diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,25934 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.9746364046453279, + "eval_steps": 500, + "global_step": 37000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.00026341524449873725, + "grad_norm": 1240.0467529296875, + "learning_rate": 1.5803336259877089e-07, + "loss": 2.360714340209961, + "step": 10 + }, + { + "epoch": 0.0005268304889974745, + "grad_norm": 51.95969772338867, + "learning_rate": 3.336259877085163e-07, + "loss": 0.679102611541748, + "step": 20 + }, + { + "epoch": 0.0007902457334962118, + "grad_norm": 30.754972457885742, + "learning_rate": 5.092186128182617e-07, + "loss": 0.2968589305877686, + "step": 30 + }, + { + "epoch": 0.001053660977994949, + "grad_norm": 13.564824104309082, + "learning_rate": 6.848112379280071e-07, + "loss": 0.24760403633117675, + "step": 40 + }, + { + "epoch": 0.0013170762224936862, + "grad_norm": 20.055593490600586, + "learning_rate": 8.604038630377525e-07, + "loss": 0.2550562143325806, + "step": 50 + }, + { + "epoch": 0.0015804914669924235, + "grad_norm": 14.47984790802002, + "learning_rate": 1.035996488147498e-06, + "loss": 0.234423828125, + "step": 60 + }, + { + "epoch": 0.0018439067114911607, + "grad_norm": 11.39301586151123, + "learning_rate": 1.2115891132572432e-06, + "loss": 0.2202540397644043, + "step": 70 + }, + { + "epoch": 0.002107321955989898, + "grad_norm": 9.123312950134277, + "learning_rate": 1.3871817383669885e-06, + "loss": 0.22982375621795653, + "step": 80 + }, + { + "epoch": 0.002370737200488635, + "grad_norm": 13.056436538696289, + "learning_rate": 1.5627743634767342e-06, + "loss": 0.2120755910873413, + "step": 90 + }, + { + "epoch": 0.0026341524449873723, + "grad_norm": 13.004779815673828, + "learning_rate": 1.7383669885864795e-06, + "loss": 0.20886459350585937, + "step": 100 + }, + { + "epoch": 0.00289756768948611, + "grad_norm": 10.194211959838867, + "learning_rate": 1.913959613696225e-06, + "loss": 0.21107115745544433, + "step": 110 + }, + { + "epoch": 0.003160982933984847, + "grad_norm": 17.870107650756836, + "learning_rate": 2.08955223880597e-06, + "loss": 0.22807445526123046, + "step": 120 + }, + { + "epoch": 0.003424398178483584, + "grad_norm": 10.447258949279785, + "learning_rate": 2.265144863915716e-06, + "loss": 0.19623351097106934, + "step": 130 + }, + { + "epoch": 0.0036878134229823213, + "grad_norm": 15.005298614501953, + "learning_rate": 2.440737489025461e-06, + "loss": 0.2115079402923584, + "step": 140 + }, + { + "epoch": 0.0039512286674810585, + "grad_norm": 8.877615928649902, + "learning_rate": 2.6163301141352064e-06, + "loss": 0.21258363723754883, + "step": 150 + }, + { + "epoch": 0.004214643911979796, + "grad_norm": 16.82979965209961, + "learning_rate": 2.791922739244952e-06, + "loss": 0.22850914001464845, + "step": 160 + }, + { + "epoch": 0.004478059156478534, + "grad_norm": 12.637763023376465, + "learning_rate": 2.9675153643546974e-06, + "loss": 0.22500476837158204, + "step": 170 + }, + { + "epoch": 0.00474147440097727, + "grad_norm": 11.108145713806152, + "learning_rate": 3.1431079894644427e-06, + "loss": 0.22509567737579345, + "step": 180 + }, + { + "epoch": 0.005004889645476008, + "grad_norm": 7.171646595001221, + "learning_rate": 3.318700614574188e-06, + "loss": 0.20883538722991943, + "step": 190 + }, + { + "epoch": 0.005268304889974745, + "grad_norm": 12.457633018493652, + "learning_rate": 3.4942932396839337e-06, + "loss": 0.21761527061462402, + "step": 200 + }, + { + "epoch": 0.005531720134473482, + "grad_norm": 10.639338493347168, + "learning_rate": 3.669885864793679e-06, + "loss": 0.20572023391723632, + "step": 210 + }, + { + "epoch": 0.00579513537897222, + "grad_norm": 17.81014060974121, + "learning_rate": 3.845478489903424e-06, + "loss": 0.1994907259941101, + "step": 220 + }, + { + "epoch": 0.0060585506234709565, + "grad_norm": 10.481728553771973, + "learning_rate": 4.0210711150131695e-06, + "loss": 0.21125555038452148, + "step": 230 + }, + { + "epoch": 0.006321965867969694, + "grad_norm": 12.868586540222168, + "learning_rate": 4.196663740122916e-06, + "loss": 0.21690988540649414, + "step": 240 + }, + { + "epoch": 0.006585381112468432, + "grad_norm": 16.4504337310791, + "learning_rate": 4.37225636523266e-06, + "loss": 0.174403178691864, + "step": 250 + }, + { + "epoch": 0.006848796356967168, + "grad_norm": 14.680349349975586, + "learning_rate": 4.547848990342406e-06, + "loss": 0.21403129100799562, + "step": 260 + }, + { + "epoch": 0.007112211601465906, + "grad_norm": 6.039432048797607, + "learning_rate": 4.7234416154521515e-06, + "loss": 0.19299013614654542, + "step": 270 + }, + { + "epoch": 0.007375626845964643, + "grad_norm": 11.484891891479492, + "learning_rate": 4.899034240561897e-06, + "loss": 0.19696568250656127, + "step": 280 + }, + { + "epoch": 0.00763904209046338, + "grad_norm": 8.588067054748535, + "learning_rate": 5.074626865671642e-06, + "loss": 0.18822193145751953, + "step": 290 + }, + { + "epoch": 0.007902457334962117, + "grad_norm": 11.45522689819336, + "learning_rate": 5.250219490781387e-06, + "loss": 0.19861156940460206, + "step": 300 + }, + { + "epoch": 0.008165872579460855, + "grad_norm": 8.510603904724121, + "learning_rate": 5.425812115891133e-06, + "loss": 0.18854843378067015, + "step": 310 + }, + { + "epoch": 0.008429287823959592, + "grad_norm": 12.61191177368164, + "learning_rate": 5.601404741000879e-06, + "loss": 0.1974649667739868, + "step": 320 + }, + { + "epoch": 0.008692703068458329, + "grad_norm": 7.218238353729248, + "learning_rate": 5.776997366110624e-06, + "loss": 0.20580873489379883, + "step": 330 + }, + { + "epoch": 0.008956118312957067, + "grad_norm": 14.211240768432617, + "learning_rate": 5.9525899912203685e-06, + "loss": 0.19162254333496093, + "step": 340 + }, + { + "epoch": 0.009219533557455804, + "grad_norm": 10.297245979309082, + "learning_rate": 6.128182616330115e-06, + "loss": 0.21373462677001953, + "step": 350 + }, + { + "epoch": 0.00948294880195454, + "grad_norm": 9.502967834472656, + "learning_rate": 6.30377524143986e-06, + "loss": 0.1999206304550171, + "step": 360 + }, + { + "epoch": 0.00974636404645328, + "grad_norm": 8.1216459274292, + "learning_rate": 6.479367866549605e-06, + "loss": 0.2027184009552002, + "step": 370 + }, + { + "epoch": 0.010009779290952016, + "grad_norm": 5.453703880310059, + "learning_rate": 6.654960491659351e-06, + "loss": 0.20155735015869142, + "step": 380 + }, + { + "epoch": 0.010273194535450753, + "grad_norm": 12.771483421325684, + "learning_rate": 6.830553116769097e-06, + "loss": 0.1958877205848694, + "step": 390 + }, + { + "epoch": 0.01053660977994949, + "grad_norm": 6.244961738586426, + "learning_rate": 7.006145741878841e-06, + "loss": 0.1929736852645874, + "step": 400 + }, + { + "epoch": 0.010800025024448228, + "grad_norm": 4.223571300506592, + "learning_rate": 7.181738366988587e-06, + "loss": 0.18852312564849855, + "step": 410 + }, + { + "epoch": 0.011063440268946964, + "grad_norm": 6.50313663482666, + "learning_rate": 7.3573309920983325e-06, + "loss": 0.1951861023902893, + "step": 420 + }, + { + "epoch": 0.011326855513445701, + "grad_norm": 12.231367111206055, + "learning_rate": 7.532923617208078e-06, + "loss": 0.1786510705947876, + "step": 430 + }, + { + "epoch": 0.01159027075794444, + "grad_norm": 3.998512029647827, + "learning_rate": 7.708516242317823e-06, + "loss": 0.18833776712417602, + "step": 440 + }, + { + "epoch": 0.011853686002443176, + "grad_norm": 6.612815856933594, + "learning_rate": 7.88410886742757e-06, + "loss": 0.19663529396057128, + "step": 450 + }, + { + "epoch": 0.012117101246941913, + "grad_norm": 7.270054817199707, + "learning_rate": 8.059701492537314e-06, + "loss": 0.19169166088104247, + "step": 460 + }, + { + "epoch": 0.012380516491440651, + "grad_norm": 4.480962753295898, + "learning_rate": 8.23529411764706e-06, + "loss": 0.1817769765853882, + "step": 470 + }, + { + "epoch": 0.012643931735939388, + "grad_norm": 5.346428394317627, + "learning_rate": 8.410886742756806e-06, + "loss": 0.18909978866577148, + "step": 480 + }, + { + "epoch": 0.012907346980438125, + "grad_norm": 8.30456829071045, + "learning_rate": 8.58647936786655e-06, + "loss": 0.1927554965019226, + "step": 490 + }, + { + "epoch": 0.013170762224936863, + "grad_norm": 14.385656356811523, + "learning_rate": 8.762071992976295e-06, + "loss": 0.2078390598297119, + "step": 500 + }, + { + "epoch": 0.0134341774694356, + "grad_norm": 6.696582317352295, + "learning_rate": 8.937664618086041e-06, + "loss": 0.2070148229598999, + "step": 510 + }, + { + "epoch": 0.013697592713934337, + "grad_norm": 10.482494354248047, + "learning_rate": 9.113257243195787e-06, + "loss": 0.20279884338378906, + "step": 520 + }, + { + "epoch": 0.013961007958433075, + "grad_norm": 6.145002365112305, + "learning_rate": 9.288849868305532e-06, + "loss": 0.19506795406341554, + "step": 530 + }, + { + "epoch": 0.014224423202931812, + "grad_norm": 10.24714183807373, + "learning_rate": 9.464442493415278e-06, + "loss": 0.1966523051261902, + "step": 540 + }, + { + "epoch": 0.014487838447430549, + "grad_norm": 3.129317045211792, + "learning_rate": 9.640035118525024e-06, + "loss": 0.18744311332702637, + "step": 550 + }, + { + "epoch": 0.014751253691929285, + "grad_norm": 5.289870738983154, + "learning_rate": 9.815627743634768e-06, + "loss": 0.18227899074554443, + "step": 560 + }, + { + "epoch": 0.015014668936428024, + "grad_norm": 5.129745960235596, + "learning_rate": 9.991220368744513e-06, + "loss": 0.1917642831802368, + "step": 570 + }, + { + "epoch": 0.01527808418092676, + "grad_norm": 6.940797805786133, + "learning_rate": 1.0166812993854259e-05, + "loss": 0.1948378324508667, + "step": 580 + }, + { + "epoch": 0.015541499425425497, + "grad_norm": 7.858885288238525, + "learning_rate": 1.0342405618964005e-05, + "loss": 0.1823291540145874, + "step": 590 + }, + { + "epoch": 0.015804914669924234, + "grad_norm": 9.534207344055176, + "learning_rate": 1.0517998244073751e-05, + "loss": 0.18598108291625975, + "step": 600 + }, + { + "epoch": 0.016068329914422974, + "grad_norm": 5.870530605316162, + "learning_rate": 1.0693590869183494e-05, + "loss": 0.19221689701080322, + "step": 610 + }, + { + "epoch": 0.01633174515892171, + "grad_norm": 4.762645244598389, + "learning_rate": 1.086918349429324e-05, + "loss": 0.18140406608581544, + "step": 620 + }, + { + "epoch": 0.016595160403420448, + "grad_norm": 9.293000221252441, + "learning_rate": 1.1044776119402986e-05, + "loss": 0.20683131217956544, + "step": 630 + }, + { + "epoch": 0.016858575647919184, + "grad_norm": 7.308525085449219, + "learning_rate": 1.122036874451273e-05, + "loss": 0.1841539978981018, + "step": 640 + }, + { + "epoch": 0.01712199089241792, + "grad_norm": 7.309910297393799, + "learning_rate": 1.1395961369622477e-05, + "loss": 0.1871891975402832, + "step": 650 + }, + { + "epoch": 0.017385406136916658, + "grad_norm": 5.154829502105713, + "learning_rate": 1.1571553994732223e-05, + "loss": 0.21009836196899415, + "step": 660 + }, + { + "epoch": 0.017648821381415394, + "grad_norm": 4.8346686363220215, + "learning_rate": 1.1747146619841967e-05, + "loss": 0.19911305904388427, + "step": 670 + }, + { + "epoch": 0.017912236625914135, + "grad_norm": 4.647324085235596, + "learning_rate": 1.1922739244951713e-05, + "loss": 0.1919912338256836, + "step": 680 + }, + { + "epoch": 0.01817565187041287, + "grad_norm": 8.402735710144043, + "learning_rate": 1.209833187006146e-05, + "loss": 0.19697179794311523, + "step": 690 + }, + { + "epoch": 0.018439067114911608, + "grad_norm": 5.751430511474609, + "learning_rate": 1.2273924495171204e-05, + "loss": 0.1968095898628235, + "step": 700 + }, + { + "epoch": 0.018702482359410345, + "grad_norm": 8.435208320617676, + "learning_rate": 1.244951712028095e-05, + "loss": 0.1818108320236206, + "step": 710 + }, + { + "epoch": 0.01896589760390908, + "grad_norm": 7.318845748901367, + "learning_rate": 1.2625109745390696e-05, + "loss": 0.18747050762176515, + "step": 720 + }, + { + "epoch": 0.019229312848407818, + "grad_norm": 3.714348554611206, + "learning_rate": 1.2800702370500439e-05, + "loss": 0.18208258152008056, + "step": 730 + }, + { + "epoch": 0.01949272809290656, + "grad_norm": 7.584974765777588, + "learning_rate": 1.2976294995610185e-05, + "loss": 0.191927170753479, + "step": 740 + }, + { + "epoch": 0.019756143337405295, + "grad_norm": 5.960648059844971, + "learning_rate": 1.315188762071993e-05, + "loss": 0.2015866756439209, + "step": 750 + }, + { + "epoch": 0.02001955858190403, + "grad_norm": 8.6318941116333, + "learning_rate": 1.3327480245829676e-05, + "loss": 0.1974639415740967, + "step": 760 + }, + { + "epoch": 0.02028297382640277, + "grad_norm": 7.8056206703186035, + "learning_rate": 1.3503072870939422e-05, + "loss": 0.20706455707550048, + "step": 770 + }, + { + "epoch": 0.020546389070901505, + "grad_norm": 3.3132433891296387, + "learning_rate": 1.3678665496049166e-05, + "loss": 0.19320933818817138, + "step": 780 + }, + { + "epoch": 0.020809804315400242, + "grad_norm": 4.5514678955078125, + "learning_rate": 1.3854258121158912e-05, + "loss": 0.18315668106079103, + "step": 790 + }, + { + "epoch": 0.02107321955989898, + "grad_norm": 14.86097240447998, + "learning_rate": 1.4029850746268658e-05, + "loss": 0.20814051628112792, + "step": 800 + }, + { + "epoch": 0.02133663480439772, + "grad_norm": 4.936925888061523, + "learning_rate": 1.4205443371378403e-05, + "loss": 0.18723890781402588, + "step": 810 + }, + { + "epoch": 0.021600050048896455, + "grad_norm": 3.4291982650756836, + "learning_rate": 1.4381035996488149e-05, + "loss": 0.18011690378189088, + "step": 820 + }, + { + "epoch": 0.021863465293395192, + "grad_norm": 2.619142532348633, + "learning_rate": 1.4556628621597895e-05, + "loss": 0.18511075973510743, + "step": 830 + }, + { + "epoch": 0.02212688053789393, + "grad_norm": 3.4120090007781982, + "learning_rate": 1.4732221246707638e-05, + "loss": 0.18523793220520018, + "step": 840 + }, + { + "epoch": 0.022390295782392666, + "grad_norm": 5.595027446746826, + "learning_rate": 1.4907813871817384e-05, + "loss": 0.18780069351196288, + "step": 850 + }, + { + "epoch": 0.022653711026891402, + "grad_norm": 3.570603847503662, + "learning_rate": 1.508340649692713e-05, + "loss": 0.1813866376876831, + "step": 860 + }, + { + "epoch": 0.022917126271390142, + "grad_norm": 5.158076286315918, + "learning_rate": 1.5258999122036876e-05, + "loss": 0.1915258526802063, + "step": 870 + }, + { + "epoch": 0.02318054151588888, + "grad_norm": 4.592512607574463, + "learning_rate": 1.5434591747146622e-05, + "loss": 0.20339436531066896, + "step": 880 + }, + { + "epoch": 0.023443956760387616, + "grad_norm": 6.081757068634033, + "learning_rate": 1.561018437225637e-05, + "loss": 0.17982156276702882, + "step": 890 + }, + { + "epoch": 0.023707372004886353, + "grad_norm": 5.188660621643066, + "learning_rate": 1.578577699736611e-05, + "loss": 0.1751444935798645, + "step": 900 + }, + { + "epoch": 0.02397078724938509, + "grad_norm": 2.592183828353882, + "learning_rate": 1.5961369622475857e-05, + "loss": 0.19086190462112426, + "step": 910 + }, + { + "epoch": 0.024234202493883826, + "grad_norm": 10.466522216796875, + "learning_rate": 1.6136962247585604e-05, + "loss": 0.18863316774368286, + "step": 920 + }, + { + "epoch": 0.024497617738382566, + "grad_norm": 3.484884023666382, + "learning_rate": 1.6312554872695346e-05, + "loss": 0.19139750003814698, + "step": 930 + }, + { + "epoch": 0.024761032982881303, + "grad_norm": 3.475672483444214, + "learning_rate": 1.6488147497805093e-05, + "loss": 0.18417433500289918, + "step": 940 + }, + { + "epoch": 0.02502444822738004, + "grad_norm": 4.305137634277344, + "learning_rate": 1.666374012291484e-05, + "loss": 0.1737154722213745, + "step": 950 + }, + { + "epoch": 0.025287863471878776, + "grad_norm": 4.229822635650635, + "learning_rate": 1.6839332748024585e-05, + "loss": 0.1825109004974365, + "step": 960 + }, + { + "epoch": 0.025551278716377513, + "grad_norm": 2.1458518505096436, + "learning_rate": 1.701492537313433e-05, + "loss": 0.1716059923171997, + "step": 970 + }, + { + "epoch": 0.02581469396087625, + "grad_norm": 8.936372756958008, + "learning_rate": 1.7190517998244077e-05, + "loss": 0.20058987140655518, + "step": 980 + }, + { + "epoch": 0.026078109205374986, + "grad_norm": 4.288111209869385, + "learning_rate": 1.736611062335382e-05, + "loss": 0.18617603778839112, + "step": 990 + }, + { + "epoch": 0.026341524449873727, + "grad_norm": 6.355926513671875, + "learning_rate": 1.7541703248463566e-05, + "loss": 0.19094316959381102, + "step": 1000 + }, + { + "epoch": 0.026604939694372463, + "grad_norm": 4.050109386444092, + "learning_rate": 1.771729587357331e-05, + "loss": 0.19923408031463624, + "step": 1010 + }, + { + "epoch": 0.0268683549388712, + "grad_norm": 5.037466526031494, + "learning_rate": 1.7892888498683055e-05, + "loss": 0.19317023754119872, + "step": 1020 + }, + { + "epoch": 0.027131770183369937, + "grad_norm": 4.274409770965576, + "learning_rate": 1.80684811237928e-05, + "loss": 0.17795734405517577, + "step": 1030 + }, + { + "epoch": 0.027395185427868674, + "grad_norm": 3.30890154838562, + "learning_rate": 1.8244073748902547e-05, + "loss": 0.19251251220703125, + "step": 1040 + }, + { + "epoch": 0.02765860067236741, + "grad_norm": 2.9566683769226074, + "learning_rate": 1.8419666374012293e-05, + "loss": 0.18300242424011232, + "step": 1050 + }, + { + "epoch": 0.02792201591686615, + "grad_norm": 3.87607741355896, + "learning_rate": 1.859525899912204e-05, + "loss": 0.19716136455535888, + "step": 1060 + }, + { + "epoch": 0.028185431161364887, + "grad_norm": 3.8118443489074707, + "learning_rate": 1.8770851624231782e-05, + "loss": 0.1801258683204651, + "step": 1070 + }, + { + "epoch": 0.028448846405863624, + "grad_norm": 2.6945271492004395, + "learning_rate": 1.8946444249341528e-05, + "loss": 0.20070993900299072, + "step": 1080 + }, + { + "epoch": 0.02871226165036236, + "grad_norm": 6.464297294616699, + "learning_rate": 1.9122036874451274e-05, + "loss": 0.19830548763275146, + "step": 1090 + }, + { + "epoch": 0.028975676894861097, + "grad_norm": 2.0310046672821045, + "learning_rate": 1.929762949956102e-05, + "loss": 0.1799211621284485, + "step": 1100 + }, + { + "epoch": 0.029239092139359834, + "grad_norm": 3.397616386413574, + "learning_rate": 1.9473222124670767e-05, + "loss": 0.19357128143310548, + "step": 1110 + }, + { + "epoch": 0.02950250738385857, + "grad_norm": 5.939323425292969, + "learning_rate": 1.9648814749780513e-05, + "loss": 0.1766756296157837, + "step": 1120 + }, + { + "epoch": 0.02976592262835731, + "grad_norm": 5.818624019622803, + "learning_rate": 1.9824407374890256e-05, + "loss": 0.18121392726898194, + "step": 1130 + }, + { + "epoch": 0.030029337872856048, + "grad_norm": 3.6339993476867676, + "learning_rate": 2e-05, + "loss": 0.18885889053344726, + "step": 1140 + }, + { + "epoch": 0.030292753117354784, + "grad_norm": 9.329225540161133, + "learning_rate": 1.9999996360784282e-05, + "loss": 0.1862187147140503, + "step": 1150 + }, + { + "epoch": 0.03055616836185352, + "grad_norm": 5.06416654586792, + "learning_rate": 1.9999985443139778e-05, + "loss": 0.17984529733657836, + "step": 1160 + }, + { + "epoch": 0.030819583606352258, + "grad_norm": 3.781578302383423, + "learning_rate": 1.999996724707443e-05, + "loss": 0.1888970375061035, + "step": 1170 + }, + { + "epoch": 0.031082998850850994, + "grad_norm": 2.128148317337036, + "learning_rate": 1.9999941772601482e-05, + "loss": 0.1762686252593994, + "step": 1180 + }, + { + "epoch": 0.03134641409534973, + "grad_norm": 3.6858041286468506, + "learning_rate": 1.9999909019739482e-05, + "loss": 0.18129703998565674, + "step": 1190 + }, + { + "epoch": 0.03160982933984847, + "grad_norm": 5.046562671661377, + "learning_rate": 1.9999868988512265e-05, + "loss": 0.17452647686004638, + "step": 1200 + }, + { + "epoch": 0.031873244584347205, + "grad_norm": 12.171921730041504, + "learning_rate": 1.9999821678948964e-05, + "loss": 0.176239275932312, + "step": 1210 + }, + { + "epoch": 0.03213665982884595, + "grad_norm": 4.554821014404297, + "learning_rate": 1.999976709108402e-05, + "loss": 0.18586684465408326, + "step": 1220 + }, + { + "epoch": 0.032400075073344685, + "grad_norm": 5.561591148376465, + "learning_rate": 1.9999705224957156e-05, + "loss": 0.19549739360809326, + "step": 1230 + }, + { + "epoch": 0.03266349031784342, + "grad_norm": 11.971210479736328, + "learning_rate": 1.9999636080613406e-05, + "loss": 0.19237958192825316, + "step": 1240 + }, + { + "epoch": 0.03292690556234216, + "grad_norm": 4.842509746551514, + "learning_rate": 1.9999559658103097e-05, + "loss": 0.1939957857131958, + "step": 1250 + }, + { + "epoch": 0.033190320806840895, + "grad_norm": 5.927211761474609, + "learning_rate": 1.9999475957481848e-05, + "loss": 0.2022526502609253, + "step": 1260 + }, + { + "epoch": 0.03345373605133963, + "grad_norm": 2.038816452026367, + "learning_rate": 1.999938497881059e-05, + "loss": 0.2052863359451294, + "step": 1270 + }, + { + "epoch": 0.03371715129583837, + "grad_norm": 2.925954818725586, + "learning_rate": 1.999928672215553e-05, + "loss": 0.18391671180725097, + "step": 1280 + }, + { + "epoch": 0.033980566540337105, + "grad_norm": 3.7110981941223145, + "learning_rate": 1.9999181187588187e-05, + "loss": 0.19377336502075196, + "step": 1290 + }, + { + "epoch": 0.03424398178483584, + "grad_norm": 3.913609027862549, + "learning_rate": 1.9999068375185372e-05, + "loss": 0.18016549348831176, + "step": 1300 + }, + { + "epoch": 0.03450739702933458, + "grad_norm": 8.4606294631958, + "learning_rate": 1.99989482850292e-05, + "loss": 0.184322988986969, + "step": 1310 + }, + { + "epoch": 0.034770812273833315, + "grad_norm": 2.021946668624878, + "learning_rate": 1.9998820917207076e-05, + "loss": 0.18283509016036986, + "step": 1320 + }, + { + "epoch": 0.03503422751833205, + "grad_norm": 3.1540169715881348, + "learning_rate": 1.99986862718117e-05, + "loss": 0.18447697162628174, + "step": 1330 + }, + { + "epoch": 0.03529764276283079, + "grad_norm": 4.01518440246582, + "learning_rate": 1.999854434894108e-05, + "loss": 0.187074875831604, + "step": 1340 + }, + { + "epoch": 0.03556105800732953, + "grad_norm": 5.297163009643555, + "learning_rate": 1.9998395148698505e-05, + "loss": 0.18541643619537354, + "step": 1350 + }, + { + "epoch": 0.03582447325182827, + "grad_norm": 10.064289093017578, + "learning_rate": 1.9998238671192575e-05, + "loss": 0.1915461778640747, + "step": 1360 + }, + { + "epoch": 0.036087888496327006, + "grad_norm": 5.913508892059326, + "learning_rate": 1.999807491653718e-05, + "loss": 0.19750117063522338, + "step": 1370 + }, + { + "epoch": 0.03635130374082574, + "grad_norm": 4.206572532653809, + "learning_rate": 1.9997903884851503e-05, + "loss": 0.1824948310852051, + "step": 1380 + }, + { + "epoch": 0.03661471898532448, + "grad_norm": 3.5462112426757812, + "learning_rate": 1.999772557626004e-05, + "loss": 0.17996599674224853, + "step": 1390 + }, + { + "epoch": 0.036878134229823216, + "grad_norm": 3.841313123703003, + "learning_rate": 1.9997539990892555e-05, + "loss": 0.1717366099357605, + "step": 1400 + }, + { + "epoch": 0.03714154947432195, + "grad_norm": 3.5583274364471436, + "learning_rate": 1.9997347128884138e-05, + "loss": 0.18444688320159913, + "step": 1410 + }, + { + "epoch": 0.03740496471882069, + "grad_norm": 2.3911099433898926, + "learning_rate": 1.9997146990375162e-05, + "loss": 0.18556758165359497, + "step": 1420 + }, + { + "epoch": 0.037668379963319426, + "grad_norm": 4.502468109130859, + "learning_rate": 1.999693957551129e-05, + "loss": 0.17516961097717285, + "step": 1430 + }, + { + "epoch": 0.03793179520781816, + "grad_norm": 6.034434795379639, + "learning_rate": 1.999672488444349e-05, + "loss": 0.17402315139770508, + "step": 1440 + }, + { + "epoch": 0.0381952104523169, + "grad_norm": 4.432474136352539, + "learning_rate": 1.999650291732802e-05, + "loss": 0.19600434303283693, + "step": 1450 + }, + { + "epoch": 0.038458625696815636, + "grad_norm": 1.9279738664627075, + "learning_rate": 1.9996273674326446e-05, + "loss": 0.18723634481430054, + "step": 1460 + }, + { + "epoch": 0.03872204094131437, + "grad_norm": 4.016931533813477, + "learning_rate": 1.9996037155605615e-05, + "loss": 0.17172617912292482, + "step": 1470 + }, + { + "epoch": 0.03898545618581312, + "grad_norm": 3.3585681915283203, + "learning_rate": 1.9995793361337676e-05, + "loss": 0.18191821575164796, + "step": 1480 + }, + { + "epoch": 0.03924887143031185, + "grad_norm": 4.338852882385254, + "learning_rate": 1.9995542291700075e-05, + "loss": 0.18500828742980957, + "step": 1490 + }, + { + "epoch": 0.03951228667481059, + "grad_norm": 4.091066360473633, + "learning_rate": 1.9995283946875548e-05, + "loss": 0.1791035294532776, + "step": 1500 + }, + { + "epoch": 0.03977570191930933, + "grad_norm": 6.628575801849365, + "learning_rate": 1.999501832705213e-05, + "loss": 0.16450824737548828, + "step": 1510 + }, + { + "epoch": 0.04003911716380806, + "grad_norm": 2.334413766860962, + "learning_rate": 1.9994745432423154e-05, + "loss": 0.1717955470085144, + "step": 1520 + }, + { + "epoch": 0.0403025324083068, + "grad_norm": 3.7677855491638184, + "learning_rate": 1.9994465263187242e-05, + "loss": 0.16275398731231688, + "step": 1530 + }, + { + "epoch": 0.04056594765280554, + "grad_norm": 4.688307762145996, + "learning_rate": 1.9994177819548313e-05, + "loss": 0.2020111560821533, + "step": 1540 + }, + { + "epoch": 0.040829362897304274, + "grad_norm": 3.4726829528808594, + "learning_rate": 1.999388310171558e-05, + "loss": 0.18940417766571044, + "step": 1550 + }, + { + "epoch": 0.04109277814180301, + "grad_norm": 4.971543312072754, + "learning_rate": 1.9993581109903557e-05, + "loss": 0.16627306938171388, + "step": 1560 + }, + { + "epoch": 0.04135619338630175, + "grad_norm": 8.080140113830566, + "learning_rate": 1.999327184433204e-05, + "loss": 0.1768871545791626, + "step": 1570 + }, + { + "epoch": 0.041619608630800484, + "grad_norm": 3.340677261352539, + "learning_rate": 1.9992955305226127e-05, + "loss": 0.1750815510749817, + "step": 1580 + }, + { + "epoch": 0.04188302387529922, + "grad_norm": 2.0620100498199463, + "learning_rate": 1.9992631492816214e-05, + "loss": 0.17990264892578126, + "step": 1590 + }, + { + "epoch": 0.04214643911979796, + "grad_norm": 3.0989537239074707, + "learning_rate": 1.9992300407337977e-05, + "loss": 0.19354822635650634, + "step": 1600 + }, + { + "epoch": 0.0424098543642967, + "grad_norm": 9.38802433013916, + "learning_rate": 1.9991962049032403e-05, + "loss": 0.17305214405059816, + "step": 1610 + }, + { + "epoch": 0.04267326960879544, + "grad_norm": 6.4716668128967285, + "learning_rate": 1.9991616418145757e-05, + "loss": 0.18459359407424927, + "step": 1620 + }, + { + "epoch": 0.042936684853294174, + "grad_norm": 3.147315263748169, + "learning_rate": 1.9991263514929606e-05, + "loss": 0.17241122722625732, + "step": 1630 + }, + { + "epoch": 0.04320010009779291, + "grad_norm": 7.0495734214782715, + "learning_rate": 1.999090333964081e-05, + "loss": 0.18629064559936523, + "step": 1640 + }, + { + "epoch": 0.04346351534229165, + "grad_norm": 2.039573907852173, + "learning_rate": 1.9990535892541524e-05, + "loss": 0.16833927631378173, + "step": 1650 + }, + { + "epoch": 0.043726930586790384, + "grad_norm": 4.071380138397217, + "learning_rate": 1.999016117389918e-05, + "loss": 0.1974048614501953, + "step": 1660 + }, + { + "epoch": 0.04399034583128912, + "grad_norm": 5.368656158447266, + "learning_rate": 1.998977918398653e-05, + "loss": 0.1949012279510498, + "step": 1670 + }, + { + "epoch": 0.04425376107578786, + "grad_norm": 2.2139081954956055, + "learning_rate": 1.9989389923081587e-05, + "loss": 0.17000781297683715, + "step": 1680 + }, + { + "epoch": 0.044517176320286594, + "grad_norm": 3.2063956260681152, + "learning_rate": 1.998899339146768e-05, + "loss": 0.16955630779266356, + "step": 1690 + }, + { + "epoch": 0.04478059156478533, + "grad_norm": 1.7960844039916992, + "learning_rate": 1.9988589589433423e-05, + "loss": 0.1880159854888916, + "step": 1700 + }, + { + "epoch": 0.04504400680928407, + "grad_norm": 5.123932361602783, + "learning_rate": 1.9988178517272717e-05, + "loss": 0.18648149967193603, + "step": 1710 + }, + { + "epoch": 0.045307422053782805, + "grad_norm": 3.0537796020507812, + "learning_rate": 1.998776017528476e-05, + "loss": 0.18290233612060547, + "step": 1720 + }, + { + "epoch": 0.04557083729828155, + "grad_norm": 3.9228076934814453, + "learning_rate": 1.9987334563774038e-05, + "loss": 0.19368939399719237, + "step": 1730 + }, + { + "epoch": 0.045834252542780285, + "grad_norm": 5.3459367752075195, + "learning_rate": 1.998690168305033e-05, + "loss": 0.1755902051925659, + "step": 1740 + }, + { + "epoch": 0.04609766778727902, + "grad_norm": 1.7226766347885132, + "learning_rate": 1.9986461533428708e-05, + "loss": 0.18514509201049806, + "step": 1750 + }, + { + "epoch": 0.04636108303177776, + "grad_norm": 3.2564897537231445, + "learning_rate": 1.9986014115229525e-05, + "loss": 0.18136851787567138, + "step": 1760 + }, + { + "epoch": 0.046624498276276495, + "grad_norm": 1.9003889560699463, + "learning_rate": 1.998555942877844e-05, + "loss": 0.18272826671600342, + "step": 1770 + }, + { + "epoch": 0.04688791352077523, + "grad_norm": 2.818310022354126, + "learning_rate": 1.9985097474406387e-05, + "loss": 0.18274883031845093, + "step": 1780 + }, + { + "epoch": 0.04715132876527397, + "grad_norm": 3.761615753173828, + "learning_rate": 1.99846282524496e-05, + "loss": 0.1706752061843872, + "step": 1790 + }, + { + "epoch": 0.047414744009772705, + "grad_norm": 4.160594940185547, + "learning_rate": 1.9984151763249598e-05, + "loss": 0.18383340835571288, + "step": 1800 + }, + { + "epoch": 0.04767815925427144, + "grad_norm": 2.4480063915252686, + "learning_rate": 1.9983668007153185e-05, + "loss": 0.18391181230545045, + "step": 1810 + }, + { + "epoch": 0.04794157449877018, + "grad_norm": 2.8958194255828857, + "learning_rate": 1.998317698451247e-05, + "loss": 0.17987383604049684, + "step": 1820 + }, + { + "epoch": 0.048204989743268915, + "grad_norm": 2.7527029514312744, + "learning_rate": 1.9982678695684833e-05, + "loss": 0.17583736181259155, + "step": 1830 + }, + { + "epoch": 0.04846840498776765, + "grad_norm": 2.10933256149292, + "learning_rate": 1.998217314103295e-05, + "loss": 0.18025974035263062, + "step": 1840 + }, + { + "epoch": 0.04873182023226639, + "grad_norm": 2.4016599655151367, + "learning_rate": 1.998166032092479e-05, + "loss": 0.1809309244155884, + "step": 1850 + }, + { + "epoch": 0.04899523547676513, + "grad_norm": 2.077577829360962, + "learning_rate": 1.99811402357336e-05, + "loss": 0.16272881031036376, + "step": 1860 + }, + { + "epoch": 0.04925865072126387, + "grad_norm": 3.2457082271575928, + "learning_rate": 1.9980612885837925e-05, + "loss": 0.1844792604446411, + "step": 1870 + }, + { + "epoch": 0.049522065965762606, + "grad_norm": 2.239504098892212, + "learning_rate": 1.998007827162159e-05, + "loss": 0.17323575019836426, + "step": 1880 + }, + { + "epoch": 0.04978548121026134, + "grad_norm": 5.093751430511475, + "learning_rate": 1.9979536393473715e-05, + "loss": 0.18002724647521973, + "step": 1890 + }, + { + "epoch": 0.05004889645476008, + "grad_norm": 4.157344818115234, + "learning_rate": 1.9978987251788696e-05, + "loss": 0.17924578189849855, + "step": 1900 + }, + { + "epoch": 0.050312311699258816, + "grad_norm": 4.624197483062744, + "learning_rate": 1.997843084696623e-05, + "loss": 0.18008124828338623, + "step": 1910 + }, + { + "epoch": 0.05057572694375755, + "grad_norm": 6.328530788421631, + "learning_rate": 1.9977867179411282e-05, + "loss": 0.15612313747406006, + "step": 1920 + }, + { + "epoch": 0.05083914218825629, + "grad_norm": 3.762817621231079, + "learning_rate": 1.997729624953412e-05, + "loss": 0.1849362850189209, + "step": 1930 + }, + { + "epoch": 0.051102557432755026, + "grad_norm": 2.637850522994995, + "learning_rate": 1.997671805775029e-05, + "loss": 0.1814112901687622, + "step": 1940 + }, + { + "epoch": 0.05136597267725376, + "grad_norm": 4.267578125, + "learning_rate": 1.9976132604480625e-05, + "loss": 0.18101271390914916, + "step": 1950 + }, + { + "epoch": 0.0516293879217525, + "grad_norm": 4.239940166473389, + "learning_rate": 1.997553989015125e-05, + "loss": 0.19434611797332763, + "step": 1960 + }, + { + "epoch": 0.051892803166251236, + "grad_norm": 3.3949615955352783, + "learning_rate": 1.9974939915193557e-05, + "loss": 0.1743127942085266, + "step": 1970 + }, + { + "epoch": 0.05215621841074997, + "grad_norm": 2.3918778896331787, + "learning_rate": 1.9974332680044237e-05, + "loss": 0.18094581365585327, + "step": 1980 + }, + { + "epoch": 0.05241963365524872, + "grad_norm": 2.401001214981079, + "learning_rate": 1.9973718185145262e-05, + "loss": 0.17097023725509644, + "step": 1990 + }, + { + "epoch": 0.05268304889974745, + "grad_norm": 5.569243431091309, + "learning_rate": 1.997309643094389e-05, + "loss": 0.1768436074256897, + "step": 2000 + }, + { + "epoch": 0.05294646414424619, + "grad_norm": 1.2817565202713013, + "learning_rate": 1.997246741789266e-05, + "loss": 0.1727694272994995, + "step": 2010 + }, + { + "epoch": 0.05320987938874493, + "grad_norm": 5.285900592803955, + "learning_rate": 1.99718311464494e-05, + "loss": 0.18553755283355713, + "step": 2020 + }, + { + "epoch": 0.05347329463324366, + "grad_norm": 4.141730785369873, + "learning_rate": 1.9971187617077204e-05, + "loss": 0.17676810026168824, + "step": 2030 + }, + { + "epoch": 0.0537367098777424, + "grad_norm": 3.247800350189209, + "learning_rate": 1.9970536830244466e-05, + "loss": 0.1788102865219116, + "step": 2040 + }, + { + "epoch": 0.05400012512224114, + "grad_norm": 2.415637493133545, + "learning_rate": 1.9969878786424863e-05, + "loss": 0.17100372314453124, + "step": 2050 + }, + { + "epoch": 0.054263540366739874, + "grad_norm": 2.5893208980560303, + "learning_rate": 1.9969213486097335e-05, + "loss": 0.18333752155303956, + "step": 2060 + }, + { + "epoch": 0.05452695561123861, + "grad_norm": 5.285122394561768, + "learning_rate": 1.996854092974613e-05, + "loss": 0.18512225151062012, + "step": 2070 + }, + { + "epoch": 0.05479037085573735, + "grad_norm": 3.9135308265686035, + "learning_rate": 1.9967861117860754e-05, + "loss": 0.17024166584014894, + "step": 2080 + }, + { + "epoch": 0.055053786100236084, + "grad_norm": 2.5465779304504395, + "learning_rate": 1.9967174050936008e-05, + "loss": 0.18028564453125, + "step": 2090 + }, + { + "epoch": 0.05531720134473482, + "grad_norm": 2.3952853679656982, + "learning_rate": 1.9966479729471966e-05, + "loss": 0.18981683254241943, + "step": 2100 + }, + { + "epoch": 0.05558061658923356, + "grad_norm": 4.664034366607666, + "learning_rate": 1.9965778153973986e-05, + "loss": 0.16761740446090698, + "step": 2110 + }, + { + "epoch": 0.0558440318337323, + "grad_norm": 2.906691074371338, + "learning_rate": 1.9965069324952703e-05, + "loss": 0.17800801992416382, + "step": 2120 + }, + { + "epoch": 0.05610744707823104, + "grad_norm": 6.599113941192627, + "learning_rate": 1.9964353242924038e-05, + "loss": 0.18182945251464844, + "step": 2130 + }, + { + "epoch": 0.056370862322729774, + "grad_norm": 2.465280055999756, + "learning_rate": 1.9963629908409184e-05, + "loss": 0.17824724912643433, + "step": 2140 + }, + { + "epoch": 0.05663427756722851, + "grad_norm": 2.4029107093811035, + "learning_rate": 1.9962899321934616e-05, + "loss": 0.16545947790145873, + "step": 2150 + }, + { + "epoch": 0.05689769281172725, + "grad_norm": 2.548790693283081, + "learning_rate": 1.9962161484032085e-05, + "loss": 0.1624548316001892, + "step": 2160 + }, + { + "epoch": 0.057161108056225984, + "grad_norm": 3.268845558166504, + "learning_rate": 1.9961416395238623e-05, + "loss": 0.15270982980728148, + "step": 2170 + }, + { + "epoch": 0.05742452330072472, + "grad_norm": 3.8069839477539062, + "learning_rate": 1.9960664056096534e-05, + "loss": 0.19497629404067993, + "step": 2180 + }, + { + "epoch": 0.05768793854522346, + "grad_norm": 7.550158500671387, + "learning_rate": 1.9959904467153408e-05, + "loss": 0.19241867065429688, + "step": 2190 + }, + { + "epoch": 0.057951353789722194, + "grad_norm": 2.32344651222229, + "learning_rate": 1.99591376289621e-05, + "loss": 0.16824026107788087, + "step": 2200 + }, + { + "epoch": 0.05821476903422093, + "grad_norm": 3.568920135498047, + "learning_rate": 1.9958363542080758e-05, + "loss": 0.16973953247070311, + "step": 2210 + }, + { + "epoch": 0.05847818427871967, + "grad_norm": 2.7857272624969482, + "learning_rate": 1.9957582207072784e-05, + "loss": 0.1622350335121155, + "step": 2220 + }, + { + "epoch": 0.058741599523218405, + "grad_norm": 4.109652519226074, + "learning_rate": 1.9956793624506875e-05, + "loss": 0.17969547510147094, + "step": 2230 + }, + { + "epoch": 0.05900501476771714, + "grad_norm": 5.810640811920166, + "learning_rate": 1.9955997794956994e-05, + "loss": 0.16846954822540283, + "step": 2240 + }, + { + "epoch": 0.059268430012215885, + "grad_norm": 4.013246059417725, + "learning_rate": 1.995519471900238e-05, + "loss": 0.17391908168792725, + "step": 2250 + }, + { + "epoch": 0.05953184525671462, + "grad_norm": 1.8151252269744873, + "learning_rate": 1.9954384397227545e-05, + "loss": 0.1717991828918457, + "step": 2260 + }, + { + "epoch": 0.05979526050121336, + "grad_norm": 5.252824306488037, + "learning_rate": 1.9953566830222276e-05, + "loss": 0.17196133136749267, + "step": 2270 + }, + { + "epoch": 0.060058675745712095, + "grad_norm": 3.414438009262085, + "learning_rate": 1.9952742018581634e-05, + "loss": 0.1794467091560364, + "step": 2280 + }, + { + "epoch": 0.06032209099021083, + "grad_norm": 3.8314356803894043, + "learning_rate": 1.9951909962905956e-05, + "loss": 0.17155954837799073, + "step": 2290 + }, + { + "epoch": 0.06058550623470957, + "grad_norm": 4.092149257659912, + "learning_rate": 1.9951070663800843e-05, + "loss": 0.16831212043762206, + "step": 2300 + }, + { + "epoch": 0.060848921479208305, + "grad_norm": 1.8720239400863647, + "learning_rate": 1.995022412187718e-05, + "loss": 0.16441915035247803, + "step": 2310 + }, + { + "epoch": 0.06111233672370704, + "grad_norm": 2.944154977798462, + "learning_rate": 1.9949370337751105e-05, + "loss": 0.17715238332748412, + "step": 2320 + }, + { + "epoch": 0.06137575196820578, + "grad_norm": 3.2072885036468506, + "learning_rate": 1.994850931204405e-05, + "loss": 0.17858015298843383, + "step": 2330 + }, + { + "epoch": 0.061639167212704515, + "grad_norm": 7.501430988311768, + "learning_rate": 1.99476410453827e-05, + "loss": 0.18364896774291992, + "step": 2340 + }, + { + "epoch": 0.06190258245720325, + "grad_norm": 5.096051216125488, + "learning_rate": 1.994676553839902e-05, + "loss": 0.18501732349395753, + "step": 2350 + }, + { + "epoch": 0.06216599770170199, + "grad_norm": 3.6222963333129883, + "learning_rate": 1.9945882791730246e-05, + "loss": 0.18310348987579345, + "step": 2360 + }, + { + "epoch": 0.06242941294620073, + "grad_norm": 5.054588794708252, + "learning_rate": 1.994499280601887e-05, + "loss": 0.16762070655822753, + "step": 2370 + }, + { + "epoch": 0.06269282819069946, + "grad_norm": 4.246995449066162, + "learning_rate": 1.994409558191267e-05, + "loss": 0.16547105312347413, + "step": 2380 + }, + { + "epoch": 0.0629562434351982, + "grad_norm": 6.085341453552246, + "learning_rate": 1.9943191120064672e-05, + "loss": 0.18195632696151734, + "step": 2390 + }, + { + "epoch": 0.06321965867969694, + "grad_norm": 2.5149712562561035, + "learning_rate": 1.99422794211332e-05, + "loss": 0.16690210103988648, + "step": 2400 + }, + { + "epoch": 0.06348307392419568, + "grad_norm": 1.8314058780670166, + "learning_rate": 1.9941360485781814e-05, + "loss": 0.17921217679977416, + "step": 2410 + }, + { + "epoch": 0.06374648916869441, + "grad_norm": 3.0319581031799316, + "learning_rate": 1.9940434314679363e-05, + "loss": 0.17374268770217896, + "step": 2420 + }, + { + "epoch": 0.06400990441319315, + "grad_norm": 2.614964723587036, + "learning_rate": 1.9939500908499947e-05, + "loss": 0.16636699438095093, + "step": 2430 + }, + { + "epoch": 0.0642733196576919, + "grad_norm": 8.1782865524292, + "learning_rate": 1.9938560267922948e-05, + "loss": 0.1790091276168823, + "step": 2440 + }, + { + "epoch": 0.06453673490219063, + "grad_norm": 2.309563636779785, + "learning_rate": 1.9937612393632995e-05, + "loss": 0.15551881790161132, + "step": 2450 + }, + { + "epoch": 0.06480015014668937, + "grad_norm": 2.14058780670166, + "learning_rate": 1.9936657286320003e-05, + "loss": 0.16118932962417604, + "step": 2460 + }, + { + "epoch": 0.0650635653911881, + "grad_norm": 2.0489773750305176, + "learning_rate": 1.993569494667913e-05, + "loss": 0.17816853523254395, + "step": 2470 + }, + { + "epoch": 0.06532698063568684, + "grad_norm": 3.904982805252075, + "learning_rate": 1.9934725375410812e-05, + "loss": 0.18489534854888917, + "step": 2480 + }, + { + "epoch": 0.06559039588018557, + "grad_norm": 2.64485502243042, + "learning_rate": 1.9933748573220745e-05, + "loss": 0.18557517528533934, + "step": 2490 + }, + { + "epoch": 0.06585381112468432, + "grad_norm": 5.070641994476318, + "learning_rate": 1.993276454081989e-05, + "loss": 0.17167556285858154, + "step": 2500 + }, + { + "epoch": 0.06611722636918305, + "grad_norm": 3.228343963623047, + "learning_rate": 1.9931773278924467e-05, + "loss": 0.18222241401672362, + "step": 2510 + }, + { + "epoch": 0.06638064161368179, + "grad_norm": 3.964406728744507, + "learning_rate": 1.9930774788255957e-05, + "loss": 0.18239933252334595, + "step": 2520 + }, + { + "epoch": 0.06664405685818052, + "grad_norm": 2.248955488204956, + "learning_rate": 1.992976906954111e-05, + "loss": 0.1838900327682495, + "step": 2530 + }, + { + "epoch": 0.06690747210267926, + "grad_norm": 2.444645881652832, + "learning_rate": 1.992875612351192e-05, + "loss": 0.17198894023895264, + "step": 2540 + }, + { + "epoch": 0.067170887347178, + "grad_norm": 1.0445919036865234, + "learning_rate": 1.9927735950905663e-05, + "loss": 0.16035268306732178, + "step": 2550 + }, + { + "epoch": 0.06743430259167674, + "grad_norm": 1.4774821996688843, + "learning_rate": 1.9926708552464865e-05, + "loss": 0.19084625244140624, + "step": 2560 + }, + { + "epoch": 0.06769771783617548, + "grad_norm": 6.330926418304443, + "learning_rate": 1.9925673928937307e-05, + "loss": 0.18249220848083497, + "step": 2570 + }, + { + "epoch": 0.06796113308067421, + "grad_norm": 3.022345542907715, + "learning_rate": 1.9924632081076032e-05, + "loss": 0.17329739332199096, + "step": 2580 + }, + { + "epoch": 0.06822454832517295, + "grad_norm": 2.528435468673706, + "learning_rate": 1.9923583009639344e-05, + "loss": 0.1920970559120178, + "step": 2590 + }, + { + "epoch": 0.06848796356967168, + "grad_norm": 4.425650119781494, + "learning_rate": 1.9922526715390802e-05, + "loss": 0.1737358331680298, + "step": 2600 + }, + { + "epoch": 0.06875137881417043, + "grad_norm": 4.2841010093688965, + "learning_rate": 1.9921463199099222e-05, + "loss": 0.1759103059768677, + "step": 2610 + }, + { + "epoch": 0.06901479405866916, + "grad_norm": 2.7117884159088135, + "learning_rate": 1.992039246153868e-05, + "loss": 0.18489656448364258, + "step": 2620 + }, + { + "epoch": 0.0692782093031679, + "grad_norm": 4.911075592041016, + "learning_rate": 1.99193145034885e-05, + "loss": 0.16783981323242186, + "step": 2630 + }, + { + "epoch": 0.06954162454766663, + "grad_norm": 1.9133150577545166, + "learning_rate": 1.9918229325733267e-05, + "loss": 0.17042827606201172, + "step": 2640 + }, + { + "epoch": 0.06980503979216537, + "grad_norm": 1.6695719957351685, + "learning_rate": 1.9917136929062824e-05, + "loss": 0.17181049585342406, + "step": 2650 + }, + { + "epoch": 0.0700684550366641, + "grad_norm": 4.871783256530762, + "learning_rate": 1.991603731427226e-05, + "loss": 0.17624785900115966, + "step": 2660 + }, + { + "epoch": 0.07033187028116285, + "grad_norm": 6.5822954177856445, + "learning_rate": 1.9914930482161928e-05, + "loss": 0.16645238399505616, + "step": 2670 + }, + { + "epoch": 0.07059528552566158, + "grad_norm": 2.7118725776672363, + "learning_rate": 1.9913816433537425e-05, + "loss": 0.15975658893585204, + "step": 2680 + }, + { + "epoch": 0.07085870077016032, + "grad_norm": 2.9324259757995605, + "learning_rate": 1.99126951692096e-05, + "loss": 0.17494509220123292, + "step": 2690 + }, + { + "epoch": 0.07112211601465906, + "grad_norm": 3.2424631118774414, + "learning_rate": 1.991156668999456e-05, + "loss": 0.1830177903175354, + "step": 2700 + }, + { + "epoch": 0.0713855312591578, + "grad_norm": 2.379058599472046, + "learning_rate": 1.9910430996713668e-05, + "loss": 0.1679988384246826, + "step": 2710 + }, + { + "epoch": 0.07164894650365654, + "grad_norm": 3.8943114280700684, + "learning_rate": 1.9909288090193518e-05, + "loss": 0.18313868045806886, + "step": 2720 + }, + { + "epoch": 0.07191236174815527, + "grad_norm": 1.817726731300354, + "learning_rate": 1.9908137971265977e-05, + "loss": 0.19061301946640014, + "step": 2730 + }, + { + "epoch": 0.07217577699265401, + "grad_norm": 3.5006699562072754, + "learning_rate": 1.9906980640768146e-05, + "loss": 0.1798560619354248, + "step": 2740 + }, + { + "epoch": 0.07243919223715274, + "grad_norm": 5.4646196365356445, + "learning_rate": 1.9905816099542382e-05, + "loss": 0.1740372896194458, + "step": 2750 + }, + { + "epoch": 0.07270260748165149, + "grad_norm": 3.6449296474456787, + "learning_rate": 1.9904644348436282e-05, + "loss": 0.1744105577468872, + "step": 2760 + }, + { + "epoch": 0.07296602272615021, + "grad_norm": 2.6258902549743652, + "learning_rate": 1.990346538830271e-05, + "loss": 0.17291841506958008, + "step": 2770 + }, + { + "epoch": 0.07322943797064896, + "grad_norm": 3.0395848751068115, + "learning_rate": 1.9902279219999752e-05, + "loss": 0.1657947778701782, + "step": 2780 + }, + { + "epoch": 0.07349285321514769, + "grad_norm": 5.427474498748779, + "learning_rate": 1.9901085844390757e-05, + "loss": 0.17203714847564697, + "step": 2790 + }, + { + "epoch": 0.07375626845964643, + "grad_norm": 7.625851154327393, + "learning_rate": 1.9899885262344316e-05, + "loss": 0.1807958722114563, + "step": 2800 + }, + { + "epoch": 0.07401968370414516, + "grad_norm": 3.7625389099121094, + "learning_rate": 1.9898677474734266e-05, + "loss": 0.17288472652435302, + "step": 2810 + }, + { + "epoch": 0.0742830989486439, + "grad_norm": 2.4761409759521484, + "learning_rate": 1.9897462482439684e-05, + "loss": 0.1710650682449341, + "step": 2820 + }, + { + "epoch": 0.07454651419314265, + "grad_norm": 4.0331549644470215, + "learning_rate": 1.9896240286344892e-05, + "loss": 0.18755892515182496, + "step": 2830 + }, + { + "epoch": 0.07480992943764138, + "grad_norm": 3.9232523441314697, + "learning_rate": 1.989501088733946e-05, + "loss": 0.17383863925933837, + "step": 2840 + }, + { + "epoch": 0.07507334468214012, + "grad_norm": 2.079674243927002, + "learning_rate": 1.9893774286318198e-05, + "loss": 0.17691659927368164, + "step": 2850 + }, + { + "epoch": 0.07533675992663885, + "grad_norm": 8.49146556854248, + "learning_rate": 1.9892530484181155e-05, + "loss": 0.18154418468475342, + "step": 2860 + }, + { + "epoch": 0.0756001751711376, + "grad_norm": 4.2099609375, + "learning_rate": 1.989127948183363e-05, + "loss": 0.17273948192596436, + "step": 2870 + }, + { + "epoch": 0.07586359041563633, + "grad_norm": 4.139365196228027, + "learning_rate": 1.9890021280186148e-05, + "loss": 0.18434247970581055, + "step": 2880 + }, + { + "epoch": 0.07612700566013507, + "grad_norm": 5.7023749351501465, + "learning_rate": 1.9888755880154485e-05, + "loss": 0.1615874171257019, + "step": 2890 + }, + { + "epoch": 0.0763904209046338, + "grad_norm": 4.332421779632568, + "learning_rate": 1.988748328265966e-05, + "loss": 0.15035327672958373, + "step": 2900 + }, + { + "epoch": 0.07665383614913254, + "grad_norm": 1.7322015762329102, + "learning_rate": 1.9886203488627918e-05, + "loss": 0.18242958784103394, + "step": 2910 + }, + { + "epoch": 0.07691725139363127, + "grad_norm": 4.378740310668945, + "learning_rate": 1.988491649899075e-05, + "loss": 0.1772255778312683, + "step": 2920 + }, + { + "epoch": 0.07718066663813002, + "grad_norm": 5.687978744506836, + "learning_rate": 1.988362231468488e-05, + "loss": 0.1645892858505249, + "step": 2930 + }, + { + "epoch": 0.07744408188262875, + "grad_norm": 1.2320036888122559, + "learning_rate": 1.9882320936652278e-05, + "loss": 0.18189287185668945, + "step": 2940 + }, + { + "epoch": 0.07770749712712749, + "grad_norm": 3.2876033782958984, + "learning_rate": 1.9881012365840138e-05, + "loss": 0.16775233745574952, + "step": 2950 + }, + { + "epoch": 0.07797091237162623, + "grad_norm": 3.265852689743042, + "learning_rate": 1.987969660320089e-05, + "loss": 0.17740464210510254, + "step": 2960 + }, + { + "epoch": 0.07823432761612496, + "grad_norm": 1.0840257406234741, + "learning_rate": 1.9878373649692213e-05, + "loss": 0.1837003469467163, + "step": 2970 + }, + { + "epoch": 0.0784977428606237, + "grad_norm": 1.6622782945632935, + "learning_rate": 1.9877043506277004e-05, + "loss": 0.1839388132095337, + "step": 2980 + }, + { + "epoch": 0.07876115810512244, + "grad_norm": 3.7889814376831055, + "learning_rate": 1.9875706173923396e-05, + "loss": 0.17191548347473146, + "step": 2990 + }, + { + "epoch": 0.07902457334962118, + "grad_norm": 1.8319087028503418, + "learning_rate": 1.9874361653604765e-05, + "loss": 0.16178125143051147, + "step": 3000 + }, + { + "epoch": 0.07928798859411991, + "grad_norm": 3.1240715980529785, + "learning_rate": 1.9873009946299702e-05, + "loss": 0.18039338588714598, + "step": 3010 + }, + { + "epoch": 0.07955140383861865, + "grad_norm": 3.4955434799194336, + "learning_rate": 1.9871651052992046e-05, + "loss": 0.16119122505187988, + "step": 3020 + }, + { + "epoch": 0.07981481908311738, + "grad_norm": 15.529653549194336, + "learning_rate": 1.987028497467085e-05, + "loss": 0.15447007417678832, + "step": 3030 + }, + { + "epoch": 0.08007823432761613, + "grad_norm": 21.28272247314453, + "learning_rate": 1.986891171233041e-05, + "loss": 2.8393976211547853, + "step": 3040 + }, + { + "epoch": 0.08034164957211486, + "grad_norm": 6.518364906311035, + "learning_rate": 1.9867531266970247e-05, + "loss": 0.269659161567688, + "step": 3050 + }, + { + "epoch": 0.0806050648166136, + "grad_norm": 3.319887638092041, + "learning_rate": 1.9866143639595107e-05, + "loss": 0.24596257209777833, + "step": 3060 + }, + { + "epoch": 0.08086848006111233, + "grad_norm": 4.384130477905273, + "learning_rate": 1.9864748831214964e-05, + "loss": 0.23252687454223633, + "step": 3070 + }, + { + "epoch": 0.08113189530561107, + "grad_norm": 4.095651626586914, + "learning_rate": 1.986334684284502e-05, + "loss": 0.24272022247314454, + "step": 3080 + }, + { + "epoch": 0.08139531055010982, + "grad_norm": 7.813477516174316, + "learning_rate": 1.98619376755057e-05, + "loss": 0.23909568786621094, + "step": 3090 + }, + { + "epoch": 0.08165872579460855, + "grad_norm": 3.0808095932006836, + "learning_rate": 1.986052133022266e-05, + "loss": 0.24618134498596192, + "step": 3100 + }, + { + "epoch": 0.08192214103910729, + "grad_norm": 5.515805244445801, + "learning_rate": 1.985909780802678e-05, + "loss": 0.21603891849517823, + "step": 3110 + }, + { + "epoch": 0.08218555628360602, + "grad_norm": 1.5073351860046387, + "learning_rate": 1.985766710995416e-05, + "loss": 0.19726396799087526, + "step": 3120 + }, + { + "epoch": 0.08244897152810476, + "grad_norm": 5.752342224121094, + "learning_rate": 1.9856229237046115e-05, + "loss": 0.20675277709960938, + "step": 3130 + }, + { + "epoch": 0.0827123867726035, + "grad_norm": 2.6994307041168213, + "learning_rate": 1.9854784190349203e-05, + "loss": 0.21076593399047852, + "step": 3140 + }, + { + "epoch": 0.08297580201710224, + "grad_norm": 3.161623954772949, + "learning_rate": 1.985333197091518e-05, + "loss": 0.20023956298828124, + "step": 3150 + }, + { + "epoch": 0.08323921726160097, + "grad_norm": 5.005593299865723, + "learning_rate": 1.9851872579801044e-05, + "loss": 0.1948699712753296, + "step": 3160 + }, + { + "epoch": 0.08350263250609971, + "grad_norm": 3.1755945682525635, + "learning_rate": 1.9850406018068997e-05, + "loss": 0.1830970048904419, + "step": 3170 + }, + { + "epoch": 0.08376604775059844, + "grad_norm": 2.322216510772705, + "learning_rate": 1.984893228678647e-05, + "loss": 0.18794100284576415, + "step": 3180 + }, + { + "epoch": 0.08402946299509718, + "grad_norm": 3.9853222370147705, + "learning_rate": 1.9847451387026103e-05, + "loss": 0.1933680295944214, + "step": 3190 + }, + { + "epoch": 0.08429287823959591, + "grad_norm": 3.685981512069702, + "learning_rate": 1.9845963319865758e-05, + "loss": 0.20287303924560546, + "step": 3200 + }, + { + "epoch": 0.08455629348409466, + "grad_norm": 1.3483988046646118, + "learning_rate": 1.9844468086388523e-05, + "loss": 0.184690523147583, + "step": 3210 + }, + { + "epoch": 0.0848197087285934, + "grad_norm": 4.629469394683838, + "learning_rate": 1.9842965687682686e-05, + "loss": 0.18942369222640992, + "step": 3220 + }, + { + "epoch": 0.08508312397309213, + "grad_norm": 2.0157630443573, + "learning_rate": 1.9841456124841763e-05, + "loss": 0.20020670890808107, + "step": 3230 + }, + { + "epoch": 0.08534653921759087, + "grad_norm": 1.8733210563659668, + "learning_rate": 1.983993939896447e-05, + "loss": 0.17808420658111573, + "step": 3240 + }, + { + "epoch": 0.0856099544620896, + "grad_norm": 2.170532703399658, + "learning_rate": 1.9838415511154755e-05, + "loss": 0.2060643196105957, + "step": 3250 + }, + { + "epoch": 0.08587336970658835, + "grad_norm": 2.7236168384552, + "learning_rate": 1.9836884462521768e-05, + "loss": 0.18426477909088135, + "step": 3260 + }, + { + "epoch": 0.08613678495108708, + "grad_norm": 3.115349054336548, + "learning_rate": 1.9835346254179865e-05, + "loss": 0.18889079093933106, + "step": 3270 + }, + { + "epoch": 0.08640020019558582, + "grad_norm": 5.8399271965026855, + "learning_rate": 1.9833800887248628e-05, + "loss": 0.19885026216506957, + "step": 3280 + }, + { + "epoch": 0.08666361544008455, + "grad_norm": 8.020423889160156, + "learning_rate": 1.983224836285284e-05, + "loss": 0.20701828002929687, + "step": 3290 + }, + { + "epoch": 0.0869270306845833, + "grad_norm": 8.319822311401367, + "learning_rate": 1.9830688682122494e-05, + "loss": 0.21213173866271973, + "step": 3300 + }, + { + "epoch": 0.08719044592908202, + "grad_norm": 1.4663560390472412, + "learning_rate": 1.982912184619279e-05, + "loss": 0.19522721767425538, + "step": 3310 + }, + { + "epoch": 0.08745386117358077, + "grad_norm": 8.025879859924316, + "learning_rate": 1.9827547856204147e-05, + "loss": 0.1873168468475342, + "step": 3320 + }, + { + "epoch": 0.0877172764180795, + "grad_norm": 3.7237467765808105, + "learning_rate": 1.9825966713302172e-05, + "loss": 0.1921435832977295, + "step": 3330 + }, + { + "epoch": 0.08798069166257824, + "grad_norm": 3.4259660243988037, + "learning_rate": 1.98243784186377e-05, + "loss": 0.19060527086257933, + "step": 3340 + }, + { + "epoch": 0.08824410690707699, + "grad_norm": 3.2376763820648193, + "learning_rate": 1.982278297336675e-05, + "loss": 0.19539921283721923, + "step": 3350 + }, + { + "epoch": 0.08850752215157572, + "grad_norm": 2.6323204040527344, + "learning_rate": 1.9821180378650563e-05, + "loss": 0.20277647972106932, + "step": 3360 + }, + { + "epoch": 0.08877093739607446, + "grad_norm": 3.8610823154449463, + "learning_rate": 1.9819570635655577e-05, + "loss": 0.19387149810791016, + "step": 3370 + }, + { + "epoch": 0.08903435264057319, + "grad_norm": 3.749545097351074, + "learning_rate": 1.981795374555343e-05, + "loss": 0.18470561504364014, + "step": 3380 + }, + { + "epoch": 0.08929776788507193, + "grad_norm": 3.628415584564209, + "learning_rate": 1.981632970952096e-05, + "loss": 0.1806620955467224, + "step": 3390 + }, + { + "epoch": 0.08956118312957066, + "grad_norm": 1.8304684162139893, + "learning_rate": 1.9814698528740222e-05, + "loss": 0.19420531988143921, + "step": 3400 + }, + { + "epoch": 0.0898245983740694, + "grad_norm": 3.8019587993621826, + "learning_rate": 1.9813060204398448e-05, + "loss": 0.19016759395599364, + "step": 3410 + }, + { + "epoch": 0.09008801361856814, + "grad_norm": 2.6580960750579834, + "learning_rate": 1.9811414737688088e-05, + "loss": 0.1930164575576782, + "step": 3420 + }, + { + "epoch": 0.09035142886306688, + "grad_norm": 3.0833935737609863, + "learning_rate": 1.9809762129806777e-05, + "loss": 0.19644736051559447, + "step": 3430 + }, + { + "epoch": 0.09061484410756561, + "grad_norm": 3.616730213165283, + "learning_rate": 1.9808102381957364e-05, + "loss": 0.19193400144577027, + "step": 3440 + }, + { + "epoch": 0.09087825935206435, + "grad_norm": 2.6576435565948486, + "learning_rate": 1.980643549534788e-05, + "loss": 0.18914273977279664, + "step": 3450 + }, + { + "epoch": 0.0911416745965631, + "grad_norm": 0.8002331256866455, + "learning_rate": 1.980476147119155e-05, + "loss": 0.17592965364456176, + "step": 3460 + }, + { + "epoch": 0.09140508984106183, + "grad_norm": 1.3671228885650635, + "learning_rate": 1.980308031070682e-05, + "loss": 0.1741530179977417, + "step": 3470 + }, + { + "epoch": 0.09166850508556057, + "grad_norm": 2.183088541030884, + "learning_rate": 1.9801392015117286e-05, + "loss": 0.19551424980163573, + "step": 3480 + }, + { + "epoch": 0.0919319203300593, + "grad_norm": 3.970794916152954, + "learning_rate": 1.9799696585651782e-05, + "loss": 0.1722075819969177, + "step": 3490 + }, + { + "epoch": 0.09219533557455804, + "grad_norm": 0.7683253884315491, + "learning_rate": 1.9797994023544307e-05, + "loss": 0.20143322944641112, + "step": 3500 + }, + { + "epoch": 0.09245875081905677, + "grad_norm": 1.922876000404358, + "learning_rate": 1.979628433003406e-05, + "loss": 0.18738627433776855, + "step": 3510 + }, + { + "epoch": 0.09272216606355552, + "grad_norm": 1.2646851539611816, + "learning_rate": 1.9794567506365432e-05, + "loss": 0.18937720060348512, + "step": 3520 + }, + { + "epoch": 0.09298558130805425, + "grad_norm": 1.9259783029556274, + "learning_rate": 1.9792843553788e-05, + "loss": 0.18392776250839232, + "step": 3530 + }, + { + "epoch": 0.09324899655255299, + "grad_norm": 2.326076030731201, + "learning_rate": 1.9791112473556528e-05, + "loss": 0.19404315948486328, + "step": 3540 + }, + { + "epoch": 0.09351241179705172, + "grad_norm": 2.3911843299865723, + "learning_rate": 1.978937426693097e-05, + "loss": 0.18724722862243653, + "step": 3550 + }, + { + "epoch": 0.09377582704155046, + "grad_norm": 2.4326820373535156, + "learning_rate": 1.9787628935176476e-05, + "loss": 0.19222370386123658, + "step": 3560 + }, + { + "epoch": 0.0940392422860492, + "grad_norm": 3.5273001194000244, + "learning_rate": 1.9785876479563367e-05, + "loss": 0.19501863718032836, + "step": 3570 + }, + { + "epoch": 0.09430265753054794, + "grad_norm": 1.9283238649368286, + "learning_rate": 1.978411690136716e-05, + "loss": 0.18611249923706055, + "step": 3580 + }, + { + "epoch": 0.09456607277504668, + "grad_norm": 3.3371431827545166, + "learning_rate": 1.9782350201868547e-05, + "loss": 0.1871795654296875, + "step": 3590 + }, + { + "epoch": 0.09482948801954541, + "grad_norm": 3.065276861190796, + "learning_rate": 1.978057638235341e-05, + "loss": 0.18282206058502198, + "step": 3600 + }, + { + "epoch": 0.09509290326404415, + "grad_norm": 3.8498787879943848, + "learning_rate": 1.9778795444112812e-05, + "loss": 0.19380820989608766, + "step": 3610 + }, + { + "epoch": 0.09535631850854288, + "grad_norm": 7.0834550857543945, + "learning_rate": 1.9777007388443e-05, + "loss": 0.1770959973335266, + "step": 3620 + }, + { + "epoch": 0.09561973375304163, + "grad_norm": 3.4550278186798096, + "learning_rate": 1.9775212216645393e-05, + "loss": 0.17588636875152588, + "step": 3630 + }, + { + "epoch": 0.09588314899754036, + "grad_norm": 1.65609872341156, + "learning_rate": 1.977340993002659e-05, + "loss": 0.19630712270736694, + "step": 3640 + }, + { + "epoch": 0.0961465642420391, + "grad_norm": 3.0210111141204834, + "learning_rate": 1.977160052989839e-05, + "loss": 0.19723811149597167, + "step": 3650 + }, + { + "epoch": 0.09640997948653783, + "grad_norm": 1.8660608530044556, + "learning_rate": 1.9769784017577734e-05, + "loss": 0.1801268458366394, + "step": 3660 + }, + { + "epoch": 0.09667339473103657, + "grad_norm": 8.654324531555176, + "learning_rate": 1.976796039438677e-05, + "loss": 0.19204331636428834, + "step": 3670 + }, + { + "epoch": 0.0969368099755353, + "grad_norm": 7.131392002105713, + "learning_rate": 1.9766129661652803e-05, + "loss": 0.200388765335083, + "step": 3680 + }, + { + "epoch": 0.09720022522003405, + "grad_norm": 2.8265419006347656, + "learning_rate": 1.9764291820708324e-05, + "loss": 0.19815735816955565, + "step": 3690 + }, + { + "epoch": 0.09746364046453278, + "grad_norm": 3.4900243282318115, + "learning_rate": 1.9762446872890984e-05, + "loss": 0.17856037616729736, + "step": 3700 + }, + { + "epoch": 0.09772705570903152, + "grad_norm": 1.7796211242675781, + "learning_rate": 1.976059481954363e-05, + "loss": 0.19718677997589112, + "step": 3710 + }, + { + "epoch": 0.09799047095353026, + "grad_norm": 1.7027003765106201, + "learning_rate": 1.9758735662014253e-05, + "loss": 0.17620015144348145, + "step": 3720 + }, + { + "epoch": 0.098253886198029, + "grad_norm": 2.49672532081604, + "learning_rate": 1.975686940165604e-05, + "loss": 0.1677696466445923, + "step": 3730 + }, + { + "epoch": 0.09851730144252774, + "grad_norm": 1.4921077489852905, + "learning_rate": 1.9754996039827322e-05, + "loss": 0.1704113721847534, + "step": 3740 + }, + { + "epoch": 0.09878071668702647, + "grad_norm": 2.664116621017456, + "learning_rate": 1.9753115577891623e-05, + "loss": 0.1789764404296875, + "step": 3750 + }, + { + "epoch": 0.09904413193152521, + "grad_norm": 1.8090711832046509, + "learning_rate": 1.9751228017217623e-05, + "loss": 0.17567682266235352, + "step": 3760 + }, + { + "epoch": 0.09930754717602394, + "grad_norm": 2.3013875484466553, + "learning_rate": 1.9749333359179166e-05, + "loss": 0.17986319065093995, + "step": 3770 + }, + { + "epoch": 0.09957096242052269, + "grad_norm": 1.5938514471054077, + "learning_rate": 1.974743160515527e-05, + "loss": 0.17964928150177, + "step": 3780 + }, + { + "epoch": 0.09983437766502141, + "grad_norm": 2.513519048690796, + "learning_rate": 1.974552275653011e-05, + "loss": 0.18609919548034667, + "step": 3790 + }, + { + "epoch": 0.10009779290952016, + "grad_norm": 2.6827147006988525, + "learning_rate": 1.974360681469303e-05, + "loss": 0.18490129709243774, + "step": 3800 + }, + { + "epoch": 0.10036120815401889, + "grad_norm": 4.033178806304932, + "learning_rate": 1.9741683781038538e-05, + "loss": 0.17747622728347778, + "step": 3810 + }, + { + "epoch": 0.10062462339851763, + "grad_norm": 2.258007287979126, + "learning_rate": 1.9739753656966294e-05, + "loss": 0.17886345386505126, + "step": 3820 + }, + { + "epoch": 0.10088803864301636, + "grad_norm": 3.19892954826355, + "learning_rate": 1.9737816443881134e-05, + "loss": 0.1952717423439026, + "step": 3830 + }, + { + "epoch": 0.1011514538875151, + "grad_norm": 2.4676148891448975, + "learning_rate": 1.973587214319304e-05, + "loss": 0.16900460720062255, + "step": 3840 + }, + { + "epoch": 0.10141486913201385, + "grad_norm": 1.7283899784088135, + "learning_rate": 1.9733920756317157e-05, + "loss": 0.178424072265625, + "step": 3850 + }, + { + "epoch": 0.10167828437651258, + "grad_norm": 1.6810351610183716, + "learning_rate": 1.973196228467379e-05, + "loss": 0.1836153745651245, + "step": 3860 + }, + { + "epoch": 0.10194169962101132, + "grad_norm": 2.7280619144439697, + "learning_rate": 1.9729996729688403e-05, + "loss": 0.19159855842590331, + "step": 3870 + }, + { + "epoch": 0.10220511486551005, + "grad_norm": 2.2106828689575195, + "learning_rate": 1.9728024092791606e-05, + "loss": 0.18581626415252686, + "step": 3880 + }, + { + "epoch": 0.1024685301100088, + "grad_norm": 1.1353429555892944, + "learning_rate": 1.9726044375419172e-05, + "loss": 0.18125925064086915, + "step": 3890 + }, + { + "epoch": 0.10273194535450753, + "grad_norm": 3.961158514022827, + "learning_rate": 1.9724057579012026e-05, + "loss": 0.18128154277801514, + "step": 3900 + }, + { + "epoch": 0.10299536059900627, + "grad_norm": 3.1606342792510986, + "learning_rate": 1.972206370501624e-05, + "loss": 0.18525553941726686, + "step": 3910 + }, + { + "epoch": 0.103258775843505, + "grad_norm": 2.8953857421875, + "learning_rate": 1.9720062754883048e-05, + "loss": 0.18219292163848877, + "step": 3920 + }, + { + "epoch": 0.10352219108800374, + "grad_norm": 0.9955203533172607, + "learning_rate": 1.971805473006882e-05, + "loss": 0.17435874938964843, + "step": 3930 + }, + { + "epoch": 0.10378560633250247, + "grad_norm": 3.057771682739258, + "learning_rate": 1.9716039632035086e-05, + "loss": 0.17562313079833985, + "step": 3940 + }, + { + "epoch": 0.10404902157700122, + "grad_norm": 2.341850519180298, + "learning_rate": 1.9714017462248523e-05, + "loss": 0.20631422996520996, + "step": 3950 + }, + { + "epoch": 0.10431243682149995, + "grad_norm": 2.598541498184204, + "learning_rate": 1.9711988222180958e-05, + "loss": 0.1972409725189209, + "step": 3960 + }, + { + "epoch": 0.10457585206599869, + "grad_norm": 2.954911708831787, + "learning_rate": 1.9709951913309347e-05, + "loss": 0.18587663173675537, + "step": 3970 + }, + { + "epoch": 0.10483926731049743, + "grad_norm": 2.358367919921875, + "learning_rate": 1.9707908537115816e-05, + "loss": 0.1837489128112793, + "step": 3980 + }, + { + "epoch": 0.10510268255499616, + "grad_norm": 3.806717872619629, + "learning_rate": 1.9705858095087613e-05, + "loss": 0.18874624967575074, + "step": 3990 + }, + { + "epoch": 0.1053660977994949, + "grad_norm": 2.6119537353515625, + "learning_rate": 1.9703800588717144e-05, + "loss": 0.1722928285598755, + "step": 4000 + }, + { + "epoch": 0.10562951304399364, + "grad_norm": 3.14023756980896, + "learning_rate": 1.9701736019501948e-05, + "loss": 0.1944958209991455, + "step": 4010 + }, + { + "epoch": 0.10589292828849238, + "grad_norm": 3.7122998237609863, + "learning_rate": 1.969966438894471e-05, + "loss": 0.17903214693069458, + "step": 4020 + }, + { + "epoch": 0.10615634353299111, + "grad_norm": 2.487043857574463, + "learning_rate": 1.969758569855325e-05, + "loss": 0.1871708631515503, + "step": 4030 + }, + { + "epoch": 0.10641975877748985, + "grad_norm": 1.3900234699249268, + "learning_rate": 1.969549994984053e-05, + "loss": 0.18428175449371337, + "step": 4040 + }, + { + "epoch": 0.10668317402198858, + "grad_norm": 5.028898239135742, + "learning_rate": 1.9693407144324642e-05, + "loss": 0.16265769004821778, + "step": 4050 + }, + { + "epoch": 0.10694658926648733, + "grad_norm": 5.211652755737305, + "learning_rate": 1.9691307283528834e-05, + "loss": 0.1874739408493042, + "step": 4060 + }, + { + "epoch": 0.10721000451098606, + "grad_norm": 3.925504207611084, + "learning_rate": 1.968920036898146e-05, + "loss": 0.1864060640335083, + "step": 4070 + }, + { + "epoch": 0.1074734197554848, + "grad_norm": 3.6029045581817627, + "learning_rate": 1.9687086402216033e-05, + "loss": 0.17147278785705566, + "step": 4080 + }, + { + "epoch": 0.10773683499998353, + "grad_norm": 1.027235507965088, + "learning_rate": 1.968496538477118e-05, + "loss": 0.1782658576965332, + "step": 4090 + }, + { + "epoch": 0.10800025024448227, + "grad_norm": 3.672891855239868, + "learning_rate": 1.9682837318190683e-05, + "loss": 0.1989532470703125, + "step": 4100 + }, + { + "epoch": 0.10826366548898102, + "grad_norm": 1.7158297300338745, + "learning_rate": 1.968070220402343e-05, + "loss": 0.15799436569213868, + "step": 4110 + }, + { + "epoch": 0.10852708073347975, + "grad_norm": 1.2952675819396973, + "learning_rate": 1.967856004382345e-05, + "loss": 0.1822509288787842, + "step": 4120 + }, + { + "epoch": 0.10879049597797849, + "grad_norm": 5.32310676574707, + "learning_rate": 1.96764108391499e-05, + "loss": 0.17983125448226928, + "step": 4130 + }, + { + "epoch": 0.10905391122247722, + "grad_norm": 7.700556755065918, + "learning_rate": 1.967425459156707e-05, + "loss": 0.1873644232749939, + "step": 4140 + }, + { + "epoch": 0.10931732646697596, + "grad_norm": 2.700819492340088, + "learning_rate": 1.9672091302644363e-05, + "loss": 0.175858211517334, + "step": 4150 + }, + { + "epoch": 0.1095807417114747, + "grad_norm": 5.8064494132995605, + "learning_rate": 1.966992097395632e-05, + "loss": 0.17721139192581176, + "step": 4160 + }, + { + "epoch": 0.10984415695597344, + "grad_norm": 1.3422290086746216, + "learning_rate": 1.9667743607082593e-05, + "loss": 0.17760924100875855, + "step": 4170 + }, + { + "epoch": 0.11010757220047217, + "grad_norm": 2.621978998184204, + "learning_rate": 1.9665559203607965e-05, + "loss": 0.170654296875, + "step": 4180 + }, + { + "epoch": 0.11037098744497091, + "grad_norm": 2.2802352905273438, + "learning_rate": 1.9663367765122346e-05, + "loss": 0.1804550528526306, + "step": 4190 + }, + { + "epoch": 0.11063440268946964, + "grad_norm": 1.4888993501663208, + "learning_rate": 1.966116929322075e-05, + "loss": 0.1824851632118225, + "step": 4200 + }, + { + "epoch": 0.11089781793396838, + "grad_norm": 2.390995979309082, + "learning_rate": 1.9658963789503328e-05, + "loss": 0.16869405508041382, + "step": 4210 + }, + { + "epoch": 0.11116123317846711, + "grad_norm": 1.5486677885055542, + "learning_rate": 1.9656751255575336e-05, + "loss": 0.15823016166687012, + "step": 4220 + }, + { + "epoch": 0.11142464842296586, + "grad_norm": 3.886641025543213, + "learning_rate": 1.965453169304715e-05, + "loss": 0.1804514169692993, + "step": 4230 + }, + { + "epoch": 0.1116880636674646, + "grad_norm": 4.962596893310547, + "learning_rate": 1.9652305103534265e-05, + "loss": 0.1831153392791748, + "step": 4240 + }, + { + "epoch": 0.11195147891196333, + "grad_norm": 2.0381572246551514, + "learning_rate": 1.9650071488657293e-05, + "loss": 0.17015918493270873, + "step": 4250 + }, + { + "epoch": 0.11221489415646207, + "grad_norm": 1.5309396982192993, + "learning_rate": 1.964783085004195e-05, + "loss": 0.17528939247131348, + "step": 4260 + }, + { + "epoch": 0.1124783094009608, + "grad_norm": 2.1703784465789795, + "learning_rate": 1.9645583189319074e-05, + "loss": 0.18402817249298095, + "step": 4270 + }, + { + "epoch": 0.11274172464545955, + "grad_norm": 5.970995903015137, + "learning_rate": 1.9643328508124605e-05, + "loss": 0.18465309143066405, + "step": 4280 + }, + { + "epoch": 0.11300513988995828, + "grad_norm": 3.0648131370544434, + "learning_rate": 1.9641066808099597e-05, + "loss": 0.1818598985671997, + "step": 4290 + }, + { + "epoch": 0.11326855513445702, + "grad_norm": 1.562493920326233, + "learning_rate": 1.963879809089022e-05, + "loss": 0.16390377283096313, + "step": 4300 + }, + { + "epoch": 0.11353197037895575, + "grad_norm": 3.747511863708496, + "learning_rate": 1.9636522358147737e-05, + "loss": 0.1843700647354126, + "step": 4310 + }, + { + "epoch": 0.1137953856234545, + "grad_norm": 3.20627760887146, + "learning_rate": 1.9634239611528527e-05, + "loss": 0.17664997577667235, + "step": 4320 + }, + { + "epoch": 0.11405880086795323, + "grad_norm": 2.003528118133545, + "learning_rate": 1.9631949852694073e-05, + "loss": 0.18137218952178955, + "step": 4330 + }, + { + "epoch": 0.11432221611245197, + "grad_norm": 3.667776107788086, + "learning_rate": 1.9629653083310956e-05, + "loss": 0.18416948318481446, + "step": 4340 + }, + { + "epoch": 0.1145856313569507, + "grad_norm": 3.3599917888641357, + "learning_rate": 1.9627349305050867e-05, + "loss": 0.18841917514801027, + "step": 4350 + }, + { + "epoch": 0.11484904660144944, + "grad_norm": 3.1923553943634033, + "learning_rate": 1.9625038519590596e-05, + "loss": 0.18241087198257447, + "step": 4360 + }, + { + "epoch": 0.11511246184594819, + "grad_norm": 7.041860580444336, + "learning_rate": 1.962272072861203e-05, + "loss": 0.19077067375183104, + "step": 4370 + }, + { + "epoch": 0.11537587709044692, + "grad_norm": 2.4864025115966797, + "learning_rate": 1.9620395933802158e-05, + "loss": 0.19220278263092042, + "step": 4380 + }, + { + "epoch": 0.11563929233494566, + "grad_norm": 2.319230079650879, + "learning_rate": 1.9618064136853066e-05, + "loss": 0.16494144201278688, + "step": 4390 + }, + { + "epoch": 0.11590270757944439, + "grad_norm": 2.3102433681488037, + "learning_rate": 1.9615725339461936e-05, + "loss": 0.18102731704711914, + "step": 4400 + }, + { + "epoch": 0.11616612282394313, + "grad_norm": 4.698543548583984, + "learning_rate": 1.961337954333105e-05, + "loss": 0.17309901714324952, + "step": 4410 + }, + { + "epoch": 0.11642953806844186, + "grad_norm": 1.887073040008545, + "learning_rate": 1.961102675016777e-05, + "loss": 0.17926366329193116, + "step": 4420 + }, + { + "epoch": 0.1166929533129406, + "grad_norm": 4.0832200050354, + "learning_rate": 1.960866696168457e-05, + "loss": 0.16114299297332763, + "step": 4430 + }, + { + "epoch": 0.11695636855743934, + "grad_norm": 3.5319437980651855, + "learning_rate": 1.9606300179599004e-05, + "loss": 0.16761524677276612, + "step": 4440 + }, + { + "epoch": 0.11721978380193808, + "grad_norm": 2.911208391189575, + "learning_rate": 1.9603926405633712e-05, + "loss": 0.1807692050933838, + "step": 4450 + }, + { + "epoch": 0.11748319904643681, + "grad_norm": 2.4235546588897705, + "learning_rate": 1.9601545641516434e-05, + "loss": 0.1848861575126648, + "step": 4460 + }, + { + "epoch": 0.11774661429093555, + "grad_norm": 3.4266066551208496, + "learning_rate": 1.9599157888979997e-05, + "loss": 0.17216044664382935, + "step": 4470 + }, + { + "epoch": 0.11801002953543428, + "grad_norm": 1.6716322898864746, + "learning_rate": 1.95967631497623e-05, + "loss": 0.17106916904449462, + "step": 4480 + }, + { + "epoch": 0.11827344477993303, + "grad_norm": 2.740271806716919, + "learning_rate": 1.9594361425606348e-05, + "loss": 0.17745668888092042, + "step": 4490 + }, + { + "epoch": 0.11853686002443177, + "grad_norm": 1.654542088508606, + "learning_rate": 1.959195271826021e-05, + "loss": 0.1688701868057251, + "step": 4500 + }, + { + "epoch": 0.1188002752689305, + "grad_norm": 3.3781795501708984, + "learning_rate": 1.9589537029477052e-05, + "loss": 0.1634523391723633, + "step": 4510 + }, + { + "epoch": 0.11906369051342924, + "grad_norm": 2.0217766761779785, + "learning_rate": 1.9587114361015118e-05, + "loss": 0.177728271484375, + "step": 4520 + }, + { + "epoch": 0.11932710575792797, + "grad_norm": 2.83096981048584, + "learning_rate": 1.958468471463773e-05, + "loss": 0.17717102766036988, + "step": 4530 + }, + { + "epoch": 0.11959052100242672, + "grad_norm": 3.5877749919891357, + "learning_rate": 1.9582248092113286e-05, + "loss": 0.17598862648010255, + "step": 4540 + }, + { + "epoch": 0.11985393624692545, + "grad_norm": 1.8635174036026, + "learning_rate": 1.9579804495215268e-05, + "loss": 0.16526496410369873, + "step": 4550 + }, + { + "epoch": 0.12011735149142419, + "grad_norm": 1.4270195960998535, + "learning_rate": 1.9577353925722228e-05, + "loss": 0.17655625343322753, + "step": 4560 + }, + { + "epoch": 0.12038076673592292, + "grad_norm": 1.4144021272659302, + "learning_rate": 1.9574896385417803e-05, + "loss": 0.18099462985992432, + "step": 4570 + }, + { + "epoch": 0.12064418198042166, + "grad_norm": 2.40080189704895, + "learning_rate": 1.957243187609069e-05, + "loss": 0.17441370487213134, + "step": 4580 + }, + { + "epoch": 0.1209075972249204, + "grad_norm": 2.7048280239105225, + "learning_rate": 1.9569960399534668e-05, + "loss": 0.17117936611175538, + "step": 4590 + }, + { + "epoch": 0.12117101246941914, + "grad_norm": 3.6085686683654785, + "learning_rate": 1.9567481957548585e-05, + "loss": 0.18381413221359252, + "step": 4600 + }, + { + "epoch": 0.12143442771391787, + "grad_norm": 4.486020088195801, + "learning_rate": 1.9564996551936354e-05, + "loss": 0.17837073802947997, + "step": 4610 + }, + { + "epoch": 0.12169784295841661, + "grad_norm": 3.4809787273406982, + "learning_rate": 1.9562504184506966e-05, + "loss": 0.17322902679443358, + "step": 4620 + }, + { + "epoch": 0.12196125820291535, + "grad_norm": 3.2241952419281006, + "learning_rate": 1.9560004857074472e-05, + "loss": 0.16990643739700317, + "step": 4630 + }, + { + "epoch": 0.12222467344741408, + "grad_norm": 2.378389835357666, + "learning_rate": 1.955749857145799e-05, + "loss": 0.1687392473220825, + "step": 4640 + }, + { + "epoch": 0.12248808869191283, + "grad_norm": 3.1733615398406982, + "learning_rate": 1.95549853294817e-05, + "loss": 0.17134263515472412, + "step": 4650 + }, + { + "epoch": 0.12275150393641156, + "grad_norm": 1.2213780879974365, + "learning_rate": 1.955246513297485e-05, + "loss": 0.188274085521698, + "step": 4660 + }, + { + "epoch": 0.1230149191809103, + "grad_norm": 1.2766227722167969, + "learning_rate": 1.9549937983771748e-05, + "loss": 0.18044764995574952, + "step": 4670 + }, + { + "epoch": 0.12327833442540903, + "grad_norm": 7.036100387573242, + "learning_rate": 1.954740388371176e-05, + "loss": 0.16549757719039918, + "step": 4680 + }, + { + "epoch": 0.12354174966990777, + "grad_norm": 1.956167221069336, + "learning_rate": 1.9544862834639316e-05, + "loss": 0.17114813327789308, + "step": 4690 + }, + { + "epoch": 0.1238051649144065, + "grad_norm": 1.9366813898086548, + "learning_rate": 1.9542314838403907e-05, + "loss": 0.16718895435333253, + "step": 4700 + }, + { + "epoch": 0.12406858015890525, + "grad_norm": 2.898319959640503, + "learning_rate": 1.9539759896860064e-05, + "loss": 0.18049087524414062, + "step": 4710 + }, + { + "epoch": 0.12433199540340398, + "grad_norm": 4.346360683441162, + "learning_rate": 1.9537198011867383e-05, + "loss": 0.18031893968582152, + "step": 4720 + }, + { + "epoch": 0.12459541064790272, + "grad_norm": 3.6843791007995605, + "learning_rate": 1.9534629185290524e-05, + "loss": 0.18354978561401367, + "step": 4730 + }, + { + "epoch": 0.12485882589240146, + "grad_norm": 2.3384857177734375, + "learning_rate": 1.9532053418999186e-05, + "loss": 0.1806301712989807, + "step": 4740 + }, + { + "epoch": 0.1251222411369002, + "grad_norm": 3.5594980716705322, + "learning_rate": 1.952947071486812e-05, + "loss": 0.18392765522003174, + "step": 4750 + }, + { + "epoch": 0.12538565638139892, + "grad_norm": 3.4826865196228027, + "learning_rate": 1.952688107477713e-05, + "loss": 0.16455860137939454, + "step": 4760 + }, + { + "epoch": 0.12564907162589767, + "grad_norm": 1.4731063842773438, + "learning_rate": 1.9524284500611076e-05, + "loss": 0.18932244777679444, + "step": 4770 + }, + { + "epoch": 0.1259124868703964, + "grad_norm": 2.245819330215454, + "learning_rate": 1.9521680994259845e-05, + "loss": 0.18019413948059082, + "step": 4780 + }, + { + "epoch": 0.12617590211489516, + "grad_norm": 2.0097711086273193, + "learning_rate": 1.9519070557618386e-05, + "loss": 0.18016436100006103, + "step": 4790 + }, + { + "epoch": 0.12643931735939387, + "grad_norm": 1.514467716217041, + "learning_rate": 1.9516453192586686e-05, + "loss": 0.1703273296356201, + "step": 4800 + }, + { + "epoch": 0.12670273260389261, + "grad_norm": 1.9529038667678833, + "learning_rate": 1.951382890106978e-05, + "loss": 0.17025848627090454, + "step": 4810 + }, + { + "epoch": 0.12696614784839136, + "grad_norm": 2.240128755569458, + "learning_rate": 1.951119768497774e-05, + "loss": 0.18745911121368408, + "step": 4820 + }, + { + "epoch": 0.1272295630928901, + "grad_norm": 1.7611162662506104, + "learning_rate": 1.9508559546225672e-05, + "loss": 0.16638420820236205, + "step": 4830 + }, + { + "epoch": 0.12749297833738882, + "grad_norm": 1.076245903968811, + "learning_rate": 1.9505914486733735e-05, + "loss": 0.17454710006713867, + "step": 4840 + }, + { + "epoch": 0.12775639358188756, + "grad_norm": 1.4830067157745361, + "learning_rate": 1.9503262508427115e-05, + "loss": 0.17827236652374268, + "step": 4850 + }, + { + "epoch": 0.1280198088263863, + "grad_norm": 3.162735939025879, + "learning_rate": 1.9500603613236034e-05, + "loss": 0.16573346853256227, + "step": 4860 + }, + { + "epoch": 0.12828322407088505, + "grad_norm": 1.4853084087371826, + "learning_rate": 1.9497937803095756e-05, + "loss": 0.1786785125732422, + "step": 4870 + }, + { + "epoch": 0.1285466393153838, + "grad_norm": 4.742806911468506, + "learning_rate": 1.9495265079946565e-05, + "loss": 0.1745316505432129, + "step": 4880 + }, + { + "epoch": 0.1288100545598825, + "grad_norm": 2.6637141704559326, + "learning_rate": 1.9492585445733788e-05, + "loss": 0.1887979745864868, + "step": 4890 + }, + { + "epoch": 0.12907346980438125, + "grad_norm": 3.2127342224121094, + "learning_rate": 1.9489898902407783e-05, + "loss": 0.1802167773246765, + "step": 4900 + }, + { + "epoch": 0.12933688504888, + "grad_norm": 4.673598289489746, + "learning_rate": 1.9487205451923925e-05, + "loss": 0.1724701404571533, + "step": 4910 + }, + { + "epoch": 0.12960030029337874, + "grad_norm": 6.352458477020264, + "learning_rate": 1.9484505096242628e-05, + "loss": 0.18303260803222657, + "step": 4920 + }, + { + "epoch": 0.12986371553787746, + "grad_norm": 4.700098037719727, + "learning_rate": 1.9481797837329323e-05, + "loss": 0.1692683219909668, + "step": 4930 + }, + { + "epoch": 0.1301271307823762, + "grad_norm": 2.313249111175537, + "learning_rate": 1.9479083677154474e-05, + "loss": 0.1677984595298767, + "step": 4940 + }, + { + "epoch": 0.13039054602687494, + "grad_norm": 1.1354424953460693, + "learning_rate": 1.9476362617693564e-05, + "loss": 0.1733534812927246, + "step": 4950 + }, + { + "epoch": 0.1306539612713737, + "grad_norm": 1.2695558071136475, + "learning_rate": 1.9473634660927092e-05, + "loss": 0.18294107913970947, + "step": 4960 + }, + { + "epoch": 0.1309173765158724, + "grad_norm": 5.287885665893555, + "learning_rate": 1.9470899808840585e-05, + "loss": 0.18378654718399048, + "step": 4970 + }, + { + "epoch": 0.13118079176037115, + "grad_norm": 2.3802990913391113, + "learning_rate": 1.9468158063424593e-05, + "loss": 0.1844257116317749, + "step": 4980 + }, + { + "epoch": 0.1314442070048699, + "grad_norm": 3.628051280975342, + "learning_rate": 1.946540942667467e-05, + "loss": 0.17003254890441893, + "step": 4990 + }, + { + "epoch": 0.13170762224936863, + "grad_norm": 1.0280840396881104, + "learning_rate": 1.9462653900591393e-05, + "loss": 0.1738024353981018, + "step": 5000 + }, + { + "epoch": 0.13197103749386738, + "grad_norm": 3.0836308002471924, + "learning_rate": 1.9459891487180353e-05, + "loss": 0.1703585386276245, + "step": 5010 + }, + { + "epoch": 0.1322344527383661, + "grad_norm": 1.4050794839859009, + "learning_rate": 1.9457122188452156e-05, + "loss": 0.15929946899414063, + "step": 5020 + }, + { + "epoch": 0.13249786798286484, + "grad_norm": 4.2166290283203125, + "learning_rate": 1.9454346006422414e-05, + "loss": 0.1726327657699585, + "step": 5030 + }, + { + "epoch": 0.13276128322736358, + "grad_norm": 1.7488353252410889, + "learning_rate": 1.9451562943111753e-05, + "loss": 0.18393173217773437, + "step": 5040 + }, + { + "epoch": 0.13302469847186232, + "grad_norm": 2.5261988639831543, + "learning_rate": 1.944877300054581e-05, + "loss": 0.186913001537323, + "step": 5050 + }, + { + "epoch": 0.13328811371636104, + "grad_norm": 2.451140880584717, + "learning_rate": 1.944597618075522e-05, + "loss": 0.17099488973617555, + "step": 5060 + }, + { + "epoch": 0.13355152896085978, + "grad_norm": 2.809166193008423, + "learning_rate": 1.944317248577563e-05, + "loss": 0.16449750661849977, + "step": 5070 + }, + { + "epoch": 0.13381494420535853, + "grad_norm": 2.565145254135132, + "learning_rate": 1.9440361917647693e-05, + "loss": 0.18006199598312378, + "step": 5080 + }, + { + "epoch": 0.13407835944985727, + "grad_norm": 1.4677820205688477, + "learning_rate": 1.943754447841706e-05, + "loss": 0.16598637104034425, + "step": 5090 + }, + { + "epoch": 0.134341774694356, + "grad_norm": 3.338259696960449, + "learning_rate": 1.9434720170134387e-05, + "loss": 0.17088088989257813, + "step": 5100 + }, + { + "epoch": 0.13460518993885473, + "grad_norm": 3.479022264480591, + "learning_rate": 1.9431888994855324e-05, + "loss": 0.17977800369262695, + "step": 5110 + }, + { + "epoch": 0.13486860518335347, + "grad_norm": 3.6745517253875732, + "learning_rate": 1.9429050954640528e-05, + "loss": 0.17024381160736085, + "step": 5120 + }, + { + "epoch": 0.13513202042785222, + "grad_norm": 3.9598851203918457, + "learning_rate": 1.942620605155564e-05, + "loss": 0.16883533000946044, + "step": 5130 + }, + { + "epoch": 0.13539543567235096, + "grad_norm": 1.4584859609603882, + "learning_rate": 1.9423354287671307e-05, + "loss": 0.1740166425704956, + "step": 5140 + }, + { + "epoch": 0.13565885091684968, + "grad_norm": 2.1775765419006348, + "learning_rate": 1.9420495665063166e-05, + "loss": 0.16992764472961425, + "step": 5150 + }, + { + "epoch": 0.13592226616134842, + "grad_norm": 1.1604734659194946, + "learning_rate": 1.9417630185811846e-05, + "loss": 0.1713045835494995, + "step": 5160 + }, + { + "epoch": 0.13618568140584716, + "grad_norm": 2.9775733947753906, + "learning_rate": 1.9414757852002962e-05, + "loss": 0.18422749042510986, + "step": 5170 + }, + { + "epoch": 0.1364490966503459, + "grad_norm": 3.200249433517456, + "learning_rate": 1.941187866572713e-05, + "loss": 0.16946697235107422, + "step": 5180 + }, + { + "epoch": 0.13671251189484462, + "grad_norm": 0.9890790581703186, + "learning_rate": 1.940899262907994e-05, + "loss": 0.1737157940864563, + "step": 5190 + }, + { + "epoch": 0.13697592713934337, + "grad_norm": 2.2065675258636475, + "learning_rate": 1.940609974416198e-05, + "loss": 0.17493653297424316, + "step": 5200 + }, + { + "epoch": 0.1372393423838421, + "grad_norm": 8.239404678344727, + "learning_rate": 1.940320001307881e-05, + "loss": 0.17767457962036132, + "step": 5210 + }, + { + "epoch": 0.13750275762834085, + "grad_norm": 2.7690041065216064, + "learning_rate": 1.940029343794098e-05, + "loss": 0.1830669164657593, + "step": 5220 + }, + { + "epoch": 0.13776617287283957, + "grad_norm": 1.3944640159606934, + "learning_rate": 1.9397380020864024e-05, + "loss": 0.1763041615486145, + "step": 5230 + }, + { + "epoch": 0.13802958811733831, + "grad_norm": 3.972710132598877, + "learning_rate": 1.939445976396845e-05, + "loss": 0.16932206153869628, + "step": 5240 + }, + { + "epoch": 0.13829300336183706, + "grad_norm": 1.6734015941619873, + "learning_rate": 1.939153266937975e-05, + "loss": 0.17236708402633666, + "step": 5250 + }, + { + "epoch": 0.1385564186063358, + "grad_norm": 1.7830928564071655, + "learning_rate": 1.938859873922839e-05, + "loss": 0.18013837337493896, + "step": 5260 + }, + { + "epoch": 0.13881983385083455, + "grad_norm": 2.1915645599365234, + "learning_rate": 1.9385657975649807e-05, + "loss": 0.17325162887573242, + "step": 5270 + }, + { + "epoch": 0.13908324909533326, + "grad_norm": 1.6690959930419922, + "learning_rate": 1.9382710380784418e-05, + "loss": 0.16862272024154662, + "step": 5280 + }, + { + "epoch": 0.139346664339832, + "grad_norm": 1.6115663051605225, + "learning_rate": 1.937975595677761e-05, + "loss": 0.16437160968780518, + "step": 5290 + }, + { + "epoch": 0.13961007958433075, + "grad_norm": 2.042649745941162, + "learning_rate": 1.9376794705779735e-05, + "loss": 0.1682317852973938, + "step": 5300 + }, + { + "epoch": 0.1398734948288295, + "grad_norm": 2.540118932723999, + "learning_rate": 1.937382662994613e-05, + "loss": 0.17348843812942505, + "step": 5310 + }, + { + "epoch": 0.1401369100733282, + "grad_norm": 2.478205442428589, + "learning_rate": 1.9370851731437077e-05, + "loss": 0.1747550845146179, + "step": 5320 + }, + { + "epoch": 0.14040032531782695, + "grad_norm": 5.557038307189941, + "learning_rate": 1.9367870012417844e-05, + "loss": 0.17934465408325195, + "step": 5330 + }, + { + "epoch": 0.1406637405623257, + "grad_norm": 1.008366346359253, + "learning_rate": 1.936488147505865e-05, + "loss": 0.17692177295684813, + "step": 5340 + }, + { + "epoch": 0.14092715580682444, + "grad_norm": 3.1162211894989014, + "learning_rate": 1.9361886121534686e-05, + "loss": 0.16576319932937622, + "step": 5350 + }, + { + "epoch": 0.14119057105132315, + "grad_norm": 1.4552398920059204, + "learning_rate": 1.93588839540261e-05, + "loss": 0.17411779165267943, + "step": 5360 + }, + { + "epoch": 0.1414539862958219, + "grad_norm": 2.2250707149505615, + "learning_rate": 1.9355874974717988e-05, + "loss": 0.16682896614074708, + "step": 5370 + }, + { + "epoch": 0.14171740154032064, + "grad_norm": 2.0385568141937256, + "learning_rate": 1.9352859185800424e-05, + "loss": 0.16301342248916625, + "step": 5380 + }, + { + "epoch": 0.14198081678481939, + "grad_norm": 1.7465003728866577, + "learning_rate": 1.934983658946843e-05, + "loss": 0.1749427080154419, + "step": 5390 + }, + { + "epoch": 0.14224423202931813, + "grad_norm": 2.8983042240142822, + "learning_rate": 1.9346807187921983e-05, + "loss": 0.16984254121780396, + "step": 5400 + }, + { + "epoch": 0.14250764727381685, + "grad_norm": 5.220624923706055, + "learning_rate": 1.9343770983366004e-05, + "loss": 0.1648175001144409, + "step": 5410 + }, + { + "epoch": 0.1427710625183156, + "grad_norm": 2.539048671722412, + "learning_rate": 1.9340727978010377e-05, + "loss": 0.1724213719367981, + "step": 5420 + }, + { + "epoch": 0.14303447776281433, + "grad_norm": 2.1186509132385254, + "learning_rate": 1.933767817406994e-05, + "loss": 0.16887811422348023, + "step": 5430 + }, + { + "epoch": 0.14329789300731308, + "grad_norm": 3.070924997329712, + "learning_rate": 1.9334621573764462e-05, + "loss": 0.17736185789108277, + "step": 5440 + }, + { + "epoch": 0.1435613082518118, + "grad_norm": 2.453111171722412, + "learning_rate": 1.9331558179318676e-05, + "loss": 0.17534419298171997, + "step": 5450 + }, + { + "epoch": 0.14382472349631054, + "grad_norm": 2.170525074005127, + "learning_rate": 1.9328487992962247e-05, + "loss": 0.17725106477737426, + "step": 5460 + }, + { + "epoch": 0.14408813874080928, + "grad_norm": 2.8742613792419434, + "learning_rate": 1.9325411016929793e-05, + "loss": 0.17953864336013795, + "step": 5470 + }, + { + "epoch": 0.14435155398530802, + "grad_norm": 1.3532480001449585, + "learning_rate": 1.9322327253460867e-05, + "loss": 0.17182319164276122, + "step": 5480 + }, + { + "epoch": 0.14461496922980674, + "grad_norm": 2.2898712158203125, + "learning_rate": 1.931923670479997e-05, + "loss": 0.16850945949554444, + "step": 5490 + }, + { + "epoch": 0.14487838447430548, + "grad_norm": 1.0998563766479492, + "learning_rate": 1.9316139373196533e-05, + "loss": 0.17255064249038696, + "step": 5500 + }, + { + "epoch": 0.14514179971880423, + "grad_norm": 2.68060564994812, + "learning_rate": 1.9313035260904923e-05, + "loss": 0.179382061958313, + "step": 5510 + }, + { + "epoch": 0.14540521496330297, + "grad_norm": 2.4581732749938965, + "learning_rate": 1.9309924370184455e-05, + "loss": 0.17087984085083008, + "step": 5520 + }, + { + "epoch": 0.1456686302078017, + "grad_norm": 4.17896032333374, + "learning_rate": 1.9306806703299367e-05, + "loss": 0.1826640248298645, + "step": 5530 + }, + { + "epoch": 0.14593204545230043, + "grad_norm": 2.7782535552978516, + "learning_rate": 1.9303682262518828e-05, + "loss": 0.18576438426971437, + "step": 5540 + }, + { + "epoch": 0.14619546069679917, + "grad_norm": 2.4362518787384033, + "learning_rate": 1.9300551050116946e-05, + "loss": 0.17349045276641845, + "step": 5550 + }, + { + "epoch": 0.14645887594129792, + "grad_norm": 3.102874517440796, + "learning_rate": 1.9297413068372745e-05, + "loss": 0.160721492767334, + "step": 5560 + }, + { + "epoch": 0.14672229118579666, + "grad_norm": 4.1335248947143555, + "learning_rate": 1.9294268319570195e-05, + "loss": 0.16471575498580932, + "step": 5570 + }, + { + "epoch": 0.14698570643029538, + "grad_norm": 3.112025022506714, + "learning_rate": 1.9291116805998168e-05, + "loss": 0.18155699968338013, + "step": 5580 + }, + { + "epoch": 0.14724912167479412, + "grad_norm": 1.4145865440368652, + "learning_rate": 1.9287958529950477e-05, + "loss": 0.17428386211395264, + "step": 5590 + }, + { + "epoch": 0.14751253691929286, + "grad_norm": 3.9216887950897217, + "learning_rate": 1.9284793493725852e-05, + "loss": 0.17609132528305055, + "step": 5600 + }, + { + "epoch": 0.1477759521637916, + "grad_norm": 3.4289698600769043, + "learning_rate": 1.9281621699627943e-05, + "loss": 0.17075462341308595, + "step": 5610 + }, + { + "epoch": 0.14803936740829032, + "grad_norm": 1.6094110012054443, + "learning_rate": 1.927844314996532e-05, + "loss": 0.15933669805526735, + "step": 5620 + }, + { + "epoch": 0.14830278265278907, + "grad_norm": 4.1563897132873535, + "learning_rate": 1.9275257847051464e-05, + "loss": 0.1707752227783203, + "step": 5630 + }, + { + "epoch": 0.1485661978972878, + "grad_norm": 4.421563625335693, + "learning_rate": 1.9272065793204778e-05, + "loss": 0.16583887338638306, + "step": 5640 + }, + { + "epoch": 0.14882961314178655, + "grad_norm": 4.686405181884766, + "learning_rate": 1.9268866990748578e-05, + "loss": 0.16623942852020263, + "step": 5650 + }, + { + "epoch": 0.1490930283862853, + "grad_norm": 2.036879062652588, + "learning_rate": 1.9265661442011087e-05, + "loss": 0.19060674905776978, + "step": 5660 + }, + { + "epoch": 0.149356443630784, + "grad_norm": 6.762901306152344, + "learning_rate": 1.9262449149325445e-05, + "loss": 0.18554489612579345, + "step": 5670 + }, + { + "epoch": 0.14961985887528276, + "grad_norm": 1.0937964916229248, + "learning_rate": 1.92592301150297e-05, + "loss": 0.16549305915832518, + "step": 5680 + }, + { + "epoch": 0.1498832741197815, + "grad_norm": 1.4695836305618286, + "learning_rate": 1.9256004341466795e-05, + "loss": 0.16770973205566406, + "step": 5690 + }, + { + "epoch": 0.15014668936428024, + "grad_norm": 3.7794127464294434, + "learning_rate": 1.9252771830984594e-05, + "loss": 0.1526745319366455, + "step": 5700 + }, + { + "epoch": 0.15041010460877896, + "grad_norm": 6.774730682373047, + "learning_rate": 1.9249532585935855e-05, + "loss": 0.1697704792022705, + "step": 5710 + }, + { + "epoch": 0.1506735198532777, + "grad_norm": 5.374912738800049, + "learning_rate": 1.9246286608678243e-05, + "loss": 0.1672649621963501, + "step": 5720 + }, + { + "epoch": 0.15093693509777645, + "grad_norm": 2.166128396987915, + "learning_rate": 1.9243033901574316e-05, + "loss": 0.1733180284500122, + "step": 5730 + }, + { + "epoch": 0.1512003503422752, + "grad_norm": 5.7467427253723145, + "learning_rate": 1.923977446699154e-05, + "loss": 0.17819337844848632, + "step": 5740 + }, + { + "epoch": 0.1514637655867739, + "grad_norm": 4.536844730377197, + "learning_rate": 1.923650830730227e-05, + "loss": 0.16845345497131348, + "step": 5750 + }, + { + "epoch": 0.15172718083127265, + "grad_norm": 4.168144702911377, + "learning_rate": 1.923323542488375e-05, + "loss": 0.16591516733169556, + "step": 5760 + }, + { + "epoch": 0.1519905960757714, + "grad_norm": 1.8950287103652954, + "learning_rate": 1.9229955822118138e-05, + "loss": 0.17191482782363893, + "step": 5770 + }, + { + "epoch": 0.15225401132027014, + "grad_norm": 2.992539882659912, + "learning_rate": 1.9226669501392465e-05, + "loss": 0.17090003490447997, + "step": 5780 + }, + { + "epoch": 0.15251742656476888, + "grad_norm": 2.3274121284484863, + "learning_rate": 1.9223376465098657e-05, + "loss": 0.16439394950866698, + "step": 5790 + }, + { + "epoch": 0.1527808418092676, + "grad_norm": 7.540190696716309, + "learning_rate": 1.9220076715633526e-05, + "loss": 0.17066435813903807, + "step": 5800 + }, + { + "epoch": 0.15304425705376634, + "grad_norm": 3.5165903568267822, + "learning_rate": 1.921677025539877e-05, + "loss": 0.16616518497467042, + "step": 5810 + }, + { + "epoch": 0.15330767229826509, + "grad_norm": 1.3686691522598267, + "learning_rate": 1.921345708680098e-05, + "loss": 0.1653651237487793, + "step": 5820 + }, + { + "epoch": 0.15357108754276383, + "grad_norm": 1.8440113067626953, + "learning_rate": 1.921013721225162e-05, + "loss": 0.1650429129600525, + "step": 5830 + }, + { + "epoch": 0.15383450278726254, + "grad_norm": 3.6867387294769287, + "learning_rate": 1.9206810634167036e-05, + "loss": 0.15265412330627443, + "step": 5840 + }, + { + "epoch": 0.1540979180317613, + "grad_norm": 1.6172244548797607, + "learning_rate": 1.920347735496846e-05, + "loss": 0.16188249588012696, + "step": 5850 + }, + { + "epoch": 0.15436133327626003, + "grad_norm": 1.2509324550628662, + "learning_rate": 1.9200137377081993e-05, + "loss": 0.16903424263000488, + "step": 5860 + }, + { + "epoch": 0.15462474852075878, + "grad_norm": 2.343607187271118, + "learning_rate": 1.919679070293861e-05, + "loss": 0.16258904933929444, + "step": 5870 + }, + { + "epoch": 0.1548881637652575, + "grad_norm": 3.9457740783691406, + "learning_rate": 1.9193437334974178e-05, + "loss": 0.15988292694091796, + "step": 5880 + }, + { + "epoch": 0.15515157900975624, + "grad_norm": 5.432035446166992, + "learning_rate": 1.9190077275629408e-05, + "loss": 0.16355471611022948, + "step": 5890 + }, + { + "epoch": 0.15541499425425498, + "grad_norm": 1.3304089307785034, + "learning_rate": 1.9186710527349907e-05, + "loss": 0.16574712991714477, + "step": 5900 + }, + { + "epoch": 0.15567840949875372, + "grad_norm": 5.167713642120361, + "learning_rate": 1.9183337092586137e-05, + "loss": 0.17007663249969482, + "step": 5910 + }, + { + "epoch": 0.15594182474325247, + "grad_norm": 1.8150618076324463, + "learning_rate": 1.9179956973793422e-05, + "loss": 0.16641359329223632, + "step": 5920 + }, + { + "epoch": 0.15620523998775118, + "grad_norm": 1.3111382722854614, + "learning_rate": 1.917657017343197e-05, + "loss": 0.17059872150421143, + "step": 5930 + }, + { + "epoch": 0.15646865523224993, + "grad_norm": 2.8257627487182617, + "learning_rate": 1.9173176693966835e-05, + "loss": 0.1743929386138916, + "step": 5940 + }, + { + "epoch": 0.15673207047674867, + "grad_norm": 3.2543110847473145, + "learning_rate": 1.9169776537867936e-05, + "loss": 0.17329413890838624, + "step": 5950 + }, + { + "epoch": 0.1569954857212474, + "grad_norm": 2.730203866958618, + "learning_rate": 1.9166369707610055e-05, + "loss": 0.16494884490966796, + "step": 5960 + }, + { + "epoch": 0.15725890096574613, + "grad_norm": 1.9956934452056885, + "learning_rate": 1.9162956205672833e-05, + "loss": 0.17909886837005615, + "step": 5970 + }, + { + "epoch": 0.15752231621024487, + "grad_norm": 0.9318256378173828, + "learning_rate": 1.9159536034540757e-05, + "loss": 0.17097859382629393, + "step": 5980 + }, + { + "epoch": 0.15778573145474362, + "grad_norm": 2.7959866523742676, + "learning_rate": 1.9156109196703177e-05, + "loss": 0.17129623889923096, + "step": 5990 + }, + { + "epoch": 0.15804914669924236, + "grad_norm": 1.1599644422531128, + "learning_rate": 1.91526756946543e-05, + "loss": 0.17501790523529054, + "step": 6000 + }, + { + "epoch": 0.15831256194374108, + "grad_norm": 6.316742420196533, + "learning_rate": 1.9149235530893172e-05, + "loss": 0.1769660472869873, + "step": 6010 + }, + { + "epoch": 0.15857597718823982, + "grad_norm": 2.1791491508483887, + "learning_rate": 1.9145788707923692e-05, + "loss": 0.18230912685394288, + "step": 6020 + }, + { + "epoch": 0.15883939243273856, + "grad_norm": 2.027712106704712, + "learning_rate": 1.914233522825461e-05, + "loss": 0.17117310762405397, + "step": 6030 + }, + { + "epoch": 0.1591028076772373, + "grad_norm": 3.6611745357513428, + "learning_rate": 1.913887509439951e-05, + "loss": 0.16736054420471191, + "step": 6040 + }, + { + "epoch": 0.15936622292173605, + "grad_norm": 3.948150396347046, + "learning_rate": 1.9135408308876834e-05, + "loss": 0.17750407457351686, + "step": 6050 + }, + { + "epoch": 0.15962963816623477, + "grad_norm": 1.2617881298065186, + "learning_rate": 1.913193487420986e-05, + "loss": 0.1658021926879883, + "step": 6060 + }, + { + "epoch": 0.1598930534107335, + "grad_norm": 1.5301584005355835, + "learning_rate": 1.9128454792926694e-05, + "loss": 0.16524035930633546, + "step": 6070 + }, + { + "epoch": 0.16015646865523225, + "grad_norm": 4.333000659942627, + "learning_rate": 1.9124968067560296e-05, + "loss": 0.17392778396606445, + "step": 6080 + }, + { + "epoch": 0.160419883899731, + "grad_norm": 4.03311824798584, + "learning_rate": 1.9121474700648454e-05, + "loss": 0.16975760459899902, + "step": 6090 + }, + { + "epoch": 0.1606832991442297, + "grad_norm": 3.593534469604492, + "learning_rate": 1.9117974694733795e-05, + "loss": 0.17366198301315308, + "step": 6100 + }, + { + "epoch": 0.16094671438872846, + "grad_norm": 3.1581127643585205, + "learning_rate": 1.9114468052363764e-05, + "loss": 0.18065620660781861, + "step": 6110 + }, + { + "epoch": 0.1612101296332272, + "grad_norm": 1.6253399848937988, + "learning_rate": 1.9110954776090656e-05, + "loss": 0.17199116945266724, + "step": 6120 + }, + { + "epoch": 0.16147354487772594, + "grad_norm": 3.1436872482299805, + "learning_rate": 1.910743486847158e-05, + "loss": 0.1631827473640442, + "step": 6130 + }, + { + "epoch": 0.16173696012222466, + "grad_norm": 1.5395588874816895, + "learning_rate": 1.9103908332068483e-05, + "loss": 0.17242989540100098, + "step": 6140 + }, + { + "epoch": 0.1620003753667234, + "grad_norm": 3.797245740890503, + "learning_rate": 1.910037516944812e-05, + "loss": 0.15920908451080323, + "step": 6150 + }, + { + "epoch": 0.16226379061122215, + "grad_norm": 3.6731274127960205, + "learning_rate": 1.909683538318209e-05, + "loss": 0.17110397815704345, + "step": 6160 + }, + { + "epoch": 0.1625272058557209, + "grad_norm": 3.396333694458008, + "learning_rate": 1.9093288975846797e-05, + "loss": 0.1648148775100708, + "step": 6170 + }, + { + "epoch": 0.16279062110021963, + "grad_norm": 1.5526561737060547, + "learning_rate": 1.9089735950023463e-05, + "loss": 0.17363948822021485, + "step": 6180 + }, + { + "epoch": 0.16305403634471835, + "grad_norm": 1.4899340867996216, + "learning_rate": 1.9086176308298147e-05, + "loss": 0.1698089838027954, + "step": 6190 + }, + { + "epoch": 0.1633174515892171, + "grad_norm": 8.666157722473145, + "learning_rate": 1.90826100532617e-05, + "loss": 0.1837103009223938, + "step": 6200 + }, + { + "epoch": 0.16358086683371584, + "grad_norm": 3.0068657398223877, + "learning_rate": 1.90790371875098e-05, + "loss": 0.16790030002593995, + "step": 6210 + }, + { + "epoch": 0.16384428207821458, + "grad_norm": 1.4281753301620483, + "learning_rate": 1.9075457713642934e-05, + "loss": 0.1774059534072876, + "step": 6220 + }, + { + "epoch": 0.1641076973227133, + "grad_norm": 1.5582479238510132, + "learning_rate": 1.9071871634266392e-05, + "loss": 0.16697818040847778, + "step": 6230 + }, + { + "epoch": 0.16437111256721204, + "grad_norm": 1.5157357454299927, + "learning_rate": 1.9068278951990282e-05, + "loss": 0.1590636134147644, + "step": 6240 + }, + { + "epoch": 0.16463452781171078, + "grad_norm": 4.909310817718506, + "learning_rate": 1.9064679669429515e-05, + "loss": 0.1702186107635498, + "step": 6250 + }, + { + "epoch": 0.16489794305620953, + "grad_norm": 3.177954912185669, + "learning_rate": 1.9061073789203798e-05, + "loss": 0.1665342330932617, + "step": 6260 + }, + { + "epoch": 0.16516135830070824, + "grad_norm": 1.1850922107696533, + "learning_rate": 1.9057461313937646e-05, + "loss": 0.18224472999572755, + "step": 6270 + }, + { + "epoch": 0.165424773545207, + "grad_norm": 1.9878414869308472, + "learning_rate": 1.905384224626038e-05, + "loss": 0.18065929412841797, + "step": 6280 + }, + { + "epoch": 0.16568818878970573, + "grad_norm": 3.8000614643096924, + "learning_rate": 1.9050216588806113e-05, + "loss": 0.17412657737731935, + "step": 6290 + }, + { + "epoch": 0.16595160403420448, + "grad_norm": 4.074619770050049, + "learning_rate": 1.9046584344213754e-05, + "loss": 0.16042897701263428, + "step": 6300 + }, + { + "epoch": 0.16621501927870322, + "grad_norm": 3.3307929039001465, + "learning_rate": 1.9042945515127004e-05, + "loss": 0.16181868314743042, + "step": 6310 + }, + { + "epoch": 0.16647843452320193, + "grad_norm": 2.1914682388305664, + "learning_rate": 1.9039300104194366e-05, + "loss": 0.16467502117156982, + "step": 6320 + }, + { + "epoch": 0.16674184976770068, + "grad_norm": 2.866262197494507, + "learning_rate": 1.903564811406912e-05, + "loss": 0.17197684049606324, + "step": 6330 + }, + { + "epoch": 0.16700526501219942, + "grad_norm": 4.397765636444092, + "learning_rate": 1.9031989547409347e-05, + "loss": 0.16158585548400878, + "step": 6340 + }, + { + "epoch": 0.16726868025669817, + "grad_norm": 1.1589778661727905, + "learning_rate": 1.9028324406877905e-05, + "loss": 0.15754142999649048, + "step": 6350 + }, + { + "epoch": 0.16753209550119688, + "grad_norm": 1.994996190071106, + "learning_rate": 1.9024652695142447e-05, + "loss": 0.1568400740623474, + "step": 6360 + }, + { + "epoch": 0.16779551074569563, + "grad_norm": 1.7435288429260254, + "learning_rate": 1.90209744148754e-05, + "loss": 0.17211060523986815, + "step": 6370 + }, + { + "epoch": 0.16805892599019437, + "grad_norm": 3.781378984451294, + "learning_rate": 1.901728956875398e-05, + "loss": 0.16037373542785643, + "step": 6380 + }, + { + "epoch": 0.1683223412346931, + "grad_norm": 1.4683865308761597, + "learning_rate": 1.9013598159460168e-05, + "loss": 0.1542722463607788, + "step": 6390 + }, + { + "epoch": 0.16858575647919183, + "grad_norm": 4.501943588256836, + "learning_rate": 1.900990018968074e-05, + "loss": 0.16164885759353637, + "step": 6400 + }, + { + "epoch": 0.16884917172369057, + "grad_norm": 1.534532070159912, + "learning_rate": 1.900619566210723e-05, + "loss": 0.14781076908111573, + "step": 6410 + }, + { + "epoch": 0.16911258696818932, + "grad_norm": 1.4801349639892578, + "learning_rate": 1.900248457943596e-05, + "loss": 0.16195423603057862, + "step": 6420 + }, + { + "epoch": 0.16937600221268806, + "grad_norm": 5.090292453765869, + "learning_rate": 1.899876694436801e-05, + "loss": 0.1718473792076111, + "step": 6430 + }, + { + "epoch": 0.1696394174571868, + "grad_norm": 1.795758843421936, + "learning_rate": 1.899504275960924e-05, + "loss": 0.16729098558425903, + "step": 6440 + }, + { + "epoch": 0.16990283270168552, + "grad_norm": 4.113773822784424, + "learning_rate": 1.899131202787027e-05, + "loss": 0.17048470973968505, + "step": 6450 + }, + { + "epoch": 0.17016624794618426, + "grad_norm": 1.6368587017059326, + "learning_rate": 1.8987574751866488e-05, + "loss": 0.19026503562927247, + "step": 6460 + }, + { + "epoch": 0.170429663190683, + "grad_norm": 5.414082050323486, + "learning_rate": 1.8983830934318044e-05, + "loss": 0.17780473232269287, + "step": 6470 + }, + { + "epoch": 0.17069307843518175, + "grad_norm": 4.594887733459473, + "learning_rate": 1.8980080577949848e-05, + "loss": 0.17586157321929932, + "step": 6480 + }, + { + "epoch": 0.17095649367968047, + "grad_norm": 3.959243059158325, + "learning_rate": 1.8976323685491577e-05, + "loss": 0.1751037836074829, + "step": 6490 + }, + { + "epoch": 0.1712199089241792, + "grad_norm": 2.8979671001434326, + "learning_rate": 1.8972560259677658e-05, + "loss": 0.169608736038208, + "step": 6500 + }, + { + "epoch": 0.17148332416867795, + "grad_norm": 5.959170341491699, + "learning_rate": 1.896879030324727e-05, + "loss": 0.165480637550354, + "step": 6510 + }, + { + "epoch": 0.1717467394131767, + "grad_norm": 1.1680361032485962, + "learning_rate": 1.8965013818944357e-05, + "loss": 0.17120769023895263, + "step": 6520 + }, + { + "epoch": 0.1720101546576754, + "grad_norm": 2.350128650665283, + "learning_rate": 1.89612308095176e-05, + "loss": 0.17026820182800292, + "step": 6530 + }, + { + "epoch": 0.17227356990217416, + "grad_norm": 2.1626827716827393, + "learning_rate": 1.8957441277720438e-05, + "loss": 0.17408860921859742, + "step": 6540 + }, + { + "epoch": 0.1725369851466729, + "grad_norm": 1.796492338180542, + "learning_rate": 1.8953645226311056e-05, + "loss": 0.15839898586273193, + "step": 6550 + }, + { + "epoch": 0.17280040039117164, + "grad_norm": 1.6793272495269775, + "learning_rate": 1.894984265805239e-05, + "loss": 0.18677401542663574, + "step": 6560 + }, + { + "epoch": 0.1730638156356704, + "grad_norm": 1.949769377708435, + "learning_rate": 1.8946033575712104e-05, + "loss": 0.15959689617156983, + "step": 6570 + }, + { + "epoch": 0.1733272308801691, + "grad_norm": 2.258028030395508, + "learning_rate": 1.894221798206262e-05, + "loss": 0.1761033058166504, + "step": 6580 + }, + { + "epoch": 0.17359064612466785, + "grad_norm": 4.173799991607666, + "learning_rate": 1.8938395879881086e-05, + "loss": 0.17830733060836793, + "step": 6590 + }, + { + "epoch": 0.1738540613691666, + "grad_norm": 1.2828105688095093, + "learning_rate": 1.8934567271949397e-05, + "loss": 0.1847146987915039, + "step": 6600 + }, + { + "epoch": 0.17411747661366533, + "grad_norm": 2.092668294906616, + "learning_rate": 1.8930732161054176e-05, + "loss": 0.15951205492019654, + "step": 6610 + }, + { + "epoch": 0.17438089185816405, + "grad_norm": 4.46002197265625, + "learning_rate": 1.8926890549986786e-05, + "loss": 0.16805381774902345, + "step": 6620 + }, + { + "epoch": 0.1746443071026628, + "grad_norm": 2.486891746520996, + "learning_rate": 1.8923042441543313e-05, + "loss": 0.16804299354553223, + "step": 6630 + }, + { + "epoch": 0.17490772234716154, + "grad_norm": 5.411587238311768, + "learning_rate": 1.891918783852458e-05, + "loss": 0.18035318851470947, + "step": 6640 + }, + { + "epoch": 0.17517113759166028, + "grad_norm": 0.804013729095459, + "learning_rate": 1.8915326743736134e-05, + "loss": 0.15399959087371826, + "step": 6650 + }, + { + "epoch": 0.175434552836159, + "grad_norm": 4.3981404304504395, + "learning_rate": 1.8911459159988238e-05, + "loss": 0.17601872682571412, + "step": 6660 + }, + { + "epoch": 0.17569796808065774, + "grad_norm": 1.339983344078064, + "learning_rate": 1.8907585090095897e-05, + "loss": 0.16894770860671998, + "step": 6670 + }, + { + "epoch": 0.17596138332515648, + "grad_norm": 3.78580904006958, + "learning_rate": 1.890370453687882e-05, + "loss": 0.17066690921783448, + "step": 6680 + }, + { + "epoch": 0.17622479856965523, + "grad_norm": 2.7853739261627197, + "learning_rate": 1.889981750316144e-05, + "loss": 0.1842102289199829, + "step": 6690 + }, + { + "epoch": 0.17648821381415397, + "grad_norm": 1.7369493246078491, + "learning_rate": 1.8895923991772913e-05, + "loss": 0.17612870931625366, + "step": 6700 + }, + { + "epoch": 0.1767516290586527, + "grad_norm": 2.2136363983154297, + "learning_rate": 1.8892024005547104e-05, + "loss": 0.18044610023498536, + "step": 6710 + }, + { + "epoch": 0.17701504430315143, + "grad_norm": 5.091931343078613, + "learning_rate": 1.8888117547322586e-05, + "loss": 0.1652448892593384, + "step": 6720 + }, + { + "epoch": 0.17727845954765017, + "grad_norm": 2.5525481700897217, + "learning_rate": 1.888420461994265e-05, + "loss": 0.1728058338165283, + "step": 6730 + }, + { + "epoch": 0.17754187479214892, + "grad_norm": 3.6554055213928223, + "learning_rate": 1.88802852262553e-05, + "loss": 0.16762619018554686, + "step": 6740 + }, + { + "epoch": 0.17780529003664763, + "grad_norm": 3.4655208587646484, + "learning_rate": 1.887635936911323e-05, + "loss": 0.17684130668640136, + "step": 6750 + }, + { + "epoch": 0.17806870528114638, + "grad_norm": 1.632101058959961, + "learning_rate": 1.8872427051373852e-05, + "loss": 0.16861413717269896, + "step": 6760 + }, + { + "epoch": 0.17833212052564512, + "grad_norm": 1.4289665222167969, + "learning_rate": 1.886848827589928e-05, + "loss": 0.1662585496902466, + "step": 6770 + }, + { + "epoch": 0.17859553577014387, + "grad_norm": 2.975188732147217, + "learning_rate": 1.8864543045556318e-05, + "loss": 0.17488429546356202, + "step": 6780 + }, + { + "epoch": 0.1788589510146426, + "grad_norm": 6.324111461639404, + "learning_rate": 1.8860591363216482e-05, + "loss": 0.1771680235862732, + "step": 6790 + }, + { + "epoch": 0.17912236625914132, + "grad_norm": 1.9030544757843018, + "learning_rate": 1.8856633231755972e-05, + "loss": 0.17557811737060547, + "step": 6800 + }, + { + "epoch": 0.17938578150364007, + "grad_norm": 1.8994438648223877, + "learning_rate": 1.8852668654055686e-05, + "loss": 0.1684272289276123, + "step": 6810 + }, + { + "epoch": 0.1796491967481388, + "grad_norm": 2.3021652698516846, + "learning_rate": 1.884869763300122e-05, + "loss": 0.16774260997772217, + "step": 6820 + }, + { + "epoch": 0.17991261199263756, + "grad_norm": 1.7774169445037842, + "learning_rate": 1.8844720171482854e-05, + "loss": 0.15951805114746093, + "step": 6830 + }, + { + "epoch": 0.18017602723713627, + "grad_norm": 2.262357711791992, + "learning_rate": 1.884073627239555e-05, + "loss": 0.17045761346817018, + "step": 6840 + }, + { + "epoch": 0.18043944248163502, + "grad_norm": 4.096415996551514, + "learning_rate": 1.8836745938638963e-05, + "loss": 0.1755443334579468, + "step": 6850 + }, + { + "epoch": 0.18070285772613376, + "grad_norm": 1.9613577127456665, + "learning_rate": 1.8832749173117435e-05, + "loss": 0.17232749462127686, + "step": 6860 + }, + { + "epoch": 0.1809662729706325, + "grad_norm": 2.6770877838134766, + "learning_rate": 1.8828745978739982e-05, + "loss": 0.17023165225982667, + "step": 6870 + }, + { + "epoch": 0.18122968821513122, + "grad_norm": 0.9537907838821411, + "learning_rate": 1.88247363584203e-05, + "loss": 0.16179245710372925, + "step": 6880 + }, + { + "epoch": 0.18149310345962996, + "grad_norm": 4.312086582183838, + "learning_rate": 1.8820720315076767e-05, + "loss": 0.17039222717285157, + "step": 6890 + }, + { + "epoch": 0.1817565187041287, + "grad_norm": 2.052088975906372, + "learning_rate": 1.8816697851632425e-05, + "loss": 0.1823539137840271, + "step": 6900 + }, + { + "epoch": 0.18201993394862745, + "grad_norm": 1.8238223791122437, + "learning_rate": 1.8812668971015006e-05, + "loss": 0.16873868703842163, + "step": 6910 + }, + { + "epoch": 0.1822833491931262, + "grad_norm": 1.29923415184021, + "learning_rate": 1.88086336761569e-05, + "loss": 0.16440414190292357, + "step": 6920 + }, + { + "epoch": 0.1825467644376249, + "grad_norm": 1.475290060043335, + "learning_rate": 1.8804591969995167e-05, + "loss": 0.17073442935943603, + "step": 6930 + }, + { + "epoch": 0.18281017968212365, + "grad_norm": 2.1828582286834717, + "learning_rate": 1.8800543855471535e-05, + "loss": 0.1700200319290161, + "step": 6940 + }, + { + "epoch": 0.1830735949266224, + "grad_norm": 1.740129828453064, + "learning_rate": 1.87964893355324e-05, + "loss": 0.16085003614425658, + "step": 6950 + }, + { + "epoch": 0.18333701017112114, + "grad_norm": 3.5461034774780273, + "learning_rate": 1.879242841312881e-05, + "loss": 0.16853599548339843, + "step": 6960 + }, + { + "epoch": 0.18360042541561986, + "grad_norm": 4.716802597045898, + "learning_rate": 1.878836109121648e-05, + "loss": 0.17098926305770873, + "step": 6970 + }, + { + "epoch": 0.1838638406601186, + "grad_norm": 2.5136687755584717, + "learning_rate": 1.8784287372755795e-05, + "loss": 0.18581616878509521, + "step": 6980 + }, + { + "epoch": 0.18412725590461734, + "grad_norm": 5.607929229736328, + "learning_rate": 1.8780207260711767e-05, + "loss": 0.17855950593948364, + "step": 6990 + }, + { + "epoch": 0.1843906711491161, + "grad_norm": 1.8399549722671509, + "learning_rate": 1.8776120758054087e-05, + "loss": 0.17461702823638917, + "step": 7000 + }, + { + "epoch": 0.1846540863936148, + "grad_norm": 3.180877447128296, + "learning_rate": 1.8772027867757082e-05, + "loss": 0.1682128667831421, + "step": 7010 + }, + { + "epoch": 0.18491750163811355, + "grad_norm": 4.2219438552856445, + "learning_rate": 1.8767928592799742e-05, + "loss": 0.16773054599761963, + "step": 7020 + }, + { + "epoch": 0.1851809168826123, + "grad_norm": 4.523402690887451, + "learning_rate": 1.876382293616569e-05, + "loss": 0.17730321884155273, + "step": 7030 + }, + { + "epoch": 0.18544433212711103, + "grad_norm": 1.3975764513015747, + "learning_rate": 1.8759710900843202e-05, + "loss": 0.16810944080352783, + "step": 7040 + }, + { + "epoch": 0.18570774737160978, + "grad_norm": 1.929778814315796, + "learning_rate": 1.8755592489825193e-05, + "loss": 0.152152419090271, + "step": 7050 + }, + { + "epoch": 0.1859711626161085, + "grad_norm": 2.1797986030578613, + "learning_rate": 1.8751467706109224e-05, + "loss": 0.18132548332214354, + "step": 7060 + }, + { + "epoch": 0.18623457786060724, + "grad_norm": 4.448251247406006, + "learning_rate": 1.8747336552697483e-05, + "loss": 0.17027875185012817, + "step": 7070 + }, + { + "epoch": 0.18649799310510598, + "grad_norm": 1.3856507539749146, + "learning_rate": 1.874319903259681e-05, + "loss": 0.1688234329223633, + "step": 7080 + }, + { + "epoch": 0.18676140834960472, + "grad_norm": 1.338340163230896, + "learning_rate": 1.873905514881867e-05, + "loss": 0.18039542436599731, + "step": 7090 + }, + { + "epoch": 0.18702482359410344, + "grad_norm": 4.548517227172852, + "learning_rate": 1.873490490437915e-05, + "loss": 0.16035857200622558, + "step": 7100 + }, + { + "epoch": 0.18728823883860218, + "grad_norm": 5.830115795135498, + "learning_rate": 1.8730748302298993e-05, + "loss": 0.173580801486969, + "step": 7110 + }, + { + "epoch": 0.18755165408310093, + "grad_norm": 2.3270809650421143, + "learning_rate": 1.8726585345603544e-05, + "loss": 0.20042080879211427, + "step": 7120 + }, + { + "epoch": 0.18781506932759967, + "grad_norm": 1.6321041584014893, + "learning_rate": 1.8722416037322778e-05, + "loss": 0.17558739185333253, + "step": 7130 + }, + { + "epoch": 0.1880784845720984, + "grad_norm": 1.6950098276138306, + "learning_rate": 1.871824038049131e-05, + "loss": 0.18087623119354249, + "step": 7140 + }, + { + "epoch": 0.18834189981659713, + "grad_norm": 1.9401136636734009, + "learning_rate": 1.871405837814835e-05, + "loss": 0.16209206581115723, + "step": 7150 + }, + { + "epoch": 0.18860531506109587, + "grad_norm": 2.157315492630005, + "learning_rate": 1.8709870033337753e-05, + "loss": 0.18236569166183472, + "step": 7160 + }, + { + "epoch": 0.18886873030559462, + "grad_norm": 4.938285827636719, + "learning_rate": 1.8705675349107964e-05, + "loss": 0.1716249942779541, + "step": 7170 + }, + { + "epoch": 0.18913214555009336, + "grad_norm": 3.4710168838500977, + "learning_rate": 1.870147432851206e-05, + "loss": 0.17087876796722412, + "step": 7180 + }, + { + "epoch": 0.18939556079459208, + "grad_norm": 3.788525342941284, + "learning_rate": 1.8697266974607732e-05, + "loss": 0.17649215459823608, + "step": 7190 + }, + { + "epoch": 0.18965897603909082, + "grad_norm": 2.3726043701171875, + "learning_rate": 1.8693053290457264e-05, + "loss": 0.1622039794921875, + "step": 7200 + }, + { + "epoch": 0.18992239128358956, + "grad_norm": 3.2989308834075928, + "learning_rate": 1.868883327912756e-05, + "loss": 0.17068650722503662, + "step": 7210 + }, + { + "epoch": 0.1901858065280883, + "grad_norm": 1.267765998840332, + "learning_rate": 1.868460694369013e-05, + "loss": 0.17668700218200684, + "step": 7220 + }, + { + "epoch": 0.19044922177258702, + "grad_norm": 2.0194320678710938, + "learning_rate": 1.868037428722108e-05, + "loss": 0.1699920654296875, + "step": 7230 + }, + { + "epoch": 0.19071263701708577, + "grad_norm": 2.5008106231689453, + "learning_rate": 1.867613531280112e-05, + "loss": 0.16978907585144043, + "step": 7240 + }, + { + "epoch": 0.1909760522615845, + "grad_norm": 2.1206414699554443, + "learning_rate": 1.867189002351556e-05, + "loss": 0.1701862096786499, + "step": 7250 + }, + { + "epoch": 0.19123946750608325, + "grad_norm": 4.403330326080322, + "learning_rate": 1.8667638422454306e-05, + "loss": 0.16980268955230712, + "step": 7260 + }, + { + "epoch": 0.19150288275058197, + "grad_norm": 2.9617466926574707, + "learning_rate": 1.8663380512711852e-05, + "loss": 0.16657600402832032, + "step": 7270 + }, + { + "epoch": 0.19176629799508071, + "grad_norm": 1.6474148035049438, + "learning_rate": 1.865911629738729e-05, + "loss": 0.16871960163116456, + "step": 7280 + }, + { + "epoch": 0.19202971323957946, + "grad_norm": 1.9593570232391357, + "learning_rate": 1.86548457795843e-05, + "loss": 0.17759356498718262, + "step": 7290 + }, + { + "epoch": 0.1922931284840782, + "grad_norm": 1.8544610738754272, + "learning_rate": 1.8650568962411154e-05, + "loss": 0.16894125938415527, + "step": 7300 + }, + { + "epoch": 0.19255654372857695, + "grad_norm": 2.9292774200439453, + "learning_rate": 1.8646285848980693e-05, + "loss": 0.16140581369400026, + "step": 7310 + }, + { + "epoch": 0.19281995897307566, + "grad_norm": 6.881701946258545, + "learning_rate": 1.8641996442410365e-05, + "loss": 0.14970903396606444, + "step": 7320 + }, + { + "epoch": 0.1930833742175744, + "grad_norm": 0.7535760402679443, + "learning_rate": 1.8637700745822176e-05, + "loss": 0.17968838214874266, + "step": 7330 + }, + { + "epoch": 0.19334678946207315, + "grad_norm": 0.5784302949905396, + "learning_rate": 1.8633398762342722e-05, + "loss": 0.17113159894943236, + "step": 7340 + }, + { + "epoch": 0.1936102047065719, + "grad_norm": 3.432145357131958, + "learning_rate": 1.8629090495103175e-05, + "loss": 0.16940593719482422, + "step": 7350 + }, + { + "epoch": 0.1938736199510706, + "grad_norm": 2.7787270545959473, + "learning_rate": 1.8624775947239272e-05, + "loss": 0.15147242546081544, + "step": 7360 + }, + { + "epoch": 0.19413703519556935, + "grad_norm": 3.536893606185913, + "learning_rate": 1.862045512189133e-05, + "loss": 0.15813528299331664, + "step": 7370 + }, + { + "epoch": 0.1944004504400681, + "grad_norm": 2.813382863998413, + "learning_rate": 1.8616128022204233e-05, + "loss": 0.17074451446533204, + "step": 7380 + }, + { + "epoch": 0.19466386568456684, + "grad_norm": 1.823862075805664, + "learning_rate": 1.8611794651327432e-05, + "loss": 0.17314213514328003, + "step": 7390 + }, + { + "epoch": 0.19492728092906556, + "grad_norm": 1.2219538688659668, + "learning_rate": 1.8607455012414936e-05, + "loss": 0.16074463129043579, + "step": 7400 + }, + { + "epoch": 0.1951906961735643, + "grad_norm": 1.3485084772109985, + "learning_rate": 1.8603109108625327e-05, + "loss": 0.16705360412597656, + "step": 7410 + }, + { + "epoch": 0.19545411141806304, + "grad_norm": 2.6942336559295654, + "learning_rate": 1.859875694312174e-05, + "loss": 0.15758218765258789, + "step": 7420 + }, + { + "epoch": 0.19571752666256179, + "grad_norm": 1.054936170578003, + "learning_rate": 1.8594398519071867e-05, + "loss": 0.17418150901794432, + "step": 7430 + }, + { + "epoch": 0.19598094190706053, + "grad_norm": 1.2174276113510132, + "learning_rate": 1.8590033839647955e-05, + "loss": 0.1775670051574707, + "step": 7440 + }, + { + "epoch": 0.19624435715155925, + "grad_norm": 0.8863564133644104, + "learning_rate": 1.8585662908026814e-05, + "loss": 0.1678775429725647, + "step": 7450 + }, + { + "epoch": 0.196507772396058, + "grad_norm": 1.2479194402694702, + "learning_rate": 1.8581285727389787e-05, + "loss": 0.1716856598854065, + "step": 7460 + }, + { + "epoch": 0.19677118764055673, + "grad_norm": 5.347126483917236, + "learning_rate": 1.857690230092278e-05, + "loss": 0.17964506149291992, + "step": 7470 + }, + { + "epoch": 0.19703460288505548, + "grad_norm": 2.3655402660369873, + "learning_rate": 1.8572512631816243e-05, + "loss": 0.1659895658493042, + "step": 7480 + }, + { + "epoch": 0.1972980181295542, + "grad_norm": 1.4043081998825073, + "learning_rate": 1.856811672326516e-05, + "loss": 0.15814836025238038, + "step": 7490 + }, + { + "epoch": 0.19756143337405294, + "grad_norm": 2.003373384475708, + "learning_rate": 1.8563714578469064e-05, + "loss": 0.1712430477142334, + "step": 7500 + }, + { + "epoch": 0.19782484861855168, + "grad_norm": 4.684846878051758, + "learning_rate": 1.8559306200632036e-05, + "loss": 0.17329723834991456, + "step": 7510 + }, + { + "epoch": 0.19808826386305042, + "grad_norm": 1.708345890045166, + "learning_rate": 1.8554891592962666e-05, + "loss": 0.15741639137268065, + "step": 7520 + }, + { + "epoch": 0.19835167910754914, + "grad_norm": 2.4214718341827393, + "learning_rate": 1.8550470758674112e-05, + "loss": 0.164920175075531, + "step": 7530 + }, + { + "epoch": 0.19861509435204788, + "grad_norm": 1.3073595762252808, + "learning_rate": 1.8546043700984036e-05, + "loss": 0.1709401488304138, + "step": 7540 + }, + { + "epoch": 0.19887850959654663, + "grad_norm": 2.415018320083618, + "learning_rate": 1.8541610423114647e-05, + "loss": 0.17312939167022706, + "step": 7550 + }, + { + "epoch": 0.19914192484104537, + "grad_norm": 2.4456396102905273, + "learning_rate": 1.853717092829268e-05, + "loss": 0.173726487159729, + "step": 7560 + }, + { + "epoch": 0.1994053400855441, + "grad_norm": 1.81907320022583, + "learning_rate": 1.853272521974938e-05, + "loss": 0.17627353668212892, + "step": 7570 + }, + { + "epoch": 0.19966875533004283, + "grad_norm": 1.260183334350586, + "learning_rate": 1.8528273300720538e-05, + "loss": 0.16582975387573243, + "step": 7580 + }, + { + "epoch": 0.19993217057454157, + "grad_norm": 1.4951329231262207, + "learning_rate": 1.8523815174446444e-05, + "loss": 0.16517119407653807, + "step": 7590 + }, + { + "epoch": 0.20019558581904032, + "grad_norm": 0.9507079720497131, + "learning_rate": 1.851935084417192e-05, + "loss": 0.16164355278015136, + "step": 7600 + }, + { + "epoch": 0.20045900106353906, + "grad_norm": 2.002885103225708, + "learning_rate": 1.8514880313146295e-05, + "loss": 0.17967453002929687, + "step": 7610 + }, + { + "epoch": 0.20072241630803778, + "grad_norm": 2.073059320449829, + "learning_rate": 1.8510403584623414e-05, + "loss": 0.17459397315979003, + "step": 7620 + }, + { + "epoch": 0.20098583155253652, + "grad_norm": 6.2530436515808105, + "learning_rate": 1.850592066186163e-05, + "loss": 0.16354663372039796, + "step": 7630 + }, + { + "epoch": 0.20124924679703526, + "grad_norm": 2.6243183612823486, + "learning_rate": 1.850143154812382e-05, + "loss": 0.15856837034225463, + "step": 7640 + }, + { + "epoch": 0.201512662041534, + "grad_norm": 1.8908188343048096, + "learning_rate": 1.849693624667734e-05, + "loss": 0.16302520036697388, + "step": 7650 + }, + { + "epoch": 0.20177607728603272, + "grad_norm": 1.4392446279525757, + "learning_rate": 1.8492434760794072e-05, + "loss": 0.17663487195968627, + "step": 7660 + }, + { + "epoch": 0.20203949253053147, + "grad_norm": 2.43588924407959, + "learning_rate": 1.848792709375039e-05, + "loss": 0.16520986557006836, + "step": 7670 + }, + { + "epoch": 0.2023029077750302, + "grad_norm": 1.8082846403121948, + "learning_rate": 1.848341324882717e-05, + "loss": 0.15692648887634278, + "step": 7680 + }, + { + "epoch": 0.20256632301952895, + "grad_norm": 0.828616201877594, + "learning_rate": 1.8478893229309784e-05, + "loss": 0.17404477596282958, + "step": 7690 + }, + { + "epoch": 0.2028297382640277, + "grad_norm": 1.2157355546951294, + "learning_rate": 1.847436703848809e-05, + "loss": 0.16804027557373047, + "step": 7700 + }, + { + "epoch": 0.20309315350852641, + "grad_norm": 3.058650493621826, + "learning_rate": 1.8469834679656452e-05, + "loss": 0.17467894554138183, + "step": 7710 + }, + { + "epoch": 0.20335656875302516, + "grad_norm": 1.91413414478302, + "learning_rate": 1.8465296156113715e-05, + "loss": 0.1716160297393799, + "step": 7720 + }, + { + "epoch": 0.2036199839975239, + "grad_norm": 2.487852096557617, + "learning_rate": 1.8460751471163212e-05, + "loss": 0.18112963438034058, + "step": 7730 + }, + { + "epoch": 0.20388339924202264, + "grad_norm": 2.5864555835723877, + "learning_rate": 1.845620062811276e-05, + "loss": 0.17997872829437256, + "step": 7740 + }, + { + "epoch": 0.20414681448652136, + "grad_norm": 2.985783815383911, + "learning_rate": 1.8451643630274664e-05, + "loss": 0.1660379409790039, + "step": 7750 + }, + { + "epoch": 0.2044102297310201, + "grad_norm": 1.523375391960144, + "learning_rate": 1.8447080480965694e-05, + "loss": 0.16855190992355346, + "step": 7760 + }, + { + "epoch": 0.20467364497551885, + "grad_norm": 2.8014121055603027, + "learning_rate": 1.844251118350711e-05, + "loss": 0.16131818294525146, + "step": 7770 + }, + { + "epoch": 0.2049370602200176, + "grad_norm": 2.0758681297302246, + "learning_rate": 1.8437935741224652e-05, + "loss": 0.1804158091545105, + "step": 7780 + }, + { + "epoch": 0.2052004754645163, + "grad_norm": 1.1975489854812622, + "learning_rate": 1.8433354157448513e-05, + "loss": 0.1744102120399475, + "step": 7790 + }, + { + "epoch": 0.20546389070901505, + "grad_norm": 4.444065093994141, + "learning_rate": 1.8428766435513373e-05, + "loss": 0.17064743041992186, + "step": 7800 + }, + { + "epoch": 0.2057273059535138, + "grad_norm": 5.801499843597412, + "learning_rate": 1.8424172578758374e-05, + "loss": 0.167653226852417, + "step": 7810 + }, + { + "epoch": 0.20599072119801254, + "grad_norm": 3.624890089035034, + "learning_rate": 1.841957259052712e-05, + "loss": 0.1615079879760742, + "step": 7820 + }, + { + "epoch": 0.20625413644251128, + "grad_norm": 1.5724962949752808, + "learning_rate": 1.8414966474167688e-05, + "loss": 0.1745395541191101, + "step": 7830 + }, + { + "epoch": 0.20651755168701, + "grad_norm": 5.580723285675049, + "learning_rate": 1.8410354233032604e-05, + "loss": 0.16142590045928956, + "step": 7840 + }, + { + "epoch": 0.20678096693150874, + "grad_norm": 1.6538622379302979, + "learning_rate": 1.840573587047885e-05, + "loss": 0.1686989188194275, + "step": 7850 + }, + { + "epoch": 0.20704438217600749, + "grad_norm": 1.7280665636062622, + "learning_rate": 1.8401111389867882e-05, + "loss": 0.1745375871658325, + "step": 7860 + }, + { + "epoch": 0.20730779742050623, + "grad_norm": 3.5799858570098877, + "learning_rate": 1.8396480794565583e-05, + "loss": 0.16591970920562743, + "step": 7870 + }, + { + "epoch": 0.20757121266500494, + "grad_norm": 2.9077281951904297, + "learning_rate": 1.839184408794231e-05, + "loss": 0.1735524296760559, + "step": 7880 + }, + { + "epoch": 0.2078346279095037, + "grad_norm": 1.7181355953216553, + "learning_rate": 1.8387201273372855e-05, + "loss": 0.1699329137802124, + "step": 7890 + }, + { + "epoch": 0.20809804315400243, + "grad_norm": 5.141706943511963, + "learning_rate": 1.838255235423646e-05, + "loss": 0.16158028841018676, + "step": 7900 + }, + { + "epoch": 0.20836145839850118, + "grad_norm": 1.576851487159729, + "learning_rate": 1.8377897333916808e-05, + "loss": 0.16760889291763306, + "step": 7910 + }, + { + "epoch": 0.2086248736429999, + "grad_norm": 6.123061656951904, + "learning_rate": 1.837323621580202e-05, + "loss": 0.1597365617752075, + "step": 7920 + }, + { + "epoch": 0.20888828888749864, + "grad_norm": 3.8493268489837646, + "learning_rate": 1.8368569003284662e-05, + "loss": 0.16725847721099854, + "step": 7930 + }, + { + "epoch": 0.20915170413199738, + "grad_norm": 2.1557397842407227, + "learning_rate": 1.8363895699761736e-05, + "loss": 0.17427263259887696, + "step": 7940 + }, + { + "epoch": 0.20941511937649612, + "grad_norm": 5.269211292266846, + "learning_rate": 1.8359216308634668e-05, + "loss": 0.16679720878601073, + "step": 7950 + }, + { + "epoch": 0.20967853462099487, + "grad_norm": 0.6166566610336304, + "learning_rate": 1.8354530833309323e-05, + "loss": 0.1716822862625122, + "step": 7960 + }, + { + "epoch": 0.20994194986549358, + "grad_norm": 3.5395615100860596, + "learning_rate": 1.8349839277195993e-05, + "loss": 0.15502511262893676, + "step": 7970 + }, + { + "epoch": 0.21020536510999233, + "grad_norm": 2.2356109619140625, + "learning_rate": 1.8345141643709394e-05, + "loss": 0.165336275100708, + "step": 7980 + }, + { + "epoch": 0.21046878035449107, + "grad_norm": 3.1790695190429688, + "learning_rate": 1.8340437936268667e-05, + "loss": 0.1709498167037964, + "step": 7990 + }, + { + "epoch": 0.2107321955989898, + "grad_norm": 4.481166362762451, + "learning_rate": 1.8335728158297374e-05, + "loss": 0.17622543573379518, + "step": 8000 + }, + { + "epoch": 0.21099561084348853, + "grad_norm": 0.9332616925239563, + "learning_rate": 1.833101231322349e-05, + "loss": 0.17516076564788818, + "step": 8010 + }, + { + "epoch": 0.21125902608798727, + "grad_norm": 3.243543863296509, + "learning_rate": 1.832629040447942e-05, + "loss": 0.18097747564315797, + "step": 8020 + }, + { + "epoch": 0.21152244133248602, + "grad_norm": 2.1820068359375, + "learning_rate": 1.8321562435501963e-05, + "loss": 0.18786931037902832, + "step": 8030 + }, + { + "epoch": 0.21178585657698476, + "grad_norm": 2.834824800491333, + "learning_rate": 1.831682840973234e-05, + "loss": 0.1605833053588867, + "step": 8040 + }, + { + "epoch": 0.21204927182148348, + "grad_norm": 2.302321672439575, + "learning_rate": 1.8312088330616188e-05, + "loss": 0.16982619762420653, + "step": 8050 + }, + { + "epoch": 0.21231268706598222, + "grad_norm": 1.5783329010009766, + "learning_rate": 1.8307342201603533e-05, + "loss": 0.16248438358306885, + "step": 8060 + }, + { + "epoch": 0.21257610231048096, + "grad_norm": 1.408844232559204, + "learning_rate": 1.830259002614881e-05, + "loss": 0.15146074295043946, + "step": 8070 + }, + { + "epoch": 0.2128395175549797, + "grad_norm": 1.3815752267837524, + "learning_rate": 1.829783180771087e-05, + "loss": 0.17852706909179689, + "step": 8080 + }, + { + "epoch": 0.21310293279947845, + "grad_norm": 2.0581421852111816, + "learning_rate": 1.8293067549752937e-05, + "loss": 0.17140390872955322, + "step": 8090 + }, + { + "epoch": 0.21336634804397717, + "grad_norm": 1.1516538858413696, + "learning_rate": 1.8288297255742646e-05, + "loss": 0.1916679859161377, + "step": 8100 + }, + { + "epoch": 0.2136297632884759, + "grad_norm": 3.0807809829711914, + "learning_rate": 1.828352092915203e-05, + "loss": 0.17085427045822144, + "step": 8110 + }, + { + "epoch": 0.21389317853297465, + "grad_norm": 1.0461969375610352, + "learning_rate": 1.8278738573457496e-05, + "loss": 0.16013662815093993, + "step": 8120 + }, + { + "epoch": 0.2141565937774734, + "grad_norm": 1.1890825033187866, + "learning_rate": 1.8273950192139854e-05, + "loss": 0.16641087532043458, + "step": 8130 + }, + { + "epoch": 0.2144200090219721, + "grad_norm": 3.4797708988189697, + "learning_rate": 1.826915578868429e-05, + "loss": 0.1805914044380188, + "step": 8140 + }, + { + "epoch": 0.21468342426647086, + "grad_norm": 3.52163028717041, + "learning_rate": 1.8264355366580387e-05, + "loss": 0.1646334409713745, + "step": 8150 + }, + { + "epoch": 0.2149468395109696, + "grad_norm": 1.1108171939849854, + "learning_rate": 1.825954892932209e-05, + "loss": 0.16286721229553222, + "step": 8160 + }, + { + "epoch": 0.21521025475546834, + "grad_norm": 1.9880704879760742, + "learning_rate": 1.8254736480407738e-05, + "loss": 0.15940263271331787, + "step": 8170 + }, + { + "epoch": 0.21547366999996706, + "grad_norm": 1.2636542320251465, + "learning_rate": 1.8249918023340036e-05, + "loss": 0.15693287849426268, + "step": 8180 + }, + { + "epoch": 0.2157370852444658, + "grad_norm": 4.015563488006592, + "learning_rate": 1.8245093561626063e-05, + "loss": 0.1755892276763916, + "step": 8190 + }, + { + "epoch": 0.21600050048896455, + "grad_norm": 4.204861164093018, + "learning_rate": 1.824026309877727e-05, + "loss": 0.1776719093322754, + "step": 8200 + }, + { + "epoch": 0.2162639157334633, + "grad_norm": 3.930189847946167, + "learning_rate": 1.8235426638309483e-05, + "loss": 0.17433196306228638, + "step": 8210 + }, + { + "epoch": 0.21652733097796203, + "grad_norm": 1.1858417987823486, + "learning_rate": 1.823058418374288e-05, + "loss": 0.1776374101638794, + "step": 8220 + }, + { + "epoch": 0.21679074622246075, + "grad_norm": 2.381281614303589, + "learning_rate": 1.822573573860201e-05, + "loss": 0.16156091690063476, + "step": 8230 + }, + { + "epoch": 0.2170541614669595, + "grad_norm": 0.8454897999763489, + "learning_rate": 1.8220881306415783e-05, + "loss": 0.161505126953125, + "step": 8240 + }, + { + "epoch": 0.21731757671145824, + "grad_norm": 1.657363772392273, + "learning_rate": 1.821602089071746e-05, + "loss": 0.160190212726593, + "step": 8250 + }, + { + "epoch": 0.21758099195595698, + "grad_norm": 1.5874501466751099, + "learning_rate": 1.8211154495044666e-05, + "loss": 0.1763451337814331, + "step": 8260 + }, + { + "epoch": 0.2178444072004557, + "grad_norm": 1.3686648607254028, + "learning_rate": 1.8206282122939368e-05, + "loss": 0.15405559539794922, + "step": 8270 + }, + { + "epoch": 0.21810782244495444, + "grad_norm": 2.455176591873169, + "learning_rate": 1.8201403777947896e-05, + "loss": 0.15980288982391358, + "step": 8280 + }, + { + "epoch": 0.21837123768945318, + "grad_norm": 1.0419600009918213, + "learning_rate": 1.8196519463620912e-05, + "loss": 0.17014437913894653, + "step": 8290 + }, + { + "epoch": 0.21863465293395193, + "grad_norm": 1.7923341989517212, + "learning_rate": 1.8191629183513435e-05, + "loss": 0.16139771938323974, + "step": 8300 + }, + { + "epoch": 0.21889806817845064, + "grad_norm": 1.2861281633377075, + "learning_rate": 1.8186732941184824e-05, + "loss": 0.16867351531982422, + "step": 8310 + }, + { + "epoch": 0.2191614834229494, + "grad_norm": 1.2197952270507812, + "learning_rate": 1.818183074019877e-05, + "loss": 0.16637442111968995, + "step": 8320 + }, + { + "epoch": 0.21942489866744813, + "grad_norm": 2.092991352081299, + "learning_rate": 1.8176922584123308e-05, + "loss": 0.16264939308166504, + "step": 8330 + }, + { + "epoch": 0.21968831391194688, + "grad_norm": 3.2119240760803223, + "learning_rate": 1.817200847653081e-05, + "loss": 0.1815420866012573, + "step": 8340 + }, + { + "epoch": 0.21995172915644562, + "grad_norm": 1.6750664710998535, + "learning_rate": 1.8167088420997975e-05, + "loss": 0.17943034172058106, + "step": 8350 + }, + { + "epoch": 0.22021514440094433, + "grad_norm": 2.3476693630218506, + "learning_rate": 1.816216242110582e-05, + "loss": 0.17644221782684327, + "step": 8360 + }, + { + "epoch": 0.22047855964544308, + "grad_norm": 1.0248236656188965, + "learning_rate": 1.815723048043972e-05, + "loss": 0.15980939865112304, + "step": 8370 + }, + { + "epoch": 0.22074197488994182, + "grad_norm": 1.5431236028671265, + "learning_rate": 1.8152292602589336e-05, + "loss": 0.16916911602020263, + "step": 8380 + }, + { + "epoch": 0.22100539013444057, + "grad_norm": 0.959420382976532, + "learning_rate": 1.8147348791148678e-05, + "loss": 0.17721104621887207, + "step": 8390 + }, + { + "epoch": 0.22126880537893928, + "grad_norm": 0.9468563199043274, + "learning_rate": 1.8142399049716065e-05, + "loss": 0.15881028175354003, + "step": 8400 + }, + { + "epoch": 0.22153222062343803, + "grad_norm": 1.661454200744629, + "learning_rate": 1.8137443381894126e-05, + "loss": 0.17958208322525024, + "step": 8410 + }, + { + "epoch": 0.22179563586793677, + "grad_norm": 0.9187719225883484, + "learning_rate": 1.813248179128982e-05, + "loss": 0.15948023796081542, + "step": 8420 + }, + { + "epoch": 0.2220590511124355, + "grad_norm": 1.1451939344406128, + "learning_rate": 1.81275142815144e-05, + "loss": 0.1784202814102173, + "step": 8430 + }, + { + "epoch": 0.22232246635693423, + "grad_norm": 3.987659215927124, + "learning_rate": 1.812254085618343e-05, + "loss": 0.18151220083236694, + "step": 8440 + }, + { + "epoch": 0.22258588160143297, + "grad_norm": 3.994075059890747, + "learning_rate": 1.8117561518916792e-05, + "loss": 0.1678159713745117, + "step": 8450 + }, + { + "epoch": 0.22284929684593172, + "grad_norm": 2.8410162925720215, + "learning_rate": 1.811257627333866e-05, + "loss": 0.16332905292510985, + "step": 8460 + }, + { + "epoch": 0.22311271209043046, + "grad_norm": 1.047262191772461, + "learning_rate": 1.810758512307751e-05, + "loss": 0.16453289985656738, + "step": 8470 + }, + { + "epoch": 0.2233761273349292, + "grad_norm": 4.944979190826416, + "learning_rate": 1.8102588071766116e-05, + "loss": 0.17180402278900148, + "step": 8480 + }, + { + "epoch": 0.22363954257942792, + "grad_norm": 2.468308448791504, + "learning_rate": 1.8097585123041547e-05, + "loss": 0.16345694065093994, + "step": 8490 + }, + { + "epoch": 0.22390295782392666, + "grad_norm": 2.9842779636383057, + "learning_rate": 1.809257628054517e-05, + "loss": 0.16652379035949708, + "step": 8500 + }, + { + "epoch": 0.2241663730684254, + "grad_norm": 1.1200640201568604, + "learning_rate": 1.808756154792263e-05, + "loss": 0.16154274940490723, + "step": 8510 + }, + { + "epoch": 0.22442978831292415, + "grad_norm": 2.885934591293335, + "learning_rate": 1.8082540928823866e-05, + "loss": 0.16689971685409546, + "step": 8520 + }, + { + "epoch": 0.22469320355742287, + "grad_norm": 1.0953487157821655, + "learning_rate": 1.8077514426903107e-05, + "loss": 0.17047452926635742, + "step": 8530 + }, + { + "epoch": 0.2249566188019216, + "grad_norm": 2.7019665241241455, + "learning_rate": 1.8072482045818852e-05, + "loss": 0.15445919036865235, + "step": 8540 + }, + { + "epoch": 0.22522003404642035, + "grad_norm": 2.7076876163482666, + "learning_rate": 1.806744378923389e-05, + "loss": 0.16946548223495483, + "step": 8550 + }, + { + "epoch": 0.2254834492909191, + "grad_norm": 2.103975296020508, + "learning_rate": 1.8062399660815277e-05, + "loss": 0.17313816547393798, + "step": 8560 + }, + { + "epoch": 0.2257468645354178, + "grad_norm": 1.6260203123092651, + "learning_rate": 1.8057349664234348e-05, + "loss": 0.1576596975326538, + "step": 8570 + }, + { + "epoch": 0.22601027977991656, + "grad_norm": 2.122241735458374, + "learning_rate": 1.8052293803166714e-05, + "loss": 0.1758165717124939, + "step": 8580 + }, + { + "epoch": 0.2262736950244153, + "grad_norm": 1.8157440423965454, + "learning_rate": 1.8047232081292242e-05, + "loss": 0.16199681758880616, + "step": 8590 + }, + { + "epoch": 0.22653711026891404, + "grad_norm": 1.9818861484527588, + "learning_rate": 1.8042164502295074e-05, + "loss": 0.18239927291870117, + "step": 8600 + }, + { + "epoch": 0.2268005255134128, + "grad_norm": 0.5456166863441467, + "learning_rate": 1.8037091069863615e-05, + "loss": 0.16270630359649657, + "step": 8610 + }, + { + "epoch": 0.2270639407579115, + "grad_norm": 2.034343957901001, + "learning_rate": 1.8032011787690523e-05, + "loss": 0.16258046627044678, + "step": 8620 + }, + { + "epoch": 0.22732735600241025, + "grad_norm": 3.8354651927948, + "learning_rate": 1.8026926659472727e-05, + "loss": 0.1612878680229187, + "step": 8630 + }, + { + "epoch": 0.227590771246909, + "grad_norm": 1.308855414390564, + "learning_rate": 1.802183568891139e-05, + "loss": 0.16899081468582153, + "step": 8640 + }, + { + "epoch": 0.22785418649140773, + "grad_norm": 2.491147756576538, + "learning_rate": 1.801673887971195e-05, + "loss": 0.17585294246673583, + "step": 8650 + }, + { + "epoch": 0.22811760173590645, + "grad_norm": 1.3311322927474976, + "learning_rate": 1.801163623558408e-05, + "loss": 0.17406296730041504, + "step": 8660 + }, + { + "epoch": 0.2283810169804052, + "grad_norm": 1.1154834032058716, + "learning_rate": 1.800652776024171e-05, + "loss": 0.16373727321624756, + "step": 8670 + }, + { + "epoch": 0.22864443222490394, + "grad_norm": 1.8622162342071533, + "learning_rate": 1.8001413457403005e-05, + "loss": 0.1663828134536743, + "step": 8680 + }, + { + "epoch": 0.22890784746940268, + "grad_norm": 2.0145375728607178, + "learning_rate": 1.7996293330790374e-05, + "loss": 0.1708606958389282, + "step": 8690 + }, + { + "epoch": 0.2291712627139014, + "grad_norm": 1.1102200746536255, + "learning_rate": 1.7991167384130466e-05, + "loss": 0.16543641090393066, + "step": 8700 + }, + { + "epoch": 0.22943467795840014, + "grad_norm": 1.4417059421539307, + "learning_rate": 1.798603562115417e-05, + "loss": 0.16601539850234986, + "step": 8710 + }, + { + "epoch": 0.22969809320289888, + "grad_norm": 1.6073282957077026, + "learning_rate": 1.79808980455966e-05, + "loss": 0.1634589433670044, + "step": 8720 + }, + { + "epoch": 0.22996150844739763, + "grad_norm": 1.1768248081207275, + "learning_rate": 1.7975754661197106e-05, + "loss": 0.16483421325683595, + "step": 8730 + }, + { + "epoch": 0.23022492369189637, + "grad_norm": 1.9127576351165771, + "learning_rate": 1.797060547169927e-05, + "loss": 0.16938071250915526, + "step": 8740 + }, + { + "epoch": 0.2304883389363951, + "grad_norm": 1.8922983407974243, + "learning_rate": 1.796545048085089e-05, + "loss": 0.1673765540122986, + "step": 8750 + }, + { + "epoch": 0.23075175418089383, + "grad_norm": 1.2700562477111816, + "learning_rate": 1.796028969240399e-05, + "loss": 0.17090284824371338, + "step": 8760 + }, + { + "epoch": 0.23101516942539257, + "grad_norm": 1.1772215366363525, + "learning_rate": 1.7955123110114815e-05, + "loss": 0.171226966381073, + "step": 8770 + }, + { + "epoch": 0.23127858466989132, + "grad_norm": 1.2644784450531006, + "learning_rate": 1.794995073774383e-05, + "loss": 0.15555286407470703, + "step": 8780 + }, + { + "epoch": 0.23154199991439003, + "grad_norm": 0.8993234634399414, + "learning_rate": 1.7944772579055705e-05, + "loss": 0.17595932483673096, + "step": 8790 + }, + { + "epoch": 0.23180541515888878, + "grad_norm": 2.7910408973693848, + "learning_rate": 1.793958863781933e-05, + "loss": 0.16504385471343994, + "step": 8800 + }, + { + "epoch": 0.23206883040338752, + "grad_norm": 2.5594749450683594, + "learning_rate": 1.7934398917807802e-05, + "loss": 0.19950308799743652, + "step": 8810 + }, + { + "epoch": 0.23233224564788627, + "grad_norm": 1.5980995893478394, + "learning_rate": 1.7929203422798426e-05, + "loss": 0.16588497161865234, + "step": 8820 + }, + { + "epoch": 0.23259566089238498, + "grad_norm": 1.5760592222213745, + "learning_rate": 1.79240021565727e-05, + "loss": 0.16078193187713624, + "step": 8830 + }, + { + "epoch": 0.23285907613688372, + "grad_norm": 5.905186653137207, + "learning_rate": 1.7918795122916337e-05, + "loss": 0.1665295124053955, + "step": 8840 + }, + { + "epoch": 0.23312249138138247, + "grad_norm": 0.7906802892684937, + "learning_rate": 1.7913582325619236e-05, + "loss": 0.16812641620635987, + "step": 8850 + }, + { + "epoch": 0.2333859066258812, + "grad_norm": 4.158473014831543, + "learning_rate": 1.7908363768475498e-05, + "loss": 0.1745143413543701, + "step": 8860 + }, + { + "epoch": 0.23364932187037996, + "grad_norm": 2.5124568939208984, + "learning_rate": 1.7903139455283412e-05, + "loss": 0.16939520835876465, + "step": 8870 + }, + { + "epoch": 0.23391273711487867, + "grad_norm": 5.189577579498291, + "learning_rate": 1.7897909389845462e-05, + "loss": 0.17069075107574463, + "step": 8880 + }, + { + "epoch": 0.23417615235937742, + "grad_norm": 5.837613105773926, + "learning_rate": 1.789267357596831e-05, + "loss": 0.1610799551010132, + "step": 8890 + }, + { + "epoch": 0.23443956760387616, + "grad_norm": 1.536623239517212, + "learning_rate": 1.7887432017462812e-05, + "loss": 0.17057119607925414, + "step": 8900 + }, + { + "epoch": 0.2347029828483749, + "grad_norm": 4.994365215301514, + "learning_rate": 1.7882184718144002e-05, + "loss": 0.1818924903869629, + "step": 8910 + }, + { + "epoch": 0.23496639809287362, + "grad_norm": 3.165142059326172, + "learning_rate": 1.7876931681831087e-05, + "loss": 0.16904531717300414, + "step": 8920 + }, + { + "epoch": 0.23522981333737236, + "grad_norm": 1.5859851837158203, + "learning_rate": 1.787167291234745e-05, + "loss": 0.1652522325515747, + "step": 8930 + }, + { + "epoch": 0.2354932285818711, + "grad_norm": 5.673972129821777, + "learning_rate": 1.7866408413520662e-05, + "loss": 0.15671926736831665, + "step": 8940 + }, + { + "epoch": 0.23575664382636985, + "grad_norm": 2.017590284347534, + "learning_rate": 1.786113818918244e-05, + "loss": 0.1809386730194092, + "step": 8950 + }, + { + "epoch": 0.23602005907086857, + "grad_norm": 2.9026882648468018, + "learning_rate": 1.7855862243168684e-05, + "loss": 0.16011519432067872, + "step": 8960 + }, + { + "epoch": 0.2362834743153673, + "grad_norm": 1.9918408393859863, + "learning_rate": 1.785058057931946e-05, + "loss": 0.17151141166687012, + "step": 8970 + }, + { + "epoch": 0.23654688955986605, + "grad_norm": 2.9654719829559326, + "learning_rate": 1.7845293201478985e-05, + "loss": 0.18467168807983397, + "step": 8980 + }, + { + "epoch": 0.2368103048043648, + "grad_norm": 1.2674046754837036, + "learning_rate": 1.7840000113495647e-05, + "loss": 0.1681634306907654, + "step": 8990 + }, + { + "epoch": 0.23707372004886354, + "grad_norm": 6.172347068786621, + "learning_rate": 1.7834701319221978e-05, + "loss": 0.1718947172164917, + "step": 9000 + }, + { + "epoch": 0.23733713529336226, + "grad_norm": 1.4084432125091553, + "learning_rate": 1.782939682251467e-05, + "loss": 0.17439498901367187, + "step": 9010 + }, + { + "epoch": 0.237600550537861, + "grad_norm": 2.726349353790283, + "learning_rate": 1.7824086627234565e-05, + "loss": 0.16478637456893921, + "step": 9020 + }, + { + "epoch": 0.23786396578235974, + "grad_norm": 4.304076671600342, + "learning_rate": 1.7818770737246653e-05, + "loss": 0.1735978603363037, + "step": 9030 + }, + { + "epoch": 0.2381273810268585, + "grad_norm": 1.3137340545654297, + "learning_rate": 1.7813449156420072e-05, + "loss": 0.14409762620925903, + "step": 9040 + }, + { + "epoch": 0.2383907962713572, + "grad_norm": 2.127565860748291, + "learning_rate": 1.7808121888628093e-05, + "loss": 0.1733271837234497, + "step": 9050 + }, + { + "epoch": 0.23865421151585595, + "grad_norm": 1.3313630819320679, + "learning_rate": 1.780278893774813e-05, + "loss": 0.1641359806060791, + "step": 9060 + }, + { + "epoch": 0.2389176267603547, + "grad_norm": 1.9735864400863647, + "learning_rate": 1.7797450307661733e-05, + "loss": 0.1629354476928711, + "step": 9070 + }, + { + "epoch": 0.23918104200485343, + "grad_norm": 3.0851693153381348, + "learning_rate": 1.77921060022546e-05, + "loss": 0.17087717056274415, + "step": 9080 + }, + { + "epoch": 0.23944445724935215, + "grad_norm": 1.521122932434082, + "learning_rate": 1.7786756025416535e-05, + "loss": 0.1789768695831299, + "step": 9090 + }, + { + "epoch": 0.2397078724938509, + "grad_norm": 2.29960036277771, + "learning_rate": 1.7781400381041484e-05, + "loss": 0.17247596979141236, + "step": 9100 + }, + { + "epoch": 0.23997128773834964, + "grad_norm": 1.2741082906723022, + "learning_rate": 1.777603907302752e-05, + "loss": 0.16218769550323486, + "step": 9110 + }, + { + "epoch": 0.24023470298284838, + "grad_norm": 1.3085131645202637, + "learning_rate": 1.7770672105276833e-05, + "loss": 0.14968500137329102, + "step": 9120 + }, + { + "epoch": 0.24049811822734712, + "grad_norm": 2.388035535812378, + "learning_rate": 1.776529948169573e-05, + "loss": 0.17083903551101684, + "step": 9130 + }, + { + "epoch": 0.24076153347184584, + "grad_norm": 1.3554624319076538, + "learning_rate": 1.775992120619464e-05, + "loss": 0.1769646406173706, + "step": 9140 + }, + { + "epoch": 0.24102494871634458, + "grad_norm": 1.64016854763031, + "learning_rate": 1.7754537282688104e-05, + "loss": 0.1693008542060852, + "step": 9150 + }, + { + "epoch": 0.24128836396084333, + "grad_norm": 3.705508232116699, + "learning_rate": 1.7749147715094777e-05, + "loss": 0.15701783895492555, + "step": 9160 + }, + { + "epoch": 0.24155177920534207, + "grad_norm": 1.049346685409546, + "learning_rate": 1.7743752507337414e-05, + "loss": 0.17568821907043458, + "step": 9170 + }, + { + "epoch": 0.2418151944498408, + "grad_norm": 3.434546947479248, + "learning_rate": 1.7738351663342887e-05, + "loss": 0.17099368572235107, + "step": 9180 + }, + { + "epoch": 0.24207860969433953, + "grad_norm": 1.9312384128570557, + "learning_rate": 1.7732945187042155e-05, + "loss": 0.1609158158302307, + "step": 9190 + }, + { + "epoch": 0.24234202493883827, + "grad_norm": 1.5324152708053589, + "learning_rate": 1.7727533082370288e-05, + "loss": 0.1552303910255432, + "step": 9200 + }, + { + "epoch": 0.24260544018333702, + "grad_norm": 0.8496096730232239, + "learning_rate": 1.772211535326645e-05, + "loss": 0.1829839825630188, + "step": 9210 + }, + { + "epoch": 0.24286885542783573, + "grad_norm": 2.6993422508239746, + "learning_rate": 1.7716692003673898e-05, + "loss": 0.16176064014434816, + "step": 9220 + }, + { + "epoch": 0.24313227067233448, + "grad_norm": 2.3521485328674316, + "learning_rate": 1.7711263037539978e-05, + "loss": 0.17962357997894288, + "step": 9230 + }, + { + "epoch": 0.24339568591683322, + "grad_norm": 2.857625722885132, + "learning_rate": 1.770582845881613e-05, + "loss": 0.17189524173736573, + "step": 9240 + }, + { + "epoch": 0.24365910116133196, + "grad_norm": 1.152662992477417, + "learning_rate": 1.7700388271457872e-05, + "loss": 0.16647419929504395, + "step": 9250 + }, + { + "epoch": 0.2439225164058307, + "grad_norm": 0.8443275690078735, + "learning_rate": 1.7694942479424803e-05, + "loss": 0.16322238445281984, + "step": 9260 + }, + { + "epoch": 0.24418593165032942, + "grad_norm": 1.8076926469802856, + "learning_rate": 1.768949108668061e-05, + "loss": 0.16212935447692872, + "step": 9270 + }, + { + "epoch": 0.24444934689482817, + "grad_norm": 1.6279770135879517, + "learning_rate": 1.7684034097193055e-05, + "loss": 0.1555098533630371, + "step": 9280 + }, + { + "epoch": 0.2447127621393269, + "grad_norm": 5.117138862609863, + "learning_rate": 1.7678571514933966e-05, + "loss": 0.17470474243164064, + "step": 9290 + }, + { + "epoch": 0.24497617738382566, + "grad_norm": 4.41643762588501, + "learning_rate": 1.7673103343879242e-05, + "loss": 0.16322546005249022, + "step": 9300 + }, + { + "epoch": 0.24523959262832437, + "grad_norm": 2.0286550521850586, + "learning_rate": 1.7667629588008862e-05, + "loss": 0.1589446783065796, + "step": 9310 + }, + { + "epoch": 0.24550300787282311, + "grad_norm": 1.5142842531204224, + "learning_rate": 1.766215025130686e-05, + "loss": 0.15982253551483155, + "step": 9320 + }, + { + "epoch": 0.24576642311732186, + "grad_norm": 1.105963945388794, + "learning_rate": 1.7656665337761327e-05, + "loss": 0.16539645195007324, + "step": 9330 + }, + { + "epoch": 0.2460298383618206, + "grad_norm": 1.093591332435608, + "learning_rate": 1.7651174851364427e-05, + "loss": 0.16007717847824096, + "step": 9340 + }, + { + "epoch": 0.24629325360631932, + "grad_norm": 6.915426254272461, + "learning_rate": 1.764567879611237e-05, + "loss": 0.16880688667297364, + "step": 9350 + }, + { + "epoch": 0.24655666885081806, + "grad_norm": 3.2144203186035156, + "learning_rate": 1.7640177176005425e-05, + "loss": 0.15764827728271485, + "step": 9360 + }, + { + "epoch": 0.2468200840953168, + "grad_norm": 1.1754294633865356, + "learning_rate": 1.7634669995047904e-05, + "loss": 0.1819005012512207, + "step": 9370 + }, + { + "epoch": 0.24708349933981555, + "grad_norm": 1.9150971174240112, + "learning_rate": 1.7629157257248175e-05, + "loss": 0.16156461238861083, + "step": 9380 + }, + { + "epoch": 0.2473469145843143, + "grad_norm": 1.041055679321289, + "learning_rate": 1.762363896661864e-05, + "loss": 0.1629374861717224, + "step": 9390 + }, + { + "epoch": 0.247610329828813, + "grad_norm": 4.329281330108643, + "learning_rate": 1.761811512717576e-05, + "loss": 0.16649868488311767, + "step": 9400 + }, + { + "epoch": 0.24787374507331175, + "grad_norm": 0.9003462195396423, + "learning_rate": 1.761258574294001e-05, + "loss": 0.14600822925567628, + "step": 9410 + }, + { + "epoch": 0.2481371603178105, + "grad_norm": 2.314706802368164, + "learning_rate": 1.760705081793593e-05, + "loss": 0.1628469467163086, + "step": 9420 + }, + { + "epoch": 0.24840057556230924, + "grad_norm": 3.247070550918579, + "learning_rate": 1.760151035619206e-05, + "loss": 0.1757439613342285, + "step": 9430 + }, + { + "epoch": 0.24866399080680796, + "grad_norm": 1.0492268800735474, + "learning_rate": 1.7595964361741e-05, + "loss": 0.16031429767608643, + "step": 9440 + }, + { + "epoch": 0.2489274060513067, + "grad_norm": 4.544323444366455, + "learning_rate": 1.759041283861936e-05, + "loss": 0.17225009202957153, + "step": 9450 + }, + { + "epoch": 0.24919082129580544, + "grad_norm": 1.0400464534759521, + "learning_rate": 1.7584855790867773e-05, + "loss": 0.16639275550842286, + "step": 9460 + }, + { + "epoch": 0.2494542365403042, + "grad_norm": 1.4093406200408936, + "learning_rate": 1.7579293222530907e-05, + "loss": 0.16496472358703612, + "step": 9470 + }, + { + "epoch": 0.24971765178480293, + "grad_norm": 1.5801032781600952, + "learning_rate": 1.7573725137657432e-05, + "loss": 0.1664199113845825, + "step": 9480 + }, + { + "epoch": 0.24998106702930165, + "grad_norm": 3.252880334854126, + "learning_rate": 1.756815154030005e-05, + "loss": 0.16422860622406005, + "step": 9490 + }, + { + "epoch": 0.2502444822738004, + "grad_norm": 1.3494372367858887, + "learning_rate": 1.756257243451545e-05, + "loss": 0.1568368911743164, + "step": 9500 + }, + { + "epoch": 0.2505078975182991, + "grad_norm": 1.2839263677597046, + "learning_rate": 1.7556987824364363e-05, + "loss": 0.16363422870635985, + "step": 9510 + }, + { + "epoch": 0.25077131276279785, + "grad_norm": 2.9571659564971924, + "learning_rate": 1.75513977139115e-05, + "loss": 0.16724445819854736, + "step": 9520 + }, + { + "epoch": 0.2510347280072966, + "grad_norm": 5.429722785949707, + "learning_rate": 1.7545802107225585e-05, + "loss": 0.18005069494247436, + "step": 9530 + }, + { + "epoch": 0.25129814325179534, + "grad_norm": 0.9683855175971985, + "learning_rate": 1.754020100837934e-05, + "loss": 0.17200334072113038, + "step": 9540 + }, + { + "epoch": 0.2515615584962941, + "grad_norm": 1.3464741706848145, + "learning_rate": 1.753459442144949e-05, + "loss": 0.17369855642318727, + "step": 9550 + }, + { + "epoch": 0.2518249737407928, + "grad_norm": 4.674990177154541, + "learning_rate": 1.7528982350516754e-05, + "loss": 0.1638195753097534, + "step": 9560 + }, + { + "epoch": 0.25208838898529157, + "grad_norm": 2.2213902473449707, + "learning_rate": 1.7523364799665832e-05, + "loss": 0.15869691371917724, + "step": 9570 + }, + { + "epoch": 0.2523518042297903, + "grad_norm": 2.1054718494415283, + "learning_rate": 1.7517741772985424e-05, + "loss": 0.17035949230194092, + "step": 9580 + }, + { + "epoch": 0.252615219474289, + "grad_norm": 0.6201252341270447, + "learning_rate": 1.751211327456821e-05, + "loss": 0.17110464572906495, + "step": 9590 + }, + { + "epoch": 0.25287863471878774, + "grad_norm": 5.214323043823242, + "learning_rate": 1.7506479308510855e-05, + "loss": 0.16662635803222656, + "step": 9600 + }, + { + "epoch": 0.2531420499632865, + "grad_norm": 4.655108451843262, + "learning_rate": 1.7500839878914004e-05, + "loss": 0.16358193159103393, + "step": 9610 + }, + { + "epoch": 0.25340546520778523, + "grad_norm": 2.3508002758026123, + "learning_rate": 1.749519498988227e-05, + "loss": 0.178320050239563, + "step": 9620 + }, + { + "epoch": 0.253668880452284, + "grad_norm": 4.056017875671387, + "learning_rate": 1.748954464552426e-05, + "loss": 0.1705488920211792, + "step": 9630 + }, + { + "epoch": 0.2539322956967827, + "grad_norm": 36.46140670776367, + "learning_rate": 1.7483888849952525e-05, + "loss": 0.17811455726623535, + "step": 9640 + }, + { + "epoch": 0.25419571094128146, + "grad_norm": 1.896009922027588, + "learning_rate": 1.7478227607283604e-05, + "loss": 0.7065652847290039, + "step": 9650 + }, + { + "epoch": 0.2544591261857802, + "grad_norm": 2.5364885330200195, + "learning_rate": 1.7472560921637993e-05, + "loss": 0.17789881229400634, + "step": 9660 + }, + { + "epoch": 0.25472254143027895, + "grad_norm": 3.122502565383911, + "learning_rate": 1.746688879714015e-05, + "loss": 0.16588733196258545, + "step": 9670 + }, + { + "epoch": 0.25498595667477764, + "grad_norm": 2.2115306854248047, + "learning_rate": 1.746121123791849e-05, + "loss": 0.1728952169418335, + "step": 9680 + }, + { + "epoch": 0.2552493719192764, + "grad_norm": 4.740716934204102, + "learning_rate": 1.7455528248105385e-05, + "loss": 0.18714640140533448, + "step": 9690 + }, + { + "epoch": 0.2555127871637751, + "grad_norm": 1.1171883344650269, + "learning_rate": 1.7449839831837166e-05, + "loss": 0.17156922817230225, + "step": 9700 + }, + { + "epoch": 0.25577620240827387, + "grad_norm": 1.9390515089035034, + "learning_rate": 1.7444145993254105e-05, + "loss": 0.15038692951202393, + "step": 9710 + }, + { + "epoch": 0.2560396176527726, + "grad_norm": 1.1696727275848389, + "learning_rate": 1.7438446736500418e-05, + "loss": 0.17869341373443604, + "step": 9720 + }, + { + "epoch": 0.25630303289727135, + "grad_norm": 3.153860092163086, + "learning_rate": 1.7432742065724276e-05, + "loss": 0.1776731014251709, + "step": 9730 + }, + { + "epoch": 0.2565664481417701, + "grad_norm": 1.8851313591003418, + "learning_rate": 1.7427031985077785e-05, + "loss": 0.1718169093132019, + "step": 9740 + }, + { + "epoch": 0.25682986338626884, + "grad_norm": 2.572056293487549, + "learning_rate": 1.7421316498716986e-05, + "loss": 0.14862059354782103, + "step": 9750 + }, + { + "epoch": 0.2570932786307676, + "grad_norm": 4.337028980255127, + "learning_rate": 1.741559561080185e-05, + "loss": 0.15519614219665528, + "step": 9760 + }, + { + "epoch": 0.2573566938752663, + "grad_norm": 1.823828101158142, + "learning_rate": 1.74098693254963e-05, + "loss": 0.1460455060005188, + "step": 9770 + }, + { + "epoch": 0.257620109119765, + "grad_norm": 1.918368935585022, + "learning_rate": 1.7404137646968162e-05, + "loss": 0.16474132537841796, + "step": 9780 + }, + { + "epoch": 0.25788352436426376, + "grad_norm": 3.5668070316314697, + "learning_rate": 1.7398400579389204e-05, + "loss": 0.15777206420898438, + "step": 9790 + }, + { + "epoch": 0.2581469396087625, + "grad_norm": 3.7563867568969727, + "learning_rate": 1.739265812693511e-05, + "loss": 0.1705336332321167, + "step": 9800 + }, + { + "epoch": 0.25841035485326125, + "grad_norm": 1.9741572141647339, + "learning_rate": 1.7386910293785485e-05, + "loss": 0.16098759174346924, + "step": 9810 + }, + { + "epoch": 0.25867377009776, + "grad_norm": 3.406445264816284, + "learning_rate": 1.738115708412385e-05, + "loss": 0.15561914443969727, + "step": 9820 + }, + { + "epoch": 0.25893718534225874, + "grad_norm": 1.8321276903152466, + "learning_rate": 1.737539850213764e-05, + "loss": 0.17006561756134034, + "step": 9830 + }, + { + "epoch": 0.2592006005867575, + "grad_norm": 1.8626508712768555, + "learning_rate": 1.7369634552018196e-05, + "loss": 0.1636008381843567, + "step": 9840 + }, + { + "epoch": 0.25946401583125617, + "grad_norm": 2.8165507316589355, + "learning_rate": 1.7363865237960773e-05, + "loss": 0.1616446256637573, + "step": 9850 + }, + { + "epoch": 0.2597274310757549, + "grad_norm": 2.434441328048706, + "learning_rate": 1.7358090564164524e-05, + "loss": 0.16161463260650635, + "step": 9860 + }, + { + "epoch": 0.25999084632025365, + "grad_norm": 1.4848533868789673, + "learning_rate": 1.735231053483251e-05, + "loss": 0.1729748249053955, + "step": 9870 + }, + { + "epoch": 0.2602542615647524, + "grad_norm": 0.7964158058166504, + "learning_rate": 1.734652515417168e-05, + "loss": 0.17737159729003907, + "step": 9880 + }, + { + "epoch": 0.26051767680925114, + "grad_norm": 0.9846078157424927, + "learning_rate": 1.7340734426392894e-05, + "loss": 0.1651587128639221, + "step": 9890 + }, + { + "epoch": 0.2607810920537499, + "grad_norm": 2.473421573638916, + "learning_rate": 1.7334938355710876e-05, + "loss": 0.17304928302764894, + "step": 9900 + }, + { + "epoch": 0.26104450729824863, + "grad_norm": 2.0938100814819336, + "learning_rate": 1.7329136946344272e-05, + "loss": 0.1713578701019287, + "step": 9910 + }, + { + "epoch": 0.2613079225427474, + "grad_norm": 1.1600371599197388, + "learning_rate": 1.7323330202515592e-05, + "loss": 0.16025201082229615, + "step": 9920 + }, + { + "epoch": 0.2615713377872461, + "grad_norm": 3.907545804977417, + "learning_rate": 1.731751812845124e-05, + "loss": 0.17861415147781373, + "step": 9930 + }, + { + "epoch": 0.2618347530317448, + "grad_norm": 1.0781487226486206, + "learning_rate": 1.7311700728381486e-05, + "loss": 0.15549813508987426, + "step": 9940 + }, + { + "epoch": 0.26209816827624355, + "grad_norm": 2.9965620040893555, + "learning_rate": 1.7305878006540483e-05, + "loss": 0.16777470111846923, + "step": 9950 + }, + { + "epoch": 0.2623615835207423, + "grad_norm": 1.2340980768203735, + "learning_rate": 1.7300049967166273e-05, + "loss": 0.16619398593902587, + "step": 9960 + }, + { + "epoch": 0.26262499876524104, + "grad_norm": 1.463624119758606, + "learning_rate": 1.729421661450074e-05, + "loss": 0.15763309001922607, + "step": 9970 + }, + { + "epoch": 0.2628884140097398, + "grad_norm": 1.3264646530151367, + "learning_rate": 1.728837795278966e-05, + "loss": 0.16335705518722535, + "step": 9980 + }, + { + "epoch": 0.2631518292542385, + "grad_norm": 3.475846529006958, + "learning_rate": 1.7282533986282656e-05, + "loss": 0.17093327045440673, + "step": 9990 + }, + { + "epoch": 0.26341524449873727, + "grad_norm": 8.702486038208008, + "learning_rate": 1.7276684719233222e-05, + "loss": 0.17402191162109376, + "step": 10000 + }, + { + "epoch": 0.263678659743236, + "grad_norm": 2.967327833175659, + "learning_rate": 1.7270830155898704e-05, + "loss": 0.18402110338211058, + "step": 10010 + }, + { + "epoch": 0.26394207498773475, + "grad_norm": 1.225803256034851, + "learning_rate": 1.726497030054031e-05, + "loss": 0.16686680316925048, + "step": 10020 + }, + { + "epoch": 0.26420549023223344, + "grad_norm": 0.752476692199707, + "learning_rate": 1.72591051574231e-05, + "loss": 0.16316511631011962, + "step": 10030 + }, + { + "epoch": 0.2644689054767322, + "grad_norm": 1.983931541442871, + "learning_rate": 1.7253234730815964e-05, + "loss": 0.17867097854614258, + "step": 10040 + }, + { + "epoch": 0.26473232072123093, + "grad_norm": 1.2743942737579346, + "learning_rate": 1.724735902499166e-05, + "loss": 0.17276759147644044, + "step": 10050 + }, + { + "epoch": 0.2649957359657297, + "grad_norm": 0.9014014005661011, + "learning_rate": 1.7241478044226782e-05, + "loss": 0.16782515048980712, + "step": 10060 + }, + { + "epoch": 0.2652591512102284, + "grad_norm": 1.461192011833191, + "learning_rate": 1.723559179280176e-05, + "loss": 0.16669453382492067, + "step": 10070 + }, + { + "epoch": 0.26552256645472716, + "grad_norm": 2.261979341506958, + "learning_rate": 1.722970027500086e-05, + "loss": 0.1807500123977661, + "step": 10080 + }, + { + "epoch": 0.2657859816992259, + "grad_norm": 3.8714711666107178, + "learning_rate": 1.7223803495112185e-05, + "loss": 0.16808569431304932, + "step": 10090 + }, + { + "epoch": 0.26604939694372465, + "grad_norm": 0.9468282461166382, + "learning_rate": 1.7217901457427667e-05, + "loss": 0.15341663360595703, + "step": 10100 + }, + { + "epoch": 0.26631281218822334, + "grad_norm": 3.7509710788726807, + "learning_rate": 1.7211994166243058e-05, + "loss": 0.16726315021514893, + "step": 10110 + }, + { + "epoch": 0.2665762274327221, + "grad_norm": 1.7511935234069824, + "learning_rate": 1.7206081625857943e-05, + "loss": 0.17361044883728027, + "step": 10120 + }, + { + "epoch": 0.2668396426772208, + "grad_norm": 3.348145008087158, + "learning_rate": 1.7200163840575728e-05, + "loss": 0.17009975910186767, + "step": 10130 + }, + { + "epoch": 0.26710305792171957, + "grad_norm": 2.962674140930176, + "learning_rate": 1.7194240814703626e-05, + "loss": 0.16599173545837403, + "step": 10140 + }, + { + "epoch": 0.2673664731662183, + "grad_norm": 1.2494072914123535, + "learning_rate": 1.7188312552552672e-05, + "loss": 0.17172447443008423, + "step": 10150 + }, + { + "epoch": 0.26762988841071705, + "grad_norm": 2.3965790271759033, + "learning_rate": 1.718237905843771e-05, + "loss": 0.1677739977836609, + "step": 10160 + }, + { + "epoch": 0.2678933036552158, + "grad_norm": 3.26468563079834, + "learning_rate": 1.71764403366774e-05, + "loss": 0.17013306617736818, + "step": 10170 + }, + { + "epoch": 0.26815671889971454, + "grad_norm": 3.2682793140411377, + "learning_rate": 1.7170496391594194e-05, + "loss": 0.16103256940841676, + "step": 10180 + }, + { + "epoch": 0.2684201341442133, + "grad_norm": 1.7906304597854614, + "learning_rate": 1.716454722751435e-05, + "loss": 0.15973858833312987, + "step": 10190 + }, + { + "epoch": 0.268683549388712, + "grad_norm": 1.736289620399475, + "learning_rate": 1.715859284876793e-05, + "loss": 0.1677062392234802, + "step": 10200 + }, + { + "epoch": 0.2689469646332107, + "grad_norm": 0.47359856963157654, + "learning_rate": 1.7152633259688788e-05, + "loss": 0.16129859685897827, + "step": 10210 + }, + { + "epoch": 0.26921037987770946, + "grad_norm": 5.013718605041504, + "learning_rate": 1.714666846461457e-05, + "loss": 0.15962156057357788, + "step": 10220 + }, + { + "epoch": 0.2694737951222082, + "grad_norm": 1.5661720037460327, + "learning_rate": 1.7140698467886707e-05, + "loss": 0.1508167028427124, + "step": 10230 + }, + { + "epoch": 0.26973721036670695, + "grad_norm": 1.1465344429016113, + "learning_rate": 1.713472327385042e-05, + "loss": 0.16048423051834107, + "step": 10240 + }, + { + "epoch": 0.2700006256112057, + "grad_norm": 1.9026316404342651, + "learning_rate": 1.7128742886854723e-05, + "loss": 0.1541124939918518, + "step": 10250 + }, + { + "epoch": 0.27026404085570443, + "grad_norm": 2.43469500541687, + "learning_rate": 1.712275731125239e-05, + "loss": 0.17140823602676392, + "step": 10260 + }, + { + "epoch": 0.2705274561002032, + "grad_norm": 6.105805397033691, + "learning_rate": 1.7116766551399986e-05, + "loss": 0.1634799838066101, + "step": 10270 + }, + { + "epoch": 0.2707908713447019, + "grad_norm": 1.938686728477478, + "learning_rate": 1.711077061165784e-05, + "loss": 0.1655953884124756, + "step": 10280 + }, + { + "epoch": 0.2710542865892006, + "grad_norm": 4.508531093597412, + "learning_rate": 1.710476949639006e-05, + "loss": 0.1758341073989868, + "step": 10290 + }, + { + "epoch": 0.27131770183369935, + "grad_norm": 2.6813724040985107, + "learning_rate": 1.7098763209964514e-05, + "loss": 0.15596837997436525, + "step": 10300 + }, + { + "epoch": 0.2715811170781981, + "grad_norm": 1.799993872642517, + "learning_rate": 1.7092751756752838e-05, + "loss": 0.15480451583862304, + "step": 10310 + }, + { + "epoch": 0.27184453232269684, + "grad_norm": 1.7224605083465576, + "learning_rate": 1.7086735141130423e-05, + "loss": 0.1648601174354553, + "step": 10320 + }, + { + "epoch": 0.2721079475671956, + "grad_norm": 2.978431463241577, + "learning_rate": 1.7080713367476428e-05, + "loss": 0.15981528759002686, + "step": 10330 + }, + { + "epoch": 0.27237136281169433, + "grad_norm": 0.6369432806968689, + "learning_rate": 1.7074686440173752e-05, + "loss": 0.17378594875335693, + "step": 10340 + }, + { + "epoch": 0.2726347780561931, + "grad_norm": 2.043173313140869, + "learning_rate": 1.7068654363609058e-05, + "loss": 0.15432771444320678, + "step": 10350 + }, + { + "epoch": 0.2728981933006918, + "grad_norm": 2.752129316329956, + "learning_rate": 1.7062617142172754e-05, + "loss": 0.1774064540863037, + "step": 10360 + }, + { + "epoch": 0.2731616085451905, + "grad_norm": 2.3817272186279297, + "learning_rate": 1.7056574780258986e-05, + "loss": 0.1632688045501709, + "step": 10370 + }, + { + "epoch": 0.27342502378968925, + "grad_norm": 2.2778074741363525, + "learning_rate": 1.705052728226564e-05, + "loss": 0.15889192819595338, + "step": 10380 + }, + { + "epoch": 0.273688439034188, + "grad_norm": 1.4330559968948364, + "learning_rate": 1.704447465259436e-05, + "loss": 0.15565969944000244, + "step": 10390 + }, + { + "epoch": 0.27395185427868674, + "grad_norm": 3.334307909011841, + "learning_rate": 1.7038416895650504e-05, + "loss": 0.16733351945877076, + "step": 10400 + }, + { + "epoch": 0.2742152695231855, + "grad_norm": 1.9446099996566772, + "learning_rate": 1.7032354015843162e-05, + "loss": 0.16590291261672974, + "step": 10410 + }, + { + "epoch": 0.2744786847676842, + "grad_norm": 4.908642292022705, + "learning_rate": 1.7026286017585168e-05, + "loss": 0.15705243349075318, + "step": 10420 + }, + { + "epoch": 0.27474210001218297, + "grad_norm": 1.9492592811584473, + "learning_rate": 1.7020212905293072e-05, + "loss": 0.17008140087127685, + "step": 10430 + }, + { + "epoch": 0.2750055152566817, + "grad_norm": 1.793287754058838, + "learning_rate": 1.7014134683387147e-05, + "loss": 0.17017580270767213, + "step": 10440 + }, + { + "epoch": 0.27526893050118045, + "grad_norm": 1.4260156154632568, + "learning_rate": 1.7008051356291382e-05, + "loss": 0.15449953079223633, + "step": 10450 + }, + { + "epoch": 0.27553234574567914, + "grad_norm": 1.4016035795211792, + "learning_rate": 1.7001962928433488e-05, + "loss": 0.17537989616394042, + "step": 10460 + }, + { + "epoch": 0.2757957609901779, + "grad_norm": 2.730642318725586, + "learning_rate": 1.699586940424488e-05, + "loss": 0.17933493852615356, + "step": 10470 + }, + { + "epoch": 0.27605917623467663, + "grad_norm": 1.2780147790908813, + "learning_rate": 1.6989770788160697e-05, + "loss": 0.17381551265716552, + "step": 10480 + }, + { + "epoch": 0.2763225914791754, + "grad_norm": 4.233339786529541, + "learning_rate": 1.6983667084619768e-05, + "loss": 0.16962260007858276, + "step": 10490 + }, + { + "epoch": 0.2765860067236741, + "grad_norm": 2.5890400409698486, + "learning_rate": 1.6977558298064636e-05, + "loss": 0.16460258960723878, + "step": 10500 + }, + { + "epoch": 0.27684942196817286, + "grad_norm": 2.09346866607666, + "learning_rate": 1.6971444432941533e-05, + "loss": 0.15855820178985597, + "step": 10510 + }, + { + "epoch": 0.2771128372126716, + "grad_norm": 1.6417956352233887, + "learning_rate": 1.69653254937004e-05, + "loss": 0.16787092685699462, + "step": 10520 + }, + { + "epoch": 0.27737625245717035, + "grad_norm": 1.064825177192688, + "learning_rate": 1.6959201484794867e-05, + "loss": 0.16029760837554932, + "step": 10530 + }, + { + "epoch": 0.2776396677016691, + "grad_norm": 1.6606754064559937, + "learning_rate": 1.6953072410682243e-05, + "loss": 0.17648932933807374, + "step": 10540 + }, + { + "epoch": 0.2779030829461678, + "grad_norm": 4.053995609283447, + "learning_rate": 1.694693827582354e-05, + "loss": 0.17371286153793336, + "step": 10550 + }, + { + "epoch": 0.2781664981906665, + "grad_norm": 2.4995157718658447, + "learning_rate": 1.694079908468344e-05, + "loss": 0.15562156438827515, + "step": 10560 + }, + { + "epoch": 0.27842991343516527, + "grad_norm": 2.386960029602051, + "learning_rate": 1.6934654841730318e-05, + "loss": 0.16840798854827882, + "step": 10570 + }, + { + "epoch": 0.278693328679664, + "grad_norm": 2.0186102390289307, + "learning_rate": 1.6928505551436215e-05, + "loss": 0.16891305446624755, + "step": 10580 + }, + { + "epoch": 0.27895674392416275, + "grad_norm": 2.8074491024017334, + "learning_rate": 1.6922351218276852e-05, + "loss": 0.1625191330909729, + "step": 10590 + }, + { + "epoch": 0.2792201591686615, + "grad_norm": 3.491753339767456, + "learning_rate": 1.6916191846731615e-05, + "loss": 0.15475211143493653, + "step": 10600 + }, + { + "epoch": 0.27948357441316024, + "grad_norm": 3.0825541019439697, + "learning_rate": 1.6910027441283562e-05, + "loss": 0.17147622108459473, + "step": 10610 + }, + { + "epoch": 0.279746989657659, + "grad_norm": 2.4547269344329834, + "learning_rate": 1.6903858006419413e-05, + "loss": 0.18026995658874512, + "step": 10620 + }, + { + "epoch": 0.2800104049021577, + "grad_norm": 2.4258809089660645, + "learning_rate": 1.689768354662955e-05, + "loss": 0.16826764345169068, + "step": 10630 + }, + { + "epoch": 0.2802738201466564, + "grad_norm": 2.5427980422973633, + "learning_rate": 1.689150406640801e-05, + "loss": 0.16762207746505736, + "step": 10640 + }, + { + "epoch": 0.28053723539115516, + "grad_norm": 1.6073648929595947, + "learning_rate": 1.6885319570252485e-05, + "loss": 0.16751723289489745, + "step": 10650 + }, + { + "epoch": 0.2808006506356539, + "grad_norm": 3.7220256328582764, + "learning_rate": 1.687913006266432e-05, + "loss": 0.16273918151855468, + "step": 10660 + }, + { + "epoch": 0.28106406588015265, + "grad_norm": 2.928100109100342, + "learning_rate": 1.6872935548148503e-05, + "loss": 0.17194337844848634, + "step": 10670 + }, + { + "epoch": 0.2813274811246514, + "grad_norm": 3.047515630722046, + "learning_rate": 1.686673603121367e-05, + "loss": 0.16286033391952515, + "step": 10680 + }, + { + "epoch": 0.28159089636915013, + "grad_norm": 3.7232882976531982, + "learning_rate": 1.6860531516372102e-05, + "loss": 0.16922259330749512, + "step": 10690 + }, + { + "epoch": 0.2818543116136489, + "grad_norm": 1.5876383781433105, + "learning_rate": 1.6854322008139704e-05, + "loss": 0.16132194995880128, + "step": 10700 + }, + { + "epoch": 0.2821177268581476, + "grad_norm": 0.9653616547584534, + "learning_rate": 1.6848107511036026e-05, + "loss": 0.16511409282684325, + "step": 10710 + }, + { + "epoch": 0.2823811421026463, + "grad_norm": 2.3763375282287598, + "learning_rate": 1.6841888029584255e-05, + "loss": 0.14753872156143188, + "step": 10720 + }, + { + "epoch": 0.28264455734714505, + "grad_norm": 1.9268609285354614, + "learning_rate": 1.6835663568311187e-05, + "loss": 0.1503894805908203, + "step": 10730 + }, + { + "epoch": 0.2829079725916438, + "grad_norm": 1.9043006896972656, + "learning_rate": 1.682943413174726e-05, + "loss": 0.16788947582244873, + "step": 10740 + }, + { + "epoch": 0.28317138783614254, + "grad_norm": 1.7059781551361084, + "learning_rate": 1.6823199724426523e-05, + "loss": 0.17295582294464112, + "step": 10750 + }, + { + "epoch": 0.2834348030806413, + "grad_norm": 1.5484586954116821, + "learning_rate": 1.6816960350886653e-05, + "loss": 0.17311544418334962, + "step": 10760 + }, + { + "epoch": 0.28369821832514003, + "grad_norm": 3.034515619277954, + "learning_rate": 1.681071601566893e-05, + "loss": 0.16882672309875488, + "step": 10770 + }, + { + "epoch": 0.28396163356963877, + "grad_norm": 2.0294792652130127, + "learning_rate": 1.6804466723318248e-05, + "loss": 0.16841418743133546, + "step": 10780 + }, + { + "epoch": 0.2842250488141375, + "grad_norm": 3.959886312484741, + "learning_rate": 1.6798212478383118e-05, + "loss": 0.1645594835281372, + "step": 10790 + }, + { + "epoch": 0.28448846405863626, + "grad_norm": 3.074615478515625, + "learning_rate": 1.6791953285415645e-05, + "loss": 0.15847090482711793, + "step": 10800 + }, + { + "epoch": 0.28475187930313495, + "grad_norm": 2.1334962844848633, + "learning_rate": 1.678568914897154e-05, + "loss": 0.1653167724609375, + "step": 10810 + }, + { + "epoch": 0.2850152945476337, + "grad_norm": 5.066307067871094, + "learning_rate": 1.6779420073610116e-05, + "loss": 0.16236342191696168, + "step": 10820 + }, + { + "epoch": 0.28527870979213243, + "grad_norm": 6.05610466003418, + "learning_rate": 1.6773146063894273e-05, + "loss": 0.17876428365707397, + "step": 10830 + }, + { + "epoch": 0.2855421250366312, + "grad_norm": 4.845612049102783, + "learning_rate": 1.6766867124390505e-05, + "loss": 0.1621859550476074, + "step": 10840 + }, + { + "epoch": 0.2858055402811299, + "grad_norm": 3.89675235748291, + "learning_rate": 1.6760583259668897e-05, + "loss": 0.16310319900512696, + "step": 10850 + }, + { + "epoch": 0.28606895552562867, + "grad_norm": 1.7111464738845825, + "learning_rate": 1.6754294474303116e-05, + "loss": 0.1666949987411499, + "step": 10860 + }, + { + "epoch": 0.2863323707701274, + "grad_norm": 0.7771022319793701, + "learning_rate": 1.674800077287041e-05, + "loss": 0.17018003463745118, + "step": 10870 + }, + { + "epoch": 0.28659578601462615, + "grad_norm": 2.980215549468994, + "learning_rate": 1.6741702159951613e-05, + "loss": 0.17035326957702637, + "step": 10880 + }, + { + "epoch": 0.2868592012591249, + "grad_norm": 1.4614256620407104, + "learning_rate": 1.6735398640131117e-05, + "loss": 0.16818625926971437, + "step": 10890 + }, + { + "epoch": 0.2871226165036236, + "grad_norm": 1.1185781955718994, + "learning_rate": 1.6729090217996904e-05, + "loss": 0.17180737257003784, + "step": 10900 + }, + { + "epoch": 0.28738603174812233, + "grad_norm": 1.6062567234039307, + "learning_rate": 1.6722776898140513e-05, + "loss": 0.16105318069458008, + "step": 10910 + }, + { + "epoch": 0.28764944699262107, + "grad_norm": 6.148932933807373, + "learning_rate": 1.6716458685157047e-05, + "loss": 0.16838667392730713, + "step": 10920 + }, + { + "epoch": 0.2879128622371198, + "grad_norm": 2.257540464401245, + "learning_rate": 1.671013558364518e-05, + "loss": 0.16308460235595704, + "step": 10930 + }, + { + "epoch": 0.28817627748161856, + "grad_norm": 6.425755023956299, + "learning_rate": 1.6703807598207132e-05, + "loss": 0.16461313962936402, + "step": 10940 + }, + { + "epoch": 0.2884396927261173, + "grad_norm": 1.1854498386383057, + "learning_rate": 1.669747473344869e-05, + "loss": 0.15650880336761475, + "step": 10950 + }, + { + "epoch": 0.28870310797061605, + "grad_norm": 3.585944175720215, + "learning_rate": 1.6691136993979183e-05, + "loss": 0.1490423560142517, + "step": 10960 + }, + { + "epoch": 0.2889665232151148, + "grad_norm": 2.156951427459717, + "learning_rate": 1.6684794384411494e-05, + "loss": 0.15593162775039673, + "step": 10970 + }, + { + "epoch": 0.2892299384596135, + "grad_norm": 1.3406567573547363, + "learning_rate": 1.667844690936204e-05, + "loss": 0.16920124292373656, + "step": 10980 + }, + { + "epoch": 0.2894933537041122, + "grad_norm": 1.8499528169631958, + "learning_rate": 1.6672094573450798e-05, + "loss": 0.18240888118743898, + "step": 10990 + }, + { + "epoch": 0.28975676894861097, + "grad_norm": 1.945177435874939, + "learning_rate": 1.6665737381301263e-05, + "loss": 0.16866406202316284, + "step": 11000 + }, + { + "epoch": 0.2900201841931097, + "grad_norm": 1.1978505849838257, + "learning_rate": 1.665937533754048e-05, + "loss": 0.15739843845367432, + "step": 11010 + }, + { + "epoch": 0.29028359943760845, + "grad_norm": 1.9324181079864502, + "learning_rate": 1.6653008446799014e-05, + "loss": 0.16808900833129883, + "step": 11020 + }, + { + "epoch": 0.2905470146821072, + "grad_norm": 0.9644923210144043, + "learning_rate": 1.6646636713710967e-05, + "loss": 0.15285654067993165, + "step": 11030 + }, + { + "epoch": 0.29081042992660594, + "grad_norm": 2.9204294681549072, + "learning_rate": 1.664026014291396e-05, + "loss": 0.17728879451751708, + "step": 11040 + }, + { + "epoch": 0.2910738451711047, + "grad_norm": 3.86358380317688, + "learning_rate": 1.6633878739049135e-05, + "loss": 0.16395483016967774, + "step": 11050 + }, + { + "epoch": 0.2913372604156034, + "grad_norm": 2.285890817642212, + "learning_rate": 1.6627492506761155e-05, + "loss": 0.1678805470466614, + "step": 11060 + }, + { + "epoch": 0.2916006756601021, + "grad_norm": 1.6460524797439575, + "learning_rate": 1.6621101450698193e-05, + "loss": 0.17867043018341064, + "step": 11070 + }, + { + "epoch": 0.29186409090460086, + "grad_norm": 3.0653505325317383, + "learning_rate": 1.6614705575511937e-05, + "loss": 0.16762268543243408, + "step": 11080 + }, + { + "epoch": 0.2921275061490996, + "grad_norm": 2.8461790084838867, + "learning_rate": 1.660830488585758e-05, + "loss": 0.16133394241333007, + "step": 11090 + }, + { + "epoch": 0.29239092139359835, + "grad_norm": 2.3109350204467773, + "learning_rate": 1.660189938639382e-05, + "loss": 0.15947847366333007, + "step": 11100 + }, + { + "epoch": 0.2926543366380971, + "grad_norm": 0.9776355028152466, + "learning_rate": 1.6595489081782858e-05, + "loss": 0.1657570481300354, + "step": 11110 + }, + { + "epoch": 0.29291775188259583, + "grad_norm": 1.6997944116592407, + "learning_rate": 1.6589073976690387e-05, + "loss": 0.1711916446685791, + "step": 11120 + }, + { + "epoch": 0.2931811671270946, + "grad_norm": 3.0427398681640625, + "learning_rate": 1.65826540757856e-05, + "loss": 0.1619178056716919, + "step": 11130 + }, + { + "epoch": 0.2934445823715933, + "grad_norm": 1.7185020446777344, + "learning_rate": 1.6576229383741177e-05, + "loss": 0.16749472618103028, + "step": 11140 + }, + { + "epoch": 0.29370799761609206, + "grad_norm": 2.1597628593444824, + "learning_rate": 1.6569799905233285e-05, + "loss": 0.15980730056762696, + "step": 11150 + }, + { + "epoch": 0.29397141286059075, + "grad_norm": 1.63338303565979, + "learning_rate": 1.6563365644941578e-05, + "loss": 0.16434140205383302, + "step": 11160 + }, + { + "epoch": 0.2942348281050895, + "grad_norm": 3.3085994720458984, + "learning_rate": 1.6556926607549186e-05, + "loss": 0.15781755447387696, + "step": 11170 + }, + { + "epoch": 0.29449824334958824, + "grad_norm": 2.722555637359619, + "learning_rate": 1.655048279774272e-05, + "loss": 0.1720254898071289, + "step": 11180 + }, + { + "epoch": 0.294761658594087, + "grad_norm": 2.791558027267456, + "learning_rate": 1.6544034220212262e-05, + "loss": 0.1631596326828003, + "step": 11190 + }, + { + "epoch": 0.2950250738385857, + "grad_norm": 1.0485073328018188, + "learning_rate": 1.6537580879651364e-05, + "loss": 0.16433389186859132, + "step": 11200 + }, + { + "epoch": 0.29528848908308447, + "grad_norm": 1.3080395460128784, + "learning_rate": 1.6531122780757046e-05, + "loss": 0.1703435182571411, + "step": 11210 + }, + { + "epoch": 0.2955519043275832, + "grad_norm": 4.229607105255127, + "learning_rate": 1.6524659928229792e-05, + "loss": 0.17260711193084716, + "step": 11220 + }, + { + "epoch": 0.29581531957208196, + "grad_norm": 4.757648468017578, + "learning_rate": 1.6518192326773546e-05, + "loss": 0.17303617000579835, + "step": 11230 + }, + { + "epoch": 0.29607873481658065, + "grad_norm": 2.9214701652526855, + "learning_rate": 1.6511719981095703e-05, + "loss": 0.17856764793395996, + "step": 11240 + }, + { + "epoch": 0.2963421500610794, + "grad_norm": 3.771855592727661, + "learning_rate": 1.6505242895907122e-05, + "loss": 0.16803815364837646, + "step": 11250 + }, + { + "epoch": 0.29660556530557813, + "grad_norm": 2.7159695625305176, + "learning_rate": 1.64987610759221e-05, + "loss": 0.16644749641418458, + "step": 11260 + }, + { + "epoch": 0.2968689805500769, + "grad_norm": 1.4975543022155762, + "learning_rate": 1.6492274525858386e-05, + "loss": 0.16396504640579224, + "step": 11270 + }, + { + "epoch": 0.2971323957945756, + "grad_norm": 1.7197980880737305, + "learning_rate": 1.648578325043717e-05, + "loss": 0.16342889070510863, + "step": 11280 + }, + { + "epoch": 0.29739581103907436, + "grad_norm": 0.7188948392868042, + "learning_rate": 1.6479287254383084e-05, + "loss": 0.17266550064086914, + "step": 11290 + }, + { + "epoch": 0.2976592262835731, + "grad_norm": 2.1265695095062256, + "learning_rate": 1.6472786542424196e-05, + "loss": 0.1490931749343872, + "step": 11300 + }, + { + "epoch": 0.29792264152807185, + "grad_norm": 2.434117317199707, + "learning_rate": 1.6466281119292004e-05, + "loss": 0.1716543197631836, + "step": 11310 + }, + { + "epoch": 0.2981860567725706, + "grad_norm": 1.999320387840271, + "learning_rate": 1.6459770989721433e-05, + "loss": 0.16990287303924562, + "step": 11320 + }, + { + "epoch": 0.2984494720170693, + "grad_norm": 3.125255584716797, + "learning_rate": 1.6453256158450835e-05, + "loss": 0.1631378412246704, + "step": 11330 + }, + { + "epoch": 0.298712887261568, + "grad_norm": 1.368018388748169, + "learning_rate": 1.644673663022199e-05, + "loss": 0.170165491104126, + "step": 11340 + }, + { + "epoch": 0.29897630250606677, + "grad_norm": 2.717869520187378, + "learning_rate": 1.6440212409780088e-05, + "loss": 0.1671891212463379, + "step": 11350 + }, + { + "epoch": 0.2992397177505655, + "grad_norm": 1.194761037826538, + "learning_rate": 1.643368350187374e-05, + "loss": 0.1767759680747986, + "step": 11360 + }, + { + "epoch": 0.29950313299506426, + "grad_norm": 2.625335693359375, + "learning_rate": 1.6427149911254973e-05, + "loss": 0.15918991565704346, + "step": 11370 + }, + { + "epoch": 0.299766548239563, + "grad_norm": 1.8642637729644775, + "learning_rate": 1.64206116426792e-05, + "loss": 0.15451186895370483, + "step": 11380 + }, + { + "epoch": 0.30002996348406175, + "grad_norm": 1.5237486362457275, + "learning_rate": 1.641406870090527e-05, + "loss": 0.15578148365020753, + "step": 11390 + }, + { + "epoch": 0.3002933787285605, + "grad_norm": 1.3648377656936646, + "learning_rate": 1.640752109069541e-05, + "loss": 0.16496151685714722, + "step": 11400 + }, + { + "epoch": 0.30055679397305923, + "grad_norm": 0.7198021411895752, + "learning_rate": 1.640096881681526e-05, + "loss": 0.16492170095443726, + "step": 11410 + }, + { + "epoch": 0.3008202092175579, + "grad_norm": 2.127000570297241, + "learning_rate": 1.6394411884033844e-05, + "loss": 0.17192739248275757, + "step": 11420 + }, + { + "epoch": 0.30108362446205666, + "grad_norm": 1.0075953006744385, + "learning_rate": 1.638785029712358e-05, + "loss": 0.16300554275512696, + "step": 11430 + }, + { + "epoch": 0.3013470397065554, + "grad_norm": 2.8779633045196533, + "learning_rate": 1.638128406086027e-05, + "loss": 0.17133902311325072, + "step": 11440 + }, + { + "epoch": 0.30161045495105415, + "grad_norm": 1.0915799140930176, + "learning_rate": 1.6374713180023114e-05, + "loss": 0.15675240755081177, + "step": 11450 + }, + { + "epoch": 0.3018738701955529, + "grad_norm": 1.4045281410217285, + "learning_rate": 1.6368137659394673e-05, + "loss": 0.16971807479858397, + "step": 11460 + }, + { + "epoch": 0.30213728544005164, + "grad_norm": 4.1956682205200195, + "learning_rate": 1.63615575037609e-05, + "loss": 0.1889129638671875, + "step": 11470 + }, + { + "epoch": 0.3024007006845504, + "grad_norm": 5.0809006690979, + "learning_rate": 1.6354972717911115e-05, + "loss": 0.1501060962677002, + "step": 11480 + }, + { + "epoch": 0.3026641159290491, + "grad_norm": 4.371751308441162, + "learning_rate": 1.634838330663801e-05, + "loss": 0.17013825178146363, + "step": 11490 + }, + { + "epoch": 0.3029275311735478, + "grad_norm": 2.2258849143981934, + "learning_rate": 1.6341789274737644e-05, + "loss": 0.16531975269317628, + "step": 11500 + }, + { + "epoch": 0.30319094641804656, + "grad_norm": 1.9783573150634766, + "learning_rate": 1.6335190627009432e-05, + "loss": 0.17594610452651976, + "step": 11510 + }, + { + "epoch": 0.3034543616625453, + "grad_norm": 5.304379940032959, + "learning_rate": 1.632858736825616e-05, + "loss": 0.16333131790161132, + "step": 11520 + }, + { + "epoch": 0.30371777690704405, + "grad_norm": 1.9075709581375122, + "learning_rate": 1.632197950328396e-05, + "loss": 0.15945775508880616, + "step": 11530 + }, + { + "epoch": 0.3039811921515428, + "grad_norm": 3.2281694412231445, + "learning_rate": 1.631536703690233e-05, + "loss": 0.162673020362854, + "step": 11540 + }, + { + "epoch": 0.30424460739604153, + "grad_norm": 1.9339172840118408, + "learning_rate": 1.6308749973924096e-05, + "loss": 0.16387953758239746, + "step": 11550 + }, + { + "epoch": 0.3045080226405403, + "grad_norm": 1.674332618713379, + "learning_rate": 1.6302128319165454e-05, + "loss": 0.17316954135894774, + "step": 11560 + }, + { + "epoch": 0.304771437885039, + "grad_norm": 1.8994656801223755, + "learning_rate": 1.629550207744592e-05, + "loss": 0.1728541612625122, + "step": 11570 + }, + { + "epoch": 0.30503485312953776, + "grad_norm": 1.0488888025283813, + "learning_rate": 1.6288871253588366e-05, + "loss": 0.16703497171401976, + "step": 11580 + }, + { + "epoch": 0.30529826837403645, + "grad_norm": 4.005836486816406, + "learning_rate": 1.6282235852418983e-05, + "loss": 0.1798107385635376, + "step": 11590 + }, + { + "epoch": 0.3055616836185352, + "grad_norm": 1.9339059591293335, + "learning_rate": 1.6275595878767315e-05, + "loss": 0.16658738851547242, + "step": 11600 + }, + { + "epoch": 0.30582509886303394, + "grad_norm": 1.6809768676757812, + "learning_rate": 1.6268951337466212e-05, + "loss": 0.16311601400375367, + "step": 11610 + }, + { + "epoch": 0.3060885141075327, + "grad_norm": 1.1647820472717285, + "learning_rate": 1.626230223335186e-05, + "loss": 0.1774993658065796, + "step": 11620 + }, + { + "epoch": 0.3063519293520314, + "grad_norm": 1.360097885131836, + "learning_rate": 1.6255648571263762e-05, + "loss": 0.1719681978225708, + "step": 11630 + }, + { + "epoch": 0.30661534459653017, + "grad_norm": 3.530965566635132, + "learning_rate": 1.6248990356044742e-05, + "loss": 0.1749415159225464, + "step": 11640 + }, + { + "epoch": 0.3068787598410289, + "grad_norm": 2.485177993774414, + "learning_rate": 1.6242327592540942e-05, + "loss": 0.15750596523284913, + "step": 11650 + }, + { + "epoch": 0.30714217508552766, + "grad_norm": 1.4429829120635986, + "learning_rate": 1.62356602856018e-05, + "loss": 0.16543300151824952, + "step": 11660 + }, + { + "epoch": 0.3074055903300264, + "grad_norm": 1.0211246013641357, + "learning_rate": 1.6228988440080068e-05, + "loss": 0.16060178279876708, + "step": 11670 + }, + { + "epoch": 0.3076690055745251, + "grad_norm": 3.0444605350494385, + "learning_rate": 1.6222312060831814e-05, + "loss": 0.1530345559120178, + "step": 11680 + }, + { + "epoch": 0.30793242081902383, + "grad_norm": 2.8523387908935547, + "learning_rate": 1.6215631152716387e-05, + "loss": 0.17052619457244872, + "step": 11690 + }, + { + "epoch": 0.3081958360635226, + "grad_norm": 2.02268648147583, + "learning_rate": 1.620894572059644e-05, + "loss": 0.16898614168167114, + "step": 11700 + }, + { + "epoch": 0.3084592513080213, + "grad_norm": 1.786167025566101, + "learning_rate": 1.6202255769337924e-05, + "loss": 0.15555659532546998, + "step": 11710 + }, + { + "epoch": 0.30872266655252006, + "grad_norm": 1.22487473487854, + "learning_rate": 1.619556130381007e-05, + "loss": 0.16601090431213378, + "step": 11720 + }, + { + "epoch": 0.3089860817970188, + "grad_norm": 2.440392017364502, + "learning_rate": 1.61888623288854e-05, + "loss": 0.17595984935760497, + "step": 11730 + }, + { + "epoch": 0.30924949704151755, + "grad_norm": 2.724210739135742, + "learning_rate": 1.6182158849439718e-05, + "loss": 0.1554914116859436, + "step": 11740 + }, + { + "epoch": 0.3095129122860163, + "grad_norm": 2.659330368041992, + "learning_rate": 1.61754508703521e-05, + "loss": 0.16678513288497926, + "step": 11750 + }, + { + "epoch": 0.309776327530515, + "grad_norm": 2.0172038078308105, + "learning_rate": 1.6168738396504912e-05, + "loss": 0.17562490701675415, + "step": 11760 + }, + { + "epoch": 0.3100397427750137, + "grad_norm": 3.8281707763671875, + "learning_rate": 1.6162021432783773e-05, + "loss": 0.17031266689300537, + "step": 11770 + }, + { + "epoch": 0.31030315801951247, + "grad_norm": 2.769963026046753, + "learning_rate": 1.6155299984077584e-05, + "loss": 0.1619594693183899, + "step": 11780 + }, + { + "epoch": 0.3105665732640112, + "grad_norm": 0.8592987060546875, + "learning_rate": 1.6148574055278507e-05, + "loss": 0.14064273834228516, + "step": 11790 + }, + { + "epoch": 0.31082998850850996, + "grad_norm": 1.416334867477417, + "learning_rate": 1.6141843651281955e-05, + "loss": 0.1600215196609497, + "step": 11800 + }, + { + "epoch": 0.3110934037530087, + "grad_norm": 2.1182026863098145, + "learning_rate": 1.6135108776986615e-05, + "loss": 0.17031853199005126, + "step": 11810 + }, + { + "epoch": 0.31135681899750745, + "grad_norm": 0.9389249086380005, + "learning_rate": 1.612836943729442e-05, + "loss": 0.1463878870010376, + "step": 11820 + }, + { + "epoch": 0.3116202342420062, + "grad_norm": 1.7438561916351318, + "learning_rate": 1.6121625637110546e-05, + "loss": 0.17514048814773558, + "step": 11830 + }, + { + "epoch": 0.31188364948650493, + "grad_norm": 1.8737670183181763, + "learning_rate": 1.6114877381343422e-05, + "loss": 0.16191134452819825, + "step": 11840 + }, + { + "epoch": 0.3121470647310036, + "grad_norm": 1.6040972471237183, + "learning_rate": 1.610812467490472e-05, + "loss": 0.17908995151519774, + "step": 11850 + }, + { + "epoch": 0.31241047997550236, + "grad_norm": 2.5281789302825928, + "learning_rate": 1.6101367522709356e-05, + "loss": 0.1572499990463257, + "step": 11860 + }, + { + "epoch": 0.3126738952200011, + "grad_norm": 2.344128370285034, + "learning_rate": 1.6094605929675476e-05, + "loss": 0.15562047958374023, + "step": 11870 + }, + { + "epoch": 0.31293731046449985, + "grad_norm": 2.052171468734741, + "learning_rate": 1.608783990072445e-05, + "loss": 0.16465867757797242, + "step": 11880 + }, + { + "epoch": 0.3132007257089986, + "grad_norm": 3.6695945262908936, + "learning_rate": 1.6081069440780895e-05, + "loss": 0.16564571857452393, + "step": 11890 + }, + { + "epoch": 0.31346414095349734, + "grad_norm": 2.645644426345825, + "learning_rate": 1.607429455477264e-05, + "loss": 0.15045626163482667, + "step": 11900 + }, + { + "epoch": 0.3137275561979961, + "grad_norm": 1.562927484512329, + "learning_rate": 1.606751524763075e-05, + "loss": 0.152282977104187, + "step": 11910 + }, + { + "epoch": 0.3139909714424948, + "grad_norm": 1.0183966159820557, + "learning_rate": 1.606073152428948e-05, + "loss": 0.15882717370986937, + "step": 11920 + }, + { + "epoch": 0.31425438668699357, + "grad_norm": 2.47676944732666, + "learning_rate": 1.6053943389686323e-05, + "loss": 0.15989025831222534, + "step": 11930 + }, + { + "epoch": 0.31451780193149226, + "grad_norm": 2.306567430496216, + "learning_rate": 1.604715084876198e-05, + "loss": 0.16103131771087648, + "step": 11940 + }, + { + "epoch": 0.314781217175991, + "grad_norm": 0.6556203365325928, + "learning_rate": 1.6040353906460353e-05, + "loss": 0.1704404830932617, + "step": 11950 + }, + { + "epoch": 0.31504463242048975, + "grad_norm": 1.7158825397491455, + "learning_rate": 1.6033552567728553e-05, + "loss": 0.15168464183807373, + "step": 11960 + }, + { + "epoch": 0.3153080476649885, + "grad_norm": 2.291569471359253, + "learning_rate": 1.602674683751688e-05, + "loss": 0.17278625965118408, + "step": 11970 + }, + { + "epoch": 0.31557146290948723, + "grad_norm": 1.9693660736083984, + "learning_rate": 1.6019936720778845e-05, + "loss": 0.166748309135437, + "step": 11980 + }, + { + "epoch": 0.315834878153986, + "grad_norm": 2.916064500808716, + "learning_rate": 1.601312222247114e-05, + "loss": 0.16318831443786622, + "step": 11990 + }, + { + "epoch": 0.3160982933984847, + "grad_norm": 1.3091145753860474, + "learning_rate": 1.6006303347553655e-05, + "loss": 0.17221720218658448, + "step": 12000 + }, + { + "epoch": 0.31636170864298346, + "grad_norm": 1.1554275751113892, + "learning_rate": 1.5999480100989458e-05, + "loss": 0.1671406149864197, + "step": 12010 + }, + { + "epoch": 0.31662512388748215, + "grad_norm": 2.4813544750213623, + "learning_rate": 1.5992652487744806e-05, + "loss": 0.1669934034347534, + "step": 12020 + }, + { + "epoch": 0.3168885391319809, + "grad_norm": 1.0720349550247192, + "learning_rate": 1.5985820512789128e-05, + "loss": 0.15382648706436158, + "step": 12030 + }, + { + "epoch": 0.31715195437647964, + "grad_norm": 2.153705358505249, + "learning_rate": 1.5978984181095034e-05, + "loss": 0.17027074098587036, + "step": 12040 + }, + { + "epoch": 0.3174153696209784, + "grad_norm": 1.257405161857605, + "learning_rate": 1.5972143497638294e-05, + "loss": 0.16688921451568603, + "step": 12050 + }, + { + "epoch": 0.3176787848654771, + "grad_norm": 1.9108648300170898, + "learning_rate": 1.5965298467397855e-05, + "loss": 0.15784056186676027, + "step": 12060 + }, + { + "epoch": 0.31794220010997587, + "grad_norm": 1.135555386543274, + "learning_rate": 1.595844909535583e-05, + "loss": 0.1606656312942505, + "step": 12070 + }, + { + "epoch": 0.3182056153544746, + "grad_norm": 1.3785191774368286, + "learning_rate": 1.595159538649748e-05, + "loss": 0.1637197494506836, + "step": 12080 + }, + { + "epoch": 0.31846903059897336, + "grad_norm": 3.840480089187622, + "learning_rate": 1.5944737345811238e-05, + "loss": 0.1654859185218811, + "step": 12090 + }, + { + "epoch": 0.3187324458434721, + "grad_norm": 0.8346825838088989, + "learning_rate": 1.5937874978288676e-05, + "loss": 0.15820553302764892, + "step": 12100 + }, + { + "epoch": 0.3189958610879708, + "grad_norm": 1.5829136371612549, + "learning_rate": 1.5931008288924523e-05, + "loss": 0.15952603816986083, + "step": 12110 + }, + { + "epoch": 0.31925927633246953, + "grad_norm": 1.2521353960037231, + "learning_rate": 1.5924137282716653e-05, + "loss": 0.15865278244018555, + "step": 12120 + }, + { + "epoch": 0.3195226915769683, + "grad_norm": 1.90872061252594, + "learning_rate": 1.591726196466608e-05, + "loss": 0.17575249671936036, + "step": 12130 + }, + { + "epoch": 0.319786106821467, + "grad_norm": 0.6674976348876953, + "learning_rate": 1.5910382339776953e-05, + "loss": 0.1575980305671692, + "step": 12140 + }, + { + "epoch": 0.32004952206596576, + "grad_norm": 3.209744930267334, + "learning_rate": 1.5903498413056568e-05, + "loss": 0.16615101099014282, + "step": 12150 + }, + { + "epoch": 0.3203129373104645, + "grad_norm": 1.8076069355010986, + "learning_rate": 1.5896610189515335e-05, + "loss": 0.1611255407333374, + "step": 12160 + }, + { + "epoch": 0.32057635255496325, + "grad_norm": 1.727213978767395, + "learning_rate": 1.588971767416681e-05, + "loss": 0.16150546073913574, + "step": 12170 + }, + { + "epoch": 0.320839767799462, + "grad_norm": 5.160546779632568, + "learning_rate": 1.5882820872027654e-05, + "loss": 0.16852226257324218, + "step": 12180 + }, + { + "epoch": 0.32110318304396074, + "grad_norm": 0.8550304770469666, + "learning_rate": 1.5875919788117662e-05, + "loss": 0.16070616245269775, + "step": 12190 + }, + { + "epoch": 0.3213665982884594, + "grad_norm": 1.172480821609497, + "learning_rate": 1.586901442745974e-05, + "loss": 0.16748132705688476, + "step": 12200 + }, + { + "epoch": 0.32163001353295817, + "grad_norm": 2.032978057861328, + "learning_rate": 1.5862104795079903e-05, + "loss": 0.15195621252059938, + "step": 12210 + }, + { + "epoch": 0.3218934287774569, + "grad_norm": 1.312203288078308, + "learning_rate": 1.5855190896007287e-05, + "loss": 0.18289544582366943, + "step": 12220 + }, + { + "epoch": 0.32215684402195566, + "grad_norm": 0.8713375926017761, + "learning_rate": 1.584827273527412e-05, + "loss": 0.14678397178649902, + "step": 12230 + }, + { + "epoch": 0.3224202592664544, + "grad_norm": 4.486700534820557, + "learning_rate": 1.5841350317915745e-05, + "loss": 0.16222200393676758, + "step": 12240 + }, + { + "epoch": 0.32268367451095314, + "grad_norm": 1.5093228816986084, + "learning_rate": 1.5834423648970588e-05, + "loss": 0.1745450258255005, + "step": 12250 + }, + { + "epoch": 0.3229470897554519, + "grad_norm": 2.8938376903533936, + "learning_rate": 1.5827492733480183e-05, + "loss": 0.1714477062225342, + "step": 12260 + }, + { + "epoch": 0.32321050499995063, + "grad_norm": 1.8304592370986938, + "learning_rate": 1.5820557576489145e-05, + "loss": 0.1494492769241333, + "step": 12270 + }, + { + "epoch": 0.3234739202444493, + "grad_norm": 1.2277522087097168, + "learning_rate": 1.581361818304518e-05, + "loss": 0.1596914768218994, + "step": 12280 + }, + { + "epoch": 0.32373733548894806, + "grad_norm": 1.3385345935821533, + "learning_rate": 1.5806674558199088e-05, + "loss": 0.1729437828063965, + "step": 12290 + }, + { + "epoch": 0.3240007507334468, + "grad_norm": 0.8806704878807068, + "learning_rate": 1.5799726707004728e-05, + "loss": 0.17392619848251342, + "step": 12300 + }, + { + "epoch": 0.32426416597794555, + "grad_norm": 0.9253001809120178, + "learning_rate": 1.5792774634519046e-05, + "loss": 0.1747748374938965, + "step": 12310 + }, + { + "epoch": 0.3245275812224443, + "grad_norm": 2.724848985671997, + "learning_rate": 1.5785818345802068e-05, + "loss": 0.16253660917282103, + "step": 12320 + }, + { + "epoch": 0.32479099646694304, + "grad_norm": 0.9329707622528076, + "learning_rate": 1.5778857845916874e-05, + "loss": 0.16540203094482422, + "step": 12330 + }, + { + "epoch": 0.3250544117114418, + "grad_norm": 1.5239917039871216, + "learning_rate": 1.577189313992962e-05, + "loss": 0.1595819592475891, + "step": 12340 + }, + { + "epoch": 0.3253178269559405, + "grad_norm": 0.7669548392295837, + "learning_rate": 1.5764924232909518e-05, + "loss": 0.17054165601730348, + "step": 12350 + }, + { + "epoch": 0.32558124220043927, + "grad_norm": 0.9223381876945496, + "learning_rate": 1.5757951129928837e-05, + "loss": 0.16015908718109131, + "step": 12360 + }, + { + "epoch": 0.32584465744493796, + "grad_norm": 0.772723913192749, + "learning_rate": 1.5750973836062907e-05, + "loss": 0.15618748664855958, + "step": 12370 + }, + { + "epoch": 0.3261080726894367, + "grad_norm": 2.877375364303589, + "learning_rate": 1.57439923563901e-05, + "loss": 0.1711950421333313, + "step": 12380 + }, + { + "epoch": 0.32637148793393544, + "grad_norm": 1.585689902305603, + "learning_rate": 1.573700669599184e-05, + "loss": 0.154817795753479, + "step": 12390 + }, + { + "epoch": 0.3266349031784342, + "grad_norm": 1.7530324459075928, + "learning_rate": 1.573001685995259e-05, + "loss": 0.1512624979019165, + "step": 12400 + }, + { + "epoch": 0.32689831842293293, + "grad_norm": 4.1038408279418945, + "learning_rate": 1.572302285335985e-05, + "loss": 0.1759546399116516, + "step": 12410 + }, + { + "epoch": 0.3271617336674317, + "grad_norm": 1.5943409204483032, + "learning_rate": 1.5716024681304177e-05, + "loss": 0.15971808433532714, + "step": 12420 + }, + { + "epoch": 0.3274251489119304, + "grad_norm": 2.9541358947753906, + "learning_rate": 1.5709022348879124e-05, + "loss": 0.1637795686721802, + "step": 12430 + }, + { + "epoch": 0.32768856415642916, + "grad_norm": 1.5326303243637085, + "learning_rate": 1.5702015861181295e-05, + "loss": 0.159002685546875, + "step": 12440 + }, + { + "epoch": 0.3279519794009279, + "grad_norm": 4.3393988609313965, + "learning_rate": 1.569500522331032e-05, + "loss": 0.1640324592590332, + "step": 12450 + }, + { + "epoch": 0.3282153946454266, + "grad_norm": 5.520726203918457, + "learning_rate": 1.568799044036884e-05, + "loss": 0.16121139526367187, + "step": 12460 + }, + { + "epoch": 0.32847880988992534, + "grad_norm": 0.946117639541626, + "learning_rate": 1.5680971517462516e-05, + "loss": 0.1471323251724243, + "step": 12470 + }, + { + "epoch": 0.3287422251344241, + "grad_norm": 4.419226169586182, + "learning_rate": 1.5673948459700024e-05, + "loss": 0.14972081184387206, + "step": 12480 + }, + { + "epoch": 0.3290056403789228, + "grad_norm": 2.0303003787994385, + "learning_rate": 1.5666921272193046e-05, + "loss": 0.1642866849899292, + "step": 12490 + }, + { + "epoch": 0.32926905562342157, + "grad_norm": 2.8176403045654297, + "learning_rate": 1.565988996005628e-05, + "loss": 0.170470666885376, + "step": 12500 + }, + { + "epoch": 0.3295324708679203, + "grad_norm": 1.9970959424972534, + "learning_rate": 1.5652854528407407e-05, + "loss": 0.16080517768859864, + "step": 12510 + }, + { + "epoch": 0.32979588611241906, + "grad_norm": 1.555129885673523, + "learning_rate": 1.5645814982367125e-05, + "loss": 0.1623956322669983, + "step": 12520 + }, + { + "epoch": 0.3300593013569178, + "grad_norm": 0.8437623977661133, + "learning_rate": 1.5638771327059117e-05, + "loss": 0.16722278594970702, + "step": 12530 + }, + { + "epoch": 0.3303227166014165, + "grad_norm": 2.2068533897399902, + "learning_rate": 1.563172356761006e-05, + "loss": 0.15940253734588622, + "step": 12540 + }, + { + "epoch": 0.33058613184591523, + "grad_norm": 0.6867446303367615, + "learning_rate": 1.5624671709149618e-05, + "loss": 0.1537413239479065, + "step": 12550 + }, + { + "epoch": 0.330849547090414, + "grad_norm": 2.044262409210205, + "learning_rate": 1.561761575681044e-05, + "loss": 0.15259177684783937, + "step": 12560 + }, + { + "epoch": 0.3311129623349127, + "grad_norm": 1.8643264770507812, + "learning_rate": 1.5610555715728146e-05, + "loss": 0.15341873168945314, + "step": 12570 + }, + { + "epoch": 0.33137637757941146, + "grad_norm": 2.2367396354675293, + "learning_rate": 1.560349159104134e-05, + "loss": 0.16090472936630248, + "step": 12580 + }, + { + "epoch": 0.3316397928239102, + "grad_norm": 2.6326286792755127, + "learning_rate": 1.55964233878916e-05, + "loss": 0.16225993633270264, + "step": 12590 + }, + { + "epoch": 0.33190320806840895, + "grad_norm": 0.6998537182807922, + "learning_rate": 1.5589351111423474e-05, + "loss": 0.1577354907989502, + "step": 12600 + }, + { + "epoch": 0.3321666233129077, + "grad_norm": 1.9655646085739136, + "learning_rate": 1.5582274766784457e-05, + "loss": 0.17782853841781615, + "step": 12610 + }, + { + "epoch": 0.33243003855740644, + "grad_norm": 2.807861328125, + "learning_rate": 1.557519435912503e-05, + "loss": 0.15079200267791748, + "step": 12620 + }, + { + "epoch": 0.3326934538019051, + "grad_norm": 1.0507616996765137, + "learning_rate": 1.556810989359861e-05, + "loss": 0.1646086573600769, + "step": 12630 + }, + { + "epoch": 0.33295686904640387, + "grad_norm": 3.524193286895752, + "learning_rate": 1.5561021375361583e-05, + "loss": 0.16478803157806396, + "step": 12640 + }, + { + "epoch": 0.3332202842909026, + "grad_norm": 1.2690186500549316, + "learning_rate": 1.5553928809573275e-05, + "loss": 0.1654997706413269, + "step": 12650 + }, + { + "epoch": 0.33348369953540136, + "grad_norm": 1.6596980094909668, + "learning_rate": 1.5546832201395964e-05, + "loss": 0.1599993109703064, + "step": 12660 + }, + { + "epoch": 0.3337471147799001, + "grad_norm": 2.3701679706573486, + "learning_rate": 1.5539731555994864e-05, + "loss": 0.15994062423706054, + "step": 12670 + }, + { + "epoch": 0.33401053002439884, + "grad_norm": 2.056980609893799, + "learning_rate": 1.5532626878538137e-05, + "loss": 0.16137834787368774, + "step": 12680 + }, + { + "epoch": 0.3342739452688976, + "grad_norm": 2.894155979156494, + "learning_rate": 1.5525518174196867e-05, + "loss": 0.15564736127853393, + "step": 12690 + }, + { + "epoch": 0.33453736051339633, + "grad_norm": 2.1170570850372314, + "learning_rate": 1.551840544814508e-05, + "loss": 0.1553105115890503, + "step": 12700 + }, + { + "epoch": 0.3348007757578951, + "grad_norm": 1.3458045721054077, + "learning_rate": 1.551128870555972e-05, + "loss": 0.16175575256347657, + "step": 12710 + }, + { + "epoch": 0.33506419100239376, + "grad_norm": 1.8287495374679565, + "learning_rate": 1.550416795162067e-05, + "loss": 0.16637661457061767, + "step": 12720 + }, + { + "epoch": 0.3353276062468925, + "grad_norm": 2.209925413131714, + "learning_rate": 1.549704319151071e-05, + "loss": 0.16328543424606323, + "step": 12730 + }, + { + "epoch": 0.33559102149139125, + "grad_norm": 1.2938861846923828, + "learning_rate": 1.548991443041555e-05, + "loss": 0.15909361839294434, + "step": 12740 + }, + { + "epoch": 0.33585443673589, + "grad_norm": 2.484092950820923, + "learning_rate": 1.5482781673523816e-05, + "loss": 0.16873061656951904, + "step": 12750 + }, + { + "epoch": 0.33611785198038874, + "grad_norm": 1.9406691789627075, + "learning_rate": 1.5475644926027035e-05, + "loss": 0.1504009485244751, + "step": 12760 + }, + { + "epoch": 0.3363812672248875, + "grad_norm": 3.34218168258667, + "learning_rate": 1.5468504193119636e-05, + "loss": 0.17943694591522216, + "step": 12770 + }, + { + "epoch": 0.3366446824693862, + "grad_norm": 3.061660051345825, + "learning_rate": 1.5461359479998953e-05, + "loss": 0.161957848072052, + "step": 12780 + }, + { + "epoch": 0.33690809771388497, + "grad_norm": 1.4755637645721436, + "learning_rate": 1.5454210791865213e-05, + "loss": 0.16277238130569457, + "step": 12790 + }, + { + "epoch": 0.33717151295838366, + "grad_norm": 3.246464490890503, + "learning_rate": 1.5447058133921552e-05, + "loss": 0.1761563539505005, + "step": 12800 + }, + { + "epoch": 0.3374349282028824, + "grad_norm": 2.3984580039978027, + "learning_rate": 1.543990151137397e-05, + "loss": 0.16497646570205687, + "step": 12810 + }, + { + "epoch": 0.33769834344738114, + "grad_norm": 1.1797159910202026, + "learning_rate": 1.5432740929431372e-05, + "loss": 0.14909846782684327, + "step": 12820 + }, + { + "epoch": 0.3379617586918799, + "grad_norm": 3.093390703201294, + "learning_rate": 1.5425576393305537e-05, + "loss": 0.15970450639724731, + "step": 12830 + }, + { + "epoch": 0.33822517393637863, + "grad_norm": 1.3253108263015747, + "learning_rate": 1.5418407908211125e-05, + "loss": 0.1578690528869629, + "step": 12840 + }, + { + "epoch": 0.3384885891808774, + "grad_norm": 2.085578203201294, + "learning_rate": 1.5411235479365664e-05, + "loss": 0.1538609027862549, + "step": 12850 + }, + { + "epoch": 0.3387520044253761, + "grad_norm": 1.974002718925476, + "learning_rate": 1.5404059111989565e-05, + "loss": 0.17184979915618898, + "step": 12860 + }, + { + "epoch": 0.33901541966987486, + "grad_norm": 3.145155906677246, + "learning_rate": 1.5396878811306088e-05, + "loss": 0.16290626525878907, + "step": 12870 + }, + { + "epoch": 0.3392788349143736, + "grad_norm": 1.106196403503418, + "learning_rate": 1.5389694582541376e-05, + "loss": 0.16741195917129517, + "step": 12880 + }, + { + "epoch": 0.3395422501588723, + "grad_norm": 0.7461680173873901, + "learning_rate": 1.5382506430924414e-05, + "loss": 0.1566668391227722, + "step": 12890 + }, + { + "epoch": 0.33980566540337104, + "grad_norm": 2.1409449577331543, + "learning_rate": 1.5375314361687053e-05, + "loss": 0.15983819961547852, + "step": 12900 + }, + { + "epoch": 0.3400690806478698, + "grad_norm": 2.9824843406677246, + "learning_rate": 1.5368118380063984e-05, + "loss": 0.16839112043380738, + "step": 12910 + }, + { + "epoch": 0.3403324958923685, + "grad_norm": 2.536487102508545, + "learning_rate": 1.536091849129276e-05, + "loss": 0.1665171980857849, + "step": 12920 + }, + { + "epoch": 0.34059591113686727, + "grad_norm": 1.1583400964736938, + "learning_rate": 1.5353714700613768e-05, + "loss": 0.15846843719482423, + "step": 12930 + }, + { + "epoch": 0.340859326381366, + "grad_norm": 3.927927017211914, + "learning_rate": 1.5346507013270237e-05, + "loss": 0.15360299348831177, + "step": 12940 + }, + { + "epoch": 0.34112274162586476, + "grad_norm": 3.900160551071167, + "learning_rate": 1.5339295434508238e-05, + "loss": 0.15601952075958253, + "step": 12950 + }, + { + "epoch": 0.3413861568703635, + "grad_norm": 2.2197422981262207, + "learning_rate": 1.533207996957666e-05, + "loss": 0.15795159339904785, + "step": 12960 + }, + { + "epoch": 0.34164957211486224, + "grad_norm": 1.2409968376159668, + "learning_rate": 1.5324860623727236e-05, + "loss": 0.16267358064651488, + "step": 12970 + }, + { + "epoch": 0.34191298735936093, + "grad_norm": 3.0172178745269775, + "learning_rate": 1.5317637402214515e-05, + "loss": 0.16109175682067872, + "step": 12980 + }, + { + "epoch": 0.3421764026038597, + "grad_norm": 1.5983353853225708, + "learning_rate": 1.5310410310295876e-05, + "loss": 0.14954946041107178, + "step": 12990 + }, + { + "epoch": 0.3424398178483584, + "grad_norm": 1.9156423807144165, + "learning_rate": 1.5303179353231494e-05, + "loss": 0.16045722961425782, + "step": 13000 + }, + { + "epoch": 0.34270323309285716, + "grad_norm": 2.40293550491333, + "learning_rate": 1.529594453628439e-05, + "loss": 0.1605028986930847, + "step": 13010 + }, + { + "epoch": 0.3429666483373559, + "grad_norm": 1.6374614238739014, + "learning_rate": 1.5288705864720356e-05, + "loss": 0.16268960237503052, + "step": 13020 + }, + { + "epoch": 0.34323006358185465, + "grad_norm": 2.7389020919799805, + "learning_rate": 1.528146334380803e-05, + "loss": 0.185835599899292, + "step": 13030 + }, + { + "epoch": 0.3434934788263534, + "grad_norm": 1.7967411279678345, + "learning_rate": 1.5274216978818813e-05, + "loss": 0.15608874559402466, + "step": 13040 + }, + { + "epoch": 0.34375689407085214, + "grad_norm": 5.535101413726807, + "learning_rate": 1.526696677502693e-05, + "loss": 0.15928881168365477, + "step": 13050 + }, + { + "epoch": 0.3440203093153508, + "grad_norm": 2.938918352127075, + "learning_rate": 1.525971273770939e-05, + "loss": 0.15901625156402588, + "step": 13060 + }, + { + "epoch": 0.34428372455984957, + "grad_norm": 2.915039300918579, + "learning_rate": 1.5252454872146003e-05, + "loss": 0.1580638885498047, + "step": 13070 + }, + { + "epoch": 0.3445471398043483, + "grad_norm": 3.1087443828582764, + "learning_rate": 1.5245193183619347e-05, + "loss": 0.16780526638031007, + "step": 13080 + }, + { + "epoch": 0.34481055504884706, + "grad_norm": 0.7963283658027649, + "learning_rate": 1.5237927677414797e-05, + "loss": 0.16344361305236815, + "step": 13090 + }, + { + "epoch": 0.3450739702933458, + "grad_norm": 0.8021800518035889, + "learning_rate": 1.5230658358820493e-05, + "loss": 0.1558628797531128, + "step": 13100 + }, + { + "epoch": 0.34533738553784454, + "grad_norm": 3.1785812377929688, + "learning_rate": 1.5223385233127372e-05, + "loss": 0.1542830228805542, + "step": 13110 + }, + { + "epoch": 0.3456008007823433, + "grad_norm": 3.7884557247161865, + "learning_rate": 1.5216108305629123e-05, + "loss": 0.15545141696929932, + "step": 13120 + }, + { + "epoch": 0.34586421602684203, + "grad_norm": 3.336463689804077, + "learning_rate": 1.5208827581622205e-05, + "loss": 0.18085801601409912, + "step": 13130 + }, + { + "epoch": 0.3461276312713408, + "grad_norm": 1.03438401222229, + "learning_rate": 1.5201543066405844e-05, + "loss": 0.149586284160614, + "step": 13140 + }, + { + "epoch": 0.34639104651583946, + "grad_norm": 1.8687132596969604, + "learning_rate": 1.5194254765282025e-05, + "loss": 0.1672002911567688, + "step": 13150 + }, + { + "epoch": 0.3466544617603382, + "grad_norm": 2.2782304286956787, + "learning_rate": 1.5186962683555493e-05, + "loss": 0.1628553032875061, + "step": 13160 + }, + { + "epoch": 0.34691787700483695, + "grad_norm": 1.171496868133545, + "learning_rate": 1.517966682653373e-05, + "loss": 0.1521201491355896, + "step": 13170 + }, + { + "epoch": 0.3471812922493357, + "grad_norm": 3.6065218448638916, + "learning_rate": 1.5172367199526985e-05, + "loss": 0.15939595699310302, + "step": 13180 + }, + { + "epoch": 0.34744470749383444, + "grad_norm": 2.9318933486938477, + "learning_rate": 1.5165063807848234e-05, + "loss": 0.1728064775466919, + "step": 13190 + }, + { + "epoch": 0.3477081227383332, + "grad_norm": 4.013476848602295, + "learning_rate": 1.5157756656813207e-05, + "loss": 0.17528828382492065, + "step": 13200 + }, + { + "epoch": 0.3479715379828319, + "grad_norm": 3.087224006652832, + "learning_rate": 1.5150445751740356e-05, + "loss": 0.1524742841720581, + "step": 13210 + }, + { + "epoch": 0.34823495322733067, + "grad_norm": 1.6064430475234985, + "learning_rate": 1.5143131097950882e-05, + "loss": 0.15364410877227783, + "step": 13220 + }, + { + "epoch": 0.3484983684718294, + "grad_norm": 2.0919885635375977, + "learning_rate": 1.5135812700768702e-05, + "loss": 0.17493629455566406, + "step": 13230 + }, + { + "epoch": 0.3487617837163281, + "grad_norm": 0.8896381855010986, + "learning_rate": 1.512849056552046e-05, + "loss": 0.1534730911254883, + "step": 13240 + }, + { + "epoch": 0.34902519896082684, + "grad_norm": 2.0714194774627686, + "learning_rate": 1.512116469753552e-05, + "loss": 0.15756642818450928, + "step": 13250 + }, + { + "epoch": 0.3492886142053256, + "grad_norm": 2.51999831199646, + "learning_rate": 1.5113835102145968e-05, + "loss": 0.15981428623199462, + "step": 13260 + }, + { + "epoch": 0.34955202944982433, + "grad_norm": 1.2751866579055786, + "learning_rate": 1.51065017846866e-05, + "loss": 0.1564551830291748, + "step": 13270 + }, + { + "epoch": 0.3498154446943231, + "grad_norm": 2.0626587867736816, + "learning_rate": 1.509916475049492e-05, + "loss": 0.15018190145492555, + "step": 13280 + }, + { + "epoch": 0.3500788599388218, + "grad_norm": 4.556334495544434, + "learning_rate": 1.5091824004911137e-05, + "loss": 0.18787047863006592, + "step": 13290 + }, + { + "epoch": 0.35034227518332056, + "grad_norm": 3.608231782913208, + "learning_rate": 1.5084479553278164e-05, + "loss": 0.16101791858673095, + "step": 13300 + }, + { + "epoch": 0.3506056904278193, + "grad_norm": 2.0719916820526123, + "learning_rate": 1.5077131400941609e-05, + "loss": 0.15561792850494385, + "step": 13310 + }, + { + "epoch": 0.350869105672318, + "grad_norm": 0.6682707071304321, + "learning_rate": 1.5069779553249775e-05, + "loss": 0.15185623168945311, + "step": 13320 + }, + { + "epoch": 0.35113252091681674, + "grad_norm": 2.3800504207611084, + "learning_rate": 1.5062424015553653e-05, + "loss": 0.18110270500183107, + "step": 13330 + }, + { + "epoch": 0.3513959361613155, + "grad_norm": 2.605989694595337, + "learning_rate": 1.505506479320692e-05, + "loss": 0.1526637077331543, + "step": 13340 + }, + { + "epoch": 0.3516593514058142, + "grad_norm": 2.656242847442627, + "learning_rate": 1.5047701891565938e-05, + "loss": 0.1627279758453369, + "step": 13350 + }, + { + "epoch": 0.35192276665031297, + "grad_norm": 1.1875625848770142, + "learning_rate": 1.504033531598974e-05, + "loss": 0.15445574522018432, + "step": 13360 + }, + { + "epoch": 0.3521861818948117, + "grad_norm": 3.046935796737671, + "learning_rate": 1.5032965071840045e-05, + "loss": 0.16264586448669432, + "step": 13370 + }, + { + "epoch": 0.35244959713931046, + "grad_norm": 0.81063312292099, + "learning_rate": 1.5025591164481226e-05, + "loss": 0.1737267255783081, + "step": 13380 + }, + { + "epoch": 0.3527130123838092, + "grad_norm": 3.8570234775543213, + "learning_rate": 1.5018213599280339e-05, + "loss": 0.15850013494491577, + "step": 13390 + }, + { + "epoch": 0.35297642762830794, + "grad_norm": 1.981008768081665, + "learning_rate": 1.5010832381607085e-05, + "loss": 0.166035258769989, + "step": 13400 + }, + { + "epoch": 0.35323984287280663, + "grad_norm": 4.30646276473999, + "learning_rate": 1.500344751683384e-05, + "loss": 0.1662672758102417, + "step": 13410 + }, + { + "epoch": 0.3535032581173054, + "grad_norm": 4.994363307952881, + "learning_rate": 1.4996059010335626e-05, + "loss": 0.16438839435577393, + "step": 13420 + }, + { + "epoch": 0.3537666733618041, + "grad_norm": 1.4153422117233276, + "learning_rate": 1.4988666867490117e-05, + "loss": 0.15160995721817017, + "step": 13430 + }, + { + "epoch": 0.35403008860630286, + "grad_norm": 5.023283004760742, + "learning_rate": 1.4981271093677632e-05, + "loss": 0.15431373119354247, + "step": 13440 + }, + { + "epoch": 0.3542935038508016, + "grad_norm": 4.023380279541016, + "learning_rate": 1.4973871694281133e-05, + "loss": 0.1668138861656189, + "step": 13450 + }, + { + "epoch": 0.35455691909530035, + "grad_norm": 2.8145172595977783, + "learning_rate": 1.4966468674686226e-05, + "loss": 0.15174081325531005, + "step": 13460 + }, + { + "epoch": 0.3548203343397991, + "grad_norm": 3.650362968444824, + "learning_rate": 1.4959062040281143e-05, + "loss": 0.15896077156066896, + "step": 13470 + }, + { + "epoch": 0.35508374958429784, + "grad_norm": 2.6771116256713867, + "learning_rate": 1.4951651796456758e-05, + "loss": 0.16511976718902588, + "step": 13480 + }, + { + "epoch": 0.3553471648287966, + "grad_norm": 1.6405799388885498, + "learning_rate": 1.4944237948606561e-05, + "loss": 0.15167560577392578, + "step": 13490 + }, + { + "epoch": 0.35561058007329527, + "grad_norm": 1.416763424873352, + "learning_rate": 1.4936820502126676e-05, + "loss": 0.14866940975189208, + "step": 13500 + }, + { + "epoch": 0.355873995317794, + "grad_norm": 2.9921517372131348, + "learning_rate": 1.4929399462415832e-05, + "loss": 0.1611330270767212, + "step": 13510 + }, + { + "epoch": 0.35613741056229276, + "grad_norm": 1.8662618398666382, + "learning_rate": 1.492197483487539e-05, + "loss": 0.15894768238067628, + "step": 13520 + }, + { + "epoch": 0.3564008258067915, + "grad_norm": 3.4335973262786865, + "learning_rate": 1.491454662490931e-05, + "loss": 0.1668680191040039, + "step": 13530 + }, + { + "epoch": 0.35666424105129024, + "grad_norm": 1.9441567659378052, + "learning_rate": 1.4907114837924168e-05, + "loss": 0.15988341569900513, + "step": 13540 + }, + { + "epoch": 0.356927656295789, + "grad_norm": 1.0773409605026245, + "learning_rate": 1.4899679479329133e-05, + "loss": 0.15524792671203613, + "step": 13550 + }, + { + "epoch": 0.35719107154028773, + "grad_norm": 2.8398571014404297, + "learning_rate": 1.4892240554535981e-05, + "loss": 0.15231478214263916, + "step": 13560 + }, + { + "epoch": 0.3574544867847865, + "grad_norm": 0.968870997428894, + "learning_rate": 1.4884798068959085e-05, + "loss": 0.15070971250534057, + "step": 13570 + }, + { + "epoch": 0.3577179020292852, + "grad_norm": 1.6262853145599365, + "learning_rate": 1.4877352028015408e-05, + "loss": 0.15795416831970216, + "step": 13580 + }, + { + "epoch": 0.3579813172737839, + "grad_norm": 1.0119630098342896, + "learning_rate": 1.4869902437124499e-05, + "loss": 0.1664564609527588, + "step": 13590 + }, + { + "epoch": 0.35824473251828265, + "grad_norm": 3.1567630767822266, + "learning_rate": 1.486244930170849e-05, + "loss": 0.1601466417312622, + "step": 13600 + }, + { + "epoch": 0.3585081477627814, + "grad_norm": 1.4302423000335693, + "learning_rate": 1.4854992627192096e-05, + "loss": 0.16933691501617432, + "step": 13610 + }, + { + "epoch": 0.35877156300728014, + "grad_norm": 3.644655704498291, + "learning_rate": 1.4847532419002602e-05, + "loss": 0.15899367332458497, + "step": 13620 + }, + { + "epoch": 0.3590349782517789, + "grad_norm": 0.960704505443573, + "learning_rate": 1.4840068682569877e-05, + "loss": 0.16334836483001708, + "step": 13630 + }, + { + "epoch": 0.3592983934962776, + "grad_norm": 3.4693617820739746, + "learning_rate": 1.4832601423326346e-05, + "loss": 0.13824304342269897, + "step": 13640 + }, + { + "epoch": 0.35956180874077637, + "grad_norm": 1.6642053127288818, + "learning_rate": 1.4825130646707002e-05, + "loss": 0.13877997398376465, + "step": 13650 + }, + { + "epoch": 0.3598252239852751, + "grad_norm": 2.690894603729248, + "learning_rate": 1.4817656358149398e-05, + "loss": 0.164223313331604, + "step": 13660 + }, + { + "epoch": 0.3600886392297738, + "grad_norm": 0.9582263827323914, + "learning_rate": 1.4810178563093648e-05, + "loss": 0.16622185707092285, + "step": 13670 + }, + { + "epoch": 0.36035205447427254, + "grad_norm": 1.7703144550323486, + "learning_rate": 1.4802697266982411e-05, + "loss": 0.1696224331855774, + "step": 13680 + }, + { + "epoch": 0.3606154697187713, + "grad_norm": 2.6746201515197754, + "learning_rate": 1.4795212475260896e-05, + "loss": 0.1658730149269104, + "step": 13690 + }, + { + "epoch": 0.36087888496327003, + "grad_norm": 1.4631884098052979, + "learning_rate": 1.4787724193376858e-05, + "loss": 0.1535053849220276, + "step": 13700 + }, + { + "epoch": 0.3611423002077688, + "grad_norm": 3.784512758255005, + "learning_rate": 1.4780232426780592e-05, + "loss": 0.15698496103286744, + "step": 13710 + }, + { + "epoch": 0.3614057154522675, + "grad_norm": 1.292128086090088, + "learning_rate": 1.477273718092493e-05, + "loss": 0.1622738480567932, + "step": 13720 + }, + { + "epoch": 0.36166913069676626, + "grad_norm": 1.646433711051941, + "learning_rate": 1.4765238461265235e-05, + "loss": 0.1633553385734558, + "step": 13730 + }, + { + "epoch": 0.361932545941265, + "grad_norm": 2.133492946624756, + "learning_rate": 1.4757736273259396e-05, + "loss": 0.16902320384979247, + "step": 13740 + }, + { + "epoch": 0.36219596118576375, + "grad_norm": 1.7885822057724, + "learning_rate": 1.4750230622367831e-05, + "loss": 0.153229284286499, + "step": 13750 + }, + { + "epoch": 0.36245937643026244, + "grad_norm": 4.212640285491943, + "learning_rate": 1.474272151405348e-05, + "loss": 0.15916681289672852, + "step": 13760 + }, + { + "epoch": 0.3627227916747612, + "grad_norm": 1.9911668300628662, + "learning_rate": 1.473520895378179e-05, + "loss": 0.16323832273483277, + "step": 13770 + }, + { + "epoch": 0.3629862069192599, + "grad_norm": 2.7139556407928467, + "learning_rate": 1.472769294702073e-05, + "loss": 0.173483407497406, + "step": 13780 + }, + { + "epoch": 0.36324962216375867, + "grad_norm": 4.811188697814941, + "learning_rate": 1.4720173499240771e-05, + "loss": 0.1607419729232788, + "step": 13790 + }, + { + "epoch": 0.3635130374082574, + "grad_norm": 3.2883107662200928, + "learning_rate": 1.4712650615914899e-05, + "loss": 0.16939907073974608, + "step": 13800 + }, + { + "epoch": 0.36377645265275615, + "grad_norm": 3.447277545928955, + "learning_rate": 1.4705124302518581e-05, + "loss": 0.1761792778968811, + "step": 13810 + }, + { + "epoch": 0.3640398678972549, + "grad_norm": 2.5899293422698975, + "learning_rate": 1.4697594564529805e-05, + "loss": 0.17301149368286134, + "step": 13820 + }, + { + "epoch": 0.36430328314175364, + "grad_norm": 1.5445622205734253, + "learning_rate": 1.4690061407429028e-05, + "loss": 0.1558293581008911, + "step": 13830 + }, + { + "epoch": 0.3645666983862524, + "grad_norm": 1.4138950109481812, + "learning_rate": 1.468252483669922e-05, + "loss": 0.17048161029815673, + "step": 13840 + }, + { + "epoch": 0.3648301136307511, + "grad_norm": 1.8431432247161865, + "learning_rate": 1.4674984857825808e-05, + "loss": 0.17019615173339844, + "step": 13850 + }, + { + "epoch": 0.3650935288752498, + "grad_norm": 2.6899023056030273, + "learning_rate": 1.4667441476296723e-05, + "loss": 0.16184841394424437, + "step": 13860 + }, + { + "epoch": 0.36535694411974856, + "grad_norm": 1.5707675218582153, + "learning_rate": 1.4659894697602358e-05, + "loss": 0.1585773229598999, + "step": 13870 + }, + { + "epoch": 0.3656203593642473, + "grad_norm": 3.037916660308838, + "learning_rate": 1.465234452723559e-05, + "loss": 0.15600180625915527, + "step": 13880 + }, + { + "epoch": 0.36588377460874605, + "grad_norm": 2.3809237480163574, + "learning_rate": 1.4644790970691755e-05, + "loss": 0.15218397378921508, + "step": 13890 + }, + { + "epoch": 0.3661471898532448, + "grad_norm": 2.3208444118499756, + "learning_rate": 1.4637234033468658e-05, + "loss": 0.15720827579498292, + "step": 13900 + }, + { + "epoch": 0.36641060509774354, + "grad_norm": 0.6232905387878418, + "learning_rate": 1.4629673721066566e-05, + "loss": 0.1573716640472412, + "step": 13910 + }, + { + "epoch": 0.3666740203422423, + "grad_norm": 3.22369122505188, + "learning_rate": 1.4622110038988196e-05, + "loss": 0.1689432144165039, + "step": 13920 + }, + { + "epoch": 0.36693743558674097, + "grad_norm": 0.9221498966217041, + "learning_rate": 1.4614542992738727e-05, + "loss": 0.16958801746368407, + "step": 13930 + }, + { + "epoch": 0.3672008508312397, + "grad_norm": 0.6792687773704529, + "learning_rate": 1.4606972587825779e-05, + "loss": 0.16753238439559937, + "step": 13940 + }, + { + "epoch": 0.36746426607573845, + "grad_norm": 2.944368839263916, + "learning_rate": 1.4599398829759419e-05, + "loss": 0.15743356943130493, + "step": 13950 + }, + { + "epoch": 0.3677276813202372, + "grad_norm": 1.915217399597168, + "learning_rate": 1.4591821724052155e-05, + "loss": 0.15311715602874756, + "step": 13960 + }, + { + "epoch": 0.36799109656473594, + "grad_norm": 1.228477120399475, + "learning_rate": 1.4584241276218935e-05, + "loss": 0.16933056116104125, + "step": 13970 + }, + { + "epoch": 0.3682545118092347, + "grad_norm": 0.9250273108482361, + "learning_rate": 1.4576657491777134e-05, + "loss": 0.16354241371154785, + "step": 13980 + }, + { + "epoch": 0.36851792705373343, + "grad_norm": 2.0059986114501953, + "learning_rate": 1.4569070376246552e-05, + "loss": 0.16388382911682128, + "step": 13990 + }, + { + "epoch": 0.3687813422982322, + "grad_norm": 1.0544192790985107, + "learning_rate": 1.4561479935149426e-05, + "loss": 0.1644890308380127, + "step": 14000 + }, + { + "epoch": 0.3690447575427309, + "grad_norm": 2.7023472785949707, + "learning_rate": 1.4553886174010405e-05, + "loss": 0.17325005531311036, + "step": 14010 + }, + { + "epoch": 0.3693081727872296, + "grad_norm": 0.8957515358924866, + "learning_rate": 1.4546289098356556e-05, + "loss": 0.14678511619567872, + "step": 14020 + }, + { + "epoch": 0.36957158803172835, + "grad_norm": 1.2469077110290527, + "learning_rate": 1.4538688713717357e-05, + "loss": 0.15160919427871705, + "step": 14030 + }, + { + "epoch": 0.3698350032762271, + "grad_norm": 4.098679542541504, + "learning_rate": 1.4531085025624694e-05, + "loss": 0.16026103496551514, + "step": 14040 + }, + { + "epoch": 0.37009841852072584, + "grad_norm": 0.753747284412384, + "learning_rate": 1.4523478039612863e-05, + "loss": 0.18051421642303467, + "step": 14050 + }, + { + "epoch": 0.3703618337652246, + "grad_norm": 2.01808500289917, + "learning_rate": 1.4515867761218558e-05, + "loss": 0.16089764833450318, + "step": 14060 + }, + { + "epoch": 0.3706252490097233, + "grad_norm": 1.9346779584884644, + "learning_rate": 1.4508254195980861e-05, + "loss": 0.15458993911743163, + "step": 14070 + }, + { + "epoch": 0.37088866425422207, + "grad_norm": 1.6162859201431274, + "learning_rate": 1.450063734944126e-05, + "loss": 0.1641873836517334, + "step": 14080 + }, + { + "epoch": 0.3711520794987208, + "grad_norm": 1.8483895063400269, + "learning_rate": 1.449301722714362e-05, + "loss": 0.16251323223114014, + "step": 14090 + }, + { + "epoch": 0.37141549474321955, + "grad_norm": 1.3683427572250366, + "learning_rate": 1.4485393834634202e-05, + "loss": 0.16002782583236694, + "step": 14100 + }, + { + "epoch": 0.37167890998771824, + "grad_norm": 2.6834895610809326, + "learning_rate": 1.447776717746163e-05, + "loss": 0.15898462533950805, + "step": 14110 + }, + { + "epoch": 0.371942325232217, + "grad_norm": 1.6313940286636353, + "learning_rate": 1.4470137261176922e-05, + "loss": 0.16718592643737792, + "step": 14120 + }, + { + "epoch": 0.37220574047671573, + "grad_norm": 5.210270404815674, + "learning_rate": 1.4462504091333454e-05, + "loss": 0.1639876365661621, + "step": 14130 + }, + { + "epoch": 0.3724691557212145, + "grad_norm": 1.9529401063919067, + "learning_rate": 1.4454867673486983e-05, + "loss": 0.16868382692337036, + "step": 14140 + }, + { + "epoch": 0.3727325709657132, + "grad_norm": 2.664555072784424, + "learning_rate": 1.4447228013195616e-05, + "loss": 0.1628589630126953, + "step": 14150 + }, + { + "epoch": 0.37299598621021196, + "grad_norm": 1.6162710189819336, + "learning_rate": 1.4439585116019835e-05, + "loss": 0.16540224552154542, + "step": 14160 + }, + { + "epoch": 0.3732594014547107, + "grad_norm": 2.549570322036743, + "learning_rate": 1.4431938987522463e-05, + "loss": 0.17100212574005128, + "step": 14170 + }, + { + "epoch": 0.37352281669920945, + "grad_norm": 2.287505865097046, + "learning_rate": 1.4424289633268689e-05, + "loss": 0.1748489260673523, + "step": 14180 + }, + { + "epoch": 0.37378623194370814, + "grad_norm": 1.4811744689941406, + "learning_rate": 1.4416637058826038e-05, + "loss": 0.1523346185684204, + "step": 14190 + }, + { + "epoch": 0.3740496471882069, + "grad_norm": 2.65850567817688, + "learning_rate": 1.4408981269764386e-05, + "loss": 0.16391570568084718, + "step": 14200 + }, + { + "epoch": 0.3743130624327056, + "grad_norm": 3.1254734992980957, + "learning_rate": 1.4401322271655947e-05, + "loss": 0.15574090480804442, + "step": 14210 + }, + { + "epoch": 0.37457647767720437, + "grad_norm": 1.7017269134521484, + "learning_rate": 1.4393660070075267e-05, + "loss": 0.1653306484222412, + "step": 14220 + }, + { + "epoch": 0.3748398929217031, + "grad_norm": 4.008907794952393, + "learning_rate": 1.4385994670599233e-05, + "loss": 0.1614455461502075, + "step": 14230 + }, + { + "epoch": 0.37510330816620185, + "grad_norm": 2.281111478805542, + "learning_rate": 1.4378326078807048e-05, + "loss": 0.1513892650604248, + "step": 14240 + }, + { + "epoch": 0.3753667234107006, + "grad_norm": 0.8905887007713318, + "learning_rate": 1.4370654300280246e-05, + "loss": 0.1675179719924927, + "step": 14250 + }, + { + "epoch": 0.37563013865519934, + "grad_norm": 1.8461934328079224, + "learning_rate": 1.4362979340602677e-05, + "loss": 0.1649237871170044, + "step": 14260 + }, + { + "epoch": 0.3758935538996981, + "grad_norm": 1.2684061527252197, + "learning_rate": 1.4355301205360513e-05, + "loss": 0.16813116073608397, + "step": 14270 + }, + { + "epoch": 0.3761569691441968, + "grad_norm": 1.21127188205719, + "learning_rate": 1.4347619900142223e-05, + "loss": 0.17758638858795167, + "step": 14280 + }, + { + "epoch": 0.3764203843886955, + "grad_norm": 2.815572500228882, + "learning_rate": 1.43399354305386e-05, + "loss": 0.1486414074897766, + "step": 14290 + }, + { + "epoch": 0.37668379963319426, + "grad_norm": 4.958432674407959, + "learning_rate": 1.4332247802142729e-05, + "loss": 0.18005688190460206, + "step": 14300 + }, + { + "epoch": 0.376947214877693, + "grad_norm": 1.2346845865249634, + "learning_rate": 1.4324557020550001e-05, + "loss": 0.1622421622276306, + "step": 14310 + }, + { + "epoch": 0.37721063012219175, + "grad_norm": 1.2694391012191772, + "learning_rate": 1.4316863091358093e-05, + "loss": 0.15512681007385254, + "step": 14320 + }, + { + "epoch": 0.3774740453666905, + "grad_norm": 1.0510679483413696, + "learning_rate": 1.4309166020166984e-05, + "loss": 0.1447918653488159, + "step": 14330 + }, + { + "epoch": 0.37773746061118924, + "grad_norm": 1.152713656425476, + "learning_rate": 1.430146581257893e-05, + "loss": 0.1583552837371826, + "step": 14340 + }, + { + "epoch": 0.378000875855688, + "grad_norm": 3.0620813369750977, + "learning_rate": 1.4293762474198478e-05, + "loss": 0.15753687620162965, + "step": 14350 + }, + { + "epoch": 0.3782642911001867, + "grad_norm": 2.7572989463806152, + "learning_rate": 1.428605601063245e-05, + "loss": 0.16000353097915648, + "step": 14360 + }, + { + "epoch": 0.3785277063446854, + "grad_norm": 1.1954624652862549, + "learning_rate": 1.427834642748994e-05, + "loss": 0.15904098749160767, + "step": 14370 + }, + { + "epoch": 0.37879112158918415, + "grad_norm": 1.827528953552246, + "learning_rate": 1.4270633730382315e-05, + "loss": 0.17010276317596434, + "step": 14380 + }, + { + "epoch": 0.3790545368336829, + "grad_norm": 2.87675404548645, + "learning_rate": 1.4262917924923212e-05, + "loss": 0.15329018831253052, + "step": 14390 + }, + { + "epoch": 0.37931795207818164, + "grad_norm": 2.336273670196533, + "learning_rate": 1.4255199016728527e-05, + "loss": 0.16068037748336791, + "step": 14400 + }, + { + "epoch": 0.3795813673226804, + "grad_norm": 1.6187307834625244, + "learning_rate": 1.4247477011416409e-05, + "loss": 0.16208419799804688, + "step": 14410 + }, + { + "epoch": 0.37984478256717913, + "grad_norm": 0.8680592179298401, + "learning_rate": 1.4239751914607274e-05, + "loss": 0.16738697290420532, + "step": 14420 + }, + { + "epoch": 0.3801081978116779, + "grad_norm": 3.5835933685302734, + "learning_rate": 1.4232023731923772e-05, + "loss": 0.14882376194000244, + "step": 14430 + }, + { + "epoch": 0.3803716130561766, + "grad_norm": 1.0542619228363037, + "learning_rate": 1.422429246899082e-05, + "loss": 0.16604315042495726, + "step": 14440 + }, + { + "epoch": 0.3806350283006753, + "grad_norm": 2.0738022327423096, + "learning_rate": 1.4216558131435551e-05, + "loss": 0.1598573923110962, + "step": 14450 + }, + { + "epoch": 0.38089844354517405, + "grad_norm": 1.2943708896636963, + "learning_rate": 1.420882072488736e-05, + "loss": 0.15874643325805665, + "step": 14460 + }, + { + "epoch": 0.3811618587896728, + "grad_norm": 2.983804941177368, + "learning_rate": 1.420108025497786e-05, + "loss": 0.16908255815505982, + "step": 14470 + }, + { + "epoch": 0.38142527403417154, + "grad_norm": 1.1933165788650513, + "learning_rate": 1.41933367273409e-05, + "loss": 0.1581798315048218, + "step": 14480 + }, + { + "epoch": 0.3816886892786703, + "grad_norm": 1.5747376680374146, + "learning_rate": 1.4185590147612555e-05, + "loss": 0.15923535823822021, + "step": 14490 + }, + { + "epoch": 0.381952104523169, + "grad_norm": 4.817906856536865, + "learning_rate": 1.4177840521431117e-05, + "loss": 0.15523736476898192, + "step": 14500 + }, + { + "epoch": 0.38221551976766777, + "grad_norm": 5.585853576660156, + "learning_rate": 1.4170087854437102e-05, + "loss": 0.1499537467956543, + "step": 14510 + }, + { + "epoch": 0.3824789350121665, + "grad_norm": 2.1137115955352783, + "learning_rate": 1.4162332152273233e-05, + "loss": 0.17018527984619142, + "step": 14520 + }, + { + "epoch": 0.38274235025666525, + "grad_norm": 2.152005195617676, + "learning_rate": 1.4154573420584448e-05, + "loss": 0.15803229808807373, + "step": 14530 + }, + { + "epoch": 0.38300576550116394, + "grad_norm": 1.3991780281066895, + "learning_rate": 1.414681166501788e-05, + "loss": 0.16172605752944946, + "step": 14540 + }, + { + "epoch": 0.3832691807456627, + "grad_norm": 1.1816176176071167, + "learning_rate": 1.4139046891222877e-05, + "loss": 0.17250592708587648, + "step": 14550 + }, + { + "epoch": 0.38353259599016143, + "grad_norm": 1.585795521736145, + "learning_rate": 1.4131279104850968e-05, + "loss": 0.1866520643234253, + "step": 14560 + }, + { + "epoch": 0.3837960112346602, + "grad_norm": 1.194443702697754, + "learning_rate": 1.4123508311555894e-05, + "loss": 0.15434713363647462, + "step": 14570 + }, + { + "epoch": 0.3840594264791589, + "grad_norm": 1.8122276067733765, + "learning_rate": 1.4115734516993562e-05, + "loss": 0.1633039355278015, + "step": 14580 + }, + { + "epoch": 0.38432284172365766, + "grad_norm": 5.219485282897949, + "learning_rate": 1.4107957726822084e-05, + "loss": 0.1631326675415039, + "step": 14590 + }, + { + "epoch": 0.3845862569681564, + "grad_norm": 3.865084648132324, + "learning_rate": 1.4100177946701738e-05, + "loss": 0.16815831661224365, + "step": 14600 + }, + { + "epoch": 0.38484967221265515, + "grad_norm": 1.1253814697265625, + "learning_rate": 1.4092395182294989e-05, + "loss": 0.16539604663848878, + "step": 14610 + }, + { + "epoch": 0.3851130874571539, + "grad_norm": 1.7343097925186157, + "learning_rate": 1.4084609439266464e-05, + "loss": 0.16718039512634278, + "step": 14620 + }, + { + "epoch": 0.3853765027016526, + "grad_norm": 1.5060738325119019, + "learning_rate": 1.4076820723282962e-05, + "loss": 0.1583859443664551, + "step": 14630 + }, + { + "epoch": 0.3856399179461513, + "grad_norm": 1.0225528478622437, + "learning_rate": 1.4069029040013448e-05, + "loss": 0.17784805297851564, + "step": 14640 + }, + { + "epoch": 0.38590333319065007, + "grad_norm": 2.295485496520996, + "learning_rate": 1.4061234395129049e-05, + "loss": 0.16835296154022217, + "step": 14650 + }, + { + "epoch": 0.3861667484351488, + "grad_norm": 2.874044418334961, + "learning_rate": 1.405343679430304e-05, + "loss": 0.15912588834762573, + "step": 14660 + }, + { + "epoch": 0.38643016367964755, + "grad_norm": 1.7891255617141724, + "learning_rate": 1.4045636243210852e-05, + "loss": 0.16723607778549193, + "step": 14670 + }, + { + "epoch": 0.3866935789241463, + "grad_norm": 1.0418012142181396, + "learning_rate": 1.4037832747530063e-05, + "loss": 0.1545453429222107, + "step": 14680 + }, + { + "epoch": 0.38695699416864504, + "grad_norm": 1.2165640592575073, + "learning_rate": 1.403002631294039e-05, + "loss": 0.1597032904624939, + "step": 14690 + }, + { + "epoch": 0.3872204094131438, + "grad_norm": 1.069140076637268, + "learning_rate": 1.40222169451237e-05, + "loss": 0.15132637023925782, + "step": 14700 + }, + { + "epoch": 0.3874838246576425, + "grad_norm": 1.5442016124725342, + "learning_rate": 1.4014404649763983e-05, + "loss": 0.16818814277648925, + "step": 14710 + }, + { + "epoch": 0.3877472399021412, + "grad_norm": 3.66473126411438, + "learning_rate": 1.4006589432547367e-05, + "loss": 0.16642171144485474, + "step": 14720 + }, + { + "epoch": 0.38801065514663996, + "grad_norm": 2.6610519886016846, + "learning_rate": 1.3998771299162099e-05, + "loss": 0.16236573457717896, + "step": 14730 + }, + { + "epoch": 0.3882740703911387, + "grad_norm": 3.119645595550537, + "learning_rate": 1.3990950255298565e-05, + "loss": 0.16514413356781005, + "step": 14740 + }, + { + "epoch": 0.38853748563563745, + "grad_norm": 1.039487361907959, + "learning_rate": 1.3983126306649243e-05, + "loss": 0.17968955039978027, + "step": 14750 + }, + { + "epoch": 0.3888009008801362, + "grad_norm": 3.853801727294922, + "learning_rate": 1.3975299458908753e-05, + "loss": 0.155118727684021, + "step": 14760 + }, + { + "epoch": 0.38906431612463493, + "grad_norm": 2.1599533557891846, + "learning_rate": 1.3967469717773807e-05, + "loss": 0.16734516620635986, + "step": 14770 + }, + { + "epoch": 0.3893277313691337, + "grad_norm": 1.430975079536438, + "learning_rate": 1.395963708894323e-05, + "loss": 0.15001704692840576, + "step": 14780 + }, + { + "epoch": 0.3895911466136324, + "grad_norm": 3.091114044189453, + "learning_rate": 1.3951801578117946e-05, + "loss": 0.16479849815368652, + "step": 14790 + }, + { + "epoch": 0.3898545618581311, + "grad_norm": 0.9745585322380066, + "learning_rate": 1.3943963191000976e-05, + "loss": 0.15710713863372802, + "step": 14800 + }, + { + "epoch": 0.39011797710262985, + "grad_norm": 2.475353479385376, + "learning_rate": 1.393612193329744e-05, + "loss": 0.1726353645324707, + "step": 14810 + }, + { + "epoch": 0.3903813923471286, + "grad_norm": 4.893131256103516, + "learning_rate": 1.3928277810714542e-05, + "loss": 0.16303844451904298, + "step": 14820 + }, + { + "epoch": 0.39064480759162734, + "grad_norm": 3.118406057357788, + "learning_rate": 1.3920430828961576e-05, + "loss": 0.15522629022598267, + "step": 14830 + }, + { + "epoch": 0.3909082228361261, + "grad_norm": 1.3194411993026733, + "learning_rate": 1.3912580993749907e-05, + "loss": 0.15890185832977294, + "step": 14840 + }, + { + "epoch": 0.39117163808062483, + "grad_norm": 1.885522723197937, + "learning_rate": 1.3904728310792988e-05, + "loss": 0.16213818788528442, + "step": 14850 + }, + { + "epoch": 0.39143505332512357, + "grad_norm": 2.0854644775390625, + "learning_rate": 1.3896872785806342e-05, + "loss": 0.16000847816467284, + "step": 14860 + }, + { + "epoch": 0.3916984685696223, + "grad_norm": 4.021148204803467, + "learning_rate": 1.3889014424507557e-05, + "loss": 0.14992682933807372, + "step": 14870 + }, + { + "epoch": 0.39196188381412106, + "grad_norm": 1.7427293062210083, + "learning_rate": 1.3881153232616284e-05, + "loss": 0.15682170391082764, + "step": 14880 + }, + { + "epoch": 0.39222529905861975, + "grad_norm": 1.5437815189361572, + "learning_rate": 1.3873289215854247e-05, + "loss": 0.16317398548126222, + "step": 14890 + }, + { + "epoch": 0.3924887143031185, + "grad_norm": 2.3758580684661865, + "learning_rate": 1.3865422379945205e-05, + "loss": 0.1464613199234009, + "step": 14900 + }, + { + "epoch": 0.39275212954761723, + "grad_norm": 1.4198546409606934, + "learning_rate": 1.3857552730614989e-05, + "loss": 0.17815663814544677, + "step": 14910 + }, + { + "epoch": 0.393015544792116, + "grad_norm": 1.6661925315856934, + "learning_rate": 1.3849680273591468e-05, + "loss": 0.1637780785560608, + "step": 14920 + }, + { + "epoch": 0.3932789600366147, + "grad_norm": 1.8396483659744263, + "learning_rate": 1.3841805014604553e-05, + "loss": 0.16942863464355468, + "step": 14930 + }, + { + "epoch": 0.39354237528111347, + "grad_norm": 3.1518585681915283, + "learning_rate": 1.3833926959386199e-05, + "loss": 0.16120415925979614, + "step": 14940 + }, + { + "epoch": 0.3938057905256122, + "grad_norm": 1.932082176208496, + "learning_rate": 1.3826046113670395e-05, + "loss": 0.16232476234436036, + "step": 14950 + }, + { + "epoch": 0.39406920577011095, + "grad_norm": 1.070197582244873, + "learning_rate": 1.3818162483193159e-05, + "loss": 0.160104501247406, + "step": 14960 + }, + { + "epoch": 0.39433262101460964, + "grad_norm": 0.8047518134117126, + "learning_rate": 1.3810276073692537e-05, + "loss": 0.14788737297058105, + "step": 14970 + }, + { + "epoch": 0.3945960362591084, + "grad_norm": 0.9767709374427795, + "learning_rate": 1.3802386890908602e-05, + "loss": 0.1574622869491577, + "step": 14980 + }, + { + "epoch": 0.39485945150360713, + "grad_norm": 1.8476247787475586, + "learning_rate": 1.3794494940583437e-05, + "loss": 0.16136462688446046, + "step": 14990 + }, + { + "epoch": 0.39512286674810587, + "grad_norm": 1.8728665113449097, + "learning_rate": 1.3786600228461147e-05, + "loss": 0.16258159875869752, + "step": 15000 + }, + { + "epoch": 0.3953862819926046, + "grad_norm": 3.8335442543029785, + "learning_rate": 1.377870276028784e-05, + "loss": 0.1611560583114624, + "step": 15010 + }, + { + "epoch": 0.39564969723710336, + "grad_norm": 1.550518274307251, + "learning_rate": 1.377080254181164e-05, + "loss": 0.1683631420135498, + "step": 15020 + }, + { + "epoch": 0.3959131124816021, + "grad_norm": 1.183109998703003, + "learning_rate": 1.3762899578782662e-05, + "loss": 0.16866258382797242, + "step": 15030 + }, + { + "epoch": 0.39617652772610085, + "grad_norm": 3.876025915145874, + "learning_rate": 1.375499387695303e-05, + "loss": 0.15990071296691893, + "step": 15040 + }, + { + "epoch": 0.3964399429705996, + "grad_norm": 1.6096203327178955, + "learning_rate": 1.3747085442076844e-05, + "loss": 0.15472798347473143, + "step": 15050 + }, + { + "epoch": 0.3967033582150983, + "grad_norm": 0.8998991847038269, + "learning_rate": 1.3739174279910214e-05, + "loss": 0.15954914093017578, + "step": 15060 + }, + { + "epoch": 0.396966773459597, + "grad_norm": 1.7496753931045532, + "learning_rate": 1.3731260396211219e-05, + "loss": 0.14827136993408202, + "step": 15070 + }, + { + "epoch": 0.39723018870409577, + "grad_norm": 3.5599164962768555, + "learning_rate": 1.372334379673993e-05, + "loss": 0.16268184185028076, + "step": 15080 + }, + { + "epoch": 0.3974936039485945, + "grad_norm": 0.7996748685836792, + "learning_rate": 1.3715424487258386e-05, + "loss": 0.15387920141220093, + "step": 15090 + }, + { + "epoch": 0.39775701919309325, + "grad_norm": 2.931562662124634, + "learning_rate": 1.37075024735306e-05, + "loss": 0.17519910335540773, + "step": 15100 + }, + { + "epoch": 0.398020434437592, + "grad_norm": 1.430924654006958, + "learning_rate": 1.3699577761322563e-05, + "loss": 0.16185038089752196, + "step": 15110 + }, + { + "epoch": 0.39828384968209074, + "grad_norm": 1.3743400573730469, + "learning_rate": 1.3691650356402212e-05, + "loss": 0.16592153310775756, + "step": 15120 + }, + { + "epoch": 0.3985472649265895, + "grad_norm": 1.120642066001892, + "learning_rate": 1.3683720264539469e-05, + "loss": 0.14836671352386474, + "step": 15130 + }, + { + "epoch": 0.3988106801710882, + "grad_norm": 2.320157051086426, + "learning_rate": 1.3675787491506184e-05, + "loss": 0.15730634927749634, + "step": 15140 + }, + { + "epoch": 0.3990740954155869, + "grad_norm": 1.1405949592590332, + "learning_rate": 1.3667852043076173e-05, + "loss": 0.15911877155303955, + "step": 15150 + }, + { + "epoch": 0.39933751066008566, + "grad_norm": 2.946908950805664, + "learning_rate": 1.3659913925025203e-05, + "loss": 0.16931018829345704, + "step": 15160 + }, + { + "epoch": 0.3996009259045844, + "grad_norm": 3.2568747997283936, + "learning_rate": 1.3651973143130979e-05, + "loss": 0.15586941242218016, + "step": 15170 + }, + { + "epoch": 0.39986434114908315, + "grad_norm": 2.1314356327056885, + "learning_rate": 1.3644029703173139e-05, + "loss": 0.15220351219177247, + "step": 15180 + }, + { + "epoch": 0.4001277563935819, + "grad_norm": 1.4562536478042603, + "learning_rate": 1.3636083610933262e-05, + "loss": 0.15674118995666503, + "step": 15190 + }, + { + "epoch": 0.40039117163808063, + "grad_norm": 2.6970291137695312, + "learning_rate": 1.3628134872194861e-05, + "loss": 0.16037545204162598, + "step": 15200 + }, + { + "epoch": 0.4006545868825794, + "grad_norm": 1.5283162593841553, + "learning_rate": 1.362018349274337e-05, + "loss": 0.15105776786804198, + "step": 15210 + }, + { + "epoch": 0.4009180021270781, + "grad_norm": 2.7593142986297607, + "learning_rate": 1.3612229478366144e-05, + "loss": 0.15022321939468383, + "step": 15220 + }, + { + "epoch": 0.4011814173715768, + "grad_norm": 2.4657273292541504, + "learning_rate": 1.360427283485246e-05, + "loss": 0.16150012016296386, + "step": 15230 + }, + { + "epoch": 0.40144483261607555, + "grad_norm": 1.5177191495895386, + "learning_rate": 1.3596313567993504e-05, + "loss": 0.16362261772155762, + "step": 15240 + }, + { + "epoch": 0.4017082478605743, + "grad_norm": 1.2831443548202515, + "learning_rate": 1.3588351683582374e-05, + "loss": 0.14856951236724852, + "step": 15250 + }, + { + "epoch": 0.40197166310507304, + "grad_norm": 1.495530605316162, + "learning_rate": 1.3580387187414077e-05, + "loss": 0.15219683647155763, + "step": 15260 + }, + { + "epoch": 0.4022350783495718, + "grad_norm": 2.2346324920654297, + "learning_rate": 1.3572420085285512e-05, + "loss": 0.1517975926399231, + "step": 15270 + }, + { + "epoch": 0.4024984935940705, + "grad_norm": 1.7088243961334229, + "learning_rate": 1.3564450382995482e-05, + "loss": 0.16070547103881835, + "step": 15280 + }, + { + "epoch": 0.40276190883856927, + "grad_norm": 1.3421777486801147, + "learning_rate": 1.3556478086344678e-05, + "loss": 0.15042333602905272, + "step": 15290 + }, + { + "epoch": 0.403025324083068, + "grad_norm": 3.779463291168213, + "learning_rate": 1.3548503201135687e-05, + "loss": 0.13534078598022461, + "step": 15300 + }, + { + "epoch": 0.40328873932756676, + "grad_norm": 1.4087412357330322, + "learning_rate": 1.3540525733172966e-05, + "loss": 0.16739248037338256, + "step": 15310 + }, + { + "epoch": 0.40355215457206545, + "grad_norm": 0.9496793150901794, + "learning_rate": 1.353254568826287e-05, + "loss": 0.1592487096786499, + "step": 15320 + }, + { + "epoch": 0.4038155698165642, + "grad_norm": 1.9337328672409058, + "learning_rate": 1.3524563072213612e-05, + "loss": 0.1609489917755127, + "step": 15330 + }, + { + "epoch": 0.40407898506106293, + "grad_norm": 2.4506397247314453, + "learning_rate": 1.3516577890835292e-05, + "loss": 0.15960073471069336, + "step": 15340 + }, + { + "epoch": 0.4043424003055617, + "grad_norm": 1.0813474655151367, + "learning_rate": 1.350859014993986e-05, + "loss": 0.16202502250671386, + "step": 15350 + }, + { + "epoch": 0.4046058155500604, + "grad_norm": 1.1117557287216187, + "learning_rate": 1.3500599855341146e-05, + "loss": 0.1518958568572998, + "step": 15360 + }, + { + "epoch": 0.40486923079455917, + "grad_norm": 2.825270414352417, + "learning_rate": 1.3492607012854824e-05, + "loss": 0.16900527477264404, + "step": 15370 + }, + { + "epoch": 0.4051326460390579, + "grad_norm": 1.4888981580734253, + "learning_rate": 1.3484611628298438e-05, + "loss": 0.1506636381149292, + "step": 15380 + }, + { + "epoch": 0.40539606128355665, + "grad_norm": 3.4700260162353516, + "learning_rate": 1.3476613707491372e-05, + "loss": 0.1539484977722168, + "step": 15390 + }, + { + "epoch": 0.4056594765280554, + "grad_norm": 0.937309741973877, + "learning_rate": 1.3468613256254852e-05, + "loss": 0.14905575513839722, + "step": 15400 + }, + { + "epoch": 0.4059228917725541, + "grad_norm": 1.22142493724823, + "learning_rate": 1.3460610280411957e-05, + "loss": 0.1649417757987976, + "step": 15410 + }, + { + "epoch": 0.40618630701705283, + "grad_norm": 2.7011523246765137, + "learning_rate": 1.3452604785787595e-05, + "loss": 0.1561107635498047, + "step": 15420 + }, + { + "epoch": 0.40644972226155157, + "grad_norm": 2.3537020683288574, + "learning_rate": 1.3444596778208516e-05, + "loss": 0.16336412429809571, + "step": 15430 + }, + { + "epoch": 0.4067131375060503, + "grad_norm": 1.8011703491210938, + "learning_rate": 1.3436586263503286e-05, + "loss": 0.15739808082580567, + "step": 15440 + }, + { + "epoch": 0.40697655275054906, + "grad_norm": 2.1092655658721924, + "learning_rate": 1.3428573247502307e-05, + "loss": 0.16967337131500243, + "step": 15450 + }, + { + "epoch": 0.4072399679950478, + "grad_norm": 3.597790002822876, + "learning_rate": 1.3420557736037796e-05, + "loss": 0.16125917434692383, + "step": 15460 + }, + { + "epoch": 0.40750338323954655, + "grad_norm": 0.811501681804657, + "learning_rate": 1.3412539734943792e-05, + "loss": 0.14117342233657837, + "step": 15470 + }, + { + "epoch": 0.4077667984840453, + "grad_norm": 1.1118788719177246, + "learning_rate": 1.3404519250056139e-05, + "loss": 0.15143053531646727, + "step": 15480 + }, + { + "epoch": 0.408030213728544, + "grad_norm": 2.5886106491088867, + "learning_rate": 1.3396496287212491e-05, + "loss": 0.15183370113372802, + "step": 15490 + }, + { + "epoch": 0.4082936289730427, + "grad_norm": 2.855050563812256, + "learning_rate": 1.3388470852252307e-05, + "loss": 0.16254086494445802, + "step": 15500 + }, + { + "epoch": 0.40855704421754147, + "grad_norm": 0.966523289680481, + "learning_rate": 1.3380442951016846e-05, + "loss": 0.14198529720306396, + "step": 15510 + }, + { + "epoch": 0.4088204594620402, + "grad_norm": 2.8404319286346436, + "learning_rate": 1.3372412589349164e-05, + "loss": 0.15785501003265381, + "step": 15520 + }, + { + "epoch": 0.40908387470653895, + "grad_norm": 2.675081968307495, + "learning_rate": 1.3364379773094099e-05, + "loss": 0.16033878326416015, + "step": 15530 + }, + { + "epoch": 0.4093472899510377, + "grad_norm": 2.603215456008911, + "learning_rate": 1.335634450809828e-05, + "loss": 0.16615533828735352, + "step": 15540 + }, + { + "epoch": 0.40961070519553644, + "grad_norm": 1.0019960403442383, + "learning_rate": 1.3348306800210127e-05, + "loss": 0.1619364619255066, + "step": 15550 + }, + { + "epoch": 0.4098741204400352, + "grad_norm": 1.4611668586730957, + "learning_rate": 1.3340266655279828e-05, + "loss": 0.15746042728424073, + "step": 15560 + }, + { + "epoch": 0.4101375356845339, + "grad_norm": 1.0391004085540771, + "learning_rate": 1.3332224079159342e-05, + "loss": 0.15804357528686525, + "step": 15570 + }, + { + "epoch": 0.4104009509290326, + "grad_norm": 0.6960294842720032, + "learning_rate": 1.3324179077702408e-05, + "loss": 0.16233930587768555, + "step": 15580 + }, + { + "epoch": 0.41066436617353136, + "grad_norm": 3.5685510635375977, + "learning_rate": 1.3316131656764523e-05, + "loss": 0.14220736026763917, + "step": 15590 + }, + { + "epoch": 0.4109277814180301, + "grad_norm": 1.0885134935379028, + "learning_rate": 1.3308081822202953e-05, + "loss": 0.16248902082443237, + "step": 15600 + }, + { + "epoch": 0.41119119666252885, + "grad_norm": 3.461158275604248, + "learning_rate": 1.3300029579876705e-05, + "loss": 0.16798219680786133, + "step": 15610 + }, + { + "epoch": 0.4114546119070276, + "grad_norm": 1.3515574932098389, + "learning_rate": 1.3291974935646554e-05, + "loss": 0.15843404531478883, + "step": 15620 + }, + { + "epoch": 0.41171802715152633, + "grad_norm": 1.1590218544006348, + "learning_rate": 1.3283917895375018e-05, + "loss": 0.15248254537582398, + "step": 15630 + }, + { + "epoch": 0.4119814423960251, + "grad_norm": 1.4552500247955322, + "learning_rate": 1.327585846492636e-05, + "loss": 0.1492213010787964, + "step": 15640 + }, + { + "epoch": 0.4122448576405238, + "grad_norm": 2.813021183013916, + "learning_rate": 1.3267796650166576e-05, + "loss": 0.16660749912261963, + "step": 15650 + }, + { + "epoch": 0.41250827288502256, + "grad_norm": 1.1179732084274292, + "learning_rate": 1.3259732456963406e-05, + "loss": 0.1678708553314209, + "step": 15660 + }, + { + "epoch": 0.41277168812952125, + "grad_norm": 3.302461862564087, + "learning_rate": 1.3251665891186313e-05, + "loss": 0.14790809154510498, + "step": 15670 + }, + { + "epoch": 0.41303510337402, + "grad_norm": 1.6907111406326294, + "learning_rate": 1.32435969587065e-05, + "loss": 0.15158390998840332, + "step": 15680 + }, + { + "epoch": 0.41329851861851874, + "grad_norm": 1.1835286617279053, + "learning_rate": 1.323552566539688e-05, + "loss": 0.1584031105041504, + "step": 15690 + }, + { + "epoch": 0.4135619338630175, + "grad_norm": 3.8104629516601562, + "learning_rate": 1.3227452017132085e-05, + "loss": 0.1575599193572998, + "step": 15700 + }, + { + "epoch": 0.4138253491075162, + "grad_norm": 3.516876220703125, + "learning_rate": 1.321937601978847e-05, + "loss": 0.15305399894714355, + "step": 15710 + }, + { + "epoch": 0.41408876435201497, + "grad_norm": 1.4132704734802246, + "learning_rate": 1.3211297679244093e-05, + "loss": 0.1664426565170288, + "step": 15720 + }, + { + "epoch": 0.4143521795965137, + "grad_norm": 1.9237918853759766, + "learning_rate": 1.3203217001378716e-05, + "loss": 0.16904199123382568, + "step": 15730 + }, + { + "epoch": 0.41461559484101246, + "grad_norm": 2.047954559326172, + "learning_rate": 1.3195133992073808e-05, + "loss": 0.15652600526809693, + "step": 15740 + }, + { + "epoch": 0.41487901008551115, + "grad_norm": 2.564643144607544, + "learning_rate": 1.318704865721253e-05, + "loss": 0.1600573778152466, + "step": 15750 + }, + { + "epoch": 0.4151424253300099, + "grad_norm": 2.0069291591644287, + "learning_rate": 1.3178961002679738e-05, + "loss": 0.15994648933410643, + "step": 15760 + }, + { + "epoch": 0.41540584057450863, + "grad_norm": 2.668060541152954, + "learning_rate": 1.317087103436198e-05, + "loss": 0.1704960584640503, + "step": 15770 + }, + { + "epoch": 0.4156692558190074, + "grad_norm": 1.147977590560913, + "learning_rate": 1.3162778758147477e-05, + "loss": 0.15942211151123048, + "step": 15780 + }, + { + "epoch": 0.4159326710635061, + "grad_norm": 1.8006469011306763, + "learning_rate": 1.315468417992614e-05, + "loss": 0.1678765296936035, + "step": 15790 + }, + { + "epoch": 0.41619608630800486, + "grad_norm": 0.9667121767997742, + "learning_rate": 1.3146587305589552e-05, + "loss": 0.16053922176361085, + "step": 15800 + }, + { + "epoch": 0.4164595015525036, + "grad_norm": 1.359053134918213, + "learning_rate": 1.3138488141030967e-05, + "loss": 0.15073554515838622, + "step": 15810 + }, + { + "epoch": 0.41672291679700235, + "grad_norm": 3.7596263885498047, + "learning_rate": 1.313038669214531e-05, + "loss": 0.16381676197052003, + "step": 15820 + }, + { + "epoch": 0.4169863320415011, + "grad_norm": 0.7880494594573975, + "learning_rate": 1.312228296482916e-05, + "loss": 0.15678962469100952, + "step": 15830 + }, + { + "epoch": 0.4172497472859998, + "grad_norm": 0.9280499219894409, + "learning_rate": 1.311417696498076e-05, + "loss": 0.1581402063369751, + "step": 15840 + }, + { + "epoch": 0.4175131625304985, + "grad_norm": 2.0727524757385254, + "learning_rate": 1.310606869850001e-05, + "loss": 0.14034979343414306, + "step": 15850 + }, + { + "epoch": 0.41777657777499727, + "grad_norm": 1.150648832321167, + "learning_rate": 1.3097958171288453e-05, + "loss": 0.155013108253479, + "step": 15860 + }, + { + "epoch": 0.418039993019496, + "grad_norm": 4.710323810577393, + "learning_rate": 1.3089845389249282e-05, + "loss": 0.15226640701293945, + "step": 15870 + }, + { + "epoch": 0.41830340826399476, + "grad_norm": 1.4413129091262817, + "learning_rate": 1.308173035828733e-05, + "loss": 0.16164941787719728, + "step": 15880 + }, + { + "epoch": 0.4185668235084935, + "grad_norm": 2.0450587272644043, + "learning_rate": 1.3073613084309065e-05, + "loss": 0.16129881143569946, + "step": 15890 + }, + { + "epoch": 0.41883023875299225, + "grad_norm": 3.154676675796509, + "learning_rate": 1.3065493573222594e-05, + "loss": 0.16715312004089355, + "step": 15900 + }, + { + "epoch": 0.419093653997491, + "grad_norm": 1.3110663890838623, + "learning_rate": 1.305737183093764e-05, + "loss": 0.15343143939971923, + "step": 15910 + }, + { + "epoch": 0.41935706924198973, + "grad_norm": 1.9772621393203735, + "learning_rate": 1.3049247863365562e-05, + "loss": 0.15402576923370362, + "step": 15920 + }, + { + "epoch": 0.4196204844864884, + "grad_norm": 1.4819034337997437, + "learning_rate": 1.3041121676419333e-05, + "loss": 0.1534045934677124, + "step": 15930 + }, + { + "epoch": 0.41988389973098716, + "grad_norm": 1.760694980621338, + "learning_rate": 1.3032993276013546e-05, + "loss": 0.15160133838653564, + "step": 15940 + }, + { + "epoch": 0.4201473149754859, + "grad_norm": 0.7558428049087524, + "learning_rate": 1.3024862668064393e-05, + "loss": 0.14873344898223878, + "step": 15950 + }, + { + "epoch": 0.42041073021998465, + "grad_norm": 1.0987017154693604, + "learning_rate": 1.3016729858489689e-05, + "loss": 0.1728153944015503, + "step": 15960 + }, + { + "epoch": 0.4206741454644834, + "grad_norm": 1.0380008220672607, + "learning_rate": 1.300859485320884e-05, + "loss": 0.16615829467773438, + "step": 15970 + }, + { + "epoch": 0.42093756070898214, + "grad_norm": 2.8251612186431885, + "learning_rate": 1.3000457658142858e-05, + "loss": 0.1642277717590332, + "step": 15980 + }, + { + "epoch": 0.4212009759534809, + "grad_norm": 2.2946290969848633, + "learning_rate": 1.299231827921434e-05, + "loss": 0.16109743118286132, + "step": 15990 + }, + { + "epoch": 0.4214643911979796, + "grad_norm": 1.3908578157424927, + "learning_rate": 1.2984176722347478e-05, + "loss": 0.15582258701324464, + "step": 16000 + }, + { + "epoch": 0.4217278064424783, + "grad_norm": 0.9212650656700134, + "learning_rate": 1.2976032993468049e-05, + "loss": 0.15686706304550171, + "step": 16010 + }, + { + "epoch": 0.42199122168697706, + "grad_norm": 3.4316463470458984, + "learning_rate": 1.296788709850341e-05, + "loss": 0.14522716999053956, + "step": 16020 + }, + { + "epoch": 0.4222546369314758, + "grad_norm": 1.6338825225830078, + "learning_rate": 1.29597390433825e-05, + "loss": 0.14500095844268798, + "step": 16030 + }, + { + "epoch": 0.42251805217597455, + "grad_norm": 1.304457664489746, + "learning_rate": 1.2951588834035813e-05, + "loss": 0.15137314796447754, + "step": 16040 + }, + { + "epoch": 0.4227814674204733, + "grad_norm": 1.4979201555252075, + "learning_rate": 1.2943436476395435e-05, + "loss": 0.1661139965057373, + "step": 16050 + }, + { + "epoch": 0.42304488266497203, + "grad_norm": 0.9115285873413086, + "learning_rate": 1.2935281976394998e-05, + "loss": 0.16102445125579834, + "step": 16060 + }, + { + "epoch": 0.4233082979094708, + "grad_norm": 3.0328447818756104, + "learning_rate": 1.2927125339969701e-05, + "loss": 0.15149662494659424, + "step": 16070 + }, + { + "epoch": 0.4235717131539695, + "grad_norm": 1.4161286354064941, + "learning_rate": 1.2918966573056291e-05, + "loss": 0.15039018392562867, + "step": 16080 + }, + { + "epoch": 0.42383512839846826, + "grad_norm": 1.7117106914520264, + "learning_rate": 1.2910805681593078e-05, + "loss": 0.16847959756851197, + "step": 16090 + }, + { + "epoch": 0.42409854364296695, + "grad_norm": 0.863828182220459, + "learning_rate": 1.2902642671519903e-05, + "loss": 0.1537746787071228, + "step": 16100 + }, + { + "epoch": 0.4243619588874657, + "grad_norm": 0.8783108592033386, + "learning_rate": 1.2894477548778164e-05, + "loss": 0.1654977321624756, + "step": 16110 + }, + { + "epoch": 0.42462537413196444, + "grad_norm": 0.8521326780319214, + "learning_rate": 1.2886310319310787e-05, + "loss": 0.15521103143692017, + "step": 16120 + }, + { + "epoch": 0.4248887893764632, + "grad_norm": 1.7508410215377808, + "learning_rate": 1.2878140989062233e-05, + "loss": 0.16662421226501464, + "step": 16130 + }, + { + "epoch": 0.4251522046209619, + "grad_norm": 2.220085382461548, + "learning_rate": 1.2869969563978493e-05, + "loss": 0.1634347677230835, + "step": 16140 + }, + { + "epoch": 0.42541561986546067, + "grad_norm": 2.292022228240967, + "learning_rate": 1.2861796050007082e-05, + "loss": 0.15511910915374755, + "step": 16150 + }, + { + "epoch": 0.4256790351099594, + "grad_norm": 1.6157565116882324, + "learning_rate": 1.2853620453097039e-05, + "loss": 0.17001636028289796, + "step": 16160 + }, + { + "epoch": 0.42594245035445816, + "grad_norm": 1.2734569311141968, + "learning_rate": 1.2845442779198916e-05, + "loss": 0.15546820163726807, + "step": 16170 + }, + { + "epoch": 0.4262058655989569, + "grad_norm": 0.8812159895896912, + "learning_rate": 1.2837263034264773e-05, + "loss": 0.14568672180175782, + "step": 16180 + }, + { + "epoch": 0.4264692808434556, + "grad_norm": 0.9944833517074585, + "learning_rate": 1.2829081224248183e-05, + "loss": 0.1602702856063843, + "step": 16190 + }, + { + "epoch": 0.42673269608795433, + "grad_norm": 2.143244743347168, + "learning_rate": 1.2820897355104224e-05, + "loss": 0.16800901889801026, + "step": 16200 + }, + { + "epoch": 0.4269961113324531, + "grad_norm": 1.4418374300003052, + "learning_rate": 1.2812711432789464e-05, + "loss": 0.16217210292816162, + "step": 16210 + }, + { + "epoch": 0.4272595265769518, + "grad_norm": 1.7578306198120117, + "learning_rate": 1.2804523463261975e-05, + "loss": 0.1475135087966919, + "step": 16220 + }, + { + "epoch": 0.42752294182145056, + "grad_norm": 2.3281915187835693, + "learning_rate": 1.2796333452481309e-05, + "loss": 0.1587705135345459, + "step": 16230 + }, + { + "epoch": 0.4277863570659493, + "grad_norm": 2.3321268558502197, + "learning_rate": 1.2788141406408516e-05, + "loss": 0.1596773386001587, + "step": 16240 + }, + { + "epoch": 0.42804977231044805, + "grad_norm": 0.9681317806243896, + "learning_rate": 1.2779947331006113e-05, + "loss": 0.1565258741378784, + "step": 16250 + }, + { + "epoch": 0.4283131875549468, + "grad_norm": 1.3185616731643677, + "learning_rate": 1.2771751232238107e-05, + "loss": 0.14218628406524658, + "step": 16260 + }, + { + "epoch": 0.42857660279944554, + "grad_norm": 4.543839454650879, + "learning_rate": 1.2763553116069969e-05, + "loss": 0.1536334991455078, + "step": 16270 + }, + { + "epoch": 0.4288400180439442, + "grad_norm": 0.9844002723693848, + "learning_rate": 1.2755352988468646e-05, + "loss": 0.14347877502441406, + "step": 16280 + }, + { + "epoch": 0.42910343328844297, + "grad_norm": 2.0776889324188232, + "learning_rate": 1.2747150855402541e-05, + "loss": 0.16178817749023439, + "step": 16290 + }, + { + "epoch": 0.4293668485329417, + "grad_norm": 1.1388351917266846, + "learning_rate": 1.2738946722841519e-05, + "loss": 0.15092459917068482, + "step": 16300 + }, + { + "epoch": 0.42963026377744046, + "grad_norm": 3.952554225921631, + "learning_rate": 1.2730740596756908e-05, + "loss": 0.14522212743759155, + "step": 16310 + }, + { + "epoch": 0.4298936790219392, + "grad_norm": 3.792088031768799, + "learning_rate": 1.2722532483121475e-05, + "loss": 0.16409151554107665, + "step": 16320 + }, + { + "epoch": 0.43015709426643794, + "grad_norm": 1.8820555210113525, + "learning_rate": 1.271432238790944e-05, + "loss": 0.15875685214996338, + "step": 16330 + }, + { + "epoch": 0.4304205095109367, + "grad_norm": 2.7368478775024414, + "learning_rate": 1.2706110317096463e-05, + "loss": 0.1524024248123169, + "step": 16340 + }, + { + "epoch": 0.43068392475543543, + "grad_norm": 3.729574680328369, + "learning_rate": 1.269789627665965e-05, + "loss": 0.16900596618652344, + "step": 16350 + }, + { + "epoch": 0.4309473399999341, + "grad_norm": 1.8489389419555664, + "learning_rate": 1.2689680272577527e-05, + "loss": 0.17208330631256102, + "step": 16360 + }, + { + "epoch": 0.43121075524443286, + "grad_norm": 1.7357921600341797, + "learning_rate": 1.2681462310830061e-05, + "loss": 0.15926226377487182, + "step": 16370 + }, + { + "epoch": 0.4314741704889316, + "grad_norm": 2.8678689002990723, + "learning_rate": 1.2673242397398635e-05, + "loss": 0.16326615810394288, + "step": 16380 + }, + { + "epoch": 0.43173758573343035, + "grad_norm": 0.9289158582687378, + "learning_rate": 1.266502053826606e-05, + "loss": 0.1655275583267212, + "step": 16390 + }, + { + "epoch": 0.4320010009779291, + "grad_norm": 3.0142760276794434, + "learning_rate": 1.2656796739416557e-05, + "loss": 0.16683927774429322, + "step": 16400 + }, + { + "epoch": 0.43226441622242784, + "grad_norm": 1.59852933883667, + "learning_rate": 1.2648571006835764e-05, + "loss": 0.14478163719177245, + "step": 16410 + }, + { + "epoch": 0.4325278314669266, + "grad_norm": 1.3471509218215942, + "learning_rate": 1.2640343346510724e-05, + "loss": 0.1537543296813965, + "step": 16420 + }, + { + "epoch": 0.4327912467114253, + "grad_norm": 0.7227351665496826, + "learning_rate": 1.2632113764429884e-05, + "loss": 0.16060515642166137, + "step": 16430 + }, + { + "epoch": 0.43305466195592407, + "grad_norm": 1.3213040828704834, + "learning_rate": 1.2623882266583083e-05, + "loss": 0.1600586175918579, + "step": 16440 + }, + { + "epoch": 0.43331807720042276, + "grad_norm": 1.7010314464569092, + "learning_rate": 1.2615648858961564e-05, + "loss": 0.15424916744232178, + "step": 16450 + }, + { + "epoch": 0.4335814924449215, + "grad_norm": 2.9537482261657715, + "learning_rate": 1.260741354755796e-05, + "loss": 0.16018736362457275, + "step": 16460 + }, + { + "epoch": 0.43384490768942024, + "grad_norm": 1.3403676748275757, + "learning_rate": 1.2599176338366282e-05, + "loss": 0.16938875913619994, + "step": 16470 + }, + { + "epoch": 0.434108322933919, + "grad_norm": 2.3439080715179443, + "learning_rate": 1.2590937237381927e-05, + "loss": 0.14432430267333984, + "step": 16480 + }, + { + "epoch": 0.43437173817841773, + "grad_norm": 0.9049398899078369, + "learning_rate": 1.2582696250601667e-05, + "loss": 0.16370956897735595, + "step": 16490 + }, + { + "epoch": 0.4346351534229165, + "grad_norm": 2.135629653930664, + "learning_rate": 1.257445338402365e-05, + "loss": 0.15900068283081054, + "step": 16500 + }, + { + "epoch": 0.4348985686674152, + "grad_norm": 0.5737161040306091, + "learning_rate": 1.256620864364739e-05, + "loss": 0.16476501226425172, + "step": 16510 + }, + { + "epoch": 0.43516198391191396, + "grad_norm": 1.0121902227401733, + "learning_rate": 1.255796203547376e-05, + "loss": 0.15439974069595336, + "step": 16520 + }, + { + "epoch": 0.4354253991564127, + "grad_norm": 1.3221663236618042, + "learning_rate": 1.2549713565505001e-05, + "loss": 0.15375421047210694, + "step": 16530 + }, + { + "epoch": 0.4356888144009114, + "grad_norm": 1.6791268587112427, + "learning_rate": 1.254146323974471e-05, + "loss": 0.1510796070098877, + "step": 16540 + }, + { + "epoch": 0.43595222964541014, + "grad_norm": 1.8416796922683716, + "learning_rate": 1.253321106419782e-05, + "loss": 0.14541809558868407, + "step": 16550 + }, + { + "epoch": 0.4362156448899089, + "grad_norm": 0.7232946753501892, + "learning_rate": 1.2524957044870627e-05, + "loss": 0.15708858966827394, + "step": 16560 + }, + { + "epoch": 0.4364790601344076, + "grad_norm": 1.2400181293487549, + "learning_rate": 1.2516701187770761e-05, + "loss": 0.1543187379837036, + "step": 16570 + }, + { + "epoch": 0.43674247537890637, + "grad_norm": 1.5587527751922607, + "learning_rate": 1.250844349890719e-05, + "loss": 0.1592945098876953, + "step": 16580 + }, + { + "epoch": 0.4370058906234051, + "grad_norm": 1.1324422359466553, + "learning_rate": 1.250018398429022e-05, + "loss": 0.16867531538009645, + "step": 16590 + }, + { + "epoch": 0.43726930586790386, + "grad_norm": 4.256040573120117, + "learning_rate": 1.2491922649931475e-05, + "loss": 0.16578010320663453, + "step": 16600 + }, + { + "epoch": 0.4375327211124026, + "grad_norm": 3.3228137493133545, + "learning_rate": 1.2483659501843916e-05, + "loss": 0.1637904405593872, + "step": 16610 + }, + { + "epoch": 0.4377961363569013, + "grad_norm": 0.7832944989204407, + "learning_rate": 1.2475394546041822e-05, + "loss": 0.1458975076675415, + "step": 16620 + }, + { + "epoch": 0.43805955160140003, + "grad_norm": 2.0160279273986816, + "learning_rate": 1.2467127788540774e-05, + "loss": 0.14525007009506224, + "step": 16630 + }, + { + "epoch": 0.4383229668458988, + "grad_norm": 1.3399499654769897, + "learning_rate": 1.245885923535768e-05, + "loss": 0.15708847045898439, + "step": 16640 + }, + { + "epoch": 0.4385863820903975, + "grad_norm": 0.77265465259552, + "learning_rate": 1.2450588892510756e-05, + "loss": 0.15859448909759521, + "step": 16650 + }, + { + "epoch": 0.43884979733489626, + "grad_norm": 1.0252106189727783, + "learning_rate": 1.2442316766019502e-05, + "loss": 0.1608428716659546, + "step": 16660 + }, + { + "epoch": 0.439113212579395, + "grad_norm": 2.726357936859131, + "learning_rate": 1.2434042861904743e-05, + "loss": 0.152876877784729, + "step": 16670 + }, + { + "epoch": 0.43937662782389375, + "grad_norm": 1.408489465713501, + "learning_rate": 1.2425767186188569e-05, + "loss": 0.1503664255142212, + "step": 16680 + }, + { + "epoch": 0.4396400430683925, + "grad_norm": 1.558113932609558, + "learning_rate": 1.2417489744894381e-05, + "loss": 0.1520521879196167, + "step": 16690 + }, + { + "epoch": 0.43990345831289124, + "grad_norm": 2.351691246032715, + "learning_rate": 1.2409210544046859e-05, + "loss": 0.1453655481338501, + "step": 16700 + }, + { + "epoch": 0.4401668735573899, + "grad_norm": 2.3554537296295166, + "learning_rate": 1.2400929589671962e-05, + "loss": 0.15786617994308472, + "step": 16710 + }, + { + "epoch": 0.44043028880188867, + "grad_norm": 3.577662467956543, + "learning_rate": 1.2392646887796924e-05, + "loss": 0.16981258392333984, + "step": 16720 + }, + { + "epoch": 0.4406937040463874, + "grad_norm": 0.8845790028572083, + "learning_rate": 1.2384362444450253e-05, + "loss": 0.15169229507446289, + "step": 16730 + }, + { + "epoch": 0.44095711929088616, + "grad_norm": 0.9211968779563904, + "learning_rate": 1.2376076265661722e-05, + "loss": 0.16699542999267578, + "step": 16740 + }, + { + "epoch": 0.4412205345353849, + "grad_norm": 3.07585072517395, + "learning_rate": 1.2367788357462375e-05, + "loss": 0.15944290161132812, + "step": 16750 + }, + { + "epoch": 0.44148394977988364, + "grad_norm": 2.9037723541259766, + "learning_rate": 1.2359498725884509e-05, + "loss": 0.16058270931243895, + "step": 16760 + }, + { + "epoch": 0.4417473650243824, + "grad_norm": 2.358760356903076, + "learning_rate": 1.2351207376961666e-05, + "loss": 0.1638423204421997, + "step": 16770 + }, + { + "epoch": 0.44201078026888113, + "grad_norm": 1.9604889154434204, + "learning_rate": 1.234291431672866e-05, + "loss": 0.16349823474884034, + "step": 16780 + }, + { + "epoch": 0.4422741955133799, + "grad_norm": 1.5194076299667358, + "learning_rate": 1.2334619551221531e-05, + "loss": 0.15058577060699463, + "step": 16790 + }, + { + "epoch": 0.44253761075787856, + "grad_norm": 2.3775405883789062, + "learning_rate": 1.232632308647757e-05, + "loss": 0.15821993350982666, + "step": 16800 + }, + { + "epoch": 0.4428010260023773, + "grad_norm": 3.1628782749176025, + "learning_rate": 1.2318024928535296e-05, + "loss": 0.15999042987823486, + "step": 16810 + }, + { + "epoch": 0.44306444124687605, + "grad_norm": 2.0952723026275635, + "learning_rate": 1.2309725083434477e-05, + "loss": 0.15628563165664672, + "step": 16820 + }, + { + "epoch": 0.4433278564913748, + "grad_norm": 1.0502691268920898, + "learning_rate": 1.2301423557216088e-05, + "loss": 0.1617302417755127, + "step": 16830 + }, + { + "epoch": 0.44359127173587354, + "grad_norm": 2.2797155380249023, + "learning_rate": 1.2293120355922346e-05, + "loss": 0.14310643672943116, + "step": 16840 + }, + { + "epoch": 0.4438546869803723, + "grad_norm": 0.6413837671279907, + "learning_rate": 1.2284815485596671e-05, + "loss": 0.14379615783691407, + "step": 16850 + }, + { + "epoch": 0.444118102224871, + "grad_norm": 2.1480560302734375, + "learning_rate": 1.2276508952283712e-05, + "loss": 0.1506582975387573, + "step": 16860 + }, + { + "epoch": 0.44438151746936977, + "grad_norm": 3.2240607738494873, + "learning_rate": 1.2268200762029319e-05, + "loss": 0.15275473594665528, + "step": 16870 + }, + { + "epoch": 0.44464493271386846, + "grad_norm": 1.6737252473831177, + "learning_rate": 1.2259890920880555e-05, + "loss": 0.15548573732376098, + "step": 16880 + }, + { + "epoch": 0.4449083479583672, + "grad_norm": 3.1859290599823, + "learning_rate": 1.2251579434885679e-05, + "loss": 0.15575871467590333, + "step": 16890 + }, + { + "epoch": 0.44517176320286594, + "grad_norm": 1.2047410011291504, + "learning_rate": 1.2243266310094146e-05, + "loss": 0.15061795711517334, + "step": 16900 + }, + { + "epoch": 0.4454351784473647, + "grad_norm": 1.7560594081878662, + "learning_rate": 1.2234951552556613e-05, + "loss": 0.15685230493545532, + "step": 16910 + }, + { + "epoch": 0.44569859369186343, + "grad_norm": 1.343847393989563, + "learning_rate": 1.2226635168324916e-05, + "loss": 0.151579749584198, + "step": 16920 + }, + { + "epoch": 0.4459620089363622, + "grad_norm": 1.8434875011444092, + "learning_rate": 1.2218317163452078e-05, + "loss": 0.14654133319854737, + "step": 16930 + }, + { + "epoch": 0.4462254241808609, + "grad_norm": 2.5810189247131348, + "learning_rate": 1.2209997543992301e-05, + "loss": 0.15223758220672606, + "step": 16940 + }, + { + "epoch": 0.44648883942535966, + "grad_norm": 4.089043617248535, + "learning_rate": 1.2201676316000964e-05, + "loss": 0.15677835941314697, + "step": 16950 + }, + { + "epoch": 0.4467522546698584, + "grad_norm": 1.507109522819519, + "learning_rate": 1.2193353485534614e-05, + "loss": 0.1550657033920288, + "step": 16960 + }, + { + "epoch": 0.4470156699143571, + "grad_norm": 1.0712004899978638, + "learning_rate": 1.2185029058650972e-05, + "loss": 0.15033295154571533, + "step": 16970 + }, + { + "epoch": 0.44727908515885584, + "grad_norm": 1.2962602376937866, + "learning_rate": 1.2176703041408908e-05, + "loss": 0.1551538348197937, + "step": 16980 + }, + { + "epoch": 0.4475425004033546, + "grad_norm": 3.1185057163238525, + "learning_rate": 1.2168375439868464e-05, + "loss": 0.1484465479850769, + "step": 16990 + }, + { + "epoch": 0.4478059156478533, + "grad_norm": 0.6280648112297058, + "learning_rate": 1.2160046260090816e-05, + "loss": 0.14259257316589355, + "step": 17000 + }, + { + "epoch": 0.44806933089235207, + "grad_norm": 2.9803130626678467, + "learning_rate": 1.2151715508138313e-05, + "loss": 0.14653512239456176, + "step": 17010 + }, + { + "epoch": 0.4483327461368508, + "grad_norm": 3.13938045501709, + "learning_rate": 1.2143383190074429e-05, + "loss": 0.15497925281524658, + "step": 17020 + }, + { + "epoch": 0.44859616138134956, + "grad_norm": 3.910656213760376, + "learning_rate": 1.2135049311963787e-05, + "loss": 0.16258184909820556, + "step": 17030 + }, + { + "epoch": 0.4488595766258483, + "grad_norm": 1.067094087600708, + "learning_rate": 1.2126713879872138e-05, + "loss": 0.16146260499954224, + "step": 17040 + }, + { + "epoch": 0.44912299187034704, + "grad_norm": 1.0420509576797485, + "learning_rate": 1.2118376899866375e-05, + "loss": 0.17237170934677123, + "step": 17050 + }, + { + "epoch": 0.44938640711484573, + "grad_norm": 4.71498966217041, + "learning_rate": 1.2110038378014512e-05, + "loss": 0.15846014022827148, + "step": 17060 + }, + { + "epoch": 0.4496498223593445, + "grad_norm": 1.3583329916000366, + "learning_rate": 1.210169832038568e-05, + "loss": 0.15201218128204347, + "step": 17070 + }, + { + "epoch": 0.4499132376038432, + "grad_norm": 1.643551230430603, + "learning_rate": 1.2093356733050137e-05, + "loss": 0.1460457682609558, + "step": 17080 + }, + { + "epoch": 0.45017665284834196, + "grad_norm": 3.7497622966766357, + "learning_rate": 1.2085013622079248e-05, + "loss": 0.16421205997467042, + "step": 17090 + }, + { + "epoch": 0.4504400680928407, + "grad_norm": 1.5383836030960083, + "learning_rate": 1.2076668993545491e-05, + "loss": 0.15214041471481324, + "step": 17100 + }, + { + "epoch": 0.45070348333733945, + "grad_norm": 4.063543796539307, + "learning_rate": 1.206832285352244e-05, + "loss": 0.17085912227630615, + "step": 17110 + }, + { + "epoch": 0.4509668985818382, + "grad_norm": 1.5928771495819092, + "learning_rate": 1.2059975208084787e-05, + "loss": 0.15533337593078614, + "step": 17120 + }, + { + "epoch": 0.45123031382633694, + "grad_norm": 3.54727840423584, + "learning_rate": 1.2051626063308299e-05, + "loss": 0.14924774169921876, + "step": 17130 + }, + { + "epoch": 0.4514937290708356, + "grad_norm": 0.9585392475128174, + "learning_rate": 1.2043275425269854e-05, + "loss": 0.15019567012786866, + "step": 17140 + }, + { + "epoch": 0.45175714431533437, + "grad_norm": 1.3212496042251587, + "learning_rate": 1.2034923300047391e-05, + "loss": 0.1542275905609131, + "step": 17150 + }, + { + "epoch": 0.4520205595598331, + "grad_norm": 3.026092529296875, + "learning_rate": 1.2026569693719963e-05, + "loss": 0.14751508235931396, + "step": 17160 + }, + { + "epoch": 0.45228397480433186, + "grad_norm": 1.264483094215393, + "learning_rate": 1.2018214612367673e-05, + "loss": 0.14753332138061523, + "step": 17170 + }, + { + "epoch": 0.4525473900488306, + "grad_norm": 1.0697717666625977, + "learning_rate": 1.200985806207172e-05, + "loss": 0.14955343008041383, + "step": 17180 + }, + { + "epoch": 0.45281080529332934, + "grad_norm": 2.8215856552124023, + "learning_rate": 1.2001500048914354e-05, + "loss": 0.14118080139160155, + "step": 17190 + }, + { + "epoch": 0.4530742205378281, + "grad_norm": 1.002766489982605, + "learning_rate": 1.1993140578978898e-05, + "loss": 0.15804443359375, + "step": 17200 + }, + { + "epoch": 0.45333763578232683, + "grad_norm": 1.551566243171692, + "learning_rate": 1.1984779658349743e-05, + "loss": 0.15658769607543946, + "step": 17210 + }, + { + "epoch": 0.4536010510268256, + "grad_norm": 2.5837044715881348, + "learning_rate": 1.1976417293112318e-05, + "loss": 0.14618022441864015, + "step": 17220 + }, + { + "epoch": 0.45386446627132426, + "grad_norm": 1.0982544422149658, + "learning_rate": 1.1968053489353122e-05, + "loss": 0.14167003631591796, + "step": 17230 + }, + { + "epoch": 0.454127881515823, + "grad_norm": 2.7502872943878174, + "learning_rate": 1.195968825315968e-05, + "loss": 0.15783772468566895, + "step": 17240 + }, + { + "epoch": 0.45439129676032175, + "grad_norm": 2.4946205615997314, + "learning_rate": 1.1951321590620583e-05, + "loss": 0.16078360080718995, + "step": 17250 + }, + { + "epoch": 0.4546547120048205, + "grad_norm": 2.8402481079101562, + "learning_rate": 1.1942953507825446e-05, + "loss": 0.15795955657958985, + "step": 17260 + }, + { + "epoch": 0.45491812724931924, + "grad_norm": 1.3818917274475098, + "learning_rate": 1.1934584010864921e-05, + "loss": 0.15267064571380615, + "step": 17270 + }, + { + "epoch": 0.455181542493818, + "grad_norm": 1.1723880767822266, + "learning_rate": 1.1926213105830683e-05, + "loss": 0.15370278358459472, + "step": 17280 + }, + { + "epoch": 0.4554449577383167, + "grad_norm": 1.5822356939315796, + "learning_rate": 1.1917840798815448e-05, + "loss": 0.17035675048828125, + "step": 17290 + }, + { + "epoch": 0.45570837298281547, + "grad_norm": 1.8395482301712036, + "learning_rate": 1.1909467095912932e-05, + "loss": 0.16685545444488525, + "step": 17300 + }, + { + "epoch": 0.4559717882273142, + "grad_norm": 2.8542943000793457, + "learning_rate": 1.1901092003217885e-05, + "loss": 0.1518098831176758, + "step": 17310 + }, + { + "epoch": 0.4562352034718129, + "grad_norm": 1.5584897994995117, + "learning_rate": 1.1892715526826058e-05, + "loss": 0.16363046169281006, + "step": 17320 + }, + { + "epoch": 0.45649861871631164, + "grad_norm": 0.6702388525009155, + "learning_rate": 1.1884337672834212e-05, + "loss": 0.14870381355285645, + "step": 17330 + }, + { + "epoch": 0.4567620339608104, + "grad_norm": 1.5773581266403198, + "learning_rate": 1.1875958447340108e-05, + "loss": 0.14815051555633546, + "step": 17340 + }, + { + "epoch": 0.45702544920530913, + "grad_norm": 3.651599645614624, + "learning_rate": 1.1867577856442514e-05, + "loss": 0.15149140357971191, + "step": 17350 + }, + { + "epoch": 0.4572888644498079, + "grad_norm": 1.7512463331222534, + "learning_rate": 1.1859195906241182e-05, + "loss": 0.15762484073638916, + "step": 17360 + }, + { + "epoch": 0.4575522796943066, + "grad_norm": 4.566371440887451, + "learning_rate": 1.1850812602836851e-05, + "loss": 0.16118390560150148, + "step": 17370 + }, + { + "epoch": 0.45781569493880536, + "grad_norm": 1.3424749374389648, + "learning_rate": 1.1842427952331263e-05, + "loss": 0.14656128883361816, + "step": 17380 + }, + { + "epoch": 0.4580791101833041, + "grad_norm": 2.1990678310394287, + "learning_rate": 1.183404196082712e-05, + "loss": 0.1627406358718872, + "step": 17390 + }, + { + "epoch": 0.4583425254278028, + "grad_norm": 1.7389707565307617, + "learning_rate": 1.1825654634428111e-05, + "loss": 0.14895224571228027, + "step": 17400 + }, + { + "epoch": 0.45860594067230154, + "grad_norm": 3.420145273208618, + "learning_rate": 1.1817265979238892e-05, + "loss": 0.16065609455108643, + "step": 17410 + }, + { + "epoch": 0.4588693559168003, + "grad_norm": 0.5141645073890686, + "learning_rate": 1.1808876001365093e-05, + "loss": 0.16511306762695313, + "step": 17420 + }, + { + "epoch": 0.459132771161299, + "grad_norm": 4.023585319519043, + "learning_rate": 1.1800484706913294e-05, + "loss": 0.15396008491516114, + "step": 17430 + }, + { + "epoch": 0.45939618640579777, + "grad_norm": 3.257535934448242, + "learning_rate": 1.1792092101991052e-05, + "loss": 0.15420238971710204, + "step": 17440 + }, + { + "epoch": 0.4596596016502965, + "grad_norm": 4.561605930328369, + "learning_rate": 1.1783698192706856e-05, + "loss": 0.15497398376464844, + "step": 17450 + }, + { + "epoch": 0.45992301689479526, + "grad_norm": 2.502244710922241, + "learning_rate": 1.1775302985170161e-05, + "loss": 0.15690728425979614, + "step": 17460 + }, + { + "epoch": 0.460186432139294, + "grad_norm": 1.248986840248108, + "learning_rate": 1.1766906485491361e-05, + "loss": 0.16173287630081176, + "step": 17470 + }, + { + "epoch": 0.46044984738379274, + "grad_norm": 1.0628888607025146, + "learning_rate": 1.175850869978179e-05, + "loss": 0.16524906158447267, + "step": 17480 + }, + { + "epoch": 0.46071326262829143, + "grad_norm": 1.0567646026611328, + "learning_rate": 1.1750109634153719e-05, + "loss": 0.14236325025558472, + "step": 17490 + }, + { + "epoch": 0.4609766778727902, + "grad_norm": 3.739332437515259, + "learning_rate": 1.1741709294720346e-05, + "loss": 0.1432879686355591, + "step": 17500 + }, + { + "epoch": 0.4612400931172889, + "grad_norm": 3.5751266479492188, + "learning_rate": 1.173330768759581e-05, + "loss": 0.15466618537902832, + "step": 17510 + }, + { + "epoch": 0.46150350836178766, + "grad_norm": 2.145519733428955, + "learning_rate": 1.1724904818895158e-05, + "loss": 0.14844638109207153, + "step": 17520 + }, + { + "epoch": 0.4617669236062864, + "grad_norm": 1.7272312641143799, + "learning_rate": 1.1716500694734359e-05, + "loss": 0.16101584434509278, + "step": 17530 + }, + { + "epoch": 0.46203033885078515, + "grad_norm": 1.141391396522522, + "learning_rate": 1.1708095321230295e-05, + "loss": 0.15032494068145752, + "step": 17540 + }, + { + "epoch": 0.4622937540952839, + "grad_norm": 2.0215003490448, + "learning_rate": 1.1699688704500765e-05, + "loss": 0.14555583000183106, + "step": 17550 + }, + { + "epoch": 0.46255716933978264, + "grad_norm": 1.7067536115646362, + "learning_rate": 1.1691280850664463e-05, + "loss": 0.15407795906066896, + "step": 17560 + }, + { + "epoch": 0.4628205845842814, + "grad_norm": 1.5726145505905151, + "learning_rate": 1.1682871765840996e-05, + "loss": 0.15440326929092407, + "step": 17570 + }, + { + "epoch": 0.46308399982878007, + "grad_norm": 2.0750601291656494, + "learning_rate": 1.1674461456150845e-05, + "loss": 0.15966498851776123, + "step": 17580 + }, + { + "epoch": 0.4633474150732788, + "grad_norm": 3.0505483150482178, + "learning_rate": 1.1666049927715408e-05, + "loss": 0.15533552169799805, + "step": 17590 + }, + { + "epoch": 0.46361083031777756, + "grad_norm": 1.8854368925094604, + "learning_rate": 1.1657637186656952e-05, + "loss": 0.14425480365753174, + "step": 17600 + }, + { + "epoch": 0.4638742455622763, + "grad_norm": 1.4432621002197266, + "learning_rate": 1.1649223239098639e-05, + "loss": 0.16128876209259033, + "step": 17610 + }, + { + "epoch": 0.46413766080677504, + "grad_norm": 3.0689659118652344, + "learning_rate": 1.1640808091164495e-05, + "loss": 0.15863369703292846, + "step": 17620 + }, + { + "epoch": 0.4644010760512738, + "grad_norm": 2.287428379058838, + "learning_rate": 1.1632391748979435e-05, + "loss": 0.14503343105316163, + "step": 17630 + }, + { + "epoch": 0.46466449129577253, + "grad_norm": 1.1321524381637573, + "learning_rate": 1.162397421866923e-05, + "loss": 0.15569480657577514, + "step": 17640 + }, + { + "epoch": 0.4649279065402713, + "grad_norm": 2.243319272994995, + "learning_rate": 1.1615555506360526e-05, + "loss": 0.15334529876708985, + "step": 17650 + }, + { + "epoch": 0.46519132178476996, + "grad_norm": 1.2475773096084595, + "learning_rate": 1.1607135618180823e-05, + "loss": 0.1661982297897339, + "step": 17660 + }, + { + "epoch": 0.4654547370292687, + "grad_norm": 1.1676806211471558, + "learning_rate": 1.1598714560258477e-05, + "loss": 0.15314946174621583, + "step": 17670 + }, + { + "epoch": 0.46571815227376745, + "grad_norm": 2.0490949153900146, + "learning_rate": 1.1590292338722702e-05, + "loss": 0.1592707872390747, + "step": 17680 + }, + { + "epoch": 0.4659815675182662, + "grad_norm": 0.9818809032440186, + "learning_rate": 1.1581868959703553e-05, + "loss": 0.14878494739532472, + "step": 17690 + }, + { + "epoch": 0.46624498276276494, + "grad_norm": 3.9627227783203125, + "learning_rate": 1.1573444429331926e-05, + "loss": 0.14982779026031495, + "step": 17700 + }, + { + "epoch": 0.4665083980072637, + "grad_norm": 1.262900948524475, + "learning_rate": 1.1565018753739559e-05, + "loss": 0.15341397523880004, + "step": 17710 + }, + { + "epoch": 0.4667718132517624, + "grad_norm": 0.9258854985237122, + "learning_rate": 1.1556591939059024e-05, + "loss": 0.15082868337631225, + "step": 17720 + }, + { + "epoch": 0.46703522849626117, + "grad_norm": 4.20810604095459, + "learning_rate": 1.1548163991423716e-05, + "loss": 0.1485476016998291, + "step": 17730 + }, + { + "epoch": 0.4672986437407599, + "grad_norm": 1.3817286491394043, + "learning_rate": 1.1539734916967864e-05, + "loss": 0.14985930919647217, + "step": 17740 + }, + { + "epoch": 0.4675620589852586, + "grad_norm": 1.1596143245697021, + "learning_rate": 1.1531304721826506e-05, + "loss": 0.14508838653564454, + "step": 17750 + }, + { + "epoch": 0.46782547422975734, + "grad_norm": 2.9170081615448, + "learning_rate": 1.1522873412135509e-05, + "loss": 0.14976389408111573, + "step": 17760 + }, + { + "epoch": 0.4680888894742561, + "grad_norm": 1.7168699502944946, + "learning_rate": 1.1514440994031536e-05, + "loss": 0.14381440877914428, + "step": 17770 + }, + { + "epoch": 0.46835230471875483, + "grad_norm": 0.821213960647583, + "learning_rate": 1.1506007473652075e-05, + "loss": 0.15619027614593506, + "step": 17780 + }, + { + "epoch": 0.4686157199632536, + "grad_norm": 4.671971797943115, + "learning_rate": 1.1497572857135397e-05, + "loss": 0.15569875240325928, + "step": 17790 + }, + { + "epoch": 0.4688791352077523, + "grad_norm": 3.3345112800598145, + "learning_rate": 1.1489137150620582e-05, + "loss": 0.16098741292953492, + "step": 17800 + }, + { + "epoch": 0.46914255045225106, + "grad_norm": 0.8613945245742798, + "learning_rate": 1.1480700360247504e-05, + "loss": 0.15666470527648926, + "step": 17810 + }, + { + "epoch": 0.4694059656967498, + "grad_norm": 2.4395601749420166, + "learning_rate": 1.147226249215682e-05, + "loss": 0.16322462558746337, + "step": 17820 + }, + { + "epoch": 0.46966938094124855, + "grad_norm": 1.5228322744369507, + "learning_rate": 1.1463823552489977e-05, + "loss": 0.16073241233825683, + "step": 17830 + }, + { + "epoch": 0.46993279618574724, + "grad_norm": 2.66923451423645, + "learning_rate": 1.1455383547389193e-05, + "loss": 0.15022573471069336, + "step": 17840 + }, + { + "epoch": 0.470196211430246, + "grad_norm": 2.042550563812256, + "learning_rate": 1.1446942482997479e-05, + "loss": 0.14393603801727295, + "step": 17850 + }, + { + "epoch": 0.4704596266747447, + "grad_norm": 1.1591393947601318, + "learning_rate": 1.1438500365458593e-05, + "loss": 0.14727914333343506, + "step": 17860 + }, + { + "epoch": 0.47072304191924347, + "grad_norm": 1.6289873123168945, + "learning_rate": 1.1430057200917084e-05, + "loss": 0.1541304588317871, + "step": 17870 + }, + { + "epoch": 0.4709864571637422, + "grad_norm": 2.1480634212493896, + "learning_rate": 1.1421612995518243e-05, + "loss": 0.15001378059387208, + "step": 17880 + }, + { + "epoch": 0.47124987240824096, + "grad_norm": 1.4128506183624268, + "learning_rate": 1.1413167755408132e-05, + "loss": 0.16379134654998778, + "step": 17890 + }, + { + "epoch": 0.4715132876527397, + "grad_norm": 2.093247652053833, + "learning_rate": 1.1404721486733558e-05, + "loss": 0.15469896793365479, + "step": 17900 + }, + { + "epoch": 0.47177670289723844, + "grad_norm": 1.6161643266677856, + "learning_rate": 1.1396274195642085e-05, + "loss": 0.14906933307647705, + "step": 17910 + }, + { + "epoch": 0.47204011814173713, + "grad_norm": 1.948172688484192, + "learning_rate": 1.1387825888282006e-05, + "loss": 0.15030336380004883, + "step": 17920 + }, + { + "epoch": 0.4723035333862359, + "grad_norm": 1.3184270858764648, + "learning_rate": 1.1379376570802373e-05, + "loss": 0.13992198705673217, + "step": 17930 + }, + { + "epoch": 0.4725669486307346, + "grad_norm": 0.8461310863494873, + "learning_rate": 1.137092624935296e-05, + "loss": 0.149918532371521, + "step": 17940 + }, + { + "epoch": 0.47283036387523336, + "grad_norm": 3.919799566268921, + "learning_rate": 1.1362474930084278e-05, + "loss": 0.17018002271652222, + "step": 17950 + }, + { + "epoch": 0.4730937791197321, + "grad_norm": 0.9279007911682129, + "learning_rate": 1.1354022619147557e-05, + "loss": 0.14624167680740358, + "step": 17960 + }, + { + "epoch": 0.47335719436423085, + "grad_norm": 1.9835418462753296, + "learning_rate": 1.1345569322694756e-05, + "loss": 0.15407028198242187, + "step": 17970 + }, + { + "epoch": 0.4736206096087296, + "grad_norm": 3.4828622341156006, + "learning_rate": 1.133711504687855e-05, + "loss": 0.16472094058990477, + "step": 17980 + }, + { + "epoch": 0.47388402485322834, + "grad_norm": 1.0894211530685425, + "learning_rate": 1.1328659797852325e-05, + "loss": 0.14639554023742676, + "step": 17990 + }, + { + "epoch": 0.4741474400977271, + "grad_norm": 2.632457971572876, + "learning_rate": 1.1320203581770177e-05, + "loss": 0.16240501403808594, + "step": 18000 + }, + { + "epoch": 0.47441085534222577, + "grad_norm": 4.44385290145874, + "learning_rate": 1.1311746404786903e-05, + "loss": 0.16344332695007324, + "step": 18010 + }, + { + "epoch": 0.4746742705867245, + "grad_norm": 2.7959907054901123, + "learning_rate": 1.1303288273058002e-05, + "loss": 0.1472707748413086, + "step": 18020 + }, + { + "epoch": 0.47493768583122326, + "grad_norm": 1.6029959917068481, + "learning_rate": 1.1294829192739665e-05, + "loss": 0.1504539966583252, + "step": 18030 + }, + { + "epoch": 0.475201101075722, + "grad_norm": 0.6915399432182312, + "learning_rate": 1.1286369169988782e-05, + "loss": 0.16439175605773926, + "step": 18040 + }, + { + "epoch": 0.47546451632022074, + "grad_norm": 1.626025676727295, + "learning_rate": 1.1277908210962916e-05, + "loss": 0.16345667839050293, + "step": 18050 + }, + { + "epoch": 0.4757279315647195, + "grad_norm": 3.152981758117676, + "learning_rate": 1.126944632182032e-05, + "loss": 0.15254082679748535, + "step": 18060 + }, + { + "epoch": 0.47599134680921823, + "grad_norm": 1.559854507446289, + "learning_rate": 1.1260983508719922e-05, + "loss": 0.14053142070770264, + "step": 18070 + }, + { + "epoch": 0.476254762053717, + "grad_norm": 2.292491912841797, + "learning_rate": 1.1252519777821324e-05, + "loss": 0.1566366195678711, + "step": 18080 + }, + { + "epoch": 0.4765181772982157, + "grad_norm": 3.350506544113159, + "learning_rate": 1.1244055135284795e-05, + "loss": 0.154224693775177, + "step": 18090 + }, + { + "epoch": 0.4767815925427144, + "grad_norm": 1.4111663103103638, + "learning_rate": 1.123558958727126e-05, + "loss": 0.15744714736938475, + "step": 18100 + }, + { + "epoch": 0.47704500778721315, + "grad_norm": 2.1482930183410645, + "learning_rate": 1.1227123139942322e-05, + "loss": 0.14814527034759523, + "step": 18110 + }, + { + "epoch": 0.4773084230317119, + "grad_norm": 0.9631815552711487, + "learning_rate": 1.1218655799460216e-05, + "loss": 0.15726189613342284, + "step": 18120 + }, + { + "epoch": 0.47757183827621064, + "grad_norm": 1.4226658344268799, + "learning_rate": 1.1210187571987845e-05, + "loss": 0.15461375713348388, + "step": 18130 + }, + { + "epoch": 0.4778352535207094, + "grad_norm": 4.593253135681152, + "learning_rate": 1.1201718463688745e-05, + "loss": 0.16165547370910643, + "step": 18140 + }, + { + "epoch": 0.4780986687652081, + "grad_norm": 1.7888740301132202, + "learning_rate": 1.1193248480727101e-05, + "loss": 0.1572542667388916, + "step": 18150 + }, + { + "epoch": 0.47836208400970687, + "grad_norm": 1.6417293548583984, + "learning_rate": 1.118477762926773e-05, + "loss": 0.1422085165977478, + "step": 18160 + }, + { + "epoch": 0.4786254992542056, + "grad_norm": 2.136979818344116, + "learning_rate": 1.117630591547609e-05, + "loss": 0.15954264402389526, + "step": 18170 + }, + { + "epoch": 0.4788889144987043, + "grad_norm": 1.712295413017273, + "learning_rate": 1.1167833345518248e-05, + "loss": 0.1449791431427002, + "step": 18180 + }, + { + "epoch": 0.47915232974320304, + "grad_norm": 1.8807224035263062, + "learning_rate": 1.1159359925560918e-05, + "loss": 0.15322797298431395, + "step": 18190 + }, + { + "epoch": 0.4794157449877018, + "grad_norm": 4.083950519561768, + "learning_rate": 1.115088566177141e-05, + "loss": 0.15588765144348143, + "step": 18200 + }, + { + "epoch": 0.47967916023220053, + "grad_norm": 1.9321285486221313, + "learning_rate": 1.1142410560317669e-05, + "loss": 0.15532091856002808, + "step": 18210 + }, + { + "epoch": 0.4799425754766993, + "grad_norm": 1.5147652626037598, + "learning_rate": 1.113393462736823e-05, + "loss": 0.16533432006835938, + "step": 18220 + }, + { + "epoch": 0.480205990721198, + "grad_norm": 1.1109064817428589, + "learning_rate": 1.1125457869092251e-05, + "loss": 0.1578255772590637, + "step": 18230 + }, + { + "epoch": 0.48046940596569676, + "grad_norm": 1.3178790807724, + "learning_rate": 1.1116980291659474e-05, + "loss": 0.15549148321151735, + "step": 18240 + }, + { + "epoch": 0.4807328212101955, + "grad_norm": 2.115896701812744, + "learning_rate": 1.1108501901240254e-05, + "loss": 0.14846646785736084, + "step": 18250 + }, + { + "epoch": 0.48099623645469425, + "grad_norm": 1.7174875736236572, + "learning_rate": 1.1100022704005524e-05, + "loss": 0.15033965110778807, + "step": 18260 + }, + { + "epoch": 0.48125965169919294, + "grad_norm": 3.597693681716919, + "learning_rate": 1.109154270612681e-05, + "loss": 0.1529923915863037, + "step": 18270 + }, + { + "epoch": 0.4815230669436917, + "grad_norm": 1.17400324344635, + "learning_rate": 1.1083061913776224e-05, + "loss": 0.14264774322509766, + "step": 18280 + }, + { + "epoch": 0.4817864821881904, + "grad_norm": 2.8394477367401123, + "learning_rate": 1.1074580333126448e-05, + "loss": 0.15206401348114013, + "step": 18290 + }, + { + "epoch": 0.48204989743268917, + "grad_norm": 0.9735348224639893, + "learning_rate": 1.1066097970350747e-05, + "loss": 0.15296398401260375, + "step": 18300 + }, + { + "epoch": 0.4823133126771879, + "grad_norm": 0.9847986102104187, + "learning_rate": 1.1057614831622942e-05, + "loss": 0.1451619029045105, + "step": 18310 + }, + { + "epoch": 0.48257672792168665, + "grad_norm": 0.7013766765594482, + "learning_rate": 1.1049130923117437e-05, + "loss": 0.15064951181411743, + "step": 18320 + }, + { + "epoch": 0.4828401431661854, + "grad_norm": 1.091356873512268, + "learning_rate": 1.1040646251009181e-05, + "loss": 0.15780696868896485, + "step": 18330 + }, + { + "epoch": 0.48310355841068414, + "grad_norm": 1.275473713874817, + "learning_rate": 1.1032160821473692e-05, + "loss": 0.1584775447845459, + "step": 18340 + }, + { + "epoch": 0.4833669736551829, + "grad_norm": 3.483849048614502, + "learning_rate": 1.1023674640687019e-05, + "loss": 0.15945882797241212, + "step": 18350 + }, + { + "epoch": 0.4836303888996816, + "grad_norm": 1.1538023948669434, + "learning_rate": 1.1015187714825782e-05, + "loss": 0.15324833393096923, + "step": 18360 + }, + { + "epoch": 0.4838938041441803, + "grad_norm": 0.963645339012146, + "learning_rate": 1.1006700050067122e-05, + "loss": 0.1378652572631836, + "step": 18370 + }, + { + "epoch": 0.48415721938867906, + "grad_norm": 1.8987176418304443, + "learning_rate": 1.0998211652588737e-05, + "loss": 0.1459387421607971, + "step": 18380 + }, + { + "epoch": 0.4844206346331778, + "grad_norm": 1.9667937755584717, + "learning_rate": 1.0989722528568845e-05, + "loss": 0.15722167491912842, + "step": 18390 + }, + { + "epoch": 0.48468404987767655, + "grad_norm": 2.7628750801086426, + "learning_rate": 1.0981232684186191e-05, + "loss": 0.1589844346046448, + "step": 18400 + }, + { + "epoch": 0.4849474651221753, + "grad_norm": 1.5833640098571777, + "learning_rate": 1.0972742125620062e-05, + "loss": 0.152923321723938, + "step": 18410 + }, + { + "epoch": 0.48521088036667404, + "grad_norm": 1.7823517322540283, + "learning_rate": 1.0964250859050245e-05, + "loss": 0.15118865966796874, + "step": 18420 + }, + { + "epoch": 0.4854742956111728, + "grad_norm": 1.1554049253463745, + "learning_rate": 1.0955758890657051e-05, + "loss": 0.14468057155609132, + "step": 18430 + }, + { + "epoch": 0.48573771085567147, + "grad_norm": 3.3886289596557617, + "learning_rate": 1.0947266226621298e-05, + "loss": 0.13269903659820556, + "step": 18440 + }, + { + "epoch": 0.4860011261001702, + "grad_norm": 0.924567461013794, + "learning_rate": 1.0938772873124321e-05, + "loss": 0.15945146083831788, + "step": 18450 + }, + { + "epoch": 0.48626454134466895, + "grad_norm": 1.2905631065368652, + "learning_rate": 1.0930278836347943e-05, + "loss": 0.14568889141082764, + "step": 18460 + }, + { + "epoch": 0.4865279565891677, + "grad_norm": 1.1000081300735474, + "learning_rate": 1.0921784122474495e-05, + "loss": 0.14859650135040284, + "step": 18470 + }, + { + "epoch": 0.48679137183366644, + "grad_norm": 2.6194779872894287, + "learning_rate": 1.0913288737686789e-05, + "loss": 0.15786585807800294, + "step": 18480 + }, + { + "epoch": 0.4870547870781652, + "grad_norm": 2.2194314002990723, + "learning_rate": 1.090479268816814e-05, + "loss": 0.15917823314666749, + "step": 18490 + }, + { + "epoch": 0.48731820232266393, + "grad_norm": 1.810748815536499, + "learning_rate": 1.0896295980102334e-05, + "loss": 0.1576259732246399, + "step": 18500 + }, + { + "epoch": 0.4875816175671627, + "grad_norm": 1.0585203170776367, + "learning_rate": 1.0887798619673648e-05, + "loss": 0.14439606666564941, + "step": 18510 + }, + { + "epoch": 0.4878450328116614, + "grad_norm": 1.1931428909301758, + "learning_rate": 1.087930061306682e-05, + "loss": 0.15117177963256836, + "step": 18520 + }, + { + "epoch": 0.4881084480561601, + "grad_norm": 0.8473861813545227, + "learning_rate": 1.0870801966467073e-05, + "loss": 0.15287456512451172, + "step": 18530 + }, + { + "epoch": 0.48837186330065885, + "grad_norm": 1.3272427320480347, + "learning_rate": 1.0862302686060082e-05, + "loss": 0.1426910161972046, + "step": 18540 + }, + { + "epoch": 0.4886352785451576, + "grad_norm": 2.9460387229919434, + "learning_rate": 1.0853802778031995e-05, + "loss": 0.15294713973999025, + "step": 18550 + }, + { + "epoch": 0.48889869378965634, + "grad_norm": 3.9390225410461426, + "learning_rate": 1.0845302248569411e-05, + "loss": 0.14955015182495118, + "step": 18560 + }, + { + "epoch": 0.4891621090341551, + "grad_norm": 1.8925321102142334, + "learning_rate": 1.0836801103859377e-05, + "loss": 0.15192300081253052, + "step": 18570 + }, + { + "epoch": 0.4894255242786538, + "grad_norm": 1.3245209455490112, + "learning_rate": 1.0828299350089401e-05, + "loss": 0.1423677086830139, + "step": 18580 + }, + { + "epoch": 0.48968893952315257, + "grad_norm": 1.4239168167114258, + "learning_rate": 1.0819796993447422e-05, + "loss": 0.1551332950592041, + "step": 18590 + }, + { + "epoch": 0.4899523547676513, + "grad_norm": 1.812379002571106, + "learning_rate": 1.081129404012182e-05, + "loss": 0.1493332028388977, + "step": 18600 + }, + { + "epoch": 0.49021577001215005, + "grad_norm": 1.8795262575149536, + "learning_rate": 1.0802790496301414e-05, + "loss": 0.1550887107849121, + "step": 18610 + }, + { + "epoch": 0.49047918525664874, + "grad_norm": 2.8536102771759033, + "learning_rate": 1.079428636817545e-05, + "loss": 0.15208611488342286, + "step": 18620 + }, + { + "epoch": 0.4907426005011475, + "grad_norm": 0.6890729069709778, + "learning_rate": 1.0785781661933596e-05, + "loss": 0.14488821029663085, + "step": 18630 + }, + { + "epoch": 0.49100601574564623, + "grad_norm": 0.6721867322921753, + "learning_rate": 1.0777276383765953e-05, + "loss": 0.14683911800384522, + "step": 18640 + }, + { + "epoch": 0.491269430990145, + "grad_norm": 2.292140483856201, + "learning_rate": 1.0768770539863017e-05, + "loss": 0.16190038919448851, + "step": 18650 + }, + { + "epoch": 0.4915328462346437, + "grad_norm": 0.840707540512085, + "learning_rate": 1.076026413641572e-05, + "loss": 0.15714601278305054, + "step": 18660 + }, + { + "epoch": 0.49179626147914246, + "grad_norm": 2.1310598850250244, + "learning_rate": 1.0751757179615378e-05, + "loss": 0.1645306706428528, + "step": 18670 + }, + { + "epoch": 0.4920596767236412, + "grad_norm": 3.859764575958252, + "learning_rate": 1.0743249675653733e-05, + "loss": 0.1559451103210449, + "step": 18680 + }, + { + "epoch": 0.49232309196813995, + "grad_norm": 1.0686813592910767, + "learning_rate": 1.0734741630722904e-05, + "loss": 0.15666595697402955, + "step": 18690 + }, + { + "epoch": 0.49258650721263864, + "grad_norm": 0.641873836517334, + "learning_rate": 1.0726233051015415e-05, + "loss": 0.15374504327774047, + "step": 18700 + }, + { + "epoch": 0.4928499224571374, + "grad_norm": 1.9054124355316162, + "learning_rate": 1.0717723942724182e-05, + "loss": 0.14103367328643798, + "step": 18710 + }, + { + "epoch": 0.4931133377016361, + "grad_norm": 2.8336031436920166, + "learning_rate": 1.07092143120425e-05, + "loss": 0.14544992446899413, + "step": 18720 + }, + { + "epoch": 0.49337675294613487, + "grad_norm": 3.7412562370300293, + "learning_rate": 1.0700704165164042e-05, + "loss": 0.14236844778060914, + "step": 18730 + }, + { + "epoch": 0.4936401681906336, + "grad_norm": 2.526545286178589, + "learning_rate": 1.069219350828286e-05, + "loss": 0.16090660095214843, + "step": 18740 + }, + { + "epoch": 0.49390358343513235, + "grad_norm": 1.1642675399780273, + "learning_rate": 1.068368234759338e-05, + "loss": 0.14623030424118041, + "step": 18750 + }, + { + "epoch": 0.4941669986796311, + "grad_norm": 2.751481533050537, + "learning_rate": 1.0675170689290393e-05, + "loss": 0.15296798944473267, + "step": 18760 + }, + { + "epoch": 0.49443041392412984, + "grad_norm": 2.95401930809021, + "learning_rate": 1.0666658539569048e-05, + "loss": 0.15585758686065673, + "step": 18770 + }, + { + "epoch": 0.4946938291686286, + "grad_norm": 3.1897006034851074, + "learning_rate": 1.0658145904624852e-05, + "loss": 0.16022499799728393, + "step": 18780 + }, + { + "epoch": 0.4949572444131273, + "grad_norm": 2.9920334815979004, + "learning_rate": 1.0649632790653677e-05, + "loss": 0.15415759086608888, + "step": 18790 + }, + { + "epoch": 0.495220659657626, + "grad_norm": 1.1532328128814697, + "learning_rate": 1.0641119203851724e-05, + "loss": 0.15790050029754638, + "step": 18800 + }, + { + "epoch": 0.49548407490212476, + "grad_norm": 1.5754544734954834, + "learning_rate": 1.0632605150415561e-05, + "loss": 0.1618953227996826, + "step": 18810 + }, + { + "epoch": 0.4957474901466235, + "grad_norm": 1.611019253730774, + "learning_rate": 1.062409063654207e-05, + "loss": 0.1647101402282715, + "step": 18820 + }, + { + "epoch": 0.49601090539112225, + "grad_norm": 5.259434700012207, + "learning_rate": 1.0615575668428494e-05, + "loss": 0.14894865751266478, + "step": 18830 + }, + { + "epoch": 0.496274320635621, + "grad_norm": 2.029409885406494, + "learning_rate": 1.0607060252272383e-05, + "loss": 0.15324459075927735, + "step": 18840 + }, + { + "epoch": 0.49653773588011973, + "grad_norm": 2.663134813308716, + "learning_rate": 1.0598544394271633e-05, + "loss": 0.13564808368682862, + "step": 18850 + }, + { + "epoch": 0.4968011511246185, + "grad_norm": 1.2556356191635132, + "learning_rate": 1.0590028100624448e-05, + "loss": 0.15316635370254517, + "step": 18860 + }, + { + "epoch": 0.4970645663691172, + "grad_norm": 3.9869110584259033, + "learning_rate": 1.0581511377529354e-05, + "loss": 0.16093536615371704, + "step": 18870 + }, + { + "epoch": 0.4973279816136159, + "grad_norm": 1.2686913013458252, + "learning_rate": 1.0572994231185194e-05, + "loss": 0.16273109912872313, + "step": 18880 + }, + { + "epoch": 0.49759139685811465, + "grad_norm": 0.8274340629577637, + "learning_rate": 1.056447666779111e-05, + "loss": 0.15027689933776855, + "step": 18890 + }, + { + "epoch": 0.4978548121026134, + "grad_norm": 1.1062766313552856, + "learning_rate": 1.0555958693546552e-05, + "loss": 0.147349214553833, + "step": 18900 + }, + { + "epoch": 0.49811822734711214, + "grad_norm": 1.3075840473175049, + "learning_rate": 1.0547440314651274e-05, + "loss": 0.13858819007873535, + "step": 18910 + }, + { + "epoch": 0.4983816425916109, + "grad_norm": 2.031595230102539, + "learning_rate": 1.0538921537305315e-05, + "loss": 0.15762057304382324, + "step": 18920 + }, + { + "epoch": 0.49864505783610963, + "grad_norm": 2.6959786415100098, + "learning_rate": 1.0530402367709007e-05, + "loss": 0.1507290005683899, + "step": 18930 + }, + { + "epoch": 0.4989084730806084, + "grad_norm": 2.739394187927246, + "learning_rate": 1.0521882812062978e-05, + "loss": 0.1512363076210022, + "step": 18940 + }, + { + "epoch": 0.4991718883251071, + "grad_norm": 0.8324770927429199, + "learning_rate": 1.0513362876568118e-05, + "loss": 0.14861708879470825, + "step": 18950 + }, + { + "epoch": 0.49943530356960586, + "grad_norm": 2.764322280883789, + "learning_rate": 1.0504842567425611e-05, + "loss": 0.15912808179855348, + "step": 18960 + }, + { + "epoch": 0.49969871881410455, + "grad_norm": 0.8420987129211426, + "learning_rate": 1.0496321890836901e-05, + "loss": 0.1509975552558899, + "step": 18970 + }, + { + "epoch": 0.4999621340586033, + "grad_norm": 1.6821141242980957, + "learning_rate": 1.0487800853003706e-05, + "loss": 0.15393611192703247, + "step": 18980 + }, + { + "epoch": 0.500225549303102, + "grad_norm": 3.3981645107269287, + "learning_rate": 1.0479279460128005e-05, + "loss": 0.151292085647583, + "step": 18990 + }, + { + "epoch": 0.5004889645476008, + "grad_norm": 2.6047379970550537, + "learning_rate": 1.0470757718412035e-05, + "loss": 0.14450426101684571, + "step": 19000 + }, + { + "epoch": 0.5007523797920995, + "grad_norm": 1.877776861190796, + "learning_rate": 1.0462235634058287e-05, + "loss": 0.14993395805358886, + "step": 19010 + }, + { + "epoch": 0.5010157950365982, + "grad_norm": 1.1405893564224243, + "learning_rate": 1.0453713213269503e-05, + "loss": 0.15476009845733643, + "step": 19020 + }, + { + "epoch": 0.501279210281097, + "grad_norm": 1.8707350492477417, + "learning_rate": 1.0445190462248667e-05, + "loss": 0.14522665739059448, + "step": 19030 + }, + { + "epoch": 0.5015426255255957, + "grad_norm": 2.3489415645599365, + "learning_rate": 1.0436667387199004e-05, + "loss": 0.1515674114227295, + "step": 19040 + }, + { + "epoch": 0.5018060407700945, + "grad_norm": 1.5449211597442627, + "learning_rate": 1.0428143994323978e-05, + "loss": 0.15496447086334228, + "step": 19050 + }, + { + "epoch": 0.5020694560145932, + "grad_norm": 1.6887036561965942, + "learning_rate": 1.041962028982728e-05, + "loss": 0.14969265460968018, + "step": 19060 + }, + { + "epoch": 0.502332871259092, + "grad_norm": 3.437704086303711, + "learning_rate": 1.0411096279912835e-05, + "loss": 0.16827548742294313, + "step": 19070 + }, + { + "epoch": 0.5025962865035907, + "grad_norm": 1.0531585216522217, + "learning_rate": 1.0402571970784778e-05, + "loss": 0.15571458339691163, + "step": 19080 + }, + { + "epoch": 0.5028597017480895, + "grad_norm": 1.4149327278137207, + "learning_rate": 1.0394047368647473e-05, + "loss": 0.14546139240264894, + "step": 19090 + }, + { + "epoch": 0.5031231169925882, + "grad_norm": 2.1914889812469482, + "learning_rate": 1.0385522479705493e-05, + "loss": 0.14251078367233277, + "step": 19100 + }, + { + "epoch": 0.5033865322370868, + "grad_norm": 0.6615564227104187, + "learning_rate": 1.0376997310163622e-05, + "loss": 0.16003574132919313, + "step": 19110 + }, + { + "epoch": 0.5036499474815856, + "grad_norm": 1.7406820058822632, + "learning_rate": 1.0368471866226837e-05, + "loss": 0.1458576202392578, + "step": 19120 + }, + { + "epoch": 0.5039133627260843, + "grad_norm": 4.413959980010986, + "learning_rate": 1.0359946154100336e-05, + "loss": 0.14773013591766357, + "step": 19130 + }, + { + "epoch": 0.5041767779705831, + "grad_norm": 1.397755742073059, + "learning_rate": 1.035142017998949e-05, + "loss": 0.1510316848754883, + "step": 19140 + }, + { + "epoch": 0.5044401932150818, + "grad_norm": 1.2370930910110474, + "learning_rate": 1.0342893950099879e-05, + "loss": 0.14860148429870607, + "step": 19150 + }, + { + "epoch": 0.5047036084595806, + "grad_norm": 0.9773162007331848, + "learning_rate": 1.0334367470637257e-05, + "loss": 0.1510108232498169, + "step": 19160 + }, + { + "epoch": 0.5049670237040793, + "grad_norm": 1.5205739736557007, + "learning_rate": 1.0325840747807562e-05, + "loss": 0.15114011764526367, + "step": 19170 + }, + { + "epoch": 0.505230438948578, + "grad_norm": 1.7201714515686035, + "learning_rate": 1.0317313787816915e-05, + "loss": 0.1532581329345703, + "step": 19180 + }, + { + "epoch": 0.5054938541930768, + "grad_norm": 2.2158167362213135, + "learning_rate": 1.0308786596871606e-05, + "loss": 0.1583150863647461, + "step": 19190 + }, + { + "epoch": 0.5057572694375755, + "grad_norm": 1.9414952993392944, + "learning_rate": 1.0300259181178086e-05, + "loss": 0.14917930364608764, + "step": 19200 + }, + { + "epoch": 0.5060206846820743, + "grad_norm": 0.8030292987823486, + "learning_rate": 1.0291731546942977e-05, + "loss": 0.1432624101638794, + "step": 19210 + }, + { + "epoch": 0.506284099926573, + "grad_norm": 1.6809905767440796, + "learning_rate": 1.0283203700373065e-05, + "loss": 0.14732029438018798, + "step": 19220 + }, + { + "epoch": 0.5065475151710718, + "grad_norm": 0.7435778379440308, + "learning_rate": 1.0274675647675281e-05, + "loss": 0.16089013814926148, + "step": 19230 + }, + { + "epoch": 0.5068109304155705, + "grad_norm": 2.4948434829711914, + "learning_rate": 1.0266147395056712e-05, + "loss": 0.15120396614074708, + "step": 19240 + }, + { + "epoch": 0.5070743456600693, + "grad_norm": 2.4109861850738525, + "learning_rate": 1.0257618948724583e-05, + "loss": 0.15643504858016968, + "step": 19250 + }, + { + "epoch": 0.507337760904568, + "grad_norm": 1.3695958852767944, + "learning_rate": 1.024909031488627e-05, + "loss": 0.16018238067626953, + "step": 19260 + }, + { + "epoch": 0.5076011761490666, + "grad_norm": 1.9655468463897705, + "learning_rate": 1.0240561499749278e-05, + "loss": 0.13718252182006835, + "step": 19270 + }, + { + "epoch": 0.5078645913935654, + "grad_norm": 0.7304215431213379, + "learning_rate": 1.023203250952125e-05, + "loss": 0.14792541265487671, + "step": 19280 + }, + { + "epoch": 0.5081280066380641, + "grad_norm": 1.8636192083358765, + "learning_rate": 1.0223503350409949e-05, + "loss": 0.14306578636169434, + "step": 19290 + }, + { + "epoch": 0.5083914218825629, + "grad_norm": 1.1902563571929932, + "learning_rate": 1.0214974028623268e-05, + "loss": 0.16245076656341553, + "step": 19300 + }, + { + "epoch": 0.5086548371270616, + "grad_norm": 1.7365609407424927, + "learning_rate": 1.0206444550369212e-05, + "loss": 0.16373276710510254, + "step": 19310 + }, + { + "epoch": 0.5089182523715604, + "grad_norm": 1.734985589981079, + "learning_rate": 1.0197914921855909e-05, + "loss": 0.15351606607437135, + "step": 19320 + }, + { + "epoch": 0.5091816676160591, + "grad_norm": 2.1506760120391846, + "learning_rate": 1.0189385149291586e-05, + "loss": 0.15751049518585206, + "step": 19330 + }, + { + "epoch": 0.5094450828605579, + "grad_norm": 1.0210728645324707, + "learning_rate": 1.018085523888458e-05, + "loss": 0.15337830781936646, + "step": 19340 + }, + { + "epoch": 0.5097084981050566, + "grad_norm": 3.630755662918091, + "learning_rate": 1.0172325196843328e-05, + "loss": 0.14808616638183594, + "step": 19350 + }, + { + "epoch": 0.5099719133495553, + "grad_norm": 1.3301416635513306, + "learning_rate": 1.0163795029376362e-05, + "loss": 0.14218244552612305, + "step": 19360 + }, + { + "epoch": 0.5102353285940541, + "grad_norm": 1.5397474765777588, + "learning_rate": 1.015526474269231e-05, + "loss": 0.1513054609298706, + "step": 19370 + }, + { + "epoch": 0.5104987438385528, + "grad_norm": 2.607743501663208, + "learning_rate": 1.0146734342999875e-05, + "loss": 0.14503084421157836, + "step": 19380 + }, + { + "epoch": 0.5107621590830516, + "grad_norm": 1.9455803632736206, + "learning_rate": 1.0138203836507859e-05, + "loss": 0.16207147836685182, + "step": 19390 + }, + { + "epoch": 0.5110255743275502, + "grad_norm": 1.3873592615127563, + "learning_rate": 1.0129673229425125e-05, + "loss": 0.15713565349578856, + "step": 19400 + }, + { + "epoch": 0.511288989572049, + "grad_norm": 2.7797069549560547, + "learning_rate": 1.0121142527960624e-05, + "loss": 0.14391715526580812, + "step": 19410 + }, + { + "epoch": 0.5115524048165477, + "grad_norm": 2.528665781021118, + "learning_rate": 1.0112611738323362e-05, + "loss": 0.1418299674987793, + "step": 19420 + }, + { + "epoch": 0.5118158200610465, + "grad_norm": 1.6892849206924438, + "learning_rate": 1.0104080866722419e-05, + "loss": 0.1405206084251404, + "step": 19430 + }, + { + "epoch": 0.5120792353055452, + "grad_norm": 2.6812210083007812, + "learning_rate": 1.0095549919366928e-05, + "loss": 0.14283456802368164, + "step": 19440 + }, + { + "epoch": 0.5123426505500439, + "grad_norm": 2.028449535369873, + "learning_rate": 1.0087018902466086e-05, + "loss": 0.16155834197998048, + "step": 19450 + }, + { + "epoch": 0.5126060657945427, + "grad_norm": 3.1139235496520996, + "learning_rate": 1.0078487822229132e-05, + "loss": 0.16164784431457518, + "step": 19460 + }, + { + "epoch": 0.5128694810390414, + "grad_norm": 1.0991437435150146, + "learning_rate": 1.0069956684865354e-05, + "loss": 0.15612831115722656, + "step": 19470 + }, + { + "epoch": 0.5131328962835402, + "grad_norm": 1.7656172513961792, + "learning_rate": 1.0061425496584084e-05, + "loss": 0.1443232774734497, + "step": 19480 + }, + { + "epoch": 0.5133963115280389, + "grad_norm": 3.0055248737335205, + "learning_rate": 1.0052894263594687e-05, + "loss": 0.1328594446182251, + "step": 19490 + }, + { + "epoch": 0.5136597267725377, + "grad_norm": 1.1190121173858643, + "learning_rate": 1.0044362992106563e-05, + "loss": 0.14415603876113892, + "step": 19500 + }, + { + "epoch": 0.5139231420170364, + "grad_norm": 1.5299254655838013, + "learning_rate": 1.0035831688329139e-05, + "loss": 0.15121160745620726, + "step": 19510 + }, + { + "epoch": 0.5141865572615352, + "grad_norm": 1.7458570003509521, + "learning_rate": 1.0027300358471865e-05, + "loss": 0.15349026918411254, + "step": 19520 + }, + { + "epoch": 0.5144499725060339, + "grad_norm": 3.0793912410736084, + "learning_rate": 1.0018769008744214e-05, + "loss": 0.1598818302154541, + "step": 19530 + }, + { + "epoch": 0.5147133877505325, + "grad_norm": 1.13520348072052, + "learning_rate": 1.0010237645355669e-05, + "loss": 0.15261731147766114, + "step": 19540 + }, + { + "epoch": 0.5149768029950313, + "grad_norm": 1.2502754926681519, + "learning_rate": 1.0001706274515724e-05, + "loss": 0.153913152217865, + "step": 19550 + }, + { + "epoch": 0.51524021823953, + "grad_norm": 2.3605403900146484, + "learning_rate": 9.993174902433877e-06, + "loss": 0.14079036712646484, + "step": 19560 + }, + { + "epoch": 0.5155036334840288, + "grad_norm": 0.8114306330680847, + "learning_rate": 9.984643535319632e-06, + "loss": 0.1511770963668823, + "step": 19570 + }, + { + "epoch": 0.5157670487285275, + "grad_norm": 3.9862313270568848, + "learning_rate": 9.976112179382483e-06, + "loss": 0.16307482719421387, + "step": 19580 + }, + { + "epoch": 0.5160304639730263, + "grad_norm": 1.2031644582748413, + "learning_rate": 9.967580840831922e-06, + "loss": 0.15624452829360963, + "step": 19590 + }, + { + "epoch": 0.516293879217525, + "grad_norm": 1.614613652229309, + "learning_rate": 9.959049525877425e-06, + "loss": 0.15057613849639892, + "step": 19600 + }, + { + "epoch": 0.5165572944620238, + "grad_norm": 3.043444871902466, + "learning_rate": 9.950518240728446e-06, + "loss": 0.15843660831451417, + "step": 19610 + }, + { + "epoch": 0.5168207097065225, + "grad_norm": 0.8924649357795715, + "learning_rate": 9.94198699159443e-06, + "loss": 0.14863486289978028, + "step": 19620 + }, + { + "epoch": 0.5170841249510212, + "grad_norm": 1.318077802658081, + "learning_rate": 9.93345578468478e-06, + "loss": 0.15775115489959718, + "step": 19630 + }, + { + "epoch": 0.51734754019552, + "grad_norm": 1.9204890727996826, + "learning_rate": 9.924924626208884e-06, + "loss": 0.1418859601020813, + "step": 19640 + }, + { + "epoch": 0.5176109554400187, + "grad_norm": 1.5320758819580078, + "learning_rate": 9.916393522376087e-06, + "loss": 0.15289084911346434, + "step": 19650 + }, + { + "epoch": 0.5178743706845175, + "grad_norm": 2.4725213050842285, + "learning_rate": 9.907862479395684e-06, + "loss": 0.15217175483703613, + "step": 19660 + }, + { + "epoch": 0.5181377859290162, + "grad_norm": 1.6496870517730713, + "learning_rate": 9.899331503476951e-06, + "loss": 0.15979795455932616, + "step": 19670 + }, + { + "epoch": 0.518401201173515, + "grad_norm": 1.1783877611160278, + "learning_rate": 9.890800600829086e-06, + "loss": 0.1389568567276001, + "step": 19680 + }, + { + "epoch": 0.5186646164180136, + "grad_norm": 3.1684632301330566, + "learning_rate": 9.882269777661261e-06, + "loss": 0.13829348087310792, + "step": 19690 + }, + { + "epoch": 0.5189280316625123, + "grad_norm": 1.3073264360427856, + "learning_rate": 9.873739040182569e-06, + "loss": 0.14591498374938966, + "step": 19700 + }, + { + "epoch": 0.5191914469070111, + "grad_norm": 1.8517711162567139, + "learning_rate": 9.86520839460205e-06, + "loss": 0.16317975521087646, + "step": 19710 + }, + { + "epoch": 0.5194548621515098, + "grad_norm": 3.621812105178833, + "learning_rate": 9.856677847128678e-06, + "loss": 0.1437302589416504, + "step": 19720 + }, + { + "epoch": 0.5197182773960086, + "grad_norm": 1.1408345699310303, + "learning_rate": 9.848147403971353e-06, + "loss": 0.14324636459350587, + "step": 19730 + }, + { + "epoch": 0.5199816926405073, + "grad_norm": 3.075996160507202, + "learning_rate": 9.839617071338897e-06, + "loss": 0.15422736406326293, + "step": 19740 + }, + { + "epoch": 0.5202451078850061, + "grad_norm": 1.8465067148208618, + "learning_rate": 9.831086855440056e-06, + "loss": 0.16270096302032472, + "step": 19750 + }, + { + "epoch": 0.5205085231295048, + "grad_norm": 1.8349863290786743, + "learning_rate": 9.82255676248349e-06, + "loss": 0.14087529182434083, + "step": 19760 + }, + { + "epoch": 0.5207719383740036, + "grad_norm": 2.9180309772491455, + "learning_rate": 9.814026798677767e-06, + "loss": 0.1387271285057068, + "step": 19770 + }, + { + "epoch": 0.5210353536185023, + "grad_norm": 0.9653788208961487, + "learning_rate": 9.805496970231367e-06, + "loss": 0.14327144622802734, + "step": 19780 + }, + { + "epoch": 0.521298768863001, + "grad_norm": 1.12789785861969, + "learning_rate": 9.796967283352657e-06, + "loss": 0.13996831178665162, + "step": 19790 + }, + { + "epoch": 0.5215621841074998, + "grad_norm": 1.4600239992141724, + "learning_rate": 9.788437744249925e-06, + "loss": 0.14753828048706055, + "step": 19800 + }, + { + "epoch": 0.5218255993519985, + "grad_norm": 2.994171619415283, + "learning_rate": 9.779908359131327e-06, + "loss": 0.1454757571220398, + "step": 19810 + }, + { + "epoch": 0.5220890145964973, + "grad_norm": 1.552266240119934, + "learning_rate": 9.77137913420492e-06, + "loss": 0.14196969270706178, + "step": 19820 + }, + { + "epoch": 0.522352429840996, + "grad_norm": 1.8833706378936768, + "learning_rate": 9.762850075678644e-06, + "loss": 0.15706683397293092, + "step": 19830 + }, + { + "epoch": 0.5226158450854947, + "grad_norm": 2.7846360206604004, + "learning_rate": 9.754321189760312e-06, + "loss": 0.14188101291656494, + "step": 19840 + }, + { + "epoch": 0.5228792603299934, + "grad_norm": 0.7984705567359924, + "learning_rate": 9.74579248265762e-06, + "loss": 0.14016458988189698, + "step": 19850 + }, + { + "epoch": 0.5231426755744922, + "grad_norm": 1.0785188674926758, + "learning_rate": 9.737263960578127e-06, + "loss": 0.14552472829818724, + "step": 19860 + }, + { + "epoch": 0.5234060908189909, + "grad_norm": 1.1710593700408936, + "learning_rate": 9.728735629729256e-06, + "loss": 0.15032246112823486, + "step": 19870 + }, + { + "epoch": 0.5236695060634896, + "grad_norm": 1.699578046798706, + "learning_rate": 9.7202074963183e-06, + "loss": 0.14772374629974366, + "step": 19880 + }, + { + "epoch": 0.5239329213079884, + "grad_norm": 1.6510086059570312, + "learning_rate": 9.711679566552396e-06, + "loss": 0.14762755632400512, + "step": 19890 + }, + { + "epoch": 0.5241963365524871, + "grad_norm": 1.7631961107254028, + "learning_rate": 9.703151846638545e-06, + "loss": 0.15428080558776855, + "step": 19900 + }, + { + "epoch": 0.5244597517969859, + "grad_norm": 1.438719630241394, + "learning_rate": 9.694624342783591e-06, + "loss": 0.15370659828186034, + "step": 19910 + }, + { + "epoch": 0.5247231670414846, + "grad_norm": 1.8142485618591309, + "learning_rate": 9.686097061194209e-06, + "loss": 0.16767258644104005, + "step": 19920 + }, + { + "epoch": 0.5249865822859834, + "grad_norm": 2.1469507217407227, + "learning_rate": 9.677570008076933e-06, + "loss": 0.14947314262390138, + "step": 19930 + }, + { + "epoch": 0.5252499975304821, + "grad_norm": 1.1872823238372803, + "learning_rate": 9.669043189638115e-06, + "loss": 0.1518467664718628, + "step": 19940 + }, + { + "epoch": 0.5255134127749809, + "grad_norm": 1.3587907552719116, + "learning_rate": 9.66051661208394e-06, + "loss": 0.16042499542236327, + "step": 19950 + }, + { + "epoch": 0.5257768280194796, + "grad_norm": 1.1917998790740967, + "learning_rate": 9.651990281620423e-06, + "loss": 0.1409102201461792, + "step": 19960 + }, + { + "epoch": 0.5260402432639782, + "grad_norm": 3.6775527000427246, + "learning_rate": 9.643464204453392e-06, + "loss": 0.14888967275619508, + "step": 19970 + }, + { + "epoch": 0.526303658508477, + "grad_norm": 2.3463945388793945, + "learning_rate": 9.634938386788497e-06, + "loss": 0.15932149887084962, + "step": 19980 + }, + { + "epoch": 0.5265670737529757, + "grad_norm": 1.4691152572631836, + "learning_rate": 9.626412834831193e-06, + "loss": 0.16016432046890258, + "step": 19990 + }, + { + "epoch": 0.5268304889974745, + "grad_norm": 2.7667124271392822, + "learning_rate": 9.617887554786746e-06, + "loss": 0.13903367519378662, + "step": 20000 + }, + { + "epoch": 0.5270939042419732, + "grad_norm": 2.0913541316986084, + "learning_rate": 9.609362552860225e-06, + "loss": 0.13841686248779297, + "step": 20010 + }, + { + "epoch": 0.527357319486472, + "grad_norm": 3.1243131160736084, + "learning_rate": 9.600837835256486e-06, + "loss": 0.14975866079330444, + "step": 20020 + }, + { + "epoch": 0.5276207347309707, + "grad_norm": 1.5658342838287354, + "learning_rate": 9.592313408180199e-06, + "loss": 0.1307140588760376, + "step": 20030 + }, + { + "epoch": 0.5278841499754695, + "grad_norm": 5.826523303985596, + "learning_rate": 9.583789277835798e-06, + "loss": 0.1638022780418396, + "step": 20040 + }, + { + "epoch": 0.5281475652199682, + "grad_norm": 1.7647778987884521, + "learning_rate": 9.575265450427517e-06, + "loss": 0.15824333429336548, + "step": 20050 + }, + { + "epoch": 0.5284109804644669, + "grad_norm": 1.5214742422103882, + "learning_rate": 9.566741932159369e-06, + "loss": 0.1561541438102722, + "step": 20060 + }, + { + "epoch": 0.5286743957089657, + "grad_norm": 1.8043266534805298, + "learning_rate": 9.558218729235132e-06, + "loss": 0.1570636510848999, + "step": 20070 + }, + { + "epoch": 0.5289378109534644, + "grad_norm": 4.210882663726807, + "learning_rate": 9.549695847858363e-06, + "loss": 0.15076112747192383, + "step": 20080 + }, + { + "epoch": 0.5292012261979632, + "grad_norm": 1.9036006927490234, + "learning_rate": 9.541173294232387e-06, + "loss": 0.15411808490753173, + "step": 20090 + }, + { + "epoch": 0.5294646414424619, + "grad_norm": 1.21097731590271, + "learning_rate": 9.53265107456028e-06, + "loss": 0.15460859537124633, + "step": 20100 + }, + { + "epoch": 0.5297280566869607, + "grad_norm": 0.9268825650215149, + "learning_rate": 9.524129195044886e-06, + "loss": 0.1363750457763672, + "step": 20110 + }, + { + "epoch": 0.5299914719314593, + "grad_norm": 2.1646125316619873, + "learning_rate": 9.515607661888799e-06, + "loss": 0.15048515796661377, + "step": 20120 + }, + { + "epoch": 0.5302548871759581, + "grad_norm": 2.5554685592651367, + "learning_rate": 9.507086481294345e-06, + "loss": 0.17467155456542968, + "step": 20130 + }, + { + "epoch": 0.5305183024204568, + "grad_norm": 0.9073246717453003, + "learning_rate": 9.498565659463624e-06, + "loss": 0.14312201738357544, + "step": 20140 + }, + { + "epoch": 0.5307817176649555, + "grad_norm": 1.0557907819747925, + "learning_rate": 9.490045202598442e-06, + "loss": 0.14242768287658691, + "step": 20150 + }, + { + "epoch": 0.5310451329094543, + "grad_norm": 0.8773637413978577, + "learning_rate": 9.48152511690037e-06, + "loss": 0.15273351669311525, + "step": 20160 + }, + { + "epoch": 0.531308548153953, + "grad_norm": 2.51826810836792, + "learning_rate": 9.473005408570685e-06, + "loss": 0.14272309541702272, + "step": 20170 + }, + { + "epoch": 0.5315719633984518, + "grad_norm": 1.5484079122543335, + "learning_rate": 9.464486083810397e-06, + "loss": 0.13582628965377808, + "step": 20180 + }, + { + "epoch": 0.5318353786429505, + "grad_norm": 1.9150656461715698, + "learning_rate": 9.455967148820243e-06, + "loss": 0.14825716018676757, + "step": 20190 + }, + { + "epoch": 0.5320987938874493, + "grad_norm": 3.2279438972473145, + "learning_rate": 9.44744860980067e-06, + "loss": 0.156512451171875, + "step": 20200 + }, + { + "epoch": 0.532362209131948, + "grad_norm": 0.8479232788085938, + "learning_rate": 9.438930472951838e-06, + "loss": 0.1349220395088196, + "step": 20210 + }, + { + "epoch": 0.5326256243764467, + "grad_norm": 1.271417498588562, + "learning_rate": 9.430412744473615e-06, + "loss": 0.14556360244750977, + "step": 20220 + }, + { + "epoch": 0.5328890396209455, + "grad_norm": 1.6307005882263184, + "learning_rate": 9.421895430565568e-06, + "loss": 0.1510610342025757, + "step": 20230 + }, + { + "epoch": 0.5331524548654442, + "grad_norm": 2.253476619720459, + "learning_rate": 9.413378537426972e-06, + "loss": 0.1421764850616455, + "step": 20240 + }, + { + "epoch": 0.533415870109943, + "grad_norm": 0.8070362210273743, + "learning_rate": 9.404862071256785e-06, + "loss": 0.1434248685836792, + "step": 20250 + }, + { + "epoch": 0.5336792853544416, + "grad_norm": 2.268548011779785, + "learning_rate": 9.396346038253654e-06, + "loss": 0.16117963790893555, + "step": 20260 + }, + { + "epoch": 0.5339427005989404, + "grad_norm": 4.327115058898926, + "learning_rate": 9.387830444615926e-06, + "loss": 0.13640806674957276, + "step": 20270 + }, + { + "epoch": 0.5342061158434391, + "grad_norm": 4.666041851043701, + "learning_rate": 9.379315296541611e-06, + "loss": 0.14479348659515381, + "step": 20280 + }, + { + "epoch": 0.5344695310879379, + "grad_norm": 1.4965832233428955, + "learning_rate": 9.370800600228399e-06, + "loss": 0.15653204917907715, + "step": 20290 + }, + { + "epoch": 0.5347329463324366, + "grad_norm": 0.671990156173706, + "learning_rate": 9.362286361873658e-06, + "loss": 0.15212355852127074, + "step": 20300 + }, + { + "epoch": 0.5349963615769353, + "grad_norm": 2.1503217220306396, + "learning_rate": 9.353772587674413e-06, + "loss": 0.15267419815063477, + "step": 20310 + }, + { + "epoch": 0.5352597768214341, + "grad_norm": 2.073683738708496, + "learning_rate": 9.345259283827362e-06, + "loss": 0.1401435613632202, + "step": 20320 + }, + { + "epoch": 0.5355231920659328, + "grad_norm": 1.647687315940857, + "learning_rate": 9.336746456528853e-06, + "loss": 0.1500331163406372, + "step": 20330 + }, + { + "epoch": 0.5357866073104316, + "grad_norm": 0.760737955570221, + "learning_rate": 9.328234111974885e-06, + "loss": 0.14130369424819947, + "step": 20340 + }, + { + "epoch": 0.5360500225549303, + "grad_norm": 0.9913352727890015, + "learning_rate": 9.319722256361117e-06, + "loss": 0.15511466264724733, + "step": 20350 + }, + { + "epoch": 0.5363134377994291, + "grad_norm": 5.195424556732178, + "learning_rate": 9.311210895882835e-06, + "loss": 0.14771417379379273, + "step": 20360 + }, + { + "epoch": 0.5365768530439278, + "grad_norm": 1.582372784614563, + "learning_rate": 9.302700036734985e-06, + "loss": 0.14598548412322998, + "step": 20370 + }, + { + "epoch": 0.5368402682884266, + "grad_norm": 2.042304277420044, + "learning_rate": 9.294189685112133e-06, + "loss": 0.15454158782958985, + "step": 20380 + }, + { + "epoch": 0.5371036835329253, + "grad_norm": 2.0021021366119385, + "learning_rate": 9.285679847208474e-06, + "loss": 0.1361936569213867, + "step": 20390 + }, + { + "epoch": 0.537367098777424, + "grad_norm": 1.8028041124343872, + "learning_rate": 9.277170529217847e-06, + "loss": 0.15407943725585938, + "step": 20400 + }, + { + "epoch": 0.5376305140219227, + "grad_norm": 1.0218807458877563, + "learning_rate": 9.268661737333691e-06, + "loss": 0.14535975456237793, + "step": 20410 + }, + { + "epoch": 0.5378939292664214, + "grad_norm": 1.8712348937988281, + "learning_rate": 9.260153477749072e-06, + "loss": 0.1497753381729126, + "step": 20420 + }, + { + "epoch": 0.5381573445109202, + "grad_norm": 2.1980068683624268, + "learning_rate": 9.251645756656676e-06, + "loss": 0.15375897884368897, + "step": 20430 + }, + { + "epoch": 0.5384207597554189, + "grad_norm": 2.071403980255127, + "learning_rate": 9.243138580248782e-06, + "loss": 0.1473431706428528, + "step": 20440 + }, + { + "epoch": 0.5386841749999177, + "grad_norm": 4.3607072830200195, + "learning_rate": 9.234631954717285e-06, + "loss": 0.1514992117881775, + "step": 20450 + }, + { + "epoch": 0.5389475902444164, + "grad_norm": 1.8466129302978516, + "learning_rate": 9.226125886253673e-06, + "loss": 0.15416309833526612, + "step": 20460 + }, + { + "epoch": 0.5392110054889152, + "grad_norm": 0.8675938844680786, + "learning_rate": 9.217620381049026e-06, + "loss": 0.14995532035827636, + "step": 20470 + }, + { + "epoch": 0.5394744207334139, + "grad_norm": 0.7887395620346069, + "learning_rate": 9.209115445294022e-06, + "loss": 0.14256045818328858, + "step": 20480 + }, + { + "epoch": 0.5397378359779126, + "grad_norm": 0.8381207585334778, + "learning_rate": 9.200611085178917e-06, + "loss": 0.13956921100616454, + "step": 20490 + }, + { + "epoch": 0.5400012512224114, + "grad_norm": 1.1248928308486938, + "learning_rate": 9.192107306893556e-06, + "loss": 0.1488182783126831, + "step": 20500 + }, + { + "epoch": 0.5402646664669101, + "grad_norm": 4.169008255004883, + "learning_rate": 9.183604116627354e-06, + "loss": 0.15875413417816162, + "step": 20510 + }, + { + "epoch": 0.5405280817114089, + "grad_norm": 1.1508829593658447, + "learning_rate": 9.175101520569293e-06, + "loss": 0.15402891635894775, + "step": 20520 + }, + { + "epoch": 0.5407914969559076, + "grad_norm": 2.7639448642730713, + "learning_rate": 9.166599524907937e-06, + "loss": 0.14669585227966309, + "step": 20530 + }, + { + "epoch": 0.5410549122004064, + "grad_norm": 3.027893304824829, + "learning_rate": 9.158098135831404e-06, + "loss": 0.15006096363067628, + "step": 20540 + }, + { + "epoch": 0.541318327444905, + "grad_norm": 1.5320595502853394, + "learning_rate": 9.149597359527368e-06, + "loss": 0.1438932776451111, + "step": 20550 + }, + { + "epoch": 0.5415817426894038, + "grad_norm": 0.8944485783576965, + "learning_rate": 9.141097202183063e-06, + "loss": 0.15185142755508424, + "step": 20560 + }, + { + "epoch": 0.5418451579339025, + "grad_norm": 1.2006646394729614, + "learning_rate": 9.13259766998527e-06, + "loss": 0.1429642081260681, + "step": 20570 + }, + { + "epoch": 0.5421085731784012, + "grad_norm": 2.785670757293701, + "learning_rate": 9.124098769120318e-06, + "loss": 0.14321818351745605, + "step": 20580 + }, + { + "epoch": 0.5423719884229, + "grad_norm": 0.9249302744865417, + "learning_rate": 9.115600505774072e-06, + "loss": 0.15495259761810304, + "step": 20590 + }, + { + "epoch": 0.5426354036673987, + "grad_norm": 1.2122645378112793, + "learning_rate": 9.10710288613193e-06, + "loss": 0.15641043186187745, + "step": 20600 + }, + { + "epoch": 0.5428988189118975, + "grad_norm": 1.3687288761138916, + "learning_rate": 9.098605916378837e-06, + "loss": 0.14467068910598754, + "step": 20610 + }, + { + "epoch": 0.5431622341563962, + "grad_norm": 2.516977548599243, + "learning_rate": 9.090109602699239e-06, + "loss": 0.14520392417907715, + "step": 20620 + }, + { + "epoch": 0.543425649400895, + "grad_norm": 3.2032580375671387, + "learning_rate": 9.081613951277135e-06, + "loss": 0.1454416275024414, + "step": 20630 + }, + { + "epoch": 0.5436890646453937, + "grad_norm": 1.0720607042312622, + "learning_rate": 9.07311896829602e-06, + "loss": 0.14403748512268066, + "step": 20640 + }, + { + "epoch": 0.5439524798898925, + "grad_norm": 1.395155668258667, + "learning_rate": 9.064624659938902e-06, + "loss": 0.13254339694976808, + "step": 20650 + }, + { + "epoch": 0.5442158951343912, + "grad_norm": 1.7973721027374268, + "learning_rate": 9.056131032388318e-06, + "loss": 0.1536765694618225, + "step": 20660 + }, + { + "epoch": 0.5444793103788899, + "grad_norm": 3.1863319873809814, + "learning_rate": 9.047638091826291e-06, + "loss": 0.1510603904724121, + "step": 20670 + }, + { + "epoch": 0.5447427256233887, + "grad_norm": 0.9731155037879944, + "learning_rate": 9.039145844434344e-06, + "loss": 0.15710983276367188, + "step": 20680 + }, + { + "epoch": 0.5450061408678873, + "grad_norm": 2.7286109924316406, + "learning_rate": 9.030654296393512e-06, + "loss": 0.14759641885757446, + "step": 20690 + }, + { + "epoch": 0.5452695561123861, + "grad_norm": 1.4948736429214478, + "learning_rate": 9.022163453884299e-06, + "loss": 0.14151123762130738, + "step": 20700 + }, + { + "epoch": 0.5455329713568848, + "grad_norm": 1.3140703439712524, + "learning_rate": 9.013673323086716e-06, + "loss": 0.14905529022216796, + "step": 20710 + }, + { + "epoch": 0.5457963866013836, + "grad_norm": 2.55560564994812, + "learning_rate": 9.005183910180243e-06, + "loss": 0.1484734296798706, + "step": 20720 + }, + { + "epoch": 0.5460598018458823, + "grad_norm": 1.837162971496582, + "learning_rate": 8.996695221343834e-06, + "loss": 0.14953410625457764, + "step": 20730 + }, + { + "epoch": 0.546323217090381, + "grad_norm": 1.645748257637024, + "learning_rate": 8.988207262755936e-06, + "loss": 0.1449885606765747, + "step": 20740 + }, + { + "epoch": 0.5465866323348798, + "grad_norm": 1.1076549291610718, + "learning_rate": 8.979720040594438e-06, + "loss": 0.15542398691177367, + "step": 20750 + }, + { + "epoch": 0.5468500475793785, + "grad_norm": 3.0000391006469727, + "learning_rate": 8.971233561036718e-06, + "loss": 0.14200279712677003, + "step": 20760 + }, + { + "epoch": 0.5471134628238773, + "grad_norm": 1.26378333568573, + "learning_rate": 8.962747830259598e-06, + "loss": 0.14239777326583863, + "step": 20770 + }, + { + "epoch": 0.547376878068376, + "grad_norm": 1.955225944519043, + "learning_rate": 8.954262854439354e-06, + "loss": 0.15499541759490967, + "step": 20780 + }, + { + "epoch": 0.5476402933128748, + "grad_norm": 2.1217734813690186, + "learning_rate": 8.945778639751723e-06, + "loss": 0.14988842010498046, + "step": 20790 + }, + { + "epoch": 0.5479037085573735, + "grad_norm": 2.766911745071411, + "learning_rate": 8.937295192371882e-06, + "loss": 0.13798151016235352, + "step": 20800 + }, + { + "epoch": 0.5481671238018723, + "grad_norm": 1.0762141942977905, + "learning_rate": 8.928812518474445e-06, + "loss": 0.12716048955917358, + "step": 20810 + }, + { + "epoch": 0.548430539046371, + "grad_norm": 1.6689727306365967, + "learning_rate": 8.920330624233477e-06, + "loss": 0.1366144299507141, + "step": 20820 + }, + { + "epoch": 0.5486939542908696, + "grad_norm": 1.9647561311721802, + "learning_rate": 8.911849515822457e-06, + "loss": 0.14782001972198486, + "step": 20830 + }, + { + "epoch": 0.5489573695353684, + "grad_norm": 1.1421020030975342, + "learning_rate": 8.903369199414311e-06, + "loss": 0.15308423042297364, + "step": 20840 + }, + { + "epoch": 0.5492207847798671, + "grad_norm": 1.196864366531372, + "learning_rate": 8.894889681181374e-06, + "loss": 0.1434711217880249, + "step": 20850 + }, + { + "epoch": 0.5494842000243659, + "grad_norm": 3.087095022201538, + "learning_rate": 8.886410967295401e-06, + "loss": 0.15229035615921022, + "step": 20860 + }, + { + "epoch": 0.5497476152688646, + "grad_norm": 1.4528872966766357, + "learning_rate": 8.877933063927576e-06, + "loss": 0.13050740957260132, + "step": 20870 + }, + { + "epoch": 0.5500110305133634, + "grad_norm": 1.3137937784194946, + "learning_rate": 8.869455977248473e-06, + "loss": 0.14141480922698973, + "step": 20880 + }, + { + "epoch": 0.5502744457578621, + "grad_norm": 2.9413983821868896, + "learning_rate": 8.860979713428085e-06, + "loss": 0.13075659275054932, + "step": 20890 + }, + { + "epoch": 0.5505378610023609, + "grad_norm": 2.187072992324829, + "learning_rate": 8.852504278635806e-06, + "loss": 0.14814119338989257, + "step": 20900 + }, + { + "epoch": 0.5508012762468596, + "grad_norm": 2.2644853591918945, + "learning_rate": 8.844029679040418e-06, + "loss": 0.1396593928337097, + "step": 20910 + }, + { + "epoch": 0.5510646914913583, + "grad_norm": 2.2756223678588867, + "learning_rate": 8.835555920810104e-06, + "loss": 0.14719418287277222, + "step": 20920 + }, + { + "epoch": 0.5513281067358571, + "grad_norm": 1.7641741037368774, + "learning_rate": 8.827083010112429e-06, + "loss": 0.13072391748428344, + "step": 20930 + }, + { + "epoch": 0.5515915219803558, + "grad_norm": 0.9556217789649963, + "learning_rate": 8.818610953114342e-06, + "loss": 0.1382274866104126, + "step": 20940 + }, + { + "epoch": 0.5518549372248546, + "grad_norm": 2.440429210662842, + "learning_rate": 8.810139755982173e-06, + "loss": 0.14079344272613525, + "step": 20950 + }, + { + "epoch": 0.5521183524693533, + "grad_norm": 2.197925567626953, + "learning_rate": 8.801669424881624e-06, + "loss": 0.15188088417053222, + "step": 20960 + }, + { + "epoch": 0.5523817677138521, + "grad_norm": 0.8414325714111328, + "learning_rate": 8.79319996597777e-06, + "loss": 0.13869950771331788, + "step": 20970 + }, + { + "epoch": 0.5526451829583507, + "grad_norm": 2.114098072052002, + "learning_rate": 8.784731385435049e-06, + "loss": 0.14270566701889037, + "step": 20980 + }, + { + "epoch": 0.5529085982028495, + "grad_norm": 1.2355846166610718, + "learning_rate": 8.77626368941725e-06, + "loss": 0.12979985475540162, + "step": 20990 + }, + { + "epoch": 0.5531720134473482, + "grad_norm": 1.9584819078445435, + "learning_rate": 8.767796884087539e-06, + "loss": 0.1443095922470093, + "step": 21000 + }, + { + "epoch": 0.5534354286918469, + "grad_norm": 2.5708303451538086, + "learning_rate": 8.759330975608418e-06, + "loss": 0.14637815952301025, + "step": 21010 + }, + { + "epoch": 0.5536988439363457, + "grad_norm": 1.0604673624038696, + "learning_rate": 8.750865970141736e-06, + "loss": 0.14230458736419677, + "step": 21020 + }, + { + "epoch": 0.5539622591808444, + "grad_norm": 1.1940425634384155, + "learning_rate": 8.742401873848696e-06, + "loss": 0.14554939270019532, + "step": 21030 + }, + { + "epoch": 0.5542256744253432, + "grad_norm": 1.6265305280685425, + "learning_rate": 8.733938692889827e-06, + "loss": 0.15123180150985718, + "step": 21040 + }, + { + "epoch": 0.5544890896698419, + "grad_norm": 1.768066167831421, + "learning_rate": 8.725476433425003e-06, + "loss": 0.14297451972961425, + "step": 21050 + }, + { + "epoch": 0.5547525049143407, + "grad_norm": 2.704347610473633, + "learning_rate": 8.717015101613417e-06, + "loss": 0.14960274696350098, + "step": 21060 + }, + { + "epoch": 0.5550159201588394, + "grad_norm": 1.185866117477417, + "learning_rate": 8.708554703613592e-06, + "loss": 0.14365140199661255, + "step": 21070 + }, + { + "epoch": 0.5552793354033382, + "grad_norm": 1.3126977682113647, + "learning_rate": 8.700095245583374e-06, + "loss": 0.12437348365783692, + "step": 21080 + }, + { + "epoch": 0.5555427506478369, + "grad_norm": 1.2678942680358887, + "learning_rate": 8.691636733679915e-06, + "loss": 0.14453461170196533, + "step": 21090 + }, + { + "epoch": 0.5558061658923356, + "grad_norm": 1.4278889894485474, + "learning_rate": 8.683179174059691e-06, + "loss": 0.14709386825561524, + "step": 21100 + }, + { + "epoch": 0.5560695811368344, + "grad_norm": 1.4097157716751099, + "learning_rate": 8.67472257287848e-06, + "loss": 0.13973368406295777, + "step": 21110 + }, + { + "epoch": 0.556332996381333, + "grad_norm": 2.39982533454895, + "learning_rate": 8.666266936291354e-06, + "loss": 0.14864025115966797, + "step": 21120 + }, + { + "epoch": 0.5565964116258318, + "grad_norm": 2.2976722717285156, + "learning_rate": 8.657812270452695e-06, + "loss": 0.1528327226638794, + "step": 21130 + }, + { + "epoch": 0.5568598268703305, + "grad_norm": 3.144921064376831, + "learning_rate": 8.649358581516171e-06, + "loss": 0.14806679487228394, + "step": 21140 + }, + { + "epoch": 0.5571232421148293, + "grad_norm": 1.3857184648513794, + "learning_rate": 8.640905875634744e-06, + "loss": 0.146455979347229, + "step": 21150 + }, + { + "epoch": 0.557386657359328, + "grad_norm": 2.050049304962158, + "learning_rate": 8.632454158960656e-06, + "loss": 0.14265081882476807, + "step": 21160 + }, + { + "epoch": 0.5576500726038268, + "grad_norm": 2.141723394393921, + "learning_rate": 8.624003437645431e-06, + "loss": 0.12839069366455078, + "step": 21170 + }, + { + "epoch": 0.5579134878483255, + "grad_norm": 3.41788911819458, + "learning_rate": 8.615553717839873e-06, + "loss": 0.13141226768493652, + "step": 21180 + }, + { + "epoch": 0.5581769030928242, + "grad_norm": 1.6771339178085327, + "learning_rate": 8.607105005694048e-06, + "loss": 0.15483478307724, + "step": 21190 + }, + { + "epoch": 0.558440318337323, + "grad_norm": 2.831346035003662, + "learning_rate": 8.598657307357293e-06, + "loss": 0.14960618019104005, + "step": 21200 + }, + { + "epoch": 0.5587037335818217, + "grad_norm": 1.3686697483062744, + "learning_rate": 8.590210628978213e-06, + "loss": 0.13718738555908203, + "step": 21210 + }, + { + "epoch": 0.5589671488263205, + "grad_norm": 2.4408740997314453, + "learning_rate": 8.581764976704653e-06, + "loss": 0.15208723545074462, + "step": 21220 + }, + { + "epoch": 0.5592305640708192, + "grad_norm": 1.487444281578064, + "learning_rate": 8.573320356683739e-06, + "loss": 0.1448686122894287, + "step": 21230 + }, + { + "epoch": 0.559493979315318, + "grad_norm": 1.6113229990005493, + "learning_rate": 8.564876775061819e-06, + "loss": 0.13391214609146118, + "step": 21240 + }, + { + "epoch": 0.5597573945598167, + "grad_norm": 2.674628973007202, + "learning_rate": 8.556434237984496e-06, + "loss": 0.12622861862182616, + "step": 21250 + }, + { + "epoch": 0.5600208098043153, + "grad_norm": 2.816848039627075, + "learning_rate": 8.547992751596617e-06, + "loss": 0.1492697238922119, + "step": 21260 + }, + { + "epoch": 0.5602842250488141, + "grad_norm": 0.7032254934310913, + "learning_rate": 8.53955232204226e-06, + "loss": 0.15109466314315795, + "step": 21270 + }, + { + "epoch": 0.5605476402933128, + "grad_norm": 1.9566556215286255, + "learning_rate": 8.531112955464728e-06, + "loss": 0.14732959270477294, + "step": 21280 + }, + { + "epoch": 0.5608110555378116, + "grad_norm": 0.7365887761116028, + "learning_rate": 8.522674658006564e-06, + "loss": 0.13900787830352784, + "step": 21290 + }, + { + "epoch": 0.5610744707823103, + "grad_norm": 1.7768200635910034, + "learning_rate": 8.514237435809517e-06, + "loss": 0.13309884071350098, + "step": 21300 + }, + { + "epoch": 0.5613378860268091, + "grad_norm": 0.9830077290534973, + "learning_rate": 8.505801295014569e-06, + "loss": 0.13333265781402587, + "step": 21310 + }, + { + "epoch": 0.5616013012713078, + "grad_norm": 2.476543426513672, + "learning_rate": 8.497366241761907e-06, + "loss": 0.14623563289642333, + "step": 21320 + }, + { + "epoch": 0.5618647165158066, + "grad_norm": 0.8215112686157227, + "learning_rate": 8.488932282190914e-06, + "loss": 0.14069904088974, + "step": 21330 + }, + { + "epoch": 0.5621281317603053, + "grad_norm": 0.7417834997177124, + "learning_rate": 8.480499422440207e-06, + "loss": 0.14149219989776612, + "step": 21340 + }, + { + "epoch": 0.562391547004804, + "grad_norm": 1.8595393896102905, + "learning_rate": 8.472067668647571e-06, + "loss": 0.14403085708618163, + "step": 21350 + }, + { + "epoch": 0.5626549622493028, + "grad_norm": 1.6216130256652832, + "learning_rate": 8.463637026950013e-06, + "loss": 0.14204106330871583, + "step": 21360 + }, + { + "epoch": 0.5629183774938015, + "grad_norm": 1.2684329748153687, + "learning_rate": 8.455207503483708e-06, + "loss": 0.14577219486236573, + "step": 21370 + }, + { + "epoch": 0.5631817927383003, + "grad_norm": 4.683033466339111, + "learning_rate": 8.446779104384028e-06, + "loss": 0.149788236618042, + "step": 21380 + }, + { + "epoch": 0.563445207982799, + "grad_norm": 1.8134779930114746, + "learning_rate": 8.43835183578553e-06, + "loss": 0.13846781253814697, + "step": 21390 + }, + { + "epoch": 0.5637086232272978, + "grad_norm": 0.9615273475646973, + "learning_rate": 8.429925703821941e-06, + "loss": 0.15682209730148317, + "step": 21400 + }, + { + "epoch": 0.5639720384717964, + "grad_norm": 1.6720410585403442, + "learning_rate": 8.421500714626164e-06, + "loss": 0.14811196327209472, + "step": 21410 + }, + { + "epoch": 0.5642354537162952, + "grad_norm": 2.6564948558807373, + "learning_rate": 8.413076874330266e-06, + "loss": 0.14629483222961426, + "step": 21420 + }, + { + "epoch": 0.5644988689607939, + "grad_norm": 0.9261010885238647, + "learning_rate": 8.404654189065486e-06, + "loss": 0.13897862434387206, + "step": 21430 + }, + { + "epoch": 0.5647622842052926, + "grad_norm": 1.2306358814239502, + "learning_rate": 8.396232664962218e-06, + "loss": 0.14984960556030275, + "step": 21440 + }, + { + "epoch": 0.5650256994497914, + "grad_norm": 1.0163781642913818, + "learning_rate": 8.387812308150011e-06, + "loss": 0.12390830516815185, + "step": 21450 + }, + { + "epoch": 0.5652891146942901, + "grad_norm": 1.409242868423462, + "learning_rate": 8.379393124757554e-06, + "loss": 0.16381604671478273, + "step": 21460 + }, + { + "epoch": 0.5655525299387889, + "grad_norm": 0.9843032360076904, + "learning_rate": 8.370975120912708e-06, + "loss": 0.1478442907333374, + "step": 21470 + }, + { + "epoch": 0.5658159451832876, + "grad_norm": 3.593583106994629, + "learning_rate": 8.362558302742447e-06, + "loss": 0.13487639427185058, + "step": 21480 + }, + { + "epoch": 0.5660793604277864, + "grad_norm": 4.590746879577637, + "learning_rate": 8.354142676372898e-06, + "loss": 0.14573183059692382, + "step": 21490 + }, + { + "epoch": 0.5663427756722851, + "grad_norm": 0.9912018179893494, + "learning_rate": 8.345728247929319e-06, + "loss": 0.13684476613998414, + "step": 21500 + }, + { + "epoch": 0.5666061909167839, + "grad_norm": 1.9980136156082153, + "learning_rate": 8.33731502353609e-06, + "loss": 0.1410775065422058, + "step": 21510 + }, + { + "epoch": 0.5668696061612826, + "grad_norm": 2.1643905639648438, + "learning_rate": 8.328903009316724e-06, + "loss": 0.1630345344543457, + "step": 21520 + }, + { + "epoch": 0.5671330214057813, + "grad_norm": 2.1693203449249268, + "learning_rate": 8.320492211393844e-06, + "loss": 0.1396716356277466, + "step": 21530 + }, + { + "epoch": 0.5673964366502801, + "grad_norm": 3.985605001449585, + "learning_rate": 8.312082635889193e-06, + "loss": 0.147645902633667, + "step": 21540 + }, + { + "epoch": 0.5676598518947787, + "grad_norm": 1.3176409006118774, + "learning_rate": 8.303674288923621e-06, + "loss": 0.15445961952209472, + "step": 21550 + }, + { + "epoch": 0.5679232671392775, + "grad_norm": 2.547032356262207, + "learning_rate": 8.29526717661709e-06, + "loss": 0.14077210426330566, + "step": 21560 + }, + { + "epoch": 0.5681866823837762, + "grad_norm": 1.8179384469985962, + "learning_rate": 8.286861305088655e-06, + "loss": 0.1411992311477661, + "step": 21570 + }, + { + "epoch": 0.568450097628275, + "grad_norm": 1.061055064201355, + "learning_rate": 8.278456680456476e-06, + "loss": 0.13582290410995485, + "step": 21580 + }, + { + "epoch": 0.5687135128727737, + "grad_norm": 1.9348623752593994, + "learning_rate": 8.270053308837791e-06, + "loss": 0.13533287048339843, + "step": 21590 + }, + { + "epoch": 0.5689769281172725, + "grad_norm": 1.9148125648498535, + "learning_rate": 8.261651196348953e-06, + "loss": 0.1391640305519104, + "step": 21600 + }, + { + "epoch": 0.5692403433617712, + "grad_norm": 2.2309772968292236, + "learning_rate": 8.253250349105369e-06, + "loss": 0.14293016195297242, + "step": 21610 + }, + { + "epoch": 0.5695037586062699, + "grad_norm": 1.1504936218261719, + "learning_rate": 8.24485077322154e-06, + "loss": 0.14539017677307128, + "step": 21620 + }, + { + "epoch": 0.5697671738507687, + "grad_norm": 1.653061032295227, + "learning_rate": 8.236452474811046e-06, + "loss": 0.1474233388900757, + "step": 21630 + }, + { + "epoch": 0.5700305890952674, + "grad_norm": 1.4097166061401367, + "learning_rate": 8.228055459986521e-06, + "loss": 0.14775147438049316, + "step": 21640 + }, + { + "epoch": 0.5702940043397662, + "grad_norm": 2.1024110317230225, + "learning_rate": 8.219659734859686e-06, + "loss": 0.13425831794738768, + "step": 21650 + }, + { + "epoch": 0.5705574195842649, + "grad_norm": 1.946166753768921, + "learning_rate": 8.211265305541304e-06, + "loss": 0.14790737628936768, + "step": 21660 + }, + { + "epoch": 0.5708208348287637, + "grad_norm": 2.3956305980682373, + "learning_rate": 8.202872178141204e-06, + "loss": 0.14793856143951417, + "step": 21670 + }, + { + "epoch": 0.5710842500732624, + "grad_norm": 5.550455093383789, + "learning_rate": 8.19448035876827e-06, + "loss": 0.13450241088867188, + "step": 21680 + }, + { + "epoch": 0.5713476653177612, + "grad_norm": 1.5088399648666382, + "learning_rate": 8.186089853530426e-06, + "loss": 0.14640257358551026, + "step": 21690 + }, + { + "epoch": 0.5716110805622598, + "grad_norm": 4.2838640213012695, + "learning_rate": 8.177700668534646e-06, + "loss": 0.14784672260284423, + "step": 21700 + }, + { + "epoch": 0.5718744958067585, + "grad_norm": 1.3914507627487183, + "learning_rate": 8.169312809886944e-06, + "loss": 0.1389720320701599, + "step": 21710 + }, + { + "epoch": 0.5721379110512573, + "grad_norm": 1.883269190788269, + "learning_rate": 8.160926283692355e-06, + "loss": 0.1411662220954895, + "step": 21720 + }, + { + "epoch": 0.572401326295756, + "grad_norm": 1.8045552968978882, + "learning_rate": 8.152541096054965e-06, + "loss": 0.1347530484199524, + "step": 21730 + }, + { + "epoch": 0.5726647415402548, + "grad_norm": 1.4020146131515503, + "learning_rate": 8.144157253077871e-06, + "loss": 0.1423660159111023, + "step": 21740 + }, + { + "epoch": 0.5729281567847535, + "grad_norm": 1.36337149143219, + "learning_rate": 8.135774760863193e-06, + "loss": 0.15066616535186766, + "step": 21750 + }, + { + "epoch": 0.5731915720292523, + "grad_norm": 0.9919650554656982, + "learning_rate": 8.127393625512077e-06, + "loss": 0.13566431999206544, + "step": 21760 + }, + { + "epoch": 0.573454987273751, + "grad_norm": 1.626855492591858, + "learning_rate": 8.11901385312467e-06, + "loss": 0.14926079511642457, + "step": 21770 + }, + { + "epoch": 0.5737184025182498, + "grad_norm": 1.497727632522583, + "learning_rate": 8.110635449800132e-06, + "loss": 0.14667437076568604, + "step": 21780 + }, + { + "epoch": 0.5739818177627485, + "grad_norm": 1.5982110500335693, + "learning_rate": 8.10225842163663e-06, + "loss": 0.1436631202697754, + "step": 21790 + }, + { + "epoch": 0.5742452330072472, + "grad_norm": 1.6620408296585083, + "learning_rate": 8.093882774731322e-06, + "loss": 0.15494861602783203, + "step": 21800 + }, + { + "epoch": 0.574508648251746, + "grad_norm": 2.80298113822937, + "learning_rate": 8.085508515180373e-06, + "loss": 0.14500644207000732, + "step": 21810 + }, + { + "epoch": 0.5747720634962447, + "grad_norm": 1.799157977104187, + "learning_rate": 8.077135649078917e-06, + "loss": 0.13523218631744385, + "step": 21820 + }, + { + "epoch": 0.5750354787407435, + "grad_norm": 2.5170514583587646, + "learning_rate": 8.068764182521102e-06, + "loss": 0.1479992985725403, + "step": 21830 + }, + { + "epoch": 0.5752988939852421, + "grad_norm": 2.192117929458618, + "learning_rate": 8.060394121600033e-06, + "loss": 0.1433668851852417, + "step": 21840 + }, + { + "epoch": 0.5755623092297409, + "grad_norm": 0.9620199799537659, + "learning_rate": 8.052025472407802e-06, + "loss": 0.13941656351089476, + "step": 21850 + }, + { + "epoch": 0.5758257244742396, + "grad_norm": 1.6182867288589478, + "learning_rate": 8.043658241035477e-06, + "loss": 0.14990029335021973, + "step": 21860 + }, + { + "epoch": 0.5760891397187383, + "grad_norm": 0.8902259469032288, + "learning_rate": 8.035292433573088e-06, + "loss": 0.1344151020050049, + "step": 21870 + }, + { + "epoch": 0.5763525549632371, + "grad_norm": 1.8648197650909424, + "learning_rate": 8.026928056109628e-06, + "loss": 0.13932644128799437, + "step": 21880 + }, + { + "epoch": 0.5766159702077358, + "grad_norm": 2.6698038578033447, + "learning_rate": 8.018565114733056e-06, + "loss": 0.1523970365524292, + "step": 21890 + }, + { + "epoch": 0.5768793854522346, + "grad_norm": 1.2656781673431396, + "learning_rate": 8.010203615530279e-06, + "loss": 0.13836041688919068, + "step": 21900 + }, + { + "epoch": 0.5771428006967333, + "grad_norm": 2.9047603607177734, + "learning_rate": 8.00184356458716e-06, + "loss": 0.14403009414672852, + "step": 21910 + }, + { + "epoch": 0.5774062159412321, + "grad_norm": 2.8234176635742188, + "learning_rate": 7.993484967988504e-06, + "loss": 0.14103100299835206, + "step": 21920 + }, + { + "epoch": 0.5776696311857308, + "grad_norm": 1.2861300706863403, + "learning_rate": 7.98512783181805e-06, + "loss": 0.13157203197479247, + "step": 21930 + }, + { + "epoch": 0.5779330464302296, + "grad_norm": 0.892197847366333, + "learning_rate": 7.976772162158498e-06, + "loss": 0.1567770481109619, + "step": 21940 + }, + { + "epoch": 0.5781964616747283, + "grad_norm": 1.6031218767166138, + "learning_rate": 7.968417965091449e-06, + "loss": 0.12900313138961791, + "step": 21950 + }, + { + "epoch": 0.578459876919227, + "grad_norm": 3.4371166229248047, + "learning_rate": 7.960065246697456e-06, + "loss": 0.14414687156677247, + "step": 21960 + }, + { + "epoch": 0.5787232921637258, + "grad_norm": 1.0754612684249878, + "learning_rate": 7.951714013055988e-06, + "loss": 0.1490562915802002, + "step": 21970 + }, + { + "epoch": 0.5789867074082244, + "grad_norm": 2.092714786529541, + "learning_rate": 7.94336427024543e-06, + "loss": 0.1413975477218628, + "step": 21980 + }, + { + "epoch": 0.5792501226527232, + "grad_norm": 2.0482113361358643, + "learning_rate": 7.93501602434309e-06, + "loss": 0.14299347400665283, + "step": 21990 + }, + { + "epoch": 0.5795135378972219, + "grad_norm": 1.4665892124176025, + "learning_rate": 7.926669281425177e-06, + "loss": 0.1469014525413513, + "step": 22000 + }, + { + "epoch": 0.5797769531417207, + "grad_norm": 2.2121827602386475, + "learning_rate": 7.91832404756681e-06, + "loss": 0.1542726516723633, + "step": 22010 + }, + { + "epoch": 0.5800403683862194, + "grad_norm": 2.275954246520996, + "learning_rate": 7.909980328842015e-06, + "loss": 0.14812233448028564, + "step": 22020 + }, + { + "epoch": 0.5803037836307182, + "grad_norm": 1.5012000799179077, + "learning_rate": 7.901638131323705e-06, + "loss": 0.13522322177886964, + "step": 22030 + }, + { + "epoch": 0.5805671988752169, + "grad_norm": 2.081918478012085, + "learning_rate": 7.893297461083697e-06, + "loss": 0.14270479679107667, + "step": 22040 + }, + { + "epoch": 0.5808306141197156, + "grad_norm": 2.1052303314208984, + "learning_rate": 7.884958324192689e-06, + "loss": 0.15447330474853516, + "step": 22050 + }, + { + "epoch": 0.5810940293642144, + "grad_norm": 2.020350456237793, + "learning_rate": 7.876620726720257e-06, + "loss": 0.1644977569580078, + "step": 22060 + }, + { + "epoch": 0.5813574446087131, + "grad_norm": 0.9711400270462036, + "learning_rate": 7.868284674734877e-06, + "loss": 0.14652594327926635, + "step": 22070 + }, + { + "epoch": 0.5816208598532119, + "grad_norm": 0.9338367581367493, + "learning_rate": 7.859950174303879e-06, + "loss": 0.1452716588973999, + "step": 22080 + }, + { + "epoch": 0.5818842750977106, + "grad_norm": 1.1187154054641724, + "learning_rate": 7.851617231493472e-06, + "loss": 0.14396626949310304, + "step": 22090 + }, + { + "epoch": 0.5821476903422094, + "grad_norm": 3.4363608360290527, + "learning_rate": 7.843285852368734e-06, + "loss": 0.1452549457550049, + "step": 22100 + }, + { + "epoch": 0.5824111055867081, + "grad_norm": 0.9405121803283691, + "learning_rate": 7.8349560429936e-06, + "loss": 0.1389599323272705, + "step": 22110 + }, + { + "epoch": 0.5826745208312069, + "grad_norm": 2.1479709148406982, + "learning_rate": 7.826627809430867e-06, + "loss": 0.14706158638000488, + "step": 22120 + }, + { + "epoch": 0.5829379360757055, + "grad_norm": 1.4259692430496216, + "learning_rate": 7.818301157742182e-06, + "loss": 0.14699466228485109, + "step": 22130 + }, + { + "epoch": 0.5832013513202042, + "grad_norm": 2.6266515254974365, + "learning_rate": 7.809976093988039e-06, + "loss": 0.1445547103881836, + "step": 22140 + }, + { + "epoch": 0.583464766564703, + "grad_norm": 2.1515979766845703, + "learning_rate": 7.801652624227782e-06, + "loss": 0.15275577306747437, + "step": 22150 + }, + { + "epoch": 0.5837281818092017, + "grad_norm": 2.2453622817993164, + "learning_rate": 7.79333075451959e-06, + "loss": 0.13521158695220947, + "step": 22160 + }, + { + "epoch": 0.5839915970537005, + "grad_norm": 2.1808154582977295, + "learning_rate": 7.785010490920477e-06, + "loss": 0.14260942935943605, + "step": 22170 + }, + { + "epoch": 0.5842550122981992, + "grad_norm": 2.67992901802063, + "learning_rate": 7.776691839486297e-06, + "loss": 0.13862948417663573, + "step": 22180 + }, + { + "epoch": 0.584518427542698, + "grad_norm": 1.1465553045272827, + "learning_rate": 7.76837480627171e-06, + "loss": 0.14144099950790406, + "step": 22190 + }, + { + "epoch": 0.5847818427871967, + "grad_norm": 2.872953176498413, + "learning_rate": 7.760059397330225e-06, + "loss": 0.13000528812408446, + "step": 22200 + }, + { + "epoch": 0.5850452580316955, + "grad_norm": 2.5530576705932617, + "learning_rate": 7.75174561871415e-06, + "loss": 0.1379254102706909, + "step": 22210 + }, + { + "epoch": 0.5853086732761942, + "grad_norm": 1.7738046646118164, + "learning_rate": 7.743433476474608e-06, + "loss": 0.1383064031600952, + "step": 22220 + }, + { + "epoch": 0.5855720885206929, + "grad_norm": 3.4087798595428467, + "learning_rate": 7.73512297666154e-06, + "loss": 0.13759169578552247, + "step": 22230 + }, + { + "epoch": 0.5858355037651917, + "grad_norm": 1.2587449550628662, + "learning_rate": 7.726814125323683e-06, + "loss": 0.1306830406188965, + "step": 22240 + }, + { + "epoch": 0.5860989190096904, + "grad_norm": 3.8443896770477295, + "learning_rate": 7.718506928508583e-06, + "loss": 0.1373496651649475, + "step": 22250 + }, + { + "epoch": 0.5863623342541892, + "grad_norm": 1.5308045148849487, + "learning_rate": 7.71020139226257e-06, + "loss": 0.1365554690361023, + "step": 22260 + }, + { + "epoch": 0.5866257494986878, + "grad_norm": 1.1082593202590942, + "learning_rate": 7.701897522630772e-06, + "loss": 0.13776953220367433, + "step": 22270 + }, + { + "epoch": 0.5868891647431866, + "grad_norm": 2.23307204246521, + "learning_rate": 7.693595325657109e-06, + "loss": 0.14689230918884277, + "step": 22280 + }, + { + "epoch": 0.5871525799876853, + "grad_norm": 0.7898665070533752, + "learning_rate": 7.685294807384273e-06, + "loss": 0.1355324864387512, + "step": 22290 + }, + { + "epoch": 0.5874159952321841, + "grad_norm": 1.8944190740585327, + "learning_rate": 7.676995973853743e-06, + "loss": 0.13793805837631226, + "step": 22300 + }, + { + "epoch": 0.5876794104766828, + "grad_norm": 1.9833056926727295, + "learning_rate": 7.668698831105768e-06, + "loss": 0.12725415229797363, + "step": 22310 + }, + { + "epoch": 0.5879428257211815, + "grad_norm": 0.9674490690231323, + "learning_rate": 7.66040338517936e-06, + "loss": 0.15150362253189087, + "step": 22320 + }, + { + "epoch": 0.5882062409656803, + "grad_norm": 1.0702906847000122, + "learning_rate": 7.652109642112308e-06, + "loss": 0.14388723373413087, + "step": 22330 + }, + { + "epoch": 0.588469656210179, + "grad_norm": 0.7703921794891357, + "learning_rate": 7.643817607941157e-06, + "loss": 0.1264678955078125, + "step": 22340 + }, + { + "epoch": 0.5887330714546778, + "grad_norm": 1.8987869024276733, + "learning_rate": 7.635527288701201e-06, + "loss": 0.1320556402206421, + "step": 22350 + }, + { + "epoch": 0.5889964866991765, + "grad_norm": 2.0078439712524414, + "learning_rate": 7.6272386904265005e-06, + "loss": 0.1439530849456787, + "step": 22360 + }, + { + "epoch": 0.5892599019436753, + "grad_norm": 1.1273033618927002, + "learning_rate": 7.618951819149848e-06, + "loss": 0.1328580856323242, + "step": 22370 + }, + { + "epoch": 0.589523317188174, + "grad_norm": 1.2614485025405884, + "learning_rate": 7.610666680902791e-06, + "loss": 0.12047417163848877, + "step": 22380 + }, + { + "epoch": 0.5897867324326727, + "grad_norm": 1.0242893695831299, + "learning_rate": 7.602383281715608e-06, + "loss": 0.1306384325027466, + "step": 22390 + }, + { + "epoch": 0.5900501476771715, + "grad_norm": 3.46950101852417, + "learning_rate": 7.594101627617312e-06, + "loss": 0.13199341297149658, + "step": 22400 + }, + { + "epoch": 0.5903135629216701, + "grad_norm": 2.7481963634490967, + "learning_rate": 7.585821724635656e-06, + "loss": 0.14063292741775513, + "step": 22410 + }, + { + "epoch": 0.5905769781661689, + "grad_norm": 2.1436681747436523, + "learning_rate": 7.577543578797097e-06, + "loss": 0.137488055229187, + "step": 22420 + }, + { + "epoch": 0.5908403934106676, + "grad_norm": 4.7881999015808105, + "learning_rate": 7.569267196126844e-06, + "loss": 0.13316417932510377, + "step": 22430 + }, + { + "epoch": 0.5911038086551664, + "grad_norm": 1.6547141075134277, + "learning_rate": 7.560992582648793e-06, + "loss": 0.15307302474975587, + "step": 22440 + }, + { + "epoch": 0.5913672238996651, + "grad_norm": 0.8127312660217285, + "learning_rate": 7.552719744385566e-06, + "loss": 0.13826363086700438, + "step": 22450 + }, + { + "epoch": 0.5916306391441639, + "grad_norm": 1.0936634540557861, + "learning_rate": 7.544448687358495e-06, + "loss": 0.14720991849899293, + "step": 22460 + }, + { + "epoch": 0.5918940543886626, + "grad_norm": 1.0550200939178467, + "learning_rate": 7.536179417587612e-06, + "loss": 0.14581931829452516, + "step": 22470 + }, + { + "epoch": 0.5921574696331613, + "grad_norm": 2.1047282218933105, + "learning_rate": 7.527911941091644e-06, + "loss": 0.13667354583740235, + "step": 22480 + }, + { + "epoch": 0.5924208848776601, + "grad_norm": 2.4748458862304688, + "learning_rate": 7.519646263888023e-06, + "loss": 0.1502878785133362, + "step": 22490 + }, + { + "epoch": 0.5926843001221588, + "grad_norm": 1.3406200408935547, + "learning_rate": 7.511382391992861e-06, + "loss": 0.13202352523803712, + "step": 22500 + }, + { + "epoch": 0.5929477153666576, + "grad_norm": 2.366527795791626, + "learning_rate": 7.503120331420963e-06, + "loss": 0.14055700302124025, + "step": 22510 + }, + { + "epoch": 0.5932111306111563, + "grad_norm": 1.506205439567566, + "learning_rate": 7.494860088185815e-06, + "loss": 0.13953958749771117, + "step": 22520 + }, + { + "epoch": 0.5934745458556551, + "grad_norm": 1.0939046144485474, + "learning_rate": 7.48660166829957e-06, + "loss": 0.1406635046005249, + "step": 22530 + }, + { + "epoch": 0.5937379611001538, + "grad_norm": 1.8596951961517334, + "learning_rate": 7.478345077773075e-06, + "loss": 0.1388508439064026, + "step": 22540 + }, + { + "epoch": 0.5940013763446526, + "grad_norm": 1.9868288040161133, + "learning_rate": 7.470090322615822e-06, + "loss": 0.15146961212158203, + "step": 22550 + }, + { + "epoch": 0.5942647915891512, + "grad_norm": 3.519893169403076, + "learning_rate": 7.4618374088359805e-06, + "loss": 0.14383569955825806, + "step": 22560 + }, + { + "epoch": 0.5945282068336499, + "grad_norm": 0.9257417917251587, + "learning_rate": 7.45358634244038e-06, + "loss": 0.1353355050086975, + "step": 22570 + }, + { + "epoch": 0.5947916220781487, + "grad_norm": 2.752671003341675, + "learning_rate": 7.4453371294345e-06, + "loss": 0.13987044095993043, + "step": 22580 + }, + { + "epoch": 0.5950550373226474, + "grad_norm": 1.3669841289520264, + "learning_rate": 7.437089775822477e-06, + "loss": 0.1574385643005371, + "step": 22590 + }, + { + "epoch": 0.5953184525671462, + "grad_norm": 2.649716377258301, + "learning_rate": 7.4288442876070865e-06, + "loss": 0.14281003475189208, + "step": 22600 + }, + { + "epoch": 0.5955818678116449, + "grad_norm": 1.809186339378357, + "learning_rate": 7.420600670789753e-06, + "loss": 0.14608203172683715, + "step": 22610 + }, + { + "epoch": 0.5958452830561437, + "grad_norm": 1.5172547101974487, + "learning_rate": 7.412358931370536e-06, + "loss": 0.13955216407775878, + "step": 22620 + }, + { + "epoch": 0.5961086983006424, + "grad_norm": 1.1923365592956543, + "learning_rate": 7.404119075348129e-06, + "loss": 0.13791170120239257, + "step": 22630 + }, + { + "epoch": 0.5963721135451412, + "grad_norm": 1.155256748199463, + "learning_rate": 7.3958811087198555e-06, + "loss": 0.13895280361175538, + "step": 22640 + }, + { + "epoch": 0.5966355287896399, + "grad_norm": 1.5260686874389648, + "learning_rate": 7.3876450374816644e-06, + "loss": 0.1286543846130371, + "step": 22650 + }, + { + "epoch": 0.5968989440341386, + "grad_norm": 2.426647901535034, + "learning_rate": 7.379410867628115e-06, + "loss": 0.1391974925994873, + "step": 22660 + }, + { + "epoch": 0.5971623592786374, + "grad_norm": 1.2296149730682373, + "learning_rate": 7.371178605152405e-06, + "loss": 0.14388325214385986, + "step": 22670 + }, + { + "epoch": 0.597425774523136, + "grad_norm": 1.1288949251174927, + "learning_rate": 7.362948256046321e-06, + "loss": 0.125488018989563, + "step": 22680 + }, + { + "epoch": 0.5976891897676349, + "grad_norm": 1.915586233139038, + "learning_rate": 7.354719826300266e-06, + "loss": 0.14120378494262695, + "step": 22690 + }, + { + "epoch": 0.5979526050121335, + "grad_norm": 3.396127223968506, + "learning_rate": 7.346493321903251e-06, + "loss": 0.13166370391845703, + "step": 22700 + }, + { + "epoch": 0.5982160202566323, + "grad_norm": 2.5027034282684326, + "learning_rate": 7.338268748842876e-06, + "loss": 0.13278889656066895, + "step": 22710 + }, + { + "epoch": 0.598479435501131, + "grad_norm": 4.3556809425354, + "learning_rate": 7.330046113105345e-06, + "loss": 0.14672384262084961, + "step": 22720 + }, + { + "epoch": 0.5987428507456298, + "grad_norm": 2.191594123840332, + "learning_rate": 7.321825420675444e-06, + "loss": 0.12191802263259888, + "step": 22730 + }, + { + "epoch": 0.5990062659901285, + "grad_norm": 1.0853569507598877, + "learning_rate": 7.3136066775365455e-06, + "loss": 0.1309959650039673, + "step": 22740 + }, + { + "epoch": 0.5992696812346272, + "grad_norm": 1.4965956211090088, + "learning_rate": 7.305389889670611e-06, + "loss": 0.12787082195281982, + "step": 22750 + }, + { + "epoch": 0.599533096479126, + "grad_norm": 2.431267261505127, + "learning_rate": 7.2971750630581675e-06, + "loss": 0.14571905136108398, + "step": 22760 + }, + { + "epoch": 0.5997965117236247, + "grad_norm": 1.583493709564209, + "learning_rate": 7.2889622036783245e-06, + "loss": 0.13373968601226807, + "step": 22770 + }, + { + "epoch": 0.6000599269681235, + "grad_norm": 1.4441906213760376, + "learning_rate": 7.280751317508756e-06, + "loss": 0.13663046360015868, + "step": 22780 + }, + { + "epoch": 0.6003233422126222, + "grad_norm": 2.163827896118164, + "learning_rate": 7.272542410525691e-06, + "loss": 0.13905130624771117, + "step": 22790 + }, + { + "epoch": 0.600586757457121, + "grad_norm": 3.2356021404266357, + "learning_rate": 7.2643354887039395e-06, + "loss": 0.14402172565460206, + "step": 22800 + }, + { + "epoch": 0.6008501727016197, + "grad_norm": 1.9611421823501587, + "learning_rate": 7.256130558016843e-06, + "loss": 0.1314738392829895, + "step": 22810 + }, + { + "epoch": 0.6011135879461185, + "grad_norm": 1.369958519935608, + "learning_rate": 7.247927624436304e-06, + "loss": 0.1422540783882141, + "step": 22820 + }, + { + "epoch": 0.6013770031906172, + "grad_norm": 3.498368263244629, + "learning_rate": 7.239726693932779e-06, + "loss": 0.14050543308258057, + "step": 22830 + }, + { + "epoch": 0.6016404184351158, + "grad_norm": 1.1335395574569702, + "learning_rate": 7.2315277724752495e-06, + "loss": 0.12277098894119262, + "step": 22840 + }, + { + "epoch": 0.6019038336796146, + "grad_norm": 1.5519561767578125, + "learning_rate": 7.223330866031253e-06, + "loss": 0.13981045484542848, + "step": 22850 + }, + { + "epoch": 0.6021672489241133, + "grad_norm": 1.5562304258346558, + "learning_rate": 7.215135980566847e-06, + "loss": 0.12953239679336548, + "step": 22860 + }, + { + "epoch": 0.6024306641686121, + "grad_norm": 2.1523730754852295, + "learning_rate": 7.2069431220466215e-06, + "loss": 0.14343478679656982, + "step": 22870 + }, + { + "epoch": 0.6026940794131108, + "grad_norm": 2.5678439140319824, + "learning_rate": 7.198752296433697e-06, + "loss": 0.13722772598266603, + "step": 22880 + }, + { + "epoch": 0.6029574946576096, + "grad_norm": 1.0840282440185547, + "learning_rate": 7.190563509689705e-06, + "loss": 0.1326589822769165, + "step": 22890 + }, + { + "epoch": 0.6032209099021083, + "grad_norm": 2.5575218200683594, + "learning_rate": 7.182376767774803e-06, + "loss": 0.1361541748046875, + "step": 22900 + }, + { + "epoch": 0.603484325146607, + "grad_norm": 2.312211751937866, + "learning_rate": 7.174192076647653e-06, + "loss": 0.14027760028839112, + "step": 22910 + }, + { + "epoch": 0.6037477403911058, + "grad_norm": 1.1340315341949463, + "learning_rate": 7.166009442265423e-06, + "loss": 0.1324375867843628, + "step": 22920 + }, + { + "epoch": 0.6040111556356045, + "grad_norm": 2.077333450317383, + "learning_rate": 7.15782887058379e-06, + "loss": 0.1285334587097168, + "step": 22930 + }, + { + "epoch": 0.6042745708801033, + "grad_norm": 2.273592472076416, + "learning_rate": 7.149650367556926e-06, + "loss": 0.13435734510421754, + "step": 22940 + }, + { + "epoch": 0.604537986124602, + "grad_norm": 1.3225858211517334, + "learning_rate": 7.141473939137498e-06, + "loss": 0.1347590446472168, + "step": 22950 + }, + { + "epoch": 0.6048014013691008, + "grad_norm": 2.232646942138672, + "learning_rate": 7.133299591276665e-06, + "loss": 0.1392536163330078, + "step": 22960 + }, + { + "epoch": 0.6050648166135995, + "grad_norm": 4.303405284881592, + "learning_rate": 7.12512732992407e-06, + "loss": 0.13265571594238282, + "step": 22970 + }, + { + "epoch": 0.6053282318580983, + "grad_norm": 2.1359996795654297, + "learning_rate": 7.1169571610278375e-06, + "loss": 0.1523197054862976, + "step": 22980 + }, + { + "epoch": 0.6055916471025969, + "grad_norm": 1.4149458408355713, + "learning_rate": 7.108789090534569e-06, + "loss": 0.129486083984375, + "step": 22990 + }, + { + "epoch": 0.6058550623470956, + "grad_norm": 1.5227822065353394, + "learning_rate": 7.100623124389334e-06, + "loss": 0.12914557456970216, + "step": 23000 + }, + { + "epoch": 0.6061184775915944, + "grad_norm": 1.860114574432373, + "learning_rate": 7.092459268535685e-06, + "loss": 0.1379807710647583, + "step": 23010 + }, + { + "epoch": 0.6063818928360931, + "grad_norm": 2.470343828201294, + "learning_rate": 7.084297528915616e-06, + "loss": 0.14903533458709717, + "step": 23020 + }, + { + "epoch": 0.6066453080805919, + "grad_norm": 1.687584638595581, + "learning_rate": 7.076137911469603e-06, + "loss": 0.13147525787353515, + "step": 23030 + }, + { + "epoch": 0.6069087233250906, + "grad_norm": 2.4215316772460938, + "learning_rate": 7.067980422136565e-06, + "loss": 0.13636424541473388, + "step": 23040 + }, + { + "epoch": 0.6071721385695894, + "grad_norm": 3.1996517181396484, + "learning_rate": 7.059825066853872e-06, + "loss": 0.1468991756439209, + "step": 23050 + }, + { + "epoch": 0.6074355538140881, + "grad_norm": 1.674701452255249, + "learning_rate": 7.0516718515573445e-06, + "loss": 0.14027292728424073, + "step": 23060 + }, + { + "epoch": 0.6076989690585869, + "grad_norm": 1.2875831127166748, + "learning_rate": 7.043520782181248e-06, + "loss": 0.1321595549583435, + "step": 23070 + }, + { + "epoch": 0.6079623843030856, + "grad_norm": 1.137254238128662, + "learning_rate": 7.035371864658276e-06, + "loss": 0.14105639457702637, + "step": 23080 + }, + { + "epoch": 0.6082257995475843, + "grad_norm": 3.110307216644287, + "learning_rate": 7.0272251049195684e-06, + "loss": 0.13282785415649415, + "step": 23090 + }, + { + "epoch": 0.6084892147920831, + "grad_norm": 1.6299216747283936, + "learning_rate": 7.019080508894683e-06, + "loss": 0.14454102516174316, + "step": 23100 + }, + { + "epoch": 0.6087526300365818, + "grad_norm": 1.070438027381897, + "learning_rate": 7.010938082511614e-06, + "loss": 0.1334700584411621, + "step": 23110 + }, + { + "epoch": 0.6090160452810806, + "grad_norm": 1.2572892904281616, + "learning_rate": 7.0027978316967705e-06, + "loss": 0.12693685293197632, + "step": 23120 + }, + { + "epoch": 0.6092794605255792, + "grad_norm": 2.1366989612579346, + "learning_rate": 6.994659762374967e-06, + "loss": 0.12994059324264526, + "step": 23130 + }, + { + "epoch": 0.609542875770078, + "grad_norm": 1.7777999639511108, + "learning_rate": 6.986523880469458e-06, + "loss": 0.14147584438323973, + "step": 23140 + }, + { + "epoch": 0.6098062910145767, + "grad_norm": 1.7626532316207886, + "learning_rate": 6.978390191901878e-06, + "loss": 0.15412578582763672, + "step": 23150 + }, + { + "epoch": 0.6100697062590755, + "grad_norm": 2.307338237762451, + "learning_rate": 6.970258702592277e-06, + "loss": 0.14650497436523438, + "step": 23160 + }, + { + "epoch": 0.6103331215035742, + "grad_norm": 1.1785070896148682, + "learning_rate": 6.962129418459109e-06, + "loss": 0.1301561713218689, + "step": 23170 + }, + { + "epoch": 0.6105965367480729, + "grad_norm": 1.2437578439712524, + "learning_rate": 6.954002345419214e-06, + "loss": 0.1371382713317871, + "step": 23180 + }, + { + "epoch": 0.6108599519925717, + "grad_norm": 1.7616289854049683, + "learning_rate": 6.945877489387829e-06, + "loss": 0.1446597695350647, + "step": 23190 + }, + { + "epoch": 0.6111233672370704, + "grad_norm": 0.9418478608131409, + "learning_rate": 6.9377548562785735e-06, + "loss": 0.1419245719909668, + "step": 23200 + }, + { + "epoch": 0.6113867824815692, + "grad_norm": 3.2602951526641846, + "learning_rate": 6.929634452003447e-06, + "loss": 0.13840678930282593, + "step": 23210 + }, + { + "epoch": 0.6116501977260679, + "grad_norm": 2.0433974266052246, + "learning_rate": 6.921516282472837e-06, + "loss": 0.14857593774795533, + "step": 23220 + }, + { + "epoch": 0.6119136129705667, + "grad_norm": 2.117244243621826, + "learning_rate": 6.91340035359549e-06, + "loss": 0.14924811124801635, + "step": 23230 + }, + { + "epoch": 0.6121770282150654, + "grad_norm": 1.4525434970855713, + "learning_rate": 6.905286671278536e-06, + "loss": 0.1421501398086548, + "step": 23240 + }, + { + "epoch": 0.6124404434595642, + "grad_norm": 3.0995590686798096, + "learning_rate": 6.897175241427462e-06, + "loss": 0.13361562490463258, + "step": 23250 + }, + { + "epoch": 0.6127038587040629, + "grad_norm": 1.2748833894729614, + "learning_rate": 6.889066069946107e-06, + "loss": 0.14257099628448486, + "step": 23260 + }, + { + "epoch": 0.6129672739485615, + "grad_norm": 1.3205608129501343, + "learning_rate": 6.880959162736691e-06, + "loss": 0.1445458173751831, + "step": 23270 + }, + { + "epoch": 0.6132306891930603, + "grad_norm": 1.6428927183151245, + "learning_rate": 6.872854525699761e-06, + "loss": 0.14134978055953978, + "step": 23280 + }, + { + "epoch": 0.613494104437559, + "grad_norm": 2.1511592864990234, + "learning_rate": 6.8647521647342205e-06, + "loss": 0.13768391609191893, + "step": 23290 + }, + { + "epoch": 0.6137575196820578, + "grad_norm": 1.6408647298812866, + "learning_rate": 6.856652085737323e-06, + "loss": 0.14617904424667358, + "step": 23300 + }, + { + "epoch": 0.6140209349265565, + "grad_norm": 2.0751960277557373, + "learning_rate": 6.848554294604652e-06, + "loss": 0.13581932783126832, + "step": 23310 + }, + { + "epoch": 0.6142843501710553, + "grad_norm": 1.047861099243164, + "learning_rate": 6.8404587972301316e-06, + "loss": 0.12902594804763795, + "step": 23320 + }, + { + "epoch": 0.614547765415554, + "grad_norm": 1.8386720418930054, + "learning_rate": 6.8323655995060135e-06, + "loss": 0.15323374271392823, + "step": 23330 + }, + { + "epoch": 0.6148111806600528, + "grad_norm": 1.575285792350769, + "learning_rate": 6.824274707322874e-06, + "loss": 0.1460517168045044, + "step": 23340 + }, + { + "epoch": 0.6150745959045515, + "grad_norm": 2.85732364654541, + "learning_rate": 6.816186126569615e-06, + "loss": 0.14838056564331054, + "step": 23350 + }, + { + "epoch": 0.6153380111490502, + "grad_norm": 3.768015146255493, + "learning_rate": 6.808099863133453e-06, + "loss": 0.15101315975189208, + "step": 23360 + }, + { + "epoch": 0.615601426393549, + "grad_norm": 1.258283257484436, + "learning_rate": 6.800015922899923e-06, + "loss": 0.1415132164955139, + "step": 23370 + }, + { + "epoch": 0.6158648416380477, + "grad_norm": 1.549623966217041, + "learning_rate": 6.7919343117528655e-06, + "loss": 0.1413954019546509, + "step": 23380 + }, + { + "epoch": 0.6161282568825465, + "grad_norm": 2.6203057765960693, + "learning_rate": 6.783855035574418e-06, + "loss": 0.1400364875793457, + "step": 23390 + }, + { + "epoch": 0.6163916721270452, + "grad_norm": 2.141921281814575, + "learning_rate": 6.775778100245038e-06, + "loss": 0.1464184284210205, + "step": 23400 + }, + { + "epoch": 0.616655087371544, + "grad_norm": 3.0080294609069824, + "learning_rate": 6.76770351164346e-06, + "loss": 0.13260605335235595, + "step": 23410 + }, + { + "epoch": 0.6169185026160426, + "grad_norm": 3.822356700897217, + "learning_rate": 6.759631275646715e-06, + "loss": 0.13888609409332275, + "step": 23420 + }, + { + "epoch": 0.6171819178605413, + "grad_norm": 2.8329851627349854, + "learning_rate": 6.751561398130133e-06, + "loss": 0.14069108963012694, + "step": 23430 + }, + { + "epoch": 0.6174453331050401, + "grad_norm": 1.759352207183838, + "learning_rate": 6.743493884967312e-06, + "loss": 0.1383940815925598, + "step": 23440 + }, + { + "epoch": 0.6177087483495388, + "grad_norm": 2.4267189502716064, + "learning_rate": 6.735428742030141e-06, + "loss": 0.12434496879577636, + "step": 23450 + }, + { + "epoch": 0.6179721635940376, + "grad_norm": 1.7214387655258179, + "learning_rate": 6.7273659751887775e-06, + "loss": 0.13996971845626832, + "step": 23460 + }, + { + "epoch": 0.6182355788385363, + "grad_norm": 1.5868018865585327, + "learning_rate": 6.719305590311649e-06, + "loss": 0.14228546619415283, + "step": 23470 + }, + { + "epoch": 0.6184989940830351, + "grad_norm": 2.0611324310302734, + "learning_rate": 6.7112475932654554e-06, + "loss": 0.13636562824249268, + "step": 23480 + }, + { + "epoch": 0.6187624093275338, + "grad_norm": 1.9623517990112305, + "learning_rate": 6.70319198991515e-06, + "loss": 0.13627111911773682, + "step": 23490 + }, + { + "epoch": 0.6190258245720326, + "grad_norm": 1.8680540323257446, + "learning_rate": 6.695138786123952e-06, + "loss": 0.13240907192230225, + "step": 23500 + }, + { + "epoch": 0.6192892398165313, + "grad_norm": 1.1381889581680298, + "learning_rate": 6.687087987753331e-06, + "loss": 0.13277069330215455, + "step": 23510 + }, + { + "epoch": 0.61955265506103, + "grad_norm": 2.5393826961517334, + "learning_rate": 6.679039600663001e-06, + "loss": 0.14505046606063843, + "step": 23520 + }, + { + "epoch": 0.6198160703055288, + "grad_norm": 0.9177750945091248, + "learning_rate": 6.670993630710929e-06, + "loss": 0.13561534881591797, + "step": 23530 + }, + { + "epoch": 0.6200794855500275, + "grad_norm": 2.530397891998291, + "learning_rate": 6.662950083753319e-06, + "loss": 0.12293534278869629, + "step": 23540 + }, + { + "epoch": 0.6203429007945263, + "grad_norm": 1.850788950920105, + "learning_rate": 6.654908965644609e-06, + "loss": 0.1451024055480957, + "step": 23550 + }, + { + "epoch": 0.6206063160390249, + "grad_norm": 2.1883656978607178, + "learning_rate": 6.646870282237474e-06, + "loss": 0.13079228401184081, + "step": 23560 + }, + { + "epoch": 0.6208697312835237, + "grad_norm": 1.631291151046753, + "learning_rate": 6.6388340393828125e-06, + "loss": 0.1421767830848694, + "step": 23570 + }, + { + "epoch": 0.6211331465280224, + "grad_norm": 2.2631242275238037, + "learning_rate": 6.6308002429297516e-06, + "loss": 0.1322676420211792, + "step": 23580 + }, + { + "epoch": 0.6213965617725212, + "grad_norm": 1.5092486143112183, + "learning_rate": 6.622768898725634e-06, + "loss": 0.12232050895690919, + "step": 23590 + }, + { + "epoch": 0.6216599770170199, + "grad_norm": 2.8269600868225098, + "learning_rate": 6.614740012616017e-06, + "loss": 0.1418905258178711, + "step": 23600 + }, + { + "epoch": 0.6219233922615186, + "grad_norm": 2.9493091106414795, + "learning_rate": 6.606713590444673e-06, + "loss": 0.13100000619888305, + "step": 23610 + }, + { + "epoch": 0.6221868075060174, + "grad_norm": 2.7755534648895264, + "learning_rate": 6.598689638053577e-06, + "loss": 0.1367103099822998, + "step": 23620 + }, + { + "epoch": 0.6224502227505161, + "grad_norm": 1.7308557033538818, + "learning_rate": 6.590668161282903e-06, + "loss": 0.13652374744415283, + "step": 23630 + }, + { + "epoch": 0.6227136379950149, + "grad_norm": 1.887300729751587, + "learning_rate": 6.582649165971037e-06, + "loss": 0.1373499870300293, + "step": 23640 + }, + { + "epoch": 0.6229770532395136, + "grad_norm": 0.871224582195282, + "learning_rate": 6.574632657954542e-06, + "loss": 0.12844521999359132, + "step": 23650 + }, + { + "epoch": 0.6232404684840124, + "grad_norm": 1.35122811794281, + "learning_rate": 6.5666186430681835e-06, + "loss": 0.1271742582321167, + "step": 23660 + }, + { + "epoch": 0.6235038837285111, + "grad_norm": 1.7217661142349243, + "learning_rate": 6.558607127144905e-06, + "loss": 0.12657018899917602, + "step": 23670 + }, + { + "epoch": 0.6237672989730099, + "grad_norm": 2.6903531551361084, + "learning_rate": 6.550598116015832e-06, + "loss": 0.132509183883667, + "step": 23680 + }, + { + "epoch": 0.6240307142175086, + "grad_norm": 4.978670597076416, + "learning_rate": 6.54259161551027e-06, + "loss": 0.14487918615341186, + "step": 23690 + }, + { + "epoch": 0.6242941294620072, + "grad_norm": 1.5900905132293701, + "learning_rate": 6.534587631455696e-06, + "loss": 0.13520666360855102, + "step": 23700 + }, + { + "epoch": 0.624557544706506, + "grad_norm": 1.000401258468628, + "learning_rate": 6.5265861696777556e-06, + "loss": 0.14731879234313966, + "step": 23710 + }, + { + "epoch": 0.6248209599510047, + "grad_norm": 2.9675021171569824, + "learning_rate": 6.518587236000259e-06, + "loss": 0.13353707790374755, + "step": 23720 + }, + { + "epoch": 0.6250843751955035, + "grad_norm": 2.379964828491211, + "learning_rate": 6.510590836245166e-06, + "loss": 0.12668304443359374, + "step": 23730 + }, + { + "epoch": 0.6253477904400022, + "grad_norm": 1.407075047492981, + "learning_rate": 6.502596976232616e-06, + "loss": 0.13311684131622314, + "step": 23740 + }, + { + "epoch": 0.625611205684501, + "grad_norm": 2.5020546913146973, + "learning_rate": 6.494605661780874e-06, + "loss": 0.13292922973632812, + "step": 23750 + }, + { + "epoch": 0.6258746209289997, + "grad_norm": 1.9719411134719849, + "learning_rate": 6.486616898706365e-06, + "loss": 0.1364532470703125, + "step": 23760 + }, + { + "epoch": 0.6261380361734985, + "grad_norm": 1.0826867818832397, + "learning_rate": 6.47863069282366e-06, + "loss": 0.13406839370727539, + "step": 23770 + }, + { + "epoch": 0.6264014514179972, + "grad_norm": 1.718273401260376, + "learning_rate": 6.47064704994546e-06, + "loss": 0.13121000528335572, + "step": 23780 + }, + { + "epoch": 0.6266648666624959, + "grad_norm": 1.6053166389465332, + "learning_rate": 6.462665975882607e-06, + "loss": 0.14020037651062012, + "step": 23790 + }, + { + "epoch": 0.6269282819069947, + "grad_norm": 2.723588466644287, + "learning_rate": 6.454687476444071e-06, + "loss": 0.13197112083435059, + "step": 23800 + }, + { + "epoch": 0.6271916971514934, + "grad_norm": 1.447771668434143, + "learning_rate": 6.446711557436944e-06, + "loss": 0.12310125827789306, + "step": 23810 + }, + { + "epoch": 0.6274551123959922, + "grad_norm": 1.4374295473098755, + "learning_rate": 6.438738224666452e-06, + "loss": 0.13401618003845214, + "step": 23820 + }, + { + "epoch": 0.6277185276404909, + "grad_norm": 1.5970851182937622, + "learning_rate": 6.4307674839359225e-06, + "loss": 0.13583881855010987, + "step": 23830 + }, + { + "epoch": 0.6279819428849897, + "grad_norm": 1.459290862083435, + "learning_rate": 6.42279934104681e-06, + "loss": 0.1332627773284912, + "step": 23840 + }, + { + "epoch": 0.6282453581294883, + "grad_norm": 2.1865975856781006, + "learning_rate": 6.4148338017986725e-06, + "loss": 0.13154213428497313, + "step": 23850 + }, + { + "epoch": 0.6285087733739871, + "grad_norm": 1.3428059816360474, + "learning_rate": 6.406870871989166e-06, + "loss": 0.12098588943481445, + "step": 23860 + }, + { + "epoch": 0.6287721886184858, + "grad_norm": 1.5922534465789795, + "learning_rate": 6.398910557414064e-06, + "loss": 0.1393895149230957, + "step": 23870 + }, + { + "epoch": 0.6290356038629845, + "grad_norm": 1.440024971961975, + "learning_rate": 6.390952863867224e-06, + "loss": 0.12373380661010742, + "step": 23880 + }, + { + "epoch": 0.6292990191074833, + "grad_norm": 2.75650691986084, + "learning_rate": 6.382997797140591e-06, + "loss": 0.1447679042816162, + "step": 23890 + }, + { + "epoch": 0.629562434351982, + "grad_norm": 1.4112900495529175, + "learning_rate": 6.375045363024217e-06, + "loss": 0.13446015119552612, + "step": 23900 + }, + { + "epoch": 0.6298258495964808, + "grad_norm": 2.5232276916503906, + "learning_rate": 6.367095567306219e-06, + "loss": 0.1448499321937561, + "step": 23910 + }, + { + "epoch": 0.6300892648409795, + "grad_norm": 1.5831512212753296, + "learning_rate": 6.359148415772805e-06, + "loss": 0.11496567726135254, + "step": 23920 + }, + { + "epoch": 0.6303526800854783, + "grad_norm": 2.815788745880127, + "learning_rate": 6.351203914208254e-06, + "loss": 0.14093097448348998, + "step": 23930 + }, + { + "epoch": 0.630616095329977, + "grad_norm": 1.1019055843353271, + "learning_rate": 6.343262068394914e-06, + "loss": 0.13895230293273925, + "step": 23940 + }, + { + "epoch": 0.6308795105744757, + "grad_norm": 2.6164774894714355, + "learning_rate": 6.335322884113208e-06, + "loss": 0.14676634073257447, + "step": 23950 + }, + { + "epoch": 0.6311429258189745, + "grad_norm": 2.8587896823883057, + "learning_rate": 6.3273863671416146e-06, + "loss": 0.13942902088165282, + "step": 23960 + }, + { + "epoch": 0.6314063410634732, + "grad_norm": 3.404776096343994, + "learning_rate": 6.319452523256673e-06, + "loss": 0.1304668068885803, + "step": 23970 + }, + { + "epoch": 0.631669756307972, + "grad_norm": 1.9187800884246826, + "learning_rate": 6.31152135823298e-06, + "loss": 0.13686647415161132, + "step": 23980 + }, + { + "epoch": 0.6319331715524706, + "grad_norm": 1.1005117893218994, + "learning_rate": 6.303592877843171e-06, + "loss": 0.1328800916671753, + "step": 23990 + }, + { + "epoch": 0.6321965867969694, + "grad_norm": 2.238279342651367, + "learning_rate": 6.295667087857948e-06, + "loss": 0.13052172660827638, + "step": 24000 + }, + { + "epoch": 0.6324600020414681, + "grad_norm": 3.5310909748077393, + "learning_rate": 6.287743994046035e-06, + "loss": 0.1326054334640503, + "step": 24010 + }, + { + "epoch": 0.6327234172859669, + "grad_norm": 4.462266445159912, + "learning_rate": 6.279823602174202e-06, + "loss": 0.1318842887878418, + "step": 24020 + }, + { + "epoch": 0.6329868325304656, + "grad_norm": 3.313356876373291, + "learning_rate": 6.271905918007255e-06, + "loss": 0.14427293539047242, + "step": 24030 + }, + { + "epoch": 0.6332502477749643, + "grad_norm": 2.022413969039917, + "learning_rate": 6.263990947308022e-06, + "loss": 0.14116895198822021, + "step": 24040 + }, + { + "epoch": 0.6335136630194631, + "grad_norm": 1.2693536281585693, + "learning_rate": 6.2560786958373645e-06, + "loss": 0.14537339210510253, + "step": 24050 + }, + { + "epoch": 0.6337770782639618, + "grad_norm": 1.305526614189148, + "learning_rate": 6.248169169354159e-06, + "loss": 0.13658607006072998, + "step": 24060 + }, + { + "epoch": 0.6340404935084606, + "grad_norm": 1.4738647937774658, + "learning_rate": 6.240262373615298e-06, + "loss": 0.1283488988876343, + "step": 24070 + }, + { + "epoch": 0.6343039087529593, + "grad_norm": 1.0887134075164795, + "learning_rate": 6.232358314375693e-06, + "loss": 0.12539262771606446, + "step": 24080 + }, + { + "epoch": 0.6345673239974581, + "grad_norm": 1.6956242322921753, + "learning_rate": 6.224456997388254e-06, + "loss": 0.12633587121963502, + "step": 24090 + }, + { + "epoch": 0.6348307392419568, + "grad_norm": 2.2638680934906006, + "learning_rate": 6.2165584284039046e-06, + "loss": 0.1262298583984375, + "step": 24100 + }, + { + "epoch": 0.6350941544864556, + "grad_norm": 3.3484742641448975, + "learning_rate": 6.208662613171566e-06, + "loss": 0.1397782802581787, + "step": 24110 + }, + { + "epoch": 0.6353575697309543, + "grad_norm": 3.2166261672973633, + "learning_rate": 6.200769557438144e-06, + "loss": 0.13639755249023439, + "step": 24120 + }, + { + "epoch": 0.6356209849754529, + "grad_norm": 1.2867481708526611, + "learning_rate": 6.192879266948554e-06, + "loss": 0.13259371519088745, + "step": 24130 + }, + { + "epoch": 0.6358844002199517, + "grad_norm": 1.7410553693771362, + "learning_rate": 6.184991747445686e-06, + "loss": 0.13306361436843872, + "step": 24140 + }, + { + "epoch": 0.6361478154644504, + "grad_norm": 2.7994141578674316, + "learning_rate": 6.1771070046704165e-06, + "loss": 0.12709860801696776, + "step": 24150 + }, + { + "epoch": 0.6364112307089492, + "grad_norm": 1.792803168296814, + "learning_rate": 6.169225044361604e-06, + "loss": 0.14054362773895263, + "step": 24160 + }, + { + "epoch": 0.6366746459534479, + "grad_norm": 3.1670761108398438, + "learning_rate": 6.161345872256075e-06, + "loss": 0.13797638416290284, + "step": 24170 + }, + { + "epoch": 0.6369380611979467, + "grad_norm": 1.5826307535171509, + "learning_rate": 6.153469494088638e-06, + "loss": 0.1339413642883301, + "step": 24180 + }, + { + "epoch": 0.6372014764424454, + "grad_norm": 2.9239697456359863, + "learning_rate": 6.1455959155920555e-06, + "loss": 0.1279139280319214, + "step": 24190 + }, + { + "epoch": 0.6374648916869442, + "grad_norm": 1.7218010425567627, + "learning_rate": 6.1377251424970555e-06, + "loss": 0.12655272483825683, + "step": 24200 + }, + { + "epoch": 0.6377283069314429, + "grad_norm": 2.213919162750244, + "learning_rate": 6.1298571805323344e-06, + "loss": 0.13593080043792724, + "step": 24210 + }, + { + "epoch": 0.6379917221759416, + "grad_norm": 1.7792781591415405, + "learning_rate": 6.121992035424529e-06, + "loss": 0.12173371315002442, + "step": 24220 + }, + { + "epoch": 0.6382551374204404, + "grad_norm": 1.3213690519332886, + "learning_rate": 6.114129712898227e-06, + "loss": 0.13941333293914795, + "step": 24230 + }, + { + "epoch": 0.6385185526649391, + "grad_norm": 1.7713841199874878, + "learning_rate": 6.106270218675975e-06, + "loss": 0.13567465543746948, + "step": 24240 + }, + { + "epoch": 0.6387819679094379, + "grad_norm": 1.8865396976470947, + "learning_rate": 6.098413558478244e-06, + "loss": 0.13933823108673096, + "step": 24250 + }, + { + "epoch": 0.6390453831539366, + "grad_norm": 1.271030306816101, + "learning_rate": 6.090559738023456e-06, + "loss": 0.1278850555419922, + "step": 24260 + }, + { + "epoch": 0.6393087983984354, + "grad_norm": 1.6737171411514282, + "learning_rate": 6.0827087630279605e-06, + "loss": 0.1389690399169922, + "step": 24270 + }, + { + "epoch": 0.639572213642934, + "grad_norm": 1.5963928699493408, + "learning_rate": 6.074860639206028e-06, + "loss": 0.12713227272033692, + "step": 24280 + }, + { + "epoch": 0.6398356288874328, + "grad_norm": 1.3352234363555908, + "learning_rate": 6.067015372269872e-06, + "loss": 0.12758007049560546, + "step": 24290 + }, + { + "epoch": 0.6400990441319315, + "grad_norm": 1.5285828113555908, + "learning_rate": 6.059172967929609e-06, + "loss": 0.13203625679016112, + "step": 24300 + }, + { + "epoch": 0.6403624593764302, + "grad_norm": 1.3045217990875244, + "learning_rate": 6.0513334318932825e-06, + "loss": 0.12618367671966552, + "step": 24310 + }, + { + "epoch": 0.640625874620929, + "grad_norm": 1.6120918989181519, + "learning_rate": 6.043496769866848e-06, + "loss": 0.13684310913085937, + "step": 24320 + }, + { + "epoch": 0.6408892898654277, + "grad_norm": 1.5523791313171387, + "learning_rate": 6.035662987554154e-06, + "loss": 0.1257244110107422, + "step": 24330 + }, + { + "epoch": 0.6411527051099265, + "grad_norm": 2.2033839225769043, + "learning_rate": 6.027832090656982e-06, + "loss": 0.13762186765670775, + "step": 24340 + }, + { + "epoch": 0.6414161203544252, + "grad_norm": 1.502098798751831, + "learning_rate": 6.020004084874984e-06, + "loss": 0.13737202882766725, + "step": 24350 + }, + { + "epoch": 0.641679535598924, + "grad_norm": 1.4007431268692017, + "learning_rate": 6.012178975905725e-06, + "loss": 0.13659565448760985, + "step": 24360 + }, + { + "epoch": 0.6419429508434227, + "grad_norm": 2.870152473449707, + "learning_rate": 6.0043567694446566e-06, + "loss": 0.11918835639953614, + "step": 24370 + }, + { + "epoch": 0.6422063660879215, + "grad_norm": 2.084540605545044, + "learning_rate": 5.9965374711851175e-06, + "loss": 0.12667852640151978, + "step": 24380 + }, + { + "epoch": 0.6424697813324202, + "grad_norm": 1.2166314125061035, + "learning_rate": 5.9887210868183325e-06, + "loss": 0.13674747943878174, + "step": 24390 + }, + { + "epoch": 0.6427331965769189, + "grad_norm": 1.2317956686019897, + "learning_rate": 5.9809076220334024e-06, + "loss": 0.13177137374877929, + "step": 24400 + }, + { + "epoch": 0.6429966118214177, + "grad_norm": 2.2454824447631836, + "learning_rate": 5.973097082517301e-06, + "loss": 0.1354832649230957, + "step": 24410 + }, + { + "epoch": 0.6432600270659163, + "grad_norm": 4.381562232971191, + "learning_rate": 5.965289473954879e-06, + "loss": 0.12144508361816406, + "step": 24420 + }, + { + "epoch": 0.6435234423104151, + "grad_norm": 2.2292072772979736, + "learning_rate": 5.95748480202885e-06, + "loss": 0.13876508474349974, + "step": 24430 + }, + { + "epoch": 0.6437868575549138, + "grad_norm": 1.8806253671646118, + "learning_rate": 5.949683072419792e-06, + "loss": 0.1272729754447937, + "step": 24440 + }, + { + "epoch": 0.6440502727994126, + "grad_norm": 2.2012064456939697, + "learning_rate": 5.941884290806142e-06, + "loss": 0.12997076511383057, + "step": 24450 + }, + { + "epoch": 0.6443136880439113, + "grad_norm": 2.051976442337036, + "learning_rate": 5.934088462864181e-06, + "loss": 0.13854691982269288, + "step": 24460 + }, + { + "epoch": 0.6445771032884101, + "grad_norm": 2.1803886890411377, + "learning_rate": 5.926295594268059e-06, + "loss": 0.1410223960876465, + "step": 24470 + }, + { + "epoch": 0.6448405185329088, + "grad_norm": 2.523216962814331, + "learning_rate": 5.918505690689759e-06, + "loss": 0.1287736177444458, + "step": 24480 + }, + { + "epoch": 0.6451039337774075, + "grad_norm": 2.6607606410980225, + "learning_rate": 5.910718757799105e-06, + "loss": 0.14075417518615724, + "step": 24490 + }, + { + "epoch": 0.6453673490219063, + "grad_norm": 1.8810338973999023, + "learning_rate": 5.902934801263766e-06, + "loss": 0.14395354986190795, + "step": 24500 + }, + { + "epoch": 0.645630764266405, + "grad_norm": 1.1218173503875732, + "learning_rate": 5.89515382674924e-06, + "loss": 0.12791303396224976, + "step": 24510 + }, + { + "epoch": 0.6458941795109038, + "grad_norm": 1.7858893871307373, + "learning_rate": 5.88737583991886e-06, + "loss": 0.1539613962173462, + "step": 24520 + }, + { + "epoch": 0.6461575947554025, + "grad_norm": 1.7881817817687988, + "learning_rate": 5.879600846433777e-06, + "loss": 0.12808167934417725, + "step": 24530 + }, + { + "epoch": 0.6464210099999013, + "grad_norm": 2.394242525100708, + "learning_rate": 5.871828851952965e-06, + "loss": 0.13699729442596437, + "step": 24540 + }, + { + "epoch": 0.6466844252444, + "grad_norm": 2.589207887649536, + "learning_rate": 5.864059862133221e-06, + "loss": 0.13807885646820067, + "step": 24550 + }, + { + "epoch": 0.6469478404888986, + "grad_norm": 1.3353780508041382, + "learning_rate": 5.856293882629147e-06, + "loss": 0.12268950939178466, + "step": 24560 + }, + { + "epoch": 0.6472112557333974, + "grad_norm": 1.9795188903808594, + "learning_rate": 5.848530919093162e-06, + "loss": 0.13317196369171141, + "step": 24570 + }, + { + "epoch": 0.6474746709778961, + "grad_norm": 1.9011614322662354, + "learning_rate": 5.840770977175485e-06, + "loss": 0.13114914894104004, + "step": 24580 + }, + { + "epoch": 0.6477380862223949, + "grad_norm": 2.3116583824157715, + "learning_rate": 5.833014062524129e-06, + "loss": 0.12124675512313843, + "step": 24590 + }, + { + "epoch": 0.6480015014668936, + "grad_norm": 1.887340784072876, + "learning_rate": 5.825260180784923e-06, + "loss": 0.12589058876037598, + "step": 24600 + }, + { + "epoch": 0.6482649167113924, + "grad_norm": 2.1323037147521973, + "learning_rate": 5.817509337601475e-06, + "loss": 0.13147741556167603, + "step": 24610 + }, + { + "epoch": 0.6485283319558911, + "grad_norm": 2.307000160217285, + "learning_rate": 5.8097615386151686e-06, + "loss": 0.13600587844848633, + "step": 24620 + }, + { + "epoch": 0.6487917472003899, + "grad_norm": 2.48944091796875, + "learning_rate": 5.8020167894652035e-06, + "loss": 0.13014633655548097, + "step": 24630 + }, + { + "epoch": 0.6490551624448886, + "grad_norm": 2.229933261871338, + "learning_rate": 5.79427509578853e-06, + "loss": 0.13549177646636962, + "step": 24640 + }, + { + "epoch": 0.6493185776893873, + "grad_norm": 2.0379881858825684, + "learning_rate": 5.786536463219899e-06, + "loss": 0.13245067596435547, + "step": 24650 + }, + { + "epoch": 0.6495819929338861, + "grad_norm": 1.7165244817733765, + "learning_rate": 5.778800897391812e-06, + "loss": 0.12420871257781982, + "step": 24660 + }, + { + "epoch": 0.6498454081783848, + "grad_norm": 1.310441493988037, + "learning_rate": 5.771068403934545e-06, + "loss": 0.13638441562652587, + "step": 24670 + }, + { + "epoch": 0.6501088234228836, + "grad_norm": 2.029022693634033, + "learning_rate": 5.76333898847615e-06, + "loss": 0.12920522689819336, + "step": 24680 + }, + { + "epoch": 0.6503722386673823, + "grad_norm": 1.697629451751709, + "learning_rate": 5.755612656642424e-06, + "loss": 0.1317318558692932, + "step": 24690 + }, + { + "epoch": 0.650635653911881, + "grad_norm": 2.803190231323242, + "learning_rate": 5.747889414056925e-06, + "loss": 0.12691123485565187, + "step": 24700 + }, + { + "epoch": 0.6508990691563797, + "grad_norm": 2.8606488704681396, + "learning_rate": 5.740169266340963e-06, + "loss": 0.12792572975158692, + "step": 24710 + }, + { + "epoch": 0.6511624844008785, + "grad_norm": 1.3603699207305908, + "learning_rate": 5.732452219113589e-06, + "loss": 0.13621312379837036, + "step": 24720 + }, + { + "epoch": 0.6514258996453772, + "grad_norm": 3.0921502113342285, + "learning_rate": 5.724738277991611e-06, + "loss": 0.13150129318237305, + "step": 24730 + }, + { + "epoch": 0.6516893148898759, + "grad_norm": 1.5941011905670166, + "learning_rate": 5.717027448589568e-06, + "loss": 0.1416204333305359, + "step": 24740 + }, + { + "epoch": 0.6519527301343747, + "grad_norm": 1.2659862041473389, + "learning_rate": 5.709319736519725e-06, + "loss": 0.12160996198654175, + "step": 24750 + }, + { + "epoch": 0.6522161453788734, + "grad_norm": 1.623497724533081, + "learning_rate": 5.701615147392096e-06, + "loss": 0.14979522228240966, + "step": 24760 + }, + { + "epoch": 0.6524795606233722, + "grad_norm": 1.4536168575286865, + "learning_rate": 5.6939136868144075e-06, + "loss": 0.13744478225708007, + "step": 24770 + }, + { + "epoch": 0.6527429758678709, + "grad_norm": 1.405125617980957, + "learning_rate": 5.686215360392127e-06, + "loss": 0.12743618488311767, + "step": 24780 + }, + { + "epoch": 0.6530063911123697, + "grad_norm": 5.709386825561523, + "learning_rate": 5.678520173728414e-06, + "loss": 0.13084194660186768, + "step": 24790 + }, + { + "epoch": 0.6532698063568684, + "grad_norm": 2.28610897064209, + "learning_rate": 5.67082813242416e-06, + "loss": 0.13961296081542968, + "step": 24800 + }, + { + "epoch": 0.6535332216013672, + "grad_norm": 0.9124532341957092, + "learning_rate": 5.663139242077971e-06, + "loss": 0.1403242826461792, + "step": 24810 + }, + { + "epoch": 0.6537966368458659, + "grad_norm": 1.2751017808914185, + "learning_rate": 5.6554535082861496e-06, + "loss": 0.1457698106765747, + "step": 24820 + }, + { + "epoch": 0.6540600520903646, + "grad_norm": 1.639346957206726, + "learning_rate": 5.647770936642705e-06, + "loss": 0.13016829490661622, + "step": 24830 + }, + { + "epoch": 0.6543234673348634, + "grad_norm": 1.6819461584091187, + "learning_rate": 5.6400915327393435e-06, + "loss": 0.13253800868988036, + "step": 24840 + }, + { + "epoch": 0.654586882579362, + "grad_norm": 1.2634174823760986, + "learning_rate": 5.632415302165464e-06, + "loss": 0.1314258337020874, + "step": 24850 + }, + { + "epoch": 0.6548502978238608, + "grad_norm": 1.8643667697906494, + "learning_rate": 5.624742250508164e-06, + "loss": 0.12299454212188721, + "step": 24860 + }, + { + "epoch": 0.6551137130683595, + "grad_norm": 1.6672914028167725, + "learning_rate": 5.617072383352222e-06, + "loss": 0.1308366298675537, + "step": 24870 + }, + { + "epoch": 0.6553771283128583, + "grad_norm": 1.6924079656600952, + "learning_rate": 5.6094057062800886e-06, + "loss": 0.12606921195983886, + "step": 24880 + }, + { + "epoch": 0.655640543557357, + "grad_norm": 1.5589250326156616, + "learning_rate": 5.601742224871911e-06, + "loss": 0.1371609687805176, + "step": 24890 + }, + { + "epoch": 0.6559039588018558, + "grad_norm": 2.4261434078216553, + "learning_rate": 5.594081944705496e-06, + "loss": 0.1384268045425415, + "step": 24900 + }, + { + "epoch": 0.6561673740463545, + "grad_norm": 2.2880752086639404, + "learning_rate": 5.586424871356337e-06, + "loss": 0.12381449937820435, + "step": 24910 + }, + { + "epoch": 0.6564307892908532, + "grad_norm": 1.4532544612884521, + "learning_rate": 5.578771010397572e-06, + "loss": 0.13362683057785035, + "step": 24920 + }, + { + "epoch": 0.656694204535352, + "grad_norm": 1.4138152599334717, + "learning_rate": 5.571120367400008e-06, + "loss": 0.1355597972869873, + "step": 24930 + }, + { + "epoch": 0.6569576197798507, + "grad_norm": 3.526632308959961, + "learning_rate": 5.563472947932123e-06, + "loss": 0.12687042951583863, + "step": 24940 + }, + { + "epoch": 0.6572210350243495, + "grad_norm": 1.5611504316329956, + "learning_rate": 5.555828757560035e-06, + "loss": 0.1322413444519043, + "step": 24950 + }, + { + "epoch": 0.6574844502688482, + "grad_norm": 1.2827401161193848, + "learning_rate": 5.548187801847513e-06, + "loss": 0.13164730072021485, + "step": 24960 + }, + { + "epoch": 0.657747865513347, + "grad_norm": 1.3520678281784058, + "learning_rate": 5.5405500863559795e-06, + "loss": 0.13929452896118164, + "step": 24970 + }, + { + "epoch": 0.6580112807578457, + "grad_norm": 1.980705738067627, + "learning_rate": 5.532915616644484e-06, + "loss": 0.128364098072052, + "step": 24980 + }, + { + "epoch": 0.6582746960023445, + "grad_norm": 1.3464552164077759, + "learning_rate": 5.525284398269733e-06, + "loss": 0.11934142112731934, + "step": 24990 + }, + { + "epoch": 0.6585381112468431, + "grad_norm": 1.7729099988937378, + "learning_rate": 5.517656436786059e-06, + "loss": 0.13095258474349974, + "step": 25000 + }, + { + "epoch": 0.6588015264913418, + "grad_norm": 2.73870587348938, + "learning_rate": 5.510031737745405e-06, + "loss": 0.13141311407089235, + "step": 25010 + }, + { + "epoch": 0.6590649417358406, + "grad_norm": 1.6452128887176514, + "learning_rate": 5.50241030669737e-06, + "loss": 0.13083635568618773, + "step": 25020 + }, + { + "epoch": 0.6593283569803393, + "grad_norm": 1.5080146789550781, + "learning_rate": 5.4947921491891565e-06, + "loss": 0.13297886848449708, + "step": 25030 + }, + { + "epoch": 0.6595917722248381, + "grad_norm": 1.4485372304916382, + "learning_rate": 5.487177270765588e-06, + "loss": 0.13110953569412231, + "step": 25040 + }, + { + "epoch": 0.6598551874693368, + "grad_norm": 1.4952484369277954, + "learning_rate": 5.479565676969101e-06, + "loss": 0.12726955413818358, + "step": 25050 + }, + { + "epoch": 0.6601186027138356, + "grad_norm": 1.1663979291915894, + "learning_rate": 5.47195737333974e-06, + "loss": 0.11983405351638794, + "step": 25060 + }, + { + "epoch": 0.6603820179583343, + "grad_norm": 1.2042052745819092, + "learning_rate": 5.464352365415162e-06, + "loss": 0.13239039182662965, + "step": 25070 + }, + { + "epoch": 0.660645433202833, + "grad_norm": 1.8329789638519287, + "learning_rate": 5.456750658730616e-06, + "loss": 0.14161735773086548, + "step": 25080 + }, + { + "epoch": 0.6609088484473318, + "grad_norm": 1.4280349016189575, + "learning_rate": 5.449152258818955e-06, + "loss": 0.13099944591522217, + "step": 25090 + }, + { + "epoch": 0.6611722636918305, + "grad_norm": 3.1302452087402344, + "learning_rate": 5.4415571712106184e-06, + "loss": 0.13577942848205565, + "step": 25100 + }, + { + "epoch": 0.6614356789363293, + "grad_norm": 2.258941411972046, + "learning_rate": 5.433965401433639e-06, + "loss": 0.12222174406051636, + "step": 25110 + }, + { + "epoch": 0.661699094180828, + "grad_norm": 1.6579668521881104, + "learning_rate": 5.426376955013641e-06, + "loss": 0.12920050621032714, + "step": 25120 + }, + { + "epoch": 0.6619625094253268, + "grad_norm": 1.8793742656707764, + "learning_rate": 5.4187918374738154e-06, + "loss": 0.1374260663986206, + "step": 25130 + }, + { + "epoch": 0.6622259246698254, + "grad_norm": 1.376941204071045, + "learning_rate": 5.4112100543349365e-06, + "loss": 0.12202659845352173, + "step": 25140 + }, + { + "epoch": 0.6624893399143242, + "grad_norm": 2.4289844036102295, + "learning_rate": 5.403631611115362e-06, + "loss": 0.13023073673248292, + "step": 25150 + }, + { + "epoch": 0.6627527551588229, + "grad_norm": 2.1073296070098877, + "learning_rate": 5.396056513331005e-06, + "loss": 0.13214601278305055, + "step": 25160 + }, + { + "epoch": 0.6630161704033216, + "grad_norm": 1.9068315029144287, + "learning_rate": 5.388484766495348e-06, + "loss": 0.1410287141799927, + "step": 25170 + }, + { + "epoch": 0.6632795856478204, + "grad_norm": 2.458360195159912, + "learning_rate": 5.3809163761194364e-06, + "loss": 0.13660567998886108, + "step": 25180 + }, + { + "epoch": 0.6635430008923191, + "grad_norm": 1.0994071960449219, + "learning_rate": 5.373351347711865e-06, + "loss": 0.11652270555496216, + "step": 25190 + }, + { + "epoch": 0.6638064161368179, + "grad_norm": 2.1196765899658203, + "learning_rate": 5.365789686778797e-06, + "loss": 0.1388843536376953, + "step": 25200 + }, + { + "epoch": 0.6640698313813166, + "grad_norm": 2.2791812419891357, + "learning_rate": 5.358231398823937e-06, + "loss": 0.12458174228668213, + "step": 25210 + }, + { + "epoch": 0.6643332466258154, + "grad_norm": 2.297106981277466, + "learning_rate": 5.350676489348521e-06, + "loss": 0.13075881004333495, + "step": 25220 + }, + { + "epoch": 0.6645966618703141, + "grad_norm": 1.899437665939331, + "learning_rate": 5.3431249638513475e-06, + "loss": 0.14028871059417725, + "step": 25230 + }, + { + "epoch": 0.6648600771148129, + "grad_norm": 1.6026115417480469, + "learning_rate": 5.335576827828737e-06, + "loss": 0.13006434440612794, + "step": 25240 + }, + { + "epoch": 0.6651234923593116, + "grad_norm": 2.400099992752075, + "learning_rate": 5.328032086774557e-06, + "loss": 0.12756762504577637, + "step": 25250 + }, + { + "epoch": 0.6653869076038103, + "grad_norm": 3.3062548637390137, + "learning_rate": 5.320490746180189e-06, + "loss": 0.12573859691619874, + "step": 25260 + }, + { + "epoch": 0.665650322848309, + "grad_norm": 2.58324933052063, + "learning_rate": 5.312952811534542e-06, + "loss": 0.12915254831314088, + "step": 25270 + }, + { + "epoch": 0.6659137380928077, + "grad_norm": 2.0124402046203613, + "learning_rate": 5.305418288324058e-06, + "loss": 0.1253851532936096, + "step": 25280 + }, + { + "epoch": 0.6661771533373065, + "grad_norm": 1.8414723873138428, + "learning_rate": 5.297887182032686e-06, + "loss": 0.13310680389404297, + "step": 25290 + }, + { + "epoch": 0.6664405685818052, + "grad_norm": 2.41351056098938, + "learning_rate": 5.29035949814189e-06, + "loss": 0.12173905372619628, + "step": 25300 + }, + { + "epoch": 0.666703983826304, + "grad_norm": 2.6791083812713623, + "learning_rate": 5.2828352421306415e-06, + "loss": 0.12899575233459473, + "step": 25310 + }, + { + "epoch": 0.6669673990708027, + "grad_norm": 1.7966035604476929, + "learning_rate": 5.275314419475417e-06, + "loss": 0.1192636251449585, + "step": 25320 + }, + { + "epoch": 0.6672308143153015, + "grad_norm": 2.291858673095703, + "learning_rate": 5.267797035650202e-06, + "loss": 0.12332535982131958, + "step": 25330 + }, + { + "epoch": 0.6674942295598002, + "grad_norm": 1.1458030939102173, + "learning_rate": 5.2602830961264704e-06, + "loss": 0.13520365953445435, + "step": 25340 + }, + { + "epoch": 0.6677576448042989, + "grad_norm": 2.3721702098846436, + "learning_rate": 5.252772606373186e-06, + "loss": 0.1300657033920288, + "step": 25350 + }, + { + "epoch": 0.6680210600487977, + "grad_norm": 2.130174160003662, + "learning_rate": 5.245265571856812e-06, + "loss": 0.1195662260055542, + "step": 25360 + }, + { + "epoch": 0.6682844752932964, + "grad_norm": 1.9424993991851807, + "learning_rate": 5.237761998041291e-06, + "loss": 0.1294872760772705, + "step": 25370 + }, + { + "epoch": 0.6685478905377952, + "grad_norm": 1.4057939052581787, + "learning_rate": 5.230261890388053e-06, + "loss": 0.14345974922180177, + "step": 25380 + }, + { + "epoch": 0.6688113057822939, + "grad_norm": 1.3446969985961914, + "learning_rate": 5.2227652543559924e-06, + "loss": 0.12499780654907226, + "step": 25390 + }, + { + "epoch": 0.6690747210267927, + "grad_norm": 2.3217098712921143, + "learning_rate": 5.215272095401483e-06, + "loss": 0.13025436401367188, + "step": 25400 + }, + { + "epoch": 0.6693381362712914, + "grad_norm": 1.7457010746002197, + "learning_rate": 5.207782418978378e-06, + "loss": 0.12822134494781495, + "step": 25410 + }, + { + "epoch": 0.6696015515157901, + "grad_norm": 1.5481938123703003, + "learning_rate": 5.200296230537982e-06, + "loss": 0.12609055042266845, + "step": 25420 + }, + { + "epoch": 0.6698649667602888, + "grad_norm": 1.7505950927734375, + "learning_rate": 5.192813535529066e-06, + "loss": 0.13643052577972412, + "step": 25430 + }, + { + "epoch": 0.6701283820047875, + "grad_norm": 1.3429715633392334, + "learning_rate": 5.1853343393978585e-06, + "loss": 0.12670886516571045, + "step": 25440 + }, + { + "epoch": 0.6703917972492863, + "grad_norm": 1.783089280128479, + "learning_rate": 5.177858647588037e-06, + "loss": 0.1306181311607361, + "step": 25450 + }, + { + "epoch": 0.670655212493785, + "grad_norm": 1.8717070817947388, + "learning_rate": 5.170386465540741e-06, + "loss": 0.12298781871795654, + "step": 25460 + }, + { + "epoch": 0.6709186277382838, + "grad_norm": 1.6588422060012817, + "learning_rate": 5.162917798694546e-06, + "loss": 0.12497611045837402, + "step": 25470 + }, + { + "epoch": 0.6711820429827825, + "grad_norm": 2.155613660812378, + "learning_rate": 5.15545265248546e-06, + "loss": 0.13800846338272094, + "step": 25480 + }, + { + "epoch": 0.6714454582272813, + "grad_norm": 1.5045251846313477, + "learning_rate": 5.147991032346948e-06, + "loss": 0.11993607282638549, + "step": 25490 + }, + { + "epoch": 0.67170887347178, + "grad_norm": 1.269142508506775, + "learning_rate": 5.140532943709893e-06, + "loss": 0.12991983890533448, + "step": 25500 + }, + { + "epoch": 0.6719722887162788, + "grad_norm": 1.488168478012085, + "learning_rate": 5.133078392002623e-06, + "loss": 0.13342472314834594, + "step": 25510 + }, + { + "epoch": 0.6722357039607775, + "grad_norm": 2.0284454822540283, + "learning_rate": 5.125627382650873e-06, + "loss": 0.12573618888854982, + "step": 25520 + }, + { + "epoch": 0.6724991192052762, + "grad_norm": 3.8415136337280273, + "learning_rate": 5.118179921077808e-06, + "loss": 0.13221687078475952, + "step": 25530 + }, + { + "epoch": 0.672762534449775, + "grad_norm": 2.369098663330078, + "learning_rate": 5.1107360127040206e-06, + "loss": 0.13527467250823974, + "step": 25540 + }, + { + "epoch": 0.6730259496942737, + "grad_norm": 1.797027826309204, + "learning_rate": 5.103295662947504e-06, + "loss": 0.1261106014251709, + "step": 25550 + }, + { + "epoch": 0.6732893649387724, + "grad_norm": 3.8057236671447754, + "learning_rate": 5.095858877223665e-06, + "loss": 0.12498772144317627, + "step": 25560 + }, + { + "epoch": 0.6735527801832711, + "grad_norm": 1.5639567375183105, + "learning_rate": 5.088425660945319e-06, + "loss": 0.12648916244506836, + "step": 25570 + }, + { + "epoch": 0.6738161954277699, + "grad_norm": 1.4558689594268799, + "learning_rate": 5.080996019522675e-06, + "loss": 0.12021915912628174, + "step": 25580 + }, + { + "epoch": 0.6740796106722686, + "grad_norm": 3.490931749343872, + "learning_rate": 5.073569958363356e-06, + "loss": 0.128900146484375, + "step": 25590 + }, + { + "epoch": 0.6743430259167673, + "grad_norm": 1.340933918952942, + "learning_rate": 5.06614748287237e-06, + "loss": 0.11912744045257569, + "step": 25600 + }, + { + "epoch": 0.6746064411612661, + "grad_norm": 2.1723031997680664, + "learning_rate": 5.058728598452103e-06, + "loss": 0.1413840889930725, + "step": 25610 + }, + { + "epoch": 0.6748698564057648, + "grad_norm": 1.8538182973861694, + "learning_rate": 5.05131331050235e-06, + "loss": 0.11695785522460937, + "step": 25620 + }, + { + "epoch": 0.6751332716502636, + "grad_norm": 1.5360100269317627, + "learning_rate": 5.043901624420273e-06, + "loss": 0.1379454255104065, + "step": 25630 + }, + { + "epoch": 0.6753966868947623, + "grad_norm": 2.5603387355804443, + "learning_rate": 5.036493545600421e-06, + "loss": 0.12424347400665284, + "step": 25640 + }, + { + "epoch": 0.6756601021392611, + "grad_norm": 1.0794901847839355, + "learning_rate": 5.029089079434708e-06, + "loss": 0.1258098840713501, + "step": 25650 + }, + { + "epoch": 0.6759235173837598, + "grad_norm": 2.2018795013427734, + "learning_rate": 5.021688231312423e-06, + "loss": 0.11699883937835694, + "step": 25660 + }, + { + "epoch": 0.6761869326282586, + "grad_norm": 1.2291730642318726, + "learning_rate": 5.014291006620229e-06, + "loss": 0.13019800186157227, + "step": 25670 + }, + { + "epoch": 0.6764503478727573, + "grad_norm": 3.392538547515869, + "learning_rate": 5.006897410742142e-06, + "loss": 0.12773337364196777, + "step": 25680 + }, + { + "epoch": 0.676713763117256, + "grad_norm": 1.2877130508422852, + "learning_rate": 4.999507449059542e-06, + "loss": 0.12771573066711425, + "step": 25690 + }, + { + "epoch": 0.6769771783617548, + "grad_norm": 1.493895411491394, + "learning_rate": 4.992121126951159e-06, + "loss": 0.1334062933921814, + "step": 25700 + }, + { + "epoch": 0.6772405936062534, + "grad_norm": 1.9804935455322266, + "learning_rate": 4.984738449793074e-06, + "loss": 0.12453010082244872, + "step": 25710 + }, + { + "epoch": 0.6775040088507522, + "grad_norm": 2.129753351211548, + "learning_rate": 4.977359422958731e-06, + "loss": 0.12513575553894044, + "step": 25720 + }, + { + "epoch": 0.6777674240952509, + "grad_norm": 1.4186307191848755, + "learning_rate": 4.96998405181889e-06, + "loss": 0.13320140838623046, + "step": 25730 + }, + { + "epoch": 0.6780308393397497, + "grad_norm": 2.0219082832336426, + "learning_rate": 4.962612341741666e-06, + "loss": 0.1161996603012085, + "step": 25740 + }, + { + "epoch": 0.6782942545842484, + "grad_norm": 1.775639533996582, + "learning_rate": 4.955244298092514e-06, + "loss": 0.12613322734832763, + "step": 25750 + }, + { + "epoch": 0.6785576698287472, + "grad_norm": 1.494731068611145, + "learning_rate": 4.947879926234212e-06, + "loss": 0.13795970678329467, + "step": 25760 + }, + { + "epoch": 0.6788210850732459, + "grad_norm": 2.2704029083251953, + "learning_rate": 4.940519231526867e-06, + "loss": 0.12754318714141846, + "step": 25770 + }, + { + "epoch": 0.6790845003177446, + "grad_norm": 1.2924214601516724, + "learning_rate": 4.933162219327909e-06, + "loss": 0.11388198137283326, + "step": 25780 + }, + { + "epoch": 0.6793479155622434, + "grad_norm": 1.7936962842941284, + "learning_rate": 4.925808894992087e-06, + "loss": 0.12623152732849122, + "step": 25790 + }, + { + "epoch": 0.6796113308067421, + "grad_norm": 1.3384201526641846, + "learning_rate": 4.918459263871474e-06, + "loss": 0.12634038925170898, + "step": 25800 + }, + { + "epoch": 0.6798747460512409, + "grad_norm": 1.8524893522262573, + "learning_rate": 4.911113331315448e-06, + "loss": 0.11375504732131958, + "step": 25810 + }, + { + "epoch": 0.6801381612957396, + "grad_norm": 3.909694194793701, + "learning_rate": 4.903771102670686e-06, + "loss": 0.12788107395172119, + "step": 25820 + }, + { + "epoch": 0.6804015765402384, + "grad_norm": 1.5324046611785889, + "learning_rate": 4.896432583281189e-06, + "loss": 0.12752649784088135, + "step": 25830 + }, + { + "epoch": 0.680664991784737, + "grad_norm": 1.9918378591537476, + "learning_rate": 4.88909777848824e-06, + "loss": 0.12874687910079957, + "step": 25840 + }, + { + "epoch": 0.6809284070292358, + "grad_norm": 1.8962688446044922, + "learning_rate": 4.881766693630439e-06, + "loss": 0.13202145099639892, + "step": 25850 + }, + { + "epoch": 0.6811918222737345, + "grad_norm": 1.7243987321853638, + "learning_rate": 4.874439334043654e-06, + "loss": 0.12332782745361329, + "step": 25860 + }, + { + "epoch": 0.6814552375182332, + "grad_norm": 1.200007677078247, + "learning_rate": 4.8671157050610525e-06, + "loss": 0.13454337120056153, + "step": 25870 + }, + { + "epoch": 0.681718652762732, + "grad_norm": 2.241781711578369, + "learning_rate": 4.859795812013099e-06, + "loss": 0.14014177322387694, + "step": 25880 + }, + { + "epoch": 0.6819820680072307, + "grad_norm": 1.8308086395263672, + "learning_rate": 4.852479660227518e-06, + "loss": 0.1250972032546997, + "step": 25890 + }, + { + "epoch": 0.6822454832517295, + "grad_norm": 1.8949426412582397, + "learning_rate": 4.8451672550293236e-06, + "loss": 0.11453173160552979, + "step": 25900 + }, + { + "epoch": 0.6825088984962282, + "grad_norm": 1.7976726293563843, + "learning_rate": 4.837858601740799e-06, + "loss": 0.12262119054794311, + "step": 25910 + }, + { + "epoch": 0.682772313740727, + "grad_norm": 2.4796974658966064, + "learning_rate": 4.830553705681496e-06, + "loss": 0.12234218120574951, + "step": 25920 + }, + { + "epoch": 0.6830357289852257, + "grad_norm": 1.9589786529541016, + "learning_rate": 4.823252572168236e-06, + "loss": 0.12920804023742677, + "step": 25930 + }, + { + "epoch": 0.6832991442297245, + "grad_norm": 1.234521746635437, + "learning_rate": 4.815955206515104e-06, + "loss": 0.12472419738769532, + "step": 25940 + }, + { + "epoch": 0.6835625594742232, + "grad_norm": 3.373856782913208, + "learning_rate": 4.808661614033422e-06, + "loss": 0.12593798637390136, + "step": 25950 + }, + { + "epoch": 0.6838259747187219, + "grad_norm": 1.2695002555847168, + "learning_rate": 4.801371800031796e-06, + "loss": 0.12524933815002443, + "step": 25960 + }, + { + "epoch": 0.6840893899632207, + "grad_norm": 1.743187665939331, + "learning_rate": 4.794085769816057e-06, + "loss": 0.10233479738235474, + "step": 25970 + }, + { + "epoch": 0.6843528052077194, + "grad_norm": 1.9050437211990356, + "learning_rate": 4.786803528689304e-06, + "loss": 0.13357897996902465, + "step": 25980 + }, + { + "epoch": 0.6846162204522181, + "grad_norm": 4.022175312042236, + "learning_rate": 4.779525081951857e-06, + "loss": 0.13815125226974487, + "step": 25990 + }, + { + "epoch": 0.6848796356967168, + "grad_norm": 2.248281955718994, + "learning_rate": 4.772250434901281e-06, + "loss": 0.12030645608901977, + "step": 26000 + }, + { + "epoch": 0.6851430509412156, + "grad_norm": 2.2890219688415527, + "learning_rate": 4.7649795928323875e-06, + "loss": 0.11388468742370605, + "step": 26010 + }, + { + "epoch": 0.6854064661857143, + "grad_norm": 1.4510376453399658, + "learning_rate": 4.757712561037203e-06, + "loss": 0.12377902269363403, + "step": 26020 + }, + { + "epoch": 0.6856698814302131, + "grad_norm": 3.04190993309021, + "learning_rate": 4.750449344804989e-06, + "loss": 0.11781233549118042, + "step": 26030 + }, + { + "epoch": 0.6859332966747118, + "grad_norm": 5.123847961425781, + "learning_rate": 4.743189949422227e-06, + "loss": 0.14818997383117677, + "step": 26040 + }, + { + "epoch": 0.6861967119192105, + "grad_norm": 2.042891025543213, + "learning_rate": 4.735934380172613e-06, + "loss": 0.12007951736450195, + "step": 26050 + }, + { + "epoch": 0.6864601271637093, + "grad_norm": 2.0521225929260254, + "learning_rate": 4.728682642337072e-06, + "loss": 0.12081928253173828, + "step": 26060 + }, + { + "epoch": 0.686723542408208, + "grad_norm": 1.1114120483398438, + "learning_rate": 4.7214347411937325e-06, + "loss": 0.12369352579116821, + "step": 26070 + }, + { + "epoch": 0.6869869576527068, + "grad_norm": 1.4293690919876099, + "learning_rate": 4.714190682017918e-06, + "loss": 0.13445353507995605, + "step": 26080 + }, + { + "epoch": 0.6872503728972055, + "grad_norm": 1.797214388847351, + "learning_rate": 4.706950470082177e-06, + "loss": 0.12318563461303711, + "step": 26090 + }, + { + "epoch": 0.6875137881417043, + "grad_norm": 1.720319151878357, + "learning_rate": 4.699714110656242e-06, + "loss": 0.1171583890914917, + "step": 26100 + }, + { + "epoch": 0.687777203386203, + "grad_norm": 1.6333545446395874, + "learning_rate": 4.6924816090070575e-06, + "loss": 0.12445545196533203, + "step": 26110 + }, + { + "epoch": 0.6880406186307017, + "grad_norm": 1.8079514503479004, + "learning_rate": 4.685252970398742e-06, + "loss": 0.1382610321044922, + "step": 26120 + }, + { + "epoch": 0.6883040338752004, + "grad_norm": 1.4077295064926147, + "learning_rate": 4.678028200092607e-06, + "loss": 0.12871141433715821, + "step": 26130 + }, + { + "epoch": 0.6885674491196991, + "grad_norm": 1.8897806406021118, + "learning_rate": 4.670807303347161e-06, + "loss": 0.1278050422668457, + "step": 26140 + }, + { + "epoch": 0.6888308643641979, + "grad_norm": 2.3707480430603027, + "learning_rate": 4.663590285418082e-06, + "loss": 0.12754204273223876, + "step": 26150 + }, + { + "epoch": 0.6890942796086966, + "grad_norm": 2.836160898208618, + "learning_rate": 4.656377151558224e-06, + "loss": 0.1136633276939392, + "step": 26160 + }, + { + "epoch": 0.6893576948531954, + "grad_norm": 1.6652731895446777, + "learning_rate": 4.64916790701762e-06, + "loss": 0.1398106336593628, + "step": 26170 + }, + { + "epoch": 0.6896211100976941, + "grad_norm": 2.559479236602783, + "learning_rate": 4.641962557043462e-06, + "loss": 0.11484137773513795, + "step": 26180 + }, + { + "epoch": 0.6898845253421929, + "grad_norm": 2.519879102706909, + "learning_rate": 4.634761106880127e-06, + "loss": 0.1213154673576355, + "step": 26190 + }, + { + "epoch": 0.6901479405866916, + "grad_norm": 2.3399505615234375, + "learning_rate": 4.627563561769136e-06, + "loss": 0.12091805934906005, + "step": 26200 + }, + { + "epoch": 0.6904113558311903, + "grad_norm": 1.7298312187194824, + "learning_rate": 4.620369926949164e-06, + "loss": 0.12394771575927735, + "step": 26210 + }, + { + "epoch": 0.6906747710756891, + "grad_norm": 1.3920811414718628, + "learning_rate": 4.613180207656061e-06, + "loss": 0.11069111824035645, + "step": 26220 + }, + { + "epoch": 0.6909381863201878, + "grad_norm": 1.5509004592895508, + "learning_rate": 4.60599440912281e-06, + "loss": 0.12767703533172609, + "step": 26230 + }, + { + "epoch": 0.6912016015646866, + "grad_norm": 2.573535680770874, + "learning_rate": 4.598812536579545e-06, + "loss": 0.11602911949157715, + "step": 26240 + }, + { + "epoch": 0.6914650168091853, + "grad_norm": 1.6734099388122559, + "learning_rate": 4.591634595253541e-06, + "loss": 0.1196373462677002, + "step": 26250 + }, + { + "epoch": 0.6917284320536841, + "grad_norm": 2.1661152839660645, + "learning_rate": 4.584460590369214e-06, + "loss": 0.12832157611846923, + "step": 26260 + }, + { + "epoch": 0.6919918472981827, + "grad_norm": 2.3321237564086914, + "learning_rate": 4.577290527148118e-06, + "loss": 0.12734506130218506, + "step": 26270 + }, + { + "epoch": 0.6922552625426815, + "grad_norm": 1.6326546669006348, + "learning_rate": 4.5701244108089325e-06, + "loss": 0.12892568111419678, + "step": 26280 + }, + { + "epoch": 0.6925186777871802, + "grad_norm": 1.6847262382507324, + "learning_rate": 4.562962246567466e-06, + "loss": 0.13253918886184693, + "step": 26290 + }, + { + "epoch": 0.6927820930316789, + "grad_norm": 3.297942876815796, + "learning_rate": 4.555804039636649e-06, + "loss": 0.1314547538757324, + "step": 26300 + }, + { + "epoch": 0.6930455082761777, + "grad_norm": 5.319476127624512, + "learning_rate": 4.548649795226532e-06, + "loss": 0.12391246557235717, + "step": 26310 + }, + { + "epoch": 0.6933089235206764, + "grad_norm": 1.5717527866363525, + "learning_rate": 4.541499518544292e-06, + "loss": 0.12464072704315185, + "step": 26320 + }, + { + "epoch": 0.6935723387651752, + "grad_norm": 2.2667927742004395, + "learning_rate": 4.5343532147941985e-06, + "loss": 0.1235917329788208, + "step": 26330 + }, + { + "epoch": 0.6938357540096739, + "grad_norm": 1.7554293870925903, + "learning_rate": 4.527210889177639e-06, + "loss": 0.13122339248657228, + "step": 26340 + }, + { + "epoch": 0.6940991692541727, + "grad_norm": 4.1936187744140625, + "learning_rate": 4.520072546893115e-06, + "loss": 0.1331318974494934, + "step": 26350 + }, + { + "epoch": 0.6943625844986714, + "grad_norm": 1.5481935739517212, + "learning_rate": 4.512938193136216e-06, + "loss": 0.11920256614685058, + "step": 26360 + }, + { + "epoch": 0.6946259997431702, + "grad_norm": 1.3999332189559937, + "learning_rate": 4.5058078330996315e-06, + "loss": 0.12774291038513183, + "step": 26370 + }, + { + "epoch": 0.6948894149876689, + "grad_norm": 2.324660062789917, + "learning_rate": 4.498681471973145e-06, + "loss": 0.12270166873931884, + "step": 26380 + }, + { + "epoch": 0.6951528302321676, + "grad_norm": 2.019185781478882, + "learning_rate": 4.491559114943628e-06, + "loss": 0.11729981899261474, + "step": 26390 + }, + { + "epoch": 0.6954162454766664, + "grad_norm": 2.0001399517059326, + "learning_rate": 4.484440767195044e-06, + "loss": 0.1314157485961914, + "step": 26400 + }, + { + "epoch": 0.695679660721165, + "grad_norm": 1.8167296648025513, + "learning_rate": 4.477326433908435e-06, + "loss": 0.12074465751647949, + "step": 26410 + }, + { + "epoch": 0.6959430759656638, + "grad_norm": 0.9957918524742126, + "learning_rate": 4.4702161202619085e-06, + "loss": 0.12109637260437012, + "step": 26420 + }, + { + "epoch": 0.6962064912101625, + "grad_norm": 1.5943559408187866, + "learning_rate": 4.463109831430668e-06, + "loss": 0.12313277721405029, + "step": 26430 + }, + { + "epoch": 0.6964699064546613, + "grad_norm": 2.588819742202759, + "learning_rate": 4.456007572586971e-06, + "loss": 0.12539902925491334, + "step": 26440 + }, + { + "epoch": 0.69673332169916, + "grad_norm": 2.432908058166504, + "learning_rate": 4.448909348900159e-06, + "loss": 0.12543704509735107, + "step": 26450 + }, + { + "epoch": 0.6969967369436588, + "grad_norm": 2.2302777767181396, + "learning_rate": 4.441815165536612e-06, + "loss": 0.1382764458656311, + "step": 26460 + }, + { + "epoch": 0.6972601521881575, + "grad_norm": 1.9267276525497437, + "learning_rate": 4.434725027659786e-06, + "loss": 0.12438304424285888, + "step": 26470 + }, + { + "epoch": 0.6975235674326562, + "grad_norm": 2.160097599029541, + "learning_rate": 4.427638940430192e-06, + "loss": 0.11275174617767333, + "step": 26480 + }, + { + "epoch": 0.697786982677155, + "grad_norm": 1.8869092464447021, + "learning_rate": 4.420556909005392e-06, + "loss": 0.12502150535583495, + "step": 26490 + }, + { + "epoch": 0.6980503979216537, + "grad_norm": 3.462843656539917, + "learning_rate": 4.41347893853999e-06, + "loss": 0.13482294082641602, + "step": 26500 + }, + { + "epoch": 0.6983138131661525, + "grad_norm": 3.4367330074310303, + "learning_rate": 4.40640503418564e-06, + "loss": 0.11310168504714965, + "step": 26510 + }, + { + "epoch": 0.6985772284106512, + "grad_norm": 1.4830509424209595, + "learning_rate": 4.39933520109103e-06, + "loss": 0.11204725503921509, + "step": 26520 + }, + { + "epoch": 0.69884064365515, + "grad_norm": 2.758413553237915, + "learning_rate": 4.3922694444018985e-06, + "loss": 0.11532481908798217, + "step": 26530 + }, + { + "epoch": 0.6991040588996487, + "grad_norm": 2.2051820755004883, + "learning_rate": 4.385207769261007e-06, + "loss": 0.11392016410827636, + "step": 26540 + }, + { + "epoch": 0.6993674741441475, + "grad_norm": 3.0922088623046875, + "learning_rate": 4.378150180808136e-06, + "loss": 0.12577866315841674, + "step": 26550 + }, + { + "epoch": 0.6996308893886461, + "grad_norm": 1.5778255462646484, + "learning_rate": 4.371096684180114e-06, + "loss": 0.12512559890747071, + "step": 26560 + }, + { + "epoch": 0.6998943046331448, + "grad_norm": 1.7936575412750244, + "learning_rate": 4.364047284510774e-06, + "loss": 0.10628471374511719, + "step": 26570 + }, + { + "epoch": 0.7001577198776436, + "grad_norm": 1.6378873586654663, + "learning_rate": 4.357001986930982e-06, + "loss": 0.1190016508102417, + "step": 26580 + }, + { + "epoch": 0.7004211351221423, + "grad_norm": 3.0848822593688965, + "learning_rate": 4.349960796568601e-06, + "loss": 0.12344317436218262, + "step": 26590 + }, + { + "epoch": 0.7006845503666411, + "grad_norm": 2.5492029190063477, + "learning_rate": 4.342923718548512e-06, + "loss": 0.12260513305664063, + "step": 26600 + }, + { + "epoch": 0.7009479656111398, + "grad_norm": 1.7489057779312134, + "learning_rate": 4.335890757992611e-06, + "loss": 0.1149555802345276, + "step": 26610 + }, + { + "epoch": 0.7012113808556386, + "grad_norm": 2.2251529693603516, + "learning_rate": 4.328861920019787e-06, + "loss": 0.11700115203857422, + "step": 26620 + }, + { + "epoch": 0.7014747961001373, + "grad_norm": 4.124988555908203, + "learning_rate": 4.321837209745931e-06, + "loss": 0.12826552391052246, + "step": 26630 + }, + { + "epoch": 0.701738211344636, + "grad_norm": 1.8050569295883179, + "learning_rate": 4.3148166322839305e-06, + "loss": 0.11841685771942138, + "step": 26640 + }, + { + "epoch": 0.7020016265891348, + "grad_norm": 3.286198854446411, + "learning_rate": 4.3078001927436605e-06, + "loss": 0.11644930839538574, + "step": 26650 + }, + { + "epoch": 0.7022650418336335, + "grad_norm": 1.5489692687988281, + "learning_rate": 4.300787896231997e-06, + "loss": 0.11453173160552979, + "step": 26660 + }, + { + "epoch": 0.7025284570781323, + "grad_norm": 3.0210037231445312, + "learning_rate": 4.293779747852791e-06, + "loss": 0.12340993881225586, + "step": 26670 + }, + { + "epoch": 0.702791872322631, + "grad_norm": 3.568045139312744, + "learning_rate": 4.286775752706865e-06, + "loss": 0.12373065948486328, + "step": 26680 + }, + { + "epoch": 0.7030552875671298, + "grad_norm": 2.357720136642456, + "learning_rate": 4.279775915892038e-06, + "loss": 0.12434806823730468, + "step": 26690 + }, + { + "epoch": 0.7033187028116284, + "grad_norm": 1.5401982069015503, + "learning_rate": 4.2727802425030885e-06, + "loss": 0.1109654188156128, + "step": 26700 + }, + { + "epoch": 0.7035821180561272, + "grad_norm": 1.4726052284240723, + "learning_rate": 4.265788737631779e-06, + "loss": 0.11062666177749633, + "step": 26710 + }, + { + "epoch": 0.7038455333006259, + "grad_norm": 1.555052399635315, + "learning_rate": 4.258801406366817e-06, + "loss": 0.12758088111877441, + "step": 26720 + }, + { + "epoch": 0.7041089485451246, + "grad_norm": 3.5088677406311035, + "learning_rate": 4.251818253793885e-06, + "loss": 0.12719898223876952, + "step": 26730 + }, + { + "epoch": 0.7043723637896234, + "grad_norm": 1.952720284461975, + "learning_rate": 4.244839284995627e-06, + "loss": 0.1294842004776001, + "step": 26740 + }, + { + "epoch": 0.7046357790341221, + "grad_norm": 2.541106939315796, + "learning_rate": 4.237864505051638e-06, + "loss": 0.12144107818603515, + "step": 26750 + }, + { + "epoch": 0.7048991942786209, + "grad_norm": 2.1261236667633057, + "learning_rate": 4.230893919038461e-06, + "loss": 0.1345946431159973, + "step": 26760 + }, + { + "epoch": 0.7051626095231196, + "grad_norm": 2.012589931488037, + "learning_rate": 4.223927532029592e-06, + "loss": 0.10921218395233154, + "step": 26770 + }, + { + "epoch": 0.7054260247676184, + "grad_norm": 1.6391212940216064, + "learning_rate": 4.21696534909546e-06, + "loss": 0.13766399621963502, + "step": 26780 + }, + { + "epoch": 0.7056894400121171, + "grad_norm": 2.2703144550323486, + "learning_rate": 4.2100073753034535e-06, + "loss": 0.11982831954956055, + "step": 26790 + }, + { + "epoch": 0.7059528552566159, + "grad_norm": 2.8428666591644287, + "learning_rate": 4.203053615717884e-06, + "loss": 0.12256717681884766, + "step": 26800 + }, + { + "epoch": 0.7062162705011146, + "grad_norm": 1.7090121507644653, + "learning_rate": 4.196104075399986e-06, + "loss": 0.13435567617416383, + "step": 26810 + }, + { + "epoch": 0.7064796857456133, + "grad_norm": 1.5776444673538208, + "learning_rate": 4.1891587594079485e-06, + "loss": 0.12831175327301025, + "step": 26820 + }, + { + "epoch": 0.7067431009901121, + "grad_norm": 1.9801814556121826, + "learning_rate": 4.182217672796867e-06, + "loss": 0.1293949842453003, + "step": 26830 + }, + { + "epoch": 0.7070065162346107, + "grad_norm": 1.3276187181472778, + "learning_rate": 4.175280820618763e-06, + "loss": 0.11909120082855225, + "step": 26840 + }, + { + "epoch": 0.7072699314791095, + "grad_norm": 1.7764521837234497, + "learning_rate": 4.168348207922577e-06, + "loss": 0.10656702518463135, + "step": 26850 + }, + { + "epoch": 0.7075333467236082, + "grad_norm": 2.4440834522247314, + "learning_rate": 4.16141983975416e-06, + "loss": 0.1202507734298706, + "step": 26860 + }, + { + "epoch": 0.707796761968107, + "grad_norm": 3.0066120624542236, + "learning_rate": 4.154495721156287e-06, + "loss": 0.12616740465164183, + "step": 26870 + }, + { + "epoch": 0.7080601772126057, + "grad_norm": 3.144502639770508, + "learning_rate": 4.147575857168623e-06, + "loss": 0.12280927896499634, + "step": 26880 + }, + { + "epoch": 0.7083235924571045, + "grad_norm": 1.4385367631912231, + "learning_rate": 4.140660252827746e-06, + "loss": 0.10667153596878051, + "step": 26890 + }, + { + "epoch": 0.7085870077016032, + "grad_norm": 2.1413791179656982, + "learning_rate": 4.13374891316713e-06, + "loss": 0.12131205797195435, + "step": 26900 + }, + { + "epoch": 0.7088504229461019, + "grad_norm": 1.5054233074188232, + "learning_rate": 4.126841843217144e-06, + "loss": 0.12172902822494507, + "step": 26910 + }, + { + "epoch": 0.7091138381906007, + "grad_norm": 2.5925302505493164, + "learning_rate": 4.1199390480050605e-06, + "loss": 0.12055661678314208, + "step": 26920 + }, + { + "epoch": 0.7093772534350994, + "grad_norm": 1.99134361743927, + "learning_rate": 4.113040532555021e-06, + "loss": 0.11421711444854736, + "step": 26930 + }, + { + "epoch": 0.7096406686795982, + "grad_norm": 1.9981968402862549, + "learning_rate": 4.106146301888065e-06, + "loss": 0.12709773778915406, + "step": 26940 + }, + { + "epoch": 0.7099040839240969, + "grad_norm": 1.6902313232421875, + "learning_rate": 4.0992563610221155e-06, + "loss": 0.11238172054290771, + "step": 26950 + }, + { + "epoch": 0.7101674991685957, + "grad_norm": 2.811741590499878, + "learning_rate": 4.092370714971966e-06, + "loss": 0.12160882949829102, + "step": 26960 + }, + { + "epoch": 0.7104309144130944, + "grad_norm": 2.5249977111816406, + "learning_rate": 4.085489368749287e-06, + "loss": 0.12666621208190917, + "step": 26970 + }, + { + "epoch": 0.7106943296575932, + "grad_norm": 1.857527494430542, + "learning_rate": 4.07861232736262e-06, + "loss": 0.12547557353973388, + "step": 26980 + }, + { + "epoch": 0.7109577449020918, + "grad_norm": 3.9893124103546143, + "learning_rate": 4.0717395958173665e-06, + "loss": 0.123392653465271, + "step": 26990 + }, + { + "epoch": 0.7112211601465905, + "grad_norm": 4.076525688171387, + "learning_rate": 4.064871179115807e-06, + "loss": 0.12511038780212402, + "step": 27000 + }, + { + "epoch": 0.7114845753910893, + "grad_norm": 2.443826198577881, + "learning_rate": 4.0580070822570694e-06, + "loss": 0.12057487964630127, + "step": 27010 + }, + { + "epoch": 0.711747990635588, + "grad_norm": 1.3156691789627075, + "learning_rate": 4.051147310237129e-06, + "loss": 0.12211661338806153, + "step": 27020 + }, + { + "epoch": 0.7120114058800868, + "grad_norm": 1.56373131275177, + "learning_rate": 4.044291868048837e-06, + "loss": 0.10941388607025146, + "step": 27030 + }, + { + "epoch": 0.7122748211245855, + "grad_norm": 2.0344319343566895, + "learning_rate": 4.0374407606818684e-06, + "loss": 0.11895418167114258, + "step": 27040 + }, + { + "epoch": 0.7125382363690843, + "grad_norm": 3.2524263858795166, + "learning_rate": 4.03059399312277e-06, + "loss": 0.14040277004241944, + "step": 27050 + }, + { + "epoch": 0.712801651613583, + "grad_norm": 2.5406880378723145, + "learning_rate": 4.0237515703549015e-06, + "loss": 0.11775069236755371, + "step": 27060 + }, + { + "epoch": 0.7130650668580818, + "grad_norm": 1.8621793985366821, + "learning_rate": 4.0169134973584726e-06, + "loss": 0.10709633827209472, + "step": 27070 + }, + { + "epoch": 0.7133284821025805, + "grad_norm": 1.851054310798645, + "learning_rate": 4.010079779110537e-06, + "loss": 0.12014293670654297, + "step": 27080 + }, + { + "epoch": 0.7135918973470792, + "grad_norm": 1.447886347770691, + "learning_rate": 4.003250420584966e-06, + "loss": 0.13499667644500732, + "step": 27090 + }, + { + "epoch": 0.713855312591578, + "grad_norm": 3.467099905014038, + "learning_rate": 3.996425426752459e-06, + "loss": 0.13257734775543212, + "step": 27100 + }, + { + "epoch": 0.7141187278360767, + "grad_norm": 2.0391347408294678, + "learning_rate": 3.989604802580545e-06, + "loss": 0.12530720233917236, + "step": 27110 + }, + { + "epoch": 0.7143821430805755, + "grad_norm": 2.0103931427001953, + "learning_rate": 3.982788553033561e-06, + "loss": 0.1182249903678894, + "step": 27120 + }, + { + "epoch": 0.7146455583250741, + "grad_norm": 2.4394376277923584, + "learning_rate": 3.975976683072676e-06, + "loss": 0.1267741084098816, + "step": 27130 + }, + { + "epoch": 0.714908973569573, + "grad_norm": 2.288142204284668, + "learning_rate": 3.9691691976558636e-06, + "loss": 0.13226401805877686, + "step": 27140 + }, + { + "epoch": 0.7151723888140716, + "grad_norm": 2.040182590484619, + "learning_rate": 3.9623661017378946e-06, + "loss": 0.11685832738876342, + "step": 27150 + }, + { + "epoch": 0.7154358040585704, + "grad_norm": 2.045790433883667, + "learning_rate": 3.955567400270365e-06, + "loss": 0.12473411560058593, + "step": 27160 + }, + { + "epoch": 0.7156992193030691, + "grad_norm": 1.968262791633606, + "learning_rate": 3.948773098201657e-06, + "loss": 0.11133127212524414, + "step": 27170 + }, + { + "epoch": 0.7159626345475678, + "grad_norm": 3.0325751304626465, + "learning_rate": 3.941983200476969e-06, + "loss": 0.12660932540893555, + "step": 27180 + }, + { + "epoch": 0.7162260497920666, + "grad_norm": 2.2582433223724365, + "learning_rate": 3.935197712038267e-06, + "loss": 0.12011477947235108, + "step": 27190 + }, + { + "epoch": 0.7164894650365653, + "grad_norm": 1.339880108833313, + "learning_rate": 3.928416637824326e-06, + "loss": 0.11319893598556519, + "step": 27200 + }, + { + "epoch": 0.7167528802810641, + "grad_norm": 2.0244088172912598, + "learning_rate": 3.921639982770709e-06, + "loss": 0.1320235013961792, + "step": 27210 + }, + { + "epoch": 0.7170162955255628, + "grad_norm": 1.3608853816986084, + "learning_rate": 3.914867751809756e-06, + "loss": 0.11979753971099853, + "step": 27220 + }, + { + "epoch": 0.7172797107700616, + "grad_norm": 2.4791576862335205, + "learning_rate": 3.908099949870588e-06, + "loss": 0.12551684379577638, + "step": 27230 + }, + { + "epoch": 0.7175431260145603, + "grad_norm": 2.282327175140381, + "learning_rate": 3.901336581879102e-06, + "loss": 0.11630229949951172, + "step": 27240 + }, + { + "epoch": 0.717806541259059, + "grad_norm": 2.3499505519866943, + "learning_rate": 3.894577652757968e-06, + "loss": 0.12728703022003174, + "step": 27250 + }, + { + "epoch": 0.7180699565035578, + "grad_norm": 1.475678563117981, + "learning_rate": 3.887823167426633e-06, + "loss": 0.12068568468093872, + "step": 27260 + }, + { + "epoch": 0.7183333717480564, + "grad_norm": 1.4407235383987427, + "learning_rate": 3.8810731308013005e-06, + "loss": 0.11354503631591797, + "step": 27270 + }, + { + "epoch": 0.7185967869925552, + "grad_norm": 1.7043343782424927, + "learning_rate": 3.874327547794931e-06, + "loss": 0.11947088241577149, + "step": 27280 + }, + { + "epoch": 0.7188602022370539, + "grad_norm": 2.194348096847534, + "learning_rate": 3.8675864233172575e-06, + "loss": 0.11811175346374511, + "step": 27290 + }, + { + "epoch": 0.7191236174815527, + "grad_norm": 2.2106809616088867, + "learning_rate": 3.860849762274759e-06, + "loss": 0.11653542518615723, + "step": 27300 + }, + { + "epoch": 0.7193870327260514, + "grad_norm": 2.580744981765747, + "learning_rate": 3.854117569570674e-06, + "loss": 0.1189916491508484, + "step": 27310 + }, + { + "epoch": 0.7196504479705502, + "grad_norm": 1.6707990169525146, + "learning_rate": 3.8473898501049745e-06, + "loss": 0.11945688724517822, + "step": 27320 + }, + { + "epoch": 0.7199138632150489, + "grad_norm": 1.9711782932281494, + "learning_rate": 3.840666608774385e-06, + "loss": 0.11008464097976685, + "step": 27330 + }, + { + "epoch": 0.7201772784595476, + "grad_norm": 2.8936479091644287, + "learning_rate": 3.833947850472378e-06, + "loss": 0.11834237575531006, + "step": 27340 + }, + { + "epoch": 0.7204406937040464, + "grad_norm": 2.555882215499878, + "learning_rate": 3.827233580089152e-06, + "loss": 0.13846733570098876, + "step": 27350 + }, + { + "epoch": 0.7207041089485451, + "grad_norm": 2.795130968093872, + "learning_rate": 3.82052380251164e-06, + "loss": 0.12253649234771728, + "step": 27360 + }, + { + "epoch": 0.7209675241930439, + "grad_norm": 1.598934531211853, + "learning_rate": 3.8138185226235114e-06, + "loss": 0.11128857135772705, + "step": 27370 + }, + { + "epoch": 0.7212309394375426, + "grad_norm": 3.322619915008545, + "learning_rate": 3.8071177453051523e-06, + "loss": 0.1278993606567383, + "step": 27380 + }, + { + "epoch": 0.7214943546820414, + "grad_norm": 2.5924112796783447, + "learning_rate": 3.8004214754336853e-06, + "loss": 0.11856110095977783, + "step": 27390 + }, + { + "epoch": 0.7217577699265401, + "grad_norm": 1.6891505718231201, + "learning_rate": 3.7937297178829458e-06, + "loss": 0.10729761123657226, + "step": 27400 + }, + { + "epoch": 0.7220211851710389, + "grad_norm": 2.5940134525299072, + "learning_rate": 3.7870424775234703e-06, + "loss": 0.11109875440597534, + "step": 27410 + }, + { + "epoch": 0.7222846004155375, + "grad_norm": 1.4519281387329102, + "learning_rate": 3.780359759222534e-06, + "loss": 0.1306913137435913, + "step": 27420 + }, + { + "epoch": 0.7225480156600362, + "grad_norm": 1.9856945276260376, + "learning_rate": 3.773681567844103e-06, + "loss": 0.12837846279144288, + "step": 27430 + }, + { + "epoch": 0.722811430904535, + "grad_norm": 1.5893417596817017, + "learning_rate": 3.767007908248852e-06, + "loss": 0.11391992568969726, + "step": 27440 + }, + { + "epoch": 0.7230748461490337, + "grad_norm": 2.167315721511841, + "learning_rate": 3.7603387852941608e-06, + "loss": 0.12183933258056641, + "step": 27450 + }, + { + "epoch": 0.7233382613935325, + "grad_norm": 2.343026876449585, + "learning_rate": 3.753674203834099e-06, + "loss": 0.10509202480316163, + "step": 27460 + }, + { + "epoch": 0.7236016766380312, + "grad_norm": 2.546464204788208, + "learning_rate": 3.747014168719445e-06, + "loss": 0.11198716163635254, + "step": 27470 + }, + { + "epoch": 0.72386509188253, + "grad_norm": 2.689006805419922, + "learning_rate": 3.740358684797656e-06, + "loss": 0.11870479583740234, + "step": 27480 + }, + { + "epoch": 0.7241285071270287, + "grad_norm": 2.029829978942871, + "learning_rate": 3.7337077569128797e-06, + "loss": 0.11470954418182373, + "step": 27490 + }, + { + "epoch": 0.7243919223715275, + "grad_norm": 2.271672248840332, + "learning_rate": 3.7270613899059495e-06, + "loss": 0.10412956476211548, + "step": 27500 + }, + { + "epoch": 0.7246553376160262, + "grad_norm": 2.0853962898254395, + "learning_rate": 3.7204195886143736e-06, + "loss": 0.10953856706619262, + "step": 27510 + }, + { + "epoch": 0.7249187528605249, + "grad_norm": 2.3700003623962402, + "learning_rate": 3.713782357872352e-06, + "loss": 0.12143164873123169, + "step": 27520 + }, + { + "epoch": 0.7251821681050237, + "grad_norm": 1.7592706680297852, + "learning_rate": 3.7071497025107373e-06, + "loss": 0.12474308013916016, + "step": 27530 + }, + { + "epoch": 0.7254455833495224, + "grad_norm": 2.031510829925537, + "learning_rate": 3.7005216273570623e-06, + "loss": 0.10719377994537353, + "step": 27540 + }, + { + "epoch": 0.7257089985940212, + "grad_norm": 3.017385244369507, + "learning_rate": 3.693898137235532e-06, + "loss": 0.10991458892822266, + "step": 27550 + }, + { + "epoch": 0.7259724138385198, + "grad_norm": 2.213019847869873, + "learning_rate": 3.6872792369670074e-06, + "loss": 0.12189595699310303, + "step": 27560 + }, + { + "epoch": 0.7262358290830186, + "grad_norm": 1.8190817832946777, + "learning_rate": 3.6806649313690066e-06, + "loss": 0.12225488424301148, + "step": 27570 + }, + { + "epoch": 0.7264992443275173, + "grad_norm": 1.5902248620986938, + "learning_rate": 3.6740552252557093e-06, + "loss": 0.12539159059524535, + "step": 27580 + }, + { + "epoch": 0.7267626595720161, + "grad_norm": 1.4588388204574585, + "learning_rate": 3.667450123437939e-06, + "loss": 0.11465888023376465, + "step": 27590 + }, + { + "epoch": 0.7270260748165148, + "grad_norm": 1.6508336067199707, + "learning_rate": 3.6608496307231823e-06, + "loss": 0.11578365564346313, + "step": 27600 + }, + { + "epoch": 0.7272894900610135, + "grad_norm": 2.2258894443511963, + "learning_rate": 3.6542537519155617e-06, + "loss": 0.10102317333221436, + "step": 27610 + }, + { + "epoch": 0.7275529053055123, + "grad_norm": 2.1271777153015137, + "learning_rate": 3.647662491815833e-06, + "loss": 0.12745194435119628, + "step": 27620 + }, + { + "epoch": 0.727816320550011, + "grad_norm": 2.2298097610473633, + "learning_rate": 3.641075855221409e-06, + "loss": 0.12908843755722046, + "step": 27630 + }, + { + "epoch": 0.7280797357945098, + "grad_norm": 1.5264685153961182, + "learning_rate": 3.634493846926321e-06, + "loss": 0.1284632444381714, + "step": 27640 + }, + { + "epoch": 0.7283431510390085, + "grad_norm": 1.8772872686386108, + "learning_rate": 3.62791647172125e-06, + "loss": 0.12904932498931884, + "step": 27650 + }, + { + "epoch": 0.7286065662835073, + "grad_norm": 2.997443437576294, + "learning_rate": 3.621343734393482e-06, + "loss": 0.12207945585250854, + "step": 27660 + }, + { + "epoch": 0.728869981528006, + "grad_norm": 1.960983157157898, + "learning_rate": 3.6147756397269397e-06, + "loss": 0.11739420890808105, + "step": 27670 + }, + { + "epoch": 0.7291333967725048, + "grad_norm": 2.8179759979248047, + "learning_rate": 3.6082121925021717e-06, + "loss": 0.11512085199356079, + "step": 27680 + }, + { + "epoch": 0.7293968120170035, + "grad_norm": 1.6591345071792603, + "learning_rate": 3.601653397496335e-06, + "loss": 0.11137815713882446, + "step": 27690 + }, + { + "epoch": 0.7296602272615021, + "grad_norm": 1.739301323890686, + "learning_rate": 3.5950992594832044e-06, + "loss": 0.1216360330581665, + "step": 27700 + }, + { + "epoch": 0.729923642506001, + "grad_norm": 1.888721227645874, + "learning_rate": 3.588549783233164e-06, + "loss": 0.10083291530609131, + "step": 27710 + }, + { + "epoch": 0.7301870577504996, + "grad_norm": 2.555622100830078, + "learning_rate": 3.582004973513201e-06, + "loss": 0.11899766921997071, + "step": 27720 + }, + { + "epoch": 0.7304504729949984, + "grad_norm": 3.156766891479492, + "learning_rate": 3.575464835086917e-06, + "loss": 0.12469673156738281, + "step": 27730 + }, + { + "epoch": 0.7307138882394971, + "grad_norm": 3.022470474243164, + "learning_rate": 3.568929372714508e-06, + "loss": 0.1385214924812317, + "step": 27740 + }, + { + "epoch": 0.7309773034839959, + "grad_norm": 3.247267484664917, + "learning_rate": 3.562398591152755e-06, + "loss": 0.1230552077293396, + "step": 27750 + }, + { + "epoch": 0.7312407187284946, + "grad_norm": 2.0307517051696777, + "learning_rate": 3.5558724951550515e-06, + "loss": 0.11038093566894532, + "step": 27760 + }, + { + "epoch": 0.7315041339729933, + "grad_norm": 1.7537480592727661, + "learning_rate": 3.549351089471367e-06, + "loss": 0.10726062059402466, + "step": 27770 + }, + { + "epoch": 0.7317675492174921, + "grad_norm": 2.3625705242156982, + "learning_rate": 3.54283437884827e-06, + "loss": 0.1259612798690796, + "step": 27780 + }, + { + "epoch": 0.7320309644619908, + "grad_norm": 2.389314651489258, + "learning_rate": 3.536322368028895e-06, + "loss": 0.1241368293762207, + "step": 27790 + }, + { + "epoch": 0.7322943797064896, + "grad_norm": 2.17557430267334, + "learning_rate": 3.5298150617529635e-06, + "loss": 0.12591085433959961, + "step": 27800 + }, + { + "epoch": 0.7325577949509883, + "grad_norm": 2.488030195236206, + "learning_rate": 3.5233124647567817e-06, + "loss": 0.12855825424194336, + "step": 27810 + }, + { + "epoch": 0.7328212101954871, + "grad_norm": 1.784310221672058, + "learning_rate": 3.516814581773216e-06, + "loss": 0.1197272539138794, + "step": 27820 + }, + { + "epoch": 0.7330846254399858, + "grad_norm": 2.3673925399780273, + "learning_rate": 3.5103214175317056e-06, + "loss": 0.11585829257965088, + "step": 27830 + }, + { + "epoch": 0.7333480406844846, + "grad_norm": 1.5819833278656006, + "learning_rate": 3.5038329767582567e-06, + "loss": 0.12029169797897339, + "step": 27840 + }, + { + "epoch": 0.7336114559289832, + "grad_norm": 2.4856081008911133, + "learning_rate": 3.4973492641754315e-06, + "loss": 0.11129606962203979, + "step": 27850 + }, + { + "epoch": 0.7338748711734819, + "grad_norm": 2.348841905593872, + "learning_rate": 3.4908702845023643e-06, + "loss": 0.11572437286376953, + "step": 27860 + }, + { + "epoch": 0.7341382864179807, + "grad_norm": 1.8997622728347778, + "learning_rate": 3.484396042454734e-06, + "loss": 0.11634625196456909, + "step": 27870 + }, + { + "epoch": 0.7344017016624794, + "grad_norm": 2.1162328720092773, + "learning_rate": 3.477926542744766e-06, + "loss": 0.12197237014770508, + "step": 27880 + }, + { + "epoch": 0.7346651169069782, + "grad_norm": 3.328286647796631, + "learning_rate": 3.4714617900812485e-06, + "loss": 0.13137012720108032, + "step": 27890 + }, + { + "epoch": 0.7349285321514769, + "grad_norm": 3.474308490753174, + "learning_rate": 3.465001789169502e-06, + "loss": 0.1266784906387329, + "step": 27900 + }, + { + "epoch": 0.7351919473959757, + "grad_norm": 2.0428249835968018, + "learning_rate": 3.458546544711404e-06, + "loss": 0.11857596635818482, + "step": 27910 + }, + { + "epoch": 0.7354553626404744, + "grad_norm": 2.7461633682250977, + "learning_rate": 3.45209606140535e-06, + "loss": 0.11612544059753419, + "step": 27920 + }, + { + "epoch": 0.7357187778849732, + "grad_norm": 2.1998343467712402, + "learning_rate": 3.4456503439462797e-06, + "loss": 0.11919455528259278, + "step": 27930 + }, + { + "epoch": 0.7359821931294719, + "grad_norm": 1.9298081398010254, + "learning_rate": 3.4392093970256703e-06, + "loss": 0.1362772822380066, + "step": 27940 + }, + { + "epoch": 0.7362456083739706, + "grad_norm": 2.0755512714385986, + "learning_rate": 3.4327732253315195e-06, + "loss": 0.12882509231567382, + "step": 27950 + }, + { + "epoch": 0.7365090236184694, + "grad_norm": 2.20558500289917, + "learning_rate": 3.4263418335483477e-06, + "loss": 0.1188694953918457, + "step": 27960 + }, + { + "epoch": 0.7367724388629681, + "grad_norm": 1.6727770566940308, + "learning_rate": 3.419915226357203e-06, + "loss": 0.11101734638214111, + "step": 27970 + }, + { + "epoch": 0.7370358541074669, + "grad_norm": 2.180335283279419, + "learning_rate": 3.413493408435642e-06, + "loss": 0.12124950885772705, + "step": 27980 + }, + { + "epoch": 0.7372992693519655, + "grad_norm": 2.1919219493865967, + "learning_rate": 3.407076384457747e-06, + "loss": 0.10855158567428588, + "step": 27990 + }, + { + "epoch": 0.7375626845964643, + "grad_norm": 1.7051347494125366, + "learning_rate": 3.400664159094106e-06, + "loss": 0.10191359519958496, + "step": 28000 + }, + { + "epoch": 0.737826099840963, + "grad_norm": 1.9416378736495972, + "learning_rate": 3.394256737011803e-06, + "loss": 0.11091425418853759, + "step": 28010 + }, + { + "epoch": 0.7380895150854618, + "grad_norm": 2.092737913131714, + "learning_rate": 3.3878541228744455e-06, + "loss": 0.10375857353210449, + "step": 28020 + }, + { + "epoch": 0.7383529303299605, + "grad_norm": 1.942246437072754, + "learning_rate": 3.38145632134213e-06, + "loss": 0.11361316442489625, + "step": 28030 + }, + { + "epoch": 0.7386163455744592, + "grad_norm": 2.4419429302215576, + "learning_rate": 3.3750633370714527e-06, + "loss": 0.11355415582656861, + "step": 28040 + }, + { + "epoch": 0.738879760818958, + "grad_norm": 1.8917027711868286, + "learning_rate": 3.368675174715501e-06, + "loss": 0.11543427705764771, + "step": 28050 + }, + { + "epoch": 0.7391431760634567, + "grad_norm": 3.1051034927368164, + "learning_rate": 3.3622918389238544e-06, + "loss": 0.11629109382629395, + "step": 28060 + }, + { + "epoch": 0.7394065913079555, + "grad_norm": 1.9480150938034058, + "learning_rate": 3.3559133343425865e-06, + "loss": 0.12893034219741822, + "step": 28070 + }, + { + "epoch": 0.7396700065524542, + "grad_norm": 2.36897611618042, + "learning_rate": 3.3495396656142442e-06, + "loss": 0.11220893859863282, + "step": 28080 + }, + { + "epoch": 0.739933421796953, + "grad_norm": 1.6527416706085205, + "learning_rate": 3.343170837377858e-06, + "loss": 0.12132742404937744, + "step": 28090 + }, + { + "epoch": 0.7401968370414517, + "grad_norm": 2.1033756732940674, + "learning_rate": 3.336806854268936e-06, + "loss": 0.10846874713897706, + "step": 28100 + }, + { + "epoch": 0.7404602522859505, + "grad_norm": 2.213559150695801, + "learning_rate": 3.3304477209194574e-06, + "loss": 0.12202613353729248, + "step": 28110 + }, + { + "epoch": 0.7407236675304492, + "grad_norm": 1.9922734498977661, + "learning_rate": 3.3240934419578828e-06, + "loss": 0.11584949493408203, + "step": 28120 + }, + { + "epoch": 0.7409870827749478, + "grad_norm": 1.3553136587142944, + "learning_rate": 3.3177440220091207e-06, + "loss": 0.11099700927734375, + "step": 28130 + }, + { + "epoch": 0.7412504980194466, + "grad_norm": 1.9503971338272095, + "learning_rate": 3.3113994656945537e-06, + "loss": 0.11193999052047729, + "step": 28140 + }, + { + "epoch": 0.7415139132639453, + "grad_norm": 2.921067714691162, + "learning_rate": 3.3050597776320293e-06, + "loss": 0.10863912105560303, + "step": 28150 + }, + { + "epoch": 0.7417773285084441, + "grad_norm": 1.990405797958374, + "learning_rate": 3.298724962435842e-06, + "loss": 0.10606728792190552, + "step": 28160 + }, + { + "epoch": 0.7420407437529428, + "grad_norm": 3.4758801460266113, + "learning_rate": 3.292395024716746e-06, + "loss": 0.12185709476470948, + "step": 28170 + }, + { + "epoch": 0.7423041589974416, + "grad_norm": 2.3218963146209717, + "learning_rate": 3.2860699690819407e-06, + "loss": 0.11246616840362549, + "step": 28180 + }, + { + "epoch": 0.7425675742419403, + "grad_norm": 2.546417474746704, + "learning_rate": 3.2797498001350737e-06, + "loss": 0.1138612151145935, + "step": 28190 + }, + { + "epoch": 0.7428309894864391, + "grad_norm": 2.6018333435058594, + "learning_rate": 3.27343452247624e-06, + "loss": 0.10571072101593018, + "step": 28200 + }, + { + "epoch": 0.7430944047309378, + "grad_norm": 2.0279781818389893, + "learning_rate": 3.267124140701974e-06, + "loss": 0.11523489952087403, + "step": 28210 + }, + { + "epoch": 0.7433578199754365, + "grad_norm": 1.6594549417495728, + "learning_rate": 3.2608186594052314e-06, + "loss": 0.10535740852355957, + "step": 28220 + }, + { + "epoch": 0.7436212352199353, + "grad_norm": 2.2031846046447754, + "learning_rate": 3.2545180831754264e-06, + "loss": 0.11648386716842651, + "step": 28230 + }, + { + "epoch": 0.743884650464434, + "grad_norm": 2.107219696044922, + "learning_rate": 3.248222416598381e-06, + "loss": 0.09860738515853881, + "step": 28240 + }, + { + "epoch": 0.7441480657089328, + "grad_norm": 1.7066762447357178, + "learning_rate": 3.2419316642563646e-06, + "loss": 0.10705703496932983, + "step": 28250 + }, + { + "epoch": 0.7444114809534315, + "grad_norm": 2.317791223526001, + "learning_rate": 3.235645830728047e-06, + "loss": 0.11860129833221436, + "step": 28260 + }, + { + "epoch": 0.7446748961979303, + "grad_norm": 1.8128211498260498, + "learning_rate": 3.2293649205885293e-06, + "loss": 0.1057919979095459, + "step": 28270 + }, + { + "epoch": 0.744938311442429, + "grad_norm": 1.4090392589569092, + "learning_rate": 3.2230889384093355e-06, + "loss": 0.10545306205749512, + "step": 28280 + }, + { + "epoch": 0.7452017266869276, + "grad_norm": 3.704796314239502, + "learning_rate": 3.2168178887583947e-06, + "loss": 0.1114679217338562, + "step": 28290 + }, + { + "epoch": 0.7454651419314264, + "grad_norm": 5.21004056930542, + "learning_rate": 3.210551776200044e-06, + "loss": 0.1063854455947876, + "step": 28300 + }, + { + "epoch": 0.7457285571759251, + "grad_norm": 2.47674298286438, + "learning_rate": 3.204290605295033e-06, + "loss": 0.11737663745880127, + "step": 28310 + }, + { + "epoch": 0.7459919724204239, + "grad_norm": 3.175201177597046, + "learning_rate": 3.1980343806005077e-06, + "loss": 0.11716958284378051, + "step": 28320 + }, + { + "epoch": 0.7462553876649226, + "grad_norm": 2.5023279190063477, + "learning_rate": 3.1917831066700255e-06, + "loss": 0.11865862607955932, + "step": 28330 + }, + { + "epoch": 0.7465188029094214, + "grad_norm": 1.88677179813385, + "learning_rate": 3.1855367880535336e-06, + "loss": 0.12504181861877442, + "step": 28340 + }, + { + "epoch": 0.7467822181539201, + "grad_norm": 2.461745023727417, + "learning_rate": 3.179295429297361e-06, + "loss": 0.12324681282043456, + "step": 28350 + }, + { + "epoch": 0.7470456333984189, + "grad_norm": 2.3057684898376465, + "learning_rate": 3.173059034944248e-06, + "loss": 0.1121549129486084, + "step": 28360 + }, + { + "epoch": 0.7473090486429176, + "grad_norm": 1.9620640277862549, + "learning_rate": 3.1668276095333063e-06, + "loss": 0.11108381748199463, + "step": 28370 + }, + { + "epoch": 0.7475724638874163, + "grad_norm": 1.620008111000061, + "learning_rate": 3.160601157600045e-06, + "loss": 0.11809587478637695, + "step": 28380 + }, + { + "epoch": 0.7478358791319151, + "grad_norm": 2.0724828243255615, + "learning_rate": 3.154379683676334e-06, + "loss": 0.10042437314987182, + "step": 28390 + }, + { + "epoch": 0.7480992943764138, + "grad_norm": 2.713256597518921, + "learning_rate": 3.148163192290432e-06, + "loss": 0.1099932074546814, + "step": 28400 + }, + { + "epoch": 0.7483627096209126, + "grad_norm": 2.97103214263916, + "learning_rate": 3.1419516879669744e-06, + "loss": 0.11471521854400635, + "step": 28410 + }, + { + "epoch": 0.7486261248654112, + "grad_norm": 2.2565765380859375, + "learning_rate": 3.1357451752269595e-06, + "loss": 0.11897482872009277, + "step": 28420 + }, + { + "epoch": 0.74888954010991, + "grad_norm": 2.228559970855713, + "learning_rate": 3.1295436585877558e-06, + "loss": 0.12785837650299073, + "step": 28430 + }, + { + "epoch": 0.7491529553544087, + "grad_norm": 5.651340484619141, + "learning_rate": 3.1233471425630947e-06, + "loss": 0.11633776426315308, + "step": 28440 + }, + { + "epoch": 0.7494163705989075, + "grad_norm": 2.7284231185913086, + "learning_rate": 3.1171556316630626e-06, + "loss": 0.10986015796661378, + "step": 28450 + }, + { + "epoch": 0.7496797858434062, + "grad_norm": 2.599918842315674, + "learning_rate": 3.110969130394117e-06, + "loss": 0.09533497095108032, + "step": 28460 + }, + { + "epoch": 0.7499432010879049, + "grad_norm": 2.038268804550171, + "learning_rate": 3.1047876432590596e-06, + "loss": 0.11859664916992188, + "step": 28470 + }, + { + "epoch": 0.7502066163324037, + "grad_norm": 2.185943603515625, + "learning_rate": 3.0986111747570335e-06, + "loss": 0.11069562435150146, + "step": 28480 + }, + { + "epoch": 0.7504700315769024, + "grad_norm": 2.1472835540771484, + "learning_rate": 3.092439729383548e-06, + "loss": 0.11790274381637574, + "step": 28490 + }, + { + "epoch": 0.7507334468214012, + "grad_norm": 2.39027738571167, + "learning_rate": 3.0862733116304435e-06, + "loss": 0.11101759672164917, + "step": 28500 + }, + { + "epoch": 0.7509968620658999, + "grad_norm": 2.1168699264526367, + "learning_rate": 3.0801119259859115e-06, + "loss": 0.12099113464355468, + "step": 28510 + }, + { + "epoch": 0.7512602773103987, + "grad_norm": 2.844245195388794, + "learning_rate": 3.0739555769344666e-06, + "loss": 0.10415678024291992, + "step": 28520 + }, + { + "epoch": 0.7515236925548974, + "grad_norm": 2.1987814903259277, + "learning_rate": 3.067804268956964e-06, + "loss": 0.11616944074630738, + "step": 28530 + }, + { + "epoch": 0.7517871077993962, + "grad_norm": 2.598059892654419, + "learning_rate": 3.0616580065305977e-06, + "loss": 0.11030938625335693, + "step": 28540 + }, + { + "epoch": 0.7520505230438949, + "grad_norm": 1.9699128866195679, + "learning_rate": 3.055516794128881e-06, + "loss": 0.11766586303710938, + "step": 28550 + }, + { + "epoch": 0.7523139382883935, + "grad_norm": 3.4788661003112793, + "learning_rate": 3.0493806362216516e-06, + "loss": 0.10998309850692749, + "step": 28560 + }, + { + "epoch": 0.7525773535328923, + "grad_norm": 2.9202020168304443, + "learning_rate": 3.0432495372750692e-06, + "loss": 0.10635056495666503, + "step": 28570 + }, + { + "epoch": 0.752840768777391, + "grad_norm": 2.5616695880889893, + "learning_rate": 3.037123501751611e-06, + "loss": 0.11015079021453858, + "step": 28580 + }, + { + "epoch": 0.7531041840218898, + "grad_norm": 3.8711302280426025, + "learning_rate": 3.0310025341100757e-06, + "loss": 0.11429985761642455, + "step": 28590 + }, + { + "epoch": 0.7533675992663885, + "grad_norm": 2.3962244987487793, + "learning_rate": 3.0248866388055674e-06, + "loss": 0.11821801662445068, + "step": 28600 + }, + { + "epoch": 0.7536310145108873, + "grad_norm": 1.817092776298523, + "learning_rate": 3.018775820289488e-06, + "loss": 0.09194302558898926, + "step": 28610 + }, + { + "epoch": 0.753894429755386, + "grad_norm": 2.7037010192871094, + "learning_rate": 3.0126700830095668e-06, + "loss": 0.11291483640670777, + "step": 28620 + }, + { + "epoch": 0.7541578449998848, + "grad_norm": 3.5120296478271484, + "learning_rate": 3.006569431409817e-06, + "loss": 0.10337754487991332, + "step": 28630 + }, + { + "epoch": 0.7544212602443835, + "grad_norm": 2.7358076572418213, + "learning_rate": 3.000473869930558e-06, + "loss": 0.10415666103363037, + "step": 28640 + }, + { + "epoch": 0.7546846754888822, + "grad_norm": 2.2263593673706055, + "learning_rate": 2.9943834030084006e-06, + "loss": 0.12715294361114501, + "step": 28650 + }, + { + "epoch": 0.754948090733381, + "grad_norm": 2.1171228885650635, + "learning_rate": 2.988298035076249e-06, + "loss": 0.10066454410552979, + "step": 28660 + }, + { + "epoch": 0.7552115059778797, + "grad_norm": 2.462890148162842, + "learning_rate": 2.9822177705633005e-06, + "loss": 0.10910975933074951, + "step": 28670 + }, + { + "epoch": 0.7554749212223785, + "grad_norm": 2.352280378341675, + "learning_rate": 2.976142613895031e-06, + "loss": 0.1053269624710083, + "step": 28680 + }, + { + "epoch": 0.7557383364668772, + "grad_norm": 2.054515838623047, + "learning_rate": 2.970072569493202e-06, + "loss": 0.10272719860076904, + "step": 28690 + }, + { + "epoch": 0.756001751711376, + "grad_norm": 3.1268043518066406, + "learning_rate": 2.9640076417758546e-06, + "loss": 0.1109468936920166, + "step": 28700 + }, + { + "epoch": 0.7562651669558746, + "grad_norm": 2.64375638961792, + "learning_rate": 2.9579478351573e-06, + "loss": 0.11090574264526368, + "step": 28710 + }, + { + "epoch": 0.7565285822003734, + "grad_norm": 2.992016077041626, + "learning_rate": 2.9518931540481377e-06, + "loss": 0.11675618886947632, + "step": 28720 + }, + { + "epoch": 0.7567919974448721, + "grad_norm": 1.4418258666992188, + "learning_rate": 2.945843602855215e-06, + "loss": 0.10776818990707397, + "step": 28730 + }, + { + "epoch": 0.7570554126893708, + "grad_norm": 2.691709518432617, + "learning_rate": 2.9397991859816588e-06, + "loss": 0.12032103538513184, + "step": 28740 + }, + { + "epoch": 0.7573188279338696, + "grad_norm": 2.2372989654541016, + "learning_rate": 2.9337599078268597e-06, + "loss": 0.11322319507598877, + "step": 28750 + }, + { + "epoch": 0.7575822431783683, + "grad_norm": 2.413177967071533, + "learning_rate": 2.927725772786464e-06, + "loss": 0.12143168449401856, + "step": 28760 + }, + { + "epoch": 0.7578456584228671, + "grad_norm": 1.7885555028915405, + "learning_rate": 2.9216967852523747e-06, + "loss": 0.10440824031829835, + "step": 28770 + }, + { + "epoch": 0.7581090736673658, + "grad_norm": 3.0227367877960205, + "learning_rate": 2.915672949612749e-06, + "loss": 0.10822809934616089, + "step": 28780 + }, + { + "epoch": 0.7583724889118646, + "grad_norm": 1.9576395750045776, + "learning_rate": 2.9096542702519904e-06, + "loss": 0.10745427608489991, + "step": 28790 + }, + { + "epoch": 0.7586359041563633, + "grad_norm": 2.208000421524048, + "learning_rate": 2.9036407515507613e-06, + "loss": 0.10682599544525147, + "step": 28800 + }, + { + "epoch": 0.758899319400862, + "grad_norm": 1.7568933963775635, + "learning_rate": 2.89763239788596e-06, + "loss": 0.10932720899581909, + "step": 28810 + }, + { + "epoch": 0.7591627346453608, + "grad_norm": 2.95475697517395, + "learning_rate": 2.8916292136307147e-06, + "loss": 0.11770594120025635, + "step": 28820 + }, + { + "epoch": 0.7594261498898595, + "grad_norm": 2.113166332244873, + "learning_rate": 2.885631203154412e-06, + "loss": 0.11966009140014648, + "step": 28830 + }, + { + "epoch": 0.7596895651343583, + "grad_norm": 2.5020804405212402, + "learning_rate": 2.879638370822656e-06, + "loss": 0.11431496143341065, + "step": 28840 + }, + { + "epoch": 0.759952980378857, + "grad_norm": 1.8628836870193481, + "learning_rate": 2.8736507209972996e-06, + "loss": 0.1055791974067688, + "step": 28850 + }, + { + "epoch": 0.7602163956233557, + "grad_norm": 2.678143262863159, + "learning_rate": 2.867668258036401e-06, + "loss": 0.11627008914947509, + "step": 28860 + }, + { + "epoch": 0.7604798108678544, + "grad_norm": 2.1036880016326904, + "learning_rate": 2.8616909862942565e-06, + "loss": 0.11081373691558838, + "step": 28870 + }, + { + "epoch": 0.7607432261123532, + "grad_norm": 2.086621046066284, + "learning_rate": 2.855718910121389e-06, + "loss": 0.10621250867843628, + "step": 28880 + }, + { + "epoch": 0.7610066413568519, + "grad_norm": 1.9281421899795532, + "learning_rate": 2.8497520338645278e-06, + "loss": 0.09438964128494262, + "step": 28890 + }, + { + "epoch": 0.7612700566013506, + "grad_norm": 1.9136593341827393, + "learning_rate": 2.843790361866625e-06, + "loss": 0.10351718664169311, + "step": 28900 + }, + { + "epoch": 0.7615334718458494, + "grad_norm": 1.8397423028945923, + "learning_rate": 2.837833898466842e-06, + "loss": 0.10649740695953369, + "step": 28910 + }, + { + "epoch": 0.7617968870903481, + "grad_norm": 2.0611207485198975, + "learning_rate": 2.831882648000547e-06, + "loss": 0.10676255226135253, + "step": 28920 + }, + { + "epoch": 0.7620603023348469, + "grad_norm": 2.358098030090332, + "learning_rate": 2.8259366147993217e-06, + "loss": 0.10361164808273315, + "step": 28930 + }, + { + "epoch": 0.7623237175793456, + "grad_norm": 2.1827902793884277, + "learning_rate": 2.819995803190948e-06, + "loss": 0.11374306678771973, + "step": 28940 + }, + { + "epoch": 0.7625871328238444, + "grad_norm": 3.3835904598236084, + "learning_rate": 2.8140602174993927e-06, + "loss": 0.10448834896087647, + "step": 28950 + }, + { + "epoch": 0.7628505480683431, + "grad_norm": 2.453949213027954, + "learning_rate": 2.8081298620448427e-06, + "loss": 0.10882518291473389, + "step": 28960 + }, + { + "epoch": 0.7631139633128419, + "grad_norm": 1.7577091455459595, + "learning_rate": 2.8022047411436604e-06, + "loss": 0.09835269451141357, + "step": 28970 + }, + { + "epoch": 0.7633773785573406, + "grad_norm": 2.8192009925842285, + "learning_rate": 2.7962848591084122e-06, + "loss": 0.12625932693481445, + "step": 28980 + }, + { + "epoch": 0.7636407938018392, + "grad_norm": 2.531100034713745, + "learning_rate": 2.7903702202478355e-06, + "loss": 0.1128846287727356, + "step": 28990 + }, + { + "epoch": 0.763904209046338, + "grad_norm": 4.788426876068115, + "learning_rate": 2.7844608288668586e-06, + "loss": 0.10368287563323975, + "step": 29000 + }, + { + "epoch": 0.7641676242908367, + "grad_norm": 1.7313182353973389, + "learning_rate": 2.7785566892665993e-06, + "loss": 0.11727116107940674, + "step": 29010 + }, + { + "epoch": 0.7644310395353355, + "grad_norm": 2.5190751552581787, + "learning_rate": 2.77265780574434e-06, + "loss": 0.12259666919708252, + "step": 29020 + }, + { + "epoch": 0.7646944547798342, + "grad_norm": 2.4139492511749268, + "learning_rate": 2.766764182593544e-06, + "loss": 0.1112183690071106, + "step": 29030 + }, + { + "epoch": 0.764957870024333, + "grad_norm": 1.7942827939987183, + "learning_rate": 2.760875824103846e-06, + "loss": 0.10873684883117676, + "step": 29040 + }, + { + "epoch": 0.7652212852688317, + "grad_norm": 3.2319607734680176, + "learning_rate": 2.7549927345610407e-06, + "loss": 0.1078611135482788, + "step": 29050 + }, + { + "epoch": 0.7654847005133305, + "grad_norm": 2.5782546997070312, + "learning_rate": 2.749114918247102e-06, + "loss": 0.1071505069732666, + "step": 29060 + }, + { + "epoch": 0.7657481157578292, + "grad_norm": 1.991343379020691, + "learning_rate": 2.7432423794401598e-06, + "loss": 0.09548298716545105, + "step": 29070 + }, + { + "epoch": 0.7660115310023279, + "grad_norm": 2.1509194374084473, + "learning_rate": 2.7373751224144905e-06, + "loss": 0.10765364170074462, + "step": 29080 + }, + { + "epoch": 0.7662749462468267, + "grad_norm": 2.2037036418914795, + "learning_rate": 2.731513151440547e-06, + "loss": 0.10936481952667236, + "step": 29090 + }, + { + "epoch": 0.7665383614913254, + "grad_norm": 1.7178722620010376, + "learning_rate": 2.7256564707849165e-06, + "loss": 0.11378359794616699, + "step": 29100 + }, + { + "epoch": 0.7668017767358242, + "grad_norm": 3.3877999782562256, + "learning_rate": 2.719805084710356e-06, + "loss": 0.12234458923339844, + "step": 29110 + }, + { + "epoch": 0.7670651919803229, + "grad_norm": 2.1996004581451416, + "learning_rate": 2.7139589974757454e-06, + "loss": 0.09587026238441468, + "step": 29120 + }, + { + "epoch": 0.7673286072248217, + "grad_norm": 2.319058895111084, + "learning_rate": 2.7081182133361207e-06, + "loss": 0.10682792663574218, + "step": 29130 + }, + { + "epoch": 0.7675920224693203, + "grad_norm": 1.7965444326400757, + "learning_rate": 2.7022827365426606e-06, + "loss": 0.11196770668029785, + "step": 29140 + }, + { + "epoch": 0.7678554377138191, + "grad_norm": 1.8581444025039673, + "learning_rate": 2.696452571342676e-06, + "loss": 0.11477812528610229, + "step": 29150 + }, + { + "epoch": 0.7681188529583178, + "grad_norm": 2.16499400138855, + "learning_rate": 2.6906277219796116e-06, + "loss": 0.09675438404083252, + "step": 29160 + }, + { + "epoch": 0.7683822682028165, + "grad_norm": 2.1936724185943604, + "learning_rate": 2.6848081926930447e-06, + "loss": 0.09673025608062744, + "step": 29170 + }, + { + "epoch": 0.7686456834473153, + "grad_norm": 2.4259588718414307, + "learning_rate": 2.6789939877186753e-06, + "loss": 0.10829490423202515, + "step": 29180 + }, + { + "epoch": 0.768909098691814, + "grad_norm": 2.300069570541382, + "learning_rate": 2.67318511128834e-06, + "loss": 0.10750792026519776, + "step": 29190 + }, + { + "epoch": 0.7691725139363128, + "grad_norm": 2.990957498550415, + "learning_rate": 2.66738156762999e-06, + "loss": 0.09994562268257141, + "step": 29200 + }, + { + "epoch": 0.7694359291808115, + "grad_norm": 3.1971774101257324, + "learning_rate": 2.6615833609676846e-06, + "loss": 0.11228828430175782, + "step": 29210 + }, + { + "epoch": 0.7696993444253103, + "grad_norm": 1.9332016706466675, + "learning_rate": 2.6557904955216185e-06, + "loss": 0.1132003903388977, + "step": 29220 + }, + { + "epoch": 0.769962759669809, + "grad_norm": 2.2459774017333984, + "learning_rate": 2.650002975508086e-06, + "loss": 0.0991871178150177, + "step": 29230 + }, + { + "epoch": 0.7702261749143078, + "grad_norm": 2.271679639816284, + "learning_rate": 2.6442208051394936e-06, + "loss": 0.10542857646942139, + "step": 29240 + }, + { + "epoch": 0.7704895901588065, + "grad_norm": 1.8365098237991333, + "learning_rate": 2.638443988624355e-06, + "loss": 0.09838100671768188, + "step": 29250 + }, + { + "epoch": 0.7707530054033052, + "grad_norm": 1.6570764780044556, + "learning_rate": 2.632672530167283e-06, + "loss": 0.0978689432144165, + "step": 29260 + }, + { + "epoch": 0.771016420647804, + "grad_norm": 2.1386752128601074, + "learning_rate": 2.6269064339689986e-06, + "loss": 0.10669133663177491, + "step": 29270 + }, + { + "epoch": 0.7712798358923026, + "grad_norm": 1.8295000791549683, + "learning_rate": 2.6211457042263157e-06, + "loss": 0.10960543155670166, + "step": 29280 + }, + { + "epoch": 0.7715432511368014, + "grad_norm": 2.4679512977600098, + "learning_rate": 2.6153903451321384e-06, + "loss": 0.10737919807434082, + "step": 29290 + }, + { + "epoch": 0.7718066663813001, + "grad_norm": 2.6014797687530518, + "learning_rate": 2.6096403608754685e-06, + "loss": 0.09928510189056397, + "step": 29300 + }, + { + "epoch": 0.7720700816257989, + "grad_norm": 2.6088900566101074, + "learning_rate": 2.603895755641387e-06, + "loss": 0.12148603200912475, + "step": 29310 + }, + { + "epoch": 0.7723334968702976, + "grad_norm": 1.6765676736831665, + "learning_rate": 2.598156533611077e-06, + "loss": 0.10209417343139648, + "step": 29320 + }, + { + "epoch": 0.7725969121147963, + "grad_norm": 3.7853498458862305, + "learning_rate": 2.592422698961781e-06, + "loss": 0.10178289413452149, + "step": 29330 + }, + { + "epoch": 0.7728603273592951, + "grad_norm": 2.369936466217041, + "learning_rate": 2.586694255866831e-06, + "loss": 0.12237434387207032, + "step": 29340 + }, + { + "epoch": 0.7731237426037938, + "grad_norm": 2.714228391647339, + "learning_rate": 2.580971208495642e-06, + "loss": 0.10207786560058593, + "step": 29350 + }, + { + "epoch": 0.7733871578482926, + "grad_norm": 2.8468620777130127, + "learning_rate": 2.575253561013692e-06, + "loss": 0.10504772663116455, + "step": 29360 + }, + { + "epoch": 0.7736505730927913, + "grad_norm": 2.0115020275115967, + "learning_rate": 2.56954131758253e-06, + "loss": 0.09764404296875, + "step": 29370 + }, + { + "epoch": 0.7739139883372901, + "grad_norm": 3.0523557662963867, + "learning_rate": 2.5638344823597762e-06, + "loss": 0.10703903436660767, + "step": 29380 + }, + { + "epoch": 0.7741774035817888, + "grad_norm": 2.0612897872924805, + "learning_rate": 2.558133059499105e-06, + "loss": 0.11590051651000977, + "step": 29390 + }, + { + "epoch": 0.7744408188262876, + "grad_norm": 2.5620453357696533, + "learning_rate": 2.552437053150266e-06, + "loss": 0.10244138240814209, + "step": 29400 + }, + { + "epoch": 0.7747042340707863, + "grad_norm": 1.6147369146347046, + "learning_rate": 2.546746467459058e-06, + "loss": 0.10191892385482788, + "step": 29410 + }, + { + "epoch": 0.774967649315285, + "grad_norm": 3.798386812210083, + "learning_rate": 2.541061306567325e-06, + "loss": 0.11903448104858398, + "step": 29420 + }, + { + "epoch": 0.7752310645597837, + "grad_norm": 2.514331102371216, + "learning_rate": 2.5353815746129827e-06, + "loss": 0.10510845184326172, + "step": 29430 + }, + { + "epoch": 0.7754944798042824, + "grad_norm": 2.7960243225097656, + "learning_rate": 2.529707275729979e-06, + "loss": 0.10595155954360962, + "step": 29440 + }, + { + "epoch": 0.7757578950487812, + "grad_norm": 2.961278200149536, + "learning_rate": 2.524038414048321e-06, + "loss": 0.1189367413520813, + "step": 29450 + }, + { + "epoch": 0.7760213102932799, + "grad_norm": 2.4825856685638428, + "learning_rate": 2.518374993694043e-06, + "loss": 0.10330641269683838, + "step": 29460 + }, + { + "epoch": 0.7762847255377787, + "grad_norm": 2.3879129886627197, + "learning_rate": 2.5127170187892257e-06, + "loss": 0.10337074995040893, + "step": 29470 + }, + { + "epoch": 0.7765481407822774, + "grad_norm": 2.7880403995513916, + "learning_rate": 2.507064493451995e-06, + "loss": 0.10043458938598633, + "step": 29480 + }, + { + "epoch": 0.7768115560267762, + "grad_norm": 2.605165719985962, + "learning_rate": 2.501417421796498e-06, + "loss": 0.11003365516662597, + "step": 29490 + }, + { + "epoch": 0.7770749712712749, + "grad_norm": 2.500290870666504, + "learning_rate": 2.4957758079329173e-06, + "loss": 0.10914342403411866, + "step": 29500 + }, + { + "epoch": 0.7773383865157736, + "grad_norm": 2.0862910747528076, + "learning_rate": 2.490139655967465e-06, + "loss": 0.1104726791381836, + "step": 29510 + }, + { + "epoch": 0.7776018017602724, + "grad_norm": 1.5494083166122437, + "learning_rate": 2.4845089700023694e-06, + "loss": 0.10943143367767334, + "step": 29520 + }, + { + "epoch": 0.7778652170047711, + "grad_norm": 1.9544038772583008, + "learning_rate": 2.4788837541358957e-06, + "loss": 0.10035377740859985, + "step": 29530 + }, + { + "epoch": 0.7781286322492699, + "grad_norm": 2.4971771240234375, + "learning_rate": 2.4732640124623152e-06, + "loss": 0.10469467639923095, + "step": 29540 + }, + { + "epoch": 0.7783920474937686, + "grad_norm": 2.099591016769409, + "learning_rate": 2.467649749071913e-06, + "loss": 0.10801444053649903, + "step": 29550 + }, + { + "epoch": 0.7786554627382674, + "grad_norm": 2.2934656143188477, + "learning_rate": 2.4620409680509993e-06, + "loss": 0.1071326494216919, + "step": 29560 + }, + { + "epoch": 0.778918877982766, + "grad_norm": 2.2565231323242188, + "learning_rate": 2.4564376734818818e-06, + "loss": 0.11529922485351562, + "step": 29570 + }, + { + "epoch": 0.7791822932272648, + "grad_norm": 1.5265114307403564, + "learning_rate": 2.450839869442887e-06, + "loss": 0.09918317794799805, + "step": 29580 + }, + { + "epoch": 0.7794457084717635, + "grad_norm": 2.9246344566345215, + "learning_rate": 2.4452475600083323e-06, + "loss": 0.09692617654800414, + "step": 29590 + }, + { + "epoch": 0.7797091237162622, + "grad_norm": 2.3693196773529053, + "learning_rate": 2.4396607492485393e-06, + "loss": 0.10343327522277831, + "step": 29600 + }, + { + "epoch": 0.779972538960761, + "grad_norm": 2.380563735961914, + "learning_rate": 2.4340794412298364e-06, + "loss": 0.11702461242675781, + "step": 29610 + }, + { + "epoch": 0.7802359542052597, + "grad_norm": 2.8338441848754883, + "learning_rate": 2.428503640014539e-06, + "loss": 0.10326707363128662, + "step": 29620 + }, + { + "epoch": 0.7804993694497585, + "grad_norm": 1.6100231409072876, + "learning_rate": 2.4229333496609532e-06, + "loss": 0.1095805048942566, + "step": 29630 + }, + { + "epoch": 0.7807627846942572, + "grad_norm": 2.8282220363616943, + "learning_rate": 2.4173685742233787e-06, + "loss": 0.10760390758514404, + "step": 29640 + }, + { + "epoch": 0.781026199938756, + "grad_norm": 2.5959534645080566, + "learning_rate": 2.411809317752095e-06, + "loss": 0.1050339937210083, + "step": 29650 + }, + { + "epoch": 0.7812896151832547, + "grad_norm": 2.366518020629883, + "learning_rate": 2.4062555842933753e-06, + "loss": 0.10169177055358887, + "step": 29660 + }, + { + "epoch": 0.7815530304277535, + "grad_norm": 2.252856731414795, + "learning_rate": 2.4007073778894665e-06, + "loss": 0.09772607088088989, + "step": 29670 + }, + { + "epoch": 0.7818164456722522, + "grad_norm": 1.9755494594573975, + "learning_rate": 2.395164702578583e-06, + "loss": 0.10486779212951661, + "step": 29680 + }, + { + "epoch": 0.7820798609167509, + "grad_norm": 2.7908689975738525, + "learning_rate": 2.3896275623949327e-06, + "loss": 0.10790781974792481, + "step": 29690 + }, + { + "epoch": 0.7823432761612497, + "grad_norm": 2.547146797180176, + "learning_rate": 2.38409596136868e-06, + "loss": 0.10579084157943726, + "step": 29700 + }, + { + "epoch": 0.7826066914057483, + "grad_norm": 1.9605520963668823, + "learning_rate": 2.378569903525969e-06, + "loss": 0.1096808910369873, + "step": 29710 + }, + { + "epoch": 0.7828701066502471, + "grad_norm": 3.482095718383789, + "learning_rate": 2.3730493928888963e-06, + "loss": 0.11458050012588501, + "step": 29720 + }, + { + "epoch": 0.7831335218947458, + "grad_norm": 2.5684244632720947, + "learning_rate": 2.367534433475527e-06, + "loss": 0.10702211856842041, + "step": 29730 + }, + { + "epoch": 0.7833969371392446, + "grad_norm": 2.5373597145080566, + "learning_rate": 2.3620250292998925e-06, + "loss": 0.11295833587646484, + "step": 29740 + }, + { + "epoch": 0.7836603523837433, + "grad_norm": 2.0560262203216553, + "learning_rate": 2.356521184371973e-06, + "loss": 0.10672816038131713, + "step": 29750 + }, + { + "epoch": 0.7839237676282421, + "grad_norm": 1.926749587059021, + "learning_rate": 2.3510229026977018e-06, + "loss": 0.10053611993789673, + "step": 29760 + }, + { + "epoch": 0.7841871828727408, + "grad_norm": 2.5205535888671875, + "learning_rate": 2.345530188278967e-06, + "loss": 0.10578505992889405, + "step": 29770 + }, + { + "epoch": 0.7844505981172395, + "grad_norm": 2.9257490634918213, + "learning_rate": 2.3400430451136e-06, + "loss": 0.10366299152374267, + "step": 29780 + }, + { + "epoch": 0.7847140133617383, + "grad_norm": 3.890284776687622, + "learning_rate": 2.334561477195387e-06, + "loss": 0.09329728484153747, + "step": 29790 + }, + { + "epoch": 0.784977428606237, + "grad_norm": 2.233510732650757, + "learning_rate": 2.329085488514049e-06, + "loss": 0.10152137279510498, + "step": 29800 + }, + { + "epoch": 0.7852408438507358, + "grad_norm": 2.5453286170959473, + "learning_rate": 2.3236150830552374e-06, + "loss": 0.09723150730133057, + "step": 29810 + }, + { + "epoch": 0.7855042590952345, + "grad_norm": 3.1797115802764893, + "learning_rate": 2.3181502648005593e-06, + "loss": 0.10419720411300659, + "step": 29820 + }, + { + "epoch": 0.7857676743397333, + "grad_norm": 2.6128158569335938, + "learning_rate": 2.3126910377275415e-06, + "loss": 0.09732639789581299, + "step": 29830 + }, + { + "epoch": 0.786031089584232, + "grad_norm": 2.6414105892181396, + "learning_rate": 2.3072374058096447e-06, + "loss": 0.1206304669380188, + "step": 29840 + }, + { + "epoch": 0.7862945048287308, + "grad_norm": 2.37416410446167, + "learning_rate": 2.301789373016259e-06, + "loss": 0.10089867115020752, + "step": 29850 + }, + { + "epoch": 0.7865579200732294, + "grad_norm": 2.854691505432129, + "learning_rate": 2.296346943312693e-06, + "loss": 0.10735208988189697, + "step": 29860 + }, + { + "epoch": 0.7868213353177281, + "grad_norm": 2.828103542327881, + "learning_rate": 2.290910120660188e-06, + "loss": 0.1117136001586914, + "step": 29870 + }, + { + "epoch": 0.7870847505622269, + "grad_norm": 2.6025240421295166, + "learning_rate": 2.285478909015896e-06, + "loss": 0.09624631404876709, + "step": 29880 + }, + { + "epoch": 0.7873481658067256, + "grad_norm": 2.4076333045959473, + "learning_rate": 2.2800533123328874e-06, + "loss": 0.12056015729904175, + "step": 29890 + }, + { + "epoch": 0.7876115810512244, + "grad_norm": 2.583597183227539, + "learning_rate": 2.2746333345601435e-06, + "loss": 0.10620609521865845, + "step": 29900 + }, + { + "epoch": 0.7878749962957231, + "grad_norm": 2.083920478820801, + "learning_rate": 2.269218979642558e-06, + "loss": 0.10186402797698975, + "step": 29910 + }, + { + "epoch": 0.7881384115402219, + "grad_norm": 2.469878673553467, + "learning_rate": 2.2638102515209382e-06, + "loss": 0.10707249641418456, + "step": 29920 + }, + { + "epoch": 0.7884018267847206, + "grad_norm": 3.0258519649505615, + "learning_rate": 2.258407154131982e-06, + "loss": 0.10077005624771118, + "step": 29930 + }, + { + "epoch": 0.7886652420292193, + "grad_norm": 2.8931632041931152, + "learning_rate": 2.2530096914082965e-06, + "loss": 0.1181487798690796, + "step": 29940 + }, + { + "epoch": 0.7889286572737181, + "grad_norm": 1.846309781074524, + "learning_rate": 2.2476178672783943e-06, + "loss": 0.10267398357391358, + "step": 29950 + }, + { + "epoch": 0.7891920725182168, + "grad_norm": 2.8318498134613037, + "learning_rate": 2.242231685666675e-06, + "loss": 0.10139364004135132, + "step": 29960 + }, + { + "epoch": 0.7894554877627156, + "grad_norm": 3.277092695236206, + "learning_rate": 2.2368511504934333e-06, + "loss": 0.10618481636047364, + "step": 29970 + }, + { + "epoch": 0.7897189030072143, + "grad_norm": 2.1572771072387695, + "learning_rate": 2.2314762656748544e-06, + "loss": 0.09985011219978332, + "step": 29980 + }, + { + "epoch": 0.7899823182517131, + "grad_norm": 2.130484104156494, + "learning_rate": 2.2261070351230107e-06, + "loss": 0.10213338136672974, + "step": 29990 + }, + { + "epoch": 0.7902457334962117, + "grad_norm": 2.4754655361175537, + "learning_rate": 2.2207434627458626e-06, + "loss": 0.11357920169830323, + "step": 30000 + }, + { + "epoch": 0.7905091487407105, + "grad_norm": 2.5307085514068604, + "learning_rate": 2.2153855524472513e-06, + "loss": 0.11849948167800903, + "step": 30010 + }, + { + "epoch": 0.7907725639852092, + "grad_norm": 1.966380000114441, + "learning_rate": 2.2100333081268853e-06, + "loss": 0.09856631755828857, + "step": 30020 + }, + { + "epoch": 0.7910359792297079, + "grad_norm": 2.6938114166259766, + "learning_rate": 2.204686733680369e-06, + "loss": 0.10647413730621338, + "step": 30030 + }, + { + "epoch": 0.7912993944742067, + "grad_norm": 2.3374381065368652, + "learning_rate": 2.1993458329991635e-06, + "loss": 0.1041305661201477, + "step": 30040 + }, + { + "epoch": 0.7915628097187054, + "grad_norm": 1.7037140130996704, + "learning_rate": 2.1940106099706137e-06, + "loss": 0.10678884983062745, + "step": 30050 + }, + { + "epoch": 0.7918262249632042, + "grad_norm": 1.69705069065094, + "learning_rate": 2.1886810684779204e-06, + "loss": 0.10515414476394654, + "step": 30060 + }, + { + "epoch": 0.7920896402077029, + "grad_norm": 2.4213860034942627, + "learning_rate": 2.18335721240015e-06, + "loss": 0.10291564464569092, + "step": 30070 + }, + { + "epoch": 0.7923530554522017, + "grad_norm": 2.9970948696136475, + "learning_rate": 2.1780390456122424e-06, + "loss": 0.0909004807472229, + "step": 30080 + }, + { + "epoch": 0.7926164706967004, + "grad_norm": 2.569342851638794, + "learning_rate": 2.1727265719849853e-06, + "loss": 0.10564754009246827, + "step": 30090 + }, + { + "epoch": 0.7928798859411992, + "grad_norm": 1.882119059562683, + "learning_rate": 2.1674197953850274e-06, + "loss": 0.09796984195709228, + "step": 30100 + }, + { + "epoch": 0.7931433011856979, + "grad_norm": 2.7177345752716064, + "learning_rate": 2.162118719674867e-06, + "loss": 0.11103873252868653, + "step": 30110 + }, + { + "epoch": 0.7934067164301966, + "grad_norm": 2.7934441566467285, + "learning_rate": 2.1568233487128556e-06, + "loss": 0.09664603471755981, + "step": 30120 + }, + { + "epoch": 0.7936701316746954, + "grad_norm": 3.138277292251587, + "learning_rate": 2.1515336863531976e-06, + "loss": 0.11434612274169922, + "step": 30130 + }, + { + "epoch": 0.793933546919194, + "grad_norm": 2.781381368637085, + "learning_rate": 2.1462497364459356e-06, + "loss": 0.10660459995269775, + "step": 30140 + }, + { + "epoch": 0.7941969621636928, + "grad_norm": 2.122896194458008, + "learning_rate": 2.1409715028369504e-06, + "loss": 0.09622383117675781, + "step": 30150 + }, + { + "epoch": 0.7944603774081915, + "grad_norm": 2.1920950412750244, + "learning_rate": 2.135698989367976e-06, + "loss": 0.10186517238616943, + "step": 30160 + }, + { + "epoch": 0.7947237926526903, + "grad_norm": 3.591695547103882, + "learning_rate": 2.1304321998765685e-06, + "loss": 0.0964763343334198, + "step": 30170 + }, + { + "epoch": 0.794987207897189, + "grad_norm": 2.504103660583496, + "learning_rate": 2.1251711381961336e-06, + "loss": 0.11194210052490235, + "step": 30180 + }, + { + "epoch": 0.7952506231416878, + "grad_norm": 3.861851930618286, + "learning_rate": 2.11991580815589e-06, + "loss": 0.10469574928283691, + "step": 30190 + }, + { + "epoch": 0.7955140383861865, + "grad_norm": 2.454982042312622, + "learning_rate": 2.1146662135808948e-06, + "loss": 0.10376131534576416, + "step": 30200 + }, + { + "epoch": 0.7957774536306852, + "grad_norm": 2.77408504486084, + "learning_rate": 2.1094223582920346e-06, + "loss": 0.10155937671661378, + "step": 30210 + }, + { + "epoch": 0.796040868875184, + "grad_norm": 2.772961139678955, + "learning_rate": 2.1041842461060103e-06, + "loss": 0.10219388008117676, + "step": 30220 + }, + { + "epoch": 0.7963042841196827, + "grad_norm": 2.2180521488189697, + "learning_rate": 2.098951880835346e-06, + "loss": 0.10754778385162353, + "step": 30230 + }, + { + "epoch": 0.7965676993641815, + "grad_norm": 2.5286777019500732, + "learning_rate": 2.093725266288382e-06, + "loss": 0.10817101001739501, + "step": 30240 + }, + { + "epoch": 0.7968311146086802, + "grad_norm": 2.591078996658325, + "learning_rate": 2.088504406269274e-06, + "loss": 0.10906763076782226, + "step": 30250 + }, + { + "epoch": 0.797094529853179, + "grad_norm": 2.8958966732025146, + "learning_rate": 2.083289304577992e-06, + "loss": 0.10168473720550537, + "step": 30260 + }, + { + "epoch": 0.7973579450976777, + "grad_norm": 2.9004645347595215, + "learning_rate": 2.0780799650103123e-06, + "loss": 0.09553328752517701, + "step": 30270 + }, + { + "epoch": 0.7976213603421765, + "grad_norm": 2.617055892944336, + "learning_rate": 2.0728763913578097e-06, + "loss": 0.10431647300720215, + "step": 30280 + }, + { + "epoch": 0.7978847755866751, + "grad_norm": 2.5605013370513916, + "learning_rate": 2.0676785874078773e-06, + "loss": 0.09678224325180054, + "step": 30290 + }, + { + "epoch": 0.7981481908311738, + "grad_norm": 1.9871913194656372, + "learning_rate": 2.062486556943699e-06, + "loss": 0.10465483665466309, + "step": 30300 + }, + { + "epoch": 0.7984116060756726, + "grad_norm": 2.688141345977783, + "learning_rate": 2.0573003037442575e-06, + "loss": 0.11760351657867432, + "step": 30310 + }, + { + "epoch": 0.7986750213201713, + "grad_norm": 2.454638957977295, + "learning_rate": 2.0521198315843315e-06, + "loss": 0.09818471670150757, + "step": 30320 + }, + { + "epoch": 0.7989384365646701, + "grad_norm": 2.2396464347839355, + "learning_rate": 2.046945144234491e-06, + "loss": 0.10850400924682617, + "step": 30330 + }, + { + "epoch": 0.7992018518091688, + "grad_norm": 2.0988404750823975, + "learning_rate": 2.0417762454611012e-06, + "loss": 0.11646751165390015, + "step": 30340 + }, + { + "epoch": 0.7994652670536676, + "grad_norm": 2.2964046001434326, + "learning_rate": 2.0366131390263055e-06, + "loss": 0.10442326068878174, + "step": 30350 + }, + { + "epoch": 0.7997286822981663, + "grad_norm": 2.6679303646087646, + "learning_rate": 2.0314558286880383e-06, + "loss": 0.10766558647155762, + "step": 30360 + }, + { + "epoch": 0.7999920975426651, + "grad_norm": 2.8615047931671143, + "learning_rate": 2.026304318200011e-06, + "loss": 0.11062862873077392, + "step": 30370 + }, + { + "epoch": 0.8002555127871638, + "grad_norm": 2.0368452072143555, + "learning_rate": 2.021158611311712e-06, + "loss": 0.09943974018096924, + "step": 30380 + }, + { + "epoch": 0.8005189280316625, + "grad_norm": 1.842964768409729, + "learning_rate": 2.016018711768416e-06, + "loss": 0.11500153541564942, + "step": 30390 + }, + { + "epoch": 0.8007823432761613, + "grad_norm": 2.379216432571411, + "learning_rate": 2.010884623311162e-06, + "loss": 0.10982255935668946, + "step": 30400 + }, + { + "epoch": 0.80104575852066, + "grad_norm": 2.9312610626220703, + "learning_rate": 2.005756349676753e-06, + "loss": 0.11181745529174805, + "step": 30410 + }, + { + "epoch": 0.8013091737651588, + "grad_norm": 1.951024055480957, + "learning_rate": 2.0006338945977777e-06, + "loss": 0.09987835884094239, + "step": 30420 + }, + { + "epoch": 0.8015725890096574, + "grad_norm": 2.3941848278045654, + "learning_rate": 1.9955172618025743e-06, + "loss": 0.09316412210464478, + "step": 30430 + }, + { + "epoch": 0.8018360042541562, + "grad_norm": 2.617851495742798, + "learning_rate": 1.9904064550152515e-06, + "loss": 0.08970715999603271, + "step": 30440 + }, + { + "epoch": 0.8020994194986549, + "grad_norm": 2.1766350269317627, + "learning_rate": 1.9853014779556734e-06, + "loss": 0.0981758177280426, + "step": 30450 + }, + { + "epoch": 0.8023628347431536, + "grad_norm": 2.289214849472046, + "learning_rate": 1.9802023343394606e-06, + "loss": 0.09463894367218018, + "step": 30460 + }, + { + "epoch": 0.8026262499876524, + "grad_norm": 3.0735182762145996, + "learning_rate": 1.975109027877995e-06, + "loss": 0.11400902271270752, + "step": 30470 + }, + { + "epoch": 0.8028896652321511, + "grad_norm": 2.8219165802001953, + "learning_rate": 1.9700215622784026e-06, + "loss": 0.08814204335212708, + "step": 30480 + }, + { + "epoch": 0.8031530804766499, + "grad_norm": 4.343008995056152, + "learning_rate": 1.96493994124356e-06, + "loss": 0.10833534002304077, + "step": 30490 + }, + { + "epoch": 0.8034164957211486, + "grad_norm": 2.0072598457336426, + "learning_rate": 1.9598641684720897e-06, + "loss": 0.09561420679092407, + "step": 30500 + }, + { + "epoch": 0.8036799109656474, + "grad_norm": 1.8328014612197876, + "learning_rate": 1.954794247658356e-06, + "loss": 0.1171347975730896, + "step": 30510 + }, + { + "epoch": 0.8039433262101461, + "grad_norm": 2.0299172401428223, + "learning_rate": 1.949730182492473e-06, + "loss": 0.09234933257102966, + "step": 30520 + }, + { + "epoch": 0.8042067414546449, + "grad_norm": 2.5840299129486084, + "learning_rate": 1.9446719766602793e-06, + "loss": 0.09388642311096192, + "step": 30530 + }, + { + "epoch": 0.8044701566991436, + "grad_norm": 4.075571537017822, + "learning_rate": 1.9396196338433547e-06, + "loss": 0.10177830457687378, + "step": 30540 + }, + { + "epoch": 0.8047335719436423, + "grad_norm": 2.667022705078125, + "learning_rate": 1.9345731577190187e-06, + "loss": 0.09689171314239502, + "step": 30550 + }, + { + "epoch": 0.804996987188141, + "grad_norm": 3.121891975402832, + "learning_rate": 1.929532551960309e-06, + "loss": 0.10689686536788941, + "step": 30560 + }, + { + "epoch": 0.8052604024326397, + "grad_norm": 2.2691843509674072, + "learning_rate": 1.9244978202359974e-06, + "loss": 0.08815123438835144, + "step": 30570 + }, + { + "epoch": 0.8055238176771385, + "grad_norm": 1.9257245063781738, + "learning_rate": 1.9194689662105803e-06, + "loss": 0.09292150735855102, + "step": 30580 + }, + { + "epoch": 0.8057872329216372, + "grad_norm": 2.6829171180725098, + "learning_rate": 1.9144459935442705e-06, + "loss": 0.10261683464050293, + "step": 30590 + }, + { + "epoch": 0.806050648166136, + "grad_norm": 2.127746820449829, + "learning_rate": 1.9094289058930095e-06, + "loss": 0.09911723136901855, + "step": 30600 + }, + { + "epoch": 0.8063140634106347, + "grad_norm": 2.1037678718566895, + "learning_rate": 1.9044177069084502e-06, + "loss": 0.09517464637756348, + "step": 30610 + }, + { + "epoch": 0.8065774786551335, + "grad_norm": 2.23226261138916, + "learning_rate": 1.899412400237951e-06, + "loss": 0.09779354333877563, + "step": 30620 + }, + { + "epoch": 0.8068408938996322, + "grad_norm": 2.8154726028442383, + "learning_rate": 1.8944129895246e-06, + "loss": 0.10118200778961181, + "step": 30630 + }, + { + "epoch": 0.8071043091441309, + "grad_norm": 2.604771137237549, + "learning_rate": 1.8894194784071761e-06, + "loss": 0.10235118865966797, + "step": 30640 + }, + { + "epoch": 0.8073677243886297, + "grad_norm": 2.595907688140869, + "learning_rate": 1.8844318705201815e-06, + "loss": 0.1025397777557373, + "step": 30650 + }, + { + "epoch": 0.8076311396331284, + "grad_norm": 2.9412050247192383, + "learning_rate": 1.879450169493805e-06, + "loss": 0.10422976016998291, + "step": 30660 + }, + { + "epoch": 0.8078945548776272, + "grad_norm": 2.640108346939087, + "learning_rate": 1.874474378953942e-06, + "loss": 0.11244480609893799, + "step": 30670 + }, + { + "epoch": 0.8081579701221259, + "grad_norm": 3.6098430156707764, + "learning_rate": 1.8695045025221936e-06, + "loss": 0.09203046560287476, + "step": 30680 + }, + { + "epoch": 0.8084213853666247, + "grad_norm": 2.220684766769409, + "learning_rate": 1.864540543815848e-06, + "loss": 0.0908686637878418, + "step": 30690 + }, + { + "epoch": 0.8086848006111234, + "grad_norm": 3.4180355072021484, + "learning_rate": 1.859582506447889e-06, + "loss": 0.10386002063751221, + "step": 30700 + }, + { + "epoch": 0.8089482158556222, + "grad_norm": 2.6170005798339844, + "learning_rate": 1.8546303940269883e-06, + "loss": 0.09624334573745727, + "step": 30710 + }, + { + "epoch": 0.8092116311001208, + "grad_norm": 3.1627533435821533, + "learning_rate": 1.8496842101575053e-06, + "loss": 0.09810062646865844, + "step": 30720 + }, + { + "epoch": 0.8094750463446195, + "grad_norm": 2.7935574054718018, + "learning_rate": 1.844743958439491e-06, + "loss": 0.11098016500473022, + "step": 30730 + }, + { + "epoch": 0.8097384615891183, + "grad_norm": 2.9776036739349365, + "learning_rate": 1.839809642468674e-06, + "loss": 0.10253689289093018, + "step": 30740 + }, + { + "epoch": 0.810001876833617, + "grad_norm": 1.7603973150253296, + "learning_rate": 1.8348812658364557e-06, + "loss": 0.10446474552154542, + "step": 30750 + }, + { + "epoch": 0.8102652920781158, + "grad_norm": 2.1191413402557373, + "learning_rate": 1.8299588321299266e-06, + "loss": 0.09071592092514039, + "step": 30760 + }, + { + "epoch": 0.8105287073226145, + "grad_norm": 2.240161657333374, + "learning_rate": 1.8250423449318434e-06, + "loss": 0.09549994468688965, + "step": 30770 + }, + { + "epoch": 0.8107921225671133, + "grad_norm": 2.141131639480591, + "learning_rate": 1.8201318078206443e-06, + "loss": 0.1090738296508789, + "step": 30780 + }, + { + "epoch": 0.811055537811612, + "grad_norm": 2.535543203353882, + "learning_rate": 1.8152272243704216e-06, + "loss": 0.11368459463119507, + "step": 30790 + }, + { + "epoch": 0.8113189530561108, + "grad_norm": 2.7985916137695312, + "learning_rate": 1.8103285981509432e-06, + "loss": 0.10812461376190186, + "step": 30800 + }, + { + "epoch": 0.8115823683006095, + "grad_norm": 2.118370771408081, + "learning_rate": 1.8054359327276473e-06, + "loss": 0.10020149946212768, + "step": 30810 + }, + { + "epoch": 0.8118457835451082, + "grad_norm": 2.0047829151153564, + "learning_rate": 1.8005492316616224e-06, + "loss": 0.10143258571624755, + "step": 30820 + }, + { + "epoch": 0.812109198789607, + "grad_norm": 2.5310420989990234, + "learning_rate": 1.7956684985096217e-06, + "loss": 0.09013472199440002, + "step": 30830 + }, + { + "epoch": 0.8123726140341057, + "grad_norm": 2.307105541229248, + "learning_rate": 1.7907937368240525e-06, + "loss": 0.09439090490341187, + "step": 30840 + }, + { + "epoch": 0.8126360292786045, + "grad_norm": 3.4833290576934814, + "learning_rate": 1.7859249501529742e-06, + "loss": 0.1038731575012207, + "step": 30850 + }, + { + "epoch": 0.8128994445231031, + "grad_norm": 2.838317632675171, + "learning_rate": 1.7810621420401054e-06, + "loss": 0.09495738744735718, + "step": 30860 + }, + { + "epoch": 0.8131628597676019, + "grad_norm": 2.5386040210723877, + "learning_rate": 1.776205316024806e-06, + "loss": 0.09882755875587464, + "step": 30870 + }, + { + "epoch": 0.8134262750121006, + "grad_norm": 1.904213786125183, + "learning_rate": 1.7713544756420785e-06, + "loss": 0.08724591135978699, + "step": 30880 + }, + { + "epoch": 0.8136896902565994, + "grad_norm": 2.895726442337036, + "learning_rate": 1.76650962442258e-06, + "loss": 0.10342975854873657, + "step": 30890 + }, + { + "epoch": 0.8139531055010981, + "grad_norm": 3.7152669429779053, + "learning_rate": 1.761670765892599e-06, + "loss": 0.10221347808837891, + "step": 30900 + }, + { + "epoch": 0.8142165207455968, + "grad_norm": 2.3250396251678467, + "learning_rate": 1.7568379035740669e-06, + "loss": 0.10407196283340454, + "step": 30910 + }, + { + "epoch": 0.8144799359900956, + "grad_norm": 2.7358028888702393, + "learning_rate": 1.7520110409845493e-06, + "loss": 0.09351401329040528, + "step": 30920 + }, + { + "epoch": 0.8147433512345943, + "grad_norm": 4.105398654937744, + "learning_rate": 1.7471901816372416e-06, + "loss": 0.1023678183555603, + "step": 30930 + }, + { + "epoch": 0.8150067664790931, + "grad_norm": 2.669689893722534, + "learning_rate": 1.7423753290409784e-06, + "loss": 0.09631807804107666, + "step": 30940 + }, + { + "epoch": 0.8152701817235918, + "grad_norm": 3.552760124206543, + "learning_rate": 1.7375664867002151e-06, + "loss": 0.0958160102367401, + "step": 30950 + }, + { + "epoch": 0.8155335969680906, + "grad_norm": 1.7587215900421143, + "learning_rate": 1.7327636581150343e-06, + "loss": 0.10062299966812134, + "step": 30960 + }, + { + "epoch": 0.8157970122125893, + "grad_norm": 2.98443341255188, + "learning_rate": 1.7279668467811427e-06, + "loss": 0.09323989748954772, + "step": 30970 + }, + { + "epoch": 0.816060427457088, + "grad_norm": 2.8802919387817383, + "learning_rate": 1.7231760561898625e-06, + "loss": 0.10445778369903565, + "step": 30980 + }, + { + "epoch": 0.8163238427015868, + "grad_norm": 2.4013547897338867, + "learning_rate": 1.718391289828144e-06, + "loss": 0.09353861808776856, + "step": 30990 + }, + { + "epoch": 0.8165872579460854, + "grad_norm": 2.4476892948150635, + "learning_rate": 1.7136125511785462e-06, + "loss": 0.09303069710731507, + "step": 31000 + }, + { + "epoch": 0.8168506731905842, + "grad_norm": 2.1959261894226074, + "learning_rate": 1.7088398437192344e-06, + "loss": 0.10058534145355225, + "step": 31010 + }, + { + "epoch": 0.8171140884350829, + "grad_norm": 2.140393018722534, + "learning_rate": 1.7040731709239967e-06, + "loss": 0.09071755409240723, + "step": 31020 + }, + { + "epoch": 0.8173775036795817, + "grad_norm": 2.865306854248047, + "learning_rate": 1.6993125362622231e-06, + "loss": 0.10071582794189453, + "step": 31030 + }, + { + "epoch": 0.8176409189240804, + "grad_norm": 2.8918306827545166, + "learning_rate": 1.6945579431989067e-06, + "loss": 0.09408020973205566, + "step": 31040 + }, + { + "epoch": 0.8179043341685792, + "grad_norm": 2.6730973720550537, + "learning_rate": 1.689809395194647e-06, + "loss": 0.0944307267665863, + "step": 31050 + }, + { + "epoch": 0.8181677494130779, + "grad_norm": 3.3615620136260986, + "learning_rate": 1.6850668957056393e-06, + "loss": 0.1079858422279358, + "step": 31060 + }, + { + "epoch": 0.8184311646575766, + "grad_norm": 3.0385358333587646, + "learning_rate": 1.6803304481836835e-06, + "loss": 0.0967553734779358, + "step": 31070 + }, + { + "epoch": 0.8186945799020754, + "grad_norm": 3.5228161811828613, + "learning_rate": 1.6756000560761687e-06, + "loss": 0.10038232803344727, + "step": 31080 + }, + { + "epoch": 0.8189579951465741, + "grad_norm": 2.4686086177825928, + "learning_rate": 1.6708757228260786e-06, + "loss": 0.09443095922470093, + "step": 31090 + }, + { + "epoch": 0.8192214103910729, + "grad_norm": 2.2701773643493652, + "learning_rate": 1.6661574518719859e-06, + "loss": 0.08886488676071166, + "step": 31100 + }, + { + "epoch": 0.8194848256355716, + "grad_norm": 3.2575011253356934, + "learning_rate": 1.6614452466480502e-06, + "loss": 0.09249666929244996, + "step": 31110 + }, + { + "epoch": 0.8197482408800704, + "grad_norm": 2.291822671890259, + "learning_rate": 1.6567391105840247e-06, + "loss": 0.08911730647087097, + "step": 31120 + }, + { + "epoch": 0.820011656124569, + "grad_norm": 2.264866352081299, + "learning_rate": 1.6520390471052306e-06, + "loss": 0.09007474780082703, + "step": 31130 + }, + { + "epoch": 0.8202750713690679, + "grad_norm": 2.013707399368286, + "learning_rate": 1.647345059632578e-06, + "loss": 0.08621320724487305, + "step": 31140 + }, + { + "epoch": 0.8205384866135665, + "grad_norm": 3.0965676307678223, + "learning_rate": 1.6426571515825563e-06, + "loss": 0.10268726348876953, + "step": 31150 + }, + { + "epoch": 0.8208019018580652, + "grad_norm": 2.7091152667999268, + "learning_rate": 1.6379753263672272e-06, + "loss": 0.0956764578819275, + "step": 31160 + }, + { + "epoch": 0.821065317102564, + "grad_norm": 3.0782246589660645, + "learning_rate": 1.6332995873942237e-06, + "loss": 0.09544395208358765, + "step": 31170 + }, + { + "epoch": 0.8213287323470627, + "grad_norm": 2.6849801540374756, + "learning_rate": 1.6286299380667504e-06, + "loss": 0.09766727685928345, + "step": 31180 + }, + { + "epoch": 0.8215921475915615, + "grad_norm": 2.600214719772339, + "learning_rate": 1.6239663817835783e-06, + "loss": 0.0954663634300232, + "step": 31190 + }, + { + "epoch": 0.8218555628360602, + "grad_norm": 3.208660364151001, + "learning_rate": 1.6193089219390478e-06, + "loss": 0.09592362642288207, + "step": 31200 + }, + { + "epoch": 0.822118978080559, + "grad_norm": 2.3347742557525635, + "learning_rate": 1.6146575619230597e-06, + "loss": 0.10723881721496582, + "step": 31210 + }, + { + "epoch": 0.8223823933250577, + "grad_norm": 2.26339054107666, + "learning_rate": 1.610012305121068e-06, + "loss": 0.09247217178344727, + "step": 31220 + }, + { + "epoch": 0.8226458085695565, + "grad_norm": 2.5863518714904785, + "learning_rate": 1.6053731549140982e-06, + "loss": 0.10398697853088379, + "step": 31230 + }, + { + "epoch": 0.8229092238140552, + "grad_norm": 2.6512680053710938, + "learning_rate": 1.6007401146787195e-06, + "loss": 0.09604309797286988, + "step": 31240 + }, + { + "epoch": 0.8231726390585539, + "grad_norm": 2.5529515743255615, + "learning_rate": 1.596113187787064e-06, + "loss": 0.0910347044467926, + "step": 31250 + }, + { + "epoch": 0.8234360543030527, + "grad_norm": 3.039975881576538, + "learning_rate": 1.5914923776068025e-06, + "loss": 0.10282801389694214, + "step": 31260 + }, + { + "epoch": 0.8236994695475514, + "grad_norm": 2.440168619155884, + "learning_rate": 1.5868776875011604e-06, + "loss": 0.08939815163612366, + "step": 31270 + }, + { + "epoch": 0.8239628847920502, + "grad_norm": 3.9061203002929688, + "learning_rate": 1.582269120828912e-06, + "loss": 0.10450479984283448, + "step": 31280 + }, + { + "epoch": 0.8242263000365488, + "grad_norm": 2.2245635986328125, + "learning_rate": 1.5776666809443697e-06, + "loss": 0.0855478286743164, + "step": 31290 + }, + { + "epoch": 0.8244897152810476, + "grad_norm": 2.376559257507324, + "learning_rate": 1.5730703711973883e-06, + "loss": 0.09390913844108581, + "step": 31300 + }, + { + "epoch": 0.8247531305255463, + "grad_norm": 2.30993914604187, + "learning_rate": 1.5684801949333584e-06, + "loss": 0.09362049102783203, + "step": 31310 + }, + { + "epoch": 0.8250165457700451, + "grad_norm": 2.3145241737365723, + "learning_rate": 1.5638961554932064e-06, + "loss": 0.11074192523956299, + "step": 31320 + }, + { + "epoch": 0.8252799610145438, + "grad_norm": 2.749227285385132, + "learning_rate": 1.5593182562134002e-06, + "loss": 0.09878183603286743, + "step": 31330 + }, + { + "epoch": 0.8255433762590425, + "grad_norm": 2.600839138031006, + "learning_rate": 1.554746500425931e-06, + "loss": 0.09051791429519654, + "step": 31340 + }, + { + "epoch": 0.8258067915035413, + "grad_norm": 2.0401172637939453, + "learning_rate": 1.5501808914583139e-06, + "loss": 0.09087826609611512, + "step": 31350 + }, + { + "epoch": 0.82607020674804, + "grad_norm": 3.238875389099121, + "learning_rate": 1.5456214326336017e-06, + "loss": 0.0973188579082489, + "step": 31360 + }, + { + "epoch": 0.8263336219925388, + "grad_norm": 2.1365721225738525, + "learning_rate": 1.5410681272703643e-06, + "loss": 0.08399013280868531, + "step": 31370 + }, + { + "epoch": 0.8265970372370375, + "grad_norm": 3.1137588024139404, + "learning_rate": 1.536520978682693e-06, + "loss": 0.09276201128959656, + "step": 31380 + }, + { + "epoch": 0.8268604524815363, + "grad_norm": 1.9693509340286255, + "learning_rate": 1.531979990180199e-06, + "loss": 0.08514498472213745, + "step": 31390 + }, + { + "epoch": 0.827123867726035, + "grad_norm": 2.268947124481201, + "learning_rate": 1.5274451650680066e-06, + "loss": 0.11317533254623413, + "step": 31400 + }, + { + "epoch": 0.8273872829705338, + "grad_norm": 3.210847854614258, + "learning_rate": 1.5229165066467632e-06, + "loss": 0.09838460683822632, + "step": 31410 + }, + { + "epoch": 0.8276506982150325, + "grad_norm": 2.0385725498199463, + "learning_rate": 1.5183940182126189e-06, + "loss": 0.09104366302490234, + "step": 31420 + }, + { + "epoch": 0.8279141134595311, + "grad_norm": 2.1743690967559814, + "learning_rate": 1.513877703057235e-06, + "loss": 0.0891028881072998, + "step": 31430 + }, + { + "epoch": 0.8281775287040299, + "grad_norm": 2.3527987003326416, + "learning_rate": 1.509367564467782e-06, + "loss": 0.09143124222755432, + "step": 31440 + }, + { + "epoch": 0.8284409439485286, + "grad_norm": 2.8060455322265625, + "learning_rate": 1.5048636057269294e-06, + "loss": 0.10030558109283447, + "step": 31450 + }, + { + "epoch": 0.8287043591930274, + "grad_norm": 2.5397398471832275, + "learning_rate": 1.50036583011286e-06, + "loss": 0.10158647298812866, + "step": 31460 + }, + { + "epoch": 0.8289677744375261, + "grad_norm": 2.0586929321289062, + "learning_rate": 1.4958742408992454e-06, + "loss": 0.09609794616699219, + "step": 31470 + }, + { + "epoch": 0.8292311896820249, + "grad_norm": 2.16595458984375, + "learning_rate": 1.4913888413552536e-06, + "loss": 0.089745032787323, + "step": 31480 + }, + { + "epoch": 0.8294946049265236, + "grad_norm": 4.28193998336792, + "learning_rate": 1.486909634745557e-06, + "loss": 0.09334198832511902, + "step": 31490 + }, + { + "epoch": 0.8297580201710223, + "grad_norm": 2.167095422744751, + "learning_rate": 1.4824366243303156e-06, + "loss": 0.09402472972869873, + "step": 31500 + }, + { + "epoch": 0.8300214354155211, + "grad_norm": 2.390947103500366, + "learning_rate": 1.4779698133651766e-06, + "loss": 0.09824639558792114, + "step": 31510 + }, + { + "epoch": 0.8302848506600198, + "grad_norm": 2.878258466720581, + "learning_rate": 1.47350920510128e-06, + "loss": 0.09834996461868287, + "step": 31520 + }, + { + "epoch": 0.8305482659045186, + "grad_norm": 2.8324320316314697, + "learning_rate": 1.469054802785246e-06, + "loss": 0.10469850301742553, + "step": 31530 + }, + { + "epoch": 0.8308116811490173, + "grad_norm": 3.454210042953491, + "learning_rate": 1.4646066096591838e-06, + "loss": 0.09929529428482056, + "step": 31540 + }, + { + "epoch": 0.8310750963935161, + "grad_norm": 2.6556901931762695, + "learning_rate": 1.4601646289606809e-06, + "loss": 0.09962403178215026, + "step": 31550 + }, + { + "epoch": 0.8313385116380148, + "grad_norm": 2.743283748626709, + "learning_rate": 1.4557288639227995e-06, + "loss": 0.08729730248451233, + "step": 31560 + }, + { + "epoch": 0.8316019268825136, + "grad_norm": 2.6119890213012695, + "learning_rate": 1.451299317774083e-06, + "loss": 0.08525025248527526, + "step": 31570 + }, + { + "epoch": 0.8318653421270122, + "grad_norm": 2.2103848457336426, + "learning_rate": 1.4468759937385423e-06, + "loss": 0.08679014444351196, + "step": 31580 + }, + { + "epoch": 0.8321287573715109, + "grad_norm": 2.4755454063415527, + "learning_rate": 1.4424588950356688e-06, + "loss": 0.09784281849861146, + "step": 31590 + }, + { + "epoch": 0.8323921726160097, + "grad_norm": 2.3507957458496094, + "learning_rate": 1.4380480248804185e-06, + "loss": 0.11400649547576905, + "step": 31600 + }, + { + "epoch": 0.8326555878605084, + "grad_norm": 3.210017681121826, + "learning_rate": 1.4336433864832033e-06, + "loss": 0.10196816921234131, + "step": 31610 + }, + { + "epoch": 0.8329190031050072, + "grad_norm": 3.008943796157837, + "learning_rate": 1.4292449830499189e-06, + "loss": 0.09726625680923462, + "step": 31620 + }, + { + "epoch": 0.8331824183495059, + "grad_norm": 2.2935378551483154, + "learning_rate": 1.4248528177819098e-06, + "loss": 0.08794136643409729, + "step": 31630 + }, + { + "epoch": 0.8334458335940047, + "grad_norm": 4.242031574249268, + "learning_rate": 1.4204668938759835e-06, + "loss": 0.09324506521224976, + "step": 31640 + }, + { + "epoch": 0.8337092488385034, + "grad_norm": 3.3895344734191895, + "learning_rate": 1.416087214524403e-06, + "loss": 0.09843721389770507, + "step": 31650 + }, + { + "epoch": 0.8339726640830022, + "grad_norm": 2.7560200691223145, + "learning_rate": 1.4117137829148885e-06, + "loss": 0.09861487150192261, + "step": 31660 + }, + { + "epoch": 0.8342360793275009, + "grad_norm": 2.0826268196105957, + "learning_rate": 1.4073466022306148e-06, + "loss": 0.08571858406066894, + "step": 31670 + }, + { + "epoch": 0.8344994945719996, + "grad_norm": 2.9610722064971924, + "learning_rate": 1.4029856756502026e-06, + "loss": 0.1021299958229065, + "step": 31680 + }, + { + "epoch": 0.8347629098164984, + "grad_norm": 2.82110857963562, + "learning_rate": 1.3986310063477215e-06, + "loss": 0.09528750777244568, + "step": 31690 + }, + { + "epoch": 0.835026325060997, + "grad_norm": 2.8091628551483154, + "learning_rate": 1.3942825974926888e-06, + "loss": 0.09438509941101074, + "step": 31700 + }, + { + "epoch": 0.8352897403054959, + "grad_norm": 2.924680471420288, + "learning_rate": 1.3899404522500626e-06, + "loss": 0.09681832194328308, + "step": 31710 + }, + { + "epoch": 0.8355531555499945, + "grad_norm": 2.8224925994873047, + "learning_rate": 1.385604573780248e-06, + "loss": 0.0863483190536499, + "step": 31720 + }, + { + "epoch": 0.8358165707944933, + "grad_norm": 2.6975252628326416, + "learning_rate": 1.38127496523908e-06, + "loss": 0.11061984300613403, + "step": 31730 + }, + { + "epoch": 0.836079986038992, + "grad_norm": 2.30928897857666, + "learning_rate": 1.376951629777833e-06, + "loss": 0.09769997596740723, + "step": 31740 + }, + { + "epoch": 0.8363434012834908, + "grad_norm": 2.696685314178467, + "learning_rate": 1.372634570543222e-06, + "loss": 0.09412550926208496, + "step": 31750 + }, + { + "epoch": 0.8366068165279895, + "grad_norm": 1.9905240535736084, + "learning_rate": 1.3683237906773873e-06, + "loss": 0.09072552919387818, + "step": 31760 + }, + { + "epoch": 0.8368702317724882, + "grad_norm": 2.3340466022491455, + "learning_rate": 1.3640192933178997e-06, + "loss": 0.09157328605651856, + "step": 31770 + }, + { + "epoch": 0.837133647016987, + "grad_norm": 2.916508436203003, + "learning_rate": 1.359721081597758e-06, + "loss": 0.1039898157119751, + "step": 31780 + }, + { + "epoch": 0.8373970622614857, + "grad_norm": 2.76381516456604, + "learning_rate": 1.355429158645384e-06, + "loss": 0.09036235809326172, + "step": 31790 + }, + { + "epoch": 0.8376604775059845, + "grad_norm": 2.403029441833496, + "learning_rate": 1.3511435275846297e-06, + "loss": 0.10470151901245117, + "step": 31800 + }, + { + "epoch": 0.8379238927504832, + "grad_norm": 4.835330963134766, + "learning_rate": 1.3468641915347613e-06, + "loss": 0.09940251111984252, + "step": 31810 + }, + { + "epoch": 0.838187307994982, + "grad_norm": 3.256408452987671, + "learning_rate": 1.3425911536104573e-06, + "loss": 0.09098913073539734, + "step": 31820 + }, + { + "epoch": 0.8384507232394807, + "grad_norm": 2.816784381866455, + "learning_rate": 1.3383244169218268e-06, + "loss": 0.08938727974891662, + "step": 31830 + }, + { + "epoch": 0.8387141384839795, + "grad_norm": 3.305464029312134, + "learning_rate": 1.3340639845743797e-06, + "loss": 0.10417000055313111, + "step": 31840 + }, + { + "epoch": 0.8389775537284782, + "grad_norm": 3.3853840827941895, + "learning_rate": 1.32980985966905e-06, + "loss": 0.11112897396087647, + "step": 31850 + }, + { + "epoch": 0.8392409689729768, + "grad_norm": 2.9715147018432617, + "learning_rate": 1.3255620453021656e-06, + "loss": 0.09381628632545472, + "step": 31860 + }, + { + "epoch": 0.8395043842174756, + "grad_norm": 3.718665599822998, + "learning_rate": 1.3213205445654688e-06, + "loss": 0.08886138200759888, + "step": 31870 + }, + { + "epoch": 0.8397677994619743, + "grad_norm": 2.914806842803955, + "learning_rate": 1.3170853605461132e-06, + "loss": 0.09915677309036255, + "step": 31880 + }, + { + "epoch": 0.8400312147064731, + "grad_norm": 3.7859818935394287, + "learning_rate": 1.3128564963266444e-06, + "loss": 0.09422934055328369, + "step": 31890 + }, + { + "epoch": 0.8402946299509718, + "grad_norm": 3.6316444873809814, + "learning_rate": 1.3086339549850125e-06, + "loss": 0.10054954290390014, + "step": 31900 + }, + { + "epoch": 0.8405580451954706, + "grad_norm": 2.801081895828247, + "learning_rate": 1.3044177395945657e-06, + "loss": 0.08703817129135132, + "step": 31910 + }, + { + "epoch": 0.8408214604399693, + "grad_norm": 2.2162201404571533, + "learning_rate": 1.3002078532240447e-06, + "loss": 0.09831899404525757, + "step": 31920 + }, + { + "epoch": 0.8410848756844681, + "grad_norm": 2.748303174972534, + "learning_rate": 1.296004298937592e-06, + "loss": 0.09348570108413697, + "step": 31930 + }, + { + "epoch": 0.8413482909289668, + "grad_norm": 3.640702962875366, + "learning_rate": 1.2918070797947346e-06, + "loss": 0.07695497274398803, + "step": 31940 + }, + { + "epoch": 0.8416117061734655, + "grad_norm": 2.473499298095703, + "learning_rate": 1.2876161988503843e-06, + "loss": 0.10556498765945435, + "step": 31950 + }, + { + "epoch": 0.8418751214179643, + "grad_norm": 3.0333046913146973, + "learning_rate": 1.2834316591548502e-06, + "loss": 0.09029422402381897, + "step": 31960 + }, + { + "epoch": 0.842138536662463, + "grad_norm": 3.1886541843414307, + "learning_rate": 1.279253463753819e-06, + "loss": 0.10032658576965332, + "step": 31970 + }, + { + "epoch": 0.8424019519069618, + "grad_norm": 3.0496695041656494, + "learning_rate": 1.2750816156883627e-06, + "loss": 0.09770067930221557, + "step": 31980 + }, + { + "epoch": 0.8426653671514605, + "grad_norm": 3.0296425819396973, + "learning_rate": 1.2709161179949303e-06, + "loss": 0.09172361493110656, + "step": 31990 + }, + { + "epoch": 0.8429287823959593, + "grad_norm": 2.842625379562378, + "learning_rate": 1.266756973705351e-06, + "loss": 0.09949113130569458, + "step": 32000 + }, + { + "epoch": 0.8431921976404579, + "grad_norm": 2.0945377349853516, + "learning_rate": 1.2626041858468318e-06, + "loss": 0.09936209321022034, + "step": 32010 + }, + { + "epoch": 0.8434556128849566, + "grad_norm": 2.577378511428833, + "learning_rate": 1.2584577574419499e-06, + "loss": 0.08917063474655151, + "step": 32020 + }, + { + "epoch": 0.8437190281294554, + "grad_norm": 3.115636110305786, + "learning_rate": 1.2543176915086542e-06, + "loss": 0.10362435579299926, + "step": 32030 + }, + { + "epoch": 0.8439824433739541, + "grad_norm": 3.783900499343872, + "learning_rate": 1.2501839910602632e-06, + "loss": 0.08882284164428711, + "step": 32040 + }, + { + "epoch": 0.8442458586184529, + "grad_norm": 3.2886297702789307, + "learning_rate": 1.24605665910546e-06, + "loss": 0.10188018083572388, + "step": 32050 + }, + { + "epoch": 0.8445092738629516, + "grad_norm": 2.823124408721924, + "learning_rate": 1.2419356986483e-06, + "loss": 0.08267722129821778, + "step": 32060 + }, + { + "epoch": 0.8447726891074504, + "grad_norm": 3.082852602005005, + "learning_rate": 1.2378211126881956e-06, + "loss": 0.09565993547439575, + "step": 32070 + }, + { + "epoch": 0.8450361043519491, + "grad_norm": 2.6502888202667236, + "learning_rate": 1.2337129042199147e-06, + "loss": 0.08680728673934937, + "step": 32080 + }, + { + "epoch": 0.8452995195964479, + "grad_norm": 3.1734061241149902, + "learning_rate": 1.2296110762335933e-06, + "loss": 0.08536245822906494, + "step": 32090 + }, + { + "epoch": 0.8455629348409466, + "grad_norm": 3.0250675678253174, + "learning_rate": 1.2255156317147165e-06, + "loss": 0.09021503925323486, + "step": 32100 + }, + { + "epoch": 0.8458263500854453, + "grad_norm": 2.707343816757202, + "learning_rate": 1.2214265736441277e-06, + "loss": 0.09431515336036682, + "step": 32110 + }, + { + "epoch": 0.8460897653299441, + "grad_norm": 3.4488039016723633, + "learning_rate": 1.2173439049980174e-06, + "loss": 0.09072630405426026, + "step": 32120 + }, + { + "epoch": 0.8463531805744428, + "grad_norm": 2.5344271659851074, + "learning_rate": 1.2132676287479273e-06, + "loss": 0.08708333969116211, + "step": 32130 + }, + { + "epoch": 0.8466165958189416, + "grad_norm": 3.3699326515197754, + "learning_rate": 1.209197747860752e-06, + "loss": 0.09022250175476074, + "step": 32140 + }, + { + "epoch": 0.8468800110634402, + "grad_norm": 2.8691623210906982, + "learning_rate": 1.2051342652987218e-06, + "loss": 0.08761894702911377, + "step": 32150 + }, + { + "epoch": 0.847143426307939, + "grad_norm": 3.1099801063537598, + "learning_rate": 1.2010771840194168e-06, + "loss": 0.093716561794281, + "step": 32160 + }, + { + "epoch": 0.8474068415524377, + "grad_norm": 2.7673745155334473, + "learning_rate": 1.1970265069757548e-06, + "loss": 0.09899194836616516, + "step": 32170 + }, + { + "epoch": 0.8476702567969365, + "grad_norm": 3.4169809818267822, + "learning_rate": 1.1929822371159915e-06, + "loss": 0.09449652433395386, + "step": 32180 + }, + { + "epoch": 0.8479336720414352, + "grad_norm": 1.9965338706970215, + "learning_rate": 1.188944377383725e-06, + "loss": 0.09456847310066223, + "step": 32190 + }, + { + "epoch": 0.8481970872859339, + "grad_norm": 1.8978941440582275, + "learning_rate": 1.1849129307178841e-06, + "loss": 0.0848076581954956, + "step": 32200 + }, + { + "epoch": 0.8484605025304327, + "grad_norm": 2.4781811237335205, + "learning_rate": 1.180887900052724e-06, + "loss": 0.09457845091819764, + "step": 32210 + }, + { + "epoch": 0.8487239177749314, + "grad_norm": 3.950805187225342, + "learning_rate": 1.17686928831784e-06, + "loss": 0.09850364923477173, + "step": 32220 + }, + { + "epoch": 0.8489873330194302, + "grad_norm": 2.6489500999450684, + "learning_rate": 1.1728570984381505e-06, + "loss": 0.10502493381500244, + "step": 32230 + }, + { + "epoch": 0.8492507482639289, + "grad_norm": 3.1179251670837402, + "learning_rate": 1.1688513333339003e-06, + "loss": 0.1087291955947876, + "step": 32240 + }, + { + "epoch": 0.8495141635084277, + "grad_norm": 2.4884374141693115, + "learning_rate": 1.1648519959206584e-06, + "loss": 0.08456183671951294, + "step": 32250 + }, + { + "epoch": 0.8497775787529264, + "grad_norm": 3.393989086151123, + "learning_rate": 1.1608590891093118e-06, + "loss": 0.08505107760429383, + "step": 32260 + }, + { + "epoch": 0.8500409939974252, + "grad_norm": 2.86142635345459, + "learning_rate": 1.156872615806075e-06, + "loss": 0.09721573591232299, + "step": 32270 + }, + { + "epoch": 0.8503044092419239, + "grad_norm": 2.310731887817383, + "learning_rate": 1.1528925789124734e-06, + "loss": 0.09403331875801087, + "step": 32280 + }, + { + "epoch": 0.8505678244864225, + "grad_norm": 2.5843143463134766, + "learning_rate": 1.1489189813253499e-06, + "loss": 0.09738047122955322, + "step": 32290 + }, + { + "epoch": 0.8508312397309213, + "grad_norm": 2.4077913761138916, + "learning_rate": 1.1449518259368596e-06, + "loss": 0.08665544986724853, + "step": 32300 + }, + { + "epoch": 0.85109465497542, + "grad_norm": 2.749081611633301, + "learning_rate": 1.1409911156344677e-06, + "loss": 0.08973259925842285, + "step": 32310 + }, + { + "epoch": 0.8513580702199188, + "grad_norm": 3.078742742538452, + "learning_rate": 1.1370368533009557e-06, + "loss": 0.08851167559623718, + "step": 32320 + }, + { + "epoch": 0.8516214854644175, + "grad_norm": 2.1707944869995117, + "learning_rate": 1.1330890418144014e-06, + "loss": 0.08186575174331664, + "step": 32330 + }, + { + "epoch": 0.8518849007089163, + "grad_norm": 3.904430389404297, + "learning_rate": 1.1291476840481908e-06, + "loss": 0.09806355237960815, + "step": 32340 + }, + { + "epoch": 0.852148315953415, + "grad_norm": 2.5323638916015625, + "learning_rate": 1.125212782871019e-06, + "loss": 0.09636491537094116, + "step": 32350 + }, + { + "epoch": 0.8524117311979138, + "grad_norm": 3.0442373752593994, + "learning_rate": 1.1212843411468755e-06, + "loss": 0.08716462850570679, + "step": 32360 + }, + { + "epoch": 0.8526751464424125, + "grad_norm": 3.1411843299865723, + "learning_rate": 1.1173623617350472e-06, + "loss": 0.08374248147010803, + "step": 32370 + }, + { + "epoch": 0.8529385616869112, + "grad_norm": 2.8382465839385986, + "learning_rate": 1.1134468474901227e-06, + "loss": 0.08676354885101319, + "step": 32380 + }, + { + "epoch": 0.85320197693141, + "grad_norm": 3.314818859100342, + "learning_rate": 1.1095378012619796e-06, + "loss": 0.08865727186203003, + "step": 32390 + }, + { + "epoch": 0.8534653921759087, + "grad_norm": 2.4749529361724854, + "learning_rate": 1.1056352258957924e-06, + "loss": 0.09147441983222962, + "step": 32400 + }, + { + "epoch": 0.8537288074204075, + "grad_norm": 3.2287988662719727, + "learning_rate": 1.1017391242320263e-06, + "loss": 0.08807107210159301, + "step": 32410 + }, + { + "epoch": 0.8539922226649062, + "grad_norm": 2.5333008766174316, + "learning_rate": 1.0978494991064248e-06, + "loss": 0.09830349683761597, + "step": 32420 + }, + { + "epoch": 0.854255637909405, + "grad_norm": 2.6093642711639404, + "learning_rate": 1.0939663533500311e-06, + "loss": 0.08493547439575196, + "step": 32430 + }, + { + "epoch": 0.8545190531539036, + "grad_norm": 2.188157320022583, + "learning_rate": 1.0900896897891634e-06, + "loss": 0.09279501438140869, + "step": 32440 + }, + { + "epoch": 0.8547824683984024, + "grad_norm": 3.1662585735321045, + "learning_rate": 1.0862195112454289e-06, + "loss": 0.09418617486953736, + "step": 32450 + }, + { + "epoch": 0.8550458836429011, + "grad_norm": 2.855330228805542, + "learning_rate": 1.0823558205357054e-06, + "loss": 0.08274784088134765, + "step": 32460 + }, + { + "epoch": 0.8553092988873998, + "grad_norm": 3.831960439682007, + "learning_rate": 1.0784986204721537e-06, + "loss": 0.09601414799690247, + "step": 32470 + }, + { + "epoch": 0.8555727141318986, + "grad_norm": 4.508949279785156, + "learning_rate": 1.0746479138622134e-06, + "loss": 0.09537210464477539, + "step": 32480 + }, + { + "epoch": 0.8558361293763973, + "grad_norm": 3.281848669052124, + "learning_rate": 1.070803703508595e-06, + "loss": 0.09963751435279847, + "step": 32490 + }, + { + "epoch": 0.8560995446208961, + "grad_norm": 2.8651018142700195, + "learning_rate": 1.066965992209279e-06, + "loss": 0.08906693458557129, + "step": 32500 + }, + { + "epoch": 0.8563629598653948, + "grad_norm": 3.749058723449707, + "learning_rate": 1.0631347827575179e-06, + "loss": 0.09104868173599243, + "step": 32510 + }, + { + "epoch": 0.8566263751098936, + "grad_norm": 3.639958381652832, + "learning_rate": 1.0593100779418286e-06, + "loss": 0.08675020933151245, + "step": 32520 + }, + { + "epoch": 0.8568897903543923, + "grad_norm": 3.2243337631225586, + "learning_rate": 1.0554918805460013e-06, + "loss": 0.1055908441543579, + "step": 32530 + }, + { + "epoch": 0.8571532055988911, + "grad_norm": 3.267773151397705, + "learning_rate": 1.051680193349084e-06, + "loss": 0.09441837072372436, + "step": 32540 + }, + { + "epoch": 0.8574166208433898, + "grad_norm": 2.9819767475128174, + "learning_rate": 1.0478750191253817e-06, + "loss": 0.092238849401474, + "step": 32550 + }, + { + "epoch": 0.8576800360878885, + "grad_norm": 3.1032159328460693, + "learning_rate": 1.0440763606444694e-06, + "loss": 0.08441115021705628, + "step": 32560 + }, + { + "epoch": 0.8579434513323873, + "grad_norm": 2.8293964862823486, + "learning_rate": 1.0402842206711728e-06, + "loss": 0.10488336086273194, + "step": 32570 + }, + { + "epoch": 0.8582068665768859, + "grad_norm": 4.440855503082275, + "learning_rate": 1.0364986019655765e-06, + "loss": 0.09008986949920654, + "step": 32580 + }, + { + "epoch": 0.8584702818213847, + "grad_norm": 3.1748995780944824, + "learning_rate": 1.0327195072830153e-06, + "loss": 0.09772027134895325, + "step": 32590 + }, + { + "epoch": 0.8587336970658834, + "grad_norm": 2.7506790161132812, + "learning_rate": 1.028946939374076e-06, + "loss": 0.10051932334899902, + "step": 32600 + }, + { + "epoch": 0.8589971123103822, + "grad_norm": 3.026794910430908, + "learning_rate": 1.0251809009846003e-06, + "loss": 0.08333194851875306, + "step": 32610 + }, + { + "epoch": 0.8592605275548809, + "grad_norm": 2.3127853870391846, + "learning_rate": 1.0214213948556718e-06, + "loss": 0.08777538537979127, + "step": 32620 + }, + { + "epoch": 0.8595239427993796, + "grad_norm": 2.835169553756714, + "learning_rate": 1.0176684237236212e-06, + "loss": 0.09629465341567993, + "step": 32630 + }, + { + "epoch": 0.8597873580438784, + "grad_norm": 2.6187071800231934, + "learning_rate": 1.0139219903200227e-06, + "loss": 0.09569077491760254, + "step": 32640 + }, + { + "epoch": 0.8600507732883771, + "grad_norm": 3.189018964767456, + "learning_rate": 1.0101820973716891e-06, + "loss": 0.10341213941574097, + "step": 32650 + }, + { + "epoch": 0.8603141885328759, + "grad_norm": 3.280672788619995, + "learning_rate": 1.0064487476006812e-06, + "loss": 0.09243636131286621, + "step": 32660 + }, + { + "epoch": 0.8605776037773746, + "grad_norm": 3.0809671878814697, + "learning_rate": 1.0027219437242907e-06, + "loss": 0.090659499168396, + "step": 32670 + }, + { + "epoch": 0.8608410190218734, + "grad_norm": 3.083965301513672, + "learning_rate": 9.990016884550425e-07, + "loss": 0.09528309106826782, + "step": 32680 + }, + { + "epoch": 0.8611044342663721, + "grad_norm": 2.9919588565826416, + "learning_rate": 9.952879845007024e-07, + "loss": 0.08630480766296386, + "step": 32690 + }, + { + "epoch": 0.8613678495108709, + "grad_norm": 4.974435329437256, + "learning_rate": 9.915808345642642e-07, + "loss": 0.08518406152725219, + "step": 32700 + }, + { + "epoch": 0.8616312647553696, + "grad_norm": 3.2687153816223145, + "learning_rate": 9.878802413439514e-07, + "loss": 0.09389680624008179, + "step": 32710 + }, + { + "epoch": 0.8618946799998682, + "grad_norm": 3.269413471221924, + "learning_rate": 9.841862075332143e-07, + "loss": 0.08683829307556153, + "step": 32720 + }, + { + "epoch": 0.862158095244367, + "grad_norm": 2.1130051612854004, + "learning_rate": 9.804987358207274e-07, + "loss": 0.08717564344406128, + "step": 32730 + }, + { + "epoch": 0.8624215104888657, + "grad_norm": 2.2284178733825684, + "learning_rate": 9.768178288903985e-07, + "loss": 0.09053283929824829, + "step": 32740 + }, + { + "epoch": 0.8626849257333645, + "grad_norm": 2.964956760406494, + "learning_rate": 9.731434894213465e-07, + "loss": 0.0924189567565918, + "step": 32750 + }, + { + "epoch": 0.8629483409778632, + "grad_norm": 2.818732500076294, + "learning_rate": 9.694757200879146e-07, + "loss": 0.09887105226516724, + "step": 32760 + }, + { + "epoch": 0.863211756222362, + "grad_norm": 3.0482699871063232, + "learning_rate": 9.658145235596628e-07, + "loss": 0.0984571635723114, + "step": 32770 + }, + { + "epoch": 0.8634751714668607, + "grad_norm": 3.3049094676971436, + "learning_rate": 9.621599025013672e-07, + "loss": 0.09350149035453796, + "step": 32780 + }, + { + "epoch": 0.8637385867113595, + "grad_norm": 3.3450093269348145, + "learning_rate": 9.585118595730226e-07, + "loss": 0.09969736337661743, + "step": 32790 + }, + { + "epoch": 0.8640020019558582, + "grad_norm": 3.010702610015869, + "learning_rate": 9.548703974298311e-07, + "loss": 0.09053313732147217, + "step": 32800 + }, + { + "epoch": 0.8642654172003569, + "grad_norm": 4.267056941986084, + "learning_rate": 9.512355187222e-07, + "loss": 0.09062435030937195, + "step": 32810 + }, + { + "epoch": 0.8645288324448557, + "grad_norm": 2.522667169570923, + "learning_rate": 9.47607226095758e-07, + "loss": 0.09739378690719605, + "step": 32820 + }, + { + "epoch": 0.8647922476893544, + "grad_norm": 3.249929666519165, + "learning_rate": 9.439855221913297e-07, + "loss": 0.09295490980148316, + "step": 32830 + }, + { + "epoch": 0.8650556629338532, + "grad_norm": 3.04577374458313, + "learning_rate": 9.403704096449484e-07, + "loss": 0.089294695854187, + "step": 32840 + }, + { + "epoch": 0.8653190781783519, + "grad_norm": 3.546205520629883, + "learning_rate": 9.367618910878484e-07, + "loss": 0.09278282523155212, + "step": 32850 + }, + { + "epoch": 0.8655824934228507, + "grad_norm": 2.8939449787139893, + "learning_rate": 9.331599691464644e-07, + "loss": 0.0969402015209198, + "step": 32860 + }, + { + "epoch": 0.8658459086673493, + "grad_norm": 3.0040645599365234, + "learning_rate": 9.295646464424323e-07, + "loss": 0.0852716624736786, + "step": 32870 + }, + { + "epoch": 0.8661093239118481, + "grad_norm": 3.804758071899414, + "learning_rate": 9.259759255925827e-07, + "loss": 0.08663511276245117, + "step": 32880 + }, + { + "epoch": 0.8663727391563468, + "grad_norm": 3.1563785076141357, + "learning_rate": 9.223938092089413e-07, + "loss": 0.08772512674331664, + "step": 32890 + }, + { + "epoch": 0.8666361544008455, + "grad_norm": 2.911977529525757, + "learning_rate": 9.188182998987261e-07, + "loss": 0.0893083930015564, + "step": 32900 + }, + { + "epoch": 0.8668995696453443, + "grad_norm": 3.0248661041259766, + "learning_rate": 9.152494002643464e-07, + "loss": 0.09258517622947693, + "step": 32910 + }, + { + "epoch": 0.867162984889843, + "grad_norm": 2.71309757232666, + "learning_rate": 9.116871129034066e-07, + "loss": 0.08721237182617188, + "step": 32920 + }, + { + "epoch": 0.8674264001343418, + "grad_norm": 2.675579071044922, + "learning_rate": 9.081314404086872e-07, + "loss": 0.09345157146453857, + "step": 32930 + }, + { + "epoch": 0.8676898153788405, + "grad_norm": 3.364332914352417, + "learning_rate": 9.045823853681602e-07, + "loss": 0.09568181037902831, + "step": 32940 + }, + { + "epoch": 0.8679532306233393, + "grad_norm": 2.8271567821502686, + "learning_rate": 9.010399503649836e-07, + "loss": 0.09254084229469299, + "step": 32950 + }, + { + "epoch": 0.868216645867838, + "grad_norm": 3.780181884765625, + "learning_rate": 8.975041379774951e-07, + "loss": 0.11153990030288696, + "step": 32960 + }, + { + "epoch": 0.8684800611123368, + "grad_norm": 3.5503885746002197, + "learning_rate": 8.939749507792094e-07, + "loss": 0.08651825189590454, + "step": 32970 + }, + { + "epoch": 0.8687434763568355, + "grad_norm": 3.1480464935302734, + "learning_rate": 8.904523913388218e-07, + "loss": 0.08173489570617676, + "step": 32980 + }, + { + "epoch": 0.8690068916013342, + "grad_norm": 2.5804443359375, + "learning_rate": 8.869364622202015e-07, + "loss": 0.08510965704917908, + "step": 32990 + }, + { + "epoch": 0.869270306845833, + "grad_norm": 3.0227744579315186, + "learning_rate": 8.834271659823967e-07, + "loss": 0.09088825583457946, + "step": 33000 + }, + { + "epoch": 0.8695337220903316, + "grad_norm": 2.510936737060547, + "learning_rate": 8.799245051796246e-07, + "loss": 0.08831479549407958, + "step": 33010 + }, + { + "epoch": 0.8697971373348304, + "grad_norm": 2.319561004638672, + "learning_rate": 8.764284823612667e-07, + "loss": 0.0977380096912384, + "step": 33020 + }, + { + "epoch": 0.8700605525793291, + "grad_norm": 3.514066696166992, + "learning_rate": 8.729391000718867e-07, + "loss": 0.10509551763534546, + "step": 33030 + }, + { + "epoch": 0.8703239678238279, + "grad_norm": 2.6805834770202637, + "learning_rate": 8.694563608512041e-07, + "loss": 0.0954094648361206, + "step": 33040 + }, + { + "epoch": 0.8705873830683266, + "grad_norm": 2.7591593265533447, + "learning_rate": 8.659802672341078e-07, + "loss": 0.08769781589508056, + "step": 33050 + }, + { + "epoch": 0.8708507983128254, + "grad_norm": 2.450892686843872, + "learning_rate": 8.625108217506472e-07, + "loss": 0.08800445199012756, + "step": 33060 + }, + { + "epoch": 0.8711142135573241, + "grad_norm": 3.151442766189575, + "learning_rate": 8.590480269260337e-07, + "loss": 0.09832297563552857, + "step": 33070 + }, + { + "epoch": 0.8713776288018228, + "grad_norm": 3.1990106105804443, + "learning_rate": 8.555918852806421e-07, + "loss": 0.09540647864341736, + "step": 33080 + }, + { + "epoch": 0.8716410440463216, + "grad_norm": 2.726900339126587, + "learning_rate": 8.521423993300005e-07, + "loss": 0.09101966619491578, + "step": 33090 + }, + { + "epoch": 0.8719044592908203, + "grad_norm": 3.9439103603363037, + "learning_rate": 8.486995715847923e-07, + "loss": 0.0902621328830719, + "step": 33100 + }, + { + "epoch": 0.8721678745353191, + "grad_norm": 3.1467573642730713, + "learning_rate": 8.452634045508557e-07, + "loss": 0.08556361794471741, + "step": 33110 + }, + { + "epoch": 0.8724312897798178, + "grad_norm": 2.8918988704681396, + "learning_rate": 8.41833900729182e-07, + "loss": 0.08823853731155396, + "step": 33120 + }, + { + "epoch": 0.8726947050243166, + "grad_norm": 2.6564698219299316, + "learning_rate": 8.384110626159126e-07, + "loss": 0.09400023818016053, + "step": 33130 + }, + { + "epoch": 0.8729581202688153, + "grad_norm": 2.0437591075897217, + "learning_rate": 8.349948927023388e-07, + "loss": 0.08657661676406861, + "step": 33140 + }, + { + "epoch": 0.8732215355133139, + "grad_norm": 2.673616647720337, + "learning_rate": 8.315853934748908e-07, + "loss": 0.09517983198165894, + "step": 33150 + }, + { + "epoch": 0.8734849507578127, + "grad_norm": 2.9588677883148193, + "learning_rate": 8.281825674151533e-07, + "loss": 0.10126006603240967, + "step": 33160 + }, + { + "epoch": 0.8737483660023114, + "grad_norm": 2.237642526626587, + "learning_rate": 8.247864169998499e-07, + "loss": 0.08141086101531983, + "step": 33170 + }, + { + "epoch": 0.8740117812468102, + "grad_norm": 2.782707691192627, + "learning_rate": 8.213969447008441e-07, + "loss": 0.0846743106842041, + "step": 33180 + }, + { + "epoch": 0.8742751964913089, + "grad_norm": 2.940558910369873, + "learning_rate": 8.180141529851404e-07, + "loss": 0.09448177218437195, + "step": 33190 + }, + { + "epoch": 0.8745386117358077, + "grad_norm": 3.990419864654541, + "learning_rate": 8.146380443148794e-07, + "loss": 0.09198277592658996, + "step": 33200 + }, + { + "epoch": 0.8748020269803064, + "grad_norm": 4.355365753173828, + "learning_rate": 8.11268621147342e-07, + "loss": 0.09124784469604492, + "step": 33210 + }, + { + "epoch": 0.8750654422248052, + "grad_norm": 3.217268705368042, + "learning_rate": 8.079058859349364e-07, + "loss": 0.09819191694259644, + "step": 33220 + }, + { + "epoch": 0.8753288574693039, + "grad_norm": 2.503653049468994, + "learning_rate": 8.045498411252095e-07, + "loss": 0.09199391603469849, + "step": 33230 + }, + { + "epoch": 0.8755922727138026, + "grad_norm": 4.333323955535889, + "learning_rate": 8.012004891608327e-07, + "loss": 0.10191622972488404, + "step": 33240 + }, + { + "epoch": 0.8758556879583014, + "grad_norm": 3.788987874984741, + "learning_rate": 7.978578324796072e-07, + "loss": 0.09258716702461242, + "step": 33250 + }, + { + "epoch": 0.8761191032028001, + "grad_norm": 2.841064214706421, + "learning_rate": 7.945218735144689e-07, + "loss": 0.08670987486839295, + "step": 33260 + }, + { + "epoch": 0.8763825184472989, + "grad_norm": 3.063732385635376, + "learning_rate": 7.91192614693469e-07, + "loss": 0.08308436274528504, + "step": 33270 + }, + { + "epoch": 0.8766459336917976, + "grad_norm": 3.5911357402801514, + "learning_rate": 7.878700584397836e-07, + "loss": 0.09154269695281983, + "step": 33280 + }, + { + "epoch": 0.8769093489362964, + "grad_norm": 3.8533663749694824, + "learning_rate": 7.84554207171716e-07, + "loss": 0.08195815086364747, + "step": 33290 + }, + { + "epoch": 0.877172764180795, + "grad_norm": 3.3662235736846924, + "learning_rate": 7.812450633026847e-07, + "loss": 0.0779221534729004, + "step": 33300 + }, + { + "epoch": 0.8774361794252938, + "grad_norm": 3.3317434787750244, + "learning_rate": 7.779426292412273e-07, + "loss": 0.08401200771331788, + "step": 33310 + }, + { + "epoch": 0.8776995946697925, + "grad_norm": 2.5025198459625244, + "learning_rate": 7.746469073909979e-07, + "loss": 0.07913843393325806, + "step": 33320 + }, + { + "epoch": 0.8779630099142912, + "grad_norm": 2.581843137741089, + "learning_rate": 7.713579001507632e-07, + "loss": 0.09350779056549072, + "step": 33330 + }, + { + "epoch": 0.87822642515879, + "grad_norm": 3.079846143722534, + "learning_rate": 7.680756099144072e-07, + "loss": 0.08742387890815735, + "step": 33340 + }, + { + "epoch": 0.8784898404032887, + "grad_norm": 3.4663469791412354, + "learning_rate": 7.648000390709221e-07, + "loss": 0.10125529766082764, + "step": 33350 + }, + { + "epoch": 0.8787532556477875, + "grad_norm": 2.724555730819702, + "learning_rate": 7.615311900044087e-07, + "loss": 0.0801609992980957, + "step": 33360 + }, + { + "epoch": 0.8790166708922862, + "grad_norm": 3.8816425800323486, + "learning_rate": 7.582690650940771e-07, + "loss": 0.08759927749633789, + "step": 33370 + }, + { + "epoch": 0.879280086136785, + "grad_norm": 2.7854957580566406, + "learning_rate": 7.550136667142393e-07, + "loss": 0.08779460191726685, + "step": 33380 + }, + { + "epoch": 0.8795435013812837, + "grad_norm": 2.9395711421966553, + "learning_rate": 7.517649972343199e-07, + "loss": 0.07762956619262695, + "step": 33390 + }, + { + "epoch": 0.8798069166257825, + "grad_norm": 2.4419424533843994, + "learning_rate": 7.485230590188397e-07, + "loss": 0.08556549549102783, + "step": 33400 + }, + { + "epoch": 0.8800703318702812, + "grad_norm": 3.1790149211883545, + "learning_rate": 7.452878544274167e-07, + "loss": 0.09188237190246581, + "step": 33410 + }, + { + "epoch": 0.8803337471147799, + "grad_norm": 3.4276013374328613, + "learning_rate": 7.420593858147773e-07, + "loss": 0.1004730224609375, + "step": 33420 + }, + { + "epoch": 0.8805971623592787, + "grad_norm": 3.09948992729187, + "learning_rate": 7.388376555307386e-07, + "loss": 0.09586588144302369, + "step": 33430 + }, + { + "epoch": 0.8808605776037773, + "grad_norm": 3.011385440826416, + "learning_rate": 7.356226659202137e-07, + "loss": 0.09139069318771362, + "step": 33440 + }, + { + "epoch": 0.8811239928482761, + "grad_norm": 2.9434092044830322, + "learning_rate": 7.324144193232119e-07, + "loss": 0.08222222924232483, + "step": 33450 + }, + { + "epoch": 0.8813874080927748, + "grad_norm": 3.461017608642578, + "learning_rate": 7.29212918074833e-07, + "loss": 0.08140824437141418, + "step": 33460 + }, + { + "epoch": 0.8816508233372736, + "grad_norm": 3.4094762802124023, + "learning_rate": 7.260181645052689e-07, + "loss": 0.09209970235824586, + "step": 33470 + }, + { + "epoch": 0.8819142385817723, + "grad_norm": 2.956716775894165, + "learning_rate": 7.228301609397981e-07, + "loss": 0.0940753698348999, + "step": 33480 + }, + { + "epoch": 0.8821776538262711, + "grad_norm": 2.7445716857910156, + "learning_rate": 7.196489096987869e-07, + "loss": 0.09359192848205566, + "step": 33490 + }, + { + "epoch": 0.8824410690707698, + "grad_norm": 3.565687417984009, + "learning_rate": 7.164744130976886e-07, + "loss": 0.07051714658737182, + "step": 33500 + }, + { + "epoch": 0.8827044843152685, + "grad_norm": 2.9255950450897217, + "learning_rate": 7.133066734470351e-07, + "loss": 0.09146180152893066, + "step": 33510 + }, + { + "epoch": 0.8829678995597673, + "grad_norm": 3.3785321712493896, + "learning_rate": 7.101456930524509e-07, + "loss": 0.09216414093971252, + "step": 33520 + }, + { + "epoch": 0.883231314804266, + "grad_norm": 2.956625461578369, + "learning_rate": 7.069914742146267e-07, + "loss": 0.09294200539588929, + "step": 33530 + }, + { + "epoch": 0.8834947300487648, + "grad_norm": 3.3643321990966797, + "learning_rate": 7.038440192293416e-07, + "loss": 0.09469785690307617, + "step": 33540 + }, + { + "epoch": 0.8837581452932635, + "grad_norm": 2.3630645275115967, + "learning_rate": 7.007033303874511e-07, + "loss": 0.09044077396392822, + "step": 33550 + }, + { + "epoch": 0.8840215605377623, + "grad_norm": 2.3828125, + "learning_rate": 6.97569409974882e-07, + "loss": 0.0920715093612671, + "step": 33560 + }, + { + "epoch": 0.884284975782261, + "grad_norm": 2.940476655960083, + "learning_rate": 6.944422602726375e-07, + "loss": 0.08115366101264954, + "step": 33570 + }, + { + "epoch": 0.8845483910267598, + "grad_norm": 2.262655258178711, + "learning_rate": 6.913218835567925e-07, + "loss": 0.08868597149848938, + "step": 33580 + }, + { + "epoch": 0.8848118062712584, + "grad_norm": 3.8854804039001465, + "learning_rate": 6.882082820984892e-07, + "loss": 0.0899362564086914, + "step": 33590 + }, + { + "epoch": 0.8850752215157571, + "grad_norm": 2.4198811054229736, + "learning_rate": 6.851014581639437e-07, + "loss": 0.0861936092376709, + "step": 33600 + }, + { + "epoch": 0.8853386367602559, + "grad_norm": 2.940662145614624, + "learning_rate": 6.820014140144382e-07, + "loss": 0.08639075756072997, + "step": 33610 + }, + { + "epoch": 0.8856020520047546, + "grad_norm": 3.4123270511627197, + "learning_rate": 6.789081519063134e-07, + "loss": 0.0884084939956665, + "step": 33620 + }, + { + "epoch": 0.8858654672492534, + "grad_norm": 2.7637171745300293, + "learning_rate": 6.758216740909829e-07, + "loss": 0.09473817944526672, + "step": 33630 + }, + { + "epoch": 0.8861288824937521, + "grad_norm": 3.9326062202453613, + "learning_rate": 6.727419828149173e-07, + "loss": 0.09054059982299804, + "step": 33640 + }, + { + "epoch": 0.8863922977382509, + "grad_norm": 2.9211432933807373, + "learning_rate": 6.696690803196482e-07, + "loss": 0.08608226180076599, + "step": 33650 + }, + { + "epoch": 0.8866557129827496, + "grad_norm": 2.9137332439422607, + "learning_rate": 6.666029688417674e-07, + "loss": 0.09524034261703491, + "step": 33660 + }, + { + "epoch": 0.8869191282272483, + "grad_norm": 2.506002187728882, + "learning_rate": 6.635436506129223e-07, + "loss": 0.08343989849090576, + "step": 33670 + }, + { + "epoch": 0.8871825434717471, + "grad_norm": 3.4398488998413086, + "learning_rate": 6.60491127859818e-07, + "loss": 0.08801136016845704, + "step": 33680 + }, + { + "epoch": 0.8874459587162458, + "grad_norm": 3.086602210998535, + "learning_rate": 6.57445402804211e-07, + "loss": 0.08585988879203796, + "step": 33690 + }, + { + "epoch": 0.8877093739607446, + "grad_norm": 3.644207000732422, + "learning_rate": 6.544064776629123e-07, + "loss": 0.08669039011001586, + "step": 33700 + }, + { + "epoch": 0.8879727892052433, + "grad_norm": 3.130675792694092, + "learning_rate": 6.513743546477824e-07, + "loss": 0.08281627297401428, + "step": 33710 + }, + { + "epoch": 0.888236204449742, + "grad_norm": 3.5040981769561768, + "learning_rate": 6.483490359657296e-07, + "loss": 0.09753240942955017, + "step": 33720 + }, + { + "epoch": 0.8884996196942407, + "grad_norm": 2.509805202484131, + "learning_rate": 6.453305238187147e-07, + "loss": 0.07626189589500428, + "step": 33730 + }, + { + "epoch": 0.8887630349387395, + "grad_norm": 3.297980546951294, + "learning_rate": 6.423188204037412e-07, + "loss": 0.089473295211792, + "step": 33740 + }, + { + "epoch": 0.8890264501832382, + "grad_norm": 3.609055519104004, + "learning_rate": 6.393139279128513e-07, + "loss": 0.08234286308288574, + "step": 33750 + }, + { + "epoch": 0.8892898654277369, + "grad_norm": 3.1050009727478027, + "learning_rate": 6.3631584853314e-07, + "loss": 0.09177621603012084, + "step": 33760 + }, + { + "epoch": 0.8895532806722357, + "grad_norm": 4.912407398223877, + "learning_rate": 6.333245844467384e-07, + "loss": 0.08577954173088073, + "step": 33770 + }, + { + "epoch": 0.8898166959167344, + "grad_norm": 2.1349265575408936, + "learning_rate": 6.303401378308172e-07, + "loss": 0.08301413059234619, + "step": 33780 + }, + { + "epoch": 0.8900801111612332, + "grad_norm": 2.6511449813842773, + "learning_rate": 6.273625108575843e-07, + "loss": 0.08608705997467041, + "step": 33790 + }, + { + "epoch": 0.8903435264057319, + "grad_norm": 2.6987340450286865, + "learning_rate": 6.243917056942849e-07, + "loss": 0.08819245100021363, + "step": 33800 + }, + { + "epoch": 0.8906069416502307, + "grad_norm": 3.332184314727783, + "learning_rate": 6.214277245032019e-07, + "loss": 0.0826616883277893, + "step": 33810 + }, + { + "epoch": 0.8908703568947294, + "grad_norm": 3.1226818561553955, + "learning_rate": 6.184705694416471e-07, + "loss": 0.08658822774887084, + "step": 33820 + }, + { + "epoch": 0.8911337721392282, + "grad_norm": 3.1416993141174316, + "learning_rate": 6.155202426619655e-07, + "loss": 0.09397135972976685, + "step": 33830 + }, + { + "epoch": 0.8913971873837269, + "grad_norm": 3.1700146198272705, + "learning_rate": 6.125767463115317e-07, + "loss": 0.08617582321166992, + "step": 33840 + }, + { + "epoch": 0.8916606026282256, + "grad_norm": 3.659766912460327, + "learning_rate": 6.096400825327487e-07, + "loss": 0.1005324363708496, + "step": 33850 + }, + { + "epoch": 0.8919240178727244, + "grad_norm": 3.5798540115356445, + "learning_rate": 6.067102534630498e-07, + "loss": 0.09116754531860352, + "step": 33860 + }, + { + "epoch": 0.892187433117223, + "grad_norm": 4.612022876739502, + "learning_rate": 6.037872612348905e-07, + "loss": 0.08430334329605102, + "step": 33870 + }, + { + "epoch": 0.8924508483617218, + "grad_norm": 3.2607076168060303, + "learning_rate": 6.008711079757467e-07, + "loss": 0.08617499470710754, + "step": 33880 + }, + { + "epoch": 0.8927142636062205, + "grad_norm": 3.1663031578063965, + "learning_rate": 5.979617958081241e-07, + "loss": 0.08005425930023194, + "step": 33890 + }, + { + "epoch": 0.8929776788507193, + "grad_norm": 2.7275381088256836, + "learning_rate": 5.950593268495464e-07, + "loss": 0.08028669357299804, + "step": 33900 + }, + { + "epoch": 0.893241094095218, + "grad_norm": 2.798659563064575, + "learning_rate": 5.921637032125538e-07, + "loss": 0.08099162578582764, + "step": 33910 + }, + { + "epoch": 0.8935045093397168, + "grad_norm": 3.4171090126037598, + "learning_rate": 5.89274927004706e-07, + "loss": 0.09170220494270324, + "step": 33920 + }, + { + "epoch": 0.8937679245842155, + "grad_norm": 2.7121992111206055, + "learning_rate": 5.863930003285779e-07, + "loss": 0.07527078986167908, + "step": 33930 + }, + { + "epoch": 0.8940313398287142, + "grad_norm": 3.8034987449645996, + "learning_rate": 5.835179252817635e-07, + "loss": 0.08907216787338257, + "step": 33940 + }, + { + "epoch": 0.894294755073213, + "grad_norm": 2.9947826862335205, + "learning_rate": 5.806497039568648e-07, + "loss": 0.07973228693008423, + "step": 33950 + }, + { + "epoch": 0.8945581703177117, + "grad_norm": 2.947913885116577, + "learning_rate": 5.777883384414973e-07, + "loss": 0.08191275596618652, + "step": 33960 + }, + { + "epoch": 0.8948215855622105, + "grad_norm": 2.398035764694214, + "learning_rate": 5.749338308182851e-07, + "loss": 0.0796295702457428, + "step": 33970 + }, + { + "epoch": 0.8950850008067092, + "grad_norm": 3.1365795135498047, + "learning_rate": 5.720861831648617e-07, + "loss": 0.0893513798713684, + "step": 33980 + }, + { + "epoch": 0.895348416051208, + "grad_norm": 3.040403366088867, + "learning_rate": 5.692453975538703e-07, + "loss": 0.07733850479125977, + "step": 33990 + }, + { + "epoch": 0.8956118312957067, + "grad_norm": 2.356736660003662, + "learning_rate": 5.664114760529571e-07, + "loss": 0.08532350659370422, + "step": 34000 + }, + { + "epoch": 0.8958752465402054, + "grad_norm": 2.647516965866089, + "learning_rate": 5.635844207247687e-07, + "loss": 0.08528587222099304, + "step": 34010 + }, + { + "epoch": 0.8961386617847041, + "grad_norm": 2.8468241691589355, + "learning_rate": 5.607642336269615e-07, + "loss": 0.08043513894081115, + "step": 34020 + }, + { + "epoch": 0.8964020770292028, + "grad_norm": 2.8875701427459717, + "learning_rate": 5.579509168121888e-07, + "loss": 0.07911034822463989, + "step": 34030 + }, + { + "epoch": 0.8966654922737016, + "grad_norm": 3.164689779281616, + "learning_rate": 5.551444723281029e-07, + "loss": 0.08771675825119019, + "step": 34040 + }, + { + "epoch": 0.8969289075182003, + "grad_norm": 3.606766939163208, + "learning_rate": 5.523449022173566e-07, + "loss": 0.10364255905151368, + "step": 34050 + }, + { + "epoch": 0.8971923227626991, + "grad_norm": 3.046910285949707, + "learning_rate": 5.495522085175942e-07, + "loss": 0.08228116035461426, + "step": 34060 + }, + { + "epoch": 0.8974557380071978, + "grad_norm": 2.889198064804077, + "learning_rate": 5.467663932614642e-07, + "loss": 0.07738556265830994, + "step": 34070 + }, + { + "epoch": 0.8977191532516966, + "grad_norm": 2.5148823261260986, + "learning_rate": 5.439874584766014e-07, + "loss": 0.08565489053726197, + "step": 34080 + }, + { + "epoch": 0.8979825684961953, + "grad_norm": 2.8561153411865234, + "learning_rate": 5.412154061856301e-07, + "loss": 0.07297351360321044, + "step": 34090 + }, + { + "epoch": 0.8982459837406941, + "grad_norm": 2.9823615550994873, + "learning_rate": 5.38450238406174e-07, + "loss": 0.09023510217666626, + "step": 34100 + }, + { + "epoch": 0.8985093989851928, + "grad_norm": 2.7235589027404785, + "learning_rate": 5.356919571508401e-07, + "loss": 0.09568157196044921, + "step": 34110 + }, + { + "epoch": 0.8987728142296915, + "grad_norm": 3.1792116165161133, + "learning_rate": 5.329405644272268e-07, + "loss": 0.08732208013534545, + "step": 34120 + }, + { + "epoch": 0.8990362294741903, + "grad_norm": 3.8919825553894043, + "learning_rate": 5.301960622379132e-07, + "loss": 0.08982914686203003, + "step": 34130 + }, + { + "epoch": 0.899299644718689, + "grad_norm": 3.639899492263794, + "learning_rate": 5.27458452580466e-07, + "loss": 0.09040303230285644, + "step": 34140 + }, + { + "epoch": 0.8995630599631877, + "grad_norm": 3.3732199668884277, + "learning_rate": 5.24727737447438e-07, + "loss": 0.08651951551437378, + "step": 34150 + }, + { + "epoch": 0.8998264752076864, + "grad_norm": 3.2725319862365723, + "learning_rate": 5.220039188263615e-07, + "loss": 0.09009863138198852, + "step": 34160 + }, + { + "epoch": 0.9000898904521852, + "grad_norm": 3.2424583435058594, + "learning_rate": 5.192869986997484e-07, + "loss": 0.08922330141067505, + "step": 34170 + }, + { + "epoch": 0.9003533056966839, + "grad_norm": 3.9026777744293213, + "learning_rate": 5.165769790450903e-07, + "loss": 0.08428667187690735, + "step": 34180 + }, + { + "epoch": 0.9006167209411826, + "grad_norm": 2.7002367973327637, + "learning_rate": 5.138738618348548e-07, + "loss": 0.08676460385322571, + "step": 34190 + }, + { + "epoch": 0.9008801361856814, + "grad_norm": 3.4640839099884033, + "learning_rate": 5.111776490364906e-07, + "loss": 0.09998248815536499, + "step": 34200 + }, + { + "epoch": 0.9011435514301801, + "grad_norm": 3.589803695678711, + "learning_rate": 5.084883426124165e-07, + "loss": 0.07095450162887573, + "step": 34210 + }, + { + "epoch": 0.9014069666746789, + "grad_norm": 3.6323933601379395, + "learning_rate": 5.058059445200237e-07, + "loss": 0.09183912873268127, + "step": 34220 + }, + { + "epoch": 0.9016703819191776, + "grad_norm": 2.136220693588257, + "learning_rate": 5.031304567116802e-07, + "loss": 0.08723145723342896, + "step": 34230 + }, + { + "epoch": 0.9019337971636764, + "grad_norm": 4.554092884063721, + "learning_rate": 5.004618811347195e-07, + "loss": 0.09918276071548462, + "step": 34240 + }, + { + "epoch": 0.9021972124081751, + "grad_norm": 4.433627605438232, + "learning_rate": 4.978002197314469e-07, + "loss": 0.09099435210227966, + "step": 34250 + }, + { + "epoch": 0.9024606276526739, + "grad_norm": 3.177689552307129, + "learning_rate": 4.951454744391338e-07, + "loss": 0.08232357501983642, + "step": 34260 + }, + { + "epoch": 0.9027240428971726, + "grad_norm": 3.513439416885376, + "learning_rate": 4.924976471900178e-07, + "loss": 0.08061438798904419, + "step": 34270 + }, + { + "epoch": 0.9029874581416713, + "grad_norm": 4.3444504737854, + "learning_rate": 4.89856739911303e-07, + "loss": 0.07662336230278015, + "step": 34280 + }, + { + "epoch": 0.90325087338617, + "grad_norm": 3.666475296020508, + "learning_rate": 4.872227545251562e-07, + "loss": 0.08087790012359619, + "step": 34290 + }, + { + "epoch": 0.9035142886306687, + "grad_norm": 3.1013076305389404, + "learning_rate": 4.84595692948705e-07, + "loss": 0.08061681985855103, + "step": 34300 + }, + { + "epoch": 0.9037777038751675, + "grad_norm": 4.525918960571289, + "learning_rate": 4.819755570940377e-07, + "loss": 0.08898134231567383, + "step": 34310 + }, + { + "epoch": 0.9040411191196662, + "grad_norm": 3.047653913497925, + "learning_rate": 4.793623488682008e-07, + "loss": 0.09368298053741456, + "step": 34320 + }, + { + "epoch": 0.904304534364165, + "grad_norm": 3.835718870162964, + "learning_rate": 4.7675607017320323e-07, + "loss": 0.08204132318496704, + "step": 34330 + }, + { + "epoch": 0.9045679496086637, + "grad_norm": 4.734767436981201, + "learning_rate": 4.741567229060062e-07, + "loss": 0.08209804296493531, + "step": 34340 + }, + { + "epoch": 0.9048313648531625, + "grad_norm": 3.642352819442749, + "learning_rate": 4.715643089585242e-07, + "loss": 0.08672599196434021, + "step": 34350 + }, + { + "epoch": 0.9050947800976612, + "grad_norm": 3.0590031147003174, + "learning_rate": 4.6897883021763013e-07, + "loss": 0.0785254716873169, + "step": 34360 + }, + { + "epoch": 0.9053581953421599, + "grad_norm": 3.8977413177490234, + "learning_rate": 4.664002885651475e-07, + "loss": 0.08096829056739807, + "step": 34370 + }, + { + "epoch": 0.9056216105866587, + "grad_norm": 3.535943031311035, + "learning_rate": 4.638286858778496e-07, + "loss": 0.08702375888824462, + "step": 34380 + }, + { + "epoch": 0.9058850258311574, + "grad_norm": 3.3159475326538086, + "learning_rate": 4.612640240274591e-07, + "loss": 0.09486638903617858, + "step": 34390 + }, + { + "epoch": 0.9061484410756562, + "grad_norm": 3.6709792613983154, + "learning_rate": 4.587063048806462e-07, + "loss": 0.09028800725936889, + "step": 34400 + }, + { + "epoch": 0.9064118563201549, + "grad_norm": 4.422862529754639, + "learning_rate": 4.561555302990328e-07, + "loss": 0.09550145268440247, + "step": 34410 + }, + { + "epoch": 0.9066752715646537, + "grad_norm": 3.2406539916992188, + "learning_rate": 4.536117021391806e-07, + "loss": 0.0909227192401886, + "step": 34420 + }, + { + "epoch": 0.9069386868091524, + "grad_norm": 3.9687139987945557, + "learning_rate": 4.5107482225259735e-07, + "loss": 0.09125137925148011, + "step": 34430 + }, + { + "epoch": 0.9072021020536511, + "grad_norm": 3.5226128101348877, + "learning_rate": 4.4854489248573494e-07, + "loss": 0.08169687986373901, + "step": 34440 + }, + { + "epoch": 0.9074655172981498, + "grad_norm": 3.3040473461151123, + "learning_rate": 4.460219146799838e-07, + "loss": 0.07790647745132447, + "step": 34450 + }, + { + "epoch": 0.9077289325426485, + "grad_norm": 2.919196605682373, + "learning_rate": 4.4350589067167717e-07, + "loss": 0.08542875051498414, + "step": 34460 + }, + { + "epoch": 0.9079923477871473, + "grad_norm": 3.5072696208953857, + "learning_rate": 4.409968222920868e-07, + "loss": 0.10299406051635743, + "step": 34470 + }, + { + "epoch": 0.908255763031646, + "grad_norm": 3.3141977787017822, + "learning_rate": 4.384947113674165e-07, + "loss": 0.09077048301696777, + "step": 34480 + }, + { + "epoch": 0.9085191782761448, + "grad_norm": 2.8847885131835938, + "learning_rate": 4.35999559718816e-07, + "loss": 0.08016586303710938, + "step": 34490 + }, + { + "epoch": 0.9087825935206435, + "grad_norm": 3.3175978660583496, + "learning_rate": 4.3351136916236156e-07, + "loss": 0.09030872583389282, + "step": 34500 + }, + { + "epoch": 0.9090460087651423, + "grad_norm": 4.125988483428955, + "learning_rate": 4.3103014150906675e-07, + "loss": 0.07947187423706055, + "step": 34510 + }, + { + "epoch": 0.909309424009641, + "grad_norm": 3.0547385215759277, + "learning_rate": 4.285558785648747e-07, + "loss": 0.08567538857460022, + "step": 34520 + }, + { + "epoch": 0.9095728392541398, + "grad_norm": 2.4659345149993896, + "learning_rate": 4.260885821306604e-07, + "loss": 0.0782891869544983, + "step": 34530 + }, + { + "epoch": 0.9098362544986385, + "grad_norm": 2.9520230293273926, + "learning_rate": 4.236282540022307e-07, + "loss": 0.08332614898681641, + "step": 34540 + }, + { + "epoch": 0.9100996697431372, + "grad_norm": 2.914689302444458, + "learning_rate": 4.2117489597031767e-07, + "loss": 0.09373865723609924, + "step": 34550 + }, + { + "epoch": 0.910363084987636, + "grad_norm": 3.3988354206085205, + "learning_rate": 4.1872850982058067e-07, + "loss": 0.08931816816329956, + "step": 34560 + }, + { + "epoch": 0.9106265002321347, + "grad_norm": 3.0300910472869873, + "learning_rate": 4.1628909733360534e-07, + "loss": 0.08607436418533325, + "step": 34570 + }, + { + "epoch": 0.9108899154766334, + "grad_norm": 3.259099006652832, + "learning_rate": 4.1385666028490036e-07, + "loss": 0.08593692779541015, + "step": 34580 + }, + { + "epoch": 0.9111533307211321, + "grad_norm": 2.8752355575561523, + "learning_rate": 4.114312004449006e-07, + "loss": 0.07754545211791992, + "step": 34590 + }, + { + "epoch": 0.9114167459656309, + "grad_norm": 3.9250106811523438, + "learning_rate": 4.090127195789606e-07, + "loss": 0.08809812068939209, + "step": 34600 + }, + { + "epoch": 0.9116801612101296, + "grad_norm": 3.0396037101745605, + "learning_rate": 4.066012194473512e-07, + "loss": 0.09812246561050415, + "step": 34610 + }, + { + "epoch": 0.9119435764546284, + "grad_norm": 2.8047337532043457, + "learning_rate": 4.041967018052706e-07, + "loss": 0.08855500221252441, + "step": 34620 + }, + { + "epoch": 0.9122069916991271, + "grad_norm": 3.4276540279388428, + "learning_rate": 4.0179916840282776e-07, + "loss": 0.07851216197013855, + "step": 34630 + }, + { + "epoch": 0.9124704069436258, + "grad_norm": 3.3260245323181152, + "learning_rate": 3.9940862098505117e-07, + "loss": 0.08380018472671509, + "step": 34640 + }, + { + "epoch": 0.9127338221881246, + "grad_norm": 3.2529170513153076, + "learning_rate": 3.970250612918858e-07, + "loss": 0.09415042400360107, + "step": 34650 + }, + { + "epoch": 0.9129972374326233, + "grad_norm": 3.0640816688537598, + "learning_rate": 3.946484910581871e-07, + "loss": 0.08776201009750366, + "step": 34660 + }, + { + "epoch": 0.9132606526771221, + "grad_norm": 3.4164910316467285, + "learning_rate": 3.9227891201372827e-07, + "loss": 0.08057547211647034, + "step": 34670 + }, + { + "epoch": 0.9135240679216208, + "grad_norm": 3.4903430938720703, + "learning_rate": 3.899163258831895e-07, + "loss": 0.08529804944992066, + "step": 34680 + }, + { + "epoch": 0.9137874831661196, + "grad_norm": 2.3379082679748535, + "learning_rate": 3.8756073438616203e-07, + "loss": 0.08512848615646362, + "step": 34690 + }, + { + "epoch": 0.9140508984106183, + "grad_norm": 3.0583620071411133, + "learning_rate": 3.8521213923714775e-07, + "loss": 0.08457539081573487, + "step": 34700 + }, + { + "epoch": 0.914314313655117, + "grad_norm": 2.4475715160369873, + "learning_rate": 3.8287054214555694e-07, + "loss": 0.08075867295265197, + "step": 34710 + }, + { + "epoch": 0.9145777288996157, + "grad_norm": 2.2949345111846924, + "learning_rate": 3.805359448157031e-07, + "loss": 0.08054609894752503, + "step": 34720 + }, + { + "epoch": 0.9148411441441144, + "grad_norm": 3.6749932765960693, + "learning_rate": 3.78208348946808e-07, + "loss": 0.08090617060661316, + "step": 34730 + }, + { + "epoch": 0.9151045593886132, + "grad_norm": 2.8682212829589844, + "learning_rate": 3.7588775623299433e-07, + "loss": 0.0919027328491211, + "step": 34740 + }, + { + "epoch": 0.9153679746331119, + "grad_norm": 3.7178449630737305, + "learning_rate": 3.735741683632932e-07, + "loss": 0.08541535139083863, + "step": 34750 + }, + { + "epoch": 0.9156313898776107, + "grad_norm": 2.4408204555511475, + "learning_rate": 3.712675870216309e-07, + "loss": 0.08207824230194091, + "step": 34760 + }, + { + "epoch": 0.9158948051221094, + "grad_norm": 4.000577449798584, + "learning_rate": 3.689680138868379e-07, + "loss": 0.08217844963073731, + "step": 34770 + }, + { + "epoch": 0.9161582203666082, + "grad_norm": 2.8707523345947266, + "learning_rate": 3.66675450632642e-07, + "loss": 0.0790333867073059, + "step": 34780 + }, + { + "epoch": 0.9164216356111069, + "grad_norm": 2.8341212272644043, + "learning_rate": 3.643898989276684e-07, + "loss": 0.08879296779632569, + "step": 34790 + }, + { + "epoch": 0.9166850508556056, + "grad_norm": 3.5799801349639893, + "learning_rate": 3.6211136043544425e-07, + "loss": 0.0860048770904541, + "step": 34800 + }, + { + "epoch": 0.9169484661001044, + "grad_norm": 3.2639358043670654, + "learning_rate": 3.598398368143863e-07, + "loss": 0.09017524719238282, + "step": 34810 + }, + { + "epoch": 0.9172118813446031, + "grad_norm": 2.117288827896118, + "learning_rate": 3.575753297178053e-07, + "loss": 0.07934668064117431, + "step": 34820 + }, + { + "epoch": 0.9174752965891019, + "grad_norm": 3.1589303016662598, + "learning_rate": 3.553178407939095e-07, + "loss": 0.08012961149215699, + "step": 34830 + }, + { + "epoch": 0.9177387118336006, + "grad_norm": 2.319383382797241, + "learning_rate": 3.5306737168579685e-07, + "loss": 0.07514309883117676, + "step": 34840 + }, + { + "epoch": 0.9180021270780994, + "grad_norm": 3.3474016189575195, + "learning_rate": 3.5082392403145373e-07, + "loss": 0.08532885909080505, + "step": 34850 + }, + { + "epoch": 0.918265542322598, + "grad_norm": 3.710047721862793, + "learning_rate": 3.485874994637606e-07, + "loss": 0.08599490523338318, + "step": 34860 + }, + { + "epoch": 0.9185289575670968, + "grad_norm": 3.8279218673706055, + "learning_rate": 3.463580996104821e-07, + "loss": 0.09103659391403199, + "step": 34870 + }, + { + "epoch": 0.9187923728115955, + "grad_norm": 3.1961450576782227, + "learning_rate": 3.4413572609427235e-07, + "loss": 0.09026677012443543, + "step": 34880 + }, + { + "epoch": 0.9190557880560942, + "grad_norm": 5.260765075683594, + "learning_rate": 3.4192038053267095e-07, + "loss": 0.09015097618103027, + "step": 34890 + }, + { + "epoch": 0.919319203300593, + "grad_norm": 3.2651538848876953, + "learning_rate": 3.3971206453810133e-07, + "loss": 0.07386161088943481, + "step": 34900 + }, + { + "epoch": 0.9195826185450917, + "grad_norm": 2.685894727706909, + "learning_rate": 3.3751077971787115e-07, + "loss": 0.07826831340789794, + "step": 34910 + }, + { + "epoch": 0.9198460337895905, + "grad_norm": 3.9342615604400635, + "learning_rate": 3.353165276741699e-07, + "loss": 0.08399295210838317, + "step": 34920 + }, + { + "epoch": 0.9201094490340892, + "grad_norm": 3.6597211360931396, + "learning_rate": 3.331293100040711e-07, + "loss": 0.07790507674217224, + "step": 34930 + }, + { + "epoch": 0.920372864278588, + "grad_norm": 3.717237710952759, + "learning_rate": 3.3094912829952477e-07, + "loss": 0.09116606712341309, + "step": 34940 + }, + { + "epoch": 0.9206362795230867, + "grad_norm": 3.8569538593292236, + "learning_rate": 3.2877598414736033e-07, + "loss": 0.08034144043922424, + "step": 34950 + }, + { + "epoch": 0.9208996947675855, + "grad_norm": 3.9546608924865723, + "learning_rate": 3.266098791292871e-07, + "loss": 0.0916824460029602, + "step": 34960 + }, + { + "epoch": 0.9211631100120842, + "grad_norm": 2.2304117679595947, + "learning_rate": 3.244508148218895e-07, + "loss": 0.06886508464813232, + "step": 34970 + }, + { + "epoch": 0.9214265252565829, + "grad_norm": 3.654021739959717, + "learning_rate": 3.22298792796627e-07, + "loss": 0.09011399745941162, + "step": 34980 + }, + { + "epoch": 0.9216899405010817, + "grad_norm": 2.9298200607299805, + "learning_rate": 3.2015381461983577e-07, + "loss": 0.09375889301300049, + "step": 34990 + }, + { + "epoch": 0.9219533557455803, + "grad_norm": 2.915745258331299, + "learning_rate": 3.1801588185272016e-07, + "loss": 0.07374373674392701, + "step": 35000 + }, + { + "epoch": 0.9222167709900791, + "grad_norm": 2.8780250549316406, + "learning_rate": 3.1588499605136435e-07, + "loss": 0.09162523150444031, + "step": 35010 + }, + { + "epoch": 0.9224801862345778, + "grad_norm": 3.673145294189453, + "learning_rate": 3.1376115876671553e-07, + "loss": 0.08796296715736389, + "step": 35020 + }, + { + "epoch": 0.9227436014790766, + "grad_norm": 2.194580316543579, + "learning_rate": 3.1164437154459494e-07, + "loss": 0.08083784580230713, + "step": 35030 + }, + { + "epoch": 0.9230070167235753, + "grad_norm": 4.728752613067627, + "learning_rate": 3.0953463592569343e-07, + "loss": 0.08136105537414551, + "step": 35040 + }, + { + "epoch": 0.9232704319680741, + "grad_norm": 3.203824281692505, + "learning_rate": 3.074319534455639e-07, + "loss": 0.09154722094535828, + "step": 35050 + }, + { + "epoch": 0.9235338472125728, + "grad_norm": 2.8871078491210938, + "learning_rate": 3.053363256346331e-07, + "loss": 0.09584543704986573, + "step": 35060 + }, + { + "epoch": 0.9237972624570715, + "grad_norm": 4.758818626403809, + "learning_rate": 3.0324775401818884e-07, + "loss": 0.08051756620407105, + "step": 35070 + }, + { + "epoch": 0.9240606777015703, + "grad_norm": 4.0753021240234375, + "learning_rate": 3.0116624011637953e-07, + "loss": 0.09277215003967285, + "step": 35080 + }, + { + "epoch": 0.924324092946069, + "grad_norm": 3.182389736175537, + "learning_rate": 2.9909178544422656e-07, + "loss": 0.07685382962226868, + "step": 35090 + }, + { + "epoch": 0.9245875081905678, + "grad_norm": 3.438662528991699, + "learning_rate": 2.9702439151160334e-07, + "loss": 0.08587347269058228, + "step": 35100 + }, + { + "epoch": 0.9248509234350665, + "grad_norm": 3.1514713764190674, + "learning_rate": 2.949640598232517e-07, + "loss": 0.07920421957969666, + "step": 35110 + }, + { + "epoch": 0.9251143386795653, + "grad_norm": 3.9428534507751465, + "learning_rate": 2.9291079187876656e-07, + "loss": 0.09536715745925903, + "step": 35120 + }, + { + "epoch": 0.925377753924064, + "grad_norm": 4.046616077423096, + "learning_rate": 2.9086458917260585e-07, + "loss": 0.0780355155467987, + "step": 35130 + }, + { + "epoch": 0.9256411691685628, + "grad_norm": 3.5762743949890137, + "learning_rate": 2.8882545319408706e-07, + "loss": 0.07884734272956848, + "step": 35140 + }, + { + "epoch": 0.9259045844130614, + "grad_norm": 2.744865655899048, + "learning_rate": 2.8679338542737745e-07, + "loss": 0.08550113439559937, + "step": 35150 + }, + { + "epoch": 0.9261679996575601, + "grad_norm": 4.032914638519287, + "learning_rate": 2.8476838735150613e-07, + "loss": 0.07993297576904297, + "step": 35160 + }, + { + "epoch": 0.9264314149020589, + "grad_norm": 3.747814416885376, + "learning_rate": 2.8275046044035304e-07, + "loss": 0.07986670732498169, + "step": 35170 + }, + { + "epoch": 0.9266948301465576, + "grad_norm": 2.8504292964935303, + "learning_rate": 2.8073960616265215e-07, + "loss": 0.0877582848072052, + "step": 35180 + }, + { + "epoch": 0.9269582453910564, + "grad_norm": 3.639679193496704, + "learning_rate": 2.7873582598199056e-07, + "loss": 0.09818193316459656, + "step": 35190 + }, + { + "epoch": 0.9272216606355551, + "grad_norm": 3.0701212882995605, + "learning_rate": 2.7673912135680714e-07, + "loss": 0.07075505256652832, + "step": 35200 + }, + { + "epoch": 0.9274850758800539, + "grad_norm": 3.5033814907073975, + "learning_rate": 2.7474949374038717e-07, + "loss": 0.07938226461410522, + "step": 35210 + }, + { + "epoch": 0.9277484911245526, + "grad_norm": 2.3465144634246826, + "learning_rate": 2.7276694458086894e-07, + "loss": 0.08473074436187744, + "step": 35220 + }, + { + "epoch": 0.9280119063690514, + "grad_norm": 3.7491815090179443, + "learning_rate": 2.7079147532123704e-07, + "loss": 0.09437916278839112, + "step": 35230 + }, + { + "epoch": 0.9282753216135501, + "grad_norm": 2.998117446899414, + "learning_rate": 2.688230873993225e-07, + "loss": 0.06754139661788941, + "step": 35240 + }, + { + "epoch": 0.9285387368580488, + "grad_norm": 3.8980607986450195, + "learning_rate": 2.6686178224780367e-07, + "loss": 0.0745414137840271, + "step": 35250 + }, + { + "epoch": 0.9288021521025476, + "grad_norm": 3.6264164447784424, + "learning_rate": 2.6490756129420313e-07, + "loss": 0.09022721648216248, + "step": 35260 + }, + { + "epoch": 0.9290655673470463, + "grad_norm": 3.6303157806396484, + "learning_rate": 2.6296042596088643e-07, + "loss": 0.08179405331611633, + "step": 35270 + }, + { + "epoch": 0.9293289825915451, + "grad_norm": 3.2234721183776855, + "learning_rate": 2.610203776650655e-07, + "loss": 0.08703947067260742, + "step": 35280 + }, + { + "epoch": 0.9295923978360437, + "grad_norm": 4.058443069458008, + "learning_rate": 2.5908741781878744e-07, + "loss": 0.08165610432624817, + "step": 35290 + }, + { + "epoch": 0.9298558130805425, + "grad_norm": 3.4541196823120117, + "learning_rate": 2.5716154782894573e-07, + "loss": 0.08090918064117432, + "step": 35300 + }, + { + "epoch": 0.9301192283250412, + "grad_norm": 4.408613204956055, + "learning_rate": 2.552427690972703e-07, + "loss": 0.07848947048187256, + "step": 35310 + }, + { + "epoch": 0.9303826435695399, + "grad_norm": 3.5581841468811035, + "learning_rate": 2.5333108302033394e-07, + "loss": 0.08197213411331176, + "step": 35320 + }, + { + "epoch": 0.9306460588140387, + "grad_norm": 3.8104805946350098, + "learning_rate": 2.514264909895414e-07, + "loss": 0.08282191753387451, + "step": 35330 + }, + { + "epoch": 0.9309094740585374, + "grad_norm": 3.007112741470337, + "learning_rate": 2.495289943911361e-07, + "loss": 0.09306011199951172, + "step": 35340 + }, + { + "epoch": 0.9311728893030362, + "grad_norm": 3.3257217407226562, + "learning_rate": 2.476385946062021e-07, + "loss": 0.08321922421455383, + "step": 35350 + }, + { + "epoch": 0.9314363045475349, + "grad_norm": 3.368159770965576, + "learning_rate": 2.4575529301065104e-07, + "loss": 0.08963929414749146, + "step": 35360 + }, + { + "epoch": 0.9316997197920337, + "grad_norm": 4.778257369995117, + "learning_rate": 2.43879090975232e-07, + "loss": 0.08702784776687622, + "step": 35370 + }, + { + "epoch": 0.9319631350365324, + "grad_norm": 4.080934524536133, + "learning_rate": 2.4200998986552484e-07, + "loss": 0.08376226425170899, + "step": 35380 + }, + { + "epoch": 0.9322265502810312, + "grad_norm": 4.291754722595215, + "learning_rate": 2.401479910419413e-07, + "loss": 0.08626672029495239, + "step": 35390 + }, + { + "epoch": 0.9324899655255299, + "grad_norm": 3.040850877761841, + "learning_rate": 2.3829309585972738e-07, + "loss": 0.08492223024368287, + "step": 35400 + }, + { + "epoch": 0.9327533807700286, + "grad_norm": 4.045016765594482, + "learning_rate": 2.364453056689553e-07, + "loss": 0.07744430303573609, + "step": 35410 + }, + { + "epoch": 0.9330167960145274, + "grad_norm": 3.5754144191741943, + "learning_rate": 2.3460462181452258e-07, + "loss": 0.08965497016906739, + "step": 35420 + }, + { + "epoch": 0.933280211259026, + "grad_norm": 2.975832939147949, + "learning_rate": 2.3277104563616205e-07, + "loss": 0.09426172375679016, + "step": 35430 + }, + { + "epoch": 0.9335436265035248, + "grad_norm": 2.4494810104370117, + "learning_rate": 2.3094457846842943e-07, + "loss": 0.07939582467079162, + "step": 35440 + }, + { + "epoch": 0.9338070417480235, + "grad_norm": 5.184408664703369, + "learning_rate": 2.291252216407047e-07, + "loss": 0.07262312173843384, + "step": 35450 + }, + { + "epoch": 0.9340704569925223, + "grad_norm": 3.5664632320404053, + "learning_rate": 2.273129764771953e-07, + "loss": 0.0773284375667572, + "step": 35460 + }, + { + "epoch": 0.934333872237021, + "grad_norm": 4.060650825500488, + "learning_rate": 2.2550784429693052e-07, + "loss": 0.08531564474105835, + "step": 35470 + }, + { + "epoch": 0.9345972874815198, + "grad_norm": 3.4591619968414307, + "learning_rate": 2.2370982641376494e-07, + "loss": 0.06935621500015259, + "step": 35480 + }, + { + "epoch": 0.9348607027260185, + "grad_norm": 5.154021739959717, + "learning_rate": 2.2191892413637173e-07, + "loss": 0.07920868396759033, + "step": 35490 + }, + { + "epoch": 0.9351241179705172, + "grad_norm": 3.9261136054992676, + "learning_rate": 2.2013513876824933e-07, + "loss": 0.07650346755981445, + "step": 35500 + }, + { + "epoch": 0.935387533215016, + "grad_norm": 3.955589771270752, + "learning_rate": 2.183584716077114e-07, + "loss": 0.08930530548095703, + "step": 35510 + }, + { + "epoch": 0.9356509484595147, + "grad_norm": 3.378135919570923, + "learning_rate": 2.1658892394789244e-07, + "loss": 0.0870656967163086, + "step": 35520 + }, + { + "epoch": 0.9359143637040135, + "grad_norm": 4.072518348693848, + "learning_rate": 2.1482649707674775e-07, + "loss": 0.08003019094467163, + "step": 35530 + }, + { + "epoch": 0.9361777789485122, + "grad_norm": 2.946012258529663, + "learning_rate": 2.13071192277049e-07, + "loss": 0.07830334901809692, + "step": 35540 + }, + { + "epoch": 0.936441194193011, + "grad_norm": 2.590907573699951, + "learning_rate": 2.1132301082637752e-07, + "loss": 0.08750003576278687, + "step": 35550 + }, + { + "epoch": 0.9367046094375097, + "grad_norm": 2.990302085876465, + "learning_rate": 2.0958195399713888e-07, + "loss": 0.08712869882583618, + "step": 35560 + }, + { + "epoch": 0.9369680246820085, + "grad_norm": 3.4447948932647705, + "learning_rate": 2.0784802305654938e-07, + "loss": 0.07839518785476685, + "step": 35570 + }, + { + "epoch": 0.9372314399265071, + "grad_norm": 4.332442760467529, + "learning_rate": 2.061212192666373e-07, + "loss": 0.08971573114395141, + "step": 35580 + }, + { + "epoch": 0.9374948551710058, + "grad_norm": 2.749143600463867, + "learning_rate": 2.0440154388424616e-07, + "loss": 0.08754003047943115, + "step": 35590 + }, + { + "epoch": 0.9377582704155046, + "grad_norm": 2.927546739578247, + "learning_rate": 2.0268899816102916e-07, + "loss": 0.0785633683204651, + "step": 35600 + }, + { + "epoch": 0.9380216856600033, + "grad_norm": 3.532088041305542, + "learning_rate": 2.0098358334345147e-07, + "loss": 0.08401823043823242, + "step": 35610 + }, + { + "epoch": 0.9382851009045021, + "grad_norm": 3.187556028366089, + "learning_rate": 1.9928530067278684e-07, + "loss": 0.07669014930725097, + "step": 35620 + }, + { + "epoch": 0.9385485161490008, + "grad_norm": 2.6920106410980225, + "learning_rate": 1.9759415138511984e-07, + "loss": 0.08418586254119872, + "step": 35630 + }, + { + "epoch": 0.9388119313934996, + "grad_norm": 4.2160797119140625, + "learning_rate": 1.959101367113414e-07, + "loss": 0.09396454095840454, + "step": 35640 + }, + { + "epoch": 0.9390753466379983, + "grad_norm": 3.0394904613494873, + "learning_rate": 1.9423325787714774e-07, + "loss": 0.0792531967163086, + "step": 35650 + }, + { + "epoch": 0.9393387618824971, + "grad_norm": 2.6805026531219482, + "learning_rate": 1.9256351610304814e-07, + "loss": 0.07476949691772461, + "step": 35660 + }, + { + "epoch": 0.9396021771269958, + "grad_norm": 2.5930001735687256, + "learning_rate": 1.909009126043504e-07, + "loss": 0.08326936960220337, + "step": 35670 + }, + { + "epoch": 0.9398655923714945, + "grad_norm": 2.6263363361358643, + "learning_rate": 1.8924544859116656e-07, + "loss": 0.08831897377967834, + "step": 35680 + }, + { + "epoch": 0.9401290076159933, + "grad_norm": 3.397096633911133, + "learning_rate": 1.8759712526841943e-07, + "loss": 0.08533579707145691, + "step": 35690 + }, + { + "epoch": 0.940392422860492, + "grad_norm": 3.057429790496826, + "learning_rate": 1.8595594383582605e-07, + "loss": 0.08875679969787598, + "step": 35700 + }, + { + "epoch": 0.9406558381049908, + "grad_norm": 3.3069798946380615, + "learning_rate": 1.8432190548791197e-07, + "loss": 0.08948233127593994, + "step": 35710 + }, + { + "epoch": 0.9409192533494894, + "grad_norm": 2.6773791313171387, + "learning_rate": 1.8269501141399915e-07, + "loss": 0.08357824087142944, + "step": 35720 + }, + { + "epoch": 0.9411826685939882, + "grad_norm": 3.4415228366851807, + "learning_rate": 1.8107526279821041e-07, + "loss": 0.08246555924415588, + "step": 35730 + }, + { + "epoch": 0.9414460838384869, + "grad_norm": 3.3099188804626465, + "learning_rate": 1.794626608194705e-07, + "loss": 0.08040993809700012, + "step": 35740 + }, + { + "epoch": 0.9417094990829857, + "grad_norm": 3.0002310276031494, + "learning_rate": 1.7785720665150054e-07, + "loss": 0.08492194414138794, + "step": 35750 + }, + { + "epoch": 0.9419729143274844, + "grad_norm": 3.571011781692505, + "learning_rate": 1.7625890146281798e-07, + "loss": 0.08621615171432495, + "step": 35760 + }, + { + "epoch": 0.9422363295719831, + "grad_norm": 2.9340009689331055, + "learning_rate": 1.7466774641674012e-07, + "loss": 0.07213945388793945, + "step": 35770 + }, + { + "epoch": 0.9424997448164819, + "grad_norm": 3.554152011871338, + "learning_rate": 1.7308374267137606e-07, + "loss": 0.0857938289642334, + "step": 35780 + }, + { + "epoch": 0.9427631600609806, + "grad_norm": 3.5777173042297363, + "learning_rate": 1.715068913796336e-07, + "loss": 0.08814908266067505, + "step": 35790 + }, + { + "epoch": 0.9430265753054794, + "grad_norm": 3.6281986236572266, + "learning_rate": 1.6993719368921468e-07, + "loss": 0.08689531683921814, + "step": 35800 + }, + { + "epoch": 0.9432899905499781, + "grad_norm": 3.8593194484710693, + "learning_rate": 1.683746507426087e-07, + "loss": 0.07296856641769409, + "step": 35810 + }, + { + "epoch": 0.9435534057944769, + "grad_norm": 4.059710502624512, + "learning_rate": 1.66819263677106e-07, + "loss": 0.08675963878631592, + "step": 35820 + }, + { + "epoch": 0.9438168210389756, + "grad_norm": 2.736088514328003, + "learning_rate": 1.6527103362478202e-07, + "loss": 0.08935936093330384, + "step": 35830 + }, + { + "epoch": 0.9440802362834743, + "grad_norm": 3.0965464115142822, + "learning_rate": 1.6372996171250544e-07, + "loss": 0.09971678256988525, + "step": 35840 + }, + { + "epoch": 0.9443436515279731, + "grad_norm": 3.029658794403076, + "learning_rate": 1.6219604906193677e-07, + "loss": 0.07428082823753357, + "step": 35850 + }, + { + "epoch": 0.9446070667724717, + "grad_norm": 4.0001983642578125, + "learning_rate": 1.606692967895207e-07, + "loss": 0.08609504103660584, + "step": 35860 + }, + { + "epoch": 0.9448704820169705, + "grad_norm": 4.174829006195068, + "learning_rate": 1.5914970600649615e-07, + "loss": 0.09754637479782105, + "step": 35870 + }, + { + "epoch": 0.9451338972614692, + "grad_norm": 3.3369531631469727, + "learning_rate": 1.5763727781888726e-07, + "loss": 0.06865531206130981, + "step": 35880 + }, + { + "epoch": 0.945397312505968, + "grad_norm": 2.7387948036193848, + "learning_rate": 1.5613201332750127e-07, + "loss": 0.08175322413444519, + "step": 35890 + }, + { + "epoch": 0.9456607277504667, + "grad_norm": 3.5467112064361572, + "learning_rate": 1.5463391362793735e-07, + "loss": 0.078273344039917, + "step": 35900 + }, + { + "epoch": 0.9459241429949655, + "grad_norm": 2.473376989364624, + "learning_rate": 1.5314297981057658e-07, + "loss": 0.08549461364746094, + "step": 35910 + }, + { + "epoch": 0.9461875582394642, + "grad_norm": 3.4955010414123535, + "learning_rate": 1.5165921296058427e-07, + "loss": 0.08397986888885497, + "step": 35920 + }, + { + "epoch": 0.9464509734839629, + "grad_norm": 2.184948682785034, + "learning_rate": 1.501826141579099e-07, + "loss": 0.07227013111114503, + "step": 35930 + }, + { + "epoch": 0.9467143887284617, + "grad_norm": 2.3793978691101074, + "learning_rate": 1.48713184477286e-07, + "loss": 0.07976223826408387, + "step": 35940 + }, + { + "epoch": 0.9469778039729604, + "grad_norm": 4.323724269866943, + "learning_rate": 1.4725092498822814e-07, + "loss": 0.08139788508415222, + "step": 35950 + }, + { + "epoch": 0.9472412192174592, + "grad_norm": 3.7213127613067627, + "learning_rate": 1.457958367550305e-07, + "loss": 0.084447181224823, + "step": 35960 + }, + { + "epoch": 0.9475046344619579, + "grad_norm": 2.254516839981079, + "learning_rate": 1.443479208367693e-07, + "loss": 0.07382199168205261, + "step": 35970 + }, + { + "epoch": 0.9477680497064567, + "grad_norm": 3.6569080352783203, + "learning_rate": 1.4290717828730037e-07, + "loss": 0.08451644182205201, + "step": 35980 + }, + { + "epoch": 0.9480314649509554, + "grad_norm": 4.592838764190674, + "learning_rate": 1.4147361015525717e-07, + "loss": 0.08421679735183715, + "step": 35990 + }, + { + "epoch": 0.9482948801954542, + "grad_norm": 3.530100107192993, + "learning_rate": 1.4004721748405392e-07, + "loss": 0.10214877128601074, + "step": 36000 + }, + { + "epoch": 0.9485582954399528, + "grad_norm": 4.11027193069458, + "learning_rate": 1.3862800131188237e-07, + "loss": 0.09025883674621582, + "step": 36010 + }, + { + "epoch": 0.9488217106844515, + "grad_norm": 4.2960615158081055, + "learning_rate": 1.37215962671704e-07, + "loss": 0.08418797850608825, + "step": 36020 + }, + { + "epoch": 0.9490851259289503, + "grad_norm": 3.516362190246582, + "learning_rate": 1.3581110259126674e-07, + "loss": 0.0800922155380249, + "step": 36030 + }, + { + "epoch": 0.949348541173449, + "grad_norm": 3.230654001235962, + "learning_rate": 1.3441342209308595e-07, + "loss": 0.08695355653762818, + "step": 36040 + }, + { + "epoch": 0.9496119564179478, + "grad_norm": 4.088803768157959, + "learning_rate": 1.3302292219445457e-07, + "loss": 0.08544000387191772, + "step": 36050 + }, + { + "epoch": 0.9498753716624465, + "grad_norm": 3.3295881748199463, + "learning_rate": 1.3163960390743746e-07, + "loss": 0.07673301696777343, + "step": 36060 + }, + { + "epoch": 0.9501387869069453, + "grad_norm": 3.865393877029419, + "learning_rate": 1.302634682388726e-07, + "loss": 0.0691301167011261, + "step": 36070 + }, + { + "epoch": 0.950402202151444, + "grad_norm": 3.059763193130493, + "learning_rate": 1.2889451619037318e-07, + "loss": 0.08133491277694702, + "step": 36080 + }, + { + "epoch": 0.9506656173959428, + "grad_norm": 4.078855037689209, + "learning_rate": 1.2753274875832112e-07, + "loss": 0.08215219974517822, + "step": 36090 + }, + { + "epoch": 0.9509290326404415, + "grad_norm": 2.079050302505493, + "learning_rate": 1.2617816693386907e-07, + "loss": 0.07495338916778564, + "step": 36100 + }, + { + "epoch": 0.9511924478849402, + "grad_norm": 3.2780537605285645, + "learning_rate": 1.2483077170293954e-07, + "loss": 0.08882225751876831, + "step": 36110 + }, + { + "epoch": 0.951455863129439, + "grad_norm": 3.6124536991119385, + "learning_rate": 1.2349056404622472e-07, + "loss": 0.07228685617446899, + "step": 36120 + }, + { + "epoch": 0.9517192783739377, + "grad_norm": 2.95235538482666, + "learning_rate": 1.2215754493918763e-07, + "loss": 0.08821086883544922, + "step": 36130 + }, + { + "epoch": 0.9519826936184365, + "grad_norm": 3.007786273956299, + "learning_rate": 1.208317153520555e-07, + "loss": 0.09799190759658813, + "step": 36140 + }, + { + "epoch": 0.9522461088629351, + "grad_norm": 3.6415412425994873, + "learning_rate": 1.195130762498231e-07, + "loss": 0.07750015258789063, + "step": 36150 + }, + { + "epoch": 0.952509524107434, + "grad_norm": 3.42032527923584, + "learning_rate": 1.1820162859225493e-07, + "loss": 0.08089326620101929, + "step": 36160 + }, + { + "epoch": 0.9527729393519326, + "grad_norm": 2.4192352294921875, + "learning_rate": 1.1689737333387852e-07, + "loss": 0.0768781065940857, + "step": 36170 + }, + { + "epoch": 0.9530363545964314, + "grad_norm": 3.1803667545318604, + "learning_rate": 1.1560031142398787e-07, + "loss": 0.07857604622840882, + "step": 36180 + }, + { + "epoch": 0.9532997698409301, + "grad_norm": 2.9735517501831055, + "learning_rate": 1.1431044380663891e-07, + "loss": 0.08995000123977662, + "step": 36190 + }, + { + "epoch": 0.9535631850854288, + "grad_norm": 2.8042662143707275, + "learning_rate": 1.1302777142065402e-07, + "loss": 0.09115211963653565, + "step": 36200 + }, + { + "epoch": 0.9538266003299276, + "grad_norm": 4.019819736480713, + "learning_rate": 1.117522951996175e-07, + "loss": 0.08938701152801513, + "step": 36210 + }, + { + "epoch": 0.9540900155744263, + "grad_norm": 3.830817937850952, + "learning_rate": 1.1048401607187565e-07, + "loss": 0.0883729875087738, + "step": 36220 + }, + { + "epoch": 0.9543534308189251, + "grad_norm": 3.3281593322753906, + "learning_rate": 1.0922293496053671e-07, + "loss": 0.09427074193954468, + "step": 36230 + }, + { + "epoch": 0.9546168460634238, + "grad_norm": 3.2076969146728516, + "learning_rate": 1.0796905278347092e-07, + "loss": 0.08058937788009643, + "step": 36240 + }, + { + "epoch": 0.9548802613079226, + "grad_norm": 3.318570852279663, + "learning_rate": 1.0672237045330491e-07, + "loss": 0.07468730211257935, + "step": 36250 + }, + { + "epoch": 0.9551436765524213, + "grad_norm": 3.6576366424560547, + "learning_rate": 1.0548288887743174e-07, + "loss": 0.07270472049713135, + "step": 36260 + }, + { + "epoch": 0.9554070917969201, + "grad_norm": 4.028535842895508, + "learning_rate": 1.0425060895799865e-07, + "loss": 0.08274773955345154, + "step": 36270 + }, + { + "epoch": 0.9556705070414188, + "grad_norm": 3.9358878135681152, + "learning_rate": 1.0302553159190932e-07, + "loss": 0.08901549577713012, + "step": 36280 + }, + { + "epoch": 0.9559339222859174, + "grad_norm": 3.3334083557128906, + "learning_rate": 1.0180765767083045e-07, + "loss": 0.08110134601593018, + "step": 36290 + }, + { + "epoch": 0.9561973375304162, + "grad_norm": 3.35532546043396, + "learning_rate": 1.00596988081183e-07, + "loss": 0.07306838631629944, + "step": 36300 + }, + { + "epoch": 0.9564607527749149, + "grad_norm": 3.415783405303955, + "learning_rate": 9.939352370414435e-08, + "loss": 0.07844064235687256, + "step": 36310 + }, + { + "epoch": 0.9567241680194137, + "grad_norm": 2.5848941802978516, + "learning_rate": 9.819726541564711e-08, + "loss": 0.08685501813888549, + "step": 36320 + }, + { + "epoch": 0.9569875832639124, + "grad_norm": 3.358569622039795, + "learning_rate": 9.700821408638039e-08, + "loss": 0.0871274471282959, + "step": 36330 + }, + { + "epoch": 0.9572509985084112, + "grad_norm": 2.1462488174438477, + "learning_rate": 9.582637058178635e-08, + "loss": 0.0806852400302887, + "step": 36340 + }, + { + "epoch": 0.9575144137529099, + "grad_norm": 3.50710129737854, + "learning_rate": 9.465173576206355e-08, + "loss": 0.08141102194786072, + "step": 36350 + }, + { + "epoch": 0.9577778289974086, + "grad_norm": 3.9189441204071045, + "learning_rate": 9.348431048215923e-08, + "loss": 0.08635926246643066, + "step": 36360 + }, + { + "epoch": 0.9580412442419074, + "grad_norm": 3.1705801486968994, + "learning_rate": 9.232409559177701e-08, + "loss": 0.08082292079925538, + "step": 36370 + }, + { + "epoch": 0.9583046594864061, + "grad_norm": 3.6148178577423096, + "learning_rate": 9.117109193537254e-08, + "loss": 0.08813682794570923, + "step": 36380 + }, + { + "epoch": 0.9585680747309049, + "grad_norm": 3.4696545600891113, + "learning_rate": 9.002530035215007e-08, + "loss": 0.08665171265602112, + "step": 36390 + }, + { + "epoch": 0.9588314899754036, + "grad_norm": 3.5750224590301514, + "learning_rate": 8.888672167606693e-08, + "loss": 0.0845930516719818, + "step": 36400 + }, + { + "epoch": 0.9590949052199024, + "grad_norm": 3.2835988998413086, + "learning_rate": 8.775535673582913e-08, + "loss": 0.07250134944915772, + "step": 36410 + }, + { + "epoch": 0.9593583204644011, + "grad_norm": 4.413487911224365, + "learning_rate": 8.663120635489241e-08, + "loss": 0.08059373497962952, + "step": 36420 + }, + { + "epoch": 0.9596217357088999, + "grad_norm": 3.376332998275757, + "learning_rate": 8.55142713514645e-08, + "loss": 0.08823221325874328, + "step": 36430 + }, + { + "epoch": 0.9598851509533985, + "grad_norm": 3.240321159362793, + "learning_rate": 8.440455253849622e-08, + "loss": 0.07824410200119018, + "step": 36440 + }, + { + "epoch": 0.9601485661978972, + "grad_norm": 2.689415693283081, + "learning_rate": 8.330205072369035e-08, + "loss": 0.08759328126907348, + "step": 36450 + }, + { + "epoch": 0.960411981442396, + "grad_norm": 3.6855602264404297, + "learning_rate": 8.22067667094939e-08, + "loss": 0.08704519271850586, + "step": 36460 + }, + { + "epoch": 0.9606753966868947, + "grad_norm": 3.1361541748046875, + "learning_rate": 8.11187012931025e-08, + "loss": 0.08616516590118409, + "step": 36470 + }, + { + "epoch": 0.9609388119313935, + "grad_norm": 2.9094643592834473, + "learning_rate": 8.003785526645824e-08, + "loss": 0.07651432156562805, + "step": 36480 + }, + { + "epoch": 0.9612022271758922, + "grad_norm": 4.9116644859313965, + "learning_rate": 7.896422941624626e-08, + "loss": 0.07787474989891052, + "step": 36490 + }, + { + "epoch": 0.961465642420391, + "grad_norm": 3.6026175022125244, + "learning_rate": 7.789782452389705e-08, + "loss": 0.08877738118171692, + "step": 36500 + }, + { + "epoch": 0.9617290576648897, + "grad_norm": 3.443798303604126, + "learning_rate": 7.683864136558749e-08, + "loss": 0.07672094106674195, + "step": 36510 + }, + { + "epoch": 0.9619924729093885, + "grad_norm": 3.748326063156128, + "learning_rate": 7.578668071223649e-08, + "loss": 0.08077262043952942, + "step": 36520 + }, + { + "epoch": 0.9622558881538872, + "grad_norm": 3.920647144317627, + "learning_rate": 7.474194332950602e-08, + "loss": 0.08188680410385132, + "step": 36530 + }, + { + "epoch": 0.9625193033983859, + "grad_norm": 3.1353626251220703, + "learning_rate": 7.370442997780114e-08, + "loss": 0.07856382131576538, + "step": 36540 + }, + { + "epoch": 0.9627827186428847, + "grad_norm": 2.3608505725860596, + "learning_rate": 7.267414141227002e-08, + "loss": 0.07214556336402893, + "step": 36550 + }, + { + "epoch": 0.9630461338873834, + "grad_norm": 2.985487937927246, + "learning_rate": 7.165107838279838e-08, + "loss": 0.08765004873275757, + "step": 36560 + }, + { + "epoch": 0.9633095491318822, + "grad_norm": 3.393420696258545, + "learning_rate": 7.063524163401836e-08, + "loss": 0.0862686038017273, + "step": 36570 + }, + { + "epoch": 0.9635729643763808, + "grad_norm": 4.5645880699157715, + "learning_rate": 6.962663190529851e-08, + "loss": 0.08534841537475586, + "step": 36580 + }, + { + "epoch": 0.9638363796208796, + "grad_norm": 3.4809412956237793, + "learning_rate": 6.862524993074825e-08, + "loss": 0.07310707569122314, + "step": 36590 + }, + { + "epoch": 0.9640997948653783, + "grad_norm": 4.459349155426025, + "learning_rate": 6.763109643921683e-08, + "loss": 0.07446432709693909, + "step": 36600 + }, + { + "epoch": 0.9643632101098771, + "grad_norm": 2.702246904373169, + "learning_rate": 6.664417215429319e-08, + "loss": 0.0904299795627594, + "step": 36610 + }, + { + "epoch": 0.9646266253543758, + "grad_norm": 3.412749767303467, + "learning_rate": 6.566447779430163e-08, + "loss": 0.08332064151763915, + "step": 36620 + }, + { + "epoch": 0.9648900405988745, + "grad_norm": 2.9370551109313965, + "learning_rate": 6.469201407230729e-08, + "loss": 0.0853144347667694, + "step": 36630 + }, + { + "epoch": 0.9651534558433733, + "grad_norm": 4.072447299957275, + "learning_rate": 6.372678169611068e-08, + "loss": 0.08554071784019471, + "step": 36640 + }, + { + "epoch": 0.965416871087872, + "grad_norm": 3.9238545894622803, + "learning_rate": 6.276878136824982e-08, + "loss": 0.08157247304916382, + "step": 36650 + }, + { + "epoch": 0.9656802863323708, + "grad_norm": 3.971055746078491, + "learning_rate": 6.181801378599805e-08, + "loss": 0.07479498386383057, + "step": 36660 + }, + { + "epoch": 0.9659437015768695, + "grad_norm": 3.1646523475646973, + "learning_rate": 6.087447964136517e-08, + "loss": 0.07456987500190734, + "step": 36670 + }, + { + "epoch": 0.9662071168213683, + "grad_norm": 3.8882508277893066, + "learning_rate": 5.993817962109627e-08, + "loss": 0.09351956844329834, + "step": 36680 + }, + { + "epoch": 0.966470532065867, + "grad_norm": 3.752134323120117, + "learning_rate": 5.9009114406671787e-08, + "loss": 0.08450908660888672, + "step": 36690 + }, + { + "epoch": 0.9667339473103658, + "grad_norm": 3.4013006687164307, + "learning_rate": 5.8087284674304136e-08, + "loss": 0.08548129200935364, + "step": 36700 + }, + { + "epoch": 0.9669973625548645, + "grad_norm": 3.795179843902588, + "learning_rate": 5.7172691094942165e-08, + "loss": 0.07082120180130005, + "step": 36710 + }, + { + "epoch": 0.9672607777993631, + "grad_norm": 3.1464622020721436, + "learning_rate": 5.626533433426451e-08, + "loss": 0.08485739827156066, + "step": 36720 + }, + { + "epoch": 0.967524193043862, + "grad_norm": 3.7063071727752686, + "learning_rate": 5.5365215052686216e-08, + "loss": 0.08527722358703613, + "step": 36730 + }, + { + "epoch": 0.9677876082883606, + "grad_norm": 3.2720746994018555, + "learning_rate": 5.4472333905352115e-08, + "loss": 0.0766224205493927, + "step": 36740 + }, + { + "epoch": 0.9680510235328594, + "grad_norm": 3.035940647125244, + "learning_rate": 5.3586691542140135e-08, + "loss": 0.07074192762374878, + "step": 36750 + }, + { + "epoch": 0.9683144387773581, + "grad_norm": 3.073568344116211, + "learning_rate": 5.2708288607659084e-08, + "loss": 0.07084767818450928, + "step": 36760 + }, + { + "epoch": 0.9685778540218569, + "grad_norm": 3.9952454566955566, + "learning_rate": 5.183712574124755e-08, + "loss": 0.078428453207016, + "step": 36770 + }, + { + "epoch": 0.9688412692663556, + "grad_norm": 3.3293302059173584, + "learning_rate": 5.0973203576976104e-08, + "loss": 0.0775246262550354, + "step": 36780 + }, + { + "epoch": 0.9691046845108544, + "grad_norm": 3.8461673259735107, + "learning_rate": 5.011652274364509e-08, + "loss": 0.08336674571037292, + "step": 36790 + }, + { + "epoch": 0.9693680997553531, + "grad_norm": 3.9163217544555664, + "learning_rate": 4.926708386478352e-08, + "loss": 0.07140924930572509, + "step": 36800 + }, + { + "epoch": 0.9696315149998518, + "grad_norm": 3.380279541015625, + "learning_rate": 4.8424887558649045e-08, + "loss": 0.08106863498687744, + "step": 36810 + }, + { + "epoch": 0.9698949302443506, + "grad_norm": 2.2135894298553467, + "learning_rate": 4.7589934438228017e-08, + "loss": 0.08188547492027283, + "step": 36820 + }, + { + "epoch": 0.9701583454888493, + "grad_norm": 3.461358070373535, + "learning_rate": 4.676222511123762e-08, + "loss": 0.09048053622245789, + "step": 36830 + }, + { + "epoch": 0.9704217607333481, + "grad_norm": 4.639743804931641, + "learning_rate": 4.59417601801182e-08, + "loss": 0.09762736558914184, + "step": 36840 + }, + { + "epoch": 0.9706851759778468, + "grad_norm": 3.442808151245117, + "learning_rate": 4.512854024203983e-08, + "loss": 0.10220340490341187, + "step": 36850 + }, + { + "epoch": 0.9709485912223456, + "grad_norm": 2.256014585494995, + "learning_rate": 4.432256588890016e-08, + "loss": 0.07872591018676758, + "step": 36860 + }, + { + "epoch": 0.9712120064668442, + "grad_norm": 3.3270103931427, + "learning_rate": 4.352383770732216e-08, + "loss": 0.07538859248161316, + "step": 36870 + }, + { + "epoch": 0.9714754217113429, + "grad_norm": 4.60349178314209, + "learning_rate": 4.273235627865302e-08, + "loss": 0.07270563244819642, + "step": 36880 + }, + { + "epoch": 0.9717388369558417, + "grad_norm": 2.6745288372039795, + "learning_rate": 4.19481221789686e-08, + "loss": 0.08623796105384826, + "step": 36890 + }, + { + "epoch": 0.9720022522003404, + "grad_norm": 3.3050827980041504, + "learning_rate": 4.117113597906675e-08, + "loss": 0.0770674467086792, + "step": 36900 + }, + { + "epoch": 0.9722656674448392, + "grad_norm": 3.156118154525757, + "learning_rate": 4.040139824447398e-08, + "loss": 0.07851160764694214, + "step": 36910 + }, + { + "epoch": 0.9725290826893379, + "grad_norm": 3.6100010871887207, + "learning_rate": 3.963890953543659e-08, + "loss": 0.0853744089603424, + "step": 36920 + }, + { + "epoch": 0.9727924979338367, + "grad_norm": 3.411668539047241, + "learning_rate": 3.88836704069262e-08, + "loss": 0.0797040581703186, + "step": 36930 + }, + { + "epoch": 0.9730559131783354, + "grad_norm": 3.192444324493408, + "learning_rate": 3.813568140864088e-08, + "loss": 0.08731818795204163, + "step": 36940 + }, + { + "epoch": 0.9733193284228342, + "grad_norm": 3.726130485534668, + "learning_rate": 3.739494308499736e-08, + "loss": 0.09442147016525268, + "step": 36950 + }, + { + "epoch": 0.9735827436673329, + "grad_norm": 3.273164987564087, + "learning_rate": 3.66614559751377e-08, + "loss": 0.08505194187164307, + "step": 36960 + }, + { + "epoch": 0.9738461589118316, + "grad_norm": 3.6091010570526123, + "learning_rate": 3.593522061292598e-08, + "loss": 0.06708827614784241, + "step": 36970 + }, + { + "epoch": 0.9741095741563304, + "grad_norm": 4.1784868240356445, + "learning_rate": 3.521623752694603e-08, + "loss": 0.08353699445724487, + "step": 36980 + }, + { + "epoch": 0.9743729894008291, + "grad_norm": 2.719046115875244, + "learning_rate": 3.4504507240507027e-08, + "loss": 0.07681536674499512, + "step": 36990 + }, + { + "epoch": 0.9746364046453279, + "grad_norm": 3.293401002883911, + "learning_rate": 3.380003027163459e-08, + "loss": 0.07417883872985839, + "step": 37000 + } + ], + "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": false + }, + "attributes": {} + } + }, + "total_flos": 1.9693793009551698e+20, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +}