{ "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": 1152.328369140625, "learning_rate": 1.5803336259877089e-07, "loss": 2.2921533584594727, "step": 10 }, { "epoch": 0.0005268304889974745, "grad_norm": 53.24566650390625, "learning_rate": 3.336259877085163e-07, "loss": 0.6023933410644531, "step": 20 }, { "epoch": 0.0007902457334962118, "grad_norm": 40.87922286987305, "learning_rate": 5.092186128182617e-07, "loss": 0.296391224861145, "step": 30 }, { "epoch": 0.001053660977994949, "grad_norm": 18.022502899169922, "learning_rate": 6.848112379280071e-07, "loss": 0.2472477912902832, "step": 40 }, { "epoch": 0.0013170762224936862, "grad_norm": 24.571449279785156, "learning_rate": 8.604038630377525e-07, "loss": 0.255388879776001, "step": 50 }, { "epoch": 0.0015804914669924235, "grad_norm": 16.29840087890625, "learning_rate": 1.035996488147498e-06, "loss": 0.235915207862854, "step": 60 }, { "epoch": 0.0018439067114911607, "grad_norm": 14.414058685302734, "learning_rate": 1.2115891132572432e-06, "loss": 0.22037596702575685, "step": 70 }, { "epoch": 0.002107321955989898, "grad_norm": 11.300154685974121, "learning_rate": 1.3871817383669885e-06, "loss": 0.2314291477203369, "step": 80 }, { "epoch": 0.002370737200488635, "grad_norm": 15.075602531433105, "learning_rate": 1.5627743634767342e-06, "loss": 0.21451361179351808, "step": 90 }, { "epoch": 0.0026341524449873723, "grad_norm": 15.50602912902832, "learning_rate": 1.7383669885864795e-06, "loss": 0.21277594566345215, "step": 100 }, { "epoch": 0.00289756768948611, "grad_norm": 13.98180103302002, "learning_rate": 1.913959613696225e-06, "loss": 0.21569375991821288, "step": 110 }, { "epoch": 0.003160982933984847, "grad_norm": 21.818260192871094, "learning_rate": 2.08955223880597e-06, "loss": 0.23122456073760986, "step": 120 }, { "epoch": 0.003424398178483584, "grad_norm": 13.584999084472656, "learning_rate": 2.265144863915716e-06, "loss": 0.20022356510162354, "step": 130 }, { "epoch": 0.0036878134229823213, "grad_norm": 15.440423011779785, "learning_rate": 2.440737489025461e-06, "loss": 0.21240477561950682, "step": 140 }, { "epoch": 0.0039512286674810585, "grad_norm": 13.260856628417969, "learning_rate": 2.6163301141352064e-06, "loss": 0.21666340827941893, "step": 150 }, { "epoch": 0.004214643911979796, "grad_norm": 20.54210662841797, "learning_rate": 2.791922739244952e-06, "loss": 0.23008432388305664, "step": 160 }, { "epoch": 0.004478059156478534, "grad_norm": 14.393362045288086, "learning_rate": 2.9675153643546974e-06, "loss": 0.2277599334716797, "step": 170 }, { "epoch": 0.00474147440097727, "grad_norm": 13.047211647033691, "learning_rate": 3.1431079894644427e-06, "loss": 0.22721343040466307, "step": 180 }, { "epoch": 0.005004889645476008, "grad_norm": 9.628622055053711, "learning_rate": 3.318700614574188e-06, "loss": 0.20976624488830567, "step": 190 }, { "epoch": 0.005268304889974745, "grad_norm": 14.27336597442627, "learning_rate": 3.4942932396839337e-06, "loss": 0.22278010845184326, "step": 200 }, { "epoch": 0.005531720134473482, "grad_norm": 12.907462120056152, "learning_rate": 3.669885864793679e-06, "loss": 0.20920848846435547, "step": 210 }, { "epoch": 0.00579513537897222, "grad_norm": 23.269777297973633, "learning_rate": 3.845478489903424e-06, "loss": 0.20175089836120605, "step": 220 }, { "epoch": 0.0060585506234709565, "grad_norm": 12.854060173034668, "learning_rate": 4.0210711150131695e-06, "loss": 0.21234264373779296, "step": 230 }, { "epoch": 0.006321965867969694, "grad_norm": 14.025676727294922, "learning_rate": 4.196663740122916e-06, "loss": 0.21939070224761964, "step": 240 }, { "epoch": 0.006585381112468432, "grad_norm": 18.676380157470703, "learning_rate": 4.37225636523266e-06, "loss": 0.17606027126312257, "step": 250 }, { "epoch": 0.006848796356967168, "grad_norm": 15.576834678649902, "learning_rate": 4.547848990342406e-06, "loss": 0.21418380737304688, "step": 260 }, { "epoch": 0.007112211601465906, "grad_norm": 7.936060905456543, "learning_rate": 4.7234416154521515e-06, "loss": 0.19268882274627686, "step": 270 }, { "epoch": 0.007375626845964643, "grad_norm": 17.274049758911133, "learning_rate": 4.899034240561897e-06, "loss": 0.19788388013839722, "step": 280 }, { "epoch": 0.00763904209046338, "grad_norm": 11.116101264953613, "learning_rate": 5.074626865671642e-06, "loss": 0.19115767478942872, "step": 290 }, { "epoch": 0.007902457334962117, "grad_norm": 11.506976127624512, "learning_rate": 5.250219490781387e-06, "loss": 0.1990084171295166, "step": 300 }, { "epoch": 0.008165872579460855, "grad_norm": 10.175557136535645, "learning_rate": 5.425812115891133e-06, "loss": 0.1898605465888977, "step": 310 }, { "epoch": 0.008429287823959592, "grad_norm": 13.681167602539062, "learning_rate": 5.601404741000879e-06, "loss": 0.19849462509155275, "step": 320 }, { "epoch": 0.008692703068458329, "grad_norm": 7.95233154296875, "learning_rate": 5.776997366110624e-06, "loss": 0.20587432384490967, "step": 330 }, { "epoch": 0.008956118312957067, "grad_norm": 15.228087425231934, "learning_rate": 5.9525899912203685e-06, "loss": 0.190270733833313, "step": 340 }, { "epoch": 0.009219533557455804, "grad_norm": 10.498289108276367, "learning_rate": 6.128182616330115e-06, "loss": 0.2132331132888794, "step": 350 }, { "epoch": 0.00948294880195454, "grad_norm": 9.87523365020752, "learning_rate": 6.30377524143986e-06, "loss": 0.19962100982666015, "step": 360 }, { "epoch": 0.00974636404645328, "grad_norm": 8.791894912719727, "learning_rate": 6.479367866549605e-06, "loss": 0.20329029560089112, "step": 370 }, { "epoch": 0.010009779290952016, "grad_norm": 7.236010551452637, "learning_rate": 6.654960491659351e-06, "loss": 0.2059494972229004, "step": 380 }, { "epoch": 0.010273194535450753, "grad_norm": 13.132527351379395, "learning_rate": 6.830553116769097e-06, "loss": 0.19868574142456055, "step": 390 }, { "epoch": 0.01053660977994949, "grad_norm": 7.777538299560547, "learning_rate": 7.006145741878841e-06, "loss": 0.19532399177551268, "step": 400 }, { "epoch": 0.010800025024448228, "grad_norm": 4.6001434326171875, "learning_rate": 7.181738366988587e-06, "loss": 0.18898736238479613, "step": 410 }, { "epoch": 0.011063440268946964, "grad_norm": 6.627692699432373, "learning_rate": 7.3573309920983325e-06, "loss": 0.19578258991241454, "step": 420 }, { "epoch": 0.011326855513445701, "grad_norm": 12.506808280944824, "learning_rate": 7.532923617208078e-06, "loss": 0.17923154830932617, "step": 430 }, { "epoch": 0.01159027075794444, "grad_norm": 5.822325229644775, "learning_rate": 7.708516242317823e-06, "loss": 0.18727375268936158, "step": 440 }, { "epoch": 0.011853686002443176, "grad_norm": 7.834691524505615, "learning_rate": 7.88410886742757e-06, "loss": 0.19795167446136475, "step": 450 }, { "epoch": 0.012117101246941913, "grad_norm": 7.860893249511719, "learning_rate": 8.059701492537314e-06, "loss": 0.19256291389465333, "step": 460 }, { "epoch": 0.012380516491440651, "grad_norm": 6.469077110290527, "learning_rate": 8.23529411764706e-06, "loss": 0.18206284046173096, "step": 470 }, { "epoch": 0.012643931735939388, "grad_norm": 7.27107572555542, "learning_rate": 8.410886742756806e-06, "loss": 0.19195011854171753, "step": 480 }, { "epoch": 0.012907346980438125, "grad_norm": 8.728857040405273, "learning_rate": 8.58647936786655e-06, "loss": 0.19165066480636597, "step": 490 }, { "epoch": 0.013170762224936863, "grad_norm": 15.100836753845215, "learning_rate": 8.762071992976295e-06, "loss": 0.20890650749206544, "step": 500 }, { "epoch": 0.0134341774694356, "grad_norm": 7.575761795043945, "learning_rate": 8.937664618086041e-06, "loss": 0.2074503183364868, "step": 510 }, { "epoch": 0.013697592713934337, "grad_norm": 11.206635475158691, "learning_rate": 9.113257243195787e-06, "loss": 0.20359530448913574, "step": 520 }, { "epoch": 0.013961007958433075, "grad_norm": 6.768627166748047, "learning_rate": 9.288849868305532e-06, "loss": 0.19439611434936524, "step": 530 }, { "epoch": 0.014224423202931812, "grad_norm": 10.721158981323242, "learning_rate": 9.464442493415278e-06, "loss": 0.19293508529663086, "step": 540 }, { "epoch": 0.014487838447430549, "grad_norm": 4.952611446380615, "learning_rate": 9.640035118525024e-06, "loss": 0.18917618989944457, "step": 550 }, { "epoch": 0.014751253691929285, "grad_norm": 6.42905855178833, "learning_rate": 9.815627743634768e-06, "loss": 0.1814356803894043, "step": 560 }, { "epoch": 0.015014668936428024, "grad_norm": 5.743300914764404, "learning_rate": 9.991220368744513e-06, "loss": 0.1905747890472412, "step": 570 }, { "epoch": 0.01527808418092676, "grad_norm": 8.628737449645996, "learning_rate": 1.0166812993854259e-05, "loss": 0.2002119541168213, "step": 580 }, { "epoch": 0.015541499425425497, "grad_norm": 10.12346076965332, "learning_rate": 1.0342405618964005e-05, "loss": 0.1829677939414978, "step": 590 }, { "epoch": 0.015804914669924234, "grad_norm": 10.190818786621094, "learning_rate": 1.0517998244073751e-05, "loss": 0.18982141017913817, "step": 600 }, { "epoch": 0.016068329914422974, "grad_norm": 6.896692276000977, "learning_rate": 1.0693590869183494e-05, "loss": 0.1902841091156006, "step": 610 }, { "epoch": 0.01633174515892171, "grad_norm": 5.719489574432373, "learning_rate": 1.086918349429324e-05, "loss": 0.18204526901245116, "step": 620 }, { "epoch": 0.016595160403420448, "grad_norm": 12.944291114807129, "learning_rate": 1.1044776119402986e-05, "loss": 0.20504298210144042, "step": 630 }, { "epoch": 0.016858575647919184, "grad_norm": 7.089725971221924, "learning_rate": 1.122036874451273e-05, "loss": 0.18443610668182372, "step": 640 }, { "epoch": 0.01712199089241792, "grad_norm": 8.072535514831543, "learning_rate": 1.1395961369622477e-05, "loss": 0.18972475528717042, "step": 650 }, { "epoch": 0.017385406136916658, "grad_norm": 4.6718058586120605, "learning_rate": 1.1571553994732223e-05, "loss": 0.21578578948974608, "step": 660 }, { "epoch": 0.017648821381415394, "grad_norm": 6.208983421325684, "learning_rate": 1.1747146619841967e-05, "loss": 0.19793095588684081, "step": 670 }, { "epoch": 0.017912236625914135, "grad_norm": 6.143943786621094, "learning_rate": 1.1922739244951713e-05, "loss": 0.18916568756103516, "step": 680 }, { "epoch": 0.01817565187041287, "grad_norm": 8.256867408752441, "learning_rate": 1.209833187006146e-05, "loss": 0.198850417137146, "step": 690 }, { "epoch": 0.018439067114911608, "grad_norm": 7.259819984436035, "learning_rate": 1.2273924495171204e-05, "loss": 0.19684672355651855, "step": 700 }, { "epoch": 0.018702482359410345, "grad_norm": 7.9615797996521, "learning_rate": 1.244951712028095e-05, "loss": 0.18070366382598876, "step": 710 }, { "epoch": 0.01896589760390908, "grad_norm": 7.950150012969971, "learning_rate": 1.2625109745390696e-05, "loss": 0.1864357352256775, "step": 720 }, { "epoch": 0.019229312848407818, "grad_norm": 4.535582542419434, "learning_rate": 1.2800702370500439e-05, "loss": 0.18751511573791504, "step": 730 }, { "epoch": 0.01949272809290656, "grad_norm": 8.837357521057129, "learning_rate": 1.2976294995610185e-05, "loss": 0.19537696838378907, "step": 740 }, { "epoch": 0.019756143337405295, "grad_norm": 5.999636173248291, "learning_rate": 1.315188762071993e-05, "loss": 0.20178117752075195, "step": 750 }, { "epoch": 0.02001955858190403, "grad_norm": 8.001970291137695, "learning_rate": 1.3327480245829676e-05, "loss": 0.19602066278457642, "step": 760 }, { "epoch": 0.02028297382640277, "grad_norm": 8.192914009094238, "learning_rate": 1.3503072870939422e-05, "loss": 0.20744049549102783, "step": 770 }, { "epoch": 0.020546389070901505, "grad_norm": 3.5170412063598633, "learning_rate": 1.3678665496049166e-05, "loss": 0.19172439575195313, "step": 780 }, { "epoch": 0.020809804315400242, "grad_norm": 4.965949058532715, "learning_rate": 1.3854258121158912e-05, "loss": 0.18401081562042237, "step": 790 }, { "epoch": 0.02107321955989898, "grad_norm": 16.927871704101562, "learning_rate": 1.4029850746268658e-05, "loss": 0.21336770057678223, "step": 800 }, { "epoch": 0.02133663480439772, "grad_norm": 4.541973114013672, "learning_rate": 1.4205443371378403e-05, "loss": 0.18735945224761963, "step": 810 }, { "epoch": 0.021600050048896455, "grad_norm": 5.559776782989502, "learning_rate": 1.4381035996488149e-05, "loss": 0.18534114360809326, "step": 820 }, { "epoch": 0.021863465293395192, "grad_norm": 3.7679874897003174, "learning_rate": 1.4556628621597895e-05, "loss": 0.1903618335723877, "step": 830 }, { "epoch": 0.02212688053789393, "grad_norm": 4.4505438804626465, "learning_rate": 1.4732221246707638e-05, "loss": 0.1874035120010376, "step": 840 }, { "epoch": 0.022390295782392666, "grad_norm": 5.4106526374816895, "learning_rate": 1.4907813871817384e-05, "loss": 0.1886696219444275, "step": 850 }, { "epoch": 0.022653711026891402, "grad_norm": 4.4938764572143555, "learning_rate": 1.508340649692713e-05, "loss": 0.18186020851135254, "step": 860 }, { "epoch": 0.022917126271390142, "grad_norm": 5.467157363891602, "learning_rate": 1.5258999122036876e-05, "loss": 0.18999900817871093, "step": 870 }, { "epoch": 0.02318054151588888, "grad_norm": 4.836270332336426, "learning_rate": 1.5434591747146622e-05, "loss": 0.19789689779281616, "step": 880 }, { "epoch": 0.023443956760387616, "grad_norm": 4.793219566345215, "learning_rate": 1.561018437225637e-05, "loss": 0.17963579893112183, "step": 890 }, { "epoch": 0.023707372004886353, "grad_norm": 4.370285987854004, "learning_rate": 1.578577699736611e-05, "loss": 0.1766575813293457, "step": 900 }, { "epoch": 0.02397078724938509, "grad_norm": 2.7577741146087646, "learning_rate": 1.5961369622475857e-05, "loss": 0.19224463701248168, "step": 910 }, { "epoch": 0.024234202493883826, "grad_norm": 10.941360473632812, "learning_rate": 1.6136962247585604e-05, "loss": 0.193387770652771, "step": 920 }, { "epoch": 0.024497617738382566, "grad_norm": 4.191863059997559, "learning_rate": 1.6312554872695346e-05, "loss": 0.1966615915298462, "step": 930 }, { "epoch": 0.024761032982881303, "grad_norm": 3.8482110500335693, "learning_rate": 1.6488147497805093e-05, "loss": 0.18322458267211914, "step": 940 }, { "epoch": 0.02502444822738004, "grad_norm": 3.763852834701538, "learning_rate": 1.666374012291484e-05, "loss": 0.17601606845855713, "step": 950 }, { "epoch": 0.025287863471878776, "grad_norm": 4.1300764083862305, "learning_rate": 1.6839332748024585e-05, "loss": 0.18427841663360595, "step": 960 }, { "epoch": 0.025551278716377513, "grad_norm": 2.6957623958587646, "learning_rate": 1.701492537313433e-05, "loss": 0.17454148530960084, "step": 970 }, { "epoch": 0.02581469396087625, "grad_norm": 8.33347225189209, "learning_rate": 1.7190517998244077e-05, "loss": 0.19688470363616944, "step": 980 }, { "epoch": 0.026078109205374986, "grad_norm": 3.952312469482422, "learning_rate": 1.736611062335382e-05, "loss": 0.18648204803466797, "step": 990 }, { "epoch": 0.026341524449873727, "grad_norm": 7.210516452789307, "learning_rate": 1.7541703248463566e-05, "loss": 0.19502214193344117, "step": 1000 }, { "epoch": 0.026604939694372463, "grad_norm": 5.301218032836914, "learning_rate": 1.771729587357331e-05, "loss": 0.1977167844772339, "step": 1010 }, { "epoch": 0.0268683549388712, "grad_norm": 5.971017360687256, "learning_rate": 1.7892888498683055e-05, "loss": 0.19554243087768555, "step": 1020 }, { "epoch": 0.027131770183369937, "grad_norm": 3.021047592163086, "learning_rate": 1.80684811237928e-05, "loss": 0.17699878215789794, "step": 1030 }, { "epoch": 0.027395185427868674, "grad_norm": 3.62266206741333, "learning_rate": 1.8244073748902547e-05, "loss": 0.18992218971252442, "step": 1040 }, { "epoch": 0.02765860067236741, "grad_norm": 4.98801851272583, "learning_rate": 1.8419666374012293e-05, "loss": 0.18033676147460936, "step": 1050 }, { "epoch": 0.02792201591686615, "grad_norm": 4.392043590545654, "learning_rate": 1.859525899912204e-05, "loss": 0.19512935876846313, "step": 1060 }, { "epoch": 0.028185431161364887, "grad_norm": 3.946234941482544, "learning_rate": 1.8770851624231782e-05, "loss": 0.17863694429397584, "step": 1070 }, { "epoch": 0.028448846405863624, "grad_norm": 2.5624539852142334, "learning_rate": 1.8946444249341528e-05, "loss": 0.20050721168518065, "step": 1080 }, { "epoch": 0.02871226165036236, "grad_norm": 7.076059818267822, "learning_rate": 1.9122036874451274e-05, "loss": 0.1963038682937622, "step": 1090 }, { "epoch": 0.028975676894861097, "grad_norm": 3.339928388595581, "learning_rate": 1.929762949956102e-05, "loss": 0.18488069772720336, "step": 1100 }, { "epoch": 0.029239092139359834, "grad_norm": 4.308832168579102, "learning_rate": 1.9473222124670767e-05, "loss": 0.19442265033721923, "step": 1110 }, { "epoch": 0.02950250738385857, "grad_norm": 4.275976181030273, "learning_rate": 1.9648814749780513e-05, "loss": 0.18063654899597167, "step": 1120 }, { "epoch": 0.02976592262835731, "grad_norm": 5.688209533691406, "learning_rate": 1.9824407374890256e-05, "loss": 0.18211650848388672, "step": 1130 }, { "epoch": 0.030029337872856048, "grad_norm": 3.5611820220947266, "learning_rate": 2e-05, "loss": 0.18691667318344116, "step": 1140 }, { "epoch": 0.030292753117354784, "grad_norm": 7.6922736167907715, "learning_rate": 1.9999996360784282e-05, "loss": 0.186564040184021, "step": 1150 }, { "epoch": 0.03055616836185352, "grad_norm": 6.01164436340332, "learning_rate": 1.9999985443139778e-05, "loss": 0.1817225217819214, "step": 1160 }, { "epoch": 0.030819583606352258, "grad_norm": 3.347769021987915, "learning_rate": 1.999996724707443e-05, "loss": 0.18834023475646972, "step": 1170 }, { "epoch": 0.031082998850850994, "grad_norm": 3.8269312381744385, "learning_rate": 1.9999941772601482e-05, "loss": 0.1785709023475647, "step": 1180 }, { "epoch": 0.03134641409534973, "grad_norm": 6.100703716278076, "learning_rate": 1.9999909019739482e-05, "loss": 0.1834322690963745, "step": 1190 }, { "epoch": 0.03160982933984847, "grad_norm": 5.194270610809326, "learning_rate": 1.9999868988512265e-05, "loss": 0.17318860292434693, "step": 1200 }, { "epoch": 0.031873244584347205, "grad_norm": 10.909791946411133, "learning_rate": 1.9999821678948964e-05, "loss": 0.17627689838409424, "step": 1210 }, { "epoch": 0.03213665982884595, "grad_norm": 7.049479007720947, "learning_rate": 1.999976709108402e-05, "loss": 0.19043148756027223, "step": 1220 }, { "epoch": 0.032400075073344685, "grad_norm": 7.8877482414245605, "learning_rate": 1.9999705224957156e-05, "loss": 0.19992196559906006, "step": 1230 }, { "epoch": 0.03266349031784342, "grad_norm": 13.307353019714355, "learning_rate": 1.9999636080613406e-05, "loss": 0.19174755811691285, "step": 1240 }, { "epoch": 0.03292690556234216, "grad_norm": 5.622386455535889, "learning_rate": 1.9999559658103097e-05, "loss": 0.18984978199005126, "step": 1250 }, { "epoch": 0.033190320806840895, "grad_norm": 6.2683868408203125, "learning_rate": 1.9999475957481848e-05, "loss": 0.19611685276031493, "step": 1260 }, { "epoch": 0.03345373605133963, "grad_norm": 2.1613709926605225, "learning_rate": 1.999938497881059e-05, "loss": 0.19802818298339844, "step": 1270 }, { "epoch": 0.03371715129583837, "grad_norm": 3.495927095413208, "learning_rate": 1.999928672215553e-05, "loss": 0.1806408166885376, "step": 1280 }, { "epoch": 0.033980566540337105, "grad_norm": 2.8071630001068115, "learning_rate": 1.9999181187588187e-05, "loss": 0.19101006984710694, "step": 1290 }, { "epoch": 0.03424398178483584, "grad_norm": 5.270023822784424, "learning_rate": 1.9999068375185372e-05, "loss": 0.1793213129043579, "step": 1300 }, { "epoch": 0.03450739702933458, "grad_norm": 8.78927993774414, "learning_rate": 1.99989482850292e-05, "loss": 0.18469719886779784, "step": 1310 }, { "epoch": 0.034770812273833315, "grad_norm": 2.6327006816864014, "learning_rate": 1.9998820917207076e-05, "loss": 0.18053780794143676, "step": 1320 }, { "epoch": 0.03503422751833205, "grad_norm": 3.03661847114563, "learning_rate": 1.99986862718117e-05, "loss": 0.18466596603393554, "step": 1330 }, { "epoch": 0.03529764276283079, "grad_norm": 4.018127918243408, "learning_rate": 1.999854434894108e-05, "loss": 0.19183743000030518, "step": 1340 }, { "epoch": 0.03556105800732953, "grad_norm": 5.794654846191406, "learning_rate": 1.9998395148698505e-05, "loss": 0.18500161170959473, "step": 1350 }, { "epoch": 0.03582447325182827, "grad_norm": 10.268134117126465, "learning_rate": 1.9998238671192575e-05, "loss": 0.1844891667366028, "step": 1360 }, { "epoch": 0.036087888496327006, "grad_norm": 6.5594587326049805, "learning_rate": 1.999807491653718e-05, "loss": 0.18833087682723998, "step": 1370 }, { "epoch": 0.03635130374082574, "grad_norm": 4.935270309448242, "learning_rate": 1.9997903884851503e-05, "loss": 0.18165212869644165, "step": 1380 }, { "epoch": 0.03661471898532448, "grad_norm": 2.7506721019744873, "learning_rate": 1.999772557626004e-05, "loss": 0.18111629486083985, "step": 1390 }, { "epoch": 0.036878134229823216, "grad_norm": 3.6175403594970703, "learning_rate": 1.9997539990892555e-05, "loss": 0.17043757438659668, "step": 1400 }, { "epoch": 0.03714154947432195, "grad_norm": 3.795468807220459, "learning_rate": 1.9997347128884138e-05, "loss": 0.18552510738372802, "step": 1410 }, { "epoch": 0.03740496471882069, "grad_norm": 2.2657439708709717, "learning_rate": 1.9997146990375162e-05, "loss": 0.18940937519073486, "step": 1420 }, { "epoch": 0.037668379963319426, "grad_norm": 5.322626113891602, "learning_rate": 1.999693957551129e-05, "loss": 0.1799783945083618, "step": 1430 }, { "epoch": 0.03793179520781816, "grad_norm": 5.264209270477295, "learning_rate": 1.999672488444349e-05, "loss": 0.17427442073822022, "step": 1440 }, { "epoch": 0.0381952104523169, "grad_norm": 4.974515438079834, "learning_rate": 1.999650291732802e-05, "loss": 0.20000867843627929, "step": 1450 }, { "epoch": 0.038458625696815636, "grad_norm": 2.055105447769165, "learning_rate": 1.9996273674326446e-05, "loss": 0.1915614366531372, "step": 1460 }, { "epoch": 0.03872204094131437, "grad_norm": 3.892733573913574, "learning_rate": 1.9996037155605615e-05, "loss": 0.17680710554122925, "step": 1470 }, { "epoch": 0.03898545618581312, "grad_norm": 4.135645389556885, "learning_rate": 1.9995793361337676e-05, "loss": 0.18537298440933228, "step": 1480 }, { "epoch": 0.03924887143031185, "grad_norm": 4.757589340209961, "learning_rate": 1.9995542291700075e-05, "loss": 0.18772364854812623, "step": 1490 }, { "epoch": 0.03951228667481059, "grad_norm": 3.802520275115967, "learning_rate": 1.9995283946875548e-05, "loss": 0.1810229539871216, "step": 1500 }, { "epoch": 0.03977570191930933, "grad_norm": 6.306492805480957, "learning_rate": 1.999501832705213e-05, "loss": 0.16938204765319825, "step": 1510 }, { "epoch": 0.04003911716380806, "grad_norm": 2.684715986251831, "learning_rate": 1.9994745432423154e-05, "loss": 0.16983799934387206, "step": 1520 }, { "epoch": 0.0403025324083068, "grad_norm": 5.0829339027404785, "learning_rate": 1.9994465263187242e-05, "loss": 0.16737825870513917, "step": 1530 }, { "epoch": 0.04056594765280554, "grad_norm": 5.963609218597412, "learning_rate": 1.9994177819548313e-05, "loss": 0.19454680681228637, "step": 1540 }, { "epoch": 0.040829362897304274, "grad_norm": 3.9762139320373535, "learning_rate": 1.999388310171558e-05, "loss": 0.18905284404754638, "step": 1550 }, { "epoch": 0.04109277814180301, "grad_norm": 5.083252429962158, "learning_rate": 1.9993581109903557e-05, "loss": 0.16726791858673096, "step": 1560 }, { "epoch": 0.04135619338630175, "grad_norm": 7.779130935668945, "learning_rate": 1.999327184433204e-05, "loss": 0.17934694290161132, "step": 1570 }, { "epoch": 0.041619608630800484, "grad_norm": 3.100452184677124, "learning_rate": 1.9992955305226127e-05, "loss": 0.18014085292816162, "step": 1580 }, { "epoch": 0.04188302387529922, "grad_norm": 2.0381829738616943, "learning_rate": 1.9992631492816214e-05, "loss": 0.177481746673584, "step": 1590 }, { "epoch": 0.04214643911979796, "grad_norm": 2.282123565673828, "learning_rate": 1.9992300407337977e-05, "loss": 0.1934949517250061, "step": 1600 }, { "epoch": 0.0424098543642967, "grad_norm": 8.144119262695312, "learning_rate": 1.9991962049032403e-05, "loss": 0.17203974723815918, "step": 1610 }, { "epoch": 0.04267326960879544, "grad_norm": 7.016752243041992, "learning_rate": 1.9991616418145757e-05, "loss": 0.17974287271499634, "step": 1620 }, { "epoch": 0.042936684853294174, "grad_norm": 3.7450339794158936, "learning_rate": 1.9991263514929606e-05, "loss": 0.17064465284347535, "step": 1630 }, { "epoch": 0.04320010009779291, "grad_norm": 6.8023362159729, "learning_rate": 1.999090333964081e-05, "loss": 0.19019980430603028, "step": 1640 }, { "epoch": 0.04346351534229165, "grad_norm": 2.3212924003601074, "learning_rate": 1.9990535892541524e-05, "loss": 0.17250568866729737, "step": 1650 }, { "epoch": 0.043726930586790384, "grad_norm": 4.346439838409424, "learning_rate": 1.999016117389918e-05, "loss": 0.19721512794494628, "step": 1660 }, { "epoch": 0.04399034583128912, "grad_norm": 4.991761207580566, "learning_rate": 1.998977918398653e-05, "loss": 0.1916358709335327, "step": 1670 }, { "epoch": 0.04425376107578786, "grad_norm": 2.4757213592529297, "learning_rate": 1.9989389923081587e-05, "loss": 0.1718665599822998, "step": 1680 }, { "epoch": 0.044517176320286594, "grad_norm": 3.323908805847168, "learning_rate": 1.998899339146768e-05, "loss": 0.17189912796020507, "step": 1690 }, { "epoch": 0.04478059156478533, "grad_norm": 2.2458603382110596, "learning_rate": 1.9988589589433423e-05, "loss": 0.18796297311782836, "step": 1700 }, { "epoch": 0.04504400680928407, "grad_norm": 5.711018085479736, "learning_rate": 1.9988178517272717e-05, "loss": 0.1873321056365967, "step": 1710 }, { "epoch": 0.045307422053782805, "grad_norm": 3.353743076324463, "learning_rate": 1.998776017528476e-05, "loss": 0.18296506404876708, "step": 1720 }, { "epoch": 0.04557083729828155, "grad_norm": 5.4835638999938965, "learning_rate": 1.9987334563774038e-05, "loss": 0.18708770275115966, "step": 1730 }, { "epoch": 0.045834252542780285, "grad_norm": 5.980207443237305, "learning_rate": 1.998690168305033e-05, "loss": 0.17635705471038818, "step": 1740 }, { "epoch": 0.04609766778727902, "grad_norm": 2.4297149181365967, "learning_rate": 1.9986461533428708e-05, "loss": 0.18557387590408325, "step": 1750 }, { "epoch": 0.04636108303177776, "grad_norm": 3.991760492324829, "learning_rate": 1.9986014115229525e-05, "loss": 0.17830429077148438, "step": 1760 }, { "epoch": 0.046624498276276495, "grad_norm": 2.5410196781158447, "learning_rate": 1.998555942877844e-05, "loss": 0.18656494617462158, "step": 1770 }, { "epoch": 0.04688791352077523, "grad_norm": 3.4378178119659424, "learning_rate": 1.9985097474406387e-05, "loss": 0.18236327171325684, "step": 1780 }, { "epoch": 0.04715132876527397, "grad_norm": 3.7632594108581543, "learning_rate": 1.99846282524496e-05, "loss": 0.1710899829864502, "step": 1790 }, { "epoch": 0.047414744009772705, "grad_norm": 4.051877021789551, "learning_rate": 1.9984151763249598e-05, "loss": 0.18267159461975097, "step": 1800 }, { "epoch": 0.04767815925427144, "grad_norm": 2.482800245285034, "learning_rate": 1.9983668007153185e-05, "loss": 0.18755812644958497, "step": 1810 }, { "epoch": 0.04794157449877018, "grad_norm": 3.5258572101593018, "learning_rate": 1.998317698451247e-05, "loss": 0.17787575721740723, "step": 1820 }, { "epoch": 0.048204989743268915, "grad_norm": 2.8729662895202637, "learning_rate": 1.9982678695684833e-05, "loss": 0.1760495901107788, "step": 1830 }, { "epoch": 0.04846840498776765, "grad_norm": 2.32561993598938, "learning_rate": 1.998217314103295e-05, "loss": 0.18195037841796874, "step": 1840 }, { "epoch": 0.04873182023226639, "grad_norm": 2.518605947494507, "learning_rate": 1.998166032092479e-05, "loss": 0.18083409070968628, "step": 1850 }, { "epoch": 0.04899523547676513, "grad_norm": 2.134213924407959, "learning_rate": 1.99811402357336e-05, "loss": 0.16671687364578247, "step": 1860 }, { "epoch": 0.04925865072126387, "grad_norm": 3.77367901802063, "learning_rate": 1.9980612885837925e-05, "loss": 0.18224892616271973, "step": 1870 }, { "epoch": 0.049522065965762606, "grad_norm": 2.437218189239502, "learning_rate": 1.998007827162159e-05, "loss": 0.17267928123474122, "step": 1880 }, { "epoch": 0.04978548121026134, "grad_norm": 5.750367641448975, "learning_rate": 1.9979536393473715e-05, "loss": 0.17804194688796998, "step": 1890 }, { "epoch": 0.05004889645476008, "grad_norm": 3.634906053543091, "learning_rate": 1.9978987251788696e-05, "loss": 0.17651128768920898, "step": 1900 }, { "epoch": 0.050312311699258816, "grad_norm": 5.417503356933594, "learning_rate": 1.997843084696623e-05, "loss": 0.18060888051986695, "step": 1910 }, { "epoch": 0.05057572694375755, "grad_norm": 6.066929817199707, "learning_rate": 1.9977867179411282e-05, "loss": 0.15588653087615967, "step": 1920 }, { "epoch": 0.05083914218825629, "grad_norm": 3.1211531162261963, "learning_rate": 1.997729624953412e-05, "loss": 0.18444516658782958, "step": 1930 }, { "epoch": 0.051102557432755026, "grad_norm": 2.0735909938812256, "learning_rate": 1.997671805775029e-05, "loss": 0.1793699860572815, "step": 1940 }, { "epoch": 0.05136597267725376, "grad_norm": 4.4888410568237305, "learning_rate": 1.9976132604480625e-05, "loss": 0.17873330116271974, "step": 1950 }, { "epoch": 0.0516293879217525, "grad_norm": 3.4976069927215576, "learning_rate": 1.997553989015125e-05, "loss": 0.1917967438697815, "step": 1960 }, { "epoch": 0.051892803166251236, "grad_norm": 4.024819850921631, "learning_rate": 1.9974939915193557e-05, "loss": 0.17278671264648438, "step": 1970 }, { "epoch": 0.05215621841074997, "grad_norm": 2.090906858444214, "learning_rate": 1.9974332680044237e-05, "loss": 0.18849658966064453, "step": 1980 }, { "epoch": 0.05241963365524872, "grad_norm": 1.8004672527313232, "learning_rate": 1.9973718185145262e-05, "loss": 0.17083253860473632, "step": 1990 }, { "epoch": 0.05268304889974745, "grad_norm": 5.4738945960998535, "learning_rate": 1.997309643094389e-05, "loss": 0.1764271855354309, "step": 2000 }, { "epoch": 0.05294646414424619, "grad_norm": 1.4949977397918701, "learning_rate": 1.997246741789266e-05, "loss": 0.17532150745391845, "step": 2010 }, { "epoch": 0.05320987938874493, "grad_norm": 4.458802700042725, "learning_rate": 1.99718311464494e-05, "loss": 0.18669071197509765, "step": 2020 }, { "epoch": 0.05347329463324366, "grad_norm": 3.317828416824341, "learning_rate": 1.9971187617077204e-05, "loss": 0.17434899806976317, "step": 2030 }, { "epoch": 0.0537367098777424, "grad_norm": 3.6948554515838623, "learning_rate": 1.9970536830244466e-05, "loss": 0.18317074775695802, "step": 2040 }, { "epoch": 0.05400012512224114, "grad_norm": 1.7569202184677124, "learning_rate": 1.9969878786424863e-05, "loss": 0.17278764247894288, "step": 2050 }, { "epoch": 0.054263540366739874, "grad_norm": 2.873075485229492, "learning_rate": 1.9969213486097335e-05, "loss": 0.18763284683227538, "step": 2060 }, { "epoch": 0.05452695561123861, "grad_norm": 5.63405704498291, "learning_rate": 1.996854092974613e-05, "loss": 0.18789607286453247, "step": 2070 }, { "epoch": 0.05479037085573735, "grad_norm": 4.186635494232178, "learning_rate": 1.9967861117860754e-05, "loss": 0.1719210147857666, "step": 2080 }, { "epoch": 0.055053786100236084, "grad_norm": 2.9324874877929688, "learning_rate": 1.9967174050936008e-05, "loss": 0.17930634021759034, "step": 2090 }, { "epoch": 0.05531720134473482, "grad_norm": 2.3921971321105957, "learning_rate": 1.9966479729471966e-05, "loss": 0.18952457904815673, "step": 2100 }, { "epoch": 0.05558061658923356, "grad_norm": 5.487048625946045, "learning_rate": 1.9965778153973986e-05, "loss": 0.17647802829742432, "step": 2110 }, { "epoch": 0.0558440318337323, "grad_norm": 3.37650990486145, "learning_rate": 1.9965069324952703e-05, "loss": 0.18460919857025146, "step": 2120 }, { "epoch": 0.05610744707823104, "grad_norm": 6.557918548583984, "learning_rate": 1.9964353242924038e-05, "loss": 0.1848422646522522, "step": 2130 }, { "epoch": 0.056370862322729774, "grad_norm": 2.1119277477264404, "learning_rate": 1.9963629908409184e-05, "loss": 0.17867151498794556, "step": 2140 }, { "epoch": 0.05663427756722851, "grad_norm": 2.011385917663574, "learning_rate": 1.9962899321934616e-05, "loss": 0.1687416672706604, "step": 2150 }, { "epoch": 0.05689769281172725, "grad_norm": 3.723785638809204, "learning_rate": 1.9962161484032085e-05, "loss": 0.16300532817840577, "step": 2160 }, { "epoch": 0.057161108056225984, "grad_norm": 3.484182119369507, "learning_rate": 1.9961416395238623e-05, "loss": 0.1595086932182312, "step": 2170 }, { "epoch": 0.05742452330072472, "grad_norm": 3.789327621459961, "learning_rate": 1.9960664056096534e-05, "loss": 0.1948120951652527, "step": 2180 }, { "epoch": 0.05768793854522346, "grad_norm": 8.10851001739502, "learning_rate": 1.9959904467153408e-05, "loss": 0.18976277112960815, "step": 2190 }, { "epoch": 0.057951353789722194, "grad_norm": 1.5798625946044922, "learning_rate": 1.99591376289621e-05, "loss": 0.17186059951782226, "step": 2200 }, { "epoch": 0.05821476903422093, "grad_norm": 4.5596089363098145, "learning_rate": 1.9958363542080758e-05, "loss": 0.1741294026374817, "step": 2210 }, { "epoch": 0.05847818427871967, "grad_norm": 2.616273880004883, "learning_rate": 1.9957582207072784e-05, "loss": 0.16312651634216307, "step": 2220 }, { "epoch": 0.058741599523218405, "grad_norm": 4.992913722991943, "learning_rate": 1.9956793624506875e-05, "loss": 0.18014094829559327, "step": 2230 }, { "epoch": 0.05900501476771714, "grad_norm": 5.733574867248535, "learning_rate": 1.9955997794956994e-05, "loss": 0.16741901636123657, "step": 2240 }, { "epoch": 0.059268430012215885, "grad_norm": 4.148486614227295, "learning_rate": 1.995519471900238e-05, "loss": 0.17606362104415893, "step": 2250 }, { "epoch": 0.05953184525671462, "grad_norm": 2.7025930881500244, "learning_rate": 1.9954384397227545e-05, "loss": 0.17655752897262572, "step": 2260 }, { "epoch": 0.05979526050121336, "grad_norm": 4.61444091796875, "learning_rate": 1.9953566830222276e-05, "loss": 0.17268358469009398, "step": 2270 }, { "epoch": 0.060058675745712095, "grad_norm": 3.0413057804107666, "learning_rate": 1.9952742018581634e-05, "loss": 0.18095309734344484, "step": 2280 }, { "epoch": 0.06032209099021083, "grad_norm": 3.9616031646728516, "learning_rate": 1.9951909962905956e-05, "loss": 0.1733607530593872, "step": 2290 }, { "epoch": 0.06058550623470957, "grad_norm": 4.53289794921875, "learning_rate": 1.9951070663800843e-05, "loss": 0.17100341320037843, "step": 2300 }, { "epoch": 0.060848921479208305, "grad_norm": 1.641640543937683, "learning_rate": 1.995022412187718e-05, "loss": 0.17253880500793456, "step": 2310 }, { "epoch": 0.06111233672370704, "grad_norm": 3.070230722427368, "learning_rate": 1.9949370337751105e-05, "loss": 0.17532062530517578, "step": 2320 }, { "epoch": 0.06137575196820578, "grad_norm": 3.503784418106079, "learning_rate": 1.994850931204405e-05, "loss": 0.1754580020904541, "step": 2330 }, { "epoch": 0.061639167212704515, "grad_norm": 8.64353084564209, "learning_rate": 1.99476410453827e-05, "loss": 0.1808714747428894, "step": 2340 }, { "epoch": 0.06190258245720325, "grad_norm": 5.480358600616455, "learning_rate": 1.994676553839902e-05, "loss": 0.18993451595306396, "step": 2350 }, { "epoch": 0.06216599770170199, "grad_norm": 3.213327407836914, "learning_rate": 1.9945882791730246e-05, "loss": 0.19034571647644044, "step": 2360 }, { "epoch": 0.06242941294620073, "grad_norm": 4.61894416809082, "learning_rate": 1.994499280601887e-05, "loss": 0.17468146085739136, "step": 2370 }, { "epoch": 0.06269282819069946, "grad_norm": 4.142031192779541, "learning_rate": 1.994409558191267e-05, "loss": 0.16522612571716308, "step": 2380 }, { "epoch": 0.0629562434351982, "grad_norm": 5.989138126373291, "learning_rate": 1.9943191120064672e-05, "loss": 0.18381273746490479, "step": 2390 }, { "epoch": 0.06321965867969694, "grad_norm": 2.901590347290039, "learning_rate": 1.99422794211332e-05, "loss": 0.17069602012634277, "step": 2400 }, { "epoch": 0.06348307392419568, "grad_norm": 2.072641372680664, "learning_rate": 1.9941360485781814e-05, "loss": 0.17681868076324464, "step": 2410 }, { "epoch": 0.06374648916869441, "grad_norm": 2.1396398544311523, "learning_rate": 1.9940434314679363e-05, "loss": 0.17210711240768434, "step": 2420 }, { "epoch": 0.06400990441319315, "grad_norm": 2.26560640335083, "learning_rate": 1.9939500908499947e-05, "loss": 0.16813669204711915, "step": 2430 }, { "epoch": 0.0642733196576919, "grad_norm": 8.008849143981934, "learning_rate": 1.9938560267922948e-05, "loss": 0.18098126649856566, "step": 2440 }, { "epoch": 0.06453673490219063, "grad_norm": 2.0089199542999268, "learning_rate": 1.9937612393632995e-05, "loss": 0.15457288026809693, "step": 2450 }, { "epoch": 0.06480015014668937, "grad_norm": 2.8384549617767334, "learning_rate": 1.9936657286320003e-05, "loss": 0.16238348484039306, "step": 2460 }, { "epoch": 0.0650635653911881, "grad_norm": 2.881422519683838, "learning_rate": 1.993569494667913e-05, "loss": 0.17691339254379274, "step": 2470 }, { "epoch": 0.06532698063568684, "grad_norm": 4.515578746795654, "learning_rate": 1.9934725375410812e-05, "loss": 0.18680375814437866, "step": 2480 }, { "epoch": 0.06559039588018557, "grad_norm": 2.62628436088562, "learning_rate": 1.9933748573220745e-05, "loss": 0.18407900333404542, "step": 2490 }, { "epoch": 0.06585381112468432, "grad_norm": 4.831400394439697, "learning_rate": 1.993276454081989e-05, "loss": 0.17046328783035278, "step": 2500 }, { "epoch": 0.06611722636918305, "grad_norm": 3.9050111770629883, "learning_rate": 1.9931773278924467e-05, "loss": 0.1819709300994873, "step": 2510 }, { "epoch": 0.06638064161368179, "grad_norm": 3.71109676361084, "learning_rate": 1.9930774788255957e-05, "loss": 0.1815136194229126, "step": 2520 }, { "epoch": 0.06664405685818052, "grad_norm": 2.489482879638672, "learning_rate": 1.992976906954111e-05, "loss": 0.18606064319610596, "step": 2530 }, { "epoch": 0.06690747210267926, "grad_norm": 2.457472324371338, "learning_rate": 1.992875612351192e-05, "loss": 0.17208917140960694, "step": 2540 }, { "epoch": 0.067170887347178, "grad_norm": 1.3004209995269775, "learning_rate": 1.9927735950905663e-05, "loss": 0.16000609397888182, "step": 2550 }, { "epoch": 0.06743430259167674, "grad_norm": 1.3555023670196533, "learning_rate": 1.9926708552464865e-05, "loss": 0.19217021465301515, "step": 2560 }, { "epoch": 0.06769771783617548, "grad_norm": 6.2479681968688965, "learning_rate": 1.9925673928937307e-05, "loss": 0.1808590054512024, "step": 2570 }, { "epoch": 0.06796113308067421, "grad_norm": 2.9503400325775146, "learning_rate": 1.9924632081076032e-05, "loss": 0.17420923709869385, "step": 2580 }, { "epoch": 0.06822454832517295, "grad_norm": 2.7353217601776123, "learning_rate": 1.9923583009639344e-05, "loss": 0.1892242193222046, "step": 2590 }, { "epoch": 0.06848796356967168, "grad_norm": 4.0648298263549805, "learning_rate": 1.9922526715390802e-05, "loss": 0.1771496057510376, "step": 2600 }, { "epoch": 0.06875137881417043, "grad_norm": 4.616885662078857, "learning_rate": 1.9921463199099222e-05, "loss": 0.1761288285255432, "step": 2610 }, { "epoch": 0.06901479405866916, "grad_norm": 3.145737886428833, "learning_rate": 1.992039246153868e-05, "loss": 0.18354207277297974, "step": 2620 }, { "epoch": 0.0692782093031679, "grad_norm": 5.019693374633789, "learning_rate": 1.99193145034885e-05, "loss": 0.1677647352218628, "step": 2630 }, { "epoch": 0.06954162454766663, "grad_norm": 2.3524997234344482, "learning_rate": 1.9918229325733267e-05, "loss": 0.1712871789932251, "step": 2640 }, { "epoch": 0.06980503979216537, "grad_norm": 2.0392887592315674, "learning_rate": 1.9917136929062824e-05, "loss": 0.1712141752243042, "step": 2650 }, { "epoch": 0.0700684550366641, "grad_norm": 5.247335910797119, "learning_rate": 1.991603731427226e-05, "loss": 0.17658209800720215, "step": 2660 }, { "epoch": 0.07033187028116285, "grad_norm": 6.426676273345947, "learning_rate": 1.9914930482161928e-05, "loss": 0.16792676448822022, "step": 2670 }, { "epoch": 0.07059528552566158, "grad_norm": 1.8738642930984497, "learning_rate": 1.9913816433537425e-05, "loss": 0.15786761045455933, "step": 2680 }, { "epoch": 0.07085870077016032, "grad_norm": 2.951785087585449, "learning_rate": 1.99126951692096e-05, "loss": 0.1775432348251343, "step": 2690 }, { "epoch": 0.07112211601465906, "grad_norm": 3.7512433528900146, "learning_rate": 1.991156668999456e-05, "loss": 0.178620445728302, "step": 2700 }, { "epoch": 0.0713855312591578, "grad_norm": 3.120042562484741, "learning_rate": 1.9910430996713668e-05, "loss": 0.16860768795013428, "step": 2710 }, { "epoch": 0.07164894650365654, "grad_norm": 3.3676984310150146, "learning_rate": 1.9909288090193518e-05, "loss": 0.18325366973876953, "step": 2720 }, { "epoch": 0.07191236174815527, "grad_norm": 1.9209805727005005, "learning_rate": 1.9908137971265977e-05, "loss": 0.18560723066329957, "step": 2730 }, { "epoch": 0.07217577699265401, "grad_norm": 3.8658828735351562, "learning_rate": 1.9906980640768146e-05, "loss": 0.17366669178009034, "step": 2740 }, { "epoch": 0.07243919223715274, "grad_norm": 5.231012344360352, "learning_rate": 1.9905816099542382e-05, "loss": 0.1777738571166992, "step": 2750 }, { "epoch": 0.07270260748165149, "grad_norm": 4.587658405303955, "learning_rate": 1.9904644348436282e-05, "loss": 0.17792006731033325, "step": 2760 }, { "epoch": 0.07296602272615021, "grad_norm": 3.067178964614868, "learning_rate": 1.990346538830271e-05, "loss": 0.1707811713218689, "step": 2770 }, { "epoch": 0.07322943797064896, "grad_norm": 3.2902185916900635, "learning_rate": 1.9902279219999752e-05, "loss": 0.1681424617767334, "step": 2780 }, { "epoch": 0.07349285321514769, "grad_norm": 5.043569087982178, "learning_rate": 1.9901085844390757e-05, "loss": 0.17096856832504273, "step": 2790 }, { "epoch": 0.07375626845964643, "grad_norm": 6.7907514572143555, "learning_rate": 1.9899885262344316e-05, "loss": 0.180657696723938, "step": 2800 }, { "epoch": 0.07401968370414516, "grad_norm": 2.6912126541137695, "learning_rate": 1.9898677474734266e-05, "loss": 0.1735234260559082, "step": 2810 }, { "epoch": 0.0742830989486439, "grad_norm": 2.958493947982788, "learning_rate": 1.9897462482439684e-05, "loss": 0.17359732389450072, "step": 2820 }, { "epoch": 0.07454651419314265, "grad_norm": 4.76692008972168, "learning_rate": 1.9896240286344892e-05, "loss": 0.1881261348724365, "step": 2830 }, { "epoch": 0.07480992943764138, "grad_norm": 4.098763465881348, "learning_rate": 1.989501088733946e-05, "loss": 0.17516133785247803, "step": 2840 }, { "epoch": 0.07507334468214012, "grad_norm": 2.250487804412842, "learning_rate": 1.9893774286318198e-05, "loss": 0.1729832887649536, "step": 2850 }, { "epoch": 0.07533675992663885, "grad_norm": 8.362164497375488, "learning_rate": 1.9892530484181155e-05, "loss": 0.18024682998657227, "step": 2860 }, { "epoch": 0.0756001751711376, "grad_norm": 3.4269020557403564, "learning_rate": 1.989127948183363e-05, "loss": 0.17086901664733886, "step": 2870 }, { "epoch": 0.07586359041563633, "grad_norm": 4.559165954589844, "learning_rate": 1.9890021280186148e-05, "loss": 0.1902235746383667, "step": 2880 }, { "epoch": 0.07612700566013507, "grad_norm": 2.6864426136016846, "learning_rate": 1.9888755880154485e-05, "loss": 0.16129323244094848, "step": 2890 }, { "epoch": 0.0763904209046338, "grad_norm": 5.316067218780518, "learning_rate": 1.988748328265966e-05, "loss": 0.15233986377716063, "step": 2900 }, { "epoch": 0.07665383614913254, "grad_norm": 2.6330599784851074, "learning_rate": 1.9886203488627918e-05, "loss": 0.18195056915283203, "step": 2910 }, { "epoch": 0.07691725139363127, "grad_norm": 3.2509405612945557, "learning_rate": 1.988491649899075e-05, "loss": 0.17461755275726318, "step": 2920 }, { "epoch": 0.07718066663813002, "grad_norm": 5.570692539215088, "learning_rate": 1.988362231468488e-05, "loss": 0.16273342370986937, "step": 2930 }, { "epoch": 0.07744408188262875, "grad_norm": 2.4209914207458496, "learning_rate": 1.9882320936652278e-05, "loss": 0.18293979167938232, "step": 2940 }, { "epoch": 0.07770749712712749, "grad_norm": 2.6601483821868896, "learning_rate": 1.9881012365840138e-05, "loss": 0.16551487445831298, "step": 2950 }, { "epoch": 0.07797091237162623, "grad_norm": 3.4437880516052246, "learning_rate": 1.987969660320089e-05, "loss": 0.17204588651657104, "step": 2960 }, { "epoch": 0.07823432761612496, "grad_norm": 1.1793370246887207, "learning_rate": 1.9878373649692213e-05, "loss": 0.18510870933532714, "step": 2970 }, { "epoch": 0.0784977428606237, "grad_norm": 1.5268296003341675, "learning_rate": 1.9877043506277004e-05, "loss": 0.18761029243469238, "step": 2980 }, { "epoch": 0.07876115810512244, "grad_norm": 3.550591468811035, "learning_rate": 1.9875706173923396e-05, "loss": 0.17621891498565673, "step": 2990 }, { "epoch": 0.07902457334962118, "grad_norm": 2.5785014629364014, "learning_rate": 1.9874361653604765e-05, "loss": 0.16148853302001953, "step": 3000 }, { "epoch": 0.07928798859411991, "grad_norm": 3.2898190021514893, "learning_rate": 1.9873009946299702e-05, "loss": 0.18031697273254393, "step": 3010 }, { "epoch": 0.07955140383861865, "grad_norm": 3.906134843826294, "learning_rate": 1.9871651052992046e-05, "loss": 0.163431978225708, "step": 3020 }, { "epoch": 0.07981481908311738, "grad_norm": 4.695989608764648, "learning_rate": 1.987028497467085e-05, "loss": 0.1563675284385681, "step": 3030 }, { "epoch": 0.08007823432761613, "grad_norm": 2.627660036087036, "learning_rate": 1.986891171233041e-05, "loss": 0.16314175128936767, "step": 3040 }, { "epoch": 0.08034164957211486, "grad_norm": 4.297965049743652, "learning_rate": 1.9867531266970247e-05, "loss": 0.17625702619552613, "step": 3050 }, { "epoch": 0.0806050648166136, "grad_norm": 1.8680833578109741, "learning_rate": 1.9866143639595107e-05, "loss": 0.17714593410491944, "step": 3060 }, { "epoch": 0.08086848006111233, "grad_norm": 2.6812148094177246, "learning_rate": 1.9864748831214964e-05, "loss": 0.16127517223358154, "step": 3070 }, { "epoch": 0.08113189530561107, "grad_norm": 2.5916965007781982, "learning_rate": 1.986334684284502e-05, "loss": 0.18602524995803832, "step": 3080 }, { "epoch": 0.08139531055010982, "grad_norm": 5.094557285308838, "learning_rate": 1.98619376755057e-05, "loss": 0.17858562469482422, "step": 3090 }, { "epoch": 0.08165872579460855, "grad_norm": 1.6852796077728271, "learning_rate": 1.986052133022266e-05, "loss": 0.19167789220809936, "step": 3100 }, { "epoch": 0.08192214103910729, "grad_norm": 3.395786762237549, "learning_rate": 1.985909780802678e-05, "loss": 0.1657545804977417, "step": 3110 }, { "epoch": 0.08218555628360602, "grad_norm": 3.009049654006958, "learning_rate": 1.985766710995416e-05, "loss": 0.16537847518920898, "step": 3120 }, { "epoch": 0.08244897152810476, "grad_norm": 7.040225982666016, "learning_rate": 1.9856229237046115e-05, "loss": 0.17790117263793945, "step": 3130 }, { "epoch": 0.0827123867726035, "grad_norm": 3.518348455429077, "learning_rate": 1.9854784190349203e-05, "loss": 0.1682518720626831, "step": 3140 }, { "epoch": 0.08297580201710224, "grad_norm": 5.3045830726623535, "learning_rate": 1.985333197091518e-05, "loss": 0.1774523973464966, "step": 3150 }, { "epoch": 0.08323921726160097, "grad_norm": 4.842081069946289, "learning_rate": 1.9851872579801044e-05, "loss": 0.16894032955169677, "step": 3160 }, { "epoch": 0.08350263250609971, "grad_norm": 1.3791265487670898, "learning_rate": 1.9850406018068997e-05, "loss": 0.16362123489379882, "step": 3170 }, { "epoch": 0.08376604775059844, "grad_norm": 4.844457626342773, "learning_rate": 1.984893228678647e-05, "loss": 0.16251940727233888, "step": 3180 }, { "epoch": 0.08402946299509718, "grad_norm": 4.47798490524292, "learning_rate": 1.9847451387026103e-05, "loss": 0.17158786058425904, "step": 3190 }, { "epoch": 0.08429287823959591, "grad_norm": 4.031536102294922, "learning_rate": 1.9845963319865758e-05, "loss": 0.1822810411453247, "step": 3200 }, { "epoch": 0.08455629348409466, "grad_norm": 1.835968255996704, "learning_rate": 1.9844468086388523e-05, "loss": 0.16297447681427002, "step": 3210 }, { "epoch": 0.0848197087285934, "grad_norm": 6.039210796356201, "learning_rate": 1.9842965687682686e-05, "loss": 0.16359148025512696, "step": 3220 }, { "epoch": 0.08508312397309213, "grad_norm": 2.432617664337158, "learning_rate": 1.9841456124841763e-05, "loss": 0.17819063663482665, "step": 3230 }, { "epoch": 0.08534653921759087, "grad_norm": 3.4029951095581055, "learning_rate": 1.983993939896447e-05, "loss": 0.16387171745300294, "step": 3240 }, { "epoch": 0.0856099544620896, "grad_norm": 2.858003616333008, "learning_rate": 1.9838415511154755e-05, "loss": 0.18370270729064941, "step": 3250 }, { "epoch": 0.08587336970658835, "grad_norm": 3.8369314670562744, "learning_rate": 1.9836884462521768e-05, "loss": 0.1609644651412964, "step": 3260 }, { "epoch": 0.08613678495108708, "grad_norm": 2.536306142807007, "learning_rate": 1.9835346254179865e-05, "loss": 0.1752711057662964, "step": 3270 }, { "epoch": 0.08640020019558582, "grad_norm": 3.0543346405029297, "learning_rate": 1.9833800887248628e-05, "loss": 0.17627136707305907, "step": 3280 }, { "epoch": 0.08666361544008455, "grad_norm": 5.7723069190979, "learning_rate": 1.983224836285284e-05, "loss": 0.17592887878417968, "step": 3290 }, { "epoch": 0.0869270306845833, "grad_norm": 6.413233280181885, "learning_rate": 1.9830688682122494e-05, "loss": 0.18242404460906983, "step": 3300 }, { "epoch": 0.08719044592908202, "grad_norm": 4.198527812957764, "learning_rate": 1.982912184619279e-05, "loss": 0.16997164487838745, "step": 3310 }, { "epoch": 0.08745386117358077, "grad_norm": 8.266045570373535, "learning_rate": 1.9827547856204147e-05, "loss": 0.1653149127960205, "step": 3320 }, { "epoch": 0.0877172764180795, "grad_norm": 3.1471633911132812, "learning_rate": 1.9825966713302172e-05, "loss": 0.17106237411499023, "step": 3330 }, { "epoch": 0.08798069166257824, "grad_norm": 3.9877264499664307, "learning_rate": 1.98243784186377e-05, "loss": 0.16934133768081666, "step": 3340 }, { "epoch": 0.08824410690707699, "grad_norm": 4.233682155609131, "learning_rate": 1.982278297336675e-05, "loss": 0.17890199422836303, "step": 3350 }, { "epoch": 0.08850752215157572, "grad_norm": 2.335585594177246, "learning_rate": 1.9821180378650563e-05, "loss": 0.18082044124603272, "step": 3360 }, { "epoch": 0.08877093739607446, "grad_norm": 4.758769512176514, "learning_rate": 1.9819570635655577e-05, "loss": 0.17483808994293212, "step": 3370 }, { "epoch": 0.08903435264057319, "grad_norm": 4.552272796630859, "learning_rate": 1.981795374555343e-05, "loss": 0.16557178497314454, "step": 3380 }, { "epoch": 0.08929776788507193, "grad_norm": 2.1353912353515625, "learning_rate": 1.981632970952096e-05, "loss": 0.16436500549316407, "step": 3390 }, { "epoch": 0.08956118312957066, "grad_norm": 1.4865621328353882, "learning_rate": 1.9814698528740222e-05, "loss": 0.1703031063079834, "step": 3400 }, { "epoch": 0.0898245983740694, "grad_norm": 4.681690216064453, "learning_rate": 1.9813060204398448e-05, "loss": 0.17173923254013063, "step": 3410 }, { "epoch": 0.09008801361856814, "grad_norm": 3.613391637802124, "learning_rate": 1.9811414737688088e-05, "loss": 0.1711006760597229, "step": 3420 }, { "epoch": 0.09035142886306688, "grad_norm": 2.5374746322631836, "learning_rate": 1.9809762129806777e-05, "loss": 0.17804735898971558, "step": 3430 }, { "epoch": 0.09061484410756561, "grad_norm": 3.0925421714782715, "learning_rate": 1.9808102381957364e-05, "loss": 0.18218311071395873, "step": 3440 }, { "epoch": 0.09087825935206435, "grad_norm": 3.349320650100708, "learning_rate": 1.980643549534788e-05, "loss": 0.1723076343536377, "step": 3450 }, { "epoch": 0.0911416745965631, "grad_norm": 2.1947596073150635, "learning_rate": 1.980476147119155e-05, "loss": 0.15935126543045045, "step": 3460 }, { "epoch": 0.09140508984106183, "grad_norm": 1.086989402770996, "learning_rate": 1.980308031070682e-05, "loss": 0.15949246883392335, "step": 3470 }, { "epoch": 0.09166850508556057, "grad_norm": 2.2655980587005615, "learning_rate": 1.9801392015117286e-05, "loss": 0.1816182851791382, "step": 3480 }, { "epoch": 0.0919319203300593, "grad_norm": 4.004907608032227, "learning_rate": 1.9799696585651782e-05, "loss": 0.155777645111084, "step": 3490 }, { "epoch": 0.09219533557455804, "grad_norm": 1.4335718154907227, "learning_rate": 1.9797994023544307e-05, "loss": 0.18433351516723634, "step": 3500 }, { "epoch": 0.09245875081905677, "grad_norm": 2.2632503509521484, "learning_rate": 1.979628433003406e-05, "loss": 0.1719195604324341, "step": 3510 }, { "epoch": 0.09272216606355552, "grad_norm": 1.5530011653900146, "learning_rate": 1.9794567506365432e-05, "loss": 0.17292529344558716, "step": 3520 }, { "epoch": 0.09298558130805425, "grad_norm": 3.6977782249450684, "learning_rate": 1.9792843553788e-05, "loss": 0.17024219036102295, "step": 3530 }, { "epoch": 0.09324899655255299, "grad_norm": 2.9129791259765625, "learning_rate": 1.9791112473556528e-05, "loss": 0.18272478580474855, "step": 3540 }, { "epoch": 0.09351241179705172, "grad_norm": 1.9461795091629028, "learning_rate": 1.978937426693097e-05, "loss": 0.1778128981590271, "step": 3550 }, { "epoch": 0.09377582704155046, "grad_norm": 2.3384053707122803, "learning_rate": 1.9787628935176476e-05, "loss": 0.17771589756011963, "step": 3560 }, { "epoch": 0.0940392422860492, "grad_norm": 4.476064205169678, "learning_rate": 1.9785876479563367e-05, "loss": 0.18107062578201294, "step": 3570 }, { "epoch": 0.09430265753054794, "grad_norm": 1.050663709640503, "learning_rate": 1.978411690136716e-05, "loss": 0.17528812885284423, "step": 3580 }, { "epoch": 0.09456607277504668, "grad_norm": 2.7553138732910156, "learning_rate": 1.9782350201868547e-05, "loss": 0.17033900022506715, "step": 3590 }, { "epoch": 0.09482948801954541, "grad_norm": 3.9455323219299316, "learning_rate": 1.978057638235341e-05, "loss": 0.1651599645614624, "step": 3600 }, { "epoch": 0.09509290326404415, "grad_norm": 3.7094411849975586, "learning_rate": 1.9778795444112812e-05, "loss": 0.17019755840301515, "step": 3610 }, { "epoch": 0.09535631850854288, "grad_norm": 7.700146675109863, "learning_rate": 1.9777007388443e-05, "loss": 0.1697389602661133, "step": 3620 }, { "epoch": 0.09561973375304163, "grad_norm": 2.6274774074554443, "learning_rate": 1.9775212216645393e-05, "loss": 0.16202213764190673, "step": 3630 }, { "epoch": 0.09588314899754036, "grad_norm": 1.2037841081619263, "learning_rate": 1.977340993002659e-05, "loss": 0.17769014835357666, "step": 3640 }, { "epoch": 0.0961465642420391, "grad_norm": 4.711949348449707, "learning_rate": 1.977160052989839e-05, "loss": 0.1812081217765808, "step": 3650 }, { "epoch": 0.09640997948653783, "grad_norm": 2.0244901180267334, "learning_rate": 1.9769784017577734e-05, "loss": 0.16496962308883667, "step": 3660 }, { "epoch": 0.09667339473103657, "grad_norm": 9.259961128234863, "learning_rate": 1.976796039438677e-05, "loss": 0.18384387493133544, "step": 3670 }, { "epoch": 0.0969368099755353, "grad_norm": 5.980914115905762, "learning_rate": 1.9766129661652803e-05, "loss": 0.18516829013824462, "step": 3680 }, { "epoch": 0.09720022522003405, "grad_norm": 5.269677639007568, "learning_rate": 1.9764291820708324e-05, "loss": 0.17961907386779785, "step": 3690 }, { "epoch": 0.09746364046453278, "grad_norm": 3.752739906311035, "learning_rate": 1.9762446872890984e-05, "loss": 0.1666250228881836, "step": 3700 }, { "epoch": 0.09772705570903152, "grad_norm": 2.861088514328003, "learning_rate": 1.976059481954363e-05, "loss": 0.1820595979690552, "step": 3710 }, { "epoch": 0.09799047095353026, "grad_norm": 2.309786558151245, "learning_rate": 1.9758735662014253e-05, "loss": 0.16194965839385986, "step": 3720 }, { "epoch": 0.098253886198029, "grad_norm": 2.840271472930908, "learning_rate": 1.975686940165604e-05, "loss": 0.15821869373321534, "step": 3730 }, { "epoch": 0.09851730144252774, "grad_norm": 1.734982967376709, "learning_rate": 1.9754996039827322e-05, "loss": 0.1562150239944458, "step": 3740 }, { "epoch": 0.09878071668702647, "grad_norm": 3.1049816608428955, "learning_rate": 1.9753115577891623e-05, "loss": 0.1673650026321411, "step": 3750 }, { "epoch": 0.09904413193152521, "grad_norm": 1.7939774990081787, "learning_rate": 1.9751228017217623e-05, "loss": 0.15565838813781738, "step": 3760 }, { "epoch": 0.09930754717602394, "grad_norm": 2.4716339111328125, "learning_rate": 1.9749333359179166e-05, "loss": 0.1717700481414795, "step": 3770 }, { "epoch": 0.09957096242052269, "grad_norm": 2.0183722972869873, "learning_rate": 1.974743160515527e-05, "loss": 0.1674565315246582, "step": 3780 }, { "epoch": 0.09983437766502141, "grad_norm": 4.1907477378845215, "learning_rate": 1.974552275653011e-05, "loss": 0.17433459758758546, "step": 3790 }, { "epoch": 0.10009779290952016, "grad_norm": 3.112013578414917, "learning_rate": 1.974360681469303e-05, "loss": 0.17303165197372436, "step": 3800 }, { "epoch": 0.10036120815401889, "grad_norm": 2.2174038887023926, "learning_rate": 1.9741683781038538e-05, "loss": 0.1740640878677368, "step": 3810 }, { "epoch": 0.10062462339851763, "grad_norm": 4.024685859680176, "learning_rate": 1.9739753656966294e-05, "loss": 0.17245259284973144, "step": 3820 }, { "epoch": 0.10088803864301636, "grad_norm": 2.6979150772094727, "learning_rate": 1.9737816443881134e-05, "loss": 0.18186430931091307, "step": 3830 }, { "epoch": 0.1011514538875151, "grad_norm": 3.9280519485473633, "learning_rate": 1.973587214319304e-05, "loss": 0.16206233501434325, "step": 3840 }, { "epoch": 0.10141486913201385, "grad_norm": 2.1714742183685303, "learning_rate": 1.9733920756317157e-05, "loss": 0.16689469814300537, "step": 3850 }, { "epoch": 0.10167828437651258, "grad_norm": 1.80259108543396, "learning_rate": 1.973196228467379e-05, "loss": 0.16823095083236694, "step": 3860 }, { "epoch": 0.10194169962101132, "grad_norm": 2.290109872817993, "learning_rate": 1.9729996729688403e-05, "loss": 0.1784023404121399, "step": 3870 }, { "epoch": 0.10220511486551005, "grad_norm": 2.846423625946045, "learning_rate": 1.9728024092791606e-05, "loss": 0.17834924459457396, "step": 3880 }, { "epoch": 0.1024685301100088, "grad_norm": 1.4606678485870361, "learning_rate": 1.9726044375419172e-05, "loss": 0.17467039823532104, "step": 3890 }, { "epoch": 0.10273194535450753, "grad_norm": 4.260433197021484, "learning_rate": 1.9724057579012026e-05, "loss": 0.17408515214920045, "step": 3900 }, { "epoch": 0.10299536059900627, "grad_norm": 2.5575196743011475, "learning_rate": 1.972206370501624e-05, "loss": 0.17659531831741332, "step": 3910 }, { "epoch": 0.103258775843505, "grad_norm": 3.320159435272217, "learning_rate": 1.9720062754883048e-05, "loss": 0.17536463737487792, "step": 3920 }, { "epoch": 0.10352219108800374, "grad_norm": 1.5681967735290527, "learning_rate": 1.971805473006882e-05, "loss": 0.16720738410949706, "step": 3930 }, { "epoch": 0.10378560633250247, "grad_norm": 3.3278982639312744, "learning_rate": 1.9716039632035086e-05, "loss": 0.16129713058471679, "step": 3940 }, { "epoch": 0.10404902157700122, "grad_norm": 2.271549940109253, "learning_rate": 1.9714017462248523e-05, "loss": 0.19769939184188842, "step": 3950 }, { "epoch": 0.10431243682149995, "grad_norm": 3.0460426807403564, "learning_rate": 1.9711988222180958e-05, "loss": 0.18344171047210694, "step": 3960 }, { "epoch": 0.10457585206599869, "grad_norm": 4.281373977661133, "learning_rate": 1.9709951913309347e-05, "loss": 0.18256943225860595, "step": 3970 }, { "epoch": 0.10483926731049743, "grad_norm": 2.2282028198242188, "learning_rate": 1.9707908537115816e-05, "loss": 0.1720556855201721, "step": 3980 }, { "epoch": 0.10510268255499616, "grad_norm": 2.8792648315429688, "learning_rate": 1.9705858095087613e-05, "loss": 0.1791140079498291, "step": 3990 }, { "epoch": 0.1053660977994949, "grad_norm": 2.722794532775879, "learning_rate": 1.9703800588717144e-05, "loss": 0.1623035788536072, "step": 4000 }, { "epoch": 0.10562951304399364, "grad_norm": 2.554260730743408, "learning_rate": 1.9701736019501948e-05, "loss": 0.1821276903152466, "step": 4010 }, { "epoch": 0.10589292828849238, "grad_norm": 4.012267589569092, "learning_rate": 1.969966438894471e-05, "loss": 0.1675316333770752, "step": 4020 }, { "epoch": 0.10615634353299111, "grad_norm": 2.3353111743927, "learning_rate": 1.969758569855325e-05, "loss": 0.1742751955986023, "step": 4030 }, { "epoch": 0.10641975877748985, "grad_norm": 1.2253715991973877, "learning_rate": 1.969549994984053e-05, "loss": 0.17779165506362915, "step": 4040 }, { "epoch": 0.10668317402198858, "grad_norm": 4.985834121704102, "learning_rate": 1.9693407144324642e-05, "loss": 0.15298010110855104, "step": 4050 }, { "epoch": 0.10694658926648733, "grad_norm": 3.8436806201934814, "learning_rate": 1.9691307283528834e-05, "loss": 0.17984732389450073, "step": 4060 }, { "epoch": 0.10721000451098606, "grad_norm": 2.7200019359588623, "learning_rate": 1.968920036898146e-05, "loss": 0.17747189998626708, "step": 4070 }, { "epoch": 0.1074734197554848, "grad_norm": 2.534663200378418, "learning_rate": 1.9687086402216033e-05, "loss": 0.16457457542419435, "step": 4080 }, { "epoch": 0.10773683499998353, "grad_norm": 2.1377432346343994, "learning_rate": 1.968496538477118e-05, "loss": 0.1649470567703247, "step": 4090 }, { "epoch": 0.10800025024448227, "grad_norm": 3.9717965126037598, "learning_rate": 1.9682837318190683e-05, "loss": 0.18999936580657958, "step": 4100 }, { "epoch": 0.10826366548898102, "grad_norm": 2.420260190963745, "learning_rate": 1.968070220402343e-05, "loss": 0.15235259532928466, "step": 4110 }, { "epoch": 0.10852708073347975, "grad_norm": 1.35859215259552, "learning_rate": 1.967856004382345e-05, "loss": 0.17624285221099853, "step": 4120 }, { "epoch": 0.10879049597797849, "grad_norm": 5.045132637023926, "learning_rate": 1.96764108391499e-05, "loss": 0.1698990821838379, "step": 4130 }, { "epoch": 0.10905391122247722, "grad_norm": 5.997868061065674, "learning_rate": 1.967425459156707e-05, "loss": 0.1762904167175293, "step": 4140 }, { "epoch": 0.10931732646697596, "grad_norm": 2.0113329887390137, "learning_rate": 1.9672091302644363e-05, "loss": 0.17373712062835694, "step": 4150 }, { "epoch": 0.1095807417114747, "grad_norm": 6.490437984466553, "learning_rate": 1.966992097395632e-05, "loss": 0.1732953429222107, "step": 4160 }, { "epoch": 0.10984415695597344, "grad_norm": 1.5028724670410156, "learning_rate": 1.9667743607082593e-05, "loss": 0.17089614868164063, "step": 4170 }, { "epoch": 0.11010757220047217, "grad_norm": 3.0163753032684326, "learning_rate": 1.9665559203607965e-05, "loss": 0.16535618305206298, "step": 4180 }, { "epoch": 0.11037098744497091, "grad_norm": 2.3037948608398438, "learning_rate": 1.9663367765122346e-05, "loss": 0.1684015393257141, "step": 4190 }, { "epoch": 0.11063440268946964, "grad_norm": 1.2293727397918701, "learning_rate": 1.966116929322075e-05, "loss": 0.17464110851287842, "step": 4200 }, { "epoch": 0.11089781793396838, "grad_norm": 2.5168912410736084, "learning_rate": 1.9658963789503328e-05, "loss": 0.1625833511352539, "step": 4210 }, { "epoch": 0.11116123317846711, "grad_norm": 1.6633086204528809, "learning_rate": 1.9656751255575336e-05, "loss": 0.15286695957183838, "step": 4220 }, { "epoch": 0.11142464842296586, "grad_norm": 4.454507827758789, "learning_rate": 1.965453169304715e-05, "loss": 0.1728376865386963, "step": 4230 }, { "epoch": 0.1116880636674646, "grad_norm": 3.897305727005005, "learning_rate": 1.9652305103534265e-05, "loss": 0.17852697372436524, "step": 4240 }, { "epoch": 0.11195147891196333, "grad_norm": 3.1083319187164307, "learning_rate": 1.9650071488657293e-05, "loss": 0.1622702360153198, "step": 4250 }, { "epoch": 0.11221489415646207, "grad_norm": 2.357553243637085, "learning_rate": 1.964783085004195e-05, "loss": 0.168349027633667, "step": 4260 }, { "epoch": 0.1124783094009608, "grad_norm": 2.401501178741455, "learning_rate": 1.9645583189319074e-05, "loss": 0.17769347429275512, "step": 4270 }, { "epoch": 0.11274172464545955, "grad_norm": 5.036011695861816, "learning_rate": 1.9643328508124605e-05, "loss": 0.17642536163330078, "step": 4280 }, { "epoch": 0.11300513988995828, "grad_norm": 3.490665912628174, "learning_rate": 1.9641066808099597e-05, "loss": 0.17678734064102172, "step": 4290 }, { "epoch": 0.11326855513445702, "grad_norm": 1.738020658493042, "learning_rate": 1.963879809089022e-05, "loss": 0.15592508316040038, "step": 4300 }, { "epoch": 0.11353197037895575, "grad_norm": 3.9844837188720703, "learning_rate": 1.9636522358147737e-05, "loss": 0.17508833408355712, "step": 4310 }, { "epoch": 0.1137953856234545, "grad_norm": 3.6114442348480225, "learning_rate": 1.9634239611528527e-05, "loss": 0.17240018844604493, "step": 4320 }, { "epoch": 0.11405880086795323, "grad_norm": 2.6976778507232666, "learning_rate": 1.9631949852694073e-05, "loss": 0.17911844253540038, "step": 4330 }, { "epoch": 0.11432221611245197, "grad_norm": 2.9453542232513428, "learning_rate": 1.9629653083310956e-05, "loss": 0.18035888671875, "step": 4340 }, { "epoch": 0.1145856313569507, "grad_norm": 2.8634798526763916, "learning_rate": 1.9627349305050867e-05, "loss": 0.17954376935958863, "step": 4350 }, { "epoch": 0.11484904660144944, "grad_norm": 3.292043924331665, "learning_rate": 1.9625038519590596e-05, "loss": 0.17503640651702881, "step": 4360 }, { "epoch": 0.11511246184594819, "grad_norm": 6.6897664070129395, "learning_rate": 1.962272072861203e-05, "loss": 0.17956483364105225, "step": 4370 }, { "epoch": 0.11537587709044692, "grad_norm": 3.6474010944366455, "learning_rate": 1.9620395933802158e-05, "loss": 0.18612439632415773, "step": 4380 }, { "epoch": 0.11563929233494566, "grad_norm": 1.4399446249008179, "learning_rate": 1.9618064136853066e-05, "loss": 0.159867525100708, "step": 4390 }, { "epoch": 0.11590270757944439, "grad_norm": 2.081092596054077, "learning_rate": 1.9615725339461936e-05, "loss": 0.17567360401153564, "step": 4400 }, { "epoch": 0.11616612282394313, "grad_norm": 4.536656379699707, "learning_rate": 1.961337954333105e-05, "loss": 0.1650662899017334, "step": 4410 }, { "epoch": 0.11642953806844186, "grad_norm": 2.0923447608947754, "learning_rate": 1.961102675016777e-05, "loss": 0.17326884269714354, "step": 4420 }, { "epoch": 0.1166929533129406, "grad_norm": 3.9211032390594482, "learning_rate": 1.960866696168457e-05, "loss": 0.15869226455688476, "step": 4430 }, { "epoch": 0.11695636855743934, "grad_norm": 4.674497127532959, "learning_rate": 1.9606300179599004e-05, "loss": 0.16153194904327392, "step": 4440 }, { "epoch": 0.11721978380193808, "grad_norm": 2.219001531600952, "learning_rate": 1.9603926405633712e-05, "loss": 0.17727296352386473, "step": 4450 }, { "epoch": 0.11748319904643681, "grad_norm": 1.7324433326721191, "learning_rate": 1.9601545641516434e-05, "loss": 0.17860946655273438, "step": 4460 }, { "epoch": 0.11774661429093555, "grad_norm": 4.262462139129639, "learning_rate": 1.9599157888979997e-05, "loss": 0.17437138557434081, "step": 4470 }, { "epoch": 0.11801002953543428, "grad_norm": 1.9242116212844849, "learning_rate": 1.95967631497623e-05, "loss": 0.1677917242050171, "step": 4480 }, { "epoch": 0.11827344477993303, "grad_norm": 2.5671064853668213, "learning_rate": 1.9594361425606348e-05, "loss": 0.17460231781005858, "step": 4490 }, { "epoch": 0.11853686002443177, "grad_norm": 1.978698492050171, "learning_rate": 1.959195271826021e-05, "loss": 0.16342027187347413, "step": 4500 }, { "epoch": 0.1188002752689305, "grad_norm": 3.306943416595459, "learning_rate": 1.9589537029477052e-05, "loss": 0.16117432117462158, "step": 4510 }, { "epoch": 0.11906369051342924, "grad_norm": 3.427915334701538, "learning_rate": 1.9587114361015118e-05, "loss": 0.1700894355773926, "step": 4520 }, { "epoch": 0.11932710575792797, "grad_norm": 2.676492214202881, "learning_rate": 1.958468471463773e-05, "loss": 0.17243685722351074, "step": 4530 }, { "epoch": 0.11959052100242672, "grad_norm": 2.8428094387054443, "learning_rate": 1.9582248092113286e-05, "loss": 0.17552597522735597, "step": 4540 }, { "epoch": 0.11985393624692545, "grad_norm": 2.129549503326416, "learning_rate": 1.9579804495215268e-05, "loss": 0.1635894775390625, "step": 4550 }, { "epoch": 0.12011735149142419, "grad_norm": 2.0078234672546387, "learning_rate": 1.9577353925722228e-05, "loss": 0.17493181228637694, "step": 4560 }, { "epoch": 0.12038076673592292, "grad_norm": 1.8076927661895752, "learning_rate": 1.9574896385417803e-05, "loss": 0.1793774127960205, "step": 4570 }, { "epoch": 0.12064418198042166, "grad_norm": 1.8266024589538574, "learning_rate": 1.957243187609069e-05, "loss": 0.17072606086730957, "step": 4580 }, { "epoch": 0.1209075972249204, "grad_norm": 1.6489548683166504, "learning_rate": 1.9569960399534668e-05, "loss": 0.16950554847717286, "step": 4590 }, { "epoch": 0.12117101246941914, "grad_norm": 3.5417325496673584, "learning_rate": 1.9567481957548585e-05, "loss": 0.1801328182220459, "step": 4600 }, { "epoch": 0.12143442771391787, "grad_norm": 3.2218539714813232, "learning_rate": 1.9564996551936354e-05, "loss": 0.17555596828460693, "step": 4610 }, { "epoch": 0.12169784295841661, "grad_norm": 3.0498197078704834, "learning_rate": 1.9562504184506966e-05, "loss": 0.16508339643478392, "step": 4620 }, { "epoch": 0.12196125820291535, "grad_norm": 3.59855318069458, "learning_rate": 1.9560004857074472e-05, "loss": 0.1681840419769287, "step": 4630 }, { "epoch": 0.12222467344741408, "grad_norm": 2.394068479537964, "learning_rate": 1.955749857145799e-05, "loss": 0.1695917010307312, "step": 4640 }, { "epoch": 0.12248808869191283, "grad_norm": 3.217271566390991, "learning_rate": 1.95549853294817e-05, "loss": 0.16670637130737304, "step": 4650 }, { "epoch": 0.12275150393641156, "grad_norm": 1.3357664346694946, "learning_rate": 1.955246513297485e-05, "loss": 0.1823087215423584, "step": 4660 }, { "epoch": 0.1230149191809103, "grad_norm": 1.7085297107696533, "learning_rate": 1.9549937983771748e-05, "loss": 0.18020365238189698, "step": 4670 }, { "epoch": 0.12327833442540903, "grad_norm": 6.612557411193848, "learning_rate": 1.954740388371176e-05, "loss": 0.1611422300338745, "step": 4680 }, { "epoch": 0.12354174966990777, "grad_norm": 1.551203727722168, "learning_rate": 1.9544862834639316e-05, "loss": 0.16571117639541627, "step": 4690 }, { "epoch": 0.1238051649144065, "grad_norm": 1.747071385383606, "learning_rate": 1.9542314838403907e-05, "loss": 0.16095192432403566, "step": 4700 }, { "epoch": 0.12406858015890525, "grad_norm": 3.1804375648498535, "learning_rate": 1.9539759896860064e-05, "loss": 0.1746140480041504, "step": 4710 }, { "epoch": 0.12433199540340398, "grad_norm": 3.9847934246063232, "learning_rate": 1.9537198011867383e-05, "loss": 0.17563059329986572, "step": 4720 }, { "epoch": 0.12459541064790272, "grad_norm": 3.9221537113189697, "learning_rate": 1.9534629185290524e-05, "loss": 0.18369977474212645, "step": 4730 }, { "epoch": 0.12485882589240146, "grad_norm": 2.302168846130371, "learning_rate": 1.9532053418999186e-05, "loss": 0.1805165648460388, "step": 4740 }, { "epoch": 0.1251222411369002, "grad_norm": 3.525416374206543, "learning_rate": 1.952947071486812e-05, "loss": 0.18940200805664062, "step": 4750 }, { "epoch": 0.12538565638139892, "grad_norm": 3.078427791595459, "learning_rate": 1.952688107477713e-05, "loss": 0.1639944791793823, "step": 4760 }, { "epoch": 0.12564907162589767, "grad_norm": 1.7100762128829956, "learning_rate": 1.9524284500611076e-05, "loss": 0.18245161771774293, "step": 4770 }, { "epoch": 0.1259124868703964, "grad_norm": 24.769611358642578, "learning_rate": 1.9521680994259845e-05, "loss": 0.1736838102340698, "step": 4780 }, { "epoch": 0.12617590211489516, "grad_norm": 2.3905069828033447, "learning_rate": 1.9519070557618386e-05, "loss": 3.7142181396484375, "step": 4790 }, { "epoch": 0.12643931735939387, "grad_norm": 2.693229913711548, "learning_rate": 1.9516453192586686e-05, "loss": 0.22038984298706055, "step": 4800 }, { "epoch": 0.12670273260389261, "grad_norm": 4.328276634216309, "learning_rate": 1.951382890106978e-05, "loss": 0.19407424926757813, "step": 4810 }, { "epoch": 0.12696614784839136, "grad_norm": 2.713364839553833, "learning_rate": 1.951119768497774e-05, "loss": 0.20595960617065429, "step": 4820 }, { "epoch": 0.1272295630928901, "grad_norm": 3.1007537841796875, "learning_rate": 1.9508559546225672e-05, "loss": 0.1867194652557373, "step": 4830 }, { "epoch": 0.12749297833738882, "grad_norm": 2.1146600246429443, "learning_rate": 1.9505914486733735e-05, "loss": 0.182800030708313, "step": 4840 }, { "epoch": 0.12775639358188756, "grad_norm": 1.6339370012283325, "learning_rate": 1.9503262508427115e-05, "loss": 0.20050432682037353, "step": 4850 }, { "epoch": 0.1280198088263863, "grad_norm": 3.0191924571990967, "learning_rate": 1.9500603613236034e-05, "loss": 0.17831366062164306, "step": 4860 }, { "epoch": 0.12828322407088505, "grad_norm": 2.8091824054718018, "learning_rate": 1.9497937803095756e-05, "loss": 0.19510102272033691, "step": 4870 }, { "epoch": 0.1285466393153838, "grad_norm": 4.196601390838623, "learning_rate": 1.9495265079946565e-05, "loss": 0.18831653594970704, "step": 4880 }, { "epoch": 0.1288100545598825, "grad_norm": 4.394210338592529, "learning_rate": 1.9492585445733788e-05, "loss": 0.20420005321502685, "step": 4890 }, { "epoch": 0.12907346980438125, "grad_norm": 3.489626407623291, "learning_rate": 1.9489898902407783e-05, "loss": 0.18862123489379884, "step": 4900 }, { "epoch": 0.12933688504888, "grad_norm": 4.735525131225586, "learning_rate": 1.9487205451923925e-05, "loss": 0.18216466903686523, "step": 4910 }, { "epoch": 0.12960030029337874, "grad_norm": 6.517829418182373, "learning_rate": 1.9484505096242628e-05, "loss": 0.19044101238250732, "step": 4920 }, { "epoch": 0.12986371553787746, "grad_norm": 5.790692329406738, "learning_rate": 1.9481797837329323e-05, "loss": 0.183174729347229, "step": 4930 }, { "epoch": 0.1301271307823762, "grad_norm": 1.79916512966156, "learning_rate": 1.9479083677154474e-05, "loss": 0.17323999404907225, "step": 4940 }, { "epoch": 0.13039054602687494, "grad_norm": 1.3160834312438965, "learning_rate": 1.9476362617693564e-05, "loss": 0.1815234899520874, "step": 4950 }, { "epoch": 0.1306539612713737, "grad_norm": 1.67411470413208, "learning_rate": 1.9473634660927092e-05, "loss": 0.18733208179473876, "step": 4960 }, { "epoch": 0.1309173765158724, "grad_norm": 5.037570953369141, "learning_rate": 1.9470899808840585e-05, "loss": 0.1846956729888916, "step": 4970 }, { "epoch": 0.13118079176037115, "grad_norm": 1.8619064092636108, "learning_rate": 1.9468158063424593e-05, "loss": 0.18975462913513183, "step": 4980 }, { "epoch": 0.1314442070048699, "grad_norm": 3.225700616836548, "learning_rate": 1.946540942667467e-05, "loss": 0.16946625709533691, "step": 4990 }, { "epoch": 0.13170762224936863, "grad_norm": 1.1881866455078125, "learning_rate": 1.9462653900591393e-05, "loss": 0.173594868183136, "step": 5000 }, { "epoch": 0.13197103749386738, "grad_norm": 2.857513904571533, "learning_rate": 1.9459891487180353e-05, "loss": 0.17550170421600342, "step": 5010 }, { "epoch": 0.1322344527383661, "grad_norm": 2.1959967613220215, "learning_rate": 1.9457122188452156e-05, "loss": 0.16599507331848146, "step": 5020 }, { "epoch": 0.13249786798286484, "grad_norm": 3.824434518814087, "learning_rate": 1.9454346006422414e-05, "loss": 0.17282904386520387, "step": 5030 }, { "epoch": 0.13276128322736358, "grad_norm": 1.980412244796753, "learning_rate": 1.9451562943111753e-05, "loss": 0.18114445209503174, "step": 5040 }, { "epoch": 0.13302469847186232, "grad_norm": 2.4351959228515625, "learning_rate": 1.944877300054581e-05, "loss": 0.18607349395751954, "step": 5050 }, { "epoch": 0.13328811371636104, "grad_norm": 2.467095136642456, "learning_rate": 1.944597618075522e-05, "loss": 0.16894615888595582, "step": 5060 }, { "epoch": 0.13355152896085978, "grad_norm": 2.6105570793151855, "learning_rate": 1.944317248577563e-05, "loss": 0.16367750167846679, "step": 5070 }, { "epoch": 0.13381494420535853, "grad_norm": 2.1722967624664307, "learning_rate": 1.9440361917647693e-05, "loss": 0.18050326108932496, "step": 5080 }, { "epoch": 0.13407835944985727, "grad_norm": 1.4496557712554932, "learning_rate": 1.943754447841706e-05, "loss": 0.1651815414428711, "step": 5090 }, { "epoch": 0.134341774694356, "grad_norm": 3.2407619953155518, "learning_rate": 1.9434720170134387e-05, "loss": 0.16921218633651733, "step": 5100 }, { "epoch": 0.13460518993885473, "grad_norm": 3.3515398502349854, "learning_rate": 1.9431888994855324e-05, "loss": 0.18301103115081788, "step": 5110 }, { "epoch": 0.13486860518335347, "grad_norm": 5.332551956176758, "learning_rate": 1.9429050954640528e-05, "loss": 0.17564899921417237, "step": 5120 }, { "epoch": 0.13513202042785222, "grad_norm": 2.9540841579437256, "learning_rate": 1.942620605155564e-05, "loss": 0.17125153541564941, "step": 5130 }, { "epoch": 0.13539543567235096, "grad_norm": 1.8847217559814453, "learning_rate": 1.9423354287671307e-05, "loss": 0.1797324538230896, "step": 5140 }, { "epoch": 0.13565885091684968, "grad_norm": 1.8396559953689575, "learning_rate": 1.9420495665063166e-05, "loss": 0.17330050468444824, "step": 5150 }, { "epoch": 0.13592226616134842, "grad_norm": 1.524906873703003, "learning_rate": 1.9417630185811846e-05, "loss": 0.16882662773132323, "step": 5160 }, { "epoch": 0.13618568140584716, "grad_norm": 2.5440614223480225, "learning_rate": 1.9414757852002962e-05, "loss": 0.18318302631378175, "step": 5170 }, { "epoch": 0.1364490966503459, "grad_norm": 2.817826986312866, "learning_rate": 1.941187866572713e-05, "loss": 0.1678206205368042, "step": 5180 }, { "epoch": 0.13671251189484462, "grad_norm": 1.4059306383132935, "learning_rate": 1.940899262907994e-05, "loss": 0.17017884254455568, "step": 5190 }, { "epoch": 0.13697592713934337, "grad_norm": 2.7491772174835205, "learning_rate": 1.940609974416198e-05, "loss": 0.17544410228729249, "step": 5200 }, { "epoch": 0.1372393423838421, "grad_norm": 8.5287504196167, "learning_rate": 1.940320001307881e-05, "loss": 0.17877840995788574, "step": 5210 }, { "epoch": 0.13750275762834085, "grad_norm": 1.8168407678604126, "learning_rate": 1.940029343794098e-05, "loss": 0.17951881885528564, "step": 5220 }, { "epoch": 0.13776617287283957, "grad_norm": 1.3033761978149414, "learning_rate": 1.9397380020864024e-05, "loss": 0.18127357959747314, "step": 5230 }, { "epoch": 0.13802958811733831, "grad_norm": 2.273768663406372, "learning_rate": 1.939445976396845e-05, "loss": 0.17169294357299805, "step": 5240 }, { "epoch": 0.13829300336183706, "grad_norm": 1.6593101024627686, "learning_rate": 1.939153266937975e-05, "loss": 0.17472355365753173, "step": 5250 }, { "epoch": 0.1385564186063358, "grad_norm": 2.837193727493286, "learning_rate": 1.938859873922839e-05, "loss": 0.18481550216674805, "step": 5260 }, { "epoch": 0.13881983385083455, "grad_norm": 2.3597686290740967, "learning_rate": 1.9385657975649807e-05, "loss": 0.17701208591461182, "step": 5270 }, { "epoch": 0.13908324909533326, "grad_norm": 1.5679123401641846, "learning_rate": 1.9382710380784418e-05, "loss": 0.16644446849822997, "step": 5280 }, { "epoch": 0.139346664339832, "grad_norm": 2.433248519897461, "learning_rate": 1.937975595677761e-05, "loss": 0.1641322135925293, "step": 5290 }, { "epoch": 0.13961007958433075, "grad_norm": 2.8176355361938477, "learning_rate": 1.9376794705779735e-05, "loss": 0.16657586097717286, "step": 5300 }, { "epoch": 0.1398734948288295, "grad_norm": 2.1801562309265137, "learning_rate": 1.937382662994613e-05, "loss": 0.17274742126464843, "step": 5310 }, { "epoch": 0.1401369100733282, "grad_norm": 4.081486701965332, "learning_rate": 1.9370851731437077e-05, "loss": 0.17577544450759888, "step": 5320 }, { "epoch": 0.14040032531782695, "grad_norm": 4.06489896774292, "learning_rate": 1.9367870012417844e-05, "loss": 0.18243703842163086, "step": 5330 }, { "epoch": 0.1406637405623257, "grad_norm": 2.8595187664031982, "learning_rate": 1.936488147505865e-05, "loss": 0.17486560344696045, "step": 5340 }, { "epoch": 0.14092715580682444, "grad_norm": 2.256521224975586, "learning_rate": 1.9361886121534686e-05, "loss": 0.16600306034088136, "step": 5350 }, { "epoch": 0.14119057105132315, "grad_norm": 1.9577568769454956, "learning_rate": 1.93588839540261e-05, "loss": 0.17151427268981934, "step": 5360 }, { "epoch": 0.1414539862958219, "grad_norm": 3.149722099304199, "learning_rate": 1.9355874974717988e-05, "loss": 0.16600780487060546, "step": 5370 }, { "epoch": 0.14171740154032064, "grad_norm": 1.9859907627105713, "learning_rate": 1.9352859185800424e-05, "loss": 0.15928300619125366, "step": 5380 }, { "epoch": 0.14198081678481939, "grad_norm": 2.565993309020996, "learning_rate": 1.934983658946843e-05, "loss": 0.17912847995758058, "step": 5390 }, { "epoch": 0.14224423202931813, "grad_norm": 2.4322712421417236, "learning_rate": 1.9346807187921983e-05, "loss": 0.17009468078613282, "step": 5400 }, { "epoch": 0.14250764727381685, "grad_norm": 5.3307719230651855, "learning_rate": 1.9343770983366004e-05, "loss": 0.1629367470741272, "step": 5410 }, { "epoch": 0.1427710625183156, "grad_norm": 3.3527145385742188, "learning_rate": 1.9340727978010377e-05, "loss": 0.1758910059928894, "step": 5420 }, { "epoch": 0.14303447776281433, "grad_norm": 2.937940835952759, "learning_rate": 1.933767817406994e-05, "loss": 0.1715101957321167, "step": 5430 }, { "epoch": 0.14329789300731308, "grad_norm": 2.9165844917297363, "learning_rate": 1.9334621573764462e-05, "loss": 0.16880466938018798, "step": 5440 }, { "epoch": 0.1435613082518118, "grad_norm": 1.2169981002807617, "learning_rate": 1.9331558179318676e-05, "loss": 0.17317895889282225, "step": 5450 }, { "epoch": 0.14382472349631054, "grad_norm": 2.834941864013672, "learning_rate": 1.9328487992962247e-05, "loss": 0.17429842948913574, "step": 5460 }, { "epoch": 0.14408813874080928, "grad_norm": 2.6817190647125244, "learning_rate": 1.9325411016929793e-05, "loss": 0.17860908508300782, "step": 5470 }, { "epoch": 0.14435155398530802, "grad_norm": 1.8139413595199585, "learning_rate": 1.9322327253460867e-05, "loss": 0.16991548538208007, "step": 5480 }, { "epoch": 0.14461496922980674, "grad_norm": 2.0931267738342285, "learning_rate": 1.931923670479997e-05, "loss": 0.1658308744430542, "step": 5490 }, { "epoch": 0.14487838447430548, "grad_norm": 1.275233268737793, "learning_rate": 1.9316139373196533e-05, "loss": 0.17527616024017334, "step": 5500 }, { "epoch": 0.14514179971880423, "grad_norm": 1.7908324003219604, "learning_rate": 1.9313035260904923e-05, "loss": 0.1747829556465149, "step": 5510 }, { "epoch": 0.14540521496330297, "grad_norm": 2.631655216217041, "learning_rate": 1.9309924370184455e-05, "loss": 0.17286276817321777, "step": 5520 }, { "epoch": 0.1456686302078017, "grad_norm": 4.172806262969971, "learning_rate": 1.9306806703299367e-05, "loss": 0.17544846534729003, "step": 5530 }, { "epoch": 0.14593204545230043, "grad_norm": 2.252880334854126, "learning_rate": 1.9303682262518828e-05, "loss": 0.1822710871696472, "step": 5540 }, { "epoch": 0.14619546069679917, "grad_norm": 1.667072057723999, "learning_rate": 1.9300551050116946e-05, "loss": 0.17063034772872926, "step": 5550 }, { "epoch": 0.14645887594129792, "grad_norm": 2.6356050968170166, "learning_rate": 1.9297413068372745e-05, "loss": 0.16601455211639404, "step": 5560 }, { "epoch": 0.14672229118579666, "grad_norm": 4.675506114959717, "learning_rate": 1.9294268319570195e-05, "loss": 0.16532421112060547, "step": 5570 }, { "epoch": 0.14698570643029538, "grad_norm": 2.181631088256836, "learning_rate": 1.9291116805998168e-05, "loss": 0.18210327625274658, "step": 5580 }, { "epoch": 0.14724912167479412, "grad_norm": 1.1931922435760498, "learning_rate": 1.9287958529950477e-05, "loss": 0.17008395195007325, "step": 5590 }, { "epoch": 0.14751253691929286, "grad_norm": 2.6179239749908447, "learning_rate": 1.9284793493725852e-05, "loss": 0.1734899878501892, "step": 5600 }, { "epoch": 0.1477759521637916, "grad_norm": 2.259885549545288, "learning_rate": 1.9281621699627943e-05, "loss": 0.17243847846984864, "step": 5610 }, { "epoch": 0.14803936740829032, "grad_norm": 1.1944204568862915, "learning_rate": 1.927844314996532e-05, "loss": 0.1536738634109497, "step": 5620 }, { "epoch": 0.14830278265278907, "grad_norm": 3.260671854019165, "learning_rate": 1.9275257847051464e-05, "loss": 0.16193888187408448, "step": 5630 }, { "epoch": 0.1485661978972878, "grad_norm": 4.402671813964844, "learning_rate": 1.9272065793204778e-05, "loss": 0.16629443168640137, "step": 5640 }, { "epoch": 0.14882961314178655, "grad_norm": 5.667645454406738, "learning_rate": 1.9268866990748578e-05, "loss": 0.16817145347595214, "step": 5650 }, { "epoch": 0.1490930283862853, "grad_norm": 2.1732397079467773, "learning_rate": 1.9265661442011087e-05, "loss": 0.18714728355407714, "step": 5660 }, { "epoch": 0.149356443630784, "grad_norm": 4.236209392547607, "learning_rate": 1.9262449149325445e-05, "loss": 0.18157315254211426, "step": 5670 }, { "epoch": 0.14961985887528276, "grad_norm": 1.7208629846572876, "learning_rate": 1.92592301150297e-05, "loss": 0.16355228424072266, "step": 5680 }, { "epoch": 0.1498832741197815, "grad_norm": 2.0241734981536865, "learning_rate": 1.9256004341466795e-05, "loss": 0.1583286166191101, "step": 5690 }, { "epoch": 0.15014668936428024, "grad_norm": 2.6650590896606445, "learning_rate": 1.9252771830984594e-05, "loss": 0.15145057439804077, "step": 5700 }, { "epoch": 0.15041010460877896, "grad_norm": 6.8562846183776855, "learning_rate": 1.9249532585935855e-05, "loss": 0.169380521774292, "step": 5710 }, { "epoch": 0.1506735198532777, "grad_norm": 4.9568915367126465, "learning_rate": 1.9246286608678243e-05, "loss": 0.16459800004959108, "step": 5720 }, { "epoch": 0.15093693509777645, "grad_norm": 3.0124645233154297, "learning_rate": 1.9243033901574316e-05, "loss": 0.17216343879699708, "step": 5730 }, { "epoch": 0.1512003503422752, "grad_norm": 3.7983531951904297, "learning_rate": 1.923977446699154e-05, "loss": 0.17335137128829955, "step": 5740 }, { "epoch": 0.1514637655867739, "grad_norm": 5.465087890625, "learning_rate": 1.923650830730227e-05, "loss": 0.16456804275512696, "step": 5750 }, { "epoch": 0.15172718083127265, "grad_norm": 2.9005091190338135, "learning_rate": 1.923323542488375e-05, "loss": 0.16241971254348755, "step": 5760 }, { "epoch": 0.1519905960757714, "grad_norm": 1.2566637992858887, "learning_rate": 1.9229955822118138e-05, "loss": 0.16815674304962158, "step": 5770 }, { "epoch": 0.15225401132027014, "grad_norm": 2.0080513954162598, "learning_rate": 1.9226669501392465e-05, "loss": 0.1646472692489624, "step": 5780 }, { "epoch": 0.15251742656476888, "grad_norm": 2.2740871906280518, "learning_rate": 1.9223376465098657e-05, "loss": 0.16322988271713257, "step": 5790 }, { "epoch": 0.1527808418092676, "grad_norm": 7.668686866760254, "learning_rate": 1.9220076715633526e-05, "loss": 0.1678305983543396, "step": 5800 }, { "epoch": 0.15304425705376634, "grad_norm": 3.7402803897857666, "learning_rate": 1.921677025539877e-05, "loss": 0.16645054817199706, "step": 5810 }, { "epoch": 0.15330767229826509, "grad_norm": 1.405545711517334, "learning_rate": 1.921345708680098e-05, "loss": 0.16109809875488282, "step": 5820 }, { "epoch": 0.15357108754276383, "grad_norm": 2.0498173236846924, "learning_rate": 1.921013721225162e-05, "loss": 0.16458098888397216, "step": 5830 }, { "epoch": 0.15383450278726254, "grad_norm": 3.3330576419830322, "learning_rate": 1.9206810634167036e-05, "loss": 0.1527179002761841, "step": 5840 }, { "epoch": 0.1540979180317613, "grad_norm": 2.0103061199188232, "learning_rate": 1.920347735496846e-05, "loss": 0.16349775791168214, "step": 5850 }, { "epoch": 0.15436133327626003, "grad_norm": 1.5941537618637085, "learning_rate": 1.9200137377081993e-05, "loss": 0.1653532028198242, "step": 5860 }, { "epoch": 0.15462474852075878, "grad_norm": 2.1088438034057617, "learning_rate": 1.919679070293861e-05, "loss": 0.16074947118759156, "step": 5870 }, { "epoch": 0.1548881637652575, "grad_norm": 4.1618804931640625, "learning_rate": 1.9193437334974178e-05, "loss": 0.15755188465118408, "step": 5880 }, { "epoch": 0.15515157900975624, "grad_norm": 4.56793212890625, "learning_rate": 1.9190077275629408e-05, "loss": 0.16214089393615722, "step": 5890 }, { "epoch": 0.15541499425425498, "grad_norm": 1.7585976123809814, "learning_rate": 1.9186710527349907e-05, "loss": 0.1655438780784607, "step": 5900 }, { "epoch": 0.15567840949875372, "grad_norm": 3.724648952484131, "learning_rate": 1.9183337092586137e-05, "loss": 0.1679638147354126, "step": 5910 }, { "epoch": 0.15594182474325247, "grad_norm": 1.7517844438552856, "learning_rate": 1.9179956973793422e-05, "loss": 0.16804733276367187, "step": 5920 }, { "epoch": 0.15620523998775118, "grad_norm": 1.7509723901748657, "learning_rate": 1.917657017343197e-05, "loss": 0.1711927533149719, "step": 5930 }, { "epoch": 0.15646865523224993, "grad_norm": 2.4012229442596436, "learning_rate": 1.9173176693966835e-05, "loss": 0.17096556425094606, "step": 5940 }, { "epoch": 0.15673207047674867, "grad_norm": 1.599761962890625, "learning_rate": 1.9169776537867936e-05, "loss": 0.16796581745147704, "step": 5950 }, { "epoch": 0.1569954857212474, "grad_norm": 2.025940418243408, "learning_rate": 1.9166369707610055e-05, "loss": 0.15596766471862794, "step": 5960 }, { "epoch": 0.15725890096574613, "grad_norm": 2.5390448570251465, "learning_rate": 1.9162956205672833e-05, "loss": 0.16899288892745973, "step": 5970 }, { "epoch": 0.15752231621024487, "grad_norm": 2.062568426132202, "learning_rate": 1.9159536034540757e-05, "loss": 0.16742510795593263, "step": 5980 }, { "epoch": 0.15778573145474362, "grad_norm": 2.7181360721588135, "learning_rate": 1.9156109196703177e-05, "loss": 0.17004867792129516, "step": 5990 }, { "epoch": 0.15804914669924236, "grad_norm": 1.5434995889663696, "learning_rate": 1.91526756946543e-05, "loss": 0.1760188579559326, "step": 6000 }, { "epoch": 0.15831256194374108, "grad_norm": 6.026331424713135, "learning_rate": 1.9149235530893172e-05, "loss": 0.17598656415939332, "step": 6010 }, { "epoch": 0.15857597718823982, "grad_norm": 1.5573654174804688, "learning_rate": 1.9145788707923692e-05, "loss": 0.1767571449279785, "step": 6020 }, { "epoch": 0.15883939243273856, "grad_norm": 2.0415823459625244, "learning_rate": 1.914233522825461e-05, "loss": 0.16683809757232665, "step": 6030 }, { "epoch": 0.1591028076772373, "grad_norm": 2.33962082862854, "learning_rate": 1.913887509439951e-05, "loss": 0.1674262285232544, "step": 6040 }, { "epoch": 0.15936622292173605, "grad_norm": 3.491574287414551, "learning_rate": 1.9135408308876834e-05, "loss": 0.17102857828140258, "step": 6050 }, { "epoch": 0.15962963816623477, "grad_norm": 3.8703691959381104, "learning_rate": 1.913193487420986e-05, "loss": 0.16185617446899414, "step": 6060 }, { "epoch": 0.1598930534107335, "grad_norm": 1.6344120502471924, "learning_rate": 1.9128454792926694e-05, "loss": 0.16544573307037352, "step": 6070 }, { "epoch": 0.16015646865523225, "grad_norm": 4.014691352844238, "learning_rate": 1.9124968067560296e-05, "loss": 0.17168040275573732, "step": 6080 }, { "epoch": 0.160419883899731, "grad_norm": 4.4546356201171875, "learning_rate": 1.9121474700648454e-05, "loss": 0.16958110332489013, "step": 6090 }, { "epoch": 0.1606832991442297, "grad_norm": 2.8906989097595215, "learning_rate": 1.9117974694733795e-05, "loss": 0.16886957883834838, "step": 6100 }, { "epoch": 0.16094671438872846, "grad_norm": 1.653565764427185, "learning_rate": 1.9114468052363764e-05, "loss": 0.1717803955078125, "step": 6110 }, { "epoch": 0.1612101296332272, "grad_norm": 1.8023502826690674, "learning_rate": 1.9110954776090656e-05, "loss": 0.17511296272277832, "step": 6120 }, { "epoch": 0.16147354487772594, "grad_norm": 2.449190378189087, "learning_rate": 1.910743486847158e-05, "loss": 0.16130332946777343, "step": 6130 }, { "epoch": 0.16173696012222466, "grad_norm": 2.076946496963501, "learning_rate": 1.9103908332068483e-05, "loss": 0.17460484504699708, "step": 6140 }, { "epoch": 0.1620003753667234, "grad_norm": 3.5501556396484375, "learning_rate": 1.910037516944812e-05, "loss": 0.1576932907104492, "step": 6150 }, { "epoch": 0.16226379061122215, "grad_norm": 3.194321870803833, "learning_rate": 1.909683538318209e-05, "loss": 0.16315851211547852, "step": 6160 }, { "epoch": 0.1625272058557209, "grad_norm": 3.070650577545166, "learning_rate": 1.9093288975846797e-05, "loss": 0.1610170364379883, "step": 6170 }, { "epoch": 0.16279062110021963, "grad_norm": 2.4313089847564697, "learning_rate": 1.9089735950023463e-05, "loss": 0.17812445163726806, "step": 6180 }, { "epoch": 0.16305403634471835, "grad_norm": 1.8644837141036987, "learning_rate": 1.9086176308298147e-05, "loss": 0.1673278331756592, "step": 6190 }, { "epoch": 0.1633174515892171, "grad_norm": 7.395666599273682, "learning_rate": 1.90826100532617e-05, "loss": 0.18234014511108398, "step": 6200 }, { "epoch": 0.16358086683371584, "grad_norm": 4.232606887817383, "learning_rate": 1.90790371875098e-05, "loss": 0.16605186462402344, "step": 6210 }, { "epoch": 0.16384428207821458, "grad_norm": 2.195112705230713, "learning_rate": 1.9075457713642934e-05, "loss": 0.17565006017684937, "step": 6220 }, { "epoch": 0.1641076973227133, "grad_norm": 1.1855543851852417, "learning_rate": 1.9071871634266392e-05, "loss": 0.16857655048370362, "step": 6230 }, { "epoch": 0.16437111256721204, "grad_norm": 1.1885030269622803, "learning_rate": 1.9068278951990282e-05, "loss": 0.16012897491455078, "step": 6240 }, { "epoch": 0.16463452781171078, "grad_norm": 4.977087497711182, "learning_rate": 1.9064679669429515e-05, "loss": 0.16823840141296387, "step": 6250 }, { "epoch": 0.16489794305620953, "grad_norm": 1.736676573753357, "learning_rate": 1.9061073789203798e-05, "loss": 0.16622426509857177, "step": 6260 }, { "epoch": 0.16516135830070824, "grad_norm": 1.8090811967849731, "learning_rate": 1.9057461313937646e-05, "loss": 0.18401665687561036, "step": 6270 }, { "epoch": 0.165424773545207, "grad_norm": 1.9850068092346191, "learning_rate": 1.905384224626038e-05, "loss": 0.17966148853302003, "step": 6280 }, { "epoch": 0.16568818878970573, "grad_norm": 5.202455043792725, "learning_rate": 1.9050216588806113e-05, "loss": 0.1733680009841919, "step": 6290 }, { "epoch": 0.16595160403420448, "grad_norm": 3.423353910446167, "learning_rate": 1.9046584344213754e-05, "loss": 0.15421415567398072, "step": 6300 }, { "epoch": 0.16621501927870322, "grad_norm": 4.508658409118652, "learning_rate": 1.9042945515127004e-05, "loss": 0.1616438865661621, "step": 6310 }, { "epoch": 0.16647843452320193, "grad_norm": 2.083101511001587, "learning_rate": 1.9039300104194366e-05, "loss": 0.16008297204971314, "step": 6320 }, { "epoch": 0.16674184976770068, "grad_norm": 3.094064235687256, "learning_rate": 1.903564811406912e-05, "loss": 0.17567126750946044, "step": 6330 }, { "epoch": 0.16700526501219942, "grad_norm": 1.9155021905899048, "learning_rate": 1.9031989547409347e-05, "loss": 0.15738580226898194, "step": 6340 }, { "epoch": 0.16726868025669817, "grad_norm": 1.683387279510498, "learning_rate": 1.9028324406877905e-05, "loss": 0.15834457874298097, "step": 6350 }, { "epoch": 0.16753209550119688, "grad_norm": 1.9250668287277222, "learning_rate": 1.9024652695142447e-05, "loss": 0.1485285997390747, "step": 6360 }, { "epoch": 0.16779551074569563, "grad_norm": 2.8460636138916016, "learning_rate": 1.90209744148754e-05, "loss": 0.1670464515686035, "step": 6370 }, { "epoch": 0.16805892599019437, "grad_norm": 3.627331495285034, "learning_rate": 1.901728956875398e-05, "loss": 0.1597130537033081, "step": 6380 }, { "epoch": 0.1683223412346931, "grad_norm": 2.9211559295654297, "learning_rate": 1.9013598159460168e-05, "loss": 0.15170284509658813, "step": 6390 }, { "epoch": 0.16858575647919183, "grad_norm": 4.204162120819092, "learning_rate": 1.900990018968074e-05, "loss": 0.1606121063232422, "step": 6400 }, { "epoch": 0.16884917172369057, "grad_norm": 1.4400683641433716, "learning_rate": 1.900619566210723e-05, "loss": 0.1479124903678894, "step": 6410 }, { "epoch": 0.16911258696818932, "grad_norm": 1.1471730470657349, "learning_rate": 1.900248457943596e-05, "loss": 0.15518677234649658, "step": 6420 }, { "epoch": 0.16937600221268806, "grad_norm": 3.837933301925659, "learning_rate": 1.899876694436801e-05, "loss": 0.16649779081344604, "step": 6430 }, { "epoch": 0.1696394174571868, "grad_norm": 1.9743930101394653, "learning_rate": 1.899504275960924e-05, "loss": 0.16189520359039306, "step": 6440 }, { "epoch": 0.16990283270168552, "grad_norm": 4.61109733581543, "learning_rate": 1.899131202787027e-05, "loss": 0.16709916591644286, "step": 6450 }, { "epoch": 0.17016624794618426, "grad_norm": 1.4629868268966675, "learning_rate": 1.8987574751866488e-05, "loss": 0.18981029987335205, "step": 6460 }, { "epoch": 0.170429663190683, "grad_norm": 5.109851837158203, "learning_rate": 1.8983830934318044e-05, "loss": 0.17718074321746827, "step": 6470 }, { "epoch": 0.17069307843518175, "grad_norm": 3.4982848167419434, "learning_rate": 1.8980080577949848e-05, "loss": 0.16982486248016357, "step": 6480 }, { "epoch": 0.17095649367968047, "grad_norm": 4.421207904815674, "learning_rate": 1.8976323685491577e-05, "loss": 0.1697385311126709, "step": 6490 }, { "epoch": 0.1712199089241792, "grad_norm": 2.544130325317383, "learning_rate": 1.8972560259677658e-05, "loss": 0.17274026870727538, "step": 6500 }, { "epoch": 0.17148332416867795, "grad_norm": 6.575688362121582, "learning_rate": 1.896879030324727e-05, "loss": 0.16297225952148436, "step": 6510 }, { "epoch": 0.1717467394131767, "grad_norm": 1.5775753259658813, "learning_rate": 1.8965013818944357e-05, "loss": 0.16951922178268433, "step": 6520 }, { "epoch": 0.1720101546576754, "grad_norm": 1.9414620399475098, "learning_rate": 1.89612308095176e-05, "loss": 0.17020161151885987, "step": 6530 }, { "epoch": 0.17227356990217416, "grad_norm": 1.870597243309021, "learning_rate": 1.8957441277720438e-05, "loss": 0.1716052770614624, "step": 6540 }, { "epoch": 0.1725369851466729, "grad_norm": 1.1751890182495117, "learning_rate": 1.8953645226311056e-05, "loss": 0.15890636444091796, "step": 6550 }, { "epoch": 0.17280040039117164, "grad_norm": 2.0884132385253906, "learning_rate": 1.894984265805239e-05, "loss": 0.18127732276916503, "step": 6560 }, { "epoch": 0.1730638156356704, "grad_norm": 1.609208345413208, "learning_rate": 1.8946033575712104e-05, "loss": 0.1604251742362976, "step": 6570 }, { "epoch": 0.1733272308801691, "grad_norm": 1.629771113395691, "learning_rate": 1.894221798206262e-05, "loss": 0.1723095655441284, "step": 6580 }, { "epoch": 0.17359064612466785, "grad_norm": 3.357144832611084, "learning_rate": 1.8938395879881086e-05, "loss": 0.17733144760131836, "step": 6590 }, { "epoch": 0.1738540613691666, "grad_norm": 1.3412683010101318, "learning_rate": 1.8934567271949397e-05, "loss": 0.17680714130401612, "step": 6600 }, { "epoch": 0.17411747661366533, "grad_norm": 3.142411947250366, "learning_rate": 1.8930732161054176e-05, "loss": 0.16094101667404176, "step": 6610 }, { "epoch": 0.17438089185816405, "grad_norm": 4.867790699005127, "learning_rate": 1.8926890549986786e-05, "loss": 0.16634843349456788, "step": 6620 }, { "epoch": 0.1746443071026628, "grad_norm": 3.4914278984069824, "learning_rate": 1.8923042441543313e-05, "loss": 0.16267772912979125, "step": 6630 }, { "epoch": 0.17490772234716154, "grad_norm": 5.881528854370117, "learning_rate": 1.891918783852458e-05, "loss": 0.1746336817741394, "step": 6640 }, { "epoch": 0.17517113759166028, "grad_norm": 1.0424351692199707, "learning_rate": 1.8915326743736134e-05, "loss": 0.14768381118774415, "step": 6650 }, { "epoch": 0.175434552836159, "grad_norm": 4.726067543029785, "learning_rate": 1.8911459159988238e-05, "loss": 0.17255690097808837, "step": 6660 }, { "epoch": 0.17569796808065774, "grad_norm": 2.824528217315674, "learning_rate": 1.8907585090095897e-05, "loss": 0.16900087594985963, "step": 6670 }, { "epoch": 0.17596138332515648, "grad_norm": 4.180332660675049, "learning_rate": 1.890370453687882e-05, "loss": 0.16901683807373047, "step": 6680 }, { "epoch": 0.17622479856965523, "grad_norm": 3.8776910305023193, "learning_rate": 1.889981750316144e-05, "loss": 0.17774875164031984, "step": 6690 }, { "epoch": 0.17648821381415397, "grad_norm": 2.9989404678344727, "learning_rate": 1.8895923991772913e-05, "loss": 0.16902604103088378, "step": 6700 }, { "epoch": 0.1767516290586527, "grad_norm": 2.5293660163879395, "learning_rate": 1.8892024005547104e-05, "loss": 0.18483250141143798, "step": 6710 }, { "epoch": 0.17701504430315143, "grad_norm": 6.220608234405518, "learning_rate": 1.8888117547322586e-05, "loss": 0.16430799961090087, "step": 6720 }, { "epoch": 0.17727845954765017, "grad_norm": 3.614666223526001, "learning_rate": 1.888420461994265e-05, "loss": 0.16829793453216552, "step": 6730 }, { "epoch": 0.17754187479214892, "grad_norm": 3.590768575668335, "learning_rate": 1.88802852262553e-05, "loss": 0.1603969693183899, "step": 6740 }, { "epoch": 0.17780529003664763, "grad_norm": 4.85261344909668, "learning_rate": 1.887635936911323e-05, "loss": 0.17536864280700684, "step": 6750 }, { "epoch": 0.17806870528114638, "grad_norm": 3.0255179405212402, "learning_rate": 1.8872427051373852e-05, "loss": 0.1678982973098755, "step": 6760 }, { "epoch": 0.17833212052564512, "grad_norm": 1.4360747337341309, "learning_rate": 1.886848827589928e-05, "loss": 0.1614221453666687, "step": 6770 }, { "epoch": 0.17859553577014387, "grad_norm": 2.207235097885132, "learning_rate": 1.8864543045556318e-05, "loss": 0.16603639125823974, "step": 6780 }, { "epoch": 0.1788589510146426, "grad_norm": 5.60127592086792, "learning_rate": 1.8860591363216482e-05, "loss": 0.1683640480041504, "step": 6790 }, { "epoch": 0.17912236625914132, "grad_norm": 2.255174160003662, "learning_rate": 1.8856633231755972e-05, "loss": 0.16753889322280885, "step": 6800 }, { "epoch": 0.17938578150364007, "grad_norm": 1.1590032577514648, "learning_rate": 1.8852668654055686e-05, "loss": 0.1688519835472107, "step": 6810 }, { "epoch": 0.1796491967481388, "grad_norm": 2.893728494644165, "learning_rate": 1.884869763300122e-05, "loss": 0.1713976502418518, "step": 6820 }, { "epoch": 0.17991261199263756, "grad_norm": 2.2206146717071533, "learning_rate": 1.8844720171482854e-05, "loss": 0.1604481339454651, "step": 6830 }, { "epoch": 0.18017602723713627, "grad_norm": 2.511606454849243, "learning_rate": 1.884073627239555e-05, "loss": 0.17083828449249266, "step": 6840 }, { "epoch": 0.18043944248163502, "grad_norm": 3.612034320831299, "learning_rate": 1.8836745938638963e-05, "loss": 0.17726129293441772, "step": 6850 }, { "epoch": 0.18070285772613376, "grad_norm": 2.5946240425109863, "learning_rate": 1.8832749173117435e-05, "loss": 0.1767590045928955, "step": 6860 }, { "epoch": 0.1809662729706325, "grad_norm": 2.1195859909057617, "learning_rate": 1.8828745978739982e-05, "loss": 0.16807641983032226, "step": 6870 }, { "epoch": 0.18122968821513122, "grad_norm": 1.3928111791610718, "learning_rate": 1.88247363584203e-05, "loss": 0.16512480974197388, "step": 6880 }, { "epoch": 0.18149310345962996, "grad_norm": 3.0671212673187256, "learning_rate": 1.8820720315076767e-05, "loss": 0.16269617080688475, "step": 6890 }, { "epoch": 0.1817565187041287, "grad_norm": 2.9817705154418945, "learning_rate": 1.8816697851632425e-05, "loss": 0.17684656381607056, "step": 6900 }, { "epoch": 0.18201993394862745, "grad_norm": 1.9153692722320557, "learning_rate": 1.8812668971015006e-05, "loss": 0.16586185693740846, "step": 6910 }, { "epoch": 0.1822833491931262, "grad_norm": 1.4657378196716309, "learning_rate": 1.88086336761569e-05, "loss": 0.16534814834594727, "step": 6920 }, { "epoch": 0.1825467644376249, "grad_norm": 1.4667502641677856, "learning_rate": 1.8804591969995167e-05, "loss": 0.1709376573562622, "step": 6930 }, { "epoch": 0.18281017968212365, "grad_norm": 2.831915855407715, "learning_rate": 1.8800543855471535e-05, "loss": 0.16813244819641113, "step": 6940 }, { "epoch": 0.1830735949266224, "grad_norm": 2.626401901245117, "learning_rate": 1.87964893355324e-05, "loss": 0.16098377704620362, "step": 6950 }, { "epoch": 0.18333701017112114, "grad_norm": 2.9156668186187744, "learning_rate": 1.879242841312881e-05, "loss": 0.1658147931098938, "step": 6960 }, { "epoch": 0.18360042541561986, "grad_norm": 4.831404685974121, "learning_rate": 1.878836109121648e-05, "loss": 0.16169953346252441, "step": 6970 }, { "epoch": 0.1838638406601186, "grad_norm": 2.0507233142852783, "learning_rate": 1.8784287372755795e-05, "loss": 0.1782217502593994, "step": 6980 }, { "epoch": 0.18412725590461734, "grad_norm": 5.629910469055176, "learning_rate": 1.8780207260711767e-05, "loss": 0.1725234031677246, "step": 6990 }, { "epoch": 0.1843906711491161, "grad_norm": 1.5877234935760498, "learning_rate": 1.8776120758054087e-05, "loss": 0.17216978073120118, "step": 7000 }, { "epoch": 0.1846540863936148, "grad_norm": 3.471071481704712, "learning_rate": 1.8772027867757082e-05, "loss": 0.1632932424545288, "step": 7010 }, { "epoch": 0.18491750163811355, "grad_norm": 4.4580607414245605, "learning_rate": 1.8767928592799742e-05, "loss": 0.16594367027282714, "step": 7020 }, { "epoch": 0.1851809168826123, "grad_norm": 3.4564599990844727, "learning_rate": 1.876382293616569e-05, "loss": 0.1781316041946411, "step": 7030 }, { "epoch": 0.18544433212711103, "grad_norm": 1.3190487623214722, "learning_rate": 1.8759710900843202e-05, "loss": 0.17143559455871582, "step": 7040 }, { "epoch": 0.18570774737160978, "grad_norm": 2.3162484169006348, "learning_rate": 1.8755592489825193e-05, "loss": 0.1535707473754883, "step": 7050 }, { "epoch": 0.1859711626161085, "grad_norm": 2.422478675842285, "learning_rate": 1.8751467706109224e-05, "loss": 0.17743272781372071, "step": 7060 }, { "epoch": 0.18623457786060724, "grad_norm": 4.517858028411865, "learning_rate": 1.8747336552697483e-05, "loss": 0.16838529109954833, "step": 7070 }, { "epoch": 0.18649799310510598, "grad_norm": 1.736438274383545, "learning_rate": 1.874319903259681e-05, "loss": 0.16544835567474364, "step": 7080 }, { "epoch": 0.18676140834960472, "grad_norm": 1.2915745973587036, "learning_rate": 1.873905514881867e-05, "loss": 0.1783340573310852, "step": 7090 }, { "epoch": 0.18702482359410344, "grad_norm": 3.8017547130584717, "learning_rate": 1.873490490437915e-05, "loss": 0.16258809566497803, "step": 7100 }, { "epoch": 0.18728823883860218, "grad_norm": 5.146725177764893, "learning_rate": 1.8730748302298993e-05, "loss": 0.16960835456848145, "step": 7110 }, { "epoch": 0.18755165408310093, "grad_norm": 4.269168376922607, "learning_rate": 1.8726585345603544e-05, "loss": 0.19395184516906738, "step": 7120 }, { "epoch": 0.18781506932759967, "grad_norm": 1.4230202436447144, "learning_rate": 1.8722416037322778e-05, "loss": 0.17387229204177856, "step": 7130 }, { "epoch": 0.1880784845720984, "grad_norm": 2.4117202758789062, "learning_rate": 1.871824038049131e-05, "loss": 0.17967348098754882, "step": 7140 }, { "epoch": 0.18834189981659713, "grad_norm": 2.7643206119537354, "learning_rate": 1.871405837814835e-05, "loss": 0.16199285984039308, "step": 7150 }, { "epoch": 0.18860531506109587, "grad_norm": 1.7345882654190063, "learning_rate": 1.8709870033337753e-05, "loss": 0.18592183589935302, "step": 7160 }, { "epoch": 0.18886873030559462, "grad_norm": 4.404778957366943, "learning_rate": 1.8705675349107964e-05, "loss": 0.16570388078689574, "step": 7170 }, { "epoch": 0.18913214555009336, "grad_norm": 2.9908127784729004, "learning_rate": 1.870147432851206e-05, "loss": 0.16444692611694336, "step": 7180 }, { "epoch": 0.18939556079459208, "grad_norm": 3.122044563293457, "learning_rate": 1.8697266974607732e-05, "loss": 0.17177481651306153, "step": 7190 }, { "epoch": 0.18965897603909082, "grad_norm": 3.2723615169525146, "learning_rate": 1.8693053290457264e-05, "loss": 0.160370397567749, "step": 7200 }, { "epoch": 0.18992239128358956, "grad_norm": 2.1007938385009766, "learning_rate": 1.868883327912756e-05, "loss": 0.17296547889709474, "step": 7210 }, { "epoch": 0.1901858065280883, "grad_norm": 1.6526496410369873, "learning_rate": 1.868460694369013e-05, "loss": 0.17775368690490723, "step": 7220 }, { "epoch": 0.19044922177258702, "grad_norm": 2.7214109897613525, "learning_rate": 1.868037428722108e-05, "loss": 0.1657734990119934, "step": 7230 }, { "epoch": 0.19071263701708577, "grad_norm": 3.3503100872039795, "learning_rate": 1.867613531280112e-05, "loss": 0.16402392387390136, "step": 7240 }, { "epoch": 0.1909760522615845, "grad_norm": 2.213954210281372, "learning_rate": 1.867189002351556e-05, "loss": 0.1669006586074829, "step": 7250 }, { "epoch": 0.19123946750608325, "grad_norm": 4.105005264282227, "learning_rate": 1.8667638422454306e-05, "loss": 0.17047407627105712, "step": 7260 }, { "epoch": 0.19150288275058197, "grad_norm": 1.9198588132858276, "learning_rate": 1.8663380512711852e-05, "loss": 0.16588655710220337, "step": 7270 }, { "epoch": 0.19176629799508071, "grad_norm": 1.587346076965332, "learning_rate": 1.865911629738729e-05, "loss": 0.16331765651702881, "step": 7280 }, { "epoch": 0.19202971323957946, "grad_norm": 2.0019118785858154, "learning_rate": 1.86548457795843e-05, "loss": 0.17304718494415283, "step": 7290 }, { "epoch": 0.1922931284840782, "grad_norm": 2.127439022064209, "learning_rate": 1.8650568962411154e-05, "loss": 0.16698362827301025, "step": 7300 }, { "epoch": 0.19255654372857695, "grad_norm": 2.1485657691955566, "learning_rate": 1.8646285848980693e-05, "loss": 0.1593562126159668, "step": 7310 }, { "epoch": 0.19281995897307566, "grad_norm": 6.84826135635376, "learning_rate": 1.8641996442410365e-05, "loss": 0.1482262372970581, "step": 7320 }, { "epoch": 0.1930833742175744, "grad_norm": 1.1645580530166626, "learning_rate": 1.8637700745822176e-05, "loss": 0.17384862899780273, "step": 7330 }, { "epoch": 0.19334678946207315, "grad_norm": 0.621807873249054, "learning_rate": 1.8633398762342722e-05, "loss": 0.16933817863464357, "step": 7340 }, { "epoch": 0.1936102047065719, "grad_norm": 3.396801710128784, "learning_rate": 1.8629090495103175e-05, "loss": 0.1693096160888672, "step": 7350 }, { "epoch": 0.1938736199510706, "grad_norm": 2.3611135482788086, "learning_rate": 1.8624775947239272e-05, "loss": 0.14679422378540039, "step": 7360 }, { "epoch": 0.19413703519556935, "grad_norm": 3.043168544769287, "learning_rate": 1.862045512189133e-05, "loss": 0.16477795839309692, "step": 7370 }, { "epoch": 0.1944004504400681, "grad_norm": 3.0012407302856445, "learning_rate": 1.8616128022204233e-05, "loss": 0.16934062242507936, "step": 7380 }, { "epoch": 0.19466386568456684, "grad_norm": 1.9098526239395142, "learning_rate": 1.8611794651327432e-05, "loss": 0.17128474712371827, "step": 7390 }, { "epoch": 0.19492728092906556, "grad_norm": 1.7310293912887573, "learning_rate": 1.8607455012414936e-05, "loss": 0.161135733127594, "step": 7400 }, { "epoch": 0.1951906961735643, "grad_norm": 1.1125842332839966, "learning_rate": 1.8603109108625327e-05, "loss": 0.17037283182144164, "step": 7410 }, { "epoch": 0.19545411141806304, "grad_norm": 3.622380256652832, "learning_rate": 1.859875694312174e-05, "loss": 0.15808781385421752, "step": 7420 }, { "epoch": 0.19571752666256179, "grad_norm": 1.0731163024902344, "learning_rate": 1.8594398519071867e-05, "loss": 0.17566795349121095, "step": 7430 }, { "epoch": 0.19598094190706053, "grad_norm": 1.5883700847625732, "learning_rate": 1.8590033839647955e-05, "loss": 0.17601234912872316, "step": 7440 }, { "epoch": 0.19624435715155925, "grad_norm": 0.9166069030761719, "learning_rate": 1.8585662908026814e-05, "loss": 0.15982954502105712, "step": 7450 }, { "epoch": 0.196507772396058, "grad_norm": 1.1139459609985352, "learning_rate": 1.8581285727389787e-05, "loss": 0.16959104537963868, "step": 7460 }, { "epoch": 0.19677118764055673, "grad_norm": 5.590317249298096, "learning_rate": 1.857690230092278e-05, "loss": 0.17953134775161744, "step": 7470 }, { "epoch": 0.19703460288505548, "grad_norm": 2.6799237728118896, "learning_rate": 1.8572512631816243e-05, "loss": 0.16493169069290162, "step": 7480 }, { "epoch": 0.1972980181295542, "grad_norm": 1.339092493057251, "learning_rate": 1.856811672326516e-05, "loss": 0.15501129627227783, "step": 7490 }, { "epoch": 0.19756143337405294, "grad_norm": 2.285064220428467, "learning_rate": 1.8563714578469064e-05, "loss": 0.1703229069709778, "step": 7500 }, { "epoch": 0.19782484861855168, "grad_norm": 3.6907711029052734, "learning_rate": 1.8559306200632036e-05, "loss": 0.16858415603637694, "step": 7510 }, { "epoch": 0.19808826386305042, "grad_norm": 1.6432993412017822, "learning_rate": 1.8554891592962666e-05, "loss": 0.15639195442199708, "step": 7520 }, { "epoch": 0.19835167910754914, "grad_norm": 1.6998028755187988, "learning_rate": 1.8550470758674112e-05, "loss": 0.1613994359970093, "step": 7530 }, { "epoch": 0.19861509435204788, "grad_norm": 1.548936367034912, "learning_rate": 1.8546043700984036e-05, "loss": 0.1699316143989563, "step": 7540 }, { "epoch": 0.19887850959654663, "grad_norm": 2.0332326889038086, "learning_rate": 1.8541610423114647e-05, "loss": 0.17319233417510987, "step": 7550 }, { "epoch": 0.19914192484104537, "grad_norm": 3.3424320220947266, "learning_rate": 1.853717092829268e-05, "loss": 0.17201372385025024, "step": 7560 }, { "epoch": 0.1994053400855441, "grad_norm": 1.8622283935546875, "learning_rate": 1.853272521974938e-05, "loss": 0.17176756858825684, "step": 7570 }, { "epoch": 0.19966875533004283, "grad_norm": 1.7094417810440063, "learning_rate": 1.8528273300720538e-05, "loss": 0.1654815912246704, "step": 7580 }, { "epoch": 0.19993217057454157, "grad_norm": 2.0944252014160156, "learning_rate": 1.8523815174446444e-05, "loss": 0.16835209131240844, "step": 7590 }, { "epoch": 0.20019558581904032, "grad_norm": 1.4034842252731323, "learning_rate": 1.851935084417192e-05, "loss": 0.15724371671676635, "step": 7600 }, { "epoch": 0.20045900106353906, "grad_norm": 1.5546433925628662, "learning_rate": 1.8514880313146295e-05, "loss": 0.178194260597229, "step": 7610 }, { "epoch": 0.20072241630803778, "grad_norm": 1.8767272233963013, "learning_rate": 1.8510403584623414e-05, "loss": 0.17449839115142823, "step": 7620 }, { "epoch": 0.20098583155253652, "grad_norm": 6.026278018951416, "learning_rate": 1.850592066186163e-05, "loss": 0.15971677303314208, "step": 7630 }, { "epoch": 0.20124924679703526, "grad_norm": 2.864448308944702, "learning_rate": 1.850143154812382e-05, "loss": 0.15981996059417725, "step": 7640 }, { "epoch": 0.201512662041534, "grad_norm": 2.3485989570617676, "learning_rate": 1.849693624667734e-05, "loss": 0.15783698558807374, "step": 7650 }, { "epoch": 0.20177607728603272, "grad_norm": 1.5421316623687744, "learning_rate": 1.8492434760794072e-05, "loss": 0.17067272663116456, "step": 7660 }, { "epoch": 0.20203949253053147, "grad_norm": 1.5930255651474, "learning_rate": 1.848792709375039e-05, "loss": 0.16022138595581054, "step": 7670 }, { "epoch": 0.2023029077750302, "grad_norm": 1.0202730894088745, "learning_rate": 1.848341324882717e-05, "loss": 0.15607467889785767, "step": 7680 }, { "epoch": 0.20256632301952895, "grad_norm": 1.013343334197998, "learning_rate": 1.8478893229309784e-05, "loss": 0.17362754344940184, "step": 7690 }, { "epoch": 0.2028297382640277, "grad_norm": 1.4814964532852173, "learning_rate": 1.847436703848809e-05, "loss": 0.165701961517334, "step": 7700 }, { "epoch": 0.20309315350852641, "grad_norm": 3.288787603378296, "learning_rate": 1.8469834679656452e-05, "loss": 0.17550396919250488, "step": 7710 }, { "epoch": 0.20335656875302516, "grad_norm": 2.145028591156006, "learning_rate": 1.8465296156113715e-05, "loss": 0.16549068689346313, "step": 7720 }, { "epoch": 0.2036199839975239, "grad_norm": 1.8425043821334839, "learning_rate": 1.8460751471163212e-05, "loss": 0.1743842124938965, "step": 7730 }, { "epoch": 0.20388339924202264, "grad_norm": 2.042139768600464, "learning_rate": 1.845620062811276e-05, "loss": 0.17596144676208497, "step": 7740 }, { "epoch": 0.20414681448652136, "grad_norm": 3.183049440383911, "learning_rate": 1.8451643630274664e-05, "loss": 0.16308995485305786, "step": 7750 }, { "epoch": 0.2044102297310201, "grad_norm": 1.5489860773086548, "learning_rate": 1.8447080480965694e-05, "loss": 0.1677347779273987, "step": 7760 }, { "epoch": 0.20467364497551885, "grad_norm": 2.8942673206329346, "learning_rate": 1.844251118350711e-05, "loss": 0.16435208320617675, "step": 7770 }, { "epoch": 0.2049370602200176, "grad_norm": 2.353044271469116, "learning_rate": 1.8437935741224652e-05, "loss": 0.17543631792068481, "step": 7780 }, { "epoch": 0.2052004754645163, "grad_norm": 1.5347014665603638, "learning_rate": 1.8433354157448513e-05, "loss": 0.17307164669036865, "step": 7790 }, { "epoch": 0.20546389070901505, "grad_norm": 3.9742798805236816, "learning_rate": 1.8428766435513373e-05, "loss": 0.17278904914855958, "step": 7800 }, { "epoch": 0.2057273059535138, "grad_norm": 5.510382652282715, "learning_rate": 1.8424172578758374e-05, "loss": 0.16817884445190429, "step": 7810 }, { "epoch": 0.20599072119801254, "grad_norm": 3.832531690597534, "learning_rate": 1.841957259052712e-05, "loss": 0.16358866691589355, "step": 7820 }, { "epoch": 0.20625413644251128, "grad_norm": 1.7288804054260254, "learning_rate": 1.8414966474167688e-05, "loss": 0.1737713098526001, "step": 7830 }, { "epoch": 0.20651755168701, "grad_norm": 5.008682727813721, "learning_rate": 1.8410354233032604e-05, "loss": 0.15977586507797242, "step": 7840 }, { "epoch": 0.20678096693150874, "grad_norm": 1.0487114191055298, "learning_rate": 1.840573587047885e-05, "loss": 0.1645817756652832, "step": 7850 }, { "epoch": 0.20704438217600749, "grad_norm": 1.7940717935562134, "learning_rate": 1.8401111389867882e-05, "loss": 0.176603102684021, "step": 7860 }, { "epoch": 0.20730779742050623, "grad_norm": 3.948103904724121, "learning_rate": 1.8396480794565583e-05, "loss": 0.16182925701141357, "step": 7870 }, { "epoch": 0.20757121266500494, "grad_norm": 3.251309633255005, "learning_rate": 1.839184408794231e-05, "loss": 0.17302812337875367, "step": 7880 }, { "epoch": 0.2078346279095037, "grad_norm": 1.5409390926361084, "learning_rate": 1.8387201273372855e-05, "loss": 0.16679840087890624, "step": 7890 }, { "epoch": 0.20809804315400243, "grad_norm": 4.928312301635742, "learning_rate": 1.838255235423646e-05, "loss": 0.1576216459274292, "step": 7900 }, { "epoch": 0.20836145839850118, "grad_norm": 1.3344881534576416, "learning_rate": 1.8377897333916808e-05, "loss": 0.16644176244735717, "step": 7910 }, { "epoch": 0.2086248736429999, "grad_norm": 5.89364767074585, "learning_rate": 1.837323621580202e-05, "loss": 0.15821770429611207, "step": 7920 }, { "epoch": 0.20888828888749864, "grad_norm": 3.4641494750976562, "learning_rate": 1.8368569003284662e-05, "loss": 0.16312217712402344, "step": 7930 }, { "epoch": 0.20915170413199738, "grad_norm": 2.005476236343384, "learning_rate": 1.8363895699761736e-05, "loss": 0.16748005151748657, "step": 7940 }, { "epoch": 0.20941511937649612, "grad_norm": 5.866455554962158, "learning_rate": 1.8359216308634668e-05, "loss": 0.16239014863967896, "step": 7950 }, { "epoch": 0.20967853462099487, "grad_norm": 0.6929324269294739, "learning_rate": 1.8354530833309323e-05, "loss": 0.16845121383666992, "step": 7960 }, { "epoch": 0.20994194986549358, "grad_norm": 3.7118189334869385, "learning_rate": 1.8349839277195993e-05, "loss": 0.15725793838500976, "step": 7970 }, { "epoch": 0.21020536510999233, "grad_norm": 2.428861618041992, "learning_rate": 1.8345141643709394e-05, "loss": 0.16366167068481446, "step": 7980 }, { "epoch": 0.21046878035449107, "grad_norm": 3.1248464584350586, "learning_rate": 1.8340437936268667e-05, "loss": 0.16965696811676026, "step": 7990 }, { "epoch": 0.2107321955989898, "grad_norm": 3.9007816314697266, "learning_rate": 1.8335728158297374e-05, "loss": 0.1753111958503723, "step": 8000 }, { "epoch": 0.21099561084348853, "grad_norm": 1.2851256132125854, "learning_rate": 1.833101231322349e-05, "loss": 0.17173566818237304, "step": 8010 }, { "epoch": 0.21125902608798727, "grad_norm": 3.157607316970825, "learning_rate": 1.832629040447942e-05, "loss": 0.1757857084274292, "step": 8020 }, { "epoch": 0.21152244133248602, "grad_norm": 1.9891291856765747, "learning_rate": 1.8321562435501963e-05, "loss": 0.1832936406135559, "step": 8030 }, { "epoch": 0.21178585657698476, "grad_norm": 3.16850209236145, "learning_rate": 1.831682840973234e-05, "loss": 0.16396868228912354, "step": 8040 }, { "epoch": 0.21204927182148348, "grad_norm": 2.0873146057128906, "learning_rate": 1.8312088330616188e-05, "loss": 0.16990110874176026, "step": 8050 }, { "epoch": 0.21231268706598222, "grad_norm": 1.5688426494598389, "learning_rate": 1.8307342201603533e-05, "loss": 0.16079044342041016, "step": 8060 }, { "epoch": 0.21257610231048096, "grad_norm": 2.2465457916259766, "learning_rate": 1.830259002614881e-05, "loss": 0.15069980621337892, "step": 8070 }, { "epoch": 0.2128395175549797, "grad_norm": 1.3416025638580322, "learning_rate": 1.829783180771087e-05, "loss": 0.17285652160644532, "step": 8080 }, { "epoch": 0.21310293279947845, "grad_norm": 1.4384974241256714, "learning_rate": 1.8293067549752937e-05, "loss": 0.1694039821624756, "step": 8090 }, { "epoch": 0.21336634804397717, "grad_norm": 1.0770974159240723, "learning_rate": 1.8288297255742646e-05, "loss": 0.18700462579727173, "step": 8100 }, { "epoch": 0.2136297632884759, "grad_norm": 3.0609841346740723, "learning_rate": 1.828352092915203e-05, "loss": 0.17289925813674928, "step": 8110 }, { "epoch": 0.21389317853297465, "grad_norm": 1.5945316553115845, "learning_rate": 1.8278738573457496e-05, "loss": 0.15527472496032715, "step": 8120 }, { "epoch": 0.2141565937774734, "grad_norm": 1.2506096363067627, "learning_rate": 1.8273950192139854e-05, "loss": 0.16812877655029296, "step": 8130 }, { "epoch": 0.2144200090219721, "grad_norm": 3.609356641769409, "learning_rate": 1.826915578868429e-05, "loss": 0.172845721244812, "step": 8140 }, { "epoch": 0.21468342426647086, "grad_norm": 3.5277345180511475, "learning_rate": 1.8264355366580387e-05, "loss": 0.16352579593658448, "step": 8150 }, { "epoch": 0.2149468395109696, "grad_norm": 1.1969075202941895, "learning_rate": 1.825954892932209e-05, "loss": 0.16508721113204955, "step": 8160 }, { "epoch": 0.21521025475546834, "grad_norm": 1.7953121662139893, "learning_rate": 1.8254736480407738e-05, "loss": 0.16437215805053712, "step": 8170 }, { "epoch": 0.21547366999996706, "grad_norm": 1.8100930452346802, "learning_rate": 1.8249918023340036e-05, "loss": 0.15855648517608642, "step": 8180 }, { "epoch": 0.2157370852444658, "grad_norm": 3.6664044857025146, "learning_rate": 1.8245093561626063e-05, "loss": 0.1707010507583618, "step": 8190 }, { "epoch": 0.21600050048896455, "grad_norm": 4.795748233795166, "learning_rate": 1.824026309877727e-05, "loss": 0.1738038420677185, "step": 8200 }, { "epoch": 0.2162639157334633, "grad_norm": 3.424812078475952, "learning_rate": 1.8235426638309483e-05, "loss": 0.17486697435379028, "step": 8210 }, { "epoch": 0.21652733097796203, "grad_norm": 1.9768824577331543, "learning_rate": 1.823058418374288e-05, "loss": 0.175701105594635, "step": 8220 }, { "epoch": 0.21679074622246075, "grad_norm": 2.4561522006988525, "learning_rate": 1.822573573860201e-05, "loss": 0.16262245178222656, "step": 8230 }, { "epoch": 0.2170541614669595, "grad_norm": 0.702596127986908, "learning_rate": 1.8220881306415783e-05, "loss": 0.1600751042366028, "step": 8240 }, { "epoch": 0.21731757671145824, "grad_norm": 1.439475178718567, "learning_rate": 1.821602089071746e-05, "loss": 0.15883415937423706, "step": 8250 }, { "epoch": 0.21758099195595698, "grad_norm": 2.1011569499969482, "learning_rate": 1.8211154495044666e-05, "loss": 0.17233335971832275, "step": 8260 }, { "epoch": 0.2178444072004557, "grad_norm": 1.3548786640167236, "learning_rate": 1.8206282122939368e-05, "loss": 0.15652003288269042, "step": 8270 }, { "epoch": 0.21810782244495444, "grad_norm": 2.367825984954834, "learning_rate": 1.8201403777947896e-05, "loss": 0.16075009107589722, "step": 8280 }, { "epoch": 0.21837123768945318, "grad_norm": 1.1987932920455933, "learning_rate": 1.8196519463620912e-05, "loss": 0.1687687397003174, "step": 8290 }, { "epoch": 0.21863465293395193, "grad_norm": 2.017791986465454, "learning_rate": 1.8191629183513435e-05, "loss": 0.16607096195220947, "step": 8300 }, { "epoch": 0.21889806817845064, "grad_norm": 1.3950587511062622, "learning_rate": 1.8186732941184824e-05, "loss": 0.16602795124053954, "step": 8310 }, { "epoch": 0.2191614834229494, "grad_norm": 1.2408851385116577, "learning_rate": 1.818183074019877e-05, "loss": 0.1644521951675415, "step": 8320 }, { "epoch": 0.21942489866744813, "grad_norm": 2.8829662799835205, "learning_rate": 1.8176922584123308e-05, "loss": 0.15963393449783325, "step": 8330 }, { "epoch": 0.21968831391194688, "grad_norm": 3.49430251121521, "learning_rate": 1.817200847653081e-05, "loss": 0.18338887691497802, "step": 8340 }, { "epoch": 0.21995172915644562, "grad_norm": 1.922873854637146, "learning_rate": 1.8167088420997975e-05, "loss": 0.18248612880706788, "step": 8350 }, { "epoch": 0.22021514440094433, "grad_norm": 2.5819191932678223, "learning_rate": 1.816216242110582e-05, "loss": 0.17005912065505982, "step": 8360 }, { "epoch": 0.22047855964544308, "grad_norm": 1.081637978553772, "learning_rate": 1.815723048043972e-05, "loss": 0.15965933799743653, "step": 8370 }, { "epoch": 0.22074197488994182, "grad_norm": 1.296785831451416, "learning_rate": 1.8152292602589336e-05, "loss": 0.1670219659805298, "step": 8380 }, { "epoch": 0.22100539013444057, "grad_norm": 1.687735915184021, "learning_rate": 1.8147348791148678e-05, "loss": 0.17716505527496337, "step": 8390 }, { "epoch": 0.22126880537893928, "grad_norm": 0.9895438551902771, "learning_rate": 1.8142399049716065e-05, "loss": 0.16179924011230468, "step": 8400 }, { "epoch": 0.22153222062343803, "grad_norm": 1.7005356550216675, "learning_rate": 1.8137443381894126e-05, "loss": 0.1751154661178589, "step": 8410 }, { "epoch": 0.22179563586793677, "grad_norm": 1.0185707807540894, "learning_rate": 1.813248179128982e-05, "loss": 0.16117196083068847, "step": 8420 }, { "epoch": 0.2220590511124355, "grad_norm": 1.2613309621810913, "learning_rate": 1.81275142815144e-05, "loss": 0.17357947826385497, "step": 8430 }, { "epoch": 0.22232246635693423, "grad_norm": 3.6929891109466553, "learning_rate": 1.812254085618343e-05, "loss": 0.17285339832305907, "step": 8440 }, { "epoch": 0.22258588160143297, "grad_norm": 3.691772937774658, "learning_rate": 1.8117561518916792e-05, "loss": 0.1655185580253601, "step": 8450 }, { "epoch": 0.22284929684593172, "grad_norm": 2.026895761489868, "learning_rate": 1.811257627333866e-05, "loss": 0.1670604944229126, "step": 8460 }, { "epoch": 0.22311271209043046, "grad_norm": 0.7056310772895813, "learning_rate": 1.810758512307751e-05, "loss": 0.16574815511703492, "step": 8470 }, { "epoch": 0.2233761273349292, "grad_norm": 4.772472381591797, "learning_rate": 1.8102588071766116e-05, "loss": 0.16918550729751586, "step": 8480 }, { "epoch": 0.22363954257942792, "grad_norm": 2.958854913711548, "learning_rate": 1.8097585123041547e-05, "loss": 0.15863226652145385, "step": 8490 }, { "epoch": 0.22390295782392666, "grad_norm": 3.7478930950164795, "learning_rate": 1.809257628054517e-05, "loss": 0.15998196601867676, "step": 8500 }, { "epoch": 0.2241663730684254, "grad_norm": 1.569686770439148, "learning_rate": 1.808756154792263e-05, "loss": 0.16526247262954713, "step": 8510 }, { "epoch": 0.22442978831292415, "grad_norm": 3.3055508136749268, "learning_rate": 1.8082540928823866e-05, "loss": 0.17167898416519164, "step": 8520 }, { "epoch": 0.22469320355742287, "grad_norm": 1.1258971691131592, "learning_rate": 1.8077514426903107e-05, "loss": 0.16882200241088868, "step": 8530 }, { "epoch": 0.2249566188019216, "grad_norm": 2.304652452468872, "learning_rate": 1.8072482045818852e-05, "loss": 0.15322906970977784, "step": 8540 }, { "epoch": 0.22522003404642035, "grad_norm": 2.6191112995147705, "learning_rate": 1.806744378923389e-05, "loss": 0.16137789487838744, "step": 8550 }, { "epoch": 0.2254834492909191, "grad_norm": 2.500746488571167, "learning_rate": 1.8062399660815277e-05, "loss": 0.17000045776367187, "step": 8560 }, { "epoch": 0.2257468645354178, "grad_norm": 1.6301957368850708, "learning_rate": 1.8057349664234348e-05, "loss": 0.15356210470199586, "step": 8570 }, { "epoch": 0.22601027977991656, "grad_norm": 2.0379834175109863, "learning_rate": 1.8052293803166714e-05, "loss": 0.16949565410614015, "step": 8580 }, { "epoch": 0.2262736950244153, "grad_norm": 1.3554596900939941, "learning_rate": 1.8047232081292242e-05, "loss": 0.16646510362625122, "step": 8590 }, { "epoch": 0.22653711026891404, "grad_norm": 2.0863664150238037, "learning_rate": 1.8042164502295074e-05, "loss": 0.18160871267318726, "step": 8600 }, { "epoch": 0.2268005255134128, "grad_norm": 0.6402860283851624, "learning_rate": 1.8037091069863615e-05, "loss": 0.16525942087173462, "step": 8610 }, { "epoch": 0.2270639407579115, "grad_norm": 1.8418846130371094, "learning_rate": 1.8032011787690523e-05, "loss": 0.15705316066741942, "step": 8620 }, { "epoch": 0.22732735600241025, "grad_norm": 3.4998011589050293, "learning_rate": 1.8026926659472727e-05, "loss": 0.15644218921661376, "step": 8630 }, { "epoch": 0.227590771246909, "grad_norm": 1.8136348724365234, "learning_rate": 1.802183568891139e-05, "loss": 0.16772842407226562, "step": 8640 }, { "epoch": 0.22785418649140773, "grad_norm": 2.9936470985412598, "learning_rate": 1.801673887971195e-05, "loss": 0.17361016273498536, "step": 8650 }, { "epoch": 0.22811760173590645, "grad_norm": 1.7538431882858276, "learning_rate": 1.801163623558408e-05, "loss": 0.17402690649032593, "step": 8660 }, { "epoch": 0.2283810169804052, "grad_norm": 1.6880005598068237, "learning_rate": 1.800652776024171e-05, "loss": 0.1672511339187622, "step": 8670 }, { "epoch": 0.22864443222490394, "grad_norm": 2.087268352508545, "learning_rate": 1.8001413457403005e-05, "loss": 0.16715874671936035, "step": 8680 }, { "epoch": 0.22890784746940268, "grad_norm": 2.1578896045684814, "learning_rate": 1.7996293330790374e-05, "loss": 0.16884129047393798, "step": 8690 }, { "epoch": 0.2291712627139014, "grad_norm": 1.6546615362167358, "learning_rate": 1.7991167384130466e-05, "loss": 0.15945804119110107, "step": 8700 }, { "epoch": 0.22943467795840014, "grad_norm": 1.6826118230819702, "learning_rate": 1.798603562115417e-05, "loss": 0.16339805126190185, "step": 8710 }, { "epoch": 0.22969809320289888, "grad_norm": 1.463374137878418, "learning_rate": 1.79808980455966e-05, "loss": 0.16220271587371826, "step": 8720 }, { "epoch": 0.22996150844739763, "grad_norm": 1.16975998878479, "learning_rate": 1.7975754661197106e-05, "loss": 0.16448845863342285, "step": 8730 }, { "epoch": 0.23022492369189637, "grad_norm": 2.4568445682525635, "learning_rate": 1.797060547169927e-05, "loss": 0.1700390934944153, "step": 8740 }, { "epoch": 0.2304883389363951, "grad_norm": 2.2129790782928467, "learning_rate": 1.796545048085089e-05, "loss": 0.16340579986572265, "step": 8750 }, { "epoch": 0.23075175418089383, "grad_norm": 1.1669270992279053, "learning_rate": 1.796028969240399e-05, "loss": 0.1664592742919922, "step": 8760 }, { "epoch": 0.23101516942539257, "grad_norm": 1.5703670978546143, "learning_rate": 1.7955123110114815e-05, "loss": 0.16898078918457032, "step": 8770 }, { "epoch": 0.23127858466989132, "grad_norm": 1.199865460395813, "learning_rate": 1.794995073774383e-05, "loss": 0.1571760892868042, "step": 8780 }, { "epoch": 0.23154199991439003, "grad_norm": 0.7730661034584045, "learning_rate": 1.7944772579055705e-05, "loss": 0.175445294380188, "step": 8790 }, { "epoch": 0.23180541515888878, "grad_norm": 2.980748176574707, "learning_rate": 1.793958863781933e-05, "loss": 0.16276946067810058, "step": 8800 }, { "epoch": 0.23206883040338752, "grad_norm": 2.8583364486694336, "learning_rate": 1.7934398917807802e-05, "loss": 0.19856734275817872, "step": 8810 }, { "epoch": 0.23233224564788627, "grad_norm": 1.9120938777923584, "learning_rate": 1.7929203422798426e-05, "loss": 0.1656585454940796, "step": 8820 }, { "epoch": 0.23259566089238498, "grad_norm": 1.7188349962234497, "learning_rate": 1.79240021565727e-05, "loss": 0.160136079788208, "step": 8830 }, { "epoch": 0.23285907613688372, "grad_norm": 6.145048141479492, "learning_rate": 1.7918795122916337e-05, "loss": 0.16921738386154175, "step": 8840 }, { "epoch": 0.23312249138138247, "grad_norm": 1.335062861442566, "learning_rate": 1.7913582325619236e-05, "loss": 0.16983689069747926, "step": 8850 }, { "epoch": 0.2333859066258812, "grad_norm": 4.299654960632324, "learning_rate": 1.7908363768475498e-05, "loss": 0.17150342464447021, "step": 8860 }, { "epoch": 0.23364932187037996, "grad_norm": 2.4743857383728027, "learning_rate": 1.7903139455283412e-05, "loss": 0.1714864730834961, "step": 8870 }, { "epoch": 0.23391273711487867, "grad_norm": 4.783401966094971, "learning_rate": 1.7897909389845462e-05, "loss": 0.17152836322784423, "step": 8880 }, { "epoch": 0.23417615235937742, "grad_norm": 5.3615264892578125, "learning_rate": 1.789267357596831e-05, "loss": 0.16432632207870485, "step": 8890 }, { "epoch": 0.23443956760387616, "grad_norm": 0.9846181273460388, "learning_rate": 1.7887432017462812e-05, "loss": 0.17594892978668214, "step": 8900 }, { "epoch": 0.2347029828483749, "grad_norm": 5.416113376617432, "learning_rate": 1.7882184718144002e-05, "loss": 0.17631438970565796, "step": 8910 }, { "epoch": 0.23496639809287362, "grad_norm": 4.554300785064697, "learning_rate": 1.7876931681831087e-05, "loss": 0.17245500087738036, "step": 8920 }, { "epoch": 0.23522981333737236, "grad_norm": 1.81010103225708, "learning_rate": 1.787167291234745e-05, "loss": 0.16770720481872559, "step": 8930 }, { "epoch": 0.2354932285818711, "grad_norm": 6.710333824157715, "learning_rate": 1.7866408413520662e-05, "loss": 0.15937756299972533, "step": 8940 }, { "epoch": 0.23575664382636985, "grad_norm": 1.8284590244293213, "learning_rate": 1.786113818918244e-05, "loss": 0.1779787063598633, "step": 8950 }, { "epoch": 0.23602005907086857, "grad_norm": 3.3370988368988037, "learning_rate": 1.7855862243168684e-05, "loss": 0.16015002727508545, "step": 8960 }, { "epoch": 0.2362834743153673, "grad_norm": 2.181126117706299, "learning_rate": 1.785058057931946e-05, "loss": 0.17128994464874267, "step": 8970 }, { "epoch": 0.23654688955986605, "grad_norm": 3.6284303665161133, "learning_rate": 1.7845293201478985e-05, "loss": 0.18359588384628295, "step": 8980 }, { "epoch": 0.2368103048043648, "grad_norm": 1.7385653257369995, "learning_rate": 1.7840000113495647e-05, "loss": 0.16845088005065917, "step": 8990 }, { "epoch": 0.23707372004886354, "grad_norm": 5.79549503326416, "learning_rate": 1.7834701319221978e-05, "loss": 0.1703871250152588, "step": 9000 }, { "epoch": 0.23733713529336226, "grad_norm": 1.5731345415115356, "learning_rate": 1.782939682251467e-05, "loss": 0.17035086154937745, "step": 9010 }, { "epoch": 0.237600550537861, "grad_norm": 2.585512399673462, "learning_rate": 1.7824086627234565e-05, "loss": 0.16240739822387695, "step": 9020 }, { "epoch": 0.23786396578235974, "grad_norm": 4.492061614990234, "learning_rate": 1.7818770737246653e-05, "loss": 0.17434961795806886, "step": 9030 }, { "epoch": 0.2381273810268585, "grad_norm": 1.1835885047912598, "learning_rate": 1.7813449156420072e-05, "loss": 0.14742234945297242, "step": 9040 }, { "epoch": 0.2383907962713572, "grad_norm": 2.7473268508911133, "learning_rate": 1.7808121888628093e-05, "loss": 0.16984528303146362, "step": 9050 }, { "epoch": 0.23865421151585595, "grad_norm": 1.0305126905441284, "learning_rate": 1.780278893774813e-05, "loss": 0.16533594131469725, "step": 9060 }, { "epoch": 0.2389176267603547, "grad_norm": 2.001854419708252, "learning_rate": 1.7797450307661733e-05, "loss": 0.1655232787132263, "step": 9070 }, { "epoch": 0.23918104200485343, "grad_norm": 3.66428804397583, "learning_rate": 1.77921060022546e-05, "loss": 0.17150919437408446, "step": 9080 }, { "epoch": 0.23944445724935215, "grad_norm": 1.2165192365646362, "learning_rate": 1.7786756025416535e-05, "loss": 0.17276318073272706, "step": 9090 }, { "epoch": 0.2397078724938509, "grad_norm": 2.4863736629486084, "learning_rate": 1.7781400381041484e-05, "loss": 0.16713489294052125, "step": 9100 }, { "epoch": 0.23997128773834964, "grad_norm": 1.3731706142425537, "learning_rate": 1.777603907302752e-05, "loss": 0.16973283290863037, "step": 9110 }, { "epoch": 0.24023470298284838, "grad_norm": 1.395102858543396, "learning_rate": 1.7770672105276833e-05, "loss": 0.15100579261779784, "step": 9120 }, { "epoch": 0.24049811822734712, "grad_norm": 2.433039426803589, "learning_rate": 1.776529948169573e-05, "loss": 0.1689704179763794, "step": 9130 }, { "epoch": 0.24076153347184584, "grad_norm": 1.8142529726028442, "learning_rate": 1.775992120619464e-05, "loss": 0.17224065065383912, "step": 9140 }, { "epoch": 0.24102494871634458, "grad_norm": 1.05403470993042, "learning_rate": 1.7754537282688104e-05, "loss": 0.16173650026321412, "step": 9150 }, { "epoch": 0.24128836396084333, "grad_norm": 3.40757155418396, "learning_rate": 1.7749147715094777e-05, "loss": 0.1603859543800354, "step": 9160 }, { "epoch": 0.24155177920534207, "grad_norm": 1.6184163093566895, "learning_rate": 1.7743752507337414e-05, "loss": 0.18062272071838378, "step": 9170 }, { "epoch": 0.2418151944498408, "grad_norm": 3.407419443130493, "learning_rate": 1.7738351663342887e-05, "loss": 0.17785732746124266, "step": 9180 }, { "epoch": 0.24207860969433953, "grad_norm": 2.3222384452819824, "learning_rate": 1.7732945187042155e-05, "loss": 0.16393716335296632, "step": 9190 }, { "epoch": 0.24234202493883827, "grad_norm": 1.7515875101089478, "learning_rate": 1.7727533082370288e-05, "loss": 0.15402138233184814, "step": 9200 }, { "epoch": 0.24260544018333702, "grad_norm": 0.8800183534622192, "learning_rate": 1.772211535326645e-05, "loss": 0.18602280616760253, "step": 9210 }, { "epoch": 0.24286885542783573, "grad_norm": 2.4260220527648926, "learning_rate": 1.7716692003673898e-05, "loss": 0.15949110984802245, "step": 9220 }, { "epoch": 0.24313227067233448, "grad_norm": 2.710947036743164, "learning_rate": 1.7711263037539978e-05, "loss": 0.18011007308959961, "step": 9230 }, { "epoch": 0.24339568591683322, "grad_norm": 3.1459927558898926, "learning_rate": 1.770582845881613e-05, "loss": 0.17059041261672975, "step": 9240 }, { "epoch": 0.24365910116133196, "grad_norm": 2.000945806503296, "learning_rate": 1.7700388271457872e-05, "loss": 0.16891000270843506, "step": 9250 }, { "epoch": 0.2439225164058307, "grad_norm": 1.1562334299087524, "learning_rate": 1.7694942479424803e-05, "loss": 0.16891708374023437, "step": 9260 }, { "epoch": 0.24418593165032942, "grad_norm": 1.6990251541137695, "learning_rate": 1.768949108668061e-05, "loss": 0.15522934198379518, "step": 9270 }, { "epoch": 0.24444934689482817, "grad_norm": 1.0558345317840576, "learning_rate": 1.7684034097193055e-05, "loss": 0.1551721215248108, "step": 9280 }, { "epoch": 0.2447127621393269, "grad_norm": 4.623561382293701, "learning_rate": 1.7678571514933966e-05, "loss": 0.17181110382080078, "step": 9290 }, { "epoch": 0.24497617738382566, "grad_norm": 3.8408164978027344, "learning_rate": 1.7673103343879242e-05, "loss": 0.1591391921043396, "step": 9300 }, { "epoch": 0.24523959262832437, "grad_norm": 2.9450268745422363, "learning_rate": 1.7667629588008862e-05, "loss": 0.15912177562713622, "step": 9310 }, { "epoch": 0.24550300787282311, "grad_norm": 1.107922077178955, "learning_rate": 1.766215025130686e-05, "loss": 0.16112756729125977, "step": 9320 }, { "epoch": 0.24576642311732186, "grad_norm": 1.2271803617477417, "learning_rate": 1.7656665337761327e-05, "loss": 0.1650542736053467, "step": 9330 }, { "epoch": 0.2460298383618206, "grad_norm": 0.9114519953727722, "learning_rate": 1.7651174851364427e-05, "loss": 0.15941166877746582, "step": 9340 }, { "epoch": 0.24629325360631932, "grad_norm": 7.022656440734863, "learning_rate": 1.764567879611237e-05, "loss": 0.16794562339782715, "step": 9350 }, { "epoch": 0.24655666885081806, "grad_norm": 2.831185817718506, "learning_rate": 1.7640177176005425e-05, "loss": 0.1585421085357666, "step": 9360 }, { "epoch": 0.2468200840953168, "grad_norm": 1.3124788999557495, "learning_rate": 1.7634669995047904e-05, "loss": 0.17551641464233397, "step": 9370 }, { "epoch": 0.24708349933981555, "grad_norm": 1.6164741516113281, "learning_rate": 1.7629157257248175e-05, "loss": 0.15637083053588868, "step": 9380 }, { "epoch": 0.2473469145843143, "grad_norm": 1.3366557359695435, "learning_rate": 1.762363896661864e-05, "loss": 0.16157846450805663, "step": 9390 }, { "epoch": 0.247610329828813, "grad_norm": 4.435426235198975, "learning_rate": 1.761811512717576e-05, "loss": 0.16659053564071655, "step": 9400 }, { "epoch": 0.24787374507331175, "grad_norm": 1.4459891319274902, "learning_rate": 1.761258574294001e-05, "loss": 0.1485545516014099, "step": 9410 }, { "epoch": 0.2481371603178105, "grad_norm": 2.1477339267730713, "learning_rate": 1.760705081793593e-05, "loss": 0.16222636699676513, "step": 9420 }, { "epoch": 0.24840057556230924, "grad_norm": 3.602921485900879, "learning_rate": 1.760151035619206e-05, "loss": 0.1745921015739441, "step": 9430 }, { "epoch": 0.24866399080680796, "grad_norm": 1.1533336639404297, "learning_rate": 1.7595964361741e-05, "loss": 0.15866591930389404, "step": 9440 }, { "epoch": 0.2489274060513067, "grad_norm": 3.9899418354034424, "learning_rate": 1.759041283861936e-05, "loss": 0.1729160189628601, "step": 9450 }, { "epoch": 0.24919082129580544, "grad_norm": 0.7947150468826294, "learning_rate": 1.7584855790867773e-05, "loss": 0.1633564591407776, "step": 9460 }, { "epoch": 0.2494542365403042, "grad_norm": 1.3398782014846802, "learning_rate": 1.7579293222530907e-05, "loss": 0.16870534420013428, "step": 9470 }, { "epoch": 0.24971765178480293, "grad_norm": 1.3661308288574219, "learning_rate": 1.7573725137657432e-05, "loss": 0.16464576721191407, "step": 9480 }, { "epoch": 0.24998106702930165, "grad_norm": 3.386193037033081, "learning_rate": 1.756815154030005e-05, "loss": 0.1631955623626709, "step": 9490 }, { "epoch": 0.2502444822738004, "grad_norm": 0.8659093379974365, "learning_rate": 1.756257243451545e-05, "loss": 0.15487229824066162, "step": 9500 }, { "epoch": 0.2505078975182991, "grad_norm": 1.4883315563201904, "learning_rate": 1.7556987824364363e-05, "loss": 0.1678086042404175, "step": 9510 }, { "epoch": 0.25077131276279785, "grad_norm": 2.507408380508423, "learning_rate": 1.75513977139115e-05, "loss": 0.16780393123626708, "step": 9520 }, { "epoch": 0.2510347280072966, "grad_norm": 5.826882362365723, "learning_rate": 1.7545802107225585e-05, "loss": 0.173481547832489, "step": 9530 }, { "epoch": 0.25129814325179534, "grad_norm": 1.0698617696762085, "learning_rate": 1.754020100837934e-05, "loss": 0.1683891534805298, "step": 9540 }, { "epoch": 0.2515615584962941, "grad_norm": 1.240858793258667, "learning_rate": 1.753459442144949e-05, "loss": 0.1728266716003418, "step": 9550 }, { "epoch": 0.2518249737407928, "grad_norm": 4.923933506011963, "learning_rate": 1.7528982350516754e-05, "loss": 0.16356316804885865, "step": 9560 }, { "epoch": 0.25208838898529157, "grad_norm": 3.0306291580200195, "learning_rate": 1.7523364799665832e-05, "loss": 0.1596743106842041, "step": 9570 }, { "epoch": 0.2523518042297903, "grad_norm": 2.385200023651123, "learning_rate": 1.7517741772985424e-05, "loss": 0.17027523517608642, "step": 9580 }, { "epoch": 0.252615219474289, "grad_norm": 1.1493698358535767, "learning_rate": 1.751211327456821e-05, "loss": 0.16631211042404176, "step": 9590 }, { "epoch": 0.25287863471878774, "grad_norm": 5.781593322753906, "learning_rate": 1.7506479308510855e-05, "loss": 0.16614222526550293, "step": 9600 }, { "epoch": 0.2531420499632865, "grad_norm": 4.587834358215332, "learning_rate": 1.7500839878914004e-05, "loss": 0.16367944478988647, "step": 9610 }, { "epoch": 0.25340546520778523, "grad_norm": 2.466442346572876, "learning_rate": 1.749519498988227e-05, "loss": 0.17922860383987427, "step": 9620 }, { "epoch": 0.253668880452284, "grad_norm": 3.487891912460327, "learning_rate": 1.748954464552426e-05, "loss": 0.15931909084320067, "step": 9630 }, { "epoch": 0.2539322956967827, "grad_norm": 1.1308667659759521, "learning_rate": 1.7483888849952525e-05, "loss": 0.1745511770248413, "step": 9640 }, { "epoch": 0.25419571094128146, "grad_norm": 3.0129544734954834, "learning_rate": 1.7478227607283604e-05, "loss": 0.1561269998550415, "step": 9650 }, { "epoch": 0.2544591261857802, "grad_norm": 1.1312857866287231, "learning_rate": 1.7472560921637993e-05, "loss": 0.16497210264205933, "step": 9660 }, { "epoch": 0.25472254143027895, "grad_norm": 5.16348934173584, "learning_rate": 1.746688879714015e-05, "loss": 0.16489684581756592, "step": 9670 }, { "epoch": 0.25498595667477764, "grad_norm": 1.5051692724227905, "learning_rate": 1.746121123791849e-05, "loss": 0.16575284004211427, "step": 9680 }, { "epoch": 0.2552493719192764, "grad_norm": 5.362798690795898, "learning_rate": 1.7455528248105385e-05, "loss": 0.1795041561126709, "step": 9690 }, { "epoch": 0.2555127871637751, "grad_norm": 0.9900004267692566, "learning_rate": 1.7449839831837166e-05, "loss": 0.1732058644294739, "step": 9700 }, { "epoch": 0.25577620240827387, "grad_norm": 1.8600823879241943, "learning_rate": 1.7444145993254105e-05, "loss": 0.15079221725463868, "step": 9710 }, { "epoch": 0.2560396176527726, "grad_norm": 0.809789776802063, "learning_rate": 1.7438446736500418e-05, "loss": 0.17631235122680664, "step": 9720 }, { "epoch": 0.25630303289727135, "grad_norm": 3.413461685180664, "learning_rate": 1.7432742065724276e-05, "loss": 0.1671010971069336, "step": 9730 }, { "epoch": 0.2565664481417701, "grad_norm": 3.0549001693725586, "learning_rate": 1.7427031985077785e-05, "loss": 0.16985907554626464, "step": 9740 }, { "epoch": 0.25682986338626884, "grad_norm": 1.8445491790771484, "learning_rate": 1.7421316498716986e-05, "loss": 0.14904501438140869, "step": 9750 }, { "epoch": 0.2570932786307676, "grad_norm": 3.419191598892212, "learning_rate": 1.741559561080185e-05, "loss": 0.15429506301879883, "step": 9760 }, { "epoch": 0.2573566938752663, "grad_norm": 1.6862506866455078, "learning_rate": 1.74098693254963e-05, "loss": 0.13276591300964355, "step": 9770 }, { "epoch": 0.257620109119765, "grad_norm": 1.7907254695892334, "learning_rate": 1.7404137646968162e-05, "loss": 0.1659785270690918, "step": 9780 }, { "epoch": 0.25788352436426376, "grad_norm": 2.3701634407043457, "learning_rate": 1.7398400579389204e-05, "loss": 0.16141457557678224, "step": 9790 }, { "epoch": 0.2581469396087625, "grad_norm": 2.735910177230835, "learning_rate": 1.739265812693511e-05, "loss": 0.16952834129333497, "step": 9800 }, { "epoch": 0.25841035485326125, "grad_norm": 3.2619314193725586, "learning_rate": 1.7386910293785485e-05, "loss": 0.1617741107940674, "step": 9810 }, { "epoch": 0.25867377009776, "grad_norm": 3.2186970710754395, "learning_rate": 1.738115708412385e-05, "loss": 0.15411369800567626, "step": 9820 }, { "epoch": 0.25893718534225874, "grad_norm": 1.6634825468063354, "learning_rate": 1.737539850213764e-05, "loss": 0.16489121913909913, "step": 9830 }, { "epoch": 0.2592006005867575, "grad_norm": 1.5876173973083496, "learning_rate": 1.7369634552018196e-05, "loss": 0.15637811422348022, "step": 9840 }, { "epoch": 0.25946401583125617, "grad_norm": 2.2926511764526367, "learning_rate": 1.7363865237960773e-05, "loss": 0.15822126865386962, "step": 9850 }, { "epoch": 0.2597274310757549, "grad_norm": 2.7093505859375, "learning_rate": 1.7358090564164524e-05, "loss": 0.1568864941596985, "step": 9860 }, { "epoch": 0.25999084632025365, "grad_norm": 1.5934948921203613, "learning_rate": 1.735231053483251e-05, "loss": 0.16860791444778442, "step": 9870 }, { "epoch": 0.2602542615647524, "grad_norm": 1.0615029335021973, "learning_rate": 1.734652515417168e-05, "loss": 0.17914618253707887, "step": 9880 }, { "epoch": 0.26051767680925114, "grad_norm": 1.0519152879714966, "learning_rate": 1.7340734426392894e-05, "loss": 0.16382884979248047, "step": 9890 }, { "epoch": 0.2607810920537499, "grad_norm": 2.518009662628174, "learning_rate": 1.7334938355710876e-05, "loss": 0.17667146921157836, "step": 9900 }, { "epoch": 0.26104450729824863, "grad_norm": 2.316612482070923, "learning_rate": 1.7329136946344272e-05, "loss": 0.1700221300125122, "step": 9910 }, { "epoch": 0.2613079225427474, "grad_norm": 1.4516922235488892, "learning_rate": 1.7323330202515592e-05, "loss": 0.1601678252220154, "step": 9920 }, { "epoch": 0.2615713377872461, "grad_norm": 1.430235743522644, "learning_rate": 1.731751812845124e-05, "loss": 0.18406275510787964, "step": 9930 }, { "epoch": 0.2618347530317448, "grad_norm": 1.1660380363464355, "learning_rate": 1.7311700728381486e-05, "loss": 0.1555093765258789, "step": 9940 }, { "epoch": 0.26209816827624355, "grad_norm": 3.0793612003326416, "learning_rate": 1.7305878006540483e-05, "loss": 0.16744575500488282, "step": 9950 }, { "epoch": 0.2623615835207423, "grad_norm": 2.1287801265716553, "learning_rate": 1.7300049967166273e-05, "loss": 0.15922318696975707, "step": 9960 }, { "epoch": 0.26262499876524104, "grad_norm": 1.2566365003585815, "learning_rate": 1.729421661450074e-05, "loss": 0.14974489212036132, "step": 9970 }, { "epoch": 0.2628884140097398, "grad_norm": 1.5693022012710571, "learning_rate": 1.728837795278966e-05, "loss": 0.15741076469421386, "step": 9980 }, { "epoch": 0.2631518292542385, "grad_norm": 4.340803146362305, "learning_rate": 1.7282533986282656e-05, "loss": 0.17295827865600585, "step": 9990 }, { "epoch": 0.26341524449873727, "grad_norm": 9.762492179870605, "learning_rate": 1.7276684719233222e-05, "loss": 0.18023250102996827, "step": 10000 }, { "epoch": 0.263678659743236, "grad_norm": 2.7699785232543945, "learning_rate": 1.7270830155898704e-05, "loss": 0.18219728469848634, "step": 10010 }, { "epoch": 0.26394207498773475, "grad_norm": 1.6922520399093628, "learning_rate": 1.726497030054031e-05, "loss": 0.1665285587310791, "step": 10020 }, { "epoch": 0.26420549023223344, "grad_norm": 0.8887388706207275, "learning_rate": 1.72591051574231e-05, "loss": 0.16054270267486573, "step": 10030 }, { "epoch": 0.2644689054767322, "grad_norm": 2.054616928100586, "learning_rate": 1.7253234730815964e-05, "loss": 0.17909250259399415, "step": 10040 }, { "epoch": 0.26473232072123093, "grad_norm": 1.6924982070922852, "learning_rate": 1.724735902499166e-05, "loss": 0.1752676010131836, "step": 10050 }, { "epoch": 0.2649957359657297, "grad_norm": 0.8662326335906982, "learning_rate": 1.7241478044226782e-05, "loss": 0.1645512819290161, "step": 10060 }, { "epoch": 0.2652591512102284, "grad_norm": 1.3489165306091309, "learning_rate": 1.723559179280176e-05, "loss": 0.16222437620162963, "step": 10070 }, { "epoch": 0.26552256645472716, "grad_norm": 2.497847080230713, "learning_rate": 1.722970027500086e-05, "loss": 0.1803062915802002, "step": 10080 }, { "epoch": 0.2657859816992259, "grad_norm": 3.7857584953308105, "learning_rate": 1.7223803495112185e-05, "loss": 0.16481571197509765, "step": 10090 }, { "epoch": 0.26604939694372465, "grad_norm": 0.8247717618942261, "learning_rate": 1.7217901457427667e-05, "loss": 0.1541861414909363, "step": 10100 }, { "epoch": 0.26631281218822334, "grad_norm": 2.728693962097168, "learning_rate": 1.7211994166243058e-05, "loss": 0.16817734241485596, "step": 10110 }, { "epoch": 0.2665762274327221, "grad_norm": 1.538205623626709, "learning_rate": 1.7206081625857943e-05, "loss": 0.17027392387390136, "step": 10120 }, { "epoch": 0.2668396426772208, "grad_norm": 3.0102698802948, "learning_rate": 1.7200163840575728e-05, "loss": 0.16805872917175294, "step": 10130 }, { "epoch": 0.26710305792171957, "grad_norm": 3.4441022872924805, "learning_rate": 1.7194240814703626e-05, "loss": 0.16949177980422975, "step": 10140 }, { "epoch": 0.2673664731662183, "grad_norm": 1.4483190774917603, "learning_rate": 1.7188312552552672e-05, "loss": 0.16780569553375244, "step": 10150 }, { "epoch": 0.26762988841071705, "grad_norm": 3.4730045795440674, "learning_rate": 1.718237905843771e-05, "loss": 0.16562325954437257, "step": 10160 }, { "epoch": 0.2678933036552158, "grad_norm": 2.4625236988067627, "learning_rate": 1.71764403366774e-05, "loss": 0.1685797691345215, "step": 10170 }, { "epoch": 0.26815671889971454, "grad_norm": 3.6230123043060303, "learning_rate": 1.7170496391594194e-05, "loss": 0.16283204555511474, "step": 10180 }, { "epoch": 0.2684201341442133, "grad_norm": 1.9572492837905884, "learning_rate": 1.716454722751435e-05, "loss": 0.1619403600692749, "step": 10190 }, { "epoch": 0.268683549388712, "grad_norm": 1.90475332736969, "learning_rate": 1.715859284876793e-05, "loss": 0.16416118144989014, "step": 10200 }, { "epoch": 0.2689469646332107, "grad_norm": 0.6766147017478943, "learning_rate": 1.7152633259688788e-05, "loss": 0.16499967575073243, "step": 10210 }, { "epoch": 0.26921037987770946, "grad_norm": 5.512386798858643, "learning_rate": 1.714666846461457e-05, "loss": 0.16085669994354249, "step": 10220 }, { "epoch": 0.2694737951222082, "grad_norm": 1.2205923795700073, "learning_rate": 1.7140698467886707e-05, "loss": 0.15756011009216309, "step": 10230 }, { "epoch": 0.26973721036670695, "grad_norm": 1.427659034729004, "learning_rate": 1.713472327385042e-05, "loss": 0.1628410816192627, "step": 10240 }, { "epoch": 0.2700006256112057, "grad_norm": 2.3802313804626465, "learning_rate": 1.7128742886854723e-05, "loss": 0.15495188236236573, "step": 10250 }, { "epoch": 0.27026404085570443, "grad_norm": 2.745756149291992, "learning_rate": 1.712275731125239e-05, "loss": 0.17041118144989015, "step": 10260 }, { "epoch": 0.2705274561002032, "grad_norm": 7.105286121368408, "learning_rate": 1.7116766551399986e-05, "loss": 0.1616368293762207, "step": 10270 }, { "epoch": 0.2707908713447019, "grad_norm": 1.712129831314087, "learning_rate": 1.711077061165784e-05, "loss": 0.16302791833877564, "step": 10280 }, { "epoch": 0.2710542865892006, "grad_norm": 4.115485668182373, "learning_rate": 1.710476949639006e-05, "loss": 0.17180657386779785, "step": 10290 }, { "epoch": 0.27131770183369935, "grad_norm": 2.6162567138671875, "learning_rate": 1.7098763209964514e-05, "loss": 0.15625433921813964, "step": 10300 }, { "epoch": 0.2715811170781981, "grad_norm": 1.547921061515808, "learning_rate": 1.7092751756752838e-05, "loss": 0.15650256872177123, "step": 10310 }, { "epoch": 0.27184453232269684, "grad_norm": 1.7093976736068726, "learning_rate": 1.7086735141130423e-05, "loss": 0.16822359561920167, "step": 10320 }, { "epoch": 0.2721079475671956, "grad_norm": 2.9611024856567383, "learning_rate": 1.7080713367476428e-05, "loss": 0.16597148180007934, "step": 10330 }, { "epoch": 0.27237136281169433, "grad_norm": 0.9416398406028748, "learning_rate": 1.7074686440173752e-05, "loss": 0.17039273977279662, "step": 10340 }, { "epoch": 0.2726347780561931, "grad_norm": 1.8727692365646362, "learning_rate": 1.7068654363609058e-05, "loss": 0.1564027786254883, "step": 10350 }, { "epoch": 0.2728981933006918, "grad_norm": 3.1324610710144043, "learning_rate": 1.7062617142172754e-05, "loss": 0.17431529760360717, "step": 10360 }, { "epoch": 0.2731616085451905, "grad_norm": 2.489434242248535, "learning_rate": 1.7056574780258986e-05, "loss": 0.16800279617309571, "step": 10370 }, { "epoch": 0.27342502378968925, "grad_norm": 2.4953627586364746, "learning_rate": 1.705052728226564e-05, "loss": 0.15717192888259887, "step": 10380 }, { "epoch": 0.273688439034188, "grad_norm": 1.3526575565338135, "learning_rate": 1.704447465259436e-05, "loss": 0.15778739452362062, "step": 10390 }, { "epoch": 0.27395185427868674, "grad_norm": 2.915863275527954, "learning_rate": 1.7038416895650504e-05, "loss": 0.16832199096679687, "step": 10400 }, { "epoch": 0.2742152695231855, "grad_norm": 2.255072832107544, "learning_rate": 1.7032354015843162e-05, "loss": 0.16737570762634277, "step": 10410 }, { "epoch": 0.2744786847676842, "grad_norm": 5.569680690765381, "learning_rate": 1.7026286017585168e-05, "loss": 0.15581192970275878, "step": 10420 }, { "epoch": 0.27474210001218297, "grad_norm": 1.8782156705856323, "learning_rate": 1.7020212905293072e-05, "loss": 0.1659844398498535, "step": 10430 }, { "epoch": 0.2750055152566817, "grad_norm": 1.276254415512085, "learning_rate": 1.7014134683387147e-05, "loss": 0.17057547569274903, "step": 10440 }, { "epoch": 0.27526893050118045, "grad_norm": 1.4099823236465454, "learning_rate": 1.7008051356291382e-05, "loss": 0.15595911741256713, "step": 10450 }, { "epoch": 0.27553234574567914, "grad_norm": 1.6229416131973267, "learning_rate": 1.7001962928433488e-05, "loss": 0.1722460389137268, "step": 10460 }, { "epoch": 0.2757957609901779, "grad_norm": 3.2423243522644043, "learning_rate": 1.699586940424488e-05, "loss": 0.18192787170410157, "step": 10470 }, { "epoch": 0.27605917623467663, "grad_norm": 1.0879122018814087, "learning_rate": 1.6989770788160697e-05, "loss": 0.17398216724395751, "step": 10480 }, { "epoch": 0.2763225914791754, "grad_norm": 4.334569931030273, "learning_rate": 1.6983667084619768e-05, "loss": 0.1729522705078125, "step": 10490 }, { "epoch": 0.2765860067236741, "grad_norm": 2.9098877906799316, "learning_rate": 1.6977558298064636e-05, "loss": 0.16542277336120606, "step": 10500 }, { "epoch": 0.27684942196817286, "grad_norm": 2.8276844024658203, "learning_rate": 1.6971444432941533e-05, "loss": 0.1629910945892334, "step": 10510 }, { "epoch": 0.2771128372126716, "grad_norm": 1.5926227569580078, "learning_rate": 1.69653254937004e-05, "loss": 0.16571733951568604, "step": 10520 }, { "epoch": 0.27737625245717035, "grad_norm": 1.5957578420639038, "learning_rate": 1.6959201484794867e-05, "loss": 0.16562641859054567, "step": 10530 }, { "epoch": 0.2776396677016691, "grad_norm": 1.7257236242294312, "learning_rate": 1.6953072410682243e-05, "loss": 0.1763966679573059, "step": 10540 }, { "epoch": 0.2779030829461678, "grad_norm": 3.911327362060547, "learning_rate": 1.694693827582354e-05, "loss": 0.173819899559021, "step": 10550 }, { "epoch": 0.2781664981906665, "grad_norm": 2.3413383960723877, "learning_rate": 1.694079908468344e-05, "loss": 0.16208033561706542, "step": 10560 }, { "epoch": 0.27842991343516527, "grad_norm": 2.207401752471924, "learning_rate": 1.6934654841730318e-05, "loss": 0.16967480182647704, "step": 10570 }, { "epoch": 0.278693328679664, "grad_norm": 1.7805358171463013, "learning_rate": 1.6928505551436215e-05, "loss": 0.16946470737457275, "step": 10580 }, { "epoch": 0.27895674392416275, "grad_norm": 2.95813250541687, "learning_rate": 1.6922351218276852e-05, "loss": 0.16774113178253175, "step": 10590 }, { "epoch": 0.2792201591686615, "grad_norm": 3.605607748031616, "learning_rate": 1.6916191846731615e-05, "loss": 0.15588322877883912, "step": 10600 }, { "epoch": 0.27948357441316024, "grad_norm": 3.011329174041748, "learning_rate": 1.6910027441283562e-05, "loss": 0.1731510877609253, "step": 10610 }, { "epoch": 0.279746989657659, "grad_norm": 2.474174976348877, "learning_rate": 1.6903858006419413e-05, "loss": 0.17992446422576905, "step": 10620 }, { "epoch": 0.2800104049021577, "grad_norm": 2.0365681648254395, "learning_rate": 1.689768354662955e-05, "loss": 0.17030186653137208, "step": 10630 }, { "epoch": 0.2802738201466564, "grad_norm": 2.7914915084838867, "learning_rate": 1.689150406640801e-05, "loss": 0.16537280082702638, "step": 10640 }, { "epoch": 0.28053723539115516, "grad_norm": 1.3176796436309814, "learning_rate": 1.6885319570252485e-05, "loss": 0.16637375354766845, "step": 10650 }, { "epoch": 0.2808006506356539, "grad_norm": 2.899141550064087, "learning_rate": 1.687913006266432e-05, "loss": 0.16394987106323242, "step": 10660 }, { "epoch": 0.28106406588015265, "grad_norm": 2.6074812412261963, "learning_rate": 1.6872935548148503e-05, "loss": 0.16957148313522338, "step": 10670 }, { "epoch": 0.2813274811246514, "grad_norm": 3.8935329914093018, "learning_rate": 1.686673603121367e-05, "loss": 0.16366980075836182, "step": 10680 }, { "epoch": 0.28159089636915013, "grad_norm": 4.04357385635376, "learning_rate": 1.6860531516372102e-05, "loss": 0.1762535333633423, "step": 10690 }, { "epoch": 0.2818543116136489, "grad_norm": 1.647749423980713, "learning_rate": 1.6854322008139704e-05, "loss": 0.1602853298187256, "step": 10700 }, { "epoch": 0.2821177268581476, "grad_norm": 0.7663295269012451, "learning_rate": 1.6848107511036026e-05, "loss": 0.1649836540222168, "step": 10710 }, { "epoch": 0.2823811421026463, "grad_norm": 2.3689827919006348, "learning_rate": 1.6841888029584255e-05, "loss": 0.14791914224624633, "step": 10720 }, { "epoch": 0.28264455734714505, "grad_norm": 1.8727763891220093, "learning_rate": 1.6835663568311187e-05, "loss": 0.14848747253417968, "step": 10730 }, { "epoch": 0.2829079725916438, "grad_norm": 1.6879758834838867, "learning_rate": 1.682943413174726e-05, "loss": 0.16914057731628418, "step": 10740 }, { "epoch": 0.28317138783614254, "grad_norm": 1.5439136028289795, "learning_rate": 1.6823199724426523e-05, "loss": 0.174540114402771, "step": 10750 }, { "epoch": 0.2834348030806413, "grad_norm": 1.5240633487701416, "learning_rate": 1.6816960350886653e-05, "loss": 0.1761704444885254, "step": 10760 }, { "epoch": 0.28369821832514003, "grad_norm": 3.4023447036743164, "learning_rate": 1.681071601566893e-05, "loss": 0.1761087417602539, "step": 10770 }, { "epoch": 0.28396163356963877, "grad_norm": 1.107810616493225, "learning_rate": 1.6804466723318248e-05, "loss": 0.1709235429763794, "step": 10780 }, { "epoch": 0.2842250488141375, "grad_norm": 3.6266820430755615, "learning_rate": 1.6798212478383118e-05, "loss": 0.16116355657577514, "step": 10790 }, { "epoch": 0.28448846405863626, "grad_norm": 3.2287237644195557, "learning_rate": 1.6791953285415645e-05, "loss": 0.16479766368865967, "step": 10800 }, { "epoch": 0.28475187930313495, "grad_norm": 1.960256576538086, "learning_rate": 1.678568914897154e-05, "loss": 0.16838881969451905, "step": 10810 }, { "epoch": 0.2850152945476337, "grad_norm": 5.585461139678955, "learning_rate": 1.6779420073610116e-05, "loss": 0.1656888723373413, "step": 10820 }, { "epoch": 0.28527870979213243, "grad_norm": 6.163236141204834, "learning_rate": 1.6773146063894273e-05, "loss": 0.18085355758666993, "step": 10830 }, { "epoch": 0.2855421250366312, "grad_norm": 5.598367691040039, "learning_rate": 1.6766867124390505e-05, "loss": 0.16364619731903077, "step": 10840 }, { "epoch": 0.2858055402811299, "grad_norm": 3.8092877864837646, "learning_rate": 1.6760583259668897e-05, "loss": 0.15902485847473144, "step": 10850 }, { "epoch": 0.28606895552562867, "grad_norm": 1.7525676488876343, "learning_rate": 1.6754294474303116e-05, "loss": 0.16465868949890136, "step": 10860 }, { "epoch": 0.2863323707701274, "grad_norm": 0.7432495355606079, "learning_rate": 1.674800077287041e-05, "loss": 0.1704012632369995, "step": 10870 }, { "epoch": 0.28659578601462615, "grad_norm": 2.061554431915283, "learning_rate": 1.6741702159951613e-05, "loss": 0.17162511348724366, "step": 10880 }, { "epoch": 0.2868592012591249, "grad_norm": 1.8531501293182373, "learning_rate": 1.6735398640131117e-05, "loss": 0.16887344121932985, "step": 10890 }, { "epoch": 0.2871226165036236, "grad_norm": 1.4178473949432373, "learning_rate": 1.6729090217996904e-05, "loss": 0.17284703254699707, "step": 10900 }, { "epoch": 0.28738603174812233, "grad_norm": 1.882224202156067, "learning_rate": 1.6722776898140513e-05, "loss": 0.16319178342819213, "step": 10910 }, { "epoch": 0.28764944699262107, "grad_norm": 6.326511859893799, "learning_rate": 1.6716458685157047e-05, "loss": 0.17056385278701783, "step": 10920 }, { "epoch": 0.2879128622371198, "grad_norm": 2.8658244609832764, "learning_rate": 1.671013558364518e-05, "loss": 0.16641669273376464, "step": 10930 }, { "epoch": 0.28817627748161856, "grad_norm": 7.26046085357666, "learning_rate": 1.6703807598207132e-05, "loss": 0.16646103858947753, "step": 10940 }, { "epoch": 0.2884396927261173, "grad_norm": 1.1053900718688965, "learning_rate": 1.669747473344869e-05, "loss": 0.15505836009979249, "step": 10950 }, { "epoch": 0.28870310797061605, "grad_norm": 4.04313850402832, "learning_rate": 1.6691136993979183e-05, "loss": 0.1492897391319275, "step": 10960 }, { "epoch": 0.2889665232151148, "grad_norm": 1.8202848434448242, "learning_rate": 1.6684794384411494e-05, "loss": 0.15672047138214112, "step": 10970 }, { "epoch": 0.2892299384596135, "grad_norm": 1.5597196817398071, "learning_rate": 1.667844690936204e-05, "loss": 0.16649835109710692, "step": 10980 }, { "epoch": 0.2894933537041122, "grad_norm": 2.161638021469116, "learning_rate": 1.6672094573450798e-05, "loss": 0.18306628465652466, "step": 10990 }, { "epoch": 0.28975676894861097, "grad_norm": 1.8803339004516602, "learning_rate": 1.6665737381301263e-05, "loss": 0.17355082035064698, "step": 11000 }, { "epoch": 0.2900201841931097, "grad_norm": 1.1596728563308716, "learning_rate": 1.665937533754048e-05, "loss": 0.16120260953903198, "step": 11010 }, { "epoch": 0.29028359943760845, "grad_norm": 1.3311069011688232, "learning_rate": 1.6653008446799014e-05, "loss": 0.16630706787109376, "step": 11020 }, { "epoch": 0.2905470146821072, "grad_norm": 1.186118721961975, "learning_rate": 1.6646636713710967e-05, "loss": 0.15472831726074218, "step": 11030 }, { "epoch": 0.29081042992660594, "grad_norm": 2.6080033779144287, "learning_rate": 1.664026014291396e-05, "loss": 0.17354084253311158, "step": 11040 }, { "epoch": 0.2910738451711047, "grad_norm": 3.593088150024414, "learning_rate": 1.6633878739049135e-05, "loss": 0.16269590854644775, "step": 11050 }, { "epoch": 0.2913372604156034, "grad_norm": 2.7622008323669434, "learning_rate": 1.6627492506761155e-05, "loss": 0.1697235345840454, "step": 11060 }, { "epoch": 0.2916006756601021, "grad_norm": 1.1990017890930176, "learning_rate": 1.6621101450698193e-05, "loss": 0.1777369737625122, "step": 11070 }, { "epoch": 0.29186409090460086, "grad_norm": 4.056821346282959, "learning_rate": 1.6614705575511937e-05, "loss": 0.17348352670669556, "step": 11080 }, { "epoch": 0.2921275061490996, "grad_norm": 3.8929336071014404, "learning_rate": 1.660830488585758e-05, "loss": 0.1618555188179016, "step": 11090 }, { "epoch": 0.29239092139359835, "grad_norm": 1.7924915552139282, "learning_rate": 1.660189938639382e-05, "loss": 0.15748941898345947, "step": 11100 }, { "epoch": 0.2926543366380971, "grad_norm": 1.4881186485290527, "learning_rate": 1.6595489081782858e-05, "loss": 0.16878366470336914, "step": 11110 }, { "epoch": 0.29291775188259583, "grad_norm": 1.2647178173065186, "learning_rate": 1.6589073976690387e-05, "loss": 0.17016613483428955, "step": 11120 }, { "epoch": 0.2931811671270946, "grad_norm": 3.217775821685791, "learning_rate": 1.65826540757856e-05, "loss": 0.1668480157852173, "step": 11130 }, { "epoch": 0.2934445823715933, "grad_norm": 1.1251964569091797, "learning_rate": 1.6576229383741177e-05, "loss": 0.17272701263427734, "step": 11140 }, { "epoch": 0.29370799761609206, "grad_norm": 2.402942180633545, "learning_rate": 1.6569799905233285e-05, "loss": 0.1571812152862549, "step": 11150 }, { "epoch": 0.29397141286059075, "grad_norm": 2.032918691635132, "learning_rate": 1.6563365644941578e-05, "loss": 0.16231048107147217, "step": 11160 }, { "epoch": 0.2942348281050895, "grad_norm": 3.58001446723938, "learning_rate": 1.6556926607549186e-05, "loss": 0.15981104373931884, "step": 11170 }, { "epoch": 0.29449824334958824, "grad_norm": 3.853156805038452, "learning_rate": 1.655048279774272e-05, "loss": 0.1729426145553589, "step": 11180 }, { "epoch": 0.294761658594087, "grad_norm": 2.4938278198242188, "learning_rate": 1.6544034220212262e-05, "loss": 0.16601123809814453, "step": 11190 }, { "epoch": 0.2950250738385857, "grad_norm": 1.0947108268737793, "learning_rate": 1.6537580879651364e-05, "loss": 0.16185758113861085, "step": 11200 }, { "epoch": 0.29528848908308447, "grad_norm": 1.7148208618164062, "learning_rate": 1.6531122780757046e-05, "loss": 0.16983562707901, "step": 11210 }, { "epoch": 0.2955519043275832, "grad_norm": 4.049745559692383, "learning_rate": 1.6524659928229792e-05, "loss": 0.17423783540725707, "step": 11220 }, { "epoch": 0.29581531957208196, "grad_norm": 5.287744045257568, "learning_rate": 1.6518192326773546e-05, "loss": 0.17422239780426024, "step": 11230 }, { "epoch": 0.29607873481658065, "grad_norm": 3.261887550354004, "learning_rate": 1.6511719981095703e-05, "loss": 0.1795516014099121, "step": 11240 }, { "epoch": 0.2963421500610794, "grad_norm": 3.725733995437622, "learning_rate": 1.6505242895907122e-05, "loss": 0.1678837537765503, "step": 11250 }, { "epoch": 0.29660556530557813, "grad_norm": 2.457517147064209, "learning_rate": 1.64987610759221e-05, "loss": 0.16840299367904663, "step": 11260 }, { "epoch": 0.2968689805500769, "grad_norm": 1.942630410194397, "learning_rate": 1.6492274525858386e-05, "loss": 0.16761445999145508, "step": 11270 }, { "epoch": 0.2971323957945756, "grad_norm": 1.6081838607788086, "learning_rate": 1.648578325043717e-05, "loss": 0.16129491329193116, "step": 11280 }, { "epoch": 0.29739581103907436, "grad_norm": 0.8572188019752502, "learning_rate": 1.6479287254383084e-05, "loss": 0.17357850074768066, "step": 11290 }, { "epoch": 0.2976592262835731, "grad_norm": 2.4591262340545654, "learning_rate": 1.6472786542424196e-05, "loss": 0.14940792322158813, "step": 11300 }, { "epoch": 0.29792264152807185, "grad_norm": 2.7247278690338135, "learning_rate": 1.6466281119292004e-05, "loss": 0.17510435581207276, "step": 11310 }, { "epoch": 0.2981860567725706, "grad_norm": 2.0386106967926025, "learning_rate": 1.6459770989721433e-05, "loss": 0.1724250078201294, "step": 11320 }, { "epoch": 0.2984494720170693, "grad_norm": 3.5580849647521973, "learning_rate": 1.6453256158450835e-05, "loss": 0.16654244661331177, "step": 11330 }, { "epoch": 0.298712887261568, "grad_norm": 1.8895870447158813, "learning_rate": 1.644673663022199e-05, "loss": 0.17124334573745728, "step": 11340 }, { "epoch": 0.29897630250606677, "grad_norm": 2.6261038780212402, "learning_rate": 1.6440212409780088e-05, "loss": 0.1690125823020935, "step": 11350 }, { "epoch": 0.2992397177505655, "grad_norm": 1.0005329847335815, "learning_rate": 1.643368350187374e-05, "loss": 0.17663142681121827, "step": 11360 }, { "epoch": 0.29950313299506426, "grad_norm": 2.531780481338501, "learning_rate": 1.6427149911254973e-05, "loss": 0.159660005569458, "step": 11370 }, { "epoch": 0.299766548239563, "grad_norm": 1.5762965679168701, "learning_rate": 1.64206116426792e-05, "loss": 0.1542515277862549, "step": 11380 }, { "epoch": 0.30002996348406175, "grad_norm": 1.5223002433776855, "learning_rate": 1.641406870090527e-05, "loss": 0.16297547817230223, "step": 11390 }, { "epoch": 0.3002933787285605, "grad_norm": 2.0835111141204834, "learning_rate": 1.640752109069541e-05, "loss": 0.16794254779815673, "step": 11400 }, { "epoch": 0.30055679397305923, "grad_norm": 0.8624632954597473, "learning_rate": 1.640096881681526e-05, "loss": 0.16991864442825316, "step": 11410 }, { "epoch": 0.3008202092175579, "grad_norm": 24.65991973876953, "learning_rate": 1.6394411884033844e-05, "loss": 0.531645393371582, "step": 11420 }, { "epoch": 0.30108362446205666, "grad_norm": 8.889766693115234, "learning_rate": 1.638785029712358e-05, "loss": 1.2073392868041992, "step": 11430 }, { "epoch": 0.3013470397065554, "grad_norm": 3.556133270263672, "learning_rate": 1.638128406086027e-05, "loss": 0.19152204990386962, "step": 11440 }, { "epoch": 0.30161045495105415, "grad_norm": 1.396327018737793, "learning_rate": 1.6374713180023114e-05, "loss": 0.1693344831466675, "step": 11450 }, { "epoch": 0.3018738701955529, "grad_norm": 1.726304531097412, "learning_rate": 1.6368137659394673e-05, "loss": 0.1686856985092163, "step": 11460 }, { "epoch": 0.30213728544005164, "grad_norm": 2.531750440597534, "learning_rate": 1.63615575037609e-05, "loss": 0.18362590074539184, "step": 11470 }, { "epoch": 0.3024007006845504, "grad_norm": 4.111328601837158, "learning_rate": 1.6354972717911115e-05, "loss": 0.15158660411834718, "step": 11480 }, { "epoch": 0.3026641159290491, "grad_norm": 1.8728941679000854, "learning_rate": 1.634838330663801e-05, "loss": 0.16558566093444824, "step": 11490 }, { "epoch": 0.3029275311735478, "grad_norm": 1.3489431142807007, "learning_rate": 1.6341789274737644e-05, "loss": 0.16409695148468018, "step": 11500 }, { "epoch": 0.30319094641804656, "grad_norm": 2.6017377376556396, "learning_rate": 1.6335190627009432e-05, "loss": 0.18045250177383423, "step": 11510 }, { "epoch": 0.3034543616625453, "grad_norm": 6.208148002624512, "learning_rate": 1.632858736825616e-05, "loss": 0.16921770572662354, "step": 11520 }, { "epoch": 0.30371777690704405, "grad_norm": 1.773626685142517, "learning_rate": 1.632197950328396e-05, "loss": 0.16119561195373536, "step": 11530 }, { "epoch": 0.3039811921515428, "grad_norm": 3.407862901687622, "learning_rate": 1.631536703690233e-05, "loss": 0.1642417550086975, "step": 11540 }, { "epoch": 0.30424460739604153, "grad_norm": 1.6669870615005493, "learning_rate": 1.6308749973924096e-05, "loss": 0.17084741592407227, "step": 11550 }, { "epoch": 0.3045080226405403, "grad_norm": 1.3529757261276245, "learning_rate": 1.6302128319165454e-05, "loss": 0.17439889907836914, "step": 11560 }, { "epoch": 0.304771437885039, "grad_norm": 1.696756362915039, "learning_rate": 1.629550207744592e-05, "loss": 0.1739212155342102, "step": 11570 }, { "epoch": 0.30503485312953776, "grad_norm": 0.9522674679756165, "learning_rate": 1.6288871253588366e-05, "loss": 0.17025165557861327, "step": 11580 }, { "epoch": 0.30529826837403645, "grad_norm": 3.557136297225952, "learning_rate": 1.6282235852418983e-05, "loss": 0.17270635366439818, "step": 11590 }, { "epoch": 0.3055616836185352, "grad_norm": 1.65994393825531, "learning_rate": 1.6275595878767315e-05, "loss": 0.1684000611305237, "step": 11600 }, { "epoch": 0.30582509886303394, "grad_norm": 1.5503636598587036, "learning_rate": 1.6268951337466212e-05, "loss": 0.16444238424301147, "step": 11610 }, { "epoch": 0.3060885141075327, "grad_norm": 0.8927469849586487, "learning_rate": 1.626230223335186e-05, "loss": 0.17427680492401124, "step": 11620 }, { "epoch": 0.3063519293520314, "grad_norm": 1.396451473236084, "learning_rate": 1.6255648571263762e-05, "loss": 0.16955091953277587, "step": 11630 }, { "epoch": 0.30661534459653017, "grad_norm": 3.5563251972198486, "learning_rate": 1.6248990356044742e-05, "loss": 0.17594707012176514, "step": 11640 }, { "epoch": 0.3068787598410289, "grad_norm": 2.701951265335083, "learning_rate": 1.6242327592540942e-05, "loss": 0.15931928157806396, "step": 11650 }, { "epoch": 0.30714217508552766, "grad_norm": 1.70096755027771, "learning_rate": 1.62356602856018e-05, "loss": 0.16665834188461304, "step": 11660 }, { "epoch": 0.3074055903300264, "grad_norm": 1.3457056283950806, "learning_rate": 1.6228988440080068e-05, "loss": 0.1636185884475708, "step": 11670 }, { "epoch": 0.3076690055745251, "grad_norm": 3.5992562770843506, "learning_rate": 1.6222312060831814e-05, "loss": 0.1510030746459961, "step": 11680 }, { "epoch": 0.30793242081902383, "grad_norm": 3.434920310974121, "learning_rate": 1.6215631152716387e-05, "loss": 0.16966400146484376, "step": 11690 }, { "epoch": 0.3081958360635226, "grad_norm": 2.1868181228637695, "learning_rate": 1.620894572059644e-05, "loss": 0.16853713989257812, "step": 11700 }, { "epoch": 0.3084592513080213, "grad_norm": 1.648666262626648, "learning_rate": 1.6202255769337924e-05, "loss": 0.1489039897918701, "step": 11710 }, { "epoch": 0.30872266655252006, "grad_norm": 1.743798851966858, "learning_rate": 1.619556130381007e-05, "loss": 0.17196906805038453, "step": 11720 }, { "epoch": 0.3089860817970188, "grad_norm": 2.1497929096221924, "learning_rate": 1.61888623288854e-05, "loss": 0.17752912044525146, "step": 11730 }, { "epoch": 0.30924949704151755, "grad_norm": 3.405186653137207, "learning_rate": 1.6182158849439718e-05, "loss": 0.15658183097839357, "step": 11740 }, { "epoch": 0.3095129122860163, "grad_norm": 2.4676098823547363, "learning_rate": 1.61754508703521e-05, "loss": 0.1660766839981079, "step": 11750 }, { "epoch": 0.309776327530515, "grad_norm": 1.750555396080017, "learning_rate": 1.6168738396504912e-05, "loss": 0.17030371427536012, "step": 11760 }, { "epoch": 0.3100397427750137, "grad_norm": 4.5202412605285645, "learning_rate": 1.6162021432783773e-05, "loss": 0.1652486205101013, "step": 11770 }, { "epoch": 0.31030315801951247, "grad_norm": 2.6887130737304688, "learning_rate": 1.6155299984077584e-05, "loss": 0.1617575168609619, "step": 11780 }, { "epoch": 0.3105665732640112, "grad_norm": 1.4296220541000366, "learning_rate": 1.6148574055278507e-05, "loss": 0.1444973587989807, "step": 11790 }, { "epoch": 0.31082998850850996, "grad_norm": 1.208160400390625, "learning_rate": 1.6141843651281955e-05, "loss": 0.15928184986114502, "step": 11800 }, { "epoch": 0.3110934037530087, "grad_norm": 3.0110065937042236, "learning_rate": 1.6135108776986615e-05, "loss": 0.17220042943954467, "step": 11810 }, { "epoch": 0.31135681899750745, "grad_norm": 1.2829954624176025, "learning_rate": 1.612836943729442e-05, "loss": 0.14790618419647217, "step": 11820 }, { "epoch": 0.3116202342420062, "grad_norm": 1.6845296621322632, "learning_rate": 1.6121625637110546e-05, "loss": 0.17163844108581544, "step": 11830 }, { "epoch": 0.31188364948650493, "grad_norm": 1.9430111646652222, "learning_rate": 1.6114877381343422e-05, "loss": 0.1583188533782959, "step": 11840 }, { "epoch": 0.3121470647310036, "grad_norm": 1.4413580894470215, "learning_rate": 1.610812467490472e-05, "loss": 0.17790491580963136, "step": 11850 }, { "epoch": 0.31241047997550236, "grad_norm": 3.3095672130584717, "learning_rate": 1.6101367522709356e-05, "loss": 0.15778627395629882, "step": 11860 }, { "epoch": 0.3126738952200011, "grad_norm": 1.9229635000228882, "learning_rate": 1.6094605929675476e-05, "loss": 0.15313551425933838, "step": 11870 }, { "epoch": 0.31293731046449985, "grad_norm": 1.272693395614624, "learning_rate": 1.608783990072445e-05, "loss": 0.16470350027084352, "step": 11880 }, { "epoch": 0.3132007257089986, "grad_norm": 4.674856662750244, "learning_rate": 1.6081069440780895e-05, "loss": 0.16558074951171875, "step": 11890 }, { "epoch": 0.31346414095349734, "grad_norm": 2.3408267498016357, "learning_rate": 1.607429455477264e-05, "loss": 0.15257437229156495, "step": 11900 }, { "epoch": 0.3137275561979961, "grad_norm": 1.044440746307373, "learning_rate": 1.606751524763075e-05, "loss": 0.15175209045410157, "step": 11910 }, { "epoch": 0.3139909714424948, "grad_norm": 0.9212856292724609, "learning_rate": 1.606073152428948e-05, "loss": 0.1566387891769409, "step": 11920 }, { "epoch": 0.31425438668699357, "grad_norm": 2.4608242511749268, "learning_rate": 1.6053943389686323e-05, "loss": 0.15977518558502196, "step": 11930 }, { "epoch": 0.31451780193149226, "grad_norm": 2.045259952545166, "learning_rate": 1.604715084876198e-05, "loss": 0.1571756362915039, "step": 11940 }, { "epoch": 0.314781217175991, "grad_norm": 1.3288992643356323, "learning_rate": 1.6040353906460353e-05, "loss": 0.16874393224716186, "step": 11950 }, { "epoch": 0.31504463242048975, "grad_norm": 1.3705263137817383, "learning_rate": 1.6033552567728553e-05, "loss": 0.14605543613433838, "step": 11960 }, { "epoch": 0.3153080476649885, "grad_norm": 1.6168590784072876, "learning_rate": 1.602674683751688e-05, "loss": 0.171294903755188, "step": 11970 }, { "epoch": 0.31557146290948723, "grad_norm": 1.7866275310516357, "learning_rate": 1.6019936720778845e-05, "loss": 0.1661598801612854, "step": 11980 }, { "epoch": 0.315834878153986, "grad_norm": 1.3748222589492798, "learning_rate": 1.601312222247114e-05, "loss": 0.16002442836761474, "step": 11990 }, { "epoch": 0.3160982933984847, "grad_norm": 1.1955655813217163, "learning_rate": 1.6006303347553655e-05, "loss": 0.16933490037918092, "step": 12000 }, { "epoch": 0.31636170864298346, "grad_norm": 1.2006319761276245, "learning_rate": 1.5999480100989458e-05, "loss": 0.16893668174743653, "step": 12010 }, { "epoch": 0.31662512388748215, "grad_norm": 2.421680450439453, "learning_rate": 1.5992652487744806e-05, "loss": 0.1729865312576294, "step": 12020 }, { "epoch": 0.3168885391319809, "grad_norm": 0.7297250032424927, "learning_rate": 1.5985820512789128e-05, "loss": 0.15002074241638183, "step": 12030 }, { "epoch": 0.31715195437647964, "grad_norm": 2.3237831592559814, "learning_rate": 1.5978984181095034e-05, "loss": 0.16624855995178223, "step": 12040 }, { "epoch": 0.3174153696209784, "grad_norm": 1.3116823434829712, "learning_rate": 1.5972143497638294e-05, "loss": 0.16967558860778809, "step": 12050 }, { "epoch": 0.3176787848654771, "grad_norm": 1.4799073934555054, "learning_rate": 1.5965298467397855e-05, "loss": 0.16057286262512208, "step": 12060 }, { "epoch": 0.31794220010997587, "grad_norm": 1.3258254528045654, "learning_rate": 1.595844909535583e-05, "loss": 0.15433071851730346, "step": 12070 }, { "epoch": 0.3182056153544746, "grad_norm": 0.7848514914512634, "learning_rate": 1.595159538649748e-05, "loss": 0.1630272626876831, "step": 12080 }, { "epoch": 0.31846903059897336, "grad_norm": 3.991100549697876, "learning_rate": 1.5944737345811238e-05, "loss": 0.16619470119476318, "step": 12090 }, { "epoch": 0.3187324458434721, "grad_norm": 0.662146806716919, "learning_rate": 1.5937874978288676e-05, "loss": 0.15668749809265137, "step": 12100 }, { "epoch": 0.3189958610879708, "grad_norm": 1.9934775829315186, "learning_rate": 1.5931008288924523e-05, "loss": 0.15391333103179933, "step": 12110 }, { "epoch": 0.31925927633246953, "grad_norm": 1.2029038667678833, "learning_rate": 1.5924137282716653e-05, "loss": 0.15488913059234619, "step": 12120 }, { "epoch": 0.3195226915769683, "grad_norm": 1.5079964399337769, "learning_rate": 1.591726196466608e-05, "loss": 0.17353090047836303, "step": 12130 }, { "epoch": 0.319786106821467, "grad_norm": 0.9240995645523071, "learning_rate": 1.5910382339776953e-05, "loss": 0.15490736961364746, "step": 12140 }, { "epoch": 0.32004952206596576, "grad_norm": 2.8684613704681396, "learning_rate": 1.5903498413056568e-05, "loss": 0.15966206789016724, "step": 12150 }, { "epoch": 0.3203129373104645, "grad_norm": 2.295173168182373, "learning_rate": 1.5896610189515335e-05, "loss": 0.15905685424804689, "step": 12160 }, { "epoch": 0.32057635255496325, "grad_norm": 2.5690670013427734, "learning_rate": 1.588971767416681e-05, "loss": 0.16274576187133788, "step": 12170 }, { "epoch": 0.320839767799462, "grad_norm": 5.329929828643799, "learning_rate": 1.5882820872027654e-05, "loss": 0.1664417266845703, "step": 12180 }, { "epoch": 0.32110318304396074, "grad_norm": 1.0088609457015991, "learning_rate": 1.5875919788117662e-05, "loss": 0.16211652755737305, "step": 12190 }, { "epoch": 0.3213665982884594, "grad_norm": 0.9202170968055725, "learning_rate": 1.586901442745974e-05, "loss": 0.16409807205200194, "step": 12200 }, { "epoch": 0.32163001353295817, "grad_norm": 2.489638328552246, "learning_rate": 1.5862104795079903e-05, "loss": 0.15297367572784423, "step": 12210 }, { "epoch": 0.3218934287774569, "grad_norm": 1.247287631034851, "learning_rate": 1.5855190896007287e-05, "loss": 0.1784900665283203, "step": 12220 }, { "epoch": 0.32215684402195566, "grad_norm": 0.7396259307861328, "learning_rate": 1.584827273527412e-05, "loss": 0.14499177932739257, "step": 12230 }, { "epoch": 0.3224202592664544, "grad_norm": 3.9155311584472656, "learning_rate": 1.5841350317915745e-05, "loss": 0.164611554145813, "step": 12240 }, { "epoch": 0.32268367451095314, "grad_norm": 1.1213852167129517, "learning_rate": 1.5834423648970588e-05, "loss": 0.17299330234527588, "step": 12250 }, { "epoch": 0.3229470897554519, "grad_norm": 3.1068074703216553, "learning_rate": 1.5827492733480183e-05, "loss": 0.1731787919998169, "step": 12260 }, { "epoch": 0.32321050499995063, "grad_norm": 1.9104045629501343, "learning_rate": 1.5820557576489145e-05, "loss": 0.14513008594512938, "step": 12270 }, { "epoch": 0.3234739202444493, "grad_norm": 1.401347279548645, "learning_rate": 1.581361818304518e-05, "loss": 0.15652110576629638, "step": 12280 }, { "epoch": 0.32373733548894806, "grad_norm": 0.9282483458518982, "learning_rate": 1.5806674558199088e-05, "loss": 0.1724330186843872, "step": 12290 }, { "epoch": 0.3240007507334468, "grad_norm": 1.0573439598083496, "learning_rate": 1.5799726707004728e-05, "loss": 0.1680769681930542, "step": 12300 }, { "epoch": 0.32426416597794555, "grad_norm": 1.8326505422592163, "learning_rate": 1.5792774634519046e-05, "loss": 0.17123379707336425, "step": 12310 }, { "epoch": 0.3245275812224443, "grad_norm": 3.0849430561065674, "learning_rate": 1.5785818345802068e-05, "loss": 0.16264810562133789, "step": 12320 }, { "epoch": 0.32479099646694304, "grad_norm": 1.507127046585083, "learning_rate": 1.5778857845916874e-05, "loss": 0.16208704710006713, "step": 12330 }, { "epoch": 0.3250544117114418, "grad_norm": 1.6345640420913696, "learning_rate": 1.577189313992962e-05, "loss": 0.15421707630157472, "step": 12340 }, { "epoch": 0.3253178269559405, "grad_norm": 0.5575732588768005, "learning_rate": 1.5764924232909518e-05, "loss": 0.16689867973327638, "step": 12350 }, { "epoch": 0.32558124220043927, "grad_norm": 0.927710771560669, "learning_rate": 1.5757951129928837e-05, "loss": 0.15894936323165892, "step": 12360 }, { "epoch": 0.32584465744493796, "grad_norm": 0.61835116147995, "learning_rate": 1.5750973836062907e-05, "loss": 0.15562939643859863, "step": 12370 }, { "epoch": 0.3261080726894367, "grad_norm": 3.6078386306762695, "learning_rate": 1.57439923563901e-05, "loss": 0.1702427864074707, "step": 12380 }, { "epoch": 0.32637148793393544, "grad_norm": 2.057495594024658, "learning_rate": 1.573700669599184e-05, "loss": 0.15577244758605957, "step": 12390 }, { "epoch": 0.3266349031784342, "grad_norm": 1.1171272993087769, "learning_rate": 1.573001685995259e-05, "loss": 0.15392599105834961, "step": 12400 }, { "epoch": 0.32689831842293293, "grad_norm": 3.052410125732422, "learning_rate": 1.572302285335985e-05, "loss": 0.17259879112243653, "step": 12410 }, { "epoch": 0.3271617336674317, "grad_norm": 1.7858326435089111, "learning_rate": 1.5716024681304177e-05, "loss": 0.1557581067085266, "step": 12420 }, { "epoch": 0.3274251489119304, "grad_norm": 2.703058958053589, "learning_rate": 1.5709022348879124e-05, "loss": 0.16173291206359863, "step": 12430 }, { "epoch": 0.32768856415642916, "grad_norm": 1.5215518474578857, "learning_rate": 1.5702015861181295e-05, "loss": 0.15830638408660888, "step": 12440 }, { "epoch": 0.3279519794009279, "grad_norm": 4.6997480392456055, "learning_rate": 1.569500522331032e-05, "loss": 0.16842238903045653, "step": 12450 }, { "epoch": 0.3282153946454266, "grad_norm": 5.822527885437012, "learning_rate": 1.568799044036884e-05, "loss": 0.15744292736053467, "step": 12460 }, { "epoch": 0.32847880988992534, "grad_norm": 0.7652601003646851, "learning_rate": 1.5680971517462516e-05, "loss": 0.15028157234191894, "step": 12470 }, { "epoch": 0.3287422251344241, "grad_norm": 3.8224244117736816, "learning_rate": 1.5673948459700024e-05, "loss": 0.14940791130065917, "step": 12480 }, { "epoch": 0.3290056403789228, "grad_norm": 2.039870262145996, "learning_rate": 1.5666921272193046e-05, "loss": 0.15793116092681886, "step": 12490 }, { "epoch": 0.32926905562342157, "grad_norm": 2.004145860671997, "learning_rate": 1.565988996005628e-05, "loss": 0.16312990188598633, "step": 12500 }, { "epoch": 0.3295324708679203, "grad_norm": 1.6877480745315552, "learning_rate": 1.5652854528407407e-05, "loss": 0.15487008094787597, "step": 12510 }, { "epoch": 0.32979588611241906, "grad_norm": 1.6468505859375, "learning_rate": 1.5645814982367125e-05, "loss": 0.16241683959960937, "step": 12520 }, { "epoch": 0.3300593013569178, "grad_norm": 0.8714964985847473, "learning_rate": 1.5638771327059117e-05, "loss": 0.1655595302581787, "step": 12530 }, { "epoch": 0.3303227166014165, "grad_norm": 2.3351449966430664, "learning_rate": 1.563172356761006e-05, "loss": 0.1613940954208374, "step": 12540 }, { "epoch": 0.33058613184591523, "grad_norm": 0.978262186050415, "learning_rate": 1.5624671709149618e-05, "loss": 0.1525665044784546, "step": 12550 }, { "epoch": 0.330849547090414, "grad_norm": 1.9291038513183594, "learning_rate": 1.561761575681044e-05, "loss": 0.15273256301879884, "step": 12560 }, { "epoch": 0.3311129623349127, "grad_norm": 2.308380365371704, "learning_rate": 1.5610555715728146e-05, "loss": 0.1536446690559387, "step": 12570 }, { "epoch": 0.33137637757941146, "grad_norm": 2.7208399772644043, "learning_rate": 1.560349159104134e-05, "loss": 0.16356472969055175, "step": 12580 }, { "epoch": 0.3316397928239102, "grad_norm": 2.6498453617095947, "learning_rate": 1.55964233878916e-05, "loss": 0.16001975536346436, "step": 12590 }, { "epoch": 0.33190320806840895, "grad_norm": 0.9311385154724121, "learning_rate": 1.5589351111423474e-05, "loss": 0.1576171875, "step": 12600 }, { "epoch": 0.3321666233129077, "grad_norm": 2.105259656906128, "learning_rate": 1.5582274766784457e-05, "loss": 0.18024256229400634, "step": 12610 }, { "epoch": 0.33243003855740644, "grad_norm": 1.2579883337020874, "learning_rate": 1.557519435912503e-05, "loss": 0.14746205806732177, "step": 12620 }, { "epoch": 0.3326934538019051, "grad_norm": 1.3126137256622314, "learning_rate": 1.556810989359861e-05, "loss": 0.15832147598266602, "step": 12630 }, { "epoch": 0.33295686904640387, "grad_norm": 3.0909667015075684, "learning_rate": 1.5561021375361583e-05, "loss": 0.1657378554344177, "step": 12640 }, { "epoch": 0.3332202842909026, "grad_norm": 1.8154317140579224, "learning_rate": 1.5553928809573275e-05, "loss": 0.16000847816467284, "step": 12650 }, { "epoch": 0.33348369953540136, "grad_norm": 1.5213961601257324, "learning_rate": 1.5546832201395964e-05, "loss": 0.15937163829803466, "step": 12660 }, { "epoch": 0.3337471147799001, "grad_norm": 2.356102705001831, "learning_rate": 1.5539731555994864e-05, "loss": 0.15750577449798583, "step": 12670 }, { "epoch": 0.33401053002439884, "grad_norm": 1.6471339464187622, "learning_rate": 1.5532626878538137e-05, "loss": 0.15519886016845702, "step": 12680 }, { "epoch": 0.3342739452688976, "grad_norm": 3.1120364665985107, "learning_rate": 1.5525518174196867e-05, "loss": 0.15617221593856812, "step": 12690 }, { "epoch": 0.33453736051339633, "grad_norm": 2.259647846221924, "learning_rate": 1.551840544814508e-05, "loss": 0.14818639755249025, "step": 12700 }, { "epoch": 0.3348007757578951, "grad_norm": 1.5117875337600708, "learning_rate": 1.551128870555972e-05, "loss": 0.1619633674621582, "step": 12710 }, { "epoch": 0.33506419100239376, "grad_norm": 2.496619701385498, "learning_rate": 1.550416795162067e-05, "loss": 0.16561417579650878, "step": 12720 }, { "epoch": 0.3353276062468925, "grad_norm": 1.6728893518447876, "learning_rate": 1.549704319151071e-05, "loss": 0.16066070795059204, "step": 12730 }, { "epoch": 0.33559102149139125, "grad_norm": 1.2107070684432983, "learning_rate": 1.548991443041555e-05, "loss": 0.15927586555480958, "step": 12740 }, { "epoch": 0.33585443673589, "grad_norm": 3.2977702617645264, "learning_rate": 1.5482781673523816e-05, "loss": 0.1641996383666992, "step": 12750 }, { "epoch": 0.33611785198038874, "grad_norm": 2.3060953617095947, "learning_rate": 1.5475644926027035e-05, "loss": 0.147590970993042, "step": 12760 }, { "epoch": 0.3363812672248875, "grad_norm": 3.4531052112579346, "learning_rate": 1.5468504193119636e-05, "loss": 0.17946617603302, "step": 12770 }, { "epoch": 0.3366446824693862, "grad_norm": 3.353425979614258, "learning_rate": 1.5461359479998953e-05, "loss": 0.15674725770950318, "step": 12780 }, { "epoch": 0.33690809771388497, "grad_norm": 2.772580146789551, "learning_rate": 1.5454210791865213e-05, "loss": 0.16420307159423828, "step": 12790 }, { "epoch": 0.33717151295838366, "grad_norm": 2.2733805179595947, "learning_rate": 1.5447058133921552e-05, "loss": 0.17272318601608277, "step": 12800 }, { "epoch": 0.3374349282028824, "grad_norm": 2.2998299598693848, "learning_rate": 1.543990151137397e-05, "loss": 0.16033573150634767, "step": 12810 }, { "epoch": 0.33769834344738114, "grad_norm": 1.0490620136260986, "learning_rate": 1.5432740929431372e-05, "loss": 0.14263365268707276, "step": 12820 }, { "epoch": 0.3379617586918799, "grad_norm": 2.596773386001587, "learning_rate": 1.5425576393305537e-05, "loss": 0.15590591430664064, "step": 12830 }, { "epoch": 0.33822517393637863, "grad_norm": 1.3078886270523071, "learning_rate": 1.5418407908211125e-05, "loss": 0.15920485258102418, "step": 12840 }, { "epoch": 0.3384885891808774, "grad_norm": 2.6117193698883057, "learning_rate": 1.5411235479365664e-05, "loss": 0.15295491218566895, "step": 12850 }, { "epoch": 0.3387520044253761, "grad_norm": 1.9291168451309204, "learning_rate": 1.5404059111989565e-05, "loss": 0.1664342164993286, "step": 12860 }, { "epoch": 0.33901541966987486, "grad_norm": 2.172530174255371, "learning_rate": 1.5396878811306088e-05, "loss": 0.16144734621047974, "step": 12870 }, { "epoch": 0.3392788349143736, "grad_norm": 1.4492079019546509, "learning_rate": 1.5389694582541376e-05, "loss": 0.1681617259979248, "step": 12880 }, { "epoch": 0.3395422501588723, "grad_norm": 1.0379042625427246, "learning_rate": 1.5382506430924414e-05, "loss": 0.15808167457580566, "step": 12890 }, { "epoch": 0.33980566540337104, "grad_norm": 1.847463607788086, "learning_rate": 1.5375314361687053e-05, "loss": 0.16192831993103027, "step": 12900 }, { "epoch": 0.3400690806478698, "grad_norm": 3.7765843868255615, "learning_rate": 1.5368118380063984e-05, "loss": 0.16592605113983155, "step": 12910 }, { "epoch": 0.3403324958923685, "grad_norm": 2.988604784011841, "learning_rate": 1.536091849129276e-05, "loss": 0.1659018039703369, "step": 12920 }, { "epoch": 0.34059591113686727, "grad_norm": 1.1552904844284058, "learning_rate": 1.5353714700613768e-05, "loss": 0.161027991771698, "step": 12930 }, { "epoch": 0.340859326381366, "grad_norm": 4.0622358322143555, "learning_rate": 1.5346507013270237e-05, "loss": 0.15525089502334594, "step": 12940 }, { "epoch": 0.34112274162586476, "grad_norm": 4.207780838012695, "learning_rate": 1.5339295434508238e-05, "loss": 0.1510990858078003, "step": 12950 }, { "epoch": 0.3413861568703635, "grad_norm": 0.9811456799507141, "learning_rate": 1.533207996957666e-05, "loss": 0.15369040966033937, "step": 12960 }, { "epoch": 0.34164957211486224, "grad_norm": 1.6550929546356201, "learning_rate": 1.5324860623727236e-05, "loss": 0.16122424602508545, "step": 12970 }, { "epoch": 0.34191298735936093, "grad_norm": 3.1475532054901123, "learning_rate": 1.5317637402214515e-05, "loss": 0.15632253885269165, "step": 12980 }, { "epoch": 0.3421764026038597, "grad_norm": 1.577987790107727, "learning_rate": 1.5310410310295876e-05, "loss": 0.14626179933547973, "step": 12990 }, { "epoch": 0.3424398178483584, "grad_norm": 2.1335220336914062, "learning_rate": 1.5303179353231494e-05, "loss": 0.16377050876617433, "step": 13000 }, { "epoch": 0.34270323309285716, "grad_norm": 3.5071277618408203, "learning_rate": 1.529594453628439e-05, "loss": 0.1606379508972168, "step": 13010 }, { "epoch": 0.3429666483373559, "grad_norm": 2.400578022003174, "learning_rate": 1.5288705864720356e-05, "loss": 0.16394290924072266, "step": 13020 }, { "epoch": 0.34323006358185465, "grad_norm": 3.0037364959716797, "learning_rate": 1.528146334380803e-05, "loss": 0.18191850185394287, "step": 13030 }, { "epoch": 0.3434934788263534, "grad_norm": 2.274841547012329, "learning_rate": 1.5274216978818813e-05, "loss": 0.15699976682662964, "step": 13040 }, { "epoch": 0.34375689407085214, "grad_norm": 4.8331098556518555, "learning_rate": 1.526696677502693e-05, "loss": 0.1610248327255249, "step": 13050 }, { "epoch": 0.3440203093153508, "grad_norm": 3.2868635654449463, "learning_rate": 1.525971273770939e-05, "loss": 0.1579851031303406, "step": 13060 }, { "epoch": 0.34428372455984957, "grad_norm": 2.855036973953247, "learning_rate": 1.5252454872146003e-05, "loss": 0.15719873905181886, "step": 13070 }, { "epoch": 0.3445471398043483, "grad_norm": 2.7071337699890137, "learning_rate": 1.5245193183619347e-05, "loss": 0.1662978172302246, "step": 13080 }, { "epoch": 0.34481055504884706, "grad_norm": 0.8102446794509888, "learning_rate": 1.5237927677414797e-05, "loss": 0.15780177116394042, "step": 13090 }, { "epoch": 0.3450739702933458, "grad_norm": 0.9327265024185181, "learning_rate": 1.5230658358820493e-05, "loss": 0.15603781938552858, "step": 13100 }, { "epoch": 0.34533738553784454, "grad_norm": 2.9138035774230957, "learning_rate": 1.5223385233127372e-05, "loss": 0.15237600803375245, "step": 13110 }, { "epoch": 0.3456008007823433, "grad_norm": 3.549675464630127, "learning_rate": 1.5216108305629123e-05, "loss": 0.15693202018737792, "step": 13120 }, { "epoch": 0.34586421602684203, "grad_norm": 2.3692333698272705, "learning_rate": 1.5208827581622205e-05, "loss": 0.1800450325012207, "step": 13130 }, { "epoch": 0.3461276312713408, "grad_norm": 1.5167356729507446, "learning_rate": 1.5201543066405844e-05, "loss": 0.1518951177597046, "step": 13140 }, { "epoch": 0.34639104651583946, "grad_norm": 1.7540887594223022, "learning_rate": 1.5194254765282025e-05, "loss": 0.16819779872894286, "step": 13150 }, { "epoch": 0.3466544617603382, "grad_norm": 1.5094738006591797, "learning_rate": 1.5186962683555493e-05, "loss": 0.1589136838912964, "step": 13160 }, { "epoch": 0.34691787700483695, "grad_norm": 1.6537319421768188, "learning_rate": 1.517966682653373e-05, "loss": 0.14669014215469361, "step": 13170 }, { "epoch": 0.3471812922493357, "grad_norm": 3.9618449211120605, "learning_rate": 1.5172367199526985e-05, "loss": 0.15858371257781984, "step": 13180 }, { "epoch": 0.34744470749383444, "grad_norm": 3.2050368785858154, "learning_rate": 1.5165063807848234e-05, "loss": 0.17041635513305664, "step": 13190 }, { "epoch": 0.3477081227383332, "grad_norm": 3.9991378784179688, "learning_rate": 1.5157756656813207e-05, "loss": 0.17361006736755372, "step": 13200 }, { "epoch": 0.3479715379828319, "grad_norm": 2.7857234477996826, "learning_rate": 1.5150445751740356e-05, "loss": 0.15261584520339966, "step": 13210 }, { "epoch": 0.34823495322733067, "grad_norm": 1.6494743824005127, "learning_rate": 1.5143131097950882e-05, "loss": 0.15254027843475343, "step": 13220 }, { "epoch": 0.3484983684718294, "grad_norm": 2.7133917808532715, "learning_rate": 1.5135812700768702e-05, "loss": 0.1714639902114868, "step": 13230 }, { "epoch": 0.3487617837163281, "grad_norm": 1.1578922271728516, "learning_rate": 1.512849056552046e-05, "loss": 0.1510182499885559, "step": 13240 }, { "epoch": 0.34902519896082684, "grad_norm": 2.296895742416382, "learning_rate": 1.512116469753552e-05, "loss": 0.15925652980804444, "step": 13250 }, { "epoch": 0.3492886142053256, "grad_norm": 2.7513797283172607, "learning_rate": 1.5113835102145968e-05, "loss": 0.15751463174819946, "step": 13260 }, { "epoch": 0.34955202944982433, "grad_norm": 1.5912257432937622, "learning_rate": 1.51065017846866e-05, "loss": 0.15580229759216307, "step": 13270 }, { "epoch": 0.3498154446943231, "grad_norm": 1.5548375844955444, "learning_rate": 1.509916475049492e-05, "loss": 0.14984889030456544, "step": 13280 }, { "epoch": 0.3500788599388218, "grad_norm": 4.508754253387451, "learning_rate": 1.5091824004911137e-05, "loss": 0.1826821446418762, "step": 13290 }, { "epoch": 0.35034227518332056, "grad_norm": 4.373196601867676, "learning_rate": 1.5084479553278164e-05, "loss": 0.16000396013259888, "step": 13300 }, { "epoch": 0.3506056904278193, "grad_norm": 1.6869169473648071, "learning_rate": 1.5077131400941609e-05, "loss": 0.15466063022613524, "step": 13310 }, { "epoch": 0.350869105672318, "grad_norm": 0.8416905403137207, "learning_rate": 1.5069779553249775e-05, "loss": 0.15629178285598755, "step": 13320 }, { "epoch": 0.35113252091681674, "grad_norm": 2.554279327392578, "learning_rate": 1.5062424015553653e-05, "loss": 0.17580580711364746, "step": 13330 }, { "epoch": 0.3513959361613155, "grad_norm": 2.8681602478027344, "learning_rate": 1.505506479320692e-05, "loss": 0.14878604412078858, "step": 13340 }, { "epoch": 0.3516593514058142, "grad_norm": 2.240858793258667, "learning_rate": 1.5047701891565938e-05, "loss": 0.16265437602996827, "step": 13350 }, { "epoch": 0.35192276665031297, "grad_norm": 1.40277099609375, "learning_rate": 1.504033531598974e-05, "loss": 0.15359225273132324, "step": 13360 }, { "epoch": 0.3521861818948117, "grad_norm": 2.488462448120117, "learning_rate": 1.5032965071840045e-05, "loss": 0.16154942512512208, "step": 13370 }, { "epoch": 0.35244959713931046, "grad_norm": 1.0833241939544678, "learning_rate": 1.5025591164481226e-05, "loss": 0.17312953472137452, "step": 13380 }, { "epoch": 0.3527130123838092, "grad_norm": 2.9258158206939697, "learning_rate": 1.5018213599280339e-05, "loss": 0.16039738655090333, "step": 13390 }, { "epoch": 0.35297642762830794, "grad_norm": 1.6592719554901123, "learning_rate": 1.5010832381607085e-05, "loss": 0.15730655193328857, "step": 13400 }, { "epoch": 0.35323984287280663, "grad_norm": 4.314607620239258, "learning_rate": 1.500344751683384e-05, "loss": 0.16337258815765382, "step": 13410 }, { "epoch": 0.3535032581173054, "grad_norm": 4.974173069000244, "learning_rate": 1.4996059010335626e-05, "loss": 0.17047961950302123, "step": 13420 }, { "epoch": 0.3537666733618041, "grad_norm": 1.5248643159866333, "learning_rate": 1.4988666867490117e-05, "loss": 0.15116615295410157, "step": 13430 }, { "epoch": 0.35403008860630286, "grad_norm": 5.758083343505859, "learning_rate": 1.4981271093677632e-05, "loss": 0.1542057991027832, "step": 13440 }, { "epoch": 0.3542935038508016, "grad_norm": 3.8536837100982666, "learning_rate": 1.4973871694281133e-05, "loss": 0.16725125312805175, "step": 13450 }, { "epoch": 0.35455691909530035, "grad_norm": 2.350957155227661, "learning_rate": 1.4966468674686226e-05, "loss": 0.15110912322998046, "step": 13460 }, { "epoch": 0.3548203343397991, "grad_norm": 4.276898384094238, "learning_rate": 1.4959062040281143e-05, "loss": 0.1592295289039612, "step": 13470 }, { "epoch": 0.35508374958429784, "grad_norm": 1.5741161108016968, "learning_rate": 1.4951651796456758e-05, "loss": 0.16823420524597169, "step": 13480 }, { "epoch": 0.3553471648287966, "grad_norm": 1.1693323850631714, "learning_rate": 1.4944237948606561e-05, "loss": 0.15385782718658447, "step": 13490 }, { "epoch": 0.35561058007329527, "grad_norm": 1.1458889245986938, "learning_rate": 1.4936820502126676e-05, "loss": 0.1452336072921753, "step": 13500 }, { "epoch": 0.355873995317794, "grad_norm": 3.3649935722351074, "learning_rate": 1.4929399462415832e-05, "loss": 0.15587780475616456, "step": 13510 }, { "epoch": 0.35613741056229276, "grad_norm": 1.7445478439331055, "learning_rate": 1.492197483487539e-05, "loss": 0.15437772274017333, "step": 13520 }, { "epoch": 0.3564008258067915, "grad_norm": 2.96597957611084, "learning_rate": 1.491454662490931e-05, "loss": 0.16230967044830322, "step": 13530 }, { "epoch": 0.35666424105129024, "grad_norm": 1.4207254648208618, "learning_rate": 1.4907114837924168e-05, "loss": 0.15627477169036866, "step": 13540 }, { "epoch": 0.356927656295789, "grad_norm": 1.3539060354232788, "learning_rate": 1.4899679479329133e-05, "loss": 0.15171437263488768, "step": 13550 }, { "epoch": 0.35719107154028773, "grad_norm": 2.6191773414611816, "learning_rate": 1.4892240554535981e-05, "loss": 0.14973585605621337, "step": 13560 }, { "epoch": 0.3574544867847865, "grad_norm": 0.9476029276847839, "learning_rate": 1.4884798068959085e-05, "loss": 0.15274058580398558, "step": 13570 }, { "epoch": 0.3577179020292852, "grad_norm": 1.6126043796539307, "learning_rate": 1.4877352028015408e-05, "loss": 0.15787405967712403, "step": 13580 }, { "epoch": 0.3579813172737839, "grad_norm": 1.5129671096801758, "learning_rate": 1.4869902437124499e-05, "loss": 0.16451139450073243, "step": 13590 }, { "epoch": 0.35824473251828265, "grad_norm": 1.8411132097244263, "learning_rate": 1.486244930170849e-05, "loss": 0.16159818172454835, "step": 13600 }, { "epoch": 0.3585081477627814, "grad_norm": 2.2291460037231445, "learning_rate": 1.4854992627192096e-05, "loss": 0.16351745128631592, "step": 13610 }, { "epoch": 0.35877156300728014, "grad_norm": 3.2741968631744385, "learning_rate": 1.4847532419002602e-05, "loss": 0.15576688051223755, "step": 13620 }, { "epoch": 0.3590349782517789, "grad_norm": 0.8076201677322388, "learning_rate": 1.4840068682569877e-05, "loss": 0.1634470820426941, "step": 13630 }, { "epoch": 0.3592983934962776, "grad_norm": 3.0120785236358643, "learning_rate": 1.4832601423326346e-05, "loss": 0.13304557800292968, "step": 13640 }, { "epoch": 0.35956180874077637, "grad_norm": 1.2052834033966064, "learning_rate": 1.4825130646707002e-05, "loss": 0.13874657154083253, "step": 13650 }, { "epoch": 0.3598252239852751, "grad_norm": 3.113389492034912, "learning_rate": 1.4817656358149398e-05, "loss": 0.16459939479827881, "step": 13660 }, { "epoch": 0.3600886392297738, "grad_norm": 1.2737958431243896, "learning_rate": 1.4810178563093648e-05, "loss": 0.1681502342224121, "step": 13670 }, { "epoch": 0.36035205447427254, "grad_norm": 1.399552822113037, "learning_rate": 1.4802697266982411e-05, "loss": 0.17053868770599365, "step": 13680 }, { "epoch": 0.3606154697187713, "grad_norm": 3.2309372425079346, "learning_rate": 1.4795212475260896e-05, "loss": 0.16112104654312134, "step": 13690 }, { "epoch": 0.36087888496327003, "grad_norm": 1.2670496702194214, "learning_rate": 1.4787724193376858e-05, "loss": 0.1553349494934082, "step": 13700 }, { "epoch": 0.3611423002077688, "grad_norm": 3.679805278778076, "learning_rate": 1.4780232426780592e-05, "loss": 0.1580265760421753, "step": 13710 }, { "epoch": 0.3614057154522675, "grad_norm": 1.2814724445343018, "learning_rate": 1.477273718092493e-05, "loss": 0.164237380027771, "step": 13720 }, { "epoch": 0.36166913069676626, "grad_norm": 1.3968729972839355, "learning_rate": 1.4765238461265235e-05, "loss": 0.16158157587051392, "step": 13730 }, { "epoch": 0.361932545941265, "grad_norm": 3.002592086791992, "learning_rate": 1.4757736273259396e-05, "loss": 0.1704047679901123, "step": 13740 }, { "epoch": 0.36219596118576375, "grad_norm": 1.7589398622512817, "learning_rate": 1.4750230622367831e-05, "loss": 0.1517457365989685, "step": 13750 }, { "epoch": 0.36245937643026244, "grad_norm": 5.170746326446533, "learning_rate": 1.474272151405348e-05, "loss": 0.15843842029571534, "step": 13760 }, { "epoch": 0.3627227916747612, "grad_norm": 1.6912128925323486, "learning_rate": 1.473520895378179e-05, "loss": 0.16373400688171386, "step": 13770 }, { "epoch": 0.3629862069192599, "grad_norm": 2.489180564880371, "learning_rate": 1.472769294702073e-05, "loss": 0.17241696119308472, "step": 13780 }, { "epoch": 0.36324962216375867, "grad_norm": 5.205865383148193, "learning_rate": 1.4720173499240771e-05, "loss": 0.1604029655456543, "step": 13790 }, { "epoch": 0.3635130374082574, "grad_norm": 2.9069058895111084, "learning_rate": 1.4712650615914899e-05, "loss": 0.1615748643875122, "step": 13800 }, { "epoch": 0.36377645265275615, "grad_norm": 3.2219998836517334, "learning_rate": 1.4705124302518581e-05, "loss": 0.17289801836013793, "step": 13810 }, { "epoch": 0.3640398678972549, "grad_norm": 3.437138319015503, "learning_rate": 1.4697594564529805e-05, "loss": 0.1674578905105591, "step": 13820 }, { "epoch": 0.36430328314175364, "grad_norm": 1.4291118383407593, "learning_rate": 1.4690061407429028e-05, "loss": 0.15147583484649657, "step": 13830 }, { "epoch": 0.3645666983862524, "grad_norm": 1.230968952178955, "learning_rate": 1.468252483669922e-05, "loss": 0.16596583127975464, "step": 13840 }, { "epoch": 0.3648301136307511, "grad_norm": 1.8615069389343262, "learning_rate": 1.4674984857825808e-05, "loss": 0.17266263961791992, "step": 13850 }, { "epoch": 0.3650935288752498, "grad_norm": 2.8067967891693115, "learning_rate": 1.4667441476296723e-05, "loss": 0.16018860340118407, "step": 13860 }, { "epoch": 0.36535694411974856, "grad_norm": 1.4214255809783936, "learning_rate": 1.4659894697602358e-05, "loss": 0.15826835632324218, "step": 13870 }, { "epoch": 0.3656203593642473, "grad_norm": 3.3747196197509766, "learning_rate": 1.465234452723559e-05, "loss": 0.15639755725860596, "step": 13880 }, { "epoch": 0.36588377460874605, "grad_norm": 2.4673943519592285, "learning_rate": 1.4644790970691755e-05, "loss": 0.1479099750518799, "step": 13890 }, { "epoch": 0.3661471898532448, "grad_norm": 2.6965322494506836, "learning_rate": 1.4637234033468658e-05, "loss": 0.1569599151611328, "step": 13900 }, { "epoch": 0.36641060509774354, "grad_norm": 0.862841784954071, "learning_rate": 1.4629673721066566e-05, "loss": 0.1578845977783203, "step": 13910 }, { "epoch": 0.3666740203422423, "grad_norm": 3.6370809078216553, "learning_rate": 1.4622110038988196e-05, "loss": 0.16477272510528565, "step": 13920 }, { "epoch": 0.36693743558674097, "grad_norm": 1.1513028144836426, "learning_rate": 1.4614542992738727e-05, "loss": 0.16894347667694093, "step": 13930 }, { "epoch": 0.3672008508312397, "grad_norm": 0.7810608148574829, "learning_rate": 1.4606972587825779e-05, "loss": 0.16512428522109984, "step": 13940 }, { "epoch": 0.36746426607573845, "grad_norm": 2.781390905380249, "learning_rate": 1.4599398829759419e-05, "loss": 0.15790910720825196, "step": 13950 }, { "epoch": 0.3677276813202372, "grad_norm": 1.6093757152557373, "learning_rate": 1.4591821724052155e-05, "loss": 0.1530538320541382, "step": 13960 }, { "epoch": 0.36799109656473594, "grad_norm": 1.363867998123169, "learning_rate": 1.4584241276218935e-05, "loss": 0.1704336404800415, "step": 13970 }, { "epoch": 0.3682545118092347, "grad_norm": 0.9817802309989929, "learning_rate": 1.4576657491777134e-05, "loss": 0.16021158695220947, "step": 13980 }, { "epoch": 0.36851792705373343, "grad_norm": 1.959911584854126, "learning_rate": 1.4569070376246552e-05, "loss": 0.163731849193573, "step": 13990 }, { "epoch": 0.3687813422982322, "grad_norm": 0.9785584211349487, "learning_rate": 1.4561479935149426e-05, "loss": 0.15927039384841918, "step": 14000 }, { "epoch": 0.3690447575427309, "grad_norm": 2.0504658222198486, "learning_rate": 1.4553886174010405e-05, "loss": 0.17570149898529053, "step": 14010 }, { "epoch": 0.3693081727872296, "grad_norm": 0.666264533996582, "learning_rate": 1.4546289098356556e-05, "loss": 0.14760524034500122, "step": 14020 }, { "epoch": 0.36957158803172835, "grad_norm": 1.0558174848556519, "learning_rate": 1.4538688713717357e-05, "loss": 0.14737305641174317, "step": 14030 }, { "epoch": 0.3698350032762271, "grad_norm": 4.759511470794678, "learning_rate": 1.4531085025624694e-05, "loss": 0.16379616260528565, "step": 14040 }, { "epoch": 0.37009841852072584, "grad_norm": 1.2178043127059937, "learning_rate": 1.4523478039612863e-05, "loss": 0.17567323446273803, "step": 14050 }, { "epoch": 0.3703618337652246, "grad_norm": 1.9211039543151855, "learning_rate": 1.4515867761218558e-05, "loss": 0.16012808084487914, "step": 14060 }, { "epoch": 0.3706252490097233, "grad_norm": 2.115614891052246, "learning_rate": 1.4508254195980861e-05, "loss": 0.15323616266250611, "step": 14070 }, { "epoch": 0.37088866425422207, "grad_norm": 1.3816845417022705, "learning_rate": 1.450063734944126e-05, "loss": 0.16782057285308838, "step": 14080 }, { "epoch": 0.3711520794987208, "grad_norm": 1.8535281419754028, "learning_rate": 1.449301722714362e-05, "loss": 0.1636006474494934, "step": 14090 }, { "epoch": 0.37141549474321955, "grad_norm": 1.512150764465332, "learning_rate": 1.4485393834634202e-05, "loss": 0.15981277227401733, "step": 14100 }, { "epoch": 0.37167890998771824, "grad_norm": 2.3241958618164062, "learning_rate": 1.447776717746163e-05, "loss": 0.1536604642868042, "step": 14110 }, { "epoch": 0.371942325232217, "grad_norm": 1.689979910850525, "learning_rate": 1.4470137261176922e-05, "loss": 0.16261379718780516, "step": 14120 }, { "epoch": 0.37220574047671573, "grad_norm": 4.621491432189941, "learning_rate": 1.4462504091333454e-05, "loss": 0.16060540676116944, "step": 14130 }, { "epoch": 0.3724691557212145, "grad_norm": 1.6623435020446777, "learning_rate": 1.4454867673486983e-05, "loss": 0.1614415764808655, "step": 14140 }, { "epoch": 0.3727325709657132, "grad_norm": 2.006497383117676, "learning_rate": 1.4447228013195616e-05, "loss": 0.15829625129699706, "step": 14150 }, { "epoch": 0.37299598621021196, "grad_norm": 1.0645620822906494, "learning_rate": 1.4439585116019835e-05, "loss": 0.1629784345626831, "step": 14160 }, { "epoch": 0.3732594014547107, "grad_norm": 2.4035580158233643, "learning_rate": 1.4431938987522463e-05, "loss": 0.16518287658691405, "step": 14170 }, { "epoch": 0.37352281669920945, "grad_norm": 1.5943456888198853, "learning_rate": 1.4424289633268689e-05, "loss": 0.17188730239868164, "step": 14180 }, { "epoch": 0.37378623194370814, "grad_norm": 1.3842244148254395, "learning_rate": 1.4416637058826038e-05, "loss": 0.14701359272003173, "step": 14190 }, { "epoch": 0.3740496471882069, "grad_norm": 2.724670171737671, "learning_rate": 1.4408981269764386e-05, "loss": 0.15982205867767335, "step": 14200 }, { "epoch": 0.3743130624327056, "grad_norm": 3.4785871505737305, "learning_rate": 1.4401322271655947e-05, "loss": 0.15324802398681642, "step": 14210 }, { "epoch": 0.37457647767720437, "grad_norm": 1.2485357522964478, "learning_rate": 1.4393660070075267e-05, "loss": 0.1657063126564026, "step": 14220 }, { "epoch": 0.3748398929217031, "grad_norm": 3.6526713371276855, "learning_rate": 1.4385994670599233e-05, "loss": 0.15707308053970337, "step": 14230 }, { "epoch": 0.37510330816620185, "grad_norm": 2.056427001953125, "learning_rate": 1.4378326078807048e-05, "loss": 0.1573472499847412, "step": 14240 }, { "epoch": 0.3753667234107006, "grad_norm": 1.1022026538848877, "learning_rate": 1.4370654300280246e-05, "loss": 0.1629125475883484, "step": 14250 }, { "epoch": 0.37563013865519934, "grad_norm": 1.6389265060424805, "learning_rate": 1.4362979340602677e-05, "loss": 0.165444278717041, "step": 14260 }, { "epoch": 0.3758935538996981, "grad_norm": 1.539693832397461, "learning_rate": 1.4355301205360513e-05, "loss": 0.16524276733398438, "step": 14270 }, { "epoch": 0.3761569691441968, "grad_norm": 1.19831120967865, "learning_rate": 1.4347619900142223e-05, "loss": 0.17629740238189698, "step": 14280 }, { "epoch": 0.3764203843886955, "grad_norm": 3.3784737586975098, "learning_rate": 1.43399354305386e-05, "loss": 0.14657105207443238, "step": 14290 }, { "epoch": 0.37668379963319426, "grad_norm": 5.036738872528076, "learning_rate": 1.4332247802142729e-05, "loss": 0.1759623885154724, "step": 14300 }, { "epoch": 0.376947214877693, "grad_norm": 1.1066523790359497, "learning_rate": 1.4324557020550001e-05, "loss": 0.15777672529220582, "step": 14310 }, { "epoch": 0.37721063012219175, "grad_norm": 1.560503602027893, "learning_rate": 1.4316863091358093e-05, "loss": 0.15208611488342286, "step": 14320 }, { "epoch": 0.3774740453666905, "grad_norm": 1.7352256774902344, "learning_rate": 1.4309166020166984e-05, "loss": 0.1440818190574646, "step": 14330 }, { "epoch": 0.37773746061118924, "grad_norm": 1.0832479000091553, "learning_rate": 1.430146581257893e-05, "loss": 0.15781214237213134, "step": 14340 }, { "epoch": 0.378000875855688, "grad_norm": 2.7672743797302246, "learning_rate": 1.4293762474198478e-05, "loss": 0.16154085397720336, "step": 14350 }, { "epoch": 0.3782642911001867, "grad_norm": 2.627849817276001, "learning_rate": 1.428605601063245e-05, "loss": 0.155913782119751, "step": 14360 }, { "epoch": 0.3785277063446854, "grad_norm": 1.473291277885437, "learning_rate": 1.427834642748994e-05, "loss": 0.15812067985534667, "step": 14370 }, { "epoch": 0.37879112158918415, "grad_norm": 2.0499441623687744, "learning_rate": 1.4270633730382315e-05, "loss": 0.1647390604019165, "step": 14380 }, { "epoch": 0.3790545368336829, "grad_norm": 2.9803826808929443, "learning_rate": 1.4262917924923212e-05, "loss": 0.14823155403137206, "step": 14390 }, { "epoch": 0.37931795207818164, "grad_norm": 2.479318618774414, "learning_rate": 1.4255199016728527e-05, "loss": 0.15921130180358886, "step": 14400 }, { "epoch": 0.3795813673226804, "grad_norm": 1.768227219581604, "learning_rate": 1.4247477011416409e-05, "loss": 0.1604311227798462, "step": 14410 }, { "epoch": 0.37984478256717913, "grad_norm": 0.9915137887001038, "learning_rate": 1.4239751914607274e-05, "loss": 0.1666959524154663, "step": 14420 }, { "epoch": 0.3801081978116779, "grad_norm": 3.8443822860717773, "learning_rate": 1.4232023731923772e-05, "loss": 0.14343565702438354, "step": 14430 }, { "epoch": 0.3803716130561766, "grad_norm": 0.949441134929657, "learning_rate": 1.422429246899082e-05, "loss": 0.16786718368530273, "step": 14440 }, { "epoch": 0.3806350283006753, "grad_norm": 2.3500211238861084, "learning_rate": 1.4216558131435551e-05, "loss": 0.15511682033538818, "step": 14450 }, { "epoch": 0.38089844354517405, "grad_norm": 1.4051313400268555, "learning_rate": 1.420882072488736e-05, "loss": 0.15422044992446898, "step": 14460 }, { "epoch": 0.3811618587896728, "grad_norm": 2.7151999473571777, "learning_rate": 1.420108025497786e-05, "loss": 0.16974589824676514, "step": 14470 }, { "epoch": 0.38142527403417154, "grad_norm": 1.0133333206176758, "learning_rate": 1.41933367273409e-05, "loss": 0.1596132278442383, "step": 14480 }, { "epoch": 0.3816886892786703, "grad_norm": 1.4748766422271729, "learning_rate": 1.4185590147612555e-05, "loss": 0.15678553581237792, "step": 14490 }, { "epoch": 0.381952104523169, "grad_norm": 5.148837089538574, "learning_rate": 1.4177840521431117e-05, "loss": 0.1545060992240906, "step": 14500 }, { "epoch": 0.38221551976766777, "grad_norm": 6.032548904418945, "learning_rate": 1.4170087854437102e-05, "loss": 0.15046563148498535, "step": 14510 }, { "epoch": 0.3824789350121665, "grad_norm": 3.0742909908294678, "learning_rate": 1.4162332152273233e-05, "loss": 0.1708683967590332, "step": 14520 }, { "epoch": 0.38274235025666525, "grad_norm": 2.5285043716430664, "learning_rate": 1.4154573420584448e-05, "loss": 0.14951255321502685, "step": 14530 }, { "epoch": 0.38300576550116394, "grad_norm": 1.0771830081939697, "learning_rate": 1.414681166501788e-05, "loss": 0.16247870922088622, "step": 14540 }, { "epoch": 0.3832691807456627, "grad_norm": 1.200103759765625, "learning_rate": 1.4139046891222877e-05, "loss": 0.16982555389404297, "step": 14550 }, { "epoch": 0.38353259599016143, "grad_norm": 1.128592610359192, "learning_rate": 1.4131279104850968e-05, "loss": 0.18184273242950438, "step": 14560 }, { "epoch": 0.3837960112346602, "grad_norm": 1.1413217782974243, "learning_rate": 1.4123508311555894e-05, "loss": 0.1552832841873169, "step": 14570 }, { "epoch": 0.3840594264791589, "grad_norm": 1.7595350742340088, "learning_rate": 1.4115734516993562e-05, "loss": 0.1684280514717102, "step": 14580 }, { "epoch": 0.38432284172365766, "grad_norm": 5.18871545791626, "learning_rate": 1.4107957726822084e-05, "loss": 0.16218349933624268, "step": 14590 }, { "epoch": 0.3845862569681564, "grad_norm": 3.8760569095611572, "learning_rate": 1.4100177946701738e-05, "loss": 0.16552958488464356, "step": 14600 }, { "epoch": 0.38484967221265515, "grad_norm": 1.1924980878829956, "learning_rate": 1.4092395182294989e-05, "loss": 0.1605541944503784, "step": 14610 }, { "epoch": 0.3851130874571539, "grad_norm": 1.2744816541671753, "learning_rate": 1.4084609439266464e-05, "loss": 0.16657575368881225, "step": 14620 }, { "epoch": 0.3853765027016526, "grad_norm": 2.062944173812866, "learning_rate": 1.4076820723282962e-05, "loss": 0.15980315208435059, "step": 14630 }, { "epoch": 0.3856399179461513, "grad_norm": 1.3884375095367432, "learning_rate": 1.4069029040013448e-05, "loss": 0.1711089015007019, "step": 14640 }, { "epoch": 0.38590333319065007, "grad_norm": 1.7210917472839355, "learning_rate": 1.4061234395129049e-05, "loss": 0.15961074829101562, "step": 14650 }, { "epoch": 0.3861667484351488, "grad_norm": 2.797461748123169, "learning_rate": 1.405343679430304e-05, "loss": 0.16171481609344482, "step": 14660 }, { "epoch": 0.38643016367964755, "grad_norm": 2.118846893310547, "learning_rate": 1.4045636243210852e-05, "loss": 0.16472985744476318, "step": 14670 }, { "epoch": 0.3866935789241463, "grad_norm": 1.6010947227478027, "learning_rate": 1.4037832747530063e-05, "loss": 0.14690585136413575, "step": 14680 }, { "epoch": 0.38695699416864504, "grad_norm": 1.6369959115982056, "learning_rate": 1.403002631294039e-05, "loss": 0.15343093872070312, "step": 14690 }, { "epoch": 0.3872204094131438, "grad_norm": 1.5257538557052612, "learning_rate": 1.40222169451237e-05, "loss": 0.14870972633361818, "step": 14700 }, { "epoch": 0.3874838246576425, "grad_norm": 1.442521333694458, "learning_rate": 1.4014404649763983e-05, "loss": 0.16199653148651122, "step": 14710 }, { "epoch": 0.3877472399021412, "grad_norm": 3.310291290283203, "learning_rate": 1.4006589432547367e-05, "loss": 0.16171224117279054, "step": 14720 }, { "epoch": 0.38801065514663996, "grad_norm": 2.5675952434539795, "learning_rate": 1.3998771299162099e-05, "loss": 0.1625523328781128, "step": 14730 }, { "epoch": 0.3882740703911387, "grad_norm": 2.671154499053955, "learning_rate": 1.3990950255298565e-05, "loss": 0.159955894947052, "step": 14740 }, { "epoch": 0.38853748563563745, "grad_norm": 1.1514859199523926, "learning_rate": 1.3983126306649243e-05, "loss": 0.17662596702575684, "step": 14750 }, { "epoch": 0.3888009008801362, "grad_norm": 3.808117628097534, "learning_rate": 1.3975299458908753e-05, "loss": 0.1507678747177124, "step": 14760 }, { "epoch": 0.38906431612463493, "grad_norm": 2.178919792175293, "learning_rate": 1.3967469717773807e-05, "loss": 0.16728146076202394, "step": 14770 }, { "epoch": 0.3893277313691337, "grad_norm": 1.8595620393753052, "learning_rate": 1.395963708894323e-05, "loss": 0.15019235610961915, "step": 14780 }, { "epoch": 0.3895911466136324, "grad_norm": 3.5673186779022217, "learning_rate": 1.3951801578117946e-05, "loss": 0.16242350339889527, "step": 14790 }, { "epoch": 0.3898545618581311, "grad_norm": 0.967399537563324, "learning_rate": 1.3943963191000976e-05, "loss": 0.1539109468460083, "step": 14800 }, { "epoch": 0.39011797710262985, "grad_norm": 2.3631088733673096, "learning_rate": 1.393612193329744e-05, "loss": 0.16953096389770508, "step": 14810 }, { "epoch": 0.3903813923471286, "grad_norm": 4.901872158050537, "learning_rate": 1.3928277810714542e-05, "loss": 0.1600364327430725, "step": 14820 }, { "epoch": 0.39064480759162734, "grad_norm": 3.703784942626953, "learning_rate": 1.3920430828961576e-05, "loss": 0.1589793562889099, "step": 14830 }, { "epoch": 0.3909082228361261, "grad_norm": 1.4982322454452515, "learning_rate": 1.3912580993749907e-05, "loss": 0.15891288518905639, "step": 14840 }, { "epoch": 0.39117163808062483, "grad_norm": 1.5926371812820435, "learning_rate": 1.3904728310792988e-05, "loss": 0.16124531030654907, "step": 14850 }, { "epoch": 0.39143505332512357, "grad_norm": 2.4811606407165527, "learning_rate": 1.3896872785806342e-05, "loss": 0.15599894523620605, "step": 14860 }, { "epoch": 0.3916984685696223, "grad_norm": 3.6083927154541016, "learning_rate": 1.3889014424507557e-05, "loss": 0.1529940128326416, "step": 14870 }, { "epoch": 0.39196188381412106, "grad_norm": 2.0242459774017334, "learning_rate": 1.3881153232616284e-05, "loss": 0.15902193784713745, "step": 14880 }, { "epoch": 0.39222529905861975, "grad_norm": 1.2690616846084595, "learning_rate": 1.3873289215854247e-05, "loss": 0.1584039568901062, "step": 14890 }, { "epoch": 0.3924887143031185, "grad_norm": 2.5568671226501465, "learning_rate": 1.3865422379945205e-05, "loss": 0.14635078907012938, "step": 14900 }, { "epoch": 0.39275212954761723, "grad_norm": 1.79790198802948, "learning_rate": 1.3857552730614989e-05, "loss": 0.17087142467498778, "step": 14910 }, { "epoch": 0.393015544792116, "grad_norm": 1.824770450592041, "learning_rate": 1.3849680273591468e-05, "loss": 0.16054561138153076, "step": 14920 }, { "epoch": 0.3932789600366147, "grad_norm": 1.742627739906311, "learning_rate": 1.3841805014604553e-05, "loss": 0.16993348598480223, "step": 14930 }, { "epoch": 0.39354237528111347, "grad_norm": 3.0533995628356934, "learning_rate": 1.3833926959386199e-05, "loss": 0.16019976139068604, "step": 14940 }, { "epoch": 0.3938057905256122, "grad_norm": 2.1299166679382324, "learning_rate": 1.3826046113670395e-05, "loss": 0.16624672412872316, "step": 14950 }, { "epoch": 0.39406920577011095, "grad_norm": 1.162654161453247, "learning_rate": 1.3818162483193159e-05, "loss": 0.15703234672546387, "step": 14960 }, { "epoch": 0.39433262101460964, "grad_norm": 1.243863821029663, "learning_rate": 1.3810276073692537e-05, "loss": 0.14666136503219604, "step": 14970 }, { "epoch": 0.3945960362591084, "grad_norm": 1.0581907033920288, "learning_rate": 1.3802386890908602e-05, "loss": 0.15453500747680665, "step": 14980 }, { "epoch": 0.39485945150360713, "grad_norm": 1.782218337059021, "learning_rate": 1.3794494940583437e-05, "loss": 0.16078600883483887, "step": 14990 }, { "epoch": 0.39512286674810587, "grad_norm": 1.1909539699554443, "learning_rate": 1.3786600228461147e-05, "loss": 0.16131398677825928, "step": 15000 }, { "epoch": 0.3953862819926046, "grad_norm": 3.846465587615967, "learning_rate": 1.377870276028784e-05, "loss": 0.15857396125793458, "step": 15010 }, { "epoch": 0.39564969723710336, "grad_norm": 1.3952217102050781, "learning_rate": 1.377080254181164e-05, "loss": 0.16953203678131104, "step": 15020 }, { "epoch": 0.3959131124816021, "grad_norm": 1.3148268461227417, "learning_rate": 1.3762899578782662e-05, "loss": 0.16704320907592773, "step": 15030 }, { "epoch": 0.39617652772610085, "grad_norm": 3.7494139671325684, "learning_rate": 1.375499387695303e-05, "loss": 0.16067204475402833, "step": 15040 }, { "epoch": 0.3964399429705996, "grad_norm": 1.4528555870056152, "learning_rate": 1.3747085442076844e-05, "loss": 0.1540750741958618, "step": 15050 }, { "epoch": 0.3967033582150983, "grad_norm": 1.0391533374786377, "learning_rate": 1.3739174279910214e-05, "loss": 0.15533171892166137, "step": 15060 }, { "epoch": 0.396966773459597, "grad_norm": 1.6441750526428223, "learning_rate": 1.3731260396211219e-05, "loss": 0.14133117198944092, "step": 15070 }, { "epoch": 0.39723018870409577, "grad_norm": 3.383928060531616, "learning_rate": 1.372334379673993e-05, "loss": 0.16007041931152344, "step": 15080 }, { "epoch": 0.3974936039485945, "grad_norm": 1.0477783679962158, "learning_rate": 1.3715424487258386e-05, "loss": 0.15036731958389282, "step": 15090 }, { "epoch": 0.39775701919309325, "grad_norm": 2.9328155517578125, "learning_rate": 1.37075024735306e-05, "loss": 0.1690086841583252, "step": 15100 }, { "epoch": 0.398020434437592, "grad_norm": 1.5472793579101562, "learning_rate": 1.3699577761322563e-05, "loss": 0.16049761772155763, "step": 15110 }, { "epoch": 0.39828384968209074, "grad_norm": 1.6671178340911865, "learning_rate": 1.3691650356402212e-05, "loss": 0.16591136455535888, "step": 15120 }, { "epoch": 0.3985472649265895, "grad_norm": 1.418971061706543, "learning_rate": 1.3683720264539469e-05, "loss": 0.14458789825439453, "step": 15130 }, { "epoch": 0.3988106801710882, "grad_norm": 2.40997052192688, "learning_rate": 1.3675787491506184e-05, "loss": 0.1538228750228882, "step": 15140 }, { "epoch": 0.3990740954155869, "grad_norm": 1.53390371799469, "learning_rate": 1.3667852043076173e-05, "loss": 0.15694559812545777, "step": 15150 }, { "epoch": 0.39933751066008566, "grad_norm": 3.2445197105407715, "learning_rate": 1.3659913925025203e-05, "loss": 0.16500037908554077, "step": 15160 }, { "epoch": 0.3996009259045844, "grad_norm": 3.998479127883911, "learning_rate": 1.3651973143130979e-05, "loss": 0.15226938724517822, "step": 15170 }, { "epoch": 0.39986434114908315, "grad_norm": 1.8225268125534058, "learning_rate": 1.3644029703173139e-05, "loss": 0.14883028268814086, "step": 15180 }, { "epoch": 0.4001277563935819, "grad_norm": 1.359062671661377, "learning_rate": 1.3636083610933262e-05, "loss": 0.15788183212280274, "step": 15190 }, { "epoch": 0.40039117163808063, "grad_norm": 2.7848572731018066, "learning_rate": 1.3628134872194861e-05, "loss": 0.15969011783599854, "step": 15200 }, { "epoch": 0.4006545868825794, "grad_norm": 1.7498366832733154, "learning_rate": 1.362018349274337e-05, "loss": 0.15291560888290406, "step": 15210 }, { "epoch": 0.4009180021270781, "grad_norm": 2.7801787853240967, "learning_rate": 1.3612229478366144e-05, "loss": 0.15409090518951415, "step": 15220 }, { "epoch": 0.4011814173715768, "grad_norm": 2.9895615577697754, "learning_rate": 1.360427283485246e-05, "loss": 0.15917093753814698, "step": 15230 }, { "epoch": 0.40144483261607555, "grad_norm": 1.64254891872406, "learning_rate": 1.3596313567993504e-05, "loss": 0.1620260000228882, "step": 15240 }, { "epoch": 0.4017082478605743, "grad_norm": 1.6572551727294922, "learning_rate": 1.3588351683582374e-05, "loss": 0.1507791757583618, "step": 15250 }, { "epoch": 0.40197166310507304, "grad_norm": 1.2379554510116577, "learning_rate": 1.3580387187414077e-05, "loss": 0.1510254979133606, "step": 15260 }, { "epoch": 0.4022350783495718, "grad_norm": 2.984525203704834, "learning_rate": 1.3572420085285512e-05, "loss": 0.15230637788772583, "step": 15270 }, { "epoch": 0.4024984935940705, "grad_norm": 2.180574655532837, "learning_rate": 1.3564450382995482e-05, "loss": 0.15894075632095336, "step": 15280 }, { "epoch": 0.40276190883856927, "grad_norm": 1.6114373207092285, "learning_rate": 1.3556478086344678e-05, "loss": 0.14573376178741454, "step": 15290 }, { "epoch": 0.403025324083068, "grad_norm": 4.450488567352295, "learning_rate": 1.3548503201135687e-05, "loss": 0.13826676607131957, "step": 15300 }, { "epoch": 0.40328873932756676, "grad_norm": 1.534326195716858, "learning_rate": 1.3540525733172966e-05, "loss": 0.16706380844116211, "step": 15310 }, { "epoch": 0.40355215457206545, "grad_norm": 1.082912564277649, "learning_rate": 1.353254568826287e-05, "loss": 0.15773417949676513, "step": 15320 }, { "epoch": 0.4038155698165642, "grad_norm": 2.1823604106903076, "learning_rate": 1.3524563072213612e-05, "loss": 0.15869476795196533, "step": 15330 }, { "epoch": 0.40407898506106293, "grad_norm": 2.6143991947174072, "learning_rate": 1.3516577890835292e-05, "loss": 0.15746926069259642, "step": 15340 }, { "epoch": 0.4043424003055617, "grad_norm": 1.1718958616256714, "learning_rate": 1.350859014993986e-05, "loss": 0.16412045955657958, "step": 15350 }, { "epoch": 0.4046058155500604, "grad_norm": 0.964106559753418, "learning_rate": 1.3500599855341146e-05, "loss": 0.15213282108306886, "step": 15360 }, { "epoch": 0.40486923079455917, "grad_norm": 3.36811900138855, "learning_rate": 1.3492607012854824e-05, "loss": 0.16791504621505737, "step": 15370 }, { "epoch": 0.4051326460390579, "grad_norm": 2.0190179347991943, "learning_rate": 1.3484611628298438e-05, "loss": 0.14716031551361083, "step": 15380 }, { "epoch": 0.40539606128355665, "grad_norm": 3.9474923610687256, "learning_rate": 1.3476613707491372e-05, "loss": 0.14961695671081543, "step": 15390 }, { "epoch": 0.4056594765280554, "grad_norm": 1.2071101665496826, "learning_rate": 1.3468613256254852e-05, "loss": 0.14634755849838257, "step": 15400 }, { "epoch": 0.4059228917725541, "grad_norm": 1.4622379541397095, "learning_rate": 1.3460610280411957e-05, "loss": 0.1658925175666809, "step": 15410 }, { "epoch": 0.40618630701705283, "grad_norm": 2.4624578952789307, "learning_rate": 1.3452604785787595e-05, "loss": 0.1541474223136902, "step": 15420 }, { "epoch": 0.40644972226155157, "grad_norm": 2.638230323791504, "learning_rate": 1.3444596778208516e-05, "loss": 0.16513596773147582, "step": 15430 }, { "epoch": 0.4067131375060503, "grad_norm": 2.2766971588134766, "learning_rate": 1.3436586263503286e-05, "loss": 0.1545332670211792, "step": 15440 }, { "epoch": 0.40697655275054906, "grad_norm": 2.1176092624664307, "learning_rate": 1.3428573247502307e-05, "loss": 0.16908299922943115, "step": 15450 }, { "epoch": 0.4072399679950478, "grad_norm": 3.9783034324645996, "learning_rate": 1.3420557736037796e-05, "loss": 0.16041033267974852, "step": 15460 }, { "epoch": 0.40750338323954655, "grad_norm": 1.1058838367462158, "learning_rate": 1.3412539734943792e-05, "loss": 0.1403239607810974, "step": 15470 }, { "epoch": 0.4077667984840453, "grad_norm": 1.0821478366851807, "learning_rate": 1.3404519250056139e-05, "loss": 0.15126826763153076, "step": 15480 }, { "epoch": 0.408030213728544, "grad_norm": 2.8137574195861816, "learning_rate": 1.3396496287212491e-05, "loss": 0.14997904300689696, "step": 15490 }, { "epoch": 0.4082936289730427, "grad_norm": 3.1658334732055664, "learning_rate": 1.3388470852252307e-05, "loss": 0.15826936960220336, "step": 15500 }, { "epoch": 0.40855704421754147, "grad_norm": 1.366135835647583, "learning_rate": 1.3380442951016846e-05, "loss": 0.14203083515167236, "step": 15510 }, { "epoch": 0.4088204594620402, "grad_norm": 2.39990234375, "learning_rate": 1.3372412589349164e-05, "loss": 0.15541350841522217, "step": 15520 }, { "epoch": 0.40908387470653895, "grad_norm": 2.118067979812622, "learning_rate": 1.3364379773094099e-05, "loss": 0.15978001356124877, "step": 15530 }, { "epoch": 0.4093472899510377, "grad_norm": 1.5658795833587646, "learning_rate": 1.335634450809828e-05, "loss": 0.1612769842147827, "step": 15540 }, { "epoch": 0.40961070519553644, "grad_norm": 1.0575807094573975, "learning_rate": 1.3348306800210127e-05, "loss": 0.15726358890533448, "step": 15550 }, { "epoch": 0.4098741204400352, "grad_norm": 1.3648037910461426, "learning_rate": 1.3340266655279828e-05, "loss": 0.15852506160736085, "step": 15560 }, { "epoch": 0.4101375356845339, "grad_norm": 1.2493823766708374, "learning_rate": 1.3332224079159342e-05, "loss": 0.15233199596405028, "step": 15570 }, { "epoch": 0.4104009509290326, "grad_norm": 0.7505952715873718, "learning_rate": 1.3324179077702408e-05, "loss": 0.15922908782958983, "step": 15580 }, { "epoch": 0.41066436617353136, "grad_norm": 3.438309669494629, "learning_rate": 1.3316131656764523e-05, "loss": 0.13891427516937255, "step": 15590 }, { "epoch": 0.4109277814180301, "grad_norm": 2.1918845176696777, "learning_rate": 1.3308081822202953e-05, "loss": 0.16688706874847412, "step": 15600 }, { "epoch": 0.41119119666252885, "grad_norm": 3.1545825004577637, "learning_rate": 1.3300029579876705e-05, "loss": 0.16632318496704102, "step": 15610 }, { "epoch": 0.4114546119070276, "grad_norm": 1.5551623106002808, "learning_rate": 1.3291974935646554e-05, "loss": 0.1575329542160034, "step": 15620 }, { "epoch": 0.41171802715152633, "grad_norm": 1.5047698020935059, "learning_rate": 1.3283917895375018e-05, "loss": 0.15119123458862305, "step": 15630 }, { "epoch": 0.4119814423960251, "grad_norm": 1.514114260673523, "learning_rate": 1.327585846492636e-05, "loss": 0.1473496675491333, "step": 15640 }, { "epoch": 0.4122448576405238, "grad_norm": 3.2116692066192627, "learning_rate": 1.3267796650166576e-05, "loss": 0.16574447154998778, "step": 15650 }, { "epoch": 0.41250827288502256, "grad_norm": 1.4739993810653687, "learning_rate": 1.3259732456963406e-05, "loss": 0.16158599853515626, "step": 15660 }, { "epoch": 0.41277168812952125, "grad_norm": 2.609041452407837, "learning_rate": 1.3251665891186313e-05, "loss": 0.14836628437042237, "step": 15670 }, { "epoch": 0.41303510337402, "grad_norm": 3.1157925128936768, "learning_rate": 1.32435969587065e-05, "loss": 0.14812450408935546, "step": 15680 }, { "epoch": 0.41329851861851874, "grad_norm": 2.1525752544403076, "learning_rate": 1.323552566539688e-05, "loss": 0.15600073337554932, "step": 15690 }, { "epoch": 0.4135619338630175, "grad_norm": 3.9913535118103027, "learning_rate": 1.3227452017132085e-05, "loss": 0.1590052604675293, "step": 15700 }, { "epoch": 0.4138253491075162, "grad_norm": 4.172314643859863, "learning_rate": 1.321937601978847e-05, "loss": 0.1552678108215332, "step": 15710 }, { "epoch": 0.41408876435201497, "grad_norm": 1.8040648698806763, "learning_rate": 1.3211297679244093e-05, "loss": 0.16420326232910157, "step": 15720 }, { "epoch": 0.4143521795965137, "grad_norm": 1.8970303535461426, "learning_rate": 1.3203217001378716e-05, "loss": 0.1706709384918213, "step": 15730 }, { "epoch": 0.41461559484101246, "grad_norm": 2.612912178039551, "learning_rate": 1.3195133992073808e-05, "loss": 0.1559166431427002, "step": 15740 }, { "epoch": 0.41487901008551115, "grad_norm": 2.4222357273101807, "learning_rate": 1.318704865721253e-05, "loss": 0.15240130424499512, "step": 15750 }, { "epoch": 0.4151424253300099, "grad_norm": 2.866710901260376, "learning_rate": 1.3178961002679738e-05, "loss": 0.16336419582366943, "step": 15760 }, { "epoch": 0.41540584057450863, "grad_norm": 2.2122085094451904, "learning_rate": 1.317087103436198e-05, "loss": 0.16675169467926027, "step": 15770 }, { "epoch": 0.4156692558190074, "grad_norm": 1.9132623672485352, "learning_rate": 1.3162778758147477e-05, "loss": 0.15795884132385254, "step": 15780 }, { "epoch": 0.4159326710635061, "grad_norm": 1.3575239181518555, "learning_rate": 1.315468417992614e-05, "loss": 0.16643534898757933, "step": 15790 }, { "epoch": 0.41619608630800486, "grad_norm": 1.2532196044921875, "learning_rate": 1.3146587305589552e-05, "loss": 0.15945684909820557, "step": 15800 }, { "epoch": 0.4164595015525036, "grad_norm": 1.6412190198898315, "learning_rate": 1.3138488141030967e-05, "loss": 0.1530603885650635, "step": 15810 }, { "epoch": 0.41672291679700235, "grad_norm": 3.977388620376587, "learning_rate": 1.313038669214531e-05, "loss": 0.16721708774566652, "step": 15820 }, { "epoch": 0.4169863320415011, "grad_norm": 1.259612798690796, "learning_rate": 1.312228296482916e-05, "loss": 0.15464837551116944, "step": 15830 }, { "epoch": 0.4172497472859998, "grad_norm": 1.131041407585144, "learning_rate": 1.311417696498076e-05, "loss": 0.15309510231018067, "step": 15840 }, { "epoch": 0.4175131625304985, "grad_norm": 1.6333165168762207, "learning_rate": 1.310606869850001e-05, "loss": 0.1409689426422119, "step": 15850 }, { "epoch": 0.41777657777499727, "grad_norm": 1.1971694231033325, "learning_rate": 1.3097958171288453e-05, "loss": 0.15929949283599854, "step": 15860 }, { "epoch": 0.418039993019496, "grad_norm": 5.512283802032471, "learning_rate": 1.3089845389249282e-05, "loss": 0.15576045513153075, "step": 15870 }, { "epoch": 0.41830340826399476, "grad_norm": 1.5426267385482788, "learning_rate": 1.308173035828733e-05, "loss": 0.165060293674469, "step": 15880 }, { "epoch": 0.4185668235084935, "grad_norm": 1.1938704252243042, "learning_rate": 1.3073613084309065e-05, "loss": 0.1614086389541626, "step": 15890 }, { "epoch": 0.41883023875299225, "grad_norm": 3.1029396057128906, "learning_rate": 1.3065493573222594e-05, "loss": 0.16065995693206786, "step": 15900 }, { "epoch": 0.419093653997491, "grad_norm": 1.2007269859313965, "learning_rate": 1.305737183093764e-05, "loss": 0.1566516637802124, "step": 15910 }, { "epoch": 0.41935706924198973, "grad_norm": 1.6347784996032715, "learning_rate": 1.3049247863365562e-05, "loss": 0.15184805393218995, "step": 15920 }, { "epoch": 0.4196204844864884, "grad_norm": 1.3549771308898926, "learning_rate": 1.3041121676419333e-05, "loss": 0.15667027235031128, "step": 15930 }, { "epoch": 0.41988389973098716, "grad_norm": 1.112658977508545, "learning_rate": 1.3032993276013546e-05, "loss": 0.15204071998596191, "step": 15940 }, { "epoch": 0.4201473149754859, "grad_norm": 0.8151575922966003, "learning_rate": 1.3024862668064393e-05, "loss": 0.15079913139343262, "step": 15950 }, { "epoch": 0.42041073021998465, "grad_norm": 0.8945287466049194, "learning_rate": 1.3016729858489689e-05, "loss": 0.16913001537322997, "step": 15960 }, { "epoch": 0.4206741454644834, "grad_norm": 1.2679989337921143, "learning_rate": 1.300859485320884e-05, "loss": 0.16577606201171874, "step": 15970 }, { "epoch": 0.42093756070898214, "grad_norm": 3.1424803733825684, "learning_rate": 1.3000457658142858e-05, "loss": 0.16321210861206054, "step": 15980 }, { "epoch": 0.4212009759534809, "grad_norm": 2.1093006134033203, "learning_rate": 1.299231827921434e-05, "loss": 0.1610027074813843, "step": 15990 }, { "epoch": 0.4214643911979796, "grad_norm": 2.1681597232818604, "learning_rate": 1.2984176722347478e-05, "loss": 0.15712908506393433, "step": 16000 }, { "epoch": 0.4217278064424783, "grad_norm": 1.1225603818893433, "learning_rate": 1.2976032993468049e-05, "loss": 0.16006659269332885, "step": 16010 }, { "epoch": 0.42199122168697706, "grad_norm": 3.2723870277404785, "learning_rate": 1.296788709850341e-05, "loss": 0.14388504028320312, "step": 16020 }, { "epoch": 0.4222546369314758, "grad_norm": 1.5090786218643188, "learning_rate": 1.29597390433825e-05, "loss": 0.14282917976379395, "step": 16030 }, { "epoch": 0.42251805217597455, "grad_norm": 1.6493394374847412, "learning_rate": 1.2951588834035813e-05, "loss": 0.15339262485504152, "step": 16040 }, { "epoch": 0.4227814674204733, "grad_norm": 1.2919631004333496, "learning_rate": 1.2943436476395435e-05, "loss": 0.1678244113922119, "step": 16050 }, { "epoch": 0.42304488266497203, "grad_norm": 0.9469822645187378, "learning_rate": 1.2935281976394998e-05, "loss": 0.1623077630996704, "step": 16060 }, { "epoch": 0.4233082979094708, "grad_norm": 2.871307373046875, "learning_rate": 1.2927125339969701e-05, "loss": 0.15667492151260376, "step": 16070 }, { "epoch": 0.4235717131539695, "grad_norm": 1.5482304096221924, "learning_rate": 1.2918966573056291e-05, "loss": 0.14953796863555907, "step": 16080 }, { "epoch": 0.42383512839846826, "grad_norm": 1.9509609937667847, "learning_rate": 1.2910805681593078e-05, "loss": 0.16551586389541625, "step": 16090 }, { "epoch": 0.42409854364296695, "grad_norm": 0.9165215492248535, "learning_rate": 1.2902642671519903e-05, "loss": 0.1509498953819275, "step": 16100 }, { "epoch": 0.4243619588874657, "grad_norm": 0.6161694526672363, "learning_rate": 1.2894477548778164e-05, "loss": 0.16238949298858643, "step": 16110 }, { "epoch": 0.42462537413196444, "grad_norm": 1.1001226902008057, "learning_rate": 1.2886310319310787e-05, "loss": 0.15858452320098876, "step": 16120 }, { "epoch": 0.4248887893764632, "grad_norm": 2.271212339401245, "learning_rate": 1.2878140989062233e-05, "loss": 0.16526122093200685, "step": 16130 }, { "epoch": 0.4251522046209619, "grad_norm": 1.9352915287017822, "learning_rate": 1.2869969563978493e-05, "loss": 0.16358444690704346, "step": 16140 }, { "epoch": 0.42541561986546067, "grad_norm": 2.3219287395477295, "learning_rate": 1.2861796050007082e-05, "loss": 0.15115556716918946, "step": 16150 }, { "epoch": 0.4256790351099594, "grad_norm": 2.1218690872192383, "learning_rate": 1.2853620453097039e-05, "loss": 0.17125887870788575, "step": 16160 }, { "epoch": 0.42594245035445816, "grad_norm": 1.1628432273864746, "learning_rate": 1.2845442779198916e-05, "loss": 0.15161523818969727, "step": 16170 }, { "epoch": 0.4262058655989569, "grad_norm": 0.6703410744667053, "learning_rate": 1.2837263034264773e-05, "loss": 0.14365029335021973, "step": 16180 }, { "epoch": 0.4264692808434556, "grad_norm": 1.3911399841308594, "learning_rate": 1.2829081224248183e-05, "loss": 0.15607850551605223, "step": 16190 }, { "epoch": 0.42673269608795433, "grad_norm": 2.088107109069824, "learning_rate": 1.2820897355104224e-05, "loss": 0.1695912003517151, "step": 16200 }, { "epoch": 0.4269961113324531, "grad_norm": 1.5882540941238403, "learning_rate": 1.2812711432789464e-05, "loss": 0.15938180685043335, "step": 16210 }, { "epoch": 0.4272595265769518, "grad_norm": 2.130225419998169, "learning_rate": 1.2804523463261975e-05, "loss": 0.14295580387115478, "step": 16220 }, { "epoch": 0.42752294182145056, "grad_norm": 1.6990611553192139, "learning_rate": 1.2796333452481309e-05, "loss": 0.1598968029022217, "step": 16230 }, { "epoch": 0.4277863570659493, "grad_norm": 2.5199832916259766, "learning_rate": 1.2788141406408516e-05, "loss": 0.16489577293395996, "step": 16240 }, { "epoch": 0.42804977231044805, "grad_norm": 0.8657703995704651, "learning_rate": 1.2779947331006113e-05, "loss": 0.150496768951416, "step": 16250 }, { "epoch": 0.4283131875549468, "grad_norm": 1.48533296585083, "learning_rate": 1.2771751232238107e-05, "loss": 0.14486509561538696, "step": 16260 }, { "epoch": 0.42857660279944554, "grad_norm": 5.4344401359558105, "learning_rate": 1.2763553116069969e-05, "loss": 0.15568660497665404, "step": 16270 }, { "epoch": 0.4288400180439442, "grad_norm": 1.5432227849960327, "learning_rate": 1.2755352988468646e-05, "loss": 0.14619338512420654, "step": 16280 }, { "epoch": 0.42910343328844297, "grad_norm": 2.690412759780884, "learning_rate": 1.2747150855402541e-05, "loss": 0.16346964836120606, "step": 16290 }, { "epoch": 0.4293668485329417, "grad_norm": 1.2178256511688232, "learning_rate": 1.2738946722841519e-05, "loss": 0.15042538642883302, "step": 16300 }, { "epoch": 0.42963026377744046, "grad_norm": 3.7438533306121826, "learning_rate": 1.2730740596756908e-05, "loss": 0.14227490425109862, "step": 16310 }, { "epoch": 0.4298936790219392, "grad_norm": 4.0962700843811035, "learning_rate": 1.2722532483121475e-05, "loss": 0.16051967144012452, "step": 16320 }, { "epoch": 0.43015709426643794, "grad_norm": 1.6544036865234375, "learning_rate": 1.271432238790944e-05, "loss": 0.15602684020996094, "step": 16330 }, { "epoch": 0.4304205095109367, "grad_norm": 2.7611258029937744, "learning_rate": 1.2706110317096463e-05, "loss": 0.15222347974777223, "step": 16340 }, { "epoch": 0.43068392475543543, "grad_norm": 3.697237968444824, "learning_rate": 1.269789627665965e-05, "loss": 0.1665552020072937, "step": 16350 }, { "epoch": 0.4309473399999341, "grad_norm": 1.7537156343460083, "learning_rate": 1.2689680272577527e-05, "loss": 0.17554166316986083, "step": 16360 }, { "epoch": 0.43121075524443286, "grad_norm": 1.7937908172607422, "learning_rate": 1.2681462310830061e-05, "loss": 0.15441317558288575, "step": 16370 }, { "epoch": 0.4314741704889316, "grad_norm": 2.3646929264068604, "learning_rate": 1.2673242397398635e-05, "loss": 0.15836877822875978, "step": 16380 }, { "epoch": 0.43173758573343035, "grad_norm": 0.9639731645584106, "learning_rate": 1.266502053826606e-05, "loss": 0.16086291074752807, "step": 16390 }, { "epoch": 0.4320010009779291, "grad_norm": 3.550212860107422, "learning_rate": 1.2656796739416557e-05, "loss": 0.16197526454925537, "step": 16400 }, { "epoch": 0.43226441622242784, "grad_norm": 1.4565311670303345, "learning_rate": 1.2648571006835764e-05, "loss": 0.14561922550201417, "step": 16410 }, { "epoch": 0.4325278314669266, "grad_norm": 1.042940616607666, "learning_rate": 1.2640343346510724e-05, "loss": 0.1554659128189087, "step": 16420 }, { "epoch": 0.4327912467114253, "grad_norm": 0.6557489037513733, "learning_rate": 1.2632113764429884e-05, "loss": 0.15790307521820068, "step": 16430 }, { "epoch": 0.43305466195592407, "grad_norm": 1.515228033065796, "learning_rate": 1.2623882266583083e-05, "loss": 0.15804719924926758, "step": 16440 }, { "epoch": 0.43331807720042276, "grad_norm": 1.9285097122192383, "learning_rate": 1.2615648858961564e-05, "loss": 0.1498339891433716, "step": 16450 }, { "epoch": 0.4335814924449215, "grad_norm": 2.60168719291687, "learning_rate": 1.260741354755796e-05, "loss": 0.15736546516418456, "step": 16460 }, { "epoch": 0.43384490768942024, "grad_norm": 0.7001813650131226, "learning_rate": 1.2599176338366282e-05, "loss": 0.16831963062286376, "step": 16470 }, { "epoch": 0.434108322933919, "grad_norm": 3.397522449493408, "learning_rate": 1.2590937237381927e-05, "loss": 0.1477783203125, "step": 16480 }, { "epoch": 0.43437173817841773, "grad_norm": 1.4849143028259277, "learning_rate": 1.2582696250601667e-05, "loss": 0.16257691383361816, "step": 16490 }, { "epoch": 0.4346351534229165, "grad_norm": 2.6116015911102295, "learning_rate": 1.257445338402365e-05, "loss": 0.15912990570068358, "step": 16500 }, { "epoch": 0.4348985686674152, "grad_norm": 0.7836470007896423, "learning_rate": 1.256620864364739e-05, "loss": 0.1614063024520874, "step": 16510 }, { "epoch": 0.43516198391191396, "grad_norm": 1.1753970384597778, "learning_rate": 1.255796203547376e-05, "loss": 0.15014877319335937, "step": 16520 }, { "epoch": 0.4354253991564127, "grad_norm": 1.409611463546753, "learning_rate": 1.2549713565505001e-05, "loss": 0.1511407732963562, "step": 16530 }, { "epoch": 0.4356888144009114, "grad_norm": 1.8745590448379517, "learning_rate": 1.254146323974471e-05, "loss": 0.14770905971527098, "step": 16540 }, { "epoch": 0.43595222964541014, "grad_norm": 2.11601185798645, "learning_rate": 1.253321106419782e-05, "loss": 0.1437983274459839, "step": 16550 }, { "epoch": 0.4362156448899089, "grad_norm": 0.8396399021148682, "learning_rate": 1.2524957044870627e-05, "loss": 0.155448317527771, "step": 16560 }, { "epoch": 0.4364790601344076, "grad_norm": 1.0349730253219604, "learning_rate": 1.2516701187770761e-05, "loss": 0.15187687873840333, "step": 16570 }, { "epoch": 0.43674247537890637, "grad_norm": 2.00941801071167, "learning_rate": 1.250844349890719e-05, "loss": 0.15818049907684326, "step": 16580 }, { "epoch": 0.4370058906234051, "grad_norm": 1.212266445159912, "learning_rate": 1.250018398429022e-05, "loss": 0.16992175579071045, "step": 16590 }, { "epoch": 0.43726930586790386, "grad_norm": 3.7703781127929688, "learning_rate": 1.2491922649931475e-05, "loss": 0.1668122887611389, "step": 16600 }, { "epoch": 0.4375327211124026, "grad_norm": 3.936406373977661, "learning_rate": 1.2483659501843916e-05, "loss": 0.15646789073944092, "step": 16610 }, { "epoch": 0.4377961363569013, "grad_norm": 0.8056411743164062, "learning_rate": 1.2475394546041822e-05, "loss": 0.14660277366638183, "step": 16620 }, { "epoch": 0.43805955160140003, "grad_norm": 1.8079756498336792, "learning_rate": 1.2467127788540774e-05, "loss": 0.14338220357894899, "step": 16630 }, { "epoch": 0.4383229668458988, "grad_norm": 2.0318763256073, "learning_rate": 1.245885923535768e-05, "loss": 0.15707893371582032, "step": 16640 }, { "epoch": 0.4385863820903975, "grad_norm": 0.9354209303855896, "learning_rate": 1.2450588892510756e-05, "loss": 0.1523742437362671, "step": 16650 }, { "epoch": 0.43884979733489626, "grad_norm": 1.1750320196151733, "learning_rate": 1.2442316766019502e-05, "loss": 0.16480355262756347, "step": 16660 }, { "epoch": 0.439113212579395, "grad_norm": 2.5508551597595215, "learning_rate": 1.2434042861904743e-05, "loss": 0.14883568286895751, "step": 16670 }, { "epoch": 0.43937662782389375, "grad_norm": 1.6356611251831055, "learning_rate": 1.2425767186188569e-05, "loss": 0.15219457149505616, "step": 16680 }, { "epoch": 0.4396400430683925, "grad_norm": 1.7863452434539795, "learning_rate": 1.2417489744894381e-05, "loss": 0.1526647686958313, "step": 16690 }, { "epoch": 0.43990345831289124, "grad_norm": 2.2071375846862793, "learning_rate": 1.2409210544046859e-05, "loss": 0.14533441066741942, "step": 16700 }, { "epoch": 0.4401668735573899, "grad_norm": 2.5460519790649414, "learning_rate": 1.2400929589671962e-05, "loss": 0.15717296600341796, "step": 16710 }, { "epoch": 0.44043028880188867, "grad_norm": 3.385284900665283, "learning_rate": 1.2392646887796924e-05, "loss": 0.165934419631958, "step": 16720 }, { "epoch": 0.4406937040463874, "grad_norm": 1.1267379522323608, "learning_rate": 1.2384362444450253e-05, "loss": 0.15075194835662842, "step": 16730 }, { "epoch": 0.44095711929088616, "grad_norm": 1.52692711353302, "learning_rate": 1.2376076265661722e-05, "loss": 0.1664757251739502, "step": 16740 }, { "epoch": 0.4412205345353849, "grad_norm": 2.4064269065856934, "learning_rate": 1.2367788357462375e-05, "loss": 0.15832142829895018, "step": 16750 }, { "epoch": 0.44148394977988364, "grad_norm": 3.814232110977173, "learning_rate": 1.2359498725884509e-05, "loss": 0.15829418897628783, "step": 16760 }, { "epoch": 0.4417473650243824, "grad_norm": 2.1325478553771973, "learning_rate": 1.2351207376961666e-05, "loss": 0.16048043966293335, "step": 16770 }, { "epoch": 0.44201078026888113, "grad_norm": 2.3439905643463135, "learning_rate": 1.234291431672866e-05, "loss": 0.16256933212280272, "step": 16780 }, { "epoch": 0.4422741955133799, "grad_norm": 1.5603235960006714, "learning_rate": 1.2334619551221531e-05, "loss": 0.1489524245262146, "step": 16790 }, { "epoch": 0.44253761075787856, "grad_norm": 1.9997000694274902, "learning_rate": 1.232632308647757e-05, "loss": 0.1585424304008484, "step": 16800 }, { "epoch": 0.4428010260023773, "grad_norm": 3.508033037185669, "learning_rate": 1.2318024928535296e-05, "loss": 0.15317168235778808, "step": 16810 }, { "epoch": 0.44306444124687605, "grad_norm": 1.557861089706421, "learning_rate": 1.2309725083434477e-05, "loss": 0.15419974327087402, "step": 16820 }, { "epoch": 0.4433278564913748, "grad_norm": 1.1665585041046143, "learning_rate": 1.2301423557216088e-05, "loss": 0.1571541905403137, "step": 16830 }, { "epoch": 0.44359127173587354, "grad_norm": 2.2400662899017334, "learning_rate": 1.2293120355922346e-05, "loss": 0.14883594512939452, "step": 16840 }, { "epoch": 0.4438546869803723, "grad_norm": 0.760226845741272, "learning_rate": 1.2284815485596671e-05, "loss": 0.1388306736946106, "step": 16850 }, { "epoch": 0.444118102224871, "grad_norm": 2.532831907272339, "learning_rate": 1.2276508952283712e-05, "loss": 0.15533778667449952, "step": 16860 }, { "epoch": 0.44438151746936977, "grad_norm": 2.809915542602539, "learning_rate": 1.2268200762029319e-05, "loss": 0.15016436576843262, "step": 16870 }, { "epoch": 0.44464493271386846, "grad_norm": 1.7426577806472778, "learning_rate": 1.2259890920880555e-05, "loss": 0.15364935398101806, "step": 16880 }, { "epoch": 0.4449083479583672, "grad_norm": 2.8148703575134277, "learning_rate": 1.2251579434885679e-05, "loss": 0.15197408199310303, "step": 16890 }, { "epoch": 0.44517176320286594, "grad_norm": 1.048153042793274, "learning_rate": 1.2243266310094146e-05, "loss": 0.1469741940498352, "step": 16900 }, { "epoch": 0.4454351784473647, "grad_norm": 1.8285155296325684, "learning_rate": 1.2234951552556613e-05, "loss": 0.15793948173522948, "step": 16910 }, { "epoch": 0.44569859369186343, "grad_norm": 1.7763044834136963, "learning_rate": 1.2226635168324916e-05, "loss": 0.15218764543533325, "step": 16920 }, { "epoch": 0.4459620089363622, "grad_norm": 1.4055720567703247, "learning_rate": 1.2218317163452078e-05, "loss": 0.1420309066772461, "step": 16930 }, { "epoch": 0.4462254241808609, "grad_norm": 2.0574228763580322, "learning_rate": 1.2209997543992301e-05, "loss": 0.1509220838546753, "step": 16940 }, { "epoch": 0.44648883942535966, "grad_norm": 4.514794826507568, "learning_rate": 1.2201676316000964e-05, "loss": 0.15954864025115967, "step": 16950 }, { "epoch": 0.4467522546698584, "grad_norm": 1.2354834079742432, "learning_rate": 1.2193353485534614e-05, "loss": 0.15140347480773925, "step": 16960 }, { "epoch": 0.4470156699143571, "grad_norm": 0.9404045343399048, "learning_rate": 1.2185029058650972e-05, "loss": 0.14943840503692626, "step": 16970 }, { "epoch": 0.44727908515885584, "grad_norm": 1.668919563293457, "learning_rate": 1.2176703041408908e-05, "loss": 0.15305508375167848, "step": 16980 }, { "epoch": 0.4475425004033546, "grad_norm": 3.320655345916748, "learning_rate": 1.2168375439868464e-05, "loss": 0.15734139680862427, "step": 16990 }, { "epoch": 0.4478059156478533, "grad_norm": 0.8656436204910278, "learning_rate": 1.2160046260090816e-05, "loss": 0.14107741117477418, "step": 17000 }, { "epoch": 0.44806933089235207, "grad_norm": 2.5499863624572754, "learning_rate": 1.2151715508138313e-05, "loss": 0.14303795099258423, "step": 17010 }, { "epoch": 0.4483327461368508, "grad_norm": 3.654669761657715, "learning_rate": 1.2143383190074429e-05, "loss": 0.15588021278381348, "step": 17020 }, { "epoch": 0.44859616138134956, "grad_norm": 3.5761048793792725, "learning_rate": 1.2135049311963787e-05, "loss": 0.16106927394866943, "step": 17030 }, { "epoch": 0.4488595766258483, "grad_norm": 1.156817078590393, "learning_rate": 1.2126713879872138e-05, "loss": 0.15855567455291747, "step": 17040 }, { "epoch": 0.44912299187034704, "grad_norm": 1.1861586570739746, "learning_rate": 1.2118376899866375e-05, "loss": 0.16582155227661133, "step": 17050 }, { "epoch": 0.44938640711484573, "grad_norm": 5.1541571617126465, "learning_rate": 1.2110038378014512e-05, "loss": 0.15457282066345215, "step": 17060 }, { "epoch": 0.4496498223593445, "grad_norm": 1.5380417108535767, "learning_rate": 1.210169832038568e-05, "loss": 0.15240786075592042, "step": 17070 }, { "epoch": 0.4499132376038432, "grad_norm": 1.7226371765136719, "learning_rate": 1.2093356733050137e-05, "loss": 0.14246550798416138, "step": 17080 }, { "epoch": 0.45017665284834196, "grad_norm": 4.284316062927246, "learning_rate": 1.2085013622079248e-05, "loss": 0.1647144675254822, "step": 17090 }, { "epoch": 0.4504400680928407, "grad_norm": 1.6376136541366577, "learning_rate": 1.2076668993545491e-05, "loss": 0.1507442831993103, "step": 17100 }, { "epoch": 0.45070348333733945, "grad_norm": 4.0754218101501465, "learning_rate": 1.206832285352244e-05, "loss": 0.16919618844985962, "step": 17110 }, { "epoch": 0.4509668985818382, "grad_norm": 1.5145442485809326, "learning_rate": 1.2059975208084787e-05, "loss": 0.158736515045166, "step": 17120 }, { "epoch": 0.45123031382633694, "grad_norm": 4.301379203796387, "learning_rate": 1.2051626063308299e-05, "loss": 0.15175249576568603, "step": 17130 }, { "epoch": 0.4514937290708356, "grad_norm": 1.0274739265441895, "learning_rate": 1.2043275425269854e-05, "loss": 0.15184235572814941, "step": 17140 }, { "epoch": 0.45175714431533437, "grad_norm": 1.545817494392395, "learning_rate": 1.2034923300047391e-05, "loss": 0.14848012924194337, "step": 17150 }, { "epoch": 0.4520205595598331, "grad_norm": 2.5371062755584717, "learning_rate": 1.2026569693719963e-05, "loss": 0.14434609413146973, "step": 17160 }, { "epoch": 0.45228397480433186, "grad_norm": 1.2162970304489136, "learning_rate": 1.2018214612367673e-05, "loss": 0.15101388692855836, "step": 17170 }, { "epoch": 0.4525473900488306, "grad_norm": 1.0275274515151978, "learning_rate": 1.200985806207172e-05, "loss": 0.15006164312362671, "step": 17180 }, { "epoch": 0.45281080529332934, "grad_norm": 3.0804569721221924, "learning_rate": 1.2001500048914354e-05, "loss": 0.1404451847076416, "step": 17190 }, { "epoch": 0.4530742205378281, "grad_norm": 0.6648167967796326, "learning_rate": 1.1993140578978898e-05, "loss": 0.15683066844940186, "step": 17200 }, { "epoch": 0.45333763578232683, "grad_norm": 1.346746802330017, "learning_rate": 1.1984779658349743e-05, "loss": 0.1578765630722046, "step": 17210 }, { "epoch": 0.4536010510268256, "grad_norm": 2.189789056777954, "learning_rate": 1.1976417293112318e-05, "loss": 0.14413584470748902, "step": 17220 }, { "epoch": 0.45386446627132426, "grad_norm": 1.736456274986267, "learning_rate": 1.1968053489353122e-05, "loss": 0.1421433448791504, "step": 17230 }, { "epoch": 0.454127881515823, "grad_norm": 2.236889362335205, "learning_rate": 1.195968825315968e-05, "loss": 0.15542534589767457, "step": 17240 }, { "epoch": 0.45439129676032175, "grad_norm": 2.6046712398529053, "learning_rate": 1.1951321590620583e-05, "loss": 0.15463911294937133, "step": 17250 }, { "epoch": 0.4546547120048205, "grad_norm": 3.195147752761841, "learning_rate": 1.1942953507825446e-05, "loss": 0.15486407279968262, "step": 17260 }, { "epoch": 0.45491812724931924, "grad_norm": 1.8419455289840698, "learning_rate": 1.1934584010864921e-05, "loss": 0.15364439487457277, "step": 17270 }, { "epoch": 0.455181542493818, "grad_norm": 0.8945298790931702, "learning_rate": 1.1926213105830683e-05, "loss": 0.15587685108184815, "step": 17280 }, { "epoch": 0.4554449577383167, "grad_norm": 1.6104570627212524, "learning_rate": 1.1917840798815448e-05, "loss": 0.1681126594543457, "step": 17290 }, { "epoch": 0.45570837298281547, "grad_norm": 1.6152132749557495, "learning_rate": 1.1909467095912932e-05, "loss": 0.16995372772216796, "step": 17300 }, { "epoch": 0.4559717882273142, "grad_norm": 2.9857981204986572, "learning_rate": 1.1901092003217885e-05, "loss": 0.15086650848388672, "step": 17310 }, { "epoch": 0.4562352034718129, "grad_norm": 1.6727803945541382, "learning_rate": 1.1892715526826058e-05, "loss": 0.16174957752227784, "step": 17320 }, { "epoch": 0.45649861871631164, "grad_norm": 0.5933118462562561, "learning_rate": 1.1884337672834212e-05, "loss": 0.14448877573013305, "step": 17330 }, { "epoch": 0.4567620339608104, "grad_norm": 1.9478734731674194, "learning_rate": 1.1875958447340108e-05, "loss": 0.14333455562591552, "step": 17340 }, { "epoch": 0.45702544920530913, "grad_norm": 4.44288969039917, "learning_rate": 1.1867577856442514e-05, "loss": 0.15417487621307374, "step": 17350 }, { "epoch": 0.4572888644498079, "grad_norm": 1.6958166360855103, "learning_rate": 1.1859195906241182e-05, "loss": 0.15081453323364258, "step": 17360 }, { "epoch": 0.4575522796943066, "grad_norm": 5.621847629547119, "learning_rate": 1.1850812602836851e-05, "loss": 0.1595436453819275, "step": 17370 }, { "epoch": 0.45781569493880536, "grad_norm": 1.468120813369751, "learning_rate": 1.1842427952331263e-05, "loss": 0.15486751794815062, "step": 17380 }, { "epoch": 0.4580791101833041, "grad_norm": 2.136945962905884, "learning_rate": 1.183404196082712e-05, "loss": 0.1641899824142456, "step": 17390 }, { "epoch": 0.4583425254278028, "grad_norm": 1.5485758781433105, "learning_rate": 1.1825654634428111e-05, "loss": 0.1474777340888977, "step": 17400 }, { "epoch": 0.45860594067230154, "grad_norm": 2.927480459213257, "learning_rate": 1.1817265979238892e-05, "loss": 0.15977882146835326, "step": 17410 }, { "epoch": 0.4588693559168003, "grad_norm": 0.8518860340118408, "learning_rate": 1.1808876001365093e-05, "loss": 0.16243958473205566, "step": 17420 }, { "epoch": 0.459132771161299, "grad_norm": 4.164353847503662, "learning_rate": 1.1800484706913294e-05, "loss": 0.15554482936859132, "step": 17430 }, { "epoch": 0.45939618640579777, "grad_norm": 2.869448661804199, "learning_rate": 1.1792092101991052e-05, "loss": 0.15326801538467408, "step": 17440 }, { "epoch": 0.4596596016502965, "grad_norm": 4.633493900299072, "learning_rate": 1.1783698192706856e-05, "loss": 0.15662978887557982, "step": 17450 }, { "epoch": 0.45992301689479526, "grad_norm": 2.460253953933716, "learning_rate": 1.1775302985170161e-05, "loss": 0.15299644470214843, "step": 17460 }, { "epoch": 0.460186432139294, "grad_norm": 1.5410771369934082, "learning_rate": 1.1766906485491361e-05, "loss": 0.16276494264602662, "step": 17470 }, { "epoch": 0.46044984738379274, "grad_norm": 1.4526876211166382, "learning_rate": 1.175850869978179e-05, "loss": 0.16646761894226075, "step": 17480 }, { "epoch": 0.46071326262829143, "grad_norm": 1.0261658430099487, "learning_rate": 1.1750109634153719e-05, "loss": 0.1396243929862976, "step": 17490 }, { "epoch": 0.4609766778727902, "grad_norm": 3.691807508468628, "learning_rate": 1.1741709294720346e-05, "loss": 0.1409089207649231, "step": 17500 }, { "epoch": 0.4612400931172889, "grad_norm": 3.1384334564208984, "learning_rate": 1.173330768759581e-05, "loss": 0.147768235206604, "step": 17510 }, { "epoch": 0.46150350836178766, "grad_norm": 1.8695298433303833, "learning_rate": 1.1724904818895158e-05, "loss": 0.14496386051177979, "step": 17520 }, { "epoch": 0.4617669236062864, "grad_norm": 1.3464291095733643, "learning_rate": 1.1716500694734359e-05, "loss": 0.1606682777404785, "step": 17530 }, { "epoch": 0.46203033885078515, "grad_norm": 1.815765619277954, "learning_rate": 1.1708095321230295e-05, "loss": 0.14999713897705078, "step": 17540 }, { "epoch": 0.4622937540952839, "grad_norm": 2.3977928161621094, "learning_rate": 1.1699688704500765e-05, "loss": 0.14356672763824463, "step": 17550 }, { "epoch": 0.46255716933978264, "grad_norm": 2.3598735332489014, "learning_rate": 1.1691280850664463e-05, "loss": 0.15466682910919188, "step": 17560 }, { "epoch": 0.4628205845842814, "grad_norm": 2.4956204891204834, "learning_rate": 1.1682871765840996e-05, "loss": 0.15175845623016357, "step": 17570 }, { "epoch": 0.46308399982878007, "grad_norm": 2.3795504570007324, "learning_rate": 1.1674461456150845e-05, "loss": 0.15567885637283324, "step": 17580 }, { "epoch": 0.4633474150732788, "grad_norm": 4.095776557922363, "learning_rate": 1.1666049927715408e-05, "loss": 0.15780513286590575, "step": 17590 }, { "epoch": 0.46361083031777756, "grad_norm": 1.1310771703720093, "learning_rate": 1.1657637186656952e-05, "loss": 0.14254138469696045, "step": 17600 }, { "epoch": 0.4638742455622763, "grad_norm": 1.5997593402862549, "learning_rate": 1.1649223239098639e-05, "loss": 0.15365511178970337, "step": 17610 }, { "epoch": 0.46413766080677504, "grad_norm": 3.423727512359619, "learning_rate": 1.1640808091164495e-05, "loss": 0.15658434629440307, "step": 17620 }, { "epoch": 0.4644010760512738, "grad_norm": 3.002230167388916, "learning_rate": 1.1632391748979435e-05, "loss": 0.14603960514068604, "step": 17630 }, { "epoch": 0.46466449129577253, "grad_norm": 1.2198796272277832, "learning_rate": 1.162397421866923e-05, "loss": 0.1548284411430359, "step": 17640 }, { "epoch": 0.4649279065402713, "grad_norm": 2.6891894340515137, "learning_rate": 1.1615555506360526e-05, "loss": 0.15195513963699342, "step": 17650 }, { "epoch": 0.46519132178476996, "grad_norm": 1.8933161497116089, "learning_rate": 1.1607135618180823e-05, "loss": 0.163409423828125, "step": 17660 }, { "epoch": 0.4654547370292687, "grad_norm": 2.392296552658081, "learning_rate": 1.1598714560258477e-05, "loss": 0.14913451671600342, "step": 17670 }, { "epoch": 0.46571815227376745, "grad_norm": 1.8850008249282837, "learning_rate": 1.1590292338722702e-05, "loss": 0.15556907653808594, "step": 17680 }, { "epoch": 0.4659815675182662, "grad_norm": 1.6008729934692383, "learning_rate": 1.1581868959703553e-05, "loss": 0.1478629231452942, "step": 17690 }, { "epoch": 0.46624498276276494, "grad_norm": 3.73388409614563, "learning_rate": 1.1573444429331926e-05, "loss": 0.14492721557617189, "step": 17700 }, { "epoch": 0.4665083980072637, "grad_norm": 1.4206384420394897, "learning_rate": 1.1565018753739559e-05, "loss": 0.15342350006103517, "step": 17710 }, { "epoch": 0.4667718132517624, "grad_norm": 1.1505632400512695, "learning_rate": 1.1556591939059024e-05, "loss": 0.15025155544281005, "step": 17720 }, { "epoch": 0.46703522849626117, "grad_norm": 4.592880725860596, "learning_rate": 1.1548163991423716e-05, "loss": 0.14866397380828858, "step": 17730 }, { "epoch": 0.4672986437407599, "grad_norm": 1.5478802919387817, "learning_rate": 1.1539734916967864e-05, "loss": 0.15178451538085938, "step": 17740 }, { "epoch": 0.4675620589852586, "grad_norm": 1.3008482456207275, "learning_rate": 1.1531304721826506e-05, "loss": 0.14175326824188234, "step": 17750 }, { "epoch": 0.46782547422975734, "grad_norm": 2.3335015773773193, "learning_rate": 1.1522873412135509e-05, "loss": 0.15152777433395387, "step": 17760 }, { "epoch": 0.4680888894742561, "grad_norm": 1.351057529449463, "learning_rate": 1.1514440994031536e-05, "loss": 0.14477300643920898, "step": 17770 }, { "epoch": 0.46835230471875483, "grad_norm": 0.9229457974433899, "learning_rate": 1.1506007473652075e-05, "loss": 0.15559895038604737, "step": 17780 }, { "epoch": 0.4686157199632536, "grad_norm": 4.42130708694458, "learning_rate": 1.1497572857135397e-05, "loss": 0.15679354667663575, "step": 17790 }, { "epoch": 0.4688791352077523, "grad_norm": 3.2427563667297363, "learning_rate": 1.1489137150620582e-05, "loss": 0.1658451199531555, "step": 17800 }, { "epoch": 0.46914255045225106, "grad_norm": 0.8832017183303833, "learning_rate": 1.1480700360247504e-05, "loss": 0.15986504554748535, "step": 17810 }, { "epoch": 0.4694059656967498, "grad_norm": 2.53617262840271, "learning_rate": 1.147226249215682e-05, "loss": 0.16035568714141846, "step": 17820 }, { "epoch": 0.46966938094124855, "grad_norm": 1.9290162324905396, "learning_rate": 1.1463823552489977e-05, "loss": 0.15516204833984376, "step": 17830 }, { "epoch": 0.46993279618574724, "grad_norm": 2.530945062637329, "learning_rate": 1.1455383547389193e-05, "loss": 0.15146286487579347, "step": 17840 }, { "epoch": 0.470196211430246, "grad_norm": 1.2230780124664307, "learning_rate": 1.1446942482997479e-05, "loss": 0.14464617967605592, "step": 17850 }, { "epoch": 0.4704596266747447, "grad_norm": 1.1083898544311523, "learning_rate": 1.1438500365458593e-05, "loss": 0.14617717266082764, "step": 17860 }, { "epoch": 0.47072304191924347, "grad_norm": 1.6999573707580566, "learning_rate": 1.1430057200917084e-05, "loss": 0.15079450607299805, "step": 17870 }, { "epoch": 0.4709864571637422, "grad_norm": 1.501018762588501, "learning_rate": 1.1421612995518243e-05, "loss": 0.1425314426422119, "step": 17880 }, { "epoch": 0.47124987240824096, "grad_norm": 1.5468533039093018, "learning_rate": 1.1413167755408132e-05, "loss": 0.159220552444458, "step": 17890 }, { "epoch": 0.4715132876527397, "grad_norm": 2.1574907302856445, "learning_rate": 1.1404721486733558e-05, "loss": 0.15643953084945678, "step": 17900 }, { "epoch": 0.47177670289723844, "grad_norm": 1.3413221836090088, "learning_rate": 1.1396274195642085e-05, "loss": 0.1462104082107544, "step": 17910 }, { "epoch": 0.47204011814173713, "grad_norm": 1.9813793897628784, "learning_rate": 1.1387825888282006e-05, "loss": 0.15173981189727784, "step": 17920 }, { "epoch": 0.4723035333862359, "grad_norm": 1.6517291069030762, "learning_rate": 1.1379376570802373e-05, "loss": 0.14015519618988037, "step": 17930 }, { "epoch": 0.4725669486307346, "grad_norm": 1.0787506103515625, "learning_rate": 1.137092624935296e-05, "loss": 0.14379255771636962, "step": 17940 }, { "epoch": 0.47283036387523336, "grad_norm": 4.211395740509033, "learning_rate": 1.1362474930084278e-05, "loss": 0.1669153690338135, "step": 17950 }, { "epoch": 0.4730937791197321, "grad_norm": 1.2553256750106812, "learning_rate": 1.1354022619147557e-05, "loss": 0.14375418424606323, "step": 17960 }, { "epoch": 0.47335719436423085, "grad_norm": 2.1988112926483154, "learning_rate": 1.1345569322694756e-05, "loss": 0.15550391674041747, "step": 17970 }, { "epoch": 0.4736206096087296, "grad_norm": 2.889254570007324, "learning_rate": 1.133711504687855e-05, "loss": 0.15858149528503418, "step": 17980 }, { "epoch": 0.47388402485322834, "grad_norm": 1.8310811519622803, "learning_rate": 1.1328659797852325e-05, "loss": 0.1444251775741577, "step": 17990 }, { "epoch": 0.4741474400977271, "grad_norm": 2.4594950675964355, "learning_rate": 1.1320203581770177e-05, "loss": 0.16284908056259156, "step": 18000 }, { "epoch": 0.47441085534222577, "grad_norm": 4.513452529907227, "learning_rate": 1.1311746404786903e-05, "loss": 0.1601824164390564, "step": 18010 }, { "epoch": 0.4746742705867245, "grad_norm": 2.930063486099243, "learning_rate": 1.1303288273058002e-05, "loss": 0.14759578704833984, "step": 18020 }, { "epoch": 0.47493768583122326, "grad_norm": 2.1415138244628906, "learning_rate": 1.1294829192739665e-05, "loss": 0.15077409744262696, "step": 18030 }, { "epoch": 0.475201101075722, "grad_norm": 1.0839760303497314, "learning_rate": 1.1286369169988782e-05, "loss": 0.15543529987335206, "step": 18040 }, { "epoch": 0.47546451632022074, "grad_norm": 1.9588181972503662, "learning_rate": 1.1277908210962916e-05, "loss": 0.16128263473510743, "step": 18050 }, { "epoch": 0.4757279315647195, "grad_norm": 2.9340434074401855, "learning_rate": 1.126944632182032e-05, "loss": 0.15315259695053102, "step": 18060 }, { "epoch": 0.47599134680921823, "grad_norm": 1.8488330841064453, "learning_rate": 1.1260983508719922e-05, "loss": 0.13568508625030518, "step": 18070 }, { "epoch": 0.476254762053717, "grad_norm": 2.348237991333008, "learning_rate": 1.1252519777821324e-05, "loss": 0.15585193634033204, "step": 18080 }, { "epoch": 0.4765181772982157, "grad_norm": 2.984783887863159, "learning_rate": 1.1244055135284795e-05, "loss": 0.14858829975128174, "step": 18090 }, { "epoch": 0.4767815925427144, "grad_norm": 1.9838366508483887, "learning_rate": 1.123558958727126e-05, "loss": 0.15472196340560912, "step": 18100 }, { "epoch": 0.47704500778721315, "grad_norm": 2.4229907989501953, "learning_rate": 1.1227123139942322e-05, "loss": 0.14921170473098755, "step": 18110 }, { "epoch": 0.4773084230317119, "grad_norm": 1.1431751251220703, "learning_rate": 1.1218655799460216e-05, "loss": 0.15778546333312987, "step": 18120 }, { "epoch": 0.47757183827621064, "grad_norm": 1.916442632675171, "learning_rate": 1.1210187571987845e-05, "loss": 0.15187995433807372, "step": 18130 }, { "epoch": 0.4778352535207094, "grad_norm": 4.560173034667969, "learning_rate": 1.1201718463688745e-05, "loss": 0.16127102375030516, "step": 18140 }, { "epoch": 0.4780986687652081, "grad_norm": 2.0625946521759033, "learning_rate": 1.1193248480727101e-05, "loss": 0.1524353504180908, "step": 18150 }, { "epoch": 0.47836208400970687, "grad_norm": 1.8124313354492188, "learning_rate": 1.118477762926773e-05, "loss": 0.14360418319702148, "step": 18160 }, { "epoch": 0.4786254992542056, "grad_norm": 2.346139430999756, "learning_rate": 1.117630591547609e-05, "loss": 0.1583205819129944, "step": 18170 }, { "epoch": 0.4788889144987043, "grad_norm": 1.7868889570236206, "learning_rate": 1.1167833345518248e-05, "loss": 0.14373605251312255, "step": 18180 }, { "epoch": 0.47915232974320304, "grad_norm": 1.6953699588775635, "learning_rate": 1.1159359925560918e-05, "loss": 0.15125881433486937, "step": 18190 }, { "epoch": 0.4794157449877018, "grad_norm": 3.9657087326049805, "learning_rate": 1.115088566177141e-05, "loss": 0.15501124858856202, "step": 18200 }, { "epoch": 0.47967916023220053, "grad_norm": 1.8421427011489868, "learning_rate": 1.1142410560317669e-05, "loss": 0.15389506816864013, "step": 18210 }, { "epoch": 0.4799425754766993, "grad_norm": 1.6649041175842285, "learning_rate": 1.113393462736823e-05, "loss": 0.16412949562072754, "step": 18220 }, { "epoch": 0.480205990721198, "grad_norm": 1.079190731048584, "learning_rate": 1.1125457869092251e-05, "loss": 0.1553109407424927, "step": 18230 }, { "epoch": 0.48046940596569676, "grad_norm": 1.3432776927947998, "learning_rate": 1.1116980291659474e-05, "loss": 0.15690159797668457, "step": 18240 }, { "epoch": 0.4807328212101955, "grad_norm": 2.062835454940796, "learning_rate": 1.1108501901240254e-05, "loss": 0.14372735023498534, "step": 18250 }, { "epoch": 0.48099623645469425, "grad_norm": 1.822687029838562, "learning_rate": 1.1100022704005524e-05, "loss": 0.1518087148666382, "step": 18260 }, { "epoch": 0.48125965169919294, "grad_norm": 4.065635681152344, "learning_rate": 1.109154270612681e-05, "loss": 0.15125951766967774, "step": 18270 }, { "epoch": 0.4815230669436917, "grad_norm": 1.22953462600708, "learning_rate": 1.1083061913776224e-05, "loss": 0.14459948539733886, "step": 18280 }, { "epoch": 0.4817864821881904, "grad_norm": 2.1333446502685547, "learning_rate": 1.1074580333126448e-05, "loss": 0.15320847034454346, "step": 18290 }, { "epoch": 0.48204989743268917, "grad_norm": 1.1647471189498901, "learning_rate": 1.1066097970350747e-05, "loss": 0.15227870941162108, "step": 18300 }, { "epoch": 0.4823133126771879, "grad_norm": 0.9540969729423523, "learning_rate": 1.1057614831622942e-05, "loss": 0.14591057300567628, "step": 18310 }, { "epoch": 0.48257672792168665, "grad_norm": 0.6988644003868103, "learning_rate": 1.1049130923117437e-05, "loss": 0.14878396987915038, "step": 18320 }, { "epoch": 0.4828401431661854, "grad_norm": 1.1794672012329102, "learning_rate": 1.1040646251009181e-05, "loss": 0.1571000337600708, "step": 18330 }, { "epoch": 0.48310355841068414, "grad_norm": 1.0825790166854858, "learning_rate": 1.1032160821473692e-05, "loss": 0.15925517082214355, "step": 18340 }, { "epoch": 0.4833669736551829, "grad_norm": 3.3989975452423096, "learning_rate": 1.1023674640687019e-05, "loss": 0.15715229511260986, "step": 18350 }, { "epoch": 0.4836303888996816, "grad_norm": 1.327684760093689, "learning_rate": 1.1015187714825782e-05, "loss": 0.14607582092285157, "step": 18360 }, { "epoch": 0.4838938041441803, "grad_norm": 1.0227867364883423, "learning_rate": 1.1006700050067122e-05, "loss": 0.1409260630607605, "step": 18370 }, { "epoch": 0.48415721938867906, "grad_norm": 2.014120578765869, "learning_rate": 1.0998211652588737e-05, "loss": 0.14446213245391845, "step": 18380 }, { "epoch": 0.4844206346331778, "grad_norm": 2.3058149814605713, "learning_rate": 1.0989722528568845e-05, "loss": 0.15437610149383546, "step": 18390 }, { "epoch": 0.48468404987767655, "grad_norm": 2.3912806510925293, "learning_rate": 1.0981232684186191e-05, "loss": 0.158417809009552, "step": 18400 }, { "epoch": 0.4849474651221753, "grad_norm": 1.753167748451233, "learning_rate": 1.0972742125620062e-05, "loss": 0.14919042587280273, "step": 18410 }, { "epoch": 0.48521088036667404, "grad_norm": 2.589138984680176, "learning_rate": 1.0964250859050245e-05, "loss": 0.14793246984481812, "step": 18420 }, { "epoch": 0.4854742956111728, "grad_norm": 1.2413376569747925, "learning_rate": 1.0955758890657051e-05, "loss": 0.14577803611755372, "step": 18430 }, { "epoch": 0.48573771085567147, "grad_norm": 3.391089677810669, "learning_rate": 1.0947266226621298e-05, "loss": 0.1347484827041626, "step": 18440 }, { "epoch": 0.4860011261001702, "grad_norm": 0.869446873664856, "learning_rate": 1.0938772873124321e-05, "loss": 0.15647361278533936, "step": 18450 }, { "epoch": 0.48626454134466895, "grad_norm": 1.2492997646331787, "learning_rate": 1.0930278836347943e-05, "loss": 0.14130815267562866, "step": 18460 }, { "epoch": 0.4865279565891677, "grad_norm": 1.153387188911438, "learning_rate": 1.0921784122474495e-05, "loss": 0.1477504253387451, "step": 18470 }, { "epoch": 0.48679137183366644, "grad_norm": 2.4675827026367188, "learning_rate": 1.0913288737686789e-05, "loss": 0.15981907844543458, "step": 18480 }, { "epoch": 0.4870547870781652, "grad_norm": 1.9880423545837402, "learning_rate": 1.090479268816814e-05, "loss": 0.15468873977661132, "step": 18490 }, { "epoch": 0.48731820232266393, "grad_norm": 2.8381969928741455, "learning_rate": 1.0896295980102334e-05, "loss": 0.15858993530273438, "step": 18500 }, { "epoch": 0.4875816175671627, "grad_norm": 1.611324667930603, "learning_rate": 1.0887798619673648e-05, "loss": 0.14372290372848512, "step": 18510 }, { "epoch": 0.4878450328116614, "grad_norm": 1.1751866340637207, "learning_rate": 1.087930061306682e-05, "loss": 0.1512036442756653, "step": 18520 }, { "epoch": 0.4881084480561601, "grad_norm": 0.9670770764350891, "learning_rate": 1.0870801966467073e-05, "loss": 0.15630772113800048, "step": 18530 }, { "epoch": 0.48837186330065885, "grad_norm": 1.6376339197158813, "learning_rate": 1.0862302686060082e-05, "loss": 0.14483993053436278, "step": 18540 }, { "epoch": 0.4886352785451576, "grad_norm": 2.982804536819458, "learning_rate": 1.0853802778031995e-05, "loss": 0.15326886177062987, "step": 18550 }, { "epoch": 0.48889869378965634, "grad_norm": 4.29051399230957, "learning_rate": 1.0845302248569411e-05, "loss": 0.15446577072143555, "step": 18560 }, { "epoch": 0.4891621090341551, "grad_norm": 1.9008228778839111, "learning_rate": 1.0836801103859377e-05, "loss": 0.1548550009727478, "step": 18570 }, { "epoch": 0.4894255242786538, "grad_norm": 1.3289551734924316, "learning_rate": 1.0828299350089401e-05, "loss": 0.14465038776397704, "step": 18580 }, { "epoch": 0.48968893952315257, "grad_norm": 1.7853409051895142, "learning_rate": 1.0819796993447422e-05, "loss": 0.15053610801696776, "step": 18590 }, { "epoch": 0.4899523547676513, "grad_norm": 2.042677164077759, "learning_rate": 1.081129404012182e-05, "loss": 0.15019314289093016, "step": 18600 }, { "epoch": 0.49021577001215005, "grad_norm": 2.2787041664123535, "learning_rate": 1.0802790496301414e-05, "loss": 0.15574630498886108, "step": 18610 }, { "epoch": 0.49047918525664874, "grad_norm": 2.757503032684326, "learning_rate": 1.079428636817545e-05, "loss": 0.14752404689788817, "step": 18620 }, { "epoch": 0.4907426005011475, "grad_norm": 0.6755043268203735, "learning_rate": 1.0785781661933596e-05, "loss": 0.14444341659545898, "step": 18630 }, { "epoch": 0.49100601574564623, "grad_norm": 0.915384829044342, "learning_rate": 1.0777276383765953e-05, "loss": 0.1471014976501465, "step": 18640 }, { "epoch": 0.491269430990145, "grad_norm": 1.656694769859314, "learning_rate": 1.0768770539863017e-05, "loss": 0.15762451887130738, "step": 18650 }, { "epoch": 0.4915328462346437, "grad_norm": 0.9779417514801025, "learning_rate": 1.076026413641572e-05, "loss": 0.15299077033996583, "step": 18660 }, { "epoch": 0.49179626147914246, "grad_norm": 1.7630821466445923, "learning_rate": 1.0751757179615378e-05, "loss": 0.16221768856048585, "step": 18670 }, { "epoch": 0.4920596767236412, "grad_norm": 3.8173680305480957, "learning_rate": 1.0743249675653733e-05, "loss": 0.16032781600952148, "step": 18680 }, { "epoch": 0.49232309196813995, "grad_norm": 1.2800040245056152, "learning_rate": 1.0734741630722904e-05, "loss": 0.15514081716537476, "step": 18690 }, { "epoch": 0.49258650721263864, "grad_norm": 0.9110012054443359, "learning_rate": 1.0726233051015415e-05, "loss": 0.1502981185913086, "step": 18700 }, { "epoch": 0.4928499224571374, "grad_norm": 2.3433170318603516, "learning_rate": 1.0717723942724182e-05, "loss": 0.14103347063064575, "step": 18710 }, { "epoch": 0.4931133377016361, "grad_norm": 2.759817600250244, "learning_rate": 1.07092143120425e-05, "loss": 0.14331669807434083, "step": 18720 }, { "epoch": 0.49337675294613487, "grad_norm": 4.010676860809326, "learning_rate": 1.0700704165164042e-05, "loss": 0.14215950965881347, "step": 18730 }, { "epoch": 0.4936401681906336, "grad_norm": 2.2685611248016357, "learning_rate": 1.069219350828286e-05, "loss": 0.16013188362121583, "step": 18740 }, { "epoch": 0.49390358343513235, "grad_norm": 1.5485634803771973, "learning_rate": 1.068368234759338e-05, "loss": 0.14823615550994873, "step": 18750 }, { "epoch": 0.4941669986796311, "grad_norm": 1.7238999605178833, "learning_rate": 1.0675170689290393e-05, "loss": 0.15622180700302124, "step": 18760 }, { "epoch": 0.49443041392412984, "grad_norm": 2.6966943740844727, "learning_rate": 1.0666658539569048e-05, "loss": 0.15169110298156738, "step": 18770 }, { "epoch": 0.4946938291686286, "grad_norm": 3.105707883834839, "learning_rate": 1.0658145904624852e-05, "loss": 0.16437352895736695, "step": 18780 }, { "epoch": 0.4949572444131273, "grad_norm": 2.3432979583740234, "learning_rate": 1.0649632790653677e-05, "loss": 0.1516926407814026, "step": 18790 }, { "epoch": 0.495220659657626, "grad_norm": 1.261526346206665, "learning_rate": 1.0641119203851724e-05, "loss": 0.15418943166732788, "step": 18800 }, { "epoch": 0.49548407490212476, "grad_norm": 1.4165865182876587, "learning_rate": 1.0632605150415561e-05, "loss": 0.16214652061462403, "step": 18810 }, { "epoch": 0.4957474901466235, "grad_norm": 1.789797306060791, "learning_rate": 1.062409063654207e-05, "loss": 0.1630618691444397, "step": 18820 }, { "epoch": 0.49601090539112225, "grad_norm": 5.5591959953308105, "learning_rate": 1.0615575668428494e-05, "loss": 0.1461625576019287, "step": 18830 }, { "epoch": 0.496274320635621, "grad_norm": 1.1392030715942383, "learning_rate": 1.0607060252272383e-05, "loss": 0.14868592023849486, "step": 18840 }, { "epoch": 0.49653773588011973, "grad_norm": 3.048738718032837, "learning_rate": 1.0598544394271633e-05, "loss": 0.13674777746200562, "step": 18850 }, { "epoch": 0.4968011511246185, "grad_norm": 1.4911168813705444, "learning_rate": 1.0590028100624448e-05, "loss": 0.1489378809928894, "step": 18860 }, { "epoch": 0.4970645663691172, "grad_norm": 3.7172439098358154, "learning_rate": 1.0581511377529354e-05, "loss": 0.16072983741760255, "step": 18870 }, { "epoch": 0.4973279816136159, "grad_norm": 1.3517475128173828, "learning_rate": 1.0572994231185194e-05, "loss": 0.1633277177810669, "step": 18880 }, { "epoch": 0.49759139685811465, "grad_norm": 0.8299849629402161, "learning_rate": 1.056447666779111e-05, "loss": 0.15460147857666015, "step": 18890 }, { "epoch": 0.4978548121026134, "grad_norm": 1.6908766031265259, "learning_rate": 1.0555958693546552e-05, "loss": 0.14803993701934814, "step": 18900 }, { "epoch": 0.49811822734711214, "grad_norm": 1.4406987428665161, "learning_rate": 1.0547440314651274e-05, "loss": 0.1344010829925537, "step": 18910 }, { "epoch": 0.4983816425916109, "grad_norm": 2.286829710006714, "learning_rate": 1.0538921537305315e-05, "loss": 0.1612061381340027, "step": 18920 }, { "epoch": 0.49864505783610963, "grad_norm": 3.5775227546691895, "learning_rate": 1.0530402367709007e-05, "loss": 0.15201566219329835, "step": 18930 }, { "epoch": 0.4989084730806084, "grad_norm": 2.3842101097106934, "learning_rate": 1.0521882812062978e-05, "loss": 0.1476964235305786, "step": 18940 }, { "epoch": 0.4991718883251071, "grad_norm": 1.7394089698791504, "learning_rate": 1.0513362876568118e-05, "loss": 0.14842209815979004, "step": 18950 }, { "epoch": 0.49943530356960586, "grad_norm": 3.1617345809936523, "learning_rate": 1.0504842567425611e-05, "loss": 0.16001513004302978, "step": 18960 }, { "epoch": 0.49969871881410455, "grad_norm": 0.8326540589332581, "learning_rate": 1.0496321890836901e-05, "loss": 0.15042771100997926, "step": 18970 }, { "epoch": 0.4999621340586033, "grad_norm": 1.9427924156188965, "learning_rate": 1.0487800853003706e-05, "loss": 0.1521357297897339, "step": 18980 }, { "epoch": 0.500225549303102, "grad_norm": 2.746302366256714, "learning_rate": 1.0479279460128005e-05, "loss": 0.15350751876831054, "step": 18990 }, { "epoch": 0.5004889645476008, "grad_norm": 2.6246120929718018, "learning_rate": 1.0470757718412035e-05, "loss": 0.1438712239265442, "step": 19000 }, { "epoch": 0.5007523797920995, "grad_norm": 1.7106847763061523, "learning_rate": 1.0462235634058287e-05, "loss": 0.14541152715682984, "step": 19010 }, { "epoch": 0.5010157950365982, "grad_norm": 1.7620234489440918, "learning_rate": 1.0453713213269503e-05, "loss": 0.15672210454940796, "step": 19020 }, { "epoch": 0.501279210281097, "grad_norm": 1.8717246055603027, "learning_rate": 1.0445190462248667e-05, "loss": 0.14734511375427245, "step": 19030 }, { "epoch": 0.5015426255255957, "grad_norm": 2.783531427383423, "learning_rate": 1.0436667387199004e-05, "loss": 0.1523361921310425, "step": 19040 }, { "epoch": 0.5018060407700945, "grad_norm": 1.9270339012145996, "learning_rate": 1.0428143994323978e-05, "loss": 0.15495588779449462, "step": 19050 }, { "epoch": 0.5020694560145932, "grad_norm": 1.9313080310821533, "learning_rate": 1.041962028982728e-05, "loss": 0.1477288246154785, "step": 19060 }, { "epoch": 0.502332871259092, "grad_norm": 3.112851619720459, "learning_rate": 1.0411096279912835e-05, "loss": 0.16215767860412597, "step": 19070 }, { "epoch": 0.5025962865035907, "grad_norm": 1.3337591886520386, "learning_rate": 1.0402571970784778e-05, "loss": 0.15485882759094238, "step": 19080 }, { "epoch": 0.5028597017480895, "grad_norm": 1.2421989440917969, "learning_rate": 1.0394047368647473e-05, "loss": 0.1410317063331604, "step": 19090 }, { "epoch": 0.5031231169925882, "grad_norm": 2.0490550994873047, "learning_rate": 1.0385522479705493e-05, "loss": 0.14032626152038574, "step": 19100 }, { "epoch": 0.5033865322370868, "grad_norm": 0.8578566908836365, "learning_rate": 1.0376997310163622e-05, "loss": 0.1535954713821411, "step": 19110 }, { "epoch": 0.5036499474815856, "grad_norm": 1.4236096143722534, "learning_rate": 1.0368471866226837e-05, "loss": 0.14511040449142457, "step": 19120 }, { "epoch": 0.5039133627260843, "grad_norm": 4.904054641723633, "learning_rate": 1.0359946154100336e-05, "loss": 0.14747812747955322, "step": 19130 }, { "epoch": 0.5041767779705831, "grad_norm": 1.458019495010376, "learning_rate": 1.035142017998949e-05, "loss": 0.14835655689239502, "step": 19140 }, { "epoch": 0.5044401932150818, "grad_norm": 1.1402918100357056, "learning_rate": 1.0342893950099879e-05, "loss": 0.15296945571899415, "step": 19150 }, { "epoch": 0.5047036084595806, "grad_norm": 1.2650742530822754, "learning_rate": 1.0334367470637257e-05, "loss": 0.14838608503341674, "step": 19160 }, { "epoch": 0.5049670237040793, "grad_norm": 0.8469447493553162, "learning_rate": 1.0325840747807562e-05, "loss": 0.14514410495758057, "step": 19170 }, { "epoch": 0.505230438948578, "grad_norm": 1.9592435359954834, "learning_rate": 1.0317313787816915e-05, "loss": 0.1569833278656006, "step": 19180 }, { "epoch": 0.5054938541930768, "grad_norm": 2.1361491680145264, "learning_rate": 1.0308786596871606e-05, "loss": 0.15659935474395753, "step": 19190 }, { "epoch": 0.5057572694375755, "grad_norm": 1.7959967851638794, "learning_rate": 1.0300259181178086e-05, "loss": 0.14818015098571777, "step": 19200 }, { "epoch": 0.5060206846820743, "grad_norm": 0.7847166061401367, "learning_rate": 1.0291731546942977e-05, "loss": 0.14268791675567627, "step": 19210 }, { "epoch": 0.506284099926573, "grad_norm": 1.2981293201446533, "learning_rate": 1.0283203700373065e-05, "loss": 0.14852484464645385, "step": 19220 }, { "epoch": 0.5065475151710718, "grad_norm": 0.8303238153457642, "learning_rate": 1.0274675647675281e-05, "loss": 0.15977600812911988, "step": 19230 }, { "epoch": 0.5068109304155705, "grad_norm": 2.968410015106201, "learning_rate": 1.0266147395056712e-05, "loss": 0.1515573501586914, "step": 19240 }, { "epoch": 0.5070743456600693, "grad_norm": 2.204643964767456, "learning_rate": 1.0257618948724583e-05, "loss": 0.1522369146347046, "step": 19250 }, { "epoch": 0.507337760904568, "grad_norm": 1.608620285987854, "learning_rate": 1.024909031488627e-05, "loss": 0.15812282562255858, "step": 19260 }, { "epoch": 0.5076011761490666, "grad_norm": 1.6134130954742432, "learning_rate": 1.0240561499749278e-05, "loss": 0.13889663219451903, "step": 19270 }, { "epoch": 0.5078645913935654, "grad_norm": 1.2370295524597168, "learning_rate": 1.023203250952125e-05, "loss": 0.14642223119735717, "step": 19280 }, { "epoch": 0.5081280066380641, "grad_norm": 1.8005298376083374, "learning_rate": 1.0223503350409949e-05, "loss": 0.14440598487854003, "step": 19290 }, { "epoch": 0.5083914218825629, "grad_norm": 1.054296612739563, "learning_rate": 1.0214974028623268e-05, "loss": 0.1622214674949646, "step": 19300 }, { "epoch": 0.5086548371270616, "grad_norm": 1.568660855293274, "learning_rate": 1.0206444550369212e-05, "loss": 0.16154277324676514, "step": 19310 }, { "epoch": 0.5089182523715604, "grad_norm": 1.6731512546539307, "learning_rate": 1.0197914921855909e-05, "loss": 0.15122551918029786, "step": 19320 }, { "epoch": 0.5091816676160591, "grad_norm": 1.396100401878357, "learning_rate": 1.0189385149291586e-05, "loss": 0.1547834038734436, "step": 19330 }, { "epoch": 0.5094450828605579, "grad_norm": 1.0740479230880737, "learning_rate": 1.018085523888458e-05, "loss": 0.14531333446502687, "step": 19340 }, { "epoch": 0.5097084981050566, "grad_norm": 4.050624847412109, "learning_rate": 1.0172325196843328e-05, "loss": 0.1457369327545166, "step": 19350 }, { "epoch": 0.5099719133495553, "grad_norm": 1.4262527227401733, "learning_rate": 1.0163795029376362e-05, "loss": 0.14113917350769042, "step": 19360 }, { "epoch": 0.5102353285940541, "grad_norm": 2.2941229343414307, "learning_rate": 1.015526474269231e-05, "loss": 0.14933968782424928, "step": 19370 }, { "epoch": 0.5104987438385528, "grad_norm": 2.9002606868743896, "learning_rate": 1.0146734342999875e-05, "loss": 0.1446614980697632, "step": 19380 }, { "epoch": 0.5107621590830516, "grad_norm": 1.9593751430511475, "learning_rate": 1.0138203836507859e-05, "loss": 0.15716607570648194, "step": 19390 }, { "epoch": 0.5110255743275502, "grad_norm": 1.8741748332977295, "learning_rate": 1.0129673229425125e-05, "loss": 0.1550332188606262, "step": 19400 }, { "epoch": 0.511288989572049, "grad_norm": 2.5224409103393555, "learning_rate": 1.0121142527960624e-05, "loss": 0.13862007856369019, "step": 19410 }, { "epoch": 0.5115524048165477, "grad_norm": 2.511568069458008, "learning_rate": 1.0112611738323362e-05, "loss": 0.13975961208343507, "step": 19420 }, { "epoch": 0.5118158200610465, "grad_norm": 1.8610302209854126, "learning_rate": 1.0104080866722419e-05, "loss": 0.1371997356414795, "step": 19430 }, { "epoch": 0.5120792353055452, "grad_norm": 2.843611240386963, "learning_rate": 1.0095549919366928e-05, "loss": 0.1363823413848877, "step": 19440 }, { "epoch": 0.5123426505500439, "grad_norm": 1.7104110717773438, "learning_rate": 1.0087018902466086e-05, "loss": 0.16177875995635987, "step": 19450 }, { "epoch": 0.5126060657945427, "grad_norm": 3.0086376667022705, "learning_rate": 1.0078487822229132e-05, "loss": 0.16183071136474608, "step": 19460 }, { "epoch": 0.5128694810390414, "grad_norm": 0.9233871698379517, "learning_rate": 1.0069956684865354e-05, "loss": 0.15211894512176513, "step": 19470 }, { "epoch": 0.5131328962835402, "grad_norm": 1.892513632774353, "learning_rate": 1.0061425496584084e-05, "loss": 0.14478483200073242, "step": 19480 }, { "epoch": 0.5133963115280389, "grad_norm": 3.5502545833587646, "learning_rate": 1.0052894263594687e-05, "loss": 0.13465322256088258, "step": 19490 }, { "epoch": 0.5136597267725377, "grad_norm": 1.4764572381973267, "learning_rate": 1.0044362992106563e-05, "loss": 0.14417703151702882, "step": 19500 }, { "epoch": 0.5139231420170364, "grad_norm": 1.198499321937561, "learning_rate": 1.0035831688329139e-05, "loss": 0.14684696197509767, "step": 19510 }, { "epoch": 0.5141865572615352, "grad_norm": 1.8289612531661987, "learning_rate": 1.0027300358471865e-05, "loss": 0.14994525909423828, "step": 19520 }, { "epoch": 0.5144499725060339, "grad_norm": 2.885932683944702, "learning_rate": 1.0018769008744214e-05, "loss": 0.15793485641479493, "step": 19530 }, { "epoch": 0.5147133877505325, "grad_norm": 1.2779453992843628, "learning_rate": 1.0010237645355669e-05, "loss": 0.15395740270614625, "step": 19540 }, { "epoch": 0.5149768029950313, "grad_norm": 1.849679946899414, "learning_rate": 1.0001706274515724e-05, "loss": 0.15658326148986818, "step": 19550 }, { "epoch": 0.51524021823953, "grad_norm": 2.613114356994629, "learning_rate": 9.993174902433877e-06, "loss": 0.13815804719924926, "step": 19560 }, { "epoch": 0.5155036334840288, "grad_norm": 1.0161300897598267, "learning_rate": 9.984643535319632e-06, "loss": 0.15304181575775147, "step": 19570 }, { "epoch": 0.5157670487285275, "grad_norm": 4.472378730773926, "learning_rate": 9.976112179382483e-06, "loss": 0.1616726040840149, "step": 19580 }, { "epoch": 0.5160304639730263, "grad_norm": 1.02987802028656, "learning_rate": 9.967580840831922e-06, "loss": 0.15389981269836425, "step": 19590 }, { "epoch": 0.516293879217525, "grad_norm": 1.3153833150863647, "learning_rate": 9.959049525877425e-06, "loss": 0.15291297435760498, "step": 19600 }, { "epoch": 0.5165572944620238, "grad_norm": 2.620709180831909, "learning_rate": 9.950518240728446e-06, "loss": 0.157227623462677, "step": 19610 }, { "epoch": 0.5168207097065225, "grad_norm": 0.824658215045929, "learning_rate": 9.94198699159443e-06, "loss": 0.14949507713317872, "step": 19620 }, { "epoch": 0.5170841249510212, "grad_norm": 1.6348958015441895, "learning_rate": 9.93345578468478e-06, "loss": 0.15230648517608641, "step": 19630 }, { "epoch": 0.51734754019552, "grad_norm": 1.8029756546020508, "learning_rate": 9.924924626208884e-06, "loss": 0.14250258207321168, "step": 19640 }, { "epoch": 0.5176109554400187, "grad_norm": 1.1978631019592285, "learning_rate": 9.916393522376087e-06, "loss": 0.15387731790542603, "step": 19650 }, { "epoch": 0.5178743706845175, "grad_norm": 2.985598564147949, "learning_rate": 9.907862479395684e-06, "loss": 0.15196164846420288, "step": 19660 }, { "epoch": 0.5181377859290162, "grad_norm": 2.2986645698547363, "learning_rate": 9.899331503476951e-06, "loss": 0.15603492259979249, "step": 19670 }, { "epoch": 0.518401201173515, "grad_norm": 1.6402835845947266, "learning_rate": 9.890800600829086e-06, "loss": 0.14013367891311646, "step": 19680 }, { "epoch": 0.5186646164180136, "grad_norm": 3.0208606719970703, "learning_rate": 9.882269777661261e-06, "loss": 0.14061453342437744, "step": 19690 }, { "epoch": 0.5189280316625123, "grad_norm": 1.3027905225753784, "learning_rate": 9.873739040182569e-06, "loss": 0.14117958545684814, "step": 19700 }, { "epoch": 0.5191914469070111, "grad_norm": 1.7234883308410645, "learning_rate": 9.86520839460205e-06, "loss": 0.16160013675689697, "step": 19710 }, { "epoch": 0.5194548621515098, "grad_norm": 3.377652406692505, "learning_rate": 9.856677847128678e-06, "loss": 0.1384030818939209, "step": 19720 }, { "epoch": 0.5197182773960086, "grad_norm": 1.3195608854293823, "learning_rate": 9.848147403971353e-06, "loss": 0.14494949579238892, "step": 19730 }, { "epoch": 0.5199816926405073, "grad_norm": 3.6491165161132812, "learning_rate": 9.839617071338897e-06, "loss": 0.1516500949859619, "step": 19740 }, { "epoch": 0.5202451078850061, "grad_norm": 1.8058571815490723, "learning_rate": 9.831086855440056e-06, "loss": 0.15849080085754394, "step": 19750 }, { "epoch": 0.5205085231295048, "grad_norm": 1.5765862464904785, "learning_rate": 9.82255676248349e-06, "loss": 0.139889132976532, "step": 19760 }, { "epoch": 0.5207719383740036, "grad_norm": 3.595402956008911, "learning_rate": 9.814026798677767e-06, "loss": 0.14078807830810547, "step": 19770 }, { "epoch": 0.5210353536185023, "grad_norm": 0.9130980968475342, "learning_rate": 9.805496970231367e-06, "loss": 0.13916926383972167, "step": 19780 }, { "epoch": 0.521298768863001, "grad_norm": 0.8482924699783325, "learning_rate": 9.796967283352657e-06, "loss": 0.1416097640991211, "step": 19790 }, { "epoch": 0.5215621841074998, "grad_norm": 1.488692045211792, "learning_rate": 9.788437744249925e-06, "loss": 0.1477830410003662, "step": 19800 }, { "epoch": 0.5218255993519985, "grad_norm": 2.4786717891693115, "learning_rate": 9.779908359131327e-06, "loss": 0.14272873401641845, "step": 19810 }, { "epoch": 0.5220890145964973, "grad_norm": 1.0257289409637451, "learning_rate": 9.77137913420492e-06, "loss": 0.13878140449523926, "step": 19820 }, { "epoch": 0.522352429840996, "grad_norm": 1.1079179048538208, "learning_rate": 9.762850075678644e-06, "loss": 0.1573284387588501, "step": 19830 }, { "epoch": 0.5226158450854947, "grad_norm": 3.531754970550537, "learning_rate": 9.754321189760312e-06, "loss": 0.13955872058868407, "step": 19840 }, { "epoch": 0.5228792603299934, "grad_norm": 1.1177923679351807, "learning_rate": 9.74579248265762e-06, "loss": 0.13908709287643434, "step": 19850 }, { "epoch": 0.5231426755744922, "grad_norm": 0.8094909191131592, "learning_rate": 9.737263960578127e-06, "loss": 0.15120973587036132, "step": 19860 }, { "epoch": 0.5234060908189909, "grad_norm": 1.1501209735870361, "learning_rate": 9.728735629729256e-06, "loss": 0.15018682479858397, "step": 19870 }, { "epoch": 0.5236695060634896, "grad_norm": 1.8736356496810913, "learning_rate": 9.7202074963183e-06, "loss": 0.1488809108734131, "step": 19880 }, { "epoch": 0.5239329213079884, "grad_norm": 1.4790704250335693, "learning_rate": 9.711679566552396e-06, "loss": 0.14634426832199096, "step": 19890 }, { "epoch": 0.5241963365524871, "grad_norm": 2.043567419052124, "learning_rate": 9.703151846638545e-06, "loss": 0.15305771827697753, "step": 19900 }, { "epoch": 0.5244597517969859, "grad_norm": 1.76250159740448, "learning_rate": 9.694624342783591e-06, "loss": 0.1511755108833313, "step": 19910 }, { "epoch": 0.5247231670414846, "grad_norm": 1.234163761138916, "learning_rate": 9.686097061194209e-06, "loss": 0.16338231563568115, "step": 19920 }, { "epoch": 0.5249865822859834, "grad_norm": 2.0679187774658203, "learning_rate": 9.677570008076933e-06, "loss": 0.14196035861968995, "step": 19930 }, { "epoch": 0.5252499975304821, "grad_norm": 1.971527099609375, "learning_rate": 9.669043189638115e-06, "loss": 0.1464618682861328, "step": 19940 }, { "epoch": 0.5255134127749809, "grad_norm": 1.2873704433441162, "learning_rate": 9.66051661208394e-06, "loss": 0.1609327793121338, "step": 19950 }, { "epoch": 0.5257768280194796, "grad_norm": 1.7200616598129272, "learning_rate": 9.651990281620423e-06, "loss": 0.1386871099472046, "step": 19960 }, { "epoch": 0.5260402432639782, "grad_norm": 3.5366711616516113, "learning_rate": 9.643464204453392e-06, "loss": 0.14305496215820312, "step": 19970 }, { "epoch": 0.526303658508477, "grad_norm": 2.8148374557495117, "learning_rate": 9.634938386788497e-06, "loss": 0.1639392375946045, "step": 19980 }, { "epoch": 0.5265670737529757, "grad_norm": 1.6736161708831787, "learning_rate": 9.626412834831193e-06, "loss": 0.154992413520813, "step": 19990 }, { "epoch": 0.5268304889974745, "grad_norm": 2.865802526473999, "learning_rate": 9.617887554786746e-06, "loss": 0.13268111944198607, "step": 20000 }, { "epoch": 0.5270939042419732, "grad_norm": 2.3740782737731934, "learning_rate": 9.609362552860225e-06, "loss": 0.13908073902130128, "step": 20010 }, { "epoch": 0.527357319486472, "grad_norm": 2.3607778549194336, "learning_rate": 9.600837835256486e-06, "loss": 0.1475742816925049, "step": 20020 }, { "epoch": 0.5276207347309707, "grad_norm": 2.524005174636841, "learning_rate": 9.592313408180199e-06, "loss": 0.13441228866577148, "step": 20030 }, { "epoch": 0.5278841499754695, "grad_norm": 5.77451229095459, "learning_rate": 9.583789277835798e-06, "loss": 0.16155765056610108, "step": 20040 }, { "epoch": 0.5281475652199682, "grad_norm": 1.6869187355041504, "learning_rate": 9.575265450427517e-06, "loss": 0.15693260431289674, "step": 20050 }, { "epoch": 0.5284109804644669, "grad_norm": 1.410447120666504, "learning_rate": 9.566741932159369e-06, "loss": 0.15283230543136597, "step": 20060 }, { "epoch": 0.5286743957089657, "grad_norm": 1.2426077127456665, "learning_rate": 9.558218729235132e-06, "loss": 0.15905451774597168, "step": 20070 }, { "epoch": 0.5289378109534644, "grad_norm": 4.140406608581543, "learning_rate": 9.549695847858363e-06, "loss": 0.15052201747894287, "step": 20080 }, { "epoch": 0.5292012261979632, "grad_norm": 1.9783015251159668, "learning_rate": 9.541173294232387e-06, "loss": 0.15466305017471313, "step": 20090 }, { "epoch": 0.5294646414424619, "grad_norm": 1.0121181011199951, "learning_rate": 9.53265107456028e-06, "loss": 0.15120456218719483, "step": 20100 }, { "epoch": 0.5297280566869607, "grad_norm": 0.7986055612564087, "learning_rate": 9.524129195044886e-06, "loss": 0.13761502504348755, "step": 20110 }, { "epoch": 0.5299914719314593, "grad_norm": 2.9031240940093994, "learning_rate": 9.515607661888799e-06, "loss": 0.15223073959350586, "step": 20120 }, { "epoch": 0.5302548871759581, "grad_norm": 2.55381441116333, "learning_rate": 9.507086481294345e-06, "loss": 0.17392330169677733, "step": 20130 }, { "epoch": 0.5305183024204568, "grad_norm": 1.0957683324813843, "learning_rate": 9.498565659463624e-06, "loss": 0.14546335935592652, "step": 20140 }, { "epoch": 0.5307817176649555, "grad_norm": 1.0752477645874023, "learning_rate": 9.490045202598442e-06, "loss": 0.142478609085083, "step": 20150 }, { "epoch": 0.5310451329094543, "grad_norm": 1.4033870697021484, "learning_rate": 9.48152511690037e-06, "loss": 0.1482825517654419, "step": 20160 }, { "epoch": 0.531308548153953, "grad_norm": 2.280871629714966, "learning_rate": 9.473005408570685e-06, "loss": 0.144431734085083, "step": 20170 }, { "epoch": 0.5315719633984518, "grad_norm": 1.79402494430542, "learning_rate": 9.464486083810397e-06, "loss": 0.13846700191497802, "step": 20180 }, { "epoch": 0.5318353786429505, "grad_norm": 2.440382242202759, "learning_rate": 9.455967148820243e-06, "loss": 0.14434008598327636, "step": 20190 }, { "epoch": 0.5320987938874493, "grad_norm": 3.5710549354553223, "learning_rate": 9.44744860980067e-06, "loss": 0.15313129425048827, "step": 20200 }, { "epoch": 0.532362209131948, "grad_norm": 1.4785511493682861, "learning_rate": 9.438930472951838e-06, "loss": 0.13587828874588012, "step": 20210 }, { "epoch": 0.5326256243764467, "grad_norm": 1.7744923830032349, "learning_rate": 9.430412744473615e-06, "loss": 0.14387166500091553, "step": 20220 }, { "epoch": 0.5328890396209455, "grad_norm": 1.7108676433563232, "learning_rate": 9.421895430565568e-06, "loss": 0.14803094863891603, "step": 20230 }, { "epoch": 0.5331524548654442, "grad_norm": 2.300088882446289, "learning_rate": 9.413378537426972e-06, "loss": 0.1413062572479248, "step": 20240 }, { "epoch": 0.533415870109943, "grad_norm": 1.1075398921966553, "learning_rate": 9.404862071256785e-06, "loss": 0.14312875270843506, "step": 20250 }, { "epoch": 0.5336792853544416, "grad_norm": 2.4182443618774414, "learning_rate": 9.396346038253654e-06, "loss": 0.1620453953742981, "step": 20260 }, { "epoch": 0.5339427005989404, "grad_norm": 4.313726902008057, "learning_rate": 9.387830444615926e-06, "loss": 0.13847110271453858, "step": 20270 }, { "epoch": 0.5342061158434391, "grad_norm": 5.126481533050537, "learning_rate": 9.379315296541611e-06, "loss": 0.14250104427337645, "step": 20280 }, { "epoch": 0.5344695310879379, "grad_norm": 1.8366636037826538, "learning_rate": 9.370800600228399e-06, "loss": 0.1512735366821289, "step": 20290 }, { "epoch": 0.5347329463324366, "grad_norm": 1.0646424293518066, "learning_rate": 9.362286361873658e-06, "loss": 0.15154638290405273, "step": 20300 }, { "epoch": 0.5349963615769353, "grad_norm": 2.090090274810791, "learning_rate": 9.353772587674413e-06, "loss": 0.1504178524017334, "step": 20310 }, { "epoch": 0.5352597768214341, "grad_norm": 1.877052903175354, "learning_rate": 9.345259283827362e-06, "loss": 0.14119358062744142, "step": 20320 }, { "epoch": 0.5355231920659328, "grad_norm": 1.785345435142517, "learning_rate": 9.336746456528853e-06, "loss": 0.14522836208343506, "step": 20330 }, { "epoch": 0.5357866073104316, "grad_norm": 0.8245441317558289, "learning_rate": 9.328234111974885e-06, "loss": 0.13902480602264405, "step": 20340 }, { "epoch": 0.5360500225549303, "grad_norm": 0.8620415925979614, "learning_rate": 9.319722256361117e-06, "loss": 0.1563551664352417, "step": 20350 }, { "epoch": 0.5363134377994291, "grad_norm": 5.289443492889404, "learning_rate": 9.311210895882835e-06, "loss": 0.1485649824142456, "step": 20360 }, { "epoch": 0.5365768530439278, "grad_norm": 1.8197811841964722, "learning_rate": 9.302700036734985e-06, "loss": 0.14446600675582885, "step": 20370 }, { "epoch": 0.5368402682884266, "grad_norm": 1.672603726387024, "learning_rate": 9.294189685112133e-06, "loss": 0.14907171726226806, "step": 20380 }, { "epoch": 0.5371036835329253, "grad_norm": 1.4667975902557373, "learning_rate": 9.285679847208474e-06, "loss": 0.13331379890441894, "step": 20390 }, { "epoch": 0.537367098777424, "grad_norm": 1.722071647644043, "learning_rate": 9.277170529217847e-06, "loss": 0.1491900086402893, "step": 20400 }, { "epoch": 0.5376305140219227, "grad_norm": 0.7416393756866455, "learning_rate": 9.268661737333691e-06, "loss": 0.1400340437889099, "step": 20410 }, { "epoch": 0.5378939292664214, "grad_norm": 1.7606945037841797, "learning_rate": 9.260153477749072e-06, "loss": 0.14499454498291015, "step": 20420 }, { "epoch": 0.5381573445109202, "grad_norm": 2.338789224624634, "learning_rate": 9.251645756656676e-06, "loss": 0.1530721068382263, "step": 20430 }, { "epoch": 0.5384207597554189, "grad_norm": 1.9440562725067139, "learning_rate": 9.243138580248782e-06, "loss": 0.14249829053878785, "step": 20440 }, { "epoch": 0.5386841749999177, "grad_norm": 4.961899757385254, "learning_rate": 9.234631954717285e-06, "loss": 0.1511752963066101, "step": 20450 }, { "epoch": 0.5389475902444164, "grad_norm": 2.213035821914673, "learning_rate": 9.226125886253673e-06, "loss": 0.1517271637916565, "step": 20460 }, { "epoch": 0.5392110054889152, "grad_norm": 0.8693720102310181, "learning_rate": 9.217620381049026e-06, "loss": 0.14499667882919312, "step": 20470 }, { "epoch": 0.5394744207334139, "grad_norm": 1.2049248218536377, "learning_rate": 9.209115445294022e-06, "loss": 0.1421457290649414, "step": 20480 }, { "epoch": 0.5397378359779126, "grad_norm": 0.8323115110397339, "learning_rate": 9.200611085178917e-06, "loss": 0.13983142375946045, "step": 20490 }, { "epoch": 0.5400012512224114, "grad_norm": 1.4219354391098022, "learning_rate": 9.192107306893556e-06, "loss": 0.14258501529693604, "step": 20500 }, { "epoch": 0.5402646664669101, "grad_norm": 4.796818733215332, "learning_rate": 9.183604116627354e-06, "loss": 0.15824518203735352, "step": 20510 }, { "epoch": 0.5405280817114089, "grad_norm": 1.548359751701355, "learning_rate": 9.175101520569293e-06, "loss": 0.15671970844268798, "step": 20520 }, { "epoch": 0.5407914969559076, "grad_norm": 2.5210814476013184, "learning_rate": 9.166599524907937e-06, "loss": 0.14572296142578126, "step": 20530 }, { "epoch": 0.5410549122004064, "grad_norm": 3.076974630355835, "learning_rate": 9.158098135831404e-06, "loss": 0.15168240070343017, "step": 20540 }, { "epoch": 0.541318327444905, "grad_norm": 2.0092933177948, "learning_rate": 9.149597359527368e-06, "loss": 0.14738544225692748, "step": 20550 }, { "epoch": 0.5415817426894038, "grad_norm": 1.1098583936691284, "learning_rate": 9.141097202183063e-06, "loss": 0.14872634410858154, "step": 20560 }, { "epoch": 0.5418451579339025, "grad_norm": 1.4816752672195435, "learning_rate": 9.13259766998527e-06, "loss": 0.13893131017684937, "step": 20570 }, { "epoch": 0.5421085731784012, "grad_norm": 2.752422332763672, "learning_rate": 9.124098769120318e-06, "loss": 0.14134843349456788, "step": 20580 }, { "epoch": 0.5423719884229, "grad_norm": 0.9620932936668396, "learning_rate": 9.115600505774072e-06, "loss": 0.14981493949890137, "step": 20590 }, { "epoch": 0.5426354036673987, "grad_norm": 1.3784884214401245, "learning_rate": 9.10710288613193e-06, "loss": 0.14732713699340821, "step": 20600 }, { "epoch": 0.5428988189118975, "grad_norm": 1.229271411895752, "learning_rate": 9.098605916378837e-06, "loss": 0.14359753131866454, "step": 20610 }, { "epoch": 0.5431622341563962, "grad_norm": 2.191663980484009, "learning_rate": 9.090109602699239e-06, "loss": 0.14115462303161622, "step": 20620 }, { "epoch": 0.543425649400895, "grad_norm": 3.1173908710479736, "learning_rate": 9.081613951277135e-06, "loss": 0.14498474597930908, "step": 20630 }, { "epoch": 0.5436890646453937, "grad_norm": 1.4801175594329834, "learning_rate": 9.07311896829602e-06, "loss": 0.1434754252433777, "step": 20640 }, { "epoch": 0.5439524798898925, "grad_norm": 1.4501726627349854, "learning_rate": 9.064624659938902e-06, "loss": 0.1362076997756958, "step": 20650 }, { "epoch": 0.5442158951343912, "grad_norm": 1.8334254026412964, "learning_rate": 9.056131032388318e-06, "loss": 0.15180710554122925, "step": 20660 }, { "epoch": 0.5444793103788899, "grad_norm": 3.5424880981445312, "learning_rate": 9.047638091826291e-06, "loss": 0.1454113245010376, "step": 20670 }, { "epoch": 0.5447427256233887, "grad_norm": 1.2134159803390503, "learning_rate": 9.039145844434344e-06, "loss": 0.1533818244934082, "step": 20680 }, { "epoch": 0.5450061408678873, "grad_norm": 2.3140311241149902, "learning_rate": 9.030654296393512e-06, "loss": 0.14346907138824463, "step": 20690 }, { "epoch": 0.5452695561123861, "grad_norm": 1.646511435508728, "learning_rate": 9.022163453884299e-06, "loss": 0.14062026739120484, "step": 20700 }, { "epoch": 0.5455329713568848, "grad_norm": 1.1660511493682861, "learning_rate": 9.013673323086716e-06, "loss": 0.1495838165283203, "step": 20710 }, { "epoch": 0.5457963866013836, "grad_norm": 2.5529732704162598, "learning_rate": 9.005183910180243e-06, "loss": 0.145039701461792, "step": 20720 }, { "epoch": 0.5460598018458823, "grad_norm": 1.7233998775482178, "learning_rate": 8.996695221343834e-06, "loss": 0.14628045558929442, "step": 20730 }, { "epoch": 0.546323217090381, "grad_norm": 1.9669753313064575, "learning_rate": 8.988207262755936e-06, "loss": 0.14334036111831666, "step": 20740 }, { "epoch": 0.5465866323348798, "grad_norm": 1.3599997758865356, "learning_rate": 8.979720040594438e-06, "loss": 0.15770106315612792, "step": 20750 }, { "epoch": 0.5468500475793785, "grad_norm": 3.371966600418091, "learning_rate": 8.971233561036718e-06, "loss": 0.13841230869293214, "step": 20760 }, { "epoch": 0.5471134628238773, "grad_norm": 1.259408950805664, "learning_rate": 8.962747830259598e-06, "loss": 0.14503049850463867, "step": 20770 }, { "epoch": 0.547376878068376, "grad_norm": 1.0769424438476562, "learning_rate": 8.954262854439354e-06, "loss": 0.1537938356399536, "step": 20780 }, { "epoch": 0.5476402933128748, "grad_norm": 1.7200392484664917, "learning_rate": 8.945778639751723e-06, "loss": 0.14934319257736206, "step": 20790 }, { "epoch": 0.5479037085573735, "grad_norm": 3.280984401702881, "learning_rate": 8.937295192371882e-06, "loss": 0.1335212230682373, "step": 20800 }, { "epoch": 0.5481671238018723, "grad_norm": 1.0212525129318237, "learning_rate": 8.928812518474445e-06, "loss": 0.1244123101234436, "step": 20810 }, { "epoch": 0.548430539046371, "grad_norm": 1.6531898975372314, "learning_rate": 8.920330624233477e-06, "loss": 0.13719420433044432, "step": 20820 }, { "epoch": 0.5486939542908696, "grad_norm": 2.2257697582244873, "learning_rate": 8.911849515822457e-06, "loss": 0.14972730875015258, "step": 20830 }, { "epoch": 0.5489573695353684, "grad_norm": 1.0479432344436646, "learning_rate": 8.903369199414311e-06, "loss": 0.149874746799469, "step": 20840 }, { "epoch": 0.5492207847798671, "grad_norm": 1.5384891033172607, "learning_rate": 8.894889681181374e-06, "loss": 0.14668365716934204, "step": 20850 }, { "epoch": 0.5494842000243659, "grad_norm": 3.0264432430267334, "learning_rate": 8.886410967295401e-06, "loss": 0.15152406692504883, "step": 20860 }, { "epoch": 0.5497476152688646, "grad_norm": 1.7717911005020142, "learning_rate": 8.877933063927576e-06, "loss": 0.13001163005828859, "step": 20870 }, { "epoch": 0.5500110305133634, "grad_norm": 2.142303466796875, "learning_rate": 8.869455977248473e-06, "loss": 0.14536246061325073, "step": 20880 }, { "epoch": 0.5502744457578621, "grad_norm": 2.4534153938293457, "learning_rate": 8.860979713428085e-06, "loss": 0.1338377594947815, "step": 20890 }, { "epoch": 0.5505378610023609, "grad_norm": 2.0292818546295166, "learning_rate": 8.852504278635806e-06, "loss": 0.14805428981781005, "step": 20900 }, { "epoch": 0.5508012762468596, "grad_norm": 1.809794545173645, "learning_rate": 8.844029679040418e-06, "loss": 0.13689656257629396, "step": 20910 }, { "epoch": 0.5510646914913583, "grad_norm": 2.222609043121338, "learning_rate": 8.835555920810104e-06, "loss": 0.13870773315429688, "step": 20920 }, { "epoch": 0.5513281067358571, "grad_norm": 1.631691336631775, "learning_rate": 8.827083010112429e-06, "loss": 0.12882883548736573, "step": 20930 }, { "epoch": 0.5515915219803558, "grad_norm": 1.0666322708129883, "learning_rate": 8.818610953114342e-06, "loss": 0.13509325981140136, "step": 20940 }, { "epoch": 0.5518549372248546, "grad_norm": 2.3072290420532227, "learning_rate": 8.810139755982173e-06, "loss": 0.13495441675186157, "step": 20950 }, { "epoch": 0.5521183524693533, "grad_norm": 2.660555839538574, "learning_rate": 8.801669424881624e-06, "loss": 0.15165935754776, "step": 20960 }, { "epoch": 0.5523817677138521, "grad_norm": 1.131927728652954, "learning_rate": 8.79319996597777e-06, "loss": 0.13886313438415526, "step": 20970 }, { "epoch": 0.5526451829583507, "grad_norm": 1.9273040294647217, "learning_rate": 8.784731385435049e-06, "loss": 0.13874428272247313, "step": 20980 }, { "epoch": 0.5529085982028495, "grad_norm": 1.5741939544677734, "learning_rate": 8.77626368941725e-06, "loss": 0.126040780544281, "step": 20990 }, { "epoch": 0.5531720134473482, "grad_norm": 1.7881758213043213, "learning_rate": 8.767796884087539e-06, "loss": 0.1411255717277527, "step": 21000 }, { "epoch": 0.5534354286918469, "grad_norm": 2.892745018005371, "learning_rate": 8.759330975608418e-06, "loss": 0.14488428831100464, "step": 21010 }, { "epoch": 0.5536988439363457, "grad_norm": 1.0806857347488403, "learning_rate": 8.750865970141736e-06, "loss": 0.13910698890686035, "step": 21020 }, { "epoch": 0.5539622591808444, "grad_norm": 1.7669082880020142, "learning_rate": 8.742401873848696e-06, "loss": 0.14037785530090333, "step": 21030 }, { "epoch": 0.5542256744253432, "grad_norm": 2.3522655963897705, "learning_rate": 8.733938692889827e-06, "loss": 0.14897757768630981, "step": 21040 }, { "epoch": 0.5544890896698419, "grad_norm": 1.6973563432693481, "learning_rate": 8.725476433425003e-06, "loss": 0.1434385061264038, "step": 21050 }, { "epoch": 0.5547525049143407, "grad_norm": 3.225053310394287, "learning_rate": 8.717015101613417e-06, "loss": 0.14656603336334229, "step": 21060 }, { "epoch": 0.5550159201588394, "grad_norm": 1.3370206356048584, "learning_rate": 8.708554703613592e-06, "loss": 0.14293451309204103, "step": 21070 }, { "epoch": 0.5552793354033382, "grad_norm": 1.8564631938934326, "learning_rate": 8.700095245583374e-06, "loss": 0.12805578708648682, "step": 21080 }, { "epoch": 0.5555427506478369, "grad_norm": 1.1603786945343018, "learning_rate": 8.691636733679915e-06, "loss": 0.14366674423217773, "step": 21090 }, { "epoch": 0.5558061658923356, "grad_norm": 1.2090137004852295, "learning_rate": 8.683179174059691e-06, "loss": 0.1457334041595459, "step": 21100 }, { "epoch": 0.5560695811368344, "grad_norm": 1.2610669136047363, "learning_rate": 8.67472257287848e-06, "loss": 0.14108891487121583, "step": 21110 }, { "epoch": 0.556332996381333, "grad_norm": 2.7410130500793457, "learning_rate": 8.666266936291354e-06, "loss": 0.1503204107284546, "step": 21120 }, { "epoch": 0.5565964116258318, "grad_norm": 2.227957248687744, "learning_rate": 8.657812270452695e-06, "loss": 0.15283974409103393, "step": 21130 }, { "epoch": 0.5568598268703305, "grad_norm": 2.946908950805664, "learning_rate": 8.649358581516171e-06, "loss": 0.14804325103759766, "step": 21140 }, { "epoch": 0.5571232421148293, "grad_norm": 1.3403877019882202, "learning_rate": 8.640905875634744e-06, "loss": 0.14461030960083007, "step": 21150 }, { "epoch": 0.557386657359328, "grad_norm": 2.2670738697052, "learning_rate": 8.632454158960656e-06, "loss": 0.14227975606918336, "step": 21160 }, { "epoch": 0.5576500726038268, "grad_norm": 2.3777215480804443, "learning_rate": 8.624003437645431e-06, "loss": 0.12164924144744874, "step": 21170 }, { "epoch": 0.5579134878483255, "grad_norm": 3.71161150932312, "learning_rate": 8.615553717839873e-06, "loss": 0.13403699398040772, "step": 21180 }, { "epoch": 0.5581769030928242, "grad_norm": 1.8379285335540771, "learning_rate": 8.607105005694048e-06, "loss": 0.15943078994750975, "step": 21190 }, { "epoch": 0.558440318337323, "grad_norm": 1.657612919807434, "learning_rate": 8.598657307357293e-06, "loss": 0.1456581711769104, "step": 21200 }, { "epoch": 0.5587037335818217, "grad_norm": 1.0790845155715942, "learning_rate": 8.590210628978213e-06, "loss": 0.13906172513961793, "step": 21210 }, { "epoch": 0.5589671488263205, "grad_norm": 2.354119300842285, "learning_rate": 8.581764976704653e-06, "loss": 0.1480809450149536, "step": 21220 }, { "epoch": 0.5592305640708192, "grad_norm": 1.6062363386154175, "learning_rate": 8.573320356683739e-06, "loss": 0.14497379064559937, "step": 21230 }, { "epoch": 0.559493979315318, "grad_norm": 1.5016268491744995, "learning_rate": 8.564876775061819e-06, "loss": 0.13598532676696778, "step": 21240 }, { "epoch": 0.5597573945598167, "grad_norm": 2.805615186691284, "learning_rate": 8.556434237984496e-06, "loss": 0.1308077573776245, "step": 21250 }, { "epoch": 0.5600208098043153, "grad_norm": 2.8405110836029053, "learning_rate": 8.547992751596617e-06, "loss": 0.15049387216567994, "step": 21260 }, { "epoch": 0.5602842250488141, "grad_norm": 1.0493501424789429, "learning_rate": 8.53955232204226e-06, "loss": 0.14464433193206788, "step": 21270 }, { "epoch": 0.5605476402933128, "grad_norm": 1.8412740230560303, "learning_rate": 8.531112955464728e-06, "loss": 0.1423697829246521, "step": 21280 }, { "epoch": 0.5608110555378116, "grad_norm": 0.902775764465332, "learning_rate": 8.522674658006564e-06, "loss": 0.13255139589309692, "step": 21290 }, { "epoch": 0.5610744707823103, "grad_norm": 1.8765844106674194, "learning_rate": 8.514237435809517e-06, "loss": 0.13392655849456786, "step": 21300 }, { "epoch": 0.5613378860268091, "grad_norm": 1.2442905902862549, "learning_rate": 8.505801295014569e-06, "loss": 0.12731893062591554, "step": 21310 }, { "epoch": 0.5616013012713078, "grad_norm": 2.8252649307250977, "learning_rate": 8.497366241761907e-06, "loss": 0.1440313935279846, "step": 21320 }, { "epoch": 0.5618647165158066, "grad_norm": 1.152809977531433, "learning_rate": 8.488932282190914e-06, "loss": 0.14209243059158325, "step": 21330 }, { "epoch": 0.5621281317603053, "grad_norm": 0.784774661064148, "learning_rate": 8.480499422440207e-06, "loss": 0.13682312965393068, "step": 21340 }, { "epoch": 0.562391547004804, "grad_norm": 2.2977607250213623, "learning_rate": 8.472067668647571e-06, "loss": 0.1422706127166748, "step": 21350 }, { "epoch": 0.5626549622493028, "grad_norm": 1.7862918376922607, "learning_rate": 8.463637026950013e-06, "loss": 0.1438812494277954, "step": 21360 }, { "epoch": 0.5629183774938015, "grad_norm": 1.2837775945663452, "learning_rate": 8.455207503483708e-06, "loss": 0.13855996131896972, "step": 21370 }, { "epoch": 0.5631817927383003, "grad_norm": 4.231773376464844, "learning_rate": 8.446779104384028e-06, "loss": 0.14958813190460205, "step": 21380 }, { "epoch": 0.563445207982799, "grad_norm": 1.7278904914855957, "learning_rate": 8.43835183578553e-06, "loss": 0.13348129987716675, "step": 21390 }, { "epoch": 0.5637086232272978, "grad_norm": 1.0349653959274292, "learning_rate": 8.429925703821941e-06, "loss": 0.15483717918395995, "step": 21400 }, { "epoch": 0.5639720384717964, "grad_norm": 2.9543776512145996, "learning_rate": 8.421500714626164e-06, "loss": 0.1491797685623169, "step": 21410 }, { "epoch": 0.5642354537162952, "grad_norm": 1.9067773818969727, "learning_rate": 8.413076874330266e-06, "loss": 0.14802258014678954, "step": 21420 }, { "epoch": 0.5644988689607939, "grad_norm": 1.248184323310852, "learning_rate": 8.404654189065486e-06, "loss": 0.14163333177566528, "step": 21430 }, { "epoch": 0.5647622842052926, "grad_norm": 1.245596170425415, "learning_rate": 8.396232664962218e-06, "loss": 0.15422275066375732, "step": 21440 }, { "epoch": 0.5650256994497914, "grad_norm": 0.7712500095367432, "learning_rate": 8.387812308150011e-06, "loss": 0.12276947498321533, "step": 21450 }, { "epoch": 0.5652891146942901, "grad_norm": 1.4611167907714844, "learning_rate": 8.379393124757554e-06, "loss": 0.15910027027130128, "step": 21460 }, { "epoch": 0.5655525299387889, "grad_norm": 1.2725504636764526, "learning_rate": 8.370975120912708e-06, "loss": 0.15093109607696534, "step": 21470 }, { "epoch": 0.5658159451832876, "grad_norm": 3.4717438220977783, "learning_rate": 8.362558302742447e-06, "loss": 0.1337334394454956, "step": 21480 }, { "epoch": 0.5660793604277864, "grad_norm": 3.5617318153381348, "learning_rate": 8.354142676372898e-06, "loss": 0.14648321866989136, "step": 21490 }, { "epoch": 0.5663427756722851, "grad_norm": 1.3237030506134033, "learning_rate": 8.345728247929319e-06, "loss": 0.13224444389343262, "step": 21500 }, { "epoch": 0.5666061909167839, "grad_norm": 2.0740325450897217, "learning_rate": 8.33731502353609e-06, "loss": 0.14314329624176025, "step": 21510 }, { "epoch": 0.5668696061612826, "grad_norm": 1.9246050119400024, "learning_rate": 8.328903009316724e-06, "loss": 0.16157556772232057, "step": 21520 }, { "epoch": 0.5671330214057813, "grad_norm": 1.607194423675537, "learning_rate": 8.320492211393844e-06, "loss": 0.14262175559997559, "step": 21530 }, { "epoch": 0.5673964366502801, "grad_norm": 3.2594666481018066, "learning_rate": 8.312082635889193e-06, "loss": 0.15042200088500976, "step": 21540 }, { "epoch": 0.5676598518947787, "grad_norm": 1.2088980674743652, "learning_rate": 8.303674288923621e-06, "loss": 0.1519657254219055, "step": 21550 }, { "epoch": 0.5679232671392775, "grad_norm": 2.5542402267456055, "learning_rate": 8.29526717661709e-06, "loss": 0.14420833587646484, "step": 21560 }, { "epoch": 0.5681866823837762, "grad_norm": 1.9219850301742554, "learning_rate": 8.286861305088655e-06, "loss": 0.13678343296051027, "step": 21570 }, { "epoch": 0.568450097628275, "grad_norm": 1.1924426555633545, "learning_rate": 8.278456680456476e-06, "loss": 0.13353686332702636, "step": 21580 }, { "epoch": 0.5687135128727737, "grad_norm": 1.4431250095367432, "learning_rate": 8.270053308837791e-06, "loss": 0.13076269626617432, "step": 21590 }, { "epoch": 0.5689769281172725, "grad_norm": 1.7427561283111572, "learning_rate": 8.261651196348953e-06, "loss": 0.13707287311553956, "step": 21600 }, { "epoch": 0.5692403433617712, "grad_norm": 2.141451597213745, "learning_rate": 8.253250349105369e-06, "loss": 0.1423823595046997, "step": 21610 }, { "epoch": 0.5695037586062699, "grad_norm": 1.469038486480713, "learning_rate": 8.24485077322154e-06, "loss": 0.14267370700836182, "step": 21620 }, { "epoch": 0.5697671738507687, "grad_norm": 1.7643768787384033, "learning_rate": 8.236452474811046e-06, "loss": 0.14344208240509032, "step": 21630 }, { "epoch": 0.5700305890952674, "grad_norm": 1.4078924655914307, "learning_rate": 8.228055459986521e-06, "loss": 0.14423422813415526, "step": 21640 }, { "epoch": 0.5702940043397662, "grad_norm": 2.708272933959961, "learning_rate": 8.219659734859686e-06, "loss": 0.1327930808067322, "step": 21650 }, { "epoch": 0.5705574195842649, "grad_norm": 2.643217086791992, "learning_rate": 8.211265305541304e-06, "loss": 0.14001076221466063, "step": 21660 }, { "epoch": 0.5708208348287637, "grad_norm": 2.9370954036712646, "learning_rate": 8.202872178141204e-06, "loss": 0.14579174518585206, "step": 21670 }, { "epoch": 0.5710842500732624, "grad_norm": 6.020047187805176, "learning_rate": 8.19448035876827e-06, "loss": 0.13250213861465454, "step": 21680 }, { "epoch": 0.5713476653177612, "grad_norm": 2.1349234580993652, "learning_rate": 8.186089853530426e-06, "loss": 0.14782435894012452, "step": 21690 }, { "epoch": 0.5716110805622598, "grad_norm": 4.885651111602783, "learning_rate": 8.177700668534646e-06, "loss": 0.1478786587715149, "step": 21700 }, { "epoch": 0.5718744958067585, "grad_norm": 1.2089589834213257, "learning_rate": 8.169312809886944e-06, "loss": 0.13889373540878297, "step": 21710 }, { "epoch": 0.5721379110512573, "grad_norm": 1.6750292778015137, "learning_rate": 8.160926283692355e-06, "loss": 0.14023040533065795, "step": 21720 }, { "epoch": 0.572401326295756, "grad_norm": 2.4051156044006348, "learning_rate": 8.152541096054965e-06, "loss": 0.1286496639251709, "step": 21730 }, { "epoch": 0.5726647415402548, "grad_norm": 1.7987178564071655, "learning_rate": 8.144157253077871e-06, "loss": 0.14381103515625, "step": 21740 }, { "epoch": 0.5729281567847535, "grad_norm": 1.4698773622512817, "learning_rate": 8.135774760863193e-06, "loss": 0.14788858890533446, "step": 21750 }, { "epoch": 0.5731915720292523, "grad_norm": 1.098967432975769, "learning_rate": 8.127393625512077e-06, "loss": 0.14012300968170166, "step": 21760 }, { "epoch": 0.573454987273751, "grad_norm": 1.2897335290908813, "learning_rate": 8.11901385312467e-06, "loss": 0.14474020004272461, "step": 21770 }, { "epoch": 0.5737184025182498, "grad_norm": 1.6256589889526367, "learning_rate": 8.110635449800132e-06, "loss": 0.14658674001693725, "step": 21780 }, { "epoch": 0.5739818177627485, "grad_norm": 1.504141926765442, "learning_rate": 8.10225842163663e-06, "loss": 0.14726014137268068, "step": 21790 }, { "epoch": 0.5742452330072472, "grad_norm": 1.2804535627365112, "learning_rate": 8.093882774731322e-06, "loss": 0.1538918733596802, "step": 21800 }, { "epoch": 0.574508648251746, "grad_norm": 2.7749266624450684, "learning_rate": 8.085508515180373e-06, "loss": 0.14414138793945314, "step": 21810 }, { "epoch": 0.5747720634962447, "grad_norm": 1.544080138206482, "learning_rate": 8.077135649078917e-06, "loss": 0.13512532711029052, "step": 21820 }, { "epoch": 0.5750354787407435, "grad_norm": 2.0579240322113037, "learning_rate": 8.068764182521102e-06, "loss": 0.15057728290557862, "step": 21830 }, { "epoch": 0.5752988939852421, "grad_norm": 1.663588047027588, "learning_rate": 8.060394121600033e-06, "loss": 0.13729771375656127, "step": 21840 }, { "epoch": 0.5755623092297409, "grad_norm": 0.8384369015693665, "learning_rate": 8.052025472407802e-06, "loss": 0.13718347549438475, "step": 21850 }, { "epoch": 0.5758257244742396, "grad_norm": 1.3091685771942139, "learning_rate": 8.043658241035477e-06, "loss": 0.14897470474243163, "step": 21860 }, { "epoch": 0.5760891397187383, "grad_norm": 0.9822778701782227, "learning_rate": 8.035292433573088e-06, "loss": 0.1302346706390381, "step": 21870 }, { "epoch": 0.5763525549632371, "grad_norm": 2.014143466949463, "learning_rate": 8.026928056109628e-06, "loss": 0.13120930194854735, "step": 21880 }, { "epoch": 0.5766159702077358, "grad_norm": 2.6840105056762695, "learning_rate": 8.018565114733056e-06, "loss": 0.1560115933418274, "step": 21890 }, { "epoch": 0.5768793854522346, "grad_norm": 1.0778522491455078, "learning_rate": 8.010203615530279e-06, "loss": 0.13489533662796022, "step": 21900 }, { "epoch": 0.5771428006967333, "grad_norm": 2.6054935455322266, "learning_rate": 8.00184356458716e-06, "loss": 0.143388032913208, "step": 21910 }, { "epoch": 0.5774062159412321, "grad_norm": 3.047520637512207, "learning_rate": 7.993484967988504e-06, "loss": 0.14362941980361937, "step": 21920 }, { "epoch": 0.5776696311857308, "grad_norm": 1.6592599153518677, "learning_rate": 7.98512783181805e-06, "loss": 0.13122003078460692, "step": 21930 }, { "epoch": 0.5779330464302296, "grad_norm": 1.2666429281234741, "learning_rate": 7.976772162158498e-06, "loss": 0.15366096496582032, "step": 21940 }, { "epoch": 0.5781964616747283, "grad_norm": 1.7177362442016602, "learning_rate": 7.968417965091449e-06, "loss": 0.13317110538482665, "step": 21950 }, { "epoch": 0.578459876919227, "grad_norm": 3.021562099456787, "learning_rate": 7.960065246697456e-06, "loss": 0.14072277545928955, "step": 21960 }, { "epoch": 0.5787232921637258, "grad_norm": 1.1649587154388428, "learning_rate": 7.951714013055988e-06, "loss": 0.1429387331008911, "step": 21970 }, { "epoch": 0.5789867074082244, "grad_norm": 2.08249831199646, "learning_rate": 7.94336427024543e-06, "loss": 0.13785550594329835, "step": 21980 }, { "epoch": 0.5792501226527232, "grad_norm": 1.9279241561889648, "learning_rate": 7.93501602434309e-06, "loss": 0.1382765769958496, "step": 21990 }, { "epoch": 0.5795135378972219, "grad_norm": 1.400913953781128, "learning_rate": 7.926669281425177e-06, "loss": 0.14505621194839477, "step": 22000 }, { "epoch": 0.5797769531417207, "grad_norm": 1.9903488159179688, "learning_rate": 7.91832404756681e-06, "loss": 0.1508869409561157, "step": 22010 }, { "epoch": 0.5800403683862194, "grad_norm": 3.5714104175567627, "learning_rate": 7.909980328842015e-06, "loss": 0.14244810342788697, "step": 22020 }, { "epoch": 0.5803037836307182, "grad_norm": 1.592347502708435, "learning_rate": 7.901638131323705e-06, "loss": 0.14054585695266725, "step": 22030 }, { "epoch": 0.5805671988752169, "grad_norm": 1.6753326654434204, "learning_rate": 7.893297461083697e-06, "loss": 0.1434148907661438, "step": 22040 }, { "epoch": 0.5808306141197156, "grad_norm": 2.1273603439331055, "learning_rate": 7.884958324192689e-06, "loss": 0.15336287021636963, "step": 22050 }, { "epoch": 0.5810940293642144, "grad_norm": 2.3004209995269775, "learning_rate": 7.876620726720257e-06, "loss": 0.16143786907196045, "step": 22060 }, { "epoch": 0.5813574446087131, "grad_norm": 1.1344289779663086, "learning_rate": 7.868284674734877e-06, "loss": 0.14408729076385499, "step": 22070 }, { "epoch": 0.5816208598532119, "grad_norm": 1.0064433813095093, "learning_rate": 7.859950174303879e-06, "loss": 0.14414505958557128, "step": 22080 }, { "epoch": 0.5818842750977106, "grad_norm": 1.0492030382156372, "learning_rate": 7.851617231493472e-06, "loss": 0.14123495817184448, "step": 22090 }, { "epoch": 0.5821476903422094, "grad_norm": 2.216630697250366, "learning_rate": 7.843285852368734e-06, "loss": 0.13997535705566405, "step": 22100 }, { "epoch": 0.5824111055867081, "grad_norm": 0.9829292297363281, "learning_rate": 7.8349560429936e-06, "loss": 0.13537092208862306, "step": 22110 }, { "epoch": 0.5826745208312069, "grad_norm": 2.198000192642212, "learning_rate": 7.826627809430867e-06, "loss": 0.14418892860412597, "step": 22120 }, { "epoch": 0.5829379360757055, "grad_norm": 1.6304913759231567, "learning_rate": 7.818301157742182e-06, "loss": 0.13649579286575317, "step": 22130 }, { "epoch": 0.5832013513202042, "grad_norm": 2.407407522201538, "learning_rate": 7.809976093988039e-06, "loss": 0.14323530197143555, "step": 22140 }, { "epoch": 0.583464766564703, "grad_norm": 2.495483160018921, "learning_rate": 7.801652624227782e-06, "loss": 0.15078948736190795, "step": 22150 }, { "epoch": 0.5837281818092017, "grad_norm": 2.070927143096924, "learning_rate": 7.79333075451959e-06, "loss": 0.13313164710998535, "step": 22160 }, { "epoch": 0.5839915970537005, "grad_norm": 1.6029773950576782, "learning_rate": 7.785010490920477e-06, "loss": 0.14160951375961303, "step": 22170 }, { "epoch": 0.5842550122981992, "grad_norm": 3.0713908672332764, "learning_rate": 7.776691839486297e-06, "loss": 0.13393135070800782, "step": 22180 }, { "epoch": 0.584518427542698, "grad_norm": 1.4439380168914795, "learning_rate": 7.76837480627171e-06, "loss": 0.1437251925468445, "step": 22190 }, { "epoch": 0.5847818427871967, "grad_norm": 2.4114913940429688, "learning_rate": 7.760059397330225e-06, "loss": 0.12840898036956788, "step": 22200 }, { "epoch": 0.5850452580316955, "grad_norm": 2.3954689502716064, "learning_rate": 7.75174561871415e-06, "loss": 0.1410682201385498, "step": 22210 }, { "epoch": 0.5853086732761942, "grad_norm": 1.3756992816925049, "learning_rate": 7.743433476474608e-06, "loss": 0.13621237277984619, "step": 22220 }, { "epoch": 0.5855720885206929, "grad_norm": 3.478839159011841, "learning_rate": 7.73512297666154e-06, "loss": 0.13746176958084105, "step": 22230 }, { "epoch": 0.5858355037651917, "grad_norm": 1.2560418844223022, "learning_rate": 7.726814125323683e-06, "loss": 0.13181164264678955, "step": 22240 }, { "epoch": 0.5860989190096904, "grad_norm": 3.8917076587677, "learning_rate": 7.718506928508583e-06, "loss": 0.13495171070098877, "step": 22250 }, { "epoch": 0.5863623342541892, "grad_norm": 1.9605416059494019, "learning_rate": 7.71020139226257e-06, "loss": 0.130181884765625, "step": 22260 }, { "epoch": 0.5866257494986878, "grad_norm": 1.5873034000396729, "learning_rate": 7.701897522630772e-06, "loss": 0.13403995037078859, "step": 22270 }, { "epoch": 0.5868891647431866, "grad_norm": 1.822157859802246, "learning_rate": 7.693595325657109e-06, "loss": 0.1479096531867981, "step": 22280 }, { "epoch": 0.5871525799876853, "grad_norm": 1.0901025533676147, "learning_rate": 7.685294807384273e-06, "loss": 0.14305154085159302, "step": 22290 }, { "epoch": 0.5874159952321841, "grad_norm": 2.452662944793701, "learning_rate": 7.676995973853743e-06, "loss": 0.13558242321014405, "step": 22300 }, { "epoch": 0.5876794104766828, "grad_norm": 1.93693208694458, "learning_rate": 7.668698831105768e-06, "loss": 0.12266781330108642, "step": 22310 }, { "epoch": 0.5879428257211815, "grad_norm": 0.9869850873947144, "learning_rate": 7.66040338517936e-06, "loss": 0.1472939968109131, "step": 22320 }, { "epoch": 0.5882062409656803, "grad_norm": 0.9716719388961792, "learning_rate": 7.652109642112308e-06, "loss": 0.1478618860244751, "step": 22330 }, { "epoch": 0.588469656210179, "grad_norm": 1.0063369274139404, "learning_rate": 7.643817607941157e-06, "loss": 0.12518105506896973, "step": 22340 }, { "epoch": 0.5887330714546778, "grad_norm": 1.6990529298782349, "learning_rate": 7.635527288701201e-06, "loss": 0.13088009357452393, "step": 22350 }, { "epoch": 0.5889964866991765, "grad_norm": 1.9307259321212769, "learning_rate": 7.6272386904265005e-06, "loss": 0.1391417860984802, "step": 22360 }, { "epoch": 0.5892599019436753, "grad_norm": 1.2809548377990723, "learning_rate": 7.618951819149848e-06, "loss": 0.13124468326568603, "step": 22370 }, { "epoch": 0.589523317188174, "grad_norm": 1.3683974742889404, "learning_rate": 7.610666680902791e-06, "loss": 0.11812108755111694, "step": 22380 }, { "epoch": 0.5897867324326727, "grad_norm": 1.483941912651062, "learning_rate": 7.602383281715608e-06, "loss": 0.1314542531967163, "step": 22390 }, { "epoch": 0.5900501476771715, "grad_norm": 4.474586486816406, "learning_rate": 7.594101627617312e-06, "loss": 0.13332138061523438, "step": 22400 }, { "epoch": 0.5903135629216701, "grad_norm": 3.5831406116485596, "learning_rate": 7.585821724635656e-06, "loss": 0.1379106044769287, "step": 22410 }, { "epoch": 0.5905769781661689, "grad_norm": 2.564894676208496, "learning_rate": 7.577543578797097e-06, "loss": 0.13316528797149657, "step": 22420 }, { "epoch": 0.5908403934106676, "grad_norm": 4.604330539703369, "learning_rate": 7.569267196126844e-06, "loss": 0.12885720729827882, "step": 22430 }, { "epoch": 0.5911038086551664, "grad_norm": 1.4172629117965698, "learning_rate": 7.560992582648793e-06, "loss": 0.1475963592529297, "step": 22440 }, { "epoch": 0.5913672238996651, "grad_norm": 0.8075081706047058, "learning_rate": 7.552719744385566e-06, "loss": 0.14158983230590821, "step": 22450 }, { "epoch": 0.5916306391441639, "grad_norm": 1.3962252140045166, "learning_rate": 7.544448687358495e-06, "loss": 0.14100382328033448, "step": 22460 }, { "epoch": 0.5918940543886626, "grad_norm": 0.9321791529655457, "learning_rate": 7.536179417587612e-06, "loss": 0.14232985973358153, "step": 22470 }, { "epoch": 0.5921574696331613, "grad_norm": 2.332939386367798, "learning_rate": 7.527911941091644e-06, "loss": 0.13378028869628905, "step": 22480 }, { "epoch": 0.5924208848776601, "grad_norm": 2.5185718536376953, "learning_rate": 7.519646263888023e-06, "loss": 0.14460217952728271, "step": 22490 }, { "epoch": 0.5926843001221588, "grad_norm": 1.079343318939209, "learning_rate": 7.511382391992861e-06, "loss": 0.13311028480529785, "step": 22500 }, { "epoch": 0.5929477153666576, "grad_norm": 1.9228135347366333, "learning_rate": 7.503120331420963e-06, "loss": 0.13911445140838624, "step": 22510 }, { "epoch": 0.5932111306111563, "grad_norm": 1.4101320505142212, "learning_rate": 7.494860088185815e-06, "loss": 0.13674529790878295, "step": 22520 }, { "epoch": 0.5934745458556551, "grad_norm": 1.39107084274292, "learning_rate": 7.48660166829957e-06, "loss": 0.14006348848342895, "step": 22530 }, { "epoch": 0.5937379611001538, "grad_norm": 2.2209556102752686, "learning_rate": 7.478345077773075e-06, "loss": 0.13478124141693115, "step": 22540 }, { "epoch": 0.5940013763446526, "grad_norm": 2.688272476196289, "learning_rate": 7.470090322615822e-06, "loss": 0.14678621292114258, "step": 22550 }, { "epoch": 0.5942647915891512, "grad_norm": 3.3901402950286865, "learning_rate": 7.4618374088359805e-06, "loss": 0.14383790493011475, "step": 22560 }, { "epoch": 0.5945282068336499, "grad_norm": 1.150425910949707, "learning_rate": 7.45358634244038e-06, "loss": 0.1303415060043335, "step": 22570 }, { "epoch": 0.5947916220781487, "grad_norm": 2.697671413421631, "learning_rate": 7.4453371294345e-06, "loss": 0.1355839490890503, "step": 22580 }, { "epoch": 0.5950550373226474, "grad_norm": 1.2967753410339355, "learning_rate": 7.437089775822477e-06, "loss": 0.1550910711288452, "step": 22590 }, { "epoch": 0.5953184525671462, "grad_norm": 3.84210205078125, "learning_rate": 7.4288442876070865e-06, "loss": 0.13968451023101808, "step": 22600 }, { "epoch": 0.5955818678116449, "grad_norm": 1.6008884906768799, "learning_rate": 7.420600670789753e-06, "loss": 0.146233069896698, "step": 22610 }, { "epoch": 0.5958452830561437, "grad_norm": 2.119495391845703, "learning_rate": 7.412358931370536e-06, "loss": 0.135200035572052, "step": 22620 }, { "epoch": 0.5961086983006424, "grad_norm": 1.346866250038147, "learning_rate": 7.404119075348129e-06, "loss": 0.1345914125442505, "step": 22630 }, { "epoch": 0.5963721135451412, "grad_norm": 1.3197778463363647, "learning_rate": 7.3958811087198555e-06, "loss": 0.141022253036499, "step": 22640 }, { "epoch": 0.5966355287896399, "grad_norm": 1.3655354976654053, "learning_rate": 7.3876450374816644e-06, "loss": 0.12459039688110352, "step": 22650 }, { "epoch": 0.5968989440341386, "grad_norm": 2.43582820892334, "learning_rate": 7.379410867628115e-06, "loss": 0.1380171537399292, "step": 22660 }, { "epoch": 0.5971623592786374, "grad_norm": 2.1982126235961914, "learning_rate": 7.371178605152405e-06, "loss": 0.13955957889556886, "step": 22670 }, { "epoch": 0.597425774523136, "grad_norm": 1.9086530208587646, "learning_rate": 7.362948256046321e-06, "loss": 0.12534399032592775, "step": 22680 }, { "epoch": 0.5976891897676349, "grad_norm": 1.8895699977874756, "learning_rate": 7.354719826300266e-06, "loss": 0.14359574317932128, "step": 22690 }, { "epoch": 0.5979526050121335, "grad_norm": 4.2843241691589355, "learning_rate": 7.346493321903251e-06, "loss": 0.13096948862075805, "step": 22700 }, { "epoch": 0.5982160202566323, "grad_norm": 2.1211800575256348, "learning_rate": 7.338268748842876e-06, "loss": 0.1324922800064087, "step": 22710 }, { "epoch": 0.598479435501131, "grad_norm": 4.1453046798706055, "learning_rate": 7.330046113105345e-06, "loss": 0.14965102672576905, "step": 22720 }, { "epoch": 0.5987428507456298, "grad_norm": 1.796135425567627, "learning_rate": 7.321825420675444e-06, "loss": 0.12539145946502686, "step": 22730 }, { "epoch": 0.5990062659901285, "grad_norm": 1.344254732131958, "learning_rate": 7.3136066775365455e-06, "loss": 0.1264049530029297, "step": 22740 }, { "epoch": 0.5992696812346272, "grad_norm": 1.995759129524231, "learning_rate": 7.305389889670611e-06, "loss": 0.12686008214950562, "step": 22750 }, { "epoch": 0.599533096479126, "grad_norm": 2.8383147716522217, "learning_rate": 7.2971750630581675e-06, "loss": 0.1434401750564575, "step": 22760 }, { "epoch": 0.5997965117236247, "grad_norm": 2.242617130279541, "learning_rate": 7.2889622036783245e-06, "loss": 0.13119027614593506, "step": 22770 }, { "epoch": 0.6000599269681235, "grad_norm": 1.7598474025726318, "learning_rate": 7.280751317508756e-06, "loss": 0.1340182065963745, "step": 22780 }, { "epoch": 0.6003233422126222, "grad_norm": 2.6252481937408447, "learning_rate": 7.272542410525691e-06, "loss": 0.1379840850830078, "step": 22790 }, { "epoch": 0.600586757457121, "grad_norm": 3.5062448978424072, "learning_rate": 7.2643354887039395e-06, "loss": 0.1501749873161316, "step": 22800 }, { "epoch": 0.6008501727016197, "grad_norm": 1.6030250787734985, "learning_rate": 7.256130558016843e-06, "loss": 0.13294801712036133, "step": 22810 }, { "epoch": 0.6011135879461185, "grad_norm": 1.0581181049346924, "learning_rate": 7.247927624436304e-06, "loss": 0.13467874526977539, "step": 22820 }, { "epoch": 0.6013770031906172, "grad_norm": 3.5453672409057617, "learning_rate": 7.239726693932779e-06, "loss": 0.13964409828186036, "step": 22830 }, { "epoch": 0.6016404184351158, "grad_norm": 1.106894612312317, "learning_rate": 7.2315277724752495e-06, "loss": 0.12614212036132813, "step": 22840 }, { "epoch": 0.6019038336796146, "grad_norm": 2.182436466217041, "learning_rate": 7.223330866031253e-06, "loss": 0.13934667110443116, "step": 22850 }, { "epoch": 0.6021672489241133, "grad_norm": 1.9273924827575684, "learning_rate": 7.215135980566847e-06, "loss": 0.12868934869766235, "step": 22860 }, { "epoch": 0.6024306641686121, "grad_norm": 1.833559274673462, "learning_rate": 7.2069431220466215e-06, "loss": 0.1405894637107849, "step": 22870 }, { "epoch": 0.6026940794131108, "grad_norm": 2.423560380935669, "learning_rate": 7.198752296433697e-06, "loss": 0.13778550624847413, "step": 22880 }, { "epoch": 0.6029574946576096, "grad_norm": 1.1707684993743896, "learning_rate": 7.190563509689705e-06, "loss": 0.13231147527694703, "step": 22890 }, { "epoch": 0.6032209099021083, "grad_norm": 2.0792741775512695, "learning_rate": 7.182376767774803e-06, "loss": 0.1355958938598633, "step": 22900 }, { "epoch": 0.603484325146607, "grad_norm": 1.7644784450531006, "learning_rate": 7.174192076647653e-06, "loss": 0.13537803888320923, "step": 22910 }, { "epoch": 0.6037477403911058, "grad_norm": 1.0290899276733398, "learning_rate": 7.166009442265423e-06, "loss": 0.1238730788230896, "step": 22920 }, { "epoch": 0.6040111556356045, "grad_norm": 1.7026091814041138, "learning_rate": 7.15782887058379e-06, "loss": 0.12539321184158325, "step": 22930 }, { "epoch": 0.6042745708801033, "grad_norm": 2.530768394470215, "learning_rate": 7.149650367556926e-06, "loss": 0.13197875022888184, "step": 22940 }, { "epoch": 0.604537986124602, "grad_norm": 1.3148891925811768, "learning_rate": 7.141473939137498e-06, "loss": 0.13350509405136107, "step": 22950 }, { "epoch": 0.6048014013691008, "grad_norm": 2.1494576930999756, "learning_rate": 7.133299591276665e-06, "loss": 0.14067089557647705, "step": 22960 }, { "epoch": 0.6050648166135995, "grad_norm": 3.996091365814209, "learning_rate": 7.12512732992407e-06, "loss": 0.131087327003479, "step": 22970 }, { "epoch": 0.6053282318580983, "grad_norm": 3.0581116676330566, "learning_rate": 7.1169571610278375e-06, "loss": 0.14994275569915771, "step": 22980 }, { "epoch": 0.6055916471025969, "grad_norm": 1.6706568002700806, "learning_rate": 7.108789090534569e-06, "loss": 0.1260170340538025, "step": 22990 }, { "epoch": 0.6058550623470956, "grad_norm": 1.3875712156295776, "learning_rate": 7.100623124389334e-06, "loss": 0.13039451837539673, "step": 23000 }, { "epoch": 0.6061184775915944, "grad_norm": 1.4271996021270752, "learning_rate": 7.092459268535685e-06, "loss": 0.13684940338134766, "step": 23010 }, { "epoch": 0.6063818928360931, "grad_norm": 2.717250347137451, "learning_rate": 7.084297528915616e-06, "loss": 0.14379068613052368, "step": 23020 }, { "epoch": 0.6066453080805919, "grad_norm": 1.5731843709945679, "learning_rate": 7.076137911469603e-06, "loss": 0.13386814594268798, "step": 23030 }, { "epoch": 0.6069087233250906, "grad_norm": 2.3466849327087402, "learning_rate": 7.067980422136565e-06, "loss": 0.1356213092803955, "step": 23040 }, { "epoch": 0.6071721385695894, "grad_norm": 4.378043174743652, "learning_rate": 7.059825066853872e-06, "loss": 0.1458061456680298, "step": 23050 }, { "epoch": 0.6074355538140881, "grad_norm": 1.684226632118225, "learning_rate": 7.0516718515573445e-06, "loss": 0.13643072843551635, "step": 23060 }, { "epoch": 0.6076989690585869, "grad_norm": 1.3434169292449951, "learning_rate": 7.043520782181248e-06, "loss": 0.12718154191970826, "step": 23070 }, { "epoch": 0.6079623843030856, "grad_norm": 1.6881122589111328, "learning_rate": 7.035371864658276e-06, "loss": 0.14375481605529786, "step": 23080 }, { "epoch": 0.6082257995475843, "grad_norm": 2.593827724456787, "learning_rate": 7.0272251049195684e-06, "loss": 0.12631626129150392, "step": 23090 }, { "epoch": 0.6084892147920831, "grad_norm": 1.8978620767593384, "learning_rate": 7.019080508894683e-06, "loss": 0.1386675477027893, "step": 23100 }, { "epoch": 0.6087526300365818, "grad_norm": 1.2484378814697266, "learning_rate": 7.010938082511614e-06, "loss": 0.13375499248504638, "step": 23110 }, { "epoch": 0.6090160452810806, "grad_norm": 1.1914927959442139, "learning_rate": 7.0027978316967705e-06, "loss": 0.1269080638885498, "step": 23120 }, { "epoch": 0.6092794605255792, "grad_norm": 1.6321614980697632, "learning_rate": 6.994659762374967e-06, "loss": 0.13450502157211303, "step": 23130 }, { "epoch": 0.609542875770078, "grad_norm": 2.339128017425537, "learning_rate": 6.986523880469458e-06, "loss": 0.1349124550819397, "step": 23140 }, { "epoch": 0.6098062910145767, "grad_norm": 1.8911360502243042, "learning_rate": 6.978390191901878e-06, "loss": 0.1529719352722168, "step": 23150 }, { "epoch": 0.6100697062590755, "grad_norm": 3.0500056743621826, "learning_rate": 6.970258702592277e-06, "loss": 0.14494163990020753, "step": 23160 }, { "epoch": 0.6103331215035742, "grad_norm": 1.025847315788269, "learning_rate": 6.962129418459109e-06, "loss": 0.1269016981124878, "step": 23170 }, { "epoch": 0.6105965367480729, "grad_norm": 1.7302321195602417, "learning_rate": 6.954002345419214e-06, "loss": 0.13632490634918212, "step": 23180 }, { "epoch": 0.6108599519925717, "grad_norm": 1.5275684595108032, "learning_rate": 6.945877489387829e-06, "loss": 0.14547549486160277, "step": 23190 }, { "epoch": 0.6111233672370704, "grad_norm": 1.013899564743042, "learning_rate": 6.9377548562785735e-06, "loss": 0.13988676071166992, "step": 23200 }, { "epoch": 0.6113867824815692, "grad_norm": 2.2280185222625732, "learning_rate": 6.929634452003447e-06, "loss": 0.13484642505645753, "step": 23210 }, { "epoch": 0.6116501977260679, "grad_norm": 2.2380902767181396, "learning_rate": 6.921516282472837e-06, "loss": 0.14889390468597413, "step": 23220 }, { "epoch": 0.6119136129705667, "grad_norm": 2.1727912425994873, "learning_rate": 6.91340035359549e-06, "loss": 0.1448598623275757, "step": 23230 }, { "epoch": 0.6121770282150654, "grad_norm": 1.9304324388504028, "learning_rate": 6.905286671278536e-06, "loss": 0.14151265621185302, "step": 23240 }, { "epoch": 0.6124404434595642, "grad_norm": 3.016475200653076, "learning_rate": 6.897175241427462e-06, "loss": 0.12956702709197998, "step": 23250 }, { "epoch": 0.6127038587040629, "grad_norm": 1.3723551034927368, "learning_rate": 6.889066069946107e-06, "loss": 0.14073760509490968, "step": 23260 }, { "epoch": 0.6129672739485615, "grad_norm": 1.1435035467147827, "learning_rate": 6.880959162736691e-06, "loss": 0.13621170520782472, "step": 23270 }, { "epoch": 0.6132306891930603, "grad_norm": 1.9321669340133667, "learning_rate": 6.872854525699761e-06, "loss": 0.14087339639663696, "step": 23280 }, { "epoch": 0.613494104437559, "grad_norm": 1.972037434577942, "learning_rate": 6.8647521647342205e-06, "loss": 0.132607102394104, "step": 23290 }, { "epoch": 0.6137575196820578, "grad_norm": 1.6136209964752197, "learning_rate": 6.856652085737323e-06, "loss": 0.143936824798584, "step": 23300 }, { "epoch": 0.6140209349265565, "grad_norm": 1.601549506187439, "learning_rate": 6.848554294604652e-06, "loss": 0.1272900342941284, "step": 23310 }, { "epoch": 0.6142843501710553, "grad_norm": 1.0508836507797241, "learning_rate": 6.8404587972301316e-06, "loss": 0.12803523540496825, "step": 23320 }, { "epoch": 0.614547765415554, "grad_norm": 2.2401020526885986, "learning_rate": 6.8323655995060135e-06, "loss": 0.14963711500167848, "step": 23330 }, { "epoch": 0.6148111806600528, "grad_norm": 2.0401575565338135, "learning_rate": 6.824274707322874e-06, "loss": 0.13992650508880616, "step": 23340 }, { "epoch": 0.6150745959045515, "grad_norm": 2.425795078277588, "learning_rate": 6.816186126569615e-06, "loss": 0.14299449920654297, "step": 23350 }, { "epoch": 0.6153380111490502, "grad_norm": 4.139267921447754, "learning_rate": 6.808099863133453e-06, "loss": 0.1464477300643921, "step": 23360 }, { "epoch": 0.615601426393549, "grad_norm": 1.4089066982269287, "learning_rate": 6.800015922899923e-06, "loss": 0.13766062259674072, "step": 23370 }, { "epoch": 0.6158648416380477, "grad_norm": 1.5197685956954956, "learning_rate": 6.7919343117528655e-06, "loss": 0.13569262027740478, "step": 23380 }, { "epoch": 0.6161282568825465, "grad_norm": 3.214470624923706, "learning_rate": 6.783855035574418e-06, "loss": 0.14016003608703614, "step": 23390 }, { "epoch": 0.6163916721270452, "grad_norm": 1.7456014156341553, "learning_rate": 6.775778100245038e-06, "loss": 0.14325554370880128, "step": 23400 }, { "epoch": 0.616655087371544, "grad_norm": 3.173039197921753, "learning_rate": 6.76770351164346e-06, "loss": 0.13178255558013915, "step": 23410 }, { "epoch": 0.6169185026160426, "grad_norm": 3.975569725036621, "learning_rate": 6.759631275646715e-06, "loss": 0.13476746082305907, "step": 23420 }, { "epoch": 0.6171819178605413, "grad_norm": 2.5677037239074707, "learning_rate": 6.751561398130133e-06, "loss": 0.13939285278320312, "step": 23430 }, { "epoch": 0.6174453331050401, "grad_norm": 2.5268373489379883, "learning_rate": 6.743493884967312e-06, "loss": 0.13822048902511597, "step": 23440 }, { "epoch": 0.6177087483495388, "grad_norm": 2.091121196746826, "learning_rate": 6.735428742030141e-06, "loss": 0.12088127136230468, "step": 23450 }, { "epoch": 0.6179721635940376, "grad_norm": 1.3831844329833984, "learning_rate": 6.7273659751887775e-06, "loss": 0.13680993318557738, "step": 23460 }, { "epoch": 0.6182355788385363, "grad_norm": 2.40653920173645, "learning_rate": 6.719305590311649e-06, "loss": 0.13713886737823486, "step": 23470 }, { "epoch": 0.6184989940830351, "grad_norm": 3.0446314811706543, "learning_rate": 6.7112475932654554e-06, "loss": 0.1376265048980713, "step": 23480 }, { "epoch": 0.6187624093275338, "grad_norm": 2.185944080352783, "learning_rate": 6.70319198991515e-06, "loss": 0.13580527305603027, "step": 23490 }, { "epoch": 0.6190258245720326, "grad_norm": 1.537619948387146, "learning_rate": 6.695138786123952e-06, "loss": 0.12598180770874023, "step": 23500 }, { "epoch": 0.6192892398165313, "grad_norm": 1.421324610710144, "learning_rate": 6.687087987753331e-06, "loss": 0.12927324771881105, "step": 23510 }, { "epoch": 0.61955265506103, "grad_norm": 2.562445640563965, "learning_rate": 6.679039600663001e-06, "loss": 0.14503915309906007, "step": 23520 }, { "epoch": 0.6198160703055288, "grad_norm": 1.3798733949661255, "learning_rate": 6.670993630710929e-06, "loss": 0.12877674102783204, "step": 23530 }, { "epoch": 0.6200794855500275, "grad_norm": 2.495429515838623, "learning_rate": 6.662950083753319e-06, "loss": 0.12336808443069458, "step": 23540 }, { "epoch": 0.6203429007945263, "grad_norm": 1.8605527877807617, "learning_rate": 6.654908965644609e-06, "loss": 0.1374637484550476, "step": 23550 }, { "epoch": 0.6206063160390249, "grad_norm": 2.8418476581573486, "learning_rate": 6.646870282237474e-06, "loss": 0.1289588689804077, "step": 23560 }, { "epoch": 0.6208697312835237, "grad_norm": 1.5646756887435913, "learning_rate": 6.6388340393828125e-06, "loss": 0.1410160779953003, "step": 23570 }, { "epoch": 0.6211331465280224, "grad_norm": 1.8152170181274414, "learning_rate": 6.6308002429297516e-06, "loss": 0.12911182641983032, "step": 23580 }, { "epoch": 0.6213965617725212, "grad_norm": 1.6263149976730347, "learning_rate": 6.622768898725634e-06, "loss": 0.12072880268096924, "step": 23590 }, { "epoch": 0.6216599770170199, "grad_norm": 2.6485657691955566, "learning_rate": 6.614740012616017e-06, "loss": 0.13704383373260498, "step": 23600 }, { "epoch": 0.6219233922615186, "grad_norm": 4.513274192810059, "learning_rate": 6.606713590444673e-06, "loss": 0.13212889432907104, "step": 23610 }, { "epoch": 0.6221868075060174, "grad_norm": 3.1553001403808594, "learning_rate": 6.598689638053577e-06, "loss": 0.13583652973175048, "step": 23620 }, { "epoch": 0.6224502227505161, "grad_norm": 1.9191932678222656, "learning_rate": 6.590668161282903e-06, "loss": 0.13355889320373535, "step": 23630 }, { "epoch": 0.6227136379950149, "grad_norm": 1.824464201927185, "learning_rate": 6.582649165971037e-06, "loss": 0.1317456841468811, "step": 23640 }, { "epoch": 0.6229770532395136, "grad_norm": 1.193006992340088, "learning_rate": 6.574632657954542e-06, "loss": 0.1219285249710083, "step": 23650 }, { "epoch": 0.6232404684840124, "grad_norm": 1.6263452768325806, "learning_rate": 6.5666186430681835e-06, "loss": 0.13171777725219727, "step": 23660 }, { "epoch": 0.6235038837285111, "grad_norm": 2.319789409637451, "learning_rate": 6.558607127144905e-06, "loss": 0.1298205852508545, "step": 23670 }, { "epoch": 0.6237672989730099, "grad_norm": 2.2049386501312256, "learning_rate": 6.550598116015832e-06, "loss": 0.1323573350906372, "step": 23680 }, { "epoch": 0.6240307142175086, "grad_norm": 5.243276596069336, "learning_rate": 6.54259161551027e-06, "loss": 0.14407172203063964, "step": 23690 }, { "epoch": 0.6242941294620072, "grad_norm": 2.104513645172119, "learning_rate": 6.534587631455696e-06, "loss": 0.1326206922531128, "step": 23700 }, { "epoch": 0.624557544706506, "grad_norm": 1.007358431816101, "learning_rate": 6.5265861696777556e-06, "loss": 0.14154564142227172, "step": 23710 }, { "epoch": 0.6248209599510047, "grad_norm": 3.1516425609588623, "learning_rate": 6.518587236000259e-06, "loss": 0.13329309225082397, "step": 23720 }, { "epoch": 0.6250843751955035, "grad_norm": 2.2348403930664062, "learning_rate": 6.510590836245166e-06, "loss": 0.12158057689666749, "step": 23730 }, { "epoch": 0.6253477904400022, "grad_norm": 1.7913466691970825, "learning_rate": 6.502596976232616e-06, "loss": 0.1299106001853943, "step": 23740 }, { "epoch": 0.625611205684501, "grad_norm": 2.91825008392334, "learning_rate": 6.494605661780874e-06, "loss": 0.13902764320373534, "step": 23750 }, { "epoch": 0.6258746209289997, "grad_norm": 1.5855042934417725, "learning_rate": 6.486616898706365e-06, "loss": 0.1321015477180481, "step": 23760 }, { "epoch": 0.6261380361734985, "grad_norm": 1.1117304563522339, "learning_rate": 6.47863069282366e-06, "loss": 0.12788832187652588, "step": 23770 }, { "epoch": 0.6264014514179972, "grad_norm": 1.8410205841064453, "learning_rate": 6.47064704994546e-06, "loss": 0.1301414966583252, "step": 23780 }, { "epoch": 0.6266648666624959, "grad_norm": 1.491723656654358, "learning_rate": 6.462665975882607e-06, "loss": 0.13388724327087403, "step": 23790 }, { "epoch": 0.6269282819069947, "grad_norm": 2.5580556392669678, "learning_rate": 6.454687476444071e-06, "loss": 0.13222528696060182, "step": 23800 }, { "epoch": 0.6271916971514934, "grad_norm": 1.972542405128479, "learning_rate": 6.446711557436944e-06, "loss": 0.1212841510772705, "step": 23810 }, { "epoch": 0.6274551123959922, "grad_norm": 1.9002398252487183, "learning_rate": 6.438738224666452e-06, "loss": 0.13365960121154785, "step": 23820 }, { "epoch": 0.6277185276404909, "grad_norm": 1.3735443353652954, "learning_rate": 6.4307674839359225e-06, "loss": 0.13927438259124755, "step": 23830 }, { "epoch": 0.6279819428849897, "grad_norm": 2.068290948867798, "learning_rate": 6.42279934104681e-06, "loss": 0.12858080863952637, "step": 23840 }, { "epoch": 0.6282453581294883, "grad_norm": 2.785038709640503, "learning_rate": 6.4148338017986725e-06, "loss": 0.12675809860229492, "step": 23850 }, { "epoch": 0.6285087733739871, "grad_norm": 1.8948830366134644, "learning_rate": 6.406870871989166e-06, "loss": 0.12323253154754639, "step": 23860 }, { "epoch": 0.6287721886184858, "grad_norm": 1.699054479598999, "learning_rate": 6.398910557414064e-06, "loss": 0.13357701301574706, "step": 23870 }, { "epoch": 0.6290356038629845, "grad_norm": 1.8275678157806396, "learning_rate": 6.390952863867224e-06, "loss": 0.11929709911346435, "step": 23880 }, { "epoch": 0.6292990191074833, "grad_norm": 3.232290506362915, "learning_rate": 6.382997797140591e-06, "loss": 0.13424063920974733, "step": 23890 }, { "epoch": 0.629562434351982, "grad_norm": 1.226946234703064, "learning_rate": 6.375045363024217e-06, "loss": 0.1277064323425293, "step": 23900 }, { "epoch": 0.6298258495964808, "grad_norm": 2.4847965240478516, "learning_rate": 6.367095567306219e-06, "loss": 0.14572107791900635, "step": 23910 }, { "epoch": 0.6300892648409795, "grad_norm": 1.255802035331726, "learning_rate": 6.359148415772805e-06, "loss": 0.11517641544342042, "step": 23920 }, { "epoch": 0.6303526800854783, "grad_norm": 3.1774768829345703, "learning_rate": 6.351203914208254e-06, "loss": 0.1395902991294861, "step": 23930 }, { "epoch": 0.630616095329977, "grad_norm": 1.2932969331741333, "learning_rate": 6.343262068394914e-06, "loss": 0.13530632257461547, "step": 23940 }, { "epoch": 0.6308795105744757, "grad_norm": 2.0751473903656006, "learning_rate": 6.335322884113208e-06, "loss": 0.1419090747833252, "step": 23950 }, { "epoch": 0.6311429258189745, "grad_norm": 2.571265459060669, "learning_rate": 6.3273863671416146e-06, "loss": 0.13996527194976807, "step": 23960 }, { "epoch": 0.6314063410634732, "grad_norm": 3.522761583328247, "learning_rate": 6.319452523256673e-06, "loss": 0.12922863960266112, "step": 23970 }, { "epoch": 0.631669756307972, "grad_norm": 1.834774374961853, "learning_rate": 6.31152135823298e-06, "loss": 0.1334446668624878, "step": 23980 }, { "epoch": 0.6319331715524706, "grad_norm": 1.3301129341125488, "learning_rate": 6.303592877843171e-06, "loss": 0.1289549469947815, "step": 23990 }, { "epoch": 0.6321965867969694, "grad_norm": 1.8420592546463013, "learning_rate": 6.295667087857948e-06, "loss": 0.12978371381759643, "step": 24000 }, { "epoch": 0.6324600020414681, "grad_norm": 3.491814374923706, "learning_rate": 6.287743994046035e-06, "loss": 0.12733969688415528, "step": 24010 }, { "epoch": 0.6327234172859669, "grad_norm": 5.4128289222717285, "learning_rate": 6.279823602174202e-06, "loss": 0.1342423677444458, "step": 24020 }, { "epoch": 0.6329868325304656, "grad_norm": 2.9964544773101807, "learning_rate": 6.271905918007255e-06, "loss": 0.13787245750427246, "step": 24030 }, { "epoch": 0.6332502477749643, "grad_norm": 1.7378638982772827, "learning_rate": 6.263990947308022e-06, "loss": 0.13630412817001342, "step": 24040 }, { "epoch": 0.6335136630194631, "grad_norm": 1.2725496292114258, "learning_rate": 6.2560786958373645e-06, "loss": 0.13997739553451538, "step": 24050 }, { "epoch": 0.6337770782639618, "grad_norm": 1.711808443069458, "learning_rate": 6.248169169354159e-06, "loss": 0.13671025037765502, "step": 24060 }, { "epoch": 0.6340404935084606, "grad_norm": 2.073143720626831, "learning_rate": 6.240262373615298e-06, "loss": 0.13107942342758178, "step": 24070 }, { "epoch": 0.6343039087529593, "grad_norm": 1.7659382820129395, "learning_rate": 6.232358314375693e-06, "loss": 0.128756844997406, "step": 24080 }, { "epoch": 0.6345673239974581, "grad_norm": 2.475923538208008, "learning_rate": 6.224456997388254e-06, "loss": 0.12405611276626587, "step": 24090 }, { "epoch": 0.6348307392419568, "grad_norm": 1.8043580055236816, "learning_rate": 6.2165584284039046e-06, "loss": 0.12875719070434571, "step": 24100 }, { "epoch": 0.6350941544864556, "grad_norm": 3.2536964416503906, "learning_rate": 6.208662613171566e-06, "loss": 0.1290964126586914, "step": 24110 }, { "epoch": 0.6353575697309543, "grad_norm": 3.3394291400909424, "learning_rate": 6.200769557438144e-06, "loss": 0.13555305004119872, "step": 24120 }, { "epoch": 0.6356209849754529, "grad_norm": 1.0087779760360718, "learning_rate": 6.192879266948554e-06, "loss": 0.13207581043243408, "step": 24130 }, { "epoch": 0.6358844002199517, "grad_norm": 1.3436305522918701, "learning_rate": 6.184991747445686e-06, "loss": 0.13731856346130372, "step": 24140 }, { "epoch": 0.6361478154644504, "grad_norm": 2.2971794605255127, "learning_rate": 6.1771070046704165e-06, "loss": 0.12696986198425292, "step": 24150 }, { "epoch": 0.6364112307089492, "grad_norm": 1.6123943328857422, "learning_rate": 6.169225044361604e-06, "loss": 0.14081110954284667, "step": 24160 }, { "epoch": 0.6366746459534479, "grad_norm": 3.812631607055664, "learning_rate": 6.161345872256075e-06, "loss": 0.1343659281730652, "step": 24170 }, { "epoch": 0.6369380611979467, "grad_norm": 1.4216790199279785, "learning_rate": 6.153469494088638e-06, "loss": 0.13146893978118895, "step": 24180 }, { "epoch": 0.6372014764424454, "grad_norm": 1.8986432552337646, "learning_rate": 6.1455959155920555e-06, "loss": 0.12904940843582152, "step": 24190 }, { "epoch": 0.6374648916869442, "grad_norm": 2.7000505924224854, "learning_rate": 6.1377251424970555e-06, "loss": 0.12352800369262695, "step": 24200 }, { "epoch": 0.6377283069314429, "grad_norm": 2.900724411010742, "learning_rate": 6.1298571805323344e-06, "loss": 0.1337544322013855, "step": 24210 }, { "epoch": 0.6379917221759416, "grad_norm": 2.0684237480163574, "learning_rate": 6.121992035424529e-06, "loss": 0.12194957733154296, "step": 24220 }, { "epoch": 0.6382551374204404, "grad_norm": 1.7146700620651245, "learning_rate": 6.114129712898227e-06, "loss": 0.13655010461807252, "step": 24230 }, { "epoch": 0.6385185526649391, "grad_norm": 2.422619342803955, "learning_rate": 6.106270218675975e-06, "loss": 0.124226975440979, "step": 24240 }, { "epoch": 0.6387819679094379, "grad_norm": 2.0748839378356934, "learning_rate": 6.098413558478244e-06, "loss": 0.13580055236816407, "step": 24250 }, { "epoch": 0.6390453831539366, "grad_norm": 1.2343336343765259, "learning_rate": 6.090559738023456e-06, "loss": 0.12499800920486451, "step": 24260 }, { "epoch": 0.6393087983984354, "grad_norm": 2.2212328910827637, "learning_rate": 6.0827087630279605e-06, "loss": 0.13646378517150878, "step": 24270 }, { "epoch": 0.639572213642934, "grad_norm": 2.2380635738372803, "learning_rate": 6.074860639206028e-06, "loss": 0.1296303987503052, "step": 24280 }, { "epoch": 0.6398356288874328, "grad_norm": 1.1984124183654785, "learning_rate": 6.067015372269872e-06, "loss": 0.1264725685119629, "step": 24290 }, { "epoch": 0.6400990441319315, "grad_norm": 1.9888418912887573, "learning_rate": 6.059172967929609e-06, "loss": 0.12859340906143188, "step": 24300 }, { "epoch": 0.6403624593764302, "grad_norm": 1.2758114337921143, "learning_rate": 6.0513334318932825e-06, "loss": 0.125271475315094, "step": 24310 }, { "epoch": 0.640625874620929, "grad_norm": 1.9805370569229126, "learning_rate": 6.043496769866848e-06, "loss": 0.13922395706176757, "step": 24320 }, { "epoch": 0.6408892898654277, "grad_norm": 1.4451749324798584, "learning_rate": 6.035662987554154e-06, "loss": 0.1278216004371643, "step": 24330 }, { "epoch": 0.6411527051099265, "grad_norm": 1.966525912284851, "learning_rate": 6.027832090656982e-06, "loss": 0.13776025772094727, "step": 24340 }, { "epoch": 0.6414161203544252, "grad_norm": 1.7976866960525513, "learning_rate": 6.020004084874984e-06, "loss": 0.13217394351959227, "step": 24350 }, { "epoch": 0.641679535598924, "grad_norm": 1.446486234664917, "learning_rate": 6.012178975905725e-06, "loss": 0.12788469791412355, "step": 24360 }, { "epoch": 0.6419429508434227, "grad_norm": 3.1582415103912354, "learning_rate": 6.0043567694446566e-06, "loss": 0.11674304008483886, "step": 24370 }, { "epoch": 0.6422063660879215, "grad_norm": 2.3948869705200195, "learning_rate": 5.9965374711851175e-06, "loss": 0.1256544589996338, "step": 24380 }, { "epoch": 0.6424697813324202, "grad_norm": 1.3678661584854126, "learning_rate": 5.9887210868183325e-06, "loss": 0.1367666244506836, "step": 24390 }, { "epoch": 0.6427331965769189, "grad_norm": 1.280441164970398, "learning_rate": 5.9809076220334024e-06, "loss": 0.12849130630493164, "step": 24400 }, { "epoch": 0.6429966118214177, "grad_norm": 2.038336753845215, "learning_rate": 5.973097082517301e-06, "loss": 0.1297145128250122, "step": 24410 }, { "epoch": 0.6432600270659163, "grad_norm": 4.147812366485596, "learning_rate": 5.965289473954879e-06, "loss": 0.12024813890457153, "step": 24420 }, { "epoch": 0.6435234423104151, "grad_norm": 1.9974199533462524, "learning_rate": 5.95748480202885e-06, "loss": 0.13996121883392335, "step": 24430 }, { "epoch": 0.6437868575549138, "grad_norm": 2.1103768348693848, "learning_rate": 5.949683072419792e-06, "loss": 0.13103976249694824, "step": 24440 }, { "epoch": 0.6440502727994126, "grad_norm": 2.5153024196624756, "learning_rate": 5.941884290806142e-06, "loss": 0.1270591974258423, "step": 24450 }, { "epoch": 0.6443136880439113, "grad_norm": 1.6376796960830688, "learning_rate": 5.934088462864181e-06, "loss": 0.1341926336288452, "step": 24460 }, { "epoch": 0.6445771032884101, "grad_norm": 2.919337272644043, "learning_rate": 5.926295594268059e-06, "loss": 0.13512077331542968, "step": 24470 }, { "epoch": 0.6448405185329088, "grad_norm": 2.7911226749420166, "learning_rate": 5.918505690689759e-06, "loss": 0.12628669738769532, "step": 24480 }, { "epoch": 0.6451039337774075, "grad_norm": 2.883136510848999, "learning_rate": 5.910718757799105e-06, "loss": 0.14044207334518433, "step": 24490 }, { "epoch": 0.6453673490219063, "grad_norm": 2.1522581577301025, "learning_rate": 5.902934801263766e-06, "loss": 0.143804669380188, "step": 24500 }, { "epoch": 0.645630764266405, "grad_norm": 1.180483341217041, "learning_rate": 5.89515382674924e-06, "loss": 0.12707782983779908, "step": 24510 }, { "epoch": 0.6458941795109038, "grad_norm": 2.40232253074646, "learning_rate": 5.88737583991886e-06, "loss": 0.15196152925491332, "step": 24520 }, { "epoch": 0.6461575947554025, "grad_norm": 2.402045965194702, "learning_rate": 5.879600846433777e-06, "loss": 0.12838971614837646, "step": 24530 }, { "epoch": 0.6464210099999013, "grad_norm": 2.753762722015381, "learning_rate": 5.871828851952965e-06, "loss": 0.13566702604293823, "step": 24540 }, { "epoch": 0.6466844252444, "grad_norm": 2.323172092437744, "learning_rate": 5.864059862133221e-06, "loss": 0.1330738067626953, "step": 24550 }, { "epoch": 0.6469478404888986, "grad_norm": 1.0267027616500854, "learning_rate": 5.856293882629147e-06, "loss": 0.11966187953948974, "step": 24560 }, { "epoch": 0.6472112557333974, "grad_norm": 2.175837516784668, "learning_rate": 5.848530919093162e-06, "loss": 0.12840889692306517, "step": 24570 }, { "epoch": 0.6474746709778961, "grad_norm": 1.9355578422546387, "learning_rate": 5.840770977175485e-06, "loss": 0.12507891654968262, "step": 24580 }, { "epoch": 0.6477380862223949, "grad_norm": 2.889035940170288, "learning_rate": 5.833014062524129e-06, "loss": 0.12743431329727173, "step": 24590 }, { "epoch": 0.6480015014668936, "grad_norm": 2.0259299278259277, "learning_rate": 5.825260180784923e-06, "loss": 0.1245743989944458, "step": 24600 }, { "epoch": 0.6482649167113924, "grad_norm": 2.0134685039520264, "learning_rate": 5.817509337601475e-06, "loss": 0.1294093370437622, "step": 24610 }, { "epoch": 0.6485283319558911, "grad_norm": 2.788296699523926, "learning_rate": 5.8097615386151686e-06, "loss": 0.13083953857421876, "step": 24620 }, { "epoch": 0.6487917472003899, "grad_norm": 2.6942739486694336, "learning_rate": 5.8020167894652035e-06, "loss": 0.130181622505188, "step": 24630 }, { "epoch": 0.6490551624448886, "grad_norm": 2.23017954826355, "learning_rate": 5.79427509578853e-06, "loss": 0.13134605884552003, "step": 24640 }, { "epoch": 0.6493185776893873, "grad_norm": 2.1562767028808594, "learning_rate": 5.786536463219899e-06, "loss": 0.13239922523498535, "step": 24650 }, { "epoch": 0.6495819929338861, "grad_norm": 1.7293927669525146, "learning_rate": 5.778800897391812e-06, "loss": 0.11975600719451904, "step": 24660 }, { "epoch": 0.6498454081783848, "grad_norm": 1.1979467868804932, "learning_rate": 5.771068403934545e-06, "loss": 0.14218097925186157, "step": 24670 }, { "epoch": 0.6501088234228836, "grad_norm": 2.494457244873047, "learning_rate": 5.76333898847615e-06, "loss": 0.13004250526428224, "step": 24680 }, { "epoch": 0.6503722386673823, "grad_norm": 1.611355185508728, "learning_rate": 5.755612656642424e-06, "loss": 0.1254101037979126, "step": 24690 }, { "epoch": 0.650635653911881, "grad_norm": 2.758192300796509, "learning_rate": 5.747889414056925e-06, "loss": 0.12240171432495117, "step": 24700 }, { "epoch": 0.6508990691563797, "grad_norm": 2.361844062805176, "learning_rate": 5.740169266340963e-06, "loss": 0.1262230634689331, "step": 24710 }, { "epoch": 0.6511624844008785, "grad_norm": 1.3915367126464844, "learning_rate": 5.732452219113589e-06, "loss": 0.13309482336044312, "step": 24720 }, { "epoch": 0.6514258996453772, "grad_norm": 3.076021194458008, "learning_rate": 5.724738277991611e-06, "loss": 0.1251436710357666, "step": 24730 }, { "epoch": 0.6516893148898759, "grad_norm": 1.989406943321228, "learning_rate": 5.717027448589568e-06, "loss": 0.13985366821289064, "step": 24740 }, { "epoch": 0.6519527301343747, "grad_norm": 1.160333514213562, "learning_rate": 5.709319736519725e-06, "loss": 0.12144136428833008, "step": 24750 }, { "epoch": 0.6522161453788734, "grad_norm": 1.8474445343017578, "learning_rate": 5.701615147392096e-06, "loss": 0.1451274037361145, "step": 24760 }, { "epoch": 0.6524795606233722, "grad_norm": 1.5898326635360718, "learning_rate": 5.6939136868144075e-06, "loss": 0.13345763683319092, "step": 24770 }, { "epoch": 0.6527429758678709, "grad_norm": 1.4656144380569458, "learning_rate": 5.686215360392127e-06, "loss": 0.1259496808052063, "step": 24780 }, { "epoch": 0.6530063911123697, "grad_norm": 5.771688461303711, "learning_rate": 5.678520173728414e-06, "loss": 0.12906501293182374, "step": 24790 }, { "epoch": 0.6532698063568684, "grad_norm": 1.9430004358291626, "learning_rate": 5.67082813242416e-06, "loss": 0.13465688228607178, "step": 24800 }, { "epoch": 0.6535332216013672, "grad_norm": 1.1937010288238525, "learning_rate": 5.663139242077971e-06, "loss": 0.1353611707687378, "step": 24810 }, { "epoch": 0.6537966368458659, "grad_norm": 1.493874430656433, "learning_rate": 5.6554535082861496e-06, "loss": 0.14274773597717286, "step": 24820 }, { "epoch": 0.6540600520903646, "grad_norm": 2.4533843994140625, "learning_rate": 5.647770936642705e-06, "loss": 0.1261414647102356, "step": 24830 }, { "epoch": 0.6543234673348634, "grad_norm": 2.3556551933288574, "learning_rate": 5.6400915327393435e-06, "loss": 0.13247766494750976, "step": 24840 }, { "epoch": 0.654586882579362, "grad_norm": 1.1864373683929443, "learning_rate": 5.632415302165464e-06, "loss": 0.1283403754234314, "step": 24850 }, { "epoch": 0.6548502978238608, "grad_norm": 2.196153402328491, "learning_rate": 5.624742250508164e-06, "loss": 0.12520289421081543, "step": 24860 }, { "epoch": 0.6551137130683595, "grad_norm": 1.7835100889205933, "learning_rate": 5.617072383352222e-06, "loss": 0.13029322624206544, "step": 24870 }, { "epoch": 0.6553771283128583, "grad_norm": 1.930707335472107, "learning_rate": 5.6094057062800886e-06, "loss": 0.12471168041229248, "step": 24880 }, { "epoch": 0.655640543557357, "grad_norm": 1.3032314777374268, "learning_rate": 5.601742224871911e-06, "loss": 0.13882253170013428, "step": 24890 }, { "epoch": 0.6559039588018558, "grad_norm": 2.7639782428741455, "learning_rate": 5.594081944705496e-06, "loss": 0.1300508737564087, "step": 24900 }, { "epoch": 0.6561673740463545, "grad_norm": 2.3413474559783936, "learning_rate": 5.586424871356337e-06, "loss": 0.12314949035644532, "step": 24910 }, { "epoch": 0.6564307892908532, "grad_norm": 1.751784086227417, "learning_rate": 5.578771010397572e-06, "loss": 0.13390302658081055, "step": 24920 }, { "epoch": 0.656694204535352, "grad_norm": 1.4943387508392334, "learning_rate": 5.571120367400008e-06, "loss": 0.134382963180542, "step": 24930 }, { "epoch": 0.6569576197798507, "grad_norm": 4.340025901794434, "learning_rate": 5.563472947932123e-06, "loss": 0.12617727518081664, "step": 24940 }, { "epoch": 0.6572210350243495, "grad_norm": 1.89308500289917, "learning_rate": 5.555828757560035e-06, "loss": 0.12860698699951173, "step": 24950 }, { "epoch": 0.6574844502688482, "grad_norm": 1.4106342792510986, "learning_rate": 5.548187801847513e-06, "loss": 0.13682950735092164, "step": 24960 }, { "epoch": 0.657747865513347, "grad_norm": 1.2691173553466797, "learning_rate": 5.5405500863559795e-06, "loss": 0.1330959677696228, "step": 24970 }, { "epoch": 0.6580112807578457, "grad_norm": 2.046901226043701, "learning_rate": 5.532915616644484e-06, "loss": 0.12580628395080568, "step": 24980 }, { "epoch": 0.6582746960023445, "grad_norm": 1.684776782989502, "learning_rate": 5.525284398269733e-06, "loss": 0.11701213121414185, "step": 24990 }, { "epoch": 0.6585381112468431, "grad_norm": 2.1192214488983154, "learning_rate": 5.517656436786059e-06, "loss": 0.12980759143829346, "step": 25000 }, { "epoch": 0.6588015264913418, "grad_norm": 2.3965578079223633, "learning_rate": 5.510031737745405e-06, "loss": 0.12191979885101319, "step": 25010 }, { "epoch": 0.6590649417358406, "grad_norm": 1.752271294593811, "learning_rate": 5.50241030669737e-06, "loss": 0.12898566722869872, "step": 25020 }, { "epoch": 0.6593283569803393, "grad_norm": 2.2170584201812744, "learning_rate": 5.4947921491891565e-06, "loss": 0.1310732364654541, "step": 25030 }, { "epoch": 0.6595917722248381, "grad_norm": 1.1052030324935913, "learning_rate": 5.487177270765588e-06, "loss": 0.12376270294189454, "step": 25040 }, { "epoch": 0.6598551874693368, "grad_norm": 1.6727482080459595, "learning_rate": 5.479565676969101e-06, "loss": 0.12560949325561524, "step": 25050 }, { "epoch": 0.6601186027138356, "grad_norm": 1.1346989870071411, "learning_rate": 5.47195737333974e-06, "loss": 0.11838046312332154, "step": 25060 }, { "epoch": 0.6603820179583343, "grad_norm": 1.2086113691329956, "learning_rate": 5.464352365415162e-06, "loss": 0.13163533210754394, "step": 25070 }, { "epoch": 0.660645433202833, "grad_norm": 1.4120088815689087, "learning_rate": 5.456750658730616e-06, "loss": 0.13677810430526732, "step": 25080 }, { "epoch": 0.6609088484473318, "grad_norm": 1.5078744888305664, "learning_rate": 5.449152258818955e-06, "loss": 0.1298375606536865, "step": 25090 }, { "epoch": 0.6611722636918305, "grad_norm": 3.5070652961730957, "learning_rate": 5.4415571712106184e-06, "loss": 0.1312755823135376, "step": 25100 }, { "epoch": 0.6614356789363293, "grad_norm": 2.139953851699829, "learning_rate": 5.433965401433639e-06, "loss": 0.12207468748092651, "step": 25110 }, { "epoch": 0.661699094180828, "grad_norm": 1.4003115892410278, "learning_rate": 5.426376955013641e-06, "loss": 0.12540192604064943, "step": 25120 }, { "epoch": 0.6619625094253268, "grad_norm": 2.0140628814697266, "learning_rate": 5.4187918374738154e-06, "loss": 0.13716113567352295, "step": 25130 }, { "epoch": 0.6622259246698254, "grad_norm": 1.0709857940673828, "learning_rate": 5.4112100543349365e-06, "loss": 0.1217609167098999, "step": 25140 }, { "epoch": 0.6624893399143242, "grad_norm": 2.213170051574707, "learning_rate": 5.403631611115362e-06, "loss": 0.13366844654083251, "step": 25150 }, { "epoch": 0.6627527551588229, "grad_norm": 1.771912932395935, "learning_rate": 5.396056513331005e-06, "loss": 0.12874397039413452, "step": 25160 }, { "epoch": 0.6630161704033216, "grad_norm": 2.2390494346618652, "learning_rate": 5.388484766495348e-06, "loss": 0.1430896043777466, "step": 25170 }, { "epoch": 0.6632795856478204, "grad_norm": 3.064640522003174, "learning_rate": 5.3809163761194364e-06, "loss": 0.13481814861297609, "step": 25180 }, { "epoch": 0.6635430008923191, "grad_norm": 1.237969994544983, "learning_rate": 5.373351347711865e-06, "loss": 0.10965129137039184, "step": 25190 }, { "epoch": 0.6638064161368179, "grad_norm": 2.659374713897705, "learning_rate": 5.365789686778797e-06, "loss": 0.13645381927490235, "step": 25200 }, { "epoch": 0.6640698313813166, "grad_norm": 2.699413776397705, "learning_rate": 5.358231398823937e-06, "loss": 0.11787506341934204, "step": 25210 }, { "epoch": 0.6643332466258154, "grad_norm": 2.317333459854126, "learning_rate": 5.350676489348521e-06, "loss": 0.1261438846588135, "step": 25220 }, { "epoch": 0.6645966618703141, "grad_norm": 1.672515869140625, "learning_rate": 5.3431249638513475e-06, "loss": 0.14378669261932372, "step": 25230 }, { "epoch": 0.6648600771148129, "grad_norm": 2.0133445262908936, "learning_rate": 5.335576827828737e-06, "loss": 0.128617262840271, "step": 25240 }, { "epoch": 0.6651234923593116, "grad_norm": 2.5744173526763916, "learning_rate": 5.328032086774557e-06, "loss": 0.11897732019424438, "step": 25250 }, { "epoch": 0.6653869076038103, "grad_norm": 2.593395233154297, "learning_rate": 5.320490746180189e-06, "loss": 0.11428614854812622, "step": 25260 }, { "epoch": 0.665650322848309, "grad_norm": 2.3439347743988037, "learning_rate": 5.312952811534542e-06, "loss": 0.12599024772644044, "step": 25270 }, { "epoch": 0.6659137380928077, "grad_norm": 2.1506338119506836, "learning_rate": 5.305418288324058e-06, "loss": 0.12771453857421874, "step": 25280 }, { "epoch": 0.6661771533373065, "grad_norm": 2.013507127761841, "learning_rate": 5.297887182032686e-06, "loss": 0.12513647079467774, "step": 25290 }, { "epoch": 0.6664405685818052, "grad_norm": 1.9908636808395386, "learning_rate": 5.29035949814189e-06, "loss": 0.1197293758392334, "step": 25300 }, { "epoch": 0.666703983826304, "grad_norm": 2.4376139640808105, "learning_rate": 5.2828352421306415e-06, "loss": 0.12762266397476196, "step": 25310 }, { "epoch": 0.6669673990708027, "grad_norm": 2.163377285003662, "learning_rate": 5.275314419475417e-06, "loss": 0.11610825061798095, "step": 25320 }, { "epoch": 0.6672308143153015, "grad_norm": 2.2170512676239014, "learning_rate": 5.267797035650202e-06, "loss": 0.12171125411987305, "step": 25330 }, { "epoch": 0.6674942295598002, "grad_norm": 1.4618148803710938, "learning_rate": 5.2602830961264704e-06, "loss": 0.13123645782470703, "step": 25340 }, { "epoch": 0.6677576448042989, "grad_norm": 2.5360820293426514, "learning_rate": 5.252772606373186e-06, "loss": 0.12327852249145507, "step": 25350 }, { "epoch": 0.6680210600487977, "grad_norm": 1.973042607307434, "learning_rate": 5.245265571856812e-06, "loss": 0.11934485435485839, "step": 25360 }, { "epoch": 0.6682844752932964, "grad_norm": 2.0833656787872314, "learning_rate": 5.237761998041291e-06, "loss": 0.13165898323059083, "step": 25370 }, { "epoch": 0.6685478905377952, "grad_norm": 1.5867940187454224, "learning_rate": 5.230261890388053e-06, "loss": 0.13407182693481445, "step": 25380 }, { "epoch": 0.6688113057822939, "grad_norm": 2.1814606189727783, "learning_rate": 5.2227652543559924e-06, "loss": 0.12399698495864868, "step": 25390 }, { "epoch": 0.6690747210267927, "grad_norm": 2.6975197792053223, "learning_rate": 5.215272095401483e-06, "loss": 0.13079400062561036, "step": 25400 }, { "epoch": 0.6693381362712914, "grad_norm": 1.6143878698349, "learning_rate": 5.207782418978378e-06, "loss": 0.12492165565490723, "step": 25410 }, { "epoch": 0.6696015515157901, "grad_norm": 2.193239212036133, "learning_rate": 5.200296230537982e-06, "loss": 0.12554248571395873, "step": 25420 }, { "epoch": 0.6698649667602888, "grad_norm": 1.8350647687911987, "learning_rate": 5.192813535529066e-06, "loss": 0.13346328735351562, "step": 25430 }, { "epoch": 0.6701283820047875, "grad_norm": 1.2782431840896606, "learning_rate": 5.1853343393978585e-06, "loss": 0.12088682651519775, "step": 25440 }, { "epoch": 0.6703917972492863, "grad_norm": 2.1159896850585938, "learning_rate": 5.177858647588037e-06, "loss": 0.12740762233734132, "step": 25450 }, { "epoch": 0.670655212493785, "grad_norm": 1.9556351900100708, "learning_rate": 5.170386465540741e-06, "loss": 0.11999534368515015, "step": 25460 }, { "epoch": 0.6709186277382838, "grad_norm": 1.6217342615127563, "learning_rate": 5.162917798694546e-06, "loss": 0.12082363367080688, "step": 25470 }, { "epoch": 0.6711820429827825, "grad_norm": 2.5088510513305664, "learning_rate": 5.15545265248546e-06, "loss": 0.13281023502349854, "step": 25480 }, { "epoch": 0.6714454582272813, "grad_norm": 1.733744502067566, "learning_rate": 5.147991032346948e-06, "loss": 0.11465080976486205, "step": 25490 }, { "epoch": 0.67170887347178, "grad_norm": 1.7596796751022339, "learning_rate": 5.140532943709893e-06, "loss": 0.1337079167366028, "step": 25500 }, { "epoch": 0.6719722887162788, "grad_norm": 1.6178165674209595, "learning_rate": 5.133078392002623e-06, "loss": 0.13537509441375734, "step": 25510 }, { "epoch": 0.6722357039607775, "grad_norm": 1.725416660308838, "learning_rate": 5.125627382650873e-06, "loss": 0.12156558036804199, "step": 25520 }, { "epoch": 0.6724991192052762, "grad_norm": 3.246931314468384, "learning_rate": 5.118179921077808e-06, "loss": 0.13194814920425416, "step": 25530 }, { "epoch": 0.672762534449775, "grad_norm": 2.4095163345336914, "learning_rate": 5.1107360127040206e-06, "loss": 0.12952966690063478, "step": 25540 }, { "epoch": 0.6730259496942737, "grad_norm": 2.000789165496826, "learning_rate": 5.103295662947504e-06, "loss": 0.1251981258392334, "step": 25550 }, { "epoch": 0.6732893649387724, "grad_norm": 4.084661960601807, "learning_rate": 5.095858877223665e-06, "loss": 0.11721367835998535, "step": 25560 }, { "epoch": 0.6735527801832711, "grad_norm": 1.5544350147247314, "learning_rate": 5.088425660945319e-06, "loss": 0.12817554473876952, "step": 25570 }, { "epoch": 0.6738161954277699, "grad_norm": 1.8803746700286865, "learning_rate": 5.080996019522675e-06, "loss": 0.11735271215438843, "step": 25580 }, { "epoch": 0.6740796106722686, "grad_norm": 3.2741665840148926, "learning_rate": 5.073569958363356e-06, "loss": 0.13284690380096437, "step": 25590 }, { "epoch": 0.6743430259167673, "grad_norm": 1.8518338203430176, "learning_rate": 5.06614748287237e-06, "loss": 0.11651338338851928, "step": 25600 }, { "epoch": 0.6746064411612661, "grad_norm": 1.956048607826233, "learning_rate": 5.058728598452103e-06, "loss": 0.1326156497001648, "step": 25610 }, { "epoch": 0.6748698564057648, "grad_norm": 1.980670690536499, "learning_rate": 5.05131331050235e-06, "loss": 0.11326308250427246, "step": 25620 }, { "epoch": 0.6751332716502636, "grad_norm": 1.7232425212860107, "learning_rate": 5.043901624420273e-06, "loss": 0.129158878326416, "step": 25630 }, { "epoch": 0.6753966868947623, "grad_norm": 2.965353488922119, "learning_rate": 5.036493545600421e-06, "loss": 0.12006962299346924, "step": 25640 }, { "epoch": 0.6756601021392611, "grad_norm": 1.7375879287719727, "learning_rate": 5.029089079434708e-06, "loss": 0.11950781345367431, "step": 25650 }, { "epoch": 0.6759235173837598, "grad_norm": 1.7164090871810913, "learning_rate": 5.021688231312423e-06, "loss": 0.11351621150970459, "step": 25660 }, { "epoch": 0.6761869326282586, "grad_norm": 1.2775717973709106, "learning_rate": 5.014291006620229e-06, "loss": 0.12208459377288819, "step": 25670 }, { "epoch": 0.6764503478727573, "grad_norm": 1.2091182470321655, "learning_rate": 5.006897410742142e-06, "loss": 0.12752645015716552, "step": 25680 }, { "epoch": 0.676713763117256, "grad_norm": 1.5378831624984741, "learning_rate": 4.999507449059542e-06, "loss": 0.11702570915222169, "step": 25690 }, { "epoch": 0.6769771783617548, "grad_norm": 1.5661667585372925, "learning_rate": 4.992121126951159e-06, "loss": 0.12776558399200438, "step": 25700 }, { "epoch": 0.6772405936062534, "grad_norm": 1.6906994581222534, "learning_rate": 4.984738449793074e-06, "loss": 0.12680916786193847, "step": 25710 }, { "epoch": 0.6775040088507522, "grad_norm": 2.372471570968628, "learning_rate": 4.977359422958731e-06, "loss": 0.12438621520996093, "step": 25720 }, { "epoch": 0.6777674240952509, "grad_norm": 1.4728449583053589, "learning_rate": 4.96998405181889e-06, "loss": 0.130432653427124, "step": 25730 }, { "epoch": 0.6780308393397497, "grad_norm": 2.0444018840789795, "learning_rate": 4.962612341741666e-06, "loss": 0.10929815769195557, "step": 25740 }, { "epoch": 0.6782942545842484, "grad_norm": 1.9644142389297485, "learning_rate": 4.955244298092514e-06, "loss": 0.11820098161697387, "step": 25750 }, { "epoch": 0.6785576698287472, "grad_norm": 1.8170255422592163, "learning_rate": 4.947879926234212e-06, "loss": 0.13273589611053466, "step": 25760 }, { "epoch": 0.6788210850732459, "grad_norm": 3.0021750926971436, "learning_rate": 4.940519231526867e-06, "loss": 0.125988233089447, "step": 25770 }, { "epoch": 0.6790845003177446, "grad_norm": 1.4715160131454468, "learning_rate": 4.933162219327909e-06, "loss": 0.11931881904602051, "step": 25780 }, { "epoch": 0.6793479155622434, "grad_norm": 1.5691847801208496, "learning_rate": 4.925808894992087e-06, "loss": 0.12216781377792359, "step": 25790 }, { "epoch": 0.6796113308067421, "grad_norm": 1.353743553161621, "learning_rate": 4.918459263871474e-06, "loss": 0.12409648895263672, "step": 25800 }, { "epoch": 0.6798747460512409, "grad_norm": 1.788814663887024, "learning_rate": 4.911113331315448e-06, "loss": 0.10957249402999877, "step": 25810 }, { "epoch": 0.6801381612957396, "grad_norm": 4.473914623260498, "learning_rate": 4.903771102670686e-06, "loss": 0.12787349224090577, "step": 25820 }, { "epoch": 0.6804015765402384, "grad_norm": 1.4976993799209595, "learning_rate": 4.896432583281189e-06, "loss": 0.12458195686340331, "step": 25830 }, { "epoch": 0.680664991784737, "grad_norm": 1.8903135061264038, "learning_rate": 4.88909777848824e-06, "loss": 0.12345458269119262, "step": 25840 }, { "epoch": 0.6809284070292358, "grad_norm": 2.216174364089966, "learning_rate": 4.881766693630439e-06, "loss": 0.13239420652389527, "step": 25850 }, { "epoch": 0.6811918222737345, "grad_norm": 1.6763803958892822, "learning_rate": 4.874439334043654e-06, "loss": 0.11905602216720582, "step": 25860 }, { "epoch": 0.6814552375182332, "grad_norm": 1.3827444314956665, "learning_rate": 4.8671157050610525e-06, "loss": 0.12761656045913697, "step": 25870 }, { "epoch": 0.681718652762732, "grad_norm": 2.1799354553222656, "learning_rate": 4.859795812013099e-06, "loss": 0.13050510883331298, "step": 25880 }, { "epoch": 0.6819820680072307, "grad_norm": 1.772984266281128, "learning_rate": 4.852479660227518e-06, "loss": 0.12020308971405029, "step": 25890 }, { "epoch": 0.6822454832517295, "grad_norm": 2.158005475997925, "learning_rate": 4.8451672550293236e-06, "loss": 0.11236072778701782, "step": 25900 }, { "epoch": 0.6825088984962282, "grad_norm": 2.2076375484466553, "learning_rate": 4.837858601740799e-06, "loss": 0.11868202686309814, "step": 25910 }, { "epoch": 0.682772313740727, "grad_norm": 2.393004894256592, "learning_rate": 4.830553705681496e-06, "loss": 0.12295454740524292, "step": 25920 }, { "epoch": 0.6830357289852257, "grad_norm": 1.9114913940429688, "learning_rate": 4.823252572168236e-06, "loss": 0.13024792671203614, "step": 25930 }, { "epoch": 0.6832991442297245, "grad_norm": 1.2344367504119873, "learning_rate": 4.815955206515104e-06, "loss": 0.12130583524703979, "step": 25940 }, { "epoch": 0.6835625594742232, "grad_norm": 3.531301498413086, "learning_rate": 4.808661614033422e-06, "loss": 0.1251712679862976, "step": 25950 }, { "epoch": 0.6838259747187219, "grad_norm": 1.1298044919967651, "learning_rate": 4.801371800031796e-06, "loss": 0.12135320901870728, "step": 25960 }, { "epoch": 0.6840893899632207, "grad_norm": 2.032155752182007, "learning_rate": 4.794085769816057e-06, "loss": 0.10451215505599976, "step": 25970 }, { "epoch": 0.6843528052077194, "grad_norm": 1.8818185329437256, "learning_rate": 4.786803528689304e-06, "loss": 0.13290719985961913, "step": 25980 }, { "epoch": 0.6846162204522181, "grad_norm": 3.436281204223633, "learning_rate": 4.779525081951857e-06, "loss": 0.1280445337295532, "step": 25990 }, { "epoch": 0.6848796356967168, "grad_norm": 2.5037190914154053, "learning_rate": 4.772250434901281e-06, "loss": 0.12375850677490234, "step": 26000 }, { "epoch": 0.6851430509412156, "grad_norm": 1.6202713251113892, "learning_rate": 4.7649795928323875e-06, "loss": 0.11450966596603393, "step": 26010 }, { "epoch": 0.6854064661857143, "grad_norm": 1.804693341255188, "learning_rate": 4.757712561037203e-06, "loss": 0.12172110080718994, "step": 26020 }, { "epoch": 0.6856698814302131, "grad_norm": 2.737114429473877, "learning_rate": 4.750449344804989e-06, "loss": 0.12376172542572021, "step": 26030 }, { "epoch": 0.6859332966747118, "grad_norm": 4.542108058929443, "learning_rate": 4.743189949422227e-06, "loss": 0.14596214294433593, "step": 26040 }, { "epoch": 0.6861967119192105, "grad_norm": 2.0848426818847656, "learning_rate": 4.735934380172613e-06, "loss": 0.12027740478515625, "step": 26050 }, { "epoch": 0.6864601271637093, "grad_norm": 2.1186773777008057, "learning_rate": 4.728682642337072e-06, "loss": 0.12360453605651855, "step": 26060 }, { "epoch": 0.686723542408208, "grad_norm": 1.8360328674316406, "learning_rate": 4.7214347411937325e-06, "loss": 0.12967324256896973, "step": 26070 }, { "epoch": 0.6869869576527068, "grad_norm": 1.1216509342193604, "learning_rate": 4.714190682017918e-06, "loss": 0.12462942600250244, "step": 26080 }, { "epoch": 0.6872503728972055, "grad_norm": 1.6691501140594482, "learning_rate": 4.706950470082177e-06, "loss": 0.11941962242126465, "step": 26090 }, { "epoch": 0.6875137881417043, "grad_norm": 2.2850539684295654, "learning_rate": 4.699714110656242e-06, "loss": 0.11675863265991211, "step": 26100 }, { "epoch": 0.687777203386203, "grad_norm": 1.9370723962783813, "learning_rate": 4.6924816090070575e-06, "loss": 0.1273660659790039, "step": 26110 }, { "epoch": 0.6880406186307017, "grad_norm": 2.7019436359405518, "learning_rate": 4.685252970398742e-06, "loss": 0.13324601650238038, "step": 26120 }, { "epoch": 0.6883040338752004, "grad_norm": 1.5368582010269165, "learning_rate": 4.678028200092607e-06, "loss": 0.12456917762756348, "step": 26130 }, { "epoch": 0.6885674491196991, "grad_norm": 1.4287878274917603, "learning_rate": 4.670807303347161e-06, "loss": 0.12167155742645264, "step": 26140 }, { "epoch": 0.6888308643641979, "grad_norm": 2.481729030609131, "learning_rate": 4.663590285418082e-06, "loss": 0.12699625492095948, "step": 26150 }, { "epoch": 0.6890942796086966, "grad_norm": 3.373718023300171, "learning_rate": 4.656377151558224e-06, "loss": 0.11187907457351684, "step": 26160 }, { "epoch": 0.6893576948531954, "grad_norm": 1.465585470199585, "learning_rate": 4.64916790701762e-06, "loss": 0.1337738037109375, "step": 26170 }, { "epoch": 0.6896211100976941, "grad_norm": 2.8149235248565674, "learning_rate": 4.641962557043462e-06, "loss": 0.11439747810363769, "step": 26180 }, { "epoch": 0.6898845253421929, "grad_norm": 2.904231071472168, "learning_rate": 4.634761106880127e-06, "loss": 0.1159211277961731, "step": 26190 }, { "epoch": 0.6901479405866916, "grad_norm": 2.4520530700683594, "learning_rate": 4.627563561769136e-06, "loss": 0.11590389013290406, "step": 26200 }, { "epoch": 0.6904113558311903, "grad_norm": 1.7171440124511719, "learning_rate": 4.620369926949164e-06, "loss": 0.12243932485580444, "step": 26210 }, { "epoch": 0.6906747710756891, "grad_norm": 1.892367959022522, "learning_rate": 4.613180207656061e-06, "loss": 0.1088054895401001, "step": 26220 }, { "epoch": 0.6909381863201878, "grad_norm": 1.3000959157943726, "learning_rate": 4.60599440912281e-06, "loss": 0.12052830457687377, "step": 26230 }, { "epoch": 0.6912016015646866, "grad_norm": 2.28590989112854, "learning_rate": 4.598812536579545e-06, "loss": 0.11552619934082031, "step": 26240 }, { "epoch": 0.6914650168091853, "grad_norm": 1.7057980298995972, "learning_rate": 4.591634595253541e-06, "loss": 0.11032028198242187, "step": 26250 }, { "epoch": 0.6917284320536841, "grad_norm": 2.269038200378418, "learning_rate": 4.584460590369214e-06, "loss": 0.1325057029724121, "step": 26260 }, { "epoch": 0.6919918472981827, "grad_norm": 1.8336213827133179, "learning_rate": 4.577290527148118e-06, "loss": 0.1256972074508667, "step": 26270 }, { "epoch": 0.6922552625426815, "grad_norm": 1.7620259523391724, "learning_rate": 4.5701244108089325e-06, "loss": 0.12582999467849731, "step": 26280 }, { "epoch": 0.6925186777871802, "grad_norm": 1.8952890634536743, "learning_rate": 4.562962246567466e-06, "loss": 0.1280970811843872, "step": 26290 }, { "epoch": 0.6927820930316789, "grad_norm": 2.875483751296997, "learning_rate": 4.555804039636649e-06, "loss": 0.12116649150848388, "step": 26300 }, { "epoch": 0.6930455082761777, "grad_norm": 5.255125999450684, "learning_rate": 4.548649795226532e-06, "loss": 0.11964194774627686, "step": 26310 }, { "epoch": 0.6933089235206764, "grad_norm": 1.6912431716918945, "learning_rate": 4.541499518544292e-06, "loss": 0.11799527406692505, "step": 26320 }, { "epoch": 0.6935723387651752, "grad_norm": 3.794062852859497, "learning_rate": 4.5343532147941985e-06, "loss": 0.1241004467010498, "step": 26330 }, { "epoch": 0.6938357540096739, "grad_norm": 1.8534700870513916, "learning_rate": 4.527210889177639e-06, "loss": 0.12852514982223512, "step": 26340 }, { "epoch": 0.6940991692541727, "grad_norm": 3.8717775344848633, "learning_rate": 4.520072546893115e-06, "loss": 0.13203078508377075, "step": 26350 }, { "epoch": 0.6943625844986714, "grad_norm": 2.040114164352417, "learning_rate": 4.512938193136216e-06, "loss": 0.11563601493835449, "step": 26360 }, { "epoch": 0.6946259997431702, "grad_norm": 1.407114863395691, "learning_rate": 4.5058078330996315e-06, "loss": 0.12353546619415283, "step": 26370 }, { "epoch": 0.6948894149876689, "grad_norm": 2.1362695693969727, "learning_rate": 4.498681471973145e-06, "loss": 0.11633710861206055, "step": 26380 }, { "epoch": 0.6951528302321676, "grad_norm": 2.0555579662323, "learning_rate": 4.491559114943628e-06, "loss": 0.1207036018371582, "step": 26390 }, { "epoch": 0.6954162454766664, "grad_norm": 2.1427762508392334, "learning_rate": 4.484440767195044e-06, "loss": 0.1295487880706787, "step": 26400 }, { "epoch": 0.695679660721165, "grad_norm": 1.3545873165130615, "learning_rate": 4.477326433908435e-06, "loss": 0.11854559183120728, "step": 26410 }, { "epoch": 0.6959430759656638, "grad_norm": 1.037865161895752, "learning_rate": 4.4702161202619085e-06, "loss": 0.12033686637878419, "step": 26420 }, { "epoch": 0.6962064912101625, "grad_norm": 1.7579319477081299, "learning_rate": 4.463109831430668e-06, "loss": 0.11953799724578858, "step": 26430 }, { "epoch": 0.6964699064546613, "grad_norm": 2.3128902912139893, "learning_rate": 4.456007572586971e-06, "loss": 0.12289603948593139, "step": 26440 }, { "epoch": 0.69673332169916, "grad_norm": 3.302377939224243, "learning_rate": 4.448909348900159e-06, "loss": 0.12105990648269653, "step": 26450 }, { "epoch": 0.6969967369436588, "grad_norm": 2.0492687225341797, "learning_rate": 4.441815165536612e-06, "loss": 0.1441558837890625, "step": 26460 }, { "epoch": 0.6972601521881575, "grad_norm": 2.047536611557007, "learning_rate": 4.434725027659786e-06, "loss": 0.11619887351989747, "step": 26470 }, { "epoch": 0.6975235674326562, "grad_norm": 2.2583823204040527, "learning_rate": 4.427638940430192e-06, "loss": 0.11390492916107178, "step": 26480 }, { "epoch": 0.697786982677155, "grad_norm": 1.6283906698226929, "learning_rate": 4.420556909005392e-06, "loss": 0.11876590251922607, "step": 26490 }, { "epoch": 0.6980503979216537, "grad_norm": 4.093177318572998, "learning_rate": 4.41347893853999e-06, "loss": 0.13037559986114503, "step": 26500 }, { "epoch": 0.6983138131661525, "grad_norm": 3.6358225345611572, "learning_rate": 4.40640503418564e-06, "loss": 0.1110456109046936, "step": 26510 }, { "epoch": 0.6985772284106512, "grad_norm": 1.7168653011322021, "learning_rate": 4.39933520109103e-06, "loss": 0.10842888355255127, "step": 26520 }, { "epoch": 0.69884064365515, "grad_norm": 2.9122073650360107, "learning_rate": 4.3922694444018985e-06, "loss": 0.11768146753311157, "step": 26530 }, { "epoch": 0.6991040588996487, "grad_norm": 2.316253423690796, "learning_rate": 4.385207769261007e-06, "loss": 0.11818331480026245, "step": 26540 }, { "epoch": 0.6993674741441475, "grad_norm": 3.698465585708618, "learning_rate": 4.378150180808136e-06, "loss": 0.12274245023727418, "step": 26550 }, { "epoch": 0.6996308893886461, "grad_norm": 2.089400053024292, "learning_rate": 4.371096684180114e-06, "loss": 0.1272583484649658, "step": 26560 }, { "epoch": 0.6998943046331448, "grad_norm": 2.2182676792144775, "learning_rate": 4.364047284510774e-06, "loss": 0.10338400602340699, "step": 26570 }, { "epoch": 0.7001577198776436, "grad_norm": 1.8318662643432617, "learning_rate": 4.357001986930982e-06, "loss": 0.10921673774719239, "step": 26580 }, { "epoch": 0.7004211351221423, "grad_norm": 3.3936960697174072, "learning_rate": 4.349960796568601e-06, "loss": 0.12262170314788819, "step": 26590 }, { "epoch": 0.7006845503666411, "grad_norm": 1.954695701599121, "learning_rate": 4.342923718548512e-06, "loss": 0.1145561695098877, "step": 26600 }, { "epoch": 0.7009479656111398, "grad_norm": 2.0422232151031494, "learning_rate": 4.335890757992611e-06, "loss": 0.11324523687362671, "step": 26610 }, { "epoch": 0.7012113808556386, "grad_norm": 2.6669981479644775, "learning_rate": 4.328861920019787e-06, "loss": 0.11343731880187988, "step": 26620 }, { "epoch": 0.7014747961001373, "grad_norm": 3.6952407360076904, "learning_rate": 4.321837209745931e-06, "loss": 0.11504724025726318, "step": 26630 }, { "epoch": 0.701738211344636, "grad_norm": 1.8185707330703735, "learning_rate": 4.3148166322839305e-06, "loss": 0.11628437042236328, "step": 26640 }, { "epoch": 0.7020016265891348, "grad_norm": 2.0889742374420166, "learning_rate": 4.3078001927436605e-06, "loss": 0.1066625952720642, "step": 26650 }, { "epoch": 0.7022650418336335, "grad_norm": 2.208969831466675, "learning_rate": 4.300787896231997e-06, "loss": 0.11117961406707763, "step": 26660 }, { "epoch": 0.7025284570781323, "grad_norm": 3.4926350116729736, "learning_rate": 4.293779747852791e-06, "loss": 0.12112195491790771, "step": 26670 }, { "epoch": 0.702791872322631, "grad_norm": 3.4444773197174072, "learning_rate": 4.286775752706865e-06, "loss": 0.12442445755004883, "step": 26680 }, { "epoch": 0.7030552875671298, "grad_norm": 2.664132833480835, "learning_rate": 4.279775915892038e-06, "loss": 0.12013448476791382, "step": 26690 }, { "epoch": 0.7033187028116284, "grad_norm": 1.8657097816467285, "learning_rate": 4.2727802425030885e-06, "loss": 0.10677926540374756, "step": 26700 }, { "epoch": 0.7035821180561272, "grad_norm": 1.6496355533599854, "learning_rate": 4.265788737631779e-06, "loss": 0.1099427342414856, "step": 26710 }, { "epoch": 0.7038455333006259, "grad_norm": 1.4497063159942627, "learning_rate": 4.258801406366817e-06, "loss": 0.12086975574493408, "step": 26720 }, { "epoch": 0.7041089485451246, "grad_norm": 3.8659303188323975, "learning_rate": 4.251818253793885e-06, "loss": 0.119498872756958, "step": 26730 }, { "epoch": 0.7043723637896234, "grad_norm": 1.8630094528198242, "learning_rate": 4.244839284995627e-06, "loss": 0.12573350667953492, "step": 26740 }, { "epoch": 0.7046357790341221, "grad_norm": 2.269444465637207, "learning_rate": 4.237864505051638e-06, "loss": 0.126607084274292, "step": 26750 }, { "epoch": 0.7048991942786209, "grad_norm": 1.9730706214904785, "learning_rate": 4.230893919038461e-06, "loss": 0.12338886260986329, "step": 26760 }, { "epoch": 0.7051626095231196, "grad_norm": 1.8239703178405762, "learning_rate": 4.223927532029592e-06, "loss": 0.1086989164352417, "step": 26770 }, { "epoch": 0.7054260247676184, "grad_norm": 2.027935743331909, "learning_rate": 4.21696534909546e-06, "loss": 0.13012611865997314, "step": 26780 }, { "epoch": 0.7056894400121171, "grad_norm": 1.952977180480957, "learning_rate": 4.2100073753034535e-06, "loss": 0.12340890169143677, "step": 26790 }, { "epoch": 0.7059528552566159, "grad_norm": 2.451727867126465, "learning_rate": 4.203053615717884e-06, "loss": 0.11967780590057372, "step": 26800 }, { "epoch": 0.7062162705011146, "grad_norm": 1.8471583127975464, "learning_rate": 4.196104075399986e-06, "loss": 0.1257554531097412, "step": 26810 }, { "epoch": 0.7064796857456133, "grad_norm": 1.8613475561141968, "learning_rate": 4.1891587594079485e-06, "loss": 0.12757556438446044, "step": 26820 }, { "epoch": 0.7067431009901121, "grad_norm": 2.0140485763549805, "learning_rate": 4.182217672796867e-06, "loss": 0.12108402252197266, "step": 26830 }, { "epoch": 0.7070065162346107, "grad_norm": 2.537404775619507, "learning_rate": 4.175280820618763e-06, "loss": 0.12180197238922119, "step": 26840 }, { "epoch": 0.7072699314791095, "grad_norm": 2.3537163734436035, "learning_rate": 4.168348207922577e-06, "loss": 0.10205235481262206, "step": 26850 }, { "epoch": 0.7075333467236082, "grad_norm": 2.4961700439453125, "learning_rate": 4.16141983975416e-06, "loss": 0.1200857162475586, "step": 26860 }, { "epoch": 0.707796761968107, "grad_norm": 3.3632216453552246, "learning_rate": 4.154495721156287e-06, "loss": 0.12598834037780762, "step": 26870 }, { "epoch": 0.7080601772126057, "grad_norm": 2.8951714038848877, "learning_rate": 4.147575857168623e-06, "loss": 0.11884348392486573, "step": 26880 }, { "epoch": 0.7083235924571045, "grad_norm": 1.6499073505401611, "learning_rate": 4.140660252827746e-06, "loss": 0.10634064674377441, "step": 26890 }, { "epoch": 0.7085870077016032, "grad_norm": 2.620936632156372, "learning_rate": 4.13374891316713e-06, "loss": 0.1138524055480957, "step": 26900 }, { "epoch": 0.7088504229461019, "grad_norm": 1.6431175470352173, "learning_rate": 4.126841843217144e-06, "loss": 0.10902796983718872, "step": 26910 }, { "epoch": 0.7091138381906007, "grad_norm": 2.7639756202697754, "learning_rate": 4.1199390480050605e-06, "loss": 0.11918041706085206, "step": 26920 }, { "epoch": 0.7093772534350994, "grad_norm": 2.3704559803009033, "learning_rate": 4.113040532555021e-06, "loss": 0.11574499607086182, "step": 26930 }, { "epoch": 0.7096406686795982, "grad_norm": 2.035290002822876, "learning_rate": 4.106146301888065e-06, "loss": 0.12331919670104981, "step": 26940 }, { "epoch": 0.7099040839240969, "grad_norm": 1.4049837589263916, "learning_rate": 4.0992563610221155e-06, "loss": 0.11348507404327393, "step": 26950 }, { "epoch": 0.7101674991685957, "grad_norm": 2.306044816970825, "learning_rate": 4.092370714971966e-06, "loss": 0.1130854606628418, "step": 26960 }, { "epoch": 0.7104309144130944, "grad_norm": 2.768244981765747, "learning_rate": 4.085489368749287e-06, "loss": 0.12600769996643066, "step": 26970 }, { "epoch": 0.7106943296575932, "grad_norm": 1.8466508388519287, "learning_rate": 4.07861232736262e-06, "loss": 0.11902087926864624, "step": 26980 }, { "epoch": 0.7109577449020918, "grad_norm": 3.837592363357544, "learning_rate": 4.0717395958173665e-06, "loss": 0.12085504531860351, "step": 26990 }, { "epoch": 0.7112211601465905, "grad_norm": 3.642681360244751, "learning_rate": 4.064871179115807e-06, "loss": 0.12092745304107666, "step": 27000 }, { "epoch": 0.7114845753910893, "grad_norm": 2.284507989883423, "learning_rate": 4.0580070822570694e-06, "loss": 0.11881099939346314, "step": 27010 }, { "epoch": 0.711747990635588, "grad_norm": 1.640337586402893, "learning_rate": 4.051147310237129e-06, "loss": 0.11330009698867798, "step": 27020 }, { "epoch": 0.7120114058800868, "grad_norm": 1.5526736974716187, "learning_rate": 4.044291868048837e-06, "loss": 0.10736293792724609, "step": 27030 }, { "epoch": 0.7122748211245855, "grad_norm": 2.2466132640838623, "learning_rate": 4.0374407606818684e-06, "loss": 0.11898529529571533, "step": 27040 }, { "epoch": 0.7125382363690843, "grad_norm": 2.4998207092285156, "learning_rate": 4.03059399312277e-06, "loss": 0.13691201210021972, "step": 27050 }, { "epoch": 0.712801651613583, "grad_norm": 2.292668342590332, "learning_rate": 4.0237515703549015e-06, "loss": 0.11655626296997071, "step": 27060 }, { "epoch": 0.7130650668580818, "grad_norm": 1.3303741216659546, "learning_rate": 4.0169134973584726e-06, "loss": 0.10680043697357178, "step": 27070 }, { "epoch": 0.7133284821025805, "grad_norm": 2.2989699840545654, "learning_rate": 4.010079779110537e-06, "loss": 0.11910378932952881, "step": 27080 }, { "epoch": 0.7135918973470792, "grad_norm": 1.609040379524231, "learning_rate": 4.003250420584966e-06, "loss": 0.13427588939666749, "step": 27090 }, { "epoch": 0.713855312591578, "grad_norm": 2.7562851905822754, "learning_rate": 3.996425426752459e-06, "loss": 0.13219802379608153, "step": 27100 }, { "epoch": 0.7141187278360767, "grad_norm": 2.8153634071350098, "learning_rate": 3.989604802580545e-06, "loss": 0.12330658435821533, "step": 27110 }, { "epoch": 0.7143821430805755, "grad_norm": 1.7776134014129639, "learning_rate": 3.982788553033561e-06, "loss": 0.11771135330200196, "step": 27120 }, { "epoch": 0.7146455583250741, "grad_norm": 2.6481831073760986, "learning_rate": 3.975976683072676e-06, "loss": 0.12527189254760743, "step": 27130 }, { "epoch": 0.714908973569573, "grad_norm": 1.7080787420272827, "learning_rate": 3.9691691976558636e-06, "loss": 0.130148446559906, "step": 27140 }, { "epoch": 0.7151723888140716, "grad_norm": 1.702757716178894, "learning_rate": 3.9623661017378946e-06, "loss": 0.11597278118133544, "step": 27150 }, { "epoch": 0.7154358040585704, "grad_norm": 2.305845022201538, "learning_rate": 3.955567400270365e-06, "loss": 0.12041558027267456, "step": 27160 }, { "epoch": 0.7156992193030691, "grad_norm": 2.215188503265381, "learning_rate": 3.948773098201657e-06, "loss": 0.11499714851379395, "step": 27170 }, { "epoch": 0.7159626345475678, "grad_norm": 3.128521203994751, "learning_rate": 3.941983200476969e-06, "loss": 0.12410346269607545, "step": 27180 }, { "epoch": 0.7162260497920666, "grad_norm": 3.0618152618408203, "learning_rate": 3.935197712038267e-06, "loss": 0.1176764726638794, "step": 27190 }, { "epoch": 0.7164894650365653, "grad_norm": 1.4966659545898438, "learning_rate": 3.928416637824326e-06, "loss": 0.10844321250915527, "step": 27200 }, { "epoch": 0.7167528802810641, "grad_norm": 2.2830586433410645, "learning_rate": 3.921639982770709e-06, "loss": 0.12604944705963134, "step": 27210 }, { "epoch": 0.7170162955255628, "grad_norm": 1.9704365730285645, "learning_rate": 3.914867751809756e-06, "loss": 0.12214329242706298, "step": 27220 }, { "epoch": 0.7172797107700616, "grad_norm": 2.914777994155884, "learning_rate": 3.908099949870588e-06, "loss": 0.1186862587928772, "step": 27230 }, { "epoch": 0.7175431260145603, "grad_norm": 2.173161268234253, "learning_rate": 3.901336581879102e-06, "loss": 0.11180191040039063, "step": 27240 }, { "epoch": 0.717806541259059, "grad_norm": 2.806457757949829, "learning_rate": 3.894577652757968e-06, "loss": 0.1229783535003662, "step": 27250 }, { "epoch": 0.7180699565035578, "grad_norm": 1.8316577672958374, "learning_rate": 3.887823167426633e-06, "loss": 0.11711606979370118, "step": 27260 }, { "epoch": 0.7183333717480564, "grad_norm": 1.28608238697052, "learning_rate": 3.8810731308013005e-06, "loss": 0.11530792713165283, "step": 27270 }, { "epoch": 0.7185967869925552, "grad_norm": 1.9602203369140625, "learning_rate": 3.874327547794931e-06, "loss": 0.11444146633148193, "step": 27280 }, { "epoch": 0.7188602022370539, "grad_norm": 2.028308391571045, "learning_rate": 3.8675864233172575e-06, "loss": 0.12067440748214722, "step": 27290 }, { "epoch": 0.7191236174815527, "grad_norm": 1.588750958442688, "learning_rate": 3.860849762274759e-06, "loss": 0.11140980720520019, "step": 27300 }, { "epoch": 0.7193870327260514, "grad_norm": 2.612454652786255, "learning_rate": 3.854117569570674e-06, "loss": 0.11961381435394287, "step": 27310 }, { "epoch": 0.7196504479705502, "grad_norm": 1.608583688735962, "learning_rate": 3.8473898501049745e-06, "loss": 0.11637678146362304, "step": 27320 }, { "epoch": 0.7199138632150489, "grad_norm": 2.1317973136901855, "learning_rate": 3.840666608774385e-06, "loss": 0.10888699293136597, "step": 27330 }, { "epoch": 0.7201772784595476, "grad_norm": 2.273522138595581, "learning_rate": 3.833947850472378e-06, "loss": 0.11614391803741456, "step": 27340 }, { "epoch": 0.7204406937040464, "grad_norm": 3.422173023223877, "learning_rate": 3.827233580089152e-06, "loss": 0.12660032510757446, "step": 27350 }, { "epoch": 0.7207041089485451, "grad_norm": 3.4146227836608887, "learning_rate": 3.82052380251164e-06, "loss": 0.12602605819702148, "step": 27360 }, { "epoch": 0.7209675241930439, "grad_norm": 2.606353282928467, "learning_rate": 3.8138185226235114e-06, "loss": 0.111050283908844, "step": 27370 }, { "epoch": 0.7212309394375426, "grad_norm": 3.037069320678711, "learning_rate": 3.8071177453051523e-06, "loss": 0.12639312744140624, "step": 27380 }, { "epoch": 0.7214943546820414, "grad_norm": 2.4512882232666016, "learning_rate": 3.8004214754336853e-06, "loss": 0.11725349426269531, "step": 27390 }, { "epoch": 0.7217577699265401, "grad_norm": 2.2892284393310547, "learning_rate": 3.7937297178829458e-06, "loss": 0.10771315097808838, "step": 27400 }, { "epoch": 0.7220211851710389, "grad_norm": 2.093550205230713, "learning_rate": 3.7870424775234703e-06, "loss": 0.10945441722869872, "step": 27410 }, { "epoch": 0.7222846004155375, "grad_norm": 1.5276533365249634, "learning_rate": 3.780359759222534e-06, "loss": 0.1271977186203003, "step": 27420 }, { "epoch": 0.7225480156600362, "grad_norm": 2.116854190826416, "learning_rate": 3.773681567844103e-06, "loss": 0.12090098857879639, "step": 27430 }, { "epoch": 0.722811430904535, "grad_norm": 1.5775858163833618, "learning_rate": 3.767007908248852e-06, "loss": 0.11325161457061768, "step": 27440 }, { "epoch": 0.7230748461490337, "grad_norm": 3.127258777618408, "learning_rate": 3.7603387852941608e-06, "loss": 0.11302859783172607, "step": 27450 }, { "epoch": 0.7233382613935325, "grad_norm": 2.5898797512054443, "learning_rate": 3.753674203834099e-06, "loss": 0.10894417762756348, "step": 27460 }, { "epoch": 0.7236016766380312, "grad_norm": 3.157992124557495, "learning_rate": 3.747014168719445e-06, "loss": 0.11047008037567138, "step": 27470 }, { "epoch": 0.72386509188253, "grad_norm": 2.050549268722534, "learning_rate": 3.740358684797656e-06, "loss": 0.11585736274719238, "step": 27480 }, { "epoch": 0.7241285071270287, "grad_norm": 1.7023861408233643, "learning_rate": 3.7337077569128797e-06, "loss": 0.11184508800506592, "step": 27490 }, { "epoch": 0.7243919223715275, "grad_norm": 2.113564968109131, "learning_rate": 3.7270613899059495e-06, "loss": 0.1057029128074646, "step": 27500 }, { "epoch": 0.7246553376160262, "grad_norm": 1.9088757038116455, "learning_rate": 3.7204195886143736e-06, "loss": 0.10692421197891236, "step": 27510 }, { "epoch": 0.7249187528605249, "grad_norm": 2.243518114089966, "learning_rate": 3.713782357872352e-06, "loss": 0.12149257659912109, "step": 27520 }, { "epoch": 0.7251821681050237, "grad_norm": 2.227588653564453, "learning_rate": 3.7071497025107373e-06, "loss": 0.12282897233963012, "step": 27530 }, { "epoch": 0.7254455833495224, "grad_norm": 2.034132957458496, "learning_rate": 3.7005216273570623e-06, "loss": 0.09906233549118042, "step": 27540 }, { "epoch": 0.7257089985940212, "grad_norm": 3.1146936416625977, "learning_rate": 3.693898137235532e-06, "loss": 0.11175247430801391, "step": 27550 }, { "epoch": 0.7259724138385198, "grad_norm": 2.619507312774658, "learning_rate": 3.6872792369670074e-06, "loss": 0.12138533592224121, "step": 27560 }, { "epoch": 0.7262358290830186, "grad_norm": 1.6831495761871338, "learning_rate": 3.6806649313690066e-06, "loss": 0.11880645751953126, "step": 27570 }, { "epoch": 0.7264992443275173, "grad_norm": 1.738491177558899, "learning_rate": 3.6740552252557093e-06, "loss": 0.12168548107147217, "step": 27580 }, { "epoch": 0.7267626595720161, "grad_norm": 1.504196047782898, "learning_rate": 3.667450123437939e-06, "loss": 0.11416280269622803, "step": 27590 }, { "epoch": 0.7270260748165148, "grad_norm": 1.553892731666565, "learning_rate": 3.6608496307231823e-06, "loss": 0.1141118049621582, "step": 27600 }, { "epoch": 0.7272894900610135, "grad_norm": 2.412071704864502, "learning_rate": 3.6542537519155617e-06, "loss": 0.09614930152893067, "step": 27610 }, { "epoch": 0.7275529053055123, "grad_norm": 2.307502508163452, "learning_rate": 3.647662491815833e-06, "loss": 0.11986812353134155, "step": 27620 }, { "epoch": 0.727816320550011, "grad_norm": 2.0428600311279297, "learning_rate": 3.641075855221409e-06, "loss": 0.12943905591964722, "step": 27630 }, { "epoch": 0.7280797357945098, "grad_norm": 1.760228157043457, "learning_rate": 3.634493846926321e-06, "loss": 0.12556304931640624, "step": 27640 }, { "epoch": 0.7283431510390085, "grad_norm": 1.862084984779358, "learning_rate": 3.62791647172125e-06, "loss": 0.12642072439193724, "step": 27650 }, { "epoch": 0.7286065662835073, "grad_norm": 3.369708299636841, "learning_rate": 3.621343734393482e-06, "loss": 0.11977866888046265, "step": 27660 }, { "epoch": 0.728869981528006, "grad_norm": 1.8033630847930908, "learning_rate": 3.6147756397269397e-06, "loss": 0.12258311510086059, "step": 27670 }, { "epoch": 0.7291333967725048, "grad_norm": 1.9465311765670776, "learning_rate": 3.6082121925021717e-06, "loss": 0.10919594764709473, "step": 27680 }, { "epoch": 0.7293968120170035, "grad_norm": 1.6121233701705933, "learning_rate": 3.601653397496335e-06, "loss": 0.11094956398010254, "step": 27690 }, { "epoch": 0.7296602272615021, "grad_norm": 1.8580366373062134, "learning_rate": 3.5950992594832044e-06, "loss": 0.12219004631042481, "step": 27700 }, { "epoch": 0.729923642506001, "grad_norm": 1.680714726448059, "learning_rate": 3.588549783233164e-06, "loss": 0.09826242923736572, "step": 27710 }, { "epoch": 0.7301870577504996, "grad_norm": 2.820449113845825, "learning_rate": 3.582004973513201e-06, "loss": 0.11475749015808105, "step": 27720 }, { "epoch": 0.7304504729949984, "grad_norm": 3.2887401580810547, "learning_rate": 3.575464835086917e-06, "loss": 0.12094709873199463, "step": 27730 }, { "epoch": 0.7307138882394971, "grad_norm": 3.9353747367858887, "learning_rate": 3.568929372714508e-06, "loss": 0.13394556045532227, "step": 27740 }, { "epoch": 0.7309773034839959, "grad_norm": 4.158006191253662, "learning_rate": 3.562398591152755e-06, "loss": 0.11635229587554932, "step": 27750 }, { "epoch": 0.7312407187284946, "grad_norm": 1.5722386837005615, "learning_rate": 3.5558724951550515e-06, "loss": 0.11010942459106446, "step": 27760 }, { "epoch": 0.7315041339729933, "grad_norm": 1.4629744291305542, "learning_rate": 3.549351089471367e-06, "loss": 0.10464197397232056, "step": 27770 }, { "epoch": 0.7317675492174921, "grad_norm": 2.6219990253448486, "learning_rate": 3.54283437884827e-06, "loss": 0.1156963586807251, "step": 27780 }, { "epoch": 0.7320309644619908, "grad_norm": 1.8033485412597656, "learning_rate": 3.536322368028895e-06, "loss": 0.11308473348617554, "step": 27790 }, { "epoch": 0.7322943797064896, "grad_norm": 2.539767026901245, "learning_rate": 3.5298150617529635e-06, "loss": 0.11732995510101318, "step": 27800 }, { "epoch": 0.7325577949509883, "grad_norm": 2.686767339706421, "learning_rate": 3.5233124647567817e-06, "loss": 0.13129724264144899, "step": 27810 }, { "epoch": 0.7328212101954871, "grad_norm": 1.9250283241271973, "learning_rate": 3.516814581773216e-06, "loss": 0.11896510124206543, "step": 27820 }, { "epoch": 0.7330846254399858, "grad_norm": 2.689452886581421, "learning_rate": 3.5103214175317056e-06, "loss": 0.11598386764526367, "step": 27830 }, { "epoch": 0.7333480406844846, "grad_norm": 1.7790195941925049, "learning_rate": 3.5038329767582567e-06, "loss": 0.12042083740234374, "step": 27840 }, { "epoch": 0.7336114559289832, "grad_norm": 1.9293131828308105, "learning_rate": 3.4973492641754315e-06, "loss": 0.10935003757476806, "step": 27850 }, { "epoch": 0.7338748711734819, "grad_norm": 2.1586570739746094, "learning_rate": 3.4908702845023643e-06, "loss": 0.10789308547973633, "step": 27860 }, { "epoch": 0.7341382864179807, "grad_norm": 2.270688533782959, "learning_rate": 3.484396042454734e-06, "loss": 0.11904678344726563, "step": 27870 }, { "epoch": 0.7344017016624794, "grad_norm": 2.26582670211792, "learning_rate": 3.477926542744766e-06, "loss": 0.11810382604598998, "step": 27880 }, { "epoch": 0.7346651169069782, "grad_norm": 3.6506950855255127, "learning_rate": 3.4714617900812485e-06, "loss": 0.11792330741882324, "step": 27890 }, { "epoch": 0.7349285321514769, "grad_norm": 3.1305196285247803, "learning_rate": 3.465001789169502e-06, "loss": 0.11963694095611573, "step": 27900 }, { "epoch": 0.7351919473959757, "grad_norm": 2.2437665462493896, "learning_rate": 3.458546544711404e-06, "loss": 0.10657951831817628, "step": 27910 }, { "epoch": 0.7354553626404744, "grad_norm": 3.243807554244995, "learning_rate": 3.45209606140535e-06, "loss": 0.11622375249862671, "step": 27920 }, { "epoch": 0.7357187778849732, "grad_norm": 2.361266851425171, "learning_rate": 3.4456503439462797e-06, "loss": 0.11346714496612549, "step": 27930 }, { "epoch": 0.7359821931294719, "grad_norm": 1.9549546241760254, "learning_rate": 3.4392093970256703e-06, "loss": 0.13111215829849243, "step": 27940 }, { "epoch": 0.7362456083739706, "grad_norm": 1.8540914058685303, "learning_rate": 3.4327732253315195e-06, "loss": 0.12131855487823487, "step": 27950 }, { "epoch": 0.7365090236184694, "grad_norm": 2.150757312774658, "learning_rate": 3.4263418335483477e-06, "loss": 0.11857209205627442, "step": 27960 }, { "epoch": 0.7367724388629681, "grad_norm": 2.1395130157470703, "learning_rate": 3.419915226357203e-06, "loss": 0.11156437397003174, "step": 27970 }, { "epoch": 0.7370358541074669, "grad_norm": 2.1161112785339355, "learning_rate": 3.413493408435642e-06, "loss": 0.12061219215393067, "step": 27980 }, { "epoch": 0.7372992693519655, "grad_norm": 1.9851971864700317, "learning_rate": 3.407076384457747e-06, "loss": 0.11137653589248657, "step": 27990 }, { "epoch": 0.7375626845964643, "grad_norm": 1.6245335340499878, "learning_rate": 3.400664159094106e-06, "loss": 0.09677697420120239, "step": 28000 }, { "epoch": 0.737826099840963, "grad_norm": 2.8780033588409424, "learning_rate": 3.394256737011803e-06, "loss": 0.11905922889709472, "step": 28010 }, { "epoch": 0.7380895150854618, "grad_norm": 2.175405263900757, "learning_rate": 3.3878541228744455e-06, "loss": 0.09696484804153442, "step": 28020 }, { "epoch": 0.7383529303299605, "grad_norm": 2.2076966762542725, "learning_rate": 3.38145632134213e-06, "loss": 0.10750731229782104, "step": 28030 }, { "epoch": 0.7386163455744592, "grad_norm": 2.1678950786590576, "learning_rate": 3.3750633370714527e-06, "loss": 0.11262630224227906, "step": 28040 }, { "epoch": 0.738879760818958, "grad_norm": 1.7162021398544312, "learning_rate": 3.368675174715501e-06, "loss": 0.11604769229888916, "step": 28050 }, { "epoch": 0.7391431760634567, "grad_norm": 2.8968634605407715, "learning_rate": 3.3622918389238544e-06, "loss": 0.11323734521865844, "step": 28060 }, { "epoch": 0.7394065913079555, "grad_norm": 2.1852524280548096, "learning_rate": 3.3559133343425865e-06, "loss": 0.12027927637100219, "step": 28070 }, { "epoch": 0.7396700065524542, "grad_norm": 2.0149433612823486, "learning_rate": 3.3495396656142442e-06, "loss": 0.1109731912612915, "step": 28080 }, { "epoch": 0.739933421796953, "grad_norm": 2.3945257663726807, "learning_rate": 3.343170837377858e-06, "loss": 0.11414506435394287, "step": 28090 }, { "epoch": 0.7401968370414517, "grad_norm": 2.7414145469665527, "learning_rate": 3.336806854268936e-06, "loss": 0.10494086742401124, "step": 28100 }, { "epoch": 0.7404602522859505, "grad_norm": 2.9468727111816406, "learning_rate": 3.3304477209194574e-06, "loss": 0.11993104219436646, "step": 28110 }, { "epoch": 0.7407236675304492, "grad_norm": 2.4926602840423584, "learning_rate": 3.3240934419578828e-06, "loss": 0.11695023775100707, "step": 28120 }, { "epoch": 0.7409870827749478, "grad_norm": 1.477337121963501, "learning_rate": 3.3177440220091207e-06, "loss": 0.11567959785461426, "step": 28130 }, { "epoch": 0.7412504980194466, "grad_norm": 1.8095290660858154, "learning_rate": 3.3113994656945537e-06, "loss": 0.10739989280700683, "step": 28140 }, { "epoch": 0.7415139132639453, "grad_norm": 3.303297519683838, "learning_rate": 3.3050597776320293e-06, "loss": 0.10168728828430176, "step": 28150 }, { "epoch": 0.7417773285084441, "grad_norm": 2.7728984355926514, "learning_rate": 3.298724962435842e-06, "loss": 0.10463587045669556, "step": 28160 }, { "epoch": 0.7420407437529428, "grad_norm": 3.5139336585998535, "learning_rate": 3.292395024716746e-06, "loss": 0.11796506643295288, "step": 28170 }, { "epoch": 0.7423041589974416, "grad_norm": 2.0061488151550293, "learning_rate": 3.2860699690819407e-06, "loss": 0.11067337989807129, "step": 28180 }, { "epoch": 0.7425675742419403, "grad_norm": 2.888375997543335, "learning_rate": 3.2797498001350737e-06, "loss": 0.10838557481765747, "step": 28190 }, { "epoch": 0.7428309894864391, "grad_norm": 2.942737579345703, "learning_rate": 3.27343452247624e-06, "loss": 0.10631660223007203, "step": 28200 }, { "epoch": 0.7430944047309378, "grad_norm": 2.1780624389648438, "learning_rate": 3.267124140701974e-06, "loss": 0.11464296579360962, "step": 28210 }, { "epoch": 0.7433578199754365, "grad_norm": 1.9958412647247314, "learning_rate": 3.2608186594052314e-06, "loss": 0.10507645606994628, "step": 28220 }, { "epoch": 0.7436212352199353, "grad_norm": 2.7290165424346924, "learning_rate": 3.2545180831754264e-06, "loss": 0.11766657829284669, "step": 28230 }, { "epoch": 0.743884650464434, "grad_norm": 2.362501382827759, "learning_rate": 3.248222416598381e-06, "loss": 0.09772932529449463, "step": 28240 }, { "epoch": 0.7441480657089328, "grad_norm": 2.1228156089782715, "learning_rate": 3.2419316642563646e-06, "loss": 0.1053774356842041, "step": 28250 }, { "epoch": 0.7444114809534315, "grad_norm": 2.6000540256500244, "learning_rate": 3.235645830728047e-06, "loss": 0.11478091478347778, "step": 28260 }, { "epoch": 0.7446748961979303, "grad_norm": 1.8644509315490723, "learning_rate": 3.2293649205885293e-06, "loss": 0.11004642248153687, "step": 28270 }, { "epoch": 0.744938311442429, "grad_norm": 1.3974204063415527, "learning_rate": 3.2230889384093355e-06, "loss": 0.09976094365119934, "step": 28280 }, { "epoch": 0.7452017266869276, "grad_norm": 2.8899242877960205, "learning_rate": 3.2168178887583947e-06, "loss": 0.11146472692489624, "step": 28290 }, { "epoch": 0.7454651419314264, "grad_norm": 3.4219894409179688, "learning_rate": 3.210551776200044e-06, "loss": 0.09860923290252685, "step": 28300 }, { "epoch": 0.7457285571759251, "grad_norm": 2.73883056640625, "learning_rate": 3.204290605295033e-06, "loss": 0.1146081566810608, "step": 28310 }, { "epoch": 0.7459919724204239, "grad_norm": 2.716050386428833, "learning_rate": 3.1980343806005077e-06, "loss": 0.109317946434021, "step": 28320 }, { "epoch": 0.7462553876649226, "grad_norm": 2.2993593215942383, "learning_rate": 3.1917831066700255e-06, "loss": 0.1153179407119751, "step": 28330 }, { "epoch": 0.7465188029094214, "grad_norm": 2.368767738342285, "learning_rate": 3.1855367880535336e-06, "loss": 0.11989669799804688, "step": 28340 }, { "epoch": 0.7467822181539201, "grad_norm": 2.705239772796631, "learning_rate": 3.179295429297361e-06, "loss": 0.12392721176147461, "step": 28350 }, { "epoch": 0.7470456333984189, "grad_norm": 2.255113124847412, "learning_rate": 3.173059034944248e-06, "loss": 0.10581635236740113, "step": 28360 }, { "epoch": 0.7473090486429176, "grad_norm": 2.0057194232940674, "learning_rate": 3.1668276095333063e-06, "loss": 0.10697410106658936, "step": 28370 }, { "epoch": 0.7475724638874163, "grad_norm": 1.7072101831436157, "learning_rate": 3.160601157600045e-06, "loss": 0.11555836200714112, "step": 28380 }, { "epoch": 0.7478358791319151, "grad_norm": 2.2226903438568115, "learning_rate": 3.154379683676334e-06, "loss": 0.09963278770446778, "step": 28390 }, { "epoch": 0.7480992943764138, "grad_norm": 2.2591450214385986, "learning_rate": 3.148163192290432e-06, "loss": 0.10532755851745605, "step": 28400 }, { "epoch": 0.7483627096209126, "grad_norm": 3.8206372261047363, "learning_rate": 3.1419516879669744e-06, "loss": 0.11225606203079223, "step": 28410 }, { "epoch": 0.7486261248654112, "grad_norm": 1.7153035402297974, "learning_rate": 3.1357451752269595e-06, "loss": 0.11435861587524414, "step": 28420 }, { "epoch": 0.74888954010991, "grad_norm": 2.352046489715576, "learning_rate": 3.1295436585877558e-06, "loss": 0.12441807985305786, "step": 28430 }, { "epoch": 0.7491529553544087, "grad_norm": 5.7395853996276855, "learning_rate": 3.1233471425630947e-06, "loss": 0.11970760822296142, "step": 28440 }, { "epoch": 0.7494163705989075, "grad_norm": 2.5452980995178223, "learning_rate": 3.1171556316630626e-06, "loss": 0.10263001918792725, "step": 28450 }, { "epoch": 0.7496797858434062, "grad_norm": 2.4333908557891846, "learning_rate": 3.110969130394117e-06, "loss": 0.10292954444885254, "step": 28460 }, { "epoch": 0.7499432010879049, "grad_norm": 2.044245719909668, "learning_rate": 3.1047876432590596e-06, "loss": 0.11816179752349854, "step": 28470 }, { "epoch": 0.7502066163324037, "grad_norm": 2.796865224838257, "learning_rate": 3.0986111747570335e-06, "loss": 0.11739137172698974, "step": 28480 }, { "epoch": 0.7504700315769024, "grad_norm": 2.5703961849212646, "learning_rate": 3.092439729383548e-06, "loss": 0.12064225673675537, "step": 28490 }, { "epoch": 0.7507334468214012, "grad_norm": 2.1784424781799316, "learning_rate": 3.0862733116304435e-06, "loss": 0.1030731439590454, "step": 28500 }, { "epoch": 0.7509968620658999, "grad_norm": 2.704134464263916, "learning_rate": 3.0801119259859115e-06, "loss": 0.11935839653015137, "step": 28510 }, { "epoch": 0.7512602773103987, "grad_norm": 2.764292001724243, "learning_rate": 3.0739555769344666e-06, "loss": 0.09811877012252808, "step": 28520 }, { "epoch": 0.7515236925548974, "grad_norm": 2.356726884841919, "learning_rate": 3.067804268956964e-06, "loss": 0.11932450532913208, "step": 28530 }, { "epoch": 0.7517871077993962, "grad_norm": 2.686866283416748, "learning_rate": 3.0616580065305977e-06, "loss": 0.11156609058380126, "step": 28540 }, { "epoch": 0.7520505230438949, "grad_norm": 1.6383637189865112, "learning_rate": 3.055516794128881e-06, "loss": 0.11283009052276612, "step": 28550 }, { "epoch": 0.7523139382883935, "grad_norm": 3.6679632663726807, "learning_rate": 3.0493806362216516e-06, "loss": 0.10800158977508545, "step": 28560 }, { "epoch": 0.7525773535328923, "grad_norm": 2.3603885173797607, "learning_rate": 3.0432495372750692e-06, "loss": 0.1114461064338684, "step": 28570 }, { "epoch": 0.752840768777391, "grad_norm": 2.1773085594177246, "learning_rate": 3.037123501751611e-06, "loss": 0.10376884937286376, "step": 28580 }, { "epoch": 0.7531041840218898, "grad_norm": 3.583765745162964, "learning_rate": 3.0310025341100757e-06, "loss": 0.11549009084701538, "step": 28590 }, { "epoch": 0.7533675992663885, "grad_norm": 2.1853675842285156, "learning_rate": 3.0248866388055674e-06, "loss": 0.11786285638809205, "step": 28600 }, { "epoch": 0.7536310145108873, "grad_norm": 2.23671817779541, "learning_rate": 3.018775820289488e-06, "loss": 0.09516096711158753, "step": 28610 }, { "epoch": 0.753894429755386, "grad_norm": 2.743872880935669, "learning_rate": 3.0126700830095668e-06, "loss": 0.11164462566375732, "step": 28620 }, { "epoch": 0.7541578449998848, "grad_norm": 2.965693712234497, "learning_rate": 3.006569431409817e-06, "loss": 0.10422165393829345, "step": 28630 }, { "epoch": 0.7544212602443835, "grad_norm": 2.634626865386963, "learning_rate": 3.000473869930558e-06, "loss": 0.10708086490631104, "step": 28640 }, { "epoch": 0.7546846754888822, "grad_norm": 2.045654296875, "learning_rate": 2.9943834030084006e-06, "loss": 0.11768980026245117, "step": 28650 }, { "epoch": 0.754948090733381, "grad_norm": 2.0757508277893066, "learning_rate": 2.988298035076249e-06, "loss": 0.09576982855796815, "step": 28660 }, { "epoch": 0.7552115059778797, "grad_norm": 2.3830440044403076, "learning_rate": 2.9822177705633005e-06, "loss": 0.10797507762908935, "step": 28670 }, { "epoch": 0.7554749212223785, "grad_norm": 2.653588056564331, "learning_rate": 2.976142613895031e-06, "loss": 0.10350757837295532, "step": 28680 }, { "epoch": 0.7557383364668772, "grad_norm": 2.2332828044891357, "learning_rate": 2.970072569493202e-06, "loss": 0.1007194995880127, "step": 28690 }, { "epoch": 0.756001751711376, "grad_norm": 3.5676019191741943, "learning_rate": 2.9640076417758546e-06, "loss": 0.11289653778076172, "step": 28700 }, { "epoch": 0.7562651669558746, "grad_norm": 3.042318820953369, "learning_rate": 2.9579478351573e-06, "loss": 0.10465327501296998, "step": 28710 }, { "epoch": 0.7565285822003734, "grad_norm": 2.6518266201019287, "learning_rate": 2.9518931540481377e-06, "loss": 0.11123971939086914, "step": 28720 }, { "epoch": 0.7567919974448721, "grad_norm": 1.9314626455307007, "learning_rate": 2.945843602855215e-06, "loss": 0.11462053060531616, "step": 28730 }, { "epoch": 0.7570554126893708, "grad_norm": 2.3471288681030273, "learning_rate": 2.9397991859816588e-06, "loss": 0.12008394002914428, "step": 28740 }, { "epoch": 0.7573188279338696, "grad_norm": 2.114701986312866, "learning_rate": 2.9337599078268597e-06, "loss": 0.1111328125, "step": 28750 }, { "epoch": 0.7575822431783683, "grad_norm": 1.9153140783309937, "learning_rate": 2.927725772786464e-06, "loss": 0.11192314624786377, "step": 28760 }, { "epoch": 0.7578456584228671, "grad_norm": 1.408592700958252, "learning_rate": 2.9216967852523747e-06, "loss": 0.10182015895843506, "step": 28770 }, { "epoch": 0.7581090736673658, "grad_norm": 2.8868112564086914, "learning_rate": 2.915672949612749e-06, "loss": 0.10589675903320313, "step": 28780 }, { "epoch": 0.7583724889118646, "grad_norm": 2.0162503719329834, "learning_rate": 2.9096542702519904e-06, "loss": 0.10263439416885375, "step": 28790 }, { "epoch": 0.7586359041563633, "grad_norm": 2.3354508876800537, "learning_rate": 2.9036407515507613e-06, "loss": 0.10826673507690429, "step": 28800 }, { "epoch": 0.758899319400862, "grad_norm": 1.8749221563339233, "learning_rate": 2.89763239788596e-06, "loss": 0.10808737277984619, "step": 28810 }, { "epoch": 0.7591627346453608, "grad_norm": 3.1519722938537598, "learning_rate": 2.8916292136307147e-06, "loss": 0.12254884243011474, "step": 28820 }, { "epoch": 0.7594261498898595, "grad_norm": 1.769697904586792, "learning_rate": 2.885631203154412e-06, "loss": 0.12230651378631592, "step": 28830 }, { "epoch": 0.7596895651343583, "grad_norm": 2.5704383850097656, "learning_rate": 2.879638370822656e-06, "loss": 0.1119236707687378, "step": 28840 }, { "epoch": 0.759952980378857, "grad_norm": 1.9605469703674316, "learning_rate": 2.8736507209972996e-06, "loss": 0.09927954673767089, "step": 28850 }, { "epoch": 0.7602163956233557, "grad_norm": 3.2990856170654297, "learning_rate": 2.867668258036401e-06, "loss": 0.11095004081726074, "step": 28860 }, { "epoch": 0.7604798108678544, "grad_norm": 1.901349663734436, "learning_rate": 2.8616909862942565e-06, "loss": 0.1097725510597229, "step": 28870 }, { "epoch": 0.7607432261123532, "grad_norm": 1.5263487100601196, "learning_rate": 2.855718910121389e-06, "loss": 0.09662339687347413, "step": 28880 }, { "epoch": 0.7610066413568519, "grad_norm": 1.8883674144744873, "learning_rate": 2.8497520338645278e-06, "loss": 0.09294755458831787, "step": 28890 }, { "epoch": 0.7612700566013506, "grad_norm": 2.307509422302246, "learning_rate": 2.843790361866625e-06, "loss": 0.1043461799621582, "step": 28900 }, { "epoch": 0.7615334718458494, "grad_norm": 1.7123709917068481, "learning_rate": 2.837833898466842e-06, "loss": 0.09962539076805114, "step": 28910 }, { "epoch": 0.7617968870903481, "grad_norm": 2.252530336380005, "learning_rate": 2.831882648000547e-06, "loss": 0.10121194124221802, "step": 28920 }, { "epoch": 0.7620603023348469, "grad_norm": 2.7539193630218506, "learning_rate": 2.8259366147993217e-06, "loss": 0.09961608648300171, "step": 28930 }, { "epoch": 0.7623237175793456, "grad_norm": 2.153559923171997, "learning_rate": 2.819995803190948e-06, "loss": 0.11787517070770263, "step": 28940 }, { "epoch": 0.7625871328238444, "grad_norm": 3.663261651992798, "learning_rate": 2.8140602174993927e-06, "loss": 0.10597577095031738, "step": 28950 }, { "epoch": 0.7628505480683431, "grad_norm": 2.2988367080688477, "learning_rate": 2.8081298620448427e-06, "loss": 0.10316389799118042, "step": 28960 }, { "epoch": 0.7631139633128419, "grad_norm": 1.7484097480773926, "learning_rate": 2.8022047411436604e-06, "loss": 0.08858197927474976, "step": 28970 }, { "epoch": 0.7633773785573406, "grad_norm": 2.9995696544647217, "learning_rate": 2.7962848591084122e-06, "loss": 0.11471686363220215, "step": 28980 }, { "epoch": 0.7636407938018392, "grad_norm": 2.759803295135498, "learning_rate": 2.7903702202478355e-06, "loss": 0.10538597106933593, "step": 28990 }, { "epoch": 0.763904209046338, "grad_norm": 4.1938252449035645, "learning_rate": 2.7844608288668586e-06, "loss": 0.09982691407203674, "step": 29000 }, { "epoch": 0.7641676242908367, "grad_norm": 2.2502071857452393, "learning_rate": 2.7785566892665993e-06, "loss": 0.11603376865386963, "step": 29010 }, { "epoch": 0.7644310395353355, "grad_norm": 2.7633495330810547, "learning_rate": 2.77265780574434e-06, "loss": 0.11243966817855836, "step": 29020 }, { "epoch": 0.7646944547798342, "grad_norm": 2.883598804473877, "learning_rate": 2.766764182593544e-06, "loss": 0.11381475925445557, "step": 29030 }, { "epoch": 0.764957870024333, "grad_norm": 2.023949146270752, "learning_rate": 2.760875824103846e-06, "loss": 0.10873709917068482, "step": 29040 }, { "epoch": 0.7652212852688317, "grad_norm": 2.0595881938934326, "learning_rate": 2.7549927345610407e-06, "loss": 0.10038840770721436, "step": 29050 }, { "epoch": 0.7654847005133305, "grad_norm": 2.444664239883423, "learning_rate": 2.749114918247102e-06, "loss": 0.10729103088378907, "step": 29060 }, { "epoch": 0.7657481157578292, "grad_norm": 1.9679092168807983, "learning_rate": 2.7432423794401598e-06, "loss": 0.09562249779701233, "step": 29070 }, { "epoch": 0.7660115310023279, "grad_norm": 2.6365509033203125, "learning_rate": 2.7373751224144905e-06, "loss": 0.1086490273475647, "step": 29080 }, { "epoch": 0.7662749462468267, "grad_norm": 2.161958932876587, "learning_rate": 2.731513151440547e-06, "loss": 0.10515178442001342, "step": 29090 }, { "epoch": 0.7665383614913254, "grad_norm": 1.6600340604782104, "learning_rate": 2.7256564707849165e-06, "loss": 0.10529828071594238, "step": 29100 }, { "epoch": 0.7668017767358242, "grad_norm": 3.2353837490081787, "learning_rate": 2.719805084710356e-06, "loss": 0.11351773738861085, "step": 29110 }, { "epoch": 0.7670651919803229, "grad_norm": 2.108356475830078, "learning_rate": 2.7139589974757454e-06, "loss": 0.09950892925262451, "step": 29120 }, { "epoch": 0.7673286072248217, "grad_norm": 2.3235416412353516, "learning_rate": 2.7081182133361207e-06, "loss": 0.1036847710609436, "step": 29130 }, { "epoch": 0.7675920224693203, "grad_norm": 2.255338191986084, "learning_rate": 2.7022827365426606e-06, "loss": 0.11577515602111817, "step": 29140 }, { "epoch": 0.7678554377138191, "grad_norm": 1.6550941467285156, "learning_rate": 2.696452571342676e-06, "loss": 0.10409971475601196, "step": 29150 }, { "epoch": 0.7681188529583178, "grad_norm": 1.9966447353363037, "learning_rate": 2.6906277219796116e-06, "loss": 0.0921284556388855, "step": 29160 }, { "epoch": 0.7683822682028165, "grad_norm": 2.7844643592834473, "learning_rate": 2.6848081926930447e-06, "loss": 0.09154273271560669, "step": 29170 }, { "epoch": 0.7686456834473153, "grad_norm": 2.284398317337036, "learning_rate": 2.6789939877186753e-06, "loss": 0.11034582853317261, "step": 29180 }, { "epoch": 0.768909098691814, "grad_norm": 2.206732749938965, "learning_rate": 2.67318511128834e-06, "loss": 0.09825051426887513, "step": 29190 }, { "epoch": 0.7691725139363128, "grad_norm": 3.6704635620117188, "learning_rate": 2.66738156762999e-06, "loss": 0.09856427907943725, "step": 29200 }, { "epoch": 0.7694359291808115, "grad_norm": 3.056009292602539, "learning_rate": 2.6615833609676846e-06, "loss": 0.10277936458587647, "step": 29210 }, { "epoch": 0.7696993444253103, "grad_norm": 2.5698564052581787, "learning_rate": 2.6557904955216185e-06, "loss": 0.10990549325942993, "step": 29220 }, { "epoch": 0.769962759669809, "grad_norm": 1.7697595357894897, "learning_rate": 2.650002975508086e-06, "loss": 0.10832103490829467, "step": 29230 }, { "epoch": 0.7702261749143078, "grad_norm": 2.0347344875335693, "learning_rate": 2.6442208051394936e-06, "loss": 0.09921174049377442, "step": 29240 }, { "epoch": 0.7704895901588065, "grad_norm": 2.1822376251220703, "learning_rate": 2.638443988624355e-06, "loss": 0.09582136273384094, "step": 29250 }, { "epoch": 0.7707530054033052, "grad_norm": 2.180994987487793, "learning_rate": 2.632672530167283e-06, "loss": 0.09171658158302307, "step": 29260 }, { "epoch": 0.771016420647804, "grad_norm": 2.0512707233428955, "learning_rate": 2.6269064339689986e-06, "loss": 0.10007542371749878, "step": 29270 }, { "epoch": 0.7712798358923026, "grad_norm": 1.7984670400619507, "learning_rate": 2.6211457042263157e-06, "loss": 0.10929744243621826, "step": 29280 }, { "epoch": 0.7715432511368014, "grad_norm": 1.891234278678894, "learning_rate": 2.6153903451321384e-06, "loss": 0.09915659427642823, "step": 29290 }, { "epoch": 0.7718066663813001, "grad_norm": 2.4172520637512207, "learning_rate": 2.6096403608754685e-06, "loss": 0.0928540825843811, "step": 29300 }, { "epoch": 0.7720700816257989, "grad_norm": 3.5524368286132812, "learning_rate": 2.603895755641387e-06, "loss": 0.11691970825195312, "step": 29310 }, { "epoch": 0.7723334968702976, "grad_norm": 1.6545408964157104, "learning_rate": 2.598156533611077e-06, "loss": 0.10386155843734741, "step": 29320 }, { "epoch": 0.7725969121147963, "grad_norm": 3.357712984085083, "learning_rate": 2.592422698961781e-06, "loss": 0.10098321437835693, "step": 29330 }, { "epoch": 0.7728603273592951, "grad_norm": 2.81974720954895, "learning_rate": 2.586694255866831e-06, "loss": 0.11606758832931519, "step": 29340 }, { "epoch": 0.7731237426037938, "grad_norm": 2.475933074951172, "learning_rate": 2.580971208495642e-06, "loss": 0.09522231221199036, "step": 29350 }, { "epoch": 0.7733871578482926, "grad_norm": 3.061854600906372, "learning_rate": 2.575253561013692e-06, "loss": 0.0978116512298584, "step": 29360 }, { "epoch": 0.7736505730927913, "grad_norm": 2.3163912296295166, "learning_rate": 2.56954131758253e-06, "loss": 0.09138562679290771, "step": 29370 }, { "epoch": 0.7739139883372901, "grad_norm": 3.602783679962158, "learning_rate": 2.5638344823597762e-06, "loss": 0.10845835208892822, "step": 29380 }, { "epoch": 0.7741774035817888, "grad_norm": 2.3402037620544434, "learning_rate": 2.558133059499105e-06, "loss": 0.10891485214233398, "step": 29390 }, { "epoch": 0.7744408188262876, "grad_norm": 2.7686755657196045, "learning_rate": 2.552437053150266e-06, "loss": 0.09881072044372559, "step": 29400 }, { "epoch": 0.7747042340707863, "grad_norm": 2.1414191722869873, "learning_rate": 2.546746467459058e-06, "loss": 0.10429540872573853, "step": 29410 }, { "epoch": 0.774967649315285, "grad_norm": 3.838721990585327, "learning_rate": 2.541061306567325e-06, "loss": 0.11066834926605225, "step": 29420 }, { "epoch": 0.7752310645597837, "grad_norm": 2.400519847869873, "learning_rate": 2.5353815746129827e-06, "loss": 0.10016038417816162, "step": 29430 }, { "epoch": 0.7754944798042824, "grad_norm": 3.100158452987671, "learning_rate": 2.529707275729979e-06, "loss": 0.10752429962158203, "step": 29440 }, { "epoch": 0.7757578950487812, "grad_norm": 3.6025173664093018, "learning_rate": 2.524038414048321e-06, "loss": 0.1096623420715332, "step": 29450 }, { "epoch": 0.7760213102932799, "grad_norm": 2.9812891483306885, "learning_rate": 2.518374993694043e-06, "loss": 0.10200324058532714, "step": 29460 }, { "epoch": 0.7762847255377787, "grad_norm": 2.247539758682251, "learning_rate": 2.5127170187892257e-06, "loss": 0.10217461585998536, "step": 29470 }, { "epoch": 0.7765481407822774, "grad_norm": 2.649362564086914, "learning_rate": 2.507064493451995e-06, "loss": 0.09038604497909546, "step": 29480 }, { "epoch": 0.7768115560267762, "grad_norm": 2.644320011138916, "learning_rate": 2.501417421796498e-06, "loss": 0.1058276653289795, "step": 29490 }, { "epoch": 0.7770749712712749, "grad_norm": 2.60488224029541, "learning_rate": 2.4957758079329173e-06, "loss": 0.10369610786437988, "step": 29500 }, { "epoch": 0.7773383865157736, "grad_norm": 2.4266912937164307, "learning_rate": 2.490139655967465e-06, "loss": 0.10918962955474854, "step": 29510 }, { "epoch": 0.7776018017602724, "grad_norm": 1.909536361694336, "learning_rate": 2.4845089700023694e-06, "loss": 0.10473952293395997, "step": 29520 }, { "epoch": 0.7778652170047711, "grad_norm": 2.282759189605713, "learning_rate": 2.4788837541358957e-06, "loss": 0.10252273082733154, "step": 29530 }, { "epoch": 0.7781286322492699, "grad_norm": 2.9764063358306885, "learning_rate": 2.4732640124623152e-06, "loss": 0.11008028984069824, "step": 29540 }, { "epoch": 0.7783920474937686, "grad_norm": 2.505176067352295, "learning_rate": 2.467649749071913e-06, "loss": 0.10872839689254761, "step": 29550 }, { "epoch": 0.7786554627382674, "grad_norm": 2.2435765266418457, "learning_rate": 2.4620409680509993e-06, "loss": 0.10713464021682739, "step": 29560 }, { "epoch": 0.778918877982766, "grad_norm": 2.2254340648651123, "learning_rate": 2.4564376734818818e-06, "loss": 0.11157847642898559, "step": 29570 }, { "epoch": 0.7791822932272648, "grad_norm": 1.5601329803466797, "learning_rate": 2.450839869442887e-06, "loss": 0.09256991744041443, "step": 29580 }, { "epoch": 0.7794457084717635, "grad_norm": 3.0326662063598633, "learning_rate": 2.4452475600083323e-06, "loss": 0.09439142942428588, "step": 29590 }, { "epoch": 0.7797091237162622, "grad_norm": 1.9556654691696167, "learning_rate": 2.4396607492485393e-06, "loss": 0.10146557092666626, "step": 29600 }, { "epoch": 0.779972538960761, "grad_norm": 2.1354520320892334, "learning_rate": 2.4340794412298364e-06, "loss": 0.11182875633239746, "step": 29610 }, { "epoch": 0.7802359542052597, "grad_norm": 2.4716806411743164, "learning_rate": 2.428503640014539e-06, "loss": 0.09817959666252137, "step": 29620 }, { "epoch": 0.7804993694497585, "grad_norm": 1.9975446462631226, "learning_rate": 2.4229333496609532e-06, "loss": 0.10294058322906494, "step": 29630 }, { "epoch": 0.7807627846942572, "grad_norm": 3.099565267562866, "learning_rate": 2.4173685742233787e-06, "loss": 0.09914603233337402, "step": 29640 }, { "epoch": 0.781026199938756, "grad_norm": 2.7444188594818115, "learning_rate": 2.411809317752095e-06, "loss": 0.09842485785484315, "step": 29650 }, { "epoch": 0.7812896151832547, "grad_norm": 2.2517123222351074, "learning_rate": 2.4062555842933753e-06, "loss": 0.09744746088981629, "step": 29660 }, { "epoch": 0.7815530304277535, "grad_norm": 2.0974700450897217, "learning_rate": 2.4007073778894665e-06, "loss": 0.09923091530799866, "step": 29670 }, { "epoch": 0.7818164456722522, "grad_norm": 2.2137153148651123, "learning_rate": 2.395164702578583e-06, "loss": 0.10213557481765748, "step": 29680 }, { "epoch": 0.7820798609167509, "grad_norm": 4.088891983032227, "learning_rate": 2.3896275623949327e-06, "loss": 0.10735218524932862, "step": 29690 }, { "epoch": 0.7823432761612497, "grad_norm": 2.5747785568237305, "learning_rate": 2.38409596136868e-06, "loss": 0.10654714107513427, "step": 29700 }, { "epoch": 0.7826066914057483, "grad_norm": 2.2222440242767334, "learning_rate": 2.378569903525969e-06, "loss": 0.10899795293807983, "step": 29710 }, { "epoch": 0.7828701066502471, "grad_norm": 3.7405827045440674, "learning_rate": 2.3730493928888963e-06, "loss": 0.1139366626739502, "step": 29720 }, { "epoch": 0.7831335218947458, "grad_norm": 2.291313648223877, "learning_rate": 2.367534433475527e-06, "loss": 0.10048555135726929, "step": 29730 }, { "epoch": 0.7833969371392446, "grad_norm": 2.7869069576263428, "learning_rate": 2.3620250292998925e-06, "loss": 0.10638096332550048, "step": 29740 }, { "epoch": 0.7836603523837433, "grad_norm": 2.1471457481384277, "learning_rate": 2.356521184371973e-06, "loss": 0.10196717977523803, "step": 29750 }, { "epoch": 0.7839237676282421, "grad_norm": 1.9460792541503906, "learning_rate": 2.3510229026977018e-06, "loss": 0.09554510116577149, "step": 29760 }, { "epoch": 0.7841871828727408, "grad_norm": 3.650329113006592, "learning_rate": 2.345530188278967e-06, "loss": 0.1056856632232666, "step": 29770 }, { "epoch": 0.7844505981172395, "grad_norm": 1.904505729675293, "learning_rate": 2.3400430451136e-06, "loss": 0.10303242206573486, "step": 29780 }, { "epoch": 0.7847140133617383, "grad_norm": 3.7206695079803467, "learning_rate": 2.334561477195387e-06, "loss": 0.09055495262145996, "step": 29790 }, { "epoch": 0.784977428606237, "grad_norm": 2.4749603271484375, "learning_rate": 2.329085488514049e-06, "loss": 0.09469698667526245, "step": 29800 }, { "epoch": 0.7852408438507358, "grad_norm": 2.4942212104797363, "learning_rate": 2.3236150830552374e-06, "loss": 0.0982345998287201, "step": 29810 }, { "epoch": 0.7855042590952345, "grad_norm": 3.664398193359375, "learning_rate": 2.3181502648005593e-06, "loss": 0.1042325496673584, "step": 29820 }, { "epoch": 0.7857676743397333, "grad_norm": 2.561300754547119, "learning_rate": 2.3126910377275415e-06, "loss": 0.09419879913330079, "step": 29830 }, { "epoch": 0.786031089584232, "grad_norm": 2.856940269470215, "learning_rate": 2.3072374058096447e-06, "loss": 0.11634151935577393, "step": 29840 }, { "epoch": 0.7862945048287308, "grad_norm": 2.228686571121216, "learning_rate": 2.301789373016259e-06, "loss": 0.09906910061836242, "step": 29850 }, { "epoch": 0.7865579200732294, "grad_norm": 2.2904393672943115, "learning_rate": 2.296346943312693e-06, "loss": 0.09928237795829772, "step": 29860 }, { "epoch": 0.7868213353177281, "grad_norm": 3.535670757293701, "learning_rate": 2.290910120660188e-06, "loss": 0.10761458873748779, "step": 29870 }, { "epoch": 0.7870847505622269, "grad_norm": 2.536269426345825, "learning_rate": 2.285478909015896e-06, "loss": 0.0927540123462677, "step": 29880 }, { "epoch": 0.7873481658067256, "grad_norm": 3.1441924571990967, "learning_rate": 2.2800533123328874e-06, "loss": 0.11588709354400635, "step": 29890 }, { "epoch": 0.7876115810512244, "grad_norm": 2.718297004699707, "learning_rate": 2.2746333345601435e-06, "loss": 0.1026069164276123, "step": 29900 }, { "epoch": 0.7878749962957231, "grad_norm": 2.318330764770508, "learning_rate": 2.269218979642558e-06, "loss": 0.09587740898132324, "step": 29910 }, { "epoch": 0.7881384115402219, "grad_norm": 2.1704354286193848, "learning_rate": 2.2638102515209382e-06, "loss": 0.10703418254852295, "step": 29920 }, { "epoch": 0.7884018267847206, "grad_norm": 3.1871471405029297, "learning_rate": 2.258407154131982e-06, "loss": 0.09458117485046387, "step": 29930 }, { "epoch": 0.7886652420292193, "grad_norm": 3.113297939300537, "learning_rate": 2.2530096914082965e-06, "loss": 0.11429524421691895, "step": 29940 }, { "epoch": 0.7889286572737181, "grad_norm": 2.5023903846740723, "learning_rate": 2.2476178672783943e-06, "loss": 0.09619373679161072, "step": 29950 }, { "epoch": 0.7891920725182168, "grad_norm": 2.4235987663269043, "learning_rate": 2.242231685666675e-06, "loss": 0.10269303321838379, "step": 29960 }, { "epoch": 0.7894554877627156, "grad_norm": 2.884469747543335, "learning_rate": 2.2368511504934333e-06, "loss": 0.1074988842010498, "step": 29970 }, { "epoch": 0.7897189030072143, "grad_norm": 2.0993826389312744, "learning_rate": 2.2314762656748544e-06, "loss": 0.09632360935211182, "step": 29980 }, { "epoch": 0.7899823182517131, "grad_norm": 2.1515653133392334, "learning_rate": 2.2261070351230107e-06, "loss": 0.10460811853408813, "step": 29990 }, { "epoch": 0.7902457334962117, "grad_norm": 2.950430154800415, "learning_rate": 2.2207434627458626e-06, "loss": 0.10714054107666016, "step": 30000 }, { "epoch": 0.7905091487407105, "grad_norm": 2.6495983600616455, "learning_rate": 2.2153855524472513e-06, "loss": 0.11717712879180908, "step": 30010 }, { "epoch": 0.7907725639852092, "grad_norm": 1.595211386680603, "learning_rate": 2.2100333081268853e-06, "loss": 0.0968432068824768, "step": 30020 }, { "epoch": 0.7910359792297079, "grad_norm": 2.5460457801818848, "learning_rate": 2.204686733680369e-06, "loss": 0.10151463747024536, "step": 30030 }, { "epoch": 0.7912993944742067, "grad_norm": 2.6396682262420654, "learning_rate": 2.1993458329991635e-06, "loss": 0.10002212524414063, "step": 30040 }, { "epoch": 0.7915628097187054, "grad_norm": 2.582612991333008, "learning_rate": 2.1940106099706137e-06, "loss": 0.10509301424026489, "step": 30050 }, { "epoch": 0.7918262249632042, "grad_norm": 1.597108006477356, "learning_rate": 2.1886810684779204e-06, "loss": 0.09787994623184204, "step": 30060 }, { "epoch": 0.7920896402077029, "grad_norm": 3.83121657371521, "learning_rate": 2.18335721240015e-06, "loss": 0.10678735971450806, "step": 30070 }, { "epoch": 0.7923530554522017, "grad_norm": 2.930661916732788, "learning_rate": 2.1780390456122424e-06, "loss": 0.09114636778831482, "step": 30080 }, { "epoch": 0.7926164706967004, "grad_norm": 2.488654136657715, "learning_rate": 2.1727265719849853e-06, "loss": 0.1023168683052063, "step": 30090 }, { "epoch": 0.7928798859411992, "grad_norm": 2.307115316390991, "learning_rate": 2.1674197953850274e-06, "loss": 0.09650250673294067, "step": 30100 }, { "epoch": 0.7931433011856979, "grad_norm": 2.3505730628967285, "learning_rate": 2.162118719674867e-06, "loss": 0.10759183168411254, "step": 30110 }, { "epoch": 0.7934067164301966, "grad_norm": 3.039865493774414, "learning_rate": 2.1568233487128556e-06, "loss": 0.09282082319259644, "step": 30120 }, { "epoch": 0.7936701316746954, "grad_norm": 2.780367374420166, "learning_rate": 2.1515336863531976e-06, "loss": 0.10141334533691407, "step": 30130 }, { "epoch": 0.793933546919194, "grad_norm": 3.211916208267212, "learning_rate": 2.1462497364459356e-06, "loss": 0.10169368982315063, "step": 30140 }, { "epoch": 0.7941969621636928, "grad_norm": 2.4988982677459717, "learning_rate": 2.1409715028369504e-06, "loss": 0.09530332684516907, "step": 30150 }, { "epoch": 0.7944603774081915, "grad_norm": 2.790571689605713, "learning_rate": 2.135698989367976e-06, "loss": 0.11098856925964355, "step": 30160 }, { "epoch": 0.7947237926526903, "grad_norm": 3.127795457839966, "learning_rate": 2.1304321998765685e-06, "loss": 0.09779836535453797, "step": 30170 }, { "epoch": 0.794987207897189, "grad_norm": 2.073948860168457, "learning_rate": 2.1251711381961336e-06, "loss": 0.10568143129348755, "step": 30180 }, { "epoch": 0.7952506231416878, "grad_norm": 3.466186046600342, "learning_rate": 2.11991580815589e-06, "loss": 0.10224857330322265, "step": 30190 }, { "epoch": 0.7955140383861865, "grad_norm": 2.5870578289031982, "learning_rate": 2.1146662135808948e-06, "loss": 0.10088796615600586, "step": 30200 }, { "epoch": 0.7957774536306852, "grad_norm": 3.664719581604004, "learning_rate": 2.1094223582920346e-06, "loss": 0.09838424921035767, "step": 30210 }, { "epoch": 0.796040868875184, "grad_norm": 2.8716464042663574, "learning_rate": 2.1041842461060103e-06, "loss": 0.10233678817749023, "step": 30220 }, { "epoch": 0.7963042841196827, "grad_norm": 2.1612603664398193, "learning_rate": 2.098951880835346e-06, "loss": 0.10424715280532837, "step": 30230 }, { "epoch": 0.7965676993641815, "grad_norm": 2.256784677505493, "learning_rate": 2.093725266288382e-06, "loss": 0.10204557180404664, "step": 30240 }, { "epoch": 0.7968311146086802, "grad_norm": 3.4603002071380615, "learning_rate": 2.088504406269274e-06, "loss": 0.11249909400939942, "step": 30250 }, { "epoch": 0.797094529853179, "grad_norm": 3.2352354526519775, "learning_rate": 2.083289304577992e-06, "loss": 0.09968087673187256, "step": 30260 }, { "epoch": 0.7973579450976777, "grad_norm": 2.709477663040161, "learning_rate": 2.0780799650103123e-06, "loss": 0.09353912472724915, "step": 30270 }, { "epoch": 0.7976213603421765, "grad_norm": 3.349909543991089, "learning_rate": 2.0728763913578097e-06, "loss": 0.10021003484725952, "step": 30280 }, { "epoch": 0.7978847755866751, "grad_norm": 2.7528300285339355, "learning_rate": 2.0676785874078773e-06, "loss": 0.0932681679725647, "step": 30290 }, { "epoch": 0.7981481908311738, "grad_norm": 2.1198205947875977, "learning_rate": 2.062486556943699e-06, "loss": 0.10481417179107666, "step": 30300 }, { "epoch": 0.7984116060756726, "grad_norm": 2.536238193511963, "learning_rate": 2.0573003037442575e-06, "loss": 0.10384995937347412, "step": 30310 }, { "epoch": 0.7986750213201713, "grad_norm": 2.572091579437256, "learning_rate": 2.0521198315843315e-06, "loss": 0.10052930116653443, "step": 30320 }, { "epoch": 0.7989384365646701, "grad_norm": 2.4923818111419678, "learning_rate": 2.046945144234491e-06, "loss": 0.10642353296279908, "step": 30330 }, { "epoch": 0.7992018518091688, "grad_norm": 3.3130178451538086, "learning_rate": 2.0417762454611012e-06, "loss": 0.11691091060638428, "step": 30340 }, { "epoch": 0.7994652670536676, "grad_norm": 2.2763097286224365, "learning_rate": 2.0366131390263055e-06, "loss": 0.0938005805015564, "step": 30350 }, { "epoch": 0.7997286822981663, "grad_norm": 2.470073699951172, "learning_rate": 2.0314558286880383e-06, "loss": 0.09769619703292846, "step": 30360 }, { "epoch": 0.7999920975426651, "grad_norm": 3.1806557178497314, "learning_rate": 2.026304318200011e-06, "loss": 0.1108576774597168, "step": 30370 }, { "epoch": 0.8002555127871638, "grad_norm": 2.7527103424072266, "learning_rate": 2.021158611311712e-06, "loss": 0.10291061401367188, "step": 30380 }, { "epoch": 0.8005189280316625, "grad_norm": 2.2240588665008545, "learning_rate": 2.016018711768416e-06, "loss": 0.09928432106971741, "step": 30390 }, { "epoch": 0.8007823432761613, "grad_norm": 2.6307313442230225, "learning_rate": 2.010884623311162e-06, "loss": 0.1056209683418274, "step": 30400 }, { "epoch": 0.80104575852066, "grad_norm": 2.476088285446167, "learning_rate": 2.005756349676753e-06, "loss": 0.10575317144393921, "step": 30410 }, { "epoch": 0.8013091737651588, "grad_norm": 2.1374149322509766, "learning_rate": 2.0006338945977777e-06, "loss": 0.09488512277603149, "step": 30420 }, { "epoch": 0.8015725890096574, "grad_norm": 2.492119312286377, "learning_rate": 1.9955172618025743e-06, "loss": 0.09238582849502563, "step": 30430 }, { "epoch": 0.8018360042541562, "grad_norm": 2.8077552318573, "learning_rate": 1.9904064550152515e-06, "loss": 0.08814902901649475, "step": 30440 }, { "epoch": 0.8020994194986549, "grad_norm": 1.9619741439819336, "learning_rate": 1.9853014779556734e-06, "loss": 0.09457663893699646, "step": 30450 }, { "epoch": 0.8023628347431536, "grad_norm": 2.7252917289733887, "learning_rate": 1.9802023343394606e-06, "loss": 0.09827115535736083, "step": 30460 }, { "epoch": 0.8026262499876524, "grad_norm": 3.013024091720581, "learning_rate": 1.975109027877995e-06, "loss": 0.11206744909286499, "step": 30470 }, { "epoch": 0.8028896652321511, "grad_norm": 2.701106309890747, "learning_rate": 1.9700215622784026e-06, "loss": 0.07936249375343322, "step": 30480 }, { "epoch": 0.8031530804766499, "grad_norm": 4.32535457611084, "learning_rate": 1.96493994124356e-06, "loss": 0.09822758436203002, "step": 30490 }, { "epoch": 0.8034164957211486, "grad_norm": 2.8034284114837646, "learning_rate": 1.9598641684720897e-06, "loss": 0.09236825704574585, "step": 30500 }, { "epoch": 0.8036799109656474, "grad_norm": 1.9240646362304688, "learning_rate": 1.954794247658356e-06, "loss": 0.10824140310287475, "step": 30510 }, { "epoch": 0.8039433262101461, "grad_norm": 2.4001522064208984, "learning_rate": 1.949730182492473e-06, "loss": 0.0923038363456726, "step": 30520 }, { "epoch": 0.8042067414546449, "grad_norm": 2.988293170928955, "learning_rate": 1.9446719766602793e-06, "loss": 0.09458569884300232, "step": 30530 }, { "epoch": 0.8044701566991436, "grad_norm": 3.7206480503082275, "learning_rate": 1.9396196338433547e-06, "loss": 0.1024365782737732, "step": 30540 }, { "epoch": 0.8047335719436423, "grad_norm": 2.5054523944854736, "learning_rate": 1.9345731577190187e-06, "loss": 0.09325479865074157, "step": 30550 }, { "epoch": 0.804996987188141, "grad_norm": 2.686267375946045, "learning_rate": 1.929532551960309e-06, "loss": 0.1010021448135376, "step": 30560 }, { "epoch": 0.8052604024326397, "grad_norm": 3.034846782684326, "learning_rate": 1.9244978202359974e-06, "loss": 0.08057697415351868, "step": 30570 }, { "epoch": 0.8055238176771385, "grad_norm": 1.956581473350525, "learning_rate": 1.9194689662105803e-06, "loss": 0.0910660207271576, "step": 30580 }, { "epoch": 0.8057872329216372, "grad_norm": 2.467956304550171, "learning_rate": 1.9144459935442705e-06, "loss": 0.10367544889450073, "step": 30590 }, { "epoch": 0.806050648166136, "grad_norm": 2.387284278869629, "learning_rate": 1.9094289058930095e-06, "loss": 0.09152739644050598, "step": 30600 }, { "epoch": 0.8063140634106347, "grad_norm": 2.4548428058624268, "learning_rate": 1.9044177069084502e-06, "loss": 0.09042552709579468, "step": 30610 }, { "epoch": 0.8065774786551335, "grad_norm": 2.7395477294921875, "learning_rate": 1.899412400237951e-06, "loss": 0.10116773843765259, "step": 30620 }, { "epoch": 0.8068408938996322, "grad_norm": 2.342313766479492, "learning_rate": 1.8944129895246e-06, "loss": 0.1036409616470337, "step": 30630 }, { "epoch": 0.8071043091441309, "grad_norm": 2.6218295097351074, "learning_rate": 1.8894194784071761e-06, "loss": 0.0989854097366333, "step": 30640 }, { "epoch": 0.8073677243886297, "grad_norm": 3.0676138401031494, "learning_rate": 1.8844318705201815e-06, "loss": 0.10716586112976074, "step": 30650 }, { "epoch": 0.8076311396331284, "grad_norm": 2.7637510299682617, "learning_rate": 1.879450169493805e-06, "loss": 0.10426152944564819, "step": 30660 }, { "epoch": 0.8078945548776272, "grad_norm": 1.9283548593521118, "learning_rate": 1.874474378953942e-06, "loss": 0.10641329288482666, "step": 30670 }, { "epoch": 0.8081579701221259, "grad_norm": 4.071557521820068, "learning_rate": 1.8695045025221936e-06, "loss": 0.08916231393814086, "step": 30680 }, { "epoch": 0.8084213853666247, "grad_norm": 2.245530605316162, "learning_rate": 1.864540543815848e-06, "loss": 0.0893309772014618, "step": 30690 }, { "epoch": 0.8086848006111234, "grad_norm": 3.9504923820495605, "learning_rate": 1.859582506447889e-06, "loss": 0.1015709638595581, "step": 30700 }, { "epoch": 0.8089482158556222, "grad_norm": 2.6874101161956787, "learning_rate": 1.8546303940269883e-06, "loss": 0.09994968175888061, "step": 30710 }, { "epoch": 0.8092116311001208, "grad_norm": 2.957608222961426, "learning_rate": 1.8496842101575053e-06, "loss": 0.0977719783782959, "step": 30720 }, { "epoch": 0.8094750463446195, "grad_norm": 3.3895134925842285, "learning_rate": 1.844743958439491e-06, "loss": 0.10587010383605958, "step": 30730 }, { "epoch": 0.8097384615891183, "grad_norm": 2.6271913051605225, "learning_rate": 1.839809642468674e-06, "loss": 0.09324031472206115, "step": 30740 }, { "epoch": 0.810001876833617, "grad_norm": 2.015333890914917, "learning_rate": 1.8348812658364557e-06, "loss": 0.10969742536544799, "step": 30750 }, { "epoch": 0.8102652920781158, "grad_norm": 2.040105104446411, "learning_rate": 1.8299588321299266e-06, "loss": 0.09074721336364747, "step": 30760 }, { "epoch": 0.8105287073226145, "grad_norm": 2.4194700717926025, "learning_rate": 1.8250423449318434e-06, "loss": 0.10139127969741821, "step": 30770 }, { "epoch": 0.8107921225671133, "grad_norm": 2.5803046226501465, "learning_rate": 1.8201318078206443e-06, "loss": 0.10258971452713013, "step": 30780 }, { "epoch": 0.811055537811612, "grad_norm": 2.5577874183654785, "learning_rate": 1.8152272243704216e-06, "loss": 0.11165409088134766, "step": 30790 }, { "epoch": 0.8113189530561108, "grad_norm": 2.7143495082855225, "learning_rate": 1.8103285981509432e-06, "loss": 0.1002543330192566, "step": 30800 }, { "epoch": 0.8115823683006095, "grad_norm": 2.577420711517334, "learning_rate": 1.8054359327276473e-06, "loss": 0.09292222261428833, "step": 30810 }, { "epoch": 0.8118457835451082, "grad_norm": 2.4340879917144775, "learning_rate": 1.8005492316616224e-06, "loss": 0.10226681232452392, "step": 30820 }, { "epoch": 0.812109198789607, "grad_norm": 1.8867952823638916, "learning_rate": 1.7956684985096217e-06, "loss": 0.08606173992156982, "step": 30830 }, { "epoch": 0.8123726140341057, "grad_norm": 2.797616720199585, "learning_rate": 1.7907937368240525e-06, "loss": 0.09132012128829955, "step": 30840 }, { "epoch": 0.8126360292786045, "grad_norm": 3.3253660202026367, "learning_rate": 1.7859249501529742e-06, "loss": 0.10803258419036865, "step": 30850 }, { "epoch": 0.8128994445231031, "grad_norm": 3.204704761505127, "learning_rate": 1.7810621420401054e-06, "loss": 0.09735462665557862, "step": 30860 }, { "epoch": 0.8131628597676019, "grad_norm": 2.5410773754119873, "learning_rate": 1.776205316024806e-06, "loss": 0.09154222011566163, "step": 30870 }, { "epoch": 0.8134262750121006, "grad_norm": 2.9023163318634033, "learning_rate": 1.7713544756420785e-06, "loss": 0.08947622179985046, "step": 30880 }, { "epoch": 0.8136896902565994, "grad_norm": 2.9054648876190186, "learning_rate": 1.76650962442258e-06, "loss": 0.09189578890800476, "step": 30890 }, { "epoch": 0.8139531055010981, "grad_norm": 3.0281436443328857, "learning_rate": 1.761670765892599e-06, "loss": 0.09854156374931336, "step": 30900 }, { "epoch": 0.8142165207455968, "grad_norm": 3.053354263305664, "learning_rate": 1.7568379035740669e-06, "loss": 0.10368887186050416, "step": 30910 }, { "epoch": 0.8144799359900956, "grad_norm": 2.8746678829193115, "learning_rate": 1.7520110409845493e-06, "loss": 0.09154638051986694, "step": 30920 }, { "epoch": 0.8147433512345943, "grad_norm": 3.7915894985198975, "learning_rate": 1.7471901816372416e-06, "loss": 0.09269033670425415, "step": 30930 }, { "epoch": 0.8150067664790931, "grad_norm": 2.911078691482544, "learning_rate": 1.7423753290409784e-06, "loss": 0.10281584262847901, "step": 30940 }, { "epoch": 0.8152701817235918, "grad_norm": 2.9215481281280518, "learning_rate": 1.7375664867002151e-06, "loss": 0.0930462658405304, "step": 30950 }, { "epoch": 0.8155335969680906, "grad_norm": 1.9292019605636597, "learning_rate": 1.7327636581150343e-06, "loss": 0.09873565435409545, "step": 30960 }, { "epoch": 0.8157970122125893, "grad_norm": 3.184523105621338, "learning_rate": 1.7279668467811427e-06, "loss": 0.0959493339061737, "step": 30970 }, { "epoch": 0.816060427457088, "grad_norm": 2.4993553161621094, "learning_rate": 1.7231760561898625e-06, "loss": 0.09481889605522156, "step": 30980 }, { "epoch": 0.8163238427015868, "grad_norm": 2.6109414100646973, "learning_rate": 1.718391289828144e-06, "loss": 0.09090780615806579, "step": 30990 }, { "epoch": 0.8165872579460854, "grad_norm": 2.539360284805298, "learning_rate": 1.7136125511785462e-06, "loss": 0.08940721750259399, "step": 31000 }, { "epoch": 0.8168506731905842, "grad_norm": 2.526603937149048, "learning_rate": 1.7088398437192344e-06, "loss": 0.10360062122344971, "step": 31010 }, { "epoch": 0.8171140884350829, "grad_norm": 2.239875316619873, "learning_rate": 1.7040731709239967e-06, "loss": 0.09201693534851074, "step": 31020 }, { "epoch": 0.8173775036795817, "grad_norm": 3.10837984085083, "learning_rate": 1.6993125362622231e-06, "loss": 0.09629506468772889, "step": 31030 }, { "epoch": 0.8176409189240804, "grad_norm": 2.4178402423858643, "learning_rate": 1.6945579431989067e-06, "loss": 0.09446513652801514, "step": 31040 }, { "epoch": 0.8179043341685792, "grad_norm": 1.8361387252807617, "learning_rate": 1.689809395194647e-06, "loss": 0.08634142875671387, "step": 31050 }, { "epoch": 0.8181677494130779, "grad_norm": 4.253039360046387, "learning_rate": 1.6850668957056393e-06, "loss": 0.10642150640487671, "step": 31060 }, { "epoch": 0.8184311646575766, "grad_norm": 2.925239086151123, "learning_rate": 1.6803304481836835e-06, "loss": 0.09691294431686401, "step": 31070 }, { "epoch": 0.8186945799020754, "grad_norm": 2.7793922424316406, "learning_rate": 1.6756000560761687e-06, "loss": 0.08860089778900146, "step": 31080 }, { "epoch": 0.8189579951465741, "grad_norm": 2.327289342880249, "learning_rate": 1.6708757228260786e-06, "loss": 0.09721096158027649, "step": 31090 }, { "epoch": 0.8192214103910729, "grad_norm": 2.166386127471924, "learning_rate": 1.6661574518719859e-06, "loss": 0.0860556721687317, "step": 31100 }, { "epoch": 0.8194848256355716, "grad_norm": 2.882319211959839, "learning_rate": 1.6614452466480502e-06, "loss": 0.08990329504013062, "step": 31110 }, { "epoch": 0.8197482408800704, "grad_norm": 2.944000244140625, "learning_rate": 1.6567391105840247e-06, "loss": 0.09298897981643676, "step": 31120 }, { "epoch": 0.820011656124569, "grad_norm": 3.0493788719177246, "learning_rate": 1.6520390471052306e-06, "loss": 0.09311634302139282, "step": 31130 }, { "epoch": 0.8202750713690679, "grad_norm": 2.1942760944366455, "learning_rate": 1.647345059632578e-06, "loss": 0.08998218774795533, "step": 31140 }, { "epoch": 0.8205384866135665, "grad_norm": 2.7829575538635254, "learning_rate": 1.6426571515825563e-06, "loss": 0.09551002383232117, "step": 31150 }, { "epoch": 0.8208019018580652, "grad_norm": 3.0136125087738037, "learning_rate": 1.6379753263672272e-06, "loss": 0.09090236425399781, "step": 31160 }, { "epoch": 0.821065317102564, "grad_norm": 2.572028875350952, "learning_rate": 1.6332995873942237e-06, "loss": 0.08248096704483032, "step": 31170 }, { "epoch": 0.8213287323470627, "grad_norm": 3.04836368560791, "learning_rate": 1.6286299380667504e-06, "loss": 0.09463915824890137, "step": 31180 }, { "epoch": 0.8215921475915615, "grad_norm": 3.1784627437591553, "learning_rate": 1.6239663817835783e-06, "loss": 0.10013211965560913, "step": 31190 }, { "epoch": 0.8218555628360602, "grad_norm": 2.858044147491455, "learning_rate": 1.6193089219390478e-06, "loss": 0.09708229303359986, "step": 31200 }, { "epoch": 0.822118978080559, "grad_norm": 2.1299996376037598, "learning_rate": 1.6146575619230597e-06, "loss": 0.09990543127059937, "step": 31210 }, { "epoch": 0.8223823933250577, "grad_norm": 2.4338815212249756, "learning_rate": 1.610012305121068e-06, "loss": 0.08626278638839721, "step": 31220 }, { "epoch": 0.8226458085695565, "grad_norm": 2.646470785140991, "learning_rate": 1.6053731549140982e-06, "loss": 0.10160280466079712, "step": 31230 }, { "epoch": 0.8229092238140552, "grad_norm": 2.785996198654175, "learning_rate": 1.6007401146787195e-06, "loss": 0.09772066473960876, "step": 31240 }, { "epoch": 0.8231726390585539, "grad_norm": 2.273759126663208, "learning_rate": 1.596113187787064e-06, "loss": 0.09062532186508179, "step": 31250 }, { "epoch": 0.8234360543030527, "grad_norm": 3.053584575653076, "learning_rate": 1.5914923776068025e-06, "loss": 0.09898535013198853, "step": 31260 }, { "epoch": 0.8236994695475514, "grad_norm": 2.993293046951294, "learning_rate": 1.5868776875011604e-06, "loss": 0.08629781007766724, "step": 31270 }, { "epoch": 0.8239628847920502, "grad_norm": 3.9406497478485107, "learning_rate": 1.582269120828912e-06, "loss": 0.10609774589538574, "step": 31280 }, { "epoch": 0.8242263000365488, "grad_norm": 2.3454458713531494, "learning_rate": 1.5776666809443697e-06, "loss": 0.087696772813797, "step": 31290 }, { "epoch": 0.8244897152810476, "grad_norm": 2.5838844776153564, "learning_rate": 1.5730703711973883e-06, "loss": 0.09014293551445007, "step": 31300 }, { "epoch": 0.8247531305255463, "grad_norm": 2.205440044403076, "learning_rate": 1.5684801949333584e-06, "loss": 0.09577813744544983, "step": 31310 }, { "epoch": 0.8250165457700451, "grad_norm": 2.518202066421509, "learning_rate": 1.5638961554932064e-06, "loss": 0.1030962347984314, "step": 31320 }, { "epoch": 0.8252799610145438, "grad_norm": 3.3504865169525146, "learning_rate": 1.5593182562134002e-06, "loss": 0.10023380517959594, "step": 31330 }, { "epoch": 0.8255433762590425, "grad_norm": 3.0419411659240723, "learning_rate": 1.554746500425931e-06, "loss": 0.0979425072669983, "step": 31340 }, { "epoch": 0.8258067915035413, "grad_norm": 1.8139435052871704, "learning_rate": 1.5501808914583139e-06, "loss": 0.08379030227661133, "step": 31350 }, { "epoch": 0.82607020674804, "grad_norm": 3.1180577278137207, "learning_rate": 1.5456214326336017e-06, "loss": 0.09169161319732666, "step": 31360 }, { "epoch": 0.8263336219925388, "grad_norm": 2.0802595615386963, "learning_rate": 1.5410681272703643e-06, "loss": 0.08745007514953614, "step": 31370 }, { "epoch": 0.8265970372370375, "grad_norm": 2.667269706726074, "learning_rate": 1.536520978682693e-06, "loss": 0.08899636268615722, "step": 31380 }, { "epoch": 0.8268604524815363, "grad_norm": 2.1716737747192383, "learning_rate": 1.531979990180199e-06, "loss": 0.08771735429763794, "step": 31390 }, { "epoch": 0.827123867726035, "grad_norm": 2.9558234214782715, "learning_rate": 1.5274451650680066e-06, "loss": 0.10723339319229126, "step": 31400 }, { "epoch": 0.8273872829705338, "grad_norm": 2.8216021060943604, "learning_rate": 1.5229165066467632e-06, "loss": 0.09533407688140869, "step": 31410 }, { "epoch": 0.8276506982150325, "grad_norm": 2.10180401802063, "learning_rate": 1.5183940182126189e-06, "loss": 0.08372776508331299, "step": 31420 }, { "epoch": 0.8279141134595311, "grad_norm": 2.8334412574768066, "learning_rate": 1.513877703057235e-06, "loss": 0.09754269123077393, "step": 31430 }, { "epoch": 0.8281775287040299, "grad_norm": 2.703045129776001, "learning_rate": 1.509367564467782e-06, "loss": 0.08811682462692261, "step": 31440 }, { "epoch": 0.8284409439485286, "grad_norm": 2.9357659816741943, "learning_rate": 1.5048636057269294e-06, "loss": 0.09812031388282776, "step": 31450 }, { "epoch": 0.8287043591930274, "grad_norm": 2.4847307205200195, "learning_rate": 1.50036583011286e-06, "loss": 0.0937902569770813, "step": 31460 }, { "epoch": 0.8289677744375261, "grad_norm": 2.663884401321411, "learning_rate": 1.4958742408992454e-06, "loss": 0.09382380843162537, "step": 31470 }, { "epoch": 0.8292311896820249, "grad_norm": 2.4058611392974854, "learning_rate": 1.4913888413552536e-06, "loss": 0.09245985150337219, "step": 31480 }, { "epoch": 0.8294946049265236, "grad_norm": 3.0190048217773438, "learning_rate": 1.486909634745557e-06, "loss": 0.09063085317611694, "step": 31490 }, { "epoch": 0.8297580201710223, "grad_norm": 2.3740084171295166, "learning_rate": 1.4824366243303156e-06, "loss": 0.09309451580047608, "step": 31500 }, { "epoch": 0.8300214354155211, "grad_norm": 2.3305819034576416, "learning_rate": 1.4779698133651766e-06, "loss": 0.09240049123764038, "step": 31510 }, { "epoch": 0.8302848506600198, "grad_norm": 3.0213959217071533, "learning_rate": 1.47350920510128e-06, "loss": 0.09630792140960694, "step": 31520 }, { "epoch": 0.8305482659045186, "grad_norm": 2.757730722427368, "learning_rate": 1.469054802785246e-06, "loss": 0.09781612157821655, "step": 31530 }, { "epoch": 0.8308116811490173, "grad_norm": 3.667731761932373, "learning_rate": 1.4646066096591838e-06, "loss": 0.10265524387359619, "step": 31540 }, { "epoch": 0.8310750963935161, "grad_norm": 2.3839268684387207, "learning_rate": 1.4601646289606809e-06, "loss": 0.09289059638977051, "step": 31550 }, { "epoch": 0.8313385116380148, "grad_norm": 2.8694636821746826, "learning_rate": 1.4557288639227995e-06, "loss": 0.08062921762466431, "step": 31560 }, { "epoch": 0.8316019268825136, "grad_norm": 2.7592809200286865, "learning_rate": 1.451299317774083e-06, "loss": 0.0863462209701538, "step": 31570 }, { "epoch": 0.8318653421270122, "grad_norm": 2.8989663124084473, "learning_rate": 1.4468759937385423e-06, "loss": 0.08256853818893432, "step": 31580 }, { "epoch": 0.8321287573715109, "grad_norm": 2.6955912113189697, "learning_rate": 1.4424588950356688e-06, "loss": 0.08909855484962463, "step": 31590 }, { "epoch": 0.8323921726160097, "grad_norm": 2.287877082824707, "learning_rate": 1.4380480248804185e-06, "loss": 0.10789285898208618, "step": 31600 }, { "epoch": 0.8326555878605084, "grad_norm": 3.1645419597625732, "learning_rate": 1.4336433864832033e-06, "loss": 0.09034264087677002, "step": 31610 }, { "epoch": 0.8329190031050072, "grad_norm": 3.28701114654541, "learning_rate": 1.4292449830499189e-06, "loss": 0.09236731529235839, "step": 31620 }, { "epoch": 0.8331824183495059, "grad_norm": 2.6041207313537598, "learning_rate": 1.4248528177819098e-06, "loss": 0.08599579334259033, "step": 31630 }, { "epoch": 0.8334458335940047, "grad_norm": 4.465145111083984, "learning_rate": 1.4204668938759835e-06, "loss": 0.08864498138427734, "step": 31640 }, { "epoch": 0.8337092488385034, "grad_norm": 3.1196537017822266, "learning_rate": 1.416087214524403e-06, "loss": 0.09463412165641785, "step": 31650 }, { "epoch": 0.8339726640830022, "grad_norm": 2.4571783542633057, "learning_rate": 1.4117137829148885e-06, "loss": 0.09267030954360962, "step": 31660 }, { "epoch": 0.8342360793275009, "grad_norm": 2.1699140071868896, "learning_rate": 1.4073466022306148e-06, "loss": 0.09213330745697021, "step": 31670 }, { "epoch": 0.8344994945719996, "grad_norm": 2.8548202514648438, "learning_rate": 1.4029856756502026e-06, "loss": 0.08906298875808716, "step": 31680 }, { "epoch": 0.8347629098164984, "grad_norm": 2.7775137424468994, "learning_rate": 1.3986310063477215e-06, "loss": 0.1003427505493164, "step": 31690 }, { "epoch": 0.835026325060997, "grad_norm": 2.9104933738708496, "learning_rate": 1.3942825974926888e-06, "loss": 0.09478731155395508, "step": 31700 }, { "epoch": 0.8352897403054959, "grad_norm": 2.8120973110198975, "learning_rate": 1.3899404522500626e-06, "loss": 0.08821655511856079, "step": 31710 }, { "epoch": 0.8355531555499945, "grad_norm": 2.5403881072998047, "learning_rate": 1.385604573780248e-06, "loss": 0.080320805311203, "step": 31720 }, { "epoch": 0.8358165707944933, "grad_norm": 2.508396625518799, "learning_rate": 1.38127496523908e-06, "loss": 0.10328054428100586, "step": 31730 }, { "epoch": 0.836079986038992, "grad_norm": 2.507641553878784, "learning_rate": 1.376951629777833e-06, "loss": 0.0903585433959961, "step": 31740 }, { "epoch": 0.8363434012834908, "grad_norm": 1.9856983423233032, "learning_rate": 1.372634570543222e-06, "loss": 0.09504212141036987, "step": 31750 }, { "epoch": 0.8366068165279895, "grad_norm": 2.1139702796936035, "learning_rate": 1.3683237906773873e-06, "loss": 0.09193559885025024, "step": 31760 }, { "epoch": 0.8368702317724882, "grad_norm": 2.883849859237671, "learning_rate": 1.3640192933178997e-06, "loss": 0.09337139129638672, "step": 31770 }, { "epoch": 0.837133647016987, "grad_norm": 2.609987258911133, "learning_rate": 1.359721081597758e-06, "loss": 0.09937549233436585, "step": 31780 }, { "epoch": 0.8373970622614857, "grad_norm": 2.398963689804077, "learning_rate": 1.355429158645384e-06, "loss": 0.08591742515563965, "step": 31790 }, { "epoch": 0.8376604775059845, "grad_norm": 2.6143550872802734, "learning_rate": 1.3511435275846297e-06, "loss": 0.10097318887710571, "step": 31800 }, { "epoch": 0.8379238927504832, "grad_norm": 3.8716416358947754, "learning_rate": 1.3468641915347613e-06, "loss": 0.0977815866470337, "step": 31810 }, { "epoch": 0.838187307994982, "grad_norm": 3.202150344848633, "learning_rate": 1.3425911536104573e-06, "loss": 0.087603360414505, "step": 31820 }, { "epoch": 0.8384507232394807, "grad_norm": 2.26426100730896, "learning_rate": 1.3383244169218268e-06, "loss": 0.0811897337436676, "step": 31830 }, { "epoch": 0.8387141384839795, "grad_norm": 3.231029510498047, "learning_rate": 1.3340639845743797e-06, "loss": 0.09923813343048096, "step": 31840 }, { "epoch": 0.8389775537284782, "grad_norm": 4.05213737487793, "learning_rate": 1.32980985966905e-06, "loss": 0.10015764236450195, "step": 31850 }, { "epoch": 0.8392409689729768, "grad_norm": 3.174537420272827, "learning_rate": 1.3255620453021656e-06, "loss": 0.10488483905792237, "step": 31860 }, { "epoch": 0.8395043842174756, "grad_norm": 3.330568552017212, "learning_rate": 1.3213205445654688e-06, "loss": 0.09040071964263915, "step": 31870 }, { "epoch": 0.8397677994619743, "grad_norm": 2.6492648124694824, "learning_rate": 1.3170853605461132e-06, "loss": 0.09081581830978394, "step": 31880 }, { "epoch": 0.8400312147064731, "grad_norm": 2.995795488357544, "learning_rate": 1.3128564963266444e-06, "loss": 0.0970467209815979, "step": 31890 }, { "epoch": 0.8402946299509718, "grad_norm": 3.9345998764038086, "learning_rate": 1.3086339549850125e-06, "loss": 0.09780036211013794, "step": 31900 }, { "epoch": 0.8405580451954706, "grad_norm": 3.0761139392852783, "learning_rate": 1.3044177395945657e-06, "loss": 0.0860569417476654, "step": 31910 }, { "epoch": 0.8408214604399693, "grad_norm": 2.354823589324951, "learning_rate": 1.3002078532240447e-06, "loss": 0.103157377243042, "step": 31920 }, { "epoch": 0.8410848756844681, "grad_norm": 2.739769458770752, "learning_rate": 1.296004298937592e-06, "loss": 0.09106968641281128, "step": 31930 }, { "epoch": 0.8413482909289668, "grad_norm": 2.8682022094726562, "learning_rate": 1.2918070797947346e-06, "loss": 0.07297152280807495, "step": 31940 }, { "epoch": 0.8416117061734655, "grad_norm": 2.441664457321167, "learning_rate": 1.2876161988503843e-06, "loss": 0.09800957441329956, "step": 31950 }, { "epoch": 0.8418751214179643, "grad_norm": 2.825857400894165, "learning_rate": 1.2834316591548502e-06, "loss": 0.09286340475082397, "step": 31960 }, { "epoch": 0.842138536662463, "grad_norm": 2.8484673500061035, "learning_rate": 1.279253463753819e-06, "loss": 0.09213622212409973, "step": 31970 }, { "epoch": 0.8424019519069618, "grad_norm": 3.0931968688964844, "learning_rate": 1.2750816156883627e-06, "loss": 0.08948628306388855, "step": 31980 }, { "epoch": 0.8426653671514605, "grad_norm": 2.5542192459106445, "learning_rate": 1.2709161179949303e-06, "loss": 0.08488245010375976, "step": 31990 }, { "epoch": 0.8429287823959593, "grad_norm": 2.965385913848877, "learning_rate": 1.266756973705351e-06, "loss": 0.09720894694328308, "step": 32000 }, { "epoch": 0.8431921976404579, "grad_norm": 2.6722729206085205, "learning_rate": 1.2626041858468318e-06, "loss": 0.09223483204841613, "step": 32010 }, { "epoch": 0.8434556128849566, "grad_norm": 3.058316707611084, "learning_rate": 1.2584577574419499e-06, "loss": 0.08763957023620605, "step": 32020 }, { "epoch": 0.8437190281294554, "grad_norm": 3.1932520866394043, "learning_rate": 1.2543176915086542e-06, "loss": 0.10584629774093628, "step": 32030 }, { "epoch": 0.8439824433739541, "grad_norm": 3.2959299087524414, "learning_rate": 1.2501839910602632e-06, "loss": 0.08594993948936462, "step": 32040 }, { "epoch": 0.8442458586184529, "grad_norm": 2.9462761878967285, "learning_rate": 1.24605665910546e-06, "loss": 0.09892609715461731, "step": 32050 }, { "epoch": 0.8445092738629516, "grad_norm": 3.233171224594116, "learning_rate": 1.2419356986483e-06, "loss": 0.09488378763198853, "step": 32060 }, { "epoch": 0.8447726891074504, "grad_norm": 2.867380142211914, "learning_rate": 1.2378211126881956e-06, "loss": 0.09011580944061279, "step": 32070 }, { "epoch": 0.8450361043519491, "grad_norm": 3.164931297302246, "learning_rate": 1.2337129042199147e-06, "loss": 0.08624527454376221, "step": 32080 }, { "epoch": 0.8452995195964479, "grad_norm": 2.3789799213409424, "learning_rate": 1.2296110762335933e-06, "loss": 0.08558144569396972, "step": 32090 }, { "epoch": 0.8455629348409466, "grad_norm": 2.903695821762085, "learning_rate": 1.2255156317147165e-06, "loss": 0.0843706727027893, "step": 32100 }, { "epoch": 0.8458263500854453, "grad_norm": 2.4823875427246094, "learning_rate": 1.2214265736441277e-06, "loss": 0.09174721240997315, "step": 32110 }, { "epoch": 0.8460897653299441, "grad_norm": 3.0608506202697754, "learning_rate": 1.2173439049980174e-06, "loss": 0.08922343254089356, "step": 32120 }, { "epoch": 0.8463531805744428, "grad_norm": 3.4762067794799805, "learning_rate": 1.2132676287479273e-06, "loss": 0.08573474287986756, "step": 32130 }, { "epoch": 0.8466165958189416, "grad_norm": 3.2869739532470703, "learning_rate": 1.209197747860752e-06, "loss": 0.08689513802528381, "step": 32140 }, { "epoch": 0.8468800110634402, "grad_norm": 2.2017078399658203, "learning_rate": 1.2051342652987218e-06, "loss": 0.09234862327575684, "step": 32150 }, { "epoch": 0.847143426307939, "grad_norm": 3.2745046615600586, "learning_rate": 1.2010771840194168e-06, "loss": 0.08928439617156983, "step": 32160 }, { "epoch": 0.8474068415524377, "grad_norm": 2.6998023986816406, "learning_rate": 1.1970265069757548e-06, "loss": 0.09632024765014649, "step": 32170 }, { "epoch": 0.8476702567969365, "grad_norm": 2.530871629714966, "learning_rate": 1.1929822371159915e-06, "loss": 0.09520634412765502, "step": 32180 }, { "epoch": 0.8479336720414352, "grad_norm": 2.735085964202881, "learning_rate": 1.188944377383725e-06, "loss": 0.0966878890991211, "step": 32190 }, { "epoch": 0.8481970872859339, "grad_norm": 2.300551176071167, "learning_rate": 1.1849129307178841e-06, "loss": 0.08062283992767334, "step": 32200 }, { "epoch": 0.8484605025304327, "grad_norm": 2.806135416030884, "learning_rate": 1.180887900052724e-06, "loss": 0.08237143754959106, "step": 32210 }, { "epoch": 0.8487239177749314, "grad_norm": 3.4056713581085205, "learning_rate": 1.17686928831784e-06, "loss": 0.09998158216476441, "step": 32220 }, { "epoch": 0.8489873330194302, "grad_norm": 2.7797131538391113, "learning_rate": 1.1728570984381505e-06, "loss": 0.0934590220451355, "step": 32230 }, { "epoch": 0.8492507482639289, "grad_norm": 2.7816195487976074, "learning_rate": 1.1688513333339003e-06, "loss": 0.0969113826751709, "step": 32240 }, { "epoch": 0.8495141635084277, "grad_norm": 2.1711041927337646, "learning_rate": 1.1648519959206584e-06, "loss": 0.08231713771820068, "step": 32250 }, { "epoch": 0.8497775787529264, "grad_norm": 3.7808096408843994, "learning_rate": 1.1608590891093118e-06, "loss": 0.08189399242401123, "step": 32260 }, { "epoch": 0.8500409939974252, "grad_norm": 3.5198497772216797, "learning_rate": 1.156872615806075e-06, "loss": 0.09123708009719848, "step": 32270 }, { "epoch": 0.8503044092419239, "grad_norm": 2.949235677719116, "learning_rate": 1.1528925789124734e-06, "loss": 0.08859055042266846, "step": 32280 }, { "epoch": 0.8505678244864225, "grad_norm": 2.2262392044067383, "learning_rate": 1.1489189813253499e-06, "loss": 0.08816961646080017, "step": 32290 }, { "epoch": 0.8508312397309213, "grad_norm": 3.2346715927124023, "learning_rate": 1.1449518259368596e-06, "loss": 0.08228282928466797, "step": 32300 }, { "epoch": 0.85109465497542, "grad_norm": 2.9517695903778076, "learning_rate": 1.1409911156344677e-06, "loss": 0.08799546957015991, "step": 32310 }, { "epoch": 0.8513580702199188, "grad_norm": 3.041588544845581, "learning_rate": 1.1370368533009557e-06, "loss": 0.08754298686981202, "step": 32320 }, { "epoch": 0.8516214854644175, "grad_norm": 2.7088489532470703, "learning_rate": 1.1330890418144014e-06, "loss": 0.08284364938735962, "step": 32330 }, { "epoch": 0.8518849007089163, "grad_norm": 4.788604259490967, "learning_rate": 1.1291476840481908e-06, "loss": 0.09575401544570923, "step": 32340 }, { "epoch": 0.852148315953415, "grad_norm": 2.8867130279541016, "learning_rate": 1.125212782871019e-06, "loss": 0.09033082723617554, "step": 32350 }, { "epoch": 0.8524117311979138, "grad_norm": 2.692941427230835, "learning_rate": 1.1212843411468755e-06, "loss": 0.09568871259689331, "step": 32360 }, { "epoch": 0.8526751464424125, "grad_norm": 3.3971364498138428, "learning_rate": 1.1173623617350472e-06, "loss": 0.08620694875717164, "step": 32370 }, { "epoch": 0.8529385616869112, "grad_norm": 2.5423552989959717, "learning_rate": 1.1134468474901227e-06, "loss": 0.08830447793006897, "step": 32380 }, { "epoch": 0.85320197693141, "grad_norm": 2.6549065113067627, "learning_rate": 1.1095378012619796e-06, "loss": 0.0758131206035614, "step": 32390 }, { "epoch": 0.8534653921759087, "grad_norm": 2.2150943279266357, "learning_rate": 1.1056352258957924e-06, "loss": 0.08616618514060974, "step": 32400 }, { "epoch": 0.8537288074204075, "grad_norm": 3.8038885593414307, "learning_rate": 1.1017391242320263e-06, "loss": 0.08892855048179626, "step": 32410 }, { "epoch": 0.8539922226649062, "grad_norm": 3.197146415710449, "learning_rate": 1.0978494991064248e-06, "loss": 0.09576656222343445, "step": 32420 }, { "epoch": 0.854255637909405, "grad_norm": 2.375127077102661, "learning_rate": 1.0939663533500311e-06, "loss": 0.08151317834854126, "step": 32430 }, { "epoch": 0.8545190531539036, "grad_norm": 2.785122871398926, "learning_rate": 1.0900896897891634e-06, "loss": 0.0986911416053772, "step": 32440 }, { "epoch": 0.8547824683984024, "grad_norm": 3.6989784240722656, "learning_rate": 1.0862195112454289e-06, "loss": 0.08874974846839905, "step": 32450 }, { "epoch": 0.8550458836429011, "grad_norm": 1.8290084600448608, "learning_rate": 1.0823558205357054e-06, "loss": 0.07582336068153381, "step": 32460 }, { "epoch": 0.8553092988873998, "grad_norm": 4.1615309715271, "learning_rate": 1.0784986204721537e-06, "loss": 0.08925153613090515, "step": 32470 }, { "epoch": 0.8555727141318986, "grad_norm": 4.389146327972412, "learning_rate": 1.0746479138622134e-06, "loss": 0.09577749967575074, "step": 32480 }, { "epoch": 0.8558361293763973, "grad_norm": 3.181905508041382, "learning_rate": 1.070803703508595e-06, "loss": 0.10089679956436157, "step": 32490 }, { "epoch": 0.8560995446208961, "grad_norm": 3.137723207473755, "learning_rate": 1.066965992209279e-06, "loss": 0.078213369846344, "step": 32500 }, { "epoch": 0.8563629598653948, "grad_norm": 2.8316423892974854, "learning_rate": 1.0631347827575179e-06, "loss": 0.08290903568267823, "step": 32510 }, { "epoch": 0.8566263751098936, "grad_norm": 3.4813730716705322, "learning_rate": 1.0593100779418286e-06, "loss": 0.08574831485748291, "step": 32520 }, { "epoch": 0.8568897903543923, "grad_norm": 4.004169940948486, "learning_rate": 1.0554918805460013e-06, "loss": 0.1021581768989563, "step": 32530 }, { "epoch": 0.8571532055988911, "grad_norm": 3.1221837997436523, "learning_rate": 1.051680193349084e-06, "loss": 0.0900246798992157, "step": 32540 }, { "epoch": 0.8574166208433898, "grad_norm": 3.056314468383789, "learning_rate": 1.0478750191253817e-06, "loss": 0.08575124144554139, "step": 32550 }, { "epoch": 0.8576800360878885, "grad_norm": 3.0542638301849365, "learning_rate": 1.0440763606444694e-06, "loss": 0.08063366413116455, "step": 32560 }, { "epoch": 0.8579434513323873, "grad_norm": 3.0438966751098633, "learning_rate": 1.0402842206711728e-06, "loss": 0.10674437284469604, "step": 32570 }, { "epoch": 0.8582068665768859, "grad_norm": 4.585306644439697, "learning_rate": 1.0364986019655765e-06, "loss": 0.09235278964042663, "step": 32580 }, { "epoch": 0.8584702818213847, "grad_norm": 3.020968198776245, "learning_rate": 1.0327195072830153e-06, "loss": 0.09338023662567138, "step": 32590 }, { "epoch": 0.8587336970658834, "grad_norm": 2.4640886783599854, "learning_rate": 1.028946939374076e-06, "loss": 0.10037720203399658, "step": 32600 }, { "epoch": 0.8589971123103822, "grad_norm": 3.2039833068847656, "learning_rate": 1.0251809009846003e-06, "loss": 0.08920050859451294, "step": 32610 }, { "epoch": 0.8592605275548809, "grad_norm": 2.199112892150879, "learning_rate": 1.0214213948556718e-06, "loss": 0.07883176803588868, "step": 32620 }, { "epoch": 0.8595239427993796, "grad_norm": 2.9971423149108887, "learning_rate": 1.0176684237236212e-06, "loss": 0.09104140996932983, "step": 32630 }, { "epoch": 0.8597873580438784, "grad_norm": 2.982630491256714, "learning_rate": 1.0139219903200227e-06, "loss": 0.08842053413391113, "step": 32640 }, { "epoch": 0.8600507732883771, "grad_norm": 2.710325002670288, "learning_rate": 1.0101820973716891e-06, "loss": 0.0955734372138977, "step": 32650 }, { "epoch": 0.8603141885328759, "grad_norm": 3.3860409259796143, "learning_rate": 1.0064487476006812e-06, "loss": 0.08669326305389405, "step": 32660 }, { "epoch": 0.8605776037773746, "grad_norm": 2.5106115341186523, "learning_rate": 1.0027219437242907e-06, "loss": 0.0889466404914856, "step": 32670 }, { "epoch": 0.8608410190218734, "grad_norm": 2.7231998443603516, "learning_rate": 9.990016884550425e-07, "loss": 0.09223927855491638, "step": 32680 }, { "epoch": 0.8611044342663721, "grad_norm": 2.933701992034912, "learning_rate": 9.952879845007024e-07, "loss": 0.08801427483558655, "step": 32690 }, { "epoch": 0.8613678495108709, "grad_norm": 4.7756428718566895, "learning_rate": 9.915808345642642e-07, "loss": 0.08389148116111755, "step": 32700 }, { "epoch": 0.8616312647553696, "grad_norm": 3.6905558109283447, "learning_rate": 9.878802413439514e-07, "loss": 0.08894308805465698, "step": 32710 }, { "epoch": 0.8618946799998682, "grad_norm": 3.8245530128479004, "learning_rate": 9.841862075332143e-07, "loss": 0.08153004646301269, "step": 32720 }, { "epoch": 0.862158095244367, "grad_norm": 1.7937746047973633, "learning_rate": 9.804987358207274e-07, "loss": 0.08218700885772705, "step": 32730 }, { "epoch": 0.8624215104888657, "grad_norm": 3.0946924686431885, "learning_rate": 9.768178288903985e-07, "loss": 0.0990536391735077, "step": 32740 }, { "epoch": 0.8626849257333645, "grad_norm": 2.9138779640197754, "learning_rate": 9.731434894213465e-07, "loss": 0.09229787588119506, "step": 32750 }, { "epoch": 0.8629483409778632, "grad_norm": 2.796353578567505, "learning_rate": 9.694757200879146e-07, "loss": 0.0962436318397522, "step": 32760 }, { "epoch": 0.863211756222362, "grad_norm": 3.0355894565582275, "learning_rate": 9.658145235596628e-07, "loss": 0.09681029319763183, "step": 32770 }, { "epoch": 0.8634751714668607, "grad_norm": 4.103431224822998, "learning_rate": 9.621599025013672e-07, "loss": 0.09177099466323853, "step": 32780 }, { "epoch": 0.8637385867113595, "grad_norm": 2.846986770629883, "learning_rate": 9.585118595730226e-07, "loss": 0.10183346271514893, "step": 32790 }, { "epoch": 0.8640020019558582, "grad_norm": 3.125056505203247, "learning_rate": 9.548703974298311e-07, "loss": 0.08817275762557983, "step": 32800 }, { "epoch": 0.8642654172003569, "grad_norm": 3.931093215942383, "learning_rate": 9.512355187222e-07, "loss": 0.09189233779907227, "step": 32810 }, { "epoch": 0.8645288324448557, "grad_norm": 2.475339651107788, "learning_rate": 9.47607226095758e-07, "loss": 0.08796191215515137, "step": 32820 }, { "epoch": 0.8647922476893544, "grad_norm": 4.151906490325928, "learning_rate": 9.439855221913297e-07, "loss": 0.0943767249584198, "step": 32830 }, { "epoch": 0.8650556629338532, "grad_norm": 2.5837721824645996, "learning_rate": 9.403704096449484e-07, "loss": 0.0846022605895996, "step": 32840 }, { "epoch": 0.8653190781783519, "grad_norm": 3.4563910961151123, "learning_rate": 9.367618910878484e-07, "loss": 0.09141744375228882, "step": 32850 }, { "epoch": 0.8655824934228507, "grad_norm": 3.3639988899230957, "learning_rate": 9.331599691464644e-07, "loss": 0.09926159381866455, "step": 32860 }, { "epoch": 0.8658459086673493, "grad_norm": 2.68133282661438, "learning_rate": 9.295646464424323e-07, "loss": 0.07868162989616394, "step": 32870 }, { "epoch": 0.8661093239118481, "grad_norm": 3.4418070316314697, "learning_rate": 9.259759255925827e-07, "loss": 0.0839989185333252, "step": 32880 }, { "epoch": 0.8663727391563468, "grad_norm": 3.2081265449523926, "learning_rate": 9.223938092089413e-07, "loss": 0.08773126006126404, "step": 32890 }, { "epoch": 0.8666361544008455, "grad_norm": 2.454683780670166, "learning_rate": 9.188182998987261e-07, "loss": 0.0820432186126709, "step": 32900 }, { "epoch": 0.8668995696453443, "grad_norm": 3.289794445037842, "learning_rate": 9.152494002643464e-07, "loss": 0.088738614320755, "step": 32910 }, { "epoch": 0.867162984889843, "grad_norm": 3.3731651306152344, "learning_rate": 9.116871129034066e-07, "loss": 0.08697800040245056, "step": 32920 }, { "epoch": 0.8674264001343418, "grad_norm": 2.857538938522339, "learning_rate": 9.081314404086872e-07, "loss": 0.08273639678955078, "step": 32930 }, { "epoch": 0.8676898153788405, "grad_norm": 3.0904202461242676, "learning_rate": 9.045823853681602e-07, "loss": 0.09320009350776673, "step": 32940 }, { "epoch": 0.8679532306233393, "grad_norm": 2.693464517593384, "learning_rate": 9.010399503649836e-07, "loss": 0.08980790376663209, "step": 32950 }, { "epoch": 0.868216645867838, "grad_norm": 3.465594530105591, "learning_rate": 8.975041379774951e-07, "loss": 0.10310330390930175, "step": 32960 }, { "epoch": 0.8684800611123368, "grad_norm": 2.7278904914855957, "learning_rate": 8.939749507792094e-07, "loss": 0.09242959022521972, "step": 32970 }, { "epoch": 0.8687434763568355, "grad_norm": 2.8747365474700928, "learning_rate": 8.904523913388218e-07, "loss": 0.07709479928016663, "step": 32980 }, { "epoch": 0.8690068916013342, "grad_norm": 2.721312999725342, "learning_rate": 8.869364622202015e-07, "loss": 0.08395496606826783, "step": 32990 }, { "epoch": 0.869270306845833, "grad_norm": 2.258620500564575, "learning_rate": 8.834271659823967e-07, "loss": 0.08888227343559266, "step": 33000 }, { "epoch": 0.8695337220903316, "grad_norm": 2.208204746246338, "learning_rate": 8.799245051796246e-07, "loss": 0.08902350664138795, "step": 33010 }, { "epoch": 0.8697971373348304, "grad_norm": 2.6384472846984863, "learning_rate": 8.764284823612667e-07, "loss": 0.08656991720199585, "step": 33020 }, { "epoch": 0.8700605525793291, "grad_norm": 3.4460151195526123, "learning_rate": 8.729391000718867e-07, "loss": 0.0963441252708435, "step": 33030 }, { "epoch": 0.8703239678238279, "grad_norm": 2.9477736949920654, "learning_rate": 8.694563608512041e-07, "loss": 0.09387299418449402, "step": 33040 }, { "epoch": 0.8705873830683266, "grad_norm": 3.3285911083221436, "learning_rate": 8.659802672341078e-07, "loss": 0.09106696844100952, "step": 33050 }, { "epoch": 0.8708507983128254, "grad_norm": 2.3664605617523193, "learning_rate": 8.625108217506472e-07, "loss": 0.08322448134422303, "step": 33060 }, { "epoch": 0.8711142135573241, "grad_norm": 3.7617034912109375, "learning_rate": 8.590480269260337e-07, "loss": 0.09010876417160034, "step": 33070 }, { "epoch": 0.8713776288018228, "grad_norm": 3.8081319332122803, "learning_rate": 8.555918852806421e-07, "loss": 0.09080619812011718, "step": 33080 }, { "epoch": 0.8716410440463216, "grad_norm": 3.099125623703003, "learning_rate": 8.521423993300005e-07, "loss": 0.09265373349189758, "step": 33090 }, { "epoch": 0.8719044592908203, "grad_norm": 3.82966685295105, "learning_rate": 8.486995715847923e-07, "loss": 0.0958257794380188, "step": 33100 }, { "epoch": 0.8721678745353191, "grad_norm": 2.663174629211426, "learning_rate": 8.452634045508557e-07, "loss": 0.07829921245574951, "step": 33110 }, { "epoch": 0.8724312897798178, "grad_norm": 3.19863224029541, "learning_rate": 8.41833900729182e-07, "loss": 0.08812856674194336, "step": 33120 }, { "epoch": 0.8726947050243166, "grad_norm": 3.1322178840637207, "learning_rate": 8.384110626159126e-07, "loss": 0.09103515148162841, "step": 33130 }, { "epoch": 0.8729581202688153, "grad_norm": 2.898738145828247, "learning_rate": 8.349948927023388e-07, "loss": 0.08207857608795166, "step": 33140 }, { "epoch": 0.8732215355133139, "grad_norm": 3.0527703762054443, "learning_rate": 8.315853934748908e-07, "loss": 0.09289202690124512, "step": 33150 }, { "epoch": 0.8734849507578127, "grad_norm": 2.5004055500030518, "learning_rate": 8.281825674151533e-07, "loss": 0.09376102089881896, "step": 33160 }, { "epoch": 0.8737483660023114, "grad_norm": 2.747701644897461, "learning_rate": 8.247864169998499e-07, "loss": 0.08988502621650696, "step": 33170 }, { "epoch": 0.8740117812468102, "grad_norm": 2.471207857131958, "learning_rate": 8.213969447008441e-07, "loss": 0.07651523351669312, "step": 33180 }, { "epoch": 0.8742751964913089, "grad_norm": 2.876281499862671, "learning_rate": 8.180141529851404e-07, "loss": 0.08601000308990478, "step": 33190 }, { "epoch": 0.8745386117358077, "grad_norm": 3.3386693000793457, "learning_rate": 8.146380443148794e-07, "loss": 0.08774766921997071, "step": 33200 }, { "epoch": 0.8748020269803064, "grad_norm": 4.429173946380615, "learning_rate": 8.11268621147342e-07, "loss": 0.08983232975006103, "step": 33210 }, { "epoch": 0.8750654422248052, "grad_norm": 3.1978249549865723, "learning_rate": 8.079058859349364e-07, "loss": 0.09865237474441528, "step": 33220 }, { "epoch": 0.8753288574693039, "grad_norm": 2.776215076446533, "learning_rate": 8.045498411252095e-07, "loss": 0.08836942911148071, "step": 33230 }, { "epoch": 0.8755922727138026, "grad_norm": 4.2545647621154785, "learning_rate": 8.012004891608327e-07, "loss": 0.10253009796142579, "step": 33240 }, { "epoch": 0.8758556879583014, "grad_norm": 3.142240524291992, "learning_rate": 7.978578324796072e-07, "loss": 0.09346885681152343, "step": 33250 }, { "epoch": 0.8761191032028001, "grad_norm": 2.4203219413757324, "learning_rate": 7.945218735144689e-07, "loss": 0.0876720666885376, "step": 33260 }, { "epoch": 0.8763825184472989, "grad_norm": 2.680615186691284, "learning_rate": 7.91192614693469e-07, "loss": 0.07844331264495849, "step": 33270 }, { "epoch": 0.8766459336917976, "grad_norm": 4.084168910980225, "learning_rate": 7.878700584397836e-07, "loss": 0.0904033899307251, "step": 33280 }, { "epoch": 0.8769093489362964, "grad_norm": 3.2061004638671875, "learning_rate": 7.84554207171716e-07, "loss": 0.07571238279342651, "step": 33290 }, { "epoch": 0.877172764180795, "grad_norm": 3.0252633094787598, "learning_rate": 7.812450633026847e-07, "loss": 0.0806653618812561, "step": 33300 }, { "epoch": 0.8774361794252938, "grad_norm": 3.7704968452453613, "learning_rate": 7.779426292412273e-07, "loss": 0.07912169694900513, "step": 33310 }, { "epoch": 0.8776995946697925, "grad_norm": 2.351949691772461, "learning_rate": 7.746469073909979e-07, "loss": 0.08100239038467408, "step": 33320 }, { "epoch": 0.8779630099142912, "grad_norm": 2.3370180130004883, "learning_rate": 7.713579001507632e-07, "loss": 0.09470623731613159, "step": 33330 }, { "epoch": 0.87822642515879, "grad_norm": 2.4741339683532715, "learning_rate": 7.680756099144072e-07, "loss": 0.08651316165924072, "step": 33340 }, { "epoch": 0.8784898404032887, "grad_norm": 3.6233773231506348, "learning_rate": 7.648000390709221e-07, "loss": 0.09087352752685547, "step": 33350 }, { "epoch": 0.8787532556477875, "grad_norm": 3.0098185539245605, "learning_rate": 7.615311900044087e-07, "loss": 0.0727483868598938, "step": 33360 }, { "epoch": 0.8790166708922862, "grad_norm": 3.930030584335327, "learning_rate": 7.582690650940771e-07, "loss": 0.08658503293991089, "step": 33370 }, { "epoch": 0.879280086136785, "grad_norm": 2.536533832550049, "learning_rate": 7.550136667142393e-07, "loss": 0.08407927751541137, "step": 33380 }, { "epoch": 0.8795435013812837, "grad_norm": 2.61950421333313, "learning_rate": 7.517649972343199e-07, "loss": 0.07774099111557006, "step": 33390 }, { "epoch": 0.8798069166257825, "grad_norm": 2.062739610671997, "learning_rate": 7.485230590188397e-07, "loss": 0.07658504843711852, "step": 33400 }, { "epoch": 0.8800703318702812, "grad_norm": 3.3957290649414062, "learning_rate": 7.452878544274167e-07, "loss": 0.08277466297149658, "step": 33410 }, { "epoch": 0.8803337471147799, "grad_norm": 3.2577383518218994, "learning_rate": 7.420593858147773e-07, "loss": 0.08509113788604736, "step": 33420 }, { "epoch": 0.8805971623592787, "grad_norm": 3.1876611709594727, "learning_rate": 7.388376555307386e-07, "loss": 0.0930111587047577, "step": 33430 }, { "epoch": 0.8808605776037773, "grad_norm": 3.6059939861297607, "learning_rate": 7.356226659202137e-07, "loss": 0.09433202743530274, "step": 33440 }, { "epoch": 0.8811239928482761, "grad_norm": 2.130681037902832, "learning_rate": 7.324144193232119e-07, "loss": 0.07947914600372315, "step": 33450 }, { "epoch": 0.8813874080927748, "grad_norm": 4.078615188598633, "learning_rate": 7.29212918074833e-07, "loss": 0.08237431645393371, "step": 33460 }, { "epoch": 0.8816508233372736, "grad_norm": 2.5744826793670654, "learning_rate": 7.260181645052689e-07, "loss": 0.0892946720123291, "step": 33470 }, { "epoch": 0.8819142385817723, "grad_norm": 3.1801135540008545, "learning_rate": 7.228301609397981e-07, "loss": 0.09688584804534912, "step": 33480 }, { "epoch": 0.8821776538262711, "grad_norm": 2.979107618331909, "learning_rate": 7.196489096987869e-07, "loss": 0.09564831256866455, "step": 33490 }, { "epoch": 0.8824410690707698, "grad_norm": 3.5899975299835205, "learning_rate": 7.164744130976886e-07, "loss": 0.07234923839569092, "step": 33500 }, { "epoch": 0.8827044843152685, "grad_norm": 3.533519744873047, "learning_rate": 7.133066734470351e-07, "loss": 0.09257988929748535, "step": 33510 }, { "epoch": 0.8829678995597673, "grad_norm": 3.1265056133270264, "learning_rate": 7.101456930524509e-07, "loss": 0.0902636170387268, "step": 33520 }, { "epoch": 0.883231314804266, "grad_norm": 2.9531257152557373, "learning_rate": 7.069914742146267e-07, "loss": 0.08576451539993286, "step": 33530 }, { "epoch": 0.8834947300487648, "grad_norm": 3.312082529067993, "learning_rate": 7.038440192293416e-07, "loss": 0.08315178155899047, "step": 33540 }, { "epoch": 0.8837581452932635, "grad_norm": 2.8581430912017822, "learning_rate": 7.007033303874511e-07, "loss": 0.08873788118362427, "step": 33550 }, { "epoch": 0.8840215605377623, "grad_norm": 2.2090659141540527, "learning_rate": 6.97569409974882e-07, "loss": 0.0881535530090332, "step": 33560 }, { "epoch": 0.884284975782261, "grad_norm": 3.3284502029418945, "learning_rate": 6.944422602726375e-07, "loss": 0.07573654651641845, "step": 33570 }, { "epoch": 0.8845483910267598, "grad_norm": 3.075071334838867, "learning_rate": 6.913218835567925e-07, "loss": 0.09423752427101136, "step": 33580 }, { "epoch": 0.8848118062712584, "grad_norm": 3.3416695594787598, "learning_rate": 6.882082820984892e-07, "loss": 0.08886175155639649, "step": 33590 }, { "epoch": 0.8850752215157571, "grad_norm": 3.094182252883911, "learning_rate": 6.851014581639437e-07, "loss": 0.07777606844902038, "step": 33600 }, { "epoch": 0.8853386367602559, "grad_norm": 2.8987481594085693, "learning_rate": 6.820014140144382e-07, "loss": 0.08347312211990357, "step": 33610 }, { "epoch": 0.8856020520047546, "grad_norm": 3.8380250930786133, "learning_rate": 6.789081519063134e-07, "loss": 0.09110926389694214, "step": 33620 }, { "epoch": 0.8858654672492534, "grad_norm": 2.577397346496582, "learning_rate": 6.758216740909829e-07, "loss": 0.0922200322151184, "step": 33630 }, { "epoch": 0.8861288824937521, "grad_norm": 3.6527347564697266, "learning_rate": 6.727419828149173e-07, "loss": 0.0938602089881897, "step": 33640 }, { "epoch": 0.8863922977382509, "grad_norm": 3.466562271118164, "learning_rate": 6.696690803196482e-07, "loss": 0.08979893326759339, "step": 33650 }, { "epoch": 0.8866557129827496, "grad_norm": 2.9820618629455566, "learning_rate": 6.666029688417674e-07, "loss": 0.09374719858169556, "step": 33660 }, { "epoch": 0.8869191282272483, "grad_norm": 2.0097410678863525, "learning_rate": 6.635436506129223e-07, "loss": 0.08514592647552491, "step": 33670 }, { "epoch": 0.8871825434717471, "grad_norm": 3.7405121326446533, "learning_rate": 6.60491127859818e-07, "loss": 0.0853306531906128, "step": 33680 }, { "epoch": 0.8874459587162458, "grad_norm": 2.6622934341430664, "learning_rate": 6.57445402804211e-07, "loss": 0.07905309200286866, "step": 33690 }, { "epoch": 0.8877093739607446, "grad_norm": 3.4130747318267822, "learning_rate": 6.544064776629123e-07, "loss": 0.08580552339553833, "step": 33700 }, { "epoch": 0.8879727892052433, "grad_norm": 2.1667416095733643, "learning_rate": 6.513743546477824e-07, "loss": 0.07473984956741334, "step": 33710 }, { "epoch": 0.888236204449742, "grad_norm": 3.1489193439483643, "learning_rate": 6.483490359657296e-07, "loss": 0.08496062755584717, "step": 33720 }, { "epoch": 0.8884996196942407, "grad_norm": 2.5333380699157715, "learning_rate": 6.453305238187147e-07, "loss": 0.0721280574798584, "step": 33730 }, { "epoch": 0.8887630349387395, "grad_norm": 3.403695583343506, "learning_rate": 6.423188204037412e-07, "loss": 0.08447886705398559, "step": 33740 }, { "epoch": 0.8890264501832382, "grad_norm": 3.616903305053711, "learning_rate": 6.393139279128513e-07, "loss": 0.08639968633651733, "step": 33750 }, { "epoch": 0.8892898654277369, "grad_norm": 3.2610082626342773, "learning_rate": 6.3631584853314e-07, "loss": 0.0886648416519165, "step": 33760 }, { "epoch": 0.8895532806722357, "grad_norm": 4.342204570770264, "learning_rate": 6.333245844467384e-07, "loss": 0.08771692514419556, "step": 33770 }, { "epoch": 0.8898166959167344, "grad_norm": 2.9558451175689697, "learning_rate": 6.303401378308172e-07, "loss": 0.08513480424880981, "step": 33780 }, { "epoch": 0.8900801111612332, "grad_norm": 3.3833584785461426, "learning_rate": 6.273625108575843e-07, "loss": 0.08831550478935242, "step": 33790 }, { "epoch": 0.8903435264057319, "grad_norm": 2.4837746620178223, "learning_rate": 6.243917056942849e-07, "loss": 0.08225837945938111, "step": 33800 }, { "epoch": 0.8906069416502307, "grad_norm": 3.5455639362335205, "learning_rate": 6.214277245032019e-07, "loss": 0.08585963249206544, "step": 33810 }, { "epoch": 0.8908703568947294, "grad_norm": 2.867846965789795, "learning_rate": 6.184705694416471e-07, "loss": 0.08504784107208252, "step": 33820 }, { "epoch": 0.8911337721392282, "grad_norm": 3.2729697227478027, "learning_rate": 6.155202426619655e-07, "loss": 0.08707352876663207, "step": 33830 }, { "epoch": 0.8913971873837269, "grad_norm": 2.7022464275360107, "learning_rate": 6.125767463115317e-07, "loss": 0.08570126891136169, "step": 33840 }, { "epoch": 0.8916606026282256, "grad_norm": 3.7274389266967773, "learning_rate": 6.096400825327487e-07, "loss": 0.0864186704158783, "step": 33850 }, { "epoch": 0.8919240178727244, "grad_norm": 3.6806533336639404, "learning_rate": 6.067102534630498e-07, "loss": 0.08243987560272217, "step": 33860 }, { "epoch": 0.892187433117223, "grad_norm": 3.4134750366210938, "learning_rate": 6.037872612348905e-07, "loss": 0.08662645816802979, "step": 33870 }, { "epoch": 0.8924508483617218, "grad_norm": 3.3761091232299805, "learning_rate": 6.008711079757467e-07, "loss": 0.08861711025238037, "step": 33880 }, { "epoch": 0.8927142636062205, "grad_norm": 3.8590023517608643, "learning_rate": 5.979617958081241e-07, "loss": 0.08228753805160523, "step": 33890 }, { "epoch": 0.8929776788507193, "grad_norm": 2.5920283794403076, "learning_rate": 5.950593268495464e-07, "loss": 0.07754223346710205, "step": 33900 }, { "epoch": 0.893241094095218, "grad_norm": 3.191586971282959, "learning_rate": 5.921637032125538e-07, "loss": 0.08434219360351562, "step": 33910 }, { "epoch": 0.8935045093397168, "grad_norm": 3.006871461868286, "learning_rate": 5.89274927004706e-07, "loss": 0.08247482776641846, "step": 33920 }, { "epoch": 0.8937679245842155, "grad_norm": 3.201061725616455, "learning_rate": 5.863930003285779e-07, "loss": 0.07925151586532593, "step": 33930 }, { "epoch": 0.8940313398287142, "grad_norm": 3.2484443187713623, "learning_rate": 5.835179252817635e-07, "loss": 0.08889644145965576, "step": 33940 }, { "epoch": 0.894294755073213, "grad_norm": 3.1078145503997803, "learning_rate": 5.806497039568648e-07, "loss": 0.08446462750434876, "step": 33950 }, { "epoch": 0.8945581703177117, "grad_norm": 3.555105686187744, "learning_rate": 5.777883384414973e-07, "loss": 0.07739390134811401, "step": 33960 }, { "epoch": 0.8948215855622105, "grad_norm": 2.650618314743042, "learning_rate": 5.749338308182851e-07, "loss": 0.08098430633544922, "step": 33970 }, { "epoch": 0.8950850008067092, "grad_norm": 3.9091811180114746, "learning_rate": 5.720861831648617e-07, "loss": 0.09174574613571167, "step": 33980 }, { "epoch": 0.895348416051208, "grad_norm": 3.762361526489258, "learning_rate": 5.692453975538703e-07, "loss": 0.08094145059585571, "step": 33990 }, { "epoch": 0.8956118312957067, "grad_norm": 2.5618748664855957, "learning_rate": 5.664114760529571e-07, "loss": 0.0791934609413147, "step": 34000 }, { "epoch": 0.8958752465402054, "grad_norm": 2.804961919784546, "learning_rate": 5.635844207247687e-07, "loss": 0.08356852531433105, "step": 34010 }, { "epoch": 0.8961386617847041, "grad_norm": 3.255131483078003, "learning_rate": 5.607642336269615e-07, "loss": 0.08215757012367249, "step": 34020 }, { "epoch": 0.8964020770292028, "grad_norm": 3.2720890045166016, "learning_rate": 5.579509168121888e-07, "loss": 0.0820299506187439, "step": 34030 }, { "epoch": 0.8966654922737016, "grad_norm": 4.233641147613525, "learning_rate": 5.551444723281029e-07, "loss": 0.07894362211227417, "step": 34040 }, { "epoch": 0.8969289075182003, "grad_norm": 3.622255325317383, "learning_rate": 5.523449022173566e-07, "loss": 0.09294643402099609, "step": 34050 }, { "epoch": 0.8971923227626991, "grad_norm": 2.470841884613037, "learning_rate": 5.495522085175942e-07, "loss": 0.07745720148086548, "step": 34060 }, { "epoch": 0.8974557380071978, "grad_norm": 3.39422345161438, "learning_rate": 5.467663932614642e-07, "loss": 0.07891461849212647, "step": 34070 }, { "epoch": 0.8977191532516966, "grad_norm": 3.165367603302002, "learning_rate": 5.439874584766014e-07, "loss": 0.0947558581829071, "step": 34080 }, { "epoch": 0.8979825684961953, "grad_norm": 3.5732998847961426, "learning_rate": 5.412154061856301e-07, "loss": 0.07709518671035767, "step": 34090 }, { "epoch": 0.8982459837406941, "grad_norm": 2.571470260620117, "learning_rate": 5.38450238406174e-07, "loss": 0.08872199058532715, "step": 34100 }, { "epoch": 0.8985093989851928, "grad_norm": 2.8618483543395996, "learning_rate": 5.356919571508401e-07, "loss": 0.0915326714515686, "step": 34110 }, { "epoch": 0.8987728142296915, "grad_norm": 4.355645656585693, "learning_rate": 5.329405644272268e-07, "loss": 0.0845130980014801, "step": 34120 }, { "epoch": 0.8990362294741903, "grad_norm": 3.5744502544403076, "learning_rate": 5.301960622379132e-07, "loss": 0.07691995501518249, "step": 34130 }, { "epoch": 0.899299644718689, "grad_norm": 3.605327606201172, "learning_rate": 5.27458452580466e-07, "loss": 0.0885988712310791, "step": 34140 }, { "epoch": 0.8995630599631877, "grad_norm": 3.2719509601593018, "learning_rate": 5.24727737447438e-07, "loss": 0.08142627477645874, "step": 34150 }, { "epoch": 0.8998264752076864, "grad_norm": 3.3356800079345703, "learning_rate": 5.220039188263615e-07, "loss": 0.08494765758514404, "step": 34160 }, { "epoch": 0.9000898904521852, "grad_norm": 3.272571563720703, "learning_rate": 5.192869986997484e-07, "loss": 0.08318660259246827, "step": 34170 }, { "epoch": 0.9003533056966839, "grad_norm": 3.054112195968628, "learning_rate": 5.165769790450903e-07, "loss": 0.08590464591979981, "step": 34180 }, { "epoch": 0.9006167209411826, "grad_norm": 2.7261688709259033, "learning_rate": 5.138738618348548e-07, "loss": 0.0747373342514038, "step": 34190 }, { "epoch": 0.9008801361856814, "grad_norm": 3.451206684112549, "learning_rate": 5.111776490364906e-07, "loss": 0.08322645425796509, "step": 34200 }, { "epoch": 0.9011435514301801, "grad_norm": 3.1690359115600586, "learning_rate": 5.084883426124165e-07, "loss": 0.06670655608177185, "step": 34210 }, { "epoch": 0.9014069666746789, "grad_norm": 3.080108642578125, "learning_rate": 5.058059445200237e-07, "loss": 0.08162369728088378, "step": 34220 }, { "epoch": 0.9016703819191776, "grad_norm": 2.728167772293091, "learning_rate": 5.031304567116802e-07, "loss": 0.08675851821899414, "step": 34230 }, { "epoch": 0.9019337971636764, "grad_norm": 4.3546319007873535, "learning_rate": 5.004618811347195e-07, "loss": 0.09387590885162353, "step": 34240 }, { "epoch": 0.9021972124081751, "grad_norm": 3.5529592037200928, "learning_rate": 4.978002197314469e-07, "loss": 0.08580414056777955, "step": 34250 }, { "epoch": 0.9024606276526739, "grad_norm": 3.6288952827453613, "learning_rate": 4.951454744391338e-07, "loss": 0.08337353467941284, "step": 34260 }, { "epoch": 0.9027240428971726, "grad_norm": 2.876065969467163, "learning_rate": 4.924976471900178e-07, "loss": 0.07400510907173156, "step": 34270 }, { "epoch": 0.9029874581416713, "grad_norm": 4.327888011932373, "learning_rate": 4.89856739911303e-07, "loss": 0.07898850440979004, "step": 34280 }, { "epoch": 0.90325087338617, "grad_norm": 5.125980377197266, "learning_rate": 4.872227545251562e-07, "loss": 0.07621689438819886, "step": 34290 }, { "epoch": 0.9035142886306687, "grad_norm": 2.8650689125061035, "learning_rate": 4.84595692948705e-07, "loss": 0.08053421974182129, "step": 34300 }, { "epoch": 0.9037777038751675, "grad_norm": 3.1784636974334717, "learning_rate": 4.819755570940377e-07, "loss": 0.09256478548049926, "step": 34310 }, { "epoch": 0.9040411191196662, "grad_norm": 3.4430346488952637, "learning_rate": 4.793623488682008e-07, "loss": 0.09212582111358643, "step": 34320 }, { "epoch": 0.904304534364165, "grad_norm": 4.799671649932861, "learning_rate": 4.7675607017320323e-07, "loss": 0.08183037042617798, "step": 34330 }, { "epoch": 0.9045679496086637, "grad_norm": 4.785935401916504, "learning_rate": 4.741567229060062e-07, "loss": 0.07596416473388672, "step": 34340 }, { "epoch": 0.9048313648531625, "grad_norm": 3.4324591159820557, "learning_rate": 4.715643089585242e-07, "loss": 0.07960377931594849, "step": 34350 }, { "epoch": 0.9050947800976612, "grad_norm": 3.113072156906128, "learning_rate": 4.6897883021763013e-07, "loss": 0.07639552354812622, "step": 34360 }, { "epoch": 0.9053581953421599, "grad_norm": 3.807981014251709, "learning_rate": 4.664002885651475e-07, "loss": 0.08166407942771911, "step": 34370 }, { "epoch": 0.9056216105866587, "grad_norm": 4.980324745178223, "learning_rate": 4.638286858778496e-07, "loss": 0.08832589387893677, "step": 34380 }, { "epoch": 0.9058850258311574, "grad_norm": 2.5279386043548584, "learning_rate": 4.612640240274591e-07, "loss": 0.08098738193511963, "step": 34390 }, { "epoch": 0.9061484410756562, "grad_norm": 2.658425807952881, "learning_rate": 4.587063048806462e-07, "loss": 0.08618833422660828, "step": 34400 }, { "epoch": 0.9064118563201549, "grad_norm": 3.293240785598755, "learning_rate": 4.561555302990328e-07, "loss": 0.09159172177314759, "step": 34410 }, { "epoch": 0.9066752715646537, "grad_norm": 3.528641700744629, "learning_rate": 4.536117021391806e-07, "loss": 0.0895397424697876, "step": 34420 }, { "epoch": 0.9069386868091524, "grad_norm": 3.8428449630737305, "learning_rate": 4.5107482225259735e-07, "loss": 0.08772671222686768, "step": 34430 }, { "epoch": 0.9072021020536511, "grad_norm": 3.4045557975769043, "learning_rate": 4.4854489248573494e-07, "loss": 0.08292000889778137, "step": 34440 }, { "epoch": 0.9074655172981498, "grad_norm": 3.5913116931915283, "learning_rate": 4.460219146799838e-07, "loss": 0.07479780912399292, "step": 34450 }, { "epoch": 0.9077289325426485, "grad_norm": 3.0088114738464355, "learning_rate": 4.4350589067167717e-07, "loss": 0.07705175876617432, "step": 34460 }, { "epoch": 0.9079923477871473, "grad_norm": 3.806795358657837, "learning_rate": 4.409968222920868e-07, "loss": 0.08993245959281922, "step": 34470 }, { "epoch": 0.908255763031646, "grad_norm": 2.7671923637390137, "learning_rate": 4.384947113674165e-07, "loss": 0.0839149534702301, "step": 34480 }, { "epoch": 0.9085191782761448, "grad_norm": 3.3305420875549316, "learning_rate": 4.35999559718816e-07, "loss": 0.08590950965881347, "step": 34490 }, { "epoch": 0.9087825935206435, "grad_norm": 2.8584537506103516, "learning_rate": 4.3351136916236156e-07, "loss": 0.08750685453414916, "step": 34500 }, { "epoch": 0.9090460087651423, "grad_norm": 3.7898828983306885, "learning_rate": 4.3103014150906675e-07, "loss": 0.08020315766334533, "step": 34510 }, { "epoch": 0.909309424009641, "grad_norm": 3.5840959548950195, "learning_rate": 4.285558785648747e-07, "loss": 0.07868388295173645, "step": 34520 }, { "epoch": 0.9095728392541398, "grad_norm": 2.4898359775543213, "learning_rate": 4.260885821306604e-07, "loss": 0.07393144369125366, "step": 34530 }, { "epoch": 0.9098362544986385, "grad_norm": 3.5966269969940186, "learning_rate": 4.236282540022307e-07, "loss": 0.08008968830108643, "step": 34540 }, { "epoch": 0.9100996697431372, "grad_norm": 2.7266252040863037, "learning_rate": 4.2117489597031767e-07, "loss": 0.08197280764579773, "step": 34550 }, { "epoch": 0.910363084987636, "grad_norm": 3.826730251312256, "learning_rate": 4.1872850982058067e-07, "loss": 0.07699528932571412, "step": 34560 }, { "epoch": 0.9106265002321347, "grad_norm": 3.55062198638916, "learning_rate": 4.1628909733360534e-07, "loss": 0.08169797658920289, "step": 34570 }, { "epoch": 0.9108899154766334, "grad_norm": 3.6435155868530273, "learning_rate": 4.1385666028490036e-07, "loss": 0.08992097973823547, "step": 34580 }, { "epoch": 0.9111533307211321, "grad_norm": 3.5133721828460693, "learning_rate": 4.114312004449006e-07, "loss": 0.0761521577835083, "step": 34590 }, { "epoch": 0.9114167459656309, "grad_norm": 3.859178066253662, "learning_rate": 4.090127195789606e-07, "loss": 0.0869214415550232, "step": 34600 }, { "epoch": 0.9116801612101296, "grad_norm": 3.1446430683135986, "learning_rate": 4.066012194473512e-07, "loss": 0.09349464178085327, "step": 34610 }, { "epoch": 0.9119435764546284, "grad_norm": 3.1985557079315186, "learning_rate": 4.041967018052706e-07, "loss": 0.08613032102584839, "step": 34620 }, { "epoch": 0.9122069916991271, "grad_norm": 3.4931671619415283, "learning_rate": 4.0179916840282776e-07, "loss": 0.07631266117095947, "step": 34630 }, { "epoch": 0.9124704069436258, "grad_norm": 2.9332387447357178, "learning_rate": 3.9940862098505117e-07, "loss": 0.07311130166053773, "step": 34640 }, { "epoch": 0.9127338221881246, "grad_norm": 4.795865058898926, "learning_rate": 3.970250612918858e-07, "loss": 0.09792664647102356, "step": 34650 }, { "epoch": 0.9129972374326233, "grad_norm": 3.3335072994232178, "learning_rate": 3.946484910581871e-07, "loss": 0.07830094695091247, "step": 34660 }, { "epoch": 0.9132606526771221, "grad_norm": 3.6562342643737793, "learning_rate": 3.9227891201372827e-07, "loss": 0.07793920040130616, "step": 34670 }, { "epoch": 0.9135240679216208, "grad_norm": 4.062324523925781, "learning_rate": 3.899163258831895e-07, "loss": 0.08973205089569092, "step": 34680 }, { "epoch": 0.9137874831661196, "grad_norm": 2.913686513900757, "learning_rate": 3.8756073438616203e-07, "loss": 0.08321585655212402, "step": 34690 }, { "epoch": 0.9140508984106183, "grad_norm": 2.940764904022217, "learning_rate": 3.8521213923714775e-07, "loss": 0.0883119523525238, "step": 34700 }, { "epoch": 0.914314313655117, "grad_norm": 2.479753255844116, "learning_rate": 3.8287054214555694e-07, "loss": 0.0725214958190918, "step": 34710 }, { "epoch": 0.9145777288996157, "grad_norm": 2.3723669052124023, "learning_rate": 3.805359448157031e-07, "loss": 0.07924695611000061, "step": 34720 }, { "epoch": 0.9148411441441144, "grad_norm": 2.673440456390381, "learning_rate": 3.78208348946808e-07, "loss": 0.07786771655082703, "step": 34730 }, { "epoch": 0.9151045593886132, "grad_norm": 3.4135947227478027, "learning_rate": 3.7588775623299433e-07, "loss": 0.08614643216133118, "step": 34740 }, { "epoch": 0.9153679746331119, "grad_norm": 3.0617635250091553, "learning_rate": 3.735741683632932e-07, "loss": 0.08145022392272949, "step": 34750 }, { "epoch": 0.9156313898776107, "grad_norm": 2.2422149181365967, "learning_rate": 3.712675870216309e-07, "loss": 0.07848646640777587, "step": 34760 }, { "epoch": 0.9158948051221094, "grad_norm": 4.117730617523193, "learning_rate": 3.689680138868379e-07, "loss": 0.08356014490127564, "step": 34770 }, { "epoch": 0.9161582203666082, "grad_norm": 2.6450095176696777, "learning_rate": 3.66675450632642e-07, "loss": 0.07291778326034545, "step": 34780 }, { "epoch": 0.9164216356111069, "grad_norm": 3.702530860900879, "learning_rate": 3.643898989276684e-07, "loss": 0.09753494262695313, "step": 34790 }, { "epoch": 0.9166850508556056, "grad_norm": 4.208644390106201, "learning_rate": 3.6211136043544425e-07, "loss": 0.0768026888370514, "step": 34800 }, { "epoch": 0.9169484661001044, "grad_norm": 4.414900302886963, "learning_rate": 3.598398368143863e-07, "loss": 0.08935600519180298, "step": 34810 }, { "epoch": 0.9172118813446031, "grad_norm": 2.382289409637451, "learning_rate": 3.575753297178053e-07, "loss": 0.07582857608795165, "step": 34820 }, { "epoch": 0.9174752965891019, "grad_norm": 3.526064395904541, "learning_rate": 3.553178407939095e-07, "loss": 0.08213526010513306, "step": 34830 }, { "epoch": 0.9177387118336006, "grad_norm": 2.4225335121154785, "learning_rate": 3.5306737168579685e-07, "loss": 0.0753321886062622, "step": 34840 }, { "epoch": 0.9180021270780994, "grad_norm": 3.159782886505127, "learning_rate": 3.5082392403145373e-07, "loss": 0.08387211561203003, "step": 34850 }, { "epoch": 0.918265542322598, "grad_norm": 2.9781694412231445, "learning_rate": 3.485874994637606e-07, "loss": 0.08767216205596924, "step": 34860 }, { "epoch": 0.9185289575670968, "grad_norm": 2.8042893409729004, "learning_rate": 3.463580996104821e-07, "loss": 0.08773433566093444, "step": 34870 }, { "epoch": 0.9187923728115955, "grad_norm": 3.52388334274292, "learning_rate": 3.4413572609427235e-07, "loss": 0.0834645688533783, "step": 34880 }, { "epoch": 0.9190557880560942, "grad_norm": 4.163448333740234, "learning_rate": 3.4192038053267095e-07, "loss": 0.08666800260543824, "step": 34890 }, { "epoch": 0.919319203300593, "grad_norm": 3.303874969482422, "learning_rate": 3.3971206453810133e-07, "loss": 0.07679834961891174, "step": 34900 }, { "epoch": 0.9195826185450917, "grad_norm": 3.4751601219177246, "learning_rate": 3.3751077971787115e-07, "loss": 0.0759615182876587, "step": 34910 }, { "epoch": 0.9198460337895905, "grad_norm": 3.013002872467041, "learning_rate": 3.353165276741699e-07, "loss": 0.0781369686126709, "step": 34920 }, { "epoch": 0.9201094490340892, "grad_norm": 3.6066360473632812, "learning_rate": 3.331293100040711e-07, "loss": 0.08583108186721802, "step": 34930 }, { "epoch": 0.920372864278588, "grad_norm": 4.178858757019043, "learning_rate": 3.3094912829952477e-07, "loss": 0.08738164901733399, "step": 34940 }, { "epoch": 0.9206362795230867, "grad_norm": 3.5088565349578857, "learning_rate": 3.2877598414736033e-07, "loss": 0.0819533109664917, "step": 34950 }, { "epoch": 0.9208996947675855, "grad_norm": 3.702573537826538, "learning_rate": 3.266098791292871e-07, "loss": 0.0861507773399353, "step": 34960 }, { "epoch": 0.9211631100120842, "grad_norm": 2.9307031631469727, "learning_rate": 3.244508148218895e-07, "loss": 0.07113102674484253, "step": 34970 }, { "epoch": 0.9214265252565829, "grad_norm": 3.7733571529388428, "learning_rate": 3.22298792796627e-07, "loss": 0.08451238870620728, "step": 34980 }, { "epoch": 0.9216899405010817, "grad_norm": 2.9650747776031494, "learning_rate": 3.2015381461983577e-07, "loss": 0.08986079692840576, "step": 34990 }, { "epoch": 0.9219533557455803, "grad_norm": 2.901705741882324, "learning_rate": 3.1801588185272016e-07, "loss": 0.07863502502441407, "step": 35000 }, { "epoch": 0.9222167709900791, "grad_norm": 3.4274184703826904, "learning_rate": 3.1588499605136435e-07, "loss": 0.08521322011947632, "step": 35010 }, { "epoch": 0.9224801862345778, "grad_norm": 3.4532361030578613, "learning_rate": 3.1376115876671553e-07, "loss": 0.08930236101150513, "step": 35020 }, { "epoch": 0.9227436014790766, "grad_norm": 2.677497148513794, "learning_rate": 3.1164437154459494e-07, "loss": 0.07726163864135742, "step": 35030 }, { "epoch": 0.9230070167235753, "grad_norm": 4.6823039054870605, "learning_rate": 3.0953463592569343e-07, "loss": 0.07625153660774231, "step": 35040 }, { "epoch": 0.9232704319680741, "grad_norm": 3.783653736114502, "learning_rate": 3.074319534455639e-07, "loss": 0.08540564775466919, "step": 35050 }, { "epoch": 0.9235338472125728, "grad_norm": 2.891902208328247, "learning_rate": 3.053363256346331e-07, "loss": 0.0885211706161499, "step": 35060 }, { "epoch": 0.9237972624570715, "grad_norm": 5.231968402862549, "learning_rate": 3.0324775401818884e-07, "loss": 0.08330981731414795, "step": 35070 }, { "epoch": 0.9240606777015703, "grad_norm": 3.8283002376556396, "learning_rate": 3.0116624011637953e-07, "loss": 0.08966223001480103, "step": 35080 }, { "epoch": 0.924324092946069, "grad_norm": 2.9097790718078613, "learning_rate": 2.9909178544422656e-07, "loss": 0.0718967616558075, "step": 35090 }, { "epoch": 0.9245875081905678, "grad_norm": 3.3116655349731445, "learning_rate": 2.9702439151160334e-07, "loss": 0.08450562953948974, "step": 35100 }, { "epoch": 0.9248509234350665, "grad_norm": 3.7638099193573, "learning_rate": 2.949640598232517e-07, "loss": 0.07570406198501586, "step": 35110 }, { "epoch": 0.9251143386795653, "grad_norm": 4.373971939086914, "learning_rate": 2.9291079187876656e-07, "loss": 0.08542313575744628, "step": 35120 }, { "epoch": 0.925377753924064, "grad_norm": 3.9106688499450684, "learning_rate": 2.9086458917260585e-07, "loss": 0.08136407136917115, "step": 35130 }, { "epoch": 0.9256411691685628, "grad_norm": 3.7996585369110107, "learning_rate": 2.8882545319408706e-07, "loss": 0.07880950570106507, "step": 35140 }, { "epoch": 0.9259045844130614, "grad_norm": 3.9594037532806396, "learning_rate": 2.8679338542737745e-07, "loss": 0.08399271965026855, "step": 35150 }, { "epoch": 0.9261679996575601, "grad_norm": 5.401860237121582, "learning_rate": 2.8476838735150613e-07, "loss": 0.08326495289802552, "step": 35160 }, { "epoch": 0.9264314149020589, "grad_norm": 3.678445339202881, "learning_rate": 2.8275046044035304e-07, "loss": 0.08300691843032837, "step": 35170 }, { "epoch": 0.9266948301465576, "grad_norm": 2.8148300647735596, "learning_rate": 2.8073960616265215e-07, "loss": 0.08833566904067994, "step": 35180 }, { "epoch": 0.9269582453910564, "grad_norm": 4.183376312255859, "learning_rate": 2.7873582598199056e-07, "loss": 0.09223508238792419, "step": 35190 }, { "epoch": 0.9272216606355551, "grad_norm": 2.656059503555298, "learning_rate": 2.7673912135680714e-07, "loss": 0.06255797743797302, "step": 35200 }, { "epoch": 0.9274850758800539, "grad_norm": 3.287139654159546, "learning_rate": 2.7474949374038717e-07, "loss": 0.072490394115448, "step": 35210 }, { "epoch": 0.9277484911245526, "grad_norm": 2.979581117630005, "learning_rate": 2.7276694458086894e-07, "loss": 0.08339035511016846, "step": 35220 }, { "epoch": 0.9280119063690514, "grad_norm": 3.223867177963257, "learning_rate": 2.7079147532123704e-07, "loss": 0.08633075356483459, "step": 35230 }, { "epoch": 0.9282753216135501, "grad_norm": 2.7584614753723145, "learning_rate": 2.688230873993225e-07, "loss": 0.0661958634853363, "step": 35240 }, { "epoch": 0.9285387368580488, "grad_norm": 3.5630762577056885, "learning_rate": 2.6686178224780367e-07, "loss": 0.07255240678787231, "step": 35250 }, { "epoch": 0.9288021521025476, "grad_norm": 3.2665438652038574, "learning_rate": 2.6490756129420313e-07, "loss": 0.08207308053970337, "step": 35260 }, { "epoch": 0.9290655673470463, "grad_norm": 3.89489483833313, "learning_rate": 2.6296042596088643e-07, "loss": 0.08328218460083008, "step": 35270 }, { "epoch": 0.9293289825915451, "grad_norm": 3.529517889022827, "learning_rate": 2.610203776650655e-07, "loss": 0.08588662147521972, "step": 35280 }, { "epoch": 0.9295923978360437, "grad_norm": 4.171391487121582, "learning_rate": 2.5908741781878744e-07, "loss": 0.0692150354385376, "step": 35290 }, { "epoch": 0.9298558130805425, "grad_norm": 3.0706098079681396, "learning_rate": 2.5716154782894573e-07, "loss": 0.07974791526794434, "step": 35300 }, { "epoch": 0.9301192283250412, "grad_norm": 4.568943977355957, "learning_rate": 2.552427690972703e-07, "loss": 0.07646337747573853, "step": 35310 }, { "epoch": 0.9303826435695399, "grad_norm": 3.559438467025757, "learning_rate": 2.5333108302033394e-07, "loss": 0.07724981307983399, "step": 35320 }, { "epoch": 0.9306460588140387, "grad_norm": 3.4691405296325684, "learning_rate": 2.514264909895414e-07, "loss": 0.08535066843032837, "step": 35330 }, { "epoch": 0.9309094740585374, "grad_norm": 3.0367884635925293, "learning_rate": 2.495289943911361e-07, "loss": 0.08193172812461853, "step": 35340 }, { "epoch": 0.9311728893030362, "grad_norm": 3.584326982498169, "learning_rate": 2.476385946062021e-07, "loss": 0.08309050798416137, "step": 35350 }, { "epoch": 0.9314363045475349, "grad_norm": 3.8764662742614746, "learning_rate": 2.4575529301065104e-07, "loss": 0.07844210267066956, "step": 35360 }, { "epoch": 0.9316997197920337, "grad_norm": 3.0842337608337402, "learning_rate": 2.43879090975232e-07, "loss": 0.08860675096511841, "step": 35370 }, { "epoch": 0.9319631350365324, "grad_norm": 3.7807631492614746, "learning_rate": 2.4200998986552484e-07, "loss": 0.08189990520477294, "step": 35380 }, { "epoch": 0.9322265502810312, "grad_norm": 4.938887119293213, "learning_rate": 2.401479910419413e-07, "loss": 0.08305638432502746, "step": 35390 }, { "epoch": 0.9324899655255299, "grad_norm": 2.7442314624786377, "learning_rate": 2.3829309585972738e-07, "loss": 0.08595767617225647, "step": 35400 }, { "epoch": 0.9327533807700286, "grad_norm": 4.373393535614014, "learning_rate": 2.364453056689553e-07, "loss": 0.08209512233734131, "step": 35410 }, { "epoch": 0.9330167960145274, "grad_norm": 3.6009581089019775, "learning_rate": 2.3460462181452258e-07, "loss": 0.07746945023536682, "step": 35420 }, { "epoch": 0.933280211259026, "grad_norm": 2.5131962299346924, "learning_rate": 2.3277104563616205e-07, "loss": 0.08125864267349243, "step": 35430 }, { "epoch": 0.9335436265035248, "grad_norm": 3.0173494815826416, "learning_rate": 2.3094457846842943e-07, "loss": 0.08011082410812378, "step": 35440 }, { "epoch": 0.9338070417480235, "grad_norm": 4.693338871002197, "learning_rate": 2.291252216407047e-07, "loss": 0.07068297863006592, "step": 35450 }, { "epoch": 0.9340704569925223, "grad_norm": 3.823984146118164, "learning_rate": 2.273129764771953e-07, "loss": 0.07269293069839478, "step": 35460 }, { "epoch": 0.934333872237021, "grad_norm": 4.133332252502441, "learning_rate": 2.2550784429693052e-07, "loss": 0.08249006271362305, "step": 35470 }, { "epoch": 0.9345972874815198, "grad_norm": 4.361036777496338, "learning_rate": 2.2370982641376494e-07, "loss": 0.06637172102928161, "step": 35480 }, { "epoch": 0.9348607027260185, "grad_norm": 4.7470221519470215, "learning_rate": 2.2191892413637173e-07, "loss": 0.07679241895675659, "step": 35490 }, { "epoch": 0.9351241179705172, "grad_norm": 3.577810049057007, "learning_rate": 2.2013513876824933e-07, "loss": 0.07599198818206787, "step": 35500 }, { "epoch": 0.935387533215016, "grad_norm": 2.850602865219116, "learning_rate": 2.183584716077114e-07, "loss": 0.08530403971672058, "step": 35510 }, { "epoch": 0.9356509484595147, "grad_norm": 3.306650161743164, "learning_rate": 2.1658892394789244e-07, "loss": 0.07656578421592712, "step": 35520 }, { "epoch": 0.9359143637040135, "grad_norm": 4.234335422515869, "learning_rate": 2.1482649707674775e-07, "loss": 0.0783177673816681, "step": 35530 }, { "epoch": 0.9361777789485122, "grad_norm": 2.6654672622680664, "learning_rate": 2.13071192277049e-07, "loss": 0.07320937514305115, "step": 35540 }, { "epoch": 0.936441194193011, "grad_norm": 2.767101287841797, "learning_rate": 2.1132301082637752e-07, "loss": 0.07153128385543824, "step": 35550 }, { "epoch": 0.9367046094375097, "grad_norm": 3.270897150039673, "learning_rate": 2.0958195399713888e-07, "loss": 0.08088930249214173, "step": 35560 }, { "epoch": 0.9369680246820085, "grad_norm": 3.7242259979248047, "learning_rate": 2.0784802305654938e-07, "loss": 0.07651939392089843, "step": 35570 }, { "epoch": 0.9372314399265071, "grad_norm": 4.117777347564697, "learning_rate": 2.061212192666373e-07, "loss": 0.0813284158706665, "step": 35580 }, { "epoch": 0.9374948551710058, "grad_norm": 3.4074769020080566, "learning_rate": 2.0440154388424616e-07, "loss": 0.07720180749893188, "step": 35590 }, { "epoch": 0.9377582704155046, "grad_norm": 2.351501226425171, "learning_rate": 2.0268899816102916e-07, "loss": 0.07074620723724365, "step": 35600 }, { "epoch": 0.9380216856600033, "grad_norm": 3.3080084323883057, "learning_rate": 2.0098358334345147e-07, "loss": 0.07449942231178283, "step": 35610 }, { "epoch": 0.9382851009045021, "grad_norm": 4.0300822257995605, "learning_rate": 1.9928530067278684e-07, "loss": 0.07755244374275208, "step": 35620 }, { "epoch": 0.9385485161490008, "grad_norm": 3.1490063667297363, "learning_rate": 1.9759415138511984e-07, "loss": 0.0822323739528656, "step": 35630 }, { "epoch": 0.9388119313934996, "grad_norm": 2.938828945159912, "learning_rate": 1.959101367113414e-07, "loss": 0.08934667110443115, "step": 35640 }, { "epoch": 0.9390753466379983, "grad_norm": 2.7892751693725586, "learning_rate": 1.9423325787714774e-07, "loss": 0.08018579483032226, "step": 35650 }, { "epoch": 0.9393387618824971, "grad_norm": 3.1887128353118896, "learning_rate": 1.9256351610304814e-07, "loss": 0.08020989298820495, "step": 35660 }, { "epoch": 0.9396021771269958, "grad_norm": 2.6608972549438477, "learning_rate": 1.909009126043504e-07, "loss": 0.06912118792533875, "step": 35670 }, { "epoch": 0.9398655923714945, "grad_norm": 3.353555679321289, "learning_rate": 1.8924544859116656e-07, "loss": 0.08753131628036499, "step": 35680 }, { "epoch": 0.9401290076159933, "grad_norm": 3.919297456741333, "learning_rate": 1.8759712526841943e-07, "loss": 0.08855857849121093, "step": 35690 }, { "epoch": 0.940392422860492, "grad_norm": 4.091468811035156, "learning_rate": 1.8595594383582605e-07, "loss": 0.08229383826255798, "step": 35700 }, { "epoch": 0.9406558381049908, "grad_norm": 3.471830368041992, "learning_rate": 1.8432190548791197e-07, "loss": 0.08775599598884583, "step": 35710 }, { "epoch": 0.9409192533494894, "grad_norm": 3.0250978469848633, "learning_rate": 1.8269501141399915e-07, "loss": 0.08047012090682984, "step": 35720 }, { "epoch": 0.9411826685939882, "grad_norm": 3.4985592365264893, "learning_rate": 1.8107526279821041e-07, "loss": 0.0831346869468689, "step": 35730 }, { "epoch": 0.9414460838384869, "grad_norm": 3.2898471355438232, "learning_rate": 1.794626608194705e-07, "loss": 0.08591846227645875, "step": 35740 }, { "epoch": 0.9417094990829857, "grad_norm": 2.376610040664673, "learning_rate": 1.7785720665150054e-07, "loss": 0.0830851674079895, "step": 35750 }, { "epoch": 0.9419729143274844, "grad_norm": 3.592801809310913, "learning_rate": 1.7625890146281798e-07, "loss": 0.08124594688415528, "step": 35760 }, { "epoch": 0.9422363295719831, "grad_norm": 2.1783204078674316, "learning_rate": 1.7466774641674012e-07, "loss": 0.0724498987197876, "step": 35770 }, { "epoch": 0.9424997448164819, "grad_norm": 2.8570406436920166, "learning_rate": 1.7308374267137606e-07, "loss": 0.06992928981781006, "step": 35780 }, { "epoch": 0.9427631600609806, "grad_norm": 3.338733434677124, "learning_rate": 1.715068913796336e-07, "loss": 0.0845819354057312, "step": 35790 }, { "epoch": 0.9430265753054794, "grad_norm": 3.054316282272339, "learning_rate": 1.6993719368921468e-07, "loss": 0.08302334547042847, "step": 35800 }, { "epoch": 0.9432899905499781, "grad_norm": 3.110469341278076, "learning_rate": 1.683746507426087e-07, "loss": 0.0710926592350006, "step": 35810 }, { "epoch": 0.9435534057944769, "grad_norm": 3.6836814880371094, "learning_rate": 1.66819263677106e-07, "loss": 0.0866852879524231, "step": 35820 }, { "epoch": 0.9438168210389756, "grad_norm": 3.089726686477661, "learning_rate": 1.6527103362478202e-07, "loss": 0.08793244361877442, "step": 35830 }, { "epoch": 0.9440802362834743, "grad_norm": 2.3603460788726807, "learning_rate": 1.6372996171250544e-07, "loss": 0.08573151230812073, "step": 35840 }, { "epoch": 0.9443436515279731, "grad_norm": 2.9334616661071777, "learning_rate": 1.6219604906193677e-07, "loss": 0.0712968349456787, "step": 35850 }, { "epoch": 0.9446070667724717, "grad_norm": 3.7206623554229736, "learning_rate": 1.606692967895207e-07, "loss": 0.085934317111969, "step": 35860 }, { "epoch": 0.9448704820169705, "grad_norm": 3.266435384750366, "learning_rate": 1.5914970600649615e-07, "loss": 0.09974340796470642, "step": 35870 }, { "epoch": 0.9451338972614692, "grad_norm": 3.095750570297241, "learning_rate": 1.5763727781888726e-07, "loss": 0.07222499847412109, "step": 35880 }, { "epoch": 0.945397312505968, "grad_norm": 2.587693929672241, "learning_rate": 1.5613201332750127e-07, "loss": 0.07665959000587463, "step": 35890 }, { "epoch": 0.9456607277504667, "grad_norm": 3.726127862930298, "learning_rate": 1.5463391362793735e-07, "loss": 0.0788551151752472, "step": 35900 }, { "epoch": 0.9459241429949655, "grad_norm": 2.8302395343780518, "learning_rate": 1.5314297981057658e-07, "loss": 0.08063291311264038, "step": 35910 }, { "epoch": 0.9461875582394642, "grad_norm": 2.055081367492676, "learning_rate": 1.5165921296058427e-07, "loss": 0.077357017993927, "step": 35920 }, { "epoch": 0.9464509734839629, "grad_norm": 2.3219265937805176, "learning_rate": 1.501826141579099e-07, "loss": 0.06661049127578736, "step": 35930 }, { "epoch": 0.9467143887284617, "grad_norm": 2.8919918537139893, "learning_rate": 1.48713184477286e-07, "loss": 0.0788213551044464, "step": 35940 }, { "epoch": 0.9469778039729604, "grad_norm": 4.590354919433594, "learning_rate": 1.4725092498822814e-07, "loss": 0.08098058700561524, "step": 35950 }, { "epoch": 0.9472412192174592, "grad_norm": 3.0350377559661865, "learning_rate": 1.457958367550305e-07, "loss": 0.08318697214126587, "step": 35960 }, { "epoch": 0.9475046344619579, "grad_norm": 2.5546388626098633, "learning_rate": 1.443479208367693e-07, "loss": 0.07231478691101074, "step": 35970 }, { "epoch": 0.9477680497064567, "grad_norm": 3.9915175437927246, "learning_rate": 1.4290717828730037e-07, "loss": 0.0735990822315216, "step": 35980 }, { "epoch": 0.9480314649509554, "grad_norm": 3.7617194652557373, "learning_rate": 1.4147361015525717e-07, "loss": 0.0864393949508667, "step": 35990 }, { "epoch": 0.9482948801954542, "grad_norm": 3.703019857406616, "learning_rate": 1.4004721748405392e-07, "loss": 0.0926012635231018, "step": 36000 }, { "epoch": 0.9485582954399528, "grad_norm": 3.36833119392395, "learning_rate": 1.3862800131188237e-07, "loss": 0.08158743381500244, "step": 36010 }, { "epoch": 0.9488217106844515, "grad_norm": 2.9328463077545166, "learning_rate": 1.37215962671704e-07, "loss": 0.08166238069534301, "step": 36020 }, { "epoch": 0.9490851259289503, "grad_norm": 3.5359408855438232, "learning_rate": 1.3581110259126674e-07, "loss": 0.08173725008964539, "step": 36030 }, { "epoch": 0.949348541173449, "grad_norm": 3.8088860511779785, "learning_rate": 1.3441342209308595e-07, "loss": 0.07788649797439576, "step": 36040 }, { "epoch": 0.9496119564179478, "grad_norm": 5.3692426681518555, "learning_rate": 1.3302292219445457e-07, "loss": 0.08170855641365052, "step": 36050 }, { "epoch": 0.9498753716624465, "grad_norm": 2.820491313934326, "learning_rate": 1.3163960390743746e-07, "loss": 0.0764153778553009, "step": 36060 }, { "epoch": 0.9501387869069453, "grad_norm": 3.6313254833221436, "learning_rate": 1.302634682388726e-07, "loss": 0.07287755012512206, "step": 36070 }, { "epoch": 0.950402202151444, "grad_norm": 1.8859241008758545, "learning_rate": 1.2889451619037318e-07, "loss": 0.07605007886886597, "step": 36080 }, { "epoch": 0.9506656173959428, "grad_norm": 4.381514072418213, "learning_rate": 1.2753274875832112e-07, "loss": 0.08243867754936218, "step": 36090 }, { "epoch": 0.9509290326404415, "grad_norm": 2.4846017360687256, "learning_rate": 1.2617816693386907e-07, "loss": 0.08059709072113037, "step": 36100 }, { "epoch": 0.9511924478849402, "grad_norm": 3.6432600021362305, "learning_rate": 1.2483077170293954e-07, "loss": 0.08806056976318359, "step": 36110 }, { "epoch": 0.951455863129439, "grad_norm": 3.725806951522827, "learning_rate": 1.2349056404622472e-07, "loss": 0.077144193649292, "step": 36120 }, { "epoch": 0.9517192783739377, "grad_norm": 2.486477851867676, "learning_rate": 1.2215754493918763e-07, "loss": 0.08165484666824341, "step": 36130 }, { "epoch": 0.9519826936184365, "grad_norm": 2.8795220851898193, "learning_rate": 1.208317153520555e-07, "loss": 0.09024041891098022, "step": 36140 }, { "epoch": 0.9522461088629351, "grad_norm": 3.9442830085754395, "learning_rate": 1.195130762498231e-07, "loss": 0.07634541392326355, "step": 36150 }, { "epoch": 0.952509524107434, "grad_norm": 3.9262337684631348, "learning_rate": 1.1820162859225493e-07, "loss": 0.07660925388336182, "step": 36160 }, { "epoch": 0.9527729393519326, "grad_norm": 2.3106112480163574, "learning_rate": 1.1689737333387852e-07, "loss": 0.07074901461601257, "step": 36170 }, { "epoch": 0.9530363545964314, "grad_norm": 3.6926963329315186, "learning_rate": 1.1560031142398787e-07, "loss": 0.07507573366165161, "step": 36180 }, { "epoch": 0.9532997698409301, "grad_norm": 4.069038391113281, "learning_rate": 1.1431044380663891e-07, "loss": 0.09243282675743103, "step": 36190 }, { "epoch": 0.9535631850854288, "grad_norm": 3.346508026123047, "learning_rate": 1.1302777142065402e-07, "loss": 0.08197695016860962, "step": 36200 }, { "epoch": 0.9538266003299276, "grad_norm": 3.0777969360351562, "learning_rate": 1.117522951996175e-07, "loss": 0.08153185844421387, "step": 36210 }, { "epoch": 0.9540900155744263, "grad_norm": 3.8561391830444336, "learning_rate": 1.1048401607187565e-07, "loss": 0.08188385963439941, "step": 36220 }, { "epoch": 0.9543534308189251, "grad_norm": 3.2981224060058594, "learning_rate": 1.0922293496053671e-07, "loss": 0.0946079134941101, "step": 36230 }, { "epoch": 0.9546168460634238, "grad_norm": 3.438758373260498, "learning_rate": 1.0796905278347092e-07, "loss": 0.08123475313186646, "step": 36240 }, { "epoch": 0.9548802613079226, "grad_norm": 3.1398608684539795, "learning_rate": 1.0672237045330491e-07, "loss": 0.07501614093780518, "step": 36250 }, { "epoch": 0.9551436765524213, "grad_norm": 4.078083515167236, "learning_rate": 1.0548288887743174e-07, "loss": 0.07005045413970948, "step": 36260 }, { "epoch": 0.9554070917969201, "grad_norm": 3.430208921432495, "learning_rate": 1.0425060895799865e-07, "loss": 0.07699830532073974, "step": 36270 }, { "epoch": 0.9556705070414188, "grad_norm": 4.287264823913574, "learning_rate": 1.0302553159190932e-07, "loss": 0.08675129413604736, "step": 36280 }, { "epoch": 0.9559339222859174, "grad_norm": 3.5384368896484375, "learning_rate": 1.0180765767083045e-07, "loss": 0.08632760047912598, "step": 36290 }, { "epoch": 0.9561973375304162, "grad_norm": 2.9397473335266113, "learning_rate": 1.00596988081183e-07, "loss": 0.07309897541999817, "step": 36300 }, { "epoch": 0.9564607527749149, "grad_norm": 3.0770065784454346, "learning_rate": 9.939352370414435e-08, "loss": 0.07312554717063904, "step": 36310 }, { "epoch": 0.9567241680194137, "grad_norm": 2.5066728591918945, "learning_rate": 9.819726541564711e-08, "loss": 0.08234319686889649, "step": 36320 }, { "epoch": 0.9569875832639124, "grad_norm": 3.322861433029175, "learning_rate": 9.700821408638039e-08, "loss": 0.07578976154327392, "step": 36330 }, { "epoch": 0.9572509985084112, "grad_norm": 2.3848307132720947, "learning_rate": 9.582637058178635e-08, "loss": 0.0769990086555481, "step": 36340 }, { "epoch": 0.9575144137529099, "grad_norm": 3.131087303161621, "learning_rate": 9.465173576206355e-08, "loss": 0.07628765106201171, "step": 36350 }, { "epoch": 0.9577778289974086, "grad_norm": 2.852261543273926, "learning_rate": 9.348431048215923e-08, "loss": 0.0847012996673584, "step": 36360 }, { "epoch": 0.9580412442419074, "grad_norm": 4.168527603149414, "learning_rate": 9.232409559177701e-08, "loss": 0.08691462874412537, "step": 36370 }, { "epoch": 0.9583046594864061, "grad_norm": 3.445667028427124, "learning_rate": 9.117109193537254e-08, "loss": 0.08627705574035645, "step": 36380 }, { "epoch": 0.9585680747309049, "grad_norm": 2.528339385986328, "learning_rate": 9.002530035215007e-08, "loss": 0.07553990483283997, "step": 36390 }, { "epoch": 0.9588314899754036, "grad_norm": 3.0290210247039795, "learning_rate": 8.888672167606693e-08, "loss": 0.07539713382720947, "step": 36400 }, { "epoch": 0.9590949052199024, "grad_norm": 3.5532350540161133, "learning_rate": 8.775535673582913e-08, "loss": 0.07393468022346497, "step": 36410 }, { "epoch": 0.9593583204644011, "grad_norm": 3.272132396697998, "learning_rate": 8.663120635489241e-08, "loss": 0.06979392170906067, "step": 36420 }, { "epoch": 0.9596217357088999, "grad_norm": 3.333536148071289, "learning_rate": 8.55142713514645e-08, "loss": 0.08329839706420898, "step": 36430 }, { "epoch": 0.9598851509533985, "grad_norm": 3.388298988342285, "learning_rate": 8.440455253849622e-08, "loss": 0.07924200296401977, "step": 36440 }, { "epoch": 0.9601485661978972, "grad_norm": 2.502913475036621, "learning_rate": 8.330205072369035e-08, "loss": 0.07855167388916015, "step": 36450 }, { "epoch": 0.960411981442396, "grad_norm": 3.7618792057037354, "learning_rate": 8.22067667094939e-08, "loss": 0.08702921867370605, "step": 36460 }, { "epoch": 0.9606753966868947, "grad_norm": 2.6357014179229736, "learning_rate": 8.11187012931025e-08, "loss": 0.07891066074371338, "step": 36470 }, { "epoch": 0.9609388119313935, "grad_norm": 3.621796131134033, "learning_rate": 8.003785526645824e-08, "loss": 0.08091295957565307, "step": 36480 }, { "epoch": 0.9612022271758922, "grad_norm": 4.393803119659424, "learning_rate": 7.896422941624626e-08, "loss": 0.0785364031791687, "step": 36490 }, { "epoch": 0.961465642420391, "grad_norm": 2.8782379627227783, "learning_rate": 7.789782452389705e-08, "loss": 0.08258590698242188, "step": 36500 }, { "epoch": 0.9617290576648897, "grad_norm": 3.3955063819885254, "learning_rate": 7.683864136558749e-08, "loss": 0.07894209027290344, "step": 36510 }, { "epoch": 0.9619924729093885, "grad_norm": 4.02926778793335, "learning_rate": 7.578668071223649e-08, "loss": 0.08180314302444458, "step": 36520 }, { "epoch": 0.9622558881538872, "grad_norm": 3.34255313873291, "learning_rate": 7.474194332950602e-08, "loss": 0.08218361139297485, "step": 36530 }, { "epoch": 0.9625193033983859, "grad_norm": 2.77030873298645, "learning_rate": 7.370442997780114e-08, "loss": 0.07246441841125488, "step": 36540 }, { "epoch": 0.9627827186428847, "grad_norm": 2.890333652496338, "learning_rate": 7.267414141227002e-08, "loss": 0.06803547143936158, "step": 36550 }, { "epoch": 0.9630461338873834, "grad_norm": 2.9208192825317383, "learning_rate": 7.165107838279838e-08, "loss": 0.08337743282318115, "step": 36560 }, { "epoch": 0.9633095491318822, "grad_norm": 3.1270577907562256, "learning_rate": 7.063524163401836e-08, "loss": 0.08284040689468383, "step": 36570 }, { "epoch": 0.9635729643763808, "grad_norm": 3.690291166305542, "learning_rate": 6.962663190529851e-08, "loss": 0.08348687887191772, "step": 36580 }, { "epoch": 0.9638363796208796, "grad_norm": 3.127641201019287, "learning_rate": 6.862524993074825e-08, "loss": 0.07167301177978516, "step": 36590 }, { "epoch": 0.9640997948653783, "grad_norm": 4.8933186531066895, "learning_rate": 6.763109643921683e-08, "loss": 0.07865029573440552, "step": 36600 }, { "epoch": 0.9643632101098771, "grad_norm": 3.741194009780884, "learning_rate": 6.664417215429319e-08, "loss": 0.08897361755371094, "step": 36610 }, { "epoch": 0.9646266253543758, "grad_norm": 3.263190507888794, "learning_rate": 6.566447779430163e-08, "loss": 0.08241435289382934, "step": 36620 }, { "epoch": 0.9648900405988745, "grad_norm": 2.7277228832244873, "learning_rate": 6.469201407230729e-08, "loss": 0.08214519023895264, "step": 36630 }, { "epoch": 0.9651534558433733, "grad_norm": 3.445108652114868, "learning_rate": 6.372678169611068e-08, "loss": 0.07352313995361329, "step": 36640 }, { "epoch": 0.965416871087872, "grad_norm": 3.450192451477051, "learning_rate": 6.276878136824982e-08, "loss": 0.07435155510902405, "step": 36650 }, { "epoch": 0.9656802863323708, "grad_norm": 3.130492687225342, "learning_rate": 6.181801378599805e-08, "loss": 0.07789390087127686, "step": 36660 }, { "epoch": 0.9659437015768695, "grad_norm": 3.8167169094085693, "learning_rate": 6.087447964136517e-08, "loss": 0.07404811382293701, "step": 36670 }, { "epoch": 0.9662071168213683, "grad_norm": 4.552938938140869, "learning_rate": 5.993817962109627e-08, "loss": 0.09162539839744568, "step": 36680 }, { "epoch": 0.966470532065867, "grad_norm": 4.183182716369629, "learning_rate": 5.9009114406671787e-08, "loss": 0.076705002784729, "step": 36690 }, { "epoch": 0.9667339473103658, "grad_norm": 4.834007740020752, "learning_rate": 5.8087284674304136e-08, "loss": 0.08301629424095154, "step": 36700 }, { "epoch": 0.9669973625548645, "grad_norm": 3.182088851928711, "learning_rate": 5.7172691094942165e-08, "loss": 0.07280011177062988, "step": 36710 }, { "epoch": 0.9672607777993631, "grad_norm": 3.583500385284424, "learning_rate": 5.626533433426451e-08, "loss": 0.08204436898231507, "step": 36720 }, { "epoch": 0.967524193043862, "grad_norm": 3.486598014831543, "learning_rate": 5.5365215052686216e-08, "loss": 0.0787355661392212, "step": 36730 }, { "epoch": 0.9677876082883606, "grad_norm": 3.4781558513641357, "learning_rate": 5.4472333905352115e-08, "loss": 0.07419096827507018, "step": 36740 }, { "epoch": 0.9680510235328594, "grad_norm": 2.8844311237335205, "learning_rate": 5.3586691542140135e-08, "loss": 0.06630986332893371, "step": 36750 }, { "epoch": 0.9683144387773581, "grad_norm": 3.165076971054077, "learning_rate": 5.2708288607659084e-08, "loss": 0.061668145656585696, "step": 36760 }, { "epoch": 0.9685778540218569, "grad_norm": 3.01873779296875, "learning_rate": 5.183712574124755e-08, "loss": 0.0767469048500061, "step": 36770 }, { "epoch": 0.9688412692663556, "grad_norm": 3.9430813789367676, "learning_rate": 5.0973203576976104e-08, "loss": 0.07701430320739747, "step": 36780 }, { "epoch": 0.9691046845108544, "grad_norm": 3.1871585845947266, "learning_rate": 5.011652274364509e-08, "loss": 0.073766028881073, "step": 36790 }, { "epoch": 0.9693680997553531, "grad_norm": 4.148520469665527, "learning_rate": 4.926708386478352e-08, "loss": 0.07842659950256348, "step": 36800 }, { "epoch": 0.9696315149998518, "grad_norm": 2.805513381958008, "learning_rate": 4.8424887558649045e-08, "loss": 0.08038463592529296, "step": 36810 }, { "epoch": 0.9698949302443506, "grad_norm": 2.3946123123168945, "learning_rate": 4.7589934438228017e-08, "loss": 0.07740173935890197, "step": 36820 }, { "epoch": 0.9701583454888493, "grad_norm": 3.024116277694702, "learning_rate": 4.676222511123762e-08, "loss": 0.08344019651412964, "step": 36830 }, { "epoch": 0.9704217607333481, "grad_norm": 5.282049655914307, "learning_rate": 4.59417601801182e-08, "loss": 0.09052765369415283, "step": 36840 }, { "epoch": 0.9706851759778468, "grad_norm": 3.098071336746216, "learning_rate": 4.512854024203983e-08, "loss": 0.09533540010452271, "step": 36850 }, { "epoch": 0.9709485912223456, "grad_norm": 2.0724098682403564, "learning_rate": 4.432256588890016e-08, "loss": 0.07746735811233521, "step": 36860 }, { "epoch": 0.9712120064668442, "grad_norm": 3.2085554599761963, "learning_rate": 4.352383770732216e-08, "loss": 0.07085311412811279, "step": 36870 }, { "epoch": 0.9714754217113429, "grad_norm": 4.40797758102417, "learning_rate": 4.273235627865302e-08, "loss": 0.07046605348587036, "step": 36880 }, { "epoch": 0.9717388369558417, "grad_norm": 3.7317450046539307, "learning_rate": 4.19481221789686e-08, "loss": 0.07778849601745605, "step": 36890 }, { "epoch": 0.9720022522003404, "grad_norm": 3.7790048122406006, "learning_rate": 4.117113597906675e-08, "loss": 0.0742936372756958, "step": 36900 }, { "epoch": 0.9722656674448392, "grad_norm": 3.352574586868286, "learning_rate": 4.040139824447398e-08, "loss": 0.07595282793045044, "step": 36910 }, { "epoch": 0.9725290826893379, "grad_norm": 4.1413350105285645, "learning_rate": 3.963890953543659e-08, "loss": 0.07602373957633972, "step": 36920 }, { "epoch": 0.9727924979338367, "grad_norm": 3.9709181785583496, "learning_rate": 3.88836704069262e-08, "loss": 0.0773328721523285, "step": 36930 }, { "epoch": 0.9730559131783354, "grad_norm": 4.199800968170166, "learning_rate": 3.813568140864088e-08, "loss": 0.09370261430740356, "step": 36940 }, { "epoch": 0.9733193284228342, "grad_norm": 4.016210079193115, "learning_rate": 3.739494308499736e-08, "loss": 0.09207624793052674, "step": 36950 }, { "epoch": 0.9735827436673329, "grad_norm": 3.190653085708618, "learning_rate": 3.66614559751377e-08, "loss": 0.08349753022193909, "step": 36960 }, { "epoch": 0.9738461589118316, "grad_norm": 3.945061683654785, "learning_rate": 3.593522061292598e-08, "loss": 0.06225389838218689, "step": 36970 }, { "epoch": 0.9741095741563304, "grad_norm": 4.5246124267578125, "learning_rate": 3.521623752694603e-08, "loss": 0.08834230899810791, "step": 36980 }, { "epoch": 0.9743729894008291, "grad_norm": 3.4223575592041016, "learning_rate": 3.4504507240507027e-08, "loss": 0.07711650729179383, "step": 36990 }, { "epoch": 0.9746364046453279, "grad_norm": 3.535166025161743, "learning_rate": 3.380003027163459e-08, "loss": 0.06669443249702453, "step": 37000 }, { "epoch": 0.9748998198898265, "grad_norm": 3.6270833015441895, "learning_rate": 3.310280713307856e-08, "loss": 0.07273004055023194, "step": 37010 }, { "epoch": 0.9751632351343253, "grad_norm": 3.819063663482666, "learning_rate": 3.241283833230746e-08, "loss": 0.07382599115371705, "step": 37020 }, { "epoch": 0.975426650378824, "grad_norm": 4.087609767913818, "learning_rate": 3.1730124371511794e-08, "loss": 0.06988371610641479, "step": 37030 }, { "epoch": 0.9756900656233228, "grad_norm": 2.9559390544891357, "learning_rate": 3.1054665747599635e-08, "loss": 0.06831412315368653, "step": 37040 }, { "epoch": 0.9759534808678215, "grad_norm": 3.817091941833496, "learning_rate": 3.038646295219771e-08, "loss": 0.09696980118751526, "step": 37050 }, { "epoch": 0.9762168961123202, "grad_norm": 3.164090871810913, "learning_rate": 2.972551647165367e-08, "loss": 0.08492853641510009, "step": 37060 }, { "epoch": 0.976480311356819, "grad_norm": 3.5060603618621826, "learning_rate": 2.9071826787033797e-08, "loss": 0.08991841077804566, "step": 37070 }, { "epoch": 0.9767437266013177, "grad_norm": 2.8292593955993652, "learning_rate": 2.8425394374120842e-08, "loss": 0.07178260087966919, "step": 37080 }, { "epoch": 0.9770071418458165, "grad_norm": 2.9002645015716553, "learning_rate": 2.7786219703415108e-08, "loss": 0.07620673179626465, "step": 37090 }, { "epoch": 0.9772705570903152, "grad_norm": 2.4541521072387695, "learning_rate": 2.715430324013779e-08, "loss": 0.05855629444122314, "step": 37100 }, { "epoch": 0.977533972334814, "grad_norm": 3.6320855617523193, "learning_rate": 2.6529645444223206e-08, "loss": 0.08513114452362061, "step": 37110 }, { "epoch": 0.9777973875793127, "grad_norm": 4.192796230316162, "learning_rate": 2.5912246770324333e-08, "loss": 0.07665727734565735, "step": 37120 }, { "epoch": 0.9780608028238115, "grad_norm": 3.3813674449920654, "learning_rate": 2.5302107667811716e-08, "loss": 0.08083423972129822, "step": 37130 }, { "epoch": 0.9783242180683102, "grad_norm": 3.326260566711426, "learning_rate": 2.469922858076901e-08, "loss": 0.07541903853416443, "step": 37140 }, { "epoch": 0.9785876333128088, "grad_norm": 5.418677806854248, "learning_rate": 2.410360994799965e-08, "loss": 0.08768891096115113, "step": 37150 }, { "epoch": 0.9788510485573076, "grad_norm": 3.3006787300109863, "learning_rate": 2.351525220301909e-08, "loss": 0.0835734486579895, "step": 37160 }, { "epoch": 0.9791144638018063, "grad_norm": 3.708540439605713, "learning_rate": 2.293415577406144e-08, "loss": 0.08326730728149415, "step": 37170 }, { "epoch": 0.9793778790463051, "grad_norm": 4.1878275871276855, "learning_rate": 2.2360321084071712e-08, "loss": 0.066817706823349, "step": 37180 }, { "epoch": 0.9796412942908038, "grad_norm": 3.481401205062866, "learning_rate": 2.179374855071137e-08, "loss": 0.08952552080154419, "step": 37190 }, { "epoch": 0.9799047095353026, "grad_norm": 3.373746633529663, "learning_rate": 2.123443858635832e-08, "loss": 0.0740240216255188, "step": 37200 }, { "epoch": 0.9801681247798013, "grad_norm": 2.656270980834961, "learning_rate": 2.068239159810137e-08, "loss": 0.07666639685630798, "step": 37210 }, { "epoch": 0.9804315400243001, "grad_norm": 5.1515936851501465, "learning_rate": 2.0137607987743558e-08, "loss": 0.07475238442420959, "step": 37220 }, { "epoch": 0.9806949552687988, "grad_norm": 3.753035306930542, "learning_rate": 1.9600088151803255e-08, "loss": 0.0742489755153656, "step": 37230 }, { "epoch": 0.9809583705132975, "grad_norm": 3.378284454345703, "learning_rate": 1.906983248150973e-08, "loss": 0.07131866216659546, "step": 37240 }, { "epoch": 0.9812217857577963, "grad_norm": 3.23266863822937, "learning_rate": 1.854684136280538e-08, "loss": 0.0768273949623108, "step": 37250 }, { "epoch": 0.981485201002295, "grad_norm": 4.142930030822754, "learning_rate": 1.8031115176346815e-08, "loss": 0.08557204008102418, "step": 37260 }, { "epoch": 0.9817486162467938, "grad_norm": 3.559143304824829, "learning_rate": 1.752265429750044e-08, "loss": 0.08882511854171753, "step": 37270 }, { "epoch": 0.9820120314912925, "grad_norm": 4.47169303894043, "learning_rate": 1.7021459096348004e-08, "loss": 0.08154500722885132, "step": 37280 }, { "epoch": 0.9822754467357913, "grad_norm": 2.9045212268829346, "learning_rate": 1.652752993767992e-08, "loss": 0.07244112491607665, "step": 37290 }, { "epoch": 0.98253886198029, "grad_norm": 2.9126698970794678, "learning_rate": 1.604086718099973e-08, "loss": 0.07872928380966186, "step": 37300 }, { "epoch": 0.9828022772247887, "grad_norm": 4.144546031951904, "learning_rate": 1.556147118051965e-08, "loss": 0.07599409818649291, "step": 37310 }, { "epoch": 0.9830656924692874, "grad_norm": 3.116420269012451, "learning_rate": 1.508934228516723e-08, "loss": 0.07679245471954346, "step": 37320 }, { "epoch": 0.9833291077137861, "grad_norm": 3.3042190074920654, "learning_rate": 1.4624480838576483e-08, "loss": 0.07897761464118958, "step": 37330 }, { "epoch": 0.9835925229582849, "grad_norm": 2.803739309310913, "learning_rate": 1.4166887179094535e-08, "loss": 0.06951371431350709, "step": 37340 }, { "epoch": 0.9838559382027836, "grad_norm": 3.00223708152771, "learning_rate": 1.3716561639777192e-08, "loss": 0.0770606517791748, "step": 37350 }, { "epoch": 0.9841193534472824, "grad_norm": 3.9872593879699707, "learning_rate": 1.3273504548392269e-08, "loss": 0.08413544893264771, "step": 37360 }, { "epoch": 0.9843827686917811, "grad_norm": 2.876662492752075, "learning_rate": 1.2837716227414032e-08, "loss": 0.06648480892181396, "step": 37370 }, { "epoch": 0.9846461839362799, "grad_norm": 2.489630937576294, "learning_rate": 1.240919699402765e-08, "loss": 0.07373791933059692, "step": 37380 }, { "epoch": 0.9849095991807786, "grad_norm": 3.6133229732513428, "learning_rate": 1.1987947160130297e-08, "loss": 0.07642838954925538, "step": 37390 }, { "epoch": 0.9851730144252773, "grad_norm": 4.140716552734375, "learning_rate": 1.1573967032324495e-08, "loss": 0.08433569073677064, "step": 37400 }, { "epoch": 0.9854364296697761, "grad_norm": 3.6831257343292236, "learning_rate": 1.1167256911922552e-08, "loss": 0.07518775463104248, "step": 37410 }, { "epoch": 0.9856998449142748, "grad_norm": 3.6383676528930664, "learning_rate": 1.0767817094945454e-08, "loss": 0.07626051902770996, "step": 37420 }, { "epoch": 0.9859632601587736, "grad_norm": 3.5468087196350098, "learning_rate": 1.0375647872123974e-08, "loss": 0.07155324220657348, "step": 37430 }, { "epoch": 0.9862266754032722, "grad_norm": 4.557754039764404, "learning_rate": 9.990749528894227e-09, "loss": 0.07736894488334656, "step": 37440 }, { "epoch": 0.986490090647771, "grad_norm": 4.17971658706665, "learning_rate": 9.613122345403236e-09, "loss": 0.08173721432685851, "step": 37450 }, { "epoch": 0.9867535058922697, "grad_norm": 3.750587224960327, "learning_rate": 9.24276659650336e-09, "loss": 0.0802665650844574, "step": 37460 }, { "epoch": 0.9870169211367685, "grad_norm": 3.2527756690979004, "learning_rate": 8.879682551755643e-09, "loss": 0.09130132794380189, "step": 37470 }, { "epoch": 0.9872803363812672, "grad_norm": 2.520150661468506, "learning_rate": 8.523870475429797e-09, "loss": 0.06389182806015015, "step": 37480 }, { "epoch": 0.9875437516257659, "grad_norm": 3.1287715435028076, "learning_rate": 8.175330626498668e-09, "loss": 0.07279114127159118, "step": 37490 }, { "epoch": 0.9878071668702647, "grad_norm": 3.4946529865264893, "learning_rate": 7.834063258645997e-09, "loss": 0.08548706769943237, "step": 37500 }, { "epoch": 0.9880705821147634, "grad_norm": 3.232144594192505, "learning_rate": 7.500068620260869e-09, "loss": 0.08434526920318604, "step": 37510 }, { "epoch": 0.9883339973592622, "grad_norm": 3.5866501331329346, "learning_rate": 7.173346954438831e-09, "loss": 0.08140432834625244, "step": 37520 }, { "epoch": 0.9885974126037609, "grad_norm": 3.5442845821380615, "learning_rate": 6.853898498982992e-09, "loss": 0.07826769351959229, "step": 37530 }, { "epoch": 0.9888608278482597, "grad_norm": 3.3313863277435303, "learning_rate": 6.541723486399587e-09, "loss": 0.0695716142654419, "step": 37540 }, { "epoch": 0.9891242430927584, "grad_norm": 3.205204486846924, "learning_rate": 6.236822143904642e-09, "loss": 0.07108571529388427, "step": 37550 }, { "epoch": 0.9893876583372572, "grad_norm": 3.451540231704712, "learning_rate": 5.939194693418415e-09, "loss": 0.07588871717453002, "step": 37560 }, { "epoch": 0.9896510735817559, "grad_norm": 5.07623291015625, "learning_rate": 5.648841351566514e-09, "loss": 0.0805634617805481, "step": 37570 }, { "epoch": 0.9899144888262545, "grad_norm": 3.539172887802124, "learning_rate": 5.365762329681002e-09, "loss": 0.07214661836624145, "step": 37580 }, { "epoch": 0.9901779040707533, "grad_norm": 2.5865638256073, "learning_rate": 5.089957833798176e-09, "loss": 0.08194873332977295, "step": 37590 }, { "epoch": 0.990441319315252, "grad_norm": 2.6363794803619385, "learning_rate": 4.8214280646619035e-09, "loss": 0.08321574926376343, "step": 37600 }, { "epoch": 0.9907047345597508, "grad_norm": 3.323806047439575, "learning_rate": 4.560173217718067e-09, "loss": 0.08131537437438965, "step": 37610 }, { "epoch": 0.9909681498042495, "grad_norm": 3.869969129562378, "learning_rate": 4.3061934831212235e-09, "loss": 0.07277969121932984, "step": 37620 }, { "epoch": 0.9912315650487483, "grad_norm": 2.370100975036621, "learning_rate": 4.0594890457268384e-09, "loss": 0.06726463437080384, "step": 37630 }, { "epoch": 0.991494980293247, "grad_norm": 3.5646793842315674, "learning_rate": 3.820060085097943e-09, "loss": 0.08136144876480103, "step": 37640 }, { "epoch": 0.9917583955377458, "grad_norm": 5.559628486633301, "learning_rate": 3.587906775501804e-09, "loss": 0.07826299667358398, "step": 37650 }, { "epoch": 0.9920218107822445, "grad_norm": 3.073551654815674, "learning_rate": 3.363029285907704e-09, "loss": 0.06348711848258973, "step": 37660 }, { "epoch": 0.9922852260267432, "grad_norm": 3.7994332313537598, "learning_rate": 3.145427779993604e-09, "loss": 0.08498048186302185, "step": 37670 }, { "epoch": 0.992548641271242, "grad_norm": 2.9633126258850098, "learning_rate": 2.935102416138369e-09, "loss": 0.08448224067687989, "step": 37680 }, { "epoch": 0.9928120565157407, "grad_norm": 2.819002628326416, "learning_rate": 2.732053347425101e-09, "loss": 0.06761745810508728, "step": 37690 }, { "epoch": 0.9930754717602395, "grad_norm": 3.347784996032715, "learning_rate": 2.536280721642248e-09, "loss": 0.08447184562683105, "step": 37700 }, { "epoch": 0.9933388870047382, "grad_norm": 3.383450508117676, "learning_rate": 2.3477846812813843e-09, "loss": 0.07283848524093628, "step": 37710 }, { "epoch": 0.993602302249237, "grad_norm": 3.93935489654541, "learning_rate": 2.16656536353832e-09, "loss": 0.07824652791023254, "step": 37720 }, { "epoch": 0.9938657174937356, "grad_norm": 3.137885808944702, "learning_rate": 1.9926229003119912e-09, "loss": 0.07570983171463012, "step": 37730 }, { "epoch": 0.9941291327382344, "grad_norm": 3.743903875350952, "learning_rate": 1.8259574182055707e-09, "loss": 0.07726111412048339, "step": 37740 }, { "epoch": 0.9943925479827331, "grad_norm": 3.592741012573242, "learning_rate": 1.6665690385253564e-09, "loss": 0.071458238363266, "step": 37750 }, { "epoch": 0.9946559632272318, "grad_norm": 4.778446197509766, "learning_rate": 1.5144578772807728e-09, "loss": 0.08570523262023926, "step": 37760 }, { "epoch": 0.9949193784717306, "grad_norm": 2.4649405479431152, "learning_rate": 1.36962404518437e-09, "loss": 0.08875195980072022, "step": 37770 }, { "epoch": 0.9951827937162293, "grad_norm": 4.582870960235596, "learning_rate": 1.232067647654045e-09, "loss": 0.08691973686218261, "step": 37780 }, { "epoch": 0.9954462089607281, "grad_norm": 3.504087448120117, "learning_rate": 1.1017887848085995e-09, "loss": 0.07827926874160766, "step": 37790 }, { "epoch": 0.9957096242052268, "grad_norm": 3.4798214435577393, "learning_rate": 9.78787551469962e-10, "loss": 0.07318630218505859, "step": 37800 }, { "epoch": 0.9959730394497256, "grad_norm": 3.499976396560669, "learning_rate": 8.630640371642963e-10, "loss": 0.08560901880264282, "step": 37810 }, { "epoch": 0.9962364546942243, "grad_norm": 3.10959529876709, "learning_rate": 7.546183261197826e-10, "loss": 0.08031674027442932, "step": 37820 }, { "epoch": 0.9964998699387231, "grad_norm": 3.1060032844543457, "learning_rate": 6.534504972688372e-10, "loss": 0.07283352017402649, "step": 37830 }, { "epoch": 0.9967632851832218, "grad_norm": 3.667722702026367, "learning_rate": 5.595606242447815e-10, "loss": 0.09366916418075562, "step": 37840 }, { "epoch": 0.9970267004277205, "grad_norm": 3.5113143920898438, "learning_rate": 4.729487753851736e-10, "loss": 0.07126797437667846, "step": 37850 }, { "epoch": 0.9972901156722193, "grad_norm": 4.284871578216553, "learning_rate": 3.9361501372958686e-10, "loss": 0.08577083349227906, "step": 37860 }, { "epoch": 0.997553530916718, "grad_norm": 3.385962963104248, "learning_rate": 3.2155939702072093e-10, "loss": 0.08021430969238282, "step": 37870 }, { "epoch": 0.9978169461612167, "grad_norm": 3.0940101146698, "learning_rate": 2.5678197770329094e-10, "loss": 0.08186119794845581, "step": 37880 }, { "epoch": 0.9980803614057154, "grad_norm": 3.3375744819641113, "learning_rate": 1.9928280292624835e-10, "loss": 0.0813709855079651, "step": 37890 }, { "epoch": 0.9983437766502142, "grad_norm": 3.227263927459717, "learning_rate": 1.4906191453833985e-10, "loss": 0.07682281136512756, "step": 37900 }, { "epoch": 0.9986071918947129, "grad_norm": 3.0187644958496094, "learning_rate": 1.061193490936585e-10, "loss": 0.0740546703338623, "step": 37910 }, { "epoch": 0.9988706071392117, "grad_norm": 3.228172540664673, "learning_rate": 7.045513784831315e-11, "loss": 0.07460197210311889, "step": 37920 }, { "epoch": 0.9991340223837104, "grad_norm": 2.3687469959259033, "learning_rate": 4.206930675820786e-11, "loss": 0.06468831300735474, "step": 37930 }, { "epoch": 0.9993974376282091, "grad_norm": 3.550696611404419, "learning_rate": 2.0961876485703358e-11, "loss": 0.0682080090045929, "step": 37940 }, { "epoch": 0.9996608528727079, "grad_norm": 1.8239085674285889, "learning_rate": 7.132862392955631e-12, "loss": 0.07592484951019288, "step": 37950 }, { "epoch": 0.9999242681172066, "grad_norm": 3.4450058937072754, "learning_rate": 5.822745441363964e-13, "loss": 0.07168815135955811, "step": 37960 }, { "epoch": 1.0, "step": 37963, "total_flos": 2.0205788732727283e+20, "train_loss": 0.14090690907605397, "train_runtime": 337693.1582, "train_samples_per_second": 7.195, "train_steps_per_second": 0.112 } ], "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": 2.0205788732727283e+20, "train_batch_size": 1, "trial_name": null, "trial_params": null }