{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 37963, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00026341524449873725, "grad_norm": 1216.858154296875, "learning_rate": 1.5803336259877089e-07, "loss": 2.372263717651367, "step": 10 }, { "epoch": 0.0005268304889974745, "grad_norm": 58.291683197021484, "learning_rate": 3.336259877085163e-07, "loss": 0.6805263996124268, "step": 20 }, { "epoch": 0.0007902457334962118, "grad_norm": 35.27565002441406, "learning_rate": 5.092186128182617e-07, "loss": 0.3130784511566162, "step": 30 }, { "epoch": 0.001053660977994949, "grad_norm": 16.055953979492188, "learning_rate": 6.848112379280071e-07, "loss": 0.2575071811676025, "step": 40 }, { "epoch": 0.0013170762224936862, "grad_norm": 21.586252212524414, "learning_rate": 8.604038630377525e-07, "loss": 0.26515803337097166, "step": 50 }, { "epoch": 0.0015804914669924235, "grad_norm": 14.836892127990723, "learning_rate": 1.035996488147498e-06, "loss": 0.2501066207885742, "step": 60 }, { "epoch": 0.0018439067114911607, "grad_norm": 15.245532035827637, "learning_rate": 1.2115891132572432e-06, "loss": 0.22771573066711426, "step": 70 }, { "epoch": 0.002107321955989898, "grad_norm": 10.363273620605469, "learning_rate": 1.3871817383669885e-06, "loss": 0.24467992782592773, "step": 80 }, { "epoch": 0.002370737200488635, "grad_norm": 13.13510513305664, "learning_rate": 1.5627743634767342e-06, "loss": 0.2221926212310791, "step": 90 }, { "epoch": 0.0026341524449873723, "grad_norm": 12.594985008239746, "learning_rate": 1.7383669885864795e-06, "loss": 0.22232375144958497, "step": 100 }, { "epoch": 0.00289756768948611, "grad_norm": 11.869989395141602, "learning_rate": 1.913959613696225e-06, "loss": 0.2270669937133789, "step": 110 }, { "epoch": 0.003160982933984847, "grad_norm": 19.094928741455078, "learning_rate": 2.08955223880597e-06, "loss": 0.24924917221069337, "step": 120 }, { "epoch": 0.003424398178483584, "grad_norm": 14.594099044799805, "learning_rate": 2.265144863915716e-06, "loss": 0.2181525707244873, "step": 130 }, { "epoch": 0.0036878134229823213, "grad_norm": 16.19100570678711, "learning_rate": 2.440737489025461e-06, "loss": 0.22549042701721192, "step": 140 }, { "epoch": 0.0039512286674810585, "grad_norm": 8.812203407287598, "learning_rate": 2.6163301141352064e-06, "loss": 0.22353053092956543, "step": 150 }, { "epoch": 0.004214643911979796, "grad_norm": 17.371822357177734, "learning_rate": 2.791922739244952e-06, "loss": 0.2520279884338379, "step": 160 }, { "epoch": 0.004478059156478534, "grad_norm": 10.19724178314209, "learning_rate": 2.9675153643546974e-06, "loss": 0.24660415649414064, "step": 170 }, { "epoch": 0.00474147440097727, "grad_norm": 7.54756498336792, "learning_rate": 3.1431079894644427e-06, "loss": 0.24439642429351807, "step": 180 }, { "epoch": 0.005004889645476008, "grad_norm": 7.050674915313721, "learning_rate": 3.318700614574188e-06, "loss": 0.2212231159210205, "step": 190 }, { "epoch": 0.005268304889974745, "grad_norm": 13.394369125366211, "learning_rate": 3.4942932396839337e-06, "loss": 0.23461651802062988, "step": 200 }, { "epoch": 0.005531720134473482, "grad_norm": 17.462080001831055, "learning_rate": 3.669885864793679e-06, "loss": 0.22870469093322754, "step": 210 }, { "epoch": 0.00579513537897222, "grad_norm": 26.927310943603516, "learning_rate": 3.845478489903424e-06, "loss": 0.21369614601135253, "step": 220 }, { "epoch": 0.0060585506234709565, "grad_norm": 14.947160720825195, "learning_rate": 4.0210711150131695e-06, "loss": 0.21966190338134767, "step": 230 }, { "epoch": 0.006321965867969694, "grad_norm": 12.736027717590332, "learning_rate": 4.196663740122916e-06, "loss": 0.24580941200256348, "step": 240 }, { "epoch": 0.006585381112468432, "grad_norm": 9.799609184265137, "learning_rate": 4.37225636523266e-06, "loss": 0.19318530559539795, "step": 250 }, { "epoch": 0.006848796356967168, "grad_norm": 9.492569923400879, "learning_rate": 4.547848990342406e-06, "loss": 0.22372145652770997, "step": 260 }, { "epoch": 0.007112211601465906, "grad_norm": 7.549798011779785, "learning_rate": 4.7234416154521515e-06, "loss": 0.20140402317047118, "step": 270 }, { "epoch": 0.007375626845964643, "grad_norm": 13.578999519348145, "learning_rate": 4.899034240561897e-06, "loss": 0.20918593406677247, "step": 280 }, { "epoch": 0.00763904209046338, "grad_norm": 9.166558265686035, "learning_rate": 5.074626865671642e-06, "loss": 0.1972205400466919, "step": 290 }, { "epoch": 0.007902457334962117, "grad_norm": 8.107062339782715, "learning_rate": 5.250219490781387e-06, "loss": 0.20403773784637452, "step": 300 }, { "epoch": 0.008165872579460855, "grad_norm": 8.722216606140137, "learning_rate": 5.425812115891133e-06, "loss": 0.2011693000793457, "step": 310 }, { "epoch": 0.008429287823959592, "grad_norm": 12.69148063659668, "learning_rate": 5.601404741000879e-06, "loss": 0.2078859329223633, "step": 320 }, { "epoch": 0.008692703068458329, "grad_norm": 7.064239978790283, "learning_rate": 5.776997366110624e-06, "loss": 0.21744611263275146, "step": 330 }, { "epoch": 0.008956118312957067, "grad_norm": 11.75399112701416, "learning_rate": 5.9525899912203685e-06, "loss": 0.20960135459899903, "step": 340 }, { "epoch": 0.009219533557455804, "grad_norm": 8.743265151977539, "learning_rate": 6.128182616330115e-06, "loss": 0.22711172103881835, "step": 350 }, { "epoch": 0.00948294880195454, "grad_norm": 7.5478668212890625, "learning_rate": 6.30377524143986e-06, "loss": 0.21109576225280763, "step": 360 }, { "epoch": 0.00974636404645328, "grad_norm": 6.639023303985596, "learning_rate": 6.479367866549605e-06, "loss": 0.21695730686187745, "step": 370 }, { "epoch": 0.010009779290952016, "grad_norm": 9.527194023132324, "learning_rate": 6.654960491659351e-06, "loss": 0.21946520805358888, "step": 380 }, { "epoch": 0.010273194535450753, "grad_norm": 12.758036613464355, "learning_rate": 6.830553116769097e-06, "loss": 0.21301827430725098, "step": 390 }, { "epoch": 0.01053660977994949, "grad_norm": 6.767460346221924, "learning_rate": 7.006145741878841e-06, "loss": 0.20237975120544432, "step": 400 }, { "epoch": 0.010800025024448228, "grad_norm": 5.646766662597656, "learning_rate": 7.181738366988587e-06, "loss": 0.19278799295425414, "step": 410 }, { "epoch": 0.011063440268946964, "grad_norm": 6.160284996032715, "learning_rate": 7.3573309920983325e-06, "loss": 0.2051539897918701, "step": 420 }, { "epoch": 0.011326855513445701, "grad_norm": 14.325827598571777, "learning_rate": 7.532923617208078e-06, "loss": 0.19082332849502565, "step": 430 }, { "epoch": 0.01159027075794444, "grad_norm": 4.854583263397217, "learning_rate": 7.708516242317823e-06, "loss": 0.1994854688644409, "step": 440 }, { "epoch": 0.011853686002443176, "grad_norm": 8.881163597106934, "learning_rate": 7.88410886742757e-06, "loss": 0.20647990703582764, "step": 450 }, { "epoch": 0.012117101246941913, "grad_norm": 7.325456142425537, "learning_rate": 8.059701492537314e-06, "loss": 0.1984945058822632, "step": 460 }, { "epoch": 0.012380516491440651, "grad_norm": 4.576322555541992, "learning_rate": 8.23529411764706e-06, "loss": 0.1902897596359253, "step": 470 }, { "epoch": 0.012643931735939388, "grad_norm": 6.590148448944092, "learning_rate": 8.410886742756806e-06, "loss": 0.2038738250732422, "step": 480 }, { "epoch": 0.012907346980438125, "grad_norm": 9.073351860046387, "learning_rate": 8.58647936786655e-06, "loss": 0.20971860885620117, "step": 490 }, { "epoch": 0.013170762224936863, "grad_norm": 9.728736877441406, "learning_rate": 8.762071992976295e-06, "loss": 0.23039441108703612, "step": 500 }, { "epoch": 0.0134341774694356, "grad_norm": 8.736261367797852, "learning_rate": 8.937664618086041e-06, "loss": 0.2229452610015869, "step": 510 }, { "epoch": 0.013697592713934337, "grad_norm": 6.253108978271484, "learning_rate": 9.113257243195787e-06, "loss": 0.2116520881652832, "step": 520 }, { "epoch": 0.013961007958433075, "grad_norm": 5.372574329376221, "learning_rate": 9.288849868305532e-06, "loss": 0.2011418342590332, "step": 530 }, { "epoch": 0.014224423202931812, "grad_norm": 10.353477478027344, "learning_rate": 9.464442493415278e-06, "loss": 0.20820722579956055, "step": 540 }, { "epoch": 0.014487838447430549, "grad_norm": 3.489136219024658, "learning_rate": 9.640035118525024e-06, "loss": 0.199906063079834, "step": 550 }, { "epoch": 0.014751253691929285, "grad_norm": 7.704500675201416, "learning_rate": 9.815627743634768e-06, "loss": 0.19410412311553954, "step": 560 }, { "epoch": 0.015014668936428024, "grad_norm": 4.957610607147217, "learning_rate": 9.991220368744513e-06, "loss": 0.2025153160095215, "step": 570 }, { "epoch": 0.01527808418092676, "grad_norm": 6.147917747497559, "learning_rate": 1.0166812993854259e-05, "loss": 0.2030883550643921, "step": 580 }, { "epoch": 0.015541499425425497, "grad_norm": 8.839682579040527, "learning_rate": 1.0342405618964005e-05, "loss": 0.1908982515335083, "step": 590 }, { "epoch": 0.015804914669924234, "grad_norm": 10.765022277832031, "learning_rate": 1.0517998244073751e-05, "loss": 0.20895752906799317, "step": 600 }, { "epoch": 0.016068329914422974, "grad_norm": 4.1301188468933105, "learning_rate": 1.0693590869183494e-05, "loss": 0.2104104995727539, "step": 610 }, { "epoch": 0.01633174515892171, "grad_norm": 5.00321626663208, "learning_rate": 1.086918349429324e-05, "loss": 0.19376648664474488, "step": 620 }, { "epoch": 0.016595160403420448, "grad_norm": 10.098116874694824, "learning_rate": 1.1044776119402986e-05, "loss": 0.2220301389694214, "step": 630 }, { "epoch": 0.016858575647919184, "grad_norm": 8.164828300476074, "learning_rate": 1.122036874451273e-05, "loss": 0.186989164352417, "step": 640 }, { "epoch": 0.01712199089241792, "grad_norm": 6.0607500076293945, "learning_rate": 1.1395961369622477e-05, "loss": 0.1909113883972168, "step": 650 }, { "epoch": 0.017385406136916658, "grad_norm": 4.371347427368164, "learning_rate": 1.1571553994732223e-05, "loss": 0.214566969871521, "step": 660 }, { "epoch": 0.017648821381415394, "grad_norm": 5.701393127441406, "learning_rate": 1.1747146619841967e-05, "loss": 0.20785298347473144, "step": 670 }, { "epoch": 0.017912236625914135, "grad_norm": 6.087163925170898, "learning_rate": 1.1922739244951713e-05, "loss": 0.1986930251121521, "step": 680 }, { "epoch": 0.01817565187041287, "grad_norm": 9.385022163391113, "learning_rate": 1.209833187006146e-05, "loss": 0.22220907211303711, "step": 690 }, { "epoch": 0.018439067114911608, "grad_norm": 9.611412048339844, "learning_rate": 1.2273924495171204e-05, "loss": 0.21005072593688964, "step": 700 }, { "epoch": 0.018702482359410345, "grad_norm": 7.795034885406494, "learning_rate": 1.244951712028095e-05, "loss": 0.1845225214958191, "step": 710 }, { "epoch": 0.01896589760390908, "grad_norm": 6.859914302825928, "learning_rate": 1.2625109745390696e-05, "loss": 0.19467897415161134, "step": 720 }, { "epoch": 0.019229312848407818, "grad_norm": 5.119071960449219, "learning_rate": 1.2800702370500439e-05, "loss": 0.19577403068542482, "step": 730 }, { "epoch": 0.01949272809290656, "grad_norm": 8.040349960327148, "learning_rate": 1.2976294995610185e-05, "loss": 0.20361974239349365, "step": 740 }, { "epoch": 0.019756143337405295, "grad_norm": 6.102715492248535, "learning_rate": 1.315188762071993e-05, "loss": 0.2220911979675293, "step": 750 }, { "epoch": 0.02001955858190403, "grad_norm": 6.165253162384033, "learning_rate": 1.3327480245829676e-05, "loss": 0.20718011856079102, "step": 760 }, { "epoch": 0.02028297382640277, "grad_norm": 8.203484535217285, "learning_rate": 1.3503072870939422e-05, "loss": 0.21474499702453614, "step": 770 }, { "epoch": 0.020546389070901505, "grad_norm": 4.87082052230835, "learning_rate": 1.3678665496049166e-05, "loss": 0.20563530921936035, "step": 780 }, { "epoch": 0.020809804315400242, "grad_norm": 7.674990177154541, "learning_rate": 1.3854258121158912e-05, "loss": 0.19210031032562255, "step": 790 }, { "epoch": 0.02107321955989898, "grad_norm": 14.870322227478027, "learning_rate": 1.4029850746268658e-05, "loss": 0.214613676071167, "step": 800 }, { "epoch": 0.02133663480439772, "grad_norm": 4.547909259796143, "learning_rate": 1.4205443371378403e-05, "loss": 0.1957342505455017, "step": 810 }, { "epoch": 0.021600050048896455, "grad_norm": 3.57409405708313, "learning_rate": 1.4381035996488149e-05, "loss": 0.18840723037719725, "step": 820 }, { "epoch": 0.021863465293395192, "grad_norm": 3.5035839080810547, "learning_rate": 1.4556628621597895e-05, "loss": 0.1914868474006653, "step": 830 }, { "epoch": 0.02212688053789393, "grad_norm": 3.8623862266540527, "learning_rate": 1.4732221246707638e-05, "loss": 0.1913320779800415, "step": 840 }, { "epoch": 0.022390295782392666, "grad_norm": 7.156163215637207, "learning_rate": 1.4907813871817384e-05, "loss": 0.19116511344909667, "step": 850 }, { "epoch": 0.022653711026891402, "grad_norm": 5.988685131072998, "learning_rate": 1.508340649692713e-05, "loss": 0.19063524007797242, "step": 860 }, { "epoch": 0.022917126271390142, "grad_norm": 4.694153308868408, "learning_rate": 1.5258999122036876e-05, "loss": 0.20948052406311035, "step": 870 }, { "epoch": 0.02318054151588888, "grad_norm": 5.238402843475342, "learning_rate": 1.5434591747146622e-05, "loss": 0.21998288631439208, "step": 880 }, { "epoch": 0.023443956760387616, "grad_norm": 5.852624893188477, "learning_rate": 1.561018437225637e-05, "loss": 0.19518139362335205, "step": 890 }, { "epoch": 0.023707372004886353, "grad_norm": 4.7149200439453125, "learning_rate": 1.578577699736611e-05, "loss": 0.18113348484039307, "step": 900 }, { "epoch": 0.02397078724938509, "grad_norm": 3.8622028827667236, "learning_rate": 1.5961369622475857e-05, "loss": 0.19703441858291626, "step": 910 }, { "epoch": 0.024234202493883826, "grad_norm": 11.124042510986328, "learning_rate": 1.6136962247585604e-05, "loss": 0.2054431200027466, "step": 920 }, { "epoch": 0.024497617738382566, "grad_norm": 3.727165699005127, "learning_rate": 1.6312554872695346e-05, "loss": 0.20793991088867186, "step": 930 }, { "epoch": 0.024761032982881303, "grad_norm": 3.1881048679351807, "learning_rate": 1.6488147497805093e-05, "loss": 0.19128236770629883, "step": 940 }, { "epoch": 0.02502444822738004, "grad_norm": 4.896829128265381, "learning_rate": 1.666374012291484e-05, "loss": 0.18748389482498168, "step": 950 }, { "epoch": 0.025287863471878776, "grad_norm": 4.7075066566467285, "learning_rate": 1.6839332748024585e-05, "loss": 0.1943873643875122, "step": 960 }, { "epoch": 0.025551278716377513, "grad_norm": 2.350215196609497, "learning_rate": 1.701492537313433e-05, "loss": 0.18150179386138915, "step": 970 }, { "epoch": 0.02581469396087625, "grad_norm": 9.136993408203125, "learning_rate": 1.7190517998244077e-05, "loss": 0.2112889766693115, "step": 980 }, { "epoch": 0.026078109205374986, "grad_norm": 3.859117269515991, "learning_rate": 1.736611062335382e-05, "loss": 0.18779940605163575, "step": 990 }, { "epoch": 0.026341524449873727, "grad_norm": 5.580035209655762, "learning_rate": 1.7541703248463566e-05, "loss": 0.19592103958129883, "step": 1000 }, { "epoch": 0.026604939694372463, "grad_norm": 4.203274726867676, "learning_rate": 1.771729587357331e-05, "loss": 0.20215299129486083, "step": 1010 }, { "epoch": 0.0268683549388712, "grad_norm": 7.283274173736572, "learning_rate": 1.7892888498683055e-05, "loss": 0.21270380020141602, "step": 1020 }, { "epoch": 0.027131770183369937, "grad_norm": 4.384298801422119, "learning_rate": 1.80684811237928e-05, "loss": 0.1900118589401245, "step": 1030 }, { "epoch": 0.027395185427868674, "grad_norm": 3.447072982788086, "learning_rate": 1.8244073748902547e-05, "loss": 0.2004077434539795, "step": 1040 }, { "epoch": 0.02765860067236741, "grad_norm": 2.647733211517334, "learning_rate": 1.8419666374012293e-05, "loss": 0.18839190006256104, "step": 1050 }, { "epoch": 0.02792201591686615, "grad_norm": 3.636383533477783, "learning_rate": 1.859525899912204e-05, "loss": 0.19927127361297609, "step": 1060 }, { "epoch": 0.028185431161364887, "grad_norm": 3.881094217300415, "learning_rate": 1.8770851624231782e-05, "loss": 0.18835257291793822, "step": 1070 }, { "epoch": 0.028448846405863624, "grad_norm": 4.69539737701416, "learning_rate": 1.8946444249341528e-05, "loss": 0.21528162956237792, "step": 1080 }, { "epoch": 0.02871226165036236, "grad_norm": 6.701592445373535, "learning_rate": 1.9122036874451274e-05, "loss": 0.21031253337860106, "step": 1090 }, { "epoch": 0.028975676894861097, "grad_norm": 3.599555492401123, "learning_rate": 1.929762949956102e-05, "loss": 0.19361751079559325, "step": 1100 }, { "epoch": 0.029239092139359834, "grad_norm": 4.312601566314697, "learning_rate": 1.9473222124670767e-05, "loss": 0.20636086463928222, "step": 1110 }, { "epoch": 0.02950250738385857, "grad_norm": 4.312938213348389, "learning_rate": 1.9648814749780513e-05, "loss": 0.18446682691574096, "step": 1120 }, { "epoch": 0.02976592262835731, "grad_norm": 5.664150714874268, "learning_rate": 1.9824407374890256e-05, "loss": 0.19502415657043456, "step": 1130 }, { "epoch": 0.030029337872856048, "grad_norm": 2.9939327239990234, "learning_rate": 2e-05, "loss": 0.1956365704536438, "step": 1140 }, { "epoch": 0.030292753117354784, "grad_norm": 7.477921009063721, "learning_rate": 1.9999996360784282e-05, "loss": 0.1976664900779724, "step": 1150 }, { "epoch": 0.03055616836185352, "grad_norm": 4.456860542297363, "learning_rate": 1.9999985443139778e-05, "loss": 0.1880091905593872, "step": 1160 }, { "epoch": 0.030819583606352258, "grad_norm": 3.214022159576416, "learning_rate": 1.999996724707443e-05, "loss": 0.19554107189178466, "step": 1170 }, { "epoch": 0.031082998850850994, "grad_norm": 3.645238161087036, "learning_rate": 1.9999941772601482e-05, "loss": 0.18527443408966066, "step": 1180 }, { "epoch": 0.03134641409534973, "grad_norm": 2.852123260498047, "learning_rate": 1.9999909019739482e-05, "loss": 0.2074648380279541, "step": 1190 }, { "epoch": 0.03160982933984847, "grad_norm": 5.052933692932129, "learning_rate": 1.9999868988512265e-05, "loss": 0.18114173412322998, "step": 1200 }, { "epoch": 0.031873244584347205, "grad_norm": 12.089102745056152, "learning_rate": 1.9999821678948964e-05, "loss": 0.1952626347541809, "step": 1210 }, { "epoch": 0.03213665982884595, "grad_norm": 6.772907733917236, "learning_rate": 1.999976709108402e-05, "loss": 0.20682845115661622, "step": 1220 }, { "epoch": 0.032400075073344685, "grad_norm": 7.1226935386657715, "learning_rate": 1.9999705224957156e-05, "loss": 0.21533613204956054, "step": 1230 }, { "epoch": 0.03266349031784342, "grad_norm": 11.651985168457031, "learning_rate": 1.9999636080613406e-05, "loss": 0.21128971576690675, "step": 1240 }, { "epoch": 0.03292690556234216, "grad_norm": 4.417197227478027, "learning_rate": 1.9999559658103097e-05, "loss": 0.19935188293457032, "step": 1250 }, { "epoch": 0.033190320806840895, "grad_norm": 4.602780342102051, "learning_rate": 1.9999475957481848e-05, "loss": 0.20724015235900878, "step": 1260 }, { "epoch": 0.03345373605133963, "grad_norm": 1.7654412984848022, "learning_rate": 1.999938497881059e-05, "loss": 0.2055220127105713, "step": 1270 }, { "epoch": 0.03371715129583837, "grad_norm": 3.016294479370117, "learning_rate": 1.999928672215553e-05, "loss": 0.1855383038520813, "step": 1280 }, { "epoch": 0.033980566540337105, "grad_norm": 3.1926941871643066, "learning_rate": 1.9999181187588187e-05, "loss": 0.21109287738800048, "step": 1290 }, { "epoch": 0.03424398178483584, "grad_norm": 12.131499290466309, "learning_rate": 1.9999068375185372e-05, "loss": 0.1965908408164978, "step": 1300 }, { "epoch": 0.03450739702933458, "grad_norm": 7.6271162033081055, "learning_rate": 1.99989482850292e-05, "loss": 0.2013707160949707, "step": 1310 }, { "epoch": 0.034770812273833315, "grad_norm": 2.8277170658111572, "learning_rate": 1.9998820917207076e-05, "loss": 0.19567304849624634, "step": 1320 }, { "epoch": 0.03503422751833205, "grad_norm": 2.3858845233917236, "learning_rate": 1.99986862718117e-05, "loss": 0.19793484210968018, "step": 1330 }, { "epoch": 0.03529764276283079, "grad_norm": 4.106758117675781, "learning_rate": 1.999854434894108e-05, "loss": 0.20077452659606934, "step": 1340 }, { "epoch": 0.03556105800732953, "grad_norm": 4.606701374053955, "learning_rate": 1.9998395148698505e-05, "loss": 0.2005112886428833, "step": 1350 }, { "epoch": 0.03582447325182827, "grad_norm": 11.171460151672363, "learning_rate": 1.9998238671192575e-05, "loss": 0.19315193891525267, "step": 1360 }, { "epoch": 0.036087888496327006, "grad_norm": 5.521824836730957, "learning_rate": 1.999807491653718e-05, "loss": 0.19576207399368287, "step": 1370 }, { "epoch": 0.03635130374082574, "grad_norm": 4.200732231140137, "learning_rate": 1.9997903884851503e-05, "loss": 0.18546109199523925, "step": 1380 }, { "epoch": 0.03661471898532448, "grad_norm": 2.621633768081665, "learning_rate": 1.999772557626004e-05, "loss": 0.1919564962387085, "step": 1390 }, { "epoch": 0.036878134229823216, "grad_norm": 3.527102470397949, "learning_rate": 1.9997539990892555e-05, "loss": 0.18324241638183594, "step": 1400 }, { "epoch": 0.03714154947432195, "grad_norm": 4.960463523864746, "learning_rate": 1.9997347128884138e-05, "loss": 0.19865188598632813, "step": 1410 }, { "epoch": 0.03740496471882069, "grad_norm": 1.776876449584961, "learning_rate": 1.9997146990375162e-05, "loss": 0.2116234302520752, "step": 1420 }, { "epoch": 0.037668379963319426, "grad_norm": 4.12790584564209, "learning_rate": 1.999693957551129e-05, "loss": 0.18611186742782593, "step": 1430 }, { "epoch": 0.03793179520781816, "grad_norm": 6.804045677185059, "learning_rate": 1.999672488444349e-05, "loss": 0.17933913469314575, "step": 1440 }, { "epoch": 0.0381952104523169, "grad_norm": 3.4403610229492188, "learning_rate": 1.999650291732802e-05, "loss": 0.2063199520111084, "step": 1450 }, { "epoch": 0.038458625696815636, "grad_norm": 3.17010760307312, "learning_rate": 1.9996273674326446e-05, "loss": 0.19650416374206542, "step": 1460 }, { "epoch": 0.03872204094131437, "grad_norm": 2.8838183879852295, "learning_rate": 1.9996037155605615e-05, "loss": 0.18679815530776978, "step": 1470 }, { "epoch": 0.03898545618581312, "grad_norm": 4.627749919891357, "learning_rate": 1.9995793361337676e-05, "loss": 0.1966300129890442, "step": 1480 }, { "epoch": 0.03924887143031185, "grad_norm": 5.271109104156494, "learning_rate": 1.9995542291700075e-05, "loss": 0.19165982007980348, "step": 1490 }, { "epoch": 0.03951228667481059, "grad_norm": 6.724852085113525, "learning_rate": 1.9995283946875548e-05, "loss": 0.18388543128967286, "step": 1500 }, { "epoch": 0.03977570191930933, "grad_norm": 5.314005374908447, "learning_rate": 1.999501832705213e-05, "loss": 0.1727055788040161, "step": 1510 }, { "epoch": 0.04003911716380806, "grad_norm": 2.0413990020751953, "learning_rate": 1.9994745432423154e-05, "loss": 0.18105911016464232, "step": 1520 }, { "epoch": 0.0403025324083068, "grad_norm": 6.113137722015381, "learning_rate": 1.9994465263187242e-05, "loss": 0.1780048727989197, "step": 1530 }, { "epoch": 0.04056594765280554, "grad_norm": 6.594235897064209, "learning_rate": 1.9994177819548313e-05, "loss": 0.20702047348022462, "step": 1540 }, { "epoch": 0.040829362897304274, "grad_norm": 4.571872234344482, "learning_rate": 1.999388310171558e-05, "loss": 0.19601716995239257, "step": 1550 }, { "epoch": 0.04109277814180301, "grad_norm": 6.710470199584961, "learning_rate": 1.9993581109903557e-05, "loss": 0.17355554103851317, "step": 1560 }, { "epoch": 0.04135619338630175, "grad_norm": 6.709134578704834, "learning_rate": 1.999327184433204e-05, "loss": 0.1859331965446472, "step": 1570 }, { "epoch": 0.041619608630800484, "grad_norm": 3.222343921661377, "learning_rate": 1.9992955305226127e-05, "loss": 0.1865220069885254, "step": 1580 }, { "epoch": 0.04188302387529922, "grad_norm": 2.6437699794769287, "learning_rate": 1.9992631492816214e-05, "loss": 0.19359207153320312, "step": 1590 }, { "epoch": 0.04214643911979796, "grad_norm": 1.6836724281311035, "learning_rate": 1.9992300407337977e-05, "loss": 0.2042041540145874, "step": 1600 }, { "epoch": 0.0424098543642967, "grad_norm": 7.666203022003174, "learning_rate": 1.9991962049032403e-05, "loss": 0.1845729112625122, "step": 1610 }, { "epoch": 0.04267326960879544, "grad_norm": 6.829221725463867, "learning_rate": 1.9991616418145757e-05, "loss": 0.19091721773147582, "step": 1620 }, { "epoch": 0.042936684853294174, "grad_norm": 3.5164124965667725, "learning_rate": 1.9991263514929606e-05, "loss": 0.1826985478401184, "step": 1630 }, { "epoch": 0.04320010009779291, "grad_norm": 7.890981197357178, "learning_rate": 1.999090333964081e-05, "loss": 0.1953359365463257, "step": 1640 }, { "epoch": 0.04346351534229165, "grad_norm": 2.073434352874756, "learning_rate": 1.9990535892541524e-05, "loss": 0.1782887101173401, "step": 1650 }, { "epoch": 0.043726930586790384, "grad_norm": 4.243154525756836, "learning_rate": 1.999016117389918e-05, "loss": 0.20971808433532715, "step": 1660 }, { "epoch": 0.04399034583128912, "grad_norm": 4.5022873878479, "learning_rate": 1.998977918398653e-05, "loss": 0.20120930671691895, "step": 1670 }, { "epoch": 0.04425376107578786, "grad_norm": 3.691859245300293, "learning_rate": 1.9989389923081587e-05, "loss": 0.17786755561828613, "step": 1680 }, { "epoch": 0.044517176320286594, "grad_norm": 3.1013376712799072, "learning_rate": 1.998899339146768e-05, "loss": 0.1793448209762573, "step": 1690 }, { "epoch": 0.04478059156478533, "grad_norm": 1.9693052768707275, "learning_rate": 1.9988589589433423e-05, "loss": 0.196533727645874, "step": 1700 }, { "epoch": 0.04504400680928407, "grad_norm": 5.608611106872559, "learning_rate": 1.9988178517272717e-05, "loss": 0.19871165752410888, "step": 1710 }, { "epoch": 0.045307422053782805, "grad_norm": 3.2578063011169434, "learning_rate": 1.998776017528476e-05, "loss": 0.1923875093460083, "step": 1720 }, { "epoch": 0.04557083729828155, "grad_norm": 4.231794834136963, "learning_rate": 1.9987334563774038e-05, "loss": 0.1948142647743225, "step": 1730 }, { "epoch": 0.045834252542780285, "grad_norm": 5.5231804847717285, "learning_rate": 1.998690168305033e-05, "loss": 0.18305881023406984, "step": 1740 }, { "epoch": 0.04609766778727902, "grad_norm": 2.1611623764038086, "learning_rate": 1.9986461533428708e-05, "loss": 0.19997794628143312, "step": 1750 }, { "epoch": 0.04636108303177776, "grad_norm": 3.206763982772827, "learning_rate": 1.9986014115229525e-05, "loss": 0.19234614372253417, "step": 1760 }, { "epoch": 0.046624498276276495, "grad_norm": 2.1135056018829346, "learning_rate": 1.998555942877844e-05, "loss": 0.19878032207489013, "step": 1770 }, { "epoch": 0.04688791352077523, "grad_norm": 2.811004877090454, "learning_rate": 1.9985097474406387e-05, "loss": 0.19255785942077636, "step": 1780 }, { "epoch": 0.04715132876527397, "grad_norm": 3.90399432182312, "learning_rate": 1.99846282524496e-05, "loss": 0.18011757135391235, "step": 1790 }, { "epoch": 0.047414744009772705, "grad_norm": 4.001239776611328, "learning_rate": 1.9984151763249598e-05, "loss": 0.19527348279953002, "step": 1800 }, { "epoch": 0.04767815925427144, "grad_norm": 3.0939128398895264, "learning_rate": 1.9983668007153185e-05, "loss": 0.19197678565979004, "step": 1810 }, { "epoch": 0.04794157449877018, "grad_norm": 2.9096763134002686, "learning_rate": 1.998317698451247e-05, "loss": 0.18201600313186644, "step": 1820 }, { "epoch": 0.048204989743268915, "grad_norm": 3.691051721572876, "learning_rate": 1.9982678695684833e-05, "loss": 0.19060544967651366, "step": 1830 }, { "epoch": 0.04846840498776765, "grad_norm": 1.9144105911254883, "learning_rate": 1.998217314103295e-05, "loss": 0.193269681930542, "step": 1840 }, { "epoch": 0.04873182023226639, "grad_norm": 2.116708755493164, "learning_rate": 1.998166032092479e-05, "loss": 0.19523890018463136, "step": 1850 }, { "epoch": 0.04899523547676513, "grad_norm": 3.4366774559020996, "learning_rate": 1.99811402357336e-05, "loss": 0.1730170488357544, "step": 1860 }, { "epoch": 0.04925865072126387, "grad_norm": 3.611096143722534, "learning_rate": 1.9980612885837925e-05, "loss": 0.19901690483093262, "step": 1870 }, { "epoch": 0.049522065965762606, "grad_norm": 2.729809522628784, "learning_rate": 1.998007827162159e-05, "loss": 0.18706135749816893, "step": 1880 }, { "epoch": 0.04978548121026134, "grad_norm": 6.06675910949707, "learning_rate": 1.9979536393473715e-05, "loss": 0.19614908695220948, "step": 1890 }, { "epoch": 0.05004889645476008, "grad_norm": 3.3115999698638916, "learning_rate": 1.9978987251788696e-05, "loss": 0.18416833877563477, "step": 1900 }, { "epoch": 0.050312311699258816, "grad_norm": 5.498225688934326, "learning_rate": 1.997843084696623e-05, "loss": 0.19507501125335694, "step": 1910 }, { "epoch": 0.05057572694375755, "grad_norm": 5.157803535461426, "learning_rate": 1.9977867179411282e-05, "loss": 0.1633998155593872, "step": 1920 }, { "epoch": 0.05083914218825629, "grad_norm": 3.130387783050537, "learning_rate": 1.997729624953412e-05, "loss": 0.20046682357788087, "step": 1930 }, { "epoch": 0.051102557432755026, "grad_norm": 1.4384745359420776, "learning_rate": 1.997671805775029e-05, "loss": 0.19507005214691162, "step": 1940 }, { "epoch": 0.05136597267725376, "grad_norm": 6.55009126663208, "learning_rate": 1.9976132604480625e-05, "loss": 0.18887476921081542, "step": 1950 }, { "epoch": 0.0516293879217525, "grad_norm": 3.253206729888916, "learning_rate": 1.997553989015125e-05, "loss": 0.2038723945617676, "step": 1960 }, { "epoch": 0.051892803166251236, "grad_norm": 5.493484973907471, "learning_rate": 1.9974939915193557e-05, "loss": 0.17812100648880005, "step": 1970 }, { "epoch": 0.05215621841074997, "grad_norm": 2.7087044715881348, "learning_rate": 1.9974332680044237e-05, "loss": 0.19548389911651612, "step": 1980 }, { "epoch": 0.05241963365524872, "grad_norm": 2.520240545272827, "learning_rate": 1.9973718185145262e-05, "loss": 0.1738547444343567, "step": 1990 }, { "epoch": 0.05268304889974745, "grad_norm": 4.827975749969482, "learning_rate": 1.997309643094389e-05, "loss": 0.18644872903823853, "step": 2000 }, { "epoch": 0.05294646414424619, "grad_norm": 1.6542577743530273, "learning_rate": 1.997246741789266e-05, "loss": 0.18310902118682862, "step": 2010 }, { "epoch": 0.05320987938874493, "grad_norm": 4.345027446746826, "learning_rate": 1.99718311464494e-05, "loss": 0.2018181324005127, "step": 2020 }, { "epoch": 0.05347329463324366, "grad_norm": 3.2264015674591064, "learning_rate": 1.9971187617077204e-05, "loss": 0.18996564149856568, "step": 2030 }, { "epoch": 0.0537367098777424, "grad_norm": 3.7870612144470215, "learning_rate": 1.9970536830244466e-05, "loss": 0.1861804485321045, "step": 2040 }, { "epoch": 0.05400012512224114, "grad_norm": 2.755880117416382, "learning_rate": 1.9969878786424863e-05, "loss": 0.18633283376693727, "step": 2050 }, { "epoch": 0.054263540366739874, "grad_norm": 2.044163227081299, "learning_rate": 1.9969213486097335e-05, "loss": 0.1985015392303467, "step": 2060 }, { "epoch": 0.05452695561123861, "grad_norm": 6.059758186340332, "learning_rate": 1.996854092974613e-05, "loss": 0.1941745400428772, "step": 2070 }, { "epoch": 0.05479037085573735, "grad_norm": 4.195821285247803, "learning_rate": 1.9967861117860754e-05, "loss": 0.17809646129608153, "step": 2080 }, { "epoch": 0.055053786100236084, "grad_norm": 3.4521713256835938, "learning_rate": 1.9967174050936008e-05, "loss": 0.18540620803833008, "step": 2090 }, { "epoch": 0.05531720134473482, "grad_norm": 2.540928363800049, "learning_rate": 1.9966479729471966e-05, "loss": 0.19492077827453613, "step": 2100 }, { "epoch": 0.05558061658923356, "grad_norm": 5.30415678024292, "learning_rate": 1.9965778153973986e-05, "loss": 0.17823656797409057, "step": 2110 }, { "epoch": 0.0558440318337323, "grad_norm": 3.231379747390747, "learning_rate": 1.9965069324952703e-05, "loss": 0.1924843430519104, "step": 2120 }, { "epoch": 0.05610744707823104, "grad_norm": 7.249763011932373, "learning_rate": 1.9964353242924038e-05, "loss": 0.1980189085006714, "step": 2130 }, { "epoch": 0.056370862322729774, "grad_norm": 3.2448153495788574, "learning_rate": 1.9963629908409184e-05, "loss": 0.18364591598510743, "step": 2140 }, { "epoch": 0.05663427756722851, "grad_norm": 3.1840872764587402, "learning_rate": 1.9962899321934616e-05, "loss": 0.17144757509231567, "step": 2150 }, { "epoch": 0.05689769281172725, "grad_norm": 2.7897820472717285, "learning_rate": 1.9962161484032085e-05, "loss": 0.16851568222045898, "step": 2160 }, { "epoch": 0.057161108056225984, "grad_norm": 2.870610237121582, "learning_rate": 1.9961416395238623e-05, "loss": 0.1650080442428589, "step": 2170 }, { "epoch": 0.05742452330072472, "grad_norm": 3.162261724472046, "learning_rate": 1.9960664056096534e-05, "loss": 0.21215529441833497, "step": 2180 }, { "epoch": 0.05768793854522346, "grad_norm": 8.104076385498047, "learning_rate": 1.9959904467153408e-05, "loss": 0.20582923889160157, "step": 2190 }, { "epoch": 0.057951353789722194, "grad_norm": 1.8404182195663452, "learning_rate": 1.99591376289621e-05, "loss": 0.17758100032806395, "step": 2200 }, { "epoch": 0.05821476903422093, "grad_norm": 4.464293479919434, "learning_rate": 1.9958363542080758e-05, "loss": 0.19636000394821168, "step": 2210 }, { "epoch": 0.05847818427871967, "grad_norm": 2.418220043182373, "learning_rate": 1.9957582207072784e-05, "loss": 0.17413172721862794, "step": 2220 }, { "epoch": 0.058741599523218405, "grad_norm": 4.321222305297852, "learning_rate": 1.9956793624506875e-05, "loss": 0.19459298849105836, "step": 2230 }, { "epoch": 0.05900501476771714, "grad_norm": 4.644446849822998, "learning_rate": 1.9955997794956994e-05, "loss": 0.1730610966682434, "step": 2240 }, { "epoch": 0.059268430012215885, "grad_norm": 3.0418155193328857, "learning_rate": 1.995519471900238e-05, "loss": 0.18072221279144288, "step": 2250 }, { "epoch": 0.05953184525671462, "grad_norm": 1.782165288925171, "learning_rate": 1.9954384397227545e-05, "loss": 0.18120296001434327, "step": 2260 }, { "epoch": 0.05979526050121336, "grad_norm": 5.117288112640381, "learning_rate": 1.9953566830222276e-05, "loss": 0.17586038112640381, "step": 2270 }, { "epoch": 0.060058675745712095, "grad_norm": 3.8470613956451416, "learning_rate": 1.9952742018581634e-05, "loss": 0.19392603635787964, "step": 2280 }, { "epoch": 0.06032209099021083, "grad_norm": 5.0190300941467285, "learning_rate": 1.9951909962905956e-05, "loss": 0.17820771932601928, "step": 2290 }, { "epoch": 0.06058550623470957, "grad_norm": 5.736504077911377, "learning_rate": 1.9951070663800843e-05, "loss": 0.17343271970748902, "step": 2300 }, { "epoch": 0.060848921479208305, "grad_norm": 2.0889179706573486, "learning_rate": 1.995022412187718e-05, "loss": 0.17610299587249756, "step": 2310 }, { "epoch": 0.06111233672370704, "grad_norm": 4.063083648681641, "learning_rate": 1.9949370337751105e-05, "loss": 0.19263689517974852, "step": 2320 }, { "epoch": 0.06137575196820578, "grad_norm": 3.1233415603637695, "learning_rate": 1.994850931204405e-05, "loss": 0.18283427953720094, "step": 2330 }, { "epoch": 0.061639167212704515, "grad_norm": 7.465311527252197, "learning_rate": 1.99476410453827e-05, "loss": 0.19444198608398439, "step": 2340 }, { "epoch": 0.06190258245720325, "grad_norm": 4.379313945770264, "learning_rate": 1.994676553839902e-05, "loss": 0.19880189895629882, "step": 2350 }, { "epoch": 0.06216599770170199, "grad_norm": 5.917383670806885, "learning_rate": 1.9945882791730246e-05, "loss": 0.20259828567504884, "step": 2360 }, { "epoch": 0.06242941294620073, "grad_norm": 5.0624847412109375, "learning_rate": 1.994499280601887e-05, "loss": 0.18192431926727295, "step": 2370 }, { "epoch": 0.06269282819069946, "grad_norm": 4.571087837219238, "learning_rate": 1.994409558191267e-05, "loss": 0.17711472511291504, "step": 2380 }, { "epoch": 0.0629562434351982, "grad_norm": 6.050675392150879, "learning_rate": 1.9943191120064672e-05, "loss": 0.1919628143310547, "step": 2390 }, { "epoch": 0.06321965867969694, "grad_norm": 4.64731502532959, "learning_rate": 1.99422794211332e-05, "loss": 0.17945259809494019, "step": 2400 }, { "epoch": 0.06348307392419568, "grad_norm": 3.47151780128479, "learning_rate": 1.9941360485781814e-05, "loss": 0.18658344745635985, "step": 2410 }, { "epoch": 0.06374648916869441, "grad_norm": 1.881614327430725, "learning_rate": 1.9940434314679363e-05, "loss": 0.1865686893463135, "step": 2420 }, { "epoch": 0.06400990441319315, "grad_norm": 3.808119773864746, "learning_rate": 1.9939500908499947e-05, "loss": 0.17774194478988647, "step": 2430 }, { "epoch": 0.0642733196576919, "grad_norm": 8.280099868774414, "learning_rate": 1.9938560267922948e-05, "loss": 0.1943115472793579, "step": 2440 }, { "epoch": 0.06453673490219063, "grad_norm": 2.305104970932007, "learning_rate": 1.9937612393632995e-05, "loss": 0.1630925416946411, "step": 2450 }, { "epoch": 0.06480015014668937, "grad_norm": 2.684286117553711, "learning_rate": 1.9936657286320003e-05, "loss": 0.16919898986816406, "step": 2460 }, { "epoch": 0.0650635653911881, "grad_norm": 2.8962996006011963, "learning_rate": 1.993569494667913e-05, "loss": 0.18471906185150147, "step": 2470 }, { "epoch": 0.06532698063568684, "grad_norm": 3.6993110179901123, "learning_rate": 1.9934725375410812e-05, "loss": 0.19146593809127807, "step": 2480 }, { "epoch": 0.06559039588018557, "grad_norm": 2.390958786010742, "learning_rate": 1.9933748573220745e-05, "loss": 0.1938166618347168, "step": 2490 }, { "epoch": 0.06585381112468432, "grad_norm": 7.483299732208252, "learning_rate": 1.993276454081989e-05, "loss": 0.18031926155090333, "step": 2500 }, { "epoch": 0.06611722636918305, "grad_norm": 5.489187240600586, "learning_rate": 1.9931773278924467e-05, "loss": 0.18941657543182372, "step": 2510 }, { "epoch": 0.06638064161368179, "grad_norm": 4.617403984069824, "learning_rate": 1.9930774788255957e-05, "loss": 0.1923074722290039, "step": 2520 }, { "epoch": 0.06664405685818052, "grad_norm": 2.564993381500244, "learning_rate": 1.992976906954111e-05, "loss": 0.20013747215270997, "step": 2530 }, { "epoch": 0.06690747210267926, "grad_norm": 3.267894744873047, "learning_rate": 1.992875612351192e-05, "loss": 0.1821206331253052, "step": 2540 }, { "epoch": 0.067170887347178, "grad_norm": 2.2305121421813965, "learning_rate": 1.9927735950905663e-05, "loss": 0.16415390968322754, "step": 2550 }, { "epoch": 0.06743430259167674, "grad_norm": 2.0436975955963135, "learning_rate": 1.9926708552464865e-05, "loss": 0.20756180286407472, "step": 2560 }, { "epoch": 0.06769771783617548, "grad_norm": 5.966801166534424, "learning_rate": 1.9925673928937307e-05, "loss": 0.19010705947875978, "step": 2570 }, { "epoch": 0.06796113308067421, "grad_norm": 3.2199292182922363, "learning_rate": 1.9924632081076032e-05, "loss": 0.18300753831863403, "step": 2580 }, { "epoch": 0.06822454832517295, "grad_norm": 3.194340705871582, "learning_rate": 1.9923583009639344e-05, "loss": 0.20225744247436522, "step": 2590 }, { "epoch": 0.06848796356967168, "grad_norm": 3.341642379760742, "learning_rate": 1.9922526715390802e-05, "loss": 0.18709290027618408, "step": 2600 }, { "epoch": 0.06875137881417043, "grad_norm": 4.478317737579346, "learning_rate": 1.9921463199099222e-05, "loss": 0.18174588680267334, "step": 2610 }, { "epoch": 0.06901479405866916, "grad_norm": 3.0433528423309326, "learning_rate": 1.992039246153868e-05, "loss": 0.1877084493637085, "step": 2620 }, { "epoch": 0.0692782093031679, "grad_norm": 4.565948963165283, "learning_rate": 1.99193145034885e-05, "loss": 0.17374887466430664, "step": 2630 }, { "epoch": 0.06954162454766663, "grad_norm": 1.9712069034576416, "learning_rate": 1.9918229325733267e-05, "loss": 0.18082921504974364, "step": 2640 }, { "epoch": 0.06980503979216537, "grad_norm": 1.8594028949737549, "learning_rate": 1.9917136929062824e-05, "loss": 0.17794979810714723, "step": 2650 }, { "epoch": 0.0700684550366641, "grad_norm": 4.027826309204102, "learning_rate": 1.991603731427226e-05, "loss": 0.1831676721572876, "step": 2660 }, { "epoch": 0.07033187028116285, "grad_norm": 5.407297611236572, "learning_rate": 1.9914930482161928e-05, "loss": 0.1803300380706787, "step": 2670 }, { "epoch": 0.07059528552566158, "grad_norm": 2.3734257221221924, "learning_rate": 1.9913816433537425e-05, "loss": 0.1683368444442749, "step": 2680 }, { "epoch": 0.07085870077016032, "grad_norm": 3.139096260070801, "learning_rate": 1.99126951692096e-05, "loss": 0.1918026924133301, "step": 2690 }, { "epoch": 0.07112211601465906, "grad_norm": 3.7017223834991455, "learning_rate": 1.991156668999456e-05, "loss": 0.19312978982925416, "step": 2700 }, { "epoch": 0.0713855312591578, "grad_norm": 3.014127254486084, "learning_rate": 1.9910430996713668e-05, "loss": 0.17089459896087647, "step": 2710 }, { "epoch": 0.07164894650365654, "grad_norm": 5.492636680603027, "learning_rate": 1.9909288090193518e-05, "loss": 0.19119589328765868, "step": 2720 }, { "epoch": 0.07191236174815527, "grad_norm": 2.1030006408691406, "learning_rate": 1.9908137971265977e-05, "loss": 0.19488012790679932, "step": 2730 }, { "epoch": 0.07217577699265401, "grad_norm": 3.184659481048584, "learning_rate": 1.9906980640768146e-05, "loss": 0.18685141801834107, "step": 2740 }, { "epoch": 0.07243919223715274, "grad_norm": 6.022740364074707, "learning_rate": 1.9905816099542382e-05, "loss": 0.1844327211380005, "step": 2750 }, { "epoch": 0.07270260748165149, "grad_norm": 3.3105337619781494, "learning_rate": 1.9904644348436282e-05, "loss": 0.18523452281951905, "step": 2760 }, { "epoch": 0.07296602272615021, "grad_norm": 2.7014458179473877, "learning_rate": 1.990346538830271e-05, "loss": 0.18013062477111816, "step": 2770 }, { "epoch": 0.07322943797064896, "grad_norm": 2.3197498321533203, "learning_rate": 1.9902279219999752e-05, "loss": 0.1764710783958435, "step": 2780 }, { "epoch": 0.07349285321514769, "grad_norm": 5.595232963562012, "learning_rate": 1.9901085844390757e-05, "loss": 0.18347619771957396, "step": 2790 }, { "epoch": 0.07375626845964643, "grad_norm": 6.556450843811035, "learning_rate": 1.9899885262344316e-05, "loss": 0.1898115396499634, "step": 2800 }, { "epoch": 0.07401968370414516, "grad_norm": 3.6051034927368164, "learning_rate": 1.9898677474734266e-05, "loss": 0.18249425888061524, "step": 2810 }, { "epoch": 0.0742830989486439, "grad_norm": 3.0657918453216553, "learning_rate": 1.9897462482439684e-05, "loss": 0.1841177225112915, "step": 2820 }, { "epoch": 0.07454651419314265, "grad_norm": 4.426631450653076, "learning_rate": 1.9896240286344892e-05, "loss": 0.19920854568481444, "step": 2830 }, { "epoch": 0.07480992943764138, "grad_norm": 3.9778947830200195, "learning_rate": 1.989501088733946e-05, "loss": 0.18430956602096557, "step": 2840 }, { "epoch": 0.07507334468214012, "grad_norm": 1.963383436203003, "learning_rate": 1.9893774286318198e-05, "loss": 0.18080658912658693, "step": 2850 }, { "epoch": 0.07533675992663885, "grad_norm": 7.972723484039307, "learning_rate": 1.9892530484181155e-05, "loss": 0.18941159248352052, "step": 2860 }, { "epoch": 0.0756001751711376, "grad_norm": 3.3552310466766357, "learning_rate": 1.989127948183363e-05, "loss": 0.18045167922973632, "step": 2870 }, { "epoch": 0.07586359041563633, "grad_norm": 4.646827220916748, "learning_rate": 1.9890021280186148e-05, "loss": 0.2019951343536377, "step": 2880 }, { "epoch": 0.07612700566013507, "grad_norm": 4.084786891937256, "learning_rate": 1.9888755880154485e-05, "loss": 0.1767280340194702, "step": 2890 }, { "epoch": 0.0763904209046338, "grad_norm": 3.976330041885376, "learning_rate": 1.988748328265966e-05, "loss": 0.1631881833076477, "step": 2900 }, { "epoch": 0.07665383614913254, "grad_norm": 3.900550127029419, "learning_rate": 1.9886203488627918e-05, "loss": 0.18796756267547607, "step": 2910 }, { "epoch": 0.07691725139363127, "grad_norm": 4.5621232986450195, "learning_rate": 1.988491649899075e-05, "loss": 0.1848163604736328, "step": 2920 }, { "epoch": 0.07718066663813002, "grad_norm": 4.938278675079346, "learning_rate": 1.988362231468488e-05, "loss": 0.17839505672454833, "step": 2930 }, { "epoch": 0.07744408188262875, "grad_norm": 1.928694725036621, "learning_rate": 1.9882320936652278e-05, "loss": 0.18899123668670653, "step": 2940 }, { "epoch": 0.07770749712712749, "grad_norm": 2.9503560066223145, "learning_rate": 1.9881012365840138e-05, "loss": 0.17419015169143676, "step": 2950 }, { "epoch": 0.07797091237162623, "grad_norm": 2.7028496265411377, "learning_rate": 1.987969660320089e-05, "loss": 0.18425474166870118, "step": 2960 }, { "epoch": 0.07823432761612496, "grad_norm": 2.510359764099121, "learning_rate": 1.9878373649692213e-05, "loss": 0.1989011287689209, "step": 2970 }, { "epoch": 0.0784977428606237, "grad_norm": 2.642591953277588, "learning_rate": 1.9877043506277004e-05, "loss": 0.1935784101486206, "step": 2980 }, { "epoch": 0.07876115810512244, "grad_norm": 4.389028549194336, "learning_rate": 1.9875706173923396e-05, "loss": 0.18192651271820068, "step": 2990 }, { "epoch": 0.07902457334962118, "grad_norm": 4.135473251342773, "learning_rate": 1.9874361653604765e-05, "loss": 0.17450098991394042, "step": 3000 }, { "epoch": 0.07928798859411991, "grad_norm": 3.695692539215088, "learning_rate": 1.9873009946299702e-05, "loss": 0.19595288038253783, "step": 3010 }, { "epoch": 0.07955140383861865, "grad_norm": 4.751458644866943, "learning_rate": 1.9871651052992046e-05, "loss": 0.1743461847305298, "step": 3020 }, { "epoch": 0.07981481908311738, "grad_norm": 21.81496238708496, "learning_rate": 1.987028497467085e-05, "loss": 0.16754767894744874, "step": 3030 }, { "epoch": 0.08007823432761613, "grad_norm": 41.40000915527344, "learning_rate": 1.986891171233041e-05, "loss": 3.1389392852783202, "step": 3040 }, { "epoch": 0.08034164957211486, "grad_norm": 7.899384498596191, "learning_rate": 1.9867531266970247e-05, "loss": 0.9128328323364258, "step": 3050 }, { "epoch": 0.0806050648166136, "grad_norm": 6.970911026000977, "learning_rate": 1.9866143639595107e-05, "loss": 0.40287299156188966, "step": 3060 }, { "epoch": 0.08086848006111233, "grad_norm": 3.6127257347106934, "learning_rate": 1.9864748831214964e-05, "loss": 0.2828946828842163, "step": 3070 }, { "epoch": 0.08113189530561107, "grad_norm": 5.091655254364014, "learning_rate": 1.986334684284502e-05, "loss": 0.2540289402008057, "step": 3080 }, { "epoch": 0.08139531055010982, "grad_norm": 10.519774436950684, "learning_rate": 1.98619376755057e-05, "loss": 0.2523433923721313, "step": 3090 }, { "epoch": 0.08165872579460855, "grad_norm": 4.363302707672119, "learning_rate": 1.986052133022266e-05, "loss": 0.2477975606918335, "step": 3100 }, { "epoch": 0.08192214103910729, "grad_norm": 8.223128318786621, "learning_rate": 1.985909780802678e-05, "loss": 0.21361303329467773, "step": 3110 }, { "epoch": 0.08218555628360602, "grad_norm": 2.258450508117676, "learning_rate": 1.985766710995416e-05, "loss": 0.20169546604156494, "step": 3120 }, { "epoch": 0.08244897152810476, "grad_norm": 7.401748180389404, "learning_rate": 1.9856229237046115e-05, "loss": 0.21518549919128419, "step": 3130 }, { "epoch": 0.0827123867726035, "grad_norm": 4.1699018478393555, "learning_rate": 1.9854784190349203e-05, "loss": 0.2182821273803711, "step": 3140 }, { "epoch": 0.08297580201710224, "grad_norm": 4.526690483093262, "learning_rate": 1.985333197091518e-05, "loss": 0.2142322540283203, "step": 3150 }, { "epoch": 0.08323921726160097, "grad_norm": 3.4284474849700928, "learning_rate": 1.9851872579801044e-05, "loss": 0.20082199573516846, "step": 3160 }, { "epoch": 0.08350263250609971, "grad_norm": 4.111307621002197, "learning_rate": 1.9850406018068997e-05, "loss": 0.18752663135528563, "step": 3170 }, { "epoch": 0.08376604775059844, "grad_norm": 4.233161449432373, "learning_rate": 1.984893228678647e-05, "loss": 0.190695321559906, "step": 3180 }, { "epoch": 0.08402946299509718, "grad_norm": 4.789837837219238, "learning_rate": 1.9847451387026103e-05, "loss": 0.19633628129959108, "step": 3190 }, { "epoch": 0.08429287823959591, "grad_norm": 4.493845462799072, "learning_rate": 1.9845963319865758e-05, "loss": 0.2135075092315674, "step": 3200 }, { "epoch": 0.08455629348409466, "grad_norm": 1.6070270538330078, "learning_rate": 1.9844468086388523e-05, "loss": 0.18081240653991698, "step": 3210 }, { "epoch": 0.0848197087285934, "grad_norm": 6.352787494659424, "learning_rate": 1.9842965687682686e-05, "loss": 0.18943965435028076, "step": 3220 }, { "epoch": 0.08508312397309213, "grad_norm": 2.339141607284546, "learning_rate": 1.9841456124841763e-05, "loss": 0.20806796550750734, "step": 3230 }, { "epoch": 0.08534653921759087, "grad_norm": 2.8634114265441895, "learning_rate": 1.983993939896447e-05, "loss": 0.1809318780899048, "step": 3240 }, { "epoch": 0.0856099544620896, "grad_norm": 3.0046238899230957, "learning_rate": 1.9838415511154755e-05, "loss": 0.20622403621673585, "step": 3250 }, { "epoch": 0.08587336970658835, "grad_norm": 3.1119837760925293, "learning_rate": 1.9836884462521768e-05, "loss": 0.18334369659423827, "step": 3260 }, { "epoch": 0.08613678495108708, "grad_norm": 3.595574140548706, "learning_rate": 1.9835346254179865e-05, "loss": 0.19600123167037964, "step": 3270 }, { "epoch": 0.08640020019558582, "grad_norm": 4.395539283752441, "learning_rate": 1.9833800887248628e-05, "loss": 0.19209678173065187, "step": 3280 }, { "epoch": 0.08666361544008455, "grad_norm": 6.030745506286621, "learning_rate": 1.983224836285284e-05, "loss": 0.20327999591827392, "step": 3290 }, { "epoch": 0.0869270306845833, "grad_norm": 6.169255256652832, "learning_rate": 1.9830688682122494e-05, "loss": 0.21170496940612793, "step": 3300 }, { "epoch": 0.08719044592908202, "grad_norm": 5.457612037658691, "learning_rate": 1.982912184619279e-05, "loss": 0.19211854934692382, "step": 3310 }, { "epoch": 0.08745386117358077, "grad_norm": 5.729690074920654, "learning_rate": 1.9827547856204147e-05, "loss": 0.18160796165466309, "step": 3320 }, { "epoch": 0.0877172764180795, "grad_norm": 2.9075186252593994, "learning_rate": 1.9825966713302172e-05, "loss": 0.18688899278640747, "step": 3330 }, { "epoch": 0.08798069166257824, "grad_norm": 4.76597785949707, "learning_rate": 1.98243784186377e-05, "loss": 0.1903820276260376, "step": 3340 }, { "epoch": 0.08824410690707699, "grad_norm": 4.340602397918701, "learning_rate": 1.982278297336675e-05, "loss": 0.19768792390823364, "step": 3350 }, { "epoch": 0.08850752215157572, "grad_norm": 4.584577560424805, "learning_rate": 1.9821180378650563e-05, "loss": 0.19868557453155516, "step": 3360 }, { "epoch": 0.08877093739607446, "grad_norm": 5.179031848907471, "learning_rate": 1.9819570635655577e-05, "loss": 0.20435912609100343, "step": 3370 }, { "epoch": 0.08903435264057319, "grad_norm": 5.350759983062744, "learning_rate": 1.981795374555343e-05, "loss": 0.1958784580230713, "step": 3380 }, { "epoch": 0.08929776788507193, "grad_norm": 2.9807558059692383, "learning_rate": 1.981632970952096e-05, "loss": 0.17533090114593505, "step": 3390 }, { "epoch": 0.08956118312957066, "grad_norm": 2.556894540786743, "learning_rate": 1.9814698528740222e-05, "loss": 0.1939052700996399, "step": 3400 }, { "epoch": 0.0898245983740694, "grad_norm": 3.4649441242218018, "learning_rate": 1.9813060204398448e-05, "loss": 0.1870485544204712, "step": 3410 }, { "epoch": 0.09008801361856814, "grad_norm": 4.2834553718566895, "learning_rate": 1.9811414737688088e-05, "loss": 0.18913209438323975, "step": 3420 }, { "epoch": 0.09035142886306688, "grad_norm": 3.056201696395874, "learning_rate": 1.9809762129806777e-05, "loss": 0.1920717716217041, "step": 3430 }, { "epoch": 0.09061484410756561, "grad_norm": 3.005885124206543, "learning_rate": 1.9808102381957364e-05, "loss": 0.19147806167602538, "step": 3440 }, { "epoch": 0.09087825935206435, "grad_norm": 3.795593738555908, "learning_rate": 1.980643549534788e-05, "loss": 0.17831430435180665, "step": 3450 }, { "epoch": 0.0911416745965631, "grad_norm": 1.8621803522109985, "learning_rate": 1.980476147119155e-05, "loss": 0.17744462490081786, "step": 3460 }, { "epoch": 0.09140508984106183, "grad_norm": 2.2570722103118896, "learning_rate": 1.980308031070682e-05, "loss": 0.1825331687927246, "step": 3470 }, { "epoch": 0.09166850508556057, "grad_norm": 3.9516167640686035, "learning_rate": 1.9801392015117286e-05, "loss": 0.20918803215026854, "step": 3480 }, { "epoch": 0.0919319203300593, "grad_norm": 3.4378888607025146, "learning_rate": 1.9799696585651782e-05, "loss": 0.1719522476196289, "step": 3490 }, { "epoch": 0.09219533557455804, "grad_norm": 1.409721851348877, "learning_rate": 1.9797994023544307e-05, "loss": 0.20252008438110353, "step": 3500 }, { "epoch": 0.09245875081905677, "grad_norm": 4.277249813079834, "learning_rate": 1.979628433003406e-05, "loss": 0.18401904106140138, "step": 3510 }, { "epoch": 0.09272216606355552, "grad_norm": 2.706533670425415, "learning_rate": 1.9794567506365432e-05, "loss": 0.18073272705078125, "step": 3520 }, { "epoch": 0.09298558130805425, "grad_norm": 3.6744000911712646, "learning_rate": 1.9792843553788e-05, "loss": 0.1822329044342041, "step": 3530 }, { "epoch": 0.09324899655255299, "grad_norm": 3.335937738418579, "learning_rate": 1.9791112473556528e-05, "loss": 0.19796209335327147, "step": 3540 }, { "epoch": 0.09351241179705172, "grad_norm": 2.6437883377075195, "learning_rate": 1.978937426693097e-05, "loss": 0.18831920623779297, "step": 3550 }, { "epoch": 0.09377582704155046, "grad_norm": 3.4072747230529785, "learning_rate": 1.9787628935176476e-05, "loss": 0.185935378074646, "step": 3560 }, { "epoch": 0.0940392422860492, "grad_norm": 3.4103379249572754, "learning_rate": 1.9785876479563367e-05, "loss": 0.19327244758605958, "step": 3570 }, { "epoch": 0.09430265753054794, "grad_norm": 1.6974483728408813, "learning_rate": 1.978411690136716e-05, "loss": 0.1893221139907837, "step": 3580 }, { "epoch": 0.09456607277504668, "grad_norm": 2.9079670906066895, "learning_rate": 1.9782350201868547e-05, "loss": 0.1774933934211731, "step": 3590 }, { "epoch": 0.09482948801954541, "grad_norm": 5.662779808044434, "learning_rate": 1.978057638235341e-05, "loss": 0.1913711190223694, "step": 3600 }, { "epoch": 0.09509290326404415, "grad_norm": 3.964238166809082, "learning_rate": 1.9778795444112812e-05, "loss": 0.18812466859817506, "step": 3610 }, { "epoch": 0.09535631850854288, "grad_norm": 4.975062370300293, "learning_rate": 1.9777007388443e-05, "loss": 0.18344542980194092, "step": 3620 }, { "epoch": 0.09561973375304163, "grad_norm": 1.9209367036819458, "learning_rate": 1.9775212216645393e-05, "loss": 0.16861286163330078, "step": 3630 }, { "epoch": 0.09588314899754036, "grad_norm": 1.5856252908706665, "learning_rate": 1.977340993002659e-05, "loss": 0.19432146549224855, "step": 3640 }, { "epoch": 0.0961465642420391, "grad_norm": 5.017166614532471, "learning_rate": 1.977160052989839e-05, "loss": 0.19420454502105713, "step": 3650 }, { "epoch": 0.09640997948653783, "grad_norm": 2.5825605392456055, "learning_rate": 1.9769784017577734e-05, "loss": 0.1767353296279907, "step": 3660 }, { "epoch": 0.09667339473103657, "grad_norm": 9.985602378845215, "learning_rate": 1.976796039438677e-05, "loss": 0.19165711402893065, "step": 3670 }, { "epoch": 0.0969368099755353, "grad_norm": 5.9278974533081055, "learning_rate": 1.9766129661652803e-05, "loss": 0.19478224515914916, "step": 3680 }, { "epoch": 0.09720022522003405, "grad_norm": 5.707930088043213, "learning_rate": 1.9764291820708324e-05, "loss": 0.19843045473098755, "step": 3690 }, { "epoch": 0.09746364046453278, "grad_norm": 2.9496591091156006, "learning_rate": 1.9762446872890984e-05, "loss": 0.17706778049468994, "step": 3700 }, { "epoch": 0.09772705570903152, "grad_norm": 3.8667593002319336, "learning_rate": 1.976059481954363e-05, "loss": 0.1993479013442993, "step": 3710 }, { "epoch": 0.09799047095353026, "grad_norm": 1.6717897653579712, "learning_rate": 1.9758735662014253e-05, "loss": 0.17816649675369262, "step": 3720 }, { "epoch": 0.098253886198029, "grad_norm": 3.0397582054138184, "learning_rate": 1.975686940165604e-05, "loss": 0.1732832670211792, "step": 3730 }, { "epoch": 0.09851730144252774, "grad_norm": 2.470946788787842, "learning_rate": 1.9754996039827322e-05, "loss": 0.176236355304718, "step": 3740 }, { "epoch": 0.09878071668702647, "grad_norm": 2.8315889835357666, "learning_rate": 1.9753115577891623e-05, "loss": 0.17744022607803345, "step": 3750 }, { "epoch": 0.09904413193152521, "grad_norm": 3.0599358081817627, "learning_rate": 1.9751228017217623e-05, "loss": 0.1722869396209717, "step": 3760 }, { "epoch": 0.09930754717602394, "grad_norm": 3.4497363567352295, "learning_rate": 1.9749333359179166e-05, "loss": 0.17917091846466066, "step": 3770 }, { "epoch": 0.09957096242052269, "grad_norm": 3.229686975479126, "learning_rate": 1.974743160515527e-05, "loss": 0.18002710342407227, "step": 3780 }, { "epoch": 0.09983437766502141, "grad_norm": 1.997007131576538, "learning_rate": 1.974552275653011e-05, "loss": 0.19177277088165284, "step": 3790 }, { "epoch": 0.10009779290952016, "grad_norm": 3.7287471294403076, "learning_rate": 1.974360681469303e-05, "loss": 0.18530693054199218, "step": 3800 }, { "epoch": 0.10036120815401889, "grad_norm": 1.9283244609832764, "learning_rate": 1.9741683781038538e-05, "loss": 0.1857541561126709, "step": 3810 }, { "epoch": 0.10062462339851763, "grad_norm": 3.686837673187256, "learning_rate": 1.9739753656966294e-05, "loss": 0.18073644638061523, "step": 3820 }, { "epoch": 0.10088803864301636, "grad_norm": 4.565985202789307, "learning_rate": 1.9737816443881134e-05, "loss": 0.1966541290283203, "step": 3830 }, { "epoch": 0.1011514538875151, "grad_norm": 3.1165950298309326, "learning_rate": 1.973587214319304e-05, "loss": 0.1711329460144043, "step": 3840 }, { "epoch": 0.10141486913201385, "grad_norm": 2.526902437210083, "learning_rate": 1.9733920756317157e-05, "loss": 0.17759722471237183, "step": 3850 }, { "epoch": 0.10167828437651258, "grad_norm": 1.9559698104858398, "learning_rate": 1.973196228467379e-05, "loss": 0.18347470760345458, "step": 3860 }, { "epoch": 0.10194169962101132, "grad_norm": 3.2591896057128906, "learning_rate": 1.9729996729688403e-05, "loss": 0.1918001651763916, "step": 3870 }, { "epoch": 0.10220511486551005, "grad_norm": 3.238548517227173, "learning_rate": 1.9728024092791606e-05, "loss": 0.19188899993896485, "step": 3880 }, { "epoch": 0.1024685301100088, "grad_norm": 1.6570371389389038, "learning_rate": 1.9726044375419172e-05, "loss": 0.18097162246704102, "step": 3890 }, { "epoch": 0.10273194535450753, "grad_norm": 6.12017822265625, "learning_rate": 1.9724057579012026e-05, "loss": 0.18911347389221192, "step": 3900 }, { "epoch": 0.10299536059900627, "grad_norm": 2.029507875442505, "learning_rate": 1.972206370501624e-05, "loss": 0.1888122320175171, "step": 3910 }, { "epoch": 0.103258775843505, "grad_norm": 2.8869853019714355, "learning_rate": 1.9720062754883048e-05, "loss": 0.18739699125289916, "step": 3920 }, { "epoch": 0.10352219108800374, "grad_norm": 2.345921754837036, "learning_rate": 1.971805473006882e-05, "loss": 0.17723631858825684, "step": 3930 }, { "epoch": 0.10378560633250247, "grad_norm": 3.0287649631500244, "learning_rate": 1.9716039632035086e-05, "loss": 0.1785689949989319, "step": 3940 }, { "epoch": 0.10404902157700122, "grad_norm": 2.3496556282043457, "learning_rate": 1.9714017462248523e-05, "loss": 0.20363311767578124, "step": 3950 }, { "epoch": 0.10431243682149995, "grad_norm": 2.7647407054901123, "learning_rate": 1.9711988222180958e-05, "loss": 0.18993332386016845, "step": 3960 }, { "epoch": 0.10457585206599869, "grad_norm": 5.696256160736084, "learning_rate": 1.9709951913309347e-05, "loss": 0.19243887662887574, "step": 3970 }, { "epoch": 0.10483926731049743, "grad_norm": 2.070690870285034, "learning_rate": 1.9707908537115816e-05, "loss": 0.18509730100631713, "step": 3980 }, { "epoch": 0.10510268255499616, "grad_norm": 2.5540571212768555, "learning_rate": 1.9705858095087613e-05, "loss": 0.1881028890609741, "step": 3990 }, { "epoch": 0.1053660977994949, "grad_norm": 4.609920501708984, "learning_rate": 1.9703800588717144e-05, "loss": 0.17430440187454224, "step": 4000 }, { "epoch": 0.10562951304399364, "grad_norm": 3.1286184787750244, "learning_rate": 1.9701736019501948e-05, "loss": 0.1984719753265381, "step": 4010 }, { "epoch": 0.10589292828849238, "grad_norm": 4.410133361816406, "learning_rate": 1.969966438894471e-05, "loss": 0.17521432638168336, "step": 4020 }, { "epoch": 0.10615634353299111, "grad_norm": 3.0032360553741455, "learning_rate": 1.969758569855325e-05, "loss": 0.19048229455947877, "step": 4030 }, { "epoch": 0.10641975877748985, "grad_norm": 1.2013188600540161, "learning_rate": 1.969549994984053e-05, "loss": 0.19852278232574463, "step": 4040 }, { "epoch": 0.10668317402198858, "grad_norm": 3.6427218914031982, "learning_rate": 1.9693407144324642e-05, "loss": 0.1648772120475769, "step": 4050 }, { "epoch": 0.10694658926648733, "grad_norm": 3.2199642658233643, "learning_rate": 1.9691307283528834e-05, "loss": 0.20608830451965332, "step": 4060 }, { "epoch": 0.10721000451098606, "grad_norm": 2.6749696731567383, "learning_rate": 1.968920036898146e-05, "loss": 0.18776509761810303, "step": 4070 }, { "epoch": 0.1074734197554848, "grad_norm": 2.393136501312256, "learning_rate": 1.9687086402216033e-05, "loss": 0.1821596145629883, "step": 4080 }, { "epoch": 0.10773683499998353, "grad_norm": 1.8408534526824951, "learning_rate": 1.968496538477118e-05, "loss": 0.17887433767318725, "step": 4090 }, { "epoch": 0.10800025024448227, "grad_norm": 2.3750438690185547, "learning_rate": 1.9682837318190683e-05, "loss": 0.20585079193115235, "step": 4100 }, { "epoch": 0.10826366548898102, "grad_norm": 2.8319756984710693, "learning_rate": 1.968070220402343e-05, "loss": 0.16612377166748046, "step": 4110 }, { "epoch": 0.10852708073347975, "grad_norm": 1.9069048166275024, "learning_rate": 1.967856004382345e-05, "loss": 0.17688900232315063, "step": 4120 }, { "epoch": 0.10879049597797849, "grad_norm": 5.0369439125061035, "learning_rate": 1.96764108391499e-05, "loss": 0.17638555765151978, "step": 4130 }, { "epoch": 0.10905391122247722, "grad_norm": 5.747983932495117, "learning_rate": 1.967425459156707e-05, "loss": 0.18800159692764282, "step": 4140 }, { "epoch": 0.10931732646697596, "grad_norm": 2.3741767406463623, "learning_rate": 1.9672091302644363e-05, "loss": 0.18372081518173217, "step": 4150 }, { "epoch": 0.1095807417114747, "grad_norm": 5.708831787109375, "learning_rate": 1.966992097395632e-05, "loss": 0.1822826623916626, "step": 4160 }, { "epoch": 0.10984415695597344, "grad_norm": 1.5648043155670166, "learning_rate": 1.9667743607082593e-05, "loss": 0.17678004503250122, "step": 4170 }, { "epoch": 0.11010757220047217, "grad_norm": 3.463596820831299, "learning_rate": 1.9665559203607965e-05, "loss": 0.16971532106399537, "step": 4180 }, { "epoch": 0.11037098744497091, "grad_norm": 1.4446300268173218, "learning_rate": 1.9663367765122346e-05, "loss": 0.18374334573745726, "step": 4190 }, { "epoch": 0.11063440268946964, "grad_norm": 1.2626368999481201, "learning_rate": 1.966116929322075e-05, "loss": 0.18811365365982055, "step": 4200 }, { "epoch": 0.11089781793396838, "grad_norm": 2.25801682472229, "learning_rate": 1.9658963789503328e-05, "loss": 0.17331156730651856, "step": 4210 }, { "epoch": 0.11116123317846711, "grad_norm": 2.2809176445007324, "learning_rate": 1.9656751255575336e-05, "loss": 0.1613398551940918, "step": 4220 }, { "epoch": 0.11142464842296586, "grad_norm": 3.94130539894104, "learning_rate": 1.965453169304715e-05, "loss": 0.18196384906768798, "step": 4230 }, { "epoch": 0.1116880636674646, "grad_norm": 3.281834363937378, "learning_rate": 1.9652305103534265e-05, "loss": 0.18697152137756348, "step": 4240 }, { "epoch": 0.11195147891196333, "grad_norm": 3.0692825317382812, "learning_rate": 1.9650071488657293e-05, "loss": 0.1712125778198242, "step": 4250 }, { "epoch": 0.11221489415646207, "grad_norm": 2.160331964492798, "learning_rate": 1.964783085004195e-05, "loss": 0.1793117880821228, "step": 4260 }, { "epoch": 0.1124783094009608, "grad_norm": 3.0753581523895264, "learning_rate": 1.9645583189319074e-05, "loss": 0.18885011672973634, "step": 4270 }, { "epoch": 0.11274172464545955, "grad_norm": 6.3264594078063965, "learning_rate": 1.9643328508124605e-05, "loss": 0.1951131820678711, "step": 4280 }, { "epoch": 0.11300513988995828, "grad_norm": 4.003157615661621, "learning_rate": 1.9641066808099597e-05, "loss": 0.18833229541778565, "step": 4290 }, { "epoch": 0.11326855513445702, "grad_norm": 1.5505484342575073, "learning_rate": 1.963879809089022e-05, "loss": 0.1620851993560791, "step": 4300 }, { "epoch": 0.11353197037895575, "grad_norm": 4.84819221496582, "learning_rate": 1.9636522358147737e-05, "loss": 0.1969823122024536, "step": 4310 }, { "epoch": 0.1137953856234545, "grad_norm": 3.3414950370788574, "learning_rate": 1.9634239611528527e-05, "loss": 0.1927353024482727, "step": 4320 }, { "epoch": 0.11405880086795323, "grad_norm": 2.8033154010772705, "learning_rate": 1.9631949852694073e-05, "loss": 0.18668415546417236, "step": 4330 }, { "epoch": 0.11432221611245197, "grad_norm": 3.5981478691101074, "learning_rate": 1.9629653083310956e-05, "loss": 0.19271641969680786, "step": 4340 }, { "epoch": 0.1145856313569507, "grad_norm": 3.2108070850372314, "learning_rate": 1.9627349305050867e-05, "loss": 0.19004114866256713, "step": 4350 }, { "epoch": 0.11484904660144944, "grad_norm": 4.597020149230957, "learning_rate": 1.9625038519590596e-05, "loss": 0.18823221921920777, "step": 4360 }, { "epoch": 0.11511246184594819, "grad_norm": 6.66159200668335, "learning_rate": 1.962272072861203e-05, "loss": 0.19502179622650145, "step": 4370 }, { "epoch": 0.11537587709044692, "grad_norm": 4.034911155700684, "learning_rate": 1.9620395933802158e-05, "loss": 0.19667913913726806, "step": 4380 }, { "epoch": 0.11563929233494566, "grad_norm": 1.8334640264511108, "learning_rate": 1.9618064136853066e-05, "loss": 0.1688124418258667, "step": 4390 }, { "epoch": 0.11590270757944439, "grad_norm": 2.387218952178955, "learning_rate": 1.9615725339461936e-05, "loss": 0.18490772247314452, "step": 4400 }, { "epoch": 0.11616612282394313, "grad_norm": 3.1332454681396484, "learning_rate": 1.961337954333105e-05, "loss": 0.17604663372039794, "step": 4410 }, { "epoch": 0.11642953806844186, "grad_norm": 2.4526169300079346, "learning_rate": 1.961102675016777e-05, "loss": 0.18824644088745118, "step": 4420 }, { "epoch": 0.1166929533129406, "grad_norm": 3.303798198699951, "learning_rate": 1.960866696168457e-05, "loss": 0.1682475209236145, "step": 4430 }, { "epoch": 0.11695636855743934, "grad_norm": 3.9799859523773193, "learning_rate": 1.9606300179599004e-05, "loss": 0.17307280302047728, "step": 4440 }, { "epoch": 0.11721978380193808, "grad_norm": 2.545074224472046, "learning_rate": 1.9603926405633712e-05, "loss": 0.18958544731140137, "step": 4450 }, { "epoch": 0.11748319904643681, "grad_norm": 2.413820266723633, "learning_rate": 1.9601545641516434e-05, "loss": 0.1987229585647583, "step": 4460 }, { "epoch": 0.11774661429093555, "grad_norm": 3.7804312705993652, "learning_rate": 1.9599157888979997e-05, "loss": 0.18399736881256104, "step": 4470 }, { "epoch": 0.11801002953543428, "grad_norm": 1.631490707397461, "learning_rate": 1.95967631497623e-05, "loss": 0.17620290517807008, "step": 4480 }, { "epoch": 0.11827344477993303, "grad_norm": 3.184896230697632, "learning_rate": 1.9594361425606348e-05, "loss": 0.1811963438987732, "step": 4490 }, { "epoch": 0.11853686002443177, "grad_norm": 2.2048325538635254, "learning_rate": 1.959195271826021e-05, "loss": 0.18106807470321656, "step": 4500 }, { "epoch": 0.1188002752689305, "grad_norm": 7.2130327224731445, "learning_rate": 1.9589537029477052e-05, "loss": 0.17258849143981933, "step": 4510 }, { "epoch": 0.11906369051342924, "grad_norm": 2.4077794551849365, "learning_rate": 1.9587114361015118e-05, "loss": 0.1846498131752014, "step": 4520 }, { "epoch": 0.11932710575792797, "grad_norm": 2.7301387786865234, "learning_rate": 1.958468471463773e-05, "loss": 0.18327456712722778, "step": 4530 }, { "epoch": 0.11959052100242672, "grad_norm": 2.5104012489318848, "learning_rate": 1.9582248092113286e-05, "loss": 0.18134310245513915, "step": 4540 }, { "epoch": 0.11985393624692545, "grad_norm": 1.8185328245162964, "learning_rate": 1.9579804495215268e-05, "loss": 0.1704181909561157, "step": 4550 }, { "epoch": 0.12011735149142419, "grad_norm": 2.447025775909424, "learning_rate": 1.9577353925722228e-05, "loss": 0.191862952709198, "step": 4560 }, { "epoch": 0.12038076673592292, "grad_norm": 1.858643651008606, "learning_rate": 1.9574896385417803e-05, "loss": 0.1937686800956726, "step": 4570 }, { "epoch": 0.12064418198042166, "grad_norm": 1.8188230991363525, "learning_rate": 1.957243187609069e-05, "loss": 0.18003554344177247, "step": 4580 }, { "epoch": 0.1209075972249204, "grad_norm": 2.1316845417022705, "learning_rate": 1.9569960399534668e-05, "loss": 0.17073830366134643, "step": 4590 }, { "epoch": 0.12117101246941914, "grad_norm": 3.631636381149292, "learning_rate": 1.9567481957548585e-05, "loss": 0.1929926872253418, "step": 4600 }, { "epoch": 0.12143442771391787, "grad_norm": 3.767240524291992, "learning_rate": 1.9564996551936354e-05, "loss": 0.1854413628578186, "step": 4610 }, { "epoch": 0.12169784295841661, "grad_norm": 3.299015998840332, "learning_rate": 1.9562504184506966e-05, "loss": 0.17941205501556395, "step": 4620 }, { "epoch": 0.12196125820291535, "grad_norm": 3.2886083126068115, "learning_rate": 1.9560004857074472e-05, "loss": 0.18028855323791504, "step": 4630 }, { "epoch": 0.12222467344741408, "grad_norm": 1.9190970659255981, "learning_rate": 1.955749857145799e-05, "loss": 0.1745362401008606, "step": 4640 }, { "epoch": 0.12248808869191283, "grad_norm": 4.371621131896973, "learning_rate": 1.95549853294817e-05, "loss": 0.17106374502182006, "step": 4650 }, { "epoch": 0.12275150393641156, "grad_norm": 1.5536365509033203, "learning_rate": 1.955246513297485e-05, "loss": 0.19802556037902833, "step": 4660 }, { "epoch": 0.1230149191809103, "grad_norm": 1.708240270614624, "learning_rate": 1.9549937983771748e-05, "loss": 0.19452601671218872, "step": 4670 }, { "epoch": 0.12327833442540903, "grad_norm": 7.025672912597656, "learning_rate": 1.954740388371176e-05, "loss": 0.1734345316886902, "step": 4680 }, { "epoch": 0.12354174966990777, "grad_norm": 2.1495556831359863, "learning_rate": 1.9544862834639316e-05, "loss": 0.1711448311805725, "step": 4690 }, { "epoch": 0.1238051649144065, "grad_norm": 2.5095925331115723, "learning_rate": 1.9542314838403907e-05, "loss": 0.16957744359970092, "step": 4700 }, { "epoch": 0.12406858015890525, "grad_norm": 2.407132863998413, "learning_rate": 1.9539759896860064e-05, "loss": 0.1895298957824707, "step": 4710 }, { "epoch": 0.12433199540340398, "grad_norm": 4.140638828277588, "learning_rate": 1.9537198011867383e-05, "loss": 0.1867833375930786, "step": 4720 }, { "epoch": 0.12459541064790272, "grad_norm": 3.707385301589966, "learning_rate": 1.9534629185290524e-05, "loss": 0.1890628457069397, "step": 4730 }, { "epoch": 0.12485882589240146, "grad_norm": 1.5830198526382446, "learning_rate": 1.9532053418999186e-05, "loss": 0.19023988246917725, "step": 4740 }, { "epoch": 0.1251222411369002, "grad_norm": 2.7426486015319824, "learning_rate": 1.952947071486812e-05, "loss": 0.1930983543395996, "step": 4750 }, { "epoch": 0.12538565638139892, "grad_norm": 3.518355369567871, "learning_rate": 1.952688107477713e-05, "loss": 0.16996872425079346, "step": 4760 }, { "epoch": 0.12564907162589767, "grad_norm": 1.7738209962844849, "learning_rate": 1.9524284500611076e-05, "loss": 0.19860095977783204, "step": 4770 }, { "epoch": 0.1259124868703964, "grad_norm": 3.4008848667144775, "learning_rate": 1.9521680994259845e-05, "loss": 0.188714337348938, "step": 4780 }, { "epoch": 0.12617590211489516, "grad_norm": 1.5849107503890991, "learning_rate": 1.9519070557618386e-05, "loss": 0.18569495677947997, "step": 4790 }, { "epoch": 0.12643931735939387, "grad_norm": 1.4123388528823853, "learning_rate": 1.9516453192586686e-05, "loss": 0.17660293579101563, "step": 4800 }, { "epoch": 0.12670273260389261, "grad_norm": 1.481276273727417, "learning_rate": 1.951382890106978e-05, "loss": 0.17360745668411254, "step": 4810 }, { "epoch": 0.12696614784839136, "grad_norm": 2.1321215629577637, "learning_rate": 1.951119768497774e-05, "loss": 0.20112462043762208, "step": 4820 }, { "epoch": 0.1272295630928901, "grad_norm": 2.6001508235931396, "learning_rate": 1.9508559546225672e-05, "loss": 0.17699971199035644, "step": 4830 }, { "epoch": 0.12749297833738882, "grad_norm": 1.1898219585418701, "learning_rate": 1.9505914486733735e-05, "loss": 0.1797061324119568, "step": 4840 }, { "epoch": 0.12775639358188756, "grad_norm": 1.3213948011398315, "learning_rate": 1.9503262508427115e-05, "loss": 0.183526873588562, "step": 4850 }, { "epoch": 0.1280198088263863, "grad_norm": 4.372756004333496, "learning_rate": 1.9500603613236034e-05, "loss": 0.17158956527709962, "step": 4860 }, { "epoch": 0.12828322407088505, "grad_norm": 1.508731722831726, "learning_rate": 1.9497937803095756e-05, "loss": 0.18629889488220214, "step": 4870 }, { "epoch": 0.1285466393153838, "grad_norm": 5.016032695770264, "learning_rate": 1.9495265079946565e-05, "loss": 0.17491095066070556, "step": 4880 }, { "epoch": 0.1288100545598825, "grad_norm": 2.3442511558532715, "learning_rate": 1.9492585445733788e-05, "loss": 0.20485515594482423, "step": 4890 }, { "epoch": 0.12907346980438125, "grad_norm": 3.7090020179748535, "learning_rate": 1.9489898902407783e-05, "loss": 0.18974674940109254, "step": 4900 }, { "epoch": 0.12933688504888, "grad_norm": 4.329697132110596, "learning_rate": 1.9487205451923925e-05, "loss": 0.17998819351196288, "step": 4910 }, { "epoch": 0.12960030029337874, "grad_norm": 5.712081432342529, "learning_rate": 1.9484505096242628e-05, "loss": 0.1913785457611084, "step": 4920 }, { "epoch": 0.12986371553787746, "grad_norm": 5.530656337738037, "learning_rate": 1.9481797837329323e-05, "loss": 0.1770768642425537, "step": 4930 }, { "epoch": 0.1301271307823762, "grad_norm": 2.9982786178588867, "learning_rate": 1.9479083677154474e-05, "loss": 0.177888822555542, "step": 4940 }, { "epoch": 0.13039054602687494, "grad_norm": 1.2833551168441772, "learning_rate": 1.9476362617693564e-05, "loss": 0.17825260162353515, "step": 4950 }, { "epoch": 0.1306539612713737, "grad_norm": 1.8497055768966675, "learning_rate": 1.9473634660927092e-05, "loss": 0.18914542198181153, "step": 4960 }, { "epoch": 0.1309173765158724, "grad_norm": 4.8870720863342285, "learning_rate": 1.9470899808840585e-05, "loss": 0.19089066982269287, "step": 4970 }, { "epoch": 0.13118079176037115, "grad_norm": 3.0853779315948486, "learning_rate": 1.9468158063424593e-05, "loss": 0.19616940021514892, "step": 4980 }, { "epoch": 0.1314442070048699, "grad_norm": 3.689155101776123, "learning_rate": 1.946540942667467e-05, "loss": 0.1778654098510742, "step": 4990 }, { "epoch": 0.13170762224936863, "grad_norm": 1.369022011756897, "learning_rate": 1.9462653900591393e-05, "loss": 0.1835202693939209, "step": 5000 }, { "epoch": 0.13197103749386738, "grad_norm": 2.748837471008301, "learning_rate": 1.9459891487180353e-05, "loss": 0.1783807635307312, "step": 5010 }, { "epoch": 0.1322344527383661, "grad_norm": 1.9291914701461792, "learning_rate": 1.9457122188452156e-05, "loss": 0.1710192918777466, "step": 5020 }, { "epoch": 0.13249786798286484, "grad_norm": 4.531503677368164, "learning_rate": 1.9454346006422414e-05, "loss": 0.18009450435638427, "step": 5030 }, { "epoch": 0.13276128322736358, "grad_norm": 1.2797269821166992, "learning_rate": 1.9451562943111753e-05, "loss": 0.19462995529174804, "step": 5040 }, { "epoch": 0.13302469847186232, "grad_norm": 2.6504695415496826, "learning_rate": 1.944877300054581e-05, "loss": 0.19293397665023804, "step": 5050 }, { "epoch": 0.13328811371636104, "grad_norm": 3.2647182941436768, "learning_rate": 1.944597618075522e-05, "loss": 0.17274442911148072, "step": 5060 }, { "epoch": 0.13355152896085978, "grad_norm": 2.383723020553589, "learning_rate": 1.944317248577563e-05, "loss": 0.16622562408447267, "step": 5070 }, { "epoch": 0.13381494420535853, "grad_norm": 4.081072807312012, "learning_rate": 1.9440361917647693e-05, "loss": 0.1886562705039978, "step": 5080 }, { "epoch": 0.13407835944985727, "grad_norm": 1.981072187423706, "learning_rate": 1.943754447841706e-05, "loss": 0.1750401258468628, "step": 5090 }, { "epoch": 0.134341774694356, "grad_norm": 3.5318658351898193, "learning_rate": 1.9434720170134387e-05, "loss": 0.17987370491027832, "step": 5100 }, { "epoch": 0.13460518993885473, "grad_norm": 3.0762174129486084, "learning_rate": 1.9431888994855324e-05, "loss": 0.19203416109085084, "step": 5110 }, { "epoch": 0.13486860518335347, "grad_norm": 4.639650821685791, "learning_rate": 1.9429050954640528e-05, "loss": 0.1783306360244751, "step": 5120 }, { "epoch": 0.13513202042785222, "grad_norm": 3.194140911102295, "learning_rate": 1.942620605155564e-05, "loss": 0.18034632205963136, "step": 5130 }, { "epoch": 0.13539543567235096, "grad_norm": 1.1843633651733398, "learning_rate": 1.9423354287671307e-05, "loss": 0.1839032292366028, "step": 5140 }, { "epoch": 0.13565885091684968, "grad_norm": 2.199920177459717, "learning_rate": 1.9420495665063166e-05, "loss": 0.17689507007598876, "step": 5150 }, { "epoch": 0.13592226616134842, "grad_norm": 1.4929628372192383, "learning_rate": 1.9417630185811846e-05, "loss": 0.17543377876281738, "step": 5160 }, { "epoch": 0.13618568140584716, "grad_norm": 3.8388254642486572, "learning_rate": 1.9414757852002962e-05, "loss": 0.20013115406036378, "step": 5170 }, { "epoch": 0.1364490966503459, "grad_norm": 3.4901123046875, "learning_rate": 1.941187866572713e-05, "loss": 0.17925949096679689, "step": 5180 }, { "epoch": 0.13671251189484462, "grad_norm": 1.9650816917419434, "learning_rate": 1.940899262907994e-05, "loss": 0.18361992835998536, "step": 5190 }, { "epoch": 0.13697592713934337, "grad_norm": 1.8782423734664917, "learning_rate": 1.940609974416198e-05, "loss": 0.17942779064178466, "step": 5200 }, { "epoch": 0.1372393423838421, "grad_norm": 7.367305755615234, "learning_rate": 1.940320001307881e-05, "loss": 0.17899383306503297, "step": 5210 }, { "epoch": 0.13750275762834085, "grad_norm": 3.23629093170166, "learning_rate": 1.940029343794098e-05, "loss": 0.18375343084335327, "step": 5220 }, { "epoch": 0.13776617287283957, "grad_norm": 1.2617040872573853, "learning_rate": 1.9397380020864024e-05, "loss": 0.17974655628204345, "step": 5230 }, { "epoch": 0.13802958811733831, "grad_norm": 4.224201202392578, "learning_rate": 1.939445976396845e-05, "loss": 0.17653994560241698, "step": 5240 }, { "epoch": 0.13829300336183706, "grad_norm": 1.8551445007324219, "learning_rate": 1.939153266937975e-05, "loss": 0.18462560176849366, "step": 5250 }, { "epoch": 0.1385564186063358, "grad_norm": 2.7699830532073975, "learning_rate": 1.938859873922839e-05, "loss": 0.19461669921875, "step": 5260 }, { "epoch": 0.13881983385083455, "grad_norm": 2.4307758808135986, "learning_rate": 1.9385657975649807e-05, "loss": 0.18174535036087036, "step": 5270 }, { "epoch": 0.13908324909533326, "grad_norm": 1.633845329284668, "learning_rate": 1.9382710380784418e-05, "loss": 0.17741720676422118, "step": 5280 }, { "epoch": 0.139346664339832, "grad_norm": 1.7305327653884888, "learning_rate": 1.937975595677761e-05, "loss": 0.17531025409698486, "step": 5290 }, { "epoch": 0.13961007958433075, "grad_norm": 2.037203550338745, "learning_rate": 1.9376794705779735e-05, "loss": 0.18028221130371094, "step": 5300 }, { "epoch": 0.1398734948288295, "grad_norm": 1.8693130016326904, "learning_rate": 1.937382662994613e-05, "loss": 0.17729164361953736, "step": 5310 }, { "epoch": 0.1401369100733282, "grad_norm": 1.789958119392395, "learning_rate": 1.9370851731437077e-05, "loss": 0.180820095539093, "step": 5320 }, { "epoch": 0.14040032531782695, "grad_norm": 6.009233474731445, "learning_rate": 1.9367870012417844e-05, "loss": 0.19060148000717164, "step": 5330 }, { "epoch": 0.1406637405623257, "grad_norm": 2.174441337585449, "learning_rate": 1.936488147505865e-05, "loss": 0.18544418811798097, "step": 5340 }, { "epoch": 0.14092715580682444, "grad_norm": 3.645125150680542, "learning_rate": 1.9361886121534686e-05, "loss": 0.1760026454925537, "step": 5350 }, { "epoch": 0.14119057105132315, "grad_norm": 1.5930678844451904, "learning_rate": 1.93588839540261e-05, "loss": 0.18483452796936034, "step": 5360 }, { "epoch": 0.1414539862958219, "grad_norm": 2.249593734741211, "learning_rate": 1.9355874974717988e-05, "loss": 0.17826972007751465, "step": 5370 }, { "epoch": 0.14171740154032064, "grad_norm": 1.3358795642852783, "learning_rate": 1.9352859185800424e-05, "loss": 0.16321015357971191, "step": 5380 }, { "epoch": 0.14198081678481939, "grad_norm": 1.818878412246704, "learning_rate": 1.934983658946843e-05, "loss": 0.18047977685928346, "step": 5390 }, { "epoch": 0.14224423202931813, "grad_norm": 2.799435615539551, "learning_rate": 1.9346807187921983e-05, "loss": 0.1720750093460083, "step": 5400 }, { "epoch": 0.14250764727381685, "grad_norm": 6.239834785461426, "learning_rate": 1.9343770983366004e-05, "loss": 0.17046306133270264, "step": 5410 }, { "epoch": 0.1427710625183156, "grad_norm": 3.7472567558288574, "learning_rate": 1.9340727978010377e-05, "loss": 0.17757667303085328, "step": 5420 }, { "epoch": 0.14303447776281433, "grad_norm": 1.8321607112884521, "learning_rate": 1.933767817406994e-05, "loss": 0.18186695575714112, "step": 5430 }, { "epoch": 0.14329789300731308, "grad_norm": 3.2847437858581543, "learning_rate": 1.9334621573764462e-05, "loss": 0.18298714160919188, "step": 5440 }, { "epoch": 0.1435613082518118, "grad_norm": 2.0122268199920654, "learning_rate": 1.9331558179318676e-05, "loss": 0.18022603988647462, "step": 5450 }, { "epoch": 0.14382472349631054, "grad_norm": 3.052276611328125, "learning_rate": 1.9328487992962247e-05, "loss": 0.1815226435661316, "step": 5460 }, { "epoch": 0.14408813874080928, "grad_norm": 3.7045586109161377, "learning_rate": 1.9325411016929793e-05, "loss": 0.18485679626464843, "step": 5470 }, { "epoch": 0.14435155398530802, "grad_norm": 1.423164963722229, "learning_rate": 1.9322327253460867e-05, "loss": 0.182426917552948, "step": 5480 }, { "epoch": 0.14461496922980674, "grad_norm": 1.9084197282791138, "learning_rate": 1.931923670479997e-05, "loss": 0.17390133142471315, "step": 5490 }, { "epoch": 0.14487838447430548, "grad_norm": 1.272045373916626, "learning_rate": 1.9316139373196533e-05, "loss": 0.18195159435272218, "step": 5500 }, { "epoch": 0.14514179971880423, "grad_norm": 2.771637439727783, "learning_rate": 1.9313035260904923e-05, "loss": 0.19315485954284667, "step": 5510 }, { "epoch": 0.14540521496330297, "grad_norm": 2.523651123046875, "learning_rate": 1.9309924370184455e-05, "loss": 0.1821666479110718, "step": 5520 }, { "epoch": 0.1456686302078017, "grad_norm": 3.382167339324951, "learning_rate": 1.9306806703299367e-05, "loss": 0.18235890865325927, "step": 5530 }, { "epoch": 0.14593204545230043, "grad_norm": 2.8885107040405273, "learning_rate": 1.9303682262518828e-05, "loss": 0.19163455963134765, "step": 5540 }, { "epoch": 0.14619546069679917, "grad_norm": 2.0009703636169434, "learning_rate": 1.9300551050116946e-05, "loss": 0.17521636486053466, "step": 5550 }, { "epoch": 0.14645887594129792, "grad_norm": 3.579141855239868, "learning_rate": 1.9297413068372745e-05, "loss": 0.16940791606903077, "step": 5560 }, { "epoch": 0.14672229118579666, "grad_norm": 4.429261207580566, "learning_rate": 1.9294268319570195e-05, "loss": 0.1778137803077698, "step": 5570 }, { "epoch": 0.14698570643029538, "grad_norm": 4.133678436279297, "learning_rate": 1.9291116805998168e-05, "loss": 0.1883171558380127, "step": 5580 }, { "epoch": 0.14724912167479412, "grad_norm": 1.0185624361038208, "learning_rate": 1.9287958529950477e-05, "loss": 0.18376989364624025, "step": 5590 }, { "epoch": 0.14751253691929286, "grad_norm": 3.239064931869507, "learning_rate": 1.9284793493725852e-05, "loss": 0.18566833734512328, "step": 5600 }, { "epoch": 0.1477759521637916, "grad_norm": 2.9110913276672363, "learning_rate": 1.9281621699627943e-05, "loss": 0.18174781799316406, "step": 5610 }, { "epoch": 0.14803936740829032, "grad_norm": 1.6149957180023193, "learning_rate": 1.927844314996532e-05, "loss": 0.1664803147315979, "step": 5620 }, { "epoch": 0.14830278265278907, "grad_norm": 3.3343772888183594, "learning_rate": 1.9275257847051464e-05, "loss": 0.1709155797958374, "step": 5630 }, { "epoch": 0.1485661978972878, "grad_norm": 3.8372905254364014, "learning_rate": 1.9272065793204778e-05, "loss": 0.17563915252685547, "step": 5640 }, { "epoch": 0.14882961314178655, "grad_norm": 5.585941314697266, "learning_rate": 1.9268866990748578e-05, "loss": 0.1775055408477783, "step": 5650 }, { "epoch": 0.1490930283862853, "grad_norm": 1.9721060991287231, "learning_rate": 1.9265661442011087e-05, "loss": 0.19333884716033936, "step": 5660 }, { "epoch": 0.149356443630784, "grad_norm": 6.318997383117676, "learning_rate": 1.9262449149325445e-05, "loss": 0.19914543628692627, "step": 5670 }, { "epoch": 0.14961985887528276, "grad_norm": 1.643854022026062, "learning_rate": 1.92592301150297e-05, "loss": 0.18034563064575196, "step": 5680 }, { "epoch": 0.1498832741197815, "grad_norm": 1.9192432165145874, "learning_rate": 1.9256004341466795e-05, "loss": 0.1741476058959961, "step": 5690 }, { "epoch": 0.15014668936428024, "grad_norm": 3.5381357669830322, "learning_rate": 1.9252771830984594e-05, "loss": 0.15871908664703369, "step": 5700 }, { "epoch": 0.15041010460877896, "grad_norm": 6.919475555419922, "learning_rate": 1.9249532585935855e-05, "loss": 0.175436532497406, "step": 5710 }, { "epoch": 0.1506735198532777, "grad_norm": 4.863026142120361, "learning_rate": 1.9246286608678243e-05, "loss": 0.17552826404571534, "step": 5720 }, { "epoch": 0.15093693509777645, "grad_norm": 3.111114740371704, "learning_rate": 1.9243033901574316e-05, "loss": 0.18652281761169434, "step": 5730 }, { "epoch": 0.1512003503422752, "grad_norm": 4.4947509765625, "learning_rate": 1.923977446699154e-05, "loss": 0.18723442554473876, "step": 5740 }, { "epoch": 0.1514637655867739, "grad_norm": 5.1744537353515625, "learning_rate": 1.923650830730227e-05, "loss": 0.17362666130065918, "step": 5750 }, { "epoch": 0.15172718083127265, "grad_norm": 4.5073561668396, "learning_rate": 1.923323542488375e-05, "loss": 0.17453171014785768, "step": 5760 }, { "epoch": 0.1519905960757714, "grad_norm": 1.4074112176895142, "learning_rate": 1.9229955822118138e-05, "loss": 0.1775158405303955, "step": 5770 }, { "epoch": 0.15225401132027014, "grad_norm": 2.940427303314209, "learning_rate": 1.9226669501392465e-05, "loss": 0.1808817744255066, "step": 5780 }, { "epoch": 0.15251742656476888, "grad_norm": 1.846122145652771, "learning_rate": 1.9223376465098657e-05, "loss": 0.1698350191116333, "step": 5790 }, { "epoch": 0.1527808418092676, "grad_norm": 7.964675426483154, "learning_rate": 1.9220076715633526e-05, "loss": 0.1802960991859436, "step": 5800 }, { "epoch": 0.15304425705376634, "grad_norm": 3.9675028324127197, "learning_rate": 1.921677025539877e-05, "loss": 0.17737383842468263, "step": 5810 }, { "epoch": 0.15330767229826509, "grad_norm": 1.0760884284973145, "learning_rate": 1.921345708680098e-05, "loss": 0.17074639797210694, "step": 5820 }, { "epoch": 0.15357108754276383, "grad_norm": 2.69808030128479, "learning_rate": 1.921013721225162e-05, "loss": 0.18073058128356934, "step": 5830 }, { "epoch": 0.15383450278726254, "grad_norm": 3.7863855361938477, "learning_rate": 1.9206810634167036e-05, "loss": 0.1628231406211853, "step": 5840 }, { "epoch": 0.1540979180317613, "grad_norm": 1.9263856410980225, "learning_rate": 1.920347735496846e-05, "loss": 0.17212362289428712, "step": 5850 }, { "epoch": 0.15436133327626003, "grad_norm": 1.7235687971115112, "learning_rate": 1.9200137377081993e-05, "loss": 0.17819247245788575, "step": 5860 }, { "epoch": 0.15462474852075878, "grad_norm": 1.669785976409912, "learning_rate": 1.919679070293861e-05, "loss": 0.17051491737365723, "step": 5870 }, { "epoch": 0.1548881637652575, "grad_norm": 4.709681034088135, "learning_rate": 1.9193437334974178e-05, "loss": 0.17358756065368652, "step": 5880 }, { "epoch": 0.15515157900975624, "grad_norm": 5.448112487792969, "learning_rate": 1.9190077275629408e-05, "loss": 0.17136752605438232, "step": 5890 }, { "epoch": 0.15541499425425498, "grad_norm": 1.6955738067626953, "learning_rate": 1.9186710527349907e-05, "loss": 0.1745814561843872, "step": 5900 }, { "epoch": 0.15567840949875372, "grad_norm": 4.836145877838135, "learning_rate": 1.9183337092586137e-05, "loss": 0.17380751371383668, "step": 5910 }, { "epoch": 0.15594182474325247, "grad_norm": 1.4071156978607178, "learning_rate": 1.9179956973793422e-05, "loss": 0.17932027578353882, "step": 5920 }, { "epoch": 0.15620523998775118, "grad_norm": 1.9462382793426514, "learning_rate": 1.917657017343197e-05, "loss": 0.18140355348587037, "step": 5930 }, { "epoch": 0.15646865523224993, "grad_norm": 2.4838573932647705, "learning_rate": 1.9173176693966835e-05, "loss": 0.18200838565826416, "step": 5940 }, { "epoch": 0.15673207047674867, "grad_norm": 2.127821207046509, "learning_rate": 1.9169776537867936e-05, "loss": 0.18380496501922608, "step": 5950 }, { "epoch": 0.1569954857212474, "grad_norm": 2.4472854137420654, "learning_rate": 1.9166369707610055e-05, "loss": 0.16881023645401, "step": 5960 }, { "epoch": 0.15725890096574613, "grad_norm": 2.488090753555298, "learning_rate": 1.9162956205672833e-05, "loss": 0.1880045175552368, "step": 5970 }, { "epoch": 0.15752231621024487, "grad_norm": 1.0206799507141113, "learning_rate": 1.9159536034540757e-05, "loss": 0.18317186832427979, "step": 5980 }, { "epoch": 0.15778573145474362, "grad_norm": 4.547656059265137, "learning_rate": 1.9156109196703177e-05, "loss": 0.18591307401657103, "step": 5990 }, { "epoch": 0.15804914669924236, "grad_norm": 1.71575927734375, "learning_rate": 1.91526756946543e-05, "loss": 0.1806221842765808, "step": 6000 }, { "epoch": 0.15831256194374108, "grad_norm": 8.788862228393555, "learning_rate": 1.9149235530893172e-05, "loss": 0.1840038299560547, "step": 6010 }, { "epoch": 0.15857597718823982, "grad_norm": 2.2417852878570557, "learning_rate": 1.9145788707923692e-05, "loss": 0.2040086030960083, "step": 6020 }, { "epoch": 0.15883939243273856, "grad_norm": 2.6018736362457275, "learning_rate": 1.914233522825461e-05, "loss": 0.18443677425384522, "step": 6030 }, { "epoch": 0.1591028076772373, "grad_norm": 3.929914712905884, "learning_rate": 1.913887509439951e-05, "loss": 0.17497631311416625, "step": 6040 }, { "epoch": 0.15936622292173605, "grad_norm": 3.927807092666626, "learning_rate": 1.9135408308876834e-05, "loss": 0.18157079219818115, "step": 6050 }, { "epoch": 0.15962963816623477, "grad_norm": 2.6829960346221924, "learning_rate": 1.913193487420986e-05, "loss": 0.1700118064880371, "step": 6060 }, { "epoch": 0.1598930534107335, "grad_norm": 1.7004002332687378, "learning_rate": 1.9128454792926694e-05, "loss": 0.17374904155731202, "step": 6070 }, { "epoch": 0.16015646865523225, "grad_norm": 3.7313759326934814, "learning_rate": 1.9124968067560296e-05, "loss": 0.1756625294685364, "step": 6080 }, { "epoch": 0.160419883899731, "grad_norm": 4.4033684730529785, "learning_rate": 1.9121474700648454e-05, "loss": 0.1788057804107666, "step": 6090 }, { "epoch": 0.1606832991442297, "grad_norm": 2.876469612121582, "learning_rate": 1.9117974694733795e-05, "loss": 0.17796764373779297, "step": 6100 }, { "epoch": 0.16094671438872846, "grad_norm": 3.2366857528686523, "learning_rate": 1.9114468052363764e-05, "loss": 0.18360531330108643, "step": 6110 }, { "epoch": 0.1612101296332272, "grad_norm": 2.07303524017334, "learning_rate": 1.9110954776090656e-05, "loss": 0.1929800033569336, "step": 6120 }, { "epoch": 0.16147354487772594, "grad_norm": 2.4361190795898438, "learning_rate": 1.910743486847158e-05, "loss": 0.1755685567855835, "step": 6130 }, { "epoch": 0.16173696012222466, "grad_norm": 2.1656265258789062, "learning_rate": 1.9103908332068483e-05, "loss": 0.1847392201423645, "step": 6140 }, { "epoch": 0.1620003753667234, "grad_norm": 3.336913585662842, "learning_rate": 1.910037516944812e-05, "loss": 0.16198515892028809, "step": 6150 }, { "epoch": 0.16226379061122215, "grad_norm": 3.235215425491333, "learning_rate": 1.909683538318209e-05, "loss": 0.18150067329406738, "step": 6160 }, { "epoch": 0.1625272058557209, "grad_norm": 3.1352078914642334, "learning_rate": 1.9093288975846797e-05, "loss": 0.1733603835105896, "step": 6170 }, { "epoch": 0.16279062110021963, "grad_norm": 2.196420907974243, "learning_rate": 1.9089735950023463e-05, "loss": 0.19044578075408936, "step": 6180 }, { "epoch": 0.16305403634471835, "grad_norm": 1.4295334815979004, "learning_rate": 1.9086176308298147e-05, "loss": 0.17691757678985595, "step": 6190 }, { "epoch": 0.1633174515892171, "grad_norm": 8.097099304199219, "learning_rate": 1.90826100532617e-05, "loss": 0.1869907259941101, "step": 6200 }, { "epoch": 0.16358086683371584, "grad_norm": 2.5174853801727295, "learning_rate": 1.90790371875098e-05, "loss": 0.17832096815109252, "step": 6210 }, { "epoch": 0.16384428207821458, "grad_norm": 1.863981008529663, "learning_rate": 1.9075457713642934e-05, "loss": 0.1894148111343384, "step": 6220 }, { "epoch": 0.1641076973227133, "grad_norm": 1.697574257850647, "learning_rate": 1.9071871634266392e-05, "loss": 0.17301665544509887, "step": 6230 }, { "epoch": 0.16437111256721204, "grad_norm": 1.3605891466140747, "learning_rate": 1.9068278951990282e-05, "loss": 0.17270268201828004, "step": 6240 }, { "epoch": 0.16463452781171078, "grad_norm": 5.530545234680176, "learning_rate": 1.9064679669429515e-05, "loss": 0.18301327228546144, "step": 6250 }, { "epoch": 0.16489794305620953, "grad_norm": 2.9259583950042725, "learning_rate": 1.9061073789203798e-05, "loss": 0.18135714530944824, "step": 6260 }, { "epoch": 0.16516135830070824, "grad_norm": 1.4272350072860718, "learning_rate": 1.9057461313937646e-05, "loss": 0.19034349918365479, "step": 6270 }, { "epoch": 0.165424773545207, "grad_norm": 1.822893738746643, "learning_rate": 1.905384224626038e-05, "loss": 0.19170016050338745, "step": 6280 }, { "epoch": 0.16568818878970573, "grad_norm": 4.489447116851807, "learning_rate": 1.9050216588806113e-05, "loss": 0.17925257682800294, "step": 6290 }, { "epoch": 0.16595160403420448, "grad_norm": 3.946599245071411, "learning_rate": 1.9046584344213754e-05, "loss": 0.16988165378570558, "step": 6300 }, { "epoch": 0.16621501927870322, "grad_norm": 3.6928012371063232, "learning_rate": 1.9042945515127004e-05, "loss": 0.1720714807510376, "step": 6310 }, { "epoch": 0.16647843452320193, "grad_norm": 2.7263965606689453, "learning_rate": 1.9039300104194366e-05, "loss": 0.16907591819763185, "step": 6320 }, { "epoch": 0.16674184976770068, "grad_norm": 2.084549903869629, "learning_rate": 1.903564811406912e-05, "loss": 0.18065807819366456, "step": 6330 }, { "epoch": 0.16700526501219942, "grad_norm": 3.5454201698303223, "learning_rate": 1.9031989547409347e-05, "loss": 0.16625385284423827, "step": 6340 }, { "epoch": 0.16726868025669817, "grad_norm": 1.2255574464797974, "learning_rate": 1.9028324406877905e-05, "loss": 0.1650179624557495, "step": 6350 }, { "epoch": 0.16753209550119688, "grad_norm": 2.056687116622925, "learning_rate": 1.9024652695142447e-05, "loss": 0.15664825439453126, "step": 6360 }, { "epoch": 0.16779551074569563, "grad_norm": 1.6763864755630493, "learning_rate": 1.90209744148754e-05, "loss": 0.17191965579986573, "step": 6370 }, { "epoch": 0.16805892599019437, "grad_norm": 4.70133638381958, "learning_rate": 1.901728956875398e-05, "loss": 0.16945090293884277, "step": 6380 }, { "epoch": 0.1683223412346931, "grad_norm": 2.346900224685669, "learning_rate": 1.9013598159460168e-05, "loss": 0.16121312379837036, "step": 6390 }, { "epoch": 0.16858575647919183, "grad_norm": 5.631246089935303, "learning_rate": 1.900990018968074e-05, "loss": 0.17040247917175294, "step": 6400 }, { "epoch": 0.16884917172369057, "grad_norm": 1.7377842664718628, "learning_rate": 1.900619566210723e-05, "loss": 0.15558654069900513, "step": 6410 }, { "epoch": 0.16911258696818932, "grad_norm": 1.590258240699768, "learning_rate": 1.900248457943596e-05, "loss": 0.17126671075820923, "step": 6420 }, { "epoch": 0.16937600221268806, "grad_norm": 5.393208980560303, "learning_rate": 1.899876694436801e-05, "loss": 0.18182395696640014, "step": 6430 }, { "epoch": 0.1696394174571868, "grad_norm": 2.2034103870391846, "learning_rate": 1.899504275960924e-05, "loss": 0.17607566118240356, "step": 6440 }, { "epoch": 0.16990283270168552, "grad_norm": 4.079049587249756, "learning_rate": 1.899131202787027e-05, "loss": 0.1785662889480591, "step": 6450 }, { "epoch": 0.17016624794618426, "grad_norm": 1.6862192153930664, "learning_rate": 1.8987574751866488e-05, "loss": 0.19698882102966309, "step": 6460 }, { "epoch": 0.170429663190683, "grad_norm": 5.530920505523682, "learning_rate": 1.8983830934318044e-05, "loss": 0.1893184542655945, "step": 6470 }, { "epoch": 0.17069307843518175, "grad_norm": 3.503452777862549, "learning_rate": 1.8980080577949848e-05, "loss": 0.1774419665336609, "step": 6480 }, { "epoch": 0.17095649367968047, "grad_norm": 3.200523614883423, "learning_rate": 1.8976323685491577e-05, "loss": 0.1902053952217102, "step": 6490 }, { "epoch": 0.1712199089241792, "grad_norm": 2.7144570350646973, "learning_rate": 1.8972560259677658e-05, "loss": 0.1828027129173279, "step": 6500 }, { "epoch": 0.17148332416867795, "grad_norm": 6.380727767944336, "learning_rate": 1.896879030324727e-05, "loss": 0.17056092023849487, "step": 6510 }, { "epoch": 0.1717467394131767, "grad_norm": 1.545305848121643, "learning_rate": 1.8965013818944357e-05, "loss": 0.1767922282218933, "step": 6520 }, { "epoch": 0.1720101546576754, "grad_norm": 1.8249284029006958, "learning_rate": 1.89612308095176e-05, "loss": 0.1814229369163513, "step": 6530 }, { "epoch": 0.17227356990217416, "grad_norm": 3.143597364425659, "learning_rate": 1.8957441277720438e-05, "loss": 0.19322807788848878, "step": 6540 }, { "epoch": 0.1725369851466729, "grad_norm": 1.4871182441711426, "learning_rate": 1.8953645226311056e-05, "loss": 0.17279186248779296, "step": 6550 }, { "epoch": 0.17280040039117164, "grad_norm": 2.8265464305877686, "learning_rate": 1.894984265805239e-05, "loss": 0.19253653287887573, "step": 6560 }, { "epoch": 0.1730638156356704, "grad_norm": 1.5155589580535889, "learning_rate": 1.8946033575712104e-05, "loss": 0.1708642601966858, "step": 6570 }, { "epoch": 0.1733272308801691, "grad_norm": 1.8124043941497803, "learning_rate": 1.894221798206262e-05, "loss": 0.1817082166671753, "step": 6580 }, { "epoch": 0.17359064612466785, "grad_norm": 3.47078800201416, "learning_rate": 1.8938395879881086e-05, "loss": 0.18443832397460938, "step": 6590 }, { "epoch": 0.1738540613691666, "grad_norm": 1.9204150438308716, "learning_rate": 1.8934567271949397e-05, "loss": 0.19286599159240722, "step": 6600 }, { "epoch": 0.17411747661366533, "grad_norm": 2.0755043029785156, "learning_rate": 1.8930732161054176e-05, "loss": 0.17542200088500975, "step": 6610 }, { "epoch": 0.17438089185816405, "grad_norm": 3.674717903137207, "learning_rate": 1.8926890549986786e-05, "loss": 0.1765366315841675, "step": 6620 }, { "epoch": 0.1746443071026628, "grad_norm": 2.6947410106658936, "learning_rate": 1.8923042441543313e-05, "loss": 0.18264222145080566, "step": 6630 }, { "epoch": 0.17490772234716154, "grad_norm": 5.398747444152832, "learning_rate": 1.891918783852458e-05, "loss": 0.18674042224884033, "step": 6640 }, { "epoch": 0.17517113759166028, "grad_norm": 1.285008430480957, "learning_rate": 1.8915326743736134e-05, "loss": 0.1637054443359375, "step": 6650 }, { "epoch": 0.175434552836159, "grad_norm": 3.465289831161499, "learning_rate": 1.8911459159988238e-05, "loss": 0.18807342052459716, "step": 6660 }, { "epoch": 0.17569796808065774, "grad_norm": 1.6161775588989258, "learning_rate": 1.8907585090095897e-05, "loss": 0.174782133102417, "step": 6670 }, { "epoch": 0.17596138332515648, "grad_norm": 2.5409748554229736, "learning_rate": 1.890370453687882e-05, "loss": 0.1881714344024658, "step": 6680 }, { "epoch": 0.17622479856965523, "grad_norm": 3.3372623920440674, "learning_rate": 1.889981750316144e-05, "loss": 0.1844981789588928, "step": 6690 }, { "epoch": 0.17648821381415397, "grad_norm": 2.6311469078063965, "learning_rate": 1.8895923991772913e-05, "loss": 0.1830836296081543, "step": 6700 }, { "epoch": 0.1767516290586527, "grad_norm": 2.3492867946624756, "learning_rate": 1.8892024005547104e-05, "loss": 0.19102814197540283, "step": 6710 }, { "epoch": 0.17701504430315143, "grad_norm": 5.964318752288818, "learning_rate": 1.8888117547322586e-05, "loss": 0.17706288099288942, "step": 6720 }, { "epoch": 0.17727845954765017, "grad_norm": 3.2148635387420654, "learning_rate": 1.888420461994265e-05, "loss": 0.18583123683929442, "step": 6730 }, { "epoch": 0.17754187479214892, "grad_norm": 3.6975717544555664, "learning_rate": 1.88802852262553e-05, "loss": 0.1735023856163025, "step": 6740 }, { "epoch": 0.17780529003664763, "grad_norm": 3.5513956546783447, "learning_rate": 1.887635936911323e-05, "loss": 0.19133095741271972, "step": 6750 }, { "epoch": 0.17806870528114638, "grad_norm": 2.213367223739624, "learning_rate": 1.8872427051373852e-05, "loss": 0.17970676422119142, "step": 6760 }, { "epoch": 0.17833212052564512, "grad_norm": 2.1530892848968506, "learning_rate": 1.886848827589928e-05, "loss": 0.17393827438354492, "step": 6770 }, { "epoch": 0.17859553577014387, "grad_norm": 2.4993655681610107, "learning_rate": 1.8864543045556318e-05, "loss": 0.1845411777496338, "step": 6780 }, { "epoch": 0.1788589510146426, "grad_norm": 5.780002593994141, "learning_rate": 1.8860591363216482e-05, "loss": 0.18877949714660644, "step": 6790 }, { "epoch": 0.17912236625914132, "grad_norm": 2.315476179122925, "learning_rate": 1.8856633231755972e-05, "loss": 0.18296804428100585, "step": 6800 }, { "epoch": 0.17938578150364007, "grad_norm": 1.908765196800232, "learning_rate": 1.8852668654055686e-05, "loss": 0.17126307487487794, "step": 6810 }, { "epoch": 0.1796491967481388, "grad_norm": 2.287569999694824, "learning_rate": 1.884869763300122e-05, "loss": 0.17863523960113525, "step": 6820 }, { "epoch": 0.17991261199263756, "grad_norm": 1.945277452468872, "learning_rate": 1.8844720171482854e-05, "loss": 0.1691671848297119, "step": 6830 }, { "epoch": 0.18017602723713627, "grad_norm": 2.623081684112549, "learning_rate": 1.884073627239555e-05, "loss": 0.17723450660705567, "step": 6840 }, { "epoch": 0.18043944248163502, "grad_norm": 4.383397102355957, "learning_rate": 1.8836745938638963e-05, "loss": 0.18752243518829345, "step": 6850 }, { "epoch": 0.18070285772613376, "grad_norm": 1.4518429040908813, "learning_rate": 1.8832749173117435e-05, "loss": 0.17964576482772826, "step": 6860 }, { "epoch": 0.1809662729706325, "grad_norm": 2.232084035873413, "learning_rate": 1.8828745978739982e-05, "loss": 0.18077175617218016, "step": 6870 }, { "epoch": 0.18122968821513122, "grad_norm": 1.4144248962402344, "learning_rate": 1.88247363584203e-05, "loss": 0.18070186376571656, "step": 6880 }, { "epoch": 0.18149310345962996, "grad_norm": 3.783716917037964, "learning_rate": 1.8820720315076767e-05, "loss": 0.1747065305709839, "step": 6890 }, { "epoch": 0.1817565187041287, "grad_norm": 3.617279529571533, "learning_rate": 1.8816697851632425e-05, "loss": 0.18161503076553345, "step": 6900 }, { "epoch": 0.18201993394862745, "grad_norm": 2.220163345336914, "learning_rate": 1.8812668971015006e-05, "loss": 0.18093885183334352, "step": 6910 }, { "epoch": 0.1822833491931262, "grad_norm": 2.033977508544922, "learning_rate": 1.88086336761569e-05, "loss": 0.17880477905273437, "step": 6920 }, { "epoch": 0.1825467644376249, "grad_norm": 0.9976582527160645, "learning_rate": 1.8804591969995167e-05, "loss": 0.18056353330612182, "step": 6930 }, { "epoch": 0.18281017968212365, "grad_norm": 2.3442888259887695, "learning_rate": 1.8800543855471535e-05, "loss": 0.1728384017944336, "step": 6940 }, { "epoch": 0.1830735949266224, "grad_norm": 2.0228824615478516, "learning_rate": 1.87964893355324e-05, "loss": 0.17657740116119386, "step": 6950 }, { "epoch": 0.18333701017112114, "grad_norm": 4.354208469390869, "learning_rate": 1.879242841312881e-05, "loss": 0.18518469333648682, "step": 6960 }, { "epoch": 0.18360042541561986, "grad_norm": 5.737354278564453, "learning_rate": 1.878836109121648e-05, "loss": 0.18113816976547242, "step": 6970 }, { "epoch": 0.1838638406601186, "grad_norm": 1.8749818801879883, "learning_rate": 1.8784287372755795e-05, "loss": 0.1869650363922119, "step": 6980 }, { "epoch": 0.18412725590461734, "grad_norm": 4.584497928619385, "learning_rate": 1.8780207260711767e-05, "loss": 0.18258335590362548, "step": 6990 }, { "epoch": 0.1843906711491161, "grad_norm": 2.088501214981079, "learning_rate": 1.8776120758054087e-05, "loss": 0.18976778984069825, "step": 7000 }, { "epoch": 0.1846540863936148, "grad_norm": 4.132274150848389, "learning_rate": 1.8772027867757082e-05, "loss": 0.17567107677459717, "step": 7010 }, { "epoch": 0.18491750163811355, "grad_norm": 4.033403396606445, "learning_rate": 1.8767928592799742e-05, "loss": 0.17583785057067872, "step": 7020 }, { "epoch": 0.1851809168826123, "grad_norm": 4.3038859367370605, "learning_rate": 1.876382293616569e-05, "loss": 0.18936904668807983, "step": 7030 }, { "epoch": 0.18544433212711103, "grad_norm": 1.9938068389892578, "learning_rate": 1.8759710900843202e-05, "loss": 0.1818368673324585, "step": 7040 }, { "epoch": 0.18570774737160978, "grad_norm": 1.830378770828247, "learning_rate": 1.8755592489825193e-05, "loss": 0.15985755920410155, "step": 7050 }, { "epoch": 0.1859711626161085, "grad_norm": 2.8414957523345947, "learning_rate": 1.8751467706109224e-05, "loss": 0.19400467872619628, "step": 7060 }, { "epoch": 0.18623457786060724, "grad_norm": 4.357273578643799, "learning_rate": 1.8747336552697483e-05, "loss": 0.17428827285766602, "step": 7070 }, { "epoch": 0.18649799310510598, "grad_norm": 1.6119614839553833, "learning_rate": 1.874319903259681e-05, "loss": 0.17004218101501464, "step": 7080 }, { "epoch": 0.18676140834960472, "grad_norm": 1.7253013849258423, "learning_rate": 1.873905514881867e-05, "loss": 0.19224650859832765, "step": 7090 }, { "epoch": 0.18702482359410344, "grad_norm": 5.851776599884033, "learning_rate": 1.873490490437915e-05, "loss": 0.1760157585144043, "step": 7100 }, { "epoch": 0.18728823883860218, "grad_norm": 5.3542327880859375, "learning_rate": 1.8730748302298993e-05, "loss": 0.1882850170135498, "step": 7110 }, { "epoch": 0.18755165408310093, "grad_norm": 3.6084065437316895, "learning_rate": 1.8726585345603544e-05, "loss": 0.20759997367858887, "step": 7120 }, { "epoch": 0.18781506932759967, "grad_norm": 1.2493661642074585, "learning_rate": 1.8722416037322778e-05, "loss": 0.18157811164855958, "step": 7130 }, { "epoch": 0.1880784845720984, "grad_norm": 2.520287275314331, "learning_rate": 1.871824038049131e-05, "loss": 0.1919010877609253, "step": 7140 }, { "epoch": 0.18834189981659713, "grad_norm": 2.638066053390503, "learning_rate": 1.871405837814835e-05, "loss": 0.17154104709625245, "step": 7150 }, { "epoch": 0.18860531506109587, "grad_norm": 1.3175578117370605, "learning_rate": 1.8709870033337753e-05, "loss": 0.19637441635131836, "step": 7160 }, { "epoch": 0.18886873030559462, "grad_norm": 4.867798328399658, "learning_rate": 1.8705675349107964e-05, "loss": 0.18043338060379027, "step": 7170 }, { "epoch": 0.18913214555009336, "grad_norm": 3.62184739112854, "learning_rate": 1.870147432851206e-05, "loss": 0.1763390064239502, "step": 7180 }, { "epoch": 0.18939556079459208, "grad_norm": 3.8450143337249756, "learning_rate": 1.8697266974607732e-05, "loss": 0.18399417400360107, "step": 7190 }, { "epoch": 0.18965897603909082, "grad_norm": 2.5629191398620605, "learning_rate": 1.8693053290457264e-05, "loss": 0.17024933099746703, "step": 7200 }, { "epoch": 0.18992239128358956, "grad_norm": 3.1625359058380127, "learning_rate": 1.868883327912756e-05, "loss": 0.1805041551589966, "step": 7210 }, { "epoch": 0.1901858065280883, "grad_norm": 1.1958237886428833, "learning_rate": 1.868460694369013e-05, "loss": 0.18767527341842652, "step": 7220 }, { "epoch": 0.19044922177258702, "grad_norm": 1.2842082977294922, "learning_rate": 1.868037428722108e-05, "loss": 0.18146030902862548, "step": 7230 }, { "epoch": 0.19071263701708577, "grad_norm": 2.1625897884368896, "learning_rate": 1.867613531280112e-05, "loss": 0.18299014568328859, "step": 7240 }, { "epoch": 0.1909760522615845, "grad_norm": 1.5595141649246216, "learning_rate": 1.867189002351556e-05, "loss": 0.17872996330261232, "step": 7250 }, { "epoch": 0.19123946750608325, "grad_norm": 4.36417818069458, "learning_rate": 1.8667638422454306e-05, "loss": 0.181874418258667, "step": 7260 }, { "epoch": 0.19150288275058197, "grad_norm": 2.6057581901550293, "learning_rate": 1.8663380512711852e-05, "loss": 0.17910640239715575, "step": 7270 }, { "epoch": 0.19176629799508071, "grad_norm": 1.2955493927001953, "learning_rate": 1.865911629738729e-05, "loss": 0.17751786708831788, "step": 7280 }, { "epoch": 0.19202971323957946, "grad_norm": 1.8645769357681274, "learning_rate": 1.86548457795843e-05, "loss": 0.1831357955932617, "step": 7290 }, { "epoch": 0.1922931284840782, "grad_norm": 2.4052863121032715, "learning_rate": 1.8650568962411154e-05, "loss": 0.18078296184539794, "step": 7300 }, { "epoch": 0.19255654372857695, "grad_norm": 2.2206881046295166, "learning_rate": 1.8646285848980693e-05, "loss": 0.17221521139144896, "step": 7310 }, { "epoch": 0.19281995897307566, "grad_norm": 6.361074924468994, "learning_rate": 1.8641996442410365e-05, "loss": 0.16016955375671388, "step": 7320 }, { "epoch": 0.1930833742175744, "grad_norm": 0.9186651706695557, "learning_rate": 1.8637700745822176e-05, "loss": 0.18602992296218873, "step": 7330 }, { "epoch": 0.19334678946207315, "grad_norm": 1.0072658061981201, "learning_rate": 1.8633398762342722e-05, "loss": 0.1797180414199829, "step": 7340 }, { "epoch": 0.1936102047065719, "grad_norm": 2.947813034057617, "learning_rate": 1.8629090495103175e-05, "loss": 0.1802757740020752, "step": 7350 }, { "epoch": 0.1938736199510706, "grad_norm": 2.7228240966796875, "learning_rate": 1.8624775947239272e-05, "loss": 0.16172583103179933, "step": 7360 }, { "epoch": 0.19413703519556935, "grad_norm": 4.383436679840088, "learning_rate": 1.862045512189133e-05, "loss": 0.17532564401626588, "step": 7370 }, { "epoch": 0.1944004504400681, "grad_norm": 2.517037868499756, "learning_rate": 1.8616128022204233e-05, "loss": 0.17541285753250122, "step": 7380 }, { "epoch": 0.19466386568456684, "grad_norm": 1.826460361480713, "learning_rate": 1.8611794651327432e-05, "loss": 0.1800481081008911, "step": 7390 }, { "epoch": 0.19492728092906556, "grad_norm": 1.187620997428894, "learning_rate": 1.8607455012414936e-05, "loss": 0.1734493374824524, "step": 7400 }, { "epoch": 0.1951906961735643, "grad_norm": 1.430274248123169, "learning_rate": 1.8603109108625327e-05, "loss": 0.1763633131980896, "step": 7410 }, { "epoch": 0.19545411141806304, "grad_norm": 2.884211540222168, "learning_rate": 1.859875694312174e-05, "loss": 0.16596102714538574, "step": 7420 }, { "epoch": 0.19571752666256179, "grad_norm": 1.3488999605178833, "learning_rate": 1.8594398519071867e-05, "loss": 0.18692057132720946, "step": 7430 }, { "epoch": 0.19598094190706053, "grad_norm": 1.2473856210708618, "learning_rate": 1.8590033839647955e-05, "loss": 0.18391668796539307, "step": 7440 }, { "epoch": 0.19624435715155925, "grad_norm": 0.8762473464012146, "learning_rate": 1.8585662908026814e-05, "loss": 0.16530634164810182, "step": 7450 }, { "epoch": 0.196507772396058, "grad_norm": 1.3593754768371582, "learning_rate": 1.8581285727389787e-05, "loss": 0.18369048833847046, "step": 7460 }, { "epoch": 0.19677118764055673, "grad_norm": 4.451939582824707, "learning_rate": 1.857690230092278e-05, "loss": 0.1902376651763916, "step": 7470 }, { "epoch": 0.19703460288505548, "grad_norm": 2.6925456523895264, "learning_rate": 1.8572512631816243e-05, "loss": 0.17324883937835694, "step": 7480 }, { "epoch": 0.1972980181295542, "grad_norm": 1.4972330331802368, "learning_rate": 1.856811672326516e-05, "loss": 0.16569185256958008, "step": 7490 }, { "epoch": 0.19756143337405294, "grad_norm": 1.8306926488876343, "learning_rate": 1.8563714578469064e-05, "loss": 0.18524487018585206, "step": 7500 }, { "epoch": 0.19782484861855168, "grad_norm": 4.482421398162842, "learning_rate": 1.8559306200632036e-05, "loss": 0.18542149066925048, "step": 7510 }, { "epoch": 0.19808826386305042, "grad_norm": 1.7900335788726807, "learning_rate": 1.8554891592962666e-05, "loss": 0.16782915592193604, "step": 7520 }, { "epoch": 0.19835167910754914, "grad_norm": 2.1064772605895996, "learning_rate": 1.8550470758674112e-05, "loss": 0.16778528690338135, "step": 7530 }, { "epoch": 0.19861509435204788, "grad_norm": 1.78360915184021, "learning_rate": 1.8546043700984036e-05, "loss": 0.17412524223327636, "step": 7540 }, { "epoch": 0.19887850959654663, "grad_norm": 2.384956121444702, "learning_rate": 1.8541610423114647e-05, "loss": 0.1849895238876343, "step": 7550 }, { "epoch": 0.19914192484104537, "grad_norm": 1.91452956199646, "learning_rate": 1.853717092829268e-05, "loss": 0.1845860004425049, "step": 7560 }, { "epoch": 0.1994053400855441, "grad_norm": 2.03829288482666, "learning_rate": 1.853272521974938e-05, "loss": 0.18717920780181885, "step": 7570 }, { "epoch": 0.19966875533004283, "grad_norm": 0.9642726182937622, "learning_rate": 1.8528273300720538e-05, "loss": 0.17625856399536133, "step": 7580 }, { "epoch": 0.19993217057454157, "grad_norm": 1.8118878602981567, "learning_rate": 1.8523815174446444e-05, "loss": 0.1714707612991333, "step": 7590 }, { "epoch": 0.20019558581904032, "grad_norm": 1.1768625974655151, "learning_rate": 1.851935084417192e-05, "loss": 0.16721351146698, "step": 7600 }, { "epoch": 0.20045900106353906, "grad_norm": 2.3959174156188965, "learning_rate": 1.8514880313146295e-05, "loss": 0.20215063095092772, "step": 7610 }, { "epoch": 0.20072241630803778, "grad_norm": 2.1300811767578125, "learning_rate": 1.8510403584623414e-05, "loss": 0.18263663053512574, "step": 7620 }, { "epoch": 0.20098583155253652, "grad_norm": 5.715440273284912, "learning_rate": 1.850592066186163e-05, "loss": 0.1730947732925415, "step": 7630 }, { "epoch": 0.20124924679703526, "grad_norm": 3.418882369995117, "learning_rate": 1.850143154812382e-05, "loss": 0.17087141275405884, "step": 7640 }, { "epoch": 0.201512662041534, "grad_norm": 2.515338182449341, "learning_rate": 1.849693624667734e-05, "loss": 0.17250216007232666, "step": 7650 }, { "epoch": 0.20177607728603272, "grad_norm": 2.2275071144104004, "learning_rate": 1.8492434760794072e-05, "loss": 0.17549772262573243, "step": 7660 }, { "epoch": 0.20203949253053147, "grad_norm": 2.0032095909118652, "learning_rate": 1.848792709375039e-05, "loss": 0.17599267959594728, "step": 7670 }, { "epoch": 0.2023029077750302, "grad_norm": 1.4310202598571777, "learning_rate": 1.848341324882717e-05, "loss": 0.1660658001899719, "step": 7680 }, { "epoch": 0.20256632301952895, "grad_norm": 0.9640329480171204, "learning_rate": 1.8478893229309784e-05, "loss": 0.18856385946273804, "step": 7690 }, { "epoch": 0.2028297382640277, "grad_norm": 1.2118650674819946, "learning_rate": 1.847436703848809e-05, "loss": 0.17961957454681396, "step": 7700 }, { "epoch": 0.20309315350852641, "grad_norm": 3.9127821922302246, "learning_rate": 1.8469834679656452e-05, "loss": 0.18634557723999023, "step": 7710 }, { "epoch": 0.20335656875302516, "grad_norm": 1.9664801359176636, "learning_rate": 1.8465296156113715e-05, "loss": 0.1757755994796753, "step": 7720 }, { "epoch": 0.2036199839975239, "grad_norm": 2.415213108062744, "learning_rate": 1.8460751471163212e-05, "loss": 0.19216008186340333, "step": 7730 }, { "epoch": 0.20388339924202264, "grad_norm": 2.6380348205566406, "learning_rate": 1.845620062811276e-05, "loss": 0.19100580215454102, "step": 7740 }, { "epoch": 0.20414681448652136, "grad_norm": 3.8237714767456055, "learning_rate": 1.8451643630274664e-05, "loss": 0.17323157787322999, "step": 7750 }, { "epoch": 0.2044102297310201, "grad_norm": 2.414480447769165, "learning_rate": 1.8447080480965694e-05, "loss": 0.18270835876464844, "step": 7760 }, { "epoch": 0.20467364497551885, "grad_norm": 2.580840826034546, "learning_rate": 1.844251118350711e-05, "loss": 0.1825932025909424, "step": 7770 }, { "epoch": 0.2049370602200176, "grad_norm": 2.367274522781372, "learning_rate": 1.8437935741224652e-05, "loss": 0.1824488401412964, "step": 7780 }, { "epoch": 0.2052004754645163, "grad_norm": 1.4428006410598755, "learning_rate": 1.8433354157448513e-05, "loss": 0.1852792263031006, "step": 7790 }, { "epoch": 0.20546389070901505, "grad_norm": 4.390959739685059, "learning_rate": 1.8428766435513373e-05, "loss": 0.18241257667541505, "step": 7800 }, { "epoch": 0.2057273059535138, "grad_norm": 5.333165645599365, "learning_rate": 1.8424172578758374e-05, "loss": 0.17825193405151368, "step": 7810 }, { "epoch": 0.20599072119801254, "grad_norm": 4.111297607421875, "learning_rate": 1.841957259052712e-05, "loss": 0.17152893543243408, "step": 7820 }, { "epoch": 0.20625413644251128, "grad_norm": 1.5920213460922241, "learning_rate": 1.8414966474167688e-05, "loss": 0.18052306175231933, "step": 7830 }, { "epoch": 0.20651755168701, "grad_norm": 5.933437347412109, "learning_rate": 1.8410354233032604e-05, "loss": 0.17354967594146728, "step": 7840 }, { "epoch": 0.20678096693150874, "grad_norm": 1.3040883541107178, "learning_rate": 1.840573587047885e-05, "loss": 0.17958674430847169, "step": 7850 }, { "epoch": 0.20704438217600749, "grad_norm": 1.7679927349090576, "learning_rate": 1.8401111389867882e-05, "loss": 0.18865548372268676, "step": 7860 }, { "epoch": 0.20730779742050623, "grad_norm": 3.845642328262329, "learning_rate": 1.8396480794565583e-05, "loss": 0.17668068408966064, "step": 7870 }, { "epoch": 0.20757121266500494, "grad_norm": 2.659982919692993, "learning_rate": 1.839184408794231e-05, "loss": 0.18559010028839112, "step": 7880 }, { "epoch": 0.2078346279095037, "grad_norm": 1.6284575462341309, "learning_rate": 1.8387201273372855e-05, "loss": 0.17796366214752196, "step": 7890 }, { "epoch": 0.20809804315400243, "grad_norm": 5.756608963012695, "learning_rate": 1.838255235423646e-05, "loss": 0.16659865379333497, "step": 7900 }, { "epoch": 0.20836145839850118, "grad_norm": 2.1990222930908203, "learning_rate": 1.8377897333916808e-05, "loss": 0.17429587841033936, "step": 7910 }, { "epoch": 0.2086248736429999, "grad_norm": 6.158779621124268, "learning_rate": 1.837323621580202e-05, "loss": 0.1672481656074524, "step": 7920 }, { "epoch": 0.20888828888749864, "grad_norm": 3.785738706588745, "learning_rate": 1.8368569003284662e-05, "loss": 0.17753723859786988, "step": 7930 }, { "epoch": 0.20915170413199738, "grad_norm": 1.724737524986267, "learning_rate": 1.8363895699761736e-05, "loss": 0.18115252256393433, "step": 7940 }, { "epoch": 0.20941511937649612, "grad_norm": 5.923300266265869, "learning_rate": 1.8359216308634668e-05, "loss": 0.16894410848617553, "step": 7950 }, { "epoch": 0.20967853462099487, "grad_norm": 1.0044196844100952, "learning_rate": 1.8354530833309323e-05, "loss": 0.1855271577835083, "step": 7960 }, { "epoch": 0.20994194986549358, "grad_norm": 3.59104061126709, "learning_rate": 1.8349839277195993e-05, "loss": 0.16316466331481932, "step": 7970 }, { "epoch": 0.21020536510999233, "grad_norm": 1.4269753694534302, "learning_rate": 1.8345141643709394e-05, "loss": 0.1781078815460205, "step": 7980 }, { "epoch": 0.21046878035449107, "grad_norm": 3.7057814598083496, "learning_rate": 1.8340437936268667e-05, "loss": 0.17744669914245606, "step": 7990 }, { "epoch": 0.2107321955989898, "grad_norm": 3.6147620677948, "learning_rate": 1.8335728158297374e-05, "loss": 0.1832594871520996, "step": 8000 }, { "epoch": 0.21099561084348853, "grad_norm": 0.8143312931060791, "learning_rate": 1.833101231322349e-05, "loss": 0.18268396854400634, "step": 8010 }, { "epoch": 0.21125902608798727, "grad_norm": 4.274045944213867, "learning_rate": 1.832629040447942e-05, "loss": 0.18351571559906005, "step": 8020 }, { "epoch": 0.21152244133248602, "grad_norm": 2.029146194458008, "learning_rate": 1.8321562435501963e-05, "loss": 0.19355878829956055, "step": 8030 }, { "epoch": 0.21178585657698476, "grad_norm": 2.7647223472595215, "learning_rate": 1.831682840973234e-05, "loss": 0.17343707084655763, "step": 8040 }, { "epoch": 0.21204927182148348, "grad_norm": 3.225268602371216, "learning_rate": 1.8312088330616188e-05, "loss": 0.18000868558883668, "step": 8050 }, { "epoch": 0.21231268706598222, "grad_norm": 1.5547890663146973, "learning_rate": 1.8307342201603533e-05, "loss": 0.1662738561630249, "step": 8060 }, { "epoch": 0.21257610231048096, "grad_norm": 1.7140817642211914, "learning_rate": 1.830259002614881e-05, "loss": 0.15906250476837158, "step": 8070 }, { "epoch": 0.2128395175549797, "grad_norm": 1.3796790838241577, "learning_rate": 1.829783180771087e-05, "loss": 0.18632394075393677, "step": 8080 }, { "epoch": 0.21310293279947845, "grad_norm": 1.5006600618362427, "learning_rate": 1.8293067549752937e-05, "loss": 0.18067140579223634, "step": 8090 }, { "epoch": 0.21336634804397717, "grad_norm": 1.256265640258789, "learning_rate": 1.8288297255742646e-05, "loss": 0.19308886528015137, "step": 8100 }, { "epoch": 0.2136297632884759, "grad_norm": 2.981459617614746, "learning_rate": 1.828352092915203e-05, "loss": 0.18349683284759521, "step": 8110 }, { "epoch": 0.21389317853297465, "grad_norm": 1.3726598024368286, "learning_rate": 1.8278738573457496e-05, "loss": 0.15871710777282716, "step": 8120 }, { "epoch": 0.2141565937774734, "grad_norm": 1.2386916875839233, "learning_rate": 1.8273950192139854e-05, "loss": 0.17208869457244874, "step": 8130 }, { "epoch": 0.2144200090219721, "grad_norm": 3.9373884201049805, "learning_rate": 1.826915578868429e-05, "loss": 0.19085489511489867, "step": 8140 }, { "epoch": 0.21468342426647086, "grad_norm": 2.617427110671997, "learning_rate": 1.8264355366580387e-05, "loss": 0.17592272758483887, "step": 8150 }, { "epoch": 0.2149468395109696, "grad_norm": 1.4980970621109009, "learning_rate": 1.825954892932209e-05, "loss": 0.16803033351898194, "step": 8160 }, { "epoch": 0.21521025475546834, "grad_norm": 2.444868326187134, "learning_rate": 1.8254736480407738e-05, "loss": 0.17277626991271972, "step": 8170 }, { "epoch": 0.21547366999996706, "grad_norm": 1.5495022535324097, "learning_rate": 1.8249918023340036e-05, "loss": 0.16895663738250732, "step": 8180 }, { "epoch": 0.2157370852444658, "grad_norm": 4.810120582580566, "learning_rate": 1.8245093561626063e-05, "loss": 0.17509688138961793, "step": 8190 }, { "epoch": 0.21600050048896455, "grad_norm": 4.463722229003906, "learning_rate": 1.824026309877727e-05, "loss": 0.19278261661529542, "step": 8200 }, { "epoch": 0.2162639157334633, "grad_norm": 3.5976638793945312, "learning_rate": 1.8235426638309483e-05, "loss": 0.18753540515899658, "step": 8210 }, { "epoch": 0.21652733097796203, "grad_norm": 1.6523970365524292, "learning_rate": 1.823058418374288e-05, "loss": 0.17901740074157715, "step": 8220 }, { "epoch": 0.21679074622246075, "grad_norm": 1.9603666067123413, "learning_rate": 1.822573573860201e-05, "loss": 0.16952236890792846, "step": 8230 }, { "epoch": 0.2170541614669595, "grad_norm": 1.5278851985931396, "learning_rate": 1.8220881306415783e-05, "loss": 0.17177388668060303, "step": 8240 }, { "epoch": 0.21731757671145824, "grad_norm": 1.7199827432632446, "learning_rate": 1.821602089071746e-05, "loss": 0.1746244788169861, "step": 8250 }, { "epoch": 0.21758099195595698, "grad_norm": 1.5061224699020386, "learning_rate": 1.8211154495044666e-05, "loss": 0.18921477794647218, "step": 8260 }, { "epoch": 0.2178444072004557, "grad_norm": 1.4529812335968018, "learning_rate": 1.8206282122939368e-05, "loss": 0.16534104347229003, "step": 8270 }, { "epoch": 0.21810782244495444, "grad_norm": 1.8083614110946655, "learning_rate": 1.8201403777947896e-05, "loss": 0.17731733322143556, "step": 8280 }, { "epoch": 0.21837123768945318, "grad_norm": 1.0524346828460693, "learning_rate": 1.8196519463620912e-05, "loss": 0.1805946111679077, "step": 8290 }, { "epoch": 0.21863465293395193, "grad_norm": 1.4378844499588013, "learning_rate": 1.8191629183513435e-05, "loss": 0.1747182011604309, "step": 8300 }, { "epoch": 0.21889806817845064, "grad_norm": 1.4149177074432373, "learning_rate": 1.8186732941184824e-05, "loss": 0.1775124788284302, "step": 8310 }, { "epoch": 0.2191614834229494, "grad_norm": 1.7459492683410645, "learning_rate": 1.818183074019877e-05, "loss": 0.17315294742584228, "step": 8320 }, { "epoch": 0.21942489866744813, "grad_norm": 2.650243043899536, "learning_rate": 1.8176922584123308e-05, "loss": 0.1648319721221924, "step": 8330 }, { "epoch": 0.21968831391194688, "grad_norm": 3.4491753578186035, "learning_rate": 1.817200847653081e-05, "loss": 0.1934752106666565, "step": 8340 }, { "epoch": 0.21995172915644562, "grad_norm": 1.6645146608352661, "learning_rate": 1.8167088420997975e-05, "loss": 0.19660580158233643, "step": 8350 }, { "epoch": 0.22021514440094433, "grad_norm": 3.2331318855285645, "learning_rate": 1.816216242110582e-05, "loss": 0.1840837001800537, "step": 8360 }, { "epoch": 0.22047855964544308, "grad_norm": 1.765446662902832, "learning_rate": 1.815723048043972e-05, "loss": 0.1636873722076416, "step": 8370 }, { "epoch": 0.22074197488994182, "grad_norm": 2.8227553367614746, "learning_rate": 1.8152292602589336e-05, "loss": 0.18478281497955323, "step": 8380 }, { "epoch": 0.22100539013444057, "grad_norm": 1.1731518507003784, "learning_rate": 1.8147348791148678e-05, "loss": 0.19027907848358155, "step": 8390 }, { "epoch": 0.22126880537893928, "grad_norm": 0.9337196350097656, "learning_rate": 1.8142399049716065e-05, "loss": 0.17180044651031495, "step": 8400 }, { "epoch": 0.22153222062343803, "grad_norm": 1.7544597387313843, "learning_rate": 1.8137443381894126e-05, "loss": 0.1859952688217163, "step": 8410 }, { "epoch": 0.22179563586793677, "grad_norm": 1.4610549211502075, "learning_rate": 1.813248179128982e-05, "loss": 0.17265784740447998, "step": 8420 }, { "epoch": 0.2220590511124355, "grad_norm": 1.4362907409667969, "learning_rate": 1.81275142815144e-05, "loss": 0.19077266454696656, "step": 8430 }, { "epoch": 0.22232246635693423, "grad_norm": 2.7194604873657227, "learning_rate": 1.812254085618343e-05, "loss": 0.1870494842529297, "step": 8440 }, { "epoch": 0.22258588160143297, "grad_norm": 3.651970148086548, "learning_rate": 1.8117561518916792e-05, "loss": 0.1826285243034363, "step": 8450 }, { "epoch": 0.22284929684593172, "grad_norm": 2.3415610790252686, "learning_rate": 1.811257627333866e-05, "loss": 0.17799110412597657, "step": 8460 }, { "epoch": 0.22311271209043046, "grad_norm": 1.0388944149017334, "learning_rate": 1.810758512307751e-05, "loss": 0.17244677543640136, "step": 8470 }, { "epoch": 0.2233761273349292, "grad_norm": 4.2617716789245605, "learning_rate": 1.8102588071766116e-05, "loss": 0.17887802124023439, "step": 8480 }, { "epoch": 0.22363954257942792, "grad_norm": 2.6614322662353516, "learning_rate": 1.8097585123041547e-05, "loss": 0.1736769199371338, "step": 8490 }, { "epoch": 0.22390295782392666, "grad_norm": 4.069907188415527, "learning_rate": 1.809257628054517e-05, "loss": 0.16937180757522582, "step": 8500 }, { "epoch": 0.2241663730684254, "grad_norm": 1.9057620763778687, "learning_rate": 1.808756154792263e-05, "loss": 0.17794935703277587, "step": 8510 }, { "epoch": 0.22442978831292415, "grad_norm": 4.113516807556152, "learning_rate": 1.8082540928823866e-05, "loss": 0.18284544944763184, "step": 8520 }, { "epoch": 0.22469320355742287, "grad_norm": 1.2685917615890503, "learning_rate": 1.8077514426903107e-05, "loss": 0.18190257549285888, "step": 8530 }, { "epoch": 0.2249566188019216, "grad_norm": 2.7826569080352783, "learning_rate": 1.8072482045818852e-05, "loss": 0.16682810783386232, "step": 8540 }, { "epoch": 0.22522003404642035, "grad_norm": 2.910146474838257, "learning_rate": 1.806744378923389e-05, "loss": 0.18080148696899415, "step": 8550 }, { "epoch": 0.2254834492909191, "grad_norm": 2.968749761581421, "learning_rate": 1.8062399660815277e-05, "loss": 0.1788671612739563, "step": 8560 }, { "epoch": 0.2257468645354178, "grad_norm": 2.6836888790130615, "learning_rate": 1.8057349664234348e-05, "loss": 0.1670302391052246, "step": 8570 }, { "epoch": 0.22601027977991656, "grad_norm": 1.5998092889785767, "learning_rate": 1.8052293803166714e-05, "loss": 0.19337258338928223, "step": 8580 }, { "epoch": 0.2262736950244153, "grad_norm": 1.8480579853057861, "learning_rate": 1.8047232081292242e-05, "loss": 0.17982136011123656, "step": 8590 }, { "epoch": 0.22653711026891404, "grad_norm": 1.8097938299179077, "learning_rate": 1.8042164502295074e-05, "loss": 0.1933764934539795, "step": 8600 }, { "epoch": 0.2268005255134128, "grad_norm": 0.8172866702079773, "learning_rate": 1.8037091069863615e-05, "loss": 0.17228543758392334, "step": 8610 }, { "epoch": 0.2270639407579115, "grad_norm": 2.4655511379241943, "learning_rate": 1.8032011787690523e-05, "loss": 0.17305411100387574, "step": 8620 }, { "epoch": 0.22732735600241025, "grad_norm": 3.233267068862915, "learning_rate": 1.8026926659472727e-05, "loss": 0.17404841184616088, "step": 8630 }, { "epoch": 0.227590771246909, "grad_norm": 1.5306425094604492, "learning_rate": 1.802183568891139e-05, "loss": 0.17909005880355836, "step": 8640 }, { "epoch": 0.22785418649140773, "grad_norm": 3.4158730506896973, "learning_rate": 1.801673887971195e-05, "loss": 0.19664801359176637, "step": 8650 }, { "epoch": 0.22811760173590645, "grad_norm": 2.3812716007232666, "learning_rate": 1.801163623558408e-05, "loss": 0.18578813076019288, "step": 8660 }, { "epoch": 0.2283810169804052, "grad_norm": 1.5880799293518066, "learning_rate": 1.800652776024171e-05, "loss": 0.17411861419677735, "step": 8670 }, { "epoch": 0.22864443222490394, "grad_norm": 3.1676669120788574, "learning_rate": 1.8001413457403005e-05, "loss": 0.17553358078002929, "step": 8680 }, { "epoch": 0.22890784746940268, "grad_norm": 2.0541794300079346, "learning_rate": 1.7996293330790374e-05, "loss": 0.1760097026824951, "step": 8690 }, { "epoch": 0.2291712627139014, "grad_norm": 1.281325340270996, "learning_rate": 1.7991167384130466e-05, "loss": 0.17371734380722045, "step": 8700 }, { "epoch": 0.22943467795840014, "grad_norm": 1.3670188188552856, "learning_rate": 1.798603562115417e-05, "loss": 0.17281912565231322, "step": 8710 }, { "epoch": 0.22969809320289888, "grad_norm": 1.2536063194274902, "learning_rate": 1.79808980455966e-05, "loss": 0.16401911973953248, "step": 8720 }, { "epoch": 0.22996150844739763, "grad_norm": 1.1549088954925537, "learning_rate": 1.7975754661197106e-05, "loss": 0.1806270956993103, "step": 8730 }, { "epoch": 0.23022492369189637, "grad_norm": 2.6539149284362793, "learning_rate": 1.797060547169927e-05, "loss": 0.1754382610321045, "step": 8740 }, { "epoch": 0.2304883389363951, "grad_norm": 1.4248253107070923, "learning_rate": 1.796545048085089e-05, "loss": 0.17097171545028686, "step": 8750 }, { "epoch": 0.23075175418089383, "grad_norm": 1.1385341882705688, "learning_rate": 1.796028969240399e-05, "loss": 0.17896625995635987, "step": 8760 }, { "epoch": 0.23101516942539257, "grad_norm": 1.7672652006149292, "learning_rate": 1.7955123110114815e-05, "loss": 0.17596333026885985, "step": 8770 }, { "epoch": 0.23127858466989132, "grad_norm": 1.3710050582885742, "learning_rate": 1.794995073774383e-05, "loss": 0.16349754333496094, "step": 8780 }, { "epoch": 0.23154199991439003, "grad_norm": 1.1364151239395142, "learning_rate": 1.7944772579055705e-05, "loss": 0.17881594896316527, "step": 8790 }, { "epoch": 0.23180541515888878, "grad_norm": 2.9336705207824707, "learning_rate": 1.793958863781933e-05, "loss": 0.17256773710250856, "step": 8800 }, { "epoch": 0.23206883040338752, "grad_norm": 3.8654239177703857, "learning_rate": 1.7934398917807802e-05, "loss": 0.19902503490447998, "step": 8810 }, { "epoch": 0.23233224564788627, "grad_norm": 1.5671660900115967, "learning_rate": 1.7929203422798426e-05, "loss": 0.17218968868255616, "step": 8820 }, { "epoch": 0.23259566089238498, "grad_norm": 2.119943141937256, "learning_rate": 1.79240021565727e-05, "loss": 0.16648898124694825, "step": 8830 }, { "epoch": 0.23285907613688372, "grad_norm": 5.291749000549316, "learning_rate": 1.7918795122916337e-05, "loss": 0.180749249458313, "step": 8840 }, { "epoch": 0.23312249138138247, "grad_norm": 1.2312241792678833, "learning_rate": 1.7913582325619236e-05, "loss": 0.17687859535217285, "step": 8850 }, { "epoch": 0.2333859066258812, "grad_norm": 4.036336421966553, "learning_rate": 1.7908363768475498e-05, "loss": 0.18256515264511108, "step": 8860 }, { "epoch": 0.23364932187037996, "grad_norm": 2.1628212928771973, "learning_rate": 1.7903139455283412e-05, "loss": 0.1749199390411377, "step": 8870 }, { "epoch": 0.23391273711487867, "grad_norm": 4.565154075622559, "learning_rate": 1.7897909389845462e-05, "loss": 0.17339141368865968, "step": 8880 }, { "epoch": 0.23417615235937742, "grad_norm": 5.625838279724121, "learning_rate": 1.789267357596831e-05, "loss": 0.1721548318862915, "step": 8890 }, { "epoch": 0.23443956760387616, "grad_norm": 1.8468177318572998, "learning_rate": 1.7887432017462812e-05, "loss": 0.1864609718322754, "step": 8900 }, { "epoch": 0.2347029828483749, "grad_norm": 5.640969276428223, "learning_rate": 1.7882184718144002e-05, "loss": 0.1858208417892456, "step": 8910 }, { "epoch": 0.23496639809287362, "grad_norm": 3.2828729152679443, "learning_rate": 1.7876931681831087e-05, "loss": 0.17834837436676027, "step": 8920 }, { "epoch": 0.23522981333737236, "grad_norm": 1.6279432773590088, "learning_rate": 1.787167291234745e-05, "loss": 0.17377426624298095, "step": 8930 }, { "epoch": 0.2354932285818711, "grad_norm": 5.721900463104248, "learning_rate": 1.7866408413520662e-05, "loss": 0.17425363063812255, "step": 8940 }, { "epoch": 0.23575664382636985, "grad_norm": 2.320150136947632, "learning_rate": 1.786113818918244e-05, "loss": 0.19206427335739135, "step": 8950 }, { "epoch": 0.23602005907086857, "grad_norm": 3.4263596534729004, "learning_rate": 1.7855862243168684e-05, "loss": 0.1634758472442627, "step": 8960 }, { "epoch": 0.2362834743153673, "grad_norm": 2.220841646194458, "learning_rate": 1.785058057931946e-05, "loss": 0.17972133159637452, "step": 8970 }, { "epoch": 0.23654688955986605, "grad_norm": 3.380962371826172, "learning_rate": 1.7845293201478985e-05, "loss": 0.19756007194519043, "step": 8980 }, { "epoch": 0.2368103048043648, "grad_norm": 2.213886022567749, "learning_rate": 1.7840000113495647e-05, "loss": 0.1749347925186157, "step": 8990 }, { "epoch": 0.23707372004886354, "grad_norm": 4.758527755737305, "learning_rate": 1.7834701319221978e-05, "loss": 0.17074074745178222, "step": 9000 }, { "epoch": 0.23733713529336226, "grad_norm": 1.609146237373352, "learning_rate": 1.782939682251467e-05, "loss": 0.18267062902450562, "step": 9010 }, { "epoch": 0.237600550537861, "grad_norm": 2.8563895225524902, "learning_rate": 1.7824086627234565e-05, "loss": 0.17960273027420043, "step": 9020 }, { "epoch": 0.23786396578235974, "grad_norm": 4.526554584503174, "learning_rate": 1.7818770737246653e-05, "loss": 0.18345229625701903, "step": 9030 }, { "epoch": 0.2381273810268585, "grad_norm": 1.0846542119979858, "learning_rate": 1.7813449156420072e-05, "loss": 0.14967105388641358, "step": 9040 }, { "epoch": 0.2383907962713572, "grad_norm": 1.9014506340026855, "learning_rate": 1.7808121888628093e-05, "loss": 0.18148276805877686, "step": 9050 }, { "epoch": 0.23865421151585595, "grad_norm": 2.1397290229797363, "learning_rate": 1.780278893774813e-05, "loss": 0.1788839340209961, "step": 9060 }, { "epoch": 0.2389176267603547, "grad_norm": 3.209590196609497, "learning_rate": 1.7797450307661733e-05, "loss": 0.1791624069213867, "step": 9070 }, { "epoch": 0.23918104200485343, "grad_norm": 3.197964906692505, "learning_rate": 1.77921060022546e-05, "loss": 0.17361714839935302, "step": 9080 }, { "epoch": 0.23944445724935215, "grad_norm": 1.8734122514724731, "learning_rate": 1.7786756025416535e-05, "loss": 0.1883077383041382, "step": 9090 }, { "epoch": 0.2397078724938509, "grad_norm": 2.357625961303711, "learning_rate": 1.7781400381041484e-05, "loss": 0.18341244459152223, "step": 9100 }, { "epoch": 0.23997128773834964, "grad_norm": 1.4281054735183716, "learning_rate": 1.777603907302752e-05, "loss": 0.17886494398117064, "step": 9110 }, { "epoch": 0.24023470298284838, "grad_norm": 1.8790905475616455, "learning_rate": 1.7770672105276833e-05, "loss": 0.1575465202331543, "step": 9120 }, { "epoch": 0.24049811822734712, "grad_norm": 3.839090347290039, "learning_rate": 1.776529948169573e-05, "loss": 0.17910523414611818, "step": 9130 }, { "epoch": 0.24076153347184584, "grad_norm": 1.8183900117874146, "learning_rate": 1.775992120619464e-05, "loss": 0.18334107398986815, "step": 9140 }, { "epoch": 0.24102494871634458, "grad_norm": 1.5823229551315308, "learning_rate": 1.7754537282688104e-05, "loss": 0.17507179975509643, "step": 9150 }, { "epoch": 0.24128836396084333, "grad_norm": 3.2451224327087402, "learning_rate": 1.7749147715094777e-05, "loss": 0.16973350048065186, "step": 9160 }, { "epoch": 0.24155177920534207, "grad_norm": 0.9676021933555603, "learning_rate": 1.7743752507337414e-05, "loss": 0.1811367392539978, "step": 9170 }, { "epoch": 0.2418151944498408, "grad_norm": 2.855619192123413, "learning_rate": 1.7738351663342887e-05, "loss": 0.1762377381324768, "step": 9180 }, { "epoch": 0.24207860969433953, "grad_norm": 2.346205472946167, "learning_rate": 1.7732945187042155e-05, "loss": 0.1674528479576111, "step": 9190 }, { "epoch": 0.24234202493883827, "grad_norm": 1.4201799631118774, "learning_rate": 1.7727533082370288e-05, "loss": 0.1645021915435791, "step": 9200 }, { "epoch": 0.24260544018333702, "grad_norm": 1.1332969665527344, "learning_rate": 1.772211535326645e-05, "loss": 0.19045376777648926, "step": 9210 }, { "epoch": 0.24286885542783573, "grad_norm": 2.345177173614502, "learning_rate": 1.7716692003673898e-05, "loss": 0.17151954174041747, "step": 9220 }, { "epoch": 0.24313227067233448, "grad_norm": 2.4640932083129883, "learning_rate": 1.7711263037539978e-05, "loss": 0.19792098999023439, "step": 9230 }, { "epoch": 0.24339568591683322, "grad_norm": 2.145721673965454, "learning_rate": 1.770582845881613e-05, "loss": 0.17913296222686767, "step": 9240 }, { "epoch": 0.24365910116133196, "grad_norm": 1.3234084844589233, "learning_rate": 1.7700388271457872e-05, "loss": 0.17053627967834473, "step": 9250 }, { "epoch": 0.2439225164058307, "grad_norm": 1.196284294128418, "learning_rate": 1.7694942479424803e-05, "loss": 0.17947877645492555, "step": 9260 }, { "epoch": 0.24418593165032942, "grad_norm": 1.671730399131775, "learning_rate": 1.768949108668061e-05, "loss": 0.1655193567276001, "step": 9270 }, { "epoch": 0.24444934689482817, "grad_norm": 1.4502537250518799, "learning_rate": 1.7684034097193055e-05, "loss": 0.16050630807876587, "step": 9280 }, { "epoch": 0.2447127621393269, "grad_norm": 4.736251354217529, "learning_rate": 1.7678571514933966e-05, "loss": 0.17801623344421386, "step": 9290 }, { "epoch": 0.24497617738382566, "grad_norm": 3.182889699935913, "learning_rate": 1.7673103343879242e-05, "loss": 0.1732776641845703, "step": 9300 }, { "epoch": 0.24523959262832437, "grad_norm": 1.9029914140701294, "learning_rate": 1.7667629588008862e-05, "loss": 0.16476351022720337, "step": 9310 }, { "epoch": 0.24550300787282311, "grad_norm": 1.4873477220535278, "learning_rate": 1.766215025130686e-05, "loss": 0.1772567629814148, "step": 9320 }, { "epoch": 0.24576642311732186, "grad_norm": 1.189774751663208, "learning_rate": 1.7656665337761327e-05, "loss": 0.1773827314376831, "step": 9330 }, { "epoch": 0.2460298383618206, "grad_norm": 1.1805791854858398, "learning_rate": 1.7651174851364427e-05, "loss": 0.16899609565734863, "step": 9340 }, { "epoch": 0.24629325360631932, "grad_norm": 6.904183387756348, "learning_rate": 1.764567879611237e-05, "loss": 0.17799081802368164, "step": 9350 }, { "epoch": 0.24655666885081806, "grad_norm": 2.7086477279663086, "learning_rate": 1.7640177176005425e-05, "loss": 0.16798043251037598, "step": 9360 }, { "epoch": 0.2468200840953168, "grad_norm": 1.4868760108947754, "learning_rate": 1.7634669995047904e-05, "loss": 0.18403842449188232, "step": 9370 }, { "epoch": 0.24708349933981555, "grad_norm": 1.941258430480957, "learning_rate": 1.7629157257248175e-05, "loss": 0.16389799118041992, "step": 9380 }, { "epoch": 0.2473469145843143, "grad_norm": 1.0524721145629883, "learning_rate": 1.762363896661864e-05, "loss": 0.1699079155921936, "step": 9390 }, { "epoch": 0.247610329828813, "grad_norm": 3.9324328899383545, "learning_rate": 1.761811512717576e-05, "loss": 0.17275586128234863, "step": 9400 }, { "epoch": 0.24787374507331175, "grad_norm": 1.7293500900268555, "learning_rate": 1.761258574294001e-05, "loss": 0.16042687892913818, "step": 9410 }, { "epoch": 0.2481371603178105, "grad_norm": 2.4302947521209717, "learning_rate": 1.760705081793593e-05, "loss": 0.18526241779327393, "step": 9420 }, { "epoch": 0.24840057556230924, "grad_norm": 3.69242000579834, "learning_rate": 1.760151035619206e-05, "loss": 0.19372878074645997, "step": 9430 }, { "epoch": 0.24866399080680796, "grad_norm": 1.3716877698898315, "learning_rate": 1.7595964361741e-05, "loss": 0.170262610912323, "step": 9440 }, { "epoch": 0.2489274060513067, "grad_norm": 4.110414505004883, "learning_rate": 1.759041283861936e-05, "loss": 0.17800135612487794, "step": 9450 }, { "epoch": 0.24919082129580544, "grad_norm": 1.3877794742584229, "learning_rate": 1.7584855790867773e-05, "loss": 0.17220735549926758, "step": 9460 }, { "epoch": 0.2494542365403042, "grad_norm": 1.6000826358795166, "learning_rate": 1.7579293222530907e-05, "loss": 0.17720617055892945, "step": 9470 }, { "epoch": 0.24971765178480293, "grad_norm": 1.4298291206359863, "learning_rate": 1.7573725137657432e-05, "loss": 0.17512266635894774, "step": 9480 }, { "epoch": 0.24998106702930165, "grad_norm": 2.84328556060791, "learning_rate": 1.756815154030005e-05, "loss": 0.17593467235565186, "step": 9490 }, { "epoch": 0.2502444822738004, "grad_norm": 1.1374009847640991, "learning_rate": 1.756257243451545e-05, "loss": 0.16383895874023438, "step": 9500 }, { "epoch": 0.2505078975182991, "grad_norm": 1.7199004888534546, "learning_rate": 1.7556987824364363e-05, "loss": 0.17256693840026854, "step": 9510 }, { "epoch": 0.25077131276279785, "grad_norm": 1.7729095220565796, "learning_rate": 1.75513977139115e-05, "loss": 0.17438486814498902, "step": 9520 }, { "epoch": 0.2510347280072966, "grad_norm": 4.6396403312683105, "learning_rate": 1.7545802107225585e-05, "loss": 0.1861712694168091, "step": 9530 }, { "epoch": 0.25129814325179534, "grad_norm": 1.379565715789795, "learning_rate": 1.754020100837934e-05, "loss": 0.1777747631072998, "step": 9540 }, { "epoch": 0.2515615584962941, "grad_norm": 1.7625566720962524, "learning_rate": 1.753459442144949e-05, "loss": 0.18934500217437744, "step": 9550 }, { "epoch": 0.2518249737407928, "grad_norm": 4.135609149932861, "learning_rate": 1.7528982350516754e-05, "loss": 0.1692408561706543, "step": 9560 }, { "epoch": 0.25208838898529157, "grad_norm": 2.235125780105591, "learning_rate": 1.7523364799665832e-05, "loss": 0.16721247434616088, "step": 9570 }, { "epoch": 0.2523518042297903, "grad_norm": 1.5637704133987427, "learning_rate": 1.7517741772985424e-05, "loss": 0.17692850828170775, "step": 9580 }, { "epoch": 0.252615219474289, "grad_norm": 0.8999918103218079, "learning_rate": 1.751211327456821e-05, "loss": 0.18383384943008424, "step": 9590 }, { "epoch": 0.25287863471878774, "grad_norm": 4.1175408363342285, "learning_rate": 1.7506479308510855e-05, "loss": 0.17355213165283204, "step": 9600 }, { "epoch": 0.2531420499632865, "grad_norm": 4.3445963859558105, "learning_rate": 1.7500839878914004e-05, "loss": 0.1708474040031433, "step": 9610 }, { "epoch": 0.25340546520778523, "grad_norm": 2.7495064735412598, "learning_rate": 1.749519498988227e-05, "loss": 0.19137132167816162, "step": 9620 }, { "epoch": 0.253668880452284, "grad_norm": 4.060660362243652, "learning_rate": 1.748954464552426e-05, "loss": 0.17377195358276368, "step": 9630 }, { "epoch": 0.2539322956967827, "grad_norm": 1.4273996353149414, "learning_rate": 1.7483888849952525e-05, "loss": 0.18238874673843383, "step": 9640 }, { "epoch": 0.25419571094128146, "grad_norm": 2.3983867168426514, "learning_rate": 1.7478227607283604e-05, "loss": 0.16775416135787963, "step": 9650 }, { "epoch": 0.2544591261857802, "grad_norm": 1.3844480514526367, "learning_rate": 1.7472560921637993e-05, "loss": 0.17683136463165283, "step": 9660 }, { "epoch": 0.25472254143027895, "grad_norm": 4.140798568725586, "learning_rate": 1.746688879714015e-05, "loss": 0.17453036308288575, "step": 9670 }, { "epoch": 0.25498595667477764, "grad_norm": 0.9054309129714966, "learning_rate": 1.746121123791849e-05, "loss": 0.17064249515533447, "step": 9680 }, { "epoch": 0.2552493719192764, "grad_norm": 4.46489143371582, "learning_rate": 1.7455528248105385e-05, "loss": 0.19045886993408204, "step": 9690 }, { "epoch": 0.2555127871637751, "grad_norm": 1.0649560689926147, "learning_rate": 1.7449839831837166e-05, "loss": 0.1871674656867981, "step": 9700 }, { "epoch": 0.25577620240827387, "grad_norm": 2.293290853500366, "learning_rate": 1.7444145993254105e-05, "loss": 0.1668079376220703, "step": 9710 }, { "epoch": 0.2560396176527726, "grad_norm": 1.107317566871643, "learning_rate": 1.7438446736500418e-05, "loss": 0.19088584184646606, "step": 9720 }, { "epoch": 0.25630303289727135, "grad_norm": 3.3194973468780518, "learning_rate": 1.7432742065724276e-05, "loss": 0.18241522312164307, "step": 9730 }, { "epoch": 0.2565664481417701, "grad_norm": 3.6050305366516113, "learning_rate": 1.7427031985077785e-05, "loss": 0.17914412021636963, "step": 9740 }, { "epoch": 0.25682986338626884, "grad_norm": 1.2917240858078003, "learning_rate": 1.7421316498716986e-05, "loss": 0.15559697151184082, "step": 9750 }, { "epoch": 0.2570932786307676, "grad_norm": 4.460684776306152, "learning_rate": 1.741559561080185e-05, "loss": 0.16406502723693847, "step": 9760 }, { "epoch": 0.2573566938752663, "grad_norm": 1.6017380952835083, "learning_rate": 1.74098693254963e-05, "loss": 0.14423234462738038, "step": 9770 }, { "epoch": 0.257620109119765, "grad_norm": 2.244955062866211, "learning_rate": 1.7404137646968162e-05, "loss": 0.17763330936431884, "step": 9780 }, { "epoch": 0.25788352436426376, "grad_norm": 2.3317954540252686, "learning_rate": 1.7398400579389204e-05, "loss": 0.16347768306732177, "step": 9790 }, { "epoch": 0.2581469396087625, "grad_norm": 3.0564699172973633, "learning_rate": 1.739265812693511e-05, "loss": 0.17482033967971802, "step": 9800 }, { "epoch": 0.25841035485326125, "grad_norm": 3.4123222827911377, "learning_rate": 1.7386910293785485e-05, "loss": 0.17319033145904542, "step": 9810 }, { "epoch": 0.25867377009776, "grad_norm": 2.7701714038848877, "learning_rate": 1.738115708412385e-05, "loss": 0.15891873836517334, "step": 9820 }, { "epoch": 0.25893718534225874, "grad_norm": 1.552183747291565, "learning_rate": 1.737539850213764e-05, "loss": 0.18356915712356567, "step": 9830 }, { "epoch": 0.2592006005867575, "grad_norm": 1.8902356624603271, "learning_rate": 1.7369634552018196e-05, "loss": 0.17109508514404298, "step": 9840 }, { "epoch": 0.25946401583125617, "grad_norm": 1.2652727365493774, "learning_rate": 1.7363865237960773e-05, "loss": 0.16792312860488892, "step": 9850 }, { "epoch": 0.2597274310757549, "grad_norm": 3.1060304641723633, "learning_rate": 1.7358090564164524e-05, "loss": 0.16460182666778564, "step": 9860 }, { "epoch": 0.25999084632025365, "grad_norm": 1.5393037796020508, "learning_rate": 1.735231053483251e-05, "loss": 0.17455856800079345, "step": 9870 }, { "epoch": 0.2602542615647524, "grad_norm": 1.075895071029663, "learning_rate": 1.734652515417168e-05, "loss": 0.18653212785720824, "step": 9880 }, { "epoch": 0.26051767680925114, "grad_norm": 1.6583274602890015, "learning_rate": 1.7340734426392894e-05, "loss": 0.1733178973197937, "step": 9890 }, { "epoch": 0.2607810920537499, "grad_norm": 2.2033722400665283, "learning_rate": 1.7334938355710876e-05, "loss": 0.1839218854904175, "step": 9900 }, { "epoch": 0.26104450729824863, "grad_norm": 1.1612942218780518, "learning_rate": 1.7329136946344272e-05, "loss": 0.17436248064041138, "step": 9910 }, { "epoch": 0.2613079225427474, "grad_norm": 1.1916697025299072, "learning_rate": 1.7323330202515592e-05, "loss": 0.1698922634124756, "step": 9920 }, { "epoch": 0.2615713377872461, "grad_norm": 3.754920244216919, "learning_rate": 1.731751812845124e-05, "loss": 0.1858532190322876, "step": 9930 }, { "epoch": 0.2618347530317448, "grad_norm": 1.309641718864441, "learning_rate": 1.7311700728381486e-05, "loss": 0.1649491310119629, "step": 9940 }, { "epoch": 0.26209816827624355, "grad_norm": 3.574662685394287, "learning_rate": 1.7305878006540483e-05, "loss": 0.16952252388000488, "step": 9950 }, { "epoch": 0.2623615835207423, "grad_norm": 1.4152582883834839, "learning_rate": 1.7300049967166273e-05, "loss": 0.17070726156234742, "step": 9960 }, { "epoch": 0.26262499876524104, "grad_norm": 0.9454043507575989, "learning_rate": 1.729421661450074e-05, "loss": 0.16272118091583251, "step": 9970 }, { "epoch": 0.2628884140097398, "grad_norm": 1.9373688697814941, "learning_rate": 1.728837795278966e-05, "loss": 0.171381938457489, "step": 9980 }, { "epoch": 0.2631518292542385, "grad_norm": 2.9220898151397705, "learning_rate": 1.7282533986282656e-05, "loss": 0.18054943084716796, "step": 9990 }, { "epoch": 0.26341524449873727, "grad_norm": 9.717860221862793, "learning_rate": 1.7276684719233222e-05, "loss": 0.18929489850997924, "step": 10000 }, { "epoch": 0.263678659743236, "grad_norm": 1.5384173393249512, "learning_rate": 1.7270830155898704e-05, "loss": 0.19718668460845948, "step": 10010 }, { "epoch": 0.26394207498773475, "grad_norm": 2.464986801147461, "learning_rate": 1.726497030054031e-05, "loss": 0.17723101377487183, "step": 10020 }, { "epoch": 0.26420549023223344, "grad_norm": 0.7872733473777771, "learning_rate": 1.72591051574231e-05, "loss": 0.17185455560684204, "step": 10030 }, { "epoch": 0.2644689054767322, "grad_norm": 2.509782314300537, "learning_rate": 1.7253234730815964e-05, "loss": 0.18956592082977294, "step": 10040 }, { "epoch": 0.26473232072123093, "grad_norm": 1.612607479095459, "learning_rate": 1.724735902499166e-05, "loss": 0.1820389986038208, "step": 10050 }, { "epoch": 0.2649957359657297, "grad_norm": 0.7318688035011292, "learning_rate": 1.7241478044226782e-05, "loss": 0.1744948148727417, "step": 10060 }, { "epoch": 0.2652591512102284, "grad_norm": 1.5291476249694824, "learning_rate": 1.723559179280176e-05, "loss": 0.1749386191368103, "step": 10070 }, { "epoch": 0.26552256645472716, "grad_norm": 1.771761178970337, "learning_rate": 1.722970027500086e-05, "loss": 0.19333361387252807, "step": 10080 }, { "epoch": 0.2657859816992259, "grad_norm": 3.5169498920440674, "learning_rate": 1.7223803495112185e-05, "loss": 0.17591650485992433, "step": 10090 }, { "epoch": 0.26604939694372465, "grad_norm": 1.1599398851394653, "learning_rate": 1.7217901457427667e-05, "loss": 0.16554044485092162, "step": 10100 }, { "epoch": 0.26631281218822334, "grad_norm": 2.3936407566070557, "learning_rate": 1.7211994166243058e-05, "loss": 0.17766551971435546, "step": 10110 }, { "epoch": 0.2665762274327221, "grad_norm": 1.8514487743377686, "learning_rate": 1.7206081625857943e-05, "loss": 0.18235820531845093, "step": 10120 }, { "epoch": 0.2668396426772208, "grad_norm": 2.4735031127929688, "learning_rate": 1.7200163840575728e-05, "loss": 0.17783199548721312, "step": 10130 }, { "epoch": 0.26710305792171957, "grad_norm": 3.255913496017456, "learning_rate": 1.7194240814703626e-05, "loss": 0.17770395278930665, "step": 10140 }, { "epoch": 0.2673664731662183, "grad_norm": 1.361997365951538, "learning_rate": 1.7188312552552672e-05, "loss": 0.18257691860198974, "step": 10150 }, { "epoch": 0.26762988841071705, "grad_norm": 2.7849819660186768, "learning_rate": 1.718237905843771e-05, "loss": 0.17537754774093628, "step": 10160 }, { "epoch": 0.2678933036552158, "grad_norm": 2.4645183086395264, "learning_rate": 1.71764403366774e-05, "loss": 0.17012317180633546, "step": 10170 }, { "epoch": 0.26815671889971454, "grad_norm": 2.246222972869873, "learning_rate": 1.7170496391594194e-05, "loss": 0.16769664287567138, "step": 10180 }, { "epoch": 0.2684201341442133, "grad_norm": 2.093928575515747, "learning_rate": 1.716454722751435e-05, "loss": 0.17151843309402465, "step": 10190 }, { "epoch": 0.268683549388712, "grad_norm": 1.814561367034912, "learning_rate": 1.715859284876793e-05, "loss": 0.1766250252723694, "step": 10200 }, { "epoch": 0.2689469646332107, "grad_norm": 0.9372042417526245, "learning_rate": 1.7152633259688788e-05, "loss": 0.16991734504699707, "step": 10210 }, { "epoch": 0.26921037987770946, "grad_norm": 5.108697414398193, "learning_rate": 1.714666846461457e-05, "loss": 0.16487884521484375, "step": 10220 }, { "epoch": 0.2694737951222082, "grad_norm": 1.5595147609710693, "learning_rate": 1.7140698467886707e-05, "loss": 0.16595228910446166, "step": 10230 }, { "epoch": 0.26973721036670695, "grad_norm": 1.0922976732254028, "learning_rate": 1.713472327385042e-05, "loss": 0.1780110478401184, "step": 10240 }, { "epoch": 0.2700006256112057, "grad_norm": 3.0800013542175293, "learning_rate": 1.7128742886854723e-05, "loss": 0.16550796031951903, "step": 10250 }, { "epoch": 0.27026404085570443, "grad_norm": 1.8595861196517944, "learning_rate": 1.712275731125239e-05, "loss": 0.17650455236434937, "step": 10260 }, { "epoch": 0.2705274561002032, "grad_norm": 6.162204742431641, "learning_rate": 1.7116766551399986e-05, "loss": 0.1669856548309326, "step": 10270 }, { "epoch": 0.2707908713447019, "grad_norm": 1.3557591438293457, "learning_rate": 1.711077061165784e-05, "loss": 0.17477118968963623, "step": 10280 }, { "epoch": 0.2710542865892006, "grad_norm": 3.5166690349578857, "learning_rate": 1.710476949639006e-05, "loss": 0.1841128945350647, "step": 10290 }, { "epoch": 0.27131770183369935, "grad_norm": 2.6496589183807373, "learning_rate": 1.7098763209964514e-05, "loss": 0.1704707622528076, "step": 10300 }, { "epoch": 0.2715811170781981, "grad_norm": 1.220288634300232, "learning_rate": 1.7092751756752838e-05, "loss": 0.15985463857650756, "step": 10310 }, { "epoch": 0.27184453232269684, "grad_norm": 2.6602210998535156, "learning_rate": 1.7086735141130423e-05, "loss": 0.17695775032043456, "step": 10320 }, { "epoch": 0.2721079475671956, "grad_norm": 3.658834934234619, "learning_rate": 1.7080713367476428e-05, "loss": 0.1691422700881958, "step": 10330 }, { "epoch": 0.27237136281169433, "grad_norm": 1.1311091184616089, "learning_rate": 1.7074686440173752e-05, "loss": 0.1808295726776123, "step": 10340 }, { "epoch": 0.2726347780561931, "grad_norm": 1.882101058959961, "learning_rate": 1.7068654363609058e-05, "loss": 0.1626690149307251, "step": 10350 }, { "epoch": 0.2728981933006918, "grad_norm": 3.4195749759674072, "learning_rate": 1.7062617142172754e-05, "loss": 0.19066376686096193, "step": 10360 }, { "epoch": 0.2731616085451905, "grad_norm": 1.6256362199783325, "learning_rate": 1.7056574780258986e-05, "loss": 0.1847226142883301, "step": 10370 }, { "epoch": 0.27342502378968925, "grad_norm": 2.581170082092285, "learning_rate": 1.705052728226564e-05, "loss": 0.16653411388397216, "step": 10380 }, { "epoch": 0.273688439034188, "grad_norm": 1.5861421823501587, "learning_rate": 1.704447465259436e-05, "loss": 0.1698061227798462, "step": 10390 }, { "epoch": 0.27395185427868674, "grad_norm": 3.0513288974761963, "learning_rate": 1.7038416895650504e-05, "loss": 0.18118221759796144, "step": 10400 }, { "epoch": 0.2742152695231855, "grad_norm": 2.476778030395508, "learning_rate": 1.7032354015843162e-05, "loss": 0.17754459381103516, "step": 10410 }, { "epoch": 0.2744786847676842, "grad_norm": 4.69702672958374, "learning_rate": 1.7026286017585168e-05, "loss": 0.1679738163948059, "step": 10420 }, { "epoch": 0.27474210001218297, "grad_norm": 2.3099889755249023, "learning_rate": 1.7020212905293072e-05, "loss": 0.17655518054962158, "step": 10430 }, { "epoch": 0.2750055152566817, "grad_norm": 1.6877617835998535, "learning_rate": 1.7014134683387147e-05, "loss": 0.176712965965271, "step": 10440 }, { "epoch": 0.27526893050118045, "grad_norm": 1.3665052652359009, "learning_rate": 1.7008051356291382e-05, "loss": 0.1628609776496887, "step": 10450 }, { "epoch": 0.27553234574567914, "grad_norm": 1.232194185256958, "learning_rate": 1.7001962928433488e-05, "loss": 0.1763024926185608, "step": 10460 }, { "epoch": 0.2757957609901779, "grad_norm": 2.730449676513672, "learning_rate": 1.699586940424488e-05, "loss": 0.1866118788719177, "step": 10470 }, { "epoch": 0.27605917623467663, "grad_norm": 1.4870802164077759, "learning_rate": 1.6989770788160697e-05, "loss": 0.18221583366394042, "step": 10480 }, { "epoch": 0.2763225914791754, "grad_norm": 4.48907470703125, "learning_rate": 1.6983667084619768e-05, "loss": 0.18123202323913573, "step": 10490 }, { "epoch": 0.2765860067236741, "grad_norm": 3.0900728702545166, "learning_rate": 1.6977558298064636e-05, "loss": 0.1781907320022583, "step": 10500 }, { "epoch": 0.27684942196817286, "grad_norm": 2.4332034587860107, "learning_rate": 1.6971444432941533e-05, "loss": 0.17161192893981933, "step": 10510 }, { "epoch": 0.2771128372126716, "grad_norm": 1.2439360618591309, "learning_rate": 1.69653254937004e-05, "loss": 0.17539350986480712, "step": 10520 }, { "epoch": 0.27737625245717035, "grad_norm": 1.7634233236312866, "learning_rate": 1.6959201484794867e-05, "loss": 0.1695181369781494, "step": 10530 }, { "epoch": 0.2776396677016691, "grad_norm": 1.4058648347854614, "learning_rate": 1.6953072410682243e-05, "loss": 0.18444141149520873, "step": 10540 }, { "epoch": 0.2779030829461678, "grad_norm": 3.543766736984253, "learning_rate": 1.694693827582354e-05, "loss": 0.17635841369628907, "step": 10550 }, { "epoch": 0.2781664981906665, "grad_norm": 1.996254563331604, "learning_rate": 1.694079908468344e-05, "loss": 0.16561262607574462, "step": 10560 }, { "epoch": 0.27842991343516527, "grad_norm": 2.37406325340271, "learning_rate": 1.6934654841730318e-05, "loss": 0.1784225344657898, "step": 10570 }, { "epoch": 0.278693328679664, "grad_norm": 1.9780415296554565, "learning_rate": 1.6928505551436215e-05, "loss": 0.17998249530792237, "step": 10580 }, { "epoch": 0.27895674392416275, "grad_norm": 3.12923002243042, "learning_rate": 1.6922351218276852e-05, "loss": 0.17404086589813234, "step": 10590 }, { "epoch": 0.2792201591686615, "grad_norm": 3.779499053955078, "learning_rate": 1.6916191846731615e-05, "loss": 0.16199703216552735, "step": 10600 }, { "epoch": 0.27948357441316024, "grad_norm": 4.0363078117370605, "learning_rate": 1.6910027441283562e-05, "loss": 0.17838985919952394, "step": 10610 }, { "epoch": 0.279746989657659, "grad_norm": 4.018301963806152, "learning_rate": 1.6903858006419413e-05, "loss": 0.19306461811065673, "step": 10620 }, { "epoch": 0.2800104049021577, "grad_norm": 1.7824957370758057, "learning_rate": 1.689768354662955e-05, "loss": 0.17776875495910643, "step": 10630 }, { "epoch": 0.2802738201466564, "grad_norm": 1.600566029548645, "learning_rate": 1.689150406640801e-05, "loss": 0.1730882167816162, "step": 10640 }, { "epoch": 0.28053723539115516, "grad_norm": 2.1137607097625732, "learning_rate": 1.6885319570252485e-05, "loss": 0.17609922885894774, "step": 10650 }, { "epoch": 0.2808006506356539, "grad_norm": 4.184754848480225, "learning_rate": 1.687913006266432e-05, "loss": 0.1644190788269043, "step": 10660 }, { "epoch": 0.28106406588015265, "grad_norm": 3.318202257156372, "learning_rate": 1.6872935548148503e-05, "loss": 0.1777878999710083, "step": 10670 }, { "epoch": 0.2813274811246514, "grad_norm": 4.101240634918213, "learning_rate": 1.686673603121367e-05, "loss": 0.17307600975036622, "step": 10680 }, { "epoch": 0.28159089636915013, "grad_norm": 5.0249247550964355, "learning_rate": 1.6860531516372102e-05, "loss": 0.18147884607315062, "step": 10690 }, { "epoch": 0.2818543116136489, "grad_norm": 2.1442337036132812, "learning_rate": 1.6854322008139704e-05, "loss": 0.17224507331848143, "step": 10700 }, { "epoch": 0.2821177268581476, "grad_norm": 1.137394666671753, "learning_rate": 1.6848107511036026e-05, "loss": 0.17485735416412354, "step": 10710 }, { "epoch": 0.2823811421026463, "grad_norm": 1.9236615896224976, "learning_rate": 1.6841888029584255e-05, "loss": 0.15780954360961913, "step": 10720 }, { "epoch": 0.28264455734714505, "grad_norm": 1.9226033687591553, "learning_rate": 1.6835663568311187e-05, "loss": 0.15770504474639893, "step": 10730 }, { "epoch": 0.2829079725916438, "grad_norm": 1.7929500341415405, "learning_rate": 1.682943413174726e-05, "loss": 0.17647056579589843, "step": 10740 }, { "epoch": 0.28317138783614254, "grad_norm": 1.9663877487182617, "learning_rate": 1.6823199724426523e-05, "loss": 0.18954721689224244, "step": 10750 }, { "epoch": 0.2834348030806413, "grad_norm": 2.1117730140686035, "learning_rate": 1.6816960350886653e-05, "loss": 0.1845695734024048, "step": 10760 }, { "epoch": 0.28369821832514003, "grad_norm": 3.168827533721924, "learning_rate": 1.681071601566893e-05, "loss": 0.18546123504638673, "step": 10770 }, { "epoch": 0.28396163356963877, "grad_norm": 1.6908819675445557, "learning_rate": 1.6804466723318248e-05, "loss": 0.17650444507598878, "step": 10780 }, { "epoch": 0.2842250488141375, "grad_norm": 3.4576542377471924, "learning_rate": 1.6798212478383118e-05, "loss": 0.16746512651443482, "step": 10790 }, { "epoch": 0.28448846405863626, "grad_norm": 2.898672342300415, "learning_rate": 1.6791953285415645e-05, "loss": 0.17620499134063722, "step": 10800 }, { "epoch": 0.28475187930313495, "grad_norm": 3.078176736831665, "learning_rate": 1.678568914897154e-05, "loss": 0.17570071220397948, "step": 10810 }, { "epoch": 0.2850152945476337, "grad_norm": 4.998715400695801, "learning_rate": 1.6779420073610116e-05, "loss": 0.1747354745864868, "step": 10820 }, { "epoch": 0.28527870979213243, "grad_norm": 5.346747398376465, "learning_rate": 1.6773146063894273e-05, "loss": 0.1917446494102478, "step": 10830 }, { "epoch": 0.2855421250366312, "grad_norm": 6.050363540649414, "learning_rate": 1.6766867124390505e-05, "loss": 0.1690915822982788, "step": 10840 }, { "epoch": 0.2858055402811299, "grad_norm": 1.859377384185791, "learning_rate": 1.6760583259668897e-05, "loss": 0.17506531476974488, "step": 10850 }, { "epoch": 0.28606895552562867, "grad_norm": 1.248664379119873, "learning_rate": 1.6754294474303116e-05, "loss": 0.1759138822555542, "step": 10860 }, { "epoch": 0.2863323707701274, "grad_norm": 1.3658448457717896, "learning_rate": 1.674800077287041e-05, "loss": 0.1759246826171875, "step": 10870 }, { "epoch": 0.28659578601462615, "grad_norm": 3.249368906021118, "learning_rate": 1.6741702159951613e-05, "loss": 0.18084548711776732, "step": 10880 }, { "epoch": 0.2868592012591249, "grad_norm": 1.5144351720809937, "learning_rate": 1.6735398640131117e-05, "loss": 0.17838027477264404, "step": 10890 }, { "epoch": 0.2871226165036236, "grad_norm": 1.5715539455413818, "learning_rate": 1.6729090217996904e-05, "loss": 0.17389975786209105, "step": 10900 }, { "epoch": 0.28738603174812233, "grad_norm": 2.394202947616577, "learning_rate": 1.6722776898140513e-05, "loss": 0.17515554428100585, "step": 10910 }, { "epoch": 0.28764944699262107, "grad_norm": 5.118825435638428, "learning_rate": 1.6716458685157047e-05, "loss": 0.17137498855590821, "step": 10920 }, { "epoch": 0.2879128622371198, "grad_norm": 1.3595876693725586, "learning_rate": 1.671013558364518e-05, "loss": 0.16836678981781006, "step": 10930 }, { "epoch": 0.28817627748161856, "grad_norm": 6.556189060211182, "learning_rate": 1.6703807598207132e-05, "loss": 0.16831426620483397, "step": 10940 }, { "epoch": 0.2884396927261173, "grad_norm": 1.3966282606124878, "learning_rate": 1.669747473344869e-05, "loss": 0.15858408212661743, "step": 10950 }, { "epoch": 0.28870310797061605, "grad_norm": 3.5470070838928223, "learning_rate": 1.6691136993979183e-05, "loss": 0.1641353964805603, "step": 10960 }, { "epoch": 0.2889665232151148, "grad_norm": 1.7197526693344116, "learning_rate": 1.6684794384411494e-05, "loss": 0.16540119647979737, "step": 10970 }, { "epoch": 0.2892299384596135, "grad_norm": 1.31356680393219, "learning_rate": 1.667844690936204e-05, "loss": 0.1784948468208313, "step": 10980 }, { "epoch": 0.2894933537041122, "grad_norm": 2.526379346847534, "learning_rate": 1.6672094573450798e-05, "loss": 0.18894097805023194, "step": 10990 }, { "epoch": 0.28975676894861097, "grad_norm": 1.7413474321365356, "learning_rate": 1.6665737381301263e-05, "loss": 0.17869534492492675, "step": 11000 }, { "epoch": 0.2900201841931097, "grad_norm": 1.5312601327896118, "learning_rate": 1.665937533754048e-05, "loss": 0.1642817258834839, "step": 11010 }, { "epoch": 0.29028359943760845, "grad_norm": 2.0457119941711426, "learning_rate": 1.6653008446799014e-05, "loss": 0.17334325313568116, "step": 11020 }, { "epoch": 0.2905470146821072, "grad_norm": 1.5136834383010864, "learning_rate": 1.6646636713710967e-05, "loss": 0.1630751371383667, "step": 11030 }, { "epoch": 0.29081042992660594, "grad_norm": 2.7161951065063477, "learning_rate": 1.664026014291396e-05, "loss": 0.18110960721969604, "step": 11040 }, { "epoch": 0.2910738451711047, "grad_norm": 4.667737007141113, "learning_rate": 1.6633878739049135e-05, "loss": 0.17405521869659424, "step": 11050 }, { "epoch": 0.2913372604156034, "grad_norm": 2.309051513671875, "learning_rate": 1.6627492506761155e-05, "loss": 0.18503208160400392, "step": 11060 }, { "epoch": 0.2916006756601021, "grad_norm": 1.9998284578323364, "learning_rate": 1.6621101450698193e-05, "loss": 0.18946530818939208, "step": 11070 }, { "epoch": 0.29186409090460086, "grad_norm": 2.523282766342163, "learning_rate": 1.6614705575511937e-05, "loss": 0.1812371253967285, "step": 11080 }, { "epoch": 0.2921275061490996, "grad_norm": 3.507861375808716, "learning_rate": 1.660830488585758e-05, "loss": 0.16592006683349608, "step": 11090 }, { "epoch": 0.29239092139359835, "grad_norm": 2.5271530151367188, "learning_rate": 1.660189938639382e-05, "loss": 0.1676603674888611, "step": 11100 }, { "epoch": 0.2926543366380971, "grad_norm": 1.5362398624420166, "learning_rate": 1.6595489081782858e-05, "loss": 0.1852595567703247, "step": 11110 }, { "epoch": 0.29291775188259583, "grad_norm": 1.634544849395752, "learning_rate": 1.6589073976690387e-05, "loss": 0.18500828742980957, "step": 11120 }, { "epoch": 0.2931811671270946, "grad_norm": 2.549102306365967, "learning_rate": 1.65826540757856e-05, "loss": 0.17258310317993164, "step": 11130 }, { "epoch": 0.2934445823715933, "grad_norm": 1.719428300857544, "learning_rate": 1.6576229383741177e-05, "loss": 0.17487707138061523, "step": 11140 }, { "epoch": 0.29370799761609206, "grad_norm": 2.1078052520751953, "learning_rate": 1.6569799905233285e-05, "loss": 0.17328145503997802, "step": 11150 }, { "epoch": 0.29397141286059075, "grad_norm": 2.2829933166503906, "learning_rate": 1.6563365644941578e-05, "loss": 0.16627341508865356, "step": 11160 }, { "epoch": 0.2942348281050895, "grad_norm": 3.0400118827819824, "learning_rate": 1.6556926607549186e-05, "loss": 0.17260546684265138, "step": 11170 }, { "epoch": 0.29449824334958824, "grad_norm": 2.2521181106567383, "learning_rate": 1.655048279774272e-05, "loss": 0.17748785018920898, "step": 11180 }, { "epoch": 0.294761658594087, "grad_norm": 2.8392155170440674, "learning_rate": 1.6544034220212262e-05, "loss": 0.17350627183914186, "step": 11190 }, { "epoch": 0.2950250738385857, "grad_norm": 1.39667809009552, "learning_rate": 1.6537580879651364e-05, "loss": 0.17127883434295654, "step": 11200 }, { "epoch": 0.29528848908308447, "grad_norm": 2.134629726409912, "learning_rate": 1.6531122780757046e-05, "loss": 0.17934136390686034, "step": 11210 }, { "epoch": 0.2955519043275832, "grad_norm": 3.9326601028442383, "learning_rate": 1.6524659928229792e-05, "loss": 0.18146102428436278, "step": 11220 }, { "epoch": 0.29581531957208196, "grad_norm": 4.783472537994385, "learning_rate": 1.6518192326773546e-05, "loss": 0.18176225423812867, "step": 11230 }, { "epoch": 0.29607873481658065, "grad_norm": 2.8725156784057617, "learning_rate": 1.6511719981095703e-05, "loss": 0.1869802951812744, "step": 11240 }, { "epoch": 0.2963421500610794, "grad_norm": 2.7164955139160156, "learning_rate": 1.6505242895907122e-05, "loss": 0.17155781984329224, "step": 11250 }, { "epoch": 0.29660556530557813, "grad_norm": 3.1466891765594482, "learning_rate": 1.64987610759221e-05, "loss": 0.17330386638641357, "step": 11260 }, { "epoch": 0.2968689805500769, "grad_norm": 2.0636157989501953, "learning_rate": 1.6492274525858386e-05, "loss": 0.1736930251121521, "step": 11270 }, { "epoch": 0.2971323957945756, "grad_norm": 2.61120867729187, "learning_rate": 1.648578325043717e-05, "loss": 0.17057446241378785, "step": 11280 }, { "epoch": 0.29739581103907436, "grad_norm": 0.7067874073982239, "learning_rate": 1.6479287254383084e-05, "loss": 0.17977781295776368, "step": 11290 }, { "epoch": 0.2976592262835731, "grad_norm": 1.7653344869613647, "learning_rate": 1.6472786542424196e-05, "loss": 0.1609557867050171, "step": 11300 }, { "epoch": 0.29792264152807185, "grad_norm": 2.729358196258545, "learning_rate": 1.6466281119292004e-05, "loss": 0.180289888381958, "step": 11310 }, { "epoch": 0.2981860567725706, "grad_norm": 2.2743265628814697, "learning_rate": 1.6459770989721433e-05, "loss": 0.18419808149337769, "step": 11320 }, { "epoch": 0.2984494720170693, "grad_norm": 3.1945316791534424, "learning_rate": 1.6453256158450835e-05, "loss": 0.1709047794342041, "step": 11330 }, { "epoch": 0.298712887261568, "grad_norm": 1.0187842845916748, "learning_rate": 1.644673663022199e-05, "loss": 0.17151231765747071, "step": 11340 }, { "epoch": 0.29897630250606677, "grad_norm": 2.534513235092163, "learning_rate": 1.6440212409780088e-05, "loss": 0.18290818929672242, "step": 11350 }, { "epoch": 0.2992397177505655, "grad_norm": 1.0479308366775513, "learning_rate": 1.643368350187374e-05, "loss": 0.1873087167739868, "step": 11360 }, { "epoch": 0.29950313299506426, "grad_norm": 2.7092275619506836, "learning_rate": 1.6427149911254973e-05, "loss": 0.1686504602432251, "step": 11370 }, { "epoch": 0.299766548239563, "grad_norm": 1.5787794589996338, "learning_rate": 1.64206116426792e-05, "loss": 0.16175355911254882, "step": 11380 }, { "epoch": 0.30002996348406175, "grad_norm": 1.7015562057495117, "learning_rate": 1.641406870090527e-05, "loss": 0.1621190309524536, "step": 11390 }, { "epoch": 0.3002933787285605, "grad_norm": 1.9492084980010986, "learning_rate": 1.640752109069541e-05, "loss": 0.17868666648864745, "step": 11400 }, { "epoch": 0.30055679397305923, "grad_norm": 1.0563684701919556, "learning_rate": 1.640096881681526e-05, "loss": 0.17345391511917113, "step": 11410 }, { "epoch": 0.3008202092175579, "grad_norm": 2.6932806968688965, "learning_rate": 1.6394411884033844e-05, "loss": 0.17659057378768922, "step": 11420 }, { "epoch": 0.30108362446205666, "grad_norm": 1.499502182006836, "learning_rate": 1.638785029712358e-05, "loss": 0.16850576400756836, "step": 11430 }, { "epoch": 0.3013470397065554, "grad_norm": 2.734290361404419, "learning_rate": 1.638128406086027e-05, "loss": 0.18114769458770752, "step": 11440 }, { "epoch": 0.30161045495105415, "grad_norm": 1.6366091966629028, "learning_rate": 1.6374713180023114e-05, "loss": 0.1644802212715149, "step": 11450 }, { "epoch": 0.3018738701955529, "grad_norm": 1.6777491569519043, "learning_rate": 1.6368137659394673e-05, "loss": 0.17502100467681886, "step": 11460 }, { "epoch": 0.30213728544005164, "grad_norm": 3.1021177768707275, "learning_rate": 1.63615575037609e-05, "loss": 0.19404821395874022, "step": 11470 }, { "epoch": 0.3024007006845504, "grad_norm": 4.799623012542725, "learning_rate": 1.6354972717911115e-05, "loss": 0.1543795108795166, "step": 11480 }, { "epoch": 0.3026641159290491, "grad_norm": 3.4491076469421387, "learning_rate": 1.634838330663801e-05, "loss": 0.16991103887557985, "step": 11490 }, { "epoch": 0.3029275311735478, "grad_norm": 2.5361545085906982, "learning_rate": 1.6341789274737644e-05, "loss": 0.17823277711868285, "step": 11500 }, { "epoch": 0.30319094641804656, "grad_norm": 19.424007415771484, "learning_rate": 1.6335190627009432e-05, "loss": 1.588632583618164, "step": 11510 }, { "epoch": 0.3034543616625453, "grad_norm": 5.581923961639404, "learning_rate": 1.632858736825616e-05, "loss": 0.4040719985961914, "step": 11520 }, { "epoch": 0.30371777690704405, "grad_norm": 4.552865505218506, "learning_rate": 1.632197950328396e-05, "loss": 0.21246714591979982, "step": 11530 }, { "epoch": 0.3039811921515428, "grad_norm": 1.654420256614685, "learning_rate": 1.631536703690233e-05, "loss": 0.17698569297790528, "step": 11540 }, { "epoch": 0.30424460739604153, "grad_norm": 38.0149040222168, "learning_rate": 1.6308749973924096e-05, "loss": 2.4950809478759766, "step": 11550 }, { "epoch": 0.3045080226405403, "grad_norm": 3.024904727935791, "learning_rate": 1.6302128319165454e-05, "loss": 0.5629608631134033, "step": 11560 }, { "epoch": 0.304771437885039, "grad_norm": 3.2599520683288574, "learning_rate": 1.629550207744592e-05, "loss": 0.2655982494354248, "step": 11570 }, { "epoch": 0.30503485312953776, "grad_norm": 2.3767552375793457, "learning_rate": 1.6288871253588366e-05, "loss": 0.22699313163757323, "step": 11580 }, { "epoch": 0.30529826837403645, "grad_norm": 3.0021185874938965, "learning_rate": 1.6282235852418983e-05, "loss": 0.2162613868713379, "step": 11590 }, { "epoch": 0.3055616836185352, "grad_norm": 1.279842734336853, "learning_rate": 1.6275595878767315e-05, "loss": 0.20982801914215088, "step": 11600 }, { "epoch": 0.30582509886303394, "grad_norm": 1.996922492980957, "learning_rate": 1.6268951337466212e-05, "loss": 0.2110342025756836, "step": 11610 }, { "epoch": 0.3060885141075327, "grad_norm": 1.314691185951233, "learning_rate": 1.626230223335186e-05, "loss": 0.2222665786743164, "step": 11620 }, { "epoch": 0.3063519293520314, "grad_norm": 2.169583320617676, "learning_rate": 1.6255648571263762e-05, "loss": 0.20734102725982667, "step": 11630 }, { "epoch": 0.30661534459653017, "grad_norm": 1.2429577112197876, "learning_rate": 1.6248990356044742e-05, "loss": 0.2059171199798584, "step": 11640 }, { "epoch": 0.3068787598410289, "grad_norm": 1.9830187559127808, "learning_rate": 1.6242327592540942e-05, "loss": 0.17688355445861817, "step": 11650 }, { "epoch": 0.30714217508552766, "grad_norm": 1.7741931676864624, "learning_rate": 1.62356602856018e-05, "loss": 0.19700005054473876, "step": 11660 }, { "epoch": 0.3074055903300264, "grad_norm": 4.759335994720459, "learning_rate": 1.6228988440080068e-05, "loss": 0.18085386753082275, "step": 11670 }, { "epoch": 0.3076690055745251, "grad_norm": 2.7363758087158203, "learning_rate": 1.6222312060831814e-05, "loss": 0.18508033752441405, "step": 11680 }, { "epoch": 0.30793242081902383, "grad_norm": 3.833785057067871, "learning_rate": 1.6215631152716387e-05, "loss": 0.19292072057724, "step": 11690 }, { "epoch": 0.3081958360635226, "grad_norm": 2.289790391921997, "learning_rate": 1.620894572059644e-05, "loss": 0.18963735103607177, "step": 11700 }, { "epoch": 0.3084592513080213, "grad_norm": 3.220163106918335, "learning_rate": 1.6202255769337924e-05, "loss": 0.1745438575744629, "step": 11710 }, { "epoch": 0.30872266655252006, "grad_norm": 1.474513292312622, "learning_rate": 1.619556130381007e-05, "loss": 0.1906408190727234, "step": 11720 }, { "epoch": 0.3089860817970188, "grad_norm": 2.591881513595581, "learning_rate": 1.61888623288854e-05, "loss": 0.20130350589752197, "step": 11730 }, { "epoch": 0.30924949704151755, "grad_norm": 2.2615163326263428, "learning_rate": 1.6182158849439718e-05, "loss": 0.1752530336380005, "step": 11740 }, { "epoch": 0.3095129122860163, "grad_norm": 2.1300837993621826, "learning_rate": 1.61754508703521e-05, "loss": 0.18807250261306763, "step": 11750 }, { "epoch": 0.309776327530515, "grad_norm": 6.038087844848633, "learning_rate": 1.6168738396504912e-05, "loss": 0.20772407054901124, "step": 11760 }, { "epoch": 0.3100397427750137, "grad_norm": 3.579810380935669, "learning_rate": 1.6162021432783773e-05, "loss": 0.19266215562820435, "step": 11770 }, { "epoch": 0.31030315801951247, "grad_norm": 3.818767547607422, "learning_rate": 1.6155299984077584e-05, "loss": 0.18223192691802978, "step": 11780 }, { "epoch": 0.3105665732640112, "grad_norm": 1.6919313669204712, "learning_rate": 1.6148574055278507e-05, "loss": 0.1635216236114502, "step": 11790 }, { "epoch": 0.31082998850850996, "grad_norm": 0.9894604682922363, "learning_rate": 1.6141843651281955e-05, "loss": 0.18315293788909912, "step": 11800 }, { "epoch": 0.3110934037530087, "grad_norm": 2.9781594276428223, "learning_rate": 1.6135108776986615e-05, "loss": 0.19391193389892578, "step": 11810 }, { "epoch": 0.31135681899750745, "grad_norm": 1.1383711099624634, "learning_rate": 1.612836943729442e-05, "loss": 0.1664895534515381, "step": 11820 }, { "epoch": 0.3116202342420062, "grad_norm": 0.8793426752090454, "learning_rate": 1.6121625637110546e-05, "loss": 0.19523717164993287, "step": 11830 }, { "epoch": 0.31188364948650493, "grad_norm": 4.123857498168945, "learning_rate": 1.6114877381343422e-05, "loss": 0.18655582666397094, "step": 11840 }, { "epoch": 0.3121470647310036, "grad_norm": 2.4716949462890625, "learning_rate": 1.610812467490472e-05, "loss": 0.20267767906188966, "step": 11850 }, { "epoch": 0.31241047997550236, "grad_norm": 3.5565953254699707, "learning_rate": 1.6101367522709356e-05, "loss": 0.1824507713317871, "step": 11860 }, { "epoch": 0.3126738952200011, "grad_norm": 2.3587398529052734, "learning_rate": 1.6094605929675476e-05, "loss": 0.1765931725502014, "step": 11870 }, { "epoch": 0.31293731046449985, "grad_norm": 2.2139158248901367, "learning_rate": 1.608783990072445e-05, "loss": 0.18725301027297975, "step": 11880 }, { "epoch": 0.3132007257089986, "grad_norm": 4.827059268951416, "learning_rate": 1.6081069440780895e-05, "loss": 0.181646990776062, "step": 11890 }, { "epoch": 0.31346414095349734, "grad_norm": 2.754523515701294, "learning_rate": 1.607429455477264e-05, "loss": 0.1683582305908203, "step": 11900 }, { "epoch": 0.3137275561979961, "grad_norm": 1.7297654151916504, "learning_rate": 1.606751524763075e-05, "loss": 0.17711174488067627, "step": 11910 }, { "epoch": 0.3139909714424948, "grad_norm": 2.335810661315918, "learning_rate": 1.606073152428948e-05, "loss": 0.1882438063621521, "step": 11920 }, { "epoch": 0.31425438668699357, "grad_norm": 1.6738989353179932, "learning_rate": 1.6053943389686323e-05, "loss": 0.18364436626434327, "step": 11930 }, { "epoch": 0.31451780193149226, "grad_norm": 1.8335570096969604, "learning_rate": 1.604715084876198e-05, "loss": 0.18092772960662842, "step": 11940 }, { "epoch": 0.314781217175991, "grad_norm": 1.5093140602111816, "learning_rate": 1.6040353906460353e-05, "loss": 0.19241368770599365, "step": 11950 }, { "epoch": 0.31504463242048975, "grad_norm": 3.776172161102295, "learning_rate": 1.6033552567728553e-05, "loss": 0.17035586833953859, "step": 11960 }, { "epoch": 0.3153080476649885, "grad_norm": 2.40541410446167, "learning_rate": 1.602674683751688e-05, "loss": 0.19812469482421874, "step": 11970 }, { "epoch": 0.31557146290948723, "grad_norm": 2.773387908935547, "learning_rate": 1.6019936720778845e-05, "loss": 0.18730226755142212, "step": 11980 }, { "epoch": 0.315834878153986, "grad_norm": 2.861768960952759, "learning_rate": 1.601312222247114e-05, "loss": 0.17813037633895873, "step": 11990 }, { "epoch": 0.3160982933984847, "grad_norm": 2.2202067375183105, "learning_rate": 1.6006303347553655e-05, "loss": 0.19649685621261598, "step": 12000 }, { "epoch": 0.31636170864298346, "grad_norm": 2.477783203125, "learning_rate": 1.5999480100989458e-05, "loss": 0.19622480869293213, "step": 12010 }, { "epoch": 0.31662512388748215, "grad_norm": 2.862853527069092, "learning_rate": 1.5992652487744806e-05, "loss": 0.20149569511413573, "step": 12020 }, { "epoch": 0.3168885391319809, "grad_norm": 2.1359333992004395, "learning_rate": 1.5985820512789128e-05, "loss": 0.16951849460601806, "step": 12030 }, { "epoch": 0.31715195437647964, "grad_norm": 2.7514021396636963, "learning_rate": 1.5978984181095034e-05, "loss": 0.1935945749282837, "step": 12040 }, { "epoch": 0.3174153696209784, "grad_norm": 2.232344627380371, "learning_rate": 1.5972143497638294e-05, "loss": 0.189836323261261, "step": 12050 }, { "epoch": 0.3176787848654771, "grad_norm": 2.0591611862182617, "learning_rate": 1.5965298467397855e-05, "loss": 0.18150885105133058, "step": 12060 }, { "epoch": 0.31794220010997587, "grad_norm": 1.5350804328918457, "learning_rate": 1.595844909535583e-05, "loss": 0.1805711269378662, "step": 12070 }, { "epoch": 0.3182056153544746, "grad_norm": 2.37802791595459, "learning_rate": 1.595159538649748e-05, "loss": 0.1856452226638794, "step": 12080 }, { "epoch": 0.31846903059897336, "grad_norm": 4.663433074951172, "learning_rate": 1.5944737345811238e-05, "loss": 0.18978564739227294, "step": 12090 }, { "epoch": 0.3187324458434721, "grad_norm": 0.9344496130943298, "learning_rate": 1.5937874978288676e-05, "loss": 0.17246285676956177, "step": 12100 }, { "epoch": 0.3189958610879708, "grad_norm": 2.0968947410583496, "learning_rate": 1.5931008288924523e-05, "loss": 0.18097398281097413, "step": 12110 }, { "epoch": 0.31925927633246953, "grad_norm": 1.578900933265686, "learning_rate": 1.5924137282716653e-05, "loss": 0.17832480669021605, "step": 12120 }, { "epoch": 0.3195226915769683, "grad_norm": 1.9329750537872314, "learning_rate": 1.591726196466608e-05, "loss": 0.20239567756652832, "step": 12130 }, { "epoch": 0.319786106821467, "grad_norm": 1.2938774824142456, "learning_rate": 1.5910382339776953e-05, "loss": 0.18155834674835206, "step": 12140 }, { "epoch": 0.32004952206596576, "grad_norm": 3.1146538257598877, "learning_rate": 1.5903498413056568e-05, "loss": 0.1850644826889038, "step": 12150 }, { "epoch": 0.3203129373104645, "grad_norm": 1.6465729475021362, "learning_rate": 1.5896610189515335e-05, "loss": 0.17645007371902466, "step": 12160 }, { "epoch": 0.32057635255496325, "grad_norm": 2.397923469543457, "learning_rate": 1.588971767416681e-05, "loss": 0.18865768909454345, "step": 12170 }, { "epoch": 0.320839767799462, "grad_norm": 5.593918323516846, "learning_rate": 1.5882820872027654e-05, "loss": 0.1972930908203125, "step": 12180 }, { "epoch": 0.32110318304396074, "grad_norm": 1.6560051441192627, "learning_rate": 1.5875919788117662e-05, "loss": 0.1893925666809082, "step": 12190 }, { "epoch": 0.3213665982884594, "grad_norm": 1.3297841548919678, "learning_rate": 1.586901442745974e-05, "loss": 0.19036054611206055, "step": 12200 }, { "epoch": 0.32163001353295817, "grad_norm": 2.209409475326538, "learning_rate": 1.5862104795079903e-05, "loss": 0.17221620082855224, "step": 12210 }, { "epoch": 0.3218934287774569, "grad_norm": 3.6661036014556885, "learning_rate": 1.5855190896007287e-05, "loss": 0.20152873992919923, "step": 12220 }, { "epoch": 0.32215684402195566, "grad_norm": 3.0797269344329834, "learning_rate": 1.584827273527412e-05, "loss": 0.1576531410217285, "step": 12230 }, { "epoch": 0.3224202592664544, "grad_norm": 3.393460512161255, "learning_rate": 1.5841350317915745e-05, "loss": 0.17906289100646972, "step": 12240 }, { "epoch": 0.32268367451095314, "grad_norm": 0.9368900060653687, "learning_rate": 1.5834423648970588e-05, "loss": 0.19591717720031737, "step": 12250 }, { "epoch": 0.3229470897554519, "grad_norm": 2.9191524982452393, "learning_rate": 1.5827492733480183e-05, "loss": 0.1861851692199707, "step": 12260 }, { "epoch": 0.32321050499995063, "grad_norm": 2.2250568866729736, "learning_rate": 1.5820557576489145e-05, "loss": 0.1629551410675049, "step": 12270 }, { "epoch": 0.3234739202444493, "grad_norm": 1.463942527770996, "learning_rate": 1.581361818304518e-05, "loss": 0.16981034278869628, "step": 12280 }, { "epoch": 0.32373733548894806, "grad_norm": 10.07772445678711, "learning_rate": 1.5806674558199088e-05, "loss": 0.1941838026046753, "step": 12290 }, { "epoch": 0.3240007507334468, "grad_norm": 1.2919303178787231, "learning_rate": 1.5799726707004728e-05, "loss": 0.1973050832748413, "step": 12300 }, { "epoch": 0.32426416597794555, "grad_norm": 1.6899330615997314, "learning_rate": 1.5792774634519046e-05, "loss": 0.19603347778320312, "step": 12310 }, { "epoch": 0.3245275812224443, "grad_norm": 2.7164363861083984, "learning_rate": 1.5785818345802068e-05, "loss": 0.18604322671890258, "step": 12320 }, { "epoch": 0.32479099646694304, "grad_norm": 1.8307275772094727, "learning_rate": 1.5778857845916874e-05, "loss": 0.17477145195007324, "step": 12330 }, { "epoch": 0.3250544117114418, "grad_norm": 1.5658226013183594, "learning_rate": 1.577189313992962e-05, "loss": 0.16727356910705565, "step": 12340 }, { "epoch": 0.3253178269559405, "grad_norm": 1.0500984191894531, "learning_rate": 1.5764924232909518e-05, "loss": 0.201423716545105, "step": 12350 }, { "epoch": 0.32558124220043927, "grad_norm": 1.6722815036773682, "learning_rate": 1.5757951129928837e-05, "loss": 0.17699074745178223, "step": 12360 }, { "epoch": 0.32584465744493796, "grad_norm": 0.7058488130569458, "learning_rate": 1.5750973836062907e-05, "loss": 0.17969331741333008, "step": 12370 }, { "epoch": 0.3261080726894367, "grad_norm": 4.606058597564697, "learning_rate": 1.57439923563901e-05, "loss": 0.20190672874450682, "step": 12380 }, { "epoch": 0.32637148793393544, "grad_norm": 1.3247215747833252, "learning_rate": 1.573700669599184e-05, "loss": 0.1727010726928711, "step": 12390 }, { "epoch": 0.3266349031784342, "grad_norm": 1.2113783359527588, "learning_rate": 1.573001685995259e-05, "loss": 0.16817522048950195, "step": 12400 }, { "epoch": 0.32689831842293293, "grad_norm": 3.4951059818267822, "learning_rate": 1.572302285335985e-05, "loss": 0.19238476753234862, "step": 12410 }, { "epoch": 0.3271617336674317, "grad_norm": 1.7229505777359009, "learning_rate": 1.5716024681304177e-05, "loss": 0.1827389717102051, "step": 12420 }, { "epoch": 0.3274251489119304, "grad_norm": 3.9429357051849365, "learning_rate": 1.5709022348879124e-05, "loss": 0.18471795320510864, "step": 12430 }, { "epoch": 0.32768856415642916, "grad_norm": 3.016324758529663, "learning_rate": 1.5702015861181295e-05, "loss": 0.1782430648803711, "step": 12440 }, { "epoch": 0.3279519794009279, "grad_norm": 3.4164350032806396, "learning_rate": 1.569500522331032e-05, "loss": 0.18741292953491212, "step": 12450 }, { "epoch": 0.3282153946454266, "grad_norm": 3.8984477519989014, "learning_rate": 1.568799044036884e-05, "loss": 0.18215131759643555, "step": 12460 }, { "epoch": 0.32847880988992534, "grad_norm": 1.2472952604293823, "learning_rate": 1.5680971517462516e-05, "loss": 0.1615942597389221, "step": 12470 }, { "epoch": 0.3287422251344241, "grad_norm": 3.348358154296875, "learning_rate": 1.5673948459700024e-05, "loss": 0.15835826396942138, "step": 12480 }, { "epoch": 0.3290056403789228, "grad_norm": 2.4278454780578613, "learning_rate": 1.5666921272193046e-05, "loss": 0.16986584663391113, "step": 12490 }, { "epoch": 0.32926905562342157, "grad_norm": 1.9199434518814087, "learning_rate": 1.565988996005628e-05, "loss": 0.17954367399215698, "step": 12500 }, { "epoch": 0.3295324708679203, "grad_norm": 2.0094168186187744, "learning_rate": 1.5652854528407407e-05, "loss": 0.17517966032028198, "step": 12510 }, { "epoch": 0.32979588611241906, "grad_norm": 2.1369080543518066, "learning_rate": 1.5645814982367125e-05, "loss": 0.173681640625, "step": 12520 }, { "epoch": 0.3300593013569178, "grad_norm": 0.7346780300140381, "learning_rate": 1.5638771327059117e-05, "loss": 0.18123526573181153, "step": 12530 }, { "epoch": 0.3303227166014165, "grad_norm": 2.9580986499786377, "learning_rate": 1.563172356761006e-05, "loss": 0.18826054334640502, "step": 12540 }, { "epoch": 0.33058613184591523, "grad_norm": 1.5193113088607788, "learning_rate": 1.5624671709149618e-05, "loss": 0.17209519147872926, "step": 12550 }, { "epoch": 0.330849547090414, "grad_norm": 1.6838910579681396, "learning_rate": 1.561761575681044e-05, "loss": 0.16919245719909667, "step": 12560 }, { "epoch": 0.3311129623349127, "grad_norm": 1.4526879787445068, "learning_rate": 1.5610555715728146e-05, "loss": 0.16993315219879152, "step": 12570 }, { "epoch": 0.33137637757941146, "grad_norm": 2.469179153442383, "learning_rate": 1.560349159104134e-05, "loss": 0.1783144235610962, "step": 12580 }, { "epoch": 0.3316397928239102, "grad_norm": 2.721414566040039, "learning_rate": 1.55964233878916e-05, "loss": 0.17876558303833007, "step": 12590 }, { "epoch": 0.33190320806840895, "grad_norm": 1.23512864112854, "learning_rate": 1.5589351111423474e-05, "loss": 0.17039891481399536, "step": 12600 }, { "epoch": 0.3321666233129077, "grad_norm": 2.2303647994995117, "learning_rate": 1.5582274766784457e-05, "loss": 0.19073809385299684, "step": 12610 }, { "epoch": 0.33243003855740644, "grad_norm": 1.460414171218872, "learning_rate": 1.557519435912503e-05, "loss": 0.16249940395355225, "step": 12620 }, { "epoch": 0.3326934538019051, "grad_norm": 2.0626707077026367, "learning_rate": 1.556810989359861e-05, "loss": 0.18231170177459716, "step": 12630 }, { "epoch": 0.33295686904640387, "grad_norm": 3.79715633392334, "learning_rate": 1.5561021375361583e-05, "loss": 0.19059034585952758, "step": 12640 }, { "epoch": 0.3332202842909026, "grad_norm": 1.645234227180481, "learning_rate": 1.5553928809573275e-05, "loss": 0.1833071231842041, "step": 12650 }, { "epoch": 0.33348369953540136, "grad_norm": 1.9209109544754028, "learning_rate": 1.5546832201395964e-05, "loss": 0.18272631168365477, "step": 12660 }, { "epoch": 0.3337471147799001, "grad_norm": 2.6839818954467773, "learning_rate": 1.5539731555994864e-05, "loss": 0.17818992137908934, "step": 12670 }, { "epoch": 0.33401053002439884, "grad_norm": 1.5445427894592285, "learning_rate": 1.5532626878538137e-05, "loss": 0.17567989826202393, "step": 12680 }, { "epoch": 0.3342739452688976, "grad_norm": 3.904625177383423, "learning_rate": 1.5525518174196867e-05, "loss": 0.17505133152008057, "step": 12690 }, { "epoch": 0.33453736051339633, "grad_norm": 2.9744460582733154, "learning_rate": 1.551840544814508e-05, "loss": 0.16773617267608643, "step": 12700 }, { "epoch": 0.3348007757578951, "grad_norm": 1.0284092426300049, "learning_rate": 1.551128870555972e-05, "loss": 0.1728747606277466, "step": 12710 }, { "epoch": 0.33506419100239376, "grad_norm": 3.226684331893921, "learning_rate": 1.550416795162067e-05, "loss": 0.18371503353118895, "step": 12720 }, { "epoch": 0.3353276062468925, "grad_norm": 2.291261672973633, "learning_rate": 1.549704319151071e-05, "loss": 0.1835680603981018, "step": 12730 }, { "epoch": 0.33559102149139125, "grad_norm": 1.0861209630966187, "learning_rate": 1.548991443041555e-05, "loss": 0.18247843980789186, "step": 12740 }, { "epoch": 0.33585443673589, "grad_norm": 2.739576578140259, "learning_rate": 1.5482781673523816e-05, "loss": 0.18610976934432982, "step": 12750 }, { "epoch": 0.33611785198038874, "grad_norm": 1.5827041864395142, "learning_rate": 1.5475644926027035e-05, "loss": 0.16284761428833008, "step": 12760 }, { "epoch": 0.3363812672248875, "grad_norm": 2.670408248901367, "learning_rate": 1.5468504193119636e-05, "loss": 0.19947829246520996, "step": 12770 }, { "epoch": 0.3366446824693862, "grad_norm": 2.911447525024414, "learning_rate": 1.5461359479998953e-05, "loss": 0.17293668985366822, "step": 12780 }, { "epoch": 0.33690809771388497, "grad_norm": 2.441829204559326, "learning_rate": 1.5454210791865213e-05, "loss": 0.18091795444488526, "step": 12790 }, { "epoch": 0.33717151295838366, "grad_norm": 2.0564424991607666, "learning_rate": 1.5447058133921552e-05, "loss": 0.19152361154556274, "step": 12800 }, { "epoch": 0.3374349282028824, "grad_norm": 2.422574281692505, "learning_rate": 1.543990151137397e-05, "loss": 0.17642321586608886, "step": 12810 }, { "epoch": 0.33769834344738114, "grad_norm": 1.3862190246582031, "learning_rate": 1.5432740929431372e-05, "loss": 0.16174442768096925, "step": 12820 }, { "epoch": 0.3379617586918799, "grad_norm": 3.040651559829712, "learning_rate": 1.5425576393305537e-05, "loss": 0.16939914226531982, "step": 12830 }, { "epoch": 0.33822517393637863, "grad_norm": 1.5488568544387817, "learning_rate": 1.5418407908211125e-05, "loss": 0.17889020442962647, "step": 12840 }, { "epoch": 0.3384885891808774, "grad_norm": 2.219973087310791, "learning_rate": 1.5411235479365664e-05, "loss": 0.16694798469543456, "step": 12850 }, { "epoch": 0.3387520044253761, "grad_norm": 1.2192537784576416, "learning_rate": 1.5404059111989565e-05, "loss": 0.1799445629119873, "step": 12860 }, { "epoch": 0.33901541966987486, "grad_norm": 2.5201892852783203, "learning_rate": 1.5396878811306088e-05, "loss": 0.17883164882659913, "step": 12870 }, { "epoch": 0.3392788349143736, "grad_norm": 1.1712652444839478, "learning_rate": 1.5389694582541376e-05, "loss": 0.17658421993255616, "step": 12880 }, { "epoch": 0.3395422501588723, "grad_norm": 0.7239189743995667, "learning_rate": 1.5382506430924414e-05, "loss": 0.18241262435913086, "step": 12890 }, { "epoch": 0.33980566540337104, "grad_norm": 2.241931200027466, "learning_rate": 1.5375314361687053e-05, "loss": 0.1758476734161377, "step": 12900 }, { "epoch": 0.3400690806478698, "grad_norm": 3.2468833923339844, "learning_rate": 1.5368118380063984e-05, "loss": 0.18414816856384278, "step": 12910 }, { "epoch": 0.3403324958923685, "grad_norm": 3.2150747776031494, "learning_rate": 1.536091849129276e-05, "loss": 0.19106291532516478, "step": 12920 }, { "epoch": 0.34059591113686727, "grad_norm": 1.270937204360962, "learning_rate": 1.5353714700613768e-05, "loss": 0.17604513168334962, "step": 12930 }, { "epoch": 0.340859326381366, "grad_norm": 4.120845317840576, "learning_rate": 1.5346507013270237e-05, "loss": 0.18284752368927001, "step": 12940 }, { "epoch": 0.34112274162586476, "grad_norm": 3.723259210586548, "learning_rate": 1.5339295434508238e-05, "loss": 0.172753643989563, "step": 12950 }, { "epoch": 0.3413861568703635, "grad_norm": 1.2255444526672363, "learning_rate": 1.533207996957666e-05, "loss": 0.17508050203323364, "step": 12960 }, { "epoch": 0.34164957211486224, "grad_norm": 1.6183662414550781, "learning_rate": 1.5324860623727236e-05, "loss": 0.18318684101104737, "step": 12970 }, { "epoch": 0.34191298735936093, "grad_norm": 2.363431215286255, "learning_rate": 1.5317637402214515e-05, "loss": 0.1689462423324585, "step": 12980 }, { "epoch": 0.3421764026038597, "grad_norm": 1.571887493133545, "learning_rate": 1.5310410310295876e-05, "loss": 0.16199755668640137, "step": 12990 }, { "epoch": 0.3424398178483584, "grad_norm": 2.078132152557373, "learning_rate": 1.5303179353231494e-05, "loss": 0.18172141313552856, "step": 13000 }, { "epoch": 0.34270323309285716, "grad_norm": 3.6557021141052246, "learning_rate": 1.529594453628439e-05, "loss": 0.18166438341140748, "step": 13010 }, { "epoch": 0.3429666483373559, "grad_norm": 2.0125973224639893, "learning_rate": 1.5288705864720356e-05, "loss": 0.17909111976623535, "step": 13020 }, { "epoch": 0.34323006358185465, "grad_norm": 2.802680015563965, "learning_rate": 1.528146334380803e-05, "loss": 0.20308115482330322, "step": 13030 }, { "epoch": 0.3434934788263534, "grad_norm": 2.271308183670044, "learning_rate": 1.5274216978818813e-05, "loss": 0.1734297513961792, "step": 13040 }, { "epoch": 0.34375689407085214, "grad_norm": 4.965381622314453, "learning_rate": 1.526696677502693e-05, "loss": 0.1756630063056946, "step": 13050 }, { "epoch": 0.3440203093153508, "grad_norm": 3.023407459259033, "learning_rate": 1.525971273770939e-05, "loss": 0.18334972858428955, "step": 13060 }, { "epoch": 0.34428372455984957, "grad_norm": 2.7510998249053955, "learning_rate": 1.5252454872146003e-05, "loss": 0.19309982061386108, "step": 13070 }, { "epoch": 0.3445471398043483, "grad_norm": 2.4645228385925293, "learning_rate": 1.5245193183619347e-05, "loss": 0.1829069137573242, "step": 13080 }, { "epoch": 0.34481055504884706, "grad_norm": 1.6244924068450928, "learning_rate": 1.5237927677414797e-05, "loss": 0.17620563507080078, "step": 13090 }, { "epoch": 0.3450739702933458, "grad_norm": 0.8489829897880554, "learning_rate": 1.5230658358820493e-05, "loss": 0.17435287237167357, "step": 13100 }, { "epoch": 0.34533738553784454, "grad_norm": 2.9000325202941895, "learning_rate": 1.5223385233127372e-05, "loss": 0.17601732015609742, "step": 13110 }, { "epoch": 0.3456008007823433, "grad_norm": 3.055295467376709, "learning_rate": 1.5216108305629123e-05, "loss": 0.16737909317016603, "step": 13120 }, { "epoch": 0.34586421602684203, "grad_norm": 2.9998528957366943, "learning_rate": 1.5208827581622205e-05, "loss": 0.19119586944580078, "step": 13130 }, { "epoch": 0.3461276312713408, "grad_norm": 1.7417030334472656, "learning_rate": 1.5201543066405844e-05, "loss": 0.16351394653320311, "step": 13140 }, { "epoch": 0.34639104651583946, "grad_norm": 31.347244262695312, "learning_rate": 1.5194254765282025e-05, "loss": 0.18680838346481324, "step": 13150 }, { "epoch": 0.3466544617603382, "grad_norm": 1.19352388381958, "learning_rate": 1.5186962683555493e-05, "loss": 0.16938804388046264, "step": 13160 }, { "epoch": 0.34691787700483695, "grad_norm": 1.3841090202331543, "learning_rate": 1.517966682653373e-05, "loss": 0.16159439086914062, "step": 13170 }, { "epoch": 0.3471812922493357, "grad_norm": 3.585858106613159, "learning_rate": 1.5172367199526985e-05, "loss": 0.17650552988052368, "step": 13180 }, { "epoch": 0.34744470749383444, "grad_norm": 3.4243595600128174, "learning_rate": 1.5165063807848234e-05, "loss": 0.1850489377975464, "step": 13190 }, { "epoch": 0.3477081227383332, "grad_norm": 3.833935022354126, "learning_rate": 1.5157756656813207e-05, "loss": 0.19421675205230712, "step": 13200 }, { "epoch": 0.3479715379828319, "grad_norm": 2.419516086578369, "learning_rate": 1.5150445751740356e-05, "loss": 0.16437315940856934, "step": 13210 }, { "epoch": 0.34823495322733067, "grad_norm": 1.4645287990570068, "learning_rate": 1.5143131097950882e-05, "loss": 0.16785577535629273, "step": 13220 }, { "epoch": 0.3484983684718294, "grad_norm": 2.755218505859375, "learning_rate": 1.5135812700768702e-05, "loss": 0.18851103782653808, "step": 13230 }, { "epoch": 0.3487617837163281, "grad_norm": 1.2550195455551147, "learning_rate": 1.512849056552046e-05, "loss": 0.1649218440055847, "step": 13240 }, { "epoch": 0.34902519896082684, "grad_norm": 2.4086575508117676, "learning_rate": 1.512116469753552e-05, "loss": 0.1753202795982361, "step": 13250 }, { "epoch": 0.3492886142053256, "grad_norm": 3.0663788318634033, "learning_rate": 1.5113835102145968e-05, "loss": 0.17951560020446777, "step": 13260 }, { "epoch": 0.34955202944982433, "grad_norm": 2.4858453273773193, "learning_rate": 1.51065017846866e-05, "loss": 0.17641062736511232, "step": 13270 }, { "epoch": 0.3498154446943231, "grad_norm": 1.439605951309204, "learning_rate": 1.509916475049492e-05, "loss": 0.1645904541015625, "step": 13280 }, { "epoch": 0.3500788599388218, "grad_norm": 4.357490539550781, "learning_rate": 1.5091824004911137e-05, "loss": 0.2060936450958252, "step": 13290 }, { "epoch": 0.35034227518332056, "grad_norm": 3.5110082626342773, "learning_rate": 1.5084479553278164e-05, "loss": 0.1763577938079834, "step": 13300 }, { "epoch": 0.3506056904278193, "grad_norm": 2.024142265319824, "learning_rate": 1.5077131400941609e-05, "loss": 0.17455803155899047, "step": 13310 }, { "epoch": 0.350869105672318, "grad_norm": 1.2730791568756104, "learning_rate": 1.5069779553249775e-05, "loss": 0.1670079708099365, "step": 13320 }, { "epoch": 0.35113252091681674, "grad_norm": 2.458183765411377, "learning_rate": 1.5062424015553653e-05, "loss": 0.2021024227142334, "step": 13330 }, { "epoch": 0.3513959361613155, "grad_norm": 1.9243935346603394, "learning_rate": 1.505506479320692e-05, "loss": 0.16383280754089355, "step": 13340 }, { "epoch": 0.3516593514058142, "grad_norm": 1.8801243305206299, "learning_rate": 1.5047701891565938e-05, "loss": 0.1848810911178589, "step": 13350 }, { "epoch": 0.35192276665031297, "grad_norm": 1.6597808599472046, "learning_rate": 1.504033531598974e-05, "loss": 0.17065610885620117, "step": 13360 }, { "epoch": 0.3521861818948117, "grad_norm": 2.3016481399536133, "learning_rate": 1.5032965071840045e-05, "loss": 0.1769503951072693, "step": 13370 }, { "epoch": 0.35244959713931046, "grad_norm": 0.9627740979194641, "learning_rate": 1.5025591164481226e-05, "loss": 0.19264508485794068, "step": 13380 }, { "epoch": 0.3527130123838092, "grad_norm": 3.1099209785461426, "learning_rate": 1.5018213599280339e-05, "loss": 0.17924758195877075, "step": 13390 }, { "epoch": 0.35297642762830794, "grad_norm": 2.190075635910034, "learning_rate": 1.5010832381607085e-05, "loss": 0.17309703826904296, "step": 13400 }, { "epoch": 0.35323984287280663, "grad_norm": 4.27790641784668, "learning_rate": 1.500344751683384e-05, "loss": 0.18132983446121215, "step": 13410 }, { "epoch": 0.3535032581173054, "grad_norm": 4.73012638092041, "learning_rate": 1.4996059010335626e-05, "loss": 0.1842617630958557, "step": 13420 }, { "epoch": 0.3537666733618041, "grad_norm": 1.492867112159729, "learning_rate": 1.4988666867490117e-05, "loss": 0.17364208698272704, "step": 13430 }, { "epoch": 0.35403008860630286, "grad_norm": 3.4389822483062744, "learning_rate": 1.4981271093677632e-05, "loss": 0.1630117654800415, "step": 13440 }, { "epoch": 0.3542935038508016, "grad_norm": 2.984853744506836, "learning_rate": 1.4973871694281133e-05, "loss": 0.1805511236190796, "step": 13450 }, { "epoch": 0.35455691909530035, "grad_norm": 2.152420997619629, "learning_rate": 1.4966468674686226e-05, "loss": 0.16272828578948975, "step": 13460 }, { "epoch": 0.3548203343397991, "grad_norm": 4.695549011230469, "learning_rate": 1.4959062040281143e-05, "loss": 0.17875113487243652, "step": 13470 }, { "epoch": 0.35508374958429784, "grad_norm": 2.0016729831695557, "learning_rate": 1.4951651796456758e-05, "loss": 0.18086705207824708, "step": 13480 }, { "epoch": 0.3553471648287966, "grad_norm": 1.1547391414642334, "learning_rate": 1.4944237948606561e-05, "loss": 0.16321851015090943, "step": 13490 }, { "epoch": 0.35561058007329527, "grad_norm": 0.9589448571205139, "learning_rate": 1.4936820502126676e-05, "loss": 0.16560784578323365, "step": 13500 }, { "epoch": 0.355873995317794, "grad_norm": 1.9077152013778687, "learning_rate": 1.4929399462415832e-05, "loss": 0.17694411277770997, "step": 13510 }, { "epoch": 0.35613741056229276, "grad_norm": 1.214753270149231, "learning_rate": 1.492197483487539e-05, "loss": 0.16394978761672974, "step": 13520 }, { "epoch": 0.3564008258067915, "grad_norm": 2.969041109085083, "learning_rate": 1.491454662490931e-05, "loss": 0.17767624855041503, "step": 13530 }, { "epoch": 0.35666424105129024, "grad_norm": 6.862597465515137, "learning_rate": 1.4907114837924168e-05, "loss": 0.17159799337387086, "step": 13540 }, { "epoch": 0.356927656295789, "grad_norm": 1.019300103187561, "learning_rate": 1.4899679479329133e-05, "loss": 0.1649252414703369, "step": 13550 }, { "epoch": 0.35719107154028773, "grad_norm": 2.835045576095581, "learning_rate": 1.4892240554535981e-05, "loss": 0.1658864974975586, "step": 13560 }, { "epoch": 0.3574544867847865, "grad_norm": 1.246625304222107, "learning_rate": 1.4884798068959085e-05, "loss": 0.16799709796905518, "step": 13570 }, { "epoch": 0.3577179020292852, "grad_norm": 1.5306698083877563, "learning_rate": 1.4877352028015408e-05, "loss": 0.183358097076416, "step": 13580 }, { "epoch": 0.3579813172737839, "grad_norm": 1.6551965475082397, "learning_rate": 1.4869902437124499e-05, "loss": 0.19456355571746825, "step": 13590 }, { "epoch": 0.35824473251828265, "grad_norm": 1.722974181175232, "learning_rate": 1.486244930170849e-05, "loss": 0.18188570737838744, "step": 13600 }, { "epoch": 0.3585081477627814, "grad_norm": 1.4645273685455322, "learning_rate": 1.4854992627192096e-05, "loss": 0.17642672061920167, "step": 13610 }, { "epoch": 0.35877156300728014, "grad_norm": 2.929185390472412, "learning_rate": 1.4847532419002602e-05, "loss": 0.17506096363067628, "step": 13620 }, { "epoch": 0.3590349782517789, "grad_norm": 1.091428279876709, "learning_rate": 1.4840068682569877e-05, "loss": 0.18321537971496582, "step": 13630 }, { "epoch": 0.3592983934962776, "grad_norm": 3.49985408782959, "learning_rate": 1.4832601423326346e-05, "loss": 0.1599287986755371, "step": 13640 }, { "epoch": 0.35956180874077637, "grad_norm": 1.9443508386611938, "learning_rate": 1.4825130646707002e-05, "loss": 0.14993423223495483, "step": 13650 }, { "epoch": 0.3598252239852751, "grad_norm": 2.027369260787964, "learning_rate": 1.4817656358149398e-05, "loss": 0.18686468601226808, "step": 13660 }, { "epoch": 0.3600886392297738, "grad_norm": 1.6312741041183472, "learning_rate": 1.4810178563093648e-05, "loss": 0.17834455966949464, "step": 13670 }, { "epoch": 0.36035205447427254, "grad_norm": 1.911319375038147, "learning_rate": 1.4802697266982411e-05, "loss": 0.18439233303070068, "step": 13680 }, { "epoch": 0.3606154697187713, "grad_norm": 2.9120428562164307, "learning_rate": 1.4795212475260896e-05, "loss": 0.18051773309707642, "step": 13690 }, { "epoch": 0.36087888496327003, "grad_norm": 2.1217615604400635, "learning_rate": 1.4787724193376858e-05, "loss": 0.16833999156951904, "step": 13700 }, { "epoch": 0.3611423002077688, "grad_norm": 2.9514384269714355, "learning_rate": 1.4780232426780592e-05, "loss": 0.1669293761253357, "step": 13710 }, { "epoch": 0.3614057154522675, "grad_norm": 1.3177789449691772, "learning_rate": 1.477273718092493e-05, "loss": 0.17717446088790895, "step": 13720 }, { "epoch": 0.36166913069676626, "grad_norm": 1.5838464498519897, "learning_rate": 1.4765238461265235e-05, "loss": 0.18068480491638184, "step": 13730 }, { "epoch": 0.361932545941265, "grad_norm": 2.705155611038208, "learning_rate": 1.4757736273259396e-05, "loss": 0.1870967149734497, "step": 13740 }, { "epoch": 0.36219596118576375, "grad_norm": 1.9277290105819702, "learning_rate": 1.4750230622367831e-05, "loss": 0.1774294376373291, "step": 13750 }, { "epoch": 0.36245937643026244, "grad_norm": 4.3243794441223145, "learning_rate": 1.474272151405348e-05, "loss": 0.17754411697387695, "step": 13760 }, { "epoch": 0.3627227916747612, "grad_norm": 1.6988685131072998, "learning_rate": 1.473520895378179e-05, "loss": 0.18587102890014648, "step": 13770 }, { "epoch": 0.3629862069192599, "grad_norm": 2.256678581237793, "learning_rate": 1.472769294702073e-05, "loss": 0.2001034736633301, "step": 13780 }, { "epoch": 0.36324962216375867, "grad_norm": 4.8723273277282715, "learning_rate": 1.4720173499240771e-05, "loss": 0.17847115993499757, "step": 13790 }, { "epoch": 0.3635130374082574, "grad_norm": 2.828245162963867, "learning_rate": 1.4712650615914899e-05, "loss": 0.18516237735748292, "step": 13800 }, { "epoch": 0.36377645265275615, "grad_norm": 3.544750928878784, "learning_rate": 1.4705124302518581e-05, "loss": 0.19686685800552367, "step": 13810 }, { "epoch": 0.3640398678972549, "grad_norm": 3.1968672275543213, "learning_rate": 1.4697594564529805e-05, "loss": 0.1885266423225403, "step": 13820 }, { "epoch": 0.36430328314175364, "grad_norm": 1.0631296634674072, "learning_rate": 1.4690061407429028e-05, "loss": 0.17325570583343505, "step": 13830 }, { "epoch": 0.3645666983862524, "grad_norm": 1.374678134918213, "learning_rate": 1.468252483669922e-05, "loss": 0.1892564296722412, "step": 13840 }, { "epoch": 0.3648301136307511, "grad_norm": 9.228960990905762, "learning_rate": 1.4674984857825808e-05, "loss": 0.19584553241729735, "step": 13850 }, { "epoch": 0.3650935288752498, "grad_norm": 3.4635376930236816, "learning_rate": 1.4667441476296723e-05, "loss": 0.1896182894706726, "step": 13860 }, { "epoch": 0.36535694411974856, "grad_norm": 1.3389685153961182, "learning_rate": 1.4659894697602358e-05, "loss": 0.18262739181518556, "step": 13870 }, { "epoch": 0.3656203593642473, "grad_norm": 2.9314141273498535, "learning_rate": 1.465234452723559e-05, "loss": 0.1690659999847412, "step": 13880 }, { "epoch": 0.36588377460874605, "grad_norm": 2.0952696800231934, "learning_rate": 1.4644790970691755e-05, "loss": 0.1695370078086853, "step": 13890 }, { "epoch": 0.3661471898532448, "grad_norm": 2.6981639862060547, "learning_rate": 1.4637234033468658e-05, "loss": 0.1685786485671997, "step": 13900 }, { "epoch": 0.36641060509774354, "grad_norm": 1.4132907390594482, "learning_rate": 1.4629673721066566e-05, "loss": 0.18397583961486816, "step": 13910 }, { "epoch": 0.3666740203422423, "grad_norm": 1.951647400856018, "learning_rate": 1.4622110038988196e-05, "loss": 0.18415147066116333, "step": 13920 }, { "epoch": 0.36693743558674097, "grad_norm": 1.5461570024490356, "learning_rate": 1.4614542992738727e-05, "loss": 0.19179947376251222, "step": 13930 }, { "epoch": 0.3672008508312397, "grad_norm": 1.4186636209487915, "learning_rate": 1.4606972587825779e-05, "loss": 0.17973723411560058, "step": 13940 }, { "epoch": 0.36746426607573845, "grad_norm": 2.744619369506836, "learning_rate": 1.4599398829759419e-05, "loss": 0.1705977201461792, "step": 13950 }, { "epoch": 0.3677276813202372, "grad_norm": 1.1742850542068481, "learning_rate": 1.4591821724052155e-05, "loss": 0.1744845151901245, "step": 13960 }, { "epoch": 0.36799109656473594, "grad_norm": 1.7637357711791992, "learning_rate": 1.4584241276218935e-05, "loss": 0.17893624305725098, "step": 13970 }, { "epoch": 0.3682545118092347, "grad_norm": 1.8025087118148804, "learning_rate": 1.4576657491777134e-05, "loss": 0.18446192741394044, "step": 13980 }, { "epoch": 0.36851792705373343, "grad_norm": 1.806636095046997, "learning_rate": 1.4569070376246552e-05, "loss": 0.182353413105011, "step": 13990 }, { "epoch": 0.3687813422982322, "grad_norm": 1.4342446327209473, "learning_rate": 1.4561479935149426e-05, "loss": 0.17758095264434814, "step": 14000 }, { "epoch": 0.3690447575427309, "grad_norm": 2.284421682357788, "learning_rate": 1.4553886174010405e-05, "loss": 0.19463223218917847, "step": 14010 }, { "epoch": 0.3693081727872296, "grad_norm": 0.9198055863380432, "learning_rate": 1.4546289098356556e-05, "loss": 0.1613471031188965, "step": 14020 }, { "epoch": 0.36957158803172835, "grad_norm": 1.0152322053909302, "learning_rate": 1.4538688713717357e-05, "loss": 0.17846918106079102, "step": 14030 }, { "epoch": 0.3698350032762271, "grad_norm": 4.009786605834961, "learning_rate": 1.4531085025624694e-05, "loss": 0.18012341260910034, "step": 14040 }, { "epoch": 0.37009841852072584, "grad_norm": 2.0352208614349365, "learning_rate": 1.4523478039612863e-05, "loss": 0.1976252794265747, "step": 14050 }, { "epoch": 0.3703618337652246, "grad_norm": 1.4286140203475952, "learning_rate": 1.4515867761218558e-05, "loss": 0.1773144006729126, "step": 14060 }, { "epoch": 0.3706252490097233, "grad_norm": 2.540377378463745, "learning_rate": 1.4508254195980861e-05, "loss": 0.16634659767150878, "step": 14070 }, { "epoch": 0.37088866425422207, "grad_norm": 1.6697752475738525, "learning_rate": 1.450063734944126e-05, "loss": 0.1939622402191162, "step": 14080 }, { "epoch": 0.3711520794987208, "grad_norm": 1.771666407585144, "learning_rate": 1.449301722714362e-05, "loss": 0.17174761295318602, "step": 14090 }, { "epoch": 0.37141549474321955, "grad_norm": 2.0876100063323975, "learning_rate": 1.4485393834634202e-05, "loss": 0.1822780728340149, "step": 14100 }, { "epoch": 0.37167890998771824, "grad_norm": 2.0115866661071777, "learning_rate": 1.447776717746163e-05, "loss": 0.17347532510757446, "step": 14110 }, { "epoch": 0.371942325232217, "grad_norm": 1.199824571609497, "learning_rate": 1.4470137261176922e-05, "loss": 0.17679591178894044, "step": 14120 }, { "epoch": 0.37220574047671573, "grad_norm": 4.53000020980835, "learning_rate": 1.4462504091333454e-05, "loss": 0.17770609855651856, "step": 14130 }, { "epoch": 0.3724691557212145, "grad_norm": 1.2624868154525757, "learning_rate": 1.4454867673486983e-05, "loss": 0.1802108645439148, "step": 14140 }, { "epoch": 0.3727325709657132, "grad_norm": 1.6515958309173584, "learning_rate": 1.4447228013195616e-05, "loss": 0.1860613465309143, "step": 14150 }, { "epoch": 0.37299598621021196, "grad_norm": 1.1961920261383057, "learning_rate": 1.4439585116019835e-05, "loss": 0.17851009368896484, "step": 14160 }, { "epoch": 0.3732594014547107, "grad_norm": 2.2280306816101074, "learning_rate": 1.4431938987522463e-05, "loss": 0.1894020676612854, "step": 14170 }, { "epoch": 0.37352281669920945, "grad_norm": 2.467132568359375, "learning_rate": 1.4424289633268689e-05, "loss": 0.1912940263748169, "step": 14180 }, { "epoch": 0.37378623194370814, "grad_norm": 1.0196329355239868, "learning_rate": 1.4416637058826038e-05, "loss": 0.1647688627243042, "step": 14190 }, { "epoch": 0.3740496471882069, "grad_norm": 2.2714481353759766, "learning_rate": 1.4408981269764386e-05, "loss": 0.1825883150100708, "step": 14200 }, { "epoch": 0.3743130624327056, "grad_norm": 2.9740664958953857, "learning_rate": 1.4401322271655947e-05, "loss": 0.1781484842300415, "step": 14210 }, { "epoch": 0.37457647767720437, "grad_norm": 1.9755860567092896, "learning_rate": 1.4393660070075267e-05, "loss": 0.1780252456665039, "step": 14220 }, { "epoch": 0.3748398929217031, "grad_norm": 2.7239482402801514, "learning_rate": 1.4385994670599233e-05, "loss": 0.178538715839386, "step": 14230 }, { "epoch": 0.37510330816620185, "grad_norm": 1.567705512046814, "learning_rate": 1.4378326078807048e-05, "loss": 0.1788170576095581, "step": 14240 }, { "epoch": 0.3753667234107006, "grad_norm": 1.0296019315719604, "learning_rate": 1.4370654300280246e-05, "loss": 0.18709529638290406, "step": 14250 }, { "epoch": 0.37563013865519934, "grad_norm": 1.723708987236023, "learning_rate": 1.4362979340602677e-05, "loss": 0.18076547384262084, "step": 14260 }, { "epoch": 0.3758935538996981, "grad_norm": 1.1453943252563477, "learning_rate": 1.4355301205360513e-05, "loss": 0.17582371234893798, "step": 14270 }, { "epoch": 0.3761569691441968, "grad_norm": 1.0228322744369507, "learning_rate": 1.4347619900142223e-05, "loss": 0.20957860946655274, "step": 14280 }, { "epoch": 0.3764203843886955, "grad_norm": 2.2065680027008057, "learning_rate": 1.43399354305386e-05, "loss": 0.1657517671585083, "step": 14290 }, { "epoch": 0.37668379963319426, "grad_norm": 4.523097038269043, "learning_rate": 1.4332247802142729e-05, "loss": 0.1965700626373291, "step": 14300 }, { "epoch": 0.376947214877693, "grad_norm": 1.2722824811935425, "learning_rate": 1.4324557020550001e-05, "loss": 0.17417627573013306, "step": 14310 }, { "epoch": 0.37721063012219175, "grad_norm": 1.4134076833724976, "learning_rate": 1.4316863091358093e-05, "loss": 0.17208629846572876, "step": 14320 }, { "epoch": 0.3774740453666905, "grad_norm": 0.8680151700973511, "learning_rate": 1.4309166020166984e-05, "loss": 0.15696561336517334, "step": 14330 }, { "epoch": 0.37773746061118924, "grad_norm": 1.2299469709396362, "learning_rate": 1.430146581257893e-05, "loss": 0.17729976177215576, "step": 14340 }, { "epoch": 0.378000875855688, "grad_norm": 3.5236475467681885, "learning_rate": 1.4293762474198478e-05, "loss": 0.17370107173919677, "step": 14350 }, { "epoch": 0.3782642911001867, "grad_norm": 2.248525857925415, "learning_rate": 1.428605601063245e-05, "loss": 0.16842501163482665, "step": 14360 }, { "epoch": 0.3785277063446854, "grad_norm": 1.218157410621643, "learning_rate": 1.427834642748994e-05, "loss": 0.1715567708015442, "step": 14370 }, { "epoch": 0.37879112158918415, "grad_norm": 1.7554597854614258, "learning_rate": 1.4270633730382315e-05, "loss": 0.179646897315979, "step": 14380 }, { "epoch": 0.3790545368336829, "grad_norm": 3.1658289432525635, "learning_rate": 1.4262917924923212e-05, "loss": 0.16792491674423218, "step": 14390 }, { "epoch": 0.37931795207818164, "grad_norm": 2.7526602745056152, "learning_rate": 1.4255199016728527e-05, "loss": 0.18049395084381104, "step": 14400 }, { "epoch": 0.3795813673226804, "grad_norm": 1.1425248384475708, "learning_rate": 1.4247477011416409e-05, "loss": 0.18504834175109863, "step": 14410 }, { "epoch": 0.37984478256717913, "grad_norm": 1.2786535024642944, "learning_rate": 1.4239751914607274e-05, "loss": 0.18211698532104492, "step": 14420 }, { "epoch": 0.3801081978116779, "grad_norm": 3.2687041759490967, "learning_rate": 1.4232023731923772e-05, "loss": 0.15926840305328369, "step": 14430 }, { "epoch": 0.3803716130561766, "grad_norm": 4.270437717437744, "learning_rate": 1.422429246899082e-05, "loss": 0.19389872550964354, "step": 14440 }, { "epoch": 0.3806350283006753, "grad_norm": 2.267202854156494, "learning_rate": 1.4216558131435551e-05, "loss": 0.179032039642334, "step": 14450 }, { "epoch": 0.38089844354517405, "grad_norm": 2.2588469982147217, "learning_rate": 1.420882072488736e-05, "loss": 0.17170683145523072, "step": 14460 }, { "epoch": 0.3811618587896728, "grad_norm": 2.2212421894073486, "learning_rate": 1.420108025497786e-05, "loss": 0.192425000667572, "step": 14470 }, { "epoch": 0.38142527403417154, "grad_norm": 1.3390552997589111, "learning_rate": 1.41933367273409e-05, "loss": 0.17350034713745116, "step": 14480 }, { "epoch": 0.3816886892786703, "grad_norm": 1.366836428642273, "learning_rate": 1.4185590147612555e-05, "loss": 0.181963050365448, "step": 14490 }, { "epoch": 0.381952104523169, "grad_norm": 4.645387649536133, "learning_rate": 1.4177840521431117e-05, "loss": 0.17644420862197877, "step": 14500 }, { "epoch": 0.38221551976766777, "grad_norm": 5.551546573638916, "learning_rate": 1.4170087854437102e-05, "loss": 0.1659933567047119, "step": 14510 }, { "epoch": 0.3824789350121665, "grad_norm": 3.1508867740631104, "learning_rate": 1.4162332152273233e-05, "loss": 0.19190326929092408, "step": 14520 }, { "epoch": 0.38274235025666525, "grad_norm": 2.523735284805298, "learning_rate": 1.4154573420584448e-05, "loss": 0.16799418926239013, "step": 14530 }, { "epoch": 0.38300576550116394, "grad_norm": 1.276528000831604, "learning_rate": 1.414681166501788e-05, "loss": 0.1767870545387268, "step": 14540 }, { "epoch": 0.3832691807456627, "grad_norm": 1.0367372035980225, "learning_rate": 1.4139046891222877e-05, "loss": 0.20049829483032228, "step": 14550 }, { "epoch": 0.38353259599016143, "grad_norm": 1.1977585554122925, "learning_rate": 1.4131279104850968e-05, "loss": 0.20447869300842286, "step": 14560 }, { "epoch": 0.3837960112346602, "grad_norm": 1.0546820163726807, "learning_rate": 1.4123508311555894e-05, "loss": 0.17189326286315917, "step": 14570 }, { "epoch": 0.3840594264791589, "grad_norm": 2.4039125442504883, "learning_rate": 1.4115734516993562e-05, "loss": 0.18335604667663574, "step": 14580 }, { "epoch": 0.38432284172365766, "grad_norm": 4.795260906219482, "learning_rate": 1.4107957726822084e-05, "loss": 0.1825007438659668, "step": 14590 }, { "epoch": 0.3845862569681564, "grad_norm": 3.655802011489868, "learning_rate": 1.4100177946701738e-05, "loss": 0.18846161365509034, "step": 14600 }, { "epoch": 0.38484967221265515, "grad_norm": 1.452875018119812, "learning_rate": 1.4092395182294989e-05, "loss": 0.17894858121871948, "step": 14610 }, { "epoch": 0.3851130874571539, "grad_norm": 1.2618865966796875, "learning_rate": 1.4084609439266464e-05, "loss": 0.18969030380249025, "step": 14620 }, { "epoch": 0.3853765027016526, "grad_norm": 2.3781440258026123, "learning_rate": 1.4076820723282962e-05, "loss": 0.1736894130706787, "step": 14630 }, { "epoch": 0.3856399179461513, "grad_norm": 1.3749862909317017, "learning_rate": 1.4069029040013448e-05, "loss": 0.20233848094940185, "step": 14640 }, { "epoch": 0.38590333319065007, "grad_norm": 2.146799087524414, "learning_rate": 1.4061234395129049e-05, "loss": 0.18557729721069335, "step": 14650 }, { "epoch": 0.3861667484351488, "grad_norm": 2.1530630588531494, "learning_rate": 1.405343679430304e-05, "loss": 0.17715946435928345, "step": 14660 }, { "epoch": 0.38643016367964755, "grad_norm": 1.854191780090332, "learning_rate": 1.4045636243210852e-05, "loss": 0.18509243726730346, "step": 14670 }, { "epoch": 0.3866935789241463, "grad_norm": 0.8441782593727112, "learning_rate": 1.4037832747530063e-05, "loss": 0.16197564601898193, "step": 14680 }, { "epoch": 0.38695699416864504, "grad_norm": 1.4459177255630493, "learning_rate": 1.403002631294039e-05, "loss": 0.17293448448181153, "step": 14690 }, { "epoch": 0.3872204094131438, "grad_norm": 1.9358233213424683, "learning_rate": 1.40222169451237e-05, "loss": 0.15965094566345214, "step": 14700 }, { "epoch": 0.3874838246576425, "grad_norm": 0.8645235896110535, "learning_rate": 1.4014404649763983e-05, "loss": 0.17921953201293944, "step": 14710 }, { "epoch": 0.3877472399021412, "grad_norm": 3.2861852645874023, "learning_rate": 1.4006589432547367e-05, "loss": 0.17544753551483155, "step": 14720 }, { "epoch": 0.38801065514663996, "grad_norm": 2.8116018772125244, "learning_rate": 1.3998771299162099e-05, "loss": 0.177012038230896, "step": 14730 }, { "epoch": 0.3882740703911387, "grad_norm": 2.9405529499053955, "learning_rate": 1.3990950255298565e-05, "loss": 0.1784972906112671, "step": 14740 }, { "epoch": 0.38853748563563745, "grad_norm": 1.187866449356079, "learning_rate": 1.3983126306649243e-05, "loss": 0.19698739051818848, "step": 14750 }, { "epoch": 0.3888009008801362, "grad_norm": 3.3484678268432617, "learning_rate": 1.3975299458908753e-05, "loss": 0.16687976121902465, "step": 14760 }, { "epoch": 0.38906431612463493, "grad_norm": 2.213905096054077, "learning_rate": 1.3967469717773807e-05, "loss": 0.1746382713317871, "step": 14770 }, { "epoch": 0.3893277313691337, "grad_norm": 1.7430022954940796, "learning_rate": 1.395963708894323e-05, "loss": 0.16971678733825685, "step": 14780 }, { "epoch": 0.3895911466136324, "grad_norm": 3.0406417846679688, "learning_rate": 1.3951801578117946e-05, "loss": 0.18852503299713136, "step": 14790 }, { "epoch": 0.3898545618581311, "grad_norm": 0.9979509115219116, "learning_rate": 1.3943963191000976e-05, "loss": 0.17447800636291505, "step": 14800 }, { "epoch": 0.39011797710262985, "grad_norm": 2.5065605640411377, "learning_rate": 1.393612193329744e-05, "loss": 0.1943203926086426, "step": 14810 }, { "epoch": 0.3903813923471286, "grad_norm": 4.445403575897217, "learning_rate": 1.3928277810714542e-05, "loss": 0.17682074308395385, "step": 14820 }, { "epoch": 0.39064480759162734, "grad_norm": 2.4583303928375244, "learning_rate": 1.3920430828961576e-05, "loss": 0.16664910316467285, "step": 14830 }, { "epoch": 0.3909082228361261, "grad_norm": 1.8124672174453735, "learning_rate": 1.3912580993749907e-05, "loss": 0.17071974277496338, "step": 14840 }, { "epoch": 0.39117163808062483, "grad_norm": 2.190932035446167, "learning_rate": 1.3904728310792988e-05, "loss": 0.1854402780532837, "step": 14850 }, { "epoch": 0.39143505332512357, "grad_norm": 1.440690279006958, "learning_rate": 1.3896872785806342e-05, "loss": 0.17143096923828124, "step": 14860 }, { "epoch": 0.3916984685696223, "grad_norm": 3.98307204246521, "learning_rate": 1.3889014424507557e-05, "loss": 0.16993486881256104, "step": 14870 }, { "epoch": 0.39196188381412106, "grad_norm": 1.6713179349899292, "learning_rate": 1.3881153232616284e-05, "loss": 0.17508437633514404, "step": 14880 }, { "epoch": 0.39222529905861975, "grad_norm": 1.2886433601379395, "learning_rate": 1.3873289215854247e-05, "loss": 0.17676153182983398, "step": 14890 }, { "epoch": 0.3924887143031185, "grad_norm": 2.373667001724243, "learning_rate": 1.3865422379945205e-05, "loss": 0.16521501541137695, "step": 14900 }, { "epoch": 0.39275212954761723, "grad_norm": 1.451012134552002, "learning_rate": 1.3857552730614989e-05, "loss": 0.2034604787826538, "step": 14910 }, { "epoch": 0.393015544792116, "grad_norm": 1.6663886308670044, "learning_rate": 1.3849680273591468e-05, "loss": 0.17126225233078002, "step": 14920 }, { "epoch": 0.3932789600366147, "grad_norm": 2.3925893306732178, "learning_rate": 1.3841805014604553e-05, "loss": 0.18704500198364257, "step": 14930 }, { "epoch": 0.39354237528111347, "grad_norm": 2.6039845943450928, "learning_rate": 1.3833926959386199e-05, "loss": 0.17937991619110108, "step": 14940 }, { "epoch": 0.3938057905256122, "grad_norm": 2.2728350162506104, "learning_rate": 1.3826046113670395e-05, "loss": 0.1867887020111084, "step": 14950 }, { "epoch": 0.39406920577011095, "grad_norm": 1.4760841131210327, "learning_rate": 1.3818162483193159e-05, "loss": 0.1759882688522339, "step": 14960 }, { "epoch": 0.39433262101460964, "grad_norm": 1.1043930053710938, "learning_rate": 1.3810276073692537e-05, "loss": 0.1584068775177002, "step": 14970 }, { "epoch": 0.3945960362591084, "grad_norm": 1.4351797103881836, "learning_rate": 1.3802386890908602e-05, "loss": 0.17329423427581786, "step": 14980 }, { "epoch": 0.39485945150360713, "grad_norm": 2.1902573108673096, "learning_rate": 1.3794494940583437e-05, "loss": 0.18261487483978273, "step": 14990 }, { "epoch": 0.39512286674810587, "grad_norm": 1.845354676246643, "learning_rate": 1.3786600228461147e-05, "loss": 0.18555394411087037, "step": 15000 }, { "epoch": 0.3953862819926046, "grad_norm": 3.149162530899048, "learning_rate": 1.377870276028784e-05, "loss": 0.16789793968200684, "step": 15010 }, { "epoch": 0.39564969723710336, "grad_norm": 1.6211767196655273, "learning_rate": 1.377080254181164e-05, "loss": 0.18959145545959472, "step": 15020 }, { "epoch": 0.3959131124816021, "grad_norm": 1.5801867246627808, "learning_rate": 1.3762899578782662e-05, "loss": 0.18773093223571777, "step": 15030 }, { "epoch": 0.39617652772610085, "grad_norm": 3.9069511890411377, "learning_rate": 1.375499387695303e-05, "loss": 0.17772629261016845, "step": 15040 }, { "epoch": 0.3964399429705996, "grad_norm": 1.576292872428894, "learning_rate": 1.3747085442076844e-05, "loss": 0.17474972009658812, "step": 15050 }, { "epoch": 0.3967033582150983, "grad_norm": 0.8301007747650146, "learning_rate": 1.3739174279910214e-05, "loss": 0.17766108512878417, "step": 15060 }, { "epoch": 0.396966773459597, "grad_norm": 1.4516812562942505, "learning_rate": 1.3731260396211219e-05, "loss": 0.16301745176315308, "step": 15070 }, { "epoch": 0.39723018870409577, "grad_norm": 3.3551995754241943, "learning_rate": 1.372334379673993e-05, "loss": 0.18070412874221803, "step": 15080 }, { "epoch": 0.3974936039485945, "grad_norm": 0.9678198099136353, "learning_rate": 1.3715424487258386e-05, "loss": 0.16579755544662475, "step": 15090 }, { "epoch": 0.39775701919309325, "grad_norm": 2.0561625957489014, "learning_rate": 1.37075024735306e-05, "loss": 0.18159104585647584, "step": 15100 }, { "epoch": 0.398020434437592, "grad_norm": 1.4270095825195312, "learning_rate": 1.3699577761322563e-05, "loss": 0.17488161325454712, "step": 15110 }, { "epoch": 0.39828384968209074, "grad_norm": 1.5594359636306763, "learning_rate": 1.3691650356402212e-05, "loss": 0.18533861637115479, "step": 15120 }, { "epoch": 0.3985472649265895, "grad_norm": 1.413525104522705, "learning_rate": 1.3683720264539469e-05, "loss": 0.16227113008499144, "step": 15130 }, { "epoch": 0.3988106801710882, "grad_norm": 1.85318124294281, "learning_rate": 1.3675787491506184e-05, "loss": 0.17325127124786377, "step": 15140 }, { "epoch": 0.3990740954155869, "grad_norm": 1.4001044034957886, "learning_rate": 1.3667852043076173e-05, "loss": 0.17343976497650146, "step": 15150 }, { "epoch": 0.39933751066008566, "grad_norm": 3.2805333137512207, "learning_rate": 1.3659913925025203e-05, "loss": 0.19328365325927735, "step": 15160 }, { "epoch": 0.3996009259045844, "grad_norm": 3.477764368057251, "learning_rate": 1.3651973143130979e-05, "loss": 0.1753884196281433, "step": 15170 }, { "epoch": 0.39986434114908315, "grad_norm": 1.80180823802948, "learning_rate": 1.3644029703173139e-05, "loss": 0.1622506022453308, "step": 15180 }, { "epoch": 0.4001277563935819, "grad_norm": 1.4251396656036377, "learning_rate": 1.3636083610933262e-05, "loss": 0.17928144931793213, "step": 15190 }, { "epoch": 0.40039117163808063, "grad_norm": 2.894774913787842, "learning_rate": 1.3628134872194861e-05, "loss": 0.17723913192749025, "step": 15200 }, { "epoch": 0.4006545868825794, "grad_norm": 2.4909894466400146, "learning_rate": 1.362018349274337e-05, "loss": 0.17343759536743164, "step": 15210 }, { "epoch": 0.4009180021270781, "grad_norm": 3.168888807296753, "learning_rate": 1.3612229478366144e-05, "loss": 0.17257814407348632, "step": 15220 }, { "epoch": 0.4011814173715768, "grad_norm": 2.261807441711426, "learning_rate": 1.360427283485246e-05, "loss": 0.1775970697402954, "step": 15230 }, { "epoch": 0.40144483261607555, "grad_norm": 0.8062502145767212, "learning_rate": 1.3596313567993504e-05, "loss": 0.17988035678863526, "step": 15240 }, { "epoch": 0.4017082478605743, "grad_norm": 1.377397060394287, "learning_rate": 1.3588351683582374e-05, "loss": 0.16083812713623047, "step": 15250 }, { "epoch": 0.40197166310507304, "grad_norm": 1.4973891973495483, "learning_rate": 1.3580387187414077e-05, "loss": 0.1715841054916382, "step": 15260 }, { "epoch": 0.4022350783495718, "grad_norm": 2.5318217277526855, "learning_rate": 1.3572420085285512e-05, "loss": 0.17030395269393922, "step": 15270 }, { "epoch": 0.4024984935940705, "grad_norm": 2.037083625793457, "learning_rate": 1.3564450382995482e-05, "loss": 0.1772638201713562, "step": 15280 }, { "epoch": 0.40276190883856927, "grad_norm": 1.8187894821166992, "learning_rate": 1.3556478086344678e-05, "loss": 0.16361876726150512, "step": 15290 }, { "epoch": 0.403025324083068, "grad_norm": 3.0293548107147217, "learning_rate": 1.3548503201135687e-05, "loss": 0.14734237194061278, "step": 15300 }, { "epoch": 0.40328873932756676, "grad_norm": 2.1533265113830566, "learning_rate": 1.3540525733172966e-05, "loss": 0.19285213947296143, "step": 15310 }, { "epoch": 0.40355215457206545, "grad_norm": 1.0574487447738647, "learning_rate": 1.353254568826287e-05, "loss": 0.17075965404510499, "step": 15320 }, { "epoch": 0.4038155698165642, "grad_norm": 2.660959005355835, "learning_rate": 1.3524563072213612e-05, "loss": 0.18781666755676268, "step": 15330 }, { "epoch": 0.40407898506106293, "grad_norm": 1.5512830018997192, "learning_rate": 1.3516577890835292e-05, "loss": 0.17299675941467285, "step": 15340 }, { "epoch": 0.4043424003055617, "grad_norm": 1.0949249267578125, "learning_rate": 1.350859014993986e-05, "loss": 0.18575432300567626, "step": 15350 }, { "epoch": 0.4046058155500604, "grad_norm": 1.1538299322128296, "learning_rate": 1.3500599855341146e-05, "loss": 0.1753160238265991, "step": 15360 }, { "epoch": 0.40486923079455917, "grad_norm": 3.681753158569336, "learning_rate": 1.3492607012854824e-05, "loss": 0.18030353784561157, "step": 15370 }, { "epoch": 0.4051326460390579, "grad_norm": 1.5603190660476685, "learning_rate": 1.3484611628298438e-05, "loss": 0.16052650213241576, "step": 15380 }, { "epoch": 0.40539606128355665, "grad_norm": 5.771909713745117, "learning_rate": 1.3476613707491372e-05, "loss": 0.16646180152893067, "step": 15390 }, { "epoch": 0.4056594765280554, "grad_norm": 0.5745401978492737, "learning_rate": 1.3468613256254852e-05, "loss": 0.15984324216842652, "step": 15400 }, { "epoch": 0.4059228917725541, "grad_norm": 1.8917344808578491, "learning_rate": 1.3460610280411957e-05, "loss": 0.18412952423095702, "step": 15410 }, { "epoch": 0.40618630701705283, "grad_norm": 2.17368745803833, "learning_rate": 1.3452604785787595e-05, "loss": 0.17663817405700682, "step": 15420 }, { "epoch": 0.40644972226155157, "grad_norm": 1.8421632051467896, "learning_rate": 1.3444596778208516e-05, "loss": 0.17601417303085326, "step": 15430 }, { "epoch": 0.4067131375060503, "grad_norm": 1.8334579467773438, "learning_rate": 1.3436586263503286e-05, "loss": 0.16824667453765868, "step": 15440 }, { "epoch": 0.40697655275054906, "grad_norm": 2.012174367904663, "learning_rate": 1.3428573247502307e-05, "loss": 0.18104147911071777, "step": 15450 }, { "epoch": 0.4072399679950478, "grad_norm": 3.48150372505188, "learning_rate": 1.3420557736037796e-05, "loss": 0.18166885375976563, "step": 15460 }, { "epoch": 0.40750338323954655, "grad_norm": 1.4528682231903076, "learning_rate": 1.3412539734943792e-05, "loss": 0.16083680391311644, "step": 15470 }, { "epoch": 0.4077667984840453, "grad_norm": 1.2245419025421143, "learning_rate": 1.3404519250056139e-05, "loss": 0.16763445138931274, "step": 15480 }, { "epoch": 0.408030213728544, "grad_norm": 2.297454357147217, "learning_rate": 1.3396496287212491e-05, "loss": 0.17226316928863525, "step": 15490 }, { "epoch": 0.4082936289730427, "grad_norm": 2.7081363201141357, "learning_rate": 1.3388470852252307e-05, "loss": 0.17569575309753419, "step": 15500 }, { "epoch": 0.40855704421754147, "grad_norm": 1.7869937419891357, "learning_rate": 1.3380442951016846e-05, "loss": 0.1644347071647644, "step": 15510 }, { "epoch": 0.4088204594620402, "grad_norm": 2.8126988410949707, "learning_rate": 1.3372412589349164e-05, "loss": 0.17785706520080566, "step": 15520 }, { "epoch": 0.40908387470653895, "grad_norm": 3.0396010875701904, "learning_rate": 1.3364379773094099e-05, "loss": 0.18173391819000245, "step": 15530 }, { "epoch": 0.4093472899510377, "grad_norm": 2.792931079864502, "learning_rate": 1.335634450809828e-05, "loss": 0.18704177141189576, "step": 15540 }, { "epoch": 0.40961070519553644, "grad_norm": 1.3731138706207275, "learning_rate": 1.3348306800210127e-05, "loss": 0.17369611263275148, "step": 15550 }, { "epoch": 0.4098741204400352, "grad_norm": 1.8987396955490112, "learning_rate": 1.3340266655279828e-05, "loss": 0.16791067123413086, "step": 15560 }, { "epoch": 0.4101375356845339, "grad_norm": 1.131542682647705, "learning_rate": 1.3332224079159342e-05, "loss": 0.17670743465423583, "step": 15570 }, { "epoch": 0.4104009509290326, "grad_norm": 1.3619425296783447, "learning_rate": 1.3324179077702408e-05, "loss": 0.1813466429710388, "step": 15580 }, { "epoch": 0.41066436617353136, "grad_norm": 3.922915458679199, "learning_rate": 1.3316131656764523e-05, "loss": 0.15228664875030518, "step": 15590 }, { "epoch": 0.4109277814180301, "grad_norm": 2.550011157989502, "learning_rate": 1.3308081822202953e-05, "loss": 0.17774453163146972, "step": 15600 }, { "epoch": 0.41119119666252885, "grad_norm": 3.625101327896118, "learning_rate": 1.3300029579876705e-05, "loss": 0.19138699769973755, "step": 15610 }, { "epoch": 0.4114546119070276, "grad_norm": 2.170959234237671, "learning_rate": 1.3291974935646554e-05, "loss": 0.1808995485305786, "step": 15620 }, { "epoch": 0.41171802715152633, "grad_norm": 1.0363433361053467, "learning_rate": 1.3283917895375018e-05, "loss": 0.17055089473724366, "step": 15630 }, { "epoch": 0.4119814423960251, "grad_norm": 0.9719519019126892, "learning_rate": 1.327585846492636e-05, "loss": 0.16699440479278566, "step": 15640 }, { "epoch": 0.4122448576405238, "grad_norm": 3.742572069168091, "learning_rate": 1.3267796650166576e-05, "loss": 0.1831895589828491, "step": 15650 }, { "epoch": 0.41250827288502256, "grad_norm": 0.8444315195083618, "learning_rate": 1.3259732456963406e-05, "loss": 0.18436520099639891, "step": 15660 }, { "epoch": 0.41277168812952125, "grad_norm": 2.6703362464904785, "learning_rate": 1.3251665891186313e-05, "loss": 0.1727433681488037, "step": 15670 }, { "epoch": 0.41303510337402, "grad_norm": 1.9420878887176514, "learning_rate": 1.32435969587065e-05, "loss": 0.17243406772613526, "step": 15680 }, { "epoch": 0.41329851861851874, "grad_norm": 1.5737868547439575, "learning_rate": 1.323552566539688e-05, "loss": 0.17960892915725707, "step": 15690 }, { "epoch": 0.4135619338630175, "grad_norm": 2.604276418685913, "learning_rate": 1.3227452017132085e-05, "loss": 0.17744343280792235, "step": 15700 }, { "epoch": 0.4138253491075162, "grad_norm": 3.5401930809020996, "learning_rate": 1.321937601978847e-05, "loss": 0.16366046667099, "step": 15710 }, { "epoch": 0.41408876435201497, "grad_norm": 1.7713778018951416, "learning_rate": 1.3211297679244093e-05, "loss": 0.17841200828552245, "step": 15720 }, { "epoch": 0.4143521795965137, "grad_norm": 2.1572771072387695, "learning_rate": 1.3203217001378716e-05, "loss": 0.18505197763442993, "step": 15730 }, { "epoch": 0.41461559484101246, "grad_norm": 2.246473789215088, "learning_rate": 1.3195133992073808e-05, "loss": 0.1685664176940918, "step": 15740 }, { "epoch": 0.41487901008551115, "grad_norm": 3.724618911743164, "learning_rate": 1.318704865721253e-05, "loss": 0.17617878913879395, "step": 15750 }, { "epoch": 0.4151424253300099, "grad_norm": 2.0997121334075928, "learning_rate": 1.3178961002679738e-05, "loss": 0.17665802240371703, "step": 15760 }, { "epoch": 0.41540584057450863, "grad_norm": 2.547701835632324, "learning_rate": 1.317087103436198e-05, "loss": 0.1889127016067505, "step": 15770 }, { "epoch": 0.4156692558190074, "grad_norm": 1.5773721933364868, "learning_rate": 1.3162778758147477e-05, "loss": 0.16751161813735962, "step": 15780 }, { "epoch": 0.4159326710635061, "grad_norm": 1.8050707578659058, "learning_rate": 1.315468417992614e-05, "loss": 0.19035847187042237, "step": 15790 }, { "epoch": 0.41619608630800486, "grad_norm": 1.3298825025558472, "learning_rate": 1.3146587305589552e-05, "loss": 0.17597843408584596, "step": 15800 }, { "epoch": 0.4164595015525036, "grad_norm": 1.1634563207626343, "learning_rate": 1.3138488141030967e-05, "loss": 0.16891517639160156, "step": 15810 }, { "epoch": 0.41672291679700235, "grad_norm": 3.381333589553833, "learning_rate": 1.313038669214531e-05, "loss": 0.17447613477706908, "step": 15820 }, { "epoch": 0.4169863320415011, "grad_norm": 2.0099146366119385, "learning_rate": 1.312228296482916e-05, "loss": 0.1793687343597412, "step": 15830 }, { "epoch": 0.4172497472859998, "grad_norm": 2.2660443782806396, "learning_rate": 1.311417696498076e-05, "loss": 0.18052488565444946, "step": 15840 }, { "epoch": 0.4175131625304985, "grad_norm": 1.7564001083374023, "learning_rate": 1.310606869850001e-05, "loss": 0.16297950744628906, "step": 15850 }, { "epoch": 0.41777657777499727, "grad_norm": 1.0540050268173218, "learning_rate": 1.3097958171288453e-05, "loss": 0.16991889476776123, "step": 15860 }, { "epoch": 0.418039993019496, "grad_norm": 5.179036617279053, "learning_rate": 1.3089845389249282e-05, "loss": 0.1687081575393677, "step": 15870 }, { "epoch": 0.41830340826399476, "grad_norm": 1.4456008672714233, "learning_rate": 1.308173035828733e-05, "loss": 0.18423328399658204, "step": 15880 }, { "epoch": 0.4185668235084935, "grad_norm": 1.7608599662780762, "learning_rate": 1.3073613084309065e-05, "loss": 0.18013331890106202, "step": 15890 }, { "epoch": 0.41883023875299225, "grad_norm": 3.533249616622925, "learning_rate": 1.3065493573222594e-05, "loss": 0.1761803150177002, "step": 15900 }, { "epoch": 0.419093653997491, "grad_norm": 1.5565742254257202, "learning_rate": 1.305737183093764e-05, "loss": 0.1658438563346863, "step": 15910 }, { "epoch": 0.41935706924198973, "grad_norm": 2.263533592224121, "learning_rate": 1.3049247863365562e-05, "loss": 0.16708149909973144, "step": 15920 }, { "epoch": 0.4196204844864884, "grad_norm": 1.7524269819259644, "learning_rate": 1.3041121676419333e-05, "loss": 0.17091131210327148, "step": 15930 }, { "epoch": 0.41988389973098716, "grad_norm": 1.4046229124069214, "learning_rate": 1.3032993276013546e-05, "loss": 0.17280259132385253, "step": 15940 }, { "epoch": 0.4201473149754859, "grad_norm": 0.9606296420097351, "learning_rate": 1.3024862668064393e-05, "loss": 0.16766839027404784, "step": 15950 }, { "epoch": 0.42041073021998465, "grad_norm": 1.1102269887924194, "learning_rate": 1.3016729858489689e-05, "loss": 0.17936705350875853, "step": 15960 }, { "epoch": 0.4206741454644834, "grad_norm": 1.456336259841919, "learning_rate": 1.300859485320884e-05, "loss": 0.18170176744461058, "step": 15970 }, { "epoch": 0.42093756070898214, "grad_norm": 2.972379684448242, "learning_rate": 1.3000457658142858e-05, "loss": 0.18027610778808595, "step": 15980 }, { "epoch": 0.4212009759534809, "grad_norm": 2.4431588649749756, "learning_rate": 1.299231827921434e-05, "loss": 0.17614065408706664, "step": 15990 }, { "epoch": 0.4214643911979796, "grad_norm": 1.8348748683929443, "learning_rate": 1.2984176722347478e-05, "loss": 0.1704084038734436, "step": 16000 }, { "epoch": 0.4217278064424783, "grad_norm": 1.1147788763046265, "learning_rate": 1.2976032993468049e-05, "loss": 0.17823464870452882, "step": 16010 }, { "epoch": 0.42199122168697706, "grad_norm": 3.512554168701172, "learning_rate": 1.296788709850341e-05, "loss": 0.1602123737335205, "step": 16020 }, { "epoch": 0.4222546369314758, "grad_norm": 1.8064013719558716, "learning_rate": 1.29597390433825e-05, "loss": 0.15364668369293213, "step": 16030 }, { "epoch": 0.42251805217597455, "grad_norm": 1.7493150234222412, "learning_rate": 1.2951588834035813e-05, "loss": 0.1742260217666626, "step": 16040 }, { "epoch": 0.4227814674204733, "grad_norm": 1.1211100816726685, "learning_rate": 1.2943436476395435e-05, "loss": 0.18010447025299073, "step": 16050 }, { "epoch": 0.42304488266497203, "grad_norm": 0.6941617727279663, "learning_rate": 1.2935281976394998e-05, "loss": 0.19311790466308593, "step": 16060 }, { "epoch": 0.4233082979094708, "grad_norm": 1.8690435886383057, "learning_rate": 1.2927125339969701e-05, "loss": 0.16906516551971434, "step": 16070 }, { "epoch": 0.4235717131539695, "grad_norm": 3.270442008972168, "learning_rate": 1.2918966573056291e-05, "loss": 0.17732049226760865, "step": 16080 }, { "epoch": 0.42383512839846826, "grad_norm": 2.5508739948272705, "learning_rate": 1.2910805681593078e-05, "loss": 0.18751633167266846, "step": 16090 }, { "epoch": 0.42409854364296695, "grad_norm": 1.5888500213623047, "learning_rate": 1.2902642671519903e-05, "loss": 0.1678393602371216, "step": 16100 }, { "epoch": 0.4243619588874657, "grad_norm": 1.474325180053711, "learning_rate": 1.2894477548778164e-05, "loss": 0.18304729461669922, "step": 16110 }, { "epoch": 0.42462537413196444, "grad_norm": 0.9732087850570679, "learning_rate": 1.2886310319310787e-05, "loss": 0.1767248749732971, "step": 16120 }, { "epoch": 0.4248887893764632, "grad_norm": 2.3543267250061035, "learning_rate": 1.2878140989062233e-05, "loss": 0.18512485027313233, "step": 16130 }, { "epoch": 0.4251522046209619, "grad_norm": 2.4081456661224365, "learning_rate": 1.2869969563978493e-05, "loss": 0.17778847217559815, "step": 16140 }, { "epoch": 0.42541561986546067, "grad_norm": 1.5264918804168701, "learning_rate": 1.2861796050007082e-05, "loss": 0.16484813690185546, "step": 16150 }, { "epoch": 0.4256790351099594, "grad_norm": 2.533240795135498, "learning_rate": 1.2853620453097039e-05, "loss": 0.19985523223876953, "step": 16160 }, { "epoch": 0.42594245035445816, "grad_norm": 1.10588800907135, "learning_rate": 1.2845442779198916e-05, "loss": 0.17163480520248414, "step": 16170 }, { "epoch": 0.4262058655989569, "grad_norm": 1.0058865547180176, "learning_rate": 1.2837263034264773e-05, "loss": 0.16450486183166504, "step": 16180 }, { "epoch": 0.4264692808434556, "grad_norm": 0.9439525008201599, "learning_rate": 1.2829081224248183e-05, "loss": 0.18478550910949706, "step": 16190 }, { "epoch": 0.42673269608795433, "grad_norm": 2.4113669395446777, "learning_rate": 1.2820897355104224e-05, "loss": 0.18301570415496826, "step": 16200 }, { "epoch": 0.4269961113324531, "grad_norm": 1.0258780717849731, "learning_rate": 1.2812711432789464e-05, "loss": 0.17347068786621095, "step": 16210 }, { "epoch": 0.4272595265769518, "grad_norm": 2.64677095413208, "learning_rate": 1.2804523463261975e-05, "loss": 0.15404300689697265, "step": 16220 }, { "epoch": 0.42752294182145056, "grad_norm": 2.080031156539917, "learning_rate": 1.2796333452481309e-05, "loss": 0.17253546714782714, "step": 16230 }, { "epoch": 0.4277863570659493, "grad_norm": 2.7219085693359375, "learning_rate": 1.2788141406408516e-05, "loss": 0.1783612608909607, "step": 16240 }, { "epoch": 0.42804977231044805, "grad_norm": 1.1443654298782349, "learning_rate": 1.2779947331006113e-05, "loss": 0.16576497554779052, "step": 16250 }, { "epoch": 0.4283131875549468, "grad_norm": 1.3989720344543457, "learning_rate": 1.2771751232238107e-05, "loss": 0.16110901832580565, "step": 16260 }, { "epoch": 0.42857660279944554, "grad_norm": 4.9421281814575195, "learning_rate": 1.2763553116069969e-05, "loss": 0.16851506233215333, "step": 16270 }, { "epoch": 0.4288400180439442, "grad_norm": 1.0107544660568237, "learning_rate": 1.2755352988468646e-05, "loss": 0.17021539211273193, "step": 16280 }, { "epoch": 0.42910343328844297, "grad_norm": 2.173802137374878, "learning_rate": 1.2747150855402541e-05, "loss": 0.18430426120758056, "step": 16290 }, { "epoch": 0.4293668485329417, "grad_norm": 1.5801364183425903, "learning_rate": 1.2738946722841519e-05, "loss": 0.16526556015014648, "step": 16300 }, { "epoch": 0.42963026377744046, "grad_norm": 3.9050779342651367, "learning_rate": 1.2730740596756908e-05, "loss": 0.1513571858406067, "step": 16310 }, { "epoch": 0.4298936790219392, "grad_norm": 3.6321957111358643, "learning_rate": 1.2722532483121475e-05, "loss": 0.17885863780975342, "step": 16320 }, { "epoch": 0.43015709426643794, "grad_norm": 1.668014645576477, "learning_rate": 1.271432238790944e-05, "loss": 0.1690671920776367, "step": 16330 }, { "epoch": 0.4304205095109367, "grad_norm": 2.418766975402832, "learning_rate": 1.2706110317096463e-05, "loss": 0.1712494134902954, "step": 16340 }, { "epoch": 0.43068392475543543, "grad_norm": 2.8945138454437256, "learning_rate": 1.269789627665965e-05, "loss": 0.18294241428375244, "step": 16350 }, { "epoch": 0.4309473399999341, "grad_norm": 1.708100438117981, "learning_rate": 1.2689680272577527e-05, "loss": 0.18818670511245728, "step": 16360 }, { "epoch": 0.43121075524443286, "grad_norm": 2.4441869258880615, "learning_rate": 1.2681462310830061e-05, "loss": 0.17612918615341186, "step": 16370 }, { "epoch": 0.4314741704889316, "grad_norm": 3.556774616241455, "learning_rate": 1.2673242397398635e-05, "loss": 0.18572568893432617, "step": 16380 }, { "epoch": 0.43173758573343035, "grad_norm": 1.1726183891296387, "learning_rate": 1.266502053826606e-05, "loss": 0.1798325300216675, "step": 16390 }, { "epoch": 0.4320010009779291, "grad_norm": 3.6370487213134766, "learning_rate": 1.2656796739416557e-05, "loss": 0.18072874546051027, "step": 16400 }, { "epoch": 0.43226441622242784, "grad_norm": 1.4401739835739136, "learning_rate": 1.2648571006835764e-05, "loss": 0.15924620628356934, "step": 16410 }, { "epoch": 0.4325278314669266, "grad_norm": 1.4976667165756226, "learning_rate": 1.2640343346510724e-05, "loss": 0.17050776481628419, "step": 16420 }, { "epoch": 0.4327912467114253, "grad_norm": 0.8370130658149719, "learning_rate": 1.2632113764429884e-05, "loss": 0.1833980441093445, "step": 16430 }, { "epoch": 0.43305466195592407, "grad_norm": 1.4675699472427368, "learning_rate": 1.2623882266583083e-05, "loss": 0.17943886518478394, "step": 16440 }, { "epoch": 0.43331807720042276, "grad_norm": 1.5372447967529297, "learning_rate": 1.2615648858961564e-05, "loss": 0.16915289163589478, "step": 16450 }, { "epoch": 0.4335814924449215, "grad_norm": 2.8903403282165527, "learning_rate": 1.260741354755796e-05, "loss": 0.1758727788925171, "step": 16460 }, { "epoch": 0.43384490768942024, "grad_norm": 1.0774588584899902, "learning_rate": 1.2599176338366282e-05, "loss": 0.1841222643852234, "step": 16470 }, { "epoch": 0.434108322933919, "grad_norm": 3.1631760597229004, "learning_rate": 1.2590937237381927e-05, "loss": 0.16499829292297363, "step": 16480 }, { "epoch": 0.43437173817841773, "grad_norm": 1.4107781648635864, "learning_rate": 1.2582696250601667e-05, "loss": 0.186468505859375, "step": 16490 }, { "epoch": 0.4346351534229165, "grad_norm": 2.4524004459381104, "learning_rate": 1.257445338402365e-05, "loss": 0.18214039802551268, "step": 16500 }, { "epoch": 0.4348985686674152, "grad_norm": 1.3370707035064697, "learning_rate": 1.256620864364739e-05, "loss": 0.18013166189193724, "step": 16510 }, { "epoch": 0.43516198391191396, "grad_norm": 1.5073109865188599, "learning_rate": 1.255796203547376e-05, "loss": 0.16915924549102784, "step": 16520 }, { "epoch": 0.4354253991564127, "grad_norm": 1.0785716772079468, "learning_rate": 1.2549713565505001e-05, "loss": 0.1686413049697876, "step": 16530 }, { "epoch": 0.4356888144009114, "grad_norm": 1.0610026121139526, "learning_rate": 1.254146323974471e-05, "loss": 0.16775450706481934, "step": 16540 }, { "epoch": 0.43595222964541014, "grad_norm": 1.9317487478256226, "learning_rate": 1.253321106419782e-05, "loss": 0.16121233701705934, "step": 16550 }, { "epoch": 0.4362156448899089, "grad_norm": 1.016005516052246, "learning_rate": 1.2524957044870627e-05, "loss": 0.1695927858352661, "step": 16560 }, { "epoch": 0.4364790601344076, "grad_norm": 1.3103923797607422, "learning_rate": 1.2516701187770761e-05, "loss": 0.1718219757080078, "step": 16570 }, { "epoch": 0.43674247537890637, "grad_norm": 2.381049156188965, "learning_rate": 1.250844349890719e-05, "loss": 0.18044748306274414, "step": 16580 }, { "epoch": 0.4370058906234051, "grad_norm": 1.4223213195800781, "learning_rate": 1.250018398429022e-05, "loss": 0.1804347038269043, "step": 16590 }, { "epoch": 0.43726930586790386, "grad_norm": 4.271573066711426, "learning_rate": 1.2491922649931475e-05, "loss": 0.19651103019714355, "step": 16600 }, { "epoch": 0.4375327211124026, "grad_norm": 4.011427402496338, "learning_rate": 1.2483659501843916e-05, "loss": 0.17990036010742189, "step": 16610 }, { "epoch": 0.4377961363569013, "grad_norm": 1.3552829027175903, "learning_rate": 1.2475394546041822e-05, "loss": 0.16138395071029663, "step": 16620 }, { "epoch": 0.43805955160140003, "grad_norm": 1.9853801727294922, "learning_rate": 1.2467127788540774e-05, "loss": 0.16224949359893798, "step": 16630 }, { "epoch": 0.4383229668458988, "grad_norm": 1.2822707891464233, "learning_rate": 1.245885923535768e-05, "loss": 0.1794661521911621, "step": 16640 }, { "epoch": 0.4385863820903975, "grad_norm": 0.9368155598640442, "learning_rate": 1.2450588892510756e-05, "loss": 0.17643799781799316, "step": 16650 }, { "epoch": 0.43884979733489626, "grad_norm": 1.4444326162338257, "learning_rate": 1.2442316766019502e-05, "loss": 0.1814626693725586, "step": 16660 }, { "epoch": 0.439113212579395, "grad_norm": 2.2621562480926514, "learning_rate": 1.2434042861904743e-05, "loss": 0.1696681261062622, "step": 16670 }, { "epoch": 0.43937662782389375, "grad_norm": 1.552234411239624, "learning_rate": 1.2425767186188569e-05, "loss": 0.1698589324951172, "step": 16680 }, { "epoch": 0.4396400430683925, "grad_norm": 2.6902406215667725, "learning_rate": 1.2417489744894381e-05, "loss": 0.1726980209350586, "step": 16690 }, { "epoch": 0.43990345831289124, "grad_norm": 2.370231866836548, "learning_rate": 1.2409210544046859e-05, "loss": 0.16114946603775024, "step": 16700 }, { "epoch": 0.4401668735573899, "grad_norm": 2.1338255405426025, "learning_rate": 1.2400929589671962e-05, "loss": 0.16408376693725585, "step": 16710 }, { "epoch": 0.44043028880188867, "grad_norm": 4.218188762664795, "learning_rate": 1.2392646887796924e-05, "loss": 0.19405421018600463, "step": 16720 }, { "epoch": 0.4406937040463874, "grad_norm": 1.0984797477722168, "learning_rate": 1.2384362444450253e-05, "loss": 0.16893577575683594, "step": 16730 }, { "epoch": 0.44095711929088616, "grad_norm": 1.125999927520752, "learning_rate": 1.2376076265661722e-05, "loss": 0.19346282482147217, "step": 16740 }, { "epoch": 0.4412205345353849, "grad_norm": 2.212229013442993, "learning_rate": 1.2367788357462375e-05, "loss": 0.1810660481452942, "step": 16750 }, { "epoch": 0.44148394977988364, "grad_norm": 2.9323976039886475, "learning_rate": 1.2359498725884509e-05, "loss": 0.17537519931793213, "step": 16760 }, { "epoch": 0.4417473650243824, "grad_norm": 2.556318759918213, "learning_rate": 1.2351207376961666e-05, "loss": 0.19232641458511351, "step": 16770 }, { "epoch": 0.44201078026888113, "grad_norm": 2.3096072673797607, "learning_rate": 1.234291431672866e-05, "loss": 0.17845994234085083, "step": 16780 }, { "epoch": 0.4422741955133799, "grad_norm": 1.9657198190689087, "learning_rate": 1.2334619551221531e-05, "loss": 0.16929484605789186, "step": 16790 }, { "epoch": 0.44253761075787856, "grad_norm": 2.7697994709014893, "learning_rate": 1.232632308647757e-05, "loss": 0.17874534130096437, "step": 16800 }, { "epoch": 0.4428010260023773, "grad_norm": 2.6637470722198486, "learning_rate": 1.2318024928535296e-05, "loss": 0.16966047286987304, "step": 16810 }, { "epoch": 0.44306444124687605, "grad_norm": 2.256373167037964, "learning_rate": 1.2309725083434477e-05, "loss": 0.16958181858062743, "step": 16820 }, { "epoch": 0.4433278564913748, "grad_norm": 0.9429588913917542, "learning_rate": 1.2301423557216088e-05, "loss": 0.19172945022583007, "step": 16830 }, { "epoch": 0.44359127173587354, "grad_norm": 2.2404279708862305, "learning_rate": 1.2293120355922346e-05, "loss": 0.1683651924133301, "step": 16840 }, { "epoch": 0.4438546869803723, "grad_norm": 1.5654133558273315, "learning_rate": 1.2284815485596671e-05, "loss": 0.15848469734191895, "step": 16850 }, { "epoch": 0.444118102224871, "grad_norm": 2.293668031692505, "learning_rate": 1.2276508952283712e-05, "loss": 0.18374359607696533, "step": 16860 }, { "epoch": 0.44438151746936977, "grad_norm": 3.2795705795288086, "learning_rate": 1.2268200762029319e-05, "loss": 0.16693612337112426, "step": 16870 }, { "epoch": 0.44464493271386846, "grad_norm": 1.495031476020813, "learning_rate": 1.2259890920880555e-05, "loss": 0.16917208433151246, "step": 16880 }, { "epoch": 0.4449083479583672, "grad_norm": 2.909221887588501, "learning_rate": 1.2251579434885679e-05, "loss": 0.16882060766220092, "step": 16890 }, { "epoch": 0.44517176320286594, "grad_norm": 1.0349332094192505, "learning_rate": 1.2243266310094146e-05, "loss": 0.16456301212310792, "step": 16900 }, { "epoch": 0.4454351784473647, "grad_norm": 2.2167959213256836, "learning_rate": 1.2234951552556613e-05, "loss": 0.17186362743377687, "step": 16910 }, { "epoch": 0.44569859369186343, "grad_norm": 1.9844732284545898, "learning_rate": 1.2226635168324916e-05, "loss": 0.17274624109268188, "step": 16920 }, { "epoch": 0.4459620089363622, "grad_norm": 1.778106451034546, "learning_rate": 1.2218317163452078e-05, "loss": 0.16244553327560424, "step": 16930 }, { "epoch": 0.4462254241808609, "grad_norm": 3.4970436096191406, "learning_rate": 1.2209997543992301e-05, "loss": 0.16630055904388427, "step": 16940 }, { "epoch": 0.44648883942535966, "grad_norm": 4.101444721221924, "learning_rate": 1.2201676316000964e-05, "loss": 0.18151865005493165, "step": 16950 }, { "epoch": 0.4467522546698584, "grad_norm": 1.891074299812317, "learning_rate": 1.2193353485534614e-05, "loss": 0.17179088592529296, "step": 16960 }, { "epoch": 0.4470156699143571, "grad_norm": 0.9945513606071472, "learning_rate": 1.2185029058650972e-05, "loss": 0.167551851272583, "step": 16970 }, { "epoch": 0.44727908515885584, "grad_norm": 1.9937174320220947, "learning_rate": 1.2176703041408908e-05, "loss": 0.16636965274810792, "step": 16980 }, { "epoch": 0.4475425004033546, "grad_norm": 3.4567031860351562, "learning_rate": 1.2168375439868464e-05, "loss": 0.17585277557373047, "step": 16990 }, { "epoch": 0.4478059156478533, "grad_norm": 1.563748836517334, "learning_rate": 1.2160046260090816e-05, "loss": 0.1599268674850464, "step": 17000 }, { "epoch": 0.44806933089235207, "grad_norm": 3.175241231918335, "learning_rate": 1.2151715508138313e-05, "loss": 0.16587672233581544, "step": 17010 }, { "epoch": 0.4483327461368508, "grad_norm": 2.664095163345337, "learning_rate": 1.2143383190074429e-05, "loss": 0.17715824842453004, "step": 17020 }, { "epoch": 0.44859616138134956, "grad_norm": 2.573857069015503, "learning_rate": 1.2135049311963787e-05, "loss": 0.18066428899765014, "step": 17030 }, { "epoch": 0.4488595766258483, "grad_norm": 1.201318383216858, "learning_rate": 1.2126713879872138e-05, "loss": 0.18058578968048095, "step": 17040 }, { "epoch": 0.44912299187034704, "grad_norm": 1.5761966705322266, "learning_rate": 1.2118376899866375e-05, "loss": 0.1884979486465454, "step": 17050 }, { "epoch": 0.44938640711484573, "grad_norm": 4.646324634552002, "learning_rate": 1.2110038378014512e-05, "loss": 0.18036531209945678, "step": 17060 }, { "epoch": 0.4496498223593445, "grad_norm": 1.0772100687026978, "learning_rate": 1.210169832038568e-05, "loss": 0.16615655422210693, "step": 17070 }, { "epoch": 0.4499132376038432, "grad_norm": 1.407227873802185, "learning_rate": 1.2093356733050137e-05, "loss": 0.1657041549682617, "step": 17080 }, { "epoch": 0.45017665284834196, "grad_norm": 3.235029458999634, "learning_rate": 1.2085013622079248e-05, "loss": 0.17936851978302001, "step": 17090 }, { "epoch": 0.4504400680928407, "grad_norm": 2.1410105228424072, "learning_rate": 1.2076668993545491e-05, "loss": 0.16344854831695557, "step": 17100 }, { "epoch": 0.45070348333733945, "grad_norm": 4.1087188720703125, "learning_rate": 1.206832285352244e-05, "loss": 0.19326183795928956, "step": 17110 }, { "epoch": 0.4509668985818382, "grad_norm": 1.4342926740646362, "learning_rate": 1.2059975208084787e-05, "loss": 0.17636239528656006, "step": 17120 }, { "epoch": 0.45123031382633694, "grad_norm": 3.751789093017578, "learning_rate": 1.2051626063308299e-05, "loss": 0.17080938816070557, "step": 17130 }, { "epoch": 0.4514937290708356, "grad_norm": 0.9515185356140137, "learning_rate": 1.2043275425269854e-05, "loss": 0.1645756483078003, "step": 17140 }, { "epoch": 0.45175714431533437, "grad_norm": 2.315664768218994, "learning_rate": 1.2034923300047391e-05, "loss": 0.17313122749328613, "step": 17150 }, { "epoch": 0.4520205595598331, "grad_norm": 3.181027412414551, "learning_rate": 1.2026569693719963e-05, "loss": 0.16262632608413696, "step": 17160 }, { "epoch": 0.45228397480433186, "grad_norm": 2.15622878074646, "learning_rate": 1.2018214612367673e-05, "loss": 0.16536409854888917, "step": 17170 }, { "epoch": 0.4525473900488306, "grad_norm": 1.0520271062850952, "learning_rate": 1.200985806207172e-05, "loss": 0.155819833278656, "step": 17180 }, { "epoch": 0.45281080529332934, "grad_norm": 2.4900741577148438, "learning_rate": 1.2001500048914354e-05, "loss": 0.1544504404067993, "step": 17190 }, { "epoch": 0.4530742205378281, "grad_norm": 1.6877870559692383, "learning_rate": 1.1993140578978898e-05, "loss": 0.17245056629180908, "step": 17200 }, { "epoch": 0.45333763578232683, "grad_norm": 1.2987040281295776, "learning_rate": 1.1984779658349743e-05, "loss": 0.17125205993652343, "step": 17210 }, { "epoch": 0.4536010510268256, "grad_norm": 2.4988114833831787, "learning_rate": 1.1976417293112318e-05, "loss": 0.16405344009399414, "step": 17220 }, { "epoch": 0.45386446627132426, "grad_norm": 1.0598678588867188, "learning_rate": 1.1968053489353122e-05, "loss": 0.1630624771118164, "step": 17230 }, { "epoch": 0.454127881515823, "grad_norm": 2.6588857173919678, "learning_rate": 1.195968825315968e-05, "loss": 0.1777212142944336, "step": 17240 }, { "epoch": 0.45439129676032175, "grad_norm": 2.2645463943481445, "learning_rate": 1.1951321590620583e-05, "loss": 0.1860855460166931, "step": 17250 }, { "epoch": 0.4546547120048205, "grad_norm": 2.334970474243164, "learning_rate": 1.1942953507825446e-05, "loss": 0.18194512128829957, "step": 17260 }, { "epoch": 0.45491812724931924, "grad_norm": 1.4316296577453613, "learning_rate": 1.1934584010864921e-05, "loss": 0.17910406589508057, "step": 17270 }, { "epoch": 0.455181542493818, "grad_norm": 0.8623569011688232, "learning_rate": 1.1926213105830683e-05, "loss": 0.17750251293182373, "step": 17280 }, { "epoch": 0.4554449577383167, "grad_norm": 1.4482276439666748, "learning_rate": 1.1917840798815448e-05, "loss": 0.18472671508789062, "step": 17290 }, { "epoch": 0.45570837298281547, "grad_norm": 1.4941611289978027, "learning_rate": 1.1909467095912932e-05, "loss": 0.19500732421875, "step": 17300 }, { "epoch": 0.4559717882273142, "grad_norm": 2.5638697147369385, "learning_rate": 1.1901092003217885e-05, "loss": 0.1712160587310791, "step": 17310 }, { "epoch": 0.4562352034718129, "grad_norm": 2.1947438716888428, "learning_rate": 1.1892715526826058e-05, "loss": 0.18300278186798097, "step": 17320 }, { "epoch": 0.45649861871631164, "grad_norm": 0.9254606366157532, "learning_rate": 1.1884337672834212e-05, "loss": 0.16643478870391845, "step": 17330 }, { "epoch": 0.4567620339608104, "grad_norm": 1.8243801593780518, "learning_rate": 1.1875958447340108e-05, "loss": 0.16046006679534913, "step": 17340 }, { "epoch": 0.45702544920530913, "grad_norm": 3.931354522705078, "learning_rate": 1.1867577856442514e-05, "loss": 0.1756762981414795, "step": 17350 }, { "epoch": 0.4572888644498079, "grad_norm": 2.3167972564697266, "learning_rate": 1.1859195906241182e-05, "loss": 0.1822901964187622, "step": 17360 }, { "epoch": 0.4575522796943066, "grad_norm": 5.004260540008545, "learning_rate": 1.1850812602836851e-05, "loss": 0.18014273643493653, "step": 17370 }, { "epoch": 0.45781569493880536, "grad_norm": 1.1897482872009277, "learning_rate": 1.1842427952331263e-05, "loss": 0.17011475563049316, "step": 17380 }, { "epoch": 0.4580791101833041, "grad_norm": 2.391775369644165, "learning_rate": 1.183404196082712e-05, "loss": 0.17878652811050416, "step": 17390 }, { "epoch": 0.4583425254278028, "grad_norm": 2.38238263130188, "learning_rate": 1.1825654634428111e-05, "loss": 0.1720033884048462, "step": 17400 }, { "epoch": 0.45860594067230154, "grad_norm": 2.889423131942749, "learning_rate": 1.1817265979238892e-05, "loss": 0.17693554162979125, "step": 17410 }, { "epoch": 0.4588693559168003, "grad_norm": 0.7400386333465576, "learning_rate": 1.1808876001365093e-05, "loss": 0.1634977340698242, "step": 17420 }, { "epoch": 0.459132771161299, "grad_norm": 3.7518365383148193, "learning_rate": 1.1800484706913294e-05, "loss": 0.17240018844604493, "step": 17430 }, { "epoch": 0.45939618640579777, "grad_norm": 2.562833786010742, "learning_rate": 1.1792092101991052e-05, "loss": 0.16232995986938475, "step": 17440 }, { "epoch": 0.4596596016502965, "grad_norm": 5.337149143218994, "learning_rate": 1.1783698192706856e-05, "loss": 0.17456971406936644, "step": 17450 }, { "epoch": 0.45992301689479526, "grad_norm": 2.473876476287842, "learning_rate": 1.1775302985170161e-05, "loss": 0.17244868278503417, "step": 17460 }, { "epoch": 0.460186432139294, "grad_norm": 1.9070401191711426, "learning_rate": 1.1766906485491361e-05, "loss": 0.188873291015625, "step": 17470 }, { "epoch": 0.46044984738379274, "grad_norm": 1.0019125938415527, "learning_rate": 1.175850869978179e-05, "loss": 0.18105510473251343, "step": 17480 }, { "epoch": 0.46071326262829143, "grad_norm": 1.3276439905166626, "learning_rate": 1.1750109634153719e-05, "loss": 0.15637423992156982, "step": 17490 }, { "epoch": 0.4609766778727902, "grad_norm": 3.0039944648742676, "learning_rate": 1.1741709294720346e-05, "loss": 0.16535418033599852, "step": 17500 }, { "epoch": 0.4612400931172889, "grad_norm": 2.530869722366333, "learning_rate": 1.173330768759581e-05, "loss": 0.16933715343475342, "step": 17510 }, { "epoch": 0.46150350836178766, "grad_norm": 2.1491808891296387, "learning_rate": 1.1724904818895158e-05, "loss": 0.16472806930541992, "step": 17520 }, { "epoch": 0.4617669236062864, "grad_norm": 1.3991883993148804, "learning_rate": 1.1716500694734359e-05, "loss": 0.1801194429397583, "step": 17530 }, { "epoch": 0.46203033885078515, "grad_norm": 1.9796322584152222, "learning_rate": 1.1708095321230295e-05, "loss": 0.17325681447982788, "step": 17540 }, { "epoch": 0.4622937540952839, "grad_norm": 2.3675343990325928, "learning_rate": 1.1699688704500765e-05, "loss": 0.16170481443405152, "step": 17550 }, { "epoch": 0.46255716933978264, "grad_norm": 2.58241605758667, "learning_rate": 1.1691280850664463e-05, "loss": 0.17277448177337645, "step": 17560 }, { "epoch": 0.4628205845842814, "grad_norm": 2.0703125, "learning_rate": 1.1682871765840996e-05, "loss": 0.17619426250457765, "step": 17570 }, { "epoch": 0.46308399982878007, "grad_norm": 1.774964451789856, "learning_rate": 1.1674461456150845e-05, "loss": 0.16868579387664795, "step": 17580 }, { "epoch": 0.4633474150732788, "grad_norm": 3.06599760055542, "learning_rate": 1.1666049927715408e-05, "loss": 0.17147539854049682, "step": 17590 }, { "epoch": 0.46361083031777756, "grad_norm": 2.0206282138824463, "learning_rate": 1.1657637186656952e-05, "loss": 0.15704272985458373, "step": 17600 }, { "epoch": 0.4638742455622763, "grad_norm": 1.4176549911499023, "learning_rate": 1.1649223239098639e-05, "loss": 0.17982927560806275, "step": 17610 }, { "epoch": 0.46413766080677504, "grad_norm": 3.1910312175750732, "learning_rate": 1.1640808091164495e-05, "loss": 0.17129920721054076, "step": 17620 }, { "epoch": 0.4644010760512738, "grad_norm": 2.3978958129882812, "learning_rate": 1.1632391748979435e-05, "loss": 0.1581118106842041, "step": 17630 }, { "epoch": 0.46466449129577253, "grad_norm": 1.1643688678741455, "learning_rate": 1.162397421866923e-05, "loss": 0.16801415681838988, "step": 17640 }, { "epoch": 0.4649279065402713, "grad_norm": 3.043912887573242, "learning_rate": 1.1615555506360526e-05, "loss": 0.17273762226104736, "step": 17650 }, { "epoch": 0.46519132178476996, "grad_norm": 1.819854974746704, "learning_rate": 1.1607135618180823e-05, "loss": 0.18016564846038818, "step": 17660 }, { "epoch": 0.4654547370292687, "grad_norm": 2.5527946949005127, "learning_rate": 1.1598714560258477e-05, "loss": 0.17369697093963624, "step": 17670 }, { "epoch": 0.46571815227376745, "grad_norm": 1.9508273601531982, "learning_rate": 1.1590292338722702e-05, "loss": 0.1724379062652588, "step": 17680 }, { "epoch": 0.4659815675182662, "grad_norm": 1.6284246444702148, "learning_rate": 1.1581868959703553e-05, "loss": 0.1633872628211975, "step": 17690 }, { "epoch": 0.46624498276276494, "grad_norm": 4.408462047576904, "learning_rate": 1.1573444429331926e-05, "loss": 0.1634888768196106, "step": 17700 }, { "epoch": 0.4665083980072637, "grad_norm": 1.9799416065216064, "learning_rate": 1.1565018753739559e-05, "loss": 0.1703490972518921, "step": 17710 }, { "epoch": 0.4667718132517624, "grad_norm": 1.7077350616455078, "learning_rate": 1.1556591939059024e-05, "loss": 0.16397016048431395, "step": 17720 }, { "epoch": 0.46703522849626117, "grad_norm": 4.204740047454834, "learning_rate": 1.1548163991423716e-05, "loss": 0.16114832162857057, "step": 17730 }, { "epoch": 0.4672986437407599, "grad_norm": 1.5503315925598145, "learning_rate": 1.1539734916967864e-05, "loss": 0.16675748825073242, "step": 17740 }, { "epoch": 0.4675620589852586, "grad_norm": 1.4906221628189087, "learning_rate": 1.1531304721826506e-05, "loss": 0.1596430778503418, "step": 17750 }, { "epoch": 0.46782547422975734, "grad_norm": 2.8029532432556152, "learning_rate": 1.1522873412135509e-05, "loss": 0.16370638608932495, "step": 17760 }, { "epoch": 0.4680888894742561, "grad_norm": 1.5315933227539062, "learning_rate": 1.1514440994031536e-05, "loss": 0.15183804035186768, "step": 17770 }, { "epoch": 0.46835230471875483, "grad_norm": 1.3731112480163574, "learning_rate": 1.1506007473652075e-05, "loss": 0.16984047889709472, "step": 17780 }, { "epoch": 0.4686157199632536, "grad_norm": 4.030615329742432, "learning_rate": 1.1497572857135397e-05, "loss": 0.16272578239440919, "step": 17790 }, { "epoch": 0.4688791352077523, "grad_norm": 3.0680184364318848, "learning_rate": 1.1489137150620582e-05, "loss": 0.1765734910964966, "step": 17800 }, { "epoch": 0.46914255045225106, "grad_norm": 0.8224918842315674, "learning_rate": 1.1480700360247504e-05, "loss": 0.17049150466918944, "step": 17810 }, { "epoch": 0.4694059656967498, "grad_norm": 2.8141911029815674, "learning_rate": 1.147226249215682e-05, "loss": 0.17514309883117676, "step": 17820 }, { "epoch": 0.46966938094124855, "grad_norm": 1.289320468902588, "learning_rate": 1.1463823552489977e-05, "loss": 0.17950448989868165, "step": 17830 }, { "epoch": 0.46993279618574724, "grad_norm": 2.43760347366333, "learning_rate": 1.1455383547389193e-05, "loss": 0.16424791812896727, "step": 17840 }, { "epoch": 0.470196211430246, "grad_norm": 1.7965645790100098, "learning_rate": 1.1446942482997479e-05, "loss": 0.1510935068130493, "step": 17850 }, { "epoch": 0.4704596266747447, "grad_norm": 1.8660502433776855, "learning_rate": 1.1438500365458593e-05, "loss": 0.1588861107826233, "step": 17860 }, { "epoch": 0.47072304191924347, "grad_norm": 2.574497699737549, "learning_rate": 1.1430057200917084e-05, "loss": 0.1706599473953247, "step": 17870 }, { "epoch": 0.4709864571637422, "grad_norm": 1.1563524007797241, "learning_rate": 1.1421612995518243e-05, "loss": 0.1570691704750061, "step": 17880 }, { "epoch": 0.47124987240824096, "grad_norm": 2.150265693664551, "learning_rate": 1.1413167755408132e-05, "loss": 0.17709863185882568, "step": 17890 }, { "epoch": 0.4715132876527397, "grad_norm": 1.6722248792648315, "learning_rate": 1.1404721486733558e-05, "loss": 0.17439029216766358, "step": 17900 }, { "epoch": 0.47177670289723844, "grad_norm": 1.4922417402267456, "learning_rate": 1.1396274195642085e-05, "loss": 0.15833659172058107, "step": 17910 }, { "epoch": 0.47204011814173713, "grad_norm": 2.0732932090759277, "learning_rate": 1.1387825888282006e-05, "loss": 0.17785029411315917, "step": 17920 }, { "epoch": 0.4723035333862359, "grad_norm": 1.7560993432998657, "learning_rate": 1.1379376570802373e-05, "loss": 0.1617656707763672, "step": 17930 }, { "epoch": 0.4725669486307346, "grad_norm": 1.0080510377883911, "learning_rate": 1.137092624935296e-05, "loss": 0.16439650058746338, "step": 17940 }, { "epoch": 0.47283036387523336, "grad_norm": 5.2435221672058105, "learning_rate": 1.1362474930084278e-05, "loss": 0.17855020761489868, "step": 17950 }, { "epoch": 0.4730937791197321, "grad_norm": 1.2297052145004272, "learning_rate": 1.1354022619147557e-05, "loss": 0.15585048198699952, "step": 17960 }, { "epoch": 0.47335719436423085, "grad_norm": 1.8875024318695068, "learning_rate": 1.1345569322694756e-05, "loss": 0.17434155941009521, "step": 17970 }, { "epoch": 0.4736206096087296, "grad_norm": 3.2239134311676025, "learning_rate": 1.133711504687855e-05, "loss": 0.178518545627594, "step": 17980 }, { "epoch": 0.47388402485322834, "grad_norm": 1.8271642923355103, "learning_rate": 1.1328659797852325e-05, "loss": 0.16610476970672608, "step": 17990 }, { "epoch": 0.4741474400977271, "grad_norm": 2.228762149810791, "learning_rate": 1.1320203581770177e-05, "loss": 0.17353860139846802, "step": 18000 }, { "epoch": 0.47441085534222577, "grad_norm": 4.319690227508545, "learning_rate": 1.1311746404786903e-05, "loss": 0.17955597639083862, "step": 18010 }, { "epoch": 0.4746742705867245, "grad_norm": 2.261308431625366, "learning_rate": 1.1303288273058002e-05, "loss": 0.15451536178588868, "step": 18020 }, { "epoch": 0.47493768583122326, "grad_norm": 1.3954415321350098, "learning_rate": 1.1294829192739665e-05, "loss": 0.1672551989555359, "step": 18030 }, { "epoch": 0.475201101075722, "grad_norm": 0.984168291091919, "learning_rate": 1.1286369169988782e-05, "loss": 0.17263622283935548, "step": 18040 }, { "epoch": 0.47546451632022074, "grad_norm": 1.990071177482605, "learning_rate": 1.1277908210962916e-05, "loss": 0.17867755889892578, "step": 18050 }, { "epoch": 0.4757279315647195, "grad_norm": 2.7845802307128906, "learning_rate": 1.126944632182032e-05, "loss": 0.16408121585845947, "step": 18060 }, { "epoch": 0.47599134680921823, "grad_norm": 1.4715521335601807, "learning_rate": 1.1260983508719922e-05, "loss": 0.15089802742004393, "step": 18070 }, { "epoch": 0.476254762053717, "grad_norm": 2.908473491668701, "learning_rate": 1.1252519777821324e-05, "loss": 0.1633727431297302, "step": 18080 }, { "epoch": 0.4765181772982157, "grad_norm": 3.212637186050415, "learning_rate": 1.1244055135284795e-05, "loss": 0.16429075002670288, "step": 18090 }, { "epoch": 0.4767815925427144, "grad_norm": 2.6969199180603027, "learning_rate": 1.123558958727126e-05, "loss": 0.17045462131500244, "step": 18100 }, { "epoch": 0.47704500778721315, "grad_norm": 2.030350923538208, "learning_rate": 1.1227123139942322e-05, "loss": 0.15777034759521485, "step": 18110 }, { "epoch": 0.4773084230317119, "grad_norm": 1.5075547695159912, "learning_rate": 1.1218655799460216e-05, "loss": 0.1699126958847046, "step": 18120 }, { "epoch": 0.47757183827621064, "grad_norm": 1.4019031524658203, "learning_rate": 1.1210187571987845e-05, "loss": 0.16378824710845946, "step": 18130 }, { "epoch": 0.4778352535207094, "grad_norm": 3.3193492889404297, "learning_rate": 1.1201718463688745e-05, "loss": 0.16708862781524658, "step": 18140 }, { "epoch": 0.4780986687652081, "grad_norm": 2.102113723754883, "learning_rate": 1.1193248480727101e-05, "loss": 0.17133784294128418, "step": 18150 }, { "epoch": 0.47836208400970687, "grad_norm": 2.6963260173797607, "learning_rate": 1.118477762926773e-05, "loss": 0.15720348358154296, "step": 18160 }, { "epoch": 0.4786254992542056, "grad_norm": 1.7083075046539307, "learning_rate": 1.117630591547609e-05, "loss": 0.16592352390289306, "step": 18170 }, { "epoch": 0.4788889144987043, "grad_norm": 1.7021559476852417, "learning_rate": 1.1167833345518248e-05, "loss": 0.15155948400497438, "step": 18180 }, { "epoch": 0.47915232974320304, "grad_norm": 2.5987744331359863, "learning_rate": 1.1159359925560918e-05, "loss": 0.17036659717559816, "step": 18190 }, { "epoch": 0.4794157449877018, "grad_norm": 3.059680223464966, "learning_rate": 1.115088566177141e-05, "loss": 0.16443588733673095, "step": 18200 }, { "epoch": 0.47967916023220053, "grad_norm": 1.7719395160675049, "learning_rate": 1.1142410560317669e-05, "loss": 0.16137309074401857, "step": 18210 }, { "epoch": 0.4799425754766993, "grad_norm": 1.8316943645477295, "learning_rate": 1.113393462736823e-05, "loss": 0.1919570803642273, "step": 18220 }, { "epoch": 0.480205990721198, "grad_norm": 1.2686058282852173, "learning_rate": 1.1125457869092251e-05, "loss": 0.17120450735092163, "step": 18230 }, { "epoch": 0.48046940596569676, "grad_norm": 1.4392706155776978, "learning_rate": 1.1116980291659474e-05, "loss": 0.16766961812973022, "step": 18240 }, { "epoch": 0.4807328212101955, "grad_norm": 2.5325472354888916, "learning_rate": 1.1108501901240254e-05, "loss": 0.15856101512908935, "step": 18250 }, { "epoch": 0.48099623645469425, "grad_norm": 1.8310080766677856, "learning_rate": 1.1100022704005524e-05, "loss": 0.16487313508987428, "step": 18260 }, { "epoch": 0.48125965169919294, "grad_norm": 3.3295795917510986, "learning_rate": 1.109154270612681e-05, "loss": 0.16408485174179077, "step": 18270 }, { "epoch": 0.4815230669436917, "grad_norm": 1.0047740936279297, "learning_rate": 1.1083061913776224e-05, "loss": 0.16017093658447265, "step": 18280 }, { "epoch": 0.4817864821881904, "grad_norm": 2.380215644836426, "learning_rate": 1.1074580333126448e-05, "loss": 0.16753706932067872, "step": 18290 }, { "epoch": 0.48204989743268917, "grad_norm": 1.4238752126693726, "learning_rate": 1.1066097970350747e-05, "loss": 0.1632701873779297, "step": 18300 }, { "epoch": 0.4823133126771879, "grad_norm": 1.7516393661499023, "learning_rate": 1.1057614831622942e-05, "loss": 0.1547827959060669, "step": 18310 }, { "epoch": 0.48257672792168665, "grad_norm": 0.8351757526397705, "learning_rate": 1.1049130923117437e-05, "loss": 0.15018483400344848, "step": 18320 }, { "epoch": 0.4828401431661854, "grad_norm": 0.9272449612617493, "learning_rate": 1.1040646251009181e-05, "loss": 0.1782886028289795, "step": 18330 }, { "epoch": 0.48310355841068414, "grad_norm": 2.0272021293640137, "learning_rate": 1.1032160821473692e-05, "loss": 0.16949005126953126, "step": 18340 }, { "epoch": 0.4833669736551829, "grad_norm": 3.624986410140991, "learning_rate": 1.1023674640687019e-05, "loss": 0.17087329626083375, "step": 18350 }, { "epoch": 0.4836303888996816, "grad_norm": 0.8397311568260193, "learning_rate": 1.1015187714825782e-05, "loss": 0.16269418001174926, "step": 18360 }, { "epoch": 0.4838938041441803, "grad_norm": 1.0159715414047241, "learning_rate": 1.1006700050067122e-05, "loss": 0.14036070108413695, "step": 18370 }, { "epoch": 0.48415721938867906, "grad_norm": 2.231295347213745, "learning_rate": 1.0998211652588737e-05, "loss": 0.15333800315856932, "step": 18380 }, { "epoch": 0.4844206346331778, "grad_norm": 1.4776235818862915, "learning_rate": 1.0989722528568845e-05, "loss": 0.1690879225730896, "step": 18390 }, { "epoch": 0.48468404987767655, "grad_norm": 2.156628131866455, "learning_rate": 1.0981232684186191e-05, "loss": 0.16790368556976318, "step": 18400 }, { "epoch": 0.4849474651221753, "grad_norm": 2.979234218597412, "learning_rate": 1.0972742125620062e-05, "loss": 0.1562070369720459, "step": 18410 }, { "epoch": 0.48521088036667404, "grad_norm": 2.1461682319641113, "learning_rate": 1.0964250859050245e-05, "loss": 0.15951411724090575, "step": 18420 }, { "epoch": 0.4854742956111728, "grad_norm": 1.9177145957946777, "learning_rate": 1.0955758890657051e-05, "loss": 0.15301513671875, "step": 18430 }, { "epoch": 0.48573771085567147, "grad_norm": 3.39852237701416, "learning_rate": 1.0947266226621298e-05, "loss": 0.14523826837539672, "step": 18440 }, { "epoch": 0.4860011261001702, "grad_norm": 1.1638059616088867, "learning_rate": 1.0938772873124321e-05, "loss": 0.16901602745056152, "step": 18450 }, { "epoch": 0.48626454134466895, "grad_norm": 1.3023028373718262, "learning_rate": 1.0930278836347943e-05, "loss": 0.15315905809402466, "step": 18460 }, { "epoch": 0.4865279565891677, "grad_norm": 1.0247039794921875, "learning_rate": 1.0921784122474495e-05, "loss": 0.15926752090454102, "step": 18470 }, { "epoch": 0.48679137183366644, "grad_norm": 1.8837745189666748, "learning_rate": 1.0913288737686789e-05, "loss": 0.16808794736862182, "step": 18480 }, { "epoch": 0.4870547870781652, "grad_norm": 1.8987613916397095, "learning_rate": 1.090479268816814e-05, "loss": 0.1656394600868225, "step": 18490 }, { "epoch": 0.48731820232266393, "grad_norm": 2.6601665019989014, "learning_rate": 1.0896295980102334e-05, "loss": 0.17076656818389893, "step": 18500 }, { "epoch": 0.4875816175671627, "grad_norm": 1.1414077281951904, "learning_rate": 1.0887798619673648e-05, "loss": 0.1471010208129883, "step": 18510 }, { "epoch": 0.4878450328116614, "grad_norm": 0.9577065706253052, "learning_rate": 1.087930061306682e-05, "loss": 0.16271796226501464, "step": 18520 }, { "epoch": 0.4881084480561601, "grad_norm": 0.8618561029434204, "learning_rate": 1.0870801966467073e-05, "loss": 0.16390035152435303, "step": 18530 }, { "epoch": 0.48837186330065885, "grad_norm": 1.3517752885818481, "learning_rate": 1.0862302686060082e-05, "loss": 0.1471879243850708, "step": 18540 }, { "epoch": 0.4886352785451576, "grad_norm": 2.2548279762268066, "learning_rate": 1.0853802778031995e-05, "loss": 0.16306289434432983, "step": 18550 }, { "epoch": 0.48889869378965634, "grad_norm": 3.130863904953003, "learning_rate": 1.0845302248569411e-05, "loss": 0.15726678371429442, "step": 18560 }, { "epoch": 0.4891621090341551, "grad_norm": 1.8232866525650024, "learning_rate": 1.0836801103859377e-05, "loss": 0.15999088287353516, "step": 18570 }, { "epoch": 0.4894255242786538, "grad_norm": 1.6189179420471191, "learning_rate": 1.0828299350089401e-05, "loss": 0.15101827383041383, "step": 18580 }, { "epoch": 0.48968893952315257, "grad_norm": 2.0921084880828857, "learning_rate": 1.0819796993447422e-05, "loss": 0.16534013748168946, "step": 18590 }, { "epoch": 0.4899523547676513, "grad_norm": 2.295651435852051, "learning_rate": 1.081129404012182e-05, "loss": 0.160641086101532, "step": 18600 }, { "epoch": 0.49021577001215005, "grad_norm": 1.3907279968261719, "learning_rate": 1.0802790496301414e-05, "loss": 0.1693619966506958, "step": 18610 }, { "epoch": 0.49047918525664874, "grad_norm": 2.336686372756958, "learning_rate": 1.079428636817545e-05, "loss": 0.16612231731414795, "step": 18620 }, { "epoch": 0.4907426005011475, "grad_norm": 0.9481478929519653, "learning_rate": 1.0785781661933596e-05, "loss": 0.1526033878326416, "step": 18630 }, { "epoch": 0.49100601574564623, "grad_norm": 0.9416671395301819, "learning_rate": 1.0777276383765953e-05, "loss": 0.15722726583480834, "step": 18640 }, { "epoch": 0.491269430990145, "grad_norm": 2.4606711864471436, "learning_rate": 1.0768770539863017e-05, "loss": 0.16634793281555177, "step": 18650 }, { "epoch": 0.4915328462346437, "grad_norm": 0.9393767714500427, "learning_rate": 1.076026413641572e-05, "loss": 0.15816742181777954, "step": 18660 }, { "epoch": 0.49179626147914246, "grad_norm": 1.9817348718643188, "learning_rate": 1.0751757179615378e-05, "loss": 0.18321602344512938, "step": 18670 }, { "epoch": 0.4920596767236412, "grad_norm": 2.989551544189453, "learning_rate": 1.0743249675653733e-05, "loss": 0.16822476387023927, "step": 18680 }, { "epoch": 0.49232309196813995, "grad_norm": 1.7050141096115112, "learning_rate": 1.0734741630722904e-05, "loss": 0.16722307205200196, "step": 18690 }, { "epoch": 0.49258650721263864, "grad_norm": 1.4252686500549316, "learning_rate": 1.0726233051015415e-05, "loss": 0.1591216206550598, "step": 18700 }, { "epoch": 0.4928499224571374, "grad_norm": 1.571723222732544, "learning_rate": 1.0717723942724182e-05, "loss": 0.14825305938720704, "step": 18710 }, { "epoch": 0.4931133377016361, "grad_norm": 2.301736831665039, "learning_rate": 1.07092143120425e-05, "loss": 0.1538274645805359, "step": 18720 }, { "epoch": 0.49337675294613487, "grad_norm": 3.9426586627960205, "learning_rate": 1.0700704165164042e-05, "loss": 0.15886027812957765, "step": 18730 }, { "epoch": 0.4936401681906336, "grad_norm": 2.4484472274780273, "learning_rate": 1.069219350828286e-05, "loss": 0.16306500434875487, "step": 18740 }, { "epoch": 0.49390358343513235, "grad_norm": 1.1851714849472046, "learning_rate": 1.068368234759338e-05, "loss": 0.15390665531158448, "step": 18750 }, { "epoch": 0.4941669986796311, "grad_norm": 2.0625929832458496, "learning_rate": 1.0675170689290393e-05, "loss": 0.1625291347503662, "step": 18760 }, { "epoch": 0.49443041392412984, "grad_norm": 2.5479252338409424, "learning_rate": 1.0666658539569048e-05, "loss": 0.1622879385948181, "step": 18770 }, { "epoch": 0.4946938291686286, "grad_norm": 2.756211042404175, "learning_rate": 1.0658145904624852e-05, "loss": 0.17172422409057617, "step": 18780 }, { "epoch": 0.4949572444131273, "grad_norm": 1.768528699874878, "learning_rate": 1.0649632790653677e-05, "loss": 0.16264033317565918, "step": 18790 }, { "epoch": 0.495220659657626, "grad_norm": 1.0048705339431763, "learning_rate": 1.0641119203851724e-05, "loss": 0.1664189577102661, "step": 18800 }, { "epoch": 0.49548407490212476, "grad_norm": 1.3492763042449951, "learning_rate": 1.0632605150415561e-05, "loss": 0.1730271100997925, "step": 18810 }, { "epoch": 0.4957474901466235, "grad_norm": 1.4835103750228882, "learning_rate": 1.062409063654207e-05, "loss": 0.17560381889343263, "step": 18820 }, { "epoch": 0.49601090539112225, "grad_norm": 5.150135517120361, "learning_rate": 1.0615575668428494e-05, "loss": 0.1567671537399292, "step": 18830 }, { "epoch": 0.496274320635621, "grad_norm": 2.2721569538116455, "learning_rate": 1.0607060252272383e-05, "loss": 0.16531093120574952, "step": 18840 }, { "epoch": 0.49653773588011973, "grad_norm": 1.9671077728271484, "learning_rate": 1.0598544394271633e-05, "loss": 0.1449050188064575, "step": 18850 }, { "epoch": 0.4968011511246185, "grad_norm": 1.2638661861419678, "learning_rate": 1.0590028100624448e-05, "loss": 0.1539365291595459, "step": 18860 }, { "epoch": 0.4970645663691172, "grad_norm": 3.6724140644073486, "learning_rate": 1.0581511377529354e-05, "loss": 0.16910886764526367, "step": 18870 }, { "epoch": 0.4973279816136159, "grad_norm": 1.3582401275634766, "learning_rate": 1.0572994231185194e-05, "loss": 0.17650125026702881, "step": 18880 }, { "epoch": 0.49759139685811465, "grad_norm": 1.5483627319335938, "learning_rate": 1.056447666779111e-05, "loss": 0.16517012119293212, "step": 18890 }, { "epoch": 0.4978548121026134, "grad_norm": 1.0217875242233276, "learning_rate": 1.0555958693546552e-05, "loss": 0.15541415214538573, "step": 18900 }, { "epoch": 0.49811822734711214, "grad_norm": 1.5329941511154175, "learning_rate": 1.0547440314651274e-05, "loss": 0.14388564825057984, "step": 18910 }, { "epoch": 0.4983816425916109, "grad_norm": 2.0067222118377686, "learning_rate": 1.0538921537305315e-05, "loss": 0.17595664262771607, "step": 18920 }, { "epoch": 0.49864505783610963, "grad_norm": 3.0256104469299316, "learning_rate": 1.0530402367709007e-05, "loss": 0.16048583984375, "step": 18930 }, { "epoch": 0.4989084730806084, "grad_norm": 2.7758586406707764, "learning_rate": 1.0521882812062978e-05, "loss": 0.16005315780639648, "step": 18940 }, { "epoch": 0.4991718883251071, "grad_norm": 1.5875416994094849, "learning_rate": 1.0513362876568118e-05, "loss": 0.1579769253730774, "step": 18950 }, { "epoch": 0.49943530356960586, "grad_norm": 2.983616352081299, "learning_rate": 1.0504842567425611e-05, "loss": 0.16579537391662597, "step": 18960 }, { "epoch": 0.49969871881410455, "grad_norm": 1.4959874153137207, "learning_rate": 1.0496321890836901e-05, "loss": 0.15809946060180663, "step": 18970 }, { "epoch": 0.4999621340586033, "grad_norm": 1.654444694519043, "learning_rate": 1.0487800853003706e-05, "loss": 0.16071486473083496, "step": 18980 }, { "epoch": 0.500225549303102, "grad_norm": 2.603365659713745, "learning_rate": 1.0479279460128005e-05, "loss": 0.16158571243286132, "step": 18990 }, { "epoch": 0.5004889645476008, "grad_norm": 1.702472448348999, "learning_rate": 1.0470757718412035e-05, "loss": 0.15206140279769897, "step": 19000 }, { "epoch": 0.5007523797920995, "grad_norm": 1.7067499160766602, "learning_rate": 1.0462235634058287e-05, "loss": 0.16008846759796141, "step": 19010 }, { "epoch": 0.5010157950365982, "grad_norm": 1.572457194328308, "learning_rate": 1.0453713213269503e-05, "loss": 0.16136465072631836, "step": 19020 }, { "epoch": 0.501279210281097, "grad_norm": 2.2606773376464844, "learning_rate": 1.0445190462248667e-05, "loss": 0.1571868896484375, "step": 19030 }, { "epoch": 0.5015426255255957, "grad_norm": 2.3187332153320312, "learning_rate": 1.0436667387199004e-05, "loss": 0.16190389394760132, "step": 19040 }, { "epoch": 0.5018060407700945, "grad_norm": 1.686772108078003, "learning_rate": 1.0428143994323978e-05, "loss": 0.1653367519378662, "step": 19050 }, { "epoch": 0.5020694560145932, "grad_norm": 1.8152234554290771, "learning_rate": 1.041962028982728e-05, "loss": 0.16057556867599487, "step": 19060 }, { "epoch": 0.502332871259092, "grad_norm": 3.8370888233184814, "learning_rate": 1.0411096279912835e-05, "loss": 0.17386534214019775, "step": 19070 }, { "epoch": 0.5025962865035907, "grad_norm": 1.2691700458526611, "learning_rate": 1.0402571970784778e-05, "loss": 0.16171514987945557, "step": 19080 }, { "epoch": 0.5028597017480895, "grad_norm": 1.5934542417526245, "learning_rate": 1.0394047368647473e-05, "loss": 0.1495189666748047, "step": 19090 }, { "epoch": 0.5031231169925882, "grad_norm": 2.0588016510009766, "learning_rate": 1.0385522479705493e-05, "loss": 0.149137544631958, "step": 19100 }, { "epoch": 0.5033865322370868, "grad_norm": 0.9324582815170288, "learning_rate": 1.0376997310163622e-05, "loss": 0.17044850587844848, "step": 19110 }, { "epoch": 0.5036499474815856, "grad_norm": 1.3783502578735352, "learning_rate": 1.0368471866226837e-05, "loss": 0.1519399642944336, "step": 19120 }, { "epoch": 0.5039133627260843, "grad_norm": 4.247899055480957, "learning_rate": 1.0359946154100336e-05, "loss": 0.16030112504959107, "step": 19130 }, { "epoch": 0.5041767779705831, "grad_norm": 1.8646626472473145, "learning_rate": 1.035142017998949e-05, "loss": 0.16097164154052734, "step": 19140 }, { "epoch": 0.5044401932150818, "grad_norm": 1.534378170967102, "learning_rate": 1.0342893950099879e-05, "loss": 0.16760338544845582, "step": 19150 }, { "epoch": 0.5047036084595806, "grad_norm": 1.6652705669403076, "learning_rate": 1.0334367470637257e-05, "loss": 0.15588891506195068, "step": 19160 }, { "epoch": 0.5049670237040793, "grad_norm": 1.1501758098602295, "learning_rate": 1.0325840747807562e-05, "loss": 0.1538923740386963, "step": 19170 }, { "epoch": 0.505230438948578, "grad_norm": 1.4627457857131958, "learning_rate": 1.0317313787816915e-05, "loss": 0.1664123773574829, "step": 19180 }, { "epoch": 0.5054938541930768, "grad_norm": 2.2418692111968994, "learning_rate": 1.0308786596871606e-05, "loss": 0.16999258995056152, "step": 19190 }, { "epoch": 0.5057572694375755, "grad_norm": 1.2957733869552612, "learning_rate": 1.0300259181178086e-05, "loss": 0.16487257480621337, "step": 19200 }, { "epoch": 0.5060206846820743, "grad_norm": 0.868187665939331, "learning_rate": 1.0291731546942977e-05, "loss": 0.1495111107826233, "step": 19210 }, { "epoch": 0.506284099926573, "grad_norm": 1.7198151350021362, "learning_rate": 1.0283203700373065e-05, "loss": 0.1547337770462036, "step": 19220 }, { "epoch": 0.5065475151710718, "grad_norm": 1.045255422592163, "learning_rate": 1.0274675647675281e-05, "loss": 0.1690202236175537, "step": 19230 }, { "epoch": 0.5068109304155705, "grad_norm": 3.0218398571014404, "learning_rate": 1.0266147395056712e-05, "loss": 0.16071462631225586, "step": 19240 }, { "epoch": 0.5070743456600693, "grad_norm": 2.7811684608459473, "learning_rate": 1.0257618948724583e-05, "loss": 0.1637493848800659, "step": 19250 }, { "epoch": 0.507337760904568, "grad_norm": 1.1074501276016235, "learning_rate": 1.024909031488627e-05, "loss": 0.16678531169891359, "step": 19260 }, { "epoch": 0.5076011761490666, "grad_norm": 2.107475757598877, "learning_rate": 1.0240561499749278e-05, "loss": 0.14738453626632692, "step": 19270 }, { "epoch": 0.5078645913935654, "grad_norm": 1.4920716285705566, "learning_rate": 1.023203250952125e-05, "loss": 0.1708441734313965, "step": 19280 }, { "epoch": 0.5081280066380641, "grad_norm": 1.4919390678405762, "learning_rate": 1.0223503350409949e-05, "loss": 0.15932146310806275, "step": 19290 }, { "epoch": 0.5083914218825629, "grad_norm": 1.4463746547698975, "learning_rate": 1.0214974028623268e-05, "loss": 0.16626032590866088, "step": 19300 }, { "epoch": 0.5086548371270616, "grad_norm": 1.8780683279037476, "learning_rate": 1.0206444550369212e-05, "loss": 0.17246003150939943, "step": 19310 }, { "epoch": 0.5089182523715604, "grad_norm": 2.134315252304077, "learning_rate": 1.0197914921855909e-05, "loss": 0.1557589054107666, "step": 19320 }, { "epoch": 0.5091816676160591, "grad_norm": 1.5609729290008545, "learning_rate": 1.0189385149291586e-05, "loss": 0.16406428813934326, "step": 19330 }, { "epoch": 0.5094450828605579, "grad_norm": 1.4931503534317017, "learning_rate": 1.018085523888458e-05, "loss": 0.1594549298286438, "step": 19340 }, { "epoch": 0.5097084981050566, "grad_norm": 3.7944178581237793, "learning_rate": 1.0172325196843328e-05, "loss": 0.1560646414756775, "step": 19350 }, { "epoch": 0.5099719133495553, "grad_norm": 1.4877095222473145, "learning_rate": 1.0163795029376362e-05, "loss": 0.14433122873306276, "step": 19360 }, { "epoch": 0.5102353285940541, "grad_norm": 1.8624098300933838, "learning_rate": 1.015526474269231e-05, "loss": 0.15480337142944336, "step": 19370 }, { "epoch": 0.5104987438385528, "grad_norm": 2.4697771072387695, "learning_rate": 1.0146734342999875e-05, "loss": 0.14906381368637084, "step": 19380 }, { "epoch": 0.5107621590830516, "grad_norm": 2.283022165298462, "learning_rate": 1.0138203836507859e-05, "loss": 0.17514058351516723, "step": 19390 }, { "epoch": 0.5110255743275502, "grad_norm": 1.7878395318984985, "learning_rate": 1.0129673229425125e-05, "loss": 0.16465848684310913, "step": 19400 }, { "epoch": 0.511288989572049, "grad_norm": 3.135988473892212, "learning_rate": 1.0121142527960624e-05, "loss": 0.15382955074310303, "step": 19410 }, { "epoch": 0.5115524048165477, "grad_norm": 2.3224587440490723, "learning_rate": 1.0112611738323362e-05, "loss": 0.1489412546157837, "step": 19420 }, { "epoch": 0.5118158200610465, "grad_norm": 2.83717679977417, "learning_rate": 1.0104080866722419e-05, "loss": 0.1537434220314026, "step": 19430 }, { "epoch": 0.5120792353055452, "grad_norm": 2.520373582839966, "learning_rate": 1.0095549919366928e-05, "loss": 0.1504810333251953, "step": 19440 }, { "epoch": 0.5123426505500439, "grad_norm": 2.7386133670806885, "learning_rate": 1.0087018902466086e-05, "loss": 0.1713522791862488, "step": 19450 }, { "epoch": 0.5126060657945427, "grad_norm": 3.719709634780884, "learning_rate": 1.0078487822229132e-05, "loss": 0.16978754997253417, "step": 19460 }, { "epoch": 0.5128694810390414, "grad_norm": 1.455356240272522, "learning_rate": 1.0069956684865354e-05, "loss": 0.16453614234924316, "step": 19470 }, { "epoch": 0.5131328962835402, "grad_norm": 2.1439857482910156, "learning_rate": 1.0061425496584084e-05, "loss": 0.15665242671966553, "step": 19480 }, { "epoch": 0.5133963115280389, "grad_norm": 3.038930654525757, "learning_rate": 1.0052894263594687e-05, "loss": 0.14242053031921387, "step": 19490 }, { "epoch": 0.5136597267725377, "grad_norm": 2.1399946212768555, "learning_rate": 1.0044362992106563e-05, "loss": 0.15103678703308104, "step": 19500 }, { "epoch": 0.5139231420170364, "grad_norm": 1.2852637767791748, "learning_rate": 1.0035831688329139e-05, "loss": 0.1605614423751831, "step": 19510 }, { "epoch": 0.5141865572615352, "grad_norm": 1.4252841472625732, "learning_rate": 1.0027300358471865e-05, "loss": 0.1616081953048706, "step": 19520 }, { "epoch": 0.5144499725060339, "grad_norm": 3.0312442779541016, "learning_rate": 1.0018769008744214e-05, "loss": 0.1675998568534851, "step": 19530 }, { "epoch": 0.5147133877505325, "grad_norm": 1.5208508968353271, "learning_rate": 1.0010237645355669e-05, "loss": 0.1601786732673645, "step": 19540 }, { "epoch": 0.5149768029950313, "grad_norm": 1.7555261850357056, "learning_rate": 1.0001706274515724e-05, "loss": 0.1687162399291992, "step": 19550 }, { "epoch": 0.51524021823953, "grad_norm": 2.690321922302246, "learning_rate": 9.993174902433877e-06, "loss": 0.1552799701690674, "step": 19560 }, { "epoch": 0.5155036334840288, "grad_norm": 1.2675676345825195, "learning_rate": 9.984643535319632e-06, "loss": 0.1626521587371826, "step": 19570 }, { "epoch": 0.5157670487285275, "grad_norm": 3.7708988189697266, "learning_rate": 9.976112179382483e-06, "loss": 0.1740155339241028, "step": 19580 }, { "epoch": 0.5160304639730263, "grad_norm": 1.3950282335281372, "learning_rate": 9.967580840831922e-06, "loss": 0.1718793988227844, "step": 19590 }, { "epoch": 0.516293879217525, "grad_norm": 1.8306978940963745, "learning_rate": 9.959049525877425e-06, "loss": 0.16579385995864868, "step": 19600 }, { "epoch": 0.5165572944620238, "grad_norm": 2.294154405593872, "learning_rate": 9.950518240728446e-06, "loss": 0.16998569965362548, "step": 19610 }, { "epoch": 0.5168207097065225, "grad_norm": 1.1418389081954956, "learning_rate": 9.94198699159443e-06, "loss": 0.1528834581375122, "step": 19620 }, { "epoch": 0.5170841249510212, "grad_norm": 1.4513050317764282, "learning_rate": 9.93345578468478e-06, "loss": 0.1645357847213745, "step": 19630 }, { "epoch": 0.51734754019552, "grad_norm": 1.9365788698196411, "learning_rate": 9.924924626208884e-06, "loss": 0.15034772157669068, "step": 19640 }, { "epoch": 0.5176109554400187, "grad_norm": 1.7406482696533203, "learning_rate": 9.916393522376087e-06, "loss": 0.16265277862548827, "step": 19650 }, { "epoch": 0.5178743706845175, "grad_norm": 2.9840688705444336, "learning_rate": 9.907862479395684e-06, "loss": 0.1624795079231262, "step": 19660 }, { "epoch": 0.5181377859290162, "grad_norm": 1.8688379526138306, "learning_rate": 9.899331503476951e-06, "loss": 0.16685118675231933, "step": 19670 }, { "epoch": 0.518401201173515, "grad_norm": 1.442759394645691, "learning_rate": 9.890800600829086e-06, "loss": 0.14804475307464598, "step": 19680 }, { "epoch": 0.5186646164180136, "grad_norm": 2.9345364570617676, "learning_rate": 9.882269777661261e-06, "loss": 0.14970898628234863, "step": 19690 }, { "epoch": 0.5189280316625123, "grad_norm": 1.7612321376800537, "learning_rate": 9.873739040182569e-06, "loss": 0.15070275068283082, "step": 19700 }, { "epoch": 0.5191914469070111, "grad_norm": 1.4937562942504883, "learning_rate": 9.86520839460205e-06, "loss": 0.17168737649917604, "step": 19710 }, { "epoch": 0.5194548621515098, "grad_norm": 3.8432257175445557, "learning_rate": 9.856677847128678e-06, "loss": 0.15280687808990479, "step": 19720 }, { "epoch": 0.5197182773960086, "grad_norm": 1.0572052001953125, "learning_rate": 9.848147403971353e-06, "loss": 0.15694018602371215, "step": 19730 }, { "epoch": 0.5199816926405073, "grad_norm": 3.3467607498168945, "learning_rate": 9.839617071338897e-06, "loss": 0.16920461654663085, "step": 19740 }, { "epoch": 0.5202451078850061, "grad_norm": 2.0915443897247314, "learning_rate": 9.831086855440056e-06, "loss": 0.17744877338409423, "step": 19750 }, { "epoch": 0.5205085231295048, "grad_norm": 1.871233344078064, "learning_rate": 9.82255676248349e-06, "loss": 0.15066211223602294, "step": 19760 }, { "epoch": 0.5207719383740036, "grad_norm": 2.8749332427978516, "learning_rate": 9.814026798677767e-06, "loss": 0.15159156322479247, "step": 19770 }, { "epoch": 0.5210353536185023, "grad_norm": 1.05247962474823, "learning_rate": 9.805496970231367e-06, "loss": 0.1547696828842163, "step": 19780 }, { "epoch": 0.521298768863001, "grad_norm": 0.9934360980987549, "learning_rate": 9.796967283352657e-06, "loss": 0.14665076732635499, "step": 19790 }, { "epoch": 0.5215621841074998, "grad_norm": 2.0522007942199707, "learning_rate": 9.788437744249925e-06, "loss": 0.16359362602233887, "step": 19800 }, { "epoch": 0.5218255993519985, "grad_norm": 2.445829391479492, "learning_rate": 9.779908359131327e-06, "loss": 0.15340303182601928, "step": 19810 }, { "epoch": 0.5220890145964973, "grad_norm": 1.5580782890319824, "learning_rate": 9.77137913420492e-06, "loss": 0.14907510280609132, "step": 19820 }, { "epoch": 0.522352429840996, "grad_norm": 1.6351038217544556, "learning_rate": 9.762850075678644e-06, "loss": 0.16914464235305787, "step": 19830 }, { "epoch": 0.5226158450854947, "grad_norm": 2.5148603916168213, "learning_rate": 9.754321189760312e-06, "loss": 0.15362412929534913, "step": 19840 }, { "epoch": 0.5228792603299934, "grad_norm": 0.9569868445396423, "learning_rate": 9.74579248265762e-06, "loss": 0.1481872320175171, "step": 19850 }, { "epoch": 0.5231426755744922, "grad_norm": 1.0826976299285889, "learning_rate": 9.737263960578127e-06, "loss": 0.15733740329742432, "step": 19860 }, { "epoch": 0.5234060908189909, "grad_norm": 1.4061822891235352, "learning_rate": 9.728735629729256e-06, "loss": 0.16261996030807496, "step": 19870 }, { "epoch": 0.5236695060634896, "grad_norm": 2.308547019958496, "learning_rate": 9.7202074963183e-06, "loss": 0.15657432079315187, "step": 19880 }, { "epoch": 0.5239329213079884, "grad_norm": 1.685750961303711, "learning_rate": 9.711679566552396e-06, "loss": 0.16202130317687988, "step": 19890 }, { "epoch": 0.5241963365524871, "grad_norm": 1.8855832815170288, "learning_rate": 9.703151846638545e-06, "loss": 0.16209917068481444, "step": 19900 }, { "epoch": 0.5244597517969859, "grad_norm": 1.2373605966567993, "learning_rate": 9.694624342783591e-06, "loss": 0.16335772275924682, "step": 19910 }, { "epoch": 0.5247231670414846, "grad_norm": 2.023212194442749, "learning_rate": 9.686097061194209e-06, "loss": 0.17910908460617064, "step": 19920 }, { "epoch": 0.5249865822859834, "grad_norm": 2.4532663822174072, "learning_rate": 9.677570008076933e-06, "loss": 0.15959715843200684, "step": 19930 }, { "epoch": 0.5252499975304821, "grad_norm": 1.9255107641220093, "learning_rate": 9.669043189638115e-06, "loss": 0.15155636072158812, "step": 19940 }, { "epoch": 0.5255134127749809, "grad_norm": 1.43503737449646, "learning_rate": 9.66051661208394e-06, "loss": 0.17412989139556884, "step": 19950 }, { "epoch": 0.5257768280194796, "grad_norm": 2.1126246452331543, "learning_rate": 9.651990281620423e-06, "loss": 0.14905407428741455, "step": 19960 }, { "epoch": 0.5260402432639782, "grad_norm": 3.8426411151885986, "learning_rate": 9.643464204453392e-06, "loss": 0.14753823280334472, "step": 19970 }, { "epoch": 0.526303658508477, "grad_norm": 2.6530513763427734, "learning_rate": 9.634938386788497e-06, "loss": 0.1732232689857483, "step": 19980 }, { "epoch": 0.5265670737529757, "grad_norm": 2.258434534072876, "learning_rate": 9.626412834831193e-06, "loss": 0.16906592845916749, "step": 19990 }, { "epoch": 0.5268304889974745, "grad_norm": 2.9179916381835938, "learning_rate": 9.617887554786746e-06, "loss": 0.14801794290542603, "step": 20000 }, { "epoch": 0.5270939042419732, "grad_norm": 2.0544769763946533, "learning_rate": 9.609362552860225e-06, "loss": 0.14485849142074586, "step": 20010 }, { "epoch": 0.527357319486472, "grad_norm": 2.5241501331329346, "learning_rate": 9.600837835256486e-06, "loss": 0.1649012088775635, "step": 20020 }, { "epoch": 0.5276207347309707, "grad_norm": 2.024671792984009, "learning_rate": 9.592313408180199e-06, "loss": 0.13670549392700196, "step": 20030 }, { "epoch": 0.5278841499754695, "grad_norm": 5.415407180786133, "learning_rate": 9.583789277835798e-06, "loss": 0.17682486772537231, "step": 20040 }, { "epoch": 0.5281475652199682, "grad_norm": 1.3940690755844116, "learning_rate": 9.575265450427517e-06, "loss": 0.1676391363143921, "step": 20050 }, { "epoch": 0.5284109804644669, "grad_norm": 1.2612311840057373, "learning_rate": 9.566741932159369e-06, "loss": 0.167276132106781, "step": 20060 }, { "epoch": 0.5286743957089657, "grad_norm": 1.3548164367675781, "learning_rate": 9.558218729235132e-06, "loss": 0.1697009563446045, "step": 20070 }, { "epoch": 0.5289378109534644, "grad_norm": 4.319390773773193, "learning_rate": 9.549695847858363e-06, "loss": 0.16106482744216918, "step": 20080 }, { "epoch": 0.5292012261979632, "grad_norm": 2.099909782409668, "learning_rate": 9.541173294232387e-06, "loss": 0.1621546983718872, "step": 20090 }, { "epoch": 0.5294646414424619, "grad_norm": 2.1925108432769775, "learning_rate": 9.53265107456028e-06, "loss": 0.16282265186309813, "step": 20100 }, { "epoch": 0.5297280566869607, "grad_norm": 1.1918811798095703, "learning_rate": 9.524129195044886e-06, "loss": 0.14310731887817382, "step": 20110 }, { "epoch": 0.5299914719314593, "grad_norm": 2.817106246948242, "learning_rate": 9.515607661888799e-06, "loss": 0.1636163115501404, "step": 20120 }, { "epoch": 0.5302548871759581, "grad_norm": 2.2813611030578613, "learning_rate": 9.507086481294345e-06, "loss": 0.18055202960968017, "step": 20130 }, { "epoch": 0.5305183024204568, "grad_norm": 1.360292673110962, "learning_rate": 9.498565659463624e-06, "loss": 0.1539899706840515, "step": 20140 }, { "epoch": 0.5307817176649555, "grad_norm": 1.1594547033309937, "learning_rate": 9.490045202598442e-06, "loss": 0.15235869884490966, "step": 20150 }, { "epoch": 0.5310451329094543, "grad_norm": 1.3379945755004883, "learning_rate": 9.48152511690037e-06, "loss": 0.16299896240234374, "step": 20160 }, { "epoch": 0.531308548153953, "grad_norm": 2.9823219776153564, "learning_rate": 9.473005408570685e-06, "loss": 0.15238332748413086, "step": 20170 }, { "epoch": 0.5315719633984518, "grad_norm": 1.4813201427459717, "learning_rate": 9.464486083810397e-06, "loss": 0.14336390495300294, "step": 20180 }, { "epoch": 0.5318353786429505, "grad_norm": 1.6534258127212524, "learning_rate": 9.455967148820243e-06, "loss": 0.15623552799224855, "step": 20190 }, { "epoch": 0.5320987938874493, "grad_norm": 3.737717628479004, "learning_rate": 9.44744860980067e-06, "loss": 0.16474788188934325, "step": 20200 }, { "epoch": 0.532362209131948, "grad_norm": 1.1788315773010254, "learning_rate": 9.438930472951838e-06, "loss": 0.1453916072845459, "step": 20210 }, { "epoch": 0.5326256243764467, "grad_norm": 1.8627351522445679, "learning_rate": 9.430412744473615e-06, "loss": 0.15066165924072267, "step": 20220 }, { "epoch": 0.5328890396209455, "grad_norm": 2.3431472778320312, "learning_rate": 9.421895430565568e-06, "loss": 0.15793383121490479, "step": 20230 }, { "epoch": 0.5331524548654442, "grad_norm": 2.5435221195220947, "learning_rate": 9.413378537426972e-06, "loss": 0.15182168483734132, "step": 20240 }, { "epoch": 0.533415870109943, "grad_norm": 1.3810820579528809, "learning_rate": 9.404862071256785e-06, "loss": 0.1558550000190735, "step": 20250 }, { "epoch": 0.5336792853544416, "grad_norm": 2.6208207607269287, "learning_rate": 9.396346038253654e-06, "loss": 0.17657803297042846, "step": 20260 }, { "epoch": 0.5339427005989404, "grad_norm": 3.789059638977051, "learning_rate": 9.387830444615926e-06, "loss": 0.14784380197525024, "step": 20270 }, { "epoch": 0.5342061158434391, "grad_norm": 4.904918193817139, "learning_rate": 9.379315296541611e-06, "loss": 0.1475001096725464, "step": 20280 }, { "epoch": 0.5344695310879379, "grad_norm": 1.801716923713684, "learning_rate": 9.370800600228399e-06, "loss": 0.1643697738647461, "step": 20290 }, { "epoch": 0.5347329463324366, "grad_norm": 0.8282959461212158, "learning_rate": 9.362286361873658e-06, "loss": 0.1556757926940918, "step": 20300 }, { "epoch": 0.5349963615769353, "grad_norm": 2.593029737472534, "learning_rate": 9.353772587674413e-06, "loss": 0.16057705879211426, "step": 20310 }, { "epoch": 0.5352597768214341, "grad_norm": 1.4390743970870972, "learning_rate": 9.345259283827362e-06, "loss": 0.1529288411140442, "step": 20320 }, { "epoch": 0.5355231920659328, "grad_norm": 2.6207385063171387, "learning_rate": 9.336746456528853e-06, "loss": 0.15327625274658202, "step": 20330 }, { "epoch": 0.5357866073104316, "grad_norm": 0.8622437715530396, "learning_rate": 9.328234111974885e-06, "loss": 0.15043339729309083, "step": 20340 }, { "epoch": 0.5360500225549303, "grad_norm": 1.179559588432312, "learning_rate": 9.319722256361117e-06, "loss": 0.17072219848632814, "step": 20350 }, { "epoch": 0.5363134377994291, "grad_norm": 4.925013542175293, "learning_rate": 9.311210895882835e-06, "loss": 0.1603996753692627, "step": 20360 }, { "epoch": 0.5365768530439278, "grad_norm": 1.908305048942566, "learning_rate": 9.302700036734985e-06, "loss": 0.15658645629882811, "step": 20370 }, { "epoch": 0.5368402682884266, "grad_norm": 2.0604021549224854, "learning_rate": 9.294189685112133e-06, "loss": 0.16420031785964967, "step": 20380 }, { "epoch": 0.5371036835329253, "grad_norm": 1.6538084745407104, "learning_rate": 9.285679847208474e-06, "loss": 0.143086838722229, "step": 20390 }, { "epoch": 0.537367098777424, "grad_norm": 1.515937089920044, "learning_rate": 9.277170529217847e-06, "loss": 0.1635167717933655, "step": 20400 }, { "epoch": 0.5376305140219227, "grad_norm": 0.799386203289032, "learning_rate": 9.268661737333691e-06, "loss": 0.1594010829925537, "step": 20410 }, { "epoch": 0.5378939292664214, "grad_norm": 1.334677815437317, "learning_rate": 9.260153477749072e-06, "loss": 0.1563480257987976, "step": 20420 }, { "epoch": 0.5381573445109202, "grad_norm": 2.5109951496124268, "learning_rate": 9.251645756656676e-06, "loss": 0.16040502786636351, "step": 20430 }, { "epoch": 0.5384207597554189, "grad_norm": 1.7522660493850708, "learning_rate": 9.243138580248782e-06, "loss": 0.14978501796722413, "step": 20440 }, { "epoch": 0.5386841749999177, "grad_norm": 4.344269275665283, "learning_rate": 9.234631954717285e-06, "loss": 0.1627308964729309, "step": 20450 }, { "epoch": 0.5389475902444164, "grad_norm": 2.3147337436676025, "learning_rate": 9.226125886253673e-06, "loss": 0.16199085712432862, "step": 20460 }, { "epoch": 0.5392110054889152, "grad_norm": 1.0713818073272705, "learning_rate": 9.217620381049026e-06, "loss": 0.15329086780548096, "step": 20470 }, { "epoch": 0.5394744207334139, "grad_norm": 1.4891091585159302, "learning_rate": 9.209115445294022e-06, "loss": 0.14839699268341064, "step": 20480 }, { "epoch": 0.5397378359779126, "grad_norm": 1.536225438117981, "learning_rate": 9.200611085178917e-06, "loss": 0.1505847692489624, "step": 20490 }, { "epoch": 0.5400012512224114, "grad_norm": 1.509139060974121, "learning_rate": 9.192107306893556e-06, "loss": 0.15256298780441285, "step": 20500 }, { "epoch": 0.5402646664669101, "grad_norm": 3.880758762359619, "learning_rate": 9.183604116627354e-06, "loss": 0.1687747836112976, "step": 20510 }, { "epoch": 0.5405280817114089, "grad_norm": 1.7888318300247192, "learning_rate": 9.175101520569293e-06, "loss": 0.1695014238357544, "step": 20520 }, { "epoch": 0.5407914969559076, "grad_norm": 2.4950711727142334, "learning_rate": 9.166599524907937e-06, "loss": 0.162015700340271, "step": 20530 }, { "epoch": 0.5410549122004064, "grad_norm": 2.5274291038513184, "learning_rate": 9.158098135831404e-06, "loss": 0.15924204587936402, "step": 20540 }, { "epoch": 0.541318327444905, "grad_norm": 1.328564167022705, "learning_rate": 9.149597359527368e-06, "loss": 0.16390247344970704, "step": 20550 }, { "epoch": 0.5415817426894038, "grad_norm": 1.3596957921981812, "learning_rate": 9.141097202183063e-06, "loss": 0.16233358383178711, "step": 20560 }, { "epoch": 0.5418451579339025, "grad_norm": 1.87162184715271, "learning_rate": 9.13259766998527e-06, "loss": 0.15864961147308348, "step": 20570 }, { "epoch": 0.5421085731784012, "grad_norm": 2.7559609413146973, "learning_rate": 9.124098769120318e-06, "loss": 0.1525573253631592, "step": 20580 }, { "epoch": 0.5423719884229, "grad_norm": 1.1147100925445557, "learning_rate": 9.115600505774072e-06, "loss": 0.15991265773773194, "step": 20590 }, { "epoch": 0.5426354036673987, "grad_norm": 1.0502277612686157, "learning_rate": 9.10710288613193e-06, "loss": 0.1526641845703125, "step": 20600 }, { "epoch": 0.5428988189118975, "grad_norm": 1.4558076858520508, "learning_rate": 9.098605916378837e-06, "loss": 0.1493664264678955, "step": 20610 }, { "epoch": 0.5431622341563962, "grad_norm": 2.0174548625946045, "learning_rate": 9.090109602699239e-06, "loss": 0.14919276237487794, "step": 20620 }, { "epoch": 0.543425649400895, "grad_norm": 2.8219850063323975, "learning_rate": 9.081613951277135e-06, "loss": 0.1565387725830078, "step": 20630 }, { "epoch": 0.5436890646453937, "grad_norm": 1.2119215726852417, "learning_rate": 9.07311896829602e-06, "loss": 0.1549063205718994, "step": 20640 }, { "epoch": 0.5439524798898925, "grad_norm": 1.2772560119628906, "learning_rate": 9.064624659938902e-06, "loss": 0.14116290807724, "step": 20650 }, { "epoch": 0.5442158951343912, "grad_norm": 1.7110161781311035, "learning_rate": 9.056131032388318e-06, "loss": 0.1647857904434204, "step": 20660 }, { "epoch": 0.5444793103788899, "grad_norm": 3.3792455196380615, "learning_rate": 9.047638091826291e-06, "loss": 0.15479159355163574, "step": 20670 }, { "epoch": 0.5447427256233887, "grad_norm": 1.367091178894043, "learning_rate": 9.039145844434344e-06, "loss": 0.17168424129486085, "step": 20680 }, { "epoch": 0.5450061408678873, "grad_norm": 1.8095715045928955, "learning_rate": 9.030654296393512e-06, "loss": 0.150681209564209, "step": 20690 }, { "epoch": 0.5452695561123861, "grad_norm": 1.397249698638916, "learning_rate": 9.022163453884299e-06, "loss": 0.1505736827850342, "step": 20700 }, { "epoch": 0.5455329713568848, "grad_norm": 1.0211765766143799, "learning_rate": 9.013673323086716e-06, "loss": 0.15388789176940917, "step": 20710 }, { "epoch": 0.5457963866013836, "grad_norm": 2.27984881401062, "learning_rate": 9.005183910180243e-06, "loss": 0.1505514860153198, "step": 20720 }, { "epoch": 0.5460598018458823, "grad_norm": 1.2511247396469116, "learning_rate": 8.996695221343834e-06, "loss": 0.1547664999961853, "step": 20730 }, { "epoch": 0.546323217090381, "grad_norm": 2.139737844467163, "learning_rate": 8.988207262755936e-06, "loss": 0.14741284847259523, "step": 20740 }, { "epoch": 0.5465866323348798, "grad_norm": 1.9609532356262207, "learning_rate": 8.979720040594438e-06, "loss": 0.1665189027786255, "step": 20750 }, { "epoch": 0.5468500475793785, "grad_norm": 2.539050579071045, "learning_rate": 8.971233561036718e-06, "loss": 0.14586422443389893, "step": 20760 }, { "epoch": 0.5471134628238773, "grad_norm": 1.3833597898483276, "learning_rate": 8.962747830259598e-06, "loss": 0.14612798690795897, "step": 20770 }, { "epoch": 0.547376878068376, "grad_norm": 1.3974275588989258, "learning_rate": 8.954262854439354e-06, "loss": 0.1669377565383911, "step": 20780 }, { "epoch": 0.5476402933128748, "grad_norm": 1.2609375715255737, "learning_rate": 8.945778639751723e-06, "loss": 0.1637174367904663, "step": 20790 }, { "epoch": 0.5479037085573735, "grad_norm": 2.910423994064331, "learning_rate": 8.937295192371882e-06, "loss": 0.150505530834198, "step": 20800 }, { "epoch": 0.5481671238018723, "grad_norm": 1.556558609008789, "learning_rate": 8.928812518474445e-06, "loss": 0.1300429582595825, "step": 20810 }, { "epoch": 0.548430539046371, "grad_norm": 1.66160249710083, "learning_rate": 8.920330624233477e-06, "loss": 0.13597511053085326, "step": 20820 }, { "epoch": 0.5486939542908696, "grad_norm": 1.5723053216934204, "learning_rate": 8.911849515822457e-06, "loss": 0.15617597103118896, "step": 20830 }, { "epoch": 0.5489573695353684, "grad_norm": 1.1280449628829956, "learning_rate": 8.903369199414311e-06, "loss": 0.15798003673553468, "step": 20840 }, { "epoch": 0.5492207847798671, "grad_norm": 0.8941916227340698, "learning_rate": 8.894889681181374e-06, "loss": 0.15108835697174072, "step": 20850 }, { "epoch": 0.5494842000243659, "grad_norm": 2.1905910968780518, "learning_rate": 8.886410967295401e-06, "loss": 0.16114156246185302, "step": 20860 }, { "epoch": 0.5497476152688646, "grad_norm": 1.3693886995315552, "learning_rate": 8.877933063927576e-06, "loss": 0.1388913869857788, "step": 20870 }, { "epoch": 0.5500110305133634, "grad_norm": 2.3389103412628174, "learning_rate": 8.869455977248473e-06, "loss": 0.1519389867782593, "step": 20880 }, { "epoch": 0.5502744457578621, "grad_norm": 3.052349805831909, "learning_rate": 8.860979713428085e-06, "loss": 0.1391865611076355, "step": 20890 }, { "epoch": 0.5505378610023609, "grad_norm": 2.1393191814422607, "learning_rate": 8.852504278635806e-06, "loss": 0.14989393949508667, "step": 20900 }, { "epoch": 0.5508012762468596, "grad_norm": 2.496567726135254, "learning_rate": 8.844029679040418e-06, "loss": 0.1447909116744995, "step": 20910 }, { "epoch": 0.5510646914913583, "grad_norm": 2.160331964492798, "learning_rate": 8.835555920810104e-06, "loss": 0.1450098752975464, "step": 20920 }, { "epoch": 0.5513281067358571, "grad_norm": 1.6645318269729614, "learning_rate": 8.827083010112429e-06, "loss": 0.13801221847534179, "step": 20930 }, { "epoch": 0.5515915219803558, "grad_norm": 1.264369249343872, "learning_rate": 8.818610953114342e-06, "loss": 0.13787084817886353, "step": 20940 }, { "epoch": 0.5518549372248546, "grad_norm": 1.4914475679397583, "learning_rate": 8.810139755982173e-06, "loss": 0.14095628261566162, "step": 20950 }, { "epoch": 0.5521183524693533, "grad_norm": 1.9963555335998535, "learning_rate": 8.801669424881624e-06, "loss": 0.1587359070777893, "step": 20960 }, { "epoch": 0.5523817677138521, "grad_norm": 1.1625484228134155, "learning_rate": 8.79319996597777e-06, "loss": 0.14938271045684814, "step": 20970 }, { "epoch": 0.5526451829583507, "grad_norm": 2.859764337539673, "learning_rate": 8.784731385435049e-06, "loss": 0.1484687566757202, "step": 20980 }, { "epoch": 0.5529085982028495, "grad_norm": 0.9605329632759094, "learning_rate": 8.77626368941725e-06, "loss": 0.14126973152160643, "step": 20990 }, { "epoch": 0.5531720134473482, "grad_norm": 1.8981870412826538, "learning_rate": 8.767796884087539e-06, "loss": 0.14888336658477783, "step": 21000 }, { "epoch": 0.5534354286918469, "grad_norm": 2.501767158508301, "learning_rate": 8.759330975608418e-06, "loss": 0.15664386749267578, "step": 21010 }, { "epoch": 0.5536988439363457, "grad_norm": 1.608886957168579, "learning_rate": 8.750865970141736e-06, "loss": 0.15053164958953857, "step": 21020 }, { "epoch": 0.5539622591808444, "grad_norm": 1.194658637046814, "learning_rate": 8.742401873848696e-06, "loss": 0.1543223261833191, "step": 21030 }, { "epoch": 0.5542256744253432, "grad_norm": 1.363739013671875, "learning_rate": 8.733938692889827e-06, "loss": 0.1545048952102661, "step": 21040 }, { "epoch": 0.5544890896698419, "grad_norm": 1.1441094875335693, "learning_rate": 8.725476433425003e-06, "loss": 0.15273171663284302, "step": 21050 }, { "epoch": 0.5547525049143407, "grad_norm": 2.157519578933716, "learning_rate": 8.717015101613417e-06, "loss": 0.15603153705596923, "step": 21060 }, { "epoch": 0.5550159201588394, "grad_norm": 1.0963715314865112, "learning_rate": 8.708554703613592e-06, "loss": 0.15238429307937623, "step": 21070 }, { "epoch": 0.5552793354033382, "grad_norm": 1.8103610277175903, "learning_rate": 8.700095245583374e-06, "loss": 0.13462436199188232, "step": 21080 }, { "epoch": 0.5555427506478369, "grad_norm": 1.5299246311187744, "learning_rate": 8.691636733679915e-06, "loss": 0.14925527572631836, "step": 21090 }, { "epoch": 0.5558061658923356, "grad_norm": 1.8846828937530518, "learning_rate": 8.683179174059691e-06, "loss": 0.16486502885818483, "step": 21100 }, { "epoch": 0.5560695811368344, "grad_norm": 2.0616066455841064, "learning_rate": 8.67472257287848e-06, "loss": 0.15265092849731446, "step": 21110 }, { "epoch": 0.556332996381333, "grad_norm": 3.6940436363220215, "learning_rate": 8.666266936291354e-06, "loss": 0.15718095302581786, "step": 21120 }, { "epoch": 0.5565964116258318, "grad_norm": 2.495020866394043, "learning_rate": 8.657812270452695e-06, "loss": 0.1631619453430176, "step": 21130 }, { "epoch": 0.5568598268703305, "grad_norm": 2.884885787963867, "learning_rate": 8.649358581516171e-06, "loss": 0.15888277292251587, "step": 21140 }, { "epoch": 0.5571232421148293, "grad_norm": 1.1035664081573486, "learning_rate": 8.640905875634744e-06, "loss": 0.15265328884124757, "step": 21150 }, { "epoch": 0.557386657359328, "grad_norm": 2.1372199058532715, "learning_rate": 8.632454158960656e-06, "loss": 0.14980511665344237, "step": 21160 }, { "epoch": 0.5576500726038268, "grad_norm": 1.861580729484558, "learning_rate": 8.624003437645431e-06, "loss": 0.1336355447769165, "step": 21170 }, { "epoch": 0.5579134878483255, "grad_norm": 4.058391094207764, "learning_rate": 8.615553717839873e-06, "loss": 0.13089311122894287, "step": 21180 }, { "epoch": 0.5581769030928242, "grad_norm": 1.8697351217269897, "learning_rate": 8.607105005694048e-06, "loss": 0.1699019193649292, "step": 21190 }, { "epoch": 0.558440318337323, "grad_norm": 1.4119104146957397, "learning_rate": 8.598657307357293e-06, "loss": 0.1532852053642273, "step": 21200 }, { "epoch": 0.5587037335818217, "grad_norm": 1.2555187940597534, "learning_rate": 8.590210628978213e-06, "loss": 0.14619138240814208, "step": 21210 }, { "epoch": 0.5589671488263205, "grad_norm": 2.362445592880249, "learning_rate": 8.581764976704653e-06, "loss": 0.155700945854187, "step": 21220 }, { "epoch": 0.5592305640708192, "grad_norm": 1.7455013990402222, "learning_rate": 8.573320356683739e-06, "loss": 0.15817434787750245, "step": 21230 }, { "epoch": 0.559493979315318, "grad_norm": 1.1404563188552856, "learning_rate": 8.564876775061819e-06, "loss": 0.141549015045166, "step": 21240 }, { "epoch": 0.5597573945598167, "grad_norm": 2.338893175125122, "learning_rate": 8.556434237984496e-06, "loss": 0.13511965274810792, "step": 21250 }, { "epoch": 0.5600208098043153, "grad_norm": 2.740928888320923, "learning_rate": 8.547992751596617e-06, "loss": 0.15503990650177002, "step": 21260 }, { "epoch": 0.5602842250488141, "grad_norm": 1.2564418315887451, "learning_rate": 8.53955232204226e-06, "loss": 0.15346722602844237, "step": 21270 }, { "epoch": 0.5605476402933128, "grad_norm": 1.9735493659973145, "learning_rate": 8.531112955464728e-06, "loss": 0.14962049722671508, "step": 21280 }, { "epoch": 0.5608110555378116, "grad_norm": 1.212249755859375, "learning_rate": 8.522674658006564e-06, "loss": 0.13763593435287474, "step": 21290 }, { "epoch": 0.5610744707823103, "grad_norm": 2.134455680847168, "learning_rate": 8.514237435809517e-06, "loss": 0.13914434909820556, "step": 21300 }, { "epoch": 0.5613378860268091, "grad_norm": 0.9556203484535217, "learning_rate": 8.505801295014569e-06, "loss": 0.1410364866256714, "step": 21310 }, { "epoch": 0.5616013012713078, "grad_norm": 2.169667959213257, "learning_rate": 8.497366241761907e-06, "loss": 0.14645113945007324, "step": 21320 }, { "epoch": 0.5618647165158066, "grad_norm": 1.0189564228057861, "learning_rate": 8.488932282190914e-06, "loss": 0.1515304684638977, "step": 21330 }, { "epoch": 0.5621281317603053, "grad_norm": 1.1086182594299316, "learning_rate": 8.480499422440207e-06, "loss": 0.14560036659240722, "step": 21340 }, { "epoch": 0.562391547004804, "grad_norm": 2.849871873855591, "learning_rate": 8.472067668647571e-06, "loss": 0.152410089969635, "step": 21350 }, { "epoch": 0.5626549622493028, "grad_norm": 1.3940744400024414, "learning_rate": 8.463637026950013e-06, "loss": 0.14482229948043823, "step": 21360 }, { "epoch": 0.5629183774938015, "grad_norm": 1.9042760133743286, "learning_rate": 8.455207503483708e-06, "loss": 0.14181623458862305, "step": 21370 }, { "epoch": 0.5631817927383003, "grad_norm": 5.018985748291016, "learning_rate": 8.446779104384028e-06, "loss": 0.15894765853881837, "step": 21380 }, { "epoch": 0.563445207982799, "grad_norm": 1.3773530721664429, "learning_rate": 8.43835183578553e-06, "loss": 0.1411137580871582, "step": 21390 }, { "epoch": 0.5637086232272978, "grad_norm": 1.688269853591919, "learning_rate": 8.429925703821941e-06, "loss": 0.1655648946762085, "step": 21400 }, { "epoch": 0.5639720384717964, "grad_norm": 2.761652708053589, "learning_rate": 8.421500714626164e-06, "loss": 0.1647416591644287, "step": 21410 }, { "epoch": 0.5642354537162952, "grad_norm": 2.8237862586975098, "learning_rate": 8.413076874330266e-06, "loss": 0.15733802318572998, "step": 21420 }, { "epoch": 0.5644988689607939, "grad_norm": 1.3306758403778076, "learning_rate": 8.404654189065486e-06, "loss": 0.15244755744934083, "step": 21430 }, { "epoch": 0.5647622842052926, "grad_norm": 1.156952142715454, "learning_rate": 8.396232664962218e-06, "loss": 0.16047805547714233, "step": 21440 }, { "epoch": 0.5650256994497914, "grad_norm": 0.842793345451355, "learning_rate": 8.387812308150011e-06, "loss": 0.12693930864334108, "step": 21450 }, { "epoch": 0.5652891146942901, "grad_norm": 1.3786693811416626, "learning_rate": 8.379393124757554e-06, "loss": 0.17660685777664184, "step": 21460 }, { "epoch": 0.5655525299387889, "grad_norm": 1.0627182722091675, "learning_rate": 8.370975120912708e-06, "loss": 0.158439302444458, "step": 21470 }, { "epoch": 0.5658159451832876, "grad_norm": 3.6676371097564697, "learning_rate": 8.362558302742447e-06, "loss": 0.1500684380531311, "step": 21480 }, { "epoch": 0.5660793604277864, "grad_norm": 3.273338794708252, "learning_rate": 8.354142676372898e-06, "loss": 0.15721867084503174, "step": 21490 }, { "epoch": 0.5663427756722851, "grad_norm": 0.8739500045776367, "learning_rate": 8.345728247929319e-06, "loss": 0.14943571090698243, "step": 21500 }, { "epoch": 0.5666061909167839, "grad_norm": 2.351999044418335, "learning_rate": 8.33731502353609e-06, "loss": 0.1508263349533081, "step": 21510 }, { "epoch": 0.5668696061612826, "grad_norm": 2.4633004665374756, "learning_rate": 8.328903009316724e-06, "loss": 0.17388648986816407, "step": 21520 }, { "epoch": 0.5671330214057813, "grad_norm": 2.475206136703491, "learning_rate": 8.320492211393844e-06, "loss": 0.15280675888061523, "step": 21530 }, { "epoch": 0.5673964366502801, "grad_norm": 3.398625612258911, "learning_rate": 8.312082635889193e-06, "loss": 0.15959800481796266, "step": 21540 }, { "epoch": 0.5676598518947787, "grad_norm": 1.234135627746582, "learning_rate": 8.303674288923621e-06, "loss": 0.15231044292449952, "step": 21550 }, { "epoch": 0.5679232671392775, "grad_norm": 1.4154012203216553, "learning_rate": 8.29526717661709e-06, "loss": 0.15728222131729125, "step": 21560 }, { "epoch": 0.5681866823837762, "grad_norm": 1.9816855192184448, "learning_rate": 8.286861305088655e-06, "loss": 0.14375197887420654, "step": 21570 }, { "epoch": 0.568450097628275, "grad_norm": 1.1379021406173706, "learning_rate": 8.278456680456476e-06, "loss": 0.1527440667152405, "step": 21580 }, { "epoch": 0.5687135128727737, "grad_norm": 1.607303261756897, "learning_rate": 8.270053308837791e-06, "loss": 0.14522018432617187, "step": 21590 }, { "epoch": 0.5689769281172725, "grad_norm": 1.9323077201843262, "learning_rate": 8.261651196348953e-06, "loss": 0.15076882839202882, "step": 21600 }, { "epoch": 0.5692403433617712, "grad_norm": 2.6531119346618652, "learning_rate": 8.253250349105369e-06, "loss": 0.15037128925323487, "step": 21610 }, { "epoch": 0.5695037586062699, "grad_norm": 1.8953895568847656, "learning_rate": 8.24485077322154e-06, "loss": 0.15414042472839357, "step": 21620 }, { "epoch": 0.5697671738507687, "grad_norm": 1.3131239414215088, "learning_rate": 8.236452474811046e-06, "loss": 0.15227892398834228, "step": 21630 }, { "epoch": 0.5700305890952674, "grad_norm": 1.7242991924285889, "learning_rate": 8.228055459986521e-06, "loss": 0.15299687385559083, "step": 21640 }, { "epoch": 0.5702940043397662, "grad_norm": 2.5663130283355713, "learning_rate": 8.219659734859686e-06, "loss": 0.13918042182922363, "step": 21650 }, { "epoch": 0.5705574195842649, "grad_norm": 2.1068780422210693, "learning_rate": 8.211265305541304e-06, "loss": 0.15599123239517212, "step": 21660 }, { "epoch": 0.5708208348287637, "grad_norm": 2.684523820877075, "learning_rate": 8.202872178141204e-06, "loss": 0.16083693504333496, "step": 21670 }, { "epoch": 0.5710842500732624, "grad_norm": 4.650776386260986, "learning_rate": 8.19448035876827e-06, "loss": 0.1374622941017151, "step": 21680 }, { "epoch": 0.5713476653177612, "grad_norm": 1.7828888893127441, "learning_rate": 8.186089853530426e-06, "loss": 0.15060557126998902, "step": 21690 }, { "epoch": 0.5716110805622598, "grad_norm": 4.7199578285217285, "learning_rate": 8.177700668534646e-06, "loss": 0.15529375076293944, "step": 21700 }, { "epoch": 0.5718744958067585, "grad_norm": 1.1355528831481934, "learning_rate": 8.169312809886944e-06, "loss": 0.1443256616592407, "step": 21710 }, { "epoch": 0.5721379110512573, "grad_norm": 2.4722752571105957, "learning_rate": 8.160926283692355e-06, "loss": 0.14855830669403075, "step": 21720 }, { "epoch": 0.572401326295756, "grad_norm": 2.173485040664673, "learning_rate": 8.152541096054965e-06, "loss": 0.1421190857887268, "step": 21730 }, { "epoch": 0.5726647415402548, "grad_norm": 1.801295518875122, "learning_rate": 8.144157253077871e-06, "loss": 0.15059318542480468, "step": 21740 }, { "epoch": 0.5729281567847535, "grad_norm": 1.6947262287139893, "learning_rate": 8.135774760863193e-06, "loss": 0.15137977600097657, "step": 21750 }, { "epoch": 0.5731915720292523, "grad_norm": 1.1068843603134155, "learning_rate": 8.127393625512077e-06, "loss": 0.1459658622741699, "step": 21760 }, { "epoch": 0.573454987273751, "grad_norm": 2.066021680831909, "learning_rate": 8.11901385312467e-06, "loss": 0.15394141674041747, "step": 21770 }, { "epoch": 0.5737184025182498, "grad_norm": 1.4724466800689697, "learning_rate": 8.110635449800132e-06, "loss": 0.15565274953842162, "step": 21780 }, { "epoch": 0.5739818177627485, "grad_norm": 1.7074573040008545, "learning_rate": 8.10225842163663e-06, "loss": 0.15496214628219604, "step": 21790 }, { "epoch": 0.5742452330072472, "grad_norm": 1.554263949394226, "learning_rate": 8.093882774731322e-06, "loss": 0.16163214445114135, "step": 21800 }, { "epoch": 0.574508648251746, "grad_norm": 2.452634572982788, "learning_rate": 8.085508515180373e-06, "loss": 0.1534902811050415, "step": 21810 }, { "epoch": 0.5747720634962447, "grad_norm": 1.8458257913589478, "learning_rate": 8.077135649078917e-06, "loss": 0.14405357837677002, "step": 21820 }, { "epoch": 0.5750354787407435, "grad_norm": 2.2213239669799805, "learning_rate": 8.068764182521102e-06, "loss": 0.16010674238204955, "step": 21830 }, { "epoch": 0.5752988939852421, "grad_norm": 2.098254919052124, "learning_rate": 8.060394121600033e-06, "loss": 0.14626736640930177, "step": 21840 }, { "epoch": 0.5755623092297409, "grad_norm": 1.427281141281128, "learning_rate": 8.052025472407802e-06, "loss": 0.15278565883636475, "step": 21850 }, { "epoch": 0.5758257244742396, "grad_norm": 1.6200398206710815, "learning_rate": 8.043658241035477e-06, "loss": 0.1586440086364746, "step": 21860 }, { "epoch": 0.5760891397187383, "grad_norm": 1.0467792749404907, "learning_rate": 8.035292433573088e-06, "loss": 0.14516010284423828, "step": 21870 }, { "epoch": 0.5763525549632371, "grad_norm": 1.90810227394104, "learning_rate": 8.026928056109628e-06, "loss": 0.14732160568237304, "step": 21880 }, { "epoch": 0.5766159702077358, "grad_norm": 2.5917110443115234, "learning_rate": 8.018565114733056e-06, "loss": 0.1596977710723877, "step": 21890 }, { "epoch": 0.5768793854522346, "grad_norm": 1.0988473892211914, "learning_rate": 8.010203615530279e-06, "loss": 0.14932458400726317, "step": 21900 }, { "epoch": 0.5771428006967333, "grad_norm": 2.1641101837158203, "learning_rate": 8.00184356458716e-06, "loss": 0.15316474437713623, "step": 21910 }, { "epoch": 0.5774062159412321, "grad_norm": 1.9858789443969727, "learning_rate": 7.993484967988504e-06, "loss": 0.15254414081573486, "step": 21920 }, { "epoch": 0.5776696311857308, "grad_norm": 1.3008300065994263, "learning_rate": 7.98512783181805e-06, "loss": 0.1396868944168091, "step": 21930 }, { "epoch": 0.5779330464302296, "grad_norm": 1.275011658668518, "learning_rate": 7.976772162158498e-06, "loss": 0.1707656741142273, "step": 21940 }, { "epoch": 0.5781964616747283, "grad_norm": 1.130198359489441, "learning_rate": 7.968417965091449e-06, "loss": 0.13965941667556764, "step": 21950 }, { "epoch": 0.578459876919227, "grad_norm": 2.8277933597564697, "learning_rate": 7.960065246697456e-06, "loss": 0.14474520683288575, "step": 21960 }, { "epoch": 0.5787232921637258, "grad_norm": 1.3166108131408691, "learning_rate": 7.951714013055988e-06, "loss": 0.15836355686187745, "step": 21970 }, { "epoch": 0.5789867074082244, "grad_norm": 2.257054328918457, "learning_rate": 7.94336427024543e-06, "loss": 0.152459454536438, "step": 21980 }, { "epoch": 0.5792501226527232, "grad_norm": 1.6497747898101807, "learning_rate": 7.93501602434309e-06, "loss": 0.14459902048110962, "step": 21990 }, { "epoch": 0.5795135378972219, "grad_norm": 1.69979989528656, "learning_rate": 7.926669281425177e-06, "loss": 0.15647454261779786, "step": 22000 }, { "epoch": 0.5797769531417207, "grad_norm": 1.692007064819336, "learning_rate": 7.91832404756681e-06, "loss": 0.16169894933700563, "step": 22010 }, { "epoch": 0.5800403683862194, "grad_norm": 2.3823914527893066, "learning_rate": 7.909980328842015e-06, "loss": 0.15055183172225953, "step": 22020 }, { "epoch": 0.5803037836307182, "grad_norm": 1.2822022438049316, "learning_rate": 7.901638131323705e-06, "loss": 0.14386255741119386, "step": 22030 }, { "epoch": 0.5805671988752169, "grad_norm": 1.4576858282089233, "learning_rate": 7.893297461083697e-06, "loss": 0.15304899215698242, "step": 22040 }, { "epoch": 0.5808306141197156, "grad_norm": 2.858038902282715, "learning_rate": 7.884958324192689e-06, "loss": 0.1601717948913574, "step": 22050 }, { "epoch": 0.5810940293642144, "grad_norm": 1.7704893350601196, "learning_rate": 7.876620726720257e-06, "loss": 0.16607980728149413, "step": 22060 }, { "epoch": 0.5813574446087131, "grad_norm": 1.5701237916946411, "learning_rate": 7.868284674734877e-06, "loss": 0.15746374130249025, "step": 22070 }, { "epoch": 0.5816208598532119, "grad_norm": 0.9994221329689026, "learning_rate": 7.859950174303879e-06, "loss": 0.15289721488952637, "step": 22080 }, { "epoch": 0.5818842750977106, "grad_norm": 1.515264630317688, "learning_rate": 7.851617231493472e-06, "loss": 0.15431476831436158, "step": 22090 }, { "epoch": 0.5821476903422094, "grad_norm": 2.6085097789764404, "learning_rate": 7.843285852368734e-06, "loss": 0.1478745937347412, "step": 22100 }, { "epoch": 0.5824111055867081, "grad_norm": 1.108466386795044, "learning_rate": 7.8349560429936e-06, "loss": 0.14808273315429688, "step": 22110 }, { "epoch": 0.5826745208312069, "grad_norm": 3.47076153755188, "learning_rate": 7.826627809430867e-06, "loss": 0.15813822746276857, "step": 22120 }, { "epoch": 0.5829379360757055, "grad_norm": 1.3112080097198486, "learning_rate": 7.818301157742182e-06, "loss": 0.14495353698730468, "step": 22130 }, { "epoch": 0.5832013513202042, "grad_norm": 2.1763858795166016, "learning_rate": 7.809976093988039e-06, "loss": 0.14851994514465333, "step": 22140 }, { "epoch": 0.583464766564703, "grad_norm": 2.4375948905944824, "learning_rate": 7.801652624227782e-06, "loss": 0.16536809206008912, "step": 22150 }, { "epoch": 0.5837281818092017, "grad_norm": 2.5043697357177734, "learning_rate": 7.79333075451959e-06, "loss": 0.14712488651275635, "step": 22160 }, { "epoch": 0.5839915970537005, "grad_norm": 1.3465851545333862, "learning_rate": 7.785010490920477e-06, "loss": 0.1476643443107605, "step": 22170 }, { "epoch": 0.5842550122981992, "grad_norm": 3.344003915786743, "learning_rate": 7.776691839486297e-06, "loss": 0.14453173875808717, "step": 22180 }, { "epoch": 0.584518427542698, "grad_norm": 1.615980625152588, "learning_rate": 7.76837480627171e-06, "loss": 0.1521540403366089, "step": 22190 }, { "epoch": 0.5847818427871967, "grad_norm": 2.8994576930999756, "learning_rate": 7.760059397330225e-06, "loss": 0.13817684650421141, "step": 22200 }, { "epoch": 0.5850452580316955, "grad_norm": 2.469525098800659, "learning_rate": 7.75174561871415e-06, "loss": 0.14497947692871094, "step": 22210 }, { "epoch": 0.5853086732761942, "grad_norm": 1.3600075244903564, "learning_rate": 7.743433476474608e-06, "loss": 0.14482923746109008, "step": 22220 }, { "epoch": 0.5855720885206929, "grad_norm": 3.38305926322937, "learning_rate": 7.73512297666154e-06, "loss": 0.14813109636306762, "step": 22230 }, { "epoch": 0.5858355037651917, "grad_norm": 1.4895377159118652, "learning_rate": 7.726814125323683e-06, "loss": 0.14238460063934327, "step": 22240 }, { "epoch": 0.5860989190096904, "grad_norm": 3.657975196838379, "learning_rate": 7.718506928508583e-06, "loss": 0.1407778024673462, "step": 22250 }, { "epoch": 0.5863623342541892, "grad_norm": 1.5785173177719116, "learning_rate": 7.71020139226257e-06, "loss": 0.1420443058013916, "step": 22260 }, { "epoch": 0.5866257494986878, "grad_norm": 0.929151177406311, "learning_rate": 7.701897522630772e-06, "loss": 0.1452374815940857, "step": 22270 }, { "epoch": 0.5868891647431866, "grad_norm": 1.851035237312317, "learning_rate": 7.693595325657109e-06, "loss": 0.1570465087890625, "step": 22280 }, { "epoch": 0.5871525799876853, "grad_norm": 1.139420747756958, "learning_rate": 7.685294807384273e-06, "loss": 0.14931998252868653, "step": 22290 }, { "epoch": 0.5874159952321841, "grad_norm": 2.014949083328247, "learning_rate": 7.676995973853743e-06, "loss": 0.14188685417175292, "step": 22300 }, { "epoch": 0.5876794104766828, "grad_norm": 2.180875301361084, "learning_rate": 7.668698831105768e-06, "loss": 0.13391382694244386, "step": 22310 }, { "epoch": 0.5879428257211815, "grad_norm": 2.1112873554229736, "learning_rate": 7.66040338517936e-06, "loss": 0.15775548219680785, "step": 22320 }, { "epoch": 0.5882062409656803, "grad_norm": 1.2052031755447388, "learning_rate": 7.652109642112308e-06, "loss": 0.14956908226013182, "step": 22330 }, { "epoch": 0.588469656210179, "grad_norm": 0.9659489989280701, "learning_rate": 7.643817607941157e-06, "loss": 0.1315206527709961, "step": 22340 }, { "epoch": 0.5887330714546778, "grad_norm": 1.784672737121582, "learning_rate": 7.635527288701201e-06, "loss": 0.13727030754089356, "step": 22350 }, { "epoch": 0.5889964866991765, "grad_norm": 1.7830249071121216, "learning_rate": 7.6272386904265005e-06, "loss": 0.15568249225616454, "step": 22360 }, { "epoch": 0.5892599019436753, "grad_norm": 1.2267553806304932, "learning_rate": 7.618951819149848e-06, "loss": 0.13978500366210939, "step": 22370 }, { "epoch": 0.589523317188174, "grad_norm": 1.4520193338394165, "learning_rate": 7.610666680902791e-06, "loss": 0.12428560256958007, "step": 22380 }, { "epoch": 0.5897867324326727, "grad_norm": 1.163927674293518, "learning_rate": 7.602383281715608e-06, "loss": 0.13405494689941405, "step": 22390 }, { "epoch": 0.5900501476771715, "grad_norm": 4.182002067565918, "learning_rate": 7.594101627617312e-06, "loss": 0.1427021622657776, "step": 22400 }, { "epoch": 0.5903135629216701, "grad_norm": 3.5605571269989014, "learning_rate": 7.585821724635656e-06, "loss": 0.14996615648269654, "step": 22410 }, { "epoch": 0.5905769781661689, "grad_norm": 2.684140682220459, "learning_rate": 7.577543578797097e-06, "loss": 0.14454375505447387, "step": 22420 }, { "epoch": 0.5908403934106676, "grad_norm": 4.810328006744385, "learning_rate": 7.569267196126844e-06, "loss": 0.14960017204284667, "step": 22430 }, { "epoch": 0.5911038086551664, "grad_norm": 1.4968206882476807, "learning_rate": 7.560992582648793e-06, "loss": 0.15861241817474364, "step": 22440 }, { "epoch": 0.5913672238996651, "grad_norm": 0.871313214302063, "learning_rate": 7.552719744385566e-06, "loss": 0.14994168281555176, "step": 22450 }, { "epoch": 0.5916306391441639, "grad_norm": 1.3117263317108154, "learning_rate": 7.544448687358495e-06, "loss": 0.14903331995010377, "step": 22460 }, { "epoch": 0.5918940543886626, "grad_norm": 1.1415622234344482, "learning_rate": 7.536179417587612e-06, "loss": 0.15793852806091307, "step": 22470 }, { "epoch": 0.5921574696331613, "grad_norm": 2.477957248687744, "learning_rate": 7.527911941091644e-06, "loss": 0.14844436645507814, "step": 22480 }, { "epoch": 0.5924208848776601, "grad_norm": 1.8930084705352783, "learning_rate": 7.519646263888023e-06, "loss": 0.15693881511688232, "step": 22490 }, { "epoch": 0.5926843001221588, "grad_norm": 1.6962405443191528, "learning_rate": 7.511382391992861e-06, "loss": 0.13903785943984986, "step": 22500 }, { "epoch": 0.5929477153666576, "grad_norm": 2.9225947856903076, "learning_rate": 7.503120331420963e-06, "loss": 0.15195162296295167, "step": 22510 }, { "epoch": 0.5932111306111563, "grad_norm": 1.8557738065719604, "learning_rate": 7.494860088185815e-06, "loss": 0.14568432569503784, "step": 22520 }, { "epoch": 0.5934745458556551, "grad_norm": 1.2538342475891113, "learning_rate": 7.48660166829957e-06, "loss": 0.14409466981887817, "step": 22530 }, { "epoch": 0.5937379611001538, "grad_norm": 1.6550335884094238, "learning_rate": 7.478345077773075e-06, "loss": 0.14275012016296387, "step": 22540 }, { "epoch": 0.5940013763446526, "grad_norm": 1.5266855955123901, "learning_rate": 7.470090322615822e-06, "loss": 0.15627313852310182, "step": 22550 }, { "epoch": 0.5942647915891512, "grad_norm": 2.697666883468628, "learning_rate": 7.4618374088359805e-06, "loss": 0.15600982904434205, "step": 22560 }, { "epoch": 0.5945282068336499, "grad_norm": 1.617370843887329, "learning_rate": 7.45358634244038e-06, "loss": 0.14037845134735108, "step": 22570 }, { "epoch": 0.5947916220781487, "grad_norm": 3.3359153270721436, "learning_rate": 7.4453371294345e-06, "loss": 0.14367227554321288, "step": 22580 }, { "epoch": 0.5950550373226474, "grad_norm": 1.2580533027648926, "learning_rate": 7.437089775822477e-06, "loss": 0.16309473514556885, "step": 22590 }, { "epoch": 0.5953184525671462, "grad_norm": 3.5278103351593018, "learning_rate": 7.4288442876070865e-06, "loss": 0.1447066068649292, "step": 22600 }, { "epoch": 0.5955818678116449, "grad_norm": 2.0824854373931885, "learning_rate": 7.420600670789753e-06, "loss": 0.15929713249206542, "step": 22610 }, { "epoch": 0.5958452830561437, "grad_norm": 1.3629480600357056, "learning_rate": 7.412358931370536e-06, "loss": 0.14168138504028321, "step": 22620 }, { "epoch": 0.5961086983006424, "grad_norm": 0.9786011576652527, "learning_rate": 7.404119075348129e-06, "loss": 0.138591992855072, "step": 22630 }, { "epoch": 0.5963721135451412, "grad_norm": 1.336868405342102, "learning_rate": 7.3958811087198555e-06, "loss": 0.15037250518798828, "step": 22640 }, { "epoch": 0.5966355287896399, "grad_norm": 2.4528181552886963, "learning_rate": 7.3876450374816644e-06, "loss": 0.13693100214004517, "step": 22650 }, { "epoch": 0.5968989440341386, "grad_norm": 2.0146303176879883, "learning_rate": 7.379410867628115e-06, "loss": 0.14092170000076293, "step": 22660 }, { "epoch": 0.5971623592786374, "grad_norm": 2.94650936126709, "learning_rate": 7.371178605152405e-06, "loss": 0.16098971366882325, "step": 22670 }, { "epoch": 0.597425774523136, "grad_norm": 1.8171688318252563, "learning_rate": 7.362948256046321e-06, "loss": 0.1335107445716858, "step": 22680 }, { "epoch": 0.5976891897676349, "grad_norm": 1.7063241004943848, "learning_rate": 7.354719826300266e-06, "loss": 0.14667452573776246, "step": 22690 }, { "epoch": 0.5979526050121335, "grad_norm": 3.91907000541687, "learning_rate": 7.346493321903251e-06, "loss": 0.14511308670043946, "step": 22700 }, { "epoch": 0.5982160202566323, "grad_norm": 2.269705295562744, "learning_rate": 7.338268748842876e-06, "loss": 0.1488993763923645, "step": 22710 }, { "epoch": 0.598479435501131, "grad_norm": 3.924834966659546, "learning_rate": 7.330046113105345e-06, "loss": 0.16069278717041016, "step": 22720 }, { "epoch": 0.5987428507456298, "grad_norm": 2.643754482269287, "learning_rate": 7.321825420675444e-06, "loss": 0.13953560590744019, "step": 22730 }, { "epoch": 0.5990062659901285, "grad_norm": 1.6370291709899902, "learning_rate": 7.3136066775365455e-06, "loss": 0.12960360050201417, "step": 22740 }, { "epoch": 0.5992696812346272, "grad_norm": 1.5974769592285156, "learning_rate": 7.305389889670611e-06, "loss": 0.13465906381607057, "step": 22750 }, { "epoch": 0.599533096479126, "grad_norm": 2.3585703372955322, "learning_rate": 7.2971750630581675e-06, "loss": 0.159224271774292, "step": 22760 }, { "epoch": 0.5997965117236247, "grad_norm": 2.322598457336426, "learning_rate": 7.2889622036783245e-06, "loss": 0.1429004430770874, "step": 22770 }, { "epoch": 0.6000599269681235, "grad_norm": 1.2016079425811768, "learning_rate": 7.280751317508756e-06, "loss": 0.1390307664871216, "step": 22780 }, { "epoch": 0.6003233422126222, "grad_norm": 2.7242579460144043, "learning_rate": 7.272542410525691e-06, "loss": 0.1472085475921631, "step": 22790 }, { "epoch": 0.600586757457121, "grad_norm": 4.469705104827881, "learning_rate": 7.2643354887039395e-06, "loss": 0.15751090049743652, "step": 22800 }, { "epoch": 0.6008501727016197, "grad_norm": 1.9763599634170532, "learning_rate": 7.256130558016843e-06, "loss": 0.1414959192276001, "step": 22810 }, { "epoch": 0.6011135879461185, "grad_norm": 2.178252696990967, "learning_rate": 7.247927624436304e-06, "loss": 0.14920396804809571, "step": 22820 }, { "epoch": 0.6013770031906172, "grad_norm": 3.7500345706939697, "learning_rate": 7.239726693932779e-06, "loss": 0.14840128421783447, "step": 22830 }, { "epoch": 0.6016404184351158, "grad_norm": 1.5202512741088867, "learning_rate": 7.2315277724752495e-06, "loss": 0.1346398711204529, "step": 22840 }, { "epoch": 0.6019038336796146, "grad_norm": 2.223884105682373, "learning_rate": 7.223330866031253e-06, "loss": 0.14373687505722046, "step": 22850 }, { "epoch": 0.6021672489241133, "grad_norm": 2.045170545578003, "learning_rate": 7.215135980566847e-06, "loss": 0.1402364492416382, "step": 22860 }, { "epoch": 0.6024306641686121, "grad_norm": 2.059710741043091, "learning_rate": 7.2069431220466215e-06, "loss": 0.14510976076126098, "step": 22870 }, { "epoch": 0.6026940794131108, "grad_norm": 1.9782174825668335, "learning_rate": 7.198752296433697e-06, "loss": 0.1488866090774536, "step": 22880 }, { "epoch": 0.6029574946576096, "grad_norm": 0.9300306439399719, "learning_rate": 7.190563509689705e-06, "loss": 0.14214866161346434, "step": 22890 }, { "epoch": 0.6032209099021083, "grad_norm": 1.8252007961273193, "learning_rate": 7.182376767774803e-06, "loss": 0.1454352021217346, "step": 22900 }, { "epoch": 0.603484325146607, "grad_norm": 1.3516958951950073, "learning_rate": 7.174192076647653e-06, "loss": 0.1487440824508667, "step": 22910 }, { "epoch": 0.6037477403911058, "grad_norm": 0.9324631690979004, "learning_rate": 7.166009442265423e-06, "loss": 0.136998450756073, "step": 22920 }, { "epoch": 0.6040111556356045, "grad_norm": 1.7175514698028564, "learning_rate": 7.15782887058379e-06, "loss": 0.12935173511505127, "step": 22930 }, { "epoch": 0.6042745708801033, "grad_norm": 2.343916416168213, "learning_rate": 7.149650367556926e-06, "loss": 0.1382911443710327, "step": 22940 }, { "epoch": 0.604537986124602, "grad_norm": 1.559006690979004, "learning_rate": 7.141473939137498e-06, "loss": 0.14165475368499755, "step": 22950 }, { "epoch": 0.6048014013691008, "grad_norm": 1.8927712440490723, "learning_rate": 7.133299591276665e-06, "loss": 0.14985394477844238, "step": 22960 }, { "epoch": 0.6050648166135995, "grad_norm": 4.164748668670654, "learning_rate": 7.12512732992407e-06, "loss": 0.13801054954528807, "step": 22970 }, { "epoch": 0.6053282318580983, "grad_norm": 2.837956428527832, "learning_rate": 7.1169571610278375e-06, "loss": 0.1590420961380005, "step": 22980 }, { "epoch": 0.6055916471025969, "grad_norm": 1.9007610082626343, "learning_rate": 7.108789090534569e-06, "loss": 0.13161201477050782, "step": 22990 }, { "epoch": 0.6058550623470956, "grad_norm": 1.2265169620513916, "learning_rate": 7.100623124389334e-06, "loss": 0.13448684215545653, "step": 23000 }, { "epoch": 0.6061184775915944, "grad_norm": 1.6626315116882324, "learning_rate": 7.092459268535685e-06, "loss": 0.1539805054664612, "step": 23010 }, { "epoch": 0.6063818928360931, "grad_norm": 2.3342041969299316, "learning_rate": 7.084297528915616e-06, "loss": 0.15376026630401612, "step": 23020 }, { "epoch": 0.6066453080805919, "grad_norm": 1.436623215675354, "learning_rate": 7.076137911469603e-06, "loss": 0.1370314598083496, "step": 23030 }, { "epoch": 0.6069087233250906, "grad_norm": 2.154818058013916, "learning_rate": 7.067980422136565e-06, "loss": 0.14388267993927, "step": 23040 }, { "epoch": 0.6071721385695894, "grad_norm": 3.3133187294006348, "learning_rate": 7.059825066853872e-06, "loss": 0.14516947269439698, "step": 23050 }, { "epoch": 0.6074355538140881, "grad_norm": 2.1618828773498535, "learning_rate": 7.0516718515573445e-06, "loss": 0.14303712844848632, "step": 23060 }, { "epoch": 0.6076989690585869, "grad_norm": 1.768021821975708, "learning_rate": 7.043520782181248e-06, "loss": 0.13656917810440064, "step": 23070 }, { "epoch": 0.6079623843030856, "grad_norm": 1.1324745416641235, "learning_rate": 7.035371864658276e-06, "loss": 0.15719785690307617, "step": 23080 }, { "epoch": 0.6082257995475843, "grad_norm": 2.7079155445098877, "learning_rate": 7.0272251049195684e-06, "loss": 0.13902642726898193, "step": 23090 }, { "epoch": 0.6084892147920831, "grad_norm": 1.6138129234313965, "learning_rate": 7.019080508894683e-06, "loss": 0.14991475343704225, "step": 23100 }, { "epoch": 0.6087526300365818, "grad_norm": 1.2982001304626465, "learning_rate": 7.010938082511614e-06, "loss": 0.14402225017547607, "step": 23110 }, { "epoch": 0.6090160452810806, "grad_norm": 1.7836382389068604, "learning_rate": 7.0027978316967705e-06, "loss": 0.1307303786277771, "step": 23120 }, { "epoch": 0.6092794605255792, "grad_norm": 2.207327365875244, "learning_rate": 6.994659762374967e-06, "loss": 0.14308372735977173, "step": 23130 }, { "epoch": 0.609542875770078, "grad_norm": 2.458028793334961, "learning_rate": 6.986523880469458e-06, "loss": 0.148895525932312, "step": 23140 }, { "epoch": 0.6098062910145767, "grad_norm": 1.6203421354293823, "learning_rate": 6.978390191901878e-06, "loss": 0.1674268960952759, "step": 23150 }, { "epoch": 0.6100697062590755, "grad_norm": 1.5597175359725952, "learning_rate": 6.970258702592277e-06, "loss": 0.15015215873718263, "step": 23160 }, { "epoch": 0.6103331215035742, "grad_norm": 2.037933826446533, "learning_rate": 6.962129418459109e-06, "loss": 0.14227330684661865, "step": 23170 }, { "epoch": 0.6105965367480729, "grad_norm": 1.3981856107711792, "learning_rate": 6.954002345419214e-06, "loss": 0.1406956672668457, "step": 23180 }, { "epoch": 0.6108599519925717, "grad_norm": 2.3642921447753906, "learning_rate": 6.945877489387829e-06, "loss": 0.15346566438674927, "step": 23190 }, { "epoch": 0.6111233672370704, "grad_norm": 1.2353862524032593, "learning_rate": 6.9377548562785735e-06, "loss": 0.1519009590148926, "step": 23200 }, { "epoch": 0.6113867824815692, "grad_norm": 3.2323830127716064, "learning_rate": 6.929634452003447e-06, "loss": 0.14658772945404053, "step": 23210 }, { "epoch": 0.6116501977260679, "grad_norm": 2.100294589996338, "learning_rate": 6.921516282472837e-06, "loss": 0.15594058036804198, "step": 23220 }, { "epoch": 0.6119136129705667, "grad_norm": 2.0103042125701904, "learning_rate": 6.91340035359549e-06, "loss": 0.15532379150390624, "step": 23230 }, { "epoch": 0.6121770282150654, "grad_norm": 1.3653777837753296, "learning_rate": 6.905286671278536e-06, "loss": 0.152099609375, "step": 23240 }, { "epoch": 0.6124404434595642, "grad_norm": 2.6993508338928223, "learning_rate": 6.897175241427462e-06, "loss": 0.13589770793914796, "step": 23250 }, { "epoch": 0.6127038587040629, "grad_norm": 1.6227701902389526, "learning_rate": 6.889066069946107e-06, "loss": 0.14953906536102296, "step": 23260 }, { "epoch": 0.6129672739485615, "grad_norm": 1.2748507261276245, "learning_rate": 6.880959162736691e-06, "loss": 0.14498871564865112, "step": 23270 }, { "epoch": 0.6132306891930603, "grad_norm": 1.9305678606033325, "learning_rate": 6.872854525699761e-06, "loss": 0.14380415678024291, "step": 23280 }, { "epoch": 0.613494104437559, "grad_norm": 2.230907917022705, "learning_rate": 6.8647521647342205e-06, "loss": 0.13420393466949462, "step": 23290 }, { "epoch": 0.6137575196820578, "grad_norm": 1.576470971107483, "learning_rate": 6.856652085737323e-06, "loss": 0.14136301279067992, "step": 23300 }, { "epoch": 0.6140209349265565, "grad_norm": 1.8301745653152466, "learning_rate": 6.848554294604652e-06, "loss": 0.1362842082977295, "step": 23310 }, { "epoch": 0.6142843501710553, "grad_norm": 1.220761775970459, "learning_rate": 6.8404587972301316e-06, "loss": 0.138979172706604, "step": 23320 }, { "epoch": 0.614547765415554, "grad_norm": 1.6911739110946655, "learning_rate": 6.8323655995060135e-06, "loss": 0.15891034603118898, "step": 23330 }, { "epoch": 0.6148111806600528, "grad_norm": 1.5462980270385742, "learning_rate": 6.824274707322874e-06, "loss": 0.15319013595581055, "step": 23340 }, { "epoch": 0.6150745959045515, "grad_norm": 2.3425402641296387, "learning_rate": 6.816186126569615e-06, "loss": 0.15574485063552856, "step": 23350 }, { "epoch": 0.6153380111490502, "grad_norm": 3.9082226753234863, "learning_rate": 6.808099863133453e-06, "loss": 0.15663607120513917, "step": 23360 }, { "epoch": 0.615601426393549, "grad_norm": 1.626569151878357, "learning_rate": 6.800015922899923e-06, "loss": 0.1458722472190857, "step": 23370 }, { "epoch": 0.6158648416380477, "grad_norm": 1.567157506942749, "learning_rate": 6.7919343117528655e-06, "loss": 0.14584171772003174, "step": 23380 }, { "epoch": 0.6161282568825465, "grad_norm": 2.7933273315429688, "learning_rate": 6.783855035574418e-06, "loss": 0.14873523712158204, "step": 23390 }, { "epoch": 0.6163916721270452, "grad_norm": 1.098400592803955, "learning_rate": 6.775778100245038e-06, "loss": 0.14740021228790284, "step": 23400 }, { "epoch": 0.616655087371544, "grad_norm": 2.442129373550415, "learning_rate": 6.76770351164346e-06, "loss": 0.1377851366996765, "step": 23410 }, { "epoch": 0.6169185026160426, "grad_norm": 4.196123123168945, "learning_rate": 6.759631275646715e-06, "loss": 0.14080123901367186, "step": 23420 }, { "epoch": 0.6171819178605413, "grad_norm": 2.560654878616333, "learning_rate": 6.751561398130133e-06, "loss": 0.1465640425682068, "step": 23430 }, { "epoch": 0.6174453331050401, "grad_norm": 2.331252336502075, "learning_rate": 6.743493884967312e-06, "loss": 0.15427942276000978, "step": 23440 }, { "epoch": 0.6177087483495388, "grad_norm": 2.090568780899048, "learning_rate": 6.735428742030141e-06, "loss": 0.13972575664520265, "step": 23450 }, { "epoch": 0.6179721635940376, "grad_norm": 2.196913242340088, "learning_rate": 6.7273659751887775e-06, "loss": 0.14496853351593017, "step": 23460 }, { "epoch": 0.6182355788385363, "grad_norm": 1.667687177658081, "learning_rate": 6.719305590311649e-06, "loss": 0.1487153172492981, "step": 23470 }, { "epoch": 0.6184989940830351, "grad_norm": 2.4606192111968994, "learning_rate": 6.7112475932654554e-06, "loss": 0.13766400814056395, "step": 23480 }, { "epoch": 0.6187624093275338, "grad_norm": 2.0081112384796143, "learning_rate": 6.70319198991515e-06, "loss": 0.14084045886993407, "step": 23490 }, { "epoch": 0.6190258245720326, "grad_norm": 2.3109567165374756, "learning_rate": 6.695138786123952e-06, "loss": 0.12763088941574097, "step": 23500 }, { "epoch": 0.6192892398165313, "grad_norm": 1.6922729015350342, "learning_rate": 6.687087987753331e-06, "loss": 0.13850779533386232, "step": 23510 }, { "epoch": 0.61955265506103, "grad_norm": 2.183448314666748, "learning_rate": 6.679039600663001e-06, "loss": 0.1582246780395508, "step": 23520 }, { "epoch": 0.6198160703055288, "grad_norm": 1.2797973155975342, "learning_rate": 6.670993630710929e-06, "loss": 0.13869116306304932, "step": 23530 }, { "epoch": 0.6200794855500275, "grad_norm": 2.7716598510742188, "learning_rate": 6.662950083753319e-06, "loss": 0.12669495344161988, "step": 23540 }, { "epoch": 0.6203429007945263, "grad_norm": 1.5875529050827026, "learning_rate": 6.654908965644609e-06, "loss": 0.15529999732971192, "step": 23550 }, { "epoch": 0.6206063160390249, "grad_norm": 2.0880656242370605, "learning_rate": 6.646870282237474e-06, "loss": 0.13606040477752684, "step": 23560 }, { "epoch": 0.6208697312835237, "grad_norm": 1.330278992652893, "learning_rate": 6.6388340393828125e-06, "loss": 0.1527197241783142, "step": 23570 }, { "epoch": 0.6211331465280224, "grad_norm": 2.2242891788482666, "learning_rate": 6.6308002429297516e-06, "loss": 0.13727577924728393, "step": 23580 }, { "epoch": 0.6213965617725212, "grad_norm": 1.5698277950286865, "learning_rate": 6.622768898725634e-06, "loss": 0.12996227741241456, "step": 23590 }, { "epoch": 0.6216599770170199, "grad_norm": 2.731976270675659, "learning_rate": 6.614740012616017e-06, "loss": 0.14647245407104492, "step": 23600 }, { "epoch": 0.6219233922615186, "grad_norm": 2.9593684673309326, "learning_rate": 6.606713590444673e-06, "loss": 0.13984596729278564, "step": 23610 }, { "epoch": 0.6221868075060174, "grad_norm": 2.712939500808716, "learning_rate": 6.598689638053577e-06, "loss": 0.1428668737411499, "step": 23620 }, { "epoch": 0.6224502227505161, "grad_norm": 1.7023111581802368, "learning_rate": 6.590668161282903e-06, "loss": 0.14748268127441405, "step": 23630 }, { "epoch": 0.6227136379950149, "grad_norm": 2.375697135925293, "learning_rate": 6.582649165971037e-06, "loss": 0.1390436887741089, "step": 23640 }, { "epoch": 0.6229770532395136, "grad_norm": 0.997202455997467, "learning_rate": 6.574632657954542e-06, "loss": 0.13634012937545775, "step": 23650 }, { "epoch": 0.6232404684840124, "grad_norm": 1.9453229904174805, "learning_rate": 6.5666186430681835e-06, "loss": 0.1380814790725708, "step": 23660 }, { "epoch": 0.6235038837285111, "grad_norm": 2.0675859451293945, "learning_rate": 6.558607127144905e-06, "loss": 0.13278427124023437, "step": 23670 }, { "epoch": 0.6237672989730099, "grad_norm": 2.067091226577759, "learning_rate": 6.550598116015832e-06, "loss": 0.13574492931365967, "step": 23680 }, { "epoch": 0.6240307142175086, "grad_norm": 4.17357873916626, "learning_rate": 6.54259161551027e-06, "loss": 0.15680614709854127, "step": 23690 }, { "epoch": 0.6242941294620072, "grad_norm": 1.6112396717071533, "learning_rate": 6.534587631455696e-06, "loss": 0.13937036991119384, "step": 23700 }, { "epoch": 0.624557544706506, "grad_norm": 1.4371219873428345, "learning_rate": 6.5265861696777556e-06, "loss": 0.15908024311065674, "step": 23710 }, { "epoch": 0.6248209599510047, "grad_norm": 3.6843886375427246, "learning_rate": 6.518587236000259e-06, "loss": 0.13803019523620605, "step": 23720 }, { "epoch": 0.6250843751955035, "grad_norm": 2.409865379333496, "learning_rate": 6.510590836245166e-06, "loss": 0.13017888069152833, "step": 23730 }, { "epoch": 0.6253477904400022, "grad_norm": 1.8937571048736572, "learning_rate": 6.502596976232616e-06, "loss": 0.1386413812637329, "step": 23740 }, { "epoch": 0.625611205684501, "grad_norm": 2.764737367630005, "learning_rate": 6.494605661780874e-06, "loss": 0.14245612621307374, "step": 23750 }, { "epoch": 0.6258746209289997, "grad_norm": 2.074209213256836, "learning_rate": 6.486616898706365e-06, "loss": 0.14888513088226318, "step": 23760 }, { "epoch": 0.6261380361734985, "grad_norm": 1.8255398273468018, "learning_rate": 6.47863069282366e-06, "loss": 0.14368681907653807, "step": 23770 }, { "epoch": 0.6264014514179972, "grad_norm": 1.3012917041778564, "learning_rate": 6.47064704994546e-06, "loss": 0.14696388244628905, "step": 23780 }, { "epoch": 0.6266648666624959, "grad_norm": 1.6147539615631104, "learning_rate": 6.462665975882607e-06, "loss": 0.14477112293243408, "step": 23790 }, { "epoch": 0.6269282819069947, "grad_norm": 2.2686665058135986, "learning_rate": 6.454687476444071e-06, "loss": 0.14158968925476073, "step": 23800 }, { "epoch": 0.6271916971514934, "grad_norm": 1.7398149967193604, "learning_rate": 6.446711557436944e-06, "loss": 0.12932660579681396, "step": 23810 }, { "epoch": 0.6274551123959922, "grad_norm": 1.92982017993927, "learning_rate": 6.438738224666452e-06, "loss": 0.14512816667556763, "step": 23820 }, { "epoch": 0.6277185276404909, "grad_norm": 1.6865075826644897, "learning_rate": 6.4307674839359225e-06, "loss": 0.14462382793426515, "step": 23830 }, { "epoch": 0.6279819428849897, "grad_norm": 2.326875686645508, "learning_rate": 6.42279934104681e-06, "loss": 0.1356255292892456, "step": 23840 }, { "epoch": 0.6282453581294883, "grad_norm": 2.434551477432251, "learning_rate": 6.4148338017986725e-06, "loss": 0.13498698472976683, "step": 23850 }, { "epoch": 0.6285087733739871, "grad_norm": 2.1972739696502686, "learning_rate": 6.406870871989166e-06, "loss": 0.12688057422637938, "step": 23860 }, { "epoch": 0.6287721886184858, "grad_norm": 1.6215006113052368, "learning_rate": 6.398910557414064e-06, "loss": 0.14225409030914307, "step": 23870 }, { "epoch": 0.6290356038629845, "grad_norm": 2.091836929321289, "learning_rate": 6.390952863867224e-06, "loss": 0.13115792274475097, "step": 23880 }, { "epoch": 0.6292990191074833, "grad_norm": 3.227696180343628, "learning_rate": 6.382997797140591e-06, "loss": 0.14984586238861083, "step": 23890 }, { "epoch": 0.629562434351982, "grad_norm": 1.404751181602478, "learning_rate": 6.375045363024217e-06, "loss": 0.1373382568359375, "step": 23900 }, { "epoch": 0.6298258495964808, "grad_norm": 2.9436731338500977, "learning_rate": 6.367095567306219e-06, "loss": 0.15281281471252442, "step": 23910 }, { "epoch": 0.6300892648409795, "grad_norm": 1.3317595720291138, "learning_rate": 6.359148415772805e-06, "loss": 0.11811977624893188, "step": 23920 }, { "epoch": 0.6303526800854783, "grad_norm": 2.9358339309692383, "learning_rate": 6.351203914208254e-06, "loss": 0.14760026931762696, "step": 23930 }, { "epoch": 0.630616095329977, "grad_norm": 1.1782221794128418, "learning_rate": 6.343262068394914e-06, "loss": 0.14985579252243042, "step": 23940 }, { "epoch": 0.6308795105744757, "grad_norm": 2.2007317543029785, "learning_rate": 6.335322884113208e-06, "loss": 0.15700275897979737, "step": 23950 }, { "epoch": 0.6311429258189745, "grad_norm": 2.296405553817749, "learning_rate": 6.3273863671416146e-06, "loss": 0.1429065465927124, "step": 23960 }, { "epoch": 0.6314063410634732, "grad_norm": 3.516213893890381, "learning_rate": 6.319452523256673e-06, "loss": 0.13798691034317018, "step": 23970 }, { "epoch": 0.631669756307972, "grad_norm": 1.592969536781311, "learning_rate": 6.31152135823298e-06, "loss": 0.14033071994781493, "step": 23980 }, { "epoch": 0.6319331715524706, "grad_norm": 1.2415523529052734, "learning_rate": 6.303592877843171e-06, "loss": 0.14063881635665892, "step": 23990 }, { "epoch": 0.6321965867969694, "grad_norm": 1.6163406372070312, "learning_rate": 6.295667087857948e-06, "loss": 0.14298993349075317, "step": 24000 }, { "epoch": 0.6324600020414681, "grad_norm": 3.2079951763153076, "learning_rate": 6.287743994046035e-06, "loss": 0.13583297729492189, "step": 24010 }, { "epoch": 0.6327234172859669, "grad_norm": 5.356350421905518, "learning_rate": 6.279823602174202e-06, "loss": 0.136150586605072, "step": 24020 }, { "epoch": 0.6329868325304656, "grad_norm": 3.903125524520874, "learning_rate": 6.271905918007255e-06, "loss": 0.15035332441329957, "step": 24030 }, { "epoch": 0.6332502477749643, "grad_norm": 2.131291151046753, "learning_rate": 6.263990947308022e-06, "loss": 0.14237618446350098, "step": 24040 }, { "epoch": 0.6335136630194631, "grad_norm": 1.2873637676239014, "learning_rate": 6.2560786958373645e-06, "loss": 0.1531292200088501, "step": 24050 }, { "epoch": 0.6337770782639618, "grad_norm": 1.6212061643600464, "learning_rate": 6.248169169354159e-06, "loss": 0.14522581100463866, "step": 24060 }, { "epoch": 0.6340404935084606, "grad_norm": 1.995742917060852, "learning_rate": 6.240262373615298e-06, "loss": 0.1392803192138672, "step": 24070 }, { "epoch": 0.6343039087529593, "grad_norm": 1.7642818689346313, "learning_rate": 6.232358314375693e-06, "loss": 0.1275707721710205, "step": 24080 }, { "epoch": 0.6345673239974581, "grad_norm": 1.9320789575576782, "learning_rate": 6.224456997388254e-06, "loss": 0.12924697399139404, "step": 24090 }, { "epoch": 0.6348307392419568, "grad_norm": 2.0490987300872803, "learning_rate": 6.2165584284039046e-06, "loss": 0.13881685733795165, "step": 24100 }, { "epoch": 0.6350941544864556, "grad_norm": 3.2747018337249756, "learning_rate": 6.208662613171566e-06, "loss": 0.13899784088134765, "step": 24110 }, { "epoch": 0.6353575697309543, "grad_norm": 3.357234477996826, "learning_rate": 6.200769557438144e-06, "loss": 0.14572958946228026, "step": 24120 }, { "epoch": 0.6356209849754529, "grad_norm": 1.2355059385299683, "learning_rate": 6.192879266948554e-06, "loss": 0.13770493268966674, "step": 24130 }, { "epoch": 0.6358844002199517, "grad_norm": 1.5859113931655884, "learning_rate": 6.184991747445686e-06, "loss": 0.14286887645721436, "step": 24140 }, { "epoch": 0.6361478154644504, "grad_norm": 2.7184414863586426, "learning_rate": 6.1771070046704165e-06, "loss": 0.1363132953643799, "step": 24150 }, { "epoch": 0.6364112307089492, "grad_norm": 1.6516433954238892, "learning_rate": 6.169225044361604e-06, "loss": 0.14798818826675414, "step": 24160 }, { "epoch": 0.6366746459534479, "grad_norm": 3.183387279510498, "learning_rate": 6.161345872256075e-06, "loss": 0.14393315315246583, "step": 24170 }, { "epoch": 0.6369380611979467, "grad_norm": 1.8947347402572632, "learning_rate": 6.153469494088638e-06, "loss": 0.14006675481796266, "step": 24180 }, { "epoch": 0.6372014764424454, "grad_norm": 2.223742961883545, "learning_rate": 6.1455959155920555e-06, "loss": 0.1355870008468628, "step": 24190 }, { "epoch": 0.6374648916869442, "grad_norm": 2.3586671352386475, "learning_rate": 6.1377251424970555e-06, "loss": 0.13816308975219727, "step": 24200 }, { "epoch": 0.6377283069314429, "grad_norm": 2.4537642002105713, "learning_rate": 6.1298571805323344e-06, "loss": 0.13764294385910034, "step": 24210 }, { "epoch": 0.6379917221759416, "grad_norm": 2.2376158237457275, "learning_rate": 6.121992035424529e-06, "loss": 0.1319904088973999, "step": 24220 }, { "epoch": 0.6382551374204404, "grad_norm": 2.277024030685425, "learning_rate": 6.114129712898227e-06, "loss": 0.14942370653152465, "step": 24230 }, { "epoch": 0.6385185526649391, "grad_norm": 1.859134554862976, "learning_rate": 6.106270218675975e-06, "loss": 0.13797614574432374, "step": 24240 }, { "epoch": 0.6387819679094379, "grad_norm": 2.387338161468506, "learning_rate": 6.098413558478244e-06, "loss": 0.1454103708267212, "step": 24250 }, { "epoch": 0.6390453831539366, "grad_norm": 1.5031404495239258, "learning_rate": 6.090559738023456e-06, "loss": 0.13996398448944092, "step": 24260 }, { "epoch": 0.6393087983984354, "grad_norm": 2.693175792694092, "learning_rate": 6.0827087630279605e-06, "loss": 0.14362845420837403, "step": 24270 }, { "epoch": 0.639572213642934, "grad_norm": 1.9359045028686523, "learning_rate": 6.074860639206028e-06, "loss": 0.13666350841522218, "step": 24280 }, { "epoch": 0.6398356288874328, "grad_norm": 1.111184000968933, "learning_rate": 6.067015372269872e-06, "loss": 0.1426556944847107, "step": 24290 }, { "epoch": 0.6400990441319315, "grad_norm": 1.8732831478118896, "learning_rate": 6.059172967929609e-06, "loss": 0.1377884864807129, "step": 24300 }, { "epoch": 0.6403624593764302, "grad_norm": 1.2383121252059937, "learning_rate": 6.0513334318932825e-06, "loss": 0.12896674871444702, "step": 24310 }, { "epoch": 0.640625874620929, "grad_norm": 2.0533852577209473, "learning_rate": 6.043496769866848e-06, "loss": 0.14538594484329223, "step": 24320 }, { "epoch": 0.6408892898654277, "grad_norm": 1.3838186264038086, "learning_rate": 6.035662987554154e-06, "loss": 0.13979089260101318, "step": 24330 }, { "epoch": 0.6411527051099265, "grad_norm": 2.3755087852478027, "learning_rate": 6.027832090656982e-06, "loss": 0.1446695566177368, "step": 24340 }, { "epoch": 0.6414161203544252, "grad_norm": 1.5609709024429321, "learning_rate": 6.020004084874984e-06, "loss": 0.14032304286956787, "step": 24350 }, { "epoch": 0.641679535598924, "grad_norm": 1.360190510749817, "learning_rate": 6.012178975905725e-06, "loss": 0.14182969331741332, "step": 24360 }, { "epoch": 0.6419429508434227, "grad_norm": 2.1627283096313477, "learning_rate": 6.0043567694446566e-06, "loss": 0.12609782218933105, "step": 24370 }, { "epoch": 0.6422063660879215, "grad_norm": 2.6158666610717773, "learning_rate": 5.9965374711851175e-06, "loss": 0.14132137298583985, "step": 24380 }, { "epoch": 0.6424697813324202, "grad_norm": 1.1752569675445557, "learning_rate": 5.9887210868183325e-06, "loss": 0.15071251392364501, "step": 24390 }, { "epoch": 0.6427331965769189, "grad_norm": 1.6251379251480103, "learning_rate": 5.9809076220334024e-06, "loss": 0.1358732223510742, "step": 24400 }, { "epoch": 0.6429966118214177, "grad_norm": 1.9536516666412354, "learning_rate": 5.973097082517301e-06, "loss": 0.1399837851524353, "step": 24410 }, { "epoch": 0.6432600270659163, "grad_norm": 3.650721311569214, "learning_rate": 5.965289473954879e-06, "loss": 0.12602134943008422, "step": 24420 }, { "epoch": 0.6435234423104151, "grad_norm": 2.430985450744629, "learning_rate": 5.95748480202885e-06, "loss": 0.1431636095046997, "step": 24430 }, { "epoch": 0.6437868575549138, "grad_norm": 2.592683792114258, "learning_rate": 5.949683072419792e-06, "loss": 0.13545247316360473, "step": 24440 }, { "epoch": 0.6440502727994126, "grad_norm": 2.458422899246216, "learning_rate": 5.941884290806142e-06, "loss": 0.13385404348373414, "step": 24450 }, { "epoch": 0.6443136880439113, "grad_norm": 1.8012207746505737, "learning_rate": 5.934088462864181e-06, "loss": 0.1436646580696106, "step": 24460 }, { "epoch": 0.6445771032884101, "grad_norm": 2.5120553970336914, "learning_rate": 5.926295594268059e-06, "loss": 0.14087704420089722, "step": 24470 }, { "epoch": 0.6448405185329088, "grad_norm": 2.5149483680725098, "learning_rate": 5.918505690689759e-06, "loss": 0.131751549243927, "step": 24480 }, { "epoch": 0.6451039337774075, "grad_norm": 2.564323902130127, "learning_rate": 5.910718757799105e-06, "loss": 0.1435558319091797, "step": 24490 }, { "epoch": 0.6453673490219063, "grad_norm": 2.10391902923584, "learning_rate": 5.902934801263766e-06, "loss": 0.14759451150894165, "step": 24500 }, { "epoch": 0.645630764266405, "grad_norm": 1.4032996892929077, "learning_rate": 5.89515382674924e-06, "loss": 0.1310071349143982, "step": 24510 }, { "epoch": 0.6458941795109038, "grad_norm": 1.8643888235092163, "learning_rate": 5.88737583991886e-06, "loss": 0.16191158294677735, "step": 24520 }, { "epoch": 0.6461575947554025, "grad_norm": 1.793576955795288, "learning_rate": 5.879600846433777e-06, "loss": 0.13666512966156005, "step": 24530 }, { "epoch": 0.6464210099999013, "grad_norm": 2.3413326740264893, "learning_rate": 5.871828851952965e-06, "loss": 0.14252132177352905, "step": 24540 }, { "epoch": 0.6466844252444, "grad_norm": 2.182408571243286, "learning_rate": 5.864059862133221e-06, "loss": 0.13728630542755127, "step": 24550 }, { "epoch": 0.6469478404888986, "grad_norm": 1.1397889852523804, "learning_rate": 5.856293882629147e-06, "loss": 0.12930960655212403, "step": 24560 }, { "epoch": 0.6472112557333974, "grad_norm": 2.6330525875091553, "learning_rate": 5.848530919093162e-06, "loss": 0.14287929534912108, "step": 24570 }, { "epoch": 0.6474746709778961, "grad_norm": 1.6525115966796875, "learning_rate": 5.840770977175485e-06, "loss": 0.13693615198135375, "step": 24580 }, { "epoch": 0.6477380862223949, "grad_norm": 2.4851410388946533, "learning_rate": 5.833014062524129e-06, "loss": 0.13842182159423827, "step": 24590 }, { "epoch": 0.6480015014668936, "grad_norm": 1.529087781906128, "learning_rate": 5.825260180784923e-06, "loss": 0.13602795600891113, "step": 24600 }, { "epoch": 0.6482649167113924, "grad_norm": 2.8831000328063965, "learning_rate": 5.817509337601475e-06, "loss": 0.1435348153114319, "step": 24610 }, { "epoch": 0.6485283319558911, "grad_norm": 2.0978965759277344, "learning_rate": 5.8097615386151686e-06, "loss": 0.1445143699645996, "step": 24620 }, { "epoch": 0.6487917472003899, "grad_norm": 2.4543955326080322, "learning_rate": 5.8020167894652035e-06, "loss": 0.13830976486206054, "step": 24630 }, { "epoch": 0.6490551624448886, "grad_norm": 2.0738120079040527, "learning_rate": 5.79427509578853e-06, "loss": 0.14274550676345826, "step": 24640 }, { "epoch": 0.6493185776893873, "grad_norm": 2.006953477859497, "learning_rate": 5.786536463219899e-06, "loss": 0.13835721015930175, "step": 24650 }, { "epoch": 0.6495819929338861, "grad_norm": 1.4700852632522583, "learning_rate": 5.778800897391812e-06, "loss": 0.12859160900115968, "step": 24660 }, { "epoch": 0.6498454081783848, "grad_norm": 1.3009859323501587, "learning_rate": 5.771068403934545e-06, "loss": 0.1469485878944397, "step": 24670 }, { "epoch": 0.6501088234228836, "grad_norm": 1.5310999155044556, "learning_rate": 5.76333898847615e-06, "loss": 0.13461849689483643, "step": 24680 }, { "epoch": 0.6503722386673823, "grad_norm": 1.607694387435913, "learning_rate": 5.755612656642424e-06, "loss": 0.13181931972503663, "step": 24690 }, { "epoch": 0.650635653911881, "grad_norm": 3.114487886428833, "learning_rate": 5.747889414056925e-06, "loss": 0.1354024291038513, "step": 24700 }, { "epoch": 0.6508990691563797, "grad_norm": 1.4951481819152832, "learning_rate": 5.740169266340963e-06, "loss": 0.12994004487991334, "step": 24710 }, { "epoch": 0.6511624844008785, "grad_norm": 1.2941431999206543, "learning_rate": 5.732452219113589e-06, "loss": 0.13842320442199707, "step": 24720 }, { "epoch": 0.6514258996453772, "grad_norm": 3.0678212642669678, "learning_rate": 5.724738277991611e-06, "loss": 0.1332806944847107, "step": 24730 }, { "epoch": 0.6516893148898759, "grad_norm": 2.051542282104492, "learning_rate": 5.717027448589568e-06, "loss": 0.15206483602523804, "step": 24740 }, { "epoch": 0.6519527301343747, "grad_norm": 1.2322791814804077, "learning_rate": 5.709319736519725e-06, "loss": 0.12868866920471192, "step": 24750 }, { "epoch": 0.6522161453788734, "grad_norm": 1.669326663017273, "learning_rate": 5.701615147392096e-06, "loss": 0.1474974513053894, "step": 24760 }, { "epoch": 0.6524795606233722, "grad_norm": 1.6498825550079346, "learning_rate": 5.6939136868144075e-06, "loss": 0.14312105178833007, "step": 24770 }, { "epoch": 0.6527429758678709, "grad_norm": 1.7681376934051514, "learning_rate": 5.686215360392127e-06, "loss": 0.12734408378601075, "step": 24780 }, { "epoch": 0.6530063911123697, "grad_norm": 6.892552852630615, "learning_rate": 5.678520173728414e-06, "loss": 0.14106919765472412, "step": 24790 }, { "epoch": 0.6532698063568684, "grad_norm": 2.445577383041382, "learning_rate": 5.67082813242416e-06, "loss": 0.1446218252182007, "step": 24800 }, { "epoch": 0.6535332216013672, "grad_norm": 1.292442798614502, "learning_rate": 5.663139242077971e-06, "loss": 0.14161331653594972, "step": 24810 }, { "epoch": 0.6537966368458659, "grad_norm": 1.532670497894287, "learning_rate": 5.6554535082861496e-06, "loss": 0.15386171340942384, "step": 24820 }, { "epoch": 0.6540600520903646, "grad_norm": 2.02508282661438, "learning_rate": 5.647770936642705e-06, "loss": 0.12927980422973634, "step": 24830 }, { "epoch": 0.6543234673348634, "grad_norm": 2.9284443855285645, "learning_rate": 5.6400915327393435e-06, "loss": 0.137223482131958, "step": 24840 }, { "epoch": 0.654586882579362, "grad_norm": 1.4343067407608032, "learning_rate": 5.632415302165464e-06, "loss": 0.14281172752380372, "step": 24850 }, { "epoch": 0.6548502978238608, "grad_norm": 2.098095655441284, "learning_rate": 5.624742250508164e-06, "loss": 0.1278897762298584, "step": 24860 }, { "epoch": 0.6551137130683595, "grad_norm": 2.083678722381592, "learning_rate": 5.617072383352222e-06, "loss": 0.14089593887329102, "step": 24870 }, { "epoch": 0.6553771283128583, "grad_norm": 1.3689844608306885, "learning_rate": 5.6094057062800886e-06, "loss": 0.13246314525604247, "step": 24880 }, { "epoch": 0.655640543557357, "grad_norm": 1.596967101097107, "learning_rate": 5.601742224871911e-06, "loss": 0.15335050821304322, "step": 24890 }, { "epoch": 0.6559039588018558, "grad_norm": 3.732055425643921, "learning_rate": 5.594081944705496e-06, "loss": 0.14060747623443604, "step": 24900 }, { "epoch": 0.6561673740463545, "grad_norm": 1.8445812463760376, "learning_rate": 5.586424871356337e-06, "loss": 0.13375011682510377, "step": 24910 }, { "epoch": 0.6564307892908532, "grad_norm": 1.7072690725326538, "learning_rate": 5.578771010397572e-06, "loss": 0.13712334632873535, "step": 24920 }, { "epoch": 0.656694204535352, "grad_norm": 2.006814479827881, "learning_rate": 5.571120367400008e-06, "loss": 0.13767836093902588, "step": 24930 }, { "epoch": 0.6569576197798507, "grad_norm": 3.47794246673584, "learning_rate": 5.563472947932123e-06, "loss": 0.13498804569244385, "step": 24940 }, { "epoch": 0.6572210350243495, "grad_norm": 2.0669546127319336, "learning_rate": 5.555828757560035e-06, "loss": 0.1392390489578247, "step": 24950 }, { "epoch": 0.6574844502688482, "grad_norm": 1.524154543876648, "learning_rate": 5.548187801847513e-06, "loss": 0.1392725706100464, "step": 24960 }, { "epoch": 0.657747865513347, "grad_norm": 1.3643351793289185, "learning_rate": 5.5405500863559795e-06, "loss": 0.14173531532287598, "step": 24970 }, { "epoch": 0.6580112807578457, "grad_norm": 2.1354923248291016, "learning_rate": 5.532915616644484e-06, "loss": 0.13765509128570558, "step": 24980 }, { "epoch": 0.6582746960023445, "grad_norm": 2.562069892883301, "learning_rate": 5.525284398269733e-06, "loss": 0.12189801931381225, "step": 24990 }, { "epoch": 0.6585381112468431, "grad_norm": 2.5067970752716064, "learning_rate": 5.517656436786059e-06, "loss": 0.13712718486785888, "step": 25000 }, { "epoch": 0.6588015264913418, "grad_norm": 3.9140830039978027, "learning_rate": 5.510031737745405e-06, "loss": 0.13904058933258057, "step": 25010 }, { "epoch": 0.6590649417358406, "grad_norm": 2.223228931427002, "learning_rate": 5.50241030669737e-06, "loss": 0.1390578031539917, "step": 25020 }, { "epoch": 0.6593283569803393, "grad_norm": 2.2085390090942383, "learning_rate": 5.4947921491891565e-06, "loss": 0.13715927600860595, "step": 25030 }, { "epoch": 0.6595917722248381, "grad_norm": 1.3406833410263062, "learning_rate": 5.487177270765588e-06, "loss": 0.14008901119232178, "step": 25040 }, { "epoch": 0.6598551874693368, "grad_norm": 1.5470160245895386, "learning_rate": 5.479565676969101e-06, "loss": 0.12719330787658692, "step": 25050 }, { "epoch": 0.6601186027138356, "grad_norm": 1.5640532970428467, "learning_rate": 5.47195737333974e-06, "loss": 0.1311184883117676, "step": 25060 }, { "epoch": 0.6603820179583343, "grad_norm": 1.0065288543701172, "learning_rate": 5.464352365415162e-06, "loss": 0.1368672490119934, "step": 25070 }, { "epoch": 0.660645433202833, "grad_norm": 1.556189775466919, "learning_rate": 5.456750658730616e-06, "loss": 0.14450427293777465, "step": 25080 }, { "epoch": 0.6609088484473318, "grad_norm": 1.3270645141601562, "learning_rate": 5.449152258818955e-06, "loss": 0.13646650314331055, "step": 25090 }, { "epoch": 0.6611722636918305, "grad_norm": 2.956058979034424, "learning_rate": 5.4415571712106184e-06, "loss": 0.13952697515487672, "step": 25100 }, { "epoch": 0.6614356789363293, "grad_norm": 1.3919132947921753, "learning_rate": 5.433965401433639e-06, "loss": 0.12452958822250366, "step": 25110 }, { "epoch": 0.661699094180828, "grad_norm": 1.4824289083480835, "learning_rate": 5.426376955013641e-06, "loss": 0.13600215911865235, "step": 25120 }, { "epoch": 0.6619625094253268, "grad_norm": 2.137773036956787, "learning_rate": 5.4187918374738154e-06, "loss": 0.14577937126159668, "step": 25130 }, { "epoch": 0.6622259246698254, "grad_norm": 1.1034185886383057, "learning_rate": 5.4112100543349365e-06, "loss": 0.12682180404663085, "step": 25140 }, { "epoch": 0.6624893399143242, "grad_norm": 1.8931728601455688, "learning_rate": 5.403631611115362e-06, "loss": 0.13703978061676025, "step": 25150 }, { "epoch": 0.6627527551588229, "grad_norm": 2.114881992340088, "learning_rate": 5.396056513331005e-06, "loss": 0.12870125770568847, "step": 25160 }, { "epoch": 0.6630161704033216, "grad_norm": 2.2048816680908203, "learning_rate": 5.388484766495348e-06, "loss": 0.15284996032714843, "step": 25170 }, { "epoch": 0.6632795856478204, "grad_norm": 2.382073402404785, "learning_rate": 5.3809163761194364e-06, "loss": 0.1438948392868042, "step": 25180 }, { "epoch": 0.6635430008923191, "grad_norm": 1.7130136489868164, "learning_rate": 5.373351347711865e-06, "loss": 0.11740198135375976, "step": 25190 }, { "epoch": 0.6638064161368179, "grad_norm": 2.4459071159362793, "learning_rate": 5.365789686778797e-06, "loss": 0.14121490716934204, "step": 25200 }, { "epoch": 0.6640698313813166, "grad_norm": 1.8445204496383667, "learning_rate": 5.358231398823937e-06, "loss": 0.13155869245529175, "step": 25210 }, { "epoch": 0.6643332466258154, "grad_norm": 2.4096808433532715, "learning_rate": 5.350676489348521e-06, "loss": 0.13767975568771362, "step": 25220 }, { "epoch": 0.6645966618703141, "grad_norm": 2.2196686267852783, "learning_rate": 5.3431249638513475e-06, "loss": 0.15240572690963744, "step": 25230 }, { "epoch": 0.6648600771148129, "grad_norm": 1.6567013263702393, "learning_rate": 5.335576827828737e-06, "loss": 0.13571717739105224, "step": 25240 }, { "epoch": 0.6651234923593116, "grad_norm": 3.3091819286346436, "learning_rate": 5.328032086774557e-06, "loss": 0.13666675090789795, "step": 25250 }, { "epoch": 0.6653869076038103, "grad_norm": 1.9754585027694702, "learning_rate": 5.320490746180189e-06, "loss": 0.12664808034896852, "step": 25260 }, { "epoch": 0.665650322848309, "grad_norm": 2.9509494304656982, "learning_rate": 5.312952811534542e-06, "loss": 0.14645323753356934, "step": 25270 }, { "epoch": 0.6659137380928077, "grad_norm": 1.9967987537384033, "learning_rate": 5.305418288324058e-06, "loss": 0.13503297567367553, "step": 25280 }, { "epoch": 0.6661771533373065, "grad_norm": 2.1161985397338867, "learning_rate": 5.297887182032686e-06, "loss": 0.13783122301101686, "step": 25290 }, { "epoch": 0.6664405685818052, "grad_norm": 2.523787498474121, "learning_rate": 5.29035949814189e-06, "loss": 0.13038008213043212, "step": 25300 }, { "epoch": 0.666703983826304, "grad_norm": 2.2408738136291504, "learning_rate": 5.2828352421306415e-06, "loss": 0.14041075706481934, "step": 25310 }, { "epoch": 0.6669673990708027, "grad_norm": 1.296110987663269, "learning_rate": 5.275314419475417e-06, "loss": 0.1284629225730896, "step": 25320 }, { "epoch": 0.6672308143153015, "grad_norm": 2.226987838745117, "learning_rate": 5.267797035650202e-06, "loss": 0.12653623819351195, "step": 25330 }, { "epoch": 0.6674942295598002, "grad_norm": 1.639304757118225, "learning_rate": 5.2602830961264704e-06, "loss": 0.13569397926330568, "step": 25340 }, { "epoch": 0.6677576448042989, "grad_norm": 2.472886800765991, "learning_rate": 5.252772606373186e-06, "loss": 0.13013051748275756, "step": 25350 }, { "epoch": 0.6680210600487977, "grad_norm": 1.993100881576538, "learning_rate": 5.245265571856812e-06, "loss": 0.120896315574646, "step": 25360 }, { "epoch": 0.6682844752932964, "grad_norm": 2.3490560054779053, "learning_rate": 5.237761998041291e-06, "loss": 0.1349652051925659, "step": 25370 }, { "epoch": 0.6685478905377952, "grad_norm": 1.4415194988250732, "learning_rate": 5.230261890388053e-06, "loss": 0.1443002700805664, "step": 25380 }, { "epoch": 0.6688113057822939, "grad_norm": 2.239523410797119, "learning_rate": 5.2227652543559924e-06, "loss": 0.13099520206451415, "step": 25390 }, { "epoch": 0.6690747210267927, "grad_norm": 1.9202672243118286, "learning_rate": 5.215272095401483e-06, "loss": 0.13593021631240845, "step": 25400 }, { "epoch": 0.6693381362712914, "grad_norm": 1.7920634746551514, "learning_rate": 5.207782418978378e-06, "loss": 0.13524483442306517, "step": 25410 }, { "epoch": 0.6696015515157901, "grad_norm": 1.9791600704193115, "learning_rate": 5.200296230537982e-06, "loss": 0.13560060262680054, "step": 25420 }, { "epoch": 0.6698649667602888, "grad_norm": 1.482654333114624, "learning_rate": 5.192813535529066e-06, "loss": 0.13864212036132811, "step": 25430 }, { "epoch": 0.6701283820047875, "grad_norm": 1.6013976335525513, "learning_rate": 5.1853343393978585e-06, "loss": 0.12602567672729492, "step": 25440 }, { "epoch": 0.6703917972492863, "grad_norm": 2.1127374172210693, "learning_rate": 5.177858647588037e-06, "loss": 0.1364603877067566, "step": 25450 }, { "epoch": 0.670655212493785, "grad_norm": 1.3345807790756226, "learning_rate": 5.170386465540741e-06, "loss": 0.12252075672149658, "step": 25460 }, { "epoch": 0.6709186277382838, "grad_norm": 2.1149513721466064, "learning_rate": 5.162917798694546e-06, "loss": 0.13596595525741578, "step": 25470 }, { "epoch": 0.6711820429827825, "grad_norm": 1.3194674253463745, "learning_rate": 5.15545265248546e-06, "loss": 0.14481351375579835, "step": 25480 }, { "epoch": 0.6714454582272813, "grad_norm": 1.8065658807754517, "learning_rate": 5.147991032346948e-06, "loss": 0.11773269176483155, "step": 25490 }, { "epoch": 0.67170887347178, "grad_norm": 1.7099649906158447, "learning_rate": 5.140532943709893e-06, "loss": 0.13771796226501465, "step": 25500 }, { "epoch": 0.6719722887162788, "grad_norm": 1.788379430770874, "learning_rate": 5.133078392002623e-06, "loss": 0.13616960048675536, "step": 25510 }, { "epoch": 0.6722357039607775, "grad_norm": 1.667105793952942, "learning_rate": 5.125627382650873e-06, "loss": 0.13518099784851073, "step": 25520 }, { "epoch": 0.6724991192052762, "grad_norm": 3.4110164642333984, "learning_rate": 5.118179921077808e-06, "loss": 0.14128880500793456, "step": 25530 }, { "epoch": 0.672762534449775, "grad_norm": 2.19138765335083, "learning_rate": 5.1107360127040206e-06, "loss": 0.1386500358581543, "step": 25540 }, { "epoch": 0.6730259496942737, "grad_norm": 2.3307747840881348, "learning_rate": 5.103295662947504e-06, "loss": 0.13577741384506226, "step": 25550 }, { "epoch": 0.6732893649387724, "grad_norm": 3.5931525230407715, "learning_rate": 5.095858877223665e-06, "loss": 0.12658811807632447, "step": 25560 }, { "epoch": 0.6735527801832711, "grad_norm": 1.5469611883163452, "learning_rate": 5.088425660945319e-06, "loss": 0.1361326217651367, "step": 25570 }, { "epoch": 0.6738161954277699, "grad_norm": 1.3560559749603271, "learning_rate": 5.080996019522675e-06, "loss": 0.12270870208740234, "step": 25580 }, { "epoch": 0.6740796106722686, "grad_norm": 3.39471435546875, "learning_rate": 5.073569958363356e-06, "loss": 0.13709229230880737, "step": 25590 }, { "epoch": 0.6743430259167673, "grad_norm": 1.816014051437378, "learning_rate": 5.06614748287237e-06, "loss": 0.12190659046173095, "step": 25600 }, { "epoch": 0.6746064411612661, "grad_norm": 2.186880350112915, "learning_rate": 5.058728598452103e-06, "loss": 0.14012426137924194, "step": 25610 }, { "epoch": 0.6748698564057648, "grad_norm": 2.3255326747894287, "learning_rate": 5.05131331050235e-06, "loss": 0.12842627763748168, "step": 25620 }, { "epoch": 0.6751332716502636, "grad_norm": 1.5470114946365356, "learning_rate": 5.043901624420273e-06, "loss": 0.14733710289001464, "step": 25630 }, { "epoch": 0.6753966868947623, "grad_norm": 3.183274269104004, "learning_rate": 5.036493545600421e-06, "loss": 0.1320556879043579, "step": 25640 }, { "epoch": 0.6756601021392611, "grad_norm": 1.3753634691238403, "learning_rate": 5.029089079434708e-06, "loss": 0.12760488986968993, "step": 25650 }, { "epoch": 0.6759235173837598, "grad_norm": 2.161471366882324, "learning_rate": 5.021688231312423e-06, "loss": 0.12034705877304078, "step": 25660 }, { "epoch": 0.6761869326282586, "grad_norm": 1.490592360496521, "learning_rate": 5.014291006620229e-06, "loss": 0.12764570713043213, "step": 25670 }, { "epoch": 0.6764503478727573, "grad_norm": 1.4033888578414917, "learning_rate": 5.006897410742142e-06, "loss": 0.1296623468399048, "step": 25680 }, { "epoch": 0.676713763117256, "grad_norm": 2.395350694656372, "learning_rate": 4.999507449059542e-06, "loss": 0.12172707319259643, "step": 25690 }, { "epoch": 0.6769771783617548, "grad_norm": 1.962580919265747, "learning_rate": 4.992121126951159e-06, "loss": 0.13791619539260863, "step": 25700 }, { "epoch": 0.6772405936062534, "grad_norm": 2.1027824878692627, "learning_rate": 4.984738449793074e-06, "loss": 0.13880865573883056, "step": 25710 }, { "epoch": 0.6775040088507522, "grad_norm": 1.8158479928970337, "learning_rate": 4.977359422958731e-06, "loss": 0.13268171548843383, "step": 25720 }, { "epoch": 0.6777674240952509, "grad_norm": 1.471930742263794, "learning_rate": 4.96998405181889e-06, "loss": 0.14149153232574463, "step": 25730 }, { "epoch": 0.6780308393397497, "grad_norm": 2.0645861625671387, "learning_rate": 4.962612341741666e-06, "loss": 0.12129881381988525, "step": 25740 }, { "epoch": 0.6782942545842484, "grad_norm": 2.2710297107696533, "learning_rate": 4.955244298092514e-06, "loss": 0.14076365232467652, "step": 25750 }, { "epoch": 0.6785576698287472, "grad_norm": 1.742219090461731, "learning_rate": 4.947879926234212e-06, "loss": 0.14165738821029664, "step": 25760 }, { "epoch": 0.6788210850732459, "grad_norm": 2.1872456073760986, "learning_rate": 4.940519231526867e-06, "loss": 0.13372431993484496, "step": 25770 }, { "epoch": 0.6790845003177446, "grad_norm": 1.8838437795639038, "learning_rate": 4.933162219327909e-06, "loss": 0.12451380491256714, "step": 25780 }, { "epoch": 0.6793479155622434, "grad_norm": 2.141467571258545, "learning_rate": 4.925808894992087e-06, "loss": 0.11885191202163696, "step": 25790 }, { "epoch": 0.6796113308067421, "grad_norm": 1.55570387840271, "learning_rate": 4.918459263871474e-06, "loss": 0.13123589754104614, "step": 25800 }, { "epoch": 0.6798747460512409, "grad_norm": 2.6305525302886963, "learning_rate": 4.911113331315448e-06, "loss": 0.11886521577835082, "step": 25810 }, { "epoch": 0.6801381612957396, "grad_norm": 3.4264471530914307, "learning_rate": 4.903771102670686e-06, "loss": 0.1282457709312439, "step": 25820 }, { "epoch": 0.6804015765402384, "grad_norm": 1.9046764373779297, "learning_rate": 4.896432583281189e-06, "loss": 0.12643213272094728, "step": 25830 }, { "epoch": 0.680664991784737, "grad_norm": 2.2594974040985107, "learning_rate": 4.88909777848824e-06, "loss": 0.12926957607269288, "step": 25840 }, { "epoch": 0.6809284070292358, "grad_norm": 2.486872434616089, "learning_rate": 4.881766693630439e-06, "loss": 0.145823073387146, "step": 25850 }, { "epoch": 0.6811918222737345, "grad_norm": 1.9431557655334473, "learning_rate": 4.874439334043654e-06, "loss": 0.12455518245697021, "step": 25860 }, { "epoch": 0.6814552375182332, "grad_norm": 1.6216596364974976, "learning_rate": 4.8671157050610525e-06, "loss": 0.13374794721603395, "step": 25870 }, { "epoch": 0.681718652762732, "grad_norm": 2.6359140872955322, "learning_rate": 4.859795812013099e-06, "loss": 0.14651904106140137, "step": 25880 }, { "epoch": 0.6819820680072307, "grad_norm": 2.0689549446105957, "learning_rate": 4.852479660227518e-06, "loss": 0.12965573072433473, "step": 25890 }, { "epoch": 0.6822454832517295, "grad_norm": 2.830522060394287, "learning_rate": 4.8451672550293236e-06, "loss": 0.11456241607666015, "step": 25900 }, { "epoch": 0.6825088984962282, "grad_norm": 2.256282329559326, "learning_rate": 4.837858601740799e-06, "loss": 0.12883870601654052, "step": 25910 }, { "epoch": 0.682772313740727, "grad_norm": 2.0426430702209473, "learning_rate": 4.830553705681496e-06, "loss": 0.12662856578826903, "step": 25920 }, { "epoch": 0.6830357289852257, "grad_norm": 2.0198559761047363, "learning_rate": 4.823252572168236e-06, "loss": 0.1317821979522705, "step": 25930 }, { "epoch": 0.6832991442297245, "grad_norm": 1.1020013093948364, "learning_rate": 4.815955206515104e-06, "loss": 0.12649459838867189, "step": 25940 }, { "epoch": 0.6835625594742232, "grad_norm": 3.48006010055542, "learning_rate": 4.808661614033422e-06, "loss": 0.13034363985061645, "step": 25950 }, { "epoch": 0.6838259747187219, "grad_norm": 1.337556004524231, "learning_rate": 4.801371800031796e-06, "loss": 0.1374469518661499, "step": 25960 }, { "epoch": 0.6840893899632207, "grad_norm": 1.6626062393188477, "learning_rate": 4.794085769816057e-06, "loss": 0.11562528610229492, "step": 25970 }, { "epoch": 0.6843528052077194, "grad_norm": 2.1965129375457764, "learning_rate": 4.786803528689304e-06, "loss": 0.14105262756347656, "step": 25980 }, { "epoch": 0.6846162204522181, "grad_norm": 3.7374815940856934, "learning_rate": 4.779525081951857e-06, "loss": 0.13755416870117188, "step": 25990 }, { "epoch": 0.6848796356967168, "grad_norm": 2.03157377243042, "learning_rate": 4.772250434901281e-06, "loss": 0.12089352607727051, "step": 26000 }, { "epoch": 0.6851430509412156, "grad_norm": 2.4877607822418213, "learning_rate": 4.7649795928323875e-06, "loss": 0.1196978211402893, "step": 26010 }, { "epoch": 0.6854064661857143, "grad_norm": 1.9740869998931885, "learning_rate": 4.757712561037203e-06, "loss": 0.1311385750770569, "step": 26020 }, { "epoch": 0.6856698814302131, "grad_norm": 2.6603598594665527, "learning_rate": 4.750449344804989e-06, "loss": 0.12382135391235352, "step": 26030 }, { "epoch": 0.6859332966747118, "grad_norm": 5.300684452056885, "learning_rate": 4.743189949422227e-06, "loss": 0.14582890272140503, "step": 26040 }, { "epoch": 0.6861967119192105, "grad_norm": 2.3901541233062744, "learning_rate": 4.735934380172613e-06, "loss": 0.11994497776031494, "step": 26050 }, { "epoch": 0.6864601271637093, "grad_norm": 2.8389313220977783, "learning_rate": 4.728682642337072e-06, "loss": 0.12890408039093018, "step": 26060 }, { "epoch": 0.686723542408208, "grad_norm": 1.8910671472549438, "learning_rate": 4.7214347411937325e-06, "loss": 0.1392799735069275, "step": 26070 }, { "epoch": 0.6869869576527068, "grad_norm": 1.3072376251220703, "learning_rate": 4.714190682017918e-06, "loss": 0.13572986125946046, "step": 26080 }, { "epoch": 0.6872503728972055, "grad_norm": 2.077669620513916, "learning_rate": 4.706950470082177e-06, "loss": 0.12679564952850342, "step": 26090 }, { "epoch": 0.6875137881417043, "grad_norm": 2.421854019165039, "learning_rate": 4.699714110656242e-06, "loss": 0.12060567140579223, "step": 26100 }, { "epoch": 0.687777203386203, "grad_norm": 2.100238800048828, "learning_rate": 4.6924816090070575e-06, "loss": 0.13937201499938964, "step": 26110 }, { "epoch": 0.6880406186307017, "grad_norm": 2.3384552001953125, "learning_rate": 4.685252970398742e-06, "loss": 0.1422940731048584, "step": 26120 }, { "epoch": 0.6883040338752004, "grad_norm": 1.4034349918365479, "learning_rate": 4.678028200092607e-06, "loss": 0.1319751262664795, "step": 26130 }, { "epoch": 0.6885674491196991, "grad_norm": 1.5970114469528198, "learning_rate": 4.670807303347161e-06, "loss": 0.12898743152618408, "step": 26140 }, { "epoch": 0.6888308643641979, "grad_norm": 3.233220100402832, "learning_rate": 4.663590285418082e-06, "loss": 0.130381977558136, "step": 26150 }, { "epoch": 0.6890942796086966, "grad_norm": 3.1935136318206787, "learning_rate": 4.656377151558224e-06, "loss": 0.11741485595703124, "step": 26160 }, { "epoch": 0.6893576948531954, "grad_norm": 2.3086812496185303, "learning_rate": 4.64916790701762e-06, "loss": 0.13923773765563965, "step": 26170 }, { "epoch": 0.6896211100976941, "grad_norm": 3.570732593536377, "learning_rate": 4.641962557043462e-06, "loss": 0.12289612293243408, "step": 26180 }, { "epoch": 0.6898845253421929, "grad_norm": 2.1623666286468506, "learning_rate": 4.634761106880127e-06, "loss": 0.12794969081878663, "step": 26190 }, { "epoch": 0.6901479405866916, "grad_norm": 1.8056031465530396, "learning_rate": 4.627563561769136e-06, "loss": 0.13098247051239015, "step": 26200 }, { "epoch": 0.6904113558311903, "grad_norm": 1.5793615579605103, "learning_rate": 4.620369926949164e-06, "loss": 0.12650901079177856, "step": 26210 }, { "epoch": 0.6906747710756891, "grad_norm": 1.753039836883545, "learning_rate": 4.613180207656061e-06, "loss": 0.11267906427383423, "step": 26220 }, { "epoch": 0.6909381863201878, "grad_norm": 1.5850797891616821, "learning_rate": 4.60599440912281e-06, "loss": 0.12473773956298828, "step": 26230 }, { "epoch": 0.6912016015646866, "grad_norm": 3.6848063468933105, "learning_rate": 4.598812536579545e-06, "loss": 0.12391815185546876, "step": 26240 }, { "epoch": 0.6914650168091853, "grad_norm": 1.9935423135757446, "learning_rate": 4.591634595253541e-06, "loss": 0.124055016040802, "step": 26250 }, { "epoch": 0.6917284320536841, "grad_norm": 1.9530961513519287, "learning_rate": 4.584460590369214e-06, "loss": 0.1342921733856201, "step": 26260 }, { "epoch": 0.6919918472981827, "grad_norm": 1.93272066116333, "learning_rate": 4.577290527148118e-06, "loss": 0.13036434650421141, "step": 26270 }, { "epoch": 0.6922552625426815, "grad_norm": 2.3334197998046875, "learning_rate": 4.5701244108089325e-06, "loss": 0.13974781036376954, "step": 26280 }, { "epoch": 0.6925186777871802, "grad_norm": 1.6536011695861816, "learning_rate": 4.562962246567466e-06, "loss": 0.13592331409454345, "step": 26290 }, { "epoch": 0.6927820930316789, "grad_norm": 3.467686414718628, "learning_rate": 4.555804039636649e-06, "loss": 0.13042404651641845, "step": 26300 }, { "epoch": 0.6930455082761777, "grad_norm": 4.79103422164917, "learning_rate": 4.548649795226532e-06, "loss": 0.12849488258361816, "step": 26310 }, { "epoch": 0.6933089235206764, "grad_norm": 1.5319290161132812, "learning_rate": 4.541499518544292e-06, "loss": 0.13182289600372316, "step": 26320 }, { "epoch": 0.6935723387651752, "grad_norm": 2.7932512760162354, "learning_rate": 4.5343532147941985e-06, "loss": 0.13107655048370362, "step": 26330 }, { "epoch": 0.6938357540096739, "grad_norm": 1.9134095907211304, "learning_rate": 4.527210889177639e-06, "loss": 0.1394984245300293, "step": 26340 }, { "epoch": 0.6940991692541727, "grad_norm": 3.494183301925659, "learning_rate": 4.520072546893115e-06, "loss": 0.14108283519744874, "step": 26350 }, { "epoch": 0.6943625844986714, "grad_norm": 1.8695812225341797, "learning_rate": 4.512938193136216e-06, "loss": 0.12775381803512573, "step": 26360 }, { "epoch": 0.6946259997431702, "grad_norm": 1.36142098903656, "learning_rate": 4.5058078330996315e-06, "loss": 0.1271292209625244, "step": 26370 }, { "epoch": 0.6948894149876689, "grad_norm": 2.0642049312591553, "learning_rate": 4.498681471973145e-06, "loss": 0.12947604656219483, "step": 26380 }, { "epoch": 0.6951528302321676, "grad_norm": 2.127567768096924, "learning_rate": 4.491559114943628e-06, "loss": 0.12507343292236328, "step": 26390 }, { "epoch": 0.6954162454766664, "grad_norm": 1.7812960147857666, "learning_rate": 4.484440767195044e-06, "loss": 0.14492454528808593, "step": 26400 }, { "epoch": 0.695679660721165, "grad_norm": 1.648618459701538, "learning_rate": 4.477326433908435e-06, "loss": 0.12851827144622802, "step": 26410 }, { "epoch": 0.6959430759656638, "grad_norm": 1.1585171222686768, "learning_rate": 4.4702161202619085e-06, "loss": 0.131928551197052, "step": 26420 }, { "epoch": 0.6962064912101625, "grad_norm": 1.2481154203414917, "learning_rate": 4.463109831430668e-06, "loss": 0.1286790370941162, "step": 26430 }, { "epoch": 0.6964699064546613, "grad_norm": 1.8354549407958984, "learning_rate": 4.456007572586971e-06, "loss": 0.12692601680755616, "step": 26440 }, { "epoch": 0.69673332169916, "grad_norm": 2.551865816116333, "learning_rate": 4.448909348900159e-06, "loss": 0.12757575511932373, "step": 26450 }, { "epoch": 0.6969967369436588, "grad_norm": 2.2781405448913574, "learning_rate": 4.441815165536612e-06, "loss": 0.13813257217407227, "step": 26460 }, { "epoch": 0.6972601521881575, "grad_norm": 2.2056455612182617, "learning_rate": 4.434725027659786e-06, "loss": 0.13249012231826782, "step": 26470 }, { "epoch": 0.6975235674326562, "grad_norm": 2.665870428085327, "learning_rate": 4.427638940430192e-06, "loss": 0.1203934907913208, "step": 26480 }, { "epoch": 0.697786982677155, "grad_norm": 2.2185428142547607, "learning_rate": 4.420556909005392e-06, "loss": 0.13218129873275758, "step": 26490 }, { "epoch": 0.6980503979216537, "grad_norm": 2.939619302749634, "learning_rate": 4.41347893853999e-06, "loss": 0.14305559396743775, "step": 26500 }, { "epoch": 0.6983138131661525, "grad_norm": 3.4107892513275146, "learning_rate": 4.40640503418564e-06, "loss": 0.12065484523773193, "step": 26510 }, { "epoch": 0.6985772284106512, "grad_norm": 1.9436159133911133, "learning_rate": 4.39933520109103e-06, "loss": 0.11668217182159424, "step": 26520 }, { "epoch": 0.69884064365515, "grad_norm": 3.9934725761413574, "learning_rate": 4.3922694444018985e-06, "loss": 0.12266271114349366, "step": 26530 }, { "epoch": 0.6991040588996487, "grad_norm": 3.1371874809265137, "learning_rate": 4.385207769261007e-06, "loss": 0.12169344425201416, "step": 26540 }, { "epoch": 0.6993674741441475, "grad_norm": 4.207334518432617, "learning_rate": 4.378150180808136e-06, "loss": 0.1378082275390625, "step": 26550 }, { "epoch": 0.6996308893886461, "grad_norm": 2.0985138416290283, "learning_rate": 4.371096684180114e-06, "loss": 0.13119562864303588, "step": 26560 }, { "epoch": 0.6998943046331448, "grad_norm": 1.6139988899230957, "learning_rate": 4.364047284510774e-06, "loss": 0.1078732967376709, "step": 26570 }, { "epoch": 0.7001577198776436, "grad_norm": 1.6786175966262817, "learning_rate": 4.357001986930982e-06, "loss": 0.12050902843475342, "step": 26580 }, { "epoch": 0.7004211351221423, "grad_norm": 2.9565529823303223, "learning_rate": 4.349960796568601e-06, "loss": 0.126115620136261, "step": 26590 }, { "epoch": 0.7006845503666411, "grad_norm": 1.7983038425445557, "learning_rate": 4.342923718548512e-06, "loss": 0.12370152473449707, "step": 26600 }, { "epoch": 0.7009479656111398, "grad_norm": 1.6351481676101685, "learning_rate": 4.335890757992611e-06, "loss": 0.12131609916687011, "step": 26610 }, { "epoch": 0.7012113808556386, "grad_norm": 1.555045247077942, "learning_rate": 4.328861920019787e-06, "loss": 0.1169790506362915, "step": 26620 }, { "epoch": 0.7014747961001373, "grad_norm": 3.103348970413208, "learning_rate": 4.321837209745931e-06, "loss": 0.12408034801483155, "step": 26630 }, { "epoch": 0.701738211344636, "grad_norm": 1.6203829050064087, "learning_rate": 4.3148166322839305e-06, "loss": 0.1263548493385315, "step": 26640 }, { "epoch": 0.7020016265891348, "grad_norm": 1.9362592697143555, "learning_rate": 4.3078001927436605e-06, "loss": 0.12072362899780273, "step": 26650 }, { "epoch": 0.7022650418336335, "grad_norm": 2.1045820713043213, "learning_rate": 4.300787896231997e-06, "loss": 0.11856260299682617, "step": 26660 }, { "epoch": 0.7025284570781323, "grad_norm": 3.1594934463500977, "learning_rate": 4.293779747852791e-06, "loss": 0.12358498573303223, "step": 26670 }, { "epoch": 0.702791872322631, "grad_norm": 3.3807907104492188, "learning_rate": 4.286775752706865e-06, "loss": 0.12335718870162964, "step": 26680 }, { "epoch": 0.7030552875671298, "grad_norm": 3.113083839416504, "learning_rate": 4.279775915892038e-06, "loss": 0.13083329200744628, "step": 26690 }, { "epoch": 0.7033187028116284, "grad_norm": 1.6428978443145752, "learning_rate": 4.2727802425030885e-06, "loss": 0.11181845664978027, "step": 26700 }, { "epoch": 0.7035821180561272, "grad_norm": 1.7914539575576782, "learning_rate": 4.265788737631779e-06, "loss": 0.1103587031364441, "step": 26710 }, { "epoch": 0.7038455333006259, "grad_norm": 2.371513605117798, "learning_rate": 4.258801406366817e-06, "loss": 0.13570210933685303, "step": 26720 }, { "epoch": 0.7041089485451246, "grad_norm": 3.4533910751342773, "learning_rate": 4.251818253793885e-06, "loss": 0.1406776189804077, "step": 26730 }, { "epoch": 0.7043723637896234, "grad_norm": 2.1078550815582275, "learning_rate": 4.244839284995627e-06, "loss": 0.1339285135269165, "step": 26740 }, { "epoch": 0.7046357790341221, "grad_norm": 3.0231995582580566, "learning_rate": 4.237864505051638e-06, "loss": 0.13203163146972657, "step": 26750 }, { "epoch": 0.7048991942786209, "grad_norm": 2.1137752532958984, "learning_rate": 4.230893919038461e-06, "loss": 0.13284714221954347, "step": 26760 }, { "epoch": 0.7051626095231196, "grad_norm": 1.8601187467575073, "learning_rate": 4.223927532029592e-06, "loss": 0.11525154113769531, "step": 26770 }, { "epoch": 0.7054260247676184, "grad_norm": 2.7552926540374756, "learning_rate": 4.21696534909546e-06, "loss": 0.14255441427230836, "step": 26780 }, { "epoch": 0.7056894400121171, "grad_norm": 1.8275222778320312, "learning_rate": 4.2100073753034535e-06, "loss": 0.12475419044494629, "step": 26790 }, { "epoch": 0.7059528552566159, "grad_norm": 2.4489476680755615, "learning_rate": 4.203053615717884e-06, "loss": 0.1261476993560791, "step": 26800 }, { "epoch": 0.7062162705011146, "grad_norm": 1.7699565887451172, "learning_rate": 4.196104075399986e-06, "loss": 0.13922574520111083, "step": 26810 }, { "epoch": 0.7064796857456133, "grad_norm": 2.0835041999816895, "learning_rate": 4.1891587594079485e-06, "loss": 0.1339637041091919, "step": 26820 }, { "epoch": 0.7067431009901121, "grad_norm": 1.8757370710372925, "learning_rate": 4.182217672796867e-06, "loss": 0.1259255290031433, "step": 26830 }, { "epoch": 0.7070065162346107, "grad_norm": 1.4603394269943237, "learning_rate": 4.175280820618763e-06, "loss": 0.12973591089248657, "step": 26840 }, { "epoch": 0.7072699314791095, "grad_norm": 2.0848467350006104, "learning_rate": 4.168348207922577e-06, "loss": 0.11305248737335205, "step": 26850 }, { "epoch": 0.7075333467236082, "grad_norm": 3.034919500350952, "learning_rate": 4.16141983975416e-06, "loss": 0.12886970043182372, "step": 26860 }, { "epoch": 0.707796761968107, "grad_norm": 3.080284357070923, "learning_rate": 4.154495721156287e-06, "loss": 0.12849273681640624, "step": 26870 }, { "epoch": 0.7080601772126057, "grad_norm": 2.5319647789001465, "learning_rate": 4.147575857168623e-06, "loss": 0.12133002281188965, "step": 26880 }, { "epoch": 0.7083235924571045, "grad_norm": 1.5519593954086304, "learning_rate": 4.140660252827746e-06, "loss": 0.11225550174713135, "step": 26890 }, { "epoch": 0.7085870077016032, "grad_norm": 2.9240245819091797, "learning_rate": 4.13374891316713e-06, "loss": 0.12198654413223267, "step": 26900 }, { "epoch": 0.7088504229461019, "grad_norm": 1.679378867149353, "learning_rate": 4.126841843217144e-06, "loss": 0.13282037973403932, "step": 26910 }, { "epoch": 0.7091138381906007, "grad_norm": 2.7658050060272217, "learning_rate": 4.1199390480050605e-06, "loss": 0.12534494400024415, "step": 26920 }, { "epoch": 0.7093772534350994, "grad_norm": 2.3503100872039795, "learning_rate": 4.113040532555021e-06, "loss": 0.1185654878616333, "step": 26930 }, { "epoch": 0.7096406686795982, "grad_norm": 3.393669366836548, "learning_rate": 4.106146301888065e-06, "loss": 0.13894922733306886, "step": 26940 }, { "epoch": 0.7099040839240969, "grad_norm": 1.8273921012878418, "learning_rate": 4.0992563610221155e-06, "loss": 0.11514692306518555, "step": 26950 }, { "epoch": 0.7101674991685957, "grad_norm": 2.228851556777954, "learning_rate": 4.092370714971966e-06, "loss": 0.1271723508834839, "step": 26960 }, { "epoch": 0.7104309144130944, "grad_norm": 2.3353769779205322, "learning_rate": 4.085489368749287e-06, "loss": 0.12739841938018798, "step": 26970 }, { "epoch": 0.7106943296575932, "grad_norm": 2.3917462825775146, "learning_rate": 4.07861232736262e-06, "loss": 0.13114361763000487, "step": 26980 }, { "epoch": 0.7109577449020918, "grad_norm": 4.083085536956787, "learning_rate": 4.0717395958173665e-06, "loss": 0.12709147930145265, "step": 26990 }, { "epoch": 0.7112211601465905, "grad_norm": 3.93658709526062, "learning_rate": 4.064871179115807e-06, "loss": 0.13175179958343505, "step": 27000 }, { "epoch": 0.7114845753910893, "grad_norm": 2.6281015872955322, "learning_rate": 4.0580070822570694e-06, "loss": 0.12263141870498658, "step": 27010 }, { "epoch": 0.711747990635588, "grad_norm": 1.7049773931503296, "learning_rate": 4.051147310237129e-06, "loss": 0.12646443843841554, "step": 27020 }, { "epoch": 0.7120114058800868, "grad_norm": 1.8857557773590088, "learning_rate": 4.044291868048837e-06, "loss": 0.11836047172546386, "step": 27030 }, { "epoch": 0.7122748211245855, "grad_norm": 2.729722023010254, "learning_rate": 4.0374407606818684e-06, "loss": 0.13139986991882324, "step": 27040 }, { "epoch": 0.7125382363690843, "grad_norm": 2.561631679534912, "learning_rate": 4.03059399312277e-06, "loss": 0.15160002708435058, "step": 27050 }, { "epoch": 0.712801651613583, "grad_norm": 2.376431703567505, "learning_rate": 4.0237515703549015e-06, "loss": 0.12261762619018554, "step": 27060 }, { "epoch": 0.7130650668580818, "grad_norm": 1.5978031158447266, "learning_rate": 4.0169134973584726e-06, "loss": 0.11433976888656616, "step": 27070 }, { "epoch": 0.7133284821025805, "grad_norm": 1.8167953491210938, "learning_rate": 4.010079779110537e-06, "loss": 0.12868287563323974, "step": 27080 }, { "epoch": 0.7135918973470792, "grad_norm": 1.6787364482879639, "learning_rate": 4.003250420584966e-06, "loss": 0.13813371658325196, "step": 27090 }, { "epoch": 0.713855312591578, "grad_norm": 2.511669635772705, "learning_rate": 3.996425426752459e-06, "loss": 0.13897154331207276, "step": 27100 }, { "epoch": 0.7141187278360767, "grad_norm": 2.180558443069458, "learning_rate": 3.989604802580545e-06, "loss": 0.12546925544738768, "step": 27110 }, { "epoch": 0.7143821430805755, "grad_norm": 1.8234881162643433, "learning_rate": 3.982788553033561e-06, "loss": 0.12262818813323975, "step": 27120 }, { "epoch": 0.7146455583250741, "grad_norm": 2.1695377826690674, "learning_rate": 3.975976683072676e-06, "loss": 0.1344813108444214, "step": 27130 }, { "epoch": 0.714908973569573, "grad_norm": 2.329129219055176, "learning_rate": 3.9691691976558636e-06, "loss": 0.14063794612884523, "step": 27140 }, { "epoch": 0.7151723888140716, "grad_norm": 2.0837936401367188, "learning_rate": 3.9623661017378946e-06, "loss": 0.12877953052520752, "step": 27150 }, { "epoch": 0.7154358040585704, "grad_norm": 2.086613655090332, "learning_rate": 3.955567400270365e-06, "loss": 0.12686259746551515, "step": 27160 }, { "epoch": 0.7156992193030691, "grad_norm": 2.207915782928467, "learning_rate": 3.948773098201657e-06, "loss": 0.11699705123901367, "step": 27170 }, { "epoch": 0.7159626345475678, "grad_norm": 2.9946210384368896, "learning_rate": 3.941983200476969e-06, "loss": 0.12474590539932251, "step": 27180 }, { "epoch": 0.7162260497920666, "grad_norm": 2.4746949672698975, "learning_rate": 3.935197712038267e-06, "loss": 0.11840057373046875, "step": 27190 }, { "epoch": 0.7164894650365653, "grad_norm": 1.5092371702194214, "learning_rate": 3.928416637824326e-06, "loss": 0.11455458402633667, "step": 27200 }, { "epoch": 0.7167528802810641, "grad_norm": 2.7356958389282227, "learning_rate": 3.921639982770709e-06, "loss": 0.13071117401123047, "step": 27210 }, { "epoch": 0.7170162955255628, "grad_norm": 1.5331411361694336, "learning_rate": 3.914867751809756e-06, "loss": 0.129394793510437, "step": 27220 }, { "epoch": 0.7172797107700616, "grad_norm": 2.523557186126709, "learning_rate": 3.908099949870588e-06, "loss": 0.12697205543518067, "step": 27230 }, { "epoch": 0.7175431260145603, "grad_norm": 2.0871925354003906, "learning_rate": 3.901336581879102e-06, "loss": 0.12007498741149902, "step": 27240 }, { "epoch": 0.717806541259059, "grad_norm": 2.108488082885742, "learning_rate": 3.894577652757968e-06, "loss": 0.13268465995788575, "step": 27250 }, { "epoch": 0.7180699565035578, "grad_norm": 1.7211796045303345, "learning_rate": 3.887823167426633e-06, "loss": 0.12560064792633058, "step": 27260 }, { "epoch": 0.7183333717480564, "grad_norm": 1.3802917003631592, "learning_rate": 3.8810731308013005e-06, "loss": 0.11177425384521485, "step": 27270 }, { "epoch": 0.7185967869925552, "grad_norm": 1.3732608556747437, "learning_rate": 3.874327547794931e-06, "loss": 0.11951870918273926, "step": 27280 }, { "epoch": 0.7188602022370539, "grad_norm": 2.743987560272217, "learning_rate": 3.8675864233172575e-06, "loss": 0.1263663053512573, "step": 27290 }, { "epoch": 0.7191236174815527, "grad_norm": 1.4458520412445068, "learning_rate": 3.860849762274759e-06, "loss": 0.1145302414894104, "step": 27300 }, { "epoch": 0.7193870327260514, "grad_norm": 1.997813105583191, "learning_rate": 3.854117569570674e-06, "loss": 0.13440098762512206, "step": 27310 }, { "epoch": 0.7196504479705502, "grad_norm": 1.719280481338501, "learning_rate": 3.8473898501049745e-06, "loss": 0.1221724271774292, "step": 27320 }, { "epoch": 0.7199138632150489, "grad_norm": 1.8664685487747192, "learning_rate": 3.840666608774385e-06, "loss": 0.11965910196304322, "step": 27330 }, { "epoch": 0.7201772784595476, "grad_norm": 2.605727434158325, "learning_rate": 3.833947850472378e-06, "loss": 0.12000195980072022, "step": 27340 }, { "epoch": 0.7204406937040464, "grad_norm": 3.2546744346618652, "learning_rate": 3.827233580089152e-06, "loss": 0.13966094255447387, "step": 27350 }, { "epoch": 0.7207041089485451, "grad_norm": 3.0369017124176025, "learning_rate": 3.82052380251164e-06, "loss": 0.12899715900421144, "step": 27360 }, { "epoch": 0.7209675241930439, "grad_norm": 2.211146593093872, "learning_rate": 3.8138185226235114e-06, "loss": 0.11850711107254028, "step": 27370 }, { "epoch": 0.7212309394375426, "grad_norm": 2.463955879211426, "learning_rate": 3.8071177453051523e-06, "loss": 0.1305699348449707, "step": 27380 }, { "epoch": 0.7214943546820414, "grad_norm": 3.1168906688690186, "learning_rate": 3.8004214754336853e-06, "loss": 0.12957961559295655, "step": 27390 }, { "epoch": 0.7217577699265401, "grad_norm": 2.4218227863311768, "learning_rate": 3.7937297178829458e-06, "loss": 0.11502560377120971, "step": 27400 }, { "epoch": 0.7220211851710389, "grad_norm": 2.7724609375, "learning_rate": 3.7870424775234703e-06, "loss": 0.11876897811889649, "step": 27410 }, { "epoch": 0.7222846004155375, "grad_norm": 1.2698166370391846, "learning_rate": 3.780359759222534e-06, "loss": 0.12964173555374145, "step": 27420 }, { "epoch": 0.7225480156600362, "grad_norm": 2.049455165863037, "learning_rate": 3.773681567844103e-06, "loss": 0.12580127716064454, "step": 27430 }, { "epoch": 0.722811430904535, "grad_norm": 1.8707528114318848, "learning_rate": 3.767007908248852e-06, "loss": 0.12704267501831054, "step": 27440 }, { "epoch": 0.7230748461490337, "grad_norm": 2.5649685859680176, "learning_rate": 3.7603387852941608e-06, "loss": 0.1205400824546814, "step": 27450 }, { "epoch": 0.7233382613935325, "grad_norm": 2.4782330989837646, "learning_rate": 3.753674203834099e-06, "loss": 0.11247961521148682, "step": 27460 }, { "epoch": 0.7236016766380312, "grad_norm": 3.1084532737731934, "learning_rate": 3.747014168719445e-06, "loss": 0.11528878211975098, "step": 27470 }, { "epoch": 0.72386509188253, "grad_norm": 2.9042773246765137, "learning_rate": 3.740358684797656e-06, "loss": 0.12799676656723022, "step": 27480 }, { "epoch": 0.7241285071270287, "grad_norm": 1.7792513370513916, "learning_rate": 3.7337077569128797e-06, "loss": 0.12344682216644287, "step": 27490 }, { "epoch": 0.7243919223715275, "grad_norm": 3.0840137004852295, "learning_rate": 3.7270613899059495e-06, "loss": 0.11055762767791748, "step": 27500 }, { "epoch": 0.7246553376160262, "grad_norm": 2.2448649406433105, "learning_rate": 3.7204195886143736e-06, "loss": 0.11250364780426025, "step": 27510 }, { "epoch": 0.7249187528605249, "grad_norm": 2.6653573513031006, "learning_rate": 3.713782357872352e-06, "loss": 0.12412221431732177, "step": 27520 }, { "epoch": 0.7251821681050237, "grad_norm": 2.784731388092041, "learning_rate": 3.7071497025107373e-06, "loss": 0.1267947554588318, "step": 27530 }, { "epoch": 0.7254455833495224, "grad_norm": 2.0468318462371826, "learning_rate": 3.7005216273570623e-06, "loss": 0.11415042877197265, "step": 27540 }, { "epoch": 0.7257089985940212, "grad_norm": 2.363417625427246, "learning_rate": 3.693898137235532e-06, "loss": 0.12244791984558105, "step": 27550 }, { "epoch": 0.7259724138385198, "grad_norm": 2.1282827854156494, "learning_rate": 3.6872792369670074e-06, "loss": 0.121429705619812, "step": 27560 }, { "epoch": 0.7262358290830186, "grad_norm": 1.4969696998596191, "learning_rate": 3.6806649313690066e-06, "loss": 0.1202396273612976, "step": 27570 }, { "epoch": 0.7264992443275173, "grad_norm": 1.6459083557128906, "learning_rate": 3.6740552252557093e-06, "loss": 0.12921652793884278, "step": 27580 }, { "epoch": 0.7267626595720161, "grad_norm": 1.4309989213943481, "learning_rate": 3.667450123437939e-06, "loss": 0.13196494579315185, "step": 27590 }, { "epoch": 0.7270260748165148, "grad_norm": 1.865199089050293, "learning_rate": 3.6608496307231823e-06, "loss": 0.11636639833450317, "step": 27600 }, { "epoch": 0.7272894900610135, "grad_norm": 1.8824511766433716, "learning_rate": 3.6542537519155617e-06, "loss": 0.11038496494293212, "step": 27610 }, { "epoch": 0.7275529053055123, "grad_norm": 2.587639093399048, "learning_rate": 3.647662491815833e-06, "loss": 0.12481789588928223, "step": 27620 }, { "epoch": 0.727816320550011, "grad_norm": 1.916255235671997, "learning_rate": 3.641075855221409e-06, "loss": 0.135107684135437, "step": 27630 }, { "epoch": 0.7280797357945098, "grad_norm": 1.7707706689834595, "learning_rate": 3.634493846926321e-06, "loss": 0.1323853015899658, "step": 27640 }, { "epoch": 0.7283431510390085, "grad_norm": 2.6606693267822266, "learning_rate": 3.62791647172125e-06, "loss": 0.13330211639404296, "step": 27650 }, { "epoch": 0.7286065662835073, "grad_norm": 2.7702252864837646, "learning_rate": 3.621343734393482e-06, "loss": 0.13105560541152955, "step": 27660 }, { "epoch": 0.728869981528006, "grad_norm": 1.921453833580017, "learning_rate": 3.6147756397269397e-06, "loss": 0.12434623241424561, "step": 27670 }, { "epoch": 0.7291333967725048, "grad_norm": 2.8965272903442383, "learning_rate": 3.6082121925021717e-06, "loss": 0.12267184257507324, "step": 27680 }, { "epoch": 0.7293968120170035, "grad_norm": 1.8276113271713257, "learning_rate": 3.601653397496335e-06, "loss": 0.11598846912384034, "step": 27690 }, { "epoch": 0.7296602272615021, "grad_norm": 1.9913052320480347, "learning_rate": 3.5950992594832044e-06, "loss": 0.12442805767059326, "step": 27700 }, { "epoch": 0.729923642506001, "grad_norm": 2.346356153488159, "learning_rate": 3.588549783233164e-06, "loss": 0.11131479740142822, "step": 27710 }, { "epoch": 0.7301870577504996, "grad_norm": 2.4962234497070312, "learning_rate": 3.582004973513201e-06, "loss": 0.12587311267852783, "step": 27720 }, { "epoch": 0.7304504729949984, "grad_norm": 4.926432132720947, "learning_rate": 3.575464835086917e-06, "loss": 0.13185852766036987, "step": 27730 }, { "epoch": 0.7307138882394971, "grad_norm": 3.7772886753082275, "learning_rate": 3.568929372714508e-06, "loss": 0.14498796463012695, "step": 27740 }, { "epoch": 0.7309773034839959, "grad_norm": 2.9336397647857666, "learning_rate": 3.562398591152755e-06, "loss": 0.12512447834014892, "step": 27750 }, { "epoch": 0.7312407187284946, "grad_norm": 2.057337522506714, "learning_rate": 3.5558724951550515e-06, "loss": 0.12149609327316284, "step": 27760 }, { "epoch": 0.7315041339729933, "grad_norm": 1.6912169456481934, "learning_rate": 3.549351089471367e-06, "loss": 0.11537425518035889, "step": 27770 }, { "epoch": 0.7317675492174921, "grad_norm": 3.0191993713378906, "learning_rate": 3.54283437884827e-06, "loss": 0.12501201629638672, "step": 27780 }, { "epoch": 0.7320309644619908, "grad_norm": 2.24043869972229, "learning_rate": 3.536322368028895e-06, "loss": 0.12483590841293335, "step": 27790 }, { "epoch": 0.7322943797064896, "grad_norm": 2.882416009902954, "learning_rate": 3.5298150617529635e-06, "loss": 0.1298593759536743, "step": 27800 }, { "epoch": 0.7325577949509883, "grad_norm": 2.640341281890869, "learning_rate": 3.5233124647567817e-06, "loss": 0.1384586811065674, "step": 27810 }, { "epoch": 0.7328212101954871, "grad_norm": 1.7522492408752441, "learning_rate": 3.516814581773216e-06, "loss": 0.118773353099823, "step": 27820 }, { "epoch": 0.7330846254399858, "grad_norm": 2.1985936164855957, "learning_rate": 3.5103214175317056e-06, "loss": 0.11761863231658935, "step": 27830 }, { "epoch": 0.7333480406844846, "grad_norm": 2.0492138862609863, "learning_rate": 3.5038329767582567e-06, "loss": 0.12718544006347657, "step": 27840 }, { "epoch": 0.7336114559289832, "grad_norm": 2.1592743396759033, "learning_rate": 3.4973492641754315e-06, "loss": 0.1167148232460022, "step": 27850 }, { "epoch": 0.7338748711734819, "grad_norm": 2.2309024333953857, "learning_rate": 3.4908702845023643e-06, "loss": 0.11794106960296631, "step": 27860 }, { "epoch": 0.7341382864179807, "grad_norm": 2.1205742359161377, "learning_rate": 3.484396042454734e-06, "loss": 0.1273207426071167, "step": 27870 }, { "epoch": 0.7344017016624794, "grad_norm": 2.086106300354004, "learning_rate": 3.477926542744766e-06, "loss": 0.12062928676605225, "step": 27880 }, { "epoch": 0.7346651169069782, "grad_norm": 3.583400011062622, "learning_rate": 3.4714617900812485e-06, "loss": 0.13649301528930663, "step": 27890 }, { "epoch": 0.7349285321514769, "grad_norm": 3.712921619415283, "learning_rate": 3.465001789169502e-06, "loss": 0.12089052200317382, "step": 27900 }, { "epoch": 0.7351919473959757, "grad_norm": 2.2505240440368652, "learning_rate": 3.458546544711404e-06, "loss": 0.12655550241470337, "step": 27910 }, { "epoch": 0.7354553626404744, "grad_norm": 2.037625789642334, "learning_rate": 3.45209606140535e-06, "loss": 0.1157951831817627, "step": 27920 }, { "epoch": 0.7357187778849732, "grad_norm": 2.3300464153289795, "learning_rate": 3.4456503439462797e-06, "loss": 0.11829931735992431, "step": 27930 }, { "epoch": 0.7359821931294719, "grad_norm": 2.4028725624084473, "learning_rate": 3.4392093970256703e-06, "loss": 0.13696064949035644, "step": 27940 }, { "epoch": 0.7362456083739706, "grad_norm": 2.1234827041625977, "learning_rate": 3.4327732253315195e-06, "loss": 0.13201115131378174, "step": 27950 }, { "epoch": 0.7365090236184694, "grad_norm": 1.9247244596481323, "learning_rate": 3.4263418335483477e-06, "loss": 0.12104233503341674, "step": 27960 }, { "epoch": 0.7367724388629681, "grad_norm": 2.152416467666626, "learning_rate": 3.419915226357203e-06, "loss": 0.10875089168548584, "step": 27970 }, { "epoch": 0.7370358541074669, "grad_norm": 2.7312161922454834, "learning_rate": 3.413493408435642e-06, "loss": 0.12240619659423828, "step": 27980 }, { "epoch": 0.7372992693519655, "grad_norm": 2.3106813430786133, "learning_rate": 3.407076384457747e-06, "loss": 0.11111165285110473, "step": 27990 }, { "epoch": 0.7375626845964643, "grad_norm": 1.7933475971221924, "learning_rate": 3.400664159094106e-06, "loss": 0.10788981914520264, "step": 28000 }, { "epoch": 0.737826099840963, "grad_norm": 1.9450124502182007, "learning_rate": 3.394256737011803e-06, "loss": 0.1095770239830017, "step": 28010 }, { "epoch": 0.7380895150854618, "grad_norm": 2.0326757431030273, "learning_rate": 3.3878541228744455e-06, "loss": 0.10574278831481934, "step": 28020 }, { "epoch": 0.7383529303299605, "grad_norm": 1.748637080192566, "learning_rate": 3.38145632134213e-06, "loss": 0.11835110187530518, "step": 28030 }, { "epoch": 0.7386163455744592, "grad_norm": 3.032410144805908, "learning_rate": 3.3750633370714527e-06, "loss": 0.11760175228118896, "step": 28040 }, { "epoch": 0.738879760818958, "grad_norm": 1.5715725421905518, "learning_rate": 3.368675174715501e-06, "loss": 0.12205849885940552, "step": 28050 }, { "epoch": 0.7391431760634567, "grad_norm": 3.7094619274139404, "learning_rate": 3.3622918389238544e-06, "loss": 0.12052245140075683, "step": 28060 }, { "epoch": 0.7394065913079555, "grad_norm": 1.7881040573120117, "learning_rate": 3.3559133343425865e-06, "loss": 0.1297991991043091, "step": 28070 }, { "epoch": 0.7396700065524542, "grad_norm": 2.6684420108795166, "learning_rate": 3.3495396656142442e-06, "loss": 0.12518045902252198, "step": 28080 }, { "epoch": 0.739933421796953, "grad_norm": 2.653926134109497, "learning_rate": 3.343170837377858e-06, "loss": 0.12330400943756104, "step": 28090 }, { "epoch": 0.7401968370414517, "grad_norm": 2.3162620067596436, "learning_rate": 3.336806854268936e-06, "loss": 0.11045894622802735, "step": 28100 }, { "epoch": 0.7404602522859505, "grad_norm": 2.101665735244751, "learning_rate": 3.3304477209194574e-06, "loss": 0.12350244522094726, "step": 28110 }, { "epoch": 0.7407236675304492, "grad_norm": 2.249006986618042, "learning_rate": 3.3240934419578828e-06, "loss": 0.12132606506347657, "step": 28120 }, { "epoch": 0.7409870827749478, "grad_norm": 1.6865530014038086, "learning_rate": 3.3177440220091207e-06, "loss": 0.11514096260070801, "step": 28130 }, { "epoch": 0.7412504980194466, "grad_norm": 1.676722526550293, "learning_rate": 3.3113994656945537e-06, "loss": 0.11407766342163086, "step": 28140 }, { "epoch": 0.7415139132639453, "grad_norm": 3.0163607597351074, "learning_rate": 3.3050597776320293e-06, "loss": 0.11540033817291259, "step": 28150 }, { "epoch": 0.7417773285084441, "grad_norm": 2.4945318698883057, "learning_rate": 3.298724962435842e-06, "loss": 0.10597082376480102, "step": 28160 }, { "epoch": 0.7420407437529428, "grad_norm": 3.4957549571990967, "learning_rate": 3.292395024716746e-06, "loss": 0.1252351999282837, "step": 28170 }, { "epoch": 0.7423041589974416, "grad_norm": 2.473736047744751, "learning_rate": 3.2860699690819407e-06, "loss": 0.1172598123550415, "step": 28180 }, { "epoch": 0.7425675742419403, "grad_norm": 2.900479793548584, "learning_rate": 3.2797498001350737e-06, "loss": 0.1281599521636963, "step": 28190 }, { "epoch": 0.7428309894864391, "grad_norm": 3.1395835876464844, "learning_rate": 3.27343452247624e-06, "loss": 0.11093426942825317, "step": 28200 }, { "epoch": 0.7430944047309378, "grad_norm": 2.3122568130493164, "learning_rate": 3.267124140701974e-06, "loss": 0.12515852451324463, "step": 28210 }, { "epoch": 0.7433578199754365, "grad_norm": 1.9639606475830078, "learning_rate": 3.2608186594052314e-06, "loss": 0.11628085374832153, "step": 28220 }, { "epoch": 0.7436212352199353, "grad_norm": 2.4722096920013428, "learning_rate": 3.2545180831754264e-06, "loss": 0.1284090757369995, "step": 28230 }, { "epoch": 0.743884650464434, "grad_norm": 2.0174379348754883, "learning_rate": 3.248222416598381e-06, "loss": 0.10639231204986573, "step": 28240 }, { "epoch": 0.7441480657089328, "grad_norm": 1.821362853050232, "learning_rate": 3.2419316642563646e-06, "loss": 0.11315898895263672, "step": 28250 }, { "epoch": 0.7444114809534315, "grad_norm": 2.3399722576141357, "learning_rate": 3.235645830728047e-06, "loss": 0.1242445945739746, "step": 28260 }, { "epoch": 0.7446748961979303, "grad_norm": 1.5865986347198486, "learning_rate": 3.2293649205885293e-06, "loss": 0.10401179790496826, "step": 28270 }, { "epoch": 0.744938311442429, "grad_norm": 1.7964364290237427, "learning_rate": 3.2230889384093355e-06, "loss": 0.10837385654449463, "step": 28280 }, { "epoch": 0.7452017266869276, "grad_norm": 2.660982370376587, "learning_rate": 3.2168178887583947e-06, "loss": 0.11495282649993896, "step": 28290 }, { "epoch": 0.7454651419314264, "grad_norm": 3.7643890380859375, "learning_rate": 3.210551776200044e-06, "loss": 0.11184637546539307, "step": 28300 }, { "epoch": 0.7457285571759251, "grad_norm": 2.4784162044525146, "learning_rate": 3.204290605295033e-06, "loss": 0.12133021354675293, "step": 28310 }, { "epoch": 0.7459919724204239, "grad_norm": 2.6764302253723145, "learning_rate": 3.1980343806005077e-06, "loss": 0.1206396460533142, "step": 28320 }, { "epoch": 0.7462553876649226, "grad_norm": 2.241766929626465, "learning_rate": 3.1917831066700255e-06, "loss": 0.13148105144500732, "step": 28330 }, { "epoch": 0.7465188029094214, "grad_norm": 2.397270917892456, "learning_rate": 3.1855367880535336e-06, "loss": 0.13019320964813233, "step": 28340 }, { "epoch": 0.7467822181539201, "grad_norm": 2.2066164016723633, "learning_rate": 3.179295429297361e-06, "loss": 0.130704927444458, "step": 28350 }, { "epoch": 0.7470456333984189, "grad_norm": 2.1150805950164795, "learning_rate": 3.173059034944248e-06, "loss": 0.11153819561004638, "step": 28360 }, { "epoch": 0.7473090486429176, "grad_norm": 2.622462272644043, "learning_rate": 3.1668276095333063e-06, "loss": 0.12336583137512207, "step": 28370 }, { "epoch": 0.7475724638874163, "grad_norm": 2.0047993659973145, "learning_rate": 3.160601157600045e-06, "loss": 0.12738373279571533, "step": 28380 }, { "epoch": 0.7478358791319151, "grad_norm": 2.8534305095672607, "learning_rate": 3.154379683676334e-06, "loss": 0.10547405481338501, "step": 28390 }, { "epoch": 0.7480992943764138, "grad_norm": 2.978388786315918, "learning_rate": 3.148163192290432e-06, "loss": 0.11134934425354004, "step": 28400 }, { "epoch": 0.7483627096209126, "grad_norm": 2.2084341049194336, "learning_rate": 3.1419516879669744e-06, "loss": 0.11232075691223145, "step": 28410 }, { "epoch": 0.7486261248654112, "grad_norm": 2.474980354309082, "learning_rate": 3.1357451752269595e-06, "loss": 0.11047379970550537, "step": 28420 }, { "epoch": 0.74888954010991, "grad_norm": 2.009077787399292, "learning_rate": 3.1295436585877558e-06, "loss": 0.13073945045471191, "step": 28430 }, { "epoch": 0.7491529553544087, "grad_norm": 6.1512451171875, "learning_rate": 3.1233471425630947e-06, "loss": 0.12663640975952148, "step": 28440 }, { "epoch": 0.7494163705989075, "grad_norm": 2.2685225009918213, "learning_rate": 3.1171556316630626e-06, "loss": 0.11528844833374023, "step": 28450 }, { "epoch": 0.7496797858434062, "grad_norm": 2.3764679431915283, "learning_rate": 3.110969130394117e-06, "loss": 0.10492380857467651, "step": 28460 }, { "epoch": 0.7499432010879049, "grad_norm": 2.013015031814575, "learning_rate": 3.1047876432590596e-06, "loss": 0.12479019165039062, "step": 28470 }, { "epoch": 0.7502066163324037, "grad_norm": 2.3597538471221924, "learning_rate": 3.0986111747570335e-06, "loss": 0.12043893337249756, "step": 28480 }, { "epoch": 0.7504700315769024, "grad_norm": 2.6710612773895264, "learning_rate": 3.092439729383548e-06, "loss": 0.12899259328842164, "step": 28490 }, { "epoch": 0.7507334468214012, "grad_norm": 1.7170896530151367, "learning_rate": 3.0862733116304435e-06, "loss": 0.11207287311553955, "step": 28500 }, { "epoch": 0.7509968620658999, "grad_norm": 2.254316568374634, "learning_rate": 3.0801119259859115e-06, "loss": 0.12329511642456055, "step": 28510 }, { "epoch": 0.7512602773103987, "grad_norm": 2.5856878757476807, "learning_rate": 3.0739555769344666e-06, "loss": 0.10059999227523804, "step": 28520 }, { "epoch": 0.7515236925548974, "grad_norm": 2.6730995178222656, "learning_rate": 3.067804268956964e-06, "loss": 0.12572684288024902, "step": 28530 }, { "epoch": 0.7517871077993962, "grad_norm": 3.234936237335205, "learning_rate": 3.0616580065305977e-06, "loss": 0.11403517723083496, "step": 28540 }, { "epoch": 0.7520505230438949, "grad_norm": 1.7618201971054077, "learning_rate": 3.055516794128881e-06, "loss": 0.11804466247558594, "step": 28550 }, { "epoch": 0.7523139382883935, "grad_norm": 3.0234971046447754, "learning_rate": 3.0493806362216516e-06, "loss": 0.11571762561798096, "step": 28560 }, { "epoch": 0.7525773535328923, "grad_norm": 2.5136375427246094, "learning_rate": 3.0432495372750692e-06, "loss": 0.1192370057106018, "step": 28570 }, { "epoch": 0.752840768777391, "grad_norm": 2.5876593589782715, "learning_rate": 3.037123501751611e-06, "loss": 0.1172797441482544, "step": 28580 }, { "epoch": 0.7531041840218898, "grad_norm": 2.838146924972534, "learning_rate": 3.0310025341100757e-06, "loss": 0.12628397941589356, "step": 28590 }, { "epoch": 0.7533675992663885, "grad_norm": 2.09342098236084, "learning_rate": 3.0248866388055674e-06, "loss": 0.12289602756500244, "step": 28600 }, { "epoch": 0.7536310145108873, "grad_norm": 1.8049896955490112, "learning_rate": 3.018775820289488e-06, "loss": 0.10224072933197022, "step": 28610 }, { "epoch": 0.753894429755386, "grad_norm": 2.5406785011291504, "learning_rate": 3.0126700830095668e-06, "loss": 0.11847739219665528, "step": 28620 }, { "epoch": 0.7541578449998848, "grad_norm": 2.7523033618927, "learning_rate": 3.006569431409817e-06, "loss": 0.11090502738952637, "step": 28630 }, { "epoch": 0.7544212602443835, "grad_norm": 3.0836896896362305, "learning_rate": 3.000473869930558e-06, "loss": 0.11365814208984375, "step": 28640 }, { "epoch": 0.7546846754888822, "grad_norm": 2.1474192142486572, "learning_rate": 2.9943834030084006e-06, "loss": 0.12300286293029786, "step": 28650 }, { "epoch": 0.754948090733381, "grad_norm": 2.1502583026885986, "learning_rate": 2.988298035076249e-06, "loss": 0.11247246265411377, "step": 28660 }, { "epoch": 0.7552115059778797, "grad_norm": 2.5884459018707275, "learning_rate": 2.9822177705633005e-06, "loss": 0.1185571789741516, "step": 28670 }, { "epoch": 0.7554749212223785, "grad_norm": 3.269700527191162, "learning_rate": 2.976142613895031e-06, "loss": 0.11125520467758179, "step": 28680 }, { "epoch": 0.7557383364668772, "grad_norm": 2.686096668243408, "learning_rate": 2.970072569493202e-06, "loss": 0.10665868520736695, "step": 28690 }, { "epoch": 0.756001751711376, "grad_norm": 2.1038119792938232, "learning_rate": 2.9640076417758546e-06, "loss": 0.1264824628829956, "step": 28700 }, { "epoch": 0.7562651669558746, "grad_norm": 2.562220335006714, "learning_rate": 2.9579478351573e-06, "loss": 0.10971908569335938, "step": 28710 }, { "epoch": 0.7565285822003734, "grad_norm": 3.055809736251831, "learning_rate": 2.9518931540481377e-06, "loss": 0.11320481300354004, "step": 28720 }, { "epoch": 0.7567919974448721, "grad_norm": 1.9217923879623413, "learning_rate": 2.945843602855215e-06, "loss": 0.11583073139190674, "step": 28730 }, { "epoch": 0.7570554126893708, "grad_norm": 2.4820563793182373, "learning_rate": 2.9397991859816588e-06, "loss": 0.1262313961982727, "step": 28740 }, { "epoch": 0.7573188279338696, "grad_norm": 2.1924362182617188, "learning_rate": 2.9337599078268597e-06, "loss": 0.1113595962524414, "step": 28750 }, { "epoch": 0.7575822431783683, "grad_norm": 3.3483166694641113, "learning_rate": 2.927725772786464e-06, "loss": 0.1300480008125305, "step": 28760 }, { "epoch": 0.7578456584228671, "grad_norm": 1.4967621564865112, "learning_rate": 2.9216967852523747e-06, "loss": 0.11118640899658203, "step": 28770 }, { "epoch": 0.7581090736673658, "grad_norm": 2.4320836067199707, "learning_rate": 2.915672949612749e-06, "loss": 0.11584283113479614, "step": 28780 }, { "epoch": 0.7583724889118646, "grad_norm": 1.7884893417358398, "learning_rate": 2.9096542702519904e-06, "loss": 0.10830364227294922, "step": 28790 }, { "epoch": 0.7586359041563633, "grad_norm": 1.993897795677185, "learning_rate": 2.9036407515507613e-06, "loss": 0.1033976674079895, "step": 28800 }, { "epoch": 0.758899319400862, "grad_norm": 1.7900561094284058, "learning_rate": 2.89763239788596e-06, "loss": 0.12433733940124511, "step": 28810 }, { "epoch": 0.7591627346453608, "grad_norm": 3.4135828018188477, "learning_rate": 2.8916292136307147e-06, "loss": 0.11879667043685913, "step": 28820 }, { "epoch": 0.7594261498898595, "grad_norm": 2.0647644996643066, "learning_rate": 2.885631203154412e-06, "loss": 0.12654017210006713, "step": 28830 }, { "epoch": 0.7596895651343583, "grad_norm": 3.9343743324279785, "learning_rate": 2.879638370822656e-06, "loss": 0.1240350604057312, "step": 28840 }, { "epoch": 0.759952980378857, "grad_norm": 2.178617000579834, "learning_rate": 2.8736507209972996e-06, "loss": 0.10103265047073365, "step": 28850 }, { "epoch": 0.7602163956233557, "grad_norm": 2.625030040740967, "learning_rate": 2.867668258036401e-06, "loss": 0.10894306898117065, "step": 28860 }, { "epoch": 0.7604798108678544, "grad_norm": 2.0165348052978516, "learning_rate": 2.8616909862942565e-06, "loss": 0.11187762022018433, "step": 28870 }, { "epoch": 0.7607432261123532, "grad_norm": 2.1292896270751953, "learning_rate": 2.855718910121389e-06, "loss": 0.10809912681579589, "step": 28880 }, { "epoch": 0.7610066413568519, "grad_norm": 2.4068758487701416, "learning_rate": 2.8497520338645278e-06, "loss": 0.09654272198677064, "step": 28890 }, { "epoch": 0.7612700566013506, "grad_norm": 2.4564967155456543, "learning_rate": 2.843790361866625e-06, "loss": 0.10263981819152831, "step": 28900 }, { "epoch": 0.7615334718458494, "grad_norm": 1.8024176359176636, "learning_rate": 2.837833898466842e-06, "loss": 0.10493454933166504, "step": 28910 }, { "epoch": 0.7617968870903481, "grad_norm": 2.3879141807556152, "learning_rate": 2.831882648000547e-06, "loss": 0.11217247247695923, "step": 28920 }, { "epoch": 0.7620603023348469, "grad_norm": 2.5662920475006104, "learning_rate": 2.8259366147993217e-06, "loss": 0.11630409955978394, "step": 28930 }, { "epoch": 0.7623237175793456, "grad_norm": 2.0968868732452393, "learning_rate": 2.819995803190948e-06, "loss": 0.12080391645431518, "step": 28940 }, { "epoch": 0.7625871328238444, "grad_norm": 2.8533499240875244, "learning_rate": 2.8140602174993927e-06, "loss": 0.10860865116119385, "step": 28950 }, { "epoch": 0.7628505480683431, "grad_norm": 2.897923707962036, "learning_rate": 2.8081298620448427e-06, "loss": 0.1143236517906189, "step": 28960 }, { "epoch": 0.7631139633128419, "grad_norm": 2.4706432819366455, "learning_rate": 2.8022047411436604e-06, "loss": 0.09931314587593079, "step": 28970 }, { "epoch": 0.7633773785573406, "grad_norm": 3.0773112773895264, "learning_rate": 2.7962848591084122e-06, "loss": 0.12099273204803467, "step": 28980 }, { "epoch": 0.7636407938018392, "grad_norm": 2.8326029777526855, "learning_rate": 2.7903702202478355e-06, "loss": 0.1126554250717163, "step": 28990 }, { "epoch": 0.763904209046338, "grad_norm": 3.459909200668335, "learning_rate": 2.7844608288668586e-06, "loss": 0.10473830699920654, "step": 29000 }, { "epoch": 0.7641676242908367, "grad_norm": 2.5066370964050293, "learning_rate": 2.7785566892665993e-06, "loss": 0.1242222785949707, "step": 29010 }, { "epoch": 0.7644310395353355, "grad_norm": 2.243896484375, "learning_rate": 2.77265780574434e-06, "loss": 0.1154629945755005, "step": 29020 }, { "epoch": 0.7646944547798342, "grad_norm": 2.033327341079712, "learning_rate": 2.766764182593544e-06, "loss": 0.11100831031799316, "step": 29030 }, { "epoch": 0.764957870024333, "grad_norm": 1.9738445281982422, "learning_rate": 2.760875824103846e-06, "loss": 0.11293950080871581, "step": 29040 }, { "epoch": 0.7652212852688317, "grad_norm": 3.134652853012085, "learning_rate": 2.7549927345610407e-06, "loss": 0.10852689743041992, "step": 29050 }, { "epoch": 0.7654847005133305, "grad_norm": 2.4934139251708984, "learning_rate": 2.749114918247102e-06, "loss": 0.11488784551620483, "step": 29060 }, { "epoch": 0.7657481157578292, "grad_norm": 2.397547721862793, "learning_rate": 2.7432423794401598e-06, "loss": 0.10694299936294556, "step": 29070 }, { "epoch": 0.7660115310023279, "grad_norm": 1.896187663078308, "learning_rate": 2.7373751224144905e-06, "loss": 0.1167593240737915, "step": 29080 }, { "epoch": 0.7662749462468267, "grad_norm": 2.0560879707336426, "learning_rate": 2.731513151440547e-06, "loss": 0.11243681907653809, "step": 29090 }, { "epoch": 0.7665383614913254, "grad_norm": 1.6918058395385742, "learning_rate": 2.7256564707849165e-06, "loss": 0.11035823822021484, "step": 29100 }, { "epoch": 0.7668017767358242, "grad_norm": 3.231649398803711, "learning_rate": 2.719805084710356e-06, "loss": 0.13140416145324707, "step": 29110 }, { "epoch": 0.7670651919803229, "grad_norm": 2.006213665008545, "learning_rate": 2.7139589974757454e-06, "loss": 0.1012069821357727, "step": 29120 }, { "epoch": 0.7673286072248217, "grad_norm": 1.9243407249450684, "learning_rate": 2.7081182133361207e-06, "loss": 0.11563574075698853, "step": 29130 }, { "epoch": 0.7675920224693203, "grad_norm": 1.804592490196228, "learning_rate": 2.7022827365426606e-06, "loss": 0.12162514925003051, "step": 29140 }, { "epoch": 0.7678554377138191, "grad_norm": 1.7549705505371094, "learning_rate": 2.696452571342676e-06, "loss": 0.11550238132476806, "step": 29150 }, { "epoch": 0.7681188529583178, "grad_norm": 1.9671651124954224, "learning_rate": 2.6906277219796116e-06, "loss": 0.10846978425979614, "step": 29160 }, { "epoch": 0.7683822682028165, "grad_norm": 2.3293073177337646, "learning_rate": 2.6848081926930447e-06, "loss": 0.09691280722618104, "step": 29170 }, { "epoch": 0.7686456834473153, "grad_norm": 3.44612979888916, "learning_rate": 2.6789939877186753e-06, "loss": 0.11552159786224366, "step": 29180 }, { "epoch": 0.768909098691814, "grad_norm": 2.296377658843994, "learning_rate": 2.67318511128834e-06, "loss": 0.11114543676376343, "step": 29190 }, { "epoch": 0.7691725139363128, "grad_norm": 3.061239719390869, "learning_rate": 2.66738156762999e-06, "loss": 0.10022697448730469, "step": 29200 }, { "epoch": 0.7694359291808115, "grad_norm": 3.656698703765869, "learning_rate": 2.6615833609676846e-06, "loss": 0.11863152980804444, "step": 29210 }, { "epoch": 0.7696993444253103, "grad_norm": 2.2706425189971924, "learning_rate": 2.6557904955216185e-06, "loss": 0.11827702522277832, "step": 29220 }, { "epoch": 0.769962759669809, "grad_norm": 2.0737667083740234, "learning_rate": 2.650002975508086e-06, "loss": 0.10497589111328125, "step": 29230 }, { "epoch": 0.7702261749143078, "grad_norm": 2.3975417613983154, "learning_rate": 2.6442208051394936e-06, "loss": 0.11058510541915893, "step": 29240 }, { "epoch": 0.7704895901588065, "grad_norm": 3.0971128940582275, "learning_rate": 2.638443988624355e-06, "loss": 0.108311128616333, "step": 29250 }, { "epoch": 0.7707530054033052, "grad_norm": 2.15234375, "learning_rate": 2.632672530167283e-06, "loss": 0.1030653476715088, "step": 29260 }, { "epoch": 0.771016420647804, "grad_norm": 2.191063642501831, "learning_rate": 2.6269064339689986e-06, "loss": 0.10965341329574585, "step": 29270 }, { "epoch": 0.7712798358923026, "grad_norm": 1.9766387939453125, "learning_rate": 2.6211457042263157e-06, "loss": 0.11151266098022461, "step": 29280 }, { "epoch": 0.7715432511368014, "grad_norm": 2.005147695541382, "learning_rate": 2.6153903451321384e-06, "loss": 0.1110039234161377, "step": 29290 }, { "epoch": 0.7718066663813001, "grad_norm": 2.536606550216675, "learning_rate": 2.6096403608754685e-06, "loss": 0.09875606298446656, "step": 29300 }, { "epoch": 0.7720700816257989, "grad_norm": 2.2655258178710938, "learning_rate": 2.603895755641387e-06, "loss": 0.11897778511047363, "step": 29310 }, { "epoch": 0.7723334968702976, "grad_norm": 1.7213408946990967, "learning_rate": 2.598156533611077e-06, "loss": 0.10123322010040284, "step": 29320 }, { "epoch": 0.7725969121147963, "grad_norm": 4.293367385864258, "learning_rate": 2.592422698961781e-06, "loss": 0.11270835399627685, "step": 29330 }, { "epoch": 0.7728603273592951, "grad_norm": 2.3319203853607178, "learning_rate": 2.586694255866831e-06, "loss": 0.11728527545928955, "step": 29340 }, { "epoch": 0.7731237426037938, "grad_norm": 2.9056122303009033, "learning_rate": 2.580971208495642e-06, "loss": 0.10089658498764038, "step": 29350 }, { "epoch": 0.7733871578482926, "grad_norm": 3.7555150985717773, "learning_rate": 2.575253561013692e-06, "loss": 0.10769236087799072, "step": 29360 }, { "epoch": 0.7736505730927913, "grad_norm": 2.1866583824157715, "learning_rate": 2.56954131758253e-06, "loss": 0.09659603834152222, "step": 29370 }, { "epoch": 0.7739139883372901, "grad_norm": 3.6087825298309326, "learning_rate": 2.5638344823597762e-06, "loss": 0.11496410369873047, "step": 29380 }, { "epoch": 0.7741774035817888, "grad_norm": 2.062878131866455, "learning_rate": 2.558133059499105e-06, "loss": 0.11737377643585205, "step": 29390 }, { "epoch": 0.7744408188262876, "grad_norm": 3.376284599304199, "learning_rate": 2.552437053150266e-06, "loss": 0.10573241710662842, "step": 29400 }, { "epoch": 0.7747042340707863, "grad_norm": 1.9481908082962036, "learning_rate": 2.546746467459058e-06, "loss": 0.10678396224975586, "step": 29410 }, { "epoch": 0.774967649315285, "grad_norm": 2.9938035011291504, "learning_rate": 2.541061306567325e-06, "loss": 0.11831717491149903, "step": 29420 }, { "epoch": 0.7752310645597837, "grad_norm": 2.7529876232147217, "learning_rate": 2.5353815746129827e-06, "loss": 0.11008305549621582, "step": 29430 }, { "epoch": 0.7754944798042824, "grad_norm": 3.1456923484802246, "learning_rate": 2.529707275729979e-06, "loss": 0.11764094829559327, "step": 29440 }, { "epoch": 0.7757578950487812, "grad_norm": 4.011447906494141, "learning_rate": 2.524038414048321e-06, "loss": 0.12493470907211304, "step": 29450 }, { "epoch": 0.7760213102932799, "grad_norm": 3.200214147567749, "learning_rate": 2.518374993694043e-06, "loss": 0.11393265724182129, "step": 29460 }, { "epoch": 0.7762847255377787, "grad_norm": 2.8193418979644775, "learning_rate": 2.5127170187892257e-06, "loss": 0.11201090812683105, "step": 29470 }, { "epoch": 0.7765481407822774, "grad_norm": 2.821925401687622, "learning_rate": 2.507064493451995e-06, "loss": 0.10073133707046508, "step": 29480 }, { "epoch": 0.7768115560267762, "grad_norm": 2.3521409034729004, "learning_rate": 2.501417421796498e-06, "loss": 0.11593252420425415, "step": 29490 }, { "epoch": 0.7770749712712749, "grad_norm": 2.349168062210083, "learning_rate": 2.4957758079329173e-06, "loss": 0.11096611022949218, "step": 29500 }, { "epoch": 0.7773383865157736, "grad_norm": 2.3547239303588867, "learning_rate": 2.490139655967465e-06, "loss": 0.11986613273620605, "step": 29510 }, { "epoch": 0.7776018017602724, "grad_norm": 2.3140182495117188, "learning_rate": 2.4845089700023694e-06, "loss": 0.10894260406494141, "step": 29520 }, { "epoch": 0.7778652170047711, "grad_norm": 3.27461314201355, "learning_rate": 2.4788837541358957e-06, "loss": 0.10703049898147583, "step": 29530 }, { "epoch": 0.7781286322492699, "grad_norm": 3.224018096923828, "learning_rate": 2.4732640124623152e-06, "loss": 0.11081829071044921, "step": 29540 }, { "epoch": 0.7783920474937686, "grad_norm": 2.461188793182373, "learning_rate": 2.467649749071913e-06, "loss": 0.10352194309234619, "step": 29550 }, { "epoch": 0.7786554627382674, "grad_norm": 1.9940780401229858, "learning_rate": 2.4620409680509993e-06, "loss": 0.11066265106201172, "step": 29560 }, { "epoch": 0.778918877982766, "grad_norm": 3.4894983768463135, "learning_rate": 2.4564376734818818e-06, "loss": 0.11850197315216064, "step": 29570 }, { "epoch": 0.7791822932272648, "grad_norm": 2.288437604904175, "learning_rate": 2.450839869442887e-06, "loss": 0.10499577522277832, "step": 29580 }, { "epoch": 0.7794457084717635, "grad_norm": 3.409237861633301, "learning_rate": 2.4452475600083323e-06, "loss": 0.09759896993637085, "step": 29590 }, { "epoch": 0.7797091237162622, "grad_norm": 2.4664804935455322, "learning_rate": 2.4396607492485393e-06, "loss": 0.106037175655365, "step": 29600 }, { "epoch": 0.779972538960761, "grad_norm": 2.680910110473633, "learning_rate": 2.4340794412298364e-06, "loss": 0.12185890674591064, "step": 29610 }, { "epoch": 0.7802359542052597, "grad_norm": 2.783550977706909, "learning_rate": 2.428503640014539e-06, "loss": 0.10264465808868409, "step": 29620 }, { "epoch": 0.7804993694497585, "grad_norm": 2.6537022590637207, "learning_rate": 2.4229333496609532e-06, "loss": 0.10883820056915283, "step": 29630 }, { "epoch": 0.7807627846942572, "grad_norm": 3.022343635559082, "learning_rate": 2.4173685742233787e-06, "loss": 0.11048700809478759, "step": 29640 }, { "epoch": 0.781026199938756, "grad_norm": 2.6531906127929688, "learning_rate": 2.411809317752095e-06, "loss": 0.10439932346343994, "step": 29650 }, { "epoch": 0.7812896151832547, "grad_norm": 2.3318631649017334, "learning_rate": 2.4062555842933753e-06, "loss": 0.10644493103027344, "step": 29660 }, { "epoch": 0.7815530304277535, "grad_norm": 2.3605947494506836, "learning_rate": 2.4007073778894665e-06, "loss": 0.09790964722633362, "step": 29670 }, { "epoch": 0.7818164456722522, "grad_norm": 2.4237351417541504, "learning_rate": 2.395164702578583e-06, "loss": 0.10879884958267212, "step": 29680 }, { "epoch": 0.7820798609167509, "grad_norm": 3.734762191772461, "learning_rate": 2.3896275623949327e-06, "loss": 0.10747514963150025, "step": 29690 }, { "epoch": 0.7823432761612497, "grad_norm": 2.6911673545837402, "learning_rate": 2.38409596136868e-06, "loss": 0.10710868835449219, "step": 29700 }, { "epoch": 0.7826066914057483, "grad_norm": 2.1213698387145996, "learning_rate": 2.378569903525969e-06, "loss": 0.12418334484100342, "step": 29710 }, { "epoch": 0.7828701066502471, "grad_norm": 2.937364101409912, "learning_rate": 2.3730493928888963e-06, "loss": 0.12063025236129761, "step": 29720 }, { "epoch": 0.7831335218947458, "grad_norm": 2.9223787784576416, "learning_rate": 2.367534433475527e-06, "loss": 0.11166620254516602, "step": 29730 }, { "epoch": 0.7833969371392446, "grad_norm": 2.568140983581543, "learning_rate": 2.3620250292998925e-06, "loss": 0.11525968313217164, "step": 29740 }, { "epoch": 0.7836603523837433, "grad_norm": 2.016450881958008, "learning_rate": 2.356521184371973e-06, "loss": 0.11536678075790405, "step": 29750 }, { "epoch": 0.7839237676282421, "grad_norm": 2.142179489135742, "learning_rate": 2.3510229026977018e-06, "loss": 0.10072354078292847, "step": 29760 }, { "epoch": 0.7841871828727408, "grad_norm": 2.963527202606201, "learning_rate": 2.345530188278967e-06, "loss": 0.10107452869415283, "step": 29770 }, { "epoch": 0.7844505981172395, "grad_norm": 1.5970536470413208, "learning_rate": 2.3400430451136e-06, "loss": 0.10615906715393067, "step": 29780 }, { "epoch": 0.7847140133617383, "grad_norm": 4.012382507324219, "learning_rate": 2.334561477195387e-06, "loss": 0.09698750972747802, "step": 29790 }, { "epoch": 0.784977428606237, "grad_norm": 2.520275831222534, "learning_rate": 2.329085488514049e-06, "loss": 0.10940322875976563, "step": 29800 }, { "epoch": 0.7852408438507358, "grad_norm": 2.6226890087127686, "learning_rate": 2.3236150830552374e-06, "loss": 0.09872499704360962, "step": 29810 }, { "epoch": 0.7855042590952345, "grad_norm": 3.6652467250823975, "learning_rate": 2.3181502648005593e-06, "loss": 0.10678308010101319, "step": 29820 }, { "epoch": 0.7857676743397333, "grad_norm": 2.8423733711242676, "learning_rate": 2.3126910377275415e-06, "loss": 0.09675276279449463, "step": 29830 }, { "epoch": 0.786031089584232, "grad_norm": 3.01383113861084, "learning_rate": 2.3072374058096447e-06, "loss": 0.12717561721801757, "step": 29840 }, { "epoch": 0.7862945048287308, "grad_norm": 2.2628254890441895, "learning_rate": 2.301789373016259e-06, "loss": 0.1010403037071228, "step": 29850 }, { "epoch": 0.7865579200732294, "grad_norm": 3.0944559574127197, "learning_rate": 2.296346943312693e-06, "loss": 0.10773305892944336, "step": 29860 }, { "epoch": 0.7868213353177281, "grad_norm": 3.671419620513916, "learning_rate": 2.290910120660188e-06, "loss": 0.11660845279693603, "step": 29870 }, { "epoch": 0.7870847505622269, "grad_norm": 2.9823126792907715, "learning_rate": 2.285478909015896e-06, "loss": 0.10420269966125488, "step": 29880 }, { "epoch": 0.7873481658067256, "grad_norm": 2.6680471897125244, "learning_rate": 2.2800533123328874e-06, "loss": 0.11924365758895875, "step": 29890 }, { "epoch": 0.7876115810512244, "grad_norm": 2.4877419471740723, "learning_rate": 2.2746333345601435e-06, "loss": 0.10594706535339356, "step": 29900 }, { "epoch": 0.7878749962957231, "grad_norm": 2.8672549724578857, "learning_rate": 2.269218979642558e-06, "loss": 0.09330859184265136, "step": 29910 }, { "epoch": 0.7881384115402219, "grad_norm": 2.630664110183716, "learning_rate": 2.2638102515209382e-06, "loss": 0.10929331779479981, "step": 29920 }, { "epoch": 0.7884018267847206, "grad_norm": 2.7214691638946533, "learning_rate": 2.258407154131982e-06, "loss": 0.10106037855148316, "step": 29930 }, { "epoch": 0.7886652420292193, "grad_norm": 2.3114869594573975, "learning_rate": 2.2530096914082965e-06, "loss": 0.1170762062072754, "step": 29940 }, { "epoch": 0.7889286572737181, "grad_norm": 2.354624032974243, "learning_rate": 2.2476178672783943e-06, "loss": 0.10511891841888428, "step": 29950 }, { "epoch": 0.7891920725182168, "grad_norm": 3.261267900466919, "learning_rate": 2.242231685666675e-06, "loss": 0.09963949322700501, "step": 29960 }, { "epoch": 0.7894554877627156, "grad_norm": 3.3288581371307373, "learning_rate": 2.2368511504934333e-06, "loss": 0.1086653709411621, "step": 29970 }, { "epoch": 0.7897189030072143, "grad_norm": 2.2864418029785156, "learning_rate": 2.2314762656748544e-06, "loss": 0.09772620797157287, "step": 29980 }, { "epoch": 0.7899823182517131, "grad_norm": 2.034999370574951, "learning_rate": 2.2261070351230107e-06, "loss": 0.11046223640441895, "step": 29990 }, { "epoch": 0.7902457334962117, "grad_norm": 2.6005492210388184, "learning_rate": 2.2207434627458626e-06, "loss": 0.11600444316864014, "step": 30000 }, { "epoch": 0.7905091487407105, "grad_norm": 3.192762851715088, "learning_rate": 2.2153855524472513e-06, "loss": 0.12208173274993897, "step": 30010 }, { "epoch": 0.7907725639852092, "grad_norm": 1.8556777238845825, "learning_rate": 2.2100333081268853e-06, "loss": 0.10692219734191895, "step": 30020 }, { "epoch": 0.7910359792297079, "grad_norm": 2.685199499130249, "learning_rate": 2.204686733680369e-06, "loss": 0.10056805610656738, "step": 30030 }, { "epoch": 0.7912993944742067, "grad_norm": 3.1381890773773193, "learning_rate": 2.1993458329991635e-06, "loss": 0.10735368728637695, "step": 30040 }, { "epoch": 0.7915628097187054, "grad_norm": 2.313900947570801, "learning_rate": 2.1940106099706137e-06, "loss": 0.1072955846786499, "step": 30050 }, { "epoch": 0.7918262249632042, "grad_norm": 2.1363394260406494, "learning_rate": 2.1886810684779204e-06, "loss": 0.11036525964736939, "step": 30060 }, { "epoch": 0.7920896402077029, "grad_norm": 2.485612392425537, "learning_rate": 2.18335721240015e-06, "loss": 0.1125748634338379, "step": 30070 }, { "epoch": 0.7923530554522017, "grad_norm": 2.8685975074768066, "learning_rate": 2.1780390456122424e-06, "loss": 0.10036498308181763, "step": 30080 }, { "epoch": 0.7926164706967004, "grad_norm": 2.445363759994507, "learning_rate": 2.1727265719849853e-06, "loss": 0.11574465036392212, "step": 30090 }, { "epoch": 0.7928798859411992, "grad_norm": 2.278172254562378, "learning_rate": 2.1674197953850274e-06, "loss": 0.1053774118423462, "step": 30100 }, { "epoch": 0.7931433011856979, "grad_norm": 2.887505292892456, "learning_rate": 2.162118719674867e-06, "loss": 0.11153759956359863, "step": 30110 }, { "epoch": 0.7934067164301966, "grad_norm": 2.434471607208252, "learning_rate": 2.1568233487128556e-06, "loss": 0.09627209901809693, "step": 30120 }, { "epoch": 0.7936701316746954, "grad_norm": 4.547861099243164, "learning_rate": 2.1515336863531976e-06, "loss": 0.11105449199676513, "step": 30130 }, { "epoch": 0.793933546919194, "grad_norm": 2.2925662994384766, "learning_rate": 2.1462497364459356e-06, "loss": 0.10409353971481324, "step": 30140 }, { "epoch": 0.7941969621636928, "grad_norm": 2.7002015113830566, "learning_rate": 2.1409715028369504e-06, "loss": 0.09813432693481446, "step": 30150 }, { "epoch": 0.7944603774081915, "grad_norm": 2.3236987590789795, "learning_rate": 2.135698989367976e-06, "loss": 0.10631202459335327, "step": 30160 }, { "epoch": 0.7947237926526903, "grad_norm": 2.616173505783081, "learning_rate": 2.1304321998765685e-06, "loss": 0.09367958307266236, "step": 30170 }, { "epoch": 0.794987207897189, "grad_norm": 2.2020304203033447, "learning_rate": 2.1251711381961336e-06, "loss": 0.10579628944396972, "step": 30180 }, { "epoch": 0.7952506231416878, "grad_norm": 3.454751491546631, "learning_rate": 2.11991580815589e-06, "loss": 0.10579404830932618, "step": 30190 }, { "epoch": 0.7955140383861865, "grad_norm": 3.1630706787109375, "learning_rate": 2.1146662135808948e-06, "loss": 0.1172917366027832, "step": 30200 }, { "epoch": 0.7957774536306852, "grad_norm": 2.855215311050415, "learning_rate": 2.1094223582920346e-06, "loss": 0.1035490870475769, "step": 30210 }, { "epoch": 0.796040868875184, "grad_norm": 3.468106985092163, "learning_rate": 2.1041842461060103e-06, "loss": 0.1011164903640747, "step": 30220 }, { "epoch": 0.7963042841196827, "grad_norm": 2.0357789993286133, "learning_rate": 2.098951880835346e-06, "loss": 0.10744737386703491, "step": 30230 }, { "epoch": 0.7965676993641815, "grad_norm": 3.0111899375915527, "learning_rate": 2.093725266288382e-06, "loss": 0.10809216499328614, "step": 30240 }, { "epoch": 0.7968311146086802, "grad_norm": 3.7896180152893066, "learning_rate": 2.088504406269274e-06, "loss": 0.11107795238494873, "step": 30250 }, { "epoch": 0.797094529853179, "grad_norm": 3.480806350708008, "learning_rate": 2.083289304577992e-06, "loss": 0.102664315700531, "step": 30260 }, { "epoch": 0.7973579450976777, "grad_norm": 2.9756250381469727, "learning_rate": 2.0780799650103123e-06, "loss": 0.09968578815460205, "step": 30270 }, { "epoch": 0.7976213603421765, "grad_norm": 3.595695734024048, "learning_rate": 2.0728763913578097e-06, "loss": 0.11000127792358398, "step": 30280 }, { "epoch": 0.7978847755866751, "grad_norm": 3.0821237564086914, "learning_rate": 2.0676785874078773e-06, "loss": 0.0945256233215332, "step": 30290 }, { "epoch": 0.7981481908311738, "grad_norm": 2.4611165523529053, "learning_rate": 2.062486556943699e-06, "loss": 0.1084458589553833, "step": 30300 }, { "epoch": 0.7984116060756726, "grad_norm": 2.517066478729248, "learning_rate": 2.0573003037442575e-06, "loss": 0.11476397514343262, "step": 30310 }, { "epoch": 0.7986750213201713, "grad_norm": 2.4839847087860107, "learning_rate": 2.0521198315843315e-06, "loss": 0.10807243585586548, "step": 30320 }, { "epoch": 0.7989384365646701, "grad_norm": 2.576936960220337, "learning_rate": 2.046945144234491e-06, "loss": 0.10655298233032226, "step": 30330 }, { "epoch": 0.7992018518091688, "grad_norm": 2.318402051925659, "learning_rate": 2.0417762454611012e-06, "loss": 0.122242271900177, "step": 30340 }, { "epoch": 0.7994652670536676, "grad_norm": 2.898024797439575, "learning_rate": 2.0366131390263055e-06, "loss": 0.10488355159759521, "step": 30350 }, { "epoch": 0.7997286822981663, "grad_norm": 2.5165042877197266, "learning_rate": 2.0314558286880383e-06, "loss": 0.10178505182266236, "step": 30360 }, { "epoch": 0.7999920975426651, "grad_norm": 3.0491652488708496, "learning_rate": 2.026304318200011e-06, "loss": 0.11178822517395019, "step": 30370 }, { "epoch": 0.8002555127871638, "grad_norm": 2.5125551223754883, "learning_rate": 2.021158611311712e-06, "loss": 0.1088036298751831, "step": 30380 }, { "epoch": 0.8005189280316625, "grad_norm": 2.2249317169189453, "learning_rate": 2.016018711768416e-06, "loss": 0.11351873874664306, "step": 30390 }, { "epoch": 0.8007823432761613, "grad_norm": 2.4616634845733643, "learning_rate": 2.010884623311162e-06, "loss": 0.11286170482635498, "step": 30400 }, { "epoch": 0.80104575852066, "grad_norm": 2.510443925857544, "learning_rate": 2.005756349676753e-06, "loss": 0.11005617380142212, "step": 30410 }, { "epoch": 0.8013091737651588, "grad_norm": 2.519986629486084, "learning_rate": 2.0006338945977777e-06, "loss": 0.09244897365570068, "step": 30420 }, { "epoch": 0.8015725890096574, "grad_norm": 2.9474809169769287, "learning_rate": 1.9955172618025743e-06, "loss": 0.10150737762451172, "step": 30430 }, { "epoch": 0.8018360042541562, "grad_norm": 2.9609932899475098, "learning_rate": 1.9904064550152515e-06, "loss": 0.08904974460601807, "step": 30440 }, { "epoch": 0.8020994194986549, "grad_norm": 2.7996585369110107, "learning_rate": 1.9853014779556734e-06, "loss": 0.10335123538970947, "step": 30450 }, { "epoch": 0.8023628347431536, "grad_norm": 2.9989302158355713, "learning_rate": 1.9802023343394606e-06, "loss": 0.09866828918457031, "step": 30460 }, { "epoch": 0.8026262499876524, "grad_norm": 3.12789249420166, "learning_rate": 1.975109027877995e-06, "loss": 0.11745095252990723, "step": 30470 }, { "epoch": 0.8028896652321511, "grad_norm": 2.2997303009033203, "learning_rate": 1.9700215622784026e-06, "loss": 0.09210169315338135, "step": 30480 }, { "epoch": 0.8031530804766499, "grad_norm": 4.367586135864258, "learning_rate": 1.96493994124356e-06, "loss": 0.10476797819137573, "step": 30490 }, { "epoch": 0.8034164957211486, "grad_norm": 2.509192943572998, "learning_rate": 1.9598641684720897e-06, "loss": 0.09971163272857667, "step": 30500 }, { "epoch": 0.8036799109656474, "grad_norm": 2.294102668762207, "learning_rate": 1.954794247658356e-06, "loss": 0.10897142887115478, "step": 30510 }, { "epoch": 0.8039433262101461, "grad_norm": 1.9165219068527222, "learning_rate": 1.949730182492473e-06, "loss": 0.09406293034553528, "step": 30520 }, { "epoch": 0.8042067414546449, "grad_norm": 2.4537103176116943, "learning_rate": 1.9446719766602793e-06, "loss": 0.0989997148513794, "step": 30530 }, { "epoch": 0.8044701566991436, "grad_norm": 4.4989776611328125, "learning_rate": 1.9396196338433547e-06, "loss": 0.09893377423286438, "step": 30540 }, { "epoch": 0.8047335719436423, "grad_norm": 2.8482415676116943, "learning_rate": 1.9345731577190187e-06, "loss": 0.10804245471954346, "step": 30550 }, { "epoch": 0.804996987188141, "grad_norm": 2.868272542953491, "learning_rate": 1.929532551960309e-06, "loss": 0.10072876214981079, "step": 30560 }, { "epoch": 0.8052604024326397, "grad_norm": 2.9341847896575928, "learning_rate": 1.9244978202359974e-06, "loss": 0.08783427476882935, "step": 30570 }, { "epoch": 0.8055238176771385, "grad_norm": 1.9072400331497192, "learning_rate": 1.9194689662105803e-06, "loss": 0.09450969099998474, "step": 30580 }, { "epoch": 0.8057872329216372, "grad_norm": 3.070880889892578, "learning_rate": 1.9144459935442705e-06, "loss": 0.10543134212493896, "step": 30590 }, { "epoch": 0.806050648166136, "grad_norm": 2.946713924407959, "learning_rate": 1.9094289058930095e-06, "loss": 0.1052016019821167, "step": 30600 }, { "epoch": 0.8063140634106347, "grad_norm": 2.391190528869629, "learning_rate": 1.9044177069084502e-06, "loss": 0.09661779999732971, "step": 30610 }, { "epoch": 0.8065774786551335, "grad_norm": 2.56866192817688, "learning_rate": 1.899412400237951e-06, "loss": 0.10393301248550416, "step": 30620 }, { "epoch": 0.8068408938996322, "grad_norm": 2.9661948680877686, "learning_rate": 1.8944129895246e-06, "loss": 0.10675811767578125, "step": 30630 }, { "epoch": 0.8071043091441309, "grad_norm": 2.4333927631378174, "learning_rate": 1.8894194784071761e-06, "loss": 0.10269018411636352, "step": 30640 }, { "epoch": 0.8073677243886297, "grad_norm": 3.3723835945129395, "learning_rate": 1.8844318705201815e-06, "loss": 0.10546915531158448, "step": 30650 }, { "epoch": 0.8076311396331284, "grad_norm": 2.9314193725585938, "learning_rate": 1.879450169493805e-06, "loss": 0.10539920330047607, "step": 30660 }, { "epoch": 0.8078945548776272, "grad_norm": 2.8310084342956543, "learning_rate": 1.874474378953942e-06, "loss": 0.11325669288635254, "step": 30670 }, { "epoch": 0.8081579701221259, "grad_norm": 4.254630088806152, "learning_rate": 1.8695045025221936e-06, "loss": 0.09810227751731873, "step": 30680 }, { "epoch": 0.8084213853666247, "grad_norm": 2.698230266571045, "learning_rate": 1.864540543815848e-06, "loss": 0.09105302095413208, "step": 30690 }, { "epoch": 0.8086848006111234, "grad_norm": 3.334937334060669, "learning_rate": 1.859582506447889e-06, "loss": 0.11127852201461792, "step": 30700 }, { "epoch": 0.8089482158556222, "grad_norm": 2.3899734020233154, "learning_rate": 1.8546303940269883e-06, "loss": 0.09654811024665833, "step": 30710 }, { "epoch": 0.8092116311001208, "grad_norm": 3.3910324573516846, "learning_rate": 1.8496842101575053e-06, "loss": 0.09783517122268677, "step": 30720 }, { "epoch": 0.8094750463446195, "grad_norm": 3.546205520629883, "learning_rate": 1.844743958439491e-06, "loss": 0.11701444387435914, "step": 30730 }, { "epoch": 0.8097384615891183, "grad_norm": 2.7835636138916016, "learning_rate": 1.839809642468674e-06, "loss": 0.09858607053756714, "step": 30740 }, { "epoch": 0.810001876833617, "grad_norm": 2.567781925201416, "learning_rate": 1.8348812658364557e-06, "loss": 0.1131644606590271, "step": 30750 }, { "epoch": 0.8102652920781158, "grad_norm": 2.0309479236602783, "learning_rate": 1.8299588321299266e-06, "loss": 0.10377429723739624, "step": 30760 }, { "epoch": 0.8105287073226145, "grad_norm": 2.8520472049713135, "learning_rate": 1.8250423449318434e-06, "loss": 0.10393083095550537, "step": 30770 }, { "epoch": 0.8107921225671133, "grad_norm": 2.665163040161133, "learning_rate": 1.8201318078206443e-06, "loss": 0.10926332473754882, "step": 30780 }, { "epoch": 0.811055537811612, "grad_norm": 2.530588150024414, "learning_rate": 1.8152272243704216e-06, "loss": 0.12028154134750366, "step": 30790 }, { "epoch": 0.8113189530561108, "grad_norm": 2.8257434368133545, "learning_rate": 1.8103285981509432e-06, "loss": 0.11227561235427856, "step": 30800 }, { "epoch": 0.8115823683006095, "grad_norm": 2.439746618270874, "learning_rate": 1.8054359327276473e-06, "loss": 0.10675585269927979, "step": 30810 }, { "epoch": 0.8118457835451082, "grad_norm": 2.3767528533935547, "learning_rate": 1.8005492316616224e-06, "loss": 0.10927174091339112, "step": 30820 }, { "epoch": 0.812109198789607, "grad_norm": 2.262613534927368, "learning_rate": 1.7956684985096217e-06, "loss": 0.08772720098495483, "step": 30830 }, { "epoch": 0.8123726140341057, "grad_norm": 2.8434524536132812, "learning_rate": 1.7907937368240525e-06, "loss": 0.09598644375801087, "step": 30840 }, { "epoch": 0.8126360292786045, "grad_norm": 2.717573404312134, "learning_rate": 1.7859249501529742e-06, "loss": 0.10657749176025391, "step": 30850 }, { "epoch": 0.8128994445231031, "grad_norm": 2.471322774887085, "learning_rate": 1.7810621420401054e-06, "loss": 0.10268096923828125, "step": 30860 }, { "epoch": 0.8131628597676019, "grad_norm": 2.8309786319732666, "learning_rate": 1.776205316024806e-06, "loss": 0.09967578053474427, "step": 30870 }, { "epoch": 0.8134262750121006, "grad_norm": 2.806623697280884, "learning_rate": 1.7713544756420785e-06, "loss": 0.09303456544876099, "step": 30880 }, { "epoch": 0.8136896902565994, "grad_norm": 2.068650245666504, "learning_rate": 1.76650962442258e-06, "loss": 0.09950463771820069, "step": 30890 }, { "epoch": 0.8139531055010981, "grad_norm": 3.1238865852355957, "learning_rate": 1.761670765892599e-06, "loss": 0.10960904359817505, "step": 30900 }, { "epoch": 0.8142165207455968, "grad_norm": 2.4190266132354736, "learning_rate": 1.7568379035740669e-06, "loss": 0.10383388996124268, "step": 30910 }, { "epoch": 0.8144799359900956, "grad_norm": 2.779986619949341, "learning_rate": 1.7520110409845493e-06, "loss": 0.09823440313339234, "step": 30920 }, { "epoch": 0.8147433512345943, "grad_norm": 4.431118011474609, "learning_rate": 1.7471901816372416e-06, "loss": 0.10847399234771729, "step": 30930 }, { "epoch": 0.8150067664790931, "grad_norm": 3.3221309185028076, "learning_rate": 1.7423753290409784e-06, "loss": 0.09764881730079651, "step": 30940 }, { "epoch": 0.8152701817235918, "grad_norm": 3.6144332885742188, "learning_rate": 1.7375664867002151e-06, "loss": 0.10380934476852417, "step": 30950 }, { "epoch": 0.8155335969680906, "grad_norm": 2.0358259677886963, "learning_rate": 1.7327636581150343e-06, "loss": 0.10285823345184326, "step": 30960 }, { "epoch": 0.8157970122125893, "grad_norm": 3.0127549171447754, "learning_rate": 1.7279668467811427e-06, "loss": 0.09986757636070251, "step": 30970 }, { "epoch": 0.816060427457088, "grad_norm": 3.439753293991089, "learning_rate": 1.7231760561898625e-06, "loss": 0.10693702697753907, "step": 30980 }, { "epoch": 0.8163238427015868, "grad_norm": 2.6502888202667236, "learning_rate": 1.718391289828144e-06, "loss": 0.08962771892547608, "step": 30990 }, { "epoch": 0.8165872579460854, "grad_norm": 2.782750129699707, "learning_rate": 1.7136125511785462e-06, "loss": 0.09536300897598267, "step": 31000 }, { "epoch": 0.8168506731905842, "grad_norm": 2.0651023387908936, "learning_rate": 1.7088398437192344e-06, "loss": 0.1086016297340393, "step": 31010 }, { "epoch": 0.8171140884350829, "grad_norm": 2.5064477920532227, "learning_rate": 1.7040731709239967e-06, "loss": 0.10460610389709472, "step": 31020 }, { "epoch": 0.8173775036795817, "grad_norm": 3.26031756401062, "learning_rate": 1.6993125362622231e-06, "loss": 0.10223197937011719, "step": 31030 }, { "epoch": 0.8176409189240804, "grad_norm": 2.871619701385498, "learning_rate": 1.6945579431989067e-06, "loss": 0.0966681718826294, "step": 31040 }, { "epoch": 0.8179043341685792, "grad_norm": 2.462939739227295, "learning_rate": 1.689809395194647e-06, "loss": 0.09830474257469177, "step": 31050 }, { "epoch": 0.8181677494130779, "grad_norm": 4.635070323944092, "learning_rate": 1.6850668957056393e-06, "loss": 0.11202728748321533, "step": 31060 }, { "epoch": 0.8184311646575766, "grad_norm": 2.9536008834838867, "learning_rate": 1.6803304481836835e-06, "loss": 0.09839040637016297, "step": 31070 }, { "epoch": 0.8186945799020754, "grad_norm": 3.3137617111206055, "learning_rate": 1.6756000560761687e-06, "loss": 0.09292766451835632, "step": 31080 }, { "epoch": 0.8189579951465741, "grad_norm": 2.108473777770996, "learning_rate": 1.6708757228260786e-06, "loss": 0.102767014503479, "step": 31090 }, { "epoch": 0.8192214103910729, "grad_norm": 2.558774471282959, "learning_rate": 1.6661574518719859e-06, "loss": 0.09790033102035522, "step": 31100 }, { "epoch": 0.8194848256355716, "grad_norm": 3.314307928085327, "learning_rate": 1.6614452466480502e-06, "loss": 0.09656890630722045, "step": 31110 }, { "epoch": 0.8197482408800704, "grad_norm": 2.7774038314819336, "learning_rate": 1.6567391105840247e-06, "loss": 0.09223272800445556, "step": 31120 }, { "epoch": 0.820011656124569, "grad_norm": 3.0064544677734375, "learning_rate": 1.6520390471052306e-06, "loss": 0.09546909332275391, "step": 31130 }, { "epoch": 0.8202750713690679, "grad_norm": 2.3583292961120605, "learning_rate": 1.647345059632578e-06, "loss": 0.08761156797409057, "step": 31140 }, { "epoch": 0.8205384866135665, "grad_norm": 3.1745731830596924, "learning_rate": 1.6426571515825563e-06, "loss": 0.11134417057037353, "step": 31150 }, { "epoch": 0.8208019018580652, "grad_norm": 3.7013065814971924, "learning_rate": 1.6379753263672272e-06, "loss": 0.10134365558624267, "step": 31160 }, { "epoch": 0.821065317102564, "grad_norm": 2.857827663421631, "learning_rate": 1.6332995873942237e-06, "loss": 0.09153382182121277, "step": 31170 }, { "epoch": 0.8213287323470627, "grad_norm": 3.5964601039886475, "learning_rate": 1.6286299380667504e-06, "loss": 0.09680215120315552, "step": 31180 }, { "epoch": 0.8215921475915615, "grad_norm": 3.6507956981658936, "learning_rate": 1.6239663817835783e-06, "loss": 0.099502694606781, "step": 31190 }, { "epoch": 0.8218555628360602, "grad_norm": 2.461007595062256, "learning_rate": 1.6193089219390478e-06, "loss": 0.10584900379180909, "step": 31200 }, { "epoch": 0.822118978080559, "grad_norm": 2.5571248531341553, "learning_rate": 1.6146575619230597e-06, "loss": 0.10369465351104737, "step": 31210 }, { "epoch": 0.8223823933250577, "grad_norm": 2.517164468765259, "learning_rate": 1.610012305121068e-06, "loss": 0.09121586084365844, "step": 31220 }, { "epoch": 0.8226458085695565, "grad_norm": 2.721524953842163, "learning_rate": 1.6053731549140982e-06, "loss": 0.11511334180831909, "step": 31230 }, { "epoch": 0.8229092238140552, "grad_norm": 2.647033929824829, "learning_rate": 1.6007401146787195e-06, "loss": 0.09477361440658569, "step": 31240 }, { "epoch": 0.8231726390585539, "grad_norm": 2.0958306789398193, "learning_rate": 1.596113187787064e-06, "loss": 0.08960262537002564, "step": 31250 }, { "epoch": 0.8234360543030527, "grad_norm": 2.9606618881225586, "learning_rate": 1.5914923776068025e-06, "loss": 0.0970408022403717, "step": 31260 }, { "epoch": 0.8236994695475514, "grad_norm": 3.0676398277282715, "learning_rate": 1.5868776875011604e-06, "loss": 0.09540683031082153, "step": 31270 }, { "epoch": 0.8239628847920502, "grad_norm": 3.267726421356201, "learning_rate": 1.582269120828912e-06, "loss": 0.11058946847915649, "step": 31280 }, { "epoch": 0.8242263000365488, "grad_norm": 2.885470390319824, "learning_rate": 1.5776666809443697e-06, "loss": 0.08832311630249023, "step": 31290 }, { "epoch": 0.8244897152810476, "grad_norm": 2.223916530609131, "learning_rate": 1.5730703711973883e-06, "loss": 0.0950249433517456, "step": 31300 }, { "epoch": 0.8247531305255463, "grad_norm": 2.500239133834839, "learning_rate": 1.5684801949333584e-06, "loss": 0.09579026103019714, "step": 31310 }, { "epoch": 0.8250165457700451, "grad_norm": 2.773388147354126, "learning_rate": 1.5638961554932064e-06, "loss": 0.10901179313659667, "step": 31320 }, { "epoch": 0.8252799610145438, "grad_norm": 3.070828437805176, "learning_rate": 1.5593182562134002e-06, "loss": 0.10586061477661132, "step": 31330 }, { "epoch": 0.8255433762590425, "grad_norm": 3.1626152992248535, "learning_rate": 1.554746500425931e-06, "loss": 0.09410136938095093, "step": 31340 }, { "epoch": 0.8258067915035413, "grad_norm": 1.904381513595581, "learning_rate": 1.5501808914583139e-06, "loss": 0.09774287939071655, "step": 31350 }, { "epoch": 0.82607020674804, "grad_norm": 2.494175910949707, "learning_rate": 1.5456214326336017e-06, "loss": 0.09666645526885986, "step": 31360 }, { "epoch": 0.8263336219925388, "grad_norm": 2.7019870281219482, "learning_rate": 1.5410681272703643e-06, "loss": 0.09711928963661194, "step": 31370 }, { "epoch": 0.8265970372370375, "grad_norm": 2.640991449356079, "learning_rate": 1.536520978682693e-06, "loss": 0.09706169366836548, "step": 31380 }, { "epoch": 0.8268604524815363, "grad_norm": 2.4714391231536865, "learning_rate": 1.531979990180199e-06, "loss": 0.09508795738220215, "step": 31390 }, { "epoch": 0.827123867726035, "grad_norm": 3.0128958225250244, "learning_rate": 1.5274451650680066e-06, "loss": 0.11591100692749023, "step": 31400 }, { "epoch": 0.8273872829705338, "grad_norm": 2.8003594875335693, "learning_rate": 1.5229165066467632e-06, "loss": 0.09871978759765625, "step": 31410 }, { "epoch": 0.8276506982150325, "grad_norm": 2.667059898376465, "learning_rate": 1.5183940182126189e-06, "loss": 0.09227006435394287, "step": 31420 }, { "epoch": 0.8279141134595311, "grad_norm": 2.6176540851593018, "learning_rate": 1.513877703057235e-06, "loss": 0.09697645902633667, "step": 31430 }, { "epoch": 0.8281775287040299, "grad_norm": 2.6246402263641357, "learning_rate": 1.509367564467782e-06, "loss": 0.09465501308441163, "step": 31440 }, { "epoch": 0.8284409439485286, "grad_norm": 3.3514156341552734, "learning_rate": 1.5048636057269294e-06, "loss": 0.1062354326248169, "step": 31450 }, { "epoch": 0.8287043591930274, "grad_norm": 2.6026742458343506, "learning_rate": 1.50036583011286e-06, "loss": 0.09965273141860961, "step": 31460 }, { "epoch": 0.8289677744375261, "grad_norm": 2.4056131839752197, "learning_rate": 1.4958742408992454e-06, "loss": 0.1022361159324646, "step": 31470 }, { "epoch": 0.8292311896820249, "grad_norm": 2.2939743995666504, "learning_rate": 1.4913888413552536e-06, "loss": 0.0958327293395996, "step": 31480 }, { "epoch": 0.8294946049265236, "grad_norm": 4.296524524688721, "learning_rate": 1.486909634745557e-06, "loss": 0.09076435565948486, "step": 31490 }, { "epoch": 0.8297580201710223, "grad_norm": 2.304107189178467, "learning_rate": 1.4824366243303156e-06, "loss": 0.09693774580955505, "step": 31500 }, { "epoch": 0.8300214354155211, "grad_norm": 2.7280690670013428, "learning_rate": 1.4779698133651766e-06, "loss": 0.0931594967842102, "step": 31510 }, { "epoch": 0.8302848506600198, "grad_norm": 3.146967887878418, "learning_rate": 1.47350920510128e-06, "loss": 0.09008336067199707, "step": 31520 }, { "epoch": 0.8305482659045186, "grad_norm": 2.9644622802734375, "learning_rate": 1.469054802785246e-06, "loss": 0.11130319833755493, "step": 31530 }, { "epoch": 0.8308116811490173, "grad_norm": 3.099189281463623, "learning_rate": 1.4646066096591838e-06, "loss": 0.10838453769683838, "step": 31540 }, { "epoch": 0.8310750963935161, "grad_norm": 3.1051485538482666, "learning_rate": 1.4601646289606809e-06, "loss": 0.10713218450546265, "step": 31550 }, { "epoch": 0.8313385116380148, "grad_norm": 2.7562172412872314, "learning_rate": 1.4557288639227995e-06, "loss": 0.08258674144744874, "step": 31560 }, { "epoch": 0.8316019268825136, "grad_norm": 2.4432590007781982, "learning_rate": 1.451299317774083e-06, "loss": 0.10032507181167602, "step": 31570 }, { "epoch": 0.8318653421270122, "grad_norm": 2.508864402770996, "learning_rate": 1.4468759937385423e-06, "loss": 0.08555483818054199, "step": 31580 }, { "epoch": 0.8321287573715109, "grad_norm": 3.6624183654785156, "learning_rate": 1.4424588950356688e-06, "loss": 0.10055992603302003, "step": 31590 }, { "epoch": 0.8323921726160097, "grad_norm": 2.5352301597595215, "learning_rate": 1.4380480248804185e-06, "loss": 0.12572193145751953, "step": 31600 }, { "epoch": 0.8326555878605084, "grad_norm": 3.3924739360809326, "learning_rate": 1.4336433864832033e-06, "loss": 0.10116487741470337, "step": 31610 }, { "epoch": 0.8329190031050072, "grad_norm": 3.5446181297302246, "learning_rate": 1.4292449830499189e-06, "loss": 0.09900656342506409, "step": 31620 }, { "epoch": 0.8331824183495059, "grad_norm": 2.388611316680908, "learning_rate": 1.4248528177819098e-06, "loss": 0.09408687353134156, "step": 31630 }, { "epoch": 0.8334458335940047, "grad_norm": 4.367746353149414, "learning_rate": 1.4204668938759835e-06, "loss": 0.08993055820465087, "step": 31640 }, { "epoch": 0.8337092488385034, "grad_norm": 3.0293173789978027, "learning_rate": 1.416087214524403e-06, "loss": 0.09723834991455078, "step": 31650 }, { "epoch": 0.8339726640830022, "grad_norm": 2.9721126556396484, "learning_rate": 1.4117137829148885e-06, "loss": 0.09694210290908814, "step": 31660 }, { "epoch": 0.8342360793275009, "grad_norm": 2.303568124771118, "learning_rate": 1.4073466022306148e-06, "loss": 0.09727546572685242, "step": 31670 }, { "epoch": 0.8344994945719996, "grad_norm": 3.0749824047088623, "learning_rate": 1.4029856756502026e-06, "loss": 0.10071616172790528, "step": 31680 }, { "epoch": 0.8347629098164984, "grad_norm": 2.6757242679595947, "learning_rate": 1.3986310063477215e-06, "loss": 0.10368860960006714, "step": 31690 }, { "epoch": 0.835026325060997, "grad_norm": 3.803001880645752, "learning_rate": 1.3942825974926888e-06, "loss": 0.0991098403930664, "step": 31700 }, { "epoch": 0.8352897403054959, "grad_norm": 2.6424739360809326, "learning_rate": 1.3899404522500626e-06, "loss": 0.09818837642669678, "step": 31710 }, { "epoch": 0.8355531555499945, "grad_norm": 3.034985303878784, "learning_rate": 1.385604573780248e-06, "loss": 0.08997452259063721, "step": 31720 }, { "epoch": 0.8358165707944933, "grad_norm": 2.5950565338134766, "learning_rate": 1.38127496523908e-06, "loss": 0.10266939401626587, "step": 31730 }, { "epoch": 0.836079986038992, "grad_norm": 2.3186094760894775, "learning_rate": 1.376951629777833e-06, "loss": 0.08941137790679932, "step": 31740 }, { "epoch": 0.8363434012834908, "grad_norm": 3.3729772567749023, "learning_rate": 1.372634570543222e-06, "loss": 0.09791977405548095, "step": 31750 }, { "epoch": 0.8366068165279895, "grad_norm": 2.451815128326416, "learning_rate": 1.3683237906773873e-06, "loss": 0.10602235794067383, "step": 31760 }, { "epoch": 0.8368702317724882, "grad_norm": 2.61027455329895, "learning_rate": 1.3640192933178997e-06, "loss": 0.0956148624420166, "step": 31770 }, { "epoch": 0.837133647016987, "grad_norm": 3.0501058101654053, "learning_rate": 1.359721081597758e-06, "loss": 0.09801044464111328, "step": 31780 }, { "epoch": 0.8373970622614857, "grad_norm": 3.625415563583374, "learning_rate": 1.355429158645384e-06, "loss": 0.09642842411994934, "step": 31790 }, { "epoch": 0.8376604775059845, "grad_norm": 2.5537993907928467, "learning_rate": 1.3511435275846297e-06, "loss": 0.10891262292861939, "step": 31800 }, { "epoch": 0.8379238927504832, "grad_norm": 4.405274391174316, "learning_rate": 1.3468641915347613e-06, "loss": 0.09578357934951783, "step": 31810 }, { "epoch": 0.838187307994982, "grad_norm": 5.091944694519043, "learning_rate": 1.3425911536104573e-06, "loss": 0.09241069555282592, "step": 31820 }, { "epoch": 0.8384507232394807, "grad_norm": 2.596323251724243, "learning_rate": 1.3383244169218268e-06, "loss": 0.08631106615066528, "step": 31830 }, { "epoch": 0.8387141384839795, "grad_norm": 3.8578381538391113, "learning_rate": 1.3340639845743797e-06, "loss": 0.1068005919456482, "step": 31840 }, { "epoch": 0.8389775537284782, "grad_norm": 3.4320621490478516, "learning_rate": 1.32980985966905e-06, "loss": 0.10400252342224121, "step": 31850 }, { "epoch": 0.8392409689729768, "grad_norm": 3.18393874168396, "learning_rate": 1.3255620453021656e-06, "loss": 0.10617237091064453, "step": 31860 }, { "epoch": 0.8395043842174756, "grad_norm": 4.313125133514404, "learning_rate": 1.3213205445654688e-06, "loss": 0.10099806785583496, "step": 31870 }, { "epoch": 0.8397677994619743, "grad_norm": 2.4946305751800537, "learning_rate": 1.3170853605461132e-06, "loss": 0.09285851120948792, "step": 31880 }, { "epoch": 0.8400312147064731, "grad_norm": 3.3654661178588867, "learning_rate": 1.3128564963266444e-06, "loss": 0.09898821711540222, "step": 31890 }, { "epoch": 0.8402946299509718, "grad_norm": 3.7701895236968994, "learning_rate": 1.3086339549850125e-06, "loss": 0.10347284078598022, "step": 31900 }, { "epoch": 0.8405580451954706, "grad_norm": 2.4257328510284424, "learning_rate": 1.3044177395945657e-06, "loss": 0.08472048044204712, "step": 31910 }, { "epoch": 0.8408214604399693, "grad_norm": 1.5888557434082031, "learning_rate": 1.3002078532240447e-06, "loss": 0.09408417940139771, "step": 31920 }, { "epoch": 0.8410848756844681, "grad_norm": 2.9525957107543945, "learning_rate": 1.296004298937592e-06, "loss": 0.09435389637947082, "step": 31930 }, { "epoch": 0.8413482909289668, "grad_norm": 2.5203909873962402, "learning_rate": 1.2918070797947346e-06, "loss": 0.09054456949234009, "step": 31940 }, { "epoch": 0.8416117061734655, "grad_norm": 2.5018014907836914, "learning_rate": 1.2876161988503843e-06, "loss": 0.10937116146087647, "step": 31950 }, { "epoch": 0.8418751214179643, "grad_norm": 3.387070894241333, "learning_rate": 1.2834316591548502e-06, "loss": 0.10193407535552979, "step": 31960 }, { "epoch": 0.842138536662463, "grad_norm": 3.277313709259033, "learning_rate": 1.279253463753819e-06, "loss": 0.10718719959259033, "step": 31970 }, { "epoch": 0.8424019519069618, "grad_norm": 2.7867484092712402, "learning_rate": 1.2750816156883627e-06, "loss": 0.09150842428207398, "step": 31980 }, { "epoch": 0.8426653671514605, "grad_norm": 2.9819233417510986, "learning_rate": 1.2709161179949303e-06, "loss": 0.08674167990684509, "step": 31990 }, { "epoch": 0.8429287823959593, "grad_norm": 3.3282487392425537, "learning_rate": 1.266756973705351e-06, "loss": 0.10014175176620484, "step": 32000 }, { "epoch": 0.8431921976404579, "grad_norm": 2.4463448524475098, "learning_rate": 1.2626041858468318e-06, "loss": 0.10225516557693481, "step": 32010 }, { "epoch": 0.8434556128849566, "grad_norm": 3.178802251815796, "learning_rate": 1.2584577574419499e-06, "loss": 0.09081330299377441, "step": 32020 }, { "epoch": 0.8437190281294554, "grad_norm": 2.572329044342041, "learning_rate": 1.2543176915086542e-06, "loss": 0.10528621673583985, "step": 32030 }, { "epoch": 0.8439824433739541, "grad_norm": 3.9513964653015137, "learning_rate": 1.2501839910602632e-06, "loss": 0.09379813671112061, "step": 32040 }, { "epoch": 0.8442458586184529, "grad_norm": 4.616683483123779, "learning_rate": 1.24605665910546e-06, "loss": 0.11002428531646728, "step": 32050 }, { "epoch": 0.8445092738629516, "grad_norm": 2.9063212871551514, "learning_rate": 1.2419356986483e-06, "loss": 0.0923873782157898, "step": 32060 }, { "epoch": 0.8447726891074504, "grad_norm": 2.842634916305542, "learning_rate": 1.2378211126881956e-06, "loss": 0.0952265739440918, "step": 32070 }, { "epoch": 0.8450361043519491, "grad_norm": 3.021512269973755, "learning_rate": 1.2337129042199147e-06, "loss": 0.08740826845169067, "step": 32080 }, { "epoch": 0.8452995195964479, "grad_norm": 3.163238048553467, "learning_rate": 1.2296110762335933e-06, "loss": 0.08300177454948425, "step": 32090 }, { "epoch": 0.8455629348409466, "grad_norm": 3.1313252449035645, "learning_rate": 1.2255156317147165e-06, "loss": 0.08923964500427246, "step": 32100 }, { "epoch": 0.8458263500854453, "grad_norm": 2.2423455715179443, "learning_rate": 1.2214265736441277e-06, "loss": 0.09286532402038575, "step": 32110 }, { "epoch": 0.8460897653299441, "grad_norm": 3.242619514465332, "learning_rate": 1.2173439049980174e-06, "loss": 0.08890159130096435, "step": 32120 }, { "epoch": 0.8463531805744428, "grad_norm": 4.498661041259766, "learning_rate": 1.2132676287479273e-06, "loss": 0.09349433183670045, "step": 32130 }, { "epoch": 0.8466165958189416, "grad_norm": 3.2329766750335693, "learning_rate": 1.209197747860752e-06, "loss": 0.09052103757858276, "step": 32140 }, { "epoch": 0.8468800110634402, "grad_norm": 2.8950369358062744, "learning_rate": 1.2051342652987218e-06, "loss": 0.09524500370025635, "step": 32150 }, { "epoch": 0.847143426307939, "grad_norm": 3.507214307785034, "learning_rate": 1.2010771840194168e-06, "loss": 0.08927430510520935, "step": 32160 }, { "epoch": 0.8474068415524377, "grad_norm": 2.7009949684143066, "learning_rate": 1.1970265069757548e-06, "loss": 0.10133577585220337, "step": 32170 }, { "epoch": 0.8476702567969365, "grad_norm": 2.8484127521514893, "learning_rate": 1.1929822371159915e-06, "loss": 0.09776394367218018, "step": 32180 }, { "epoch": 0.8479336720414352, "grad_norm": 2.57047176361084, "learning_rate": 1.188944377383725e-06, "loss": 0.09421412944793701, "step": 32190 }, { "epoch": 0.8481970872859339, "grad_norm": 2.2561919689178467, "learning_rate": 1.1849129307178841e-06, "loss": 0.08713370561599731, "step": 32200 }, { "epoch": 0.8484605025304327, "grad_norm": 2.3316705226898193, "learning_rate": 1.180887900052724e-06, "loss": 0.09209222793579101, "step": 32210 }, { "epoch": 0.8487239177749314, "grad_norm": 3.7147316932678223, "learning_rate": 1.17686928831784e-06, "loss": 0.10052947998046875, "step": 32220 }, { "epoch": 0.8489873330194302, "grad_norm": 2.9175825119018555, "learning_rate": 1.1728570984381505e-06, "loss": 0.10805363655090332, "step": 32230 }, { "epoch": 0.8492507482639289, "grad_norm": 2.873422861099243, "learning_rate": 1.1688513333339003e-06, "loss": 0.09449650645256043, "step": 32240 }, { "epoch": 0.8495141635084277, "grad_norm": 3.165595054626465, "learning_rate": 1.1648519959206584e-06, "loss": 0.0875783920288086, "step": 32250 }, { "epoch": 0.8497775787529264, "grad_norm": 4.455848693847656, "learning_rate": 1.1608590891093118e-06, "loss": 0.08554964065551758, "step": 32260 }, { "epoch": 0.8500409939974252, "grad_norm": 3.177588701248169, "learning_rate": 1.156872615806075e-06, "loss": 0.09029372930526733, "step": 32270 }, { "epoch": 0.8503044092419239, "grad_norm": 2.6499269008636475, "learning_rate": 1.1528925789124734e-06, "loss": 0.09653273224830627, "step": 32280 }, { "epoch": 0.8505678244864225, "grad_norm": 2.5339725017547607, "learning_rate": 1.1489189813253499e-06, "loss": 0.09877057075500488, "step": 32290 }, { "epoch": 0.8508312397309213, "grad_norm": 3.966057300567627, "learning_rate": 1.1449518259368596e-06, "loss": 0.09148743152618408, "step": 32300 }, { "epoch": 0.85109465497542, "grad_norm": 3.1455187797546387, "learning_rate": 1.1409911156344677e-06, "loss": 0.09859116077423095, "step": 32310 }, { "epoch": 0.8513580702199188, "grad_norm": 3.3148820400238037, "learning_rate": 1.1370368533009557e-06, "loss": 0.0919530749320984, "step": 32320 }, { "epoch": 0.8516214854644175, "grad_norm": 2.720465898513794, "learning_rate": 1.1330890418144014e-06, "loss": 0.09733401536941529, "step": 32330 }, { "epoch": 0.8518849007089163, "grad_norm": 3.9238269329071045, "learning_rate": 1.1291476840481908e-06, "loss": 0.09466272592544556, "step": 32340 }, { "epoch": 0.852148315953415, "grad_norm": 4.217270374298096, "learning_rate": 1.125212782871019e-06, "loss": 0.09438129663467407, "step": 32350 }, { "epoch": 0.8524117311979138, "grad_norm": 4.078958034515381, "learning_rate": 1.1212843411468755e-06, "loss": 0.1011093020439148, "step": 32360 }, { "epoch": 0.8526751464424125, "grad_norm": 3.422332286834717, "learning_rate": 1.1173623617350472e-06, "loss": 0.09399918913841247, "step": 32370 }, { "epoch": 0.8529385616869112, "grad_norm": 2.5954833030700684, "learning_rate": 1.1134468474901227e-06, "loss": 0.09863237738609314, "step": 32380 }, { "epoch": 0.85320197693141, "grad_norm": 2.9985086917877197, "learning_rate": 1.1095378012619796e-06, "loss": 0.09249123334884643, "step": 32390 }, { "epoch": 0.8534653921759087, "grad_norm": 2.0514776706695557, "learning_rate": 1.1056352258957924e-06, "loss": 0.08576788902282714, "step": 32400 }, { "epoch": 0.8537288074204075, "grad_norm": 4.1969194412231445, "learning_rate": 1.1017391242320263e-06, "loss": 0.09204506278038024, "step": 32410 }, { "epoch": 0.8539922226649062, "grad_norm": 3.6815025806427, "learning_rate": 1.0978494991064248e-06, "loss": 0.10107440948486328, "step": 32420 }, { "epoch": 0.854255637909405, "grad_norm": 2.233654022216797, "learning_rate": 1.0939663533500311e-06, "loss": 0.08743367195129395, "step": 32430 }, { "epoch": 0.8545190531539036, "grad_norm": 2.1622865200042725, "learning_rate": 1.0900896897891634e-06, "loss": 0.10060634613037109, "step": 32440 }, { "epoch": 0.8547824683984024, "grad_norm": 4.1610236167907715, "learning_rate": 1.0862195112454289e-06, "loss": 0.0978428840637207, "step": 32450 }, { "epoch": 0.8550458836429011, "grad_norm": 3.1400983333587646, "learning_rate": 1.0823558205357054e-06, "loss": 0.082077956199646, "step": 32460 }, { "epoch": 0.8553092988873998, "grad_norm": 3.455265760421753, "learning_rate": 1.0784986204721537e-06, "loss": 0.09723509550094604, "step": 32470 }, { "epoch": 0.8555727141318986, "grad_norm": 4.324492931365967, "learning_rate": 1.0746479138622134e-06, "loss": 0.09435479044914245, "step": 32480 }, { "epoch": 0.8558361293763973, "grad_norm": 4.279577255249023, "learning_rate": 1.070803703508595e-06, "loss": 0.10635780096054077, "step": 32490 }, { "epoch": 0.8560995446208961, "grad_norm": 2.893826961517334, "learning_rate": 1.066965992209279e-06, "loss": 0.08272877931594849, "step": 32500 }, { "epoch": 0.8563629598653948, "grad_norm": 3.086109161376953, "learning_rate": 1.0631347827575179e-06, "loss": 0.09827871918678284, "step": 32510 }, { "epoch": 0.8566263751098936, "grad_norm": 4.013503551483154, "learning_rate": 1.0593100779418286e-06, "loss": 0.09373946785926819, "step": 32520 }, { "epoch": 0.8568897903543923, "grad_norm": 3.234527349472046, "learning_rate": 1.0554918805460013e-06, "loss": 0.10563516616821289, "step": 32530 }, { "epoch": 0.8571532055988911, "grad_norm": 4.107700824737549, "learning_rate": 1.051680193349084e-06, "loss": 0.09215670824050903, "step": 32540 }, { "epoch": 0.8574166208433898, "grad_norm": 3.0857138633728027, "learning_rate": 1.0478750191253817e-06, "loss": 0.0980964481830597, "step": 32550 }, { "epoch": 0.8576800360878885, "grad_norm": 3.5534565448760986, "learning_rate": 1.0440763606444694e-06, "loss": 0.09108269810676575, "step": 32560 }, { "epoch": 0.8579434513323873, "grad_norm": 2.6182332038879395, "learning_rate": 1.0402842206711728e-06, "loss": 0.10657415390014649, "step": 32570 }, { "epoch": 0.8582068665768859, "grad_norm": 5.087111949920654, "learning_rate": 1.0364986019655765e-06, "loss": 0.09467771053314208, "step": 32580 }, { "epoch": 0.8584702818213847, "grad_norm": 3.1608715057373047, "learning_rate": 1.0327195072830153e-06, "loss": 0.09563338756561279, "step": 32590 }, { "epoch": 0.8587336970658834, "grad_norm": 3.0786561965942383, "learning_rate": 1.028946939374076e-06, "loss": 0.09977784752845764, "step": 32600 }, { "epoch": 0.8589971123103822, "grad_norm": 3.2337167263031006, "learning_rate": 1.0251809009846003e-06, "loss": 0.09600920677185058, "step": 32610 }, { "epoch": 0.8592605275548809, "grad_norm": 2.1908841133117676, "learning_rate": 1.0214213948556718e-06, "loss": 0.0786016345024109, "step": 32620 }, { "epoch": 0.8595239427993796, "grad_norm": 3.7261123657226562, "learning_rate": 1.0176684237236212e-06, "loss": 0.10227055549621582, "step": 32630 }, { "epoch": 0.8597873580438784, "grad_norm": 3.8062760829925537, "learning_rate": 1.0139219903200227e-06, "loss": 0.09454357624053955, "step": 32640 }, { "epoch": 0.8600507732883771, "grad_norm": 2.787571430206299, "learning_rate": 1.0101820973716891e-06, "loss": 0.09949631690979004, "step": 32650 }, { "epoch": 0.8603141885328759, "grad_norm": 3.046992778778076, "learning_rate": 1.0064487476006812e-06, "loss": 0.09153032302856445, "step": 32660 }, { "epoch": 0.8605776037773746, "grad_norm": 2.6377618312835693, "learning_rate": 1.0027219437242907e-06, "loss": 0.09557466506958008, "step": 32670 }, { "epoch": 0.8608410190218734, "grad_norm": 2.6089863777160645, "learning_rate": 9.990016884550425e-07, "loss": 0.09402128458023071, "step": 32680 }, { "epoch": 0.8611044342663721, "grad_norm": 3.0407540798187256, "learning_rate": 9.952879845007024e-07, "loss": 0.09025635719299316, "step": 32690 }, { "epoch": 0.8613678495108709, "grad_norm": 3.703721284866333, "learning_rate": 9.915808345642642e-07, "loss": 0.08563517332077027, "step": 32700 }, { "epoch": 0.8616312647553696, "grad_norm": 3.0487380027770996, "learning_rate": 9.878802413439514e-07, "loss": 0.09078384637832641, "step": 32710 }, { "epoch": 0.8618946799998682, "grad_norm": 3.628058433532715, "learning_rate": 9.841862075332143e-07, "loss": 0.0942264199256897, "step": 32720 }, { "epoch": 0.862158095244367, "grad_norm": 2.6281065940856934, "learning_rate": 9.804987358207274e-07, "loss": 0.0852974534034729, "step": 32730 }, { "epoch": 0.8624215104888657, "grad_norm": 3.0884718894958496, "learning_rate": 9.768178288903985e-07, "loss": 0.09714342355728149, "step": 32740 }, { "epoch": 0.8626849257333645, "grad_norm": 2.8360843658447266, "learning_rate": 9.731434894213465e-07, "loss": 0.09784061908721924, "step": 32750 }, { "epoch": 0.8629483409778632, "grad_norm": 2.959538221359253, "learning_rate": 9.694757200879146e-07, "loss": 0.098641037940979, "step": 32760 }, { "epoch": 0.863211756222362, "grad_norm": 2.938962936401367, "learning_rate": 9.658145235596628e-07, "loss": 0.1024125337600708, "step": 32770 }, { "epoch": 0.8634751714668607, "grad_norm": 4.127781867980957, "learning_rate": 9.621599025013672e-07, "loss": 0.09738211631774903, "step": 32780 }, { "epoch": 0.8637385867113595, "grad_norm": 2.808476448059082, "learning_rate": 9.585118595730226e-07, "loss": 0.09797812700271606, "step": 32790 }, { "epoch": 0.8640020019558582, "grad_norm": 3.0298516750335693, "learning_rate": 9.548703974298311e-07, "loss": 0.09240719079971313, "step": 32800 }, { "epoch": 0.8642654172003569, "grad_norm": 3.8873000144958496, "learning_rate": 9.512355187222e-07, "loss": 0.09259140491485596, "step": 32810 }, { "epoch": 0.8645288324448557, "grad_norm": 2.51027512550354, "learning_rate": 9.47607226095758e-07, "loss": 0.09605299234390259, "step": 32820 }, { "epoch": 0.8647922476893544, "grad_norm": 4.4521918296813965, "learning_rate": 9.439855221913297e-07, "loss": 0.09567086100578308, "step": 32830 }, { "epoch": 0.8650556629338532, "grad_norm": 3.49064302444458, "learning_rate": 9.403704096449484e-07, "loss": 0.0864011824131012, "step": 32840 }, { "epoch": 0.8653190781783519, "grad_norm": 3.999778985977173, "learning_rate": 9.367618910878484e-07, "loss": 0.09548619985580445, "step": 32850 }, { "epoch": 0.8655824934228507, "grad_norm": 3.4259207248687744, "learning_rate": 9.331599691464644e-07, "loss": 0.09955734014511108, "step": 32860 }, { "epoch": 0.8658459086673493, "grad_norm": 2.5781471729278564, "learning_rate": 9.295646464424323e-07, "loss": 0.0884982705116272, "step": 32870 }, { "epoch": 0.8661093239118481, "grad_norm": 4.981335163116455, "learning_rate": 9.259759255925827e-07, "loss": 0.08344166874885559, "step": 32880 }, { "epoch": 0.8663727391563468, "grad_norm": 3.513228416442871, "learning_rate": 9.223938092089413e-07, "loss": 0.09701043367385864, "step": 32890 }, { "epoch": 0.8666361544008455, "grad_norm": 3.0919597148895264, "learning_rate": 9.188182998987261e-07, "loss": 0.09666211605072021, "step": 32900 }, { "epoch": 0.8668995696453443, "grad_norm": 2.6260340213775635, "learning_rate": 9.152494002643464e-07, "loss": 0.09035321474075317, "step": 32910 }, { "epoch": 0.867162984889843, "grad_norm": 3.771613121032715, "learning_rate": 9.116871129034066e-07, "loss": 0.10014878511428833, "step": 32920 }, { "epoch": 0.8674264001343418, "grad_norm": 2.75187349319458, "learning_rate": 9.081314404086872e-07, "loss": 0.09253677129745483, "step": 32930 }, { "epoch": 0.8676898153788405, "grad_norm": 3.095564365386963, "learning_rate": 9.045823853681602e-07, "loss": 0.10060853958129883, "step": 32940 }, { "epoch": 0.8679532306233393, "grad_norm": 2.6872305870056152, "learning_rate": 9.010399503649836e-07, "loss": 0.0914117455482483, "step": 32950 }, { "epoch": 0.868216645867838, "grad_norm": 4.001628398895264, "learning_rate": 8.975041379774951e-07, "loss": 0.10837643146514893, "step": 32960 }, { "epoch": 0.8684800611123368, "grad_norm": 3.5230278968811035, "learning_rate": 8.939749507792094e-07, "loss": 0.10339630842208862, "step": 32970 }, { "epoch": 0.8687434763568355, "grad_norm": 2.888472557067871, "learning_rate": 8.904523913388218e-07, "loss": 0.08353308439254761, "step": 32980 }, { "epoch": 0.8690068916013342, "grad_norm": 2.8995018005371094, "learning_rate": 8.869364622202015e-07, "loss": 0.08582178354263306, "step": 32990 }, { "epoch": 0.869270306845833, "grad_norm": 2.832827091217041, "learning_rate": 8.834271659823967e-07, "loss": 0.09524689912796021, "step": 33000 }, { "epoch": 0.8695337220903316, "grad_norm": 3.379378318786621, "learning_rate": 8.799245051796246e-07, "loss": 0.08863604068756104, "step": 33010 }, { "epoch": 0.8697971373348304, "grad_norm": 2.7509024143218994, "learning_rate": 8.764284823612667e-07, "loss": 0.08813015222549439, "step": 33020 }, { "epoch": 0.8700605525793291, "grad_norm": 2.895264148712158, "learning_rate": 8.729391000718867e-07, "loss": 0.09697383642196655, "step": 33030 }, { "epoch": 0.8703239678238279, "grad_norm": 3.3012540340423584, "learning_rate": 8.694563608512041e-07, "loss": 0.09689740538597107, "step": 33040 }, { "epoch": 0.8705873830683266, "grad_norm": 3.3242783546447754, "learning_rate": 8.659802672341078e-07, "loss": 0.08751915097236633, "step": 33050 }, { "epoch": 0.8708507983128254, "grad_norm": 3.116929054260254, "learning_rate": 8.625108217506472e-07, "loss": 0.09177837371826172, "step": 33060 }, { "epoch": 0.8711142135573241, "grad_norm": 3.7297630310058594, "learning_rate": 8.590480269260337e-07, "loss": 0.09425595998764039, "step": 33070 }, { "epoch": 0.8713776288018228, "grad_norm": 3.5259485244750977, "learning_rate": 8.555918852806421e-07, "loss": 0.08675494194030761, "step": 33080 }, { "epoch": 0.8716410440463216, "grad_norm": 3.0966272354125977, "learning_rate": 8.521423993300005e-07, "loss": 0.09188753962516785, "step": 33090 }, { "epoch": 0.8719044592908203, "grad_norm": 3.5983715057373047, "learning_rate": 8.486995715847923e-07, "loss": 0.09124575853347779, "step": 33100 }, { "epoch": 0.8721678745353191, "grad_norm": 3.201040744781494, "learning_rate": 8.452634045508557e-07, "loss": 0.08652768135070801, "step": 33110 }, { "epoch": 0.8724312897798178, "grad_norm": 2.875333070755005, "learning_rate": 8.41833900729182e-07, "loss": 0.09306132793426514, "step": 33120 }, { "epoch": 0.8726947050243166, "grad_norm": 2.769075632095337, "learning_rate": 8.384110626159126e-07, "loss": 0.0909961462020874, "step": 33130 }, { "epoch": 0.8729581202688153, "grad_norm": 2.867682695388794, "learning_rate": 8.349948927023388e-07, "loss": 0.0811956286430359, "step": 33140 }, { "epoch": 0.8732215355133139, "grad_norm": 3.3257524967193604, "learning_rate": 8.315853934748908e-07, "loss": 0.0955632209777832, "step": 33150 }, { "epoch": 0.8734849507578127, "grad_norm": 3.869414806365967, "learning_rate": 8.281825674151533e-07, "loss": 0.09750907421112061, "step": 33160 }, { "epoch": 0.8737483660023114, "grad_norm": 2.7762246131896973, "learning_rate": 8.247864169998499e-07, "loss": 0.08974570035934448, "step": 33170 }, { "epoch": 0.8740117812468102, "grad_norm": 2.7864508628845215, "learning_rate": 8.213969447008441e-07, "loss": 0.08823457360267639, "step": 33180 }, { "epoch": 0.8742751964913089, "grad_norm": 3.226013660430908, "learning_rate": 8.180141529851404e-07, "loss": 0.08906721472740173, "step": 33190 }, { "epoch": 0.8745386117358077, "grad_norm": 3.5093014240264893, "learning_rate": 8.146380443148794e-07, "loss": 0.08846228122711182, "step": 33200 }, { "epoch": 0.8748020269803064, "grad_norm": 4.082247257232666, "learning_rate": 8.11268621147342e-07, "loss": 0.0905129075050354, "step": 33210 }, { "epoch": 0.8750654422248052, "grad_norm": 3.4987869262695312, "learning_rate": 8.079058859349364e-07, "loss": 0.10280086994171142, "step": 33220 }, { "epoch": 0.8753288574693039, "grad_norm": 3.955538749694824, "learning_rate": 8.045498411252095e-07, "loss": 0.09603239297866821, "step": 33230 }, { "epoch": 0.8755922727138026, "grad_norm": 4.063048362731934, "learning_rate": 8.012004891608327e-07, "loss": 0.10375138521194457, "step": 33240 }, { "epoch": 0.8758556879583014, "grad_norm": 4.079432010650635, "learning_rate": 7.978578324796072e-07, "loss": 0.08909070491790771, "step": 33250 }, { "epoch": 0.8761191032028001, "grad_norm": 2.9911234378814697, "learning_rate": 7.945218735144689e-07, "loss": 0.08930408358573913, "step": 33260 }, { "epoch": 0.8763825184472989, "grad_norm": 2.752302408218384, "learning_rate": 7.91192614693469e-07, "loss": 0.09423456192016602, "step": 33270 }, { "epoch": 0.8766459336917976, "grad_norm": 3.215351104736328, "learning_rate": 7.878700584397836e-07, "loss": 0.0886797308921814, "step": 33280 }, { "epoch": 0.8769093489362964, "grad_norm": 3.9837372303009033, "learning_rate": 7.84554207171716e-07, "loss": 0.08106666803359985, "step": 33290 }, { "epoch": 0.877172764180795, "grad_norm": 3.1975951194763184, "learning_rate": 7.812450633026847e-07, "loss": 0.07914093732833863, "step": 33300 }, { "epoch": 0.8774361794252938, "grad_norm": 3.6851303577423096, "learning_rate": 7.779426292412273e-07, "loss": 0.0798010528087616, "step": 33310 }, { "epoch": 0.8776995946697925, "grad_norm": 3.849900960922241, "learning_rate": 7.746469073909979e-07, "loss": 0.08511782288551331, "step": 33320 }, { "epoch": 0.8779630099142912, "grad_norm": 3.5349032878875732, "learning_rate": 7.713579001507632e-07, "loss": 0.1057771921157837, "step": 33330 }, { "epoch": 0.87822642515879, "grad_norm": 2.5102293491363525, "learning_rate": 7.680756099144072e-07, "loss": 0.0876925230026245, "step": 33340 }, { "epoch": 0.8784898404032887, "grad_norm": 4.0107550621032715, "learning_rate": 7.648000390709221e-07, "loss": 0.10064256191253662, "step": 33350 }, { "epoch": 0.8787532556477875, "grad_norm": 3.2747421264648438, "learning_rate": 7.615311900044087e-07, "loss": 0.07689244151115418, "step": 33360 }, { "epoch": 0.8790166708922862, "grad_norm": 4.497520446777344, "learning_rate": 7.582690650940771e-07, "loss": 0.09203946590423584, "step": 33370 }, { "epoch": 0.879280086136785, "grad_norm": 2.439319610595703, "learning_rate": 7.550136667142393e-07, "loss": 0.08631829619407654, "step": 33380 }, { "epoch": 0.8795435013812837, "grad_norm": 2.8069677352905273, "learning_rate": 7.517649972343199e-07, "loss": 0.0863680362701416, "step": 33390 }, { "epoch": 0.8798069166257825, "grad_norm": 2.391458511352539, "learning_rate": 7.485230590188397e-07, "loss": 0.07831871509552002, "step": 33400 }, { "epoch": 0.8800703318702812, "grad_norm": 3.5386290550231934, "learning_rate": 7.452878544274167e-07, "loss": 0.09449558258056641, "step": 33410 }, { "epoch": 0.8803337471147799, "grad_norm": 3.6095945835113525, "learning_rate": 7.420593858147773e-07, "loss": 0.0947471022605896, "step": 33420 }, { "epoch": 0.8805971623592787, "grad_norm": 3.499666929244995, "learning_rate": 7.388376555307386e-07, "loss": 0.09914920330047608, "step": 33430 }, { "epoch": 0.8808605776037773, "grad_norm": 3.002483367919922, "learning_rate": 7.356226659202137e-07, "loss": 0.08989368677139283, "step": 33440 }, { "epoch": 0.8811239928482761, "grad_norm": 3.0576770305633545, "learning_rate": 7.324144193232119e-07, "loss": 0.08034428358078002, "step": 33450 }, { "epoch": 0.8813874080927748, "grad_norm": 2.9425535202026367, "learning_rate": 7.29212918074833e-07, "loss": 0.08524180650711059, "step": 33460 }, { "epoch": 0.8816508233372736, "grad_norm": 3.077925443649292, "learning_rate": 7.260181645052689e-07, "loss": 0.09505313634872437, "step": 33470 }, { "epoch": 0.8819142385817723, "grad_norm": 3.4472415447235107, "learning_rate": 7.228301609397981e-07, "loss": 0.09771208763122559, "step": 33480 }, { "epoch": 0.8821776538262711, "grad_norm": 4.154006004333496, "learning_rate": 7.196489096987869e-07, "loss": 0.10023117065429688, "step": 33490 }, { "epoch": 0.8824410690707698, "grad_norm": 3.2081096172332764, "learning_rate": 7.164744130976886e-07, "loss": 0.0747557520866394, "step": 33500 }, { "epoch": 0.8827044843152685, "grad_norm": 3.075795888900757, "learning_rate": 7.133066734470351e-07, "loss": 0.09545258283615113, "step": 33510 }, { "epoch": 0.8829678995597673, "grad_norm": 3.1279118061065674, "learning_rate": 7.101456930524509e-07, "loss": 0.09523372650146485, "step": 33520 }, { "epoch": 0.883231314804266, "grad_norm": 3.2942683696746826, "learning_rate": 7.069914742146267e-07, "loss": 0.09158120155334473, "step": 33530 }, { "epoch": 0.8834947300487648, "grad_norm": 3.5227246284484863, "learning_rate": 7.038440192293416e-07, "loss": 0.09411636590957642, "step": 33540 }, { "epoch": 0.8837581452932635, "grad_norm": 3.163032054901123, "learning_rate": 7.007033303874511e-07, "loss": 0.08949615955352783, "step": 33550 }, { "epoch": 0.8840215605377623, "grad_norm": 2.811896562576294, "learning_rate": 6.97569409974882e-07, "loss": 0.10277917385101318, "step": 33560 }, { "epoch": 0.884284975782261, "grad_norm": 3.380265712738037, "learning_rate": 6.944422602726375e-07, "loss": 0.08930705189704895, "step": 33570 }, { "epoch": 0.8845483910267598, "grad_norm": 3.101896286010742, "learning_rate": 6.913218835567925e-07, "loss": 0.09309741854667664, "step": 33580 }, { "epoch": 0.8848118062712584, "grad_norm": 4.897694110870361, "learning_rate": 6.882082820984892e-07, "loss": 0.08905119895935058, "step": 33590 }, { "epoch": 0.8850752215157571, "grad_norm": 2.3872389793395996, "learning_rate": 6.851014581639437e-07, "loss": 0.0817844569683075, "step": 33600 }, { "epoch": 0.8853386367602559, "grad_norm": 3.144144058227539, "learning_rate": 6.820014140144382e-07, "loss": 0.08224602937698364, "step": 33610 }, { "epoch": 0.8856020520047546, "grad_norm": 3.189739942550659, "learning_rate": 6.789081519063134e-07, "loss": 0.09406375288963317, "step": 33620 }, { "epoch": 0.8858654672492534, "grad_norm": 3.1826510429382324, "learning_rate": 6.758216740909829e-07, "loss": 0.09949304461479187, "step": 33630 }, { "epoch": 0.8861288824937521, "grad_norm": 3.6155025959014893, "learning_rate": 6.727419828149173e-07, "loss": 0.10079312324523926, "step": 33640 }, { "epoch": 0.8863922977382509, "grad_norm": 2.5783138275146484, "learning_rate": 6.696690803196482e-07, "loss": 0.08757216334342957, "step": 33650 }, { "epoch": 0.8866557129827496, "grad_norm": 3.399517297744751, "learning_rate": 6.666029688417674e-07, "loss": 0.09392325878143311, "step": 33660 }, { "epoch": 0.8869191282272483, "grad_norm": 2.812974452972412, "learning_rate": 6.635436506129223e-07, "loss": 0.08656464219093322, "step": 33670 }, { "epoch": 0.8871825434717471, "grad_norm": 3.129350423812866, "learning_rate": 6.60491127859818e-07, "loss": 0.09565852880477906, "step": 33680 }, { "epoch": 0.8874459587162458, "grad_norm": 3.1127490997314453, "learning_rate": 6.57445402804211e-07, "loss": 0.08293832540512085, "step": 33690 }, { "epoch": 0.8877093739607446, "grad_norm": 4.839345932006836, "learning_rate": 6.544064776629123e-07, "loss": 0.08597288131713868, "step": 33700 }, { "epoch": 0.8879727892052433, "grad_norm": 2.854949712753296, "learning_rate": 6.513743546477824e-07, "loss": 0.07376391291618348, "step": 33710 }, { "epoch": 0.888236204449742, "grad_norm": 5.016962051391602, "learning_rate": 6.483490359657296e-07, "loss": 0.09717488288879395, "step": 33720 }, { "epoch": 0.8884996196942407, "grad_norm": 2.0760087966918945, "learning_rate": 6.453305238187147e-07, "loss": 0.07725675106048584, "step": 33730 }, { "epoch": 0.8887630349387395, "grad_norm": 2.8179826736450195, "learning_rate": 6.423188204037412e-07, "loss": 0.08982771039009094, "step": 33740 }, { "epoch": 0.8890264501832382, "grad_norm": 3.292743682861328, "learning_rate": 6.393139279128513e-07, "loss": 0.08159380555152893, "step": 33750 }, { "epoch": 0.8892898654277369, "grad_norm": 3.3405418395996094, "learning_rate": 6.3631584853314e-07, "loss": 0.08689085841178894, "step": 33760 }, { "epoch": 0.8895532806722357, "grad_norm": 4.572497367858887, "learning_rate": 6.333245844467384e-07, "loss": 0.08931238651275634, "step": 33770 }, { "epoch": 0.8898166959167344, "grad_norm": 2.497572422027588, "learning_rate": 6.303401378308172e-07, "loss": 0.08291392922401428, "step": 33780 }, { "epoch": 0.8900801111612332, "grad_norm": 3.0304346084594727, "learning_rate": 6.273625108575843e-07, "loss": 0.09409054517745971, "step": 33790 }, { "epoch": 0.8903435264057319, "grad_norm": 2.738354444503784, "learning_rate": 6.243917056942849e-07, "loss": 0.0864205539226532, "step": 33800 }, { "epoch": 0.8906069416502307, "grad_norm": 3.1968111991882324, "learning_rate": 6.214277245032019e-07, "loss": 0.08354384303092957, "step": 33810 }, { "epoch": 0.8908703568947294, "grad_norm": 3.5035531520843506, "learning_rate": 6.184705694416471e-07, "loss": 0.08862295150756835, "step": 33820 }, { "epoch": 0.8911337721392282, "grad_norm": 4.2114434242248535, "learning_rate": 6.155202426619655e-07, "loss": 0.10057644844055176, "step": 33830 }, { "epoch": 0.8913971873837269, "grad_norm": 2.999209403991699, "learning_rate": 6.125767463115317e-07, "loss": 0.08645391464233398, "step": 33840 }, { "epoch": 0.8916606026282256, "grad_norm": 4.000265121459961, "learning_rate": 6.096400825327487e-07, "loss": 0.10068552494049073, "step": 33850 }, { "epoch": 0.8919240178727244, "grad_norm": 3.479809284210205, "learning_rate": 6.067102534630498e-07, "loss": 0.08643823862075806, "step": 33860 }, { "epoch": 0.892187433117223, "grad_norm": 2.5303268432617188, "learning_rate": 6.037872612348905e-07, "loss": 0.07913478016853333, "step": 33870 }, { "epoch": 0.8924508483617218, "grad_norm": 3.6835238933563232, "learning_rate": 6.008711079757467e-07, "loss": 0.0875701904296875, "step": 33880 }, { "epoch": 0.8927142636062205, "grad_norm": 3.2934534549713135, "learning_rate": 5.979617958081241e-07, "loss": 0.07560775876045227, "step": 33890 }, { "epoch": 0.8929776788507193, "grad_norm": 2.512282609939575, "learning_rate": 5.950593268495464e-07, "loss": 0.08238592147827148, "step": 33900 }, { "epoch": 0.893241094095218, "grad_norm": 2.919076919555664, "learning_rate": 5.921637032125538e-07, "loss": 0.08320280313491821, "step": 33910 }, { "epoch": 0.8935045093397168, "grad_norm": 3.6241965293884277, "learning_rate": 5.89274927004706e-07, "loss": 0.08860147595405579, "step": 33920 }, { "epoch": 0.8937679245842155, "grad_norm": 2.5138514041900635, "learning_rate": 5.863930003285779e-07, "loss": 0.07800444364547729, "step": 33930 }, { "epoch": 0.8940313398287142, "grad_norm": 3.2965521812438965, "learning_rate": 5.835179252817635e-07, "loss": 0.0939902663230896, "step": 33940 }, { "epoch": 0.894294755073213, "grad_norm": 3.019381284713745, "learning_rate": 5.806497039568648e-07, "loss": 0.07974421977996826, "step": 33950 }, { "epoch": 0.8945581703177117, "grad_norm": 3.5967371463775635, "learning_rate": 5.777883384414973e-07, "loss": 0.07851769924163818, "step": 33960 }, { "epoch": 0.8948215855622105, "grad_norm": 2.8174242973327637, "learning_rate": 5.749338308182851e-07, "loss": 0.07739078402519226, "step": 33970 }, { "epoch": 0.8950850008067092, "grad_norm": 4.201671123504639, "learning_rate": 5.720861831648617e-07, "loss": 0.09269450306892395, "step": 33980 }, { "epoch": 0.895348416051208, "grad_norm": 3.399172782897949, "learning_rate": 5.692453975538703e-07, "loss": 0.08438794612884522, "step": 33990 }, { "epoch": 0.8956118312957067, "grad_norm": 3.4623701572418213, "learning_rate": 5.664114760529571e-07, "loss": 0.08390151262283325, "step": 34000 }, { "epoch": 0.8958752465402054, "grad_norm": 2.8204994201660156, "learning_rate": 5.635844207247687e-07, "loss": 0.08664155006408691, "step": 34010 }, { "epoch": 0.8961386617847041, "grad_norm": 3.7158985137939453, "learning_rate": 5.607642336269615e-07, "loss": 0.08213915824890136, "step": 34020 }, { "epoch": 0.8964020770292028, "grad_norm": 3.3220572471618652, "learning_rate": 5.579509168121888e-07, "loss": 0.07756505012512208, "step": 34030 }, { "epoch": 0.8966654922737016, "grad_norm": 3.033911943435669, "learning_rate": 5.551444723281029e-07, "loss": 0.07960489392280579, "step": 34040 }, { "epoch": 0.8969289075182003, "grad_norm": 5.297225475311279, "learning_rate": 5.523449022173566e-07, "loss": 0.10617910623550415, "step": 34050 }, { "epoch": 0.8971923227626991, "grad_norm": 2.4764060974121094, "learning_rate": 5.495522085175942e-07, "loss": 0.0855095386505127, "step": 34060 }, { "epoch": 0.8974557380071978, "grad_norm": 3.4381840229034424, "learning_rate": 5.467663932614642e-07, "loss": 0.0814032256603241, "step": 34070 }, { "epoch": 0.8977191532516966, "grad_norm": 2.92694091796875, "learning_rate": 5.439874584766014e-07, "loss": 0.09316105246543885, "step": 34080 }, { "epoch": 0.8979825684961953, "grad_norm": 3.0084803104400635, "learning_rate": 5.412154061856301e-07, "loss": 0.07237008213996887, "step": 34090 }, { "epoch": 0.8982459837406941, "grad_norm": 2.898942470550537, "learning_rate": 5.38450238406174e-07, "loss": 0.10283119678497314, "step": 34100 }, { "epoch": 0.8985093989851928, "grad_norm": 3.432744264602661, "learning_rate": 5.356919571508401e-07, "loss": 0.10021836757659912, "step": 34110 }, { "epoch": 0.8987728142296915, "grad_norm": 4.689160346984863, "learning_rate": 5.329405644272268e-07, "loss": 0.09144507646560669, "step": 34120 }, { "epoch": 0.8990362294741903, "grad_norm": 3.6323859691619873, "learning_rate": 5.301960622379132e-07, "loss": 0.08151757717132568, "step": 34130 }, { "epoch": 0.899299644718689, "grad_norm": 3.8476145267486572, "learning_rate": 5.27458452580466e-07, "loss": 0.08699052333831787, "step": 34140 }, { "epoch": 0.8995630599631877, "grad_norm": 3.5179672241210938, "learning_rate": 5.24727737447438e-07, "loss": 0.0852517306804657, "step": 34150 }, { "epoch": 0.8998264752076864, "grad_norm": 3.3987836837768555, "learning_rate": 5.220039188263615e-07, "loss": 0.08896442651748657, "step": 34160 }, { "epoch": 0.9000898904521852, "grad_norm": 3.128739595413208, "learning_rate": 5.192869986997484e-07, "loss": 0.08000218868255615, "step": 34170 }, { "epoch": 0.9003533056966839, "grad_norm": 3.0112569332122803, "learning_rate": 5.165769790450903e-07, "loss": 0.08368462324142456, "step": 34180 }, { "epoch": 0.9006167209411826, "grad_norm": 3.0043015480041504, "learning_rate": 5.138738618348548e-07, "loss": 0.08599801659584046, "step": 34190 }, { "epoch": 0.9008801361856814, "grad_norm": 3.691706895828247, "learning_rate": 5.111776490364906e-07, "loss": 0.09790818691253662, "step": 34200 }, { "epoch": 0.9011435514301801, "grad_norm": 3.751107931137085, "learning_rate": 5.084883426124165e-07, "loss": 0.07874753475189208, "step": 34210 }, { "epoch": 0.9014069666746789, "grad_norm": 2.7518179416656494, "learning_rate": 5.058059445200237e-07, "loss": 0.08827481269836426, "step": 34220 }, { "epoch": 0.9016703819191776, "grad_norm": 2.8434407711029053, "learning_rate": 5.031304567116802e-07, "loss": 0.09264191389083862, "step": 34230 }, { "epoch": 0.9019337971636764, "grad_norm": 3.5623862743377686, "learning_rate": 5.004618811347195e-07, "loss": 0.09471728801727294, "step": 34240 }, { "epoch": 0.9021972124081751, "grad_norm": 3.4860894680023193, "learning_rate": 4.978002197314469e-07, "loss": 0.08607452511787414, "step": 34250 }, { "epoch": 0.9024606276526739, "grad_norm": 2.7543022632598877, "learning_rate": 4.951454744391338e-07, "loss": 0.08257204294204712, "step": 34260 }, { "epoch": 0.9027240428971726, "grad_norm": 3.591611623764038, "learning_rate": 4.924976471900178e-07, "loss": 0.08494234085083008, "step": 34270 }, { "epoch": 0.9029874581416713, "grad_norm": 4.471585750579834, "learning_rate": 4.89856739911303e-07, "loss": 0.07940179705619813, "step": 34280 }, { "epoch": 0.90325087338617, "grad_norm": 4.307602405548096, "learning_rate": 4.872227545251562e-07, "loss": 0.08766630291938782, "step": 34290 }, { "epoch": 0.9035142886306687, "grad_norm": 3.088010787963867, "learning_rate": 4.84595692948705e-07, "loss": 0.08915607929229737, "step": 34300 }, { "epoch": 0.9037777038751675, "grad_norm": 3.9818055629730225, "learning_rate": 4.819755570940377e-07, "loss": 0.0913796603679657, "step": 34310 }, { "epoch": 0.9040411191196662, "grad_norm": 3.2396581172943115, "learning_rate": 4.793623488682008e-07, "loss": 0.09144996404647827, "step": 34320 }, { "epoch": 0.904304534364165, "grad_norm": 4.352632999420166, "learning_rate": 4.7675607017320323e-07, "loss": 0.08261058926582336, "step": 34330 }, { "epoch": 0.9045679496086637, "grad_norm": 4.995295524597168, "learning_rate": 4.741567229060062e-07, "loss": 0.08318189382553101, "step": 34340 }, { "epoch": 0.9048313648531625, "grad_norm": 3.260899782180786, "learning_rate": 4.715643089585242e-07, "loss": 0.08924617767333984, "step": 34350 }, { "epoch": 0.9050947800976612, "grad_norm": 4.0044636726379395, "learning_rate": 4.6897883021763013e-07, "loss": 0.08364477753639221, "step": 34360 }, { "epoch": 0.9053581953421599, "grad_norm": 3.9589812755584717, "learning_rate": 4.664002885651475e-07, "loss": 0.08676178455352783, "step": 34370 }, { "epoch": 0.9056216105866587, "grad_norm": 4.77268648147583, "learning_rate": 4.638286858778496e-07, "loss": 0.09614666104316712, "step": 34380 }, { "epoch": 0.9058850258311574, "grad_norm": 3.9360530376434326, "learning_rate": 4.612640240274591e-07, "loss": 0.08495822548866272, "step": 34390 }, { "epoch": 0.9061484410756562, "grad_norm": 3.4628560543060303, "learning_rate": 4.587063048806462e-07, "loss": 0.09898831844329833, "step": 34400 }, { "epoch": 0.9064118563201549, "grad_norm": 2.305750846862793, "learning_rate": 4.561555302990328e-07, "loss": 0.09796891808509826, "step": 34410 }, { "epoch": 0.9066752715646537, "grad_norm": 2.8854150772094727, "learning_rate": 4.536117021391806e-07, "loss": 0.08691774606704712, "step": 34420 }, { "epoch": 0.9069386868091524, "grad_norm": 2.890357732772827, "learning_rate": 4.5107482225259735e-07, "loss": 0.08702151775360108, "step": 34430 }, { "epoch": 0.9072021020536511, "grad_norm": 4.15429162979126, "learning_rate": 4.4854489248573494e-07, "loss": 0.08603067994117737, "step": 34440 }, { "epoch": 0.9074655172981498, "grad_norm": 3.4129340648651123, "learning_rate": 4.460219146799838e-07, "loss": 0.08342140913009644, "step": 34450 }, { "epoch": 0.9077289325426485, "grad_norm": 3.361520290374756, "learning_rate": 4.4350589067167717e-07, "loss": 0.08995670080184937, "step": 34460 }, { "epoch": 0.9079923477871473, "grad_norm": 4.514564037322998, "learning_rate": 4.409968222920868e-07, "loss": 0.09654566645622253, "step": 34470 }, { "epoch": 0.908255763031646, "grad_norm": 4.548159122467041, "learning_rate": 4.384947113674165e-07, "loss": 0.09423476457595825, "step": 34480 }, { "epoch": 0.9085191782761448, "grad_norm": 3.0828442573547363, "learning_rate": 4.35999559718816e-07, "loss": 0.08631596565246583, "step": 34490 }, { "epoch": 0.9087825935206435, "grad_norm": 3.2974255084991455, "learning_rate": 4.3351136916236156e-07, "loss": 0.09381859302520752, "step": 34500 }, { "epoch": 0.9090460087651423, "grad_norm": 3.241365432739258, "learning_rate": 4.3103014150906675e-07, "loss": 0.07554358243942261, "step": 34510 }, { "epoch": 0.909309424009641, "grad_norm": 3.6293766498565674, "learning_rate": 4.285558785648747e-07, "loss": 0.09760642051696777, "step": 34520 }, { "epoch": 0.9095728392541398, "grad_norm": 3.114748239517212, "learning_rate": 4.260885821306604e-07, "loss": 0.07752300500869751, "step": 34530 }, { "epoch": 0.9098362544986385, "grad_norm": 2.567119836807251, "learning_rate": 4.236282540022307e-07, "loss": 0.08172827959060669, "step": 34540 }, { "epoch": 0.9100996697431372, "grad_norm": 2.4011149406433105, "learning_rate": 4.2117489597031767e-07, "loss": 0.087453293800354, "step": 34550 }, { "epoch": 0.910363084987636, "grad_norm": 5.243180751800537, "learning_rate": 4.1872850982058067e-07, "loss": 0.08985509872436523, "step": 34560 }, { "epoch": 0.9106265002321347, "grad_norm": 3.4648830890655518, "learning_rate": 4.1628909733360534e-07, "loss": 0.08979392051696777, "step": 34570 }, { "epoch": 0.9108899154766334, "grad_norm": 2.9039199352264404, "learning_rate": 4.1385666028490036e-07, "loss": 0.09447096586227417, "step": 34580 }, { "epoch": 0.9111533307211321, "grad_norm": 3.6918351650238037, "learning_rate": 4.114312004449006e-07, "loss": 0.07227696180343628, "step": 34590 }, { "epoch": 0.9114167459656309, "grad_norm": 3.606501817703247, "learning_rate": 4.090127195789606e-07, "loss": 0.08282716274261474, "step": 34600 }, { "epoch": 0.9116801612101296, "grad_norm": 3.1511096954345703, "learning_rate": 4.066012194473512e-07, "loss": 0.09553505182266235, "step": 34610 }, { "epoch": 0.9119435764546284, "grad_norm": 3.819887161254883, "learning_rate": 4.041967018052706e-07, "loss": 0.09346411228179932, "step": 34620 }, { "epoch": 0.9122069916991271, "grad_norm": 3.228548526763916, "learning_rate": 4.0179916840282776e-07, "loss": 0.08070815801620483, "step": 34630 }, { "epoch": 0.9124704069436258, "grad_norm": 2.955960273742676, "learning_rate": 3.9940862098505117e-07, "loss": 0.08405004739761353, "step": 34640 }, { "epoch": 0.9127338221881246, "grad_norm": 4.564055442810059, "learning_rate": 3.970250612918858e-07, "loss": 0.1055219292640686, "step": 34650 }, { "epoch": 0.9129972374326233, "grad_norm": 2.5890605449676514, "learning_rate": 3.946484910581871e-07, "loss": 0.07887378334999084, "step": 34660 }, { "epoch": 0.9132606526771221, "grad_norm": 2.747901439666748, "learning_rate": 3.9227891201372827e-07, "loss": 0.08548101186752319, "step": 34670 }, { "epoch": 0.9135240679216208, "grad_norm": 3.0815012454986572, "learning_rate": 3.899163258831895e-07, "loss": 0.09165558815002442, "step": 34680 }, { "epoch": 0.9137874831661196, "grad_norm": 3.0540664196014404, "learning_rate": 3.8756073438616203e-07, "loss": 0.08788999319076538, "step": 34690 }, { "epoch": 0.9140508984106183, "grad_norm": 3.0170834064483643, "learning_rate": 3.8521213923714775e-07, "loss": 0.09765267372131348, "step": 34700 }, { "epoch": 0.914314313655117, "grad_norm": 2.5139827728271484, "learning_rate": 3.8287054214555694e-07, "loss": 0.0784426212310791, "step": 34710 }, { "epoch": 0.9145777288996157, "grad_norm": 3.0750980377197266, "learning_rate": 3.805359448157031e-07, "loss": 0.07888532876968384, "step": 34720 }, { "epoch": 0.9148411441441144, "grad_norm": 3.341757297515869, "learning_rate": 3.78208348946808e-07, "loss": 0.07856959700584412, "step": 34730 }, { "epoch": 0.9151045593886132, "grad_norm": 2.3020057678222656, "learning_rate": 3.7588775623299433e-07, "loss": 0.07902618050575257, "step": 34740 }, { "epoch": 0.9153679746331119, "grad_norm": 2.390141725540161, "learning_rate": 3.735741683632932e-07, "loss": 0.08202335238456726, "step": 34750 }, { "epoch": 0.9156313898776107, "grad_norm": 2.8591463565826416, "learning_rate": 3.712675870216309e-07, "loss": 0.08802300691604614, "step": 34760 }, { "epoch": 0.9158948051221094, "grad_norm": 3.39674711227417, "learning_rate": 3.689680138868379e-07, "loss": 0.08118580579757691, "step": 34770 }, { "epoch": 0.9161582203666082, "grad_norm": 2.938375234603882, "learning_rate": 3.66675450632642e-07, "loss": 0.07707728147506714, "step": 34780 }, { "epoch": 0.9164216356111069, "grad_norm": 3.298567533493042, "learning_rate": 3.643898989276684e-07, "loss": 0.09594370126724243, "step": 34790 }, { "epoch": 0.9166850508556056, "grad_norm": 4.36620569229126, "learning_rate": 3.6211136043544425e-07, "loss": 0.08596215248107911, "step": 34800 }, { "epoch": 0.9169484661001044, "grad_norm": 4.295729160308838, "learning_rate": 3.598398368143863e-07, "loss": 0.08385931253433228, "step": 34810 }, { "epoch": 0.9172118813446031, "grad_norm": 2.5831117630004883, "learning_rate": 3.575753297178053e-07, "loss": 0.08350439071655273, "step": 34820 }, { "epoch": 0.9174752965891019, "grad_norm": 3.6707592010498047, "learning_rate": 3.553178407939095e-07, "loss": 0.0927233338356018, "step": 34830 }, { "epoch": 0.9177387118336006, "grad_norm": 2.7763943672180176, "learning_rate": 3.5306737168579685e-07, "loss": 0.08285140991210938, "step": 34840 }, { "epoch": 0.9180021270780994, "grad_norm": 3.225613594055176, "learning_rate": 3.5082392403145373e-07, "loss": 0.08340762853622437, "step": 34850 }, { "epoch": 0.918265542322598, "grad_norm": 3.0817551612854004, "learning_rate": 3.485874994637606e-07, "loss": 0.0823857069015503, "step": 34860 }, { "epoch": 0.9185289575670968, "grad_norm": 3.7787697315216064, "learning_rate": 3.463580996104821e-07, "loss": 0.08521089553833008, "step": 34870 }, { "epoch": 0.9187923728115955, "grad_norm": 3.5316710472106934, "learning_rate": 3.4413572609427235e-07, "loss": 0.09422429203987122, "step": 34880 }, { "epoch": 0.9190557880560942, "grad_norm": 3.902033805847168, "learning_rate": 3.4192038053267095e-07, "loss": 0.08932816982269287, "step": 34890 }, { "epoch": 0.919319203300593, "grad_norm": 2.909271717071533, "learning_rate": 3.3971206453810133e-07, "loss": 0.0775286853313446, "step": 34900 }, { "epoch": 0.9195826185450917, "grad_norm": 2.87044095993042, "learning_rate": 3.3751077971787115e-07, "loss": 0.08158560395240784, "step": 34910 }, { "epoch": 0.9198460337895905, "grad_norm": 3.715365409851074, "learning_rate": 3.353165276741699e-07, "loss": 0.08147878646850586, "step": 34920 }, { "epoch": 0.9201094490340892, "grad_norm": 3.764178991317749, "learning_rate": 3.331293100040711e-07, "loss": 0.08309420347213745, "step": 34930 }, { "epoch": 0.920372864278588, "grad_norm": 4.1235880851745605, "learning_rate": 3.3094912829952477e-07, "loss": 0.09059157967567444, "step": 34940 }, { "epoch": 0.9206362795230867, "grad_norm": 3.966954231262207, "learning_rate": 3.2877598414736033e-07, "loss": 0.07863707542419433, "step": 34950 }, { "epoch": 0.9208996947675855, "grad_norm": 4.014176845550537, "learning_rate": 3.266098791292871e-07, "loss": 0.09499541521072388, "step": 34960 }, { "epoch": 0.9211631100120842, "grad_norm": 3.239708423614502, "learning_rate": 3.244508148218895e-07, "loss": 0.07584614753723144, "step": 34970 }, { "epoch": 0.9214265252565829, "grad_norm": 4.047436237335205, "learning_rate": 3.22298792796627e-07, "loss": 0.09254616498947144, "step": 34980 }, { "epoch": 0.9216899405010817, "grad_norm": 3.019744634628296, "learning_rate": 3.2015381461983577e-07, "loss": 0.08433105945587158, "step": 34990 }, { "epoch": 0.9219533557455803, "grad_norm": 3.250565767288208, "learning_rate": 3.1801588185272016e-07, "loss": 0.08402584195137024, "step": 35000 }, { "epoch": 0.9222167709900791, "grad_norm": 3.3826653957366943, "learning_rate": 3.1588499605136435e-07, "loss": 0.08576569557189942, "step": 35010 }, { "epoch": 0.9224801862345778, "grad_norm": 3.3631913661956787, "learning_rate": 3.1376115876671553e-07, "loss": 0.0919402778148651, "step": 35020 }, { "epoch": 0.9227436014790766, "grad_norm": 2.6574199199676514, "learning_rate": 3.1164437154459494e-07, "loss": 0.07615161538124085, "step": 35030 }, { "epoch": 0.9230070167235753, "grad_norm": 5.680471420288086, "learning_rate": 3.0953463592569343e-07, "loss": 0.07751066088676453, "step": 35040 }, { "epoch": 0.9232704319680741, "grad_norm": 3.8190364837646484, "learning_rate": 3.074319534455639e-07, "loss": 0.09189126491546631, "step": 35050 }, { "epoch": 0.9235338472125728, "grad_norm": 3.7673075199127197, "learning_rate": 3.053363256346331e-07, "loss": 0.09196407794952392, "step": 35060 }, { "epoch": 0.9237972624570715, "grad_norm": 5.124591827392578, "learning_rate": 3.0324775401818884e-07, "loss": 0.07704642415046692, "step": 35070 }, { "epoch": 0.9240606777015703, "grad_norm": 3.4596781730651855, "learning_rate": 3.0116624011637953e-07, "loss": 0.0897938847541809, "step": 35080 }, { "epoch": 0.924324092946069, "grad_norm": 3.4642271995544434, "learning_rate": 2.9909178544422656e-07, "loss": 0.0808876633644104, "step": 35090 }, { "epoch": 0.9245875081905678, "grad_norm": 3.3686089515686035, "learning_rate": 2.9702439151160334e-07, "loss": 0.09032883644104003, "step": 35100 }, { "epoch": 0.9248509234350665, "grad_norm": 4.220890045166016, "learning_rate": 2.949640598232517e-07, "loss": 0.08163931369781494, "step": 35110 }, { "epoch": 0.9251143386795653, "grad_norm": 3.983877658843994, "learning_rate": 2.9291079187876656e-07, "loss": 0.09298641681671142, "step": 35120 }, { "epoch": 0.925377753924064, "grad_norm": 4.219701290130615, "learning_rate": 2.9086458917260585e-07, "loss": 0.09044439792633056, "step": 35130 }, { "epoch": 0.9256411691685628, "grad_norm": 3.9533848762512207, "learning_rate": 2.8882545319408706e-07, "loss": 0.07398993372917176, "step": 35140 }, { "epoch": 0.9259045844130614, "grad_norm": 3.076317071914673, "learning_rate": 2.8679338542737745e-07, "loss": 0.09360878467559815, "step": 35150 }, { "epoch": 0.9261679996575601, "grad_norm": 5.358429908752441, "learning_rate": 2.8476838735150613e-07, "loss": 0.0820543646812439, "step": 35160 }, { "epoch": 0.9264314149020589, "grad_norm": 3.138561248779297, "learning_rate": 2.8275046044035304e-07, "loss": 0.0786907434463501, "step": 35170 }, { "epoch": 0.9266948301465576, "grad_norm": 3.3934662342071533, "learning_rate": 2.8073960616265215e-07, "loss": 0.09883753061294556, "step": 35180 }, { "epoch": 0.9269582453910564, "grad_norm": 3.9386441707611084, "learning_rate": 2.7873582598199056e-07, "loss": 0.08781647682189941, "step": 35190 }, { "epoch": 0.9272216606355551, "grad_norm": 2.4442944526672363, "learning_rate": 2.7673912135680714e-07, "loss": 0.07351301908493042, "step": 35200 }, { "epoch": 0.9274850758800539, "grad_norm": 3.9154365062713623, "learning_rate": 2.7474949374038717e-07, "loss": 0.07554945349693298, "step": 35210 }, { "epoch": 0.9277484911245526, "grad_norm": 2.4718570709228516, "learning_rate": 2.7276694458086894e-07, "loss": 0.08849283456802368, "step": 35220 }, { "epoch": 0.9280119063690514, "grad_norm": 3.308765411376953, "learning_rate": 2.7079147532123704e-07, "loss": 0.0982850432395935, "step": 35230 }, { "epoch": 0.9282753216135501, "grad_norm": 2.6897995471954346, "learning_rate": 2.688230873993225e-07, "loss": 0.06926339864730835, "step": 35240 }, { "epoch": 0.9285387368580488, "grad_norm": 3.650346040725708, "learning_rate": 2.6686178224780367e-07, "loss": 0.08119605779647827, "step": 35250 }, { "epoch": 0.9288021521025476, "grad_norm": 2.9021408557891846, "learning_rate": 2.6490756129420313e-07, "loss": 0.08435753583908082, "step": 35260 }, { "epoch": 0.9290655673470463, "grad_norm": 3.413626194000244, "learning_rate": 2.6296042596088643e-07, "loss": 0.08593282103538513, "step": 35270 }, { "epoch": 0.9293289825915451, "grad_norm": 4.042736530303955, "learning_rate": 2.610203776650655e-07, "loss": 0.08786964416503906, "step": 35280 }, { "epoch": 0.9295923978360437, "grad_norm": 3.822274923324585, "learning_rate": 2.5908741781878744e-07, "loss": 0.07977925539016724, "step": 35290 }, { "epoch": 0.9298558130805425, "grad_norm": 2.934555768966675, "learning_rate": 2.5716154782894573e-07, "loss": 0.08152265548706054, "step": 35300 }, { "epoch": 0.9301192283250412, "grad_norm": 4.309278964996338, "learning_rate": 2.552427690972703e-07, "loss": 0.07851482629776001, "step": 35310 }, { "epoch": 0.9303826435695399, "grad_norm": 3.592188835144043, "learning_rate": 2.5333108302033394e-07, "loss": 0.08045831322669983, "step": 35320 }, { "epoch": 0.9306460588140387, "grad_norm": 5.080312728881836, "learning_rate": 2.514264909895414e-07, "loss": 0.09193810820579529, "step": 35330 }, { "epoch": 0.9309094740585374, "grad_norm": 3.603349447250366, "learning_rate": 2.495289943911361e-07, "loss": 0.09593908786773682, "step": 35340 }, { "epoch": 0.9311728893030362, "grad_norm": 3.4403791427612305, "learning_rate": 2.476385946062021e-07, "loss": 0.08729465007781982, "step": 35350 }, { "epoch": 0.9314363045475349, "grad_norm": 3.0933663845062256, "learning_rate": 2.4575529301065104e-07, "loss": 0.08391414880752564, "step": 35360 }, { "epoch": 0.9316997197920337, "grad_norm": 4.046285152435303, "learning_rate": 2.43879090975232e-07, "loss": 0.08303158283233643, "step": 35370 }, { "epoch": 0.9319631350365324, "grad_norm": 6.206757068634033, "learning_rate": 2.4200998986552484e-07, "loss": 0.09350431561470032, "step": 35380 }, { "epoch": 0.9322265502810312, "grad_norm": 4.042505264282227, "learning_rate": 2.401479910419413e-07, "loss": 0.08669503927230834, "step": 35390 }, { "epoch": 0.9324899655255299, "grad_norm": 3.3467252254486084, "learning_rate": 2.3829309585972738e-07, "loss": 0.09128780364990234, "step": 35400 }, { "epoch": 0.9327533807700286, "grad_norm": 4.036329746246338, "learning_rate": 2.364453056689553e-07, "loss": 0.08371238708496094, "step": 35410 }, { "epoch": 0.9330167960145274, "grad_norm": 3.215907096862793, "learning_rate": 2.3460462181452258e-07, "loss": 0.08321308493614196, "step": 35420 }, { "epoch": 0.933280211259026, "grad_norm": 2.52276873588562, "learning_rate": 2.3277104563616205e-07, "loss": 0.08608095049858093, "step": 35430 }, { "epoch": 0.9335436265035248, "grad_norm": 2.7298171520233154, "learning_rate": 2.3094457846842943e-07, "loss": 0.07686167359352111, "step": 35440 }, { "epoch": 0.9338070417480235, "grad_norm": 4.737026691436768, "learning_rate": 2.291252216407047e-07, "loss": 0.07531114816665649, "step": 35450 }, { "epoch": 0.9340704569925223, "grad_norm": 5.21198844909668, "learning_rate": 2.273129764771953e-07, "loss": 0.07870177626609802, "step": 35460 }, { "epoch": 0.934333872237021, "grad_norm": 3.552905559539795, "learning_rate": 2.2550784429693052e-07, "loss": 0.08875997066497802, "step": 35470 }, { "epoch": 0.9345972874815198, "grad_norm": 2.960439682006836, "learning_rate": 2.2370982641376494e-07, "loss": 0.07506343126296997, "step": 35480 }, { "epoch": 0.9348607027260185, "grad_norm": 4.901406764984131, "learning_rate": 2.2191892413637173e-07, "loss": 0.0812563955783844, "step": 35490 }, { "epoch": 0.9351241179705172, "grad_norm": 2.8541386127471924, "learning_rate": 2.2013513876824933e-07, "loss": 0.07577197551727295, "step": 35500 }, { "epoch": 0.935387533215016, "grad_norm": 3.193404197692871, "learning_rate": 2.183584716077114e-07, "loss": 0.08035542964935302, "step": 35510 }, { "epoch": 0.9356509484595147, "grad_norm": 2.3108410835266113, "learning_rate": 2.1658892394789244e-07, "loss": 0.07870726585388184, "step": 35520 }, { "epoch": 0.9359143637040135, "grad_norm": 3.871729612350464, "learning_rate": 2.1482649707674775e-07, "loss": 0.07570463418960571, "step": 35530 }, { "epoch": 0.9361777789485122, "grad_norm": 3.179340362548828, "learning_rate": 2.13071192277049e-07, "loss": 0.07787681818008423, "step": 35540 }, { "epoch": 0.936441194193011, "grad_norm": 3.042442560195923, "learning_rate": 2.1132301082637752e-07, "loss": 0.08204807043075561, "step": 35550 }, { "epoch": 0.9367046094375097, "grad_norm": 3.400836706161499, "learning_rate": 2.0958195399713888e-07, "loss": 0.08817117214202881, "step": 35560 }, { "epoch": 0.9369680246820085, "grad_norm": 3.3523664474487305, "learning_rate": 2.0784802305654938e-07, "loss": 0.07667633891105652, "step": 35570 }, { "epoch": 0.9372314399265071, "grad_norm": 4.7928690910339355, "learning_rate": 2.061212192666373e-07, "loss": 0.08588166236877441, "step": 35580 }, { "epoch": 0.9374948551710058, "grad_norm": 2.6011431217193604, "learning_rate": 2.0440154388424616e-07, "loss": 0.07943220734596253, "step": 35590 }, { "epoch": 0.9377582704155046, "grad_norm": 2.815976619720459, "learning_rate": 2.0268899816102916e-07, "loss": 0.07548009157180786, "step": 35600 }, { "epoch": 0.9380216856600033, "grad_norm": 3.3688182830810547, "learning_rate": 2.0098358334345147e-07, "loss": 0.08236407041549683, "step": 35610 }, { "epoch": 0.9382851009045021, "grad_norm": 3.7306954860687256, "learning_rate": 1.9928530067278684e-07, "loss": 0.08273921608924865, "step": 35620 }, { "epoch": 0.9385485161490008, "grad_norm": 3.8284285068511963, "learning_rate": 1.9759415138511984e-07, "loss": 0.07994861602783203, "step": 35630 }, { "epoch": 0.9388119313934996, "grad_norm": 3.2158749103546143, "learning_rate": 1.959101367113414e-07, "loss": 0.0804501473903656, "step": 35640 }, { "epoch": 0.9390753466379983, "grad_norm": 3.659038543701172, "learning_rate": 1.9423325787714774e-07, "loss": 0.09184874296188354, "step": 35650 }, { "epoch": 0.9393387618824971, "grad_norm": 3.585344076156616, "learning_rate": 1.9256351610304814e-07, "loss": 0.08698840141296386, "step": 35660 }, { "epoch": 0.9396021771269958, "grad_norm": 2.744462251663208, "learning_rate": 1.909009126043504e-07, "loss": 0.07162058353424072, "step": 35670 }, { "epoch": 0.9398655923714945, "grad_norm": 2.822319269180298, "learning_rate": 1.8924544859116656e-07, "loss": 0.09116084575653076, "step": 35680 }, { "epoch": 0.9401290076159933, "grad_norm": 4.051775932312012, "learning_rate": 1.8759712526841943e-07, "loss": 0.08856608867645263, "step": 35690 }, { "epoch": 0.940392422860492, "grad_norm": 3.9210100173950195, "learning_rate": 1.8595594383582605e-07, "loss": 0.08982034921646118, "step": 35700 }, { "epoch": 0.9406558381049908, "grad_norm": 3.2568719387054443, "learning_rate": 1.8432190548791197e-07, "loss": 0.08506582975387574, "step": 35710 }, { "epoch": 0.9409192533494894, "grad_norm": 2.460819959640503, "learning_rate": 1.8269501141399915e-07, "loss": 0.07042222023010254, "step": 35720 }, { "epoch": 0.9411826685939882, "grad_norm": 3.8477725982666016, "learning_rate": 1.8107526279821041e-07, "loss": 0.08178185820579528, "step": 35730 }, { "epoch": 0.9414460838384869, "grad_norm": 2.711212158203125, "learning_rate": 1.794626608194705e-07, "loss": 0.08770062923431396, "step": 35740 }, { "epoch": 0.9417094990829857, "grad_norm": 2.7258124351501465, "learning_rate": 1.7785720665150054e-07, "loss": 0.0872636616230011, "step": 35750 }, { "epoch": 0.9419729143274844, "grad_norm": 4.738504409790039, "learning_rate": 1.7625890146281798e-07, "loss": 0.08558440804481507, "step": 35760 }, { "epoch": 0.9422363295719831, "grad_norm": 2.6235580444335938, "learning_rate": 1.7466774641674012e-07, "loss": 0.06591365337371827, "step": 35770 }, { "epoch": 0.9424997448164819, "grad_norm": 3.913238048553467, "learning_rate": 1.7308374267137606e-07, "loss": 0.0827100157737732, "step": 35780 }, { "epoch": 0.9427631600609806, "grad_norm": 3.306812047958374, "learning_rate": 1.715068913796336e-07, "loss": 0.09218102693557739, "step": 35790 }, { "epoch": 0.9430265753054794, "grad_norm": 3.551117181777954, "learning_rate": 1.6993719368921468e-07, "loss": 0.08376181125640869, "step": 35800 }, { "epoch": 0.9432899905499781, "grad_norm": 4.245507717132568, "learning_rate": 1.683746507426087e-07, "loss": 0.07174279093742371, "step": 35810 }, { "epoch": 0.9435534057944769, "grad_norm": 3.8738937377929688, "learning_rate": 1.66819263677106e-07, "loss": 0.08998479843139648, "step": 35820 }, { "epoch": 0.9438168210389756, "grad_norm": 3.872239112854004, "learning_rate": 1.6527103362478202e-07, "loss": 0.08983467817306519, "step": 35830 }, { "epoch": 0.9440802362834743, "grad_norm": 3.113593101501465, "learning_rate": 1.6372996171250544e-07, "loss": 0.08798668384552003, "step": 35840 }, { "epoch": 0.9443436515279731, "grad_norm": 3.4319710731506348, "learning_rate": 1.6219604906193677e-07, "loss": 0.07706575393676758, "step": 35850 }, { "epoch": 0.9446070667724717, "grad_norm": 3.052165985107422, "learning_rate": 1.606692967895207e-07, "loss": 0.0879701852798462, "step": 35860 }, { "epoch": 0.9448704820169705, "grad_norm": 3.8080427646636963, "learning_rate": 1.5914970600649615e-07, "loss": 0.09553242921829223, "step": 35870 }, { "epoch": 0.9451338972614692, "grad_norm": 3.2420108318328857, "learning_rate": 1.5763727781888726e-07, "loss": 0.06664088964462281, "step": 35880 }, { "epoch": 0.945397312505968, "grad_norm": 2.890103578567505, "learning_rate": 1.5613201332750127e-07, "loss": 0.07854794263839722, "step": 35890 }, { "epoch": 0.9456607277504667, "grad_norm": 4.011524677276611, "learning_rate": 1.5463391362793735e-07, "loss": 0.07835544347763061, "step": 35900 }, { "epoch": 0.9459241429949655, "grad_norm": 2.962677001953125, "learning_rate": 1.5314297981057658e-07, "loss": 0.08217702507972717, "step": 35910 }, { "epoch": 0.9461875582394642, "grad_norm": 2.326568126678467, "learning_rate": 1.5165921296058427e-07, "loss": 0.08941963911056519, "step": 35920 }, { "epoch": 0.9464509734839629, "grad_norm": 2.2458436489105225, "learning_rate": 1.501826141579099e-07, "loss": 0.07643402814865112, "step": 35930 }, { "epoch": 0.9467143887284617, "grad_norm": 3.334146738052368, "learning_rate": 1.48713184477286e-07, "loss": 0.08404995799064637, "step": 35940 }, { "epoch": 0.9469778039729604, "grad_norm": 4.457504749298096, "learning_rate": 1.4725092498822814e-07, "loss": 0.08819855451583862, "step": 35950 }, { "epoch": 0.9472412192174592, "grad_norm": 3.7349424362182617, "learning_rate": 1.457958367550305e-07, "loss": 0.08844594955444336, "step": 35960 }, { "epoch": 0.9475046344619579, "grad_norm": 2.575580596923828, "learning_rate": 1.443479208367693e-07, "loss": 0.07125333547592164, "step": 35970 }, { "epoch": 0.9477680497064567, "grad_norm": 3.203908681869507, "learning_rate": 1.4290717828730037e-07, "loss": 0.08226296901702881, "step": 35980 }, { "epoch": 0.9480314649509554, "grad_norm": 4.012980937957764, "learning_rate": 1.4147361015525717e-07, "loss": 0.08861250281333924, "step": 35990 }, { "epoch": 0.9482948801954542, "grad_norm": 4.21993350982666, "learning_rate": 1.4004721748405392e-07, "loss": 0.09899994730949402, "step": 36000 }, { "epoch": 0.9485582954399528, "grad_norm": 3.4078710079193115, "learning_rate": 1.3862800131188237e-07, "loss": 0.08754963278770447, "step": 36010 }, { "epoch": 0.9488217106844515, "grad_norm": 3.0088276863098145, "learning_rate": 1.37215962671704e-07, "loss": 0.09014828205108642, "step": 36020 }, { "epoch": 0.9490851259289503, "grad_norm": 3.439385175704956, "learning_rate": 1.3581110259126674e-07, "loss": 0.08359556198120117, "step": 36030 }, { "epoch": 0.949348541173449, "grad_norm": 3.191356658935547, "learning_rate": 1.3441342209308595e-07, "loss": 0.08492059707641601, "step": 36040 }, { "epoch": 0.9496119564179478, "grad_norm": 4.071418762207031, "learning_rate": 1.3302292219445457e-07, "loss": 0.08427830934524536, "step": 36050 }, { "epoch": 0.9498753716624465, "grad_norm": 3.5692126750946045, "learning_rate": 1.3163960390743746e-07, "loss": 0.08838143348693847, "step": 36060 }, { "epoch": 0.9501387869069453, "grad_norm": 3.467014789581299, "learning_rate": 1.302634682388726e-07, "loss": 0.08199480772018433, "step": 36070 }, { "epoch": 0.950402202151444, "grad_norm": 3.731837511062622, "learning_rate": 1.2889451619037318e-07, "loss": 0.08311796188354492, "step": 36080 }, { "epoch": 0.9506656173959428, "grad_norm": 4.163635730743408, "learning_rate": 1.2753274875832112e-07, "loss": 0.08623354434967041, "step": 36090 }, { "epoch": 0.9509290326404415, "grad_norm": 2.2847225666046143, "learning_rate": 1.2617816693386907e-07, "loss": 0.07878321409225464, "step": 36100 }, { "epoch": 0.9511924478849402, "grad_norm": 3.721102237701416, "learning_rate": 1.2483077170293954e-07, "loss": 0.09590353965759277, "step": 36110 }, { "epoch": 0.951455863129439, "grad_norm": 3.608051300048828, "learning_rate": 1.2349056404622472e-07, "loss": 0.06376389265060425, "step": 36120 }, { "epoch": 0.9517192783739377, "grad_norm": 3.1441047191619873, "learning_rate": 1.2215754493918763e-07, "loss": 0.08109477758407593, "step": 36130 }, { "epoch": 0.9519826936184365, "grad_norm": 4.73338508605957, "learning_rate": 1.208317153520555e-07, "loss": 0.09561525583267212, "step": 36140 }, { "epoch": 0.9522461088629351, "grad_norm": 3.6228814125061035, "learning_rate": 1.195130762498231e-07, "loss": 0.0803760051727295, "step": 36150 }, { "epoch": 0.952509524107434, "grad_norm": 3.845306396484375, "learning_rate": 1.1820162859225493e-07, "loss": 0.08150204420089721, "step": 36160 }, { "epoch": 0.9527729393519326, "grad_norm": 2.453352212905884, "learning_rate": 1.1689737333387852e-07, "loss": 0.07721207141876221, "step": 36170 }, { "epoch": 0.9530363545964314, "grad_norm": 3.693514108657837, "learning_rate": 1.1560031142398787e-07, "loss": 0.08853591680526733, "step": 36180 }, { "epoch": 0.9532997698409301, "grad_norm": 4.257310390472412, "learning_rate": 1.1431044380663891e-07, "loss": 0.08957104682922364, "step": 36190 }, { "epoch": 0.9535631850854288, "grad_norm": 2.937533140182495, "learning_rate": 1.1302777142065402e-07, "loss": 0.0871384620666504, "step": 36200 }, { "epoch": 0.9538266003299276, "grad_norm": 3.430797815322876, "learning_rate": 1.117522951996175e-07, "loss": 0.0876677930355072, "step": 36210 }, { "epoch": 0.9540900155744263, "grad_norm": 4.412285804748535, "learning_rate": 1.1048401607187565e-07, "loss": 0.08828964233398437, "step": 36220 }, { "epoch": 0.9543534308189251, "grad_norm": 3.9341251850128174, "learning_rate": 1.0922293496053671e-07, "loss": 0.09850395917892456, "step": 36230 }, { "epoch": 0.9546168460634238, "grad_norm": 3.9545650482177734, "learning_rate": 1.0796905278347092e-07, "loss": 0.08144553303718567, "step": 36240 }, { "epoch": 0.9548802613079226, "grad_norm": 3.4662528038024902, "learning_rate": 1.0672237045330491e-07, "loss": 0.08197720646858216, "step": 36250 }, { "epoch": 0.9551436765524213, "grad_norm": 3.6108462810516357, "learning_rate": 1.0548288887743174e-07, "loss": 0.07557978630065917, "step": 36260 }, { "epoch": 0.9554070917969201, "grad_norm": 4.04219913482666, "learning_rate": 1.0425060895799865e-07, "loss": 0.07872780561447143, "step": 36270 }, { "epoch": 0.9556705070414188, "grad_norm": 3.5448224544525146, "learning_rate": 1.0302553159190932e-07, "loss": 0.08445541858673096, "step": 36280 }, { "epoch": 0.9559339222859174, "grad_norm": 3.625551223754883, "learning_rate": 1.0180765767083045e-07, "loss": 0.09312731623649598, "step": 36290 }, { "epoch": 0.9561973375304162, "grad_norm": 3.905733346939087, "learning_rate": 1.00596988081183e-07, "loss": 0.07255728840827942, "step": 36300 }, { "epoch": 0.9564607527749149, "grad_norm": 4.045905590057373, "learning_rate": 9.939352370414435e-08, "loss": 0.07479498386383057, "step": 36310 }, { "epoch": 0.9567241680194137, "grad_norm": 2.4476020336151123, "learning_rate": 9.819726541564711e-08, "loss": 0.07386789321899415, "step": 36320 }, { "epoch": 0.9569875832639124, "grad_norm": 3.1886303424835205, "learning_rate": 9.700821408638039e-08, "loss": 0.08455114364624024, "step": 36330 }, { "epoch": 0.9572509985084112, "grad_norm": 2.1996798515319824, "learning_rate": 9.582637058178635e-08, "loss": 0.0746106505393982, "step": 36340 }, { "epoch": 0.9575144137529099, "grad_norm": 2.6651458740234375, "learning_rate": 9.465173576206355e-08, "loss": 0.08681378364562989, "step": 36350 }, { "epoch": 0.9577778289974086, "grad_norm": 3.9532010555267334, "learning_rate": 9.348431048215923e-08, "loss": 0.09432764649391175, "step": 36360 }, { "epoch": 0.9580412442419074, "grad_norm": 3.585094451904297, "learning_rate": 9.232409559177701e-08, "loss": 0.08474764823913575, "step": 36370 }, { "epoch": 0.9583046594864061, "grad_norm": 5.249664783477783, "learning_rate": 9.117109193537254e-08, "loss": 0.09249668121337891, "step": 36380 }, { "epoch": 0.9585680747309049, "grad_norm": 3.482311248779297, "learning_rate": 9.002530035215007e-08, "loss": 0.08982397317886352, "step": 36390 }, { "epoch": 0.9588314899754036, "grad_norm": 4.877847194671631, "learning_rate": 8.888672167606693e-08, "loss": 0.0793363869190216, "step": 36400 }, { "epoch": 0.9590949052199024, "grad_norm": 3.7395777702331543, "learning_rate": 8.775535673582913e-08, "loss": 0.076598060131073, "step": 36410 }, { "epoch": 0.9593583204644011, "grad_norm": 3.4096601009368896, "learning_rate": 8.663120635489241e-08, "loss": 0.06928057074546815, "step": 36420 }, { "epoch": 0.9596217357088999, "grad_norm": 3.0512166023254395, "learning_rate": 8.55142713514645e-08, "loss": 0.0871402382850647, "step": 36430 }, { "epoch": 0.9598851509533985, "grad_norm": 3.3982033729553223, "learning_rate": 8.440455253849622e-08, "loss": 0.08486660718917846, "step": 36440 }, { "epoch": 0.9601485661978972, "grad_norm": 2.862462043762207, "learning_rate": 8.330205072369035e-08, "loss": 0.089476478099823, "step": 36450 }, { "epoch": 0.960411981442396, "grad_norm": 3.6315507888793945, "learning_rate": 8.22067667094939e-08, "loss": 0.09405800700187683, "step": 36460 }, { "epoch": 0.9606753966868947, "grad_norm": 2.784045934677124, "learning_rate": 8.11187012931025e-08, "loss": 0.08469266891479492, "step": 36470 }, { "epoch": 0.9609388119313935, "grad_norm": 3.3602986335754395, "learning_rate": 8.003785526645824e-08, "loss": 0.07985009551048279, "step": 36480 }, { "epoch": 0.9612022271758922, "grad_norm": 5.066387176513672, "learning_rate": 7.896422941624626e-08, "loss": 0.08420447111129761, "step": 36490 }, { "epoch": 0.961465642420391, "grad_norm": 3.381828784942627, "learning_rate": 7.789782452389705e-08, "loss": 0.08400583267211914, "step": 36500 }, { "epoch": 0.9617290576648897, "grad_norm": 3.455932378768921, "learning_rate": 7.683864136558749e-08, "loss": 0.07780486345291138, "step": 36510 }, { "epoch": 0.9619924729093885, "grad_norm": 3.844573974609375, "learning_rate": 7.578668071223649e-08, "loss": 0.07991347312927247, "step": 36520 }, { "epoch": 0.9622558881538872, "grad_norm": 3.864488363265991, "learning_rate": 7.474194332950602e-08, "loss": 0.07749319076538086, "step": 36530 }, { "epoch": 0.9625193033983859, "grad_norm": 2.811718702316284, "learning_rate": 7.370442997780114e-08, "loss": 0.07241283059120178, "step": 36540 }, { "epoch": 0.9627827186428847, "grad_norm": 3.217768430709839, "learning_rate": 7.267414141227002e-08, "loss": 0.07094998359680176, "step": 36550 }, { "epoch": 0.9630461338873834, "grad_norm": 2.118063449859619, "learning_rate": 7.165107838279838e-08, "loss": 0.09143892526626587, "step": 36560 }, { "epoch": 0.9633095491318822, "grad_norm": 3.2817776203155518, "learning_rate": 7.063524163401836e-08, "loss": 0.09078450798988343, "step": 36570 }, { "epoch": 0.9635729643763808, "grad_norm": 3.9557137489318848, "learning_rate": 6.962663190529851e-08, "loss": 0.08677943348884583, "step": 36580 }, { "epoch": 0.9638363796208796, "grad_norm": 3.7139530181884766, "learning_rate": 6.862524993074825e-08, "loss": 0.07233318090438842, "step": 36590 }, { "epoch": 0.9640997948653783, "grad_norm": 3.5036795139312744, "learning_rate": 6.763109643921683e-08, "loss": 0.0744744062423706, "step": 36600 }, { "epoch": 0.9643632101098771, "grad_norm": 3.9194674491882324, "learning_rate": 6.664417215429319e-08, "loss": 0.09141429662704467, "step": 36610 }, { "epoch": 0.9646266253543758, "grad_norm": 2.860395669937134, "learning_rate": 6.566447779430163e-08, "loss": 0.07996866703033448, "step": 36620 }, { "epoch": 0.9648900405988745, "grad_norm": 3.767771005630493, "learning_rate": 6.469201407230729e-08, "loss": 0.08762515783309936, "step": 36630 }, { "epoch": 0.9651534558433733, "grad_norm": 4.15059757232666, "learning_rate": 6.372678169611068e-08, "loss": 0.08217288255691528, "step": 36640 }, { "epoch": 0.965416871087872, "grad_norm": 4.544426918029785, "learning_rate": 6.276878136824982e-08, "loss": 0.07482454776763917, "step": 36650 }, { "epoch": 0.9656802863323708, "grad_norm": 2.6366777420043945, "learning_rate": 6.181801378599805e-08, "loss": 0.07626769542694092, "step": 36660 }, { "epoch": 0.9659437015768695, "grad_norm": 3.168576717376709, "learning_rate": 6.087447964136517e-08, "loss": 0.06619606018066407, "step": 36670 }, { "epoch": 0.9662071168213683, "grad_norm": 3.2696211338043213, "learning_rate": 5.993817962109627e-08, "loss": 0.08863307237625122, "step": 36680 }, { "epoch": 0.966470532065867, "grad_norm": 3.16584849357605, "learning_rate": 5.9009114406671787e-08, "loss": 0.08352949619293212, "step": 36690 }, { "epoch": 0.9667339473103658, "grad_norm": 4.412471771240234, "learning_rate": 5.8087284674304136e-08, "loss": 0.08502880930900573, "step": 36700 }, { "epoch": 0.9669973625548645, "grad_norm": 4.033834457397461, "learning_rate": 5.7172691094942165e-08, "loss": 0.08050459623336792, "step": 36710 }, { "epoch": 0.9672607777993631, "grad_norm": 3.7639758586883545, "learning_rate": 5.626533433426451e-08, "loss": 0.08850472569465637, "step": 36720 }, { "epoch": 0.967524193043862, "grad_norm": 3.3412997722625732, "learning_rate": 5.5365215052686216e-08, "loss": 0.08847858309745789, "step": 36730 }, { "epoch": 0.9677876082883606, "grad_norm": 5.814715385437012, "learning_rate": 5.4472333905352115e-08, "loss": 0.08090094327926636, "step": 36740 }, { "epoch": 0.9680510235328594, "grad_norm": 3.1189610958099365, "learning_rate": 5.3586691542140135e-08, "loss": 0.07091937065124512, "step": 36750 }, { "epoch": 0.9683144387773581, "grad_norm": 3.1517767906188965, "learning_rate": 5.2708288607659084e-08, "loss": 0.07067171931266784, "step": 36760 }, { "epoch": 0.9685778540218569, "grad_norm": 3.383194923400879, "learning_rate": 5.183712574124755e-08, "loss": 0.07500823736190795, "step": 36770 }, { "epoch": 0.9688412692663556, "grad_norm": 4.062409400939941, "learning_rate": 5.0973203576976104e-08, "loss": 0.07507648468017578, "step": 36780 }, { "epoch": 0.9691046845108544, "grad_norm": 2.9999969005584717, "learning_rate": 5.011652274364509e-08, "loss": 0.07415382862091065, "step": 36790 }, { "epoch": 0.9693680997553531, "grad_norm": 4.069616317749023, "learning_rate": 4.926708386478352e-08, "loss": 0.08441691398620606, "step": 36800 }, { "epoch": 0.9696315149998518, "grad_norm": 3.774214744567871, "learning_rate": 4.8424887558649045e-08, "loss": 0.08694685697555542, "step": 36810 }, { "epoch": 0.9698949302443506, "grad_norm": 2.7483034133911133, "learning_rate": 4.7589934438228017e-08, "loss": 0.08937894105911255, "step": 36820 }, { "epoch": 0.9701583454888493, "grad_norm": 4.071099281311035, "learning_rate": 4.676222511123762e-08, "loss": 0.09397693276405335, "step": 36830 }, { "epoch": 0.9704217607333481, "grad_norm": 5.498852252960205, "learning_rate": 4.59417601801182e-08, "loss": 0.10436527729034424, "step": 36840 }, { "epoch": 0.9706851759778468, "grad_norm": 3.497896671295166, "learning_rate": 4.512854024203983e-08, "loss": 0.09694119691848754, "step": 36850 }, { "epoch": 0.9709485912223456, "grad_norm": 3.203036308288574, "learning_rate": 4.432256588890016e-08, "loss": 0.07720106840133667, "step": 36860 }, { "epoch": 0.9712120064668442, "grad_norm": 3.7742919921875, "learning_rate": 4.352383770732216e-08, "loss": 0.07581187486648559, "step": 36870 }, { "epoch": 0.9714754217113429, "grad_norm": 4.0668487548828125, "learning_rate": 4.273235627865302e-08, "loss": 0.08095640540122986, "step": 36880 }, { "epoch": 0.9717388369558417, "grad_norm": 2.6946113109588623, "learning_rate": 4.19481221789686e-08, "loss": 0.0854729175567627, "step": 36890 }, { "epoch": 0.9720022522003404, "grad_norm": 3.725881576538086, "learning_rate": 4.117113597906675e-08, "loss": 0.08470847606658935, "step": 36900 }, { "epoch": 0.9722656674448392, "grad_norm": 3.586763620376587, "learning_rate": 4.040139824447398e-08, "loss": 0.07579809427261353, "step": 36910 }, { "epoch": 0.9725290826893379, "grad_norm": 4.000285625457764, "learning_rate": 3.963890953543659e-08, "loss": 0.07874160408973693, "step": 36920 }, { "epoch": 0.9727924979338367, "grad_norm": 4.554439067840576, "learning_rate": 3.88836704069262e-08, "loss": 0.08477487564086914, "step": 36930 }, { "epoch": 0.9730559131783354, "grad_norm": 3.5277724266052246, "learning_rate": 3.813568140864088e-08, "loss": 0.09378408789634704, "step": 36940 }, { "epoch": 0.9733193284228342, "grad_norm": 3.1907386779785156, "learning_rate": 3.739494308499736e-08, "loss": 0.09605742692947387, "step": 36950 }, { "epoch": 0.9735827436673329, "grad_norm": 3.3305604457855225, "learning_rate": 3.66614559751377e-08, "loss": 0.07937285900115967, "step": 36960 }, { "epoch": 0.9738461589118316, "grad_norm": 3.4270524978637695, "learning_rate": 3.593522061292598e-08, "loss": 0.06135839223861694, "step": 36970 }, { "epoch": 0.9741095741563304, "grad_norm": 4.440671920776367, "learning_rate": 3.521623752694603e-08, "loss": 0.08360040187835693, "step": 36980 }, { "epoch": 0.9743729894008291, "grad_norm": 4.155294895172119, "learning_rate": 3.4504507240507027e-08, "loss": 0.079620623588562, "step": 36990 }, { "epoch": 0.9746364046453279, "grad_norm": 3.2655599117279053, "learning_rate": 3.380003027163459e-08, "loss": 0.07276584506034851, "step": 37000 }, { "epoch": 0.9748998198898265, "grad_norm": 4.329150199890137, "learning_rate": 3.310280713307856e-08, "loss": 0.08436558842658996, "step": 37010 }, { "epoch": 0.9751632351343253, "grad_norm": 3.6330695152282715, "learning_rate": 3.241283833230746e-08, "loss": 0.0873410701751709, "step": 37020 }, { "epoch": 0.975426650378824, "grad_norm": 4.343902587890625, "learning_rate": 3.1730124371511794e-08, "loss": 0.07808708548545837, "step": 37030 }, { "epoch": 0.9756900656233228, "grad_norm": 3.7799882888793945, "learning_rate": 3.1054665747599635e-08, "loss": 0.07013757228851318, "step": 37040 }, { "epoch": 0.9759534808678215, "grad_norm": 3.7400689125061035, "learning_rate": 3.038646295219771e-08, "loss": 0.1048505425453186, "step": 37050 }, { "epoch": 0.9762168961123202, "grad_norm": 3.1824984550476074, "learning_rate": 2.972551647165367e-08, "loss": 0.09036006927490234, "step": 37060 }, { "epoch": 0.976480311356819, "grad_norm": 4.60860538482666, "learning_rate": 2.9071826787033797e-08, "loss": 0.09596133232116699, "step": 37070 }, { "epoch": 0.9767437266013177, "grad_norm": 3.0890350341796875, "learning_rate": 2.8425394374120842e-08, "loss": 0.07035738229751587, "step": 37080 }, { "epoch": 0.9770071418458165, "grad_norm": 3.9954686164855957, "learning_rate": 2.7786219703415108e-08, "loss": 0.08393914699554443, "step": 37090 }, { "epoch": 0.9772705570903152, "grad_norm": 2.5189878940582275, "learning_rate": 2.715430324013779e-08, "loss": 0.06350737810134888, "step": 37100 }, { "epoch": 0.977533972334814, "grad_norm": 3.623678207397461, "learning_rate": 2.6529645444223206e-08, "loss": 0.09094778895378113, "step": 37110 }, { "epoch": 0.9777973875793127, "grad_norm": 3.224180221557617, "learning_rate": 2.5912246770324333e-08, "loss": 0.07949261665344239, "step": 37120 }, { "epoch": 0.9780608028238115, "grad_norm": 3.5352237224578857, "learning_rate": 2.5302107667811716e-08, "loss": 0.09002360105514526, "step": 37130 }, { "epoch": 0.9783242180683102, "grad_norm": 3.747380495071411, "learning_rate": 2.469922858076901e-08, "loss": 0.08378738164901733, "step": 37140 }, { "epoch": 0.9785876333128088, "grad_norm": 5.853700160980225, "learning_rate": 2.410360994799965e-08, "loss": 0.08653590679168702, "step": 37150 }, { "epoch": 0.9788510485573076, "grad_norm": 3.9639973640441895, "learning_rate": 2.351525220301909e-08, "loss": 0.08241533637046813, "step": 37160 }, { "epoch": 0.9791144638018063, "grad_norm": 3.81485652923584, "learning_rate": 2.293415577406144e-08, "loss": 0.08589037656784057, "step": 37170 }, { "epoch": 0.9793778790463051, "grad_norm": 4.110016822814941, "learning_rate": 2.2360321084071712e-08, "loss": 0.0708618402481079, "step": 37180 }, { "epoch": 0.9796412942908038, "grad_norm": 3.494791030883789, "learning_rate": 2.179374855071137e-08, "loss": 0.09244299530982972, "step": 37190 }, { "epoch": 0.9799047095353026, "grad_norm": 3.900062084197998, "learning_rate": 2.123443858635832e-08, "loss": 0.07841167449951172, "step": 37200 }, { "epoch": 0.9801681247798013, "grad_norm": 3.1805226802825928, "learning_rate": 2.068239159810137e-08, "loss": 0.08180774450302124, "step": 37210 }, { "epoch": 0.9804315400243001, "grad_norm": 5.14160680770874, "learning_rate": 2.0137607987743558e-08, "loss": 0.07862262725830078, "step": 37220 }, { "epoch": 0.9806949552687988, "grad_norm": 4.182098865509033, "learning_rate": 1.9600088151803255e-08, "loss": 0.07307621240615844, "step": 37230 }, { "epoch": 0.9809583705132975, "grad_norm": 3.911027193069458, "learning_rate": 1.906983248150973e-08, "loss": 0.07538008689880371, "step": 37240 }, { "epoch": 0.9812217857577963, "grad_norm": 2.6285107135772705, "learning_rate": 1.854684136280538e-08, "loss": 0.07702048420906067, "step": 37250 }, { "epoch": 0.981485201002295, "grad_norm": 4.185477256774902, "learning_rate": 1.8031115176346815e-08, "loss": 0.09656757116317749, "step": 37260 }, { "epoch": 0.9817486162467938, "grad_norm": 3.451186418533325, "learning_rate": 1.752265429750044e-08, "loss": 0.08813580870628357, "step": 37270 }, { "epoch": 0.9820120314912925, "grad_norm": 5.002378463745117, "learning_rate": 1.7021459096348004e-08, "loss": 0.08359521627426147, "step": 37280 }, { "epoch": 0.9822754467357913, "grad_norm": 3.674851417541504, "learning_rate": 1.652752993767992e-08, "loss": 0.08657060861587525, "step": 37290 }, { "epoch": 0.98253886198029, "grad_norm": 3.324910879135132, "learning_rate": 1.604086718099973e-08, "loss": 0.07822253108024597, "step": 37300 }, { "epoch": 0.9828022772247887, "grad_norm": 4.523258686065674, "learning_rate": 1.556147118051965e-08, "loss": 0.0820618748664856, "step": 37310 }, { "epoch": 0.9830656924692874, "grad_norm": 3.818019151687622, "learning_rate": 1.508934228516723e-08, "loss": 0.0822156548500061, "step": 37320 }, { "epoch": 0.9833291077137861, "grad_norm": 4.794683933258057, "learning_rate": 1.4624480838576483e-08, "loss": 0.0897565245628357, "step": 37330 }, { "epoch": 0.9835925229582849, "grad_norm": 3.0090925693511963, "learning_rate": 1.4166887179094535e-08, "loss": 0.07400492429733277, "step": 37340 }, { "epoch": 0.9838559382027836, "grad_norm": 3.3949666023254395, "learning_rate": 1.3716561639777192e-08, "loss": 0.07866249084472657, "step": 37350 }, { "epoch": 0.9841193534472824, "grad_norm": 3.692749261856079, "learning_rate": 1.3273504548392269e-08, "loss": 0.08149996995925904, "step": 37360 }, { "epoch": 0.9843827686917811, "grad_norm": 3.57616925239563, "learning_rate": 1.2837716227414032e-08, "loss": 0.0743363380432129, "step": 37370 }, { "epoch": 0.9846461839362799, "grad_norm": 2.636650562286377, "learning_rate": 1.240919699402765e-08, "loss": 0.07401443719863891, "step": 37380 }, { "epoch": 0.9849095991807786, "grad_norm": 3.69291615486145, "learning_rate": 1.1987947160130297e-08, "loss": 0.08261181116104126, "step": 37390 }, { "epoch": 0.9851730144252773, "grad_norm": 3.3338537216186523, "learning_rate": 1.1573967032324495e-08, "loss": 0.09007580280303955, "step": 37400 }, { "epoch": 0.9854364296697761, "grad_norm": 3.7413315773010254, "learning_rate": 1.1167256911922552e-08, "loss": 0.0744357943534851, "step": 37410 }, { "epoch": 0.9856998449142748, "grad_norm": 3.2866930961608887, "learning_rate": 1.0767817094945454e-08, "loss": 0.08242416381835938, "step": 37420 }, { "epoch": 0.9859632601587736, "grad_norm": 4.327996253967285, "learning_rate": 1.0375647872123974e-08, "loss": 0.08037994503974914, "step": 37430 }, { "epoch": 0.9862266754032722, "grad_norm": 5.695195198059082, "learning_rate": 9.990749528894227e-09, "loss": 0.08177554607391357, "step": 37440 }, { "epoch": 0.986490090647771, "grad_norm": 3.111041307449341, "learning_rate": 9.613122345403236e-09, "loss": 0.07754155993461609, "step": 37450 }, { "epoch": 0.9867535058922697, "grad_norm": 3.612027168273926, "learning_rate": 9.24276659650336e-09, "loss": 0.0766432523727417, "step": 37460 }, { "epoch": 0.9870169211367685, "grad_norm": 3.3463571071624756, "learning_rate": 8.879682551755643e-09, "loss": 0.08723953366279602, "step": 37470 }, { "epoch": 0.9872803363812672, "grad_norm": 2.5656228065490723, "learning_rate": 8.523870475429797e-09, "loss": 0.08061854839324951, "step": 37480 }, { "epoch": 0.9875437516257659, "grad_norm": 3.1794028282165527, "learning_rate": 8.175330626498668e-09, "loss": 0.06627641320228576, "step": 37490 }, { "epoch": 0.9878071668702647, "grad_norm": 4.609250068664551, "learning_rate": 7.834063258645997e-09, "loss": 0.09483454823493957, "step": 37500 }, { "epoch": 0.9880705821147634, "grad_norm": 3.3669216632843018, "learning_rate": 7.500068620260869e-09, "loss": 0.09351987838745117, "step": 37510 }, { "epoch": 0.9883339973592622, "grad_norm": 4.221609115600586, "learning_rate": 7.173346954438831e-09, "loss": 0.08530002236366271, "step": 37520 }, { "epoch": 0.9885974126037609, "grad_norm": 3.5954926013946533, "learning_rate": 6.853898498982992e-09, "loss": 0.08344804048538208, "step": 37530 }, { "epoch": 0.9888608278482597, "grad_norm": 4.151520252227783, "learning_rate": 6.541723486399587e-09, "loss": 0.07839307188987732, "step": 37540 }, { "epoch": 0.9891242430927584, "grad_norm": 3.627149820327759, "learning_rate": 6.236822143904642e-09, "loss": 0.06852294206619262, "step": 37550 }, { "epoch": 0.9893876583372572, "grad_norm": 3.301591157913208, "learning_rate": 5.939194693418415e-09, "loss": 0.07867291569709778, "step": 37560 }, { "epoch": 0.9896510735817559, "grad_norm": 5.242670059204102, "learning_rate": 5.648841351566514e-09, "loss": 0.08091820478439331, "step": 37570 }, { "epoch": 0.9899144888262545, "grad_norm": 4.044809341430664, "learning_rate": 5.365762329681002e-09, "loss": 0.08063054084777832, "step": 37580 }, { "epoch": 0.9901779040707533, "grad_norm": 3.164921522140503, "learning_rate": 5.089957833798176e-09, "loss": 0.08187781572341919, "step": 37590 }, { "epoch": 0.990441319315252, "grad_norm": 2.6316277980804443, "learning_rate": 4.8214280646619035e-09, "loss": 0.08725019693374633, "step": 37600 }, { "epoch": 0.9907047345597508, "grad_norm": 3.1826586723327637, "learning_rate": 4.560173217718067e-09, "loss": 0.07772318720817566, "step": 37610 }, { "epoch": 0.9909681498042495, "grad_norm": 4.563265323638916, "learning_rate": 4.3061934831212235e-09, "loss": 0.07278558611869812, "step": 37620 }, { "epoch": 0.9912315650487483, "grad_norm": 3.39392352104187, "learning_rate": 4.0594890457268384e-09, "loss": 0.07898827791213989, "step": 37630 }, { "epoch": 0.991494980293247, "grad_norm": 4.88504695892334, "learning_rate": 3.820060085097943e-09, "loss": 0.08067145943641663, "step": 37640 }, { "epoch": 0.9917583955377458, "grad_norm": 5.390471458435059, "learning_rate": 3.587906775501804e-09, "loss": 0.07907751202583313, "step": 37650 }, { "epoch": 0.9920218107822445, "grad_norm": 3.1749935150146484, "learning_rate": 3.363029285907704e-09, "loss": 0.06658693552017211, "step": 37660 }, { "epoch": 0.9922852260267432, "grad_norm": 4.138550758361816, "learning_rate": 3.145427779993604e-09, "loss": 0.09762750267982483, "step": 37670 }, { "epoch": 0.992548641271242, "grad_norm": 2.6504440307617188, "learning_rate": 2.935102416138369e-09, "loss": 0.09829806089401245, "step": 37680 }, { "epoch": 0.9928120565157407, "grad_norm": 3.5712220668792725, "learning_rate": 2.732053347425101e-09, "loss": 0.07435964941978454, "step": 37690 }, { "epoch": 0.9930754717602395, "grad_norm": 3.483119487762451, "learning_rate": 2.536280721642248e-09, "loss": 0.08386969566345215, "step": 37700 }, { "epoch": 0.9933388870047382, "grad_norm": 3.2898802757263184, "learning_rate": 2.3477846812813843e-09, "loss": 0.07648922204971313, "step": 37710 }, { "epoch": 0.993602302249237, "grad_norm": 3.677853584289551, "learning_rate": 2.16656536353832e-09, "loss": 0.08067063093185425, "step": 37720 }, { "epoch": 0.9938657174937356, "grad_norm": 3.759587526321411, "learning_rate": 1.9926229003119912e-09, "loss": 0.07484335899353027, "step": 37730 }, { "epoch": 0.9941291327382344, "grad_norm": 3.45226788520813, "learning_rate": 1.8259574182055707e-09, "loss": 0.08393220901489258, "step": 37740 }, { "epoch": 0.9943925479827331, "grad_norm": 3.2924747467041016, "learning_rate": 1.6665690385253564e-09, "loss": 0.07610039710998535, "step": 37750 }, { "epoch": 0.9946559632272318, "grad_norm": 2.974241018295288, "learning_rate": 1.5144578772807728e-09, "loss": 0.08602898716926574, "step": 37760 }, { "epoch": 0.9949193784717306, "grad_norm": 2.5531136989593506, "learning_rate": 1.36962404518437e-09, "loss": 0.09532580375671387, "step": 37770 }, { "epoch": 0.9951827937162293, "grad_norm": 4.683758735656738, "learning_rate": 1.232067647654045e-09, "loss": 0.08535735607147217, "step": 37780 }, { "epoch": 0.9954462089607281, "grad_norm": 3.42356538772583, "learning_rate": 1.1017887848085995e-09, "loss": 0.08086590766906739, "step": 37790 }, { "epoch": 0.9957096242052268, "grad_norm": 3.1828930377960205, "learning_rate": 9.78787551469962e-10, "loss": 0.07838058471679688, "step": 37800 }, { "epoch": 0.9959730394497256, "grad_norm": 3.7607669830322266, "learning_rate": 8.630640371642963e-10, "loss": 0.09159834980964661, "step": 37810 }, { "epoch": 0.9962364546942243, "grad_norm": 3.714362382888794, "learning_rate": 7.546183261197826e-10, "loss": 0.08276631832122802, "step": 37820 }, { "epoch": 0.9964998699387231, "grad_norm": 4.2483625411987305, "learning_rate": 6.534504972688372e-10, "loss": 0.07708231210708619, "step": 37830 }, { "epoch": 0.9967632851832218, "grad_norm": 3.214555501937866, "learning_rate": 5.595606242447815e-10, "loss": 0.09526807069778442, "step": 37840 }, { "epoch": 0.9970267004277205, "grad_norm": 4.889127731323242, "learning_rate": 4.729487753851736e-10, "loss": 0.07770683765411376, "step": 37850 }, { "epoch": 0.9972901156722193, "grad_norm": 3.810105323791504, "learning_rate": 3.9361501372958686e-10, "loss": 0.08389710187911988, "step": 37860 }, { "epoch": 0.997553530916718, "grad_norm": 3.8718090057373047, "learning_rate": 3.2155939702072093e-10, "loss": 0.08138742446899414, "step": 37870 }, { "epoch": 0.9978169461612167, "grad_norm": 2.7505040168762207, "learning_rate": 2.5678197770329094e-10, "loss": 0.08818286657333374, "step": 37880 }, { "epoch": 0.9980803614057154, "grad_norm": 2.8323912620544434, "learning_rate": 1.9928280292624835e-10, "loss": 0.08287738561630249, "step": 37890 }, { "epoch": 0.9983437766502142, "grad_norm": 4.0904412269592285, "learning_rate": 1.4906191453833985e-10, "loss": 0.0857631504535675, "step": 37900 }, { "epoch": 0.9986071918947129, "grad_norm": 2.9502291679382324, "learning_rate": 1.061193490936585e-10, "loss": 0.07299172878265381, "step": 37910 }, { "epoch": 0.9988706071392117, "grad_norm": 3.2955667972564697, "learning_rate": 7.045513784831315e-11, "loss": 0.07273847460746766, "step": 37920 }, { "epoch": 0.9991340223837104, "grad_norm": 3.432220935821533, "learning_rate": 4.206930675820786e-11, "loss": 0.07065805196762084, "step": 37930 }, { "epoch": 0.9993974376282091, "grad_norm": 3.4736156463623047, "learning_rate": 2.0961876485703358e-11, "loss": 0.07516711950302124, "step": 37940 }, { "epoch": 0.9996608528727079, "grad_norm": 1.761084794998169, "learning_rate": 7.132862392955631e-12, "loss": 0.08214507102966309, "step": 37950 }, { "epoch": 0.9999242681172066, "grad_norm": 4.49199914932251, "learning_rate": 5.822745441363964e-13, "loss": 0.07835026383399964, "step": 37960 }, { "epoch": 1.0, "step": 37963, "total_flos": 2.04116045707887e+20, "train_loss": 0.15196019393188215, "train_runtime": 390703.741, "train_samples_per_second": 6.219, "train_steps_per_second": 0.097 } ], "logging_steps": 10, "max_steps": 37963, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 1000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 2.04116045707887e+20, "train_batch_size": 1, "trial_name": null, "trial_params": null }