170581817627, "learning_rate": 4.729487753851736e-10, "loss": 0.0843148410320282, "step": 37850 }, { "epoch": 0.9972901156722193, "grad_norm": 4.17692232131958, "learning_rate": 3.9361501372958686e-10, "loss": 0.08913063406944274, "step": 37860 }, { "epoch": 0.997553530916718, "grad_norm": 3.526484727859497, "learning_rate": 3.2155939702072093e-10, "loss": 0.08268949985504151, "step": 37870 }, { "epoch": 0.9978169461612167, "grad_norm": 2.720268487930298, "learning_rate": 2.5678197770329094e-10, "loss": 0.09056673645973205, "step": 37880 }, { "epoch": 0.9980803614057154, "grad_norm": 4.037978172302246, "learning_rate": 1.9928280292624835e-10, "loss": 0.09141225814819336, "step": 37890 }, { "epoch": 0.9983437766502142, "grad_norm": 3.0480599403381348, "learning_rate": 1.4906191453833985e-10, "loss": 0.08581740856170654, "step": 37900 }, { "epoch": 0.9986071918947129, "grad_norm": 3.957517147064209, "learning_rate": 1.061193490936585e-10, "loss": 0.07078821659088134, "step": 37910 }, { "epoch": 0.9988706071392117, "grad_norm": 3.383472204208374, "learning_rate": 7.045513784831315e-11, "loss": 0.07636631727218628, "step": 37920 }, { "epoch": 0.9991340223837104, "grad_norm": 3.056386709213257, "learning_rate": 4.206930675820786e-11, "loss": 0.0739935040473938, "step": 37930 }, { "epoch": 0.9993974376282091, "grad_norm": 3.414250135421753, "learning_rate": 2.0961876485703358e-11, "loss": 0.07236528992652894, "step": 37940 }, { "epoch": 0.9996608528727079, "grad_norm": 2.9931747913360596, "learning_rate": 7.132862392955631e-12, "loss": 0.07893968820571899, "step": 37950 }, { "epoch": 0.9999242681172066, "grad_norm": 3.3766281604766846, "learning_rate": 5.822745441363964e-13, "loss": 0.07546517848968506, "step": 37960 }, { "epoch": 1.0, "step": 37963, "total_flos": 1.991421634663741e+20, "train_loss": 0.14788081611765483, "train_runtime": 310786.6532, "train_samples_per_second": 7.818, "train_steps_per_second": 0.122 } ], "logging_steps": 10, "max_steps": 37963, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 1000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 1.991421634663741e+20, "train_batch_size": 1, "trial_name": null, "trial_params": null }